DeepMindとAlphaFold、心臓病の鍵となるタンパク質を解明 Revealing a key protein behind heart disease

Google DeepMind and University College London (UCL) have announced a collaborative breakthrough in which AlphaFold was used to reveal the structure of a protein closely implicated in heart disease. Cardiovascular disease remains the leading cause of death worldwide, and a deeper molecular understanding of the proteins involved is essential for designing the next generation of therapies.

The research tackled a target whose structure had proven difficult to resolve through conventional experimental approaches such as X-ray crystallography or cryo-electron microscopy. By combining AlphaFold's high-accuracy structure prediction with experimental validation, the team was able to generate a reliable three-dimensional model. Knowing the precise shape of a protein is foundational to understanding how it functions, how mutations may disrupt it, and how small-molecule drugs might be designed to bind it.

AlphaFold, originally developed by DeepMind, made headlines in 2020 when it effectively solved the long-standing protein folding problem at the CASP14 competition, achieving accuracies comparable to experimental methods. Since then, the AlphaFold Protein Structure Database has grown to host more than 200 million predicted structures, freely accessible to researchers worldwide. The 2024 release of AlphaFold 3 extended capabilities to predicting interactions between proteins, DNA, RNA, and small-molecule ligands, broadening its relevance for drug discovery.

Applications in cardiology represent a particularly compelling validation of the technology's translational value. Heart disease involves a complex web of receptors, ion channels, and signaling proteins, many of which have been challenging to characterize. Structural insights of the kind produced here could, over time, support the identification of new drug binding sites and inform the design of more selective therapeutics, although the path from a predicted structure to an approved medicine remains long.

The broader ecosystem around AI-driven structural biology continues to expand. Isomorphic Labs, the drug discovery spinout from Alphabet, has announced partnerships with major pharmaceutical companies including Novartis and Eli Lilly, aiming to apply AlphaFold-derived insights at industrial scale. Competing efforts such as Meta's ESMFold and academic tools like RoseTTAFold from the Baker lab also contribute to a rapidly maturing field. Taken together, these developments suggest that structure prediction AI is transitioning from a research curiosity to a standard component of biomedical discovery pipelines, and similar breakthroughs in other disease areas appear likely in the coming years.