Basic fibroblast growth factor binds its receptors, is internalized, and stimulates DNA synthesis in Balb/c3T3 cells in the absence of heparan sulfate.

We have investigated the interaction of basic fibroblast growth factor (bFGF) with its receptors and heparan sulfate proteoglycans (HSPG). It has been suggested that in the absence of HSPG, cells are not able to bind bFGF or respond to treatment with bFGF. In our studies, Balb/c3T3 fibroblasts were treated with 50 mM sodium chlorate to completely inhibit (99%) sulfation of proteoglycans. We found that bFGF was able to bind, be internalized, and stimulate DNA synthesis in the absence of HSPG in a dose-dependent manner. bFGF bound to its receptors on chlorate-treated cells with a lower apparent affinity and no change in receptor number. To determine if this decreased affinity bFGF-receptor interaction is functional, we quantitatively analyzed bFGF internalization and stimulation of DNA synthesis in control and chlorate-treated cells. Endocytotic rate constants (ke) for chlorate-treated and control cells were ke = 0.078 ± 0.022 min−1 and ke = 0.043 ± 0.012 min−1, respectively, suggesting that the process of bFGF internalization is not dramatically altered by HSPG. bFGF stimulated DNA synthesis to the same maximal level under both conditions, but chlorate-treated cells were significantly less responsive at low bFGF doses (∼10-fold increase in ED50). The differences observed for control and chlorate-treated cells in the dose-response curves for stimulation of DNA synthesis and receptor binding correlated directly, suggesting that receptors are equally capable of eliciting a mitogenic signal under both conditions. It is unlikely that these results are due to residual HSPG since heparinase (I and III) digestion of chlorate-treated cells had little effect. Although the presence of HSPG on the cell surface increases the affinity of bFGF for its receptors, our observations suggest that HSPG are not “absolutely” required for binding, internalization, or stimulation of mitogenic activity.

Basic fibroblast growth factor (bFGF) 1 is one of the first isolated and best studied members of the large family of heparin-binding growth factors (1)(2)(3). The biological activity of bFGF is mediated by interaction with high affinity cell surface receptors (4 -6). In addition to binding to receptors, bFGF binds to heparan sulfate proteoglycans (HSPG) on the cell surface (7)(8)(9)(10). Studies have indicated that binding to HSPG facilitates bFGF receptor binding and activation. bFGF receptor binding to cells that do not express HSPG is significantly reduced when compared with cells that express HSPG (11)(12)(13)(14)(15)(16)(17)(18). Our previous studies demonstrate that HSPG act by decreasing the rate at which bFGF is released from its receptors without altering the rate of bFGF association (13). These results suggest that HSPG stabilize bFGF-receptor complexes. The mechanism for how the interaction of bFGF with both HSPG and receptors impacts subsequent steps in the pathway of bFGF cell stimulation remains poorly understood. Heparin, bFGF, and FGF receptors have been shown to form a trimolecular complex in cell-free systems (19,20). There may also be direct interactions between heparin/heparan sulfate and FGF receptors that facilitate bFGF binding and receptor dimerization (20,21). In addition, heparin/heparan sulfate might play a role by facilitating bFGF dimerization (19,22,23).
While the role of HSPG in controlling bFGF has been an area of intense research, there remains considerable question as to the function of the bFGF-receptor complexes that form in the absence of HSPG. Recent studies have reported heparin/heparan sulfate-independent binding of bFGF to its receptors (13,17,20). In these studies it has been demonstrated that bFGF binds its receptors with decreased affinity (2-20-fold) in the absence of heparin/HSPG. In one report it was shown that this heparin/HSPG-independent interaction is able to induce the expression of c-fos mRNA in a myeloid-derived cell line expressing FGF-R1 (17). It is generally believed that full biological activity of bFGF absolutely requires the presence of HSPG. However, the possibility that bFGF can function in the absence of HSPG has not been rigorously tested.
The exact signaling mechanisms used for bFGF-induced mitogenesis are not known; however, it appears that bFGF acts through multiple pathways. For example, bFGF stimulates phosphatidylinositol turnover and Ca 2ϩ flux, but these responses are not required for bFGF to stimulate cell proliferation or differentiation (24,25). Furthermore, bFGF is processed by at least two distinct pathways after internalization (26). While a considerable amount of bFGF is degraded in lysosomes, a fraction is translocated to the nucleus where it appears to be resistant to degradation (27)(28)(29)(30). The factors that control bFGF internalization and processing are not well understood, and the role of HSPG has not been fully explored.
