Targeting multiple response regulators of Mycobacterium tuberculosis augments the host immune response to infection

The genome of M. tuberculosis (Mtb) encodes eleven paired two component systems (TCSs) consisting of a sensor kinase (SK) and a response regulator (RR). The SKs sense environmental signals triggering RR-dependent gene expression pathways that enable the bacterium to adapt in the host milieu. We demonstrate that a conserved motif present in the C-terminal domain regulates the DNA binding functions of the OmpR family of Mtb RRs. Molecular docking studies against this motif helped to identify two molecules with a thiazolidine scaffold capable of targeting multiple RRs, and modulating their regulons to attenuate bacterial replication in macrophages. The changes in the bacterial transcriptome extended to an altered immune response with increased autophagy and NO production, leading to compromised survival of Mtb in macrophages. Our findings underscore the promise of targeting multiple RRs as a novel yet unexplored approach for development of new anti-mycobacterial agents particularly against drug-resistant Mtb.

in the DNA recognition helix 11,12 . The PhoP regulon includes genes involved in the synthesis of complex cell wall lipids 13 . SenX3-RegX3 is expressed during phosphate starvation and is required for phosphate uptake and aerobic respiration 14 . PrrAB is required early during intracellular infection 15 . MprAB responds to envelope stress and regulates stress-responsive and virulence-associated genes 16,17 . MtrAB is the only essential TCS known so far. It regulates DNA replication and cell division 18,19 . We hypothesized that owing to similarities in domain structure and catalytic features, families of these RRs could be targeted by a single molecule resulting in a downstream effect extending across multiple processes such as persistence, reactivation and tuning of host immune responses. Simultaneous disabling of multiple RRs would in turn, compromise bacterial replication and survival.
Using biochemical assays as well as chemical and computational tools we demonstrate that three selected RRs of Mtb, namely MtrA, RegX3 and MprA, belonging to the OmpR family share a common DNA-binding motif and can be targeted by a single molecule, thereby, leading to downstream effects on their regulons, impairment of the type VII ESX-1 secretion machinery, and attenuation of the ability of Mtb to replicate and survive in macrophages. We bring into context how this molecule influences bacterial fate in macrophages by demonstrating that it augments Mtb-induced autophagy and the release of the effector nitric oxide. Its effects on Mtb translate into changes in the immune response to infection.

Results
Mutating a common motif in MtrA, MprA and RegX3 abrogates their DNA binding activity.
Several RRs from Mtb have structural homology to PhoP, a response regulator of the OmpR family, especially at the C-terminal DNA binding domain 20 . Sequence alignments of eight such RRs from Mtb showed conservation of the motif LRXK at the C-terminal end (Fig. 1A) which was unique to this family of RRs. The positions of these residues in MtrA and RegX3 were: L201, R202, X203 (where X is A for MtrA and S for RegX3) and K204. In MprA, the residues were L202, R203, R204 and K205 (Fig. 1B). Since, two of these were positively charged (R202/203 and K204/205) the probability of direct interaction with the negatively charged DNA backbone was high 21 . To confirm the role of these residues in DNA binding, we chose three representative RRs of the OmpR family, MprA, RegX3 and MtrA and performed EMSAs using purified wild type (WT) or mutant proteins and their respective target promoter DNAs, ppK1 (for RegX3) 22 , mprA (for MprA) 23 and fbpB (for MtrA) 24 . Each WT protein bound to its target protein in a concentration-dependent manner ( Fig. 2A-C). We observed that the mutations L201A in the case of RegX3 and MtrA (L202A for MprA) and K204A in the case of RegX3, MtrA (K205A for MprA) abolished the DNA binding activities of these RRs to their respective target DNA (Fig. 2D-H). However, the mutation R202A in the case of RegX3, MtrA (R203A for MprA) did not abolish DNA binding activity of these RRs (Fig. 2I-K). We further analyzed these interactions by surface plasmon resonance (SPR). Biotin-labelled DNA fragments were immobilized on streptavidin-coated CM5 surfaces and the WT or mutant proteins were allowed to flow over these surfaces. As expected, the WT or R202A (R203A for MprA) mutant proteins showed higher responses than the K204A or L201A (K205A or L202A for MprA) mutants ( Supplementary  Fig. 1). These results strengthened our contention that the LRXK motif ( Fig. 2L) is essential to the DNA binding ability of this set of RRs from the OmpR family, in vitro. This motif is unique to the OmpR family and absent in the NarL family of RRs such as DosR.
