Regulation of Organic Hydroperoxide Stress Response by Two OhrR Homologs in Pseudomonas aeruginosa

Pseudomonas aeruginosa ohrR and ospR are gene homologs encoding oxidant sensing transcription regulators. OspR is known to regulate gpx, encoding a glutathione peroxidase, while OhrR regulates the expression of ohr that encodes an organic peroxide specific peroxiredoxin. Here, we show that ospR mediated gpx expression, like ohrR and ohr, specifically responds to organic hydroperoxides as compared to hydrogen peroxide and superoxide anion. Furthermore, the regulation of these two systems is interconnected. OspR is able to functionally complement an ohrR mutant, i.e. it regulates ohr in an oxidant dependent manner. In an ohrR mutant, in which ohr is derepressed, the induction of gpx expression by organic hydroperoxide is reduced. Likewise, in an ospR mutant, where gpx expression is constitutively high, oxidant dependent induction of ohr expression is reduced. Moreover, in vitro binding assays show that OspR binds the ohr promoter, while OhrR binds the gpx promoter, albeit with lower affinity. The binding of OhrR to the gpx promoter may not be physiologically relevant; however, OspR is shown to mediate oxidant-inducible expression at both promoters. Interestingly, the mechanism of OspR-mediated, oxidant-dependent induction at the two promoters appears to be distinct. OspR required two conserved cysteines (C24 and C134) for oxidant-dependent induction of the gpx promoter, while only C24 is essential at the ohr promoter. Overall, this study illustrates possible connection between two regulatory switches in response to oxidative stress.

in triplicate for each sample. At least 3 biological replicates were performed in all experiments. Ct value from each reaction was obtained. Ct value is the cycle number intersecting with the threshold line. The difference in Ct values of each gene (ΔCt) between the 16s rRNA gene and target gene was expressed as ΔCt = Ct target -Ct 16s rRNA . The difference in ΔCt values (ΔΔCt) between treated sample and control sample was calculated. The point estimation of the expression ratio was calculated using 2 -ΔΔCt .

Determination of oxidant resistance levels
The exponential-phase cells were treated with oxidant at the indicated concentrations for 10 min. Washed cells were 10-fold serially diluted, and 10 μl of each dilution was spotted onto an LB agar plate. The plates were incubated at 37°C overnight before observation. At least 3 biological replicates were performed.

Construction of the ospR mutant strain
The unmarked deletion of ospR in PAO1 was constructed using a Cre-lox system [15]. A 1.6-kp fragment containing ospR was amplified with the primers BT189 (5'-CTCAACAAG GAAGACAAG-3') and BT1446 (5'-GGACTGTCTGATGGGACTGC-3') and was cloned into pKNOCK-tet at the SmaI sites, yielding pKNOCKospR. An EcoRI and EcoICRI fragment containing the Gm resistance cassette flanked with lox sites from pCM357 [15] was gap-filled using Klenow fragment polymerase. This fragment was cloned into pKNOCKospR at bluntend PvuII-PvuII sites, yielding pKNOCKΔospR::Gm. The PvuII digestion deleted 327 bp of the coding region of ospR. The pKNOCKΔospR::Gm plasmid was transformed into PAO1. To select for double cross-over events, gentamicin-resistant and carbenicillin-sensitive colonies were chosen. The ospR mutant with the antibiotic cassette replacement was then transformed with the pCM157 [15] vector containing the Cre-encoding gene. Cre, a site-specific recombinase, recognizes lox sites, and recombination between these sites deletes the DNA between the two sites. Transformants with the gentamicin cassette deletion or a gentamicin-sensitive phenotype contained the unmarked deletion of ospR. pCM157 was removed by growing cells under a nonselective condition. The deletion of ospR was confirmed by PCR and Southern blot analysis.

Complementation plasmids
A 517-bp DNA fragment containing full-length ospR and a ribosome binding site was generated by PCR using the primers BT191 (5'-GATCGAGGCGCTGCTTGA-3') and BT192 (5'-GACGACGGCTAGCCTCCTA-3'). This fragment was first cloned into a broad-host-range vector, pBBR1MCS-5 [16], at the SmaI site to yield the plasmid pospR. The direction of fusion was checked by PCR to ensure that the gene was fused to the Escherichia coli lacUV5 promoter present in the pBBR1MCS-5 vector. The insertions were checked by sequencing. Then, pospR was introduced into PAO1 strains by electroporation and selected for gentamicin resistance colonies.

