Modulation of Serpin Reaction through Stabilization of Transient Intermediate by Ligands Bound to α-Helix F*

Mechanism-based inhibition of proteinases by serpins involves enzyme acylation and fast insertion of the reactive center loop (RCL) into the central β-sheet of the serpin, resulting in mechanical inactivation of the proteinase. We examined the effects of ligands specific to α-helix F (αHF) of plasminogen activator inhibitor-1 (PAI-1) on the stoichiometry of inhibition (SI) and limiting rate constant (klim) of RCL insertion for reactions with β-trypsin, tissue-type plasminogen activator (tPA), and urokinase. The somatomedin B domain of vitronectin (SMBD) did not affect SI for any proteinase or klim for tPA but decreased the klim for β-trypsin. In contrast to SMBD, monoclonal antibodies MA-55F4C12 and MA-33H1F7, the epitopes of which are located at the opposite side of αHF, decreased klim and increased SI for every enzyme. These effects were enhanced in the presence of SMBD. RCL insertion for β-trypsin and tPA is limited by different subsequent steps of PAI-1 mechanism as follows: enzyme acylation and formation of a loop-displaced acyl complex (LDA), respectively. Stabilization of LDA through the disruption of the exosite interactions between PAI-1 and tPA induced an increase in the klim but did not affect the SI. Thus it is unlikely that LDA contributes significantly to the outcome of the serpin reaction. These results demonstrate that the rate of RCL insertion is not necessarily correlated with SI and indicate that an intermediate, different from LDA, which forms during the late steps of PAI-1 mechanism, and could be stabilized by ligands specific to αHF, controls bifurcation between the inhibitory and the substrate pathways.

Like other serpins, PAI-1 (Scheme 1, I) (31) inactivates a target proteinase (E) through a unique conformational mechanism (Fig. 1). In this mechanism, the formation of the acylenzyme intermediate (EϳIЈ) triggers the spontaneous insertion of a reactive center loop (RCL) as a strand 4 into the central ␤-sheet A, the 70 Å translocation of the acyl-enzyme to the opposite pole of the PAI-1 molecule, and enzyme inactivation because of the mechanical distortion of its catalytic site ( Fig. 1) (32)(33)(34). Despite the fact that the structures of both the Michaelis complex (Scheme 1, E⅐I) (35) and the final inhibitory complex (E-I*) (34,36) are known for several serpins and proteinases, it is not clear what transient intermediates form on the reaction pathway from E⅐I to E-I*, as well as how bifurcation is regulated between the inhibitory and substrate pathways of the serpin reaction (Scheme 1) (31).
The binding of proteinase to the initial docking site at the top of the serpin molecule ( Fig. 1) results in the Michaelis complex E⅐I (Scheme 1), cleavage of the scissile bond of the RCL, and acylation of the enzyme (EϳIЈ). The formation of a loop-displaced acyl complex (LDA (E-IЈ; Scheme 1)) was proposed as a limiting step for the reaction between PAI-1 and tPA (31). Because LDA forms prior to RCL insertion, enzyme translocation requires disruption of exosite interactions at the PAI-1/ tPA interface and dissociation of the primed part of cleaved RCL from the active site of the enzyme (31). In contrast to tPA, ␤-trypsin lacks exosite interactions with PAI-1. As a result, the rate of RCL insertion for the reaction of PAI-1 with ␤-trypsin is limited by formation of the acyl-enzyme EϳI (31). Thus, because of the lack of exosite interactions at the proteinase/ serpin interface, the limiting rate of RCL insertion (k lim ) for the reaction of PAI-1 with ␤-trypsin is 30 times higher than that for tPA. The substrate pathway (Scheme 1), where proteinase completes the normal catalytic cycle and the hydrolysis of the acylenzyme results in an inactive cleaved PAI-1 with inserted RCL * This work was supported by American Heart Association Grant-in-aid 0550085Z (to A. A. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed: Dept. (I*), is an alternative to the inhibitory reaction. The stoichiometry of inhibition (SI ϭ 1 ϩ k s /k i ; the number of moles of PAI-1 required for inactivation of 1 mol of proteinase) quantitatively expresses the distribution between the inhibitory and the substrate pathways (Scheme 1) (37). The binding of ligands to PAI-1 often affects kinetics of the reaction with proteinase or (and) results in an increase in the fraction of the substrate reaction. Active PAI-1 circulates in the bloodstream as a high affinity complex with vitronectin (Vn) (38), a cell adhesive glycoprotein, which is present in plasma in micromolar concentrations (39,40), and stabilizes the active conformation of PAI-1 through a decrease in the rate of its spontaneous inactivation (41,42). Although Vn does not significantly affect the SI for the reactions of PAI-1 with uPA and tPA (43)(44)(45), under physiological pH it decreases the k lim for both enzymes and dramatically increases the rate of the reaction with activated protein C (46). On the other hand, the binding of Vn results in an increase in both the rate and SI for the reaction of PAI-1 with thrombin (43,47,48). Moreover, Vn dramatically enhances the fraction of the substrate reaction between tPA or uPA and complexes of PAI-1 with mAbs (MA-55F4C12, mAb-2) directed against ␣HF (44,45,49). Therefore, the binding of Vn and anti-␣HF mAbs to PAI-1 affects steps of the serpin mechanism, which contribute to distribution between substrate and inhibitory reactions.