DNA binding mutants of RegX3 or MprA of Mtb are compromised in gene regulatory function.
In order to investigate whether the role of the above amino acids extend to the ability of the RRs to regulate gene expression in vivo, we generated knockout strains of Mtb H37Rv in which the chromosomal copy of either regX3 or mprA had been replaced by the hygromycin resistance gene ( Supplementary Fig. 2a,b) using temperature sensitive mycobacteriophages 25 . Since mtrA is essential, it was not possible to inactivate this gene. The replacement of regX3 or mprA by the hygromycin resistance gene in the respective mutant strains was confirmed by PCR ( Supplementary Fig. 2c,d) and immunoblotting ( Supplementary Fig. 2e,f). We complemented the regX3 deletion strain (Δ regX3) with a copy of wild type regX3 integrated into the chromosome under the control of its native promoter (Δ regX3::regX3). Compared to the wild type, Δ regX3 showed upregulation of gltA1 (or prpC) a RegX3-repressible gene 26 , under nutrition-sufficient conditions (Fig. 3A). This could be reversed by complementation with WT regX3 but not with regX3 K204A. RegX3 regulates a distinct set of genes under phosphate starvation including pstS3 14 . The Δ regX3 strain was compromised in its ability to induce the expression of pstS3 under phosphate starvation. regX3 WT but not regX3 K204A complemented Δ regX3 restoring pstS3 expression (Fig. 3B). Taken together, these results suggested that the RegX3 K204A mutant is functionally compromised in terms of regulating RegX3-dependent gene expression in Mtb grown in vitro.
To further strengthen our argument that K204 (K205 for MprA) was likely to be a functionally important residue across a set of RRs, we extended our studies to analyzing gene expression under SDS stress attributable to MprA. The transcript levels of two sigma factors, sigE and sigB, both MprA targets 16,17 , were significantly reduced in the mprA mutant strain (Δ mprA). These levels could be restored in the mutant strain upon complementation with wild type mprA but not with mprA K205A (Fig. 3C,D). These observations reinforced the view that amino acid residues conserved in the DNA recognition helix of a set of RRs likely play an important functional role across this family of RRs.
A DNA binding mutant of Mtb RegX3 is compromised for survival in macrophages. Based on the report that survival of the Δ regX3 strain is compromised in macrophages 10 , we infected the murine macrophage like cell line RAW264.7 with each of the three strains containing variants of regX3 described above. In concordance with previously published observations, we observed that in comparison to the parental strain, the survival of the Δ regX3 strain was compromised by ~2 fold in RAW264.7. The observed growth defect was restored by complementation of the mutant with WT regX3 but not K204A regX3 (Fig. 3E). This suggested that the LRXK   motif plays a role in bacterial survival in macrophages. Mutation of residue R202 to alanine did not compromise DNA binding ability of RegX3 (Fig. 2I). In harmony with this, the ΔregX3::regX3R202A strain showed growth characteristics similar to the wild type ( Supplementary Fig. S2g).
Targeting the LRXK motif by molecular docking. Taking into account that the conserved LRXK motif was critical for DNA binding and function of each RRs, we reasoned that an inhibitor docking to this region could potentially target multiple RRs of this family. We performed molecular docking studies to identify inhibitors that targeted this motif. We hypothesized that such inhibitors would compromise survival of Mtb both in vitro as well as in vivo.
