Uncovering Biphasic Catalytic Mode of C5-epimerase in Heparan Sulfate Biosynthesis*

Background: C5-epimerase converts a glucuronic acid to an iduronic acid residue in the heparan sulfate biosynthetic pathway. Results: C5-epimerase displays both “reversible” and “irreversible” catalytic modes. Conclusion: C5-epimerase recognizes the saccharide sequence of the substrate to position the iduronic acid. Significance: The biphasic catalytic mode of C5-epimerase reveals a unique control mechanism in the biosynthesis of heparan sulfate. Heparan sulfate (HS), a highly sulfated polysaccharide, is biosynthesized through a pathway involving several enzymes. C5-epimerase (C5-epi) is a key enzyme in this pathway. C5-epi is known for being a two-way catalytic enzyme, displaying a “reversible” catalytic mode by converting a glucuronic acid to an iduronic acid residue, and vice versa. Here, we discovered that C5-epi can also serve as a one-way catalyst to convert a glucuronic acid to an iduronic acid residue, displaying an “irreversible” catalytic mode. Our data indicated that the reversible or irreversible catalytic mode strictly depends on the saccharide substrate structures. The biphasic mode of C5-epi offers a novel mechanism to regulate the biosynthesis of HS with the desired biological functions.

Heparan sulfate (HS), a highly sulfated polysaccharide, is biosynthesized through a pathway involving several enzymes. C 5 -epimerase (C 5 -epi) is a key enzyme in this pathway. C 5 -epi is known for being a two-way catalytic enzyme, displaying a "reversible" catalytic mode by converting a glucuronic acid to an iduronic acid residue, and vice versa. Here, we discovered that C 5 -epi can also serve as a one-way catalyst to convert a glucuronic acid to an iduronic acid residue, displaying an "irreversible" catalytic mode. Our data indicated that the reversible or irreversible catalytic mode strictly depends on the saccharide substrate structures. The biphasic mode of C 5 -epi offers a novel mechanism to regulate the biosynthesis of HS with the desired biological functions.
Heparan sulfate (HS) 4 is an essential glycan that is found on the mammalian cell surface and in the extracellular matrix. HS participates in a wide range of physiological and pathophysiological functions, including embryonic development, inflammatory responses, blood coagulation, and viral/bacterial infections (1). HS consists of a disaccharide repeating unit of glucuronic acid (GlcA) or iduronic acid (IdoA) and glucosamine, each of which is capable of carrying sulfo groups. The biosynthetic pathway of HS includes HS polymerases, sulfotransferases, and C 5 -epi. Collective actions of these enzymes result in matured HS with biological functions (supplemental Fig. S1). Uniquely distributed sulfo groups and IdoA residues play critical roles in determining the functions of HS (2); however, it remains unclear how to regulate the biosynthesis of these specific sulfated saccharide sequences in vivo.
Understanding the biosynthetic mechanism also aids in developing a chemoenzymatic method to synthesize HS-based drugs. Heparin, a commonly used anticoagulant drug, is a special form of HS with higher levels of sulfation and IdoA. Heparin is currently isolated from animal tissues through a long and poorly regulated supply chain. Worldwide distribution of contaminated heparin in 2007 has raised the concerns over the safety of animal-sourced heparins (3). A cost-effective method to prepare synthetic heparin is desirable. Using HS biosynthetic enzymes, we developed a chemoenzymatic approach to synthesize structurally defined HS and heparin oligosaccharides in high efficiency, which significantly expands the synthetic capability by a purely chemical approach (4,5). C 5 -epi converts a GlcA unit to an IdoA by forming a putative carbanion intermediate at the C 5 -position of a GlcA unit (see Fig. 1A) (6). Only a single isoform of C 5 -epi is in human genome, and C 5 -epi knock-out mice are neonatal lethal, suggesting its essential physiological roles in vivo (7). C 5 -epi reportedly is a two-way catalytic enzyme, performing both the forward and reverse epimerization. Namely, the IdoA unit reverts back to a GlcA unit in the presence of C 5 -epi (8). The nature of the reaction renders unique challenges to study the mode of action of C 5 -epi. To date, three approaches have been reported to measure the activity of C 5 -epi (9 -11). Because all these methods utilized structurally heterogeneous polysaccharide substrates and the products were identified by a disaccharide analysis, none of these approaches is capable of locating the GlcA residues participated in the epimerization beyond a disaccharide domain. Therefore, the understanding for the mechanism of action of C 5 -epi was limited, especially the effects of neighboring saccharide structures on the action of C 5 -epi.
