Drug discovery and development is a very expensive and long process. From target identification until marketing, approximately $1.3 billion and 12 years pass by (Wouters et al. 2020). Before entering the clinical phase, a drug candidate has to pass a broad battery of in silico-, in vitro-, and in vivo-based tests to determine its efficacy, metabolism, pharmacokinetic toxicity, and safety pharmacology (DiMasi 2001). Nevertheless, nine out of ten candidates fail in the clinical phases due to unexpected low efficacy or intolerable adverse effects (Harrison 2016). The most frequent cause for drug failures in clinical trials and for post-approval withdrawals in the United States and Europe is drug-induced liver injury (DILI) (Walker et al. 2020; Ostapowicz et al. 2002; Larry and Pageauz 2005)

The human liver is responsible for managing the endogenous and exogenous metabolism in the body, including glucose homeostasis, xenobiotic metabolism, and detoxification (Xu et al. 2005). Due to its location, blood flow, and functional role, it is the first organ, after the gastrointestinal system, to be confronted with enterally applied xenobiotics. Thus, it is also the initial target for medication-induced damage (David and Hamilton 2010). However, no biomarker or model exists for the reliable prediction of DILI in preclinical studies.

The current methods of predicting DILI are animal studies and in vitro hepatotoxicity testing, which often do not translate to humans (Godoy et al. 2013; Lauschke et al. 2016; Lauschke et al. 2019; Bell et al. 2017). Animals are of limited use to predict the response in humans because they differ from humans in several respects, such as the expression of metabolizing enzymes (Lossi et al. 2016). The gold standard of in vitro hepatotoxicity testing is 2D monoculture of primary human hepatocytes (PHH) (Lauschke et al. 2019; Lauschke et al. 2016; Godoy et al. 2013). In this model, PHHs grow flat on optimized plastic surfaces under static conditions, without contact with other liver cell types. This leads to a rapid loss of hepatocyte physiological functions, such as phase I and II enzyme and clearance activities, within 48–72 hr of culture (Godoy et al. 2013).

Given the shortfalls of animal testing and 2D monoculture, there is an urgent need for new, more predictive non-animal hepatotoxicity testing methods. Novel methods for the early assessment of drug toxicities could save money, resources, and working time. In addition, a reliable in vitro hepatotoxicity testing model would reduce the number of animal experiments within the framework of the 3R (Replacement, Reduction, and Refinement) animal welfare principles and, above all, save human lives.

Advanced 3D in vitro systems, such as liver organoids, are described as a very promising tool to better predict the efficacy and toxicity of preclinical drug candidates in humans, as they retain hepatocyte function long-term. Human induced pluripotent stem cells (iPSCs) could provide a suitable cell source with unlimited supply for the establishment of these advanced in vitro cell models (Robinton and Daley 2013). IPSCs have the capability to differentiate into nearly every cell type, organize into organ-specific architecture, self-renew, and self-organize, making them an attractive option for studying the reaction of human organs to xenobiotics in preclinical phases (Inoue et al. 2014).

In this work, liver organoids were differentiated from ChiPSC18 cells cultivated with the Cellartis DEF-CS 500 Culture System—a robust media for efficient expansion of human iPSCs in a feeder-free and defined environment. The differentiation was based on the publication of Wang et al. (2019) and others (Wang et al. 2018; Wu et al. 2019; Gieseck et al. 2014; Meier et al. 2017; Gerbal-Chaloin et al. 2014; Olgasi et al. 2020).

Differentiation of ChiPSCs into liver organoids

Two differentiation protocols were tested for the generation of liver spheroids from ChiPSC18 cells (Figure 1). For both methods, ChiPSC18 cells were thawed, cultured, and embedded into a Matrigel dome. In Method 1, cells were pre-cultivated for 5 days in DEF-CS medium to regenerate after the detachment and seeding stress prior to embedding (Figure 1). In Method 2, cells were allowed to form embryoid bodies (EB) in U-bottomed plates prior to embedding (Figure 1). For cells cultured using Method 1, there was an even distribution of single cells in the Matrigel dome (Figure 2, Panel A). For cells cultured using Method 2, the EBs appear as clumps (Figure 2, Panel B).

