Role of Deacetylase Activity of N-Deacetylase/N-Sulfotransferase 1 in Forming N-Sulfated Domain in Heparan Sulfate*

Background: N-Deacetylase/N-Sulfotransferase is a bifunctional enzyme in the heparan sulfate biosynthetic pathway. Results: N-Deacetylase and N-sulfotransferase domains display cooperative effects as a bifunctional enzyme; individually expressed N-deacetylase and N-sulfotransferase domains do not exhibit the same level of cooperativity. Conclusion: The bifunctionality of N-deacetylase/N-sulfotransferase is required to form N-S domain in heparan sulfate. Significance: This work uncovers the process by which NDST-1 constructs functional heparan sulfate. Heparan sulfate (HS) is a highly sulfated polysaccharide that plays important physiological roles. The biosynthesis of HS involves a series of enzymes, including glycosyltransferases (or HS polymerase), epimerase, and sulfotransferases. N-Deacetylase/N-Sulfotransferase isoform 1 (NDST-1) is a critical enzyme in this pathway. NDST-1, a bifunctional enzyme, displays N-deacetylase and N-sulfotransferase activities to convert an N-acetylated glucosamine residue to an N-sulfo glucosamine residue. Here, we report the cooperative effects between N-deacetylase and N-sulfotransferase activities. Using baculovirus expression in insect cells, we obtained three recombinant proteins: full-length NDST-1 and the individual N-deacetylase and N-sulfotransferase domains. Structurally defined oligosaccharide substrates were synthesized to test the substrate specificities of the enzymes. We discovered that N-deacetylation is the limiting step and that interplay between the N-sulfotransferase and N-deacetylase accelerates the reaction. Furthermore, combining the individually expressed N-deacetylase and N-sulfotransferase domains produced different sulfation patterns when compared with that made by the NDST-1 enzyme. Our data demonstrate the essential role of domain cooperation within NDST-1 in producing HS with specific domain structures.

Heparan sulfate (HS) 3 is a highly sulfated linear polysaccharide ubiquitously present on the cell surface and in the extra-cellular matrix. HS has a wide range of biological functions, including embryonic development, inflammatory response, response to viral and bacterial infection, and regulation of blood coagulation (1). The disaccharide repeating unit of HS consists of glucosamine or either glucuronic acid (GlcA) or iduronic acid. Both the iduronic acid (less frequently for GlcA) and glucosamine saccharides carry sulfo groups. The location of the sulfo groups and the placement of iduronic acid play a crucial role in determining the functions of HS (2). Specialized enzymes such as HS polymerase, C 5 -epimerase, and sulfotransferases participate in synthesizing HS (Fig. 1A). HS polymerase is responsible for producing the polysaccharide backbone with a repeating unit of -GlcA-GlcNAc-. The backbone is then modified by N-deacetylase/N-sulfotransferase (NDST), C 5 -epimerase, 2-O-sulfotransferase, 6-O-sulfotransferase, and 3-Osulfotransferase. The substrate specificities of the biosynthetic enzymes ultimately dictate the structures of HS (3). Understanding the biosynthetic mechanism of HS is critically important for delineating the structure/function relationship of HS as well as for improving a chemoenzymatic approach to synthesize HS-based therapeutics (4,5).
Structural analysis of HS isolated from natural sources indicates that the GlcNS residues are not evenly distributed along the HS chain, instead forming clusters known as N-S domains (21). Formation of the N-S domain enhances the downstream modification reactions, resulting in functional HS that displays high affinity binding to proteins (22). We previously demonstrated that NDST-1 has the ability to form the N-S domains, providing the means to synthesize biologically active HS (23). In this work, we investigate the relationship between N-deacetylase and N-sulfotransferase activities in the process of converting GlcNAc residues to GlcNS, as well as the ability to form N-S domain structures. The present study utilizes highly purified recombinant proteins and structurally defined oligosaccharide substrates to provide accurate structural analysis of the products. In addition to expressing full-length NDST-1, we successfully expressed the N-deacetylase and N-sulfotransferase domains separately, allowing us to identify cooperation between N-deacetylase and N-sulfotransferase activities. Although both individually expressed domains display their respective activities, when combined, they do not generate the product that contains N-S domains. This result demonstrates the cooperative nature of the bifunctional NDST-1 and its critical role in generating functional HS.

