Distinct Substrate Specificities of Bacterial Heparinases against N-Unsubstituted Glucosamine Residues in Heparan Sulfate*

The rare N-unsubstituted glucosamine (\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{GlcNH}_{3}^{+}\) \end{document}) residues in heparan sulfate have important biological and pathophysiological roles. In this study, four \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{GlcNH}_{3}^{+}\) \end{document}-containing disaccharides were obtained from partially de-N-sulfated forms of heparin and the N-sulfated K5 polysaccharide by digestion with combined heparinases I, II, and III. These were identified as \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \({\Delta}\mathrm{HexA}\mathrm{-}\mathrm{GlcNH}_{3}^{+},{\Delta}\mathrm{HexA}\mathrm{-}\mathrm{GlcNH}_{3}^{+}(6S),{\Delta}\mathrm{HexA}(2S)\mathrm{-}\mathrm{GlcNH}_{3}^{+}\) \end{document}, and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \({\Delta}\mathrm{HexA}(2S)\mathrm{-}\mathrm{GlcNH}_{3}^{+}(6S)\) \end{document}. Digestions with individual enzymes revealed that heparinase I did not cleave at \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{GlcNH}_{3}^{+}\) \end{document} residues; however, heparinases II and III showed selective and distinct activities. Heparinase II generated \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \({\Delta}\mathrm{HexA}\mathrm{-}\mathrm{GlcNH}_{3}^{+}(6S),{\Delta}\mathrm{HexA}(2S)\mathrm{-}\mathrm{GlcNH}_{3}^{+}\) \end{document}, and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \({\Delta}\mathrm{HexA}(2S)\mathrm{-}\mathrm{GlcNH}_{3}^{+}(6S)\) \end{document} disaccharides, whereas heparinase III yielded only the \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \({\Delta}\mathrm{HexA}\mathrm{-}\mathrm{GlcNH}_{3}^{+}\) \end{document} unit. Thus, the action of heparinase II requires O-sulfation, whereas heparinase III acts only on the corresponding non-sulfated unit. These striking distinctions in substrate specificities of heparinases could be used to isolate oligosaccharides with novel sequences of \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{GlcNH}_{3}^{+}\) \end{document} residues. Finally, heparinases were used to identify and quantify \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{GlcNH}_{3}^{+}\) \end{document}-containing disaccharides in native bovine kidney and porcine intestinal mucosal heparan sulfates. The relatively high content of O-sulfated \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{GlcNH}_{3}^{+}\) \end{document}-disaccharides in kidney HS raises questions about how these sequences are generated.

NAc) (8,10,13,14). This precursor is first modified by coordinated N-deacetylation and N-sulfation of GlcNAc residues forming N-sulfoglucosamine (GlcNS). The N-sulfated polymer then undergoes a further series of modifications, i.e. C5-epimerization of GlcA to iduronic acid (IdoA), O-sulfation at C2 of hexuronic acid (primarily IdoA), and O-sulfation at C6 of Gl-cNS or GlcNAc. Additionally, a rare but functionally important step, 3-O-sulfation of the GlcN residue, can also occur late in the modification process. These modifications create a complex sulfated domain structure in HS, whereas heparin is extensively and uniformly sulfated.
Because of the varied structure of mature HS chains and their extensive repertoire of interactions with different proteins, the rare structural components presumably contribute to selective protein binding. Recently, the N-unsubstituted GlcN (GlcNH 3 ϩ ) unit was found in native HS structures (15,16) and was implicated in important cell biological and pathophysiological phenomena. Levels of GlcNH 3 ϩ residues were found to correlate with the ability of bovine and human endothelial HS to bind L-selectin (17). GlcNH 3 ϩ residues were also identified as targets for an HS 3-O-sulfotransferase isoform (HS 3-OST-3A) that generates a sequence in cell surface HS that is utilized as a binding site by the herpes simplex virus glycoprotein D, thus making cells susceptible to herpes simplex virus-1 entry (18 -20). Moreover, two monoclonal antibodies that recognize Glc-NH 3 ϩ units in HS were reported, and they highlighted distinctive localizations of their respective epitopes in tissues. One monoclonal antibody bound to extracellular tissue components in rat kidney (21), whereas the other reacted with scrapie lesions in murine brain (22,23). Finally, it has been proposed that GlcNH 3 ϩ residues provide cleavage sites in HS chains for endogenous NO-derived nitroxyl anion and thus contribute to recycling of glypican-1 (24).
