Not All Perlecans Are Created Equal

Human basement membrane heparan sulfate proteoglycan (HSPG) perlecan binds and activates fibroblast growth factor (FGF)-2 through its heparan sulfate (HS) chains. Here we show that perlecans immunopurified from three cellular sources possess different HS structures and subsequently different FGF-2 binding and activating capabilities. Perlecan isolated from human umbilical arterial endothelial cells (HUAEC) and a continuous endothelial cell line (C11 STH) bound similar amounts of FGF-2 either alone or complexed with FGFRα1-IIIc or FGFR3α-IIIc. Both perlecans stimulated the growth of BaF3 cell lines expressing FGFR1b/c; however, only HUAEC perlecan stimulated those cells expressing FGFR3c, suggesting that the source of perlecan confers FGF and FGFR binding specificity. Despite these differences in FGF-2 activation, the level of 2-O-and 6-O-sulfation was similar for both perlecans. Interestingly, perlecan isolated from a colon carcinoma cell line that was capable of binding FGF-2 was incapable of activating any BaF3 cell line unless the HS was removed from the protein core. The HS chains also exhibited greater bioactivity after digestion with heparinase III. Collectively, these data clearly demonstrate that the bioactivity of HS decorating a single PG is dependent on its cell source and that subtle changes in structure including secondary interactions have a profound effect on biological activity.

The importance of perlecan to mammalian development has been demonstrated recently (17,18) by the results of two perlecan gene knockout experiments in mice. Nearly half of all perlecan null mice die at embryonic day 10.5, the normal onset of perlecan expression, or just after birth with severe defects including aberrant basement membrane formation (particularly in the heart), defective cephalic and long bone development, and achondroplasia (17,18). Similar abnormalities in cartilage development and bone ossification have been identified in mice with activating mutations in fibroblast growth factor (FGF) receptor 3. As perlecan co-localizes with this receptor (17), it has been suggested to act as a negative regulator of FGFR3 signaling, possibly by sequestering and inactivating its ligand, fibroblast growth factor 1 (FGF-1), via the HS chains.
Perlecan has also been shown to regulate FGF-2 activity in vitro (16) and in vivo through its HS chains (5,16). HS (or heparin, a widely available analogue of sulfated domains in HS) is essential to FGF-2 cell signaling, as demonstrated by the fact that addition of heparin or HSPG was shown to restore growth in FGF-responsive HSPG-deficient cells (19 -21). FGF-2 interacts with a specific HS sequence that consists of a hexasaccharide containing 2-O-sulfated iduronic acid (IdoUA(2S)) and N-sulfated glucosamine (GlcNS) (22). For receptor signaling, a dodecasaccharide containing this binding region as well as 6-O-sulfated GlcNS residues is also required (23). The role of heparin/HS in enhancing FGF-2 binding and signaling remains controversial. Some studies (22,24) suggest that heparin/HS interacts with both FGF-2 and FGFR, whereas others (25)(26)(27) suggest that heparin/HS induces FGF-2 dimerization and increases the affinity of the complex for FGFRs. In either case heparin/HS is believed to facilitate FGFR dimerization and subsequent activation.
The ability to stimulate FGF-2 activity relies on the primary sequence, structure, and organization of HS, which is dependent on cell type and differentiation state (16,28). This has been shown conclusively using HS from different cell and tissue types (29,30); however, little work has emerged on the individual contributions to growth factor binding and signaling by specific HSPGs derived from different cell types. We have demonstrated previously (2) that perlecan HS chains isolated from arterial and venous endothelial cells differed in their ability to bind FGF-2 and to adhere to vascular cells. In this study we * 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.
* have investigated the HS chains of perlecans isolated from three different cell types using binding assays, the BaF3 cell system, and HS structural analysis techniques. We report that the source of perlecan significantly influences its HS substructure and subsequently its ability to bind FGF-2 and to promote FGF-2 signaling through FGFRs.

Cell Culture
The spontaneously transformed human umbilical venous endothelial cell line, C11 STH, was provided by Dr. J. Gamble of the Hanson Cancer Research Center, Adelaide, South Australia (31). Human umbilical arterial endothelial cells (HUAEC) and C11 STH were cultured as described (6

Perlecan Isolation and Characterization
Human perlecan was immunopurified from conditioned medium, characterized using enzyme-linked immunosorbent assay, and checked for 125 I-FGF-2 binding as described by Whitelock et al. (2).

