Flexible Fitting of Small Molecules into Electron Microscopy Maps Using Molecular Dynamics Simulations with Neural...

Area of Science:

• Structural Biology
• Computational Chemistry
• Biophysics

Background:

• Cryo-electron microscopy (cryo-EM) has advanced, but determining protein-drug complex structures remains challenging.
• Accurate structure determination requires fitting models, sampling ligand geometries, and optimizing side-chain rearrangements for binding energy.

Purpose of the Study:

• To present a flexible-fitting pipeline for refining protein-ligand complex structures using cryo-EM data.
• To enable accurate structure determination of protein-drug complexes, even with medium to low-resolution data.

Main Methods:

• Utilized molecular dynamics flexible fitting (MDFF) to refine structures against 3-5 Å electron density data.
• Employed enhanced sampling to explore binding pocket rearrangements.
• Integrated quantum mechanics/molecular mechanics (QM/MM) and neural-network potential (NNP)/MM models for chemically diverse ligands.

Main Results:

• The pipeline produced structures comparable to or better than high-resolution cryo-EM or X-ray models, even with low-resolution input.
• Neural-network potentials (NNPs) improved robustness and provided a 6.5 Å radius of convergence for ligand structure determination.
• Density functional theory (DFT) calculations confirmed that improved fitting and coordination reduced ligand strain energy, indicating correct binding.

Conclusions:

• The developed flexible-fitting pipeline effectively refines protein-ligand complex structures from cryo-EM data.
• The integration of QM/MM and NNP/MM models enhances accuracy and robustness for diverse ligands.
• A computationally inexpensive strain energy calculation protocol aids in monitoring model quality and binding interactions.