Primer extension
The primer used to detect the start site of gpx mRNAs was BT214 (5 0 -gtcttctgttcgcc ctt -3 0 ), which is complementary to nucleotide sequences between +40 to +56, relative to the start codon of Gpx. Primer extension was performed as described previously [1]. A sequence ladder was generated using fmol DNA cycle sequencing system (Promega). The primer for sequencing was the same as used in primer extension reaction, and the template was a DNA fragment containing the promoter region.

Purification of OspR
Primers BT192 and BT430 (5 0 GCTTCCATGGGCACCCGGGGA 3 0 , NcoI site underlined) were designed and used in a PCR reaction with pOspR, pOspR-C24S, or pOspR-C134S as template. The purified ospR coding region digested with NcoI was ligated into the pETBlue-2 vector (Novagen) digested with NcoI and HincII to form the recombinant plasmid pET-OspR. The ospR gene and the promoter region were sequenced to ensure accuracy. This vector places the ospR gene under the control of the T7lac promoter and allows for the expression of ospR only in an Escherichia coli strain containing T7 RNA polymerase (BL21(DE3)). The pETBlue-2 vector also contains the E. coli lac operator in the opposite direction, allowing for blue/white screening for cloning. BL21(DE3) cells containing pET-OspR were grown in 200 mL Luria-Bertani broth containing 100 μg/mL ampicillin at 37°C on a rotary shaker until the cell density reached an OD 600 of approximately 0.6-0.8. Then, IPTG (0.2 mM) was added to the culture, and growth was continued for 2 h. Cells were harvested by centrifugation, washed, and resuspended in KEGD buffer (25 mM KPO 4 , 2 mM EDTA, 5% glycerol, and 1 mM DTT) at pH 7.0. Cells were the disrupted using a French press. The lysates were clarified by centrifugation and loaded onto a DEAE column that had been equilibrated with the same resuspension buffer. The flow-through fraction was dialyzed in the KEGD buffer, pH 8.0, and then loaded onto a heparin sepharose column pre-equilibrated with KEGD buffer, pH 8.0. Non-absorbed materials were removed by passage of resuspension buffer through the column. OspR was gradienteluted with KEGD buffer, pH 8.0, containing 50 and 500 mM KPO 4 . OspR was eluted at the appropriated salt concentration. After dialysis in KEGD buffer, pH 8.0, protein fractions were pooled, concentrated, and stored in KEGD buffer containing 20% glycerol. The purity of OspR was estimated to be at approximately 90% on a Coomassie blue stained SDS-polyacrylamide gel.

DNaseI footprinting assay
The gpx promoter fragment was single end-labeled [γ-32 P]ATP using T4 polynucleotidyl kinase. The fragment was 228 bp long and was produced by PCR using 32 P-labeled BT 214 and BT462 (5 0 CACCGGGTTGGGATCCTG 3 0 ) primers. The ohr and ohrR promoter fragments were prepared using 32 P-labeled BT 484 and BT468 primers and 32 P-labeled BT 467 and BT483 [1] primers, respectively. The DNA binding reaction and DNaseI footprinting assay were conducted as previously described [1]. A sequence ladder was generated using fmol DNA cycle sequencing system (Promega). The template for the sequencing ladder was the corresponding promoter fragment itself, and the primer was the same 32 P-labeled oligo used in the footprinting experiment.

Gel shift assay
Promoter fragments were prepared and the DNA binding reaction was conducted as described above. Samples were then analyzed on a 5% polyacrylamide gel in 1x Tris borate-EDTA buffer containing 2.5% glycerol. The gel was dried and analyzed by autoradiography. The specific binding of OpsR to the gpx promoter fragment was proven by the inability of salmon sperm DNA at high concentration to compete with the low amount of gpx promoter fragment (data not shown).

Site-directed mutagenesis of C24 and C134 of OspR
Site-directed mutagenesis was achieved via the PCR mutagenesis method described previously [1]. Primers containing mutational sites at the C24 and C134 were designed. The sequences of the primers used were as follows: BT396, 5 0 -ACCAGCTGTCTTTCAAGCTGT-3 0 and BT397, 5 0 -ACAGCTTGAAAGACAGCTGGT-3 0 for mutation of cysteine 24 into serine, and BT398, 5 0 -GCTGATCTCCAGCACCGGTTT-3 0 and BT399, 5 0 AAACCGGTGCTGGAGATCAGC 3 0 for mutation of cysteine 124 into serine. The changed bases are underlined, and each pair was complementary to each other. The plasmids were sequenced and designated pospR-C24S and pospR-C134S.

