Catechol and 2-methylhydroquinone (2-MHQ) cause the induction of the thiol-specific stress response and four dioxygenases/glyoxalases in Bacillus subtilis. Using transcription factor arrays, the MarR-type regulator YkvE was identified as a repressor of the dioxygenase/glyoxalase-encoding mhqE gene. Transcriptional and proteome analyses of the ΔykvE mutant revealed the upregulation of ykcA (mhqA), ydfNOP (mhqNOP), yodED (mhqED) and yvaB (azoR2) encoding multiple dioxygenases/glyoxalases, oxidoreductases and an azoreductase. Primer extension experiments identified σA-type promoter sequences upstream of mhqA, mhqNOP, mhqED and azoR2 from which transcription is elevated after thiol stress. DNase I footprinting analysis showed that YkvE protects a primary imperfect inverted repeat with the consensus sequence of tATCTcgaAtTCgAGATaaaa in the azoR2, mhqE and mhqN promoter regions. Analysis of mhqE-promoter–bgaB fusions confirmed the significance of YkvE binding to this operator in vivo. Adjacent secondary repeats were protected by YkvE in the azoR2 and mhqN promoter regions consistent with multiple DNA–protein binding complexes. DNA-binding activity of YkvE was not directly affected by thiol-reactive compounds in vitro. Mutational analyses showed that MhqA, MhqO and AzoR2 confer resistance to 2-MHQ. Moreover, the ΔykvE mutant displayed a 2-MHQ and catechol resistant phenotype. YkvE was renamed as MhqR controlling a 2-MHQ and catechol-resistance regulon of B. subtilis.
The soil-dwelling bacterium Bacillus subtilis must cope with a variety of antimicrobial agents that include toxic compounds produced by plants, fungi and other bacteria, or derived from industrial contamination. Aromatic compounds have been shown to induce stress responses in microorganisms. They also can provide alternative carbon sources for bacterial species that have the metabolic capability to degrade these compounds. Catecholic or hydroquinone-like intermediates are common in the metabolism of nitroaromatic compounds or organic insecticides (Vaillancourt et al., 2006). There is considerable existing knowledge of the regulation that governs the catabolic pathways involved in degradation of aromatic compounds (Tropel and van der Meer, 2004). In previous studies, we have analysed the gene expression profile of B. subtilis in response to phenolic compounds such as catechol and 2-methylhydroquinone (2-MHQ) by proteomics and transcriptomics (Tam et al., 2006; Duy et al., 2007). The results showed a strong overlap in the response of B. subtilis to catechol, 2-MHQ, diamide and the antibiotic nitrofurantoin as revealed by the common upregulation of the Spx-, CtsR-, PerR- and CymR-dependent thiol-specific oxidative stress response (Leichert et al., 2003; Nakano et al., 2003; Zuber, 2004; Tam et al., 2006; Duy et al., 2007).
Spx is a regulatory protein that activates transcription of genes whose products function in thiol homeostasis if cells encounter oxidative protein stress by non-native disulphide bond formation (Nakano et al., 2003; 2005; Zuber, 2004). Transcription of the spx gene in response to diamide is initiated at the P3 promoter which was shown to depend on two transcriptional regulators, the peroxide regulon repressor PerR and the novel MarR-type/DUF24 family repressor YodB (Leelakriangsak and Zuber, 2007; Leelakriangsak et al., 2007). Furthermore, YodB is suggested as repressor of the divergently transcribed nitroreductase-encoding yodC gene which was induced in response to thiol-stress conditions and derepressed in ΔyodB mutant cells (Duy et al., 2007; Leelakriangsak et al., 2007). Thus, YodB might serve along with Spx, CtsR, PerR and CymR as another global regulator of the thiol-specific stress response in B. subtilis if cells encounter toxic oxidants.
In addition to the thiol-specific oxidative stress response, four putative ring-cleavage dioxygenases/glyoxalases are induced by 2-MHQ, catechol and diamide stress. Enzymes that function in these systems include the previously characterized catechol-2,3-dioxygenase encoded by yfiE (renamed catE) and three putative hydroquinone-type dioxygenases/glyoxalases encoded by mhqA, mhqO and mhqE (Tam et al., 2006; Duy et al., 2007). MhqO, MhqE and MhqA respond most strongly to 2-MHQ stress in B. subtilis and share similarities to the chlorohydroquinone/hydroquinone 1,2-dioxygenase of Sphingomonas paucimobilis LinE (Miyauchi et al., 1999; Endo et al., 2005). Moreover, growth of the ΔmhqO and ΔmhqA mutants of B. subtilis was impaired by 2-MHQ and partly by catechol stress compared with wild-type cells (Duy et al., 2007). These putative dioxygenase enzymes also share conserved glyoxalase domains. The glyoxalase system is a critical detoxification system that is widely distributed among prokaryotic and eukaryotic organisms which functions in the detoxification of cytotoxic methylglyoxal (Ferguson et al., 1998; Booth et al., 2003).
In the present study, genomic, transcriptional and DNA-binding approaches were used to show that the novel MarR-type regulator YkvE controls three putative hydroquinone-specific dioxygenases/glyoxalases and an azoreductase encoded by mhqA, mhqE, mhqO and yvaB (azoR2) which confer resistance to 2-MHQ and catechol stress in B. subtilis. Thus, we have renamed YkvE as Methylhydroquinone-specific repressor (MhqR). We speculate that these multiple dioxygenases/glyoxalases regulated by MhqR could be involved in the detoxification of the natural thiol-reactive electrophile compound methylglyoxal which is produced as a by-product of the glycolysis.
Identification of YkvE as repressor of mhqE–lacZ using transcription factor array analysis
In previous studies, we have identified four putative dioxygenases/glyoxalases encoded by mhqA, mhqE, mhqO and yfiE (catE) (Tam et al., 2006; Duy et al., 2007). The mhqO gene is co-transcribed with mhqN encoding a NAD(P)H nitro/flavin oxidoreductase and mhqP encoding a DoxX-like oxidoreductase. The mhqE gene is co-transcribed with mhqD encoding a phospholipase/carboxylesterase. Expression of mhqA, mhqED and mhqNOP was strongly elevated in cells treated with 2-MHQ and less strongly induced after catechol stress. We sought to identify potential repressors involved in regulation of these 2-MHQ-responsive genes in B. subtilis by using the transcription factor/transformation array technology (Hayashi et al., 2006; Kobayashi, 2007). The B. subtilisΔmhqE mutant was constructed by integration of the non-replicating pMutin4 plasmid in the target gene via a single cross-over recombination resulting in a mhqE–lacZ fusion (Vagner et al., 1998). Chromosomal DNA of each of the 287 transcription factor deletion mutants was transformed into the ΔmhqE mutant and the resulting transformants were screened for increased β-galactosidase activity on plates containing Xgal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside). The ΔykvE deletion mutant resulted in derepression of mhqE–lacZ, suggesting that YkvE is the transcriptional repressor of mhqED operon expression (Fig. 1). YkvE is similar to the MarR family of winged helix transcriptional regulators that control the catabolism of aromatic compounds in different bacteria (Tropel and van der Meer, 2004).
Transcriptome and proteome analysis of the ΔykvE mutant identifies YkvE as repressor of mhqA, mhqED, mhqNOP and yvaB (azoR2)
To identify other genes belonging to the YkvE regulon, microarray and proteome experiments were performed to compare gene expression in B. subtilisΔykvE mutant cells with that of the wild-type strain in minimal medium during the exponential growth phase (Table 1). The complete data set of significantly induced genes and proteins is available in Table S1.
Table 1. Induction of YkvE-regulated genes after 2-MHQ and catechol stress in the wild type (wt) and derepression in the ΔykvE mutant as revealed by transcriptome and proteome analysis.a
Function or similarity
All genes with induction ratios of at least threefold in two transcriptome experiments and proteins with protein synthesis ratios of at least twofold in two proteome experiments which are induced in the wild type (wt) after 10 min in response to 0.33 mM 2-MHQ and 2.4 mM catechol stress and derepressed in the ΔykvE mutant under control conditions (ΔykvE/wt) were listed. The gene function or similarity is derived from the SubtiList database (http://genolist.pasteur.fr/SubtiList/index.html) or related to previous publications (Duy et al., 2007). The complete transcriptome and proteome data set is available in Table S1.
The azoR2 (yvaB) gene was a missing spot in the Eurogentec microarrays and no transcriptome induction ratios were calculated for catechol and 2-MHQ. The transcriptome induction ratios for azoR2 resulting from the comparison of wild-type and ΔykvE mutant cells were derived from data sets using different microarrays as described previously (Hayashi et al., 2006).
The microarray data revealed that mhqA, mhqED and mhqNOP encoding three 2-MHQ-induced dioxygenases/glyoxalases (MhqA, MhqE and MhqO) and oxidoreductases were derepressed in the ΔykvE mutant, which suggests that YkvE functions in the control of 2-MHQ stress-responsive genes in B. subtilis (Table 1). Among these, expression of the mhqNOP operon was most strongly elevated in the ΔykvE mutant (42- to 198-fold). Notably, the thioredoxin encoding ydfQ gene downstream of mhqNOP, which was also induced in response to catechol and 2-MHQ, was derepressed sixfold in the ΔykvE mutant (Table 1). However, only a very faint band corresponding to the mhqNOP-ydfQ-specific transcript was detected in the ΔykvE mutant as revealed by Northern blots using a ydfQ-specific RNA probe (data not shown). Thus, transcription of ydfQ might result from read-through past the mhqNOP terminator.
The proteome analysis of the ΔykvE mutant verified the elevated production of MhqA, MhqE, MhqD, MhqN and MhqO, which all are produced in response to 2-MHQ and catechol stress but not after diamide treatment in the wild type (Fig. 2). The protein synthesis ratios of these YkvE-controlled proteins in the ΔykvE mutant validated also the several-fold increased production of MhqO and MhqN (17- to 39-fold) compared with the lower ratios for MhqE, MhqD and MhqA (three- to fourfold). In addition, the yvaB gene was identified as a potential member of the YkvE regulon. The yvaB gene encodes a FMN-dependent NADH azoreductase and is herewith also referred as azoR2 gene (Nishiya and Yamamoto, 2007). The production of AzoR2 protein was 12- to 34-fold induced in wild-type cells in response to catechol and 2-MHQ stress at the proteome level (Table 1, Fig. 2).
Besides these hydroquinone-induced genes and operons, the microarray data revealed also the upregulation of some SigmaF-, SigmaE- and SigmaG-dependent sporulation genes in the ΔykvE deletion mutant including genes encoding spore coat proteins (cotE, cotJABC) (Table S1) (Eichenberger et al., 2003; Feucht et al., 2003; Steil et al., 2005). As expression of these sporulation genes is not affected by 2-MHQ or catechol stress in the wild type, the increased expression seen in ΔykvE mutant might be indirectly caused by the ΔykvE mutation.
Transcription of azoR2, mhqA, mhqED and mhqNOP is induced by 2-MHQ, catechol and diamide in the wild-type strain and derepressed in the ΔykvE mutant
In previous Northern blot experiments we have studied the expression and transcriptional organization of mhqA, mhqED and mhqNOP (Duy et al., 2007). These data showed that transcription of mhqA, mhqED and mhqNOP is strongly induced by 2-MHQ and more weakly by catechol in B. subtilis wild-type cells. This is supported by the mRNA and protein induction ratios (Figs 2 and 3; Table 1). The mhqNOP-specific transcript showed mRNA degradation products that are stabilized at the level of the 16S RNA as often observed for large transcripts that are strongly induced.
Further Northern blot experiments showed strongly elevated azoR2-specific mRNA concentration in response to 2-MHQ and catechol stress and confirmed that azoR2 is transcribed monocistronically (Fig. 3). Transcription of mhqA, mhqED, mhqNOP and azoR2 was constitutively upregulated in ΔykvE mutant cells (Fig. 3). These Northern blot results suggest that YkvE negatively regulates transcription of azoR2, mhqA, mhqED and mhqNOP in B. subtilis as revealed also by the microarray and proteome experiments.
Surprisingly, we detected also increased mRNA levels of mhqA, mhqED, mhqNOP and azoR2 after exposure of wild-type cells to the thiol-reactive compound diamide (Fig. 3). This is in contrast to the proteome data which showed no protein synthesis of MhqA, MhqE, MhqD, MhqN and MhqO in response to diamide stress. We further analysed whether the YkvE-dependent dioxygenases/glyoxalases respond to oxidative stress or the natural thiol-reactive substrate of the glyoxalase system, methylglyoxal. The results showed that the YkvE regulon is induced neither by hydrogen peroxide nor by methylglyoxal (Fig. 3).
Determination of the transcriptional start sites of azoR2, mhqA, mhqED and mhqNOP
We determined the transcriptional start sites of azoR2, mhqA, mhqE and mhqN using total RNA of wild-type cells which were exposed to 2-MHQ and catechol stress as well as RNA of ΔykvE mutant cells using primer extension experiments (Fig. 4A–D). The 5′ ends of the azoR2-specific mRNA were mapped at C and T located 80 and 81 bp upstream of the azoR2 coding sequence (Fig. 4A). Transcription of mhqA in response to 2-MHQ and catechol is initiated from a transcription start site that was mapped at A located 69 bp upstream from the start codon (Fig. 4B). The start site for transcription of mhqE was mapped at A which resides 32 bp upstream from the ATG start codon (Fig. 4C). Finally, the transcription start site of mhqN was mapped at C, which resides 158 bp upstream from the start codon (Fig. 4D). These transcription start sites were detected using two end-labelled primers each complementary to different N-terminal encoding regions of the respective genes. The transcription start sites of the azoR2, mhqA, mhqE and mhqN genes are preceded by σA-type promoter sequences. The primer extension experiments further showed that transcription is initiated from these σA-dependent promoters most strongly in response to 2-MHQ in the wild-type strain and constitutively in ΔykvE mutant cells.
DNA binding experiments of YkvE to the azoR2, mhqA, mhqE and mhqN promoters
Next, gel mobility shift assays were performed using purified YkvE protein to analyse the DNA-binding activity of YkvE to the promoter regions of the azoR2, mhqA, mhqE and mhqN genes. As shown in Fig. 5A, single DNA–YkvE binding complexes were detected using mhqE and mhqA-specific promoter fragments. For the azoR2 promoter, two shifted DNA–YkvE binding complexes were detected suggesting two different DNA binding regions. In the case of the mhqN promoter, three different DNA–YkvE binding complexes were detected which accounts for multiple operator sites. To identify the cis-acting sequences which function as operator sites for repression by YkvE, DNase I footprinting analysis was performed of purified YkvE protein on the top strand of the azoR2, mhqE and mhqN promoters (Fig. 5B). YkvE protected a region upstream of the azoR2 promoter from positions −60 to −20 relative to the transcription start site. Protection was observed in the mhqE promoter in the region spanning −1 to +22 which ends close to the ribosome binding site. In the mhqN promoter protection was observed in an area from positions −9 to +26. Based on these footprinting results we have aligned the YkvE-protected regions in the azoR2, mhqE and mhqN promoters (Fig. 6). These protected regions contain a primary imperfect inverted repeat with a consensus sequence of tATCTcgaAtTCgAGATaaaa which was designated as MhqR box. The mhqA promoter bears a sequence that resembles the MhqR box consensus at positions +30 to +47. In addition, YkvE footprinting analysis on the mhqN promoter revealed that the protected DNA region also includes an overlapping direct repeat. Further inspection of the mhqN promoter region identified a second 18 bp perfect inverted repeat which is arranged downstream of the promoter in the +122 to +139 region. Thus, the presence of two MhqR boxes and an adjacent secondary direct repeat is consistent with the formation of three DNA–YkvE binding complexes as detected by gel shifts. Furthermore, YkvE footprinting analysis on the azoR2 promoter revealed that the protected area includes also an adjacent secondary repeat element overlapping the promoter region in addition to the conserved MhqR box (Figs 5B and 6). Consistent with these different operator sites in the azoR2 promoter region, two DNA–protein binding complexes were detected using gel shift experiments (Fig. 5A). Thus, it appears likely that the MhqR box detected in the azoR2, mhqA, mhqE and mhqN promoters is the primary binding site for YkvE and the adjacent secondary repeat elements in the azoR2 and mhqN promoters serve for multimeric binding.
Confirmation of in vivo roles of YkvE binding to the mhqE promoter DNA
To confirm the importance of the MhqR box for binding of YkvE in vivo, we constructed different mhqE-promoter–bgaB fusions, which carried the −203 to +22 full-length mhqE promoter region (ST1), the −203 to −1 promoter region which lacks the 18 bp MhqR (YkvE) box (ST2) and a mutant mhqE operator site where the downstream repeat element AGATA is mutagenized to AGGGG (ST3) (Fig. S1). Expression of wild-type mhqE–bgaB was low in strain ST1 and ninefold derepressed in strains ST2 and ST3 either lacking the YkvE operator sites or harbouring one mutagenized repeat element. These in vivo data confirm the footprinting results and show that the MhqR box identified in the mhqE promoter is the cis-acting site required for repression by YkvE. However, we were not able to monitor the induction of the mhqE–bgaB fusion in response to 2-MHQ and catechol stress and assume that BgaB is inactivated by these toxic oxidants as reported previously also for diamide (Leelakriangsak and Zuber, 2007).
Analysis of the effect of MHQ, catechol, diamide, H2O2 and methylglyoxal on DNA-binding activity of YkvE
Gel shift experiments were performed to analyse whether DNA-binding activity of YkvE is directly inhibited by thiol-reactive compounds or reactive oxygen species which could be produced during the metabolism of these compounds (Fig. 7). Thus, DNA-binding reactions were performed with the mhqE- and azoR2-specific promoter probes and purified YkvE protein in the presence of different concentrations of MHQ, catechol, diamide, H2O2 and methylglyoxal. The results showed that DNA-binding activity of YkvE to the azoR2 and mhqE promoters is neither directly inhibited in the presence of thiol-reactive compounds nor by oxidative stress even if high concentrations were applied (Fig. 7). Thus, YkvE is most likely not directly inactivated by thiol-reactive compounds.
Transcription of ykvE is repressed in response to catechol, 2-MHQ and diamide stress
To address the question whether ykvE might be subject to autoregulation like other MarR-type repressors, Northern blot analyses were performed using a ykvE-specific mRNA probe. Interestingly, transcription of ykvE is strongly reduced in response to MHQ, catechol and diamide stress in the wild type (Fig. 8A). In addition, a conditional IPTG-inducible ykvE mutant was constructed using the pMutin plasmid resulting in a ykvE–lacZ fusion (see Experimental procedures). Analysis of the β-galactosidase activity in the presence and absence of IPTG (Fig. 8B) revealed a slightly increased ykvE-promoter–lacZ activity in the presence of YkvE which indicates that ykvE is not subject to autorepression. This is further supported by the fact that the conserved MhqR box is not present in the ykvE promoter region.
The ΔazoR2 mutant is sensitive and the ΔykvE mutant is resistant to catechol and 2-MHQ
Previous phenotype analyses showed that the growth of ΔmhqO and ΔmhqA deletion mutants was inhibited in the presence of 0.5 mM 2-MHQ which normally resulted in a half-maximal growth rate of the wild-type strain (Duy et al., 2007). Examination of the ΔazoR2 mutant phenotype after treatment with 2-MHQ and catechol showed that the growth of the ΔazoR2 mutant was also impaired in the presence of 0.5 mM 2-MHQ and 3.6 mM catechol (Fig. 9A), which suggests that the azoreductase-encoding azoR2 gene also contributes to catechol and 2-MHQ resistance. We further analysed the growth of the ΔykvE mutant after challenge with different concentrations of 2-MHQ and catechol. The ΔykvE mutant was able to grow with 1 mM 2-MHQ and 12 mM catechol, concentrations which are toxic to wild-type cells. Thus, the ΔykvE mutant showed a 2-MHQ and catechol-resistance phenotype. The toxicity of catechol and 2-MHQ can be also monitored by the red-coloured polymers which appear if cells are not able to detoxify these compounds. In wild-type cells these polymers were visible at toxic concentrations of 4.8 mM catechol and 0.66 mM 2-MHQ. In contrast, the ΔykvE mutant was sensitive to concentrations of 14.4 mM catechol and 1.2 mM 2-MHQ as revealed by the red-coloured medium.
Transcription of YkvE-dependent genes in the ΔyodB mutant
In previous studies we found that transcription of mhqA, mhqED and mhqNOP is reduced in the ΔyodB mutant in response to catechol but not after MHQ stress (Duy et al., 2007). Further Northern blot analyses revealed that, conversely, transcription of the YodB-dependent yodC gene is reduced in the ΔykvE mutant under control conditions and in response to 2-MHQ and catechol stress (Fig. 10). The decreased expression by catechol and 2-MHQ might be attributed to the increased resistance of the ΔykvE and ΔyodB mutants. This is supported by the fact that higher growth-inhibitory concentrations lead to derepression of the YkvE-dependent mhqNOP operon by catechol stress in the ΔyodB mutant and vice versa of the YodB-dependent yodC gene in the ΔykvE mutant in response to catechol and MHQ stress (Fig. 10). The regulatory interactions between the YkvE and YodB regulons are currently unknown and require further investigations.
