The Mycobacterial LysR-Type Regulator OxyS Responds to Oxidative Stress and Negatively Regulates Expression of the Catalase-Peroxidase Gene

Protection against oxidative stress is one of the primary defense mechanisms contributing to the survival of Mycobacterium tuberculosis in the host. In this study, we provide evidence that OxyS, a LysR-type transcriptional regulator functions as an oxidative stress response regulator in mycobacteria. Overexpression of OxyS lowers expression of the catalase-peroxidase (KatG) gene in M. smegmatis. OxyS binds directly with the katG promoter region and a conserved, GC-rich T-N11-A motif for OxyS binding was successfully characterized in the core binding site. Interestingly, the DNA-binding activity of OxyS was inhibited by H2O2, but not by dithiothreitol. Cys25, which is situated at the DNA-binding domain of OxyS, was found to have a regulatory role for the DNA-binding ability of OxyS in response to oxidative stress. In contrast, the other three cysteine residues in OxyS do not appear to have this function. Furthermore, the mycobacterial strain over-expressing OxyS had a higher sensitivity to H2O2.Thus, OxyS responds to oxidative stress through a unique cysteine residue situated in its DNA-binding domain and negatively regulates expression of the katG gene. These findings uncover a specific regulatory mechanism for mycobacterial adaptation to oxidative stress.


Introduction
Mycobacterium tuberculosis, the causative microbe of tuberculosis (TB), has a unique ability to survive and persist within host cell macrophages for long periods of time [1]. It can evade the host immune system by preventing phagosomal maturation and resist killing by reactive oxygen [2] and nitrogen intermediates [3]. Protection against oxidative stress is one of the primary defense mechanisms contributing to the survival of M. tuberculosis in the host [4].
In several bacteria, KatG, which encodes a catalase-peroxidase, is important for protection against oxidative stress. katG was shown to have a role in protecting the bacterium against micromolar concentrations of H 2 O 2 in Xanthomonas campestris pv. Campestris [5]. A knockout mutation in katG that causes loss of catalaseperoxidase activity correlates with increased susceptibility to H 2 O 2 . katG expression is induced by oxidants in an OxyRdependent manner [5]. It is also required for the activation of INH (isoniazid, a first-line anti-tuberculosis drug) in M. tuberculosis [6]. Expression of katG has been found to be negatively regulated by FurA in both M. tuberculosis and M. smegmatis [7][8][9][10][11]. FurA is a homologue of the ferric uptake regulator Fur and is encoded by a gene located immediately upstream of katG. The regulation of katG expression by FurA has been shown to be induced upon oxidative stress under the control of pfurA, located immediately upstream of the furA gene, but not of pkatG, located within the terminal region of the furA coding sequence [8]. Master et al. proposed that a katG promoter was present in this region [12]. It was also argued that the 59 end of katG mRNA is generated by processing instead of transcription initiation [10]. Recently, a transcript analysis of the furA-katG loci revealed that katG is transcribed independent of furA in a fast-growing Mycobacterium sp. strain JC1 DSM 3803 [13]. However, the mechanism and the transcription factor for furAindependent regulation of katG remain to be characterized in mycobacteria [7,12].
In several bacteria the expression of katG or ahpC is regulated in a FurA-independent manner by OxyR [14,15], a well-characterized regulator responsible for adaptation to oxidative stress [16]. However, both M. tuberculosis and M. smegmatis lack a functional OxyR [11,15,17]. LysR-type transcriptional regulators (LTTRs) are thought to constitute the largest family of transcriptional regulators, and their functional orthologs have been identified in diverse bacterial genera, archaea and eukaryotic organisms. They regulate diverse functions including oxidative stress response, cell division, virulence, and secretion, among others [18]. Domenech et al. showed that OxyS, a M. tuberculosis LysR-type regulator, contains OxyR-like DNA binding domain and plays a role in response to oxidative stress in M. tuberculosis [17]. Overproduction or depletion of OxyS in M. tuberculosis did not affect susceptibility to isoniazid but increasing the concentration of OxyS lowered levels of the alkyl hydroperoxide reductase, AhpC, and rendered the tubercle bacillus more susceptible to organic hydroperoxides [17]. The introduction of oxyS on a multicopy plasmid did not affect catalase-peroxidase activity or KatG levels in the absence of peroxide. In the presence of peroxide they saw an increase in KatG protein level. They deleted the oxyS gene from the chromosome of M. tuberculosis and found no change in catalaseperoxidase activity, concluding that oxyS was not the main regulator of katG, but could be involved in a coregulatory process [17]. However, they mentioned that OxyS could not be purified successfully in E. coli and its biochemical characteristics remained unclear.
Here, we report that we have successfully purified soluble OxyS and its mutant forms from an E. coli expression system and characterized it as an oxidative stress response regulator in M. tuberculosis. OxyS has been shown to bind directly with the katG promoter region and the conserved binding site for OxyS in the promoter region of katG was successfully mapped out. A regulatory cysteine in the DNA-binding domain of OxyS was found to be important for its response to oxidative stress. Furthermore, we examined the effect of the expression level of OxyS on the mycobacterial sensitivity to H 2 O 2 . Our findings suggest that OxyS is a negative regulator of katG in mycobacteria.

