Biochemical characterization of the Campylobacter jejuni Cj1294, a novel UDP-4-keto-6-deoxy-GlcNAc aminotransferase that generates UDP-4-amino-4,6-dideoxy-GalNAc.

Campylobacter jejuni produces multiple glycoproteins whose glycans contain 4-amino 6-deoxy sugars or their derivatives, such as diacetamidobacillosamine or pseudaminic acid. Because the proteoglycans contribute to bacterial virulence and their constitutive sugars are not commonly found in humans, inhibitors developed against the enzymes that are responsible for their biosynthesis could be novel therapeutic targets to fight this important food-borne pathogen. The biosynthesis of diacetamidobacillosamine is anticipated to involve a sugar nucleotide C6 dehydratase, a C4 aminotransferase and an acetyltransferase. We have identified a set of genes (cj1293, cj1294, and cj1298) potentially encoding one of each enzymatic activity, and demonstrated earlier that Cj1293 was a UDP-GlcNAc-specific C6 dehydratase. Others have shown that Cj1293 was involved in protein glycosylation. Here, we report on our investigation of the potential activity of Cj1294 as a sugar nucleotide C4 aminotransferase. Our biochemical characterization of overexpressed and purified protein shows that Cj1294 is a pyridoxal phosphate-dependent aminotransferase specific for UDP-4-keto-6-deoxy-GlcNAc that uses preferentially glutamic acid as an amino donor. A detailed physicokinetic study of Cj1294 was performed to determine the K(m) of 1.28 +/- 0.2 mm and k(cat) of 11.5 +/- 1.3 min(-1). Also, two residues essential for protein stability and activity, Arg(228) and Lys(181), respectively, were identified by site-directed mutagenesis. Finally, we demonstrated by NMR analysis of purified reaction product that Cj1294 produces UDP-4-amino-4,6-dideoxy-GalNAc. These results indicate that Cj1294 is involved in the biosynthesis of diacetamidofucosamine, a C4 epimer of diacetamidobacillosamine not yet described in C. jejuni proteoglycans, suggesting that the composition of C. jejuni proteoglycans is more variable than anticipated.

Protein glycosylation is an important phenomenon that, in prokaryotes, was first reported for S-layer proteins (1, 2) and has now been described for many other bacterial proteins (3)(4)(5)(6)(7)(8). Bacterial protein glycosylation has become an area of intense interest since it was shown that protein components of surface appendages such as pilins and flagella that are essential for the virulence of various pathogens were glycosylated (7, 9 -19). The role for such glycosylation is not fully understood although it has been suggested that glycosylation might be essential for secretion/assembly of flagella (15).
In Campylobacter jejuni, protein glycosylation includes, but is far from being limited to, glycosylation of the flagellins. In fact, over 30 different proteins have been shown to be glycosylated (20). Although most of these glycoproteins have unknown specific function, a potential global role of their glycosylation for bacterial virulence was demonstrated, since lack of glycosylation affected interactions with host cells (21). It also resulted in lower antigenicity, which might be beneficial for evasion of host immune defenses.
Whereas flagellin glycans comprise mostly O-linked pseudaminic acid or its derivatives (18), other C. jejuni glycoproteins harbor N-linked heptasaccharides containing N-acetylgalactosamine, glucose and diacetamidobacillosamine (2,4-diacetamido-2,4,6-trideoxy-glucopyranose, DAB) 1 (22). It has been proposed that formation of pseudaminic acid might involve condensation of DAB with phosphoenolpyruvate (PEP) (18,23), so that DAB might be a precursor for the synthesis of pseudaminic acid. To date, the biosynthesis pathways for DAB and pseudaminic acid have not been demonstrated at the biochemical level. However, elucidating these pathways would not only be of fundamental interest, but would also allow for an assessment of the role of protein glycosylation on bacterial virulence by offering the possibility to inhibit specifically each step of the pathway and assess the resulting effect on bacterial virulence.
Here, we report our investigation of the potential role of the putative aminotransferase Cj1294 in DAB biosynthesis. Using pure, overexpressed enzyme, we demonstrate that Cj1294 is a pyridoxal phosphate (PLP)-dependent aminotransferase specific for UDP-4-keto-6-deoxy-GlcNAc. We also demonstrate by NMR analysis of purified reaction product that Cj1294 is involved in the biosynthesis of a sugar distinct from DAB since Cj1294 produces UDP-4-amino-4,6-dideoxy-GalNAc. These results contribute to the fundamental understanding of the biosynthesis of sugars dedicated to protein glycosylation in C. jejuni and suggest that additional sugars, other than DAB or pseudaminic acid, might be part of the C. jejuni proteoglycans.

