Clinically prevalent mutations in Mycobacterium tuberculosis alter propionate metabolism and mediate multidrug tolerance

Abstract

The global epidemic of drug-resistant tuberculosis is a catastrophic example of how antimicrobial resistance is undermining the public health gains made possible by combination drug therapy. Recent evidence points to unappreciated bacterial factors that accelerate the emergence of drug resistance. In a genome-wide association study of Mycobacterium tuberculosis isolates from China, we find mutations in the gene encoding the transcription factor prpR enriched in drug-resistant strains. prpR mutations confer conditional drug tolerance to three of the most effective classes of antibiotics by altering propionyl-CoA metabolism. prpR-mediated drug tolerance is carbon-source dependent, and while readily detectable during infection of human macrophages, is not captured by standard susceptibility testing. These data define a previously unrecognized and clinically prevalent class of M. tuberculosis variants that undermine antibiotic efficacy and drive drug resistance.

Figure 1.

Genetic associations with isoniazid resistance (A) Geographic distribution of sequenced TB strains. (B) Maximum-likelihood tree of 679 M. tuberculosis isolates based on 59,367 genome-wide SNP sites. A strain of M. canetti was used as the outgroup. Known lineages are marked outside of the tree. Sub-lineages of China strains within L2 and L4 are highlighted with background shadows (names followed as previously described by Zhang et al.) (C) Genetic associations with isoniazid resistance among 549 strains from China. Each circle represents a gene or intergenic-region in the M. tuberculosis genome. The y-axis represents uncorrected phyOverlap p-values; genes above the gray line are genome-wide significant with q < 0.05 after correction with the Benjamini-Hochberg procedure. Colors denote loci known to contribute to resistance and sizes correspond to the number of times mutations were acquired across the dataset based on parsimony. (D) Comparison of genes identified in our sequence collection (Study A, 549 strains) with those identified using the same method on strain collections from Walker et al. (Study B, 784 strains) and Zhang et al. (Study C, 161 strains). Genes are sorted by significance after performing a meta-analysis of the phyOverlap p-values using Fisher’s method and correction with the Benjamini-Hochberg procedure. Genes above the gray line are genome-wide significant in the meta-analysis with q < 0.05.

Figure 2.

Distribution and functional characterization of prpR mutants. (A) Vertical lines represent prpR mutations identified. Red sites are found exclusively in INH resistant isolates, blue sites are found exclusively in INH sensitive isolates, and purple are found in both. Positions showing convergent evolution are depicted with each instance of mutation to an alternate amino acid. (B) The distribution of prpR mutants in isoniazid sensitive and resistant isolates (C) Schematic depicting the metabolism of propionyl-CoA through the methylcitrate cycle and the B12-dependent methylmalonyl-CoA pathway. (D and E) Bacterial growth as measured by OD₆₀₀ in 7H12 media supplemented with 0.02% acetate (D) or 0.02% propionate (E). (F) Relative fold-change in expression of prpD in indicated media at two days post-exposure. Data plotted is the mean and standard deviation of three biological replicates. ** indicates differences in expression among strains of p < 0.001 by Tukey’s multiple comparisons test after one-way ANOVA.

Figure 3.

PrpR mutants display conditional multidrug tolerance. (a-c) Drug resistance measurement of growth in varying concentrations of antibiotic normalized to a no drug control for in (a) isoniazid (INH), (b) ofloxacin (OFLX) and (c) rifampin (RIF). The experiment was performed three times with similar results. (d) Survival of prpR mutants in a library of strains over time in the indicated antibiotic and media conditions. Fraction survival represents bulk library CFU divided into strain CFU by relative abundance measurement. Each point represents the mean and standard deviation of three independent measurements. (e-g) Bacterial survival after six days of treatment in single strain assays with (e) isoniazid (INH), (f) ofloxacin (OFLX) and (g) rifampin (RIF) in media with acetate (clear background marked with A) or propionate (shaded background marked with P). Each dot represents the fraction of bacteria surviving six days of antibiotic treatment in a biological replicate with the mean and standard deviation indicated with bars. N=3 for each strain in each condition. * p < 0.05 and ** p < 0.01 indicate differences in log-transformed survival determined by Tukey’s multiple comparison test after two-way ANOVA.

Figure 4.

The normalized fitness of indicated strains compared with ΔprpR::WT during antibiotic treatment in human peripheral blood mononuclear cell derived macrophages treated with (a) no antibiotic (ND), (b) isoniazid (INH) at 3 μg/ml, (c)ofloxacin (OFLX) at 2.5 μg/ml, or (d) rifampin (RIF) at 5 μg/ml. The dotted green line indicates ΔprpR::WT which is set to 1 at each timepoint. The mean and standard deviation of three replicates are shown. e) The mean normalized end-point fitness of strains after three days of antibiotic treatment in THP-1 cells (n=3 per treatment). Differences in survival compared with ΔprpR::WT (fitness = 1) were evaluated by one-sample t test, with two-tailed p-values indicated by * p < 0.05 and ** p < 0.01 after FDR correction.

Figure 5.

Metabolic rescue of propionate sensitivity suppresses drug tolerance in prpR mutants. (a-b) Growth of indicated prpR mutants in propionate-containing media with or without vitamin B12. Data represents the mean and standard deviation of three biological replicates. (c-f) Fitness of prpR mutants after antibiotic treatment with and without the supplementation of B12 normalized to input abundance and ΔprpR::WT. (g-h) The normalized fitness of strains in THP-1 cells with or without vitamin B12 supplementation in the absence (G) or presence (H) of isoniazid. The dotted green line indicates ΔprpR::WT which is set to 1 at each timepoint. Data represents the mean and standard deviation of 3 independent replicates. Significant differences in normalized fitness ± B12 supplementation were tested by Sidak’s multiple comparison test after two-way ANOVA. * p < 0.05, ** p < 0.01