Characterization and Hormonal Modulation of Anticoagulant Heparan Sulfate Proteoglycans Synthesized by Rat Ovarian Granulosa Cells*

Anticoagulant heparan sulfate proteoglycans endow the vascular endothelium with antithrombotic properties, but their role outside the vascular bed is unknown. Granulosa cells form an avascular compartment in the ovarian follicle, in which a heparin-like activity has been described. At ovulation extravascular coagulation occurs around ovulatory follicles, and after expulsion of the oocyte, a fibrin clot forms in the antral cavity. Granulosa cells synthesize two major heparan sulfate proteo- glycans, whose anticoagulant nature has not been inves-tigated. The purpose of this study was to characterize anticoagulant heparan sulfate proteoglycans synthesized by rat ovarian granulosa cells. Affinity purified 35 S-labeled anticoagulant heparan sulfate glycosaminoglycans represent 6.5% of the total heparan sulfate syn- thesized, and they contain 13% 3- O -sulfated disaccharides that are markers of the antithrombin-binding site of heparin. The biological activity of granulosa cell heparan sulfate proteoglycans was demonstrated by their ability to bind antithrombin and to accelerate the formation of thrombin-antithrombin complexes. The im- pact of hormonal stimulation on granulosa cell anticoagulant heparan sulfate proteoglycans was studied using 125 I-antithrombin binding assays. Follicle-stimulating basal and FSH-stimulated conditions. These results indicate that the 40% decrease in total cell-bound aHSPGs observed in FSH-stimu-lated granulosa cells does not involve GPI-anchored aHSPGs but is probably due to a decrease in membrane-spanning was found to decrease the shed-ding of HSPGs from granulosa cells into the medium (23). We tested whether serine proteases were involved in the release of aHSPGs from granulosa cells under basal or FSH-stimulated conditions. We observed that the tryptic inhibitor aprotinin did not affect the release of aHSPGs in granulosa cell culture medium under basal or FSH-stimulated conditions (data not shown). These observations suggest that other types of proteases might be responsible for the release of aHSPGs. Experiments are currently underway in our laboratory to investigate this matter.

Anticoagulant heparan sulfate proteoglycans endow the vascular endothelium with antithrombotic properties, but their role outside the vascular bed is unknown. Granulosa cells form an avascular compartment in the ovarian follicle, in which a heparin-like activity has been described. At ovulation extravascular coagulation occurs around ovulatory follicles, and after expulsion of the oocyte, a fibrin clot forms in the antral cavity. Granulosa cells synthesize two major heparan sulfate proteoglycans, whose anticoagulant nature has not been investigated. The purpose of this study was to characterize anticoagulant heparan sulfate proteoglycans synthesized by rat ovarian granulosa cells. Affinity purified 35 S-labeled anticoagulant heparan sulfate glycosaminoglycans represent 6.5% of the total heparan sulfate synthesized, and they contain 13% 3-O-sulfated disaccharides that are markers of the antithrombin-binding site of heparin. The biological activity of granulosa cell heparan sulfate proteoglycans was demonstrated by their ability to bind antithrombin and to accelerate the formation of thrombin-antithrombin complexes. The impact of hormonal stimulation on granulosa cell anticoagulant heparan sulfate proteoglycans was studied using 125 I-antithrombin binding assays. Folliclestimulating hormone induced a redistribution of anticoagulant heparan sulfate proteoglycans from the granulosa cell layer to the culture medium, indicating that their distribution could be modulated according to the stage of follicular development. These results suggest that anticoagulant heparan sulfate might be critically located in the follicle to maintain fluidity around the oocyte until its expulsion at ovulation.
Certain cell types synthesize a subset of heparan sulfate proteoglycans (HSPGs) 1 that bind and activate antithrombin (AT) (1,2). Like heparin, anticoagulantly active HSPGs (aHSPGs) contain a pentasaccharide with a specific AT-binding monosaccharide sequence (3)(4)(5). aHSPGs are synthesized by endothelial cells, and they accumulate in the basement mem-brane underlying the vascular endothelium (1,6). aHSPGs have also been detected in Reichert's membrane during embryonic development and in cultured fibroblasts (2,7). In addition, aHSPGs are synthesized by a variety of cell lines of fibroblastic or epithelial origin (8 -11). Thus, the biosynthesis of aHSPGs appears not to be restricted to endothelial cells but also to occur selectively in extravascular cell types.
aHSPGs have been postulated to endow the vascular endothelium with antithrombotic properties (6,12,13), but their function outside the vascular bed is unknown. To study the nature and function of aHSPGs outside the vascular bed, we have analyzed aHSPGs synthesized by ovarian granulosa cells.
Three distinct features underline the importance of ovarian granulosa cells as a model to study extravascular aHSPGs. First, granulosa cells, together with the oocyte, form an avascular compartment separated from outer theca layers by a thick basement membrane until ovulation (14). Second, the differentiation of granulosa cells is under hormonal control, and cultured granulosa cells can be stimulated to differentiate in vitro by follicle-stimulating hormone (FSH) (14,15). This profoundly affects granulosa cell metabolism and increases their synthesis of estradiol, plasminogen activators, and proteoglycans (15)(16)(17)(18). Third, ovulation is accompanied by a local inflammation that involves vascular permeabilization and fibrin deposition (19,20), a process that could be regulated by extravascular aHSPGs.
Heparin-like activity has been detected in porcine and rat follicular fluid and in rat granulosa cells by Andrade-Gordon and collaborators (21). These authors have shown that high concentrations of heparin-like glycosaminoglycans (GAG) isolated from follicular fluid accelerate thrombin inhibition by AT and that rat granulosa cell extracts accelerate the formation of thrombin-AT complexes. However, the species responsible for this activity have not been further purified.
Valuable data are available on granulosa cell total HSPGs, due to the very comprehensive characterization work of Yanagishita and Hascall (22)(23)(24). These authors have purified granulosa cell HSPGs and studied their biosynthesis and catabolism, and they have demonstrated the synthesis of two predominant types of HSPGs (22)(23)(24). One is a membranespanning HSPG that is partially released in soluble form; the other is anchored to the membrane via a glycosylphosphatidylinositol (GPI) moiety and is removed from the cell surface by internalization (25,26). These HSPGs were, however, described without reference to anticoagulant activity.
In the present study we have characterized the heparin-like activity of rat ovarian granulosa cells. We have purified and chemically analyzed 35 S-labeled aHSPGs from primary cultures of rat granulosa cells. We report that granulosa cells synthesize aHSPGs in amounts comparable with endothelial cells. Furthermore, we have studied aHSPG expression in granulosa cell monolayers and culture media using 125 I-AT binding assays. This setting was used to assess the effects of stimulations by FSH on aHSPGs expression by granulosa cells.

