A Novel OxyR Sensor and Regulator of Hydrogen Peroxide Stress with One Cysteine Residue in Deinococcus radiodurans

In bacteria, OxyR is a peroxide sensor and transcription regulator, which can sense the presence of reactive oxygen species and induce antioxidant system. When the cells are exposed to H2O2, OxyR protein is activated via the formation of a disulfide bond between the two conserved cysteine residues (C199 and C208). In Deinococcus radiodurans, a previously unreported special characteristic of DrOxyR (DR0615) is found with only one conserved cysteine. dr0615 gene mutant is hypersensitive to H2O2, but only a little to ionizing radiation. Site-directed mutagenesis and subsequent in vivo functional analyses revealed that the conserved cysteine (C210) is necessary for sensing H2O2, but its mutation did not alter the binding characteristics of OxyR on DNA. Under oxidant stress, DrOxyR is oxidized to sulfenic acid form, which can be reduced by reducing reagents. In addition, quantitative real-time PCR and global transcription profile results showed that OxyR is not only a transcriptional activator (e.g., katE, drb0125), but also a transcriptional repressor (e.g., dps, mntH). Because OxyR regulates Mn and Fe ion transporter genes, Mn/Fe ion ratio is changed in dr0615 mutant, suggesting that the genes involved in Mn/Fe ion homeostasis, and the genes involved in antioxidant mechanism are highly cooperative under extremely oxidant stress. In conclusion, these findings expand the OxyR family, which could be divided into two classes: typical 2-Cys OxyR and 1-Cys OxyR.


Introduction
Reactive oxygen species (ROS), including hydrogen peroxide (H 2 O 2 ), superoxide, and hydroxyl radical, are toxic to cells due to their ability to damage DNA and especially proteins containing iron-sulfur clusters or sulfur atoms [1]. In bacteria, many transcription factors have been found to sense the presence of ROS and induce antioxidant system. OxyR is such a peroxide sensor and transcription regulator. It was originally identified in Salmonella enterica serovar Typhimurium and Escherichia coli [2,3]. In E. coli, OxyR is a positive regulator of dps (a DNA-binding ferritinlike protein), groA (GSH), grxA (glutaredoxin), katG (catalase), ahpCF (alkylhydroperoxide-NADPH oxido-reductase subunits F and C), fur (Fe-homeostasis regulation), and oxyS (a regulatory RNA) [4]. However, OxyR acts as a repressor of catalase expression in Neisseria gonorrhoeae [5].
As a redox-responsive protein of the LysR family, OxyR has conserved regions consisting of a helix-turn-helix motif and a LysR-substrate binding domain. When the cells are exposed to H 2 O 2 , OxyR protein is thought to be activated via the formation of a disulfide bond between the two cysteine residues (C199 and C208) [2,6]. Detailed footprinting studies indicate that oxidized OxyR binds to its target promoters as a tetramer, occupying four adjacent major grooves upstream of the genes to be transcriptionally activated [7]. However, Kim et al. argued that OxyR activation does not involve disulfide bond formation at all, and that only one thiol in OxyR is critical for protecting against H 2 O 2 [8]. Their work disclosed that OxyR is not involved in disulfide bond formation when it was activated by S-nitrosylation, and that mutation of C208 (which was reported to form a disulfide bond with C199) would not result in the cell hypersensitivity to H 2 O 2 , whereas the mutation of C199 did [8].
The gram-positive bacterium Deinococcus radiodurans is well known for its extreme resistance to ionizing radiation [9,10], ultraviolet radiation [11,12], oxidizing agent [13], and desiccation [14]. It has been suggested that protective mechanisms against oxidative damage is also involved in this extreme radiation resistance [13,15]. D. radiodurans possesses a powerful enzymatic antioxidant system, including three catalases, three superoxide dismutases, two Dps, etc. However, the mechanism of its response to oxidant stress has not been well clarified. Here, we demonstrate an OxyR in D. radiodurans, which is different from all reported homologs in containing only one cysteine residue. Based on quantitative real-time PCR (QRT-PCR), we found that DrOxyR is both an activator, and a repressor. The binding of purified Histagged OxyR protein to the upstream region of the respective genes was verified in vitro by DNA band shift assays. Furthermore, we investigated the global trancriptome variation due to disruption of droxyR, and the comparative analysis reveals pathways significantly impacted either directly or indirectly by droxyR.

Identification of oxyR in D. radiodurans
In D. radiodurans genome database (TIGRE), there is a potential oxyR homolog (DR0615, designated as droxyR) [15], which encodes a protein of 317 amino acids. BLASTP analysis showed that DR0615 exhibited 31% identity to E. coli OxyR and 29% identity to N. gonorrhoeae OxyR, respectively [16]. Five conserved residues in its helix-turn-helix region (between amino acids 3 to 62) involved in DNA binding are identical ( Figure 1) [17]. Other functional domains are conserved at its LysR-substrate binding domain (between amino acids 86 to 297), including a hydrophobic core, a tetramerization domain, and a RNA polymerase binding domain [17,18]. Interestingly, DR0615 has a single sensing cysteine residue (C210), compared with other organisms. This difference in the primary structure of oxyR raised the possibility that droxyR need not, indeed can not form an intramolecular disulfide bond, and that DrOxyR activation can be caused by the modification of just one cysteine residue (C 210 ) ( Figure 1).

