waaL has been implicated as the gene that encodes the O-antigen ligase. To date, in vitro biochemical evidence to prove that WaaL possesses ligase activity has been lacking due to the difficulty of purifying WaaL and unavailability of substrates. Here we describe the purification of WaaL, a membrane protein with 11 potential transmembrane segments from Pseudomonas aeruginosa, and the development of an in vitro O-antigen ligase assay. WaaL was expressed in a P. aeruginosa wbpL knockout strain, which is defective in its initial glycosyltransferase for O-antigen biosynthesis. This approach allowed the purification of WaaL without contaminating O-antigen-undecaprenol-phosphate (Und-P) molecules. Purified WaaL resolved to a monomer (35 kDa) and a dimer (70 kDa) band in SDS-PAGE. The substrates for the O-antigen ligase assay, O-antigen-Und-P and lipid A-core were prepared from a waaL mutant. ATP at 2–4 mM is optimum for the O-ligase activity, and ATP hydrolysis by WaaL follows Michaelis–Menten kinetics. Site-directed mutagenesis analysis indicated that the periplasmic loop region of WaaL is important for ligase activity. A waaL mutant of P. aeruginosa could not be cross-complemented by waaL of Escherichia coli, which suggested that each of these proteins has specificity for its cognate core oligosaccharide.
Pseudomonas aeruginosa is a Gram-negative opportunistic pathogen that causes disease in individuals whose defences are compromised, including those with cancer, HIV infections and cystic fibrosis (CF) (Kielhofner et al., 1992; Govan and Deretic, 1996). More than 90% of CF patients are suffering from chronic pulmonary infections by P. aeruginosa; these infections remain the major cause of death in this group. P. aeruginosa is well endowed with an arsenal of virulence factors that include secreted proteins such as exotoxin A, phospholipase C, protease, and exopolysaccharides such as the alginate-containing mucoid substances and the endotoxic lipopolysaccharide (LPS) (Burrows and Lam, 1999). LPS is an integral component of the outer membrane of Gram-negative bacteria, and is composed of three regions: the hydrophobic lipid A, which serves to anchor the LPS in the outer membrane; a core oligosaccharide; and the O antigen (O polysaccharide). Interestingly, P. aeruginosa produces two forms of O antigen, known as A band (homopolymer of d-rhamnose) and B band (heteropolymer of tri- to pentasaccharides of dideoxy sugars), and their assembly processes are fundamentally different (Rocchetta et al., 1999). Sugar nucleotide precursors for both homopolymeric and heteropolymeric O antigens are synthesized in the cytoplasm of the bacterial cell and used as donor molecules for the assembly of the O-polysaccharide units (Sadovskaya et al., 1998). Biosynthesis of B-band LPS follows the Wzy-dependent pathway as proposed by Whitfield (1995) in which individual O units are transported into the periplasm by an O-antigen flippase, Wzx (Burrows and Lam, 1999). Next, these O units are polymerized by an O-antigen polymerase, Wzy (de Kievit et al., 1995), to a strain-specific range of lengths dictated by the O-antigen chain length regulator, Wzz (Burrows et al., 1997; Daniels et al., 2002). The function of Wzz accounts for an observed modal distribution of LPS bands when these molecules are analysed by SDS-PAGE and silver staining. In contrast, during the A-band biosynthesis process, the synthesis of the homopolymeric O antigen is completed in the cytoplasm, and the polymer is then exported to the periplasm via an ATP binding cassette (ABC) transport system (Rocchetta and Lam, 1997). In subsequent steps, ligation of either the homopolymeric or the heteropolymeric O polysaccharides onto lipid A-core oligosaccharides takes place in the periplasm (Mulford and Osborn, 1983). waaL has been implicated as the gene that encodes an O-antigen ligase responsible for this reaction, and waaL mutants were found to accumulate polymerized O-antigen-linked undecaprenol-phosphate (Und-P) on the periplasmic surface of the cytoplasmic membrane (Mulford and Osborn, 1983; Morona et al., 1991; Falt et al., 1993; Reeves, 1993; Manning et al., 1995). In P. aeruginosa, a null mutant of waaL (waaL::Gmr) produces rough LPS devoid of O-antigen repeat units (Abeyrathne et al., 2005), and it was found to accumulate glycolipids containing O-antigen polysaccharide linked to carrier lipid. These glycolipid molecules were proven to be distinct from LPS molecules because they are sensitive to treatment with phenol (Abeyrathne et al., 2005). In silico analysis of P. aeruginosa WaaL using the TMHMM Serve v.2.0 program (Sonnhammer et al., 1998) predicted that WaaL is an integral membrane protein with 11 predicted transmembrane-spanning segments (TMS) and a large periplasmic loop region containing 49 amino acid residues. Until this study, direct evidence to show that WaaL proteins have ligase activity (linking O antigen to lipid A-core) has not been obtained. The obvious hurdles include the difficulty in purifying WaaL, which has a high proportion of TMS, and the unavailability of the substrate and the receptor for use in the O-antigen ligase enzymatic assays.
Here we describe for the first time the purification of an enzymatically active form of WaaL from P. aeruginosa, the requirement of ATP hydrolysis as an energy source for O-antigenic ligation activity, and the development of an in vitro biochemical assay to quantitatively determine the activity of WaaL.
