Biochemical Characterization of UDP-Gal:GlcNAc-Pyrophosphate-Lipid β-1,4-Galactosyltransferase WfeD, a New Enzyme from Shigella boydii Type 14 That Catalyzes the Second Step in O-Antigen Repeating-Unit Synthesis

Abstract

The O antigen is the outer part of the lipopolysaccharide (LPS) in the outer membrane of Gram-negative bacteria and contains many repeats of an oligosaccharide unit. It contributes to antigenic variability and is essential to the full function and virulence of bacteria. Shigella is a Gram-negative human pathogen that causes diarrhea in humans. The O antigen of Shigella boydii type 14 consists of repeating oligosaccharide units with the structure [→6-d-Galpα1→4-d-GlcpAβ1→6-d-Galpβ1→4-d-Galpβ1→4-d-GlcpNAcβ1→]n. The wfeD gene in the O-antigen gene cluster of Shigella boydii type 14 was proposed to encode a galactosyltransferase (GalT) involved in O-antigen synthesis. We confirmed here that the wfeD gene product is a β4-GalT that synthesizes the Galβ1-4GlcNAcα-R linkage. WfeD was expressed in Escherichia coli, and the activity was characterized by using UDP-[³H]Gal as the donor substrate as well as the synthetic acceptor substrate GlcNAcα-pyrophosphate-(CH₂)₁₁-O-phenyl. The enzyme product was analyzed by liquid chromatography-mass spectrometry (LC-MS), high-performance liquid chromatography (HPLC), nuclear magnetic resonance (NMR), and galactosidase digestion. The enzyme was shown to be specific for the UDP-Gal donor substrate and required pyrophosphate in the acceptor substrate. Divalent metal ions such as Mn²⁺, Ni²⁺, and, surprisingly, also Pb²⁺ enhanced the enzyme activity. Mutational analysis showed that the Glu101 residue within a DxD motif is essential for activity, possibly by forming the catalytic nucleophile. The Lys211 residue was also shown to be required for activity and may be involved in the binding of the negatively charged acceptor substrate. Our study revealed that the β4-GalT WfeD is a novel enzyme that has virtually no sequence similarity to mammalian β4-GalT, although it catalyzes a similar reaction.

Lipopolysaccharides (LPSs) consist of O-polysaccharide (O-antigenic) side chains covalently linked to a core polysaccharide and lipid A. LPSs are found in the outer membranes of Gram-negative bacteria, where they contribute to the structural integrity of the membrane and interact with the external environment (9, 10, 15). In the complex and dynamic microbial ecosystem of the human intestine, the communication between microorganisms and the gastrointestinal (GI) epithelium involves O-antigen and LPS binding molecules. Thus, the elimination of the O antigen may reduce virulence (2, 16, 21). Shigella is a genus of highly adapted bacterial pathogens that cause gastrointestinal disease, such as bacillary dysentery or shigellosis. A recent survey showed that shigellosis causes approximately 165 million cases of severe dysentery and more than 1 million deaths per year, mostly in children from developing countries (10). Shigella strains are categorized into four groups: S. boydii, S. dysenteriae, S. flexneri, and S. sonnei, each containing multiple subgroups of different serotypes, based on structural variations in their O antigens.

O antigens consist of repeating units of oligosaccharides that are assembled individually, followed by the polymerization of units to form O antigens of different lengths. The glycosyltransferases involved in the biosynthesis of O antigens play a critical role in determining O-antigen structural diversity. The pentasaccharide repeating unit of S. boydii type 14 (B14) has the structure [→6-d-Galpα1→4-d-GlcpAβ1→6-d-Galpβ1→4-d-Galpβ1→4-d-GlcpNAcβ1→]n (12), suggesting the existence of five specific glycosyltransferases: a GlcNAc-phosphotransferase (WecA), three Gal-transferases, and a glucuronosyltransferase.

Three distinct processes for the synthesis and translocation of O antigens have been described: the Wzx/Wzy-dependent pathway, the ATP binding cassette transporter-dependent process, and the synthase-dependent process (20, 25, 26). The biosynthesis of the S. boydii B14 O antigen that contains a variety of different sugar residues is expected to utilize the Wzy/Wzx-dependent pathway, where the synthesis of the repeating unit is initiated by WecA, catalyzing the transfer of sugar-phosphate (GlcNAcα-phosphate) from nucleotide sugar (UDP-GlcNAc) to a lipid carrier, undecaprenol-phosphate (Und-P), at the cytoplasmic side of the inner membrane. The wecA gene is present in the S. boydii B14 genome but outside the O-antigen gene cluster (1). The wecA gene is also involved in the synthesis of bacterial polysaccharides other than the O antigen. The extension of the chain is then mediated by specific glycosyltransferases that utilize nucleotide sugar donor substrates and are thought to be loosely associated with the inner membrane. In contrast, mammalian glycosyltransferases are usually membrane-bound proteins. Bacterial and mammalian glycosyltransferases, although they may have similar substrate specificities and form the same linkage, show significantly different amino acid sequences (4). Completed repeating units are then flipped across the membrane to the periplasmic side (by the flippase Wzx) and are polymerized (by Wzy) to form the O antigen under the control of a chain length regulator (Wzz). The repeating units are initially linked to the lipid carrier through GlcNAcα-phosphate. However, the S. boydii B14 O antigen has GlcNAc in the β linkage; thus, upon the polymerization of the completed repeating units, the linkage may be inverted, probably through the specific action of the polymerase Wzy. The entire polymer is then ligated to an outer core sugar based on lipid A. Upon completion, the LPS is extruded from the inner membrane and translocated to the outer membrane (19, 26). The latter-acting enzymes have multiple transmembrane regions that integrate them into the bacterial membranes.

