Three Enzymatic Steps Required for the Galactosamine Incorporation into Core Lipopolysaccharide

Abstract

The core lipopolysaccharides (LPS) of Proteus mirabilis as well as those of Klebsiella pneumoniae and Serratia marcescens are characterized by the presence of a hexosamine-galacturonic acid disaccharide (αHexN-(1,4)-αGalA) attached by an α1,3 linkage to l-glycero-d-manno-heptopyranose II (l-glycero-α-d-manno-heptosepyranose II). In K. pneumoniae, S. marcescens, and some P. mirabilis strains, HexN is d-glucosamine, whereas in other P. mirabilis strains, it corresponds to d-galactosamine. Previously, we have shown that two enzymes are required for the incorporation of d-glucosamine into the core LPS of K. pneumoniae; the WabH enzyme catalyzes the incorporation of GlcNAc from UDP-GlcNAc to outer core LPS, and WabN catalyzes the deacetylation of the incorporated GlcNAc. Here we report the presence of two different HexNAc transferases depending on the nature of the HexN in P. mirabilis core LPS. In vivo and in vitro assays using LPS truncated at the level of galacturonic acid as acceptor show that these two enzymes differ in their specificity for the transfer of GlcNAc or GalNAc. By contrast, only one WabN homologue was found in the studied P. mirabilis strains. Similar assays suggest that the P. mirabilis WabN homologue is able to deacetylate both GlcNAc and GalNAc. We conclude that incorporation of d-galactosamine requires three enzymes: Gne epimerase for the generation of UDP-GalNAc from UDP-GlcNAc, N-acetylgalactosaminyltransferase (WabP), and LPS:HexNAc deacetylase.

Introduction

Proteus mirabilis is a common uropathogen that causes urinary tract infections, especially in individuals with functional or anatomical abnormalities of the urinary tract (1) and elderly ones undergoing long term catheterization (2), but less frequently in normal hosts (3). Potentially serious complications arising from P. mirabilis infections include bladder and kidney stone formation, catheter obstruction by formation of encrusting biofilms, and bacteremia (reviewed in Ref. 4). This bacterium is found more frequently than Escherichia coli in kidney infection (5) and may be associated to rheumatoid arthritis (6).

As in other Enterobacteriaceae, in the genus Proteus LPS, three domains are recognized: the highly conserved and hydrophobic lipid A; the hydrophilic and highly variable O-antigen polysaccharide (O-PS)³ with more than 60 serogroups recognized (7–10); and the core oligosaccharide (OS), connecting lipid A and O-antigen. The core domain is usually divided into inner and outer core on the basis of sugar composition.

The core OS structure of 34 Proteus strains of different O-serogroups has been determined (11). The core OS of these strains share a common heptasaccharide fragment that includes a 3-deoxy-α-d-manno-oct-2-ulosonic acid (Kdo) disaccharide, a l-glycero-α-d-manno-heptosepyranose trisaccharide, and one residue each of Glc, GalA, and either d-glucosamine (GlcN) or d-galactosamine (GalN) (Fig. 1) (reviewed in Ref. 11). This common fragment is also found in the core LPS of Klebsiella pneumoniae and Serratia marcescens (12–14), and we have shown that two enzymes are required for the incorporation of GlcN into the core LPS of K. pneumoniae: the WabH enzyme catalyzing the incorporation of GlcNAc from UDP-GlcNAc to outer core LPS and WabN catalyzing the deacetylation of the incorporated GlcNAc (15). The rest of the Proteus core OS is quite variable, with up to 37 different structures recognized in the genus Proteus and 11 in P. mirabilis (11).

FIGURE 1.

P. mirabilis core OS structures and genetic organization of the core OS biosynthetic clusters. A, the common part of core OS shared by all of the studied strains of P. mirabilis (11), K. pneumoniae, and S. marcescens (11, 13, 14). Broken lines denote the truncation level for the different core biosynthetic gene mutations (13, 14, 15, 16, 35). B, the core OS structures of P. mirabilis strains 50/57 and TG83 (11). LD-Hep, l-glycero-d-manno-heptopyranose DD-Hep, d-glycero-d-manno-heptopyranose; GaloNAc, N-acetylgalactosamine open chain form; Qui, quinovosamine. C, a diagram of the wa gene cluster from P. mirabilis strains 50/57 and TG83 (this work), R110 and 51/57 (16), and HI4320 (28). Common core genes (black arrows), other outer core genes (gray arrows), waaL (O-antigen:lipidA-core ligase) (striped arrows), non-core-related (stippled arrows), and genes of unknown function (white arrows) are shown.

Recently, we have identified the genes involved in the biosynthesis of the P. mirabilis common core heptasaccharide for strains containing GlcN (R110 and 51/57) (16). As in other Enterobacteriaceae, these genes were clustered in the so-called wa region, and among these genes, K. pneumoniae wabH and wabN homologues were found able to complement the corresponding non-polar mutants (16). In this work, we characterize the wa region for two P. mirabilis strains containing GalN instead of GlcN and present evidence for a specific GalNAc transferase in these strains.

EXPERIMENTAL PROCEDURES

Bacterial Strains, Plasmids, and Growth Conditions

Bacterial strains and plasmids used in this study are shown in Table 1. Bacterial strains were grown in LB broth and LB agar (17). LB medium was supplemented with kanamycin (50 μg/ml), tetracycline (20 μg/ml), and ampicillin (100 μg/ml) when needed.

TABLE 1.

