Efficacy of the β-lactam\β-lactamase inhibitor combination is linked to WhiB4 mediated changes in redox physiology of Mycobacterium tuberculosis

Aims Inhibition of β-lactamase by clavulanate (Clav) sensitizes multi-and extensively drug-resistant Mycobacterium tuberculosis (Mtb) strains towards β-lactams such as amoxicillin (Amox). However, the underlying mechanism of how Mtb responds to Amox-Clav combination (Augmentin; AG) is not characterized. Results We integrated global expression profiling with the protein-protein interaction landscape and generated a genome-scale network of Mtb in response to AG. In addition to specific targets (e.g., peptidoglycan biosynthesis and β-lactamase), the response to AG was also centered on redox-balance, central carbon metabolism (CCM), and respiration in Mtb. We discovered that AG modulates superoxide levels, NADH/NAD+ balance and mycothiol redox potential (EMSH) of Mtb. Higher intra-mycobacterial EMSH potentiates mycobactericidal efficacy of AG, whereas lower EMSH induces tolerance. Further, Mtb responds to AG via a redox-sensitive transcription factor, WhiB4. MtbΔwhiB4 displayed higher expression of genes involved in β-lactam resistance along with those mediating respiration, CCM and redox balance. Moreimportantly, WhiB4 binds to the promoter regions and represses transcription of genes involved in β-lactamase expression in a redox-dependent manner. Lastly, while MtbΔwhiB4 maintained internal EMSH, exhibited greater β-lactamase activity and displayed AG-tolerance, overexpression of WhiB4 induced oxidative shift in EMSH and repressed β-lactamase activity to aggravate AG-mediated killing of drug-sensitive and –resistant strains of Mtb. Innovation and Conclusions This work demonstrate that efficacy of β-lactam\β-lactamase inhibitor combination can be attenuated by elevating mycobacterial antioxidant capabilities and potentiated by impairing redox buffering capacity of Mtb. The functional linkage between β-lactams, redox balance, and WhiB4 can be exploited to potentiate AG action against drug-resistant Mtb.

lactams, redox balance, and WhiB4 can be exploited to potentiate AG action against drugresistant Mtb.

Introduction
Mycobacterium tuberculosis (Mtb) displays tolerance to several clinically important antibacterials such as aminoglycosides and β-lactams (14,36). Innate resistance of Mtb towards β-lactams is largely considered due to the presence of a broad-spectrum Ambler class A βlactamase (BlaC) (15). Other mechanisms such as cell envelope permeability, induction of drug efflux pumps, and variations in peptidoglycan biosynthetic enzymes also play a role in β-lactamresistance in Mtb (18,30). The Ambler class A β -lactamases are mostly susceptible to inhibition by clavulanate (Clav), sulbactam (Sub), and tazobactam (Taz) (30). Indeed, intrinsic resistance of Mtb towards β-lactams can be overcome by using the combination of β-lactams with Clav (5,22).

