Sulfate moieties in the subendothelial extracellular matrix are involved in basic fibroblast growth factor sequestration, dimerization, and stimulation of cell proliferation.

The growth promoting activity of the subendothelial extracellular matrix (ECM) is attributed to sequestration of basic fibroblast growth factor (bFGF) by heparan sulfate proteoglycans and its regulated release by heparin-like molecules and heparan sulfate (HS) degrading enzymes. HS is also involved in bFGF receptor binding and activation. The present study focuses on the growth promoting activity and bFGF binding capacity of sulfate-depleted ECM. Corneal endothelial cells (EC) maintained in the presence of chlorate, an inhibitor of phosphoadenosine phosphosulfate synthesis, produced ECM containing 10-15% of the sulfate normally present in ECM. Incorporation of sulfate into HS was reduced by more than 90%. Binding of 125I-bFGF to sulfate-depleted ECM was reduced by 50-60% and only about 10% of the ECM-bound bFGF was accessible to release by heparin. Incubation of 125I-bFGF on top of native ECM resulted in dimerization of the ECM-bound bFGF, but there was a markedly reduced binding and dimerization of bFGF on sulfate-depleted ECM. ECM produced in the presence of chlorate contained a nearly 10-fold less endogenous bFGF as compared to native ECM and exerted little or no mitogenic activity toward vascular EC and 3T3 fibroblasts. In other studies, we investigated the interaction between chlorate-treated vascular EC and either native or sulfate-depleted ECM. Exogenous heparin stimulated the proliferation of chlorate-treated EC seeded on native ECM, suggesting its interaction with ECM-bound bFGF and subsequent presentation to high affinity cell surface receptors. On the other hand, heparin had no effect on chlorate-treated cells seeded in contact with sulfate-depleted ECM or regular tissue culture plastic. Altogether, the present experiments indicate that heparan sulfate proteoglycans associated with the cell surface and ECM act in concert to regulate the bioavailability and growth promoting activity of bFGF. While HS in the subendothelial ECM functions primarily in sequestration of bFGF in the vicinity of responsive cells, HS on cell surfaces is playing a more active role in displacing the ECM-bound bFGF and its subsequent presentation to high affinity signal transducing receptors.

Heparan sulfate (HS) 1 is a most ubiquitous glycosaminogly-can present on cell surfaces, in basement membranes and extracellular matrices (Gallagher et al., 1986;Jackson et al., 1991;Kjellen and Lindahl, 1991). Recent interest in heparan sulfate proteoglycans (HSPG) stems from increasing awareness of the functional implications of their interactions with growth factors, matrix molecules, and cytoskeletal elements (Gitay-Goren et al., 1992;Jackson et al., 1991;Ruoslahti and Yamaguchi, 1991;Vlodavsky et al., 1993;Yayon et al., 1991). The HS chains have been implicated in a variety of physiological processes including the regulation of glomerular basement membrane permeability to proteins, assembly of basement membranes, regulation of nuclear metabolism, cell attachment and spreading, recruitment of inflammatory cells (chemokines), and the regulation of mammalian cell proliferation and differentiation (Gallagher et al., 1986;Jackson et al., 1991;Ruoslahti and Yamaguchi, 1991;Tanaka et al., 1993). The sulfate residues, which may be present on four different positions of the polysaccharide backbone, are of high interest, since they have been shown to be major factors in the determination of specificity in protein-polysaccharide interactions (Lindahl, 1989). Of particular significance is the interaction between HS and basic fibroblast growth factor (bFGF), involved in bFGF receptor binding and signal transduction (Ornitz et al., 1992;Rapraeger et al., 1991;Yayon et al., 1991). A unique, highly sulfated bFGF-binding fragment of HS was isolated from cell surface HSPG of fibroblasts (Turnbull et al., 1992). Sulfation in critical positions along the polysaccharide chain, particularly 2-O-sulfation, seems necessary to generate a specific bFGF binding motif that can support high affinity bFGF-receptor binding and activation (Aviezer et al., 1994b;Habuchi et al., 1992;Ishihara et al., 1993;Turnbull et al., 1992).
Chlorate, an inhibitor of ATP sulfurylase and hence of the production of phosphoadenosine phosphosulfate, the active sulfate donor for sulfotransferases (Baeuerle and Huttner, 1986), has been shown to abolish sulfation on proteins and carbohydrate residues in intact cells without inhibiting cell growth or protein synthesis (Baeuerle and Huttner, 1986;Keller et al., 1989). Exposure to chlorate markedly reduced binding of bFGF to high affinity cell surface receptors and the ability of 3T3 fibroblasts to proliferate in response to bFGF (Guimond et al., 1993;Rapraeger et al., 1991).
