WO1995030438A2 - Methods and compositions to enhance endogenous fibrinolytic activity - Google Patents

Abstract

Thrombus formation and accretion are inhibited by administering substances which decrease the accumulation of type 1 plasminogen activator inhibitor at sites of vascular injury. It has been found that PAI-1 accumulates as part of a ternary structure composed of vitronectin, PAI-1, and fibrillar protein selected from the group consisting of vimentin and fibrin. Binary complexes of PAI-1 and vitronectin in circulation can enter clot and/or injured endothelial cells and bind to the fibrillar protein. Presence of the PAI-1 inhibits endogenous plasminogen activators, allowing the accumulation of fibrin and clot. The reduction of PAI-1 accumulation thus allows endogenous and administered fibrinolytic substances to inhibit fibrin formation and in turn limit the production and/or accretion of thrombus and clot.

Description

METHODS AND COMPOSITIONS TO ENHANCE ENDOGENOUS FIBRINOLYTIC ACTIVITY

The present application is a continυation-in-part of application Serial No. 08/240,377, filed on May 10, 1994, the full disclosure of which is incorporated herein by reference.

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates generally to methods and compositions for the treatment of cardiovascular disease. More particularly, the present invention relates to novel methods and compositions for enhancing fibrinolytic activity by inhibiting the accumulation of type 1 plasminogen activator inhibitor (PAI-1) at sites of vascular injury and subsequent thrombus or clot formation. Thrombus formation is a pathological manifestation of clotting in blood vessels. The clotting cascade is a complex biological process which results in the formation of a clot or thrombus composed of platelets and fibrin. In arteries, thrombus formation that occurs on atherosclerotic plagues can accelerate plaque growth, leading to partial or total occlusion of an affected blood vessel. Clots can also form in the venous system when a vein is damaged by trauma or surgery. Such pathological clotting is a primary cause of cardiovascular diseases, including unstable angina, acute yocardial infarction (heart attack) , cerebral vascular accidents (stroke), pulmonary embolism, deep vein thrombosis, arterial thrombosis, and the like. These diseases are a major cause of disability and mortality in Western societies. Atherosclerosis is characterized by a gradual deposit of substances, such as clotting proteins, cellular debris, cholesterol, fats, calcium and the like, and cells, such as smooth muscle cells and mononuclear cells, into the walls of arteries, resulting in the formation of atherosclerotic plaque. As the plaque enlarges, there is a progressive narrowing of the artery which can restrict blood flow to muscles and vital organs. Such restricted blood flow can be exacerbated when a plaque is ruptured by increased pressure caused by blood passing through the narrowed arterial lumen. Such rupture can injure the underlying arterial wall and activate the clotting cascade which results in the deposition of clot or thrombus at the site of injury. Clots or thrombi which consist of platelets and red blood cells entrapped in a fibrin matrix can cause total blockage of the arterial lumen. The deposition of fibrin at the site of arterial injury is the primary step in clot or thrombus formation. Thrombus or clot formation also occurs in veins at injury sites. Damage to the vein stimulates blood clotting and the deposition of fibrin.

Thrombus or clot within the vasculature is normally broken down by the fibrinolytic system. The fibrinolytic system relies on the release of plasminogen activators, most notably tissue plasminogen activator (tPA) from cells lining the blood cells. tPA converts plasminogen into a clot- digesting enzyme referred to as plasmin. Plasmin slowly degrades the clot or thrombus, thereby restoring vessel patency and blood flow. The extent of fibrinolytic activity, however, is itself regulated by the presence of endogenous plasminogen activator inhibitors. PAI-1 is the major inhibitor of tPA. Urokinase (uPA) is the other naturally occurring plasminogen activator. PAI-1 accumulates at sites of vascular injury where fibrin deposits. Because PAI-1 locally inhibits tPA and uPA, the generation of plasmin is reduced. As a result, fibrinolysis of clots that form in damaged blood vessels can be significantly diminished.

Thrombus formation and growth can be inhibited by either of two basic approaches. The first is to prevent clotting with anticoagulants and other thrombus-inhibiting drugs. Many anticoagulants, such as heparin, can be employed, but such drugs are not fully effective, have serious side effects, and require careful monitoring of the patient. Newer thrombus-inhibiting drugs, such as hirulog, may be safer but have not yet proven to be more effective.

The second approach for the treatment of clots relies on the administration of pharmacologic concentrations of plasminogen activators, such as tPA, streptokinase, and urokinase. While the use of such plasminogen activators has proven to be of great value, these drugs are not suitable for all patients or for all thrombus-related conditions. Moreover, in some instances, plasminogen activators can have serious side effects.

For these reasons, it would be desirable to identify novel methods and compositions for enhancing naturally occurring fibrinolytic activity in patients suffering from thrombus-related conditions, or patients who are predisposed to acquiring thrombus-related conditions (e.g., after orthopedic surgery) . It would be particularly desirable to develop methods and compositions which are capable of inhibiting the accumulation of PAI-1 at sites of arterial or venous injury where fibrin is deposited. Inhibiting the accumulation of PAI-1 at these sites would limit PAI-1- ediated inactivation of. plasminogen activators, and thus would allow the natural fibrinolytic process to clear the fibrin.

2. Description of the Background Art

Antibodies to vitronectin have been shown to inhibit binding of PAI-1 to vitronectin immobilized on reaction wells. Seiffert et al. (1994) J. BIOL. CHEM. 269:2659-2660; and Preissner et al. (1990) J. BIOL. CHEM. 265:18490-18498. The co- localization of plasminogen activator inhibitor-1 (PAI-1) with vimentin in necrotic rat hepatocytes and evidence that such localization may be mediated by vitronectin are described by Podor et al. (1992) ANN. N.Y. ACAD. SCI. 667:173-177. Specific binding of PAI-1 and vitronectin to the extracellular matrix of TGF-/3-activated bovine aortic endothelial cells is described in Podor and Loskutoff (1992) ANN. N.Y. ACAD. SCI. 667:46-49. The interaction of PAI-1 with vitronectin is described in Declerck et al. (1988) J. BIOL. CHEM. 263:15454- 15461 and Seiffert et al. (1990) J. CELL BIOL. 111:1283-1291. Dawson and Henney (1992) ATHEROSCLEROSIS 95:105-117, is a review article discussing the role of PAI-1 in thrombotic disease. A heparin-binding form of vitronectin that forms non-covalently associated vitronectin multimers is described in Hess et al. "Multimeric Vitronectin: Structure and Function,," in Biology of Vitronectives and their Receptors, Preissner et al., eds., Elsevier Science Publishers (1993), pages 21-29. U.S. Patent No. 5,321,127, describes a platelet glycoprotein lb receptor fragment having antiplatelet and an ithro botic activity useful for blocking platelet adhesion. Grabarek et al. (1981) describes troponin fragments which formed complexes with ATP- ase inhibitory subunit. A synthetic vitronectin fragment which modulates the activity of PAI-1 to reduce excessive fibrinolysis is described in EP 589181 Shohet et al. (1994)

THROM. HAEMOST. 71:124-128 and Paoni et al. (1993) THROM. HAEMOST. 70:307-312 described PAI-1 resistant forms of tPA. Madison et al. (1990) PROC. NATL. ACAD. SCI USA 87:3530-3533 and (1989) NATURE 339:721 describe the regions of tPA which interact with PAI-1.

SUMMARY OF THE INVENTION The present invention comprises methods and compositions for inhibiting the accumulation of type 1 plasminogen activator inhibitor (PAI-1) in a ternary complex composed of a vitronectin component, a PAI-1 component, and a fibrillar protein component selected from the group consisting of intracellular vimentin and fibrin present as the major component in clot. According to the method of the present invention, at least one substance that inhibits binding between vitronectin and either PAI-1 or the fibrillar protein component, or which inhibit binding of vitronectin subunits to each other to reduce formation of a more functionally active multimeric conformation of vitronectin, is exposed to an aqueous environment which includes the ternary complex and/or each individual component of the complex. In a preferred aspect, the aqueous environment may be a blood vessel, typically the inner wall of an arterial or venous lumen, where fibrin dissolution is enhanced by reducing the accumulation of PAI-1 in a region of pre-existing clot. Compositions that enhance fibrinolysis according to the present invention will also be useful in damaged tissues which are susceptible to persistent extravascular fibrin deposition at an injury site which can lead to pathological fibrosis in blood vessels (e.g., after angioplasty) and in other tissues, including heart, lung, liver, brain, and the like. Compositions that enhance fibrinolysis according to the present invention will further be useful as prophylactic agents which may prevent or reduce the initial incorporation of PAI-I and vitronectin in a newly forming clots. Thus, the compositions may be introduced to an aqueous environment, typically the blood plasma or interstitial fluid prior to clotting, and interfere with vitronectin and PAI-1 incorporation into clots after injury to a blood vessel or tissue, or may interfere with incorporation of vitronectin and PAI-1 into newly forming fibrin during the process of fibrin accretion onto a pre-existing clot. The present invention further comprises pharmaceutical compositions comprising or consisting essentially of a substance which inhibits binding between the vitronectin and PAI-1, or vitronectin and the fibrillar protein component, or vitronectin monomeric subunits to each other which results in formation of the more functionally active multimeric conform of vitronectin, wherein the substance is present in a pharmaceutically acceptable carrier. Suitable substances which bind to PAI-1 and inhibit complex formation are selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on PAI-1, and antibody and antibody fragments that alter the conformation of PAI-1, thus reducing or eliminating its ability to form the ternary complex. Suitable substances which bind to vitronectin to inhibit self-association and multimerization are selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to vitronectin multimerization sites, and vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to vitronectin and alter the conformation of vitronectin in order to inhibit self-association. Suitable substances which bind to vitronectin to inhibit complex formation are selected from the group consisting of PAI-1 fragments, PAI-1 analogs, antibody and antibody fragments that bind to a PAI-1-binding site on vitronectin, vimentin fragments, vimentin analogs, antibodies and antibody fragments that bind to a vimentin-binding site on vitronectin, and antibody and antibody fragments that alter the conformation of vitronectin in order to inhibit formation of the complex. Suitable substances which bind to vimentin¹ to inhibit complex formation are selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin- binding site on vimentin, and antibody and antibody fragments that alter the conformation of vimentin in order to inhibit complex formation. Suitable substances which bind to fibrin to inhibit complex formation are selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to the vitronectin- binding site on fibrin, and antibody and antibody fragments that alter the conformation of fibrin in order to inhibit complex formation.

The present invention still further provides an inhibitor of PAI-1 comprising a hybrid molecule having a first moiety which binds to PAI-1 and inhibits binding of PAI-1 to vitronectin and a second moiety which binds to a reactive center on PAI-1 and prevents it from interacting with plasminogen activators. The first and second moieties are spaced-apart by distance sufficient to permit simultaneous binding of both moieties to their respective binding sites on PAI-1, usually being in the range from 5 A to 50 A. The first moiety may be selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on PAI-1, and antibody and antibody fragments that bind and alter the conformation of PAI-1. The second moiety may be selected from the group consisting of a plasminogen activator fragment, a plasminogen activator analog, and an antibody or antibody fragment which binds to the PAI-1 reactive center, and the like. The present invention still further provides an inhibitor of PAI-1 comprising a hybrid molecule having a first moiety which binds to vitronectin but does not interfere with the binding of PAI-1 to vitronectin and a second moiety which binds to the PAI-1 to prevent it from binding to plasminogen activators. The first and second moieties are spaced-apart by distance sufficient to permit simultaneous binding of both moieties to their respective binding sites on vitronectin and PAI-1, usually in the range from 5 A to 50 A. The first moiety may be selected from a group consisting of vimentin fragments, vimentin analogs, antibody or antibody fragments that bind to the vimentin binding site on vitronectin, and fibrin fragments, fibrin analogs and antibody or antibody fragments that bind to the fibrin binding site on vitronectin. The second moiety may be selected from the group consisting of a plasminogen activator fragment, a plasminogen activator analog, and a antibody or antibody fragment which binds to the PAI-1 reactive center, or binds to moieties on PAI-1 that inhibit interactions with plasminogen activators, and the like.

The present invention still further comprises methods for screening test compounds to determine whether a test compound can inhibit the accumulation of PAI-1 in the ternary complex described above. The screening method comprises exposing the test compound to an aqueous environment including the ternary complex or each component which comprise the ternary complex. Test compounds having the desired inhibition activity are identified by their ability to

(a) inhibit the incorporation of vitronectin and/or PAI-1 into the ternary complex, (b) compete with vitronectin and/or PAI-1 for incorporation into the ternary complex, and/or (c) displace PAI-1 and/or vitronectin bound within the ternary complex. Usually, the test system will comprise vimentin or fibrin immobilized on a solid phase within the aqueous environment, where the immobilized vimentin or fibrin is exposed to vitronectin, PAI-1, and the test compound within the aqueous environment. Optionally, the vitronectin and PAI -1 may be present in the aqueous environment as a preformed conjugate. Detection is usually accomplished by measuring a detectable label bound to PAI-l or vitronectin where incorporation of the label into the ternary complex and/or failure to decrease the inhibition activity of PAI-l within the ternary complex is an indication that the test compound has not inhibited accumulation of the PAI-l. The methods and compositions of the present invention will rely on the use of novel substances having the activities and binding affinities described above. The present invention specifically excludes the use of known thrombolytic agents which might have an effect on formation of maintenance of the ternary complex, specifically excluding glycosaminoglycans, such as heparin, heparin fragments, and der atan sulfate.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1A-1C illustrate the mechanism of endothelial cell injury which results in formation of a ternary vimentin- vitronectin-PAI-1 complex and which is mediated or controlled by the methods and compositions of the present invention.

