Integrin (cid:1) M (cid:2) 2 Orchestrates and Accelerates Plasminogen Activation and Fibrinolysis by Neutrophils*

Plasmin, the pivotal thrombolytic enzyme, is generated on the surface of many cell types, where urokinase receptor (uPAR)-bound urokinase (uPA) activates cell-bound plasminogen (Plg). It has been reported that neutrophils mediate endogenous thrombolysis involving a uPA-dependent mechanism, and we previously demonstrated that both uPAR and integrin (cid:1) M (cid:2) 2 recognize uPA to con- trol cell migration and adhesion. In the present study, we report that the (cid:1) M (cid:2) 2 regulates neutrophil-dependent fi- brinolysis. Phorbol 12-myristate 13-acetate (PMA)-stimu-lated but not resting neutrophils dissolved fibrin clots, and this activity was not only uPA- and Plg-dependent but also (cid:1) M (cid:2) 2 -dependent. Purified (cid:1) M (cid:2) 2 directly bound uPA ( K d (cid:3) 40 n M ) and Plg ( K d (cid:3) 1 (cid:4) M ) in a dose-dependent and saturable manner. In Plg activation assays, addition of purified (cid:1) M (cid:2) 2 , but not a control protein, to a Neutrophils Preparation— fol- sedimentation erythrocytes hypotonic lysis residual erythrocytes. Clot Lysis Assays— PMN-mediated clot lysis was measured in a system using purified reagents or in plasma. In the former assay, resting or PMA-stimulated PMNs were added at a final of 1.5 (cid:3) 10 5 cells/well to microtiter plates 2.9 M fibrinogen, M Plg, 1 m M CaCl 2 , m M MgCl 2 in HEPES buffer (10 m M HEPES, m M 1% human fresh cell-free plasma). Clot for- mation was initiated by addition of 0.5 NIH units/ml thrombin, and the total volume of the assay was 200 (cid:3) l. In inhibition experiments, cells were preincubated in the absence or presence of selected protease inhibitors or F(ab) 2 fragments of antibodies to uPA or the (cid:1) M or (cid:2) 2 integrin subunits (50 (cid:3) g/ml) for 30 min at 37 °C, prior to incorporation into clots. Clot lysis was monitored as changes in turbidity at 405 nm at 37 °C for up to 5 h. For experiments in which (cid:1) M (cid:2) 2 on PMNs was activated with KIM 185 or Mn 2 (cid:4) , the cells were pre-treated with these reagents at (cid:5) and (cid:5) nm. Plg Activation Assays— Wild-type or mutant forms of sc-uPA (23 n M ) were incubated in the absence or presence of purified, soluble (cid:1) M (cid:2) 2 integrin (0–50 n M ) for 15 min at 37 °C in 10 m M HEPES buffer, pH 7.4, containing 150 m M NaCl, 1 m M MnCl 2 , 0.07% N -octylglucopyranoside, 0.02% BSA, and then increasing concentrations of Glu-Plg (0–10 (cid:3) M ) were added. Plm generation was determined in a Molecular Devices microplate reader in the presence of the Plm chromogenic substrate S-2251 (0.5 m M ) at 37 °C by (cid:2) A 405 over each 1-min interval and con- verted to Plm concentration by reference to a linear standard curve prepared with active site-titrated Plm. In the kinetic study rates of Plm generation were plotted against Glu-Plg concentration in a double-reciprocal manner to obtain the apparent kinetic constants for the reaction. In the inhibition studies, NIF (100 n M ) or the KD of uPA (50 n M ) were preincubated with the integrin and added to the sc-uPA/Plg mixture. When the effect of (cid:1) M (cid:2) 2 on the inactivation of Plm by (cid:1) 2 antiplasmin was measured, Plm (200 n M ) was preincubated in the absence or presence of (cid:1) M (cid:2) 2 (600 n M ) for 10 min at 37 °C followed by addition of S-2251 (0.5 m M ). Hydrolysis of the chromogenic substrate was measured continuously, and at 2 min the (cid:1) 2 -antiplasmin (1 (cid:3) M ) was added. were by incubation with 10 (cid:3) g of non- immune mouse IgG and 50 (cid:3) of protein G-agarose (Zymed Laboratories, San Francisco, CA) for 1 h at 4 °C. After centrifugation, the supernatants were mixed with 5 (cid:3) g of mAb to human anti- (cid:1) M (clone OKM I) or with non-immune mouse IgG at 4 °C overnight. The immune complexes were captured by incubating with 50 (cid:3) l of protein G-Sepha-rose for 2 h at 4 °C. The resin was washed twice with ice-cold lysis buffer; twice with TBS, 1 m M CaCl 2 , 2 m M MgCl 2 and eluted with SDS-PAGE loading buffer. Immunoprecipitates were analyzed on 8% gels by SDS-PAGE, electrophoretically transferred onto nitrocellulose membranes, and stained with goat anti-Plg, or mAbs to uPA, uPAR, or the (cid:1) M subunit. Plg Binding to (cid:1) M (cid:2) 2 on Beads— Purified (cid:1) M (cid:2) 2 was biotinylated using Sulfo-Link biotin (Pierce, Rockford, IL) according to the manufacturer’s instructions and immobilized on streptavidin-conjugated beads for 2 h at room temperature. After two washings with 10 m M HEPES buffer containing 0.1% N -octylglucopyranoside and 1 m M Mn 2 (cid:4) , nonspecific binding sites were blocked with biotin, followed by five washings with the HEPES buffer. Immobilized (cid:1) M (cid:2) 2 on the beads was mixed in a total volume of 200 (cid:3) l with combinations of Plg (20 (cid:3) g) and/or sc-uPA (10 (cid:3) g) in HEPES buffer fo r 1 h at 37°C. The resins were washed four times, and formed complexes were analyzed by Western blot as above.

adhesive responses. Components of the plasminogen (Plg) system and members of the integrin family contribute to the proteolytic and adhesive potentials of the PMN (1)(2)(3)(4). Among the proteolytic responsibilities of PMNs is their contribution to the degradation of fibrin-rich blood clots (5,6). Recent data have emphasized the importance of urokinase (uPA), released from intracellular stores, to activate Plg to plasmin (Plm) and, thereby, contribute to the fibrinolytic activity of these cells (5,7). PMNs also express the uPA receptor, uPAR, on their surface (7). Most recent studies of uPAR have emphasized its participation in cell migration, but its contribution to Plg activation has been difficult to demonstrate not only in vitro but also in vivo using deficient mice (8 -10). We have recently shown that the leukocyte integrin ␣ M ␤ 2 can function as a uPA receptor on PMN (11). This recognition influences PMN adhesion and migration (11), but the role of ␣ M ␤ 2 interaction with uPA in fibrinolysis has not been investigated. Plg can also bind to PMNs via its lysine binding sites (LBS) associated with its kringle domains (12), and numerous Plg-binding proteins, some present in PMN, have been identified (13)(14)(15)(16). Thus, PMNs exhibit multiple interactions with components of the Plg system, but it is unclear if and how these interactions affect the primary function of Plg, its role in fibrinolysis.
