|Home | About | Journals | Submit | Contact Us | Français|
Pseudomonas aeruginosa in the lungs of cystic fibrosis (CF) patients is characterized by a series of genotypic and phenotypic changes that reflect the transition from acute to chronic infection. These include the overproduction of the exopolysaccharide alginate and the loss of complete lipopolysaccharide (LPS). LPS is a major component of the Gram-negative outer membrane and is composed of lipid A, core oligosaccharide, and O antigen. In this report, we show that the LPS defect of the P. aeruginosa chronic infection isolate 2192 is temperature sensitive. When grown at 25°C, 2192 expresses serotype O1 LPS with a moderate chain length and in reduced amounts relative to those of a wild-type serotype O1 laboratory strain (stO1). In contrast, 2192 expresses no LPS O antigen when grown at 37°C. This is the first time that a temperature-sensitive defect in O-antigen production has been reported. Using complementation analyses with a constructed wbpM deletion mutant of stO1, we demonstrate that the temperature-sensitive O-antigen production defect in 2192 is due to a mutation in wbpM, which encodes a UDP-4,6-GlcNAc dehydratase involved in O-antigen synthesis. The mutation, a deletion of a single amino acid (V636) from the extreme C terminus of WbpM, renders the protein less stable than its wild-type counterpart. This residue of WbpM, which is critical for stability and function, is located outside of the recognized domains of the protein and may provide insight into the structure-function relationship of this enzyme, which is found in all 20 serotypes of P. aeruginosa. We also identify a promoter of wbpM, map a transcriptional start site of wbpM, and show that mucoidy plays a role in the loss of expression of high-molecular-weight LPS in this CF isolate.
Pseudomonas aeruginosa is an opportunistic Gram-negative pathogen that is ubiquitous in the environment. It infects a variety of immunocompromised populations, including HIV patients, burn victims, and cystic fibrosis (CF) patients, and it is the fourth most commonly acquired nosocomial pathogen (1). Lipopolysaccharide (LPS) is a major component of the P. aeruginosa outer membrane and is an important virulence factor, aiding in attachment to host tissues and providing protection from complement-mediated killing (2, 3). P. aeruginosa LPS is a tripartite molecule consisting of the lipid A moiety, which is embedded in the outer leaflet of the outer membrane, a core oligosaccharide, and the O-antigen repeating unit, which extends outward from the cell. Two types of LPS are expressed on the surface of P. aeruginosa: common antigen and O antigen (also referred to as A-band and B-band LPS, respectively), which differ in their side chains. The common antigen side chain is a polymer of d-rhamnose, while O-antigen LPS consists of serotype-specific side chains, the structure of which defines the serotype of the strain of P. aeruginosa (4). Currently, 20 different International Antigenic Typing System (IATS) serotypes are recognized, and all of the O-antigen structures have been determined (5).
P. aeruginosa has a large genome (~6 to 7 Mb), reflecting its ability to occupy a wide variety of environmental niches, including the lungs of CF patients (6). It is thought that conditions encountered by P. aeruginosa in the lungs of CF patients select for a series of genetic and thus phenotypic changes that cause the infection to progress from an acute to a chronic state (7). This, along with a notable level of intrinsic and mutationally acquired antibiotic resistances, renders these infections extremely difficult to treat and, in many cases, impossible to eradicate. As such, P. aeruginosa infection of the lung is the leading cause of morbidity and mortality among CF patients (7). The phenotypes associated with chronic infection include the downregulation of exoenzymes, flagella, pili, and the type III secretion machinery, as well as the upregulation of expression of the exopolysaccharide alginate (referred to as the “mucoid” phenotype), a switch from a planktonic to a biofilm mode of growth, and a transition from having complete LPS to having LPS that lacks the O-antigen side chains (7, 8). The defect in O-antigen production in these chronic infection isolates is typically due to “loss-of-function” mutations in O-antigen biosynthetic genes that are thought to confer a selective advantage on lineages in the population of the lung as the infection progresses to a chronic state (9). The lack of O-antigen LPS in these strains renders them nonserotypeable and serum sensitive. P. aeruginosa strain 2192 is a chronic infection isolate, recovered from a CF patient, that has many of these chronic infection phenotypes, including mucoidy (10) and serum sensitivity, due to a lack of O-antigen LPS (3, 11).
Our group previously reported the repair of the LPS O-antigen defect of strain 2192 by providing the entire serotype O11 O-antigen gene cluster in trans on plasmid pLPS2 (12). 2192(pLPS2) produced both O11 and O1 O antigens, and its resistance to serum was restored (11). The fact that both O11 and O1 O antigens were produced in 2192(pLPS2) suggests that 2192 has the O1 O-antigen biosynthetic locus and that a mutation in a gene shared by the O11 and O1 loci is responsible for the 2192 LPS defect. With this in mind, we searched the published genome of 2192 (13), and we recognized the serotype O1 O-antigen locus. Serotype O1 is a clinically relevant serotype (14), and its O-antigen subunit structure is N-acetylgalactosamine, 2-acetamido-3-amino-2,3-dideoxy-d-glucuronic acid, N-acetyl-d-fucosamine, and N-acetyl-d-quinovosamine [(→4)-d-GalNAc-(1→4)-β-d-GlcNAc3NAcA-(1→3)-d-FucNAc-(1→3)-d-QuiNAc-(1→] (5).
