Previous studies showed that M. smegmatis with the gain of H₂O₂ resistance, named mc²51, improves growth fitness in mycobacteria under stress (Table 1) (Li et al., 2014a; Li et al., 2014b). Prior to selecting H₂O₂-adapted mycobacterial mutants, we first measured the MIC of wild-type mc²155 to H₂O₂, which was 0.039 mM. We used 0.0293 mM H₂O₂ for the initial screening, which was lower than the MIC of mc²155 to H₂O₂ (Table 1). The whole process was performed as follows: cultures were started from glycerol-frozen stocks and grown to log phase (OD₆₀₀ of 0.6–0.8). Then, the culture was diluted 1,000 times and grown under 0.0293 mM H₂O₂ until the OD₆₀₀ reached logarithmic (log) phase. Cultures were then further diluted 1:1,000, and an additional 0.0293 mM of H₂O₂ was added to the culture. This process was repeated until the H₂O₂ concentrations reached 0.4395 mM. In further rounds of culture, H₂O₂ was added in steps of 0.0879 mM instead of 0.0293 mM until a concentration of 1.5 mM was reached (Supplementary Figure S1; Supplementary Table S1). The actual MIC of H₂O₂ of the mutant strain, named mc²51, selected at 1.5 mM, was 3.125 mM. In a previous study, the mutated mycobacterial strain, mc²51, evolved into an M. tuberculosis-like strain that presented slow growth and improved growth fitness under stress (Li et al., 2014b), which H₂O₂ resistance in mycobacteria linked to virulence. Previous studies including ours have associated isoniazid (INH) with H₂O₂ resistance (Timmins and Deretic, 2006; Hu et al., 2021). The MIC of isoniazid (INH) for mc²51 was dramatically reduced to ∼1% of the MIC for mc²155 (Table 1). To test our hypothesis that the H₂O₂-resistant strain mc²51 would be more persistent in the host than its parental strain mc²155, we performed non-invasive intranasal infections, which can induce respiratory mucosal immune responses and is a promising way of vaccination for respiratory infection diseases. As shown in Figure 1A, intranasal infection of C57BL/6 mice with ∼1 × 10⁷/50 μl and bacterial colony-forming units (CFUs) were counted 1 day after infection. Compared to mc²155 with the percentage of survival in the infected lung of 0.03653 ± 0.01462% (n = 3), the mc²51 strain with that of 0.7153 ± 0.2597% (n = 3) had a significantly higher survival percentage (p = 0.0451) (Figure 1B). After infection with live mycobacteria, the bacilli enter alveolar macrophages (AMΦs) and persist within them. To exclude the possibility that the higher CFUs were caused by the larger number of infected bacilli being captured by AMΦs, macrophage-killing assays were explored using the THP-1 cell line. The survival percentage of mutant mc²51 is 0.3158 ± 0.1502 (n = 9), while the survival of wild-type mc²155 was 0.01921 ± 0.01223 (n = 6) (p = 0.0003) (Figure 1C). These results indicated that mc²51 exhibited enhanced virulence.

M. tuberculosis has evolved to survive in hypoxic conditions. Over more than 100 years of research, M. tuberculosis has been confirmed to be an obligate aerobic bacterium that cannot replicate under hypoxic conditions. However, M. tuberculosis has incredible survivability in long-term anaerobic environments. Evidence suggests that M. tuberculosis has the ability to reduce the respiratory system to low levels and maintain vitality (Loebel et al., 1933). Due to respiratory depression, ATP is maintained at low levels, which guarantees minimal metabolic activity to ensure membrane integrity under hypoxic conditions. We first compared the survival of mc²51 and mc²155 strains under hypoxic conditions. We established a Wayne dormancy model (Wayne and Hayes, 1996), consisting of mycobacterial strain cultures grown in the 7H9 medium to an OD₆₀₀ of 1.0, which were then transferred to anaerobic tubes containing 1 × 10⁶ cells/ml with a headspace ratio of the culture system of 0.5. Methylene blue (1.5 mg/L) was added as an indicator of oxygen status. The OD₆₀₀ and CFUs at different indicated times and the color transition time were measured. The indicator of mc²51 had become colorless on the 6th day (141 h); on the contrary, the indicator of mc²155 was still in blue at this time (Figure 2A). For mc²155, the indicator became colorless in about 8 days (189 h), indicating that it has entered an anaerobic state. The mc²51 entered the anaerobic state faster in the later stages, and the number of viable bacteria in the anaerobic state was significantly higher than mc²155 (Figures 2B,C).

