Anticoagulant Heparan Sulfate Precursor Structures in F9 Embryonal Carcinoma Cells*

To understand the mechanisms that control anticoagulant heparan sulfate (HSact) biosynthesis, we previously showed that HSact production in the F9 system is determined by the abundance of 3-O-sulfotransferase-1 as well as the size of the HSact precursor pool. In this study, HSact precursor structures have been studied by characterizing [6-3H]GlcN metabolically labeled F9 HS tagged with 3-O-sulfates in vitro by 3′-phosphoadenosine 5′-phospho-35S and purified 3-O-sulfotransferase-1. This later in vitrolabeling allows the regions of HS destined to become the antithrombin (AT)-binding sites to be tagged for subsequent structural studies. It was shown that six 3-O-sulfation sites exist per HSact precursor chain. At least five out of six 3-O-sulfate-tagged oligosaccharides in HSactprecursors bind AT, whereas none of 3-O-sulfate-tagged oligosaccharides from HSinact precursors bind AT. When treated with low pH nitrous or heparitinase, 3-O-sulfate-tagged HSact and HSinact precursors exhibit clearly different structural features. 3-O-Sulfate-tagged HSacthexasaccharides were AT affinity purified and sequenced by chemical and enzymatic degradations. The 3-O-sulfate-tagged HSact hexasaccharides exhibited the following structures, ΔUA-[6-3H]GlcNAc6S-GlcUA-[6-3H]GlcNS335S±6S-IdceA2S-[6-3H]GlcNS6S. The underlined 6- and 3-O-sulfates constitute the most critical groups for AT binding in view of the fact that the precursor hexasaccharides possess all the elements for AT binding except for the 3-O-sulfate moiety. The presence of five potential AT-binding precursor hexasaccharides in all HSact precursor chains demonstrates for the first time the processive assembly of specific sequence in HS. The difference in structures around potential 3-O-sulfate acceptor sites in HSact and HSinact precursors suggests that these precursors might be generated by different concerted assembly mechanisms in the same cell. This study permits us to understand better the nature of the HS biosynthetic pathway that leads to the generation of specific saccharide sequences.

Different heparin/heparan sulfate (HS) 1 sequences bind to a large number of growth factors and cytokines (1)(2)(3)(4)(5)(6), enzymes (7), protease inhibitors (8 -12), virus proteins (13), and selectins (14). Such sequences are usually synthesized in the right place (15) and at the right time (16,17). These HS species are involved in development, angiogenesis, lipid metabolism, coagulation, virus infection, and inflammation (18 -22). To synthesize HS oligosaccharide sequences that bind to these specific protein ligands probably requires the synthesis of specific HS precursor structures. HS precursor structures are then acted upon by a unique sulfotransferase, which is positioned at the end of the biosynthetic pathways and whose levels control the concentration of the specific HS component that is the product of the biosynthetic pathway.
The pentasaccharide sequence, GlcNAc/NS6S-GlcUA-Glc-NS3S6S-IdceA2S-GlcNS6S, represents the minimum sequence for antithrombin (AT) binding, where the boldface 3S and 6S constitute the most critical elements involved in the interaction (8,9). The first sugar residue is either N-acetylated or N-sulfated. The third sugar residue is either with or without a 6-O-sulfate, depending on the source of heparin from which these sequences were originally characterized (23). The ATbinding sequence also exists in HS, but such sequences have never been fully characterized due to limited materials. HS act produced by endothelial cells (1% total HS) is responsible in part for the nonthrombogenic properties of blood vessels (24). Even though it constitutes a small portion of HS total , the relative abundance of the AT-binding sequence is at least 10-fold greater than would be predicted by completely random assembly of disaccharide constituents (25). These observations suggest that production of the AT-binding sequence requires the coordinated action of several biosynthetic enzymes, i.e. the enzymes that catalyze chain polymerization, GlcNAc N-deacetylation and N-sulfation, GlcUA epimerization, 2-O-sulfation of uronic acid residues, and 3-O-and 6-O-sulfation of glucosaminyl residues. Multiple forms of N-sulfotransferases (26 -30) and 3-O-sulfotransferases (31) 2 have been reported recently. Due to the substrate specificity of the enzymes involved, the initial distribution of N-sulfate groups strongly influences the subsequent epimerization and O-sulfations (32,33). The downstream O-sulfation may also be able to influence N-sulfation as well (34). However, the mechanisms that control the coordinated action of these enzymes to generate AT-binding sequences or other sequences in HS and heparin are unknown.
