INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Pseudomonas aeruginosa generates metabolic energy primarily through aerobic respiration. This process, involving a four-electron reduction of molecular oxygen (O₂) to water, can be potentially dangerous to the cell. Specifically, aberrant electron flow from the electron transport chain or cellular redox enzymes to O₂ can lead to the production of reactive oxygen intermediates (ROIs). These include superoxide (O₂), hydrogen peroxide (H₂O₂), and hydroxyl radical (HO·). Furthermore, bacteria can be exposed to exogenous ROIs, especially during infection of humans, where phagocytes (e.g., neutrophils) mount a dramatic oxygen-dependent antimicrobial response (16, 38). The unchecked production or accumulation of these species can lead to cell damage, mutations, or death. The generation of HO·, the most destructive of the above-mentioned compounds, is in part dependent upon the presence of a transition metal, such as iron or copper, and H₂O₂. Defense against ROIs is provided by antioxidant enzymes (superoxide dismutase [SOD], catalase, and peroxidase), iron sequestration, free-radical-scavenging agents, DNA-binding proteins, and DNA repair enzymes (4, 25, 26, 32, 35, 62). P. aeruginosa possesses an impressive antioxidant armament for defense against ROIs, including two SODs (cofactored by either iron [Fe-SOD] or manganese [Mn-SOD] [19, 20] to disproportionate O₂ to H₂O₂ and O₂), three catalases (KatA, KatB, and KatC) (6, 32), and four alkyl hydroperoxide reductases (AhpA, AhpB, AhpCF, and Ohr) (U. A. Ochsner, D. J. Hassett, and M. L. Vasil, unpublished data).

We have now investigated the roles of individual oxidative stress defense genes by phenotypic assessment of specific mutants and have monitored the responses of these genes to oxidative stress. It appears that redundancy of oxidative stress defense systems allows P. aeruginosa to optimally cope with ROIs generated by its own vigorous aerobic metabolism and to respond rapidly to exogenous ROIs. Some of the genes involved in oxidative stress defense, including katA (encoding the major catalase in P. aeruginosa [32]), and ahpA (encoding a ferredoxin-dependent alkyl hydroperoxide reductase [U. Ochsner and D. Hassett, unpublished data]), are expressed at high levels during aerobic growth. Their activities are maintained at such high levels that even significant oxidative stress causes only a twofold increase in expression, suggesting that high KatA and AhpA activities are critical for detoxification of ROIs produced endogenously during normal aerobic growth. On the other hand, several oxidative stress defense genes, including katB-ankB (24), ahpB, and ahpC-ahpF, are dramatically induced by ROI-generating agents, suggesting a specific and tightly regulated response.

Key regulators modulating the oxidative stress response in bacteria are SoxR and OxyR, both of which are activated at the posttranslational level. O₂ activates SoxR through oxidation of its [2Fe-2S] cluster (11, 13), and oxidized SoxR induces the expression of the second transcription factor SoxS, which directly activates transcription of several genes, including sodA in Escherichia coli (27, 30, 60). H₂O₂ induces at least 30 genes in E. coli, and the response of a subset of these genes depends on OxyR, a 34-kDa LysR-type transcriptional activator (8, 53). E. coli oxyR mutants are hypersensitive to H₂O₂ and have increased rates of spontaneous mutagenesis during aerobic growth (52). OxyR-regulated genes in E. coli include katG (encoding hydroperoxidase I), gorA (encoding glutathione reductase), ahpCF (encoding alkyl hydroperoxide reductase) (52), and fur (for ferric uptake regulator) (66). Furthermore, E. coli OxyR also controls the formation of a small RNA, designated oxyS, that can act as a positive or negative regulator in response to oxidative stress. The abundant and relatively stable 109-nucleotide oxyS RNA is transcribed immediately upstream and divergently of oxyR in E. coli. Several oxyS-regulated genes were identified in E. coli, including dps (DNA-binding protein of stationary phase) and rpoS (^S) (1, 14). Recent biochemical studies have shed light on the molecular mechanism of OxyR activation in E. coli. OxyR is redox sensitive and can switch rapidly between oxidized and reduced states, but only the oxidized form of OxyR acts as a transcriptional activator (54). In the presence of H₂O₂, OxyR forms an intramolecular disulfide bond which can be deactivated by enzymatic reduction upon relief of oxidative stress (2, 65). Both the oxidized and the reduced forms of the E. coli OxyR protein have been shown to possess DNA binding activity (55). Oxidized OxyR recognizes a motif comprised of four ATAG elements spaced at 10-bp intervals (56).

A better understanding of the oxidative stress response in P. aeruginosa, a ubiquitous gram-negative opportunist, is of great industrial and clinical importance. In this work, we provide evidence for the existence of an OxyR homolog in P. aeruginosa and characterize three OxyR-regulated genes essential for the optimal defense against oxidative stress. We describe significant differences in the OxyR response between P. aeruginosa and E. coli, including a link to DNA repair since the P. aeruginosa oxyR gene is located in an operon with recG, and OxyR regulation of a novel type of alkyl hydroperoxide reductase (AhpB) not found in E. coli that is very important for resistance to H₂O₂.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, media, and chemicals. All P. aeruginosa and E. coli strains and plasmids used in this study are listed in Table 1. Luria broth (LB) was used for strain maintenance and contained 1.5% agar (Difco) in solid media. M9 minimal medium (45) was used for cultivating P. aeruginosa in the presence of oxidative stress-generating agents. Liquid cultures were grown aerobically at 37°C in shake flasks or, for smaller volumes up to 2 ml, in 14-ml plastic tubes (Fisher Scientific) shaken at 250 rpm. Antibiotics were added as follows: for E. coli, ampicillin (Sigma Chemical Co., St. Louis, Mo.) (100 µg ml¹), gentamicin (Abbott Laboratories) (15 µg ml¹), kanamycin (Sigma) (100 µg ml¹), and tetracycline (Sigma) (15 µg ml¹); for P. aeruginosa, carbenicillin (Research Products International) (750 µg ml¹), gentamicin (75 µg ml¹), and tetracycline (150 µg ml¹). Paraquat (methyl viologen), H₂O₂ (30%), cumene hydroperoxide (CHP) (80% stock solution diluted with ethanol), t-butyl hydroperoxide (tBHP) (70% solution), and o-nitrophenyl--D-galactopyranoside (ONPG) were from Sigma Chemical Co.). Bovine liver catalase was from Boehringer Mannheim, and concentrated protein dye (Bradford reagent) was from Bio-Rad. X-Gal (5-bromo-4-chloro-3-indoyl--D-galactopyranoside) was from Research Products International and was used at 40 µg ml¹.

