Comparing atomistic simulation data with the NMR experiment: how much can NOEs actually tell us?

Area of Science:

• Computational Biology
• Biophysics
• Structural Biology

Background:

• Molecular dynamics (MD) simulations are crucial for understanding protein and nucleic acid behavior.
• Nuclear magnetic resonance (NMR) data, particularly nuclear Overhauser enhancement (NOE) derived distances, are frequently used to validate simulation quality.
• NOE distance restraints are fundamental for NMR-based protein structure determination.

Purpose of the Study:

• To investigate the extent to which diverse and nonnative-like structural ensembles can accurately match experimental NOE distance upper bounds.
• To evaluate the reliability of comparing simulated ensembles with experimental NOE data for assessing force field quality or interpreting experimental results.
• To determine if matching observed NOE restraints alone is sufficient for validating simulation accuracy.

Main Methods:

• Generation of simulated molecular dynamics trajectories for protein systems (villin headpiece, lysozyme).
• Calculation of atom-atom distances and NOE intensities from simulated ensembles.
• Comparison of calculated NOE-derived distances with experimental NOE distance upper bounds.
• Analysis of agreement considering both experimentally observed and unobserved NOEs.

Main Results:

• Simulated ensembles of highly nonnative structures (unfolded villin, denatured lysozyme) showed good agreement with experimental NOE distance upper bounds.
• The villin headpiece unfolded ensemble, despite significant deviation from the native structure, closely matched experimental restraints.
• This apparent agreement is attributed to nonlinear averaging effects and selective use of observed NOE data.
• Including unobserved NOEs predicted by simulations significantly worsened the agreement with experimental data.

Conclusions:

• Comparing experimental NOE distance restraints with simulated ensembles alone provides limited information about simulation quality.
• The nonnative nature of simulated ensembles can lead to artificially good agreement with experimental data when only observed NOEs are considered.
• A comprehensive comparison, including predicted unobserved NOEs, is necessary for a more accurate assessment of simulation fidelity.
• Force field validation requires careful consideration of the ensemble's conformational diversity and comparison against the full spectrum of experimental data.