Bacterial strains and plasmids used

Strain or plasmid	Relevant characteristics	Reference or source
Strains
P. mirabilis
50/57	Serogroup O27	J. Sidorczyk
TG83	Serogroup O57	J. Sidorczyk
R110	Serogroup O3ab	J. Sidorczyk
OXK	Serogroup O3ab	J. Sidorczyk
51/57	Serogroup O28	J. Sidorczyk
14/57	Serogroup O6	J. Sidorczyk
ATCC 29906	Type strain	CECTa
CECT 170		CECTa
K. pneumoniae
52145	Serogroup O1:K2 (core type 2)	Ref. 37
52145ΔwabN	Non-polar wabN mutant	Ref. 15
52145ΔwabH	Non-polar wabH mutant	Ref. 15
52145ΔwaaL wabN	Double non-polar waaL wabN mutant	Ref. 15
52145ΔwaaL wabH	Double non-polar waaL wabH mutant	Ref. 15
E. coli
DH5α	F−endA hsdR17 (rk− mk−) supE44 thi-1 recA1 gyr-A96 φ80lacZ	Ref. 38
BL21(λD3)	F−ompT hsdSB (rB− mB−) gal dcm(λD3)	Novagen
Plasmids
pGEM-T Easy	PCR-generated DNA fragment cloning vector Ampr	Promega
pGEM-T- wabH52145	pGEM-T containing the PCR-amplified waaN52145 gene
pGEM-T-wabHR110	pGEM-T containing the PCR-amplified wabHR110 gene	Ref. 16
pGEM-T-wabP50/57	pGEM-T containing the PCR-amplified wabP50/57 gene	This study
pGEM-T-wabNR110	pGEM-T containing the PCR-amplified wabNR110 gene	Ref. 16
pGEM-T-wabN50/57	pGEM-T containing the PCR-amplified wabN50/57 gene	This study
pET28a(+)	T4-inducible expression vector, KmR	Novagen
pET28a-wabHR110	pET28a expressing His6-WabHR110	This study
pET28a-wabP50/57	pET28a expressing His6-WabP50/57	This study
pET28a-wabNR110	pET28a expressing His6-WabNR110	This study
pET28a-wabN50/57	pET28a expressing His6-WabN50/57	This study
pACYC-GNEA. hydrophila	pACYC184 with A. hydrophila AH-3 gne	Ref. 31
pACYC-GNEP. mirabilis	pACYC184 with P. mirabilis ATCC 29906 gne	This study

^a Colección Española de Cultivos Tipo (Spanish Collection of Culture Types).

General DNA Methods

Standard DNA manipulations were done essentially as described (18). DNA restriction endonucleases, T4 DNA ligase, E. coli DNA polymerase (Klenow fragment), and alkaline phosphatase were used as recommended by the suppliers.

DNA Sequencing and Computer Analysis of Sequence Data

Double-stranded DNA sequencing was performed by using the dideoxy-chain termination method (19) with the ABI Prism dye terminator cycle sequencing kit (PerkinElmer Life Sciences). Oligonucleotides used for genomic DNA amplifications and DNA sequencing were purchased from Amersham Biosciences. The DNA sequence was translated in all six frames, and all open reading frames (ORFs) were inspected. Deduced amino acid sequences were compared with those of DNA translated in all six frames from nonredundant GenBank^TM and EMBL databases by using the BLAST (20) network service at the National Center for Biotechnology Information and the European Bioinformatics Institute. ClustalW was used for multiple-sequence alignments (21).

Plasmid Constructions for Mutant Complementation Studies

The wabH_R110 (1137-bp) and wabN_R110 (963-bp) genes from P. mirabilis R110 were PCR-amplified from strain R110 genomic DNA with oligonucleotide pairs BHf (5′-TGGCGATGGCAAATTTTACT-3′)-BHr (5′-ATTCCGGCCGATAACTTAGG-3′) and BNf (5′-GTGCACGAATTGCTCTGATG-3′)-BNr (5′-ATGGGTGGCAAGATAATGCT-3′), obtaining fragments of 1310 and 1508 bp, respectively. Similarly, the wabP_50/57 and wabN_50/57 were amplified from strain 50/57 genomic DNA with oligonucleotides 50-7 (5′-TAGCTGCAGCTATTTCAGCC-3′)-50-4 (5′-GGGATAATGGTGGCGTAATG-3′) and BNf-BBNr (5′-GGCTTTCCATTGGTCAGCTA-3′), obtaining fragments of 1881 and 1621 bp, respectively. These amplicons were ligated to vector pGEM-T (Promega) and transformed into E. coli DH5α to generate plasmids pGEM-T-wabH_R110, pGEM-T-wabN_R110, pGEM-T- wabP_50/57, and pGEM-T- wabN_50/57. A DNA fragment (1400 bp) containing the gne from P. mirabilis ATCC 29906 was amplified with oligonucleotides GneExtFw (5′-GAGCTCCCATGGTGAAATGAAACGTG-3′) and GneExtRv (5′-TCTAGAACCCGTATTCGGTGGAATTT-3′), where underlined letters denote sequences cut by SacI and XbaI, respectively. The amplified fragment was cloned in pGEM-T to obtain pGEM-T-GNE_{P. mirabilis}. By SacI digestion, a 1454-bp fragment was obtained, gel-purified, and blunt-ended with T4 DNA polymerase. This fragment was ligated to SalI-digested and blunt-ended pACYC184 to obtain pACYC-GNE_{P. mirabilis}.

Gene Distribution

To determine the distribution among several P. mirabilis strains of the gne gene an internal fragment of this gene was amplified using oligonucleotides GneIntFw (5′-ACTCGCCTATCAGGCCAATT-3′) and GneIntRev (5′-TGACTATGGCGTTTGTCGAG-3′). Oligonucleotide pairs WabHIntFw (5′-TGCCAACTCGATGGATAAGA-3′)-WabHIntRv (5′-GCGTAATTTTAGGCGGGTTA-3′) and WabPIntFw (5′-ACAACCTAACCCGTTTGCAG-3′)-WabPIntRv (5′-TCTGCGAGTGAGTCTGCATC-3′) were used to determine the distribution of wabH and wabP genes, respectively.

LPS Isolation and Electrophoresis

LPS was extracted from dry cells grown in LB. The phenol/chloroform/light petroleum ether method (22) was used for strains producing rough LPS, whereas the phenol/water procedure (23) was used for the strains producing the O antigen domain (smooth LPS). For screening purposes, LPS was obtained after proteinase K digestion of whole cells (24). LPS samples were separated by SDS-PAGE or Tricine-SDS-PAGE and visualized by silver staining as described previously (25, 26).