To the best of our knowledge the contribution of the 44amino acid N-terminal SMBD to the effects of Vn on the PAI-1 mechanism has not been studied thus far. To address this question, we have evaluated the effects of recombinant SMBD on the kinetics of RCL insertion and stoichiometry of the reaction of uPA and tPA with PAI-1 or its complex with anti-␣HF mAb MA-55F4C12, and compared the results with the data obtained previously for the whole Vn isolated from human plasma (45). According to the Scheme 1, bifurcation between inhibitory and substrate pathways is controlled by LDA; however, disruption of the pattern of the exosite interactions at the PAI-1/tPA interface, which changes significantly k lim for the reaction of PAI-1 with tPA, does not affect SI (31,50,51). Therefore, a step of the serpin mechanism, which is likely different from stabilization/ destabilization of LDA, contributes to the outcome of PAI-1 reaction.
Effects of ligands, with opposite binding sites relative to ␣HF (recombinant SMBD and anti-␣HF mAbs MA-55F4C12 and MA-33H1F7) (52,53), on kinetics and the outcome of the reactions of PAI-1 with ␤-trypsin, tPA, and uPA were analyzed in order to identify steps of the serpin reaction, which could be targeted for PAI-1 neutralization through intermolecular mechanisms. Because RCL insertion for the reactions of PAI-1 with tPA and ␤-trypsin is limited by different steps of the mechanism (31), the effects of mAbs and SMBD on the kinetics and outcome of the reactions with these enzymes were measured and compared. Finally, the possible contribution of the exosite interactions to the effects of the ligands was evaluated. SI values for the reactions of a mutant variant of PAI-1 (glutamate to alanine mutations at the positions 4Ј and 5Ј of the RCL) with tPA and ␤-trypsin were measured and compared with the data obtained for wtPAI-1. The results obtained in this study strongly support the major contribution of the SMBD to the effects of Vn on the PAI-1/proteinase reaction and indicate formation of a new transition intermediate (␣HF intermediate), which appears at the late steps of the serpin mechanism and probably controls bifurcation between the inhibitory and substrate pathways of the reaction.

EXPERIMENTAL PROCEDURES
Proteins and Reagents-Bovine trypsin and analytical grade buffer reagents were from Sigma. Human plasma Vn was from Promega (Madison, WI). SMBD was obtained and characterized as described previously (53). Human recombinant PAI-1 and its S338C (P9 Cys) mutant variant were purified, labeled with an NBD group, and characterized as described elsewhere (54). NBD P9 E350A/E351A PAI-1, which was obtained and characterized as described elsewhere (50), was provided by Dr. C. Ibarra. The activity of PAI-1 was determined by titrating the Molecular models of the Michaelis complex (E⅐I, Scheme 1) and the inhibitory complex (E-I*) obtained for ␣1-proteinase inhibitor Pittsburgh and S195A trypsin (E⅐I) (35) and ␣1-proteinase inhibitor and porcine pancreatic elastase (E-I*) (34). The serpin molecule is shown in yellow; the RCL is shown in pink, and ␣HF is colored in cyan. The active proteinase bound at the initial docking site at the top of the serpin molecule is shown in blue. The inactivated acyl-enzyme, which was translocated to the opposite pole of the serpin molecule because of insertion of RCL as a strand 4 to the ␤-sheet A, is shown in red. The enzyme in the inhibitory complex is inactivated through the mechanical distortion of its structure. The figure was prepared using CHEM 3.1 software and Protein Data Bank files 1OPH and 2D26. SCHEME 1 ␣-Helix F and Modulation of Plasminogen Activator Inhibitor 1 SEPTEMBER 7, 2007 • VOLUME 282 • NUMBER 36 inhibitor with proteinase. Concentrations of wtPAI-1 were calculated from absorbance at 280 nm, using an extinction coefficient (⑀ 280 ) of 0.93 ml mg Ϫ1 cm Ϫ1 and an M r of 43,000 (55). Human recombinant tPA (activase) and uPA were provided by Genentech (San Francisco, CA) and by Abbott, respectively. Two-chain tPA was obtained from the single-chain enzyme by treatment with immobilized plasmin (56). Mouse mAbs MA-55F4C12 and MA-33H1F7 were selected from the panel raised against the human PAI-1-tPA complex (57). The concentrations of tPA, uPA, and mAbs were calculated from absorbance values at 280 nm, using M r values of 63,500, 54,000, and 150,000, and ⑀ 280 of 1.90, 1.36, and 1.3 ml mg Ϫ1 cm Ϫ1 , respectively. All experiments were performed using a 0.05 M phosphate buffer (pH 7.4 or 6.2).
