Liver organoids represent emerging human-relevant in vitro liver models that have a wide range of biomedical applications in basic medical studies and preclinical drug discovery. However, the generation of liver organoids currently relies on the conventional Matrigel dome method, which lacks precise microenvironmental control over organoid growth and results in significant heterogeneity of the formed liver organoids. Here, we demonstrate a novel high-throughput culture method to generate uniform liver organoids from human pluripotent stem cell-derived foregut stem cells in micropatterned agarose scaffold. By using this approach, more than 8000 uniformly-sized liver organoids containing liver parenchyma cells, non-parenchymal cells, and a unique stem cell niche could be efficiently and reproducibly generated in a 48-well plate with a size coefficient of variation significance smaller than that in the Matrigel dome. Additionally, the liver organoids highly expressed liver-specific markers, including albumin (ALB), hepatocyte nuclear factor 4 alpha (HNF4α), and alpha-fetoprotein (AFP), and displayed liver functions, such as lipid accumulation, glycogen synthesis, ALB secretion, and urea synthesis. As a proof of concept, we evaluated the acute hepatotoxicity of acetaminophen (APAP) in these organoids and observed APAP-induced liver fibrosis. Overall, we expect that the liver organoids will facilitate wide biomedical applications in hepatotoxicity analysis and liver disease modeling.
1.Introduction
The liver is the major organ responsible for regulating multiple complex metabolic processes in the human body. Liver malfunction causes a series of hepatic diseases, including steatohepatitis, liver fibrosis, liver cirrhosis, and liver cancer, which pose a major threat to global public health [1]. According to a WHO report, approximately 2 million people die of liver diseases per year, representing 3.5% of annual global deaths [2]. Unfortunately, the discovery of novel therapeutics for liver diseases remains an unmet task and a major challenge for the pharmaceutical industry because conventional preclinical research model systems for liver diseases cannot accurately predict the therapeutic response of drugs in human patients with liver diseases in clinical trials [3]. Additionally, in the conventional drug discovery pipeline, hepatotoxicity is one of the major causes of both drug failure in clinical trials and drug withdrawals from the market, and liver toxicity is responsible for 12% of clinical safety failures and 21% of drug withdrawals, respectively [4–6]. Therefore, there is an urgent need to develop high-fidelity human liver models that closely emulate human liver physiology and diseases, which will benefit a wide range of biomedical applications, including basic medical research, preclinical drug screening, and hepatotoxicity evaluation.
Recently, liver organoids have emerged as unique human physiologically and pathologically relevant in vitro organotypic models for liver diseases and hepatotoxicity [7–12]. So far, human liver organoids have been generated mainly by using the Matrigel dome method, in which human pluripotent stem cells (hPSCs)-derived foregut stem cells or tissue-derived liver progenitors are encapsulated in the Matrigel domes in a culture dish or a well plate for subsequent expansion and differentiation [9, 13, 14]. However, this method has limited microenvironmental control over the growth and differentiation of organoids. First, the cell encapsulation process was highly randomized in the Matrigel dome. Matrigel is a basement membrane matrix extracted from Engelbreth–Holm–Swarm mouse sarcomas and displays batch-to-batch variability in biochemical composition [15]. These issues result in distinct intra- and inter-Matrigel dome heterogeneity in organoid formation in terms of number, size, shape, location, and maturation rate. Additionally, more professional imaging and analysis techniques are required to locate individual organoids at different focal planes within each Matrigel dome, which poses a technical challenge for imaging automation and algorithms.
To address these issues, we demonstrate a high-throughput liver organoid differentiation and culture method that enables precise control of the size, density, and mass of individual liver organoids. To generate liver organoids, we initially differentiated human induced pluripotent stem cells (hiPSCs) into human foregut stem cells (hFSCs) and then differentiated hFSCs spheroids into liver organoids in a Matrigel-free environment by using agarose-based microfabricated hexagonal closely packed cavity arrays (mHCPCAs). We optimized the geometrical parameters of mHCPCAs and cell seeding density to obtain uniform hPSC-derived liver organoids. Furthermore, we characterized the liver organoids using liver-specific biomarkers and functional assays. Finally, we used liver organoids as preclinical liver models to assess the hepatic response to acute acetaminophen (APAP) exposure and developed a liver fibrosis model.