Oxidative stress and UV irradiation sensitivity assays. To test susceptibility of P. aeruginosa strains to oxidative stress agents, 100 µl of cells grown overnight in LB were inoculated into 100 ml of M9 medium, grown to mid-exponential phase (optical density at 600 nm [OD₆₀₀ = 0.5), and split into 2-ml aliquots. Paraquat, H₂O₂, CHP, or tBHP was added at appropriate concentrations, and the subcultures were shaken aerobically. For disk inhibition assays, a culture volume representing 0.2 OD₆₀₀ unit of cells was mixed with 3 ml of 0.7% low-melting-point M9 agarose at 37°C and poured onto M9 agar plates. Sterile filter disks containing 10 µl of either 2% H₂O₂ or 20% CHP were placed in triplicate on the top agar, the plates were incubated overnight at 37°C, and the zones of growth inhibition were recorded. For UV sensitivity assays, 100 µl of cells grown overnight in LB was diluted with 10 ml of M9 medium in an uncovered glass petri dish and shaken at 60 rpm at room temperature. Irradiation was performed with a UV lamp (Fotodyne model 3-6000) placed 5 cm above the cells. Samples were removed at 0, 10, 20, 30, 40, and 60 s of UV irradiation and serially diluted with LB in a microtiter dish. Catalase (1,300 U ml¹) was added for the dilution of oxyR mutant strains. Appropriate dilutions (50 µl) were spotted on LB agar, and the colonies were enumerated after overnight incubation at 37°C in the dark.

General genetic procedures. PCR was performed using Taq polymerase and custom-made primers (Bethesda Research Laboratories, Gaithersburg, Md.) in a Perkin-Elmer Cetus thermal cycler, with 30 cycles of denaturing (1 min, 94°C), annealing (1 min, 54°C), and extending (1 min per kb of DNA, 72°C). The PCR products were purified in low-melting-point agarose gels, routinely cloned into pCRII-2.1 (Invitrogen), and sequenced with Sequenase 2.0 (United States Biochemical) and M13 primers or custom-made 18-mer oligonucleotides. Published procedures were followed for Southern blot analysis, colony hybridization, end labeling of DNA fragments, and other recombinant DNA methods (45), using DNA modifying enzymes from Bethesda Research Laboratories. Standard protocols were used for the isolation of plasmid DNA (23) and chromosomal DNA (9). Plasmids were maintained in E. coli DH5-MCR (Bethesda Research Laboratories) and transformed into P. aeruginosa strains using the magnesium chloride method (42). RNA was isolated by the hot-phenol method and analyzed by RNase protection assays as described in detail elsewhere (3). Radiolabeled riboprobes were generated from cloned DNA fragments (Table 1), using an in vitro runoff transcription system (Promega), and excess probe was hybridized to 20 µg of total RNA.

Construction of isogenic mutant strains. Mutant strains affected in oxyR were constructed as follows. A 1.36-kb PCR product containing the oxyR region was generated with primers oxyR-226 (5'TGTACACCAGGTAGTCGAG) and oxyR-1585 (5'-GTTTCCAGGCCTACCCGAG), cloned into pCRII-2.1, sequenced, excised with EcoRI, and cloned into the EcoRI site of pUC19. A 0.62-kb XhoI-SstII internal fragment of the oxyR gene was removed, and the ends were blunted with Klenow enzyme and ligated to a 1.3-kb FRT-Gm^r-FRT cassette (Gm^r) excised from pPS856 (21) with BamHI and followed by end polishing. The resulting plasmid, pUCoxyR::Gm, was digested with PvuII, yielding a 2.5-kb oxyR::Gm construct which was ligated into the SmaI site of the gene replacement vector pEX100T (47). E. coli SM10 containing pEX100T-oxyR::Gm was used as the donor strain in a biparental mating with P. aeruginosa PAO1. Transconjugants were selected on brain heart infusion agar containing gentamicin (75 µg ml¹) and irgasan (50 µg ml¹) and subsequently plated on LB agar containing gentamicin (75 µg ml¹) and 5% sucrose. Successful double-crossover events leading to the replacement of the oxyR gene with the Gm^r cassette in the putative oxyR::Gm mutant strain were verified by the loss of pEX100T-encoded Cb^r and by PCR across the oxyR gene using the primers oxyR-226 and oxyR-1585. To obtain an unmarked oxyR mutant strain, E. coli SM10 harboring pFLP2 (21) was mated into oxyR::Gm, and oxyR::Gm/pFLP2 was grown overnight in LB to allow excision of the Gm^r cartridge via the adjacent FRT sequences (21). Single colonies on LB-carbenicillin were then checked for the loss of Gm^r. Finally, pFLP2 was cured from oxyR by selection for sucrose resistance, indicating the loss of the pFLP2-borne sacB gene, and the resulting unmarked oxyR mutant was also checked for loss of plasmid-encoded Cb^r.

