Cross-Molecular Active Learning for the Discovery of Antimicrobial Polyacrylamides

Area of Science:

• Polymer Chemistry
• Machine Learning
• Microbiology

Background:

• Antimicrobial resistance (AMR) is a growing global health crisis requiring novel therapeutic strategies.
• Synthetic copolymers offer a scalable approach to developing broad-spectrum antimicrobials that can bypass resistance mechanisms.
• Antimicrobial peptides (AMPs) serve as inspiration for designing synthetic antimicrobial agents.

Purpose of the Study:

• To develop potent antimicrobial polymers using a machine learning pipeline trained on AMPs.
• To synthesize and experimentally validate candidate polymers against *Escherichia coli*.
• To investigate the mechanism of action and synergistic potential of lead polymer candidates.

Main Methods:

• A cross-molecular machine learning pipeline was employed, trained on data from antimicrobial peptides.
• Candidate copolymers were synthesized and tested for antimicrobial activity against *Escherichia coli*.
• Mechanism of action studies included bacterial membrane permeabilization assays; synergy was assessed with clinical antibiotics against biofilms.

Main Results:

• A machine learning approach successfully identified potent antimicrobial polymer candidates.
• One synthesized copolymer effectively permeabilized the *E. coli* membrane, indicating a mechanism associated with reduced resistance.
• The lead copolymer exhibited significant synergy with a standard antibiotic regimen, reducing required dosage by three orders of magnitude for biofilm eradication.

Conclusions:

• Machine learning-guided development of synthetic copolymers is a viable strategy for creating novel antimicrobials.
• The identified copolymer demonstrates potential as a membrane-disrupting agent and a potentiator of existing antibiotics.
• These findings highlight a promising avenue for combating AMR, treating complex infections, and enhancing current antibiotic therapies.