Specific regulation of procoagulant activity on monocytes. Intrinsic pathway inhibition by chondroitin 4,6-disulfate.

Hypercoagulability of blood, monocytic infiltration, and changes in pericellular and extracellular matrix glycosaminoglycans (GAGs) are observed in atherosclerosis, inflammation, and neoplasia. In the present studies, monocyte procoagulants and different GAGs including chondroitin sulfate (CS) A, CSB, CSC, CSD, CSE, and heparan sulfate, were tested either in clotting assays with whole plasma or in chromogenic assays with purified coagulation proteases. Procoagulant activity in plasma was inhibited by three of the seven GAGs, including heparan sulfate, CSE, and CSB. In contrast, activity of purified coagulation protease was inhibited only by CSE, and the inhibition was observed with intrinsic (factor VIIIa/IXa) but not extrinsic (tissue factor/factor VII) components. Reciprocal titration experiments with enzyme and substrate and Scatchard type analyses were consistent with concentration-dependent inhibitory interactions between CSE and sites on both factor VIIIa and IXa. On purified phospholipids, CSE concentration resulting in half-maximal inhibition (K) was 5 ng/ml for interaction with factor IXa and >500 ng/ml for interaction with factor VIIIa. The K values were lower for reactions on purified lipid than for reactions on monocyte surfaces and for reactions on resting than on endotoxin-stimulated monocytes. Experiments with CSE oligosaccharides of defined size indicated that the smallest CSE fragment capable of inhibitory activity was composed of 12-18 monosaccharide units. Collectively, these results indicate that factor X-activating reactions are inhibited by GAGs expressed on monocyte membranes. Inhibition is specific with respect to the structure of both the GAG and the activating protease. Lack of inhibition by added CSA, CSB, and CSC in contrast to CSE strongly suggests a direct role of 4,6-di-O-sulfated N-acetylgalactosamine GAG structures in the inhibition of intrinsic pathway protease. These findings also suggest potential pharmacologic use of CSE as specific anticoagulant in the management of prothrombotic states mediated by intrinsic pathway coagulation reactions.

Hypercoagulability of blood, monocytic infiltration, and changes in pericellular and extracellular matrix glycosaminoglycans (GAGs) 1 are observed in atherosclerosis, inflammation, and neoplasia. In the present studies, monocyte procoagulants and different GAGs including chondroitin sulfate (CS) A, CSB, CSC, CSD, CSE, and heparan sulfate, were tested either in clotting assays with whole plasma or in chromogenic assays with purified coagulation proteases. Procoagulant activity in plasma was inhibited by three of the seven GAGs, including heparan sulfate, CSE, and CSB. In contrast, activity of purified coagulation protease was inhibited only by CSE, and the inhibition was observed with intrinsic (factor VIIIa/IXa) but not extrinsic (tissue factor/ factor VII) components. Reciprocal titration experiments with enzyme and substrate and Scatchard type analyses were consistent with concentration-dependent inhibitory interactions between CSE and sites on both factor VIIIa and IXa. On purified phospholipids, CSE concentration resulting in half-maximal inhibition (K i ) was 5 ng/ml for interaction with factor IXa and >500 ng/ml for interaction with factor VIIIa. The K i values were lower for reactions on purified lipid than for reactions on monocyte surfaces and for reactions on resting than on endotoxin-stimulated monocytes. Experiments with CSE oligosaccharides of defined size indicated that the smallest CSE fragment capable of inhibitory activity was composed of 12-18 monosaccharide units. Collectively, these results indicate that factor X-activating reactions are inhibited by GAGs expressed on monocyte membranes. Inhibition is specific with respect to the structure of both the GAG and the activating protease. Lack of inhibition by added CSA, CSB, and CSC in contrast to CSE strongly suggests a direct role of 4,6-di-Osulfated N-acetylgalactosamine GAG structures in the inhibition of intrinsic pathway protease. These findings also suggest potential pharmacologic use of CSE as specific anticoagulant in the management of prothrombotic states mediated by intrinsic pathway coagulation reactions.
Mononuclear phagocytes constitute one of the major components in the cellular infiltrates that characterize atherosclerotic, neoplastic, and chronic inflammatory lesions (Ross, 1993;McGee et al., 1978;Rickles et al., 1988). These lesions are also associated with localized blood coagulation reactions and both qualitative and quantitative changes in GAGs 1 content (Wagner and Salisbury, 1989;McGee et al., 1990;Wilcox et al., 1989;Dietrich et al., 1993).