For the activation experiment, bacteria cultured in the anaerobic conditions were collected and diluted to 1 × 10⁶ cells/ml in 7H9 and the brain–heart infusion medium and were then grown under aerobic conditions. The three independent mc²51 clones were set, and each clone set up three replicates, all of which were grown better than the three independent clones of mc²155 (Figure 2D). Thus, mc²51 had the growth advantage of recovering from the dormancy of the anaerobic state over mc²155. We also measured the intracellular ATP levels of mc²51 and mc²155. As expected, the abundance of ATP in mc²51 (0.1885 ± 0.0481 μM/mg) was lower than that of mc²155 (0.8138 ± 0.1324 μM/mg) (Figure 3B).

We previously performed whole-genome sequencing to compare differences in mc²155 and mc²51 at the genome level. Whole-genome sequencing revealed that there were 29 single-nucleotide polymorphisms (SNPs) in mc²51. All 29 SNPs were cloned, and each was transformed into mc²155 strains, and only the fur (msmeg_3460), located upstream of msmeg_3461, encoding KatG (Figure 4A), could restore the resistance phenotype of H₂O₂ (Li et al., 2014b). The fur-encoded protein Fur negatively regulated katG expression (Pym et al., 2001). The A28V Fur mutation (mFur) in mc²51 may also affect the expression of katG. To verify this hypothesis, we examined the binding of mFur to the target DNA (the promoter region of the fur) using electrophoretic mobility shift assays (EMSAs). Compared to wild-type Fur protein, EMSA showed that mFur decreased DNA binding (Figure 4B), which resulted in the katG transcription dysregulation by mFur. In addition, the RNA of mc²155 and mc²51 was extracted and quantified. The results showed that compared to the mc²155 strain, the expression of the catalase–peroxidase (KatG) encoding gene katG of the mc²51 strain was significantly upregulated to ∼61.82-fold that of the wild-type strain. Taken together, mFur increases the KatG protein level in mc²51. KatG is a dual enzyme for catalase and peroxidase, which hydrolyzes ROS (Ng et al., 2004). Thus, the mc²51 may maintain ATP at lower levels through KatG, compared to mc²155 levels; that is, the abundance of KatG may affect the mycobacterial redox state and, thus, change the susceptibility to H₂O₂. We then constructed the ΔkatG (mc²155 with knockout katG) and pMV261-katG/mc²155 (mc²155 with overexpression of katG) strains, and their respective ATP content was tested. As shown in Figure 3A, the KatG level negatively correlated with the ATP level. Furthermore, we constructed the specific site mutant of the fur gene (mfur) in wild-type mc²155 causing an amino acid change of A28V of Fur (Figures 5A,B) by using recombination protein gp61 from Che9c mycobacteriophage (van Kessel et al., 2008), to construct mc²155-mfur (Table 1). As we expected, the Fur mutation at A28V induced high resistance to H₂O₂ with the MIC of H₂O₂ in mc²155-mfur being 0.64 mM. We showed that the point mutation of fur dysregulation of katG expression is a major factor leading to the phenotype H₂O₂ resistance.