Preparation of Oligosaccharides

The LPS preparations (20 mg) were hydrolyzed in 1% acetic acid (100 °C, 120 min), and the precipitate was removed by centrifugation (8000 × g, 30 min) and lyophilized to give Lipid A. The supernatants were fractionated on a column (56 × 2.6 cm) of Sephadex G-50 (S) in 0.05 m pyridinium acetate buffer, pH 4.5, with monitoring using a differential refractometer to obtain the oligosaccharide fractions.

GC-MS Analysis

Partially methylated alditol acetates and methyl glycoside acetates were analyzed on a Agilent Technologies 5973N MS instrument equipped with a 6850A gas chromatograph and an RTX-5 capillary column (Restek, 30 m × 0.25-mm inner diameter, flow rate 1 ml/min, helium as carrier gas). Acetylated methyl glycosides analysis was performed with the following temperature program: 150 °C for 5 min, 150 → 250 °C at 3 °C/min, 250 °C for 10 min. For partially methylated alditol acetates, the temperature program was as follows: 90 °C for 1 min, 90 → 140 °C at 25 °C/min, 140 → 200 °C at 5 °C/min, 200 → 280 °C at 10 °C/min, 280 °C for 10 min.

Mass Spectrometry Studies

Positive and negative ion reflectron time-of-flight mass spectra (MALDI-TOF) were acquired on a Voyager DE-PRO instrument (Applied Biosystems) equipped with a delayed extraction ion source. Ion acceleration voltage was 25 kV, grid voltage was 17 kV, mirror voltage ratio was 1.12, and delay time was 150 ns. Samples were irradiated at a frequency of 5 Hz by 337-nm photons from a pulsed nitrogen laser. A solution of 2,5-dihydroxybenoic acid in 20% CH₃CN in water at a concentration of 25 mg/ml was used as the MALDI matrix. One μl of matrix solution was deposited on the target, followed by loading of 1 μl of the sample. The droplets were allowed to dry at room temperature. Spectra were calibrated and processed under computer control using the Applied Biosystems Data Explorer software.

Plasmid Constructions for Histidine-tagged Protein Overexpression

Constructs allowing the expression of N-terminal histidine-tagged WabH_R110 (His₆-WabH_R110), WabP_50/57 (His₆-WabP_50/57), WabN_R110 (His₆-WabN_R110), and WabN_50/57 (His₆-WabN_50/57) were based in pET28a. The wabH_R110 and wabP_50/57 genes were amplified from pGEM-T-wabH_R110 and pGEM-T-wabP_50/57 with primer pairs HR1 (5′-ACGCCATATGAAGGAACATATACTTTTTATT-3′)-HR2 (5′-ACGCGGATCCTAAAACGTGTTCTGACTCATTATT-3′) and PR1 (5′-ACGCCATATGAATCAACGACATATACTTTTT-3′)-PR2 (5′-ACGCGGATCCAACATGCTATCAAATTCACTTATC), respectively. The wabN_R110 and wabN_50/57 genes were amplified from pGEM-T-wabN_R110 and pGEM-T-wabN_50/57 with primer pairs NR1 (5′-ACGCCATATGAAAAAACCAGCATTTATTATCAC-3′)-NR2 (5′-ACGCGGATCCAAAGAGTGGGTTATAGAGATTTA-3′) and NRR1 (5′-ACGCCATATGAAAAAACCAGCATTTATTATC-3′)-NRR2 (5′-ACGCGGATCCATAGTGCACCCATTTTAATTTC), respectively. The four amplified DNA fragments were digested with NdeI (double underlined nucleotides in HR1, PR1, NR1, and NRR1) and HindIII (double underlined nucleotides in HR2, PR2, NR2, and NRR2) and ligated to pET28a digested with the same enzymes to obtain plasmids pET28a-wabH_R110, pET28a-wabP_50/57, pET28a-wabN_R110, and pET28a-wabN_50/57.

Preparation of Cell-free Extracts Containing Core LPS Biosynthetic Enzymes

E. coli BL21(λD3) was used to overexpress genes from the T7 promoter in pET28a-based plasmids. The cultures were grown in LB supplemented with kanamycin for 18 h at 37 °C, diluted 1:100 in fresh medium, grown until they reached an A_{600 nm} of about 0.5, induced by adding isopropyl-1-thio-β-d-galactopyranoside, and grown for an additional 2 h.

The cells from induced cultures were harvested, washed once with 50 mm Tris-HCl (pH 8.0), and frozen until used. To prepare the lysate cell pellets, they were resuspended in 50 mm Tris-HCl (pH 8.0) and sonicated on ice (for a total of 2 min using 10-s bursts followed by 10-s cooling periods). Unbroken cells, cell debris, and membrane fraction were removed by ultracentrifugation at 100,000 × g for 60 min. Protein expression was monitored by SDS-PAGE, and protein contents of cell-free extracts were determined using the Bio-Rad Bradford assay.

His-tagged Protein Purification

Cell-free lysates from E. coli BL21(λD3) harboring pET28a-wabH_R110, pET28a-wabP_50/57, pET28a-wabN_R110, and pET28a-wabN_50/57 were prepared as above using a phosphate-saline buffer (20 mm sodium phosphate buffer (pH 7.4) containing 500 mm NaCl). His₆-tagged proteins were purified by affinity chromatography on an FPLC system (Amersham Biosciences) using 1-ml HiTrap chelating HP columns (Amersham Biosciences) previously loaded with nickel sulfate and equilibrated with the phosphate-saline buffer. The columns were washed with phosphate-saline buffer containing 5 mm imidazole (10 column volumes), and the His-tagged proteins were eluted by a continuous gradient of 50–500 mm imidazole in phosphate-saline buffer. The buffer in the eluted proteins was exchanged into 50 mm Tris-HCl (pH 8.0), and the proteins were concentrated.

GlcNAc and GalNAc Transferase Assay

The transferase activity of His₆-WabH_R110 and His₆-WabP_50/57 were determined essentially as described (27). Briefly, reactions were performed in a total 0.1-ml volume at a final concentration of 50 mm Tris-HCl (pH 8.0) containing 10 mm MgCl₂ and 1 mm dithiothreitol. 1 mm UDP-GlcNAc or UDP-GalNAc was used as donor, and 0.001 mg of LPS from strain 52145ΔwabH was used as acceptor. The reactions were started by the addition of 0.02 mg of either His₆-WabH_R110 or His₆-WabP_50/57. After 2 h at 37 °C, the reactions were stopped by adding SDS-PAGE sample buffer and boiling for 10 min. The reaction products were separated in Tricine-SDS-PAGE.

