Biosynthesis of uronamide sugars in Pseudomonas aeruginosa O6 and Escherichia coli O121 O antigens


  • Jerry D. King,

    1. Department of Molecular and Cellular Biology, University of Guelph, Ontario N1G 2W1, Canada.
    Search for more papers by this author
  • Evgeny Vinogradov,

    1. Institute for Biological Sciences, National Research Council, Ottawa, Ontario K1A 0R6,Canada.
    Search for more papers by this author
  • Vanessa Tran,

    1. Department of Molecular and Cellular Biology, University of Guelph, Ontario N1G 2W1, Canada.
    Search for more papers by this author
    • Present address: Department of Molecular Genetics, University of Toronto, Toronto, Ontario M5S 1A8, Canada.

  • Joseph S. Lam

    Corresponding author
    1. Department of Molecular and Cellular Biology, University of Guelph, Ontario N1G 2W1, Canada.
      E-mail; Tel. (+1) 519 824 4120 extension 53823; Fax (+1) 519 837 1802.
    Search for more papers by this author

E-mail; Tel. (+1) 519 824 4120 extension 53823; Fax (+1) 519 837 1802.


The major component of the outer leaflet of the outer membrane of Gram-negative bacteria is lipopolysaccharide (LPS). The outermost domain of LPS is a polysaccharide called O antigen. Pseudomonas aeruginosa establishes biofilms on wet surfaces in a wide range of habitats and mutations in O-antigen biosynthesis genes affect bacterial adhesion and the structure of these biofilms. The P. aeruginosa O6 O antigen contains a 2-acetamido-2-deoxy-d-galacturonamide (d-GalNAcAN) residue. O-antigen biosynthesis in this serotype requires the wbpS gene, which encodes a protein with conserved domains of the glutamine-dependent amidotransferase family. Replacement of conserved amino acids in the N-terminal glutaminase conserved domain of WbpS inhibited O-antigen biosynthesis under restricted-ammonia conditions, but not in rich media; suggesting that this domain functions to provide ammonia for O-antigen biosynthesis under restricted-ammonia conditions, by hydrolysis of glutamine. Escherichia coli O121 also produces a d-GalNAcAN-containing O antigen, and possesses a homologue of wbpS called wbqG. An E. coli O121 wbqG mutant was cross-complemented by providing wbpS in trans, and vice versa, showing that these two genes are functionally interchangeable. The E. coli O121 wbqG mutant O antigen contains 2-acetamido-2-deoxy-d-galacturonate (d-GalNAcA), instead of d-GalNAcAN, demonstrating that wbqG is specifically required for biosynthesis of the carboxamide in this sugar.


Pseudomonas aeruginosa is a Gram-negative microorganism which causes important life-threatening infections in compromised animals and plants. It accounts for approximately one in 10 reported cases of hospital-acquired infections (Emori and Gaynes, 1993). This bacterium is metabolically diverse, enabling it to survive in a wide range of wet environments. It thrives in natural waters such as lakes and rivers, as well as in whirlpools and hot tubs (Mena and Gerba, 2009), and readily attaches to wet surfaces, establishing complex, differentiated microbial communities known as biofilms. Biofilms constitute the dominant mode of microbial growth in nature, and promote bacterial persistence in hostile environments such as in the context of an infection under antibiotic treatment (Costerton et al., 1999). Pseudomonas aeruginosa is one of the most intensively studied biofilm-forming organisms.

The cell surfaces of Gram-negative bacteria are coated with a complex glycolipid called lipopolysaccharide (LPS). This is the major molecule in the outer leaflet of the outer membrane and mediates many interactions between bacterium and environment. The LPS structure in P. aeruginosa (Knirel et al., 2006) contains three domains: lipid A, core and O antigen. Lipid A is a fatty acyl chain-substituted, phosphorylated glucosamine dimer, which anchors LPS into the lipid bilayer. The core domain is a complex, branched oligosaccharide which contains a number of non-carbohydrate substituents and is covalently bonded to lipid A. Normally, a proportion of LPS on the cell surface consists of just these two domains, and this structure is known as lipid A–core. O antigen is a carbohydrate polymer and is the outermost domain of LPS. On the basis of O-antigen serology, P. aeruginosa strains can be classified as one of 20 major O serotypes according to the International Antigenic Typing Scheme (IATS) scheme (Liu et al., 1983; Liu and Wang, 1990). The chemical structure of O antigen in each of these serotypes has been elucidated by Knirel and colleagues (2006) and the gene clusters encoding O-antigen biosynthesis have been cloned and sequenced (Burrows et al., 1996; Bélanger et al., 1999; Raymond et al., 2002). O6 is consistently one of the predominant P. aeruginosa serotypes among clinical isolates (Vieu et al., 1984; Bert and Lambert-Zechovsky, 1996; Pirnay et al., 2002; 2003).

The O antigen of P. aeruginosa is important in the context of human and animal infection (Cryz Jr et al., 1984; Dasgupta et al., 1994), and is also a major factor influencing the efficiency of bacterial adherence to surfaces (Makin and Beveridge, 1996; Beveridge et al., 1997), which is the initial step in biofilm formation. Production of O antigen is a regulated process in P. aeruginosa, with reduction in O antigen constituting a mechanism or consequence of cell differentiation during biofilm maturation (Beveridge et al., 1997; Lau et al., 2009). Our group has recently shown that mutations affecting LPS core and O-antigen biosynthesis also change mechanical and structural properties of P. aeruginosa biofilms (Lau et al., 2009).

Escherichia coli is a normal component of the human intestinal microbiota, but pathogenic strains are associated with outbreaks of diarrhoeal disease, often initiated by consumption of contaminated food or water (reviewed in Robins-Browne and Hartland, 2002). Escherichia coli serotype O121 strains have been classified as enteroinvasive E. coli (EIEC) and as enterohaemorrhagic E. coli (EHEC), being found associated with shigellosis-like diarrhoeal disease (Hiruta et al., 1991; Matsushita et al., 1993) and from patients with haemorrhagic colitis (Mccarthy et al., 2001; Stock et al., 2001; Akiba et al., 2005) respectively.