Materials
Purified rat FSH was a kind gift from Prof. M. L. Aubert (Division of Biol. Croissance et Reproduction, Geneva University, Switzerland). Human chorionic gonadotropin was from Serono. Diethylstilbestrol was obtained from Asta Medica (Switzerland). Bovine vitronectin, chicken serum, and tissue culture reagents were purchased from Life Technologies, Inc. Cloned rat epididymal fat pad microvascular endothelial cells and a subclone of murine fibroblastic L-cells that produce aHSPGs were kindly provided by Prof. R. D. Rosenberg (Massachusetts Institute of Technology, Cambridge). Purified human thrombin was a generous gift from Dr. J. Fenton (New York State Department of Health). Purified human AT was obtained from Cutter Biological. 125 I-AT was prepared as previously published with a specific activity of 5 ϫ 10 6 cpm/ng (27

Animals
In vivo stimulation of follicular development was performed according to published procedures (31,32). Briefly, immature 21-day-old female Sprague-Dawley rats, purchased from Iffa-Credo (L'Arbresle, France), were treated daily by subcutaneous injections of 1 mg of diethylstilbestrol in sesame oil for 4 days to induce granulosa cell proliferation and were sacrificed by decapitation at day 25.

Granulosa Cell Preparation
Ovaries were dissected and placed in McCoy's medium containing 100 units/ml penicillin, and 100 g/ml streptomycin. Using a dissecting microscope, granulosa cells were released by puncturing the large follicles protruding from the ovary surface with 30-gauge needles according to published procedures (33,34). After removal of remnant ovarian tissue, granulosa cells were pelleted by centrifugation (70 ϫ g, 20 min) and resuspended in medium, and an aliquot was diluted 1:1 with trypan blue to determine viability and cell number in a hemocytometer. One ovary yielded about 1 ϫ 10 6 viable granulosa cells.

Granulosa Cells Culture
Precoating of Culture Dishes-Fetal bovine serum desensitizes granulosa cells to stimulation by gonadotropins (35). Serum-free conditions have therefore been established for granulosa cell cultures, which are used for FSH bioassay (36). In the absence of serum, granulosa cells attach to the culture dish but remain rounded and do not spread, a condition that can alter their proteoglycan synthesis (37). To allow granulosa cells to spread, we have used culture dishes precoated for 2 h at 37°C with 2% chicken serum or 1 g/ml purified bovine vitronectin and subsequently incubated the cells in serum-free medium. We verified that either coating was compatible with FSH stimulation of granulosa cells, as evidenced by increased estradiol excretion, and with HSPG synthesis by comparison to cells grown in the presence of 10% fetal bovine serum (not shown).
Estradiol was measured in granulosa cell media by radioimmunoassay as described (38).

Metabolic Labeling of Granulosa Cells
Granulosa cells were seeded in chicken serum-coated wells in labeling medium at 0.26 ϫ 10 6 cells/cm 2 and incubated for 2 h at 37°C to allow cell attachment, and subsequently incubated for 24 h in labeling medium containing 400 M Na 2 [ 35 S]SO 4 (0.25 Ci/mmol) and supplemented with 50 ng/ml FSH (except for nonstimulated cultures as stated below). The spent medium was collected, and the cells were labeled for an additional 24 h with fresh labeling medium. The spent media were freed from floating cells by centrifugation, boiled at 100°C for 5 min, filtered through a 0.45-m filter, and kept frozen at Ϫ20°C until used. The pooled spent media are referred thereafter as labeled conditioned media. At the end of the 48-h labeling, the cells were washed twice with phosphate-buffered saline, and cell layer-associated GAGs were released by treatment with 0.05% trypsin, 0.53 mM EDTA as described (9). 35 S-HS-Granulosa cell 35 S-labeled heparan sulfate chains ( 35 S-HS) from conditioned media and trypsinates of cells cultured with or without 50 ng/ml FSH were purified in parallel for analysis of HS and chondroitin/dermatan sulfate (CS) distribution, and the purification was followed by liquid scintillation counting. Total 35 S-labeled GAG chains were isolated by ion exchange chromatography on DEAE-Sephacel. The labeled sample was loaded in 150 mM NaCl, 50 mM Tris, pH 7.4; the gel was washed with 10-column volumes of the same buffer followed by 2-column volumes of 150 mM NaCl, 50 mM sodium acetate, pH 5.0, and 35 S-GAGs were eluted with 1 M NaCl, 50 mM Tris, pH 7.4. The 35 S-GAGs were cleaved from proteoglycans (conditioned media) or peptide stubs (trypsinate) by ␤-elimination, proteins removed by phenol extraction, and GAGs concentrated by ethanol precipitation (39). The relative content of HS and CS was determined analytically in purified 35 S-GAGs as ethanol-soluble 35 S counts generated by the respective activities of chondroitinase ABC (0.1 unit/ml, 37°C) and Flavobacterium heparitinase (0.5 units/ml, 43°C) as described (39). Digestion by both enzymes quantitatively degraded over 95% of the 35 S-GAGs (data not shown). Alternatively, 35 S-HS chains were isolated after degradation of CS with chondroitinase ABC (0.1 unit/ml) at 37°C for 6 h, followed by phenol extraction and ethanol precipitation. Completion of CS degradation was ascertained by a second incubation with chondroitinase ABC, which did not release additional degradation products (data not shown).