Genes involved in O-antigen biosynthesis are normally clustered between galF and gnd in Escherichia coli and Shigella and are classified into three different groups: (i) nucleotide sugar synthesis genes involved in the synthesis of donor substrates, (ii) glycosyltransferase genes, and (iii) O-antigen-processing genes, such as the flippase gene wzx and the polymerase gene wzy. The O-antigen gene cluster of B14 has been sequenced and analyzed (10). Four putative glycosyltransferase genes found in the B14 O-antigen synthesis gene cluster are wfeA, wfeB, wfeD, and wfeE. WfeD shares 38% identity and 57% similarity to the putative glycosyltransferase Orf9, which is involved in the synthesis of the E. coli O136 O antigen (our unpublished data). Since the O antigens of B14 and E. coli O136 share only one common linkage, d-Galpβ1→4-d-GlcpNAc (12, 23), wfeD was proposed to encode the galactosyltransferase (GalT) that transfers Gal to GlcNAcα-PP-Und in the β1-4 linkage, which is the second step in the biosynthetic pathway of the B14 O-antigen repeating unit.

We have used biochemical approaches to assay the WfeD enzyme activity and to characterize this enzyme. The lipid carrier analog GlcNAcα-PO₃-PO₃-(CH₂)₁₁-O-phenyl [GlcNAc-PP-(CH₂)₁₁-OPh] has previously been used as a defined synthetic acceptor substrate for the characterization of glycosyltransferases from E. coli serotypes O7 (β1,3-GalT WbbD), O56 (β1,3-Glc-transferase WfaP), and O152 (β1,3-Glc-transferase WfgD) (6, 17). In this work, we showed that GlcNAc-PP-(CH₂)₁₁-OPh could also serve as an exogenous substrate for WfeD from B14. We were therefore able to prove that wfeD encodes a novel β1,4-GalT.

MATERIALS AND METHODS

Materials.

All reagents were purchased from Sigma unless indicated otherwise. Radioactive nucleotide sugars were purchased from Perkin-Elmer. Acceptor substrates were synthesized as reported previously (5, 7, 13, 18). GlcNAcα-PO₃-PO₃-(CH₂)₉-CH₃ was kindly provided by O. Hindsgaul, Carlsberg Laboratories, Copenhagen, Denmark.

Bacterial strains, plasmids, and expression of the wfeD gene.

The putative glycosyltransferase gene wfeD from B14 was amplified by PCR and cloned into expression vector pET28a (containing a cleavable His-tag-encoded sequence) to construct plasmid pLW1298. pLW1298 was then transformed into E. coli BL21 cells. Bacteria were grown in 6 ml LB broth containing 50 μg/ml kanamycin overnight at 37°C with constant shaking (17). To induce the plasmid-derived enzyme, the bacterial suspension was transferred into 125 ml LB broth containing kanamycin and incubated for 90 min at 37°C with constant shaking. When the suspension reached an absorbance at 600 nm of 0.8, IPTG (isopropyl-beta-thiogalactoside) was added to a final concentration of 1 mM to induce protein expression. Cells were grown for an additional 4 h at 30°C and were then harvested by centrifugation for 10 min at 3,145 × g (4,000 rpm) (IEC 21000R). Pellets were washed with 5 ml phosphate-buffered saline (PBS) and resuspended in 10 ml PBS containing 10% glycerol. Aliquots of bacteria were stored at −20°C for enzyme assays.

Construction of mutants.

Four primers (F1, R1, FM, and RM) and three PCRs were used to construct each site-specific mutant by overlap extension. The first PCR with primers F1 and RM was used to amplify DNA that contains the mutation site together with upstream sequences. The second PCR with primers FM and R1 was used to amplify DNA that contains the mutation site together with downstream sequences. The mutation of interest was located in the overlap region of two amplified fragments. The overlapping fragments were mixed, denatured, and annealed to generate heteroduplexes that can be extended and then amplified in the third PCR with primers F1 and R1. The primers used for the construction of mutants are listed in Table 1. For each PCR, a total of 25 cycles was performed under the following conditions: denaturation at 95°C for 30 s, annealing at 50°C for 30 s, and extension at 72°C for 1 min. Subsequent sequencing confirmed the correct sites of mutation.

TABLE 1.

Primers used for mutagenesis

Primer	Sequencea	Mutation site
F1	5′-CGCGGATCCGTGATCGATAATCTCATAAA-3′
R1	5′-CCGCTCGAGTCAATTACTTCTCGTGAATATTT-3′
FM1	5′-TTAATCGGTTCTAAGGCTTATGA-3′	D99A
RM1	5′-TCATAAGCCTTAGAACCGATTAA-3′
FM2	5′-TCGGTTCTAAGGATTATGCTAT-3′	E101A
RM2	5′-ATAGCATAATCCTTAGAACCGA-3′
FM3	5′-AGGATTATGAAATAGCGCAAGC-3′	E103A
RM3	5′-GCTTGCGCTATTTCATAATCCT-3′
FM4	5′-ATATAAACATGTTGCGAAAAG-3′	R210A
RM4	5′-CTTTTCGCAACATGTTTATAT-3′
FM5	5′-TAGGAAAGCTTTTTTTATCGAT-3′	R212A
RM5	5′-ATCGATAAAAAAAGCTTTCCTA-3′
FM6	5′-CATGTTAGGGCTAGATTTT-3′	K211A
RM6	5′-AAAATCTAGCCCTAACATG-3′

^aThe underlined sequences indicate restriction sites. The nucleic acids being mutated for generating the single-amino-acid mutations are indicated by boldface type.

Galactosyltransferase assays.