The adhesive and migratory responses of PMNs are critical to their participation in the inflammatory response, and numerous in vitro and in vivo studies have demonstrated that ␣ M ␤ 2 is an important mediator of these activities (4,17). Fibrinogen and ICAM-1 are among the ligands of this integrin that have been implicated in endothelial transmigration of PMNs (17,18). These and a myriad of other ligands for ␣ M ␤ 2 interact with its I domain, an inserted domain of ϳ200 amino acids in the ␣ M subunit (19). As noted above, uPA is a ligand for ␣ M ␤ 2 and can support both adhesion and migration of ␣ M ␤ 2bearing cells, including PMNs (11). uPAR also interacts within a 17-amino acid sequence, M25, located within W4 repeat of the ␤-propeller (20) of the ␣ M subunit, a region lying outside of the ␣ M I domain, and a physical complex between uPAR and ␣ M ␤ 2 has been demonstrated on the PMN surface (21,22). The interaction between uPAR and ␣ M ␤ 2 augments ␣ M ␤ 2 -mediated functions such as adhesion to fibrinogen, ICAM-1, and other ligands of this integrin (23,24), but uPAR contribution to Plg activation remains uncertain.
In this study, we identify a previously unrecognized function of ␣ M ␤ 2 , a direct and primary role of this integrin in Plg activation. This activity is demonstrated with purified ␣ M ␤ 2 and components of the Plg system as well as with intact PMNs and is shown to regulate the participation of PMNs in fibrinolysis. This function is particularly relevant in view of the preferential accumulation of these cells within thrombi (6) and the prominent role of Plg in the proteolytic and migratory responses of PMNs.
Neutrophils Preparation-Granulocytes were isolated from human peripheral blood of healthy volunteers drawn into sterile acid-citratedextrose (1/7 vol. 145 mM sodium citrate, pH 4.6, and 2% dextrose). Isolation was performed by means of density gradient centrifugation onto Ficoll-Hypaque (Amersham Biosciences, Uppsala, Sweden), followed by dextran sedimentation of erythrocytes and hypotonic lysis of residual erythrocytes.
Clot Lysis Assays-PMN-mediated clot lysis was measured in a system using purified reagents or in plasma. In the former assay, resting or PMA-stimulated PMNs were added at a final concentration of 1.5 ϫ 10 5 cells/well to microtiter plates containing 2.9 M fibrinogen, 0.5 M Plg, 1 mM CaCl 2 , 1 mM MgCl 2 in HEPES buffer (10 mM HEPES, 150 mM NaCl, pH 7.4, containing 1% human fresh cell-free plasma). Clot formation was initiated by addition of 0.5 NIH units/ml thrombin, and the total volume of the assay was 200 l. In inhibition experiments, cells were preincubated in the absence or presence of selected protease inhibitors or F(ab) 2 fragments of antibodies to uPA or the ␣ M or ␤ 2 integrin subunits (50 g/ml) for 30 min at 37°C, prior to incorporation into clots. Clot lysis was monitored as changes in turbidity at 405 nm at 37°C for up to 5 h. For experiments in which ␣ M ␤ 2 on PMNs was activated with KIM 185 or Mn 2ϩ , the cells were pre-treated with these reagents for 20 min, 37°C, incubated with 50 nM sc-uPA for 30 min at 37°C, and washed two to three times to remove unbound uPA before incorporation into clots.
The plasma clot also was measured turbidimetrically with some modifications of an assay previously described by Wu et al. (30,31). Human platelet-poor plasma was prepared from citrated blood by centrifugation at 1500 ϫ g for 15 min and used on the same day. PMN were pretreated with or without blocking reagents to uPA, Plm, or ␣ M ␤ 2 as specified above and added at 5 ϫ 10 5 /well in a total volume of 150 l to recalcified (20 mM CaCl 2 , final) platelet-poor plasma. Clotting was initiated by adding thrombin (2 NIH units/ml), and the plates were incubated at 37°C. Clots formed within 0.5 h and were lysed within 20 h, and these processes were monitored as the changes of absorbance at 600 nm.
Radioiodination of Proteins-Na 125 I (specific activity ϭ 15 mCi of 125 I/mg iodine) from Amersham Biosciences was used for radioiodination. HMW-tc-uPA, Plg, and goat anti-mouse IgG were radiolabeled using a modified chloramine-T method (32). The labeled HMW-tc-uPA and goat IgG were indistinguishable from the unlabeled forms upon SDS-PAGE.
Solid Phase Ligand Binding Assays-The binding of 125 I-uPA and 125 I-Plg to immobilized ␣ M ␤ 2 was performed as described by Wei et al. (33) with minor modifications. ␣ M ␤ 2 was purified on affinity chromatography column of mAb LM2/1.6.11 as previously described (34). The receptor (20 g/ml in TBS) was immobilized onto 96-well microtiter plates (Corning Costar Corp., Cambridge, MA) at 2 g/well for 20 h at 4°C. After post-coating with BSA, 125 I-HMW-tc-uPA was added (0 -1 M) in the presence of 1 mM MnCl 2 and incubated for 4 h at 37°C. Plates were then washed with TBS, and bound HMW-tc-uPA was quantitated by counting the bound radioactivity in a ␥-counter. In competition studies, various concentrations (0 -2 M) of non-labeled uPA derivatives were added together with the radiolabeled ligand to the assays, or 125 I-HMW-tc-uPA was preincubated with the ␣ M or ␣ L I domains (0 -100 nM) prior to the addition to the ␣ M ␤ 2 -coated wells. To measure Plg binding, varying concentrations of 125 I-Glu-Plg (0 -14 M) were added to ␣ M ␤ 2 -or BSA-coated plates in the presence of 1 mM MnCl 2 , with or without 50 M unlabeled Plg. After 2-h incubation at 37°C, the wells were washed, and bound 125 I-Glu-Plg was measured as described for uPA binding. Data were determined as triplicate measurements at each experimental point.
