Decision letter: GWAS and functional studies suggest a role for altered DNA repair in the evolution of drug resistance in Mycobacterium tuberculosis

Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract The emergence of drug resistance in Mycobacterium tuberculosis (Mtb) is alarming and demands in-depth knowledge for timely diagnosis. We performed genome-wide association analysis using 2237 clinical strains of Mtb to identify novel genetic factors that evoke drug resistance. In addition to the known direct targets, we identified for the first time, a strong association between mutations in DNA repair genes and the multidrug-resistant phenotype. To evaluate the impact of variants identified in the clinical samples in the evolution of drug resistance, we utilized knockouts and complemented strains in Mycobacterium smegmatis and Mtb. Results show that variant mutations compromised the functions of MutY and UvrB. MutY variant showed enhanced survival compared with wild-type (Rv) when the Mtb strains were subjected to multiple rounds of ex vivo antibiotic stress. In an in vivo guinea pig infection model, the MutY variant outcompeted the wild-type strain. We show that novel variant mutations in the DNA repair genes collectively compromise their functions and contribute to better survival under antibiotic/host stress conditions. Editor's evaluation This paper provides important evidence implicating polymorphisms in the mycobacterial adenine DNA glycosylase, MutY, in the emergence of antibiotic resistance in Mycobacterium tuberculosis. While the precise mechanism underlying this phenotype requires further investigation, the inference from genome-wide association analyses of sequenced clinical isolates, supported by laboratory experiments and animal infection models, is convincing. This work adds a new locus of interest to the list of polymorphisms associated with tuberculosis drug resistance, and is likely to be relevant to the mycobacterial research field. https://doi.org/10.7554/eLife.75860.sa0 Decision letter Reviews on Sciety eLife's review process Introduction The acquisition of drug resistance in Mycobacterium tuberculosis (Mtb) has evoked a perilous situation worldwide (WHO, 2020). Resistance to isoniazid and rifampicin, the first-line drugs, results in multidrug resistant-TB (MDR-TB). The pathogen is defined as extensively drug resistant (XDR) when it becomes resistant to first-line TB drugs, any fluoroquinolones, and at least one additional Group A drug (moxifloxacin, levofloxacin, linezolid, and bedaquiline) (WHO, 2021). Prolonged treatment duration, high drug toxicity, and the expensive drug regimen pose a challenge for treating the MDR and XDR-TB. Moreover, the inadequate treatment of drug-resistant TB leads to the augmentation of resistance to other anti-TB drugs, increasing the probability of transmission of these strains in the population (Alexander and De, 2007; Bastos et al., 2014). Seven major lineages of Mtb are present across the globe, out of which four lineages: Lineage 1-Indo Oceanic (EAI); Lineage 2- Beijing; Lineage 3- Central Asian (CAS); and Lineage 4- Euro-American (Gagneux et al., 2006) are prevalent in humans. Clinical strains belonging to lineage 2 are more prone to developing drug resistance than lineage 4 strains (Ford et al., 2013). Acquisition of drug resistance in Mtb is majorly attributed to the chromosomal mutations that either modify the antibiotic’s direct target or increase the expression of efflux pumps that helps in decreasing the effective concentration of the drug inside the cell. The expression of drug modifying/degrading enzymes also contributes to the acquisition of drug resistance (Gygli et al., 2017). Despite well-known mechanisms of drug resistance, it is difficult to predict the resistance based on direct target mutations alone, implying the presence of hitherto unknown mechanisms that impart drug resistance. In addition, the diagnosis based on mutations in a particular known target region increases the bias that may have multiple repercussions, such as misdiagnosis and eventual spread of drug resistance (CRyPTIC Consortium and the 100,000 Genomes Project et al., 2018; The CRyPTIC Consortium, 2022b; The CRyPTIC Consortium, 2022a). The current knowledge of the mechanisms and biological triggers involved in the evolution of MDR or XDR-TB is inadequate. This knowledge is crucial for developing new drug targets and improved diagnosis. Multiple