Our studies on the control of cell proliferation by its local environment focus on the interaction of cells with the extracellular matrix (ECM) produced by cultured corneal endothelial cells (EC) Vlodavsky et al., 1980Vlodavsky et al., , 1993. This ECM closely resembles the subendothelium in vivo in its morphological appearance and molecular organization. It contains collagens (mostly types III and IV, with smaller amounts of types I and V), proteoglycans (mostly HS-and dermatan sulfate-proteoglycans, with smaller amounts of chondroitin sulfate proteoglycans), laminin, fibronectin, entactin, and elastin. EC and other cell types plated in contact with this ECM no longer require the addition of soluble bFGF in order to proliferate and express their differentiated functions Vlodavsky et al., 1980). In subsequent studies bFGF was identified as a complex with HSPG in the subendothelial ECM produced in vitro (Bashkin et al., 1989;Vlodavsky et al., 1987) and on cell surfaces and basement membranes of diverse tissues and blood vessels (Cardon-Cardo et al., 1990;Gonzalez et al., 1990). HS-bound bFGF is protected against heat inactivation and proteolytic degradation (Saksela et al., 1988) and can be released in an active form by heparinlike molecules and HS degrading enzymes (Bashkin et al., 1989;Ishai-Michaeli et al., 1990Vlodavsky et al., 1991), or by proteases (Benezra et al., 1993;Saksela et al., 1988). On the basis of these results, the ECM is regarded as a storage depot for bFGF and possibly other heparin-binding growth factors and cytokines. These immobilized growth factors are held responsible for the growth promoting activity of the ECM. In the present study, corneal EC were cultured in the presence of chlorate to produce sulfate-depleted ECM. This ECM was analyzed for its ability to sequester and dimerize bFGF and its growth promoting activity toward vascular EC and 3T3 fibroblasts, in the absence and presence of exogenously added heparin. We have also analyzed the ability of chlorate-treated EC to respond to native and sulfate-depleted ECM.
Preparation of Dishes Coated with ECM-Bovine corneal EC were dissociated from stock cultures (second to fifth passage) with STV and plated into four-well plates at an initial density of 2 ϫ 10 5 cells/ml/well. Cells were maintained as described above, except that 5% dextran T-40 was included in the growth medium Vlodavsky et al., 1980). The cells were cultured in the absence and presence of sodium chlorate (30 mM, except when stated otherwise) without replacing the medium and without addition of bFGF. Ten to 12 days after seeding, the subendothelial ECM was exposed by dissolving (5 min, room temperature) the cell layer with PBS containing 0.5% Triton X-100 and 20 mM NH 4 OH, followed by four washes in PBS. The ECM remained intact, free of cellular debris, and firmly attached to the entire area of the tissue culture dish (Gospodarowicz et al., , 1983Vlo-davsky et al., 1980Vlo-davsky et al., , 1987. For preparation of sulfate-labeled ECM, corneal EC were plated into four-well plates and cultured as described above. Na 2 35 SO 4 (540 -590 mCi/mmol) was added (20 Ci/ml) 1 and 5 days after seeding, and the cultures were incubated with the label without medium change. Ten to 12 days after seeding, the cell monolayer was dissolved and the ECM exposed, as described above. To determine the total amount of sulfate labeled material, the ECM was digested with trypsin (25 g/ml, 24 h, 37°C) and the solubilized material counted in a ␤-counter. Protein was determined in aliquots of the trypsinized material using the Coomassie protein assay reagent (Pierce) according to the manufacturer's instructions. To determine the amount of sulfate labeled HS, the ECM was digested (48 h, 37°C, pH 6.2) with a human placental heparanase (endo-␤-D-glucuronidase) purified and characterized as described (Gilat et al., 1995). The estimated M r of the HS fragments was 5,000 -7,000 as compared to a M r of about 30,000 ascribed to intact HS side chains released from ECM by treatment with alkaline borohydride or papain (Vlodavsky et al., 1983). Sulfate-labeled low M r degradation products released into the incubation medium were analyzed by gel filtration on Sepharose 6B, as described (Ishai-Michaeli et al., 1990).
Iodination and Binding of bFGF-Recombinant bFGF was iodinated using chloramine T, as described (Benezra et al., 1993). The specific activity was 1.2-1.7 ϫ 10 5 cpm/ng bFGF, and the labeled preparation was kept for up to 6 weeks at Ϫ70°C. 125 I-bFGF binding to ECM was performed as described (Ishai-Michaeli et al., 1992). Unbound bFGF was removed and the remaining ECM was solubilized with 1 N NaOH and counted in a ␥-counter. Displacement of ECM-bound 125 I-bFGF by heparin was performed as described (Ishai-Michaeli et al., 1992).