Vimentin, an insoluble cytoskeleton protein, is exposed when the cell is damaged as illustrated in Fig. IA. Vitronectin (Vn) and/or complexes of Vn and PAI-l enter into the damaged cell, as illustrated Fig. IB, and bind to the vimentin, as illustrated in Fig. 1C.

Fig. ID illustrates the mechanism of ternary complex formation in the region of fibrin in clot. Fig. 2 is a schematic illustration of the ternary vimentin-vitronectin-PAI-1 complex whose formation is mediated by the methods and compositions of the present invention.

Fig. 3 is a schematic illustration of the structure of vitronectin showing the relative positions of the cell adhesion (RGD) domain, the heparin-binding domain which contains the plasmin/plasminogen-binding domain, and three

PAI-1-binding domains. Fig. 4 is a schematic illustration of the structure of latent and active PAI-l. The reactive center is masked in latent PAI-l and vitronectin cannot bind to this conformation. The reactive center is exposed in active PAI-l which binds to vitronectin. The binding of vitronectin to PAI-l stabilizes PAI-l in its active conformation and enhances its specificity toward proteinases, particularly thrombin.

Fig. 5 is a schematic illustration showing the relative positions of the heparin-binding domain, the vitronectin-binding domain, and the reactive center on PAI-l.

Fig. 6A is a schematic illustration of a bivalent molecule comprising a first moiety Ml that binds to a vitronectin-binding domain on PAI-l and a second moiety, M2 that binds to the reactive center on PAI-l. The moieties are held together by a linking region L.

Fig. 6B is a schematic illustration showing the relative positions of a bivalent molecule comprising a first moiety Ml that binds to a vimentin- or fibrin-binding domain on vitronectin and a second moiety, M2 that binds to the reactive center of PAI-l. The moieties are linked together by a linking region L.

Fig. 7 is a graph illustrating the binding of biotinylated PAI-l to immobilized vimentin in the presence of varying concentrations of vitronectin (Vn) . Fig. 8 is a graph illustrating the inhibition of tPA activity by PAI-l alone, vitronectin alone, and an equimolar combination of vitronectin and PAI-l, all in the presence of immobilized vimentin.

Fig. 9 is a graph illustrating vimentin-binding inhibition of biotinylated vitronectin in the presence of polyclonal anti-vimentin antibody.

Fig. 10 is a graph illustrating inhibition of biotinylated vitronectin binding to vimentin in the presence of polyclonal anti-vitronectin antibody. Fig. 11 is a figure illustrating the displacement of biotinylated vitronectin complexed to vimentin in the presence of F(ab)₂ fragments of polyclonal anti-vitronectin antibody. Fig. 12 is a diagram showing insertion of 1.1 kb 3¹- vimentin cDNA into the EcoRI restriction site of the multiple cloning region of pMAL-C2 vector for recombinant expression of the carboxy-terminus of human vimentin. Fig. 13 is a diagram showing insertion of 500 bp

5'-vimentin cDNA into the Hindlll/Xhol restriction sites of the multiple cloning region of pET-21(b) vector for recombinant expression of the amino-terminus of human vimentin. Fig. 14 is a representation of human vimentin protein (465 amino acids) , showing the regions encoded by the different constructs. IA, IB, 2A and 2B denote four α-helical domains in the rod (310 aa) , and flanking nonhelical head (102 aa) and tail (54 aa) regions. TH indicates the thrombin cleavage site at aa 78-79.

Fig. 15 shows the amino-terminal vimentin (VIM) sequence which binds vitronectin. Purified vimentin 133 mer was subjected to thrombin cleavage and analyzed by SDS-PAGE, followed by ligand blotting with biotinylated Vn and epitope mapping with various antibodies directed against specific sequences in the amino-terminus of human vimentin. Vn binding activity was detected in the 1 through 78 amino acid sequence of the amino-terminus and later defined to a sequence between residues 51 and 78. Figs. 16A and 16B show the binding of ¹²⁵I-urea denatured (Fig. 16A) and ¹²⁵I-native vitronectin to a recombinant N-terminal fragment of human vimentin. Microtiter plate wells were coated with purified VIM 133 and then incubated with 5 μg/ml of ¹²⁵I-labelled multimeric (urea- treated) (Panel A) or monomeric (Panel B) Vn in the presence

(open triangles) or absence (closed circles) of 20-fold excess Vn (multimeric or monomeric, respectively) . At various times, the unbound proteins were washed and the amount bound quantitated by a gamma counter. The specifically bound proportion of Vn was calculated by subtracting the (cpm bound in the presence of excess cold Vn) from (total cpm bound in the absence of cold Vn) . Fig. 17 is a Scatchard analysis of multimeric ¹²⁵I-labelled vitronectin binding to a recombinant amino- terminal vimentin fragment. Microtiter place wells were coated with purified VIM 133 and incubated with various doses of iodinated multimeric Vn for 1 hr and the dissociation constant was calculated using the Table Curve software program (Jandel Scientific) . Affinity (calculated) of multimeric Vn for the vimentin coated wells was approximately 0.5 nM and of monomeric Vn for vimentin coated wells was undetectable. Fig. 18 is an analysis of plasma proteins pre- and post-clotting. Citrated human plasma was re-calcified, allowed to clot for 1 hr at 37°C, and aliquots of the starting plasma and resulting serum were analyzed by ELISA for total antigen levels of Vn, plasminogen, fibrinogen and albumin. Fig. 19 is an analysis of the incorporation of vitronectin into fibrin clots forming in solution. Various concentrations of purified human Vn (mixture of monomeric and multimeric forms) were incubated with purified human fibrinogen in the presence of thrombin and calcium. After 1 hr at 37°C the clot was pelleted by centrifugation and the concentration of Vn antigen in the pre- and post-clotted supernatants determined by ELISA. The specifically bound Vn was calculated by subtracting the mOD/min reading of the post- clot supernatants from the mOD/min reading of the pore-clot supernatants.

Fig. 20 is an analysis of the selective depletion of monomeric vitronectin during fibrin clot formation. Aliquots of the supernatants from the samples in Figure 19 were subjected to native polyacrylamide electrophoresis (PAGE) transferred to nitrocellulose and the Vn detected by immunoblotting. Note that in the supernatants prior to the addition of thrombin, the Vn was almost evenly distributed as monomeric and multimeric forms. However, after clotting, the remaining supernatants were selectively depleted of the monomeric conformations of Vn.

Fig. 21 is an analysis of the diffusion of ¹²⁵I-vitronectin from purified clots in the presence or absence of 2M NaCl. Purified hanging fibrin clots were formed in the presence ¹²⁵I-Vn or ¹²⁵I-labelled thrombin or PPACK ARG 93-thrombin and the rate of diffusion of the radiolabelled proteins was determined. Note that the normal thrombin and PPACK-treated thrombin both show a marked increase in the rate of diffusion in the presence of buffer containing 2.0 M NaCl. In contrast, only a modest amount of the ¹²⁵I-Vn was displaced by 2.0 M NaCl (compare closed triangles to open triangles).

Fig. 22A is an analysis of the binding of vitronectin to purified fibrin clots. Microtiter well plates were coated with purified human fibrinogen treated with thrombin and calcium to form clots and allowed to air dry at 4°C overnight. The clots were then treated with various concentrations of purified biotinylated Vn in the presence or absence of 20-fold excess unlabelled human Vn. The amount of biotinylated Vn bound was determined by incubating wells with streptavidin-conjugated alkaline phosphatase and color substrates as above. The open circles represent the net specific binding of Vn with a half-maximal binding concentration of Vn binding to fibrin calculated to be approximately 20 nM.

Fig. 22B is a graph showing the time course of binding ¹²⁵I-Vn (denatured) to fibrin in the presence or absence of excess cold.

Fig. 22C is a graph showing the time course of binding ¹²⁵I-Vn (native) in the presence or absence of excess cold.

Fig. 23 is an analysis of vitronectin mAB-mediated displacement of vitronectin from fibrin. Fibrin-coated microtiter wells were incubated with saturating concentrations of biotinylated Vn in the presence or absence of various purified monoclonal antibodies directed against different epitopes on Vn. mAB 1244 is directed against an epitope in the amino-terminal half of the protein which does not interfere with the binding of Vn to PAI-l. Monoclonal antibody mAB 153 is directed against binding site in the somatomedian B domain in the amino-terminus (residues 1-40) of Vn. Monoclonal antibody mAB 8E6 is directed against an epitope in the amino-terminal half (residues 1-256) of vitronectin and inhibits PAI-l binding. NM-IgG is preim une, normal mouse IgG. Both mABs 153 and 8E6 are directed against anionic and cationic regions respectively, on vitronectin which are believed to be involved in vitronectin multimerization.

Fig. 24 is an analysis of the displacement of vitronectin bound to fibrin by purified vimentin peptides. The saturating concentrations of biotinylated-Vn were incubated in the presence (+PAI-1) or the absence (-PAI-1) of PAI-l and then added to fibrin-coated microtiter wells. After 1 hr the wells were washed and then incubated for 1 hr with various concentrations of the vimentin 133 mer (VM 133) , synthetic vimentin 28 (VM 28) , or with the synthetic scrambled peptide (SCR) . The boμnd biotinylated-Vn was then determined using streptavidin alkaline phosphatase and color substrate. Note that the VIM 28 mer peptide, and to a lesser extent the VIM 133 mer, were very effective in displacing Vn bound to fibrin and particularly Vn bound to PAI-l.

Fig. 25A is an analysis of the displacement of BAI-1 associated w.|th vitronectin and fibrin by the vimentin 133-mer peptide. Purified fibrin-coated microtiter wells were coated with saturating concentrations of purified Vn and PAI-l pre¬ formed complexes, or equi olar purified PAI-l for 1 hr at 37°C. After washing, the plates were then incubated for 1 hr in buffer containing various concentrations of the VIM 133 mer peptide. The residual functionally active PAI-l was directly detected using ¹²⁵I-MAI 12 IgG. This monoclonal antibody is selective for active PAI-l and is directed against an epitope within the carboxy-terminus of PAI-l. Note that only in the presence of Vn, there is a vimentin 133 mer dose-dependent decrease in the amount of active PAI-l present on the fibrin clot.

Fig. 25B illustrates dose-dependent enhancement of fibrinolysis by displacement of the vitronectin/PAI-1 complexes with the vimentin 133 mer peptide. Purified ¹²⁵ι- labelled fibrin coated wells were incubated with saturating concentrations of purified Vn and PAI-l, or Vn alone for 1 h at 37°C. After washing, the wells were incubated with various concentrations of VIM 133 mer for 1 h. After washing, the wells were then treated with purified tPA and plasminogen and the release of soluble ¹²⁵I-labelled fibrin degradation peptides into the buffer was quantitated as an index of net fibrinolytic activity was suppressed in the wells pretreated with Vn and PAI-l relative to the wells treated with Vn alone and that preincubation with the VIM 133 mer increased the rate of lysis in a dose-dependent manner to a level comparable to that of wells treated with Vn alone. Fig. 26A is an analysis of the displacement of fibrin-bound vitronectin by recombinant vimentin 28-mer peptide in the presence or absence of human plasma. Purified fibrin-coated microtiter wells were pre-treated with saturating concentrations of biotinylated-Vn. After washing, the clots were incubated with various concentrations of the synthetic VIM 28-mer peptide in the presence or absence of 20% citrated human plasma and the bound biotinylated-Vn was detected with streptavidin-conjugated alkaline phosphatase as above. Note that the presence of 20% serum alone displaced approximately 25% and the addition of the vimentin 28 mer decreased the bound Vn by a further 25%.

Fig. 26B illustrates the doe-dependent prothrombolytic effects of the vimentin 133 mer peptide (VM- 133) on whole platelet-rich plasma clot lysis in 96-well microtiter plates.

Fig. 27 is a schematic representation of proposed interactions between different conformations of Vn with PAI-l, fibrin, and vimentin.

Fig. 28 is a graph illustrating the inhibition of PAI-l activity in the presence of plasminogen, tPA, or plasmin chromogenic substrate and a synthetic 14-mer peptide comprising amino acids 296-308 of the tPA light chain (NH-Ile Phe Ala Lys His Arg Arg Ser Pro Gly Gly Arg Phe Leu-COOH [SEQ ID No. :2] . Fig. 29A is a graph illustrating inhibition of PAI-l binding to Vn in the presence of mABs 153 and 8E6. Microtiter plate wells were coated overnight with urea-treated Vn (1 μg/ml, 0.05 ml). After washing, the wells were blocked (3% BSA in PBS, pH 7.4) and incubated with PAI-l (2 μg/ml) in the presence of increasing concentrations of either mABs 153 (open circles) or 8E6 (closed circles) . Bound PAI-l was detected with biotin-conjugated, affinity-purified rabbit anti-human PAI-l IgG and ¹²⁵I-streptavidin. The amount of

PAI-l bound is expressed as a percentage of the PAI-l bound to Vn in the absence of the competing mABs.

