Direct fibrinolytic agents: biochemical attributes, preclinical foundation and clinical potential.

Direct fibrinolytics are proteolytic enzymes that degrade fibrin without requiring an intermediate step of plasminogen activation. This review summarizes the current information available for five such agents, namely, plasmin (the prototypical form), three derivatives of plasmin (mini-plasmin, micro-plasmin, and delta-plasmin), and alfimeprase, a recombinant variant of a snake venom alpha-fibrinogenase, fibrolase. Biochemical attributes of molecular size, fibrin binding and inhibitor neutralization are compared. Preclinical investigations that assess the potential for thrombolytic efficacy in vitro and in animal models of vascular occlusion and for hemostatic safety in animal models of bleeding are detailed. Clinical potential has been assessed in patients with peripheral arterial and graft occlusion, acute ischemic stroke, and access catheter and hemodialysis shunt occlusions. The direct fibrinolytic agents have impressive biochemical and preclinical foundations for ultimate clinical application. However, clinical trial results for micro-plasmin and alfimeprase have not measured up to their anticipated benefit. Plasmin has thus far shown encouraging hemostatic safety, but efficacy data await completion of clinical trials. Whether direct fibrinolytics will provide clinical superiority in major thrombotic disorders over currently utilized indirect fibrinolytics such as tissue plasminogen activator remains to be determined.

Thrombolytic potency of acid-stabilized plasmin: superiority over tissue-type plasminogen activator in an in vitro model of catheter-assisted thrombolysis.

Plasmin, the direct fibrinolytic enzyme, was compared with tissue plasminogen activator (t-PA) in an in vitro thrombolysis model. Plasmin has been prepared in a highly pure form from human plasma and has been stabilized against auto-degradation by low-pH formulation. This acidified formulation of plasmin has been designed to have a low buffering capacity so that it can be directly infused into clots in a stable and latently active form. This low-pH formulation has been shown to be equivalent to a neutral-pH formulation of plasmin in its extent of clot lysis. An in vitro model of catheter-assisted thrombolysis has been devised in which large (12 x 0.6 cm), retracted clots are treated with an intrathrombus thrombolytic agent via a multi-sideport catheter. Plasmin dissolves these plasminogen-deficient clots in a dose-dependent manner and is clearly superior to t-PA. In this model system, t-PA exhibits efficacy only when retracted clots are replenished with plasminogen.

Plasmin induces local thrombolysis without causing hemorrhage: a comparison with tissue plasminogen activator in the rabbit.

The direct fibrinolytic enzyme, plasmin, was compared with tissue plasminogen activator (TPA) in rabbit models of local thrombolysis and fibrinolytic hemorrhage. Plasmin was produced by solid-phase urokinase activation of plasminogen and purified on benzamidine Sepharose. Applied as an intra-arterial infusion into the thrombosed abdominal aorta under conditions of unimpeded blood flow, plasmin (4 mg/kg) and TPA (2 mg/kg) achieved equivalent clot dissolution and flow restoration. Using the model of restricted blood flow into the thrombosed aorta, which limits local plasminogen supply, plasmin was superior to TPA in clot lysis and vascular reperfusion. Using similar dosages of plasmin (2 or 4 mg/kg) and TPA (1 or 2 mg/kg) in the earpuncture rebleed model. TPA induced rebleeding in a dose-dependent manner from prior puncture sites in 9 of 10 animals, while none of the 10 animals exposed to plasmin rebled from these sites. These results suggest that plasmin is an effective, unique thrombolytic agent, distinguished from the plasminogen activators in current usage by its striking safety profile.

Thermodynamics of maltose binding protein unfolding.

