A Copper-responsive Global Repressor Regulates Expression of Diverse Membrane-associated Transporters and Bacterial Drug Resistance in Mycobacteria

Background: The regulators involved in the expression of bacterial multidrug resistance transport proteins remain largely unknown.

Results: Ms2173, a copper-responsive repressor, has been characterized as the first mycobacterial GntR/Fad-like transcription factor that regulates bacterial drug resistance.

Conclusion: Ms2173 regulates expression of diverse membrane-associated transporters and bacterial drug resistance in mycobacteria.

Significance: These findings enhance our understanding of gene regulation and drug resistance in mycobacteria.

Abstract

Sequencing of entire bacterial genomes has led to the identification of many membrane-associated transporters, including several multidrug resistance transport proteins, in recent years. However, the regulators and signaling pathways involved in the expression of these genes remain largely unknown. In this study, we have identified Ms2173, a GntR/FadR family transcription factor, as a novel global regulator in Mycobacterium smegmatis. Ms2173 was found to specifically recognize a 15-bp palindromic motif and to broadly regulate expression of 292 genes, including 37 genes that encode membrane-associated transport proteins. Copper ions induced Ms2173 to form inactive proteins lacking DNA-binding activity. Ms2173 was shown to function as a repressor of its target genes. Interestingly, we found that the function of Ms2173 was linked to mycobacterial drug resistance. Compared with the substantially enhanced drug resistance in the Ms2173-deleted mutant strain, the strains overexpressing Ms2173 were more sensitive to anti-tuberculosis drugs than the wild-type strain. Additionally, copper ions could partially counteract the in vivo function of Ms2173. We have thus characterized the first mycobacterial GntR/Fad-like transcription factor that functions as a copper ion-responsive global repressor that we have renamed GfcR. These findings further enhance our understanding of membrane-associated transporter regulation and drug resistance in mycobacteria.

Introduction

In recent decades, the appearance and spread of bacterial drug resistance has become a major health issue worldwide. Important pathogens such as Mycobacterium tuberculosis, the causative microbe for tuberculosis, have now acquired multi-drug resistance (MDR)³ (1). Understanding the regulatory mechanism of bacterial drug resistance is therefore an important and urgent goal with significant implications for human health and disease worldwide.

Several studies have shown that bacterial drug resistance occurs by numerous mechanisms, including enzymatic inactivation, drug target alteration or protection, and reduced uptake (2, 3). Among them, increase in active membrane-bound efflux pumps that transport toxic antibacterial drugs from the cell is a major concern, especially because individual MDR pumps are capable of exporting a number of structurally dissimilar drugs (4, 5). In recent years, a number of bacterial regulatory proteins that govern the expression of drug transporters have been characterized (3). Such proteins encompass both repressors and activators such as the Escherichia coli TetR and EmrR (6, 7), the Staphylococcus aureus QacR (8), and the Bacillus subtilis BmrR (9), which inhibit or stimulate the expression of their target efflux genes. Interestingly, several global regulators such as MarA, Rob, and SoxS have been shown to enhance the expression of the E. coli acrAB MDR locus (10). These studies have provided a basic picture of the regulatory pathways controlling the expression of drug efflux genes in bacteria. However, additional MDR transporter homologs have been identified by sequencing the entire bacterial genome, and their regulators and underlying regulatory mechanisms remain to be explored (11).

Mycobacterium smegmatis is a fast-growing model mycobacterium and has been widely used to study the gene regulatory mechanism of virulent and slow-growing species like M. tuberculosis (12, 13). The genomes of both M. smegmatis (GenBank^tm accession number CP000480) and M. tuberculosis encode at least two dozen putative drug efflux transporters (14). Several of these transporters have been shown to contribute to mycobacterial resistance to isoniazid (INH), rifampicin (RIF), tetracycline, and other toxic compounds (15–17). Strikingly, M. smegmatis contains more than 500 potential regulatory factors. The large number of putative drug efflux systems and their potential regulators, therefore, underscore the complexity of the mechanisms involved in regulation of drug resistance in M. smegmatis (18). However, potential regulators involved in broad regulation of expression of membrane-associated transporter genes have not been successfully isolated to date.

