Hypersulfated low molecular weight heparin with reduced affinity for antithrombin acts as an anticoagulant by inhibiting intrinsic tenase and prothrombinase.

In buffer systems, heparin and low molecular weight heparin (LMWH) directly inhibit the intrinsic factor X-activating complex (intrinsic tenase) but have no effect on the prothrombin-activating complex (prothrombinase). Although chemical modification of LMWH, to lower its affinity for antithrombin (LA-LMWH) has no effect on its ability to inhibit intrinsic tenase, N-desulfation of LMWH reduces its activity 12-fold. To further explore the role of sulfation, hypersulfated LA-LMWH was synthesized (sLA-LMWH). sLA-LMWH is not only a 32-fold more potent inhibitor of intrinsic tenase than LA-LMWH; it also acquires prothrombinase inhibitory activity. A direct correlation between the extent of sulfation of LA-LMWH and its inhibitory activity against intrinsic tenase and prothrombinase is observed. In plasma-based assays of tenase and prothrombinase, sLA-LMWH produces similar prolongation of clotting times in plasma depleted of antithrombin and/or heparin cofactor II as it does in control plasma. In contrast, heparin has no effect in antithrombin-depleted plasma. When the effect of sLA-LMWH on various components of tenase and prothrombinase was examined, its inhibitory activity was found to be cofactor-dependent (factors Va and VIIIa) and phospholipid-independent. These studies reveal that sLA-LMWH acts as a potent antithrombin-independent inhibitor of coagulation by attenuating intrinsic tenase and prothrombinase.

protein found on nonvascular cells, is exposed by vascular injury. Tissue factor binds factor VII/VIIa (f.VII/VIIa), 1 forming the extrinsic tenase complex that activates f.X (1). f.Xa generates sufficient thrombin through prothrombinase, the phospholipid membrane-bound complex of f.Xa and f.Va, to induce local aggregation of platelets and activate f.V and f.VIII (2). f.Xa generated via extrinsic tenase is insufficient to sustain hemostasis because tissue factor pathway inhibitor rapidly inactivates tissue factor-bound f.VIIa in a f.Xa-dependent fashion (1,3). To overcome this limitation, additional f.Xa is generated by intrinsic tenase, the phospholipid membrane-bound complex of f.IXa and f.VIIIa. f.IX can be activated by extrinsic tenase or by f.XIa, generated by thrombin cleavage of f.XI (1).
The critical role of intrinsic tenase and prothrombinase in coagulation makes these enzyme complexes attractive targets for inhibition. Prothrombinase and intrinsic tenase share similar properties, with each complex consisting of a vitamin Kdependent serine protease and a nonproteolytic cofactor protein. The reactions are calcium-dependent and require a negatively charged phospholipid surface for optimal expression of activity (4 -7).
Heparin and low molecular weight heparin (LMWH) act as anticoagulants by activating antithrombin, which inactivates f.Xa and thrombin (8). In buffer systems, heparin and LMWH also inhibit intrinsic tenase activity in an antithrombin-independent fashion (9,10). In plasma systems, however, the antithrombin-dependent anticoagulant effects of heparin and LMWH predominate.
The purpose of this study was to investigate methods for modifying heparin so as to maximize its antithrombin-independent effects. Starting with a size-restricted LMWH to capitalize on its decreased propensity to bind to plasma proteins, a property that endows it with pharmacokinetic advantage over unfractionated heparin (11,12), LMWH was chemically modified to reduce its affinity for antithrombin 1700-fold (from a K d value of 25 nM to 43 M) by periodate oxidation (13). Like LMWH, this low affinity LMWH (LA-LMWH), which we termed Vasoflux, inhibited intrinsic tenase but had no effect on prothrombinase, thereby confirming that its ability to inhibit intrinsic tenase is not dependent on its affinity for antithrombin. When LA-LMWH was N-desulfated, however, most of its activity was lost, suggesting that its ability to inhibit intrinsic tenase is charge-dependent. To explore this possibility, LA-LMWH was progressively hypersulfated, and the inhibitory activities of these hypersulfated LA-LMWH (sLA-LMWH) compounds against intrinsic tenase and prothrombinase were examined. Herein, we demonstrate that upon sulfation, LA-LMWH becomes a more potent inhibitor of intrinsic tenase and acquires the ability to inhibit prothrombinase.