Phenotypic characterization of MOxyR
To test its role in the antioxidant mechanism of D. radiodurans, a droxyR disruptant strain (MOxyR) was constructed ( Table 1). The coding region of the dr0615 was replaced with a kanamycin resistance cassette under a constitutively expressed D. radiodurans groEL promoter. The primers are used for mutation listed in Table 2.
Disruption of dr0615 did not show a growth defect (data not shown). However, as shown in Figure 2A, the sensitivity of the mutant after 20 mM H 2 O 2 treatment was increased, compared to the wild type strain. After complemention with wild type droxyR gene (plasmid pRADoxyR) ( Table 1), H 2 O 2 resistance of the MOxyR_wtC was significantly increased. In contrast, the MOxyR_sdC strains, which is the oxyR disruption mutant complemented with the droxyR C210A site-directed mutant (plasmid pRADoxyRsdC), still showed sensitivity to H 2 O 2 . In addition, a little difference was observed between the ionizing radiation resistance of wild type strains and that of MOxyR ( Figure 2B). Figure 1. Alignment of OxyR homologs from different organisms. Using CLUSTAL W software aligned amino acid sequences of the Streptomyces coelicolor A3(2), Neisseria gonorrhoeae, E. coli, and D. radiodurans. Identical amino acids are highlighted in black, and conserved residues are highlighted with grey. The DrOxyR helix-turn-helix region has four conserved residues (R4, L32, S33 and R50) [17]. At its LysR-substrate binding domain, D142 and R273 possibly define an activating region on OxyR (contact with RNA polymerase) [18], A233 residue is involved in tetramerization [17], V110, L124, and A233 form a hydrophobic core [58]. Numbering is based on the E. coli OxyR sequence. doi:10.1371/journal.pone.0001602.g001

Differences in catalase activities and ROS levels between the MOxyR and wild type strains
In order to investigate the regulatory role of OxyR on enzymatic antioxidants of D. radiodurans after treatment of H 2 O 2 , we assayed the catalase activity in wild type, MOxyR, MOx-yR_wtC, and MOxyR_sdC stains with H 2 O 2 treatment or not (Table 1). When log-phase cells were treated with 20 mM H 2 O 2 , the wild type showed a 1.5-fold increase in catalase activity, whilst, neither MOxyR nor MOxyR_sdC showed an increase ( Figure 3A). On the other hand, complement strain MOxyR_wtC showed a higher catalase activity than the wild-type under normal conditions. It is well accepted that oxyR expression is autoregulated via negative feedback in E.coli [19], so we presume that the droxyR gene is under the control of the stronger groEL promoter in pRADoxyR, the transcription of which destroys the negative feedback. As a result, catalase production may have been activated  by an abundance of OxyR, which is likely oxidized by H 2 O 2 from normal metabolism. This might be the reason why MOxyRC is more resistant to hydrogen peroxide than the wild type strain.
These results indicate the droxyR gene is responsible for the regulation of catalase activity.
To determine whether the droxyR disruption has effect on the total ROS scavenging ability of the cell, we also measured the intracellular ROS level. Figure 3B shows that ROS level in all the strains increased after the H 2 O 2 treatment, and that MOxyR and MOxyR_sdC accumulated more ROS than wild type and MOxyR_wtC.
Combined with the survival phenotypic data, it could be inferred that the sensitivity of mutant to H 2 O 2 is due to the loss of induction effect of oxyR on antioxidant enzymes (e.g. catalase), and that oxyR acts as a positive regulator of catalase. Moreover, the C210A mutant showed the same phenotype as MOxyR, indicating that C210 is a site key to OxyR gene regulation.

C210 is a sensing cysteine
The site-directed mutagenesis of droxyR revealed that residue C210 plays essential roles in the function of the protein. This finding poses an intriguing question that whether C210 is a sensing cysteine. To verify this hypothesis, experiments were carried out to obtain C 210 -SOH formation by either CHP or air oxidation, followed by the use of an electrophile (NBD-Cl) to trap the C-SOH [20]. As expected, OxyR treated with CHP (Cys-210SO-NBD) exhibited a maximal absorbance at 347 nm ( Figure 4), on the hand, the reducing form Cys-210S-NBD showed its maximal  absorbance at 420 nm. This data identified that C210 is the peroxidatic center of the molecular.
Because sulfenic acid is highly unstable, and reacts further to produce a disulfide, we used nonreducing SDS-PAGE to identity whether C210 is involved in intermolecular disulfide bond formation. As typical OxyR protein, DrOxyR protein did not form intermolecular disulfide linkages after CHP (organic hydroperoxide) or H 2 O 2 (inorganic hydroperoxide) treatment (data not shown).