The O antigens of the P. aeruginosa O6 serogroup contain a 2-acetamido-2-deoxy-d-galacturonamide (d-GalNAcAN) residue, which in some strains is 3-O-acetylated (d-GalNAc3OAcAN) (Knirel et al., 2006). In the IATS O6 reference strain, all of these residues are 3-O-acetylated and the O repeat also contains the closely related 2-deoxy-2-formamido-d-galacturonic acid (d-GalNFoA) (Fig. 1A). In other P. aeruginosa O6 strains this sugar can also be present in the uronamide form (d-GalNFoAN) (Vinogradov et al., 1987). To our knowledge, the pathway for biosynthesis of uronamide sugars has not been elucidated in any organism, but it has been proposed that the d-GalNAcAN in the P. aeruginosa O6 O antigen is formed by amidotransfer to the carboxyl group at carbon 6 of 2-acetamido-2-deoxy-d-galacturonic acid (d-GalNAcA) (Bélanger et al., 1999). The substrate for this reaction is presumably the sugar-nucleotide O-antigen precursor, UDP-d-GalNAcA, which, in P. aeruginosa, is synthesized from UDP-d-GlcNAc in reactions catalysed by WbpO and WbpP (Fig. 1B) (Creuzenet et al., 2000; Zhao et al., 2000; Miller et al., 2008). It has been proposed (Prior et al., 2003; Feng et al., 2004) that the amidotransferase enzyme which forms the uronamide carboxamide group may be encoded by the wbpS gene from the P. aeruginosa O-antigen biosynthesis cluster. In support of this hypothesis, E. coli O121 and Francisella tularensis ssp. tularensis strain Schu S4, which both produce O antigens containing d-GalNAcAN (Parolis et al., 1997; Prior et al., 2003), also have homologues of wbpO, wbpP and wbpS in their O-antigen biosynthesis clusters (Fratamico et al., 2003; Prior et al., 2003). Escherichia coli O121 wbqA, wbqB and wbqG encode proteins sharing 71%, 68% and 55% sequence identity with P. aeruginosa WbpO, WbpP and WbpS respectively. The F. tularensis homologues share 52%, 52% and 44% identity respectively. Furthermore, Bordetella bronchiseptica, which produces an O polysaccharide containing three different uronamide sugars (Preston et al., 2006), has three wbpS homologues in the O-antigen biosynthesis cluster, each encoding a protein sharing 27–31% identity with the P. aeruginosa sequence. Shigella dysenteriae 7 produces an O antigen and has an O-antigen biosynthesis cluster which are almost identical to those of E. coli O121 (Knirel et al., 1988; Feng et al., 2004).

Figure 1.

A. O-antigen repeating unit structures for P. aeruginosa IATS serotype O6 reference strain and E. coli O121.
B. Proposed pathway for the biosynthesis of UDP-2-acetamido-2-deoxy-d-galacturonamide (UDP-d-GalNAcAN). The reactions catalysed by WbpO and WbpP are known; the WbpS-catalysed reaction is hypothetical.

The wbpS gene encodes a protein 627 amino acids long, which contains conserved domains of the glutamine-dependent amidotransferase class II protein family. This important protein family contains enzymes which catalyse the incorporation of nitrogen in the biosynthesis of amino acids (e.g. asparagine synthetase B, glutamate synthetase), hexosamines (glucosamine-6-phosphate synthase), and purine nucleotides [phosphoribosylpyrophosphate (PRPP) amidotransferase] and has been reviewed (Massière and Badet-Denisot, 1998; Mouilleron and Golinelli-Pimpaneau, 2007). Of the well-characterized members of this family, asparagine synthetase B (glutamine hydrolysing) encoded by the asnB gene is the most closely related to WbpS. blast algorithms align the WbpS and AsnB sequences along their entire lengths, sharing 30% and 26% identity between WbpS and AsnB from Bacillus subtilis (Yoshida et al., 1999) and E. coli respectively. The E. coli AsnB protein has been crystallized (Larsen et al., 1999). The N-terminal subunit of asparagine synthetase B hydrolyses glutamine, producing ammonia which is channelled to the active site of the C-terminal synthetase module through a molecular tunnel (Mouilleron and Golinelli-Pimpaneau, 2007). In the C-terminal subunit, hydrolysis of ATP activates the acceptor molecule (aspartate) for nucleophilic attack by ammonia. Members of the class II subfamily of glutamine amidotransferases have a conserved cysteine at the very N-terminus of the peptide chain. In bacteria, this cysteine is exposed at the terminus by cleavage of the translation-initiated methionine by methionylaminopeptidase (Hirel et al., 1989). It has been proposed that the terminal position is important for the catalytic function of the cysteine in attack on the glutamine carbonyl, possibly through involvement of the protein terminal amine group in enhancing the nucleophilicity of the thiol side-chain (Isupov et al., 1996).

Here, we report our experimental investigation of the hypothesis that P. aeruginosa WbpS and its E. coli homologue, WbqG, function in the biosynthesis of O-antigen uronamide sugars.

Results and discussion

Knockout mutation of wbpS in P. aeruginosa IATS O6

To evaluate the function of wbpS in O-antigen biosynthesis in P. aeruginosa O6, we mutated the chromosomal copy of the gene in the IATS O6 reference strain by allele-exchange mutagenesis. Analysis of the LPS phenotype of this mutant by sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting indicated that the mutation abrogated O-antigen biosynthesis, as no material was present in the wbpS mutant whole-cell lysate which bound the anti-O6 monoclonal antibody (Fig. 2C). Silver-stained analysis of these LPS gels also showed that the wbpS mutant had lost the ability to produce the species consisting of lipid A–core plus one O unit, known as ‘core+1’ (Fig. 2A and B). The mutant phenotype could be complemented by providing wbpS in trans on a shuttle vector, confirming that this phenotype is a specific consequence of the wbpS mutation and not due to polar effects or chromosomal mutation at a second site. Therefore, in P. aeruginosa IATS O6 wbpS is required for O-antigen biosynthesis. The wbpS mutant could also be cross-complemented by introduction of the E. coli homologue wbqG, albeit with slightly lower intensity of the O-antigen bands (Fig. 2). This shows that WbqG is a functional homologue capable of catalysing the same reaction as WbpS in O-antigen biosynthesis.