The amino acids L201, R202 and K204 are present in the DNA binding domain of MtrA at the α 7 helix loop end which is very near the protein's inter-domain interface. In the activated state the α 7 helix is required for interaction with DNA and the α 7-α 8 transactivation loop participates in an interaction with RNA polymerase 27 . Sitemapping of the MtrA crystal structure (PDB Id -2GWR) identified a pocket close to the amino acids L201, R202 and K204. The selected sitemap was used for the generation of Grid1 (with the coordinates x = 37.4489, y = 3.0456 and z = 8.6644). The centroid of L201, R202 and K204 was selected as Grid2 (with coordinates x = 27.5382, y = − 0.136 and z = 10.2869). These two grids ( Supplementary Fig. 3a) were used to screen a BITS Pilani database for MtrA inhibitors. Similarly for RegX3, site mapping and grid generation was performed with its crystal structure (PDB Id -2OQR) 28 . We identified a pocket close to the amino acids L201, R202 and K204 ( Supplementary Fig. 3b). For MprA, the lack of a crystal structure required us to build a homology model using Prime (details of the method are provided in the supplemental section). This homology model was then used for site mapping and grid generation ( Supplementary Fig. 3c).
2500 compounds were initially docked to each grid in high throughput virtual screening (HTVS) mode and scored with glide scoring function. The top 20% scoring (∼ 500) molecules were re-docked using SP (shape and physiochemical properties) docking and the top 10% (∼ 100) were re-docked using XP (extra precision) docking. The hits obtained by XP had better G score values (− 3.5 to − 2.26 kcal/mol). The docked hits from the SP showed a sharp decrease in their G score values from − 4.09 kcal/mol to almost − 5 kcal/mol; after − 2 kcal/mol the increase in G score value was significantly smaller. Therefore, the cut off G score value was chosen as − 2 kcal/mol for docking hits of the XP run. Eight compounds (2-9) (Fig. 4A) showed G scores lower than the cut off (Supplementary Table 1). The compounds that showed up as hits against MtrA were also docked against the grids in RegX3 and MprA to check their docking energies. All the eight compounds docked at higher energies with RegX3 and MprA. The G scores were in the range of − 1.7 kcal/mol to − 2.9 kcal/mol (Supplementary Table 1). The eight inhibitors identified as high scoring hits that docked against the LRXK motifs in MtrA, RegX3 and MprA (Fig. 4A) were tested for their ability to inhibit DNA binding activity. Compound 2 [ 2IT4O or 2-iminothiazolidine-4-one] and Compound 6 [OTABA or oxo-1,3-thiazolidin-2-ylidene amino benzoic acid], (C2 and C6 respectively) could inhibit DNA binding activity of MtrA (Fig. 5A,B), whereas the other molecules were not effective ( Supplementary  Fig. 4a). C2 and C6 also inhibited the binding ability of RegX3 (Fig. 5C,D) and MprA (Fig. 5E,F). C2 was more effective than C6 (as apparent from the IC 50 values shown in Fig. 5). The docking of C2 and C6 to the DNA binding pockets of MtrA, RegX3 and MprA (including the LRXK motif), is shown in Fig. 4(B-G). The inhibitory activity of C2 was specific for the OmpR family of RRs. C2 did not inhibit binding of DosR, a member of the NarL family to an hspX promoter-derived DNA fragment ( Supplementary Fig. 4b).

C2 and C6 inhibit the growth of Mtb in vitro.