QRT-PCR analysis disclose two classes of OxyRdependent genes
As in the oxyR knockout of E. coli., disruption of oxyR makes D. radiodurans much more sensitive to H 2 O 2 . As we have known, this sensitivity is attributed to the inhibition of normal transcription of OxyR-dependent genes [16,[21][22][23]. Therefore, eight homolog transcripts which were reported as OxyR-dependent genes in other bacteria [21,[23][24][25] were selected and compared in wild type and MOxyR using QRT-PCR. QRT-PCR was also used to analyze the expression patterns of these genes in wild type and MOxyR after treatment with H 2 O 2 . Of these genes, dr1998 codes a major catalase KatE in D. radiodurans, and it has been shown to play a role in protection of D. radiodurans from oxidative stress and ionizing radiation [13]. dr2263 and drB0092 are two dps genes, with functions of protection against oxidative stress and iron uptake and storage [26,27], but their expression patterns are different under ionizing radiation [28], indicating that their regulator may be different. Furthermore, D. radiodurans accumulates high intracellular manganese and low iron levels compared with radiation-sensitive bacteria and this is regarded as an important contribution to its resistance [29]. Three ion transport genes were selected to test whether they are regulated by OxyR, including dr1219 (feoB, coding ferrous iron transport protein B), drB0125 (coding Iron(III) dicitrate-binding periplasmic protein), and dr1709 (mntH). In addition, DR1982 is an alkylhydroperoxide reductase subunits F, which could transfers electrons from NADH to AhpC. DR0865 is a Fur or PerR homolog, and Fur proteins control iron uptake in many Gram-negative bacteria, while PerR is postulated to be the peroxide regulon repressor [30].
Consistent with catalase activity assay, the katE (dr1998) transcript was repressed approximately 1.74-fold in the MOxyR relative to that of wild type, and induction of katE expression by H 2 O 2 was eliminated in strain MOxyR ( Figure 5A), suggesting that OxyR is a positive regulator of katE. In addition, both iron transporter genes (dr1219 and drB0125) showed the same expression pattern as katE, indicating that expression of these genes were also mediated by OxyR, a finding similar to the OxyR regulon in Haemophilus influenzae [23].
When the wild type cells were exposed to H 2 O 2 , the transcripts of dps (dr2263) and mntH (dr1709) decreased, whereas, transcripts of these genes in the MOxyR cell were identified either under normal growth conditions or after H 2 O 2 challenge ( Figure 5B). These data informed us that oxidized DrOxyR might act as a transcription repressor of dr2263 and dr1709. Since the expression levels of dr2263 and dr1709 are higher in MOxyR than those of in wild type strains under normal condition, we deduce that reduced DrOxyR could also be a repressor of both of them. To verify this hypothesis, we measured the expression patterns of DR1709 and DR2263 in MOxyR_sdC, which contained a reduced OxyR protein due to its C210 mutation. Opposite to their expression patterns in MOxyR, both of them did not show a significant increase under normal condition ( Figure 5C). Although they were activated under H 2 O 2 treatment in MOxyR_sdC, the change fold is less than that of in MOxyR. The data confirmed that reduced OxyR could still negatively regulate these genes expression. DrOxyR action pattern is opposite to its homologs in E. coli [21], Bacteroides fragilis [31], and Shigella flexneri [32]. In D. radiodurans, two types of Mn(II) import systems have been identified, DR1709 belongs to the Nramp family of transporters [29]. Another type of predicted Mn transporter is an ATP-dependent ABC-type transporter (DR2283-DR2284, DR2523) [29]. However, QRT-PCR results showed that the second Mn ion transporter system was not regulated by DrOxyR protein (Data not shown).
Additionally, dr1982 (ahpF) and drB0092 (dps) did not show significant changes at oxidant stress ( Figure 5D). We deduced that OxyR did not regulate both of them, and the transcription patterns of the two dps genes (dr2263 and drB0092) are different. Furthermore, we also measured the expression of dr0865, which is a putative perR homolog. We found that katE was significantly activated after deletion of dr0865, and that the mutant strain exhibited greater resistance to H 2 O 2 than wild type strain (unpublished data). Interestingly, the expression level of perR was repressed in MOxyR, as well as in the wild type strain after H 2 O 2 challenge. Given the expression pattern of perR, the oxidant stress occurred after disruption of oxyR, which was consistent with intracellular ROS accumulation assay results.
Since the Mn(II) transporter gene (dr1709) is induced, and the iron transporter genes (dr1219, drB0125) are repressed in oxyR mutant, we assayed the intracellular Fe ion and Mn ion levels in MOxyR. As expected, compared to that Fe ion levels are three times higher than Mn ion in wild type strain, Fe ion levels are only two times higher than Mn ion in MOxyR (Table 3).

