Supply_2013_Nat.Genet_45_172

Global spread and limited genetic variation are hallmarks of M. tuberculosis, the agent of human tuberculosis. In contrast, Mycobacterium canettii and related tubercle bacilli that also cause human tuberculosis and exhibit unusual smooth colony morphology are restricted to East Africa. Here, we sequenced and analyzed the whole genomes of five representative strains of smooth tubercle bacilli (STB) using Sanger (4-5x coverage), 454/Roche (13-18x coverage) and/or Illumina DNA sequencing (45-105x coverage). We show that STB isolates are highly recombinogenic and evolutionarily early branching, with larger genome sizes, higher rates of genetic variation, fewer molecular scars and distinct CRISPR-Cas systems relative to M. tuberculosis. Despite the differences, all tuberculosis-causing mycobacteria share a highly conserved core genome. Mouse infection experiments showed that STB strains are less persistent and virulent than M. tuberculosis. We conclude that M. tuberculosis emerged from an ancestral STB-like pool of mycobacteria by gain of persistence and virulence mechanisms, and we provide insights into the molecular events involved.