The Mycobacterial Transcriptional Regulator whiB7 Gene Links Redox Homeostasis and Intrinsic Antibiotic Resistance*

Background: The whiB7 gene is a multidrug resistance determinant in mycobacteria. Results: WhiB7 autoregulates its own promoter in response to both antibiotics (in a structure- and target-independent manner) and to redox changes. Conclusion: WhiB7 links cell metabolism, redox homeostasis, and antibiotic resistance. Significance: Understanding antibiotic-induced metabolic stress responses in mycobacteria may lead to therapies for mycobacterial diseases, including tuberculosis. Intrinsic drug resistance in Mycobacterium tuberculosis limits therapeutic options for treating tuberculosis. The mycobacterial transcriptional regulator whiB7 contributes to intrinsic resistance by activating its own expression and many drug resistance genes in response to antibiotics. To investigate whiB7 activation, we constructed a GFP reporter to monitor its expression, and we used it to investigate the whiB7 promoter and to screen our custom library of almost 600 bioactive compounds, including the majority of clinical antibiotics. Results showed whiB7 was transcribed from a promoter that was conserved across mycobacteria and other actinomycetes, including an AT-rich sequence that was likely targeted by WhiB7. Expression was induced by compounds having diverse structures and targets, independent of the ability of whiB7 to mediate resistance, and was dependent on media composition. Pretreatment with whiB7 activators resulted in clinically relevant increases in intrinsic drug resistance. Antibiotic-induced transcription was synergistically increased by the reductant dithiothreitol, an effect mirrored by a whiB7-dependent shift to a highly reduced cytoplasm reflected by the ratio of reduced/oxidized mycothiol. These data provided evidence that intrinsic resistance resulting from whiB7 activation is linked to fundamental changes in cell metabolism.

Intrinsic drug resistance in Mycobacterium tuberculosis limits therapeutic options for treating tuberculosis. The mycobacterial transcriptional regulator whiB7 contributes to intrinsic resistance by activating its own expression and many drug resistance genes in response to antibiotics. To investigate whiB7 activation, we constructed a GFP reporter to monitor its expression, and we used it to investigate the whiB7 promoter and to screen our custom library of almost 600 bioactive compounds, including the majority of clinical antibiotics. Results showed whiB7 was transcribed from a promoter that was conserved across mycobacteria and other actinomycetes, including an ATrich sequence that was likely targeted by WhiB7. Expression was induced by compounds having diverse structures and targets, independent of the ability of whiB7 to mediate resistance, and was dependent on media composition. Pretreatment with whiB7 activators resulted in clinically relevant increases in intrinsic drug resistance. Antibiotic-induced transcription was synergistically increased by the reductant dithiothreitol, an effect mirrored by a whiB7-dependent shift to a highly reduced cytoplasm reflected by the ratio of reduced/oxidized mycothiol. These data provided evidence that intrinsic resistance resulting from whiB7 activation is linked to fundamental changes in cell metabolism.
The continued rise and prevalence of antibiotic resistance have progressively limited the use of available antibiotics and highlighted a need for new directions in the search for novel and effective antimicrobial therapies (1). Antibiotic resistance is typically mediated by proteins that are transcriptionally activated in response to drug exposure. Chemical inhibitors of drug-activated regulatory proteins or the products of corre-sponding resistance genes can potentiate antibiotic activity (2,3). Studies of antibiotic activities have traditionally focused on their abilities to inhibit specific targets essential for bacterial growth, including cell wall biosynthesis, transcription, translation, or DNA replication, with the implicit assumption that these are the direct causes of growth arrest or cell death (4). However, antibiotics exert diverse secondary effects that might generate signals of metabolic distress. In Escherichia coli, antibiotics that inhibit ribosomal function induce heat or cold shock responses (5). Importantly, antibiotics alter cell metabolism by activating transcription of up to 1 in 20 promoters in a concentration-and medium-dependent manner (6). These global changes of gene expression can help predict the mode of action of an antibiotic (7). Furthermore, recent studies indicate that oxidative stress induced by bactericidal antibiotics might be a common cause of cell death (8). Understanding the activation and function of the global response to antibiotic exposure may provide new strategies to overcome antibiotic resistance.
Mycobacterium tuberculosis, the etiologic agent of tuberculosis, is intrinsically resistant to most clinically available antibiotics. Intrinsic resistance systems include the low permeability of the cell envelope, drug exporters, drug-modifying systems, and target modifying enzymes (2). Intrinsic antibiotic resistance to any given drug may be determined by an interactive network, including effector proteins, regulatory proteins, and inducers (9). WhiB7, the focus of our study, is a transcriptional regulator of genes that contribute to intrinsic antibiotic resistance in mycobacteria (10).
The WhiB family of transcriptional regulators is found only in actinomycetes. These include benign and pathogenic mycobacteria as well as Streptomyces, producers of the majority of known antibiotics. In mycobacteria, the WhiB family has vital roles in fundamental cell processes, including cell division, redox homeostasis, virulence, and antibiotic resistance (10 -12). Little is known about the signals that activate transcription of the whiB genes.
Upon antibiotic treatment, whiB7 has an essential role in activating transcription of antibiotic resistance systems for tetracyclines, macrolides, lincosamides, and aminoglycosides (10).
Transcription of whiB7 is induced by antibiotic treatment as well as a variety of stress conditions such as heat shock, iron starvation, and entry into stationary phase (13). Importantly, whiB7 is highly induced in clinical isolates of the M. tuberculosis complex soon after infection of resting and activated murine macrophages (14). The observation that constitutive expression of whiB7 in trans activated transcription of its native genomic allele suggested autoregulation (10). Like other WhiB proteins, WhiB7 contains four cysteine residues that can either bind an iron-sulfur cluster or form disulfide bridges within the apoprotein (15). WhiB proteins are redox-sensitive and play critical roles in transcriptional activation (12,16,17). Here, we identify the essential motifs of whiB7-dependent promoters and explore the mechanism by which the WhiB7 protein responds to diverse antibiotics and physiologies to activate expression of intrinsic drug resistance genes.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-E. coli TOP10 (Invitrogen) was used for cloning. It was cultured in LB broth (Sigma) supplemented with appropriate antibiotics shaking at 200 rpm, 37°C. Unless otherwise specified, Mycobacterium smegmatis mc 2 155 was cultured in Middlebrook 7H9 broth (Difco) supplemented with 10% albumin dextrose catalase, 0.2% (v/v) glycerol, 0.05% tyloxapol, and appropriate antibiotics at 37°C either shaking in flasks at 200 rpm or rolling in test tubes. Kanamycin was used at a final concentration of 30 g/ml, hygromycin at 50 g/ml, apramycin at 50 g/ml, and ampicillin at 200 g/ml.