Preparative Purification of 35 S-HS Chains
Granulosa cells obtained from 100 rats (about 200 ϫ 10 6 cells) were metabolically labeled in labeling medium containing 80 M Na 2 [ 35 S]SO 4 (1.25 Ci/mmol) for 48 h as described above; conditioned media and trypsinates were pooled, and 35 S-HS was purified to generate sufficient material for disaccharide analysis. The labeled pools were sequentially treated by pronase, papain, and chondroitinase ABC to eliminate proteins and chondroitin sulfate chains as described (9). 35 S-HS chains were isolated by DEAE-Sephacel chromatography as described above and the eluate cleaved from residual peptides by ␤-elimination (39). Proteins were removed by phenol extraction, and GAGs were dialyzed extensively against 10 mM NH 4 HCO 3 and evaporated to dryness in a speed-vac concentrator.

Isolation of aHS by Affinity on Immobilized AT
aHS were isolated from anticoagulantly inactive HS (iHS) by AT affinity on concanavalin A-Sepharose as described (40). Briefly, aHS-AT complexes were formed by incubating total HS chains with 2.5 M AT in 150 mM NaCl, 10 mM Tris-HCl, pH 7.4, containing 10 M dextran sulfate (M r 8000), 0.002% Triton X-100, and 1 mM of each CaCl 2 , MgCl 2 , and MnCl 2 for 1 h. The samples were subsequently admixed to a suspension of concanavalin A-Sepharose equilibrated in the same buffer; aHS-AT complexes were bound to concanavalin A, and iHS were removed by washing. aHS were eluted from the gel by dissociation of the aHS-AT complexes in buffer containing 1 M NaCl, and aHS and iHS content (% aHS and % iHS) were quantified by scintillation counting.
Using iHS the assay background was determined to be less than 0.2% of input radioactivity.
Superose 6 -The size distribution of granulosa cell total HS, aHS, and iHS was analyzed by gel filtration on a Superose 6 column using a FPLC system (Pharmacia). The column was run in phosphate-buffered saline, pH 7.2, and was loaded with about 15,000 cpm of labeled GAGs. 0.5-ml fractions were collected and GAGs quantified by scintillation counting. The column was calibrated using dextran sulfate M r 500,000 (K av ϭ 0) and free Na 2 [ 35 S]SO 4 (K av ϭ 1).
PAGE-GAGs were electrophoresed on polyacrylamide gradient (5%-15%) gels without SDS, in buffer containing 0.1 M NaCl (41,42) and stained using silver-enhanced azure A (41). GAG molecular weight standards were heparin (M r 16,400), chondroitin sulfate A (M r 21,600), heparan sulfate I (M r 15,000), and heparan sulfate III (M r 7,000) (43). 1 g of unlabeled GAGs or 35,000 cpm of 35 S-HS chains were loaded per lane, and migration profiles of 35 S-HS were recorded with a Phosphor-Imager analyzer (Molecular Dynamics, CA). Modal R F of GAGs were measured by densitometry scanning of the stained gel or of Phosphor-Imager analysis of 35 S counts. Granulosa cell HS molecular weight was determined by extrapolation from the linear regression of the standard GAGs R F and log molecular weight.

Glucosamine Content and Specific Activity of 35 S-HS
The molar content in glucosamine of purified 35 S-HS was determined by HPLC. The samples were hydrolyzed in 2 M HCl for 4 h at 100°C, derivatized with ortho-phthaldehyde, and analyzed in a C18 HPLC amino acid analyzer. 35 S was measured in aliquots of the same samples by scintillation counting, and the specific activity of the samples was calculated in dpm 35 S/pmol sulfate.

Disaccharide Analysis
Granulosa cell aHS and iHS were degraded to disaccharides by deacetylation by hydrazinolysis followed by high pH (4.5) nitrous acid and then low pH (1.5) nitrous acid treatments as described previously (1,44). The distribution of sulfated species was analyzed by reversephase ion pairing HPLC on a C18 column (0.46 ϫ 24 cm, Vydac); 0.5-ml fractions were collected, and radioactivity was quantified in a scintillation counter. Similar to the procedure of Guo and Conrad (45), the samples were eluted at 0.5 ml/min with 1 mM tetrabutylammonium phosphate, pH 3.6, containing acetonitrile at 4.8% (45 min), 9% (15 min), and 11.7% (40).

Affinity Coelectrophoresis
The binding of aHS to AT was analyzed by affinity coelectrophoresis according to Lee and Lander (46). At neutral pH the electrophoretic mobility of GAGs is much higher than that of AT and the binding of AT to GAGs retards their migration. Three different 35 S-labeled GAG preparations were loaded in separate transverse slots close to the cathode, each slot facing three rectangular saggital wells in which AT was casted. AT concentrations were 0 (control), 30, and 500 nM, respectively. Granulosa cell 35 S-iHS (65,000 cpm) and 35 S-aHS (15,900 cpm) and endothelial cell 35 S-aHS (27,200 cpm) (47) were electrophoresed through the wells containing AT at 60 V and 200 mA for 5 h. Electrophoresis end points were determined by the migration of bromphenol blue, which migrates just after HS. The gel was air-dried and autoradiographed for 48 h.