Expression and purification of WaaL
Initial attempts to express WaaL using pET and pQE80 expression systems and different Escherichia coli expression strains, such as BL21, BL21 codonplus and BL21Rosetta (Novagen), were unsuccessful. Expression of WaaL was achieved using pVLT31 (de Lorenzo et al., 1993), which was designed for expressing proteins in the native P. aeruginosa background. In addition, we used a wbpL mutant of P. aeruginosa as an expression host. wbpL encodes the initial transferase for both A-band and B-band O-antigen biosynthesis; therefore, a knockout of this gene renders the organisms devoid of both forms of LPS molecules (Rocchetta et al., 1998). This allows purified WaaL to be free of A-band and B-band LPS. The yield of WaaL expressed in the native P. aeruginosa (wbpl::GmR) background was approximately 0.86 mg l−1 of culture. The expressed protein was solubilized in 1% β-dodecyl-maltoside (DDM) and purified to > 95% homogeneity (Fig. 1A, left panel). SDS-PAGE analysis showed that the purified WaaL has an apparent molecular mass of 35 kDa (Fig. 1A, left panel), differing from the predicted mass of 46 762.34 Da. Another predominant protein band at an apparent molecular mass of 70 kDa was also observed. Further, Western blot analysis using anti-His antibody showed reactivity of the antibody to both the 35 and 70 kDa bands (Fig. 1A, left panel), thereby identifying these as WaaL and suggesting that they constitute a monomer and dimer respectively. To further verify the existence of the dimer in the purified preparation of WaaL, both the 35 kDa and the 70 kDa bands were subjected to in-gel digestion with trypsin, treatment with CNBr, and a combination of treatment with trypsin and CNBr, following established procedures (van Montfort et al., 2002). The peptide samples were analysed using Matrix-assisted Laser Desorption Ionization time-of-flight mass spectrometry (MALDI-TOF MS). Five ion peaks were detected in the trypsin-treated sample, which included two methylated peptides, and their m/z ratios are 1197.569, 1588.823, 2059.956, 2828.2000 (methylated peptide) and 3115.500 (methylated peptide). Two ion peaks were detected in the CNBr-treated samples that included one methylated peptide with m/z ratios of 2716.4681 and 2390.0465 (methylated peptide). Samples that were treated with trypsin and CNBr showed three ions with m/z ratios of 1086.622, 1437.665 and 2475.455. The detected ions from the fragmented peptides of both protein bands corresponded well with the predicted peptides of trypsin digestion and/or CNBr hydrolysis of WaaL from P. aeruginosa. These results further confirmed that the 35 and 70 kDa bands corresponded to a monomer and a dimer of WaaL. To determine the E-value of the aforementioned ion peaks, the m/z values of the ions from trypsin-digested peptides of WaaL were analysed using the Profound online server (http://prowl.rockefeller.edu) and by comparison with the NCB Inr database with all taxa. After inputting the masses of the aforementioned five ion peaks from trypsin digestion into the analysis, we obtained an E-value of 0.073 and the suggested match from the database was WaaL of P. aeruginosa PAO1. Therefore, the identity of the monomer and the dimer bands as WaaL was unequivocal. The purified WaaL was further examined by transmission electron microscopy (TEM) using negative staining, and no evidence of aggregation was discerned (data not shown). However, aggregation of WaaL was observed when the purified protein was examined after being stored at −20°C for more than 3 months.
O-antigen ligation is an energy-dependent process
Polymerization of heteropolymeric O units, and ligation of O antigen to lipid A-core molecules, are processes expected to occur in the periplasm; however, thus far, there has been no direct evidence to show the requirement of energy in the latter process. Almost all of the DNA and RNA ligase enzymes that have been characterized to date utilize ATP or NAD+ as cofactors; hence, to understand more clearly the energy requirement of the O-antigen ligation process, we performed ATPase assays with purified WaaL to determine whether WaaL possesses ATPase activity. Our results showed that WaaL could hydrolyse ATP in the presence of the divalent cation Mg2+ (10 mM). The optimum conditions for an ATPase assay were determined, and thereafter, 10–11 pmol of purified WaaL was used and the reaction mixture was incubated at 37°C for 45 min. Under these conditions, WaaL could hydrolyse up to 5 mM of ATP and showed a specific activity of 86.68 nmol min−1 mg−1 (Fig. 1B). ATP concentrations greater than 5 mM caused decreased levels of ATPase activity as shown in Fig. 1B. Because the purified WaaL has been shown to be a dimer (Fig. 1A), it is plausible that ATP concentrations of ≥ 5 mM could exert substrate inhibition due to allosteric effects. In the ATPase assays, we used a low concentration of ATP (2–4 mM) that was deemed optimal for the WaaL activity based on the kinetic analysis (Fig. 1B). Interestingly, at low ATP concentrations (< 5 mM) the kinetics of the ATPase activity of detergent-solubilized WaaL follows Michaelis–Menten kinetics with a Vmax of 362.80 nmol min−1 mg−1 and Km of 0.16 μM (Fig. 1C). This is the first in vitro assay that shows that a WaaL protein exhibits ATPase activity and suggests a possible source of energy for the O-antigen ligation reaction.