Figure 1. Workflow of the generation of liver organoids from human iPSCs. In Method 1, single iPSCs were pre-cultivated for 5 days prior to embedding into a Matrigel dome (Panel A). In Method 2, embryoid bodies (EBs) were formed in ultra-low attachment U-bottom plates prior to embedding (Panel B). The three stages of differentiation are endodermal induction, hepatic progenitor cell expansion, and hepatocyte maturation. (bFGF: basic fibroblast growth factor; HGF: human growth factor; OSM: oncostatin M, DEX: dexamethasone)

After embedding into the Matrigel dome, the organoids were imaged during the differentiation process (Figure 2, Panel C). The method 1 embedded iPSCs show spheroid-like cell aggregates immediately after endodermal induction with Activin A (Figure 2, Panel C). The progress of differentiation is evident in the formation of liver organoids having an outer and an inner circle (Gomez-Mariano et al. 2020; Mun et al. 2020). These clear structures can also be recognized in the Method 2 embedded iPSCs; however, a 2D outgrowth from the organoid center was observed after endodermal induction (Figure 2, Panel C). We suspect that the EBs were too heavy and fell to the bottom of the Matrigel dome before it solidified. Once on the bottom, cells then proliferated along the plastic surface in 2D.

Figure 2. Three-stage liver organoid differentiation. ChiPSC18 cells were embedded into a Matrigel dome as single cells (Panel A) or embryoid bodies (EBs; Panel B) before starting the differentiation. Differentiation comprises endodermal induction, generation of hepatic progenitor cells, and hepatocyte maturation (Panel C).

Localization and distribution of cell-specific structures

The differentiation of the iPSCs was traced by examining the expression and secretion of cell-specific markers. iPSCs embedded using Method 2 showed clear expression of pluripotent marker OCT4 and the endodermal marker SOX17 (Figure 3, Panel A) 5 days post-embedding. The OCT4-positive cells were distributed toward the center of the organoid, whereas the SOX17-positive cells were in the outer part of the cell aggregates (Figure 3, Panel A). Fluorescence images of Method 2 liver organoids showed expression of liver-specific markers albumin, HNF4α, MDR1, MRP2, and CYP3A4 20 days post-embedding (Figure 3, Panel B). Albumin, HNF4α, and CYP3A4 were distributed throughout the whole organoid, demonstrating cellular homogeneity (Figure 3, Panel B). These results are consistent with previous studies showing that HNF4α expression is induced by dexamethasone, which was added to the media on Day 10 (Figure 1) (Michalopoulos et al. 2003). In addition, MDR1 was localized to the outer part of the organoid, while MRP2 was localized toward the center of the organoid structure (Figure 2, Panel B). Expression of these phase III transporters suggests the formation of bile canaliculi-like structures.

Figure 3. Expression of specific features during liver organoid differentiation. Immunofluorescence of Method 2 liver organoids for the pluripotent marker OCT4 and the endodermal marker SOX17 5 days post-embedding (Panel A). Immunofluorescence of Method 2 liver organoids for hepatocyte markers (albumin and HNF4α), liver phase III transporters (MDR1 and MRP2), and the phase I enzyme CYP3A4 20 days post-embedding (Panel B). Live-cell imaging of the functional bile canalicular system of Method 1 liver organoids 20 days post-embedding (Panel C).

In order to determine the functionality of the bile canaliculi-like structures, the ability of the Method 1 liver organoids to metabolize CMFDA was investigated at Day 20 (Figure 3, Panel C). The fluorogenic substrate CMFDA actively passes through the cell membrane of hepatocytes and is degraded to produce fluorescein, which accumulates in biliary tissue. Accumulation of fluorescein in a tract-like manner was observed in the Method 1 liver organoids at Day 20, suggesting functional bile canaliculi-like structure (Figure 3, Panel C).

Gene expression profile during the differentiation process

To examine the progress of liver organoid differentiation, the expression of genes specific to liver organoid differentiation state were assayed at Days 0, 5, 10, 15, and 20. The expression of the pluripotent marker NANOG decreased from Day 0 to 20 (Figure 4, Panel A). As expected, gene expression of endodermal markers FOX2 and SOX17 increased after induction of endoderm formation by the addition of Activin A and decreased after the addition of human growth factor (HGF) and basic fibroblast growth factor (bFGF) after Day 10 (Figure 4, Panel B). The addition of HGF and bFGF promotes the development of the endodermal spheres into the hepatic progenitor population (Olgasi et al. 2020; Michalopoulos et al. 2003). Hepatocyte maturation was then induced by the addition of oncostatin M (OSM) and dexamethasone (DEX) at Day 10 (Oh et al. 2006; Kamiya et al. 2001; Figure 1). Gene expression levels of the hepatic progenitor-specific gene AFP and hepatocyte-specific gene ALB increased steadily from Day 10 to Day 20 (Figure 4, Panel C). In addition, gene expression of stellate cell-specific gene ACTA2 and cholangiocyte-specific genes CK7 and CK19 were increased on Days 10, 15, and 20 (Figures 4, Panels D and E). These data confirm the successful generation of liver organoids of a mixed population of hepatocytes, cholangiocytes, and stellate cells.

