Identification and characterization of heparin/heparan sulfate binding domains of the endoglycosidase heparanase.

The endo-beta-glucuronidase, heparanase, is an enzyme that cleaves heparan sulfate at specific intra-chain sites, yielding heparan sulfate fragments with appreciable size and biological activities. Heparanase activity has been traditionally correlated with cell invasion associated with cancer metastasis, angiogenesis, and inflammation. In addition, heparanase up-regulation has been documented in a variety of primary human tumors, correlating with increased vascular density and poor postoperative survival, suggesting that heparanase may be considered as a target for anticancer drugs. In an attempt to identify the protein motif that would serve as a target for the development of heparanase inhibitors, we looked for protein domains that mediate the interaction of heparanase with its heparan sulfate substrate. We have identified three potential heparin binding domains and provided evidence that one of these is mapped at the N terminus of the 50-kDa active heparanase subunit. A peptide corresponding to this region (Lys(158)-Asp(171)) physically associates with heparin and heparan sulfate. Moreover, the peptide inhibited heparanase enzymatic activity in a dose-responsive manner, presumably through competition with the heparan sulfate substrate. Furthermore, antibodies directed to this region inhibited heparanase activity, and a deletion construct lacking this domain exhibited no enzymatic activity. NMR titration experiments confirmed residues Lys(158)-Asn(162) as amino acids that firmly bound heparin. Deletion of a second heparin binding domain sequence (Gln(270)-Lys(280)) yielded an inactive enzyme that failed to interact with cell surface heparan sulfate and hence accumulated in the culture medium of transfected HEK 293 cells to exceptionally high levels. The two heparin/heparan sulfate recognition domains are potentially attractive targets for the development of heparanase inhibitors.

Heparan-sulfate proteoglycans (HSPGs) 1 are members of the glycosaminoglycan family, a class of molecules that consists of unbranched, repeated disaccharide units attached to a core protein. Proteoglycans are present essentially in every tissue compartment, localized in the extracellular matrix (ECM), on the cell surface, intracellularly in granules, and even in the nucleus (1)(2)(3). Virtually all cells express at least one proteoglycan on their surface. Membrane-associated proteoglycans are mostly HS that can be either transmembrane (syndecan) or glycosylphosphatidylinositol-anchored (glypican). From mice to worms, embryos that lack HS die during gastrulation (3), suggesting a critical developmental role for HSPGs. HSPGs play key roles in numerous biological settings, including cytoskeleton organization, cell/cell, and cell/ECM interactions (4 -6).
For biological function, HSPGs exert their multiple functional repertoire via several distinct mechanisms that combine structural, biochemical, and regulatory aspects. By interacting with other macromolecules such as laminin, fibronectin, and collagen IV, HSPGs contribute to the structural integrity, self-assembly, and insolubility of the ECM and basement membrane. ECM components are, however, only one class of HSPG-binding proteins. In fact, numerous enzymes, growth factors, cytokines, and chemokines are sequestered by HSPGs on the cell surface and ECM, most often as an inactive reservoir (7,8). Cleavage of HSPGs would ultimately release these polypeptides and convert them into bioactive mediators, thus ensuring rapid tissue response to local environmental alterations. The protein core of HSPGs is susceptible to cleavage by several classes of proteases (9 -11). A more delicate way to modify HSPGs is provided by the endo-␤-glucuronidase, heparanase, an enzyme that cleaves HS at specific intra-chain sites, yielding HS fragments with appreciable size and biological activities (12)(13)(14)(15).
Heparanase activity has been correlated traditionally with cell invasion associated with cancer metastasis, angiogenesis, and inflammation (16 -19). A proof of concept for this notion has been provided recently by applying short interfering RNA and ribozyme technologies (20). In addition, heparanase upregulation has been documented in a variety of primary human tumors correlating with increased vascular density and poor postoperative survival, in some cases (21)(22)(23)(24)(25). Heparanase overexpression has also been noted in several pathologies other than cancer and inflammation (26 -29), suggesting a broader pathological repertoire than originally thought. The heparanase cDNA encodes for a protein of 543 amino acids that undergoes proteolytic processing at two potential cleavage sites, Glu 109 -Ser 110 and Gln 157 -Lys 158 , yielding an 8-kDa polypeptide at the N terminus and a 50-kDa polypeptide at the C terminus that heterodimerize to form an active heparanase enzyme (30 -33).