Cloning-All PCRs were performed using Dynazyme EXT (New England Biolabs F-505S) according to the manufacturer's instructions. Reactions included 5% (v/v) dimethyl sulfoxide. Restriction enzymes were from New England Biolabs; digests were performed according to manufacturer's instructions. Ligations were performed using T4 DNA ligase (Invitrogen 15224-041) overnight at 16°C. Strain and plasmid constructions are summarized in Table 1. Oligonucleotides used are listed in supplemental Table S1.
Construction of the Lux Reporters-A DNA fragment containing nucleotides ϩ2 to ϩ674 upstream of the annotated start codon of whiB7 (Rv3197a) was amplified from M. tuberculosis H37Rv genomic DNA by PCR using the primers TBpromF and TBpromR. The fragment was digested with AclI and EcoRI and cloned into pMV361 digested with the same enzymes to create pMV361-P B7TB . The EcoRI fragment containing the luxAB-CDE genes was isolated from pAmilux (18) and cloned into EcoRI-digested pMV361 and pMV361-P B7TB to create the HSP60 promoter-driven, constitutively active pLUXon and the antibiotic-inducible whiB7 promoter-driven pTB674lux, respectively.
Construction of the GFP Reporters-A DNA fragment containing the egfp gene from pEGFP (Clontech) was amplified by PCR using the primers eGFP_F and eGFP_R. The fragment was cloned, using the polymerase-added A overhang, into pGEM-Teasy (Promega) to create pAB2. A DNA fragment containing the first three amino acids of the annotated whiB7 (MSMEG_1953) and 689 nucleotides upstream region was amplified from the genome of M. smegmatis mc 2 155 by PCR using the primers smegpromF and smegpromR. The fragment was cloned into pGEM-Teasy to create pAB6. pAB6 was digested with ClaI and HindIII, and the whiB7 promoter-containing fragment was cloned into pMycVec1 digested with the same enzymes to create pAB7. pAB2 was digested with HindIII and XbaI, and the egfp-containing DNA fragment was cloned into pAB7 digested with the same enzymes to create antibioticinducible whiB7 promoter-driven pMS689GFP.
The efgp from pMS689GFP was amplified by PCR using the primers HSPGFP_F and HSPGFP_R. The PCR fragment was digested with BamHI and EcoRI and cloned into pMV261 digested with the same enzymes, to create the HSP60 promoterdriven constitutively active pGFPon.
Subcloning the GFP Reporter Promoter-Shorter fragments of the upstream whiB7 region were amplified from pMS689GFP. The PCR product of GFPsub_R and 438_F was digested with AclI and cloned into ClaI-and EcoRV-digested pMycVec1 to create pMS438GFP. The PCR products of 497_F and 483_F in combination with GFPsubX_R were digested with ClaI and XbaI and cloned into ClaI/XbaI-digested pMycVec1 to create pMS497GFP and pMS483GFP, respectively.
Complementation of whiB7 KO Strain-The whiB7 gene, including its 520-bp upstream and 152-bp downstream regions, was amplified by PCR from the M. smegmatis genome using primers B7comp_F and B7comp_R. The PCR fragment was cloned into pMV361, replacing the HSP60 promoter, using XbaI and HindIII to construct p361comp. A 1347-bp region of pAB707 containing the apramycin resistance gene aac(3)IV was amplified by PCR using the primers 707apra_F and 707apra_R. The product was cloned into p361comp using Hin-dIII and SpeI to construct p361comp.apra. The integrative p361comp.apra was then used to complement the pMS689GFP containing M. smegmatis whiB7 KO.
Construction of ermMT Overexpression Strain-Rv1988 (ermMT) was amplified from M. tuberculosis H37Rv genomic DNA by PCR using the primers Rv1988_F and Rv1988_R. The PCR product was digested with EcoRI and HindIII and cloned into pMV361 digested with the same enzymes. The resulting vector was digested with MfeI and ClaI, and the ermMT-containing fragment was cloned into pMV361.hyg digested with the same enzymes to create pERM.hyg. M. smegmatis was transformed with the plasmid; transformants were selected by hygromycin, and an increased resistance to macrolides was checked.
Luminescence Time Course-M. smegmatis harboring pTB7lux was inoculated into 3 ml of 7H9 ϩ kanamycin and grown for 55-60 h, followed by an incubation at room temperature for 12-15 h. The culture was diluted to an A 600 nm of 0.01-0.005 into 3 ml of kanamycin-free 7H9 and grown to an A 600 nm of 0.6 -0.8 for ϳ20 -24 h. Finally, the culture was diluted to an A 600 nm of 0.2 and 200 l was distributed into a black clear bottom 96-well plate (Costar 3631). The appropriate compound was added to the desired concentration, and the plate was placed immediately into a VarioskanFlash (Thermo Scientific). Using the SkanIt RE 2.4.3 software, the Varioskan-Flash was set to raise the temperature to 37°C and without waiting the plate was shaken for 15 s at 420 rpm, followed by measuring the A 600 nm and finally measuring luminescence for 1000 ms. The shaking, absorbance reading, and luminescence reading was repeated 12 times at 30-min intervals, three times at 1-h intervals, and finally twice at 3-h intervals. M. smegmatis wild type and whiB7 KO harboring pLUXon were prepared in identical fashion but diluted to an A 600 nm of 0.2 after the room temperature incubation and the time course had run.