Bacterial homogenates containing WfeD were prepared by sonication in 0.05 M sucrose buffer or other buffers as described previously (13, 17). The standard assay mixtures for WfeD contained (in a total volume of 40 μl) acceptor substrate as indicated in Table 2, bacterial homogenates (0.5 to 0.6 μg protein), 0.075 M MES (2-morpholinoethanesulfonic acid) buffer (pH 7), 5 mM MnCl₂, and 1 mM UDP-[³H]Gal (2,000 to 3,500 cpm/nmol). In standard GalT assays, 0.25 mM GlcNAc-PP-(CH₂)₁₁-OPh (13) was used as the acceptor substrate, while the control assays lacked an acceptor substrate. Due to the low solubility of GlcNAc-PP-(CH₂)₁₁-OPh in water, 10% methanol (MeOH) was present in all assay mixtures. At least two determinations were carried out, with variation between assays of <10%. No significant inhibition of GalT activity was noted at up to 20% MeOH or up to 7% isopropanol in the assay mixtures. Mixtures were incubated for 10 min at 37°C. Reactions were quenched by the addition of 200 μl ice-cold water to the mixtures. For product isolation, 500 μl water was added, the mixture was passed through a short C₁₈ Sep-Pak column, and the column was washed with 4 ml water. Radioactive product was then eluted with 4 ml MeOH. Scintillation fluid (4.5 ml) (Ready Safe; Beckman Coulter) was added to aliquots of each 1-ml fraction, and radioactivity was determined by scintillation counting (LS6500 scintillation counter; Beckman Coulter). The radioactive product was concentrated in the first 3 ml of MeOH eluates for high-performance liquid chromatography (HPLC) analysis. Residues were taken up in MeOH. The assays for bovine β4-GalT1 were described previously (5). Kinetic parameters were determined by using the ENZFIT program.

TABLE 2.

Activities of purified WfeD with GlcNAc derivatives tested as substrates and inhibitors^a

Compound	Activity (%)
GlcNAc α-PO3-PO3-(CH2)11-OPh	100
GlcNAc α-PO3-PO3-(CH2)6-OPh	39
GlcNAc α-PO3-PO3-(CH2)16-OPh	92
GlcNAc α-PO3-(CH2)11-OPh	11
GlcNAc α-PO3-PO3-(CH2)9-CH3	74
GlcNAc α-(CH2)11-OPh	<5
GlcNAc β-(CH2)11-OPh	<5
GlcNAc α-CO-CH2-P-(CH2)11-OPh	<5
GlcNAc β-(3-isoquinolinyl)	<1
GlcNBu β-(2-naphthyl)	<1
GlcNAc β-(2-naphthyl)	<1
GlcNBu β-S-(2-naphthyl)	<1
GlcNAc β-Bn	<1
GlcNAc α-Bn	<1
GlcNAc	<1
GlcNAc β-S-Ph	<1
4-Deoxy-GlcNAc α-Bn	<1
4-F-4-Deoxy-GlcNAc α-Bn	<1
4-Deoxy-GlcNAc	<1
6-Deoxy-GlcNAc	<1
6-S-GlcNBu β-(2-naphthyl)	<1
GlcNBu β-NH-(2-naphthyl)	<1

^aGalT assays were carried out as described in Materials and Methods, using the compounds as acceptor substrates at a 0.2 mM concentration in the assay mixture.

Product identification using mass spectrometry and HPLC.

The enzyme product was prepared as a radioactive product and also as a nonradioactive product for mass spectrometry (MS) and nuclear magnetic resonance (NMR) analysis. The separation of substrates and enzyme products was achieved by HPLC using a C₁₈ column and acetonitrile-water mixtures at a flow rate of 1 ml/min. For example, the substrate GlcNAc-PP-(CH₂)₁₁-OPh and product Gal-GlcNAc-PP-(CH₂)₁₁-OPh were separated with 24% acetonitrile in water as the mobile phase. Compounds were analyzed by electrospray ionization (ESI)-MS in the negative-ion mode, as described previously (17).

Product linkage confirmation using galactosidases and NMR.

The anomeric configuration of the linkage formed in the radioactive product was determined by the incubation of aliquots with specific galactosidases: green coffee bean α-galactosidase (0.065 U/μl), jack bean (β4-specific) β-galactosidase (0.06 U/μl), E. coli (β1-4-specific) β-galactosidase, and bovine testicular (β1-3-, β1-4-, and β1-6-specific) β-galactosidase (0.095 U/μl). Aliquots of radioactive enzyme product (800 cpm) were treated in a total volume of 100 μl with 25 μl MacIlvaine buffer (0.1 M citric acid-0.2 M Na-phosphate) (pH 4.3), 10 μl of 0.1% bovine serum albumin, and galactosidase. Mixtures were incubated for 30 min at 37°C, diluted with 800 μl of water, and applied onto 0.4-ml AG1x8 columns. The released radioactive enzyme ([³H]Gal) was eluted with 2.8 ml water, while the unreacted enzyme product stayed in the AG1x8 column.

To prepare large amounts of enzyme product for NMR, GalT assays were carried out as follows. The incubation mixtures (a total of 8 ml) contained 10 ml bacterial homogenate in 50 mM sucrose (E. coli BL21 cells complemented with plasmid pLW1298), 2.5 μmol GlcNAc-PP-(CH₂)₁₁-OPh, 10 μmol UDP-Gal, 1 mmol MES buffer (pH 7), and 50 μmol MnCl₂. After incubation for 30 min at 37°C, 8 ml of cold water was added, and the mixtures were applied onto C₁₈ Sep-Pak columns. Each column was washed with 4 ml water, and the product was eluted with 4 ml MeOH. The MeOH fractions were pooled, flash-evaporated, and redissolved in 500 μl MeOH. Aliquots of the MeOH-dissolved product were purified by HPLC, using a C₁₈ column and acetonitrile-water (24:76) as the mobile phase. The enzyme product was dried, exchanged three times with 99.96% D₂O, and analyzed by 600-MHz proton and ¹³C NMR spectroscopy in CD₃OD, as described previously (6).

Enzyme purification and analysis of enzyme protein.