Results
OxyS is a conserved mycobacterial LysR-type regulator and is involved in regulation of katG Using a previously developed bacterial one-hybrid system, we have successfully isolated a number of novel transcription factors involved in the regulation of virulence genes in M. tuberculosis [19]. Interestingly, among the many newly identified transcription factors, we found OxyS was involved in the regulation of katG. The M. tuberculosis OxyS protein shares strong sequence identity with the LysR-type transcriptional regulator family (conserved domains database (CDD), NCBI). As shown in Fig. 1A and 1B, the Nterminal part of OxyS contains the helix-turn-helix motif and the C-terminal region contains a LysR substrate-binding domain which is structurally homologous to type-2 periplasmic binding proteins. Furthermore, the N-terminal helix-turn-helix motif of OxyS was found to be highly conserved among several different mycobacterial species (Fig. S1, Table 1).
A previous study implied that OxyS is involved in regulation of KatG activity in M. tuberculosis [17]. Since the regions upstream of the katG gene in M. tuberculosis and M. smegmatis are highly tuberculosis OxyS. The conserved domains in the N-terminus and C-terminus of OxyS were found by searching the CDD database on the NCBI website. (B) The modeled structure of OxyS was obtained using the automated comparative protein modeling web server SWISS-MODEL [34] and CbnR protein (a LysR family transcriptional regulator in Ralstonia eutropha NH9) [35] as a template (PDB ID: 1iz1). (C) Detection of OxyS protein by western blotting in the recombinant mycobacterial strains. An inducible system for conditional gene over-expression in mycobacteria [36] was used to over-express OxyS in M. smegmatis. Lane 1, purified His-OxyS; Lane 2, cell lysate from Msm/pMV261; Lane 3, cell lysate from Msm/pMV261-OxyS. (D) qRT-PCR assays conducted to analyze changes in gene expression in M. smegmatis. The experiment was carried out as described in the ''Materials and Methods'' section. The 16S rRNA gene, rrs, was used as an internal control for normalization. Target genes were amplified using specific primers. Expression levels of all genes were normalized to the levels of 16S rRNA gene transcripts, and fold-changes in expression of each gene were calculated. Representative data are shown. doi:10.1371/journal.pone.0030186.g001 conserved [8] (Fig. S2) we decided to explore that further by overexpressing M. tuberculosis OxyS in M. smegmatis to determine the physiological role of OxyS. As shown in Fig. 1C, OxyS reached a high level of expression in the recombinant strain as revealed by Western blotting assays using anti-OxyS serum. Notably, qRT-PCR (Quantitative real-time PCR) analysis indicated that the level of katG expression in the recombinant M. smegmatis strain was only 36% of that of the control strain (Fig. 1D). In contrast, the expression level of furA showed no significant change when compared with M. smegmatis strains harboring empty vectors. These results indicate that the expression of katG is negatively regulated by OxyS, while the furA expression is not regulated by OxyS.
OxyS directly targets the promoter region of katG in M. tuberculosis and M. smegmatis To examine whether a direct interaction occurs between OxyS and the regulatory sequence of the M. tuberculosis katG gene, a bacterial one-hybrid assay was conducted by cloning the katG promoter region upstream of HIS3-aadA reporter genes of the bacterial one-hybrid reporter vector pBXcmT ( Fig. 2A) [19]. As shown in Fig. 2B, co-transformant strains with the katG promoter  The regulatory sequence of the katG gene was cloned into the upstream of HIS3-aadA reporter genes of the bacterial one-hybrid reporter vector pBXcmT [19]. (B) The interaction between OxyS and the promter region of katG was measured by bacterial one-hybrid analysis [19]. Left panel: bacterial one-hybrid plates. Right panel: an outline of the plates in the left panel. Each unit represents the corresponding co-transformant in the plates. CK+: co-transformant containing pBX-Rv2031p and pTRG-Rv3133c as a positive control. CK2: co-transformant containing pBX-Rv2031p and pTRG-Rv3133c-deltaC as a negative control [19]. Rv3911p (the promoter of the Rv3911 gene) was also used as a negative control. (C) in vivo ChIP assays for the interaction of OxyS with the katG promoter in M. tuberculosis. DNA recovered from the immunoprecipitates was amplified with primers specific for either katGp or a negative control promoter Rv3911p. '+' refers to the immunoprecipitate obtained with OxyS antibodies, whereas '2' refers to the control in which ChIP was carried out without any primary antibodies. 'Input' refers to total genomic DNA prior to IP reaction and was used as a positive control in PCR. and OxyS grew well on selective plates. In contrast, the strains containing either OxyS alone or the katG promoter alone did not grow on selective plates. In addition, co-transformant strains containing the promoter of Rv3911 (Rv3911p), an unrelated promoter, and OxyS did not grow either (Fig. 2B). These results indicate that OxyS can specifically interact with the promoter of the katG gene. Using anti-OxyS antibodies and ChIP (chromatin immunoprecipitation) assays, we characterized the association of OxyS with the katG promoter in vivo in M. tuberculosis. Fig. 2C (top panel) shows that the signal for katG promoter was enriched in the anti-OxyS immunoprecipitate compared with that in the control sample without any antibody. In contrast, no obvious signal was observed for the promoter of Rv3911 (Rv3911p) in both the anti-OxyS and the control immunoprecipitates (Fig. 2C, bottom panel). We successfully purified soluble His-tagged OxyS from an E. coli expression system by inducing the expression of MtbOxyS protein under low temperature conditions (Fig. S3). Binding of the purified OxyS protein with the katG promoter was then confirmed by further EMSA (electrophoretic mobility shift assay) experiments. As shown in Fig. 2D, when increasing amounts of OxyS (100-600 nM) were added into the reactions, shifted bands corresponding to the OxyS/katG promoter complex were observed and a corresponding increase in the percentage of protein/DNA complexes was seen. Using competitive EMSA assays (Fig. S4C), we comfirmed that this interaction was specific. Notably, the interaction of OxyS with the promoter of MsmkatG was also confirmed in the recombinant OxyS-overexpressing M. smegmatis strain by both ChIP and EMSA assays (Fig. S5).
Taken together, these results indicate that OxyS can specifically interact with the promoter of the katG gene in vivo and in vitro, and this direct interaction is conserved in both M. tuberculosis and M. smegmatis.

