Human Plasmin Enzymatic Activity Is Inhibited by Chemically Modified Dextrans*

Some synthetic dextran derivatives that mimic the action of heparin/heparan sulfate were shown to promote in vivo tissue repair when added alone to wounds. These biofunctional mimetics were therefore designated as “regenerating agents” in regard to their in vivo properties.In vitro, these biopolymers were able to protect various heparin-binding growth factors against proteolytic degradation as well as to inhibit the enzymatic activity of neutrophil elastase. In the present work, different dextran derivatives were tested for their capacity to inhibit the enzymatic activity of human plasmin. We show that dextran containing carboxymethyl, sulfate as well as benzylamide groups (RG1192 compound), was the most efficient inhibitor of plasmin amidolytic activity. The inhibition of plasmin by RG1192 can be classified as tight binding hyperbolic noncompetitive. One molecule of RG1192 bound 20 molecules of plasmin with a K i of 2.8 × 10− 8 m. Analysis with an optical biosensor confirmed the high affinity of RG1192 for plasmin and revealed that this polymer equally binds plasminogen with a similar affinity (K d = 3 × 10− 8 m). Competitive experiments carried out with 6-aminohexanoic acid and kringle proteolytic fragments identified the lysine-binding site domains of plasmin as the RG1192 binding sites. In addition, RG1192 blocked the generation of plasmin from Glu-plasminogen and inhibited the plasmin-mediated proteolysis of fibronectin and laminin. Data from the present in vitroinvestigation thus indicated that specific dextran derivatives can contribute to the regulation of plasmin activity by impeding the plasmin generation, as a result of their binding to plasminogen and also by directly affecting the catalytic activity of the enzyme.

Some synthetic dextran derivatives that mimic the action of heparin/heparan sulfate were shown to promote in vivo tissue repair when added alone to wounds. These biofunctional mimetics were therefore designated as "regenerating agents" in regard to their in vivo properties. In vitro, these biopolymers were able to protect various heparin-binding growth factors against proteolytic degradation as well as to inhibit the enzymatic activity of neutrophil elastase. In the present work, different dextran derivatives were tested for their capacity to inhibit the enzymatic activity of human plasmin. We show that dextran containing carboxymethyl, sulfate as well as benzylamide groups (RG1192 compound), was the most efficient inhibitor of plasmin amidolytic activity. The inhibition of plasmin by RG1192 can be classified as tight binding hyperbolic noncompetitive. One molecule of RG1192 bound 20 molecules of plasmin with a K i of 2.8 ؋ 10 ؊8 M. Analysis with an optical biosensor confirmed the high affinity of RG1192 for plasmin and revealed that this polymer equally binds plasminogen with a similar affinity (K d ‫؍‬ 3 ؋ 10 ؊8 M). Competitive experiments carried out with 6-aminohexanoic acid and kringle proteolytic fragments identified the lysine-binding site domains of plasmin as the RG1192 binding sites. In addition, RG1192 blocked the generation of plasmin from Glu-plasminogen and inhibited the plasmin-mediated proteolysis of fibronectin and laminin. Data from the present in vitro investigation thus indicated that specific dextran derivatives can contribute to the regulation of plasmin activity by impeding the plasmin generation, as a result of their binding to plasminogen and also by directly affecting the catalytic activity of the enzyme.
We have previously reported that some dextran derivatives could stimulate tissue repair when applied at the site of the injury in various in vivo models such as skin (1), bone (2), colon (3), cornea (4), and muscle (5). These biopolymers were ob-tained by controlled chemical substitution of dextran polymers by defined amounts of carboxymethyl, sulfate as well as hydrophobic groups such as benzylamide. As regards their in vivo properties, these biopolymers were denominated regenerating agents (RGTA). 1 Our initial interpretation of the ability of these biopolymers to stimulate tissue repair was to postulate that these molecules acted as functional mimetics of heparin/heparan sulfate in terms of stabilizers, protectors, and potentiators of endogenously released heparin-binding growth factors. This hypothesis was supported by in vitro experiments, which showed that these biopolymers protected some heparin-binding growth factors such as fibroblast growth factors and transforming growth factor ␤ against proteolytic degradation and enhanced their bioavailability (3,6). A second interpretation of the in vivo wound healing properties of these polymers, which does not exclude the first, is that they could also act on some of the proteinases involved in tissue remodeling. This led us to report in a previous study that human neutrophil elastase was inhibited by specific dextran derivatives (7). Among other known proteinases involved in tissue remodeling, plasmin plays a key role, since it acts directly by hydrolyzing components of the basement membrane such as fibrin, fibronectin, and laminin and also acts indirectly by activating other enzymes such as matrix metalloproteases (8,9).
