Chemoenzymatic Synthesis of Classical and Non-classical Anticoagulant Heparan Sulfate Polysaccharides*

Heparan sulfate (HS) polysaccharides interact with numerous proteins at the cell surface and orchestrate many different biological functions. Though many functions of HS are well established, only a few specific structures can be attributed to HS functions. The extreme diversity of HS makes chemical synthesis of specific bioactive HS structures a cumbersome and tedious undertaking that requires laborious and careful functional group manipulations. Now that many of the enzymes involved in HS biosynthesis are characterized, we show in this study how one can rapidly and easily assemble bioactive HS structures with a set of cloned enzymes. We have demonstrated the feasibility of this new approach to rapidly assemble antithrombin III-binding classical and non-classical anticoagulant polysaccharide structures for the first time.

Heparin and heparan sulfate (HS) 1 are linear, sulfated polysaccharides that interact with numerous proteins to regulate various biological processes such as the cell cycle, cell growth, cellular differentiation, cell adhesion, anticoagulation, and lipid metabolism (1,2). Heparin and heparan sulfate are synthesized as proteoglycans that consist of multiple polysaccharide chains attached to various distinct core proteins. The extensive studies of the biosynthesis of heparin and heparan sulfate have delineated the following sequences of events (3,4): A non-sulfated polysaccharide composed of repeating disaccharide units (4GlcA␤1-4GlcNAc␣1) is extended from a common tetrasaccharide that is covalently linked to a core protein. In the presence of 3Ј-phosphoadenosine 5Ј-phosphosulfate (PAPS), a series of modifications takes place, beginning with N-deacetylation and N-sulfation of the N-acetyl glucosamine (GlcNAc) units, followed by epimerization of glucuronic acid (GlcA) to iduronic acid (IdoA), and concluded by stepwise Osulfation at several positions. Heparin, produced mainly by mast cells, contains predominately N-and O-sulfated, IdoArich sequences, whereas heparan sulfate, ubiquitously present on the surface of all cells, possesses domain structures that consist of highly sulfated, IdoA-rich regions separated by more extended, unmodified regions.
Interactions of various proteins with HS are implicated in many physiological and pathological process (1). Moreover, many pathogens, such as parasites, bacteria, and viruses are known to enter host cells through specific interactions with the cell surface HS. Many of the proteins that interact with HS are well characterized, but the polysaccharide structure that mediates a particular interaction has been defined only in a few cases such as antithrombin III (ATIII) and fibroblast growth factor 2 (FGF2) binding HS structures. This problem is primarily attributed to the structural complexity of HS, which, in turn, arises from a complex biosynthetic pathway (3,5).
Synthesis of HS fragments would lead to a deciphering of the information encoded within HS molecules that permits these components to regulate many different biological systems as described above and also aid in the design of new drugs based on HS. HS fragments can be synthesized either by chemical routes or enzymatic routes, as demonstrated in this study. Chemical synthesis would allow one to design unusual or nonnatural structural motifs to understand enzyme functions and to design drugs with improved pharmacokinetic and pharmacodynamic characteristics. However, the total chemical synthesis of HS fragments is a challenging undertaking and often involves 50 or more individual steps. The stereoselective glycosidic bond formation is a daunting task in carbohydrate chemical synthesis despite several elegant advances in glycosylation procedures. Glucosamine is ␣-glycosidically linked to adjacent glucuronic acid in HS. It is difficult to form this cis-glycosidic bond (␣-linkage) with high specificity and masking of the amino function as a non-participating group is required to generate the ␣-linkage. In addition, the carboxyl group is generally masked as a protected hydroxyl group during glycosylation, because the presence of this functional group renders GlcA/IdoA a poor glycosyl donor or acceptor.
The enzymatic method for HS assembly could allow for a more rapid synthesis of structures of interest, which would facilitate the establishment of structure-function relationships of this class of molecules. This approach requires at least a dozen enzymes/isoforms involved in HS biosynthesis, and the combination of enzymes needed to generate a specific HS structure is unknown. To examine the potential of this method, we carried out enzymatic synthesis of biologically active HS polysaccharides. Since anticoagulant HS structures are well established, we initiated our explorations to determine the practicability of generating anticoagulant polysaccharides.
