Identification of glypican as a dual modulator of the biological activity of fibroblast growth factors.

Heparan sulfate moieties of cell-surface proteoglycans modulate the biological responses to fibroblast growth factors (FGFs). We have reported previously that cell-associated heparan sulfates inhibit the binding of the keratinocyte growth factor (KGF), but enhance the binding of acidic FGF to the KGF receptor, both in keratinocytes, which naturally express this receptor, and in rat myoblasts, which ectopically express it (Reich-Slotky, R., Bonneh-Barkay, D., Shaoul, E., Berman, B., Svahn, C. M., and Ron, D. (1994) J. Biol. Chem. 269, 32279-32285). The proteoglycan bearing these modulatory heparan sulfates was purified to homogeneity from salt extracts of rat myoblasts by anion-exchange and FGF affinity chromatography and was identified as rat glypican. Affinity-purified glypican augmented the binding of acidic FGF and basic FGF to human FGF receptor-1 in a cell-free system. This effect was abolished following digestion of glypican by heparinase. Addition of purified soluble glypican effectively replaced heparin in supporting basic FGF-induced cellular proliferation of heparan sulfate-negative cells expressing recombinant FGF receptor-1. In keratinocytes, glypican strongly inhibited the mitogenic response to KGF while enhancing the response to acidic FGF. Taken together, these findings demonstrate that glypican plays an important role in regulating the biological activity of fibroblast growth factors and that, for different growth factors, glypican can either enhance or suppress cellular responsiveness.

of heparin-binding proteins such as extracellular matrix components and growth factors (1)(2)(3). Studies in recent years have strongly indicated that HSPGs are important modulators of the activity of heparin-binding growth factors (3), an issue that has been particularly well studied for fibroblast growth factors (FGFs).
FGFs constitute a large family of polypeptides that are important in the control of cell growth and differentiation and play a key role in oncogenesis and developmental processes including limb formation, mesoderm induction, and neuronal development (4). FGFs elicit their biological activities by interaction with four distinct cell-surface tyrosine kinase receptors (FGFR1-FGFR4) that display overlapping affinities for the various FGFs (5). Several members of the receptor family also exist in alternatively spliced forms that display altered ligand binding properties (5). For example, the KGF receptor (KGFR) is a splice variant of FGFR2. Whereas FGFR2 interacts with aFGF and bFGF, but not with KGF, KGFR binds aFGF and KGF and exhibits a significantly reduced affinity for bFGF (6).
Heparan sulfates or heparin can modulate the activities of FGFs by several mechanisms. They can stabilize FGFs by protecting them from proteolysis and thermal denaturation (7,8). They can also increase the affinity of FGFs for their signaling receptors (8 -10) and facilitate receptor dimerization and subsequent signaling (10 -13). On the other hand, these molecules can also inhibit the activities of FGFs (14). The mechanisms by which HS exert these multiple effects are not very well understood.
The modulatory effects of the low affinity binding sites were studied mainly with HS extracted from cells or with heparin, which shares structural similarity with HS and thus can mimic the action of cell-or matrix-associated HSPGs (15). Because the level of expression of HSPGs and the ability of cells to synthesize HS side chains of a defined structure are developmentally regulated (16 -18), it is critical to identify the core proteins bearing such modulatory side chains, to elucidate whether the cores influence the structure and function of the glycosaminoglycans attached to them, and to determine the structural requirements for growth factor interaction with the HS moiety. Until now, most of the attempts were concentrated on the identification of native HSPGs that modulate interaction of bFGF with FGFR1. Perlecan, the large basal lamina proteoglycan, was identified as a major candidate for the bFGF low affinity accessory receptor (19). In addition, syndecans and glypican can either inhibit or stimulate bFGF/FGFR1 interactions and signaling depending on the cell type in which they are expressed or their level of expression (14,20,21). Little is known about the identity of HSPGs that bind and modulate the activities of other members of the FGF family.
In a previous study, we reported that cell-associated HSPGs * This work was supported by grants from the Israeli Cancer Research Fund, the Israeli Ministry of Health, the Israel Science Foundation-Centers for Excellence Programs, and the Gesellschaft fuer Biotechnologische Forschung (to D. R.). 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.