As regards the pivotal role of this enzyme in tissue remodeling, we have further investigated the effect of these dextran derivatives on the enzymatic activity of plasmin. We report that as for neutrophil elastase, human plasmin activity is inhibited by specific dextran derivatives that contained the aromatic residue benzylamide. Complementary studies revealed that this type of biopolymer also bind plasminogen and modulate plasmin activity in a noncompetitive manner via regulatory sites involving the lysine-binding site (LBS) domains of plasmin.

Synthesis of Dextran Derivatives
Water-soluble modified dextrans were prepared from T40 dextran (average M r 37,000; Amersham Pharmacia Biotech) according to the method described by Mauzac et al. (12). Compound RG1100 (Fig. 1) was synthesized from dextran T40 by carboxymethylation of OH residues with monochloroacetic acid treatment in aqueous NaOH at 50°C, pH 10, for 20 min with constant stirring. The product was then precipitated with methanol and dried under vacuum. The presence of carboxymethyl groups was confirmed by infrared spectroscopy with the appearance of an absorption band at 1650 cm Ϫ1 . Compound RG1503 (Fig. 1) was synthesized from a carboxymethylated dextran by O-sulfonation with 2 eq of chlorosulfonic acid in dry dichloromethane at room temperature for 2 h with constant stirring. The dichloromethane was eliminated by filtration, and the powdered O-sulfonated product was dissolved in water, adjusted to pH 7.3 with 3 M NaOH, ultrafiltered with a PLCGC 10 K membrane (Millipore, France), and freeze-dried. The presence of sulfate groups was indicated by infrared spectroscopy with the appearance of two absorption bands at 1250 and 1025 cm Ϫ1 . Compound RG1192 (Fig. 1) was synthesized from a carboxymethylated dextran by amidation of the carboxylate residues with benzylamine followed by O-sulfonation. Briefly, carboxymethylated dextran was dissolved in H 2 O/EtOH, and carboxylate functions were activated with N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline for 30 min at pH 3.5 at room temperature. 2 eq of free benzylamine were then added, and the reaction mixture was kept at room temperature overnight. The product was precipitated and washed with methanol and then dried under vacuum. The infrared spectrum confirmed the presence of benzylamide functions with the appearance of a new absorption band at 1750 cm Ϫ1 , corresponding to the carbonyl bond of the amide. This derivatized carboxymethyl-benzylamide compound was then O-sulfonated according to the above protocol. In addition, we elaborated a dextran sulfate (RG1003 compound, Fig. 1) by directly substituting T40 dextran with sulfate groups according to the above O-sulfonation protocol.
It is noteworthy that the conditions of O-sulfonation we used were originally described for sulfonation of aromatic hydrocarbon. It has been also reported that under the same conditions, hydroxyl groups can compete with aromatic groups to generate a mixture of O-and Csulfonates (13). This is illustrated during the course of the RG1192 synthesis, where total reactive groups present in its precursor form were distributed among 87 and 13% of hydroxyl and aromatic groups, respectively. To thus establish whether RG1192 contained a mixture of O-and C-sulfonates, a 1 H NMR (200 MHz, D 2 O) spectroscopy study was undertaken. In the aromatic region spectrum, only a broad 1 H peak at 7.15 ppm, assigned to the aromatic protons of unsubstituted benzylamide, was detected. As control, a C-sulfonated product was synthesized, and its analysis by 1 H NMR showed two broad multiplets at 7.3 and 7.6 ppm. The absence of C-sulfonate groups was confirmed by 13 C NMR (75 MHz, D 2 O) spectrum, in which the 13 C signal at 142 ppm, present in the C-sulfonated product was not detected in the RG1192 product. We then concluded that as described previously for other dextran derivatives (14), C-sulfonate groups were undetectable in the RG1192 compound and that O-sulfonation mainly occurred during treatment of derivatized carboxymethyl-benzylamide dextrans with chlorosulfonic acid. This dominant O-sulfonate formation may thus be explained by the higher content of OH groups (87%) as compared with aromatic groups (13%) present in the RG1192 precursor.