The present study was undertaken to identify functional domains that would serve as targets for drug development. Based on published consensus sequences that mediate the interaction between polypeptides and heparin, we have identified three potential heparin binding domains mapped at Lys 158 -Asp 171 , Gln 270 -Lys 280 , and Lys 411 -Arg 432 of the 50-kDa heparanase subunit. Here we provide evidence that the Lys 158 -Asp 171 N-terminal peptide physically associates with heparin and HS more strongly than the other two peptides. Moreover, the peptide inhibits heparanase enzymatic activity in a dose-responsive manner, presumably through competition with the HS substrate. Furthermore, antibodies directed to this region inhibit heparanase activity (34), and a deletion construct lacking this domain exhibits no enzymatic activity. NMR titration experiments performed with a synthetic pentasaccharide confirmed residues Lys 158 -Asn 162 as the amino acids important for heparin binding, indicating this recognition domain as a potentially attractive target for the development of heparanase inhibitors.
Expression Vectors-The forward primers (8F and 50F) contained an inserted EcoRI restriction site, and the reverse primer (50R) contained an XhoI restriction site that enabled cloning in-frame into the pSecTag2A vector (Invitrogen). Proteins cloned in this vector contained c-Myc and His 6 tags in their C terminus. Following PCR with a proofreading enzyme (Pfu, Promega, Madison, WI), the vector and the constructs were digested with EcoRI and XhoI and ligated with T4 ligase. The clones were propagated in DH5␣ Escherichia coli strain. DNA sequencing of the constructs was performed using vector-and construct-specific primers.
Cell Lines and Transfection-HEK 293 and B16 melanoma cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, glutamine, pyruvate, and antibiotics. Transient or stable transfection was performed using FuGENE reagent, according to the manufacturer instructions (Roche Applied Science). 48 h following transfection, cells were harvested, lysed, and analyzed by immunoblotting and heparanase enzyme activity assays or were subjected to selection with 500 g/ml Zeocin (Invitrogen).  Stable transfectant pools were obtained after 2-3 weeks and used for  further experiments. Immunoblotting, Metabolic Labeling, and Immunoprecipitation-Cell extracts were prepared using a lysis buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% Triton X-100, and a mixture of protease inhibitors (Roche Applied Science). Protein concentration was determined (Bradford reagent, Bio-Rad), and 30 g of protein were resolved by SDS-PAGE under reducing conditions. After electrophoresis, proteins were transferred to polyvinylidene difluoride membrane (Bio-Rad) and probed with the appropriate antibody followed by horseradish peroxidase-conjugated secondary antibody (Jackson ImmunoResearch, West Grove, PA) and an enhanced chemiluminescent substrate (Pierce). Metabolic labeling was performed essentially as described (34). Briefly, confluent cell cultures were methionine-starved for 30 min prior to the addition of 150 Ci/ml [ 35 S]methionine (Amersham Biosciences) and pulsed for 20 min. For immunoprecipitation, equal volumes (0.1 ml) or equal trichloroacetic acid-precipitable counts/min of lysate samples were brought to a volume of 1 ml with 50 mM Tris-HCl, pH 7.4, 5 mM EDTA, 150 mM NaCl, and 0.5% Nonidet P-40 (buffer A) and incubated with anti-Myc tag antibody for 2 h at 4°C. Protein A/G-Sepharose beads (Santa Cruz Biotechnology) were then added for an additional 30 min. Beads were collected by centrifugation and washed three times with buffer A supplemented with 300 mM NaCl and 5% sucrose and once again with buffer A. Sample buffer was then added, and after boiling at 100°C for 5 min, samples were subjected to electrophoresis as described above. Gels were fixed (30 min, 25% isopropyl alcohol ϩ 10% acetic acid) and fluorographed (30 min, Amplify, Amersham Biosciences) following drying and autoradiography.
Heparanase Activity Assay-Preparation of ECM-coated 35-mm dishes and determination of heparanase activity were performed as described in detail elsewhere (14,31). For inhibition studies, heparanase (20 ng) was incubated (2 h, 37°C) with 35 S-labeled ECM in the presence of the indicated concentration of peptides. For heparanase inhibition studies with intact cells, B16 melanoma cells (2 ϫ 10 6 ) were resuspended in RPMI medium and incubated (18 h, 37°C) with 35 Slabeled ECM in the presence of the indicated peptide concentration. To evaluate heparanase activity in cell extracts, heparanasetransfected 293 cells (1 ϫ 10 6 ) expressing the WT or deletion constructs were lysed by three freeze/thaw cycles, and the resulting cell extracts were incubated (18 h, 37°C) with 35 S-labeled ECM. The incubation medium (1 ml) containing sulfate-labeled degradation fragments was subjected to gel filtration on a Sepharose CL-6B column. Fractions (0.2 ml) were eluted with PBS, and their radioactivity was counted in a ␤-scintillation counter. Degradation fragments of HS side chains are eluted at 0.5 Ͻ K av Ͻ 0.8 (peak II, fractions 15-30) and represent heparanase degradation products. Nearly intact HSPGs are eluted just after the V 0 (K av Ͻ 0.2, peak I, fractions 3-15) (14). These high molecular weight products are released by proteases that cleave the HSPG core protein.
