The Anti-angiogenic His/Pro-rich Fragment of Histidine-rich Glycoprotein Binds to Endothelial Cell Heparan Sulfate in a Zn2+-dependent Manner*

The plasma protein histidine-rich glycoprotein (HRGP), which has been identified as an angiogenesis inhibitor, binds to heparan sulfate (HS) in a Zn2+-dependent manner. We wished to test whether this interaction is mechanistically important in mediation of the anti-angiogenic effect of HRGP. Inhibition of angiogenesis by HRGP is exerted through its central His/Pro-rich domain, which is proteolytically released. A 35-amino-acid residue synthetic peptide, HRGP330, derived from the His/Pro-rich domain retains the inhibitory effect on blood vessel formation in vitro and in vivo, an effect dependent on the presence of Zn2+. We now show that HRGP330 binds heparin/HS with the same capacity as full-length HRGP, and the binding is Zn2+-dependent. Peptides derived from the His/Pro-rich domain of HRGP downstream of HRGP330 fail to inhibit endothelial cell migration and display a significantly reduced heparin-binding capacity. An even shorter peptide, HRGP335, covering a 26-amino-acid sequence within HRGP330 retains full heparin/HS-binding capacity. Characterization of the HS interaction shows that there is a tissue-specific HS pattern recognized by HRGP335 and that the minimal length of heparin/HS required for binding to HRGP335 is an 8-mer oligosaccharide. Saturation of the HS binding sites in HRGP330 by pre-incubation with heparin abrogates the HRGP330-induced rearrangement of endothelial cell focal adhesions, suggesting that interaction with cell surface HS is needed for HRGP330 to exert its anti-angiogenic effect.

Angiogenesis, the formation of new capillary blood vessels, is essential during development and physiological conditions, such as wound healing and the menstrual cycle (1). However, prolonged and excessive angiogenesis has been implicated in a number of pathological processes, for instance rheumatoid arthritis, diabetic retinopathy, and tumor growth (2,3). The adult vasculature is tightly regulated by naturally occurring pro-and anti-angiogenic factors. Fibroblast growth factor (FGF) 3 and vascular endothelial growth factor (VEGF) are well known stimulators of endothelial cells in vitro and in vivo. To date, a number of endogenous factors negatively regulating angiogenesis have also been identified (4). The inhibitors described so far mainly fall into three groups: plasma proteins, basement membrane proteins, and serine protease inhibitors (serpins). A common feature of many of these antiangiogenic molecules is inhibition of endothelial cell chemotaxis in vitro.
We have identified histidine-rich glycoprotein (HRGP) as a potent inhibitor of angiogenesis in vivo (5). HRGP is a heparin-binding plasma protein highly conserved through vertebrate species (for review, see Ref. 6). The N-terminal part of the protein contains two cysteine protease inhibitor (cystatin)-like stretches, hence the classification of HRGP as a member of the cystatin superfamily (Fig. 1A). The central domain is rich in histidine and proline residues, and the human form contains 12 more or less conserved tandem repeats of the pentapeptide HHPHG. Multiple binding partners for HRGP have been reported, such as heparin/heparan sulfate (HS), plasminogen, fibrinogen, tropomyosin, and divalent cations (6). The heparin-binding affinity of HRGP can be modulated and is increased in the presence of Zn 2ϩ and at low pH (7), a common environment in hypoxic tumors. The anti-angiogenic effect of HRGP is mediated via its His/Pro-rich domain, which needs to be released from the full-length protein to exert its effects (5). HRGP attenuates endothelial cell migration and adhesion to vitronectin in vitro and tumor vascularization and growth in vivo, by interfering with endothelial cell focal adhesion function. We have narrowed down the minimal active domain of HRGP to a 35-amino-acid (aa) peptide, HRGP330, corresponding to a sequence in the His/Pro-rich domain of HRGP (8).