In vitro ligase assay using purified WaaL
In our previous study, we observed that PAO1waaL (waaL mutant) of P. aeruginosa produced rough LPS lacking both A-band and B-band O antigens (Abeyrathne et al., 2005). In addition, we demonstrated that the synthesis of A-band and B-band O polysaccharides in this mutant was not affected, except that these polysaccharides were not ligated to the lipid A-core. Instead, these O polysaccharides remain attached to Und-P. These molecules are ideal as the substrate for an O-ligase assay. In light of this, lipid A-core, A-band-Und-P and B-band-Und-P molecules (designated as ‘crude LPS’) were prepared from the waaL mutant and used as substrate to perform the ligase assay in vitro. The optimum conditions for a ligase assay were determined, and thereafter, the assay was performed in the presence of 5 μM crude LPS, 11 pmol purified WaaL, and 2 or 4 mM ATP. After the enzyme–substrate and receptor reactions were allowed to occur (12 h), the samples were analysed by SDS-PAGE, silver staining (Fig. 2, panel A) and Western blotting using A-band-specific monoclonal antibody (mAb) N1F10 (Fig. 2, panel B), and B-band-specific mAb MF15-4 (Fig. 2, panel C). Control LPS samples from the wild-type strain resolved into a typical heterogeneous LPS-banding pattern, with high-molecular-weight O-antigen bands and low-molecular-weight bands of core-plus-one O-repeat and core-oligosaccharide bands (Fig. 2A, lane 1). In contrast, the negative-control LPS sample from the waaL mutant showed only low molecular weight bands that have been shown previously to be core oligosaccharides (Abeyrathne et al., 2005) (Fig. 2A, lane 2). In the sample containing WaaL, the crude LPS substrate, 2 mM ATP and the reaction buffer, ligation of O antigens onto lipid A-core was apparent with the LPS forming a banding pattern that has a random distribution consisting of both high- and low-molecular-weight LPS molecules (Fig. 2A, lanes 3–5). The observed LPS-banding pattern of the in vitro-ligated LPS was different compared with the wild-type LPS control (Fig. 2A, lane 1 versus lanes 3–5). This is not unexpected and could be due to two main reasons. First, the substrate we used in the ligase assay is crude LPS prepared by the Hitchcock and Brown (1983) method and may therefore contain other polysaccharide molecules. Second, the in vitro ligation assays contain only WaaL protein, contrasting the assembly of LPS in vivo that involves WaaL and other protein members of the Wzy-dependent membrane complex. To prove that the LPS bands of the in vitro-ligated reaction mixture were not due to degradation product of WaaL, the ligation reaction was digested with proteinase K. No change in the LPS banding pattern was observed after the proteinase K treatment (Fig. 2A lane 3 versus lanes 4 and 5). When ATP was omitted from the reaction, ligation of O antigen onto lipid A-core did not occur and the banding pattern in the sample was identical to that of LPS from the waaL mutant; that is, only the low-molecular-weight core-oligosaccharide bands were discerned. This is despite the presence of WaaL, which was resolved as a monomer and a dimer band with apparent molecular masses of 35 and 70 kDa as observed in the purified WaaL preparation (Fig. 2A, lane 6). This observation provided concrete evidence that a requirement of ATP was absolute, and that without ATP the ligation reaction could not proceed. In the ‘no substrate’ control, a similar result was observed, where no high-molecular-weight LPS bands could be discerned and the presence of the WaaL monomer and dimer bands was apparent (Fig. 2A, lane 7). When WaaL was omitted in the reaction, ligation of O antigen onto lipid A-core did not occur and the banding pattern in the sample was identical to the LPS from the waal mutant (data not shown). There was also no reaction with mAbs N1F10 and MF15-4 in the Western blotting (data not shown). This particular experiment proved two points: first, WaaL is specific for the crude LPS substrate used in the O-antigen ligase assay, and second, the purity of WaaL without contaminating O-antigen-Und-P substrates was ascertained. The use of mAbs N1F10 and MF15-4 in the Western blotting experiments (Figs 2B and C, lanes 3 and 4) allowed the precise probing of the presence of high-molecular-weight O-antigen-containing LPS molecules in the ligated LPS products. Interestingly, the use of either 2 or 4 mM ATP was sufficient for the in vitro ligation of O antigens to the lipid A-core.
Identification of a specific region of WaaL that correlates with the ligase activity of P. aeruginosa PAO1
Analysis of WaaL using TIGR and NCBI servers (http://cmr.tigr.org/tigr-scripts/CMR/CmrHomePage.cgi, http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) revealed that WaaL possesses three distinct motifs that constitute a glycosyltransferase domain (amino acids 24–152), an O-antigen polymerase domain (amino acids 258–319) and a ligase domain (amino acids 255–401). To identify the specific region(s) of WaaL that would correlate with the ligase activity, five truncations of waaL (amino acid residues 1–101, 1–156, 1–232, 1–301 and 1–353) were made by constructing deletions at the 3′ end with reference to the topology of the protein as predicted by the TMHMM Server v.2.0 program (Sonnhammer et al., 1998). The significance of these truncations to the ligase function of WaaL was assessed by in vivo complementation of a waaL::Gmr strain. None of the truncated versions of waaL was able to complement the mutant to restore the production of smooth LPS with A-band and B-band O antigens (data not shown). These results led us to hypothesize that either the full-length WaaL protein was required or the truncated forms had affected the secondary and tertiary structures that might play an important role in the O-antigen ligation process. To test this hypothesis, three P. aeruginosa::E. coli WaaL (K12) chimeras were constructed based on consideration of the secondary structure of WaaL analysed predicted using the online servers TMHMM (Sonnhammer et al., 1998) and ProDOM (Bru et al., 2005). Chimera 1 contained amino acids 1–233 from P. aeruginosa WaaL (1–8 TMS) and amino acids 233–419 (periplasmic loop region and 9–11 TMS) from E. coli WaaL; chimera 2 contained amino acids 1–306 from P. aeruginosa WaaL (1–8 TMS and periplasmic loop region) and amino acids 344–419 (9–11 TMS) from E. coli WaaL; chimera 3 contained amino acids 1–360 from P. aeruginosa WaaL (1–10 TMS) and amino acids 389–419 (11 TMS) from E. coli WaaL. To detect the in vivo ligase activity of these chimeras, each of the three chimeric waaL constructs was used to complement the waaL::Gmr. Interestingly, chimera 1 and chimera 2 yielded negative results (Fig. 3A, lanes 4 and 5), and only chimera 3 was capable of complementing the waaL mutant and restoring the ligase activity, as evidenced by the production of smooth LPS with high-molecular-weight LPS bands in the complemented strain (Fig. 3A, lane 6). However, all three chimeras were able to express recombinant WaaL proteins in a P. aeruginosa background as visualized by a Coomassie blue-stained SDS-PAGE (Fig. 3B) and Western blot with anti-His-tag antibodies (Fig. 3C).
