238 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS (2,3) or dimethylpolysiloxane membranes (4-7) revealed that the nature of the materials comprising such membranes limits their use as a skin model system. A few studies described the formation of membranes by using the corneocytes and lipid extracts of porcine stratum corneum (8,9). Although reaggregation of stratum corneum compo- nents resulted in viable membranes, such an approach would not obviate the use of animal skin. Abraham and Downing (10) reported the in vitro formation of membranes by use of a mixture of epidermal ceramides, cholesterol, cholesterol sulfate, and fatty acids. Although the authors demonstrated that water transport across their model membrane was similar to that across the stratum corneum, the system suffers the constraint of requiring epidermal ceramides that involve difficult and time-consuming extraction and isolation procedures. Further, no permeation studies were reported with permeants other than water. Indeed, most of the studies on model membranes prepared from lipid mixtures have been solely tested by water transport characteristics (8-11). This study reports the preparation of model membranes from a mixture of synthetic lipids and investigates the permeability characteristics of a variety of markers with a range of physicochemical properties across this model membrane. The model mem- branes prepared possessed sufficient cohesive properties so as to be easily handled when employed in diffusion studies wherein the membrane was in intimate contact with solutions in the receiver compartment. MATERIALS AND METHODS MATERIALS Palmitic acid, stearic acid, ceramides type III from bovine brain, cholesterol, cholesterol sulfate, N-palrnitoyl-DL-dihydrogalactocerebroside, N-2-hydroxyethylpiperazine-N'- 2-ethanesulfonic acid (HEPES free acid), and bovine serum albumin (BSA) were pur- chased from Sigma Chemical Co. (St. Louis, MO). The lipids were of the highest purity available and were used as received. Carnauba fatty acid was a gift from Dr. D. T. Downing, Marshall Dermatology Research Laboratories, University of Iowa. The chain lengths of fatty acid components of this mixture have been reported to be in the range C24-C3o (12). The radioactive marker, [3H]-cortisol (specific activity 134 mCi/mg) was obtained ß . ß •4 .... from ICN Rad•ochermcals (Irwne, CA). [ C]-sucrose (specific acuwty 204 •Ci/rng), [•4C]-estradiol (specific activity 204 •Ci/mg), and [3H]-progesterone (spe- cific activity 151 mCi/mg) were purchased from Amersham (Arlington Heights, IL). All other chemicals were reagent grade, and all solvents used were of HPLC grade. Water was double-distilled and deionized using a Milli-Q ion-exchange system. METHODS Preparation of liposomes Liposomes were prepared by the reverse phase evaporation method by using a cera- mide:cholesterol:fatty acid:cholesterol sulfate mixture at a weight ratio of 37.7:28.3: 17.0:17.0. This ratio was found to exhibit the strongest interlipid interaction deter- mined by monolayer studies (1). Briefly, the lipid mixture contained in a one-liter

ARTIFICIAL MEMBRANES 239 round-bottomed flask was dissolved in a chloroform-methanol solvent mixture (2:1 by volume). HEPES buffer (0.05 M), pH 7.5, was then added to the organic solution. The ratio of the organic solution to the buffer solution was 2:1 (v/v). The mixture, which was opalescent, became clear after 18 minutes sonication (Branson sonicator, E-module, Shelton, CT). The organic solvents were then removed by a rotary evaporator at 55øC for palmitic acid systems and 70-72øC for the stearic or carnauba acid systems. After solvent removal was complete, the volume of suspension was adjusted with buffer to obtain a suspension containing 30 mg/ml lipid. The liposomes were then annealed for 30 min at the appropriate elevated temperature. The liposomal suspension was then filtered through a 12-p•m pore size Nucleopore TM (Pleasanton, CA) filter to remove any particulate lipids. The filtered suspension was then used immediately to prepare the model membranes. Preparation of membranes Membranes without BSA treatment. One ml of the filtered 30 mg/ml liposomal suspension was diluted threefold with 2 ml of the buffer and extruded through a 2-Ix pore size Nucleopore filter for non-BSA treated membranes and through a 0.45-•m pore size Nylaflo filter (Gelman Sciences, Ann Arbor, MI) for BSA-treated membranes in an extruder (Lipex Biomembranes, Vancouver, BC, Canada). Extrusion was carried out by immersing the entire extruder assembly in a water bath maintained at 40øC, a temper- ature well below the phase-transition temperature of the lipid mixture. The nitrogen pressure used was approximately 40 psi. At the end of the extrusion process, typically an hour, the extruder assembly was dismantled and the lipid-covered nylon filter was carefully retrieved and dried in an oven at 60øC. The dried membrane was then weighed to ascertain the amount of liposomal lipid retained on the filter. For studies of water vapor transmission rates and electron microscopy, the dried mem- brane was treated successively with 2 ml of 3 mM, 5 mM, and 8 mM CaCI 2 solutions in the buffer. Each treatment was separated by 2-hr intervals. The treated membrane was then dried in vacuo for at least 24 hr before use. For diffusion experiments, the lipid filter following extrusion and drying was mounted on a Franz diffusion cell (Crown Glass Co., Somerville, NJ) with a nominal diameter of 2 cm and a receiver capacity of approximately 13 mi. The membrane was placed, lipid side up, on a 0.02-mm thick silastic (Dow Chemical, Midland, MI) cut to O-ring shape, with the inner and outer diameters matching those of the diffusion cell. This was done to ensure a proper seal between the donor cap and the receiver compartment. The donor cap was then placed carefully on the cell and clamped tightly with adjustable clamps. The receiver compartment was left empty and maintained at 37øC. Three ml of 10 mM CaCI 2 solution in pH 7.5 HEPES buffer were then added to the donor and the system was allowed to dry over a period of five days before testing with marker solutions. BSA-treated membranes. The lipid-covered nylon filter, following extrusion and drying, was placed carefully on a glass slide with the lipid side up. Approximately 3 ml of BSA in buffer was then gently added drop-wise to the filter and soaked for 10 min to allow the BSA to spread over the membrane. The filter was then heated at 80øC in an oven for one hour to denature the BSA. This treatment cycle was repeated twice to ensure complete coverage of the filter surface. After the BSA treatment protocol, the surface of the filter membrane exhibited a glossy appearance, indicating the presence of the