The chemical characterization of all dextran derivatives was based on the degree of substitution (d.s.) of each individual group per glucosidic unit (Table I). Each d.s. value was determined by acidimetric titration for CH 2 COONa and SO 3 Na content and elementary analysis for CH 2 CONHCH 2 C 6 H 5 and SO 3 Na content. All of these values were FIG. 1. Schematic structure of dextran derivatives. Polymers were elaborated from T40 dextran by chemical substitutions as described under "Experimental procedures." Dextrans were substituted by carboxymethylation (RG1100) or O-sulfonation (RG1003); by carboxymethylation followed by O-sulfonation (RG1503); or by carboxymethylation followed by amidation with benzylamine and O-sulfonation (RG1192). The different percentages indicated in the figure were calculated from the d.s. relative to the position of each group in a glucosidic unit, as reported in Table I. For an easy representation, the substituted glucosidic units A, B, C, and D were arranged in an arbitrary combination. Their respective proportions (percentages) within each polymer were calculated according to the nature of the group linked to the C-2 position. In addition, R represents the proportion (percentage) of each substituted group in the global C-3 ϩ C-4 positions.
confirmed by 1 H NMR. Distribution of each group among the three reactive OH groups was also reported in Table I. Results showed that reactions of carboxymethylation and sulfonation on hydroxyl functions preferentially occurred on the C-2 position. These results are in agreement with those showing that the OH on C-2 position displayed the higher rate coefficient of dextran carboxymethylation (15). The average molecular weight of each polymer (Table I) was determined by high performance size exclusion chromatography in 0.1 M NaNO 3 , using KB-804 and KB-805 aqueous gel filtration columns (Shodex, Japan) applied in series. The effluent was monitored with a mini Dawn light scattering detector and a RID 10 A refractometer (Touzard & Matignon, France). The flow rate was 0.7 ml/min. All of these polymers did not present any significant anticoagulant activity (less than 5 IU/mg as compared with 173 IU/mg for heparin) (6).

Effect of Various Polymers on the Enzymatic Activity of Proteolytic Enzymes
Enzymatic kinetics were monitored with a Philips PU8740 spectrophotometer equipped with a thermostated cell holder. The progress curves were recorded for 0.3-5 min, depending upon the reaction velocity, and less than 5% of the substrate was hydrolyzed during the rate measurement. Plasmin, trypsin, and ␣-chymotrypsin enzymatic activities were determined in 50 mM Tris/HCl buffer, pH 7.4, containing 50 mM NaCl at 37°C and steady-state velocities were measured by following the release of p-nitroaniline at 410 nm (⑀ ϭ 8800 M Ϫ1 cm Ϫ1 ). Human tPA and uPA enzymatic activities were assayed in similar conditions at pH 8.8. To determine the kinetic constants k cat and K m , initial rates were measured as a function of substrate concentrations, and the data were fitted to the Michaelis-Menten rate equation using the GraphPad Prism software (San Diego, CA). The k cat and K m values for the plasmin/ S-2251 system are 14 s Ϫ1 and 0.4 mM, and those for the plasmin/S-2444 system are 9 s Ϫ1 and 5.3 mM, respectively.
The effect of various polymers on the different enzymatic activities was determined by reacting constant concentration of enzyme with increasing concentrations of polymers for 15 min at 37°C and measuring the residual enzymatic activity with a synthetic substrate. For plasmin, the equilibrium dissociation constants (K i ) were measured with 8 nM of enzyme and 0.4 mM of substrate (S-2251). The K i values and their S.E. values were calculated by nonlinear regression using the integral equation editor of the GraphPad Prism software. For the determination of inhibitor-plasmin binding stoichiometries, 750 nM plasmin was used, and the substrate was S-2444 (0.4 mM).

Plasmon Surface Resonance Analysis
Biotinylation of RG1192-RG1192 polymer was labeled by reaction of its aldehydic reducing group with the hydrazide functional group of biocytin hydrazide using a modified method described by Nadkarni et al. (16). Briefly, 15 mg of polymer were dissolved in 50 l of formamide containing 50 mM biocytin hydrazide and heat at 37°C for 24 h. Free biocytin hydrazide and formamide were removed by gel filtration chromatography (Econo-Pac 10 DG) (Bio-Rad), and homogeneity of the labeled polymer was checked by gel permeation chromatography (TSK gel G4000 PWXL) (Tosohaas, Montgomeryville, PA).