To delineate the biosynthetic pathway that regulates HS act synthesis, our laboratory has purified as well as molecularly cloned 3-O-ST-1 (EC 2. 8.2.23). It was demonstrated that 3-O-ST-1, existing in limited amount, acts upon HS act precursor to produce HS act and HS inact precursor to produce 3-O-sulfated HS inact (31,35). When 3-O-ST-1 is no longer limiting in the F9 cell system, the capacity for HS act generation is determined by the abundance of HS act precursors (36). We have reported that overall HS act and HS inact structures are different at the disaccharide level in the F9 cell system. In vitro 3-O-sulfation with purified 3-O-ST-1 can tag the regions of the HS act precursors destined to become AT-binding sites and allow the HS act precursors to be captured. The tagged regions can then be structurally examined (36). Based on the above observations, we have studied the HS act and HS inact precursor structures by characterizing    5 , which possess the correct positioning of all critical groups except for the absence of the 3-Osulfate residue required for AT binding in all HS act precursor chains. This finding demonstrates for the first time the overall structure of the HS act precursors. This information permits us to speculate about the nature of the HS biosynthetic pathway that leads to the generation of specific oligosaccharide sequences.
HS Preparation-The methods for HS preparation have been described previously (36). In brief, cell monolayers were labeled with 100 -1000 Ci/ml sodium [ 35  H]glucosamine (40 Ci/mmol, ICN) overnight in 1 mM glucose DMEM with 10% (v/v) dialyzed fetal bovine serum. GAGs were isolated from both monolayer and media. Purified GAGs were resuspended in 0.5 M NaBH 4 in 0.4 M NaOH and incubated at 4°C for 24 h to release the chains from the core protein. ␤-Elimination was stopped by adding 5-l aliquots of 5 M acetic acid until bubble formation ceased. The released GAG chains were purified by DEAE-Sepharose (Sigma) chromatography followed by ethanol precipitation and then resuspended in water. GAGs were digested with 20 milliunits of chondroitinase ABC (Seikagaku) in 50 mM Tris-HCl and 50 mM sodium acetate buffer (pH 8.0). Complete digestion of chondroitin sulfate by chondroitinase ABC was ensured by monitoring the extent of conversion of the carrier to disaccharides (100 g ϭ 1.14 absorbance units at 232 nm). HS was purified from chondroitinase-degraded products by phenol/chloroform extraction and ethanol precipitation. After washing the pellets with 0.5 ml of 75% ethanol, HS was suspended in H 2 O for further analysis.
The quality of HS chains prepared was analyzed by anion exchange HPLC (TSK DEAESW, 8 cm ϫ 7.5 mm inner diameter, Tosohaas Inc.). Samples were eluted with a linear gradient of 0.2 to 1 M NaCl in 10 mM KH 2 PO 4 , pH 6.0, containing 0.2% CHAPS at a flow rate of 1 ml/min, and radioactivity in the effluent was determined by on-line liquid scintilla-tion spectrometry (Packard).
The absolute amounts of the unlabeled HS were determined by hydrolyzing HS for 3 h with 6 N HCl and 0.1% phenol (v/v) at 100°C and determining the levels of glucosamine on an Applied Biosystems model 420 Amino Acid Derivatizer with on-line model 130A PTC amino acid analyzer. Were Tagged by 3-O-Sulfation with  Purified 3-O-ST-1 and PAPS-The standard 50-l reaction contains  400 units of purified 3-O-sulfotransferase-1 (3-O-ST-1 35 S. The reactions were incubated at 37°C overnight and then terminated by boiling for 1 min. After adding 100 g of chondroitin sulfate as cold carrier, the radiolabeled HS was purified by phenol/chloroform extraction, DEAE-Sepharose chromatography, and ethanol precipitation. After washing the pellets with 0.5 ml of 75% ethanol, 3-O-sulfate-tagged HS act and HS inact precursors were separated by AT affinity assay as described below.

HS act and HS inact Precursors
Separation of 3-O-Sulfate-tagged HS act and HS inact Precursors by AT Affinity Assay-AT affinity assay was used as described previously (36). In brief, AT complexes were created by mixing 3-O-sulfate-tagged HStotal in 500 l of HB (150 mM NaCl, 10 mM Tris-Cl (pH 7.4)) with 2.5 mM AT, 100 g of chondroitin sulfate, 0.002% Triton X-100, 1 mM each of CaCl 2 , MgCl 2 , and MnCl 2 . 60 l of HB containing ϳ50% concanavalin A-Sepharose 4B was then added. AT complexes were bound to concanavalin A by mixing the reaction mixtures for 1 h at room temperature. Beads were pelleted by a brief centrifugation at 10,000 ϫ g. The supernatant was collected, and the beads were washed three times with 1.25 ml of HB containing 0.0004% Triton X-100. The supernatant and washing solutions were combined as 3-O-sulfate-tagged HS inact precursors. 3-O-Sulfate-tagged HS act precursors were eluted with three successive 200-l washes of HB containing 1.0 M NaCl, 0.0004% Triton X-100 and pooled. After adding 100 g of chondroitin sulfate as cold carrier to 3-O-sulfate-tagged HS act precursors, the pooled 3-O-sulfatetagged HS act and HS inact precursors were cleaned by phenol/chloroform extraction followed by DEAE-Sepharose chromatography and ethanol precipitation. After washing the pellets with 0.5 ml of 75% ethanol and dried briefly by Speed-Vac, 3-O-sulfate-tagged HS act and HS inact precursors were resuspended in H 2 O and used for chemical and enzymatic structural studies.
