A Strategy for in Silico Prediction of the Membrane Permeability of Drugs

Abstract

Abstract Parallel artificial membrane permeation assay (PAMPA) was performed for nine amine drugs and their permeability coefficient (log Ppampa) was determined at different pHs. The previously developed digital simulation method was successfully used to reproduce the sigmoid-like pH dependence of log Ppampa, and the distribution coefficient (log KD,M) to the lipid-containing dodecane membrane could be determined for all the amine drugs studied. The thus determined log KD,M values showed a linear free-energy relationship with the standard ion-transfer potential ($\Delta _{\text{O}}^{\text{W}}\phi ^{ \circ }$; R2 = 0.754) and the distribution coefficient (log KD; R2 = 0.891), which were previously determined by ion-transfer voltammetry with the 1,2-dichloroethane|water interface. On the other side, we can predict $\Delta _{\text{O}}^{\text{W}}\phi ^{ \circ }$ very accurately by a previously proposed solvation model called the “non-Bornian” model. These findings suggest a possible strategy for perfect in silico prediction of the membrane permeability of drugs.