Determining Heparan Sulfate Structure in the Vicinity of Specific Sulfotransferase Recognition Sites by Mass Spectrometry*

Sulfated motifs on heparan sulfate (HS) are involved in various extracellular processes from cell signaling to enzymatic regulation, but the structures of these motifs are obscure. We have developed a strategy to determine the structure of sulfotransferase recognition sites which constitute these motifs. Stable isotope is first introduced into specific sites on HS with HS sulfotransferases and the modified HS is then digested into oligosaccharides of differing sizes. The overlapping oligosaccharides containing the introduced stable isotope are identified by changes in the m/z profiles by mass spectrometry, and their relationships are elucidated. In this way, the HS structures in the vicinity of the sulfotransferase recognition site are quickly determined and groups on precursor structures of HS that direct the action of HS sulfotransferases are pinpointed.

Heparan sulfate (HS) 1 is one major polysaccharide found on proteoglycans. HS chains are first synthesized in the Golgi apparatus as repeated units of the disaccharide of a glucuronic acid and an N-acetylated glucosamine (GlcA-GlcNAc) n (1). The glucuronic acid can be epimerized to iduronic acid (IdoA). Incomplete sulfation at the 2-O positions of the uronic acids, and the 3-O, 6-O, and N positions of the glucosamine by various sulfotransferases results in structural diversity within HS (1,2). The sulfate groups usually cluster in small regions and form sulfated motifs.
Although HS plays important roles in various biological processes, the structures of these biological motifs are obscure, due to the difficulties involved in obtaining homogeneous components and determining their structures. Cloning, expressing, and sequencing biopolymers have tremendously advanced our understanding of DNA and protein, but no similar methods are available for studying HS.
In this report, we describe a novel strategy that permits us to quickly determine specific HS structures in the vicinity of the sulfotransferase recognition site for 3-OST-4. HS 3-OSTs are rare modification enzymes that helps to generate binding sites for proteins such as antithrombin III (2) and herpes simplex virus glycoprotein D (11). 3-OST-4 is specifically expressed in brain tissues and may play a role in neuronal development (14). The delineation of sulfated motifs on the HS chain of proteoglycans will shed light on how cells interact with extracellular proteins, respond to extracellular signals, and initiate signaling cascade beginning on the cell membrane (15). This strategy relies on introducing a stable isotope of sulfate into specific sites on the HS chain by 3-OST-4. The mass-labeled HS is then digested into oligosaccharides and the overlapping oligosaccharides containing the stable isotope are identified by changes in their masses.
Sample Preparation-Bovine kidney heparan sulfates were labeled with sulfotransferases in the presence of 34 S, 33 S, or [ 35 S]PAPS. The labeling reaction was carried out in a 20-l reaction with 10 g of HS, 2 l of 2ϫ buffer (50 mM MES (pH 7.0), 1% (w/v) Triton X-100, 5 mM MgCl 2 , 5 mM MnCl 2 , 2.5 mM CaCl 2 , 0.075 mg/ml protamine chloride, 1.5 mg/ml bovine serum albumin), 1 l of 2 mM [ 34 S]PAPS, or 2 l of [ 35 S]PAPS (about 1.0 ϫ 10 7 cpm) and 20 ng of expressed pure sulfotransferase. The reaction was incubated at 37°C for 2 h, and the modified HS was purified with a DEAE column. The HS sample was loaded onto 0.2 ml of DEAE affinity matrix in a small purification column, washed with 2 ml of 0.25 M NaCl, 20 mM NaAc (pH 6.0), and eluted with 0.4 ml of 1 M NaCl, 20 mM NaAc (pH 6.0). The HS was then precipitated with 1 ml of ethanol and centrifuged at 14,000 ϫ g for 30 min in a 4°C room. Because the incorporation of [ 35 S]PAPS into HS can be easily monitored, labeling with [ 35 S]PAPS was performed in a parallel fashion. The labeled HS samples were then subjected to heparan sulfate lyase digestion at 37°C for 2 h to completion according to Seikagaku's protocols. The digestion buffer contained 40 mM ammonium acetate (pH 7.0), 1 mM CaCl 2 , and 1 milliunit of enzyme. The digestion of the 35 S-labeled HS was analyzed by polyacrylamide gel electrophoresis and the digestion of the 34 S-or 33 S-labeled HS was analyzed by liquid chromatography-coupled mass spectrometry (LC/MS).