Promoter activity analysis
The gpx promoter-lacZ fusion was cloned into a broad host range expression vector. The 583-bp gpx promoter fragment was amplified using primers BT214 (5'-GTCTTCTGTTCGC CCTT-3') and BT215 (5'-ATCGCGAGGTAGAGCCG-3') and cloned into the pDrive cloning vector (Qiagen). The pDrive vector with the gpx promoter inserted in the direction opposite to the lacZ cassette was chosen. The gpx promoter was excised from pDrive with BamHI and EcoICRI and directionally cloned upstream of the lacZ cassette of plasmid pUC18SfilacZ [1] digested with SmaI and BamHI. The HindIII fragment from pUC18Sfi Pgpx-lacZ containing gpx-lacZ fragment was then cloned into the broad host range expression vector p027Ery [1] to produce p027Pgpx-lacZ. The promoter fragment was verified by sequencing. The plasmid was then introduced into P. aeruginosa strains by electroporation.

β-galactosidase assay
Logarithmic phase cells were induced with oxidants for 30 min. After sonication, cell lysates were then used in the β-galactosidase assay [12]. At least three independent experiments were performed.

Statistical analysis
Statistical analysis was performed using paired t-test. A P value of less than 0.05 was considered statistically significant.

Specific induction of gpx and ospR transcription by organic hydroperoxides
The induction of the gpx-ospR (PA2826-PA2825) operon was monitored by quantitative realtime PCR analysis. Logarithmic cells of the wild-type P. aeruginosa strain were exposed to various oxidants for 15 min. Total RNA was extracted for real-time PCR analysis. Treatment of the wild-type strain with 500 μM cumene hydroperoxide (CHP), 500 μM tert-butyl hydroperoxide (tBOOH), and 100 μM linoleic hydroperoxide (LOOH) resulted in 27.6±4.1, 350.5±116.0, and 216.8±55.0-fold increases in the expression of gpx compared with the untreated sample, respectively ( Fig 1A). Similarly, treatment of the wild-type strain with the same concentrations of CHP, tBOOH, and LOOH resulted in 13.3±7.1, 66.8±15.8, and 34.5±3.5-fold increases in the expression of ospR compared with the untreated sample, respectively ( Fig 1B). These results indicate that both gpx and ospR show strong induction in the presence of organic hydroperoxides and lipid hydroperoxide (P < 0.05). In addition to lipid and organic hydroperoxides, we also treated P. aeruginosa with H 2 O 2 (500 μM, 1000 μM and 4000 μM). The relative expression of gpx and ospR in cells treated with 4000 μM H 2 O 2 was 3.2±0.3 and 2.6±0.9-fold higher than the control cells, respectively, while at 500 μM and 1000 μM H 2 O 2 , induction of gpx and ospR expression was not observed. Other reactive oxygen species were also tested, including the superoxide generators menadione (50 μM and 500 μM) and paraquat (50 μM, 100 μM and 500 μM) (Fig 1A and 1B). The results of these experiments showed that the relative expression levels of gpx and ospR in the presence of these superoxide generators in the treated samples were not significantly higher than the untreated samples (Fig 1A and 1B). The katB gene reported to be induced by oxidative stress including H 2 O 2 and superoxide anion generator such as paraquat [17] was used to show that H 2 O 2 and two superoxide anion generators i.e. paraquat and menadione used in this experiment were active. Relative expression of katB gene when induced with 500 μM H 2 O 2 , 1000 μM H 2 O 2 , 4000 μM H 2 O 2 , 500 μM menadione, and 500 μM paraquat was 40.5±4.9, 39.4±6.9, 37.8±16.3, 9.9±1.8, and 84.0±7.0-fold higher than the control cells, respectively, showing that reagents used were active (data not shown). This finding strongly suggests that gpx and ospR respond more specifically to organic/lipid hydroperoxides than H 2 O 2 and superoxide anion.