Diisopropyl fluorophosphate-treated Plm was labeled with Alexafluor-488 according to the manufacturer's protocol (Molecular Probes, Eugene, OR), incubated in the presence of increasing concentrations (0 -1.2 M) of purified ␣ M ␤ 2 , ␣ M I, or ␣ L domains, and allowed to bind to ␣ 2 -antiplasmin (1 g/well)-coated 96-well plates for 2 h at 37°C. After washing, the bound Plm was quantitated by measuring fluorescence at ex 480 and em 530 nm. Plg Activation Assays-Wild-type or mutant forms of sc-uPA (23 nM) were incubated in the absence or presence of purified, soluble ␣ M ␤ 2 integrin (0 -50 nM) for 15 min at 37°C in 10 mM HEPES buffer, pH 7.4, containing 150 mM NaCl, 1 mM MnCl 2 , 0.07% N-octylglucopyranoside, 0.02% BSA, and then increasing concentrations of Glu-Plg (0 -10 M) were added. Plm generation was determined in a Molecular Devices microplate reader in the presence of the Plm chromogenic substrate S-2251 (0.5 mM) at 37°C by ⌬A 405 over each 1-min interval and converted to Plm concentration by reference to a linear standard curve prepared with active site-titrated Plm. In the kinetic study rates of Plm generation were plotted against Glu-Plg concentration in a doublereciprocal manner to obtain the apparent kinetic constants for the reaction. In the inhibition studies, NIF (100 nM) or the KD of uPA (50 nM) were preincubated with the integrin and added to the sc-uPA/Plg mixture. When the effect of ␣ M ␤ 2 on the inactivation of Plm by ␣ 2antiplasmin was measured, Plm (200 nM) was preincubated in the absence or presence of ␣ M ␤ 2 (600 nM) for 10 min at 37°C followed by addition of S-2251 (0.5 mM). Hydrolysis of the chromogenic substrate was measured continuously, and at 2 min the ␣ 2 -antiplasmin (1 M) was added.
Plg Activation on the PMN Surface-PMNs were activated with 20 nM PMA, 20 g/ml KIM 185, or 0.5 mM Mn 2ϩ for 30 min at 37°C in the presence 25 nM sc-uPA in 10 mM Tris-Cl buffer, pH 7.4, containing 0.14 mM NaCl, 0.1% BSA, 1 mM CaCl 2 , and 1 mM MgCl 2 (unless Mn 2ϩ was present). The blocking mAbs to ␣ M , ␤ 2 , or uPAR were added at final concentrations of 20 g/ml. The cells were washed three times, and 50 l of the cell suspension (1 ϫ 10 6 cells/ml) was added to each well of the microtiter plate. Then, 100 l of a Glu-Plg (1 M) and plasmin-specific fluorescent substrate H-D-Val-Leu-Lys-7-amido-4-methylcoumarin (2 mM) mixture were added, and plasmin formation was monitored over 40 min at 37°C at ␣ em ϭ 370 nm and ␣ ex ϭ 470 nm using a fluorescence plate reader (SpectraMax GeminiXS, Molecular Devices).
Plg Binding to PMNs and Immunoprecipitation-Resting or PMAstimulated PMNs were resuspended (2 ϫ 10 7 cells/ml) in 1 ml of 10 mM HEPES buffer containing 150 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 and incubated with Plg (2 M) at 37°C for 30 min. Excess ligand was removed by centrifugation. After washing with HEPES buffer, the cells were lysed with buffer containing 20 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1 mM CaCl 2 , 2 mM MgCl 2 , 2% N-octylglucopyranoside, and protease inhibitor mixture (Roche Applied Science, Indianapolis, IN) for 1 h on ice. Cell debris was removed by centrifugation at 16,000 ϫ g for 15 min at 4°C. Lysates were precleared by incubation with 10 g of nonimmune mouse IgG and 50 l of protein G-agarose (Zymed Laboratories, San Francisco, CA) for 1 h at 4°C. After centrifugation, the supernatants were mixed with 5 g of mAb to human anti-␣ M (clone OKM I) or with non-immune mouse IgG at 4°C overnight. The immune complexes were captured by incubating with 50 l of protein G-Sepharose for 2 h at 4°C. The resin was washed twice with ice-cold lysis buffer; twice with TBS, 1 mM CaCl 2 , 2 mM MgCl 2 and eluted with SDS-PAGE loading buffer. Immunoprecipitates were analyzed on 8% gels by SDS-PAGE, electrophoretically transferred onto nitrocellulose membranes, and stained with goat anti-Plg, or mAbs to uPA, uPAR, or the ␣ M subunit.
Plg Binding to ␣ M ␤ 2 on Beads-Purified ␣ M ␤ 2 was biotinylated using Sulfo-Link biotin (Pierce, Rockford, IL) according to the manufacturer's instructions and immobilized on streptavidin-conjugated beads for 2 h at room temperature. After two washings with 10 mM HEPES buffer containing 0.1% N-octylglucopyranoside and 1 mM Mn 2ϩ , nonspecific binding sites were blocked with biotin, followed by five washings with the HEPES buffer. Immobilized ␣ M ␤ 2 on the beads was mixed in a total volume of 200 l with combinations of Plg (20 g) and/or sc-uPA (10 g) in HEPES buffer for 1 h at 37°C. The resins were washed four times, and formed complexes were analyzed by Western blot as described above.
Data Analysis-The data are expressed as means Ϯ S.E. To determine the significance of differences between two groups, Student's t test was performed using the Sigma-Plot software program (SPSS Inc.); p Ͻ 0.05 was considered significant.

RESULTS
Integrin ␣ M ␤ 2 Promotes PMN-mediated Fibrinolysis-In view of the known accumulation of PMNs in thrombi, recent data implicating uPA in the capacity of these cells to spontaneously lyse blood clots (1,6,35), and our own results showing that integrin ␣ M ␤ 2 (11) contributes to the adhesive and migratory responses to uPA, we sought to determine whether the integrin might influence PMN-mediated fibrin clot lysis. Isolated human blood PMNs were incorporated into fibrin clots enriched with Plg, and clot dissolution was monitored as changes in turbidity. The capacity of resting or PMA-stimulated PMNs to mediate fibrinolysis is shown in Fig. 1A. PMA-stimulated PMNs, but not resting cells, mediated clot lysis over a 5-h period. The observed time course of clot lysis time is consistent with previous reports (5,30). Thus, activation of the PMNs is essential for efficient fibrinolysis. The role of uPA in clot lysis mediated by stimulated PMNs was demonstrable with F(ab) 2 fragments of a function blocking mAb and amiloride, a low molecular weight uPA inhibitor. Both reagents decreased PMN-mediated fibrin dissolution by ϳ60 -70% (Fig.  1B). Since exogenous uPA was not added to the cells, the source of uPA must be the sc-uPA, which is known to be rapidly secreted from specific granules upon stimulation of PMN (1,7,35), and this assumption was verified in subsequent experiments (see below). The dependence of the lysis on Plm was demonstrated by the capacity of Trasylol, a Plm active site inhibitor, and 6-aminohexoic acid (6-AHA), an inhibitor of the LBS functions of Plm, to suppress fibrinolysis induced by the FIG. 1. PMN-dependent lysis of fibrin clots is influenced by ␣ M ␤ 2 and uPA. A, resting or PMA-stimulated PMNs were incorporated into fibrin clots, which were prepared in microtiter wells. B, in the inhibition experiments, cells were preincubated with the following reagents: 6-aminohexanoic acid (5 mM), Trasylol (100 units/ml), amiloride (300 M), NIF (100 nM), or F(ab) 2 fragments of either function blocking antibodies to uPA, ␣ M (44a), ␤ 2 (TS1/18) or control non-immune mouse IgG (50 g/ml) and then were added to the wells, which also contained these reagents at the same concentrations. C, PMNs stimulated with 20 nM PMA, 0.5 mM Mn 2ϩ , or 20 g/ml KIM 185 were preincubated with sc-uPA (50 nM) in the presence or absence of the ␣ M ␤ 2 function blocking mAbs (as above), washed, and incorporated into fibrin clots. Clot lysis was monitored at 405 nm at 37°C for up to 5 h. D, PMN were pretreated with the reagents as specified in B and added at 5 ϫ 10 5 /well to a total volume of 150 l of fresh human platelet-poor plasma. Thrombin (2 NIH units/ml) was added, the plates were incubated at 37°C, and the absorbance at 600 nm was monitored for up to 20 h. The data are expressed as percent fibrinolysis calculated from the relative clot turbidity after 20 h in the absence of PMNs. The data are the means Ϯ S.E. of quadruplicates measurements and are representative of three experiments.