Combination of amoxicillin (Amox) with Clav (Augmentin; AG) was active against Mtb in vitro
and had significant early bactericidal activity in patients with drug-resistant TB (5,8). Likewise, combination of meropenem with Clav showed significant bactericidal activity against drugresistant strains of Mtb (22). With this attention, there is an urgent need to generate knowledge about the mechanisms of action of β-lactams in combination with Clav against Mtb as well as possible route(s) of resistance.
In other bacteria, β-lactams directly interact with enzymes involved in peptidoglycan (PG) synthesis, which may lead to killing via multiple mechanisms including induction of autolysin pathway, holin:antiholin pathways, DNA damage, and alterations in physiology (e.g., TCA cycle, and oxidative stress) (26,28,34,42,54). The complex effects of β-lactams on both PG biosynthesis and other processes indicate that the response to β-lactams could be mediated either through direct sensing of β-lactam molecules or by their effects on bacterial physiology. In Staphylococcus aureus (S. aureus), a transmembrane protease, BlaR1, senses β-lactam concentrations by direct binding through extracellular domain, which activates its intracytoplasmic proteolytic domain resulting in cleavage of the β-lactamase repressor, BlaI, and induction of β-lactamase expression (17). The knowledge of how β-lactams are sensed to activate adaptation programs is lacking in Mtb. It has been shown that Mtb expresses a close homolog of BlaR1 encoded by Rv1845c (blaR), which modulates the activity of BlaC by regulating BlaI repressor in a manner analogous to S. aureus BlaR1-BlaI couple (45). However, BlaR1 orthologues in all mycobacterial species lack extracellular sensor domain involved in binding with β-lactams (45), indicating that mechanisms of antibiotic sensing and BlaC regulation are likely to be distinct in Mtb. Similarly, how β-lactams influence mycobacterial physiology such as redox balance and metabolism remain unknown. Therefore, novel insights are needed to discover how the presence of β-lactams is conveyed in Mtb to activate appropriate adaptation response.
Here, we generated a system-scale understanding of how AG affects mycobacterial physiology. Exploiting a range of technologies, we explained mechanistically that AG efficacy is partly dependent upon the redox physiology of Mtb. Further, we described rationally the role of a redox-responsive transcription factor, WhiB4, in regulating tolerance to AG by recalibrating drug transcriptome, β-lactamase activity, and intramycobacterial mycothiol redox potential (E MSH ) during infection. Our study demonstrates how Mtb alters its redox physiology in response to AG and identifies MSH and WhiB4 as major contributors to β-lactam tolerance.

Network analysis revealed modulation of cell wall processes in response to AG in Mtb.
To assess the response of Mtb towards β-lactam and β-lactamase inhibitor combination(s), we generated transcriptome of mycobacterial cells upon exposure to AG. We observed that 100 µg/ml of Amox in combination with 8 µg/ml of Clav (10X MIC of AG) arrested bacterial growth at 6 h and displayed killing only after 12 h post-exposure (Supplementary Fig. 1-Inset). Therefore, expression changes at a pre-lethal phase (i.e. 6 h post-10X MIC AG treatment) can reveal significant insights into pathways Mtb exploits to tolerate AG.
A total of 481 genes were induced ( 2-fold; P value  0.05) and 461 were repressed ( 2-fold; P value  0.05) in the wt Mtb upon AG-treatment (Supplementary Table S1). While important, transcriptome alone represent only a narrow glimpse of the mechanisms exploited by a cell in order to tolerate drug exposure. Therefore, it is important to harness the power of computational approaches that combine condition-specific expression data with general protein interaction data to construct dynamic and stress response networks. We first created a comprehensive proteinprotein interactome (PPI) of Mtb using information from experimentally validated and published interactions (see Materials and Methods) and then integrated microarray data with the PPI to generate AG response network. In the network, a node represents a gene and node weight is the product of normalized intensity values with the corresponding fold change in expression upon drug exposure. Expression profiles of nodes were used to determine the edge-weight between two interacting nodes. Although a full description of the mathematical equations and algorithms used to generate AG response network is beyond the scope of this study, we encourage readers to refer our original papers for detailed methodology (39,46,47).
Supplementary Fig. 1 and Table S2 represent the 1 % top nodes, which covers a total of 806 genes which are connected through 1096 interactions to form a well-connected AG response network of Mtb. Genes belonging to diverse functional classes such as intermediary metabolism, cell wall, lipid metabolism, virulence, and information pathways featured in the response network. Since β-lactams primarily target PG-biosynthetic processes, our expression data clearly showed induction of genes associated with the biogenesis of PG and other cell processes (Fig.   1A). Network analysis confirmed that the cumulative node weight intensity (CNW) of genes belonging to cell wall related processes was highest (30616361.02) among the classes affected by AG treatment (Fig. 1B). Induction of genes involved in PG biosynthesis and β-lactamase regulation (blaR-blaI) (Fig. 1A) in response to AG validates our experimental conditions. In addition, two other mechanisms involved in tolerating β-lactams i.e. outer membrane permeability (mycolic acid biogenesis [kasA, kasB, and fabD] and omp) and efflux pumps (efpA, Rv1819c and uppP) were induced in response to AG (Supplementary Table S1). Importantly, in line with cell surface targeting activity of β-lactams, our analysis identified activation of global regulators (e.g., sigE, sigB, and mprAB) of cell envelope stress in Mtb. These regulators function as major hub nodes and form-interconnected networks of genes important to maintain cell wall integrity in response to AG (Fig. 2). Altogether, Mtb responds to AG by modulating the expression of cell envelope associated pathways including those that are the specific targets of βlactams.