Heparin accelerates the inactivation of blood coagulation enzymes by ATIII (6, 7). A unique pentasaccharide domain was found to bind to ATIII in a highly specific way and to promote rapid inhibition of the blood coagulation enzymes. The critical structural features of the pentasaccharide required for the anticoagulant function are 3-O and 6-O sulfates of residues C and A, respectively ( Fig. 1) (8 -12). These two residues are thermodynamically linked and serve as pincer to induce the conformational change in ATIII that is required for accelerating coagulation enzyme inactivation (13).
Animal-derived heparin has been in use over seven decades as an anticoagulant. Despite having undesirable side effects such as bleeding and heparin-induced thrombocytopenia (HIT) with arterial thrombosis, it is still the drug of choice in combination therapy to treat humans after strokes and heart attacks. Recent concern about the potential spread of bovine spongiphorm encephalopathy to humans has increased interest in the development of alternatives to animal sources of heparin. Sinay et al. (14) pioneered the original chemical synthesis of the ATIII-binding pentasaccharide. We describe here our enzymatic approach to rapidly synthesize ATIII-binding classical and non-classical anticoagulant polysaccharide structures for the first time. We should like to note our companion investigations, which explored different strategies and resolved different problems of this novel synthetic approach (15,16). In those studies, we have shown that enzymes whose actions do not interfere with each other can be employed simultaneously in order to shorten the time of synthesis while generating the desired products. Finally, we also demonstrated that the sequential use of enzymes whose actions facilitate each other can be utilized to achieve the regional selectivity and hence lead to the production of homogeneous oligosaccharide of defined structures.

Digestion of Polysaccharides with Heparitinase I, Heparitinase II, and Heparinase
Polysaccharides were digested with 1 mU of Hep1, -II, and -III in a total volume of 100 l of 40 mM ammonium acetate buffer (pH 7.0) containing 3.3 mM calcium chloride at 37°C overnight.

High Performance Liquid Chromatography (HPLC)
Separation and characterization of 35 S-labeled disaccharides were carried out by HPLC using a C18-reversed phase column (0.46 ϫ 25 cm) (RPIP-HPLC) (Vydac). Solvent A was double-distilled water containing 10 mM ammonium dihydrophosphate and 1 mM tetrabutylammonium dihydrophosphate. Solvent B was 40% acetonitrile containing 10 mM ammonium dihydrophosphate and 1 mM tetrabutylammonium dihydrophosphate. The RPIP-HPLC was eluted at a flow rate of 0.5 ml/min with the following stepwise gradients: 100% solvent A for 15 min; 6% solvent B for 25 min; 12% solvent B for 40 min; 40% solvent B for 60 min; 100% solvent B for 10 min and finally 100% solvent A for 20 min to reequilibrate the column.
Flow Injection Capillary Liquid Chromatography-An ultimate capillary HPLC workstation (Dionex) was used for microseparation. Ul-tiChrom software was used in data acquisition and analysis. 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 dibutylamine as an ion-pairing agent. HPLC separations were performed on a 0.3 mm ϫ 250 mm C18 polymeric silica column (Vydac). The column temperature was maintained at 25°C, and the flow rate was set to 5 l min Ϫ1 . Sample volumes of 6.3 l were injected. For disaccharide analysis, a 20-l sample injection loop was used. The chromatographic conditions were optimized for resolution of disaccharides. In brief, non-sulfated disaccharide was eluted with 100% A, single sulfated disaccharides were eluted with 10% B, isocratic elution with 20% B for double-sulfated disaccharides, followed by isocratic elution with 35% B for triple sulfated disaccharide. The column was washed and equilibrated by further elution with 100% B for 10 min, returning to 100% A for 10 min at the end of the run. The absorbance of the column eluate was monitored at 232 nm.

Mass Spectrometry
Mass spectra were acquired on a Mariner BioSpectrometry Workstation ESI time-of-flight mass spectrometer (PerSeptive Biosystems, Framingham, MA). In the negative-ion mode, the instrument was calibrated with bis-trifluoromethyl benzoic acid, heptadecafluorononanoic acid, and perfluorotetradecanoic acid. Nitrogen was used as a desolvation gas as well as a nebulizer. Conditions for ESI-MS were as follows: nebulizer flow 1 liter/min, nozzle temperature 140°C, drying gas (N 2 ) flow 0.6 liters/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 40 -2000. During analyses, the indicated vacuum was 2.1 ϫ 10 Ϫ6 Torr.