The content of GlcNH 3 ϩ residues is low but variable between HS species. Values from 1.2 to 7.5 wt% of total GlcN were calculated for a range of porcine and bovine HS species based on reaction with o-phthaldialdehyde (15). However, GlcNH 3 ϩ residues are mostly analyzed by treatment with high pH (pH 3.9) nitrous acid, leading to chain cleavage at the GlcNH 3 ϩhexosaminidic linkage. By this method the content in human aortic and kidney HS was shown to be 2-4% (16). Quantitation of GlcNH 3 ϩ by this approach requires either the analysis of metabolically radiolabeled HS, or post-cleavage, reducing end chemical radiolabeling of the cleavage products. So far there has been no identification and characterization of GlcNH 3 ϩ residues in HS by the use of degradative enzymes that could obviate the need for radiolabeling. Moreover, a better understanding of the substrate specificities of heparin/HS-degrading enzymes could be of value in sequencing GlcNH 3 ϩ -containing oligosaccharides and in the isolation of fragments with either intact internal or reducing end GlcNH 3 ϩ residues.
Glycosaminoglycan-degrading bacterial enzymes are useful analytical tools for investigating the composition and sequence of various glycosaminoglycans (25). Three heparinases (I, II, and III) have been isolated from Flavobacterium heparinum that differ from one another in terms of their substrate specificities (26). These have been fairly well defined. Heparinase I has high specificity for highly sulfated disaccharides containing iduronate 2-O-sulfate and thus cleaves at 34GlcNS(Ϯ6S)134IdoA(2S)13 linkages (27)(28)(29). Heparinase III has highest activity toward non-sulfated or N-sulfated disaccharides (GlcNAc/GlcNS-HexA) with a preference for GlcA over IdoA. It will tolerate 6-O-sulfation of the amino sugar but is inhibited by 2-O-sulfation of IdoA (30,31). Heparinase II has a broad substrate specificity, being able to cleave a wide range of disaccharide repeat units (30,32,33). It was reported that heparinase II has two distinct active sites, one of which is heparinase I-like, whereas the other is heparinase III-like (34). Moffat et al. reported that heparinase II has activity on Nunsubstituted disaccharides in chemically modified heparin (32,35). However, we are not aware of any detailed reports about the comparative substrate specificities of heparinase I, II, and III toward sulfated and non-sulfated disaccharides containing GlcNH 3 ϩ residues. Because the GlcNH 3 ϩ content is generally very low in HS, we have, in this study, used the partially de-N-sulfated forms of both heparin and the N-sulfated K5 polysaccharide (NS-K5) to test the susceptibility of N-unsubstituted disaccharides to the various heparinases. We have then demonstrated their use for the analysis of GlcNH 3 ϩ residues in HS. . Bio-Gel P2 (fine grade) was from Bio-Rad Laboratories (Hemel Hempstead, Hertfordshire, UK). ProPac PA-1 analytical columns were obtained from Dionex (Camberley, Surrey, UK). All HPLC solutions were prepared using MilliQ (Millipore; Watford, Hertfordshire, UK) ultrapure water. Amberlite IR-120 (H ϩ form) and all reagents used for de-N-sulfation and re-N-acetylation were from Sigma. N-Sulfated K5 polysaccharide (NS-K5) was kindly supplied by Dr. P. Oreste at Glycores (Milan, Italy).

