Characterization of OxyR as a Negative Transcriptional Regulator That Represses Catalase Production in Corynebacterium diphtheriae

Corynebacterium diphtheriae and Corynebacterium glutamicum each have one gene (cat) encoding catalase. In-frame Δcat mutants of C. diphtheriae and C. glutamicum were hyper-sensitive to growth inhibition and killing by H2O2. In C. diphtheriae C7(β), both catalase activity and cat transcription decreased ∼2-fold during transition from exponential growth to early stationary phase. Prototypic OxyR in Escherichia coli senses oxidative stress and it activates katG transcription and catalase production in response to H2O2. In contrast, exposure of C. diphtheriae C7(β) to H2O2 did not stimulate transcription of cat. OxyR from C. diphtheriae and C. glutamicum have 52% similarity with E. coli OxyR and contain homologs of the two cysteine residues involved in H2O2 sensing by E. coli OxyR. In-frame ΔoxyR deletion mutants of C. diphtheriae C7(β), C. diphtheriae NCTC13129, and C. glutamicum were much more resistant than their parental wild type strains to growth inhibition by H2O2. In the C. diphtheriae C7(β) ΔoxyR mutant, cat transcripts were about 8-fold more abundant and catalase activity was about 20-fold greater than in the C7(β) wild type strain. The oxyR gene from C. diphtheriae or C. glutamicum, but not from E. coli, complemented the defect in ΔoxyR mutants of C. diphtheriae and C. glutamicum and decreased their H2O2 resistance to the level of their parental strains. Gel-mobility shift, DNaseI footprint, and primer extension assays showed that purified OxyR from C. diphtheriae C7(β) bound, in the presence or absence of DTT, to a sequence in the cat promoter region that extends from nucleotide position −55 to −10 with respect to the +1 nucleotide in the cat ORF. These results demonstrate that OxyR from C. diphtheriae or C. glutamicum functions as a transcriptional repressor of the cat gene by a mechanism that is independent of oxidative stress induced by H2O2.


Introduction
Aerobic organisms have specific mechanisms to protect themselves from reactive oxygen species (ROS) such as superoxide radical (O 2 2 ) and hydrogen peroxide (H 2 O 2 ), which can be generated by incomplete reduction of O 2 during respiration. The Fenton reaction, which generates hydroxyl radicals by reduction and oxidation of Fe ions in the presence of ROS, has been proposed as a mechanism for cell damage [1]. The ROS may cause oxidative damage to DNA, protein, and lipid [2]. Major defense systems against ROS often involve superoxide dismutase (SOD), catalase, and peroxidase, which function cooperatively to convert the ROS into O 2 and H 2 O.
Escherichia coli produces two types of catalase, both of which can detoxify H 2 O 2 by converting it to H 2 O and O 2 . The first is a bifunctional catalase hydroperoxidase I (HPI) encoded by katG that has both catalase and peroxidase activities, and the katG gene is transcriptionally activated by the positive regulator OxyR [3]. A monofunctional HPII encoded by katE has catalase activity only, and katE is activated during stationary phase [3] but is not induced by OxyR [4].
The two best characterized transcriptional regulators that mediate responses to oxidative stress in E. coli are SoxRS and OxyR. SoxR is a transcriptional activator with an iron-sulfur cluster that can be oxidized to the [2Fe-2S] state by exposure to the superoxide radical or nitric oxide. Both the reduced and the oxidized forms of SoxR can bind to its DNA target sequence [5,6], but only the oxidized form of SoxR increases the transcription of soxS. The SoxS protein then activates transcription of sodA (superoxide dismutase), zwf (glucose-6-phosphate dehydrogenase), fpr (NADPH-ferredoxin reductase), nfo (DNA repair endonuclease IV), and acrAB (an efflux pump) in E. coli [7,8].
The other major redox-sensing protein from E. coli, OxyR, functions both as a hydrogen peroxide sensor and as a transcriptional activator. E.coli OxyR, a member of the LysR family of transcription factors [9], has a helix-turn-helix (HTH) DNA binding motif in its N-terminal region and regulatory and oligomerization motifs in its C-terminal region. Oxidation of OxyR by hydrogen peroxide converts two cysteine residues (Cys199 and Cys208) to a disulfide bond, and the oxidized form of OxyR can then bind to specific DNA sequences in the regulatory regions of target genes [10]. The OxyR-activated genes in E.coli include dps (a DNA-binding protein from starved cells), gorA (GSH reductase), grxA (glutaredoxin), katG (calatase/peroxidase), ahpCF (alkyl hydroperoxide-NADPH oxidoreductase), and fur (an iron-binding repressor of iron transport) [7,11,12]. Products of genes in the OxyR regulon have many important functions in antioxidant defenses, including degradation of hydrogen peroxide by catalase and protection of DNA from oxidative attack by Dps proteins. In addition, E.coli OxyR activates production of OxyS, a small protein that regulates as many as 20 additional gene products [13].
Recent studies in Neisseria gonorrhoeae and Neisseria meningitidis indicate that OxyR can function both as a repressor of catalase gene expression in the absence of oxidative stress and as an activator of catalase gene expression in response to H 2 O 2 . In an oxyR mutant of N. gonorrhoeae, the basal level of catalase activity was greatly increased, and induction of catalase by hydrogen peroxide was abolished [14,15]. In wild-type N. meningitidis, transcription of the catalase gene was activated in response to H 2 O 2 in an OxyRdependent manner, but transcription of the catalase gene in an oxyR null mutant was at a constitutive intermediate level between the uninduced and H 2 O 2 -induced levels in the wild type strain [16].
Corynebacterium diphtheriae is gram-positive, non-spore-forming aerobic bacterium that produces diphtheria toxin and causes the severe respiratory disease diphtheria [17]. The diphtheria toxin repressor (DtxR) in C. diphtheriae represses diphtheria toxin production and iron uptake in response to high-iron conditions [18,19,20]. A C. diphtheriae C7(b) DdtxR mutant exhibited increased susceptibility to growth inhibition and killing in response to oxidative stress [21]. Only a few studies other studies of oxidative stress responses have been reported in bacteria from the genus Corynebacterium. Merkamm and Guyonvarch cloned the sodA gene from C. meiassecola and reported that superoxide dismutase played a role in cell viability [22]. The sodA gene was also cloned from C. glutamcium and shown to encode a strictly manganese-dependent form of superoxide dismutase [23].
In this study, we investigated the roles of the cat and oxyR genes of C. diphtheriae and C. glutamicum in the responses of these bacteria to H 2 O 2 -induced oxidative stress. We also characterized the effects of growth conditions on cat expression and showed that OxyR in C. diphtheriae functions as a repressor, but not as an activator, of cat by a mechanism that is independent of H 2 O 2 -induced oxidative stress.

Bacterial stains, plasmids, and growth conditions
The bacterial strains used in this study are listed in Table 1, and the plasmids used in this study are listed in Table 2. Strains of C. diphtheriae and C. glutamicum were grown at 37uC and 30uC, respectively, in heart infusion broth (Difco, Detroit, Mich.) containing 0.2% Tween 80 (designated HIBTW) or in PGT medium [24] as described for specific experiments. Low-iron PGT medium was deferrated by treatment with Chelex resin (Bio-Rad, Hercules, CA) [25], and high-iron PGT medium was made by adding 10 mM ferric chloride to the low-iron PGT medium. Exponential phase cultures of C. diphtheriae were typically harvested when A600 was between 0.8 and 1.0, and early stationary phase cultures of C. diptherheriae were typically harvested when A600 was between 6 and 7. Transconjugants of C. diphtheriae and C. glutamicum were selected on HIBTW agar plates containing 50 mg/ml of kanamycin and 30 mg/ml of nalidixic acid [21]. Strains of E. coli were grown at 37uC in Luria-Bertani (LB) medium [26] to which ampicillin (Ap), chloramphenicol (Cm), kanamycin (Km), or tetracycline (Tc) was added at a final concentration of 50, 20, 25, or 10 mg/ml, respectively, when indicated. Growth of C. diphtheriae or C. glutamicum was measured by determining A600 for culture samples appropriately diluted in 16 PBS (phosphate buffered saline) [26]. Table 1. Bacteria used in this study.

