GlcNAc 2-epimerase can serve a catabolic role in sialic acid metabolism.

Sialic acid is a major determinant of carbohydrate-receptor interactions in many systems pertinent to human health and disease. N-Acetylmannosamine (ManNAc) is the first committed intermediate in the sialic acid biosynthetic pathway; thus, the mechanisms that control intracellular ManNAc levels are important regulators of sialic acid production. UDP-GlcNAc 2-epimerase and GlcNAc 2-epimerase are two enzymes capable of generating ManNAc from UDP-GlcNAc and GlcNAc, respectively. Whereas the former enzyme has been shown to direct metabolic flux toward sialic acid in vivo, the function of the latter enzyme is unclear. Here we study the effects of GlcNAc 2-epimerase expression on sialic acid production in cells. A key tool we developed for this study is a cell-permeable, small molecule inhibitor of GlcNAc 2-epimerase designed based on mechanistic principles. Our results indicate that, unlike UDP-GlcNAc 2-epimerase, which promotes biosynthesis of sialic acid, GlcNAc 2-epimerase can serve a catabolic role, diverting metabolic flux away from the sialic acid pathway.


INTRODUCTION
Carbohydrate-receptor interactions participate in numerous cell-cell recognition events in eukaryotes (1)(2)(3)(4). Of the nine monosaccharides that constitute mammalian polysaccharides, the common terminal residue sialic acid stands out as a major determinant of cell-cell interactions.
Sialic acid is a component of sialyl Lewis x (sLe x ), a tetrasaccharide that binds the selectin family of adhesion molecules and initiates inflammatory leukocyte adhesion (5). Sialic acid is also the major binding determinant of Siglecs, a family of sialic acid-binding lectins that are involved in many processes including B-cell signaling and activation (6).
Because of the prominent role of sialic acid in cell surface recognition, there is considerable interest in the regulatory mechanisms that control its biosynthesis and presentation on the cell surface ( Figure 1) (7). The first committed intermediate in the biosynthesis of sialic acid is N-acetylmannosamine (ManNAc), which is phosphorylated by ManNAc 6-kinase (8,9) to initiate sialic acid biosynthesis in the cytosol. The subsequent action of sialic acid synthase (10), followed by an unknown phosphatase, yields sialic acid, which is then transformed into CMPsialic acid by CMP-sialic acid synthetase in the nucleus (11,12). After transport into the Golgi compartment, the sialyltransferases utilize this substrate to sialylate the terminal position of oligosaccharide chains. Transcriptional regulation of sialyltransferases is one mechanism for controlling the production of certain sialylated epitopes such as polysialic acid (13), sLe x (5) or

Expression of GlcNAc 2-epimerase in Jurkat cells
For mammalian expression, the GlcNAc 2-epimerase gene was cloned into the pcDNA5/FRT vector, a component of the FlpIn system, by digestion of pUKHRB6 (20) and pcDNA5/FRT with EcoR I and subsequent ligation with T4 DNA ligase. The correct product was confirmed by DNA sequencing and the vector was denoted pSJLG2E/FRT. The gene encoding the red fluorescent protein (dsRed) was also cloned into pcDNA5/FRT. The vectors pcDNA5/FRT and pDsRed2-N1 were digested with Hind III and Not I and the gene encoding dsRed was ligated into pcDNA5/FRT with T4 DNA ligase. The correct product was confirmed by DNA sequencing and the vector was denoted pSJLRFP/FRT. Jurkat cell lines stably expressing GlcNAc 2-epimerase were generated with the FlpIn system according to the manufacturer's instructions. Briefly, Jurkat cells were transfected with pFRT/lacZeo using Lipofectamine PLUS. After selection using Zeocin, stable transfectants were isolated that contain Flp recombinase sites flanking the Zeocin resistance gene. Clonal populations were isolated using "feeder" Jurkat cells as previously described (28) and the gene encoding the red fluorescent protein (dsRed) was introduced into selected clones to determine whether the site of incorporation yielded high, homogeneous gene expression. The vectors pSJLRFP/FRT and pOG44 (encoding the Flp recombinase) were introduced into various Zeocinresistant clones and cells were selected for Hygromycin resistance. The cells were analyzed by flow cytometry and one Zeocin-resistant cell line was selected as the host for GlcNAc 2epimerase (denoted wt*). The vectors pSJLG2E/FRT and pOG44 were introduced into wt* cells using Lipofectamine PLUS. After selection in the presence of Hygromycin, the resulting population was denoted G2E*.
Jurkat and Jurkat-derived cell lines were maintained in a 5.0% CO 2 , water-saturated atmosphere at 37 ºC and grown in RPMI-1640 media supplemented with penicillin (100 units/ml) and streptomycin (0.1 mg/ml). Wt* cells were grown in the presence of Zeocin (100 µg/ml) and G2E* cells were grown in the presence of Hygromycin (300 µg/ml). Typically, both cell lines were grown in T-25 flasks and cell densities were maintained between 2.0 x 10 5 and 1.6 x 10 6 cells/ml.

