A single bifunctional UDP-GlcNAc/Glc 4-epimerase supports the synthesis of three cell surface glycoconjugates in Campylobacter jejuni.

The major cell-surface carbohydrates (lipooligosaccharide, capsule, and glycoprotein N-linked heptasaccharide) of Campylobacter jejuni NCTC 11168 contain Gal and/or GalNAc residues. GalE is the sole annotated UDP-glucose 4-epimerase in this bacterium. The presence of GalNAc residues in these carbohydrates suggested that GalE might be a UDP-GlcNAc 4-epimerase. GalE was shown to epimerize UDP-Glc and UDP-GlcNAc in coupled assays with C. jejuni glycosyltransferases and in sugar nucleotide epimerization equilibria studies. Thus, GalE possesses UDP-GlcNAc 4-epimerase activity and was renamed Gne. The Km(app) values of a purified MalE-Gne fusion protein for UDP-GlcNAc and UDP-GalNAc are 1087 and 1070 microm, whereas those for UDP-Glc and UDP-Gal are 780 and 784 microm. The kcat and kcat/Km(app) values were three to four times higher for UDP-GalNAc and UDP-Gal than for UDP-GlcNAc and UDP-Glc. The comparison of the kinetic parameters of MalE-Gne to those of other characterized bacterial UDP-GlcNAc 4-epimerases indicated that Gne is a bifunctional UDP-GlcNAc/Glc 4-epimerase. The UDP sugar-binding site of Gne was modeled by using the structure of the UDP-GlcNAc 4-epimerase WbpP from Pseudomonas aeruginosa. Small differences were noted, and these may explain the bifunctional character of the C. jejuni Gne. In a gne mutant of C. jejuni, the lipooligosaccharide was shown by capillary electrophoresis-mass spectrometry to be truncated by at least five sugars. Furthermore, both the glycoprotein N-linked heptasaccharide and capsule were no longer detectable by high resolution magic angle spinning NMR. These data indicate that Gne is the enzyme providing Gal and GalNAc residues with the synthesis of all three cell-surface carbohydrates in C. jejuni NCTC 11168.


EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-The bacterial strains and plasmids used in this work are listed in Table II. E. coli strains were maintained on 2YT plates (Bio 101, Carlsbad, CA). For growth of E. coli in liquid medium (2YT), cultures were inoculated from fresh overnight cultures, grown at 37°C for 2 h, supplemented with IPTG to a final concentration of 1 mM, and grown at 20°C for an additional 24 h before harvest. C. jejuni NCTC 11168 (27) was grown on Mueller Hinton agar (Difco) at 37°C under microaerophilic conditions. For the construction of the insertional mutant of gne, E. coli DH10B (Invitrogen) was used as the host. The plasmid pPCR-Script Amp (Stratagene) was used as the cloning vector for all cloning experi-ments for the construction of insertional mutants. When appropriate, antibiotics were added to the following final concentrations: 30 g/ml kanamycin and 150 g/ml ampicillin.
Molecular Biology-The oligonucleotides SCJ-381 (5Ј-GCTGCTGG-ACATATGAAAATTCTTATTAGCGGTGGTGCAGGTTATATAG-3Ј) and SCJ-382 (5Ј-CTTAGCGTCGACTTATTAACACTGTTTTTCCCAA-TCAAAAGCAG-3Ј), used for the amplification of the gne gene from C. jejuni strain NCTC 11168, were designed on the sequence of Cj1131c (9). NdeI and SalI restriction sites were incorporated in the sequence of the oligonucleotides to facilitate the cloning of the amplicon (underlined in the sequences, initiation and termination codons are in boldface type). The gne gene was amplified by PCR from C. jejuni NCTC 11168 chromosomal DNA that had been purified using the DNeasy tissue kit (Qiagen, Mississauga, Ontario, Canada). Amplification reactions were done using the Expand High Fidelity PCR System (Roche Applied  (6) is in italics. The outer core sugars (large box) are not present when the gene encoding the UDP sugar epimerase is insertionally inactivated (see "Results"). B, the Pgl heptasaccharide (7). C, the capsule (6). In all three diagrams, the GalNAc, Gal, GalfNAc residues appear in boldface type.