Identical reactions using 0.25 μCi of UDP-[¹⁴C]GlcNAc or UDP-[¹⁴C]GalNAc (NEN Life Science; with specific activities of 51.2 and 55.5 mCi/mmol, respectively) were performed. Unincorporated radioactivity was removed by minigel filtration in 3 ml of Sephadex G-100, and the pooled and concentrated LPS fractions were separated as above by Tricine-SDS-PAGE. The dried gel was autoradiographed to visualize the incorporation of GlcNAc and GalNAc.

For quantitative analysis, the reactions were stopped by adding an equal volume of stop solution (27) and incubated on ice for 1 h. The LPS was recovered by filtration (0.45 μm) and washed, and the amount of radioactivity incorporated into LPS was measured in a scintillation counter.

Quantitative experiments using different concentrations of UDP-GlcNAc or UDP-GalNAc (1–800 μm) and acceptor LPS (1–800 μm) were performed. These reactions were stopped at different times (expressed in minutes), and the data from three independent experiments were used to determine the apparent K_m for acceptor LPS and enzyme substrate.

HexNAc-Core LPS-Deacetylase Assay

The ability of His₆-WabN_R110 and His₆-WabN_50/57 to catalyze the deacetylation of GlcNAc or GalNAc residue in LPS was assayed by using LPS from mutant 52145ΔwabH in reactions containing 0.4 mg each of His₆-WabH and His₆-WabN or His₆-WabP and His₆-WabN. Assay reactions using UDP-[¹⁴C]GlcNAc or UDP-[¹⁴C]GalNAc were carried out in a total of 0.1 ml at a final concentration of 50 mm Tris-HCl (pH 8.0) containing 10 mm MgCl₂, 1 mm dithiothreitol, 0.03 mg of 52145ΔwabH acceptor LPS, and 0.02 mg of each histidine-tagged protein. The reactions were started by the addition of 0.25 μCi of UDP-[¹⁴C]GlaNAc or UDP-[¹⁴C]GalNAc. Assays were performed at 37 °C for 2 h and were stopped by adding 2 volumes of 0.375 m MgCl₂ in 95% ethanol and cooling at −20 °C for 2 h. The LPS was recovered by centrifugation at 12,000 × g for 15 min and suspended in 100 μl of water. The LPS was precipitated with 2 volumes of 0.375 m MgCl₂ in 95% ethanol; this step was repeated three times to eliminate the unincorporated UDP-[¹⁴C]GlcNAc or UDP-[¹⁴C]GalNAc. The LPS was hydrolyzed by resuspension in 100 μl of 0.1 m HCl and heating to 100 °C for 48 h. The labeled residues from the hydrolyzed LPS samples were separated by thin layer chromatography (TLC) (Kieselgel 60; Merck) with n-butanol, methanol, 25% ammonia solution, water (5:4:2:1, v/v/v/v). The labeled residues were detected by autoradiography, using as standards [¹⁴C]GlcNAc, [¹⁴C]GalNAc, [¹⁴C]GlcN, and [¹⁴C]GalN.

RESULTS

P. mirabilis wa Gene Cluster

Similarly to the majority of Enterobacteriaceae in the P. mirabilis strains R110 and 51/57 a gene cluster (wa) was identified containing most of the genes involved in the biosynthesis of the core OS (16). The overall structure of these gene clusters was similar to the one found in strain HI4320 for which the whole genome sequence is available (28). Because strains R110 and 51/57 contain GlcN in the outer core LPS, we determined the organization of this cluster for strains 50/57 and TG83 containing GalN instead of GlcN (Fig. 1). Oligonucleotide pairs previously used in the amplification of DNA fragments of the cluster in strains R110 and 51/57 were used to amplify fragments from genomic DNA of strains 50/57 and TG83 and to determine its nucleotide sequence. Some of these oligonucleotide pairs did work, whereas others did not when using genomic DNA from strains 50/57 and TG83, suggesting that there are conserved and unconserved regions in the wa cluster among these four P. mirabilis strains. The nucleotide sequence of the amplified fragments was used to design further primers, allowing the amplification of DNA fragments encompassing the already sequenced ones. This strategy allowed the determination of the full nucleotide sequence of the wa gene cluster from P. mirabilis strains 50/57 and TG83 (19,155 bp). These two clusters showed more than 95% identity.

The comparative analysis of these sequences revealed that in all four strains the 5′-end of the cluster contains the hldD (ADP-d-glycero-d-manno-heptose epimerase), waaF (ADP-l-glycero-α-d-manno-heptosepyranose transferase II), and waaC (ADP-l-glycero-α-d-manno-heptosepyranose transferase I) genes. The 3′-end of the cluster contains the waaA (CMP-Kdo:lipidA Kdo bifunctional transferase and waaE (glucosyltransferase) genes adjacent to core OS-unrelated genes coaD (phosphopantetheine adenylyltransferase) (29) and fpg (formamidopyrimidine-DNA glycosylase) (30). In the middle of the cluster, genes waaQ (ADP-l-glycero-α-d-manno-heptosepyranose transferase III), wamA (ADP-d-glycero-d-manno-heptopyranose transferase), wabG (UDP-GalA transferase), wabH-like (UDP-GlcNAc transferase), and wabN (LPS-GlcNAc deacetylase) were identified (Fig. 1). The wa clusters of these two P. mirabilis strains shared additional characteristic features with those of previously studied ones (16, 28), such as the location of waaL (O-PS ligase) downstream from fpg and the presence of four additional genes (walM, -N, -O, and -R) encoding putative glycosyltransferases of unknown function (Fig. 1). The main difference between the 50/57 and TG83 wa clusters and those of R110, 51/57, and HI4320 is the inversion of the three-gene waaQ-wabH region (Fig. 1), explaining why some oligonucleotide pairs derived from strain R110 sequence failed in PCR amplification experiments using genomic DNA from either 50/57 or TG83 strains.