For the purification of the fusion protein His₆-WfeD, bacteria (0.2 mg bacteria per ml buffer) were sonicated in sonication buffer (50 mM Tris base [pH 8.0], 500 mM NaCl, 10 mM imidazole, and 1 mM EGTA) for six cycles of 1-min sonication and 1-min waiting time (Misonix Sonicator 3000, program 1). The homogenate was then centrifuged at 3.9 × 10⁴ × g for 30 min. The His-tagged fusion protein in the supernatant was purified by affinity chromatography with a Ni²⁺-nitrilotriacetic acid (NTA) Sepharose Fast Flow column (Bio-Rad). Bound proteins were eluted with a gradient of 200 to 400 mM imidazole buffer containing 50 mM Tris base (pH 8.0), 500 mM NaCl, and 1 mM EGTA. The fusion protein was analyzed by SDS-PAGE, Western blots, and GalT activity assays. Gel slices were analyzed by matrix-assisted laser desorption ionization (MALDI) peptide mass fingerprinting at the Protein Function Discovery Facility at Queen's University, using MALDI-time of flight (TOF) mass spectrometry.

Western blot analysis was performed in duplicates with rabbit antibody against the His tag as the primary antibody (kindly provided by P. Davies, Queen's University) and horseradish peroxidase (HRP)-conjugated anti-rabbit IgG as the secondary antibody (from Promega). Ni²⁺-NTA-purified and diluted cell lysates of the wild type and mutants (3 to 5 μg protein in 20 μl) were diluted 5:1 in 4 μl SDS loading buffer consisting of 1 mM Tris-HCl (pH 6.8), 1% SDS, 10% glycerol, and 1% β-mercaptoethanol and run on SDS-PAGE gels (8% gel). Proteins were electrophoretically transferred onto nitrocellulose and then probed sequentially with anti-His antibody (1:5,000) and anti-rabbit IgG (1:10,000). Labeling was visualized with 10 ml of a 1:1 mixture of luminol reagent and peroxide solution (Millipore). Prestained protein standards (Fermentas) were used to calibrate the gels. The densitometric analysis of the relative band intensity of the Western blot film was performed by using ImageJ1.43 software (9), which measured the optical density of the protein bands and indicated the approximate relative amounts of His-tagged protein in wild-type and mutant WfeD. The enzyme activities were normalized according to relative amounts of protein detected on Western blots.

Effect of amino-acid-specific reagents.

To evaluate the role of amino acids in activity, the crude and purified enzyme preparations were incubated with amino-acid-specific reagents prior to the assays for 10 min at room temperature in reaction mixtures lacking UDP-Gal. The reaction was initiated by the addition of UDP-Gal to the mixtures. Diethyl pyrocarbonate (DEPC) (reacts with His), iodoacetic acid (IAA) (reacts with Cys), p-hydroxyphenylglyoxal (HPG) (reacts with Arg), and 5,5′-dithio-bis-2-nitrobenzoic acid (DTNB) (reacts with Cys) were used at a 0.2 mM concentration in the assay. The disulfide-bond-reducing agent dithiothreitol (DTT) was used at 0.2 to 3 mM in the assay mixture, and β-mercaptoethanol was used at 1 to 30 μM.

RESULTS

WfeD enzyme activity and kinetics.

Under standard assay conditions, the reaction rate of WfeD in bacterial homogenates using 1 mM UDP-Gal and 0.25 mM GlcNAc-PP-(CH₂)₁₁-OPh acceptor in the presence of 5 mM Mn²⁺ ions was linear up to at least 20 min. The reaction rate was proportional to the enzyme protein concentration up to 0.016 mg/ml (not shown). Untransformed control BL21 cells without pLW1298 showed a low background (<5%) compared to transformed BL21 cells, suggesting a high specific activity in the crude enzyme preparation (about 8 μmol/h/mg). The crude enzyme preparation retained at least 60% of the initial activity after storage at 4°C for up to 9 weeks. The decay of the activity after storage at −20°C was similar to that at 4°C. However, after 5 months of storage at −20°C, the activities were decreased to <5% of the original. The kinetic parameters for the crude enzyme were an apparent K_m for UDP-Gal of 0.1 mM and an apparent V_max of 11 μmol/h/mg. For the acceptor substrate GlcNAc-PP-PhU, the apparent K_m was 0.4 mM and the apparent V_max was 12 μmol/h/mg (data not shown).

Enzyme purification.

WfeD was purified using Ni²⁺-NTA affinity chromatography to a specific activity 16-fold higher than that of the crude bacterial homogenate, considering the total amount of protein in both preparations. Based on the amino acid composition, the molecular mass of WfeD was calculated to be 30 kDa (protein molecular mass calculator; Science Gateway). SDS-PAGE showed a band at 30 kDa that was induced by IPTG (Fig. 1A). Western blotting, using antibody to the His tag, also showed a major protein band at 30 kDa for both the solubilized crude enzyme and the purified enzyme (Fig. 1B). MALDI peptide mass fingerprinting of gel slices identified the 30-kDa band as WfeD (data not shown). For the purified enzyme, the apparent K_m for UDP-Gal was 0.25 mM, with a V_max of 40 μmol/h/mg, measured with 1 mM GlcNAc-PP-PhU (data not shown). The K_m for the acceptor substrate GlcNAc-PP-(CH₂)₁₁-OPh was 0.10 mM, with a V_max of 42 μmol/h/mg, measured with 1 mM UDP-Gal (data not shown). The kinetic studies were performed by two independent repeats, with differences of <10%.

FIG. 1.

Expression and purification of WfeD and six mutants. (A) SDS-PAGE analysis of proteins. Wild-type WfeD and six mutants, eluted from the Ni²⁺-NTA Sepharose Fast Flow column, as well as wild-type bacterial lysates (before IPTG induction) were run on reducing SDS-PAGE gels. Molecular mass (M) markers are indicated on the right. The band of the wild-type enzyme at 30 kDa was analyzed by peptide mass fingerprinting and found to be due to WfeD. (B) Western blot analysis of purified wild-type WfeD and six mutants (D99A, E101A, E103A, R210A, R212A, and K211A). An anti-His tag antibody was used to detect the WfeD protein. All of the mutants, as well as the wild-type enzyme purified on a Ni²⁺-NTA Sepharose Fast Flow column, showed a protein band of 30 kDa. Bovine β4-GalT (33 kDa) without a His tag was used as the negative control.

