AI accelerates the pace of development of new drugs for brain diseases
AI, artificial intelligence, can be used to identify molecules with great potential to be developed into new drugs for mental illnesses, according to researchers from Karolinska Institutet, affiliated to CMM, among others. The study, which has been presented in the journal Science Advances, and conducted on mice, shows that AI with high precision can contribute to more effective future treatments for conditions such as psychosis.
This text is based on a press release from Karolinska Institutet.
In modern drug development, experimental methods are often used to determine the three-dimensional structures of target proteins (the recipient protein of the drug), to understand how molecules bind to them.
With the help of such molecular information, researchers can design drug molecules in an efficient way. However, the strategy has some limitations related to the challenge of determining the structures of many important target proteins.
In recent years, major advances in AI have made it possible to predict the structures of such proteins with greater precision than before.
Target protein for the treatment of psychosis
Researchers from Karolinska Institutet and Uppsala University, among others, have now studied whether the AI structures are also good enough for the development of drugs for mental illnesses.
“The TAAR1 protein is an interesting target for the development of drugs for mental illnesses. Drug molecules that activate TAAR1 have shown promising results in the treatment of schizophrenia and psychosis,” says Marcus Saarinen, doctoral student in Per Svenningsson’s CMM Group at the Department of Clinical Neuroscience, KI, and one of the study’s first authors.
Using calculations on supercomputers, the researchers searched chemical libraries with several million molecules to find the ones that fit best in the predicted structure.
Astonishingly high precision
Molecules that were predicted to bind to the receptor were tested in experiments by Marcus Saarinen and CMM Group Leader Per Svenningsson. An unexpectedly large number of the molecules activated TAAR1, and one of the most potent also showed promising effects in animal experiments with mice.
In the final stage of the study, the researchers were able to compare experimental structures for TAAR1 with the AI models. The AI-generated structures had what the researchers describe as “an astonishingly high level of precision”.
“AI-based structure determination with Alphafold is likely to facilitate the development of new drug candidates for therapies of several diseases in the near future,” says Per Svenningsson, senior physician and professor at the same department and one of the study’s senior authors.
The study was funded by the Knut and Alice Wallenberg Foundation, the eSSENCE research programme and the Swedish Research Council. See the study for reported conflicts of interest.
Publication:
AlphaFold accelerated discovery of psychotropic agonists targeting the trace amine-associated receptor, Alejandro Díaz-Holguín, Marcus Saarinen, Duc Duy Vo, Andrea Sturchio, Niclas Branzell, Israel Cabeza de Vaca, Huabin Hu, Núria Mitjavila-Domènech, Annika Lindqvist, Pawel Baranczewski, Mark J Millan, Yunting Yang, Jens Carlsson, Per Svenningsson. Science Advances, online August 7, 2024. doi: 10.1126/sciadv.adn1524