Activation of heparin cofactor II by fibroblasts and vascular smooth muscle cells.

Inhibition of thrombin by heparin cofactor II (HCII) is accelerated by dermatan sulfate, heparan sulfate, and heparin. Purified HCII or defibrinated plasma was incubated with washed confluent cell monolayers, 125I-thrombin was added, and the rate of formation of covalent 125I-thrombin-inhibitor complexes was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography. Fibroblasts and porcine aortic smooth muscle cells accelerated inhibition of thrombin by HCII 2.3-7.5-fold but had no effect on other thrombin inhibitors in plasma. Human umbilical vein endothelial cells and mouse macrophage-derived cells did not accelerate the thrombin-HCII reaction. IMR-90 normal human fetal lung fibroblasts treated with heparinase or heparitinase accelerated the thrombin-HCII reaction to the same degree as untreated cells. In contrast, treatment with chondroitinase ABC almost totally abolished the ability of these cells to activate HCII while chondroitinase AC had little or no effect, suggesting that dermatan sulfate was responsible for the activity observed. [35S]Sulfate-labeled proteoglycans were isolated from IMR-90 fibroblast monolayers and conditioned medium and fractionated into two peaks on Sepharose CL-2B. The lower Mr proteoglycans contained 74-76% dermatan sulfate and were 11-25 times more active with HCII than the higher Mr proteoglycans which contained 68-97% heparan sulfate. The activity of the lower Mr proteoglycans decreased 70-90% by degradation of the dermatan sulfate component with chondroitinase ABC. These results confirm that dermatan sulfate proteoglycans are primarily responsible for activation of HCII by IMR-90 fibroblasts. We suggest that HCII may inhibit thrombin when plasma is exposed to vascular smooth muscle cells or fibroblasts.

Inhibition of thrombin by heparin cofactor I1 (HCII) is accelerated by dermatan sulfate, heparan sulfate, and heparin. Purified HCII or defibrinated plasma was incubated with washed confluent cell monolayers, "' Ithrombin was added, and the rate of formation of covalent '2sI-thrombin-inhibitor complexes was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography. Fibroblasts and porcine aortic smooth muscle cells accelerated inhibition of thrombin by HCII 2.3-7.5-fold but had no effect on other thrombin inhibitors in plasma. Human umbilical vein endothelial cells and mouse macrophage-derived cells did not accelerate the thrombin-HCII reaction. IMR-90 normal human fetal lung fibroblasts treated with heparinase or heparitinase accelerated the thrombin-HCII reaction to the same degree as untreated cells. In contrast, treatment with chondroitinase ABC almost totally abolished the ability of these cells to activate HCII while chondroitinase AC had little or no effect, suggesting that dermatan sulfate was responsible for the activity observed.
[S6S]Sulfate-labeled proteoglycans were isolated from IMR-90 fibroblast monolayers and conditioned medium and fractionated into two peaks on Sepharose CL-2B. The lower M , proteoglycans contained 74-76% dermatan sulfate and were 11-25 times more active with HCII than the higher M, proteoglycans which contained 68-97% heparan sulfate. The activity of the lower M, proteoglycans decreased 70-90% by degradation of the dermatan sulfate component with chondroitinase ABC. These results confirm that dermatan sulfate proteoglycans are primarily responsible for activation of HCII by IMR-90 fibroblasts. We suggest that HCII may inhibit thrombin when plasma is exposed to vascular smooth muscle cells or fibroblasts.