TABLE I Putative epimerases present in the genome of C. jejuni NCTC 11168
This search was done in the annotated genome of C. jejuni NCTC 11168 (www.ncbi.nlm.nih.gov/cgi-bin/Entrez/altik?gi ϭ 152&db ϭ G; 7) using "epimerase" as the search term. This enzyme has been shown to possess an UDP-GlcNAc-specific C 6 dehydratase activity (10).
Science) as described by the manufacturer. Amplicons were purified using the QIAquick PCR purification kit (Qiagen), digested with NdeI and SalI, and ligated in pCWoriϩ that had been linearized with the same restriction enzymes. DNA sequencing reactions were performed as described elsewhere (30). The nucleotide sequence of the Cj1131c  gene and its product are available in GenBank TM (accession number  AL111168, see Ref. 9). Similarity and identity percentages between aa sequences were determined using Pairwise BLAST (BLAST 2 Sequences at www.ncbi.nlm.nih.gov/blast/bl2seq/bl2.html).
Coupled Enzyme Assays-The acceptor oligosaccharides (see below) were labeled with FCHASE and were prepared as described elsewhere (31). Cell lysates were prepared by sonication. Enzyme assays were performed at 37°C from 5 to 30 min.
All reactions were stopped by addition of 10 l of 50% acetonitrile, 10 mM EDTA, and 1% SDS (STOP solution) and were diluted with H 2 O to obtain 10 -15 M final concentration of the FCHASE-labeled compounds. The samples were subsequently analyzed by capillary electrophoresis using manual integration with the P/ACE Station software (Beckman Instruments, see Ref. 32). The reactions were performed in duplicate by using cells from different cultures.
Epimerization of UDP Sugars by the C. jejuni Gne and Capillary Electrophoresis Quantification-Reactions were performed at 37°C in a total volume of 30 l in 50 mM Tris-HCl, pH 8.0, with 1 mM of UDP sugar (UDP-GlcNAc, UDP-GalNAc, UDP-Glc, or UDP-Gal) and 15 l of cell lysate of PL2 ϩ pCPG6. The epimerization reaction was monitored over time by collecting 10-l aliquots after 30 min of incubation and stopped after 60 min. The reactions were stopped with STOP solution and were analyzed by capillary electrophoresis as described above. The reactions were performed in duplicate using cells from different cultures.
Purification of MalE-Gne from C. jejuni-Cells were resuspended at 10% (w/v) in 20 mM NaHEPES, pH 7.0, 200 mM NaCl, 5 mM ␤-mercaptoethanol, 1 mM EDTA (Buffer A) and were lysed by two passages through an Emulsiflex (Avestin). The cell lysate was centrifuged at 20,000 ϫ g for 30 min at 4°C. The supernatant was diluted as needed in Buffer A and applied to a 20-ml column of amylose resin (New England Biolabs). After sample application, the column was washed with 2 column volumes to elute unbound proteins. The bound MalE-Gne was eluted by washing the column with Buffer B (Buffer A with 20 mM maltose), and the eluate was collected in 2-ml fractions. The fractions containing the eluted MalE-Gne were pooled and dialyzed overnight against 100 volumes of Buffer A at 4°C. Protein quantitation was done using the BCA reagent (Pierce). The purified MalE-Gne was kept at 4°C in Buffer A without loss of activity over a period of several months.
Determination of Kinetic Parameters for MalE-Gne-Enzymatic reactions were performed in a total volume of 10 l of 50 mM Tris-HCl, pH 8.0, at 37°C. The concentrations of UDP sugar varied from 0.025 to 10 mM with 6.25 ng of purified MalE-Gne in all reactions. The reactions were stopped after 5 min of incubation with 10 l of STOP solution. The reactions were analyzed by capillary electrophoresis as described above. V max and K m(app) values were calculated by using GraphPad Prism version 3.02 for Windows (GraphPad Software, San Diego, www.graphpad.com).
Modeling of the Structure of Gne from C. jejuni-Modeling of Gne was performed by using the protein structure-modeling program MOD-ELLER 6 version 2 (33) using the structure of WbpP⅐NAD ϩ ⅐UDP-Gal-NAc from P. aeruginosa (Protein Data Bank code 1SB8) as a template.