GntR, named after a repressor of the B. subtilis gluconate operon (19), is a large, poorly characterized transcriptional regulation family both in M. smegmatis and M. tuberculosis (20, 21). GntR family proteins possess a highly conserved N-terminal winged helix-turn-helix domain for DNA-binding and a diverse C-terminal ligand-binding domain for effector-binding/oligomerization (19). The variable C-terminal domain provides a basis for their classification into subfamilies such as FadR, HutC, MocR, YtrA, AraR, and PlmA (22). FadR is the largest subfamily, comprising ∼40% of all GntRs, and most of them are involved in the regulation of oxidized substrates such as pyruvate (PdhR), gluconate (GntR), glycolate (GlcC), and L-lactate (LldR) (22–25). FadR binds DNA as a homodimer and functions as a repressor of the fad regulon in E. coli (26–29). However, no GntR/Fad family regulator has been functionally identified and characterized in mycobacteria to date.

In this study, we have successfully isolated and characterized the first GntR/FadR-like transcription factor, Ms2173, in M. smegmatis. Ms2173 acts as a novel master regulator that responds to copper ions and regulates expression of a diverse set of genes that includes 37 membrane-associated transporters. Furthermore, we provide evidence to show that Ms2173 functions as a global repressor and negatively affects mycobacterial drug resistance. Thus, we have identified a new regulatory pathway for bacterial drug resistance that is mediated by a unique copper ion-responsive GntR/FadR-like regulator in mycobacteria.

EXPERIMENTAL PROCEDURES

Strains, Enzymes, Plasmids, and Reagents

E. coli BL21 strains and the pET28a plasmid were purchased from Novagen. All restriction enzymes, T4 ligase, Pyrobest DNA polymerase, dNTPs, and all antibiotics were purchased from TaKaRa Biotech. PCR primers were synthesized by Invitrogen (supplemental Tables S1–S4).

Negative Regulator Screenings for the Drug Resistance of M. smegmatis

About 500 predicted regulatory genes in the genome of M. smegmatis mc²155 (GenBank^TM accession number CP000480) were cloned downstream of the tetracycline-inducible promoter tetO in pMind (30), and the recombinant plasmids were transformed into M. smegmatis. Transformant strains were spotted on Middlebrook 7H10 medium (complemented with 0.2% glycerol) containing 10 μg/ml kanamycin and 25 μg/ml tetracycline. The INH-sensitive transformants were screened on the above medium with or without 10 μg/ml INH. The used INH concentration (10 μg/ml) for screening was chosen according to multiple experiments in which condition the INH-sensitivities of transformants were easily observed. The recombinant plasmids were isolated from the INH-sensitive M. smegmatis transformants and, therefore, the negative regulator genes could be characterized.

Expression and Purification of Recombinant Protein and PAGE Analysis

Ms2173 and its mutant genes were amplified by PCR from the genomic DNA of M. smegmatis mc²155 and were cloned into the pET28a vector to produce recombinant vectors. After being transformed with the recombinant plasmid (supplemental Table S5), E. coli BL21 cells were grown in 200 ml Luria Bertani medium up to an A₆₀₀ of 0.6. Protein expression was induced by the addition of 0.5 mm isopropyl β-d-1-thiogalactopyranoside. The cells were collected, and proteins were purified using an affinity column as described previously (31). The elution was dialyzed overnight and stored at −80 °C. A common SDS-PAGE was used to determine the protein molecular weight.

Homology Structure Modeling

The structure of Ms2173 was modeled computationally using the automated comparative protein modeling web server SWISS-MODEL (32). The E. coli FadR family of proteins (27) was used as a template (PDB code 1E2X).

DNA Substrate Preparation and EMSA

DNA fragments for the DNA-binding activity assays were amplified by PCR from M. smegmatis mc²155 genomic DNA or directly synthesized by Invitrogen (supplemental Table S5). The DNA substrates were labeled and prepared as described previously (31) and stored at −20 °C until use. EMSA experiments with labeled DNA fragments were performed as described previously (12). Images were acquired using a Typhoon scanner (GE Healthcare).

DNase I Footprinting Assays

The 150-bp promoter regions of the Ms2173 gene (coding strand and non-coding strand) (supplemental Table S3) were amplified by PCR using appropriate primers labeled with fluorescein isothiocyanate (supplemental Tables S1 and S3). The amplified products were purified with a DNA purification kit (BioFlux) and then subjected to similar binding reaction as in EMSA. DNaseI footprinting was performed as described previously (13). The ladders were produced using the Sanger dideoxy method.