Strains
Relevant characteristics Reference or source Corynebacterium C. diphtheriae C7(b) wild-type reference strain, tox + , lysogenic for phase b [24] C. diphtheriae NCTC13129, wild-type strain lab strain C. glutamicum ATCC13032, wild-type strain lab strain Construction of in-frame deletion mutations in C. diphtheriae and C. glutamicum (i) Construction of DoxyR allele for C. diphtheriae C7(b). A 789-bp upstream DNA fragment extending from nucleotide position 2685 to 104 relative to the oxyR initiation codon was amplified from C. diphtheriae C7(b) genomic DNA by PCR using the forward primer oxyR-F1 (59-CCGCATCTGCAGTAACAG-39, with the underlined mutation to introduce the PstI site shown in bold font) and the reverse primer oxyR-R1 (59-GAAATCGAAAGCTTGG-CAGC-39, containing the HindIII site indicated in bold font). A  [27] to generate pK-DoxyR.
(ii) Construction of Dcat allele C. diphtheriae C7(b). A 1050-bp upstream DNA fragment extending from nucleotide position 21072 to +33 relative to the cat initiation codon was PCR amplified from C. diphtheriae C7(b) genomic DNA by PCR using the forward primer cat-F3 (59-GCCGCCAAAAAGCTTATC-GAG-39, with the underlined mutations to introduce the HindIII site shown in bold font) and the reverse primer cat-R1 (59-CTTGTTCAGAATTCGATCGAC-39, with the underlined mutations to introduce the EcoRI site shown in bold font). A 922-bp downstream DNA fragment extending from 222 to +900 relative to the cat stop codon was amplified from C. diphtheriae C7(b) genomic DNA by PCR using the forward primer cat-F2 (59-GTGCGAGAATTCGTGAGGCATT-39, with the underlined mutations to introduce the EcoRI site shown in bold font) and the reverse primer cat-R2 (59-GCAAGGTGTTCTAGAGGTCG-39, with the underlined mutations to introduce the XbaI site shown in bold font). The 1050 bp HindIII/EcoRI-cut upstream DNA fragment and the 922 bp EcoRI/XbaI-cut downstream fragment were cloned into HindIII/XbaI-cut pK19mobsacB [27] to generate pK-Dcat.
(v) Construction of DoxyR allele for C. glutamicum ATCC13032. A 1015-bp DNA upstream DNA fragment extending from nucleotide position 2974 to +41 relative to the oxyR initiation codon was amplified from C. glutamicum genomic DNA by PCR using the forward primer cg-oxyR-F1 (59-CGGTGAAAAGCTT-GAAGCTC-39, with the underlined mutations to introduce the HindIII site shown in bold font) and the reverse primer cg-oxyR-R1 (59-CGAAGCTGCTCGAGTGTG -39 containing the underlined mutations to introduce the XhoI site indicated in bold font). A 961bp DNA downstream DNA fragment extending from 225 to +936 relative to the oxyR stop codon was amplified with the forward primer cg-oxyR-F2 (59-GCAAAATCTCGAGGTAGCGCAG-39, containing the underlined mutations to introduce the XhoI site shown in bold font) and the reverse primer cg-oxyR-R2 (59-GTACCAGAATTCCTACCAAG-39, containing the EcoRI site shown in bold font). The 1014-bp HindIII/XhoI-cut upstream DNA fragment and 960-bp XhoI/EcoRI-cut downstream DNA fragment were cloned into HindIII/EcoRI-cut pK19mobsacB [27] to generate pK-DoxyR Cg .
(vi) Construction of Dcat allele for C. glutamicum ATCC13032. A 1036-bp upstream DNA fragment extending from nucleotide position 2964 to +72 relative to the cat initiation codon was PCR amplified from C. glutamicum genomic DNA by PCR using forward primer cg-cat-F1 (59-CTTCCATTGCTGCAGGAA-39, with the underlined mutation to introduce the PstI site shown in bold font) and cg-cat-R1-(59-GTTTCCAGAAAGCTTTGGAC-39) containing the HindIII site shown in bold font). A 1026-bp downstream flanking region extending from 234 to +992 relative to the cat stop codon was PCR amplified using the forward primer cg-cat-F2 (59-CGTCAAGAAGCTTTACCTC-39, with the underlined mutation to introduce the HindIII site shown in bold) and cg-cat-R2 (59-CCTCCAGTCTAGAATGAATC-39, with the underlined mutations to introduce the XbaI site shown in bold). The 1036-bp PstI/HindIII-cut upstream DNA fragment and the 1026-bp HindIII/XbaI-cut downstream DNA fragment were cloned into PstI/XbaI-cut pK19mobsacB [27] to generate pK-Dcat Cg .
(vii) The pK19mobsacB-derived plasmids described above were transformed into E.coli S17-1 [28] and mobilized by conjugation into C. diphtheriae C7(b), C. diphtheriae NCTC13129, or C. glutamicum ATCC13032, as appropriate. Kanamycin was used to select for co-integrates, and sucrose counter-selection was used to identify resolved co-integrates which were then screened to identify isolates carrying the desired DoxyR or Dcat alleles. Details for these procedures were described previously [29].

Agar diffusion growth inhibition assays for susceptibility to H 2 O 2
Bacteria were cultured in low-iron PGT medium or PGT medium with 10 mM FeCl 3 , as indicated, and were harvested in exponential phase (A600 between 0.8 and 1.5). Samples containing approximately 10 7 bacteria were spread on the surface of HIA agar plates. A 10 ml drop of 1 M H 2 O 2 (for C. diphtheriae strains) or 4 M H 2 O 2 (for C. glutamicum strains) was spotted onto the center of the plate, and the diameter of the growth inhibition zone was measured after overnight incubation.
Construction and use of cat::lacZ reporter plasmids NotI/XbaI-cut pK-PIM [30] was treated with the Klenow fragment of DNA polymerase and ligated with the blunt-ended 3.6-kb SmaI/XmnI DNA fragment from pSPZ, which contains the lacZ gene [31], to generate pKPL. DNA fragments containing 127bp, 227-bp, or 327-bp from the regulatory region upstream from the cat gene were amplified from chromosomal DNA of C. diphtheriae C7(b) by PCR using the forward primer cat-F5 (59-AAGCTTAGGTAATACCTAACAAAA-39, added HindIII sequence in bold font), cat-F6 (59-AAGCTTGATCGCCACAGC-CAGCC-39, added HindIII sequence in bold font), and cat-F7 (59-AAGCTTCTTGGGTGAGATCGTC-39, added HindIII sequence in bold font), respectively, and the reverse primer cat-R5 (59-CAGGATTGTCTAGACTGGGC-39, with the underlined mutations to introduce the XbaI site shown in bold font). The expected sequences of the PCR products were confirmed by DNA sequence analyses. Each HindIII/XbaI-cut PCR product was cloned, together with the SpeI/HindIII-cut 2.0 kb V DNA (Sm r / Sp r ) [32], into SpeI/XbaI-cut pKPL to generate pPL-cat100, pPL-cat200, and pPL-cat300, respectively. Each of these plasmids contains transcription and translation stops in the upstream V DNA fragment that block any fortuitous transcriptional or translational read-through from the vector DNA into cat. Plasmid pPLV contains the V DNA upstream from the promoterless lacZ gene as a negative control. C. diphtheriae C7(b) cells harboring pPLV, pPL-cat100, pPL-cat200, or pPL-cat300 were grown in lowiron PGT medium or PGT medium with 10 mM FeCl 3 . bgalactosidase activity (Miller units) was determined in samples collected during logarithmic phase growth (A600 between 0.8 and 1.5) and in early stationary phase (A600 between 6 and 7) with onitrophenyl-b-galactoside substrate as described previously [33]. The enzyme activities were calculated as A42061000/[time (min)6A6006cell volume (ml)] as described [33].