MhqR (YkvE) controls glyoxalases/dioxygenases, oxidoreductases and an azoreductase in response to thiol-stress conditions which confer resistance to catechol and 2-MHQ
In this study, we have used genomic approaches to identify the MarR-type regulator MhqR (YkvE) as a repressor of the 2-MHQ and catechol resistance regulon. The MhqR regulon includes mhqA, mhqED, mhqNOP and azoR2 which encode putative dioxygenases/glyoxalases, oxidoreductases, a phospholipase and an azoreductase. Previously, the mhqA, mhqE and mhqO genes were predicted to share similarities to linE encoding a chlorohydroquinone/hydroquinone 1,2-dioxygenase of S. paucimobilis which cleaves hydroquinone-like substrates (Miyauchi et al., 1999; Endo et al., 2005). Hydroquinones are intermediates in the metabolism of nitroaromatic compounds and insecticides, such as p-nitrophenol, γ-hexachlorocyclohexane (lindane), pentachlorophenol, parathion and fenitrothion (Munnecke, 1976; Stevens et al., 1991; Miyauchi et al., 1999; Xu et al., 1999; Tago et al., 2005; Vaillancourt et al., 2006). Thus, the MhqR regulon could be required for degradation of organic environmental pollutants present in contaminated soil. However, previous studies (Tam et al., 2006) and our unpublished data revealed that B. subtilis is not able to grow in the presence of catechol or MHQ as sole carbon energy source.
The MhqR-regulated dioxygenases/glyoxalases are conserved also among pathogenic Gram-positives including Bacillus anthracis, Listeria or Staphylococcus. Dioxygenases and glyoxalases belong to the vicinal oxygen chelate (VOC) superfamily that are structurally related proteins which provide a metal coordination environment with two or three open or readily accessible coordination sites to promote direct electrophilic participation of the metal ion in catalysis (Armstrong, 2000). The glyoxalase (Glx) enzyme system consists of two metalloenzymes which detoxify toxic methyglyoxal (MG) to lactate (Ferguson et al., 1998; Booth et al., 2003). GlxI accepts MG as non-enzymatically formed hemithioacetal of glutathione which is converted to S-lactoylglutathione. The metallohydrolase GlxII regenerates the thiol co-substrate and lactate. MG is a natural thiol-reactive electrophile which is produced upon accumulation of dihydroxyacetone phosphate (DHAP) if glycolysis becomes imbalanced. Three MhqR-regulated putative glyoxalase enzymes are induced in response to the thiol-specific agents. Thus, we are tempting to speculate that MG could be produced upon exposure to thiol-reactive agents perhaps via the inactivation of the active site cysteine of GapDH which could lead to accumulation of DHAP (Ferguson et al., 1998). As B. subtilis does not possess glutathione, it remains to be determined if cysteine, CoASH or other low-molecular-weight thiols are used for detoxification of MG in B. subtilis.
Multiple GlxI enzymes have been recently characterized also in Pseudomonas aeruginosa including two non-Zn2+ and one Zn2+-dependent enzymes (Sukdeo and Honek, 2007). Interestingly, B. anthracis encodes 18 glyoxalase family enzymes, three of which are strongly induced by paraquat (Passalacqua et al., 2007). Biochemical studies are underway to characterize the enzymatic activities of the MhqR-regulated dioxygenases/glyoxalases regarding the detoxification of catechol, MHQ or MG. Preliminary experiments revealed GlxI activity for one of these MhqR-regulated enzymes (our unpublished data).
Surprisingly, the MhqR regulon was not induced by the commercially available MG in vivo, nor was MG able to prevent DNA-binding activity of MhqR in vitro. MG reacts with the side-chains of the amino acids arginine, lysine, cysteine and with the bases guanine, adenine and cytosine in Escherichia coli (Ferguson et al., 1998). In addition, MG has been shown to be mutagenic in Salmonella typhimurium and E. coli. Thus, MG might cause predominantly DNA damage in B. subtilis. Furthermore, it is described that commercially available MG is contaminated with formaldehyde, pyruvate, lactate and formate (Kalapos, 1999). Thus, detailed proteomic and transcriptomic experiments using purified MG are required to elucidate the mode of action of MG in B. subtilis.
MhqR recognizes conserved MhqR boxes in the promoter regions of mhqA, mhqED, mhqNOP and azoR2
DNase I footprinting experiments identified conserved MhqR boxes in the upstream region of all MhqR regulon members and adjacent repeat elements in the promoter regions of mhqN and azoR2 which probably serve for multimeric binding. The MhqR box contains an 18 bp imperfect palindromic sequence composed of two adjacent 9 bp half-sites. This is consistent with the homodimer structure of MarR-like regulators which is required for recognition of the two half-sites of the operator (Alekshun et al., 2001; Hong et al., 2005; Wilkinson and Grove, 2006). The centre of each half-site is separated by about 10 bp, thus positioning the half-sites for binding of MhqR in the major groove on the same face of the double helix. The crystal structure of the OhrR-ohrA operator complex revealed the chimeric nature of the winged helix–turn–helix (wHTH) DNA-binding motif, which is composed of extended eukaryotic-like wings, prokaryotic HTH motifs and helix–helix elements, which are proposed to be utilized by the entire MarR family to bind cognate DNA (Hong et al., 2005).
Autoregulation is common for many MarR family repressors which are often encoded in contiguous operons (EmrR and MarR) or divergently oriented to their target genes (e.g. HpaR, MexR, HucR) (Alekshun and Levy, 1997; Evans et al., 2001; Galan et al., 2003; Wilkinson and Grove, 2006). In contrast, transcription of mhqR is downregulated in response to thiol-reactive agents that lead to derepression of the MhqR regulon. Furthermore, the mhqR promoter region lacks the conserved MhqR box indicating that mhqR is not autoregulatory.
