Processing of Macromolecular Heparin by Heparanase*

Heparanase is an endo-glucuronidase expressed in a variety of tissues and cells that selectively cleaves ex-tracellular and cell-surface heparan sulfate. Here we propose that this enzyme is involved also in the processing of serglycin heparin proteoglycan in mouse mast cells. In this process, newly synthesized heparin chains (60–100 kDa) are degraded to fragments (10–20 kDa) similar in size to commercially available heparin (Ja-cobsson, K. G., and Lindahl, U. (1987) Biochem. J. 246, 409–415). A fraction of these fragments contains the specific pentasaccharide sequence required for high affinity binding to antithrombin implicated with anticoagulant activity. Rat skin heparin, which escapes processing in vivo , was used as a substrate in reaction with recombinant human heparanase. An incubation product of commercial heparin size retained the specific pentasaccharide sequence, although oligosaccharides (3–4 kDa) containing this sequence could be degraded by the same enzyme. Commercial heparin was found to be a powerful inhibitor (I 50 (cid:1)

Heparin is a linear polysaccharide synthesized by connective tissue-type mast cells as part of a unique proteoglycan, serglycin. The heparin chains, similar to other glycosaminoglycans, are composed of repeating hexuronic acid (HexUA) 1 and D-glucosamine (GlcN) units that are sulfated at various positions. Heparin is structurally related to heparan sulfate (HS), which is expressed by most mammalian cells but has lower N-and O-sulfate contents. More than half of the HexUA in heparin is L-iduronic acid (IdoUA), whereas generally HexUA in HS is largely D-glucuronic acid (GlcUA) (1). The well known blood anticoagulant activity of heparin depends on the occurrence of a specific pentasaccharide sequence in the polysaccharide, which is capable of interacting with the proteinase inhibitor antithrombin (AT) (2). Although heparin has been used as an anticoagulant in the clinic for a long time, the physiological function of heparin remains unclear. Recent studies suggest an essential role for heparin in the storage and activation of specific proteases in mast cell granules (3)(4)(5).
Heparin chains ("macromolecular heparin," hereon referred to as M-heparin) bound to serglycin-type core peptide can be isolated from rat skin (6) or rat peritoneal mast cells (7) and range in size from 60 to 100 kDa. By contrast, heparin ("commercial heparin") recovered from other animal tissues such as pig intestinal mucosa or bovine lung is much smaller (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). The majority of such molecules lack covalently bound peptide, and it is generally assumed that they are generated through a common processing mechanism. Fragmentation of newly synthesized macromolecular heparin was demonstrated in cultured mast cells along with the transfer of the processed products (P-heparin) to intracellular granules (8). The degradation products were similar in size to commercial heparin. An endo-␤-D-glucuronidase capable of degrading M-heparin (6,9) was tentatively implicated in the process.
Subsequently, endo-glucuronidase ("heparanase") activity associated with cleavage of HS chains was demonstrated in a variety of normal and malignant cells (10 -13) and shown to play a role in cell invasion, angiogenesis, inflammation, and tissue remodeling (12,14,15). Cloning of the enzyme from different tissues revealed a single gene (16 -18), suggesting that mammalian cells express primarily one heparanase (13,15) (however, note previous reports (19,20) claiming the occurrence of several distinct heparanases based on biochemical studies).
Early studies with platelet heparanase showed that this enzyme could cleave the glucuronidic linkage in oligosaccharides containing the AT-binding sequence of heparin, such that the cleavage products lacked affinity for AT and hence were devoid of anticoagulant activity (21). Because the endo-glucuronidase-catalyzed cleavage of M-heparin in mast cells generates P-heparin-type products that retain the AT-binding se-quence, the platelet heparanase was assumed to differ from the mast cell endo-glucuronidase. This assumption was supported by the apparent lack of effect of the mast cell enzyme on AT-binding oligosaccharide and by a preliminary comparison of substrate recognition properties of the platelet and mast cell enzymes using microsomal intermediates in heparin biosynthesis as substrates (21).