Binding to heparin/HS is a common feature of angiogenesis inhibitors as well as of pro-angiogenic factors (for review, see Ref. 9). For instance, it has been demonstrated both in vitro and in vivo that HS binding to the angiogenesis inhibitor endostatin (10) is required for its negative regulation of blood vessel formation (11)(12)(13)(14)(15). Among many other functions, HS chains act as a co-factor in FGF signaling (16 -18) and as regulators of morphogen gradients (reviewed in Ref. 19). HS chains are covalently bound to a protein core, which may be either a cell surface or an extracellular matrix protein. HS chains are long, unbranched and negatively charged polysaccharide chains composed of disaccharide units derived from a precursor (GlcNAc␣1-4GlcUA␤1-4) n structure. During HS biosynthesis, these units are modified in a celland tissue-specific manner (20 -22) by N-deacetylation and N-sulfation of N-acetyl glucosamine (GlcNAc), C5-epimerization of glucuronic acid (GlcUA) to iduronic acid, and O-sulfation in different positions (reviewed in Ref. 23). These modifications occur in clusters defined by the pattern of N-sulfation and give rise to different domains in HS: the poorly modified N-acetylated domains, the more extensively modified N-sulfated (NS) domains, and the N-acetylated/NS domains with alternating N-acetylated and NS disaccharides (20). In the present study, we have characterized the heparin/HS-binding properties of HRGP330 and the role of this interaction for its inhibitory effect on endothelial cells.

MATERIALS AND METHODS
Cell Culture-Primary bovine adrenal cortex capillary endothelial cells, a kind gift from Dr. R. Christofferson, Department of Medical Cell Biology, Uppsala University, were cultured on gelatin-coated dishes in Dulbecco's modified Eagle's medium/10% newborn calf serum and 2 ng/ml FGF-2 (Peprotech). Telomerase-immortalized endothelial (TIME) cells derived from human dermal microvascular endothelial cells were a kind gift from Dr. Martin McMahon, Cancer Research Institute, University of California-San Francisco (24). TIME cells were cultured in complete endothelial cell basal medium MV2 (Promocell) without antibiotics on gelatin-coated cell culture plastic. Human embryonic kidney 293-Epstein-Barr nuclear antigen 1 (EBNA) cells were cultured in Dulbecco's modified Eagle's medium/10% fetal calf serum. Approximately every second month the cells were given 0.25 mg/ml G418 (Calbiochem) to ensure selection of EBNA-1 expression. Dulbecco's modified Eagle's medium and serum were from Invitrogen.
Recombinant Protein Production-Full-length human HRGP and the truncation HRGP1-240 (previously referred to as His-2 (5)) were produced in human embryonic kidney 293-EBNA cells under serum-free conditions as previously described (5). The recombinant proteins contain a His tag (His 6 ) in the N terminus of the coding region, and purification was performed using nickel-nitrilotriacetic acid-agarose (Qiagen). Protein-containing fractions were pooled and dialyzed against phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 , 1.4 mM KH 2 PO 4 ) pH 7.4.
HS was quantified by colorimetric determination of hexuronic acid using the metahydroxydiphenyl method (29) with GlcUA as a standard.
GlcUA values were arbitrarily multiplied by 3 to convert values to saccharide mass.
HS N-sulfated domains were prepared by essentially complete N-deacetylation of HS in hydrazine monohydrate containing 1% hydrazine sulfate for 4 h at 96°C (27) followed by cleavage with HNO 2 at pH 3.9 (30), as described earlier (31). A radioactive label was introduced at the reducing end of the cleavage products by reduction with NaB 3 H 4 (55-65 Ci/mmol; Amersham Biosciences), yielding a labeled 2,5-anhydromannitol (aManR) residue (32). Remaining anhydromannose residues were reduced with an excess of NaBH 4 . Unincorporated radioactivity was separated from the labeled material by desalting on a column (1.5 ϫ 7 cm) containing 10 ml of Sephadex G-10 (General Electric Health Care, Uppsala, Sweden). Subsequent size fractionation was performed on a Bio-Gel P-10 (fine, Bio-Rad ) column (1 ϫ 200 cm) in 0.5 M NH 4 HCO 3 at a flow rate of 2 ml/h. Fractions corresponding to the different oligosaccharide sizes were pooled and repeatedly evaporated to dryness until neutral in pH. The specific activity for HS oligosaccharides ranged from ϳ7.5 ϫ 10 5 Ϫ 5.9 ϫ 10 6 cpm/nmol oligosaccharide. Heparin oligosaccharides were generated by partial deaminative cleavage of the polysaccharide with nitrous acid at pH 1.5, as described previously (47). The products were radiolabeled by reduction with NaB 3 H 4 to a specific activity of ϳ7.4 ϫ 10 4 cpm/nmol oligosaccharide.