In this study we have investigated the binding, internaliza-* This study was supported by American Heart Association Grantin-Aid 13-522-923, a Whitaker Foundation Biomedical Engineering Research grant, Grant-in-Aid GA93046 from the Fight For Sight Research Division of the National Society to Prevent Blindness, by Grant IRG-97 R from the American Cancer Society, and departmental grants from the Massachusetts Lions Eye Research Fund and Research to Prevent Blindness, Inc. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom communication should be addressed: Boston University School of Medicine, Room L912, 80 East Concord St., Boston, MA 02118. Tel.: 617-638-4526; Fax: 617-638-5337. 1 The abbreviations used are: bFGF, basic fibroblast growth factor; DMEM, Dulbecco's modified Eagle's medium; HSPG, heparan sulfate proteoglycans; K d , equilibrium dissociation constant; k e endocytotic rate constant; k on , association rate constant; k off , dissociation rate constant; tion, and activity of bFGF in the presence and absence of HSPG in Balb/c3T3 cells. Balb/c3T3 fibroblasts were treated with sodium chlorate to completely inhibit sulfation of proteoglycans and binding of bFGF to HSPG. bFGF bound to its receptors on chlorate-treated cells with a lower apparent affinity. We further analyzed bFGF internalization and stimulation of DNA synthesis in control and chlorate-treated cells. Endocytotic rate constants (k e ) for chlorate-treated and control cells were not statistically different, suggesting that the intrinsic process of bFGF internalization is not altered by HSPG. bFGF stimulated DNA synthesis to the same maximal level under both conditions, but chlorate-treated cells were significantly less responsive at low bFGF doses. The differences observed for control and chlorate-treated cells in the dose-response curves for stimulation of DNA synthesis and receptor binding correlated directly, suggesting that activated FGF receptors are equally capable of eliciting a mitogenic signal in the presence and absence of HSPG. While the presence of HSPG on the cell surface increases the affinity of bFGF for its receptors, our observations suggest that HSPG are not "absolutely" required for binding, internalization, or stimulation of mitogenic activity. Cell Culture-Balb/c3T3 fibroblasts were from the American Type Culture Collection. Cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Inc.), supplemented with penicillin (100 units/ml), streptomycin (100 g/ml) (Life Technologies, Inc.), glutamine (2 mM), and 10% calf serum (Hyclone, UT) (13,32).

Materials-bFGF
Chlorate Treatment-Cells were plated at 37,500/cm 2 in DMEM with 10% calf serum, penicillin, streptomycin, and glutamine at 37°C for all experiments. After 24 h, medium was changed to DMEM with 2% dialyzed calf serum (Sigma) and glutamine with or without 50 mM chlorate. Cells were incubated for 72 h at 37°C. The optimum serum concentration was arrived at by testing various concentrations over a 72-h period to determine a level at which cell number and viability remained constant. The optimum chlorate level was determined by testing various concentrations, under the same experimental conditions, to find a concentration at which sulfation of proteoglycans would be maximally inhibited. Under these conditions cell viability was maintained at Ն89% for both chlorate and control cultures.
Zeta Probe Analysis of Proteoglycan-Cells were prepared as described above. [ 35 S]Sulfate (100 Ci/ml) was added to cells, and they were incubated at 37°C for 72 h. Medium was collected and extracellular matrix and cell surface proteoglycans were extracted and quantitated by cationic nylon filtration as described previously (32)(33)(34).
bFGF Binding--Equilibrium binding of 125 I-bFGF was conducted with confluent Balb/c3T3 cells (13). Cells in 24-well plates (Costar, Cambridge, MA) were washed once with binding buffer (DMEM, 25 mM HEPES, 0.05% gelatin) at 4°C. Fresh binding buffer was added (0.5 ml/well), and cells were incubated at 4°C for 10 min. 125 I-bFGF was added at the indicated concentrations, and cells were incubated at 4°C for 2.5 h. At the end of the binding period cells were placed on ice and washed three times with ice-cold binding buffer. HSPG-bound 125 I-bFGF was removed with a quick wash (ϳ5 s) using 2 M NaCl in 20 mM HEPES (pH 7.4), followed by a wash with PBS. Cell surface receptorbound 125 I-bFGF was extracted with two washes (one 5-min and one rapid wash) at room temperature using 2 M NaCl in 20 mM sodium acetate (pH 4.0). To confirm that these washes removed all of the surface bound bFGF and that there was no significant internalization at 4°C, the remaining cell layer was extracted in 0.5% Triton X-100 and counted for 125 I in some experiments. No significant radioactivity was present in the detergent-extracted cells. Nonspecific binding was determined empirically at three concentrations of 125 I-bFGF (1, 2, and 5 ng/ml) by competing with increasing concentrations of unlabeled bFGF until no further reduction in label binding was observed (40 g/ml bFGF for control cells and 5 g/ml bFGF for chlorate-treated cells). Nonspecific binding was observed to relate linearly with the concentration of 125 I-bFGF as expected. Thus, equations were derived in each case to determine the fraction of the total bound component that constituted nonspecific binding at each 125 I-bFGF concentration (35). The specific binding data were fit to the following generalized equation (36) for the determination of binding constants and numbers of binding sites for any number (x) of separate binding site classes: where K and N represent the association constant(s) and number of binding sites, respectively, and [FGF] b and [FGF] f represent the concentration of bound and free bFGF, respectively. Curve fitting was conducted using the Levenberg-Marquardt algorithm with the program KaleidaGraph version 3.0.4, (Synergy Software, Reading, PA) on a Macintosh IIvx. bFGF Internalization-To analyze 125 I-bFGF internalization, cells were washed once with warm binding buffer (37°C), and 0.5 ml/well of binding buffer was added. 125 I-bFGF was added directly to the binding buffer and the cells incubated at 37°C for the indicated times. At each time point cells were placed on ice, washed, and the HSPG and cell surface receptor-bound 125 I-bFGF removed as described above. To extract internal 125 I-bFGF, cells were solubilized with two washes of 0.5% Triton X-100. Samples were counted in a Packard Model #5650 gamma counter.