EXPERIMENTAL PROCEDURES
Cloning of cj1294 -cj1294 was PCR-amplified from chromosomal DNA from C. jejuni subspecies doylei ATCC 49349 (kindly provided by Dr. J. Odumeru, University of Guelph, Canada) using primers designed based on genomic information available for strain ATCC 700819. The primers used were AGGGTACATCTCCATGCTTACTTATTCTCATCA and GCGTCGGATCCTTATCCACAATATCCCTTTTT. Cloning into the pET expression vector was performed by standard procedures using BamHI and AflIII sites. The resulting construct was sequenced using primer: CGATGATGCAAGCCATGC and the T7 promoter primer at the DNA sequencing facility of the Robarts Institute (London, Ontario). The sequence has been deposited in GenBank TM under the accession number AY573603.
Construction of the K181R and R228A Mutants by Site-directed Mutagenesis-Site-directed mutagenesis was performed using the QuikChange procedure following the manufacturer's instructions (Stratagene). The primers used were CTTTTCACCCTGTTGCGCCTA-TCACTACTTTTG and CAAAAGTAGTGATAGGCGCAACAGGGTGA-AAAG to generate the K181R mutant, and CTTGGCTATAATTACGC-GTTAAGTGATGTTGC and GCAACATCACTTAACGCGTAATTATAG-CCAAG to generate the R228A mutant.
Protein Expression and Purification-Protein expression was performed in Escherichia coli BL21(DE3)pLys, using Luria Bertani broth supplemented with 100 g/ml ampicillin and 34 g/ml chloramphenicol at 25°C. After 3 h of induction with 0.15 mM isopropyl ␤-D-1-thiogalactopyranoside, the cells were pelleted by centrifugation at 18,500 ϫ g for 20 min, and the pellets were stored at Ϫ20°C until needed. Protein purification was performed by nickel chelation as described previously (24) except that the Tris buffer was replaced by Hepes 20 mM, pH 7.5. At the end of the purification, PLP was added to the pooled pure fractions to a final concentration of 0.012 mM, incubated for 1 h at 37°C, and dialysis was performed overnight in Hepes 100 mM pH 7.5. The protein was concentrated using polyethylene glycol 8000 as needed. For the R228A mutant, dialysis was performed in 20 mM Tris, pH 7.5 containing 50 mM NaCl and an extra step of anion exchange chromatography was performed using a MonoQ HR 5/5 (1 ml) column (Amersham Biosciences, 17-0546-01) and a linear gradient of 20 mM Tris, pH 7.5, 1 M NaCl in 10 column volumes. The purified protein was dialyzed in Hepes 100 mM, pH 7.5 for further analysis. Western blotting was performed using a commercial mouse anti-histidine tag primary antibody (Amersham Biosciences) and a goat anti-mouse IgG conjugated with Alexa Fluor 680 secondary antibody using standard procedures. Detection was achieved by scanning with a Li-Cor Odyssey Infrared Imaging System. Expression and purification of Cj1293 and its Helicobacter pylori homologue FlaA1, were performed as reported before (24,25).
Enzyme Assays-Coupled assays between Cj1294 and Cj1293 or FlaA1 were performed by incubating 0.12 g of FlaA1 or Cj1293 and 0.8 g of Cj1294 with 0.5 mM UDP-GlcNAc, 10 mM glutamic acid (or other amino acid), and 0.1 mM PLP (for fractions not preincubated with PLP at the end of the purification) in 20 mM Hepes buffer pH 7.5, in a final volume of 20 l unless stated otherwise. Reactions were incubated for 5 h at 37°C. The reaction products were analyzed by capillary electrophoresis (CE) as described previously (25,26).
For determination of physicokinetic parameters, reactions were first performed with UDP-GlcNAc and FlaA1 to obtain the 4-keto intermediate, the dehydratase was removed by ultrafiltration (Nanosep centrifugal device 10,000 cutoff (Pall, Invitrogen)) and the filtrate was used for further reaction with the aminotransferase. This ensured that the measured parameters would only reflect intrinsic properties of Cj1294. The concentration in 4-keto intermediate in the filtrate was determined by CE using the percentage of conversion of the original UDP-GlcNAc substrate. To determine the optimum temperature and pH, the filtrate was incubated with Cj1294 at the appropriate temperature (4 -65°C) or pH (4.5-10.5). For the pH study, sodium acetate 100 mM was used for pH 4.5-6.5 and Bis-Tris-propane 100 mM was used for pH 6.5-10.5. For determination of K m , V max and turn over parameters, 11 reactions with varying concentrations of 4-keto intermediate were set up in triplicates and incubated for 1 h at 42°C, which ensured less than 10% substrate conversion over the whole range of substrate concentrations tested. The data are the average of three independent experiments.
