Alzheimer’s Drug Production a Step Closer With the Help of Bacteria and AI

Galantamine is a drug commonly used to treat the symptoms of mild to moderate dementia, a sign of Alzheimer’s disease.

Its active ingredients are extracted in a time-consuming process from daffodils. Daffodil harvesting can be unpredictable due to variations in crop yields and weather, meaning the drug’s supply and price can fluctuate. However, it is not commercially viable to produce the active compounds required at such a large scale.

Alternative methods to produce the drug are therefore being investigated. In a new study, researchers from the University of Texas (UT) at Austin used microbial fermentation – a common bioprocessing approach – to produce galantamine with the help of artificial intelligence (AI) and biosensors.

The study is published in Nature Communications.

Harnessing bacterial fermentation

Microbes have been used for hundreds of years to make useful byproducts such as alcohol from yeast and cheese and yogurt from bacteria. Now, microbial fermentation produces therapeutic molecules such as insulin, autoimmune treatments and vaccines.

The researchers in the current study used genetically modified Escherichia coli bacteria to produce 4’-O-methylnorbelladine, a chemical building block of galantamine that is extracted from daffodils.

“The goal is to eventually ferment medicines like this in large quantities,” said Andrew Ellington, a professor of molecular biosciences and author of the study. “This method creates a reliable supply that is much less expensive to produce. It doesn’t have a growing season, and it can’t be impacted by drought or floods.”

Though using microbes to produce therapeutics is not a new concept, the use of AI is, and it could help to expand on what is possible with fermentation. The UT researchers’ approach hinged on an AI system called MutComputeX, developed by UT postdoctoral fellow Danny Diaz. MutComputeX identifies how to mutate the bacterial proteins to increase efficiency and maximize the production of the desired compound.

“This system helped identify mutations that would make the bacteria more efficient at producing the target molecule,” Diaz said. “In some cases, it was up to three times as efficient as the natural system found in daffodils.”

Additionally, the researchers also developed a biosensor to detect the bacteria that were successfully producing the desired compounds. This glowed green when it came into contact with the compounds, enabling the researchers to analyze how much was being produced.

“The biosensor allows us to test and analyze samples in seconds when it used to take something like five minutes each,” said the paper’s first author and postdoctoral researcher Simon d’Oelsnitz. “And the machine learning program allows us to easily narrow candidates from tens of thousands to tens. Put together, these are really powerful tools.”

Reference: d’Oelsnitz S, Diaz DJ, Kim W, et al. Biosensor and machine learning-aided engineering of an amaryllidaceae enzyme. Nat Commun. 2024;15(1):2084. doi: 10.1038/s41467-024-46356-y

This article is a rework of a press release issued by the University of Texas at Austin. Material has been edited for length and content.