efforts have been made to determine the mechanisms for the emergence of MDR and XDR-TB. Genome-wide association studies (GWAS) identified different genes that abet the emergence of drug resistance (Farhat et al., 2013; Hicks et al., 2018; Zhang et al., 2013; Safi et al., 2019). However, only a few genes, such as ponA1, prpR, ald, glpK, and the mutation in the thyA-Rv2765 thyX-hsdS.1 loci, are validated (Farhat et al., 2013; Hicks et al., 2018; Zhang et al., 2013; Safi et al., 2019). In a quest to identify genetic triggers that aid in the evolution of antibiotic resistance in Mtb, we performed GWAS using global data set of 2237 clinical strains that consist of antibiotic susceptible, MDR, poly-drug resistant (resistant to more than one first-line anti-TB drug other than both isoniazid and rifampicin), pre-XDR, and XDR. Interestingly, we have identified mutations in the multiple DNA repair genes of Mtb associated with the MDR phenotype. Functional validation of the identified mutations in DNA repair enzymes revealed that perturbations in the DNA repair mechanisms result in the enhanced survival of strains in the presence of antibiotics ex vivo and in vivo. Results GWAS unveils mutations in the DNA repair genes To identify the genetic determinants contributing to the development of antibiotic resistance in Mtb, we performed genome-wide association analysis using the whole-genome sequences of clinical strains from nine published studies. After the quality filtering of raw reads, the dataset had 2773 clinical strains from 9 different countries belonging to all 4 lineages (Figure 1a, Figure 1—figure supplement 1; Hicks et al., 2018; Zhang et al., 2013; Casali et al., 2014; Blouin et al., 2012; Shanmugam et al., 2019; Guerra-Assunção et al., 2015; Clark et al., 2013; Bryant et al., 2013; Walker et al., 2013). The dataset chiefly represented Lineage 2 and 4 isolates that are predominant across the globe (Figure 1b). Based on the computational predictions and the phenotypes provided by the previous studies, strains were categorized as susceptible, mono-drug resistant, MDR, Poly-DR, and pre-XDR (Supplementary file 1; Supplementary file 2Manson et al., 2017). We identified ~160,000 single nucleotide polymorphisms (SNPs) and indels after mapping the short reads on the reference Rv genome. A phylogenetic tree constructed using the SNPs shows the proper clustering of lineages (Figure 1b). The total number of SNPs observed for all the strains was comparable, suggesting the absence of genetic drift (Figure 1c). Figure 1 with 1 supplement see all Download asset Open asset Genome-wide association study unveils mutations in the DNA repair genes. (a) Geographical distribution of 2773 clinical strains of Mycobacterium tuberculosis (Mtb). The donut plot represents the proportion of susceptible and drug-resistant (DR) strains in each lineage. DR includes mono-DR, poly-DR, multidrug resistant (MDR), and pre-extensively drug resistant (XDR). A detailed breakup of distribution is given in Supplementary file 1. (b) Phylogenetic tree constructed using 1,60,000 single nucleotide polymorphisms (SNPs) using Mycobacterium canetti as an outgroup. (c) Dot-plot showing the number of SNPs identified in each strain. Different colored dots indicate the drug resistance phenotype of strain. We hypothesized that the probability of finding the genetic determinants contributing to drug resistance would be higher in the strains resistant to more than two antibiotics. Thus, we performed GWAS using 1815 drug-susceptible and 422 drug-resistant strains (Figure 1—figure supplement 1, Figure 2—figure supplements 1–4 & Supplementary file 2). We employed a genome association and prediction integrated tool (GAPIT) software, with stringent false discovery rate (FDR) adjusted p-value (Gao et al., 2016; Zegeye et al., 2014). After setting the adjusted p-value cut-off at 10–5, we identified 188 mutations, including 24 intergenic regions correlated with multidrug resistance (Supplementary file 3; Supplementary file 4; Supplementary file 5; Supplementary file 6). The effect of identified SNPs on the development of MDR/XDR reveals positive or negative contributions (Figure 2a). We have identified known first- and second-line drug resistance target genes (Figure 2b & Table 1). Although we identified multiple mutations in rpoB, only p.Leu452Pro and p.Val496Met were above the cut-off. Notably, mutations in the rrs, katG (p.Ser315Thr), embB (p.Gly406Ser), pncA (p.His71Arg), gyrA (p.Ala90Val), and recently reported genetic determinants such as folC (p.Ser150Gly), and pks were part of the 164 genes, validating our approach (Figure 2b & Table 1, Supplementary file 4; Supplementary file 6). Also, we identified the known compensatory mutations in the fabG1 upstream region, eis-Rv2417c, and oxyR-ahpC loci (Coll et al., 2018; Supplementary file 6). Importantly, the above mutations were absent in the mono-DR and drug-susceptible strains (Figure 2—figure supplement 6). Table 1 Mutations identified in the direct targets of antibiotics. AntibioticGeneMutations identifiedRifampicinrpoBLeu452Pro, Val496MethIsoniazidkatGSer315ThrEthambutolembBGly406SerOfloxacingyrAAla90Val, Ser91ProKanamycinrrs7 independent mutationsPyrazinamidepncAHis71ArgEthionamideethAMet95Arg, Pro160(frame-shift)StreptomycingidBLeu35 (frame-shift)CycloserinealdThr427Pro Figure 2 with 6 supplements see all Download asset Open asset Drug-resistant strains carry mutations in the DNA repair genes. (a) Volcano plot represents the effect of identified single nucleotide polymorphisms (SNPs) on the development of multidrug resistant/extensively drug resistant TB (MDR/XDR-TB). The positive effect (green dots) shows that the identified SNPs would aid in MDR/XDR development. The negative effect (red dots) shows that the SNPs would restrain the development of MDR/XDR. (b) Manhattan plot representing the association between the genes and drug resistance phenotype. A total of 188 genes that include intergenic regions were identified above the 10–5 cut-off value through association studies. Blue dots represent mutation in the lipid metabolism, membrane proteins, intermediary metabolism genes, and others. Green dots represent mutation in the direct targets for the first- and second-line antibiotics. Red dots represent mutations associated with the DNA repair genes. A detailed list of associated genes is provided in Supplementary file 3; Supplementary file 4. (c–e) Pie chart represents the total (c), non-synonymous (d), and synonymous (e) SNPs identified in the genes that belong to different categories. (f) Bar plot represents the percentage of synonymous mutations in the genes that resulted in abundant/moderate codon usage to the rare codon compared to H37Rv. Figure 2—source data 1 Mutations identified in genes that belong to different categories. https://cdn.elifesciences.org/articles/75860/elife-75860-fig2-data1-v3.xlsx Download elife-75860-fig2-data1-v3.xlsx Among the 188 genes, 45% of the mutations resulted in non-synonymous changes, whereas 33% resulted in synonymous changes, 14% in the upstream regions of the genes, and 8% in the stop/frameshift mutations (Figure 2c–e & -- Supplementary file 3; Supplementary file 4; Supplementary file 5; Supplementary file 6). While the non-synonymous and the stop/frameshift mutations most likely affect the functions of the proteins, the intergenic region mutations may impact gene expression. We identified mutations in genes involved in lipid metabolism, intermediary metabolism and respiration, membrane transporters, cell wall and cell processes, membrane-associated proteins, and others (https://mycobrowser.epfl.ch/) (Figure 2c–e). Synonymous changes may alter the mRNA stability or stall the translation process by changing an abundant codon to a rare codon (Brandis and Hughes, 2016; Kristofich et al., 2018; Plotkin and Kudla, 2011). Analysis of the synonymous mutations for the codon bias revealed that in >50% of events, codons were converted from moderate/abundant to rare codons (compare 75% in MDR/XDR with 25% in H37Rv) (Figure 2f, Supplementary file 7). In addition to the mutations described above, we identified novel mutations in base excision repair (BER), nucleotide excision repair (NER), and homologous recombination (HR) pathway genes, mutY, uvrA, uvrB, and recF that are associated with the MDR and XDR-TB (Figure 2b & Table 2). Mutations in the DNA repair pathway genes could contribute to the selection and evolution of antibiotic resistance (Table 2; Figure 2—figure supplement 5). Analysis showed that mutations in the DNA repair genes are distributed specifically in MDR, PDR, and preXDR/XDR strains (Figure 2—figure supplement 6a–d). Furthermore, these strains also harbored mutations in the direct targets of the antibiotics (Figure 2—figure supplement 6i–k). Collectively, in addition to mutations in the direct targets, we identified