Heparin/HS Binding-Peptides (50 M) were incubated (2 h, 4°C) with heparin/HS-Sepharose beads in PBS, washed with PBS supplemented with NaCl to a final concentration of 0.35 M, followed by one wash with PBS. Dye-free sample buffer was added, and the beads were boiled for 5 min and centrifuged, and the supernatants were loaded on Tris-Tricine gel. Subsequently, gels were stained with Coomassie Blue to visualize bound peptides. For inhibition of heparanase binding to heparin, 20 ng of heparanase protein were incubated with heparin-Sepharose beads in the presence of increasing concentrations of the peptide of interest or its scrambled control peptide, followed by two washes with PBS supplemented with 0.6 M NaCl. Sample buffer was then added, and the mixture was boiled for 5 min and centrifuged, and the supernatant was subjected to immunoblotting with anti-heparanase antibodies. Extracellular accumulation of heparanase in the presence of heparin was examined as described (41). Briefly, stable transfected 293 cells expressing the 65-kDa WT, point-mutated, or the deletion (⌬10, ⌬15, and ⌬20) variants were incubated (18 h, 37°C) with heparin, HS, and hyaluronic acid (HA) or glycol-split modified heparin under serum-free conditions. Medium was then collected and subjected to immunoblotting without or with prior precipitation with 10% trichloroacetic acid. Cell extracts were prepared, and 30 g of protein were similarly analyzed.
Heparanase Uptake-Uptake experiments were carried out essentially as described (34,41). Briefly, CHO K1 cells were incubated with 1 g of 65-kDa WT or the 65⌬10 heparanase for the indicated time under serum-free conditions. At each time point the medium was aspirated, cells were washed twice with ice-cold PBS, and cell extracts were analyzed by immunoblotting as described above. For uptake inhibition studies, heparanase was iodinated to a high specific activity by the chloramine T (Sigma) method. Iodinated heparanase was added to CHO cells together with the Lys 158 -Asp 171 (KKDC) peptide or its scrambled control peptide. Following incubation (1 h, 4°C), cells were washed, lysed, and counted with ␥-counter.
NMR Spectra-NMR samples of the unlabeled Lys 158 -Asp 171 peptide (KKDC) and the corresponding scrambled sequence were dissolved in 20 mM sodium phosphate buffer, pH 5.8, supplemented with 0.15 M NaCl and 10% D 2 O. All NMR spectra were recorded on a Bruker Avance 600 spectrometer equipped with a quadruple resonance proton-carbon-nitrogen-deuterium 5-mm probe (Bruker TCI cryo-probe®), with cooled coil and preamplifier and using a selfshielded z gradient coil. Spectra were acquired nonspinning at temperatures of 295 K. For two-dimensional homonuclear 1 H experiments, DQF-COSY, TOCSY, and ROESY were performed according to the Bruker library pulse sequences. For ROESY, mixing times of 200 ms and a spin-lock of 150 ms obtained by continuous irradiation were used. COSY/TOCSY and ROESY experiments were acquired using 8 and 16 scans per series of 1Kx512W data points, respectively. Data were zero filled in F1, and a shifted squared cosine function was applied prior to Fourier transformation. Water suppression was carried out using excitation sculpting sequence with gradients (42). Two-dimensional 1 H-15 N HSQC experiments were acquired with 96 scans for each of 320 increments on F1 dimension. The matrix size 1Kx512 was zero filled to 4Kx2K by application of a squared cosine function prior to Fourier transformation.

RESULTS
Identification of Heparin Binding domains-Sequence alignment of heparin binding domains from several proteins had led to the characterization of two consensus sequences, XBBXBX and XBBBXXBX, where B is basic and X is hydropathic (neutral and hydrophobic) amino acid (43). Analysis of the primary protein sequence of heparanase revealed the existence of two domains that match the consensus sequence for heparin binding (Fig. 1). These are mapped at Lys 158 -Asp 162 (KKFKN), which is the N terminus region of the 50-kDa heparanase subunit, and at Pro 271 -Met 278 (PRRKTAKM) (the boldface letters represent basic amino acids that comprise a consensus sequence for heparin binding) (Fig. 1). A third domain contains two clusters of basic amino acids in tandem (Lys 411 -Lys 417 and Lys 427 -Arg 432 ) ( Fig. 1) and is considered as a potential heparin binding domain as well.