In this article, we report an advanced method to characterize C 5 -epi using a series of structurally defined oligosaccharide substrates coupled with tandem mass spectrometry technique. This method permits us to locate the number and position of epimerization at the oligosaccharide levels with molecular pre-cision. On contrary to the convention view, we uncovered that C 5 -epi also serves as a one-way catalyst to convert a GlcA to an IdoA residue irreversibly, displaying an irreversible catalytic mode. C 5 -epi recognizes the N-sulfation pattern of the substrates to display reversible or irreversible catalytic mode of C 5 -epi. The biphasic mode of C 5 -epi offers a novel mechanism to regulate the biosynthesis of HS with the desired biological functions.

EXPERIMENTAL PROCEDURES
Expression and Purification of C 5 -epi-Recombinant C 5 -epi was expressed and purified as described previously (10). Briefly, a fusion protein of maltose-binding protein and the human C 5 -epi catalytic domain (Glu 53 -Asn 617 ) was constructed in pMal c2x vector (New England BioLabs). The expression was carried out in Origami-B DE3 cells (Novagen), which contained pGro7 (Takara, Japan) plasmid expressing chaperonin proteins GroEL and GroES of Escherichia coli. The purification of C 5 -epi was completed using an amylose-agarose (New England Biolabs) column following the protocol provided by the manufacturer. To carry out the epimerization reaction in D 2 O buffer, D 2 O-exchanged C 5 -epi was prepared. To this end, the har-vested bacteria pellet were suspended in the buffer that was prepared in D 2 O (Ͼ99% D, Spectra Stable Isotopes). In addition, the elution buffers for amylose-agarose column purification were also prepared in D 2 O.
The activity of purified C 5 -epi (including D 2 O-exchanged C 5 -epi) was measured by coupling the reaction of C 5 -epi and 2-O-sulfotransferase (2-OST) followed by a disaccharide analysis (12). In this assay, C 5 -epi (1.5 g) was incubated with 2 g of N-sulfo heparosan in the buffer containing 50 mM MES (pH 7.0), Triton X-100, and 1 mM CaCl 2 at 37°C for 30 min. To the reaction mixture, 0.6 g of 2-OST and 3Ј-phosphoadenosine 5Ј-phospho[ 35 S]sulfate ([ 35 S]PAPS) (5-7 ϫ 10 5 cpm) were added. The reaction was incubated at 37°C for two additional hours. The 35 S-labeled polysaccharide was purified by a DEAE column and was then subjected to nitrous acid degradation (at pH 1.5) to yield the disaccharides. The identities of the resultant disaccharides were determined by reverse-phase ion pairing HPLC using a C 18 -column as described previously (13). As a fully active C 5 -epi, Ͼ90% of resultant disaccharides from the degraded polysaccharide have a structure of IdoA2S-AnMan, where AnMan represents 2,5-anhydromannitol. FIGURE 1. The reaction catalyzed by C 5 -epi and the schematic presentation for determining the activity of C 5 -epi. A shows the reaction catalyzed by C 5 -epi. A trisaccharide segment of polysaccharide is shown. C 5 -epi removes the proton from C 5 of the GlcA residue to form a putative carbanion intermediate. An H 2 O molecule then reacts with the carbanion to form an IdoA residue. Conversely, C 5 -epi can catalyze the reverse reaction, namely to convert an IdoA residue to a GlcA residue. B shows the schematic presentation for determining the activity and reversibility of C 5 -epi. The assay was developed based on the reaction mechanism of C 5 -epi. Here, an oligosaccharide was used as a substrate. After incubation with C 5 -epi in D 2 O, the obtained epi-oligosaccharide had one unit increase in M r (MW), indicating that a single GlcA residue was converted to an IdoA residue. Tandem MS was used to locate the site of epimerization because the epimerized residue should carry a deuterium. The reversibility of C 5 -epi was determined by incubating the epi-oligosaccharide with H 2 O. If the M r of the product reduces by one unit, the reaction was proved to be "reversible." The M r values of the oligosaccharides were determined by ESI-MS. The keys for shorthand representations of carbohydrate residues are shown in the box.  