Confirmation of Gal transfer and newly synthesized linkage as Galβ1-4GlcNAc.

To confirm the synthesis of enzyme product, assay mixtures were applied onto C₁₈ Sep-Pak columns, and the enzyme product and substrate were eluted with MeOH. These fractions were then subjected to HPLC analysis. Using a C₁₈ column and an acetonitrile/water ratio of 24:76 as the mobile phase, the product eluted at 25 min and the substrate eluted at 33 min, allowing a good separation of the two compounds. The fractions eluted with MeOH from the Sep-Pak column were also analyzed by electrospray-MS analysis (Fig. 2) and showed peaks at m/z 626 for the substrate GlcNAc-PP-(CH₂)₁₁-OPh and m/z 788 for the product Gal-GlcNAc-PP-(CH₂)₁₁-OPh. This showed that only one Gal residue (m/z 162) had been added to the substrate. Since the next sugar in the repeating unit is also a β1-4-linked Gal, it was considered possible that WfeD is a bifunctional enzyme. However, there was no peak at m/z 950 that could have been due to Gal-Gal-GlcNAc-PP-(CH₂)₁₁-OPh. Thus, under our reaction conditions, WfeD did not have a dual activity.

FIG. 2.

Electrospray mass spectrometry of the WfeD product Gal-GlcNAc-PP-(CH₂)₁₁-OPh. The disaccharide lipid product was synthesized from GlcNAc-PP-(CH₂)₁₁-OPh with a bacterial homogenate containing WfeD Gal-transferase, purified by Sep-Pak C₁₈ columns, and analyzed by electrospray mass spectrometry (negative-ion mode). Both m/z 626 (substrate) and m/z 788 (product) represent [M − H]. The absence of m/z 950 indicates that only one Gal residue was transferred. amu, atomic mass unit (equal to daltons).

Galactosidases were used to determine the anomeric linkage between Gal and GlcNAc in the [³H]Gal-GlcNAc-PP-(CH₂)₁₁-OPh enzyme product synthesized by purified WfeD. The released free [³H]Gal was separated by an AG1x8 column that binds the uncleaved product. We showed that the enzyme product was resistant to green coffee bean α-galactosidase, indicating that Gal was not α linked. In contrast, treatment with jack bean (β1,4-specific) β-galactosidase and bovine testicular (β1-3-, β1-4-, and β1-6-specific) β-galactosidase released most of the radioactivity (52% and 60%, respectively) of the radioactivity from the disaccharide product. Therefore, it can be concluded that Gal was β linked in the enzyme product and likely in the 1-4 linkage.

The ¹H-NMR spectrum of the product showed a new doublet at 4.345 ppm with a coupling constant of 7.6 Hz, which is indicative of H-1 of Gal in the β linkage (Table 3). The spectrum of the enzyme product of β3-GalT WbbD [Galβ1-3GlcNAc-PP-(CH₂)₁₁-OPh] also showed a doublet for Gal H-1 at 4.35 ppm (17). To determine whether Gal was β1-3, β1-4, or β1-6 linked to GlcNAc, we conducted two-dimensional NMR (2D-NMR) experiments, including correlation spectroscopy (COSY) (Fig. 3A) and heteronuclear single quantum coherence (HSQC) (Fig. 3B), which identified most of the carbon and proton shifts of the GlcNAc residue. The chemical shifts of GlcNAc H-1, H-2, and H-3 were slightly different in the product compared to the substrate. However, there was a large downfield shift of GlcNAc H-4 from 3.33 ppm in the substrate to 3.626 ppm in the product and a large shift from the C-4 signal of 71.9 in the substrate to 79.1 in the product. This indicates that Gal was linked to the 4 position of GlcNAc and the Galβ1-4GlcNAc linkage in the enzyme product. Therefore, these results show that WfeD is a UDP-Gal:GlcNAc-R β1,4-Gal-transferase.

TABLE 3.

¹H- and ¹³C-NMR parameters established from COSY, NOESY, and HSQC^a

Product	H	Shift (ppm)	J coupling (Hz)	C	Shift (ppm)
GlcNAcα-PP-(CH2)11-OPh
GlcNAc	H-1	5.53 (5.64)		C-1	96
	H-2	3.97		C-2	55
	H-3	3.75		C-3	72.9
	H-4	3.33 (3.45)		C-4	71.9
GlcNAcα-PP-decanol (7)
GlcNAc	H-1	5.45	3.2, 7.2	C-1	94.5
Galβ1-4GlcNAc-PP-(CH2)11-OPh (WfeD)
GlcNAc	H-1	5.561		C-1	94.7
	H-2	4.027		C-2	53.1
	H-3	3.846		C-3	69.4
	H-4	3.626		C-4	79.1
Gal	H-1	4.345	7.6	C-1	103.4
	H-2	3.538	6.9, 10.1	C-2	70.8
	H-3	3.465	2.9, 9.5	C-3	73.1
	H-4	3.792	2.5	C-4	68.8
Galβ1-3GlcNAc-PP-(CH2)11-OPh (17)
GlcNAc	H-1	5.58		C-1	94.8
	H-2	4.14		C-2	52.4
	H-3	3.84		C-3	80.3
Gal	H-1	4.35		C-1	104.0
Glcβ1-3GlcNAc-PP-(CH2)11-OPh (6)
GlcNAc	H-1	5.54		C-1	95.0
	H-2	4.14		C-2	54.0
	H-3	3.87		C-3	82.8
	H-4	3.42		C-4	70.3
Glc	H-1	4.41		C-1	105.0
	H-2	3.18		C-2	74.5
	H-3	3.37		C-3	77.6
	H-4	3.33

^aShown is a comparison of NMR parameters between substrates and β3-GalT, β4-GalT, and β3-Glc-transferase enzyme products. NOESY, nuclear Overhauser enhancement spectroscopy.