Materials-An
De-N-sulfation of Heparin and N-Sulfated K5 Polysaccharide-Partially and fully de-N-sulfated heparin were both prepared using Innohep as the starting material, essentially as described by Nagasawa et al. (36). The pyridinium salt of heparin was obtained by passage through an Amberlite IR-120 (H ϩ form) column and titration of the resulting protonated form with pyridine. The pyridinium salt (10 mg) was then treated with 1 ml of 95% Me 2 SO, 5% water at 20°C for 60 min (partial de-N-sulfation) or 95% Me 2 SO, 5% methanol at 50°C for 90 min (complete de-N-sulfation). The samples were then dialyzed extensively and freeze-dried. Partial de-N-sulfation of NS-K5 was similarly performed. The degrees of de-N-sulfation were calculated from the increases in N-acetylated disaccharides obtained after re-N-acetylation (see below).
Determination of Enzymatic Activities-Fifty micrograms of substrates in 1 ml of 0.1 M sodium acetate buffer, pH 7.0, containing 0.1 mM calcium acetate and 100 g/ml bovine serum albumin, were digested at 37°C with 1 mIU of each heparinase. Digestion was measured continuously by the change in absorbance at 232 nm on a spectrophotometer.
Analysis of Enzymatic Digestion Products-Two hundred micrograms of partially or fully de-N-sulfated heparin, or partially de-N-sulfated NS-K5, in 200 l of digestion buffer (as above) was treated with 6 mIU of each heparinase at 37°C. Digestion was taken to completion over 24 h and stopped by heating at 100°C for 2 min. After centrifuging the reaction mixture at 13,000 ϫ g for 15 min, the supernatant was applied to a Bio-Gel P2 column (1.7 ϫ 115 cm) eluted with 0.2 M NH 4 HCO 3 . Elution of oligosaccharides was determined by measuring absorbance at 232 nm. The NH 4 HCO 3 was removed from appropriate fractions by incubation at 55°C for 24 h followed by three cycles of freeze-drying.
Strong Anion-exchange-HPLC Analysis of Disaccharides-SAX-HPLC separation of disaccharides was performed on a HP-1100 system with a ProPac PA-1 column (4.6 ϫ 250 mm). The disaccharides from the Bio-Gel P2 column were applied in 1 ml of water pH3.5 at a flow rate of 1 ml/min. After a 2-ml wash with pH 3.5 H 2 O, a biphasic linear gradient of NaCl was applied from 0 to 0.5 M (2.1-35.1 min) and then 0.5 to 1.0 M (35.1-57.1 min). The elution profile was monitored by absorbance at 232 nm, and fractions of 0.5 ml were collected.
Preparation of Disaccharides from De-N-sulfated Heparin and De-Nsulfated NS-K5-Two hundred micrograms of de-N-sulfated heparin or de-N-sulfated NS-K5 were digested with a mixture of heparinase I, II, and III (24 milliunits of each enzyme) at 37°C for 48 h, and then applied to a Bio-Gel P2 column (1.7 ϫ 115 cm) eluted with 0.2 M NH 4 HCO 3 . Fractions of 1.5 ml were collected, and their absorbance was measured at 232 nm. The disaccharide peak was collected, and the NH 4 HCO 3 removed, as described above.
Re-N-acetylation of De-N-sulfated Heparin and Purified N-Unsubstituted Disaccharides-De-N-sulfated heparin was dissolved in 0.25 M sodium phosphate buffer, pH 7.5. A 10-fold excess of acetic anhydride was gradually added with continuous stirring for 1 h at 0°C. The re-N-acetylated heparin was then desalted on a PD-10 column (Amersham Biosciences). N-Unsubstituted disaccharides were similarly re-Nacetylated, but in 50 mM sodium phosphate buffer, pH 7.5, and the reaction mixture was then dried.