Purification of the Amination Product-The purification of the amination product was performed by anion exchange using a High Q Econopac 1-ml column (Bio-Rad) (27) and a linear gradient of 20 column volumes of triethylammonium bicarbonate pH 8.5 (50 mM to 1 M) at 1 ml/min. Fractions containing the reaction product were pooled and lyophilized twice with resuspension in water between both lyophiliza-tion steps. Fractions were resuspended in water for CE and NMR analysis.
Determination of the Structure of the Amino Product by NMR-All NMR data were collected at 25°C on a Varian INOVA 600 MHz spectrometer equipped with a pulsed-field gradient triple resonance probe. The sample was dissolved in D 2 O at a final concentration of 0.3 mM. Its final pH was 8.5. 1 H chemical shifts were externally referenced to the methyl group of 2,2-dimethyl-2-silapentane-5-sulfonate sodium salt (0.0 ppm at 25°C). The 13 C reference was calculated as described (28). One dimensional 1 H and two-dimensional TOCSY, NOESY, HSQC, and HMBC spectra were measured by using Varian standard pulse sequences. The NMR data were processed and analyzed with the VNMR software.
Reactivity of the Purified Cj1294 Reaction Product with 2,4,6-Trinitrobenzensulfonic Acid (TNBS)-The purified Cj1294 reaction product (0.25 nmol) was incubated for 1 h at 50°C with 5 nmol of TNBS (Sigma) in 100 mM Hepes pH 7.5 (total reaction volume, 20 l). Similar reactions were carried out on UDP-GlcNAc as a negative control. Samples were analyzed directly by CE after incubation.
Mass Spectrometry Analyses-Mass spectrometry analyses were performed at the Dr. Don Rix Protein Identification Facility of the University of Western Ontario. Pure Cj1294 was subjected to trypsinolysis, and the peptides were analyzed by LC-MS/MS (Q-TOF2) and MALDI-MS. Purified amino product was analyzed on a Micromass Qtof Micro mass spectrometer equipped with a Z-spray source operating in the negative ion mode (40 V, 80°C).

RESULTS
Proposed Pathway for DAB Biosynthesis-Theoretically, the biosynthesis of DAB necessitates the activity of a C6 dehydratase to form a 4-keto-6-deoxy intermediate starting from a 2-N-acetyl sugar, a C4 aminotransferase from the degT family to add an amino group on the generated reactive 4-ketone, and an N-acetyltransferase to acetylate the newly incorporated amino group. The genome of C. jejuni contains two sets of genes potentially able to encode the necessary activities, but in the absence of biochemical data, it is not clear if these two sets of genes are functionally redundant or not. One set (cj1120c, cj1121c, and cj1123c) is part of the general protein glycosylation (pgl) cluster (20) that can sustain production of N-glycosylated proteins harboring 2,4-diacetamido-2,4,6-trideoxyhexoses in E. coli (29). The other set (cj1293, cj1294, and cj1298, Fig. 1), which is the focus of this report, includes the UDP-GlcNAc-specific C6 dehydratase Cj1293, that produces a 4-keto-6-deoxy intermediate suitable for further transamination (24). It was shown that disruption of cj1293 abrogated glycosylation of flagellins by pseudaminic acid (30). Also, these genes are in close proximity to the flagellin genes and to a gene encoding a neuraminic acid synthase homologue that was re-FIG. 1. Schematic representation of the cj1293-cj1298 operon containing genes potentially involved in DAB biosynthesis. cj1293 was shown to encode for a UDP-GlcNAc C6 dehydratase (24) but the activities of Cj1294 and Cj1298 as aminotransferase and acetyltransferase, respectively, are hypothetical. Cj1296 and Cj1297 show homologies to the N-and C-terminal halves of Cj1298, respectively. Cj1295 is not homologous to any known protein.
cently proven to be essential for the production of flagella (18,31). This supports the idea of a potential role of cj1293, cj1294, and cj1298 in the biosynthesis of the DAB precursor used to form pseudaminic acid, which is subsequently used for flagellin glycosylation. Among all the enzymes potentially involved, Cj1293 is the only one that has been characterized at the biochemical level (24) and based on its UDP-GlcNAc dehydratase activity, we propose the DAB biosynthesis pathway highlighted in Fig. 2, whereby UDP-GlcNAc would be the initial precursor. We decided to pursue the investigation of the role of the cj1293-cj1298 set of genes in DAB biosynthesis by determining the biochemical activity of the putative aminotransferase Cj1294.