Having established the ability of C2 and C6 to inhibit DNA binding of MtrA, RegX3 and MprA, we next determined the ability of these compounds to inhibit Mtb growth in vitro using Alamar Blue assays. We observed that the IC 50 values of C2 and C6 against Mtb were 9 μ M (~6 μ g/ml) and 43 μ M (~20 μ g/ml), respectively (Fig. 6A,H). The efficacy of C2 was higher compared to that of C6. The lower efficacy of C6 was in harmony with its lesser ability to inhibit DNA binding (Fig. 5). The growth curve of Mtb in the presence of C2 levelled off earlier than growth in the absence of C2 (Fig. 6B). Considering that C2 restricted growth of Mtb, we assessed bacterial replication in its presence exploiting the unstable plasmid pBP10 which is lost at a quantifiable rate from dividing cells in the absence of antibiotic selection 29 . Interestingly, exposure to C2 resulted in higher levels of retention of pBP10 by Mtb after four days of growth in the absence of kanamycin (Fig. 6C), suggesting that C2 inhibits replication of Mtb in vitro, which agrees with the growth inhibition observed in the presence of C2.
C2 regulates expression of mtrA, regX3 and mprA and their targets. To confirm that C2 targets mtrA, regX3 and mprA in Mtb cultures, we analyzed the expression of these autoregulated 23,24,30 RRs in Mtb grown in the presence of C2. The expression of all three RRs was inhibited after exposure for 48 h to C2 (Fig. 6D). C2 expectedly regulated the expression of several targets of these three RRs. It augmented the expression of rpfB an MtrA-repressible target 31 , as well as gltA1 and cydA, RegX3-repressible targets ( Supplementary Fig. S4c,d). It repressed the expression of the MprA-activated targets sigB, sigE and Rv0081, the MtrA-activated targets ftsI and dnaN 24 and the RegX3-activated target pstS3 ( Supplementary Fig. S4c,d) confirming that C2 inhibits the gene regulatory functions of MtrA, RegX3 and MprA.
C2 inhibits the ESX-1 secretion apparatus. Mtb utilizes the type VII ESX-1 secretion apparatus to translocate substrates such as ESAT-6 32 . Secretion through this virulence-associated apparatus requires the unlinked espACD operon 33 . Transcription of espACD is regulated by EspR and EspR depletion leads to reduced bacterial survival 34 . Two OmpR family RRs, PhoP and MprA regulate espR 35 . In view of this, we tested the possibility that targeting of RRs by C2 impacts the expression of espR and the espACD operon. We observed that C2 reduced the expression of both espA and espR in Mtb after 48 hours of treatment ( Supplementary Fig. S4e). In harmony with this, we observed reduced amounts of ESAT-6 in the culture filtrate of Mtb Erdman treated with C2 (Fig. 6E).   C2 inhibits growth of Mtb in macrophages and augments the release of nitric oxide. As the next logical step, we tested the efficacy of C2 and C6 in infected macrophages. RAW 264.7 cells were infected with Mtb in the absence or presence of C2 or C6. Bacterial survival was assayed after 4 days. The IC 50 values of C2 and C6 against Mtb grown in RAW264.7, were 3 μ M (~0.85 μ g/ml) and 37 μ M (~12.5 μ g/ml), respectively (Fig. 6F,H). C2 was more effective than C6 in infected macrophages. C2 did not affect the viability of RAW264.7 (Fig. 6G).
In order to quantify the survival of Mtb in primary macrophages in the presence of C2, murine bone marrow derived macrophages (BMDM) or human monocyte derived macrophages (hMDM) were infected with Mtb at an MOI of 5. In BMDMs and MDMs we observed a ~8 fold decrease in CFU after 6 days in the presence of C2 (Fig. 7A,B). We argued that the effects of a compound on Mtb in infected macrophages depend not only on its ability to regulate bacterial gene expression programs, but also on associated changes in the immune response during infection. Considering that the survival of Mtb in macrophages is diminished in the presence of C2, we tested for the production of the antibacterial effector NO. NO release from infected macrophages was significantly higher in the presence of C2 than in its absence (Fig. 7C,D; Supplementary Fig. S5a). C2 alone did not affect NO release from uninfected macrophages (data not shown).

C2 augments autophagy.