Effects of SMBD on Kinetics of RCL Insertion Because of the Reaction of Proteinases with NBD P9 PAI-1 and Its Complexes with mAbs-Stopped flow fluorimetry was employed to examine the effects of SMBD on the rate of RCL insertion because of the reaction between proteinases and NBD P9 PAI-1 (10 -20 nM) or its complexes with mAbs (45). A micro-volume stopped flow reaction analyzer SF-61 DX2 double mixing stopped flow system (Hi-Tech Scientific) or SX-18MV (Applied Photophysics Ltd.) (58), equipped with a fluorescence detector and a thermostated cell, were used to monitor the changes in NBD fluorescence because of the reaction of proteinase with NBD P9 PAI-1 and its complexes with mAbs and SMBD. Briefly, SMBD (30 -80 nM) was added either to NBD P9 PAI-1 (15-30 nM) or to its preformed complex with mAb (1.5-2.0-fold molar excess over NBD P9 PAI-1) at least 10 min prior to data collection. Traces of changes in NBD fluorescence emission because of the reaction with proteinase were measured. Measurements were carried out in 50 mM phosphate buffer, pH 7.4, at 25°C. The data were analyzed using SigmaPlot 8.0 (SPSS Inc.) for Windows employing the nonlinear least squares Levenberg-Marquardt algorithm. Changes in NBD fluorescence emission (excitation at 480 nm, detection through 500 nm cutoff filter) with time were measured. A single exponential equation (F t ϭ A ϩ B(1 Ϫ e Ϫk obs t ), where F t is fluorescence emission at t seconds; A is the fluorescence at t ϭ 0; B is the amplitude of the fluorescence change, and k obs is the observed first-order rate constant) was fit to the data using either KinAssist (Hi-Tech Scientific) or SigmaPlot 8.0 software. The quality of the fit was estimated by visual analysis of plots of the residuals (deviation of the fitted function from the data). The values of k obs (average of 5-10 measurements; standard error (S.E.) less than 10%) were measured at concentration of proteinase at least 5-fold higher than that of NBD P9 PAI-1 or its complexes. The values of k obs obtained at different concentrations of uPA, tPA or ␤-trypsin were plotted against proteinase concentration using SigmaPlot 8.0, and plots were fit by a hyperbolic equation k obs ϭ k lim ϫ [E]/(K m ϩ [E]); k lim is the limiting rate constant of RCL insertion, and K m is the proteinase concentration ([E]) at halfsaturation (k obs ϭ k lim /2). Correlation coefficients (r 2 ) were used to estimate goodness of fitting (r 2 were Ն0.98 for all the data).
Effects of SMBD on Stoichiometry of Inhibition for the Reactions of Proteinases with NBD P9 PAI-1 and Their Complexes with mAb-The SI is the number of moles of PAI-1 (or its complex with ligand(s)) required for the inactivation of 1 mol of proteinase. The SI values for reactions of ␤-trypsin, tPA, or uPA were determined directly by titration of NBD P9 PAI-1 or its binary (with SMBD or mAb) or ternary (with both SMBD and mAb) complexes with proteinase using either Photon Technology International or Varian Cary Eclipse fluorescence spectrophotometer as described previously (59). Briefly, the known amount of NBD P9 PAI-1 (or its complex with ligand(s)) was titrated with small aliquots (less than 2% of NBD P9 PAI-1 volume) of proteinase. An increase in fluorescence emission because of the reaction of NBD P9 PAI-1 with the enzyme was measured and plotted against proteinase concentration. Starting from the equivalence point, the addition of proteinase did not induce changes in the NBD fluorescence. The values of SI (average of 3-5 measurements) were calculated from the amounts of proteinase required to reach the equivalence point as the ratios between the number of moles of NBD P9 PAI-1 present in solution and the number of moles of proteinase required for its complete titration.
Effects of SMBD on Stoichiometry of Inhibition of uPA with wtPAI-1 and Their Complexes with mAb-The effect of the SMBD of Vn on the SI for the reaction of uPA with wtPAI-1 and its complexes with MA-55F4C12 or MA-33H1F7 was determined from a decrease in the fluorescence emission of uPA/paminobenzamidine (PAB) complex because of the displacement of PAB by PAI-1 (58,60). The progress of the reaction between a mixture of uPA (50 -100 nM) with PAB (100 M) and increasing the amounts of wtPAI-1 or its complexes with SMBD, mAbs, or both ligands (25-500 nM) was monitored using micro-volume stopped flow reaction analyzers SF-61 DX2 (double mixing stopped flow system; Hi-Tech Scientific) or SX-18MV (Applied Photophysics Ltd.) (58). The SI was calculated from the dependences of a fraction of PAB displaced from the active site of uPA by wtPAI-1 (a decrease in PAB fluorescence, % of maximal) on the ratio of moles of wtPAI-1 (or its complexes) reacting with unchanged amount of uPA. The SI was equal to the value of [PAI-1]/[uPA] corresponding to complete displacement of PAB from uPA active site.