ʈ To whom correspondence should be addressed. Tel.: 972-4-8294-217; Fax: 972-4-8225-153; E-mail: DinaR@Techunix.Technion.ac.il. 1 The abbreviations used are: HS, heparan sulfate(s); HSPG, heparan sulfate proteoglycan; FGF, fibroblast growth factor; FGFR, fibroblast growth factor receptor; aFGF, acidic fibroblast growth factor; bFGF, basic fibroblast growth factor; KGF, keratinocyte growth factor; KGFR, keratinocyte growth factor receptor; FCS, fetal calf serum; PAGE, polyacrylamide gel electrophoresis; GAG, glycosaminoglycan; HPLC, high pressure liquid chromatography. exert a differential effect on the binding of KGF and aFGF to KGFR (22), which binds both growth factors equally well (23). Thus, treatment of cells with a metabolic inhibitor of sulfation or with HS-degrading enzymes reduced the binding of aFGF to KGFR, but enhanced the binding of KGF. Addition of heparin reversed the effect (22). This differential effect was observed both in keratinocytes, which naturally express KGFR, and in the rat myoblast cell line L6E9, which ectopically expresses this receptor (22). This study was carried out to identify the proteoglycan that may be responsible for this effect. Here, we report the purification of such a proteoglycan from rat myoblasts and its identification as glypican (24). We show that glypican exerts either a stimulatory or an inhibitory activity that is dependent on the type of growth factor.

EXPERIMENTAL PROCEDURES
Materials-Recombinant aFGF, bFGF, and KGF were produced in bacteria as described (25,26). Bovine brain aFGF was purchased from R&D Systems. Carrier-free Na 125 I and Na 2 35 SO 4 were purchased from DuPont NEN. Fetal and newborn calf sera and media were purchased from Life Technologies, Inc. Heparin from bovine lung and all other chemicals were purchased from Sigma. Chondroitinase ABC was from Seikagaku, and heparinases I and III were from Ibex Technologies. Fibronectin was purchased from Upstate Biotechnology, Inc.
Construction and Production of Soluble Human FGFR1-A soluble extracellular domain of the short isoform of human FGFR1 (hR1 (31)) was cloned into the APtag vector to produce an in-frame fusion of hR1 with secreted placental alkaline phosphatase (32). This plasmid was cotransfected with the selectable Neo R marker into NIH/3T3 cells. Conditioned medium from G418-resistant colonies was screened for alkaline phosphatase activity (32). The clone that produced the highest level of activity was expanded and used to purify the hR1/alkaline phosphatase fusion protein using bFGF affinity chromatography.
Receptor Binding Assays-Subconfluent cultures in 24-well plates were subjected to a binding assay as described previously (22,27). Binding sites were distinguished as low or high affinity by salt washes at neutral or low pH, respectively (0.5 M NaCl for KGF and 1 M NaCl for aFGF) (22). Binding of FGFs to soluble FGFR1 (hR1/alkaline phosphatase) was assessed following covalent cross-linking. Affinity-purified hR1/alkaline phosphatase (0.15 alkaline phosphatase A units/min) was incubated for 1.5 h at room temperature in 50 l of HEPES binding buffer (100 mM HEPES, 150 mM NaCl, 5 mM KCl, 1.2 mM MgCl 2 , 8.8 mM dextrose, and 0.1% bovine serum albumin) containing 2 ng/ml radioiodinated FGF. Cross-linking was performed with 0.2 mM disuccinimidyl suberate for 30 min. The reaction was quenched by adding 20 mM glycine, and ligand-receptor complexes were visualized following SDS-PAGE and autoradiography.
Preparation of FGF Affinity Column-Coupling of the growth factors to Sepharose CL-2B (Sigma) was done as described previously (35). Briefly, 1 mg of growth factor (aFGF, bFGF, or KGF) was coupled to 1 ml of packed beads in the presence of 1 mg of heparin. Following coupling, the column was extensively washed with 3 M NaCl to remove heparin and was stored in 10 mM Tris, pH 7.0, 0.5 M NaCl, and 10 mM dithiothreitol. Coupling efficiency in different preparations was between 35 and 80%.