Protein/Oligosaccharide-binding Assays-Binding between radiolabeled polysaccharides/oligosaccharides and peptides was performed in solution as described previously (33). In a typical assay, 3 H-labeled heparin, HS, or size-defined oligosaccharides were incubated with 320 pmol of peptide in 200 l of 50 mM Tris buffer, pH 7.4, containing 0.5 mg/ml bovine serum albumin, 140 mM NaCl, and 15 M ZnCl 2 . Controls were performed with carrier bovine serum albumin and radiolabeled oligosaccharides only. After 20 min, the incubation mixtures were filtered through nitrocellulose filters (25 mm diameter, 0.45 m pore size; Sartorius, Germany) that were then rapidly rinsed. Protein-bound oligosaccharides were released from the filter by 2 M NaCl, and the radioactivity was measured with a Beckman LS 6500 liquid scintillation counter. For determination of affinity constants, 25 pmol of protein/ peptide was used, and the ligand (heparin 12-mer) varied between 1.6 and 34 pmol and 3.5 and 65 pmol for measurement with HRGP and HRGP335, respectively. K D values were calculated by nonlinear regression for one binding site using the Kaleidagraph software package using the equation y ϭ [HRGP⅐heparin] max ϫ [heparin] free /(K D ϩ [heparin] free ).
Chemotaxis Assay-The chemotaxis assay was performed using a modified Boyden chamber as described earlier (34) with 8-m micropore polycarbonate filters (PFB8 -50; Neuro Probe Inc.) coated with type-1 collagen solution at 100 g/ml (Vitrogen 100; Collagen Corp.). TIME cells that had been starved overnight in 1% fetal calf serum were trypsinized and resuspended at 4 ϫ 10 5 cells/ml in endothelial cell basal medium, 0.25% bovine serum albumin, and trasylol at 1000 kallikreininhibiting units. The cell suspension was added in the upper chamber and VEGF-A (10 ng/ml) (Peprotech) in the lower chamber. HRGP335 was added both in the upper and in the lower chamber at 100 ng/ml. Zn 2ϩ was included or not in the assay medium as 10 M ZnCl 2 . After 5 h at 37°C, cells that had migrated through the filter were stained with Giemsa and counted using the Easy Image Analysis software (Tekno Optik, Sweden). All samples were analyzed in at least six wells for each treatment and at several separate occasions. Statistical significance was calculated using Student's t test.
Paxillin Staining-Bovine adrenal cortex capillary endothelial cells were seeded onto gelatin-coated glass culture slides (Falcon) and starved overnight in Dulbecco's modified Eagle's medium/1% newborn calf serum. The cells were stimulated for 15 min with 100 ng/ml HRGP330 that had been pre-incubated or not with a 50ϫ molar excess of a Ͼ14-mer heparin fragment. Control cells received vehicle or vehicle plus heparin. The slides were put on ice, washed in ice-cold TBS (20 mM Tris/HCl, pH 7.5, 150 mM NaCl), and fixed in zinc-fix (0.05% calciumacetate, 0.5% zinc-acetate, 0.5% ZnCl in TBS, pH 6.6, and 0.2% Triton X-100) for 15 min at room temperature. Paxillin was detected using a monoclonal anti-paxillin antibody (#610052; BD Biosciences) diluted 1:400 and a secondary anti-mouse Alexa 488-conjugated antibody (#A-11017; Molecular Probes) diluted 1:1000. Nuclei were stained with Hoechst 33342 (1 g/ml; Molecular Probes). The preparations were examined and microphotographed using the 60ϫ objective in a Nikon Eclipse E1000 microscope. Quantification of the number of focal adhesions in the cell body and cell edge in HRGP330-treated cells and cells treated with HRGP330 pre-incubated with heparin was performed by counting the number of focal adhesions in a square of defined size in three randomly chosen areas in eight separate photos from each treatment. Statistical significance was calculated using Student's t test.