Determination of k e -The amount of 125 I-bFGF bound on the cell surface and internalized over time at 37°C was used to determine the endocytotic rate constant (k e ; units of min Ϫ1 ) (37), as: Internalization data were fit to Equation 3, and a linear regression with correlation analysis was performed to determine k e using the software, Primer to Biostatistics Version 2.0, (McGraw-Hill, Inc.) on a Macintosh IIvx computer.
Cross-linking-Cells (100-mm dishes) were treated with and without chlorate at 37°C for 72 h. Cells were washed once in cold binding buffer (4°C), and then fresh binding buffer (4°C) was added, and the cells were incubated at 4°C for 10 min. bFGF (concentrations indicated in the figure) was added and the binding incubation carried out at 4°C for 2.5 h. Cross-linking was conducted using disuccinimidyl suberate (final concentration 0.3 mM) in PBS at room temperature for 20 min (13). The reaction was quenched with 50 mM Tris-HCl (pH 8.0) and 100 mM glycine for 10 min. Cells were washed three times with PBS and extracted with boiling sample buffer using a cell lifter (Costar). Samples were resolved on 5% SDS-PAGE gels, electroblotted onto Immobilon-N membranes, and the location of the cross-linked bFGF visualized by immunoblot analysis with an anti-bFGF antibody.
Western Blot Analyses of FGF Receptor and bFGF-For FGF receptor analysis, control and chlorate-treated cells were washed twice in PBS, and then 1 ml of HTG (20 mM HEPES, 1% Triton X-100, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride) was added, and cells were scraped with a cell lifter. Lysate was centrifuged at 10,000 ϫ g for 10 min. Some samples were precleared by incubating lysate with 100 l of bovine serum albumin-agarose beads (Sigma) for 1 h at 4°C followed by centrifugation at 10,000 ϫ g for 10 min (38). The supernatant was run on 5% polyacrylamide gels and electrotransferred to Immobilon-N (Millipore Corp, Bedford, MA). For both anti-FGF receptor and anti-bFGF analyses, membranes were blocked overnight in 5% milk and probed for 1 h at 37°C with either monoclonal antibody anti-FGF receptor or monoclonal antibody anti-bFGF type II at 1:1000 followed by 1-h incubation at 37°C with horseradish peroxidase-linked anti-mouse IgG (from sheep) at 1:1000. Bands were visualized with Renaissance Western blot chemiluminescence reagent (DuPont NEN) on Kodak O-MAT film.

H]Thymidine Incorporation-After 72 h treatment with or without chlorate, bFGF (concentration indicated in figures) and [ 3 H]thymidine
(1 Ci/ml) were added and the cells incubated for 36 h at 37°C. Cells were washed with PBS, fixed, and DNA precipitated with methanol and trichloroacetic acid (32). Total [ 3 H]thymidine incorporation was quantitated by liquid scintillation counting. In separate cultures, labeled nuclei were visualized by autoradiography. Cells were fixed in methanol and overlaid with NTB-2 (Kodak). Cells were exposed for 3 days and developed and fixed directly in wells. The number of labeled and total nuclei were counted manually in three separate (1-mm 2 ) fields in each culture under a light microscope.
