Oxidative Stress Response and Characterization of the oxyR-ahpC and furA-katG Loci in Mycobacterium marinum

Abstract

Oxidative stress response in pathogenic mycobacteria is believed to be of significance for host-pathogen interactions at various stages of infection. It also plays a role in determining the intrinsic susceptibility to isoniazid in mycobacterial species. In this work, we characterized the oxyR-ahpC and furA-katG loci in the nontuberculous pathogen Mycobacterium marinum. In contrast to Mycobacterium smegmatis and like Mycobacterium tuberculosis and Mycobacterium leprae, M. marinum was shown to possess a closely linked and divergently oriented equivalents of the regulator of peroxide stress response oxyR and its subordinate gene ahpC, encoding a homolog of alkyl hydroperoxide reductase. Purified mycobacterial OxyR was found to bind to the oxyR-ahpC promoter region from M. marinum and additional mycobacterial species. Mobility shift DNA binding analyses using OxyR binding sites from several mycobacteria and a panel of in vitro-generated mutants validated the proposed consensus mycobacterial recognition sequence. M. marinum AhpC levels detected by immunoblotting, were increased upon treatment with H₂O₂, in keeping with the presence of a functional OxyR and its binding site within the promoter region of ahpC. In contrast, OxyR did not bind to the sequences upstream of the katG structural gene, and katG expression did not follow the pattern seen with ahpC. Instead, a new open reading frame encoding a homolog of the ferric uptake regulator Fur was identified immediately upstream of katG in M. marinum. The furA-katG linkage and arrangement are ubiquitous in mycobacteria, suggesting the presence of additional regulators of oxidative stress response and potentially explaining the observed differences in ahpC and katG expression. Collectively, these findings broaden our understanding of oxidative stress response in mycobacteria. They also suggest that M. marinum will be useful as a model system for studying the role of oxidative stress response in mycobacterial physiology, intracellular survival, and other host-pathogen interactions associated with mycobacterial diseases.

Oxidative stress response and protection against reactive oxygen intermediates and reactive nitrogen intermediates have been implicated in the intracellular survival of pathogenic mycobacteria and their persistence in the host (5, 17, 20, 21, 25, 26, 46). In addition, several elements of oxidative stress response have been implicated in the innate susceptibility (9, 11) and acquired resistance (27, 53) to the front-line antituberculosis drug isonicotinic acid hydrazide (isoniazid). Recently, we have addressed the regulation of oxidative stress response in the primary mycobacterial pathogens, i.e., Mycobacterium tuberculosis and Mycobacterium leprae (10, 11, 13, 15, 37), with the rationale that a delineation of such processes may improve our understanding of host-pathogen interactions in mycobacterial disease (11). Unexpectedly, the oxyR gene, which is the mycobacterial equivalent of the central regulator of oxidative stress response in Escherichia coli, was found to be inactivated in M. tuberculosis via multiple mutations (Fig. 1A) (10, 11, 37). The alterations in oxyR are conserved in all contemporary strains of M. tuberculosis and other members of the M. tuberculosis complex (10, 11, 40), with only a single polymorphism recorded thus far among nine distinct lesions (39). The loss of M. tuberculosis oxyR appears to be related to the altered expression (15) of the closely linked and divergently transcribed ahpC gene (Fig. 1A) (10, 37, 47), encoding a homolog of alkyl hydroperoxide reductase (6, 24). In other bacteria, this antioxidant system plays a role in reducing organic peroxides (4, 24) and detoxifies targets particularly sensitive to peroxide-mediated damage, such as lipids and nucleic acids (24). The loss of oxyR in M. tuberculosis appears counterintuitive, since the tubercle bacillus is most likely subjected to oxidative damage encountered in the host phagocytic cells and inflammatory sites in addition to the endogenous oxidative metabolism of the bacterium. Surprisingly, the elimination of oxyR function is not the only lesion in oxidative stress response genes of the primary mycobacterial pathogens. It has recently been reported that M. leprae has multiple mutations in the catalase-peroxidase gene katG (18, 28) (Fig. 1B).

FIG. 1.

Genetic organization of the oxyR-ahpC and furA-katG loci in mycobacteria. (A) The genes oxyR (open boxes) and ahpC (shaded boxes) are tightly linked and divergently transcribed (arrows) in the majority of mycobacterial species with the exception of M. smegmatis (line indicates that the corresponding region upstream of ahpC has been sequenced and characterized but that no oxyR has been identified in this organism). In M. tuberculosis, oxyR has been inactivated via multiple, naturally occurring mutations (filled balloons, nonsense and frameshift mutations; open balloons, deletions). (B) Linkage of furA (encoding a homolog of the ferric uptake regulator Fur) and katG in mycobacteria. The furA and katG genes are cotranscribed in M. tuberculosis. In M. leprae, both furA and katG are inactivated via multiple mutations (balloons, insertions; triangles, deletions).

The apparent selective inactivation of parts of the oxidative stress response in two major mycobacterial pathogens, M. tuberculosis and M. leprae, suggests that these phenomena may be related to some, less obvious aspects of host-pathogen interactions during infection (11). Unfortunately, direct analyses of these phenomena in M. tuberculosis and M. leprae are precluded by the facts that M. leprae cannot be grown in vitro (50) and all strains of M. tuberculosis examined to date lack a functional oxyR (10, 40). When genetic analyses of M. tuberculosis or M. leprae are not practical or possible, it has been a tradition in mycobacterial research to resort to surrogate systems. Among these, Mycobacterium smegmatis has become very popular due to its rapid growth and relative ease of genetic manipulation (23). Unfortunately, this organism, albeit displaying a vigorous oxidative stress response (15), does not have the typical mycobacterial arrangement of oxyR-ahpC genes and, moreover, lacks a detectable homolog of mycobacterial oxyR (15) (Fig. 1A). This prompted us to explore other mycobacterial species as potential model systems to investigate the role of oxyR and other elements of oxidative stress response in M. tuberculosis and M. leprae. Here, we extended our studies to Mycobacterium marinum, a nontuberculous-disease-causing species (19). M. marinum is phylogenetically close to M. tuberculosis (32), and the two organisms appear to share at least some properties in the context of intracellular survival and infection (29, 31, 45). For example, both M. marinum (3) and M. tuberculosis (7, 8, 12, 41, 42, 44, 45, 48) avoid late endosomal/lysosomal compartments when phagocytosed by macrophages. M. marinum can also cause chronic granulomatous infection in poikilothermic animal models (31). Like tuberculosis, M. marinum infections flare up upon induction of immunosuppression (31).

Here we initiated analyses of oxidative stress response systems in M. marinum. We show that this organism possesses functional elements of oxidative stress response that closely resemble in organization those in M. tuberculosis and M. leprae. Furthermore, M. marinum has an intact oxyR gene and an inducible ahpC. We also examined whether various parts of mycobacterial oxidative stress response are coordinately regulated. In contrast to the situation in enteric bacteria, where several members of the peroxide stress response are coordinately regulated by OxyR, ahpC and katG appear to be differentially controlled in M. marinum. While OxyR binds to the ahpC promoter, it does not associate with the sequences upstream of katG. Instead, a putative novel regulator, furA, encoding a homolog of the ferric uptake regulator Fur, is located immediately upstream of the M. marinum katG gene, an arrangement which appears to be ubiquitous among mycobacterial species.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

M. marinum ATCC 15069, an isolate from a patient with an infected foot, was obtained from the American Type Culture Collection (ATCC). Mycobacterium bovis BCG Pasteur (ATCC 27291), Mycobacterium intracellulare (ATCC 13950), M. tuberculosis H37Rv (ATCC 27294), and Mycobacterium xenopi (ATCC 19250) were from the ATCC. M. smegmatis mc²155 (ahpC⁺) and VD1865-6 (mc²155 ahpC::Km^r) have been described elsewhere (38, 52). Mycobacteria were grown in Middlebrook 7H9 medium or on 7H10 plates supplemented with albumin, dextrose, catalase, and 0.05% Tween 80. Media were supplemented with kanamycin (10 μg/ml) when necessary. E. coli was grown in LB supplemented with ampicillin (100 μg/ml), kanamycin (25 μg/ml), and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal; 40 μg/ml) when required. All incubations were at 37°C. Plasmid pMthis₁₀- ahpC (14) was used to overproduce and purify M. tuberculosis His₁₀-tagged AhpC. Plasmid phsp60-gfp (14) carrying the gene for green fluorescent protein (GFP) was used to transform M. marinum for fluorescence microscopy.

Recombinant DNA techniques, genetic methods, and sequence analysis.

Chromosomal DNA isolation, Southern blotting, E. coli transformation, cloning procedures, PCR amplification, and DNA and protein sequencing were based on standard methods (2, 23). Electroporation of M. marinum was carried out as previously described for M. smegmatis (23).

Purification of M. tuberculosis AhpC and antibody production.

M. tuberculosis AhpC was overproduced and purified as a hybrid protein with a His₁₀ tag, using procedures developed for the previously reported purification of OxyR (13). Briefly, an overnight culture (1 ml) of E. coli harboring plasmid pMthis₁₀ahpC was inoculated into 200 ml of LB and grown to an optical density at 600 nm of 0.4. Isopropyl-β-d-thiogalactopyranoside was added to 1 mM, incubation at 37°C was continued for 2 h, and bacteria were harvested by centrifugation at 3,000 rpm. Cells were lysed in a French press; after fractionation by centrifugation, the AhpC-enriched pellet was resuspended in homogenization buffer (13). His₁₀-AhpC was mixed with Ni-nitrilotriacetic acid resin (Qiagen) and packed into a column, and the bound protein was eluted with elution buffer (200 mM imidazole, 8 M urea, 100 mM sodium phosphate, 10 mM Tris-HCl [pH 7.6]). The eluted protein was analyzed by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE), and the fractions enriched in AhpC were pooled. The purified product was digested with cyanogen bromide and subjected to amino acid sequence analysis (15 cycles) as previously described (13). The amino acid sequence of the peptide fragments obtained conformed with the predicted sequence of M. tuberculosis AhpC. Rabbit antisera against M. tuberculosis His₁₀-AhpC were generated by using standard immunological procedures.

Exposure of mycobacteria to hydrogen peroxide and immunoblot analyses.

