Sulfated Derivatives of Escherichia coli K5 Polysaccharides as Modulators of Fibroblast Growth Factor Signaling*

Heparan sulfate (HS) proteoglycans are intimately involved in the regulation of fibroblast growth factor (FGF) signaling. HS and the related glycosaminoglycan heparin interact with FGFs and FGF receptors (FGFRs), and it is believed that both interactions are required for productive FGF signaling. Attempts to inhibit FGF activity have been made with modified heparin preparations, various heparin-like polysaccharide analogues and other polyanionic molecules, which may all act by interfering with the physiological HS-FGF-FGFR interactions on the cell surface. Here, we have studied the potential of sulfated derivatives of a bacterial polysaccharide (capsular polysaccharide from Escherichia coli K5 (K5PS)) in the modulation of FGF-heparin/HS interactions and FGF signaling. We demonstrate that O-sulfated and N,O-sulfated species of K5PS, with high degrees of sulfation, displaced FGF-1, FGF-2, and FGF-8b from heparin. However, only O-sulfated K5PS efficiently inhibited the FGF-induced proliferation of S115 mammary carcinoma cells and 3T3 fibroblasts, whereas N,O-sulfated K5PS had little or no inhibitory effect. Studies with CHO677 cells lacking endogenous HS, as well as with chlorate-treated S115 cells expressing undersulfated HS, indicated that whereas exogenously administered heparin and N,O-sulfated K5PS restored the cellular response toward FGF stimulation, O-sulfated K5PS was largely devoid of such stimulatory activity. Our data suggest that highly O-sulfated species of K5PS may be efficient inhibitors of FGF signaling.

Fibroblast growth factors (FGFs) 1 are involved in the proliferation, migration, and differentiation of many cell types. FGFs have been implicated in physiological processes such as embryonic development (1) and tissue regeneration but also in a number of pathological conditions including cancer, inflammatory diseases, fibrosis, etc. (2). FGF signals are thought to stimulate tumor cell proliferation in an autocrine or paracrine fashion and various types of malignant tumors have been shown to overexpress FGFs and FGF receptors (FGFRs) (2,3). FGFs are also potent angiogenic factors that promote tumor neovascularization by stimulating the proliferation and migration of endothelial cells (4). Conceivably, development of agents that antagonize FGF signaling might provide means to influence multiple aspects of tumor progression.
FGFs exert their biological activities by binding to FGFRs that leads to FGFR dimerization, phosphorylation of the cytoplasmic receptor domains, and activation of intracellular signal transduction pathways (5). FGF activity is critically dependent on the sulfated glycosaminoglycan heparan sulfate (HS) (6, 7), a component of HS proteoglycans (HSPGs) found on cell surfaces and in the extracellular matrix (8,9). HS has been demonstrated to interact with FGFs as well as with the extracellular domains of FGFRs (10). HS likely plays a critical role in the formation of "ternary complexes" consisting of FGF, FGFR, and HS, in which HS concomitantly interacts with both the growth factor and receptor components of the complex (11,12). Cells lacking endogenous HS thus respond poorly to FGFs, whereas the response can be readily restored by addition of heparin (6,7,(13)(14)(15). By contrast, in cells which express endogenous HSPGs, exogenously administered heparin may inhibit rather than stimulate FGF signaling, presumably by perturbing the interactions of cellular HSPGs with FGFs/FGFRs (13)(14)(15).
In this paper, we have investigated the potential of semisynthetic, heparin-like polysaccharides as inhibitors of FGF signaling. The capsular polysaccharide of Escherichia coli K5 bacteria (hereafter denoted as K5 polysaccharide or K5PS) has the same structure as the primary polymerization product of heparin/HS biosynthesis, consisting of alternating D-glucuronic acid and D-N-acetylglucosamine units (GlcA␣1,4GlcNAc) n (16). K5PS can be subjected to various degrees of chemical N-deacetylation/N-sulfation as well as O-sulfation (17,18), to yield different heparin/HS-like semisynthetic polysaccharides (Fig. 1). Highly sulfated species of K5PS have been shown to bind to FGF-2 and inhibit FGF-2 induced endothelial cell proliferation, probably by interfering with the formation of FGF-2⅐FGFR⅐HS complexes (18). Interestingly, the N,O-sulfated polysaccharide species have also displayed anti-angiogenic activity in chick chorioallontoic membrane assays (18) (see "Discussion"). The results presented in this paper provide novel data with regard to the properties of sulfated K5PSs as FGF inhibitors. We have assessed the previously described series (18) of chemically sulfated derivatives of K5PS for their capability to interact with FGFs and modulate FGF-induced cellular responses. Our data suggest that whereas various species of sulfated K5PS bind FGF-1, FGF-2 and FGF-8b, only a highly O-sulfated preparation was an efficient inhibitor of the biological activity of FGFs at low (Յ1 g/ml) concentrations. Experiments in cells lacking endogenous active HS suggest that the differential inhibitory capacity of the preparations derive from differences in their agonist/antagonist properties with regard to FGF signaling. Thus, compounds such as O-sulfated K5PS deserve to be further explored as inhibitors of FGF-dependent tumor cell proliferation.

