Peptide and Protein Structure Prediction with a Simplified Continuum Solvent Model.

Area of Science:

• Computational Chemistry
• Structural Biology
• Biophysics

Background:

• Accurate prediction of protein structures is crucial for understanding biological function.
• Continuum solvent models offer a computationally efficient approach to simulate biomolecular systems.
• Existing models require refinement to reliably sample native-like conformations.

Purpose of the Study:

• To simplify and parametrize a continuum solvent model based on screened Coulomb potentials.
• To accurately predict native-like structures of peptides and miniproteins using replica-exchange simulations.
• To refine model parameters for improved accuracy in protein structure prediction.

Main Methods:

• Iterative refinement of 11 parameter values using low-energy, native, and non-native structures.
• Replica-exchange simulations initiated from extended conformations.
• Utilizing the CHARMM22/CMAP force field with an additional ψ torsion term.
• Screening atom self-energy and interaction energies based on group type (charged, neutral hydrogen-bonding, other neutral).

Main Results:

• The centroid of the largest cluster from simulations predicted native structures with low root-mean-square deviation (RMSD) from experimental structures (e.g., 0.47 Å for Trp-zip2, 0.86 Å for MBH12).
• The model demonstrated transferability, as conformations of MBH12 and BS1 were excluded from refinement.
• Six parameters govern dielectric response, and four surface tension values approximate nonpolar effects.

Conclusions:

• The refined continuum solvent model (SCP18) effectively samples native-like peptide and miniprotein structures.
• The model's parameters support the importance of main-chain hydrogen bonds in protein folding.
• The developed model shows promise for accurate and efficient protein structure prediction in computational studies.