Preparation of De-N-sulfated
Heparins-In this study, two de-N-sulfated heparin preparations, partial and complete, were prepared. The degrees of de-N-sulfation, calculated from the increase in N-acetylated disaccharides obtained after re-Nacetylation and digestion with heparinases, were 50 and 98% for the two preparations, respectively.
Digestion of Partially De-N-sulfated Heparin with Heparinases-Partially de-N-sulfated (50%) heparin was digested with a mixture of heparinases I, II, and III, and separated by Bio-Gel P2 gel filtration chromatography (Fig. 1). Only two major peaks were obtained, which were identified as dp4 (minor peak) and dp2 by comparing their elution positions with those of standard oligosaccharides. From the peak areas it could be calculated that 86% of the linkages in partially de-Nsulfated heparin were cleaved by the heparinases. The excess of degradation (86%) over the content of N-unsubstituted residues (50%) indicated that (i) heparinase I, II, or III must have activity toward N-unsubstituted disaccharides and (ii) N-unsubstituted oligosaccharides (probably disaccharides) can be obtained by this procedure.
Structural Analysis of N-unsubstituted Disaccharides Derived from Partially De-N-sulfated Heparin-To positively identify each of the three later-eluting unknown disaccharides (b, c, and d), they were collected individually from SAX-HPLC, re-N-acetylated, and then re-analyzed by SAX-HPLC. As shown in Fig. 3A, after re-N-acetylation, disaccharide b shifted its retention time to a later position corresponding to that of the standard disaccharide ⌬HexA-GlcNAc(6S), indicating that the original structure of b must have been ⌬HexA-GlcNH 3 ϩ (6S). In the same way, disaccharide c shifted its retention time to that of ⌬HexA(2S)-GlcNAc (Fig. 3B), and disaccharide d shifted its to that of ⌬HexA(2S)-GlcNAc(6S) (Fig. 3C). These results similarly identified disaccha-rides c and d as originally being ⌬HexA(2S)-GlcNH 3 ϩ and ⌬HexA(2S)-GlcNH 3 ϩ (6S), respectively.

Structural Analysis of N-Unsubstituted Disaccharides from
Partially De-N-sulfated NS-K5-As the content of disaccharide a is low in partially de-N-sulfated heparin, and it was eluted from the SAX-HPLC column very early at low salt concentration, it was difficult to identify its structure conclusively using heparin. According to its elution position, and the known identities of the three other species (b-d), we might expect disaccharide a to be a non-sulfated, N-unsubstituted disaccharide. To possibly prepare a larger amount of this disaccharide, we chose de-N-sulfated NS-K5 as a potential substrate. The K5 polysaccharide has the same sequence as completely unmodified HS. NS-K5 is thus comprised mainly of GlcA-GlcNS disaccharides (37).
NS-K5 was partially de-N-sulfated to generate a significant proportion of GlcA-GlcNH 3 ϩ disaccharide units. The degree of de-N-sulfation achieved was 45%. This material was then digested with a mixture of heparinases I, II, and III and separated by Bio-Gel P2 gel filtration chromatography. The dp2 fraction obtained was analyzed by SAX-HPLC. Compared with control NS-K5, which yielded almost entirely ⌬HexA-GlcNS (Fig. 4A), one additional peak was obtained from the partially de-N-sulfated NS-K5, which eluted very early at 3.1 min (Fig. 4B). Its elution position is the same as that of disaccharide a from partially de-N-sulfated heparin (Fig. 2B). This suggests that this disaccharide is likely to be ⌬HexA-GlcNH 3 ϩ . To confirm this, disaccharide a was collected, re-N-acetylated, desalted on Bio-Gel P2, and then re-analyzed by SAX-HPLC. As shown in Fig.  4C, after re-N-acetylation, disaccharide a shifted its retention time to a later position corresponding to the known elution position of ⌬HexA-GlcNAc, unequivocally confirming that the original structure of disaccharide a is ⌬HexA-GlcNH 3 ϩ .