Here we report several changes in the O-antigen locus of 2192 relative to the published representative P. aeruginosa serotype O1 sequence (15). Moreover, we demonstrate that a single deleted valine (V636) residue in WbpM is responsible for the O-antigen defect in strain 2192. We also show that wbpM-complemented 2192 lacks high-molecular-weight (HMW) LPS, corroborating a recent finding that mucoid strains of P. aeruginosa do not express HMW LPS (16). In order to assess the importance of wbpM in a wild-type O1 strain, we generated a wbpM deletion mutant in stO1 (the ΔwbpM mutant). This strain lacks O antigen, and this defect can be complemented with the wild-type wbpM gene expressed in trans but not with the wbpM gene from 2192. Real-time quantitative PCR (RT-qPCR) revealed that both stO1 and 2192 express wbpM, suggesting that the lack of O-antigen expression is not due to a transcriptional defect in 2192. We also show, using lacZ fusions, that there is a promoter upstream of wbpM in stO1and 2192. Further characterization revealed that WbpM lacking V636 is less stable than the wild-type protein and that this protein is at least partially functional in vivo; when grown at 25°C, strain 2192 expresses some O antigen with intermediate chain lengths.
The strains and plasmids used in this study are listed in Table 1. The primers used for complementation and sequencing analyses are available upon request. All Escherichia coli and P. aeruginosa strains were grown at 25°C or 37°C with shaking at 200 rpm in Lysogeny Broth (LB) (1% tryptone, 0.5% yeast extract, 0.5% NaCl [wt/vol]), supplemented with antibiotics as follows: for E. coli, 100 μg carbenicillin ml−1, 15 μg gentamicin ml−1, or 10 μg tetracycline ml−1; for P. aeruginosa, 500 μg carbenicillin ml−1, 250 μg gentamicin ml−1, 25 μg spectinomycin ml−1, or 100 μg tetracycline ml−1.
The O-antigen locus from strain 2192 (spanning open reading frames [ORFs] PA2G_2553.1 to PA2G_02568.1) from the genome sequenced at the Broad Institute (http://www.broadinstitute.org/annotation/genome/pseudomonas_group/MultiHome.html) was compared to the serotype O1 O-antigen locus (15) using BLASTn. The sequence of the 3′ region of wbpM was determined by first amplifying an ~500-bp region from each of the 20 IATS P. aeruginosa strains and 2192 with primers wbpM-Mid-F (CCGAGAAGATGATCCACCTGT) and wbpM-Mid-R (CACCAGGCCTGGTGTGACGTTTCAG). These amplicons were run on a 1% agarose gel and were purified using a gel extraction kit (Qiagen). Purified amplicons were sent to GENEWIZ for sequencing of the region that contains the codon for V636 by using primer wbpMV636 (TTGGTCATCCAGGCCGGTTC), and the sequences were analyzed using the Sequencher 4.0 software package (Gene Codes Corporation). ClustalW (http://www.ebi.ac.uk/Tools/msa/clustalw2/) was used to generate multiple sequence alignments.
P. aeruginosa strains were grown overnight at 25°C or 37°C in LB containing carbenicillin. Cells were adjusted to an optical density at 600 nm (OD600) of 0.5, and 1 ml of OD-adjusted cultures was used for LPS preparation, as described by Davis and Goldberg (21). Briefly, the cells were pelleted at 20,800 × g, and the supernatant was removed. The pellets were resuspended in 200 μl of 1× sodium dodecyl sulfate (SDS) buffer (0.1 M Tris-HCl [pH 6.8], 2% β-mercaptoethanol [wt/vol], 2% SDS [wt/vol], 10% glycerol [wt/vol]) and were boiled for 15 min. After the samples had cooled, 10 μg DNase ml−1 and 10 μg RNase ml−1 were added, and the samples were incubated at 37°C for 30 min. Proteinase K was added to a final concentration of 10 μg ml−1, and the samples were incubated overnight at 59°C. Two hundred microliters of 2× SDS buffer was added to the samples. For Western blotting, 10 μl was run on an 8% or 12% polyacrylamide gel and was transferred to a nitrocellulose membrane using the Trans-Blot Cell (Bio-Rad). Blots were analyzed using polyclonal P. aeruginosa serotype O1 antiserum (Accurate Chemical & Scientific). The secondary antibody was goat anti-rabbit immunoglobulin G coupled to either horseradish peroxidase (Sigma-Aldrich) or IRDye 680 (Li-Cor). For direct staining of LPS in gels, LPS preparations were made as described above, except that overnight cultures were adjusted to an OD600 of 1.5 before LPS was isolated.
