Protein structure prediction: do hydrogen bonding and water-mediated interactions suffice?

Area of Science:

• Computational Biology
• Biophysics
• Protein Structure Prediction

Background:

• Hydrogen bonds are crucial for stabilizing protein secondary structures like alpha-helices.
• Understanding the physics of hydrogen bond formation is key to accurate protein structure prediction.
• Previous models often rely on sequence-based information, potentially missing context-dependent interactions.

Purpose of the Study:

• To investigate the many-body physics of hydrogen bond formation in protein alpha-helices.
• To evaluate a simplified physics-based model (the "amnesiac" Hamiltonian) for predicting low-energy protein structures.
• To assess the impact of context-sensitive potentials and secondary structure information on prediction accuracy.

Main Methods:

• Utilized a context-sensitive hydrogen bond potential dependent on residue identity and solvent exposure.
• Employed molecular dynamics simulations within the Associated Memory Hamiltonian framework.
• Generated low-energy structures for three alpha-helical proteins using the "amnesiac" model.
• Compared simulated structures against X-ray crystal structures and results from the full Associated Memory Hamiltonian.

Main Results:

• The "amnesiac" model achieved moderate structural similarity (Q ≈ 0.4) to native structures when using predicted secondary structure information.
• Without secondary structure information, the "amnesiac" model yielded lower similarity (Q ≈ 0.3).
• Both "amnesiac" model results were less accurate than predictions using local-sequence structure matches.

Conclusions:

• Context-sensitive hydrogen bond potentials are important for modeling protein alpha-helices.
• Secondary structure information significantly improves structure prediction accuracy with simplified models.
• Physics-based models without sequence matching show potential but require further refinement for high accuracy.