Fluorescence Time Course-M. smegmatis harboring pMS689GFP was inoculated into 3 ml of 7H9 ϩ kanamycin and grown for 48 -54 h until a final A 600 nm of 6 -8. The culture was diluted to an A 600 nm of 0.01-0.005 into 3 ml of kanamycin-free 7H9 and grown to an A 600 nm of 0.6 -0.8 for ϳ20 -24 h. The culture was diluted to an A 600 nm of 0.2 and dispensed into a black, clear bottom 96-well plate (8 rows (A-H) each containing 12 wells). The appropriate compound was added to the desired concentration, and the plate was covered with Microseal B Film (Bio-Rad MSB1001) and placed immediately into a VarioskanFlash, and fluorescence was monitored. Briefly, the temperature was raised to 37°C, and without waiting the plate was shaken for 15 s at 420 rpm, followed by measuring the A 600 nm and finally measuring fluorescence under default settings by excitation at 488 nm and measuring emission at 509 nm. The shaking, absorbance measurement, and fluorescence measurement were repeated every 30 min 30 more times. M. smegmatis wild type and whiB7 KO harboring pGFPon was prepared in identical fashion but diluted to an A 600 nm of 0.2 after the 48 -54-h incubation and the time course was run.
WhiB7 Activation Assay-M. smegmatis harboring pMS689GFP was inoculated into 3 ml of 7H9 ϩ kanamycin and grown for 48 -54 h to a final A 600 nm of 6 -8. The inoculum was diluted to an A 600 nm of 0.0125 in 50 ml of kanamycin-free 7H9 and grown for ϳ20 h to an A 600 nm of 0.6 -0.8. The culture was diluted to an A 600 nm of 0.2 and dispensed into a black clear bottom 96-well plate. Wells 1-11 of row A received 300 l, and the rest of the plate received 200 l, except for wells in column 12 of rows G and H, which received 200 l of sterile media. For the wells that received 300 l of culture, 3 l of stock 5 mM

Insights into whiB7 Activation and Action
compounds was added, and 100 l was used to serially dilute down from rows A to H creating a concentration range of 50 to 0.02 M in a final volume 200 l. The plate was covered with a lid, wrapped in tin foil, and incubated at 37°C, 200 rpm for 5 h. After the incubation the plate was placed in a Varioskan Flask, the A 600 nm and emission at 509 nm after 488 nm excitation were measured following shaking for 15 s at 420 rpm. The assay is presented graphically in supplemental Fig. S4. The absorbance and fluorescence values were corrected by subtracting the average of the values measured in wells 12 of rows G and H (sterile media) (supplemental Fig. S4). The fluorescent values were then standardized to the absorbance by dividing the fluorescent value of a well by its corresponding absorbance value. Finally, the fold increase of standardized fluorescence was calculated by dividing values by the average of the untreated wells (wells 12 of row A to F). Because EGFP 2 continued to fluoresce in cultures whose A 600 nm had been reduced by antibiotic-induced lysis (data not shown), wells with an A 600 nm less than one-third of the untreated wells were disregarded. The concentration range for each compound was analyzed in three biologically independent runs. A compound was considered a WhiB7 activator if there was at least a Ն4-fold increase in fluorescence versus the untreated control in two out of the three replicates.
Minimum Inhibitory Concentration Determination-M. smegmatis wild type or whiB7 KO was inoculated into 3 ml of 7H9 and grown for 48 -54 h to a final A 600 nm of 6 -8. The culture was diluted to an A 600 nm of 0.005, and 100 l was added to 100 l of antibiotic containing 7H9 at 2-fold serial dilutions across a 96-well plate (Costar 3370). The plate was incubated for 48 h, 30 l of sterile 10 mg/100 ml (w/v) resazurin solution was added, and the plate was incubated for an additional 24 h. Wells that remained blue were deemed to contain an inhibitory concentration of antibiotic.
mRNA Isolation and Quantification-M. smegmatis was grown to an A 600 nm of 0.6 -0.8 and split into 30 -50-ml aliquots. The appropriate amount of antibiotic was added to selected aliquots, and the cultures were incubated for 1 h. Four times culture volume of GTC buffer (5 M guanidine thiocyanate, 17 mM sodium lauroyl sarcosinate, 28.5 mM trisodium citrate, 0.5% (v/v) Tween 80, 0.7% (v/v) 2-mercaptoethanol) was added, and the samples were incubated for 1 h at room temperature. Samples were pelleted by centrifugation for 10 min, 5000 rpm, 4°C, and the supernatant was discarded. The pellets were suspended in 1 ml of QiaZol (Qiagen 79306) and transferred to a 2-ml screw cap tube (MBP 3488) containing ϳ100 l of 0.1-mm glass beads (BioSpec 11079101). The tubes were beaten three times using an MP FastPrep-24 at 6.0 m/s for 45 s with 3-5-min ice breaks. The samples were centrifuged for 5 min at 16,000 ϫ g, 4°C, and the supernatants were transferred to phase gel lock tubes (5 Prime 2302830). The tubes were incubated at room temperature for 5 min, and 0.2 ml of ice-cold chloroform was added. The tube contents were mixed by inversion for 15 s, incubated at room temperature for 3 min, and finally centrifuged for 5 min at 16,000 ϫ g, 4°C. The upper fraction was transferred to a 1.5-ml Eppendorf tube containing 550 l of 30 mM sodium acetate in isopropyl alcohol, mixed well, and incubated overnight at Ϫ20°C. The samples were centrifuged for 10 min at 16,000 ϫ g, 4°C; the supernatant was discarded, and the pellets were washed with 1 ml of ice-cold 75% (v/v) ethanol. Samples were centrifuged for 5 min at 16,000 ϫ g, 4°C, and the supernatant was discarded. The pellets were dissolved in 90 l of RNase-free water by incubation at 65°C for 10 min. Samples were then treated for 30 min at 37°C with Turbo DNase (Ambion AM2239), and finally the RNA was isolated by RNAspin mini columns (GE Healthcare 25-0500-72) according to the manufacturer's instructions.