novel uncharacterized variants, including mutations in the DNA repair genes. Table 2 Mutations in DNA repair genes associated with drug resistance phenotype. GeneAmino acid changeWild typeMutatedFalse discovery rate-adjusted p-valuemutYArg262GlnGA3.83E-09uvrBAla524ValCT2.15E-07uvrAGln135LysCA3.83E-09RecFGly269GlyGT2.15E-07 The mutations in DNA repair genes result in their deficient function DNA repair pathways, including NER, HR, and BER, guard the genomic integrity (Cole et al., 1998; Singh, 2017). The Mtb MutY is a 302 amino acid (aa) long adenine DNA glycosylase encoded by rv3589. We identified Arg262Gln mutation at the C-terminal region of the MutY. Oxidative damage to DNA results in the formation of 7,8-dihydro 8-oxoguanine (8-oxoG). If left unrepaired by MutM (fpg), it results in 8-oxoG:A (mostly) or 8-oxoG:G base pairing. MutY removes A or G paired against 8-oxoG, allowing MutM to correct the mistake. The absence of repair leads to G:C to T:A or C:G mutations in the genome (Figure 3a; Kurthkoti et al., 2010). The analysis of the mutation spectrum in the drug-resistant clinical strains harboring mutY-R262Q mutation and closely related drug-susceptible strains showed a bias toward C→A, A→G, and C→T mutations (Figure 2—figure supplement 6l). To decipher the biological role of the identified variant, we cloned Mtb mutY and performed site-directed mutagenesis to generate the mutant allele. The wild type and mutant mutY genes were subcloned into an integrative Mtb shuttle vector. Constructs were electroporated into Mycobacterium smegmatis mutY mutant strain (msmΔmutY) to generate msmΔmutY::mutY and msmΔmutY::mutY-R262Q strains. We performed mutation frequency analysis to evaluate the impact of Arg262Gln mutation on its DNA repair function. In accordance with the published data, deletion resulted in a 4.32-fold increase in the mutation frequency (Figure 3b; Kurthkoti et al., 2010). While complementation with wild-type mutY rescued the phenotype, complementation with mutY-R262Q failed to do so (Figure 3b and c). Figure 3 with 2 supplements see all Download asset Open asset Variants identified in DNA repair genes abrogate their function. (a) A schematic representation of the base excision repair pathway that operates in mycobacteria. Oxidative damage can result in the conversion of G to 8-oxo-G. If MutM (Fpg) does not repair 8-oxo-G before replication, often an A is inserted against 8-oxo-G during replication. Under these conditions, MutM must avoid repair of 8-oxo-G until MutY removes the erroneously incorporated A. The predominant target of MutY is 8-oxo-G:A pair where it removes A; thus, the action of MutY provides another opportunity to incorporate C (the correct base) against 8-oxo-G. Now the DNA becomes a target for MutM again, leading to the removal of 8-oxo-G and allowing incorporation of G. (b) Mutation frequency was calculated using msm, msmΔmutY, msmΔmutY::mutY, msmΔmutY::mutY-R262Q. (c) Fold increase in the mutation frequency with respect to wild-type msm. (d) A schematic representation of the nucleotide excision repair pathway showing the recognition and initiation of repair by UvrA-UvrB and UvrC. (e) Mutation frequency of msm, msmΔuvrB, msmΔuvrB::uvrB, and msmΔuvrB::uvrB-A524V. (f) Fold increase in the mutation frequency with respect to wild-type msm. Two biologically independent experiment sets were performed. Each biological experiment was performed in a biological sextet. Data represent one set of experiments. Statistical analysis (two-way ANOVA) was performed using Graph pad prism software. *p<0.0001, p<0.001, and *p<0.01. (g, h & i) Mutation rate was calculated for different strains in the presence of isoniazid (g), rifampicin (h), or ciprofloxacin (i). (j) Table showing the fold increase in the mutation rate in comparison with wild-type Rv. The experiment was performed using six independent colonies. Data represent mean and standard deviation. Statistical analysis (two-way ANOVA) was performed using Graph pad prism software. *p<0.0001, p<0.001, and *p<0.01. Figure 3—source data 1 Mutation rate analysis in the presence of different drugs. https://cdn.elifesciences.org/articles/75860/elife-75860-fig3-data1-v3.xlsx Download elife-75860-fig3-data1-v3.xlsx Next, we investigated the role of mutation in the NER pathway gene UvrB. UvrA, UvrB, and UvrC recognize and initiate the NER pathway upon DNA damage. UvrB, a 698 aa long DNA helicase encoded by rv1633, plays a