Low pH Nitrous Acid Degradation-3-O-Sulfate-tagged HS act and HS inact precursors or 3-O-sulfate-tagged tetra-or hexasaccharide were mixed with 10 g of bovine kidney HS (ICN), treated at room temperature with low pH nitrous acid (pH 1.5) for 10 min, and then reduced under alkaline conditions with 0.5 M NaBH 4 for another 10 min. Bio-Gel P6 or P2 chromatography as described below was used to analyze further the resultant products.
Bio-Gel P2 and P6 Gel Filtration Chromatography of Chemically or Enzymatically Degraded 3-O-Sulfate-tagged HS act and HS inact Precursors-Bio-Gel P2 (0.75 ϫ 200 cm) and P6 (0.75 ϫ 200 cm) columns were equilibrated with 100 mM ammonium bicarbonate at a flow rate of 4 ml/h. 200 l of radiolabeled sample mixed with dextran blue (5 g) and phenol red (5 g) was loaded on the column. 0.4 ml per fraction was collected. A 10-l sample from each fraction was counted for 3 H and/or 35 S radioactivity unless otherwise indicated. The desired fractions were pooled and dried by Speed-Vac to remove ammonium bicarbonate and used for further analysis. A small amount of free sulfate generated during the experimental procedures was quantitated by re-running the collected disaccharide fractions on a polyamine HPLC. The free sulfate is subtracted from disaccharides to get the accurate di-, tetra-, and oligosaccharide quantitations indicated under "Results." Dr. Yoshida, Seikagagu Corp., Tokyo, Japan. Heparitinase I recognizes the following sequences: GlcNAc/NSϮ6S(3S?)-2GlcUA-GlcNAc/ NSϮ6S. The arrow indicates the cleavage site. Heparitinase II has broad sequence recognition, GlcNAc/NSϮ6S(3S?)-2GlcUA/IdceAϮ2S-GlcNAc/NSϮ6S. Heparinase and heparitinase IV recognizes the sequences: GlcNAc/NSϮ3SϮ6S-2IdceA2S-GlcNAc/NSϮ6S. The reaction products and references can be found in the Seikagagu's catalog.

Digestion of 3-O-Sulfate-tagged HS act and HS inact Precursors with
The digestion of 3-O-sulfate-tagged HS act and HS inact precursors was carried out in 100 l of 40 mM ammonium acetate (pH 7.0) containing 1 mM CaCl 2 with 2 milliunits of enzyme or 2 milliunits of each heparitinase I, heparitinase II, and heparinase. The digestion was incubated at 37°C overnight unless otherwise indicated.
PAMN High Performance Liquid Chromatography-Separation and characterization of 3-O-sulfate-tagged tetra-and hexasaccharides were carried out by HPLC using an amine-bound silica PA01 column (0.46 ϫ 25 cm) (PAMN-HPLC) (YMC). 100-l aliquots containing 2 nmol of each six known heparan sulfate disaccharide standards (Seikagagu) were included with all analytic runs. The PAMN-HPLC was eluted with H 2 O for 5 min followed by 0 -100% KH 2 PO 4 linear gradient for 100 min at a flow rate of 1 ml/min. The elution of cold disaccharide standards was detected by on-line UV monitoring, and radiolabeled disaccharides were monitored by on-line liquid scintillation spectrometry (Packard Instrument Co.). Each fraction represents a 0.5-ml elution volume.

The HS act Precursors Contain Six Potential AT-binding Sites Whose Structure Is Completed by in Vitro 3-O-Sulfation-We
have previously reported that F9 cells make 1% HS act and 29% HS act precursors, and the metabolically labeled GlcUA-anMan R 3 35 S (6%) and GlcUA-anMan R 3 35 S6 35 S (12%) accounted for about 18% of O-sulfated disaccharides in 1% F9 HS act (36). The high percentage of 3-O-sulfated disaccharides in the biosynthesized HS act chain suggests that there are multiple potential 3-O-sulfation sites in the F9 HS act precursor chain. To estimate this parameter, [6-3 H]GlcN metabolically labeled HS act precursor was isolated by exhaustive conversion to HS act by exposure in vitro to pure 3-O-ST-1 as well as PAPS and subsequent captured by AT affinity chromatography. The 3-O-sulfate-tagged HS act precursors were subjected to a mixture of heparitinase I, heparitinase II, and heparinase treatment. According to a previous publication (38), the combined heparitinase I, heparitinase II, and heparinase digestion should depolymerize all the 3-O-sulfate containing sites into tetrasaccharides and the remaining sugar residues into disaccharides. The depolymerized materials were sized by Bio-Gel P2 chromatography (Fig. 1). 15% of 3 H counts are resistant to digestion and remained as tetrasaccharides. In contrast, 0.5% of 3 H counts are resistant to digestion and remained as tetrasaccharides in [6-3 H]GlcN metabolically labeled F9 HS control (data not shown).