Construction and Characterization of the C. jejuni NCTC 11168 gne Mutant-The C. jejuni NCTC 11168 insertional mutants were created and verified as described previously (6). For Cj1131c mutant (gne::Km), genes Cj1129c (pglH) to Cj1132c (wlaA) were cloned using the following primers: wlaBgalE-F1 (5Ј-GCTATTTCATCATCACAACCTACC-3Ј) and wlaBgalE-R1 (5Ј-GCCAGATGTTGAGCTTATCCG-3Ј). The kanamycin resistance cassette from pILL600 (29) was inserted into the unique BstBI restriction site of gne in a nonpolar orientation generating pPLp31. The plasmid was then electroporated into C. jejuni, and the kanamycin-resistant mutant was characterized by PCR to confirm integration by a double crossover event.
O-Deacylated LOS Analysis by CE-MS and HR-MAS NMR Spectroscopy of C. jejuni Glycans-O-Deacylated LOS for CE-MS and C. jejuni whole cells for HR-MAS NMR were prepared as described in Ref. 5. Separation conditions by CE-MS were bare-fused silica (90 cm ϫ 50 m inner diameter, 190 m outer diameter), 5% methanol in 30 mM morpholine, pH 9.0, ϩ25 kV. 1 H HR-MAS NMR spectra of NCTC11168gne mutant were acquired on a Varian INOVA 500 MHz spectrometer with a spectral width of 9600 Hz and 19 K data points. The spectrum for the NCTC11168gne mutant was acquired on a Varian INOVA 500 MHz spectrometer with a spectral width of 8000 Hz and 16 K data points. The sample spinning rate was 3000 Ϯ 10 Hz, and all other acquisition and data processing parameters were as described previously (5).

Sequence Analysis of GalE from C. jejuni Strain NCTC
11168 -The Cj1131c gene was amplified by PCR from purified C. jejuni NCTC 11168 genomic DNA and cloned in the expression vector pCWoriϩ (28) to generate the plasmid pCPG6 (Table II). The nucleotide sequence of the cloned Cj1131c was found to be identical to that of Cj1131c (see Ref. 9; GenBank TM accession number AL111168). The C. jejuni NCTC 11168 GalE shares 98% identity with GalE from C. jejuni 81116 (six differences over 328 positions, none of which occur within the two conserved motifs nor among the residue parts of the predicted active site; Fig. 2 and see below). GalE from C. jejuni NCTC 11168 also shares significant aa sequence similarities with other bacterial UDP-GlcNAc epimerases (27-40% identity and 42-59% similarity; Fig. 2) as well as with GalE from E. coli (36% identity and 54% similarity; Fig. 2).
The C. jejuni NCTC 11168 GalE contains the two motifs characteristic of members of the so-called short chain dehydrogenase/reductase superfamily. The first motif is GXXGXXG (Gly 7 -Xaa-Xaa-Gly 10 -Xaa-Xaa-Gly 13 in the C. jejuni NCTC 11168 GalE, Fig. 2), which is located near the amino-terminal end of the enzyme and the cofactor-binding pocket. This motif helps to stabilize the FAD or NAD(P)-binding Rossman fold and is involved in the binding of the nucleotide cofactor to the domain (40). The second motif is Ser-Xaa 24 -Tyr-Xaa 3 -Lys (Ser 121 -Xaa 24 -Tyr 146 -Xaa 3 -Lys 150 in the C. jejuni NCTC 11168 GalE, Fig. 2), in which the seryl, tyrosinyl, and lysyl residues are directly involved in catalysis (14,41).
Production of GalE from C. jejuni in E. coli and Functional Characterization-GalE from C. jejuni was overproduced in E. coli PL2 from plasmid pCPG6 (Table II) upon induction with IPTG. A protein of the expected molecular weight (37 kDa) was visible in the cell lysate (data not shown). The galE28 mutation abrogates UDP-glucose 4-epimerase activity in E. coli PL2 so that any epimerization of UDP-Glc into UDP-Gal will be attributable to the pCPG6-encoded GalE. In addition, GalE from E. coli is unable to epimerize UDP-GlcNAc or UDP-GalNAc (16). Thus, any epimerization observed in the assays will be attributed solely to GalE from C. jejuni.
