Novel Antimicrobial Peptides Designed Using a Recurrent Neural Network Reduce Mortality in Experimental Sepsis

Area of Science:

• Microbiology
• Biotechnology
• Drug Discovery

Background:

• Antibiotic resistance is a growing global health threat.
• Novel antimicrobial peptides (AMPs) offer a promising alternative to traditional antibiotics.
• De novo design strategies are crucial for discovering new AMPs.

Purpose of the Study:

• To design and evaluate novel antimicrobial peptides using artificial intelligence.
• To assess the in vitro and in vivo efficacy of generated peptides against resistant bacteria.
• To investigate the potential of AMPs in combating carbapenem-resistant Gram-negative infections.

Main Methods:

• Utilized a long short-term memory recurrent neural network (LSTM RNN) to generate 198 novel peptide sequences.
• Synthesized and tested five generated peptides for in vitro antimicrobial activity.
• Evaluated the efficacy of promising peptides in a murine model of *Klebsiella pneumoniae*-induced sepsis.

Main Results:

• PEP-38 and PEP-137 exhibited in vitro activity against carbapenem-resistant *Klebsiella aerogenes* and *K. pneumoniae*, with PEP-137 also active against *Pseudomonas aeruginosa*.
• In vivo studies showed PEP-36 (control) yielding a 66.7% survival rate, PEP-137 a 50% survival rate, and PEP-38 being ineffective in a *K. pneumoniae* sepsis model.
• PEP-36, PEP-136, and PEP-174 showed no significant antibacterial effects.

Conclusions:

• AI-driven de novo design can generate effective antimicrobial peptides.
• PEP-137 shows potential for further development as a therapeutic agent against resistant Gram-negative bacteria.
• This approach provides a foundation for developing novel peptide-based drugs to address antibiotic resistance.