Structure-based design of a selective heparanase inhibitor as an antimetastatic agent

Heparanase is an endo-β-d-glucuronidase that degrades heparan sulfate glycosaminoglycans (1). Human heparanase cDNA encodes the latent enzyme, which consists of 543 amino acid residues, and becomes the active enzyme of 50 kDa after being processed twice at the amino terminus (2, 3). Because only one heparanase cDNA sequence coding functional enzyme has been identified to date, heparanase has been considered a major enzyme that degrades heparan sulfate glycosaminoglycans in mammalian tissues (4–7). High-level expression of heparanase enzyme has been found in highly invasive normal and malignant cells, such as activated immune cells, lymphoma, melanoma, and carcinoma cells (2, 8–11), as well as in human head and neck tumors (12). Further, heparanase expression has been correlated with the metastatic property of tumor cells (2, 10, 11, 13) and experimental metastasis in an animal model was shown to be reduced by treatment with heparanase inhibitors (14, 15). It has also been shown that patients with a high-level expression of heparanase mRNA tend to have a shorter postoperative survival term (16, 17). Heparanase also enhances angiogenesis by the release of growth factors, such as basic fibroblast growth factor and vascular endothelial growth factor, which are sequestered by heparan sulfate glycosaminoglycans in the extracellular matrix (10, 18–20). From these results, heparanase has been assumed to be a promising target for antitumor agents, especially antimetastatic agents.

Most of the heparanase inhibitors that have been reported are sulfated oligosaccharide derivatives that resemble the substrate of heparan sulfate glycosaminoglycans (15, 21–23). We have established a heparan sulfate degradation assay system to develop novel heparanase inhibitors that are not sulfated oligosaccharide derivatives and have used it to screen 10,000 microbial broths of actinomycetes, fungi, and bacteria (24). As a result, (R)-3-hexadecanoyl-5-hydroxymethyltetronic acid (RK-682; Fig. 1A), which we had purified previously as a potent inhibitor of tyrosine phosphatase (25), was found to possess a potent inhibitory activity for heparanase (24). The IC₅₀ value of RK-682 for heparanase was 17 μmol/L; this is a fairly low value compared with, for example, the IC₅₀ of suramin, a known heparanase inhibitor, which is 46 μmol/L (26).

RK-682 has already been shown to have several inhibitory activities, including phospholipase A2 inhibition (IC₅₀, 16 μmol/L; ref. 27), HIV-1 protease inhibition (IC₅₀, 84 μmol/L; refs. 28, 29), and protein tyrosine phosphatase inhibition (IC₅₀ of CD45, 54 μmol/L; IC₅₀ of vaccinia H1-related phosphatase, 2.0 μmol/L; ref. 25). In the present study, we therefore attempted to develop a heparanase selective inhibitor based on rational drug design using RK-682 as a lead structure and investigated the antitumor activity.

The amino acid sequence alignment of human heparanase and 1,4-β-xylanase from Penicillium simplicissimum (EC 3.2.1.8) was carried out manually using the Homology module of the Discover/InsightII programs (Accelrys, Inc., San Diego, CA). Structurally conserved regions were defined by referring to the results of database searches and site-directed mutagenesis by Hulett et al. (30). Structure construction and molecular dynamics calculations were carried out by the Biopolymer and Discover3 modules of Discover/InsightII program (Accelrys) using the X-ray structure of 1,4-β-xylanase (Protein Data Bank entry 1BG4) as a template (31).

4-Methyl-RK-682 (4-Me-RK-682) was prepared as described previously (32). RK-682 was treated with diazomethane in diethyl ether at 23°C for 20 minutes.