Thrombin, the serine protease derived from prothrombin during blood coagulation, has a number of biologic activities. The procoagulant activities of thrombin include activation of factors V, VII, VIII, and XIII, conversion of fibrinogen to fibrin, and stimulation of platelet aggregation and degranulation (1). Thrombin can also serve as an anticoagulant by * This work was supported by National Institutes of Health Grants HL-27589 and HL-07088 and by the Monsanto Company. Portions of this work were presented at the Annual Meeting of the American Society of Hematology, December 7-10, 1985, New Orleans and published in abstract form in Blood 66, 340A (1985). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U. S 63110. activating protein C in the presence of thrombomodulin (2). In addition, thrombin is mitogenic for fibroblasts (3) and has chemotactic activity for monocytes and macrophage-derived cells (4). These properties suggest that thrombin may be involved in tissue repair in addition to its functions in hemostasis. Inhibitors of thrombin may modulate these various functions.
Heparin cofactor I1 (HCII)' and antithrombin I11 (ATIII) are protease inhibitors in plasma that form 1:l covalent complexes with their target proteases. ATIII inhibits all of the proteases of the inirinsic coagulation pathway including thrombin, factors Xa, IXa, XIa, XIIa, and kallikrein (5); of these proteases, HCII inhibits only thrombin (6). Both HCII and ATIII inhibit thrombin -1000 times more rapidly in the presence of heparin (5, 7). Dermatan sulfate also accelerates the inhibition of thrombin by HCII to a similar degree while having no appreciable effect on inhibition of thrombin by ATIII (8). Chymotrypsin (9) and leukocyte cathepsin G (6) are also inhibited by HCII, but these reactions are not accelerated by heparin or dermatan sulfate. Thus, in the presence of either glycosaminoglycan, thrombin is the preferred target protease for HCII.
ATIII appears to be an important inhibitor of the procoagulant activity of thrombin, since deficiency of ATIII is associated with thrombosis (10). Currently, it is thought that ATIII is activated in vivo by binding to heparin-like molecules on the surface of vascular endothelial cells (11,12). Intravascular activation of HCII has not been demonstrated under physiologic conditions (13); however, intravenous administration of dermatan sulfate produces an antithrombotic effect in rabbits by activation of HCII (14). Although the physiologic function of HCII is unknown, we have postulated that HCII may inhibit thrombin in vivo in the vicinity of cells that synthesize significant quantities of dermatan sulfate. In this study, we have examined cultured cell monolayers for the ability to activate HCII. We find that HCII is activated by fibroblasts and vascular smooth muscle cells but not by endothelial cells and macrophage-derived cells. Furthermore, the ability of fibrolast monolayers and conditioned media to activate HCII is mediated primarily by dermatan sulfate proteoglycans synthesized by these cells.

EXPERIMENTAL PROCEDURES
Materials-Human HCII, ATIII, and thrombin were purified and assayed as previously described (7). Citrated normal human plasma was defibrinated by treatment with 1.0 unit/ml thrombin for 15 min at 22 "C followed by centrifugation prior to use. Thrombin was iodinated with carrier-free sodium ['251]iodide by the chloramine-T The abbreviations used are: HCII, heparin cofactor 11; ATIII, antithrombin III; Hepes, 4-(2-hydroxyethyl)-l-piperazine-ethanesulfonic acid BSA, bovine serum albumin; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis. 169 method as previously described (15 Seattle. Cells were grown in Dulbecco's modified Eagle's medium containing 4 g/liter glucose at 37 "C in a humidified 5% C02 incubator. Chinese hamster ovary cells were grown in minimum essential medium. Media were supplemented with 10% heat-inactivated fetal calf serum (Gibco), 2 mM L-glutamine, 100 units/ml penicillin, 100 pg/ml streptomycin, and 10 mM Hepes, pH 7. 4. NIH 3T3 fibroblasts were also supplemented with 1.1 mg/ml sodium pyruvate. Primary cultures of human umbilical vein endothelial cells were prepared by the method of Jaffe et al. (18) and were provided by Drs. Ikuro Maruyama and Philip Majerus, Washington University.
