The biological functions of HSPGs relay on their capacity to bind different molecules including extracellular matrix proteins, enzymes, and protease inhibitors (1, 47). Of major interest for the aim of this chapter is the capacity of HSPGs to bind several growth factors, cytokines and chemokines involved in the angiogenesis process (Tab. 2), thus affecting their biological activity.

HSPGs modulate the biological activity of heparin-binding growth factors and cytokines through different mechanisms:

• The optional binding of the growth factor to soluble, ECM-associated or cell-surface HSPGs results in a fine control of the bioavailability of the protein. This is the case for TGF-b that binds betaglycan (5), a cell-associated PG, and decorin (48), which is present in the ECM, and for FGF-2 that binds basement membrane perlecan as well as cell-membrane syndecans (28, 33).
• The association with HSPGs stabilizes the growth factor and protects it from proteolytic degradation (11, 37).
• HSPGs modulate the access of growth factors to specific signaling receptors by different mechanisms (Fig. 4) (49).
• HSPGs can control the intracellular fate of the growth factor (50). Finally, the possibility exists that:
• transmembrane HSPGs themselves may transduce an intracellular signal, as suggested by the presence of highly conserved tyrosine residues in the C-terminal of all the members of the syndecan family, one of them fitting a consensus sequence for tyrosine phosphorylation; vi) HSPGs may activate an intracellular transduction signal by interacting directly with growth factor receptors (51, 52).

Fig. 4: Modulation of growth factor binding to tyrosine-kinase (TK) receptors by cell-surface HSPGs. 1) HSPGs increase local concentration of the growth factor (GF) in close vicinity to the TK receptor; 2) HSPGs induce a conformational change of the growth factor facilitating its interaction with TK receptor; 3) HSPGs induce growth factor oligomerization leading to TK receptor dimerization and activation.

Whatever the mechanism(s) of regulation of growth factor activity by HSPGs, it is interesting to note that the binding of the same growth factor to different HSPGs may have different biological consequences. This is the case for syndecan (44, 53), betaglycan (54) and perlecan (28), all able to bind FGF-2 but with different effects. Indeed, syndecan inhibits the mitogenic activity of FGF-2 (32) while perlecan promotes FGF-2-induced cell proliferation and angiogenesis (28).

Conversely, modifications of HSPG composition can regulate the sensitivity of the cell to different growth factors. This may be of particular relevance when the spatial and temporal control of the activity of different growth factors must be tightly enforced. This possibility is exemplified by the shift in cell-surface HSPG properties from a FGF-2- to an FGF-1-binding phenotype in murine neuronal cells during embryonic development (55).

The modality by which the various HSPGs "discriminate" among the several heparin-binding growth factors is based on their different core proteins, the high heterogeneity of GAG-chain composition, and on the possibility that both the protein moiety and GAG-chains may interact with different growth factors. For instance, betaglycan can exist as a "nude" core protein and the presence and composition of the GAG-chains of this HSPG can be regulated in response to FGF-2 (8). FGF-2 itself binds the GAG-chain of betaglycan while the core protein can interact with TGF-b (54). Also, the number and fine structure of HS chains in syndecan 1 vary in different tissues and in relation to cell differentiation (56). Finally, different sulfated groups and distinct oligosaccharide sequences of the GAG-chain are responsible for the binding to different growth factors (see below).

In conclusion, HSPGs are characterized by a structural variability that appears to be highly regulated and that offers virtually unlimited possibilities for selective interactions with different growth factors, cytokines, and chemokines.

Heparin consists largely of 2-O-sulfate IdoA N,6-disulfate GlcN disaccharide units. Other disaccharides containing unsulfated IdoA or GlcA and N-sulfated or N-acetylated GlcN are also present as minor components. This heterogeneity is more pronounced in HS, where the low-sulfated disaccharides are the most abundant.

As stated above, heparin/HS interacts with various angiogenic growth factors. These interactions depend on the size of the polysaccharide chain and on the degree and distribution of sulfate groups. Interestingly, distinct oligosaccharide sequences have been identified to retain angiogenic growth factor-binding capacity. For instance, the minimal FGF-2-binding sequence in HS has been identified as a pentasaccharide which contains the disaccharide units IdoA(2-OSO₃)-GlcNSO₃ or IdoA(2-OSO₃)-GlcNSO₃(6-OSO₃) (57). Accordingly, binding studies involving chemically modified heparins or HS preparations have shown that 2-O- and N-sulfate groups are important for FGF-2 interaction (Fig. 5).
 

Fig. 5: Role of sulfate groups of heparin/HS in growth factor interaction. The sulfate groups involved in high affinity interaction with different angiogenic growth factors are indicated. The data referring to IL-8 are based on preliminary observations by the authors (see text for further details).