Binding of Biotinylated HRGP330 to Endothelial Cells-TIME cells were seeded at confluency (20,000 cells/well) in vitronectin-coated 96-well plates. Next day, the cells were put in serum-free medium (endothelial cell basal medium) for five h in 37°C to remove potentially cell-bound bovine HRGP derived from fetal calf serum in the cell culture medium. The cells were incubated in triplicate for 45 min on ice with 100 ng/ml BT-HRGP330 alone or in combination with a 10-fold molar excess of unlabeled HRGP330, HRGP398, or HRGP. One set of cells was incubated with BT-HRGP330 that had been pre-incubated with a molar excess of heparin. Another set of cells was pre-incubated with a mixture of 10 milliunits each of the heparin lyases heparinase (Seikagaku Corp., Tokyo, Japan) and heparitinase (Seikagaku) in a final volume of 200 l of buffer (final concentration 50 units/ml) for 1.5 h before incubation with BT-HRGP330. After the incubation, the cells were washed once in cold TBS and then incubated with streptavidin-conjugated Alexa 680 (Invitrogen catalog number A-21057) at a dilution of 1:1000 for 45 min on ice. After three washes in TBS, the infrared signal from Alexa 680 was detected using the Odyssey Imaging system (Westburg BV).

Determination of the Heparin-binding Region in HRGP and Correlation to Anti-angiogenesis-
We have previously demonstrated that the anti-angiogenic effect of HRGP is mediated via the His/Pro-rich domain, which can be proteolytically released from HRGP (5). HRGP binds to heparin in a Zn 2ϩ -dependent manner (7,35). To investigate the contribution of the His/Pro-rich region to this interaction, we determined the heparin-binding properties of full-length HRGP and a truncated form corresponding to the N-terminal cystatin domains (HRGP1-240), i.e. lacking the His/Pro-rich domain (Fig. 1A). Using a nitrocellulose filter trapping assay (33), we studied the binding of [ 3 H]acetyl-labeled heparin to HRGP and HRGP1-240 in a Zn 2ϩ -containing buffer (Fig. 1B). In the presence of 15 M Zn 2ϩ , HRGP1-240 showed a 9-fold lower heparin binding compared with intact HRGP. This finding indicated that, despite the ability of the N-terminal to bind heparin, the main heparin-binding domain is located outside of this region. In a parallel study, we investigated the anti-angiogenic properties of three synthetic peptides, HRGP330, -365, and -398, derived sequentially from the His/Pro-rich domain of HRGP. Of these three peptides, only HRGP330 was found to retain the anti-angiogenic properties, an effect dependent on the presence of Zn 2ϩ (8) (Fig. 1A). The heparin-binding capacity of HRGP330 was comparable with that of intact HRGP, whereas HRGP365 and -398 showed considerably lower binding (Fig. 1B). To further narrow down the heparin-binding domain, a 26-amino-acid peptide within HRGP330, HRGP335, was tested in the filter trapping assay. HRGP335 was able to bind heparin to a similar extent as intact HRGP. Shorter peptides were not retained by the nitrocellulose filter and could therefore not be evaluated in this assay. Moreover, the apparent dissociation constant (K D ) of the interaction between HRGP335 and heparin ranged in the same order as for full-length HRGP (10 nM for HRGP and 52 nM for HRGP335; Fig. 1, C and D, respectively).
HRGP335 was tested for its ability to inhibit endothelial cell chemotaxis. TIME cells were induced to migrate toward VEGF-A in a modified Boyden chamber in the absence or presence of HRGP335 and Zn 2ϩ , respectively (Fig. 1E). The VEGF-A-induced chemotactic response was inhibited by HRGP335, an effect dependent on the presence of Zn 2ϩ . These data further support a relation between the anti-angiogenic effect and the heparin-binding capacity of HRGP-derived fragments.
Influence of Divalent Cations on the Heparin-binding Ability of HRGP335-Zn 2ϩ is a known ligand for HRGP, and its importance for the binding of HRGP to heparin (7,35) and cell surface HS (36) (Fig. 2). However, already at 10 -15 M Zn 2ϩ , which represents the physiological level, a substantial amount of heparin bound to HRGP335. In contrast, Ni 2ϩ , Co 2ϩ , and Cu 2ϩ were able to mediate binding between heparin and HRGP335 only at concentrations far above physiologically relevant levels. Other divalent cations, such as Ca 2ϩ , Mg 2ϩ , Mn 2ϩ , Sr 2ϩ , and Hg 2ϩ , were not able to mediate binding between HRGP335 and heparin, even at concentrations as high as 1.5 mM for Mn 2ϩ , Sr 2ϩ , and Hg 2ϩ or 5 mM for Ca 2ϩ and Mg 2ϩ (data not shown). Taken together these results confirm the importance of Zn 2ϩ for heparin binding to the His/Pro-rich domain of HRGP under physiological conditions.
