Tuberculosis is caused by Mycobacterium tuberculosis (Mtb) and claims 1.8 million lives annually. Clinical strains of Mtb reveal diverse phenotypes that are largely determined by the state of the proteome. Therefore, the model strain H37Rv that has been used for most studies in the field does not represent the full genomic and phenotypic diversity of Mtb. Here, we aim to identify master regulators that control the Mtb proteome of genetically diverse strains driving different phenotypes in two conditions.
SWATH-MS profiled the proteome of 70 clinical strains of Mtb cultivated under normal and nitric oxide stress, a major bactericidal agent within macrophages. Moreover, we fully sequenced the genome of the respective strains. Six strains were subjected to transcriptional measurements.
We quantified ~2700 proteins corresponding to ~80% of the expressed genes, reproducibly across the large Mtb sample cohort. To address the aim of the project, we developed two exploratory computational frameworks, a genome-scale transcriptional model and dysregulation analysis for protein complexes, and analyzed the generated dataset using those two pipelines as well as QTL mapping. The data indicated that the basal expression and response of various protein functional groups such as IdeR and DosR regulon significantly differ between Mtb lineages. Twenty-eight transcription factors orchestrating the Mtb transcriptional network between lineages have been identified. Several subnetworks are regulated differently in terms of their stoichiometry across various lineages including the protein association Rv0068-Rv2187. QTL analysis, implemented for the first time in bacteria, revealed that a single mutation in KstR affects the expression of its seven targets involved in cholesterol metabolism, the main carbon source during Mtb infection. It further provided the first evidence of the existence of protein isoforms in Mtb. We have shown that genomic differences between the clinical isolates determine the state of the proteome and mediate different clinically relevant phenotypes.