Characterization of the binding sites and sequence motifs for OxyS in the promoter region of katG
To identify the binding sites for OxyS in the promoter region of katG (katGp), we first obtained two short DNA fragments (katGp1 and katGp2) of the katG promoter (Fig. S4A). As shown in Fig. S4B, OxyS could form specific protein/DNA complexes with the substrates katGp and katGp2. In contrast, no binding activity was observed for the substrate katGp1. In addition, unlabeled katGp and katGp2 could competitively inhibit the binding of OxyS with the labeled DNA substrate, while unlabeled katGp1 had no effect on OxyS binding (Fig. S4C). This indicates that the 21 to 2180 region (katGp2) of the katG promoter contains the binding site for OxyS.
Two fragments-foot1 and foot2-were further designed ( Fig.  S4A) to precisely determine the binding sites for OxyS in the katG promoter by DNase I footprinting assays. As shown in Fig. 3A, the foot1 and foot2 fragments were incubated with increasing amounts of the OxyS protein and digested with DNase I. Several regions were protected from digestion in the presence of the OxyS protein (Fig. 3A). Two OxyS binding boxes, named OxyS box1 and OxyS box2, were further found in the protected regions, as indicated in Fig. 3B. The core OxyS binding sites of M. tuberculosis katG also contained the LysR-binding motif T-N 11 -A, which has previously been shown to be a classic LTTR box [18].
A blast assay for the binding sequence for OxyS (OxyS box1) among different mycobacterial species revealed a conserved, GCrich T-N 11 (Fig. 3C). Specific binding of OxyS to the motif was then confirmed by base replacement and EMSA assays. As shown in Fig. 3D, when the conserved ''TG'' and ''GA'' in OxyS box1 were replaced by ''CC'', the OxyS protein lost its ability to bind to the mutant DNA substrate as evidenced by the result of our EMSA assay that specific protein/DNA complexes were observed for the wild type substrate (Fig. 3D, left panel), but no such complex was observed for the mutant substrate (Fig. 3D, right panel).