Assay for the Effects of Cultured CeUs on the Formation of Thrombin-Inhibitor Compleres-Human fibroblasts and porcine smooth muscle cells were assayed between 10 and 17 passages. Cells were grown to confluence in 2-cm2 wells of polystyrene 24-well tissue culture plates (Costar). Empty wells on the same plate, which served as controls, were conditioned with medium identical to that used to grow the cells. Medium was aspirated from empty and cell-containing wells. The wells were then washed three times with 1 ml of 150 mM NaCl, 1 mM CaClz, 0.1% polyethylene glycol, 1 mg/ml BSA, 20 mM Tris-HC1, pH 7.4 (Tris-saline buffer). Immediately after washing, 0.25 ml of inhibitor (HCII, ATIII, or defibrinated plasma) diluted in Tris-saline buffer was added to each well. At time zero, 0.25 ml of '251-thrombin diluted in Tris-saline buffer was added to each well. Tissue culture plates were then rotated on a Fisher clinical rotator at 80 rpm at room temperature to mix the reactants. During this procedure, cells remained adherent to tissue culture plates. At intervals, 60-pl samples were removed from each well, immediately added to an equal volume of SDS-electrophoresis sample buffer (20% glycerin, 4% SDS, 10% 2-mercaptoethanol, and 0.002% bromphenol blue in 0.125 M Tris-HC1, pH 6.8), and heated for 2 min at 100 "C. Samples were electrophoresed on SDS-polyacrylamide gels (4% stacking gel, 7.5% running gel) using the Laemmli buffer system (19) with an SE200 Mighty Small vertical slab gel apparatus (Hoefer). Gels were stained with Coomassie Brilliant Blue, dried, and autoradiographed as previously described (20). Gel regions containing radioactivity were cut from dried gels and counted in a Beckman model 300 y counter.
Metabolic Labeling of IMR-90 Fibroblasts with Sodium PSlSulfate-Cells were passaged into 850-cm2 polystyrene roller bottles, gassed with 95% air, 5% C02, and rotated 0.25 turns/min at 37 "C. When the cells reached confluence, they were incubated for 72 h with low-sulfate Dulbecco's modified Eagle's medium (MgC12 substituted for MgSO,). The medium was then replaced with 200 ml of fresh lowsulfate Dulbecco's modified Eagle's medium containing 7.5 pCi/ml carrier-free sodium [35S]sulfate (New England Nuclear), and the cells were labeled for 72 h. Radioactivity was measured in a Beckman model 6800 liquid scintillation counter using ScintiVerse I (Fisher).
Isolation of Proteoglycans from Culture Medium-Proteoglycans were extracted and isolated by the methods described by Yanagishita and Hascall (21) with some modifications. Medium was removed, and roller bottles were washed with 20 ml of 150 mM NaCl, 20 mM sodium phosphate buffer, pH 7. 4. The medium and wash fractions were combined. Solid urea, sodium chloride, sodium acetate, and N-ethylmaleimide were added to make final concentrations of 150, 50, and 10 mM, respectively. Triton X-100 was added to a final concentration of 0.5%. The mixture was then applied to a 5-ml DEAE-Sephacel column equilibrated with 8 M urea, 5 pg/ml BSA, 0.5% Triton X-100, 50 mM sodium acetate, pH 6.0 (8 M urea buffer) containing 150 mM NaCl. The column was washed with 25 ml of equilibration buffer, followed by step elution with 20 ml of 8 M urea buffer containing 0.4 M NaCl and 20 ml of 8 M urea buffer containing 0.65 M NaCl. Unincorporated [35S]sulfate did not bind to this column, while proteoglycans eluted with 0.65 M NaCl.
Size Fractionation of Proteoglycans-Proteoglycans eluted from the DEAE-Sephacel column were diluted with 5 volumes of 8 M urea buffer without sodium chloride and then applied to a l-ml DEAE Sephacel column equilibrated with 8 M urea buffer containing 150 mM NaCl. Proteoglycans were eluted with 2 ml of 4 M guanidine HCl, 50 mM sodium acetate, 50 mM Tris-HC1, pH 7.0, containing 5 pg/ml BSA and 0.5% Triton X-100 directly onto a 1 X 107-cm Sepharose CL-2B column equilibrated with the same buffer. One-ml fractions were collected at a flow rate of 4 ml/h.