Heparanase-mediated cleavage of HS yields fragments of variable size, only a limited proportion of glucuronidic linkages in HS chains being susceptible to degradation. Attempts to define the substrate specificity of heparanase pointed to the importance of sulfation but have otherwise failed to provide a unified picture (22)(23)(24).
The aim of this study was to identify the mast cell enzyme responsible for the intracellular processing of M-heparin, i.e. the polysaccharide chain constituents of the serglycin proteoglycan. An analysis of mRNA in cultured mastocytoma cells and subsequent cloning of cDNA revealed expression of a homologue of the heparanase found in other tissues. Recombinant heparanase was found to catalyze the two cleavage reactions previously ascribed to separate enzymes (21), i.e. one converting M-heparin to P-heparin, the other resulting in cleavage of the AT-binding sequence. Due to the relative resistance of this sequence to heparanase, the overall process is evolved to favor generation of P-heparin with retained AT-binding domain.

EXPERIMENTAL PROCEDURES
Materials-Columns for gel chromatography, Superose 12 TM , Superose 6, and Sephadex-30 were purchased from Amersham Biosciences. Chemical reagents and equipment were as described in the appropriate sections below.
Human Recombinant Heparanase-Human heparanase was cloned and overexpressed in Chinese hamster ovary cells as reported previously (16). The recombinant enzyme, purified from cell extracts by Mono-S HR 5/5 cation exchange chromatography followed by heparinand ConA-Sepharose chromatographies, was kindly provided by In-Sight Ltd. (Rabin Science Park, Rehovot, Israel).
Polysaccharide Substrates-Rat skin heparin (2.5 mg), prepared as described previously (25), was incubated in 0.3 ml of 0.5 M NaOH on ice overnight. The sample was neutralized with HCl and subsequently desalted on a PD-10 column (Amersham Biosciences). The resulting free M-heparin chains were radiolabeled by reduction with 1 mCi of NaB 3 H 4 (55 Ci/mmol, Amersham Biosciences) in 0.1 ml of 0.1 M NaOH. Following acidification of the reaction mixtures with 4 M acetic acid, samples were neutralized with 4 M NaOH and 3 H-labeled M-heparin (ϳ360 ϫ 10 3 dpm/g HexUA) was recovered by gel chromatography on a column (1 ϫ 100 cm) of Sephadex G-50 equilibrated with 0.2 M NaHCO 3 .
Porcine intestinal heparin was N-[ 3 H]acetyl-labeled by partial Ndeacetylation followed by N-[ 3 H]acetylation essentially as described previously (26) with the exception that the hydrazinolysis reaction time was extended to 2 h. The product had a specific activity of ϳ84 ϫ 10 3 dpm/g HexUA.
Affinity fractionation of labeled saccharides on immobilized AT was done essentially as described previously (27). The sample was applied to a 3-ml column of AT-Sepharose and eluted in stepwise fashion with NaCl of different concentrations (see "Results") in 0.25 M Tris-HCl, pH 7.4. Fractions of 1 ml were collected and analyzed for radioactivity. Components emerging at Ͻ0.5 M NaCl were designated low affinity (LA) heparin, whereas fractions retained on AT-Sepharose at 0.8 M NaCl and eluted with 2.0 M NaCl were considered high affinity (HA) heparin.
Oligosaccharide Substrates-Radiolabeled heparin oligosaccharides with high affinity for AT were prepared essentially as described previously (27). Heparin was subjected to partial deaminative cleavage with nitrous acid, and the products were reduced with NaB 3 H 4 . Labeled oligosaccharides of different molecular size were isolated by gel chromatography (Sephadex G-50) and further separated by affinity chromatography on immobilized AT-Sepharose. Two HA preparations were recovered, one (10 -12-mer) with a specific activity of ϳ1 ϫ 10 6 dpm/g HexUA, the other (12-14-mer) with a specific activity of ϳ116 ϫ 10 3 dpm/g HexUA.