Preparation of Sensor Surfaces-The immobilization procedure was (1 mg/ml) was injected, followed by two pulses of 10 mM HCl to remove noncovalently attached ligand. The amount of RG1192 immobilized on the surface was 100 resonance units. Determination of Equilibrium Dissociation Constant-Binding reactions were carried out at a flow rate of 10 l/min at 25°C. Various concentrations of plasmin or Glu-plasminogen in 50 mM Tris-HCl, pH 7.4, 50 mM NaCl, 0.005% P-20 were injected over the RG1192-immobilized surface, and change in resonance signal was monitored according to time. Equal volumes of each protein were also injected over a streptavidin-immobilized surface to serve as a blank for subtraction of nonspecific binding of analyte. The sensor surfaces were regenerated with a pulse of 2 M NaCl between each injection of analyte. Equilibrium dissociation constants were determined as described in the BIAtechnology handbook (Amersham Pharmacia Biotech).
At steady state, the following is true, which may be rearranged as follows, where K d ϭ k d /k a is the equilibrium dissociation constant, R eq is the response value at steady state, R max is the maximal capacity of the sensor chip for binding analyte, and C is the molar concentration of analyte. K d was calculated by nonlinear regression analysis by fitting the (R eq , C) pairs to Equation 2 using the GraphPad Prism software.

Western Blot Analysis of Fibronectin Proteolytic Fragments
Plasmin (1.6 M) and RG1192 (0.05-1 M) were incubated at 37°C for 30 min in 50 mM Tris/HCl, pH 7.4, 50 mM NaCl, 5 mM CaCl 2 , 0.01% Triton X-100 prior to the addition of 10 ng of fibronectin (0.5 nM). Following 4 h of incubation, the reaction was stopped by the addition of 1.6 M of aprotinin, and samples were separated by SDS-PAGE (7% (w/v) acrylamide) under reducing conditions. Proteins were electrophoretically transferred overnight at 4°C to Immobilon-P membrane in 25 mM Tris, pH 8.3, 192 mM glycine. Membranes saturated with Superblock ® blocking buffer were incubated with polyclonal rabbit anti-human fibronectin Ig at a 1:1000 dilution in phosphate-buffered saline containing 0.02% (v/v) Tween 20 and 0.3% Superblock ® blocking buffer. Antibodies were detected by using horseradish peroxidase-conjugated goat anti-rabbit IgG and ECL chemiluminescence according to the manufacturer's recommendations.  3 represents the maximum of substitution, since one glucosidic unit contains three reactive OH groups on C-2, C-3, and C-4 positions. The position of each group on the C-2 versus C-3 ϩ C-4 positions was also determined by analyzing the anomeric proton signal in 1 H NMR. In this representation, a d.s. value of 1 and 2 represents the maximum of substitution for C-2 and C-3 ϩ C-4 positions, respectively. S.D. of d.s. values were less than 5% (n ϭ 3).

Western Blot Analysis of Laminin Proteolytic Fragments
Plasmin (0.8 M) and RG1192 (0.05-2.5 M) were incubated at 37°C for 30 min in 50 mM Tris/HCl, pH 7.4, 50 mM NaCl, 0.01% Triton X-100. 50 ng of laminin (2 nM) was then added and the reaction mixture was kept for additional 60 min. The reaction was stopped with 0.8 M aprotinin, and samples were separated by SDS-PAGE (4 -15% (w/v) acrylamide gradient) under reducing conditions. Transfer and hybridization were performed in the same way as for fibronectin, except that the first incubation was performed with polyclonal rabbit anti-mouse laminin Ig at a 1:1000 dilution.

Electrophoretic Analysis of Plasminogen Activation
The generation of plasmin from Glu-plasminogen in the presence of RG1192 was analyzed in 50 mM Tris/HCl, pH 7.4, 50 mM NaCl, 0.01% Triton X-100 as follows. 3 g of Glu-plasminogen (0.8 M) was preincubated for 30 min at 37°C with increasing concentrations of RG1192 (0.05-1 M) before the addition of uPA (20 nM). After 2 h of incubation, samples were separated by SDS-PAGE (10% (w/v) acrylamide) under reducing conditions. Gels were then fixed and stained with Coomassie Brillant Blue R-250.