Assays for catalase activity
Spectrophotometric assays for catalase activity were performed by monitoring the rate of degradation of H 2 O 2 at A240 over 90 sec at room temperature. Each 1 ml reaction mixture contained 20 mg of crude soluble bacterial protein in 100 mM potassium phosphate, pH 7.4. The reactions were started by adding H 2 O 2 to a final concentration of 10 mM. Catalase specific activity is expressed as units corresponding to 1 mmol of H 2 O 2 decomposed/mg protein/min [34]. Staining of native polyacrylamide (12%) gels for catalase activity was performed as described previously [35]. Samples of crude bacterial extracts containing 20 mg of protein were electrophoretically separated on native polyacrylamide gels. Each gel was rinsed with distilled water and soaked for 10 min in 50 mM potassium phosphate buffer (pH 7.0) containing 10 mM H 2 O 2 . The gel was washed once again with distilled water, and then soaked for several minutes in 50 ml of a solution containing 1% ferric chloride and 1% potassium ferricyanide to allow the background color to develop. No color develops at the position(s) where catalase is present and H 2 O 2 is degraded. Protein concentrations were determined by a BCA protein assay kit (Pierce, Rockford, IL) using bovine serum albumin (BSA) as the standard.
RNA isolation and quantitative reverse-transcriptase PCR (qRT-PCR) RNA isolation was performed as described previously [21]. To study effects of H 2 O 2 on abundance of cat transcripts, C. diphtheriae C7(b) was grown in PGT medium with 10 mM FeCl 3 to mid exponential phase (A600 between 2.5 and 3.0) and treated with H 2 O 2 at a final concentration of 1 mM, 100 mM, 1 mM or 10 mM for 5 min or 30 min. To compare the abundance of cat transcripts in wild-type C7(b) and C7(b) DoxyR at different stages of growth, each bacterial strain was grown in low-iron PGT medium or PGT medium with 10 mM FeCl 3 and harvested either during exponential phase (A600 between 0.8 and 1.5) or early stationary phase (A600 between 6 and 7). Cells were suspended in acid phenol and disrupted by shaking three times for 1 min for each treatment in a bead-beater. After centrifugation, RNA was extracted from the supernatant with acidified phenol/chlorophorm, precipitated with ethanol, and treated with RNase free-DNaseI (Promega, Madison, WI). RNA concentration was measured with a NanoDrop spectrophotometer (Thermo Scientific, Wilmington, DE).
Each cDNA synthesis reaction mixture contained 1 mg of purified bacterial RNA, Cepheid Omnimix bead (Takara, Japan), specific cat primers, and a cat probe with 59 6-carboxyfluorescein and 39 black hole quencher 1 modifications. Reactions were performed in a Cepheid Smartcycler II (Cepheid, Sunnyvale, CA). Genomic DNA containing the gene of interest was used to generate a standard curve for each probe using identical experimental conditions. The abundance of the cat transcripts was normalized to the abundance dnaE transcripts for each RNA sample.

Primer extension analysis of cat transcripts
Primer extension analysis of cat transcripts was performed with reverse primer cat-PE1 (59-GTTCAGGATTGGATCGACTG-39), an anti-sense sequence that pairs approximately with codons 4-10 of the cat transcript. The reverse primer was 59-labeled with [c-32 P]ATP using T4 polynucleotide kinase (Promega, Madison, WI). Total RNA from C. diphtheriae C7(b) was purified from samples harvested in the exponential phase of growth (A600 between 0.8 and 1.5) and used as template for dideoxy-termination sequencing with the Thermo Sequenase TM cycle sequencing kit (Amersham pharmacia biotech, Cleveland, OH). The product of the primer extension reaction was analyzed on an 8.3 M urea-8% polyacrylamide sequencing gel, and the start site for the cat transcript was deduced from the experimentally determined sequence of the complementary primer extension product.
Production and purification of recombinant C. diphtheriae OxyR and E.coli OxyR For cloning into the C-terminal His6-fusion plasmid pET-22b (+) (Novagen, Madison, WI), the initiation and stop codons of the oxyR gene from C. diphtheriae C7(b) were modified to introduce NcoI and XhoI sites, respectively. The forward primer oxyR-F5 (59-ATAAATAGCCATGGGCAATA-39, with the underlined mutation in the oxyR start codon to introduce the NcoI site shown in bold font) was used in PCR with C. diphtheriae C7(b) genomic DNA as template and the reverse primer oxyR-R4 (59-TTATGCTACTC-GAGCGCGAT-39, with the underlined mutations in and near the OxyR stop codon to introduce the XhoI site shown in bold font). The NcoI/XhoI-cut PCR product was cloned into NcoI/XhoI-cut pET-22b (+) to generate pET-OxyR. Sequence analysis confirmed that the inserted fragment was in the correct reading frame and that no unexpected mutations were present in the amplified DNA.
E. coli Origami B (DE3) pLysS (Novagen, Madison, WI) containing pET-OxyR was grown in LB broth at 37uC to A600 between 0.5-0.7. Then IPTG was added at a final concentration of 0.1 mM to induce production of the C-terminally His6-tagged C. diphtheriae OxyR protein, and incubation was continued at room temperature. The bacteria were harvested 3 h after IPTG induction, disrupted by sonication, and centrifuged at 4uC to obtain cell-free supernatant. The His6-tagged fusion protein was purified from the cell lysate using TALON metal affinity resin (Clontech, Mountain View, CA) according to the manufacturer's protocol. The mass of the recombinant OxyR monomer was approximately 34 kDa, and the recombinant oligomeric OxyR protein was purified to apparent homogeneity as assessed by SDS-PAGE (12% polyacrylamide) (data not shown). The concentrations of purified OxyR used in gel-mobility shift assays and DNaseI footprinting assays were determined with the BCA protein assay kit (Pierce, Rockford, IL).
For cloning into pET-22b (+) (Novagen, Madison, WI), NdeI and XhoI sites were introduced at the initiation and stop codons of E.coli oxyR, respectively. The forward primer Eco oxyR-F3 (59-GAGGATGGCATATGAATATTC-39, with underlined mutations at the oxyR start codon to generate the NdeI site shown in bold font) was used with E. coli K12 genomic DNA as template and with the reverse primer Eco oxyR-R2 (59-TTAAACCTCGA-GAACCGCCTGTTTTA-39 containing the XhoI site overlapping the oxyR stop codon and shown in bold font) for PCR amplification of E. coli oxyR. The PCR product was digested with NdeI and XhoI and cloned into NdeI/XhoI-cut pET-22b (+) to generate pET-Eco OxyR. DNA sequence analysis confirmed the expected sequence and correct reading frame of the insert. The pET-Eco OxyR plasmid was introduced into E. coli Origami B (DE3) pLysS (Novagen. Madison, WI). This strain was grown at 37uC in LB medium, induced during exponential growth phase with IPTG, and the C-terminally His6-tagged E. coli OxyR protein was purified as described above for the comparable recombinant C. diphtheriae OxyR protein.

Gel-mobility shift assays
The 110-bp overlapping DNA probes designated (a), (b), and (c) are EcoRI fragments from the pT-cat A, pT-cat B, and pT-cat C plasmids described below. C. diphtheriae C7(b) genomic DNA was amplified by PCR with forward primer cat-F12 (59-TCAT-GAACTCTAGCACTC-39) and reverse primer cat-R9 (59-CCTATCAAATGTGAATC-39) to produce fragment (a), with forward primer cat-F13 (59-CAAAGCGGCGCGATCGCC-39) and cat-R10 (59-CAACTAAAGGTGGAGTGCT-39) to produce fragment (b), and with forward primer cat-F14 (59-TAAGCTAT-TAATCGATTC-39) and reverse primer cat-R11 (59-TGAG-GAGCTGCTTCGGTGCG-39) to produce fragment (c). Fragments (a), (b), and (c) were cloned into in the pCR2.1 TOPO vector (Invitrogen, Carlsbad, CA) to generate plasmids pT-cat A, pT-cat B, and pT-cat C, respectively, and the sequences of the cloned fragments (a), (b), and (c) were determined and shown to be identical with the corresponding regions of the genome sequence from C. diphtheriae NCTC13129. The EcoRI fragments corresponding to probes (a), (b), and (c) were excised from pT-cat A, pTcat B, and pT-cat C, purified, and labeled with [c-32 P]ATP using T4 polynucleotide kinase (Promega, Madison, WI). The DNA probes (approximately 10 4 cpm) were incubated at RT with varying amounts of purified OxyR protein from C. diphtheriae or E. coli in the binding reaction buffer reported previously [36]. When noted, 200 mM of dithiothreitol (DTT) was added to the reaction mixtures to reduce any disulfide bonds between cysteine residues in the OxyR protein. After incubation for 20-min, the reaction mixtures were analyzed using 5% non-denaturing polyacrylamide gels as described previously [36].