Fig. 29B is a graph illustrating the reduction of PAI-l activity in the presence of mABs 153 and 8E6. Microtiter plate wells were coated overnight with urea-treated Vn (1 μg/ml, 0.05 ml). After washing, the wells were blocked (3% BSA in PBS, pH 7.4) and incubated with increasing concentrations of PAI-l in the presence of 5 μg/ml of mAB 153 (closed circles) , mAB 8E6 (open triangles) or buffer alone (open circles) . After washing, the wells were incubated with 160 IU/ml urokinase for 30 min at 23°C and the remaining uPA activity determined following the addition of the chromogenic substrate S-2444 (final concentration, * 1 mM) and measuring "the change in absorbance at 405 mm. The concentration of functionally active PAI-l bound to Vn was calculated using a titration curve of PAI 1 with uPA in solution.

Fig. 30 is a graph showing the effects of anti-Vn mABs on PAI-l binding to native Vn. Microtiter plate wells were coated overnight with native Vn (2 μg/ml, 0.05 ml). After washing, the wells were blocked (3%, BSA in PBS, pH 7.4) and incubated with PAI-l (2 μg/ml) in the presence of increasing concentrations of either mABs 153 (open circles) or 8E6 (closed circles) . Bound PAI-l was detected with biotin- conjugated, affinity-purified rabbit anti-human PAI-l IgG and ¹²⁵I-streptavidin. The amount of PAI-l bound is expressed as a percentage of the PAI-l bound to Vn in the absence of the competing mABs.

Fig. 31A and 3IB showing the availability of different PAI-l binding sites on Vn. In Fig. 31A, microtiter wells were coated overnight with either mAB 153 (open symbols) or 8E6 (closed symbols) . The wells were blocked and then incubated for 1 hr with excess (2.0 μg/ml) urea-treated Vn (triangles) and native Vn (circles) . After washing, the bound Vn was detected using biotin-conjugated, affinity-purified rabbit anti-human Vn IgG and streptavidin-conjugated alkaline- phosphatase/pNPP substrate. The specific binding of Vn was determined by subtracting the change in absorbance at 405 nm in control wells coated with BSA alone. In Fig. 3IB, microtiter wells were coated overnight with either mAB 153 (open symbols) or 8E6 (closed symbols) . After blocking, the wells were incubated for 1 hr with urea-treated (circles) and native Vn (triangles) . After washing, the wells were incubated with increasing concentrations of PAI-l for 45 min and the bound PAI-l detected with biotin-conjugated, affinity- purified rabbit anti-human PAI-l IgG and ¹²⁵I-streptavidin.

Fig. 32 is a graph showing the availability of PAI-l binding sites following binding of urea-treated Vn with immobilized mABs 153 and 8E6. Microtiter wells were coated overnight with either mAB 153 (open circles) or mAB 8E6 (closed circles). After blocking (3% BSA in PBS, pH 7.4) and incubated with urea-treated Vn (2 μg/ml) and after washing the unbound vitronectin the wells were then incubated with PAI-l (2μg/nl) in the presence of increasing concentrations of the opposite mAB. The bound PAI-l detected with biotin- conjugated, affinity-purified rabbit anti-human PAI-l IgG and ¹²⁵I-streptavidin. The amount of PAI-l bound is expressed as a percentage of the PAI-l bound to immobilized mAB 153 or 8E6 in the absence of the second competing mAB.

Figs. 33A and 33B are graphs showing the stoichiometry of PAI-l binding to urea-treated and native Vn. In Fig. 33A, microtiter wells were coated overnight with affinity-purified, rabbit polyclonal anti-human Vn IgG (2 μg/ml) . After blocking, the wells were incubated for

45 min at 37°C with aliquots of PAI-l (1.0 μg/ml) which had been preincubated for l hr with increasing concentrations of either urea-treated (open circles) or native (closed circles) Vn. After washing,the wells were incubated with ¹²⁵I-labelled MAI-12 IgG (a mAB directed against active PAI-l) and the counts bound determined with a gamma counter. The amount of radioactivity bound to each well was determined using a gamma counter. In Fig. 33B, microtiter wells were coated overnight with affinity-purified rabbit anti-human Vn IgG (2 μg/ml) . After blocking, the wells were incubated with increasing concentrations of either urea-treated (open circles) or native (closed circles) Vn for 45 min at 37°C. After washing the wells were incubated for 1 hr with biotin-conjugated, affinity-purified rabbit anti-human PAI-l IgG and ¹²⁵ι- streptavidin conjugated to alkaline phosphatase as described above.

DESCRIPTION OF THE PREFERRED EMBODIMENT

The present invention provides methods and compositions for inhibiting the accumulation of type 1 plasminogen activator inhibitor (PAI-l) in a ternary complex which is present in preexisting or newly forming clot in a blood vessel or which is formed as a result of cell damage, particularly to cells which line the blood vessels, e.g., as the result of angioplasty. The accumulation of PAI-l at such sites of clot accumulation and vascular injury is problematic because PAI-l is a potent inhibitor of plasminogen activators, such as tissue plasminogen activators (tPA) and urokinase, which are responsible for the conversion of plasminogen to plasmin, where plasmin in turn is responsible for the degradation of fibrin in a thrombus or clot. The present invention is based, at least in part, on recognition by the inventors herein that the localized inhibition of plasminogen activators caused by the presence of PAI-l at a site of vascular injury is a significant factor in the loss of plasminogen activator activity which results in pathological persistence of fibrin deposited at the site. Fibrin is the major component of clot or thrombus, and the present invention relies on inhibition of PAI-l accumulation at sites of preexisting clot and/or vascular injury to restore the activity of endogenous and (if present) exogenous plasminogen activators, thus increasing the degradation (fibrinolysis) and/or reducing the formation of thrombus or clot.

In addition to such in vivo therapeutic uses, the methods and compositions of the present invention are useful in in vitro systems. In particular, the present invention provides in vitro assays and test systems for determining the ability of a test compound to inhibit the formation of the ternary complex and/or binding between particular components of the ternary complex. Screening assays are run by exposing a test compound to an aqueous environment including at least two components of the ternary complex and preferably the entire complex. The complex will usually be immobilized on a solid phase, usually a plastic surface such as a microtiter well. Optionally, isolated tissue or cells from blood vessels may be introduced to and/or cultured on the solid phase, and formation of the ternary complex induced. The test may then be run by measuring the ability of the test compound to inhibit formation of the complex in the aqueous cellular environment. Such assays and test systems are useful for identifying test substances which are suitable for further testing as drugs for treating patients suffering from thrombus-related conditions.

Referring now to Figs. 1A-1C, a damaged cell (Fig. IA) exposes intracellular components to plasma proteins and other substances present in the blood. In particular, vimentin, an insoluble intermediate filament cytoskeleton component, is exposed in the damaged cells and binds to plasma constituents such as complement, fibrinogen, or immunoglobulins. Of particular interest to the present invention, exposed intracellular vimentin binds vitronectin

(Vn) , a plasma protein that may be complexed with PAI-l. The binding of vitronectin to PAI-l stabilizes PAI-l in its active form (Fig. 4) . Under normal conditions, only a small amount of PAI-l is produced by endothelial cells. Relatively large amounts of PAI-l, however, accumulate within damaged endothelial cells by entering through openings in the cellular membrane (Fig. IB) and binding to the exposed vimentin (Fig. 1C) . The excess PAI-l serves to inhibit local fibrin degradation which in turn results in fibrin accumulation and persistence. As fibrin is the major constituent of a thrombus, PAI-l acts to prevent or inhibit clot breakdown, and the clot can cause narrowing of blood vessel lumen. Of further particular interest to the present invention, it has _* been determined that the binding of the vimentin, vitronectin, and PAI-l results in a ternary complex where the vitronectin binds as an intermediate to both the vimentin and the PAI-l, as illustrated in Fig. 2. The binding sites of the vimentin and the PAI-l are distinct, as illustrated.

Referring now to Fig. ID, a ternary complex comprising PAI-l and vitronectin also forms by binding to fibrin located in preexisting clot in blood vessels and elsewhere. Surprisingly, it has been fo nd that PAI-l bound to vitronectin and fibrin within preexisting clot has a substantially greater activity than unbound PAI-l, as discussed in detail in the Experimental section hereinafter. Thus, the release of PAI-l from clot and/or inhibition of accumulation of PAI-l in clot will significantly decrease the inhibition of endogenous and administered fibrinolytic agents, such as tPA, urokinase, and the like.

Vitronectin is a 78 kD adhesive glycoprotein which is produced in the liver and released into the blood circulation. Vitronectin is also known as complement "S- protein". Vitronectin has at least four distinct binding domains of interest, including a heparin-binding domain whi'ch contains the plasmin/plasminogen-binding domain, and three PAI-1-binding domains. The structure of vitronectin is shown in Fig. 3, including three amino acid sequences (including amino acids 1-40, 115-121, and 348-379) which are presently believed to contribute to PAI-1-binding, and is described in Seiffert et al. (1994), supra.; Preissner et al. (1990), supra . } and Mimuro et al. (1993) BIOCHEM. 32:2314-2320. Of particular interest to the present invention, binding between vitronectin and PAI-l can be inhibited by antibodies, such as mAB 153 and mAB 8E6 directed against the amino-terminal and carboxy-terminal binding region, respectively, as demonstrated in the Experimental section hereinafter. Other characteristics of vitronectin are well described in the literature. Vitronectin is present in plasma at concentrations of 200-400 μg/ml and as both a monomeric (>95%) and in various multimer (<5%) forms from dimers, trimers up to 18 mer of molecular weights up approximately 1200 kDa. PAI-l circulates with vitronectin multimers (dimer-tetrameres) . It is believed that the multimeric form is that form of vitronectin associated with matices and solid surfaces and is believed to be the most functionally active form with regards to binding to antithrombin Ill-thrombin complexes, complement, cell surfaces and PAI-l. Multimerization is believed to be mediated by electrostatic interactions between adjacent anionic amino-termini and cationic carboxy-terminal heparin binding domain of overlapping monomeric subunits and these multimers can be further stabilized by disulfide exchange. See, e .g. Biology of Vitronectins and their Receptors , Preissner et al., eds., Excerpta Medica, Amsterdam 1993. The cDNA sequence of vitronectin is set forth in Jenne and Stanley (1985) EMBO J. 4:3153-3157. Latent (inactive) PAI-l does not bind to vitronectin. As illustrated in Fig. 4, binding to vitronectin stabilizes the structure of active PAI-l which has the reactive center exposed near the carboxy-terminus of the molecule. The reactive center of PAI-l in the stabilized PAI- 1-vitronectin complex is able to bind to the active sites on tPA and uPA to inhibit fibrinolytic activity, as discussed above. Blocking or inhibiting binding between PAI-l and vitronectin can thus reduce or eliminate the ability of PAI-l to inhibit natural fibrinolytic activity. Vimentin is an intracellular intermediate filaments

(IFs) protein which is the primary constituent of the three- dimensional in situ cytoskeleton of many cell types, including vascular endothelial cells platelets, and smooth muscle cells. A primary therapeutic use of the methods and compositions of the present invention is in the inhibition of PAI-l at vascular sites where vimentin will be the principal exposed IF. Pathological fibrin accretion can occur in other tissues where other IF's, such as desmins, keratins, and the like, may be exposed by cellular damage and which may provide a site for formation of the ternary complex. For convenience, the remaining disclosure will be directed specifically at vascular therapy and reference will be made to vimentin. It will be appreciated, however, that the invention also includes treatment of other tissue types, such as heart, lung, liver, brain, skin, and the like, where other IF's may be responsible for binding of the ternary complex the ternary complex.

PAI-l is an approximately 50 kd (379 amino acid) serine protease inhibitor of the serpin gene family. PAI-l is produced by a variety of cells, particularly endothelial and smooth muscle cells, and is a fast-acting inhibitor of plasminogen activators, including tissue plasminogen activator (tPA) and urokinase (uPA) . PAI-l has a binding domain for heparin and vitronectin comprising particular residues among amino acids 55-123 which are exposed on the surface of the folded molecule, and a reactive center which inhibits plasminogen activator activity, as illustrated in Fig. 5. Methods for purifying PAI-l from conditioned media of particular cell lines are disclosed in U.S. Patent

Nos. 5,112,955 and 5,073,626, the disclosures of which are incorporated herein by reference.

As used herein, "fibrin" refers to polymerized fibrin monomer which accumulates in blood clot as a result of the endogenous clotting cascade. Briefly, fibrinogen hydrolyzes in the presence of thrombin into fibrin and fibrinopeptide fragments. Initially, fibrin forms soft clots which can be readily dispersed. Over time, thrombin activates fibrin-stabilizing factor and the fibrin is cross-linked, resulting in hardened clot, often referred to as plaque.

Methods and compositions of the present invention will act by reducing the presence and/or activity of PAI-l in clot (as well as at sites of vascular injury) , which in turn will increase the fibrinolytic activity of endogenous and administered fibrinolytic agents, thus reducing the presence and accumulation of fibrin in both soft and hardened clots.