AG affects pathways associated with CCM, respiration, and redox balance
Our analysis revealed enrichment of nodes belonging to "intermediary metabolism and respiration" within AG response network (CNW= 20716788.92; Fig. 1B). For example, energetically efficient respiratory complexes such as NADH dehydrogenase I (nuo operon) and ATP-synthase (atpC, atpG, atpH) were down-regulated, whereas energetically less favored NADH dehydrogenase type II (ndh), cytochrome bd oxidase (cydAB), and nitrite reductase (nirBD) were activated in response to AG (Fig. 3A). AG also substantially induced genes involved in assembly and maturation of cytochrome bd oxidase in E. coli (i.e. cydDC; ABC transporter) (7) (Fig. 3A). The transcriptional shift towards a lesser energy state is consistent with the down-regulation of several genes associated with TCA cycle (sucCD, fum, mdh, and citA), along with an induction of glycolytic (pfkA, pfkb, fba, and pgi), gluconeogenesis (pckA) and glyoxylate (icl1) pathways (Fig. 3A). Interestingly, icl1 has recently been shown to promote tolerance of Mtb towards diverse anti-TB drugs by maintaining redox homeostasis (37). These findings indicate that rather than energy generation, maintenance of redox balance is more likely to be an important cellular strategy against AG.
The influence of AG on mycobacterial redox physiology is also apparent by a significant enrichment of genes that can report exposure to oxidative stress in the response network. We found increased expression of reactive oxygen species (ROS) detoxifying enzymes (ahpCD, katG, and hpx), antioxidant buffers (trxB1, trxA, trxC, mtr), methionine sulfoxide reductase (msrA), Fe-S cluster repair system (Rv1461-Rv1466; suf), and intracellular redox sensors (whiB6, whiB2, whiB3, and pknG) (Fig. 3A). The global regulator of oxidative stress in bacteria, OxyR, is non-functional in Mtb (9). However, we earlier reported that a redox-sensitive DNA binding protein, WhiB4, functions as a negative regulator of OxyR-specific antioxidant genes (e.g., ahpCD) in Mtb (6). Consequently, Mtb lacking whiB4 (MtbΔwhiB4) displayed higher expression of antioxidants and greater resistance towards oxidative stress (6). While microarrays in response to AG showed only a modest repression of WhiB4 (~1.3 fold), our qRT-PCR under similar conditions confirmed a significant down-regulation (-5.00  0.27 fold; P value  0.001) as compared to unstressed Mtb. The breakdown of iron homeostasis is another hallmark of oxidative stress (23). Data showed induction of Fe-responsive repressors, ideR and furB, along with the down-regulation of genes encoding Fe-siderophore biosynthetic enzymes (mbt operon) and Fe-transport (Rv1348), and up-regulation of Fe-storage (bfrB) ( Fig. 2A). Studies have suggested an important role for DnaK and ClpB chaperones in promoting recovery from oxidative stress (12,57). Our analysis identified that most of the functionally diverse nodes (sigma factors, antioxidants and redox-sensors) converge at a common stress responsive chaperone, DnaK, making it a major hub node coordinating AG stress response in Mtb (Fig. 3B).
Recently, two mycobacterial redox buffers, mycothiol (MSH) and ergothioniene (EGT), were implicated in protecting against oxidants and antibiotics (44). We compared gene expression changes displayed by MSH and EGT mutants (44) with the AG transcriptome. An ~ 60% of genes regulated by MSH and EGT also displayed altered expression in response to AG ( Supplementary Fig. S2), indicating an overlapping roles of MSH and EGT in tolerating oxidative stress and AG in Mtb (44). Lastly, we performed transcriptomics of Mtb in response to a known oxidant cumene hydroperoxide (CHP; 250 μM for 1 h [non-toxic concentration]) and compared expression changes with AG-responsive regulons. As shown in Supplementary Fig.   S3, a considerable overlap in gene expression (~30%) was observed between the two conditions (Supplementary Table S4). More importantly, genes associated with β-lactams tolerance (ponA2, ispH, and kasA) and redox-metabolism (ahpCD, trxB1, trxB2, trxC, and suf) were similarly regulated under CHP and AG challenge (Supplementary Table S4). Lastly, we validated our microarray data by performing qRT-PCR on a few genes highly deregulated upon AG treatment (Supplementary Table S3). Taken together, data indicate a major recalibration of genes regulating mycobacterial redox physiology in response to AG.