DNaseI footprinting assay
A 244-bp DNA fragment was amplified from C. diphtheriae C7(b) genomic DNA by PCR using the cat-F13 forward primer described previously and the cat-R12 reverse primer (59-CTTGTTCAG-GATTGGATCGAC-39) which had been 59-labeled with 32 P by prior treatment with [c-32 P]ATP and T4 polynucleotide kinase (Promega, Madison, WI). The resulting 32 P-labeled DNA probe (approximately 10 6 cpm, final concentration 0.12 nM) was incubated at RT with varying amounts of purified C. diphtheriae OxyR either treated with 200 mM DTT (reduced form of OxyR) or un-treated (air oxidized form of OxyR) in the binding reaction mixture described previously [37]. After incubation for 20-min, each reaction mixture was digested with 0.5 U of DNaseI (Promega, Madison, WI) for 2 min at 37uC, extracted with phenol/chloroform, and precipitated with ethanol. Samples of the DNaseI treated DNA were analyzed on an 8.3 M urea-6% polyacrylamide sequencing gel [36]. The nucleotide sequence was determined in a parallel reaction by the dideoxy-termination method using as template the pT-cat foot plasmid which contains the cloned 244-bp fragment used for footprinting, the 59-32 Plabeled cat-R12 reverse primer described above, and the Thermo Sequenase TM cycle sequencing kit (Amersham pharmacia biotech, Cleveland, OH).
Western immunoblot assay E.coli GC4468 or JL102 strains (harboring plasmids as indicated in the text) were grown in LB medium at 37uC and harvested during the exponential phase of growth (A600 between 0.7 and 0.9). C. diphtheriae C7(b) or C7(b) DoxyR strains (harboring plasmids as indicated in the text) were grown in PGT medium with 10 mM FeCl 3 and harvested during the exponential phase of growth (A600 between 1.0 and 1.5). Cells were disrupted using by an ultrasonic liquid process (E.coli) or a bead-beater (C. diphtheriae), and electrophoresis (SDS-PAGE, 12% polyacrylamide) and electro-blotting of proteins were performed as described previously [26]. Each blot was treated with a 1/2000 dilution of polyclonal rabbit antibody against E.coli OxyR (kindly provided by Dr. Giesela Storz) followed by a 1/10,000 dilution of goat anti-rabbit IgG (Pierce, Rockford, IL), conjugated with horseradish peroxidase (HRP), and then visualized using a SuperSignal West Dura chemiluminescent substrate detection system supplied by Pierce. Protein concentrations of extracts were determined by NanoDrop spectrophotometer (Thermo Scientific, Wilmington, DE).

Results
The role of catalase in resistance to H 2 O 2 -induced oxidative stress in C. diphtheriae C. diphtheriae is a catalase-positive gram-positive bacterium. As a baseline for studying the role of catalase in resistance of C. diphtheriae to H 2 O 2 -induced oxidative stress, we determined the effects of H 2 O 2 on growth (Fig. 1A) and viability (Fig. 1B) of wild type C. diphtheriae C7(b). For bacteria growing in PGT medium with 10 mM FeCl 3 , adding H 2 O 2 at a final concentration of 10 mM during the exponential phase of growth did not affect either growth rate or bacterial viability after 6 hours of exposure. Increasing the H 2 O 2 concentration to 100 mM caused a lag of approximately 2 hours before resumption of normal growth but had little effect on viability after 6 hours of exposure. In contrast, when H 2 O 2 was added at 300 mM, the wild-type cells showed complete growth arrest and no viable cells were present after 6 hours exposure. We observed similar effects of H 2 O 2 on growth and viability of wild type C. diphtheriae C7(b) when the experiments were repeated in low-iron PGT medium or in HIBTW medium (data not shown). These results show that wild type C. diphtheriae C7(b) exhibits substantial tolerance to H 2 O 2 , presumably by degrading H 2 O 2 to non-toxic products.
The most obvious candidate enzymes for detoxifying H 2 O 2 are catalases and peroxidases. A search of the published C. diphtheriae NCTC13129 genome sequence for homologs of catalase identified a single gene annotated as cat (DIP0281) (http://www.ncbi.nlm. nih.gov/gene/2649075). The cat gene ORF is separated by a 127bp intervening sequence from the divergently oriented sigC ORF, which encodes an RNA polymerase sigma factor presumed to be involved in responses to environmental stress ( Fig. 2A). We constructed an in-frame deletion in the cat gene of C. diphtheriae C7(b) by allelic exchange, and we used this Dcat mutant strain to examine the role of catalase in the resistance C. diphtheriae to H 2 O 2 . In contrast with the results described previously with wild type C7(b), addition of H 2 O 2 at 10 mM, 100 mM or 300 mM to exponential phase cultures of the C7(b) Dcat mutant resulted in immediate cessation of growth and complete loss of viability after 6 hours of exposure (Figs. 1A and 1B). When methyl viologen (MV, a superoxide radical generating agent) was spotted on agar plates inoculated with sufficient numbers of wild type C. diphtheriae C7(b) or the C7(b) Dcat mutant to produce confluent lawns during subsequent incubation, no significant differences were seen in the diameters of the zones of inhibition that developed (data not shown). These results indicate that catalase confers resistance to H 2 O 2 -induced stress, but not to MV-induced stress, in C. diphtheriae.

Effects of growth phase and H 2 O 2 on activity of the cat promoter and catalase
As an initial method to localize the cat promoter and measure its activity, we constructed and characterized a set of cat::lacZ transcriptional fusion reporter plasmids ( Fig. 2A). Plasmids pPL-cat100, pPL-cat200, and pPL-cat300 contain, respectively, the 100bp segment of the sigC-cat intergenic sequence immediately upstream from the cat ORF and the longer 200-bp and 300-bp DNA sequences which extend further upstream from the cat ORF. For C. diphtheriae C7(b) strains harboring either pPL-cat100, pPL-cat200, or pPL-cat300, the b-galactosidase reporter activity was significantly greater (approximately 1.8-fold) during the exponential phase of growth than during early stationary phase (P,.05) (Fig. 2B). The control plasmid pPLV, which does not contain any of the sigC-cat intergenic sequence, showed no b-galactosidase reporter activity. These data show that the functional cat promoter is located within the 100-bp segment of the sigC-cat intergenic sequence immediately upstream from the cat ORF and that the cat promoter is slightly but significantly more active during exponential growth than in stationary phase.
We measured catalase in exponential phase and early stationary phase cultures of wild type C. diphtheriae C7(b) by preparing extracts, subjecting them to electrophoresis on non-denaturing polyacrylamide gels, and performing in gel assays for catalase activity (Fig. 2C). The catalase from wild-type C7(b) migrated as a single band on the non-denaturing polyacrylamide gels and appeared, from the relative intensity of the reaction signals, to have ,2-3 fold greater activity in the bacteria harvested during exponential phase compared to the bacteria harvested in the early stationary phase. Additional assays showed that catalase from C. diphtheriae does not have peroxidase activity and that catalase activity did not differ significantly in bacteria grown under high- iron vs. low-iron conditions (data not shown). Parallel controls performed with extracts from the C7(b) Dcat mutant failed to demonstrate any catalase activity. These results indicate that the cat gene encodes the only catalase in C. diphtheriae C7(b) and that catalase is most likely regulated at the transcriptional level in C. diphtheriae C7(b).