M. marinum and M. xenopi were grown to an optical density at 600 nm of 0.4 and aliquoted into 50-ml portions, and H₂O₂ (Sigma) was added to final concentrations of 0.02, 0.2, 2, and 20 mM. Cells were incubated in a 37°C shaker for 2 h. Crude protein extracts were obtained by homogenization in a Mini Bead-beater (Biospec Products) for 2 min. The zirconia beads (0.1 mm; Biospec Products) and cell debris were removed by centrifugation, and the resulting supernatants were used for immunoblot analysis. Aliquots of 10 or 20 μg of total protein (10 μg in Fig. 3A; 20 μg in Fig. 3B and C) from M. smegmatis, M. marinum, and M. xenopi were separated on SDS–11% polyacrylamide gels, and proteins were transferred to Immobilon-P membranes (Millipore) by electroblotting. Western blot analysis was performed with rabbit antiserum to M. tuberculosis AhpC or antiserum to M. tuberculosis KatG (from Clifton Barry). Goat anti-rabbit immunoglobulin G conjugated to peroxidase (Kirkegaard & Perry Laboratories) was used as the secondary antibody. Bound antibody was visualized by formation of diaminobenzidine precipitate.

FIG. 3.

AhpC and KatG levels in M. marinum and M. xenopi exposed to H₂O₂. (A) Western blot of cell extracts (10 μg of protein) from M. marinum and M. smegmatis with an antibody raised against M. tuberculosis AhpC. Lanes: 1, purified His₁₀-AhpC from M. tuberculosis (M. t.); 2, M. marinum (M.m.); 3, M. smegmatis mutant strain VD1865-6 (M.s. ahpC::Km^r); 4, M. smegmatis parent strain mc²155 (M.s. ahpC⁺). (B) Analysis of ahpC expression in M. marinum and M. xenopi treated with H₂O₂. Exponentially growing cultures were exposed to various concentrations of H₂O₂ (0.02 to 20 mM) for 2 h. After treatment, crude protein extracts (20 μg) were separated by SDS-PAGE and probed with anti-M. tuberculosis-AhpC antibody. Steady-state levels of AhpC increased after treatment with 2 mM H₂O₂. (C) Western blot analysis of the effects of exposure to H₂O₂ on M. marinum and M. xenopi katG expression. The samples were identical to those shown in panel B except that the blot was probed with KatG antibody.

Cloning of M. marinum ahpC-oxyR.

The oxyR-ahpC intergenic regions from M. marinum and M. xenopi were obtained by using oligonucleotides Ahp4 (specific for M. tuberculosis ahpC) and OxyRZoo1 (degenerate primer based on conserved regions of mycobacterial oxyR) (10) and cloned into plasmid pCR2.1 (Invitrogen). The cloned PCR products were characterized by sequencing, and a 515-bp SalI-EcoRI fragment from one positive clone containing the intergenic oxyR-ahpC region was used as a hybridization probe. Southern blot analysis of M. marinum genomic DNA showed hybridization to an approximately 2-kb SmaI fragment and a 3-kb BamHI fragment (data not shown). A partial genomic library was constructed by eluting SmaI- or BamHI-digested DNA (a 1.6- to 2.3-kb region for SmaI digests and a 3- to 4-kb region for BamHI digests) from agarose gels and cloning it into the respective sites of pBluescript(SK). Transformants were grown in pools of 12 to 15 (total of 200 independent clones for each digest), and plasmid DNA was extracted and subjected to Southern blot analysis using the same probe. After identification of positive pools, the pools were separated into individual clones and subjected to another round of hybridizations, resulting in the identification of one positive SmaI and one positive BamHI clone containing M. marinum ahpC and oxyR genomic sequences. Further characterization and sequencing of one of these clones (plasmid pEPR124) showed that they contained the entire M. marinum ahpC gene. However, only a partial oxyR sequence (the 5′ end) was present in this clone. To obtain the complete M. marinum oxyR, another Southern blot analysis was performed with SacII-digested M. marinum genomic DNA and a probe specific for the 3′ end of oxyR. This probe was obtained by PCR using primers Mox5 (5′ATCCGGTTCGGCATCCCC3′; positions +307 to +327 relative to the oxyR initiation codon) and Mox6 (5′GCAACTCGGACAGTGCCG3′; positions +581 to +598 relative to the oxyR initiation codon). After applying the same strategy as described for isolation of the complete ahpC gene, we obtained a clone (plasmid pEPR32) with a 1-kb SacII fragment containing the 3′ end of M. marinum oxyR. This clone was used to complete the sequence of M. marinum oxyR.

DNA mobility shift assays and site-directed mutagenesis of OxyR binding site.