AG treatment induces redox imbalance in Mtb
We next examined whether AG exposure elicits redox stress in Mtb. The down-regulation of aerobic type NADH dehydrogenase I and ATP synthase upon AG exposure may result in reduction of electron flow through respiratory complexes. It might induce back-pressure in respiratory chain causing an increase in cellular NADH/NAD + ratio. We measured NADH/NAD + ratio of Mtb exposed to 10X MIC of AG at various time points post-treatment. At pre-lethal stage (6 h post-treatment), we did not observe any change in NADH/NAD + ratios (

Mycothiol buffer protects Mtb from AG-mediated killing
Since AG induces intramycobacterial oxidative stress, it is likely that the loss of major intracellular antioxidant, mycothiol, might potentiate the antimycobacterial activity of AG. To examine this, we used a MSH negative strain (MsmΔmshA) of Mycobacterium smegmatis (Msm), an organism that is widely used as a surrogate for pathogenic strains of Mtb. Wt Msm and MsmΔmshA strains were exposed to various concentrations of Amox at a saturating concentration of Clav (8 μg/ml) and percent growth inhibition was measured using Alamar blue (AB) assay. AB is an oxidation-reduction indicator dye that changes its color from nonfluorescent blue to fluorescent pink upon reduction by actively metabolizing cells, whereas inhibition of growth by antimycobacterial compounds interferes with AB reduction and color development (55). As shown in Fig. 6A, at a fixed Clav concentration, MsmΔmshA exhibited ~ 3-and 10-fold increased inhibition at 5 μg/ml and 2.5 μg/ml of Amox as compared to wt Msm, respectively. At 10 μg/ml of Amox, both strains showed nearly complete inhibition (Fig. 6A).
Next, we measured susceptibility to Clav at a fixed concentration of Amox (10 μg/ml). Higher  6C). In sum, AG exposure triggers ROS production and mycothiol provides efficient tolerance towards AG.