1,2,3-Triisopropyl-isourea was prepared as described previously (33). A solution of RK-682 (30.0 mg, 81.4 μmol) and 1,2,3-triisopropyl-isourea (30.3 mg, 162.8 μmol) in 0.5 mL dry tetrahydrofuran was stirred under reflux for 18 hours. After cooling to room temperature, the solution was concentrated. The residue was purified by silica gel column chromatography (hexane/ethyl acetate, 5:1) and preparative thin-layer chromatography on silica gel (hexane/ethyl acetate, 1:1) to give 4-isopropyl-RK-682 (4-iPr-RK-682, 6.8 mg, 20%): [α]_D²⁴ = 5.76 (c = 0.20, CHCl₃); Fourier transformation-IR (film) 3,440, 2,917, 2,850, 1,722, 1,672, 1,603, 1,466, 1,409, 1,317, 1,097, 1,048 cm⁻¹; ¹H nuclear magnetic resonance (400 MHz, CDCl₃) δ 5.38 (1H, septet, J = 5.6 Hz), 4.70 (1H, triplet, J = 3.2 Hz), 4.05 (1H, multiplet), 3.85 (1H, multiplet), 3.00 (1H, double triplet, J = 18.0, 7.8 Hz), 2.94 (1H, double triplet, J = 18.0, 7.8 Hz), 1.87 (1H, broad), 1.57 (2H, multiplet), 1.25 (22H, multiplet), 0.87 (5H, multiplet); ¹³C nuclear magnetic resonance (100 MHz, CDCl₃) δ 197.7, 177.2, 105.3, 79.6, 78.7, 61.5, 42.5, 31.9, 29.9, 29.68, 29.66, 29.62, 29.52, 29.48, 29.4, 29.2, 27.4, 23.8, 22.7, 22.1, 21.8, 14.1; matrix-associated laser desorption/ionization time of flight mass spectroscopy (positive ion, α-cyano-4-hydroxycynnamic acid) calculated for C₂₄H₄₂O₅Na: (M + Na)⁺ 433.293; found 433.235.

2-Benzyl-1,3-dicyclohexyl-isourea was prepared as described previously (33). A solution of RK-682 (30.0 mg, 81.4 μmol) and 2-benzyl-1,3-dicyclohexyl-isourea (51.2 mg, 162.8 μmol) in 0.5 mL dry tetrahydrofuran was stirred under reflux for 24 hours. After cooling to room temperature, the solution was concentrated. The residue was purified by silica gel column chromatography (hexane/ethyl acetate, 5:1) and preparative thin-layer chromatography on silica gel (hexane/ethyl acetate, 1:1) to give 4-benzyl-RK-682 (4-Bn-RK-682, 20.1 mg, 54%): [α]_D²⁴ = −0.25 (c = 0.75, CHCl₃); Fourier transformation-IR (film) 3,422, 2,918, 2,850, 1,766, 1,607, 1,433, 1,368, 1,329, 1,232, 1,075 cm⁻¹; ¹H nuclear magnetic resonance (400 MHz, CDCl₃) δ 7.41-7.30 (5H, multiplet), 5.50 (1H, doublet, J = 12.0 Hz), 5.34 (1H, doublet, J = 12.0 Hz), 4.78 (1H, triplet, J = 3.2 Hz), 4.08 (1H, broad multiplet), 3.90 (1H, broad multiplet), 2.87 (1H, double triplet, J = 17.8, 7.6 Hz), 2.78 (1H, double triplet, J = 17.8, 7.6 Hz), 2.04 (1H, broad), 1.51 (2H, multiplet), 1.25 (22H, multiplet), 0.86 (5H, multiplet); ¹³C nuclear magnetic resonance (100 MHz, CDCl₃) δ 197.8, 177.5, 170.3, 134.2, 129.1, 128.8, 128.0, 106.9, 78.7, 77.4, 61.2, 42.4, 39.0, 31.9, 29.67, 29.66, 29.65, 29.62, 29.5, 29.45, 29.3, 29.1, 28.0, 27.4, 23.6, 22.69, 22.65, 14.1; matrix-associated laser desorption/ionization time of flight mass spectroscopy (positive ion, α-cyano-4-hydroxycynnamic acid) calculated for C₂₈H₄₂O₅Na: (M + Na)⁺ 481.293; found 481.224.