Transcriptome changes in response to disruption of droxyR
The results of intracellular ROS accumulation assay showed that MOxyR accumulates higher levels of ROS than the wild type strain. It is well known that ROS is a signaling molecule, and has an important role in the regulation of a variety of biological processes [33,34]. Therefore, in order to identify other OxyRdependent genes, and to measure the consequences of higher levels of ROS, we carried out a microarray comparison of the wild type and the OxyR mutants grown under normal conditions. Table S1 and Table S2 exhibits that a total of 280 genes showed at least a 2fold change (p,0.05). A higher percentage of genes were repressed (150 genes, Table S1), whereas, 130 genes were induced (Table  S2). Table 4 and Table 5 showed the 36 most highly repressed and induced genes, many with known roles in oxidative stress response, including catalase, oxidoreductase, N-acetyltransferase. Furthermore, functional classification of these genes revealed that signal transduction mechanisms, inorganic ion transport and metabolism, lipid metabolism, energy production and conversion, and amino acid transport and metabolism showed altered expression patterns in the oxyR mutant (Table S3). Table 6 and Table 7 show genes with the same expression pattern in wild type strain after 20 mM H 2 O 2 treatment (our unpublished data) and these in MOxyR. This confirmed that oxidative stress occurs in MOxyR.

Gel mobility shift assays to confirm OxyR-regulated genes
Our global transcriptome analysis results suggested that the expression patterns of many genes were altered as a consequence of oxyR deletion. However, microarray data could not distinguish those genes directly controlled by OxyR form those controlled by indirect mechanisms. To support the reliability of both QRT-PCR and microarray data, we used a DNA mobility shift assay to determine whether purified OxyR protein could bind in vitro to the two potentially positively regulated gene (dr1998 and drB0125) and two potentially negatively regulated genes (dr1709 and dr2263). In addition, microarray data showed that many oxidoreductase genes were repressed, so we cloned the promoter sequence of drB0036 which is induced after ionizing radiation [28] to identify whether oxyR is a regulator of oxidoreductase. dr0207 was used as a negative control, which is up-regulated after ionizing radiation [35]. As shown in Figure 6, both oxidized and reduced forms of the protein could bind these promoters. Given that OxyR was not completely reduced with DTT and the DTT was probably quickly electrophoresed away from the protein in the mobility shift assays [7], we also examined the binding of the C210A mutant protein to these genes and observed a same binding pattern (data not shown). This data indicate that OxyR protein is bound to its recognition sequences even in the absence of oxidative stress, and the binding ability supports the result that reduced OxyR could regulate dr1709 and dr2263 expression.  Discussion D. radiodurans exhibits extreme radiation resistance. In addition to its powerful DNA repair systems which include some novel components [36], free radical scavengers are regarded as important contributors to this resistant mechanism [15,37].
Recently, Daly et al. reported that accumulation of Mn (II) in D. radiodurans facilitates its radiation resistance through high levels of protein protection from oxidation [29,38]. However, despite these efforts, the molecular regulatory mechanism underlying its oxidative resistance remains poorly understood. Therefore, in this work, we demonstrate that DR0615 is an OxyR homologue with  an important role in oxidative stress sensing and regulation mechanism. Unlike many other OxyR homologs, the sequence analyses showed that DrOxyR contains only one cysteine, which is absolutely conserved in other OxyR homologs (Figure 1). Further comparisons of the amino acid sequences of DrOxyR with those from other bacteria indicated that the other activating regions are well conserved in the DrOxyR protein, except the T238 residue which is involved in C199-C208 disulfide bond formation is absent (Figure 1) [18]. This sequence characteristic informed that DrOxyR could not form intramolecular disulfide bond under oxidative stress. However, disulfide bonds are not the only cysteine oxidation product important for redox sensing. As Bacillus subtilis's OhrR protein [39], our NBD chloride assay in vitro showed that the sole cysteine could be oxidized into sulfenic acid and did not generate an intermolecular disulfide bond. The in vivo functional analyses of the cysteine mutant and wild type OxyR showed that the single C210 is sufficient for DrOxyR to act as a sensor of and a regulator responding to oxidant stress. But the protein may react with low molecular weight thiols (e.g. Cysteine) to make a mixed disulfide in vivo, such as that of BsOhrR [40]. This is a major mechanistic difference from the sensing mechanism of the 2-Cys OxyR, for which sulfenic acid is an intermediate in the pathway to intramolecular disulfide bond formation [2]. Nevertheless, this activating pathway was challenged by the report of Stamler's group, whose research showed that sulfenic acid is stable in per monomer [8].
Furthermore, the droxyR could not complement the defect in the E. coli oxyR mutant strain (GS09) ( Figure S1), even when GS09 was complemented with droxyR::T201C, which is put the missing cysteine back in DrOxyR (data not shown). One explanation for the inability of DrOxyR to complement to GS09 is an inability to properly contact or communicate with E. coli RNA polymerase. Another explanation is that these two families (based on the number of cysteine residues present) use different mechanisms to activate downstream process.