Gel Electrophoresis and Autoradiograph-35 S-Labeled HS and oligosaccharides were analyzed with 12.5% polyacrylamide gel (with 0.3% of * This work was supported in part by the National Institutes of Health Grants 1PO1 HL 66105 and 5RO1HL 59479. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. bisacrylamide) electrophoresis. About 10,000 cpm of the sample was loaded onto each lane. The gel buffer contained 10 mM Tris (pH 7.4) and 1 mM EDTA, and the electrophoresis buffer contained 40 mM Tris (pH 8.0), 40 mM acetic acid, 1 mM EDTA. The gel was run at 10 volts/cm for 30 min with a SE 250 Mighty Small II gel apparatus (Hoefer Scientific Instruments, San Francisco). After electrophoresis, the gel was transferred to a gel blotting paper and dried under vacuum. The dried gel was autoradiographed by a PhosphorImager 445SI (Amersham Biosciences).
Flow Injection Capillary Liquid Chromatography and Mass Spectrometry-HPLC and mass analysis was described previously (17). An Ultimate capillary HPLC work station (Dionex, Sunnyvale, CA) was used for separation. UltiChrom software was used in data acquisition, analysis, and management. Dibutylamine was used as an ion-pairing agent. A gradient elution was performed using a binary solvent system composed of water (eluent A) and 70% aqueous methanol (eluent B), both containing 8 mM acetic acid and 5 mM ion-pairing agent. HPLC separations were performed on a 0.3 ϫ 250-mm C18 column (MS 5 m) (Vydac (www.vydac.com)). The flow rate was set to 5 l/min. Sample volume of 6.3 l was injected. The elution profile was 0% B for 5 min, 6% B for 19 min, 18% B for 17 min, 34% B for 13 min and 55% B for 16 min. After the run, the column was washed with 90% B for 15 min and equilibrated with 100% A for 28 min.
Mass spectra were acquired on a Mariner BioSpectrometry work station electrospray ionization time-of-flight mass spectrometer (Per-Septive Biosystems, Framingham, MA). Nitrogen was used as a desolvation gas as well as a nebulizer. Conditions for electrospray ionization-MS were as follows: nebulizer flow rate, 1 liter/min; nozzle temperature, 140°C; drying gas (N 2 ) flow rate, 0.6 liter/min; spray tip potential, 2.8 kV; nozzle potential, 70 V; and skimmer potential, 9 V. Negative ion spectra were generated by scanning the range of m/z from 99 to 2000. Total ion chromatograms and mass spectra were processed with the Data Explorer software, version 3.0.

Calculation of the Molecular Weights and m/z Ratios of HS
Oligosaccharides-The molecular weight of an HS oligosaccharide is the total weight of its functional groups and backbone structure (Fig. 1A). Based on major isotopes of 12 C, 1 H, 16 O, 14 N, and 32 S, the basic disaccharide GlcA-GlcNH 2 (C 12 H 19 -O 10 N 1 ) has a molecular mass of 337.09; one sulfation increases the mass by 79.96 (-SO 3 -); one acetylation (-COCH 2 -) increases the mass by 42.01. Therefore, the molecular mass (w) of an oligosaccharide with a formula of (C 12 H 19 where p, number of disaccharide units; q, number of sulfates; t, number of acetyl groups. For an oligosaccharide with z negative charges due to loss of protons, the mass/charge ratio (m/z) will be as follows.
The m/z values of the observed oligosaccharides in our study were calculated with Equations 1 and 2 ( Table I).