PMA-stimulated PMNs. Together, these results verify the prominent role of uPA-mediated Plg activation in the fibrinolytic response of the cells.
The involvement of integrin ␣ M ␤ 2 in fibrinolysis was evaluated using a blocking approach. When neutrophil inhibitory factor (NIF), a high affinity ligand antagonist of ␣ M ␤ 2 (36,37) was added, it was as effective as the uPA and Plm inhibitors in suppressing clot lysis (Fig. 1B). Similarly, F(ab) 2 fragments of two different mAbs to ␣ M ␤ 2 , one to the ␣ M and the other to the ␤ 2 subunit, were effective in inhibiting PMN-dependent fibrinolysis (Fig. 1B), whereas non-immune mouse F(ab) 2 fragments had no effect. These data implicate not only Plg and uPA but also indicate that ␣ M ␤ 2 plays a major role in PMN-mediated lysis of fibrin clots. PMN-dependent fibrinolysis was not completely blocked by the anti-integrin reagents. The participation of plasminogen receptors (14,38), plasmin-independent fibrin degradation by neutrophil proteases (39) or activation of Plg that dissociates from cell surface by cell-bound or released uPA may all contribute to integrin-independent fibrinolysis. We directly assessed the role of two prominent Plg receptors on PMNs, annexin II (13) and ␣-enolase (14), using blocking mAbs and found that these reagents suppressed fibrinolysis by 15 and 25%, respectively (Fig. 1B), compared with the 50% inhibition by the integrin ␣ M ␤ 2 -blocking mAbs.
Because PMA not only induces integrin activation but also stimulates sc-uPA secretion, we investigated whether ␣ M ␤ 2 activation in itself is sufficient to affect PMN-mediated fibrinolysis. When PMNs were incubated with known activators of ␣ M ␤ 2 , 0.5 mM Mn 2ϩ (40), or the ␤ 2 -activating mAb, KIM 185 (27), the cells were not capable of supporting fibrin degradation (not shown); i.e. they behaved the same as non-stimulated PMN (see Fig. 1A). These reagents did induce ␣ M ␤ 2 activation as verified with mAb CBRM 1/5, which reacts with activated ␣ M ␤ 2 and did not induce sc-uPA and uPAR expression on PMN surface as assessed by FACS (not shown). However, when the cells were preincubated with exogenous sc-uPA, washed, and incorporated into fibrin clots, both Mn 2ϩ and KIM 185 induced PMN-mediated fibrinolysis (60 and 45%, respectively) as compared with unstimulated cells (Fig. 1C). The fibrinolysis induced by these reagents was dramatically reduced by ␣ M ␤ 2blocking mAbs (44a, TS1/18), confirming the crucial role of ␣ M ␤ 2 in the response. In control samples, KIM 185 did not induce fibrinolysis in the absence of cells.
We sought to determine if the role of ␣ M ␤ 2 in fibrinolysis could be demonstrated in a more physiological milieu, i.e. in plasma (Fig. 1D). Isolated PMNs lysed plasma clots within a 20-h period of time, which is a time consistent with such plasma clot lysis assays (41). Similar to the results shown in Fig. 1B, clot lysis in plasma was mediated by urokinase and plasmin, because 50 -60% inhibition was observed upon addition of 6-AHA, Trasylol, or a function-blocking mAb to uPA. F(ab) 2 fragments of mAbs to ␣ M ␤ 2 and NIF reduced the total clot lysis by 30%, which accounts for approximately 60 -70% of uPA-mediated fibrinolysis (Fig. 1D). Thus, the involvement of ␣ M ␤ 2 integrin in PMN-mediated fibrinolysis was observable even in the presence of physiological concentrations of ␣ 2 -antiplasmin, the major plasma inhibitor of plasmin, and other regulators of fibrinolysis.
Recognition of uPA and Plg by Purified ␣ M ␤ 2 -To define the mechanism by which ␣ M ␤ 2 influences fibrinolysis, we examined the capacity of the purified receptor to interact with uPA and Plg. ␣ M ␤ 2 was isolated by affinity chromatography on a mAb (LM2/1.6.11) (34) from HEK 293 cells expressing the recombinant receptor. The inset in Fig. 2A shows that the ␣ M ␤ 2 used was intact and free of detectable contaminants by SDS-PAGE. Also, in an enzyme-linked immunosorbent assay format with two different mAbs (clones 3936 and 62022.11), we excluded contamination of ␣ M ␤ 2 preparation with uPAR. The binding functions of purified receptor were evaluated by immobilizing it onto microtiter plates, a system utilized extensively to analyze ligand binding to integrins (33,42,43). The uPA ligand used was HMW-tc-uPA, and the Plg ligand was Glu-Plg. After radioiodination, both migrated as single bands with the same mobility as unlabeled proteins on SDS-PAGE (not shown). As shown in Fig. 2A, immobilized ␣ M ␤ 2 bound 125 I-HMW-tc-uPA in a saturable manner using Mn 2ϩ as both a divalent cation and activator of ␣ M ␤ 2 (40). The interaction of 125 I-HMW-tc-uPA with BSA-coated wells in Mn 2ϩ did not exceed 20% of that observed with ␣ M ␤ 2 -coated plates, and a 50-fold excess of unlabeled HMW-tc-uPA relative to 125 I-HMW-tc-uPA inhibited binding to ␣ M ␤ 2 to the same level as observed with BSA-coated wells (not shown), indicating specificity. Activation of ␣ M ␤ 2 by Mn 2ϩ was verified using the activation-specific mAb, CBRM 1/5, and little uPA binding was observed in the absence of Mn 2ϩ (see Fig. 2A) or in the presence of a non-activating cation, Ca 2ϩ . From Scatchard plots, the K d of the interaction was estimated to be 40 nM. The stoichiometry, the maximal number of HMWtc-uPA molecules specifically bound per immobilized ␣ M ␤ 2 (quantitated by protein assay), was 0.9, suggesting that each  Table I. The IC 50 values for both HMW-sc-uPA and HMW-tc-uPA (ϳ100 nM) are similar and consistent with the estimated K d (40 nM) of HMWtc-uPA for ␣ M ␤ 2 , providing validation of these data. The KD was a potent inhibitor with an IC 50 of 54 nM, whereas, the GFD, which interacts with uPAR (44), showed no propensity to interact with ␣ M ␤ 2 .
