A Non-Classical LysR-Type Transcriptional Regulator PA2206 Is Required for an Effective Oxidative Stress Response in Pseudomonas aeruginosa

LysR-type transcriptional regulators (LTTRs) are emerging as key circuit components in regulating microbial stress responses and are implicated in modulating oxidative stress in the human opportunistic pathogen Pseudomonas aeruginosa. The oxidative stress response encapsulates several strategies to overcome the deleterious effects of reactive oxygen species. However, many of the regulatory components and associated molecular mechanisms underpinning this key adaptive response remain to be characterised. Comparative analysis of publically available transcriptomic datasets led to the identification of a novel LTTR, PA2206, whose expression was altered in response to a range of host signals in addition to oxidative stress. PA2206 was found to be required for tolerance to H2O2 in vitro and lethality in vivo in the Zebrafish embryo model of infection. Transcriptomic analysis in the presence of H2O2 showed that PA2206 altered the expression of 58 genes, including a large repertoire of oxidative stress and iron responsive genes, independent of the master regulator of oxidative stress, OxyR. Contrary to the classic mechanism of LysR regulation, PA2206 did not autoregulate its own expression and did not influence expression of adjacent or divergently transcribed genes. The PA2214-15 operon was identified as a direct target of PA2206 with truncated promoter fragments revealing binding to the 5′-ATTGCCTGGGGTTAT-3′ LysR box adjacent to the predicted −35 region. PA2206 also interacted with the pvdS promoter suggesting a global dimension to the PA2206 regulon, and suggests PA2206 is an important regulatory component of P. aeruginosa adaptation during oxidative stress.


Introduction
Bacteria have evolved diverse integrated regulatory networks which are essential for their survival and persistence in a vast array of ecological niches. An efficient response to changing environmental conditions requires the coordination of signal perception and transduction, transcriptional and post-transcriptional regulation, as well as post-translational events. By far the largest known family of transcriptional regulators in prokaryotes, LysR-type transcriptional regulators (LTTRs), are emerging as key coordinators of microbial gene expression. LTTRs are ubiquitous among bacteria with orthologues also found in archaea and eukaryotic organisms [1]. These signal responsive proteins represent the primary mechanism of regulation of catabolic systems in bacteria while they have also been shown to influence key adaptive and virulence phenotypes such as biofilm formation, motility, signalling, secondary metabolite production, and oxidative stress caused by reactive oxygen species (ROS) [1,2]. Organisms are exposed to ROS, such as hydrogen peroxide (H 2 O 2 ) and the superoxide anion, during the course of normal aerobic metabolism or following exposure to radical-generating compounds, including redox-cycling drugs. ROS cause wideranging damage to macromolecules, which can eventually lead to cell death. To protect themselves against this damage, cells have evolved effective defence mechanisms, including anti-oxidant enzymes and free radical scavengers. OxyR, an LTTR that is universally encoded among Gram negative bacteria, is considered the master regulator of the oxidative stress response. In Escherichia coli, OxyR works in concert with the small RNA oxyS to coordinate expression of catalase and peroxidase genes [3]. More recently, several other LTTRs have been shown to play a role in the oxidative stress response, including Hrg in Salmonella enterica [4], HypR in Enterococcus faecalis [5], CztR in Caulobacter crescentus [6] and HcaR in E. coli [7]. However, with the exception of OxyR, the role of LTTRs during oxidative stress remains largely uncharacterised in many bacteria. This is particularly true for the important nosocomial pathogen Pseudomonas aeruginosa in which LTTRs account for more than 2% of the total gene content [8].
In addition to being a leading cause of hospital-acquired infections in immunocompromised individuals, P. aeruginosa is also the primary cause of chronic lung infections in patients suffering from cystic fibrosis [9,10,11,12,13]. OxyR has been shown to modulate the expression of several key oxidative stress response genes including katA, katB, ahpB and ahpCF in P. aeruginosa [14,15]. Recently, the OxyR regulon in P. aeruginosa was defined through Co-IP analysis and was shown to include genes involved in iron homeostasis, quorum sensing, protein synthesis, and oxidative phosphorylation [15]. Wei and colleagues proposed a multilayered response to oxidative stress which includes increased expression of antioxidants to detoxify ROS, inhibition of primary metabolism and oxidative respiration to reduce the production of ROS, and modulation of iron uptake systems to minimise the Fenton reaction [15]. This requires a complex cellular response, the control of which is likely to encompass multiple regulatory components. However, apart from OxyR, the role of LTTR proteins in the P. aeruginosa oxidative stress response has not been investigated.
In this study, we identified PA2206 as a novel LTTR transcriptionally influenced by both oxidative stress and host signals, independent of the master regulator of oxidative stress, OxyR. PA2206 was found to be required for an effective oxidative stress response as well as lethality in a zebrafish model of infection. This novel non-classical LTTR was found to have a significant influence on the expression of oxidative stress responsive genes, while the PA2214-2215 and pvdS promoters were identified as direct targets. PA2206 was shown to interact specifically with the 59-ATTGCCTGGGGTTAT-39 LysR box adjacent to the predicted 235 region of the PA2214-15 promoter. The requirement of PA2206 for tolerance to oxidative stress and zebrafish embryo lethality, independent of OxyR, implies a novel route for the management of oxidative stress in P. aeruginosa, and a role for PA2206 during the host-microbial interaction.