Key amino acid residues in the periplasmic region are important in ligase activity
Secondary structure analysis of WaaL showed that it has a large periplasmic loop with 49 amino acid residues (amino acids 257–306). By using the SAM-T02 protein-structure prediction (http://www.soe.ucsc.edu/compbio/HMM-apps/T02-query.html) and NCBI Conserved Domains servers, five amino acids were identified in the periplasmic loop that are conserved among WaaL proteins of other Gram-negative bacteria. Using an alanine-scanning approach, site-directed mutants of all five conserved residues were made, namely, R264A, W268A, G282A, G284A and H303A. The H303A mutation completely abolished the ability of waaL to complement the waaL::Gmr (Fig. 4A, lane 8), while the R264A, W268A, G282A and G284A had no effect on the function of waaLPa and these mutants were able to complement the waaL::Gmr to restore LPS production (Fig. 4A, lanes 4–7). To substantiate these findings, additional mutants H303G, H303R and H303N were constructed. The Gly mutant construct was made because the properties of Gly are similar to Ala; however, it should add flexibility to the peptide. The Arg mutation is a conserved change because this amino acid carries a positive charge at neutral pH, which is similar to His. Arg and Asn could form H-bonds to stabilize the protein structure. H303G and H303N completely abolished the activity of WaaL (Fig. 4B, lanes 4 and 6). Interestingly, H303R was able to complement the waaL mutant (Fig. 4B, lane 5). To determine whether any of these mutations affected growth, the growth rates of all the complemented strains were examined. When compared with the wild-type strain, they all reached comparable cell densities as measured by OD600 after 10 h of growth (data not shown). To investigate the protein expression levels of mutant H303G, H303R and H303N, the three mutant proteins were expressed in a P. aeruginosa background, and their expression was detected by Western blotting with His antibodies. As shown in Fig. 4C, all three mutant protein were expressed and confirmed by Western blotting (Fig. 4D). The H303R mutant protein was purified (Fig. S1) and characterized for its ATPase and ligase activities. We used similar conditions to those used for the wild-type WaaL to perform the ATPase assay (10–11 pmol of purified H303R mutant protein). The H303R mutant protein could hydrolyseup to 5 mM of ATP showing specific activity of 378.85 nmol min−1 mg−1 (Fig. S1B). The kinetics of the ATPase activity of detergent-solubilized H303R WaaL follows Michaelis–Menten kinetics with a Vmax of 362.73 nmol min−1 mg−1 and Km of 0.1436 μM (Fig. 1C). Purified H303R protein was used to perform the in vitro ligase assay, and the assay showed that the H303R protein was able to ligate the substrate molecules in the presence of 2 and 4 mM ATP (Fig. S1).
Escherichia coli waaL could not cross-complement PAO1 waaL mutants
Comparison of WaaL from P. aeruginosa PAO1 (WaaLPa) with WaaL from E. coli, K. pneumoniae (WaaLPa and WaaLKp) and P. aeruginosa PA14 (WaaLPa14) by hydropathy plots revealed that all the WaaL proteins share similar secondary structures (Abeyrathne et al., 2005). To further analyse whether waaL from E. coli K12 could be used to replace waaLPa, complementation experiments were performed by transforming waaLEc into the waaLPa knockout mutants derived from PAO1 and PA14. The positive control waaLPa successfully complemented the knockout waaLPa mutants of PAO1 or PA14 (Abeyrathne et al., 2005), while waaLEc could not (Fig. 5A lane 4 and B lane 3). These results indicated that the proteins from the two species have significant differences. Their inability to cross-complement each other is likely due to distinct substrate specificity, probably the acceptor molecule lipid A-core, because E. coli K12 can ligate other LPS O antigen onto its core (Heinrichs et al., 1998).
In previous reports, waaL has been implicated by several research groups, including our own, as the O-antigen ligase gene, but thus far, only genetic data have been obtained to implicate the involvement of this gene and its encoded protein in LPS assembly (Heinrichs et al., 1998; Kaniuk et al., 2004; Abeyrathne et al., 2005; Schild et al., 2005). There has been no direct biochemical evidence to show the enzymatic activity of WaaL. As stated by Kaniuk et al. (2004), the main factors that have hampered biochemical characterization of WaaL include the difficulty of purifying an integral membrane protein such as WaaL, and the unavailability of the substrate for the O-antigen ligation reaction. In this study, we were able to successfully express and purify an enzymatically active form of WaaL. This represents a breakthrough, because WaaL has not been purified previously from any Gram-negative bacteria. In addition, the purified WaaL was crucial for the development of an in vitro biochemical assay to show that WaaL of P. aeruginosa possesses O-antigen ligase activity to attach A-band and B-band O antigens onto lipid A-core receptors. Substrate for the O-antigen ligase assay was prepared from the waaL mutant strain. We also showed that the O-antigen ligation process requires ATP as an energy source. These findings put us in the position to investigate the proposed O-antigen ligation process in which the O-antigen transfer by WaaL would require recognition and detachment of the O sugars from the O polysaccharide-und-P precursor, followed by final transfer of the substrate onto its lipid A-core receptor.
Purified WaaL showed an apparent molecular mass of 35 kDa in SDS-PAGE, indicating that it migrates faster than its predicted size of 46.8 kDa. The aberrant migration of WaaL is typical of the relative mobility of integral transmembrane proteins in SDS-PAGE gels. They tend to migrate 65–75% faster than the relative mobility of their true molecular weights, possibly as a result of hydrophobicity, high binding of SDS or due to retention of their secondary structure, thereby accelerating the migration through the gel matrix (Ward et al., 2000; Saidijam et al., 2003). Similar migration patterns have been observed for two other proteins involved in LPS biosynthesis, namely Wzy (with 12 TMS) and Wzx (with 12 TMS) (Abeyrathne and Lam, 2007). Wzx is involved in the flipping/transport of B-band precursor molecules into the periplasm, while WaaL and Wzy may not be involved in any transport machinery based on the data obtained from studying these two proteins so far. In a recent study, Alaimo et al. (2006) showed that pglK, which encodes a putative ABC-type transporter mediating the translocation of the undercaprenylpyrophosphate-linked heptasaccharide of Campylobacter jejuni, was able to cross-complement an E. coli wzx deficiency, and vice versa. The motif identified as a putative ATP binding cassette in PglK was shown to be important for the function of this protein, suggesting that ATP hydrolysis is required to energize the oligosaccharide translocation. At present, whether Wzx requires ATP for its flippase activity has not been examined. Further, the energy requirement of the O-antigen polymerization step catalysed by Wzy is also not known. More work is underway to define the ATP requirements for both Wzy and Wzx in the LPS assembly process.