Based on QRT-PCR results, eight potential OxyR-regulated genes were classified into three classes due to their different expression pattern in wild type and mutant strains before or after H 2 O 2 stress. Particularly interesting is, excepting as a regulator of katE, DrOxyR also acts as an activator of iron transport genes and as a repressor of manganese transport gene. As seen in recent studies, the high intracellular Mn/Fe ratio in D. radiodurans plays an important role in resistance ability by protecting cells from ROS generation during recovery [15,29,38]. In addition, our findings do not preclude the existence of other regulators of Mn/ Fe transport genes, or katE. Indeed, dr0865 (fur or perR) and dr2519 (mntR or dtxR) also showed the abilities to regulate these genes (our unpublished data), indicating that the oxidative stress response network is much more complex than we initial prediction.
From the microarray data, we found a total of 280 genes (about 9% of genome) that showed at least 2-fold change, suggesting that these genes were regulated by droxyR through either direct or indirect mechanisms. Several genes annotated as N-acetyltransferase (dr0763, dr1057, dr1978, dr2441, and drA0019) were repressed. In Saccharomyces cerevisiae, N-acetyltransferase could reduce intracellular oxidant levels and protect cells from oxidative stress [41]. Furthermore, genes involved in electron transport were also repressed, including dr0343, dr1493, dr1502, dr1505, and dr2618. This may result in a decrease of the production of ROS generated from the electron transport process [1,15]. In addition, an ironsulfur protein (DR1907) was also significantly inhibited to avoid the attack by ROS. Conversely, many oxidoreductases were overexpressed, some of which are involved in regulating the oxidation state and activities of several proteins [34]. Thioredoxin (DRA0164) is such an oxidoreductase whose expression level was highly elevated in MOxyR. It directly regulates the activation of specific signal transduction proteins through hydrogen peroxidesensitive noncovalent interactions [34]. Moreover, gel mobility shift assays showed that DRB0036 (oxidoreductase) is under the control of OxyR, suggesting that OxyR may be directly involved in the oxidoreductase processes of D. radiodurans.
As a signalling molecule, hydrogen peroxide has an important roles in the regulation of a variety of biological processes, such as stimulate cell proliferation [33,34]. In this work, some genes involved in transcriptional regulation, transport and cell proliferation also showed an altered expression pattern. Thus, it was deduced that these genes were not regulated by oxyR, but were affected as a consequence of ROS accumulation. For example, we have shown the expression levels of two genes, minE (dr0751) and minD (dr2383), were changed. The MinE protein, which is known to prevent the division inhibitor from acting at internal division sites, was activated, whereas the minD gene, which is a cell division inhibitor, was repressed. Based on these expression patterns, it could be assumed that the cell was stimulated due to ROS accumulation. Other interesting genes that were down-regulated in MOxyR were some DNA damage response genes, including recA, cinA, ligT, dinB, ddrB, ddrC, ddrH, ddrJ, ddrK, ddrM, and ddrO [14], whereas they were up-regulated in MOxyR after 20 mM H 2 O 2 treatment (data not shown), indicating that they were not regulated by oxyR only. Based on the gene expression patterns, two classes of OxyRdependent genes were identified and DrOxyR can function not only as a transcriptional activator but also as a transcriptional repressor. Our DNA band shift assay showed that either reduced OxyR protein or oxidized protein can bind both classes of genes. As transcriptional activator, reduced OxyR binds to two pairs of adjacent major grooves separated by one helical turn of the DNA duplex and acts to repress its own synthesis. When oxidized, the OxyR tetramer binds four adjacent major grooves upstream of those genes that are transcriptionally activated by OxyR [4,7] (Figure 7). Whereas a repressor, the reduced form of OxyR produces a basal expression level in the absence of exogenous H 2 O 2 and the oxidized form of OxyR significantly repressed the genes' expression level. Moreover, after deletion of OxyR, the expression of these genes was significantly induced in MOxyR (Figure 7). This agrees with published data showing that oxidized OxyR also acts as a repressor [5,42]. Nonetheless, no obvious binding sites, such as those identified in E. coli, were observed in these genes. This is consistent with the results of OxyR-binding sites of genes in other bacteria [5,22].
In summary, our work presents a biochemical mechanism for hydrogen peroxide sensing of D. radiodurans OxyR, which contains only one conserved cysteine. The gene transcription induction by hydrogen peroxide requires only one cysteine that can be reversibly oxidized by peroxides to a sulfenic acid form. Moreover, based on QRT-PCR and globe transcriptome analysis, we provide evidence that DrOxyR functions as not only a positive regulator but also as a negative regulator of different classes of genes. These results show that genes participating in the Mn/Fe homeostatic and antioxidant system are highly cooperative under extreme conditions, and that cooperation contributes to resistance. More research is needed to establish the detailed mechanism of OxyR regulation of these important genes, and whether communication between OxyR and other regulators such as PerR (DR0865) existed and is required for the intricate coordination of oxygen radical detoxification.

Strains and growth conditions
Bacterial strains and plasmids are listed in Table 1. All D. radiodurans strains used in this work were grown at 30uC in TGY (0.5%Tryptone, 0.3% yeast extract, 0.1% glucose) broth or on TGY plates supplemented with 1.5% Bacto-agar. Overnight cultures were incubated into fresh TGY medium and exponentialphase cells were used in all experiments. E. coli strains were grown in Luria-Bertani (LB) broth or on LB agar plates at 37uC.