Role of gpx and ospR in peroxide resistance
Next, we evaluated the function of gpx in P. aeruginosa. An unmarked deletion of gpx was constructed, and its hydroperoxides resistance was tested using cumene hydroperoxide, tertbutyl hydroperoxide, and hydrogen peroxide; however, no significant difference was observed when compared with the wild-type strain (data not shown). We hypothesized that this lack of peroxide sensitivity phenotype might be due to redundant peroxide detoxifications systems in P. aeruginosa. We then further investigated its possible function in P. aeruginosa strains with decreased ability to survive against organic hydroperoxides (ohr mutant) or H 2 O 2 (katA mutant) [1,5,18,19]. The ohr mutant containing the plasmid overexpressing the gpx gene showed increased resistance to tert-butyl hyderoperoxide (tBOOH), an organic hydroperoxide (Fig 2A). The organic hydroperoxide resistance level of the ohr mutant was two orders of magnitude less than that of the wild-type stain when cells were treated with 1 mM tBOOH. The overexpression of gpx partially restored the organic hydroperoxide-sensitive phenotype of the ohr mutant. Likewise, we found that the overexpression of Gpx partially restored the H 2 O 2sensitive phenotype of the katA mutant ( Fig 2B).

Expression of ohrR in the presence and absence of OspR
Both OspR and OhrR belong to the MarR family of transcriptional regulators and have been shown to respond to oxidative stress, especially that caused by organic hydroperoxides [5]. We postulated that the function of OspR and OhrR might be connected. To test this hypothesis, a genetic approach was used. Substitution of OspR for OhrR and OhrR for ospR was performed in an ohrR mutant and an ospR mutant. A plasmid containing OhrR, termed pohrR, was introduced into the ohrR mutant [1] and the ospR mutant. Similarly, a plasmid containing OspR, termed pospR, was introduced into both the ospR mutant and the ohrR mutant. The expression of ohr and gpx was measured by quantitative real-time PCR to assess the function of OhrR and OspR. The level of gpx was predictably high in the ospR mutant ( Fig 3A). When pospR was introduced into the ospR mutant, the expression level of gpx was approximately the same as that of wild-type cells. However, the level of gpx was high in the ospR mutant containing pohrR; this was similar to the pattern observed in an ospR mutant, indicating that OhrR could not function as a transcriptional repressor of the gpx gene ( Fig 3A). As expected, ohr was up regulated in the ohrR mutant and was down regulated in the presence of a plasmid-borne functional copy of ohrR ( Fig 3B) [1]. Surprisingly, complementation of the ohrR mutant with a functional copy of ospR led to the repression of ohr expression (Fig 3B). The expression of ohr in the ospR mutant was slightly up regulated (2.0±0.4 fold), and introduction of either ohrR or ospR into the ospR mutant suppressed the ohr expression ( Fig 3B). These results lead us to speculate that OspR might be able to control the expression of ohr by binding to the ohr promoter.
Alteration of gpx expression in the ohrR mutant and ohr expression in the ospR mutant during oxidative stress In P. aeruginosa and other bacteria, ohr has been shown to play a crucial role in protecting cells from organic hydroperoxides [1,5,13,[20][21][22]. We have previously shown that the ohrR mutant had an increased level of organic hydroperoxide resistance in P. aeruginosa, and ohr expression was highly expressed in both reduced and oxidized conditions in the ohrR mutant [1]. We hypothesized that an ohrR mutant with an increased capacity to detoxify organic hydroperoxides might alter the expression of gpx in response to organic hydroperoxides. Expression analysis showed that when the ohrR mutant was exposed to various concentrations of CHP, the relative expression of gpx was statistically lower than that of the wild-type strain at 0.1 mM, 1 mM, and 2 mM CHP (Fig 3C). These results imply that CHP are readily detoxified by Ohr, and gpx is inducible by CHP at a concentration that will probably exceed the detoxification capacity of Ohr in the ohrR mutant. This is supported by the fact that the ohrR mutant expresses more ohr and is more resistant to organic hydroperoxide than the wild-type [1]. Furthermore, we observed a faster reduction of CHP in the ohrR mutant than the wild-type (supplement data). We also observed a change in the peroxide induction pattern of ohr in the ospR mutant. At 0.05 mM and 0.1 mM CHP, relative expression of ohr in the ospR mutant, where in the expression of gpx is high, was lower than that of the wild-type. At 1 mM and 2 mM CHP, a slight reduction in ohr relative expression was also observed in the ospR mutant, albeit not statistically significant (Fig 3D).