Secondary structure predictions indicated that WaaL contains 11 potential TMS, which are predominantly α-helices, with only a few short β-chains. Folding of membrane proteins normally follows a two-stage model, in which the membrane insertion of independently stable α-helices is followed by their mutual interactions within the membrane to give higher-order folding and oligomerization (Popot and Engelman, 1990; Engelman et al., 2003). The geometry and stability of interactions between TM α-helices are the primary determinants of tertiary and quaternary structures of membrane proteins. The TM helix–helix partnership is important in stabilizing membrane protein folding as well as contacts at interaction sites (Senes et al., 2004). Currently, several unique motifs that are responsible for helix–helix association have been identified. One of the best characterized is the G-XXX-G or GG4 motif, consisting of two Gly residues separated by three variable intervening positions. ATP synthase possesses a GG4 motif in the membrane C-terminal part of the protein, and this is important for dimerization (Bustos and Velours, 2005). Further, alteration of the GG4 motif in yeast mitochondrial ATP synthase leads to the loss of subunit g and the super molecular structure of the ATP synthase with concomitant appearance of anomalous mitochondrial morphologies (Bustos and Velours, 2005). WaaL possesses three GG4 motifs, one at the 8th TMS (amino acids 244–248) and two others at the 10th TMS (amino acids 351–355 and 355–359). They might contribute to dimerization of this protein, as suggested by the presence of the 70 kDa dimer in addition to the 35 kDa monomer in our purified preparation of WaaL. MALDI-TOF MS analysis revealed a number of ion species derived from the trypsin treatment of WaaL having an E-value of 0.073 and providing a reasonably high level of confidence of match between the observed ions from the peptide fragments and the predicted fragments of WaaL of P. aeruginosa. After repeated attempts to identify more ions in MS analysis after tryptic or CNBr digestions that might represent a near-full coverage of WaaL, we obtained similar results with a yield of only a low number of ion peaks. This is probably due to the high proportion of TMS (11 in total). The high hydrophobicity of this protein likely hindered the access of specific cleavage sites throughout the sequence of WaaL for hydrolysis by trypsin or CNBr. This is a rather common problem when performing tryptic digests of membrane proteins and performing MS analysis. To ensure that the lack of cleavage of WaaL was not due to technical artifacts, we applied the same method used to digest WaaL to two other membrane proteins, Wzz1 and Wzz2. Each of these contains two TMS and a large periplasmic loop region. Interestingly, the ions observed represent 43% coverage for Wzz1 and 38% coverage for Wzz2. This experiment confirmed that the low number of ions obtained from digestion or chemical hydrolysis of WaaL was due to the high proportion of TMS and hydrophobicity, and not due to technical difficulties. We further observed by TEM that the purified preparation of WaaL was monodispersed (data not shown); therefore, although WaaL formed dimers, it was not due to aggregation of this protein that restricted the potential to yield large numbers of tryptic/CNBr-hydrolysed peptides.
Until this report, no WaaL homologue has been purified, and studied at the enzymatic level to directly verify the widely presumed ‘ligase’ activity. Almost all of the B-band assembly enzymes are integral membrane proteins, and the assembly process of B-band LPS is predicted to occur in the periplasmic space (Rocchetta et al., 1999). An important question remains to be answered: How does this assembly process occur in the periplasm, and what would be the source of ATP for the process of ligation reaction? Other proteins, such as BtuD (accession number PO6611) and the murein peptide ligase (accession number P37773), are periplasmic proteins, possessing ATP signature motifs. However, there are exceptions. For instance, in carbamoyl-phosphate synthase and biotin carboxylase, the ATP binding site within this structure is composed of a palmate motif (Yamaguchi et al., 1993) rather than the usual Walker A/B motif (Walker et al., 1982).
Ligation of O antigen onto core-lipid A in Gram-negative bacteria has been postulated to occur in the periplasm (Whitfield, 1995). However, whether this process is energy-dependent has not been resolved. In contrast, DNA and RNA ligases have been characterized previously, and these enzymes were shown to utilize ATP or NAD+ as cofactors. The T7 DNA ligase can catalyse the ligation of DNA substrates in the presence of ATP, but this would occur at a significantly reduced rate with a deoxy-ATP (dATP) analogue (Hinkle and Richardson, 1975; Engler and Richardson, 1982). It is commonly known that DNA ligases require divalent cations for activity. This requirement appears to be fulfilled by Mg2+in vivo, although other ions, such as Ca2+, have been used as a substitute, and usually the ligation reaction occurs at a reduced rate (Engler and Richardson, 1982). Taking into consideration the findings from DNA and RNA ligase studies, we performed an ATPase assay with purified WaaL in the presence of Mg2+ (10 mM) as the divalent cation. Our results indicate that WaaL possesses ATPase activity and that the activity conforms to Michaelis–Menten kinetics at low concentrations of ATP (< 5 mM). It is interesting that the use of ATP at ≥ 5 mM showed inhibitory effects on WaaL. It is not uncommon that some enzymes are subject to substrate inhibition. In this case, because WaaL is a dimer, the enzyme is active when the first ATP-substrate binding site is occupied, but when the second site is occupied with the ATP substrate, substrate inhibition due to allosteric mechanisms could occur. Naturally, more work is underway to examine the detailed kinetic studies of this proposed allosteric inhibition mechanism. This is the first time that purified WaaL has been shown to possess ATPase activity. At present, one could speculate that the role of ATP hydrolysis in the WaaL-catalysed O-antigen ligation reaction may be similar to those of DNA and RNA ligase reactions, with WaaL using ATP as an energy source for forming glycosidic bonds between O antigen and the lipid A-core. However, it is puzzling that WaaL does not possess a conserved ATP/binding/hydrolysis motif. In search of alternative motifs that could be involved in ATP hydrolysis activities, we identified two putative motifs within WaaL, one at amino acids 204–213 with a GLLIATGSR sequence (similar to a Walker A motif and the possible location of this motif is cytoplasm according to the TMHMM), and the other at amino acids 273–286 having a RQISEHPWLGHGYD sequence, which has a similar R/KXXXGXXXL/VhhhD (X = any amino acids and h = hydrophobic amino acids) signature motif to Walker B, which is located at the periplasmic loop region. We propose that these sequences might be the ATP/binding/hydrolysis motif(s) in WaaL. A detailed analysis of these ATP/binding/hydrolysis motif(s) will be performed in the future.