Stoichiometry of Inhibition of Proteinases with wtPAI-1 and Its Complexes with mAbs Estimated from SDS-PAGE-
The values of the SI for the reactions of proteinases with wtPAI-1 and its complexes with mAbs and SMBD were also estimated directly from the results of SDS-PAGE (4 -12% gradient gel; Invitrogen) as described previously (59). The SMBD and mAb were taken in 1.5-3.0-fold molar excess over wtPAI-1. The SI was calculated as an average of 3-5 measurements Ϯ S.E.

SMBD Affects Differently the Kinetics of RCL Insertion for
Reactions of NBD P9 PAI-1 with tPA and ␤-Trypsin-The effects of recombinant SMBD and human plasma Vn on the reactions of PAI-1 with ␤-trypsin and tPA were studied in order to determine what step of the reaction (Scheme 1) is affected by the ligands, and to elucidate the mechanism of modulation of PAI-1 activity, as well as to determine contribution of SMBD to the effects of whole Vn. The insertion of RCL, which is the central event of the reaction between PAI-1 and proteinase ( Fig. ␣-Helix F and Modulation of Plasminogen Activator Inhibitor 1 1), was monitored through the increase in the fluorescence emission of the NBD group attached to a cysteine residue in S338C mutant variant of PAI-1 (NBD P9 PAI-1) (54). The binding of SMBD did not affect NBD fluorescence emission under any experimental conditions employed in this study. Thus, an increase in NBD fluorescence emission resulting from the reaction of proteinase with NBD P9 PAI-1 or its complexes with the ligands was employed for the evaluation of the kinetic parameters of RCL insertion and measurements of SI. The observed first-order rate constants (k obs ) of the RCL insertion were measured using stopped flow fluorimetry as described under "Experimental Procedures." The dependence of k obs on the enzyme concentration ( Fig. 2A) demonstrated that binding of the SMBD to NBD P9 PAI-1 results in a decrease in the limiting rate of RCL insertion (k lim ) for the reaction with ␤-trypsin. In contrast, the SMBD did not affect the k lim for the reaction with tPA (Fig. 2B). It was proposed that exosite interactions formed at the proteinase/PAI-1 interface in the Michaelis complex (EI, Scheme 1) affects the k lim (31). As a result, the k lim for tPA (2.8 s Ϫ1 ; Table 1), which forms several exosite interactions with PAI-1, is 30 times lower than that for ␤-trypsin (86.0 s Ϫ1 ; Table  1), which presumably has none. "North to South Pole" translocation of the proteinase (Fig. 1) because of RCL insertion requires breaking all of the exosite interactions. Thus, the RCL insertion for tPA is limited by the formation of the LDA (E-IЈ; Scheme 1) (31). On the other hand, because of lack of exosite interactions between serpin and proteinase, the rate of RCL insertion for the reaction of ␤-trypsin is limited by the acylation of the enzyme (formation of E ϳ IЈ), the step of the mechanism that precedes LDA formation (Scheme 1) (31). Therefore, a decrease in the k lim induced by binding of the SMBD observed for the reaction with ␤-trypsin (Table 1) could reflect changes in the limiting step of the reaction. However, in contrast to ␤-trypsin, SMBD did not affect the limiting rate of RCL insertion for tPA ( Fig. 2A; Table 1). Therefore, the limiting step for RCL insertion for the reaction of tPA with PAI-1 or its complexes with SMBD (Table 1) or Vn (45) is the same (formation of the LDA; E-IЈ; Scheme 1) (31). In contrast to physiological pH 7.4, at a pH of 6.2 SMBD induces a more than 3-fold decrease in the k lim for tPA (from 8.1 Ϯ 0.5 (51) to 2.6 Ϯ 0.2 s Ϫ1 (Table 1)). Thus at low pH, SMBD affects tPA reaction in a manner similar to that observed for uPA at pH 7.4 (a decrease in k lim from 23.0 Ϯ 2.2 (45) to 6.5 Ϯ 0.2 s Ϫ1 ). Therefore, under conditions when the pattern of the exosite interactions between PAI-1 and tPA is disrupted (51), RCL insertion becomes limited by a step of the mechanism, which is different from LDA formation.
The data obtained demonstrated that SMBD does not affect the limiting rate of RCL insertion for interaction between PAI-1 and tPA at a neutral or slightly alkaline pH, when the reaction is limited by formation of LDA. In contrast, for the reactions with ␤-trypsin (limited by enzyme acylation (31)) or with uPA (or tPA at pH 6.2) SMBD induces significant decreases in the k lim . SMBD (Fig. 3) binds far away from the initial docking site of the proteinase (Fig. 1), where enzyme acylation and the formation of LDA occur. Thus, the binding of SMBD most likely contributes to the stabilization of a transient intermediate, which is different from LDA (Scheme 1). Notably, the SMBD-binding site is located in close proximity to the epitopes of MA-55F4C12 and MA-33H1F7, and includes ␣HF (Fig. 3). SMBD interacts with the residues located at the side of ␣HF, which is opposite the epitopes of MA-55F4C12 and MA-33H1F7. Binding of these mAbs results in a decrease in the k lim for both tPA and uPA (59), which could reflect stabilization of the same transient intermediate. Previously we demonstrated additivity in the effects of Vn and MA-55F4C12 (mAb-2) on k lim for reactions of PAI-1 with tPA and uPA (45). However, to the best of our knowledge, there are no data on the effects of anti-␣HF mAbs, Vn, or SMBD on the reaction of PAI-1 with ␤-trypsin, which is limited by acylation of the enzyme (Scheme 1) (31). Thus, at the next step, the effects of SMBD on kinetics of RCL insertion for the reaction of ␤-trypsin, tPA, and uPA with complexes of NBD P9 PAI-1 and MA-55F4C12 and MA-33H1F7 were studied.