pivotal role in the NER pathway by interacting with the UvrA and UvrC (Figure 3d; Kurthkoti et al., 2008). UvrB harbors N and C-terminal helicase domain, interaction domain, YAD/RRR motif, and UVR domain. Identified UvrB variant, Ala524Val, mapped to the C-terminal helicase like domain (Theis et al., 2000). To evaluate the functional significance of the mutation of uvrB, Mtb uvrB and uvrB-A524V genes were cloned into an integrative vector. The absence of uvrB led to higher mutation frequency, which could be rescued upon complementation with the wild type but not with the variant (Figure 3e and f). Subsequently, we sought to extend our investigations in Mtb. Toward this, we generated the gene replacement mutant of mutY in laboratory strain Mtb H37Rv (Rv), wherein the mutY at native loci was disrupted with a hygromycin resistance cassette. Replacement at the native loci was confirmed by performing multiple PCRs (Figure 3—figure supplement 1b–c). Complementation constructs harboring mutY or mutY-R262Q were electroporated in the RvΔmutY to generate RvΔmutY::mutY and RvΔmutY::mutY-R262Q. Western blot analysis showed comparable expression of the MutY or MutY-R262Q (Figure 3—figure supplement 1d). We determined the mutation rates in the presence of isoniazid, rifampicin, and ciprofloxacin (Figure 3g–j). The fold increase in the mutation rates relative to Rv for RvΔmutY, RvΔmutY:mutY, and RvΔmutY::mutY-R262Q were 2.90, 0.76, and 3.0 in the presence of isoniazid; 5.62, 1.13, and 5.10 in the presence of rifampicin; and 9.14, 1.57, and 8.71 in the presence of ciprofloxacin, respectively (Figure 3j). Also, we have determined the mutation frequencies in the presence of isoniazid and rifampicin (Figure 3—figure supplement 2). Results are in line with the mutation rate experiments presented in Figure 3. Together these data suggest that variants of mutY and uvrB compromise their function. The variant of mutY resists antibiotic killing The killing kinetics in the presence and absence of isoniazid, rifampicin, ciprofloxacin, and ethambutol was performed to evaluate the effect of different drugs on the survival of RvΔmutY or RvΔmutY::mutY-R262Q (Figure 4a). In the absence of antibiotics, the growth kinetics of Rv, RvΔmutY, RvΔmutY::mutY, and RvΔmutY::mutY-R262Q were similar (Figure 4b). In the presence of isoniazid, ~2 log-fold decreases in bacterial survival was observed on day 3 in Rv and RvΔmutY::mutY; however, in RvΔmutY and RvΔmutY::mutY-R262Q, the difference was limited to ~1.5 log-fold (Figure 4c). A similar was on 6 and suggesting an increase in the survival of RvΔmutY and RvΔmutY::mutY-R262Q compared with Rv and RvΔmutY::mutY (Figure 4c). Interestingly, in the presence of we not any difference (Figure In the presence of rifampicin and ciprofloxacin, we observed an increase in the survival of RvΔmutY and RvΔmutY::mutY-R262Q compared with Rv and RvΔmutY::mutY (Figure Thus, results suggest that the absence of mutY or the presence of mutY variant in the antibiotic stress. Figure 4 Download asset Open asset kinetics in the presence of antibiotics show better survival of RvΔmutY and RvΔmutY::mutY (a) representation of killing (b) kinetics in the absence of drugs. kinetics in the presence of isoniazid, rifampicin, ciprofloxacin, and Two biologically independent sets of experiments were performed. Each biological experiment was performed in biological Data represent one set of experiments. Statistical analysis (two-way ANOVA) was performed using Graph pad prism software. *p<0.0001, p<0.001, and *p<0.01. Figure data 1 kinetics in the absence and presence of different antibiotics. Download The variant of mutY survival ex vivo Next, we the survival of Rv, RvΔmutY, RvΔmutY::mutY, and RvΔmutY::mutY-R262Q in the We not any in the survival of RvΔmutY or RvΔmutY::mutY-R262Q compared with Rv or RvΔmutY::mutY (Figure supplement We that the evolution of a strain to requires the presence of antibiotic and stress. we with Rv, RvΔmutY, RvΔmutY::mutY, and RvΔmutY::mutY-R262Q in the absence or presence of the antibiotics. The after were in for and for the of The process was for and were at 4 and during the of infection (Figure at 4 showed RvΔmutY and RvΔmutY::mutY-R262Q better survival in the absence of antibiotics than Rv and RvΔmutY::mutY (Figure was additional compared with in the presence of isoniazid (Figure However, we observed a log-fold for RvΔmutY and RvΔmutY::mutY-R262Q compared with Rv or RvΔmutY::mutY in