Genome-Edited Gut–Liver Chip Measures Drug Transfer and CYP Inhibition
Engineered human Caco-2 and HepG2 cells supported simultaneous measurement of parent drugs and metabolites in a two-channel chip. Itraconazole and bergamottin increased triazolam concentrations while reducing its CYP3A4 metabolite, with material sorption and assay timing limiting pharmacokinetic interpretation.
Linking an Intestinal Barrier With Metabolic Cells
Researchers at Ritsumeikan University, the Institute of Science Tokyo and Japan’s National Institute of Health Sciences developed a two-channel model for studying drug absorption and metabolism together. The system paired genome-edited human Caco-2 intestinal epithelial cells with genome-edited HepG2 liver cells. Both are established cell lines with engineered metabolic capacity, rather than primary intestinal or liver tissue.
Caco-2 cells occupied the upper channel and HepG2 cells the lower channel of a polydimethylsiloxane, or PDMS, device. The edited intestinal cells had changes in drug-metabolizing enzymes, while the HepG2 cells carried multiple cytochrome P450 and UGT1A1 knock-ins. Wild-type versions of both cell types formed a comparator chip.
The investigators first established that the two cell types could share the culture conditions. Barrier-marker staining and low Lucifer yellow permeability supported an intact intestinal monolayer. The characterization experiments used three devices. Co-culture did not significantly raise the measured drug-metabolizing gene expression relative to the corresponding monocultures, so the enhanced metabolism was associated with the engineered cells rather than an observed co-culture-induced maturation effect.
Comparing Parent Drugs and Metabolites in Both Channels
The drug experiments compared static conditions with rocking-driven flow. Rocking was applied during the absorption and metabolism assays, rather than throughout model formation. It generated bidirectional fluid movement. Samples from both channels were collected at two, four, 24 and 48 hours and analyzed by liquid chromatography–tandem mass spectrometry.
Six drugs covered different transport and metabolic behaviors: acetaminophen, caffeine, triazolam, diclofenac, bufuralol and propranolol. Parent-drug and metabolite comparisons generally used four devices per condition. Rocking increased delivery into the lower channel for several compounds. The authors suggest that improved mixing of stagnant medium contributed to this effect, rather than claiming that short assay-period flow had improved cellular differentiation.
The engineered chip produced higher concentrations of the measured CYP3A4, CYP2C9, CYP2D6 and CYP1A2 metabolites than the wild-type chip under flow. These included α-hydroxytriazolam, 4-hydroxydiclofenac, 1-hydroxybufuralol and desisopropylpropranolol. Parent-drug results were not uniformly different: lower-channel diclofenac concentrations, for example, did not differ significantly between wild-type and engineered chips.
Interpreting disappearance of a parent compound required a separate material control. Acetaminophen, caffeine and diclofenac showed little sorption to the PDMS device, whereas triazolam, bufuralol and propranolol showed substantial losses. For those compounds, a falling parent concentration could reflect both metabolism and uptake by the device. The source cautions that simply correcting parent concentration cannot reliably reconstruct all metabolite production.
An Inhibitor Challenge Retained the Expected Direction of Change
The CYP3A4 interaction experiment used triazolam as the substrate. The upper channel received 10 μM itraconazole or 10 μM bergamottin for 24 hours before the assay, followed by 10 μM triazolam. Bergamottin is a grapefruit constituent; the experiment tested that compound rather than whole grapefruit juice.
In three-device comparisons under flow, both inhibitors increased triazolam concentrations and decreased α-hydroxytriazolam concentrations in the upper and lower channels at 24 and 48 hours. The paired parent–metabolite pattern supports inhibition of CYP3A4-mediated metabolism within this model. It provides a more specific interpretation than parent-drug disappearance alone.
Several features still constrain translation to oral first-pass pharmacokinetics. The assay followed concentrations for up to 48 hours, considerably longer than the usual early absorption window, and the culture surface was only 0.22 cm². Rocking did not reproduce continuous unidirectional perfusion or precisely controlled shear stress. PDMS sorption remained compound dependent.
The cellular arrangement also had practical limits: higher HepG2 loading could clog the lower channel, and the intestinal cells formed a conventional monolayer without villus architecture. The major propranolol metabolite 4-hydroxypropranolol could not be measured in the serum-containing medium because of instability. The study therefore used a different propranolol metabolite while retaining the uncertainty over the full metabolic balance.