Materials
Human f.V, f.Va, and f.IXa were obtained from Hematologic Technologies Inc. (Essex Junction, VT), whereas f.X, f.Xa, prothrombin, and ␣-thrombin were obtained from Enzyme Research Laboratories (South Bend, IN). Recombinant f.VIII (Kogenate) was from Bayer Inc. (Etobicoke, Canada). Albumin-free plasma-derived f.VIII, a gift from Dr. E. Saenko (Holland Laboratory, American Red Cross, Rockville, MD), was used in some experiments. The purity of f.V and f.VIII preparations were confirmed by SDS-polyacrylamide gel electrophoresis analysis (14).
A series of low affinity low molecular weight heparins (LA-LMWH) and hypersulfated LA-LMWH derivatives (sLA-LMWH) were used in this study (see Table I). A LMWH fraction (mean molecular weight 5000) was prepared from unfractionated heparin by nitrous acid depolymerization, and its affinity for antithrombin was reduced by sodium periodate oxidation, as previously described (13). The resultant LA-LMWH was subjected to ultrafiltration using a 3000-Da cut-off membrane and lyophilized. The LA-LMWH was O-sulfated using a modification of the method described by Nagasawa et al. (18). Briefly, 200 mg of LA-LMWH dissolved in 5 ml of water was converted into acid form by ion exchange at 0°C. After neutralization with tributylamine and lyophilization, the sample was dissolved in 20 ml of dry N,N-dimethylformamide. With constant stirring, 2 g of trimethylamine sulfur trioxide was added, and the mixture was incubated for varying times and temperatures to obtain the hypersulfated LA-LMWH derivatives (sLA-LMWH) listed in Table I. At the end of each reaction, 50 ml of double distilled water was added, and the solution was dialyzed three times (12 h each) against 1000 ml of 5% NaCl and then three times against 1000 ml of double distilled water. After lyophilization, each sample was subjected to elemental analysis, and the number of sulfate residues/ disaccharide was calculated based on the percentage of carbon relative to sulfur. Five sLA-LMWH derivatives were prepared and are designated sLA-LMWH-S1 to -S5. LA-LMWH was N-desulfated (N-DS-LA-LMWH) using the solvolytic desulfation method described by Inoue and Nagasawa (19).

Methods
Effect of Glycosaminoglycans on the Activity of Prothrombinase-To examine the effects of glycosaminoglycans on the activation of prothrombin by the prothrombinase complex, the rate of thrombin generation was assayed using a modification of the method of Barrow et al. (9). Reactions were performed in 20 mM Tris-HCl, pH 7.4, 150 mM NaCl containing 0.1% polyethylene glycol (TSP buffer). Two times concentrated stock solution A was made in TSP buffer so as to give final reaction concentrations of 24 M PCPS vesicles, 0.24 nM f.Va, 1.2 M prothrombin, and 4 mM CaCl 2 . 50 l of stock A was mixed with 30 l of TSP and 10 l of glycosaminoglycan at final concentrations ranging from 1-1000 g/ml. Reactions were initiated by addition of 10 l of 1 nM f.Xa. Control samples lacking glycosaminoglycan were run in parallel. 10-l aliquots were removed at 30-s intervals into the wells of a 96-well microtitre plate containing 10 l of 10 mM EDTA, pH 7.4, to quench the activation reaction. At the end of the time course, the activity of thrombin in each well was determined by adding 190 l of 200 M tGPR-pNA containing 0.1 mg/ml polybrene. Chromogenic substrate hydrolysis was monitored at 405 nm at 10-s intervals for 5 min at 23°C in a Spectramax 340 plate reader (Molecular Devices, Sunnyvale, CA).
Effect of Glycosaminoglycans on the Activity of Intrinsic Tenase-The effects of glycosaminoglycans on the activity of intrinsic tenase were determined in a similar fashion. Two times concentrated stock B was prepared so as to contain final reaction concentrations of 300 nM f.X, 4 mM CaCl 2 , 24 M PCPS, 4 nM f.IXa, and 0.4 nM f.VIII. f.VIII was activated to f.VIIIa in stock B by incubation with 10 nM thrombin for 1 min. 50 l of stock B was mixed with 30 l of TSP containing glycosaminoglycan at final concentrations ranging from 1 to 1000 g/ml. Reactions in each well were initiated by the addition of 10 l of 8 nM f.IXa, and 10-l aliquots were removed at 30-s intervals as described above. f.Xa activity was assayed using S-2222 or Chromozym.X.