The DNA-binding ability of OxyS is inhibited by H 2 O 2 and Cys25 is a regulatory residue
We assayed the effect of redox reagents on the DNA-binding ability of OxyS. As shown in Fig. 4A, a stepwise decrease in amounts of the specific protein/DNA complex was observed as 0.01-3 mM H 2 O 2 was added to the DNA-binding reaction mixture (Fig. 4A, lanes 1-5). In contrast, no effect was observed when DTT (dithiothreitol) was added to the reaction mixture ( Fig. 4A, lanes 6-10), indicating that the ability of OxyS to bind DNA was specifically reduced by H 2 O 2 . We also examined the expression level of katG in the M. smegmatis strain under oxidative stress (due to OxyS over-expression) by qRT-PCR assays. As shown in Fig. 4B, the expression level of katG increased 2-fold after 2 mM H 2 O 2 treatment, indicating that the negative regulation of katG by OxyS was eliminated.
The OxyS protein contains four cysteines (Fig. 1B), and two of them (Cys25 and Cys113) were found to be conserved among several mycobacterial OxyS orthologs (Fig. S1). Cysteine residues have been shown to be responsible for redox-sensing in many transcriptional regulators [20][21][22]. To identify the sites responsible for sensing oxidative stress in OxyS, site-directed mutations were introduced into these cysteine residues in the oxyS gene. The mutant proteins were expressed and successfully purified from an E. coli expression system (Fig. S3). As shown in Fig. 4C, all the mutant proteins maintained their DNA-binding ability when compared with wild-type protein. We examined the effect of these mutations on the ability of OxyS to bind to DNA in response to H 2 O 2 . Cys113, Cys124 and Cys293 mutations did not change the sensitivity of OxyS to H 2 O 2 (Fig. 4D). Interestingly, the DNAbinding activity of the mutant protein OxyS-Cys25A, in which the mutation was situated in the DNA-binding domain of OxyS, was not affected by H 2 O 2 (Fig. 4D) indicating that it lacked the ability to respond to the oxidative signal. These results indicate that the Cys25 residue is involved in the regulation of the DNA-binding ability of OxyS under oxidative stress, while the other three cysteine residues are dispensable.
The fate of OxyS and its mutant variants after peroxide treatment was further determined by native-PAGE assays. As shown in Fig. S6, addition of 3 mM H 2 O 2 (lane 3) changed the electrophoretic mobility of OxyS and all its mutant variants. In contrast, addition of 3 mM DTT (lane 2) neither altered the electrophoretic pattern nor changed the electrophoretic mobility of MSMEG_6092, a DNA binding protein from M. smegmatis with no cysteine residues (Fig. S6). These results indicate that purified OxyS proteins are in a reduced state and are capable of redoxsensing.
Taken together, our results lend support to a model in which all the cysteine residues in OxyS can sense oxidative signal, but Cys25 is the only regulatory cysteine residue capable of both sensing the oxidative signal and regulating the ability of OxyS to bind to DNA under oxidative stress.
OxyS-Cys25A lacks the ability to respond to the oxidative signal in vivo in M. smegmatis We first examined the effect of over-expressing OxyS on oxidative stress response in M. smegmatis using a modified bacterial growth time course assay. As shown in Fig. 5A, we observed that the M. smegmatis strains over-expressing OxyS were more sensitive to H 2 O 2 as 2-5 mM H 2 O 2 was added to the medium if compared with the control strain. This is consistent with a previous observation [17]. The findings of our detailed bacterial growth time course assays with increasing concentrations of H 2 O 2 also support the same conclusion (Fig. S7). As shown above, overexpression of OxyS lowered expression of the katG gene in M. smegmatis (Fig. 1D) and OxyS-Cys25 is a regulatory cysteine residue (Fig. 4D). Thus, we further compared the DNA-binding ability of mutant OxyS proteins with that of the wild type protein for the promoter of katG in vivo in response to H 2 O 2 . As shown in Fig. 5B, when over-expression of OxyS-WT or OxyS-C113A in M. smegmatis, the signal for katG promoter with 2 mM H 2 O 2 treatment was reduced substantially