Figure 4. Relative gene expression of liver organoids during the differentiation process. IPSCs were expanded and embedded into a Matrigel dome. Cells were harvested at days 0, 5, 10, 15, and 20 for all methods. Undifferentiated iPSCs served as control. Graphs show the relative mRNA level of the pluripotent marker NANOG (Panel A), endodermal markers FOXA2 and SOX17 (Panel B), hepatic progenitor AFP and hepatocyte marker ALB (Panel C), Stellate cell marker ACTA2 (Panel D), and cholangiocyte markers CK7 and CK19 (Panel E). The expression was normalized to housekeeping genes PPIB, POLR2A, PPIA, EIF4E2, and B2M. n = 1.

Functional characterization of liver organoids

The functional properties of the liver organoids during the differentiation process were also examined. Levels of secretion of albumin (ALB) and αGST were measured, as mature hepatocytes secrete albumin and αGST into the extracellular space (Mun et al. 2020). On Day 25, the level of ALB in the organoid supernatant was 600 ng/ml, which is comparable to secretion levels of primary human hepatocytes (Lang et al. 2011; Bale et al. 2015) (Figure 5, Panel A). In addition, 100 ng/ml αGST was detected in the organoid supernatant at Days 20 and 25 (Figure 5A).

We tested the ability of liver organoids to secrete alanine aminotransferase (ALT) and aspartate aminotransferase (AST), as these markers are commonly used to detect DILI during preclinical safety studies and clinical trials. In addition, the measured ratio of ALT/AST is a biomarker for liver damage. AST activity was detected at Day 10 and increased to over 20 nmol/ml/day on Days 15, 20, and 25, while ALT activity was detectable on Day 25 (Figure 5, Panel B). These data suggest that liver organoids produced using these differentiation protocols can be used for toxicity studies.

Figure 5. Secretion and gene expression in liver organoids. The secretion of ALB, αGST, ALT, and AST during liver organoid differentiation. Supernatant was collected at days 5, 10, 15, 20, and 25 of differentiation. Panel A. Concentration of ALB and αGST. Panel B. ALT and AST activity. Differentiated liver organoids were treated with 750 µM metformin (MET), 30 µM bosentan (BOS), 100 µM acetaminophen (APAP) or 25 µM rifampicin (RIF). Media samples and cells were collected after 72 hours. Panel C. The viability of the cells was analyzed by measuring LDH released into the supernatant. The LDH release was normalized to the vehicle controls dimethylsulfoxide (0.2% DMSO; for RIF, BOS, and APAP) and untreated (for MET). Fold change in gene expression of ALB (Panel D) and CYP3A4 and CYP1A1 (Panel E). Gene expression was first normalized to the housekeeping genes PPIB, POLR2A, PPIA, EIF4E2, and B2M. Then, normalized gene expression was compared to vehicle control samples (0.2% DMSO; for samples treated with BOS, APAP, or RIG) or untreated samples (for samples treated with MET).

To further characterize the utility of these liver organoids as a predictive model for DILI, we treated organoids with three drugs known to cause liver damage (bosentan, acetaminophen, and rifampicin) and assayed for release of LDH into the supernatant, which is an indicator of cell injury. Metformin was used as a negative control, as it has no known toxic effect on hepatocytes. Treatment of the organoids with acetaminophen and bosentan lead to increased levels of LDH in the supernatant (Figure 5, Panel C). Interestingly, rifampicin-treated organoids secreted lower amounts of LDH into the supernatant than untreated cells or the vehicle controls (Figure 5, Panel C).