Bacterial strains, media and growth conditions
Strains and plasmids used in this study are listed in Table 1. All cultures of P. aeruginosa strain mPAO1 and associated mutants were routinely grown in Luria Bertani [16] or M9 minimal media [supplemented with 0.3% (w/v) sodium citrate, 1 mM MgSO 4 , and 0.05 mM FeCl 3 ] at 37uC [17]. Bacterial cultures were grown in Casamino Acid (CAA) medium supplemented with 100 mM FeCl 3 , for the proteomic analysis [18]. Tetracycline-resistant transposon insertion mutants (Table 1) were obtained from the University of Washington Genome Centre mutant library [19]. Transposon insertion, location, and orientation were confirmed for all mutants. E. coli strains were routinely grown in LB media at 37uC. Where appropriate, antibiotics were added to growth media at the following concentrations: tetracycline, 60 mg ml 21 , and gentamicin, 20 mg ml 21 , for P. aeruginosa; and ampicillin, 50 mg ml 21 , and gentamicin, 20 mg ml 21 , for E. coli.

Assay of survival under conditions of oxidative stress
Growth assays were performed with 20 mM H 2 O 2 or 40 mM menadione added to exponentially growing cells. Briefly, overnight cultures were transferred into fresh media at a starting OD600 nm of 0.1, and were incubated at 37uC with shaking for 3 hrs. Subsequently, at approximately OD600 nm 0.6, H 2 O 2 or menadione was added and growth monitored hourly. Survival assays were performed in the presence of 100 mM H 2 O 2 as this concentration has previously been chosen for P. aeruginosa and other species [20,21]. Overnight cultures were washed in Ringer's solution (J-strength) and resuspended to an optical density of 0.8 at 600 nm. Bacterial suspensions (5 ml) were exposed to 100 mM H 2 O 2 at 37uC with agitation (150 rpm). Viable counts were determined on LB agar with antibiotic selection where appropriate before and after 1 hour of exposure, and expressed as percentage surviving cells as described by Heeb and colleagues [22]. Zonal inhibition assays were performed on M9 plates supplemented with glucose (2% w/v). An overnight PAO1 culture was spread on the agar plate using a sterile swab and air dried in a laminar flow hood for 30 mins. Subsequently, filter paper disks (approx 7 mm) were placed in the centre of the plate and saturated with 10 ml of a 30% (w/w) solution of H 2 O 2 (Sigma) corresponding to an 8.8 M concentration. Plates were incubated at 37uC overnight and the zone of inhibition measured at 16 hrs.

Zebrafish infection assay
Zebrafish embryos were derived from adults of the Zebrafish AB line (Danio rerio) which were kept at 28uC in a 14-h/10-h light/ dark cycle in our breeding facility. Zebrafish embryos were staged at 26 hours post fertilization according to previously described developmental criteria [23]. Embryos were manually dechorionated and subsequently anaesthetized in a 0.015% aqueous solution of ethyl-3-aminobenzoate methanesulfonate salt. P. aeruginosa cells from an overnight culture in LB broth were washed twice in phosphate buffer saline and diluted in phenol red phosphate buffer saline in order to visually monitor injection under a stereo-microscope. Bacterial cells were then microinjected (1-2 nl) into the blood island of the Zebrafish embryos. The inoculum size was determined by injecting in triplicate in phosphate buffer saline followed by plating on LB agar plates. Approximately 300 colony forming units were injected into 10 embryos per group. Injected Zebrafish embryos were returned to egg water, incubated at 30uC, and monitored for mortality on a daily basis. Data represents three independent biological replicates. Kaplan-Meier curves were used to compare survival rates of zebrafish infected with wild-type and PA2206 2 . Differences between infected embryos, compared using a log-rank test in MedCalcH, were considered significant #0.05. All animal experiment protocols were approved by the Animal Experimentation Ethics Committee, University College Cork.

Bioinformatics and comparative genomic analysis
Genome sequences were obtained from the www.pseudomonas. com resource [24] and compared using the ARTEMIS Comparison Tool on the www.webact.org site [25]. Promoter predictions were performed using BProm software on the http://linux1. softberry.com site.
To screen the PAO1 genome for the LysR consensus sequences identified proximal to the PA2214 predicted 235 upstream region, nucleotide sequences from positions 2300 to 21 relative to the origin of translation of all coding sequences (CDS) were parsed from the Genbank annotated genome sequence file (version AE004091.2) of P. aeruginosa PAO1 and stored locally in FASTA format using a script in Perl. CDSs matching TTN 9 TA, where N is any nucleotide, in the 300 bp upstream of the origin of translations were identified using Bioperl [16]. Subsequently, sequences were analysed using MEME motif search software to construct a PA2206 LysR consensus motif. This analysis was restricted to sequences which had a statistical significance (p-value) lower than 1e 204 .

Complementation in trans of mutant strains
Oligonucleotide primers PA2206CCF and PA2206CCR (Table  S1) were designed to amplify the PA2206 gene and promoter region based on the PAO1 genome sequence (GenBank accession number NC002516). The DNA fragment was amplified from PAO1 genomic DNA using Platinum High Fidelity Taq polymerase (Invitrogen, UK) and subsequently cloned as a HindIII and XbaI fragment into HindIII-XbaI sites of pBBR1MCS-5 generating pBR1-PA2206 C [26]. Similarly, primer Hin2206F was designed at the PA2206 translational start site to create construct (pBR1-2) with primer Xb2206R (Table S1). Chemical transformation into competent E. coli DH5a cells and purification was followed by electroporation into the P. aeruginosa mPAO1 PA2206 Tn5 mutant strain as described by Sambrook and colleagues [17]. Construct fidelity was confirmed by sequencing in both directions.

