Computational analysis of PKA-balanol interactions

Protein kinases are important targets for designing therapeutic drugs. This paper illustrates a computational approach to extend the usefulness of a single protein-inhibitor structure in aiding the design of protein kinase inhibitors. Using the complex structure of the catalytic subunit of PKA (cPKA) and balanol as a guide, we have analyzed and compared the distribution of amino acid types near the protein-ligand interface for nearly 400 kinases. This analysis has identified a number of sites that are more variable in amino acid types among the kinases analyzed, and these are useful sites to consider in designing specific protein kinase inhibitors. On the other hand, we have found kinases whose protein-ligand interfaces are similar to that of the cPKA-balanol complex and balanol can be a useful lead compound for developing effective inhibitors for these kinases. Generally, this approach can help us discover new drug targets for an existing class of compounds that have already been well characterized pharmacologically. The relative significance of the charge/polarity of residues at the protein-ligand interface has been quantified by carrying out computational sensitivity analysis in which the charge/polarity of an atom or functional group was turned off/on, and the resulting effects on binding affinity have been examined. The binding affinity was estimated by using an implicit-solvent model in which the electrostatic contributions were obtained by solving the Poisson equation and the hydrophobic effects were accounted for by using surface-area dependent terms. The same sensitivity analysis approach was applied to the ligand balanol to develop a pharmacophoric model for searching new drug leads from small-molecule libraries. To help evaluate the binding affinity of designed inhibitors before they are made, we have developed a semiempirical approach to improve the predictive reliability of the implicit-solvent binding model.