Tuesday, January 13, 2009

Self-consistent solvation energy contribution calculation for protein-ligand complexes




Solvation energy is a major contribution to a ligand binding energy and is the interaction pretty much responsible for binding selectivity. Actual calculation of the solvation energy requires a method valid both for small molecule ligands and large proteins (and protein-ligand complexes).


Calculation of the electrostatic contributions for the binding energies in a continuous solvation energy approach imposes different problems for large and small molecules. Normally people use some kind of Generalized Born (GB) approximation. The latter is only exact for a charge in the center of a spherical cavity and thus can only be valid for a small molecule with most of the charge located within a few atoms.

If the molecule of interest is large, most of the charges are close to the molecular surface, instead. GB approximation in its most commonly accepted form fails next to a molecular surface: the Born radius is missed by a factor of 2. This means that there can be no "classic" GB model working good both for small and large molecules!

Binding affinity calculation requires calculation of differences between the energy of the complex (a large molecule) and the energies of the protein (another large molecule) and the ligand (a small molecule) at infinite separation.

If a GB model is made working by careful adjusting of "bare" Born radii to fit experimental IC50 of complexes, a good sanity check would require reproduction of experimentally known solvation energies of small molecules and ions (and the other way around). The two graphs in this post show, that this is indeed possible. A relatively large error in the small molecules solvation energies shows that although the resulting model is reasonable, the obtained GB parameters are only quantitatively transferable between large and small molecules.

No comments: