Algorithm optimises peptides to target drug-resistant bacteria

16th April 2018 (Last Updated April 16th, 2018 17:14)

Researchers from Massachusetts Institute of Technology (MIT) and the Catholic University of Brasilia have developed a computer algorithm to develop antimicrobial peptides that kill microbes more effectively than peptides synthesised by humans.

Algorithm optimises peptides to target drug-resistant bacteria
Guavanin 2 cleared skin infections more effectively than the Pg-AMP1 peptide it is based on. Credit: Wikimedia

Researchers from Massachusetts Institute of Technology (MIT) and the Catholic University of Brasilia have developed a computer algorithm to develop antimicrobial peptides that kill microbes more effectively than peptides synthesised by humans.

The team published their findings in ‘In silico optimization of a guava antimicrobial peptide enables combinatorial exploration for peptide design’, in the journal Nature Communications. The report describes their algorithm, which mimics the natural process of evolution, and has already produced a drug candidate that has successfully killed bacteria in mice.

Antimicrobial peptides kill microbes by damaging their cell membranes then entering them and disrupting cellular targets such as DNA, RNA and proteins, but producing them is time-consuming and expensive.

A British Government report estimated that antibiotic-resistant bacteria will kill 10 million people per year by 2050, ‘so coming up with new methods to generate antimicrobials is of huge interest, both from a scientific perspective and also from a global health perspective,’ said MIT postdoctoral researcher and Areces Foundation Felow Cesar de la Fuente-Nunez.

The researchers applied Charles Darwin’s theory of natural selection to their algorithm, which then generated thousands of variants from a peptide sequence and tested the variants for desired traits specified by the researchers. The team began with an antimicrobial peptide found in the seeds of a guava plant, known as Pg-AMP1, and asked the algorithm to produce peptide sequences which tend to form alpha helices and have a particular level of hydrophobicity, both features that help peptides penetrate bacterial membranes.

The algorithm then generated and tested tens of thousands of peptide sequences, and human scientists tested the top 100 candidates against bacteria grown in lab dishes. The best performer, guavanin 2, contains 20 amino acids and is rich in arginine, features that make the peptide much more potent, especially against Gram-negative bacteria, a category which includes species responsible for pneumonia and urinary tract infections.

“By using this approach, we were able to explore many, many more peptides than if we had done this manually. Then we only had to screen a tiny fraction of the entirety of the sequences that the computer was able to browse through,” said de la Fuente-Nunez.

The researchers tested guavanin 2 on mice, and found it to clear skin infections caused by a type of Gram-negative bacteria known as Pseudomonas aeruginosa more effectively than the original Pg-AMP1 peptide. The team plans to develop guavanin 2 for potential human use, and to develop other potent antimicrobial peptides.

The study was funded by the Ramón Areces Foundation and the Defense Threat Reduction Agency.