Specific glucose residues in the core-lipid A structure of the R2 core type have been implicated as being required for the O-antigen ligation process to occur (MacLachlan et al., 1991; Heinrichs et al., 1998), and these sugars have been correlated to the specificity of its cognate core oligosaccharides (Heinrichs et al., 1998). This specific recognition may account for the specificity of a WaaL ligase for the O-antigen ligation in a particular species, because there are chemical and structural differences among the core-oligosaccharide structures of various Gram-negative species. Also, the O-antigen structures of different bacteria are clearly distinct from each other. In this study, purified WaaL of P. aeruginosa exhibited ligase activity when it was incubated with homologous O antigens and the core-lipid A as a substrate in the presence of 2 or 4 mM ATP and Mg2+ (10 mM). This observation strongly supports a key role for WaaL in the ligation process. However, the banding pattern of in vitro-ligated LPS showed a difference in modal distribution of the LPS bands when compared with LPS from wild-type strain PAO1. Some of the silver-stained bands likely belong to other polysaccharide or glycolipids antigens that are present in the whole-cell lysate, which was the source for O-antigen-Und-P preparation. It has been proposed that the WaaL in P. aeruginosa, as well as the WaaL of other Gram-negative bacteria such as E. coli and Salmonella, function as part of a complex that may involve precise interactions with O-antigen intermediates on an Und-P carrier and specific lipid A-core acceptor. Feldman et al. (1999) suggested an O-antigen processing model for the Wzy-dependent O-antigen biosynthesis pathway complex in which Wzx, Wzy and WaaL are involved in recognizing the Und-P-linked sugars. In P. aeruginosa PAO1 there are two Wzz proteins, named Wzz1 and Wzz2 (Daniels et al., 2002). Therefore, both Wzz proteins are likely involved in the predicted complex formation. After adding purified Wzz1 and Wzz2 either separately or simultaneously into the ligase assay reaction mix, no change in the banding pattern of ligated LPS could be discerned (data not shown). These observations indicated that Wzz1 and Wzz2 might not possess enzymatic activity, but may function as chaperons to facilitate the assembly of a correct conformation of a membrane complex structure that would be crucial for the O-antigen ligation to occur before the process could proceed to chain length determination.
Two strategies were used to determine the region(s) of WaaL that could be important for ligase activity, and the results were assessed by in vivo complementation. Initially, five different versions of C-terminally truncated WaaL were made. None of the deletion constructs could complement a waaL mutant to restore smooth LPS production. This observation suggested that the full length of WaaL might be required for ligase activity, or that C-terminally truncated WaaL was unstable and/or incorrectly inserted into the membrane. Subsequently, chimeras of waaL between P. aeruginosa and E. coli waaL genes were made. Interestingly, chimera 3, which contains TMS 1–10 of P. aeruginosa and the TMS 11 (amino acids 389–419) from E. coli, was capable of complementing the waaL mutant and restored LPS production in the transconjugant. We also found that WaaL has three GG4 motifs, one at the 8th TMS and two at the 10th TMS. and that these motifs may play a role in stabilizing the structure of WaaL, leading to correct function in the ligation process. In addition to the two GG4 motifs at 10th TMS, WaaL has a leucine motif, Leu-X9-Leu at amino acids 342–353. A leucine-zipper motif has been shown to mediate dimerization of proteins such as DNA binding proteins, photosystem I reaction centre polypeptides of higher plants, subunits 6 and 9 of ATP synthase, subunits 3 and 4 of NADH dehydrogenase, and subunit II of cytochrome oxidase in mitochondria (Konstantinov and Moller, 1994). The leucine motif may provide specific recognition sites between membrane-spanning domains of F0-ATPase and between enzyme complexes in the inner mitochondrial membrane (Konstantinov and Moller, 1994). Chimera 3, containing 1–10 TMS of P. aeruginosa and 11 TMS from E. coli waaL, was capable of complementing the P. aeruginosa waaL mutant. Here we propose the leucine-zipper motif at the 10th TMS may contribute to the formation of an LPS assembly complex. Taken together, these results indicate that the structure of WaaL is important for the ligase activity and may also be important for forming an O-antigen assembly membrane complex required for LPS biosynthesis.
TMHMM was utilized to analyse the secondary structure of WaaL from other Gram-negative bacteria. We found that all WaaL proteins possess a putative periplasmic loop region: 84 amino acids in E. coli, 88 amino acids in Vibrio cholera, and 73 amino acids in Salmonella typhimurium. Interestingly, Pseudomonas spp. do have the predicted periplasmic loop region in WaaL, but the size of the periplasmic loop region is small compared with E. coli, V. cholera or S. typhimurium. P. aeruginosa PAO1 WaaL contains 49 amino acids and PA14 contains 48 amino acids in their periplasmic loop regions. However, Schild et al. (2005) reported that the periplasmic loop region of V. cholera WaaL contained 103 amino acids by LacZ/PhoA assays. This suggested that there may be more or less amino acids in this region of the actual structure (Schild et al., 2005).
Comparison of the large periplasmic loop region of WaaL revealed five conserved amino acid residues. By using alanine-scanning mutagenesis, we discovered that H303 was crucial for the ligase activity. To further substantiate our observation of the effect of the H303A mutant, we constructed another mutant H303G and found that the latter also totally abolished the ligase activity of WaaL. Histidine residues carry a positive charge at neutral pH, making them an ideal amino acid to participate in proton transfer in the catalytic domain or for static interactions with substrates in binding pockets. H303R is a conserved mutant because Arg, which replaces His, is also positively charged at or near neutral pH. Indeed, this mutation did not affect the activity of WaaL. Further, our ATPase assay and in vitro ligase results showed that H303R behaved similarly to the wild-type WaaL. Interestingly, the H303N affected WaaL activity and this mutant was not able to complement the waaL knockout mutant. Thus, although Asn could facilitate hydrogen bonding with the substrate at that location, it does not carry a positive charge at neutral pH. Based on these observations, we speculate that the positive charge of H303 may play a significant role in the static interactions with negatively charged phosphate groups on the lipid A-core molecules. Alternatively, the periplasmic loop region of WaaL may act as a scaffold to direct the O antigen onto the lipid A-core acceptor molecules for the ligation process. Our data showed that H303 with non-conserved amino-acid substitutions leads to complete abolition of the O-antigen ligase activity. A recent study of WaaL of V. cholerae has shown that amino acids H321 and R208, which are located in the periplasmic region, are important in ligase activity (Schild et al., 2005).
