Development of Next-Generation Antibiotics Integrating AI-Driven Design with Target Specificity, Permeability, and Synthetic Accessibility – A-IMPACT

The objective of this project is to use artificial intelligence (AI) to identify new antibacterials targeting a complex involved in the synthesis of bacterial peptidoglycans (PGN) and accounting for physiologically relevant microenvironments such as biofilms.

Our project is motivated by several considerations.

First, the PGN component of the bacterial cell wall plays a key role in shape maintenance, resistance to osmotic pressure, and cell division. It is also the target of two successful classes of antibiotics, ß-lactams and glycopeptides, for which rapidly rising resistance is a concern. Nonetheless, PGN remains a highly attractive target for the development of novel antibacterials with new modes of action.

Second, we have identified the complex formed by the proteins MreC (a structural protein that serves as platform for PGN biosynthesis factors during cell elongation) and PBP (penicillin binding protein) as a promising target for new antibiotic drugs, since disrupting this interaction kills H. pylori in vitro. Because the MreC:PBP2 interface is conserved among several bacteria of the ESKAPE group, we expect these drugs to have broad spectrum activity.

Third, antibacterial efficacy is profoundly influenced by the physiological state of bacteria, which varies with environmental cues such as pH, oxygen levels, and nutrient availability. This physiological adaptation, which is typically ignored in standard drug screens, helps explain the discrepancies observed between in vitro potency and in vivo efficacy.

Fourth, AI-guided approaches enable to select or design compounds with much higher hit rate than standar screens, and thereby offer a scalable avenue to revitalize the antibiotic pipeline in both academic and industrial settings.
Our project will tackle these challenges and opportunities by implementing methodologies to screen for inhibitors of MreC:PBP2 virtually and experimentally, in species including Helicobacter pylori and Pseudomonas aeruginosa. These targeted screens will be complemented by growth inhibition screens on large chemical libraries, including in clinically relevant conditions of biofilms (for P. aeruginosa) and low pH (for H. pylori). We will also develop generative AI methods to design compounds that can be easily synthesized and have high predicted activity against bacterial growth and/or MreC:PBP2 interactions. About a hundred of the selected or designed compounds will be studied experimentally for their antimicrobial activity.
Our integrated pipeline, grounded in clinical microbiology, AI-driven chemical prioritization and design, will deliver structurally novel antibiotic candidates optimized for the physiological conditions in which persistent infections actually occur.