Because of the existence of minor natural stable isotopes, such as 13 C, 18 O, 15 N, 33 S, and 34 S, the actual molecular mass of an oligosaccharide is a composite of w, w ϩ 1, w ϩ 2, w ϩ 3, etc. Subsequently, the m/z value will also be a cluster of (w Ϫ z)/z, (w ϩ 1)/z, (w ϩ 2 Ϫ z)/z, (w ϩ 3 Ϫ z)/z, etc., with neighboring values 1/z unit apart (Fig. 1B). Therefore, the distance between the neighboring peaks in a m/z profile reveals the number of the negative charge z on the molecule. Because the intensity of FIG. 1. Oligosaccharide structures and its m/z profiles. A, an oligosaccharide structure generated by heparan sulfate lyases. The oligosaccharide has a formula of (C 12 H 19 O 10 N 1 ) p (SO 3 ) q (COCH 3 ) t . p, number of the disaccharides; q, number of the sulfate groups; t, number of the acetyl groups. The wavy lines indicate the uncertainty of the epimerization. B, in vitro stable isotope incorporation changes the m/z profile of an oligosaccharide. The natural existence of multiple stable isotopes of carbon, hydrogen, oxygen, and sulfur cause the multiple peaks in an m/z profile. The distance between the neighboring peaks in a profile equal to 1/z. The in vitro incorporation of one 33 S or one 34 S increases the second or the third peak respectively. w, molecular mass; z, the number of negative charges.
the peaks in a m/z profile is dependent on the natural abundance of each isotope, the ratio among the intensities of individual peaks is also a constant. In vitro incorporation of 33 S or 34 S will enhance the second or third peak respectively due to 1 or 2 mass units increase over the major isotope 32 S (Fig. 1B).

LC/MS Study of 3-OST-4-modified Heparan
Sulfate-HS samples were first labeled with stable isotope 34 S or radioisotope 35 S by 3-OST-4 (16) and then digested with various heparan sulfate lyases. The digestions of the 35 S-labeled HS were analyzed by PAGE (Fig. 2). It was found that heparitinase II alone generated 35 S-labeled hexasaccharide (dp6) and tetrasaccharide (dp4); heparitinase was able to convert the dp6 to dp4; and heparinase was able to convert the dp4 to disaccharide (dp2). The sizes of the labeled oligosaccharides were determined with defined oligosaccharides (18).
In a parallel fashion, the digestions of the stable isotope 34 S-labeled HS were analyzed by LC/MS and all the 34 S-labeled oligosaccharides were located on the HPLC chromatograms by examining the m/z profiles (Fig. 3). It was noted that actually two hexasaccharides (peaks a and b) and two tetrasaccharides (peaks c and d) were labeled with 34 S in the heparitinase II digestion (Fig. 3A, upper panel). Mass profiles showed the two hexasaccharides with regular m/z of 685.62 and 725.62 and the two tetrasaccharides of 496.08 and 536.06, respectively (Fig.  3B). These results identified the two hexasaccharides as dp6 -4S1Ac (four sulfates and one acetyl group) and dp6 -5S1Ac (five sulfates and one acetyl group) and the two tetrasaccharides as dp4 -4S and dp4 -5S (four or five sulfates, respectively) ( Table  I). Addition of heparitinase to the digestion converted the two hexasaccharides to the two tetrasaccharides (Fig. 3A, middle  panel). The further addition of heparinase converted the two tetrasaccharides to two disaccharides (peaks e and f ) (Fig. 3A,  lower panel). The disaccharide in peak e exhibited a regular m/z of 576.04 and z of 1 and was then identified as dp2-3S (three sulfates). The disaccharide in peak f exhibited a regular m/z of 785.19 and z of 1 and was then identified as a quasi-complex between dp2-4S (four sulfates) and dibutylamine, an ion-pairing reagent with the molecular mass of 129.15 (785. 19 Ϫ 129.15 ϭ 656.04) ( Table I). Due to its high charge density, the dp2-4S has a strong tendency to form complexes with dibutylamine.