Labeled 8-mers were obtained through a "biosynthetic library" strategy recently described for the generation of heparin-related oli-gosaccharides (28) using capsular polysaccharide from Escherichia coli K5 as starting material. This polymer has the same structure, [4GlcA␤1,4GlcNAc␣1] n , as the precursor polysaccharide in heparin/HS biosynthesis (1) and is recognized as a substrate for the biosynthetic enzymes (29) following chemical N-deacetylation and N-sulfation (30). N-Sulfated K5 polysaccharide was incubated with mouse mastocytoma microsomal enzymes (including GlcUA C5epimerase and various O-sulfotransferases) in the presence of PAPS under the conditions described previously (28). The resulting modified product was re-isolated and partially cleaved by treatment with nitrous acid. 3 H-Labeled oligosaccharides obtained by reduction with NaB 3 H 4 were separated by gel chromatography, and an 8-mer fraction was recovered. The 8-mers were subjected to repeated incubation with enzymes, and PAPS and were finally re-isolated and desalted.
Cloning of Mouse Mastocytoma Heparanase-Cultured murine mastocytoma cells (31) contained Ͼ90% mast cells enriched through repeated subculturing (data not shown) and retained the ability to synthesize and process M-heparin as described previously (8). Total RNA was extracted using the TRIzol reagent (Invitrogen) according to the manufacturer's description. The cDNA was prepared with First Stranded cDNA synthesis kit for reverse transferase-PCR (Roche Applied Science) and used as a template for PCR cloning of heparanase. PCR primers were designed based on the cDNA sequences of rat and human heparanase.
Analysis of Heparanase Activity-Standard incubations of heparanase using radiolabeled saccharide as substrates (ϳ0.2-6 M disaccharide unit) and ϳ15 ng of protein (ϳ2.3 nM enzyme) in 100 l of 20 mM phosphate-citrate buffer, 1 mM CaCl 2 , 50 mM NaCl, pH 5.4, were used. Samples were incubated at 37°C for 16 h. Deviations from the standard conditions are defined under "Results" and the legends to figures. After incubation, samples were heated at 95°C for 5 min and centrifuged. Supernatants were analyzed by gel chromatography or affinity chromatography.
Analytical Procedures-HexUA was determined by the meta-hydroxydiphenyl method (32). Concentrations of radiolabeled saccharide preparations were assessed from determinations of specific radioactivity. Molar concentrations of oligosaccharides were calculated based on the number of disaccharide units. Molar concentrations of polysaccharides were obtained from estimated average M r , assuming a disaccharide unit M r of 600.
Analytical affinity chromatography on AT-Sepharose was similar to the preparative procedure described above with the exception that the column volume was 0.5 ml and effluent fractions of 0.5 ml were collected.
For gel chromatography, columns of Superose 12, Superose 6, and Sephadex-30 were operated according to the manufacturer's instruction in a buffer of 50 mM Tris-HCl, pH 7.4, containing 1 M NaCl and 0.1% Triton X-100. Samples of ϳ10 ϫ 10 3 dpm were generally applied.
Anion-exchange high pressure liquid chromatography of oligosaccharides was performed using a ProPac PA1 column (Dionex) equilibrated in NaCl solution, which was adjusted to pH 3 by the addition of HCl. The column effluent was analyzed by on-line scintillation counting. Sequence analysis of N-sulfated oligosaccharides was performed according to the combined chemical/enzymatic degradation scheme described previously (28,33).
High voltage paper electrophoresis was carried out as described previously (9) using pyridine-acetate buffer, pH 5.3, and Whatman 3MM paper. Gulono-␥-lactone (25 g) applied in 25 l of 0.25 M NH 4 OH (to hydrolyze the lactone to free acid) was used as a standard. Labeled samples were treated in similar fashion. After electrophoresis, the standard was detected by a silver dip procedure (34). Paper strips containing radiolabeled samples were cut into 1-cm pieces that were extracted with water and analyzed by scintillation counting.
Digestion of oligosaccharides with exo-␤-glucuronidase from bovine liver was performed as described previously (27). Deaminative cleavage of heparin-related saccharides at the sites of N-sulfated GlcN residues was achieved by treatment with nitrous acid at pH 1.5 as described previously (35).