Pronose Treatment-Pooled fractions from the Sepharose CL-2B column were concentrated with an Amicon PMlO membrane to 2 ml and dialyzed against 1 M NaCl, 20 mM CaC12, 100 mM Tris-HC1, pH 8.0. Samples were incubated at 57 "C for 24 h with 270 pg of Pronase added at 0, 8, and 16 h. Reaction mixtures were heated at 100 "C to inactivate the enzyme and dialyzed extensively against distilled water.
Glycosaminoglycan Analysis-The glycosaminoglycan composition of each Sepharose CL-2B peak was determined by measuring the percentage of [35S]sulfate susceptible to degradation by nitrous acid, chondroitinase AC, or chondroitinase ABC. Samples containing 6,000-10,000 cpm were evaporated to dryness in polypropylene microcentrifuge tubes. For nitrous acid degradation, the sample was dissolved in 0.45 ml of 10% acetic acid containing 240 mM NaNOZ. After an 80-min incubation at room temperature, the reaction was stopped by adjusting the pH to -5 with 10 N NaOH. Chondroitinase AC and ABC degradations were carried out in 0.5 ml of 55 mM Tris-HCl, pH 8.0, containing 0.1 unit/ml of enzyme. Samples were incubated at 37°C for 2 h. To separate degraded from undegraded material, each reaction mixture was chromatographed on a 0.5 X 45-cm Sephadex G-50 (fine) column equilibrated with 200 mM ammonium acetate-acetic acid, pH 5.0. Fractions 0.55 ml were collected at a flow rate of 2 ml/h. Untreated samples eluted in the void volume of the column. The percentage of radioactivity in the included volume after nitrous acid treatment was equal to that remaining in the void volume after chondroitinase ABC treatment and was considered to represent heparan sulfate. The percentage of radioactivity in the included volume after chondroitinase AC treatment was considered to represent chondroitin 4-sulfate and/or chondroitin 6-sulfate. Dermatan sulfate was estimated by subtracting the percentage of material degraded by chondroitinase AC from that degraded by chondroitinase ABC. This may result in some underestimation of dermatan sulfate content (see "Discussion").
Larger amounts of material were degraded with nitrous acid, chondroitinase AC, or chondroitinase ABC in the same fashion as described above. The Sephadex G-50 void volume peak from each degradation was lyophilized, dissolved in distilled water, and relyophilized several times to remove ammonium acetate. The material was then dissolved in 150 mM NaCl, 0.1% polyethylene glycol, 20 mM Tris-HC1, pH 7.4, and assayed as described below for the ability to accelerate thrombin inhibition by HCII.
Chromogenic Substrate Assay for Inhibition of Thrombin by HCII-Samples of proteoglycans isolated from IMR-90 fibroblasts were compared to serial dilutions of purified porcine skin dermatan sulfate for the ability to enhance the inhibition of thrombin by HCII. Five pl of HCII (80 pg/ml) was added to 85 pl of standard dermatan sulfate or the sample diluted in 150 mM NaCl, 0.1% PEG, 20 mM Tris-HC1, pH 7. 4. Ten pl of thrombin (8 units/ml) was then added. After incubating the solution for exactly 10 min at 22 "C, the residual thrombin activity was determined by the addition of 100 pl of 100 p M tosyl-Gly-Pro-Arg-p-nitroanilide. Hydrolysis of the substrate was terminated after exactly 1 min with 10 pl of hirudin (200 units/ml), and the absorbance at 405 nm was determined. The absorbance decreased with increasing concentration of dermatan sulfate from 0 to 4 pg!ml. The specific activity of a [""S]sulfate-labeled sample was calculated as units/cpm, where 1 unit is equal to the inhibitory effect produced by 1 p g of standard dermatan sulfate. This assay cannot distinguish between the effects of heparin, dermatan sulfate, and bin by HCII (8). heparan sulfate, all of which can accelerate the inactivation of throm-

Activation of HCII by Fibroblayts and Vascular Smooth
Muscle Cells-We examined a variety of cells in monolayer culture for the ability to accelerate formation of the complex between '"I-thrombin and HCII. Monolayers were washed with buffer and incubated with purified HCII and "'I-thrombin as described under "Experimental Procedures." Covalent complexes between "'I-thrombin and HCII in the supernatant buffer were detected by SDS-PAGE and autoradiography. When an equimolar amount of ATIII was substituted for HCII, there was no increase in formation of the ' "' Ithrombin-AT111 complex in the presence of cells (Fig. 1B). The time course of ""I-thrombin incorporation into t.he covalent complex with purified HCII is shown in Fig. 2A. In this experiment, the cell monolayer increased the rate of complex formation 3.4-fold. IMR-90 fibroblasts had no effect on the rate of complex formation between '"'I-thrombin and ATIII (Fig. 2B).