SDS-PAGE Analysis of Thrombin-AT Complex Formation
Thrombin (50 nM) was incubated for various times with 125 I-AT (2 nM) in the absence or in the presence of heparin (120 ng/ml, 7.3 nM) or granulosa cell HS (aHS, 49 ng/ml; iHS, 426 ng/ml) at 37°C in 0.15 M NaCl, 10 mM Tris, pH 7.5. The reaction was stopped by addition of Phe-Pro-Arg chloromethyl ketone (40 M), and SDS-PAGE loading buffer was added. The samples were submitted to SDS-PAGE on a 7.5% gel in reducing conditions; 125 I-AT was revealed with a PhosphorImager apparatus, and the signal was quantified using the ImageQuant software.

I-AT Cell Binding Assay
Granulosa cells were seeded in 96-well plates and kept in supplemented McCoy's medium for 48 h prior to the assay. 125 I-AT cell binding assay was performed in triplicate wells as described (6,27). The protein content of control wells was measured using the BCA protein assay (Pierce) and used to normalize the values of bound 125 I-AT that were expressed as cpm/mg protein.

I-AT Ligand Binding Assay
125 I-AT ligand binding assay was performed on conditioned media collected from the cells used for 125 I-AT cell binding assay. Conditioned media from triplicate wells were collected after 48 h incubation with granulosa cells, and floating cells and debris were removed by centrifugation (70 ϫ g, 10 min) prior to adsorption of soluble aHSPGs on nitrocellulose. 125 I-AT ligand binding assay was performed as described previously (27). The results were expressed as cpm/mg proteins.

Phospholipase C Treatments
Cultured granulosa cells were treated with phospholipase C to release GPI-anchored aHSPGs from the cell surface according to Yanagishita (25,26). Briefly, the cells were cultured for 48 h as described above and subsequently incubated with 0.4 units/ml phospholipase C in culture medium for 30 min at 37°C. The supernatant was collected, cleared of floating cells and debris by centrifugation, and assayed for aHSPGs by 125 I-AT ligand binding assay. In parallel, cellbound aHSPGs were assayed before and after phospholipase C treatment by 125 I-AT cell binding assay. The amounts of bound cpm found on intact granulosa cells was comparable with the sum of the counts bound to phospholipase-treated cells and to the counts bound to enzyme supernatant, indicating that aHSPGs were quantitatively recovered in this procedure (data not shown).

Statistical Analysis
Comparison of data obtained from cells grown in absence or in presence of FSH was done by Student's paired t test. p Ͻ 0.05 was considered significant.