Immunoprecipitation of Perlecans Bound to 125 I-FGF-2
Perlecan was immunoprecipitated essentially as described by Whitelock et al. (6) with the following modifications. Purified perlecan (1 g/ml, 100 l) was incubated with 125 I-FGF-2 (1 ϫ 10 5 cpm) for 1 h at RT. After washing with PBS, 100-l aliquots of rabbit anti-mouse antibody conjugated to protein A-Sepharose were incubated with either mAbA76 (20 g/tube), or perlecan⅐FGF-2 samples for 2 h. Perlecan⅐FGF-2 samples were then added to pre-washed A76-protein A-Sepharose beads and incubated at RT for 2 h. The beads were washed extensively with PBS and bound counts/min determined on an automated gamma counter (Wallac, Finland).

Biomolecular Interaction Analysis of Perlecans Using the BIAcore
Perlecan was biotinylated essentially as described by Cole et al. (32). A 100 g/ml solution of perlecan in 0.1 M NaHCO 3 , pH 8.0, was incubated with a 50-fold molar excess of N-hydroxysuccinimide-biotin for 3 h at RT. Biotinylation was stopped by the addition of 2 M NH 4 Cl, and excess biotin was removed by dialysis against PBS for 48 h. Ligand binding experiments were performed using a BIAcore 2000 instrument (Amersham Biosciences). The BIAcore technique has been described previously (33). Similar amounts of biotinylated perlecans (10 g/ml) in PBS were coupled to different flow cells of a streptavidin-derivatized sensor chip at a flow rate of 5 l/min, as determined by the change in response units. Binding experiments were performed at a flow rate of 20 l/min at 25°C. The injected volume of analyte was 55 l, and the kinject function was used with a programmed dissociation time of 150 s. For titration of FGF-2 against bound perlecans, stock solutions of FGF-2 were serially diluted in HBS-EP (0.01 M HEPES, 0.15 M NaCl, 3 mM EDTA, 0.005% (v/v) polysorbate 20, pH 7.4) running buffer to concentrations in the interval 6.25-400 nM. Competitive binding experiments were performed on 100 nM FGF-2 incubated with or without 100 mM NaCl or 0.1 g/ml heparin. FGFR1␣-IIIc and FGFR3␣-IIIc fusion proteins (50 nM) were incubated with or without 50 nM FGF-2 in 0.01 M HEPES, 0.15 M NaCl, 5 mM MgCl 2 , 0.01% Tween 20, pH 7.4, running buffer for 5 min prior to injection. The perlecan surface was regenerated with a 30-s pulse of 1 M NaCl or 100 g/ml heparin. Sensograms were analyzed using the BIAcore 2000 Evaluation Software 3.0. Sensograms were fitted with separate differential rate equations for the parts of the curve representing association and dissociation. The closeness of fit for each kinetic parameter is described by the statistical value 2 . Enzymatic Depolymerization of Perlecan HS-HS chains (1 ϫ 10 4 cpm) were depolymerized with heparinase I or heparinase III at a concentration of 0.025 units/ml in PBS. Samples were incubated for 16 h at 37°C, and the reaction was stopped by heating at 100°C for 5 min.

HS-Oligosaccharide Preparation
Deaminative Scission with Nitrous Acid-Cleavage of intact perlecan HS chains by HNO 2 treatment at low pH (pH 1.5) was carried out by the method of Shively and Conrad (34). Briefly, equal volumes of 1 M Ba(NO 2 ) 2 and 1 M H 2 SO 4 were mixed and centrifuged for 5 min at 10,000 rpm to remove precipitated BaSO 4 . The resulting HNO 2 (100 l) was added to lyophilized [ 3 H]perlecan HS samples (1 ϫ 10 4 cpm) and incubated for 30 min at RT. Reaction was stopped by the addition of 2 M Na 2 CO 3 (10 l).