When I2'I-thrombin was incubated with defibrinated plasma in the presence or absence of IMR-90 fibroblasts, 4 major complexes were detected by aut,oradiography as shown in Fig. 3. Complexes "a" and "b," previously noted in our laboratory (8), have not been identified. Complexes "c" and "d" co-migrated with the complexes formed in the presence of purified HCII and ATIII, respectively. The amount of each complex formed a t 5 , 10, and 15 min is shown in Table I. Complexes a, b, and d were not affected by the presence of cells, while 2.6 times more complex c was formed in the presence of cells. These results agree with the experiments shown in Figs. 1 and 2 in which purified thrombin inhibitors were used. The results suggest that HCII is the major thrombin inhibitor in plasma that can be stimulated by fibroblasts.
Human skin fibroblasts, Chinese hamster ovary cells, and NIH 3T3 mouse fibroblasts accelerated formation of the "'1- thrombin-HCII complex from 2.3-7.5-fold under the conditions described in Table 11. Porcine aort,ic smooth muscle cells accelerated complex formation 3.2-fold. In experiments with defibrinated plasma, smooth muscle cells only stimulated formation of the 12511-thrombin-HCII complex (data not shown). Smooth muscle cells did not accelerate complex formation between purified ATIII and "'I-thrombin (data not shown). Primary cultures of human umbilical vein endothelial cells and two mouse macrophage-derived iines iP388D1 and 5774) did not activate HCII (Table 11). In the case of endothelial cells, there was a slight but significant inhibition of HCII activity.  ' Letters refer to bands labeled in Fig. 3.
b p < 0.01 by Student's two-tailed t test in comparison to the corresponding control ("-cells").  NS, not significant ( p > 0.1).
Chondroitinase ARC Abolishes the Activation of HCII by IMR-90 Fibroblasts-We have shown that HCII is activated by porcine skin dermatan sulfate and bovine liver heparan sulfate, although a 5-fold higher concentration of the latter is required (8). Since IMR-90 fibroblasts synthesize both dermatan sulfate and heparan sulfate as components of secreted and membrane-bound proteoglycans (22), either or both of these glycosaminoglycans could potentially account for the effect observed in Figs. 1 and 2.
Washed monolayers of IMR-90 fibroblasts were preincubated with enzymes to selectively degrade heparan sulfate or dermatan sulfate. No difference in adherence between enzyme-treated and control cells was apparent by light microscopy after the 2-h incubation. Cells were then washed three times and immediately assayed with 1251-thrombin and HCII for complex formation (Fig. 4A). Preincubation with heparinase or heparitinase had no effect on the amount of complex formed. In contrast, preincubation of cell monolayers with chondroitinase ABC almost completely abolished the ability of the fibroblasts to stimulate complex formation. Chondroitinase AC had a small but reproducible effect. Since the hexosamine-uronic acid linkages in dermatan sulfate are susceptible to chondroitinase ABC and resistant to heparinase   Fig. 4A suggest that dermatan sulfate is responsible for the activation of HCII by IMR-90 fibroblasts. We obtained similar results using normal adult skin fibroblasts (not shown).