Promoter-fusion analysis
Promoter regions were amplified from PAO1 genomic DNA by PCR using Platinum High Fidelity Taq polymerase (Invitrogen). The PA2214 promoter region was amplified using PA2214TRF and PA2214TRR primers, while the PA2216 promoter was amplified using PA2216TRF and PA2216TRR (Table S1). The PA2206 promoter was amplified using PA2206TRF and PA2206TRR primers (Table S1). Inserts were digested using XbaI-KpnI and subsequently ligated directly into similarly digested pMP190 [27] plasmid at 16uC overnight followed by transformation into chemically competent DH5a E. coli cells. Promoter orientation was confirmed by PCR using a promoter specific primer and a lacZ primer and the fidelity of all transcriptional fusions was verified by sequencing (GATC). Triparental mating utilising an E. coli strain harbouring the pRK2013 helper plasmid was performed to transfer the fusions into recipient P. aeruginosa strains. The PA2214 promoter fusion was also introduced into E. coli containing pBR1-2 and pBBR1MCS5 vector control. Fusion analysis was performed in M9 media supplemented with glucose (10 mM) with cultures grown at 37uC with shaking.

Microarray analysis
Three independent overnight cultures of P. aeruginosa PA2206 2 mutant and PA2206 C grown in M9 minimal media [supplemented with 1 mM MgsO 4 and 2% glucose] with shaking at 37uC, were diluted to an optical density of 0.05 at 600 nm in fresh media. Cells were then grown to an optical density of 0.5 at 600 nm at which stage they were exposed to 1 mM H 2 O 2 for a ten minute time period. Total RNA was subsequently extracted using the QIAGEN TM RNeasy H Mini RNA extraction kit in accordance with the manufacturer's guidelines. RNA was treated with Ambion H TURBO TM DNase at 37uC for 1 hour and the RNA pellet was then resuspended in sterile diethylpyrocarbonate (DEPC)-treated water (1 ml of 0.1% DEPC) to a final volume of 25 ml. Isolated RNA was sent to DNAVision (Belgium) for further analysis. RNA quality was checked using a Bioanalyzer Agilent 2100 at DNAVision, and cDNA synthesis, fragmentation and terminal labelling preceded hybridisation on Affymetrix GeneChip P. aeruginosa Genome arrays according to the Affymetrix guidelines. The raw data was normalised using the robust multiarray average (RMA) algorithm [28] and pairwise comparison (t-test, P,0.05) was carried out to generate a list of genes with significantly altered expression between the test strains of greater than 1.5 fold. The p-value was then assessed for each gene by controlling the false discovery rate using the Benjamini Hochberg method [29]. All files have been submitted to the GEO database (Ref No: GSE23367) compliant with MIAME.

Real-time quantitative RT-PCR
cDNA was synthesised using AMV Reverse Transcriptase (Promega), RNasin (100 U/ml) (Promega), random primers (0.5 mg/ml) (Promega) and 10 mM dNTPs (Promega). Specific RT-PCR primers were designed for each gene based on the PAO1 genome sequence (Table S1). Standards were generated by PCR and quantitative real-time PCR analysis of expression was carried out using the Quantifast SYBR Green PCR kit (QIAGEN). Quantitative real time-PCR signals were normalised to the constitutively expressed housekeeping gene proC [30], whose expression was shown to be unaltered under the conditions used to generate the transcriptomic profile.
Microarray data were validated by quantitative RT-PCR analysis of RNA isolated from three independent experiments as above. For further analysis of gene expression under different oxidative stress conditions, overnight cultures of the wild-type and mutant strains were inoculated into M9 minimal media supplemented with glucose (2% w/v) at an initial OD 600 nm of 0.05. The cells were grown to an OD 600 nm of 0.4 and the cultures were subsequently divided and exposed to 1 mM and/or 5 mM concentrations of H 2 O 2 for 5 and 10 minute time periods. Untreated cells were harvested as a control against which to compare gene expression. All assays were carried out in triplicate, and the mean and standard error were calculated.

PA2206-HisTag protein purification and quantification
The PA2206 gene from P. aeruginosa PAO1 was amplified using gene-specific primers 2206HisF and 2206HisNR (Table S1) and cloned into the pET28a inducible expression plasmid (Invitrogen) using the EcoRI-HindIII restriction enzymes (New England Biolabs). Constructs were transformed into E. coli BL21 Codon Plus (DE3) cells giving rise to PA2206 fused to an N-terminal His 6 tag. Overnight LB cultures were transferred into fresh media supplemented with appropriate antibiotics and were grown at 30uC to an optical density of 0.8 at 600 nm at which time IPTG was added to a final concentration of 0.1 mM. The cultures were grown for a further 4 hrs and cells were harvested at 4uC following centrifugation at 10,000 rpm for 10 mins. Supernatants were removed and cells were stored overnight a 280uC to enhance lysis. Cell pellets were resuspended in Cell Lytic TM B, lysosyme, benzonase, and protease inhibitors (all Sigma-Aldrich), and subsequently incubated at room temperature for 15 mins with shaking (100 rpm). Following centrifugation at 10,000 rpm at 4uC, the clarified supernatant was purified on a Poly-Prep chromatography column (Bio-Rad) containing His Select nickel affinity gel (Sigma-Aldrich) and eluted using increasing concentrations of imidazole (10-500 mM). Fractions were analysed on a 10% (w/v) SDS PAGE (10%) gels (Fig. S1), and the concentration of purified PA2206-His 6 protein was determined by Bradford assay as described before [31].