Determination of the Heparin Fragment Size Required for HRGP335 Binding-To elaborate the molecular details of HRGP335 binding to heparin, size-defined [ 3 H]heparin oligosaccharides were tested for their ability to interact with HRGP335. The peptide was incubated with equimolar amounts of 3 H-labeled oligosaccharides, and those bound to the peptide were trapped along with the peptide on a nitrocellulose filter (Fig. 3). Oligosaccharides showed increased binding capacity with increasing size, reaching a plateau at the size of a dodecasaccharide. A fraction of hexasaccharides was retained on the filter by HRGP335, although ϳ8-fold less than the fraction of dodecasaccharides bound. Octasaccharides showed a 2-fold lower binding than dodecasaccharides, whereas decasaccharides reached almost the level of dodecasaccharides.
Tissue-specific Binding of HRGP335 to HS Chains and NS Domains-Because HS (and not heparin) is present on the surface of endothelial cells, we assayed equimolar amounts of HS from different porcine tissues for binding to HRGP335. HS from aorta and intestine showed a 3-6-fold lower binding capacity than HS from lung, liver, kidney, and heart (Fig. 4A). Of all tested HS preparations, liver HS showed the highest affinity for HRGP335. To further investigate the tissue preference seen with full-length HS, we tested the ability of the highly sulfated NS domains from these different HS sources for their binding to HRGP335. NS domains were prepared from the HS of porcine liver and intestine by N-deacetylation and depolymerization by complete deaminative cleavage at pH 3.9 and were then radiolabeled and size-fractionated. NS domains from both tissues were able to bind to HRGP335, although in a different fashion (Fig. 4B). NS octasaccharides from liver showed similar binding properties as heparin fragments, with an increasing proportion of the longer fragments bound. In contrast to oligosaccharides from liver, 8-mer NS domains from intestinal HS showed essentially no binding, whereas deca-and dodecasaccharides contained structural features sufficient for binding.

HRGP330-induced Rearrangement of Endothelial Cell Focal Adhesions Requires Interaction with Cell
Surface HS-To test whether the anti-angiogenic effect of HRGP330 was dependent on its interaction with HS on the endothelial cell surface, we pre-incubated HRGP330 with a molar excess of heparin. Preincubation with heparin should thus prevent the peptide from interacting with cell surface HS. Evaluation of the effect of HRGP330 pre-incubated with heparin on endothelial cell chemotaxis was not feasible, as the addition of heparin in the Boyden chamber assay neutralized the response stimulated by the growth factor. Instead, we investigated the effect of HRGP330 pre-incubated with heparin on focal adhesions, a growth factor-independent assay. We have previously shown that HRGP and HRGP330 induce endothelial cell focal adhesions (5,8), a response mechanistically connected to the anti-angiogenic properties of HRGP. In agreement, HRGP330 induced a marked increase in the number of focal adhesions, evenly distributed To the right, the anti-chemotactic activity of the respective protein/peptide is indicated by Ϯ (5). B, equimolar amounts of proteins and peptides (250 pmol) were incubated with 35 pmol of 3 H-labeled full-length heparin. Protein and bound heparin were trapped by filtration through a nitrocellulose filter and the recovered oligosaccharides quantified by scintillation counting. The assay was performed on duplicate samples. Error bars show the deviation of the samples from the average. In the same assay system, the affinity constant for HRGP (C) and HRGP335 (D) was assayed by varying the amount of ligand added to a constant amount of protein as described under "Materials and Methods." E, VEGF-A-induced chemotaxis (VEGF-A at 10 ng/ml) of TIME cells was arrested by the inclusion of HRGP335 (100 ng/ml) in a Zn 2ϩ -dependent manner. The result is depicted as fold stimulation compared with unstimulated cells (control). Error bars represent one standard deviation. Significance at the level of p Ͻ 0.01 is indicated. over the cell body (Fig. 5A). Cells treated with HRGP330 that had been pre-incubated with heparin showed a marked loss in focal adhesions. Focal adhesions remaining in this condition were specifically located to the cell edges. Heparin alone had no effect on focal adhesions compared with control (Fig. 5A). Quantification of the number of focal adhesions in the cell body and the cell edge in cells treated with HRGP330 or HRGP330 pre-incubated with heparin showed a significant decrease in cell body focal adhesions when HRGP330 binding to cell surface HS was prevented by heparin pre-incubation (Fig. 5B). In conclusion, this result strongly suggests that interaction with HS on the endothelial cell surface is required for a HRGP330-induced rearrangement of focal adhesions and for the peptide to exert its anti-angiogenic effect.