The maltose binding protein (MBP or MalE) of Escherichia coli is the periplasmic component of the transport system for malto-oligosaccharides. It is used widely as a carrier protein for the production of recombinant fusion proteins. The melting of recombinant MBP was studied by differential scanning and titration calorimetry and fluorescence spectroscopy under different solvent conditions. MBP exhibits a single peak of heat absorption with a delta(Hcal)/delta(HvH) ratio in the range of 1.3-1.5, suggesting that the protein comprises two strongly interacting thermodynamic domains. Binding of maltose resulted in elevation of the Tm by 8-15 degrees C, depending of pH. The presence of ligand at neutral pH, in addition to shifting the melting process to higher temperature, caused it to become more cooperative. The delta(Hcal)/delta(HvH) ratio decreased to unity, indicating that the two domains melt together in a single two-state transition. This ligand-induced merging of the two domains appears to occur only at neutral pH, because at low pH maltose simply stabilized MBP and did not cause a decrease of the delta(Hcal)/delta(HvH) ratio. Binding of maltose to MBP is characterized by very low enthalpy changes, approximately -1 kcal/mol. The melting of MBP is accompanied by an exceptionally large change in heat capacity. 0.16 cal/K-g, which is consistent with the high amount of nonpolar surface--0.72 A2/g--that becomes accessible to solvent in the unfolded state. The high value of delta Cp determines a very steep delta G versus T profile for this protein and predicts that cold denaturation should occur above freezing temperatures. Evidence for this was provided by changes in fluorescence intensity upon cooling the protein. A sigmoidal cooperative transition with a midpoint near 5 degrees C was observed when MBP was cooled at low pH. Analysis of the melting of several fusion proteins containing MBP illustrated the feasibility of assessing the folding integrity of recombinant products prior to separating them from the MBP carrier protein.

Tissue-type plasminogen activator (tPA) interacts with urokinase-type plasminogen activator (uPA) via tPA's lysine binding site. An explanation of the poor fibrin affinity of recombinant tPA/uPA chimeric molecules.

Differential scanning calorimetry was used to study the domain structure and intramolecular interactions of tPA/uPA chimeras. A high temperature transition centered near 90 degrees C was observed upon melting of the tPA/uPA chimera (amino acids 1-274 of tPA and 138-411 of uPA) and its variant lacking the finger and epidermal growth factor-like modules (residues 1-3 and 87-274 of tPA and 138-411 of uPA). Since neither of the two parent plasminogen activators display such a stable structure, one may suggest that a new stabilizing intramolecular interaction occurs in the chimeras. We found that occupation of the lysine binding site of tPA by a lysine or arginine side chain from the urokinase moiety is responsible for the high temperature transition as well as for the failure of the chimeras to exhibit the expected fibrin binding properties. All uPA species, single- and two-chain high molecular weight uPA (Pro-Uk and HMW-Uk) and two-chain low molecular weight uPA (LMW-Uk), interact intermolecularly with tPA and its kringle-containing derivatives. This intermolecular interaction was strongly inhibited by epsilon-aminocaproic acid indicating that the lysine binding site of tPA is involved. The binding of uPA with the fluorescein-labeled A-chain of tPA, registered by changes in fluorescence anisotropy, was estimated to have a Kd range of 1-7 microM. The interaction of tPA with uPA determined by solid-phase assays appeared to be tighter, with a Kd range of 50-300 nM. Two synthetic peptides, with and without carboxyl-terminal lysine, corresponding to urokinase residues 144-158 and 144-157, were approximately 100-fold more potent than epsilon-aminocaproic acid with respect to inhibition of the tPA-uPA interaction, indicating that the tPA binding site on urokinase is located within this sequence, close to the activation site Lys158-Ile159. The discovered intermolecular interaction may be related to the reported synergistic effect of simultaneous administration of these two plasminogen activators.

Influence of carbohydrate on structure, stability, and function of gelatin-binding fragments of fibronectin.

The gelatin-binding region of fibronectin contains three Asn-linked carbohydrate moieties, one on the second type II module and two on the eighth type I "finger" module. Carbohydrate groups were enzymatically removed from two nonoverlapping gelatin-binding fragments (GBFs), 21-kDa GBF (modular composition I8-I9) and, with much greater difficulty, 30-kDa GBF (modular composition I6-II1-II2-I7). The gelatin-binding properties of these fragments were affected only slightly or not at all. Fluorescence and calorimetric analyses indicated that module I8 was strongly destabilized by deglycosylation such that the apo form melts near physiological temperature. A similar effect was caused by decreasing the pH of the holo form to 6.0, suggesting that one or more histidines are important for stability of module I8. The 21-kDa fragment exhibited an acid-induced change in fluorescence that occurred at higher pH in the deglycosylated derivative, providing further evidence of a stabilizing role for one or both carbohydrate moieties. By contrast, the stability of module II2 was unaffected by removal of its single carbohydrate. The effects of carbohydrate on the stability of module I8 may be important in efforts to express it or fragments containing it in bacteria.