Preparation of Oligosaccharide Substrates-A total of eight oligosaccharides, differing in the size of the oligosaccharides and the distribution of N-sulfo groups, were prepared in this study. The preparation of the substrates followed essentially the same procedures described in our previous work (14 -16). The synthesis initiated from a disaccharide of GlcA-AnMan, which was prepared from nitrous acid, degraded heparosan. The elongation from disaccharide to the desired size of oligosaccharide substrates was completed by KfiA (N-acetyl glucosaminyl transferase of E. coli K5 strain) and PmHS2 (heparosan synthase 2 from Pasteurella multocida). In a typical elongation reaction from the disaccharide to tetrasaccharide, 6 mg of GlcA-AnMan was incubated with 18 mol of UDP-GlcNTFA (UDP-N-trifluoroacetyl glucosamine) and 2 mg of KfiA in 40 ml buffer containing 25 mM Tris-HCl, pH 7.2, 10 mM MgCl 2 at room temperature overnight. Upon the complete consumption of UDP-GlcNTFA, 2 mg of PmHS2, and 27 mol of UDP-GlcUA were added into the reaction mixture and allowed to incubate overnight at room temperature. The product was purified by a BioGel P-2 column (0.75 ϫ 200 cm), which was equilibrated with 0.1 M ammonium bicarbonate at a flow rate of 4 ml/h. The product fraction was located by electrospray ionization mass spectrometry (ESI-MS) analysis. For the additional elongation reaction to the hexa-, octa-, and decasaccharides, the conditions were identical essentially to the above, whereas the reaction volumes were adjusted to the smaller sizes depending on the amount of substrates. The conversion of GlcNTFA to GlcNS was performed under an alkaline condition to remove the trifluoroacetyl group followed by N-sulfation using N-sulfotransferase and PAPS. We completed the synthesis of milligrams of each oligosaccharide substrates. The structures of the oligosaccharides were confirmed by ESI-MS analysis.

M r after C 5 -epi treatment in D 2 O Reaction mode
The preparation of UDP-GlcNTFA was started from glucosamine (Sigma), which was first converted to GlcNTFA by reacting with S-ethyl trifluorothioacetate (Sigma-Aldrich) followed the protocol described previously (16). The resultant GlcNTFA was converted to GlcNTFA-1-phosphate using N-acetylhexosamine 1-kinase (17). The plasmid expressing N-acetylhexosamine 1-kinase was a generous gift from Professor Peng George Wang (Georgia State University, Atlanta, GA), and the expression of the enzyme was carried out in E. coli as reported (17). The UDP-GlcNTFA synthesis was completed by transforming GlcNTFA-1-phosphate using glucosamine-1phosphate acetyltransferase/N-acetylglucosamine-1-phosphate uridyltransferase as described (16). The protocols for the expression of other HS biosynthetic enzymes and the preparation of PAPS were described elsewhere (16,18). Additional methods are presented under supplemental "Methods."

RESULTS AND DISCUSSION
For the current work, eight structurally defined substrates were synthesized using a chemoenzymatic approach (Table 1) (4). An MS-based method to monitor the reaction was devised, which allowed us to pinpoint the epimerization site (EPS) and to examine the reversibility of C 5 -epi. A substrate is incubated with C 5 -epi and D 2 O, leading to an epimerized product with deuterium incorporated. One unit of increase in molecular weight (M r ) for epi-oligosaccharide suggests that one GlcA residue is converted. Using a tandem MS technique, we located the residue that carries the deuterium in the epi-oligosaccharide, thus identifying the EPS site. The reversibility of the epimerization is studied by incubating the deuterated epi-oligosaccharide with C 5 -epi in H 2 O followed by MS analysis (Fig. 1B).
The reversibility of C 5 -epi was confirmed using Octa-1 as a substrate, which has a structure of GlcA-GlcNS-GlcA-GlcNS-GlcA-GlcNAc-GlcA-AnMan (Fig. 2). The M r of epi-Octa-1 was increased to 1556.4 Ϯ 0.4 from 1554.5 Ϯ 0.3 (Table 1 and Fig. 2) when it was treated with C 5 -epi in the presence of D 2 O. Furthermore, the incubation of epi-Octa-1 with C 5 -epi in H 2 O resulted in the decrease of its M r by 1.9 units (from 1556.4 Ϯ 0.4 to 1554.5 Ϯ 0.5). Our data suggest that two GlcA units in Octa-1 were susceptible to C 5 -epi modification, and both EPS sites were reversible. These results were consistent with the previous report by Lindahl and co-workers (8).