Construction of the Ms2173 Deletion Mutant of M. smegmatis and Southern Blot Analysis

Knockout of the Ms2173 gene from M. smegmatis was performed as described previously (31, 33) with modifications. The recombinant plasmid pMind-Ms2173 was constructed and electroporated into M. smegmatis. Allelic exchange mutants in which the Ms2173 gene had been deleted were identified by restriction digestion and subsequent PCR analysis using the primers on each side of Ms2173 and the hygromycin gene. Deletion of Ms2173 was further verified by a previously described Southern blot analysis procedure (31). Genomic DNA was digested overnight with an excess of BsrBI, and the fragments were separated by electrophoresis through 1.0% agarose gels. The probe consisted of a 239-bp fragment from the downstream region of the Ms2173 gene amplified using appropriate primers (supplemental Table S1).

Quantitative Real-time PCR

Isolation of mRNA and cDNA preparation of the Msm/pMindD and Msm/pMmindD-Ms2173 (Msm and Msm/Ms2173::hyg) strains was performed, and real-time PCR analysis was subsequently carried out according to procedures described previously (31). The reactions were performed in a Bio-Rad IQ5 RT-PCR machine. Amplification specificity was assessed using a melting curve analysis. Gene expression levels were normalized to the levels of 16S rRNA gene transcripts. The degrees of expression change were calculated using the 2^−ΔΔCt method (31, 34). Average relative expression levels and standard deviations were determined from three independent experiments.

Chromatin Immunoprecipitation Assay

ChIP was performed as described previously (31). Briefly, exponentially growing M. smegmatis cells were fixed with 1% formaldehyde, and fixation was stopped by adding 0.125 m glycine for 5 min. Cross-linked cells were harvested and resuspended. Samples of those cells were sonicated on ice and incubated with 10 μl of antibodies against Ms2173 or preimmune serum under rotation for 3 h at 4 °C. The complexes were immunoprecipitated with protein A-agarose for 1 h. Cross-linking was reversed for 6 h at 65 °C. DNA samples of the input and the immunoprecipitates were purified and analyzed by PCR using Platinum Taq (Invitrogen). The amplification protocol included one denaturation step of 5 min at 95 °C and then 30 cycles of 1 min at 95 °C, 1 min at 60 °C, and 1 min at 72 °C.

Determination of Mycobacterial Growth Curves and the Effect of Drugs and Metal Ions

Growth patterns of the wild-type (Msm), Ms2173-deleted mutant (Msm/ΔMs2173), and overexpression (Msm/pMindD-Ms2173) mycobacterial strains were determined according to procedures described previously (12) with some modifications. M. smegmatis was grown in 100 ml of 7H9 medium when needed, and 10 μg/ml kanamycin and 25 ng/ml Tet or 30 μg/ml INH were added. Each mycobacterial strain was also freshly grown in Middlebrook 7H9 medium containing 50 μg/ml hygromycin and 10 μg/ml kanamycin if needed. When cells reached a stationary growth phase with an A₆₀₀ of 1.5 to 2.0, the cultures were diluted in Middlebrook 7H9 medium to an A₆₀₀ of 0.2 and split for induction with or without 2.5 μm metal ions (Cu²⁺ or Zn²⁺) or drugs (30 μg/ml INH or 2 μg/ml RIF). The cultures were conducted with an additional growth in 7H10 medium at 37 °C at 200 rpm for 3–4 days. Aliquots were taken at the indicated times, and A₆₀₀ was measured. At the same time, a small aliquot of the culture was plated on 7H10 medium for determining colony forming units.

RESULTS

M. smegmatis Ms2173 Potentially Contributes to Mycobacterial INH Sensitivity

To identify potential transcription factors that regulate drug resistance in M. smegmatis, we screened the transcriptional regulator library by spotting these recombinant strains on plates containing INH (10 μg/ml). A hypothetical transcription factor named Ms2173 was isolated as a potential contributor to INH sensitivity in M. smegmatis. As shown in Fig. 1A, the mycobacterial strain transformed with pMind-Ms2173 and, thus overexpressing Ms2173, grew very slowly on the screening plate and was more sensitive to 10 μg/ml of INH than the wild-type strain. In contrast, recombinant strains overexpressing two unrelated genes, Ms0895 and Ms1117, under the same condition had similar growth as the wild-type strain (Fig. 1A). This suggests that Ms2173 is potentially involved in regulating INH drug resistance in M. smegmatis. A further sequence BLAST assay found that Ms2173 contained an N-terminal winged helix-turn-helix DNA-binding domain and a typical FadR-like C-terminal domain, indicating that Ms2173 encodes a GntR/FadR-like regulator (Fig. 1B).

FIGURE 1.