HRGP330 Binds to HS on Endothelial Cells-To investigate whether direct binding of HRGP330 to the endothelial cell surface could be detected and whether this binding was dependent on HS, endothelial cells were incubated with 100 ng/ml biotinylated HRGP330 (BT-HRGP330), either pre-incubated or not with heparin. BT-HRGP330 was also added to cells that had been pretreated with a combination of heparin lyases (heparinase and heparitinase) to remove cell surface HS. Bound BT-HRGP330 was detected using streptavidin-coupled fluorescent Alexa 680. As shown in Fig. 6, lyase treatment of cells or heparin pre-incubation of BT-HRGP330 significantly reduced the binding of BT-HRGP330 to endothelial cells. The fluorescent signal detected from cells treated with BT-HRGP330 was lost in the presence of a 10-fold molar excess of unlabeled HRGP330 but not in the presence of HRGP398 (which lacks heparin-binding ability; see Fig. 1A), demonstrating specificity in the assay. In addition, the fluorescent signal was also lost in the presence of a 10-fold excess of HRGP, demonstrating that HRGP330 and HRGP compete for the same endothelial cell surface receptor.

DISCUSSION
HRGP is a potent inhibitor of angiogenesis and tumor growth in mice, an effect mediated via the proteolytically released His/Pro-rich domain of the protein (5,37,38). HRGP has been shown to interact with heparin and HS (35,36,39,40), a feature shared by many pro-and anti-angiogenic factors (for a review, see Ref. 9). To elucidate the role of HS binding for the anti-angiogenic activity and mechanism of action of HRGP, we investigated the molecular details of the HS-binding properties. By analysis of the heparin binding of truncated HRGP and synthetic peptides, we were able to ascribe a strong heparin-binding activity to the peptides HRGP330 and HRGP335, covering amino acid sequences 330 -364 and 335-360, respectively, in the His/Pro-rich domain. The N-terminal domain of the protein, as well as peptides derived from sequences downstream of HRGP330 in the His/Pro-rich region, showed much weaker heparin binding. These findings suggest that the N-terminal domain is not the main heparin-binding domain as earlier proposed (36). One possible explanation for this discrepancy may be that significant parts of the His/Pro-rich region were not included in the latter study.
To further characterize the interaction between HRGP335 and HS/heparin, we determined that heparin octasaccharides are sufficient for binding to HRGP335, although the affinity increases with oligosaccharide length up to a dodecasaccharide. The higher apparent affinity of longer fragments may be explained either by the increased probability of encountering binding motifs in larger fragments or the presence of additional binding sites that strengthen the interaction.
HS is less sulfated than heparin and therefore usually binds weaker to proteins (41). However, unlike heparin, HS is present on the cell surface and is therefore the natural ligand for most heparin-binding proteins. HRGP does not bind to cells lacking HS (36), indicating that HRGP can   make use of cellular HS as a natural cell surface ligand. Indeed, binding of biotinylated HRGP330 was competed by the removal of cell surface HS, heparin lyase treatment, heparin pre-incubation of BT-HRGP330, and by HRGP, demonstrating HS structures as receptor molecules for HRGP and HRGP330 on endothelial cells. In accordance, the 26-aminoacid peptide HRGP335 was also able to bind to HS, with the interaction capacity depending on the tissue source of the HS chains. HS species from different sources differ in charge and domain organization (20,22), which could explain their differential binding properties. Despite the markedly basic properties of HRGP335, these differential binding properties cannot be explained solely by a difference in overall charge of the HS chains. For example, intestinal HS chains are of higher overall sulfation degree and contain more highly sulfated NS domains as compared with lung HS (20), yet HRGP335 displayed a lower affinity for intestinal HS. Possible explanations for this result include the distribution (spacing), quality (degree and type of O-sulfation), and relative proportions of the highly sulfated NS domains within the different HS chains. Analysis of the binding properties of NS domains isolated from liver HS (good peptide binder) and intestinal HS (poor peptide binder) showed that liver octasaccharide NS domains bind to HRGP335, whereas intestinal NS domains do not bind unless the critical length of a decasaccharide is reached. Yet, longer NS oligosaccharides are very rare in intestinal HS (13), thus in accord with the overall weaker binding of intestinal HS chains to HRGP335. Apparently, domain organization rather than total sulfate contents of HS chains is the critical determinant in HRGP binding.  . HRGP330 binds to endothelial cells in a HS-dependent manner. TIME cells were incubated with 100 ng/ml biotinylated HRGP330 alone (Ϫ), after pretreatment of the cells with heparin lyases (heparinase and heparitinase; Lyase), after pre-incubation of BT-HRGP330 with heparin, or in the presence of a 10-fold molar excess of either unlabeled HRGP330, HRGP398, or HRGP, as indicated. Bound BT-HRGP330 was detected using streptavidin-coupled Alexa 680 and the Odysseyா infrared imaging system. The results are expressed as relative intensity. Error bars represent one standard deviation.