P. aeruginosa strains were grown overnight in LB containing carbenicillin and were back-diluted 1:20 into LB containing carbenicillin. After 2 h of growth, 2% arabinose (wt/vol) was added for induction. Induction was allowed to proceed for 2 h; then the cultures were pelleted at 12,857 × g for 10 min, washed in LB, and resuspended in LB containing carbenicillin and 1% glucose (wt/vol) for inhibition of transcription from the PBAD promoter on pHERD20T. Samples were taken at hourly intervals and were adjusted to an OD600 of 0.5. One milliliter of OD-adjusted cultures was used for preparation of whole-cell extracts. The cells were pelleted at 20,800 × g, and the supernatant was removed. The pellets were resuspended in 200 μl of 1× SDS buffer and were boiled for 15 min. One hundred microliters of the boiled samples was combined with 100 μl of 2× SDS buffer. Ten microliters was run on an 8% acrylamide gel and was transferred to a nitrocellulose membrane. The blots were analyzed using the monoclonal anti-Flag antibody M2 (Sigma) or monoclonal anti-RpoA antisera (NeoClone). The secondary antibody used was goat anti-mouse immunoglobulin G coupled to IRDye 680 (Li-Cor). The Odyssey infrared (IR) detection system was used to scan the Western blots.
In order to generate complementing plasmids, primers that flanked the predicted ORFs of the genes encoding the proteins listed in Table 2 were designed. Restriction sites were engineered to the 5′ ends of the primers to allow for cloning into pUCP18Ap (20). Once amplified, the products were cloned into plasmid pCR-TOPO2.1 (Invitrogen) and were transformed into E. coli TOP10 (Invitrogen). These plasmids were then digested with appropriate primer-encoded restriction enzymes, ligated into similarly digested pUCP18Ap, and transformed into E. coli TOP10. Plasmids were prepared and transferred into strain 2192 by electroporation. Briefly, 1 ml of an overnight culture of 2192 was pelleted, the supernatant was removed, and the pellet was washed with 1 ml of distilled water (dH2O). The pellet was washed twice more with 1 ml of dH2O and was resuspended in 300 μl of dH2O, and 100 μl of the resuspended cells was added to an electroporation cuvette along with approximately 300 ng of plasmid DNA. The cells were then electroporated at 1,800 V using an Eppendorf electroporator, model 2510. Electroporated cells were recovered by shaking at 200 rpm in 500 ml of LB at 37°C for 1 h. pHERD20T-wbpM constructs were made similarly to the pUCP18Ap constructs; the product was amplified, cloned into pCR-TOPO2.1, digested, and ligated into a similarly digested pHERD20T plasmid (19). For Flag-tagged WbpM, a similar cloning strategy was used, and the Flag sequence was added to the reverse primer, resulting in a C-terminal Flag tag.
A mutant allele of wbpM was generated using splice-by-overlap extension (SOE) PCR; the procedure was modified from that described by Horton et al. (22). Briefly, ~500-bp regions flanking wbpM were amplified. One of each of the primers in the two sets contained a 5′ tail that shares homology with a FLP recombination target (FRT)-flanked gentamicin (Gm) cassette. In a second SOE PCR, the two fragments were mixed with the FRT-gentamicin cassette, and the external primers were used to splice the three fragments together, resulting in the fragment wbpM::Gm. This fragment was cloned into pCR-TOPO 2.1. The wbpM::Gm fragment was then digested from pCR-TOPO2.1 and was ligated into a similarly digested gene replacement vector, pEX18T (18). Triparental matings were performed as described by Goldberg and Ohman (23) to mobilize pEX18T-wbpM::Gm into P. aeruginosa serotype O1 (strain stO1). Equal volumes of overnight cultures of P. aeruginosa, E. coli DH5α (Invitrogen), and E. coli HB101 (Invitrogen) with helper plasmid pRK2013 were pelleted, mixed in a volume of 30 μl of LB, and then spotted onto tryptic soy agar plates (Remel) and grown overnight. Transconjugants were selected on LB agar plates containing gentamicin and spectinomycin and were passaged onto a medium containing gentamicin and 5% sucrose (wt/vol) to select against strains containing the plasmid. Genomic DNA was prepared from strains containing the appropriate resistance profile (carbenicillin sensitive, gentamicin resistant, sucrose resistant), and PCR was used to verify the presence of the wbpM::Gm allele. In order to remove the gentamicin cassette, the plasmid encoding the FRT flippase, pFLP2 (18), was electroporated into a wbpM::Gm strain and was subjected to selection on LB agar plates containing 500 μg carbenicillin ml−1. Genomic DNA was prepared from strains showing loss of gentamicin resistance, and PCR was used to verify the loss of the Gm cassette. Serial passaging onto tryptic soy agar (TSA) plates was used to cure pFLP2 from the resultant wbpM deletion mutant (referred to as the ΔwbpM mutant). The deletion in this strain was verified by PCR and sequencing.
An approximately 1 kb sequence located directly 5′ of the wbpM start codon was scanned for promoters by using the neural network promoter prediction program BPROM (Softberry).