Gel Filtration Chromatography of Intact and Depolymerized HS Samples
Superose 6 Size Exclusion Fast Protein Liquid Chromatography-Superose 6 gel filtration was performed essentially as described by Melrose and Ghosh (35). A Superose TM 6 pre-packed HR 10/30 (Amersham Biosciences) column was equilibrated with 0.5 M CH 3 COONa, 0.05% Tween 20, pH 7.5, at 0.4 ml/min. Whole perlecan HS and heparinase I/III-cleaved chains (1 ϫ 10 4 cpm; 200 l) were injected onto the column and 2-min (0.8 ml) fractions collected. Aliquots of each fraction were measured in a Packard scintillation counter to generate an elution profile.
Bio-Gel P-10 Gel Permeation Chromatography-Separation of heparinase III and nitrous acid cleavage-resistant oligosaccharides was performed on a Bio-Gel P-10 column (120 ϫ 1 cm) equilibrated with 0.1 M NaCl, pH 8. Heparinase III-or HNO 2 -cleaved HS chains (1 ϫ 10 4 cpm; 200 l) were loaded onto the column and eluted at 4 ml/h. 1-ml fractions were collected, and radioactivity was determined using liquid scintillation. The percentage of susceptible linkages in the size groups corresponding to degree of polymerization dp2-dp12 (dp ϭ degree of polymerization, or number of saccharide units, e.g. dp2 ϭ disaccharide) resolved using Bio-Gel P-10 is given by the formula An/n, where An is the percentage of the total 3 H counts eluting in a specific peak, and n is the number of disaccharide repeat units in the oligosaccharide corresponding to that peak.
Disaccharide Analysis-Perlecan [ 3 H]HS chains (5 ϫ 10 4 cpm) were digested with a combination of heparinases I-III (0.1 unit/ml, 16 h at 37°C). Disaccharides (50 -90% of total HS counts) were recovered from a Bio-Gel P-2 column (120 ϫ 1 cm) eluted at 4 ml/h in 0.1 M NH 4 HCO 3 with collection of 1-ml fractions. Disaccharides were analyzed by strong anion-exchange high pressure liquid chromatography on a ProPac PA-1 column eluted with a linear gradient of 0 -1 M NaCl in MilliQ water, pH 3.5, at a flow rate of 1 ml/min (36). Fractions of 0.5 ml were counted, and disaccharides were identified by comparison of elution positions with known standards.
Bioactivity of Perlecans with FGF-2-The ability of the different perlecans to stimulate a heparin-dependent biological response to FGF-2 was examined in HSPG-deficient myeloid cell lines (BaF3) expressing splice variants of the FGFR1 isoform (FGFR1b and -c) and of the FGFR3 isoform (FGFR3c) (20). FGFR1b/c and FGFR3c cell lines were maintained in RPMI 1640 medium, supplemented with 10% FCS, 10% conditioned media from WEHI-3BD Ϫ cells, G418 (400 g/ml), and penicillin/streptomycin. Mitogenic assays were performed essentially as described by Ornitz et al. (20). Briefly, cells were washed, resuspended in RPMI 1640 medium containing 10% FCS, and seeded into 96-well plates at 6 ϫ 10 5 cells/50 l/well. The volume in the wells was made up to 100 l with RPMI 1640 medium containing final concentrations of 2 g/ml heparin or 1.25 g/ml perlecan and 5 nM FGF-2. For the free HS chain assay, whole HS chains or HS chains digested with heparinase I or heparinase III were added to a final concentration of 5 g/ml. Plates were then incubated at 37°C for 40 h. Cell proliferation was assayed by [ 3 H]thymidine uptake with 0.5 Ci of [ 3 H]thymidine in 20 l of media being added to each well and incubated for 6 h. Cells were pelleted in the wells by centrifugation at 1000 rpm for 5 min and the supernatant flicked into waste. Cells were then washed three times by the addition of 200 l of PBS per well and centrifuged followed by flicking to waste. After the final wash, cells were resuspended in 100 l of PBS, vortexed in 2 ml of scintillation fluid, and counted on an automated scintillation liquid analyzer (Packard Instrument Co.).

Statistical Analyses
Significant differences were determined using analysis of variance Student-Newman-Keuls tests. Unless otherwise stated, results from all static binding studies and BaF3 experiments are expressed as the means Ϯ S.E. of three and four observations, respectively. All experiments were performed at least twice.