Disruption of the dr0615 locus in D. radiodurans
A three-step gene splicing by overlap extension was used to generate the DR0615 mutant strain (designed MOxyR) [43]. Primers OxyR1 and OxyR2 were used to amplify a BamHI fragment upstream of targeted genes, and primers OxyR3 and OxyR4 were used to obtain a HindIII fragment downstream of targeted genes respectively ( Table 2). The kanamycin resistance cassette containing the groEL promoter was obtained from a shuttle plasmid, pRADK [43]. After these three DNA fragments were digested and ligated, the ligation products were used as template for PCR to amplify the resulting PCR fragment (OxyR1 and OxyR5 used as primer), which was then transformed into exponential phase cells by CaCl 2 treatment. The mutant strains were selected on TGY agar plates supplemented with 20 mg/mL kanamycin were confirmed by PCR product sizes, enzymedigested electrophoresis ( Figure S2), and DNA sequencing.

Complementation of oxyR mutant
Complementation plasmid construction was constructed as previously described by Hua et al [44]. Briefly, chromosomal DNA was isolated from wild type strains. A 1000-bp region containing the oxyR gene was amplified by OxyR com F and OxyR com R ( Table 2), and ligated to pMD18 T-Easy vector (Takara, JP), designed as pMDoxyR. After digested by NdeI and BamHI, the target gene oxyR was ligated to NdeI and BamHI-pre-digested pRADK [43], which named as pRADoxyR. The complementation plasmids were confirmed by PCR and DNA sequence analysis, and transformed into MOxyR and GS09 (K12 oxyR::Kan of E. coli, a gift from Dr Gisela Storz) [7], resulting in two functional complementation strains: MOxyR_wtC (D. radiodurans oxyR mutant strain complemented with pRADoxyR) and GS09C (GS09 complemented with pRADoxyR) ( Table 1).

PCR mutagenesis C 210 of OxyR
The first PCR fragment was obtained using primer OxyR com F and a mutagenic antisense primer C 210 AR ( Table 2, the mutated bases are underlined). The second PCR fragment was obtained using primer OxyR com R and the mutagenic sense primer C 210 AF, which is complementary to the C 210 AR. The mutagenesis PCR was generated by using OxyR com F and OxyR com R [39], and ligated to pMD18 T-Easy vector (Takara), designed as pMD18ox-yRsd. Then, pMD18oxyRsd was digested with NdeI and BamHI, and cloned into pRADK. The resulted pRADoxyRsdC plasmid was transformed into DrOxyR. The oxyR site-directed mutation sequence was verified by sequencing.

Measurement of cell survival rate
The sensitivity of D. radiodurans cells to hydrogen peroxide was assayed following the method as previously described with some modifications [45]. Cells were harvested in early stationary phase, washed twice with and re-suspended in phosphate buffer (20mM, pH 7.4). An aliquot was removed as control and the remaining aliquot was treated with hydrogen peroxide to a final concentra-tion of 20 mM. The mixture was incubated at 30uC in an orbital shaker. At the indicated recovery time points (20,40,60, and 80 min), an aliquot was removed and catalase (Sigma, St. Louis, MO) was added in excess (100 g/mL) to stop the H 2 O 2 treatment. The cells were diluted and spread on solid TGY media for determining the numbers of colonies forming units (cfu). Survival rates are defined as a percentage of the number of colonies obtained comparing with the control. Hydrogen peroxide disk assay was used to assay the survival rate of E. coli [46].
Cell survival rate after ionizing radiation was determined by the method described previously [47]. In short, after the cells were harvested, 200 mL of various MOxyR strains and wild type strains were irradiated at room temperature for 1h with 60 Co c-rays at various distances from the source, which correspond to doses from 0 to 24 kGy. After irradiation, the various MOxyR and wild type strains were plated on TGY plates and incubated at 30uC for 3 days prior to colony enumeration.
All the data provided here represent the mean and standard deviation of at least three independent experiments.

Catalase activity assay
Whole-cell protein extracts were obtained from D. radiodurans cells in exponential phase growth according to the method of Tian et al [48]. Catalase activity was determined as in [49]: 2 mL of 0.2 mg/ mL protein was diluted with 38 mL of PBS buffer, 10 mL of 250 mM H 2 O 2 was then added followed by incubation at 25uC for 2 min. The reaction was quenched with 450 mL of 1% sodium azide. Intracellular ROS accumulation assay ROS generation in cells was assayed as reported [50]. In short, 10 7 cells (100 mL, OD 600 = 1.0, washed three times with PBS) were resuspended in 1 mL DCFH-DA (10 mM) and incubated at 37uC for 20 minutes. After incubation, the cells were washed twice with PBS and re-suspended in 1 mL PBS. Then, the sample was divided in two parts, half (0.5 mL) was exposed to 1 mL of 10M H 2 O 2 for 20 minutes at room temperature, the other half served as the nonexposed control culture. The fluorescence intensities were measured using a fluorescence spectrophotometer with an excitation wavelength of 485 nm and an emission wavelength of 525 nm.

