An estimated 15-20% of patients who are treated for pulmonary tuberculosis (TB) are culture-negative at the time of diagnosis. Recent work has focused on the existence of differentially detectable Mycobacterium tuberculosis (Mtb) bacilli that do not grow under routine solid culture conditions without the addition of supplementary stimuli. We identified a cohort of TB patients in Lima, Peru, in whom acid-fast bacilli could be detected by sputum smear microscopy, but from whom Mtb could not be grown in standard solid culture media. When we attempted to re-grow Mtb from the frozen sputum samples of these patients, we found that 10 out of 15 could be grown in a glycerol-poor/lipid-rich medium. These fell into the following two groups: a subset that could be regrown in glycerol after "lipid-resuscitation", and a group that displayed a heritable glycerol-sensitive phenotype that were unable to grow in the presence of this carbon source. Notably, all of the glycerol-sensitive strains were found to be multidrug resistant. Although whole-genome sequencing of the lipid-resuscitated strains identified 20 unique mutations compared to closely related strains, no single genetic lesion could be associated with this phenotype. In summary, we found that lipid-based media effectively fostered the growth of Mtb from a series of sputum smear-positive samples that were not culturable in glycerol-based Lowenstein-Jensen or 7H9 media, which is consistent with Mtb's known preference for non-glycolytic sources during infection. Analysis of the recovered strains demonstrated that both genetic and non-genetic mechanisms contribute to the observed differential capturability, and suggested that this phenotype may be associated with drug resistance.

Drug resistance diagnostics that rely on the detection of resistance-related mutations could expedite patient care and TB eradication. We perform minimum inhibitory concentration testing for 12 anti-TB drugs together with Illumina whole-genome sequencing on 1452 clinical Mycobacterium tuberculosis (MTB) isolates. We evaluate genome-wide associations between mutations in MTB genes or non-coding regions and resistance, followed by validation in an independent data set of 792 patient isolates. We confirm associations at 13 non-canonical loci, with two involving non-coding regions. Promoter mutations are measured to have smaller average effects on resistance than gene body mutations. We estimate the heritability of the resistance phenotype to 11 anti-TB drugs and identify a lower than expected contribution from known resistance genes. This study highlights the complexity of the genomic mechanisms associated with the MTB resistance phenotype, including the relatively large number of potentially causal loci, and emphasizes the contribution of the non-coding portion of the genome.