Interaction of Perlecans with FGF-2
Perlecans derived from either HUAEC, C11 STH, or WiDr were assessed for their ability to bind soluble FGF-2 after coating wells of a 96-well microtiter plate with perlecan ( Fig.  1A) or by immunoprecipitation of the complex in conditioned medium with anti-perlecan antibodies (Fig. 1B). In both assays, WiDr perlecan bound significantly more 125 I-FGF-2 than either HUAEC or C11 STH perlecan (Fig. 1A, p Ͻ 0.05; Fig. 1B, p Ͻ 0.001), indicating that WiDr perlecan-HS possessed a greater number of FGF-2-binding sites than either HUAEC or C11 STH perlecan-HS. Perlecan derived from either endothelial cell line bound similar amounts of 125 I-FGF-2 with C11 STH perlecan binding a little more when coated onto microtiter wells (Fig. 1A). This difference was not statistically significant (p Ͼ 0.05).
BIAcore was used to assess binding of FGF-2 to immobilized perlecans under flow conditions. Fig. 2 shows a composite sensorgram of FGF-2 binding the three different perlecans. Fig. 2, curve a, shows the binding to HUAEC perlecan, and curve b indicates the binding to C11 STH perlecan, and curve c shows the binding to WiDr perlecan. Both HUAEC and C11 STH perlecans bound FGF-2 very efficiently and to a similar extent. In contrast to the static assays, immobilized WiDr perlecan exhibited very little binding of FGF-2, suggesting immobilization of WiDr to the BIAcore chip has a negative effect on FGF-2 binding, possibly due to steric constraints. The addition of 5 mM MgCl 2 to the binding buffer had no effect on binding (data not shown). The amount of binding demonstrated to WiDr perlecan was similar to that seen to the protein core of either C11 STH or HUAEC perlecan (Fig. 3, A and B). To rule out the possibility that the conjugation chemistry had removed the HS from WiDr perlecan, we confirmed the presence of HS attached to WiDr perlecan before and after the biotinylation procedure by probing with the anti-HS antibody, 10E4 (data not shown). The interaction of FGF-2 with HUAEC ( Fig. 3A) and C11 STH (Fig.  3B) perlecan was shown to be HS-dependent with binding reduced by either the addition of salt (Fig. 3, curves b) or heparin (Fig. 3, curves c) to the binding buffer. There was also limited binding to the protein core of both the perlecans (Fig. 3, curves d) most likely due to HS remaining after heparinase III digestion (see Fig. 7).
Experiments using decreasing concentrations of FGF-2 revealed two association events. An initial reaction that resulted in a rapid rise in response units followed by a secondary bind-ing that was demonstrated by the shallower gradient of the binding curve. To assess what might be causing each of the binding phases, we analyzed the binding of FGF-2 in the presence of salt (Fig. 3, curves b) or heparin (Fig. 3, curves c). The initial fast interaction was decreased quite effectively with either salt or heparin competition, indicating that this phase of binding between the HS and FGF-2 was ionic. The secondary binding phase was also reduced in the presence of heparin but increased in the presence of salt. The HS-FGF-2 interaction was assumed to be 1:1, and kinetic constants were fitted separately to the endothelial perlecan sensorgrams at FGF-2 concentrations that did not exhibit the secondary binding phase (Ͻ50 nM). HUAEC perlecan gave a k a Ϸ 1.0 ϫ 10 6 , k d Ϸ2.0 ϫ 10 Ϫ3 , and K d Ϸ2 nM ( 2 Ϸ2.9). This was similar to the binding constants calculated for C11 STH perlecan, which gave a k a Ϸ 1.4 ϫ 10 6 , k d Ϸ 1.7 ϫ 10 Ϫ3 , and K d Ϸ 1 nM ( 2 Ϸ1.8).
The binding of FGF-2 to C11 STH perlecan (Fig. 3B, curve b) was less affected by salt than FGF-2 binding to HUAEC (Fig.  3A, curve b), although this difference in affinity was not reflected in the kinetic data determined from Fig. 2. In the presence of heparin, both HUAEC (Fig. 3A, curve c) and C11 STH perlecan (Fig. 3B, curve c) bound similar amounts of FGF-2 (approximately 150 response units).