FIG. 3.

NMR spectroscopy (600 MHz) of the WfeD enzyme product. (A) ¹H-¹H correlation COSY spectrum. The WfeD enzyme product was exchanged in D₂O and dissolved in CD₃OD, and ¹H and ¹³C spectra were acquired by using a 600-MHz Bruker spectrometer. The ¹H-¹H COSY spectrum is shown. (B) HSQC spectrum. Spectra were acquired by using a 600-MHz Bruker spectrometer. The chemical shifts of H-4 and C-4 of GlcNAc exhibit a major change compared to the substrate, due to the attachment of Gal to GlcNAc in the β1-4 linkage. The ¹H-¹³C correlation spectrum (HSQC) is shown.

Donor substrate specificity.

The donor substrate specificity study of the enzyme in the crude cell lysate was carried out by using the standard assay but replacing 0.5 mM UDP-Gal with several other nucleotide sugars, i.e., 0.25 mM UDP-Glc, 0.4 mM UDP-GalNAc, 0.4 mM UDP-GlcNAc, 0.55 mM UDP-xylose, 0.4 mM GDP-Man, and 0.2 mM CMP-sialic acid. These nucleotide sugars yielded enzyme activities of less than 1% of the activity observed with UDP-Gal. This indicated a strict specificity of WfeD for UDP-Gal as the donor substrate and the absence in bacterial homogenates of other types of glycosyltransferase activities acting on the substrate GlcNAc-PP-(CH₂)₁₁-OPh. It also showed that the 4-epimerase present in E. coli did not convert significant amounts of UDP-Glc to UDP-Gal within the 10-min incubation time.

Role of pyrophosphate in acceptor substrate specificity.

In order to define further the structural requirements of WfeD for the acceptor substrate, a number of neutral and negatively charged phosphate-containing analogs of GlcNAc (5, 7, 8, 13, 18) were tested as acceptor substrates (Table 2). The neutral compounds included both α-and β-anomers of GlcNAc-benzyl; β-anomers of GlcNBu-benzyl, GlcNAc-naphthyl, GlcNAc-isoquinolinyl (5), as well as GlcNAcα-OCOCH₂-COO(CH₂)₁₁-OPh; and α- and β-anomers of GlcNAc-(CH₂)₁₁-OPh (18). Using the AG1x8 anion-exchange method to isolate the product, these neutral substrates showed activities of less than 8% of the activity with the standard substrate. The charged compounds included GlcNAcα-PP-(CH₂)₆-OPh, GlcNAcα-PP-(CH₂)₁₆-OPh, and GlcNAcα-PP-(CH₂)₉-CH₃, having a pyrophosphate group but modifications in the lipid moiety. All of these compounds containing GlcNAc-PP showed high levels of activity (39 to 92% of the activity with the standard substrate). This indicates that the structure of the lipid moiety in the acceptor is of minor importance, although a hydrophobic group may be required. In order to examine the role of the pyrophosphate in the substrate, we tested GlcNAcα-P-(CH₂)₁₁-OPh, which has only one phosphate group, and found 11% activity compared to the identical substrate having two phosphates (Table 2). WfeD thus has an absolute requirement for at least one phosphate group and requires the pyrophosphate group in the acceptor for full activity.

Effect of detergents on enzyme activity.

In order to determine whether the enzyme activity is affected by the presence of micelle-forming detergents, Triton X-100, Tween 80, or NP-40 (nonyl phenoxylpolyethoxylethanol) was added to the crude cell homogenate in the assay mixture at a 0.1% (vol/vol) concentration. The presence of detergent yielded WfeD enzyme activities between 80 and 90% of the activities measured in the absence of the detergent (Table 4). As the concentration of Triton X-100 increased from 0 to 0.5% in the reaction mixture, WfeD activity decreased but was maintained at about 60% of the original activity, up to 6% Triton X-100.

TABLE 4.

Effects of detergents on crude WfeD enzyme activity^a

Treatment	Mean activity (%)
No detergent	100 ± 6
Triton X-100	86 ± 4
Tween 80	92 ± 5
NP-40	86 ± 10

^aWfeD was assayed in the presence of 0.1 % (vol/vol) NP-40 (nonyl phenoxylpolyethoxylethanol 40), Tween 80, or Triton X-100. The standard assay mixture without any detergent added (“no detergent”) was used as a positive control. Treatments with detergents were carried out in duplicates, with the variation of the duplicates indicated.

Divalent cation requirement for WfeD enzyme activity.

The involvement of divalent cations as cofactors in GalT catalysis is suggested by the presence of the DxD motif (3). Previous reports stated that the presence of a cofactor such as Mn²⁺ is essential for the activity of bovine β4-GalT (5) and β3-GalT WbbD from E. coli (17), both of which have a GT-A fold and a DxD motif in the UDP-Gal binding site. WfeD has a DYEIE sequence resembling a DxD motif and is a member of the GT26 family of glycosyltransferases with a predicted WecG-TagA fold (www.CAZy.org).

To examine the cofactor requirement of WfeD, assays were carried out in the presence of 5 mM Mn²⁺ or, instead of Mn²⁺, with Mg²⁺, Cu²⁺, Ca²⁺, Pb²⁺, Ni²⁺, or EDTA or no addition in the negative control. The highest stimulation of activity was obtained in the presence of Mn²⁺ ions, but Ca²⁺ and Mg²⁺ were also effective. Surprisingly, Pb²⁺ ions, often inhibitory to enzymes, and Ni²⁺ ions, which are not usually studied as glycosyltransferase cofactors (22), were both shown to induce high levels of enzyme activity in the bacterial homogenate as well as in the purified enzyme (Fig. 4A and B). A further test showed that the effect of Pb²⁺, Ni²⁺, and Mn²⁺ on enzyme activity was concentration dependent (Fig. 4C). HPLC analysis confirmed the identity of the enzyme product formed in the presence of these metal ions. This finding may uncover a novel role that Pb²⁺ plays in the metabolism of bacteria. Mammalian β4-GalT was also tested in the presence of Pb²⁺ salts at a 5 mM concentration in the assay and showed a slight stimulation of activity, which was 11% of the activity found in the presence of 5 mM Mn²⁺.