We further reasoned that apart from augmented NO release, enhanced lysosomal trafficking of Mtb would also inhibit the survival of Mtb in the presence of C2. Autophagy enhances lysosomal trafficking of Mtb 36 . We therefore tested whether C2 impacts autophagy in Mtb-infected macrophages. The conversion of LC3-I to LC3-II is an established marker of autophagy. This was enhanced in Mtb infected RAW 264.7 cells in the presence of C2 (Supplementary Fig. S5b).
Autophagy is also characterized by formation of LC3 puncta. We confirmed the effect of C2 by enumerating the formation of LC3 puncta (by immunostaining) in infected macrophages. Puncta formation was enhanced in the presence of C2 (Fig. 7E-H; Supplementary Fig. S5c,d) as in the case of isoniazid ( Supplementary Fig. S5e,f). Similar results were obtained in the presence of bafilomycin A an inhibitor of autophagic flux (data not shown). C2 therefore augments the immune response by enhancing autophagy during Mtb infection.

Discussion
Tuberculosis is usually treatable with a combination of four first-line antitubercular drugs. However, poor compliance has led to the emergence of multidrug resistant (MDR) and extensively drug resistant (XDR) organisms. The need to develop alternate therapeutic strategies is therefore obvious. The TCSs allow bacteria to sense and respond to changes in the environment, including host-mediated antimicrobial activities. As sensors of environmental signals, TCSs regulate diverse physiological processes such as sporulation, the equilibrium between a dormant and an actively growing state, utilization of nutrient sources such as carbon, nitrogen and phosphate, antibiotic resistance and competence 7 . They are functionally regulated by accessory proteins positioning them as hubs in networks of cellular information flow 37 . Coupled with the fact that they are absent in eukaryotes, the TCSs represent attractive drug targets. Mtb encodes several members of the OmpR family of RRs which share a common winged helix-turn-helix motif in their output domain. Here we focused on three representative RRs, MtrA, RegX3 and MprA. MtrA is the only known essential RR of Mtb. It regulates genes associated with DNA replication, cell division and cell wall biosynthesis 24 . SenX3-RegX3 responds to phosphate starvation, and a regX3 transposon mutant is attenuated in the lungs of mice and guinea pigs following low dose aerosol infection 14 .
RegX3 regulates genes that are involved in energy metabolism, cell envelope maintenance and regulatory functions 10 . mprA-mprB is upregulated in the lungs of mice during infection 38 , and also in an artificial hollow-fiber granuloma model 39 . This TCS modulates ESX-1 function 40 which is a prototype of type VII secretion systems of Gram-positive bacteria.
We argued that based on the conservation of amino acid sequence in the DNA recognition helix of the OmpR family of RRs of Mtb, we could conceivably target all three of the above RRs simultaneously. Furthermore it is documented that inactivation of TCSs cause attenuation as in the case of Mtb ΔphoPR 41,42 and Mtb ΔprrAB 15 . Multiple sequence alignments of the DNA binding domains of the OmpR family of RRs in Mtb showed the presence of a conserved LRXK motif in their recognition helix. We tested our hypothesis that this motif is required for the DNA binding activities of MtrA, RegX3 and MprA by EMSAs and SPR-based protein-DNA interaction analyses. For each of the three RRs, mutating the conserved amino acid residues L201/202 or K204/205, compromised DNA binding ability, supporting our hypothesis (Fig. 2). In order to establish the role of the aforesaid amino acids in vivo, molecular genetic manipulations were carried out to inactivate the chromosomal copies of two (RegX3 and MprA) of these RRs. As expected, we observed that Δ regX3 and Δ mprA strains were highly compromised in terms of their ability to regulate genes known to be part of the RegX3 and MprA regulons respectively. In order to ascertain whether mutants carrying substitutions of the critical amino acid residues in the DNA recognition region could restore the functions of these RRs in vivo, we complemented each knockout strain with either the wild type or the respective DNA binding mutants of regX3 or mprA. As expected, the mutated copies of the respective proteins could not appropriately regulate the expression of downstream genes (Fig. 3). We sought to target this domain of high conservation such that a single molecule might simultaneously inhibit multiple Mtb TCSs.