Effect of sLA-LMWH-S5 on Components of the Prothrombinase Complex-Individual components of the prothrombinase complex were systematically removed or substituted to examine their susceptibility to inhibition by sLA-LMWH. In all cases, the rate of prothrombin activation was examined in the absence or presence of sLA-LMWH-S5 (S5), the most potent of the sLA-LMWH derivatives. Data Analysis-Rates of chromogenic substrate cleavage were obtained from the slope of the linear portion of the A 405 versus time plot. Standard curves of thrombin or f.Xa activity, with the respective chromogenic substrate, were used to convert slopes to enzyme concentration. The rates of activation were calculated by linear regression analysis of thrombin or factor Xa concentration versus time (Quattro Pro, version 5.0, Borland International Inc. Scott's Valley, CA). Comparison with control reactions performed in the absence of glycosaminoglycan, run in parallel, gave a rate relative to the control. The concentration of glycosaminoglycan that produced 50% inhibition of the f.X or prothrombin activation rate (IC 50 ) was calculated to compare the different glycosaminoglycans. All reactions were performed in duplicate and means from three to six experiments were calculated for each analysis.
Activated Partial Thromboplastin Time-Glycosaminoglycan was added to 50-l aliquots of control plasma, plasma immunodepleted of antithrombin, or plasma immunodepleted of both antithrombin and heparin cofactor II (Affinity Biologicals Inc., Hamilton, Canada). 50 l of Thrombosil (Ortho Clinical Diagnostics, Raritan, NJ) was added, and the sample was incubated for 5 min. Clotting was initiated with 50 l of 20 mM CaCl 2 , and the time to clot formation was determined on a ST4 coagulometer (Stago Diagnostica, Parsippany, NJ).
f.Xa Clotting Time-100 l of 1 nM f.Xa in 10 mM CaCl 2 , 0.1 mg/ml bovine serum albumin, 0.05% cephalin, 10 mM Hepes-NaOH, pH 6.8 was added to 50 l of plasma containing varying concentrations of glycosaminoglycan, and the time for clot formation was monitored using an ACL 30000 centrifugal coagulation analyzer (Beckman-Coulter Inc., Fullerton, CA). Assays were performed in normal plasma, plasma immunodepleted of antithrombin, or plasma immunodepleted of both antithrombin and heparin cofactor II.
Binding Studies-Fluorescein labeling was performed by incubating 10 mg of S5 with 15 mg of 5Ј-fluorescein isothiocyante (Sigma) in 2.5 ml of 1.0 M Na 2 CO 3 , pH 9.0, for 5 h at 23°C. After centrifugation at 13,000 ϫ g for 5 min, 1 ml of the supernatant was applied to duplicate PD-10 columns (Millipore Corp., Bedford, MA), equilibrated with H 2 O, and eluted with H 2 O under gravity. 0.5-ml fractions were collected, frozen, and lyophilized. Recovered material was pooled, weighed, and dissolved in 20 mM Tris-HCl, 150 mM NaCl, pH 7.4 (TS), to a concentration of 10 mg/ml.
The affinity of fluorescein-sLA-LMWH-S5 (Fl-S5) for f.IXa and f.Xa was determined by monitoring fluorescence of Fl-S5 upon titration with f.IXa or f.Xa. 50 nM of Fl-S5 in TS containing 2 mM CaCl 2 was stirred with a microstirbar in a 1-ml cuvette and maintained at 23°C with a circulating waterbath. Fluorescence was monitored in an LS50B luminescence spectrophotometer (Perkin Elmer, Norwalk, CT) in time drive with ex of 492 nm, em of 535 nm (slit widths of 5 and 20 nm, respectively), and a 515-nm cut-off filter in the emission beam, as 1-10-l aliquots of f.IXa or f.Xa were added to the cuvette. Additions were made at 1-2-min intervals, allowing the fluorescence signal to stabilize between additions. At the end of the titration, intensity values were determined from the time drive profiles. The ratio I/I o was calculated by dividing the fluorescence intensity after each addition (I) by the initial intensity (I o ). Plots of I/I o versus titrant concentration were analyzed by nonlinear regression fit to a binding isotherm equation, which yielded values for K d and maximal fluorescence change (20).