Molecular Size Distribution of Granulosa Cell 35 S-HSPGs-
Primary rat granulosa cells were cultured in serum-free medium to preserve their sensitivity to hormone stimulation. To verify that the size of HSPGs synthesized is not affected by the absence of serum in the culture medium, we analyzed the size distribution of HSPGs released by granulosa cells. Granulosa cells were metabolically labeled for 48 h with Na 2 [ 35 S]SO 4 (0.25 Ci/mmol) as described; 35 S-HSPGs were purified from conditioned media by DEAE-Sephacel chromatography, and CS were degraded by chondroitinase ABC treatment. 35 S-HSPGs, loaded on a Sepharose CL-4B gel filtration column, eluted as a major peak of HSPGs at K av 0.4 -0.6 and a minor peak of degradation products at K av 0.8 (Fig. 1). This size distribution is comparable with that reported by Yanagishita and Hascall (22) for HSPGs from granulosa cells cultured in medium con- taining 10% fetal bovine serum, indicating that our serum-free culture conditions do not affect the general size of HSPGs produced by granulosa cells.
Isolation of Granulosa Cell aHS and iHS- 35 S-HS chains were purified from granulosa cell layers and culture media and subsequently fractionated into aHS and iHS according to their affinity for AT as described under "Experimental Procedures." The aHS content of granulosa cell HS chains was of 6.5 Ϯ 0.4% (mean Ϯ S.D., n ϭ 3) of the total HS chains, and iHS constituted 93.5 Ϯ 0.4% of total HS, while the assay background was less than 0.2%. Comparable values were obtained for a control fibroblastic cell line (8.0 Ϯ 0.8% aHS, n ϭ 3), and endothelial cells have been shown to produce 1-10% aHS chains (48). This suggests that cultured rat granulosa cells synthesize considerable amounts of aHS.
Size Distribution of Granulosa Cell 35 S-Labeled aHS and iHS-We have examined the molecular size distribution of total 35 S-labeled HS, aHS, and iHS from granulosa cells by gel filtration on Superose 6. Fig. 2 shows that total HS, aHS, and iHS have a similar elution profile, with a major peak eluting at K av 0.52, at a higher molecular weight than heparin which is eluted at K av 0.68 (arrow). Low molecular weight oligosaccharides were present in the total HS and in iHS but were much less prominent in aHS. Therefore, HS, aHS, and iHS from granulosa cells have a similar size and are bigger than heparin. The fact that aHS and iHS have the same size distribution allowed us to determine the molecular weight of granulosa cell aHS using unfractionated HS chains, as detailed below.
Determination of Granulosa Cell HS Molecular Weight-The molecular weight of granulosa cell HS chains was determined by PAGE as described under "Experimental Procedures" using calibrated molecular weight standards of heparin, heparan sulfate, and chondroitin sulfate A with M r ranging from 7,000 to 21,600. Fig. 3 shows the PhosphorImager analysis of the migration pattern of granulosa cell 35 S-HS (35 000 cpm, 74 ng). Granulosa cell HS appear broadly smeared, reflecting a high polydispersity in molecular weight concordant with the broad peak observed by gel filtration in Fig. 2. Densitometry scanning of the tracing yielded one symmetric peak with a maximum R F of 0.4 and some tailing toward the bottom of the gel. In two independent preparations, the modal R F of granulosa cell HS corresponded to an average modal M r of 56,800 Ϯ 7,300. These data are confirmed by the analysis of one of these preparations by gel filtration techniques and by PAGE, using different GAG standards, which gave an average M r of 58,900 Ϯ 8,400 2 (42).
Degree of Sulfation of Granulosa Cell HS-The extent of sulfation of granulosa cell HS chains was measured by determining the molar content in glucosamine of 35 S-labeled HS. The ratio of SO 3 Ϫ /glucosamine was calculated using the Na 2 [ 35 S]SO 4 specific activity used for labeling granulosa cells, and we obtained ratios of 0.7 and 1.0 SO 3 Ϫ /glucosamine in two independent preparations of granulosa cell 35 S-HS. Using an average ratio of 0.85 sulfate per glucosamine and assuming that 15% of the disaccharides contain N-acetylated glucosamine, 3 we estimated the average molecular weight of HS disaccharides produced by granulosa cells to be 436 and used this value to calculate the number of disaccharides per HS chain and to convert 35 S counts in pmol and ng HS, respectively. We estimate that granulosa cell HS of M r 56,800 (as reported in the  (Table I). These results were confirmed in two independent preparations of primary granulosa cell HS that gave similar values. These data demonstrate that rat ovarian granulosa cells synthesize aHSPGs. The binding of granulosa cell aHS to AT and their capacity to activate the protease inhibitor were examined and are described in the next two sections.
Specific Binding of aHS to AT: Analysis of aHS-AT Complexes by Affinity Coelectrophoresis-The complexes formed between aHS and AT were visualized by affinity coelectrophoresis (Fig. 4); 35 S-labeled aHS and iHS from granulosa cells and aHS from endothelial cells were loaded on an agarose gel containing various amounts of AT cast in the gel. The binding of aHS to AT retarded their migration and is observed in aHS from granulosa cells and endothelial cells but not in iHS. Granulosa cell aHS formed complexes with AT at 30 and at 500 nM, with similar degrees of retardation as endothelial cell aHS, suggesting that the affinity of aHS from both cell types to AT is similar. Moreover, aHS from both cell types behaved homogeneously in the presence of AT, without smearing of the bands between the retarded and nonretarded positions, suggesting that all AT-binding species bound with high affinity.
Functional Activity of Granulosa Cell aHS in Accelerating the Formation of Complexes between Thrombin and AT-The biological activity of granulosa cell aHS, namely the activation of AT increasing its reactivity toward thrombin, was demonstrated in a purified system by SDS-PAGE. Thrombin (50 nM) was incubated for various times with 125 I-AT (2 nM, 0.5 ϫ 10 6 cpm/ml) in the absence or presence of GAGs, and the products of the reaction were analyzed by SDS-PAGE. At time t ϭ 0, AT migrates as a single band of M r 58,000 (Fig. 5A, lane 1). In control samples, incubated in the absence of GAGs, the formation of high molecular weight thrombin-AT complex occurs slowly, and only trace amounts are visible after 10 min of incubation (Fig. 5A, lanes 2-4). When thrombin and AT are incubated in the presence of granulosa cell aHS (0.86 nM, 49 ng/ml), the appearance of thrombin-AT complex is accelerated; the complex is present after 2 min (14%), and its amount is moderately increased after 10 min (16%) (Fig. 5A, lanes 5-7). Together with the formation of high molecular weight complex, an additional low molecular weight band appears, migrating under native AT, which corresponds to the cleaved form of AT (modified AT) (49). In contrast, granulosa cell iHS (7.5 nM, 426 ng/ml) do not accelerate the formation of thrombin-AT complex (Fig. 5A, lanes 8 -10), and the pattern observed is identical to that of control samples without GAG added. In order to illustrate the anticoagulant activity of granulosa cell HS, we have incubated thrombin and AT with the aHS and iHS obtained from 123,000 granulosa cells, and therefore, the amounts of iHS (8.5 ng) used in these experiments were about 10-fold higher than that of aHS (1.0 ng). To estimate the anticoagulant  activity of granulosa cell aHS, we quantified the fraction of 125 I-AT present as thrombin-AT complex formed after 2 min of incubation by densitometry (Fig. 5B). In control conditions, without GAG added, no thrombin-AT complex was detectable (lane 1), whereas in the presence of heparin (120 ng/ml, 7.3 nM, 14 ϫ 10 Ϫ3 USP units/ml) 6% of the AT was complexed to thrombin (lane 2). When granulosa cell aHS (49 ng/ml, 0.86 nM) was added to the reaction mixture 14% of the protease inhibitor was present as high molecular weight complex (lane 3), but in the presence of granulosa cell iHS (426 ng/ml, 7.5 nM) complexed 125 I-AT was at the limit of detection (0.14%) (lane 4). These data demonstrate that granulosa cell aHS are biologically active and accelerate the rate of thrombin inactivation by AT similar to heparin.
Having demonstrated the synthesis of aHSPGs by rat ovarian granulosa cells, we next examined the effects of hormonal stimulations on granulosa cell HSPGs. First, we analyzed the total 35 S-labeled GAGs synthesized by granulosa cells and compared their composition in HS/CS under basal conditions and after stimulation with FSH. Second, we used analytical detection of aHSPGs by 125 I-AT binding assays, on cell surfaces and on soluble aHSPGs released in the culture media, to further analyze the effect of FSH on granulosa cell aHSPGs. I-AT prevented its binding, and only 12 Ϯ 8% residual binding was observed. In contrast, a similar treatment with chondroitinase ABC (0.1units/ml) did not alter 125 I-AT binding (82 Ϯ 17%). These values are very similar to those obtained for endothelial cells (6).