Purification of FGF-binding HSPG from Rat Myoblasts-Subconfluent cultures of L6E9 cells (100 -200 tissue culture dishes of 150-mm diameter) were maintained for 16 h in nutrient mixture F-12 containing 10% dialyzed FCS. A portion of the cultures was metabolically labeled with Na 2 35 SO 4 (30 Ci/ml). Peripheral membrane proteins were extracted with HEPES binding buffer containing 1.5 M NaCl, and the extracts were pooled and concentrated 5-10 times in dialysis tubing with dry polyethylene glycol and dialyzed overnight against 10 mM Tris, pH 7.0, and 0.2 M NaCl. The dialyzed material was applied to a 20-ml DEAE-Sephacel column (Sigma) pre-equilibrated with 10 mM Tris, pH 7.0, and 0.2 M NaCl, and the column was washed with 10 -20 volumes of Tris buffer containing 0.3 M NaCl. Elution of the bound material was performed with a linear gradient of NaCl (0.4 -1.2 M NaCl in 10 mM Tris, pH 7.0) at a flow rate of 24 ml/h. Radiolabeled proteoglycan peaks were identified by measuring the radioactivity in a liquid scintillation counter, by SDS-PAGE and autoradiography, or by safranin O staining (36). Samples containing HSPGs were identified following nitrous acid deamination and solid-phase assay on cationic nylon (Zeta-probe) as described (37). Fractions containing HSPGs were pooled, diluted to reduce the salt concentration to 0.2 M, and applied to a 4-ml aFGF affinity column. The column was extensively washed with 0.2 M NaCl, and elution was performed by a stepwise increase in the NaCl concentration (see below).
Enzymatic and Chemical Deglycosylations-Enzymatic deglycosylation was carried out in Dulbecco's phosphate-buffered saline containing 125 I-HSPG and 0.5 unit/ml heparinases I and III or chondroitinase ABC. Incubation was for 2 h at 37°C. Deaminitive scission of HS with HNO 2 was performed as described (38). Anhydrous trifluoromethanesulfonic acid was used to strip the peripheral sugars (39).
Isolation of GAG Side Chains-The core protein of the HSPG was digested overnight with proteinase K (0.5 mg/ml). To ensure that digestion was complete, a parallel incubation was carried out in the presence of radioiodinated HSPG, and digestion was monitored by SDS-PAGE and autoradiography. The GAG side chains were then separated from the protein and degradation products by DEAE-Sephacel chromatography.
Protein Sequencing-Affinity-purified HSPG was digested with modified trypsin (Promega). The peptides were loaded onto DEAE-Sephacel in the presence of 0.2 M NaCl to absorb GAG-containing peptides. Non-absorbed peptides were resolved by reverse-phase HPLC on a 1-mm diameter Vydac C 18 column. The peptides were sequenced using standard chemistry on an Applied Biosystems sequencer (Model 476A). Mass spectrometry was done using a matrix-assisted laser desorption/ ionization time-of-flight apparatus (Fisons Instruments, Inc.).
Quantitation of GAG and HSPG-Safranin O dye, which reacts with carboxyl and sulfate groups and can detect nanogram amounts of sulfated GAGs (36), was occasionally used to monitor GAG content during purification. For the biological assays, we normalized the concentration of glypican HS relative to known concentrations of heparin on the basis of their sulfate content using the dimethylmethylene blue assay (40). We estimated that glypican HS contain ϳ50% of the sulfate content of heparin.
Mitogenic Assay-Balb/MK cells, plated on fibronectin (1 g/cm 2 )coated 96-well microtiter plates, were serum-starved for 48 h and incubated with growth factors for 16 h. BaF3 cells were washed three times with Dulbecco's phosphate-buffered saline, placed in 96-well microtiter plates (2 ϫ 10 4 cells/well) in RPMI 1640 medium containing 10% FCS, and incubated with growth factors for 36 h. [ 3 H]Thymidine incorporation was assayed as described previously (29,41). Each point in the experiment was done in duplicate or triplicate, and the results are representative of at least four different experiments.