In a parallel study, we have investigated the anti-angiogenic properties of the synthetic peptides HRGP330, -365, and -398 derived from the His/Pro-rich domain of HRGP. Of these three peptides, only HRGP330 was able to inhibit the chemotaxis of primary endothelial cells. Tumor vascularization in mice was reduced by HRGP330, demonstrating that the peptide also retains anti-angiogenic properties in vivo (8). The inhibitory effect on chemotaxis of HRGP330 is dependent on the presence of Zn 2ϩ . The fact that HRGP330 binds to heparin with a much higher affinity than HRGP365 and HRGP398 (and in a Zn 2ϩ -dependent manner) is a striking correlation. Reduction of the HRGP330 sequence by nine amino acid residues to a 26-amino-acid sequence (HRGP335) did neither impair the heparin-binding capacity nor the ability to inhibit endothelial cell chemotaxis in the presence of Zn 2ϩ . Even shorter peptides derived from HRGP335 are able to inhibit chemotaxis. However, these could not be tested for heparin binding in the nitrocellulose filter trapping assay, because they were too short to be retained by the filter (data not shown).
Zn 2ϩ is a known ligand for HRGP (42) and has recently been proposed to induce a conformational change in HRGP to expose the N-terminal domain for binding to cell surface HS (36). Our results show a direct Zn 2ϩ -dependent binding between the 26-amino-acid residue peptide HRGP335 and HS/heparin. His residues in HRGP335 are likely to chelate Zn 2ϩ to permit interaction with the negatively charged oligosaccharides. More detailed studies using NMR technique are required to understand the structural changes of HRGP335 induced by Zn 2ϩ and other divalent cations, forming a heparin-binding motif.
Several pro-angiogenic factors, such as VEGF and FGF-2, bind HS. One possible mechanism for the action of HRGP could thus be competition with pro-angiogenic factors for HS. However, we have previously demonstrated that HRGP and HRGP330 do not interfere with FGF-2induced activation of FGF receptor 1 (5) or VEGF-A-induced activation of VEGF receptor 2 (8) in cultured endothelial cells. Instead, the antiangiogenic effect is mediated by interference with focal adhesion function, leading to decreased adhesion and migration of endothelial cells (5,8). The addition of HRGP/HRGP330 alters the distribution of focal adhesion complexes in endothelial cells and affects signaling events in the focal adhesions (5,8). Saturation of the HS-binding sites in HRGP330 by pre-incubation with an excess of heparin interfered with the characteristic effect on focal adhesion distribution induced by HRGP330, indicating a role for HS on the endothelial cell surface in the mechanism of action of HRGP/HRGP330. Focal adhesion complexes are highly dynamic structures that mediate contact between surrounding substrate and the cytoskeleton (43,44). Notably, besides extracellular matrix proteins, integrins, and intracellular connector proteins, such as paxillin (45), focal adhesion sites may also contain HS proteoglycans, such as syndecan-4 (46).
In summary, the anti-angiogenic effect of HRGP-and HRGP-derived peptides requires interaction with HS/HS proteoglycan. However, the receptor mechanism may be complex and involve several interactions, for instance with integrins. The analysis of the molecular details of HRGP binding to HS will improve our understanding of its mechanism of action and may aid the development of proper glycomimetics to interfere with angiogenesis in tumor growth.