The affinities of other heparin derivatives for f.IXa and f.Xa were determined in competition experiments by monitoring displacement of Fl-S5 from each enzyme. A solution of 50 nM Fl-S5 and 80 nM f.IXa or 100 nM f.Xa was titrated with unlabeled S5 or LA-LMWH. Using I max , the change in fluorescence intensity between fully bound and unbound Fl-S5, the corrected intensity values after each addition were used to calculate the concentration of bound fluorescein-S5 (PL 1 ) using I/I max ϫ [Fl-S5] i , where [Fl-S5] i is the initial concentration of the fluorophore. Likewise, the free enzyme concentration (P) was determined using the equation, K d ϫ I/(I max Ϫ I), where K d is the dissociation constant determined above. By the mass action relationship, PL 2 , the concentration of the unlabeled heparin/enzyme complex, is given by P o Ϫ PL 1 Ϫ P, where P o is the total concentration of the enzyme. A plot of L 2 /PL 2 versus 1/P has a slope of K i , the dissociation constant for the unlabeled heparin (L 2 ) with the enzyme.
Statistical Analysis-Values for IC 50 were calculated based on the means of three to six experiments, each performed in duplicate. Means and standard deviations were determined using Quattro Pro. The correlation between the degree of sulfation of LMWH and mean IC 50 values for prothrombinase and intrinsic tenase was determined using a test of rank correlation (Pearson) and by one-way analysis of variance using Minitab (Release 11.11, Minitab Inc., State College, PA). A value of p Ͻ 0.05 was considered statistically significant.

Effect of Glycosaminoglycans on Intrinsic
Tenase-To examine the antithrombin-independent effect of LMWH on intrinsic tenase, a LA-LMWH was prepared as previously described (13). LA-LMWH produced 50% inhibition of the initial velocity of f.X activation (IC 50 ) of 16.3 Ϯ 6.1 g/ml (Fig. 1A). A commercial LMWH (enoxaparin) with normal antithrombin affinity, inhibited intrinsic tenase with an IC 50 value of 13.2 Ϯ 7.7 g/ml (Table I), a value comparable with the IC 50 value of 6 g/ml reported for LMWH by other investigators (9). To investigate the influence of sulfation of LA-LMWH on inhibition of tenase activity, LA-LMWH was N-desulfated by solvolysis. N-DS-LA-LMWH had 12-fold lower inhibitory activity, with an IC 50 value of 166 Ϯ 25 g/ml. These findings suggest that the ability of LMWH to inhibit intrinsic tenase is independent of its affinity for antithrombin but dependent on its charge.
To further investigate the importance of charge, progressively hypersulfated LA-LMWH derivatives were synthesized, and their inhibitory activity was compared with that of the starting material (Table I). The most highly sulfated LA-LMWH, designated S5, was 32-fold more potent than LA-LMWH, inhibiting intrinsic tenase with an IC 50 value of 0.47 Ϯ 0.2 g/ml. As illustrated in Fig. 1A, increasing the sulfation of LA-LMWH produces a progressive reduction in IC 50 values. When a plot of the number of sulfate residues/disaccharide versus IC 50 is subjected to regression analysis (not shown), the correlation coefficient is Ϫ0.86, a value that on one-way anal-ysis of variance is highly significant (p Ͻ 0.001), supporting the concept that the potency of LMWH derivatives is dependent on their degree of sulfation. A dextran sulfate, which contained 3.9 sulfate residues/disaccharide, inhibited intrinsic tenase with an IC 50 value of 0.4 g/ml, a value similar to that of S5. This observation provides further evidence that the extent of sulfation is an important determinant of potency in this system.