Total RNA (8 μg) from P. aeruginosa strain stO1 or the ΔwbpM mutant was isolated using the SV total-RNA isolation kit (Promega, Madison, WI). Oligonucleotide wbpMPE1 (GCAAGATCCTTTTATAACGACGCGGCAGCC) was end labeled with [γ-32P]ATP by using polynucleotide kinase and was used in a reverse transcription reaction with avian myeloblastosis virus (AMV) reverse transcriptase (Promega). A temperature of 45°C was used for the extension reaction. The same oligonucleotide used for primer extension was also employed in a sequencing reaction using the wbpM gene and upstream sequence cloned into pHERD30T by use of the Sequenase 7-deaza-dGTP kit (USB, Cleveland, OH). Both primer extension and sequencing reactions were run on a 6% denaturing acrylamide gel. The gel was dried and extension products were detected using autoradiography.
A wbpMp1 (P1)-lacZ fusion construct was made by first amplifying an ~500-bp fragment containing P1 and ~400 bases of wbpM with primers containing restriction sites on their 5′ ends. This fragment was cloned into pCR2.1-Topo. It was then digested out of pCR2.1-Topo and was ligated into a similarly digested P. aeruginosa integration vector, mini-CTX-lacZ (17). Triparental matings were used to mobilize the resultant plasmid into stO1 or 2192. Transconjugants were selected on LB agar containing 100 μg tetracycline ml−1 and 25 μg spectinomycin ml−1. In order to remove the plasmid backbone from the chromosome, pFLP2 was electroporated into positive strains, which were subjected to selection on LB agar supplemented with 500 μg carbenicillin ml−1. Colonies showing the proper resistance profile (tetracycline sensitive, carbenicillin resistant) were screened via PCR for the presence of the P1-lacZ fusion. Positive colonies were cured of pFLP2 by serial passaging on TSA plates.
P. aeruginosa P1-lacZ strains were grown to the exponential and stationary phases in LB; the OD600 was recorded, and 500-μl aliquots were taken. These aliquots were centrifuged at 20,800 × g, and the supernatant was removed. LacZ activity was quantified using a method modified from that of Miller (24). Cell pellets were resuspended in 850 μl of Z buffer (0.06 M Na2HPO4, 0.04 M NaH2PO4, 0.01 M KCl, 0.001 M MgSO4, 0.05 M β-mercaptoethanol [pH 7.0]), lysed by adding 100 μl of chloroform and 50 μl of 0.1% SDS, and mixed briefly using a vortex. Following lysis, 200 μl of a solution containing 4 mg ortho-nitrophenyl-β-galactoside ml−1 was added. When a sufficiently yellow color was obtained, reactions were stopped by adding 250 μl of 2 M sodium acetate (NaOAc), and the times of the reactions were noted. Cellular debris was removed by centrifuging the samples at 20,800 × g for 5 min. The supernatant was used for recording OD420 readings, and Miller units were calculated using the formula 1,000 × [OD420/(T · V · OD600)], where T is the reaction time and V is the volume of the culture used.
P. aeruginosa cultures were grown to mid-log phase and were normalized to an OD600 of 0.5 in 10 ml. Cells were centrifuged at 10,620 × g, and the pellet was washed in ice-cold STE buffer (10 mM Tris [pH 7.4], 100 mM NaCl, 1 mM EDTA). Cells were resuspended in 100 μl RNA prep buffer (0.5 M NaCl, 0.1 M Tris [pH 7.5], 10 mM EDTA, 1% SDS) and were lysed by adding 100 μl of phenol-chloroform-isoamyl alcohol (PCIAA) along with glass beads and agitating with a vortex five times each, for 30 s each time, with cooling on ice between rounds. After lysis, 150 μl (each) RNA prep buffer and PCIAA was added, and samples were mixed using a vortex a final time for 30 s and were centrifuged at 16,100 × g and 4°C for 5 min. The upper aqueous layer was removed; then samples were extracted with 150 μl PCIAA and centrifuged as described above, and the upper aqueous layer was reserved. Nucleic acids were precipitated by adding 3 volumes of ice-cold absolute ethanol to the extracted samples and placing them at −20°C overnight. The precipitated nucleic acids were pelleted by centrifugation at 16,100 × g and 4°C for 10 min, washed with 70% ethanol, and repelleted. The pellets were air dried, resuspended in 100 μl of diethyl pyrocarbonate (DEPC)-treated water (Sigma), and DNase treated by using a Turbo DNA-free kit (Ambion) according to the manufacturer's protocol. cDNAs were generated from 1 μg of total RNA with random hexamers by using the TaqMan reverse transcription kit (Applied Biosystems) according to the manufacturer's instructions in a total volume of 50 μl. Twenty-microliter qPCR mixtures were set up using 1 μl of cDNA, 10 μl of 2× FastStart Universal SYBR green qPCR master mix (Roche), and 4 pmol of each primer. qPCRs were performed in 96-well plates in an ABI Prism 7900HT Fast real-time PCR thermocycler. The program used consisted of an initial 10-min incubation at 95°C, followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. Primer specificity was verified by use of a denaturation step following the last amplification cycle. Data were analyzed as described by Morton et al. (25), and threshold cycle (CT) values were collected with a manual threshold of 0.2. Each target was tested in triplicate, and the average of the 4 CT values was used for analysis. The average CT was converted to a relative transcript number by the equation n = 2(40-Ct). These values were then standardized to the value determined for omlA and were reported as the number of target transcripts per 10 omlA transcripts.