Comparison of the wa cluster with the known core OS structures for these four strains (Fig. 1) shows that additional genes located outside of the cluster should be necessary to complete the biosynthesis of this structure. Only one of these genes has been identified (wabO) (16). For strains 50/57 and TG83, it can be postulated that four and three additional genes should be located outside of the wa cluster, respectively.

Two Different WabH Homologues

Gene-deduced amino acid alignment among homologue proteins shared by the five P. mirabilis strains studied so far showed amino acid identities and similarities over 95 and 98%, respectively. The only exception was for WabH and WabH-like proteins, where two groups were found, WabH-like_50/57-WabH-like_TG83 and WabH_R110-WabH_51/57, with 100 and 99% identities, respectively. Both groups shared 60 and 89% levels of identity and similarity. An alignment between these proteins and those of K. pneumoniae strain C3 and 52145, where it has been shown that this protein transfers the GlcNAc from UDP-GlcNAc to core OS (27), was done to identify residues specific for the WabH-like_50/57-WabH-like_TG83 pair (Fig. 2). This analysis suggested that these two types of proteins could recognize different HexNAc substrates, GlcNAc or GalNAc, and thus we propose to name them WabH and WabP, respectively.

FIGURE 2.

Amino acid alignment of WabH and WabP proteins. Shown are WabH homologues from P. mirabilis strains R110 (Pm R110), 51/57 (Pm 51/57), and HI4320 (Pm HI4320); and K. pneumoniae strains 52145 (Kp 52145) and C3 (Kp C3); and WabP homologues from P. mirabilis 50/57 (Pm 50/57) and TG83 (Pm TG83). White letters on black background indicate identical amino acid residues, black letters on gray background indicate similar amino acid residues, and arrows denote specific amino acid residues for WabP from strains 50/57 and TG83. *, identical residues; :, conserved residues; ·, semi-conserved residues.

WabH Complementation Assays

To test the above hypothesis, we took advantage of the fact that the saccharide structure of the core OS from K. pneumoniae and that of P. mirabilis is identical up to the first outer core OS residue (GalA) (Fig. 1) to express independently wabH_R110 and wabP_50/57 in the non-polar mutant K. pneumoniae 52145ΔwabH. Oligonucleotide pairs BNf-BNr and 50/7-50/4 were used to amplify the 1137-bp wabH_R110 and 1104-bp wabP_50/57 genes, respectively. These amplicons were cloned in vector pGEM-T, and plasmids pGEM-T-wabH_R110 and pGEM-T-wabP_50/57 were electroporated into mutant 52145ΔwabH.

LPS was isolated from the transformed strains and analyzed in both SDS-PAGE and Tricine-SDS-PAGE. The core lipid A region of an LPS preparation from 52145ΔwabH migrated faster than that of wild type 52145, in agreement with a core OS missing the outer core GlcN (Fig. 3). LPS from strain 52145ΔwabH (pGEM-T-wabH_R110) migrated to the same level as that of wild type strain 52145, strongly suggesting that WabH_R110 is a functional homologue of WabH₅₂₁₄₅ catalyzing the transfer of GlcNAc from UDP-GlcNAc to core OS. LPS was extracted from strain 52145ΔwabH (pGEM-T-WabH_R110), and the OS fraction was obtained by mild acid hydrolysis (see “Experimental Procedures”). The GC-MS analysis of acetylated methyl glycosides from the OS fraction revealed the presence of Kdo, l-glycero-d-manno-heptopyranose, GalA, Glc, and GlcN. By contrast, WabP_50/57 do not appear to compensate for the deficiency in WabH₅₂₁₄₅ because LPS from strain 52145ΔwabH (pGEM-T-wabP_50/57) migrated to the same level as that of mutant 52145ΔwabH (Fig. 3).

FIGURE 3.

Polyacrylamide gels showing the migration of LPS from 52145ΔwabH mutant and its complementation. The LPS samples were separated on SDS-PAGE (A) and SDS-Tricine-PAGE (B) and visualized by silver staining. Shown are LPS samples from K. pneumoniae 52145 (lane 1), 52145ΔwabH (lane 2), 52145ΔwabH (pGEM-T-wabH_R110) (lane 3), 52145ΔwabH (pGEM-T-wabP_50/57) (lane 4), 52145ΔwabH (pGEM-T-wabP_50/57 pACYC-GNE_{P. mirabilis}) (lane 5), and 52145ΔwabH (pACYC-GNE_{P. mirabilis}) (lane 6).

Role of Gne in WabP Complementation

One possible reason for the absence of core OS modification induced by WabP_50/57 could be the absence of UDP-GalNAc in the 52145 strain. Analysis of the whole genome of K. pneumoniae strains 342 (GenBank^TM CP000964) and MGH 78578 (GenBank^TM CP000647) reveals the presence of one galE (UDP-Gal 4-epimerase) homologue for each genome. Generation of UDP-GalNAc from UDP-GlcNAc will require the presence of a gne (UDP-GalNAc 4-epimerase) homologue. Thus, we introduced the Aeromonas hydrophila gne containing plasmid pACYC-GNE_{A. hydrophila} (31) into strain 52145ΔwabH (pGEM-T-wabP_50/57). LPS from this strain migrated into an intermediate position between those of 52145 and 52145ΔwabH (supplemental Fig. S1).

A search for gne homologues in P. mirabilis revealed such gene in the species type strain ATCC 29906, whose whole genome sequence is being completed. This gne (in contig GG668582, GenBank^TM ACLE01000000) was amplified and cloned to obtain pACYC-GNE_{P. mirabilis}. The LPS isolated from 52145ΔwabH (pGEM-T-wabP_50/57 pACYC-GNE_{P. mirabilis}) also migrated between those of 52145 and 52145ΔwabH (Fig. 3).