Identification of OspR binding site at gpx promoter region
Although putative OspR binding sites and promoter sequences were previously reported [8], these have not yet been experimentally verified. We first performed a primer extension assay to determine the transcription start site of gpx. RNA samples prepared from uninduced and 200 μM CHP-induced wild-type cultures were used. An 86-bp primer extension product was detected in the CHP-induced RNA sample. This placed the gpx transcription start site 30 bp upstream from the gpx translation initiation codon (Fig 4A). This information allowed us to predict the putative -10 (TACTTT) and -35 (TTGTGTG) boxes of the gpx promoter (Fig 4B), which concur with previously predicted -10 and -35 sequences [8]. To determine the DNA sequence bound by OspR at the gpx promoter region, a DNaseI footprinting experiment was performed using the purified OspR protein. As seen in Figs 4B and 5A, the region of the gpx operator protected by OspR extended from nucleotides +12 to -56, with respect to the gpx transcription start site.
The DNA binding site of the OhrR protein family is proposed to involve an inverted repeat with a core sequence of AATT-N-AATT (N is any nucleotide, and the number of nucleotides is variable) [11,13,20,22,23]. Inspection of the gpx operator region revealed more than one possible OspR binding site within the DNaseI-protected region of the gpx operator. A previously predicted site, GCAATT-N-AATTGC (OpsR box 1), is a perfect inverted repeat [8] and is identical to the Streptomyces coelicolor OhrR binding site (Fig 4B) [22]. We proposed an additional conserved binding sequence, AATTCAA-N-TTGAATT (OspR box 2.1 and 2.2) that also contains the core sequence of AATT-N-AATT (Fig 4B). To gain insight into which inverted repeat is the OspR binding site, we carried out gel shift assay using three different gpx operator fragments. The first fragment (nucleotides -50 to +1) contains all proposed putative OspR binding sites (Figs 4B and 6A). The second fragment (nucleotides -42 to +1) contains OspR boxes 1 and 2.2. The third fragment (nucleotides -31 to +1) contains the entire OspR binding box 1 but only half of box 2 (Fig 4B). The binding affinity of OspR to fragment 1 appeared to be slightly higher than its affinity to fragment 2 (Fig 6A and 6B). While up to 5 μM, OspR only slightly shifted the gpx operator fragment that contains a deletion of one half of box 2 but retained a complete box 1 (Fig 6C). These results showed that box 1 alone was not sufficient for OspR binding, suggesting that box 2 is more likely the OspR binding site at the gpx operator.

Analysis of the OspR operator sequences in the ohr promoter region
Next, we identified the binding site of OspR on the ohr promoter region. A DNaseI footprinting reaction containing OhrR and the ohr promoter fragment was performed as a control reaction. In agreement with previous reports, OhrR binding protected the ohr promoter fragment at nucleotides -39 to +48, with respect to the transcription start site of ohr (Fig 5B) [1]. Binding of OspR to the ohr promoter fragment created a DNaseI-protected region slightly shorter than the binding of OhrR. The protected region of OspR on the ohr operator region extended from nucleotide position -33 to +48 (Fig 5B). However, the binding patterns of the two proteins were significantly different. The OspR binding at the ohr promoter did not yield a hypersensitive site as was observed in the OhrR binding at the ohr promoter.
Surprisingly, OhrR also bound the gpx promoter at the same region that OspR bound ( Fig  5A). The OhrR binding region on the gpx promoter covered nucleotides -1 to -62 compared with nucleotides +12 to -56 bound by OspR on the same gpx promoter fragment (Fig 5A). In addition, the footprint at gpx promoter was observed at a higher concentration of OhrR than OspR (500 nM vs 100 nM, respectively, in this experiment).