The qScript cDNA synthesis kit (Quanta 95047-100) was used to reverse-transcribe a total of 100 ng of RNA per 20-l reaction (25°C for 5 min, 42°C for 30 min, and 85°C for 5 min). The cDNA samples were diluted 1:10, and 2.5 l was used per 25 l quantitative PCR. A mixture of PerfeCTa SYBR Green supermix (Quanta 95054-050), cDNA, and primers (1 M each) was run on a Bio-Rad Opticon2 (step 1: 95°C for 3 min; step 2, 95°C for 30 s; step 3, 55°C for 30 s; and step 4, 72°C for 30 s, read, and repeated 34 times from step 2). A standard curve of genomic DNA was used to calculate concentrations and a nonreverse-transcribed control used to estimate DNA contamination. Primers used for whiB7 were prAB47a and prAB48. Concentrations were standardized to an internal control, mysA, using primers prAB49 and prAB50. Fold increase of whiB7 was calculated against a nontreated sample run in parallel.
Transcription Start Site Determination-Transcriptional start site of the antibiotic-induced whiB7 promoter was identified by directed mapping of transcription start site method developed by Mendoza-Vargas et al. (19). Briefly, RNA was isolated from retapamulin-treated M. smegmatis by the method mentioned above. About 1.5 g of total RNA was mixed with 700 pmol of a random hexamer primer (NNNNNN), and the solution was heated to 70°C for 10 min followed by incubation on ice for 5 min. cDNA was generated using transcriptor reverse transcriptase (Roche Applied Science 03 351 317 001) according to the manufacturer's instructions (25°C for 20 min, 60°C for 40 min, and 85°C for 5 min) and isolated by the QIAquick PCR purification kit (Qiagen 28106). A guanine tail was added to the 3Ј end of the cDNA library using terminal transferase (Roche Applied Science 03 333 566 001) according to the manufacturer's instruction (37°C for 30 min, 70°C for 10 min), and the cDNA was once again isolated by the QIAquick PCR purification kit. Linear amplification of the tailed cDNA library was carried out using Dynazyme EXT in combination with the enrich_C primer (94°C for 10 min, 94°C for 1 min, 50°C for 1 min, and 72°C for 3 min, repeated from step two 34 times, 72°C for 10 min), and the cDNA was isolated by the QIAquick PCR purification kit. About 15 ng of the DNA was used per 50-l PCR using Dynazyme EXT with primers TSS_adaptor and whiB7_TSS58 (94°C for 3 min; 94°C for 30 s; 60°C for 10 s; 72°C for 30 s, repeated from step two 34 times, 72°C, 5.5 min). A single PCR product of about 200 bp was observed by agarose gel electrophoresis. This product was cloned, using the polymerase added A overhang, into pGEM-Teasy and transformed into E. coli TOP10. Several white transformants visualized on LB ϩ ampicillin ϩ 20 g/ml X-Gal ϩ 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside agar plates were transferred to liquid media, and plasmid DNA was isolated using the QIAprep spin miniprep kit (Qiagen 27106). The plasmids were sequenced at the University of British Columbia, Vancouver campus, Nucleic Acid Protein Service unit using the M13R primer.
Analysis of Mycothiol Content-M. smegmatis wild type or whiB7 KO was grown in 100 ml of NE (glucose 10 g/liter, yeast extract 2 g/liter, casamino acids 2 g/liter, lab lemco powder 1 g/liter) supplemented to 0.05% tyloxapol at 37°C, 200 rpm to an A 600 nm of 2.0. Cultures were then divided into two 50-ml fractions, and one fraction received erythromycin to a final concentration of 256 g/ml. All fractions were further incubated at 37°C for 1 h. Growth was arrested by the addition of an equal volume of prechilled water. The fractions were then divided for centrifugation at 4°C; 10 ml for MSH determination, 30 ml for MSSM determination, and 5 ml for NEM labeling (negative control). Pellets were flash-frozen in liquid nitrogen and stored at Ϫ80°C until analysis. HPLC analysis of MSH levels was performed as described previously (20).
Nicotinamide Adenine Dinucleotide Quantification-M. smegmatis wild type or whiB7 KO was grown in 50 ml of NE supplemented to 0.05% tyloxapol at 37°C, 200 rpm to an A 600 nm of 2.0. Cultures were then divided into two 25-ml fractions, and one fraction received erythromycin to a final concentration of 256 g/ml. All fractions were further incubated at 37°C for 1 h and samples taken. Erythromycintreated samples were left to incubate for an additional 1 h for the 2-h time point. Concentrations of reduced (NADH) and oxidized (NAD ϩ ) forms of nicotinamide adenine dinucleotide were analyzed as described previously (21). Once samples were extracted, the enzymatic reaction was carried out in a black clear bottom 96-well plate (Costar 3631). Samples and standards were analyzed in triplicate, and only 6 wells were prepared and followed at one time. The Varioskan was set to read the absorbance (570 nm) every 20 s for 5 min. Concentrations were calculated from a standard curve (R 2 ϭ 0.9996) of NADH (Sigma N6660-15VL) and standardized to dry weight/liter of culture.