2.Materials and methods
2.1.Fabrication of agarose-based mHCPCAs
Agarose-based mHCPCAs were fabricated using a micromolding method [16, 17]. Specifically, the layout of the HCPCAs was initially designed using AutoCAD, and a polymethyl methacrylate (PMMA) female mold was fabricated on a PMMA plate (CREROMEM, 200 mm × 300 mm × 2 mm) using a precision micromilling machine (JINGYAN, E3020) equipped with a parabolic spiral-flute bit (WeiTol, Φ0.5 mm). Then polydimethylsiloxane (PDMS) male mold was fabricated by pouring PDMS prepolymer (ratio 1:10, cross-linking agent: elastomer) into PMMA female mold and solidifying the PDMS prepolymer at 80 °C for 2 h. The released PDMS mold was then immersed in a preheated agarose solution (2% w/v). The agarose-based mHCPCAs were formed by gelation of the agarose solution at 4 °C for 20 min and released from the PDMS mold. The mHCPCAs were removed from the agarose construct using an 11 mm round punch and gently placed into a 48-well plate. Each mHCPCAs contained 170 microcavities 500 µm in diameter and 1 mm in height. The mHCPCAs in the plate were sterilized thoroughly in the purified water with 1% (w/v) penicillin-streptomycin, and then under ultraviolet light for more than 30 min. The sterilized mHCPCAs plate was stored at 4 °C for the subsequent liver organoid cultures.
2.2.Numerical simulation
Simulations were conducted using COMSOL Multiphysics (COMSOL Inc., Burlington, MA, USA). We used laminar flow and particle tracking to simulate medium change and cell-seeding particle motion. A time-dependent solver was used to determine the velocity, fluidic shear stress, streamlines, and pressure in the microcavity under the flow. For the simulation, a liquid flow with a thickness of 1 mm was added to the microcavity array to match the cell-seeding process. The drag force condition was applied to the flow domain to analyze the particle trajectories in the fluid.
2.3.Induction of hFSCs from hiPSCs
hiPSCs were maintained in NuwacellTM ncTarget hPSC Medium (Nuwacell Biotechnologies, RP01020), passaged via StemPro® Accutase® (Gibco, A1110501) detachment, and reseeded in a vitronectin (Gibco, A31804)-coated 6-well plate filled with NuwacellTM ncTarget hPSC Medium containing 10 µM Y-27632 (STEMCELL Technologies, 72304). hiPSCs were differentiated into hFSCs using a previously described method with minor modifications [8, 18]. In brief, hiPSCs were detached using Accutase and seeded on vitronectin (VTN-N)-coated tissue culture plates at a seeding density of 100 000 cells cm−2. When the cells reached 85%–90% confluence, the culture medium was replaced with RPMI 1640 medium containing 100 ng ml−1 Activin A (CELL guidance systems, GFH6) and 50 ng ml−1 bone morphogenetic protein 4 (BMP4, R&D Systems, 314-BP). On day two, the culture medium was replaced with RPMI 1640 medium with 100 ng ml−1 Activin A and 0.2% (v/v) knockout serum replacement (Gibco, A3181501). On day three, the culture medium was changed to RPMI 1640 medium with 100 ng ml−1 Activin A and 2% (v/v) knockout serum replacement. On day four to six, the cells were cultured in Advanced DMEM/F12 containing 1% (v/v) B27 (Life Technologies, 17504-044), 1% (v/v) N2, 500 ng ml−1 FGF2, and 3 µM CHIR99021, with medium exchange every single day. All cell cultures were maintained at 37 °C with 5% CO2 and 95% air. hFSCs were generated on day six and then detached into single cells using Accutase. Disassociated hFSCs were used for liver organoid formation in subsequent experiments.
2.4.Generation of human liver organoids using the Matrigel dome method
hFSCs suspensions were prepared in Matrigel solution (Corning, 356237) at a concentration of 2 × 106 cells ml−1. About 50 µl of cell suspension in Matrigel prepolymer was pipetted onto the center of a well in a 24-well plate to form a dome-shaped seated drop. The plate was gently placed in a 37 °C incubator for 30 min to solidify the Matrigel. The cell-encapsulating Matrigel domes were immersed in the liver organoid formation medium and replaced with culture medium every 2 d (table S1). After organoid formation, the medium was switched to liver organoid specification medium and refreshed with culture medium every 2 d for the following 4 d (table S1). On day 14, liver organoids were harvested from the Matrigel dome by physical scratching, pipetting, and filtering via a 100 µm cell strainer (FALCON, 352360). The harvested organoids were re-embedded in 10% (v/v) Matrigel on the ultra-low attachment multiwell plates (Corning, 29443-032) and cultured in complete hepatocyte culture medium (HCM) supplemented with 100 nM dexamethasone, 20 ng ml−1 recombinant human oncostatin M, 10 ng ml−1 recombinant human hepatocyte growth factor, and Lonza HCM bullet kit (excluding epidermal growth factor (EGF)) for 10 d. The organoid culture was maintained at 37 °C with 5% CO2 and 95% air, and the culture medium was replaced every 3 d.