Differential requirement of cysteines for oxidant sensing of OspR at the gpx and ohr promoters
The deduced amino acid sequence of OspR exhibited two conserved cysteine residues at position 24 and 134. It has been shown in vitro that C24 is important for oxidative sensing through sulfenic acid formation and then intermolecular disulfide bond formation with C134 [8]. In this study, we aimed to confirm the role of these two cysteines in the peroxide-sensing mechanism in vivo. We constructed serine substitutions of the OspR cysteines (OspRC24S and OspRC134S). These two ospR mutants and the wild-type ospR gene were cloned into a broadhost-range expression vector, pBBR1MCS-5, and transformed into a P. aeruginosa ospRohrR double-mutant strain that contains a gpx-lacZ reporter fusion. As expected, the ospRohrR double-mutant displayed constitutively high expression of β-galactosidase activity, and the plasmid-borne copy of wild-type OspR successfully complemented this phenotype (Fig 7A). When pospRC24S was introduced into the ospRohrR mutant, β-galactosidase activity was observed at a low level, indicating that OspRC24S repressed the expression of gpx. Thus, OspRC24S maintained DNA binding activity (Fig 7A). However, induction by either CHP or LOOH was not apparent (Fig 7A). Like all other members of OhrR, this result strongly suggests that C24 plays a role in the oxidant sensing of P. aeruginosa OspR [1,12,20]. Surprisingly, OspRC134S complementation showed a CHP induction pattern similar to that observed with the OspRC24S complementation (Fig 7A). The OspRC134S mutant protein retained DNA binding activity but did not seem to be able to sense the presence of CHP or LOOH. This finding indicates that both C24 and C134 are important for the oxidant sensing ability of OspR at the gpx promoter.
We further investigated the oxidant sensing mechanism of OspR at the ohr promoter. First, when pohrR, pohrRC19S, and pohrRC121S were introduced into the ospRohrR double-mutant  containing an ohr-lacZ reporter fusion, the same pattern of β-galactosidase activity was observed as previously reported using RT-PCR (Fig 7B) [1]. The double mutant containing pohrR could lower the β-galactosidase activity and retained the ability to be induced by organic hydroperoxide. The introduction of pohrRC19S into the double mutant resulted in a constitutive low level of β-galactosidase activity (Fig 7B). This is expected because C19 is likely to play a role in the oxidant sensing ability of P. aeruginosa OhrR, while the introduction of pohrRC121S in the double mutant showed similar β-galactosidase activity as the double mutant strain containing pohrR (Fig 7B). The pospR, pospRC24S or pospRC134S plasmids were then introduced into a P. aeruginosa ospRohrR mutant strain that contains an ohr-lacZ reporter fusion. In the presence of pospR, β-galactosidase activity produced by ohr-lacZ reporter fusion was repressed, and CHP and LBOOH treatments resulted in the production of elevated levels of β-galactosidase activity (Fig 7B). When pospRC24S was introduced into the double mutant, we observed a low level of β-galactosidase activity produced by ohr-lacZ in both untreated and organic hydroperoxide-treated conditions, emphasizing the role of C24 in redox sensing (Fig 7B). We found that the introduction of pospRC134S into the ospRohrR mutant resulted in the repression of ohr-lacZ expression. Unlike OspRC24S, OspRC134S retained the peroxide-inducible pattern at the ohr promoter ( Fig 7B).