To determine the core LPS changes produced by expression of both gne and wabP_50/57 in the 52145ΔwabH background, the LPS was extracted from strain 52145ΔwaaL wabH (pGEM-T-wabP_50/57 pACYC-GNE_{P. mirabilis}) by the phenol, chloroform, and petroleum ether method (22), and the core OS fraction was isolated by mild acid hydrolysis. Compositional analysis of this fraction by GC-MS of acetylated methyl glycosides revealed the presence of Kdo, l-glycero-d-manno-heptopyranose, GalA, Glc, and GalN. Comparative MALDI-TOF analysis of this core OS fraction with that of strain 52145ΔwaaL wabH (15) revealed major signals at 1488.19 and 1470.27 m/z, corresponding to a Kdo-Hep3-Hex-HexA2-HexN and its anhydro form. This is in agreement with a core extending up to the second outer OS residue (GalN) (Fig. 4A). Analysis of the core OS fraction from strain 52145ΔwaaL wabH (pGEM-T-wabP_50/57 pACYC-GNE_{A. hydrophila}) resulted in the same chemical composition and major m/z signals (supplemental Fig. S2). A similar analysis of the OS fraction from strain 52145ΔwaaL wabH (pGEM-T-wabH_R110) showed major signals at 1812.56 and 1650.54 corresponding to Kdo-Hep3-Hex3-HexA2-HexN and Kdo-Hep3-Hex2-HexA2-HexN (Fig. 4B). These structures are in agreement with a full type 2 K. pneumoniae core OS (13). Other signals attributable to OS with terminal Kdo modifications or sodium abducts are shown in Fig. 4.

FIGURE 4.

Positive ion MALDI-TOF of acid-released core LPS oligosaccharides. The spectrum of OS from K. pneumoniae 52145ΔwabH was previously reported (15). Shown are spectra of OS isolated from K. pneumoniae 52145ΔwabH (pGEM-T-wabP_50/57 pACYC-GNE) (A) and K. pneumoniae 52145ΔwabH (pGEM-T-wabH_R110) (B). Schematic structures of the most representative compounds are shown in the insets. Signals that are 18, 46, and 88 Da, respectively, below the described pseudomolecular ion are attributable to OS with terminal Kdo containing a ring double bond (−18 Da), a ketone at C-1 (−44 Da) or ring fragmentation (−88 Da). These artifacts have been described for LPS samples that are hydrolyzed in presence of acetic acid (36).

A search for gne homologues in P. mirabilis revealed such a gene in the species type strain ATCC 29906, whose whole genome sequence is being completed. This gne (in contig GG668582, GenBank^TM ACLE01000000) was amplified and cloned to obtain pACYC-GNE_{P. mirabilis}. The LPS isolated from 52145ΔwabH (pGEM-T-wabP_50/57 pACYC-GNE_{P. mirabilis}) behaved as that from 52145ΔwabH (pGEM-T-wabP_50/57 pACYC-GNE) in Tricine-SDS-PAGE. In addition, no differences in chemical composition or in major m/z signals could be found between these two core LPS samples (data not shown).

WabH and WabP in Vitro Activity

In vitro enzymatic assays were performed (27) to test if the WabH_R110 and WabP_50/57 are specific for the transfer of GlcNAc and GalNAc, respectively. This assay measured the amount of radiolabeled GlcNAc or GalNAc incorporated into acceptor LPS from UDP-[¹⁴C]GlcNAc or UDP-[¹⁴C]GalNAc. LPS from mutant K. pneumoniae 52145ΔwabH was used as acceptor. Recombinant plasmids expressing N-terminal histidine-tagged WabH_R110 (His₆-WabH_R110) and His₆-WabP_50/57 were constructed to facilitate the purification of the enzymes used in the assays by affinity chromatography.

Qualitative assays based on autoradiography of LPS reaction samples after Tricine-SDS-PAGE showed that His₆-WabH_R110 was able to transfer radiolabeled GlcNAc from UDP-GlcNAc to acceptor LPS, whereas little radioactivity was incorporated into acceptor LPS when using UDP-GalNAc as donor. By contrast, His₆-WabP_50/57 directed the incorporation of mainly GalNAc and little GlcNAc (Fig. 5).

FIGURE 5.

HexNAc transferase qualitative in vitro analysis by Tricine-SDS-PAGE of product reactions. Reactions contained LPS from 52145ΔwabH, 0.25 μCi of UDP-[¹⁴C]GlcNAc (lanes 1–4) or UDP-[¹⁴C]GalNAc (lanes 5–8), and the indicated histidine-tagged proteins. After a 2-h reaction at 37 °C, LPS was recovered, washed, and separated on Tricine-SDS-PAGE. The gel was stained (A), dried, and autoradiographed (B). Lanes 1 and 5, control reactions without the addition of the enzyme; lane 2, reactions with His₆-WabN_R110; lanes 3 and 7, reactions with His₆-WabH_R110; lane 6, reactions with His₆-WabN_50/57; lanes 4 and 8, reactions with His₆-WabP_50/57.

Quantitative experiments were used to determine the apparent kinetic parameters of WabH_R110 and WabP_50/57. In one set of experiments, the concentration of UDP-GlcNAc or UDP-GalNAc was maintained constant at 200 μm, and a range of acceptor LPS from mutant 52145ΔwabH was used (1–800 μm). In the other set, the acceptor LPS was held constant (200 μm), and different levels of UDP-GlcNAc or UDP-GalNAc were used (1–800 μm). The data allowed us to determine the apparent K_m of His₆-WabH_R110 and His₆-WabP_50/57 for acceptor LPS, UDP-GlcNAc, and UDP-GalNAc (Table 2). In reactions containing 100 nm enzyme preparations, the apparent K_m for acceptor LPS was essentially the same for both enzymes. By contrast, the apparent K_m of His₆-WabP_50/57 for UDP-GalNAc was 10-fold lower than that for UDP-GlcNAc, whereas the His₆-WabH_R110 behaved in a reverse way (Table 2).

TABLE 2.

Kinetics of His₆-WabH_R110 and His₆-WabP_50/57

The Prism GraphPad program was used to calculate the apparent Michaelis-Menten parameters from three replicate experiments.