The enzymatic function of GalE from C. jejuni was investigated by performing coupled enzyme assays with glycosyltransferases from C. jejuni. The rationale for these assays was that the expected product of the reaction would be synthesized by the glycosyltransferase only if the appropriate donor sugar was 2 For simplicity's sake, the acceptor sugars used in CgtA and CgtB enzyme assays throughout this report were named after their corresponding gangliosides (even if they only consist in their glycone moiety). (Fig. 3). By supplying UDP-Glc or UDP-GlcNAc and the C. jejuni GalE to the reaction, the product will be synthesized only if the UDP sugar is first converted into its C4epimer. In the case of assays using UDP-GlcNAc, the synthesis of the expected reaction product would indicate that the C. jejuni GalE also possesses UDP-GlcNAc 4-epimerase activity.
Coupled assays using the C. jejuni GalE and CgtB were performed to verify the ability of GalE to epimerize UDP-Glc into UDP-Gal. The CgtB glycosyltransferase from C. jejuni is a   (37). This alignment was made using ClustalX (38) and was reformatted in GeneDoc (39). Identical aa residues in all eight sequences are shaded in black; residues conserved in six or seven sequences are shaded in dark gray (white letters); and residues conserved in five sequences are shaded in light gray (black letters). The boxed aa residues numbered 1-6 are those of the predicted substrate-binding pocket of Gne from C. jejuni NCTC 11168 and GalE from C. jejuni 81116 based on the structure of WbpP⅐NAD ϩ ⅐UDP-GalNAc (see "Modeling of the Active Site of Gne from C. jejuni NCTC 11168" under the "Results"). ␤1,3-galactosyltransferase that adds a galactosyl residue via a ␤-1,3 linkage to the GalNAc moiety at the nonreducing end of the growing LOS chain (42). CgtB was shown to be able to use UDP-Gal to convert GM2-FCHASE into GM1a-FCHASE but was unable to use UDP-Glc as the donor sugar for the same reaction (Fig. 4A, traces A and B). If UDP-Glc were used in a GalE-CgtB coupled assay, the expected GM1a-FCHASE product was formed (Fig. 4A, trace C). These data indicated that GalE from C. jejuni NCTC 11168 epimerized UDP-Glc into UDP-Gal, as does GalE from C. jejuni 81116 (12).
Another series of coupled assays using GalE and CgtA was done to verify if GalE could epimerize UDP-GlcNAc into UDP-GalNAc. CgtA from C. jejuni is a ␤1,4-N-acetylgalactosaminyltransferase that adds a N-acetylgalactosaminyl residue via a ␤-1,4 linkage to the Gal moiety at the nonreducing end of the growing LOS chain (42). CgtA uses UDP-GalNAc as the donor sugar to convert GM3-FCHASE into GM2-FCHASE but is unable to use UDP-GlcNAc for the same reaction (Fig. 4B, traces  A and B). When UDP-GlcNAc was used in a GalE-CgtA-coupled assay, the expected GM2-FCHASE product was formed (Fig.  4B, trace C). This result indicated that GalE could epimerize UDP-GlcNAc and UDP-GalNAc in addition to being able to epimerize UDP-Glc and UDP-Gal. This epimerization of UDP-GlcNAc and UDP-GalNAc by the C. jejuni GalE has not been demonstrated before. Because GalE from C. jejuni possesses UDP-N-acetylglucosaminyl 4-epimerase activity, it will henceforth be designated Gne (UDP-GlcNAc 4-epimerase), after the enzymes with the same function in B. subtilis, E. coli, and Y. enterocolitica (20,37,43).
UDP Sugar Epimerization Equilibrium Assays-UDP sugar epimerization reactions were performed in lysates of PL2 ϩ pCPG6 so that UDP sugar conversion will be attributable only to Gne from C. jejuni NCTC 11168. The reactions supplied with UDP-GlcNAc and UDP-GalNAc contained their corresponding 4-epimer after the 60-min incubation period (29% conversion of UDP-GlcNAc and 71% conversion of UDP-GalNAc; data not shown). Epimerization of UDP-Glc and UDP-Gal was also observed after the 60-min incubation period (24% conversion of UDP-Glc and 76% conversion of UDP-Gal; data not shown).