Mtb WhiB4 modulates gene expression and maintains E MSH in response to AG
Altered expression of oxidative-stress genes, elevation of ROS, and perturbation of E MSH upon AG exposure suggest that intramycobacterial redox potential can serve as an internal cue to monitor the presence of β-lactams. Canonical redox sensors such as OxyR, SoxR, and FNR are either absent or rendered non-functional in Mtb (6,9). We have recently shown that Mtb features a Fe-S cluster containing transcription factor, WhiB4, which responds to oxidative stress by regulating the expression of antioxidant genes (6). The fact that whiB4 expression is uniformly repressed by β-lactams (e.g., meropenem and AG) (30) and oxidative stress (6) Importantly, most of these enzymatic activities are well known to confer protection against oxidative and nitrosative stress in Mtb (21,31,32,40,56,58 (6). We generated thiol-reduced and -oxidized forms of apo-WhiB4 as described previously (6). The oxidized and reduced apo-WhiB4 fractions were incubated with 32 P-labeled promoter fragments (~150 bp upstream) of blaC and blaR and complex formation was visualized using EMSA.
As shown in figure 8A and 8B, oxidized apo-WhiB4 binds DNA in a concentrationdependent manner, whereas this binding was significantly reversed in case of reduced apo-WhiB4. Since WhiB4 bind to its own promoter (6), we confirmed that oxidized apo-WhiB4 binds to its promoter in concentrations comparable to that required for binding blaC and blaR promoters (Fig. 8C). Next, we performed in vitro transcription assays using a highly sensitive Msm RNA polymerase holoenzyme containing stoichiometric concentrations of principal Sigma factor, SigA (RNAP-σA) (6) and determined the consequence of WhiB4 on blaC transcript. As shown in figure 8D, addition of oxidized apo-WhiB4 noticeably inhibited transcription from blaC promoter, whereas reduced apo-WhiB4 restored normal levels of blaC transcript. Lastly, we directly measured BlaC activity in the cell-free extracts derived from wt Mtb, MtbΔwhiB4, and whiB4-OE strains using a chromogenic β-lactam nitrocefin as a substrate (15). Using in vivo thiol-trapping assay, we have earlier shown that WhiB4 predominantly exists in an oxidized apoform upon overexpression in whiB4-OE strain (6). Therefore, whiB4-OE strain will reveal the We have previously reported that treatment with 5 mM diamide or DTT did not adversely affect growth of Mtb (50). Pretreatment of whiB4-OE with DTT largely restored BlaC activity to MtbΔwhiB4 levels, whereas diamide did not lead to further decrease in BlaC activity in whiB4-OE strain (Fig. 8F). Effective reduction of oxidized apo-WhiB4 by DTT within whiB4-OE cells may have led to loss of WhiB4 DNA binding and transcription repressor activity, thereby causing elevated blaC expression and activity. Taken together, these results led us to conclude that WhiB4 regulates β-lactamase expression and activity in a redox-dependent manner.

WhiB4 regulates survival in response to β-lactams in Mtb
Based on above results, we hypothesize that WhiB4 -sufficient and -deficient strains would have differential susceptibility towards β-lactams. We found that MtbΔwhiB4 uniformly displayed ~ 4-8 folds higher MICs against β-lactams as compared to wt Mtb (Table 1). This effect was specific to β-lactams, as the loss of WhiB4 did not alter MICs for other anti-TB drugs such as INH and RIF (Table 1). More-interestingly, over-expression of WhiB4 displayed ~ 2-4 folds greater sensitivity towards β-lactams as compared to wt Mtb (Table 1). We predicted that if WhiB4 is controlling tolerance to β-lactams by regulating blaC expression, we would see dehydrogenase I) to less energy efficient course (e.g., NDH, CydAB oxidase), and any interference with this respiratory-switch over (e.g., CydAB mutation) led to resensitization of mycobacteria to antibiotics (29,41). This seems to be a unifying theme underlying tolerance to conventional as well as the newly discovered anti-TB drugs bedaquiline (BDQ) and Q203 (27).
Agreeing to this, we found that exposure of Mtb to AG elicited a transcriptional signature indicating a shift from energy efficient respiration to energetically less favored pathways as shown by a significant induction of ndh and cydAB transcripts and a down-regulation of nuo, cydbc1, and atp A-H. In bacteria, including Mtb, cytochrome bd oxidase also displays catalase and/or quinol oxidase activity (1,29), which confers protection against oxidative stress and nitrosative stress. On this basis, upregulation of cytochrome bd oxidase in response to AG is indicative of oxidative stress in Mtb. Bactericidal antibiotics, including β-lactams, have been consistently shown to produce ROS as a maladaptive consequence of primary drug-target interaction on TCA cycle and respiration (11,26,28). While this proposal has been repeatedly questioned, it is strongly reinforced by multiple independent studies demonstrating that tolerance to antibiotics is linked to the bacterial ability to detoxify antibiotic-triggered ROS generation (19,38,49,60). We confirmed that AG stimulates oxidative stress in Mtb in vitro and during infection. However, in contrast to other studies (26) (6). Since oxidized apo-WhiB4 is known to repress its own expression (6), Mtb can down-regulate the expression of whiB4 by elevating the levels of oxidized apo-WhiB4 in response to oxidative stress caused by β-lactams. The downregulation of WhiB4 can reduce its negative influence on gene expression, necessary to adjust the expression of blaI, blaR, blaC, and genes involved in maintaining respiration and redox balance to neutralize β-lactam toxicity. However, considering that WhiB4 binds DNA nonspecifically with a preference for GC rich sequences (6), the exact molecular mechanism of how WhiB4 regulates expression in response to β-lactams will be investigated further.