RESULTS
Heparanase Cleaves M-heparin to P-heparin-M-heparin was released from the serglycin proteoglycan by alkaline ␤-elimination and then radiolabeled by reduction with NaB 3 H 4 . The product was incubated with heparanase under standard conditions (see "Experimental Procedures"). An analysis of the digest by gel chromatography revealed a single peak with an overall elution pattern similar to that of commercial heparin (peak position indicating ϳ15 kDa) (Fig.  1A), and this component remained unaffected through repeated incubation with additional enzyme (Fig. 1A). Affinity chromatography analysis of this product showed no AT-binding-labeled material (data not shown), indicating lack of any AT-binding region between the linkage region to core protein and the site of the first heparanase attack along the Mheparin chain (see Fig. 6). Because the radiolabel was located exclusively at the reducing end of the intact M-heparin chain, these results will reflect only the properties of the fragment released by heparanase from the reducing terminal portion of the chain. To analyze the total cleavage products, the labeled M-heparin was fractionated by affinity chromatography on AT-Sepharose and samples of (end group-labeled) HA-and LA-M-heparin were incubated with heparanase under standard conditions. The digestion products were reduced with NaB 3 H 4 using the same batch of borotritide as employed in the initial labeling of the M-heparin chains. The yield of labeled digestion products was 4 -5-fold higher than the amount of starting material, thus indicating the approximate number of cleavage points along the chain. The cleavage product of HA-M-heparin has a similar molecular size as commercial heparin, whereas that of LA-M-heparin was of somewhat smaller average molecular size than commercial heparin (Fig. 1B). Incubation of these fragments with heparanase again did not result in any further decrease in size (data not shown), suggesting end-point degradation under the conditions used.
The AT-binding Regions of M-heparin Escape Degradation by Heparanase-The relabeled fragments isolated from the heparanase digest of M-HA-heparin were fractionated by affinity chromatography on AT-Sepharose. Approximately 40% of the products showed high affinity for AT ( Fig. 2A), thereby indicating the proportion of heparanase-generated fragments containing the AT-binding sequence. These relations point to an average of 2-3 AT-binding sites in each chain of M-HAheparin from rat skin. The labeled HA products obtained after heparanase digestion were similar in size to commercial heparin, whereas the LA fragments were somewhat smaller (Fig.  2B). Reincubation of this HA fraction with heparanase had no effect on its affinity for AT (Fig. 2C) nor on its molecular size (Fig. 2D). Notably, the AT-binding sites in rat skin heparin were previously found to contain the critical -GlcUA-Glc-NSO 3 (3,6-di-OSO 3 )-sequence (36) present also in HA-oligosaccharides derived from commercial heparin. Such oligosaccharides were cleaved upon incubation with heparanase (21,22,24).
In accord with these findings, N-[ 3 H]acetyl-labeled commercial heparin was largely unaffected by incubation with heparanase (data not shown). Repeated incubation with additional 15 ng of enzyme again had no apparent effect, whereas incubation with 5-fold increased amount (75 ng, 11.5 nM) of heparanase resulted in a limited degradation of the polysaccharide (data not shown).