2.5.Generation of human liver organoids in agarose-based mHCPCAs
hFSCs were seeded in agarose-based mHCPCAs, with each microcavity containing 200 cells. The seeded foregut cells were cultured in the liver organoid formation medium supplemented with 10 µM Y27632 for the first 2 d and refreshed with the organoid formation medium for the other 2 d. After organoid formation, the culture medium was switched to the liver organoid specification medium for 4 d (table S1). After liver specification, the culture medium was changed to complete HCM for 10 d. Organoid induction and culture were performed at 37 °C with 5% CO2 and 95% air.
2.6.Hepatotoxicity assays with APAP
Human liver organoids in the agarose-based mHCPCAs on day 24 were separately incubated with 5, 10, and 20 mM APAP (Yuanye Biotechnology, S31044) in complete HCM for 48 h. The cell viability of the organoids was detected using a Calcein/propidium iodide (PI) Cell Viability/Cytotoxicity Assay Kit (Beyotime, C2015S) using an epifluorescence microscope (Olympus, IX-83). The percentage of area positively stained for PI was determined using ImageJ and was relative to the total area positively stained for PI and Calcein [19].
2.7.Generation of liver fibrotic models
On day 24, liver organoids in the mHCPCAs were exposed to 20 mM APAP, and organoids kept in complete HCM without compound were used as controls. After 48 h, the medium was replaced with complete HCM for 8 d. The expression of fibrotic and inflammatory genes was determined using quantitative polymerase chain reaction (qPCR). The expression of the Collagen1A1 protein was analyzed by immunofluorescence.
2.8.Physical characterization of liver organoids
The morphology of the organoids was assessed through bright-field image analysis (Olympus, IX-83). The size of the organoids was manually measured using image analysis software 'ImageJ' (National Institutes of Health, USA, http://imagej.nih.gov/ij).
Additionally, we employed a W8 device (CellDynamics iSRL) to examine the mass density, weight, and size of the individual liver organoids [20]. Briefly, liver organoids were fixed with 4% (w/v) paraformaldehyde (PFA) overnight at 4 °C, resuspended in 1× Dulbecco's phosphate buffered saline (DPBS) (Gibco, 70011-044) at a 200 organoids ml−1 concentration, and transferred to a conical centrifuge tube, and then analyzed according to the previous protocol established by Cristaldi et al [21]. A minimum of ten individual organoids were analyzed for each group. For each organoid, measurements were conducted with a technical repetition at least twice.
2.9.RNA-sequencing (RNA-seq) and data analysis
On day 24, liver organoids in the mHCPCAs and Matrigel domes were washed gently three times with 1× phosphate buffered saline (PBS) at 4 °C. Total ribonucleic acid (RNAs) from liver organoids in the mHCPCAs and Matrigel domes were extracted in accordance with the manual of Total RNA Extraction Reagent (Trizol) (Abclonal, RK30129). Library preparation and transcriptome sequencing were performed using an Illumina HiSeq X Ten (Novogene Bioinformatics Technology Co., Ltd, Beijing, China). The mapping of 100 bp paired-end reads to genes was undertaken using HTSeq v0.6.0, and fragments per kilobase of transcript per million fragments mapped were also analyzed.
2.10.Gene expression analysis with qPCR
On day 24, liver organoids in the mHCPCAs and Matrigel domes were washed gently three times with 1× PBS at 4 °C. To analyze the gene expression, at least 600 liver organoids were combined for a single RNA extraction batch. Total messenger ribonucleic acids (mRNAs) were isolated from the liver organoids using Trizol reagent, and complementary deoxyribonucleic acid (cDNA) was synthesized using ABScript II RT Master Mix for qPCR (ABCLONAL, RK20428). qPCR was performed using SYBR Green Real-time PCR Master Mix (ABCLONAL, RK20428) under the following reaction conditions (40 cycles): denaturation at 95 °C for 1 min, annealing at 58 °C for 30 s, and extension at 72 °C for 30 s. Primer sequences are listed in table S2. The expression levels were normalized relative to the expression of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) using the 2−ΔΔCt.
2.11.Protein expression analysis with immunofluorescence
Liver organoids in the mHCPCAs were washed gently three times with 1× PBS at 4 °C and fixed with 4% (w/v) PFA for 2 h at room temperature. The liver organoids were dehydrated using the standard sucrose dehydration method before sectioning. The 10 µm-thick sections from the liver organoids were obtained using a cryostat (Leica, CM1900). The sections were washed thoroughly three times with 1× PBS. Blocking and permeabilization were performed using 0.1% (w/v) saponin and 1% (v/v) normal donkey serum in PBS for 1 h. Sections were incubated overnight with primary antibodies at 4 °C. Subsequently, the sections were incubated with the secondary antibodies, as shown in table S3. Images were captured using a Leica TCS SP8 STED confocal microscope equipped with LAS X software.