Although ALB expression and albumin secretion are markers for mature, functional hepatocytes, abnormal secretion is associated with liver damage (Foster et al. 2019). Relative levels of ALB gene expression were determined for liver organoids treated with bosentan, acetaminophen, and rifampicin, while metformin was used as a negative control. All drug-treated organoids did not show increased expression of ALB compared to the untreated, vehicle, and metformin controls (Figure 5, Panel D). In addition, we tested the relative gene expression of CYP3A4 and CYP1A1, as CYP450 enzymes are essential for drug metabolism in the liver (Klomp et al. 2020). The bosentan-treated organoids showed increased relative CYP3A4 expression over the untreated, vehicle, and metformin controls (Figure 5, Panel E). Previous studies have shown that both bosentan and rifampicin are inducers of CYP3A4 expression (Sa-ngiamsuntorn et al. 2011). However, it is worth noting that a concentration of 40 µM of rifampicin was used to induce CYP3A4 expression in primary human hepatocytes, and we only used a concentration of 25 µM. Relative CYP1A1 expression was decreased in bosentan-treated organoids, but increased in metformin-, rifampicin-, and acetaminophen-treated organoids (Figure 5, Panel E). These results indicate that the functionality of liver organoids is different than the response of PHH.

In vitro human cell lines, such as HepG2, have only partial primary cell functions, and 2D primary liver cells lose their physiological functions rapidly. Hence, 3D iPSC-derived liver models are a promising tool for studying the reaction of the human liver to xenobiotics in preclinical phases, especially for prolonged drug exposure (Inoue et al. 2014). In this work, ChiPSC18 cells were successfully differentiated into liver organoids using a three-stage differentiation protocol. Differentiated liver organoids show a typical morphological shape and important hepatic characteristics, including the expression of cell-specific markers and secretion of albumin. Additionally, the liver enzymes αGST, ALT, and AST could be detected. The liver-like reaction to xenobiotics and the ability for long-term studies make liver organoids a robust model for the early assessment and mechanistic understanding of DILI.

Cell culture

Cryopreserved Cellartis human ChiPSC18 cells (Cat. # Y00300, sold as part of Cat. # Y00305; Lot AK20001S) were thawed, plated, and cultivated using the Cellartis DEF-CS 500 Culture System according to the user manual.

ChiPSC18 differentiation

For method 1, dissociated ChiPSC18 cells were resuspended in thawed Matrigel on ice. 25 µl of the cell suspension was pipetted into the middle of a 24-well plate. The plate was incubated for 10 min at 37°C to make the Matrigel solid. For method 2, dissociated ChiPSC18 cells were seeded into an ultra-low attachment U-bottom plate for EB formation. As soon as EBs could be observed, aggregates were harvested and embedded into a Matrigel dome according to method 1.

Cells in solid Matrigel were cultivated in RPMI-1640 medium (Gibco) supplemented with 1% KnockoutTM SR (Gibco), 1% B27 supplement (Gibco), 1% GlutaMAX (Gibco), and 1% penicillin/streptomycin. For the first 5 days, 10 ng/ml Activin A (PeproTech) was added to the medium to induce endodermal differentiation. After 5 days, the medium was supplemented with 100 ng/ml bFGF (PeproTech) and 100 ng/ml HGF (PeproTech) for hepatic progenitor differentiation. From day 10 on, cells were cultivated in Williams E medium supplemented with 1% GlutaMAX (Gibco), 1% penicillin/streptomycin, 4% FBS, 10 µM DEX, 0.05 mg/ml ascorbic acid (Sigma-Aldrich), 1% ITS+ (Corning) and 10 ng/ml OSM (STEMCELL) to promote hepatocyte maturation. Cells were cultivated long-term without OSM.

Gene expression analysis

For gene expression analysis, the QuantiGene Plex Gene Expression Assay kit (Invitrogen) was used. A custom-designed bead-based plex panel was created to include pluripotent, endodermal, and liver-specific cell markers. B2M, EIF4F, PPIB, PPIA, and POLR2A served as housekeeping genes. Cultivated iPSCs and differentiated liver organoids were lysed on days 0, 10, 15, 20, and 25 according to the user manual. Hybridization and signal amplification was conducted by shaking incubation with 100 µl of working bead mixture for 18–22 hr (54 ± 1°C, 600 rpm), followed by shaking incubation with 100 µl of preamplifier, amplifier, biotinylated label probe, and binding with SAPE (1 hr, 50 ± 1°C, 600 rpm). Between each new step, magnetic beads were washed three times with wash buffer. The amount of RNA for input was quantified using a Luminex FLEXMAP 3D instrument.