in the anti-OxyS immunoprecipitate if compared with that in the control sample without H 2 O 2 treatment. In contrast, no obvious signal change for the enriched katG promoter in response to H 2 O 2 was observed for over-expression of OxyS-C25A (Fig. 5B). Therefore, the mutant protein OxyS-Cys25A lacked the ability to respond to the oxidative signal in vivo in M. smegmatis. Discussion M. tuberculosis can resist the damaging effects of reactive oxygen and nitrogen intermediates produced in the host cells and thus survive successfully for long periods of time [2,3]. However, the transcriptional regulatory processes involved in this mycobacterial defense mechanism are still unclear. In this study, we show that OxyS, a LysR-type transcriptional regulator, is an oxidative stress response regulator in M. tuberculosis that binds directly with the katG promoter region, which is located in the terminal part of the furA coding region. Furthermore, the DNA-binding activity of OxyS was found to be inhibited by H 2 O 2 . We characterized a conserved residue, which is situated within the DNA-binding domain of OxyS, important for DNA binding in response to oxidative stress. Finally, the expression level of katG was greatly reduced in the mycobacterial strain overexpressing OxyS and the recombinant strain had an elevated sensitivity to H 2 O 2 .Taken together, our results indicate that OxyS functions as a negative regulator of katG in response to oxidative stress in mycobacteria.
Response to oxidative stress plays an important role in pathogen-host interaction during infection [4] and in determining the intrinsic susceptibility of mycobacterial species to INH [6]. Both in M. tuberculosis and in M. smegmatis, the expression of katG is induced upon oxidative stress under the control of pfurA, and is negatively regulated by FurA [7,8]. However, it has been suggested that katG can be transcribed independent of furA in a fast-growing Mycobacterium sp. strain [13]. It has recently been reported that in Caulobacter crescentus, the expression level of the katG gene is positively regulated by OxyR in a FurA-independent manner. OxyR was found to bind to a canonical OxyR binding site in the promoter region of katG, and this interaction was redox dependent, as only an oxidized form of the protein could bind to the katG promoter [14]. In the present study, we found that the DNA binding ability of OxyS was inhibited, not stimulated, by H 2 O 2 . Moreover, in M. smegmatis, the expression level of katG decreased when OxyS was over-expressed, while the expression level of furA showed no significant change, suggesting that the expression of katG is negatively regulated by OxyS, and the furA  (Fig. 4B). This result is consistent with an inhibitory effect of H 2 O 2 on the DNA-binding ability of OxyS that we observed. Interestingly, Domenech and his colleagues also observed an increase in KatG activity upon H 2 O 2 treatment in M. tuberculosis over-expressing OxyS [17]. In the present study, the morphology of M. smegmatis cells was also examined using scanning electron microscopy (SEM) to further investigate the reason for the increased sensitivity of M. smegmatis strains over-expressing OxyS to H 2 O 2 . Over-expression of OxyS in M. smegmatis had no obvious effect on cell morphology when compared with M. smegmatis cells harboring empty vectors (Fig. S8). Therefore, the effect of overexpression of OxyS on the sensitivity of M. smegmatis to H 2 O 2 was not due to the changes in cell morphology.
In the current study, by using a bacterial one hybrid system, we found that the reporter genes (HIS3-aadA) downstream the katG promoter could be successfully activated (Fig. 2B), which indicating the existance of regulatory elements in this region. We also found using in vivo and in vitro assays that MtbOxyS can directly interact with the promoter region of katG by binding to a typical LysR type T-N 11 -A motif in the katG promoter. Two OxyS binding boxes in the promoter region of katG were identified. However, only one major protein/DNA complex band was observed in our EMSA assays (Fig. 2D). This may due to the different binding preference of OxyS to these boxes or the cooperative binding of these two boxes by OxyS. Indeed, through a blast assay of the katG promoters from different mycobacterial species, we found that OxyS box1 was the most conserved motif (Fig. 3C). This suggests a common regulatory mechanism of katG expression by OxyS among different mycobacterial species. Interestingly, the binding motif for OxyS identified in the current study also shows similarity to the binding motif for OxyR in C. crescentus [14].