N-Deacetylation/N-Sulfation of K5 Polysaccharide
30 mg of K5 polysaccharide was dissolved in 7.5 ml of 2 N NaOH, incubated for 24 h at 60°C, and cooled to room temperature, and was adjusted to pH 7. The solution was warmed to 45-50°C. 100 mg of sodium carbonate, and 100 mg of trimethylamine-sulfur trioxide complex were added in a single step and incubated for 12 h. An equal portion of sodium carbonate and trimethylamine-sulfur trioxide was added after 12 h and the selective N-sulfation was continued for an additional 12 h at the same temperature. The solution was then brought to room temperature, dialyzed overnight against distilled wa- ter, using 1000 Da mwco cellulose membrane. The dialysate was lyophilized to obtain salt-free, N-sulfated K5 heparosan polysaccharide. The purified polysaccharide 2 was digested with heparitinases, and the resulting mixture was analyzed by LC/MS. The m/z of 416 was observed for the disaccharide corresponding to [M-H] Ϫ1 .

Purification of Polysaccharides
The reaction mixture, after termination of sulfation, was diluted to 1 ml with 0.25 M NaCl, 20 mM NaAc, 0.01% Triton X-100, pH 6.0, and then 1 mg of glycogen was added to minimize nonspecific interaction of polysaccharides with the column matrix. The diluted reaction mixture was loaded on 0.1 ml of DEAE-Sephacel column, pre-equilibrated with 2 ml of washing buffer containing 0.25 M NaCl, 20 mM NaAc, 0.01% Triton X-100, pH 6.0. The column was washed with 20 column volumes of washing buffer, and the polysaccharide was eluted from the column with 2 ml of 1 M NaCl in 20 mM NaAc, pH 6.0. 8 ml of absolute ethanol and 1 mg of glycogen were added to 2 ml of eluent in a 50-ml disposable polystyrene tube and incubated at 4°C overnight to facilitate the precipitation of polysaccharide. The precipitate was obtained by centrifuging in a RC3B centrifuge for 15 min at 3000 rpm. The obtained pellet was washed with 1 ml of 70% ethanol twice and finally dissolved in 200 l of double distilled water for subsequent characterization.

Cloning and Expression of Epimerase
cDNA Cloning of Human Glucuronyl C5 Epimerase-A cDNA clone coding for human C5 epimerase was isolated from a human fetal brain cDNA panel (Origene) by screening with PCR primers spanning nucleotides 7-157 of the coding region. A donor plasmid for the preparation of recombinant baculovirus expressing a soluble form of the epimerase was constructed in pFastBac HT plasmid modified by the insertion of honeybee melittin signal peptide ahead of the histidine tag. The construction employed a synthetic oligonucleotide adapter that also encoded amino acids 35-44 of the epimerase and two restriction fragments isolated from the cDNA clone (TaqI to EcoRI and EcoRI to SacI) that incorporate the rest of the epimerase coding region.
Baculovirus Expression and Purification of Glucuronyl C5 Epimerase-Human glucuronyl C5 epimerase recombinant baculovirus was prepared using the donor and the Bac-to-Bac baculovirus expression system (Invitrogen) according to the manufacturer's protocol, except that recombinant bacmid DNA was purified using an endotoxin-free plasmid purification kit (Qiagen, Inc.) and transfection of Sf9 cells was scaled up to employ 15 g of bacmid DNA and 2.5 ϫ 10 7 exponentially growing cells in four 100-mm dishes. Medium containing recombinant baculovirus was harvested at 3 days post-transfection and amplified twice for about 65 h each on Sf9 cells. The resulting high titer viral stock was stored in aliquots (0.75 ml) sufficient to infect 3.5 ϫ 10 8 cells, as determined by Western blotting of medium from infected cells using (His) 4 antibody (Qiagen). Infected cells were plated in ten 150-mm dishes and incubated at 26°C for 90 -96 h. The pooled medium was centrifuged at 400 ϫ g, adjusted to 10 mM in HEPES, titrated to pH 7.4, chilled on ice for 30 min, and centrifuged at 16,000 ϫ g. The clarified pool diluted in half with 10 mM HEPES, pH 7.4, made 1 mM in phenylmethylsulfonyl fluoride, and applied to an 8-ml column of ToyoPearl AF heparin 650 M. The column was washed with 40 ml of hCG 50 (10 mM HEPES, pH 7.4, 2% glycerol, 0.6% CHAPS, 50 mM NaCl) and eluted with an 80-ml linear gradient of 50 -600 mM NaCl in hCG. Aliquots of selected 1-ml fractions were analyzed by Western blotting for the presence of the histidine tag, and peak fractions were adjusted to 50% glycerol.