HS/CS Composition of Cell-bound and Soluble Granulosa Cell GAGs: Effect of FSH Stimulation-We
Thus, granulosa cell aHSPGs can be measured by 125 I-AT binding to surface-associated aHSPGs and to soluble aHSPGs. These assays were used to study the effects of hormonal stimulation of granulosa cells on their aHSPGs production.
Stimulation of Granulosa Cells by FSH: Effects on aHSPGs-Granulosa cells were cultured with or without 50 ng/ml FSH for 48 h. At the end of the incubation, aHSPGs were evaluated in conditioned media by 125 I-AT ligand-binding assay and on the cell monolayers by 125 I-AT cell-binding assay. Alternatively, control wells for each condition were used to measure the levels of estradiol secreted in the media and the cell's protein contents. The results of eight independent experiments are presented in Fig. 6. The effectiveness of FSH stimulation was verified by measuring the increase in estradiol secretion. The results show that granulosa cells respond to FSH stimulation by increasing their estradiol secretion (Fig. 6A); in the presence of FSH granulosa cells increase the amounts of soluble aHSPGs they release in the medium (Fig. 6B), and they concomitantly decrease their cell surface-bound aHSPGs (Fig.  6C). Statistical analysis showed highly significant differences between stimulated and nonstimulated conditions for soluble and cell-bound aHSPGs (Student paired t test, p Ͻ 0.01). When the partition between cell-bound and soluble aHSPGs was examined using normalized values, about 77% aHSPGs was present on the cell surfaces and 23% in the media, under basal conditions. In contrast, in FSH-stimulated conditions approxi- mately 50% aHSPG were detected on the cell layers and similar amounts in the media. These data indicate that FSH alters the partitioning of aHSPGs between cell-surface and culture media and favors the liberation of cell-bound HSPGs. To analyze whether granulosa cell GPI-anchored HSPGs were involved in aHSPGs expression and release in response to FSH, granulosa cells were incubated with phospholipase C.
Granulosa Cell GPI-anchored aHSPGs in Basal and FSHstimulated Conditions-Granulosa cell GPI-anchored HSPGs were released with phospholipase C. We quantified cell-bound aHSPGs before and after phospholipase treatment by 125 I-AT cell binding assay and GPI-anchored aHSPGs released in the enzyme incubation media by 125 I-AT ligand binding assay. In three independent experiments, aHSPGs released from the cell surface by phospholipase C represented 16 Ϯ 6% (mean Ϯ S.D.) of the cell surface aHSPGs and were quantitatively recovered in the enzyme supernatant, demonstrating that aHS chains are present on granulosa cell GPI-anchored HSPGs.
We then compared the amounts of total cell-bound aHSPGs and of GPI-anchored aHSPGs present on granulosa cells in basal and FSH-stimulated conditions (Fig. 7). The data were normalized to the total cell-bound aHSPGs present in the absence of FSH. The proportion of aHSPGs released by phospholipase C from granulosa cell surfaces did not vary between basal and FSH-stimulated conditions (16 Ϯ 6% and 20 Ϯ 7% respectively), despite the fact that the total aHSPGs were decreased by 40% in the presence of FSH. DISCUSSION The data presented in this study provide compelling evidence that considerable amounts of biologically active aHSPGs are synthesized by ovarian granulosa cells, which constitute an avascular compartment in the ovarian follicle in vivo. The aHS GAG chains synthesized by rat ovarian granulosa cells constitute 6.5% of their total HS, and aHS biological activity was shown by their ability to specifically complex AT and to accelerate the inactivation of thrombin by AT. Stimulation of cultured granulosa cells by FSH altered the distribution of aHSPGs and favored the release of soluble aHSPGs in culture media, suggesting that granulosa cells are able to regulate the localization of their aHSPGs according to the hormonal context and thus to their differentiation stage.
The presence of aHSPGs has been reported in the vascular bed (6, 50 -54), and structural analysis of aHS has been carried out for endothelial cells (1,47), nonvascular cell types like fibroblast and epithelial cell lines (8,9), and for Reichert's embryonic membrane (2). We set out to characterize the aHS synthesized by primary granulosa cells to unequivocally demonstrate their presence in an extravascular compartment. We used serum-free culture conditions, suitable to induce granulosa cell differentiation by FSH stimulation, and have shown that under these conditions the size distribution of HSPGs is similar to that reported by Yanagishita and Hascall (22) for granulosa cells grown in the presence of serum and to those reported for HSPGs from endothelial and fibroblastic cell lines that synthesize aHSPGs (27,47). We next isolated 35 S-labeled HS chains from granulosa cells stimulated by FSH, a condition known to increase the biosynthesis of proteoglycans (17,55).
Granulosa cell aHS GAG chains were fractionated according to their AT affinity and represented 6.5% of the total HS chains, a value comparable with that found in various types of endothelial cells (1,47,48). aHSPGs, constituting about 1% of bovine aortic endothelial cell HSPGs, were found to be abundant in situ in the aortic wall as evidenced by 125 I-AT binding (1,6). By analogy it seems that granulosa cells synthesize aHSPGs in sufficient amounts to justify a biological activity in vivo.
The size distribution of aHS chains was similar to that of the general HS population, and extensive size poly-dispersity was apparent both on gel filtration and on PAGE. The modal M r of granulosa cell HS was about 57,000, and this value was confirmed using two independent sets of GAG molecular weight standards (42). This modal chain size is higher than the M r of 30,000 estimated by Yanagishita and Hascall (56) using gel chromatography, and this discrepancy might be due to differences in the methods used for molecular weight determination. However, chain sizes of 30,000 were also encompassed in the population of chains resolved by PAGE. In parallel, we determined that granulosa cell HS disaccharides contain about 0.85 sulfate/disaccharide, and we calculated that granulosa cell HS contain approximately 130 disaccharides per GAG chain. For comparison, mouse L cells were shown to produce HS containing about 0.45 sulfate/disaccharide with a modal M r of 53,000. Moreover, Reichert's membrane HS, which are constituted predominantly of aHS chains, were estimated to contain a sulfate/ disaccharide ratio of 1, whereas basement membrane HS from EHS cells were mostly nonsulfated with a sulfate/disaccharide ratio of about 0.25 (2). Therefore, granulosa cells seem to produce relatively highly sulfated HS species (0.85 sulfate/disaccharide), resembling those of Reichert's membrane aHS, rather than HS from established cell lines.