Interestingly, C 5 -epi displayed an irreversible reaction mode when a different substrate (Octa-3) was used, which has the structure of GlcA-GlcNAc-GlcA-GlcNS-GlcA-GlcNS-GlcA-AnMan. Incubation of Octa-3 with C 5 -epi in D 2 O resulted in D-labeled epi-Octa-3. D-labeled epi-Octa-3 exhibited the M r of 1555.4 Ϯ 0.4 (Fig. 3A) with an increase of 0.9 units (Fig. 3B), suggesting that a single GlcA residue was converted to IdoA. The M r of epi-Octa-3 remained unchanged after incubation with C 5 -epi, suggesting that C 5 -epi was unable to reverse the reaction. To further confirm the structure of D-labeled epi-Octa-3, a series of experiments were conducted. Tandem MS analysis was employed to prove that the D-labeled residue is located at residue E. For example, the fragment Y 5 of epi-Octa-3 (m/z value of 499.23 for the doubly charged ion) is ϳ1.3 units higher than that of Octa-3 (m/z value of 498.57) (Fig. 3, C and D). NMR analyses of epi-Octa-3 confirmed the presence of IdoA at residue E (supplemental Table S1). The presence of an IdoA in epi-Octa-3 was also demonstrated after modifying epi-Octa-3 with 2-O-sulfotransferase followed by a disaccharide analysis (supplemental data). High resolution ESI-MS analysis of D-la-beled epi-Octa-3 showed a signal at a mass/charge ratio of 517.1121, consistent with [M Ϫ 3H] 3Ϫ of the anticipated structure (calculated mass/charge ratio, 517.1040).
Subjecting additional oligosaccharides to the analysis revealed a relationship between the reaction mode of C 5 -epi and the structures of substrates (Table 1). C 5 -epi appears to recognize the flanking saccharide sequence at the nonreducing end of the EPS residue. Oligosaccharides containing a nonreducing end GlcNS residue immediately adjacent to the EPS residue are reactive to C 5 -epi. In contrast, when the GlcNS is replaced with GlcNAc (e.g. Octa-5), the oligosaccharide is no longer a substrate of C 5 -epi, proving the critical role of GlcNS at the nonreducing end of the EPS (8). Whether C 5 -epi exhibits an irreversible or a reversible mode depends on the residue at the mode of reaction recognition site (MRRS) that is three residues away from the EPS site toward the nonreducing end (Fig. 4A). If a GlcNAc is present at the MRRS site, C 5 -epi displays an irreversible reaction mode (as for Octa-3 and Octa-4). If a GlcNS or a GlcNH 2 residue is present at the MRRS site (e.g. Ocat-1, Octa-2, and Octa-6) or the MRRS site is unoccupied (i.e. Hexa-7 or the second EPS site (EPS2) for Octa-1 and -2), C 5 -epi displays a reversible reaction mode (Fig. 4A).
To further strengthen our conclusion, we synthesized a decasaccharide substrate (Deca-8), permitting C 5 -epi to display The designated EPS is at the residue 0. The residue Ϫ1 must be a GlcNS residue to serve as an EPS site. The structure of the saccharide residue at the MRRS determines the mode of reaction of C 5 -epi. If the MRRS site (residue Ϫ3) is a GlcNS or GlcNH 2 residue or unoccupied, C 5 -epi displays a reversible reaction mode (like reaction a, c, and b). If a GlcNAc residue is at the MRRS site, C 5 -epi displays an irreversible reaction mod (like reaction d). If residue Ϫ1 is a GlcNAc residue, residue 0 is not an EPS site. B shows mixed reaction modes displayed by C 5 -epi when Deca-8 was used as a substrate. Two EPS sites were present in Deca-8. EPS1 is a reversible site, whereas EPS2 is an irreversible site. The gray shaded box indicates the saccharide residues recognized by C 5 -epi. The yellow shaded box indicates the saccharide residues recognized by C 5 -epi during the mixed reaction modes. The keys for shorthand representations of carbohydrate residues are shown in the box. mixed reaction modes (Fig. 4B). Here, two EPS sites were constructed at the residue 0 (EPS1) and residue Ϫ2 (EPS2), respectively. Two MRRS sites were also introduced: a GlcNS and a GlcNAc residue were placed at residue Ϫ3 and residue Ϫ5, which should make a reversible EPS1 site and an irreversible EPS2 site, respectively. Indeed, our data confirmed that EPS1 is a reversible site, whereas EPS2 is irreversible (supplemental Fig.  S5).