Glucuronyl C5 Epimerase Reaction-N-sulfated K5 polysaccharide 2 (1 g) was incubated with purified glucuronyl C5 epimerase at 37°C in a volume of 50 l containing 25 mM HEPES, 40 mM CaCl 2 , pH 6.5, or alternatively in 25 mM MES (pH 7.0) with or without polymer p40. After incubation for 12 h, the fresh epimerase was added, and incubation was extended for 24 h. The reaction mixture was diluted to 1 ml with DEAE wash buffer and purified on a 0.1 ml DEAE column. The epimerase treated N-sulfated K5 polysaccharide was treated with [ 35 S]PAPS in the presence of 2-OST sulfotransferase in total volume of 10 l for 30 min, and the reaction mixture was analyzed on 5% native polyacrylamide gel. The epimerase treated N-sulfated K5 polysaccharide was also treated with [ 35 S]PAPS in the presence of 6-OST-1 or 6-OST-2a. After DEAE purification, the polysaccharide was subjected to low pH nitrous acid and sodium borohydride treatment to obtain the disaccharide mixtures. The disaccharide profile was analyzed by HPLC on C18 column to estimate the percentage of glucuronic acid and iduronic acid. The labeling buffer contains 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. For a 25-l reaction, the following were assembled: 1 g of substrate, 12.5 l of 2ϫ labeling buffer, 70 ng of the expressed sulfotransferase, 2-10 l [ 35 S]PAPS (1 l ϳ1.0 ϫ 10 7 cpm) or [ 32 S]PAPS or [ 34 S]PAPS, and the appropriate amount of water. The reaction was incubated at 37°C for various periods of time ranging from 30 min to overnight, then diluted to 1 ml with DEAE wash buffer, and purified on a DEAE column. Alternatively, the reaction was stopped by heating at 70°C, and the reaction mixture was centrifuged at 10,000 ϫ g for 3 min, and the supernatant was used for GMSA or polyacrylamide gel analysis. Modified polysaccharide at each stage was digested with heparitinases and analyzed by IPRP-HPLC.

Gel Mobility Shift Assay
Heparin-ATIII binding buffer contained 12% glycerol, 20 mM Tris-HCl (pH 7.9), 100 mM KCl, 1 mM EDTA, and 1 mM dithiothreitol. For a typical 20-l binding reaction, radiolabeled polysaccharide (1ϳ50,000 cpm) was mixed with ATIII (1 g) in the binding buffer. The reaction mixture was incubated at room temperature (23°C) for 20 min and was then applied to a 4.5% native polyacrylamide gel (with 0.1% of bisacrylamide). The gel buffer was 10 mM Tris (pH 7.4) and 1 mM EDTA, and the electrophoresis buffer was 40 mM Tris (pH 8.0), 40 mM acetic acid, 1 mM EDTA. The gel was run at 6 volts/cm for 1-2 h with an S.E. 250 Mighty Small II gel apparatus (Hoefer Scientific Instruments, San Francisco). After electrophoresis, the gel was transferred to 3-mm paper and dried under vacuum. The dried gel was autoradiographed by a PhosphorImager 445SI (Molecular Dynamics, Sunnyvale, CA). The image was analyzed with NIH Image 1.60, and the band intensities were evaluated.

Factor Xa Inactivation Assay
Purified human factor Xa (Hematologic Technologies, Inc.) was diluted to a concentration of 2.85 mg/ml with 50% glycerol. Human antithrombin III (Cutter Laboratories) was dissolved in PBS at 5.0 mg/ml and frozen in aliquots. Chromogenic substrate S-2765 (Chromogenix) was dissolved in water at 25 mM. Porcine heparin (Sigma, 174 I.U./mg) was employed as a standard. Protein working stocks were prepared by diluting factor Xa 1:500 in PBS and ATIII 3:1000 in 1.6ϫ PBS, both containing bovine serum albumin at 5 mg/ml, and were held at 4°C. Substrate was diluted to 2.25 mM in PBS containing Polybrene at 1.0 mg/ml and held at 37°C. The assay was initiated by adding 20 l of ATIII (2.6 ϫ 10 Ϫ7 M) to 12 l of heparin standard or test polysaccharide (diluted in distilled water) with incubation at 37°C for 90 s. Factor Xa stock (20 l, 1.24 ϫ 10 Ϫ7 M) was added and mixed. After 240 s of incubation, 20 l of substrate was added and incubated for an additional 30 s. The reaction was stopped by the addition of 52 l of 60% acetic acid, and sample absorbance was measured at 405 nm.

