|Home | About | Journals | Submit | Contact Us | Français|
Inducible defenses against oxidative stress are coordinated by redox-sensitive transcription factors that transduce oxidative damage into differential gene expression. The opportunistic human pathogen Pseudomonas aeruginosa has evolved under physiological and host-derived sources of oxidative stress. Previous work showed that the pqrABC and pqrR genes of P. aeruginosa, all lacking known functions, were induced by treatment of three different isolates of P. aeruginosa with paraquat (PQ), a superoxide-producing agent. Insertional mutation of the pqrABCR genes resulted in hypersensitive phenotypes to a variety of oxidants, although the hypersensitivity to PQ was marginal. Mutation of pqrR and complementation assays showed that PqrR regulated the pqrABC genes in response to PQ. PqrR, a member of the MarR family of transcriptional regulators, contains a C-terminal region with four conserved cysteines, which suggested redox-regulated transcriptional activity. Purified PqrR bound to two discrete sites at the pqrA and pqrR regulatory regions. The in vitro DNA binding activity of PqrR was decreased by exposure to air and reconstituted by treatment with dl-dithiothreitol. Elemental analysis and preliminary electron paramagnetic resonance experiments showed that PqrR contains iron. Interestingly, site-directed mutagenesis of C-terminal cysteines demonstrated that the four conserved cysteine residues are essential for in vivo redox sensing by PqrR.
Pseudomonas aeruginosa is an opportunistic human pathogen that causes chronic infection in cystic fibrosis patients and burn victims and has emerged as an etiological agent of nosocomial infections, especially for patients in intensive care units (33). During the initial stage of the infection process, P. aeruginosa, must adapt to hostile environmental conditions, such as nutrient limitation, osmotic stress, and oxidative stress (19, 34). To colonize a host, P. aeruginosa must be able to evade, counteract, or repair oxidative stress damage.
In the past, and in order to understand the transcriptional response of P. aeruginosa to oxidative stress, genome-wide transcription profiles of P. aeruginosa exposed to oxidants such as paraquat (PQ), hydrogen peroxide, and sodium hypochlorite have been characterized (10, 24, 28, 30). In another study, the genome-wide expression of P. aeruginosa was analyzed under PQ-generated superoxide stress in three strains in parallel: a burn wound isolate, PAO1, and two cystic fibrosis isolates, P. aeruginosa TB and 892 (29). This global transcription experiment showed that only six genes were upregulated in all three strains. Among these six genes, a cluster of four genes, PA0939, PA0940, PA0941, and PA0942, showed the highest induction ratio. These PQ response genes were renamed pqrC, pqrB, pqrA, and pqrR (for paraquat response), respectively (29). In an independent study, pqrR and pqrA were induced by between two- and threefold by exposure to 1 mM hydrogen peroxide (24).
Based on sequence analysis, these pqrABC genes were predicted to form an operon under the transcriptional control of PqrR (Fig. (Fig.1A).1A). The pqrR gene is located upstream of pqrA and is divergently transcribed from the pqrABC genes. The pqrR gene codes for a predicted member of the MarR family of transcriptional regulators. Members of the MarR family are found in gram-negative and gram-positive microorganisms, and they control a variety of biological functions, such as resistance to multiple antibiotics, household disinfectants, oxidative agents, and virulence factors (35). The following proposed roles of the genes in the pqrABCR cluster were suggested by sequence analysis: PqrA contains a thioredoxin-like motif, PqrC is a predicted transcriptional antiterminator, and PqrB has no homology to genes with known functions. Thus, as suggested by structure and regulation, the pqrABCR gene cluster appeared as a candidate for a novel group of genes involved in antioxidant responses.
In this study, we demonstrate that the pqrABC genes form an operon negatively regulated by PqrR, which binds at two sites in the pqrA-pqrR intergenic region. We also establish that PqrR acts as a dimeric redox-sensing transcriptional repressor that contains approximately three iron atoms per monomer. Moreover, four cysteine residues in the conserved C terminus were necessary for in vivo repressor function and redox sensing. Based on structural similarity, we propose that PqrR exemplifies a subfamily of transcriptional regulators within the MarR family.
All bacterial strains and plasmids used in this study are listed in Table Table1,1, and the primers used are listed in Table Table2.2. For long-term storage, all strains were maintained at −70°C in Luria-Bertani (LB) medium containing 20% dimethyl sulfoxide and antibiotics when appropriate.
A broad-host-range vector, pME6001 (kindly provided by L. Quadri, Cornell University, NY), was used to construct a plasmid harboring the pqrR gene under the control of its native promoter. The full coding region of pqrR, including its promoter region (PAO1 genome positions 1030650 to 1031351), was amplified by PCR from P. aeruginosa genomic DNA with primers PA0942F HindIII and PA0942R KpnI, using PfuTurbo DNA polymerase (Invitrogen). The 702-bp fragment was gel purified with a gel extraction kit (Qiagen), digested with HindIII and KpnI, and subsequently ligated into vector pME6001 previously digested with HindIII and KpnI. The ligation mix was used to transform a P. aeruginosa pqrR mutant strain by electroporation. Transformants were selected on LB medium containing gentamicin (50 μg/ml) and Gmr colonies were repurified by single-colony isolation on LB-Gm medium. The identity of the insert was verified by nucleotide sequence determination using an automated DNA sequencer (CEQ 8000 genetic analysis system; UMass Sequencing Core Facility). The recombinant plasmid harboring the cloned pqrR gene was named pWR102 (Table (Table11).
Overnight cultures of P. aeruginosa PAO1 and pqrR were diluted 1:100 into 20 ml of LB medium in 125-ml Erlenmeyer flasks and were grown at 37°C with strong aeration (shaking at 250 rpm) to logarithmic phase. Once cultures reached an optical density at 600 nm of ~0.5, they were split into two samples; one aliquot was left untreated, and the other was treated with 500 μM PQ for 30 min. Cells were collected by centrifugation of 1.5-ml aliquots in a microcentrifuge for 1 min, and total RNA was extracted using an RNeasy minikit (Qiagen). The concentration and purity of RNA samples were quantified by absorbance at 260 and 280 nm in an Eppendorf BioPhotometer 6131. For real-time PCR experiments that required cDNA, 10 μg of total RNA was treated with DNA-free DNase (Ambion) to remove DNA contamination. Treated RNA samples were further confirmed for the absence of DNA contamination by PCR. A QuantiTect reverse transcription kit (Qiagen) was used to synthesize cDNA. One microgram of total RNA was used as a template for cDNA synthesis in a total reaction volume of 20 μl. Aliquots at 2 μl of cDNA preparation each were used for quantitative real-time PCRs.