The heparanase-overexpressing stable clone of HepG2 human hepatocellular carcinoma cell line, HepG2-HP, was cultured in DMEM supplemented with 10% FCS and 400 μg/mL G418 (geneticin, Life Technologies, Grand Island, NY; ref. 34). HepG2 cells, human fibrosarcoma HT1080 cells, and murine B16BL6 melanoma cells were grown in DMEM supplemented with 10% FCS. All cell lines were incubated at 37°C in a humidified atmosphere of 5% CO₂/95% air.

Heparan sulfate was purchased from Seikagaku Corp. (Tokyo, Japan). A mixture of 90 μL HepG2-HP cell lysate (2.0 mg/mL protein) and 10 μL heparan sulfate solution (10 mg/mL in PBS, pH 6.2) with or without compound solution was incubated for 24 hours at 37°C. After adding the 20 μL heparan sulfate sampling solution [glycerol 8 mL, bromophenol blue (5 mg/mL in H₂O) 5 mL, H₂O 19 mL], the reaction mixture (10 μL) was subjected to SDS-PAGE (20%). Electrophoresed gel was soaked in H₂O for 2 hours to remove SDS. The gel was stained with 0.5% Alcian Blue 8GX for 3 hours and destained with a 1:2:7 mixture of acetic acid/ethanol/water for 12 hours (35). The volumes of the bands were measured using a MD Scanning Imager equipped with MD ImageQuant Software version 3.22 (Molecular Dynamics, Inc., Sunnyvale, CA) for quantification.

Assay condition was described previously (32). The mixture of p-nitrophenyl phosphate (10 mmol/L) and vaccinia H1-related phosphatase (0.25 μmol/L), which was overexpressed in Escherichia coli and purified, with or without a compound was incubated at 37°C for 30 minutes and the absorbance was measured at 405 nm.

Glycosidase activities were assessed using the method described previously (36, 37). All enzymes and substrates were purchased from Sigma Chemical Co. (St. Louis, MO). α-Glucosidase (yeast) was incubated at 37°C for 30 minutes in citrate phosphate buffer (0.025 mol/L, pH 6.3) using p-nitrophenyl α-d-glucopyranoside as a substrate. β-Glucosidase (almond) or β-glucuronidase (bovine liver) was assayed at 37°C for 30 minutes in acetate buffer (0.1 mol/L, pH 5.0) using p-nitrophenyl β-d-glucopyranoside or phenolphthalein mono-β-d-glucuronic acid as substrates, respectively.

In vitro invasion and migration activities were assessed using the method of Albini et al. (38) For the invasion assay, the BD BioCoat Matrigel Invasion Chambers (Becton Dickinson Labware, Bedford, MA) were used. On the other hand, no Matrigel-coated cell culture inserts (Becton Dickinson Labware) were used for the migration assay. The chambers were placed on 24-well plates. Tumor cell suspension (0.5 mL, 5 × 10⁴ cells/mL) in DMEM (0% FCS) was added to the upper layer and 0.75 mL DMEM (10% FCS) was added to the lower layer. After adhesion for 1 hour, compounds or vehicle were added to the upper and lower layers, and for the HepG2 cells, human hepatocyte growth factor/scatter factor (R&D Systems, Inc., Minneapolis, MN) was also used as additive. After incubation for 20 hours at 37°C, the tumor cells and Matrigel on the upper surface of the membrane were completely removed by wiping the membrane with cotton swabs. The cells on the lower surface of the membrane were fixed with methanol and stained with 0.5% crystal violet. The cells from various regions of the membrane were counted and the percentage inhibition was calculated by the following equation: % inhibition = 100 − (cell number of the compound-treated experiment) / (cell number of the vehicle-treated experiment) × 100.

Wound healing assay was carried out using the method described previously (39). HT1080 cells (5 × 10⁵ cells/well in six-well plate) were incubated in dishes for 24 hours. An artificial wound was carefully created using plastic tip to scratch on the subconfluent cell monolayer. The medium was changed and the cells were treated with chemicals for 20 hours and were taken photographs.