The present invention relies on methods and compositions which inhibit the accumulation of PAI-l, typically at vascular injury sites where intracellular vimentin has been exposed as a result of the rupture of endothelial cells, platelets, and/or smooth muscle cells. Such inhibition is achieved by exposing a vascular lumen or other aqueous cellular environment to a substance which blocks or competes with binding between at least two of the three components of the ternary complex described above. For example, binding between PAI-l and vitronectin can be blocked, inhibited, or displaced by introducing a substance to the vascular or other aqueous cellular environment which binds to PAI-l or vitronectin in a manner which blocks or sterically inhibits binding between the two components. The substance may bind to PAI-l, being selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to vitronectin-binding site on PAI-l, and antibody and antibody fragments that alter the conformation of PAI-l. Alternatively, the substance may bind to vitronectin, being selected from the group consisting of PAI-l fragments, PAI-l analogs, antibody and antibody fragments that bind to a PAI-1-binding site on vitronectin, and antibody and antibody fragments that alter the conformation of vitronectin.

In either case, the substance'will have a binding affinity toward PAI-l or vitronectin which is sufficient to prevent initial binding or disrupt established binding in formed ternary complexes. Generally, a binding affinity of at least 10⁵ M^-1 will be sufficient to prevent initial binding, while a higher affinity of at least 10⁷ M^-1 will be sufficient to disrupt established equilibrium binding in a formed ternary complex. Such substances will preferably have binding affinities of at least 10⁷ M^"1, more preferably being at least 10⁸ M^"1, and still more preferably being at least 10⁹ M^"1, with higher affinity substances being capable of both blocking initial binding and disrupting established equilibrium binding in formed ternary complexes.

Substances useful in the present invention may also be selected to block initial binding between the binary complex of PAI-l and vitronectin to the intracellular vimentin or to disrupt established binding between the binary complex and the vimentin in a formed ternary complex. For example, the substance may bind to the vimentin-binding site on vitronectin, being selected from the group consisting of vimentin fragments, vimentin analogs, antibody and antibody fragments that bind to a vimentin-binding site on vitronectin, and antibody and antibody fragments that alter the conformation of vitronectin in such a way that binding to vimentin is inhibited or disrupted. Alternatively, the substance may bind to the vitronectin-binding site on vimentin, being selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on vimentin, and antibody and antibody fragments that alter the conformation of vimentin in such a way that blocks or disrupts binding to vitronectin. Alternatively, the substance may bind to the fibrin binding site on vitronectin, being selected from a group consisting of fibrin fragments, fibrin analogs, antibodies or antibody fragments which bind to the fibrin binding site on vitronectin which may alter its conformation in such a way to block or disrupt the binding interaction of vitronectin with vimentin.

The binding affinities of such vitronectin-binding and vimentin-binding substances will generally be at least 10⁵ M^"1, preferably being at least 10⁷ M^"1, more preferably being at least 10⁸ M^"1, and most preferably being at least 10⁹ M^_1. As described above, binding substances having lower affinities, generally from 10⁵ M^_1 to 10⁷ M^-1 will be suitable for blocking initial binding between vimentin and the binary complex of vitronectin and PAI-l. Disrupting established equilibrium binding between the binary complex and vimentin in formed ternary complexes will generally require higher affinities, usually greater than 10⁷ M^-1. Substances having higher binding affinities usually above 10⁷ M^"1 are preferable since they will generally be able to both block initial binding between the binary complex and vimentin as well as to disrupt equilibrium binding between the binary complex and vimentin in the formed complexes.

Substances useful in the present invention may further be selected to block initial binding between the binary complex PAI-l and vitronectin to fibrin present in existing clot, thrombus, plaque, and the like, or to disrupt established binding between the binary complex and fibrin in a formed ternary complex. For example, a substance may bind to the fibrin-binding site on vitronectin, being selected from the group consisting of vimentin fragments, vimentin analogs, fibrin fragments, fibrin analogs, antibody and antibody fragments that bind to a fibrin-binding site on vitronectin, and antibody and antibody fragments that alter the conformation of vitronectin in such a way that binding to fibrin is inhibited or disrupted. Alternatively, the substance may bind to the vimentin binding site on vitronectin, being selected from a group consisting of vimentin fragments, vimentin analogs, antibodies or antibody fragments which bind to the vimentin binding site on vitronectin, and other substances which may alter the conformation of vitronectin and block or disrupt its binding to fibrin. Alternatively, the substance may bind to the vitronectin-binding site on fibrin, being selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on fibrin, and antibody and antibody fragments that alter the conformation of fibrin in such a way that blocks or disrupts binding of the binary complex of vitronectin and PAI-l.

The binding affinities of such vitronectin-binding and fibrin-binding substances will generally be at least 10⁵ M^-1, preferably being at least 10⁷ M^"1, more preferably being at least 10⁸ M^_1, and most preferably being at least 10⁹ M^"1. As described above, binding substances having lower affinities, generally from 10⁵ M^"1 to 10⁷ M^-1 will be suitable for blocking initial binding between fibrin and the binary complex of vitronectin and PAI-l. Disrupting established equilibrium binding between the binary complex and fibrin in a formed ternary complex will generally require binding substances having affinities, usually greater that 10⁷ M^-1. Substances having higher binding affinities, usually above 10⁷ M^"1, are preferably since they will generally be able to both block initial binding between the binary complex and fibrin as well as to disrupt equilibrium binding between the binary complex and fibrin in formed complexes. Fragments and analogs of vitronectin, PAI-l, fibrin and vimentin will be selected to have a suitable binding affinity, as described above, so that they can block or displace binding between native components of the ternary complex. It will of course be necessary that the fragments and analogs not be so closely related to the native molecules that the fragments and analogs will actively participate in the fibrinolytic system. Thus, the fragments and analogs of the present invention should display binding specificity to the target component which is generally comparable to or greater than that of the native component, but which lack at least some of the other native activity or activities which are responsible for the inhibition of plasminogen activators. For example, suitable vitronectin, PAI-l, fibrin and vimentin fragments may be polypeptides which comprise or consist essentially of the target-binding region or domain of the native molecule (i.e., the region or domain which binds to the target molecule) , but which lack some or all of the other regions of the molecule. Exemplary vitronectin fragments binding to PAI-l may comprise portions or all of fragments including amino acids 1-40 and 348-379 which are brought together by folding of the molecule, as illustrated in Fig. 4. Fragments comprising amino acids 115-121 of the putative binding region (Fig. 3) presently does not appear to play a significant role in PAI-l binding and are less likely suitable for use in the present invention. These fragments may be joined by the natural molecular linking region or by any other linking sequence which maintains the binding fragments in proper orientation for binding to the vitronectin-binding region of PAI-l with the requisite affinity.

Exemplary vitronectin fragments binding to vimentin may comprise amino acids 40-348 (except RGD) and 40-379 to the C-terminus. Exemplary PAI-l fragments binding to vitronectin may comprise portions or all of fragments including amino acids 55-123, as illustrated in Fig. 5. Exemplary vimentin fragments binding to vitronectin may comprise all or portions of the 78 amino acid amino-terminal thrombin cleavage fragment of vimentin. In all cases, it will be appreciated that the fragments should include a sufficient number of amino acids so that the three-dimensional structure and binding properties of the intact molecule are retained sufficiently to preserve the desired binding characteristics. It will be appreciated that once the existence of the PAI-1-containing ternary complex of the present invention is known, the identification of substances which inhibit and/or dissociate such complexes is well within the abilities of one skilled in the art of drug screening an design. It is the recognition of the mechanism of PAI-l localization at sites of vascular injury and within preexisting clot, however, that provides the basis and motivation for such testing and identification. Testing can be performed generally by immobilizing at least two and preferably all three of the components of the ternary complex on a solid phase and exposing test compounds, usually selected from the groups defined above, for the ability to disrupt, displace, or otherwise dissociate the complex. Alternatively, testing could be performed by immobilizing or solubilizing any one or two of the components and screening for the ability of test compounds to inhibit formation of the ternary complex in the presence of the remaining components of the complex. Specific testing techniques are described in detail in the Experimental section hereinafter. An exemplary vimentin fragment which has been found to disrupt the binding of the binary complex of vitronectin and PAI-l to fibrin in clots comprises amino acid residues 51 through 78 of vimentin, according to the numbering of SEQ ID No.:l. Thus it is apparent that the vimentin-binding site on vitronectin is the same as, sufficiently close to, or of greater affinity than the fibrin-binding site to block binding of vitronectin to PAI-l.

Such peptidic analogs can be prepared by conventional solid phase synthesis or recombinant techniques, as are well described in the patent and scientific literature. Solid-phase synthesis techniques are based on the sequential addition of amino acid residues to a growing chain on a solid- phase substrate, as first described by Merrifield (1963) J. AM. CHEM. SOC. 85:2149-2156. Commercial systems for automated solid-phase synthesis are now widely available from suppliers, such as Applied BioSystems, Inc., Foster City, California. Recombinant polypeptide production techniques are widely described in the technical and scientific literature. See, for example, MOLECULAR CLONING: A LABORATORY MANUAL, Sambrook et al., Eds., Cold Spring Harbor Press, Cold Spring Harbor, New York (1989) Vol. 1-3.

Analogs of the binding sites of vitronectin, PAI-l, fibrin, and vimentin may also be prepared as small molecule mimetics. Small molecule mimetics are non-peptidic molecules, usually having molecular weight below 2 kD, more usually below 1 kd, and frequently in the range from 300 D to 1 kD, with structures which may be derived using techniques well known to those working in the area of drug design. Such techniques include, but are not limited to, self-consistent field (SCF) analysis, configuration interaction (CI) analysis, and normal mode dynamics computer programs, all of which are now readily available. See, Rein et al., Computer-Assisted Modeling of Receptor-Ligand Interactions, Alan Liss, New York, (1989) and Navia and Marcko, use of Structural Information in Drug Design in CURRENT OPINION AND INSTRUCTIONAL BIOLOGY, vol. 2, No. 2, pgs. 202-210 (1992) . The preparation of compounds identified by these and other techniques will depend on their structure and other characteristics, and such preparation may normally be achieved by standard by chemical synthesis methods as described in available texts, such as Furniss et al., VOGEL'S TEXT BOOK OF PRACTICAL ORGANIC CHEMISTRY, John Wiley and Sons, New York, (1992) and Larock, COMPREHENSIVE ORGANIC TRANSFORMATIONS, VCH Publishers, Inc., New York, (1989).

Antibodies and antibody fragments which bind to the specified binding sites (epitopes) on vitronectin, PAI-l, fibrin and vimentin, as described above, may be prepared by conventional techniques, typically using each of these molecules as an immunogen. Specific techniques for preparing polyclonal and monoclonal antibodies are well described in the scientific and patent literature. See, for example, ANTIBODIES: A LABORATORY MANUAL, Harlow and Lane, Eds., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, (1988) . Once an antibody is prepared, its sequence may be determined and peptidic analogs and mimetic compounds prepared based on such a sequence using the techniques described above. Conversely, preparation of antibodies to vitronectin, PAI-l vimentin or fibrin may allow for isolation of specific or randomly selected cDNA sequences from the total DNA of the immunized animals spleen cells. Once the sequence is determined to an antibody which interacts with any of these components of the ternary complexes, recombinant antibody fragments or chimerics may be synthesized and "humanized" as described in the scientific patent literature.

Compositions according to the present invention will also include multivalent, usually bivalent, PAI-l inhibitors comprising a hybrid molecule having a first moiety which binds to PAI-l and inhibits binding of PAI-l to vitronectin and a second moiety which binds to the reactive center on PAI-l (Fig. 6A) and inhibits plasminogen activator inhibition activity. The first moiety may comprise any of the PAI-1-' binding substances described above, or substances (such as VIM 133 described in the Experimental section) which bind the Vn moiety of the Vn-PAI-1 complex (Fig. 6B) , while the second moiety will be selected to bind to, block or sterically hinder the reactive center on PAI-l. Suitable second moieties include the sequences that include the active sites on tPA, urokinase, thrombin, or portions thereof. Also useful would be oligopeptides, as described in JP 5032695, and EP 320 840, which block binding between PAI-l and tPA.

As illustrated in Fig. 6, a bivalent molecule BV according to the present invention comprises a first moiety Ml and a second moiety M2 joined by a linking region L. The linking region or group is selected to provide the necessary covalent bridge between the first binding moiety and the second binding moiety. Frequently, the linking region will be derived from a bifunctional compound having a reactive group at one end which is capable of binding to the first binding moiety and second reactive group at the second end which is capable of binding to the second binding moiety. Alternatively, the linking region may be synthesized together with either the first binding moiety, the second binding moiety, or both, and will include only a single reactive functionality for covalent binding therebetween. The nature of the linking region is not critical, but it should provide sufficient spacing and flexibility between the two binding moieties so that the first binding moiety Ml may bind to the vitronectin-binding site on PAI-l or to the vimentin binding site on vitronectin (Fig. 6B) and the second binding moiety M2 may bind to the reactive center, as illustrated in Figs. 6A and 6B. The length of the linking region will usually be between 5 A and 50 A, preferably being between 5 A and 15 A. The linking region should be resistant to degradation when administered to a patient as part of a therapy as according to the present invention and further should not contribute to nonspecific adhesion of the hybrid molecule, i.e., adhesion or binding to other than the vitronectin binding site and reactive center on PAI-l. It will be appreciated that the hybrid molecules should also display no or minimum immunogenicity so that the drugs they comprise may be administered to a patient over an extended^' period of time without initiating an undesirable immune reaction.

