Transient culture of adult human hepatocytes as organoids improves engineered liver engraftment Authors: Sarah H Saxton, Inkyung Kang, Shannon Mitchell, Jeremy Lotto, Aditya Upadhyay, Chenhui Zou, Corrine Quirk, Tara N McCray, Olivia Prado, Fredrik Johansson, Chelsea L Fortin, Dorice L Goune, Ekaterini D Zacharis, Ludovic Vallier, Ype P de Jong, Kelly R Stevens Published: 2026 ABSTRACT AbstractBioengineered livers could serve as a bridge or alternative to whole organ transplantation for patients suffering from liver disease. However, identifying cellular configurations that promote adult human hepatocyte survival after tissue transplantation remains a critical hurdle. We hypothesized that transient culture of adult hepatocytes in 3D as organoids would enhance engraftment of these cells in bioengineered tissues in vivo. Here, we established an organoid system that supports culture of adult human hepatocytes from eight different human adult donors aged 19–49. These organoids support retention of human hepatocyte morphology found in the mature liver, express an adult liver transcriptome, and exhibit mature functions such as inducible cytochrome activity. Importantly, adult human hepatocyte organoids (AHHO) improved engraftment of hepatocytes up to 25–fold in vivo in engineered tissues compared to previous hepatocyte models. Thus, ‘priming’ human cells in organoids could have a major impact on translational bioengineered tissue applications.Keywords: Hepatocyte organoids, Bioengineered liver, Tissue engraftment Introduction As the largest solid organ in the body with hundreds of wide–ranging functions, the liver is essential for the maintenance of healthy human life [1]. Some of the liver's roles include synthesis of plasma proteins, blood detoxification, drug metabolism, storage of glycogen, bile production and excretion, and cholesterol synthesis. Due to the high functional and metabolic demands on the liver, hepatic injury can have systemic and devastating effects on human health. Liver disease causes more than 2 million deaths worldwide each year, and the incidence of both chronic and acute disease is continuing to increase annually [2,3]. Those living with liver disease experience a severely reduced quality of life, which results in a significant and inequitable healthcare burden across the US and the world [2–6]. Although whole organ transplant can be curative for many forms of liver disease, the rising rate of disease and limited donor supply is leaving an increasing number of patients either too sick or already deceased before the donor organ becomes available for transplant [7]. To address this challenge, some groups are pursuing liver cell therapies, in which hepatocytes are delivered to the diseased or damaged liver of the patient. This approach has progressed to clinical trials, with well over 100 patients having already received cell therapy transplantation procedures [8–10]. For clinical studies, the use of fully differentiated and mature human hepatocytes has been the preferred cell choice because alternative sources such as stem cell–derived hepatocytes remain too phenotypically and functionally immature for therapeutic applications [11–15]. Most commonly, mature human hepatocytes have been obtained from donor organs that were denied or discarded for organ transplantation. Hepatocyte transplantation has shown most potential for achieving at least partial correction of liver metabolic disorders, where the microenvironment of the host liver remains hospitable for graft cells. Unfortunately, this strategy is unlikely to circumvent the need for organ transplantation in patients with acute liver failure and cirrhosis, where the damaged microenvironment of the host liver limits engraftment of transplanted cells. This limitation is important because the forms of liver disease with disrupted microenvironment are also the most prevalent. Bioengineered human livers could address this human health challenge by serving as a bridge or alternative to whole organ transplantation for patients with acute and chronic liver failure. These engineered livers could be implanted in an ectopic location in a patient, thus avoiding the challenges associated with engrafting cells directly into damaged host livers [16–18]. While much progress has been made in liver tissue engineering, a critical challenge has been identifying configurations that support robust survival and functionality of engrafted hepatocytes after transplantation. To date, most studies have centered on solving this problem by identifying supportive (non–hepatocyte) cell populations or architectural configurations of the engineered tissue that enhance engraftment [17,19,20]. While these approaches have shown merit, they would require scaling up cell manufacturing of additional cell populations and/or producing specialized biofabrication equipment, respectively, both of which may impede clinical translation. We thus centered on a different approach, by instead focusing on improving engraftment through alternative methods for hepatocyte handling and culture prior to formation of the engineered tissue. This approach was inspired by findings in two fields. First, in liver cell transplantation, the quality of isolated hepatocytes has been shown to be directly correlated with clinical outcomes after orthotopic hepatocyte transplantation [15]. As a result of those findings, much of the cell therapy field is working to improve the viability, quality, and yield of hepatocytes upon isolation from donor livers. We thus wondered if improving the quality of hepatocytes via