The KKDC (Lys 158 -Asp 171 ) Peptide Physically Interacts with HS-In order to study the relevance of the above sequences for heparanase/HS interaction, we synthesized three peptides that contained the putative heparin binding domain ( Fig. 2A). Peptides (50 M), or their control scrambled (Scr) counterparts, were incubated with heparin-Sepharose beads (2 h, 4°C) and washed with 0.35 M NaCl, and the bound peptides were visualized by Tris-Tricine gel electrophoresis and Coomassie Blue staining. The KKDC peptide was found to associate specifically with the heparin beads, whereas no such binding was observed with its control scrambled peptide (Fig. 2B, heparin). In contrast, peptides KKLR (Fig. 2B) and QPLK (data not shown) exhibited only very weak interaction with heparin. As the QPLK peptide is shorter and spans only 10 amino acids, we synthesized peptides containing additional amino acids, N-(KLYGPDVGQPRRKTAKMLK) or C-terminal (QPRRKTAK-MLKSFLKA, QPKA) (the boldface letters represent amino acids added to the N and C terminus core peptide QPLK in Fig.  2A) to the core 10-amino acid QPLK sequence. The latter 16 amino acids peptide exhibited low but detectable interaction with heparin (Fig. 2B). Addition of the C-terminal cysteine residue shown to promote peptide dimerization and to improve the interaction with heparin (44) only slightly enhanced heparin binding (Fig. 2B, QPKAC). These results suggest that only the KKDC peptide exhibits high affinity to and physically interacts with heparin. Because heparin contains a higher number of sulfate groups and iduronic acid residues and its structure does not fully reflect the organization of HS, it was crucial to examine further the ability of the KKDC peptide to interact with HS. As demonstrated in Fig. 2B, the KKDC peptide specifically binds to HS-Sepharose beads, indicating that the peptide is capable of interacting with both heparin and HS to a similar extent.
The KKDC Peptide Inhibits Heparanase Enzymatic Activity-Giving the specific interaction of the KKDC peptide with HS, we hypothesized that it may compete with heparanase for interaction with HSPGs and thus inhibit its enzymatic activity. In order to test this possibility, the KKDC, or control scrambled (Scr) peptides, were applied onto 35 S-labeled HS-containing ECM substrate and incubated (2 h, 37°C) with active mouse (mHpa, Fig.  2C) or human (hHpa, Fig. 2D mouse and human heparanase in a dose-dependent manner, whereas the control scrambled peptide did not. Moreover, the peptide efficiently neutralized endogenous heparanase enzymatic activity in B16 melanoma cells (Fig. 2E). In contrast, the QPLK and KKLR peptides had no such inhibitory effect on heparanase activity (data not shown), correlating with the very low heparin binding abilities of these peptides (Fig. 2B, heparin).
The KKDC Peptide Interferes with Heparanase Binding to HS-Next, we evaluated the ability of the KKDC peptide to interfere with heparanase binding to HS. For this purpose, active heparanase (20 ng), which exhibits a high affinity toward its HS substrate, was incubated (2 h, 4°C) with HS-Sepharose beads in the absence or presence of increasing concentrations of the KKDC or control scrambled peptides (Fig. 2F). Following washes (0.6 M NaCl), bound heparanase was analyzed by immunoblotting with anti-heparanase antibodies. Heparanase exhibits a high affinity toward HS, and the bound protein was readily detected in the absence of competing peptide (0 M). The KKDC peptide at a concentration of 25 M or higher markedly inhibited the association of heparanase with HS (Fig. 2F, KKDC), whereas the scrambled peptide had no such effect at the same concentrations (Fig.  2F, sc). Previously, we have demonstrated that binding of the 65-kDa latent heparanase to cell surface HS is followed by uptake and processing into a 50-kDa active enzyme (41). In order to verify the possibility that the KKDC peptide will inhibit heparanase binding to cell surface HS, we employed 125 I-labeled heparanase. We first confirmed that heparanase iodination did not harm its ability to interact with heparin/ HS. To this end, iodinated heparanase was added to CHO cells in the absence or presence of the indicated heparin concentrations (g/ml). As shown in Fig. 2G, binding of iodinated heparanase to cell surface HS was competed almost completely with 5 g/ml heparin. Thus, iodination did not affect heparanase ability to interact with HS. Similarly, iodinated heparanase was added to CHO cells in the absence or presence of the indicated concentrations (micromolar) of KKDC or its control scrambled peptide. As demonstrated in Fig. 2H, the KKDC peptide at concentrations higher than 25 M significantly inhibited the binding of iodinated heparanase to cell surface HS, similar to the inhibitory effect of heparin (Fig. 2G). These results suggest that the ability of the KKDC peptide to inhibit heparanase enzymatic activity (Fig. 2, C-E) is due primarily to competition with the HS substrate.