Pharmaceutical compositions according to the present invention comprise a substance which binds to PAI-l, vitronectin, fibrin, or vimentin, with a specificity and affinity as described above, present in a pharmaceutical carrier. Such compositions will be useful to treat or prevent thrombus-related cardiovascular conditions, as described above, and may be administered by themselves or in combination with anticoagulants and/or exogenous fibrinolytic agents, such as tPA, streptokinase, urokinase, and the like. The pharmaceutical composition will find particular use with concurrent intravascular interventions which may cause injury to arterial endothelial and/or smooth muscle cells, such as angioplasty, atherectomy, laser ablation, ultrasound ablation, rotational ablation, and the like. Suitable pharmaceutical compositions will contain a therapeutic amount of the binding substance present in a pharmaceutically acceptable carrier. By a "therapeutic amount" it is meant that sufficient substance will be present in order to inhibit formation of the ternary complex or promote degradation of formed ternary complexes, generally as described above. Typically, substances will be present in a pharmaceutical composition in a concentration of from about 0.01 μg per dose to 10 mg per dose, usually being in the range from 1 μg per dose to 1 mg per dose. Daily dosages may vary widely, depending on the activity of the particular substance, usually being between about 1 μg per kg of body weight to about 5 mg per kg of body weight per day usually being from about 10 μg per kg of body to about 1 mg per kg of body weight per day. The pharmaceutically acceptable carrier can be any compatible, non-toxic substance suitable to deliver the ternary complex- inhibiting or displacing substance to the patient. Sterile water, alcohol, fats, waxes and inert solids, may be used as the carrier. Pharmaceutically acceptable adjuvants, buffering agents, dispersing agents and the like, may also be incorporated into the pharmaceutical compositions. Such compositions will be suitable for oral, nasal, transdermal, pulmonary, and/or parenteral administration, preferable being suitable for parenteral administration, i.e., subcutaneous, intravascular, and intravenous administration. The following examples are offered by way of illustration, not by way of limitation.

Experimental

1. DISRUPTION OF Vm-Vn-PAI-1 TERNARY COMPLEX BY EXPOSURE TO Vn AND αVm ANTIBODIES.

Chemicals. Proteins. Reagents. All chemicals were the highest analytic grade commercially available. Triton X-100, caprylic acid, normal rabbit immunoglobulin (IgG) and isotype- matched, non-specific mouse IgG were obtained from Sigma Chemical Co. (St. Louis, MO) . The monoclonal antibody to the heparin binding domain in vitronectin (clone 8E6) was obtained from Boehringer Mannheim (Montreal, Que.). The monoclonal antibody to the somatomedian B domain of vitronectin (clone 153) was kindly provided by Dr. D. Seiffert. Vimentin isolated from bovine lens tissue was obtained from Boehringer Mannheim (Montreal, Que.). Vitronectin was isolated by heparin-affinity chromatography from human serum as described by Yatohgo, T., et al. (1988) CELL STRUCT. FUNCTION 13:281-292 and by affinity chromotography using affinity purified sheep anti-human vitronectin IgG coupled to an Affigel resin. Human PAI-l purified from the conditioned media of human HT 1080 fibrosarcoma cells and Spectrolyse PI chromogenic substrate was obtained from Biopool Inc. (Burlington, Ont.).

Recombinant human PAI-l purified from E. coli lysates was kindly provided by Drs. David Ginsburg and Daniel Lawrence (University of Michigan, Ann Arbor, MI) (Lawrence, D., et al. (1989) EUR. J. BIOCHEM. l«6:523-533 and Sherman, P.M., et al. (1992) J. BIOL. CHEM. 267(11) :7588-7595) . Purified human glu- plasminogen and fibrinogen was obtained from Enzyme Research Laboratories (South Bend, IN) .

Purified IgG fractions of monoclonal and polyclonal antibodies were prepared by caprylic acid precipitation (Hum, B.A., et al. :1980) METHODS IN ENZYMOLOGY 70:104-142). Antisera to human PAI-l, human Vη and bovine vimentin were raised in rabbits and sheep using Freund's Complete Adjuvant for the initial immunization (75 μg total) and incomplete Freund's adjuvant for subsequent booster injections. The IgG fractions were further purified by affinity chromatography on immobilized PAI-l, Vn or vimentin, respectively. PAI-l and Vn (1 mg) were coupled to a 1 ml bed-volume of cyanogen bromide- activated Sepharose 4B (Pharmacia, Sweden) as described by the manufacturer. Vimentin was coupled to 0.5 ml bed-volume of Affi-Gel 15 (Bio Rad Laboratories, Mississauga, Ont.) according to manufacturer's instructions. Affinity purification of specific IgG fractions was performed by loading respective IgG fractions diluted in PBS, pH 7.4 onto equilibrated ligand affinity columns and washing out the nonspecifically bound IgG with 10 bed-volumes each of PBS, pH 7.4 with 0.5 M sodium chloride (NaCl), PBS, pH 7.4, 0.15 M NaCl and then unbuffered saline (0.15 M NaCl). The specific IgG fractions were eluted with 5 bed-volumes of 0.1 M glycine buffer containing 0.15 M NaCl, pH 2.3 and the fractions (1 ml) collected into 0.2 ml of 0.5 M glycine, 0.75 M NaCl, pH 8.9. Samples were then desalted by concentration and reconstitution (3 cycles) of centrifugation with Centricon 10 microconcentrators (Amicon Inc., Beverly, MA). Protein concentrations were determined by the bicinchoninic acid assay method (Pierce Chemical, Rockford, IL) .

The antibodies to PAI-l, Vn and vimentin were shown to be monospecific by immunoblot analysis of extracts from human and bovine endothelial cells or rat liver extracts.

Generation of F(ab')₂ fragments by pepsin digestion of rabbit IgG in 10 mM sodium citrate buffer, pH 3.5 was performed essentially as described by Mage, M.G. (1980) METHODS IN ENZYMOLOGY 70:142-150. The nature and completeness of digestion was monitored by SDS-polyacrylamide gel electrophoresis and staining of the gels with silver nitrate.

Biotinylation of purified proteins. Purified proteins (IgG, PAI-l, Vn, vimentin and BSA) were biotinylated with amino-hexanoyl-biotin-N-hydroxysuccinimide ester (AHNS; Boehringer Mannheim, Montreal, Que.). Briefly, 100 μg of protein was dialyzed against 0.1 M bicarbonate buffer, pH 8.4 and then incubated with 10 μg AHNS in dimethylformamide (25 mg/ l) for 4 h at room temperature. Samples of biotinylated PAI-l were dialyzed against 4 M guanidine hydrochloride as previously described (Hekman, CM., et al. (1988) ARCH. BIOCHEM. BIOPHYS. 262:199-210) and all the labelling mixtures were dialyzed against several volumes of PBS, pH 7.4 and aliquots were stored at -70°C until used. Protein concentrations were determined by the bicinchoninic acid assay method. The bioactivity of biotinylated proteins (e .g. , biotinylated PAI- 1) was determined by measuring their ability to bind to substrate-coated (e.g., vitronectin or tPA) 96-well plates (Seiffert and Podor (1991) FIBRINOLYSIS 5:225-231). Bound PAI-l or Vn was detected directly with streptavidin-conjugated alkaline phosphatase or indirectly with rabbit anti-PAI-1 or anti-Vn IgG and goat anti-rabbit IgG-conjugated alkaline phosphatase followed by pNPP substrate and quantitation of absorbance at 405 n using a microtiter plate reader (EL 340, Bio-tek Instruments Inc., Highland Park, VT) . Biotinylation did not significantly interfere with the binding of PAI-l to immobilized Vn or to human recombinant tPA (Eli Lilly Co., Indianapolis, IN) . Microtiter plates having 96 wells were coated with vimentin or vitronectin by incubation over night at 4°C with 50 μl of a coating solution. Non-specific binding sites were blocked with bovine serum albumin (BSA) .

The ability of vitronectin to mediate binding between PAI-l and immobilized vimentin was demonstrated by incubating microtiter plates having immobilized vimentin (varying coating concentrations) with 20 nM of biotinylated PAI-l in the presence of increasing concentrations of vitronectin. PAI-l concentrations were determined by the addition of streptavidin-conjugated alkaline phosphatase and pNPP substrate and measuring absorbency at 405 nm. Background binding of PAI-l to BSA-coated wells was subtracted. The results are set forth in Fig. 7 with each curve representing a different vitronectin concentration (o, 250 nM; • 125 nM; V, 25 nM; T, 12.5 nM; D, 2.5 nM; and^' I, buffer alone). It can be seen that increasing concentration of both vitronectin and vimentin result in increased binding of PAI-l.

The inhibition of fluid-phase tPA by PAI-l bound within a ternary complex of vimentin, vitronectin, and PAI-l, was confirmed by incubating wells of a microtiter plate having bound vimentin (100 nM coating concentration) for two hours at 37°C with vitronectin alone (o) PAI-l alone (•) or equimolar concentrations of both vitronectin and PAI-l (V). After washing to remove unbound proteins, the wells were incubated with TBS (pH 7.4) containing 50 nM tPA, 5.7 μM plasminogen, and 1.25 mM Spectrolyse (BioPool) , a chromogenic substrate for 5 minutes at 37°C. Plasmin generation was determined by measuring absorbency at 405 nM. The results are set forth in Fig. 8. The ability of anti-vimentin antibody to block binding between vitronectin and vimentin was confirmed as follows. Microtiter wells were coated with vimentin, as described above. The wells were then incubated for 1 hour at 37°C with varying concentrations of affinity-purified anti- vimentin IgG or pre-immune rabbit IgG. After washing, the wells were incubated with biotinylated vitronectin (2 μg/ml) for 2 hours, and the amount of biotinylated vitronectin determined by measuring the change in absorbency at 405 nM after the addition of streptavidin-conjugated alkaline phosphatase and pNPP substrate. Background binding of vitronectin to BSA-coated wells was subtracted from the results. Referring to Fig. 9, it can be seen that increasing concentrations of the anti-vimentin antibody were able to decrease binding of vitronectin to vimentin in a dose- dependent manner. Binding is expressed as the percentage of vitronectin binding to vimentin (mOD/min) in the presence of the vimentin-specific antibodies compared with binding in the presence of pre-immune rabbit IgG.

The ability of anti-vitronectin antibody to compete with binding of vitronectin to vimentin was confirmed as follows. Microtiter wells were coated with vimentin (lOOnM coating concentration) as described above. Varying concentrations of affinity-purified anti-vitronectin antibody or pre-immune rabbit IgG were pre-incubated for one hour with vitronectin (2 μg/ml) prior to addition of the vitronectin and antibody to the vimentin-coated wells for 2 hours at 37°C. After washing, the amount of biotinylated vitronectin was determined by measuring the change in absorbency at 405 nM after the addition of streptavidin-conjugated alkaline phosphatase and pNPP substrate. Background binding of the vitronectin to the BSA-coated wells was subtracted. Referring to Fig. 10, the decrease of relative binding in the presence of increasing antibody concentrations is illustrated. The binding is expressed as percentage of binding (mOD/min) in the presence of the vitronectin-specific antibody compared with binding in the presence of the pre-immune rabbit IgG. The ability of anti-vitronectin antibody fragments to displace vitronectin from preformed vitronectin-vimentin complexes was confirmed as follows. Microtiter wells were coated with vimentin (lOOnM coating concentration) as described above. The wells were then incubated for 2 hours at 37°C with biotinylated vitronectin (2 μg/ml) . After washing, the wells were incubated for 1 hour at 37°C with varying concentrations of affinity-purified anti-vitronectin IgG F(ab)₂ fragments or pre-immune rabbit IgG F(ab)₂ fragments.

The amount of bound biotinylated vitronectin was determined by measuring the change in absorbency at 405 nM after the addition of streptavidin-conjugated alkaline phosphatase and pNPP substrate. Background binding of biotinylated vitronectin to BSA-coated wells was subtracted. Referring to Fig. 11, it can be seen that the relative binding of vitronectin to vimentin decreases with increasing concentrations of the anti-vitronectin antibody fragment. Relative binding is expressed as percentage of vitronectin binding (mOD/min) in the presence of the anti-vitronectin antibody fragments compared with binding in the presence of the pre-immune rabbit antibody fragments.

