Transposon sequencing reveals metabolic pathways essential for Mycobacterium tuberculosis infection

Abstract

New drugs are needed to shorten and simplify treatment of tuberculosis caused by Mycobacterium tuberculosis. Metabolic pathways that M. tuberculosis requires for growth or survival during infection represent potential targets for anti-tubercular drug development. Genes and metabolic pathways essential for M. tuberculosis growth in standard laboratory culture conditions have been defined by genome-wide genetic screens. However, whether M. tuberculosis requires these essential genes during infection has not been comprehensively explored because mutant strains cannot be generated using standard methods. Here we show that M. tuberculosis requires the phenylalanine (Phe) and de novo purine and thiamine biosynthetic pathways for mammalian infection. We used a defined collection of M. tuberculosis transposon (Tn) mutants in essential genes, which we generated using a custom nutrient-rich medium, and transposon sequencing (Tn-seq) to identify multiple central metabolic pathways required for fitness in a mouse infection model. We confirmed by individual retesting and complementation that mutations in pheA (Phe biosynthesis) or purF (purine and thiamine biosynthesis) cause death of M. tuberculosis in the absence of nutrient supplementation in vitro and strong attenuation in infected mice. Our findings show that Tn-seq with defined Tn mutant pools can be used to identify M. tuberculosis genes required during mouse lung infection. Our results also demonstrate that M. tuberculosis requires Phe and purine/thiamine biosynthesis for survival in the host, implicating these metabolic pathways as prime targets for the development of new antibiotics to combat tuberculosis.

Fig 1. M. tuberculosis Tn mutants in genes encoding central metabolic enzymes exhibit growth defects in Middlebrook 7H9 medium.

Wild-type M. tuberculosis Erdman and M-ES Tn mutant strains were grown in MtbYM rich medium, washed twice in PBS-T, and diluted to OD₆₀₀ = 0.01 in Middlebrook 7H9 medium. Growth was monitored by measuring the OD₆₀₀. Tn mutants are grouped according to predicted function: (A) purine and thiamine metabolism (purF, purM, purQ), (B-C) amino acid biosynthesis (pheA, trpB, trpG, proC, gltB, ilvA), (D) pantothenate, riboflavin, and p-amino benzoic acid (PABA) biosynthesis (panC, ribA2, ribG, pabB). Data represent the mean ± standard error of three biological replicates.

Fig 2. Tn-seq screen of the M. tuberculosis M-ES Tn library identifies multiple central metabolic pathways required for fitness in mice.

(A) M-ES Tn-seq screen methods. Individual Tn mutants grown in MtbYM rich medium were mixed in equal abundance to create the M-ES Tn library. Triplicate M-ES Tn library samples were collected for Tn-seq (Library control) and the remainder was aliquoted and frozen for experiments. The M-ES Tn library was grown in MtbYM rich broth, then washed and diluted in PBS-T to OD₆₀₀ = 0.05. Triplicate samples of the diluted M-ES Tn library were collected (Liquid control) and the diluted M-ES Tn library was plated on MtbYM rich agar in triplicate (Plate control) for Tn-seq. Mice were injected via the lateral tail vein with ~10⁵ CFU of the M-ES Tn library to seed at least 10⁴ CFU in the lungs and spleen. Mice (n = 3) were euthanized at days 1, 7, 21, and 42. Lung and spleen homogenates were plated on MtbYM rich agar to recover surviving Tn mutants for Tn-seq. (B) M. tuberculosis CFU recovered from lungs (blue) and spleens (green). (C) Heat map of Log₂ fold change in Tn mutant abundance between the indicated conditions determined by TnseqDiff analysis of Tn-seq data. Positive values (red) indicate a relative fitness advantage; negative values (blue) indicate a relative fitness defect. (D-F) Volcano plots of TnseqDiff analysis of Tn-seq data for the M-ES Tn mutant library for (D) lungs at day 1 compared to the Plate control; (E) lungs at day 21 compared to lungs at day 1; or (F) spleens at day 21 compared to spleens at day 1. Dashed lines indicate cutoffs for statistical significance of ± 2 Log₂ fold change and adjusted P value <0.025. Tn mutants meeting these significance cutoffs are colored and labeled.