The viability of tumor cells was measured by a trypan blue exclusion assay. Cell suspension (500 μL, 2 × 10⁵ cells/mL) was added to the wells of a 24-well plate (1 × 10⁵ cells/well) and incubated at 37°C for 12 hours. The cells were treated with either compounds or vehicle (DMSO) at 37°C for 48 hours. After trypsination (100 μL), 400 μL medium was added and 20 μL trypan blue solution [0.05% in PBS (−)] was added to the cell suspension (80 μL). After 10 minutes at room temperature, cell number was measured using a hemocytometer.

The viability of tumor cells was also measured by a 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H tetrazolium, monosodium salt (Nacalai Tesque, Kyoto, Japan) assay (40). Cell suspension (100 μL, 1 × 10⁵ cells/mL) was added to the wells of a 96-well plate (1 × 10⁴ cells/well) and incubated at 37°C for 12 hours. The cells were treated with either compounds or vehicle (DMSO) at 37°C for 48 hours. 2-(2-Methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H tetrazolium, monosodium salt solution (10 μL, 5 mmol/L) was added to the wells and the mixtures were incubated at 37°C for 2 hours. The absorbance was measured at 450 nm. The relative cell number was calculated as a percentage by the following equation: % relative cell number = (A − B) / (C − B) × 100, where A is the absorbance of the sample solution, B is the absorbance of the negative control, and C is the absorbance of the positive control.

Tumor cells were incubated with vehicle or compounds for 48 hours. The tumor cell suspension in DMEM (0% FCS) was added to human fibronectin cellware 35 mm dishes, collagen IV cellware 35 mm dishes, laminin cellware 35 mm dishes, or poly-d-lysine cellware 35 mm dishes (Becton Dickinson Labware). The plate was incubated at 37°C for 1 hour. After washing the plate with PBS, the adherent cells were fixed with glutaraldehyde and stained with crystal violet. Cells from various regions of the wells were counted and the percentage inhibition was calculated by the following equation: % inhibition = 100 − (cell number of a chemical-treated experiment) / (cell number of the vehicle-treated experiment) × 100.

The results of database searches and site-directed mutagenesis suggested that mammalian heparanase is related to the members of glycosyl hydrolase families 10, 39, and 51. Homology modeling of heparanase was done using the X-ray structure of 1,4-β-xylanase from P. simplicissimum (Protein Data Bank entry 1BG4), which is a member of the glycosyl hydrolase family 10, as a template. The heparan sulfate dimer, GlcA-GlcNAc (where GlcA and GlcNAc represent d-glucuronic acid and N-acetyl-d-glucosamine, respectively), was manually located on the active site of heparanase and the complex model was constructed using molecular mechanics and molecular dynamics calculations in consideration of the mechanism of enzymatic glycoside hydrolysis (41, 42). Based on the results, it was reasonable to assume that the protonation of the exocyclic oxygen in the glycosidic bond by Glu²²⁵ (proton donor) and the interaction between Glu³⁴³ (catalytic nucleophile) and the anomeric carbon atom of GlcA occurred in this heparanase/heparan sulfate model and that hydrogen bonds contributed to the stabilization of the complex structure (Fig. 1B). We assumed that the chemical structure of RK-682 mimics that of GlcA. Thus, RK-682 was located at the position of GlcA in consideration of their structural similarity and the complex model of heparanase/RK-682 was constructed by the molecular mechanics/molecular dynamics calculations. Hydrogen bonds between Arg²⁷² and 5-OH of RK-682, between His²⁹⁶ and 5-OH of RK-682, between Tyr²⁹⁸ and 4-OH of RK-682, and between Glu²²⁵ and 1-O of RK-682 stabilized the heparanase/RK-682 complex, just as they have been shown to stabilize the heparanase/heparan sulfate complex (Fig. 1C). The alkyl side chain of RK-682 was located along the cleft structure of heparanase, which consists of hydrophobic amino acids such as valine, leucine, and phenylalanine.

