Evaluation of LLM-generated peptide as foundation template for discovery of effective encrypted AMPs against clinical superbugs

Novel antimicrobial agents are urgently needed to combat the antibiotic-resistance crisis, particularly in the face of multidrug-resistant (MDR) pathogens like carbapenem-resistant <i>Acinetobacter baumannii</i> (CRAB) and methicillin-resistant <i>Staphylococcus aureus</i> (MRSA). In this study, we present an approach that combines generative large language model with sequence alignment to identify promising antimicrobial peptides. With this strategy, we rapidly identified five novel encrypted peptides based on a generated template, demonstrating significant antimicrobial activity against a broad spectrum of clinical MDR pathogens. Among them, PL-15 stood out as a potent, broad-spectrum AMP with comparable therapeutic efficacy to polymyxin B against CRAB infections <i>in vivo</i>. Mechanistic investigations revealed that PL-15 exerts its bactericidal effects by disrupting both the outer and cytoplasmic membranes, causing membrane depolarization, elevating intracellular reactive oxygen species (ROS) levels, and ultimately leading to rapid bacterial cell death. Additionally, PL-15 demonstrated remarkable antibiofilm activity, inhibiting biofilm formation and eradicating pre-existing biofilms, further reducing the risk of resistance development. This work highlights the potential of using generative model combined with sequence alignment to accelerate the discovery of novel analogs with enhanced properties.IMPORTANCEThe rise of multidrug-resistant pathogens, such as carbapenem-resistant <i>Acinetobacter baumannii</i> and methicillin-resistant <i>Staphylococcus aureus</i>, poses a severe threat to public health, making the search for novel antimicrobial agents a critical priority. In this study, we present an innovative approach combining generative large language models and sequence alignment to identify promising antimicrobial peptides. This method allowed for the rapid discovery of five encrypted peptides with strong antimicrobial activity against a range of multidrug-resistant pathogens. Among them, PL-15 showed remarkable efficacy, comparable to polymyxin B, and exhibited potent antibiofilm properties, making it a strong candidate for further development. By providing a novel approach to discovering antimicrobial agents, this work presents a promising solution to the escalating crisis of antibiotic resistance.