Bacterial strains, mammalian cells and growth conditions
The mycobacterial strains were grown aerobically in 7H9 broth (Difco) or 7H11 agar  (39).

Drug sensitivity assay
Sensitivity to various drugs was determined using microplate alamar blue assay (AB). acetonitrile (0.1% trifluoroacetic acid) was applied as mobile phase while flow rate was maintained at 0.5 ml/min. The HPLC method used was as described previously (25).

Intracellular NADH/ NAD + ratio
NADH/NAD + ratios upon AG treatment were determined by NAD/NADH Quantification

Microarray hybridisation and data analysis
For microarray analyses, the wt Mtb and MtbΔwhiB4 strains were cultured to an OD 600 0.4 and exposed to AG (100 µg/ml of Amox and 8 µg/ml of Clav) for 6 h. For CHP stress, wt Mtb grown similarly was treated with 250 µM of CHP for 1 h and samples was processed for microarrays. Total RNA was isolated from samples (taken in replicates), processed and hybridized to Mtb Whole Genome Gene Expression Profiling microarray-G2509F (AMADID: G2509F_034585, Agilent Technologies PLC) and data was analysed as described (33). DNA microarrays were provided by the University of Delhi, South Campus, MicroArray Centre (UDSMAC). RNA amplification, cDNA labeling, microarray hybridization, scanning, and data analysis were performed at the UDSMAC as described (33). Slides were scanned on a microarray scanner (Agilent Technologies) and analyzed using GeneSpring software. Results were analyzed in MeV with significance analysis of microarrays considered significant at p≤0.05. The normalized data from the microarray gene expression experiment have been submitted to the NCBI Gene Expression Omnibus and can be queried via Gene Expression Omnibus series accession number GSE93091 (AG exposure) and GSE73877 (CHP exposure).

Constructing of condition-specific networks:
Global PPI network was generated using the dataset described in the studies outlined in Supplementary where i and j denotes nodes present in an edge.

Edge cost = 1/ EW
The main focus of the study was to identify the key players involved in regulating the variations in different conditions. We carried out shortest path analysis on the condition-specific networks and selected the paths that are most perturbed in these conditions. We implemented shortest path algorithm to obtain the results.

Shortest path analysis
The edge cost values were used as an input for calculating all vs. all shortest paths in each condition using Zen (http://www.networkdynamics.org/static/zen/html/api/algorithms/shortest_path.html). More than 9000000 paths were obtained for each condition. In order to analyze the more significant paths, we ordered the paths on the basis of their path scores. Path score is the summation of the edge cost that constitutes a path. Based on the formula considered for calculating edge cost, lower path score indicates that the nodes in the path have higher expression. So, instead of analyzing 9000000 paths, we considered subnetworks, which comprise of top 1% of the network. These networks were visualized using Cytoscape 3 (48). Our response networks competently explain the perturbations in the system upon exposure to different situations such as AG treatment and/or disruption of whiB4. The networks were further co-related to graph-theory based methods and differentially regulated paths were recognized in each condition to construct sub-network for each condition (Supplementary Table S8).

qRT-PCR analysis
Total RNA was isolated as described previously (6) and cDNA was synthesized (after DNase treatment) from 500 ng isolated RNA. Random oligonucleotide primers were used with iScript TM Select cDNA Synthesis Kit (Bio-Rad) for cDNA synthesis. Gene specific primers (Supplementary Table S9) were selected for RT-PCR (CFX96 RT-PCR system, Bio-Rad) and iQ TM SYBR Green Supermix (Bio-Rad) was used for gene expression analysis. In order to obtain meticulous expression levels, PCR expression was normalized and CFX Manager TM software (Bio-Rad) was utilized for data analysis. Gene expression was normalized to Mtb 16S rRNA expression.