Expression and purification of OxyR wild-type and mutant proteins
Wild-type oxyR was produced from pMDoxyR, which was digested with NdeI and BamHI. Mutant oxyR was obtained from pMD18oxyRsd digested with NdeI and BamHI. The products were ligated into pET28a (Novagen, San Diego, CA), and the resulting plasmids were transformed into E. coli BL21 for overexpression of His-tagged proteins, respectively. Protein expression and purification as previously described [22]. Briefly, an overnight culture was diluted 1: 50 in fresh media to an OD 600 of 0.3 at 37uC and followed by a shift to 4uC. After 0.5 hours incubation at 4uC, cells were induced with 0.1 mM isopropyl-b-D-thiogalactoside (IPTG) for 15 hours at 20uC. The harvested cells were resuspended in lysis buffer (20 mM NaH 2 PO 4 ; 20 mM Na 2 HPO 4 ; 400 mM NaCl; 15 mM imidazole; 1 mM DTT) and disrupted at 4uC in a sonicator. After centrifugation at 12, 000 rpm for 20 min at 4uC, the supernatant was loaded onto Ni-NTA agarose columns (Qiagen, Valencia, CA). The purified OxyR protein was applied to a Superdex 300 HR 16/70. The purity of protein samples was determined using 12% SDS-PAGE, and only fractions with pure OxyR protein were used for further experiments.
Reversible formation of cysteine-sulfenic acid trapping with NBD chloride Modification of OxyR protein by NBD chloride (Sigma) at C210 was detected as previously described [8,20,39,51]. Briefly, reduced or oxidized proteins were mixed with 1 mM NBD-Cl in dimethyl sulfoxide and incubated for 60 min at 25uC in the dark. Then the NBD-Cl was removed by ultrafiltration with YM-10 (Millipore, Bedford, MA) dialysed three times in 50 mM pH 7.0 potassium phosphate buffer containing 150 mM NaCl and 1 mM EDTA. The absorbance spectra (300 to 600 nm) of the modified proteins were measured on a ND-1000 spcetrophotometer (NanoDrop, Wilmingon, De. US).

RNA isolation, quantitative real-time PCR (QRT-PCR) experiments
The wild-type strain and MOxyR strain were grown in TGY to mid-exponential phase. For H 2 O 2 treatment, the cultures were divided in two; one half of the culture was treated with H 2 O 2 at a final concentration of 20 mM, while the other half was used as non-treatment control. RNA isolation, and QRT-PCR were carried out as previously described [28]. Total RNA was extracted from cells using TRI Reagent (Invitrogen, Carlsbad, CA), after liquid nitrogen grinding. Then the RNA samples were treated with Rnase free Dnase I (Promega, Madison, WI) and purified using phenol chloroform extraction. RNA quality and quantity were evaluated by UV absorbance at 260 and 280 nm.
QRT-PCR assay utilized RNA samples obtained from different conditions and first-strand cDNA synthesis was carried out in 20 mL of reaction containing 1 mg of RNA sample combined with 3 mg of random hexamers using SuperScript III Reverse Transcriptase kit (Invitrogen). Then Quant SYBR Green PCR kit (TIANGEN, BJ) was used to amplification following the manufacturer's instructions. As an internal control, a house-keeping gene encoding glycosyl transferase, dr0089, was used as a house-keeping gene, encoding the glycosyl transferase [28]. In our hands, DR0089 was unaffected by any of our treatments. cDNA probes for microarray hybridization were prepared from four biological replicate total RNA samples each of wild type and MOxyR cultures. All primers used in QRT-PCR are shown in Table 2.

Microarray hybridization and data analysis
Microarray design and constructions were carried out as our previous work [28]. Total RNA for microarray hybridization were obtained from four biological replicate samples of each of wild type and MOxyR under normal condition. First, RNA (4 mg) was annealed to 9 mg of random hexamer primers (Takara) in total volume of 20 mL at 70uC for 10 min and subsequent keep on ice for 2 min. cDNA was synthesized at 42uC overnight in total 31 mL using SuperScript III Reverse Transcriptase kit (Invitrogen) with 0.5 mM dNTP mix containing amino allyl-dUTP (GE, Piscataway, NJ). The reaction was terminated by adding 20 mL EDTA (0.5 M), and RNA was hydrolysed by adding 20 mL NaOH (1 M), then incubating at 65uC for 20 min. After neutralized with 50 mL Hepes (1M, pH 7.0), unincorporated free amino allyl-dUTPs were removed by ultrafiltration with YM 30 (Millipore), and resultant cDNA samples were coupled to 1 pmol Cy3 or Cy5 dyes (GE) in 0.1 M sodium carbonate buffer for 2 h at room temperature in the dark. Unincorporated free Cy3 or Cy5 were removed by ultrafiltration with YM 30. Two labeled cDNA pool (wild type and MOxyR) to be compared were mixed and hybridized simultaneously to the array in a solution containing 36saline sodium citrate (SSC), 0.3 % SDS, and 24 mg of unlabeled herring sperm DNA (Gibco BRL, Gaithersburg, MD) [52]. Following hybridization, slides were washed as published paper [52].
Measurement of spot intensity and normalization were carried out as our paper [28]. In short, slides were scanned with a GenePix 4000B imager (Axon, Union City, CA), and spot intensities were obtained by software GenePix pro 5.1. Normalization and statistical analysis were carried out in the R computing environment (2.11, Raqua on the Windows) using the linear models for microarray data package (Limma) [53]. Within Limma, prior to channel normalization, microarray outputs were filtered to remove spots of poor signal quality by excluding those data points with mean intensity less than two standard deviations above background in both channels. Then, global LOESS normalization was used to normalize all data, and the 2-replicate spots per gene in each array were used to maximize the robustness of differential expression measurement of each gene via the ''lmFit'' function [54]. The microarray data have been deposited in the Gene Expression Omnibus database under accession no. GSE9636.