FIG. 4.

Divalent cation requirement of WfeD. (A) Cofactor requirement of the crude WfeD enzyme. Freshly homogenized 1:10-diluted E. coli cells were used as the crude enzyme. The activities are shown as μmol/h/mg. EDTA (5 mM) and H₂O were used as negative controls. Assays were carried out in duplicates, and bars indicate the differences between duplicates. Cu²⁺ ions did not stimulate the activity. (B) Cofactor requirement of the purified WfeD enzyme. The purified WfeD solution (stored for 3 weeks) was used for the assay. The activities are shown as μmol/h/mg. EDTA (5 mM) and H₂O were used as negative controls. Assays were carried out in duplicates, and bars indicate the differences between duplicates. The products obtained by using Mn²⁺, Pb²⁺, and Ni²⁺ ions were confirmed by HPLC. (C) Concentration effects of divalent metal ions. The activity of WfeD (using 1:10-diluted crude enzyme homogenate) was measured after the addition of MnCl₂, Pb(OAc)₂, or NiSO₄ at different concentrations in the assay mixture. Activity is indicated as μmol/h/mg protein.

Effect of amino-acid-specific reagents on WfeD activity.

There are many His, Arg, and Lys residues in WfeD that may possibly be involved in the binding of the negatively charged substrates. In order to examine the involvement of specific amino acids in catalysis, the standard assay was carried out in the presence of amino-acid-specific reagents. The inclusion of 0.2 mM DEPC in the assay mixture showed a 31% inhibition of activity, while 0.2 mM HPG showed a 50% inhibition (Fig. 5). This suggests a potential role of His and Arg in protein structure or catalysis. The alkylating agent IAA (0.2 mM) showed a 20% inhibition, indicating that one or both of the Cys residues may be involved in protein structure or catalysis. DTNB (0.2 mM), which binds to reduced SH groups of Cys, inhibited the activity by 85%. DTT (0.2 mM), which reduces disulfide bonds, stimulated the enzyme almost 5-fold at concentrations of 0.1 to 3 mM. β-Mercaptoethanol also increased the activity 4-fold at 1 to 30 mM concentrations in the assay. This suggested that at least one reduced Cys residue may be required for full activity.

FIG. 5.

Effect of amino acid reagents on activity of WfeD. Amino-acid-specific reagents were preincubated with purified enzyme for 10 min, prior to the assays. The percentages of activities are shown relative to the positive control (no addition of amino acid reagents). The reagent treatments were applied in duplicates, and bars indicate variations between duplicates. DEPC acts on His, HPG acts on Arg, and DTNB and IAA react with the SH group of Cys.

Mutagenesis to identify crucial amino acids.

The WfeD enzyme has a DYEIE sequence resembling a DxD motif, found in virtually all putative glycosyltransferases (3). To evaluate the role of the acidic amino acids in this motif in enzyme activity, mutations of single acidic amino acids were made. We therefore mutated individually Asp99, Glu101, and Glu103 to the neutral Ala. The D99A and E103A mutants were highly active, but the E101A mutant exhibited minimal enzyme activity. This suggested that Glu101 in the DYE sequence has the catalytic function of an acidic residue in the DxD sequence observed for other glycosyltransferases. Since the donor and acceptor substrates both have a negatively charged pyrophosphate, we proposed that positively charged clusters of amino acids may be important for substrate binding. We therefore mutated Arg210, Arg212, and Lys211 of the RKR sequence at the N terminus of the enzyme to Ala. R210A and R212A mutants were active, while the K211A mutant lacked activity (Fig. 6). This indicated an important role of Lys211 in either substrate binding, catalysis, or protein structure. The GalT activity of the mutants was also tested with Pb²⁺, and product formation was confirmed by HPLC. The replacement of Glu101 by Ala greatly reduced enzyme activity in the presence of either Mn²⁺ or Pb²⁺, although a 60-fold-higher activity was seen for the Lys211Ala mutant in the presence of Pb²⁺ than in the presence of Mn²⁺. Western blots of Ni²⁺-NTA-purified bacterial extracts harboring the mutant enzymes showed that all of the mutants were expressed at 30 kDa (Fig. 1B). The single bands seen on Western blots indicated that none of the wild-type or mutant enzymes were degraded.

FIG. 6.

GalT activities of mutant enzymes. GalT activities of crude cell homogenates containing site-specific WfeD K211A, R212A, R210A, E103A, E101A, and D99A mutants as well as the wild-type enzyme were assayed using 1 mM UDP-Gal and 0.25 mM GlcNAc-PP-(CH₂)₁₁-OPh acceptor substrate in the presence of 5 mM MnCl₂. The activities are shown as μmol/h/mg. The wild-type and mutant enzymes were tested in duplicates, and bars indicate the differences between the duplicates. Activities were normalized according to the relative amount of His-tagged protein. The E101A and K211A mutants showed greatly reduced GalT activities.

DISCUSSION

We have shown that the second step of the repeating-unit assembly in B14 is catalyzed by WfeD, the gene product of wfeD. Analysis of the WfeD enzyme product structure by HPLC, mass spectrometry, NMR, and galactosidase digestion clearly showed that WfeD adds one Gal residue in the β1-4 linkage to GlcNAcα-pyrophosphate-lipid. There are several other putative glycosyltransferase genes in the B14 O-antigen gene cluster (wfeA, wfeB, and wfeE) that may encode the glycosyltransferases necessary to add the remaining sugars. Based on the structure of the B14 O antigen, these enzymes are expected to be a second β1,4-Gal-transferase, a β1,6-glucuronosyltransferase, and an α1,4-Gal-transferase. The third residue of the repeating unit, which is also a β1-4-linked Gal, is likely to be added by another enzyme that remains to be identified and is expected to have the same specificity for UDP-Gal but a different acceptor substrate specificity. As there is very little sequence similarity between the enzymes encoded by the genes in the B14 O-antigen gene cluster (wfeA, wfeB, and wfeE) that could help to predict their specific functions, analyses similar to those described here are required to determine their roles in O-antigen assembly.