Mode of Association of the Modulatory HSPG with L6E9
Cells-HSPGs exist either as integral or peripheral membrane proteins or as part of the extracellular matrix (1,42). To determine the mode of association of the HSPG with parental L6E9 cells, we treated L6E9 cells expressing KGFR (designated L6/ KGFR cells) with increasing concentrations of salt, a condition that is known to remove peripheral membrane proteins. L6/ KGFR cells were then assayed for binding of radioiodinated aFGF and KGF. We reasoned that if the HSPG is salt-extractable, its removal should differentially affect the binding of aFGF and KGF to KGFR. As shown in Fig. 1, extraction of L6/KGFR cells with increasing salt concentrations resulted in a progressive reduction of the binding of aFGF to both low (Fig.  1A) and high (Fig. 1B) affinity receptors. In samples that were extracted at 1.5 M NaCl, the binding of aFGF to low and high affinity receptors was reduced by 60 and 80%, respectively. By contrast, salt extraction enhanced the binding of KGF to KGFR by up to 1.6-fold (Fig. 1B), whereas little or no effect was observed with respect to binding to low affinity receptors (Fig.  1A). These results are in accordance with our previous observations obtained following treatment of L6/KGFR cells with heparan sulfate-degrading enzymes or with a metabolic inhibitor of sulfation (22) and further suggest that the differential effect on the binding of aFGF and KGF to KGFR is mediated by a peripheral membrane HSPG. The lack of reduction of the binding of KGF to low affinity receptors in the salt-extracted cells is probably due to the presence of non-heparan sulfatebinding sites, which account for Ͼ90% of the low affinity sites for KGF in L6E9 cells (22).
Purification of the Peripheral HSPG-The putative HSPG was purified from salt extracts of parental L6E9 cells that were metabolically labeled with [ 35 S]sulfate. Purification was carried out by anion-exchange chromatography on DEAE-Sephacel followed by affinity chromatography on an aFGF column. The elution profile from DEAE-Sephacel is shown in Fig. 2A. Over 99% of the material was bound to the column as judged by the negligible amount of radioactivity in the flow-through fraction, and the material eluted at salt concentrations of up to 0.4 M NaCl. A major peak of 35 S eluted at ϳ0.6 M NaCl. This peak contained predominantly heparan sulfates as judged by its sensitivity to deaminitive cleavage with nitrous acid (Fig. 2A). The eluted material migrated on SDS-PAGE as a broad band of ϳ200 kDa ( Fig. 2A, inset). The fractions containing heparan sulfates were pooled and subjected to aFGF affinity chromatography. About 80% of the pooled material bound to the column and could be eluted by 0.75-1 M NaCl (Fig. 2B). SDS-PAGE and silver staining of the eluted material revealed the presence of a broad high molecular mass band similar to that observed with the 35 S-labeled DEAE fractions (Fig. 2C). The material purified by aFGF affinity chromatography also bound to bFGF and KGF columns (data not shown). This ability to interact with all three growth factors is in agreement with the previously described ligand binding characteristics of L6E9 HSPG (22).
Chemical Characterization of Purified Proteoglycan-Enzymatic and chemical deglycosylations were carried out to characterize the GAG side chain of the purified material and to determine the molecular mass of the core protein. To increase the sensitivity of detection, the core protein was radioiodinated. Fig. 3A shows the results of enzymatic deglycosylation carried out with heparinase I/III or chondroitinase ABC treatment. Heparinase, which specifically cleaves heparan sulfates, shifted the molecular mass of the broad band to a single band of ϳ64 kDa (Fig. 3A, lane 3). No shift in molecular mass was detected following treatment with chondroitinase ABC of either the intact HSPG or the preparation that had been treated with heparan sulfate-degrading enzymes (Fig. 3A, lanes 2  and 4). Fig. 3B shows the results of chemical deglycosylation. Deaminitive cleavage with nitrous acid resulted in a shift in the molecular mass of the HSPG that was similar to that observed following treatment with heparinase (Fig. 3B, lane 2). Treatment with trifluoromethanesulfonic acid further shifted the molecular mass of the HSPG to 54 kDa (Fig. 3B, lane 3). We conclude that the purified HSPG does not carry chondroitin sulfate and that heparan sulfates account for about two-thirds of the mass of the protein. The results obtained with trifluoromethanesulfonic acid indicate that other types of glycosylations account for ϳ10 kDa of the molecular mass.