As a successful human pathogen, M. tuberculosis has unique respiration properties. M. tuberculosis excretes alkaline supernatants, which is in contrast to other strains that excrete acidic supernatants (Merrill, 1930). The difference between secreted compounds with different acid–base properties suggested that M. tuberculosis has a distinctive redox state. As shown in Table 2, the comparative genomic analysis shows that PhoPR and DosR/S/T, identified as virulence factors of M. tuberculosis, are present in M. smegmatis. The signaling transduction systems such as the two-component systems and the sigma factors of M. tuberculosis are homologous in M. smegmatis (Table 3). Different phenotypes might be due to different redox states. Thus, we considered that the same redox-regulated related protein might perform different functions in different redox states. To test this hypothesis, we examined the biological function of a universal stress protein Rv1996 that increases the expression of KatG, in various mycobacterial strains (Hu et al., 2015). Both previous studies and our studies have linked isoniazid action with redox states, and H₂O₂ resistance is negatively correlated with INH susceptibility in mycobacteria (Bhaskar et al., 2014; Hu et al., 2015; Vilcheze et al., 2017). We used INH as a chemical probe for monitoring mycobacterial redox states and measured the MICs of INH to the corresponding mycobacterial strains (Figure 6). As predicted, the MICs of INH differed across the tested mycobacterial strains: the MIC of INH in pMV261-rv1996/mc²7000 was equal to that of pMV261/mc²7000; the MIC of INH in pMV261-rv1996/BCG was lower than that in pMV261/BCG; the MIC of INH in pMV261-rv1996/mc²51 was equal to that of pMV261/mc²51; and in mc²155, the opposite results were observed with the MIC of INH in pMV261-rv1996/mc²155 being lower than that of INH in pMV261/mc²155 (Figure 6).

The H₂O₂-resistant Mycobacterium smegmatis strain mc²51 was screened in the Mi lab (Li et al., 2014b). Mycobacterium tuberculosis ΔpanCD (named mc²7000) (Sambandamurthy et al., 2002) was kindly gifted by J Deng. The M. smegmatis wild-type mc²155, mutant strain mc²51, M. bovis BCG Pasteur, and M. tuberculosis mc²7000 were cultured in Middlebrook 7H9 (Becton Diskinson Sparks, MD, United States) supplemented with ADS (10% albumin, dextrose, and saline), 0.05% Tween 80 (Sigma, St. Louis, MO, United States), and 0.5% glycerol (Beijing Modern Eastern Fine Chemical Co., Ltd., Beijing, China) for liquid culture and Middlebrook 7H10 (Becton Diskinson Sparks, MD, United States) supplemented with ADS for bacterial colony culture. The colony-forming units (CFUs) of mycobacterial strains were determined by plating serial dilutions of cultures on Middlebrook 7H10 agar plates and incubating at 37°C in an atmosphere of 5% CO₂ for the indicated time. For mc²7000 culture, panthothenate (24 mg/L) was added. When required, kanamycin (25 mg/L, Amresco, United States) and hygromycin (50 mg/L, Sigma, United States) were added. All bacterial strains used in this study are listed in Table 1.

The susceptibility of isoniazid (INH) or H₂O₂ of mycobacteria was evaluated using the modified broth microdilution method (Franzblau et al., 1998). In brief, INH or H₂O₂ was serial diluted using the 7H9 medium. The diluted fold was 1.25- or 2-fold, when required. Then, 40 μl of diluted INH or H₂O₂ was mixed with 40 μl of mycobacterial suspension with 1 × 10⁷ cells/ml in each well of 96-well microtiter plates and then incubated at 37°C for the indicated days. As an indicator, 0.02% resazurin was added to individual samples, and the color switches from blue to pink were recorded after 4 h. All the experiments were performed in triplicate. The abundance of the cultures was measured using a microplate reader (FLUOstar OPTIMA, BMG Labtech). A difference of two serial dilutions or more indicated a significant difference in the INH or H₂O₂ susceptibility of bacterial strains.

Female pathogen-free C57BL/6 mice (aged 6–8 weeks) were purchased from Vital River (Beijing, China). For the intranasal infection of M. smegmatis, mice were anesthetized by intraperitoneal injection of pentobarbital sodium (60 mg/kg), and ∼10⁷ CFU/50 µl PBS of mc²51 or mc²155 was introduced dropwise through the nostril of each mouse. The bacterial burden throughout the infection was monitored by collecting whole lung tissue at the indicated times after the mice were euthanized, and serial dilutions were then plated on 7H10 agar plates. The dose of infection was confirmed on day one after infection by plating whole lung homogenates from three mice on 7H10 agar. The percentage of survival was calculated as (CFUs after infection/CFUs before infection) × 100%.

