Mechanistic studies on pH-permeability relationships: Impact of the membrane polar headgroup region on pKa

Abstract

Passive permeability through biological membranes requires partitioning of drug molecules into the lipid bilayer and subsequent permeation. Most drugs are weak acids or bases, making their ionization constants (pK_a) critical for predicting permeation across biological barriers. The pH-partition hypothesis posits that only the uncharged form contributes to passive permeability, suggesting a proportional relationship between permeability and uncharged fraction. However, experimental pH-permeability profiles are not accurately predicted with neutral fractions calculated using aqueous pK_a values. Interactions between charged solutes and phospholipids are expected to alter the pK_a of drugs within the membrane. In this study, we use modeling and simulation and experimental partitioning in a biphasic surrogate phospholipid membrane system, diacetyl phosphatidylcholine (DAcPC) and n-hexane, to study pH dependent permeability. Models were constructed in which pK_a values were either shifted or distributed around the aqueous pK_a and the resulting neutral fractions were compared to pH-dependent permeabilities. For acids, models with shifted or distributed pK_a values can explain pH-dependent permeabilities in Caco-2 cells, but these models were not predictive for bases. For partitioning studies, five probe drugs, two acidic (ketoprofen, tolbutamide), two basic (metoprolol, verapamil), and one neutral (diazepam), were partitioned between n-hexane and buffer or buffer-hydrated DAcPC at different pH values. The apparent pK_a values in the surrogate phospholipid system (C6/DAcPC) were shifted from their aqueous pK_a values. However, the resulting pK_a values did not predict observed pH-dependent Caco-2 permeabilities. Models that decrease the pH-pK_a difference improve permeability predictions for both bases and acids and use of a pK_a shift or distribution can further improve predictions for acids.