KU Leuven and SandboxAQ partner to accelerate the search for Parkinson’s treatments with AI

The lab of Professor Peter Vangheluwe at KU Leuven is collaborating with SandboxAQ, an AI company, to expedite the discovery and testing of candidate drugs to restore the function of ATP10B, a protein associated with protection against Parkinson’s disease. This novel technology will accelerate and expand the search for small molecule drugs to enhance ATP10B’s function, paving the way for new treatments for Parkinson’s disease.

Building on its research grant support from The Michael J. Fox Foundation for Parkinson’s Research and the Aligning Science Across Parkinson’s Initiative, the Vangheluwe lab and its collaboration with its use of the SandboxAQ AI platform will help to better understand the underlying mechanisms of ATP13A2 and ATP10B dysfunction in Parkinson’s disease and pioneered drug discovery efforts on ATP13A2.

In Parkinson’s disease, communication and energy management in brain cells are impaired. The ATP10B protein plays a vital role in this process by aiding the transport of a specific type of lipid molecule (glucosylceramide) within cells. When this process goes awry, brain cells struggle to survive.

The ATP10B protein is highly complex, and its mechanisms for transporting lipid molecules are not yet fully understood. Nonetheless, researchers are seeking drugs (small molecules) that can enhance the protein’s function, potentially mitigating or even preventing the damage to brain cells caused by Parkinson’s disease.

The lab of Professor Peter Vangheluwe at KU Leuven is a world leader in uncovering the disease mechanisms underlying Parkinson’s and has previously made significant discoveries about the role of various genes in the development of the disease.

“With AI technology, we will now simulate and test the effects of potential drugs on a model of the ATP10B protein. This significantly accelerates our research, something that would not be feasible with laboratory experiments alone,” explains Professor Vangheluwe. “SandboxAQ’s technology offers us a novel way to design unique molecules and test our ideas more quickly and accurately. This provides insights that were previously unattainable and propels our research forward.”

For this project, the Laboratory for Cellular Transport Systems, led by Professor Vangheluwe, is collaborating with the Centre for Drug Design and Discovery (CD3) of KU Leuven Research and Development and SandboxAQ, a spin-off of Alphabet Inc. (the parent company of Google).

SandboxAQ uses advanced AI simulations and Large Quantitative Models (LQMs) to accelerate and improve drug discovery projects by utilizing methods such as protein folding, free energy calculations, and protein language models to enhance both small molecule and antibody design. Their in-house AI and physics-based structure-prediction capabilities, which have already demonstrated remarkable accuracy in multiple projects for biotech companies and large pharmaceutical firms, will be used to decipher the structure of ATP10B. This is critical for unraveling the conformational cycle of the protein, understanding the structural basis of the relevant state to target for activation, and identifying binding pockets. The structure prediction will enable the application of an integrated virtual screening workflow that combines active learning with physics-based methods in the search for activators of the ATP10B system, unlocking the potential of this target.

“SandboxAQ’s sophisticated AI techniques and models have already demonstrated their capability to swiftly design groundbreaking molecules for complex diseases including neurodegeneration,” says Nadia Harhen, General Manager of AI Simulation at SandboxAQ. “KU Leuven hosts one of the world’s leading research centres for Parkinson’s disease, and we are excited to deploy our technology to accelerate their work and develop new treatments for Parkinson’s.”

This article was first published on 13 January by KU Leuven.