Purification of MalE-Gne from C. jejuni NCTC 11168 -A MalE-Gne fusion protein was successfully overproduced in E. coli AD202 from plasmid pCPG13 (Table II) upon induction with IPTG. A protein of the expected molecular weight (76 kDa) is visible in the cell lysate (data not shown).
K m(app) for UDP sugars-The K m(app) values for UDP-Glc-NAc, UDP-GlcNAc, UDP-Gal, and UDP-Glc were determined for MalE-Gne (Table III). The K m(app) values for UDP-GlcNAc and UDP-GalNAc are 1.09 and 1.07 mM, respectively, whereas the K m(app) values for UDP-Glc and UDP-Gal are both 0.78 mM. The calculated k cat and k cat / K m(app) values for UDP-GalNAc and UDP-Gal are approximately three to four times higher than those for UDP-GlcNAc and UDP-Glc. These data suggest that MalE-Gne is more efficient for the epimerization of UDP-GalNAc and UDP-Gal.
Modeling of the Active Site of Gne from C. jejuni NCTC 11168 -The structure of WbpP⅐NAD ϩ ⅐UDP-GalNAc from P. aeruginosa (22) was used to build a structural model of Gne from C. jejuni. In the predicted structure of Gne (Fig. 5B), the GalNAc moiety of the UDP-GalNAc is in the catalytically appropriate substrate orientation and is in the proper position relatively to the catalytic base (Tyr 146 in Gne) and the NAD ϩ cofactor. The predicted distances between the hydroxyl group of Tyr 146 and the C-4 hydroxyl group of UDP-GalNAc and that between the C-4 of NAD ϩ and C-4 hydroxyl group of UDP-GalNAc are essentially identical to those measured between the corresponding atoms in the WbpP⅐NAD ϩ ⅐UDP-GalNAc structure (data not shown).
Based on the active site model (22), the six amino acid residues that make up the UDP sugar-binding pocket of Gne are Ile 82 , Thr 122 , Tyr 146 , Gln 176 , Leu 195 , and Leu 294 ( Fig. 5B; residues numbered 1-6 on Fig. 2). The three glycyl residue parts of the GXXGXXG motif that binds NAD ϩ are similarly positioned in Gne as they are in WbpP (data not shown). The predicted structure of Gne suggests that the six active site residues (yellow in Fig. 5) and NAD ϩ (blue in Fig. 5) also constitute a saccharide-binding pocket (gray in Fig. 5) sufficiently large enough to accommodate either UDP-GalNAc (solid green in Fig. 5) or UDP-GlcNAc (translucent green in Fig.  5) in addition to UDP-Gal and UDP-Glc.

Examination of the O-Deacylated LOS by CE-MS-If
Gne is the sole enzyme capable of epimerizing UDP-GlcNAc/Glc into UDP-GalNAc/Gal in C. jejuni NCTC 11168, its inactivation should affect several biosynthetic processes, most notably those of complex carbohydrates. The inactivation of Gne should therefore cause a truncation of the LOS.