Human-derived cell line THP-1 (ATCC TIB-202) was cultured in RPMI medium with 10% fetal bovine serum (FBS, GIBCO, United States). THP-1 cells were activated with 100 ng/ml phorbo-12-myristate-13-acetate (PMA, Sigma, United States) overnight. Infection was carried out at a multiplicity of infection (MOI) of 10 for 1 h at 37°C and 5% CO₂ atmosphere. The infected THP-1 cells were washed with RPMI 3 times and then chased for 1 h. The cells with intracellular bacilli were then washed and lysed in sterile cold PBST (PBS with 0.05% Tween 20). Lysates were then vortexed, diluted, and plated on 7H10 agar plates as previously described (Chan et al., 1992). The percentage of survival was calculated as (CFUs after infection/CFUs before infection) × 100%.

Mycobacterial strains were cultured under hypoxic conditions as described by Wayne and Hayes (Wayne and Hayes, 1996). In brief, cultures were initiated at an OD₆₀₀ of ∼0.01 (1 × 10⁶) and incubated in anaerobic tubes with sealed caps. The headspace ratio of the cultures was 0.5. The cultures were stirred using an 8 mm Teflon stir bar (Fisher Scientific, United States) at 200 rpm. Methylene blue (1.5 mg/L) was used as an oxygen tension indicator. It changes in color from blue to colorless under low oxygen tension. The color transition time was recorded. All experiments were performed in triplicate. Growth was monitored by measuring the OD₆₀₀ and by determination of CFUs after plating on 7H10.

The indicator methylene blue in the culture became colorless, indicating that the bacteria entered anaerobic conditions. Then, after collecting the bacteria in the anaerobic tube, they were washed with the culture medium or PBS three times and resuspended in a culture medium (7H9 and brain–heart infusion medium), the concentration was adjusted to the same amount of OD₆₀₀ (OD₆₀₀ of ∼1.0) and diluted for the CFU count, 1:100 or 1:50 into a fresh culture medium was shaken at 37°C, to monitor the status of the bacteria, and the OD₆₀₀ was measured. Three independent mycobacterial strain clones were set, and each clone set up three replicates.

The ATP Assay Kit was purchased from Beyotime Biotechnology (Beijing, China). The intracellular ATP assay was performed following the protocol provided by the manufacturer. In brief, the sample measurements were prepared as follows: cultures of indicated mycobacteria were obtained to an OD₆₀₀ of 0.8. Bacteria were collected by low-temperature centrifugation at the maximum speed, and the pellet was washed with precooled PBS buffer 3 times. A 300 μl volume of ATP detection lysate and 0.5 ml volume of glass beads were added for cell lysis. The lysate obtained was centrifuged at low temperature for 5 min, and the supernatant was placed on ice for later use. The preparation of the standard solution of gradient concentration ATP was performed as follows: the ATP standard solution was serially diluted into 7 concentrations of 10, 3.333, 1.111, 0.37, 0.1234, 0.04115, and 0.01371 μM and stored on ice for later use. The preparation of the working solution for the detection of ATP was performed as follows: an appropriate amount of ATP detection reagent was prepared according to the number of samples, and then, a 90% final volume of ATP detection reagent diluent was added. The prepared working fluid was placed on ice for further use. The determination of the ATP level was performed as follows: 1) the prepared ATP detection working solution was dispensed into 1.5 ml centrifuge tubes, 100 μl per tube, and was allowed to incubate at room temperature for 5 min to allow full reaction of the ATP in the centrifuge tube; 2) during the test, 20 μl of each sample (standard or total protein sample) was added to a 1.5 ml centrifuge tube containing 100 μl of ATP detection working solution and was mixed quickly with a pipette and incubated for 2 s to complete the reaction before using a Cytation 3 Cell Imaging Multi-Mode Reader to determine the relative light unit (RLU); 3) a standard curve was constructed to measure and determine the concentration of the sample by converting the RLU into an ATP concentration; and 4) in order to eliminate the error caused by the difference in the amount of protein during sample preparation, the BCA protein concentration determination kit produced by Beyotime Biotechnology (Beijing, China) was used to determine the protein concentration in the sample. The ATP concentration was converted to μM/mg protein.