However, FGF-1 and FGF-4 differ distinctly from each other and from FGF-2 in their interaction with selectively O-desulfated heparins, 6-O-sulfate groups being required in addition to 2-O-sulfate groups for GAG interaction (58).

Also, HGF interacts mainly with 6-O-sulfate groups of GlcNSO₃ residues while N-sulfates and IdoA(2-OSO₃) units play a limited role (59).

Recently, we have shown that HIV-Tat protein requires 2-O-, 6-O- and N-sulfate groups for optimal interaction with heparin (60).

Finally, heparin/HS subpopulations with chemokine-binding selectivity exist (13). For instance, we have found that heparin, HS, 6-O-desulfated heparin, and N-desulfated/N-acetylated heparin, but not 2-O-desulfated heparin, chondroitin-4-sulfate and nonsulfated hyaluronic acid affect IL-8 activity in endothelial GM 7373 cells, suggesting that the backbone structure, sulfation and the arrangement of the charges are of importance in determining the capacity of GAGs to bind IL-8 (M. Presta, manuscript in preparation).

It should be pointed out that the binding of growth factors to heparin/HS does not depend only on their interaction with sulfate groups of the GAG chain. For instance, it has been shown that carboxyl groups of heparin contribute to FGF-2 interaction (61) and that small non sulfated di- and tri-saccharides compete with radio-labeled heparin for the binding to FGF-2, albeit with a potency that is about 100 fold lower than that of unlabeled heparin (62).

Taken together, the data indicate that distinct structural requirements are necessary for the interaction of heparin/HS with different growth factors. Even though these specific binding sequences may be hidden in heparin due to its high degree of sulfation, the high heterogeneity in HS structure allows a more refined tailoring of selective binding regions that may influence the biological activity and bioavailability of heparin/HS-binding growth factors.

The interaction of angiogenic growth factors with heparin depends also on distinctive biochemical features of the protein. For instance, heparin-binding region(s) have been tentatively identified both in NH₂ terminus (63) and COOH terminus (64, 65) of FGF-2 where basic amino acid residues may interact with sulfate groups of heparin. X-ray crystallography has identified a cluster of basic amino acids that form a "basic task" in the 3D structure of FGF-2 that is able to interact with 1-2 sulfate groups, thus representing a putative heparin-binding region (66-68).

The above observations implicate that an appropriate 3D structure of the FGF-2 molecule is required for interaction with GAGs. Also, uncharged amino acids of FGF-2 have been shown to participate to heparin interaction, indicating that hydrogen bonds, van der Waals packing and hydrophobic interactions, as well as ionic interactions, provide a significant contribution to the formation of the FGF-2-heparin complex (62, 69).

At variance with FGF-2, the heparin-binding capacity of different angiogenic factors may depend on linear stretches of contiguous basic amino acids. These basic regions are present in the heparin-binding isoforms of VEGF and PlGF, originating from alternative splicing of their mRNAs (70), and in the first exon of HIV-Tat protein (M. Rusnati, unpublished observation).

Heparin-binding angiogenic growth factors bind to HSPGs associated with ECM and endothelial cell surface with a Kd equal to approximately 5-50 nM. HIV-Tat accumulates in the HSPGs of the ECM (71). Also, FGF-2 has been found in endothelial ECM in vitro (41, 72, 73) and basement membranes in vivo (74-76). Newly synthesized FGF-2 is stored in ECM from where it can be released to induce long-term stimulation of target cells (12, 39, 40, 41). Thus, ECM may act as a physiological reservoir for extracellular angiogenic factors.

1) Plasmin, a serine protease, releases FGF-2-GAGs complexes by degrading the core protein of cell-associated HSPGs (11). Because of its association with GAGs, released FGF-2 is protected from proteolytic inactivation and endowed with a larger radius of diffusion. Indeed, both heparin and solubilized GAGs of endothelial origin bind FGF-2 and protect it from heat and acidic inactivation (36) and from proteolytic degradation (11, 37). Moreover, soluble GAGs may favor the delivery of FGF-2 to the blood supply to stimulate angiogenesis by increasing the radius of diffusion of FGF-2 (38).

This kind of releasing mechanism is strictly controlled by cytokines such as TGF-b and FGF-2 itself that affect the synthesis of plasminogen activators and their inhibitors in endothelial cells, thus modulating the activation of the proenzyme plasminogen to plasmin (26).

2) Heparitinase, heparinase and heparanase, but not chondroitinase or hyaluronidase, also release biologically active FGF-2 from ECM by degrading the saccharidic backbone of immobilized HSPGs (12).

3) Phospholipase C releases FGF-2 as a biologically active complex with a GPI-anchored HSPG, suggesting that also endogenous phospholipase may be involved in the processes of mobilization of FGF-2 from cell-associated HSPGs (13).