Lastly, we tested the role of irreversible reaction mode of C 5 -epi in contributing to the biosynthesis of HS that binds to antithrombin (AT). HS or heparin forms a 1:1 complex with AT, which deactivates the activities of factor Xa and IIa in the blood coagulation cascade, to exhibit the anticoagulant activity. Anticoagulant HS and heparin isolated from natural sources have an AT-binding pentasaccharide with a structure of -GlcNAc6S-GlcA-GlcNS3SϮ6S-IdoA2S-GlcNS6S-(where GlcNS3Sϩ6S represents N-sulfo glucosamine 3-O-sulfate with or without 6-O-sulfate) (19,20). The IdoA2S residue in the pentasaccharide domain is known to be essential for high ATbinding affinity (21).
The uncovered irreversible reaction mode of C 5 -epi provides insight for the natural selection for a GlcNAc6S (not a GlcNS6S) residue as part of the AT-binding site for its role in positioning the IdoA residue. The GlcNAc residue is expected to serve as a MRRS site to direct C 5 -epi to synthesize an irreversible IdoA residue in the AT-binding pentasaccharide site, thus increasing the efficiency for the biosynthesis of AT-binding HS. To prove this assertion, the synthesis of AT-binding decasaccharides was completed using three decasaccharide substrates, including Deca-9, Deca-10, and Deca-11(supplemental Fig. S6). Deca-9 had no IdoA residue; Deca-10 contained two reversible IdoA residues; and Deca-11 had one irreversible IdoA residue at EPS2 site and one reversible IdoA residue at EPS1 site. The irreversible IdoA residue at EPS2 site in Deca-11 was achieved by exposing Deca-10 to C 5 -epi modification. These decasaccharides were modified by O-sulfotransferases to produce O-sulfated decasaccharides and were then fractionated by AT affinity columns to determine the amount of AT-binding site in the products. The results revealed that 40% O-sulfated Deca-11 bound to the AT affinity column, whereas the amount of the AT-binding portion for Deca-10 was determined to be 16%, a significant decrease in comparing to that of Deca-11 (Fig. 5). As expected, only 4% of O-sulfated Deca-8 bound to AT affinity column because it lacked an IdoA residue. Taken together, our data suggest that an irreversible IdoA residue at EPS2 site enhances the biosynthesis of ATbinding site.
Conclusions-Although the biosynthesis of HS is not a template-driven process, our results support the notion that HS biosynthetic pathway adopts an exquisite way to fine tune the extents of modifications through substrate control (22). Those enzymes involved in modifying the highly sulfated substrates (supplemental Fig. S1), i.e. 3-O-sulfotransferase, are able to distinguish the saccharide sequences with complicated sulfation patterns (23). In contrast, the mechanism used by those enzymes that modify the nonsulfated or low sulfated substrates, including N-deacetylase/N-sulfotransferase and C 5 -epi, is perceived to be subtle because the polysaccharide substrates have relatively simple repetitive structures. In the initial N-sulfation step, N-deacetylase/N-sulfotransferase introduces the GlcNS residues consecutively, rather than randomly, along the polysaccharide substrate (24). Esko and Selleck (20) hypothesized that the N-sulfation step introduces the codes to direct the extent in the subsequent modification steps. Indeed, our findings demonstrate that C 5 -epi recognizes the distribution of GlcNS residues in the substrate and displays distinctive catalytic modes, namely translating the N-sulfation code into the positions of GlcA/IdoA residues. Our data clearly show the impact of the irreversible mode of C 5 -epi on the efficiency for the biosynthesis of anticoagulant HS. Further investigation will unfold the full implication of the biphasic mode possessed by C 5 -epi in controlling the biosynthesis of HS.