Activation of HCII by Supernatant Medium from IMR-90 Fibroblasts-Monolayers of IMR-90 fibroblasts were washed and incubated with serum-free medium for 2 h at 37 "C. The supernatant medium was then removed and assayed with lZ5Ithrombin and HCII. As shown in Fig. 4B, the supernatant medium accelerated complex formation 3.6-fold compared to supernatant medium from empty tissue culture wells. When chondroitinase ABC was present during the 2-h incubation, the supernatant medium did not accelerate complex formation ( p > 0.1 in comparison to the "empty well" control). When chondroitinase AC was present during the 2-h incubation, the supernatant medium accelerated complex formation 2.5-fold. These results suggest that dermatan sulfate secreted by IMR-90 fibroblasts into the medium activates HCII.

Activation of HCZZ by Proteoglycans Purified from IMR-90
Fibroblasts-To confirm that dermatan sulfate from IMR-90 fibroblasts activates HCII, we fractionated the proteoglycans from these cells and determined their ability to accelerate inhibition of thrombin by HCII. The cells were metabolically labeled with [35S]sulfate and extracted with guanidine HCl and Triton X-100. The proteoglycans were then isolated by absorption to DEAE-Sephacel (see "Experimental Procedures" for details). Proteoglycans from the supernatant medium were isolated in a similar manner. Chromatography of the labeled proteoglycans on a Sepharose CL-2B column in the presence of guanidine HCl yielded two peaks for both the cell-associated (C-I and C-11) and supernatant material (M-I and "11) (Fig. 5). The peaks were pooled as shown and a  medium ( p a n e l A ) and cell extract ( p a w l  B ) as described under "Experimental Procedures." The proteoglycans were chromatographed on a 1 x 107-cm Sepharose CL-2B column equilibrated with 4 M guanidine HCl, 0.5% Triton X-100, 5 pg/ml BSA, 50 mM sodium acetate, and 50 mM Tris-HC1, pH 7.0. One-ml fractions were collected at a flow rate of 4 ml/h.

TABLE 111
Specific activities of P5Slsulfate-&kd proteoglycans Proteoglycans were pooled as shown in Fig. 5 and tested (before and after Pronase treatment) for acceleration of the thrombin-HCII reaction with the chromogenic substrate assay described under "Experimental Procedures." Pronase-treated samples were further treated with nitrous acid, chondroitinase AC, or chondroitinase ABC followed by gel filtration on a column of Sephadex G-50 to remove degraded material before being assayed. One unit of activity is defined as that equal to 1 pg of standard dermatan sulfate. Each value is the average of at least three determinations.   portion dialyzed against 150 mM NaCl, 0.1% polyethylene glycol, 20 mM Tris-HC1, pH 7. 4. Dialyzed material was assayed with HCII and thrombin in the chromogenic substrate assay described under "Experimental Procedures." The specific activities of peaks M-I1 and c-I1 were 11-25 times greater than those of peaks M-I and C-I, respectively, before Pronase treatment (Table 111).
Treatment of each of the peaks shown in Fig. 5 with Pronase yielded slightly increased K,, values compared to untreated material upon Sepharose CL-GB chromatography, suggesting the presence of a protein core in each case (data not shown). Degradation of the protein core with Pronase had little or no effect on the specific activity of each peak (Table 111).