Effect of Dextran Derivatives on the Amidolytic Activity of
Plasmin, ␣-Chymotrypsin, Trypsin, uPA, and tPA- Fig. 2 shows the influence of increasing concentrations of various dextran derivatives obtained by sequential chemical substitutions of dextran polymers on the amidolytic activity of plasmin. Dextran substituted with carboxymethyl functions (RG1100) did not affect the activity of plasmin even for concentrations greater than 1 M. Following O-sulfonation, the resulting derivatized polymer (RG1503) showed an inhibitory potential toward human plasmin (IC 50 ϭ 20 nM) with 70% of residual enzyme activity. This potential was, however, better than the one obtained with a dextran sulfate (RG1003), indicating a beneficial contribution of the carboxymethyl groups with respect to the antiproteinase activity of this polymer. Interestingly, when both carboxymethyl, sulfate and benzylamide groups were coupled to dextran glucosidic units (RG1192), a potent inhibitory activity of plasmin was observed. The IC 50 value was 2 nM with 20% of residual enzyme activity. In contrast, heparin and heparan sulfate did not affect the activity of plasmin.
Among the various dextran derivatives tested, RG1192 was the most efficient inhibitor of plasmin activity. The antiproteinase potency of this polymer was therefore investigated toward other serine proteinases such as trypsin, ␣-chymotrypsin, and the two human plasminogen activators, uPA and tPA (Fig. 3).
In the same concentration range used for plasmin, trypsin and uPA were insensitive to RG1192. ␣-Chymotrypsin activity was slightly affected by the presence of RG1192, whereas 70% of the tPA activity was inhibited at a saturating concentration of inhibitor. However, the inhibitory potential of RG1192 on tPA was lower than that of obtained with plasmin: IC 50 equal to 34 nM as compared with 2 nM for plasmin. Study of the Mechanism of Plasmin Inhibition by RG1503 and RG1192-The mechanism of the inhibition pattern of a single-substrate reaction can be schematically represented as follows, where n is the number of free or substrate-bound plasmin molecules per molecule of inhibitor, K i is the equilibrium dissociation constant of the enzyme-inhibitor complex, ␣ is a dimensionless number that affects the binding of substrate to the plasmin-inhibitor complex or that of inhibitor to enzyme-substrate complex, and ␤ is a dimensionless factor that affects the catalytic constant k cat . Hyperbolic noncompetitive inhibition is characterized by ␣ ϭ 1 and 0 Ͻ ␤ Ͻ 1. While classical inhibition does not depend upon the total enzyme concentration, tight binding inhibition does (17), since the concentration of the bound inhibitor [EI] is no longer negligible with respect to that of total inhibitor concentration [I] 0 . We therefore analyzed whether RG1192 and RG1503 behaved as tight binding inhibitors by measuring the inhibition of plas- min activity using different enzyme concentrations. The inhibition profiles obtained with RG1192 (Fig. 4) indicated tight binding inhibition, since plasmin inhibition decreased with the total enzyme concentration [E] 0 ; IC 50 ϭ 2 and 8 nM for [E] 0 ϭ 8 and 150 nM, respectively. At high enzyme concentration that ensured pseudoirreversible binding of plasmin to inhibitor, the tight binding inhibitor titrated the enzyme (17). Indeed, at [E] 0 ϭ 750 nM (Fig. 4, see inset), RG1192 titrated the enzyme with an equivalence point corresponding to the binding of about 20 molecules of plasmin per molecule of RG1192. Similar experiments demonstrated tight binding inhibition of plasmin by RG1503 and a 1:6 inhibitor/plasmin binding stoichiometry (data not shown).
Since these two polymers behave as tight binding inhibitors, the analysis of steady-state velocities data and K i determinations by means of the Dixon (18) and/or Cornish-Bowden (19) representation is unsuitable. We therefore used the complex steady-state rate equation (Equation 3) derived by Szedlacsek et al. (20), which takes into account both the tightly binding behavior of the inhibitors and the residual enzymatic activity of the plasmin/inhibitor complexes,  Table II summarizes the parameters describing the inhibition of plasmin by RG1503 and RG1192. These two compounds behaved as tight binding hyperbolic noncompetitive inhibitors (␣ϭ1, ␤ 0) with K i values in the 30 nM range for RG1192, whereas that for RG1503 is 3.7-fold higher. All measurements reported here were performed in 50mM Tris-HCl buffer containing 50mM NaCl. Increasing the ionic strength by the addition of NaCl resulted in a progressive reduction of the inhibitor potency of all dextran derivatives tested. At physiological ionic strength, only RG1192 retained an inhibitory capacity toward plasmin with 30% of residual enzymatic activity (data notshown). In these conditions, the new binding characteristics of RG1192-plasmin complex were ␣ ϭ 1, ␤ϭ0.3, and K i ϭ 3.1Ϯ0.2ϫ10 Ϫ7 M.