2. BINDING INTERACTIONS OF VIMENTIN AND VIMENTIN-DERIVED PEPTIDES WITH VITRONECTIN-PAI-1 COMPLEXES

A. Summary

In vivo and in vitro im unohistochemical studies indicated that Vn and PAI-l accumulate on insoluble tangles of the vimentin-type intermediate filament cytoskeleton in necrotic cells. Moreover, cellular autolysis or cleavage of purified vimentin by thrombin or plasmin results in the generation of soluble, amino-terminus-derived vimentin peptides which form fluid-phase, ternary complexes with Vn- PAI-1 in the conditioned media of cultured bovine aortic endothelium (BAEs) . The interaction of PAI-l with vimentin is mediated by the specific binding interaction between Vn and a 28-residue sequence within the amino-terminal head domain of vimentin within the Vn-PAI-1 complex. Vn binds vimentin in the absence of PAI-l, but in the presence of functionally active PAI-l, there is a 5- to 10-fold increase in the Vn binding to vimentin. Vimentin binds with high-affinity (Ki≡0.5 nM) to the more functionally active multimeric Vn, particularly multimeric Vn-PAI-1 complexes. In contrast, the binding of monomeric Vn to vimentin is extremely low or near absent. Various forms of soluble low molecular weight vimentin-derived peptides have been synthesized and used to competitively disrupt the binding of Vn-PAI-1 complexes to fibrin clots in vitro (see below) . Thus, soluble vimentin peptides selectively bind the multimeric Vn-bound form of PAI-l, and can block or displace Vn-PAI-1 complexes from fibrin.

B. Identification of the Vitronectin Binding Site on

Vimentin.

The effects of various mediators of endothelial cell damage on the distribution of vimentin and its degradation fragments which bind Vn and PAI-l was analyzed. In particular, it was shown that treatment of BAEs with LPS or ionizing radiation results in a dose- and time-dependent increase in cell damage as determined by (1) the specific release of ⁵¹chromium from pre-labelled cells, (2) exposure of the insoluble vimentin cytoskeleton to plasma proteins and (3) degradation of the native 57 kDa vimentin subunit protein release of low molecular weight (LMW) vimentin fragments into the BAE conditioned media. Biosynthetic radiolabelling and immunoprecipitation studies on the conditioned media of LPS- treated cells indicate the presence of complexes between ³⁵S- methionine labelled-PAI-1, LMW (<20 kDa) vimentin fragments, and serum-derived Vn. These observations were supported by Western and ligand blot data indicating the dose-dependent binding of biotinylated-Vn to the vimentin degradation fragments in the conditioned media of LPS-treated cells. Epitope mapping studies indicate that the Vn-binding fragments are not detectable with a monoclonal antibody directed against the carboxy-terminal rod domain (mAB V.9), but are detected by polyclonal antibodies directed against the amino-terminal head domain (obtained from Dr. W. Ip, University of Cincinnati) . To further define the Vn binding site on vimentin, purified bovine vimentin was cleaved using various purified proteinases, including thrombin and plasmin, and the resulting fragments analyzed by SDS-PAGE, Western and ligand blotting. These studies confirmed the fragment of bovine vimentin released from damaged BAEs contains the amino-terminus of vimentin.

Recombinant human vimentin expression systems were developed in order to further define the Vn binding site within the amino-terminal region of vimentin. Total RNA isolated umbilical vein endothelial cells was screened with specific oligonucleotide primers to develop variable length vimentin fragments using reverse transcription-polymerase chain reaction (RT-PCR) . The resulting cDNA fragments were cloned into pBLUESCRIPT and used to transform DH5α E. coli ' cells. Colonies were screened by restriction enzyme digestion and positive clones were then sequenced as follows. First a 1.1 Kb fragment which represents the 3' end of the coding region was subcloned into pMAL-c2 to produce a carboxy terminus fusion protein with maltose binding protein (MBP, M, 40 kDa) (Fig. 12) . Transformed E. coli were induced with IPTG, and screened using Western blots with a polyclonal anti- vimentin IgG and the monoclonal antibody (Mab V.9) directed against the carboxy-terminal "tail" domain of vimentin. Purified MBP-vimentin protein was isolated using amylose resin. This MBP-vimentin fragment was detected with Mab V.9 IgG. Second, to obtain the 5' end of the vimentin cDNA, methylmercury hydroxide was used to denature the RNA, prior to RT-PCR using specific primers. A 460 bp fragment was obtained, which was cloned into pBLUESCRIPT, and sequenced. The plasmid DNA was then digested with EcoRIIXhoI , and subcloned into a modified pFL-1 vector which contained a heart muscle kinase recognition site (HMK) for endogenous ³²p- labelling of the expressed amino-terminus fragment. This construct was digested with HindlllXhoI to release the HMK- 5'-vimentin fragment, which was then cloned into the pET-21(b) vector and then used to transform BL21(DE3) E. coli cells (Fig. 13) . Purified recombinant vimentin amino-terminus protein was isolated using a Ni²⁺-HIS column. Thus, we have produced two overlapping vimentin cDNA constructs which encode for the entire coding sequence (Fig. 14). First, a 1.1 Kb fragment (pMAL-c2-3'-vim) which represents approximately 80% of the vimentin coding sequence from the carboxy-terminal "tail" and central α-helical "rod" domain which does not bind biotinylated-Vn. Second, a 0.5 Kb fragment (pET-21(b)-5'-vim) which codes for the entire amino-terminal "head" domain including a 100 bp region which overlaps with the "rod" domain in pMAL-c2-3'-vim. In contrast to the carboxy-terminal pMAL- c2-3'-vim fragment, the 21 kDa amino-terminal fusion protein binds to biotinylated-Vn and competes for its binding to native VIM. Purified recombinant human vimentin peptide

(VIM 133) was used to produce an affinity purified sheep anti- vimentin IgG. Additional cleavage and epitope mapping studies with VIM 133 peptide narrowed the Vn binding site to a sequence between the (amino-terminal) major protein kinase C and A phosphorylation sites (residues 18-50) and a (carboxy- terminal) thrombin cleavage site at residue 78-79 (Fig. 15) . A 28 amino acid vimentin peptide (N-blocked) comprising residues 51-78 and a control scramble peptide which contains the same amino acids in random order were synthesized (95% purity) . The- sequence of the entire 133 residue human vimentin peptide which encompasses the specific 28 amino acid (Residues 51-78) peptide is shown in SEQ ID No.:l. The sequence begins at the N-terminal methionine of the signal peptide cleavage site and includes residues 1 though 133 (VIM 133) .

C. Scatchard Analysis of ¹²⁵I-Labelled Vitronectin Binding to Vimentin Fragment (VIM 133) For development of a vimentin solid-phase binding assay in microtiter plates, the dose-dependent coating efficiency and coating concentration of purified VIM 133 was determined using the ³²-P-HMK-labelled VIM 133 and by indirect immunologic methods. Optimal coating of VIM 133 was achieved at a coating concentration of 10 μg/ml (>400 ng bound/well at approx. 80% coating eff.) and used throughout the subsequent studies. Human Vn derived from outdated human plasma was purified by affinity chromatography using a resin coupled to affinity-purified sheep anti-human Vn IgG and was iodinated by the Bolton-Hunter method. The predominantly monomeric form of Vn was converted to the predominantly multimeric form by incubation with 6 M urea (urea-treated Vn or denatured Vn) overnight at room temperature, dialyzed, and assessed by reduced and non-reduced SDS-PAGE and native polyacrylamide get electrophoresis. Time course binding experiments indicate that, in contrast to multimeric Vn (Fig. 16A) , the affinity- purified monomeric Vn demonstrated no significant specific binding even after 2 hr incubation with VIM 133 coated surfaces (Fig. 16B) . The monomeric form of ¹²⁵I-labelled Vn has little or no affinity for vimentin. The multimeric form of Vn binds with a calculated association constant of approximately 0.5 nM (Fig. 17). This could be pathophysiologically and pharmacologically relevant if the multimeric form only represents approximately 3-6% of the total plasma Vn pool.

3. VITRONECTIN-DEPENDENT BINDING OF PAI-l TO FIBRIN CLOTS

The relative concentration of Vn in a blood clot is approximately 2- to 3-fold greater than in whole blood. The following tests were run to determine the nature and characteristics of Vn from the plasma and/or platelet pool incorporated into the clot. Immunofluorescence and three- dimensional confocal scanning laser microscopic image analysis of platelet-rich clots indicates that the majority of PAI-l is co-localized with Vn and the vimentin cytoskeleton of lysed platelets or fibrin fragments. However, there are also significant deposits of Vn associated with fibrin strands which is not always co-localized with PAI-l. In an effort to elucidate the nature of the Vn interactions with clots, we have analyzed the interactions of Vn with fibrin in three different assay models, including: (a) whole blood clots; (b) purified hanging fibrin(ogen) clots; and (c) purified fibrin clots in 96-well microtiter plates. The relative concentration of total Vn antigen in citrated human plasma is decreased by approximately 25% following recalcification and pelleting of the clot (Fig. 18) . The decrease of Vn antigen levels in serum corresponds to a 40-50% decrease in plasminogen levels, >95% decrease in fibrinogen levels and <5% decrease in albumin (Fig. 18) . These finding support the hypothesis that Vn, like plasminogen, is immobilized within the clot on specific binding sites which are generated during fibrin formation. To further characterize the nature of binding of fluid-phase Vn to fibrin, various concentrations of unlabelled Vn were incubated with excess fibrinogen (1 mg/ l) in the presence or absence of thrombin/calcium, the fibrin clot pelleted by centrifugation, and the Vn antigen levels in the pre- and post-clotting supernatants determined by an ELISA. The amount of Vn bound to fibrin (i.e., pre-clot [Vn]) - (post-clot [Vn]) indicates dose-dependent binding which saturates at approximately 85 nM with dissociation constant of approximately 40 nM (Fig. 19) . Aliquots of the pre- and post- clotting supernatants were subjected to non-reducing native, polyacrylamide gel electrophoresis followed by immunoblot analysis in order to visualize the monomeric and multimeric forms of Vn. As shown in Fig.- 20, the starting solution of Vn was a mixture of conformations, although the monomeric Vn constituted approximately 30%-50% of the total (Fig. 20, "Pre" Panel) . In contrast, the Vn conforms in the supernatants of clotted samples illustrate that the monomeric Vn conforms are depleted from the clot supernatant and that the excess predominately multimeric conform left unbound to fibrin in the fluid-phase (Fig. 9, "Post" Panel) is reduced. These results suggest that fibrin, in contrast to vimentin binds the monomeric conform of Vn, which represents >95% of the total plasma Vn pool. In order to further characterize the nature of the

Vn binding to fibrin, 100,000 cpm of ²⁵I-labelled Vn was incubated with fibrinogen in a microfuge test tube containing TBS buffer (0.125 ml), a clot was formed by the addition of thrombin and the time-dependent diffusion of ¹²⁵I-labelled Vn from the hanging clot into the buffer containing either normal saline (Fig. 21, solid symbols) or 2.0 M NaCl (Fig. 21, open circles) was determined in aliquots by a gamma counter. In parallel experiments, we evaluated the diffusion of ¹²⁵ι- labelled thrombin (squares) or ¹²⁵I-labelled PPACK thrombin (circles) from clots also incubated in normal or high salt buffer. The ¹²⁵I-Vn diffused out at approximately the same rate as the radiolabelled thrombin forms. However, in contrast to thrombin, incubation in 2.0 M NaCl had no significant effect on Vn diffusion out of clots, with a maximal (5 h) release of approximately 50-60% of the bound ¹²⁵I-Vn as compared to 90% of the radiolabelled thrombins. This novel observation suggests that a large proportion of Vn is bound to fibrin in a manner which does not involve electrostatic interactions but may involve hydrophobic or covalent binding factors which also mediate Vn multimerization.

A microtiter plate assay to detect the binding of Vn and Vn-PAI-1 complexes to fibrin clots via the specific binding of biotinylated Vn to fibrin coated wells is illustrated in Fig. 22A. Saturable binding is observed at 35 nM with a calculated half-maximal binding concentration of 15-20 nM. These numbers are generally in agreement with those obtained by t e hanging clot assay (above) ,and the differences observed may be due to the nature of the assays (i.e., fluid- phase (hanging clot) vs. solid-phase (microtiter wells)). In addition, Vn binding to clots is increased in the presence of PAI-l, and significantly greater amounts of active PAI-l can be incorporated into the clot in the presence of Vn. As in the hanging clot assay, 2.0 M NaCl had little significant effect on the displacement of Vn from fibrin-coated microtiter wells.

To further characterize the binding of different Vn conforms to fibrin, we analyzed the time course for the binding of ¹²⁵I-labelled denatured (multimeric) and native (monomeric) Vn to fibrin-coated microtitre wells in the presence or absence of 20-fold excess unlabelled Vn. Figure 22B illustrates that denatured Vn (in absence of excess cold Vn) binds rapidly to fibrin with saturation by approximately

60 min. However, addition of 20-fold excess Vn results in the potentiation of Vn binding with an approximately 2-fold relative increase in ¹²⁵I-Vn binding at a given time point. In contrast, native Vn binds with lower capacity (approximately 1/2 that of denatured) , and saturation is not reached by 2 h (Fig. 22C) . Similarly, addition of excess of cold Vn also resulted in a 2-fold increase in ^{1 5}I-Vn binding. These data are consistent with the hypothesis that both conforms of Vn bind to fibrin and that this binding is followed by a time-dependent multimerization of Vn on the fibrin surface. These results are in distinct contrast to the binding of the different Vn conforms to purified vimentin coated wells (Fig. 16A and B) . Thus, denatured (or mulitmeric Vn) binds rapidly and with a high degree of specificity to vimentin which saturates by 60 min (Fig. 16A) . In contrast, the early binding of ¹²⁵I-labelled native Vn to vimentin is virtually all non-specific and specific binding begins to be observed after 2 hr (Fig. 16B) . These data are consistent with earlier observations that only the multimeric form of Vn binds vimentin specifically, with an apparent affinity of 0.5 nM. Moreover, the behavior of native Vn binding to vimentin is consistent with the hypothesis that Vn multimerizer in a time and dose-dependent manner, and that the increase in specific binding observed part 2 h is due to gradual binding of newly forming Vn multimers.