Total RNA (5 μg) was loaded on 1.25% agarose gels containing 0.25 M formaldehyde. Millennium RNA molecular weight markers (Ambion) were used as a size reference for RNA. The RNA was transferred overnight from agarose gels to Nytran membranes using TurboBlotter setups (Schleicher and Schuell). RNA was cross-linked to Nytran membranes using UV light provided by a cross-linker illuminator (Fisher Scientific). DNA probes specific for pqrA and pqrC genes were prepared by PCR amplification of P. aeruginosa genomic DNA with primer pairs PA0941F and -R and PA0939F and -R, respectively. The 234-bp fragment of pqrA and 327-bp fragment of pqrC were gel purified (Qiagen) and labeled with [α-32P]dCTP (Perkin Elmer Life Sciences) by using a random priming and labeling kit (Stratagene). Both 32P-labeled DNA probes were purified using Sephadex G-25 columns (Pharmacia) to remove unincorporated radioactive labeling. The membranes were hybridized with a 32P-labeled DNA probe at 65°C in QuickHyb solution (Stratagene) and washed, according to the supplier's instructions. A phosphor screen (Amersham Biosciences) was exposed overnight to the dried membranes. The phosphor screen was scanned using a Typhoon 9210 multimode laser scanner (GE Healthcare).
The abundance of pqrC, pqrB, pqrA, and pqrR transcripts was measured by cDNA synthesis, followed by quantitative reverse transcription real-time PCR (qRT-PCR). The reactions were carried out in a DNA Engine Opticon 2 system (MJ Research) using iQ SYBR green supermix (Bio-Rad) with primer pairs pqrC F and pqrC R, pqrB F and pqrB R, pqrA F and pqrA R, and pqrR F and pqrR R, respectively (Table (Table2).2). Each primer set was designed to generate amplicons between 100 and 130 bp in size. Specific primers for pqrR amplification were designed to hybridize upstream of the transposon insertion site in the pqrR strain, which allowed for quantification of pqrR transcripts in this pqrR strain. All reactions were set up according to the supplier's instructions. The housekeeping gene rpsL was used as an internal control for all real-time PCR experiments and amplified by primers rpsL F and rpsL R. The cycling parameters used were as follows: initial denaturation at 95°C for 3 min, 40 cycles at 95°C for 20 s, 59°C for 30 s, and 72°C for 30 s. Fluorescence was read at the end of every cycle. Melting curve analysis was performed from 59°C to 95°C, with continuous fluorescence readings every 0.5°C increment. Standard curves for the quantitation of the number of mRNA transcripts per microgram of total RNA in the quantitative real-time PCR were plotted from the result of amplification reactions using each primer pair and a plasmid template that carried the pqrABCR gene cluster. In order to build this template, the entire pqrABCR cluster was amplified by PCR using primers pqrABCR BamHI F and pqrABCR BamHI R, which introduced convenient BamHI restriction sites. The 1.5-kb PCR product was gel purified (Qiagen gel purification kit), digested with BamHI, and subsequently ligated into plasmid vector pACYC184 digested with BamHI to yield pWR104. The ligation mix was used to transform Escherichia coli DH5α, and transformants were selected on LB plates containing chloramphenicol (50 μg/ml). Similarly, a plasmid containing a control gene, rpsL, was constructed using a TOPO TA cloning kit (Invitrogen). Briefly, a 137-bp fragment of the rpsL coding sequence was amplified by PCR using primers rpsL F and rpsL R. The 137-bp rpsL fragment was cloned into pCR 2.1-TOPO vector to yield pWR105. The pWR105 plasmid was transformed to E. coli TOP10, and colonies containing recombinant plasmid were identified by blue/white colonies on a LB plate containing 50 μg/ml ampicillin and 80 μg/ml 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal). Both pWR104 and pWR105 plasmids were extracted from 4 ml of overnight cultures in LB medium with appropriate antibiotics using a plasmid miniprep kit (Qiagen). Plasmid concentrations were quantified using the Eppendorf BioPhotometer 6131 at a wavelength of 260 nm, and the copy numbers were calculated. Five concentrations of 10-fold serial dilutions (106 to 102 copies) of pWR104 or pWR105 were used as templates to construct standard curves for pqrC, pqrB, pqrA, pqrR, and rpsL. Real-time PCR samples were set up in SYBR green reactions with primers for pqrC, pqrB, pqrA, pqrR, and rpsL. The cycling parameters used were the same as those noted earlier. The standard curve allowed for the quantification of mRNA in the samples. Data was analyzed by Opticon Monitor software. All real-time PCR experiments were performed in independent triplicate trials.
The His6-PqrR recombinant protein of P. aeruginosa was constructed by PCR amplification of genomic DNA with primers PA0942F C-terminal His and PA0942R C-terminal His using Pfu polymerase. The 583-bp fragment was gel purified and digested with NcoI and NotI and then cloned into a similarly digested pET-28b vector. This construct resulted in a six-His tag at the C terminus of PqrR under the control of a T7 promoter. This plasmid was introduced in E. coli DH5α. All transformants were selected on LB medium with 50 μg/ml of kanamycin. The in-frame fusion of the pqrR gene with the six-His tag was confirmed by sequencing both strands of the resulting plasmid, which was named pWR111. Plasmid pWR111 was used to transform E. coli BL21, and transformants were selected for kanamycin resistance. To construct six-His-tagged recombinant proteins of PqrR(Cys148Ser), PqrR(Cys151Ser), PqrR(Cys163Ser), and PqrR(Cys179Ser), plasmids (pWR108, pWR109, pWR110, and pWR106) were amplified with primer pairs PA0942F C-terminal His and PA0942R C-terminal His with Pfu DNA polymerase. The pqrR(Cys185Ser) clone was obtained by amplification of pWR102 with PA0942F C-terminal His and a mutagenic reverse primer, PA0942R Cys185Ser. All DNA fragments were digested with NcoI and NotI and ligated into pET-28b vector at the NcoI and NotI sites, as previously described for the pqrR wild type. The correct insertion of pqrR(Cys148Ser), pqrR(Cys151Ser), pqrR(Cys163Ser), pqrR(Cys179Ser), or pqrR(Cys185Ser) into pET-28b was confirmed by sequencing both strands of the resulting inserts, and the plasmids were named pWR114, pWR115, pWR116, pWR117, and pWR118, respectively. Each plasmid was introduced in E. coli BL21. Transformants were selected on LB plates containing kanamycin (50 μg/ml) and repurified by single-colony isolation.