Mobility shift assays with purified M. leprae OxyR were performed as described elsewhere (15). DNA fragments used in the binding assays were made by PCR amplification of the ahpC-oxyR intergenic regions from different mycobacterial species. The following oligonucleotides were used: Ahp4 (5′GGTGAAGTAGTCGCCGGGCT3′) and Mox4R (5′ACGAACGCGCGCAACCCG3′) for M. marinum; Ahp4 and Xox4R (5′AGAGCGCTTGCGGCGCTG3′) for M. xenopi; Ahp4 and OxyRTB7 (15) for the wild-type M. tuberculosis fragment (M.t.); TBOxyAT (5′CTAGCACCTCTTATCGGCGATGCCGATAAA3′) for the mutant fragment M.t.* (underlined nucleotides highlight changes made to the sequence); AhpML7 and OxyRML14 (13) for the wild-type M. leprae fragment (M.l.); AhpML7 and LoxyGC (5′GATGGTGGGCTGATAACTCTTAGCACTCATACCGCTAAG3′) for M. leprae mutant fragment M.l.1; LoxyGA (5′GATGGTGGGCTGATAACTCTTAGCACTCATACCGATAAG3′) for M.l.2; and LoxyTC (5′GATGGTGGGCTGATAACTCTTATCACTCATACCGCTAAG3′) for M.l.3. The 251-bp SacI-HincII fragment containing the M. intracellulare ahpC-oxyR intergenic region was from a PCR product obtained by amplification with the primers Ahp4 and OxyRZoo1 (10, 11, 13, 15, 37).

DNase I footprinting analysis.

Probes for DNase I footprinting were generated by PCR using primers MmfootEco (5′TATTGAATTCACATAACTCTCCTC3′) and MmfootBam (5′TATTGGATCCGTGCCGCCAACGC3′) and end labeled as described previously (2). OxyR-DNA binding reaction mixtures were identical to those used for gel mobility shift assays. After the binding, MgCl₂ was added to a final concentration of 5 mM followed by DNase I digestion. Reactions were terminated by addition of 40 mM EDTA, and reaction products were separated on a nondenaturing polyacrylamide gel (as described for DNA mobility shift assay). Protein-bound DNA and unbound DNA bands were excised from the gel, eluted with a solution containing 0.5 M ammonium acetate, 0.1% SDS, and 1 mM EDTA, phenol extracted, and precipitated with ethanol in the presence of 20 μg of glycogen. DNA was resuspended in formamide buffer and run on an 8% denaturing polyacrylamide gel along the side of a sequencing ladder generated by using the MmfootEco primer.

Cloning of M. marinum furA.

Inverse PCR was used to clone the region upstream of M. marinum katG based on the partial sequence of the 5′ end of M. marinum katG (38a). For the inverse PCR, M. marinum genomic DNA (2 μg) was digested to completion with KpnI. The digested DNA was purified by using Qiaex II kit (Qiagen) and self-ligated in a 20-μl reaction mixture. PCR was carried out with 5 μl of the ligation reaction mixture and primers specific for the known portion of M. marinum katG (and a small 5′ region upstream of katG) sequence (38a), MMAL3 (5′CCTCGTCGACAACGAAGCCAT3′) and MMBL4 (5′GCCACTACGGTGGCCTGTTCA3′). A single band corresponding to a 1.6-kb PCR product was cloned into pCR2.1 (plasmid pMMfurA1.6) and sequenced to obtain the complete nucleotide sequence of M. marinum furA.

Southern blotting with katG, ideR, furA, and fur.

M. tuberculosis katG, ideR, furA, and fur were used to probe PstI- and SacII-digested genomic DNA from M. marinum. Southern blot hybridization was performed under medium-stringency conditions (37°C for hybridization in 900 mM NaCl, 50% formamide, 1% SDS, 200 μg of salmon sperm DNA per ml; 37°C for posthybridization washes in 0.1× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate]–0.1% SDS). The probes were generated by PCR amplification using oligonucleotides specific for the M. tuberculosis genes: KatG1 (5′GGAGGTCGCGACCATCGA3′) and KatG2 (5′TCATGGCCATGCGCCGAA3′) for katG; Mtb-fur1 (5′GCTCATCGGAACATACGAAGG3′) and Mtb-fur3 (5′TTCCTTCCAGGAGTTGGTGTT3′) for furA; FurB1 (5′TTGGTGCTGGAGACGGGC3′) and FurB2 (5′ACATTGTGCTCGACGCCG3′) for fur; and IdeR1 (5′ATGGAGGGTGCCATATGAACGAGTTG3′) and IdeR2 (5′AACAACTCGGAATTCGACTGTCCGC3′) (designed based on the previously described TB1 and TB2 primers) for ideR (35). The identity of each PCR fragment was confirmed by partial DNA sequencing.

Cell culture, infection, and epifluorescence microscopy.

Murine macrophage-like J774 cell line ATCC TIB-67 was cultured in Dulbecco’s modified Eagle’s medium (DMEM; Bio Whittaker) supplemented with 4 mM l-glutamine (Bio Whittaker) and 5% fetal bovine serum (HyClone) at 37°C in humidified air containing 5% CO₂. M. bovis BCG, M. smegmatis mc²155, and M. marinum ATCC 15069 were used to infect J774 cells. Single-cell suspensions were produced as previously described (14). Briefly, bacterial cultures were pelleted and resuspended in DMEM, homogenized in a glass Tenbroeck homogenizer (20 strokes), and sonicated for 2 min in a water bath sonicator (Astrason, Farmingdale, N.Y.). Bacterial aggregates were further removed by low-speed centrifugation. Single-cell suspensions were verified by microscopy. The multiplicity of infection was 10 bacilli per macrophage, and bacterial uptake was allowed to take place for 1 h at 37°C. Extracellular bacteria were removed by washing with warm phosphate-buffered saline, and DMEM supplemented with gentamicin (100 μg/ml; Sigma) was added to kill the extracellular bacteria. The infected macrophage monolayers were incubated for various periods (1 to 168 h), after which cells were scraped and lysed in the presence of 0.1% Tween and bacteria were plated on 7H10 plates to determine CFU. Each value represents the mean CFU (±standard error) from at least three independent experiments. For epifluorescence analysis, J774 cells were allowed to adhere to no. 1 thickness, 12-mm-diameter glass coverslips in 24-well tissue culture plates (Costar) at a density of 2 × 10⁵ cells per coverslip. After infection with M. marinum harboring phsp60-gfp and elimination of extracellular bacilli as described above, macrophage monolayers were mounted on glass slides with PermaFluor (Lipshaw Immunon). Infected J774 cells were examined by fluorescence microscopy in an Olympus BX60 microscope as previously described (14, 43).