Domain structure of the Fib-1 and Fib-2 regions of human fibronectin. Thermodynamic properties of the type I finger module.

The stability of the Fib-1 (29 kDa) and Fib-2 (19 kDa) fragments of human fibronectin as well as several different subfragments and isolated type I "finger" modules were studied under various solvent conditions by differential scanning calorimetry and fluorescence spectroscopy. It was established that all fibronectin fingers constitute independently folded domains whose melting temperatures range from 54 to 108 degrees C. The difference between heat capacities of the native and denatured states (delta Cp) is low, about 0.03 cal/K-g, which is consistent with the relatively low percentage of hydrophobic amino acids and the consequent small change in non-polar surface area exposed to the solvent upon denaturation. The free energy of unfolding at 25 degrees, as calculated from the calorimetric data or measured directly by titration with GdmSCN is also small, in the range of 2.4 to 6.7 kcal/mol. The small delta G value and its flat dependence on temperature (determined by delta Cp) translates the small variations in delta G between fingers into large variations in tm. The small value of delta G also indicates that finger modules are structurally rather fragile which may account for their sensitivity to proteolysis; almost any cleavage within either of the two disulfide loops destroys the cooperative structure and abolishes the corresponding melting transition. The fact that some fingers exhibit large decreases in tm upon separation from more stable neighbors with which they interact can also be viewed as a consequence of the low values of delta G and delta Cp.

Effect of tethered peptidylchloromethylketone inhibitors on thermal stability and domain interactions of urokinase and other serine proteases.

The melting of several serine proteases that had been reacted with different peptidylchloromethylketone (cmk) inhibitors was studied by fluorescence spectroscopy and calorimetry. These inhibitors, which cross-link the two domains of the proteases, invariably increased the melting temperature by as much as 28.5 degrees C. The magnitude of the effect was dependent on the size and composition of the peptide moieties. The delta G of unfolding of tosyl-Phe-cmk-chymotrypsin was 13.5 kcal/mol compared to only 8.3 kcal/mol for chymotrypsin. Binding of cmk inhibitors also protected the two interacting domains of urokinase from acid-induced decooperation and caused them to merge into a highly cooperative structure upon refolding at low pH. Fluorescence-detected melting curves of Glu-Gly-Arg-cmk-urokinase indicated that unfolding/refolding at pH 4.5 is characterized by dramatic hysteresis; the cooling curves fell close to those obtained upon heating or cooling of the uninhibited enzyme. Upon second heating, the melting curves were similar to those of the original. The hysteresis effects are interpreted as follows. The tethered tripeptide binds to the active site, causing the protein to melt at much higher temperature in a single cooperative step, as if the two domains are merged into one cooperative unit. Upon cooling, the unfolded protein, with the inhibitor still attached, refolds at the same temperature as the underivatized protein. Only after the native structure is formed does the peptide moiety again bind and stabilize toward a second heating. At lower pH, second heating produced biphasic or triphasic melting curves that were attributed to differential protonation of acid-titratable groups on the enzyme and/or inhibitor at the time of refolding. Similar effects were observed with other trypsin-like proteases, indicating that the hysteresis and bi- and triphasic refolding at low pH are rather general for this class of enzyme.

Domain structure and domain-domain interactions in the carboxy-terminal heparin binding region of fibronectin.