Interaction of Perlecans with FGF-2 and FGF Receptors
In receptor binding experiments, FGFR1␣-IIIc did not bind to either perlecan (Fig. 4, A and B, curves e), whereas FGFR3␣- IIIc bound to both perlecans (Fig. 4, A and B, curve d). FGFR3␣-IIIc bound more to C11 STH perlecan (Fig. 4B, curve d) than to HUAEC (Fig. 4A, curve d). The binding of FGF-2 alone is shown by curves c. FGF-2 was complexed to either FGFR1␣-IIIc or FGFR3␣-IIIc and passed over the immobilized endothelial perlecans. Complexes of the growth factor with FGFR1␣-IIIc (Fig.   4, A and B, curves b) or FGFR3␣-IIIc (Fig. 4, A and B, curves a) bound to each of the endothelial derived perlecans. All of these binding events could be inhibited by the presence of either heparin or salt, and there was limited binding to the protein core, suggesting that they were HS-dependent (data not shown).

Characterization of HS from Each Perlecan Type Gel Filtration Chromatography of Intact and Depolymerized
Perlecan HS-The relative sizes of the HS chains from each of the affinity-purified perlecans were estimated by chromatography using Superose 6 fast protein liquid chromatography (35), Fig. 5. The HS isolated from HUAEC, C11 STH, and WiDr perlecans had similar sizes, which were determined to be ϳ40 kDa using chondroitin sulfate (CS) standards (35).
Enzymatic Cleavage-Cleavage of these chains with heparinases was also monitored using Superose 6. After incubation with heparinase I, the elution positions of the HS chains shifted to the right (Fig. 5B) indicating that each perlecan HS contained heparinase I-susceptible linkages (GlcNS(Ϯ6S)␣1-4IdoUA(2S)) (37). WiDr perlecan HS demonstrated a greater shift in elution volume in this system and had a broader peak compared with the HS samples derived from HUAEC or C11 STH perlecan, suggesting it contained more of these sequences. When the samples were digested with heparinase III (Fig. 5C), which cleaves GlcNR(Ϯ6S)␣1-4GlcA/IdoUA, where R ϭ Ac or S (38,39), all three HS species eluted in the total volume of the column indicating that each contained heparinase III-susceptible sequences.
The HS chains and their cleavage products were examined further on a Bio-Gel P-10 gel filtration column. Heparinase III-digested HS derived from HUAEC perlecan (Fig. 6A) and C11 STH perlecan (Fig. 6B) into fragments ranging from disaccharides (dp2) through decasaccharides (dp10) and larger (total void, V o ). The major cleavage products of both HUAEC and C11 STH perlecan HS were disaccharides (71% for HUAEC and 49% for C11 STH; see Table I), indicating that the less sulfated domains were mainly contiguous in both HS types. The C11 STH perlecan HS overall was less susceptible to heparinase III cleavage than HUAEC perlecan HS containing 18% more of the dp4 -dp12 heparinase-resistant sulfated domains (Table I). HS from WiDr-derived perlecan, however, was extremely resistant to heparinase III digestion with only 2% of the radioactivity being present in the disaccharide peak (dp2), 46% present in the dp4 -dp12 fractions, and the largest proportion, 52%, present in the dp12-V o fractions (Table I). Reapplication of the isolated WiDr V o peak from the P-10 to a Superose 6 column demonstrated that digestion had in fact occurred, with the material eluting in the V t . These data indicated the presence of a much greater proportion of heparinase III-resistant sites in the WiDr HS than seen in the HUAEC or C11 STH perlecan HS and that these were arranged in longer domains, supporting our initial hypothesis that WiDr HS had an higher degree of sulfation (Fig. 5B).
Nitrous Acid Cleavage-The HS chains of each of the respective perlecans were subjected to depolymerization using low pH nitrous acid which cleaves at GlcNS(Ϯ6S)␣1-4 hexuronic acid (Ϯ2S) residues (34). Nitrous acid cleaved all three types of HS as shown in Fig. 6, D-F. The three profiles were distinct with the most degradation occurring with WiDr, followed by C11 STH and then HUAEC perlecan HS. Although the dp2 peak was higher in the WiDr sample (Fig. 6F) and the dp4 peak higher in the other two samples (Fig. 6, D and E), the actual percentages of these oligosaccharides, determined by peak area (Table I), were very similar due to variations in peak width. The dp12-V o fractions of HUAEC, however, did contain a higher proportion of radioactivity, supporting the hypothesis that this HS contains a lower degree of sulfation compared with the other HS types investigated. When the total percentage of N-sulfation was calculated for each of the HS types, WiDr had the highest degree of N-sulfation at 42%, followed by 31% for C11 STH and then HUAEC at 27% (Table I). These data support that obtained in the heparinase III digestion experiments (Fig. 6, A-C).