Heparinase Treatment-Cells were washed once in prewarmed binding buffer. 0.5 ml of DMEM with 5 mg/ml bovine serum albumin was then added to cells with heparinase I, heparinase III, or both enzymes, and the cells were incubated for 1 h at 37°C. Cells were washed once in cold binding buffer, and binding experiments were conducted as described. In order to ensure that the binding buffer contained no heparan sulfate that could contaminate the experiment, control experiments were performed in which the binding buffer was pretreated with heparinase I (5 g/ml, 1 h, 37°C) prior to being used in the binding reaction. There were no differences in bFGF binding between cells with pretreated and untreated binding buffer (data not shown).

RESULTS
bFGF Binds to Its Cell Surface Receptor in the Absence of HSPG-Previous studies have suggested that bFGF binds its receptors in the absence of HSPG. We found that bFGF bound its receptors with a 5-20-fold decreased affinity after the removal of ϳ95% of the HSPG sites with heparinase I (13). However, the possibility that there were residual heparinaseresistant HSPG sites responsible for the receptor binding could not be formally ruled out. Thus, in the present study we used high concentrations (50 mM) of sodium chlorate to generate cells that contained no detectable sulfated proteoglycan (Table  I) and conducted bFGF binding studies with these cells. No specific 125 I-bFGF binding to HSPG sites was detected over a wide range of 125 I-bFGF concentrations with chlorate-treated cells (Fig. 1A). However, there was a considerable amount of specific binding in the receptor (low pH) fraction from chloratetreated cells (Fig. 1B). To ensure that this binding was not dependent on residual HSPG, or contaminating heparin/heparan sulfate within the binding reaction, chlorate-treated cells and/or the binding buffer were exhaustively digested with heparinase I and/or heparinase III prior to conducting bFGF binding. In these experiments there was no significant additional decrease in bFGF receptor binding in chlorate-treated cells (Fig. 2). The ability of these heparin/heparan sulfatespecific lyases to remove sulfated glycosaminoglycan from the surface of these cells was confirmed by quantitating total 35 Slabeled sulfate proteoglycan on treated cells (Table II). Thus, bFGF is capable of binding to its receptors in the absence of HSPG. Treatment with sodium chlorate inhibits sulfation of glycosaminoglycans, and the enzymes heparinase I and III specifically degrade heparan sulfate chains; thus the use of these treatments in combination suggest that un-sulfated heparan chains are also not required for receptor binding.
bFGF receptor binding to control and chlorate-treated cells  35 SO 4 (100 Ci/ml). The medium was collected and proteoglycan extracted from the extracellular matrix (ECM) and cell surface. 35 S-Sulfated proteoglycan was quantitated by cationic nylon filter binding (32)(33)(34). The results represent the average (Ϯ S.E.) of quadruplicate determinations. Similar results were observed in two additional experiments including one where the 35 1. Effects of chlorate treatment on the binding of 125 I-bFGF to Balb/c3T3 cells. Confluent Balb/c3T3 cells were treated with (q) or without (E) chlorate (50 mM) for 72 h at 37°C. Equilibrium binding of 125 I-bFGF was carried out with the indicated concentrations at 4°C. Specific binding was determined as described under "Experimental Procedures." A, binding to high salt releasable sites (HSPG) in control and chlorate-treated cells. B, binding to low pH releasable sites (predominately receptor) in control and chlorate-treated cells. The results presented represent the average of triplicate determinations at each bFGF concentration. Standard errors of the mean are presented for all points and are not visible if the error was smaller than the symbol. The cell number for control and chlorate-treated cells was the same (70,000/well). was further analyzed by the method of Scatchard to visualize the relative binding affinities and numbers of receptor sites on these cells (Fig. 3). While the binding of 125 I-bFGF to the chlorate-treated cells likely represents binding only to receptors, at high 125 I-bFGF concentrations the receptor-bound fraction from control cells contains some binding to HSPG (13). Therefore, the receptor-bound fraction from control cells was analyzed using a two-site model in which the high affinity component represents binding to receptors. The binding to chlorate-treated cells was analyzed as a single site. The number of receptor sites in control and chlorate-treated cells was similar; however, the dissociation constant was ϳ10-fold higher for receptors in chlorate-treated cells (0.104 nM) than for those in control cells (0.011 nM).