Construction of lacZ reporter fusions. DNA fragments containing the relevant promoter regions, including the translational start sites, were PCR amplified, cloned into pCRII-2.1, sequenced, and ligated into pPZ30. To achieve an in-frame translational fusion to the promoterless lacZ gene, a PstI site was incorporated in the primer sequence at an appropriate 3'-end position, when necessary. Specifically, plasmid pPZ-oxyR-126 contained a 126-bp EcoRI-PstI fragment harboring the oxyR promoter region plus the first six codons of the oxyR gene. In pPZ-recG-1133 and pPZ-recG-360 the first 32 codons of the recG gene were fused to lacZ, and these plasmids contained increasing upstream sequence as depicted in Fig. 1. Plasmid pPZ-katB-480 contained the katB promoter plus 59 codons of the katB gene on a 480-bp EcoRI-PstI fragment that had been generated by PCR with primers katB-38 (5'CTTGGAACTGCGCCATGCAG) and katB-514 (5'TCCTGCAGCAGCACCGAAC [the PstI site is underlined]). Construct pPZ-ahpB-377 harbored the first four codons of the ahpB gene and the ahpB promoter on a 377-bp EcoRI-PstI fragment obtained with primers ahpB-1321 (5'GATGGCGCTTCAACTCGAAG) and ahpB-1697 (5'ctgCAGTACGCTCATCGCGAGG [nonmatching nucleotides in lowercase type]). A 270-bp EcoRI-PstI fragment containing the ahpC promoter plus the first three codons of ahpC was PCR amplified with primers ahpC621 (5'GACCATCCTGGTGCTGGTC) and ahpC-890 (5'ctgCAGGGACATCAGTCGTTCCT) and cloned into pPZ30 linearized with EcoRI and PstI, yielding pPZ-ahpC-270. Besides those mentioned above, a minilibrary of several additional genes were also tested for their dependence on OxyR, and they included katA, katC, ahpA, ohr, fur, omlA, sodA, sodB, bfrA, bfrB, phuR, plcH, toxA, pvdS, rpoS, dps and ptxR.

View larger version (47K): [in this window] [in a new window]   FIG. 1.   Characterization of the oxyR-recG operon. (A) Genetic map showing the putative promoter, the overlapping TGA stop codon for oxyR and ATG start codon for recG, the locations of the riboprobes, relevant restriction sites, and the deletions and insertions made in the oxyR and recG mutants. The numbers flanking the maps indicate the coordinates of these loci in the PAO1 genome (Pathogenesis Corp., 12-15-99 release). (B) Expression of oxyR and recG as fusions to lacZ. The portions of the oxyR-recG DNA sequence contained in the lacZ fusion plasmids are indicated, together with the corresponding -galactosidase activities expressed from these constructs. Error bars indicate standard deviations. (C) RNase protection assays. Riboprobes specific for the oxyR promoter (oxyR-rp) and for the oxyR-recG overlapping region (oxyR-recG rp) were used to detect the corresponding transcripts in P. aeruginosa PAO1 total RNA isolated during the exponential growth phase in M9 medium. Paraquat (PQ) was added to final concentrations of 10 and 100 µM 1 h prior to harvest as indicated. Also shown are the digested probes in the absence of any P. aeruginosa RNA as a control. A DNA sequencing reaction was run in parallel and served as a size marker. Numbers indicate nucleotides.

Construction of complementing plasmids. The P. aeruginosa-E. coli multicopy shuttle vectors pUCP19 and pUCP22 (59) were used for the construction of recombinant plasmids containing the oxyR, katB, ahpB, ahpC-ahpF, and recG genes under the control of the plasmid-borne lac promoter that drives constitutive expression in P. aeruginosa. The complete oxyR gene was PCR amplified with primers oxyR-460 (5'GCAGTGTAGGCGTCGAATC) and oxyR-1585 (5'GTTTCCAGGCCTACCCGAG), and the PCR product was cloned into pCRII-2.1 and transferred as a 1.13-kb EcoRI fragment into pUCP19, resulting in pUCP-oxyR. Plasmid pUCP-katB was constructed similarly using a 1.9-kb PCR product obtained with primers katB-38 (see above) and katB-1887 (5'CCAGGATTGATCGCAACCGG). The ahpB gene was amplified by PCR with primers ahpB-1321 and ahpB-2537 (see above) and was directionally cloned as a 1.2-kb HindIII-XbaI fragment from pCRII-2.1 into pUCP22, yielding pUCP-ahpB. The ahpC-ahpF region was located on a 3.5-kb SphI fragment as predicted from the P. aeruginosa genome sequence. Accordingly, chromosomal DNA of P. aeruginosa PAO1 was cut with SphI, and fragments of the size range of 3 to 4 kb were cloned into pUCP19. A pUCP-ahpCF plasmid harboring the ahpC-ahpF genes under lac promoter control was subsequently isolated by colony hybridization using a 270-bp ahpC promoter fragment (see above) as a radiolabeled probe. For the construction of pUCP-recG, it had to be considered that the native recG gene lacked a Shine-Dalgarno sequence due to the overlap of its ATG start codon with the oxyR TGA stop codon (Fig. 1). Therefore, a Shine-Dalgarno motif (underlined, see below) was incorporated 7 bp upstream of the recG ATG. A 2.1-kb PCR product containing recG was obtained with primers recG-1490 (5'aggagAAATAGCATGACCGAGCTGTC) and recG-3616 (5'GCTTCAAGACTGAGACCTACG), cloned into pCRII-2.1, and directionally cloned as a HindIII-XbaI fragment into pUCP22, resulting in pUCP-recG.