Nucleotide sequence accession numbers.

The sequences reported here have been deposited in GenBank with the following accession numbers: (i) AF034861 for M. marinum ahpC and oxyR; (ii) AF038027 for M. marinum furA; and (iii) U43810 for M. xenopi ahpC and oxyR (intergenic region).

RESULTS AND DISCUSSION

M. marinum survival in macrophages is intermediate between that of M. bovis BCG and M. smegmatis.

Since the host macrophage is the primary source of exogenous reactive oxygen intermediates in vivo, we first tested suitability of the human M. marinum isolate used in this work by examining its ability to infect and persist in macrophages. GFP-labeled M. marinum ATCC 15069 (see Materials and Methods) efficiently infected macrophages (Fig. 2A). M. marinum ATCC 15069 displayed persistence in J774 cells over extended periods of time (Fig. 2B) and survived better than M. smegmatis in J774 cells. However, the intracellular environment appeared slightly more inhibitory for M. marinum than for M. bovis BCG (Fig. 2B). The appreciable persistence of M. marinum ATCC 15069 in mammalian macrophages at temperatures usually associated with infections caused by the human pathogens M. tuberculosis and M. leprae confirmed its suitability for further studies.

FIG. 2.

Survival of M. marinum ATCC 15069 in the J774 murine macrophage cell line. (A) Epifluorescence microscopy image of macrophages infected with GFP-expressing M. marinum ATCC 15069. (B) Comparison of the survival of M. bovis BCG (⧫), M. marinum (Mm; ▴), and M. smegmatis mc² 155 (Ms; ■) recovered from J774 cells over time. Each value represents the mean CFU (±standard error) from at least three independent experiments. Infection, incubation (37°C), and other techniques are described in Materials and Methods.

Detection of AhpC in M. marinum.

With respect to their production of AhpC, mycobacteria fall into three categories (15) represented by the following species: (i) M. smegmatis, which has a relatively high baseline level of AhpC production; (ii) M. tuberculosis, which has altered levels of AhpC (15), readily detectable on Western blots only upon overproduction following promoter mutations or other genetic manipulations (15, 36, 52); and (iii) Mycobacterium aurum, which has no detectable AhpC and appears to lack the corresponding gene (15). To test whether an equivalent of M. tuberculosis AhpC can be detected in M. marinum, we used antibodies raised against purified M. tuberculosis protein. To purify AhpC and generate the antibodies, the previously cloned and characterized M. tuberculosis ahpC gene (10) was placed behind T7 transcription and translation signals and overexpressed in E. coli as a His₁₀-AhpC fusion product. The protein product was purified on a Ni-nitrilotriacetic acid column, and the identity of the purified polypeptide as His₁₀-tagged AhpC was confirmed by amino acid sequence analysis. Next, polyclonal antibodies were raised (see Materials and Methods) and demonstrated to recognize M. tuberculosis and M. smegmatis AhpC (Fig. 3A, lanes 1 and 4). The specificity of the antibody for AhpC was further confirmed by the absence of the band corresponding to AhpC in the ahpC mutant M. smegmatis VD1865 (52) (Fig. 3A, lane 3). The anti-AhpC serum was then used to examine M. marinum ATCC 15069 protein extracts for the presence of AhpC. The results of these experiments revealed the presence of a 25-kDa polypeptide (Fig. 3A, lane 2) indistinguishable by its electrophoretic mobility from the previously characterized M. smegmatis AhpC (15, 52). Another slowly growing nontuberculous mycobacterial pathogen (19), M. xenopi, was included in these studies and also displayed detectable levels of AhpC (Fig. 3).

Characterization of the oxyR and ahpC genes in M. marinum and M. xenopi.