Assays for the INH sensitivity of the M. smegmatis strain overexpressing Ms2173 and identification of the conserved domain. A, INH sensitivity of the M. smegmatis strain overexpressing Ms2173. Msm/pMindD, Msm/pMindD-Ms2173, and two control strains, Msm/pMindD-Ms0895 and Msm/pMindD-Ms1117, were grown on 7H9 medium with 10 μg/ml kan for 36 h. Then, freshly grown bacteria were streaked on LB plates with kan (10 μg/ml), Tet (25 ng/ml), or INH (10 μg/ml) (right panel). The control plate (center panel) did not contain INH. B, analysis of the structural characteristics of Ms2173. The Ms2173 amino acid sequence was analyzed using the online tool NCBI-CDD. The N-terminal region of Ms2173 contains a Winged HTH-GntR DNA-binding domain, whereas its C terminus contains a variable FadR C-terminal ligand-binding domain.

Copper Ion Specifically Induces Ms2173 to Form an Inactive DNA-binding Protein

We measured the regulatory effects of several metal ions on the DNA-binding activity of Ms2173 (supplemental Fig. S1), and only copper ions (Cu²⁺) could inhibit the DNA-binding activity of Ms2173. We further examined the Cu²⁺ concentration-dependent inhibition on the DNA-binding activity of Ms2173. As shown in Fig. 2A, 6 μm of the Ms2173 protein alone could shift the mobility of DNA substrates in the reactions. However, a stepwise decrease in the formation of DNA/protein complexes was clearly observed with the addition of increasing amounts of Cu²⁺ into the reactions (Fig. 2A). This indicates that the DNA-binding activity of Ms2173 is inhibited by Cu²⁺. In contrast, zinc ions had no similar role under the same conditions (Fig. 2B).

FIGURE 2.

Effect of copper ions on DNA-binding activity of Ms2173. EMSA assays for the effects of Cu²⁺ (A) and Zn²⁺ (B) on the DNA-binding activity of Ms2173. Ms2173 (6 μm) and the labeled promoter DNA substrate of Ms2173 (about 150 bp) were coincubated with various concentrations of Cu²⁺ or Zn²⁺ (0–0.2 mm) (left panel). The right panel shows the results of EMSA assays in which Ms2173 and the DNA substrates were preincubated for 30 min before various concentrations of Cu²⁺ were added.

Alternatively, when Ms2173 was first preincubated with DNA substrates for 30 min and the same amount of Cu²⁺ as above was then added into the reactions, the inhibitory effect of Cu²⁺ was significantly lower (Fig. 2A, right panel). This indicates that the binding affinity of Cu²⁺ to Ms2173 is weak compared with that of Ms2173 to DNA.

Ms2173 Binds Target DNA Fragments Containing a Palindromic Sequence Motif

Several truncated DNA substrates covering the promoter region of Ms2173, designated as S1–S3 (Fig. 3A), were produced to characterize the DNA motif recognized by the Ms2173 protein. An obvious DNA-binding activity was observed with the 67-bp substrate S1, but not with S2 or S3 in EMSA assays (Fig. 3A). A competition assay confirmed the specificity of the binding of Ms2173 with the S1 DNA fragment (Fig. 3B). When unlabeled S1 or S2 DNA substrates were tested for their ability to compete with the labeled S1 DNA, only S1, but not S2, could inhibit the binding of Ms2173 to the labeled S1 DNA substrate (Fig. 3B).

FIGURE 3.

EMSA assays for the DNA-binding specificity of Ms2173. A, DNA substrates designed for EMSA assays. Three small fragments, S1 (67 bp), S2 (67 bp), and S3 (50 bp), that together cover the full-length of the 158-bp upstream promoter sequence of Ms2173, were synthesized. Three labeled small fragments (40 nm) were coincubated with increasing amounts of Ms2173 (0–6 μm). Ms2173 bound to the S1 fragment, but not to the S2 or S3 fragments. B, assays for DNA-binding specificity of Ms2173. The unlabeled DNA substrates S1 and S2 (40–160 nm) were tested for their ability to compete with the labeled S1 substrate (40 nm) for binding with Ms2173. Unlabeled S1, but not S2, DNA substrates could competitively inhibit the binding of Ms2173 to the labeled S1 DNA substrate.

The binding motif for Ms2173 was mapped by further DNase I footprinting assays. As shown in Fig. 4A, the region around GTGAACCAGTGGAGCACTGGCCTAG was protected on the coding strand when increasing amounts of Ms2173 protein (0–6 μm) were coincubated with DNaseI. Similarly, the region around CTAGGCCAGTGCTCCACTGGTTCAC was protected when the non-coding DNA strand was used as the substrate (Fig. 4B). Interestingly, Cu²⁺ (100 μm) was able to reduce the protection of these DNA regions by Ms2173 (Fig. 4C, lane 5), which is consistent with the inhibitory effect of Cu²⁺ on the DNA-binding activity of Ms2173 described above (Fig. 2B).