The fibrin-binding region appears to be located on the amino-terminal half of Vn, close to the PAI-l binding site(s) . Biotinylated monomeric Vn (50 nM) was preincubated with various concentrations of normal mouse IgG (NM-IgG) , mAB 8E6, mAB 153 or mAB 1244 IgG followed by incubation in fibrin coated wells. mAB 153, and to a lesser degree mAB 8E6, inhibit Vn binding in a dose-dependent manner with maximal inhibition of 50% at 10 μg/ml. The mAB 1244 and normal mouse IgG had no significant effect. These data are consistent with the location of fibrin binding sites on Vn within the amino- terminal half of Vn and possibly close to the PAI-l binding site(s) , as both mAB 153 and 8E6 impair PAI-l binding to Vn. Further studies with synthetic peptide directed against the amino terminal, somatomedian domain and mAB 153 epitope were performed to elucidate the expression of the two PAI-l binding sites on fibrin-bound Vn (See below) . An alternative explanation for these results is that MAb's 153 and 8Eb interfere with Vn multimerization. (Fig. 23) .

4. USE OF VIMENTIN PEPTIDES TO ENHANCE FIBRINOLYSIS BY BLOCKING THE BINDING OF VN-PAI-1 COMPLEXES TO FIBRIN

CLOTS

Fig. 24 illustrates the displacement of fibrin- bound, biotinylated Vn following 1 hr incubation with the VIM 133 and 28-mer peptides. In the absence of PAI-l (-PAI-l) , maximal displacement (reduction in MOD/min) of 40% bound Vn was achieved with 2.0 nM of the VIM 28-mer peptide while the VIM 133-mer displaced slightly greater Vn than the control, scrambled 27-mer peptide (SCR) . In the presence of PAI-l (+PAI-1) , both the VIM 133 and 27-mer peptides displaced maximum of 40-60% of the biotinylated Vn.

To determine whether the vimentin peptides also displace PAI-l bound to Vn-fibrin complexes, fibrin coated wells were preincubated with buffer containing PAI-l alone (Fig. 25A, open bars) or preformed Vn-PAI-1 complexes (Fig. 25A, solid bars) , the unbound ligands washed away, and the wells incubated for 1 hr with various concentrations of the VIM 133 peptide. The residual PAI-l bound to fibrin was directly detected using ¹²⁵I-labelled MAI 12 IgG. In the presence of Vn and no peptide, there is approximately two-fold greater PAI-l bound to the clots as compared to PAI-l alone. Approximately 50% of the PAI-l bound to Vn (cpm greater than PAI-l alone on fibrin) was removed in a dose-dependent manner by the vimentin peptide. These data indicate that the displacement of PAI-l and Vn from fibrin by vimentin peptides is proportional. Dose-dependent enhancement of clot lysis by displacement of the vitronectin/PAI-1 complex by VIM 133 is described in Fig. 25B.

Saturating concentrations of biotinylated Vn plus PAI-l were incubated for 1 hr with fibrin-coated wells, the wells were then washed to remove the unbound proteins, and the clots then treated for an additional 1 hr with various concentrations of the VIM 28-mer peptide in the presence of absence of 20% human (citrated) plasma. The bound biotinylated Vn was detected with alkaline phosphatase- conjugated streptavidin/color substrate and measured as a function of the rate of change in absorbance at 405 nm. The percentage (mOD/min) of Vn bound to fibrin in the absence of peptide or plasma was expressed as 100%. Figure 26A illustrates that the presence of plasma alone results in an approximately 25% decrease in the bound Vn and that the VIM 28-mer peptide displaced additional amounts of Vn in the presence of plasma compared to the buffer only controls. These data indicate that the vimentin peptide is more specific for the clot-bound Vn and that plasma Vn has a minimal effect on the ability of vimentin to bind clot-bound Vn. Fig. 26B illustrates the dose-dependent prothrombolytic effects of the vimentin 133 mer peptide (VM- 133) on whole platelet-rich plasma clot lysis in 96-well microtiter plates. Various doses of purified VM-133 mer peptide was preincubated in citrated platelet-rich human plasma containing tracer amounts of ¹²⁵I-fibrinogen, the plasma samples (total volume 0.085 ml) were then recalcificied and allowed to clot for one hour prior to the initiation of exogenous tPA mediated lysis as measurement of the release of soluble ¹²⁵I-fibrin degradation fragments into the surrounding buffer. As positive controls, plasma samples were pretreated with MAI-12 IgG, an inhibitory mAB directed against the reactive center of PAI-l and, the negative controls were plasma samples pretreated with buffer only. These data indicate that the VM-133 peptide can enhance the rate of clot lysis for a given dose of tPA and that the displacement of the Vn-PAI-1 complexes within the clots results in enhanced net tPA activity which is equivalent to the enhancement observed for the direct inhibition of PAI-l by MAI-12 IgG. Thus vimentin or vimentin-like peptides may be useful as prophylactic agents to enhance endogenous thrombolysis. Figure 27 illustrates a model of the proposed interactions between different conformations of Vn with PAI-l, fibrin and vimentin. Thus, it is proposed that vimentin has a high affinity for multimeric Vn, particularly multimeric Vn-PAI-1 complexes. In contrast, fibrin appears to have a higher affinity for monomeric Vn but once this form of Vn is bound to fibrin, it can begin the multimerization process which may include PAI-l. This model may explain why the vimentin peptides can displace fibrin-bound Vn or Vn-PAI-1 complexes.

5. CHARACTERIZATION OF A SYNTHETIC PEPTIDE DERIVED FROM THE PAI-l BINDING EXOSITE ON TISSUE-PLASMINOGEN ACTIVATOR A PAI-l inhibitor which enhances fibrinolytic activity by binding adjacent to the reactive center and negatively charged, carboxy-terminal strained-loop of PAI-l to inhibit PAI-l interactions with t-PA was developed as follows. The VR-1 region of the t-PA light chain (residues 296-308) may be modified to eliminate inhibition of tPA by PAI-l. Madison et al. (1989) NATURE 339-721 and (1990) PROC. NATL. ACAD. SCI. USA 87:3530-3533. A synthetic t-PA exosite I peptide (14 mer residues 296-308) and its scramble control peptide were synthesized. These peptides were tested using a plasminogen-dependent, tPA chromogenic substrate (S-2288) combined with the microtiter well-bound fibrin plate lysis assay. Fibrin-coated wells were pre-incubated in the presence of absence of Vn/PAI-1 complexes prior to the addition of tPA, plasminogen, the synthetic peptides, mAB-12, and S-2288.

6. IMMUNOLOGICAL CHARACTERIZATION OF PAI-l BINDING TO MONOMERIC AND MULTIMERIC VITRONECTIN

The conformation of Vn (i.e., monomeric versus multimeric) appears to determine its stoichiometric interaction with active PAI-l and with other binding substrates, such as vimentin and fibrin. Monoclonal antibodies were used to characterize various functional conformations of PAI-l and Vn. The site of Vn which binds to the proposed heparin binding region of PAI-l has been reported to lie in both (1) the N-terminus somatomedian B domain and (2) an area contained within the middle region of the molecule at or near the peparin binding domain. To resolve this apparent inconsistency, the interaction(s) of PAI-l with Vn in the presence of monoclonal antibodies (mABs) to the amino-terminal region (mAB 153) and to the carboxyl-terminal region (mAB 8E6) of Vn were examined. It was found that Vn possess multiple PAI-l binding sites (which sites may also play a role in multimerization) . In solid phase binding assays, both mAB 153 and 8E6 were inhibitory, inhibiting PAI-l binding to urea-treated as well as native Vn in a dose dependent manner (Fig. 29A and Fig. 30) . In the presence of either antibody, the bound PAI-l retained its ability to inhibit tPA (Fig. 29B) . No binding of latent PAI-l was detectable. When immobilized to plastic, mABs 153 and 8E6 bound equivalent amounts of urea-treated Vn which in turn were found to bind equivalent amounts of PAI-l (Figs. 31A and 31B) . mABs 153 and 8E6 were shown to recognize distinct PAI-l binding sites since PAI-l binding to mAB 153-immobilized Vn was inhibited 85 ± 2% by addition of mAB 8E6, and PAI-l binding to mAB 8E6-immobilized Vn was inhibited 87 ± 2% in the presence of mAB 153 (Fig. 32) . Compared to urea-treated Vn, much greater concentrations of native Vn were required to achieve equivalent saturable binding to either 153 or 8E6 (Fig. 31A) , suggesting that the expression of both epitopes is conformationally dependent. In contrast to urea-treated (multimeric) Vn, the native (monomeric) Vn associated with either 153 or 8E6 bound little PAI-l (Fig. 31B) . In solution, at equivalent concentrations, urea-treated Vn was found to bind two-fold more PAI-l than did Vn in its native conformation (Figs. 33A and 33B) . Thus, it appears that multimeric Vn possesses two distinct PAI-l binding sites which bind in a PAI-l:Vn stoichiometry which approximates the equation:

[PAI-l] _ 2N - 2

where [-RAJ-1] = number of PAI-l molecules;

[Vn] _mult = number of mul timeric Vn molecules; and N = number of Vn subuni ts in mul timer (N≤ 18 ) As the number of subunits (N) increases above 4, the ratio PAI-l to Vn approaches 2:1. Monomeric Vn binds PAI-l in a 1:1 stoichiometric ratio.

Although the foregoing invention has been described in detail for purposes of clarity of understanding, it will be obvious that certain modifications may be practiced within the scope of the appended claims.

SEQUENCE LISTING

(1) GENERAL INFORMATION:

(i) APPLICANT: Podor, Thomas J. Weitz, Jeffery I. Hirsh, Jack

(ii) TITLE OF INVENTION: Methods and Compositions to Enhance Endogenous Fibrinolytic Activity

(iii) NUMBER OF SEQUENCES: 1

(iv) CORRESPONDENCE ADDRESS:

(A) ADDRESSEE: Townsend and Townsend Khourie and Crew

(B) STREET: One Market Plaza, Steuart Tower, Suite 2000

(C) CITY: San Francisco

(D) STATE: California

(E) COUNTRY: USA

(F) ZIP: 94105

(v) COMPUTER READABLE FORM:

(A) MEDIUM TYPE: Floppy disk

(B) COMPUTER: IBM PC compatible

(C) OPERATING SYSTEM: PC-DOS/MS-DOS

(D) SOFTWARE: Patentln Release #1.0, Version #1.25

(vi) CURRENT APPLICATION DATA:

(A) APPLICATION NUMBER: US

(B) FILING DATE:

(C) CLASSIFICATION:

(viii) ATTORNEY/AGENT INFORMATION:

(A) NAME: Heslin, James M.

(B) REGISTRATION NUMBER: 29,541

(C) REFERENCE/DOCKET NUMBER: 016558-000210

(ix) TELECOMMUNICATION INFORMATION:

(A) TELEPHONE: 415-326-2400

(B) TELEFAX: 415-326-2422

(2) INFORMATION FOR SEQ ID NO:l:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 133 amino acids

(B) TYPE: amino acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:l:

Met Ser Thr Arg Ser Val Ser Ser Ser Ser Tyr Arg Arg Met Phe Gly 1 5 10 15

Gly Pro Gly Thr Ala Ser Arg Pro Ser Ser Ser Arg Ser Tyr Val Thr 20 25 30

Thr Ser Thr Arg Thr Tyr Ser Leu Gly Asp Ala Leu Arg Pro Ser Thr 35 40 45

Ser Arg Ser Leu Tyr Ala Ser Ser Pro Gly Gly Val Tyr Ala Thr Arg 50 55 60

Ser Ser Ala Val Arg Leu Arg Ser Ser Val Pro Gly Val Arg Leu Leu 65 70 75 80

Gin Asp Ser Val Asp Phe Ser Leu Ala Asp Ala lie Asn Thr Glu Phe

85 90 95

Lys Asn Thr Arg Thr Asn Glu Lys Val Glu Leu Gin Glu Leu Asn Asp 100 105 110

Arg Phe Ala Asn Tyr lie Asp Lys Val Arg Phe Leu Glu Gin Gin Asn 115 120 125

Lys lie Leu Leu Ala 130

(2) INFORMATION FOR SEQ ID NO:2:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 14 amino acids

(B) TYPE: amino acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:2: lie Phe Ala, Lys His Arg Arg Ser Pro Gly Gly Arg Phe Leu 1 . 5 10

Claims

WHAT IS CLAIMED IS;

1. A method for inhibiting the accumulation of plasminogen activator inhibitor-l (PAI-l) in a ternary complex composed of a vitronectin component, a PAI-l component, and a fibrillar protein component selected from the group consisting of intracellular vimentin and clot-bound fibrin, said method comprising exposing an aqueous environment including the complex and/or the components of the complex to at least one substance that inhibits binding between vitronectin and the fibrillar protein component, between vitronectin and PAI-l, or between monomeric units of vitronectin.

