Open-source competition finds potential candidates for Parkinson’s disease drugs

A new open-science competition has identified seven small molecules that could eventually be developed into drugs for Parkinson’s disease.

The CACHE, or Critical Assessment of Computational Hit-Finding Experiments, Challenge instructed participants to devise a small molecule that can bind to the WD40 repeat (WDR) domain of the protein abbreviated as LRRK2. Researchers have linked mutations in the gene that codes for this protein, along with dysfunctions in the protein itself, to Parkinson’s disease. CACHE organizers have made data from the open-source science project freely available on the challenge’s website.

No small molecules have been developed into drugs that can target the WDR domain. The domain has a toroidal shape, which makes small-molecule development difficult, according to Olexandr Isayev, a chemist at Carnegie Mellon University and a member of one of five academic groups that submitted a molecule. An anonymous group and a team from Merck KGaA also each submitted a molecule.

A typical small molecule would take up less than half of the available space within the hole of the torus, leaving space, which makes a potential binding “not very specific, or it’s very hard to find,” Isayev tells C&EN.

Isayev and colleagues drew from a library of existing molecules and used free-energy simulations and machine learning to determine which ones best fit the target. They then optimized those hits further.

These are very early-stage findings, and if any of these molecules progress down the drug development pipeline, they will likely need to be optimized. Matthew Todd, who researches drug discovery at University College London and was not involved with the competition, says these hits could be used to develop chemical probes. Then, pharmaceutical companies could develop those probes into drugs. Even if this part is done behind closed doors, he says, “it still means a lot of the work is open.”

The results could eventually go beyond developing new treatments for Parkinson’s disease, says Ryan Merkley, CEO of Conscience, which organized the competition. The nonprofit worked with the Structural Genomics Consortium to test the submitted molecules.

Merkley says the CACHE project could help artificial intelligence systems generate better drug candidates for other diseases by making more data available for scientists to use to train and compare their models.

“Very rarely do you get to pit every approach against a common problem and see which one works better,” Merkley says.

Open-source drug discovery, Todd says, is most promising in areas where conventional drug development has been less successful, including with neglected diseases and neurological drugs.

Merkley says his group focuses on areas of “market failure,” or areas where drug development is not getting the time, money, or attention in the traditional landscape of pharmaceutical companies developing and marketing drugs.

This first CACHE competition was geared toward Parkinson’s disease thanks to funding from the Michael J. Fox Foundation for Parkinson’s Research. The next competitions will focus on targeting three different binding sites: on SARS-CoV-2, a cancer therapy target, and a protein linked to high weight.