2.12.Liver function assays
Human albumin (ALB) production was detected using an enzyme linked immunosorbent assay (ELISA) kit (Bethyl Laboratories, E80-129). Urea production was detected using the Quantichrom urea assay kit (Bioassay Systems, DIUR-100). The culture media with 48 h incubation on days 10, 14, and 24 were collected and stored at −80 °C for the ALB and urea measurements. Liver organoids were trypsinized into single cells for cell counting using a hemocytometer (Watson, 177-112C). Liver organoid sections were stained with periodic acid-Schiff (Leagene; DG0011) to assess glycogen storage. For visualization of lipid droplets, liver organoid sections were stained with 1 µmol l−1 Nile red (Sigma-Aldrich, 19123) for 5 min at room temperature. The organoids were washed twice with PBS before imaging.
2.13.Statistical analysis
All data are presented as mean ± standard error of the mean (SEM). Statistical analyses of the experimental data were performed using GraphPad Prism 8.0.0 (GraphPad Software, San Diego, CA, USA). One-way analysis of variance with Student's t-test was used to verify statistically significant differences among groups, ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, n.s. = not significance (P > 0.05). The sample sizes were indicated in the figure legends.
3.Results
3.1.Development of agarose-based mHCPCAs
We developed agarose-based mHCPCAs in a 48-well plate for liver organoid formation and culture using the micro-molding method (figures S1(A) and (B)). Specifically, agarose is used as a biocompatible and ultra-low-affinity substance to eliminate cell and extracellular matrix (ECM) adhesion to microcavities during long-term organoid culture. Moreover, agarose exhibits good optical properties and is compatible with bright-field microscopy.
We optimized the geometrical design of the mHCPCAs to obtain uniform cell seeding and microenvironment for individual organoids. A hexagonal-closely packed format was employed for microcavity arrangement (figures S1(C) and (D)), which enabled homogeneous seeding density and intrinsic cytokine diffusion among neighboring microcavities. Moreover, the hexagonal-closely packed arrangement also maximized the packing density of microcavities, increasing the throughput of the organoid culture in each well of the 48-well plate (figures S1(G)). Additionally, a U-shaped bottom was fabricated for each microcavity (figures S1(E) and (F)), which enabled efficient cell aggregation during cell seeding. Furthermore, we examined microcavities with varying diameters and depths for organoid culture. Microcavities with diameters of 500 µm and 1000 µm displayed no distinct effect on the size and morphology of the organoids on day 24. Conversely, microcavities with a depth of less than or equal to 500 µm often lost organoids during culture medium exchange, while microcavities with a depth of 1000 µm maintained organoid culture for 24 d. Thus, we selected microcavities with a diameter of 500 µm and depth of 1000 µm for liver organoid culture.
We conducted numerical simulations to investigate cell seeding and medium exchange in organoid cultures in mHCPCAs. The simulation of cell seeding via the particle tracking module revealed that cell suspensions pipetted out through a 200 µm tip enabled uniform streamlines and velocity through individual microcavities (figures 1(D), (E) and S2(A)), resulting in uniform cell seeding (figures S2(B) and (C)). Additionally, the simulation of medium exchange indicated that a fluidic vortex formed within the microcavity and kept the microcavity bottom away from the high flow velocity, high shear stress, and high pressure (figures 1(F) and S2(D)). These results implied that the geometrical design of microcavities protected cultured organoids from the biophysical effects of fluidic flow during the medium exchange. Moreover, the vortex formed in the microcavity acted as a fluidic trapper and prevented the loss of cells during the medium exchange. These results revealed that the geometric design of the mHCPCAs facilitated uniform microenvironmental control over the individual organoids in each microcavity.
3.2.Cell seeding density on liver organoid formation in the mHCPCAs
We obtained liver organoids from hiPSCs via five different differentiation and culture stages, including (a) differentiation of hiPSCs into definitive endoderm cells in the Petri dish; (b) differentiation of definitive endoderm cells into hFSCs in the Petri dish; (c) formation and differentiation of hFSCs spheroids in the liver organoid formation medium in the mHCPCAs; (d) differentiation of liver organoids in the liver organoid specification medium; and (e) differentiation of liver organoids in the complete HCM (figures 1(A)–(C)). Immunofluorescence analysis demonstrated that hiPSCs stained positive for the PSC-specific markers OCT3/4 and Nanog, revealing their pluripotent identity (figure S3(A)). Furthermore, immunofluorescence analysis illustrated that hFSCs were positive for caudal type homeobox 2 (CDX2, hFSCs-specific markers) and EpCAM [7] (figure S3(B)).