RT-PCR on wt* and G2E* cells
Messenger RNA from 5.0 x 10 6 of Jurkat, wt* and G2E* cells was isolated using the Micro-FastTrack 2.0 mRNA Isolation kit according to the manufacturer's instructions. cDNA was synthesized using the SuperScript First-Strand Synthesis kit and oligo dT primers. The

Flow cytometry
Cells were seeded at a density of 2.0 x 10 5 cells/ml and incubated for three days with the compounds indicated. Cells were then washed and stained with either biotin hydrazide and FITC-avidin to detect ketones (29), or phosphine-FLAG and anti-FLAG-FITC to detect azides (30). For lectin staining, wt* and G2E* cells were washed twice with wash buffer (PBS, pH 7.1, containing 1.0% FCS) and incubated with TML-biotin for 1.0 h at room temperature. Following

Periodate-resorcinol determination of sialic acid concentration
Wt* or G2E* cells were incubated with the indicated substrate or with no substrate. Cells were grown to a density between 6.0 x 10 5 and 9.0 x 10 5 cells/ml over three days and the periodate-resorcinol assay was performed as described (28,31).

Synthesis of compound 1
Hydroxylamine hydrochloride (4.0 g, 0.057 mol) and sodium methoxide (3.0 g, 0.057 mol) were stirred in methanol for 30 min. The resulting white salts were removed by filtration.

Synthesis of compound 3
Acetic anhydride (5.0 ml, 0.055 mol) was added to a solution of 2 (1.0 g, 0.0045 mol) in pyridine (10 ml) and the reaction was stirred overnight at room temperature. The solution was concentrated, dissolved in CHCl 3 and washed with 1.0 M HCl, NaHCO 3 and saturated NaCl.
The organic phase was dried over MgSO 4 , filtered and concentrated to give the crude product

Protein expression and purification
For bacterial expression, the human GlcNAc 2-epimerase gene was cloned into pET28b(+). The GlcNAc 2-epimerase gene was amplified by PCR using Pfu DNA polymerase,

Colorimetric assay for GlcNAc 2-epimerase
The colorimetric assay to detect the epimerization of ManNAc by GlcNAc 2-epimerase was performed as described (32), except that the enzymatic reactions were monitored continuously on a Molecular Devices SpectraMax 190 spectrophotometer to acquire initial rates.
Trend lines for inhibition by 1 and 2 were determined from non-linear regression analysis using the program GraFit 4.0 and the data were fit to a competitive inhibition model using the nonlinear regression analysis program, SAS. Trend lines did not converge to fit a noncompetitive pattern for either compound. There was a small background rate that could be attributed to the side reaction of ManNAc with the coupling enzymes, as previously reported (32). In all assays this background signal was subtracted to establish the actual rate of the reaction. In addition, at the highest inhibitor concentrations and lowest substrate concentrations, there was a significant background signal that was attributed to reaction of the inhibitor directly with the coupling enzymes. This value was subtracted but likely contributes some error to the rates measured using these data points.