Donor	Enzyme	Kcat	K′m UDP-HexNAc	K′m LPS	Kcat/K′m UDP-HexNAc	Kcat/K′m LPS
		min−1	μm	μm
UDP-GlcNAc	His6-WabHR110	26 ± 5	34 ± 3	13 ± 3	0.76	2.0
	His6-WabP50/57	4 ± 2	120 ± 15	14 ± 2	0.03	0,28
UDP-GalNAc	His6-WabHR110	3 ± 1	130 ± 21	13 ± 5	0.02	0,23
	His6-WabP50/57	27 ± 6	38 ± 4	12 ± 4	0.71	2.2

WabN Complementation Studies

The compositional analysis of the OS fraction from strain 52145ΔwabH with pGEM-T-wabP_50/57 and pACYC-GNE or pACYC-GNE_{P. mirabilis} plasmids revealed the presence of GalN (see above), suggesting that the WabN deacetylase from strain 52145 is able to deacetylate both GlcNAc- and GalNAc-containing OS. The comparative analysis of the known WabN homologues from K. pneumoniae and P. mirabilis revealed that they share around 65% amino acid identity, whereas among those of P. mirabilis, identity is more than 95%. The WabN proteins from strains 50/57 and TG83 were identical and shared levels of residue identity of 99.7, 98.4, and 95,9% with those of strains R110, HI4320, and 51/57, respectively. These levels of identity suggest that they will have the same role.

Complementation experiments confirmed this hypothesis. The LPS from K. pneumoniae 52145ΔwabN is devoid of the last two outer core residues and O-PS and presents GlcNAc instead of GlcN (15). Introduction of pGEM-T-wabN_R110 or pGEM-T-wabN_50/57 restores wild type 52145 LPS migration and O-PS production (Fig. 6). In addition, an R110ΔwabN mutant was constructed, and LPS from this mutant was devoid of O-PS and migrated faster than that of wild-type R110. This mutant was fully complemented by wabN homologues from strain 50/57 and K. pneumoniae 52145 as judged by LPS migration in Tricine-SDS-PAGE (data not shown). These results suggest that both genes codify for enzymes able to deacetylate GlcNAc containing core OS.

FIGURE 6.

Polyacrylamide gels showing the migration of LPS from 52145ΔwabN mutant and its complementation. The LPS samples were separated on SDS-PAGE (A) and SDS-Tricine-PAGE (B) and visualized by silver staining. Shown are of LPS samples from K. pneumoniae 52145 (lane 1), 52145ΔwabN (lane 2), 52145ΔwabN (pGEM-T-wabN_R110) (lane 3), and 52145ΔwabN (pGEM-T-wabN_50/57) (lane 4).

WabN Deacetylates Both GlcNAc and GalNAc LPS in Vitro

An in vitro enzymatic assay measuring the amount of radiolabeled GlcNAc or GlcN incorporated into acceptor LPS from UDP-[¹⁴C]GlcNAc after acid hydrolysis and TLC separation of the radiolabeled residues was performed (15). LPS from mutant 52145ΔwabH was used as an acceptor with combinations of histidine-tagged enzymes WabH, WabP, and WabN. In reactions using UDP-[¹⁴C]GlcNAc as donor and both His₆-WabH_R110 and His₆-WabN_R110 or His₆-WabN_50/57, radiolabeled GlcN was detected (Table 3). Replacing His₆-WabH_R110 by His₆-WabP_50/57 did not allow a substantial detection of GlcN. In reactions using UDP-[¹⁴C]GalNAc as donor, His₆-WabP_50/57 and His₆-WabN_50/57 or His₆-WabN_R110 radioabeled GalN was detected (Table 3). In these reactions, replacement of His₆-WabP_50/57 by His₆-WabH_R110 resulted again in substantial GalN reduction detection (Table 3). These results strongly suggest that in P. mirabilis, there are two specific HexNAc transferases (WabH and WabP), whereas WabN enzymes are able to deacetylate both GlcNAc- and GalNAc-containing OS.

TABLE 3.

In vitro analysis of LPS-HexNAc deacetylation

Reactions contained LPS from K. pneumoniae 52145ΔwabH, 0.25 μCi of UDP-[¹⁴C]GlcNAc or UDP-[¹⁴C]GalNAc, and the indicated histidine-tagged proteins. After a 2-h reaction at 37 °C, LPS was recovered by centrifugation, precipitated, washed, and hydrolyzed in 100 μl of 0.1 m HCl and applied to a TLC plate. Non-radioactive controls were used to localize the positions of GlcN, GalN, GlcNAc, and GalNAc. Radioactive spots were scraped and suspended in scintillating liquid, and the radioactivity was measured.

Protein	UDP-GlcNAc radioactivity		UDP-GalNAc radioactivity
	GlcNAc	GlcN	GalNAc	GalN
	cpm		cpm
WabHR110 + WabNR110	<100	2670	<100	<100
WabHR110 + WabN51/57	<100	2345	<100	<100
WabP50/57 + WabNR110	<100	<100	<100	2565
WabP50/57 + WabN50/57	<100	<100	<100	2479

gne, wabH, and wabP Distribution

In order to determine the distribution of the genes determining the incorporation of GlcN/GalN into the core LPS, three pairs of oligonucleotides were designed to amplify internal fragments of gne (694 bp), wabH (815 bp), and wabP (417 bp). Internal gne fragments were amplified from strains containing GalN in their core LPS, such as 50/57, TG83, and 15/57, as expected, but also from two strains, R110 and OXK, containing GlcN instead of GalN (Fig. 7A). The core LPS of strain OXK contains a GalNAc residue in addition to GlcN (11), and the O-PS of strains R110 and OXK also contain GalNAc (9). Among strains of unknown LPS structure, gne was found in strain ATCC 29066 but not in CECT 170 (Fig. 7A). Thus, gne will be present in strains requiring the synthesis of UDP-GalNAc to be incorporated into either core or O-antigen LPS.

FIGURE 7.