Experimental Procedures
Preparation of NDST-1, NDase, and NST Expression Plasmids-The clone containing the sequence for full-length human NDST-1 was purchased from Open Biosystems (Huntsville, AL). Secreted forms of recombinant proteins were prepared for this study. The primers, 5Ј-ATGAATTCAGTCGC-CCCTAGCCGCCTCCT-3Ј and 5Ј-TTACTCGAGCTACC-TGGTGTTCTGGAGGT-3Ј, were used for amplifying fulllength NDST-1 (Val 62 -Arg 882 ). The primers, 5Ј-TATGAATT-CACGCCTCCTGCCAATCAAGC-3Ј and 5Ј-TTACTCG-AGCTAGAAGCGGTCACATGTCTTC-3Ј, were used for amplifying the N-deacetylase (NDase) domain (Arg 66 -Phe 604 ). The primers, 5Ј-TATGAATTCACTGCAGACTCTGC-CCC-3Ј and 5Ј-TTACTCGAGCCTGGTGTTCTGGAGGT-3Ј, were used for amplifying the N-sulfotransferase (NST) domain (Leu 557 -Arg 882 ). The restriction sites of EcoRI and XbaI used for subcloning are underlined. The PCR reactions were performed to amplify NDST-1, NDase, and NST by heating samples to 95°C for 5 min followed by 30 cycles of 95°C for FIGURE 1. HS biosynthetic pathway and reaction catalyzed by NDST-1. A, enzymatic modification steps required to produce HS. The polymerization step to form the polysaccharide backbone (-GlcA-GlcNAc-) is not shown. After NDST-1 modification, the polysaccharide undergoes C 5 -epimerization and 2-O-sulfation followed by 6-O-sulfation and more limited 3-O-sulfation. The shorthand keys for the symbols are depicted on the right side of the panel. B, reactions catalyzed by NDST-1. Two steps are involved to convert a GlcNAc residue to a GlcNS residue. First, NDST-1 removes the acetyl group to form GlcNH 2 , and the activity at this step is designated as N-deacetylase activity. Second, a sulfo group is transferred from the sulfo donor PAPS to the amino group to form GlcNS. C, locations of N-deacetylase and N-sulfotransferase domains in NDST-1. TM represents the transmembrane domain. 30 s, 55°C for 30 s, 70°C for 2 min, and 70°C for 10 min. The resulting PCR products were cloned into the pFastBac-mal-HT vector using EcoRI and XbaI sites to form NDST-1, NDase, and NST carrying a His 6 tag at the N terminus (24). All construct sequences were verified by the University of North Carolina Genome Analysis Facility. The preparation of recombinant baculovirus stocks harboring the targeted protein gene of interest was completed according to the manufacturer's instructions (Invitrogen).
Expression and Purification of NDST-1, NDase, and NST-Expression of the recombinant proteins was carried out in Sf9 cells grown in Sf-900 III SFM (Life Technologies) in 1-liter plastic shaker flasks with vented caps (200 ml of culture per flask) at 28°C. When the cell density reached 2.0 ϫ 10 6 cells ml Ϫ1 , recombinant virus was introduced to infect the cells. The target protein in the condition medium was harvested ϳ96 h after infection by centrifugation at 16,000 ϫ g for 30 min. The resultant medium was mixed with an equal volume of 20 mM MOPS, pH 7.0 buffer, and was filtered through a Whitman glass microfiber filter (1.2 m, Fisher) before the purification.