Assay of intracellular Fe and Mn ion concentration
Total iron and Mn concentration were detected as described previously with some modification [55]. Bacteria were grown aerobically to OD600 0.8 in 600 ml of TGY broth. After centrifugation at 10,000 g for 10 min at 4uC, cells were washed twice in 200 ml of phosphate-buffered saline (PBS) with 1 mM EDTA (pH 7.4) and resuspended in 200 ml of PBS without EDTA. After centrifugation, the pellet was resuspended in 10 ml of PBS, 8 ml of which was used for iron and manganese analysis. Cell dry weight was estimated with the remaining 2ml suspension. For iron and manganese analysis, pelleted bacteria were resuspended in 1 ml of Ultrex II nitric acid (Fluka AG., Buchs, Switzerland) and incubated at 80uC for 1 hour. After centrifugation at 20,000 g for 20 minutes, the supernatant was filtered against 0.45 mM membrane. The concentration of samples was analyzed for iron and Mn content by inductively coupled plasma-optical emission spectroscopy (ICP-MS, Model Agilent 7500a, Hewlett-Packard, Yokogawa Analytical Systems, Tokyo, Japan). All buffers and nitric acid solutions were analyzed as described above to correct for background.

Gel mobility shift assays
Gel mobility shift assays were performed with FITC-labeled DNA fragments (0.05 pmol) mixed with purified OxyR protein (oxidized protein or reduced protein, 200 nmol) in a total volume of 20 mL. The binding buffer contained 10 mM Tris-Cl (pH 8.0), 50 mM NaCl, 1 mM EDTA, 5% Glycerol, 50 mg/mL BSA and 5 mg/mL calf thymus DNA [22,39]. The reaction mixture was incubated at room temperature for 30 min and loaded onto a 1.5% agarose gel in 0.56TBE. Electrophoresis was performed at 80 V for 1 h at 4uC and followed by photographed [56,57]. For generation of the labeled DNA, the appropriate operator fragment of target genes (dr1709, dr1998, dr2263, drB0036, and drB0125) was amplified by PCR with genomic DNA and cloned into pMD18, followed by FITC-labeled RV-M and reverse priming of target genes. In addition, dr0207 was used as negative control. All primers are listed in Table 2. Figure S1 H2O2 disk assay. Photos showing the zones of inhibition by H2O2 in E. coli K-12 (wild type) (A), GS09 (oxyR::kan mutant) (B), and E. coli K-12 strain GS09 complemented with the droxyR gene (C). (D) Histogram showing the results of H2O2 disk assay in E. coli. In the assay, E. coli cells were grown in Luria-Bertani (LB) broth at 37 uC with overnight shaking. 200 ml of the overnight cultures were added to LB top agar and spread onto LB agar. Then, 10 ml of 3% H2O2 was pipetted onto 3MM Whatman paper disks (0.7-cm diameter), and these disks were placed on top of the agar and incubated at 37 uC overnight. The zone of inhibition, in mm, was taken as a measure of H2O2 sensitivity. The zone of inhibition was measured in three dimensions, and the mean values and standard deviations were calculated. Found at: doi:10.1371/journal.pone.0001602.s001 (4.13 MB TIF) Figure S2 Disruption of D. radiodurans droxyR gene. Verification of droxyR gene disruption by PCR analysis. Purified PCR fragments were amplified from the genomic DNA of strain R1 and MOxyR using primers (OxyR1 and OxyR5) that flank the coding sequences for droxyR. The PCR products of R1 revealed a band of ,2550 bp length (band 1), whereas those of MOxyR resulted in a ,3500 bp fragment (band 2). Bands 3 and 4 denote PCR products of R1 and MOxyR were digested with BamHI, respectively. Bands 5 and 6 denote PCR products of R1 and MOxyR were digested with HindIII, respectively. M denotes molecular weight standards. Found at: doi:10.1371/journal.pone.0001602.s002 (2.63 MB TIF)