Effect of Glycosaminoglycans on Prothrombinase-The inhibitory effect of the series of hypersulfated LA-LMWH deriv-

sLA-LMWH Inhibits Intrinsic Tenase and Prothrombinase
atives on prothrombinase activity was examined to determine whether increased inhibitory activity against intrinsic tenase conferred inhibitory properties against prothrombinase (Fig.  1B). At concentrations up to 1000 g/ml, unfractionated heparin had no effect on prothrombinase function, consistent with the results of Barrow et al. (9). Likewise, neither enoxaparin nor LA-LMWH had inhibitory activity against prothrombinase at these concentrations. In contrast, sLA-LMWH inhibited prothrombinase in a concentration-dependent fashion, and its inhibitory activity increased with progressive sulfation, as reflected by a reduction in IC 50 values (Table I). Maximum inhibition was effected by S5, which inhibited prothrombinase with an IC 50 value of 30 Ϯ 16 g/ml. Linear regression analysis of a plot of the number of sulfate residues/disaccharide versus IC 50 values yielded a correlation coefficient of Ϫ0.92 (not shown), which on one-way analysis of variance was highly significant (p Ͻ 0.001). Dextran sulfate inhibited prothrombinase with an IC 50 value of 35 g/ml, further highlighting the importance of sulfation for expression of this activity. The IC 50 values against prothrombinase were 2 orders of magnitude higher than those for intrinsic tenase, indicating that all of the hypersulfated carbohydrates have greater inhibitory activity against intrinsic tenase than prothrombinase.
Mode of Disruption of Prothrombinase and Intrinsic Tenase by S5-As the intrinsic tenase and prothrombinase complexes are composed of multiple components, various reactants could serve as targets for modulation by sLA-LMWH. To reveal the susceptible component(s), the influence of S5 on partially reconstituted activation complexes was examined. S5 was used in these experiments because it exhibited the most potent inhibitory effects (Table I).
When all components were present, S5 produced dose-dependent inhibition of both intrinsic tenase (Fig. 2) and prothrombinase (Fig. 3). Similar inhibitory effects were evident in assays devoid of phospholipid. Thus, the IC 50 values for S5 on intrinsic tenase in the presence or absence of phospholipid were 0.47 Ϯ 0.2 and 1.0 Ϯ 0.7 g/ml, respectively. Comparable IC 50 values for inhibitory activity of S5 against prothrombinase also were found in the presence or absence of phospholipid (30 Ϯ 16.3 and 68 Ϯ 22 g/ml, respectively). In contrast, in systems devoid of cofactor (f.Va/f.VIIIa), S5 had no inhibitory effect, suggesting that S5 interferes with the cofactor activity of f.Va and f.VIIIa in their respective enzyme complexes. S5 also had no inhibitory activity in prothrombinase systems devoid of cofactor and phospholipid or of cofactor, phospholipid, and calcium. For intrinsic tenase, despite increases in reactant concentrations, rates of activation remained slow in a system lacking f.VIIIa and PCPS and were unmeasurable in a system devoid of f.VIIIa, PCPS, and calcium (data not shown). The data obtained in partially reconstituted systems support the concept that the cofactor-enzyme interaction is the predominant target of S5 in both prothrombinase and intrinsic tenase. When f.V or f.VIII was substituted for f.Va or f.VIIIa, an initial lag phase was seen on plots of thrombin or f.Xa generation versus time, reflecting the positive feedback effect of thrombin or f.Xa on f.V or f.VIII activation. When the linear part of the curve was analyzed to give an apparent rate of activation, S5 had a similar inhibitory effect in systems using f.V and f.VIII as it did in those using their activated counterparts, suggesting that S5 has no effect on cofactor activation.
The effect of S5 on the chromogenic activity of f.Xa and f.IXa also was examined. The hydrolysis of S-2222 by f.Xa is unaffected by S5. Although S5 produced some inhibition of f.IXamediated hydrolysis of Pefa IXa, this assay is of limited value because f.IXa has almost no activity against chromogenic amide substrates (21)(22)(23).
Affinity of LA-LMWH and sLA-LMWH for f.IXa and f.Xa-To begin to explore why S5 has greater activity against intrinsic tenase than prothrombinase, its affinity for f.IXa and f.Xa was determined and compared with that of LA-LMWH, which only has inhibitory activity against intrinsic tenase. Binding affinities were determined from competition experiments where fluorescein-labeled sLA-LMWH-S5 was displaced from f.IXa or f.Xa by unlabeled S5 or LA-LMWH (Fig. 4). LA-LMWH displaced Fl-S5 from f.IXa and f.Xa with K i values of 1000 and 9300 nM, respectively. In contrast, S5 displaced Fl-S5 from f.IXa and f.Xa with K i values of 115 and 555 nM, respectively, consistent with its more potent inhibitory effect. The greater than 5-fold higher affinity of S5 for f.IXa relative to f.Xa could explain why S5 inhibits intrinsic tenase more effectively than prothrombinase.