Previous reports from our laboratory showed that 2192 lacks specific O antigen and that O-antigen expression in this strain can be restored by providing the serotype O11 O-antigen locus (pLPS2) in trans (11). These previous studies were performed using 2192 grown at 37°C and showed that 2192 expressed both serotype O1 and serotype O11 LPSs when provided with pLPS2 (11). While serotype O1 is a clinically relevant serotype (14), LPS synthesis in this serotype has not been well studied; however, LPS synthesis in other P. aeruginosa serotypes has been well characterized (26–28), and a similar mechanism for O1 synthesis has been proposed in a recent review by King et al. (4).
We performed Western blot analyses with serotype O1 antisera on LPS isolated from 2192 at both 25°C and 37°C, and we compared the results to those for LPS isolated from IATS serotype O1 (strain stO1) grown at both temperatures. As expected, LPS O antigen was detected in LPS isolated from stO1 at both temperatures. Interestingly, some serotype O1 O antigen was seen in 2192 grown at 25°C, but not at 37°C (Fig. 1). The O antigen detected in 2192 grown at 25°C was not wild type in terms of the amount or chain lengths detected. Specifically, 2192 grown at 25°C expressed serotype O1 LPS with an intermediate chain length and at a reduced amount relative to that of stO1.
In an attempt to identify the mutation(s) responsible for the LPS defect of strain 2192, we analyzed the genome sequence of 2192 that was made available by the Broad Institute (13). We identified the O-antigen locus by using BLAST searches and determined that it contained the stO1 O-antigen gene cluster (Fig. 2) as identified by Raymond et al. (15), with several nucleotide changes that altered the amino acid sequence. Since previous studies had suggested that the O-antigen defect of 2192 was based on a genetic mutation (11), we compared the O-antigen locus of 2192 to the serotype O1 O-antigen sequence. Comparison to the reported stO1 nucleotide sequence revealed six nucleotide changes in genes that were predicted to alter the amino acid sequence from the predicted wild-type stO1 protein sequence (Table 2). To verify that these differences were not misannotations that occurred during the sequencing process, we amplified and sequenced ~300-bp regions from both stO1 and 2192 that included the nucleotide sequence differences identified (data not shown). These sequences were compared to the reported stO1 sequence and to the published 2192 genomic sequence by using BLASTn. All of the differences in the published 2192 genome were verified except for one: a predicted nonsense mutation in wbpE. The sequence of the PCR amplicon that we generated from 2192, which was purported to contain the nonsense mutation, revealed no change relative to the nucleotide sequence of stO1. Thus, it appears that the report of a mutation in wbpE is actually due to a misannotation in the published 2192 sequence (13). Finally, the region spanning the 3′ half of wbpM and tyrB, the first gene following the O-antigen locus, was amplified and sequenced from both stO1 and 2192, because this region was not included in the O-antigen locus sequences reported by Raymond et al. (15). An additional mutation in wbpM predicted to alter the WbpM protein sequence in 2192 was found, bringing the total predicted differences in ORFs between stO1 and 2192 to six, discounting wbpE (Table 2). We hypothesized that one or more of these mutations may be responsible for the O-antigen defect in strain 2192.
Given that 2192 expressing the entire O11 O-antigen locus in trans expressed both O11 and O1 LPSs (11), we hypothesized that the mutation in 2192 that resulted in its LPS defect was likely in a gene within its O-antigen locus. In order to determine which of the differences was responsible for this defect, we first cloned wild-type copies of each of the ORFs that were altered in 2192 relative to stO1 into plasmid pUCP18Ap, then transferred these recombinant plasmids to 2192, and tested each of these strains for O-antigen expression at 25°C and 37°C.
Only one of the ORFs was able to rescue the O antigen defect of strain 2192 at both temperatures: stO1 wbpM (wbpMO1) (Fig. 3). Moreover, providing the 2192 version of wbpM (wbpM2192) to strain 2192 did not alter its O-antigen expression at either temperature (Fig. 3). Taken together, these data suggest that none of the other differences reported in Table 2 contribute to the lack of O-antigen expression in strain 2192 and that wbpMO1 alone is sufficient to restore O1 O-antigen expression in 2192. In addition to our Western blot data, compositional analyses of the LPSs expressed by stO1 and 2192(pUCP18Ap-wbpMO1) showed that the O antigens of these strains were identical (see Table S1 in the supplemental material).
Each of the 20 specific O-antigen biosynthetic loci have been sequenced (15), and while there is variation between the content and organization of their O-antigen loci, wbpM is conserved among all of the serotypes and is in a conserved position as the last gene of the locus in all serotypes, making it a good target for study in order to gain insights into LPS synthesis by P. aeruginosa strains of all 20 serotypes.
