Integrated Isogenic Human Induced Pluripotent Stem Cell–Based Liver and Heart Microphysiological Systems Predict Unsafe Drug–Drug Interaction

Three-dimensional (3D) microphysiological systems (MPSs) mimicking human organ function in vitro are an emerging alternative to conventional monolayer cell culture and animal models for drug development. Human induced pluripotent stem cells (hiPSCs) have the potential to capture the diversity of human genetics and provide an unlimited supply of cells. Combining hiPSCs with microfluidics technology in MPSs offers new perspectives for drug development. Here, the integration of a newly developed liver MPS with a cardiac MPS—both created with the same hiPSC line—to study drug–drug interaction (DDI) is reported. As a prominent example of clinically relevant DDI, the interaction of the arrhythmogenic gastroprokinetic cisapride with the fungicide ketoconazole was investigated. As seen in patients, metabolic conversion of cisapride to non-arrhythmogenic norcisapride in the liver MPS by the cytochrome P450 enzyme CYP3A4 was inhibited by ketoconazole, leading to arrhythmia in the cardiac MPS. These results establish integration of hiPSC-based liver and cardiac MPSs to facilitate screening for DDI, and thus drug efficacy and toxicity, isogenic in the same genetic background.


Supplementary Figure 2. Mass spectrometry analysis of cisapride absorption in PDMS.
Absorption of different concentrations of cisapride in PDMS in the liver MPS was analyzed to correct the concentrations used in the experiments. Ion abundances of the drug in stock samples and in efflux media was measured using LC-MS/MS and % remaining was calculated. One device was used per cisapride concentration of 10, 50, 100, 500, and 1,000 nM.

Supplementary Figure 4. Oxygen consumption rate (OCR) of hiPSC-Heps.
Direct measurement of OCR was made by sequential exposure of hiPSC-Heps to 1 μM oligomycin (1O1F and 1O2F) or 2 μM oligomycin (2O1F and 2O2F), 1 and 2 μM carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP) and rotenone/antimycin A. Basal respiration was measured for 20 min. Oligomycin was injected at 20 min to block ATP synthase (complex V), thereby leading to a decrease in electron flow through the electron transfer chain (ETC) and the observed decrease in OCR. FCCP is an uncoupling agent disrupting the mitochondrial membrane potential. FCCP was injected at 50 min and its disinhibition of the electron flow through the ETC led to the determined maximal OCR. At 80 min a combination of rotenone (complex I inhibitor) and antimycin A (complex III inhibitor) was injected to shut down mitochondrial respiration.

Bile Efflux Testing
hiPSC-Heps were tested for normal canalicular efflux of CMFDA at day 7. CMFDA (Life Technologies) was prepared at 4 μM in HCM and added to the liver MPS. The devices were removed from the perfusion pump and imaging was performed using a Nikon Eclipse microscope for collection of 488/525 nm (ex/em) after 1-hour exposure with 30-min intervals over 1 h.

Liver MPS COMSOL Multiphysics® model
The liver MPS COMSOL® model comprising PDMS slab containing media channel and cell chamber separated by a porous PET membrane (Supplementary Fig. 3A). By symmetry about the vertical center plane, the computational domain was half the physical domain. The cell chamber base was flush with base of 3.5 mm thick PDMS slab. The top of the PDMS slab was exposed to ambient air with specified oxygen (O2) concentration. The base of the device (base of cell chamber and PDMS) was assumed O2 impermeable. The O2 flux through the PDMS slab's vertical sides was neglected, which was a good assumption since these sides were sufficiently far away from the media channel and cell chamber. The liver MPS model setup of cell chamber, media channel, and membrane portions of computational domain are shown in Supplementary Fig. 3B. Media channel and cell chamber heights were 100 µm and the membrane thickness was 15 µm. Flow enters the inlet at 20 µL/h and exits at the outlet. Laminar inflow and outflow conditions were used in COMSOL®. The membrane was modeled as a 2D planar surface between the media channel and cell chamber, with effective diffusivities for O2 and small molecules. Fluid flow across the membrane was neglected, and zero flow velocity was assumed throughout the cell chamber. No-slip conditions were imposed at all walls and membrane surfaces within the medial channel. For O2 transport simulations, cells were assumed to fill the cell chamber volume at a specified density and consume O2 at a specified rate (see main text for details). Liver MPS model mesh of the cell chamber, media channel, and membrane portions of computational domain are shown in Supplementary Fig. 3C. Mapped meshes were used to control the axial and lateral mesh resolutions on the straight and curved portions of the media channel. Triangular mesh elements were used at connection points, and surface meshes were then swept downward through the thicknesses of the media channel and cell chamber (5 elements through each). Mesh sensitivity analysis was performed to show that this mesh resolution yields sufficiently accurate numerical results. Quadratic finite elements were used for the flow velocity components and species concentration, while linear elements were used for fluid pressure. For shear stress calculations on the membrane (media channel side) at 20 µL/h, no-slip conditions were assumed at all walls as well as along the membrane (Supplementary Fig. 3D). O2 saturation in PDMS was determined at 24 h for a flow rate of 20 µL/h (Supplementary Fig. 3E).

Cardiac MPS preparation
The cardiac MPS was prepared as described in detail in a separate publication (Mathur et al., 2015;Huebsch et al., 2020). In brief, standard soft lithography was used to prepare the microfluidic devices from PDMS casted onto patterned SU-8 wafers. PDMS blocks featuring the microchannels were permanently bonded to glass substrates after oxygen plasma surface activation (24s, 21W, 0.59 Torr).
hiPSCs were differentiated into cardiomyocytes and isogenic stromal cells using previously published protocols (Huebsch et al., 2020). In order to mimic the native cellular composition of the human heart (Naito et al., 2006), 80% hiPSC-derived cardiomyocytes were mixed with 20% hiPSC-derived stromal cells and loaded into the cell chambers (20,000 cells/chamber). The cardiac MPS was cultured in RPMI 1640 medium (11875-093; Gibco) supplemented with 2% B-27 (17504-044; Gibco). A detailed characterization of the cells and microtissues can be found in a separate study (Huebsch et al., 2020).