SAX-High Pressure Liquid Chromatography Compositional Analysis
Combined heparinase I-III digestion of both HUAEC and C11 STH perlecan HS was performed to obtain data on their composition. WiDr HS was not investigated for compositional analysis due to the fact that combined heparinase treatment did not digest the material to disaccharides. The results from these experiments are summarized in Table II. The percentage of N-sulfation calculated using these data agrees very closely with that obtained using nitrous acid digestion (see Table I). C11 STH HS had a slight decrease in unsulfated disaccharides compared with HUAEC HS, which was accounted for by a similar increase in N-sulfated disaccharides. The proportion of 2-O-sulfation in the endothelial derived perlecans was very similar (6.2% for HUAEC and 6.6% for C11 STH), supporting data presented earlier that they also bound similar amounts of FGF-2 ( Figs. 1 and 2).

Bioactivity of FGF-2 in the Presence of the Perlecans
The ability of the immunopurified perlecans to promote cellular proliferation in response to exogenous FGF-2 was investigated using BaF3 cell lines that were either expressing FGFR1b, FGFR1c, or FGFR3c receptor isotypes. Both the HUAEC and C11 STH perlecan stimulated the proliferation of the FGFR1b (Fig. 7A) and FGFR1c (Fig. 7B) expressing cells in response to FGF-2, whereas only the HUAEC perlecan stimulated the proliferation of the FGFR3c-expressing cells (Fig. 7C). Interestingly, under the experimental conditions used, WiDr perlecan failed to stimulate any of the three cell lines (Fig. 7,  A-C). The same results were obtained for the three perlecans when they were coated onto the surface of the tissue culture wells (data not shown).
It was of interest to determine what effect, if any, was achieved when the HS chains were removed from their protein cores, and whether heparinase digestion altered the bioactivity. HS from HUAEC perlecan had the same amount of proliferative capacity as the whole proteoglycan, which was markedly reduced when it was incubated with heparinase I (Fig. 8). Heparinase III incubation of HUAEC HS had no effect on its biological activity. HS from C11 STH perlecan had slightly

TABLE I Percentage of heparinase III susceptible linkages and N-sulfation of HS side chains of immunopurified perlecans
The oligosaccharides obtained after cleavage with heparinase III or nitrous acid at pH 1.5 (see "Experimental Procedures") were separated by gel chromatography on Bio-Gel P-10 ( Fig. 6, A-F  more activity than whole C11 STH perlecan. However, heparinase I had no effect on the ability of this HS to potentiate FGF-2 signaling via the FGFR1 receptor (Fig. 8). Heparinase III also had no effect on C11 STH perlecan HS. Interestingly, when the HS was released from the WiDr perlecan core, it became active (to roughly half the activity of heparin). This activity was not affected by heparinase I digestion and, unex-pectedly, was increased when the HS was digested with heparinase III (Fig. 8). Similar results to these were obtained in activity assays using BaF3 cells that expressed the IIIc isoform of FGFR1 (data not shown).