Figure 2.

SDS-PAGE analysis of the LPS phenotype of the P. aeruginosa wbpS mutant. LPS was visualized by silver staining (A and B), or Western blotting using a monoclonal antibody (mAb O25G3D6) specific for O antigen (C). The wbpS mutation abrogated production of O antigen and of the core+1 LPS. Both of these LPS features were restored by complementation of the mutation by the presence of wbpS or wbqG on a plasmid, but were not restored by transformation with the empty vector.

The wbpS slow-growth phenotype

We observed that when incubated on agar media, the wbpS mutant grew more slowly, producing smaller colonies than wild type after incubation for 24 or 48 h. The mutant also grew more slowly in liquid media. On agar media, faster-growing derivative clones could be identified as larger colonies which appeared after passage of the wbpS mutant strain (not shown). Eight of these fast-growing clones (called wbpS big 1–8) were isolated, each of which produced a colony size comparable with the wild type. We analysed the LPS from these clones by SDS-PAGE and silver staining, and by examining the banding patterns in the lower part of the gel we could categorize these clones as exhibiting four distinct LPS profiles (Fig. 3). Three of the LPS profiles, exemplified by wbpS big 1, 2 and 5, showed band shifts consistent with truncations of the LPS core oligosaccharide. Strains wbpS big 3, 6, 7 and 8 exhibited LPS profiles which closely resembled that of wbpS big 2. One of the clones, wbpS big 4, had an LPS profile indistinguishable from the parental wbpS mutant strain. In Western blot analysis, none of the wbpS big 1–8 samples bound O antigen-specific antibodies (data not shown).

Figure 3.

The P. aeruginosa wbpS mutant throws off variant clones with altered core LPS structures. Pseudomonas aeruginosa wbpS mutants grow more slowly than the wild type and on passage, faster-growing clones could be readily identified because they produced bigger colonies than their peers after 24 or 48 h. When the LPS from eight of these faster-growing clones was examined by SDS-PAGE and silver staining, seven showed alteration in the structure of core. One of the clones, wbpS big 4, had an LPS profile indistinguishable from the parental strain, wbpS. Arrows indicate the positions of O antigen-containing bands, lipid A–core and lipid A–core plus one O unit in the wild-type LPS profile.

It has previously been reported that chromosomal mutations in the P. aeruginosa wbpM gene resulted in small and large colonies (Dean and Goldberg, 2002). The conserved wbpM gene is required for O-antigen biosynthesis in most P. aeruginosa serotypes (Burrows et al., 2000), and mutation of this gene in P. aeruginosa O5, O6 and O11 backgrounds results in a mixture of colony morphologies. There appears to be a selective pressure for wbpM mutant strains to develop second-site mutations, some of which result in alteration of the LPS core structure (Dean and Goldberg, 2002). The slow-growing wbpS mutant appears to be subject to a similar pressure to develop second mutations. When mid-log phase cells were examined under light and electron microscopy, we observed that wbpS mutant cultures contain a proportion of abnormally short cells (Fig. 4). While this does not pinpoint the precise mechanism of the wbpS growth defect, this observation does suggest that there is a perturbation of cell envelope biogenesis and/or maintenance in this mutant. One possibility is that abortive O-unit synthesis in the wbpS mutant traps the undecaprenol-phosphate lipid carrier required for both O-antigen and peptidoglycan biosynthesis (Siewert and Strominger, 1967). There is as yet no evidence, however, to specifically validate this hypothesis.

Figure 4.

Pseudomonas aeruginosa wbpS cells are shorter than those of wild type. DIC light microscopy of wild-type (A) and wbpS (B) P. aeruginosa shows the distribution of cell sizes present in cell populations. The remaining panels show a typical wild-type cell (C) and a short cell characteristic of the wbpS mutant (D) in transmission electron micrographs of negatively stained whole mounts.

Given the variety of LPS core phenotypes exhibited by the wbpS big 1–8 clones, it appears that a number of different second-site mutations took place, each of which restored a wild type-like growth rate to the wbpS mutant. The instability and unpredictability of the wbpS knockout strain made it difficult to work with. Therefore, wbpS big 4 was used in subsequent experiments as it exhibits a stable wbpS mutant LPS phenotype, including the important property that in this strain too, the LPS defect could be complemented by wbpS expressed in trans (see below).

