Effect of solvent model when probing protein dynamics with molecular dynamics

Area of Science:

• Biophysics
• Computational Biology
• Structural Biology

Background:

• Protein dynamics are crucial for function.
• Molecular dynamics (MD) simulations are key tools for studying these dynamics.
• Accurate water models are essential for reliable simulation results.

Purpose of the Study:

• To compare the performance of three different water models in simulating protein dynamics.
• To evaluate the accuracy of atomistic, coarse-grained, and implicit solvent models against experimental data.
• To assess the reliability of computational methods for predicting protein flexibility.

Main Methods:

• Utilized molecular dynamics simulations for Bpti and Galectin-3 proteins.
• Employed three water models: TIP4P-Ewald (atomistic), Elba (coarse-grained), and generalized Born (implicit).
• Quantified dynamics using model-free order parameters (S2) from backbone NH and side-chain bonds, compared with NMR relaxation data.

Main Results:

• All three water models showed good agreement with experimental backbone order parameters (average unsigned deviations 0.03-0.06).
• Side-chain dynamics predictions exhibited larger deviations (0.13-0.17) and significant variation between models (0.11-0.15).
• No single water model demonstrated superior performance across all assessed parameters.

Conclusions:

• Current water models provide reasonable accuracy for backbone protein dynamics.
• Predicting side-chain dynamics remains challenging, with notable discrepancies among different simulation approaches.
• Further investigation is needed to improve the accuracy of computational methods for assessing protein dynamics, particularly for side-chain flexibility.