Cross-complementation of P. aeruginosa waaL::Gmr with E. coli waaL was not successful. In contrast, our previous cross-complementation work has shown that the waaL of strain PAO1 and strain PA14 are interchangeable (Abeyrathne et al., 2005). This shows that WaaL proteins are not interchangeable between Enterobacteriaceae and Pseudomonadaceae organisms, and residues specific to the WaaL of P. aeruginosa up to the region containing 1–10 TMS in WaaL. Despite similar hydropathy plots between WaaL of these two genera, there might be structural constraints for WaaL of a particular species that favours recognition of the specific core-oligosaccharide acceptors and the unique structures of O-polysaccharide substrates.
Detailed assessment of mechanisms of the catalytic function of WaaL has to await the solving of the WaaL structure at a reasonable resolution, which might be facilitated by co-crystallization of WaaL with ATP and/or LPS. Our group is currently working on these tasks.
In conclusion, we have successfully purified WaaL, which is an integral membrane protein with 11 potential TMS belonging to a family of O-antigen ligase enzymes. The use of a wbpL mutant as the expression host allowed the purification of WaaL to yield a product that is free of contaminating A-band or B-band LPS polymers. The requirement of ATP and Mg2+ in O-antigen ligase activity was observed. In addition, we have revealed important key residues in the O-antigen ligation process, and most importantly, an in vitro assay has been developed to determine the O-antigen ligase activity in WaaL. These results will impact our understanding of the LPS assembly process in Gram-negative bacteria.
Bacterial strains and culture conditions
Plasmids and strains used in this study are listed in Table 1. All strains were grown in Luria–Bertani broth (Invitrogen Canada, Burlington, ON) at 37°C. The following antibiotics were used in selection media at the indicated concentrations: ampicillin (Amp) at 100 μg ml−1 for E. coli, kanamycin (Kan) at 50 μg ml−1 for E. coli, and tetracycline (Tc) at 10 and 100 μg ml−1 for E. coli and P. aeruginosa respectively. All the primer sequences used in this study will be provided upon request.
Full-length waaL cloned into pVLT31 using EcoRI/PstI restriction sites
A 303 bp waaL-insert (aa 1–101) cloned into pUCP27 using SstI/PstI restriction sites
A 468 bp waaL-insert (aa 1–156) cloned into pUCP27 using SstI/PstI restriction sites
A 696 bp waaL-insert (aa 1–232) cloned into pUCP27 using SstI/PstI restriction sites
A 903 bp waaL-insert (aa 1–301) cloned into pUCP27 using SstI/PstI restriction sites
A 1059 bp waaL-insert (aa 1–353) cloned into pUCP27 using SstI/PstI restriction sites
A 699 bp PAO1/waaL-insert (aa 1–233) and PAO1/waaL amino acids 233–419 from E. coli/waaL cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acids 1–306 PAO1/waaL amino acids 344–419 from E coli/waaL cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acids 1–360 PAO1/waaL amino acids 389-419 from E coli/waaL cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acid R264A waaL/PAO1 cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acid W268A waaL/PAO1 cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acid G282A waaL/PAO1 cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acid G284A waaL/PAO1 cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acid H303A waaL/PAO1 cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acid H303R waaL/PAO1 cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acid H303G waaL/PAO1 cloned into pUCP27 using EcoRI and PstI restriction sites
Amino acid H303N waaL/PAO1 cloned into pUCP27 using EcoRI and PstI restriction sites
E. coli waaL cloned into pUCP27 using SstI and PstI restriction sites
Plasmid DNA was electroporated into P. aeruginosa with a Gene Pulser instrument (Bio-Rad). Genomic DNA was isolated from both P. aeruginosa PAO1 (Hancock and Carey, 1979) and E. coli K12 by the method of Ausbel et al. (1997). The method of Horton et al. (1993) was used to make site-directed mutants and chimeric constructs. Laboratory Service Division at University of Guelph was used to sequence the DNA and to purchase primers.
Lipopolysaccharide isolation and analysis by SDS-PAGE and Western blotting
The proteinase-K digestion method of Hitchcock and Brown (1983) was used to prepare the LPS. Isolated LPS was subjected to SDS-PAGE and visualized by silver staining using the rapid method of Fomsgaard et al. (1990). Western blotting was carried out using mAb A-band-specific N1F10 and B-band-specific MF15-4 as primary antibodies. The secondary antibody was an anti-goat mouse F(ab′)2-alkaline phosphatase conjugate (Jackson ImmunoResearch) diluted to 1:2000 according to the manufacturer's suggestion.
Expression and purification of WaaL
waaL was cloned into the pVLT31 (de Lorenzo et al., 1993) expression vector with an N-terminal histidine tag. Protein expression was carried out in a homologous P. aeruginosa PAO1 background (de Lorenzo et al., 1993; Ward et al., 2000). Cells were induced with isopropyl-thio-β-d-galactopyranoside at a final concentration of 1 mM, and protein expression was continued for 8 h at 37°C. Expression of WaaL was confirmed by Western blotting (Abeyrathne and Lam, 2007). Membrane pellets containing expressed WaaL were solubilized in 1% DDM (Sigma, St Louis, MO) and purified by affinity column chromatography with TALON chelating resin. Purified protein was quantified by the Bradford assay (Bio-Rad, Mississauga, ON). For in vitro ligase assay experiments, WaaL was expressed in a wbpL knockout mutant strain (wbpL::Gmr) that is devoid of both A-band and B-band O antigens (Rocchetta et al., 1998) to avoid contamination by these O antigens or O-polysaccharide-Und-P molecules when WaaL is being purified.