In the heparanase/RK-682 complex model, the 4-OH group of RK-682 looked toward the amino acids of His²⁹⁶ and Tyr²⁹⁸ (Figs. 1C and 2A) and we therefore speculated that the interaction between heparanase and RK-682 derivatives was more stabilized by the arylalkylation, as in the benzylation of the 4-OH group of RK-682 (Fig. 2B). It has been reported that 3-acyltetronic acids are extremely acidic; for example, the pKa value of 3-acetyltetronic acid is 0.8 (43). The acidic 3-acyltetronic acid structure of RK-682 acting as a phosphate mimic is very important to interact with protein tyrosine phosphatases and dual-specificity phosphatases (Fig. 2C). If the 4-OH group of RK-682 was blocked by alkylation, RK-682 was neutralized and decreased the binding strength to protein tyrosine phosphatases/dual-specificity phosphatases (Fig. 2D). Based on these data, we hypothesized that 4-Bn-RK-682 selectively inhibits heparanase.

We have carried out the chemical modification of RK-682 to obtain a specific inhibitor of heparanase. 4-Me-RK-682, 4-iPr-RK-682, and 4-Bn-RK-682 were prepared from RK-682 by the treatment of diazomethane, 1,2,3-triisopropyl-isourea, and 2-benzyl-1,3-dicyclohexyl-isourea, respectively (Fig. 1A).

The inhibitory activities of RK-682 derivatives against heparanase, vaccinia H1-related phosphatase, and glycosidases are shown in Fig. 3 and Table 1. RK-682 and 4-Bn-RK-682 inhibited heparanase equally (the IC₅₀ values of both compounds were 17 μmol/L; Fig. 3A; Table 1). 4-Me-RK-682 and 4-iPr-RK-682 scarcely showed heparanase inhibition (the IC₅₀ values of both compounds were >100 μmol/L; Fig. 3A; Table 1). RK-682 showed a potent inhibitory activity against vaccinia H1-related phosphatase (the IC₅₀ value was 6 μmol/L; Fig. 3B; Table 1). The inhibitory activities of 4-alkylated derivatives of RK-682 against vaccinia H1-related phosphatase were considerably reduced (the IC₅₀ values of these compounds were >100 μmol/L; Fig. 3B; Table 1). The selectivity (taken as the IC₅₀ value of vaccinia H1-related phosphatase divided by that of heparanase) of RK-682 was 0.4, whereas that of 4-Bn-RK-682 was >6. The heparanase selectivity of 4-Bn-RK-682 was >15-fold that of RK-682. Effects of RK-682 derivatives against other glycosidases (α-glucosidase, β-glucosidase, and β-glucuronidase) were also assayed. All compounds had no inhibitory activities against all these enzymes at a dose of 100 μmol/L (Table 1).

4-Bn-RK-682 remarkably inhibited both migration and invasion of HT1080 cells, whereas RK-682 showed no inhibitory activities against migration and invasion at 100 μmol/L (Fig. 4Aa). The IC₅₀ values of 4-Bn-RK-682 against the migration and invasion of HT1080 cells were 3.0 and 1.5 μmol/L, respectively. Whereas the invasion of HT1080 was not inhibited by treatment with vehicle (Fig. 4Ab) or 100 μmol/L RK-682 (Fig. 4Ac), 4-Bn-RK-682 at a concentration of only 10 μmol/L completely inhibited the invasion of HT1080 cells (Fig. 4Ad). In the HepG2 cell line that is heparanase deficient, 4-Bn-RK-682 did not exert any anti-invasive effect (Fig. 4B). There was no difference in the inhibition of invasion and migration of HT1080 cells by RK-682 and 4-Bn-RK-682 in the presence of heparan sulfate (Fig. 4C). Tumor cell migration was also studied using an in vitro wound healing assay (Fig. 4D). Confluent HT1080 cells were scratched (Fig. 4Da) and treated with chemicals for 20 hours. After the treatment with vehicle (DMSO) or RK-682 (100 μmol/L), HT1080 cells migrated and covered a great area of the scratch (Fig. 4Db and Dc). On the other hand, 10 μmol/L 4-Bn-RK-682 significantly inhibited the migration of HT1080 cells (Fig. 4Dd).

