Molecular Dynamics Flexible Fitting (MDFF)

The core mechanism of MDFF involves converting the cryo-EM density map into a potential energy function (U_EM). This potential is added to the standard molecular dynamics force field. The potential is defined such that atoms are driven into high-density regions, effectively steering the atomic model from its initial position into the target map's contours. The force on each atom is proportional to the gradient of the density map, ensuring a physically plausible fitting trajectory.

Many macromolecular complexes, such as viral capsids or chaperonins, exhibit symmetry. MDFF can enforce these symmetries during the fitting simulation to improve the global quality of the fit and prevent overfitting to local noise. By applying harmonic restraints that maintain the spatial relationships between symmetry-related subunits, the effective signal-to-noise ratio is increased, leading to more accurate and globally consistent atomic models.

MDFF is not limited to automated protocols. It can be run in an interactive mode using visualization tools like VMD. This allows researchers to manually apply steering forces to specific atoms or domains in real-time, guiding the model out of local minima or into ambiguous density. This human-in-the-loop approach is invaluable for interpreting complex or lower-resolution regions where automated fitting may struggle.

To preserve the integrity of the model during fitting into medium-resolution maps (4-7 Å), MDFF can apply secondary structure restraints. These harmonic potentials maintain the characteristic hydrogen bonding patterns and dihedral angles of alpha-helices and beta-sheets. This prevents the unfolding of these well-predicted structural elements while allowing flexible loops and domains to adapt to the experimental density.

For high-resolution maps, a cascade MDFF protocol is often employed. This involves a series of sequential simulations:

• Step 1: Fit the model into a map filtered to a lower resolution (e.g., 5 Å) to capture the global shape.
• Step 2: Progressively fit into maps with increasing resolution, allowing the model to settle into finer structural details like side-chain rotamers and ordered water molecules. This staged approach ensures a robust and accurate final fit.

MDFF can be combined with Quantum Mechanics/Molecular Mechanics (QM/MM) simulations. This hybrid approach is critical for studying enzymatic active sites or metalloproteins within a cryo-EM map. The chemically active region (e.g., a catalytic center with a bound drug) is treated with quantum mechanical accuracy, while the rest of the protein scaffold is fitted using classical MD. This provides an unprecedented view of reaction mechanisms in near-native states.