Purification of OxyR and DNA mobility shift assays. The oxyR gene was PCR amplified as an NdeI-BamHI fragment using primers (NdeI)-catATGACCCTCACCGAACTGC and (BamHI)-ggatCCTGGACAGCTCGGTCATG and cloned into pCRII-2.1. After verification of its sequence, the oxyR gene was cloned into the NdeI-BamHI sites of pET14b (Novagen), generating an in frame fusion with the vector-encoded His tag sequence. OxyR protein with an amino-terminal His₆ tag was overexpressed from pET-OxyR in the BL21(DE3) T7 expression strain and purified through metal affinity chromatography on Ni-nitrilotriacetic acid (Qiagen). End-labeled DNA fragments (1 to 2 ng) harboring the relevant promoter sequences (Table 1) were incubated for 15 min with increasing amounts (up to 1 µM) of freshly purified His₆-OxyR protein in 20 µl of binding buffer [20 mM bis-Tris borate (pH 7.5), 40 mM KCl, 5 mM MgCl₂, 1 mM EDTA, 1 mM dithiothreitol, 100 µg of bovine serum albumin ml¹, 50 µg of poly(dI-dC) ml¹, 10% glycerol], and 10 µl of the mixture was loaded on a 6% polyacrylamide gel in running buffer (20 mM bis-Tris borate, pH 7.5). After electrophoresis for 3 to 4 h at 250 V, the gel was dried and autoradiographed.

Biochemical procedures. -Galactosidase activities were determined as follows. Bacterial cell extracts from 2-ml cultures were prepared by centrifugation (10,000 × g, 10 min, 4°C), resuspension of the cells in 0.5 ml of 50 mM potassium phosphate buffer (pH 7.0), and sonication for 5 s (Branson Sonifier; output level 5). The insoluble fraction was removed by centrifugation (13,000 × g, 10 min, 4°C), and protein concentrations were estimated by the Bradford assay using bovine serum albumin as a standard (5). -Galactosidase assays were performed using ONPG as the substrate and expressed as international units with a millimolar extinction coefficient for ONPG of 3.1 (37). Catalase activity of normalized soluble protein samples was detected in stained 5% nondenaturing polyacrylamide gels (57).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Characterization of the oxyR-recG operon in P. aeruginosa PAO1. A putative OxyR homolog was identified using the E. coli OxyR amino acid sequence to search the P. aeruginosa genome sequence (www.pseudomonas.com). This P. aeruginosa OxyR homolog is predicted to be a 34-kDa protein with 40% amino acid sequence identity to E. coli OxyR (8). OxyR is a positive regulator of H₂O₂-inducible genes in E. coli and Salmonella enterica serovar Typhimurium and belongs to the LysR family of bacterial regulatory proteins (8). Immediately downstream of oxyR was an open reading frame encoding a 76-kDa protein with 59% amino acid sequence identity to the E. coli RecG protein, which functions as an ATP-dependent DNA helicase involved in replication and repair of DNA (29, 31). The stop codon of P. aeruginosa oxyR overlapped the start codon of recG, suggesting that oxyR and recG are organized in an operon (Fig. 1A). Expression of oxyR and recG was monitored by translational fusions to the lacZ reporter gene, as depicted in Fig. 1B. oxyR::lacZ activity was detected from pPZ-oxyR-126 containing a promoter immediately upstream of the oxyR gene. recG::lacZ expression was absent in pPZ-recG-360, harboring roughly 300 bp of recG upstream sequence, but was detected in pPZ-recG-1133, which contained, in addition, the entire oxyR gene including the oxyR promoter region, indicating that recG is coexpressed from the oxyR promoter in an operon with oxyR. This finding was supported by RNase protection assays using an intergenic 365-nucleotide oxyR-recG riboprobe (Fig. 1A), which was entirely protected (Fig. 1C). A single transcriptional start site for the oxyR-recG operon was found 21 nucleotides upstream of the oxyR translational start, as determined by RNase protection with a 384-nucleotide probe of which 141 nucleotides were protected. The oxyR-recG transcription did not respond to oxidative stress (Fig. 1C) or to other stimuli, such as growth phase or iron concentration (data not shown).

OxyR-dependent activation of katB, ahpB, and ahpC expression. To identify OxyR-regulated genes in P. aeruginosa, we screened a plasmid minilibrary of lacZ reporter fusions of about 20 candidate genes potentially involved in oxidative stress defense, iron uptake and storage, and DNA repair, as listed in Materials and Methods. The reporter activities in the wild type and a oxyR mutant were compared at mid-exponential growth phase in the presence and absence of 100 µM paraquat. Applying a fivefold difference in the expression levels of the candidate genes between wild-type and oxyR mutant cells as the cutoff, we found three fusions (pPZ-katB-480, pPZ-ahpB-377, and pPZ-ahpC-270) that were OxyR dependent (Fig. 2). Expression of katB-lacZ was not detected in unstimulated wild-type organisms and was induced 250-fold upon exposure to paraquat, while no activity was detected in the oxyR mutant (Fig. 2A). The ahpB-lacZ fusion was expressed at very low levels in both wild-type and oxyR bacteria, and a 90-fold induction by paraquat was observed in the wild type but not in the oxyR mutant (Fig. 2B). Expression of ahpC-lacZ was substantial in untreated wild-type cells and increased threefold in the presence of paraquat. In contrast, ahpC-lacZ was not expressed in the oxyR mutant (Fig. 2C). All other tested fusions did not depend on OxyR; e.g., katA-lacZ in plasmid pPZ-katA was expressed at similar levels in either wild-type or oxyR cells, although a roughly twofold response to paraquat was observed (Fig. 2D). This regulation of katA has been reported previously (32), and it appears that it involves a mechanism different from OxyR activation, which is, in part, controlled by quorum sensing and iron levels (18). Among other genes that were expressed independently of OxyR were sodA and sodB (encoding Fe-SOD and Mn-SOD, respectively), dps (encoding DNA-binding protein of stationary phase), ahpA (encoding alkyl hydroperoxide reductase A), bfrA and bfrB (encoding bacterioferritins A and B, respectively), fur, oxyR itself, and all additional genes of the minilibrary listed in Materials and Methods (data not shown).