To confirm that chlorate treatment did not disrupt expression of FGF receptors, an immunoblot analysis was conducted with anti-FGF-R1 (Fig. 4A). A single prominent band representing FGF-R1 with an apparent molecular mass of ϳ110 kDa was present in lanes containing control and chlorate-treated cell extracts. There was no discernible difference in the relative amounts of FGF-R1 in control and chlorate-treated cell extracts. Furthermore, chlorate treatment had no detectable effect on the amount or the electrophoretic profile of total cellular protein (data not shown). To ensure that the binding of 125 I-bFGF to chlorate-treated cells (Fig. 1B) represents interaction with the same receptor species as in control cells, bound bFGF was covalently cross-linked to its receptors. The presence of FIG. 4. Immunoblot analysis of FGF receptors and bFGF crosslinked receptors in control and chlorate-treated Balb/c3T3 cells. Confluent Balb/c3T3 cells were treated with (ϩ) or without (Ϫ) chlorate (50 mM) for 72 h at 37°C. A, after chlorate treatment cells were extracted and electrophoresed on a 5% SDS-PAGE. Lanes were normalized to cell number; lanes 2 and 4 were precleared prior to analysis. Membranes were hybridized with anti-bFGF receptor and then with horseradish peroxidase-conjugated anti-IgG. Bands were visualized using enhanced chemiluminescence. A single prominent band was observed with an apparent molecular mass of 110 kDa. This band was not present in blots where the primary antibody was omitted. B, after chlorate treatment bFGF binding was conducted with 2 and 10 ng/ml bFGF with control and chlorate-treated cells, respectively, for 2.5 h at 4°C. Unbound bFGF was washed from the cell layer and receptorbound bFGF cross-linked with disuccinimidyl suberate. Cross-linked cells were extracted and electrophoresed on 5% SDS-PAGE and transferred to Immobilon membranes. Membranes were hybridized with anti-bFGF and visualized using horseradish peroxidase-conjugated anti-IgG and enhanced chemiluminescence. Two specific bands were labeled by the anti-bFGF antibody that were not labeled when the primary antibody was omitted.   35 S-proteoglycan from the cell surface of Balb/c3T3 cells Control and chlorate-treated Balb/c3T3 cells were radiolabeled for 3 days with 35 SO 4 (100 Ci/ml). At the end of the labeling period, the cells were washed once with DMEM and then treated with the indicated enzyme (5 g/ml for heparinase I and 1 unit/ml for heparinase III) in DMEM, 5 mg/ml bovine serum albumin, at 37°C for 45 min. After enzyme treatment, the cell surface proteoglycan was extracted and quantitated by cationic nylon filtration (32)(33)(34). Similar results were observed in one additional experiment.  affinity cross-linked bFGF on FGF-receptors was detected by immunoblot with an anti-bFGF antibody. In both control and chlorate-treated cells two bands (apparent molecular mass 128 and 165 kDa) were labeled with bFGF (Fig. 4B). Consistent with the difference in the calculated dissociation constants, the extent of bFGF cross-linking was similar in control and chlorate-treated cells when a higher concentration of bFGF was bound to chlorate-treated cells compared with control cells. The presence of two specific bFGF cross-linked bands has been reported previously with these cells (4 -6, 13). The difference between the anti-FGF receptor analysis (Fig. 4A) and the crosslinking profile (Fig. 4B) might reflect the presence of a second FGF receptor species that is not recognized by the anti-FGF receptor antibody, or alternatively, the higher molecular weight band in Fig. 4B might represent a single FGF receptor linked to a cross-linked bFGF dimer. In either case it is clear from the results in Fig. 4 that the effects of chlorate on bFGF binding result from decreased affinity of the FGF receptors and not an alteration in the amount or type of receptors expressed under these conditions. bFGF Is Internalized in the Absence of HSPG--Upon receptor binding both bFGF and its receptors are internalized, and it has been postulated that internalized bFGF might play a regulatory role (39). The ability to stimulate internalization might be an important characteristic of a functional bFGF-FGF receptor complex. To determine if the process of bFGF internalization is affected by HSPG, we compared the rates of endocytosis in control and chlorate-treated cells (Fig. 5). These experiments were conducted at 37°C over a period of time in which no detectable bFGF degradation occurred. A considerable amount of bFGF was internalized over this period in both control and chlorate-treated cells (Fig. 5, A and B). As would be predicted from the equilibrium binding analyses (Figs. 1 and  3), the absolute amount of bFGF bound and internalized in chlorate-treated cells was less than that in control cells. However, determination of the first order endocytotic rate constants, by normalizing the amount of bFGF internalized to that bound on the cell surface (Equation 3), revealed that the k e values were not significantly different in control and chloratetreated cells (k e ϭ 0.043 Ϯ 0.012 min Ϫ1 (average Ϯ S.E. of five separate experiments, n ϭ 5) for control; k e ϭ 0.078 Ϯ 0.022 min Ϫ1 (n ϭ 4) for chlorate-treated cells; p ϭ 0.179). The endocytotic rate constant was also determined in two separate experiments for cells treated with heparinase I (5 g/ml; 30 min at 37°C), and it did not differ significantly from that of the parallel control cells (k e ϭ 0.027 Ϯ 0.006 min Ϫ1 (n ϭ 2) for heparinase I-treated cells; k e ϭ 0.022 Ϯ 0.002 min Ϫ1 (n ϭ 2) for control cells; p ϭ 0.529). Furthermore, internalization studies were conducted with a range of bFGF concentrations (0.2-4 ng/ml), to control for differences in the relative receptor occupancy in the control and HSPG-deficient states. No significant differences were noted between k e values determined over this range of bFGF concentrations.
