AI-driven pipeline designs novel antimicrobial peptides (AMPs) with broad-spectrum activity and strong biofilm inhibition

The escalating crisis of antimicrobial resistance necessitates novel therapeutic strategies beyond conventional antibiotics. Antimicrobial peptides (AMPs) offer a promising alternative due to their diverse mechanisms of action, including direct bacterial killing and immunomodulation. However, traditional AMP discovery is hampered by the high cost and low throughput of biochemical screening, while existing computational methods struggle to balance efficacy with structural novelty. This research addresses the critical gap in efficient, scalable de novo AMP discovery, aiming to accelerate the identification of potent new antimicrobial agents.

Researchers developed an integrated "generation-evaluation-validation" framework for de novo AMP discovery. First, a soft prompt-tuned ProtGPT2 model was used to generate candidate AMPs with novel structures. Next, a multiple-choice learning ensemble model provided high-confidence evaluation of these candidates through a dynamic voting network. Finally, antimicrobial experiments were conducted to validate the activity of nine top-ranked de novo AMPs. Validation involved monitoring bacterial surface changes, assessing biofilm inhibition, membrane disruption, and hemolysis using in vitro assays.

Out of nine computationally designed AMP candidates, four exhibited potent strain-specific antimicrobial activity, while two demonstrated broad-spectrum efficacy against various bacterial strains. All tested AMPs showed strong biofilm inhibition, a critical factor in combating persistent infections. Furthermore, these peptides induced potent membrane disruption in bacterial cells, a key mechanism of action for many AMPs. Importantly, the tested AMPs exhibited minimal hemolysis, indicating a favorable safety profile for potential therapeutic use. The integrated framework successfully balanced the generation of novel structures with the identification of therapeutically promising peptides.

The framework identified two de novo AMPs with broad-spectrum efficacy and strong biofilm inhibition, alongside minimal hemolysis, highlighting significant therapeutic potential.