Comparative transcriptomic analysis reveals a nonclassical host-responsive LysR-Type Transcriptional Regulator (LTTR)
The oxidative stress response is central to the survival of many microbial pathogens during infection. However, the molecular mechanisms governing this key adaptive phenotype remain largely uncharacterised. Interrogation of the publically available transcriptome datasets revealed a striking correlation between oxidative stress and the induction of genes encoding LTTRs in P. aeruginosa. Expression of 41 out of 121 genes annotated as LTTRs in the PAO1 genome was $2-fold altered (39 upregulated) at the transcriptional level in P. aeruginosa in response to oxidative stress induced by H 2 O 2 [32,33,34]. However, surprisingly only a limited number of these LTTRs that respond to oxidative stress were also altered in transcriptomic datasets designed to investigate the host interaction [35,36,37]. We considered this subset of LTTRs likely to play an important role in P. aeruginosa hostmicrobial interactions and selected PA2206 on the basis of it being regulated in response to both human (+2.82 fold) and plant (+2.7 fold) signals, in addition to oxidative stress [33,35,37]. Annotated as a PcaQ-type LTTR and a unique transcriptional unit, PA2206 was not found to be divergently transcribed with an adjacent gene, in contrast to the classic LysR topology [1]. Furthermore, quantitative RT-PCR and promoter fusion analysis revealed that PA2206 did not influence the expression of adjacently transcribed PA2205 or PA2207-09 (data not shown). Therefore, PA2206 would appear to be one of an emerging class of non-classical LTTRs, and its induction in the presence of a range of host signals suggests an important role during the host-interaction.
Expression of PA2206 is induced in response to hydrogen peroxide and menadione, independent of OxyR To confirm the induction of PA2206 in response to oxidative stress, the kinetics of PA2206 expression in the presence of H 2 O 2 was investigated. Quantitative Real Time RT-PCR analysis demonstrated that expression of PA2206 was induced in response to sublethal 1, 5, and 10 mM concentrations of H 2 O 2 , following exposure for 10, 20, 30, and 60 minute time periods (Fig. 1A). Expression of PA2206 remained elevated (2-4 fold) in a dose dependent manner for up to an hour after exposure to H 2 O 2 with the greatest induction observed in response to 10 mM H 2 O 2 (Fig. 1A). In order to assess whether this induction was specific for H 2 O 2 , expression was investigated in the presence of other oxidative stress inducing compounds. Exposure to the superoxide generator menadione (2-methyl-1,4-naphthoquinone) resulted in 1.7-fold (60.15) induction of PA2206 gene expression (Fig. S2), while the organic hydroperoxides cumene hydroperoxide and tertbutyl hydroperoxide did not influence expression of PA2206 (data not shown) suggesting a degree of specificity to the activation of PA2206 in response to oxidative stress.
To investigate the induction of PA2206 at the transcriptional level, two possibilities were considered; PA2206 may be (1) autoregulatory in the presence of a signal generated during oxidative stress, or (2) under the control of another transcriptional regulator. Although several putative LysR boxes were identified upstream of the PA2206 gene, promoter fusion analysis revealed that mutation or overexpression of PA2206 had no significant influence on its own promoter activity, with comparable Miller Units detected in wild-type, PA2206 2 and complemented PA2206 C strains (Fig. S3). This suggested that PA2206 was not autoregulatory, again in contrast to the classical mechanism of LysR regulation. To investigate whether induction of PA2206 was the result of regulation by OxyR (the master oxidative stress regulator), expression of PA2206 was investigated using qRT-PCR in wild-type and in the oxyR mutant grown in M9 media supplemented with 2% (w/v) glucose. PA2206 was induced to comparable levels in the wild-type and oxyR mutant following exposure to 5 mM or 10 mM H 2 O 2 (Haynes & O'Gara, unpublished data) suggesting that PA2206 expression is induced independently of OxyR during the oxidative stress response. PA2206 is required for an efficient oxidative stress response and pathogenesis in P. aeruginosa In order to determine whether the induction of PA2206 in response to H 2 O 2 and menadione correlated with a stress-tolerant phenotype, growth and viability of the wild-type and PA2206 2 mutant strains in the presence of exogenous ROS was investigated. Although growth in the untreated samples was comparable for mutant and wild-type strains, growth was significantly reduced in the PA2206 2 mutant compared to wild-type upon addition of H 2 O 2 (20 mM) or menadione (40 mM) to exponentially growing cells (Fig. S4). In addition, time-kill survival assays provided further evidence that PA2206 2 mutant was hypersensitive to H 2 O 2 ( Table 2). Viable counts for mutant and wild-type strains determined before (time 0) and one hour after (time 1) exposure to H 2 O 2 (a 1 hr sampling time was chosen to assess the steady state response rather than the early and acute response) revealed a significant reduction in survival in the PA2206 2 mutant compared to the wild-type strain ( Table 2). In contrast, wild-type cells were able to survive exposure to this physiologically high concentration (100 mM). Introduction of PA2206 in trans (PA2206 C ) restored H 2 O 2 tolerance of the mutant to 80% of the wild-type strain. Taken together, these data demonstrated that the PA2206 2 mutant was significantly more susceptible than the wild-type to exogenous ROS.
To assess the in vivo relevance of the oxidative stress sensitive phenotype of the PA2206 2 mutant, lethality in the zebrafish embryo model of infection was investigated. Several reports suggest that some transcriptional regulators that influence the oxidative stress response can have a significant impact on pathogenesis or lethality in animal models while others exhibit no influence [38,39,40,41]. Consistent with the stress-susceptible phenotype of PA2206 2 , the mutant exhibited significantly less killing than the wild-type in the zebrafish embryo model. Killing was reduced to 72% at 2 days post infection (d.p.i.) and 60% (3 d.p.i.) when compared to wild-type (Fig. 1B). Indeed, embryo survival remained constant at 90% in the presence of the mutant strain throughout these experiments. Interestingly, cytotoxicity, invasion and adhesion in epithelial cells was comparable between wild-type and PA2206 2 (data not shown). Therefore, to better understand the molecular mechanism underpinning the oxidative stress and zebrafish lethality phenotypes of the PA2206 2 mutant, we undertook global transcriptomic profiling.