The domain structures and stabilities of fragments isolated from the so-called 'hep 2' region of plasma fibronectin have been investigated by differential scanning calorimetry (DSC) and fluorescence spectroscopy. The 30 kDa hep-2A fragment contains three type III modules (III12 to III14), whereas the 40 kDa hep-2B fragment contains four such modules (III12 to III15). Melting of these fragments at neutral pH was irreversible and accompanied by rapid aggregation. In contrast, melting was completely reversible in 50 mM-glycine at pH 2.7, where DSC measurements revealed the presence of three independently folded domains in 30kDa hep-2A and four in 40 kDa hep-2B. That each domain represented a single module was confirmed by measurements with four single-module subfragments, all of which melted reversibly, even at neutral pH. At neutral pH in the presence of 6 M-urea, 30 kDa hep-2A melted reversibly in a sharp peak from which only two transitions could be resolved by deconvolution. Only the larger of these was stabilized by heparin and was assigned to modules III13 and III14. Upon isolation, module III13 melted at lower temperature than in the parent fragment where it is stabilized through an interaction with module III14. We conclude that all type III modules in the hep-2 region of fibronectin constitute independently folded domains. Modules III13 and III14 form a highly co-operative structure through functionally significant interactions that can be disrupted with acid or sufficient concentrations of urea or guanidinium chloride.

Reversible unfolding of an isolated heparin and DNA binding fragment, the first type III module from fibronectin.

Several fragments containing all or part of the first type III homology unit of fibronectin were isolated and their folding properties examined by fluorescence spectroscopy and differential scanning calorimetry. Each fragment exhibits a reversible unfolding transition when heated or titrated with guanidinium chloride. This indicates that an isolated type III module can fold independently in the absence of neighboring modules. A comparison of the specific enthalpies of unfolding of these fragments with those of well-studied globular proteins suggests that this type III unit is composed of a stable core flanked by less compact or unstructured regions. Comparison of the heparin-binding properties of these fragments revealed that removal of 12 amino acids from the amino terminus of the largest one (Ile-585 to Val-675) increased its affinity for immobilized heparin such that it now binds at physiological ionic strength.

Domain structure and interactions of recombinant urokinase-type plasminogen activator.

Urokinase-type plasminogen activator (uPA) is a mosaic glycoprotein composed of an epidermal growth factor-like (EGF), a kringle and a serine protease (SP) module. It exists in single and two-chain forms designated HMW pro-uPA and HMW uPA, respectively. A low molecular weight form, LMW uPA, lacks the EGF and kringle modules and is composed of the SP module alone. Recombinant-expressed proteins representing both HMW forms exhibit four reversible unfolding transitions that are resolved by deconvolution of melting curves obtained by differential scanning calorimetry at pH 4.5; no differences in the melting properties of the single and two-chain forms were found. The proteolytic fragment Ser1-Lys135 (EGF-kringle) exhibits two transitions, while the isolated EGF and kringle modules each exhibit a single two-state transition. Thus, both of these modules retain an independently folded compact structure when isolated. The isolated SP module (LMW uPA) exhibits two closely spaced transitions at low pH indicating the melting of two domains of similar stability. Fluorescence-detected melting curves of LMW uPA reveal increasing cooperativity with increasing pH, suggesting an increase in the interaction between the two SP domains. Treatment of both HMW and LMW uPA with the tripeptide inhibitor Glu-Gly-Arg-chloromethylketone dramatically increased the stability of both domains of the SP module which now melt together in a single two-state transition, even at low pH, with no effect on the EGF and kringle modules. From these data one concludes that UK consists of four independently folded domains. Two are formed by the EGF and kringle modules which do not interact with each other or with the SP module. The SP module contains two domains that are independent at low pH but exhibit a tendency to merge into a single cooperative unit at neutral pH or after treatment with the tripeptide inhibitor.

Domain structure and domain-domain interactions of recombinant tissue plasminogen activator.