All three recombinant proteins were purified following a very similar protocol at 4°C. The diluted and filtered medium was applied to a heparin-Toyopearl gel (Tosoh Bioscience) column (1.2 ϫ 10 cm) at a flow rate of 4 ml min Ϫ1 and washed with buffer 1 (20 mM MOPS, 50 mM NaCl, 2% glycerol, pH 7.0). The protein was eluted with buffer 2 (20 mM MOPS, 1 M NaCl, 2% glycerol, pH 7.0). Fractions containing NDST-1, NDase, or NST were pooled and dialyzed against a buffer containing 25 mM Tris, pH 7.5, 30 mM imidazole, and 500 mM NaCl using a 14,000 molecular weight cut-off membrane (Fisher). The dialyzed samples were applied to a nickel-agarose (GE Healthcare) column (0.6 ϫ 5.0 cm) at a flow rate of 0.5 ml min Ϫ1 and washed with a buffer containing 25 mM Tris, pH 7.5, 30 mM imidazole, and 500 mM NaCl followed by washing the column with the same buffer containing 64 mM imidazole until the optical density (280 nm) of the eluent reached baseline. The protein was then eluted with 300 mM imidazole in 25 mM Tris, pH 7.5, and 500 mM NaCl. The protein was stored at Ϫ80°C for the experiments. The protein concentrations were determined by Bradford reagent (Sigma), and the purities were determined to be greater than 90% using 12% SDS-PAGE (Bio-Rad).
Determination of N-Deacetylase/N-Sulfotransferase, N-Deacetylase, and N-Sulfotransferase Activity-To measure the activity of N-deacetylase/N-sulfotransferase, heparosan (10 g) was incubated with NDST-1 (10 ng) in a 100-l reaction volume containing 50 mM MES, pH 7.0, 10 mm MnCl 2 , PAPS (50 M and 1 ϫ 10 6 cpm of [ 35 S]PAPS) at 37°C for 1 h. The reaction was stopped by adding 100 l of a solution containing 50 mM sodium acetate, pH 5.0, 150 mM sodium chloride, 3 M urea, 1.5 mM EDTA, and 0.1% Triton X-100. The 35 S-labeled polysaccharide product was purified by a DEAE column. The extent of 35 S incorporation was determined by scintillation counting.
To measure N-deacetylase activity, heparosan (10 g) was incubated with NDST-1 (10 ng) in a 100-l reaction volume containing 50 mM MES, pH 7.0, 10 mM MnCl 2 at 37°C for 1 h. The reaction was stopped by heating at 100°C for 5 min. To the reaction mixture, 20 g of recombinant NST expressed in Escherichia coli (25), unlabeled PAPS (50 M), and [ 35 S]PAPS (1 ϫ 10 6 cpm) were added, and the reaction was incubated at 37°C for 1 h. The 35 S-labeled polysaccharide product was then purified and quantified using a DEAE column as described above. One unit of N-deacetylase activity was defined as the transfer of 1000 cpm of 35 S-labeled sulfo group to heparosan under this condition.
To measure N-sulfotransferase activity, deacetylated heparosan (10 g) that was prepared via a chemical approach (26) was incubated with the protein (10 ng) and [ 35 S]PAPS (1 ϫ 10 6 cpm) and unlabeled PAPS (50 M) in a buffer containing 50 mM MES, pH 7.0, and 10 mM MnCl 2 at 37°C for 1 h. The reaction was stopped by adding 100 l of a solution containing 50 mM sodium acetate, pH 5.0, 150 mM sodium chloride, 3 M urea, 1.5 mM EDTA, and 0.1% Triton X-100. The 35 S-labeled polysaccharide product was purified by a DEAE column. One unit of N-sulfotransferase activity was defined as the transfer of 1000 cpm of 35 S-labeled sulfo group to deacetylated heparosan under this condition.
Determination of the Effects of EDTA on the Individual N-Sulfotransferase and N-Deacetylase Activities-To determine the effect of EDTA on the N-deacetylase activity, a range of concentrations (0.25-10 mM) of EDTA was tested in the reaction mixture containing NDST-1 and heparosan. In this experiment, Mn 2ϩ (10 mM) was omitted from the reaction buffer. To measure the extent of deacetylation, recombinant N-sulfotransferase expressed in E. coli (20 g) and [ 35 S]PAPS (25) were used to sulfate the product. The level 35 S incorporation correlated to the level of deacetylation in the polysaccharide product. To determine the effect of EDTA on N-sulfotransferase activity, 1 mM of EDTA was mixed with NDST-1, deacetylated heparosan, and [ 35 S]PAPS in the reaction buffer that did not contain Mn 2ϩ .