To express each PqrR recombinant protein, overnight E. coli BL21 cultures of the corresponding strains were diluted 1:100 in 500 ml LB medium with 50 μg/ml kanamycin and incubated with strong aeration (250 rpm) at 37°C until the optical density at 600 nm reached 0.5 before adding 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). After 2 h, cells were collected by centrifugation for 20 min at 8,000 × g at 4°C. Cell pellets were resuspended in 5 ml of lysis buffer (50 mM Na2HPO4, 300 mM NaCl, 10 mM imidazole [pH 8.0], and protease inhibitor cocktail [Roche]) and sonicated for 2.5 min (10 s, pulse on; 10 s, pulse off). Cell debris was removed from the cell extract by centrifugation at 8,000 × g for 20 min. The cleared cell extracts were gently mixed with 3 ml of Ni-nitrilotriacetic acid agarose resin for 1 h. These mixtures were transferred to 5-ml polypropylene columns (Qiagen). Each column was washed twice with 6 ml imidazole washing buffer 1 (50 mM Na2HPO4, 300 mM NaCl, 20 mM imidazole [pH 8.0]) and washed twice with washing buffer 2 (50 mM Na2HPO4, 300 mM NaCl, 30 mM imidazole [pH 8.0]. Finally, PqrR recombinant protein was eluted four times with 1.5 ml of buffer containing 50 mM Na2HPO4, 300 mM NaCl, and 250 mM imidazole (pH 8.0). Each fraction was analyzed by 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis. All four fractions were combined and dialyzed overnight against 20 mM Tris-Cl, 100 mM NaCl, 1 mM dl-dithiothreitol (DTT), and 10% glycerol at pH 8.0. All steps in the protein purification were carried out at 4°C. Protein concentration was determined by the Bradford protein assay, using bovine serum albumin as a standard (Pierce). For UV-visible (UV-VIS) spectra, the concentrations of wild-type and cysteine-to-serine mutant PqrR proteins were adjusted to 0.1 mg/ml before being scanned with a UV-2401PC dual-beam spectrophotometer (Shimadzu).
Five PqrR mutant proteins were constructed by replacement of a C-terminal cysteine residue (residue 148, 151, 163, 179, or 185) by serine using PCR-based site-directed mutagenesis, with pWR102 as a template. The mutant alleles pqrR(Cys148Ser), pqrR(Cys151Ser), or pqrR(Cys163Ser) were constructed by using the QuikChange II-E site-directed mutagenesis kit (Stratagene) with mutagenic primers Cys148Ser (F and R), Cys151Ser (F and R), and Cys163Ser (F and R). Cloning of these fragments resulted in pWR108, pWR109, and pWR110, respectively. To create PqrR(Cys179Ser), the P. aeruginosa PAO1 pqrR gene was amplified from genomic DNA with a forward primer, PA0942F HindIII, and a mutagenic reverse primer, Cys179Ser KpnI, with PfuTurbo DNA polymerase (Stratagene). The PCR product was gel purified, digested with KpnI and HindIII, and cloned into pME6001 to yield pWR106. PqrR(Cys185Ser) was constructed in a manner similar to the construction of PqrR(Cys179Ser), but genomic DNA was amplified with PA0942F HindIII and Cys185Ser KpnI primer pairs, and the DNA fragment was cloned into pME6001 at the HindIII and KpnI sites to yield pWR107. E. coli XL1-Blue was transformed with these plasmids, and transformants were selected on LB medium with gentamicin (15 μg/ml). The identities of all mutant pqrR alleles were verified by determination of their nucleotide sequences using an automated DNA sequencer (CEQ 8000 genetic analysis system; UMass Sequencing Core Facility).
The molecular mass of the native His6-PqrR was determined by gel filtration chromatography on a Sephadex G-75 column connected to a fast protein liquid chromatography device (Pharmacia Biotech). The column was equilibrated with 20 mM Tris-Cl and 100 mM NaCl at pH 8.0. The molecular mass markers (Sigma) aprotinin (6.5 kDa), cytochrome C (12.4 kDa), carbonic anhydrase (29 kDa), albumin (66 kDa), and blue dextran (2,000 kDa) were independently applied to a column to generate a standard curve. Protein elution was monitored by absorbance at 280 nm.
A PerkinElmer Optima 4300 DV inductively coupled plasma (ICP) optical emission spectrometer was used to quantify metal concentrations of PqrR in PqrR and Cys mutants. This instrument is equipped with a 40-MHz free-running generator and a segmented-array charge-coupled-device detector. The sample introduction system consisted of a concentric nebulizer with a cyclonic spray chamber. The concentrations of Mn, Fe, Co, Ni, Cu, and Zn in each sample were determined at a λ of 294.920, 238.208, 228.616, 231.604, 327.393, and 206.200 nm, respectively.
X-band electron paramagnetic resonance (EPR) spectra were obtained with a Bruker ELEXSYS E500 X-band spectrometer over a magnetic field range of 3,000 ± 500 G at a microwave frequency of 9.62 GHz and 200 mW power. Samples were contained in 3-mm quartz tubes, and temperatures between 4 and 30 K were maintained using a helium flow cryostat (Oxford). A modulation frequency of 100 kHz, a modulation amplitude of 5 G, and a time constant of 10.24 ms were used to obtain the spectra reported. Both air oxidized and samples reduced with dithionite were analyzed. Spectral simulations were performed using the XSophe software package, version 1.1.4 (15), and the standard spin Hamiltonian, Ĥ = βB0gS (1), where β is the Bohr magnetron, B0 is the applied magnetic field, g is the g value, and S is the spin operator.