SMBD Decreases the Limiting Rate of RCL Insertion for Reactions of Proteinases with NBD P9 PAI-1 Complexed with Anti-␣HF mAb-Both mAbs employed in this study bind with low nanomolar affinity to both active and RCL inserted species of PAI-1 (wild type or NBD P9) (59,61). Whereas the binding of MA-33H1F7 did not affect the fluorescence for NBD P9 PAI-1, MA-55F4C12 induced ϳ20% enhancement in NBD fluorescence emission when bound to RCL inserted species of NBD P9 PAI-1 (59). An increase in NBD fluorescence emission because of the binding of MA-55F4C12 was additive to the effect reporting the RCL insertion (54), i.e. a similar final NBD fluorescence was observed as a result of the reaction of the enzyme with an NBD P9 PAI-1.MA-55F4C12 complex, and for the mAb bound to preformed RCL inserted species of NBD P9 PAI-1. Values of k obs were determined at different enzyme concentrations, and k lim and K m values were calculated from the dependences of k obs on the proteinase concentration (Fig. 4) as described under "Experimental Procedures." Similar to the effects on reactions with uPA and tPA (59), MA-55F4C12 and MA-33H1F7 significantly decreased (7.6 and 6.9 times, respectively) the k lim for the reaction with ␤-trypsin (Table 2). Thus, in contrast to SMBD (Table 1) and Vn (45) anti-␣HF mAbs similarly affect the k lim for tPA and ␤-trypsin, proteinases with different steps, which limit the rate of RCL insertion (31). On the other hand, SMBD, when bound to the PAI-1⅐mAb complex, always induced a decrease in the k lim for all three proteinases ( Table 2). The effect of the SMBD on reactions of uPA and ␤-trypsin with NBD P9 PAI-1.MA-33H1F7 was less pronounced than that observed for MA-55F4C12, which could originate from slight differences in the epitopes of the two mAbs (61). The most significant change in the limiting rate of RCL insertion was observed for the reactions of PAI-1⅐mAb complexes with ␤-trypsin (Figs. 1 and 4), where binding of MA-55F4C12 and SMBD resulted in more than a 100-fold decrease in the k lim (from 86.0 to 0.72 s Ϫ1 ) ( Table 2).
To determine the contribution of the exosite interactions formed by deprotonated histidine residues (45), k lim and K m values for the reaction of tPA with NBD P9 PAI-1⅐mAb complexes were determined at pH 6.2 in the presence of SMBD. The values of k lim at pH 6.2 were several times higher than those observed at pH 7.4 (Table 2), indicating that formation of the exosite interactions at the PAI-1/tPA interface affects the k lim value for the reaction with binary (either SMBD or mAb is bound to PAI-1) and ternary (both SMBD and mAb are bound to PAI-1) complexes. These results support the hypothesis that SMBD and mAb, when bound to PAI-1 at both sides of ␣HF (Fig. 3), affect the step of PAI-1 mechanism, which follows dissociation of the acyl-enzyme from the initial docking site, and which results in the formation of LDA (Scheme 1). Moreover, the binding of MA-55F4C12 and MA-33H1F7 to ␣HF of PAI-1 (Fig. 3) results in a decrease in the k lim and a re-direction of the reaction with tPA and uPA from the inhibitory to the substrate pathway (Scheme 1) (45,58). Thus, studies of the effects of SMBD on the SI for the reactions of PAI-1 and its complexes with anti-␣HF mAbs with ␤-trypsin and tPA were carried out in order to determine a step of the PAI-1 mechanism, which is affected by ligands interacting with ␣HF.
SMBD Potentiates Neutralization of PAI-1 by aHF mAbs in Reactions with Proteinase-The effects of SMBD on the SI for the reactions of ␤-trypsin, tPA, and uPA with PAI-1 and their complexes with mAbs were determined as described under "Experimental Procedures." Recombinant SMBD alone did not affect the partition between substrate and inhibitory pathways  Table 1.   Table 2.

␣-Helix F and Modulation of Plasminogen Activator Inhibitor 1
for the reactions of PAI-1 with any of the proteinases (Fig. 5). Therefore, although the interaction of SMBD with its binding site (Fig. 3) affects the kinetics of RCL insertion for ␤-trypsin and uPA, it does not affect the distribution between the inhibitory and substrate pathways (Scheme 1). The values of SI, measured for the reactions of proteinases with PAI-1⅐mAb complexes with and without SMBD are shown in the Table 3. In contrast to free PAI-1, the SI observed for PAI-1⅐mAb complexes increased considerably in the presence of SMBD for all three enzymes ( Fig. 5; Table 3). Therefore, the binding of SMBD not only affects the kinetics of RCL insertion for the interaction of PAI-1⅐mAb complexes with proteinases but also potentiates PAI-1 neutralization by anti-␣HF mAbs through the re-direction of the PAI-1 reaction from the inhibitory to the substrate pathway.