EXPERIMENTAL PROCEDURES
Materials-Cell culture media, chemicals for the preparation of low sulfate DMEM (LSDMEM) (6), testosterone, and the monoclonal antibodies against ERKs were from Sigma. Sodium chlorate was purchased from Merck, and [methyl-3 H]thymidine was from PerkinElmer Life Sciences. FGF-1 and FGF-2 were from Peprotech. The expression vector for FGF-8b was kindly provided by Dr. Graig MacArthur (Washington University). FGF-8b was produced and purified as described elsewhere (19,20).
The K5PS preparations used in this study were kindly provided by Drs. Giorgio Zoppetti and Pasqua Oreste. The synthesis and characterization of the K5PSs have been described elsewhere (18). Briefly, a series of sulfated K5PSs has been generated by chemical N-deacetylation/N-sulfation and/or O-sulfation of the unmodified K5PS structure (GlcA-GlcNAc) n (18). The modified compounds included fully N-sulfated K5PS without (K5-NS) or with (K5-NSOS) O-sulfation as well as Osulfated K5PS (K5-OS). The O-sulfation of K5-NSOS was somewhat lower than that of K5-OS, such that their overall degrees of sulfation were similar (3.84 versus 3.77 sulfate groups/disaccharide unit, respectively). However, in both N-sulfated and N-acetylated preparations the degree of GlcN 6-O-sulfation was nearly 100%, such that the differences in O-sulfation were found in the 2-O-and 3-O-sulfation of the GlcA units and in the 3-O-sulfation of the GlcN residues. The mean molecular masses of the preparations were in the range of 10 -15 kDa.
Cell Cultures-S115 mouse mammary carcinoma cells (21) were maintained in DMEM supplemented with 5% heat-inactivated fetal calf serum (FCS), 1 mM sodium pyruvate, 1 mM L-glutamine, 100 IU/ml penicillin, 100 g/ml streptomycin, and 10 nM testosterone. For experiments, the cells were grown in medium containing 4% dextran-coated charcoal-treated FCS (DCC-FCS) (22) instead of FCS, followed by a 24-h serum starvation in a 1:1 mixture of serum-free Ham's F-12 and DMEM (23) and treatment with testosterone/FGF and K5PSs in various combinations, as specified in the figure legends. To assess the cellular DNA synthesis, [methyl-3 H]thymidine (2 Ci/ml) was added to the culture medium for the last 2 h of the incubation. Subsequently, the cells were washed and solubilized into 1 M NaOH followed by quantification of the incorporated radioactivity in a beta counter (Wallac, Turku, Finland). Alternatively, cell growth was assessed by staining the cells with crystal violet (24) after various periods of incubation. In some experiments, the added saccharide was withdrawn after a 24-h incubation by change of the culture medium, after which the cells were allowed to grow for 48 h before crystal violet staining.
The treatment of S115 cells with sodium chlorate has been described elsewhere (25). Briefly, cells were cultured in LSDMEM containing 5% DCC-FCS and 30 mM sodium chlorate, followed by trypsinization, replating, and starvation in LSDMEM supplemented with 0.1% DCC-FCS and 30 mM sodium chlorate. Testosterone and K5PSs were added to the cells that were subsequently assessed for [methyl-3 H]thymidine incorporation as described above. HS-deficient Chinese hamster ovary (CHO) cells (line 677; Ref. 26), and the same cells stably transfected with cDNA encoding FGFR-1 or FGFR-4 (20), were cultured in ␣-minimum Eagle's medium containing 5% FCS, whereas mouse NIH-3T3 fibroblasts were cultured in DMEM supplemented with 5% FCS. For the [methyl-3 H]thymidine incorporation experiments, both types of cells were serum-starved and FGF-stimulated according to the protocol described above.