Heparin Efficiently Inhibits Cleavage of HA-Oligosaccharides by Heparanase-Previous work indicated that small heparin-derived oligosaccharides containing the AT-binding sequence can be cleaved by digestion with heparanase (21,22). This finding was confirmed in the present study, which also showed cleavage of HA-oligosaccharides to be quite inefficient. Repeated incubation of HA 12-14-mer (ϳ4.5 M disaccharide unit, ϳ0.7 M oligosaccharide) with 2.3 nM heparanase thus resulted in cleavage of only about half of the substrate molecules as shown by gel chromatography (Fig. 3A). Incubation with a 5-fold increased enzyme amount gave essentially complete degradation of the oligosaccharide (Fig. 3B). Conversely, decreasing the substrate concentration (ϳ0.5 M disaccharide unit, ϳ0.1 M oligosaccharide) likewise afforded extensive degradation (Fig. 4, absence of inhibitor). Although these findings suggest inefficient catalysis, they nevertheless stand in contrast to the apparent lack of effect of heparanase on the commercial heparin-sized HA product primarily excised by the enzyme from M-HA-heparin (Fig. 2, C and D). The main difference between the two potential substrates lies in the extended sequence of highly sulfated saccharide that surrounds the AT-binding domain in the full-sized HA-heparin molecule but is lacking in the HA-oligosaccharide preparations. These adjacent structures might inhibit heparanase action as suggested also by previous observations of inhibitory effects of heparin on heparanase (37). Indeed, the addition of 1 ng (ϳ20 nM disaccharide unit, ϳ0.7 nM polysaccharide) of heparin to a standard incubation containing 2.3 nM heparanase led to a ϳ50% inhibition of the degradation of HA 10 -12-mer (ϳ0.5 M disaccharide unit, ϳ0.1 M oligosaccharide); with 10 ng (ϳ200 nM disaccharide unit, ϳ7 nM polysaccharide) of added heparin, inhibition was virtually complete (Fig. 4). This remarkably efficient inhibition, I 50 ϳ0.7 nM heparin, would explain the relative resistance of the AT-binding sequence in HA-heparin to degradation by heparanase. It should be noted that each heparin chain presumably contains multiple inhibitory enzyme-binding sites. Substrate Recognition by Heparanase-Although the heparanase has been identified as an endo-glucuronidase (22,38), it was essential to verify that the facile degradation of M-heparin chains by the enzyme was in fact because of cleavage of glucuronidic linkages. To this end, the 3 H-labeled end groups obtained upon NaB 3 H 4 reduction of the P-heparin digestion products were released by treatment with nitrous acid under conditions resulting in deaminative cleavage at N-sulfated GlcN residues. Gel chromatography of labeled deamination products showed a distinct monosaccharide peak along with larger components that emerged at elution positions corresponding to trisaccharides and pentasaccharides (data not shown). The monosaccharide was recovered and migrated as guluronic acid on paper electrophoresis (data not shown). These findings confirm the ␤-glucuronidase action of hepara-nase and identify a -GlcNR-[HexUA-GlcNR] 2 -GlcUA-* sequence (R ϭ -SO 3 Ϫ or -COCH 3 ) upstream of the target glucuronidic linkage (indicated by asterisk), the adjacent GlcN units thus being either N-sulfated or N-acetylated.
More unexpected variability in substrate recognition was revealed by incubating recombinant heparanase with end group-labeled 8-mer library oligosaccharides (see "Experimental Procedures"). A tri-O-sulfated fraction was shifted in elution position upon anion-exchange high pressure liquid chromatography (data not shown). This fraction was isolated and separately incubated with the enzyme, resulting in partial (ϳ60%) conversion into a product (arrow 1 in Fig. 5C) more retarded on anion-exchange high pressure liquid chromatography than the undigested fraction (Fig. 5B). Sequence analysis showed that the intact fraction (mixture of 8-mers 1 and 2 in Fig. 5A) was fully N-and 6-O-sulfated and that the internal HexUA units were all IdoUA. However, 8-mer 1 contained a nonreducing-terminal GlcUA residue, whereas 8-mer 2 carried an IdoUA residue in the same position, suggesting that the heparanase had unexpectedly acted in exolytic fashion. This conclusion was supported by the finding that digestion with bovine liver ␤-glucuronidase, a recognized exo-enzyme, resulted in products with an elution profile similar to that in Fig.  5C. Moreover, L-iduronidase digestion resulted in ϳ40% of the sample being cleaved (8-mer 2) leaving ϳ60% (8-mer 1) intact (data not shown). Control experiments were performed to eliminate the possibility of an exo-␤-glucuronidase contaminant in the preparation of recombinant heparanase. The addition of 10 mM glucaric acid-1,4-lactone, a known inhibitor of ␤-glucuronidase (39), to incubations with heparanase did not prevent cleavage of 8-mer 1 but abolished the effect of liver glucuronidase (data not shown). Conversely, the addition of suramin, an inhibitor of heparanase (40), prevented the 8-mer cleavage by the recombinant heparanase. Interestingly, 8-mer 3 with nonreducing-terminal as well as internal GlcUA residues (Fig. 5A) was recognized as a far better substrate by liver ␤-glucuronidase than by heparanase (data not shown). Also, the activity of heparanase toward 8-mer 1, containing internal IdoUA units, was found to be much higher than that toward 8-mer 3, containing GlcUA units only (data not shown). We conclude that heparanase may act in exolytic as well as in endolytic fashion.