When immobilized ␣ M ␤ 2 was preincubated with mAbs (clones 44a and 904) to the ␣ M I domain, a region of ϳ200 amino acids within the ␣ M subunit that is involved in the binding of multiple ligands to the integrin (19), 125 I-HMW-tc-uPA binding was inhibited by 60 -80%, whereas normal mouse IgG had no effect (not shown). These results are consistent with the effects of these same mAbs on PMN-dependent fibrin clot lysis (Fig.  1C). Recombinant ␣ M I domain, but not ␣ L I domain, inhibited 125 I-HMW-tc-uPA binding to ␣ M ␤ 2 in a dose-dependent manner (Fig. 2B). Together, these results suggest that interaction of uPA with ␣ M ␤ 2 depends upon engagement of the K D of the ligand with the ␣ M I domain of the receptor. This profile derived with the purified receptor is entirely consistent with results previously obtained with ␣ M ␤ 2 -expressing cells (11).
Having established that u-PA can interact with purified ␣ M ␤ 2 , we sought to determine whether Plg itself also could bind to the isolated receptor. 125 I-Plg did bind to immobilized ␣ M ␤ 2 in a concentration-dependent and saturable manner (Fig. 3A). Under the same conditions, the binding of the highest concentration of 125 I-Plg (14 M) to BSA-coated wells did not exceed 20% of its binding to the ␣ M ␤ 2 -coated wells, and its binding to the receptor was abrogated (90% inhibition) by 50 M (a 3.6fold molar excess) unlabeled Plg. The data in Fig. 3A were analyzed with a Scatchard plot. The linearity of the plot indicated a single class of binding sites and a K d of 1 M, i.e. lower than the physiological concentration of Plg in plasma (2 M). When the binding of Alexa 488-labeled Plg to ␣ M ␤ 2 reached equilibrium the labeled ligand was replaced with 50 M unlabeled Plg to assess the reversibility of Plg binding to the integrin. In these kinetic dissociation experiments Plg fell off ␣ M ␤ 2 within 10 min at a dissociation rate constant k off ϭ 0.28 min Ϫ1 and a half-time of 2.5 min (not shown). Thus, Plg binding to the integrin is reversible, although dissociation seems to be rather slow for a low affinity interaction. Additionally, this binding was inhibited in a dose-dependent manner by 6-AHA (85% inhibition by 10 mM) (Fig 3B), implicating the LBS of Plg, which are known to mediate the interaction of Plg with cell surfaces (14,45), in ␣ M ␤ 2 recognition. Like u-PA, Plg binding to ␣ M ␤ 2 appeared to involve the ␣ M I domain. The interaction was inhibited by mAbs to the ␣ M I domain and by the purified ␣ M I domain, but not by the ␣ L I domain or mAbs to other regions of the ␣ M subunit (Fig. 3C). Although ␣ M I domain is the major uPA/Plg binding site, these interactions were also reduced by anti-␤ 2 mAb indicating that both subunits contribute to recognition, a pattern that is observed with certain other ␣ M ␤ 2 ligands, such as fibrinogen (46).   Fig. 2A. B, 125 I-Glu-Plg (2 M) was pretreated with varying concentrations (0 -10 mM) of 6-aminohexoic acid (6AHA) and then added to ␣ M ␤ 2 -or BSAcoated wells. After 2 h, the plates were processed as described. The background binding to BSA, which did not exceed 20% of total binding, has been subtracted. C, 125 I-Glu-Plg was incubated with recombinant ␣ L or ␣ M I domains for 1 h at 37°C and added to ␣ M ␤ 2 -or BSA-coated wells. Alternatively, ␣ M ␤ 2 was pretreated with function blocking mAbs to the integrin and subsequently allowed to bind 125 I-Glu-Plg for 2 h. The data are expressed as means Ϯ S.E. of triplets of three representative experiments.
Simultaneous Interaction of uPA and Plg with ␣ M ␤ 2 -Biotinylated ␣ M ␤ 2 , captured onto streptavidin-conjugated agarose, was incubated with Plg and/or sc-uPA. Following washing, bound proteins were eluted from the beads by boiling in SDS-PAGE sample buffer. After SDS-PAGE, nitrocellulose transfers were probed with anti-Plg, anti-uPA, or anti-␣ M mAbs (Fig.  4A). Blots developed with anti-␣ M mAb documented the capture of ␣ M ␤ 2 onto the agarose, and neither Plg nor sc-uPA affected the amount of the integrin bound. Although immobilized ␣ M ␤ 2 was capable of binding both Plg and sc-uPA, unarmed beads failed to bind either ligand. To confirm that both ligands could bind simultaneously to the same receptor, the experiment was repeated using radiolabeled Plg or sc-uPA. With 125 I-Plg, increasing amounts of uPA (DFP-inactivated) did not interfere with Plg binding to immobilized ␣ M ␤ 2 (Fig.  4B); and, conversely, with 125 I-sc-uPA, increasing amounts of Plg failed to affect the amount of sc-uPA bound (Fig. 4C). The binding of both ligands was reduced by anti-␣ M and anti-␤ 2 mAbs, indicating specificity. All of the results observed in the direct binding studies shown in Figs. 2-4 were confirmed by FACS analysis using Alexa-488-labeled uPA and Plg (not shown); i.e. PMA-stimulated PMNs bound the two ligands, and the interactions were suppressed by ␣ M ␤ 2 blocking reagents.