Analysis of ERK Phosphorylation-To analyze the phosphorylation of ERK 1/2, S115 cells were cultured as described above and lysed in Laemmli buffer at the time points of 2, 10, 20, and 60 min followed by fractionation of the lysates on 10% SDS-PAGE. After electrophoresis, the samples were transferred onto nitrocellulose by semidry blotting. Sample loading was studied by protein staining with Ponceau S. The membranes were blocked in PBS (0.15 M NaCl in 20 mM phosphate buffer, pH 7.4) containing 5% nonfat dry milk and 0.4% Tween 20 at room temperature for 1 h followed by incubation with mouse monoclonal antibodies against phosphorylated ERK 1/2 (0.1-0.2 g/ml) overnight at 4°C. After several washes with 0.4% Tween 20 in PBS, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies against mouse immunoglobulins (dilution 1:3000) for 1 h at room temperature. The membranes were washed several times followed by visualization of the bound antibodies using an enhanced chemiluminescence detection kit (Amersham Biosciences). Subsequently, the membranes were soaked in 0.1 M glycine, pH 2.5, and re-probed with monoclonal antibodies against ERK 2.
FGF Binding Assay-To study the ability of sulfated K5PSs to displace [ 3 H]heparin from FGF, [ 3 H]heparin (20,000 cpm), sulfated K5 polysaccharides (0.05-5 g/ml), and FGFs (2.5 g/ml) were incubated in PBS containing 0.1 mg/ml bovine serum albumin at room temperature for 2 h. The reaction mixtures were rapidly passed through PBS-washed nitrocellulose filters (Sartorius, 25-mm diameter, pore size: 0.45 m) using a vacuum suction apparatus followed by two washes of the filters with PBS. Proteins and protein-bound saccharides bind to the filter, whereas unbound saccharides pass through (27). The filter-bound [ 3 H]heparin was released with 2 M NaCl and quantified by scintillation counting.

Inhibition of Testosterone-induced Proliferation of S115 Cells by Sulfated
K5PSs-Stimulation of S115 cells with testosterone leads to a morphological change of the cells from epithelial to fibroblast-like phenotype, increased proliferation, ability to grow in soft agar, and disorganization of the actin filament network (28). The process has been shown to involve testosterone-induced expression of FGF-8b (29,30) and to require the presence of sulfated HSPGs (25). Previous data indicate that such proliferation, as measured by incorporation of [ 3 H]thymidine into cellular DNA, can be inhibited by heparin (25,31). We studied whether sulfated K5PSs were able to influence the testosterone-induced proliferation of S115 cells. The results indicated that K5-OS inhibited the proliferation efficiently ( Fig. 2A), whereas K5-NS, K5-NSOS, or unmodified K5PS had little or no inhibitory activity. To assess the inhibitory capacity of K5-OS in a more detail, a range of polysaccharide concentrations was studied in the same cell proliferation assay and their inhibitory capacities were compared with that of heparin ( Fig. 2B). The results showed that K5-OS caused a dose-dependent growth inhibition at low concentrations (0.1-1 g/ml), whereas ϳ10-fold higher concentrations of K5-NSOS were required for corresponding inhibitory effects. Notably, heparin had only a limited effect on cell proliferation. These data suggest that the structurally distinct preparations of sulfated K5PSs differed markedly in their inhibitory capacity, albeit their overall degrees of sulfation were similar. The results also demonstrate that sulfated K5PSs inhibit the testosterone-induced proliferation of S115 cells much more efficiently than heparin.