Murine Mastocytoma Cells Express the Heparanase-To elucidate the role of heparanase in the intracellular processing of M-heparin, we first investigated whether the heparanase, ubiquitous in many mammalian cells and tissues, is expressed also in mast cells. Mast cells from a murine mastocytoma (31) previously shown to produce and process heparin proteoglycan provided a source of cDNA that was probed with several pairs derived from a separate library preparation was essentially unaffected by heparanase but was digested by ␤-glucuronidase (data not shown). The column was eluted with a 0 -1 M NaCl gradient (total volume, 100 ml).
of primers based on the human and rat heparanase cDNA sequences. A full-length mast cell cDNA obtained by repeated PCR encoded a protein with 91 and 77% identity to rat (Gen-Bank TM accession number AF184967) and human heparanases (GenBank TM accession number AF144325), respectively, at the polypeptide level (data not shown). Incubation of N-[ 3 H]acetyllabeled HS with a mast cell lysate resulted in a gel chromatography pattern of cleavage products identical to that observed after incubation with human recombinant heparanase (data not shown).

DISCUSSION
The occurrence of an endo-glucuronidase in neoplastic mast cells, capable of converting M-heparin to P-heparin, was first described almost 30 years ago (9). At the same time, an enzyme from rat liver was found to degrade heparan sulfate to oligosaccharides (41). The latter enzyme, also an endo-glucuronidase, was subsequently found to be ubiquitously expressed in a variety of tissues and is now commonly referred to as heparanase (12,15). A comparison of the mastocytoma enzyme and heparanase derived from platelets pointed to a wider substrate specificity for the latter species that was found to degrade not only N-and O-sulfated polysaccharides generated in the course of cell-free heparin biosynthesis but also oligosaccharides containing the specific AT-binding pentasaccharide sequence. By contrast, the allegedly mast cell-specific endo-glucuronidase appeared at the time to show highly restricted specificity toward selected linkages in M-heparin and failed to cleave the AT-binding saccharide sequence (21). It was concluded that the mast cell and platelet enzymes were distinct entities.
Our present findings instead strongly suggest that the different cleavage reactions are the result of a single enzyme, i.e. heparanase. Incubation of M-heparin (60 -100 kDa) from rat skin with recombinant heparanase thus yielded P-heparin products similar in size to commercial heparin (Fig. 1). The P-heparin resists further degradation by the enzyme. These results strongly resemble those previously obtained by incubating M-heparin isolated from mouse mastocytoma with the mastocytoma endo-glucuronidase (9). Moreover, an analysis of cleavage products revealed the same mixed N-acetyl/N-sulfate substituent pattern upstream of the cleavage sites (Ref. 9 and data not shown) (for details see "Results"). The heparanase showed no apparent effect on the AT-binding regions present (Fig. 2). Notably, the number of HA-P-heparin molecules generated from the initial HA-M-chain averaged 2-3 as determined by AT-Sepharose chromatography of radiolabeled heparanase digestion products and this value agrees with the number of AT-binding regions per HA-M-chain calculated from structural data and from the rat heparin-induced enhancement of the intrinsic fluorescence of AT (36). Also, commercial pig mucosa heparin appeared unaffected by heparanase. Such heparin further resembles the rat skin P-heparin products in that only a minor proportion of the molecules contain the linkage region to core protein (42). Heparanase-catalyzed conversion of HA-M-heparin into HA-P-and LA-P-products and of LA-Mheparin into LA-P-product (Fig. 6) thus would account for the mixture of HA-and LA-components in various heparin preparations. Variations in the proportions and molecular size of these components reflect the distribution of heparanase cleavage sites in the corresponding M-heparin precursor structures. The proposed role for heparanase in processing of M-heparin is supported by our present finding that an enzyme closely related to the ubiquitous heparanase is expressed by the heparinproducing murine mastocytoma cells.