Site-directed mutagenesis of the WbpS glutaminase conserved domain

As a member of the class II glutamine-dependent amidotransferase family, WbpS contains a predicted N-terminal glutamine-hydrolysing domain. As mentioned before, asparagine synthetase B (AsnB) from E. coli is a homologue of WbpS. From the crystal structure of a Cys1Ala mutant of AsnB, it is known which amino acid side-chains interact with glutamine in the glutaminase domain, and all of these are conserved in the WbpS sequence (Fig. 5) (and in the WbqG sequence). Therefore, to validate the sequence-based annotation of WbpS as a glutamine-hydrolysing amidotransferase, we mutated the wbpS complementation plasmid, producing five vectors each encoding a mutant protein in which one of these conserved amino acids is replaced by one with a hydrophobic side-chain. The ability of these point-mutant alleles to complement the wbpS mutant in trans was then tested, using the intensity of O-antigen bands in silver-stained SDS-PAGE and Western blot as a measure of the efficiency of complementation. When the bacteria were grown in LB, all five of the point-mutant wbpS alleles were able to restore O-antigen production to the mutant strain, although in the experiments with E81A and D106L point-mutant constructs, the quantity of O antigen was considerably diminished compared with the wild type (Fig. 6). To account for the full complementation of the wbpS mutant by three of these point-mutant genes under these conditions, we hypothesized that in common with many glutamine-dependent amidotransferase enzymes (for example, see Kim et al., 1996), WbpS possesses an ammonia-dependent (glutamine-independent) transferase activity, and that there is sufficient ammonia available in rich media such as LB to enable biosynthesis of O antigen. Therefore, the experiment was repeated using minimal medium with or without the addition of ammonia. Under these conditions, the quantity of O antigen produced in complementation experiments with four of the wbpS point mutants depended upon the addition of ammonia. In contrast, the LPS profiles exhibited by the uncomplemented mutant and the mutant complemented with the wild-type wbpS allele were independent of ammonia concentration (Fig. 7A–C). When the mutant was complemented with the C2A, R53A, E81A and D106L wbpS alleles, the staining of O-antigen bands was darker when ammonia was added than when it was not. Although these are genetic experiments rather than biochemical characterization of the purified enzyme, the data suggest that for complementation with these mutant alleles, O-antigen biosynthesis occurs via an ammonia-dependent activity of WbpS. Under restricted-ammonia conditions, O-antigen synthesis requires, at least in part, the function of the WbpS N-terminal domain, which probably generates ammonia for the transferase reaction by catalysing the hydrolysis of glutamine. When this domain is mutated, O-antigen biosynthesis is inhibited. The detection of O-antigen production in the ammonia-restricted cultures may be due to the action of other ammonia-producing processes within the cell.

Figure 5.

A. Glutamine bound in the glutaminase active site of the E. coli asparagine synthase B (AsnB) C1A mutant, which is a homologue of WbpS. Atomic coordinates are from the X-ray crystal structure, PDB ID 1CT9 (Larsen et al., 1999). Potential hydrogen-bonding interactions are indicated. Glutamine carbons are shown as white sticks and carbons of the protein are pale grey. Oxygen atoms are mid-grey and nitrogen is dark grey.
B. All of these glutamine-binding amino acid side-chains are conserved in the WbpS sequence.

Figure 6.

Complementation of the wbpS mutation with point-mutant wbpS genes on a shuttle vector. LPS was prepared from bacteria grown in LB, separated by SDS-PAGE and visualized by silver staining (A) or Western blotting using an O antigen-specific primary monoclonal antibody (mAb O25G3D6) (B). The relative signal intensity of O-antigen bands in the Western blot as measured by volumes analysis is represented in (C). Each of the mutant wbpS alleles was able to restore O-antigen production. Compared with the wild-type allele, three of the mutant wbpS genes were equally able to restore O-antigen biosynthesis, resulting in near-wild-type levels. In contrast, the wbpS E81A and D106L point mutants were able to only partially restore O-antigen production.

Figure 7.

Complementation in restricted-ammonia conditions of the wbpS mutation with point-mutant wbpS genes. The wbpS mutant strain harbouring various complementation plasmids or none was grown in minimal medium with or without 50 mM NH3. LPS was prepared from these cultures, separated by SDS-PAGE and visualized by silver staining (A and D) or Western blotting using an O antigen-specific primary monoclonal antibody (mAb O25G3D6) (B and E). The relative intensities of O-antigen bands in Western blots (B) and (E) is represented in (C) and (F). These histograms show the results of volumes analysis. When complemented with the wild-type wbpS construct, O-antigen production was restored independently of ammonia supplementation. This was also observed with the N79L point mutant construct. The extent of complementation by the other point mutants was dependent on addition of exogenous ammonia. The results presented in these blots are reproducible and representative. This figure is a composite of several gels and blots.

The N79L allele complemented as well as the wild-type allele in all conditions, and O-antigen synthesis did not exhibit ammonia dependence (Fig. 7D–F). By analogy to the equivalent amino acid in E. coli AsnB, Asn74, this amino acid side-chain is predicted to be important in catalysis by hydrogen bonding via Nδ2 to the oxyanion of the tetrahedral intermediates formed during hydrolysis of glutamine (Larsen et al., 1999). Our data suggest that in WbpS such a hydrogen-bonding interaction involving the Asn79 side-chain is not important for glutaminase function. A similar observation was made in studies of another class II glutamine-dependent amidotransferase, glutamine PRPP amidotransferase, from E. coli. In the Gln PRPP amidotransferase, replacement of the equivalent asparagine side-chain, by making the N101G mutation, resulted in only a modest reduction in Vm for the glutamine-dependent enzyme activity and no significant change in Km for glutamine (Kim et al., 1996). In contrast with this, the N101D mutation decreased the glutamine-dependent activity Vm 2500-fold, consistent with destabilization of the oxyanion-like transition states through electrostatic repulsion (Kim et al., 1996). We constructed a wbpS N79D allele, and in ammonia-restricted conditions this construct had a severe defect in its ability to complement the wbpS mutant (Fig. 7D–F). These observations suggest that while hydrogen bonding by the Asn79 side-chain is not important for activity of the glutaminase domain of WbpS, this residue probably lines the oxyanion hole.

Mutation of wbqG in E. coli O121

Next we investigated the wbpS homologue in E. coli O121, wbqG, to establish whether or not the E. coli gene encodes an equivalent function in E. coli O-antigen biosynthesis. The wbqG gene was inactivated by insertion of a kanamycin resistance cassette, and the LPS phenotype was examined. In contrast to the P. aeruginosa wbpS mutant phenotype, mutation of the E. coli gene did not abrogate O antigen or core+1 production, and bands consistent with these structures are visible in silver-stained SDS-PAGE gels (Fig. 8A). However, the wbqG mutant LPS profile differs from the wild-type in several important respects: O antigen-containing bands are fainter in the mutant LPS, while the core+1 band is considerably darker; in fact the core+1 band is more intense than the core band, contrasting with the pattern produced by wild-type LPS in which the core band is darker than core+1. There is also a shift in the positions of O antigen-containing bands in the LPS prepared from the mutant, since the high-molecular-weight bands migrated further down the gel compared with the analogous species in the wild-type LPS sample. The core+1 band also migrated slightly faster in the mutant sample than the equivalent wild-type band. Western blotting analysis (Fig. 8B) revealed that antibody binding to the mutant O antigen was stronger than to wild type. This also enabled comparison of the band spacing in the O-antigen region. When the spacing was compared between lanes and at approximately the same height in the blot, we observed that the mutant O-antigen bands were more closely spaced than wild type. All of the wild-type LPS features were restored by complementation of the mutant with either wbqG or wbpS in trans showing that the two genes encode equivalent catalytic functions.