Deletion Mutants Lacking Heparin Binding Domains Exhibit No Enzymatic Activity-In order to study further the importance of the sequences identified to contain the heparin-binding motif, we have undertaken a different approach and created deletion mutants lacking the relevant sequences (Fig. 3). Deletion mutants were created for each of the three regions in the full-length 65-kDa heparanase, and the gene constructs were stably transfected into HEK 293 cells. 293 cells transfected with the wild type (WT) construct exhibited high levels of heparanase activity (Fig. 3A, 65WT). By contrast, no enzymatic activity of heparanase was detected in each of the deletion mutants (Fig. 3A, 65⌬15, 65⌬10, and 65⌬20). Cell lysates were next subjected to immunoblot analysis with antiheparanase antibodies. High levels of the 50-kDa heparanase protein were detected in 293 cells transfected with the fulllength WT heparanase gene construct (Fig. 3B, 65WT, Lysate), in agreement with high levels of heparanase activity in these cells (Fig. 3A, 65WT). Most interestingly, however, the 50-kDa heparanase protein was not detected in lysates of cells transfected with the deletion constructs (Fig. 3B, 65⌬15; 65⌬10; 65⌬20, Lysate). Instead, the 65-kDa form of heparanase was highly abundant, suggesting that the protein failed to be processed into the 50-and 8-kDa active heterodimer form (Fig. 3B, Lysate), which explains the lack of enzymatic activity in cells transfected with these deletion constructs (Fig. 3A).
QPLK Deletion Mutant Accumulates in the Culture Medium of Stably Transfected HEK 293 Cells-Next, we evaluated heparanase secretion by exposing the cell conditioned medium (CM) to immunoblot analysis with anti-heparanase (Fig. 3B,  upper panel) and anti-Myc tag antibodies (lower panel). Low levels of the 65-kDa heparanase were detected in the CM of cells transfected with the WT construct (Fig. 3B, 65WT, Me- dium), in agreement with the notion that heparanase is subjected to rapid uptake mediated by cell surface HSPGs and thus does not accumulate to high levels extracellularly (41). By contrast, exceptionally high levels of heparanase secretion were detected in the CM of cells transfected with the 65⌬10 gene construct (Fig. 3B, 65⌬10, Medium), whereas cells transfected with the 65⌬15 and 65⌬20 variants did not secrete detectable amounts of heparanase (Fig. 3B, Medium). The lack of heparanase activity in cells transfected with the deletion constructs is most likely due to lack of heparanase processing which, in turn, seems to result from a defect in heparanase secretion (65⌬15; 65⌬20) and uptake (65⌬10) (see below). Moreover, we confirmed similar synthesis levels of heparanase by employing metabolic labeling and immunoprecipitation analysis (Fig. 3B, right), indicating that different secretion and activity levels are not due to differences in heparanase synthesis. In order to overcome the lack of processing, deletion mutants were similarly generated in the 50-kDa heparanase subunit.
Deletion Mutants Lacking the KKDC or QPLK Sequences Are Capable of Heterodimer Formation-We have demonstrated previously that the 50-kDa heparanase protein lacks enzymatic activity and that co-expression of the 50-and 8-kDa subunits is necessary and sufficient for heparanase enzymatic activity (31). Indeed, co-transfection of the 50-and 8-kDa proteins into 293 cells resulted in a high level of heparanase activity (Fig.  3C, 8ϩ50). In striking contrast, co-transfection of the 8-kDa subunit with the 50-kDa deletion mutants yielded no enzymatic activity (Fig. 3C, 50⌬15, 50⌬10, and 50⌬20). Given the necessity of heterodimer formation between the 50-and 8-kDa subunits for obtaining active heparanase, it is conceivable that the deletions harmed regions that are required for this interaction. Subjecting total cell lysates to immunoblot analysis revealed similar expression levels of the 50- (Fig. 3D, upper panel) and 8-kDa (Fig. 3D, middle panel) subunits in the transfected cell lines. Next, lysate samples were immunoprecipitated with antibody (810) directed against the 8-kDa subunit followed by immunoblotting with anti-Myc tag antibodies that recognize the 50-kDa protein. Both the 50⌬15 and 50⌬10 constructs appeared to be associated with the 8-kDa subunit to levels comparable or even higher than the control 50-kDa protein (Fig. 3D, lower panel). Thus, the lack of enzymatic activity in these co-transfectants is most probably due to a reduced affinity of the heparanase protein to the HS substrate. By contrast, the 50⌬20 failed to interact with the 8-kDa subunit (Fig. 3D, 8ϩ50⌬20), suggesting that this region may be involved in heterodimer formation, resulting in a lack of enzymatic activity in this co-transfection.