RESULTS
A non-sulfated N-acetyl heparosan 1, a capsular polysaccharide of E. coli strain K5, resembles the unmodified nascent HS chain of proteoglycans (Scheme 1) (17). The production and isolation of bacterial K5 heparosan for use as a starting material in the chemoenzymatic synthesis of bioactive HS polysaccharides is an attractive alternative to the laborious chemical synthesis of the HS backbone. The partial digestion of the isolated polymer with heparitinase enzymes was carried out and analyzed by capillary reverse phase ion-pair high performance liquid chromatography coupled to micro-electrospray ionization time of flight (RP-IP-HPLC-ESI-TOF) mass spectrometry (24). The identity of HS precursor polysaccharide structure 1 was confirmed (Scheme 1).
The second step is synthesis of N-sulfated polysaccharide 2, as it is essential for subsequent enzymatic transformations. N-Deacetylase-N-sulfotransferase is a single protein that catalyzes the two initial modifications, N-deacetylation and Nsulfation, of the polysaccharide precursor in the biosynthesis of heparin and heparan sulfate (25,26). This enzyme exists as four isoforms in humans. The ratio of N-deacetylase to N-sulfotransferase activities differ dramatically among the four isoforms. For these reasons, we utilized the established simple chemical approaches to prepare the N-sulfated polysaccharide 2 (Scheme 1). N-deacetylation of polysaccharide 1 was accomplished by hydrazinolysis at 100°C or alkaline treatment with 2 M NaOH at 60 -65°C. The resulting free amino groups were selectively N-sulfated using trimethylamine sulfur trioxide, a chemoselective sulfonating agent, under controlled reaction conditions (27). Polysaccharide 2 was partially digested using heparitinases and analyzed by LC/MS to confirm its identity. Next, synthesis of polysaccharide 3 was undertaken, in which N-sulfated polysaccharide 2 was treated with heparan sulfate glucuronyl C-5 epimerase, resulting in inversion of stereochemical configuration at the C-5 carbon of many uronic acids along the chain. To date there are no efficient chemical strategies available to selectively epimerize glucuronic acid to iduronic acid. In fact the presence of both epimers along the polysaccharide adds further complexity and challenges in chemical synthesis and structure-function relationship studies (21,28).
Epimerization proceeds only at residues located at the reducing side of N-sulfated glucosamine residues, and only with uronic acids that are neither O-sulfated nor adjacent to Osulfated glucosamine residues (21,29). Therefore, epimerase treatment was carried out before any further modifications with sulfotransferases. The epimerization process has been monitored in the past by measuring the release of 3 H 2 O from 3 H-labeled (at C-5) polysaccharide 2 (21,28,30). We devised a new strategy in which we monitored the epimerization of polysaccharide 2 by determining selective incorporation of sulfate at the 2-O position of the newly generated iduronic acid using limiting amounts of [ 35 S]PAPS and HS 2-OST. The selective SCHEME 1. Enzymatic synthesis of antithrombin III-binding classical heparan sulfate polysaccharides.