The di-and tetrasaccharides indicated by solid bar in Fig. 1 were collected and subjected to polyamine HPLC chromatography ( Fig. 2). Each disaccharide peak on HPLC was assigned by comparing to standards ("Experimental Procedures"). The two tetrasaccharide peaks that are resistant to the combined treatment of heparitinase I, heparitinase II, and heparinase digestion are likely to represent the 3-O-sulfate-containing tetrasaccharides. We will subsequently show that the first tetrasaccharide peak has the structure ⌬UA-GlcNAc6S-GlcUA-GlcNS3S and the second peak has the structure ⌬UA-GlcNAc6S-GlcUA-GlcNS3S6S.
The di-and tetrasaccharides from Fig. 2 were further analyzed to estimate the number of 3-O-sulfation sites per HS act precursor chain. We first assessed the molecular weight for F9 HS by SDS-polyacrylamide gel electrophoresis as described previously (39) GlcNS3S and 2 ϫ ⌬UA-GlcNAc6S-GlcUA-GlcNS3S6S per chain, therefore a total of six 3-O-sulfate sites per HS act precursor chain.
The sum of N-sulfates, 2-O-sulfate, 6-O-sulfates, and 3-Osulfates for each sugar residues is provided in Table I and equals 100% of the total sulfate groups per 3-O-sulfate-tagged chain, which corresponds to a total of 71 sulfate groups. Of these, 37 sulfate groups or 52% are O-sulfates. Furthermore, since 3-O-sulfates comprise 14% of all metabolically labeled O-sulfates (36). We can calculate that each endogenous HS act chain contains 5.2 3-O-sulfation sites (37 ϫ 14%).
By treating cold F9 HS with PAP 35 S and 3-O-ST-1, we found that 6 pmol of [ 35 S]sulfates from PAP 35 S are transferred to 3-O-sulfate positions in 1 pmol of F9 HS act precursors (data not shown). Thus, from the evidence presented above, we concluded that there are six 3-O-sulfate acceptor sites per HS act precursor chain, which can be captured as 3-O-sulfated tetrasaccharides as indicated above.
We then estimated the number of 3-O-sulfate acceptor sites per HS inact precursor chain. We previously reported that metabolically labeled GlcUA-anMan R 3 35 S (5%) and GlcUA-anMan R 3 35 S6 35 S (5%) accounted for 10% of O-sulfated disaccharides in retinoic acid and dibutyryl cAMP plus theophilline F9 HS inact . However, endogenous retinoic acid and dibutyryl cAMP plus theophilline F9 HS inact are composed of 3-O-sulfated HS inact as well as HS act and HS inact precursors that have never been 3-O-sulfated (36). To obtain a more accurate estimate, cold F9 HS was exhaustively converted with PAP 35 35 S-tagged tetrasaccharide I and tetrasaccharide II from cold F9 HS act were separated by polyamine HPLC chromatography and collected for the tetrasaccharide sequencing analysis. A fraction of each tetrasaccharide collected was rerun on polyamine HPLC (Fig. 3, A and B). Tetrasaccharide I and tetrasaccharide II were first treated with low pH nitrous acid, which cleaves the GlcNSϮ6S-GlcUA/IdceAϮ2S linkage but not the GlcNAcϮ6S-GlcUA/IdceAϮ2S linkage in the tetrasaccharides. However, after low pH nitrous acid treatment and NaBH 4 reduction, both tetrasaccharide peaks remained as tetrasaccharides as judged by the Bio-Gel P2 profile (data not shown), but the elution time on polyamine HPLC was altered (Fig. 3, C and D). The alteration in this parameter corresponds to the lost of one N-sulfate and ring contraction from the reducing end GlcN residue of both tetrasaccharides.
The tetrasaccharides collected were then treated with  (Fig. 1) were further resolved by polyamine HPLC. The elution times of the disaccharide peaks were compared with those of authentic standards and numerically labeled accordingly. The tetrasaccharide fractions indicated by solid bars were pooled as tetrasaccharide I (1) and tetrasaccharide II (2). The numbers correspond with numbers and names in Table I, which summarizes the relative abundance of each of the di-and tetrasaccharides.
ance on polyamine HPLC. This result indicates that there are no 2-O-sulfates on ⌬ 4,5 UA in either tetrasaccharide (data not shown). Indeed, ⌬ 4,5 glycuronidase removed a sugar residue from the non-reducing end in both tetrasaccharides. The polyamine HPLC profiles of ⌬ 4,5 glycuronidase-treated samples are shown in Fig. 3, E and F.