To construct a plasmid containing a 115-bp intergenic region of pqrA and pqrR, P. aeruginosa genomic DNA was amplified with intergenic 0941-42F and -R primers using Taq polymerase. This primer pair added HindIII and KpnI sites to the ends of the intergenic region. The 115-bp PCR fragment was cloned into pCR 2.1-TOPO by using a TOPO TA cloning kit (Invitrogen) to generate pWR113. The plasmid was used to transform E. coli TOP10 (Invitrogen) into Kanr strains. Transformants were selected by blue/white colony screening on LB plates containing kanamycin (50 μg/ml) and X-Gal (80 μg/ml). Kanr transformants containing inserts were repurified by single-colony isolation. The cloning of the pqrA-pqrR intergenic region was verified by nucleotide sequence determination of both strands of the insert. To label the pqrA-pqrR intergenic region, plasmid pWR113 was digested with EcoRI, and the resulting overhanging 5′ ends were filled in with Klenow PolI fragments and [α-32P]dATP. Labeled DNA was separated from the unincorporated label using DNA purification spin columns (Qiagen). Purified proteins were diluted in 20 mM Tris-Cl (pH 8.0), 100 mM NaCl, 5 mM DTT, and 5% glycerol. Labeled DNA (14 fmol) was incubated with wild-type or mutant PqrR proteins in a 20 μl binding reaction [20 mM Tris-Cl (pH 8.0), 50 mM KCl, 10% glycerol, 100 μg/ml bovine serum albumin, 100 μg/ml salmon sperm DNA, 100 μg/ml poly(dI-dC), and 5 mM DTT]. The reactions were incubated at room temperature for 30 min before loading onto 4% native polyacrylamide TEA gels (40 mM Tris-Cl [pH 8.0], 3 mM sodium acetate, 1 mM EDTA). The gel electrophoresis was done at 200 V for 2 h at 4°C, before being exposed overnight in storage phosphor screens (Amersham Biosciences). The phosphor screens were scanned using a Typhoon 9210 multimode laser scanner. The intensity of the signal associated with each band was quantified by digital densitometry using ImageQuant 5.2 (Molecular Dynamics). To determine binding activity under oxidizing conditions, reactions were carried out as described previously, but DTT was omitted in all steps. The only initial amount of DTT present in these reactions was the residue from the protein storage buffer and was estimated to be ≤0.3 μM. To determine the reversibility of PqrR binding under different redox conditions, the PqrR binding reactions were carried out as described previously with no addition of DTT and incubated at room temperature for 15 min. Each reaction was split into two tubes, and 5 mM DTT was added to one of them. Both samples were incubated for another 15 min before loading onto a native polyacrylamide gel. Half-maximal saturations were estimated as a measure of affinity from binding curves fitted to the formula Y = Bmax × X/(KD + X), where KD is the protein concentration that binds to half the recognition sites at equilibrium, a Bmax value of 1 is maximum number of binding sites, and X and Y are the x axis and y axis values, respectively. Since PqrR binds to at least two sites, this estimated KD does not represent a true binding constant, and it is referred to as the half-maximal saturation.
The procedures for DNase I protection assays (footprinting) were performed using a standard protocol (3). Essentially, the 115-bp pqrA-pqrR intergenic region cloned in plasmid WR113 was used as the target DNA for the DNase I footprinting assay. The top strand was labeled by digestion of pWR113 with EcoRI, and the resulting overhanging 5′ ends were filled in with Klenow PolI fragment and [α-32P]dATP, plus unlabeled dTTP. The insert containing the labeled pqrA-pqrR region was separated from the vector by electrophoresis in a 2% agarose gel and recovered using a gel extraction kit (Qiagen). The purified DNA was digested with HindIII and repurified using a DNA purification spin column (Qiagen). The bottom strand was labeled by digestion of pWR113 with BamHI and XbaI. The insert containing the labeled pqrA-pqrR region was separated from the vector by electrophoresis in a 2% agarose gel and recovered using a gel extraction kit (Qiagen). The 5′-end overhangs were filled in with Klenow PolI fragment and [α-32P]dATP, plus unlabeled dTTP and dGTP. The labeled target DNA was purified using a DNA purification spin column (Qiagen). The purified DNA was digested with XhoI and repurified using a DNA purification spin column (Qiagen). The integrity of the target DNA was verified by electrophoresis. PqrR binding reactions with either a top- or bottom-strand-labeled probe (~140 fmol) were set up under conditions similar to those used for the gel mobility shift assay. The concentrations of PqrR used were 168 nM, 33.6 nM, 6.7 nM, 1.3 nM, 268.8 pM, and 53.8 pM. Reactions (total volume, 20 μl) were incubated at room temperature for 30 min, and 5 μl DNase I buffer was added immediately before treatment with Baseline-Zero DNase I (Epicentre Biotechnologies). The DNA was digested for 30 s, and the reaction was stopped by addition of 950 μl of stop solution (95% ethanol, 0.375 M ammonium acetate, 5 μg/ml yeast tRNA). Samples were precipitated at −20°C for an hour, and nucleic acids were recovered by centrifugation for 15 min at 14,000 rpm in an Eppendorf Microfuge. Nucleic acid pellets were washed with 70% ethanol and resuspended in 6 μl formamide loading buffer. DNase I digestion products were resolved by electrophoresis in 6% polyacrylamide Tris-borate-EDTA gels containing 7 M urea. The gels were transferred to a filter paper, dried at 80°C under vacuum, and used to expose phosphor screens overnight. The phosphor screens were scanned with a Typhoon 9210 multimode laser scanner, and digital images were saved as TIFF files.
Previous work has shown that exposure of Pseudomonas aeruginosa cultures to the superoxide-producing agent PQ elevated the steady-state levels of pqrABCR transcripts. Since the product of pqrA shows homology with thioredoxin, the activation of pqrABCR transcription led to the hypothesis that the pqrABCR genes might play a role in the inducible defenses of P. aeruginosa against oxidative stress. To determine if the pqrABCR genes play a role in defense against oxidative stress, we compared the growth of P. aeruginosa strains carrying insertional mutations in each individual gene of the pqrABCR cluster with that of an isogenic wild-type strain. Cultures of each P. aeruginosa strain were either grown untreated or exposed to the redox-cycling agents PQ and plumbagin or the thiol-specific oxidant diamide (Fig. (Fig.22).
