Cholesterol

Classical artificial membranes used for skin permeation studies are not suitable to study dextrose’s buccal absorption since they do not reflect the mucosal barrier. However, the efficacy of an in vitro procedure to predict drug buccal absorption using an artificial membrane impregnated with lipids has previously been described. Dextrose permeation was determined according to this previously reported procedure with minor modifications [22 (link)]. Briefly, a lipid mixture constituted of 4.7 g of 1-octanol (Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany), 0.15 g of phosphatidylcholine (Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany) and 0.15 g of cholesterol (Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany) was prepared in a beaker (octanol/phosphatidylcholine/cholesterol ratio 94%/3%/3% w/w/w). The mixture was placed under magnetic stirring for 1 h. The cellulose acetate–nitrate mixture membrane (0.025 µm MCE membrane^®, Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany) was weighed and impregnated by submersion for 1 h in the lipid mixture. Impregnated membranes were then interposed between two absorbing papers to eliminate the excess lipids, and were weighed again to determine the percentage of lipid impregnation. The membranes were used on the day of the experiment to avoid membrane drying or lipid degradation.
The permeation of 40% (w/v) dextrose gel was studied with this artificial membrane. Although this artificial membrane’s efficacy had already been proven and compared with porcine buccal mucosa with naproxen [22 (link)], and since paracetamol and caffeine can be administered through buccal mucosa, positive permeation controls were performed with compounded 14% (w/v) paracetamol gel and 20% (w/v) caffeine gel using the same gel formula used for dextrose gel compounding.
An in vitro permeation study was conducted using a Phoenix RDS Automated Diffusion platform (model DB-6 Manuel, Teledyne Hanson Research, Chatsworth, USA) equipped with twelve in-line vertical diffusion cells enabling simultaneous in vitro release experiments. Each 15 mL cell receptor compartment, Franz cell, was filled with simulated plasma consisting of a phosphate-buffer solution (0.8 g/L Na₂HPO₄, 0.15 g/L KH₂PO₄ and 9 g/L NaCl) at pH 7.4 for the experiments with dextrose and caffeine gels and at pH 5.8 for the paracetamol to ensure good enough solubility to maintain sink condition. The simulated plasma was maintained at 32 ± 1 °C by the dry heat block and continuously stirred at 200 rpm. For all 3 gels, approximately 1.15 g (1 mL) was uniformly spread on the dosage chamber with an extemporaneously impregnated membrane placed between the dosage chamber and the simulated plasma. The cell was sealed by placing a cover on top of the dosage chamber. At given time intervals (0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 6, 8 and 10 h), aliquots (0.5 mL) of simulated plasma were withdrawn from the cell receptor compartment through the sampling port and replaced by an equal volume of fresh pre-warmed simulated plasma, to keep the volume constant and sink conditions constant. The experiments were reproduced 8 times for the 3 different gels and the dilution effect was considered to determine the cumulative amounts of dextrose, caffeine and paracetamol. Their concentrations in each sample of withdrawn simulated plasma were determined.
The glucose contents were determined as previously described in this work. The direct determination of paracetamol content was performed at 243 nm using a UV–visible spectrophotometer (UV 2401PC, Shimadzu Scient. Inst., Colombia, SC, USA) according to a previously published method [42 ]. The caffeine contents were measured using the HPLC-UV method validated to be stability-indicating in our laboratory. Briefly, 10 µL of sample was injected into an automatic HPLC-UV-DAD apparatus (Dionex Ultimate 3000, Dionex Softron GmbH, Germering, Germany) with an RP18 column (1000 Å, 5 μm, 4 × 250 mm) (Lichrospher^®, Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany). Data analysis was performed using Chromeleon^® software (Version 7.2.8, Thermo Fisher Scientific, Waltham, USA). The mobile phase consisted of a mixture of ammonium acetate buffer adjusted to pH 4 (88%, v/v) and acetonitrile (12%, v/v), and the wavelength for caffeine detection was 274 nm. The permeated profiles of the gels were plotted as the cumulative amount of dextrose, paracetamol and caffeine diffused per unit area of membrane versus time. The flux (μg/cm²/h) and lag time (h) estimates were generated using the free Skin and Membrane Permeation Data Analysis (SAMPA) software (version 1.04) developed by Bezrouk et al. for skin and membrane permeation data analysis [43 (link)].

The preparation procedure and formulation parameters remained unchanged for all LNPs, while the nucleic acid cargo varied for the different studies. 16 LNPs with different barcoded ssDNA (IDT, Table S3, Supporting Information) were formulated for the biodistribution study, and LNPs with Cre mRNA (Etherna) were formulated for the functional delivery study. Lipid components were: DLin‐MC3‐DMA (chemically synthesized in house), cholesterol (C8667, Sigma‐Aldrich), DSPC (LP‐R4‐076, Corden Pharma) and DMPE‐PEG₂₀₀₀ (PM‐020CN, NOF Corporation). Prior to particle assembly, the nucleic acid was solubilized in 50 mM citrate buffer pH 3 (Q2445, Teknova), and the lipids were dissolved in ethanol and mixed with a molar ratio of 50:38.5:10:1.5 (MC3:cholesterol:DSPC:DMPE‐PEG₂₀₀₀). LNPs were prepared with a NanoAssemblr Ignite (Precision NanoSystems Inc.) by mixing the two solutions at a flow rate of 12 ml min⁻¹ with a 3:1 volume ratio (nucleic acid:lipid solution) to form LNPs with a lipid/nucleic acid weight ratio of 10:1. Particles were loaded into a Slide‐A‐Lyzer G2 dialysis cassette (Thermo Scientific), dialyzed in PBS pH 7.4 overnight, and then sterile filtered using a 0.2 µm filter.
The encapsulation efficiency and nucleic acid concentration were determined with either the RiboGreen (mRNA) or OliGreen (ssDNA) Assay (Thermo Fisher Scientific), according to the manufacturer's guidelines. Particle size was determined with dynamic light scattering using a DynaPro Plate Reader III (Wyatt Technology Corporation) for ssDNA LNPs, and Zetasizer Nano ZSP (Malvern Panalytical Ltd) for mRNA LNPs. Encapsulation efficiency for all LNPs was above 90% and particle size varied from 60 to 90 nm.