Protein Expression and Purification-cj1294 was cloned from C. jejuni strain ATCC 49349 using primers based on genomic information available for strain ATCC 700819 (32). DNA sequencing revealed that the sequences of Cj1294 from both strains were 98.7% identical at the protein level. Identities included residues shown to be essential for cofactor binding in related enzymes, such as Asp 152 , Ser 176 , and Lys 181 , and residues essential for dimerization such as Asn 226 and Arg 228 (33).
The Cj1294 protein (wild-type and K181R mutant, see rationale for mutant below) could be readily overexpressed in a soluble form and purified to homogeneity in a single step of metal chelation (Fig. 3, panel A for wild-type). The identity of the proteins was confirmed by Western blotting with an anti-histidine tag monoclonal antibody (Fig. 3, panel B for wild-type) as well as by sequencing by mass spectrometry. Sixteen peptides covering 49% of the protein were sequenced (data not shown). The R228A mutant (see rationale below) was mostly expressed in an insoluble form. Purification of the soluble fraction of R228A by metal chelation was inefficient because of poor binding to the resin. Further attempts at purifying it by anion-exchange chromatography were unsuccessful because of protein precipitation. This indicates that Arg 228 is essential for protein folding and stability but prevented further characterization.
The Substrate for Cj1294 is UDP-4-keto-6-deoxy-GlcNAc-The ability of Cj1294 to use UDP-4-keto-6-deoxy-GlcNAc as a transamination substrate as predicted in Fig. 2 was demonstrated by performing coupled assays with the UDP-GlcNAc C6 dehydratase/C4 reductase Cj1293 (24) or its Helicobacter pylori homologue FlaA1 (25), in the presence of UDP-GlcNAc as the substrate for the dehydratase, and in the presence of glutamate as an amino donor for the transaminase (Fig. 4). Cj1293 and FlaA1 convert UDP-GlcNAc (peak A) into the UDP-4-keto-6deoxy-GlcNAc intermediate of interest (peak B) and further reduce it into peak C (Fig. 4, lines a and c). In the presence of Cj1294, and in the presence of the appropriate amino donor (glutamate preferentially, see below), the 4-keto intermediate is used up to generate peak D, potentially corresponding to the amination reaction product (Fig. 4, lines d-h). No reaction was observed in the absence of dehydratase, indicating that UDP-GlcNAc is not a substrate for Cj1294 (Fig. 4, line b). The transamination reaction drives the conversion of UDP-GlcNAc by FlaA1/Cj1293 almost to completion so that overall, at equilibrium, 80% of the original substrate had been converted into amino product, and only 11% of UDP-GlcNAc, 5% of peak C and 4% of the 4-keto intermediate were left under the conditions tested (Fig. 4, line h). The same results were obtained whether Cj1293 or FlaA1 was used as the original dehydratase, and the final amination reaction products obtained using either dehydratase to form the 4-keto intermediate co-eluted in a single sharp peak by CE when co-injected as a mix (data not shown). This indicates that the amination reaction products are identical notwithstanding which dehydratase is used to generate the 4-keto intermediate. This is consistent with the demonstrated identical biochemical activities of the dehydratases (24,25). FlaA1 being more stable than Cj1293, all subsequent experiments were performed with FlaA1.
The production of the putative amino product (peak D) by Cj1294 was directly dependent on the amount of 4-keto intermediate generated, as controlled by the amount of dehydratase present in the coupled assay (Fig. 5, panel A). It was also dependent on and saturable with the amount of Cj1294 (Fig. 5,  panel B).  figure) to this reaction mixture. Incubation of UDP-GlcNAc (peak A) with the dehydratase (shown here using FlaA1) generates UDP-4-keto-6-deoxy-GlcNAc (peak B, 4-keto intermediate) and its reduced derivative (peak C). Upon addition of Cj1294 and glutamate, the 4-keto intermediate is modified into peak D. The yield of peak D is considerably improved if Cj1294 is preincubated with PLP at the end of its purification, or if PLP is added directly to the transamination reaction.