Analysis of Plasmin Binding to RG1192 by Surface Plasmon Resonance-Surface plasmon resonance performed on a BIAcore TM system was used to confirm the interaction between plasmin and RG1192. The biotinylated polymer was immobilized to a streptavidin-bound carboxymethyl dextran surface, and this biosensor chip was used to determine the equilibrium dissociation constant of the plasmin-RG1192 complex. Increasing concentrations of plasmin were injected over the sensor surface, and surface plasmon resonance response values were reached at steady states, representing the equilibrium binding of plasmin to immobilized RG1192 (Fig. 5A). The equilibrium dissociation constant (K d ) was calculated by a nonlinear least square fit of the data to Equation 2, as described under "Experimental Procedures" (Fig. 5B). K d was found to be 6.0 Ϯ 0.3 ϫ 10 Ϫ8 M, which is closed to the enzymatically determined constant (2.8 ϫ 10 Ϫ8 M).
Identification of the Kringle Domains of Plasmin as the RG1192 Binding Sites-The plasmin molecule contains in its A-heavy chain five homologous kringle structures in which LBS are located. Kringles 1, 4, and 5 possess LBS that mediate interactions of plasmin(ogen) with fibrin (21), laminin, and fibronectin (22) as well as ␣ 2 -antiplasmin (23) and possess significant affinity for -amino acids such as 6-aminohexanoic acid (6-AHA) (24,25). Furthermore, it has been previously shown that fibrinogen (26), penicillin (27), and oleic acid (28) modulate plasmin activity by interacting through the LBS modules. We therefore investigated whether the ability of  RG1192 to suppress plasmin activity depends on its interaction with LBS in plasmin. To explore such a possibility, plasmin activity was first measured in the presence of increasing concentrations of 6-AHA and a constant RG1192 concentration known to produce 50% inhibition. As shown in Fig. 6A, increasing concentrations of 6-AHA relieve the inhibition of plasmin activity, and the inhibitory effect of RG1192 was completely abolished at 5 mM 6-AHA. We then analyzed the effect of plasmin-derived kringle fragments on the plasmin inhibition by RG1192. As described above, plasmin activity was measured in the presence of a constant concentration of RG1192, yielding 50% of enzyme inhibition and an increasing excess of kringle 1-3, kringle 4, or recombinant kringle 5. As shown in Fig. 6B, a 300-fold molar excess of isolated kringle 1-3 as well as kringle 5 restored up to 100% of plasmin activity despite the presence of RG1192. In contrast, only a slight restoration of plamin activity was observed in the presence of kringle 4. All of these results indicate that RG1192 mediates its inhibitory effect through binding to the LBS-containing kringles but also suggest that these kringles display different affinities toward RG1192.
Interaction of Plasminogen with RG1192-The experiments mentioned above indicate that RG1192 interacts with the krin-gle domains of plasmin. Since these characteristic structures are equally present within plasminogen, the inactive zymogen of plasmin, we analyzed the capacity of the RG1192 polymer to interact with plasminogen. To investigate such a possibility, we first analyzed the effect of RG1192 on the proteolytic cascade of plasminogen activation. uPA was chosen as the plasminogen activator, since the polymer did not affect its enzymatic activity (Fig. 3). Direct analysis by SDS-PAGE of Glu-plasminogen activation by uPA was documented by the disappearance of the Glu-plasminogen band (95 kDa) and the emergence of the plasmin light B-chain band (25 kDa) and the NH 2 -terminal activation peptide (9 kDa) (Fig. 7, lane 2). The inclusion of increasing concentrations of RG1192 in the activation process impeded the conversion of Glu-plasminogen to plasmin in a concentrationdependent manner (Fig. 7, lanes 3-7). Surprisingly, in the presence of RG1192, a new 85-kDa band was observed instead of the native 95-kDa Glu-plasminogen band, and RG1192 did not reduce the level of the 9-kDa activation peptide. The 85-kDa polypeptide may represent Lys-plasminogen, the shortened form of the inactive protein, generated by trace amounts of active plasmin. These results therefore indicated that the presence of RG1192 in the reaction mixture rendered the cleav-  Fig. 4A was determined by measuring the enzyme velocity in the absence ( 0 ) and in the presence ( i ) of RG1192 for each 6-AHA concentration. All experiments were done in duplicate.