Peak II (56%) and peak III (29%) have the same retention time as the tetrasaccharides in Fig. 3, C and D. The identical polyamine HPLC profiles as in Fig. 3, E-J, were observed upon sequential ⌬ 4,5 glycuronidase, 6-O-sulfatase, and ␣-Nacetylglucosaminidase treatments of the collected peak II and peak III (data not shown). Therefore, peak II has the structure ⌬ 4,5 UA-GlcNAc6S-GlcUA-anMan R 3S and the peak III has the structure ⌬ 4,5 UA-GlcNAc6S-GlcUA-anMan R 3S6S. This result indicates that 34% (85 ϫ 40%) of the 3-O-sulfate acceptor sites in HS inact precursors are the same at the ⌬ 4,5 tetrasaccharide level as in HS act precursors. The 3-O-sulfate-tagged HS act and HS inact precursor ⌬ 4,5 tetrasaccharides do not bind to AT (data not shown). In view of the fact that GlcNAc6S-GlcUA-GlcNS3S6S-IdceA2S has 150-fold less affinity for AT (K d The elution position of peak I on polyamine HPLC suggests that this tetrasaccharide may have only one sulfate group. After ⌬ 4,5 glycuronidase treatment, the resulting products eluted at the same position as GlcNAc-GlcUA-anMan R 3S in Fig. 3G. After ␣-N-acetylglucosaminidase treatment, it coeluted with GlcUA-[ 3 H]anMan R 3S standard. From the above evidence, we concluded that peak I has the structure of ⌬ 4,5 UA-GlcNAc-GlcUA-anMan R 3S. The critical 6-O-sulfate group at residue 2 is missing in this tetrasaccharide. This observation explains the fact that a population of 3-O-sulfate-tagged HS inact precursors within the GlcNAc-GlcUA-GlcNS3S context failed to bind to AT. However, we cannot quantitate how many 6-O-sulfates are missing in the 3-O-sulfate-tagged HS inact precursors within the GlcNS-GlcUA-GlcNS3 35 S context because heparitinase treatment does not preserve all the 3-O-sulfate acceptor sites as tetrasaccharides as compared with the 3-Osulfate-tagged HS act precursors. Our inability to capture all of the 3-O-sulfate-tagged tetrasaccharides from the HS inact precursors is probably due to the heterogeneity in the structure of these precursors. This heterogeneity results from the fact that we have not, as yet, identified the ligands for 3-O-sulfated HS inact materials and hence cannot purify the 3-O-sulfated HS inact precursors.

Five of Six 3-O-Sulfate-tagged Oligosaccharides in HS act Precursors Bind to AT-Both HS act and HS inact precursors contain about six 3-O-sulfation sites per chain. 3-O-Sulfate-tagged HSact precursors binds to AT, and 3-O-sulfate-tagged HS inact precursors do not. The next question we asked is how many of the 3-O-sulfate acceptor sites (potential AT-binding sites) in 3-O-
sulfate-tagged HS act precursors actually bind to AT. We reasoned that heparitinase I digestion should leave AT-binding sites intact since all the 3-O-sulfate-tagged HS act precursors remained as 3-O-sulfate-tagged tetrasaccharide after heparitinase I, heparitinase II, and heparinase digestion (Fig. 1). If HS has the same AT-binding sequence as in heparin, i.e. GlcNAc6S-GlcUA-GlcNS3SϮ6S-IdceA2S-GlcNS6S, the reducing end IdceA2S-GlcNS6S cannot be cleaved by heparitinase I ("Experimental Procedures" (38)). To this end, 3-O-ST-1 and PAP 35 S-labeled, AT affinity purified HS act and HS inact were digested with 2 milliunits of heparitinase I overnight. A fraction of the digested materials from the 3-O-sulfate-tagged HSact and HS inact precursors were used in the AT affinity assay. We found that, for the digested HS act , the extent of AT binding was approximately equal to the undigested HS act control. To check the heparitinase I digestion reaction, the rest of the digested 3-O-sulfate-tagged HS act and HS inact materials were H]HS act precursors were depolymerized with a mixture of heparitinase I, heparitinase II, and heparinase. The resulting unsaturated di-and tetrasaccharides were originally separated on a Bio-Gel P2 column (Fig. 1) and were then further resolved by polyamine HPLC (Fig. 2). The number of each di-and tetrasaccharide per chain was determined in two steps. Initially, we determined the radioactivity per di-or tetrasaccharide. To this end, the sum of radioactivity in all diand tetrasaccharide peaks was divided by 82 to determine the average radioactivity per disaccharide and 41 to determine the average radioactivity per tetrasaccharide. Subsequently, the total radioactivity in each of various di-or tetrasaccharide peak was divided by the average radioactivity per di-or tetrasaccharide to determine the number of a particular di-or tetrasaccharide in a specific peak. The numbers represent the average of two separate experiments. Total N-sulfates, 2-Osulfate, 6-O-sulfates, and 3-O-sulfates per chain were added according to the number of these sulfates in each di-and tetrasaccharide species. It should be noted that these data are consistent with previously published data from our laboratory, which indicates that 