Electrophoretic Mobility Shift Assays (EMSA)
The histidine-tagged WhiB4 purification and generation of reduced or oxidized apo-WhiB4 was done as described previously (6). For EMSA assays, the promoter fragments of whiB4, blaC, and blaR (~150 bp upstream of translational start codon) were PCR amplified from the Mtb genome and the 5' end was labelled using [γ-32 P]-ATP labelled oligonucleotides by using T4 polynucleotide kinase (MBI Fermentas) as per the manufacturer's instructions. Binding reactions were performed in 1X TBE buffer (89 mM Tris, 89 mM boric acid and 1 mM EDTA; pH 8.4) for 30 min and 5% polyacrylamide gel was used to resolve protein-DNA complexes. Gels were exposed to auto radiographic film and visualized via phosphoimaging (GE). and heated at 95°C for 5 minute followed by snap chilling in ice for 2 minutes. The transcripts were resolved by loading samples on to 6% urea-PAGE.

Intramycobacterial E MSH measurement in vitro and during infection
Mycobacterial strains expressing Mrx1-roGFP2 were grown in 7H9 medium till an OD 600 0.4 and exposed to various concentration of AG. For measurements, cell were treated with 10 mM N-Ethylmaleimide (NEM) for 5 min at room temperature (RT) followed by fixation with 4% paraformaldehyde (PFA) for 15 min at RT. After washing thrice with 1X PBS, bacilli were analyzed using BD FACS Verse Flow cytometer (BD Biosciences). The biosensor response was measured by analyzing the ratio at a fixed emission (510/10 nm) after excitation at 405 and 488 nm as described (4). Data was analyzed using the FACSuite software. For measuring intramycobacterial E MSH during infection, PMA-differentiated THP-1 cells were infected with Mtb strains expressing Mrx1-roGFP2 (moi: 10). Infected macrophages were treated with NEM/PFA, washed with 1X PBS, and analyzed by flow cytometry as described previously (39).

Survival assay upon AG treatment in vitro and during infection.
Mtb strains were grown aerobically till OD 600 0.

Vancomycin-BODIPY staining
The pattern of nascent peptidoglycan synthesis was observed by fluorescent staining as described (53). Mtb strains were grown to exponential phase (OD 600 0.6) in 7H9 medium. 1 ml of culture was incubated with 1 µg/ml Vancomycin-BODIPY (BODIPY ® FL Vancomycin, Thermo Fisher Scientific) for 16 h under standard growth conditions. Cells were pelleted to remove excess stain and fixed with PFA. After washing with 1X PBS, culture aliquots (20 µl) were spread on slides and allowed to air dry. The bacterial cells were visualized for BODIPY ® FL Vancomycin fluorescence (excitation at 560 nm and emission at 590 nm) in a Leica TCS Sp5 confocal microscope under a 63X oil immersion objective. Staining pattern of more than 150 cells was observed for each strain and cell length was measured as well using Image J software.

Statistical Analysis
Statistical analyses were performed using GraphPad Prism software. The statistical significance of the differences between experimental groups was determined by two-tailed, unpaired Student's t test. Differences with a p value of ≤0.05 were considered significant.

Ethics statement
This study was carried out in strict accordance with the guidelines provided by the

Author Disclosure Statement
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We confirm that no competing financial interests exist.       60. Wang X, Zhao X. Contribution of oxidative damage to antimicrobial lethality.