DISCUSSION
This paper describes the biological and biochemical differences seen in the HS chains of perlecan immunopurified from two endothelial cell lines and one colon carcinoma cell line. It is clear from this study that not only does the structure and bioactivity of perlecan HS vary depending on cell source but that some forms of perlecan carry HS chains that bind FGF-2 but do not possess biological activity unless they are released from the protein core.
Interactions of FGF-2 and FGF-2 Receptors with Perlecans-The binding affinity of FGF-2 for C11 STH perlecan, observed here using the BIAcore, was twice that calculated for HUAEC perlecan, which is consistent with our previous study (2). Interestingly, the affinity of endothelial perlecans for FGF-2 (1-2 nM) was similar to that observed (using a BIAcore) for immobilized agrin (2.5 nM) (40), as well as that observed (using binding of 125 I-FGF-2) to ryudocan (0.5 nM) (41). However, it should be noted that binding for each of these interactions was assumed to have a stoichiometry of 1:1, whereas HS is likely to possess more than one interactive site. When WiDr perlecan was bound to a microtiter plate, it possessed significantly more FGF-2-binding sites than either of the endothelial perlecans. It has been suggested previously that carcinomas produce HSPGs that can out-compete normal tissue for essential growth factors such as FGF-2, resulting in the formation of new blood vessels supporting tumor growth and progression (42,43). However the FGF-2 binding ability of WiDr perlecan HS was diminished when it was modified by biotin and coupled to the BIAcore surface, highlighting the necessity to use a variety of techniques to assess the bioactivity of these molecules.
In this study, we have shown that endothelial derived perlecan HS binds to FGFR3␣-IIIc but not FGFR1␣-IIIc in the absence of FGF. Previously, heparin has been shown to bind to FGFR1 and FGFR2 through a site located in the second immunoglobulin-like domain of the receptor (24,44,45), which has led investigators to hypothesize that HS from different HSPGs regulates the activity of each of the FGFs in different ways (46). This has been supported by recent data where glypican-1 bound to FGF-1 but not to FGFR2 (KGFR) (47) and to syndecan-1, which was shown to interact with FGFR1 (48). HS oligosaccharides have been shown to either activate or inhibit cell growth depending on which FGF was used in combination with which FGFR isotype (30,49,50). Because both HUAEC and C11 STH perlecans bind FGF-2-FGFR1c and FGF-2-FGFR3c complexes with virtually the same affinity, but only HUAEC perlecan stimulates FGF-2-FGFR3c signaling, then C11 STH HS contains sequences that do not form efficient FGF-2-FGFR3 signaling complexes. However, it should be noted that differences between the results obtained using the BaF3 cells and those obtained using the BIAcore could be due to the fact that cell-based receptors are likely to be monomeric, whereas the recombinant FGF receptor fusion proteins used in this study are dimeric.
HS Substructure-The differences in growth factor binding observed in this study were due to different HS structures within the HS chains, as the HS from each of the perlecans were similar in size. The chain lengths were comparable with that of perlecanenriched HS fractions isolated from normal and transformed epithelial cells (51). Depolymerization of the HS from the endothelial derived perlecan by enzymatic and by chemical scission suggested that they were predominantly N-acetylated. This was supported by the compositional data, which showed that the predominant disaccharide in both types of HS was GlcNAc ␣1,4uronic acid. The high proportion of this disaccharide, together with the low frequency of GlcNS(Ϯ6S)␣1,4 IdoUA2S disaccharides evident from our data are all consistent with previous reports (23,52) on endothelial derived HS. The depolymerized HUAEC and C11 STH perlecan HS samples also exhibited typical elution profiles as seen for endothelial cell-derived HS chains cleaved with nitrous acid or heparinase III; relatively short domains of N-sulfated disaccharides were separated by extended sequences of predominantly N-acetylated disaccharides (52). Collectively, our data suggested that C11 STH perlecan HS contained more and larger N-sulfated domains than the HUAEC HS. It has been suggested by previous investigators (23) that the presence of correctly positioned 6-O-sulfated GlcNS residues adjacent to classic FGF-2-binding sequences (IdoUA(2S)-GlcNS repeat regions) may control the ability of the HS to potentiate FGF-2 signaling via its receptor. Different FGFR-FGF combinations may have different specific requirements for the positioning of such residues, and these differences may explain why C11 STH did not signal through the FGFR3 receptor.