Monocytes and macrophages can accelerate membrane-dependent coagulation reactions via expression of TF (tissue factor) and membrane assembly sites for coagulation protease complexes. Membrane TF interacts with plasma factor VII with high affinity (K d , approximately 10 Ϫ9 M), forming functional protease complexes that cleave plasma coagulation factors IX and X and generate the serine esterases, factors IXa and Xa (Bach et al., 1986). Factor IXa in turn assembles on acidic phospholipids with coagulation factor VIIIa (K d , approximately 10 Ϫ9 ) to form the "intrinsic pathway" protease that also generates factor Xa (Ahmad et al., 1989). Thus, this protease complex constitutes a seemingly redundant amplifying loop for activation of factor X to Xa. However, both the extrinsic and intrinsic coagulation proteases are essential for normal hemostasis, as evidenced by the coagulation deficiencies exhibited by patients with single genetic defects in either plasma factor VII, VIII, or IX. (Brigg and Nossel, 1961;Malar et al., 1982).
Chondroitin sulfates are the predominant GAGs comprising the proteoglycans produced by monocyte/macrophages (Owens and Wagner, 1992;Kolset et al., 1983;McQuillan et al., 1989;Uhlin-Hansen et al., 1989;Edwards et al. 1995). Transition from monocyte to macrophage upon exposure to differentiating stimuli is often accompanied by the appearance on the membrane and pericellular spaces of an oversulfated chondroitin sulfate containing 4,6-di-O-sulfated N-acetylgalactosamine residues (i.e. CSE). Whereas previous studies address the importance of CSE during cell differentiation, no information is available on the effects of this unique GAG on procoagulant activity. In the present study we demonstrate that CSE specifically modulates the rate of factor X-activating reaction on monocyte/macrophage membranes.

MATERIALS AND METHODS
Cells and Procoagulant Lipid Surfaces-Human blood monocytes, the human monocytoid cell line THP-1, and human brain homogenates were used as sources of natural procoagulant surfaces. Lipid extracts from rabbit brain (Sigma) were used as a source of procoagulant lipids essentially devoid of either protein or sugars. The THP-1 cells were maintained in culture at 37°C, 5% CO 2 atmosphere in RPMI 1640 media supplemented with 10% fetal calf serum and 2 ϫ 10 Ϫ5 M ␤-mercaptoethanol. Cells were propagated by serial passage at 3-5-day intervals. Before use in kinetic experiments, cells were washed with serum-free medium, M-199 (Life Technologies, Inc.) containing 0.5% low protein serum replacement (Sigma) and 2 g/ml endotoxin (Escherichia coli Serotype 0127:B8, Sigma), and incubated at 5 ϫ 10 6 /ml in slanted polypropylene tissue culture tubes. The pattern of GAG synthesis by THP-1 cells is very similar to that of human, rat, and mouse * This study was supported by Grants HL-42812 (to M. P. M.) and HL-25161 (to W. W.) from the National Institutes of Health and by Grant 93007890 (to M. P. M.) from the American Heart Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Dept. of Medicine, Rheumatology Section, Medical Center Blvd., Winston-Salem, NC 27157-1058. monocyte/macrophage. These cells synthesize and secrete into the medium HS and CS containing proteoglycans. Upon differentiation, the density of sulfation, the size heterogeneity, and the average polymer length of the CS GAG increase (Edwards et al., 1995). Mononuclear cells were isolated from the blood of human volunteers by centrifugation through discontinuous density gradients formed with Isolymph. Monocytes were further separated from the mononuclear cell population by centrifugation through hypertonic Percoll gradients (45% Percoll in NaCl at 9.3 mg/ml) as described previously (McGee and Li, 1991). Contaminating platelets were removed using differential centrifugation. Populations utilized in kinetic experiments contained less than 1 platelet/10 nucleated cells. Nucleated cells in these populations were 85-95% monocytes and 5-15% lymphocytes by morphologic criteria as determined using Wright's and nonspecific esterase stains (Li et al., 1973). Purified lipid and THP-1 cells were used for standardization of experimental conditions and for the bulk of kinetics measurements. All key experiments relating to the specificity of GAG inhibition were confirmed with human blood monocytes.