Target based approaches have been associated with high attrition rates 43 . Here we have taken as a starting point, lead structures obtained from whole cell screens to search a library of compounds developed at BITS Pilani with known anti-mycobacterial activity. The crystal structure of one of the representative RRs was used as a template in energy-based pharmacophore modelling and in silico docking to the DNA binding pocket. Eight molecules were selected for further analysis, of which two, C2 and C6 both based on a thiazolidine scaffold, could effectively inhibit the three RRs chosen by us. Here we show that the 2-imino thiazolidine-4-one derivative, C2 has higher efficacy in terms of inhibition of DNA binding activity of MtrA, MprA and RegX3 (as evaluated by EMSA), and lower MIC compared to C6 (Fig. 6). We therefore took up C2 for further detailed studies. For Mtb grown in vitro, C2 could inhibit the transcription of all three of the aforesaid RRs, and modulate downstream genes regulated by these RRs. It is logical to assume that C2 will also target other OmpR family RRs PrrA and PhoP and their regulons, since both these RRs also contain the LRXK motif.
The type VII ESX-1 secretion system is encoded by the esx-1 genetic locus, and is essential for the export of two major virulence factors of Mtb, namely ESAT-6 (EsxA) and CFP-10 (EsxB) 32 . The secretion of ESX-1 substrates also requires the products of the espACD operon. EspR is a transcriptional regulator that activates the transcription of espA, espC and espD 34 . Recent studies have linked MprAB and PhoPR to the expression of EspR 35 and the regulation of the esx-1 region 40 . Based on these reports, we hypothesized that C2 could potentially regulate espR transcription. The downregulation of espR in the presence of C2 confirmed our hypothesis. In harmony with this, we observed that C2 inhibits the expression of espA. Pang et al. 40 have reported that MprAB regulates the ESX-1 region, whereas Rybnicker et al. 44 have reported that a compound which deregulates MprAB, affects the ESX-1 region and the ESX-1 dependent secretion of ESAT-6. Taking together the fact that C2 inhibits MprA and that secretion of ESAT-6 depends on products of the esx-1 and the espACD operons, we tested the secretion of ESAT-6. For this purpose, we took Mtb Erdman as the strain of choice, since Mtb H37Rv shows limited secretion of ESAT-6 45 . We observed reduced secretion of ESAT-6 in the culture filtrate of C2-treated Mtb Erdman cells (Fig. 6E). In addition, C2 treatment lowered esxA expression and lower ESAT-6 was observed in the lysates of C2-treated cells. The mechanism of reduced expression of esxA, remains unclear at present. Reduction in esxA expression and a defective secretion machinery could together contribute to the diminished amounts of ESAT-6 in the culture filtrate of C2-treated cells. ESAT-6 is a central molecule in host-pathogen interactions. For example, it promotes pathogen survival and dissemination by arresting phagosome maturation and facilitating necrosis. Defective production and secretion of ESAT-6 is therefore likely to have wide-ranging implications in the context of infection.