FIGURE 4.

Identification of the DNA-binding motif for Ms2173. DNaseI footprinting assays for the coding strand (A) and the non-coding strand (B) and the effect of Cu²⁺ (100 μm) on the ability of increasing amounts of Ms2173 (0–6 μm) (lanes 1–4) to protect the target DNA against DNaseI digestion (C). The ladders are shown, and the corresponding nucleotide sequence is listed (lanes 5–8). The protected region on the coding strand and non-coding strand are indicated by a black bar. The protected sequences are underlined. D, sequence and structural characteristics of the promoter DNA region protected by Ms2173. The protected regions are underlined, and the 15-bp sequences containing inverted repeats (IR) separated by 3 bp are shown in boldface. The translation start codons of Ms2173 and Ms2172 are indicated in boldface. E, EMSA assays for the DNA-binding activity of Ms2173 on DNA substrates without (S4) or with mutations within the IR sequence (S6 and S7) or the separation sequence (S5). Each DNA substrate was coincubated with 0–6 μm Ms2173 protein (right panel). Ms2173 could bind to the S4 and S5 substrates but not the S6 and S7 substrates.

The protected DNA region extended from positions −127 to −113 (relative to the translational start codon) in the coding strand and from positions −46 to −32 in the non-coding strand (Fig. 4D). A palindromic motif formed by two inverted repeats (5′-CCAGTG-3′) separated from each other by three nucleotides (Fig. 4D) was found in this protected sequence. Further EMSA assays confirmed the significance of this motif for specific recognition by Ms2173 (Fig. 4E). Ms2173 was incapable of binding with the S6 and S7 fragments, in which the two inverted repeats were mutated. In contrast, Ms2173 bound well with the S5 DNA substrate in which only the three nucleotides between the two inverted repeats were replaced (Fig. 4E). These results indicate that Ms2173 recognizes and binds to an essential palindromic sequence motif.

Ms2173 broadly Recognizes the Promoters of Diverse Genes, Including 37 Membrane-associated Transporter Genes

A total of 292 potential target genes were identified when the intergenic regions of the M. smegmatis genome were searched on the basis of the sequence motif identified above (supplemental Table S6). We then analyzed the classification and number of these target genes in the context of clusters of orthologous groups (COG) categories. As shown in Fig. 5, A and B, among several defined COG categories, strikingly, the potential targets included promoters of 37 membrane-associated transporter genes (supplemental Table S7). Using the WebLogo tool (35), a more general conserved motif sequence for Ms2173 binding was characterized, as shown in Fig. 5C. Furthermore, target promoters of all 37 of these genes could be specifically recovered by chromatin immunoprecipitation using the Ms2173 antibody (Fig. 5D). Importantly, the Ms2173 antibody could not recover Ms0103p, a negative control. Further EMSA experiments also confirmed that Ms2173 could bind with many upstream DNA regions of these target genes (supplemental Fig. S2). These results indicate that Ms2173 can broadly recognize the promoters of a large number of membrane-associated transporter genes.

FIGURE 5.

Functional categories of Ms2173 target genes and assays for DNA-binding activity of Ms2173 in vivo. A, functional categories of Ms2173 target genes in M. smegmatis. A functional classification of the target genes was conducted in the context of COG categories. B, the inverted repeat motif sequence CCAGTGGAGCACTGG was used to search the intergenic regions in the M. smegmatis genome. Promoters of 37 membrane-associated transporter genes were identified and listed. Conserved sequences are highlighted. Other genes included in the same operon are shown in parentheses to the right. C, logo assays for identifying a common motif. The logos were generated by using the MEME software suite. D, ChIP assays for the association of Ms2173 with the target promoter DNA of 37 membrane-associated transporter genes. ChIP using preimmune (P) or immune sera (I) raised against Ms2173. DNA recovered from the immunoprecipitates was amplified with primers specific to the target genes or to the promoter of Ms0103, an unrelated mycobacterial gene.