The ahpC gene is tightly linked and divergently transcribed from oxyR in M. leprae and M. tuberculosis (Fig. 1A). In other species, such as M. smegmatis, oxyR is not detected anywhere on the chromosome. To examine the situation in M. marinum and M. xenopi, a set of degenerate primers based on highly conserved regions of the known mycobacterial oxyR and ahpC genes (10) was used to amplify the putative oxyR-ahpC region in these organisms. The resulting PCR fragments from M. marinum and M. xenopi were subjected to sequence analysis, and the presence of both oxyR and ahpC in a divergent arrangement was confirmed (data not shown; GenBank accession no. AF034861 and AF038027). Next, the M. marinum PCR fragment containing a partial sequence of oxyR and ahpC was used as a hybridization probe to clone additional genomic fragments to complete the characterization of the oxyR and ahpC genes from this organism (see Materials and Methods). The complete nucleotide sequence of the ahpC and oxyR genes from M. marinum was determined (GenBank accession no. AF034861). The predicted translated product of M. marinum ahpC showed 87 to 91% identity and 93 to 95% overall similarity with the M. tuberculosis, M. avium, and M. leprae ahpC gene products (data not shown), and the predicted product of M. marinum oxyR showed 72 to 75% identity and 88 to 91% overall similarity with the M. leprae and M. avium oxyR gene products (Fig. 4). These results indicate that in contrast to M. smegmatis, M. marinum has a complete oxyR gene, closely linked to and divergently transcribed from ahpC, resembling the prototypical arrangement observed in the primary mycobacterial pathogens M. tuberculosis and M. leprae.

FIG. 4.

Alignment of OxyR sequences from M. marinum (M.m.) (this work), M. avium (M.a.) (37), and M. leprae (M.l.) (10). Asterisks, identical amino acids; periods, conserved amino acid substitutions.

Binding of OxyR to the oxyR-ahpC intergenic region of M. marinum.

The availability of a purified mycobacterial OxyR (13) and the cloning of the intergenic oxyR-ahpC region from M. marinum presented an opportunity to examine whether OxyR binds to the putative ahpC-oxyR promoter region in this organism. To carry out these analyses, DNA fragments containing the oxyR-ahpC intergenic region from M. marinum were subjected to electrophoretic DNA mobility shift assays using purified M. leprae OxyR. The results of these studies are illustrated in Fig. 5A. OxyR was able to bind to the M. marinum oxyR-ahpC intergenic region in a sequence-specific fashion, as judged by the ability of specific competitor DNA (unlabeled DNA fragment identical to the probe used in the binding reaction) to reduce the amount of radiolabeled probe in the bound state (Fig. 5A). Equivalent amounts of nonspecific competitor DNA were not able to reduce the formation of OxyR-DNA complexes (Fig. 5A, lane 4).

FIG. 5.

Binding of M. leprae OxyR to the oxyR-ahpC intergenic region of M. marinum, M. xenopi, and M. intracellulare and the consensus mycobacterial OxyR binding sequence. Purified M. leprae OxyR (see Materials and Methods) was incubated with ³²P-labeled DNA fragments containing the oxyR-ahpC intergenic region from M. marinum (−120 to +50 relative to the oxyR start codon) (A), M. xenopi (−193 to +96) (B), and M. intracellulare (−108 to +193) (C). Protein-DNA complexes (open triangles) were separated from unbound probes (filled triangles) by electrophoresis on a 4% native polyacrylamide gel and analyzed by autoradiography. The specificity of the binding was tested by competition assays using specific and nonspecific competitor DNAs in the reactions. Lanes: 1, radiolabeled probe alone; 2, probe incubated with His₁₀-OxyR; 3, same as lane 2 plus 0.5 μg of cold specific competitor (unlabeled DNA identical to the radiolabeled probe); 4, same as lane 2 plus 0.5 μg of cold nonspecific competitor DNA (a 324-bp fragment from the M. bovis BCG ahpC structural gene). (D) Consensus sequence of the mycobacterial OxyR binding site within the oxyR-ahpC region (Myc.) and OxyR sequences from M. leprae (M.l.), M. tuberculosis (M.t.), M. marinum (M.m.), M. xenopi (M.x.), M. avium (M.a.), and M. intracellulare (M.i.). The sequence exhibits twofold dyad symmetry (ATC-N₉-GAT; bold letters) and contains the T-N₁₁-A core motif typical of recognition sequences of the LysR-type transcriptional regulators (34). Init., initiation.

OxyR binds to a site within the M. leprae and M. tuberculosis oxyR-ahpC promoter region which has been delimited to a 30-bp region centered 65 bp upstream of the ahpC mRNA start site P₁ (13). This was also found to be the case in M. marinum (data not shown). The availability of the sequences from the oxyR-ahpC intergenic regions of M. marinum and M. xenopi enlarged the pool of mycobacterial sequences corresponding to the OxyR binding site (Fig. 5D). The corresponding sequence including the core palindrome ATC-N₉-GAT, which contains the T-N₁₁-A motif characteristic of the recognition sequences of the LysR-type transcriptional regulators (34), displays a twofold dyad symmetry. The proposed recognition sequences in M. marinum and M. xenopi are located at positions similar to those in M. leprae and M. tuberculosis, and they also contain the T-N₁₁-A motif (Fig. 5D). The results of DNase I footprinting analyses when 50% of the DNA probe was in the bound state (Fig. 6) suggested the presence of protected bases coinciding with the ATC and GAT half-sites (underlined nucleotides correspond to dots in Fig. 6) demarcated by residues displaying hypersensitivity to DNase I. However, the results of the DNase I footprinting analyses should be interpreted with caution due to difficulties in obtaining full protection of the OxyR binding site.

FIG. 6.