As results of trypan blue exclusion assay (Fig. 5Aa) and 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H tetrazolium, monosodium salt assay (Fig. 5Ab), neither RK-682 nor 4-Bn-RK-682 showed cytotoxicity and cell growth inhibition at a dose of <100 μmol/L. It is very interesting that the treatment of 4-Bn-RK-682 makes HT1080 cells adhesive (Fig. 5B). Adherent cell number was increased by the treatment of 4-Bn-RK-682 against all of the tested adhesion molecules (i.e., fibronectin, collagen IV, laminin, poly-d-lysine, and plastic). The increase was most significant against collagen IV.

RK-682 was known to inhibit the dual-specificity protein phosphatase vaccinia H1-related phosphatase (25). 4-Bn-RK-682, which was rationally designed based on the docking model with heparanase, inhibited heparanase (IC₅₀, 17 μmol/L) without inhibiting vaccinia H1-related phosphatase at dose of up to 100 μmol/L. The level of heparanase inhibition by 4-Bn-RK-682 was comparable or more potent than that by the other small molecular heparanase inhibitors reported to date. The small molecular heparanase inhibitors have been reported by other groups as follows. The IC₅₀ values of suramin, Evans blue, and trypan blue against murine B16BL6 melanoma heparanase were 46, 320, and 320 μmol/L, respectively (26). Trachyspic acid purified from the culture broth of Talaromyces trachyspermus SANK 12191 inhibited murine B16BL6 melanoma heparanase at IC₅₀ value of 36 μmol/L (44). A-72363 A-1, A-2, and C (the diastereomers of siastatin B, a neuraminidase inhibitor) were purified from the culture filtrate of Streptomyces nobilis SANK 60192 in the screening for heparanase inhibitors. A-72363 C inhibited heparanase from B16BL6 cells (IC₅₀, 12 μmol/L) and β-glucuronidase from bovine liver (IC₅₀, 1.6 μmol/L; refs. 45, 46). A transition state mimic of glycosidase, d-galacturonic acid–type gem-diamine 1-N-iminosugar, also potently inhibited heparanase from human melanoma A375M cells (IC₅₀, 1.0 μmol/L) and β-glucuronidase from bovine liver (IC₅₀, 0.065 μmol/L; ref. 47).

In either cases, A-72363 C and d-galacturonic acid–type gem-diamine 1-N-iminosugar have potent inhibitory activity for heparanase than 4-Bn-RK-682, but both compounds are iminosugar derivatives having strong inhibitory activity for β-glucuronidase rather than heparanase (45–47). 4-Bn-RK-682, however, is not a sugar mimic and does not show inhibitory activity for other glycosidases (Table 1).

4-Bn-RK-682 inhibited the invasion and migration of human fibrosarcoma HT1080 cells at low concentration (the IC₅₀ values against the invasion and migration were 1.5 and 3.0 μmol/L, respectively), whereas the compounds did not show any inhibitory activities against migration and invasion of HepG2 cells (Fig. 4A and B). We have reported previously that heparanase mRNA and protein expression levels were hardly detectable in HepG2 cells (12). These results indicate that antimigrative and anti-invasive effects of 4-Bn-RK-682 are related with the heparanase inhibition by the compound. Additive heparan sulfate did not affect the inhibitory activities of 4-Bn-RK-682 against migration and invasion (Fig. 4C). This result suggests that 4-Bn-RK-682 does not interact with heparan sulfate and the drug may bind to heparanase rather than substrate heparan sulfate.

Heparan sulfate is one of the main components of extracellular matrix (Matrigel) and is ubiquitously existent at the cell surface. Therefore, the accessibility of additive heparan sulfate to the Matrigel and the cell surface is not well because of the electrostatic repulsion of negative charge. Heparanase is existent at the cell surface and in the medium. Additive heparan sulfate is possible to inhibit heparanase in the medium but not at the cell surface and did not inhibit the migration and invasion of HT1080 cells (Fig. 4C). From these aspects, it is suggested that heparanase at the cell surface but not in the medium is important for the migration and invasion of HT1080 cells. It is reported by other groups that addition of heparanase stimulated phosphoinositide 3-kinase–dependent endothelial cell migration and invasion (48).