Surprisingly, extensive regions of the HS from WiDr perlecan were resistant to heparinase III digestion. This is unusual for HS isolated from a matrix HSPG and is something that has not been reported previously. The level of N-sulfation observed in this study was greater than the two endothelial derived perlecans and was similar to that reported for perlecan-enriched fractions isolated from other human carcinoma cell lines (53) suggesting that this may be a tumor-specific phenomenon. As both 2-O-and 6-O-sulfations occur predominantly within extended N-sulfated regions of HS chains (54), it may be possible that WiDr HS also has increased levels of O-sulfation. This is supported by our findings that coated WiDr perlecan binds more FGF-2 and is more susceptible to heparinase I activity. However, it should be noted that structural variation in HS between cell types does not always correspond to differences in binding FGF-2, as was observed previously (55) for syndecan-1. Also, it should be noted that the suggested differences in sulfate content in WiDr HS need to be confirmed by the use of an alternative approach to disaccharide analysis, which might include chemical N-deacetylation followed by both low and high pH nitrous acid treatment.
FGF-2 Mitogenesis-FGFR1/3 "b" and "c" are two of three splice variants that differ in the amino acid sequence in the third IgG domain conferring on them different ligand-binding properties (56). In FGFR-expressing BaF3 cells, responses to FGF are dependent on the addition of exogenous activating compounds such as heparin. Perlecan immunopurified from fibroblasts has been shown previously to stimulate cell signaling of FGFR1c cells in the presence of FGF-2 (16). In our study, we also show that perlecan from two endothelial cells stimulated FGF-2 signaling of the FGFR1c in BaF3 cell lines, whereas only the HUAEC-derived perlecan stimulated the growth of FGFR3c-expressing BaF3 cells. When activity was detected in the BaF3 cell assays, it was found to be present in heparinase III-resistant segments of the HS chains, which is consistent with previous studies (30) on HS chain structure. In contrast, we found that treatment with heparinase I, which cleaves the HS chain at putative FGF-2-binding sites, does not always result in reduced bioactivity of the digested HS chain. HUAEC lost activity after heparinase I digestion, but there was no reduction in the bioactivity of HS chains from C11 STH perlecan following heparinase I digestion, suggesting that there may be other FGF-2 signaling sequences within these HS chains.
WiDr-derived perlecan did not stimulate growth of any of the BaF3 cell lines, even though it was very efficient at binding FGF-2. However, once the HS was released from the protein core it became active, suggesting that the presence of the protein core was inhibitory. Local formation of polysaccharide secondary structure has been proposed to govern the binding and catalytic interactions between proteins and glycosaminoglycans (57), but surprisingly, there has been little evaluation on the effect that protein core-HS interactions might have on HS ligands. The WiDr results are similar to those reported previously for syndecan-1 where the proteolytically released ectodomains potently inhibited FGF-2-mediated mitogenesis until heparinase III or heparinases from platelets cleaved the HS chains to release fragments capable of potentiating the action of FGF-2 (58,59). Another interpretation of the data could be that the HS chains attached to the protein core interacted with each other in a fashion that inhibited their bioactivity. HS chains have long been known to possess the ability to interact with each other (60,61). These HS-HS or protein-HS interactions within HSPGs may therefore provide a physiological mechanism whereby the activities of FGFs in wound healing and tumorigenesis are modulated. FIG. 8. Bioactivity of whole perlecan and perlecan HS ؎ heparinases in the presence of FGF-2 in FGFR1bexpressing cells. Cells were washed, resuspended in RPMI 1640 medium containing 10% FCS, and seeded into 96-well plates at 6 ϫ 10 5 cells/50 l/well. The volume in the wells was made up to 100 l with RPMI medium containing final concentrations of 2 g/ml heparin, 1.25 g/ml perlecan, or 5 g/ml HS chains Ϯ heparinase I/III. All wells contained 5 nM FGF-2 unless stated otherwise. Plates were incubated at 37°C for 40 h and assayed as described in the legend of Fig. 7. The result is representative of two separate experiments. hep, heparin; ϩ, FGF-2; H, HUAEC; C, C11 STH; W, WiDr; HS, heparan sulfate chains; hepI, heparinase I-treated; hepIII, heparinase III-treated.
In summary, this work has demonstrated that perlecans isolated from different cell types differ significantly in their ability to interact with FGF-2 and FGF receptors, which in turn allows differential regulation of FGF-2-mediated cell signaling. HS structural analysis suggests that these activities are mediated by differences in the HS sequences of the various perlecans. It is also interesting to find that HS-HS and HS-protein interactions may influence HSPG growth factor activities and that FGF-2 mitogenic sequences other than those already identified may exist in native HS. If we are to understand how perlecan and other types of HSPGs affect the signaling of the different growth factor receptors, it is essential to identify the exact HS sequences that are responsible for the differential binding.