MhqR and YodB control paralogous azoreductases with overlapping functions
Another MarR family repressor (YodB) regulates transcription of spx and the nitroreductase-encoding yodC gene in response to thiol-stress conditions (Leelakriangsak and Zuber, 2007; Leelakriangsak et al., 2007; Duy et al., 2007). In addition, the azoreductase YocJ (AzoR1) is regulated by YodB which is a paralogue of the MhqR-regulated enzyme YvaB (AzoR2) (M. Leelakriangsak et al., submitted). Azoreductases are flavin-containing enzymes which catalyse the NADH-dependent two-electron reductions of quinones to protect cells against damage by free radicals and reactive oxygen species (Fischl and Kennedy, 1990). Recently, it has been reported that AzoR2 has NADH:2,6-dichloroindophenol oxidoreductase and azoreductase activity in B. subtilis (Nishiya and Yamamoto, 2007). The ΔazoR2 mutant showed a growth defect in the presence of 2-MHQ and catechol. In addition, the ΔmhqR mutant conferred a catechol and 2-MHQ-resistance phenotype due to overproduction of MhqNOP, MhqA and/or AzoR2. These enzymes are derepressed in response to the thiol-specific compounds MHQ, catechol, diamide at the transcriptional level. However, increased protein synthesis of the MhqR-regulated proteins was observed only after MHQ and catechol stress, not in response to diamide stress. Thus, future studies are aimed to study whether mRNA stability of the MhqR regulon genes is affected by diamide stress.
Interestingly, the ΔyodB mutant also acquired a similar resistance to catechol and 2-MHQ stress as the ΔmhqR mutant. Moreover, the ΔyodBΔmhqR double mutant showed an additive hyper-resistance to catechol and 2-MHQ stress in B. subtilis (M. Leelakriangsak et al., submitted). These phenotype results suggest that the paralogous azoreductases AzoR1 and AzoR2 encoded by the MhqR and YodB regulons might have similar detoxification functions. As these azoreductases are co-regulated with the glyoxalases, they could be analogous to the E. coli aldo-keto-reductase systems involved in the GSH-independent pathway for detoxification of MG.
Mechanisms for regulation of YodB and MhqR in response to thiol-stress conditions
The YodB and MhqR regulons are both induced under thiol-stress conditions and confer resistance to catechol and MHQ via paralogue azoreductases YocJ (AzoR1) and YvaB (AzoR2) (M. Leelakriangsak et al., submitted). YodB belongs to the MarR/DUF24 family of repressors and the conserved C6 residue was shown to directly sense thiol-reactive compounds. Cysteine residues are strong nucleophiles which are shown to be conjugated with electrophilic compounds such as p-benzoquinone and hydroquinones to form thiol-S-adducts (Giles et al., 2003; Rodriguez et al., 2005; Dennehy et al., 2006). Mass spectrometry analyses identified the formation of Cys6-S-adducts upon treatment of YodB with the thiol-reactive compounds MHQ and catechol in vitro leading to the loss of DNA-binding activity of YodB (M. Leelakriangsak et al., submitted). Thus, YodB is directly sensing thiol-reactive compounds via the conserved C6 residue in contrast to MhqR which does not loose DNA-binding activity in response to thiol-reactive compounds in vitro. One cysteine residue is present in MhqR (C128) which is located in the dimerization domain and accessible for thiol modification according to published structures for B. subtilis OhrR and E. coli MarR (Alekshun et al., 2001; Hong et al., 2005). In contrast to C6 in YodB orthologues, C128 of MhqR is not conserved among orthologues in other Gram-positive bacteria. Thus, there might be unknown metabolites produced upon exposure to thiol-reactive compounds that could be sensed by MhqR, which might lead to derepression of the MhqR regulon.
In conclusion, we show herein that B. subtilis encodes a further thiol stress-responsive MarR-type repressor (MhqR) which controls expression of three putative dioxygenase/glyoxalase enzymes and an azoreductase in B. subtilis while conferring resistance to 2-MHQ and catechol. This regulatory mechanism is coupled to the complex network of control that senses and responds to thiol-specific stress in B. subtilis. Hence, it is tempting to speculate that the MhqR regulon has evolved to detoxify the natural thiol-reactive compound methylglyoxal which is produced as a by-product of the glycolytic pathway from bacteria to humankind.
Bacterial strains and growth conditions
The bacterial strains used were B. subtilis 168 (trpC2), ΔmhqE (trpC2,mhqE::pMutin4; Emr), ΔykvE (trpC2,ykvE::Cmr), ΔyvaB (trpC2,yvaB::pMutin4; Emr), ST1 (trpC2,amyE::mhqE1–bgaB; Cmr), ST2 (trpC2,amyE::mhqE2–bgaB; Cmr), ST3 (trpC2,amyE::mhqE3–bgaB; Cmr), TFSP176 (trpC2,IykvE::pMutinNC; Emr). B. subtilis strains were cultivated under vigorous agitation at 37°C in Belitsky minimal medium described previously (Stülke et al., 1993). E. coli strains were grown in LB for DNA manipulation. The antibiotics were used at the following concentrations: 1 μg ml−1 erythromycin and 5 μg ml−1 chloramphenicol. The compounds used were 2-MHQ (Acros), catechol (Fluka), diamide, H2O2 and methylglyoxal (Sigma).