Glycosaminoglycans-Chondroitin sulfate A, B, C, D, and E and HS used in these studies were "Super Special grade" from Seikagaku Kogyo Co. Ltd. The CSE preparation was purified from the squid cartilage. It gave a single band following electrophoresis on cellulose acetate, and the monosaccharide units were identified as glucuronic acid and galactosamine. The preparation had a 1.5 sulfate to hexosamine ratio with sulfate groups present both with axial and equatorial orientation and located at positions C-4 and C-6 of the galactosamine residue (Kawai et al., 1966). These GAG structures corresponded to those identified in oversulfated polysaccharides expressed by monocyte/macrophage (Edwards et al., 1995). Oligosaccharides were isolated by partial cleavage with hydrazine/hydrazine sulfate treatment (90 min at 100°C), high pH nitrous acid treatment, and NaBH 4 reduction. (Maimone and Tollefsen, 1990;Shaklee and Conrad, 1984). Tritium-labeled CSE oligosaccharides were also obtained by employing [ 3 H]NaBH 4 (Parthasarathy et al., 1994). Labeled oligosaccharides were mixed with unlabeled oligosaccharides at a ratio of 1:400 and eluted from Bio-Gel P-6 (Bio-Rad) column (120 ϫ 1 cm) with 0.5 M NH 4 HCO 3 to separate oligosaccharides containing 2, 6, 8, 10, 12, or higher numbers of monosaccharide units (Gallagher and Walker, 1985). The approximate sizes of oligosaccharides with higher than 12 units and that of nonhydrolyzed CSE preparations were determined by Sepharose 6B gel filtration on a column (53 ϫ 0.9 cm) eluted with 0.2 M NaCl (Wasteson, 1971). The concentration of CSE in each fraction was normalized with respect to uronic acid content (Blumenkrantz and Asboe-Hansen, 1973) .
Coagulation Factors-Immunoaffinity-purified human factor VIII with specific activity greater than 3,000 units/mg of protein was obtained from Armour Pharmaceutical Company (Kankakee, IL). Purity and electrophoretic characteristics of this protein have been described before (Zimmerman, 1988). Purified human factors X and IXa with specific activity of 125-200 units/mg and 200 units/mg of protein, respectively, were obtained from Enzyme Research Laboratories, Inc. (South Bend, IN). Human recombinant factor VIIa was donated by Dr. Ulla Hedner (Novo Nordisk, Copenhagen, Denmark). These factors were electrophoretically homogeneous and functionally pure as determined in coagulation tests. Thrombin (2,720 NIH units/mg) was a generous gift from D. J. Fenton (The Albany Medical College of Union University, Albany, NY). Factor Xa was generated from factor X by limited proteolysis with Russell's viper venom as described previously (Smith, 1973).
Measurement of Procoagulant Activity-Procoagulant activity was quantified using a one-stage clotting assay with pooled human plasma (George King, Bio-Medical, Inc., Overland Park, KS) as described previously (Rothberger et al., 1984). 60 l of sample were mixed with 60 l of 0.025 M CaCl 2 and 60 l of citrated plasma. Clotting times were measured using a fibrometer (Becton Dickinson & Co., Mountain View, CA) and converted to arbitrary TF units using logarithmic plots of clotting times versus dilution of a standard TF solution prepared with crude human brain homogenate. By definition, 1,000 milliunits/ml of procoagulant is the concentration of TF giving a coagulation time of 50 s in this test.
