AI drug discovery firm Valence joins Parkinson’s disease push

Illustration of the thought processes in the brain

Canada’s Valence Discovery has joined a University of Montreal-led project to try to find new drugs to treat the involuntary movement complications that can accompany treatments for Parkinson’s disease. 

The go-to treatment for the five million people worldwide with Parkinson’s is levodopa, which replaces the activity of the dopamine neurotransmitter dopamine that gets depleted by the degeneration of nerve cells in the brain in the disease.

While very effective, long-term levodopa therapy over several years will almost inevitably result in disabling motor fluctuations – known as dyskinesias – that take the form of involuntary, erratic, writhing movements of the face, limbs or trunk.

At the moment, the main treatment is to adjust the dose of levodopa and other Parkinson’s drugs to balance the movement difficulties associated with the disease itself with those caused by treatment.

Dopamine antagonists or amantadine can be added in to tackle symptoms that occur when levodopa levels are high in the body, but these often only provide short-term relief and have their own side effects, including fainting, dizziness, and hallucinations.

Enter the alliance between Valence (formerly InVivo AI), Montreal University and Canada’s Institute for Research in Immunology and Cancer (IRIC), which will use AI to design drug candidates against a new target discovered in the lab of Dr Daniel Levesque of the university’s pharmacy faculty.

Levesque has been working on the Nur77/RXR nuclear receptor complex, a promising new pharmacological target for movement disorders, and the partners will attempt to design highly selective drugs that can bind to and affect the activity of that receptor.

In animal models of Parkinson’s, Nur77/RXR seems to be directly involved in levodopa-induced dyskinesias. Rats engineered to lack the receptor were less likely to have involuntary movements compared to normal rats after long-term treatment with levodopa.

Valence’s machine-learning drug discovery platform, which the company says can design candidates based on very small data pools, will underpin the project. The platform is powered by technologies developed at Quebec’s deep learning research institute Mila.

“We’re thrilled to be working with Dr Levesque and the world-class team at IRIC, who have an extensive track record of collaborating with leading industry partners,” said Daniel Cohen, Valence’s CEO.

“This collaboration is an important example of how we’re bringing modern machine learning methods, custom-built for drug discovery, to innovative R&D organisations of all shapes and sizes,” he added.