3-O-sulfated disaccharides comprise 18% of all metabolic labeled O-sulfated disaccharides in F9 HS act (GlcUA-anMan R 3 35 S, 6%, and GlcUA-anMan R 3 35  analyzed by Bio-Gel P-6 chromatography. The digestion patterns were reproducibly obtained, and sample profiles are provided (Fig. 5). For 3-O-sulfate-tagged HS act precursors, the data show that 85% remained larger than hexasaccharides, 14% were tetrasaccharides, and 0.5% were disaccharides (Fig.  5A). For 3-O-sulfate-tagged HS inact precursors, only 14% remained larger than hexasaccharides, 52% were tetrasaccharides, 33% were disaccharides (Fig. 5B). Each fraction of oligosaccharides equal to or larger than hexasaccharides in 3-O-sulfate-tagged HS act and HS inact precursors were collected and lyophilized for the AT affinity assay. All the oligosaccharides from 3-O-sulfate-tagged HS act precursors bound to AT, but the oligosaccharides from 3-O-sulfatetagged HS inact precursors did not. Since 85% of the 3-O-sulfatetagged HS act oligosaccharides bind to AT and there are six 3-O-sulfation sites per chain, 6 ϫ 85% ϭ 5.1. It implies that at least five out of six 3-O-sulfate acceptor sites possess the correct positioning of all critical groups except for the 3-O-sulfate for AT binding in all HS act precursor chains. This experiment also suggests that HS act and HS inact precursors have distinct sulfation patterns around the potential 3-O-sulfate acceptor sites.
It is interesting to note that 33% of 3-O-sulfate-tagged HS inact precursors can be degraded into disaccharides and 0.5% of 3-O-sulfate-tagged HS act precursors can be degraded into disaccharides. This result implies that 3-O-sulfation is not the only factor that determines whether 3-O-sulfated sequences are preserved as tetrasaccharides following the digestion with heparitinase. PAP 35 S. The 3 H and 35 S double-labeled HS was AT affinity fractionated into 3-O-sulfate-tagged HS act and HS inact precursors. The low pH nitrous-treated and NaBH 4 reduced products were analyzed by Bio-Gel P-6 chromatography (Fig. 6). These conditions depolymerize HS between GlcNS and GlcUA/IdceA residues, liberating oligosaccharides whose length reflects the spacing between N-sulfated GlcN units. The 3 H radioactivity profiles are similar; therefore, the distribution of GlcNS residues are similar in HS act and HS inact precursors outside 3-O-sulfate acceptor sites. However, 3-O-sulfate acceptor sites in HS act and HS inact precursors are different because their 35 S radioactivity profiles are distinct. The 3-O-sulfate acceptor sites in HS act precursors should be located in GlcUA/IdceA-GlcNAc6S-GlcUA-GlcNS3 35 SϮ6S sequences since 88% of 3-Osulfated materials remained as tetrasaccharides (Fig. 6A). In contrast, 3-O-sulfate acceptor sites in HS inact precursors should 35 S-Tagged HS inact precursors were digested with a mixture of heparitinase I, heparitinase II, and heparinase. Tetrasaccharides obtained after Bio-Gel P2 chromatography were treated with low pH nitrous acid and NaBH 4 and subjected to Bio-Gel P2 chromatography again. Tetrasaccharides recovered after low pH nitrous acid and NaBH 4 treatment were resolved by polyamine HPLC. Peak I, peak II, and peak III were collected for further structural analysis. be located mainly in GlcUA/IdceA-GlcNSϮ6S-GlcUA-Glc NS3 35 SϮ6S sequences since 60% of 3-O-sulfated materials was cleaved into disaccharides (Fig. 6B).

FIG. 4. Polyamine HPLC profile of 3-O-35 S-tagged HS inact precursor tetrasaccharides. 3-O-
3-O-Sulfate-tagged HS act Hexasaccharide Structures-The tetrasaccharide analysis, heparitinase I, and low pH nitrous acid treatment data imply that the dominant 3-O-sulfate acceptor sites in HS act precursor are GlcUA/IdceA-GlcNAc6S-GlcUA-GlcNS3 35 SϮ6S-IdceA؎2S, whereas the dominant 3-Osulfate acceptor sites in HS inact precursor are GlcUA/IdceA-GlcNS/Ac؎6S-GlcUA-GlcNS3 35 SϮ6S-GlcUA. These results further clarify why 3-O-sulfate-tagged HS act precursors bind AT and 3-O-sulfate-tagged HS inact precursors do not. To obtain the minimum AT-binding sequence in 3-O-sulfate-tagged HS act precursors, the 3 H and 35 S double-labeled HS act and HS inact precursors were subjected to limited digestion with heparitinase I and heparitinase II. According to recently published work on the exolytic and processive mechanism of depolymerization of HS by heparitinase II (41), this digestion condition should preserve some of the AT-binding hexasaccharide structures. The digested products were analyzed by Bio-Gel P-6 chromatography (Fig. 7). The major hexasaccharide peak from 3-O-sulfate-tagged HS act precursors was collected and lyophilized. AT affinity purified, 3 H and 35 S double-labeled hexasaccharides were loaded on a polyamine HPLC column (Fig. 8). Two peaks (hexasaccharide I and hexasaccharide II) were found and collected for further analysis.