2-O sulfation of iduronic acid on polysaccharide 3 was observed to be predominant whereas incorporation of sulfate on polysaccharide 2 was observed to be minor (Fig. 2). Synthesis of polysaccharide 3 containing iduronic acid was confirmed on 5% native polyacrylamide gel through selective radioactive 2-O sulfation of iduronic acid. This approach is better than traditional approaches, which involve measuring the release of tritium as 3 H 2 O to monitor the epimerization process. Our approach is superior because, in our opinion, the classical approach may be biased by isotope effects, and furthermore the assay is relatively slow and therefore could not be employed in real-time monitoring of our synthetic strategy. Next the extent of epimerization was determined by treating the polysaccharide 3 with [ 35 S]PAPS in the presence of 6-OSTs to prepare radioactive 6-O sulfated polysaccharide 7, which was then subjected to low pH nitrous acid and sodium borohydride treatment (31) to obtain two disaccharides, GluA-anMan R [6 35 S] and IdoA-anMan R [6 35 S]. These two disaccharides were then resolved on C18 reverse phase HPLC to determine the percentage of each epimer (Fig. 3). Normally one quantifies the percentage of iduronic acid and glucuronic acid by nitrous degradation coupled to [ 3 H]borohydride end labeling. However, the traditional approach can be utilized only in the analysis of polymers that are fully N-and O-sulfated. In the case of polysaccharide 3, the above approach generated two disaccharides, which do not contain any sulfate groups and hence bind very weakly to C18 columns. On the other hand, the nitrous acid degradation of polysaccharide 3 after 6-O sulfation generates two disaccharides, each containing one sulfate group, which allowed them to bind more tightly to the C18 column. Furthermore, we confirmed that the 6-O sulfation is complete using LC/MS analysis, which eliminated the potential errors of not taking into account the presence of non-6-O-sulfated disaccharides. We further validated the results by co-injection of standard disaccharides, as previously reported. The ratio of iduronic to glucuronic acid was found to be ϳ85: 15. In previous studies, epimerization of the polysaccharide 2 by epimerase resulted in generation of about 20% iduronic acid only (21). This is a striking result. We speculate that our baculovirus expression and purification system provides C5-epimerase with improved functional characteristics as compared with enzyme preparation utilized by other investigators.
It was suggested that 2-O sulfation of iduronic acid within the ATIII-binding pentasaccharide (Unit D, Fig. 1) limits 3-OST-1 mediated 3-O sulfation of glucosamine residues at the reducing side of IdoA(2S) residue, while it has no effect at its non-reducing end (32). The polysaccharide 4 was prepared from polysaccharide 3 in the presence of PAPS or radioactive [ 35 S]PAPS, catalyzed by 2-OST-1 (Scheme 1). The radioactive polysaccharide 4 was analyzed by 5% native polyacrylamide gel to confirm the action of 2-OST-1 (Fig. 2).
It was shown earlier that 3-OST-1 generally acts on glucosamine units located between GlcA (at the non-reducing side of glucosamine) and IdoA (at the reducing side of glucosamine) and generates ATIII binding structures. Polysaccharide 6 was successfully prepared from polysaccharide 5 using 3-OST-1 and radioactive [ 35 S]PAPS. The purified polysaccharide 6, containing radiolabeled sulfate at the 3-O position, was digested with heparitinases and analyzed by C18 HPLC (data not shown). This radiolabeled polysaccharide was used in gel mobility shift analysis to test its ability to bind to ATIII. Polysaccharide 5 was treated with [ 34 S]PAPS in the presence of enzyme 3-OST1 to prepare 3-O [ 34 S]sulfate-containing polysaccharide 6, which was then digested with heparitinases and analyzed by LC/MS system (Fig. 4) (24). The mass spectrometric analysis revealed the presence of stable isotope containing 3-O sulfated disaccharide and tetrasaccharide species, which conclusively demonstrated the generation of polysaccharide 6. The number of 3-O sulfates incorporated into polymer varied from 3-10 per 100 disaccharide units, depending upon the extent of epimerization. We are certain that 3-O sulfate groups are located on the glucosamine residues as indicated in Scheme 1. This conclusion is based upon the specificity of cloned and expressed recombinant 3-OST-1 enzyme, which has previously been shown to place the sulfate group exclusively at the 3-O position of glucosamine units as indicated by studies which included the use of many different model disaccharides with 3-O sulfates at different position.
The notable point here is that the 2-O sulfate of IdoA(2S) (unit D, Fig. 1) is a minor contributing factor for ATIII-binding and hence its anticoagulant action (see below). However in this manuscript, we show for the first time that the omission of 2-OST1 enzyme allows us to engineer polysaccharide whose anticoagulant properties are preserved but in which improved therapeutic characteristics should be achieved. In addition, it was shown earlier that 3-OST-1 can also act in the absence of 2-O sulfation and still generate the ATIII-binding motif. One of the functions of 2-OST-1 may be to restrict the action of 3-OST-1 at certain glucosamine units located on the reducing side of IdoA(2S) along the chain (32). Given that 2-O sulfate is not essential for anticoagulant action, we devised a parallel synthetic strategy to prepare anticoagulant HS polysaccharides devoid of IdoA(2S) residues (polysaccharide 8, Scheme 2) and to characterize anticoagulant activity in terms of their ability to inhibit factor Xa. Platelet factor 4 (PF4) is present in the ␣ granules of platelets. Activated platelets release PF4 that binds to heparan sulfate and form a complex on the surface of endothelial cells. The complex generates antibodies, which recognize the PF4-heparin complex and activate the complement system, which disrupts the control mechanism that suppresses coagulation. Therefore, this interaction results in local thrombosis and decreased levels of platelets. This rare phenomenon can lead to the death of a patient. It occurs particularly at high frequency when patients are placed on cardiac pulmonary bypass (38). The key residue on the heparin molecule, responsible for binding of platelet factor 4, appears to be IdoA(2S) (39). Therefore polysaccharide 8 that is devoid of IdoA(2S) should exhibit two beneficial therapeutic attributes. On the one hand, it acts as an anticoagulant while reducing the risk of fatal HIT.