Inactivation of the pqrA, pqrB, pqrC, or pqrR genes had no effect on the growth rate of untreated cultures (Fig. (Fig.2A).2A). When exposed to PQ (500 μM), only the pqrA strain was hypersensitive (data not shown). At a higher PQ concentration (1 mM), the pqrA strain was hypersensitive and the pqrC and pqrB strains displayed marginal hypersensitivity, while a pqrR mutant was not hypersensitive (Fig. (Fig.2B).2B). Oxidative stress induced by PQ had less impact on the growth of the pqrC, pqrB, pqrA, and pqrR strains in comparison to that of other oxidants. All mutant strains were hypersensitive to plumbagin (250 μM) and diamide (4 mM) (Fig. 2C and D). Additionally, growth inhibition assays on plates containing LB-agar medium showed that pqrC, pqrB, pqrA, and pqrR had increased sensitivity to H2O2 and sodium hypochlorite (data not shown). The pqrA mutant strain was consistently sensitive to all oxidants tested. The results indicate that the pqrCBAR genes may play a role in the resistance of P. aeruginosa to oxidative stress elicited by a variety of oxidants.
The pqrR gene codes for a predicted member of the MarR family of transcriptional regulators, and it is transcribed divergently from the pqrABC genes (Fig. (Fig.1).1). The predicted secondary structure of PqrR showed a conserved winged-helix structure common to MarR family proteins (Fig. (Fig.1B).1B). Interestingly, sequence comparisons with other predicted members of the MarR family revealed a conserved cysteine-rich region located at the C-terminal ends of 16 PqrR homologs found among gram-negative bacteria from diverse habitats, including marine bacteria, soil bacteria, thermophiles, and plant and human pathogens (Fig. (Fig.1C).1C). Cysteine residues have been known to play a major role as redox sensors for many regulatory proteins (4). Thus, the conserved cysteine domain found in PqrR may be involved in sensing oxidative stress and regulating the expression of the pqrABCR operon.
To demonstrate the regulatory role of the PqrR protein on the expression of pqrABC genes, Northern blot analysis was used to measure the expression of pqrC and pqrA genes in isogenic wild-type and pqrR strains. Northern blot analysis of PAO1 (wild type) failed to detect transcripts of pqrC or pqrA (Fig. (Fig.3).3). Conversely, in the pqrR strain, expression of both pqrC and pqrA was robust and estimated to be at least 50-fold higher than that of the wild-type strain, although precise quantitative comparison of expression was difficult due to the extremely low level of transcripts in the wild-type strain. These results suggested that the product of the pqrR gene represses the expression of the pqrABC genes. To test this hypothesis further, we performed a genetic complementation test on the pqrR strain. The transformation of the pqrR strain with plasmid pWR102 (containing the pqrR gene) resulted in a substantial decrease in the expression levels of pqrC and pqrA. Conversely, control transformation of the pqrR strain with an empty cloning vector (pME6001) resulted in no decrease in pqrC or pqrA mRNA levels. These results indicated that pqrR codes for a repressor that tightly regulates the expression of the pqrA and pqrC genes.
The analysis of pqrA and pqrC expression by Northern blotting provided insight into the transcriptional structure of the pqrABC genes. Only one transcript, with an estimated size of 0.8 kb, was detected by the pqrC probe. The size of the pqrC transcript was consistent with the sum of the sizes of the three genes (pqrA, pqrB, and pqrC), which strongly suggested that the pqrABC genes are cotranscribed. Alternatively, Northern blotting showed two transcripts for pqrA, with estimated sizes of 0.2 and 0.8 kb. The sizes of these transcripts suggested that pqrA could be transcribed as a monocistronic message or a pqrABC polycistronic message. These results are consistent with previous sequence analysis that suggested that the pqrABC genes form an operon with a promoter located in the pqrA-pqrR intergenic region (29).
Given the repressor role of PqrR, we hypothesized that PqrR mediates the induction of the pqrABC operon with PQ. To test the potential role of PqrR in modulating the expression of the pqrABC operon in response to oxidative stress, quantitative real-time PCR was used to measure the expression levels of the pqrA, pqrB, pqrC, and pqrR transcripts in both untreated cultures or cultures treated with 500 μM PQ, as described in Materials and Methods (Fig. (Fig.4).4). This concentration of PQ was chosen since it induces oxidative stress but has a minor effect on aerobic growth in LB broth when added to cultures during logarithmic growth (data not shown). The basal expression levels of pqrC, pqrB, pqrA, and pqrR were relatively low in the PAO1 wild type (Fig. (Fig.4A).4A). The relative abundance of all transcripts increased with the addition of PQ, with significant differences for the pqrC, pqrB, and pqrA genes. Plumbagin (250 μM) and diamide (4 mM) were comparatively weak inducers under the same conditions used for PQ treatment, enhancing the expression of pqrA by twofold or less (data not shown).
To determine whether this induction of the pqrABC operon under oxidative stress depended on pqrR, the copy numbers of these transcripts were measured in a pqrR strain, untreated or treated with PQ (Fig. (Fig.4B).4B). The expression of rpsL was used as the internal control (Fig. (Fig.4C).4C). In the absence of pqrR, the copy numbers of pqrABCR transcripts were not significantly altered by the treatment with PQ. However, the basal level of the pqrC, pqrB, or pqrA transcript was more than 500-fold higher in the pqrR strain with respect to that of the wild-type strain. These real-time PCR results were consistent with the substantial differences observed by Northern blot analysis (Fig. (Fig.3)3) and supported the repressor role of PqrR. In addition, the basal expression of pqrR in the pqrR strain was 16-fold higher than that of the wild type (Fig. (Fig.4B,4B, inset). This observation suggested that PqrR regulates the expression of the pqrR gene with an autoregulatory mechanism. In summary, the expression of the pqrC, pqrB, and pqrA genes in the strain lacking pqrR remained unchanged by the treatment with PQ. Therefore, the induction of the pqrABC operon by PQ depended on PqrR.