Fluorescent and Luminescent Reporters to Study whiB7
Promoter Activity-Enhanced green fluorescent protein (EGFP) and bacterial luciferase (LuxABCDE) were evaluated as reporters to study the whiB7 promoter and its response to diverse drugs. Quantitative RT-PCR confirmed that whiB7 transcription was activated in M. smegmatis by erythromycin (an inducer of whiB7 in M. tuberculosis (10)) but not isoniazid ( Table 2). To monitor the promoter controlling the M. smegmatis whiB7 gene, plasmid pMS689GFP was constructed; it encodes the first 9 bp of whiB7 and the 689-nucleotide upstream region (i.e. the entire 513-bp intergenic region and the last 176 bp of the proximal upstream gene, uvrD2) fused to EGFP. An alternative reporter plasmid was also constructed (pTB674lux) containing the upstream intergenic region of the M. tuberculosis whiB7 gene fused to the luxAB-CDE reporter system. To determine whether these constructs contained the antibiotic-inducible whiB7 promoter, erythromycin or isoniazid was added to cultures of M. smegmatis containing pMS689GFP (Fig. 1A) or pTB674lux (Fig.  1B). Erythromycin, but not isoniazid, induced signal in these cultures. Both whiB7 promoter reporter systems consistently displayed similar induction kinetics, resulting in 4 -6-fold increases in signal after 3 h of induction with erythromycin (Fig. 1). Furthermore, the level of erythromycininduced pMS689GFP fluorescence was much lower in a whiB7 mutant (Fig. 1C; whiB7 KO) but was restored upon complementation ( Fig. 1D; whiB7 KOC). This demonstrated that pMS689GFP could be used to monitor antibiotic-induced transcriptional activation of the whiB7 promoter and provided further evidence that WhiB7 played an essential role in activating its own promoter.
To test whether the EGFP or LuxABCDE reporter activities might be dependent on whiB7, both reporters were expressed constitutively from the M. tuberculosis HSP60 promoter (constructs pGFPon and pLUXon) in M. smegmatis wild type or whiB7 KO strains. As expected, pGFPon generated indistinguishable levels of fluorescence in wild type and whiB7 KO strains (Fig. 1E). In contrast, light output by pLUXon was much lower in the whiB7 KO strain background (Fig. 1F). This surprising observation revealed that optimal light production by the LuxABCDE protein complex was dependent on physiological conditions determined by whiB7. In addition, a limited screen of our antimicrobial compounds revealed that at least one drug, the IB-␣ phosphorylation inhibitor BAY 11-7085, abolished bioluminescence from both pTB674lux and pLUXon in M. smegmatis at concentrations below its minimum inhibitory concentration (supplemental Fig. S1). These results showed that the Lux-ABCDE reporter system was dependent on whiB7 and that anti-microbial compounds could also have direct or indirect inhibitory effects on its activity. Although these observations may lead to future insights into the role of whiB7 or drugs on mycobacterial physiology, it was clear that pTB674lux was not a suitable reporter of whiB7 promoter activity, and pMS689GFP was chosen to monitor whiB7 transcriptional activation.
Mapping the Antibiotic-responsive whiB7 Promoter Region-An alignment of the sequences upstream of whiB7 in the genomes of various Mycobacterium species revealed conserved sequences resembling promoter (Ϫ10 and Ϫ35) hexamers and an AT-rich region ( Fig. 2A). This motif was located between nucleotides 438 and 497 upstream of the annotated M. smeg-matis whiB7 gene. To determine whether this region contained the inducible whiB7 promoter, the 689-bp upstream sequence used to construct pMS689GFP was shortened to 497 (pMS497GFP), 483 (pMS483GFP), or 438 (pMS438GFP) nucleotides. Only pMS497GFP responded similarly to pMS689GFP when treated with erythromycin (Fig. 2B). This was also observed using a broad range of whiB7 activators, including retapamulin, tilmicosin, doxycycline, linezolid, A23187, and acivicin (supplemental Fig. S2). A considerably reduced fluorescence response was generated by pMS483GFP, a construction lacking the conserved AT-rich region (Fig. 2B), whereas no fluorescence was detected with pMS438GFP lacking the putative whiB7 promoter (supplemental Fig. S3).
The transcriptional start site of the antibiotic-induced whiB7 promoter was mapped to a position 445 nucleotides upstream of the gene (Fig. 2C). The region contained the predicted whiB7 promoter, and this conserved motif was identified upstream of whiB7 in 12 representative actinomycete genera (KEGG data base; Fig. 2D). In all cases, there was an AT-rich region 3 bases upstream of the conserved Ϫ35 site, TTGNNN, and a conserved Ϫ10 site, TANNNT. The AT-rich motif was present 3 bases upstream of the well characterized eis promoter (22) and also upstream of other genes within the whiB7 regulon (tap and ermMT) (Fig. 2E) (10). Activation of the whiB7 Promoter Is Independent of Antibiotic Structure, Primary Target of Inhibition, and Ability of whiB7 to Mediate Resistance-To investigate whiB7 transcriptional activation, a custom compound library (Sweet library) was assem-bled and screened using the pMS689GFP reporter system. The 591-compound library (listed in supplemental Table S2) included the majority of commercially available antibiotics targeting DNA, RNA, protein, cell envelope synthesis, or essential  (19), and the transcript was cloned. The sequences of two independent clones (bottom) were aligned to the M. smegmatis genome sequence (top). The matching sequence is boxed, and the transcriptional start site is marked transcription start site (TSS). D, alignment of the identified whiB7 promoter motif across 13 genera of actinomycetes. The variable distance from the start codon is boxed. E, putative WhiB7-binding site and promoter motif are present upstream of several M. tuberculosis H37Rv genes within the WhiB7 regulon as well as erm (38), the ortholog of ermMT in M. smegmatis. metabolic conversions, as well as other physiologically active compounds. Unlike other chemical libraries that are dissolved exclusively in dimethyl sulfoxide, each compound in the Sweet library was dissolved in an optimal solvent (water, dimethyl sulfoxide, ethanol, methanol, or dimethylformamide) to ensure solubility and maximize hit discovery. A semi-high throughput screen was devised to assay the library for activators of whiB7 transcription. Each compound was serially diluted in 96-well plates to survey a broad (2,500-fold) concentration range (50 to 0.02 M; supplemental Fig. S4). pMS689GFP fluorescence was monitored after 5 h, when the culture had reached a plateau of signal intensity (Fig. 1A). Hits were identified as compounds that induced a Ն4-fold increase in fluorescence in at least two out of three trials. In total, 86 whiB7 activators were identified (supplemental Table S2). The reliability of the screen was validated by quantitative RT-PCR analysis of 23 representative hits confirming that all induced whiB7 transcription ( Fig. 3 and supplemental Table S2), while four nonhits did not (supplemental Table S2).