We investigated the effect of the initial cell seeding density on liver organoid formation in mHCPCAs. hFSCs suspensions were prepared in the liver organoid formation medium with a cell concentration ranging from 0.95 × 104 to 9.5 × 104 cells ml−1 and then plated to the mHCPCAs with 200 µl in each well in the 48-well plate to achieve a seeding density from 50, 100, 200, 400, and 500 cells microcavity. Cell suspensions settled gravitationally and aggregated in the microcavities within 30 min. The hFSCs in the microcavities dynamically self-organized into 3D spheroids within 72 h after seeding (supplementary movies1 and 2).
We found that microcavities with a seeding density of 50 cells failed to form a spheroid, and microcavities with seeding densities of 500 cells formed a large spheroid surrounded by several small cell aggregates and numerous dead cells. Furthermore, we found that an initial seeding density of 100, 200, and 400 cells per microcavity resulted in a single spheroid surrounded by much fewer dead cells than those with a seeding density of 500 cells per microcavity (figure S3(C)). The average size of the formed liver organoids is 309.05 ± 32.39 µm, 332.22 ± 32.48 µm, and 329.29 ± 36.67 µm corresponding to the initial seeding density of 100, 200, and 400 cells per microcavity, respectively (figure 1(H)). Specifically, the average size of formed organoids increased from 108.97 ± 12.41 µm on day 10, 220.64 ± 24.87 µm on day 14, to 305.40 ± 23.30 µm on day 24, when the initial seeding density is 200 cells per microcavity (figures 1(G) and (I)). We selected an initial seeding density of 200 cells per microcavity for hPSC spheroid formation (figure S3(D)). These results also indicate that the initial cell number is a critical parameter for organoid formation in the mHCPCAs method.
3.3.Characterization of the homogeneity and reproducibility of liver organoids in the mHCPCAs
The homogeneity and reproducibility of liver organoid formation are essential requirements for the development of organoid-based quantitative bioassays for investigating liver diseases and hepatotoxicity. We quantitatively analyzed the organoid size in the mHCPCAs and Matrigel dome groups on day 24 using bright-field microscopy and image analysis. The results revealed that the average sizes of liver organoids are 284.17 ± 23.98 μm and 308.83 ± 39.43 μm in the mHCPCAs group and the Matrigel dome group, respectively (figures 2(A) and (B)). Notably, the liver organoids generated in the mHCPCAs showed a smaller size coefficient of variation (CV) than those generated in the Matrigel dome method (figure 2(C)). Additionally, three batches of independent organoid culture assays indicated no significant difference in the mean size of liver organoids among the different batches (figure 2(D)).
To further assess the homogeneity of liver organoids in the mHCPCAs, we employed the W8 instrument, a commercialized microfluidic-based analyzer that enables precise measurement of the mass, density, and diameter of individual organoids. The results demonstrated that the liver organoids in the mHCPCAs group showed much higher homogeneity in mass, density, and diameter than those in the Matrigel dome group (figures 2(E)–(G)). Specifically, CVs of organoid weight and density in the mHCPCAs group were 84.40% and 85.99% lower than those in the Matrigel dome group, respectively. Overall, mHCPCAs significantly increased the homogeneity and reproducibility of liver organoid formation and facilitated organoid-based quantitative bioassays.
3.4.Transcriptomic profiling of liver organoids in the mHCPCAs
We compared global RNA-seq data of liver organoids in mHCPCAs with that in the conventional Matrigel domes. We analyzed the differential gene expression (DGE) between liver organoids in mHCPCAs and Matrigel domes. We identified 14 800 differentially expressed genes, respectively, from each of the above comparisons, with 12 970 genes at the intersection (figure 3(B)). We then performed an additional DGE analysis of the 32 868 genes annotated in the genome, which indicated that 1463 were differentially expressed, with 823 and 640 genes up- and down-regulated, respectively (figure 3(B)). To gain further insights into the biological processes enriched in liver organoids in mHCPCAs, we performed Gene Ontology term enrichment analysis. Notably, the biological processes enriched in this gene set were associated with essential functions such as metabolism, transport, and liver development (figure 3(C)). The Kyoto encyclopedia of genes and genomes (KEGGs) analysis result indicated liver-specific metabolism and biosynthesis functions were enhanced in the mHCPCA group compared to the Matrigel dome group. Specifically, the activated metabolism-related signaling pathways are responsible for the hepatic metabolism of retinol, drug, xenobiotics, cholesterol, fatty acid, etc. Meanwhile, the activated biosynthesis-related signaling pathways are related to the hepatic synthesis of primary bile acid, steroid hormone, and fatty acid (figure 3(D)). We further analyzed cell-type specific gene expression profiles of liver organoids in both the mHCPCAs and the dome. The results suggested that the liver organoids in mHCPCAs expressed a number of hepatocyte markers, some cholangiocyte-specific markers, and non-parenchymal cell markers (figure 3(E)). Compared to the liver organoids growth in the Matrigel domes, the liver organoids generated by mHCPCA demonstrated upregulation in hepatocyte-specific genes such as ALB, CPS1, CEBAP, and downregulation in cholangiocyte-specific gene KRT7, KRT17, KRT19. Additionally, DGE analysis of liver function-associated genes indicated that liver organoids in mHCPCAs displayed higher maturation in drug metabolism, glucose metabolism, fat metabolism, and bile transport (figure 3(F)).