Figure 8.

Analysis of the LPS phenotype of the E. coli O121 wbqG mutant. LPS was separated by SDS-PAGE, and visualized by silver staining (A), or Western blotting using polyclonal sera raised against E. coli O121 (B). These antibodies bind both O-antigen polymers and the core+1 LPS. LPS from the wbqG mutant exhibited O antigen-containing and core+1 LPS bands with a higher electrophoretic mobility than the corresponding wild-type species. The wild phenotype was restored by introduction of either wbqG or wbpS on a plasmid.

Every aspect of the wbqG LPS phenotype is consistent with the incorporation of d-GalNAc(3OAc)A in O units, in the usual place of d-GalNAc(3OAc)AN, which could be a logical consequence of knocking out a uronamide-forming amidotransferase function. This change in the structure of the O unit would then have multiple consequences for O-antigen assembly and chemistry, particularly because the negative charge of each repeating unit will be increased by the introduction of an additional uronate. A more negatively-charged O repeat can account for the higher electrophoretic mobility of O polysaccharide and core+1 bands towards the anode in SDS-PAGE, the decreased spacing between bands, and the stronger antibody binding. All of these consequences of altered O unit charge were observed in another O antigen mutant previously investigated by our group (King et al., 2008).

We confirmed this hypothesis by elucidating the structures of O antigen from the wbqG mutant and the complemented strain using nuclear magnetic resonance (NMR) and mass spectrometry (MS) (see below). This was possible for this mutant, because, unlike the P. aeruginosa wbpS mutant, the E. coli wbqG strain was able to produce O polysaccharides.

If the P. aeruginosa wbpS mutation abrogates O-antigen biosynthesis, why is it that the E. coli wbpG mutant is still able to make O antigen? According to the current model, after biosynthesis of the sugar-nucleotide precursors for the sugars in each O-antigen repeat, the O units are assembled by glycosyltransferases in the cytoplasm, and are subsequently translocated across the inner membrane, polymerized and ligated to lipid A–core (King et al., 2009). The P. aeruginosa wbpS mutant phenotype indicates that one or more of the enzymes involved in these processes has a substrate specificity with an absolute requirement for the uronamide form of the sugar, that is, for the WbpS-catalysed reaction product. In P. aeruginosa O6, preventing uronamide biosynthesis therefore blocks O-antigen production. The inverse logic applies to the E. coli case, i.e. production of O antigen by the E. coli wbqG mutant shows that there is no such absolute uronamide requirement in this strain. Thus, every enzyme which functions downstream of wbqG in the pathway must have a relaxed substrate specificity that allows incorporation of the uronic acid in the mutant O antigen in place of the uronamide. This being said, analysis of LPS prepared from the E. coli wbqG mutant by SDS-PAGE and silver staining suggested that the wbqG mutation resulted in production of proportionally more core+1 LPS and less O antigen (high-molecular-weight bands). This suggests that the O-antigen polymerase does have a preference for the normal substrate so that mutant O repeats are less efficiently polymerized and the wbqG mutant has a higher proportion of O units ligated to lipid A–core as monomers rather than long chains.

Elucidation of the repeating unit structures of the wbqG mutant O antigen and complemented mutant

Lipopolysaccharide was purified from the E. coli O121 wbqG mutant, and also from the complemented mutant, in which the wbqG gene was provided in trans. After mild-acid hydrolysis to remove lipid A, the polysaccharide-containing fraction was isolated by size-exclusion chromatography. The wbqG mutant polysaccharides eluted as a broad peak after void volume, whereas polysaccharides from the complemented strain came as sharp peak, indicating higher molecular mass. Further purification by ion exchange chromatography was not performed to avoid preferential isolation of polysaccharide subpopulations with different charges. NMR spectra of both polysaccharides were completely assigned using 2D methods. The spectra agreed with the published structure (Parolis et al., 1997) and were generally similar to each other with the exception of different positions of the H-4 and H-5 signals of the GalNAcA(N) residue b (Fig. 9). This can be explained by differential substitutions of the C-6 carboxyl group at this position, with COOH in the wbqG mutant and CONH2 in the complemented strain. To rule out the possibility that these signal shifts may have been caused by differences in experimental conditions like salt state or pH, both polysaccharides were deacylated with aqueous ammonia, mixed together 1:1, and a new set of 2D NMR spectra was obtained. Two sets of signals for the residue b were observed with large differences in chemical shifts of the H-4 and H-5 protons (Fig. 9A). One set of signals had the same chemical shifts as previously observed in analysis of the complemented-strain polysaccharides, indicating that these signals were from a sugar with a C-6 carboxamide, insensitive to change in pH or counter-ion. The other set of signals was shifted strongly after treatment with ammonia, compared with the wbqG polysaccharide signal. Therefore in the wbqG mutant polysaccharide, residue b had an unsubstituted carboxyl group.

Figure 9.

A. Portion of the 13C-1H correlated HSQC NMR spectra of polysaccharides from the E. coli O121 wbqG mutant LPS (red), LPS from the mutant complemented in trans with wbqG on a plasmid (wbqG + wbqG, green), and the mixture of both polysaccharides after treatment with aqueous ammonia (black). This region of the HSQC spectrum shows the HC-5 signal from the GalNAc(OAc)A(N) residue b.
B. The deduced structures of O-antigen repeating units from the wbqG LPS and the complemented-wbqG LPS. Bold lower-case letters are used to identify residues. In wbqG, R = O-. In the complemented strain, R = NH2. *3-O-acetylation on residue b was present in 30–40% of O-antigen repeats.