Measurement of Factor X Activation Rates-Rates of factor X activation were measured in continuously stirred mixtures containing procoagulant membranes (with or without added GAG), CaCl 2 (5 mM), and purified human coagulation factors (IXa, VIIIa, or VIIa). Factor VIII was optimally activated to factor VIIIa with thrombin, 0.05 units/ml as before (McGee et al., 1991). This activation reaction is very fast and under the conditions of these experiments independent of factor IXa and lipid concentration. Coagulation factors were added in 50 mM Tris buffer, pH 7.3, containing 0.15 N NaCl, pH 7.3, with either ovalbumin or bovine serum albumin at 0.5 mg/ml. Components were at equilibrium before initiation of factor X activation reaction, as indicated by the fact that experimental results were not significantly different in mixtures preincubated for either 2 or 4 min. All components were kept in molar excess relative to factors IXa and VIIIa. Reactions were initiated by the addition of substrate, and 40-l samples were taken every 15-30 s for 3.5 min. The concentration of factor Xa in each sample was determined with specific factor Xa chromogenic substrate S-2222 (N-benzoyl-Lisoleucyl-L-glutamylglycyl-L-arginine--nitroanilide hydrochloride) at 0.3 mg/ml in Tris buffer and 0.25 N NaCl, pH 8.3. Initial rates of chromogenic substrate hydrolysis were followed at 405 nM using a V max kinetic microplate reader from Molecular Devices. Progression curves were analyzed using plots of factor Xa concentration versus time. Steady-state rates were determined from the slope of straight lines fitted to experimental data points manually and/or by using a computer routine for regression analysis. When progression lines deviated appreciably from linearity, the initial tangent to the curve was obtained as the second coefficient of a polynomial equation fitted to the observed progress curve (Douglas and Wilson, 1984). Inhibition (I) was calculated as I ϭ 1 Ϫ R i /R, where R i is the initial rate in the presence of GAG and R is the rate in identical reaction mixtures but without GAG.

RESULTS AND DISCUSSION
Inhibition of Monocyte Procoagulant Activity by GAGs-Procoagulant activity was induced on either blood monocytes or THP-1 cells by incubation in M-199 serum-free media supplemented with 0.4% low protein serum replacement and endotoxin (1 g/ml for 20 h). As described previously (McGee et al., 1994), under these conditions both cell types express procoagulant activity with similar time courses. Activity increased rapidly during the first 4 h of culture, afterward slowly reaching maximum levels at 18 -20 h. The increase in procoagulant activity was primarily due to expression of TF on the cytoplasmic membrane. Expression of lipid cofactor sites for assembly of IXa-VIIIa is constitutive in monocytes and changes little upon incubation with inflammatory stimuli. The possibility that added GAGs can modify the levels of procoagulant activity was investigated using one-step clotting assay with recalcified human plasma as substrate. Chondroitin sulfate A, B, C, D, and E and HS were incorporated in reaction mixtures at concentrations ranging from 0 to 134 g/ml (Table I). Clotting times were increased by CSB and CSE and by HS.
The various GAGs were also incorporated in reaction mixtures with purified coagulation factors. Extrinsic pathway protease was measured in mixtures containing either cells (monocytes or THP-1) or human brain homogenates as a source of tissue factor. Intrinsic pathway protease was measured in mixtures with either cells or rabbit brain lipid extracts as a source of lipid co-factor for intrinsic complex assembly. Of the seven structurally different types of GAG tested, only CSE induced TABLE I Anticoagulant activity of GAGs Procoagulant activity in the presence of the indicated concentration of GAGs was measured as the shortening of plasma clotting time upon mixing with cell suspensions. Activity was quantified using a reference curve constructed with human brain homogenate as a TF standard. In this assay, 1000 milliunits/ml is the concentration of TF that shortened the clotting time of recalcified plasma to 50 s. The TF activity in control cell suspensions containing no added GAGs was 396 milliunits/ml. Data are from experiments using THP-1 cells as a source of procoagulant membranes. Similar results were obtained using human brain homogenates as procoagulant. large changes in factor X activation rates as compared with reaction mixtures without GAGs (Fig. 1A).
Chondroitin sulfate E induced a modest but significant increase in the rate of factor X activation by extrinsic pathway protease. In contrast, the rate of factor X activation by the intrinsic pathway protease was markedly reduced by this GAG (Fig. 1B).
Inhibition of Intrinsic Pathway Protease as a Function of Lipid Concentration-The inhibition calculated from rates of reactions assembled on cells was always lower than in reactions assembled on lipid extracts, suggesting that baseline reactions on cells reflect partial inhibition by endogenous GAGs. This was further examined by measuring inhibition at progressively increasing concentrations of phospholipid membranes added to reaction mixtures either as rabbit brain lipid extract or as intact cells. The results in Fig. 2 indicate that inhibition changed little when purified lipids were added at concentrations ranging from 0.05 to 12 M. In contrast, similar concentration range of phospholipid added as intact cells resulted in apparent inhibition that was inversely correlated with the membrane phospholipid concentration. This is as expected if cell membranes contribute both lipid assembly sites and inhibitory GAGs.