We further assessed the effect of K5-OS on cell proliferation by treating S115 cells with K5-OS and testosterone for various periods of time and determining the relative cell numbers by crystal violet staining. Whereas testosterone treatment induced cell proliferation, cells treated with a combination of K5-OS and testosterone displayed less proliferation than control cells without testosterone treatment (Fig. 3A), confirming the idea that K5-OS inhibited testosterone-induced proliferation of S115 cells. Unmodified K5PS did not affect the testosterone response and heparin had only a minor effect, in agreement with the [ 3 H]thymidine incorporation data presented above. To study whether the growth inhibition by K5-OS was reversible, cells were grown in the presence of K5-OS and testosterone for 24 h, after which the medium were changed, and the cells were cultured in the presence of testosterone alone for additional 48 h. Following the withdrawal of K5-OS, increased cell proliferation was observed (Fig. 3B) suggesting that the antiproliferative effect was reversible in nature.
Inhibition of FGF-8b-induced Proliferation and ERK Phosphorylation by K5-OS-The transforming effect of testosterone on S115 cells has been demonstrated to involve up-regulation of FGF-8b expression (29). We proceeded to test the effects of K5PSs on S115 cells specifically stimulated with FGF-8b, which induced an increased DNA synthesis in serum-starved S115 cells (Fig. 4A). The effect of FGF-8b was efficiently inhibited by low concentrations of K5-OS (0.1-1 g/ml), whereas K5-NSOS caused only a partial inhibition (Fig. 4B). The non-O-sulfated species of K5PS did not inhibit the proliferative effect of FGF-8b.
To study the effect of K5-OS on the downstream signaling initiated by FGF stimulation, we analyzed its ability to inhibit FGF-8b induced phosphorylation of ERK, a component of the mitogen-activated protein kinase pathway. Serum-starved S115 cells were treated with FGF-8b and K5-OS, followed by fractionation of cell lysates on SDS-PAGE and Western blotting using antibodies against phosphorylated ERK 1/2. Stimulation with FGF-8b induced rapid ERK phosphorylation, such that the maximal induction was seen 10 min after the start of the FGF treatment (Fig. 4B). Concomitant treatment with K5-OS prevented the ERK phosphorylation completely, whereas heparin or unmodified K5PS did not inhibit ERK phosphorylation.
Collectively, the above data indicate that K5-OS antagonizes FGF-8b signaling in S115 cells and further suggest that the inhibitory effect of K5-OS on the proliferation of testosteronestimulated S115 cells may be mediated through inhibition of FGF-8b activity.
FGF-Antagonist Activity of K5-OS Is Influenced by the FGF Species and Cell Type Involved-Previous data suggest that the proliferation of S115 cells may be stimulated also by FGF-1 or FGF-2 (32), and we thus assessed the effect of K5PSs on the biological activity of these FGF species. The results indicated that whereas both FGF-1 and FGF-2 enhanced the DNA synthesis of S115 cells, their effects were almost completely inhibited by K5-OS at a concentration of 1 g/ml (Fig. 5). By contrast, K5-NSOS appeared to stimulate rather than inhibit the corresponding FGF responses. To assess whether sulfated K5PSs were able to influence the FGF response in other cells than S115, we performed studies with 3T3 mouse fibroblasts that were stimulated with FGF-1, FGF-2, or FGF-8b and treated with K5PSs. Analogously to the results with S115 cells, the proliferative effects of FGF-2 and FGF-8b were inhibited by K5-OS, although the extent of inhibition remained lower than that seen in S115 cells. Notably, the effect of FGF-1 was not influenced by K5-OS in 3T3 cells. Collectively, these results suggest that different cell types may differ in their sensitivity toward the inhibitory effect of K5-OS upon FGF stimulation and further that the inhibitory capacity of the polysaccharide may differ according to the target FGF species.
K5-OS and K5-NSOS Inhibit Binding of FGFs to Heparin-A possible mechanism of how sulfated K5PSs might inhibit FGF signaling involves their binding to the growth factors, hence interference with the cellular HS-FGF-FGFR interactions. We studied the binding of K5PSs to FGF-1, FGF-2, or FGF-8b by assessing their capacity to displace [ 4. Effect of K5PS on FGF-8b induced DNA synthesis and ERK phosphorylation in S115 cells. Serum-starved cells were stimulated with FGF-8b (100 ng/ml) and treated with the various polysaccharides either at a fixed concentration of 1 g/ml (A) or with different concentrations ranging from 0.001 to 100 g/ml (B), followed by measurement of [ 3 H]thymidine uptake. The results show the mean Ϯ S.D. of three incubations. In other experiments, cells grown as in A were lysed followed by fractionation of the lysates on 10% SDS-PAGE and Western blotting with a monoclonal antibody against phosphorylated ERK 1/2 (pERK 1/2) (C). The bound antibodies were detected by anti-mouse secondary antibodies and subsequent enhanced chemiluminescence reaction. The loading of the samples was studied by anti-ERK 2 antibody staining of the same filter.