Cysteine residues have been shown to be responsible for redox sensing [20][21][22]. OxyR senses H 2 O 2 and is activated through the formation of a transient disulphide bond between Cys199 and Cys208 in its regulatory domain [23]. Chen et al. have recently reported an OxyR sensor with one cysteine residue, and divided the OxyR family into two classes: typical 2-Cys OxyR and 1-Cys OxyR [20]. In the present study, we found that all the cysteine residues in OxyS could sense the oxidative signals, but Cys25 was the only regulatory cysteine residue which could both sense the oxidative signal and regulate the DNA binding ability of OxyS under oxidative stress. This represents an interesting redox sensing and regulatory mechanism through a cysteine residue in the DNA binding domain in mycobacteria. The electrophoretic mobility of OxyS and its mutant variants were changed under oxidative stress and the DNA binding ability of OxyS was inhibited after H 2 O 2 treatment. These characteristics are similar to that of FurA, an oxidative stress-responsive protein which regulates the expression of katG in M. tuberculosis [9]. The regions upstream of the katG gene in M. tuberculosis and M. smegmatis were found to be highly conserved [12], suggesting a putative common regulatory mechanism in these two species. In the present study, we found that OxyS could interact with the promoter of katG in both M. tuberculosis and M. smegmatis. Furthermore, we identified a LysR family transcriptional regulator (MSMEG_0156) which shares 62% amino acid identity with MtbOxyS in a blast assay ( Table 1). The regulatory cysteine residue Cys25 is conserved in both of these proteins.
OxyR regulates KatG and other proteins induced by exposure to H 2 O 2 in E. coli [24]. However, the oxyR gene in M. tuberculosis harbors multiple mutations [25,26] that render it inactive. The alterations in oxyR are conserved in all members of the M. tuberculosis complex, including M. africanum, M. bovis, and M. microti [25], with minor variations [27]. Moreover, no potential OxyRencoding sequences could be detected in M. smegmatis [28]. In contrast, intact oxyR genes have been found in M. leprae [25,26] and M. avium [26]. In the present study, we showed that the OxyS proteins from both M. tuberculosis and M. smegmatis can respond to H 2 O 2 stress and regulate the expression of the katG gene. This lends support to the interesting possibility that given the lack of functional OxyR, OxyS could be an alternative regulator of the response to oxidative stress in these mycobacterial species.
In conclusion, we have provided evidence to show that OxyS, a LysR-type transcriptional regulator, functions as an oxidative DNA recovered from the immunoprecipitates was amplified with primers specific for MsmkatGp. '+' refers to the immunoprecipitate obtained with OxyS antibodies, whereas '2' refers to the control in which ChIP was carried out without any primary antibodies. 'Input' refers to total genomic DNA prior to IP reaction and was used as a positive control in PCR. doi:10.1371/journal.pone.0030186.g005 stress response regulator in M. tuberculosis. OxyS negatively regulates the expression of the katG gene and responds to oxidative stress through its unique cysteine residue situated in the DNAbinding domain. The DNA binding ability of OxyS was shown to be inhibited by H 2 O 2 . A conserved T-N 11 -A and GC-rich motif for OxyS binding in the promoter region of katG was characterized. The mycobacterial strain over-expressing OxyS had increased sensitivity to H 2 O 2 . These findings suggested a new function for the mycobacterial OxyS gene and a regulatory mechanism for their adaptation to oxidative stress.