Immunofluorescence

Organoids were fixed by a 30-min incubation with 4% formaldehyde at RT. Subsequently, cells were permeabilized with 1% Triton-X100 in PBS (-/-) for 30 min at RT and blocked with 5% BSA in PBS (-/-) at 4°C overnight. The next day, organoids were incubated with primary antibodies in antibody buffer (2% BSA, 0.1% Triton X-100 in PBS (-/-)), at 4°C overnight, at the following dilutions: anti-OCT4 (1:250, Abcam), anti-SOX17 (1:250, Abcam), anti-Albumin (1:500, Santa Cruz Biotechnology), anti-HNF4α (1:500, Sigma-Aldrich), anti-MRP2 (1:250, Santa Cruz Biotechnology), and anti-MDR1 (1:250, Santa Cruz Biotechnology). After incubation with the primary antibody, organoids were washed three times with 0.01% Triton X-100 in PBS (-/-). Then, organoids were incubated with secondary antibodies in antibody buffer for 3 hr at RT at following dilutions: goat-anti-mouse AF647 (1:1000, Abcam), donkey-anti-rabbit AF488 (1:1000, Abcam), and Hoechst 33342 (1:1000). After incubation, organoids were washed again three times with 0.01% Triton X-100 in PBS (-/-) and imaged in PBS (-/-) by confocal microscopy.

CDFDA staining

Live-cell staining was performed using CellTracker Green CMFDA Dye (Invitrogen). On day 25, the organoid co-culture medium was replaced with medium containing 1:1000 CellTracker Green CMFDA and 1:1000 Hoechst 33342. After incubation in the dark (37 °C, 45 min), cells were washed three times with PBS (-/-) and imaged by confocal microscopy.

ELISA of secreted hepatic proteins

Samples were collected on days 0, 5, 10, 15, and 20. Albumin secretion was measured using the Albumin SimpleStep ELISA Kit (Abcam) according to the manufacturer's protocol. To determine the αGST secretion, the Human Alpha GST ELISA (TECO) was used. The amount of albumin and αGST was calculated through extrapolation from the generated standard curves.

ALT and AST assays

Samples were collected on days 0, 5, 10, 15, and 20. ALT activity was measured using an ALT Activity Assay (Sigma-Aldrich) and AST activity was measured using an AST Activity Assay (Sigma-Aldrich). Assays were performed according to the manufacturer's protocol.

Cell viability assay

Samples were collected on days 0, 5, 10, 15, and 20 and an LDH-Glo™ assay (Promega) was performed according to the manufacturer's protocol. 

Bale, S. S. et al. Long-term coculture strategies for primary hepatocytes and liver sinusoidal endothelial cells. Tissue Eng. Part C. Methods 21, 413–422 (2015).

Bell, C. C. et al. Transcriptional, functional, and mechanistic comparisons of stem cell-derived hepatocytes, HepaRG cells, and three-dimensional human hepatocyte spheroids as predictive in vitro systems for drug-induced Llver injury. Drug Metab. Dispos. 45, 419–429 (2017).

DiMasi, J. A. Risks in new drug development: approval success rates for investigational drugs. Clin. Pharmacol. Ther. 69, 297–307 (2001).

David S, Hamilton JP. Drug-induced liver injury. US Gastroenterol Hepatol Rev. 6, 73–80 (2010).

Foster, A. J. et al. Integrated in vitro models for hepatic safety and metabolism: evaluation of a human Liver-Chip and liver spheroid. Arch. Toxicol. 93, (2019).

Gerbal-Chaloin, S. et al. Human induced pluripotent stem cells in hepatology: beyond the proof of concept. Am. J. Pathol. 184, 332–347 (2014).

Gieseck, R. L. et al. Maturation of induced pluripotent stem cell derived hepatocytes by 3D-culture. PLoS One 9, (2014).

Godoy, P. et al. Recent advances in 2D and 3D in vitro systems using primary hepatocytes, alternative hepatocyte sources and non-parenchymal liver cells and their use in investigating mechanisms of hepatotoxicity, cell signaling and ADME. Arch. Toxicol. 87, 1315–1530 (2013).

Gómez-Mariano, G. et al. Liver organoids reproduce alpha-1 antitrypsin deficiency-related liver disease. Hepatol. Int. 14, 127–137 (2020).

Harrison, R. K. Phase II and phase III failures: 2013-2015. Nat. Rev. Drug Discov. 15, 817–818 (2016).

Inoue, H., Nagata, N., Kurokawa, H. & Yamanaka, S. iPS cells: a game changer for future medicine. EMBO J. 33, 409–417 (2014).