protein-saccharide complexes on nitrocellulose filters (see "Experimental Procedures"). Unlabeled heparin, K5-OS, and K5-NSOS all competed with [ 3 H]heparin for binding to the various FGFs species in a concentration dependent manner (Fig. 6), whereas unmodified K5PS had no inhibitory capacity. Although the sulfated K5PSs differed somewhat in their inhibitory potency with regard to the polysaccharide structure and also the FGF species involved, both the O-sulfated and N,Osulfated preparations were efficient inhibitors of the [ 3 H]heparin-FGF interactions at the concentration of 1 g/ml used in the cell culture assays. These data suggested that other reasons than differential FGF binding capacity should be explored to satisfactorily explain the different capacities of K5-OS and K5-NSOS in the inhibition of FGF-induced cell growth.

Different Effects of K5-OS and K5-NSOS on FGF Activity in Cells Lacking
Endogenous HS-Given the findings that both K5-OS and K5-NSOS bound avidly to FGFs, whereas only the former preparation inhibited FGF-induced cell proliferation, we proceeded to study whether K5-OS and K5-NSOS might differ in their effects on FGF signaling in cells lacking endogenous HSPGs. We have earlier demonstrated that S115 cells treated with sodium chlorate, an inhibitor of cellular sulfation reactions including the sulfation of HSPGs (6), require exogenous heparin/HS for their proliferation upon testosterone stimulation (25). Accordingly, the DNA synthesis of chloratetreated S115 cells was not enhanced by testosterone treatment, whereas the testosterone response was restored by addition of heparin (Fig. 7A). Addition of K5-NSOS also restored the testosterone response in a concentration dependent manner such that K5-NSOS appeared at least as potent as heparin in this regard. By contrast, K5-OS had only a weak positive effect on the [ 3 H]thymidine incorporation by chlorate-treated S115 cells (Fig. 7A) indicating that its ability to facilitate the testosteroneinduced proliferative response was clearly distinct from that of K5-NSOS.
We further studied the effects of K5PSs on FGF signaling in CHO677 cells, which are deficient in endogenous HS but responsive to FGF in the presence of exogenous heparin (33). CHO677 cells were treated with FGF-8b alone or together with heparin/K5PSs and subjected to the [ 3 H]thymidine incorporation assay. The results were essentially similar to those obtained with chlorate-treated S115 cells in that heparin and K5-NSOS elicited a marked FGF response, while K5-OS had a much weaker effect (Fig. 7B). However, the differences between the stimulatory capacities of K5-NSOS and K5-OS were more pronounced in chlorate-treated S115 cells than in CHO677 cells. To assess whether the differential ability of K5-OS and K5-NSOS to enhance FGF signaling was dependent on the FGF species used to stimulate the cells, similar experiments were performed with CHO677 cells treated with FGF-1 and FGF-2. The results indicated that K5-NSOS was consistently more potent in stimulating the FGF responses regardless of the FGF species (Fig. 7, C and D). Unmodified K5 did not enhance thymidine incorporation in either of the model systems studied (data not shown).
The FGF responses of the wild-type CHO677 are presumably mediated by FGFR-1, which is expressed at low levels by the cells (33). To investigate how the cellular FGFR expression influenced the capacity of K5PSs to modulate FGF signaling, their effects on the proliferative FGF response were studied in CHO677 cells transfected to express a higher level of FGFR-1 or FGFR-4 (20). In the FGFR-1-transfected cells, K5-NSOS and K5-OS functioned largely in the same fashion than in untransfected CHO677 cells, such that the former polysaccharide species was more potent to stimulate FGF signaling, most distinctly in cells treated with FGF8b (Fig. 8A). By contrast, in the cells expressing FGFR-4, the two polysaccharide species did not display differences in their ability to support FGF signaling (Fig. 8B). This result indicates that, in the absence of endogenous HS, the capacity of K5-OS and K5-NSOS to stimulate FGF signaling is influenced by the FGFR species expressed by the target cells. Furthermore, the stimulatory capacity may also be influenced by the level of FGFR expression, because the differences between K5-OS and K5-NSOS appeared somewhat more distinct in cells expressing low levels of FGFR-1 than in those expressing higher levels of the receptor.