In the reactions described above, the consumption of 4-keto substrate by Cj1294 prevents its reduction by FlaA1 or Cj1293 so that very little formation of peak C is observed. However, it could be argued that peak C is also a substrate for Cj1294. To ascertain that this is not the case, reactions were performed first with FlaA1 (or Cj1293) and UDP-GlcNAc, the dehydratase was removed by ultrafiltration, and Cj1294 was added to the filtrate together with glutamate and further incubated at 37°C. Under these conditions, a clear reduction of the 4-keto peak (Fig. 6, peak B) was observed concomitant to the formation of peak D, but no variation in the intensity of peak C was observed ( Fig. 6, peak C). This confirms that peak C is not a substrate for Cj1294. This experiment also further confirmed that UDP-GlcNAc was not a substrate for Cj1294 either, since the UDP-GlcNAc peak was not affected by addition of Cj1294.
Cj1294 Is Highly Specific for Glutamate as an Amino Donor-As mentioned earlier (Fig. 4, line d versus lines e-h), an amino donor is necessary to observe Cj1294-catalyzed formation of peak D. Screening of all 20 amino acids indicated that glutamate is the preferred amino donor used by Cj1294 (Fig. 7, line e) although significant levels of conversion were also obtained with methionine, alanine, and glutamine (Fig. 7, lines  b-d). No conversion was obtained using aspartate (data not shown).
The yields of reaction could be greatly enhanced by using excess amino donor over the substrate, without altering amino donor specificity. As shown in the case of the preferred amino donor, glutamate, maximal catalysis was observed for a concentration of glutamate of about 10 mM, which represents at least a 20 molar excess over the 4-keto substrate, assuming 100% conversion of the original UDP-GlcNAc (Fig. 5, panel C). This is attributed to the fact that the reaction is reversible (34, 35) but the equilibrium is not favorable for transamination. This concentration that affords maximal catalysis was chosen for all subsequent kinetic experiments.
Cj1294 Uses PLP as a Cofactor-The mechanism of transamination implies the involvement of PLP as a cofactor. When reactions were performed without addition of exogenous PLP, using enzyme that had not received any PLP after purification (Fig. 4, line e), a significant level of activity was observed (48% of original substrate converted to amino product). However, if PLP was added during the reaction (Fig. 4, line f), 79% substrate conversion could be observed. This suggested that most, but not all, of the overexpressed enzyme already possessed bound PLP in its active site, indicating that the intracellular level of PLP was probably limiting and did not allow saturation of all the enzyme molecules. Addition of PLP to the reaction allowed full saturation of the enzyme with cofactor. In contrast, when PLP had been added to the enzyme right after purification, and excess PLP had been dialyzed away, the enzyme already functioned at maximal capacity (80% conversion) without further addition of PLP during the reaction (Fig. 4, lines g  and h).

Residue Arg 228 Is Essential for Protein Stability, Probably
Affecting Protein Dimerization-Gel filtration chromatography indicated that Cj1294 forms a dimer in solution (data not shown), which is consistent with the fact that sugar nucleotide aminotransferases form dimers (36) or tetramers (37,38), and that Cj1294 shares 63% similarity and 30% identity with 3-amino-5-hydroxybenzoic acid (AHBA) synthase, a dimeric aminotransferase (33). The PLP binding site in AHBA involves residues from two subunits found at the dimerization interface. One of these residues, Arg 236 , is essential for dimerization and PLP binding. This residue is conserved in Cj1294 (Arg 228 ) as well as in Cj1121c, but is not conserved in ArnB, an aminotransferase involved in 4-amino arabinose biosynthesis (35). A R228A mutant of Cj1294 was constructed by site-directed mutagenesis to investigate the impact of this residue on the ability of the protein to dimerize. Unfortunately, the R228A mutation resulted in total unstability of the protein, which could not be purified efficiently by metal chelation or anion exchange chromatography and precipitated readily. Although this does not allow concluding as far as the role of Arg 228 in dimerization goes, this indicates that this residue is critical for protein folding and stability.