The wbpM gene from 2192 was not able to restore wild-type levels of LPS expression in strain 2192. While this suggests that WbpM2192 is not functional, we could not rule out the possibility that when 2192 is grown at 37°C, there is a transcriptional defect in wbpM2192 that may contribute to its inability to restore O antigen. Further, there are differences between 2192 and stO1 in the intergenic region between wbpL and wbpM, as well as silent changes in the coding region of the upstream gene wbpL in 2192. Thus, it was possible that one or more of these differences might be localized to regulatory regions and might be responsible for improper transcription of wbpM2192, which could account for the lack of O antigen in 2192. In order to verify that both wbpMO1 and wbpM2192 were transcribed at 37°C, RNA was isolated from stationary-phase cultures of 2192 and stO1 grown at 37°C, and RT-qPCR was performed. Both stO1 and 2192 had detectable wbpM cDNA, suggesting that the loss of O antigen by 2192 grown at 37°C was not due to a lack of wbpM transcription. In fact, strain 2192 had significantly more wbpM cDNA than stO1 (Fig. 4A). In addition to quantifying the amount of wbpM present in these strains, we also performed primer extension analysis on wbpM in stO1 in order to map the nearest transcriptional start site (+1 site). We mapped this to a guanine residue located 44 bases upstream of the predicted wbpM start codon (Fig. 4B).
While wbpM was transcribed in both strains, the fact that 2192 had significantly more wbpM RNA than stO1 suggested that there may be differential regulation of a promoter that controls wbpM expression in these two strains. We used BPROM promoter prediction software to identify potential promoters of wbpM in the region just upstream of wbpM in both stO1 and 2192. A strong potential promoter (P1) just upstream of wbpM with identical sequences in the two strains was identified and was investigated further (Fig. 4C). Transcriptional fusions of the promoter region plus ~500 bp of the wbpM coding sequence with lacZ were generated with the promoter in the forward and reverse orientations relative to the lacZ coding sequence. The forward primer used to generate these fusions is noted in Fig. 4C. We then directly assayed promoter activity by monitoring β-galactosidase activity in strains with chromosomal P1-lacZ fusions at the attB site in both stO1 and 2192. We grew these strains, derived from stO1 and 2192, at 37°C and assayed them for lacZ activity at both the exponential and stationary phases. β-Galactosidase activity was significantly higher in the forward construct than in the reverse construct in both exponential- and stationary-phase cultures of both strains containing these transcriptional fusions, suggesting that there is no growth-phase-dependent regulation of this promoter. Further, the difference in β-galactosidase activity between 2192 and stO1 was not significant (Fig. 4D and andE),E), indicating that the activities of this promoter are similar in 2192 and stO1. This suggests that the larger amounts of wbpM RNA in 2192 are not due to differences in P1 activity but may be due to other potential promoters of wbpM. In fact, we have identified other potential promoters upstream of promoter P1 that may drive the expression of wbpM, either as monocistronic or polycistronic RNA. Experiments to address whether or not these promoters are active and are differentially regulated in 2192 versus stO1, which would help to explain the differences in wbpM transcript levels in these strains, are under way.
WbpM is an inner-membrane-bound NAD+-dependent UDP-4,6-GlcNAc dehydratase (29, 30) that is found in all 20 P. aeruginosa O-antigen loci. It has been shown to be required for O-antigen expression in multiple serotypes, particularly those that contain d-QuiNAc, d-FucNAc, or derivatives of these sugars. Serotypes O3, O5, O6, O10, and O11 express O antigens that contain these sugars, and mutants with insertional mutations in wbpM of each of these serotypes lost detectable specific O-antigen LPS expression (31). In contrast, O15 and O17 strains do not contain these sugars (5), and mutants of these strains with wbpM insertional mutations still had detectable O antigen (31, 32).
In order to verify that the defect in the O antigen of 2192 was due to the wbpM2192 sequence and not due to another mutation in this strain, we generated a mutant with a wbpM deletion in stO1. As anticipated based on the composition of the stO1 O antigen, the ΔwbpM mutant does not express O antigen at either 25°C or 37°C (data not shown). We also directly stained the LPSs of stO1, 2192, and the ΔwbpM mutant grown at both temperatures and noted the presence of lipid A plus core in all of the samples (data not shown). We then cloned both wbpMO1 and wbpM2192 under the control of an arabinose-inducible/glucose-repressible PBAD promoter into plasmid pHERD20T (19) and provided these plasmids to the ΔwbpM mutant. When the complemented strains were grown at either 25°C or 37°C in the presence of 2% arabinose, wild-type O-antigen expression was restored in the ΔwbpM(pHERD20T-wbpMO1) strain, as detected by Western blotting with serotype O1 antisera (Fig. 5). In contrast, the ΔwbpM(pHERD20T-wbpM2192) strain, when grown at 25°C, exhibited O antigen similar in signal intensity to the LPS of 2192, but no O antigen was seen at 37°C. While the intensity of the signal of the LPS from the ΔwbpM(pHERD20T-wbpM2192) strain grown at 25°C was similar to that of 2192 grown at 25°C, the former did not exhibit the same chain length defect as that seen in 2192. Specifically, in the ΔwbpM(pHERD20T-wbpM2192) strain grown at 25°C, there is some restoration of HMW LPS (Fig. 5, arrow), which is absent in 2192.