Further evidence for inhibitory GAGs expressed on the monocyte membrane was obtained by removing surface proteoglycans using mild trypsinization (0.05% Trypsin and 0.53 mM EDTA solution, Life Technologies, Inc.). Intrinsic pathway protease complexes were assembled on cells, either after a limited, 0.5-min trypsin treatment or after a 10-min trypsin treatment. Cells in both preparations were subjected to the same extent of physical manipulation except for the length of incubation in the presence of trypsin. As previously reported, trypsinization effectively removes CSE-and HS-containing proteoglycans from the monocyte surface. (Edwards et al., 1995). Rates of factor X activation on cell trypsinized for 10 min were 9.8 Ϯ 0.8 nM factor Xa/min and significantly faster than rates on cells subjected to limited trypsinization (4.0 Ϯ 0.2 nM factor Xa/min). This result is also consistent with membrane expression of inhibitory GAGs by the monocyte.
Monocytes and monocytoid cell lines are known to increase synthesis and expression of CSE upon incubation in vitro in the presence of inflammatory stimuli. (Kolset et al., 1983;Uhlin-Hansen et al., 1989;Uhlin-Hansen et al., 1993;Edwards et al., 1995). To determine if this incubation-dependent increase in CSE levels is reflected on intrinsic pathway protease activity, inhibition of factor X activation with CSE was measured on either human monocytes or THP-1 incubated with endotoxin for 48 h at either 37 or 4 -8°C. The concentrations of CSE required to achieve half-maximal inhibition were significantly (p Ͻ 0.05) increased (2.7-fold for three independent experiments) on cells incubated at 37°C as compared with cells incubated at 4 -8°C. Again, results are consistent with baseline regulation of intrinsic protease by membrane GAGs on macrophages.
Dependence of Inhibition on Size of CSE Polymer-Analysis by size exclusion chromatography indicated considerable size heterogeneity in the CSE preparation, Fig. 3. The elution profile included the range of CSE sizes described for monocyte CSs (Edwards et al., 1995, Uhlin-Hansen et al., 1993. Based on peak elution profile, the estimated average molecular weight was 88,000, consistent with a distribution coefficient, k av of 0.22. The inhibition profile of CSE fractions paralleled their uronic acid concentration profile, suggesting that for the range of sizes with Ͼ12-18 monosaccharide units, inhibition may be FIG. 1. Differential modulation of intrinsic and extrinsic protease by chondroitin 4,6-disulfate. Initial rates of factor X activation were measured in reaction mixtures with cells (0.8 -1 ϫ 10 6 /ml), factor IXa (1 nM), factor VIIIa (1 nM), or factor VIIa, (3 nM), CaCl 2 in TRIS buffer, pH 7.4, 0.15 N NaCl, and 0.5 mg/ml bovine serum albumin. Glycosaminoglycans, CSA, B, C, D, and E, and HS were added at concentrations indicated on the abscissa. Rates with all GAGs except CSE were within 10% of values measured in the absence of GAGs. Only rates with CSE are shown in the figure for clarity. Cells are THP-1 cells cultured for 48 h in the presence of endotoxin. Similar results were obtained using human brain homogenates as a source of procoagulant membranes.

FIG. 2. Inhibition of intrinsic pathway protease on cells and purified lipid surfaces.
Initial rates of factor X activation were measured in reaction mixtures without GAGs and mixtures with 6 g/ml CSE. Concentration phospholipid membranes supplied either as rabbit brain lipid extract or as intact THP-1 cells was as indicated on the abscissa. Intrinsic pathway components and other reagents were as indicated in the legend to Fig. 1. Inhibition, (I) was calculated as 1 Ϫ R i /R, where R i and R are the initial rates measured in mixtures with and without GAG, respectively. Apparent inhibition decreased with cell membrane concentration (q) but not with the sugar-free, protein-free lipid extract concentration (E). correlated with polysaccharide size. Among the small oligosaccharides obtained upon fragmentation of CSE by partial hydrazinolysis/high pH nitrous acid treatment and subsequent fractionation on Bio-Gel P-6 chromatography, oligosaccharides with 12-20 monosaccharide units at 10 g/ml mediated inhibition that was 25% of the inhibition mediated by intact CSE at the same concentration (see Table II). Additional chromatography of the CSE fragments in Sepharose 6B suggested that the average size of the smaller oligosaccharide with inhibitory activity, correspond to an molecular weight of approximately 6500, consistent with a k av of 0.77.