When the above experiments were repeated with 33 S labeling, the oligosaccharides identified above showed 33 S incorporation (Fig. 3C).
Determining the Variable Sulfation on the Oligosaccharides-Two disaccharides with three or four sulfates, two tetrasaccharides with four or five sulfates and two hexasaccharides with four or five sulfates were observed to contain 3-OST-4 introduced 34 S. It was obvious that one particular sulfation was variable. To identify the site of this sulfation, HS samples were first modified by 3-OST-4, and then modified by other sulfo-transferases (16), both in the presence of [ 34 S]PAPS. The doubly modified HS samples were digested with heparan sulfate lyases and analyzed by LC/MS. It was found that 6-OST-1 was able to convert the dp2-3S to the dp2-4S and the dp4 -4S to the dp4 -5S (Fig. 4). The enhancement of the peaks at 789.17 and 538.05 in the m/z profiles of the dp2-4S and dp4 -5S indicated the incorporation of the second 34 S into the dp2-3S and dp4 -4S, respectively. This experiment showed that a 6-Osulfate on the oligosaccharides was variable for the sulfotransferase activity of 3-OST-4.
Determining the Structure of the 34 S-Labeled Oligosaccharides-Two disaccharides, dp2-3S and dp2-4S, contained the 3-OST-4-incorporated 34 S. Because only four positions on an HS disaccharide (2-O, 3-O, 6-O, and the N) can be sulfated by sulfotransferases, the dp2-4S must be fully sulfated. The fact that 6-OST-1 was capable of adding one sulfate group onto the dp2-3S indicates that the 6-O position in the disaccharide was not sulfated. Therefore, the two disaccharides have the following structure. Two tetrasaccharides dp4 -4S and dp4 -5S contained in vitro incorporated 34 S. Because these two tetrasaccharides could be further converted into the two previously identified dp2-3S and dp2-4S by heparinase digestion, the only explanation for this phenomenon is that the dp4 -4S and dp4 -5S each lost a monosulfated disaccharide during enzymatic digestion. This monosulfated disaccharide is likely to be ⌬UA-GlcNS, because this is a common disaccharide found in enzyme cleaved HS (1)  and no monosulfated free amine containing disaccharide has been reported so far. Since heparinase cuts the GlcNSϮ6S-IdoA2S linkage (19), the tetrasaccharides should have the following structure.
The order of the disaccharides in the tetrasaccharides was also confirmed by treatment with the exo-enzyme iduronate-2-sulfatase, which removed one sulfate only from the dp2-3S or dp2-4S (Fig. 5) but not from the tetrasaccharides, indicating that the dp2-3S and dp2-4S should locate at the reducing sides of the tetrasaccharides. The oligosaccharides in peaks g and h were identified as a dp2-2S and a dp2-3S, as they had regular m/z of 496.04 and 576.02, respectively (Table I). Because these two disaccharides were derived from the ⌬UA2S-GlcNS3SϮ6S by taking off the 2-O-sulfate, they should have the structures of ⌬UA-GlcNS3SϮ6S. The dp2-3S in peak h and the dp2-3S in peak e had structures of ⌬UA-GlcNS3S6S and ⌬UA2S-GlcNS3S, respectively, and were eluted at slightly different time points. Two hexasaccharides dp6 -4S1Ac and dp6 -5S1Ac contained in vitro incorporated 34 S and they could be converted by heparitinase to the previously identified dp4 -4S and dp4 -5S. It is apparent that each hexasaccharide lost one acetylated disaccharide ⌬UA-GlcNAc (dp2-1Ac) during the enzymatic diges-tion. Because heparitinase cuts the GlcNAcϮ6S-GlcA linkage (20), the two hexasaccharides should have the following structure.