View larger version (30K): [in this window] [in a new window]   FIG. 2.   OxyR-dependent gene expression in response to oxidative stress. Wild-type and oxyR bacteria containing plasmid-borne katB-lacZ (A), ahpB-lacZ (B), ahpC-lacZ (C), and katA-lacZ (D) fusions were grown in M9 medium to mid-exponential phase. The reporter activities were measured without paraquat treatment (black bars) or after treatment with 100 µM paraquat for 1 h (stippled bars). The -galactosidase activities are presented as international units, and the error bars represent the standard deviations from four independent experiments.

Dose-response effect of OxyR-dependent expression of katB, ahpB, and ahpC. To obtain dose-response curves for OxyR-mediated gene activation, mid-exponential-phase cultures of wild-type bacteria containing katB-lacZ, ahpB-lacZ, and ahpC-lacZ were exposed for a given time to oxidative stress compounds, including paraquat, CHP, tBHP, and H₂O₂, at concentrations ranging from 0.3 µM to 1 mM (Fig. 3). The strongest induction was evoked by paraquat, which caused a half-maximal response at a concentration of 1 µM and resulted in a sustained response to concentrations of up to 300 µM, above which it became lethal. Interestingly, all three fusions were responsive to organic hydroperoxides. Approximately 100 to 300 µM CHP or tBHP was typically required to elicit a significant response, and at higher concentrations (1 mM), the cells were killed. Repeated addition of H₂O₂ caused activation of ahpB-lacZ and ahpC-lacZ at a concentration of 3 µM or higher; in contrast, at least 100 µM was required for activation of katB-lacZ. Generally, H₂O₂ had a less pronounced effect on OxyR-dependent gene activation than any of the other tested compounds, presumably due to rapid detoxification by endogenous catalase. Interestingly, the ahpB-lacZ fusion responded somewhat more strongly to lower concentrations of all of the oxidative stress compounds (e.g., 1 to 10 µM paraquat) compared to katB-lacZ and ahpC-lacZ (Fig. 3).

View larger version (22K): [in this window] [in a new window]   FIG. 3.   Dose-response curves for OxyR-dependent gene expression. Wild-type cells containing plasmid-borne katB-lacZ (A), ahpB-lacZ (B), and ahpC-lacZ (C) fusions were grown in M9 medium to mid-exponential phase and treated with increasing concentrations of paraquat for 1 h (diamonds), of CHP for 30 min (squares), of tBHP for 30 min (triangles), or of H2O2 for 1 h (circles). The -galactosidase activities are shown as a function of the indicated concentrations of the oxidative stress-generating compounds and are the mean values from triplicate assays.

Time course of OxyR-mediated gene activation. The efficiency of the OxyR-mediated response was further evaluated by monitoring the timing of target gene expression. Mid-exponential-phase cultures of wild-type bacteria containing katB-lacZ, ahpB-lacZ, and ahpC-lacZ were treated with a fixed concentration of paraquat (100 µM) or CHP (300 µM), and the -galactosidase reporter activities in samples taken at several time points postinduction were determined (Fig. 4). Paraquat elicited a response within 10 min of exposure, and the katB-lacZ, ahpB-lacZ, and ahpC-lacZ activities increased for at least 1 h. This response was expected because paraquat is not degraded and is capable of continuous redox cycling in viable aerobic bacteria. CHP caused activation within 5 min of exposure, but the response reached a plateau after 30 min and declined somewhat after that. This result was also expected, since CHP can be detoxified (e.g., by Ahp activities) and thus elicits only a transient oxidative stress response.

View larger version (18K): [in this window] [in a new window]   FIG. 4.   Time-response curves for OxyR-dependent gene expression. Wild-type cells containing katB-lacZ (A), ahpB-lacZ (B), and ahpC-lacZ (C) were grown in M9 medium to mid-exponential phase and treated with 100 µM paraquat (triangles) or 300 µM CHP (squares). Samples were removed before treatment and at 2, 5, 10, 20, 30, 45, and 60 min postexposure, and their -galactosidase activities were determined. The values are the means from triplicate experiments.

Genetic analysis of the OxyR-regulated genes katB-ankB, ahpB, and ahpC-ahpF. Genetic maps and the introduced mutations in katB, ahpB, and ahpCF are depicted in Fig. 5. Also shown are the results of RNase protection assays using specific riboprobes to map the individual transcriptional start sites. The katB, ahpB, and ahpCF transcripts were strongly induced in the presence of the oxidative stress-generating agent paraquat, and this response was dependent on OxyR. A single katB transcriptional start site was detected 42 nucleotides upstream of the katB translational initiation site. The katB::Gm mutation has a polar negative effect on ankB, which is in an operon with katB and encodes an ankyrin-like protein required for optimal catalase B activity (24). Transcription of ahpB started 60 nucleotides upstream of its start codon. The ahpB gene encodes a 22-kDa protein harboring a motif typical for antioxidant reductases as determined by the e-motif search (Department of Biochemistry, Stanford University) and contains a candidate membrane-spanning helix suggesting a localization in either the cytoplasmic membrane or periplasm (data not shown). The putative AhpB protein is 60% identical at the amino acid sequence level to the product of the Legionella pneumophila alkyl hydroperoxide reductase (tsaA) gene (GenBank accession number L46863) and 51% identical to a mouse thiol-specific antioxidant (GenBank accession number X82067). This type of peroxidase reduces hydroperoxides with reducing power from thioredoxin (28). Interestingly, the ahpB gene is located immediately upstream of an open reading frame (tdr) encoding a putative thioredoxin reductase (Fig. 5). The tdr gene, however, was not in an operon with ahpB but was expressed from its own promoter and independent of OxyR (data not shown). The ahpC mRNA start site was mapped to 43 nucleotides upstream of the ahpC translational start. Low levels of this transcript were also detectable in unstimulated wild-type cells and in oxyR mutant cells, suggesting that ahpC is expressed at low basal levels in an OxyR-independent way. The ahpC and ahpF coding sequences were spaced apart by a 144-bp intergenic sequence. RNase protection assays using a 484-nucleotide ahpC-ahpF riboprobe over this region were performed to address the question of whether ahpC and ahpF form an operon. Clearly, a fraction of the probe was protected over its entire length, suggesting an organization of ahpC and ahpF in an operon. However, additional protected RNA species of 280 and 215 nucleotides were detected, indicating that some ahpC transcripts may terminate within the ahpC-ahpF intergenic region and that ahpF may be transcribed from a separate promoter and independently of ahpC. In either case, both ahpC and ahpF transcription appeared to be OxyR responsive. The P. aeruginosa ahpC and ahpF genes encode a 21- and a 56-kDa proteins, respectively, with 59 and 66% amino acid sequence identities, respectively, to the E. coli AhpC and AhpF alkyl hydroperoxide reductase subunits. This type of Ahp is widely found in most bacterial species (51) and requires NADH or NADPH for activity (41).