The LOS was extracted from C. jejuni NCTC 11168 and NCTC11168gne and was analyzed by CE-MS as described (5). The extracted mass spectra for prominent O-deacylated LOS are presented in Fig. 6. In NCTC 11168 (Fig. 6A), the predominant ions are observed at m/z 888 and 1184, which correspond to quadruply and triply charged molecules with a molecular mass of 3555 Da, a structure in which the terminal ␤-1,3-linked galactose is not present (Fig. 1A). In the gne mutant (Fig. 6B), the predominant ions are observed at m/z 611.8, 815.8, and 1224.8, respectively, corresponding to quadruply, triply, doubly FIG. 3. Coupled assays to functionally characterize the Cj1131c gene product from C. jejuni NCTC 11168. A, GM2-FCHASE is converted into the GM1a-FCHASE by CgtB (␤1,3-galactosyltransferase) if UDP-Gal is present. There is no reaction if UDP-Glc is supplied as the donor sugar. When GalE or a bifunctional Gne enzyme is present, UDP-Glc will be epimerized into UDP-Gal, which is then used by CgtB to synthesize GM1a-FCHASE. B, GM3-FCHASE is converted into GM2-FCHASE by CgtA (␤1,4-N-acetylgalactosaminyltransferase) if UDP-GalNAc is present. There is no reaction if UDP-GlcNAc is supplied as the donor sugar. When a Gne enzyme is present, UDP-GlcNAc will be epimerized into UDP-GalNAc, which is then be used by CgtA to synthesize GM2-FCHASE. charged ions with a molecular mass of 2450 Da. Based on the previous studies (6) and with the evidence of NMR data, the glycolipids having molecular masses of 2450 were assigned to a truncated LOS molecule that contains 2 3-deoxy-D-mannooctulosonic acid, 2 heptoses, 2 hexoses, and 1 PEtn. Because Gne from C. jejuni has been shown to convert Glc to Gal, the two hexoses present on the truncated LOS must be the Glc residues linked to the heptoses (Fig. 1A). These data indicate that Gne is the sole supplier of the UDP-Gal and UDP-GalNAc required for the synthesis of the LOS.
Detection of the Capsule in Whole NCTC11168gne Cells by HR-MAS NMR-Because the disruption of gne expression affects the LOS, the effect of its disruption on the capsule that contains GalfNAc (Fig. 1A) was also investigated. The HR-MAS NMR spectrum of wild-type NCTC 11168 exhibits anomeric 1 H resonances that have been assigned previously (5) to the capsule as shown in Fig. 7C. In the HR-MAS spectrum of NCTC11168gne cells (Fig. 7E)  a One unit of activity is defined as the conversion of 1 mol of UDP sugar into its 4-epimer in 1 min at 37°C/mg of enzyme. Assays done for Gne (this work), WbpP (18), and WbgU (17)  Comparison of NCTC11168kpsM (Fig. 7D) with the spectrum of NCTC11168gne (Fig. 7E) reveals that the resonances of the five GalNAc and of the Glc residues are missing (Fig. 7E). The signal of the bacillosamine residue is too broad to be clearly seen on whole cells. It is barely detectable in the kpsM mutant (Fig. 7D), and it cannot be said for certain that it is missing in the gne mutant (Fig. 7E). Nonetheless, the absence of the GalNAc residues in the Pgl glycan in the gne mutant indicates that, as expected, Gne supplies the UDP-GalNAc required for the synthesis of the Pgl glycan.

DISCUSSION
The gne (Cj1131c) gene of C. jejuni NCTC 11168 was initially annotated to encode a UDP-glucose 4-epimerase (GalE) based on functional characterization of its product (8,9). New evidence presented in this report conclusively demonstrates that gne encodes a bifunctional UDP-GlcNAc/Glc 4-epimerase. Gne was then shown to be involved in the synthesis of three glycoconjugates in C. jejuni: the capsule, the Pgl heptasaccharide, and the LOS.
Because the assembly of the Pgl glycan will be blocked after the addition of the starting bacillosamine residue in a gne mutant of C. jejuni, the data argue in favor of the glycoprotein N-linked heptasaccharide being completely synthesized before being flipped by the putative ABC transporter WlaB (45). However, the NMR signal of the bacillosamine is broad and difficult to see in whole-cell preparations. Because of this, it cannot concluded that the bacillosamine is not present in the gne mutant. Because we favor the mechanism of block transfer, we predict that the bacillosamine residue is not present in the gne mutant. Capsule synthesis should also be affected in a gne mutant because there will be no UDP-GalNAc available for conversion from the pyranose to the furanose conformation by the putative UDP-N-acetylgalactosaminylpyranose mutase (Glf, encoded by Cj1439c, see Ref. 9). This will thus block synthesis of the capsule. The insertional inactivation of Cj1439c in C. jejuni also results in an acapsular mutant (6). This absence of capsule in insertional mutants of glf and gne also argues in favor of block transfer in the case of the capsular polysaccharide.