Ms2173 Negatively Regulates the Expression of Many Membrane-associated Transporter Genes

To characterize the function of Ms2173, an Ms2173-deleted mutant strain of M. smegmatis was generated by a gene replacement strategy (supplemental Figs. S3 and S4). We then compared the expression of some membrane-associated transporter genes in both wild-type and Ms2173-deleted mutant M. smegmatis strains using quantitative real-time PCR assays. As shown in Fig. 6A, expressions of most of the tested genes were significantly up-regulated (p < 0.05) in the mutant strains compared with that in the wild-type strain. However, the expression level of the negative control gene MsSigA, which lacked the conserved motif in its promoter, did not change obviously. This finding suggested to us that Ms2173 could function as a negative regulator in M. smegmatis. Further overexpression experiments also confirmed this observation. As shown in Fig. 6B, expressions of the tested target genes were significantly down-regulated (p < 0.05) when Ms2173 was overexpressed (about 5-fold) through a pMind-derived recombinant plasmid in M. smegmatis.

FIGURE 6.

Differential expression assays of several membrane-associated transporter genes in the wild-type, mutant, and Ms2173-overexpressed strains. The mycobacterial cDNA was amplified, and quantitative real-time PCR assays were performed as described under “Experimental Procedures.” Relative expression levels of the genes were normalized using the 16S rRNA gene as an invariant transcript, and an unrelated sigA gene was used as a negative control. Relative expression levels of target genes in Msm and Msm/ΔMs2173 (A), in Msm/pMindD and Msm/pMindD-Ms2173 (B), and in Msm/pMindD-Ms2173 strains in response to different metal ions (C) were assayed. As a positive control, total DNA of each strain was used as template for PCR amplification. The cDNA of the mutant strains and the recombinant strain containing an empty pMindD vector was used as template in the negative controls. Data were analyzed using the 2^ΔΔCt method (30). Relative expression data were analyzed for statistical significance by the unpaired two-tailed Student's t test using GraphPad Prism5. *, p ≤ 0.05.

Next, we compared relative gene expression levels in response to different metal ions in the Ms2173-overexpressed strains. As shown in Fig. 6C, expressions of the tested target genes were significantly up-regulated (p < 0.05) in response to Cu²⁺ but not to Zn²⁺. This result is similar (albeit weaker) to that observed in the Ms2173-deleted mutant M. smegmatis strain shown above (Fig. 6A), suggesting that Cu²⁺, but not Zn²⁺, could counteract the repressive regulation of target genes by Ms2173. Taken together, our results indicate that Ms2173 negatively regulates the expression of its target genes and that this function of Ms2173 is sensitive to Cu²⁺.

Ms2173 Negatively Regulates Drug Resistance in M. smegmatis

We determined mycobacterial growth curves to examine the regulatory effect of Ms2173 on the growth of M. smegmatis in response to INH and RIF. As shown in Fig. 7A, compared with the wild-type strain with a minimal inhibition concentration value of 200 μg/ml, the Ms2173-overexpressed M. smegmatis strain grew extremely slowly in the Middlebrook 7H9 medium containing 30 μg/ml INH. Conversely, the Ms2173-deleted mutant exhibited faster growth compared with the wild-type strain under the same conditions (Fig. 7B). When expressing the Ms2173 gene through a pMind plasmid in the Ms2173-deleted mutant M. smegmatis, the recombinant strain reobtained a slow growth in the Middlebrook 7H9 medium containing 30 μg/ml INH (supplemental Fig. S5), indicating that the growth differences between Ms2173-deleted mutant and the wild-type strain were due to loss of Ms2173. Similarly, Ms2173-overexpressed M. smegmatis strains grew slowly in the 7H9 medium containing 2 μg/ml RIF. However, growth of the Ms2173-deleted mutant strain was comparable with that of the wild-type strain at this concentration of RIF (supplemental Fig. S6). No obvious growth difference between the wild-type and recombinant strains was observed when the drugs were removed from the medium (supplemental Fig. S6).

FIGURE 7.

Assays for the effect of Ms2173 on mycobacterial growth in response to 30 μg/ml isoniazid in the presence or absence of metal ions. Growth curves of the wild-type, Ms2173-deleted, and Ms2173-overexpressed strains were determined as described under “Experimental Procedures.” A, the Msm/pMindD and Msm/pMindD-Ms2173 strains were grown in 7H9 medium to which 10 μg/ml Kan, 25 ng/ml Tet, and 30 μg/ml INH had been added. B, the Msm and Msm/ΔMs2173 strains were grown in 7H9 medium with 30 μg/ml INH. C, for comparing the effects of different metal ions (2.5 μm Cu²⁺ or Zn²⁺) on the growth of the Ms2173-overexpressed strains in response to 30 μg/ml INH, the Msm/pMindD-Ms2173 strain was cultured in 7H9 medium with 10 μg/ml Kan and 25 ng/ml Tet in the presence of 30 μg/ml INH (left panel), and in the absence of INH (right panel). Representative growth curves are shown.