Further evidence to support these conclusions was obtained from mass spectrometry analysis of polysaccharides from the wbqG mutant and the complemented strain. Capillary electrophoresis-mass spectrometry (CE-MS) analysis using a high orifice voltage produced mass spectra containing ion peaks from fragmentation of the O polysaccharides (Fig. 10). All of those fragments which did not contain residue b had the same m/z in both spectra; but fragments containing residue b had m/z one mass unit higher in the wbqG mutant polysaccharide which corresponds with the increase in mass when NH2 is replaced by OH.

Figure 10.

Capillary electrophoresis-mass spectrometry (CE-MS) analysis of polysaccharides from the E. coli O121 wbqG mutant LPS (A) and the complemented-wbqG LPS (B). In these experiments a high orifice voltage was applied to promote collision-induced dissociation of the O polysaccharides into oligosaccharide fragments. Peaks are labelled according to the sugar residues present in these fragments.

Therefore we deduce that the wbqG mutant O-antigen repeat unit had a similar structure to the wild-type except that a uronic acid was present in the position normally occupied by the uronamide residue b (Fig. 9B). Furthermore, the uronamide was restored by complementation (Fig. 9B). The structural analyses demonstrated that wbqG is required for biosynthesis of the uronamide sugar in E. coli O121 O antigen.

Concluding remarks

We have presented genetic data which (i) show that wbpS and wbqG are functionally interchangeable, (ii) show that they are required for production of uronamide sugars in O-antigen biosynthesis, and (iii) support the assignment of the proteins encoded by these genes as class II glutamine-dependent amidotransferases, with conserved N-terminal glutaminase domains that provide ammonia for O-antigen biosynthesis. All of these data substantiate the hypothesis that WbpS and WbqG are amidotransferase enzymes which catalyse the biosynthesis of uronamides by transfer of ammonia to the carboxylate moiety of the corresponding uronic acid. To our knowledge this is the first reported experimental investigation of uronamide biosynthesis in any organism.

Experimental procedures

Bacterial strains, plasmids and culture conditions

Bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were routinely propagated in lysogeny broth (LB, also commonly known as Luria–Bertani broth) or on LB agar containing 1.5% Bacto agar (Difco). All strains were grown at 37°C with ampicillin (100 µg ml−1), carbenicillin (200 µg ml−1), kanamycin (50 µg ml−1), streptomycin (200 µg ml−1) and gentamicin (15 µg ml−1 for E. coli; 300 µg ml−1 for P. aeruginosa), added as appropriate.

Table 1.  Bacterial strains and plasmids used in this study.
Strain/plasmidGenotype/relevant characteristicsReference/source
E. coli
JM109e14-(McrA-) recA1 endA1 gyrA96 thi-1 hsdR17 (rKmK+) supE44 relA1Δ(lac-proAB) [F′traD36+proAB+lacIqZΔM15]Stratagene
DH5αλpirF-ϕ80dlacΔM15 Δ(lacZYA-argF)-U169 deoR recA1 endA1 hsdR17(rK- mK+) supE44 thi-1 gyrA96 relA1λpirGrant et al. (1990)
SM10 λpirthi-1 thr leu tonA lacY supE recA RP4-2-Tc::Mu KmrλpirSimon et al. (1983)
EcoR20E. coli wild-type strainOchman and Selander (1984)
EcoR20cSpontaneous streptomycin-resistant derivative of EcoR20, SmrThis work
qG3EcoR20c wbqG::aphA1, Kmr, SmrThis work
P. aeruginosa
IATS O6Wild-type strain, IATS serotype O6 reference strainATCC strain number 33354
Fisher 1Wild-type strain, Fisher immunotype 1 reference strain (IATS serotype O6)ATCC strain number 27312
2M1IATS O6 wbpS::aacC1, GmrThis work
wbpS big 1–8Faster-growing derivatives of 2M1, GmrThis work
pEXMCSCounterselectable, mobilizable suicide vector backbone, derived from pEX100T (Schweizer and Hoang, 1995); oriT sacB AprKing et al. (2007)
pRE112Counterselectable, mobilizable suicide vector backbone; oriT oriV sacB CmrEdwards et al. (1998)
pUCP20EscherichiaPseudomonas shuttle vector derived from pUC18, Apr, CbrWest et al. (1994)
pPS856Source of small, broad host range gentamicin resistance gene aacC1, Apr, GmrHoang et al. (1998)
pGEMTKan/KpnISource of KpnI-flanked kanamycin resistance gene aphA1. aphA1 originally derived from pUC4K (Taylor and Rose, 1988). Kmr, AprJessica Fullerton
pE3Suicide vector for allele-exchange mutagenesis of wbpS in P. aeruginosa O6. Contains the wbpS::aacC1 allele in the pEXMCS backboneThis work
pVm1Suicide vector for allele-exchange mutagenesis of wbqG in E. coli O121. Contains the wbqG::aphA1 allele in the pRE112 backboneThis work
pCRBlunt-wbpSRibosome binding site-wbpS fusion topoisomerase cloned into pCR®-Blunt II-TOPO®. wbpS amplified from Fisher 1This work
pDp1Wild-type wbpS complementation plasmidThis work
pFn1wbpS-C2A complementation plasmidThis work
pHf1wbpS-R53A complementation plasmidThis work
pHg1wbpS-N79L complementation plasmidThis work
pHi1wbpS-E81A complementation plasmidThis work
pHj1wbpS-D106L complementation plasmidThis work
pZb1wbpS-N79D complementation plasmidThis work
pHu1wbqG complementation plasmidThis work

To test the ammonia dependence of LPS phenotypes, P. aeruginosa was cultured in defined minimal medium as described by Vidaver (1967), containing (w/v) 0.02% MgSO4·7H2O, 0.3% KH2PO4, 0.6% Na2HPO4, with 0.5% glucose as carbon source, and with 0.05% l-Asn and ± 50 mM NH4Cl as nitrogen sources. The salt solution was autoclaved, cooled, and then filter-sterilized glucose solution was added.