Residue Lys 181 Is Essential for Activity, Probably via Aldimine Formation with PLP-Lys 181 corresponds to the conserved Lys 188 in AHBA synthase or ArnB that is part of a SX 5 K conserved catalytic motif (33,35,38). This residue is responsible for the formation of an internal aldimine with PLP in these enzymes. Hence, to further demonstrate the role of PLP in Cj1294-mediated catalysis, a K188R mutant was generated by site-directed mutagenesis. This K181R mutant was totally inactive under all conditions tested, although it could be ex- and its reduced derivative (peak C). The dehydratase was removed by ultrafiltration and the filtrate was incubated further in the presence (trace 2) or absence (trace 1) of Cj1294. Appearance of the transamination product (peak D) was accompanied by a significant decrease in surface area of the 4-keto intermediate peak whereas the other peaks were not affected by the presence of Cj1294. This indicates that Cj1294 uses exclusively the 4-keto intermediate as a substrate.

FIG. 5. Yields of transamination reaction as a function of the amount of 4-keto intermediate present, as controlled by the amount of FlaA1 dehydratase used in the reaction (panel A), or as a function of the amount of aminotransferase (panel B) or glutamate (panel C).
Squares, UDP-GlcNAc; circles, transamination product. Inverted triangles, 4-keto intermediate. Triangles, reduced derivative of the 4-keto intermediate.
pressed at normal levels, was purified to homogeneity in a single step of metal chelation and showed a normal wild-typelike pattern of dimerization, suggesting proper overall folding.
Physicokinetic Characteristics of Cj1294 -Contrary to what is observed for the perosamine synthase RfbE (37), addition of divalent cations such as Mg 2ϩ , Mn 2ϩ , or Ca 2ϩ did not improve the performance of Cj1294. An optimal pH of 7.5 and an optimal temperature of 37-42°C were determined for Cj1294 (data not shown). Surprisingly, significant catalysis was obtained at 4°C so that overall, the enzyme is active from 4°C to 42°C. This could reflect the fact that C. jejuni survives in environmental water or refrigerated food and activity of this enzyme might be important in the process. The optimal pH of 7.5 and temperature of 42°C were chosen for all subsequent kinetic characterizations. The determination of the kinetic parameters was performed under conditions (amount of dehydratase, aminotransferase and glutamate, Fig. 5) that result in optimal catalysis. The K m and k cat of Cj1294 for UDP-4-keto-6-deoxy-GlcNAc were 1.28 Ϯ 0.20 mM and 11.5 Ϯ 1.3 min Ϫ1 , respectively, resulting in a turnover k cat /K m of 8.99 min Ϫ1 mM Ϫ1 .
Characterization of the Reaction Product-Mass spectrometry analysis of the purified reaction product showed a peak at m/z 590 whose fractionation pattern was consistent with the pattern expected for an aminated derivative of the 4-keto intermediate (data not shown). To prove that the reaction product (peak D, Fig. 1) was an aminated derivative, it was purified by anion-exchange chromatography and tested for its reactivity with TNBS, a reagent specific for primary amino groups. The pure reaction product reacted readily with TNBS whereas none of the other sugars tested (UDP-GlcNAc, 4-keto intermediate, peak C) did (Fig. 8).
To confirm that the newly incorporated amino group was present on carbon 4 and determine its position relatively to the sugar ring (axial versus equatorial), NMR analysis was performed on the pure reaction product. The proton chemical shift assignments were done by analyzing through-bond connections in two-dimensional TOCSY experiments. Carbon chemical shift assignments were obtained from two-dimensional HSQC experiments. As shown by CE analysis, the sample that was subjected to NMR analysis contained a single sugar nucleotide species (inset in Fig. 9). Although it contained some impurities coming from the buffer as seen by NMR, the NMR spectra allowed the complete assignment of chemical shift resonances.