The consequences of inactivating galE have been investigated in other bacteria. The inactivation of galE affects pilin glycosylation in Neisseria meningitidis C311 number 3. Meningococcal pilin are normally post-translationally modified at Ser 63 by the addition of the Gal␤1,4-Gal␣1,3-Bac trisaccharide (46 -48). In a galE mutant, the pilin were glycosylated with a lone Bac residue (47,48). The pilin of Neisseria gonorrhoeae are O-glycosylated with the disaccharide Gal␣ 1,3-GlcNAc␤-at Ser 63 (49). It is then likely that the inactivation of the gonococcal galE will cause the loss of the terminal Gal residue from the pilin disaccharide.
The O-deacylated LOS core was truncated in C. jejuni NCTC11168gne compared with its wild-type parent (Fig. 6, A  and B). Fry et al. (12) reported that the lipid A-core LOS from a galE mutant of C. jejuni was smaller than that of the parental strain, which is in perfect agreement with the data presented in this report. The GalE enzyme of C. jejuni 81116 identified and characterized by Fry et al. (11,12) is most likely a Gne enzyme, considering it is almost identical to Gne from C. jejuni NCTC 11168. In addition, the amino acid residues of the C. jejuni 81116 GalE predicted to constitute the UDP sugarbinding site are the same as those of the C. jejuni NCTC 11168 Gne (aa residues numbered 1-6 in Fig. 2). In addition to its effect on pilin glycosylation in N. meningitidis C311 number 3, the inactivation of galE caused a truncation of the LOS (46,47). The insertional inactivation of galE in N. meningitidis MC58 and N. gonorrhoeae MS11 had a similar effect on the LOS (50,51). The LPS was truncated in galE mutants of Haemophilus influenzae (52,53) and Helicobacter pylori (54) as well as in a gne (lse) mutant of Y. enterocolitica of serotype O:8 (19). The O-antigen was lost in a gne mutant of E. coli strains of serotypes O55 and O157 (36).
Based on the classification of UDP-hex/hexNAc 4-epimerases (22), Gne from C. jejuni belongs to group 2. Other UDP-hexoses 4-epimerases may be assigned to a given group even though they have not been extensively characterized. For instance, Gne (Lse) is required to generate a full-length LPS capped with O-antigen in Y. enterocolitica of serotype O:8 (19,20). This enzyme probably belongs to group 3 because it possesses a FIG. 5. Saccharide-binding pocket of UDP-hexose epimerases. A, in the WbpP⅐NAD ϩ ⅐UDP-GalNAc ternary complex (Protein Data Bank code 1SB8), the saccharide-moiety of UDP-GalNAc (green) is situated in the saccharide-binding pocket (gray), which mainly consists of six active site residues (cyan) and NAD ϩ (blue). B, a predicted structure of Gne from C. jejuni NCTC 11168 with UDP-GalNAc/GlcNAc (green and translucent green) was modeled based on the structure of Protein Data Bank code 1SB8 and the human GalE⅐NADH⅐UDP-GlcNAc ternary complex (Protein Data Bank code 1HZJ). The saccharide-binding pocket (gray), the six active site residues (cyan), and NAD ϩ (blue) are also shown. The figure was prepared by using the PyMOL Molecular Graphics System (DeLano Scientific, San Carlos, CA) (44).
UDP-GlcNAc/GalNAc 4-epimerase activity but displays weak UDP-Glc/Gal 4-epimerase activities (19,20 (50). Its LOS is of immunotype L3 (58), which contains two Gal residues but no GalNAc (59). This absence of GalNAc in its LOS and in its pilin O-linked trisaccharide (46 -48) suggests GalE from N. meningitidis MC58 does not need to epimerize UDP-GlcNAc and would therefore belong to group 1. Bacteria with larger genomes such as Y.  67), and several genes are annotated as epimerases so it may contain a Gne-like enzyme if its GalE is not a bifunctional UDP-Glc/GlcNAc 4-epimerase. P. aeruginosa of serotype O6 has not undergone extensive genomic studies, but it contains the monofunctional UDP-Glc-NAc 4-epimerase WbpP (18). It therefore cannot be excluded that it also contains a monofunctional GalE as well.
In their analysis of substrate preference in UDP-hexose 4-epimerases, Ishiyama and co-workers (22) also proposed the structural features that dictate substrate preference. When modeling the three-dimensional structure of the C. jejuni Gne,