DNA methods

Standard methods were used for DNA manipulations. Enzymes and reagents were purchased from New England Biolabs or Invitrogen. Oligonucleotides were supplied, and DNA sequencing was performed, by the University of Guelph Molecular Supercentre. KOD Hot Start DNA polymerase (Novagen) was used for PCR.

Generation of the wbpS knockout

wbpS was amplified from P. aeruginosa Fisher 1 chromosomal DNA template using PCR primers 5′-AAAAAAAACTGCAGACATATGTGTGGTATCGCTGGC and 5′-AAAAAAAAAAGCTTTCATGACTCCAACCACGC (PstI and HindIII consensus sequences underlined), and was cloned into the suicide vector pEXMCS (King et al., 2007) making use of restriction endonucleases PstI and HindIII. The wbpS coding sequence was then disrupted by insertion of the gentamicin resistance cassette from pPS856 into the internal KpnI site, generating the allelic exchange suicide vector pE3. pE3 was transferred into the acceptor strain P. aeruginosa IATS O6, by biparental mating, with trans-acting transfer functions provided by the E. coli donor strain SM10 λpir (Simon et al., 1983). Allelic replacement of the mutant wbpS::aacC1 for the wild-type wbpS in the chromosome was selected for by growth of transconjugants on Pseudomonas isolation agar (Difco) supplemented with gentamicin. Double recombinants were then isolated by passage onto LB agar (without NaCl) supplemented with gentamicin and 10% (w/v) sucrose, which gives a growth advantage to clones which have lost the suicide plasmid-borne levansucrase gene, sacB, which confers sucrose sensitivity (Gay et al., 1985). Allelic exchange in putative wbpS mutant strains was confirmed by PCR. Several mutants were isolated, and after confirmation that they shared a common LPS phenotype (not shown), a representative clone was chosen. This strain was called 2M1.

Generation of the wbqG knockout mutant

wbqG was PCR-amplified from E. coli O121 chromosomal DNA template using the primers 5′-AATAAATATCTAGATGTGTGGATTAGCTGGTTTC (XbaI site underlined) and 5′-AATAATACGAGCTCACTTTCATTTCTGGTTTTCTAAC (SacI consensus underlined). The PCR product was digested with XbaI and SacI, and ligated into the multiple cloning site of pRE112 (Edwards et al., 1998). The wbqG coding sequence was then disrupted by insertion of the kanamycin resistance cassette from pGemTKan/KpnI into the internal KpnI site, generating the allelic exchange suicide vector pVm1. The plasmid pRE112 has the R6K oriV, and consequently requires in trans provision of the pi protein for plasmid maintenance. DH5αλpir was therefore used as the host strain during construction of pVm1, which cannot replicate in wild-type E. coli. Escherichia coli SM10 λpir was used as the donor for biparental conjugative transfer of pVm1 into E. coli O121. To allow for selection of transconjugants, ECOR20c was used as the acceptor strain in matings. ECOR20c is a spontaneous streptomycin-resistant clone obtained by spreading stationary phase liquid culture of ECOR20 onto LB agar supplemented with 200 µg ml−1 streptomycin. We confirmed that ECOR20c has unchanged LPS profile as judged by SDS-PAGE with silver staining (not shown). Incorporation of pVm1 DNA into the bacterial chromosome was selected for by growth on LB agar with streptomycin and kanamycin. Double recombinants were then isolated by passage onto LB agar (without NaCl) supplemented with streptomycin, kanamycin and 10% (w/v) sucrose. The expected DNA rearrangement was verified by PCR. wbqG mutants isolated from independent conjugation reactions had a common LPS phenotype (not shown) and a representative clone was chosen, called qG3.

Construction of the wild-type wbpS complementation plasmids pDp1

wbpS was PCR-amplified from a Fisher 1 DNA template using the primers 5′- AAAAAAAAACTAGTGAAGGAGGATATACATATGTGTGGTATCGCTGGC (SpeI consensus and ribosome binding site underlined) and 5′-AAAAAAAAAAGCTTTCATGACTCCAACCACGC (HindIII site underlined), and was topoisomerase-cloned into pCR®-Blunt II-TOPO® (Invitrogen) generating the plasmid pCRBlunt-wbpS. The ribosome binding site-wbpS fusion in pCRBlunt-wbpS was excised with SpeI and HindIII, and ligated into XbaI, HindIII-cut pUCP20 shuttle vector to generate the wild-type wbpS complementation vector pDp1.

Construction of the wbqG complementation plasmid pHu1

wbqG was PCR-amplified from E. coli O121 chromosomal DNA template using the primers 5′- AAAAAAAATCTAGAGAAGGAGGATATACATATGTGTGGATTAGCTGGTTTCC (XbaI consensus and ribosome binding site underlined) and 5′-AAAAAAAAAAGCTTGACTTTCATTTCTGGTTTTCTAACC (HindIII site underlined). The PCR product was then cloned into pUCP20 with the use of XbaI and HindIII, generating pHu1.

Construction of the point-mutant wbpS complementation plasmids pFn1, pHf1, pHg1, pHi1, pHj1 and pZb1

The plasmid pCRBlunt-wbpS was used as template for PCR amplification with primers which introduced single amino acid substitutions into the encoded WbpS sequence. The C2A mutation was introduced by amplification with the primer pair 5′-AAAAAAAAACTAGTGAAGGAGGATATACATATGgccGGTATCGCTGGCTTCTGG (SpeI consensus and ribosome binding site underlined, mutant codon in lower case) and 5′-AAAAAAAAAAGCTTTCATGACTCCAACCACGC (HindIII site underlined). The PCR product was digested with SpeI and HindIII, and cloned into XbaI, HindIII-cut pUCP20 to generate pFn1. The remaining substitutions were made by overlap extension PCR as described by Horton (Horton et al., 1990) using the primers listed in Table 2. The mutant PCR alleles were digested with SpeI and BglII, and ligated into pCRBlunt-wbpS cut with the same enzymes. The point-mutant wbpS genes were then excised with SpeI and HindIII, and ligated into XbaI, HindIII-cut pUCP20 to generate the plasmids pHf1 (R53A), pHg1 (N79L), pHi1 (E81A), pHj1 (D106L) and pZb1(N79D).