uPA Binding to ␣ M ␤ 2 Accelerates Plg Activation-The inhibitory effect of ␣ M ␤ 2 function blocking antibodies on PMN-dependent fibrinolysis and on direct binding of uPA and Plg to purified ␣ M ␤ 2 raised the possibility that these interactions might influence Plg activation. To test this hypothesis, we measured Plm generation in the presence of soluble, purified integrin, and sc-uPA, the inactive form of uPA secreted by PMNs (7,35). Sc-uPA was incubated in the presence or absence of soluble ␣ M ␤ 2 , followed by addition of increasing concentrations of Plg (0 -10 M) and Plm-specific chromogenic substrate (S-2251). Substrate hydrolysis was subsequently monitored at 405 nm. The kinetics of Plg activation by sc-uPA in the presence of ␣ M ␤ 2 followed a Michaelis-type mechanism as evidenced by the linearity of double-reciprocal plots of the data (Fig. 5A, inset). These data were derived at the optimal ␣ M ␤ 2 concentration of 50 nM. As the concentrations of ␣ M ␤ 2 added to sc-uPA (23 nM) and Plg (300 nM) were increased from 0 -200 nM, the rates of Plm generation also increased at integrin concentrations below 50 nM and then no further increase in Plm generation was observed. In control samples, direct activation of either Plg or sc-uPA by ␣ M ␤ 2 was not detected (not shown). These observations indicate that the potentiation of Plg activation is specifically mediated by ␣ M ␤ 2 , rather than by a protease activity contaminating the integrin preparations. In control experiments, tc-uPA activated Plg as expected, but further enhancement of plasmin formation by soluble ␣ M ␤ 2 was not observed.
When compared with Plg activation in the absence of the integrin (Fig. 5A), ␣ M ␤ 2 enhanced the overall efficiency (k cat / K m ) of the reaction (4.07 versus 0.067 M Ϫ1 s Ϫ1 ) (Table II) by ϳ60-fold. The individual kinetic constants derived from these data (Table II) indicate that this increased efficiency was primarily due to a 20-fold reduction in the K m for Plg activation from 2 to 0.1 M. These data can be most readily interpreted in terms of formation of a ternary complex between Plg, sc-uPA, and the integrin resulting in an increase in the apparent affinity of Plg for uPA, which leads to a reduced K m for Plg activation. ␣ M ␤ 2 increased the catalytic constant k cat only by 3-fold, indicating a less marked effect of the integrin on the dissociation of tc-uPA⅐plasmin(ogen) complex.
The kringle domain of u-PA is critical for its recognition by ␣ M ␤ 2 but not for its activation of Plg. We utilized this distinction to assess the importance of the uPA-␣ M ␤ 2 interaction in the enhancement of Plg activation, employing the following recombinant sc-uPAs: wild type (WT-sc-uPA), sc-uPA lacking the KD (⌬K-sc-uPA), and sc-uPA with deleted GFD (⌬GFD-sc-uPA). In the absence of the integrin, the sc-uPA forms used showed low intrinsic Plg activator activities, which were assigned a value of 1 (Fig. 5B). Integrin ␣ M ␤ 2 , but not human serum albumin, increased velocity of Plm generation by 17-to 19-fold when WT-sc-uPA or ⌬GFD-sc-uPA were added. This effect was not observed with ⌬K-sc-uPA, which is unable to bind ␣ M ␤ 2 . Moreover, consistent with the direct involvement of the interaction of sc-uPA with ␣ M ␤ 2 in the potentiation of Plg activation, preincubation of the integrin with NIF or recombi- FIG. 4. ␣ M ␤ 2 simultaneously interacts with u-PA and Plg. A, biotinylated soluble ␣ M ␤ 2 integrin was immobilized on Streptavidinagarose and nonspecific binding sites were blocked with biotin. Immobilized ␣ M ␤ 2 integrin on the beads was mixed with combinations of Plg (20 g) or/and sc-uPA (10 g) as indicated. The resins were washed and formed complexes were analyzed by Western blot as described above. Neither protease bound to Streptavidin-agarose alone after the nonspecific binding sites were blocked with biotin. Radioiodinated 125 I-Plg was preincubated with unlabeled DIP-uPA (B), or radiolabeled 125 I-DIP-uPA was preincubated with unlabeled Plg (C). Subsequently, the mixtures were added to and incubated for 2 h with ␣ M ␤ 2 immobilized on 96-well plates. In control samples, the immobilized integrin was pretreated with anti-␣ M or anti-␤ 2 mAbs, followed by addition radiolabeled uPA or Plg. After washing, the bound radioactivity was counted in a ␥-counter. The data are expressed as means Ϯ S.E. of quadruplets of two representative experiments.
nant KD suppressed the augmentation in Plg activation induced by ␣ M ␤ 2 (Fig. 5C). In control samples, when NIF or KD was incubated with sc-uPA alone, these reagents did not influence its intrinsic activator activity. Taken together, these data indicate that binding of uPA via its KD to ␣ M ␤ 2 is essential for stimulation of Plg activation.
␣ M ␤ 2 Delays Plm Inactivation by ␣ 2 -Antiplasmin-The involvement of the LBS in Plg binding to ␣ M ␤ 2 implies that Plm will also bind to the receptor and further suggests that receptor binding might influence its inactivation by its primary physiological inhibitor, ␣ 2 -antiplasmin, which interacts initially with the LBS and then with the active site of Plm (8,47,48). This prediction was tested. First, the capability of ␣ M ␤ 2 and its I domain to compete with ␣ 2 -antiplasmin for Plm binding was examined. As shown in Fig. 6A, both ␣ M ␤ 2 and ␣ M I domains, but not ␣ L I domain, suppressed Plm binding to immobilized ␣ 2 -antiplasmin. These results are consistent with the interaction of Plg with the ␣ M I domain and suggest that Plm also binds to this domain of ␣ M ␤ 2 . Next, we tested the capacity of Plm bound to ␣ M ␤ 2 to be inhibited by ␣ 2 -antiplasmin (Fig. 6B). Plm was preincubated with or without soluble ␣ M ␤ 2 followed by addition of the Plm chromogenic substrate S-2251. Plm activity was monitored by the changes of absorbance at 405 nm (Fig.  6B). The activities of free and ␣ M ␤ 2 -complexed Plm were similar, indicating that integrin binding did not influence Plm activity. At 2 min, ␣ 2 -antiplasmin was added. In the absence of ␣ M ␤ 2 , rapid and complete inhibition of Plm activity by the inhibitor was achieved. In contrast, ␣ M ␤ 2 suppressed the inhibition of Plm activity by ␣ 2 -antiplasmin. At the physiological concentration of ␣ 2 -antiplasmin (ϳ70 g/ml), the delay in inhibition by the integrin was ϳ7 min. The slow but time-dependent decrease in Plm activity in the presence of the integrin may reflect rapid neutralization as Plm dissociates from ␣ M ␤ 2 or slow interaction of ␣ 2 -antiplasmin with Plm bound to the integrin. The protective effect of ␣ M ␤ 2 on Plm inactivation by ␣ 2 -antiplasmin is consistent with that observed upon its binding to cell surfaces (8) as well as other candidate Plg receptors (14,49).
