Cryo-EM ligand building using AlphaFold3-like model and molecular dynamics.

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Cryo-electron microscopy (cryo-EM) enables high-resolution biomolecular structure determination.
• Accurate modeling of small molecule ligands within cryo-EM maps remains challenging.
• Existing modeling methods are often protein-centric, limiting their application to diverse ligands.

Purpose of the Study:

• To develop and validate an integrated approach for fitting small molecules into cryo-EM density maps.
• To improve the accuracy and resolution of ligand-protein complex structures determined by cryo-EM.
• To provide a robust pipeline for modeling diverse small molecules in cryo-EM studies.

Main Methods:

• Integration of artificial intelligence (AI) with cryo-EM density-guided simulations.
• Utilizing protein sequence, ligand information, and experimental cryo-EM maps as inputs.
• Employing flexible fitting within molecular dynamics simulations to refine ligand poses.

Main Results:

• Successful validation on diverse protein-ligand complexes (kinases, GPCRs, solute transporters).
• Significant improvement in ligand model-to-map cross-correlation, from 40-71% to 82-95% with simulations.
• Demonstrated AI's capability to predict experimental ligand poses, with simulations further enhancing accuracy.

Conclusions:

• The developed pipeline offers a straightforward method for modeling ligand-protein complexes in cryo-EM.
• This approach enhances the resolution and reliability of structural data for small molecule interactions.
• Facilitates a deeper understanding of how small molecules regulate biological functions through atomic-level insights.