Cross-linking-bFGF cross-linking experiments were carried out as described (Ornitz et al., 1992;Spivak-Kroizman et al., 1994). Briefly, 125 I-bFGF (25 ng/0.25 ml/well) was incubated (1 h, 24°C) with ECM (four-well plates) in a buffer containing 150 mM NaCl and 25 mM HEPES (pH 7.5). The incubation medium was removed, the dishes were washed three times with PBS followed by incubation (30 min, 24°C) with 0.15 mM DSS in PBS. The cross-linking reaction was quenched with 10 mM ethanolamine-HCl (pH 8.0) for 30 min, the incubation medium removed, and the remaining ECM scraped and dissolved in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer (150 l/well, 1 h, 37°C). The soluble material was analyzed by 15% SDS-PAGE. The cross-linked bFGF was visualized by autoradiography using Kodak XAR film.
Iodination of Heparin-Heparin was reacted with Bolton-Hunter reagent and iodinated as described (Dawes and Pepper, 1979). The specific activity was 6,000 cpm/ng. Binding of 125 I-heparin to ECM was performed in the absence and presence of 20 g/ml unlabeled heparin.
Growth Factor Activity-Assay for DNA synthesis in growth-arrested Balb/c 3T3 cells was performed as described (Vlodavsky et al., 1987). For measurements of EC proliferation, cells were seeded on ECM or regular tissue culture plastic at a low density (1 ϫ 10 3 /16-mm well) in 0.5 ml DMEM containing 10% heat-inactivated calf serum. Five to 6 days after seeding, the cells were dissociated with STV and counted in a Coulter counter (Coulter Electronics). Alternatively, the cells were exposed (3-4 h, 37°C) to [ 3 H]thymidine (5 Ci/well) and DNA synthesis was determined by measuring the radioactivity incorporated into trichloroacetic acid-insoluble material (Ishai-Michaeli et al., 1990). Thymidine incorporation was linearly correlated to the number of cells per well. In other experiments, the EC were seeded at a clonal cell density (300 cells/35-mm dish) and cell colonies were fixed and stained with 0.1% crystal violet, 10 days after seeding (Gospodarowicz et al., 1976;Vlodavsky et al., 1987).

Effect of Chlorate on the ECM Content of Sulfate and
Heparan Sulfate-Corneal EC were seeded at a confluent cell density into four-well plates, in the absence and presence of increasing concentrations of chlorate (0.1-60 mM). Under these conditions, a contact inhibited cell monolayer was formed within 6 h and there was no effect to chlorate on the cell number and morphological appearance, thereafter. The cells were maintained in sulfate-free Fisher medium containing 20 Ci/ml Na 2 35 SO 4 as described under "Experimental Procedures." Ten days after seeding, the cell layer was dissolved to expose the subendothelial ECM. The ECM was then subjected to complete digestion with trypsin (25 g/ml, 24 h, 37°C) and the digest counted in a ␤-counter. Eighty-90% decrease in sul-fate content was obtained in ECM produced in the presence of 30 mM chlorate ( Fig. 1, inset). Similar results were obtained when the stock of corneal EC was treated (48 h) with 30 mM chlorate before seeding into the four-well plates and thereafter. At this concentration, chlorate had little or no effect on the total amount of ECM protein deposited by the cells, nor on the morphology and organization of the ECM as revealed by phase and scanning electron microscopy. Exposure to 60 mM chlorate resulted in an almost complete inhibition of sulfate incorporation, but this was associated with a slight decrease (Ͻ20%) in ECM deposition. In subsequent experiments, sulfate-depleted ECM was produced by cells maintained in the presence of 30 mM chlorate, as described above. Specific incorporation of sulfate into HS was analyzed by measurements of sulfate-labeled material released from ECM during incubation (24 h, 37°C) with a highly purified preparation of human placental heparanase. Sulfate-labeled HS degradation products released into the incubation medium were analyzed by gel filtration on Sepharose 6B, as described (Ishai-Michaeli et al., 1990). As demonstrated in Fig. 1, there was little (Ͻ10%) release of sulfate-labeled HS degradation products (fractions 15-35, 0.5 Ͻ K av Ͻ 0.8) from ECM produced in the presence of 30 mM chlorate.