Specificity of Individual Heparinases toward N-Unsubstituted Glucosamine
Residues-To identify the specific heparinase(s) responsible for the activity toward GlcNH 3 ϩ -containing residues, de-N-sulfated heparins and partially de-N-sulfated NS-K5 were digested with heparinases I, II, or III individually. Enzymatic activities were monitored by following the increase in UV absorbance at 232 nm with time (Fig. 5A). Heparinase I

Heparinases against GlcNH 3 ϩ Residues
showed no activity toward the fully de-N-sulfated heparin, and heparinase III acted only to a small extent. In contrast, heparinase II demonstrated significant activity against this substrate. A comparison of the kinetics of heparinase II degradation of heparin to those of both partially and fully de-N-sulfated heparins showed that the presence of increasing levels of GlcNH 3 ϩ did slow the rate of degradation (Fig. 5B), although an extensive, but incomplete, level of degradation can be achieved with time with both de-N-sulfated substrates.
Heparinase II also showed activity toward partially de-Nsulfated NS-K5 (45% de-N-sulfated) (Fig. 5C). However, this polysaccharide was a much better substrate for heparinase III (Fig. 5C). This demonstrates the likely activity of heparinase III against non-sulfated GlcNH 3 ϩ residues. To further analyze the differential activities of individual heparinases toward N-unsubstituted disaccharides, the digestion products from fully de-N-sulfated heparin incubated for 48 h with either heparinase I, II, or III were separated by Bio-Gel P2 gel filtration chromatography (Fig. 6A). The disaccharide fractions were then analyzed by SAX-HPLC (Fig. 6, B  and C). No disaccharides were released by treatment with heparinase I (data not shown). After heparinase III digestion,
To confirm the differential activities of heparinases II and III toward sulfated and non-sulfated, N-unsubstituted units, the digestion products from partially de-N-sulfated NS-K5, after incubation with heparinase II or III, were also separated by Bio-Gel P2 gel filtration chromatography (Fig. 7A). Heparinase II yielded a very small amount of disaccharides and longer fragments. However, heparinase III generated a large amount of disaccharides and some tetrasaccharides, but nothing larger, implying almost quantitative cleavage of this substrate. These results suggested that heparinase III might display preferential activity toward non-sulfated, N-unsubstituted sequences.
Identification of N-Unsubstituted Disaccharides in HS Species-To assay for the presence of natural GlcNH 3 ϩ residues in HS using heparinases, bovine kidney and porcine intestinal mucosal HS were chosen. The GlcNH 3 ϩ content of these two HS species had previously been shown to be very different (15,16). Bovine kidney HS was exhaustively digested with a mixture of heparinases I, II, and III, and the disaccharide fraction was recovered by Bio-Gel P2 chromatography and analyzed by SAX-HPLC (Fig. 8A). Eight components were identified that correspond to known N-substituted disaccharide standards (labeled [1][2][3][4][5][6][7][8]. Four additional peaks were also present (comprising 12% of total disaccharides), corresponding to the known N-unsubstituted units (labeled a-d). To confirm these latter structural assignments, bovine kidney HS was first N-acetylated before enzymatic digestion and disaccharide analysis. The four Nunsubstituted disaccharides (a-d) now disappeared, but the corresponding N-acetylated derivatives (peaks 1, 3, 4, and 7) increased by 10% in total (including the significant appearance of disaccharide 7 (HexA(2S)-GlcNAc(6S), which was barely present in the native HS) (Fig. 8B). These results show that HexA-GlcNH 3 ϩ , HexA-GlcNH 3 ϩ (6S), HexA(2S)-GlcNH 3 ϩ , and HexA(2S)-GlcNH 3 ϩ (6S) disaccharides are present in bovine kidney HS at levels of 5.1%, 3.4%, 2.2%, and 1.9% of total disaccharides, respectively (calculated as an average of two analyses). In the same way, three N-unsubstituted disaccharides, HexA-GlcNH 3 ϩ , HexA-GlcNH 3 ϩ (6S), and HexA(2S)-GlcNH 3 ϩ , were detected in porcine intestinal mucosal HS at 0.4%, 0.4%, and 0.3% of total disaccharides, respectively, i.e. a total of 1.1% (data not shown). The more highly sulfated HexA(2S)-GlcNH 3 ϩ (6S) disaccharide was not detected in this HS. This compares with a total GlcNH 3 ϩ content of 0.7% determined by pH3.9 nitrous acid treatment and chromatographic analysis (16). DISCUSSION The content of GlcNH 3 ϩ residues is low and variable between HS species, ranging from 1.2% to 7.5 wt%, as determined using o-phthaldialdehyde (15). Because of this apparently low (15,16) natural content in HS, we have used chemical modification of heparin and the K5 polysaccharide to increase the content of GlcNH 3 ϩ residues to test the susceptibilities of N-unsubstituted disaccharides to the heparinases, which are widely used for analysis of HS and heparin.