Structure clustering analyses (pubchem.ncbi.nlm.nih.gov) allowed visualization of potential similarities in the chemical structures of whiB7 activators (Fig. 3). Compounds with statistically similar structure, suggesting similar bioactivities, are defined by a Tanimoto score greater than 0.7 (23). The cluster analysis identified groups of active compounds having similar structures (including macrolides, aminoglycosides, fluoroquinolones, and tetracyclines). However, the structural similarity between these major clades, as well as most of the other activators, was well below 0.7. Therefore, whiB7 transcriptional activation was not due to a common structural motif of the compounds.
The structural diversity of the identified activators was also reflected in the diversity of their documented targets. Half of the identified whiB7 activators, including macrolides, tetracyclines, lincosamides, and aminoglycosides, inhibited protein synthesis, targeting different sites within the 50 S or the 30 S subunits of the ribosome. The second major target was DNA replication, which was inhibited by fluoroquinolones and DNA intercalators, including netropsin, nogalamycin, and phleomycin. Other potent activators (including acivicin, dequalinium, beauvericin, and A23187) are known for their pleiotropic effects on cell metabolism. These data showed that induction of whiB7 transcription did not result from direct inhibition of a single target or function and implied a common downstream effect of the activators on metabolism. Curiously, cell wall biosynthesis inhibitors, including the extended family of ␤-lactams, did not induce transcription.
To provide further evidence for the concept that direct recognition of antibiotic structure was not needed for whiB7 activation, the M. tuberculosis ribosomal methyltransferase (encoded by ermMT) that confers macrolide resistance by preventing interaction between the macrolide and its ribosomal Identified activators were clustered according to chemical structure similarity using PubChem (pubchem.ncbi.nlm.nih). A Tanimoto score of Ն0.7 is indicative of highly similar structure with potentially similar bioactivity. Numerous activators clustered within distinctive structural classes are highlighted (blue, macrolides; red, aminoglycosides; yellow, fluoroquinolones; green, tetracyclines). Activators confirmed by quantitative RT-PCR are indicated by a red asterisk.
target (24) was constitutively expressed from the HSP60 promoter in M. smegmatis (ermMT OE). If the whiB7 promoter was induced in response to macrolide structure, independent of translation inhibitory effects, the level of pMS689GFP activation would be the same in wild type and in the ermMT-expressing, resistant strain. Alternatively, if whiB7 transcription responded to the stress generated by toxic macrolide-ribosome interactions, macrolides would elicit reduced whiB7 activation in the resistant strain. When M. smegmatis ermMT OE/pMS689GFP was exposed to various macrolides (erythromycin, dirithromycin, roxithromycin, or azithromycin) at concentrations that activated pMS689GFP in the wild type background, the level of induced fluorescence was ϳ5-fold lower (Fig. 4). The level of activation by doxycycline, an antibiotic whose activity is unaffected by ErmMT, was unchanged (Fig. 4).
Interestingly, several activators were not within the whiB7 sensitivity spectrum. For example, although dequalinium, linezolid, danofloxacin, and A23187 were all active inducers of whiB7 expression, the whiB7 KO mutant was not more sensitive to these compounds (Table 2). Furthermore, pretreatment with whiB7 inducers increased intrinsic resistance levels. Pretreatment with sub-inhibitory concentrations of erythromycin or clarithromycin increased resistance to those antibiotics in a whiB7-dependent manner (supplemental Table S3). Similarly, treatment with acivicin, an amino acid analog, also increased resistance to macrolides (4-fold; supplemental Table S4).
The fact that the glutamine analog acivicin was a strong inducer of whiB7 transcription (Table 2) suggested a link between amino acid metabolism and whiB7 activation. When glutamate, the only amino acid in standard 7H9 medium, was either retained or replaced with glutamine, aspartate, asparagine, arginine, or histidine and cultures were treated with erythromycin, levels of whiB7 induction varied; induced levels were similar in cultures grown in the presence of glutamate, aspartate, or histidine, whereas increased levels of induction were observed in cultures grown in the presence of asparagine with a limited increase with glutamine and a possible small increase with arginine (supplemental Fig. S6).

Redox State Modulates the Level of whiB7 Transcriptional
Activation-Based on in vitro studies showing that WhiB proteins, including WhiB7, have oxygen-sensitive iron-sulfur clusters (12,15) that may affect transcription (17), and our studies establishing that whiB7 is needed for regulation of its own promoter, experiments were carried out to explore the hypothesis that oxidizing or reducing reagents (diamide or dithiothreitol (DTT)) might affect whiB7 promoter activity in vivo. In the absence of erythromycin, the thiol reductant DTT moderately induced expression (Fig. 5A). Treatment with DTT in combination with a low dose of erythromycin generated a strong syn-  ergistic (ϳ1300-fold) increase in mRNA levels (Fig. 5A). Conversely, the thiol oxidant diamide generated a negligible increase in whiB7 mRNA and decreased the level of induction by erythromycin ϳ5-fold (Fig. 5A). These data indicate that a reduced state favors whiB7 transcriptional activation.
Kohanski et al. (8) reported that hydroxyl radical formation is a common mechanism of antibiotic-mediated bacterial cell death. Because our experiments demonstrated that reducing, rather than oxidizing, conditions enhanced whiB7 activation, it seemed unlikely that whiB7 transcription was responding to hydroxyl radicals generated by antibiotic action. In fact, adding the hydroxyl scavenger thiourea at concentrations similar to those used by Kohanski et al. (8) to prevent significant hydroxyl radical accumulation did not prevent whiB7 transcriptional activation by several antibiotics (supplemental Fig. S5). Instead, thiourea activated the pMS689GFP reporter (supplemental Fig.  S5). This provided further evidence that whiB7 activation is intimately linked with cell metabolism and activated by a reducing environment.