QPLK Deletion Mutant Fails to Bind Cell Surface HS-Studying the biosynthesis and trafficking route of heparanase, we have demonstrated recently that secretion of the 65-kDa heparanase precursor is followed by a rapid uptake mediated by cell surface HSPGs, which can be competed by soluble heparin. This efficient uptake mechanism maintains extracellular levels of heparanase, tightly regulated (41). Accumulation of the 65⌬10 protein in the culture medium to exceptionally high levels (Fig. 3B, Medium) may indicate that this protein variant was not subjected to cellular uptake. In order to explore this possibility, 293 cells stably transfected with the different deletion mutants were grown in the absence or presence of heparin, HS, or HA, and lysate (Fig. 4A, lower panels) and medium (Fig.  4A, upper panels) samples were analyzed by immunoblotting. Only low levels of heparanase were detected in the CM of cells expressing the WT 65-kDa protein in the absence of a competitor (Fig. 4A, 65WT, Con). A marked increase in heparanase accumulation in the CM was evident in the presence of heparin or HS, whereas HA had no such effect. These results indicate that sulfation is absolutely necessary for the interaction of heparanase with heparin/HS, in agreement with our previous findings (41). Only marginal accumulation of the 65⌬15 heparanase was detected in the presence of heparin or HS (Fig.  4A, 65⌬15), supporting the critical role of this protein domain in the interaction with heparin/HS. Furthermore, the 65⌬10 heparanase did not respond to heparin or HS, and the protein levels in the CM appeared unchanged (Fig. 4A, 65⌬10), suggesting that the 65⌬10 heparanase mutant lost its ability to bind heparin or HS. Similar findings were obtained by employing metabolic labeling and immunoprecipitation analysis (data not shown). In contrast, the 65⌬20 heparanase was noted to accumulate to high levels in the presence of heparin or HS (Fig.  4A, 65⌬20), clearly indicating that this domain was not involved in heparin/HS interaction. Most interestingly, the 65⌬20 protein was not detected upon re-blotting the membrane with anti-Myc tag antibodies (data not shown), suggesting that this protein variant lost its tag prior to its secretion (see also Fig. 3B, lower panel). FIG. 4. The 65⌬15 and 65⌬10 deletion mutants fail to interact with heparin/HS. A, 293 cells stable transfected with the WT 65-kDa heparanase or its deletion mutants were left untreated (Con) or incubated with heparin (Hep), HS, or HA, all at 10 g/ml. After 18 h, medium (Med., upper panels) and lysate (Lys., lower panels) samples were subjected to immunoblotting with anti-Myc antibody. Medium samples from the ⌬20 transfected cells failed to react with the anti-Myc tag antibody, and the blot was reprobed with anti-heparanase 1453 antibody (upper panel, right). Note a robust accumulation of the WT but not the ⌬15 or ⌬10 heparanase mutants, in the presence of heparin or HS. B, uptake of the WT and ⌬10-kDa heparanase proteins. Purified 65-kDa WT or 65⌬10-kDa proteins were added to the culture medium of CHO K1 cells for the times indicated. Heparanase uptake was examined by immunoblotting with anti-heparanase (1453) antibodies (ns ϭ nonspecific protein band reacting with the 1453 antibodies). Note that the 65⌬10 heparanase is not subjected to cellular uptake. C, point mutations. Heparanase was mutated at lysine 158 or both lysine 158 and 161. Stable transfected 293 cells were left untreated (Con) or incubated with HS or glycol-split modified heparin (GS), both at 10 g/ml. After 18 h, equal volumes of medium were immunoblotted with anti-heparanase antibodies.
QPLK Deletion Mutant Is Not Subjected to Cellular Uptake-Next, the 65-kDa WT and the 65⌬10 mutant proteins were purified from the CM and exogenously added to CHO K1 cells for the indicated time, and protein uptake was examined by immunoblotting with anti-heparanase antibodies (Fig. 4B). Uptake of the WT 65-kDa heparanase was noted already 30 min after its addition, followed by processing into the 50-kDa protein that continued to accumulate by 2 and 4 h (Fig. 4B, WT). In striking contrast, uptake of the 65⌬10 mutant heparanase could not be detected at all, even at 4 h of incubation (Fig. 4B,  65⌬10), supporting the notion that the QPLK sequence is critically important for the interaction of heparanase with heparin/HS.
Lysine Residues 158 and 161 Are Critically Important for Heparanase/HS Interaction-Unlike 65⌬10, the 65⌬15 protein exhibited very low secretion levels (Figs. 3B and 4A) that may result from the presence of a glycosylation site (Asn 162 ) in the KKDC sequence, a site that has been shown to be important for heparanase secretion (45). In order to verify this possibility and to better define basic amino acids within the KKDC sequence that are responsible for HS binding, we point-mutated lysine 158 or both lysine 158 and lysine 161 to alanine. Stable transfected 293 cells were then examined for their secretion levels. In addition, we examined heparanase accumulation in response to the addition of HS or glycol-split (GS)-modified heparin, a most potent heparanase inhibitor (46). Only low levels of heparanase were detected in the culture medium of WT heparanase-transfected cells as opposed to vigorous accumulation noted in the presence of HS-or GS-modified heparin (Fig. 4C, WT). Secretion of the K158A point mutant was similar to that of the WT heparanase in the absence of competitor. In contrast, no further accumulation was observed upon addition of HS, whereas GS-modified heparin yielded a 2-fold increase in heparanase accumulation (Fig. 4C, K158A), suggesting that the GS heparin has a higher affinity toward heparanase than HS. Moreover, mutating both lysine 158 and 161 abolished the ability of heparanase to interact with HS and even with the GS-modified heparin (Fig. 4C, KK:AA), pointing to these two basic amino acid residues as critically important for heparanase interaction with HS. These results further suggest that the low secretion levels of the 65⌬15 protein is in fact because of the lack of Asn 162 glycosylation sites.