Discussion
Pseudomonas aeruginosa possesses two organic hydroperoxide resistance regulators, OhrR and OspR, that belong to the OhrR super family of transcriptional regulators. In this study, we investigated the possible overlapping roles of these two oxidation-sensing transcriptional regulators. P. aeruginosa OhrR regulates the ohr gene that encodes the peroxiredoxin or thioldependent peroxidase that catalyzes the reduction of hydroperoxides, especially organic hydroperoxides [1,7]. P. aeruginosa OspR regulates many genes, including gpx (glutathione peroxidase gene), which is located upstream of ospR [8]. Glutathione peroxidase reduces hydroperoxides using electrons from glutathione or thioredoxin. Both H 2 O 2 and organic hydroperoxides are generally known to be substrates of glutathione peroxidase [24,25]. Our data suggest that overexpression of P. aeruginosa Gpx confers both H 2 O 2 and organic hydroperoxide resistance in a P. aeruginosa katA and ohr mutant, respectively (Fig 3). The H 2 O 2 detoxification ability of P. aeruginosa Gpx concurs with a previous report when Gpx was overexpressed in the wild type strain [8]. Therefore, it is conceivable that both Ohr and Gpx protect P. aeruginosa from oxidative stress.
Similar to the ohrR/ohr genes from other bacteria [13,22,23,26], P. aeruginosa ohrR/ohr transcripts are highly inducible by organic hydroperoxides [1]. It has been shown that high concentrations of H 2 O 2 upregulate the transcription of ospR [8]. Here, we demonstrated that exposing P. aeruginosa to organic hydroperoxides, including 500 μM cumene hydroperoxide, 500 μM tert-butyl hydroperoxide, and 100 μM linoleic hydroperoxide, led to the induction of gpx and ospR transcripts (Fig 1). Concentrations of linoleic hydroperoxide as low as 10 μM induced the expression of gpx (data not shown), and exposure of P. aeruginosa to 4 mM H 2 O 2 slightly elevated the expression of gpx and ospR (Fig 1). In addition, the overexpression of gpx complemented the organic hydroperoxide-sensitive phenotype of the P. aeruginosa ohr mutant (Fig 2). These results strongly suggest that both OhrR and OspR play a role in protecting P. aeruginosa from organic hydroperoxides.
This hypothesis is further supported by the observation that high expression of OspR in the ohrR mutant or the ohrRospR mutant almost completely suppressed ohr expression (Figs 3 and  7). The binding of OspR to the ohr promoter was shown by DNaseI footprinting analysis ( Fig  5), suggesting that, in addition to OhrR, OspR could be a putative transcriptional regulator of ohr. The expression of ohr in the ospR mutant was slightly higher than that of the wild-type strain; a 2-fold induction was observed (Fig 3B). We reasoned that the overexpression of Ohr protein in the ohrRospR double mutant or the ohrR mutant masked the effect of OspR on ohr expression because high Ohr could rapidly detoxify organic hydroperoxides [13,23]. This phenomenon is also shown in Fig 3. The induction level of gpx in the ohrR mutant where ohr expression is very high was significantly lower than in the wild-type strain where ohr expression is much lower [1]. These data were confirmed by promoter-fusion assays (data not shown). Furthermore, the affinity of OspR to the ohr promoter appeared to be lower than the affinity of OhrR to the same ohr promoter fragment (Fig 5B). On the other hand, the overexpression of OhrR in the ospR mutant or the ohrRospR mutant did not suppress the expression of gpx (Figs 3 and 7). Surprisingly, we observed the binding of OhrR at the gpx promoter, albeit at a lower affinity (Fig 5A).
The DNA binding sequence of OhrR homologs has been shown to be an AT-rich motif [13,20,22,23]. The P. aeruginosa OhrR box, AATTCNNGCAC-N-GTGCNNGAATT, also contains AATT-N-AATT [1]. Both OhrR and OspR bind the ohr and gpx promoters in vitro (Fig 5). Previously, it has been proposed that the GCAATT-N-AATTGC sequence at the gpx promoter is the OspR-binding motif, and it has been shown that AATT-N-AATT sequence of this motif is essential for OspR binding (Fig 4B) [8]. However, in this study, we showed that this GCAATT-N-AATTGC is not sufficient for OspR binding on the gpx promoter (Fig 6). Therefore, we proposed a new OspR binding site, AATTCAA-N-TTGAATT (Fig 4). Half of this new OspR binding site overlaps with the previously proposed binding site [8]. This might explain why the deletion of AATT-N-AATT at the gpx promoter abolished OspR binding [8]. At the gpx promoter, the OspR binding sites were mapped to 5'-cATTCAA-N 15 -TTGAATT-3' (box 2.1) and 5'-AATCAA-N 11 -TTGAATT-3' (box 2.2) (Fig 4B). Capital letters indicate the nucleotides that match the conserved binding site. There is only one mismatch for OspR box 2.1 at the first nucleotide of the conserved motif and one deletion for OspR box 2.2. At the ohr promoter, there are 5 mismatches, 5'-AATaacA-N 5 -aTaAATT-3' (Fig 4C). It has been shown experimentally that OspR also binds the hmgA and PA1897 promoters [8], and putative OspR boxes can be identified on these promoters. The OspR box at the hmgA promoter contains 3 mismatches, 5 0 -AATTCAA-N 10 -cgtAATT-3 0 . The OspR box at the PA1897 promoter contains 3 mismatches (5 0 -AATTtAA-N 12 -TTtcATT-3 0 ). The sequences protected from DNaseI by OhrR or OspR are very similar at both the gpx and ohr promoters. However, the affinity appears to be different. For the gpx promoter, the binding affinity of OspR is higher than OhrR. For the ohr promoter, the binding affinity of OhrR is higher than OspR. The difference in the binding affinity might be biologically relevant and may dictate which protein is the major regulator of each gene.