DNase I footprinting analysis of OxyR contacts with the recognition sequence. OxyR was bound to a probe containing the OxyR binding site within the M. marinum ahpC-oxyR intergenic region and subjected to DNase I footprinting as described in Materials and Methods. Lanes: OxyR +, OxyR-bound probe; OxyR −, free probe; G, A, T, and C, sequencing ladder generated with the primer used to produce the probe (see Materials and Methods). Circles, protected bases; asterisks, hypersensitive sites. The nucleotide positions of the protected and hypersensitive sites are indicated on the right; bold letters highlight the core of the proposed OxyR binding recognition sequence.

Mutational analysis of the mycobacterial OxyR recognition sequence.

OxyR binds to the oxyR-ahpC intergenic regions from M. leprae and M. marinum with higher affinity than to the corresponding region from M. tuberculosis (13) (Fig. 7A). M. xenopi (Fig. 5B) and M. intracellulare (Fig. 5C) also displayed a relatively tight association with OxyR, suggesting that the lower affinity seen in M. tuberculosis is an exception consistent with the proposal (13) that the M. tuberculosis oxyR binding site contains mutations reducing its affinity for OxyR. Furthermore, the lower affinity of the M. tuberculosis binding site for M. leprae OxyR could not be satisfactorily explained by the divergence within the first 66 amino acid residues of OxyR in different mycobacteria (Fig. 4), which in LysR family members represent the DNA binding domain (34), since the variances relative to M. leprae OxyR were quite similar in all cases and yet OxyR binding was strong in all species tested except M. tuberculosis.

FIG. 7.

Mutational analysis of the OxyR recognition sequence within the oxyR-ahpC region. DNA fragments containing mutations (listed at the bottom; introduced by site-specific mutagenesis as described in Materials and Methods) were subjected to DNA binding mobility shift assays. Open triangles, DNA-OxyR complex; closed arrows, unbound probe. Lanes: 1 and 4, radiolabeled probe alone; 2, 3, 5, and 6, probe incubated with His₁₀-OxyR. Lanes 2 and 3 and lanes 5 and 6 represent duplicate samples. The M.t. sequence is proposed to contain a natural mutation of the left half site. Quantitation of DNA-OxyR complexes by densitometric analysis: (A) lanes 2 and 3, 2% probe bound; lanes 5 and 6, 60% probe bound; (B) lanes 2 and 3, 80% probe bound; lanes 5 and 6, binding below detection limit; (C) lanes 2 and 3, 1% probe bound; lanes 5 and 6, 1% DNA-OxyR probe bound.

To address the possibility that the OxyR binding site is mutated in M. tuberculosis, we carried out site-specific mutagenesis of the M. tuberculosis sequence, which deviates from the consensus. Figure 7 lists the mutations tested in these experiments. Mobility shift assays revealed a significant increase in DNA-protein complex formation (Fig. 7A) for the mutant M.t.* site compared to the wild-type M.t. site (60% versus 2% of DNA in the bound complex, respectively, as determined by densitometry). This result suggests the importance of both half-sites within the palindromic motif ATC-N₉-GAT, explaining in part the lower affinity of the native M. tuberculosis OxyR binding site within the oxyR-ahpC intergenic region. In a converse experiment, the corresponding M. leprae sequence was modified. Three mutated sites departing from the conserved T-N₁₁-A motif were tested (Fig. 7B and C): M.l.1, with both the T and A changed, and M.l.2 and M.l.3, with either the T or the A changed one at a time. In these experiments (Fig. 7B and C), OxyR binding to the wild-type M. leprae site M.l. resulted in 80% of the probe being in the bound state, while binding of OxyR to the double mutant M.l.1 was completely abrogated (Fig. 7B, lanes 5 and 6). When only one of the nucleotides (T or A) of the T-N₁₁-A motif was changed (M.l.2 and M.l.3), a residual weak binding was observed, with only 1% of the probe localized in the protein-DNA complex (Fig. 7C). These results confirm the hypothesis that the T-N₁₁-A motif is essential for OxyR recognition in mycobacteria and strongly suggest that the naturally occurring deviations from the consensus in M. tuberculosis are the cause of its reduced affinity for OxyR. This finding indicates an added complexity of any attempts to analyze M. tuberculosis by using a heterologous functional OxyR and underscores the need for a model system such as M. marinum, where both the oxyR gene and the OxyR target binding sites are functional.

Differential expression of ahpC and katG in M. marinum.

In enteric bacteria, OxyR controls expression of a set of genes, including ahpC and the catalase-peroxidase gene katG, which are induced in response to peroxide challenge. To investigate whether M. marinum can mount a response to oxidative stress, we monitored intracellular levels of ahpC and katG gene products in response to treatment with H₂O₂. Exponential-phase M. marinum and M. xenopi cultures were incubated with H₂O₂ at concentrations ranging from 0.02 to 20 mM, and protein extracts were analyzed by Western blotting. Treatment of M. marinum and M. xenopi with hydrogen peroxide revealed that steady-state levels of AhpC can be altered compared to the basal levels in untreated cells, with an apparent peak of induction at 2 mM H₂O₂ (Fig. 3B). The levels of AhpC could be increased in response to peroxide stress in M. marinum and M. xenopi consistent with the presence of a functional oxyR gene in these organisms, although additional studies are needed to establish that the increase of ahpC expression is oxyR dependent. The maximum level observed was at concentrations of H₂O₂ much higher than needed for induction of ahpC in M. smegmatis (62.5 to 125 μM) (15). The reasons for this are not known. However, there are precedents for the requirement of relatively high peroxide concentrations for induction of specific oxidative stress response genes. Transcription of two members of the E. coli oxyR regulon, dps (encoding a nonspecific DNA binding protein) and oxyS (encoding a nontranslated RNA with proposed regulatory functions), is induced by treatment with 0.2 to 2 mM H₂O₂ with a peak at 2 mM (1). However, we cannot exclude the possibility that additional putative differences between M. marinum and M. smegmatis, e.g., properties of the cell wall envelope (51), may account for the differences in H₂O₂ concentrations needed to elevate AhpC production.

