Simple models for nonpolar solvation: Parameterization and testing | JoVE Visualize

Area of Science:

Computational chemistry
Biomolecular modeling
Molecular dynamics simulations

Background:

Implicit solvent models are crucial for biomolecular simulations, with Generalized Born (GB) capturing polarity but neglecting nonpolar effects.
Current nonpolar models using surface area (SA) energy terms have limitations, including lack of realism and mathematical singularities.
Accurate modeling of nonpolar effects is essential for predicting solvation free energies, protein stability, and conformational changes.

Purpose of the Study:

To explore Lazaridis-Karplus (LK) gaussian energy density and dispersion (DI) energy terms as alternatives to surface area (SA) for nonpolar effects in implicit solvent models.
To parameterize and evaluate combinations of GB, SA, LK, and DI energy terms using experimental data for small molecules, protein stability, and protein loop structures.
To assess the performance of these models in molecular dynamics (MD) simulations of proteins.

Main Methods:

Parameterization of various combinations of Generalized Born (GB), Surface Area (SA), Lazaridis-Karplus (LK), and Dispersion (DI) energy terms.
Validation against experimental data: 62 small molecule solvation free energies, 387 protein stability changes from point mutations, and structures of 8 protein loops.
Molecular dynamics (MD) simulations of the Trpcage mini-protein and three other small proteins using optimized GBLK parameters.

Main Results:

Optimized models (GBLK, GBDILK) showed no performance loss compared to GBSA, with mean errors of 1.7 kcal/mol for stability changes and 2 Å for loop conformations.
The GBLK model performed poorly in MD simulations with general parameters but showed good performance when parameters were optimized specifically for MD.
LK and DI nonpolar terms demonstrated validity as alternatives to SA treatments across various applications.

Conclusions:

The Lazaridis-Karplus (LK) and dispersion (DI) energy terms are effective and realistic alternatives to surface area (SA) approaches for modeling nonpolar effects in implicit solvent models.
Optimized implicit solvent models, particularly GBLK, can accurately reproduce experimental data for solvation energies, protein stability, and loop structures.
Further refinement of parameters for specific applications like molecular dynamics (MD) is beneficial for achieving optimal performance in biomolecular simulations.