Hexasaccharide I and hexasaccharide II were digested with heparitinase IV. All the hexasaccharides were degraded into di-and tetrasaccharides, which were separated and collected after Bio-Gel P2 chromatography (data not shown). The 3 H-labeled disaccharides from hexasaccharide I and hexasaccharide II both co-eluted with ⌬UA2S-GlcNS6S UV standard on polyamine HPLC (Fig. 9, A  and B). The 3 H and 35 S double-labeled tetrasaccharide from hexasaccharide I has the structure ⌬UA- To determine the identity of the UA2S in both hexasaccharides, hexasaccharide I and hexasaccharide II were treated with low pH nitrous acid and NaBH 4 reduced. The products were fractionated into disaccharide and tetrasaccharide by Bio-Gel P2 chromatography and collected for further analysis. The 3 H-labeled disaccharides from both hexasaccharide I and hexasaccharide II coeluted with IdceA2 35 S-anMan R 6 35 S standard on ion pairing reverse phase HPLC (data not shown). Tetrasaccharide from hexasaccharide I eluted at the same position as ⌬UA-GlcNAc6S-GlcUA-anMan R 3S on polyamine HPLC (Fig.  3C), and tetrasaccharide from hexasaccharide II eluted at the same position as ⌬UA-GlcNAc6S-GlcUA-anMan R 3S6S on polyamine HPLC (Fig. 3D). Combining both hexasaccharide enzymatic and low pH nitrous acid analysis data, we concluded that hexasaccharide I has the structure ⌬UA-GlcNAc6S-GlcUA-G-lcNS3S-IdceA2S-GlcNS6S and hexasaccharide II has the structure ⌬UA-GlcNAc6S-GlcUA-GlcNS3S6S-IdceA2S-GlcNS6S.
Based on the data presented in this paper, the differences between 3-O-sulfate acceptor sites in HS act and HS inact precursors are summarized in Fig. 10. It is important to note that the heterogeneity in the HS inact precursors results in a generation of a large number of potential 3-O-sulfate acceptor sites. Those structures for which we have evidence are provided in the figure. In this regard, the second sugar residue in HS inact precursors is either a GlcNS (60%) or GlcNAc (40%), and the fifth sugar residue in HS inact precursors is either a GlcUA (85%) or IdceA (15%). Furthermore, the 6-O-sulfate groups on residue 2 and 6 of the 3-O-sulfate acceptor site on the HS inact precursors can be present or absent. These alterations distinguish the potential 3-O-sulfate acceptor sites of the HS inact precursors from that of HS act precursors whose sequence is uniquely defined. The many possible combinations of alterations in the HS inact precursor acceptor sites reflect the underlying heterogeneity of the HS inact precursors. For this reason, we are unable to preserve the major 3-O-sulfate acceptor hexasaccharide sequence in HS inact precursors and cannot quantitatively provide the actual sequence of the 3-O-sulfate acceptor sites in HS inact precursors. However, it should be emphasized that the critical 6-O-sulfate is always present in HS act precursors, whereas it can be missing in HS inact precursors. DISCUSSION This article presents an analysis of AT-binding HS act precursor sequences in F9 embryonal carcinoma cells. At least five out of six 3-O-sulfate acceptor oligosaccharides possess the correct positioning of all critical AT-binding groups except 3-O-sulfate in all HS act precursor chains. It is interesting to note that four AT-binding oligosaccharides in each LTA cell HS act chain are present as shown by an AT-protected heparin lyase assay in our laboratory. 3 To possess multiple AT-binding sites in each HS act chain suggests that HS act biosynthesis is not a random process but a highly repetitive, highly organized operation. Based on the data presented in this paper, the schematic model for the structure of HS act precursors in F9 cells is advanced in 3 J. Liu and R. D. Rosenberg, unpublished results.   (Table I). 5 ϫ ⌬UA-GlcNAc6S, 3 ϫ ⌬UA2S-GlcNS, and 3 ϫ ⌬UA-GlcNS6S are the 11 O-sulfated residues outside the AT-binding domain that need to be placed to complete the primary structure of HS act precursor (Fig. 11).

