Comparative Analysis of Caco-2 Cells and Human Jejunal and Duodenal Enteroid-derived Cells in Gel- and Membrane-Based Barrier Models of Intestinal Permeability

Studies of gut absorption are necessary for characterization of drug toxicokinetics. While the human colon carcinoma cell line Caco-2 is the most used in vitro model, additional options have been proposed to represent different intestinal segments. Accordingly, we characterized the morphology and tissue specific marker expression of three human intestinal cell types: Caco-2 and primary human enteroid cells from jejunum (J2) and duodenum (D109). Cells were cultured in the OrganoPlate® 3-lane 40 microphysiological system with flow (MPS) or static 24-well Transwells™. In both formats, J2 and D109 cells formed domedlike structures compared to uniform monolayers of Caco-2 cells. In the MPS, only Caco-2 cells formed uniform tubules. Cells grown on Transwells formed a thicker monolayer than in the MPS, with the most notable difference observed using J2 cells. All cells and culture conditions exhibited robust expression of ZO-1 (tight junctions). For cell polarization markers ezrin and villin, cell staining showed that expression was greatest in J2 and D109 cells cultured in the MPS. Regarding the functional marker mucin, J2 cells exhibited the highest expression, regardless of the platform. J2 and D109 cells exhibited poor barrier (70 kDa FITCTRITC-dextran) in MPS, but the tight barriers formed in Transwells. Barrier function and drug transport were also evaluated using caffeine, indomethacin, and propranolol; these compounds showed no appreciable binding to empty MPS or Transwells. Notably, we found that the gel lane in MPS was a sink for these compounds; only a small fraction (~13% over 15 hrs) crossed into the opposite channel, even without cells present. The permeability ratios among cell types in the MPS and Transwells were used to parameterize the probabilistic environmental compartmental absorption and transit model to determine whether in vitro data could reduce uncertainty in predictions. We found that the most accurate prediction of the fraction absorbed was achieved when using Transwell-derived data (Papp) from Caco-2 cells, combined with the experimentally-derived segment-specific (Caco-2, J2, and D109 data) absorption ratios. Overall, we show that marker staining was highest in enteroid-derived cells cultured in the MPS platform, but that studies of drug permeability across cell barriers in this model are challenging.