The synthetic acceptor substrate GlcNAcα-PP-(CH₂)₁₁-OPh is an amphipathic molecule with similarity to the endogenous substrate GlcNAcα-PP-Und that acts as a lipid carrier in O-antigen biosynthesis and is exposed at the cytoplasmic side of the inner membrane (26). We have shown that GlcNAcα-PP-(CH₂)₁₁-OPh is an excellent in vitro acceptor substrate for WfeD. The advantage of GlcNAcα-PP-(CH₂)₁₁-OPh is that the substrate and the enzymatic reaction product can be readily detected and separated by HPLC. We have previously shown that this compound was also an excellent substrate for the β1,3-Gal-transferase WbbD from E. coli strain VW187 (8, 17) and the β1,3-Glc-transferases WfgD and WfaP from E. coli serogroups O152 and O56, respectively (6). A similar compound was also shown to be an acceptor for a GalNAc-transferase from E. coli O86:H2 (28). All of these enzymes add the second sugar residue to GlcNAc in the repeating-unit assembly.

Detergent treatment of crude WfeD cell homogenates suggested that the enzyme can be solubilized but is not activated by detergents. Although the purified protein appeared to be >95% pure, it was purified only 16-fold compared to the specific activity in bacterial sonicates. This suggests that the enzyme was partially insoluble under our conditions and remained associated with the membrane; thus, a substantial portion of the total enzyme was not purified. However, the intracellular localization of the WfeD protein remains to be confirmed, for example, by cell fractionation.

The WfeD enzyme was found to be surprisingly stable after cold storage. The kinetic properties of the purified WfeD β1,4-Gal-transferase resemble those of previously studied mammalian and bacterial glycosyltransferases that have similar activities but unrelated amino acid sequences (3, 5, 14, 17, 26), suggesting a common catalytic mechanism.

A number of GlcNAc-pyrophosphate-lipid analogs proved to be good substrates, independent of the lipid structure. Thus, the main groups that are expected to bind to the enzyme are the sugar and the pyrophosphate phosphate group. Modifications of the pyrophosphate group eliminated the activity and binding to the enzyme, although the compound having a single phosphate did have 11% activity. None of the GlcNAc analogs tested, which were inactive as substrates, bound to and inhibited WfeD, and we previously reported similar findings for WbbD (8) and for WfgD and WfaP (6). Thus, the pyrophosphate group may be an essential group responsible for the binding of the acceptor to the enzymes that recognize GlcNAc-PP-lipid. In the development of an enzyme inhibitor, therefore, the pyrophosphate group must be intact or present as a closely related structural analog.

The DxD motif found in many glycosyltransferases was shown previously to be involved in divalent-metal-ion-stimulated enzyme catalysis (3, 24). WfeD has a DxExE sequence in a central position of the protein, and we have shown by mutagenesis that the second acidic residue in this sequence (Glu101) is essential for activity. Thus, the divalent-metal-ion-supported nucleophilic attack by Glu (or Asp) appears to be a common mechanism in the catalysis of GalT. This motif is within a hydrophilic region of the protein and may be part of the UDP-Gal binding site.

The WfeD protein has been classified as a member of the GT26 family of glycosyltransferases with a WecG-TagA fold (www.CAZy.org), many of which act on sugar-pyrophosphate lipids. However, there is no structure available for other transferases with this predicted fold or a possible arrangement of the catalytic region. Mammalian β1,4-Gal-transferase is almost identical in its specificity but has a GT-A fold within the GT7 family. The divalent metal ion may complex the pyrophosphate group of UDP-Gal, thus facilitating the nucleophilic action of the Glu residue on the 4-hydroxyl of GlcNAc to be galactosylated (24). It is surprising that several metal ions, including Mn²⁺, Ni²⁺, and Pb²⁺, can stimulate WfeD activity, especially given that Pb²⁺ is usually toxic to enzymes. However, in a previously reported protein kinase C study, Pb²⁺ was also found to activate the enzyme at very low concentrations (11). The slight stimulation of mammalian β4GalT by Pb²⁺ salts suggests that this metal ion can also bind to the mammalian enzyme and replace the function of Mn²⁺ ions, although this had not been considered previously. Further structural analysis of the WfeD protein may confirm that Pb²⁺ is forming a complex with UDP-Gal or may reveal other functions for Pb²⁺ suggesting the mechanism of its stimulation and to determine why its stimulation is much more pronounced in the K211A mutant.

Several putative glycosyltransferases with sequence similarity to WfeD (see Fig. S1 in the supplemental material) have the RKR sequence or a similar cluster of positively charged amino acids and are also assumed to bind two substrates containing pyrophosphate bonds. The RKR motif in WfeD may interact with one or more phosphates of the acceptor pyrophosphate group. The mutation of Lys211, but not of the adjacent Arg residues, resulted in a loss of activity, suggesting that either there might be a specific interaction between the positively charged amino acid and the negatively charged acceptor or it may be directly involved in catalysis or in an essential structural function.

WfeD has two Cys residues, one of which (Cys148) is highly conserved. Cys reagents that restore free SH bonds stimulated the activity, while reagents that covalently modify Cys inhibited the activity. This suggests that at least one of the two Cys-SH groups participates in catalysis or is essential for the structure of the protein. A similar finding was made with core 2 GlcNAc-transferase, which forms intramolecular disulfide bonds but requires one Cys-SH for activity (27). Cys148 in WfeD may be this critical Cys residue. This could be verified by mutagenesis as well as peptide analysis by mass spectrometry. X-ray crystallography will help to further unravel the enzyme-substrate interactions and enzyme protein structure and mechanisms of WfeD.