As a first step in investigating whether the cat promoter in C. diphtheriae is activated in response to H 2 O 2 , we grew replicate cultures of wild-type C7(b) harboring pPL-cat200 in PGT medium with 10 mM FeCl 3 , and we compared b-galactosidase activity in individual cultures at various times after adding 10 mM or 100 mM H 2 O 2 vs. a control culture without added H 2 O 2 (Fig. 3A).
In the H 2 O 2 -treated cultures, activity of the b-galactosidase reporter for the cat promoter showed a trend toward slight and progressive decrease over time, but none of the differences between the treated cultures and the untreated control cultures was statistically significant (P.0.05). To confirm and extend these findings, we used qRT-PCR to measure directly the relative abundance of cat transcripts in untreated control cultures and in cultures exposed for 5 or 30 min to H 2 O 2 at concentrations of 1 mM, 100 mM, 1 mM and 10 mM (Fig. 3B). After 5 minutes, the relative abundance of cat transcripts decreased slightly in all of the H 2 O 2 -treated cultures, but the differences in cat transcript abundance between the treated and untreated cultures were statistically significant only for the 100 mM and 1 mM H 2 O 2 treatment groups (p,0.05). After 30 min, the relative abundance of cat transcripts in treated cultures at each concentration of H 2 O 2 did not differ significantly from the untreated controls (p.0.5).
In gel assays for catalase activity were performed on extracts prepared from control cultures of wild-type C. diphtheriae C7(b) during the exponential growth phase and from replicate cultures exposed for 30 minutes to H 2 O 2 at 1 mM, 100 mM, 1 mM or 10 mM (Fig. 3C). By image analysis, catalase activity in the cultures treated with 1 mM, 100 mM, 1 mM and 10 mM H 2 O 2 was estimated to be 87%, 85%, 92% and 96% of the catalase activity in the untreated control cultures, respectively. Extending the duration of H 2 O 2 exposure to 2 h also failed to induce catalase activity (data not shown). In summary, treating C. diphtheriae C7(b) with H 2 O 2 at concentrations from 1 mM to 10 mM resulted in insignificant to small but significant decreases in cat promoter activity and slight decreases in catalase activity. These findings differed dramatically from published results with E coli showing that exposure to H 2 O 2 at concentrations ranging from 100 mM to 1 mM activates transcription of katG and production of catalase [38]. Our results provide no evidence that cat transcription or catalase production in C. diphtheriae is activated by exposure to H 2 O 2 .
The role of OxyR in regulating the cat promoter and catalase in C. diphtheriae Examination of the published genome sequence for C. diphtheriae NCTC13129 [39] identified one ORF (DIP1421) that encodes a putative transcriptional regulator similar to the H 2 O 2 -inducible OxyR from Erwinia carotovora. The predicted amino acid sequence of the C. diphtheriae OxyR encoded by DIP1421 is 100% identical with the putative C. glutamicum OxyR, 40% identical and 62% similar with Streptomyces coelicolor OxyR, 33% identical and 62% similar with E.coli OxyR, and 27% identical and 47% similar with N. gonorrhoeae OxyR (Fig. S1A). Each of these proteins has a highly conserved DNA binding domain characteristic of the LysR family of regulatory proteins as well as two conserved cysteine residues (Cys206 and Cys215 in C. diphtheriae OxyR) homologous to the Cys199 and Cys208 residues in E.coli OxyR that participate in redox sensing. No gene encoding a homolog of the E. coli OxyS protein is present in the C. diphtheriae genome. The oxyR gene in C. diphtheriae is flanked on the upstream side by the divergently transcribed dirA gene (which encodes an iron-repressible polypeptide) and on the downstream side by the convergently transcribed DIP1422 gene (which encodes a putative membrane protein) (Fig.  S1B).
We constructed an in-frame deletion mutation in the oxyR gene of C. diphtheriae C7(b) by allelic exchange (Fig. S1B). We examined susceptibility of the C7(b) DoxyR mutant to oxidative stress by comparing the effects on growth of H 2 O 2 at final concentrations of 10 mM, 100 mM, and 300 mM in replicate exponential phase cultures vs. control cultures without exposure to H 2 O 2 . In striking contrast with the findings for wild type C7(b) shown previously (Fig. 1A), the C7(b) DoxyR mutant grew as well after exposure to 10 mM or 100 mM H 2 O 2 as it did without exposure to H 2 O 2 , and it showed only a slight decrease in growth after exposure to 300 mM H 2 O 2 (Fig. 4A). The tolerance of the C7(b) DoxyR mutant to H 2 O 2 was comparable in low-iron PGT medium, PGT medium with 10 mM FeCl 3 , and HIBTW medium (data not shown). Introducing a cloned wild type oxyR allele into the C7(b) DoxyR mutant by integrating the pK-PIM-oxyR C7(b) plasmid into a chromosomal attB site [30] complemented the DoxyR mutation and restored the growth phenotype after exposure to H 2 O 2 to wild type (data not shown).
We used two different methods to demonstrate the effects of OxyR on activity of the cat promoter. First, we introduced the pPL-cat100, pPL-cat200, or pPL-cat300 reporter plasmid (or the pPLV control plasmid) into wild type C. diphtheriae C7(b) and the C7(b) DoxyR mutant and measured b-galactosidase activity as a surrogate for cat promoter activity during exponential growth of each strain in PGT medium with 10 mM FeCl 3 (Fig. 4B). Each of the C7(b) DoxyR derivatives exhibited a statistically significant 6-8 fold increase in b-galactosidase activity compared with the corresponding isogenic wild-type C7(b) derivative (P,0.05), and no b-galactosidase activity was present in C7(b) or C7(b) DoxyR harboring pPLV. To confirm and extend these findings, we used real-time quantitative RT-PCR assays to compare the relative abundance of cat transcripts in wild type C7(b) and the isogenic C7(b) DoxyR mutant during exponential growth (Fig. 4C) and early stationary phase (Fig. 4D), both in low-iron PGT medium and in PGT medium with 10 mM FeCl 3 . The normalized abundance of cat transcripts was significantly greater by 6-8 fold in C7(b) DoxyR than in C7(b) under each set of growth conditions (P,0.05). For each strain, the normalized abundance of cat transcripts declined by two-fold or more during transition from the exponential growth phase to early stationary phase, confirming our previous finding using reporter plasmids in wild type C7(b) (Fig. 2B). Furthermore, in C7(b) and C7(b) DoxyR during exponential and early stationary phase growth, the normalized abundance of cat transcripts exhibited a small (two-fold or less) but significant (P,0.05) increase under high-iron vs. low-iron conditions. We performed direct biochemical assays for catalase activity with extracts prepared from exponential phase cultures of C. diphtheriae C7(b), C7(b) Dcat, and C7(b) DoxyR grown in low-iron PGT medium and in PGT medium with 10 mM FeCl 3 (Fig. 5A). No catalase activity was detected in C7(b) Dcat, confirming that cat is the sole gene responsible for production of catalase in C. diphtheriae C7(b). The specific activity of catalase was approximately 15-to 20-fold greater in C7(b) DoxyR than in wild type C7(b), but growth in low-iron vs. high-iron medium had no significant effect on catalase activity (P.0.05). Analysis of these extracts by non-denaturing PAGE and in gel assays for catalase activity supported these conclusions (Fig. 5B). Taken together, these findings demonstrate that OxyR functions as a repressor of cat transcription, support the conclusion that regulation of catalase occurs primarily at the level of transcription in C. diphtheriae, and provide no evidence that OxyR functions as an activator of cat transcription in C. diphtheriae.
Both in wild type C7(b) and in C7(b) DoxyR, we found small but statistically significant (P,0.05) increases in the normalized abundance of cat transcripts under high iron vs. low-iron growth conditions (Figs. 4C and 4D) that were not accompanied by increases in catalase activity (Figs. 5A and 5B). These effects were independent of OxyR, since they occurred both in C7(b) and in C7(b) DoxyR.