View larger version (73K): [in this window] [in a new window]   FIG. 5.   Genetic maps and transcripts of OxyR-regulated genes. The katB, ahpB, and ahpC-ahpF loci are shown with their coordinates in the PAO1 genome (Pathogenesis Corp., 12-15-99 release), the sites of insertions or deletions in the corresponding mutants, the locations of the riboprobes (rp), and relevant restriction sites. The katB-ankB operon encodes a previously characterized inducible catalase (6) and an ankyrin-like factor required for optimal catalase activity (24). The ahpB gene encodes a thiol-specific peroxidase and is located upstream of a thioredoxin reductase (tdr). The ahpC-ahpF operon encodes the two subunits of the classic alkyl hydroperoxide reductase. The RNase protection assays were done with total RNA isolated from PAO1 wild-type or oxyR mutant cells in the absence or presence of paraquat (PQ) as indicated. A probe specific for the constitutively expressed omlA gene (39) was used as a control. Also loaded were diluted probes (rp) and the digested probes (rp*) as controls. Arrows point to the relevant protected riboprobe bands, and their approximate sizes (in nucleotides) are given.

Characterization of the katB, ahpB, and ahpC promoters. The mapping of the transcriptional start sites for katB-ankB, ahpB, ahpC-ahpF, and oxyR allowed the localization of the corresponding 10 and 35 elements (Fig. 6). Putative OxyR-binding sites were identified upstream of the katB, ahpB, and ahpCF promoters. Four ATAG elements spaced at 10-bp intervals comprise the binding sites for oxidized E. coli OxyR (56), and such elements were found in proper spacing and distance within the katB, ahpB, and ahpCF promoters. The number of bases matching the OxyR consensus binding sequence were 9 of 16 (katB), 13 of 16 (ahpB), and 12 of 16 (ahpCF), and in all cases, the OxyR binding motif was located exactly adjacent to the 35 promoter elements (Fig. 6). Purified His₆-tagged OxyR protein at a concentration of at least 100 µM caused a mobility shift of DNA fragments containing these target promoters, indicating direct binding of OxyR. A DNA fragment harboring the oxyR promoter was not shifted by OxyR (Fig. 6), which is in agreement with the finding that oxyR expression did not respond to oxidative stress.

View larger version (34K): [in this window] [in a new window]   FIG. 6.   Binding of OxyR to the katB, ahpB, and ahpC promoters. The alignment of the OxyR-regulated promoters indicates four putative OxyR-binding tetranucleotide sequences (underlined), the residues matching the consensus sequence derived from E. coli OxyR-regulated promoters (asterisks), and the 35 promoter elements in proper distance of the mapped transcriptional start sites. The gel mobility shift assays of radiolabeled DNA fragments containing the corresponding promoter regions were performed with purified His6-tagged OxyR protein at the given concentrations.

Phenotypic comparison of mutants affected in oxyR, katB, ahpB, ahpCF, and recG. The susceptibility of a oxyR::Gm mutant to oxidative stress compounds was compared to the phenotypes of mutants affected in single OxyR-regulated genes in order to dissect their specific roles in the oxidative stress response. Wild-type, oxyR::Gm, recG::Gm, katB::Gm, ahpB::Gm, and ahpCF::Gm bacteria were tested for their susceptibilities to H₂O₂ and CHP using standardized disk inhibition assays (Table 2). The oxyR::Gm mutant containing the control plasmid pUCP19 was dramatically susceptible to both compounds. Plasmid pUCP-oxyR fully complemented this phenotype, while plasmid-borne copies of the recG gene, which is located downstream and in an operon with oxyR (see above), resulted in minimal complementation. These findings strongly suggest that the oxyR phenotype was caused by the lack of OxyR-mediated oxidative stress defense. Still, recG appeared to be essential for optimal resistance to H₂O₂ and CHP, and the recG mutant could be complemented by the recG gene in trans. Multiple copies of single OxyR-regulated genes expressed from the constitutive lac promoter on pUCP resulted in only marginal complementation. The katB::Gm mutant showed increased susceptibility to H₂O₂, and this phenotype could not be complemented by providing only katB in trans, presumably due to the polar negative effect on ankB (24). The ahpB::Gm mutant was hypersusceptible to H₂O₂ but not to CHP and was successfully complemented by ahpB in trans. The ahpCF::Gm mutant exhibited a somewhat intriguing phenotype. While ahpCF::Gm mutant cells were hypersusceptible to CHP, they were more resistant to H₂O₂ than the wild type, and plasmid pUCP-ahpCF reversed that trend. Elevated KatB catalase levels were measured in the ahpCF::Gm mutant, and the KatB catalase activity was detectable even in the absence of paraquat as an inducer (Fig. 7). In agreement with that observation was the finding that a katB-lacZ fusion was expressed at severalfold higher levels in a ahpCF::Gm background than in wild-type cells (data not shown). This compensatory mechanism between ahpCF and katB expression indicates that the absence of AhpCF leads to internal oxidative stress. During experiments to measure catalase levels in various catalase-deficient and ahp mutant strains using activity staining, we unexpectedly observed extra bands that we suspected might reflect the ability of some alkyl hydroperoxidases to also use H₂O₂ as a substrate (Fig. 7). Wild-type organisms produced KatA and KatB activities, and two additional smaller activity bands migrated between KatA and KatB. We determined that the lower band was an electrophoretic variant of KatA, since this band was absent in a katA mutant. The upper band most likely represented AhpB, since it was absent in the oxyR and ahpB mutants. Both middle bands were retained in a katB mutant. Interestingly, the katA mutant possessed a catalase activity band that migrated with KatA. We determined that this band is AhpA, because it was absent in an ahpA mutant (data not shown). Furthermore, preliminary catalase assays indicated that both AhpB and AhpA possessed some catalase activity (data not shown), while it remained uncertain whether AhpCF had such activity.