3.5.Characterization of liver-specific biomarkers and functions of liver organoids in the mHCPCAs
We investigated whether the mHCPCAs method successfully produced functional liver organoids and examined liver-specific gene expression throughout the hepatic differentiation process using qPCR analysis (figure 4(A)). qPCR analysis indicated that the mRNA levels of liver-specific markers, hepatocyte nuclear factor 4 alpha (HNF4α) and ALB, were increased from day 14 to day 24, whereas the cholangiocyte-specific marker cytokeratin 19 (CK19) exhibited a decreasing trend. Additionally, the expression levels of alpha-fetoprotein (AFP) showed a significant increase on day 24 compared to day 14. These results indicated that the formed liver organoids were still in the early maturation stage.
We performed immunofluorescence analysis of liver organoids on day 24 to investigate the protein expression of liver-specific markers (figures 4(B) and S4). The results revealed that most organoid cells illustrated hepatic fate specification and were positive for hepatocyte-specific markers ALB, AFP, and HNF4α. Additionally, the results indicated the existence of non-parenchymal cells in the outer layer of organoids expressing vimentin (VIM, a mesenchymal marker). But these VIM expressing cells stained negative for CD31 (endothelial marker) and CD68 (Kupffer cell marker). Interestingly, most liver organoids contained a unique region highly co-expressing EpCAM, AFP, CK19, and ALB, implying that a liver stem cell niche existed in the organoid. Collectively, these results demonstrated that we successfully generated human liver organoids containing multiple liver parenchymal and non-parenchymal cell types.
Furthermore, we investigated liver-specific functions of the generated liver organoids (figures 4(C)–(F)). ELISA analysis indicated that the levels of ALB secretion and urea production of the liver organoids significantly increased over the two weeks from day 10. In addition, period acid-Schiff (PAS) staining and Nile red staining assays indicated that the liver organoids on day 24 developed liver-specific functions in the liver parenchymal cell regions, including the accumulation of fatty droplets and glycogen storage.
Overall, these results clearly indicate that the mHCPCAs method generates functional liver organoids that faithfully recapitulate liver-specific cell types, organization, and function at the early stage of liver development.
3.6.Application of liver organoids in drug toxicity assessment and liver fibrosis modeling
We assessed the acute toxic effects of the food and drug administration (FDA)-approved drug APAP. On day 24, liver organoids were incubated with APAP solutions at various concentrations (5, 10, and 20 mM) for 48 h (figure 5(A)). qPCR analysis indicated that the mRNA levels of APAP-specific cytochrome P450 enzymes, CYP2E1 and CYP1A2, were upregulated in liver organoids treated with 20 mM APAP (figure 5(B)). Calcein/PI staining assays further illustrated that the acute toxic effects of APAP followed a dose-dependent relationship, with an IC50 of APAP around 20 mM (figures 5(C) and (D)), which is consistent with the current gold standard (i.e. human primary hepatocytes) for APAP toxicity assay [22]. In contrast, liver organoids generated by the Matrigel dome method demonstrated a IC50 of APAP around 10 mM (figure 5(S)).
Furthermore, we investigated APAP-induced liver fibrosis using an organoid model (figure 6(A)). On day 24, the liver organoids treated with 20 mM APAP for 48 h showed a significant change in morphology and exhibited swelling and irregular boundaries compared to those in the untreated group (figure 6(B)). qPCR analysis indicated that the mRNA levels of fibrotic markers COL1A1 and VIM were significantly increased compared to those in the untreated group (figure 6(C)). Additionally, gene expression of the pro-inflammatory cytokines TNF-α and IL-8 was upregulated compared to that in the untreated group (figure 6(D)). Furthermore, immunofluorescence analysis revealed that Collagen1A1 excessively accumulated around the hepatic cells, and cystic lumen emerged in the organoids, which were regarded as critical cytopathological features in liver fibrosis (figure 6(E)).