4-Bn-RK-682 showed neither cytotoxicity nor cell growth inhibition at a dose of <100 μmol/L (Fig. 5Aa and Ab). These results indicate that the remarkable inhibition of cell invasion and migration was caused by neither cytotoxicity nor cell growth inhibition. Heparan sulfate glycosaminoglycans act as physical barriers in the extracellular matrix and the basement membrane and tumor cells harness heparanase to destroy heparan sulfate glycosaminoglycans for invasion. Therefore, in the invasion assay using Matrigel as a barrier, it is reasonable that a selective heparanase inhibitor, 4-Bn-RK-682, inhibited the invasion of HT1080 cells. But why 4-Bn-RK-682 also inhibited the migration of HT1080 cells in the experiment using no Matrigel is unclear. One answer is that 4-Bn-RK-682 inhibited the migration of HT1080 cells by increasing adherence of the cells (Fig. 5B). It was reported previously that a heparanase inhibitor inhibited the migration and invasion of HT1080 cells by increasing the adherence of the cells (31). Furthermore, heparanase has been known to involve in tumor migration by the degradation of cell surface heparan sulfate glycosaminoglycans of heparan sulfate proteoglycans (e.g., syndecans), which is a key mechanism of cell spreading and migration. For example, syndecan-4 is located at sites of cell-matrix adhesion (49), and regulates cell spreading and migration, in cooperation with integrins (50, 51). Thus, the present findings indicating that heparanase inhibition by 4-Bn-RK-682 results in inhibition of the invasion and migration of HT1080 cells are consistent with the previous reports. The invasion and migration of HT1080 cells were also inhibited by the antiheparanase antibody treatment (31). This result supports that there is a correlation between the heparanase inhibition and the invasion/migration inhibition.

We successfully produced a heparanase selective inhibitor by the benzylation at the 4-position of RK-682 according to a working hypothesis guided by the heparanase/RK-682 docking model. In addition, we attempted to assess the biological activity of the compound for which α/β-naphthylmethyl group was substituted at the 4-position, but this biological evaluation was hard to execute because of the weak water solubility of the compounds. The molecular mechanics/molecular dynamics calculation led the speculated interaction model of heparanase/4-Bn-RK-682 (Fig. 6). Two hydrogen bonds (between Arg²⁷² and 5-OH of 4-Bn-RK-682 and between Glu²²⁵ and 1-O of the drug) stabilize the complex. The benzyl group at 4-position of 4-Bn-RK-682 contributes to stabilize the complex by the π-π stacking with Tyr²⁹⁸ and His²⁹⁶, instead of the hydrogen bonds between His²⁹⁶ and 5-OH of RK-682, and between Tyr²⁹⁸ and 4-OH of RK-682 in the heparanase/RK-682 complex model (Fig. 1C).

This is the first report of a heparanase inhibitor that has noncharged character. Negatively charged RK-682 inhibited not only heparanase activity but also other enzyme activities. In sulfated oligosaccharides, an oligosaccharide chain length and a degree of sulfation are critical to inhibit angiogenesis and heparanase activity (14). These inhibitory activities might be exhibited by not only heparanase inhibition but also interfering with the formation of a ternary complex, which consists of heparan sulfate binding growth factor, heparan sulfate, and growth factor receptor (20, 52). In iminosugar derivatives, other glycosidases, especially β-glucuronidase, were inhibited instead of heparanase (46, 47). Negative charge–dependent interaction between a heparan sulfate binding growth factor and heparan sulfate may not be interfered by noncharged 4-Bn-RK-682. Therefore, 4-Bn-RK-682 could be assumed as a pure heparanase inhibitor different from other inhibitors described previously.

Whereas the IC₅₀ values of RK-682 and 4-Bn-RK-682 against heparanase were equal, RK-682 showed no inhibitory activities for cell-based assays (migration and invasion). This result may be caused by the difference of accessibility to the cell membrane. The neutral character of 4-Bn-RK-682 was effective against cellular effects as compared with RK-682 possessing negative charge. These results indicate that 4-Bn-RK-682 could be assumed not only as a bioprobe for decipherment heparanase function but also as a promising antimetastatic agent.