We also showed by real-time quantitative RT-PCR assays that the relative abundance of sigC transcripts was about 1.6-fold greater in C7(b) DoxyR than in C7(b) during exponential phase growth in low-iron PGT medium, and about 2-fold greater in C7(b) DoxyR than in C7(b) during exponential phase growth in high-iron PGT medium (data not shown). These modest differences suggest that OxyR has little role in regulating transcription of sigC, notwithstanding the fact that cat and sigC are separated only by a short intergenic region and are divergently transcribed ( Fig. 2A).

Molecular analysis of OxyR interaction with the cat promoter in C. diphtheriae
Our previous experiments using pPL-cat100, pPL-cat200, and pPL-cat300 as reporters for cat promoter activity (Figs. 2B and Fig. 4B) showed that the cat promoter is located within the 100 bp segment of the sigC-cat intergenic region that lies immediately upstream from the cat ORF start codon. In an effort to predict a possible OxyR binding target by bio-informatic methods, we used the MEME program (http://meme.sdsc.edu./meme/) to search the sigC-cat intergenic region for homologs of the known 50 bp E.coli OxyR binding target (59-CAATAATAAGCAAATCCTT-TAATTGTACAATTATTGTTAAGAATTACCTA-39) [40]. The best match, extending from 52-103 bp upstream from the cat ORF start codon, was 75% similar to the E.coli OxyR binding target. As a first step toward testing whether C. diphtheriae OxyR can bind to this predicted site, we designed three overlapping 110 bp DNA probes for use in gel mobility shift assays. Probe (a), extending from nucleotide 2113 to 222 with respect to the +1 nucleotide of the cat ORF start codon, was centered on the predicted binding site described above (Fig. 6A). Probe (b), extending from nucleotide 2211 to 2102, and probe (c), extending from nucleotide 251 to +59, lie immediately upstream and immediately downstream, respectively, from the predicted binding site described above (Fig. 6A). Gel mobility shift assays showed that purified C. diphtheriae OxyR bound to and shifted probe (a) but not probe (b) or probe (c) (Fig. 6B). The position of the shifted probe (a) did not change as the concentration of C. diphtheriae OxyR (calculated as monomer) added to the gel shift reaction mixture increased from 56 pM to 280 pM, suggesting that OxyR formed complexes with DNA of uniform stoichiometry within this range of experimental conditions. Purified C. diphtheriae OxyR shifted probe (a) to the same extent and to the same position whether or not 200 mM DTT was present in the gel shift reaction mixture. These experiments were performed on three different occasions, with the same results. These findings suggest that purified C. diphtheriae OxyR can bind to the cat promoter region either in its reduced or oxidized form.
Next, we mapped the 59 end of the cat transcripts in total RNA isolated from exponential phase cultures of wild-type C. diphtheriae C7(b) and the C7(b) DoxyR mutant grown in low-iron PGT medium and in PGT medium with 10 mM FeCl 3 (Fig. 7A). In both bacterial strains and in both growth media, the 59 end of the cat transcript was the C residue at position 239 relative to the +1 position in the cat ORF start codon. The intensity of the band corresponding to the primer extension product was greater with RNA from C7(b) DoxyR than from wild type C7(b), reflecting the greater abundance of cat transcripts in C7(b) DoxyR (Fig. 4C).
To define more precisely the binding target for C. diphtheriae OxyR in the cat promoter region, we performed DNaseI footprinting assays. Either in the presence or absence of 200 mM DTT, binding of OxyR protected a 46-bp AT-rich sequence extending from position 255 to 210 with respect to the +1 position in the cat ORF start codon (Fig. 7B). The extent of protection increased progressively as the concentration of OxyR in the reaction mixture increased from 0.28 nM to 2.8 nM. The OxyR binding site demonstrated by the DNaseI protection assays was immediately downstream from, but did not overlap with, the putative binding site predicted by the bio-informatics analysis described above based on homology with the binding site for E. coli OxyR.
The 46 bp sequence corresponding to the DNaseI footprint is shown in bold font in Fig. 7C. Both the G residue at position 222 that corresponds to the 39 end of probe (a) and the T residue at position 251 that corresponds to the 59 end of probe (c) lie within this DNaseI footprint. The upstream and downstream vertical lines inserted into the sequence in Fig. 7C show the location of the 59 end of probe (c) and the 39 end of probe (a), respectively. The ability of C. diphtheriae OxyR to shift the mobility of probe (a) (Fig. 6B) indicates that the 13 nucleotides at the 39 end of the DNaseI footprint sequence are not required for OxyR binding to the cat promoter region, although they might contribute to optimal binding. Conversely, the failure of C. diphtheriae OxyR to shift the mobility of probe (c) (Fig. 6B) suggests that some or all of the 4 nucleotides at the 59 end of the DNaseI footprint sequence might be required for OxyR binding. To investigate this possibility further, we used PCR to prepare a nested set of seven DNA amplicons that shared the downstream end of probe (c) at position +59 with respect to the relative to the +1 position in the cat ORF start codon but differed at their upstream ends. The shortest amplicon was identical to probe (c), and the six longer amplicons extended 2 bp, 4 bp, 9 bp, 14 bp, 19 bp or 29 bp further upstream to terminate at positions 253, 255, 260, 265, 270 or 280, respectively. In gel mobility shift assays, C. diphtheriae OxyR failed to shift the mobility of the amplicon that was identical to probe (c) but did shift the mobility of all of the longer amplicons (data not shown). We conclude, therefore, that the sequence extending downstream from the T residue at position 253 in Fig. 7C contains the minimal essential binding site for C. diphtheriae OxyR.
Comparison of Dcat and DoxyR mutations in C. diphtheriae and C. glutamicum C. diphtheriae C7(b) has been maintained in laboratories since the 1950s [24] and has been used by many investigators as a reference strain for research. To determine whether our results with C7(b) are representative of other isolates of C. diphtheriae or of another Corynebacterium species, we performed confirmatory studies with the recent clinical isolate C. diphtheriae NCTC13129 (used for the first genome sequence of this species) and with C. glutamicum ATCC13120 (a nonpathogenic Corynebacterium widely used in biotechnology). For each of these reference strains, we constructed an isogenic Dcat and an isogenic DoxyR in-frame single deletion mutant, and we complemented the DoxyR mutant by integration of pK-PIM-oxyR C7(b) into a chromosomal attB site. We then compared the newly constructed wild type, mutant and complemented-mutant variants of C. diphtheriae NCTC13129 and C. glutamicum ATCC13120 with the homologous C7(b) variants described previously. For this purpose, we used agar-diffusion growth-inhibition assays to test the tolerance of each of these bacterial variants to H 2 O 2 . Fig. 8A shows photographs of representative plates from tests with wild type C7(b), with the isogenic Dcat and DoxyR mutants, and with the DoxyR mutant complemented with pK-PIM-oxyR C7(b) . Fig. 8B presents the results from quantitative measurements of the diameters of the growth inhibition zones from tests performed with the wild type C. diphtheriae C7(b), C. diphtheriae NCTC13129 and C. glutamicum ATCC13120 strains and the mutant and complemented-mutant strains derived from them. In all cases, the Dcat mutants exhibited increased susceptibility to H 2 O 2 , the DoxyR mutants showed dramatically decreased susceptibility to H 2 O 2 , and the susceptibility to H 2 O 2 of the complemented DoxyR mutants was comparable to that of their wild type parental strains. Complementation tests with pK-PIM-oxyR plasmids carrying the wild type oxyR allele from C. diphtheriae NCTC13129 or C. glutamicum ATCC13120 gave equivalent results to those described above with the pK-PIM-oxyR C7(b) plasmid (data not shown). These findings support the conclusions that, in both C. diphtheriae and C. glutamicum, catalase protects against H 2 O 2 -induced oxidative stress, oxyR functions as a repressor of cat, and oxyR clones from C. diphtheriae and C. glutamicum are expressed in a comparable manner.