Kamiya, A., Kinoshita, T. & Miyajima, A. Oncostatin M and hepatocyte growth factor induce hepatic maturation via distinct signaling pathways. FEBS Lett. 492, 90–94 (2001).

Klomp, F., Wenzel, C., Drozdzik, M. & Oswald, S. Drug–drug interactions involving intestinal and hepatic CYP1A enzymes. Pharmaceutics 12, 1–25 (2020).

Lang, R. et al. Three-dimensional culture of hepatocytes on porcine liver tissue-derived extracellular matrix. Biomaterials 32, 7042–7052 (2011).

Larrey, D. & Pageaux, G. P. Drug-induced acute liver failure. Eur. J. Gastroenterol. Hepatol. 17, 141–143 (2005).

Lauschke, V. M., Hendriks, D. F. G., Bell, C. C., Andersson, T. B. & Ingelman-Sundberg, M. Novel 3D culture systems for studies of human liver function and assessments of the hepatotoxicity of drugs and drug candidates. Chem. Res. Toxicol. 29, 1936–1955 (2016).

Lauschke, V. M. et al. Massive rearrangements of cellular MicroRNA signatures are key drivers of hepatocyte dedifferentiation. Hepatology 64, 1743–1756 (2016).

Lauschke, V. M., Shafagh, R. Z., Hendriks, D. F. G. & Ingelman-Sundberg, M. 3D primary hepatocyte culture systems for analyses of liver diseases, drug metabolism, and toxicity: emerging culture paradigms and applications. Biotechnol. J. 14, (2019).

Lossi, L., D’Angelo, L., De Girolamo, P. & Merighi, A. Anatomical features for an adequate choice of experimental animal model in biomedicine: II. Small laboratory rodents, rabbit, and pig. Ann. Anat. 204, 11–28 (2016).

Meier, F. et al. Hepatic differentiation of human iPSCs in different 3D models: A comparative study. Int. J. Mol. Med. 40, 1759–1771 (2017).

Michalopoulos, G. K., Bowen, W. C., Mule, K. & Luo, J. HGF-, EGF-, and dexamethasone-induced gene expression patterns during formation of tissue in hepatic organoid cultures. Gene Expr. 11, 55–75 (2003).

Mun, S. J. et al. Generation of expandable human pluripotent stem cell-derived hepatocyte-like liver organoids. J. Hepatol. 71, 970–985 (2019).

Oh, H. Y. et al. Dexamethasone protects primary cultured hepatocytes from death receptor-mediated apoptosis by upregulation of cFLIP. Cell Death Differ. 13, 512–523 (2006).

Olgasi, C., Cucci, A. & Follenzi, A. iPSC-derived liver organoids: a journey from drug screening, to disease modeling, arriving to regenerative medicine. Int. J. Mol. Sci. 21, 1–30 (2020).

Ostapowicz, G. et al. Results of a prospective study of acute liver failure at 17 tertiary care centers in the United States. Ann. Intern. Med. 137, 947–954 (2002).

Robinton, D. A. & Daley, G. Q. The promise of induced pluripotent stem cells in research and therapy. Nature 481, 295–305 (2012).

Sa-ngiamsuntorn, K. et al. Upregulation of CYP 450s expression of immortalized hepatocyte-like cells derived from mesenchymal stem cells by enzyme inducers. BMC Biotechnol. 11, 1–15 (2011).

Walker, P. A., Ryder, S., Lavado, A., Dilworth, C. & Riley, R. J. The evolution of strategies to minimise the risk of human drug-induced liver injury (DILI) in drug discovery and development. Arch. Toxicol. 94, 2559–2585 (2020).

Wang, Y. et al. In situ differentiation and generation of functional liver organoids from human iPSCs in a 3D perfusable chip system. Lab Chip 18, 3606–3616 (2018).

Wang, S. et al. Human ESC-derived expandable hepatic organoids enable therapeutic liver repopulation and pathophysiological modeling of alcoholic liver injury. Cell Res. 29, 1009 (2019).

Wouters, O. J., McKee, M. & Luyten, J. Estimated research and development investment needed to bring a new medicine to market, 2009-2018. JAMA 323, 844–853 (2020).

Wu, F. et al. Generation of hepatobiliary organoids from human induced pluripotent stem cells. J. Hepatol. 70, 1145–1158 (2019).

Xu, C., Li, C. Y. T. & Kong, A. N. T. Induction of phase I, II and III drug metabolism/transport by xenobiotics. Arch. Pharm. Res. 28, 249–268 (2005).