Revealing a key protein behind heart disease

Source: DeepMind Date: 2025-11-25 URL: https://deepmind.google/blog/revealing-a-key-protein-behind-heart-disease/

Summary

Researchers at Google DeepMind and collaborators used cryo-EM and AlphaFold to resolve the atomic-resolution structure of ApoB100, the structural protein of LDL (“bad cholesterol”) particles. AlphaFold generated the raw structural predictions that were then interpreted against cryo-EM experimental data. The structure reveals ApoB100’s cage-like shell and ribbon-like belt architecture — explaining how LDL particles maintain integrity in the bloodstream, and opening new drug target possibilities for atherosclerotic cardiovascular disease (ASCVD), the world’s leading cause of death.

Implications

AlphaFold as experimental-data interpreter, not just structure predictor, is the methodological advance. Prior AlphaFold framing was: “predict structure from sequence, validate with experiment.” This result uses AlphaFold outputs as the raw material to interpret experimental cryo-EM data — a tighter integration where AI and experiment are co-equal tools in the same analysis pipeline. That’s how this discovery happened: not AI replacing experiment, but AI enabling interpretation of data that was previously uninterpretable without a structural prior.

ApoB100 structure is a genuine drug target unlock. ASCVD is the #1 global cause of death. ApoB100 is the structural protein of every LDL particle. Knowing its atomic architecture means medicinal chemists can design molecules that bind specific sites — allosteric inhibitors, antibodies, degraders — rather than working with mechanism-agnostic statins. That’s a new drug target surface for the most consequential disease target in cardiovascular medicine.

This is the second major AlphaFold clinical-path result after the GLYK crop enzyme. The pattern: AlphaFold generates structural predictions → wet-lab or cryo-EM validates → clinical or agricultural application follows. The pace of this pattern is accelerating. ApoB100 joins AlphaFold-3 drug discovery partnerships and the GLYK crop engineering result as evidence that AlphaFold’s output is now regularly entering clinical development pipelines.

Watch:

• Drug development programs targeting ApoB100 structural sites identified in this study — which companies file IND applications citing this structure?
• Whether the cryo-EM + AlphaFold co-interpretation method becomes a standard workflow in structural biology — methods papers that cite this work
• Competition from AlphaFold 3’s diffusion-based structure prediction for protein-ligand complexes: does the new method improve on the static structure resolved here?