C. diphtheriae OxyR and E.coli OxyR are not functionally equivalent
We used agar-diffusion growth-inhibition assays and complementation tests similar to those described above to evaluate the ability of the E. coli oxyR gene to substitute functionally for the C. diphtheriae oxyR gene. For tests in E. coli we used the wild type strain GC4468 and its isogenic oxyR mutant strain JL102 [41]. Preliminary control experiments on LB agar plates confirmed that the oxyR mutant JL102 exhibited greater susceptibility to growth inhibition by H 2 O 2 than the parental strain GC4468 (Fig. 9A) as described previously [41]. For complementation studies in E. coli, we cloned the E. coli oxyR gene with its native promoter region from GC4468 genomic DNA into pRK415 (a broad-host range, low copy number plasmid [42]) to generate pRK-oxyR Eco. We also constructed pRK-oxyR Eco2, which fuses a 500-bp oxyR promoter fragment from C. diphtheriae C7(b) to the oxyR ORF from E. coli GC4468, thereby replacing the native E. coli oxyR promoter region with the heterologous C. diphtheriae oxyR promoter region. Introducing either pRK-oxyR Eco or pRK-oxyR Eco2 into E. coli JL102 complemented its oxyR defect and restored its H 2 O 2 susceptibility phenotype to wild type (Fig. 9A). Therefore, the structural gene for E. coli oxyR is functional both in the pRK-oxyR Eco clone and pRK-oxyR Eco2 clone, and biologically active E. coli OxyR protein can be produced in E. coli when the oxyR structural gene is expressed either under control of its native promoter or the heterologous oxyR promoter from C. diphheriae C7(b).
Next, we subcloned the native and hybrid oxyR loci from pRK-oxyR Eco and pRK-oxyR Eco2 into pK-PIM to generate pK-PIM-oxyR Eco and pK-PIM-oxyR Eco2, respectively. When we introduced each of these plasmids into C. diphtheriae C7(b) DoxyR, neither was able to complement the DoxyR mutant phenotype, in contrast to the positive complementation result obtained with the pK-PIM-oxyR C7(b) control (Fig. 9B). Finally, we performed Western blots with a rabbit antibody against E. coli OxyR to investigate the amounts of E. coli OxyR protein produced in the E. coli and C. diphtheriae strains described above (Fig. 9C). OxyR was present in wild type E. coli GC4468 but not in the isogenic oxyR mutant JL102. The amounts of OxyR in extracts from E. coli JL102 harboring pRK-oxyR Eco or pRK-oxyR Eco2 were slightly greater than in wild type GC4468. Antibody to the E. coli OxyR protein failed to detect C. diphtheriae OxyR in the Western blots with extracts from C7(b) or C7(b) DoxyR. E. coli OxyR was detected in extracts from C. diphtheriae C7(b) DoxyR harboring pK-PIM-oxyR Eco or pK-PIM-oxyR Eco2, and the amount of E. coli OxyR made in C7(b) DoxyR harboring pK-PIM-oxyR Eco2 was significantly greater that the amount detected in E. coli JL102 harboring pRK-oxyR Eco or pRK-oxyR Eco2. Taken together, these results demonstrated that E. coli OxyR encoded by a functional native or hybrid oxyR locus was produced in C. diphtheriae but failed to complement the DoxyR phenotype or substitute for C. diphtheriae OxyR as a repressor of transcription of the cat gene.
Finally, we used gel mobility shift assays to test directly the ability of purified OxyR from C. diphtheriae and purified OxyR from E. coli to bind to target sequences in DNA probe (a) (Fig. 6A) from the cat promoter region of C. diphtheriae (Fig. 10A) and to a 300 bp DNA fragment from the katG regulatory region of E. coli (Fig. 10B). The purified OxyR from C. diphtheriae bound to and shifted the mobility of probe (a), in agreement with results shown previously in Fig. 6B, but OxyR from E. coli failed to shift probe (a) either with or without 200 mM DTT in the reaction mix.
Conversely, purified OxyR from E. coli bound the 300 bp E. coli target sequence and shifted its mobility progressively as the OxyR concentration increased, but no shift was seen with purified OxyR from C. diphtheriae either with or without 200 mM DTT in the reaction mix. These findings show that the purified OxyR proteins from C. diphtheriae and E. coli exhibit significant differences in their DNA binding specificities.

Discussion
There are few previous reports of enzymes and regulators involved in responses to oxidative stress in the human pathogen C. diphtheriae or the closely related soil bacterium C. glutamicum. The sodA gene that encodes the manganese-containing superoxide dismutase, an enzyme that scavenges superoxide radicals and protects against superoxide radical-induced oxidative damage, was cloned from C. glutamicum and characterized [23]. A previous study from our laboratory showed that insertional inactivation of the diphtheria toxin repressor gene (dtxR) in C. diphtheriae results in increased susceptibility to growth inhibition and killing following   [31]. In addition, exposure of C. diphtheriae to several stressors, including acid, cold, heat, ethanol or SDS, but not exposure to H 2 O 2 , increased transcription of the contiguous sigB and dtxR genes from the promoter located just upstream from sigB [21]. These studies provided initial evidence for metabolic interactions between dtxR, the iron-regulon and several nonrelated stress responses in C. diphtheriae.
In the current study, we chose two C. diphtheriae reference strains and one C. glutamicum reference strain for studies on the role of catalase in resistance to H 2 O 2 and the role of OxyR, a member of the LysR family of bacterial regulatory proteins, in regulation of catalase (cat) gene expression. Published genome sequences for C. diphtheriae and C. glutamicum have a single gene for catalase [23,43] and we confirmed production of a single catalase enzyme in C. diphtheriae by biochemical analysis (Figs. 2C and 5B.). We showed both by use of cat::lacZ reporter plasmids (Figs. 2B and 4B) and by direct quantitative RT-PCR assays of cat transcript abundance that OxyR strongly represses cat transcription in C. diphtheriae by an H 2 O 2 -independent mechanism (Figs. 4C and 4D). Increased production of catalase by DoxyR mutants of C. diphtheriae and C. glutamicum conferred a dramatically increased level of resistance to H 2 O 2 (Figs. 8A and 8B). Transcription of cat in C. diphtheriae decreased by about 2-fold during transition from the exponential growth phase to stationary phase (Figs. 2B, 4C and 4D). . OxyR from E. coli does not complement the DoxyR mutation in C. diphtheriae. A: The phenotypes of wild type E. coli strain GC4468 and its isogenic DoxyR derivative JL102 in H 2 O 2 agar-diffusion growth-inhibition assays are shown. The mutant phenotype of E. coli JL102 was complemented by the cloned E. coli oxyR gene expressed either from its native E. coli oxyR promoter (in pRK-oxyR Eco) or from the heterologous C. diphtheriae oxyR promoter (in pRK-oxyR Eco2). B: The phenotypes of wild type C. diphtheriae C7(b) containing pK-PIM and its isogenic DoxyR derivative containing the pK-PIM are shown. C. diphtheriae OxyR expressed from its native promoter in the pK-PIM-oxyR C7(b) clone complemented the mutant phenotype of C7(b) DoxyR. In contrast, E. coli OxyR expressed either from its native promoter in pK-PIM-oxyR Eco or from the heterologous C. diphtheriae oxyR promoter in pK-PIM-oxyR Eco2 failed to complement the mutant phenotype of C7(b) DoxyR. C: SDS-PAGE and Western blot analysis with rabbit antiserum against E.coli OxyR was performed on extracts from exponential phase cultures of the following bacterial stains: Transcription of cat in C. diphtheriae was slightly greater under highiron growth conditions than under low-iron growth conditions, but the difference was statistically significant (Figs. 4C and 4D). We believe it is unlikely that DtxR directly mediates this effect of iron on transcription of cat, because the sigC-cat intergenic region contains no sequences that closely match the 19 bp consensus sequence for DtxR-binding. Molecular characterization of this effect of iron on cat transcription will require additional experiments that are beyond the scope of the current study. We did not identify other environmental signals that regulated cat transcription. We found no evidence that exposure of C. diphtheriae to H 2 O 2 stimulates transcription of cat. In contrast, transcription of the catalase gene (katG) in E.coli during exponential growth is activated by OxyR in response to H 2 O 2 , and katG transcription during stationary phase is controlled by the starvation-induced sigma factor RpoS [44]. This mechanism of regulating cat transcription during stationary phase appears unlikely in C. diphtheriae or C. glutamicum, because their genomes do not encode rpoS [23,43].