Heparanase reproduces the effects previously ascribed to the mastocytoma endo-glucuronidase but also mimics the platelet enzyme in degrading AT-binding oligosaccharide. Why then was this substrate previously found resistant to digestion by the mastocytoma enzyme preparation (21) appears to be due to kinetics of the heparanase. The ATbinding sequence, apparently a poor substrate, requires relatively high enzyme concentration for cleavage to occur. Enzyme concentrations were unknown in the old experiments, and it seems likely that the amounts of mastocytoma enzyme used were sufficient for conversion of M-heparin to P-heparin but not for cleavage of AT-binding oligosaccharide. On a related note, why do the AT-binding regions in M-heparin remain intact through heparanase digestion when the AT-binding pentasaccharide is recognized as a substrate by this enzyme? Findings by our and other groups point to a broad substrate specificity of the enzyme, and an associated range of affinities toward substrate and non-substrate saccharide motifs (22)(23)(24).
Although several studies emphasize the role of O-sulfate groups for substrate recognition (22,24,43), the precise location of critical sulfate residues has not been established. N-Sulfation is required for the incorporation of O-sulfate groups during heparin/HS biosynthesis (44, 45) but seems not per se to be mandatory for heparanase action (22). A recent proposal implicating the -GlcNR(6-OSO 3 )-GlcUA-GlcNSO 3 -trisaccharide sequence as essential for substrate recognition (24) disagrees with our present finding (of unknown functional significance), that heparanase can actually cleave a GlcUA-[IdoUA-GlcNSO 3 (6-OSO 3 )-] n structure in exolytic fashion (Fig. 5). The AT-binding saccharide shown here and previously (22) to be cleaved by heparanase contains a predominant -IdoUA-GlcNAc(6-OSO 3 )-GlcUA*-GlcNSO 3 (3,6-di-OSO 3 )-IdoUA(2-OSO 3 )-sequence (the asterisk indicates the target glucuronidic linkage) (Fig. 6B). It has been proposed based on the relative susceptibility to heparanase cleavage of several related oligosaccharides that the 3-O-sulfate group on the adjacent GlcN residue may exert an inhibitory effect when located in a highly sulfated sequence (24). Even if the structural basis for the difference remains unclear, it is obvious that the AT-binding regions in M-heparin are less susceptible to cleavage by heparanase (Fig. 6A, arrowheads) than are the major sites cleaved to yield P-heparin products (Fig. 6A, arrows).
Further of importance to our understanding of heparanase action on M-heparin is the strongly inhibitory effect of heparin polysaccharide on the catalytic activity of the enzyme. This effect, also recognized in previous studies (37,46), was here observed upon the addition of heparin to incubations of ATbinding oligosaccharide at molar concentration (of intact polysaccharide) similar to that of the enzyme (Fig. 4). If inhibition, as assumed, is the result of the predominant [IdoUA(2-OSO 3 )-GlcNSO 3 (6-OSO 3 )-] n repeat structure, the heparin chain displays an array of repetitive overlapping inhibitory sites that effectively occur at higher molar concentration than that of the whole chain. In HA-P-heparin (as well as in commercial HA-heparin), the molar ratio of inhibitory to substrate (AT-binding) sequences thus is higher than in the HA-oligosaccharides found to be cleaved by the enzyme. This difference may explain the relative resistance of HA-P-heparin chains toward digestion by heparanase compared with the smaller HA-oligomers as well as the recurring finding that saccharides resisting prolonged incubation at a given enzyme concentration were susceptible to degradation at higher enzyme concentration. The effects of inhibitory motifs in relation to the catalytic efficiencies toward the various potential cleavage targets can be evaluated once the structures of these sites are elucidated. The occurrence of substrate and inhibitor structures in the same polymer molecule is unconventional, and the properties of the various domains in relation to heparanase action are not readily amenable to kinetic analysis. Factors of potential importance that need to be studied include the distribution of relevant domains along the M-heparin chain, the mode of in-teraction of the heparanase molecule with each domain type and the selectivity toward these domains.