Unexpectedly, we found that the majority of hexosaminidic linkages in a partially (50%) de-N-sulfated heparin were cleaved to disaccharides by a combination of heparinases I, II, and III (Fig. 1). This led us to surmise that (i) the heparinases could be used as alternative tools for the quantitation and analysis of GlcNH 3 ϩ residues in HS, and (ii) that N-unsubstituted disaccharides could be obtained by this procedure. SAX-HPLC analyses of the disaccharides obtained from de-N-sulfated heparin revealed four additional disaccharide peaks (a-d) (Fig. 2B), that were not present in digests of the parent heparin ( Fig. 2A), and did not correspond to any of the known N-substituted disaccharide standards. Re-N-acetylation of the partially de-N-sulfated heparin, prior to enzymatic digestion, did not give these four disaccharide peaks (Fig. 2C), but the content of the four N-acetylated disaccharides increased, suggesting that the four novel disaccharides (a-d) are indeed N-unsubstituted ones. Individual recovery of three of these disaccharides (b-d), followed by their N-acetylation, allowed their structures to be unequivocally determined (Fig. 3). These were variously O-sulfated, N-unsubstituted species, namely ⌬HexA-GlcNH 3 ϩ (6S), ⌬HexA(2S)-GlcNH 3 ϩ , and ⌬HexA(2S)-GlcNH 3 ϩ (6S). The content of one further putative N-unsubstituted disaccharide a in partially de-N-sulfated heparin was low (Fig. 2B), but from its elution position on SAX-HPLC we surmised that it might be the non-sulfated, N-unsubstituted disaccharide. To confirm this, we prepared a disaccharide with identical elution position, but in greater abundance, from a similar enzyme digest of partially de-N-sulfated NS-K5 (Fig. 4B). Collection and N-acetylation of this disaccharide yielded the structure ⌬HexA-GlcNAc (Fig. 4C), confirming that disaccharide a was indeed the non-sulfated disaccharide ⌬HexA-GlcNH 3 ϩ . N-unsubstituted disaccharides are normally identified and quantified in HS by chemical assay or by specific pH 3.9 nitrous acid degradation and chromatographic separation of the cleavage products. However, both methods destroy the GlcNH 3 ϩ residues. In this study we describe the generation of intact, unsaturated N-unsubstituted disaccharides by digestion with a combination of heparinases followed by SAX-HPLC. By their identification, and the characterization of their SAX-HPLC elution positions, it is now possible for heparinases to be conveniently used for analyzing the content of N-unsubstituted disaccharides in HS as part of a standard disaccharide analysis (Fig. 8A).