Introduction of GlcNAc 2-epimerase into human cells
Analysis of mRNA from the human T cell lymphoma line Jurkat by RT-PCR showed no detectable message for GlcNAc 2-epimerase, suggesting that ManNAc is produced exclusively by UDP-GlcNAc 2-epimerase in these cells. Accordingly, the mRNA encoding UDP-GlcNAc 2epimerase was observed by RT-PCR (data not shown).
To achieve stable expression of GlcNAc 2-epimerase in Jurkat cells, we used the twostep FlpIn system (Invitrogen). This approach to stable transfection is particularly useful with genes that do not produce a readily selectable phenotype, as is the case with GlcNAc 2- Tritrichomonas mobilensis (TML-biotin, followed by FITC-avidin), which recognize sialic acid in a linkage-independent fashion, and analyzed by flow cytometry. In all cases, G2E* cells showed very little change in fluorescence compared with wt* cells (data not shown).
Cell surface sialylation does not necessarily reflect the level of free sialic acid within cells. For example, a buildup of sialic acid will not be detected if downstream enzymes, such as the sialyltransferases, are functioning at capacity. Previously, our laboratory has demonstrated that ManNAc analogs bearing chemically detectable probes can be used as tools to monitor the metabolic flux within the sialic acid pathway. For example, analogs of ManNAc that contain a ketone (ManLev, Figure 3A) (33) or an azide (ManNAz) (26) within the N-acyl group are converted by cells to the corresponding sialic acids, SiaLev and SiaNAz, respectively. When expressed on the cell surface, these modified sialic acids can be reacted with chemical probes and quantified by flow cytometry ( Figure 3B). Typically, ketones are detected with biotin hydrazide followed by FITC-avidin (34), and azides are detected with phosphines conjugated to the FLAG peptide followed by an anti-FLAG antibody (35). Since the ManNAc analogs compete with endogenous ManNAc, changes in the intracellular concentration of native sialic acid intermediates will affect unnatural sialic acid expression on the cell surface. Thus, unnatural sialic acid expression can serve as a reporter of cellular metabolic flux. We have previously exploited this phenomenon to study regulatory mechanisms of sialoside expression (28).
In order to observe the effect of GlcNAc 2-epimerase expression on metabolic flux within the sialic acid pathway, we incubated wt* and G2E* cells with peracetylated forms of ManLev (Ac 4 ManLev, Figure 3A) and ManNAz (Ac 4 ManNAz) and then analyzed them by flow cytometry. The peracetylated sugars permeate cell membranes more readily than the unacetylated sugars and can therefore be used at lower concentrations (25,36). The acetyl groups are removed by non-specific esterases in the cytosol, liberating the free sugars inside cells. As shown in Figure 4, G2E* cells produced less unnatural sialic acid on the cell surface than wt* cells when both were treated with either Ac 4 ManLev ( Figure 4A) or Ac 4 ManNAz ( Figure 4B).
In the case of Ac 4 ManLev, the difference in cell surface fluorescence was 5-fold, whereas in the case of Ac 4 ManNAz this difference was 17-fold. Thus, the presence of the epimerase reduces unnatural sialoside expression and the effect is more dramatic with the substrate bearing the smaller N-acyl group.
We have previously found that the size of the N-acyl group on ManNAc derivatives can affect the efficiency of unnatural sialic acid biosynthesis in cells (37) and substrate activity with isolated enzymes in vitro (38). Accordingly, we postulated that GlcNAc 2-epimerase catalyzes the epimerization of ManNAz to the corresponding gluco analog more efficiently than ManLev.
To address this experimentally, we developed an in vitro 1 H NMR assay to measure the relative rates of the GlcNAc 2-epimerase-catalyzed reaction with the two unnatural substrates. As shown in Table 1, the enzyme epimerizes ManNAz 345 times more rapidly than ManLev, and both substrates react more slowly than ManNAc.

The effect of GlcNAc 2-epimerase on intracellular free sialic acid concentrations
To investigate the effect of GlcNAc 2-epimerase on the flux of natural sialic acid biosynthesis, we analyzed free sialic acid content in wt* and G2E* cells. Levels of glycoconjugate-bound and total sialic acid can be measured using the periodate-resorcinol assay and the level of free sialic acid can be determined by subtracting the former from the latter (31).
The level of free sialic acid in untreated Jurkat cells is close to the detection limit of the assay and is difficult to measure reproducibly. Nonetheless, G2E* cells appeared to have lower concentrations of free sialic acid than wt* cells ( Figure 5).
More reliable measurements were obtained by the addition of ManNAc, which increases the amount of free sialic acid in the cell well above the limit of detection. When both cell lines were treated with 10 mM ManNAc, G2E* cells clearly produced less sialic acid than wt* cells ( Figure 5). Thus, in the presence of excess ManNAc, GlcNAc 2-epimerase acts to lower free sialic acid levels, presumably by diverting ManNAc to GlcNAc. The Jurkat cell lines were also both incubated with 10 mM GlcNAc and analyzed for the effect on sialic acid content. The addition of GlcNAc did not measurably increase the level of sialic acid in G2E* cells above that in wt* cells ( Figure 5), suggesting that intracellular ManNAc concentrations did not differ significantly between the two cell lines. To rule out the possibility that the different effects of ManNAc and GlcNAc on sialic acid production were due to differential uptake by the cells, we performed the same experiment with the peracetylated derivatives. These compounds should equally penetrate cell membranes by passive diffusion (36). The peracetylated compounds had the same relative effects on sialic acid content in wt* and G2E* cells as observed with free ManNAc and GlcNAc (data not shown).