These results demonstrate that the stabilization of a transient intermediate of the PAI-1 reaction by the different ligands interacting with ␣HF through a decreasing in the limiting rate of RCL insertion could result in both an increase in SI (mAbs) and unchanged (SMBD) distribution between the substrate and inhibitory reactions. On the other hand, the ligands demonstrated additivity in both increasing the SI and in the effects on k lim (Tables 1-3). Thus, the intermediate, which is stabilized by ligands interacting with ␣HF, could be targeted for neutralization of the PAI-1 through the intermolecular mechanisms.
Disturbing the exosite interactions through either mutations at the RCL of PAI-1 (31,50) or through a decrease in pH (51) does not significantly affect the stoichiometry of inhibition for the reaction between PAI-1 and tPA. To determine whether or not exosite interactions between PAI-1 and proteinase contribute to changes in the SI for binary (PAI-1⅐mAb) and ternary (PAI-1⅐mAb⅐SMBD) complexes, a mutant variant of PAI-1 with glutamate to alanine mutation in positions P4Ј and P5Ј (E350A/ E351A) was studied.

Contribution of Exosite Interactions at the PAI-1/tPA Interface to the Effects of SMBD on SI for PAI-1⅐mAB
Complexes-To evaluate the contribution of the exosite interactions to the outcome of the reaction of PAI-1⅐mAb complexes with proteinases, the values of the SI for the reactions of NBD P9 E350A/ E351A PAI-1 with tPA were measured and compared with the SI observed for NBD P9 PAI-1 ( Table 4). Mutations of P4Ј and P5Ј glutamates (positions 350 and 351, respectively) did not significantly affect the SI for the reaction of tPA with PAI-1⅐mAb complexes (Table 4). Therefore, the effect of anti-␣HF mAbs on the outcome of the PAI-1 reaction with tPA does not depend significantly on the pattern of the exosite interactions at the PAI-1/proteinase interface. There was no significant difference between the SI values observed for wt PAI-1 and NBD P9 PAI-1 (Tables 3 and 4), demonstrating that the Ser to Cys mutation at the P9 position of RCL and the following attachment of the NBD group did not interfere with the effects of mAbs and SMBD on the PAI-1 mechanism, i.e. the conclusions drawn from the results of the experiments carried out with NBD P9 PAI-1 could be expanded to wtPAI-1.

DISCUSSION
The binding of SMBD to PAI-1 does not affect the SI for the reaction with any of the three proteinases and induces a significant decrease in the k lim for uPA and ␤-trypsin (Table  1). Such an effect could indicate the stabilization of the Michaelis complex (E⅐I; Scheme 1 and Fig. 1) through a mechanism similar to that observed for the formation of exosite interactions. However, in contrast to uPA and ␤-trypsin, SMBD does not affect the k lim for tPA, which has the highest number of exosite interactions with PAI-1 (Fig.  2B). Thus, there is no additivity in the effects on the k lim between exosite interactions formed at PAI-1/tPA interface in the Michaelis complex and the effects of SMBD, which binds to ␣-helices F and E and strand 1 of ␤-sheet A (53) (Fig.  3). Because the rate of RCL insertion for the reactions of PAI-1 with ␤-trypsin and tPA is limited by different steps of the mechanism (enzyme acylation and LDA formation, respectively (31)), thses data could indicate that SMBD affects a step of the reaction different from the formation of LDA (Scheme 1). Thus, the RCL insertion for the reaction of tPA with the PAI-1⅐SMBD complex is limited by the same step of the mechanism (formation of LDA E ϳ IЈ; Scheme 1) (31) as the reaction with free PAI-1. LDA is formed as a result of breaking the exosite interactions at the PAI-1/tPA interface and the dissociation of the primed part of the cleaved RCL from the active site of tPA (31). Disruption of these exosite interactions because of a decrease in pH results in an increase in both the k lim and K m values (51). As expected, the binding of SMBD to PAI-1 at pH 6.2 induces a decrease in the k lim for the reaction with tPA in a manner similar to that observed for uPA (Table 1), indicating that the rate of RCL insertion becomes limited by a step of the mechanism, which follows LDA formation. Results obtained in this and other studies demonstrate that the stabilization/destabilization of the LDA (Scheme 1) through the formation/disturbance of the exosite interactions, while affecting k lim , does not change the distribution between the substrate and inhibitory pathways of the PAI-1 reaction (Scheme 1). Indeed only small (if any) changes in the SI were observed for the reactions of proteinases with mutant variants of PAI-1, with mutations localized at the serpin/proteinase interface (positions P1Ј (31) and P5Ј (50)), as well as for the reaction between PAI-1 and tPA at low pH (51). Because SMBD and whole Vn affect the reactions with proteinases similarly, formation of the exosite-like interactions between enzyme interacting with PAI-1 at the initial docking site (Fig. 1) and 72-kDa Vn or 5-kDa SMBD bound at the opposite end of PAI-1 molecule (Fig. 3) is unlikely. Thus, a decrease in k lim induced by Vn and SMBD probably reflects stabilization of either LDA (a decrease in k i (and k s ); Scheme 1) or another intermediate, which is formed after LDA. The binding of MA-55F4C12 and MA-33H1F7 to their epitopes located