We further determined the lowest concentration of INH (MIC value) resulting in complete inhibition of growth or in growth of ≤ 1% of the initial inoculum of different mycobacterial strains. Compared with the wild-type strain with a MIC value of 200 μg/ml, the Ms2173-overexpressed M. smegmatis strain had a lower MIC value of 160 μg/ml. Conversely, the Ms2173-deleted mutant had a higher MIC value of 300 μg/ml.

These results indicate that Ms2173 negatively regulates mycobacterial INH and RIF resistances. Taken together, our observations support a model in which Ms2173 represses the expression of many membrane-associated drug efflux pumps and thus inhibits the bacterial ability to get rid of drugs from the cell (Figs. 5 and 6).

Cu²⁺ Can Partially Reverse the Function of Ms2173 in Vivo in M. smegmatis

We have shown that Cu²⁺ can inhibit the ability of Ms2173 to bind DNA (Fig. 2) and to negatively regulate the expression of its target genes (Fig. 6). If the above-mentioned growth phenotypic changes are mediated by the DNA-binding ability and transcriptional regulatory activity of Ms2173, Cu²⁺ should reverse or at least decrease the severity of these phenotypes. To test this possibility, we examined the effect of Cu²⁺ on the growth of the Ms2173-overexpressed M. smegmatis strain in the presence of INH. As shown in Fig. 7C, 2.5 μm Cu²⁺ could strongly reduce the inhibition of 30 μg/ml INH on the growth of the overexpression strain (Fig. 7A), although Cu²⁺ had a certain toxicity to the mycobacterial growth (supplemental Fig. S7). This effect was specific to Cu²⁺, as addition of Zn²⁺ did not produce the same result (Fig. 7C). In addition, when INH was removed from the medium, no obvious difference could be observed in the growth of the overexpression strain in response to the different ions (Fig. 7D and supplemental Fig. S8).

DISCUSSION

Membrane-associated transport proteins play important roles in bacterial adaptation to environmental stresses, including antibacterial drugs. With recent success in sequencing bacterial genomes, many of these transporters, including the MDR transport proteins, have been identified. However, the regulators and regulatory mechanisms involved in the expression of most of these genes remain largely unknown. In this study, we have identified Ms2173, a GntR/FadR family transcription factor, as a novel global transcriptional regulator in M. smegmatis. We further uncovered that Ms2173 underlies a new Cu²⁺-responsive pathway for the regulation of mycobacterial membrane-associated transporter expression and, thereby, the regulation of drug resistance in M. smegmatis.

The genome of the fast-growing bacterium M. smegmatis encodes more than 500 potential transcriptional regulation genes. Interestingly, several broad regulators have been characterized in M. smegmatis and other bacterial species (36). In a recent study, Yang et al. (37) found that M. smegmatis Ms6564, a TetR-like transcription factor, is involved in regulating the expression of a large number of DNA repair/damage genes. In this study, M. smegmatis Ms2173, a GntR/FadR-like protein, was confirmed as a candidate for broadly regulating the expression of membrane transport genes and other genes with diverse functions. Regulators of the GntR/FadR family are widely distributed among bacteria (22), and they are known to be involved in the regulation of multidrug resistance, biosynthesis of antibiotics, osmotic stress, and virulence (38, 39). In this study, using DNaseI footprinting and EMSA assays, a 15-bp palindromic sequence required for specific recognition by Ms2173 was identified. Furthermore, we identified the conserved binding motif for Ms2173 within the promoters of 292 M. smegmatis genes or operons. These potential target genes covered a variety of gene families, including membrane transport genes, transcriptional regulators, acyl-coA-related genes, and many metabolism genes (Fig. 5 and supplemental Table S6). On the basis of this result, we conclude that Ms2173 functions as a broad transcriptional regulator in M. smegmatis. Interestingly, an ortholog of Ms2173, Rv0494, was also found in the pathogen M. tuberculosis H37Rv (supplemental Fig. S9).

