Human Duodenal Monolayers Link Barrier Robustness With Oral Drug Absorption Estimates
A Chugai-led study optimized organoid-derived intestinal monolayers across seven donors. Permeability correlated with human absorption for 19 marketed drugs, while estimates for metabolized drugs depended on an explicit expression-based CYP3A scaling assumption.
Choosing Culture Conditions for Both Barrier Function and Metabolism
Researchers at Chugai Pharmaceutical, Hokuto Hospital and Keio University developed a human duodenal organoid-derived monolayer for oral drug absorption studies. The starting crypts came from adjacent normal duodenal tissue removed during pancreaticoduodenectomy. Transferring organoid-derived cells to Transwell inserts gave experimental access to both the luminal and basal sides of an intestinal epithelial barrier.
The team optimized cell preparation, surface coating, expansion medium and differentiation medium. A central trade-off emerged: removing growth factors could increase markers of differentiation while weakening resistance to fasted-state simulated intestinal fluid, or FaSSIF. The selected condition retained epidermal growth factor during differentiation, favoring a stable barrier alongside drug-metabolizing activity.
The initial condition formed a confluent monolayer during the four-day expansion phase in only two of seven donor cultures. Adding lipid-rich albumin, AlbuMAX, improved monolayer formation across the donor set. Several fatty acids could increase barrier resistance, but the investigators did not identify which components accounted for the improvement in expansion. They also added 100 nM 1α,25-dihydroxyvitamin D3 (VD3) during the final two differentiation days to increase absorption-related gene expression and CYP3A activity.
The final schedule comprised four days of expansion and six days of differentiation. Microscopy showed polarized epithelial cells, tight junctions and microvilli, with some MUC2-positive goblet cells. These observations describe an epithelial monolayer derived from organoids, with a format suited to permeability measurements.
Transport and Metabolism Remained Distinct Contributions
Functional tests examined more than barrier integrity. Inhibitors reduced the directional transport of digoxin and sulfasalazine, supporting P-glycoprotein and BCRP efflux activity, respectively. Raloxifene formed glucuronide metabolites, and irinotecan generated SN-38; a carboxylesterase inhibitor reduced the latter response.
Midazolam measurements combined transport with CYP-mediated metabolism. Adding the pan-CYP inhibitor 1-aminobenzotriazole reduced metabolite formation and increased parent-drug permeability. Rifampicin induction produced the opposite pattern: increased CYP3A-related activity, less parent-drug transfer and more hydroxylated metabolite. Metabolites appeared on both sides of the insert, so basal collection alone would not describe all intestinal metabolism.
The model retained donor differences. Nineteen marketed drugs were tested in monolayers from four donors. Differences were more apparent among lower-permeability drugs, with oxacillin an exception, than among the high-permeability group. Inhibitor comparisons also linked donor variation in digoxin, sulfasalazine and midazolam handling to transporter or enzyme function.
Barrier reproducibility was assessed separately. Within one donor experiment, the FaSSIF/lucifer-yellow assay had a coefficient of variation of 8.6% across 24 biological replicates. The source also reports repeated experiments in two donor cultures. Most gene-expression measurements and repeated midazolam permeability/metabolite measurements remained within the stated 1.5-fold bands; these measures support assay consistency without erasing biological variation among donors.
Human Absorption Estimates Required a Metabolic Scaling Step
For 19 marketed drugs, permeability in the donor-1 monolayer correlated with reported human fraction absorbed, Fa, with R² = 0.88 after fitting the model’s regression parameter. This was a retrospective relationship using clinical values from the literature, rather than a prospective test of new candidate compounds.
The investigators then examined nine drugs comprising six CYP3A substrates, one P-glycoprotein substrate and two dual substrates. Their target was FaFg: the fraction absorbed multiplied by the fraction escaping intestinal metabolism. Applying the original permeability–Fa curve tended to overestimate FaFg for the CYP3A-containing groups. An alternative calculation using permeability with and without CYP inhibition brought only three of nine drugs within 1.5-fold of the clinical values and underestimated six.
A third calculation used the ratio of CYP3A4 messenger RNA expression in primary enterocytes to that in the donor-1 monolayer. The ratio was 0.24, and the analysis assumed that it also represented the relative metabolic activity in vivo and in vitro. With this adjustment, seven of nine estimates were within 1.5-fold of the reported human values.
Rifabutin remained overestimated by all three approaches. The authors discuss metabolism outside CYP3A as one possible explanation. The study therefore combines a characterized human epithelial assay with a conditional extrapolation model: its absorption estimates depend on the modeled transport assumptions, the expression-to-activity scaling and the reference clinical data. The tested marketed drugs do not establish performance for every new oral modality or the full range of patient physiology.