Members of the MarR family of transcriptional regulators modulate transcription by binding to the promoter regions of their target genes. To characterize the potential interaction between PqrR and the regulatory region of the pqrABC operon in vitro, PqrR was purified as a C-terminal six-His-tagged recombinant protein. The His6-PqrR was purified in native form to near homogeneity from cell lysates using Ni-nitrilotriacetic acid columns (see Fig. S1A in the supplemental material). Molecular mass analysis of purified His-tagged PqrR by gel filtration revealed that PqrR exists as a dimer in solution, a characteristic oligomeric state for MarR family proteins (data not shown). Interestingly, purified His6-PqrR at a concentration of ~1 mg/ml displayed a dark red color. Moreover, the UV-VIS spectrum of wild-type PqrR featured an absorbance peak at 420 nm characteristic of proteins containing iron-sulfur (Fe-S) clusters (see Fig. S1B in the supplemental material) (16, 20). These observations suggested that PqrR contained a prosthetic group. Elemental analysis of PqrR by ICP optical emission spectrometry demonstrated that PqrR contains iron, with a stoichiometry of ~2.6 Fe atoms per monomer (Table (Table3).3). The presence of iron bound to PqrR, the red color of purified protein preparations, the presence of four conserved C-terminal cysteine residues, and the UV-VIS absorbance spectrum suggested that PqrR contains a prosthetic group that includes iron.
In order to test the ability of PqrR to bind DNA, gel mobility shift assays were performed with purified His6-PqrR and a 115-bp DNA fragment containing the entire intergenic region between pqrA and pqrR, as described in Materials and Methods. Purified His6-PqrR retarded the mobility of the pqrA-pqrR intergenic region, indicating the formation of protein-DNA complexes (Fig. (Fig.5).5). As observed for other MarR family proteins, PqrR formed more than one specific protein-DNA complex, which suggests multiple binding sites. The half-maximal saturation for the binding of PqrR to the pqrA-pqrR intergenic region was 7.3 nM (Fig. (Fig.5B5B).
To map the binding sites of PqrR at the pqrA-pqrR intergenic region, we performed DNase I protection assays. PqrR protected two distinct sites from DNase I digestion (Fig. 6A and B). Site I is a 22-bp region, proximal to the pqrA translational start site, and contains the inverted repeat sequence 5′-GTATCGataCGATAC-3′ (Fig. (Fig.6C).6C). Site II is a 23-bp region, proximal to the pqrR transcriptional start site, that contains the imperfect inverted repeat sequence 5′GTATCGactAAAAAC-3′. The presence of inverted repeats at PqrR binding sites is consistent with the dimeric structure of the protein. Thus, purified PqrR interacts specifically with the pqrA-pqrR intergenic region at two distinct sites.
Since PqrR bound to DNA at the pqrA-pqrR intergenic region repressed the expression of the pqrABC operon and was necessary for the induction of the pqrABC operon under oxidative stress, we hypothesized that the DNA binding activity of PqrR might be modulated by redox status. To test this possibility, we performed gel shift assays under reducing (5 mM DTT in the binding reaction) and oxidizing (≤0.3 μM DTT in the binding reaction, the residual concentration from dilution of protein stock) conditions (Fig. (Fig.5B).5B). PqrR under reducing conditions bound to DNA with an estimated half-maximal saturation of 7.3 nM. When PqrR was incubated with only a residual concentration of DTT (≤0.3 μM), the protein bound to DNA with a higher half-maximal saturation, estimated at 33 nM (Fig. (Fig.5B).5B). This decrease in binding affinity was reversible by reincubation of PqrR in the presence of 5 mM DTT, and samples thus treated bound to DNA with an estimated half-maximal saturation of 7.9 nM (Fig. (Fig.5B).5B). Therefore, redox conditions reversibly modulated the binding activity of PqrR to the pqrA-pqrR intergenic region.
The primary structure of PqrR displays a cysteine-rich C-terminal domain conserved through 16 species of gram-negative bacteria (Fig. (Fig.1C).1C). Since cysteine residues have been found to be involved in redox sensing in several transcription factors (13, 14, 27, 36), we hypothesized that the conserved cysteine residues play a role in the redox sensitivity of PqrR, arguably by providing sulfur ligands for the assembly of the Fe-containing prosthetic group. To determine the importance of the conserved cysteine region of PqrR in regulating the expression of the pqrABC operon, each of the five C-terminal cysteine residues in PqrR was replaced by serine using site-directed mutagenesis. The five corresponding mutant alleles of pqrR, together with their natural promoters, were cloned and expressed in a pqrR host using a complementation assay. As controls, wild-type pqrR and a deletion derivative lacking the 39 C-terminal residues were cloned and used to transform a pqrR-deficient host. The transcriptional activity of pqrA was measured as a reporter of PqrR repressor activity during undisturbed growth or when exposed to 500 μM PQ (Fig. (Fig.7).7). Transformation of a pqrR-deficient host with a plasmid-borne pqrR wild-type allele resulted in a decrease in pqrA transcription of more than 500-fold, compared with that from transformation with the empty cloning vector pME6001 (Fig. (Fig.7).7). Removal of the PqrR C-terminal domain containing the five cysteine residues resulted in a 100-fold increase in pqrA transcription, compared with the levels provided by a wild-type pqrR allele. Replacement of any of the four conserved cysteines (Cys148, Cys151, Cys163, and Cys179) resulted in decreased repressor function, compared with that of the wild type, while replacement of the nonconserved cysteine (Cys185) had only a negligible effect on repressor function (Fig. (Fig.77).
The effect of cysteine replacements on repressor function was paralleled by the results obtained after we measured the capacity to transduce oxidative stress (Fig. (Fig.7).7). The transcriptional activity of pqrA in a pqrR-deficient host transformed with wild-type pqrR was induced by exposure to PQ (Fig. (Fig.7).7). Conversely, when a pqrR host was transformed with pqrR alleles harboring either a deletion of the C-terminal domain or replacements of any of the four conserved cysteines, pqrA transcriptional activity was not induced by PQ-mediated stress (Fig. (Fig.7).7). A pqrR strain transformed with the pqrR allele harboring a replacement of the nonconserved cysteine (Cys185) displayed activation of pqrA activity similar to that of the strain transformed with the wild-type pqrR allele. These results showed that the four conserved C-terminal cysteines are essential to the ability of PqrR to regulate the transcriptional activity of pqrA and to transduce redox stress in vivo.