age site Arg 561 -Val 562 of plasminogen inaccessible and/or unrecognizable by uPA but did not prevent the release of the 9-kDa activation peptide. It is noteworthy that RG1192 has no direct effect on the activation of Glu-plasminogen (Fig. 7, lane  9). The effect of RG1192 on the plasminogen activation described above does not, however, clearly establish that RG1192 directly binds plasminogen. To raise this ambiguity, the equilibrium dissociation constant of the RG1192-plasminogen complex was determined. Optical biosensor experiments were then carried out with plasminogen and biotinylated RG1192 (data not shown), as described previously for plasmin. In this case, K d was found to be 3.0 Ϯ 0.3 ϫ 10 Ϫ8 M, indicating that RG1192 binds plasminogen as well as plasmin with a similar affinity.
Effect of RG1192 on the Plasmin-mediated Fibronectin and Laminin Proteolysis-Enzymatically active plasmin is able to hydrolyze extracellular matrix proteins such as fibronectin and laminin (8,9). Use was made of this property to analyze the effect of RG1192 on a more representative substrate of the biological activity of plasmin. Limited proteolysis of fibronectin by plasmin yielded several polypeptide fragments with apparent molecular masses ranging from 150 to 50 kDa (Fig. 8 A,  lane 2). The addition of increasing concentrations of RG1192 to the incubation mixtures completely abolished the proteolytic cleavage of fibronectin in a dose-dependent manner (Fig. 8A,  lanes 3-7). When the same experiment was done with laminin as the natural substrate, RG1192 was also able to prevent its plasmin-mediated proteolysis (Fig. 8B). However, no complete inhibition of this natural substrate against proteolysis was attained even for RG1192 concentrations up to 2.5 M (Fig. 8, lane 6).

DISCUSSION
It was previously shown that some dextran derivatives, which imitate the action of heparin/heparan sulfate, were able in vitro to protect various heparin-binding growth factors against proteolytic degradation (6) as well as to inhibit the enzymatic activity of neutrophil elastase (7). In this study, we report that these biopolymers are also able to inhibit the enzymatic activity of plasmin, another known proteinases involved in extracellular matrix and tissue remodeling. The extent of plasmin inhibition varied with the nature of the substituted chemical groups; the most efficient inhibitors were substituted with carboxymethyl, sulfate as well as benzylamine groups (RG1192); those without benzylamine were less efficient (RG1503), and those containing only carboxymethyl were ineffective (RG1100). Moreover, the natural glycosaminoglycans heparin and heparan sulfate have no effect upon the enzymatic activity of plasmin. Our data as well as those previously reported concerning heparin (29) as well as chondroitin sulfate and dermatan sulfate (30) indicated that sulfated glycosaminoglycans are devoid of significant antiplasmin activity.
To understand the mechanism by which RG1192 affected the catalytic activity of plasmin, competitive experiments were carried out with 6-AHA and kringle proteolytic fragments. The results reveal that a kringle-dependent interaction occurs between RG1192 and plasmin. RG1192 thus binds plasmin at regulatory sites that were clearly distinct from the catalytic site of the enzyme, hence confirming the noncompetitive character of the inhibition, which was originally determined enzymatically. Remarkably, tPA, whose activity was affected by RG1192, but to a lesser extent than plasmin, contained two homologous kringles that exhibited lysine and 6-AHA binding properties (31). Hence, it would be interesting to investigate whether other kringle-containing proteins possessing this characteristic property should be potential ligands of RG1192.
Inhibition of plasmin by RG1503 and RG1192 was of the reversible, tight binding hyperbolic noncompetitive type. Such a mechanism was already described concerning inhibition of elastase and cathepsin G by heparin (32). The plasmin inhibition data could be satisfactorily fitted to Equation 3, which assumes that the inhibition potency depends on the binding stoichiometry (n), the equilibrium dissociation constant (K i ), and the dimensionless numbers ␣ and ␤. The mechanism of plasmin inhibition by the two polymers was substrate-independent (␣ ϭ 1) and proceeded by a decrease of k cat (␤ Ͻ 1). This implies that at the concentration used, the substrate did not significantly affect the affinity of the inhibitor for the enzyme (K i ϭ K i Ј) and that its rate of hydrolysis was lowered by the presence of the polymer. The experimentally determined values of ␤ (Table II)  activity measured at saturating concentrations of inhibitor, thus confirming that ␣ was very close to 1 (in this case, ␤ ϭ ϱ / 0 ; see Equation 5). The best plasmin inhibitor (RG1192) had a K i of 2.8 ϫ 10 Ϫ8 M, a value similar to or lower than those previously described for plasmin synthetic inhibitors (33)(34)(35).
