Designing antimicrobials with programmable mechanism and safety

Antimicrobial peptides (AMPs) present a promising avenue for combating antimicrobial resistance; however, existing models fall short in controlling physicochemical properties and motifs that determine activity and selectivity. To address this challenge, researchers introduce OmegAMP, a conditional diffusion framework capable of regulating net charge, mean hydrophobicity, and sequence length to facilitate de novo, analog, and motif-guided design of AMPs.

Validation of OmegAMP's capabilities is demonstrated through testing on 204 AMPs characterized in wet-lab experiments. The de novo generation of AMPs via this framework yielded compounds with broad activity against multidrug-resistant Gram-negative bacteria. Moreover, the analog generation approach converted six inactive AMP prototypes into effective antimicrobials, with each analog's membrane-disruption mode being dictated by the original prototype.

The motif-guided analog generation mechanism preserves the ability of active prototypes to engage with lipopolysaccharides, while also enabling the redesign of a non-antimicrobial leucine zipper to acquire antimicrobial activity without compromising DNA-perturbing characteristics in vitro. In murine models simulating skin and thigh infections, OmegAMP-generated leads successfully reduced bacterial burdens, with a DNA-perturbing lead matching the efficacy of a fluoroquinolone control treatment when administered systemically.

By providing a programmable route to the development of new peptide antibiotics, OmegAMP promises a more targeted approach to designing AMPs with desired mechanisms and safety profiles based on the chosen prototype.