The environment of Mtb within its host is associated with oxygen depletion, nutrient stress, oxidative and nitrosative stress. Mtb must therefore be capable of efficiently sensing the environment and reprogramming its transcriptome in order to maximise its chance of enduring within this hostile environment by dampening the host immune response. C2 was capable of regulating gene expression attributable to each of the three chosen RRs. We reasoned that an impaired bacterial signal transduction machinery would most likely influence the response of macrophages to infection. For example, it has been reported that an mprAB knockout strain of Mtb elicits diminished levels of TNF-α and IL-1β from macrophages, compared to the wild type Mtb 40 . We tested whether C2 alters the host response to Mtb infection. The innate immune response requires the participation of macrophages, dendritic cells, epithelial cells, neutrophils and other immune cells culminating in events that augment cellular antimicrobial mechanisms, reduce inflammation and go on to shape the adaptive immune response. In this study we restricted ourselves to analysing the effects of C2 on infected macrophages. Autophagy has been associated with degradation of long-lived proteins to provide energy under nutritional stress and in removing misfolded or aggregated proteins or damaged organelles. In addition it plays a pivotal role in defense against intracellular pathogens 46,47 . Activation of the ubiquitin-mediated autophagy pathway results in reduced survival of Mtb in macrophages. The first-line antimycobacterial drug isoniazid activates autophagy 48 . With this background information, we tested the effect of C2 on Mtb-mediated autophagy in murine BMDM, hMDM and RAW264.7. Here we present evidence that C2 triggers enhanced autophagy during Mtb infection. Using a Western blot assay for LC3-II formation and fluorescence microscopy for enumeration of LC3 puncta, we observed enhanced autophagy in the presence of C2 early during infection. This effect was apparent even in the presence of bafilomycin A, confirming that the observed increase in LC3-II in the presence of C2 was not due to a block in autophagic flux. In addition, C2 also augmented release of NO in Mtb-infected macrophages (Fig. 7). C2 regulated bacterial pathways conceivably alter bacterial processes in a manner that influences the host immune response. While it is possible that C2 could exert effects on mycobacterial targets other than the RRs, it is tempting to speculate that simultaneous incapacitation of multiple RRs could be directly associated with enhanced autophagy. Taken together our results demonstrate that (a) DNA binding activity of multiple TCSs depends on a conserved motif which can be targeted to modulate stress responsive genes associated with the respective RRs; (b) a compound that docks to this DNA binding pocket can effectively inhibit multiple RRs simultaneously; and (c) targeting multiple RRs by a small molecule can twist the host-pathogen interaction to alter autophagy and NO production with a concomitant loss of bacterial burden in macrophages. The results described here serve as a proof of principle that TCSs can be simultaneously targeted to strike down pathways important for bacterial survival and tilt the balance of the host-pathogen interaction in favor of the host. It opens up new avenues of exploration in our quest for a new drug against TB. In the case of Mtb, resistance to drugs usually occurs through chromosomal mutations rather than through the acquisition of extrachromosomal resistance determinants or through mobile genetic elements. Resistance to a single drug requires mutation in a single gene. Combination chemotherapy makes the emergence of a spontaneous mutant resistant to all the components of the regimen, less likely. By this same token, we suggest that resistance to a drug capable of targeting multiple RRs would be far less likely, since it would necessitate mutations of multiple genes. This approach therefore appears to be an attractive option for development of new drugs.

Methods
Bacterial strains, media and growth conditions. Mycobacterial strains (H37Rv and Erdman) were routinely maintained in Middlebrook 7H9 medium supplemented with 10% ADC or OADC. Unless mentioned specifically, Mtb refers to the H37Rv strain. Detailed bacterial growth conditions have been provided in the supplemental section. Cell culture. RAW 264.7 cells were routinely maintained in DMEM with 10% FBS, 4 mM glutamine and penicillin/streptomycin. Human MDMs were derived from peripheral blood monocytes. BMDMs were obtained from C57BL/6 mice. Detailed cell culture methods have been provided in the supplemental section.

Construction of Mtb mutants and plasmids.
Mtb knock out mutants of mprA and regX3 were generated using temperature sensitive mycobacteriophages as described by Bardarov et al. 25 Detailed methods for the generation of knock out mutants and complementation of these mutants with either wild type or mutant mprA or regX3 are available in the supplemental section. A list of plasmids and bacterial strains is available on request.
Expression and purification of recombinant proteins. RegX3, MprA, MtrA and SenX3 (lacking the 78 bp transmembrane domain) were amplified by PCR, cloned and over expressed in E. coli. DosR (DevR) expression construct was obtained from Prof. Jaya Tyagi, All India Institute of Medical Sciences, New Delhi. Mutant proteins were generated by site directed mutagenesis using overlap extension PCR. Detailed methods for purification of recombinant proteins are given in the supplemental section. A list of primers is provided with the supplemental "Methods".
