Researchers have used artificial intelligence to identify 25 new protein compounds that can kill bacteria and fungi, including two that showed particular potency against skin and lung infections in mice.
The results, published in Science Advances on Feb. 5, suggest AI could be used to generate new drugs in the fight against antimicrobial resistance.
The two star antimicrobial peptides (AMPs) are called AMP-24 and AMP-29. When given to mice with skin or lung infections of the bacteria Acinetobacter baumannii, AMP-24 reduced levels of the bacteria and lessened inflammation and scarring in the lungs. A. baumannii is known to spread through healthcare settings, where it can infect vulnerable patients, and is known to develop antibiotic resistance, according to the Centers for Disease Control and Prevention.
AMP-29 was similarly able to curb mouse skin infections, this time of the fungus Nakaseomyces glabratus (previously known as Candida glabrata). After 24 hours of treatment, infected mice saw the number of fungal cells in their skin significantly reduced, according to researchers.
In addition to these two standouts, seven other compounds were able to kill fungi and bacteria in laboratory tests, according to a Feb. 5 press release.
The researchers were led by Wenqiang Chang, Ph.D., and Hongxiang Lou, Ph.D., of Shandong University in Jinan, China. The team identified the compounds using a two-step model that first generated new proteins and then assessed their molecular properties based on their structures to determine whether they could have antimicrobial effects.
The scientists then synthesized the most promising candidates and pitted them against pathogenic bacteria and fungi in the lab, including bacteria flagged by the World Health Organization as particularly troubling: Pseudomonas aeruginosa, Klebsiella pneumoniae, Acinetobacter baumannii and Escherichia coli.
The protein-designing model could also be used to discover and design new drugs targeting tumors, or even in the treatment of diabetes, the authors wrote in their paper.