On the other hand, this new non-classical anticoagulant should be more effective in areas with platelet enriched thrombi (arterial side) which release PF4 and generate a protected sanctuary for blood clotting. In addition, it should be resistant to heparanase cleavage and hence effective at lower dosages (40). This hypothesis can only be substantiated by experimental studies using thrombotic models in different vascular beds of primates. Polysaccharide 8 was prepared by 3-O sulfation of polysaccharide 7, which in turn was prepared from 6-O sulfation of polysaccharide 3 (Scheme 2). Polysaccharide 7 was also treated with [ 34 S]PAPS in the presence of enzyme 3-OST-1 to prepare 3-O [ 34 S]sulfate containing polysaccharide 8, which was then digested with heparitinases and analyzed by LC/MS system (Fig. 5) (24). The disaccharide analysis of the products was carried out by total enzymatic degradation and subsequently identified by LC/MS (Table I). The data shows the presence of expected disaccharides. Gel mobility shift assay (GMSA) was carried out to determine the ability of radiolabeled polysaccharide 6 and radiolabeled polysaccharide 8 to bind to ATIII. Radiolabeled polysaccharide 5, lacking critical 3-O sulfate residue essential for binding to ATIII and creating anticoagulant activity, failed to bind to ATIII and hence was not shifted (Fig. 6a) whereas polysaccharide 6 and polysaccharide 8 both specifically bound to ATIII (Fig. 6, b and c). The factor Xa assay confirmed the ability of polysaccharide 6 and polysaccharide 8 to bind to and accelerate the action of ATIII.
We have shown for the first time that enzymatic synthesis of bioactive HS polysaccharide structures with anticoagulant properties can be accomplished in a relatively simple and expeditious manner as compared with classical chemical synthesis. This synthetic strategy, in conjunction with exoglycosidases, which remove terminal sugar residues at the nonreducing end, can be applied to synthesize bioactive heparan sulfate of any size or structure and should also permit us to identify proteins that recognize the various HS structures. These structures can also be used as a diagnostic probe to detect functional abnormalities in a specific protein implicated in human disease, such as detection of antibodies that cause heparin induced thrombosis, detection of tumor associated heparanases or tumor suppressor gene products that are implicated in tumor metastasis, bone growth plate abnormalities, and detection of pathological alterations in growth factorgrowth factor receptor interactions that are implicated in various developmental abnormalities. DISCUSSION In this study, we described a novel method for synthesizing bioactive HS structures using a panel of recombinant HS biosynthetic enzymes, which act upon bacterial product consisting of -GlcNAc-GlcA repeating units. This method mimics, under in vitro conditions, the synthesis of HS polymers within the Golgi. We note that our other companion investigations dealt with various strategies to optimize this novel synthetic approach (15,16). The bacterial starting material represents the unmodified HS precursor structure, whose generation constitutes the first step in the biosynthesis of HS catalyzed by EXT polymerase. The major problem that remains unsolved in HS biology is how structures of HS proteoglycans regulate interactions with a specific biological target. It has been tacitly assumed that the sequence within heparan sulfate/heparin is unique and provided HS chains with the ability to interact with a specific protein. This idea is analogous to a similar paradigm widely held for DNA and proteins. Unfortunately there is little data to support this concept. Bernfield et al. (1) pointed out that heparan sulfate exhibiting differences in sequences may have similar biological functions. One can infer this concept by examining the extensive data obtained with the ATIII-binding pentasaccharides, generated over 3 decades, in which it has been shown that only 3-O and 6-O groups and their spacing are critical for the activation of ATIII. Changes outside these critical groups have little effect on the biological activity of ATIIIbinding pentasaccharides.