the presence of rifampicin or ciprofloxacin (Figure Figure with 1 supplement see all Download asset Open asset Mutations in the DNA repair genes a survival in the presence of antibiotics. (a) A schematic is representing the ex vivo infection experiment in the presence and absence of different antibiotics. of the strains in the at 4 and and with antibiotics or rifampicin, or (f) survival with respect to 4 was determined for each strain and with antibiotics or rifampicin or Two biologically independent sets of experiments were performed. Each biological experiment was performed in biological Data represent one set of experiments. Statistical analysis (two-way ANOVA) was performed using Graph pad prism software. *p<0.0001, p<0.001, and *p<0.01. Figure data 1 of different strains in the absence and presence of antibiotics ex vivo. Download Acquisition of direct target mutations ex vivo in the presence of drugs We sought to determine the improved survival of mutY mutant and mutY variant in the above experiment (Figure is to the acquisition of mutations in the direct target of antibiotics. To identify the mutations, we performed DNA from independent in was in proportion to SNPs present in of the reads were for the Analysis of Rv sequences in that the laboratory strain SNPs compared with the reference strain not The of the Rv laboratory strain was as the reference for the data for RvΔmutY, RvΔmutY::mutY, and RvΔmutY::mutY-R262Q strains in not show the presence of any mutations in the antibiotic target genes. In a similar independent each from the after the of ex vivo infection in the presence or absence of antibiotics, were for Data that in the absence of antibiotics, direct target mutations were identified in the ex vivo strains (Figure & However, in the presence of isoniazid, we mutations in the katG or in the Rv, RvΔmutY but not in RvΔmutY::mutY and RvΔmutY::mutY-R262Q (Figure & are in with the ex vivo evolution wherein we not a increase in the survival of RvΔmutY and RvΔmutY::mutY-R262Q in the presence of isoniazid (Figure 5). In the presence of ciprofloxacin and rifampicin, direct target mutations were identified in the gyrA and (Figure mutations were identified in mutations were identified in of RvΔmutY and RvΔmutY::mutY-R262Q, direct target mutations were identified in the Rv and RvΔmutY::mutY, suggesting that the DNA repair in the drug mutations in Mtb (Figure & Supplementary file Figure 6 Download asset Open asset reveals the acquisition of direct target mutations in the ex vivo strains. plot showing the analysis of the strains ex vivo in the absence (a) and in the presence of isoniazid rifampicin (c), and ciprofloxacin The represents the reference genome with the known direct target to represent Rv, RvΔmutY, RvΔmutY::mutY, and RvΔmutY::mutY-R262Q genome. (e) represents the single nucleotide polymorphisms identified in the Rv, RvΔmutY, RvΔmutY::mutY, and RvΔmutY::mutY-R262Q after ex vivo experiment ex vivo reveals MutY variant a survival Results above that the mutY variant survival when subjected to antibiotic likely to its to We this is the the mutY variant may the wild-type Rv when both the strains are present To this we with a of Rv RvΔmutY or Rv RvΔmutY::mutY or Rv RvΔmutY::mutY-R262Q. were and Mtb were on or (Rv) or hygromycin RvΔmutY::mutY, and to evaluate the survival rates (Figure supplement The survival was calculated as on or on is from the data that the survival rates of strains were comparable, suggesting that mutY deletion or complementation with variant not a (Figure supplement 1c). results indicate that in the absence of antibiotic deletion or the presence of mutY variant does not an (Figure supplement To these we performed ex vivo experiment with the strains that were subjected to rounds of selection (Figure 5). were with a of Rv RvΔmutY or Rv RvΔmutY::mutY or Rv RvΔmutY::mutY-R262Q (Figure 24 were either or not with an antibiotic for the and total were as described above to evaluate the survival rates (Figure supplement was difference in the either at 4 (Figure or 24 not RvΔmutY and RvΔmutY::mutY-R262Q strains showed a Rv both in the absence or presence of antibiotics. Importantly, RvΔmutY::mutY not show any Rv under any conditions. results suggest that deletion or variant strains to antibiotic stress in the helps in evolution of the strains that can the wild-type strain. Figure Download asset Open asset of mutY Rv in (a) representing the experiment performed in a