The pentasaccharide constituting the AT-binding site of heparin and aHS contains a cardinal 3-O-sulfated glucosamine essential for AT binding. We found that granulosa cell aHS contain markedly increased amounts of 3-O-sulfated disaccharides (13%) as compared with iHS (1%). These values are compatible with those reported for other aHS from endothelial cells (1,47), from fibroblast and epithelial cell lines (8,9), and from Reichert's embryonic membrane (2). In these cells, aHS 3-O- sulfated species range from 5.5% of the 35 S-labeled disaccharides for bovine aortic endothelial cells to 35% 3-O-sulfated tetrasaccharides in Reichert's membrane. Hence, granulosa cell aHS appear to contain substantial amounts of 3-O-sulfated disaccharides.
Complex formation between aHS and AT retards the migration of aHS in coelectrophoresis according to the protease inhibitor concentration in the gel (46,47). Granulosa cell aHS, like endothelial cell aHS, were markedly retarded by the presence of AT in the gel, showing the formation of aHS-AT complexes. Both types of aHS behaved homogeneously, suggesting that all molecules bind to AT with high affinity, in agreement with published data on endothelial cell aHS coelectrophoresis reporting a single class of high affinity binding sites for AT (47). For comparison, in the presence of fibronectin, heparin was found to become broadly smeared, as some heparin molecules were bound weakly and others were bound strongly (57).
Having shown that granulosa cell aHS form complexes with AT similar to those formed by endothelial cell aHS, we next demonstrated the functional anticoagulant activity of granulosa cell aHS by their ability to increase the reactivity of AT toward its target enzyme thrombin. Thrombin inhibition by AT results in a stoichiometric thrombin-AT complex in which both the enzyme and the inhibitor are inactivated. In addition, thrombin sometimes escapes inhibition by cleaving AT at its reactive site without being trapped in a complex, thereby generating modified inactive AT. Thus AT can be either an inhibitor or a substrate of thrombin, and the amount of cleaved inhibitor produced is markedly increased in the presence of heparin (49,58,59). We observed that, like heparin, granulosa cell aHS accelerate the formation of thrombin-AT complexes as well as the cleavage of AT in its modified form. In the presence of granulosa cell aHS, the amounts of AT recovered in thrombin-AT complexes (14%) exceeded that obtained in the presence of heparin (6%). The limited amounts of purified primary granulosa cell aHS available prevented us from performing additional experiments to determine more precisely their anticoagulant activity. However, we carried out a conservative estimate of the minimum anticoagulant activity of granulosa cell aHS using the known values for heparin. The concentration of granulosa cell aHS and heparin used in these experiments was 0.86 and 7.3 nM, respectively. Considering that about one-third of the heparin chains have high affinity for AT, the concentration of high affinity heparin (2.4 nM) was comparable with that of granulosa cell aHS. We therefore conclude that the anticoagulant activity of granulosa cell aHS is comparable with that of high affinity heparin (about 8 USP units/nmol). Similar calculations showed that microvascular endothelial cell aHS have an anticoagulant activity of 1.4 USP units/nmol (47). We have estimated the anticoagulant activity of aHS recovered from granulosa cell cultures to be at least 2.3 ϫ 10 Ϫ3 USP units/10 6 cells (by comparison with heparin), a value certainly underestimated since we did not correct for losses during purification. For comparison, bovine aortic and rat microvascular endothelial cells were found to have about 1 ϫ 10 Ϫ3 and 5 ϫ 10 Ϫ5 USP units/10 6 cells, respectively (1,47). In addition, Andrade-Gordon and collaborators (21) have reported a heparin-like activity of about 1 ϫ 10 Ϫ3 USP units/10 6 cells in rat granulosa cells. These authors used granulosa cell layers as a source of HS chains, while we purified aHS from both the cell layers and the culture medium, which explains our higher estimate of granulosa cell aHS anticoagulant activity. This suggests that granulosa cells might release aHS into their culture media and is in agreement with the observation that porcine follicular fluid contains heparin-like activity (21). Altogether, it appears that ovarian granulosa cells are endowed with considerable anticoagulant potential.
We further examined the effects of stimulation by FSH on the distribution of granulosa cell HSPGs and aHSPGs between cell layer and culture medium. We first analyzed total 35 Slabeled GAGs purified from granulosa cell surface and culture medium and found that the percentage of 35 S-labeled HS as compared with CS released in the culture medium was higher (33%) after FSH stimulation than under basal conditions (22%). These results are in agreement with data from Yanagishita and collaborators (18,33), who reported that granulosa cells release 90% dermatan sulfate proteoglycans and 10% HSPGs, while after stimulation by FSH and insulin-like growth factor 1 soluble HSPGs represent 29% of the soluble proteoglycans. FSH seems, therefore, to increase the proportion of HS in the GAGs accumulating in granulosa cell culture medium. In a second step, we analyzed the effects of FSH stimulation on aHSPGs present on granulosa cell monolayers and in their culture medium by 125 I-AT binding. These experiments demonstrated that FSH induces a decrease in cellbound aHSPGs and a concomitant increase of the aHSPGs present in the culture medium, suggesting that FSH increases the amounts of aHSPGs released from the cell surfaces into the culture medium. This observation in vitro would correspond in vivo to a release of aHSPGs toward the follicular fluid during the FSH-induced maturation of the ovarian follicle.