View larger version (80K): [in this window] [in a new window]   FIG. 7.   Catalase activity gel of soluble cell extracts. Bacterial cultures of PAO1 wild-type and oxyR, katA, katB, and ahpB mutant cells were grown under aerobic conditions in LB to mid-exponential phase and then exposed to 350 µM paraquat (PQ) for 1 h. PAO1 and the ahpCF mutant were grown to stationary phase (16 h) as indicated. Normalized amounts (15 µg) of the soluble protein fractions were separated on a nondenaturing gel and stained for catalase activity. The arrows indicate the positions of KatA, KatB, AhpA, and AhpB activities.

Role of the oxyR-recG operon in DNA repair. Since P. aeruginosa oxyR was located in an operon with the recG gene, encoding a putative DNA helicase, a possible function of the oxyR-recG locus in DNA repair was investigated. Wild-type bacteria, an unmarked oxyR mutant harboring an in-frame deletion of oxyR with an unlikely polar effect on recG, a recG::Gm mutant, and a recC::Tc mutant were compared for their sensitivity to UV irradiation-induced DNA damage (Fig. 8). Wild-type cells and the oxyR mutant showed similar killing patterns, characterized by roughly 3 to 4 log units of killing over 60 s of UV irradiation. The recG::Gm mutant was hypersensitive to UV and was killed by more than 5 log units within 20 s of irradiation. The UV sensitivity of recG::Gm was more dramatic than that of the recC::Tc mutant. Expression of the recG gene in trans partially restored UV tolerance in the recG::Gm mutant. A recG::Gm recC::Tc double mutant was slightly more sensitive than the recG::Gm single mutant, suggesting an additive effect. Clearly, our data show that recG plays an important role in DNA damage repair.

View larger version (21K): [in this window] [in a new window]   FIG. 8.   Kill curves upon exposure to UV irradiation. Wild-type PAO1 (), an unmarked nonpolar oxyR mutant (), a recG::Gm mutant (), a recG::Gm mutant complemented with pUCP-recG (), a recC::Tc mutant (), and a recG::Gm recC::Tc double mutant () were grown overnight in LB. The cells were UV irradiated while shaking, and aliquots were removed at 10-s intervals. Serial dilutions were plated on L agar to determine the viable cell counts. The UV killing assays were performed five times with independent cultures, and the outcome of one representative experiment is shown.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

H₂O₂ is a powerful antimicrobial agent commonly used in health care as a topical anti-infective, as well as in industry for the treatment of problematic bacterial biofilms. Also, human phagocytes produce H₂O₂ as a natural weapon during the respiratory burst to combat microbial infections (38). Thus, it is not surprising that microbes have evolved several strategies to cope with oxidative stress. In P. aeruginosa, the primary defense against H₂O₂ involves a constitutively expressed catalase, KatA (12, 17, 32), but little is known about a specific response of P. aeruginosa to oxidative stress agents. While the existence of a second, H₂O₂- or paraquat-inducible, catalase (KatB) has been described (6), the regulatory mechanism governing this response remained unknown. In this report, we characterize a regulatory gene, oxyR, and present evidence that the OxyR protein is involved in transcriptional activation of at least three genes encoding antioxidants. The OxyR-mediated stress response has been well studied in E. coli by both genetic and biochemical means (52-56). Also, a possible role of OxyR to combat host defense systems has been investigated in numerous pathogenic bacteria, including Enterococcus faecalis (44), Haemophilus influenzae (33), and Mycobacterium tuberculosis (10, 50). In M. tuberculosis, the OxyR-regulated genes katG and ahpC play crucial roles in isoniazid resistance, since isoniazid requires activation by KatG to exert lethal effects, while AhpC could play a detoxifying role (63, 64). Interestingly, the oxyR gene in members of the M. tuberculosis complex is located next to ahpC; however, oxyR is nonfunctional, due to numerous deletions and point mutations (10). M. tuberculosis katG mutant strains were found to have acquired a compensatory mutation resulting in an upregulation of AhpC, and it has been shown that this protein confers resistance to isoniazid and protection against H₂O₂, even in the absence of adequate catalase and peroxidase activities (48). We found a compensatory cross-regulation of OxyR-dependent katB-ankB and ahpC-ahpF expression in P. aeruginosa. A ahpCF::Gm mutant strain was more resistant to H₂O₂, and this phenotype correlated with higher levels of KatB observed in a catalase activity gel. Such an increased resistance due to elevated expression from all peroxide regulon promoters has been reported for a Bacillus subtilis ahpC mutant (7). Similarly, the lack of AhpC-AhpF peroxidase expression in E. coli has been shown to lead to constitutive OxyR activation due to the accumulation of endogenous oxidants (43). Somewhat surprising was the hypersusceptibility to H₂O₂ of strains harboring the ahpCF genes on a multicopy plasmid. However, a similar phenomenon has been observed upon overexpression of ahpCF in S. enterica serovar Typhimurium (51). The reasons for this are unclear, but possible explanations are that multiple copies of the ahpCF promoter titrate out OxyR or that increased AhpC-AhpF hydroperoxide reductase activity could somehow interfere with the proper sensing of oxidative stress, e.g., by maintaining the oxidized state of the OxyR protein.