Heparinases are very important tools in studying the composition and role of biologically relevant HS/heparin sequences. It was reported that heparinase I has specificity for relatively highly sulfated disaccharides containing both N-and 2-O-sulfation, whereas heparinase III cleaves at non-and low sulfated disaccharides predominantly containing GlcA. Heparinase II has a broader specificity, cleaving at both high sulfation and low sulfation sites, but not at all linkages. Which heparinases are responsible for cleaving at GlcNH 3 ϩ residues? We have investigated the individual activities of heparinases I, II, and III toward substrates with either a high content of O-sulfate groups (i.e. partially or fully de-N-sulfated heparin) or deficient in O-sulfation (i.e. partially de-N-sulfated NS-K5) (Fig. 5). This revealed that heparinase II was primarily responsible for the specificity of cleavage at GlcNH 3 ϩ in the de-N-sulfated heparins, although the rate and extent of depolymerization was reduced by the presence of the primary amine (Fig. 5A). Heparinase II excises N-unsubstituted disaccharides that are Osulfated at C-2 of uronic acid, or C-6 of GlcNH 3 ϩ , or both (Fig.  6B); it is inactive against the non-sulfated counterpart (Fig.  7B). This is consistent with the broad sulfation specificity of heparinase II. However, with the same substrates, heparinase III specifically excises non-sulfated, N-unsubstituted disaccharides (Fig. 7C), but has almost no activity toward mono-or di-O-sulfated ones in either fully de-N-sulfated heparin (Fig.  6C) or native HS (date not shown). By contrast, heparinase I showed no propensity to cleave at any sites of N-unsubstitution. The differences in activity of heparinase enzymes on Glc-NH 3 ϩ residues in HS are illustrated in Fig. 9. Chai et al. (38) reported that, using controlled conditions for partial heparinase III digestion of partially de-N-acetylated K5 polysaccharide, the enzyme was not able to cleave the GlcNH 3 ϩ -GlcA linkage within a ⌬HexA-GlcNH 3 ϩ -GlcA-GlcNAc sequence; even under forced conditions for exhaustive digestion, the Glc-NH 3 ϩ -GlcA linkages were still only partly cleaved. In our study, using the partially de-N-sulfated NS-K5 as a substrate, we reached a different conclusion, because this alternative substrate was highly degraded into disaccharides by heparinase III (Fig. 7A). This suggests that heparinase III has higher activity toward the GlcNH 3 ϩ -GlcA linkage within ⌬HexA-Glc-NH 3 ϩ -GlcA-GlcNS sequences. It is unlikely that this could be due to contaminating enzyme activities, because neither hepa-FIG. 8. SAX-HPLC analysis of disaccharides obtained from bovine kidney HS after digestion with heparinases. Bovine kidney HS (A) and re-N-acetylated bovine kidney HS (B) were digested with a mixture of heparinases I, II, and III. Disaccharides were recovered and analyzed by SAX-HPLC. Numbered peaks correspond to the known N-substituted disaccharide standards (see Fig. 2). Panel A shows one run from duplicate analyses; arrows indicate the elution positions of the four N-unsubstituted disaccharides (a-d) previously identified from digests of partially de-N-sulfated heparin (see Fig. 2B) and partially de-N-sulfated NS-K5 polysaccharide (see Fig. 4B).

FIG. 9. Action of bacterial heparinases on GlcNH 3
؉ residues in HS. The diagram shows a section of the HS chain with the non-sulfated regions in green, the transition zones in blue, and the S-domains in red as in the model of Murphy et al. (37). GlcNH 3 ϩ -disaccharides are found in all regions and are indicated by the striped lines. Heparinase III (Hep III) will cleave the GlcNH 3 ϩ hexosaminidic linkage in the non-sulfated regions, whereas heparinase II (Hep II) will scission this linkage if the disaccharide is O-sulfated, as is usually the case in the S-domains; the actions of the two enzymes are mutually exclusive. Heparinase I (Hep I) will not scission linkages bearing the free amine, but it can be used to excise sequences from S-domains with an internal GlcNH 3 ϩ residue. GlcNH 3 ϩ are also found in the transition zones, and scission by heparinases here will depend on whether the associated disaccharide is Osulfated (Hep II-sensitive) or non-sulfated (Hep III-sensitive).