Materials and Methods
Bacterial strains, plasmids, enzymes and chemicals Escherichia coli BL21 (DE3) strains, purchased from Novagen, were used as the host strain to express M. tuberculosis OxyS. pBT, pTRG vectors and E. coli host strains XR were purchased from Stratagene. pET28a was purchased from Novagen. Restriction enzymes, T4 ligase, and modification enzymes were from TaKaRa Biotech. Pyrobest DNA polymerase and deoxynucleoside triphosphates (dNTPs) were purchased from TaKaRa Biotech. DNA purification kits were purchased from Watson Biotechnologies. All antibiotics were purchased from TaKaRa Biotech. Ni-NTA (Ni 2+nitrilotriacetate) agarose columns were obtained from Qiagen. All DNA and oligonucleotides were synthesized by Invitrogen (Table  S1).

Site-directed mutagenesis
To identify the oxidative stress-sensing sites, site-directed mutations were introduced into the cysteine residues in the oxyS gene by overlap PCR [29]. Sequences of mutated oligonucleotides are listed in Table S1. All fragments were ligated into pET28a and were subsequently sequenced to confirm the presence of the sitedirected mutations.

Cloning, expression and purification of recombinant proteins
The M. tuberculosis oxyS gene was amplified from genomic DNA using the High Fidelity PCR system (TaKaRa) with appropriate primers and cloned into the pET28a over-expression vectors to produce recombinant vectors (Table S2). E. coli BL21 (DE3) cells transformed with the recombinant plasmid were grown at 37uC for 5 hours in 1 liter of LB medium containing 30 mg/ml 21 kanamycin, and at 20uC for 15 hours after 0.5 mM IPTG (isopropyl b-D-1-thiogalactopyranoside) was added. Purification of recombinant proteins was carried out as described in a previous study [30]. Purified proteins were more than 99% pure as determined by SDS-PAGE. The gels were stained with coomassie blue R-250 and destined with 5% methanol, 10% acetic acid until the background was clear. Protein concentrations were determined by spectrophotometric absorbance at 260 nm according to Gill and Hippel [31]. Native PAGE for OxyS and its mutant proteins were performed at room temperature with the use of 12% non-denaturing polyacrylamide gels. Electrophoresis gels were run at a constant voltage of 125 V in 37 mM Tris-glycine buffer (pH 8.9) until the tracking dye reached the bottom of the gel. The gels were then stained and destained as described above.

Bacterial one-hybrid analysis
The regulatory sequences of the katG gene or Rv3911, a sequence used as negative control, were cloned into the reporter genes upstream of HIS3-aadA of the reporter vector pBXcmT (Table S2) [19]. OxyS was fused to the N-terminal domain of the a-subunit of RNA polymerase in the pTRG vector (Stratagene). The bacterial one-hybrid assay was carried out as described in a previous study [19]. The reporter strain was co-transformed with pBX and pTRG recombinant plasmids. The colonies (reporter strain contains the indicated pBX and pTRG recombinant plasmids) were selected on plates containing 15 mM 3-AT (3-amino-1,2,4-triazole) and 8 mg/ ml streptomycin, while plates that did not contain 3-AT and streptomycin were used as controls. Detection of protein-DNA interactions was based on transcriptional activation of the reporter genes, HIS3 and aadA, which allows growth in the presence of 3-AT and streptomycin, respectively. Co-transformants containing pBX-Rv2031p and pTRG-Rv3133c were used as positive controls and were expected to grow on the selective screening medium, while cotransformants containing pBX-Rv2031p and pTRG-Rv3133c-deltaC were used as negative controls [19].

Chromatin immunoprecipitation (ChIP) assays
Actively growing cultures were treated with 1% (v/v) formaldehyde at room temperature for 10 minutes under gentle shaking and the reaction was stopped by adding glycine (at a final concentration of 0.125 M) for 10 minutes with gentle shaking. Cross-linked cells were collected by centrifugation and washed twice with PBS and once with TBSTT (150 mM NaCl, 20 mM Tris-HCl pH 7.5, 10 mM EDTA, 0.1% (v/v) Triton X-100, 0.1% (v/v) Tween 20) to remove excess formaldehyde. Cells were resuspended in 1 ml of TBSTT, sonicated on ice with 20 pulses of 20 seconds and 40% amplitude. The average DNA fragment size obtained was approximately 0.5 kb. Cell debris was removed by centrifugation and the clear supernatant was stored as total lysate. Typically, 100 ml of the supernatant was used as input. 900 ml of the supernatant was incubated for 3 h on a rocker at 4uC with 10 ml of antibodies against OxyS, and then the complexes were immunoprecipitated with 20 ml 50% protein A-agarose for 1 h on a rocker at 4uC. A parallel ChIP experiment without the OxyS antibody was set up as a negative control. The immunocomplex was recovered by centrifugation, washed five times with TBSTT, and resuspended with 100 ml TE buffer (20 mM Tris-HCl pH 7.8, 10 mM EDTA, 0.5% (w/v) SDS). Crosslinking was reversed at 65uC for 6 h. The DNA samples were purified, resuspended in 50 ml TE buffer and analyzed by PCR. Each experiment was performed in duplicate, and repeated twice. The amplified promoter fragments were 350 bp in length. The amplification protocol included one cycle of 5 min at 95uC, and 32 cycles of three steps each: 1 min at 95uC, 1 min at 60uC, and 1 min at 72uC. Primer sequences are indicated in Table S1.