Transcriptomic analysis reveals a global profile for PA2206
In order to uncover the extent of the PA2206 transcriptional regulon during oxidative stress, gene expression profiles were compared between isogenic PA2206 C and PA2206 2 in response to a sub-lethal 1 mM concentration of H 2 O 2 . This sub lethal concentration was chosen as it most accurately reflects the physiological levels found in the CF lung and other sites where the PMN inflammatory response occurs (33). Approximately 1% of genes (58 of the 5,500 P. aeruginosa PAO1 probe sets on the microarray) demonstrated significantly (.2-fold) altered expression (Table 3). Of these, 34 genes were upregulated and 24 genes were downregulated in PA2206 C relative to PA2206 2 , and the changes in gene expression ranged from +7.7 to 23.4 fold ( Table 3). The dataset was validated by qRT-PCR, including the wild-type strain for comparison. Expression of the grx glutaredoxin (PA5129) and a glycine tRNA (PA0729.1) was decreased in the PA2206 mutant, while a glutathione S-transferase (PA2821) and the snr cytochrome (PA3032) were found to be increased (Fig. 2). These results were in accordance with the microarray analysis.
Consistent with promoter fusion and expression analysis (data not shown), the genes adjacent to PA2206 were unaltered in the transcriptome profile. However, expression of the distantly encoded PA2214-15 operon was increased 2-3 fold in the PA2206 C strain as was expression of PA2216 (Table 3). Comparative genomic analysis of publically available genome sequences revealed the PA2206-2217 region to be exclusive to P. aeruginosa, while truncated segments were encoded in the P. fluorescens Pf-5 and P. fulva 126 genomes (Fig. S5). Intriguingly, homologues of the PA2214-17 genes were identified adjacent to the PA2206 homologue in P. fluorescens Pf-5 (Fig. 3A). Furthermore, hydrogen peroxide sensitivity assays revealed comparable zones of inhibition in PA2206 and PA2215 mutants (a PA2214 mutant was not available from the mutant collection), both of which were significantly more sensitive than wild-type (Table S2). Therefore, to investigate whether these loci were indeed direct targets of the PA2206 LTTR, promoter fusion and DNA-binding analyses were undertaken.
The PA2214-lacZ transcriptional reporter fusion was subsequently introduced into E. coli harbouring a pBBR1MCS plasmid constitutively expressing PA2206 (EcpBR-2) with the aim of elucidating whether the PA2214-15 promoter was a direct target of PA2206. As E. coli does not encode a PA2206 homologue, activation of the PA2214-15 promoter in this strain would most likely be the result of a direct interaction between the LysR protein  and the promoter DNA. PA2214-15 expression was significantly induced upon overexpression of PA2206, while promoter activity was not detected in the vector control suggesting that this promoter is under the direct transcriptional control of PA2206 (Fig. 3C). To confirm this direct interaction, DNA-mobility shift assays were performed.
PA2206 binds adjacent to the putative 235 region of the PA2214-15 promoter In order to provide direct evidence of the PA2206 binding interaction with the PA2214 promoter, a HisTag PA2206 construct was generated and introduced into E. coli BL21 DE cells for protein expression. After IPTG induction and purification through a nickel affinity column, protein purity and concentration were determined by SDS PAGE analysis and Bradford assay, respectively. Subsequent EMSA analysis revealed an interaction at the low nanomolar range of PA2206 protein with the 2160 to +31 (relative to start of translation) region of the PA2214-15 promoter, amplified using the PA2214F and PA2214R primers ( Fig. 4A and Table S1). Three complexes between the PA2206 protein and the PA2214 promoter were observed, formed in a concentration dependent manner (Fig. 4). A similar stoichiometry has previously been reported for other LTTRs, notably CysB from Salmonella enterica [42]. No such interaction was observed at the 2246 to 214 region of the PA2206 promoter (consistent with the lack of autoregulation observed at the transcriptional level) or the 2278 to 2106 region of the PA2216 promoter (Fig. 4A). Taken together with the promoter fusion analysis (Fig. 3), this would appear to confirm PA2214-15 as a direct target of PA2206 regulation.
Truncated PA2214 promoter fragments were generated with the aim of identifying the specific LysR motif with which the PA2206 protein interacts. The region of the PA2214 promoter to which PA2206 was shown to bind contains five putative LysR boxes upstream of the predicted (Bprom) transcriptional start site (Fig. 4B). Box 3 is situated adjacent to the predicted 235 region, while boxes 1 and 2 overlap each other directly upstream. Boxes 4 and 5 are situated between the predicted 235 and 210 regions, while a further three LysR boxes (boxes [6][7][8] were situated between the putative transcriptional and translational start sides. EMSA analysis subsequently revealed that PA2206 formed a complex with the promoter fragment containing boxes 3-8 but did not bind to the fragment containing boxes 1 & 2 (Fig. 4C). Neither did PA2206 bind to the promoter fragments located downstream of box 3 suggesting that this motif alone is sufficient for direct contact with PA2206 (Fig. 4C). Although F2-TrR (boxes 1&2) overlaps the 59 ATTGCC region of LysR box 3, and TrAF-TrR (boxes 4-8) is directly adjacent to the 39 region, no interaction with PA2206 was observed with these fragments, further evidence that the complete 59-ATTGCCTGGGGTTAT-39 LysR box is required for interaction with PA2206. It is worth noting that this sequence is highly similar to the ATAACC-N4-GGTTAT motif reported for the PcaQ LysR transcriptional regulator from Sinorhizobium meliloti [43,44], consistent with the PA2206 LysR domain annotation. The emerging global dimension to LysR regulation suggested that there may be additional targets of PA2206 regulation in P. aeruginosa. Therefore, BioPerl analysis was undertaken to identify the extent of the PA2206 regulon.