After Pronase treatment, the glycosaminoglycan composition was determined for each of the Sepharose CL-XB peaks (Table IV). The higher molecular weight peaks from both the cell monolayer (C-I) and medium ("1) contained predominantly heparan sulfate, while the lower molecular weight peaks (C-I1 and "11) contained predominantly dermatan sulfate. Treatment of C-I1 and M-I1 with chondroitinase ABC degraded 86 and 92% of the labeled material, respectively (Table IV), and decreased the specific activity of the remaining intact glycosaminoglycan chains (i.e. heparan sulfate) by 70-90% (Table 111). Nitrous acid treatment of peaks C-I and M-I degraded 97 and 68% of the labeled material, respectively (Table IV). The remaining intact glycosaminoglycans (ie. mixtures of dermatan sulfate and chondroitin 4-and/or 6sulfate) had specific activities that were 4-17 times greater than the starting material. Treatment with chondroitinase AC, which will completely degrade chondroitin 4-and 6sulfate, had relatively minor effects on the specific activities of the four proteoglycan peaks. These experiments indicate that the bulk of the activity with HCII is due to the dermatan sulfate component of the fibroblast proteoglycans.

DISCUSSION
In this paper we have shown that cultured monolayers of fibroblasts and aortic smooth muscle cells, as well as supernatant medium from fibroblasts, accelerate formation of the "SI-thrombin-HCII complex. In contrast, umbilical vein endothelial cells and macrophage-derived cells have no effect. These findings suggest that the function of HCII may be to inhibit thrombin under conditions of endothelial cell disruption which lead to contact between plasma and subendothelial tissues.
Treatment of IMR-90 fetal lung fibroblasts with chondroitinase ABC, but not heparinase or heparitinase, greatly diminishes the ability of these cells to accelerate the thrombin-HCII reaction. Similarly, chondroitinase ABC treatment abolishes the activity of the supernatant medium from these cells. These experiments suggest that the acceleration of the thrombin-HCII reaction by fibroblasts depends primarily on the presence of dermatan sulfate on the cell surface and in the supernatant medium.
Chrondroitinase AC consistently decreases '2SI-thrombin-HCII complex formation by a small degree (Fig. 4). This enzyme cleaves the N-acetylhexosamine @1-4-glucuronic acid linkage, the only linkage found in chondroitin 4-sulfate and chondroitin 6-sulfate (23). However, we have previously shown that purified chondroitin 4-sulfate and chondroitin 6sulfate do not activate HCII (8). The N-acetylhexosamine /31-4-glucuronic acid linkage also occurs in a minority of the repeating disaccharide units of dermatan sulfate (24) due to incomplete epimerization of glucuronic to iduronic acid during biosynthesis (25). Treatment of porcine skin dermatan sulfate wit.h chondroitinase AC results in limited cleavage of the polymer (-5%) even after prolonged incubation (data not shown). This rules out the possibility that the chondroitinase AC used in our laboratory was contaminated with chondroitinase ABC. Moreover, the ability of porcine skin dermatan sulfate to activate HCII decreased -5% after treatment with chondroitinase AC (data not shown). Thus, we believe that the effect of chondroitinase AC on the fibroblasts and the supernatant medium is most likely due to partial degradation of dermatan sulfate.
We confirmed the importance of dermatan sulfate by assaying the proteoglycans purified from IMR-90 fibroblasts. Vogel and Peterson (22) have previously characterized the [3sSS]sulfate-labeled proteoglycans synthesized by these cells. The gel filtration patterns they observed for the medium and cell-associated proteoglycans in the presence of guanidine HCl were very similar to those obtained in our experiments (Fig.  5, A and B). In addition, the glycosaminoglycan compositions of the individual peaks were similar, except that we observed a higher proportion of dermatan sulfate in peak C- 11. Vogel and Peterson (22) reported that the dermatan sulfate proteoglycan from conditioned medium (corresponding to "11, Fig.  5A) contains 2-4 dermatan sulfate chains (M, = 25,000-40,000) linked to a core protein. In contrast; their data suggested that dermatan sulfate in the cell extract of IMR-90 fibroblasts (corresponding to C-11, Fig. 5B) may represent either a single chain linked to a small core protein or free glycosaminoglycan chains (M, = 28,000). Despite these differences in structure, M-I1 and C-I1 have comparable specific activities with HCII (Table 111).
