Atlas of efflux and permeation determinants to advance antibacterial therapeutic discovery

Gram-negative bacteria present a critical and serious threat to public health and cause infections that are often untreatable with currently available antibiotics. The discovery and development of new antibiotics effective against these pathogens, however, is hampered by their cell envelopes which are mostly impermeable to small molecule compounds. The high-priority task as identified in the RFA-AI-24-021 is to develop novel tools and strategies that will enable antibiotic penetration across non-specific permeability barriers of Gram-negative bacteria. This project responds to this challenge and proposes the development of new tools and resources for optimization of intracellular accumulation in clinical and investigational antibacterial agents effective against Gram-negative bacteria. In the proposed research, we will take on two interconnected challenges. First, we aim to generate predictive Machine Learning models for active efflux and permeation across the outer membrane, the two critical barriers in bacterial penetration by small molecules. We will assemble a chemical library around promising scaffolds discovered by the consortium, and will analyze this library in Pseudomonas aeruginosa, Escherichia coli, Klebsiella pneumoniae and Acinetobacter baumannii strains with variable permeability barriers. These models will become powerful guidance tools for the development of compounds with desired permeation properties. Second, we will establish Machine Learning relationships between the structural properties of Resistance-Nodulation-cell Division (RND) transporters and their biological activity by assembling a library of RND transporters and by characterizing their substrate specificities in terms of interactions with specific regions/motifs of the transporters. This should both improve the accuracy of predictions of compound permeation into bacteria and expand the predictions to diverse Gram-negative pathogens. We will use the learnings from the models to improve overall activity of chemical scaffolds identified by this team as a proof-of-concept for the developed tools. The established relationships will be applied to address the urgent and broadly important problem of improving intracellular accumulation of antibiotics in Gram-Negative bacteria, by optimizing RND-efflux avoidance and permeation across the outer membrane. All information generated within the project (ligand molecular structures and descriptors, biological activities, RND transporters structures. Machine Learning models) will be freely available to the community on a dedicated web platform. The project is a collaborative effort of the team of experienced investigators in areas critical for success of the proposed studies and includes experts from academic institutions, industrial partners, and the Center for Combating Antibiotic Resistant Bacteria.