at the side of ␣HF, which is opposite to the SMBD-binding site (Fig. 3), delays RCL insertion for all three proteinases, and obstructs the serpin mechanism, re-directing the reaction from the inhibitory to the substrate pathway (Tables 2-4) (45,59). Because anti-␣HF mAbs affect the reaction of both tPA and ␤-trypsin in a similar manner, one could suggest that binding of these mAbs results in a change of the limiting step for RCL insertion for both enzymes because of stabilization of a transient intermediate, which is favorable for the substrate pathway. The close proximity between binding sites of mAbs and SMBD (Fig. 3) provides the structural basis for additivity in the effects of both ligands on the PAI-1 reaction, especially if RCL insertion requires reversible displacement of ␣HF (62,63). Although residues Arg 131 , Ile 135 , and Asp 138 of ␣HF are in close contact with SMBD (53), residues Glu 128 -Val-Glu-Arg 131 and Lys 154 contribute to the epitopes of MA-55F4C12 and MA-33H7F1 (61). Indeed, the binding of SMBD results in a further significant increase in the fraction of the substrate reaction (Tables 3 and 4) and a decrease in the k lim (Table 2). Therefore, the binding of both mAb and SMBD promotes greater restriction of the flexibility of ␣HF and ␤-sheet A and results in further stabilization of the intermediate and in an increase in the fraction of the substrate reaction. It has been shown that several ligands such as two mAbs (59,64), mAb-(Fab) and Vn (44,45,49), or even three mAbs (65) not only bind to the same PAI-1 molecule but also affect additively the kinetics of RCL insertion and the stoichiometry of the reaction between PAI-1 and the enzyme. Thus, based on the results of this study, and data obtained previously (45), additivity in the effects of SMBD and Fab or single chain Fv fragment (66) of antibody is anticipated.  Table 3.

TABLE 3
Effects of recombinant SMBD on the SI for the reactions of ␤-trypsin with wild type and NBD P9 PAI-1, tPA, uPA with wtPAI-1, and its complexes with MA-55F4C12 and MA-33H1F7 SI represents the number of moles of PAI-1 or its binary and ternary complexes with SMBD and the mAb required for inactivation of 1 mol of proteinase. PAI-1 with ligands by proteinase as described previously (45,59). c This was calculated from a decrease in fluorescence emission of uPA/PAB due to displacement of PAB with PAI-1 resulting from the reaction with wtPAI-1 and as described under "Experimental Procedures."

␣-Helix F and Modulation of Plasminogen Activator Inhibitor 1
Disruption of the exosite interactions because of Glu to Ala mutations at the positions P4Ј and P5Ј of RCL, which results in an increase in k lim (50), did not affect the SI for the reactions of tPA with complexes of PAI-1 with mAb, SMBD (Vn), or both ligands (Table 4). Thus, formation of the Michaelis complex, the acylation of the enzyme, and breaking of exosite interactions, resulting in LDA (Scheme 1), which occur at the initial docking site (Fig. 1), probably do not affect the partitioning between the inhibitory and the substrate reaction. As a result, exosite interactions as well as LDA are unlikely targets for intermolecular re-direction of the serpin reaction to the substrate pathway. Therefore, the most effective approaches to the neutralization of PAI-1 through intermolecular mechanisms aiming the initial docking site would be using of ligands, which compete with proteinases for PAI-1. An example of successful use of mAbs, which block formation of the Michaelis complex between PAI-1 and tPA or uPA, was described elsewhere (58,65,67).
The smallest of the three proteinases, ␤-trypsin, demonstrated the lowest SI and highest k lim for PAI-1⅐mAb complexes (Tables 2 and 3). The correlation between a decrease in the size (M r ) of the enzyme and a decrease in SI and an increase in k lim supports the complete insertion of the RCL and translocation of the proteinase to the opposite pole of the PAI-1 molecule rather than the reversible dissociation of the mAb. Such a mechanism was proposed for anti-␣-helix F mAb CLB-2C8 (67), which binds to both the active and RCL-inserted forms of PAI-1 with sub-nanomolar affinity (65). Low k off values (10 Ϫ3 -10 Ϫ4 s Ϫ1 (59,61,65)) make the dissociation of anti-␣-helix F mAb during the course of RCL insertion unlikely. Indeed, the results of surface plasmon resonance studies of the interaction of PAI-1 bound to immobilized CLB-2C8 with tPA have shown fast binding and dissociation (k off Ͼ 0.01 s Ϫ1 ) of tPA, whereas no dissociation of PAI-1 from the mAb was reported (67). On the other hand, the epitope of the mAb could be reversibly perturbed as a result of tremendous conformational changes accompanying RCL insertion (32, 36, 68 -70) and possible movements of ␣-helix F (62,63). To the best of our knowledge there is no molecular model of a PAI-1⅐mAb complex that could provide an insight into the possible mechanism of RCL insertion. Models of both active PAI-1 (71,72), and its complex with SMBD ( Fig. 3) (53) were obtained from studies of a "stable variant" of PAI-1, which contains K154T stabilizing mutation (73). Unfortunately, lysine 154 is a part of epitopes of anti-␣helix F mAbs (MA-55F4C12, MA-33H1F7 (61), and mAb-2 (44)), and its mutation in the stable variant of PAI-1 affects the interaction with the mAb (44).