The ΔmhqE and ΔyvaB mutants were constructed in the course of the European and Japanese B. subtilis functional analysis project as described previously (Vagner et al., 1998). For construction of the ΔmhqE mutant, an internal fragment of the mhqE coding region was amplified using the primer sets mhqE-F and mhqE-R by PCR. The corresponding PCR product was digested with HindIII and BamHI and inserted into the pMutin4 plasmid digested with the same enzymes to generate plasmid pMutin-mhqE which was integrated into the chromosomal mhqE gene via single cross-over recombination generating a mhqE–lacZ fusion (Vagner et al., 1998).
For construction of a conditional IykvE mutant (TFSP176) which contains the ykvE gene under control of the isopropyl-β-d-thiogalactoside (IPTG)-inducible Pspac promoter, the 5′ region of ykvE including the ribosome binding site was amplified by PCR with primers ykvE-F10 and ykvE-R10. The PCR product was digested with HindIII and BamHI and inserted into plasmid pMutinNC digested with the same enzymes to generate plasmid pMutin-Pspac ykvE. This plasmid was integrated into the chromosomal ykvE gene resulting in an IPTG-inducible conditional ykvE mutant containing a ykvE–lacZ fusion (strain TFSP176) which was used to study the autoregulation of ykvE.
The transcription factor deletion mutants were constructed using an overlap-extension PCR technique as described previously (Kobayashi, 2007). The cat gene was amplified from the plasmid pCBB31 by PCR with primers pUC-F and pUC-R. Upstream and downstream regions of each regulator gene were amplified by PCR using gene-specific primer sets F1/R1 and F2/R2 as described previously (Kobayashi, 2007). Specifically, the primer sets ykvE-F1/ykvE-R1 and ykvE-F2/ykvE-R2 were used for amplification of the ykvE upstream and downstream regions (Table S2). The 5′ end of primers R1 and F2 are complementary to pUC-R and pUC-F sequences respectively. Then, the three PCR fragments were mixed and used as template for a second PCR reaction with primers F1 and R2. The resultant PCR fragments were used for transformation of B. subtilis 168.
For construction of strains containing bgaB fusions of the wild-type mhqE-promoter, the 3′mhqE deletion promoter and the mhqE promoter point mutant, the promoter region of mhqE was amplified by PCR using primer MhqE_for-EcoRI in combination with primers MhqE_rev1_BamHI, MhqE_rev2_BamHI, MhqE_rev3_BamHI respectively. The corresponding PCR products were digested with EcoRI and BamHI and inserted in pDL digested with the same enzymes to generate plasmids pST1, pST2, pST3 and respectively. These plasmids were used to transform competent cells of B. subtilis 168 to generate strains ST1, ST2 and ST3 which were used for β-galactosidase assays.
Transcription factor array analysis
The transcription factor array analysis was performed as described previously (Hayashi et al., 2006; Kobayashi, 2007) using competent cells of the ΔmhqE::pMutin4 strain which carries a mhqE–lacZ fusion.
Proteome analysis and mass spectrometry
Cells grown in minimal medium to an OD500 of 0.4 were pulse-labelled for 5 min each with 5 μCi of l-[35S]methionine per ml before (control) and 10 min after 2-MHQ and catechol challenge. Preparation of cytoplasmic l-[35S]methionine-labelled proteins and separation by two-dimensional gel electrophoresis (2D-PAGE) using the immobilized pH gradients (IPG) in the pH range 4–7 was performed as described (Tam et al., 2006). The quantitative image analysis was performed with the DECODON Delta 2D software (http://www.decodon.com). Proteins showing an induction of at least twofold during the 5 min of l-[35S]methionine pulse in two independent experiments were considered as significantly induced. For identification of the proteins by mass spectrometry, non-radioactive protein samples of 200 μg were separated by 2D-PAGE and the 2D gels were stained with Colloidal Coomassie brilliant blue (Amersham Biosciences). Spot cutting, tryptic digestion of the proteins and spotting of the resulting peptides onto the MALDI targets (Voyager DE-STR, Per Septive Biosystems) were performed using the Ettan Spot Handling Workstation (Amersham-Biosciences, Uppsala, Sweden) as described previously (Eymann et al., 2004). The Matrix-assisted laser desorption/ionization – Time of flight (MALDI-TOF-TOF) measurement of spotted peptide solutions was carried out on a Proteome-Analyser 4800 (Applied Biosystems, Foster City, CA, USA) as described previously (Eymann et al., 2004).
For microarray analysis, the B. subtilis wild type and ΔykvE mutant were grown in minimal medium and harvested at OD500 of 0.4. Total RNA was isolated by the acid phenol method as described (Majumdar et al., 1991). Generation of fluorescence-labelled cDNA and hybridization with B. subtilis whole-genome microarrays (Eurogentec) was performed as described previously (Jürgen et al., 2005). Two independent hybridization experiments were performed using RNAs from two independent experiments. Genes showing induction ratios of at least threefold in two independent experiments were considered as significantly induced.
Northern blot experiments
Northern blot analyses were performed as described (Wetzstein et al., 1992) using RNA isolated from B. subtilis wild-type and ΔykvE mutant cells before (control) as well as 10 and 20 min after the treatment with 2.4 mM catechol or 0.33 mM 2-MHQ respectively. Hybridizations specific for mhqD, mhqO, mhqA, azoR2 and ykvE were performed with the digoxigenin-labelled RNA probes synthesized in vitro using T7 RNA polymerase from T7 promoter containing internal PCR products of the respective genes using the primer sets as described (Duy et al., 2007) as well as azoR2-for1 and azoR2-rev1, ykvE-for and ykvE-rev (Table S2).