Determination of the Effects of Metal Ions on the Full-length N-Deacetylase/N-Sulfotransferase Activity-Different metal ions (10 mM), including Ca 2ϩ , Mn 2ϩ , Mg 2ϩ , Zn 2ϩ , Co 2ϩ , or Cu 2ϩ , were added in the reaction mixture containing NDST-1, heparosan, and [ 35 S]PAPS. The activity was measured following the standard procedure as described above. To prepare metal-free NDST-1, purified NDST-1 protein was incubated with 1 mM EDTA and then dialyzed using a 1000 molecular weight cut-off membrane against 200 mM NaCl, 25 mM MOPS, pH 7.0 for 24 h to remove EDTA and EDTA/metal ion complex.
Preparation of 8-mer and 18-mer Substrates-8-mer and 18-mer oligosaccharide substrates were elongated from p-nitrophenyl glucuronide (GlcA-pNP) as reported previously (27). Briefly, GlcA-pNP (100 mg) was incubated with UDP-monosaccharide donor (UDP-GlcNAc, UDP-GlcNTFA, or UDP-GlcA) and 30 mg of PmHS2, the bacterial glycosyltransferase (heparosan synthase 2 from Pasteurella multocida) in 200 ml of reaction buffer containing 25 mM Tris (pH 7.2) and 10 mM MnCl 2 . The reaction was incubated for 16 h at room temperature and analyzed by a silica-based polyamine (PAMN) HPLC column (Waters). The product was purified by a C 18 -column (0.75 ϫ 20 cm, Biotage) and eluted with a solution containing 70% acetonitrile and 0.1% trifluoroacetic acid. The reaction cycle was repeated seven times to attain 8-mer with a structure of GlcNAc-GlcA-GlcNTFA-GlcA-GlcNAc-GlcA-GlcNAc-GlcA-pNP. The conversion of the GlcNTFA to GlcNH 2 residue was completed by incubating the 8-mer product with 0.1 M LiOH at 4°C for 1 h. The structure of the product was confirmed by electrospray ionization mass spectrometry (ESI-MS). Approximately 180 mg of the 8-mer substrate was obtained.
The synthesis of the 18-mer substrate essentially followed the same protocol by repeating the elongation reaction 17 times. However, a different purification method was used to prepare the 18-mer substrate. Because the binding affinity of the pNP-tagged oligosaccharides to the C18-column decreases as the size of the oligosaccharide increases, a Q-Sepharose column was introduced to isolate the oligosaccharide intermediates when the sizes were larger than octasaccharide. Under this condition, the sample was loaded onto a Q-Sepharose column that was equilibrated with 25 mM Tris, pH 7.5, and 50 mM NaCl. The oligosaccharide product and intermediates were eluted from the column (1 ϫ 30 cm) via a gradient of NaCl from 50 to 500 mM at a flow rate of 2 ml min Ϫ1 . The product was determined by exploiting the UV absorbance of the pNP tag at 310 nm, and the identity of the product was confirmed by ESI-MS.
Modification followed by reduction with sodium borohydride as described by Shively and Conrad (28). The resultant 35 S-labeled depolymerized polysaccharides were resolved on a Bio-Gel P-10 column (0.75 ϫ 200 cm, Bio-Rad), which was equilibrated with 20 mM Tris (pH 7.5) and 1 M NaCl at a flow rate of 4 ml h Ϫ1 . The amount of 35 S radioactivity from each fraction was determined by scintillation counting.

