The landscape and molecular underpinnings of mycobacterial drug interactions

Combination therapies are frequently employed against difficult-to-treat bacterial pathogens, but their
clinical application and trial results often fail to achieve the desired results. Mycobacterium abscessus,
a highly resistant pathogen responsible for an increasing number of respiratory infections worldwide,
highlights this challenge, as extensive and prolonged combination treatments typically fail. In this
project, I systematically assessed drug combinations in M. abscessus to uncover underlying
mechanisms and general principles of drug interactions.
I established a platform to measure more than 1.4 million combination assessments. I evaluated 48
drugs with activity against M. abscessus individually, in every possible pairwise combination in 9 x 9
checkerboards, and in combination with 2’514 inactive compounds. I tested 28 commonly used
combinations against respiratory clinical isolates from 269 patients with M. abscessus lung infections
in diagonal 7 x 7 checkerboards, and mapped interaction phenotypes to whole-genome sequencing
data of each isolate.
Drug interactions were common when combining drugs with anti-M. abscessus activity (> 60%),
including many of the clinically-used antibiotic combinations, and antagonistic drug interactions
prevailed. Drug interaction patterns were commonly conserved between antibiotic classes. Moreover,
inactive compounds frequently influenced the activity of active drugs. Organizing these 2’514 inactive
compounds in chemical space revealed clusters of chemically similar and functionally related groups,
enhancing or decreasing antibiotic potency. Testing clinically used antibiotic combinations on diverse
M. abscessus respiratory clinical isolates revealed heterogeneous interactions. These drug
interactions commonly converged with phylogenetic clades, and many combinations showed a high
degree of heritability, suggesting a genetic basis for these interactions. Genetic variants potentially
driving drug interactions were identified by applying genome-wide association studies and studying
knockout mutants.
This analysis of over 120’000 unique drug combinations in M. abscessus reveals the complex
landscape of drug interactions. I identified chemical and genetic determinants that drive drug
interactions and proposed strategies and targets to increase our understanding of drug interaction
mechanisms and enhance treatment efficacy against multi-resistant bacterial pathogens.