PCR-based analysis of the distribution of gne, wabH, and wabP genes among P. mirabilis strains. Shown is agarose gel electrophoresis of PCR-amplified products generated from template DNA obtained from strains R110 (lane 1), 50/57 (lane 2), 51/57 (lane 3), TG83 (lane 4), OXK (lane 5), 14/57 (lane 6), CECT170 (lane 7), and ATCC 29066 (lane 8). Amplification products obtained with oligonucleotide pairs designed to amplify internal regions of gne (GneIntFw-GneInt-Rev) (A), wabH (wabHIntFw-WabHIntRev) (B), and wabP (WabpIntFw-WabPIntRv) (C) are shown.

A wabH internal fragment was amplified from GlcN-containing core LPS strains, such as R110, 51/57, and OXK (11), as well as from strains CECT 170 and ATCC 29906 of unknown core LPS structure (Fig. 7B). A wabP internal fragment was amplified from GalN-containing core LPS strains, such as 50/57, TG83, and 14/57 (11) (Fig. 7C). After testing 25 different P. mirabilis strains, in no case was it possible to detect the simultaneous presence of wabH and wabP.

DISCUSSION

Sequence similarity suggested that, depending on the nature of the HexN in outer core OS, two different types of HexNAc transferases are present in P. mirabilis strains. Taking advantage of the identity between P. mirabilis and K. pneumoniae core OS sugar backbone up to the first outer core residue, we performed in vivo experiments using K. pneumoniae non-polar wabH mutant LPS as acceptor substrate. This new approach has the advantage that we look for a positive trait as the addition to the mutant LPS of a particular residue instead of the mutagenesis studies to determine the function of core LPS biosynthetic genes usually employed. This is only possible, as in our case, if proper surrogate LPS acceptor is available. By this methodology, we were able to identify a GlcNAc transferase (WabH_R110) in strain R110, containing GlcN as the second outer core residue. By contrast, the WabP_50/57 protein from a strain containing GalN as the second outer core residue was able to transfer GalNAc to K. pneumoniae core LPS only when the gne (UDP-GalNAc 4-epimerase) gene was also provided. An R110ΔwabH mutant was complemented (presence of O-antigen LPS) by wabH from strain 51/57 but not by wabP from strain 50/57 (absence of O-antigen LPS) (data not shown).

In vitro experiments using the purified enzymes and UDP-GlcNAc or UDP-GalNAc confirmed the GlcNAc and GalNAc transferase nature for WabH_R110 and WabP_50/57, respectively. As inferred from their functions and sequence similarity, both glycosyltransferases belong to the GT-B motif-containing glycosyltransferase family 4 according to the Carbohydrate Active Enzyme (CAZy) classification. This family contains also WabH homologues from K. pneumoniae and S. marcescens (32) (see the CAZy Web site).

The core OS structures of E. coli share a common substructure up to the first outer core residue (GlcI) (33). In E. coli strains with core OS types K-12, R1, R2, and R4, the WaaO glucosyltransferase links a second Glc residue (GlcII) to GlcI by an α1,2 linkage (33). In strains with R3 core type, the WaaI galactosyltransferase links a Gal residue to GlcI with the same α1,2 linkage (33). Both WaaO and WaaI belong to the GT-A fold containing CAZy glycosyltransferase family 8. An alignment (data not shown) between these two glycosyltransferases (WaaO, Q8KMW9; WaaI, Q9ZIT4) revealed that they share 51 and 80% amino acid residue identity and similarity, respectively. These levels of identity/similarity are similar to those shared by the P. mirabilis N-acetylglucosaminyl (WabH_R110) and N-acetylgalactosaminyl (WabP_50/57) transferases. Thus, it appears that glycosyltransferases involved in the transfer of sugar epimer residues to the same LPS acceptor can be differentiated by their levels of identity/similarity.

Once GlcNAc is incorporated from UDP-GlcNAc into LPS, the WabN deacetylase converts this residue to GlcN, as previously shown in K. pneumoniae (15). Both complementation experiments and in vitro assays with purified WabN proteins show that in P. mirabilis, these proteins can deacetylate both GlcNAc and GalNAc containing nearly identical core LPS. Furthermore, an R110ΔwabN mutant was fully complemented by WabN homologues from K. pneumoniae and P. mirabilis 50/57. The fact that the WabN deacetylases are able to act on carbohydrate backbones nearly identical but with terminal GlcNAc or GalNAc residues could represent an efficient system to reduce unnecessary biodiversity.

From the known core OS structures from the genus Proteus (11), we can predict that wabH and wabP homologues should be present in P. mirabilis strains containing GlcN and GalN in their core LPS, respectively. Two pairs of oligonucleotides specific for wabH and wabP confirmed this prediction (Fig. 7, B and C). These results suggest that a simple amplification test can be used to predict the HexN nature in P. mirabilis core LPS. This diagnostic amplification indicates the presence of GlcN in the core LPS of strains for which the chemical structures have not been determined, such as CECT 170, HI4320, and ATCC 29906. In agreement with the test, the analysis of the genome sequence available for strains HI4320 (whole genome sequence determined) (28) and ATCC 29906 (genome sequencing in progress) (ACLE01000000) confirmed the presence of wabH but not wabP.

By contrast, gne required for UDP-GalNAc generation from UDP-GlcNAc is present in strains containing GalN in their core LPS but also in strains containing GlcN when these strains present GalNAc in their O-PS. In P. mirabilis ATCC 29906, the gne is located in a gene cluster similar to those putatively involved in O-PS (contig GG668582, GenBank^TM ACLE01000000) (34). However, there are examples of P. mirabilis strains that should require gne for its O-PS biosynthesis where this gene was not found inside the O-PS biosynthetic cluster (34). Thus, gne cannot be used to predict the presence of GalN or GlcN in the core LPS. Nevertheless, the absence of gne is fully correlated with presence of wabH and thus of GlcN in the LPS core of the strains tested.

We conclude that although the incorporation of GlcN into core LPS requires N-acetylglucosaminyltransferase (WabH) and LPS:GlcNAc deacetylase (WabN), the incorporation of GalN requires three enzymes: Gne epimerase for the generation of UDP-GalNAc from UDP-GlcNAc, N-acetylgalactosaminyltransferase (WabP), and LPS:GalNAc deacetylase. Of these three enzymes, only the LPS:HexNAc deacetylase (WabN) is common to the incorporation of GlcN and GalN into core LPS.