Cambridge researchers use AI to find highly potent Parkinson's drug candidates in a tenth of the usual time

This approach could help get potential treatments to Parkinson's patients much more quickly.

One in 37 people alive in the UK today is expected to be diagnosed with Parkinson's in their lifetime. The fastest-growing neurological condition worldwide, it is believed to affect more than six million people worldwide, and that number is due to triple by 2040.

No disease-modifying treatments are currently available for the condition, which causes motor symptoms and can also affect the gastrointestinal system, nervous system, sleeping patterns, mood and cognition.

The disease is characterised by the clumping, or aggregation, of the protein alpha-synuclein.

When proteins like this misfold, they can form abnormal clusters called Lewy bodies, which build up within brain cells stopping them from functioning properly. Their presence is thought to contribute to the death of nerve cells seen in Parkinson's.

The Cambridge researchers designed an AI strategy to identify small molecules that bind to the amyloid aggregates and block their proliferation.

They screened a chemical library of hundreds of entries using machine learning to identify five highly potent compounds for investigation.

Usually the screening of large chemical libraries for drug candidates is extremely time-consuming and expensive, and often proves unsuccessful.

Prof Michele Vendruscolo, from the Yusuf Hamied Department of Chemistry, who led the research, said: "One route to search for potential treatments for Parkinson's requires the identification of small molecules that can inhibit the aggregation of alpha-synuclein, which is a protein closely associated with the disease. But this is an extremely time-consuming process - just identifying a lead candidate for further testing can take months or even years."

There are clinical trials for Parkinson's under way, but the lack of approval for any disease-modifying drug to date reflects the lack of methods to identify the correct molecular targets and engage with them.

But after using their machine learning method to identify compounds, the researchers tested a small number of the top-ranking candidates to find the most potent inhibitors of aggregation.

The data gained from these experimental assays was fed back into the machine learning model iteratively and after a few rounds, the highly potent compounds were identified.

"Instead of screening experimentally, we screen computationally," said Prof Vendruscolo, who is co-director of the Centre for Misfolding Diseases. "By using the knowledge we gained from the initial screening with our machine learning model, we were able to train the model to identify the specific regions on these small molecules responsible for binding, then we can re-screen and find more potent molecules."

The Cambridge team developed compounds to target pockets found on the surfaces of the aggregates, which are responsible for their exponential proliferation.

The compounds are hundreds of times more potent than those previously reported.

"Machine learning is having a real impact on drug discovery - it's speeding up the whole process of identifying the most promising candidates," added Prof Vendruscolo. "For us, this means we can start work on multiple drug discovery programmes - instead of just one. So much is possible due to the massive reduction in both time and cost - it's an exciting time."

The research, reported in the journal Nature Chemical Biology, was conducted in the Chemistry of Health Laboratory in Cambridge, which was established with the support of the UK Research Partnership Investment Fund (UKRPIF) to promote the translation of academic research into clinical programmes.