Results
Divalent Metal Ions Are Required for N-Deacetylase Activity, but Not Required for N-Sulfotransferase Activity-NDST-1 protein has both N-deacetylase and N-sulfotransferase activities. Divalent ions, including Mn 2ϩ , Mg 2ϩ , and Ca 2ϩ , are reportedly required for enzymatic activity (29). However, it was unclear whether the metal ions are necessary for the N-deacetylase activity or for the N-sulfotransferase activity. By choosing an appropriate polysaccharide substrate, we selectively measured the N-sulfotransferase activity or the N-deacetylase activity, respectively, as described in a previous publication (illustrated at bottom of Fig. 2) (30). For example, heparosan was chosen as a substrate to measure the N-deacetylase activity as it contains the GlcNAc residues, whereas chemically deacetylated heparosan was used to measure the activity of N-sulfotransferase as the substrate contains GlcNH 2 residues. We found no effect on the sulfotransferase activity of NDST-1 by the addition of 1 mM EDTA, a chelator for divalent ions (Fig.  2A). The deacetylase activity, on the other hand, was inhibited by the addition of EDTA in a dose-dependent manner (Fig. 2B), suggesting that divalent ions are required for N-deacetylase activity. The addition of different metal ions to the EDTAtreated, metal ion-free, NDST-1 protein demonstrated that divalent ions, including Mn 2ϩ , Mg 2ϩ, Co 2ϩ , and Ca 2ϩ , were able to rescue the N-deacetylase activity of NDST-1, whereas Zn 2ϩ and Cu 2ϩ were not (Fig. 2C). As expected, Mn 2ϩ , Mg 2ϩ , Co 2ϩ , and Ca 2ϩ also rescued the activity of N-deacetylase (data not shown). Further analysis indicated that Zn 2ϩ and Cu 2ϩ completely inhibited N-sulfotransferase activities at the concentration of 10 mM (data not shown). Also, the addition of Zn 2ϩ or Cu 2ϩ was unable to rescue the activity of N-deacetylase.
Previously, Riesenfeld et al. (31) identified the requirement of Mn 2ϩ ions for the N-deacetylase activity, but the authors claimed that Ca 2ϩ and Mg 2ϩ did not activate N-deacetylase.
Here, we demonstrate that both Ca 2ϩ and Mg 2ϩ clearly activate the NDST-1. The different results from the two studies can probably be attributed to the use of different substrates. Heparosan, an unmodified version of HS, was used as a substrate in our present study, whereas chemically modified HS was used in the Riesenfeld study.
The Participation of N-Sulfotransferase Accelerates N-Deacetylase Activity-The reactivity toward N-deacetylase and N-sulfotransferase activity of NDST-1 protein was selectively measured using a synthetic octasaccharide substrate (8-mer substrate). The substrate has a uniquely designed structure of GlcNAc-GlcA-GlcNH 2 -GlcA-GlcNAc-GlcA-GlcNAc-GlcA-pNP, which can be used to test the effect of N-sulfotransferase and N-deacetylase activity as well as both activities. The 8-mer substrate contains one GlcNH 2 residue (at residue 3, Fig. 3A) that is a modification site for N-sulfotransferase activity as well as two GlcNAc residues (residues 1 and 5, Fig. 3A) that are modification sites for N-deacetylase or N-deacetylase/N-sulfotransferase activities. The 8-mer substrate exhibited greater than 95% purity as determined by high resolution anion-exchange chromatography (PAMN-HPLC) analysis (Fig. 3A).
To measure the reactivity to NDST-1 protein displaying N-deacetylase activity, the 8-mer substrate was incubated with the protein and Mn 2ϩ but without the sulfo donor, PAPS. The lack of sulfo donor in the reaction abolished the N-sulfotransferase activity of the NDST-1 enzyme. A single product peak representing about 20% of the reaction mixture was detected by PAMN-HPLC, whereas 80% of the substrate remained unmodified (Fig. 3B). ESI-MS analysis confirmed the product to be deacetylated 8-mer, and the tandem MS analysis indicated that the deacetylation occurred at residue 5 (Fig. 4). The data suggest that N-deacetylase activity, although measurable, was minimal on the 8-mer substrate.
To measure the reactivity to NDST-1 protein displaying N-sulfotransferase activity, the 8-mer substrate was incubated with the protein and PAPS as well as EDTA, where the addition of EDTA was used to inhibit the deacetylase activity of NDST-1 enzyme. Under such condition, the 8-mer substrate was completely converted into 1NS 8-mer as determined by PAMN-HPLC (Fig. 3C). ESI-MS confirmed the product structure, and the tandem MS analysis located the N-sulfation site at residue 3 (data not shown). The data suggest that the 8-mer substrate displayed high reactivity to N-sulfotransferase activity without the participation of N-deacetylase.