4.Discussion
Physiologically and pathologically relevant human liver models are urgently required for a wide range of biomedical applications, including mechanistic studies of liver diseases, drug hepatotoxicity evaluations, and preclinical drug efficacy assessments for liver diseases. Recently, human liver organoids have been increasingly considered promising liver models because of their inherent similarity to native human livers in terms of developmental trajectory, liver-specific functions, and cell types [23–26]. Currently, liver organoids are derived from PSCs or liver tissues [27, 28]. Specifically, PSC-derived liver organoids have several advantages over liver tissue-derived liver organoids. First, the generation of PSC is a less invasive process than the harvesting of fresh liver tissue. Second, PSC-derived liver organoids inherit patient-specific genetic backgrounds and display liver disease-specific pathological features. Furthermore, PSC-derived liver organoids may generate liver-specific nonparenchymal cells that originate beyond the endoderm, and thus they more faithfully emulate liver-specific functions than tissue-derived liver organoids that contain cell types specified to the endoderm [29, 30]. The current standard for generating PSC-derived liver organoids is based on the Matrigel dome method, which cannot precisely control the microenvironment for organoid growth [31]. However, the chemically-undefined composition of Matrigel and random encapsulation of cells cause poor microenvironmental control over cell growth and differentiation, resulting in significant heterogeneity and low reproducibility of organoid formation [32, 33].
For a long time, the culture of endoderm-originated organoids, such as liver, intestine, and gut, relied on the Matrigel dome method. However, the organoids generated by the Matrigel dome method demonstrate great heterogeneity among individual organoids in terms of size, morphology, and maturation rate. One potential reason for organoid heterogeneity is attributed to varied local microenvironments over individual organoids, resulting from the randomized organoid locations and uncontrolled biochemical interactions between the neighboring organoids. Recently, geometrically-tuned microenvironments have been utilized to bioengineer organoids with controlled morphogenesis and functional maturation. Specifically, several bioengineering approaches have been reported to tailor physical cues of the microenvironments, including 3D bioprinting [34], microfluidic chips [35], and microcavity arrays [16]. For example, Brassard et al reported control over morphogenesis of intestinal organoids with a microfluidic channel. Intestinal stem cells were bioengineered to form tube-shaped epithelium cytoarchitecture with a perfusable lumen and a biomimicking spatial organization of crypt- and villus-like structure to that in vivo [36]. Duzagac et al demonstrated bioengineering of geometrically-controlled intestinal organoids with 3D bioprinting [34]. Using this technique, they generated centimeter-scale intestinal tissue with in vivo-like lumens, branched vasculature, and tubular intestinal epithelia. However, both 3D bioprinting and microfluidic chips are not high-throughput tools and cannot generate a large amount of geometrically-defined organoids simultaneously. Compared to 3D bioprinting and microfluidic chips, microcavity arrays may produce hundreds to thousands of individual organoids with deterministic locations, inter-organoid distance, and uniform morphology and size. Several types of microcavity have been employed to define the local geometric microenvironments of organoids, including V-shaped [37], U-shaped [16, 38, 39], and pyramid-shaped microcavities [40]. These microcavity arrays have been reported to produce intestinal organoids, testicular organoids, pancreatic duct-like organoids, and colorectal cancer organoids. In this study, we uniquely fabricated agarose-based U-shaped microcavities to regulate the size and morphogenesis of PSC-derived liver organoids.
Recently, several studies have explored non-Matrigel scaffolds for formation of PSC-derived liver organoids [33, 41]. The utilized scaffolds include decellularized liver extracellular matrix [42], alginate [43], collagen [44], and polyethylene glycol and their derivatives [45]. These non-Matrigel scaffolds can emulate some biochemical and biophysical features of naive extracellular matrix and improve the culture of PSC-derived liver organoids. Here we employed agarose as a scaffold material for liver organoid formation due to its low-cell-attachment property for cell aggregation. Previous studies have used agarose microwell arrays to produce uniform-sized human embryoid bodies and HepG2 Spheroids [46, 47]. Additionally, agarose has suitable mechanical strength for microfabrication as well as high optical transparency and biocompatibility.
Hence, we developed a high-throughput micropatterned agarose scaffold for consistent and reproducible hPSC-derived liver organoids. Firstly, we fabricate hexagonal-closely packed microcavity arrays with each cavity for a single liver organoid culture. The hexagonally-close packing layout allows strict control over the number of foregut cells sedimented in each cavity, and it also enables maximum packing density of microcavities, increasing organoid culture's throughput. Furthermore, the U-shaped bottom of the microcavity facilitates the aggregation of cells into a single spheroid. We dissected the agarose-based construct into round pieces to match the diameter of a 48-well plate by using a round punch. Thus, the micropatterned organoid culture device is compatible with conventional high-throughput screening tools in the pharmaceutical industry.