Levels of aHSPG expression can be modulated in vitro and vary according to the circumstances in vivo. Increased expression of HSPGs by transforming growth factor-␤ in cultured endothelial cells (60), and overexpression of the HSPG ryudocan in transfected fibroblastic cells (40), both result in a decrease in the percentage of aHSPGs. Homocysteine by altering the redox potential seems to specifically alter aHSPGs production by endothelial cells (61). Moreover, the characterization of a cell mutant specifically defective in aHSPGs suggests the presence of a regulatory component responsible for the assembly of the specific AT-binding oligosaccharide sequence in aHSPGs (9,62). Together, these data indicate that the expression of aHSPGs could be specifically regulated, independent of the general synthesis of HSPGs. We demonstrated in a previous study (27) that the distribution of aHSPGs between cell surface and culture medium varies between fibroblastic and endothelial cells, but this is the first report to show the modulation of aHSPGs distribution according to the functional state of primary cells.
Detailed metabolic studies by Yanagishita and Hascall (26,56) demonstrated that granulosa cells recycle their cell surface HSPGs mainly by internalization. Under these conditions GPIanchored HSPGs are exclusively internalized, whereas 70% of the membrane-spanning HSPGs are internalized, and 30% of these latter species are released into the culture medium supposedly by proteolytic cleavage of their core proteins (26). We tested whether or not aHS chains are present on GPI-anchored HSPGs and if the release of aHSPGs in culture medium induced by FSH involves GPI-anchored aHSPGs.
Under basal conditions about 16% of granulosa cell-bound aHSPGs were released by phospholipase C, demonstrating that aHS are attached to GPI-anchored HSPG core proteins in amounts comparable with those reported for the general population of HSPGs (25). The aHSPGs detected in granulosa cell culture media are presumably released from cell-surface aHSPGs. Phospholipase C released a similar proportion of aHSPGs from granulosa cells cultured under basal and FSHstimulated conditions. These results indicate that the 40% decrease in total cell-bound aHSPGs observed in FSH-stimulated granulosa cells does not involve GPI-anchored aHSPGs but is probably due to a decrease in membrane-spanning aHSPGs at the cell surface. Yanagishita and Hascall (56) have shown that the metabolism of granulosa cell HSPGs is complex; it includes a rapid synthesis and exposure of HSPGs at the cell surface within a few minutes, a rapid turnover with a half-life at the cell surface of about 4 h, followed by several distinct intracellular catabolic pathways. Our data were obtained from granulosa cells cultured under steady-state conditions for 48 h. Under such conditions, it is not possible to determine the origin of aHSPGs released by granulosa cells in response to FSH. It is tempting to speculate that the additional soluble aHSPGs present in FSH-stimulated conditions are due to an increased release of membrane-spanning aHSPGs. However, we cannot exclude the possibility that an increased synthesis of GPI-anchored aHSPGs coupled with the increased release of GPI-anchored aHSPGs could be responsible for the accumulation of soluble aHSPGs while keeping the membrane-bound levels of GPIanchored aHSPGs unaltered. Pulse-chase experiments with short 35 S-labeling pulses of GAGs followed by the purification and quantification of 35 S-aHS are required to discriminate among these possibilities. These experiments are impossible to perform at the present time because of limitations in cell number. FSH-sensitive granulosa cell lines such as those developed by Amsterdam and collaborators (63), which are transfected with the FSH receptor, seem promising candidates for such experiments.
Proteolytic enzymes might also be involved in the regulation of the release of aHSPGs from FSH-stimulated granulosa cells. The identity of these proteases is not yet established, and the presence of conserved basic amino acids in the C-terminal portion of the extracellular domains of HSPGs from the syndecan family suggests that tryptic proteases might be involved in this mechanism (64). In addition, plasmin has been shown to release bFGF-HSPG complexes from endothelial cells (65), and leupeptin (an inhibitor of thiol proteases and of some serine proteases including plasmin) was found to decrease the shedding of HSPGs from granulosa cells into the medium (23). We tested whether serine proteases were involved in the release of aHSPGs from granulosa cells under basal or FSH-stimulated conditions. We observed that the tryptic inhibitor aprotinin did not affect the release of aHSPGs in granulosa cell culture medium under basal or FSH-stimulated conditions (data not shown). These observations suggest that other types of proteases might be responsible for the release of aHSPGs. Experiments are currently underway in our laboratory to investigate this matter.
Data available concerning the localization of AT in tissues are limited and contradictory. Plasma AT is synthesized mainly in the liver, but small amounts of AT mRNA are also detected in the kidney and in endothelial cells (66). In the vessel wall, aHSPG-bound AT constitutes a natural anticoagulant mechanism protecting the vasculature against thrombosis (67). Immunocytochemical studies have shown that endogenous AT is detected only in association with vascular endothelial cells and their underlying matrix (6,13,51), thereby following the distribution of aHSPGs on the vascular endothelium of various tissues. On the other hand, kinetic radiotracer studies suggest that AT is distributed in vivo between three physiological pools, plasma, the vascular wall, and a third extravascular compartment (68), and AT has been found in follicular fluid (69,70). As of now the presence of AT mRNA and the localization of AT in the ovary remain to be investigated. Ovulation occurs in parallel with local inflammation, and vascular permeabilization, extravasation of plasma proteins, and fibrin deposition are observed in the outer layers of ovulatory follicles (19,71,72). Extravascular coagulation has been shown to be regulated by vascular permeability (73), and after ovulation a fibrin clot forms in the remnant antral cavity (74). The presence of large quantities of aHSPGs in the inner compartment of preovulatory follicles could serve to localize activated AT to this locale, thereby preventing the formation of a fibrin clot within the antral cavity prior to the expulsion of the oocyte. Therefore, aHSPGs could be critically located in the inner follicle to maintain fluidity in the environment of the oocyte. Alternatively, aHSPGs could serve as cofactors to activate other serine protease inhibitors present in the follicle and thus participate in the control of the proteolytic events leading to the breakdown of the follicular wall at ovulation. Further studies are underway in our laboratory to elucidate the physiological functions of aHSPGs in the extravascular compartment formed by the ovarian follicle.