Bacterial drug efflux pumps have been categorized into five families, including the ATP-binding cassette superfamily, the major facilitator superfamily, the multidrug and toxic compound extrusion family, the small multidrug resistance family, and the resistance-nodulation-division superfamily (3). These transporters export a wide array of substrates from the inside to the outside of the cell and contribute to bacterial drug resistance (40, 41). In this study, Ms2173 was found to regulate the expression of 37 potential membrane-associated transporter genes. Some of these target genes belong to several superfamilies of bacterial drug efflux pumps. For example, 13 of the 37 genes are hypothetical proteins that belong to the ATP-binding cassette superfamily. Four proteins encoded by Ms0716, Ms1329, Ms3316, and Ms6375 belong to the major facilitator superfamily (supplemental Table S7). In addition, a drug resistance transporter of the Bcr/CflA superfamily, encoded by Ms5047, was also one of the targets identified in our study (supplemental Table S7). Interestingly, the list of target genes also included Ms0695, which encodes an isoniazid inducible protein, IniA. The operon iniBAC, covering the iniA gene, was shown previously to confer multidrug resistance to Mycobacterium bovis Bacille de Calmette Guerin through an associated pump-like activity (42, 43). A recent study further found that the operon iniBAC is regulated by a two-component regulator, MtrA, which is involved in drug resistance in M. smegmatis (13). Our findings thus support our model that Ms2173 is involved in global regulation of mycobacterial drug resistance.

Many bacterial regulators have been characterized previously to regulate individual operons involved in drug resistance (3). Very few broad regulators and their associated target operons have been characterized so far. MarA, as well as its homologs, is one of the few examples of genes that have been characterized as a global regulator. MarA is known to function as an activator, and only some of their targets have been characterized clearly (3). In this study, we found that Ms2173 regulates multiple drug resistance-associated genes. Furthermore, by constructing recombinant strains overexpressing Ms2173 and Ms2173-deleted mutant strains and analyzing the expression of target genes in these strains by quantitative real-time PCR assays, we found that, unlike MarA, Ms2173 acts as a repressor for the expression of its targets. Our results support a model in which Ms2173 prevents constitutive expression of these membrane transporters, thereby conferring good drug responsiveness to wild-type M. smegmatis strains. Consistent with this, we observed that the Ms2173-overexpressed strains were more sensitive to the drug than the wild-type strains (Fig. 7). In contrast, the Ms2173 deletion strains displayed much stronger drug resistance than the wild-type strain (Fig. 6). Our results show that Ms2173 is a global repressor that negatively regulates the expression of multiple drug efflux pumps in M. smegmatis.

Metal ions, including copper, iron, and zinc, have important roles in all living cells. In intracellular pathogens such as M. tuberculosis, inorganic ions have been shown to be involved in regulating the expression or activity of some important proteins (44). For example, copper is essential for the ability of the protein SodC to control an intracellular oxidative burst during macrophage survival (45). Three major prokaryotic copper metalloregulators (CopY from Enterococcus hirae, CueR from E. coli and B. subtilis, and CsoR from M. tuberculosis) have been identified and characterized in bacteria (46–49). Until now, only a FadR-like regulator, TM0439, has been found to bind metal ions (Zn²⁺) (50). However, the regulatory mechanism and functional effects of the metal ions on the regulator remains unclear. Recently, a copper-responsive transcription factor, RicR, was found to function as a global regulator in M. tuberculosis (51), which is a dimer in solution and belongs to the CsoR family (49). However, Ms2173 had no sequence similarity to the RicR in M. tuberculosis (data not shown). In this study, Ms2173 was characterized as a Cu²⁺-responsive regulator. We confirmed that Cu²⁺ could specifically counteract the repressive effects of Ms2173 on the expression of its target genes and on bacterial drug resistance. These findings indicate that Ms2173 is different from RicR and other FadR proteins (27, 49, 50) and is a new copper ion-responsive transcription factor. Interestingly, the potential target genes of Ms2173 include a copper resistance protein, CopC (Ms6436), and a metal ion channel membrane protein (Ms1945). This implies that Ms2173 could be involved in regulation of the intracellular concentration of copper ions in M. smegmatis. Taken together, copper ions may cross-talk with Ms2173, which regulates the activity of multiple drug efflux pumps and, therefore, affect the drug resistance of M. smegmatis.

In summary, a new transcription factor belonging to the GntR/FadR-like family was successfully isolated and characterized in M. smegmatis. The promoters of about 292 M. smegmatis genes or operons were characterized as its potential targets. Notably, Ms2173 was found to be involved in the regulation of the expressions of 37 membrane transport genes and to negatively affect mycobacterial drug resistance. Copper ions could inhibit the DNA-binding activity of Ms2173 in vitro and partially counteract its in vivo functions. Our findings establish Ms2173 as a novel copper ion-responsive global repressor in mycobacteria and significantly enhance our understanding of the regulatory mechanism and signaling pathway associated with bacterial drug resistance.