Mycothiol, the major low molecular weight reducing agent in actinomycetes, provides resistance to free radicals as well as antibiotics (25,26). In mycobacteria, the ratio between its reduced form (MSH) and its oxidized disulfide form (MSSM) serves as an indicator of intracellular redox potential, with a basal reducing ratio of about 200:1 (MSH/MSSM) in M. smegmatis (20). Without an antibiotic inducer, the whiB7 KO contained ϳ10-fold less MSH and ϳ2-fold less MSSM (Fig. 5B) than the wild type, reflecting an ϳ5-fold decrease in MSH/ MSSM (Fig. 5C). After treatment with erythromycin, total levels of mycothiol were unchanged in wild type cells or the whiB7 KO background (Fig. 5B). Importantly, erythromycin treatment virtually eliminated detectable MSSM in wild type cells and had no effect on levels in the whiB7 KO (Fig. 5B). This resulted in an ϳ180-fold MSH/MSSM increase in wild type cells (Fig. 5C).
To determine whether the reduced MSH/MSSM ratio in the whiB7 KO or the increased MSH/MSSM ratio generated by treatment with erythromycin in the wild type reflected broader effects on redox metabolism, the levels of reduced (NADH) and oxidized (NAD ϩ ) nicotinamide adenine dinucleotide were quantified. During exponential growth, the whiB7 KO strains contained slightly lower levels (40% decreased) of NADH compared with wild type (supplemental Fig. S7A). Upon erythromycin treatment, the levels of both NADH and NAD ϩ increased in both strains. The increase was more pronounced in the whiB7 KO; after 2 h of erythromycin treatment, NADH levels increased ϳ3-fold and NAD ϩ levels ϳ2-fold, although the increases in wild type were ϳ1.6and ϳ1.2-fold, respectively (supplemental Fig. S7A). The NADH/NAD ϩ ratio was ϳ1.6fold lower in the whiB7 KO relative to wild type in exponentially growing cultures, and the ratio in both strains increased after erythromycin treatment (ϳ1.3-and ϳ1.7-fold after 2 h in the wild type and whiB7 KO; supplemental Fig. S7B).

DISCUSSION
To better understand factors that affect the intrinsic resistance of mycobacteria to antibiotics, we identified essential components of the whiB7 promoter and the spectrum of induc-ers that activate its transcription using an EGFP reporter system. Our chemical screens and investigations of redox metabolism provided new insights into the biological activation of whiB7 expression. We suggest that inducers generate a reductive shift of redox metabolism in response to inhibition of diverse cellular targets and that this leads to the transcriptional activation of whiB7-dependent intrinsic drug resistance.
Actinomycete genomes all encode WhiB7 orthologs as well as a conserved promoter consensus sequence (Fig. 2D), suggesting that whiB7 has an indispensable role in natural environments and that its transcriptional activation system is conserved in this diverse group of soil and pathogenic bacteria. A comparison of the whiB7 promoter to consensus sequences for all M. tuberculosis factors (27) showed that its Ϫ35 and Ϫ10 hexamer sequences and their spacing were most similar to SigA, the primary vegetative factor. By analogy to WhiB3, WhiB7 may partner with SigA to activate transcription of specialized regulons (28). Because loss of the conserved AT-rich region upstream of the Ϫ35 hexamer restricted activation (Fig.  2B) and WhiB7 contains an AT-hook (10), a motif known to bind the minor groove of AT-rich DNA (29), our results imply that the sequence serves as a WhiB7-binding site. The recognition motif is also found upstream of the genes within the proposed regulon (Fig. 2E). The M. smegmatis genome contains 5954 regions with blocks of at least five adjacent A or T nucleotides (searched with [AT]{5} in DNA Pattern Search available on line). Because many WhiB proteins do not encode an obvious AT hook motif, other conserved regions of WhiB7 may be involved in binding DNA directly or by partnering with other transcriptional activators to target promoters within its regulon. Although antibiotic induction of the whiB7 promoter was strongly reduced in the whiB7 KO background, a slow and constant increase in activity was observed (Fig. 1E). This may reflect the participation of other regulatory elements that either repress under noninducing conditions or co-activate upon induction.
The whiB7 promoter was induced by a wide variety of compounds (Fig. 3) at concentrations much lower than their minimum inhibitory concentrations (Table 2). Many whiB7 transcriptional activators perturb respiration, redox balance, or transmembrane ion flux. One of the most intriguing inducers was the glutamine analog acivicin. Acivicin inhibits glutamine amidotransferases leading to a response described as "metabolic mayhem" in E. coli (30). In our M. smegmatis cultures, acivicin did not inhibit growth at the highest concentration tested (800 M). The fact that it substantially increased whiB7 transcription even at 16-fold lower concentrations (50 M; supplemental Table S2) clearly illustrated the concept that activation was due to a drug-induced metabolic signal. The observation that not all toxic compounds induced whiB7, throughout a range of growth inhibitory and noninhibitory concentrations, demonstrated that the signal was not a common feature associated with cell death. Active compounds apparently induce a unique whiB7-specific metabolic stress signal. In fact, whiB7 inducers were enriched for compounds that inhibit protein biosynthesis, demonstrated by a dose-dependent effect on ribosomes by macrolides (Figs. 4 and 5A). Insights into specificity were also provided by the observation that none of the 50ϩ cell wall targeting compounds in the Sweet library induced whiB7 expression (supplemental Table S2). Protein synthesis and cell wall inhibitors are known to induce distinctive stress signals associated with different stress regulons (31,32).
A previous study had demonstrated that treatment of M. smegmatis with subinhibitory concentrations of macrolides greatly increased resistance to these drugs in an erm (38) (the M. smegmatis ermMT ortholog; MSMEG_1646)-dependent manner (33). Our work showed that macrolide pretreatment increased resistance to the activating macrolide in a whiB7-dependent manner (supplemental Table S3). The whiB7 promoter motif was present upstream of erm (38) (Fig. 2E) implying that the increase in resistance was mediated by erm (38), which requires whiB7 for transcriptional activation. Therefore, a whiB7 response to any activator compound or metabolic signal should similarly increase macrolide resistance. In support of this concept, our data demonstrated that pretreatment with acivicin, a nonmacrolide whiB7 activator, resulted in increased resistance to macrolides (roxithromycin and clarithromycin; supplemental Table S4). The clinical relevance of this induced intrinsic resistance phenomenon has recently been demonstrated in multidrug-resistant strains of M. tuberculosis. Pretreatment with rifampicin induced expression of multiple efflux pumps leading to increased resistance to ofloxacin, a quinolone (34). Rifampicin pretreatment of these strains increased expression of whiB7 and presumably a multidrug transporter that it activates (tap, Rv1258c) contributing to the resistance state (10,34). These observations support a clinically important concept; exposure to whiB7 inducers activates expression of genes and a metabolic state that provides crossresistance to diverse drugs.