We wanted to investigate the relationship between wbpM expression levels and O-antigen chain length in these strains. Specifically, we hypothesized that expression of wbpMO1 may lead to dose-dependent expression of O antigen missing in 2192 and the ΔwbpM mutant. To address this, wbpMO1 was cloned with a C-terminal Flag tag into pHERD20T. We provided this plasmid to both the ΔwbpM mutant and 2192 and induced expression of wbpMO1-Flag from pHERD20T by adding arabinose at concentrations ranging from 0% to 2%. LPS preparations and whole-cell lysates of these arabinose-induced cultures were prepared and were used in Western blot analyses to detect the O antigen and the Flag tag, respectively (Fig. 6). Wild-type O-antigen expression, including expression of HMW LPS, was restored at levels of arabinose as low as 0.2% in the ΔwbpM(pHERD20T-wbpMO1-Flag) strain (Fig. 6A, arrow). In contrast, levels of arabinose as high as 2% were not able to restore expression of HMW LPS in 2192(pHERD20T-wbpMO1-Flag) (Fig. 6A). In addition, while the Flag tag was detected at every level of induction above 0% in the ΔwbpM(pHERD20T-wbpMO1-Flag) strain, the Flag tag was detected only at levels of arabinose starting at 1% in strain 2192(pHERD20T-wbpMO1-Flag) (Fig. 6B), suggesting that either 2192 makes less WbpM-Flag protein than the ΔwbpM mutant or this protein is degraded at a higher rate in 2192 than in the ΔwbpM mutant. Further, we observed a positive correlation between wbpMO1 induction levels, Flag tag intensity, and the amount of O antigen. In addition, while there was a correlation between wbpMO1-Flag expression and the amount of O antigen expressed in 2192(pHERD20T-wbpMO1-Flag), no amount of arabinose could restore expression of HMW LPS to this strain. This suggested that mucoidy, which is a major phenotypic difference between 2192 and the ΔwbpM mutant, may be at least partly responsible for the lack of HMW LPS expression in 2192(pHERD20T-wbpMO1-Flag). In fact, a spontaneous nonmucoid derivative of strain 2192 provided with pHERD20T-wbpMO1-Flag was able to produce the longer chains missing from mucoid wbpMO1-complemented 2192 (data not shown).
These results are similar to those recently reported by Ma et al., who suggest a posttranslational mechanism of regulation of the synthesis of three polysaccharides (alginate, LPS, and Psl) by P. aeruginosa, proceeding in such a way that overexpression of one polysaccharide decreases the expression of the other polysaccharides (16). They showed that a constitutively mucoid mucA22 mutant of laboratory strain PAO1, termed PDO300, lacked expression of HMW LPS compared to its parent strain, and they propose that this is due to the fact that a limited pool of the precursor sugar, mannose-1-phosphate, is generated by the AlgC protein and then feeds into multiple polysaccharide synthesis pathways, including those responsible for the production of LPS and alginate (16).
We suggest that the decrease in the amount of HMW LPS produced in mucoid strains may provide the additional benefit of evasion of the host immune defenses, which the alginate provides, since LPS is a highly immunodominant molecule that is known to activate the host immune system. It is interesting to speculate that the overexpression of alginate may provide a physical barrier to mask LPS from components of the host immune system and that downregulation of HMW LPS with the longest O-antigen chains in mucoid strains allows for further concealment. Overexpression of alginate resulting in a decrease in the amount of HMW LPS may provide an advantage in the absence of, or prior to, an irreversible mutational event that results in complete loss of O antigen.
The mutation in 2192 wbpM is a 3-bp deletion that results in a deleted valine at amino acid position 636 in WbpM. This residue has not been recognized as important for function in vitro, nor is it in any recognized domains in silico (30). Further, 3-dimensional modeling of the predicted structure of WbpM against the crystal structure of a homolog, Helicobacter pylori FlaA1, places the valine residue in a region of the protein that is not in close proximity to either the catalytic or the NAD+ binding domains of the protein (data not shown). However, a ClustalW alignment of WbpM and some of its closely related homologs from other species shows that the valine is either conserved or replaced with a similar amino acid (leucine or isoleucine) (data not shown). This suggests that this residue may be important for the structure and/or function of WbpM. We also sequenced an ~500-bp region of wbpM from serotypes O2 to O20 that contains the coding region for position 636. We found that the predicted amino acid sequences of this region are nearly identical to those for stO1, and all are predicted to contain V636 (data not shown).