The melting of recombinant tissue plasminogen activator (rtPA) has been investigated by differential scanning calorimetry and fluorescence spectroscopy. At neutral pH, rtPA melts with only partial reversibility in a single sharp peak that can be deconvoluted into four transitions. By contrast, at acidic pH the melting process is spread over a broad range of temperature and is highly reversible. Under these conditions five transitions are resolved by deconvolution analysis. Additional measurements in 6 M guanidinium chloride reveal a sixth transition representing an extremely stable domain. Comparison of the melting curves of several fragments with those of the parent protein allowed all of the transitions to be assigned. The results indicate that rtPA is comprised of six independently folded domains. Two of these domains correspond to the two kringle modules whose thermodynamic properties are similar to those of the kringles in plasminogen. Two additional domains are formed by the epidermal growth factor (EGF)-like and finger modules, the latter of which is extremely stable, requiring the presence of a chemical denaturant for its melting to be observed. The serine protease module contains two more domains which at neutral pH melt cooperatively in a single transition but at low pH melt independently, accounting for the greater number of transitions observed there. Measurements with a 50-kDa fragment lacking the C-terminal half of the serine protease module and with a variant lacking the finger and EGF domains indicate that the serine protease domains interact strongly with and are stabilized by the finger and/or EGF domains in the intact protein. This interaction between domains located at opposite ends of the rtPA molecule produces a more compact structure. A better understanding of such interactions may enhance efforts to engineer plasminogen activators with improved thrombolytic properties.

Fluorescence spectroscopic analysis of ligand binding to kringle 1 + 2 + 3 and kringle 1 fragments from human plasminogen.

The ligand binding of kringle 1 + 2 + 3 and kringle 1 from human plasminogen has been investigated by fluorescence spectroscopy. Analysis of fluorescence titration of kringle 1 + 2 + 3 with 6-aminohexanoic acid shows that this fragment, besides the high-affinity lysine-binding site with Kd = 2.9 microM, contains two additional lysine-binding sites which differ in binding strength (Kd = 28 microM and Kd = 220 microM). This strongly suggests the existence of a lysine-binding site in each of the first three kringles. 6-Aminohexanoic acid, pentylamine, pentanoic acid and arginine were used for investigation of the ligand specificity of isolated kringle 1 prepared by pepsin hydrolysis of kringle 1 + 2 + 3. It has been established that kringle 1 has high affinity to 6-aminohexanoicacid, pentylamine and arginine (Kd values are 3.2 microM, 4.8 microM and 4.3 microM, respectively). At the same time pentanoic acid did not bind with kringle 1. These facts indicate, firstly, a broad ligand specificity of kringle 1 and, secondly, the paramount importance of the positively charged group of the ligand for its interaction with lysine-binding site of this kringle.

Analysis of ligand binding to kringles 4 and 5 fragments from human plasminogen.

The interaction of the isolated kringles 4 and 5 from human plasminogen with 6-aminohexanoic acid, pentylamine, pentanoic acid and arginine has been quantitatively characterized by scanning calorimetry and fluorescent spectroscopy. It has been found that the ligands with the positively charged group have a good binding ability while pentanoic acid in comparison with 6-aminohexanoic acid being devoid of amino group does not interact with the kringles under study. The positively charged group of the ligand is suggested to play a crucial role in ligand binding with the lysine-binding site.

Domains in human plasminogen.

Calorimetric studies of intramolecular melting of human plasminogen and of its fragments under various solvent conditions show that the intact plasminogen molecule consists of seven compact co-operative subunits, which can be regarded as structural domains. Five of these domains are formed by the homologous regions, the kringles, two domains are formed by the C-terminal part of the polypeptide chain that is split at activation, forming the light chain in plasmin, while the initial 76 amino acid residue peptide does not form any compact co-operative structure. The specific influence of epsilon-aminocaproic acid on the stability of the first, the fourth and, to a lesser extent, on the second kringle domain, provides evidence that these three domains in plasminogen possess lysine-binding ability. The first four kringle domains are almost independent in the molecule, while the fifth interacts with that part of the light chain not included in either of the two domains of this chain. These two domains are of different size and co-operate strongly in plasminogen, but at its activation into plasmin they decooperate and the stability of the smaller domain, which is formed by the N-terminal part of the light chain, decreases significantly. Since the light chain is responsible for the proteolytic activity of plasmin, it becomes clear that the active site of this protein is composed of two domains, as is the case for other serine proteases.