The resulting pqrR(Cys148Ser), pqrR(Cys151Ser), pqrR(Cys163Ser), pqrR- (Cys179Ser), and pqrR(Cys185Ser) mutant genes were cloned and expressed as six-His-tagged fusions. The corresponding mutant proteins were purified in their native state by Ni-affinity chromatography and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis for purity and integrity (see Fig. S1A in the supplemental material). Interestingly, while purified wild-type PqrR protein was red, Cys148Ser, Cys151Ser, and Cys163Ser were colorless. Conversely, Cys179Ser was slightly colored, and Cys185Ser was red. These differences in color among mutants suggested that an iron-containing group present in wild-type PqrR is absent or defective in these mutants. Moreover, the wild-type PqrR protein showed a broad absorption peak at 420 nm, which is common for proteins containing Fe-S clusters (20) (see Fig. S1B in the supplemental material). In contrast, Cys148Ser, Cys151Ser, and Cys163 did not show this absorption peak, while Cys179Ser and Cys185Ser displayed broad peaks at 420 nm, similar to that of a wild-type PqrR protein (see Fig. S1B in the supplemental material). Metal content analysis by ICP emission spectrometry of the purified PqrR mutants confirmed that mutants Cys148Ser, Cys151Ser, and Cys163Ser contain less than 0.1 atoms of iron per monomer (Table (Table3).3). Conversely, mutant PqrR(Cys185Ser) contained as much iron as the wild type, while the iron content of mutant Cys179Ser was 1.4 atoms per monomer of PqrR (Table (Table3).3). These results suggested that cysteine residues 148, 151, 163, and 179 participate in the assembly of the iron-containing group.
To investigate the role of the PqrR C-terminal cysteine residues in redox sensing, the promoter binding activities of PqrR and the mutant proteins were compared by gel mobility shift assay under reducing or oxidizing conditions (Fig. (Fig.8).8). Under reducing conditions, 60% of the DNA target was retarded by PqrR. Conversely, when PqrR was exposed to air with only residual amounts of DTT in the binding buffer, the binding activity decreased to 5% of the retarded DNA target. As shown previously, when DTT was added to the oxidized sample, the binding activity of PqrR increased to 40% of the shifted target. In contrast to the wild-type PqrR, the binding activities of the Cys163Ser, Cys148Ser, and Cys151Ser mutants were not affected under the redox conditions (Fig. (Fig.6B;6B; see also Fig. S2 in the supplemental material). Interestingly, the PqrR mutant carrying a mutation in Cys179, which in vivo displayed reduced activity and insensitivity to oxidation, was as sensitive to redox status in the gel mobility shift assay, although not as much as the wild-type protein (Fig. S2). Finally, the Cys185 mutant showed redox sensitivity similar to that of wild-type PqrR (see Fig. S2 in the supplemental material). These results are consistent, with the exception of mutant Cys179Ser, with the in vivo phenotypes of conserved cysteine replacements mutants in the complementation assay, i.e., the inability to induce transcription of pqrA under oxidative stress (Fig. (Fig.77).
PqrR is a member of the MarR family of transcriptional regulators, which modulate adaptive responses by binding an effector molecule or by posttranslational modification (21, 22, 26). In particular, members of the OhrR subfamily of MarR proteins transduce oxidative stress signals by reversible modification of thiol groups (14). For example, the OhrR protein of Xanthomonas campestris reacts with organic hydroperoxides in a process that produces intermolecular cross-links between two conserved cysteine residues (Cys22 and Cys127) (26). The Cys22 of OhrR acts as a sensor for organic hydroperoxide, resulting in a sulfenic acid intermediate, which rapidly reacts with the thiol group of Cys127 to form a disulfide bridge. This conformation results in the inactivation of the OhrR repressor function and induction of ohr, coding for a thiol peroxidase that catalyzes the conversion of organic hydroperoxides into alcohols (11).
In contrast to the thiol-based redox switch of OhrR, we propose that an iron-containing prosthetic group acts as a redox switch in PqrR that regulates DNA binding activity and transcriptional repression. Based on the evidence presented, we propose that this iron-containing prosthetic group is an iron-sulfur cluster, as found in other redox-sensitive transcription factors, such as SoxR, IscR, and FNR. To characterize the iron-containing group in PqrR, we have obtained preliminary EPR spectra on air-oxidized and dithionite-reduced samples of PqrR (see Fig. S3 in the supplemental material). The EPR spectra were consistent with the presence of a [3Fe-4S] cluster (6). The oxidized state (as isolated) exhibits a rhombic EPR signal centered at a g value of ~2 that decreases in intensity upon addition of dithionite. EPR spectral simulation provided the following g value: gx of 2.030, gy of 2.006, and gz of 1.955, using optimized line widths of 5.35 G (x), 16.45 G (y), and 23.75 G (z). Spectra collected over temperatures ranging from 4 to 30 K showed that the EPR signal was temperature dependent, with maximum signal intensity at 3,384 G at ~10K and decreasing at higher temperatures (data not shown). The rhombic EPR spectrum observed for PqrR is also similar to signals observed for [3Fe-4S] clusters in other proteins, such as oxidized ferredoxin II from Desulfovibrio gigas and ferredoxin DbfA3 from Terrabacter sp. (23, 32). In addition, preliminary Fe K-edge EXAFS analysis reveals two shells of scattering atoms that can be attributed to Fe-S vectors at ~2.3Å and Fe-Fe vectors at ~2.9Å (data not shown) (2, 5). These preliminary spectral data provide evidence for the presence of a [3Fe-4S] cluster associated with PqrR and, if further confirmed, may provide a mechanistic insight into the capacity of PqrR to transduce redox signals into gene regulation. However, only 15% of the aerobically purified PqrR displayed the EPR signal, suggesting either that the majority of the protein preparation had lost the iron group or that the in vivo form of the iron group is different. This observation is consistent with the fact that PqrR preparations rapidly lost DNA binding ability in the absence of chemical reductants, followed by decoloration of the protein solution and precipitation. Since His-tagged protein cannot be purified in the presence of strong reductants without interfering with the Ni binding, the determination of the iron-containing group in PqrR will demand either using a different purification method or modifying the His-tagged approach by performing the purification under anaerobic conditions.
