Optimized conditions for MDCK permeability and turbidimetric solubility studies using compounds representative of BCS classes I–IV

Abstract

The solubility enhancing effects of various excipients, including their compatibility with in vitro permeability (P_app) systems, was investigated using drugs representative of Biopharmaceutics Classification System (BCS) classes I–IV. Turbidimetric solubility determination using nephelometry and transport experiments using MDCK Strain I cell monolayers were employed. The highest usable concentration of each excipient [dimethyl sulfoxide (DMSO), ethanol, hydroxypropyl-β-cyclodextrin (HPCD), and sodium taurocholate] was determined by monitoring apical (AP) to basolateral (BL) [¹⁴C]mannitol apparent permeability (P_app) and the transepithelial electrical resistance (TEER) in transport experiments done at pH 6.0 and 7.4. The excipients were used in conjunction with compounds demonstrating relatively low aqueous solubility (amphotericin B, danazol, mefenamic acid, and phenytoin) in order to obtain a drug concentration >50 μM in the donor compartment. The addition of at least one of the selected excipients enhanced the solubility of the inherently poorly soluble compounds to >50 μM as determined via turbidimetric evaluation at pH 6.0 and 7.4. Ethanol and DMSO were found to be generally disruptive to the MDCK monolayer and were not nearly as useful as HPCD and sodium taurocholate. Sodium taurocholate (5 mM) was compatible with MDCK monolayers under all conditions investigated. Additionally, a novel in vitro system aimed at more accurately simulating in vivo conditions, i.e., a pH gradient (6.0 AP/7.4 BL), sodium taurocholate (5 mM, AP), and bovine serum albumin (0.25%, BL), was shown to generate more reliable P_app values for compounds that are poorly soluble and/or highly protein bound.

Introduction

Drugs intended for oral administration must be sufficiently hydrophilic to dissolve in gastrointestinal (GI) fluids; yet, they must also possess sufficient hydrophobic character to undergo passive transport across intestinal epithelia. It is generally accepted that transcellular passive diffusion is the most significant transport mechanism for intestinal drug absorption, and the rate of transport is determined by the physicochemical properties of the epithelium as well as the properties of the solute itself (Adson et al., 1995, Daugherty and Mrsny, 1999). In vitro models used to investigate drug transport, such as the Caco-2 and MDCK cell lines, differentiate into columnar epithelia and form tight junctions when cultured on semi-permeable membranes (Cereijido et al., 1978, Cho et al., 1989, Hidalgo et al., 1989). These types of cell monolayers are widely used in drug discovery settings in order to rank lead candidates with respect to their potential rate of absorption in vivo (Hidalgo et al., 1989, Artursson and Karlsson, 1991, Artursson and Borchardt, 1997, Irvine et al., 1999).

Caco-2 cells express several transporters, including P-glycoprotein, and interpretation of apparent permeability (P_app) data can in certain cases be fairly complex (Burton et al., 1993). In addition to their relatively long maturation time and subsequent high risk of infection, Caco-2 cultures consist of a heterogeneous mixture of subtypes and display a well-established lack of inter-experimental reproducibility (Hidalgo et al., 1989, Artursson and Borchardt, 1997). In this study, since only passive diffusion is considered, an absorption model without these complicating factors was preferred; thus, the MDCK Strain I cell line was used (Irvine et al., 1999). Caco-2 and MDCK cell lines demonstrate similar P_app values for passively absorbed drugs (Raub et al., 1993, Irvine et al., 1999), and a recent comparison of drug P_app vs. human intestinal absorption concluded that MDCK monolayers are in fact slightly more predictive of in vivo absorption vs. Caco-2 (Irvine et al., 1999). MDCK monolayers possess other advantages over Caco-2: a significantly shorter maturation time in culture, and thus less chance for infection to occur; relatively low expression of P-gp, if only passive transport is to be determined (Horio et al., 1989); and morphologic homogeneity, resulting in inter-laboratory reproducibility (Cho et al., 1989, Irvine et al., 1999). Two sub-types of MDCK cells are available: Strain I, which matures into tight monolayers with transepithelial electrical resistance (TEER) >1500 Ω cm²; and Strain II, which matures into monolayers displaying a TEER of 100–300 Ω cm² (Cho et al., 1989). MDCK Strain I cells can easily be grown on permeable polycarbonate supports and form a confluent, highly polarized monolayer within 5 days that mimics intestinal enterocytes and provides a convenient model for studying drug P_app (Cho et al., 1989, Raub et al., 1993, Irvine et al., 1999).

A drug’s aqueous solubility is a factor dictating its dissolution rate under given pH conditions and thus the amount of drug available for absorption (Amidon et al., 1995, Charman et al., 1997, Dressman et al., 1998). Low aqueous solubility at physiological pH values is therefore likely to result in poor absorption; consequently, even the presence of a high rate of permeation may not facilitate absorption if the compound possesses poor solubility. Conversely, a compound with high aqueous solubility might be well absorbed even if it demonstrates a moderate or low permeation rate (Artursson and Borchardt, 1997, Lipinski, 2000). Screening the aqueous solubility of large numbers of compounds necessitates a departure from traditional, time- and resource-consuming methods that accurately measure thermodynamic solubility (Lipinski, 2000). Nephelometry, which measures the point of precipitation at various pH values and under controlled temperatures, results in a reliable, quick, and reproducible screening system for accurately estimating solubility in the μg/ml range.

The Biopharmaceutics Classification System (BCS) is a recently developed guidance framework that classifies immediate release (IR) drug substances based on their aqueous solubility and transcellular P_app as a basis for establishing an in vitro–in vivo correlation (IVIVC), which can be extrapolated to estimate a drug’s rate and extent of GI absorption (Amidon et al., 1995, Fleisher et al., 1999). Relatively little information is available regarding the effects of various excipients or solvents on compound P_app when measured in vitro (Aungst et al., 2000). Thus, investigation of various buffer/excipient systems compatible with in vitro P_app experiments, using compounds representative of the four BCS classes, would assist in ascertaining which in vitro protocols provide the most reliable estimate of in vivo P_app for discovery-based compounds that could potentially fall into any of the four BCS classes.