Quantitative real time PCR. Total RNA was extracted from bacterial cultures, reverse transcribed and Real time PCR was performed on an Applied Biosystems 7500 machine using MesaGreen master mix (Eurogentec) following the manufacturer's instruction. A list of primers is provided with the supplemental "Methods".
Quantification of bacterial growth and replication. Mtb growth was monitored by following change in absorbance at 600 nm. Alternatively, Alamar blue assay was used. Detailed method has been provided in the supplemental section.
Surface Plasmon Resonance analysis of protein-DNA binding. Double stranded DNA was biotin labelled and immobilised on a streptavidin coated sensor chip. Wild type or mutant proteins were injected at different concentrations and binding response was measured in a BIAcore X-100 instrument (GE Healthcare). Detailed method has been provided in supplemental section.
In vitro phosphorylation and electrophoretic mobility shift (EMSA) assays. Autophosphorylation of MprB or SenX3 was carried out in the presence of 16 mM ATP followed by incubation with purified MprA or RegX3 for transphosphorylation. MtrA was phosphorylated using EnvZ as described by Sharma et al. 31 . DosR was phosphorylated using acetyl phosphate. For EMSAs, phosphorylated MtrA or RegX3 or MprA or DosR was incubated with 5′ -Cy5-or 32 P-labelled DNA and bound complex was separated on 5-6% non denaturing PAGE followed by detection using a Typhoon Trio Plus Imager (GE Healthcare). Detailed methods have been provided in the supplemental section.
Docking studies. Crystal structures of RegX3 (PDB Id 2OQR) and MtrA (PDB Id 2GWR) were obtained from PDB and energy minimised using the OPLS 2005 forcefield in Schrodinger suite version 9.3. A homology model of MprA was developed using Prime and the crystal structures of PrrA and PhoP (PDB Ids 3ROJ, 1YS7, 1YS6, 1KGS, 3F6P, 2ZWM and 1NXO) as templates. The model was energy minimised using the OPLS 2005 forcefield in Schrodinger suite version 9.3. Active site prediction was done using the Sitemap module while grid generation was done using the Glide module of Schrodinger LLC version 3.5. 2500 compounds in the BITS database were then docked against the generated sitemaps and grids by high throughput virtual screening (HTVS). Shortlisted compounds were used for further studies. Detailed methods have been provided in the supplemental section.

Survival of Mtb in macrophages.
Macrophages were grown and infected with Mtb for 4 h. Cells were lysed at the indicated times and viable bacteria were enumerated.
Culture filtrate preparation and Western blotting. Mtb Erdman culture filtrates were obtained after growing bacteria in Sauton's media. Proteins were immunoblotted with ESAT-6 or GroEL antibody (obtained through BEI Resources, NIH, NIAID). Detailed methods have been provided in the supplemental section.
Autophagy assays. For autophagy analysis, infected macrophages were lysed and immunoblotted with LC3 antibody. For the visualisation of LC3 puncta, infected macrophages were incubated with LC3 antibody followed by immunostaining with Alexa 488-conjugated secondary antibody. Images were acquired in a Leica confocal microscope. At least 100 cells were counted to enumerate the number of punctated cells. Detailed methods have been provided in the supplemental section.
Nitrite was measured using Griess reagent (Invitrogen), according to the manufacturer's instructions. Absorbance was read at 540 nm in an ELISA reader and the concentrations of nitrite were calculated against a standard curve.
Ethics statement. Animal experiments were approved by the institutional Animal Ethics Committee of the Bose Institute, Kolkata, India and were carried out in accordance with the approved guidelines.
Statistical analysis. Statistical analyses were performed using GraphPad Prism 5 software. Data have been expressed as means ± SD. Two-tailed Student t test was used when comparing two groups. *p < 0.05; **p < 0.01; ***p < 0.001.