A series of reports have been published by both our and other laboratories that attempt to solve this problem of defining structural parameters on HS chains, which are responsible for interactions with biological targets other than ATIII. This problem can now be addressed by using our novel enzymatic synthetic methodology. Underlying these differences in approach between various laboratories is a major difference in the philosophic approach of how structural differences in HS chains allow them to interact with many biological targets. We have emphasized that differences in sequences of the HS chains, which possess the same biological function, is due to the presence of common critical groups with identical spacing and that this parameter permits specific interactions with biological targets. It is perhaps worth emphasizing the difference between critical groups and critical sequences. In the case of sequences, biological activity requires alignment of multiple groups whereas in the case of critical groups, biological activity requires less stringent condition in which a fewer groups must be aligned. The difference in these two concepts has important implications for both elucidations of structure-function relationship and the manner by which respective biosynthetic pathways functions.
In addition, there are important differences between our SCHEME 2. Enzymatic synthesis of antithrombin III-binding non-classical heparan sulfate polysaccharide variant structures lacking IdoA(2S) groups. current experimental approach and those of others. Reports by other laboratories have used heparin oligosaccharides and a combination of chemical de-sulfation and enzymatic re-sulfation to generate heterogeneous oligosaccharide libraries, which were then employed to screen for interacting with growth factors (41). On the other hand, our laboratory utilized heparin oligosaccharides in conjunction with enzymatic sulfation and gel mobility shift analysis to identify critical groups on the oligosaccharides which permitted interactions with specific biological targets (11). Thus, our study provided a novel approach to obtain critical functional groups on HS oligosaccharides. A subsequent report from our laboratory analyzed in a greater detail those critical groups on HS oligosaccharide and heparan sulfate, which are involved in growth factor/growth factor receptor interactions. This information was correlated with somatic cell mutants and mitogenic activity in BAF cell systems to demonstrate those critical groups involved in binary complex (HS/growth factor or HS/growth factor receptor) and ternary complex formation (HS/growth factor/growth factor receptor) and evaluation of the biological significance of binary and ternary complex formation with regard to mitogenic activity (42). Comparison of our approach to define critical groups on heparan sulfate involved in growth factor-growth factor receptor interactions should be compared with that of Jemth et al., (41) who only considered binary interactions and failed to define experimentally those critical groups required for mitogenic activity.
We utilized both philosophical insights as outlined above and an extensive panel of isoform specific recombinant heparan sulfate biosynthetic enzymes to generate HS-like polysaccharides that bind specifically to ATIII. The enzymes employed included not only sulfotransferases and their various isoforms, but also C5 epimerase. Various steps in our synthetic process were monitored by a powerful new LC/MS technology developed by us. Our approach permitted us to generate ATIII binding polysaccharide without 2-O sulfate groups, which change the sequence of polysaccharide but which lie outside those groups required for biological activity. Our demonstration that polysaccharides that lack 2-O sulfate groups, but which exhibits biological activity, emphasizes the importance of critical groups, but not sequence, to biological activity. Finally the absence of 2-O sulfate groups may impart valuable therapeutic characteristics to the final product. These benefits include enhanced activity on the arterial side as well as prolonged in vivo half-life and decreased ability to bind PF4 with consequent absence or reduction in HIT. Our enzymatic synthetic approach can be extended to generate libraries of homogeneous oligosaccharides, which will allow the definition of critical functional groups and ultimately the design of heparin/HS based drugs for different pathological states.
One can envision that our approach can be adopted to synthesize different HS structures by adding different isoforms, or by changing the order of the addition of enzymes or omitting enzymes and by altering the structures at the non-reducing end with catabolic enzymes in conjunction with EXT polymerase. Finally, we would like to point out that this strategy can be employed in a stepwise manner with complete control over the production of homogeneous products (16). This stepwise enzymatic approach could be tailored such that one can synthesize structures with a specific N-sulfation pattern and restricted epimerization (16). Alternatively, enzymes can be added together which do or do not affect each other and thereby rapidly produce heterogeneous products in a combinatorial fashion (15). As noted above, simple stepwise or contemporaneous modifications can produce homogeneous or heterogeneous products, respectively, and hence our strategy can be expanded to accomplish the production of oligosaccharides or polysaccharides as desired.