In conclusion, the data presented above suggest that K5-NSOS and K5-OS are markedly different in their capacity to regulate FGF signaling both in the presence or absence of endogenous HS. K5-NSOS displayed a modest inhibitory effect in FGF-8b-treated S115 cells expressing HS but seemed to enhance FGF activity when the same cells were stimulated with FGF-1 or FGF-2. Furthermore, K5-NSOS efficiently restored FGF activity in HS-deficient cells. By contrast, K5-OS showed a more consistent inhibitory effect in HS-expressing cells and was less potent than K5-NSOS to enhance FGF activity in HS-deficient cells. Studies with FGFR-transfected cells suggest that the capacity of K5PSs to regulate FGF signaling depends on the FGFR expression of the target cells. DISCUSSION Heparin has been shown to display antitumor activity in various in vitro models and in animal studies. Although data from systematically conducted clinical trials are still limited, there is increasing evidence that heparin therapy may retard cancer growth and dissemination also in humans (for review, see Refs. 34 and 35). The antitumor activity of heparin probably involves several mechanisms. Inhibition of growth factor signaling may entail reduced growth factor stimulation of tumor cells and suppression of tumor angiogenesis. The anticoagulant action of heparin may also contribute to its antitumor effects, because thrombin and other procoagulant factors, found both in the circulation and extravascularly, are known to stimulate the growth of many tumor types (36,37). Moreover, heparin exerts inhibitory action on the HS-degrading enzyme heparanase, which is thought to be involved in tumor cell invasion and metastasis as well as in the release of HS-bound angiogenic and tumorigenic growth factors (38).
The clinical applicability of heparin in cancer therapy may be limited by the anticoagulant activity of the polysaccharide, especially if higher doses than those used in the prophylaxis or treatment of thrombosis would be required. Partially desulfated heparin derivatives, with reduced anticoagulant activity, have been investigated for their potential to inhibit cancer growth and metastasis in experimental models. Such heparin preparations have been shown to decrease the growth of pancreatic adenocarcinoma cells in nude mice (39) and inhibit metastasis of melanoma cells (40,41), as well as to decrease the FGF-2-induced endothelial cell proliferation and angiogenesis (33). In addition, other sulfated polymers such as laminarin sulfate (42), pentosan polysulfate (43,44), and sulfated ulvans (45) have been tested as antitumor agents.
The current study indicates that chemically sulfated derivatives of K5PS modulate cellular FGF responses and provide novel information of the structure-activity relationships of such compounds. N-Acetylated, highly O-sulfated K5PS (K5-OS) was identified as an efficient inhibitor of FGF signaling in S115 mammary carcinoma cells, whereas the polysaccharide preparation containing both N-and O-sulfate substituents (K5-NSOS) had much lower inhibitory capacity. Experiments with cell models deficient in endogenous HS (CHO677 cells and chlorate-treated S115 cells) demonstrated that K5-NSOS had a more prominent agonist effect on FGF signaling than K5-OS, albeit the polysaccharides had a similar overall degree of sulfation. Furthermore, studies with HS-deficient cells transfected to express FGFR-1 or FGFR-4 suggested that the agonist activity of K5PSs was influenced by the FGFR species expressed by the cells but was less dependent on the FGF-type functioning as the receptor ligand.
Our data differ somewhat from the results of a previous investigation, in which both K5-OS and K5-NSOS inhibited the FGF-2 induced proliferation of bovine aortic endothelial cells, but only K5-NSOS was effective as an inhibitor of the proliferation of human umbilical vein endothelial cells and angiogenesis in the chick chorioallantoic membrane assay (18). The reasons as to why different K5PS structures were regarded as the most effective FGF inhibitors in these two studies may involve the differences between the cell models in which the FGF signaling was studied. For instance, the expression of FGFRs 1-4 and their splice variants in S115 mammary carcinoma cells is probably different from endothelial cells. Different FGFRs seem to differ in their interactions with HS (46 -50) and might analogously differ in their susceptibility to inhibition by sulfated K5PSs. Second, the cell surface density of FGFRs and HSPGs as well as the fine structure of HS in mammary carcinoma cells and endothelial cells may be different, which can influence the sensitivity of HS-FGF and HS-FGFR interactions toward the perturbation caused by exogenous K5PSs. Such differential sensitivity is perhaps reflected by the findings that efficient inhibition of FGF action in S115 mammary carcinoma cells was achieved with much lower concentrations of K5PS than in endothelial cells (1 g/ml versus 100 g/ml, respectively).