In the course of this study, we learned that P. aeruginosa possesses multiple lines of OxyR-dependent, inducible oxidative stress defense systems with potentially overlapping functions. The expression of the three identified OxyR targets, katB-ankB, ahpB, and ahpC-ahpF, responded to any of the exogenously added oxidative stress compounds, including H₂O₂, paraquat, and organic hydroperoxides, suggesting that all of these agents or products derived from their action cause oxidation of the OxyR protein, which then indiscriminately activates the target promoters. However, one of the OxyR target genes, ahpB, responded to significantly lower concentrations of oxidative stress agents than the other targets. The reason for this dose-response shift is unknown, but interestingly, the putative OxyR-binding site in the ahpB promoter had a higher identity to the consensus "OxyR box" than the OxyR boxes in the katB and ahpCF promoters. The strength of an OxyR-binding site could possibly determine the affinity of OxyR to a target promoter and could allow the sequential activation of antioxidant genes with regard to the extent of oxidative stress encountered. Furthermore, we found that not only the KatB catalase, but also AhpB, which belongs to the alkyl hydroperoxide reductase family, possesses catalase activity. In fact, a ahpB::Gm mutant strain exhibited a more pronounced hypersusceptibility to H₂O₂ than to organic hydroperoxides. These findings suggest overlapping functions of KatB, AhpB, and AhpC-AhpF in the detoxification processes. Clearly, detailed biochemical studies on purified KatB, AhpB, and AhpC-AhpF are needed to investigate their potential broad substrate specificities. Multiple enzymatic activities have been demonstrated for mycobacterial KatG, which can act both as a catalase-peroxidase (34) and as a peroxynitritase (58). The observed redundancy of overlapping oxidative stress defense systems in P. aeruginosa may also be explained, in part, by the localization of the antioxidant enzymes in different cellular compartments. While KatA is found in the cytoplasm and in the extracellular milieu (see the accompanying paper by Hassett et al. [15]), KatB is found in the cytoplasm, cytoplasmic membrane, and periplasm (24). AhpB possesses a single cytoplasmic membrane-spanning domain, suggesting a function in the protection of membrane-bound respiratory chain components from H₂O₂.

We present evidence that expression of katB-ankB, ahpB, and ahpC-ahpF depends on OxyR. Putative OxyR boxes were found at the proper location within the ahpB and ahpC promoters, and binding of OxyR to these target promoters was demonstrated, indicating a direct activation of these genes by OxyR.

The phenotypes of a oxyR mutant included a dramatic susceptibility to oxidative stress agents and a low plating efficiency (see also the accompanying paper by Hassett et al. [15]). The oxidative stress susceptibility of P. aeruginosa was significantly increased in a low-iron environment compared to iron-rich conditions (data not shown). Although the presence of iron is known to trigger the formation of HO·, which has deleterious effects on the cells, iron is required for the function of the heme-containing antioxidant enzymes. The hypersusceptibility to oxidative stress in low-iron media was even more drastic in a oxyR mutant, suggesting a potential role of OxyR in iron metabolism. Moreover, none of the OxyR-regulated factors characterized in this study was capable of fully complementing the oxyR mutant phenotype, suggesting the existence of additional members of the OxyR regulon. Some genes, including fur and dps of E. coli, are expressed in both OxyR-dependent and OxyR-independent ways, and the situation in P. aeruginosa may be similar. In our screening of a minilibrary of translational fusions to the lacZ gene, we pulled out those genes that were expressed at at least a fivefold higher level in wild-type compared to oxyR mutant cells under oxidative stress conditions. As a consequence, we did not pick up those genes that are expressed OxyR independently but can be further upregulated by OxyR. Also, the possibility of indirect OxyR regulation exists and could involve the small RNA oxyS, which has been shown to regulate several genes in E. coli (1). However, we have not found an oxyS-like gene in a search of the P. aeruginosa genome (www.pseudomonas.com). A future goal is the isolation of other OxyR-regulated genes, through an in vitro cycle selection procedure that has been successful in the past to identify Fur-regulated genes of P. aeruginosa (40).

Antioxidant enzymes represent the first line of defense in the battle against oxidative stress. A second strategy to survive these harsh conditions is to maintain an efficient DNA repair system. Interestingly, the P. aeruginosa oxyR gene was found in an operon with recG, encoding a homolog of the E. coli RecG DNA helicase, which is an ATP-dependent DNA recombinase implicated in DNA replication, recombination, and repair (29, 31). To our knowledge, P. aeruginosa is the first microorganism for which such a genetic link between an oxidative stress gene and a DNA repair gene has been identified. In some other organisms, including mycobacteria, the oxyR gene is located in a cluster with genes encoding antioxidant enzymes, but we did not find any alkyl hydroperoxide reductases or catalases encoded near P. aeruginosa oxyR (data not shown). The precise role of RecG is somewhat elusive, but it has been postulated that the DNA binding and unwinding activities of RecG are involved in promoting branch migration by catalyzing the formation of four-strand Holliday junctions from three-strand junctions (36). Clearly, we demonstrated that a recG::Gm mutant of P. aeruginosa was hypersensitive to UV irradiation-induced DNA damage, indicating that recG is essential for optimal DNA repair. Also, the P. aeruginosa recG::Gm mutant was hypersusceptible to oxidative stress agents, thus directly demonstrating the DNA-damaging effects of ROIs. Taken together, the two coordinately expressed factors encoded by the oxyR-recG operon play a crucial role the survival in response to environmental challenges.