Effect of Chlorate on bFGF Binding Capacity and Growth Promoting Activity of ECM-Sulfate groups are involved in bFGF binding to isolated heparin and HS (Aviezer et al., 1994b;Habuchi et al., 1992;Ishai-Michaeli et al., 1992Turnbull et al., 1992). The availability of sulfate-deficient ECM provided an appropriate means to investigate to what extent sulfate groups are involved in bFGF binding to a multimolecular structure such as intact subendothelial ECM. Binding of 125 I-bFGF to ECM produced in the presence of 30 mM chlorate was inhibited by 50 -60%. Sulfate depletion had a more pronounced effect on bFGF binding to HS in the ECM, as revealed by 70 -80% reduction in the amount of ECM-bound bFGF displaced by heparin (Fig. 2). A similar value was obtained when the ECM-HS was degraded by human placental heparanase (data not shown). While heparin or heparanase treatment released 50 -60% of the 125 I-bFGF bound to native ECM, only 10 -15% of the bFGF bound to sulfate-depleted ECM was released under the same conditions. It is therefore conceivable that binding of bFGF to sulfatedepleted ECM produced in the presence of 30 mM chlorate was due primarily to binding to other components of the ECM (i.e. fibronectin) or to glycosaminoglycan side chains deprived of their sulfate moieties (Ornitz et al., 1995).
We have previously demonstrated that EC and other cell types plated in contact with the subendothelial ECM no longer require the addition of soluble bFGF in order to proliferate Vlodavsky et al., 1987). This mitogenic effect was attributed primarily to the presence of bFGF in ECM, although the mode of bFGF deposition was not elucidated (Vlodavsky et al., 1987). ECM produced in the absence and presence of chlorate was tested for mitogenic activity toward vascular EC and 3T3 fibroblasts. For this purpose, vascular EC were seeded at a low cell density (1,000 cells/16-mm well) on top of ECM produced in the absence and presence of 30 mM chlorate. Six days after seeding, the cultures were exposed to [ 3 H]thymidine and the amount of trichloroacetic acid-precipitable radioactivity was determined 3 h afterwards (Fig. 3A).

FIG. 1. Effect of chlorate on sulfate incorporation into HS in ECM.
Corneal EC were seeded at a confluent density (2 ϫ 10 5 cells/16-mm well) in the absence (E ) or presence of increasing concentrations (Ⅺ, 1 mM; Ç, 10 mM; f, 20 mM; q, 30 mM; å, 60 mM) of chlorate. The cells were maintained in Fisher's medium in the presence of Na 2 35 SO 4 (20 Ci/well) added on day 1 and 5. ECM was prepared on day 10, followed by incubation (24 h, 37°C, pH 6.2) with 2 g/ml of a purified preparation of human placental heparanase. Sulfate-labeled HS degradation products released into the incubation medium were analyzed by gel filtration on Sepharose 6B. Inset, total amount of labeled sulfate determined following trypsin digestion of the ECM. Aliquots of the trypsinized material were counted in a ␤-counter. Each data point (cpm/well) is the mean Ϯ S.D. of four wells. EC were also seeded at a clonal cell density (300 cells/35-mm dish) on top of native and sulfate-depleted ECM and cell colonies were stained 10 days after seeding (Fig. 3, inset). As demonstrated in Fig. 3A, ECM produced in the presence of chlorate exerted a greatly reduced mitogenic activity toward vascular EC seeded at a low or clonal cell density. In other experiments (Fig. 3B), ECM produced in the presence of increasing concentrations of chlorate was subjected to trypsin digestion and aliquots of the solubilized material were added to confluent, growth-arrested 3T3 fibroblasts (Fig. 3B) or to sparsely seeded EC (1,000 cells/16-mm well) maintained in the presence of 10% heat-inactivated calf serum (data not shown). A trypsin digest of native ECM was highly mitogenic to growth arrested 3T3 fibroblasts (Fig. 3B), and this activity was inhibited by neutralizing anti-bFGF antibodies (data not shown). In contrast, ECM produced in the presence of 30 mM chlorate was devoid of mitogenic activity toward 3T3 fibroblasts (Fig. 3B). Likewise, chlorate markedly reduced (ϳ60%) the growth promoting activity of ECM extracts toward sparsely seeded EC (data not shown). As demonstrated in Fig. 3A, EC plated on chlorate-treated ECM responded to exogenously added bFGF in a manner similar to cells plated on regular tissue culture plastic. These results indicate that sulfation is critical for the growth promoting activity of the ECM.