␣ M ␤ 2 -Mediated Assembly of the Plg System on PMNs-To determine whether ␣ M ␤ 2 , Plg, and uPA could co-associate on the leukocyte surface, Plg was added to resting or PMA-stimulated PMNs, followed by lysis of the cells and immunoprecipitation with a non-function blocking mAb (OKM1) to the ␣ M subunit. Immunoprecipitates were analyzed on Western blots with antibodies to Plg, uPAR, uPA, or ␣ M (Fig. 7A). Non-immune mouse IgG showed no protein bands in the ␣ M , Plg, uPAR, or uPA regions of the gels, and OKM1 did not react with Plg, uPA, and uPAR (not shown), establishing the specificity of the analytical system. This anti-␣ M mAb immunoprecipitated similar amounts of the subunit from both resting and stimulated PMNs as indicated in the blot probed with a second anti-␣ M mAb (ICO-GH1). A faint band of uPAR was detected in ␣ M immunoprecipitates from unstimulated cells incubated in the absence of Plg in agreement with previous data (22,23) demonstrating direct interaction between ␣ M ␤ 2 and uPAR. When Plg was added, uPAR and uPA immunoprecipitated with the integrin subunit. Because both ␣ M ␤ 2 and uPAR are translocated from intracellular granules to the cell surface in PMA-stimulated PMNs (7,50) and uPA interacts with both receptors (11), consistently ϳ20-fold more uPA was present in complexes from stimulated as compared with resting PMNs. Notably, upon stimulation with PMA, substantial amounts of Plg co-precipitated with the ␣ M subunit from the  /ml), and then Glu-Plg (300 nM) and the Plm-specific chromogenic substrate, S-2251 (0.5 mM) were added. Plm activity of Glu-Plg incubated alone or with ␣ M ␤ 2 was undetectable. The low intrinsic activities of utilized forms of sc-uPA in the absence of integrin were assigned a value 1. C, soluble ␣ M ␤ 2 was either untreated or incubated with NIF (100 nM) or the kringle domain of uPA (50 nM), followed by addition of sc-uPA/Plg mixture. To assess the effect of NIF and KD on the uPA intrinsic activity WT-sc-uPA was treated with these reagents as described above, then Plg was added. Plg activation was detected spectrophotometrically as described under "Experimental Procedures." The data are expressed as means Ϯ S.E. of quadruplets of two representative experiments. Plg-treated cells. This observation suggests the formation of a tetramolecular (␣ M ␤ 2 ⅐uPAR⅐uPA⅐Plg) complex on stimulated, Plg-treated PMNs. Next, we sought to determine the role of the various constituents of the tetramolecular complex in Plm generation on the PMN surface. Plm generation was measured using the fluorogenic plasmin-specific peptide substrate, and conditions were established under which little Plm was detected unless both Plg and sc-uPA were added to the cells. Little Plg activation was induced by non-stimulated PMNs even at added concentrations of sc-uPA as high as 25 nM, whereas PMA-stimulated PMNs enhanced the maximum velocity of Plg activation by 12-fold (Fig. 7B). The enhancement was fully blocked by NIF (100 nM) and significantly inhibited by mAbs to both the ␣ M I domain and the ␤ 2 subunit, but not by control mAb against MHC-1 (W6/32) or OKM1 (Fig. 7B). A function blocking mAb to uPAR also inhibited Plg activation. When this mAb was added together with a mAb directed against ␣ M subunit, the effect was additive and Plg activation was fully inhibited (Fig. 7B). Thus, both uPA receptors, ␣ M ␤ 2 and uPAR, cooperate for optimizing Plg activation on the PMN surface. When PMNs were stimulated with integrin-specific activators, Mn 2ϩ ions and KIM 185, the enhancement of plasmin formation was even greater (17-to 20-fold) than that observed with PMA (Fig. 7B), and it was also inhibited by ␣ M ␤ 2 -blocking reagents (not shown). KIM 185-induced Plg binding to the PMN surface by a similar increment, 18-fold, in soluble Plg binding as detected by FACS using Alexa-488-labeled Plg (Fig. 7B), and this increase was inhibited by 65-70% by ␣ M ␤ 2 -blocking mAbs. When PMNs were stimulated with PMA, this inhibition reached only 30 -40%, and the difference may reflect the contribution of other Plg-binding molecules (13)(14)(15)(16) that may be induced by PMA. Of note, Fg, at a physiological concentration of 3.5 mg/ml, did not compete with soluble Plg for binding ␣ M ␤ 2 (Fig. 7C), even though Fg also binds to the integrin. DISCUSSION In the present study, we have shown that integrin ␣ M ␤ 2 participates in Plg activation on the surface of human PMNs and, thereby, influences the fibrinolytic response of these cells. Specifically, we have provided evidence that 1) stimulated PMNs have the ability to lyse fibrin clots in an ␣ M ␤ 2 -dependent manner and 2) the mechanism underlying this involvement depends upon the direct and simultaneous binding of uPA and Plg to the integrin. These interactions were demonstrable with purified ␣ M ␤ 2 and on the surface of PMNs. The concurrent binding of sc-uPA and Plg to ␣ M ␤ 2 enhanced Plm formation by ϳ60-fold, primarily due to 20-fold decrease in K m . uPA binding to ␣ M ␤ 2 via its kringle domain is crucial for this effect; a mutant form of sc-uPA lacking the kringle, ⌬K-sc-uPA, failed to enhance Plg activation. Demonstrating the importance of Plg binding to ␣ M ␤ 2 in this effect, 6-AHA, which interferes with the LBS activities of Plg but does not inhibit uPA binding, suppressed the integrin-dependent enhancement of Plg activation. The capacity of uPA, Plg, and ␣ M ␤ 2 to interact on intact PMNs was detected by co-immunoprecipitation, and each of these interactions was necessary for ␣ M ␤ 2 to support enhanced Plg activation on the PMN surface. An additional important consequence of the binding of Plm/Plg to ␣ M ␤ 2 is a substantial delay in the inhibition of Plm activity by ␣ 2 -antiplasmin. It has been estimated that blockade of the LBS of Plm prolongs its half-life with ␣ 2 -antiplasmin by ϳ50-fold (48). Consequently, the overall effect of ␣ M ␤ 2 on Plm activity may be ϳ3000-fold. Thus, the mechanism underlying the capacity of stimulated PMNs to induce fibrinolysis depends upon the formation of a complex between Plg, sc-uPA, and ␣ M ␤ 2 , and the functional consequences of these interactions on Plm activity.