Liver organoids are evaluated using human liver-specific biomarkers and functions. RNA-seq, qPCR and immunofluorescence analysis indicate that our liver organoids display liver-specific gene and protein expression, including ALB, HNF4α, and AFP. These results imply organoid differentiation by our method demonstrates a stronger hepatic fate [44, 48]. Furthermore, VIM positive cells located at the outer shell of the organoids do not express CD31 and CD68, which suggests that these VIM positive cells are not endothelial and Kupffer cells or these cells are immature mesenchymal cells. Notably, we observed a unique self-organized cytoarchitecture in our liver organoids, in which a lumen-like liver stem cell niche is surrounded by a hepatic cell region and then a non-parenchymal cell shell in the outer layer [49, 50]. This new type of liver organoid recapitulates some critical features of the fetal liver during human developmental stage and has never been reported previously to the best of our knowledge. Additionally, our liver organoids showed liver-specific functions, including ALB secretion, urea production, glycogen synthesis, and lipid drop synthesis, which was consistent with some features of PSC-derived liver organoids in the previous reports [7, 8, 51].
The prediction of drug-induced hepatotoxicity is an essential step in preclinical drug discovery. However, current animal models and monolayer primary liver cell culture models fail to produce a convincing prediction of drug-induced hepatotoxicity in the preclinical phase, resulting in a large proportion of drug failure in clinical trials [52, 53]. The emergence of liver organoids provides a promising opportunity to improve the accuracy of preclinical hepatotoxicity predictions for drug candidates [54, 55]. We demonstrate the applicability of our liver organoids to APAP-induced hepatotoxicity and fibrotic models. APAP is a widely used pain reliever and fever reducer at therapeutic dosages. Specifically, APAP overdose can cause severe hepatotoxicity in children [56]. We examine the acute hepatotoxicity of APAP at high doses in a liver organoid model. The results indicate that high-dose APAP treatment at 20 mM caused upregulation of CYP2E1 and CYP1A2. CYP2E1 and CYP1A2 are highly expressed in human fetal livers and involve in the biotransformation of xenobiotics [57]. These results imply that this liver organoid model can be potentially used as a novel platform for pediatric or in-utero drug assessment studies. Additionally, high-dose APAP treatment also induces liver fibrotic pathologies, including upregulation of fibrotic genes COL1A1 and VIM and pro-inflammatory genes TNF-α and IL-8, excessive deposition of Collagen1A1, and formation of cystic lumens, which emulate typical cytopathological processes in liver fibrosis [9, 58–60]. These results imply that this liver organoid model could be used to study liver pathophysiological process.
However, the mHCPCAs method has certain limitations. Liver organoids produced by the mHCPCAs method are still at the early maturation stage of liver development and demonstrate a relatively low maturation rate and liver function compared to the liver organoids reported in some previous studies [8]. Moreover, the liver organoids generated by the mHCPCAs method do not generate a hepatic lobule-specific cytoarchitecture, in which hepatic sinusoids and cords form hexagonal structures. However, due to the limited knowledge of perinatal human liver development, the generation of functionally fully-matured PSC-derived liver organoids still remains an unsolved problem [31]. We still expect these limitations to be overcome by introducing dynamic and co-culture techniques.
5.Conclusions
In this study, we developed a novel PSC-derived liver organoid culture method that enables the generation of highly homogeneous and functional liver organoids in a high-throughput micropatterned agarose scaffold. More than 170 liver organoids can be generated in a single well of a 48-well plate with a single organoid in a single microcavity. Compared to the traditional Matrigel dome method, our mHCPCAs demonstrated high homogeneity in the morphological and physical features of the organoids. Additionally, mHCPCAs are compatible with the gold standard of the pharmaceutical industry and can be integrated with high-content screening because of the deterministic locations of individual organoids in the well plate. We demonstrated the applicability of our method by modeling drug-induced acute liver injury. We expect that the mHCPCAs method will be a powerful tool for facilitating mechanistic studies of liver diseases in basic medical research and hepatotoxicity evaluation in preclinical drug discovery.
Acknowledgments
The authors acknowledge the financial support from the National Key Research and Development Program of China (No. 2018YFA0109000) and the Applied Foundational Research Program of Wuhan Municipal Science and Technology Bureau (No. 2018010401011296). Thanks for technical support from Research Center for Medicine and Structural Biology of Wuhan University, and Innovations in stem cell and organoids Project (ISCO).
Data availability statement
All data needed to evaluate the conclusions in the paper are presented in the paper and/or supplementary materials. More detailed data will be made available to interested investigators upon request.
Conflict of interest
The authors state no conflict of interest.
CRediT authorship contribution statement
Conceptualization: P C; data curation: P C and S Q J; formal analysis: S Q J, F X, and Z X Z; Funding acquisition: P C; Investigation: S Q J, M L J, X D X, Y Z, L J G, and P C; Methodology: S Q J, F X, Z W, M L J, Z X Z, X D X, L J G, H F, and P C; Project administration: P C; Resources: P C; Software: S Q J, F X, Z X Z, M L J, and J B W; Supervision: P C; Validation: S Q J, X D X, and L J G; Visualization: S Q J, X F, Y H F, and Z X Z; Writing original draft: S Q J and P C; Writing-review & editing: P C, S Q J, and C Y L.