The chemical shifts of the uracyl and ribose moieties were very similar to those of UDP-GlcNAc (39), confirming that the reaction did not affect the UDP moiety. The well-isolated ano-meric signal H-1Љ and the methyl group H-6 served as starting points for the chemical shifts assignments of the pyranosyl moiety. The appearance of the anomeric proton signal H-1Љ at 5.60 ppm suggested the pyranosyl ring was in an ␣-anomeric configuration with H-1Љ proton in equatorial position (40), as in the starting material. Interestingly, the chemical shift of the C-4Љ carbon was shifted upfield to 55 ppm (Fig. 9), an indication of amino group substitution on carbon 4 (40). In the twodimensional HMBC spectrum, the H-6Љ proton exhibits a heteronuclear multiple bond correlation with a carbon resonance at 55 ppm assigned to C-4Љ carbon. This further verified that the amino group substitution happened at C-4Љ position. Furthermore, one-dimensional proton spectrum allowed estimations of J 1Љ2Љ , J 2Љ3Љ , and J 3Љ4Љ coupling constant as 2.3, 5.4, and 3.0 Hz, respectively. The J 2Љ3Љ coupling constant of 5.4 Hz indicates that both H-2Љ and H-3Љ protons are in axial positions (40), identical to the starting conformation. The H-4Љ proton is in equatorial position as evidenced by the small coupling of J 3Љ4Љ . Hence, its amino substituent assumes an axial position, corresponding to the galactose configuration.  1, trace a) was purified by FPLC and incubated with TNBS before CE analysis. Disappearance of peak 1 upon incubation with TNBS (trace b) indicates that the Cj1294 reaction product is an amino sugar, consistent with the predicted aminotransferase activity of Cj1294. No reaction was observed when the same experiment was performed with UDP-GlcNAc (peak 2, with TNBS, trace d; without TNBS, trace c) as a negative control. Peak 3 (traces b and d) is excess TNBS reagent.
FIG. 9. CE analysis (inset) and NMR analysis (HSQC spectrum) of the reaction product generated by Cj1294. Inset trace a, baseline. Inset trace b, purified amination product. The CE analysis indicates that the purified product subjected to NMR analysis contains a single sugar nucleotide. The NMR data (HSQC, TOCSY, HMBC, and NOESY) are consistent with the presence of an amino group in the reaction product, and indicate that this amino group is attached in axial position to C4 of the sugar ring. Altogether, these NMR analyses indicate that the reaction product of Cj1294 is UDP-4-amino-4,6-dideoxy-GalNAc.
The two-dimensional NOESY spectrum showed the expected strong NOE correlations between H-1Љ and H-2Љ, H-3Љ, and H-4Љ but also showed a strong and unexpected NOE between the H-1Љ and H-5Љ protons. This, combined with the relatively small J 2Љ3Љ coupling between protons of diaxial relationship, suggests that the sugar ring might be slightly distorted.
In summary, Cj1294 is a UDP-4-keto-6-deoxy-GlcNAc-specific C4 transaminase that generates UDP-4-amino-4,6dideoxy-GalNAc in the presence of glutamate and PLP. DISCUSSION Deoxyamino sugars are synthesized by plants, fungi and bacteria. In bacteria, these sugars are often found in lipopolysaccharide, extracellular polysaccharides or proteoglycans such as that of S-layers, flagella, and pili (41,42). Since the presence of LPS, flagella or pili is often correlated with an increased virulence potential in pathogenic bacteria, genes responsible for the biosynthesis of these sugars are potential targets for the development of antimicrobial drugs. Deoxyamino sugars are also building blocks of numerous macrolide antibiotics. Being able to synthesize these sugars and/or modify their final structure might allow the design of novel molecules with antimicrobial properties, which is important in the context of rapidly emerging antibiotic resistant pathogens (43). All these potential applications require understanding the molecular basis for the biosynthesis of these amino sugars.
Although genes potentially encoding for sugar-nucleotide aminotransferases necessary for the synthesis of deoxyamino sugars are present in the genome of numerous bacteria (32, 44 -46), very few of these enzymes have been studied at the biochemical level. The pioneering work on TDP-4-keto-6-deoxy-D-glucose aminotransferases from E. coli and Pasteurella pseudotuberculosis (type VO) (34,47), that has recently been confirmed using the E. coli WecE (38), has established the basic mechanism of action. Similar results have also been obtained on a dTDP-6-deoxy-D-xylohex-3-ulose transaminase from Aneurinibacillus thermoaerophilus (41) and a Vibrio cholerae GDP-4-keto-6-deoxymannose transaminase (RfbE) (37). The only sugar-nucleotide aminotransferases proven to function on a UDP-linked sugar are the ArnB proteins from Salmonella typhimurium and E. coli which produce UDP-4-amino-4-deoxy-L-arabinose (33,35). However, this report on C. jejuni Cj1294 is the first biochemical characterization of a UDP-4-keto-6-deoxy-GlcNAc-specific C4 aminotransferase available to date.
Several lines of evidence suggest that Cj1294 uses PLP for catalysis as expected: (i) its increased enzymatic efficiency upon addition of PLP after purification or during reaction (Fig. 4), (ii) the inactivity of a K181R mutant, predicted to be impaired for formation of an internal aldimine with PLP, and (iii) the instability of a R228A mutant, predicted to be impaired in its ability to dimerize and bind PLP. As seen with similar PLP binding enzymes (34), addition of PLP to purified Cj1294 considerably increased its stability upon storage. All kinetic assays were performed using Cj1294 enzyme that had been saturated with PLP after purification to ensure maximal catalytic efficiency.