Given that V636 is not located in any recognized domain of WbpM but was required for complementation of the LPS defect, and given that our data suggested that there is no transcriptional defect of wbpM2192 in 2192, we suspected that this residue might contribute to the stability of the protein. Specifically, we hypothesized that WbpM2192 may be less stable than its wild-type counterpart as a result of the V636 deletion. To address this, we provided 2192 and the ΔwbpM mutant with pHERD20T-wbpMO1-Flag or pHERD20T-wbpM2192-Flag, and we induced expression of these constructs in both 2192 and the ΔwbpM mutant with 2% arabinose for 2 h at 37°C. The cells were then washed; 1% glucose was added to inhibit transcription from the PBAD promoter on pHERD20T; and samples were taken at hourly intervals for Western blot analysis. Following glucose repression, the mutant WbpM2192-Flag protein was degraded at a much higher rate than wild-type WbpMO1-Flag protein in both strain 2192 (Fig. 7A and andB)B) and the ΔwbpM mutant (Fig. 7C and andD),D), lending support to the hypothesis that V636 contributes to the stability of WbpM. In addition, both the WbpMO1-Flag protein and the WbpM2192-Flag protein were made at higher levels and were more stable in the ΔwbpM mutant than in 2192 (Fig. 7A to toDD).
These data suggest that the extreme C-terminal region of the protein may be required for the function of WbpM. We also generated a Flag-tagged allele of wbpM that encodes all of the protein up to V636, termed wbpM635-Flag. While the Flag tag was detected in both 2192(pHERD20T-wbpM635-Flag) and the ΔwbpM(pHERD20T-wbpM635-Flag) strain when grown in 2% arabinose, no detectable LPS was seen in either recombinant strain, suggesting that the entire C-terminal region of the protein from at least position 635 onward is important for stability and/or function. It will be interesting to define more precisely the boundaries and properties of this domain. Studies to address this are under way.
Valine 636 of WbpM plays a role in the stability of the protein, so it is likely required for proper protein folding. The apparent instability of both WbpMO1 and WbpM2192 in strain 2192 may be due to a secondary mutation elsewhere in the genome of strain 2192 that is not present in the ΔwbpM mutant, and/or to upregulation of a protein degradation pathway in 2192 related to its chronic infection phenotype. It is interesting to speculate that the overproduction of alginate in strain 2192 alters the membrane in a manner that leads to an overall rapid turnover of membrane proteins or that strain 2192 specifically targets both WbpMO1-Flag and WbpM2192-Flag for degradation. Evidence that the mucoid phenotype may affect membrane-bound protein stability has been noted in the literature recently. A recent study has shown that nonmucoid strain PAO1 (serotype O5) can be induced to produce alginate when grown on Pseudomonas isolation agar supplemented with ammonium vanadate (PIA-AMV). It is thought that this reversible effect is due to membrane stresses that lead to the degradation of the membrane-bound anti-sigma factor MucA (33).
Interestingly, when we cloned truncated versions of WbpMO1-Flag and WbpM2192-Flag lacking the N-terminal membrane-spanning domains, similar to those generated by Creuzenet and Lam (29) (rendering them cytosolic), and expressed them in 2192 and the ΔwbpM mutant, they were much more stable in both strains than their full-length counterparts and seemed to be equally stable in both 2192 and stO1 (data not shown), suggesting that the observed degradation of WbpM in 2192 is dependent on localization to the membrane. This finding may offer insight into mechanisms of protein turnover, specifically the turnover of membrane proteins, not previously recognized in P. aeruginosa chronic CF isolates.
In this report, we have shown that the defect in O-antigen LPS expression, a typical phenotype for P. aeruginosa chronic infection isolates from CF patients' lungs, is due to a single mutational event in the O1 O-antigen locus of strain 2192. Further, we have also shown that this defect does not lead to complete loss of O-antigen expression but rather is a temperature-dependent defect: 2192 expresses LPS with intermediate O-antigen chain lengths at 25°C but does not express any detectable O antigen at 37°C. While the O-antigen locus of this strain contains several differences from the IATS serotype O1 strain stO1, only one difference is responsible for the LPS defect. This is the first time that this temperature-dependent O-antigen phenotype has been observed in a chronic CF isolate, and while the mutation responsible for this phenotype does not lead to complete loss of LPS expression, as in other examples in the literature (28, 34), it likely leads to a de facto loss of LPS expression in the lungs of CF patients, an environment that is close to 37°C. This supports the current understanding of how the infection progresses from an acute to a chronic state in the lungs of CF patients: mutations that confer a selective advantage allow lineages with certain phenotypes, including defects in LPS expression, to persist, and these strains make up the P. aeruginosa population found in the lungs of CF patients.
We thank Alicia Midland, Marc Schulman, Erica Kintz, F. Heath Damron, and John Varga for technical assistance.
M.R.D. was supported in part by the National Institutes of Health through University of Virginia Infectious Diseases Training Grant AI07406. I.V.L. was supported in part by a MARC grant to Florida International University (5 T34GM083688-05) and the University of Virginia School of Medicine. Structural analysis was supported by a Department of Energy grant to the CCRC (DE-FG02-93ER-20097).
Published ahead of print 25 January 2013
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.01999-12.