In other experiments, ECM produced in the absence and presence of 30 mM chlorate was digested (3 h, 37°C) with 0.1 g/ml trypsin and the amount of bFGF in the solubilized material was determined by an immunoassay (Quantikine human bFGF, R&D Systems, Minneapolis, MN). The amount of bFGF in sulfate-depleted ECM was about 10-fold lower than that determined in native ECM (i.e. 11 and 121 pg of bFGF/ECM-coated 16-mm culture well, respectively). Similar results were obtained when the ECM was digested with bacterial (Flavobacterium heparinum) heparinase I (IBEX Technologies, Montreal, Canada) rather than trypsin. The heparinase-treated ECM exerted little or no mitogenic activity on vascular EC.
Effect of Heparin on bFGF Sequestration and Growth Promoting Activity of Sulfate-depleted ECM-We investigated the effect of heparin on bFGF sequestration and the growth promoting activity of ECM produced by corneal EC maintained in the absence and presence of chlorate. Heparin was included in the cell lysis solution to prevent binding of intracellular bFGF to ECM-HS when the ECM-producing cells are lysed. When 125 I-bFGF was added to the lysis solution, it was found that only about 2 and 0.2% of the added bFGF was sequestered by  Fig. 1) was incubated (3 h, 24°C) with 125 I-bFGF (1.5 ϫ 10 4 cpm/0.25 ml/well), washed free of unbound bFGF and incubated (1 h, 24°C) with 10 g/ml heparin to displace the HS-bound bFGF. Radioactive material released into the incubation medium was counted in a ␥-counter. Released radioactivity is expressed as percent of 125 I-bound to native ECM (6.5 ϫ 10 3 cpm/well) and to ECM produced in the presence of each concentration of chlorate . Each data point is the mean of four wells. "Spontaneous" release of 125 I-bFGF in the presence of incubation medium alone was subtracted from the experimental values.

FIG. 3. Effect of chlorate on the mitogenic activity of ECM toward vascular endothelial cells and 3T3 fibroblasts.
A, vascular EC. Vascular EC were seeded at a low (1,000 cells/well) or clonal (300 cells/dish, inset) cell density on tissue culture plastic, or ECM produced in the absence (ECM) or presence (ECM ϩ chlorate) of 30 mM chlorate. The cells were maintained in DMEM containing 10% heat-inactivated calf serum and tested for (i) thymidine incorporation, in the absence (striped box) and presence (shaded box) of 5 ng/ml bFGF; and (ii) colony formation, in the absence of exogenously added bFGF (inset, 1, plastic; 2, native ECM; 3, sulfate-depleted ECM), as described under "Experimental Procedures." B, 3T3 fibroblasts. ECM produced in the absence and presence of increasing concentrations of chlorate was digested with trypsin (0.1 g/ml, 3 h, 37°C) and aliquots (25 l) of the solubilized material were tested for induction of [ 3 H]thymidine incorporation in growth arrested 3T3 fibroblasts. The basal incorporation of [ 3 H]thymidine into resting 3T3 fibroblasts was Ͻ1,000 cpm as compared to 82,000 cpm in the presence of 0.5 ng/ml bFGF. Each data point (cpm/well) is the mean Ϯ S.D. of three wells. native and sulfate-depleted ECM, respectively, during the 5-min cell lysis period (Fig. 4). Heparin (10 -20 g/ml) inhibited by about 80% the deposition of bFGF on top of ECM produced in the absence of chlorate, and there was little or no effect to heparin on the residual binding of 125 I-bFGF to ECM produced in presence of chlorate (Fig. 4).
We next analyzed the effect of heparin on the growth promoting activity of ECM produced in the absence and presence of chlorate. Heparin, present during the 5-min cell lysis period, had little or no effect on the growth promoting activity of native ECM toward vascular EC seeded on ECM at a low (Fig. 5) or clonal (data not shown) cell densities. Surprisingly, the mitogenic activity toward EC of ECM produced in the presence of chlorate was stimulated (1.5-4-fold, in different experiments) when heparin was included in the cell lysis solution. This stimulation was observed both when the endothelial cells were seeded directly on the ECM (Fig. 5) and when the ECM was first digested with trypsin and aliquots of the solubilized material were tested for mitogenic activity on vascular EC (data not shown). Measurements of 125 I-heparin binding revealed that under the experimental conditions applied in Fig. 5, Ͻ0.5% of the heparin was bound to the ECM and there was no difference in heparin binding to ECM produced in the absence or presence of chlorate (data not shown).