There are two major subfamilies of OhrR based upon the number of key cysteine residue at the reactive site: one-cys OhrR and two-cys OhrR [2,3]. The one-cys OhrR prototype is Bacillus subtilis OhrR [11,27]. Other examples of the one-cys OhrR are Streptomyces coelicolor OhrR [22], Staphylococcus aureus MgrA [28], and S. aureus SarZ [29]. This one-cys OhrR subfamily uses its conserved N-terminal cysteine to sense peroxide. The oxidized cysteine becomes sulfenic acid as intermediate and then reacts with small molecules thiols such as bacillithiol in B. subtilis to form a mixed-disulfide bond [11,27]. The two-cys OhrR was first identified in Xanthomonas campestris [9]. Similar to the one-cys OhrR, the conserved N-terminal cysteine of the two-cys OhrR acts as oxidant sensor. The conserved N-terminal cysteine is oxidized to sulfenic acid and then forms an intramolecular disulfide bond with the less conserved C-terminal cysteine. Both P. aeruginosa OhrR and OspR belong to the two-cystype OhrR subfamily. Their N-terminal cysteines are essential for oxidant sensing (Fig 7) [1,8]. For X. campestris OhrR, the C-terminal cysteine is essential for the conformational change of OhrR that leads to the inactivation of its DNA binding activity [9,10]. On the other hand, P. aeruginosa OhrR does not require C121 for its peroxide sensing mechanism [1], suggesting that C121 plays a role in preventing the overoxidation of the redox-sensing cysteine [30]. In Chromobacterium violaceum, the C126 mutation of OhrR partly affects its oxidative sensing mechanism [26]. Interestingly, we reported here that the essentiality of the C-terminal cysteine for the oxidant-sensing ability of OspR depends on the promoter it reacts with. OspR requires C134 for full oxidant sensing ability at the gpx promoter (Fig 7). In contrast, the requirement of this C-terminal cysteine for the oxidant sensing at the ohr promoter is much lower (Fig 7). This result provides insight into how a transcriptional regulator may differentially regulate genes according to differences in the target DNA. It has been demonstrated that as little as a single base pair difference in the binding site of the glucocorticoid receptor could lead to differences in protein conformation and its regulatory activity [31]. The importance of C-terminal cysteine of OhrR in the oxidative-sensing mechanism and DNA binding activity remains to be elucidated.
The biological reason that P. aeruginosa maintains two OhrR homologs is not well understood. Gene duplication is quite common in bacteria and other organisms [32,33]. It is believed that there are many benefits to gene duplication and divergence, including facilitation of adaptation and evolution [33,34]. It has been demonstrated that transcription factors could have the same, overlapping or different group of target genes [34]. According to our knowledge, OspR directly regulates additional genes apart from its own regulon, for example, the hmgA genes involved in amino acid metabolism [8]. It is possible that the gene duplication of OhrR homologs leads to the functional diversification of these transcription factors to fine-tune the regulatory network or to benefit cells in response to different environmental insults. Therefore, one hypothesis is that OspR and OhrR work together to control gene expression in a hierarchical manner. OspR might be a generalized oxidative stress regulator that represses both ohr and ohrR expression. Once exposed to oxidative stress, OspR loses DNA binding ability leading to the expression of ohrR. However, ohr is still repressed at this point. In the event that organic peroxide accumulates, OhrR becomes inactive and the more specialized Ohr can detoxify organic peroxides. This hypothesis is partially supported by the fact that OspR also binds the ohrR promoter albeit at low affinity (unpublished data).
Supporting Information S1 Fig. The ability of the ohrR mutant and the wild-type Pseudomonas aeruginosa to reduce cumene hydroperoxide. The FOX Assay was modified and used to quantify the ability of Pseudomonas strains to diminish organic hydroperoxide (Chuchue T. et al. 2006. ohrR-ohr are the primary sensor/regulator and protective genes against organic hydroperoxide stress in Agrobacterium tumefaciens. J.Bacteriol. 188:842-51). The overnight cultures were subcultured to OD 600 = 0.05 in 60 mL LB medium. Cells were grown to mid-log phase (OD 600 = 0.4). Cell pellet was resuspended in 400 μL Phosphate Buffer Saline (PBS) on ice and sonicated. After centrifugation to remove cell debris, supernatant was kept on ice. 600 μg total protein of each cell lysate was incubated with 100 μM CHP in 500 μL reaction volume, at 37°C with shaking. PBS without cell lysate and containing 100% CHP was used as control. The aliquots of 100 μL were taken out at different time points (2, 5, 10 and 15 min) to react with 900 μL FOX reagent (25 mM H 2 SO 4 , 100 μM Fe(NH 4 ) 2 (SO 4 ) 2 , 125 μM Xylenol-Oragnge) for 10 min. The remaining amount of peroxide was measured at 540 nM. Error bars represent standard deviation of three biological replicates. (PDF)