While the first order rate constants for endocytosis (k e ) were not significantly different for cells with or without HSPG, it is important to note that the differences in binding affinity between FGF receptors under these conditions translate into differing rates of internalization at subsaturating bFGF concentrations (see Equation 2). For example at a concentration of bFGF equal to the K d (0.011 nM) for FGF receptors on control cells (ϩHSPG), we would estimate that the rate of internalization in control cells would be on the order of 5-fold greater than that in HSPG-deficient cells. Our previous study using these cells determined the off-rate constants for FGF receptors under control and HSPG-deficient conditions (k off ϭ 0.003 min Ϫ1 for control; and k off ϭ 0.048 min Ϫ1 for HSPG-deficient cells) (13). These k off values, when taken together with the present k e values, suggest that a consequence of HSPG removal would be to alter the efficiency at which each receptor binding event leads to internalization as defined by the internalization probability (IP; Equation 4): IP ϭ k e k e ϩ k off (Eq. 4) This simple generalized expression represents the likelihood that an individual ligand-receptor binding event will lead to endocytosis of that ligand before it dissociates from the receptor. We calculate an IP of 0.95 and 0.55 for control and HSPGdeficient cells, respectively (based on an average of all k e values of 0.06 min Ϫ1 ). Thus essentially every bFGF-FGF receptor complex that forms on control cells would be internalized, while as much as 45% of the complexes that form on HSPG-deficient cells would be predicted to dissociate before they are internalized. It appears that the intrinsic process of FGF receptor internalization is not affected by the presence of HSPG, yet the overall rate of the internalization process is altered as a result of the decrease in bFGF-FGF receptor complex stability in the absence of HSPG. bFGF Stimulates DNA Synthesis in HSPG-deficient Cells-Many previous studies have demonstrated a loss of bFGF activity with a loss in HSPG expression (12, 14, 15, 18, 19, 23, 40 -44). However, in one report, it was demonstrated that bFGF stimulates the induction of c-fos mRNA expression in cells that do not express HSPG (17). In most of the previous studies where stimulation of mitogenesis was evaluated in the presence and absence of HSPG, activity was not analyzed over a wide range of bFGF concentrations. Thus, the possibility that the bFGF-FGF receptor complexes that form in the absence of HSPG are capable of transducing a mitogenic signal has not been formally tested. To determine if cells are capable of a mitogenic response to bFGF in the absence of HSPG, control and chlorate-treated cells were exposed to a wide range of bFGF concentrations and [ 3 H]thymidine incorporation into DNA measured after 36 h. bFGF stimulated DNA synthesis in both control and chlorate-treated cells. At low concentrations, bFGF was dramatically less effective in chlorate-treated cells compared with control cells. However, chlorate-treated cells were capable of responding to bFGF to the same maximal extent as control cells. The dose-response profiles for control and chlorate-treated cells were similar with the significant difference being an approximate 10-fold shift in the effective doses required to stimulate chlorate-treated cells (ED 50 ϭ 0.25 ng/ml, 14 pM for control; and ED 50 ϭ 2.50 ng/ml, 140 pM for chlorate-treated cells). Nuclear labeling and autoradiography demonstrated that the % labeled nuclei correlated with the thymidine incorporation, over the bFGF dose range tested, for control and chlorate-treated cells. To ensure that the bFGF responsiveness in chlorate-treated cells was not the result of small amounts of contaminating or residual heparin/heparan sulfate on the cells or in the medium, experiments were performed where heparinase I (5 g/ml) and/or heparinase III (5 units/ml) were added directly to the culture medium prior to the initiation of the chlorate treatment and just before the addition of bFGF. bFGF stimulated DNA synthesis in these combined chlorate/heparinase-treated cells to a similar extent as compared with cells treated with chlorate alone (data not shown).