Because oxidative stress generated by hydroxyl radicals has been reported to be a common metabolic effect of cidal antibiotics in some bacteria (8), we monitored whiB7 expression under reducing and oxidative culture conditions. Interestingly, DTT (providing additional reducing potential) activated a low level of whiB7 transcription (Fig. 5A). Combining reducing conditions with a low concentration of erythromycin synergistically increased whiB7 transcription, mimicking treatment by a much higher concentration of erythromycin. In contrast, oxidizing conditions (diamide) decreased induction (Fig. 5A). In addition, the hydroxyl radical scavenger thiourea did not prevent antibiotic-induced whiB7 induction (supplemental Fig.  S5). This implied that the whiB7 response to macrolides (and other drugs) was dependent on a metabolic shift to increased reducing potential that was generated independently of hydroxyl radical formation. The reducing potential of mycobacterial cells is reflected in the redox state of reducing agents, including MSH and NADH.
Analyses of MSH, MSSH (Fig. 5B), and their ratio (MSH/ MSSM) (Fig. 5C) revealed differences between wild type and whiB7 KO strain, both untreated and in response to erythromycin. These studies revealed that whiB7 participates in maintaining a reduced cytoplasmic (MSH/MSSM) environment under normal growth conditions and directly or indirectly controls the concentration of mycothiol (MSH ϩ MSSM; Fig. 5B). Furthermore, the fact that erythromycin treatment of the wild type strain resulted in a whiB7-dependent decrease in the pool of oxidized mycothiol suggested that WhiB7 had an active role in regulating MSSM reduction (Fig. 5B). The decrease in MSSM shifted the MSH/MSSM ratio to a highly reduced state (Fig.  5C). By analogy to other WhiB proteins, the activity of WhiB7 as a transcriptional activator may respond to such changes in the redox environment. In support of this model, DTT studies indicated that a highly reduced environment potentiated induction of whiB7 transcription (Fig. 5A). In contrast, erythromycin treatment of the whiB7 KO did not affect levels of oxidized mycothiol (Fig. 5B) or its MSH/MSSM ratio (Fig. 5C). By comparison, changes in NADH and NAD ϩ were relatively small, suggesting that the whiB7-dependent mycothiol redox changes were an initial response to erythromycin, reflected by slower changes in NADH/NAD ϩ pools. Although increased NADH and NAD ϩ pools might be needed to provide longer term responses to erythromycin, they were metabolically isolated from MSH/MSSM pools under the conditions of our experiment.
It has been reported that catabolism of host fatty acids in macrophages results in reductive stress by the accumulation of NADH/NADPH (12,35). The stress is dissipated by a WhiB3dependent shift to the production of virulence lipids (12), but the initial reductive burst may explain the activation of whiB7 in the M. tuberculosis complex soon after infection of macrophages (14) or in response to lipids added to the growth media (10). Additionally, macrophages also utilize glutathione as the major antioxidant for defense against M. tuberculosis infection (36). Exposure to glutathione may provide additional reductive stress to facilitate activation of whiB7. Finally, the observation that the whiB7 KO strain has a more oxidized cytoplasm (Fig.  5C) and lower levels of NADH (supplemental Fig. S6A) rationalizes the reduced output of LuxABCDE bioluminescence (Fig. 1D), which requires reducing power to generate reduced riboflavin mononucleotide, an essential cofactor for the light-generating reaction (37).
Other transcriptional regulators containing redox-sensitive disulfide bridges or iron-sulfur clusters serve to regulate critical redox-responsive functions in bacteria (38). WhiB1, WhiB2, WhiB3, and WhiBTM4 (a bacteriophage-encoded WhiB2 ortholog) proteins also interact with target promoters in a redox-sensitive manner (12,16,17). WhiB7, like all WhiB family proteins, contains four conserved cysteine residues that are able to bind a redox-sensitive iron-sulfur cluster or form disulfide bonds. By analogy to biochemical studies of WhiB1, the iron-sulfur cluster-bound form of WhiB7 may be an inactive form of the protein that does not bind DNA (17). Although our evidence supports the central role of reduced thiols, it does not necessarily rule out the involvement of oxidants. If oxidative damage is a common mechanism of bactericidal antibiotic-mediated cell death (8), the cell may respond by shifting its redox balance back to a more reduced state. This might allow activation of WhiB7 to induce expression of other genes in its regulon and thereby generate a defensive response after exposure to some antibiotics.
Studies of antibiotic resistance in bacterial pathogens have traditionally focused on genes carried by transmissible elements that provide drug resistance to pathogens. However, there is growing recognition that chromosomal genes having physiological functions can have alternative roles as antibiotic resistance genes. This new concept has important evolutionary implications, rationalizing the recruitment and evolution of metabolic genes to serve as antibiotic resistance genes or, alternatively, to provide physiological functions for genes under selective pressure as resistance genes. Here, we focused on whiB7, a transcriptional regulatory gene that determines intrinsic drug resistance in mycobacteria. Our studies revealed that WhiB7 is at a cross-road between physiology and antibiotic resistance with functions linked to the maintenance of balanced reducing potential as well as activation of resistance genes. Activation of whiB7 creates a nonspecific resistance state that provides cross-resistance to diverse drugs. Understanding how WhiB7 senses antibiotic-induced signals to activate intrinsic resistance genes may allow more effective uses of drug combinations (39) for treatment of mycobacterial diseases, including tuberculosis.