High-throughput profiling of drug interactions in Gram-positive bacteria

Antimicrobial resistance is an increasingly serious threat that influences all domains of modern medicine and public health. Many strategies are currently implemented to counteract its spread and tackle resistant strains. One of the most promising approaches is the use of drug combinations of approved compounds, bypassing the hurdles of new compound development.We tested more than 2000 drug combinations in different bacteria, including the species Staphylococcus aureus and Streptococcus pneumoniae, which are particularly relevant in the current antimicrobial resistance scenario. By probing several classes of antibiotics targeting different cellular processes, we showed how drug-drug interactions reflect the interplay between their targets, mirroring gene-gene interactions and capturing information on the mode of action of single drugs. Accordingly, drug interactions are subject to the same evolutionary constraints as cellular networks, with the conservation of drug interactions closely reflecting species phylogeny.We also investigated the impact of non-antibiotic drugs, which are commonly co-administered with antibiotics upon bacterial infections, testing these drugs in more than 3000 combinations with antibiotics in S. aureus. We highlighted both promising synergies, which were validated against multi-drug resistant S. aureus clinical isolates, and widespread antagonisms, which may hamper the efficacy of antimicrobial treatments.Finally, we integrated drug-drug and drug-gene interaction data, generated by testing single drugs on single-gene deletion mutant libraries. We were able to predict drug-drug interactions from drug-gene interactions using machine-learning approaches and uncover genes that are statistically associated with drug interactions, overall shedding light on the genetic background of drug interactions