Chemical-Genetic Inference of Antibiotic Interactions for Combination Therapies

Combination antibiotic therapies are being increasingly used in response to a global rise in drug resistance. Mapping synergistic and antagonistic interactions between antibiotics is essential for designing effective combinations that enhance potency and counter resistance. However, the large search-space of candidate drugs and dosage regimes makes such approaches highly challenging. Here, we present a computational approach that integrates publicly available chemogenomic data to predict dose-specific antibiotic interactions. We experimentally tested interactions for 171 antibiotic pairs in nine dose combinations in Escherichia coli to validate our approach. Our analysis revealed a core set of cellular pathways (e.g., central metabolism) and biophysical factors (e.g., drug lipophilicity) that are predictive of synergy and antagonism. Using an evolutionary approach, we modified our chemical-genetic model of E. coli, a Gram-negative bacterium, to accurately predict drug-drug interactions in the Gram-positive pathogen Staphylococcus aureus. This study provides a framework for chemogenomics-driven discovery and development of effective combination therapies.