Table 2.  Primers used for site-directed mutagenesis of wbpS in pCRBlunt-wbpS.
PrimerPrimer sequence (mutant codons and their complement underlined)

Electroporation protocol for P. aeruginosa

Approximately 1.5 ml of stationary-phase P. aeruginosa liquid culture was sedimented in a microfuge, washed three to four times with 1 ml of room-temperature 300 mM sterile sucrose solution and finally suspended in 100 µl of 300 mM sucrose. Approximately 250 ng of plasmid DNA was added, then the suspension was pulsed in a 1 mm gap electroporation cuvette using a Bio-Rad Micropulser™ apparatus at the pre-programmed Ec1 setting. Transformed cells were recovered for 2 h in LB at 37°C without shaking.

LPS preparation

For SDS-PAGE analysis, cell numbers were carefully normalized according to optical density measurements at 600 nm, and LPS was prepared using the proteinase K digestion method of Hitchcock and Brown (1983). For analysis of O-antigen structures by NMR and MS, LPS was prepared using the hot aqueous phenol extraction method of Westphal and Jann (1965).


SDS-PAGE was performed with a discontinuous gel system and 12.5% resolving gels (Laemmli, 1970) but with no SDS added to the resolving gel. LPS was visualized using the rapid silver staining method of Fomsgaard and colleagues (1990) or by Western blotting.

Western immunoblotting

Western transfer of LPS was performed to BioTrace®NT nitrocellulose membranes (Pall) according to standard protocols with minor modifications (Towbin et al., 1979; Burnette, 1981). Mouse monoclonal antibody (mAb) O25G3D6 (Emara et al., 1995) specific for P. aeruginosa O6 O antigen, or rabbit polyclonal serum raised against E. coli O121 was used as primary antibodies. Blots were incubated with alkaline phosphatase-conjugated secondary antibodies and then visualized by incubation with nitroblue tetrazoleum (NBT) (Blake et al., 1984) and 5-bromo-4-chloro-3-indoyl phosphate (BCIP) according to standard protocols.

The signal intensity of O antigen-containing bands in Western blots was measured using volume analysis tools in the Quantity One software package (Bio-Rad). Equivalent areas were analysed in each lane using local background subtraction.


Cells were taken from mid-log phase cultures and mounted without prior washing. Differential interference contrast (DIC) microscopy was performed using a Zeiss Axiovert 200M microscope with a 100× objective. Live cells were immobilized by pre-treatment of glass coverslips with poly-l-lysine solution (Sigma-Aldrich). Transmission electron microscopy of negatively stained whole mounts was performed using a Philips CM-10 microscope at 80 kV. Cells were transferred to copper grids directly from broth and stained with 1% aqueous uranyl acetate.

Preparation of polysaccharide samples for structural analysis

Lipid A was removed from purified LPS by mild acid hydrolysis (2% AcOH) followed by sedimentation as previously described (King et al., 2008). Molecules containing long-chain polysaccharides were then separated from those without, by size exclusion chromatography using a Biogel P10 column (Bio-Rad, Hercules, California) and a water–pyridine–AcOH (500:2:5) solvent system. To remove the non-stoichiometrically substituted O-acetyl groups, polysaccharides were incubated with 12% aqueous ammonia for 1 h at 50°C.

NMR spectroscopy

Nuclear magnetic resonance experiments were carried out with a Varian INOVA 600 MHz (1H) spectrometer with a Varian Z gradient probe at 25°C in 3 mm tubes, using standard pulse sequences: double quantum filtered correlation spectroscopy (DQCOSY), total correlation spectroscopy (TOCSY, mixing time 120 ms), nuclear Overhauser effect correlation spectroscopy (NOESY, mixing time 200 ms), rotating-frame nuclear Overhauser effect spectroscopy (ROESY), and 1H-13C heteronuclear single quantum correlation spectroscopy (HSQC). Acetone was used as an internal reference (2.225 ppm for 1H and 31.45 ppm for 13C).

Capillary electrophoresis-mass spectrometry (CE-MS)

All experiments were performed as described previously in detail (Li and Richards, 2007). Briefly, a Prince CE system (Prince Technologies, the Netherlands) was coupled to a 4000 QTRAP mass spectrometer (Applied Biosystems/MDS Sciex, Canada). A sheath solution (isopropanol–methanol, 2:1) was delivered at a flow rate of 1.0 µl min−1. Separations were obtained on ∼90 cm long, bare fused-silica capillary using 15 mM ammonium acetate in deionized water, pH 9.0. For positive and negative ion detection modes, 5 kV or −5 kV electrospray ionization voltages were used respectively. A high orifice voltage (+400 V) was used to promote in-source, collision-induced dissociation (pseudo-tandem mass spectrometry), and these spectra were acquired in enhanced resolution scan mode (ER) with a scan rate of 250 Da s−1.


Genomic DNA from E. coli O121 was a gift from Mohamed Karmali (Public Health Agency of Canada, Laboratory for Foodborne Zoonoses, Guelph, ON, Canada). Polyclonal antisera raised against E. coli O121 was a gift from Kim Ziebell (Public Health Agency of Canada, Laboratory for Foodborne Zoonoses, Guelph). We gratefully acknowledge Chris Whitfield (University of Guelph), who allowed us to conduct the wild-type E. coli O121 experiments in his lab. The plasmid pGemTKan/KpnI was a gift from Jessica Fullerton (University of Guelph). We also thank Jacek Stupak (National Research Council, Ottawa) for mass spectrometry analysis, Dianne Moyles (University of Guelph) for electron microscopy studies, and Evanna Huynh (University of Guelph) for technical assistance.

This work was supported by an operating grant from the Canadian Institute of Health Research (# MOP-14687). J.S.L. holds a Canada Research Chair in Cystic Fibrosis and Microbial Glycobiology jointly funded by the Canadian Foundation of Innovation and the Ontario Innovation Trust.