| The emergence and transmission of drug resistant strains of Mycobacterium tuberculosis (Mtb) calls for urgency in understanding the barriers to effective treatment. Here we developed a chemostat cultivation system to measure the mycobacterial mutation rate during periods of slow growth versus fast growth. We found that mutants accumulated at the same rate per unit time during slow growth and during fast growth, suggesting that mutation rate does not vary in accordance with growth rate. By competing genetically barcoded mutant strains at fast and slow growth rates and using a mathematical model to estimate mutant fitness, we confirmed that there is not aggravated loss of mutants at fast growth rates, but that equivalence in mutation rate per unit time reflects time-based mutation rather than replication-based mutation. To investigate a possible driver of replication-independent mutation, we developed a mass-spectrometry-based analytic method to detect DNA damage during metabolism of fatty acids, a major component of the bacterial diet in vivo. We discovered that a novel glyoxylate-dG adduct forms when cells use the glyoxylate shunt to metabolize fatty acids, and that the levels of this adduct increase in the absence of nucleotide excision repair. Using fluctuation analysis to measure the mutation rate in these conditions, we found that fatty acid metabolism is mutagenic in nucleotide repair deficient cells. Finally, cholesterol metabolism was mutagenic in both wild-type and repair-deficient backgrounds. These findings demonstrate that metabolic state can drive both mutation and DNA damage. Many patients develop recurrent tuberculosis (TB) despite receiving adequate treatment for TB, yet the causes of recurrent disease are poorly understood. We used whole-genome sequencing (WGS) and MIRU-VNTR typing to investigate the cause of recurrent infection in 13 HIV-infected individuals who had been successfully treated as part of their enrollment in a prospective cohort study in KwaZulu Natal, South Africa. By comparing the genetic relatedness in the presenting and recurrent strains, we found that 7 of the 13 recurrent episodes appeared to represent relapse of the primary infection, despite the force of TB transmission in KwaZulu Natal and the successful HIV and TB treatment history. Results from MIRU-VNTR typing were concordant with WGS for highly related and highly divergent strains, but only WGS analysis could resolve intermediate genetic distances. Patients presented with relapsed infection and reinfection up to 3 years after completion of treatment, and no difference was found between the timing of relapse and reinfection cases. Strains causing relapsed infection were more likely to harbor genetic polymorphisms associated with changes in INH susceptibility or acquisition of INH resistance despite being phenotypically drug sensitive, suggesting a possible role for low-level or undiagnosed drug resistance in tuberculosis relapse. |