Primer extension experiments
Two different primers each which are complementary to the N-terminal encoding region of azoR2 (azoR2-PE and azoR2-PE2), mhqA (mhqA-PE and mhqA-PE2), mhqE (mhqE-PE and mhqE-PE2) and mhqN (mhqN-PE and mhqN-PE2) (Table S2) were 5′-end labelled using T4 polynucleotide kinase (Roche Diagnostics) and 50 μCi of [γ-32P]-ATP (GE Healthcare). Primer extension analysis was performed using the labelled primers as described previously (Wetzstein et al., 1992). Sequencing of the corresponding promoter regions was performed as described (Sanger et al., 1977) using PCR products as templates containing the promoter region of the respective genes amplified with primer sets azoR2-for2 and azoR2rev2, mhqA-for2 and mhqA-rev2, mhqE-for2 and mhqA-rev2, mhqN-for2 and mhqN-rev2 respectively (Table S2).
Expression and purification of the YkvE protein
The IMPACT self-cleavable, affinity tag system (New England Biolabs) was used to purify YkvE. E. coli BL21(DE3)pLysS (Invitrogen) was used for overproduction of the protein. The ykvE coding sequence was amplified by PCR using primers oykvE-IntBs and oykvE-IntSm (Table S2). The PCR products were digested with BspHI and SmaI restriction enzymes and inserted into pTYB4 (BioLabs) digested with NcoI and SmaI to generate pML70. BL21(DE3)pLysS carrying pML70 was cultured in 3 l of LB medium, and IPTG (isopropyl-β-d-thiogalactopyranoside) (final concentration, 0.5 mM) was added at the mid-log phase (OD600 of 0.6). After 4 h, the cells were harvested by centrifugation, and re-suspended in buffer A [25 mM Tris-HCl (pH 8.0), 500 mM NaCl, 0.1 mM EDTA, 1 mM MgCl2, 10% glycerol]. The cells were disrupted using French pressure cell and the lysate was centrifuged 15 000 r.p.m. in Sorvall TI70. The lysate was mixed and passed through a chitin column, washed with buffer A and buffer B [25 mM Tris-HCl (pH 8.0), 100 mM KCl, 0.1 mM EDTA, 1 mM MgCl2, 10% glycerol] and then the column was quick flushed with buffer B containing 50 mM DTT. After incubation overnight at 4°C, protein was eluted by buffer B. The protein-containing fractions were applied to a Hi-Q column (Bio-Rad) and YkvE was eluted with a 100–500 mM KCl gradient. The purified YkvE was extensively dialysed against buffer B and stored at −80°C.
DNase I footprinting analysis
DNA probes for mhqE (position corresponding to positions −114 to +67 relative to TSS), azoR2 (positions −113 to +86) and mhqN (positions −104 to +137) were synthesized by PCR amplification using primers oML07-52,oML07-53; oML07-54,oML07-55; oML07-56,oML07-57 respectively (Table S2). Purified PCR fragments were used as sequencing templates. DNA probe purification and DNase I footprinting were performed as described previously (Leelakriangsak et al., 2007).
DNA gel mobility shift assays
DNA fragments containing the promoter regions of azoR2, mhqA, mhqE and mhqN genes were generated by PCR using the primer sets azoR2-for2 and azoR2rev2, mhqA-for2 and mhqA-rev2, mhqE-for2 and mhqA-rev2, mhqN-for2 and mhqN-rev2 respectively. Approximately 600 pmol of the purified PCR products were end-labelled using T4 polynucleotide kinase (Roche Diagnostics) and 50 μCi of [γ-32P]-ATP (GE Healthcare). The labelled DNA probes were purified by precipitation and 20 000 cpm of each probe was incubated with different amounts of purified YkvE protein for 30 min at room temperature in EMSA-binding buffer (10 mM Tris-HCl pH 7.5, 100 mM KCl, 0.1 mM EDTA, 5 mM MgCl2, 0.5 mM DTT, 10% glycerol) in the presence of 0.5 mg of BSA ml−1 and 0.05 mg of poly(dI-dC) ml−1. The concentrations of the compounds used for the DNA binding reactions were 1–20 mM MHQ, 2–50 mM catechol, 1–10 mM diamide, 1–100 mM H2O2 and 2–20 mM methylglyoxal. Samples were separated by 5% native polyacrylamide gel electrophoresis in 10 mM Tris, 1 mM EDTA buffer, pH 8 at 10°C and constant voltage (150 V) for 6 h. Gels were dried and the radiolabelled bands were visualized using phosphoimaging.
We thank the Decodon company for support with the Decodon Delta 2D software, Sebastian Grund for excellent technical assistance and Dirk Albrecht for mass spectrometry analysis. This work was supported by grants from the Deutsche Forschungsgemeinschaft, the Bundesministerium für Bildung und Forschung (BACELL-SysMo 031397A), the Fonds der Chemischen Industrie, the Bildungsministerium of the country Mecklenburg-Vorpommern and European Union grants BACELL-Health (LSHG-CT-2004−503468) and BACELL-BaSysBio (LSHG-CT-2006-037469) to M.H., by a scholarship of the ‘Ministry of Education and Training of Vietnam’ (MOET) to N.V.D, by a Royal Thai government scholarship to M.L., by Grant-in-Aid for Young Scientist Research (A) 17681023 from the Ministry of Education, Culture, Sports, Science and Technology of Japan to K.K., as well as by Grant GM45898 from the National Institutes of Health and grant from the Medical Research foundation of Oregon to P.Z.