Mitogenic Response of Chlorate-treated Endothelial Cells Plated on ECM-It has been previously demonstrated that heparin restores the ability of chlorate-treated 3T3 fibroblasts to proliferate in response to bFGF (Rapraeger et al., 1991). We investigated whether proliferation of chlorate-treated EC can be similarly restored by native ECM, in the absence and presence of added heparin. For this purpose, vascular EC were pretreated (24 h) with 30 mM chlorate and seeded in the presence of 30 mM chlorate and absence of exogenously added bFGF, on (i) tissue culture plastic, (ii) sulfate-depleted ECM, and (iii) native ECM. Heparin (1 g/ml) was added to some of the cultures and the cells counted on day 5 after seeding. Fig. 6 demonstrates that chlorate-treated EC exhibited little or no proliferative response to heparin when seeded on regular tissue culture plastic or sulfate-depleted ECM. A significant mito-genic response was obtained when the chlorate-treated EC were seeded in contact with native ECM, and best results were obtained when heparin (1 g/ml) was added to the culture medium (Fig. 6). Under these conditions (chlorate-treated EC maintained on native ECM in the presence of heparin), there was little or no further stimulation of cell proliferation in response to exogenously added bFGF (data not shown). These results suggest that sulfate moieties on cell surfaces play an active role in the presentation of ECM-bound bFGF to its high affinity cell surface receptors and that soluble heparin may exert a similar effect in chlorate-treated, undersulfated endothelial cells.
Dimerization of bFGF on Native and Sulfate-depleted ECM- FIG. 4. Effect of heparin on bFGF sequestration occurring when the EC layer is solubilized and the ECM exposed. Confluent cultures (10 days after seeding) of corneal EC maintained in the absence (E) and presence (q) of 30 mM chlorate were exposed (5 min, 24°C) to PBS containing 0.5% Triton X-100, 20 mM NH 4 OH, 125 I-bFGF (2 ϫ 10 4 cpm/0.25 ml/well) and increasing concentrations of heparin. The newly denuded ECM was washed three times, and the ECM-bound radioactivity solubilized (1 N NaOH, 3 h, 37°C) and counted in a ␥-counter. Each data point is the mean Ϯ S.D. of four wells. Heparin-mediated oligomerization of bFGF (Ornitz et al., 1992) and acidic FGF (Spivak-Kroizman et al., 1994;Mascarelli et al., 1993) is thought to be responsible for receptor activation and induction of cell proliferation by these mitogens. We investigated whether dimerization of bFGF can be induced by HS in native ECM as compared to sulfate-depleted ECM. For this purpose, 125 I-bFGF was incubated with ECM produced in the absence or presence of chlorate and the formation of 125 I-bFGF dimers analyzed by DSS cross-linking and SDS-PAGE. When similar amounts of solubilized native and sulfate-depleted-ECM were applied on the gel, dimerization of 125 I-bFGF was primarily detected on native ECM (Fig. 7A, lane 3), as compared to a markedly reduced dimer formation on sulfatedepleted ECM (lane 5). Likewise, dimerization of bFGF was not detected on ECM that was pretreated (1 h, 37°C) with bacterial heparinase I (0.1 IU/ml) so as to degrade Ͼ90% of its HS content (data not shown). There was no dimerization of 125 I-bFGF on regular tissue culture plastic (Fig. 7A, lane 1). When heparin (5 g/ml) was included in the incubation medium, 125 I-bFGF binding to plastic or native ECM was markedly reduced and hence no ECM-bound 125 I-bFGF dimers could be detected (Fig. 7A, lanes 2 and 4). Binding of 125 I-bFGF to sulfate-depleted ECM was slightly increased in the presence of heparin, but the ECM-bound bFGF was almost entirely in the form of monomers (lane 6). In a similar experiment, native and sulfate-depleted ECMs were incubated with 125 I-bFGF, the ECM-bound bFGF was subjected to cross-linking and solubilization as described above, except that similar amounts of radioactivity, rather than solubilized ECM extracts, were applied on the gel. Under these conditions (Fig. 7B), bFGF dimer formation was detected on both native (lane 2) and sulfatedepleted (lane 3) ECM, but not on regular plastic dishes (lane 1). In fact, densitometric analysis revealed a similar ratio of dimeric to monomeric 125 I-bFGF on native and sulfate-depleted ECM, indicating that the markedly reduced dimerization seen on sulfate-depleted ECM is due primarily to the decrease in bFGF sequestration by HS in this ECM.