PqrR specifically binds at two sites in the intergenic region between pqrA and pqrR (Fig. (Fig.5).5). The position of these cognate sites for PqrR binding was consistent with transcriptional repression of the pqrABC genes and autoregulation of pqrR observed in qRT-PCR assays (Fig. (Fig.3).3). The PqrR binding site proximal to pqrA is centered on the perfect hexameric inverted repeat GTATCGataCGATAC, consistent with reports for other members of the MarR family. Alternatively, the PqrR binding site proximal to the divergent pqrR gene is centered on the imperfect repeat CAAAAAtcaGCTATG. Comparison of the regulation ratios between wild-type and pqrR strains shows a smaller effect of the deletion of pqrR on pqrR expression than that on pqrABC expression (Fig. (Fig.4).4). Careful titration of PqrR binding will reveal if binding affinities differ between these cognate sites. The number of bands obtained with gel mobility shift assay experiments was consistent with the number of DNase I-protected sites. The two protected sites explain two high-affinity bands, as either one or both sites are occupied. A third shifted species becomes apparent only after more than 90% of the target is bound, which can be explained by a third, low-affinity binding site that fails to be protected from DNase I digestion. We cannot discard the possibility of looping of the intervening DNA by dimer-dimer interaction, although there is no evidence of this interaction in the literature on the MarR family, and there are no hypersensitive DNase I sites in the footprints.
We propose that the DNA binding activity of PqrR is reversibly affected by the redox status of its iron-containing prosthetic group, based on the following lines of evidence. Reduced PqrR lost 90% of its DNA binding activity after 20 min of incubation in the absence of reducing agents (Fig. (Fig.5).5). Treatment of these oxidized PqrR samples with DTT restored the DNA binding activity significantly (Fig. (Fig.5).5). Mutations that replace the four conserved, C-terminal cysteines in PqrR result in the loss of repressor function and the inability of PQ to induce transcription of pqrA in vivo (Fig. (Fig.8).8). Purified PqrR mutants carrying the same cysteine replacements lacked significant amounts of iron (Table (Table3),3), and their DNA binding activity was not affected by redox status, while replacement of the nonconserved cysteine resulted in no change of in vivo or in vitro properties (see Fig. S3 in the supplemental material).
Replacement of each of the C-terminal cysteines with serine clearly showed that the four conserved cysteines are crucial for the formation of the prosthetic group (Table (Table3)3) and for redox sensing in vivo (Fig. (Fig.7).7). Conversely, PqrR and its Cys185Ser derivative have similar Fe stoichiometry and redox-sensing/DNA binding activity, both in vivo and in vitro, indicating that the Cys185 residue does not significantly contribute to metal binding or PqrR DNA binding activity (Table (Table33 and Fig. Fig.7;7; see also Fig. S2 in the supplemental material). Given the structural conservation of the functional cysteines across a number of members of the MarR family (Fig. (Fig.1),1), we hypothesize that these uncharacterized proteins will probably be found to be redox and/or metal-responsive transcriptional regulators.
The apparent connection between the environmental redox status and the activity of PqrR as a transcriptional repressor strongly suggests that the oxidation state of the iron cluster could act as the sensor of oxidative stress. This behavior could be similar to the redox-sensing mechanism found for SoxR, a member of the MerR family of transcriptional regulators that contains a [2Fe-2S] cluster per monomer that senses superoxide and nitric oxide stress (13). In E. coli and Salmonella enterica, superoxide oxidizes the SoxR [2Fe-2S] clusters, which upregulates the expression of SoxS. Increasing levels of SoxS leads to the activation of many genes whose products are responsible for the removal of superoxide and cellular repair (7). It is interesting to note that the P. aeruginosa SoxR homolog is not involved in resistance to oxidative stress but plays a role in quorum sensing (12, 25). Another possible mechanism might involve a cluster conversion from a [3Fe-4S] to a [4Fe-4S] cluster upon reduction, similar to the cluster conversion involved in aconitase activation (8). Alternatively, the redox-sensing mechanisms of PqrR might involve the assembly/disassembly of a putative Fe-S cluster in a fashion similar to that of the E. coli FNR [4Fe-4S] clusters (31). Since Cys mutants of PqrR that do not form the Fe cluster show a substantial deficiency in transcriptional repression in vivo, it is logical to conclude that assembly/disassembly of the Fe cluster could determine transcriptional repression. In this scenario, oxidation of the Fe cluster accelerates the rate of disassembly, decreasing DNA binding activity and inducing expression of the pqrABC operon.
Finally, it is tempting to propose a simple model, in which pqrR-mediated repression is alleviated in the presence of certain oxidants, which allows expression of the pqrABC genes, which in turn are supposed to code for antioxidant functions. Indeed, several lines of experimental evidence point at the possibility of this scenario. Among the evidence presented, this model is supported by the regulation of the pqrABCR genes and the phenotypes of the corresponding mutants. However, it is possible that the hypersensitivity of the mutants in the pqrABC genes might be due to polar effects. On the other hand, the phenotypes associated with mutations in the pqrR gene are unlikely to be the consequence of polar effects, since there is no predicted downstream open reading frame in the P. aeruginosa chromosome.
Moreover, in this simple model, a mutation that eliminates pqrR function should result in lower sensitivity to oxidants. We did not see this lowered sensitivity but observed that a pqrR mutant is either as sensitive as or more sensitive than the isogenic wild-type strain. It is possible that PqrR locally represses the pqrABC genes but globally activates other unlinked genes. A mutation that abolishes PqrR would also eliminate these other potential, protective functions.
We thank Robert W. Herbst for assistance with acquiring EPR data and simulation. Tünde Mester assisted with UV-VIS spectrophotometry, and Carla Risso and Steve Sandler provided useful suggestions on the manuscript. The Pseudomonas Genome Project (University of Washington) shared mutant strains, and Luis Quadri donated plasmid pME6001. Pat Schloss shared the RT-PCR cycler.
W.R. was supported by a fellowship from the Royal Thai Government, and P.P. was supported by a UMass Amherst Faculty Research Grant.
Published ahead of print on 28 August 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.