To measure the reactivity to NDST-1 protein displaying both N-deacetylase and N-sulfotransferase activities, the 8-mer substrate was incubated with the protein and the sulfo donor, PAPS, in the presence of Mn 2ϩ (Fig. 3D). Three products that differed by the number of GlcNS residues were obtained and resolved by PAMN-HPLC (Fig. 3D). No unmodified 8-mer substrate remained after the reaction, suggesting the substrate has high reactivity to the enzyme displaying both N-deacetylase and N-sulfotransferase activities. It should be noted that to obtain a mixture of products from the 8-mer substrate is consistent with a previous study (23).
Taken together, the data suggest that N-deacetylation is the limiting step during NDST-1-catalyzed conversion of a GlcNAc residue to a GlcNS residue. The rate of N-deacetylation is accelerated by combining with the N-sulfation reaction. Our data are consistent with previous findings using structurally heterogeneous polysaccharide substrates (32,33).
Combination of N-Deacetylase and N-Sulfotransferase Activities Is Essential to Generate N-S Domains-NDST-1 is known for its capability to generate a domain that has consecutive repeating disaccharide units of -GlcA-GlcNS-, known as the N-S domain (23). Here, we examined whether both N-deacetylase and N-sulfotransferase on a single polypeptide are required to generate the N-S domain. To this end, the NDase (Arg 66 -Phe 604 ) and NST (Leu 557 -Arg 882 ) domains were expressed in insect cells using baculovirus expression (Fig. 1C). Using polysaccharide substrates, we found that NDase had comparable N-deacetylase activity to that of NDST-1 (Fig. 5A), and NST had comparable N-sulfotransferase to that of NDST-1 (Fig. 5B), suggesting that both individually expressed NDase and NST are active.
We next compared the ability to generate N-S domains from a mixture of NDase and NST proteins and the wild-type NDST-1 on a synthetic oligosaccharide substrate. To fully demonstrate the ability of the enzymes to form N-S domains, a long octadecasaccharide (18-mer, GlcNAc-(GlcA-GlcNAc) 8 -GlcA-pNP) was synthesized, which has nine potential sites for enzymatic modifications. The purity of the 18-mer substrate was greater than 95% as determined by PAMN-HPLC (Fig. 6A). ESI-MS analysis of the 18-mer revealed its molecular weight to be 3553.6 Ϯ 1, very close to the calculated the molecular weight (3553.0), confirming its structure (Fig. 6B).
Incubation of NDST-1 with the 18-mer substrate resulted in a mixture of 18-mer products that differed in the number of GlcNS residues, ranging from one to seven GlcNS residues (1NS 18-mer to 7NS 18-mer, Fig. 6C). The peaks labeled as 5NS 18-mer, 6NS 18-mer, and 7NS 18-mer were purified, and ESI-MS analyses were conducted to determine the molecular mass (Table 1). Tandem MS analysis confirmed the presence of N-S domains in the 18-mer products (data not shown). Additional products resulting from NDST-1-modified 18-mer substrate, including from 1NS 18-mer to 4NS 18-mer, were not analyzed by ESI-MS because each peak contained more than one component. Taken together, the results demonstrated that NDST-1 effectively modified the 18-mer substrate to generate the anticipated products. Modification of the 18-mer substrate with NDST-1 without PAPS only led to a minor amount of  Starkly contrasting results were obtained from the 18-mer substrate modified by the mixture of two individually active NDase and NST proteins (Fig. 6D). Only a single 1NS product was obtained based on the HPLC analysis although sufficient amounts of NDase and NST proteins were included in the reaction, based on the activity measurement using polysaccharide substrates. Adding additional NDase enzyme into this reaction mixture did not significantly improve the efficiency to generate GlcNS residues, suggesting that two separate proteins displayed very low reactivity to the oligosaccharide substrate (data not shown). Our observation suggested that there is a cooperative effect to produce the N-S domain when both N-deacetylase and N-sulfotransferase activities are present on a single polypeptide. Disruption of the linkage between the N-deacetylase and N-sulfotransferase domains resulted in the inability to generate typical N-S sulfation patterns on the substrate.
