Pennsylvania University Professor Uses Artificial Intelligence to Design New Antibiotics

2019-11-04

Since the accidental discovery of penicillin in 1928, medical researchers have been dedicated to finding other life-saving antibiotics and reusing them for humans. However, some bacteria develop resistance to drugs in less than a year after their introduction. For decades, no major new antibiotics have been developed.


Nature has become the main source of current antibiotics. Over millions of years, plants, fungi, and other organisms have evolved various clever chemical defenses against bacteria, so chemists have been tirelessly replicating them. For example, penicillin is produced by a fungus and kills bacteria by weakening their cell walls. Its discoverer, Alexander Fleming, once said, "I did not invent penicillin, I discovered it." Cesar de la Fuente, a professor at the University of Pennsylvania, believes that nature may have lost its inspiration, and now it is time to use 21st-century methods to search for new drugs: artificial intelligence.


In the laboratory at the University of Pennsylvania, Fuente and his team use computer algorithms and artificial intelligence to design and improve new drugs, screening for chemical components that have the potential to penetrate bacterial defenses.


At the Perelman School of Medicine at the University of Pennsylvania, De la Fuente's exploration of new antibiotics begins with the building blocks of life, amino acids, but he combines them in ways that nature has never thought of.


In a sense, this is a huge mathematical problem. The ways in which 20 core amino acids can be arranged into short peptide chains are potentially more numerous than the number of stars in the universe, but in nature, only 1% of these combinations exist, and only two have become actual antibiotics. De la Fuente hopes to discover new antibiotics beneficial to humans among all these combinations.In one project in the laboratory, Fuente automated the process and created a system that incorporates some principles of artificial intelligence. The system generates molecules, runs computer simulations, and then guides robotic arms to manufacture and test the winners' ability to combat live bacteria. The most effective combinations are fed back into the model for the next round of screening, resulting in the optimal combination.


In a recent study, he and his colleagues started with a naturally occurring antibiotic: a peptide found in pomegranate seeds. This peptide has weak antimicrobial properties, and they were interested in two features of this peptide. It contains certain positively charged amino acids, which means they will be attracted to the negatively charged membranes of bacteria. It includes other amino acids with chemical properties that allow them to interact with and disrupt these lipid membranes.If scientists can randomly generate hundreds of analogs of this natural peptide on a computer (by cutting it into fragments, reassembling them, and swapping different amino acids in several positions), perhaps the interaction between these two features can be optimized, resulting in a stronger microbial attack.


The researchers screened 15 of the hundreds of compounds generated by this algorithm and prepared them in the laboratory for bacterial testing. Indeed, several showed enhanced antibacterial activity, with one in particular standing out. They called it Punicalagin 2, and while it is still a peptide, it is significantly different from its natural source, with an unusually high content of arginine amino acids. Fuente and his team are working to make it the next penicillin.

RECENT AI NEWS

RECENT AI TOOLS