Identical protein extracts were tested for KatG levels by Western blotting using antibodies against M. tuberculosis KatG (Fig. 3C). In M. marinum, the levels of KatG remained unaltered up to 0.2 mM H₂O₂ but decreased significantly after exposure to 2 mM H₂O₂ (Fig. 3C). In M. xenopi, the steady-state levels of KatG remained unaltered up to a concentration of 2 mM hydrogen peroxide. At 20 mM H₂O₂, a concentration that appeared to be generally deleterious, KatG levels were diminished in this species (Fig. 3C). These results indicate differential expression or stability of AhpC and KatG in M. marinum and M. xenopi under conditions of oxidative stress. Although we cannot exclude the possibility that katG is induced at some very narrow concentration range of H₂O₂, the simplest explanation of the observed differences with AhpC and KatG levels is that they are differentially regulated.

Linkage of furA, encoding a homolog of the ferric uptake regulator Fur, and katG in M. marinum.

The differences in responses to hydrogen peroxide exposure of ahpC and katG, as detected by changes in gene product levels, prompted us to explore potential regulatory sequences upstream of the katG gene in M. marinum. The 5′ end of the katG gene was cloned based on the known mycobacterial katG sequences and appropriately designed degenerate primers (see Materials and Methods). No matches with the OxyR recognition sequences defined in this study were found in the region upstream of katG, consistent with the absence of OxyR binding (data not shown). Instead, an open reading frame encoding a homolog of the ferric uptake regulator Fur was identified immediately 5′ from the start of M. marinum katG (GenBank accession no. AF038027) (Fig. 8). Independently of the analyses reported here (38a), we have found fur homologs in a similar arrangement relative to katG in M. tuberculosis, M. smegmatis, and M. leprae. The mycobacterial fur homolog linked to the katG gene has been termed furA (to differentiate it from another fur homolog elsewhere on the M. tuberculosis chromosome). Significantly, furA and katG are cotranscribed in several mycobacterial species (38a).

FIG. 8.

M. marinum FurA and multiple sequence alignment of mycobacterial FurA homologs. Mm-FurA, M. marinum FurA (accession no. AF038027); Mt-FurA, M. tuberculosis FurA (accession no. AF002194); Ml-FurA, M. leprae FurA (accession no. AF013983 for annotation); Mt-Fur, M. tuberculosis Fur (accession no. Z95208); Ec-Fur, E. coli Fur (33).

While any role of furA in katG expression remains to be established in M. marinum, regulatory elements responsive to iron have been indirectly implicated in the regulation of katG in other mycobacteria. For example, a mutation in ideR, encoding a homolog of the iron-responsive Corynebacterium diphtheriae toxin regulator DtxR (35), has been shown to reduce katG expression in M. smegmatis (16). We also probed M. marinum for the presence of ideR and the second fur homolog annotated in the M. tuberculosis sequence databases as a product of genomic sequencing (Fig. 9). A single distinct band hybridizing with M. tuberculosis ideR was observed (Fig. 9D), suggesting the presence of an ideR homolog in M. marinum. When probed with this second homolog of fur from M. tuberculosis (GenBank accession no. Z95208), M. marinum displayed the presence of a somewhat more complex pattern, albeit exhibiting one strongly hybridizing DNA fragment (Fig. 9C). Since none of the bands were identical to the bands hybridizing with the furA and katG probes (Fig. 9A and B), it is likely that M. marinum, like M. tuberculosis, has additional fur homologs.

FIG. 9.

Southern blot analysis of PstI- or SacII-digested M. marinum genomic DNA probed with M. tuberculosis katG (22) (A), furA (accession no. AF002194) (B), fur (accession no. Z95208) (C), and ideR (35) (D).

Based on a recent report (30), gene replacements appear to be possible in M. marinum. Future experiments with cloning of the genes corresponding to ideR and furA, in addition to the planned inactivation of oxyR and furA, will advance our knowledge of the regulation of oxidative stress response in M. marinum. Based on the information provided here, the prototypical arrangement and expression characteristics of oxyR-ahpC and furA-katG in M. marinum, along with its ability to persist in macrophages (3, 29, 31, 45), suggest that M. marinum can serve as a model system for investigations of oxidative stress response in pathogenic mycobacteria. It is likely that future studies with inactivated oxyR and furA genes in M. marinum, and subsequent analyses of effects of such changes on gene expression, survival in macrophages, and persistence in animal models, will provide insights into the reasons underlying the loss of various parts of oxidative stress response in the major human pathogens M. leprae and M. tuberculosis.