Fig 3. Loss of purF causes death of M. tuberculosis in the absence of nutrient supplementation in vitro and in mice.

(A) Pathway for de novo synthesis of purine nucleotides and thiamine in M. tuberculosis. Solid lines indicate a single step; dashed lines indicate multiple steps. Proteins that catalyze the first five steps are indicated. Intermediate abbreviations: PPP, pentose phosphate pathway; PRPP, 5-phosphoribosyl pyrophosphate; PRA, 5-phospho-D-ribosylamine; GAR, 5’-phosphoribosylglycinamide; FGAR, 5’-phosphoribosyl-N-formylglycinamide; FGAM, 5’-phosphoribosyl N-formylglycinamidine; AIR, 5-aminoimidazole ribotide. (B-C) Strains grown in 7H9 supplemented with 60 μM thiamine and 150 μM hypoxanthine were washed in PBS-T and diluted to OD₆₀₀ = 0.01 in 7H9 or 7H9 supplemented with 60 μM thiamine and 150 μM hypoxanthine. Bacterial growth and survival were measured by (B) optical density at 600 nm or (C) serial dilutions and plating to recover viable CFU. (D-E) C57BL/6J mice were infected by aerosol with the indicated strains. Mice (n = 6) were euthanized at days 1, 7, 21, and 42 post-infection. Lung (D) and spleen (E) homogenates were serially diluted and plated to recover viable CFU. In (C-E), WT and purF::Tn pMV-purF were plated on 7H10 agar; purF::Tn was plated on 7H10 agar supplemented with 60 μM thiamine and 150 μM hypoxanthine. Data represent the mean ± standard error of three biological replicates (B-C) or six animals (D-E). Asterisks indicate P-value < 0.05; n.d. indicates not detected (detection limit = 1 CFU).

Fig 4. M. tuberculosis requires Phe synthesis for survival in vitro in the absence of amino acid supplementation and in mice.

(A) Phenylalanine biosynthesis in M. tuberculosis. Single arrows represent one-way reactions. Double arrows represent reversible reactions. Proteins predicted to catalyze each step are indicated. (B-C) Strains grown in 7H9 supplemented with 0.5% casamino acids were washed in PBS-T and diluted to OD₆₀₀ = 0.01 in 7H9 or 7H9 supplemented with 0.5% casamino acids. Bacterial growth and survival were measured by (B) optical density at 600 nm and (C) serial dilutions and plating to recover viable CFU. (D-E) C57BL/6J mice were infected by aerosol with the indicated strains. Mice (n = 6) were euthanized at days 1, 7, 21, and 42 post-infection. Lung (D) and spleen (E) homogenates were serially diluted and plated to recover viable CFU. In (C-E), WT and pheA::Tn pMV-pheA were plated on 7H10 agar; pheA::Tn was plated on 7H10 agar with 0.5% casamino acids. Data represent the mean ± standard error of three biological replicates (B-C) or six animals (D-E). Asterisks indicate P-value < 0.05; n.d. indicates not detected (detection limit = 1 CFU).

Fig 5. M. tuberculosis does not require IlvA in mice.

(A) Isoleucine biosynthesis in M. tuberculosis. Single arrows represent one-way reactions. Double arrows represent reversible reactions. Proteins predicted to catalyze each step are indicated. (B-C) Strains grown in 7H9 supplemented with 0.5% casamino acids were washed in PBS-T and diluted to OD₆₀₀ = 0.01 in 7H9, or 7H9 with 0.5% casamino acids. Bacterial growth and survival were measured by (B) optical density at 600 nm and (C) serial dilutions and plating to recover viable CFU. (D-E) C57BL/6J mice were infected by aerosol with WT Erdman or ilvA::Tn. Mice (n = 6) were euthanized at days 1, 7 and 21 post-infection. Lung (D) and spleen (E) homogenates were serially diluted and plated to recover viable CFU. In (C-E) WT was plated on 7H10 agar; ilvA::Tn was plated on 7H10 agar with 0.5% casamino acids. Data represent the mean ± standard error of three biological replicates (B-C) or six animals (D-E). Asterisks indicate P-value <0.05.