DISCUSSION
Heparan sulfates are heterogeneous molecules that vary both in their basic disaccharide subunits and in their degree and position of sulfation (Gallagher et al., 1986;Jackson et al., 1991;Kjellen and Lindahl, 1991). Both the level of sulfation and position of sulfate groups are major determinants in the interaction between bFGF and HS and the ability of heparin and HS to promote bFGF receptor binding and mitogenic activity (Aviezer et al., 1994b;Habuchi et al., 1992;Ishihara et al., 1993;Ornitz et al., 1992;Turnbull et al., 1992). Using chlorate, an inhibitor of phosphoadenosine phosphosulfate synthesis, we investigated the involvement of sulfate groups in the growth promoting activity of the subendothelial ECM. Sulfate-depleted ECM exhibited a greatly reduced mitogenic activity toward vascular EC and 3T3 fibroblasts, as compared to native ECM. Similar results were obtained, regardless of whether the vascular EC were seeded on top of chlorate-treated ECM or whether the sulfate-depleted ECM was first digested with trypsin and aliquots of the solubilized ECM added to EC seeded on regular tissue culture plastic. The lack or low mitogenic activity may be due to (i) reduced amounts of bFGF in ECM produced in the presence of chlorate and (ii) inability of this ECM to present the ECMbound bFGF to its high affinity cell surface receptors. Measurements of bFGF binding revealed a 50 -60% reduction in bFGF binding to ECM produced by chlorate-treated corneal EC, as compared to untreated cells. The amount of HS-bound, heparin/heparinase releasable 125 I-bFGF was reduced by 70 -80% in chlorate-treated ECM, suggesting that 125 I-bFGF may bind also to sulfate-depleted glycosaminoglycan side chains (Ornitz et al., 1995), ECM components other than HS (i.e. fibronectin), and possibly ECM-bound bFGF receptors (Hanneken et al., 1995).
Direct immunoquantitation of bFGF in solubilized ECM revealed about a 10-fold reduction in the amount of endogenous bFGF in ECM produced in the presence of chlorate as compared to native ECM. In this assay the ECM was first digested with trypsin to solubilize the matrix and hence the reduced amounts of bFGF determined in sulfate-depleted ECM may be attributed, in part, to tryptic degradation of bFGF that is no longer protected by properly sulfated HS (Saksela et al., 1988). It should be noted, however, that a similar decrease in bFGF content was obtained when the ECM was digested with bacterial heparinase, rather than with trypsin, resulting in solubilization of Ͼ90% of the ECM-resident bFGF. These results, together with the lack of or greatly reduced mitogenic activity exerted by intact, undegraded sulfate-depleted ECM, suggest that this ECM exhibit little or no growth promoting activity simply because its HS chains fail to sequester bFGF and hence can not function as a secured depot of this growth factor in the vicinity of cells. Measurements of the cellular content of bFGF revealed no difference between chlorate-treated and untreated EC, suggesting that chlorate did not affect the synthesis of bFGF. An inhibitory effect on bFGF deposition, possibly as a complex with cell-associated HS, cannot be excluded.
A major concern in the study of the growth promoting activity of the subendothelial ECM is whether the ECM-bound bFGF is deposited into the ECM by intact EC, prior to denudation of the ECM, or sequestered by HS and other components of the ECM when the bFGF containing EC are lysed and the ECM exposed. In the present study, heparin was included in the cell lysis solution in order to eliminate the latter possibility. Measurements of 125 I-bFGF binding revealed that heparin (10 g/ml), present in the cell lysis solution, inhibited by about 90% the binding of 125 I-bFGF to the newly exposed native ECM. However, there was no effect to this heparin on the growth promoting activity of the ECM, indicating that the mitogenic activity of native ECM is not due to sequestration of bFGF occurring when the ECM-producing cells are lysed. An unexpected result was obtained when the effect of heparin on the mitogenic activity of sulfate-depleted ECM, was investigated. Unlike the results with native ECM, heparin, present during the 5-min cell lysis period, stimulated the growth promoting activity of sulfate-depleted ECM. A possible explanation for this stimulation is the ability of heparin to bind to bFGF in the ECM and function in the displacement and presentation of ECM-bound bFGF to high affinity receptor sites on the cell surface. Alternatively, heparin may bind to intracellular bFGF and then to the ECM, increasing the concentration of bFGF in the undersulfated matrix. ECM binding of 125 I-heparin was low and there was no difference between native and sulfate-depleted ECM, but this may result from the iodination procedure, which could significantly alter the ability of heparin to bind to heparin-binding proteins (i.e. fibronectin, vitronectin) in the ECM. Heparin was previously shown to restore the mitogenic response of chlorate-treated 3T3 fibroblasts and endothelial cells to bFGF (Guimond et al., 1993;Rapraeger et al., 1991).
