An Integrated Platform for Finding and Developing Novel Antibiotics

Discovery of new drugs for Gram-negative pathogens is uniquely challenging, as target engagement, pharmacokinetics (PK), toxicity, and cell permeability all require optimization; in practice this has proved nearly impossible through the traditional approach. In addition, there is virtually no data on chemical traits that dictate the subcellular accumulation of compounds, complicating the design of antibiotics that engage targets at specific subcellular sites; as one key example, if an antibiotic could be designed to selectively localize to the periplasm, there are numerous periplasmic targets that could be exploited to develop novel and gut microbiome sparing antibiotics. Here we propose the development and application of a new molecular imaging technology to track the subcellular localization of compounds in Gram-negative bacterial cells. Using this method we will elucidate chemical traits that facilitate various subcellular localizations, with an emphasis on the periplasm. This data will be combined with existing information on target engagement, PK, toxicity, and influx/efflux to build the DiscoveryBox, a neural network that simultaneously considers these key parameters to suggest optimized candidate antibiotics from unoptimized starting points. The ready availability of DiscoveryBox will facilitate development of novel antibiotics for Gram- negative pathogens by us and by users around the world.