To further support the apparent requirements for a bifunctional NDST-1 (not NDase and NST) in generating N-S domains, we compared the distributions of GlcNS residues in the 35 S-labeled polysaccharide products. In this experiment, heparosan (-(GlcNAc-GlcA) n -) was incubated with either NDST-1 or a mixture of NDase and NST, in the presence of   Fig. 7. Under these conditions, nitrous acid depolymerization cleaved -GlcNAc-GlcA-linkages, whereas the -GlcNS-GlcA-linkages remained intact (34). The depolymerized products were analyzed by gel permeation chromatography (Bio-Gel P-10) to resolve the oligosaccharides based on their sizes. As expected, N-[ 35 S]sulfo-heparosan prepared by NDST-1 showed a ladder of 35 S-labeled oligosac-charides from tetrasaccharide to Ͼdecasaccharides, which contained -GlcNS-GlcA-repeating units (Fig. 7A). These data demonstrated that N-S domains were present in the N-[ 35 S]sulfo-heparosan prepared by NDST-1. In contrast, only 35 S-labeled tetrasaccharide was obtained from the depolymerized N-[ 35 S]sulfo-heparosan prepared by a mixture of NDase and NST (Fig. 7B). These data demonstrated that the GlcNS residues in the polysaccharide product occurred in isolation, not in the form of N-S domains. Taken together, our data from the analysis of GlcNS residue distribution in the polysaccharide  Table 1. D, PAMN-HPLC chromatogram of the 18-mer substrate treated with a mixture of NDase and NST. Only a single 1NS 18-mer product was observed with the majority of the substrate unmodified.  products are consistent with those obtained using the structurally defined 18-mer substrate.
We then compared the susceptibility of two differently N-sulfated heparosans to modification of other HS biosynthetic enzymes, such as C 5 -epi and 2-OST (Table 2). One type of N-sulfated heparosan was prepared by wild-type NDST-1, and another was prepared by the mixture of NDase and NST.
Indeed, the N-sulfated heparosan prepared by the mixture of two enzymes showed about 13-fold lower reactivity toward epimerase and 2-OST than the one synthesized by full-length NDST-1. In a control experiment, the two compounds exhibited comparable reactivity to 6-OST-1 modification consistent with this modification not being influenced by the level of N-sulfation (35)(36)(37). The data suggest that change in the N-S domain structures impacts the biosynthesis of HS.

Discussion
As the first modification enzyme in the HS biosynthetic pathway, NDST-1 plays a critical role in producing the GlcNS residues that direct subsequent enzymatic modifications to form HS with elaborate sulfation patterns (38). Here, utilizing highly purified recombinant enzymes and structurally defined oligosaccharide substrates, we investigated the interplay between N-deacetylase and N-sulfotransferase activity of NDST-1 and their roles in forming N-S domains. The crucial innovation in our work is the access to recombinant forms of N-deacetylase and N-sulfotransferase as well as full-length NDST-1 proteins and the use of structurally defined oligosaccharide substrates. Our data suggest that the activity of N-deacetylase can be greatly enhanced by the presence of N-sulfotransferase activity, but such enhancement is only obtained when both activities are present in a single polypeptide. When physically separated, the two independent N-deacetylase and N-sulfotransferase  1843 2489 a Heparosan (50 g) was incubated with NDST-1 (4 g) and 100 M PAPS in 1 ml of reaction buffer overnight. The product was purified by a DEAE column, dialyzed to remove NaCl, and dried. The sample was then reconstituted in 50 l of water. b Heparosan (50 g) was incubated with NDase (4 g) and NST (2 g) and 100 M PAPS in 1 ml of reaction buffer overnight. The product was purified by a DEAE column, dialyzed to remove NaCl, and dried. The sample was then reconstituted in 50 l of water. c Both 2-OST (0.2 g) and C5-epi (0.2 g) were incubated with 5 g of N-sulfated heparosan and 1 ϫ 10 5 cpm [ 35 S]PAPS and 1 M unlabeled PAPS at 37°C for 1 h. The product was loaded onto a DEAE column to determine the amount of 35 S-labeled product. domains do not display the same sulfation patterns as that of the intact NDST-1 protein, diminishing the ability to form HS with desired biological functions. In fact, the bifunctionality