Starvation induces antibiotic tolerance in diverse mycobacteria.

(A) Msmeg, (B) Mtb or (C) Mabs were grown in 7H9 rich media or starved in PBS prior to the addition of antibiotics and surviving colony forming units (CFU) enumerated. For the rapidly growing mycobacteria, Msmeg and Mabs, cells were allowed to adapt for 48h prior to antibiotics, for slow-growing Mtb, cells were allowed to adapt for 14-21d prior to antibiotics. Samples without pre-adaptation were washed and placed directly into PBS with antibiotics. Antibiotic concentrations were: Msmeg - Isoniazid (INH) 32μg/ml (8 x MIC), rifampin (RIF) 32μg/ml (8 x MIC), ethambutol (EMB) 4μg/ml (8 x MIC), tigecycline (TIG) 1.25μg/ml (8 x MIC), linezolid (LZD) 2.5μg/ml (8 x MIC). Mtb – RIF 0.1μg/ml (4 x MIC), INH at 0.1μg/ml (4 x MIC), EMB at 8μg/ml (4 x MIC). Mabs – TIG 10μg/ml (8 x MIC), LZD 100μg/ml (20 x MIC). Antibiotics with half-lives shorter than the duration of experiment were re-added at the following intervals: TIG, EMB every 3d; RIF, INH every 6d. Error bars represent SEM, statistical significance is calculated at each time point using student’s t test. ****: p<0.0001, ***: p<0.001, **: p<0.01, *: p<0.05, ns: p>0.05. Data are combined from 3 independent experiments.

Tn-Seq identifies genes required for antibiotic tolerance in Mabs.

(A) Experimental design. (B-E) Tn-Seq analysis showing relative abundance of individual genes under the indicted conditions. Genes depleted relative to the input population have negative values. All cultures were fully aerated throughout the experiment and cultures without antibiotics received and equal volume of DMSO. For (B-D) gene abundance in the indicated condition is measured relative the input log-phase population. In (E) an additional comparison is made for PBS with antibiotics relative to the PBS condition. Genes with significant decreases in abundance are shown in color (p-adj. < 0.05 and log2 fold-change > 0.5) using the Benjamini–Hochberg adjustment for multiple hypothesis testing. (F) Number of genes essential in each condition relative to the input population. (G) Pathway enrichment analysis of the essential genes in each condition using the DAVID knowledgebase (p <0.05). Screens were run as 3 independent experiments and the combined results analyzed. Antibiotic conditions were as described above.

Validation of Tn-Seq results.

(A-E) Homologous recombination was used to delete the indicated genes, or to generate a control strain targeting a distant intergenic region distal to the non-essential tRNA gene MAB_t5030c. Each strain was either grown in 7H9 rich media or starved in PBS for 48 prior to the addition of antibiotics as indicated. The conditions tested here correspond to the conditions in the Tn-Seq analysis where a phenotype was observed. Data from additional conditions are in Figure 3 Supplemental. Comparisons in panels A-C are made to the same control strain but plotted independently for clarity. Error bars represent SEM, statistical significance is calculated at each time point using student’s t test. ****: p<0.0001, ***: p<0.001, **: p<0.01, *: p<0.05, ns: p>0.05. Antibiotics were added as described above. Data are representative of 4 independent experiments.

Complementation analysis of katG and pafA mutants confirms their role in antibiotic tolerance.

(A) RT-qPCR analysis of katG expression in katG- (ΔkatG::pmv306), katG+ (ΔkatG::pmv306 katG), and control strain (ORBIT intergenic::pmv306). (B) CFU over time for katG+/katG- strains. (C) MICs for katG+/katG- strains. (D) Expression of pafA in pafA- (ΔpafA::pmv306), pafA+ (pafA::pmv306 pafA) and control strain. (E) CFU over time for pafA+/pafA- strains. (F) MICs for pafA+/pafA- strains. Antibiotic concentrations in (A-B, D-E) are as described above. Error bars represent SEM, statistical significance is calculated at each time point using student’s t test between katG+/katG- strains in (B) and between pafA+/pafA- strains in (E). ****: p<0.0001, ***: p<0.001, **: p<0.01, *: p<0.05, ns: p>0.05. Antibiotics were added as described above.

ROS-mediated toxicity following antibiotic exposure.

(A-D) Analysis of katG+/katG- cells challenged with different antibiotics. Cells were starved in PBS for 48h and then exposed to the indicated antibiotic. (E) Flow cytometry of control cells exposed to TIG/LZD (4h in 7H9 media or 72h in PBS) and then stained with DAPI and the ROS-sensitive dye cellROX green; percentage cellROX-positive cells are indicated. (F) Survival over time for aerated and hypoxic cultures of Mabs after exposure to TIG/LZD. (G) Survival over time for bipyridyl treated cells after exposure to TIG/LZD. Error bars represent SEM, statistical significance is calculated at each time point using student’s t test. ****: p<0.0001, ***: p<0.001, **: p<0.01, *: p<0.05, ns: p>0.05. (A-D, F) display combined data from 3 independent experiments. (E, G) are representative data from 3 independent experiments.

Consistent antibiotic-induced ROS production but variable protection by KatG among different Mabs strains.

(A) The indicated strains of Mabs were cultured in 7H9 media, exposed to TIG/LZD for 4h and then analyzed by cellROX staining. (B) CFU over time following TIG/LZD exposure. Error bars represent SEM, statistical significance is calculated at each time point using student’s t test. ****: p<0.0001, ***: p<0.001, **: p<0.01, *: p<0.05, ns: p>0.05. Combined data from 4 independent experiments are shown for persister survival experiments. Representative data from 2 independent experiments are shown for flow cytometry experiments. Antibiotics were added as described above.

Additional analysis of mutants.

(A) blaR or (B) recR mutants were examined under conditions where Tn-Seq did not predict a phenotype. Cells were starved in PBS for 48h prior to treatment with TIG/LZD or DMSO as described in Figure 3. (C) Growth recovery of mutants after antibiotic exposure. After 6 d of TIG/LZD exposure cells were washed twice in antibiotic free media and inoculated into 7H9 media. Growth was monitored by OD 600 of the culture. Error bars represent SEM, statistical significance is calculated at each time point using student’s t test. ****: p<0.0001, ***: p<0.001, **: p<0.01, *: p<0.05, ns: p>0.05. (A-B) are representative data from 4 independent experiments (C) is representative data from 2 independent experiments.

Effect of ROS scavengers.

ΔkatG cells were starved in PBS for 48h then treated with TIG/LZD and the indicated concentration of (A) thiourea or (B) TEMPO and surviving CFU over time were measured. Representative data from 2 independent experiments are shown.