The magnitude of the shift in the effective bFGF dose for mitogenic stimulation in HSPG-deficient cells was similar to the shift in receptor affinity (Figs. 3 and 6). The "specific activity" of activated FGF receptors for mitogenic stimulation in control and chlorate-treated cells was compared. Predicted relative receptor occupancies were calculated at each dose of bFGF that stimulated DNA synthesis and correlated with activity (Fig. 7). A direct correlation between predicted receptor occupancy and activity was observed for both control and chlorate-treated cells (R s ϭ 0.943, r ϭ 0.98, p Ͻ 0.01 for control; and R s ϭ 1.0, r ϭ 0.98, p Ͻ 0.001 for chlorate-treated cells). Furthermore, the relationship between receptor occupancy and activity was nearly identical for control and chlorate-treated cells (p ϭ 0.943). Therefore, the specific activities of bFGF-FGF receptor complexes that form in the presence and absence of HSPG are indistinguishable. DISCUSSION The role of growth factor interactions with components other than their receptors at the cell surface has been an area of intense research interest. We have investigated the role of heparan sulfate proteoglycans in controlling bFGF receptor interactions and mitogenic activity. While it has become generally believed that bFGF requires the presence of heparin/ heparan sulfate in order to stimulate mitogenesis, we show here that bFGF is able to stimulate DNA synthesis in HSPGdeficient Balb/c3T3 cells to the same maximal extent as in control cells expressing HSPG. While cells containing HSPG required significantly lower doses of bFGF for mitogenic stimulation than did HSPG-deficient cells, the difference in affinity of the FGF receptors was sufficient to account for this difference. In fact when the "specific mitogenic activity" of occupied FGF receptors was evaluated (Fig. 7) on HSPG-containing and -deficient cells, no difference was noted. Our data suggest that HSPG enhance bFGF activity by altering the affinity of FGF receptors, without directly altering the signal transduction pathway that is activated by occupied FGF receptors. Furthermore, our data also suggest that it is unlikely that binding of bFGF to HSPG directly induces a unique signal related to the mitogenic activity of bFGF.
Our conclusion that HSPG are not absolutely required for bFGF stimulation of DNA synthesis appears to contradict conclusions from several previous studies (12, 14, 15, 18, 19, 23, 40 -44). However, it is important to note that our present data are consistent with these earlier reports. The distinction in most instances results solely from the fact that a wider range of bFGF concentrations was evaluated in the present report. In most of the previous analyses a single concentration of bFGF was tested for mitogenic stimulation. The concentration of bFGF used was generally below that needed to produce a saturating response in the control HSPG expressing cells. For example, in the first report on the HSPG dependence of bFGFmediated growth stimulation, by Rapraeger et al. (12), the activity of 10 pM bFGF was investigated. In these studies little detectable stimulation of DNA synthesis by bFGF was observed in chlorate-treated Swiss 3T3 cells (approximately 10% of that in control cells). Although this exact concentration was not tested in the present study, the dose response presented in Fig. 6 predicts that we would observe only ϳ 5% of the maximum stimulation in chlorate-treated cells as compared ϳwith 50% of the maximum stimulation in control cells at a bFGF concentration of 10 pM. Thus we observed the same ϳ10-fold decrease in mitogenic activity with chlorate treatment at this bFGF concentration.
Although our results suggest that HSPG are not absolutely required for bFGF responsiveness, the 10-fold shift in receptor binding affinity and mitogenic activity highlight the importance of HSPG in this system. The highest soluble concentrations of bFGF reported to date in vivo are on the order of 1 ng/ml (0.055 nM) in the aqueous humor of the eye (45). If the physiologically relevant range of bFGF concentrations is 0 -1 ng/ml, then our results would predict that the ability to express HSPG would often be required for cells to efficiently bind and respond to bFGF in vivo. Additionally, the ability of heparin/ heparan sulfate to protect bFGF from protease degradation and physical denaturation could also significantly amplify bFGF activity (1-3, 9, 10). Thus, our results reinforce those previously reported where bFGF activity was inhibited when binding to cell-associated HSPG was prevented through degradation of HSPG, inhibition of HSPG synthesis, the use of genetically selected/engineered cells, and addition of soluble competitors (12, 14, 15, 18, 19, 23, 40 -44). Consistent with this model are our recent findings that soluble endothelial-derived HSPG are potent inhibitors of bFGF binding and mitogenesis in vascular smooth muscle cells in vitro (41) and in vivo (34,46). These previous results, combined with those presented here, suggest that regulation of cell-associated HSPG and soluble antagonists could significantly affect bFGF activity. In addition to this important natural mechanism of bFGF control, it is also likely that pharmacological agents directed at modulating HSPG expression could in turn alter the biological effectiveness of endogenous as well as exogenously administered bFGF.