Primer extension analysis showed that a (C) residue located 39 nucleotides upstream from the cat ORF is the transcriptional start for the cat transcript (Figs. 7A and 7C). A recent study characterized the sigma factors of C. glutamicum and features of the specific promoters that they recognize [45]. This analysis of 159 sequences of presumed s A -dependent promoters in C. glutamicum revealed a consensus sequence of TANAAT for the 210 hexamer, GNTANANTNG for the extended 210 region, and TTG(A/C)CA for the 235 hexamer in the subset of the corynebacterial promoters for which a 235 sequence was identified. Inspection of the sequence upstream from the cat transcriptional start site (Fig. 7C) revealed a putative 210 promoter region (TAAGCT, which contains all three of the highly conserved T, A and T residues in the first, second and sixth positions of the 210 consensus sequence). Furthermore, this putative 210 hexamer is properly located with respect to the cat transcriptional start site and is also separated by the optimal 17 nucleotides from a putative 235 promoter region (TTGCTA, which matches the consensus 235 sequence at five of six positions). The 210 region has the features of a corynebacterial 210 core hexamer but not of a corynebacterial extended 210 region. In addition, a consensus Shine-Dalgarno ribosome binding site (AGGAGG, labeled S-D in Fig. 7C) is present at the proper location upstream from the cat ORF. Although future studies will be needed to confirm the functional importance of each of these features for cat promoter activity, their presence provides a high degree of confidence that this region does represent the cat promoter.
DNaseI footprinting assays identified a 46 bp region protected by OxyR binding that extends from 255 to 210 nucleotides with respect to the +1 nucleotide in the cat ORF. This putative OxyR binding site overlaps with the putative 210 hexamer of the cat promoter, the start site for the cat transcript, and part of the putative ribosome binding site. Binding to this region is fully consistent with repression of cat transcription by OxyR, and a previous study in Pseudomonas putida showed that the LysR-type regulator CatR represses transcription of the catR gene by binding to a region located downstream from the start site of the catR transcript and overlaps with the ribosome binding site [46].
The OxyR proteins from C. diphtheriae and C. glutamicum are members of the LysR family of transcriptional regulators. A characteristic feature of binding sites for LysR-type bacterial regulatory proteins is the presence of one or more highly conserved T-N11-A motifs, each of which may serve as the core for an inverted repeat structure of variable complexity [47]. Examination of the sequence of the OxyR binding site identified by DNaseI footprinting of the cat promoter region of C. diphtheriae reveals three putative T-N11-A motifs (Fig. 7C). The upstream putative T-N11-A motif (marked with filled diamonds above the conserved T and A residues at positions 253 and 241 with respect to the +1 nucleotide in the cat ORF) completely overlaps the putative 210 hexamer of the cat. The two partially overlapping downstream putative T-N11-A motifs (marked with filled circles above the conserved T and A residues at positions 236 and 224 and with bold asterisks above the conserved T and A residues at positions 225 and 213) are located between the cat transcript start site and the putative ribosome binding site. The horizontal dashed arrows in Fig. 7C mark short palindromic sequences that are associated with, but not always precisely centered on, these putative T-N11-A motifs. Future studies will be required to test directly the functional roles of each of these putative T-N11-A motifs and their associated palindromic sequences in repression of cat transcription by C. diphtheriae OxyR.
We used complementation tests and gel mobility shift assays to compare the function of the OxyR regulator from C. diphtheriae with the well-characterized OxyR regulator from E. coli, which was previously shown to act both as a sensor of H 2 O 2 -induced oxidative stress and as transcriptional activator of the catalase gene (katG) in E. coli [10]. The cloned oxyR gene from C. diphtheriae (which encodes a protein identical to C. glutamicum OxyR) fully complemented the DoxyR mutation in C. diphtheriae or C. glutamicum (Figs. 8A and 8B). In contrast, the cloned E. coli oxyR gene was shown to be functional in E. coli and to direct the production of the identical E. coli OxyR protein in C. diphtheriae (Fig. 9). These findings indicate that the OxyR proteins from C. diphtheriae and E. coli have distinct functional specificities, and differences in their ability to interact with OxyR binding sequences from C. diphtheriae and E. coli were confirmed by gel mobility shift assays (Figs. 10A and 10B).
Bacterial regulatory proteins in the LysR family typically form oligomers. The biologically active form of LysR regulators, including the prototypical OxyR from E. coli, is most often tetrameric, although active octamers or dimers have been described in some cases [48,49]. Fig. S1 shows that primary structure and domain organization of OxyR from C. diphtheriae and C. glutamicum are homologous with those of other, better characterized, OxyR proteins from a diverse group of bacterial species. We performed limited experiments to examine the oligomeric state of the purified recombinant His6-tagged that we prepared in this study. On a gel filtration chromatography column standardized with several other reference proteins, our C. diphtheriae OxyR migrated as a single peak with a mobility consistent with the predicted mass of a trimer (data not shown), but we could not rule out an anomalous mobility because of weak interactions with the gel matrix or other unknown factors. We also made limited but unsuccessful attempts to determine the oligomeric state of C. diphtheriae OxyR by several protein crosslinking methods (data not show). Therefore, for this study, we expressed the concentration of C. diphtheriae OxyR used in individual experiments as monomer equivalents, and additional future studies will be needed establish definitively the oligomeric state(s) of biologically active C. diphtheriae OxyR.
In summary, our results demonstrate that catalase is the effector for defense against H 2 O 2 -induced oxidative stress in C. diphtheriae and C. glutamicum. They demonstrate that OxyR functions in an H 2 O 2 -independent manner as a repressor of cat transcription in C. diphtheriae, and they provide initial insights into the organization of the cat promoter and the molecular mechanisms for its interaction with OxyR in C. diphtheriae. Many questions about the role of OxyR in corynebacteria remain to be examined. For example, does corynebacterial OxyR function as a global transcriptional regulator, and if so what promoters (other than the cat promoter) does it repress or activate in corynebacteria? Does corynebacterial OxyR function as a sensor of oxidative stress in corynebacteria, and if so what promoters are preferentially regulated by, and what molecular mechanisms determine the preferential activity of, oxidized or reduced forms of corynebacterial OxyR. Does the regulatory activity of corynebacterial OxyR at some promoters involve interactions with small molecule co-regulators, and if so what is the nature and functional role of such co-regulators. The initial characterization of corynebacterial OxyR reported here should serve as a sound basis for investigating such questions in future studies. Figure S1 OxyR sequence and organization of the oxyR locus in C. diphtheriae. A: Alignment of predicted amino acid sequences of OxyR from Corynebacterium diphtheriae, Corynebacterium glutamicum, Streptomyces coelicolor, Escherichia coli, and Neisseria gonorrhoeae, assembled using the Clustal W program. The predicted helix-turn-helix motif (HTH) near the amino terminus is indicated by the dotted line; the predicted DNA binding domain lies between the two bent arrows; and the conserved cysteine residues represented by C206 and C215 in OxyR from C. diphtheriae or C. glutamicum are indicated by asterisks. The numbering of amino acids is shown on the right side. B: Organization of the oxyR locus in wild type C. diphtheriae C7(b) and in the isogenic C7(b) DoxyR mutant. The grey arrow shows the orientation of the oxyR ORF, and the white arrows show the orientations of the flanking dirA ORF (which encodes an iron-repressible polypeptide) and the DIP1422 ORF (which encodes a putative membrane protein). The ORF for the DoxyR allele, which contains an in-frame deletion, is shorter and is shown to scale in the lower diagram. (TIF)