Effect of S5 on Coagulation Assays-To determine whether the inhibitory activity of S5 observed in buffer systems also occurs in plasma systems, the effects of S5 on the APTT and f.Xa clotting time were examined (Fig. 5). The APTT was used as a measure of both intrinsic tenase and prothrombinase,

sLA-LMWH Inhibits Intrinsic Tenase and Prothrombinase
whereas the f.Xa clotting time was used as an index of prothrombinase activity. To verify that S5 was acting independent of plasma protease inhibitors, its anticoagulant activity in plasma fractions immunodepleted of antithrombin and/or of both antithrombin and heparin cofactor II was compared with that in control plasma. S5 produces concentration-dependent prolongation of the APTT and f.Xa clotting time in control plasma, plasma immunodepleted of antithrombin, and plasma depleted of both antithrombin and heparin cofactor II. These data confirm that S5 inhibits coagulation in an antithrombinand heparin cofactor II-independent manner and are consistent with its inhibitory activity against intrinsic tenase and prothrombinase. In contrast, concentrations of heparin that produce similar prolongations of the APTT and f.Xa clotting time in control plasma have no effect in either immunodepleted plasma, indicating that the anticoagulant effects of heparin are antithrombin-dependent. DISCUSSION Intrinsic tenase and prothrombinase complexes are critical for thrombin generation in the process of blood coagulation and, as such, are ideal targets for inhibitors of blood coagulation. Because these multicomponent complexes are assembled from intrinsic and activated components, several approaches can be used for their inhibition. These include direct inactivation of the enzyme or cofactor or disruption of the capacity of the complex to assemble productively. Our results demonstrate that unfractionated heparin and LMWH directly inhibit intrinsic tenase, consistent with the results of Barrow et al. (9). However, we have extended their findings in two important ways. First, we demonstrate that reducing the affinity of LMWH for antithrombin does not affect its ability to inhibit intrinsic tenase. Second, we demonstrate that progressive hypersulfation of LA-LMWH increases its potency of inhibition of intrinsic tenase and enables inhibition of prothrombinase. This endows sLA-LMWH with greater activity because it acts at two critical sites in the coagulation system.
Mechanism-The kinetics of f.X activation in the presence of unfractionated heparin have established that heparin inhibits intrinsic tenase in a noncompetitive fashion (9). The data reported here demonstrate that inhibition of the activation complexes involves the cofactor and enzyme but not the phospholipid surface. Because intrinsic tenase and prothrombinase are homologous complexes, it is conceivable that the site of action of sLA-LMWH is the same for both systems. Our data suggest that sLA-LMWH disrupts the interaction of the enzyme with its cofactor, inferring that one or both of these components bind sLA-LMWH. That heparin binds to f.IXa is well known (24 -26). Likewise, calcium-dependent heparin binding to f.Xa has recently been described (27). The competitive binding studies reported here reveal that S5 binds over 5-fold more tightly to f.IXa than to f.Xa, consistent with the greater inhibitory activity of all glycosaminoglycans tested against tenase rather than prothrombinase. These results suggest that the enzymes within the activation complexes represent at least part of the target for glycosaminoglycan binding and subsequent disruption of complex assembly. The binding site for heparin on f.IXa has not been formally identified but is presumed to be at a site homologous to the heparin-binding domain of f.Xa and thrombin because arrangements of basic residues in the protease domains are largely preserved (21, 28 -31). It is notable that the putative heparin-binding sites on f.Xa and f.IXa also comprise respective f.Va and f.VIIIa binding sites (32)(33)(34)(35). In addition to the enzyme component of the activation complexes, f.V and f.VIII also bind heparin, providing additional targets for interference by glycosaminoglycans (10, 36).

sLA-LMWH Inhibits Intrinsic Tenase and Prothrombinase
Although previous studies also suggested that the interaction of f.IXa and f.VIIIa were the likely target for glycosaminoglycan inhibition, direct binding studies did not reveal disruption of binding of the two factors on phospholipid surfaces (9). One possible explanation for this paradox is that the glycosaminoglycan may interfere with interaction of the substrate (either f.X or prothrombin) with an exosite that is only expressed on the enzyme in the presence of its cofactor. The existence of such an exosite, or alternate substrate binding mode, has been suggested for prothrombinase (37)(38)(39) and tenase (33, 40 -42). Further studies will be necessary to reveal the precise mechanism by which glycosaminoglycans inhibit these complexes.