Identification of the Purified HSPG as Glypican-The purified HSPG was digested with modified trypsin; GAG-containing peptides were removed by absorption to DEAE-Sephacel; and GAG-free peptides were resolved by reverse-phase HPLC. Analysis of two peptides gave the sequences LSDVPQAEIS-GEHLR and QAEALRPFGDAPR. A search in the GenBank™ using the Blast program revealed that the above sequences are identical to residues 46 -60 and 195-207 of rat glypican, respectively (24). It is noteworthy that even though the peptide encompassing residues 46 -60 contains the consensus sequence  2. Purification of L6E9 HSPG. Salt extracts from metabolically labeled L6E9 cells was applied to a column of DEAE-Sephacel as described under "Experimental Procedures." The column was extensively washed with 0.3 M NaCl and then eluted by a linear gradient of NaCl. Samples of 4 ml were collected, and aliquots were assayed for 35 S-labeled material (A, q) or analyzed by SDS-PAGE (3-18%) and exposure to an x-ray film (A, inset). Most of the radioactivity was associated with material that was sensitive to nitrous acid deamination as judged by the amount of radioactivity remaining on a cationic support (A, E). The fractions containing HSPGs from the DEAE column were pooled and purified by aFGF affinity chromatography (B). Fractions that eluted at 0.75 and 1 M salt were pooled, and 1 g of the purified material was separated by SDS-PAGE (6%) and silverstained (C).
for GAG attachment (24), this peptide is apparently not glycosylated. Mass spectrometry of several additional peptides confirmed that the purified HSPG is glypican (data not shown). Furthermore, the purified HSPG was immunoreactive with antibodies directed against bacterially expressed rat glypican, whereas no recognition was observed using antibodies against syndecan-1 and fibroglycan (Fig. 4).
Glypican Modulates Biological Activities of FGFs-It is well established that heparin potentiates the binding of FGFs to FGFR1 and is required for receptor-mediated signaling (8 -13). We therefore compared the ability of affinity-purified glypican and heparin to promote FGF receptor binding and mitogenic activity. Binding of bFGF and aFGF to FGFR1 was assayed in a cell-free system utilizing a soluble extracellular domain of human FGFR1 fused to alkaline phosphatase. As shown in Fig.  5, glypican augmented the binding of bFGF and aFGF to FGFR1 at concentrations as low as 10 and 25 ng/ml, respectively. Quantitation of the results from several experiments showed that both glypican and heparin gave a similar 4 -6-fold augmentation of bFGF binding, whereas heparin was somewhat more effective than glypican in stimulating the binding of aFGF. High concentrations of glypican (but not heparin) inhibited the binding of aFGF, but did not decrease the binding of bFGF. Similar results were obtained using different preparations of affinity-purified glypican as well as in binding assays performed with Chinese hamster ovary mutant cells that are defective in heparan sulfate synthesis and that ectopically express FGFR1 (data not shown) (43).
GAGs derived from glypican following digestion of the core protein were stimulatory for FGF binding, whereas digestion with heparinase abolished activity (see Fig. 5). These findings establish that the heparan sulfate moiety of glypican rather than the core protein is responsible for the modulation of growth factor-receptor interaction. Since heparinase I cleaves heparan sulfates in sulfate-rich regions (44), it is likely that such regions are responsible for the observed effects of glypican.
We next examined whether glypican-derived GAGs can promote the mitogenic response of cells to FGFs. Since Chinese hamster ovary mutant cells transfected with FGFR1 do not display a response to FGFs even in the presence of heparin (10), we have utilized the BaF3 lymphocytic cell line, which is both heparan sulfate-and FGFR-negative and engineered to express FGFR1 (designated F32 cells (29)). These cells are dependent on interleukin-3 for growth and can be relieved from interleukin-3 dependence in the presence of bFGF and heparin. Heparan sulfate derived from glypican acts like heparin to strongly stimulate the mitogenic response of F32 cells to bFGF, whereas it has no effect when added without the growth factor (Fig. 6).