In contrast to anti-␣-helix F mAbs, the affinity of Vn (and SMBD) to the active and RCL-inserted PAI-1 differs up to 3 orders of magnitude (74). Therefore, it is unlikely that SMBD is still bound to the final inhibitory complex or to the cleaved PAI-1 under conditions used in this study. On the other hand, because SMBD (Vn) affects both the k lim and SI for the reaction of PAI-1⅐mAb complexes with proteinases, dissociation most likely occurs during the very late steps of RCL insertion.
To explain the effects of SMBD (L1) and mAbs (L2) on the reactions of proteinases (E) with PAI-1 (I), we propose a transient ␣HF-intermediate (E-(IЈ*L1L2); Scheme 2), which is stabilized when mAb, SMBD, or both ligands are bound to ␣HF. E-(IЈ*L1L2) forms at the step of the reaction following formation of LDA (E-(IЈL1L2)) and prior to completing RCL insertion as a strand 4 to the ␤-sheet A of PAI-1 (Fig. 1). The limiting step of RCL insertion because of the reaction of tPA with the PAI-1⅐SMBD complex depends on a pattern of exosite interactions at the tPA/PAI-1 interface. As a result, at a pH of 7.4 and 6.2 the rate of the RCL insertion because of the reaction of tPA with PAI-1⅐SMBD complex is limited by the formation of different intermediates (Scheme 2) E-(IЈL1L2) and E-(IЈ*L1L2), respectively. Stabilization of ␣HF-intermediate by mAbs and SMBD (Vn), which affects RCL insertion, results in an increase in k s /k i (Scheme 2) and redirection of the reaction toward the substrate pathway. The localization of mAbs and SMBD-binding sites, together with the known importance of ␣HF in the serpin mechanism (63,75), supports the restriction of movements of ␣HF and ␤-sheet A during the RCL insertion as an origin of additivity in the effects of SMBD (Vn) and mAbs on the kinetics and outcome of PAI-1/proteinase reactions. Thus, the stabilization of ␣HF-intermediate(s) via intermolecular SCHEME 2

TABLE 4
Effects of recombinant SMBD and human plasma Vn on the SI for the reaction of tPA with NBD P9 PAI-1, NBD P9 E350A/E351A PAI-1, and their complexes with MA-55F4C12 and MA-33H1F7 SI, representing the number of moles of PAI-1 or its binary and ternary complexes with SMBD and mAb required for inactivation of 1 mol of proteinase was determined by titration of fixed amount of complexes of NBD P9 PAI-1 variants with ligands by proteinase as described previously (45,59

␣-Helix F and Modulation of Plasminogen Activator Inhibitor 1
mechanisms resulting in complete re-direction of the PAI-1/proteinase reaction to the substrate pathway could be considered as a strategy for PAI-1 neutralization. Because the conformational changes of PAI-1, induced by RCL insertion, result in a dramatic decrease in affinity of Vn to PAI-1 (74), SMBD and its analogs, which selectively recognize the active conformation of PAI-1, could be employed for design of PAI-1 inhibitors (71). Thus, the ␣HF-intermediate (Scheme 2) is a convenient target for in vivo neutralization of PAI-1 through the use of rationally designed multivalent inhibitors composed of ligands interacting with mAb epitopes and the SMBD-binding site of PAI-1. On the other hand, in contrast to the effects on interaction of PAI-1 with uPA and ␤-trypsin, Vn induces a significant increase in the rate of the reaction of PAI-1 with activated protein C (46) or thrombin (43,47,48). Therefore, further studies are necessary to determine whether or not different conformations of the ␣HF intermediate could contribute to the substrate and the inhibitory branches of the PAI-1 mechanism. Finally, the results of this study also demonstrate that recombinant SMBD affects the reaction of tPA and uPA with PAI-1 and its complexes with anti-␣HF mAbs in a manner similar to that for whole Vn isolated from human plasma (45). Nevertheless, the values of the SI for a whole Vn usually were higher and changes in the k lim were greater than those observed for SMBD, indicating a possible contribution of interactions of PAI-1 or proteinase with other parts of the Vn molecule. However, the data reported here and in our previous study (45) clearly demonstrate that the conformations of both recombinant SMBD and SMBD as a part of whole human plasma Vn (produced by Promega) are specific for active PAI-1 and are likely the same, and agree with the recent results on the pattern of disulfide bonds in the active conformation of SMBD (76).