The importance of the cellular FGFR expression on the capacity of K5PSs to modulate FGF signaling was more directly suggested by experiments with HS-deficient cells (i.e. chloratetreated S115 cells and CHO677 cells). In these cells K5PSs were able to stimulate FGF signaling. We note that the stimulatory and inhibitory effects of K5PSs involve two distinct processes, because in HS-expressing cells K5PSs act together with the endogenous HSPGs and may either perturb or enhance the FGF/FGFR interactions, whereas in HS-deficient cells the polysaccharides are likely to act as unphysiological substitutes for the lacking endogenous HS component. K5-NSOS was a more potent stimulator of FGF signaling than K5-OS when the cells expressed FGFR-1, whereas the two polysaccharides had an equal stimulatory capacity in cells expressing FGFR-4. Interestingly, S115 cells have been shown to lack FGFR-4 (30) but to express FGFR-1, FGFR-3, and at least some FGFR-2 (25).
The previous findings (18) and the results reported here raise interesting questions with regard to the specificity of polysaccharide structures required to bind FGFs and support FGF receptor activation. The K5PS structures investigated in the present study differ from natural heparin/HS polymers in that they (i) have a higher overall degree of sulfation, (ii) contain GlcA as their sole hexuronic acid component and thus lack IdoA residues, (iii) are abundant in 2-O-sulfated GlcA units, which occur rarely in heparin/HS, and (iv) contain 3-Osulfated GlcA residues, which are not found in heparin/HS. The results of the present study demonstrate that both K5-NSOS and K5-OS had appreciable capacity to bind FGF-1, FGF-2, and FGF-8b, although they differ markedly from the heparin/HS structures implicated in binding to FGF-1 and FGF-2. FGF-2 interacts with N-sulfated oligosaccharides with critically important IdoA(2-OSO 3 ) residues (51), whereas the binding of HS to FGF-1 requires both 2-O-sulfated IdoA and 6-O-sulfated GlcNSO 3 units (51)(52)(53). The FGF-8b binding saccharide sequence has been elucidated in some detail, and our data suggest that, with regard to the sulfate groups involved in the interaction, the binding requirements would resemble those characteristic of FGF-1. 2 The FGF-binding domains of heparin/HS encompass short, 5-7-mer oligosaccharide sequences. By contrast, the saccharide domains implicated in the ternary FGF-FGFR-HS interactions are longer and may differ from the "minimal" FGF-binding domains in their pattern of sulfation. FGF receptor activation thus seems to require the FGF-binding domain and a flanking sequence with 6-O-sulfate groups that are critically important for the biological activity (54,55). In chlorate-treated S115 cells, a 10 -12-mer heparin sequence with N-, 2-O-, and 6-Osulfate groups is required for testosterone-induced proliferation (25). Our data indicate that while K5-OS bound avidly to FGF-8b (Fig. 5), its capability to promote FGF-8b signaling in chlorate treated S115 cells or HS-deficient CHO677 cells was poor. By contrast, K5-NSOS was able to stimulate FGF-8b signaling in HS-deficient cells, despite that the polysaccharide structure differs markedly from the physiological heparin/HS structures. The differences between K5-OS and K5-NSOS in their ability to support FGF signaling may be related to the presence of N-sulfated GlcN units or, alternatively, to a more favorable pattern of O-sulfation in the latter polysaccharide species.
Taken together, the available data warrant further studies of sulfated K5PSs with regard to their potential as FGF antagonists. O-Sulfated K5PS appears to be an interesting candidate for such studies, because it had a significant FGF antagonist activity but was largely devoid of agonist activity. Importantly, the present study indicates that the biological activities of different K5PS structures may vary remarkably between different experimental systems, such as cultured endothelial cells versus mammary carcinoma cells. The saccharide structures intended for modulation of FGF signaling may thus need to be specifically tailored for different biological and pathological situations according to the target cell types and FGFR/FGF species involved.