The increase in the rate of formation of the lZ5I-thrombin-HCII complex observed in our assay system with cell monolayers is relatively modest (2.2-7.5-fold) compared to that observed with optimal concentrations of purified dermatan sulfate (1300-fold at 250 pg/ml) or heparin (800-fold at 67 pg/ ml) (8). The degree of acceleration of the thrombin-HCII reaction produced by the cell monolayers is approximately equal to that produced by 0.1-1.0 pg/ml purified porcine skin dermatan sulfate (not shown). However, local concentrations of dermatan sulfate in vivo may be much higher due to a greater ratio of cell surface (or extracellular matrix) to extracellular fluid volume. Vogel and Sapien (26) found that maintaining IMR-90 fibroblasts in 0.5% newborn calf serum resulted in viable but nonproliferating cells which secrete 40% more dermatan sulfate and proportionally less heparan sulfate compared to cells grown in 10% serum. They speculated that low serum concentrations may better approximate in vivo conditions. No acceleration of the thrombin-HCII reaction occurs in the presence of human umbilical vein endothelial cells or mouse macrophage-derived cells. Although Oohira et al. (27) found that dermatan sulfate accounts for 25% of the [Y3] sulfate-labeled proteoglycans secreted by human umbilical vein endothelial cells, their methods were not sensitive enough to determine whether or not dermatan sulfate was present in an SDS extract of the cells. Therefore, it is possible that there is no dermatan sulfate on the cell surface. In the same paper, analysis of cultured bovine aortic endothelial cells, which also secrete dermatan sulfate, showed no dermatan sulfate in an SDS extract of the cells.
We have not observed acceleration of '251-thr~mbin-ATIII complex formation in the presence of endothelial cells (data not shown). This may be due to the low surface (2 cm2) to volume (0.5 ml) ratio used in our assay system. Other investigators have shown acceleration of thrombin inhibition by ATIII using systems with large surface areas such as perfusion through a column of endothelial cells grown on microcarrier beads (11) or perfusion through a microvascular preparation (11,28). In the study by Marcum et al. (28) approximately a 19-fold rate enhancement of thrombin inhibition by ATIII was seen by perfusion through the blood vessels of a rat hindquarter preparation. This rate enhancement could be abolished by first perfusing the system with heparinase to degrade cell surface heparan sulfate. Heparan sulfate molecules purified from cloned bovine aortic endothelial cells have been shown to activate ATIII (29).
In this study, we have identified a potential physiologic site for the inhibition of thrombin by HCII. A break in the integrity of the endothelial surface would expose these proteins to fibroblasts and smooth muscle cells and their extracellular matrices which could increase the rate of thrombin inhibition by HCII. Platelet factor 4, a glycosaminoglycanbinding protein released by platelets upon degranulation (30,31), is a potential negative regulator of this process, since platelet factor 4 prevents dermatan sulfate from accelerating the inhibition of thrombin by HCII i n vitro (32).
The mitogenic effect of thrombin on human fibroblasts is dependent on its esterolytic activity (33). Since HCII inhibits both the proteolytic and esterolytic activities of thrombin (7), formation of the thrombin-HCII complex could block thrombin-induced mitogenesis. Thrombin is also mitogenic for macrophage-derived cells, but this effect is independent of esterolytic activity. However, thrombin-induced mitogenesis in these cells can be blocked by the thrombin inhibitor hirudin (34). The chemotactic effect of thrombin on monocytes and macrophages can be inhibited by formation of a thrombininhibitor complex with either hirudin or ATIII (35). Hence, it is possible that HCII may inhibit thrombin-induced macrophage chemotaxis or mitogensis. We suggest that HCII may inhibit thrombin in the subendothelial or extravascular tissues, thereby modulating thrombin's hemostatic, mitogenic, or chemotactic effects.