The ability of glypican to modulate the interaction of aFGF and KGF with KGFR was examined utilizing Balb/MK cells, which express native KGFR. In this cell line, heparin inhibits the biological activity of KGF and potentiates that of aFGF (25). As shown in Fig. 7, both glypican-derived heparan sulfate and heparin augmented the mitogenic activity of aFGF over a similar concentration range (panel A). In contrast, glypicanderived HS, like heparin, inhibited the mitogenic response of Balb/MK cells to KGF (panel B), whereas it had no effect on  2 and 3, respectively). Samples were resolved by SDS-PAGE (6%), and the gel was dried and exposed to an x-ray film. [ 3 H]thymidine incorporation in the absence of the growth factor (panel C). Compared with heparin, glypican-derived HS were significantly more potent in inhibiting the biological activity of KGF (half-maximal inhibition at 2 and 0.2 g/ml heparin and glypican-derived HS, respectively). The effect of glypican on the mitogenic response of Balb/MK cells correlates well with the observed effects of endogenous heparan sulfate in L6E9 and Balb/MK cells on binding of aFGF and KGF to KGFR (Ref. 22 and Fig. 1). In addition, similar results were obtained in the mitogenic assays using intact glypican (data not shown).

DISCUSSION
In this study, we have identified glypican as a modulator of the biological activities of FGFs. We found that affinity-purified glypican can efficiently promote the interaction of FGFR1 with bFGF and aFGF, both in cells that are defective in heparan sulfate synthesis and in a defined cell-free assay system. Furthermore, glypican effectively replaced the need for heparin in the bFGF-mediated mitogenic response of heparan sulfatedeficient lymphoid cells that express recombinant FGFR1. Glypican was also found to differentially modulate the mitogenic response of keratinocytes to KGF and aFGF. In these cells, which naturally express KGFR, glypican effectively inhibited the mitogenic response to KGF while potentiating that of aFGF. Thus, purified glypican can exert either a stimulatory or an inhibitory activity that is dependent on the identity of the growth factor. To the best of our knowledge, this is the first demonstration of such a dual effect for a native cell-surface HSPG. The dual activity of glypican points to a possible regulatory mechanism involving cell-associated HSPGs. In such a mechanism, a cell-associated HSPG, depending on its local concentration, can either restrict or promote the biological activity of a given ligand. Because FGFs and their receptors display a high degree of cross-interaction, such a regulatory mechanism may also be important in the coordination of the interaction of different FGFs with the same FGFR.
Glypicans (16) and syndecans (45) represent the two major families of cell-surface proteoglycans. While the syndecans contain a membrane-spanning domain, the glypicans are anchored to the plasma membrane by a glycosylphosphatidylinositol linkage (46). The glypican family includes several closely related members: human and rat glypican (24,47); rat cerebroglycan (48); rat OCI-5 (49) and its human homologue, glypican-3 (50); K-glypican (51); and Drosophila Dally (52). Both glypican-3 and Dally have been recently implicated in the control of cellular growth. Mutations in Dally affect cell division and produce morphological defects in certain tissues of the fly, and mutations in glypican-3 are thought to cause the Simpson-Golabi-Behmel over-growth syndrome (50,52). It was suggested that proteoglycans may act as coreceptors for growth factors. In the case of dally, the ligand was not identified. In the case of glypican-3, insulin-like growth factor II was suggested as a putative ligand (50). This work provides evidence that glypican is involved in the regulation of cellular growth by modulating the biological activities of members of the FGF family. Such a role is supported by the reported similarity in the cellular and tissue distribution of glypican, FGFs, and their receptors (6,27,31,51,(53)(54)(55)(56)(57)(58)(59)(60).