2. A method as in claim 1, wherein the substance binds to PAI-l and is selected from the group consisting of vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on PAI-l, and antibody and antibody fragments that alter the conformation of PAI-l.

3. A method as in claim 2, wherein the substance has a binding affinity to PAI-l of at least 10⁵ M^-1, whereby initial binding between PAI-l and vitronectin may be inhibited.

4. A method as in claim 2, wherein the substance has a binding affinity to PAI of at least 10⁷ M^-1 whereby equilibrium binding between PAI-l and vitronectin may be disrupted.

5. A method as in claim 1, wherein the substance binds to vitronectin and is selected from the group consisting of PAI-l fragments, PAI-l analogs, antibody and antibody fragments that bind to a PAI-1-binding site on vitronectin, vimentin fragments, vimentin analogs, antibodies and antibody fragments that bind to a vimentin-binding site on vitronectin, and antibody and antibody fragments that alter the conformation of vitronectin, and vitronectin fragments and analogs that inhibit formation of multimeric vitronectin.

6. A method as in claim 5, wherein the substance has a binding affinity to vitronectin of at least 10⁵ M^-1, whereby initial binding between vitronectin and PAI-l or vimentin may be inhibited.

7. A method as in claim 5, wherein the substance has a binding affinity to vitronectin of at least 10⁷ M^"1, whereby equilibrium binding between vitronectin and PAI-l and vitronectin may be disrupted.

8. A method as in claim 1, wherein the substance binds to vimentin and is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on vimentin, and antibody and antibody fragments that alter the conformation of vimentin.

9. A method as in claim 8, wherein the substance has a binding affinity to vimentin of at least 10⁵ M^_1, whereby initial binding between vimentin and vitronectin may be inhibited.

10. A method as in claim 8, wherein the substance has a binding affinity to vimentin of at least 10⁷ M^-1, whereby equilibrium binding between vimentin and vitronectin may be disrupted.

11. A method as in claim 1, wherein the substance binds to fibrin and is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to the vitronectin-binding site on fibrin, and antibody and antibody fragments that alter the conformation of fibrin.

12. A method as in claim 11, wherein the substance has a binding affinity to fibrin of at least 10⁵ M^-1, whereby initial binding between fibrin and vitronectin may be inhibited.

13. A method as in claim 12, wherein the substance has a binding affinity to fibrin of at least 10⁷ M^"1, whereby equilibrium binding between fibrin and vitronectin may be inhibited.

14. A method for inhibiting the accretion of fibrin at an injury site within a blood vessel, said method comprising exposing the injury site to a substance which inhibits the accumulation of plasminogen activator inhibitor-l (PAI-l) at said site whereby PAI-l inhibition of fibrinolytic activity at; the injury site is reduced.

_*

15. A method as in claim 14, wherein the substance inhibits binding between vimentin and vitronectin, between vitronectin and PAI-l, or between monomeric units of vitronectin.

16. A method as in claim 15, wherein the substance binds to PAI-l and is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on PAI-l, and antibody and antibody fragments that alter the conformation of PAI-l.

17. A method as in claim 16, wherein the substance has a binding affinity to PAI-l of at least 10⁵ M^"1, whereby initial binding between PAI-l and vitronectin may be inhibited.

18. A method as in claim 16, wherein the substance has a binding affinity to PAI of at least 10^"7 M^"1, whereby equilibrium binding between PAI-l and vitronectin may be disrupted.

19. A method as in claim 15, wherein the substance binds to vitronectin and is selected from the group consisting of PAI-l fragments, PAI-l analogs, antibody and antibody fragments that bind to a PAI-1-binding site on vitronectin, vimentin fragments, vimentin analogs, antibodies and antibody fragments that bind to a vimentin-binding site on vitronectin, and antibody and antibody fragments that alter the conformation of vitronectin, and vitronectin fragments and analogs that inhibit formation of multimeric vitronectin.

20. A method as in claim 19, wherein the substance has a binding affinity to vitronectin of at least 10⁵ M^"1, whereby initial binding between vitronectin and PAI-l or vimentin may be inhibited.

21. A method as in claim 19, wherein the substance has a binding affinity to vitronectin of at least 10⁷ M^-1, whereby equilibrium binding between vitronectin and PAI-l and vimentin may be disrupted.

22. A method as in claim 15, wherein the substance binds to vimentin and is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on vimentin, and antibody and antibody fragments that alter the conformation of vimentin.

23. A method as in claim 22, wherein the substance has a binding affinity to vimentin of at least 10⁵ M^-1, whereby initial binding between vimentin and vitronectin may be inhibited.

24. A method as in claim 22, wherein the substance has a binding affinity to vimentin of at least 10⁷ M^"1, whereby equilibrium binding between vimentin and vitronectin may be disrupted.

25. A method for inhibiting the accretion of fibrin at a fibrin site within a blood vessel, said method comprising exposing the fibrin site to a substance which inhibits the accumulation of plasminogen activator inhibitor-l (PAI-l) at said site whereby PAI-l inhibition of fibrinolytic activity is reduced.

26. a method as in claim 25, wherein the substance inhibits binding between fibrin and vitronectin, between vitronectin and PAI-l, or between monomeric units of vitronectin.

27. A method as in claim 26, wherein the substance binds to PAI-l and is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on PAI-l, and antibody and antibody fragments that alter the conformation of PAI-l.

28. A method as in claim 27, wherein the substance has a binding affinity to PAI-l of at least 10⁵ M^"1, whereby initial binding between PAI-l and vitronectin may be inhibited.

29. A method as in claim 27, wherein the substance has a binding affinity to PAI of at least 10~⁷ M^"1, whereby equilibrium binding between PAI-l and vitronectin may be disrupted.

30. A method as in claim 26, wherein the substance binds to vitronectin and is selected from the group consisting of PAI-l fragments, PAI-l analogs, antibody and antibody fragments that bind to a PAI-1-binding site on vitronectin, vimentin fragments, vimentin analogs, antibodies and antibody fragments that bind to a vimentin-binding site on vitronectin, and antibody and antibody fragments that alter the conformation of vitronectin, and vitronectin fragments and analogs that inhibit formation of multimeric vitronectin.

31. A method as in claim 30, wherein the substance has a binding affinity to vitronectin of at least 10⁵ M^_1, whereby initial binding between vitronectin and PAI-l or vimentin may be inhibited.

32. A method as in claim 30, wherein the substance has a binding affinity to vitronectin of at least 10⁷ M^"1, whereby equilibrium binding between vitronectin and PAI-l and vimentin may be disrupted.

33. A method as in claim 26, wherein the substance binds to fibrin and is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to the vitronectin-binding site on fibrin, and antibody and antibody fragments that alter the conformation of fibrin.

34. A method as in claim 33, wherein the substance has a binding affinity to fibrin of at least 10⁵ M^"1, whereby initial binding between fibrin and vitronectin may be inhibited.

35. A method as in claim 33, wherein the substance has a binding affinity to fibrin of at least 10⁷ M^"1, whereby equilibrium binding between fibrin and vitronectin may be inhibited.

36. A pharmaceutical composition comprising a substance which binds to plasminogen activator inhibitor-l (PAI-l) , vitronectin, vimentin, or fibrin, with a specificity and an affinity sufficient to inhibit accretion of a ternary complex of PAI-l, vitronectin, and a fibrillar protein selected from the group consisting of intracellular vimentin and clot-bound thrombin at a site within a blood vessel, present in a pharmaceutical carrier.

37. A pharmaceutical composition as in claim 36, wherein the substance binds to PAI-l and is selected from the group consisting of vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on PAI-l, and antibody and antibody fragments that alter the conformation of PAI-l.

38. A pharmaceutical composition as in claim 37, wherein the substance has a binding affinity to PAI-l of at least 10⁵ M^-1, whereby initial binding between PAI-l and vitronectin may be inhibited.

39. A pharmaceutical composition as in claim 37, wherein the substance has a binding affinity to PAI of at least 10⁷ M^-1 whereby equilibrium binding between PAI-l and vitronectin may be disrupted.

40. A pharmaceutical composition as in claim 36, wherein the substance binds to vitronectin and is selected from the group consisting of PAI-l fragments, PAI-l analogs,

_* antibody and antibody fragments that bind to a PAI-1-binding site on vitronectin, vimentin fragments, vimentin analogs, antibodies and antibody fragments that bind to a vimentin- binding site on vitronectin, and antibody and antibody fragments that alter the conformation of vitronectin, and vitronectin fragments and analogs that inhibit formation of multimeric vitronectin.

41. A pharmaceutical composition as in claim 40, wherein the substance has a binding affinity to vitronectin of at least 10⁵ M^"1, whereby initial binding between vitronectin and PAI-l or vimentin may be inhibited.

42. A pharmaceutical composition as in claim 40, wherein the substance has a binding affinity to vitronectin of at least 10⁷ M^"1, whereby equilibrium binding between vitronectin and PAI-l and vimentin may be disrupted.

43. A pharmaceutical composition as in claim 36, wherein the substance binds to vimentin and is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on vimentin, and antibody and antibody fragments that alter the conformation of vimentin.

44. A pharmaceutical composition as in claim 43, wherein the substance has a binding affinity to vimentin of at least 10⁵ M^"1, whereby initial binding between vimentin and vitronectin may be inhibited.

45. A pharmaceutical composition as in claim 43, wherein the substance has a binding affinity to vimentin of at least 10⁷ M^"1, whereby equilibrium binding between vimentin and vitronectin may be disrupted.

46. A pharmaceutical composition as in claim 36, wherein the substance binds to fibrin and is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to the vitronectin-binding site on fibrin, and antibody and antibody fragments that alter the conformation of fibrin.

47. A pharmaceutical composition as in claim 46, wherein the substance has a binding affinity to fibrin of at least 10⁵ M^"1, whereby initial binding between fibrin and vitronectin may be inhibited.

48. A pharmaceutical composition as in claim 46, wherein the substance has a binding affinity to fibrin of at least 10⁷ M^-1, whereby equilibrium binding between fibrin and vitronectin may be inhibited.

49. An inhibitor of plasminogen activator inhibitor-l (PAI-l) comprising a hybrid molecule having a first moiety which binds to PAI-l and inhibits binding of PAI-l to vitronectin and a second moiety which binds to a reactive center on PAI-l which inhibits plasminogen activator inhibition activity.

50. An inhibitor as in claim 49, wherein the first and second binding moieties are spaced-apart by a distance in the range from 5 A to 50 A.

51. An inhibitor as in claim 49, wherein the first moiety is selected from the group consisting of vitronectin fragments, vitronectin analogs, antibody and antibody fragments that bind to a vitronectin-binding site on PAI-l, and antibody and antibody fragments that alter conformation of PAI-l.

52. An inhibitor as in claim 51, wherein the first moiety has a binding affinity to PAI-l of at least 10⁵ M"¹, whereby initial binding.between PAI-l and vitronectin may be inhibited.

53. An inhibitor as in claim 51, wherein the first moiety has a binding affinity to PAI-l of at least 10⁷ M^-1, whereby equilibrium binding may be disrupted.

54. An inhibitor as in claim 50, wherein the second moiety is selected from the group consisting of a plasminogen activator fragment, a plasminogen activator analog, and an antibody or antibody fragment which binds to the PAI-l reactive site.

55. An inhibitor as in claim 54, wherein the second moiety has a binding affinity to the PAI-l reactive center of at least 10⁵ M^"1.

56. An inhibitor as in claim 54, wherein the second moiety has a binding affinity to the PAI-l reactive center of at least 10⁷ M^"1.

57. An inhibitor of plasminogen activator inhibitor-l (PAI-l) comprising a hybrid molecule having a first moiety which binds to vitronectin and a second moiety which binds to a reactive center on PAI-l which inhibits plasminogen activator activity.

58. A method for screening test components to determine whether the test compound can inhibit the accumulation of plasminogen activator inhibitor-l (PAI-l) in a ternary complex composed of a vitronectin component, a PAI-l component, and a fibrillar protein component selected from the group consisting of intracellular vimentin and clot-bound fibrin, said method comprising; exposing the test compound to an aqueous environment including the ternary complex and/or at least two components of the complex; and determining whether presence of the test compound inhibits the incorporation of PAI-l into the ternary complex or binding between PAI-l and vitronectin in the aqueous environment.

59. A method as in claim 58, wherein vimentin is immobilized on a solid phase within the aqueous environment and the solid phase vimentin is exposed to vitronectin, PAI-l, and the test compound.

60. A method as in claim 59, wherein the vitronectin and PAI-l are present as a preformed binary conjugate.

61. A method as in claim 58, wherein fibrin is immobilized on a solid phase within the aqueous environment and the solid phase fibrin is exposed to vitronectin, PAI-l, and the test compound.

62. A method as in claim 61, wherein the vitronectin and PAI-l are present as a preformed binary conjugate.

63. A method as in claim 58, wherein the PAI-l is attached to a detectable label, and wherein incorporation of PAI-l into the ternary complex or binding to vitronectin is determined by washing the solid phase and detecting immobilized label on the solid phase.

64. A method as in claim 58, wherein tissue or cells are cultured on a solid phase in the aqueous environment and formation of the ternary complex is induced.