A Specific Region of the KKDC Peptide (Lys 158 -Asp 171 ) Physically Interacts with AGA*IA Pentasaccharide-In order to define further the amino acids important for heparin/HS recognition, the KKDC or its scrambled control peptide were subjected to NMR analysis. The backbone NMR resonances of the peptide and scrambled control sequence were assigned for all amino acid residues and side chains by analysis of the COSY, TOCSY, and ROESY spectra (data not shown). To identify the heparin-binding site of the peptide KKDC, we studied the chemical shifts perturbation of NMR resonances in the presence of the pentasaccharide AGA*IA (asterisk indicates a trisulfated saccharide). This synthetic pentasaccharide mimics the high affinity binding of heparin/HS for antithrombin and contains the minimal sequence cleavable by heparanase (47,48). Moreover, AGA*IA is susceptible to limited cleavage by heparanase 2 and is thus most relevant and suitable for NMR studies because of its low molecular weight. The chemical shift perturbation of H N and N resonances of KKDC induced by addition of AGA*IA are shown in Fig. 5A. Most chemical shift changes were the continuous function of the amount of added pentasaccharide, indicating a fast exchange regime on the NMR time scale. The residues showing the largest shift are Phe 160 -Lys 161 -Asn 162 located in the N-terminal region of the peptide. Moreover, signals of Lys 158 and Lys 159 , not detectable in the free peptide, appeared upon binding to AGA*IA with larger shift changes for increasing addition of the ligand. During titration their line broadening decreased with a consequent increase in signal intensity because of the slower exchange of NH proton with the solvent. The interaction of AGA*IA with the peptide reached saturation at the stoichiometric ratio of 4:1. In order to verify the specificity of the binding, the same experiment was repeated with the scrambled control peptide (Fig. 5B). Although some resonance perturbations could be observed after addition of the AGA*IA pentasaccharide, their random occurrence and their small signal shifts are indicative of nonspecific interaction. Clearly, although for the scrambled peptide no linkage region can be identified, the Lys 158 -Lys 159 -Phe 160 -Lys 161 -Asn 162 N-terminal sequence of the KKDC peptide appears to mediate high affinity binding to the AGA*IA pentasaccharide.

Heparanase as a Target for the Development of Anticancer
Drugs-The mammalian endoglycosidase heparanase is the predominant enzyme responsible for degradation of HS, an activity that is thought to play a decisive role in cellular invasion associated with cancer metastasis, angiogenesis, and inflammation. Recently, this notion gained further support by employing specific anti-heparanase short interfering RNA and ribozyme strategies (20), providing a proof-of-concept for the pro-metastatic and pro-angiogenic functions of heparanase. Moreover, heparanase up-regulation has been documented in an increasing number of primary human tumors, including cancers of the bladder, colon, stomach, breast, pancreas, esophagus, multiple myeloma, and acute myeloid leukemia (20,49). These findings and the occurrence of a single functional heparanase enzyme position heparanase as an attractive target for the development of anticancer drugs. Currently available heparanase inhibitors are the various sulfated poly-and oligosaccharides such as modified species of heparin, laminaran sulfate, and PI-88 (33,50,51). These compounds were shown to inhibit heparanase activity and exert anti-metastatic and anti-angiogenic effects (52)(53)(54). Nevertheless, the lack of specificity makes interpretation questionable when using these and other polysulfated reagents (52,53).