FIG. 7. Bio-Gel P6 fractionation of limit digested 3-O-35 Stagged [ 3 H]HS act and [ 3 H]HS inact precursors by heparitinase I and heparitinase II. 3-O-35 S-Tagged [ 3 H]HS act and [ 3 H]HS
The reason we proposed that there are six instead of five potential AT-binding hexasaccharides in F9 HS act precursors is that we could place five out of six ⌬UA2S-GlcNS6S residues in potential AT-binding domains in F9 HS act precursors. We suspect that heparitinase I may contain trace heparitinase II contaminants that generate 15% of the tetrasaccharides in 3-O-sulfate-tagged HS act precursors (Fig. 5). More likely, endo-D-glucuronidases may cleave mature HS chains limitedly in cultured F9 cells before HS chains were prepared for our structural studies (see "Experimental Procedures"). The general presence of endo-D-glucuronidases in variety of tissue and cells has been reported (41)(42)(43)(44)(45)(46)(47)(48). Endo-D-glucuronidases recognize the sequences GlcUA/IdceA-GlcNAc/NSϮ6S-GlcUA-2GlcNAc/ NSϮ3SϮ6S-IdceA2S-GlcNSϮ6S. The arrow indicates the cleavage site (49,50). The cleavage eliminates both AT-and potential AT-binding sites (50). The endo-D-glucuronidases extensively cleave the newly synthesized heparin (M r 60,000 -100,000) to generate fragments that are stored in cytoplasmic granules of mast cells (M r 5,000 -25,000). This may explain why commercial heparin (M r 5,000 -25,000) contains limited numbers of AT-binding sites per chain (38). In contrast, endo-D-glucuronidases may cleave in a limited fashion the endogenous HS chains (48,49). This limited cleavage may explain why we can place five out of six ⌬UA2S-GlcNS6S residues in potential AT-binding domains in F9 HS act precursors.
It is apparent from our model that there is a template for HS act precursor formation that requires correct N-sulfation, epimerization, 2-O-, and 6-O-sulfation at all six sites along the chain during the biosynthesis. What remains to be explained is how this is accomplished. F9 HS act chain contains almost all the detectable 3-O-sulfated disaccharides (GlcUA-anMan R 3S and GlcUA-anMan R 3S6S) in HS total . The 3-O-sulfated disaccharides accounted for about 18% of O-sulfated disaccharides in the HS act chain (36). Our current study showed that there are six 3-O-sulfate acceptor sites per HS act chain. Therefore  Fig. 7 were AT affinity purified and separated by polyamine HPLC. The hexasaccharide I and the hexasaccharide II resolved by polyamine HPLC were collected for further analysis.  (52)). The structure of each subpopulation HS is determined by the sequential action of modification enzymes, i.e. pathway, in the Golgi apparatus. In other words, different pathways during HS modification generate different HS structures in the same cell.
This study shows that F9 cells produce different HS structures. The previous HS act -deficient mutants and 3-O-ST-1 studies in our laboratory suggested that the same cells make both HS act and HS inact precursors (35,53), and the same HS proteoglycan core proteins carry both HS act and HS inact chains (54). A Chinese hamster ovary cell mutant defective in N-ST makes both fully sulfated and undersulfated HS chains (55). A human colon carcinoma cell makes both fully sulfated and undersulfated HS chains (56). All these observations suggest that different biosynthesis schemes that generate different HS structures occur in the same cells. In addition to the increasing numbers of specific HS domain structural studies in different tissues and cells (57)(58)(59)(60)(61), we suggest that different HS biosynthetic pathways exist, which generate HS with different structures and biologic functions. Evaluation of the different pathways will eventually delineate how the HS biosynthesis is regulated.
Currently we do not know where HS act and HS inact precursor biosynthesis pathways diverge. The constant presence of the critical 6-O-sulfate groups only in HS act precursors indicates that the pathway is either set up by or diverges before the critical 6-O-sulfation. It is possible that specific isoforms of 6-O-ST are involved in HS act precursor pathways, and the existing structures around the 3-O-sulfate acceptor site favor the action of a 6-O-ST isoform. Furthermore, the existence of auxiliary proteins that modulate the action of N-ST, epimerase, 2-O-ST, and 6-O-ST for HS act precursor formation have not been excluded.
In conclusion, the decision to synthesize HS act or HS inact depends on the presence of specific HS precursor intermediates, specific modification enzymes, and perhaps auxiliary factors in the Golgi biosynthesis machinery. Currently, we have obtained a series of Chinese hamster ovary mutants defective in HS act precursor formation. Complementation of these mutations will provide us with molecular details about the required elements for HS act precursor biosynthesis. This information should allow us to formulate a more definitive model of the HS act biosynthetic pathway. This model will be evaluated by reconstituting the sequential biosynthetic apparatus using different recombinant modification enzymes/proteins.