4) Finally, free GAGs inhibit the binding of FGF-2 to cell-associated HSPGs with a potency that decreases in the following order: heparin > HS > dermatan sulfate. Hyaluronic acid, chondroitin-6-sulfate and chondroitin-4-sulfate are ineffective (77). This suggests that soluble HSPGs generated by partial hydrolysis or by proteolysis of cell-associated HSPGs may mobilize heparin-binding angiogenic growth factors from ECM.

From these data it appears that the balance between storage and release of angiogenic growth factors in ECM, as well as the integrity of the matrix, may regulate the biological effects of these molecules on endothelium (78, 79). For instance, the capacity of nitric oxide to mediate interleukin-1-induced degradation of ECM in cultured chondrocytes, with consequent release of extracellular FGF-2, has been implicated in the neovascularization of the synovia of arthritic patients (80).

Angiogenic growth factors induce response in target endothelial cells by binding to cognate cell-surface tyrosine kinase (TK) receptors (81). The interaction of heparin-binding growth factors to TK receptors is modulated by HSPGs. For instance, the interaction of FGF-2 or of the heparin-binding VEGF₁₆₅ isoform to TK receptors is strongly reduced in cells made HSPG-deficient by treatment with heparinase or chlorate (82, 83).

However, controversial results exist about the absolute requirement for HSPGs in FGF-2/receptor interaction. Yayon et al. (84) reported that HS-deficient CHO cells transfected with FGF receptor-1 (FGFR-1) do not bind FGF-2 unless heparin or HS are added to the cell culture medium. In contrast, Roghani et al. (85) have shown that FGFRs expressed in CHO cell mutants or myeloid cells retain the capacity to bind FGF-2 also in absence of heparin. In these experimental conditions, heparin induces a three-fold increase in the affinity of the growth factor for its receptor.

Controversial results were obtained also in cell-free systems. Ornitz et al. (86) showed that heparin represents an absolute requirement for cell-free binding of FGF-2 to a soluble form of the extracellular portion of FGFR-1, while Roghani et al. (85) reported that heparin is not necessary for the binding of FGF-2 to soluble FGFR. In agreement with these latter results, we have found that the formation of the FGF-2-FGFR complex in solution occurs also in the absence of heparin and it is enhanced by the GAG (87).

Similar results were obtained for the interaction of VEGF₁₆₅ with soluble KDR/flk-1 receptor (88). Interestingly, HSPGs are required also for receptor interaction of VEGF₁₂₁, a VEGF isoform lacking heparin binding ability (89). This latter observation, as well as the capacity of heparin to induce FGF-2-FGFR interaction in HS-deficient cells, can be interpreted on the basis of the capacity of GAGs to form ternary complexes by interacting with both ligand and receptor proteins (58, 87, 90). Indeed, a heparin binding domain has been identified in the NH₂-terminus of IgG-like domain II of FGFR-1 (91).

The puzzling observation that heparin itself can activate FGFR in the absence of the growth factor (52) further increases the complexity of the ternary interaction among GAGs, growth factors, and TK receptors.

From the above considerations it derives that:

• heparin-related molecules can be used to modulate the biological activity of heparin-binding angiogenic growth factors and cognate TK receptors;
• the structural requirements of heparin necessary to bind the growth factor and the TK receptor or to affect their mutual interaction may be different;
• heparin-related molecules with different structures able to differently affect the biological activities of angiogenic growth factors can be designed.

We have observed that heparin requires both 2-O- and 6-O-sulfate groups, as well as N-sulfate groups, to promote the binding of FGF-2 to soluble FGFR-1 (87). Thus, the binding of heparin/HS to FGF-2, which does not require 6-O-sulfate groups, is not sufficient to induce FGF-2 interaction with FGFR. Accordingly, unmodified heparin, but not 6-O-desulfated heparin, protects FGFR-1 from trypsin digestion (87).

These data support the hypothesis that HSPGs modulate the binding of FGF-2 to FGFR through the formation of a ternary complex in which the GAG chain interacts with FGF-2 via 2-O- and N-sulfate groups while 6-O-sulfate groups are required for its interaction with FGFR (58).

A single molecule of heparin/HS may bind several molecules of FGF-2 (95), suggesting that GAG induces oligomerization of FGF-2. Indeed, it has been demonstrated that heparin induces dimerization of FGF-2 in a cell free system (86).

A common theme among growth factors interacting with TK receptors is the involvement of ligand-induced receptor dimerization in receptor activation (96). It has been demonstrated that the dimerization and activation of FGFR catalyzed by heparin-dependent oligomerization of FGF-1 is required to induce a mitogenic response (97). Heparin has been hypothesized to play a similar role also for FGF-2 (49, 79, 98).