Heparin also directly inhibits activation of the f.VIII-von Willebrand factor complex by thrombin (10). This may reflect competition between heparin and f.VIII for thrombin binding because, in addition to exosite 1, exosite 2, the heparin-binding domain, has also been implicated in f.VIII recognition by thrombin (43). Although inhibition of f.VIII activation by heparin may cause additional attenuation of intrinsic tenase activity, it is not the predominant mechanism because sLA-LMWH had the same effect regardless of whether f.VIII or f.VIIIa was used.
Role of Sulfation-A series of variably sulfated LA-LMWH derivatives was used to investigate the structural requirements for inhibition of intrinsic tenase so that more potent inhibitors could be identified. In our assay of intrinsic tenase activity, LA-LMWH produced 50% inhibition of the initial velocity of fX activation at 16.3 g/ml. A LMWH with normal antithrombin affinity inhibited intrinsic tenase to a similar extent, with an IC 50 value of 13.2 g/ml (Table I). In contrast, N-DS-LMWH had less inhibitory effect, with a 12-fold increase in the IC 50 value to 166 g/ml. These findings indicate that the inhibition of intrinsic tenase by LMWH is independent of the antithrombin-binding pentasaccharide sequence. That the inhibitory activity is dependent on the charge of the glycosaminoglycan is supported by the observation that the potency for inhibition of intrinsic tenase and prothrombinase by LA-LMWH and sulfated derivatives thereof correlates with the number of sulfate residues/disaccharide. Furthermore, a dextran sulfate whose sulfate content is equivalent to that of S5 has similar inhibitory activity against intrinsic tenase and prothrombinase, indicating that this activity does not require the heparin backbone. In contrast, Barrow et al. (9) demonstrated that less sulfated carbohydrates such as dermatan sulfate, chondroitin sulfate, or keratin sulfate inhibit intrinsic tenase with IC 50 values more than 2 orders of magnitude higher than those of heparin or LMWH.
Whereas the reaction of heparin with antithrombin is dependent on charge and a specific arrangement of saccharides (44,45), interaction of heparin with thrombin is a charge-dependent phenomenon that does not involve specific saccharide sequences (46). Because sulfation of heparin increases its affinity for thrombin (47), a similar effect would be anticipated for glycosaminoglycan interactions with the homologous enzymes of the tenase and prothrombinase complex. This is what is observed because the affinities of S5 for f.IXa and f.Xa are 8and 16-fold higher, respectively, than those of the LA-LMWH starting material. Therefore, the increased potency of sulfated glycosaminoglycans in disrupting the activation complexes is consistent with increased affinity for the enzyme and/or the cofactor.
The results obtained in the present study are similar to those reported for DHG, a depolymerized glycosaminoglycan from the sea cucumber (48). DHG was reported to inhibit both tenase and prothrombinase activities in a dose-dependent manner that involved the respective cofactors. It was observed that both f.IXa and f.VIIIa bound to immobilized DHG. However, DHG also accelerated thrombin inhibition by heparin cofactor II (49). DHG displays a higher molecular weight than S5 (12, 500, and 5000, respectively) and a lower degree of sulfation (2.6 and 3.9 sulfate residues/disaccharide, respectively) (50). A phosphorothioate oligonucleotide that inhibits intrinsic tenase and activates heparin cofactor II has recently been described (36). Our observation that S5 retains its anticoagulant activity in plasma immunodepleted of antithrombin and heparin cofactor II indicates that serpin activation is not the means by which S5 exerts its inhibitory effects on coagulation.
As a potent inhibitor of intrinsic tenase and prothrombinase, sLA-LMWH inhibits coagulation in a novel fashion. sLA-LMWH can be added to a growing list of enzyme complex inhibitors, which include active site blocked f.Xa and f.IXa (51,52), f.IXa antibodies (53), and active site directed f.Xa inhibitors (54 -57). sLA-LMWH may have advantages over other agents because it simultaneously attenuates f.Xa and thrombin generation, with selectively greater inhibition of f.Xa generation by intrinsic tenase.