Receptor-mediated internalization of growth factors has been proposed to act as either a component of growth factor stimulation or as a step in the deactivation of signaling. In particular, there is considerable data suggesting that internalized bFGF plays an important role in the cell (27)(28)(29)(30). In this report we have defined the endocytotic rate constants for bFGF internalization in the presence and absence of HSPG. No significant difference was noted between the k e values in control and chlorate-treated cells suggesting that the process of internalization is not sensitive to the presence of HSPG. However, it is important to note that the lack of difference in the k e values does not rule out an important function for internal bFGF since the amount internalized in the absence of HSPG would be proportionately less than in control cells, as defined by the differences in the receptor affinities (see Equation 2). That is, the shift in the dose response observed for mitogenic activity (Fig. 6) would correlate with a predicted shift in the absolute amount of bFGF internalized over this concentration range. Thus, the molecular mechanism of internalization does not likely involve HSPG, but the role of internalized bFGF in mitogenic stimulation remains unclear.
Equilibrium binding constants generated at 4°C are widely used to predict relative binding to cells under dynamic conditions at 37°C (i.e. Fig. 7), yet these analyses must be interpreted with caution. Not only could significant deviations exist between measurements made at 4 and 37°C, but the magnitude of the individual rate constants, k on and k off , in compari- FIG. 7. Correlation between FGF receptor occupancy and stimulation of DNA synthesis by bFGF in control and chloratetreated Balb/c3T3 cells. The DNA synthesis stimulatory activity of bFGF (Fig. 6) over a range of concentrations that produced increased activity with increased dose (0 -2.5 ng/ml for control, 0 -12.5 ng/ml for chlorate-treated cells) was correlated with binding of bFGF to its receptors. The percentage of available receptors that would be occupied at each bFGF concentration was determined for control and chloratetreated cells using the following relationship: % receptor occupancy ϭ ͓bFGF͔ ͑K d ͒ ϩ ͓bFGF͔) ⅐ 100 (Eq. 5) Data from control (E) and chlorate-treated cells (q) were independently analyzed for correlation between receptor binding and activity using the Spearman rank-order correlation test (Primer to Biostatistics Version 2.0). Under both conditions binding correlated almost precisely with mitogenic activity; r ϭ 0.98, R s ϭ 0.943, p Ͻ 0.01 for control and r ϭ 0.98, R s ϭ 1.0, p Ͻ 0.001 for chlorate-treated cells. Furthermore, the relationship between binding and activity under the two conditions was almost identical (slope ϭ 1.07 Ϯ 0.10 control, and slope ϭ 1.06 Ϯ 0.09 for chlorate-treated cells, p ϭ 0.943).
son to those for subsequent stages of the dynamic process (i.e. k e ) of growth factor-cell interaction must also be taken into account to generate a more accurate picture. The binding of bFGF to its receptors has been analyzed extensively in cell and cell-free systems at several temperatures, and no significant temperature dependence has been noted for the equilibrium dissociation constant (see Refs. 4,17,20,47 for recent discussion). Thus, the relative differences in the K d values for FGF receptors with and without HSPG measured at 4°C (Figs. 1  and 3) are likely to reflect differences at 37°C. We have previously noted that the difference in the K d values for receptors in the presence and absence of HSPG results from differences in the k off values and not k on values under these conditions (13). Therefore it is critical to compare the magnitude of the k e to the k off values to determine if the change in K d (ϮHSPG) has any functional significance at 37°C. For example, if the k e values were very much greater than either (ϮHSPG) k off , then the change in k off, and in K d , would not have any impact when internalization is allowed. However, in our particular situation, the k e is not significantly greater than the k off for receptors in the absence of HSPG, suggesting that the changes in equilibrium binding at 4°C translate into significant differences at 37°C. While the analysis presented in Fig. 7 likely reflects the relative receptor occupancy over the range of bFGF concentrations tested, it does not imply that the precise number of occupied receptors is constant or known over the course of the mitogenic stimulation experiments at 37°C. In this report we have measured bFGF binding, internalization, and stimulation of DNA synthesis with Balb/c3T3 cells under control and sulfated proteoglycan-deficient states. We found that bFGF can bind its receptor, be internalized, and stimulate DNA synthesis in the absence of HSPG. HSPG removal caused a significant (ϳ10-fold) shift in the effective dose for bFGF binding, internalization, and mitogenesis. These results suggest that HSPG are not absolutely required for these processes. Instead it appears that HSPG act primarily at the cell surface to stabilize the formation of bFGF-FGF receptor complexes. Although the dependence of bFGF cell stimulation on HSPG may not be absolute, it is important to stress that our data provide additional evidence for a critical role for HSPG in the bFGF system. A 10-fold shift in bFGF binding and mitogenic activity could have a significant functional impact. This mechanism for HSPG modulation of bFGF requires only that HSPG bind bFGF on the cell surface, suggesting that HSPG could play a similar function for a wide range of heparinbinding growth factors. Thus HSPG might play a general role on the cell surface by amplifying the ability of the cell to sense the presence of dilute signaling molecules in its immediate extracellular environment.