Mechanisms of Intrinsic Pathway Protease Inhibition by Chondroitin Sulfate E-Reaction components assembled either on cells or on purified lipid were titrated with CSE over three different concentration ranges. (Table III). Apparent kinetic parameters (K i and I max ) varied with the concentration range, suggesting that inhibition values reflect more than one type of molecular interaction. Determination of relative values of apparent K i was obtained using complete titrations with CSE concentrations ranging from 0.008 to 158 g/ml. Using Scatchard type plots, values of K i approximated from the slope of the two extreme linear segments were 5 and 1000 ng/ml (Fig. 4,  insets A and B, respectively). Although there are inherent uncertainties in the analysis of multiple equilibria from kinetic data, these results identified at least two different types of inhibitory interactions with affinities differing at least by 100fold. These results, however, do not exclude the possibility of additional interactions of intermediate affinity or that the observed heterogeneity reflects the wide range of polymer sizes in the CSE preparation.
Based on these results, further investigations of GAG-protein interactions were approached using multiple titration experiments with reaction components. For these experiments, each one of the reactants was alternatively used as a titrating component, whereas the inhibitor and the rest of the reactants were maintained at constant concentrations. The rationale of this approach is that the component containing binding sites for CSE should be displaced by the inhibitor from the functional protease complex and that the empirical value of the kinetic parameter (K1 ⁄2 ), that is, concentration of component resulting in half-maximal rates, will increase in reactions with CSE as compared with reactions without the inhibitor. Simi-larly, the value of V sat , that is, the rate at saturating concentration of competing component, should be the same in reactions with and without CSE. To avoid possible interference from cell-derived GAGs, kinetic parameters were determined FIG. 3. Chondroitin sulfate E was fractionated in Sepharose 6B. Fractions were analyzed with respect to uronic acid content (q) and inhibitory activity ( ). The inhibition levels of factor X activation reactions with CSE were calculated as indicated in the legend to Fig. 2 for unfractionated CSE. Elution of markers for void volume (Vo) total included volume (Vt) and peak elution of fraction with 12-18 monosaccharide units (OS) are indicated.

TABLE II Differences between parameters of intrinsic protease inhibition on cells and on lipid surfaces
Parameters (K i and I max ) were calculated from initial rates of factor X activation measured in reaction mixtures with factor IXa (0.8 nM), factor VIIIa (1 nM), and factor X (150 nM). Data were fitted to an equation, (I max Ϫ I) (K i ϩ [CSE]) ϭ (K i )(I max ), using least square methods. In this equation, [CSE] represents concentration of CSE. I, the inhibition, calculated as 1 Ϫ R i /R where R i , and R are factor X activation rates measured with and without CSE, respectively. Values of K i , that is, the concentration of CSE resulting in half-maximal inhibition, were significantly greater for cells than for lipid surfaces and varied with the range of CSE concentration. CSE

Inhibition of intrinsic protease by CSE oligosaccharide
Oligosaccharides composed of the number of monosaccharide units indicated were generated by limited cleavage of CSE with hydrazine/ hydrazine sulfate followed by NaHB 4 reduction and fractionation on a P-6 column. Inhibition was measured after incorporating each size oligosaccharide in reaction mixtures with purified lipid and coagulation components, as described in the legend to Fig. 1. Oligosaccharides and intact CSE were at 10 g/ml. Rates of factor X activation were measured in mixtures with and without CSE at concentrations ranging from 8 to 1.6 ϫ 10 5 ng/ml. Other reagents and determination of inhibition were as indicated in the legend to Fig. 2. The ratios between inhibition (I) and CSE concentration ([CSE]) are plotted versus I. The slopes of the two extreme linear segments calculated by linear regression correspond to K i1 ϭ 5 ng/ml and K i2 ϭ 1000 ng/ml. The results are consistent with two independent sites n 1 and n 2 resulting in 0.57 and 0.43 fractional inhibition, respectively. using lipid extracts from rabbit brain as a source of protein-free and sugar-free lipid surface.