Glypican exists on the surface of cells both as a lipid-anchored form and as a peripheral membrane proteoglycan most likely due to cleavage of the lipid anchor by a specific phospholipase (61). The peripheral form of glypican is clearly present in L6E9 cells since it could be extracted at high salt concentrations in the absence of detergents. In fact, calculation of the relative amount of glypican in the extracts revealed that glypican accounts for Ͼ80% of the high salt-extracted proteoglycans. The remaining 20% were inactive in modulating FGF receptor binding and were not further characterized (data not shown). We also observed that ϳ40% of the heparan sulfate-binding sites for FGFs were retained on cells following salt extraction (see Fig. 1). Because treatment of L6E9 cells with chlorate-or heparan sulfate-degrading enzymes reduced the level of heparan sulfate-binding sites for FGFs by Ͼ95% (22) and since the effect of these treatments on binding of aFGF and KGF to KGFR was similar to that observed following salt extraction, it is possible that the remaining 40% represent the lipid-anchored form of glypican. Alternatively, these remaining binding sites could represent HSPG species that are distinct from glypican (i.e. perlecan and syndecans (14,19)) and that, like glypican, can modulate receptor binding and activity of FGFs. Molecular analysis of the HSPG species in L6E9 cells as well as in keratinocytes, in which the differential effect was also observed  (22), and assessment of their modulatory activity on FGFs are currently being performed.
Initial attempts to identify bFGF stimulatory species of HSPGs from a total extract of human lung fibroblasts indicated that syndecan-1, -2, and -4 as well as glypican were inactive and even inhibited heparin-dependent bFGF receptor binding (20). Subsequently, perlecan was identified as the stimulatory component in these extracts (19). Based on these findings, it was concluded that perlecan is the major candidate for the bFGF low affinity accessory receptor. Our study suggests that the repertoire of stimulatory species is larger than perlecan alone. In support of this conclusion is a recent study showing that human glypican and syndecans can stimulate to a certain extent bFGF/FGFR1 binding and signaling when expressed in K562 cells (21). Taken together, these findings suggest that the extent and mode of action (stimulatory or inhibitory) of HSPGs may be regulated in a cell type-dependent manner. Alternatively, the lack or low stimulatory activity may be due to differences in the methods that were utilized for extraction and purification of HSPGs. It was also reported that in a cell-free binding assay, soluble syndecan-4 can stimulate bFGF/FGFR1 binding only when both the receptor and the proteoglycan are immobilized (21). Based on these findings, the authors concluded that cell-surface HSPGs have to be associated with the cell membrane to be active. Our findings that the stimulatory activity of glypican in cells or in the cell-free binding assay can be reconstituted either with a soluble form of intact proteoglycan or with the GAG side chain alone strongly suggest that, at least for glypican, this is not necessarily the case. Moreover, previous studies showed that a lipid-anchored HSPG can stimulate the biological activity of bFGF when released by phosphatidylinositol-specific phospholipases from the surface of stromal bone marrow and endothelial cells (62,63). The issue of whether membrane HSPGs are active in a soluble form is critical since soluble forms of both syndecans and glypicans are produced by shedding from cell surfaces (45,47). It is therefore important to establish whether or not these soluble forms retain the ability to modulate growth factor responsiveness.
The mechanism by which HSPGs promote FGF receptor binding and biological activities is complex and is manifested at several levels, including stabilization of the ligand or the ligand-receptor complex or facilitation of receptor dimerization and subsequent activation (7,8,(12)(13)42). The stimulatory effect of glypican on receptor binding and mitogenic activity of aFGF and bFGF may be due to stabilization of the growth factor or its complex with the receptor, whereas the inhibitory effect of glypican on KGF may involve, for example, competition with KGF for binding to KGFR. Alternatively, it is possible that an unidentified HSPG exerts a stimulatory effect on KGF, similar to the situation described for bFGF and aFGF, and that glypican acts as a competitive inhibitor of this HSPG. If such a situation exists, the extent of cellular responses to KGF may be determined by a balance between the level of inhibitory and stimulatory species of cell-associated HSPGs. Studies are underway to address these possible mechanisms.
In summary, our data demonstrate, for the first time, that a native proteoglycan can exert both stimulatory and inhibitory activity that is dependent on the identity of the growth factor and that this activity can be reconstituted with a soluble form of the proteoglycan. The ability of glypican to modulate the interaction of FGFs with different FGFRs strongly suggests a regulatory role for glypican in the control of a variety of biological processes in which FGFs are implicated, including cellular growth and differentiation. Because glypican is a prototype of a family of HSPGs, it is likely that the other family members will play a similar role in modulating the biological activities of FGFs or other heparin-binding growth factors.