PA2206 has additional promoter targets suggesting a global regulatory role
To investigate whether PA2206 has a global regulatory role, a BioPerl script was generated to screen the PAO1 genome sequence for LysR consensus sequences with similarity to the 59-TTGCCTGGGGTTA-39 sequence of the PA2214-15 promoter. The absence of PA2206 homologues from species other than P. aeruginosa, with few exceptions, hindered the generation of a stringent consensus sequence. Furthermore, site-directed mutagenesis will be required to identify the nucleotides that are critical for PA2206 binding. However, LysR boxes with close sequence identity to the PA2206 LysR box were identified in 18 unique promoters (Fig. S6). Cross-referencing of this list with the transcriptomic data revealed several potential targets including the pvdS (PA2426, 5.26-fold upregulated) and PA4881 (7.66-fold upregulated) promoters. The pvdS promoter was chosen for further investigation on the basis of the relationship between iron metabolism and oxidative stress, and EMSA analysis revealed a specific interaction with the PA2206 protein in the nanomolar range (Fig. 5B). In contrast to the interaction of PA2206 with the PA2214 promoter (C1-C3 complexes, Fig. 4), the pvdS interaction occurred as a single complex migrating similar to the C2/C3 complex, with this differential stoichiometry possibly reflecting distinct affinities or oligomeric binding arrangements at each promoter [45]. This same pattern of interaction has previously been reported for the binding interaction of other LTTRs with target promoters, including that of the CatR LTTR to the catBC and clcABD promoters [12]. Binding analysis using a truncated pvdS promoter fragment downstream of the putative PA2206 box revealed no interaction suggesting that PA2206 binds to the identified consensus sequence in the pvdS promoter (Fig. 5B). In addition, PA2206 was also found to bind to the PA4881 promoter, consistent with its upregulation in the transcriptomic profile, forming C1 and C2/C3 complexes in a concentration dependent manner (Fig. S6). The specificity of these interactions was highlighted by the lack of any mobility shift with the PA0982 (Fig. 5B) and PA1874 (Fig. S6) promoters. Regulation of pvdS and PA4881 suggests a global role for PA2206 in P. aeruginosa, although further analysis incorporating Co-immunoprecipitation (Co-IP) analysis will be required to define the full repertoire of PA2206 targets in the cell. Surprisingly, neither pvdS nor PA4881 exhibited increased susceptibility to hydrogen peroxide using simple disk diffusion assays (data not shown). Therefore, the mechanism through which PA2206 contributes to the oxidative stress response in P. aeruginosa may involve a more complex interplay between downstream targets of PA2206.