The concentration ranges for the titrating component were: factor X, 0.1-300 nM; factor IXa, 0.05-8 nM; and factor VIII 0.05-6 nM. The results (see Table IV) from titrations with substrate are compatible with a noncompetitive type inhibition. The observed decrease in K1 ⁄2 for factor X with CSE is as expected for a diffusion-limited reaction measured at two levels of enzyme. For diffusion-limited reactions, a decrease in enzyme density decreases the extent of diffusion control, and the value of K1 ⁄2 tends to approach that of the intrinsic K m (McGee et al., 1992). Therefore, this result indicates that inhibition observed with CSE is mediated by a decrease in the effective density of the enzyme without direct interference with substrate binding and suggests that CSE binds to epitopes on the protease components differently from substrate binding sites. However, neither one of the two intrinsic protease complex component(s) added in excess to the other were able to overcome the inhibition. Because increasing either protease component increases the concentration of functional VIIIa-IXa complex, these results suggest that inhibition by CSE can be mediated by interaction with sites on factor VIIIa and IXa independently of VIIIa/IXa complex formation. When concentrations of VIIIa and IXa were increased simultaneously maintaining equimolar concentrations of each, the inhibitory effect of CSE was abolished and maximal rates were not significantly different in mixtures with CSE as compared with mixtures without CSE. However, IXa and VIIIa concentrations resulting in half-maximal rates increased 5-fold (Table III), indicating that the diffusional limit was achieved at lower IXa and VIIIa concentration in reaction mixtures without CSE.
The difference in results obtained with equimolar increments of VIIIa and IXa as compared with results obtained with one component added in excess to the other are consistent with predictions based on simple probability considerations. Assuming that both factor IXa and VIIIa bind CSE and that the IXa-CSE and VIIIa-CSE species bind to free and bound VIIIa and IXa, respectively, then when either protease component is in molar excess, the probability of nonfunctional complexes is expected to increase with the concentration of the component that is in excess. This is seen more clearly when considering the inhibition as a function of the varied component concentration. The inhibition increased only when either factor IXa or factor VIII were increased but not when both were increased. (Fig. 5,  A, B, and C). To further test the hypothesis and to determine the component with the high affinity site for CSE, titrations were repeated at CSE concentrations below the K i estimated for the low affinity site. In these experiments, increasing the concentration of VIIIa while maintaining IXa constant had little effect on the inhibition level, particularly at low concentrations of VIIIa. In contrast, increasing factor IXa while maintaining factor VIIIa concentration fixed resulted in a marked and progressive increase in the inhibition. Therefore, the results exclude the possibility that CSE inhibits the factor X-ac- parameters of inhibited and noninhibited reactions Kinetic parameters (K 1/2 ) I , (V sat ) I , K 1/2 , and V sat were determined from the initial rates of factor X activation in reaction mixtures with (subscript I) and without CSE, respectively. Parameters were calculated from rectangular hyperbolas fitted to the data by least square methods. Rates were measured at fixed concentrations of all reactants except one. Fixed concentrations were: factor IXa, 0.85 nM; factor VIIIa, 1 nM; and factor X, 150 nM. Surface for protease assembly were either cells (1 ϫ 10 6 /ml) or rabbit brain lipid extract (60 nM lipid phosphorous) suspended in Tris buffer, pH 7.3, 0.15 N NaCl, 5 mM CaCl 2 , and 0.5 mg/ml ovoalbumin. Chondroitin sulphate E was at 40 g/ml. Results are expressed as the ratios between kinetic parameters of reactions with CSE and kinetic parameters of reactions without CSE. Mean Ϯ S.E. of three independent sets of determinations. Reaction rates were also measured at fixed concentrations of all components except IXa and VIIIa, which were added at increasing concentrations while maintaining a constant molar ratio between the two. Rates of factor X activation on lipid membranes were measured in mixtures with and without CSE at 50 g/ml. Concentration of IXa and VIIIa were increased as indicated on the abscissa either independently (A and B, respectively) or simultaneously at equimolar concentrations (C). Other reagents and determination of inhibition were as indicated in the legend to Fig. 2. tivating reaction via interaction with sites expressed upon IXa-VIIIa complex formation and in their simplest interpretation are consistent with at least two independent CSE binding sites. One of the sites interact with high affinity and is located on factor IXa; the other site(s) interacts with low affinity and is located on factor VIII. Alternatively, the interaction with factor VIII may also be high affinity but involves a polysaccharide size(s) present at relatively low concentration in the CSE preparation. The results are summarized in Tables IV and Figs. 4, 5, and 6.