Matrix-assisted Laser Desorption Ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis
The method of van Montfort et al. (2002) was used to digest WaaL for MALDI-TOF MS analysis. After visualization, the bands containing the protein were excised from SDS-PAGE gels and completely destained with 50 mM ammonium hydrogencarbonate in 40% ethanol. The excised gel pieces were washed with 200 μl of 25 mM ammonium hydrogencarbonate three times for 15 min, and gel pieces were dehydrated with 100 μl of acetonitrile three times for 10 min and completely dried with a Speed Vac. Tryptic digestion was started with the addition of 5 μl of 75 ng μl−1 trypsin in 25 mM ammonium hydrogencarbonate buffer to the dried gel pieces. After swelling, the pieces were covered with an overlay of ∼20 μl of 25 mM ammonium hydrogencarbonate so that the gel pieces remained immersed throughout the digestion. The protein samples were digested for at least 14 h at 30°C without agitation. CNBr cleavages were performed for at least 14 h in the dark at room temperature by adding 25 μl of CNBr in 70% trifluoro acetic acid (TFA). The combined tryptic and CNBr cleavage was performed as follows. After tryptic digestion, the gel pieces were dried in Speed Vac. Dried gel pieces were washed and dehydrated twice by addition of 100 μl acetonitrile and followed by the addition of 25 μl of CNBr in 70% TFA as above. MS analysis was performed using a Reflex III MALDI-TOF instrument (Brüker, Germany) equipped with a 337 nm nitrogen laser.
Transmission electron microscopy
An electron microscopy (EM) sample grid was submerged in the purified WaaL for 10 s, blotted onto filter paper to remove any fluid from the grid, and negative stained with 1% aqueous uranyl acetate. WaaL on the EM grid was examined by using a Philips EM300 transmission electron microscope operating at 60 kV under standard conditions with the cold trap in place.
ATPase and ligase assays
The ATPase assay was an adaptation of the 5′-NT colorimetric Pi assay, originally described by Chifflet et al. (1988). In brief, 0.5 μg (11 pmol) of purified WaaL was used in a 100 μl reaction volume containing 20 mM Tris, 100 mM NaCl, 10 mM MgCl2 pH 7.5. The amounts of ATP used in entire experiments are as follows: 0–20 mM for establishing standards curve, and 2 or 4 mM respectively for in vitro ligase assay experiments. The reactions were incubated at 37°C for 45 min and stopped by the addition of 100 μl of 12% (w/v) SDS, 3% (w/v) ascorbic acid and 1% (w/v) ammonium molybdate tetrahydrate. After 5 min incubation at room temperature, 100 μl of a solution containing 2% (v/v) acetic acid, 2% (w/v) sodium arsenite and 2% sodium citrate was added. The reaction mixture was incubated for an additional 20 min, and the absorbance was measured at 750 nm.
For the in vitro ligase assay, crude LPS from the P. aeruginosa PAO1 waaL::Gmr strain was prepared by a modified Hitchcock and Brown method (1983). Briefly, an overnight culture of the P. aeruginosa PAO1 waaL::Gmr strain (25 ml) was centrifuged at 5 000 g for 5 min at 20°C. The cell pellet was washed with 50 ml of sterile saline and sedimented by repeating the centrifugation step. The cell pellet was then resuspended in 5 ml of Hitchcock and Brown lysis buffer (2% SDS, 10% glycerol and 1 M Tris-HCl pH 6.8) and heated at 100°C for 30 min. Then the sample was cooled to 20°C and 75 μl of proteinase K (from a 20 mg ml−1 stock solution) was added, and the mixture was incubated at 56°C for 3 h. At the end of this treatment, proteinase K was inactivated by heating at 95°C for 30 min The isolated crude LPS was quantified by 3-deoxy-d-manno-octylosonic acid (Kdo) assay (Droge et al., 1970; Skoza and Mohos, 1976). The mole-to-mass ratio of LPS was estimated based on the stoichiometry of having two Kdo residues per P. aeruginosa LPS molecule. The free Kdo would then react with thiobarbituric acid to give rise to the chromophore that facilitated spectrophotometric detection based on an optimum extinction coefficient at wavelength 552 nm. The Kdo that is glycosidically linked to heptose remained intact and would not react with the thiobarbituric acid. Therefore, the molar ratio of Kdo to LPS molecule was calculated as 1:1. The ligase assay was based on the ATPase assay in the presence of O polysaccharide. The reactants included purified WaaL, and the crude LPS as the substrate, and ATP as the cofactor. Separate sets of experiments were performed to determine the minimal amount of substrate required for the in vitro ligase assay by using different amount of substrate. A 5 μM substrate concentration was deemed optimum. The concentration of substrate used in the in vitro ligase assay was 5 μM, as well as two different ATP concentrations (2 or 4 mM) and 11 pmol of purified WaaL in 100 μl of reaction volume. The reaction mixtures were incubated at 37°C for 12 h. To verify that each of the major components in the reaction mix was important for the O-ligase assay, the following controls were used: (i) omission of ATP, (ii) omission of substrate, and (iii) omission of WaaL. To further confirm that the banding pattern observed in SDS-PAGE gels after in vitro ligation was due to the actual enzyme–substrate reaction and not due to the degradation of WaaL, we added 0.03 mg ml−1 proteinase K into the in vitro ligation reaction mixture and continued incubation at 56°C for 4 h before running samples on SDS-PAGE.
This work was supported by an operating grant from the Canadian Cystic Fibrosis Foundation (CCFF). The authors wish to thank Diane Moyles for performing electron microscopy on the purified WaaL, Dyanne Brewer for assistance with MALDI-TOF MS analysis, Jean-Louis Riguad (Institute Curie, France) and Anchi Cheng (Scripps Institute, USA) for their encouragement and the inspiration of working with membrane proteins, Hamed Ghanei for his insightful suggestions, Rod Merrill for helpful discussions on the site-directed mutagenesis, Frances Sharom for discussions with us on ATPase activities and allosteric substrate inhibition mechanisms, and Craig Daniels, Andrew Kropinski and Kasia Kaluzny for critically reading the manuscript. P.D.A. is a recipient of a CCFF fellowship, and J.S.L. holds of a Canadian Research Chair in Cystic Fibrosis and Microbial Glycobiology.