DISCUSSION
We have developed a strategy to rapidly obtain the structure of HS in the vicinity of a specific sulfotransferase recognition site. This site is labeled with a stable sulfur isotope by the sulfotransferase. The modified HS is then digested with enzymes in a controlled manner to oligosaccharides of different sizes. The overlapping oligosaccharides containing the incorporated isotope are identified by mass spectrometry. The precise mass measurement also establishes the sizes of the oligosaccharides and the number of sulfate and acetyl groups on each oligosaccharide. The disaccharide arrangement in the oligosaccharides can be distinguished by substrate specificities of the heparan sulfate lyases and with the help of exoglycosidases and sulfatases. In this manner, the structure in the vicinity of the introduced isotope can be deduced. The size of the sequence obtained is dependent on the conditions and methods used for HS cleavage. More than one stable isotope and enzyme can be employed to explore the structural information on HS; for example, 33 S can be incorporated by one sulfotransferase and 34 S by another. In this way, the distance between the two recognition sites, and thus the relationship between the two sulfotransferases, can be established.
Since most, if not all, heparan sulfate sulfotransferases have been cloned recently and all these enzymes can be assumed to possess different substrate specificities, it should be possible to obtain the structures in the vicinity of the recognition sites of all the sulfotransferases. The results will reveal the substrate specificities of those enzymes as well as how different critical groups direct the action of the subsequent sulfotransferases and eventually how the biological relevant motif structures on heparan sulfate are generated. Recently, we have also developed methods to in vitro synthesize functional HS structures with the cloned enzymes (16,21,22). With these technology advances, it is possible for us to define the biological functions of HS motifs by synthesizing the defined HS structures.
It is interesting to note that the upstream half of the hexasaccharide structures ⌬UA-GlcNAc-GlcA-GlcNS-IdoA2S-GlcNS3SϮ6S is non-sulfated and the downstream half is sulfated. At the disaccharide level, 3-OST-4 recognizes heavily sulfated disaccharides and generates tri-and tetra-sulfated disaccharides. Tetra-sulfated disaccharide is the most heavily sulfated disaccharide in HS and is also found in 3-OST-5 generated products (23). The biological significance of this disaccharide is under further investigation. The 6-O-sulfate that resides in the same glucosamine residue where 3-O sulfation occurred was found to be variable, indicating that this particular sulfate is not required for 3-OST-4 recognition. The fact that this 6-O-sulfate was also found variable in 3-OST-1 and 3-OST-3 recognition (10) suggests that 3-O-sulfotransferases do not make contacts with this 6-O-sulfate during substrate recognition. S and physically could not be separated from the one 34 S containing dp2-4S or dp4 -5S. As a result, the fifth peaks in the m/z profiles were enhanced.
FIG. 5. One sulfate could be removed from each of the dp2-3S and dp2-4S by iduronate-2-sulfatase. A, ion current extractions (XIC) of all 34 Slabeled oligosaccharides from the chromatograms of completely digested HS samples before and after the iduronate-2sulfatase treatment. The names of the peaks follow that of Fig. 3A. After the treatment, peaks e and f were converted to peaks g and h. The small peak before peak g was due to the disaccharide ⌬UA-GlcNS6S (regular mass 496.01). B, m/z profiles of the 34 S-containing oligosaccharides in the peaks of g and h. The oligosaccharides in peaks g and h had regular m/z of 496.04 and 576.02 and were identified as dp2-2S and dp2-3S, respectively.
We have also investigated the heparan sulfate structure at the vicinity of 3-OST-1 recognition site. Our current data (data not shown) confirmed the previously reported tetrasaccharide structure ⌬UA-GlcNAc6S-GlcA-GlcNS3SϮ6S (24). Comparing with the tetrasaccharide structure at 3-OST-4 recognition site (⌬UA-GlcNS-IdoA2S-GlcNS3SϮ6S), we found that the two internal sugar residues in the tetrasaccharides are different. We surmise that an N-acetylated glucosamine and a following GlcA are involved in directing the action of 3-OST-1, whereas an N-sulfated glucosamine and a following 2-O-sulfated IdoA are involved in the action of 3-OST-4. The differences on the HS structures recognized by the two sulfotransferases indicate that they have different biological functions.