Glycosaminoglycans are known to interact with and modulate the activity of many biologically important proteins, including coagulation inhibitors such as AT III (Marcum et al., 1966) and heparin cofactor II (Pratt et al., 1989), and growth factors such as fibroblast growth factor and hepatocyte growth factor (Lyon and Gallagher, 1994). Available evidence indicates that the interaction of GAGs with these proteins depends on the sequence and size of their oligosaccharide units. For example, chondroitin sulfates A, B, and C have common disaccharide units of hexuronic acid (glucuronic acid in CSA and CSE; iduronic acid in CSB) and N-acetylgalactosamine-O-sulfate. However, only CSB is known to bind to AT III and heparin cofactor II (Maimone and Tollefsen, 1990). Similarly, commercial heparin and heparan sulfates have common disaccharide units of hexuronic acid (glucuronic acid/iduronic acid) and N-acetyl (or N-sulfated) glucosamine O-sulfates, but their protein-binding characteristics are different. In this regard, the binding sites for proteins have been mapped to specific regions comprising a few oligosaccharide units (Gallager et al., 1992). The consensus from these studies is that the minimum oligosaccharide length required for binding is 5-7 disaccharide units but that the size required to elicit biological activity after binding is larger. Therefore, it is not surprising that in the present functional studies oligosaccharides longer than dodecasaccharides were required for activity. Heparin has been reported to inhibit intrinsic but not extrinsic or common proteases (Barrow et al., 1994). The saccharide sequence(s) required to mediate this inhibition was/were not determined. Because heparin sequences are heterogeneous with respect to sugar composition and sulfate substitutions, it is possible that the inhibition observed was mediated by O-sulfate groups as shown for CSE in the present studies. Alternatively, intrinsic protease components may have sites for interaction both with CSE and with N-sulfated sequences in commercial heparin. Whereas heparan sulfate GAGs are present in most tissues, heparin sequences such as those in commercial anticoagulant preparations are almost exclusively found in mast cell granules. Anticoagulant effects of CSE mediated primarily by acceleration of heparin cofactor II interaction with thrombin have been reported (Scully et al., 1986). The apparent K d for the interaction was in the g/ml range, approximately 2 orders of magnitude higher than the values obtained in the present studies for the interaction between CSE and the intrinsic pathway protease.
Considered together, the variety of different molecular interactions increasingly being described in the regulation of blood coagulation by GAGs (Bourin and Lindahl, 1993) points to complex physiochemical mechanisms and underline the importance of considering structural requirements for each interaction independently rather than generalizing from studies with heparin.
Intense efforts in coagulation research are currently being directed to identify specific inhibitors and modulators of blood coagulation that may control hypercoagulability while maintaining the vital hemostatic function of the coagulation system. Because of the existence of alternate routes for factor Xa generation, it is reasonable to expect that specific inhibition of intrinsic pathway protease would result in less hemorrhagic complications than inhibition of the initiating complex, TF/ FVIIa, or the prothrombinase complex. In this regard, studying the response of monocyte procoagulants to GAGs is of importance not only for characterization of coagulation reactions in a biologically relevant microenvironment but also for identification of inhibitors with potential use in pharmacologic control of hypercoagulable states associated with atherosclerosis, inflammation, and cancer.
The results of the present studies indicate specific modulation of the intrinsic pathway protease but not of extrinsic protease by chondroitin sulfate E. The results are consistent with modulation of baseline procoagulant activity by GAGs expressed on monocytes. Added GAGs in the extracellular medium further inhibited reactions in a manner that was strictly dependent on the spatial distribution of charges on the polysaccharide. Multiple functional titrations and Scatchard type analysis provided kinetic evidence consistent with a hypothetical model of inhibition mediated by interaction between CSE and at least two independent sites: one on factor IXa and another on VIIIa. In addition, results suggest that this GAG may provide a useful anticoagulant to manage prothrombotic states associated with activity of the intrinsic pathway amplifying loop.
FIG. 6. Inhibition as a function of protease component concentration; low inhibitor concentration regime. Rates of factor X activation of lipid membranes were measured in mixtures with and without CSE at 50 ng/ml. Concentrations of either VIIIa or IXa were increased as indicated on the abscissa of A and B, respectively.