RG1192 and RG1503 thus inhibited plasmin activity according to a similar mechanism, and the deleterious effect of ionic strength on their inhibitor potency indicated that the binding of these two polymers with plasmin involved at least ionic interactions. However, RG1503 and RG1192 differed in their ability to inhibit plasmin activity. First, the plasmin-RG1503 complex had a residual activity of 65% and a K i of 10 Ϫ7 M, while the plasmin-RG1192 complex only possessed 20% residual activity and a 3.7-fold lower K i value. Moreover, the latter complex was less sensitive to ionic strength than the plasmin-RG1503 complex, suggesting that plasmin bound to RG1192 by a combination of electrostatic and nonelectrostatic interactions. These differences between these two complexes can be attributable to the fact that RG1192 contains an additional substitution of benzylamide groups. The notion that these benzylamide groups are involved in the maintenance of the plasmin-RG1192 complex is strengthened by previous results showing that kringle 5 domain of plasminogen possesses a strong affinity for the aromatic ligand benzylamine (36). Since this kringle domain was involved in the plasmin-RG1192 complex, it was tempting to suggest that the substituted benzylamide groups in concert with the negatively charged sulfates may participate in the stabilization of the plasmin-RG1192 complex, perhaps through a kringle 5-dependent interaction. In contrast, RG1503, which did not possess these aromatic substitutions, formed a loose complex with plasmin that was easily disrupted by an increase of ionic strength. Furthermore, the observation that chemically modified N-oleoyl heparin is an effective inhibitor of plasmin (30) reinforces the notion that hydrophobic groups such as benzylamine are requisite structures for the biological activity of plasmin inhibitors.
Dextrans derivatized such as RG1192 were shown to promote tissue remodeling in various in vivo models (1)(2)(3)(4)(5). Based on their in vitro properties, we proposed that these biopolymers may act as survival and protective agents through the maintenance and protection of bioavailability of growth factors. Further data derived from the ischemic muscle model revealed that the regeneration process induced by these dextran derivatives was associated with the maintenance of the basement membrane integrity of degenerated muscle fibers. 2 This suggests that these polymers could also protect basement membrane components from proteolysis mediated by proteinases activated during the inflammatory process. To support this hypothesis, we studied the in vitro effect of RG1192 on the plasmin-mediated proteolysis of fibronectin and laminin, the major noncollagenous components of extracellular matrix and basement membranes. SDS-PAGE analyses followed by immunoblotting assays revealed that plasmin-catalyzed degradation of both fibronectin and laminin was prevented by RG1192. It is noteworthy that as for RG1192, the interaction of plasmin(ogen) with these two proteins is mediated via the LBS-containing kringle domains of plasmin(ogen) (22). In light of this observation, it is tempting to speculate that RG1192 may prevent the plasmin-induced proteolysis of laminin and fibronectin through direct inhibition of plasmin catalytic activity and/or through competition for the substrate binding site in plasmin. Evidence is provided here for an in vitro protective effect of RG1192 against plasmin-catalyzed degradation of matrix proteins. Nevertheless, further experiments are still necessary to clearly establish the linkage between the observed in vivo protective effect of this biopolymer and the inhibition of proteinase activities in vitro.
In this study, we provide in vitro evidence that some dextran derivatives could contribute to the regulation of plasmin activity by impeding the plasmin generation, as a result of their binding to plasminogen and also by directly affecting the catalytic activity of the enzyme. Interestingly, only RG1192 retained an inhibitory capacity toward plasmin when tested under physiological ionic strength (K i ϭ 3 ϫ 10 Ϫ7 M). Since RG1192 and RG1503 both present in vivo tissue repair activity, our data indicate that plasmin inhibition is not involved as a unique or key mechanism of action of these compounds. However, due to the very high complexity and the numerous factors involved in tissue healing, the use of various biopolymers with specific properties such as RG1192 and RG1503 may provide the tools required for a better understanding of the woundhealing property of these molecules.