Heparin Binding Domains as Valid Targets for the Development of Heparanase Inhibitors-In an attempt to apply a more rational approach, we sought for functional domains that would serve as targets for drug development. Most interestingly, functional domains other than the basic heterodimer structure (30 -32) and amino acids (Glu 225 and Glu 343 ) critical for the enzyme catalytic activity (55) have not been elucidated so far in the heparanase protein. In this study, we have identified and characterized two protein domains that mediate the interaction of heparanase with its HS substrate. A third domain (Lys 411 -Arg 432 ) seems not to be involved in heparin/HS binding but rather to mediate the interaction of the 50-and 8-kDa subunits to establish an active heterodimer (Fig. 3D). The KKDC peptide exhibits high affinity and physically interacts with immobilized heparin and HS. Moreover, the peptide was able to inhibit significantly the heparanase enzymatic activity (Fig. 2, C-E), an effect attributed to inhibition of heparanase interaction with the heparin/HS substrate (Fig.  2F). This aspect was further confirmed by inhibition of iodinated heparanase binding to CHO cells by the KKDC peptide (Fig. 2H). Thus, the KKDC peptide not only inhibits heparanase interaction with heparin-Sepharose or ECM deposited by cells in vitro but also inhibits the interaction of heparanase with cell surface HS. Site-directed mutagenesis (Fig. 4C) and NMR studies (Fig. 5) clearly identified Lys 158 , Lys 159 , and Lys 161 as important mediators of the heparanase-heparin/HS interaction, as predicted. Most interestingly, NMR studies have also identified Phe 160 and Asp 162 as amino acids residues important for this interaction, suggesting that nonbasic amino acids may be involved in heparin/HS binding. In agreement with these biochemical analyses, deletion of the KKDC sequence resulted in a loss of enzymatic activity (Fig. 3, A and C) without affecting proper heterodimer formation (Fig. 3D), suggesting that the loss of activity is due primarily to reduced affinity of heparanase to the HS substrate. We have demonstrated previously that a polyclonal antibody raised against the KKDC peptide inhibits heparanase enzymatic activity (34), indicating that this protein domain is a promising target for the development of neutralizing monoclonal antibodies. By contrast, the QPLK peptide exhibited low affinity to heparin or HS (Fig. 2B), and it did not inhibit heparanase enzymatic activity (data not shown). Synthesizing peptides containing additional amino acids, N-(KLYGPDVGQPRRKTAKMLK) or C-terminal (QPRRKTAKMLKSFLKA) to the core 10-amino acid QPLK sequence only slightly improved its interaction with heparin (Fig. 2B). Nevertheless, deletion of this protein domain resulted in a loss of heparanase enzymatic activity (Fig. 3A), although heterodimer formation appeared intact (Fig. 3D), and accumulation of the protein in the conditioned medium reached exceptionally high levels (Fig. 3B). We have demonstrated recently that heparanase does not normally accumulate extracellularly due to rapid and efficient cellular uptake mediated by cell surface HS (41). The high levels of the 65⌬10 protein detected in the CM (Fig. 3B) therefore suggest that this protein variant failed to interact with HS on the cell surface. Indeed, unlike the WT heparanase, addition of heparin or HS to the CM had no effect on the levels of this protein (Fig. 4A), and the addition of purified 65⌬10 protein to CHO cells resulted in no detectable uptake (Fig. 4B). Thus, deletion of one of the HS binding domains (65⌬15 or 65⌬10) resulted in practically complete loss of the ability of heparanase to interact with HS. This situation may imply that the two domains cooperate with one another to establish a single functional binding domain.
Predicted Three-dimensional Heparanase Model-A crystal structure of heparanase has not been resolved yet. Thus, localization of protein domains in a three-dimensional context cannot be performed. In order to overcome the lack of authentic three-dimensional structure, we adopted a predicted model based upon heparanase resemblance to ␤-D-xylosidase from Thermoanaerobacterium saccharolyticum, an approach that has been undertaken recently to investigate the structural requirements for proheparanase processing and activation (56). This xylosidase exhibits the best alignment with heparanase and was crystallized together with its substrate, providing a more reliable structure (57). Most interestingly, the KKDC peptide domain, and specifically Lys 158 and Lys 159 that were identified as residues important for heparin/HS interaction (Figs. 4C and 5), appear to reside in close proximity to amino acids Glu 225 and Glu 343 that comprise the active site (55) in a micro pocket domain (Fig. 6). This predicted three-dimensional model supports the importance of the KKDC sequence for substrate recognition and highlights the significance of this domain as a valid target for the development of heparanase inhibitors, such as neutralizing antibodies (34) and small molecules. Further analysis localizing the QPLK sequence in this three-dimensional model is currently in progress.
In summary, our results characterize, for the first time, functional domains in the endoglucuronidase enzyme, heparanase. Identification of heparin binding domains not only adds to our basic understanding of the biology of this apparently important enzyme but also provides attractive targets for the development of heparanase inhibitors in a rational rather than random fashion. Additional important functional domains are protein sequences that mediate the interaction between the 8and 50-kDa heparanase subunits. Studies aiming at this direction are currently underway.

FIG. 6.
A three-dimensional model of heparanase, highlighting amino acids that participate in heparanase activity and substrate recognition. The heparanase model was predicted based on similarity to ␤-D-xylosidase. Presented are the substrate (red) and Glu 225 and Glu 343 (green) identified as the proton donor and nucleophile residues, respectively, as well as Lys 158 and Lys 159 (blue) that mediate substrate binding.