Pervasive genome structure heterogeneity in Mycobacterium tuberculosis constitutively generates subpopulations with...

Bacterial infections are difficult to treat in part due to population heterogeneity, which is often masked by consensus genome representations. Using long-read sequencing of 138 Mycobacterium tuberculosis (M . tuberculosis) clinical isolates, we developed a framework to detect subpopulations with structural variants (SVs). We find that M . tuberculosis constitutively generates subpopulations bearing distinct clinical phenotypes. We identified 18 recurrent heterogeneous SVs affecting regulatory and coding regions of clinically relevant genes, including diagnostic markers (PPE59) and virulence mediators (pknH, pks12, and ppe38/ppe71). We find recombination between 13E12E repeat proteins drives frequent genome structure change, redistributing mutagenicity across the chromosome through large, heterogeneous, asymmetric inversions. Remarkably, most (10/18) SVs mirror previously evolved events, indicating fractal-like genome structural dynamics. The repeated emergence of subpopulations with these SVs reveals previously unappreciated structural plasticity in the M . tuberculosis genome. This plasticity may enable M. tuberculosis to readily adapt to diverse host microenvironments through rapid niche expansion.