Based on the average of three independent experiments, we obtained a K m of 1.28 Ϯ 0.20 mM and k cat of 11.5 min Ϫ1 Ϯ 1.3. The only other kinetic data available to date are for the TDP-4-keto-6-deoxy-D-glucose aminotransferases from E. coli (WecE) and P. pseudotuberculosis (34,47) which have a K m for the 4-keto substrate of 0.11 and 1 mM, respectively. Our K m value is fairly comparable to that obtained for the latter enzyme. Differences seen with parameters described for WecE could be due to the fact that a coupled assay was used for its characterization. However, the k cat of 11.5 min Ϫ1 and turnover of 8.99 mM Ϫ1 min Ϫ1 obtained for Cj1294 are significantly lower than the values of 22.8 min Ϫ1 and 217 mM Ϫ1 min Ϫ1 reported for the Pasteurella enzyme, suggesting that catalysis is less efficient for Cj1294.
NMR analyses indicated that the product generated is UDP-4-amino-4,6-dideoxy-GalNAc (Fig. 9). Cj1294 only carries out a transamination reaction and does not carry out an epimerization reaction per se since it is using a preformed 4-keto intermediate as a substrate. However, there is a change in epimer status along the biosynthesis pathway, switching from a glucose epimer used for the dehydration steps by Cj1293 (or FlaA1) to a galactose epimer obtained after transamination by Cj1294 (Fig. 10). Such a switch along the pathway has been observed in WecE (38) and its P. pseudotuberculosis homologue (34) that make TDP-4-amino-4,6-dideoxy-Gal in a pathway that starts from TDP-Glc. The molecular basis for such a change is not clear yet and will require structural investigation.
As explained above, we had assumed that Cj1294 was involved in DAB biosynthesis (Fig. 2) but the NMR results question this assumption. Indeed, synthesis of DAB starting from the reaction product generated by Cj1294 would require an additional C4 epimerization step but, apart from the C4 epimerase GalE, the C. jejuni genome does not contain such an epimerase (32). Also, C4 epimerization requires the presence of a C4 hydroxyl group (48) not present on the Cj1294 reaction product. We are currently examining if Cj1121c from in the pgl operon (20) is instead the aminotransferase that is specifically responsible for DAB biosynthesis (Figs. 2 and 10) by direct biochemical analysis. If Cj1121c has a different stereospecificity than Cj1294, i.e. makes UDP-4-amino-4,6-dideoxy-GlcNAc, this could explain why the apparently duplicated operons (cj1293-cj1298 on the one hand, and pgl genes on the other hand), are not functionally redundant based on mutagenesis studies (22,30).
Based on our enzymatic and NMR data, Cj1294 is involved in the production of a sugar not yet identified on C. jejuni glycoproteins. Most of the information available regarding C. jejuni proteoglycans stems from a study where the glycoproteins had The two steps proven at the biochemical level are those involving the formation of UDP-4-keto-6-deoxy-GlcNAc by Cj1293 (24) and UDP-4amino-4,6-dideoxy-GalNAc by Cj1294 (this study). The steps for formation of the final UDP-4-acetamido-4,6-dideoxy-GalNAc and UDP-DAB are still hypothetical. been extracted using soybean agglutinin (22). The use of a different lectin might have led to the identification of glycans of different composition. Hence, it is reasonable to assume that glycans containing sugars other than DAB are also present on C. jejuni glycoproteins, indicating that variability of proteoglycan composition is greater than anticipated.
In conclusion, this is the first biochemical characterization of a UDP-4-keto-6-deoxy-GlcNAc aminotransferase that generates UDP-4-amino-4,6-dideoxy-GalNAc in a glutamate-and PLP-dependent manner. Homologues of Cj1294 exist in several other bacteria (49 -51), including in the closely related bacterium H. pylori (44), indicating that Cj1294 is the prototype of a large family of C4 sugar nucleotide aminotransferases with novel substrate specificity. The methods described herein can now be used to test these homologues for different substrate and product specificities, and establish the underlying molecular basis by structural analysis of the proteins. This information will in turn be useful to modulate enzymatic activity and produce novel sugars that are not commercially available and could find useful application for the synthesis of novel antibiotics or the design of carbohydrate-based vaccines.