The coding regions of fur and mfur were amplified from mc²155 and mc²51 genomic DNA and cloned into the Escherichia coli expression vector pET23b (+) (Novagen, Madison, WI, United States) in-frame fused with a C-terminal His₆-tag sequence to construct the plasmids pET23b-fur and pET23b-mfur. The final constructs were transformed into BL21 (DE3) for expression, and recombinant Fur/mFur proteins were purified using Ni-NTA agarose (Qiagen, California, United States). The proteins were induced by the addition of 1 mM IPTG at 16°C for 12 h. Protein purification was performed as described previously (Li et al., 2014c). The protocols of the recombinant protein purification are available on request. The recombinant protein MSMEG_2415 was purified as described previously (Li et al., 2014c) and used as a negative control for EMSA, while MSMEG_2415 is irrelated to Fur. The DNA fragment containing the promoter region of fur for gel shift experiments was amplified by PCR with specific primers (forward: 5′-CGT​TGG​AAA​ACA​ACC​ATT​GCA​AG-3′, reverse: 5′-CAT​CCG​CAG​TTG​GGC​TTC​GAA​C-3′). Binding reaction mixtures in 20 μl of binding buffer (20 mM Tris HCl pH 8.0, 1 mM dithiothreitol (DTT), 50 mM KCl, and 5 mM MgCl₂) containing 0.15 pmol of the DNA fragment were incubated with purified Fur/mFur protein (0.5 nmol) for 30 min at 30°C. Reaction mixtures were loaded on a 4–20% polyacrylamide gel containing 0.5 × TBE. Gels were run at 70 V at 4°C for 3 h. The gel was stained with Good-view and photographed for the image.

The knockout katG strain was constructed using mycobacteriophage-based specialized transduction (Bardarov et al., 2002; Li et al., 2014c). The upstream and downstream sequences of katG were amplified from M. smegmatis genome DNA. The knockout vector was constructed using phAE159 (Hsu and Jacobs, unpublished data). The mycobacteriophage used for knockout was obtained using MaxPlax packaging extract (Epicentre Biotechnologies, Madison, WI, United States), and a katG knockout strain was obtained by phage transduction, named ∆katG. The KatG overexpression strain was constructed using pMV261 to yield pMV261-katG, and the constructed plasmid was electroporated into mc²155, yielding pMV261-katG/mc²155. The detailed information for construction of all the mycobacterial strains and primers for plasmid construction is available on request.

The single-strand (ss) DNA oligonucleotides used for recombineering were ordered from Genewiz (Suzhou, China). The site-directed mutagenesis of fur was obtained using Phage Che9c gp61-mediated recombination (van Kessel et al., 2008). The detailed information on primers for construction of the fur mycobacterial strain (named mc²155-mfur) is available on request. The coding region containing fur point mutation in the genome (encoding mFur) was amplified and sequenced by Genewiz (Suzhou, China).

The rv1996 gene was amplified and constructed and cloned into pMV261 to yield pMV261-rv1996. The constructed pMV261-rv1996 plasmid was transformed into mycobacterial strains, M. smegmatis mc²155 and mc²51, M. bovis BCG Pasteur, and M. tuberculosis mc²7000, and the corresponding strains, named pMV261-rv1996/mc²155, pMV261-rv1996/mc²51, pMV261-rv1996/BCG, and pMV261-rv1996/mc²7000. The empty vector pMV261 was transformed into the corresponding mycobacterial strains, named pMV261/mc²155, pMV261/mc²51, pMV261/BCG, and pMV261/mc²7000.

Each experiment was carried out at least twice with three–nine mice or samples per group. The CFUs and OD₆₀₀ were analyzed using an unpaired t-test (Version 8.0 for Windows GraphPad Software). The ATP content was analyzed using ANOVA tests (Version 8.0 for Windows GraphPad Software). ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.

This study was performed in strict accordance with the recommendations of the Ethics Committee established in the Guide for the Care and Use of Laboratory Animals of the Institute of Microbiology, Chinese Academy of Sciences (IMCAS). The protocol was approved by the Committee on the Ethics of Animal Experiments of the IMCAS. The mice were bred under specific pathogen-free conditions at the IMCAS laboratory animal facility. All animal experiments were conducted under isoflurane anesthesia, and all efforts were made to minimize suffering.