Substrate-based inhibitor of GlcNAc 2-epimerase
The previous experiments establish that GlcNAc 2-epimerase expression correlates with a reduction in sialic acid biosynthesis. In order to confirm that this reduction was the direct result of enzyme activity, we designed an inhibitor of GlcNAc 2-epimerase for use in cellular studies.
A mechanism describing the interconversion of GlcNAc and ManNAc by GlcNAc 2-epimerase was recently proposed by Samuel and Tanner ( Figure 6A The assay detects the production of GlcNAc by enzymatic oxidation of the C-6 hydroxyl group and concomitant production of hydrogen peroxide, which is subsequently consumed in a chromogenic peroxidase reaction (32). For ease of analysis we modified the assay to monitor the reaction continuously.
The value of K M obtained for ManNAc (4.3 mM) was close to the literature value (13.2 mM) obtained using the endpoint assay (46). We then measured the inhibition constants (K I ) of 1 and 2 to be 90 µM and 109 µM, respectively. By varying the concentrations of substrate and inhibitor, we obtained kinetic data consistent with a competitive inhibition mechanism for both 1 and 2, indicating that both analogs compete with ManNAc for the active site (Figure 7). To make an inhibitor useful in cell culture (36), compound 2 was acetylated to form compound 3 ( Figure 6B).

GlcNAc 2-epimerase inhibitor is active in cells
Compound 3 was used to confirm that GlcNAc 2-epimerase activity was responsible for the reduced conversion of Ac 4 ManLev to SiaLev in G2E* cells compared to wt* cells. As shown in Figure 8, compound 3 reverses the effect of GlcNAc 2-epimerase in a dose-dependent fashion, nearly restoring unnatural sialic acid biosynthesis to wt* levels at 200 µM. Importantly, the production of SiaLev in wt* cells was not significantly affected by the addition of 3.
Finally, we incubated wt* and G2E* cells with ManNAc in the presence and absence of compound 3 and analyzed free sialic acid levels using the periodate-resorcinol assay (Figure 9).
Although the sialic acid level in G2E* cells was still lower than that in wt* cells, the differential was significantly reduced by the GlcNAc 2-epimerase inhibitor (100 µM). Thus, the inhibitor reverses the low sialic acid phenotype characteristic of G2E* cells, confirming that the enzyme is directly responsible for that phenotype.  Figure 10A.

DISCUSSION
Alternatively, it may be possible that G2E* cells produce higher levels of ManNAc than wt* cells due to epimerization of GlcNAc, a model summarized in Figure 10B. Indeed, ManLev was previously used by our laboratory as a tool to select a Jurkat cell line containing mutations in the UDP-GlcNAc 2-epimerase gene that make the protein resistant to inhibition by CMP-sialic acid (28). Lacking that regulatory mechanism, these cells overproduced ManNAc, which suppressed unnatural sialic acid expression by direct competition with ManLev. The models shown in Figure 10 are not mutually exclusive; GlcNAc 2-epimerase may perform either of these functions depending on the relative availabilities of GlcNAc and ManNAc in the cell. Since the inhibitors were designed to mimic the ring-opened form of GlcNAc, this lends support to the proposed mechanism in Figure 6A (47,48). Increases in cellular GlcNAc may affect the flux in these pathways. Alternatively, phosphorylation of GlcNAc followed by deacetylation of GlcNAc 6-phosphate to glucosamine 6-phosphate generates a substrate for glycolysis (49). It is possible that GlcNAc 2-epimerase serves to increase glycolytic flux at the expense of protein glycosylation.
We conclude that GlcNAc 2-epimerase catalyzes the conversion of ManNAc (and ManNAc analogs) to GlcNAc (and GlcNAc analogs) in human cells, but the reverse process is unobservable, even in the presence of added GlcNAc. Furthermore, analogs of the ring-opened form of GlcNAc are effective inhibitors of GlcNAc 2-epimerase in vitro, lending support to a proposed chemical mechanism. One of these inhibitors reversed the sialic acid reduction induced by GlcNAc 2-epimerase, confirming that the phenotype observed was directly attributable to activity of the enzyme and demonstrating the utility of the inhibitor in human cells. Based on the results presented here, we propose that GlcNAc 2-epimerase is not an alternate route to ManNAc production in cells. Rather, the enzyme diverts metabolic flux away from sialic acid biosynthesis, performing a function distinct from that of UDP-GlcNAc 2epimerase.  yield ManNAc 6-phosphate (ManNAc 6-P). ManNAc 6-P is subsequently condensed with phosphoenolpyruvate (PEP) to yield sialic acid 9-phosphate (sialic acid 9-P) in a reaction catalyzed by sialic acid 9-P synthase. Dephosphorylation of sialic acid 9-P by an unknown phosphatase and transport to the nucleus enables CMP-sialic acid synthetase to produce CMPsialic acid. Following transport into the Golgi compartment, CMP-sialic acid is utilized by the sialyltransferases that append the sialic acid to glycoconjugates ultimately destined for the cell surface or secretion.