Discussion
Microbial pathogens are continuously faced with the threat of oxidative stress elicited by the host immune response, and also from the by-products of aerobic metabolism. Their ability to sense and rapidly respond to oxidative stress arising from elevated levels of ROS is central to their pathogenesis and survival. For example, the presence of pathogens such as P. aeruginosa in the human lower Bacterial cultures were grown to exponential phase, and aliquots were then either untreated or exposed to a 1 mM concentration of H 2 O 2 for 10 mins. Expression of PA3032, PA2821, PA0729.1 and PA5129 was measured as fold change relative to the proC housekeeping gene. Mean values are represented 6 standard error. Consistent with the array data, expression of PA3032 and PA2821 was increased in the PA2206 2 mutant (white bar) compared to PA2206 C (striped bar) and the wild-type (black bar) strain. Similarly, expression of PA0729.1 and PA5129 was approximately 3-fold less in the PA2206 2 mutant strain (p-value of ,0.05 by student's ttest). doi:10.1371/journal.pone.0054479.g002 Single genes and operons are denoted by colour and pattern fill while homology is denoted by connecting shaded regions. P. fluorescens encodes a PA2206 homologue, which is adjacent to truncated genes corresponding to fragments of PA2207 and PA2212, both of which are downstream of a conserved PA2214-2216 homologous operon. TTT denotes a tripartite tricarboxylate transport system, FAH denotes a putative fumarylacetoacetate hydrolase activity, while Aldedh denotes a putative aldehyde dehydrogenase activity. (B) Promoter-fusion analysis of the PA2214 upstream region revealed significantly increased promoter activity in the PA2206 C strain relative to PA2206 2 . The increase in promoter activity was consistently observed in three independent experiments consisting of three biological replicates (** p-value,0.01 by student's ttest). (C) PA2214-lacZ promoter fusion analysis performed in E. coli harbouring a PA2206-pBBR1MCS5 construct compared to vector control revealed direct regulation. Promoter activity was below baseline in the vector control (compared to empty pBBR1MCS5 plasmid), while an average of 6 Miller Units was consistently detected in the overexpressing strain (** p-value of ,0.01 by student's ttest). doi:10.1371/journal.pone.0054479.g003 respiratory tract leads to the induction of host-derived mediators such as pro-inflammatory cytokines and chemokines which activate phagocytic cells including alveolar macrophages and neutrophils [46,47]. The phagocytes target bacterial populations with the production of oxidative stress inducing reactive oxygen species (ROS), proteolytic enzymes and a range of toxic metabolites [48]. As a result P. aeruginosa has evolved a complex repertoire of strategies to overcome the deleterious effects of these and other toxic intermediates.
The response to oxidative stress is complex and highly coordinated, involving enzymatic and non-enzymatic activities and requiring multiple layers of regulation. In this study, PA2206 has been identified as an important component of transcriptional regulation required for the oxidative stress response and zebrafish lethality in P. aeruginosa. The dual involvement of transcriptional regulators in the oxidative stress response and pathogenesis has previously been reported. For example, OxyR is required for virulence of P. aeruginosa in insect and mouse models of infection The strong shift in lane 4 confirms that PA2206 binds to LysR box 3 overlapping the pPA2214 predicted 235 region and that this region is sufficient for the interaction to occur. PA2206 protein did not cause a shift for any of the other truncated promoter fragments. doi:10.1371/journal.pone.0054479.g004 [39] and also for secretion of cytotoxic factors [49]. In addition to this, Lan and colleagues demonstrated that another oxidative stress response regulator, OspR, influences levels of dissemination of P. aeruginosa in a murine model of acute pneumonia, with inactivation of OspR resulting in increased bacterial virulence [38]. In contrast, a study carried out in S. enterica serovar Typhimurium demonstrated that neither OxyR nor KatG (catalase) influence survival in the presence of neutrophils [40], while AhpC (peroxidise) is not required for virulence in a mouse model [41]. Therefore, microbes may have evolved distinct roles for these transcriptional regulators, consistent with their ecological and niche-specific requirements. It remains to be determined whether the altered pathogenesis observed upon mutation of PA2206 in the zebrafish embryo model of infection is a consequence of its reduced tolerance to oxidative stress or through independent alteration of its in vivo virulence profile.
The molecular mechanism underpinning the global regulatory role of PA2206 involves at least three direct targets; PA2214-15, pvdS, and PA4881, in addition to the possibly indirect regulation of 56 other genes during oxidative stress. PA4881 remains uncharacterised, although it has recently been shown to encode a tandem repeat protein also under the control of the redox responsive LysR protein MexT [50,51]. Annotated as hypothetical, the PA2206-17 genomic region encodes a large number of genes encoding proteins with an apparent metabolic function. Furthermore, the positive regulation by PA2206 of several operons involved in polyamine metabolism may be significant in light of recent reports on the protective role of polyamines against oxidative stress in E. coli [52]. Re-routing metabolic flux has recently been proposed as a key mechanism through which OxyR mediates oxidative stress tolerance in P. aeruginosa [15]. Indeed, the flux of cellular metabolic pathways is emerging as a key adaptive mechanism through which both microbes and higher organisms adapt to the threat of ROS, although the molecular mechanisms governing these relationships remain to be defined [53,54,55,56,57]. It is worth noting the magnitude of the reduction in cell viability observed for the PA2206 2 mutant following exposure to H 2 O 2 ( Table 2) in comparison with the exquisite sensitivity following mutation of OxyR, which controls the battery of enzymes required to mount an acute response to oxidative challenge [14]. Mutation of oxyR in P. aeruginosa and other pathogens has been shown to result in an inability to grow on LB agar plates, possibly due to the autoproduction of concentrations as low as 1.2 mM H 2 O 2 per minute [39,58]. In addition, micromolar concentrations of H 2 O 2 were sufficient to arrest the growth of oxyR mutants in several microbial species reinforcing its position as the master regulator of the oxidative stress response [59,60]. Therefore, PA2206 may represent an important OxyR-independent regulatory link between perception of an exogenous stress and adaptation through realignment of lower metabolic pathways to conserve energy and facilitate survival.