of NDST is highly conserved in all organisms that generate HS, supporting the necessity for linking these separate activities to generate N-S domains. The modification mechanism catalyzed by NDST-1 is a complex system involving two enzymatic activities and is influenced by both the macromolecular substrate and the PAPS cofactor. Understanding of the cooperative mechanism will be greatly enhanced by structural information of the full-length NDST-1. Our data suggest that N-deacetylase activity is the limiting step in the process of converting a GlcNAc residue to a GlcNS residue. This is also supported by the fact that NST expressed in E. coli is highly active on a polysaccharide substrate with -(GlcNH 2 -GlcA) n -repeating units (39). The N-deacetylase activity can be regulated by the presence of metal ions. Histone deacetylase, an enzyme that removes the acetyl group from acetylated lysine residues, requires a zinc ion for catalysis (40). Unlike histone deacetylase, N-deacetylase activity from NDST-1 is not mediated by zinc ions. This difference is perhaps not surprising given that these enzymes follow very different reaction mechanisms. In the case of the histone deacetylase, the acetyl group from a lysine residue is transferred to CoA to form acetyl-CoA (41), whereas there is no CoA cofactor involvement in the NDST-catalyzed reactions. Like NDST-1, other deacetylases such as peptidoglycan deacetylase from Streptococcus pneumoniae and acetylxylan esterases from Streptomyces lividans and Clostridium thermocellum reportedly do not require co-factors but do utilize divalent ions such as Zn 2ϩ or Co 2ϩ for catalysis (42,43). The activity of NDST-1 that was depleted with metal ions via dialysis can be rescued by the addition of Ca 2ϩ , Mn 2ϩ , Co 2ϩ , or Mg 2ϩ , suggesting that removal of metal ions from the protein does not cause irreversible structural changes. At the present time, we do not know whether a divalent ion also participates in direct binding to the saccharide substrate.
Although there is no clear evidence to link the concentration of divalent ions to the biosynthesis of HS in vivo, the potential impact of PAPS concentration on the N-S domain formation has been documented (44). Sasaki et al. reported that mutation or inhibition of PAPS transporter, a protein that transports PAPS into the Golgi apparatus to carry out HS biosynthesis, reduced the self-renewal and proliferation of mouse embryonic stem cells. The authors attributed these effects to the change in the activities of NDSTs, including NDST-1 and NDST-2 (44). The concentration of PAPS has been known to play a role in controlling the activity of NDST-1 using a structurally heterogeneous polysaccharide substrate in an in vitro experiment (45). Carlsson et al. (45) concluded that NDST only displayed "random and limited N-deacetylation" in the absence of PAPS, consistent with our observation. In our study, structurally defined oligosaccharide substrates and highly purified recombinant NDase, NST, and NDST-1 were utilized, allowing us to confirm their conclusion and demonstrate the necessity of the deacetylase to be coupled to the sulfotransferase as a bifunctional enzyme.
Previously, it has been suggested that the length of the N-S domains may be controlled by the presence of different isoforms of NDST (11,46). For example, HS from mast cells where NDST-2 is the dominant isoform has much larger N-S domains. Another possible mechanism for regulating the length of N-S domains may be by controlling the cellular concentration of PAPS. At higher concentrations of PAPS, longer N-S domains may be formed because such conditions enhance the activity of NDST-1, whereas shorter N-S domains may be formed if the cellular concentration of PAPS is lower due to a slower reaction rate of NDST-1. Continuing efforts toward understanding the cooperative nature of the N-deacetylase and N-sulfotransferase activities of NDST-1 and other NDST isoforms will help reveal the regulatory mechanisms of HS biosynthesis.
Author Contributions-W. D. expressed the enzymes and characterized the substrate specificity. Y. X. synthesized oligosaccharide substrates. V. P. prepared expression constructs. L. C. P. designed the experiment. J. L. designed the experiment and wrote the manuscript.