Purified and immobilized ␣ M ␤ 2 bound uPA. This interaction is mediated primarily by the ␣ M I domain based upon inhibition by: (a) recombinant ␣ M I domain but not ␣ L -I domain; (b) NIF, a high affinity ␣ M I domain ligand (11); and (c) mAbs, which bind to the domain (not shown). Optimal recognition of uPA may also involve the ␤ 2 subunit, based upon mAb inhibition, and appears to require activation of ␣ M ␤ 2 , as evidenced by the enhanced interaction observed in the presence of Mn 2ϩ . Mn 2ϩ activates ␣ M ␤ 2 by inducing an "open" conformation of the ␣ M I domain (40). In addition, purified ␣ M ␤ 2 bound Plg, and Plg co-immunoprecipitates with ␣ M ␤ 2 from the lysates of PMNs, indicating that this interaction can occur on the cell surface. It has been shown that cell binding of intact Glu-Plg facilitates its conversion to Lys-Plg, which lacks 80 -90 amino acids from the N terminus and is more readily activated to Plm (51). Whether this mechanism contributes to the increase in Plg activation induced by ␣ M ␤ 2 remains to be tested. The C-terminal lysine analog, 6-AHA, inhibited Plg binding to the integrin, implicating the LBS of Plg in the interaction (12,52). Thus, ␣ M ␤ 2 resembles other Plg/Plm receptors, including integrins ␣ IIb ␤ 3 (15) and ␣ V ␤ 3 (53), as well as many non-integrin Plg receptors, including annexin II (13) and ␣-enolase (14), with regard to involvement of the LBS in recognition. Although ␣ M ␤ 2 apparently does not contain a C-terminal lysine, certain internal basic amino acids can mimic these and interact with LBS (54). We cannot exclude limited proteolysis of the ␣ M ␤ 2 preparations that might generate a C-terminal lysine; however, the ␣ M ␤ 2 preparations seemed to be intact by SDS-PAGE, and the involvement of the internally located ␣ M I domain would argue against this possibility. The role of the LBS in ␣ M ␤ 2 recognition of Plg implicates kringles 1 and/or 5 in the interaction (55). Because the kringle of uPA lacks LBS activity (56), it is reasonable that both uPA and Plg could bind simultaneously to ␣ M ␤ 2 via their kringles as we have found and yet not compete with one another.
Although other Plg binding molecules are expressed on PMN, notably annexin II (13) or ␣-enolase (14), ␣ M ␤ 2 appears to account for ϳ40% of total Plg binding capacity of PMAstimulated PMNs under the conditions of our analyses, whereas 15 and 20% inhibition was achieved with anti-␣-enolase or anti-annexin II mAbs. Additionally, uPA-mediated Plg activation on PMN surface was reduced by 50% by ␣ M ␤ 2 -blocking reagents, whereas addition of mAbs to ␣-enolase resulted in 25% inhibition and the blocking mAb to annexin II had no effect. Thus, annexin II does not seem to mediate Plg activation on the PMN surface, although a prominent role for it in Plg activation on macrophages has been suggested (29). A profound role of ␣-enolase in Plg activation, but not Plg binding, was shown by Lopez-Alemany et al. (57) using high concentrations of a mAb to ␣-enolase. We observed a concordance between the role of enolase in Plg binding and activation on PMA-stimulated PMNs with the contribution to both functions being ϳ25%.
Our data indicate that ␣ M ␤ 2 , either expressed on cell surfaces (11) or in a purified form, can bind uPA in the absence of uPAR. Several recent reports (e.g. the mitogenic effects of uPA on human lung epithelial and melanoma cells) provide evidence for uPA-mediated cellular responses independent of uPAR (58,59), and Plg activation can be induced by uPA on the surface of cells lacking uPAR (60). Additionally, uPA was im-plicated in fibrin clearance and Plm-mediated pericellular proteolysis during vascular wound healing independent of uPAR in vivo (9,10). Hence, the capacity to potentiate sc-uPA-catalyzed Plg activation in the absence of uPAR is not without precedence. Nevertheless, on the surface of PMN, we were able to implicate uPAR in Plg activation with uPAR mAbs. This involvement is consistent with our previous data implicating uPAR in ␣ M ␤ 2 -dependent cell migration (11) as well as the results of other studies (23,24). Potentially, uPAR may serve as an enhancer of ␣ M ␤ 2 functions by presenting uPA to the integrin, or by influencing the activation of the integrin, which is necessary for uPA binding. Either of these effects of uPAR would enhance the apparent affinity of ␣ M ␤ 2 for uPA, which is relatively low (K d ϳ 40 nM) compared with that uPAR (K d ϳ 1 nM). Consequently, uPAR, uPA, ␣ M ␤ 2 , and Plg may form a multicontact, tetramolecular complex on the PMN surface, which profoundly influences the thrombolytic functions.
Plg interacted with ␣ M ␤ 2 as a soluble ligand, and this interaction was not inhibited by Fg. We have also observed that Fg does not inhibit PMN adhesion to Plg (not shown). These observations suggest that ␣ M ␤ 2 can recognize Plg even in the face of a vast excess of another ligand of the integrin. We have previously shown that Fg also is a poor inhibitor of uPA binding to ␣ M ␤ 2 , reducing the interaction by only ϳ40% at physiological concentrations of Fg (11). Our observation that ␣ M ␤ 2 can mediate plasmin-dependent fibrinolysis (Fig. 1B) adds credence to the biological significance of the recognition of fibrinolytic component occurrence by the integrin. Previously, Simon et al. (61) implicated ␣ M ␤ 2 in plasmin-independent fibrin clearance involving internalization by monocytes. Our data suggest a second role for the integrin in fibrinolysis, a Plg-dependent func- FIG. 7. Integrin ␣ M ␤ 2 co-associates with components of Plg system on PMNs. A, resting or PMA-stimulated PMNs were preincubated in the presence or absence of Glu-Plg (2 M). After washing, cells were lysed and ␣ M ␤ 2 was immunoprecipitated as described under "Experimental Procedures." The immune complexes were recovered, separated on 8% gels by SDS-PAGE, and analyzed by Western blot using the specific antibodies indicated. B, resting PMNs or PMNs stimulated with 20 nM PMA, 0.5 mM Mn 2ϩ , or KIM 185 (20 g/ml) were pretreated with 25 nM sc-uPA in the presence or absence of function-blocking mAbs to the integrin or uPAR as indicated, washed and added to a mixture of Glu-Plg and plasmin-specific fluorescent substrate. Plasmin formation was measured by the fluorescence at ex ϭ 470 nm. The data are the means Ϯ S.E. of quadruplets and are representative of three experiments. C, unstimulated or KIM 185-treated PMNs were preincubated with Fg (3.5 mg/ml) or with blocking mAbs to ␣ M and ␤ 2 subunits followed by 30-min incubation with Alexa 488-labeled Glu-Plg (2 M). The cells were washed and bound Plg was measured by FACS analysis. The data are representative of duplicates from two experiments.
tion. The capacity of ␣ M ␤ 2 to bind and activate Plg may also be particularly relevant to migration and invasion of PMNs and other leukocytes during inflammatory responses. This mechanism may contribute to the blunting of inflammatory responses observed in mice deficient in ␣ M ␤ 2 (4) and components of the Plg system (2,3).