The positive regulation of iron related genes by PA2206, including PvdS and both the FiuIRA and FemIRA ECF systems, supports a role for iron metabolism in the oxidative stress response in P. aeruginosa. Indeed, Palma and colleagues have previously reported induction of several iron-starvation responsive genes upon treatment with H 2 O 2 suggesting either that cells experience iron starvation in response to oxidative stress, or that transient loss of Fur repressor function may occur [32,33]. Furthermore, Wei and colleagues have also recently reported an OxyR-dependent link with iron homeostasis through positive regulation of pvdS during oxidative stress [15]. However, elucidating the full complexity of the role of iron uptake systems during oxidative stress will require further investigation. Notwithstanding this, the binding of PA2206 to the pvdS promoter in the nanomolar range, and the increased pvdS expression in the PA2206 C strain, support a direct role for PA2206 in regulating iron homeostasis during oxidative stress. The concentration of PA2206 protein required to cause a mobility shift of the pvdS promoter was comparable to the recently reported regulation of the pvdS promoter by the LysR regulator of sulphur metabolism, CysB [61]. More recently, OxyR has also been shown to bind to the pvdS promoter [15] and regulation of the pvdS promoter by these three distinct LTTRs highlights the coordination of regulatory inputs that underpins microbial adaptation. However, although cross-talk among LTTRs may facilitate fine-tuning of the microbial stress response, it is important to distinguish the regulatory influences of OxyR and PA2206 during oxidative stress. While 28 genes that were either induced or repressed under conditions of oxidative stress [33] were differentially expressed in the PA2206 C strain compared to an isogenic PA2206 2 mutant, only 6 of these loci were bound by OxyR in the recent study performed by Wei and co-workers [15]. Furthermore, in a separate study, proteomic analysis of the oxyR mutant under iron limitation [62] revealed no overlap with the PA2206 transcriptomic dataset. Therefore, while transcriptomic analysis of the OxyR regulon will be required to fully appreciate the distinct patterns of regulation by OxyR and PA2206, our data would suggest that significant overlap would not be expected.
Complex adaptive responses such as oxidative stress and host pathogenesis require the input of several classes of regulatory protein, transcriptional, post-transcriptional and post-translational. The COG prediction of post-transcriptional regulation or posttranslational modification for genes altered in the transcriptome, including glutathione S-transferase (PA2821), suggested that the oxidative stress phenotype associated with loss of PA2206 function may be multifactorial. Indeed, preliminary proteomic analysis has revealed reduced production of two probable peroxidases in the PA2206 2 mutant and time-kill survival assays revealed that both are required for survival in the presence of 100 mM H 2 O 2 (Haynes & O'Gara unpublished data). As expression of the genes encoding these proteins was unaltered at the transcriptional level, the influence of PA2206 on these targets is likely to be indirect. Characterizing the molecular pathways that link the transcriptional and post-transcriptional systems identified in this study will provide the focus of future investigations.
In summary, PA2206 contributes to an effective oxidative stress response and pathogenicity in P. aeruginosa. The role of PA2206 during oxidative stress appears to be independent of the OxyR pathway, although cross-talk with other transcriptional regulators must be considered. Elucidating the molecular pathways underpinning this crucial aspect of microbial pathogenesis will provide a platform for greater understanding of the process of infection in P. aeruginosa and other serious pathogens. Figure S1 HisTag protein purification of PA2206. Purified protein was loaded on an 10% SDS PAGE gel and visualised following Coomassie Blue staining. Imidazole concentrations used to elute each fraction are detailed below the gel. (PPT) Figure S2 PA2206 expression is induced in response to 1 mM menadione. Bacterial cultures grown to exponential phase were exposed to a 1 mM concentration of menadione and expression of PA2206 was found to be significantly increased. Mean values are represented +/2 standard error. (PPT) Figure S3 PA2206 does not autoregulate its own promoter activity. PA2206-lacZ promoter fusion analysis revealed no difference in b-galactosidase activity in wild-type, PA2206 2 or PA2206 C strains. Data presented contains three biological replicates and is representative of three independent experiments (p-value#0.01 by student's ttest). (PPT) Figure S4 Growth profiling of wild-type and PA2206 2 strains in the presence of ROS. Addition of 20 mM H 2 O 2 to exponentially growing cells led to significant reduction in growth rate in the PA2206 2 strain while the wild-type was unaffected. Similarly, addition of 40 mM menadione also resulted in reduced growth rate in the mutant strain. Significance of the growth impairment was confirmed by statistical analysis (Students ttest * p-value#0.05, ** p-value#0.005, *** p-value#0.001). (PPT) Figure S5 Comparative genomic analysis of the metabolic-centric PA2206 region. Analysis of all available P. aeruginosa genome sequences in which PA2206 homologues were identified, including P. fluorescens Pf-5 and P. fulva 126, based on the Pseudomonas Genome Database. Single genes and operons are denoted by colour and pattern-fill. Gene content and organisation is highly conserved in P. aeruginosa strains. P. fluorescens encodes a PA2206 homologue, which is adjacent to truncated genes corresponding to fragments of PA2207 and PA2212, both of which are downstream of a conserved PA2214-2216 homologous operon. P. fulva encodes a TRAP system inserted between homologues of the PA2213 and PA2216 genes. (PPT) Figure S6 (A) Genome-wide BioPerl analysis of the 59-TTGCCTGGGGTTA-39 sequence revealed 18 promoters that contained putative LysR boxes with similarity (p,1e 204 ) to the PA2206 motif. MEME analysis of these LysR boxes led to the construction of a PA2206 consensus motif. Genes whose expression was altered in the PA2206 C strain are highlighted in bold and the fold changes are indicated in parentheses. (B) EMSA analysis reveals a binding interaction between PA2206 and the PA4881 promoter, forming C1 and C2/C3 complexes in a concentration dependent manner. No interaction was observed at the PA1874 promoter, consistent with its lack of induction in the transcriptomic profile. (PPT)