Our experiments with chlorate-treated vascular EC plated on intact native ECM, demonstrated that heparin can restore the ability of the cells to proliferate in response to bFGF residing in the ECM. Previous studies revealed that the K d value for interaction of bFGF with the cell surface HS (2 ϫ 10 Ϫ9 M) is lower than for interaction with HS in the ECM (ϳ1 ϫ 10 Ϫ7 M) (Bashkin et al., 1989;Moscatelli, 1987;Roghani et al., 1994), suggesting that ECM-bound bFGF interacts first with HS on the cell surface and is then presented to high affinity cell surface receptors. Because the cell surface HSPG, unlike that of the ECM, is mobile in the plane of the membrane and can turn over more rapidly by shedding and internalization, it may readily replenish its bFGF from the ECM reservoir, which serves more as an efficient large capacity bFGF storage depot in the vicinity of cells (Bernfield and Hooper, 1991). Both functions (i.e. sequestration and presentation of bFGF to high affinity receptor sites) may not be fulfilled by non-sulfated HS side chains present in the ECM and surface of chlorate-treated EC. A difference between cell surface-and ECM-derived species of HS in their ability to promote bFGF mitogenicity was also demonstrated in our recent studies on the growth promoting activity of HS degradation fragments released by bacterial heparinase III from ECM and cell surfaces. Using HS-deficient lymphoid cells, we have demonstrated a stimulated cell proliferation induced by bFGF in the presence of HS degradation fragments released from cell surfaces, but not from ECM. 2 Altogether, it appears that HS in ECM, unlike on cell surfaces, may not function efficiently as an accessory low affinity receptor capable of directly accelerating the arrival of bFGF at its high affinity signaling receptor.
The essential involvement of sulfate groups in bFGF receptor binding and activation was previously demonstrated by applying undersulfated and oversulfated species of heparin. These studies utilized chlorate-treated cells and HS-deficient cell mutants (Guimond et al., 1993;Rapraeger et al., 1991;Yayon et al., 1991). Best results were achieved in the presence of oversulfated heparin fragments, regardless of whether the N-position was sulfated or acetylated (Aviezer et al., 1994b). In a recent study (Ornitz et al., 1995), a stimulatory effect was also induced by synthetic, nonsulfated heparan-derived di-and trisaccharides. Our studies with native and sulfate-depleted endothelial cells and ECM demonstrate that properly sulfated HSPG associated with the cell surface and ECM act in concert to regulate the bioavailability of active bFGF and possibly other effector molecules to their signal transducing receptors.
Perlecan, the large basement membrane proteoglycan, was recently identified as a major candidate for a bFGF low affinity accessory receptor and an angiogenic modulator. Other HSPG (e.g. syndecan, fibroglycan, and glypican) exhibited only a small activity (Aviezer et al., 1994a). Undersulfated perlecan synthesized in the presence of chlorate is likely to exhibit a much lower capacity to sequester bFGF and subsequently activate the high affinity bFGF cell surface receptor site. This may result in a marked inhibition of the ECM-induced EC proliferation and indirect involvement in neovascularization. It was also demonstrated that acidic FGF binding to its low affinity accessory receptor caused oligomerization of the FGF molecules, thereby indirectly cross-linking and activating the high affinity receptors, resulting in transmembrane signaling and cell proliferation (Spivak-Kroizman et al., 1994). A similar ligand oligomerization was observed during incubation of acidic FGF with bovine lens epithelial cells (Mascarelli et al., 1993). In the present study, intact ECM was found to induce dimerization of 125 I-bFGF to a much higher extent as compared to sulfate-depleted ECM. The markedly reduced dimerization was attributed primarily to the decrease in bFGF binding and sequestration by HSPG in the sulfate-deficient ECM. Dimerization of bFGF observed on sulfate-depleted ECM may be mediated by both sulfated and nonsulfated HS derived saccharides remaining in this ECM. The latter possibility was recently reported (Ornitz et al., 1995). The highly reduced ability of sulfate-depleted ECM to sequester and dimerize bFGF is in all likelihood responsible for the impaired mitogenic activity of this ECM. Oligomerization of ECM-bound bFGF and possibly other heparin-binding growth factors may contribute to the potent growth-and differentiation-promoting activities of the ECM. This oligomerization is induced by properly sulfated HSPG found in native, but not sulfate-depleted ECM. Specific alterations in the level and pattern of sulfation along the HS side chains may thus provide a means to modulate the involvement of HS and ECM in the control of cell proliferation and differentiation and in processes such as neovascularization and tissue remodeling.