Electrophoretic mobility shift assays (EMSA)
The binding of OxyS to the katG promoter was investigated using a modified electrophoretic mobility shift assay as previously published [30] with the following changes. The forward primers were FITC-labeled at their 59 termini (Table S1). The doublestranded substrates were prepared according to a previously published procedure [30]. The reactions (10 ml) for measuring mobility shift contained FITC-labeled DNA substrate and various amounts of OxyS at different concentrations diluted in a buffer containing 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 2 mM EDTA. Reactions were performed at room temperature for 20 min, and loaded onto 6% polyacrylamide/bis (37.5:1) gels in 0.56Tris-borate-EDTA (TBE) buffer, and run at a constant voltage of 130 V for 60 min. Images of the gels were acquired using a Typhoon scanner (GE Healthcare).

DNase I footprinting assays
The 100 bp (foot1) and 110 bp (foot2) katG promoter regions were amplified from genomic DNA using the primers foot1f and foot1r, and foot2f and foot2r, respectively. Both foot1f and foot2f were FITC-labeled (Table S1). The purified substrates were then subjected to the same binding reaction as in the electrophoretic mobility shift assay described above. DNase I footprinting assays were performed as previously described [32]. The ladders for foot1 and foot2 were produced using the Sanger dideoxy method using foot1f and foot2f primers, respectively.

Quantitative real-time PCR (qRT-PCR)
For real-time PCR analysis, gene-specific primers (Table S1) were used and first-strand cDNAs were synthesized using the SuperScript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Each PCR reaction (20 ml) contained 10 ml of 26 SYBR Green Master Mix Reagent (Applied Biosystems), 1.0 ml of cDNA samples, and 200 nM gene-specific primers. The reactions were performed in Bio-Rad IQ5 Real-time PCR machine. The thermocycling conditions were as follows: 95uC for 5 min, and 40 cycles at 95uC for 30 s, 60uC for 30 s and 72uC for 30 s. Amplification specificity was assessed by conducting a melting curve analysis. Gene expression levels were normalized to the levels of 16S rRNA transcripts. The degrees of change in expression were calculated using the 2 2DDCt method [33].

Bacterial growth time course assays
The antimicrobial activity of H 2 O 2 against M. smegmatis was determined using a modified bacterial growth time course assay. M. smegmatis was grown in LB at 37uC overnight. This culture was then diluted (1:100) in 5 ml of fresh LB broth containing the indicated concentration of H 2 O 2 , and the culture was again incubated at 37uC with shaking at 220 rpm for three days. Samples were taken at various time points (0, 12, 24, 36, 48, and 60 h). All assays were performed three times.   Figure S8 Assays of cell morphology by scanning electron microscopy. M. smegmatis cells prepared for scanning electron microscopy (SEM) were grown in LB at 37uC for 24 hours. After giving a heat shock at 42uC for 1 hour and incubating at 37uC for additional 4 hours, the cells were harvested by centrifugation. The bacterial pellets were then resuspended and incubated at 4uC for 24 hours in 2.5% (v/v) glutardialdehyde solution. The cells were washed twice in double-distilled water and then dehydrated with a series of 15 min treatments in 30, 50, 75, 85, 95 and 100% ethanol respectively. The final treatment in 100% ethanol was repeated to ensure complete dehydration. Samples were critical-point dried, sputter-coated with gold, and observed using a scanning electron microscope (S570; Hitachi, Tokyo, Japan). The images were taken at 15,0006 magnification (bars, 1 mm).

Author Contributions
Conceived and designed the experiments: YL Z-GH. Performed the experiments: YL. Analyzed the data: YL Z-GH. Contributed reagents/ materials/analysis tools: Z-GH. Wrote the paper: Z-GH YL.