advanced culture methods prior to their encapsulation in engineered tissues would improve engraftment outcomes. Second, in human cell biology, 3D culture systems have revolutionized culture of numerous primary epithelial cell types, ranging from the human intestine to kidney to liver [21–30]. We hypothesized that transient culture in 3D organoid culture could be leveraged to improve hepatocyte quality, or otherwise prime hepatocytes prior to their encapsulation in engineered tissues to improve tissue engraftment upon transplantation. To test this hypothesis, we established a method to transiently culture mature human hepatocytes isolated from adult donor livers as organoids in vitro, validated using eight adult donors, aged 19–49 years old and of varying sex and racial/ethnic demographics. The organoids supported retention of human hepatocyte morphology found in the mature liver, exhibit mature functions, and could be encapsulated in hydrogels to produce engineered tissues. Engraftment of the resultant tissues into mice with liver injury generated robust grafts with human hepatocyte morphology, phenotype, and functions. Finally, adult human hepatocyte organoids (AHHO) improved engraftment of hepatocytes in vivo in engineered tissues nearly 25–fold compared to prior state–of–the–art engineered tissue formulations. Taken together, these results demonstrated the critical importance of leveraging cell culture strategies such as organoids to produce engineered tissues, and thus could have impact on numerous fields of regenerative medicine. Results Culture of primary adult human hepatocytes as organoids We set out to achieve a culture system for adult primary human hepatocytes that met the following design criteria: First, the system should enable the culture of adult human hepatocytes in 3D conditions that prioritize cell–cell interactions between hepatocytes, as well as between hepatocytes and an extracellular matrix to support phenotypic and functional recovery of hepatocytes following cryopreservation [31, 32]. Second, the culture system must support maintenance of adult (post–pubescent) human hepatocyte phenotype and functionality for several weeks in vitro, with high well–to–well consistency across multiple different donors. Finally, the culture system should facilitate mobilization and collection of the cultures for downstream applications, such as human tissue model systems and implantable engineered tissues. To develop a culture system that met these specifications, we were inspired by recent progress in culturing adult epithelial cells by embedding in droplets of soft extracellular matrix hydrogels [21,33,34]. We postulated that culture in a physiologically relevant 3D matrix such as Matrigel, paired with stimulation from appropriate liver developmental or regenerative factors could support mature adult primary hepatocytes to form and be sustained as organoids [35] (Fig. 1a). Adult human hepatocytes exhibit significant growth and maintenance of phenotype and morphology through organoid culture. (a) Schematic illustrating the isolation, seeding, and growth of AHHO. (b) Average cross–sectional area of all cellular/multicellular objects in individual wells on days 1, 5, and 14. Data represented as mean ± SEM of 23 wells from the same donor. One–way ANOVA, Tukey's multiple comparisons test, ****p < 0.0001. (c) Representative brightfield images taken in the center of Matrigel droplets on days 1, 5, and 14. Red asterisks track individual cells/organoids over time. Scale bars = 500 μm for low magnification and 50 μm for inset images. (d) 2D confocal images of day 14 organoids stained for CK18 (cyan) and CK19 (magenta) (left), Alb (magenta) and MRP2 (cyan) (middle), or Arg1 (cyan) and E–cad (magenta) (right), together with Hoechst (yellow). Scale bars = 50 μm. (e) 3D confocal z–stack of day 14 organoids stained for E–cad (cyan), Arg1 (magenta), and Hoechst (yellow). Maximum intensity projection (left), Imaris surface rendering of E–cad and Hoechst (top right), depth color coding of Hoechst (yellow = top, pink = middle, blue = bottom) with E–cad in white (bottom right). Scale bars = 50 μm. (f) 3D confocal z–stack of day 14 organoids stained for E–cad (cyan), HNF4α (red), and Hoechst (blue). Maximum intensity projection (left), depth color coding of E–cad (top right), and HNF4α (bottom right) (yellow = top, pink = middle, blue = bottom). Scale bars = 50 μm. Towards this end, we first sought to test whether prior methods for culturing more developmentally immature human hepatic cells could also support the culture of adult hepatocytes in organoids. We cultured human hepatocytes from a 34–year–old donor in conditions previously developed to culture human fetal hepatocyte organoids [36] (Fig. S1a and b). After two weeks, immunostaining revealed that although some organoids did form, nearly all cells expressed cytokeratin–19 (CK19), a cytoskeletal protein strongly expressed by cholangiocytes and some hepatic progenitor cells, but generally absent in healthy mature hepatocytes (Fig. S1c, left). Furthermore, many CK19+ organoids were organized in a single or double layer of small cells forming a hollow cystic structure, a morphology typical of liver ductal cell or cholangiocyte organoids [23,37–39] (Fig. S1c, left, arrow). We also tested organoid formation and growth under conditions developed for human hepatoblast organoids. Here, cultures again generated CK19+ organoids with ductal morphology, though dense organoids containing cells with hepatocyte morphological signatures, such as a strong expression of epithelial marker cytokeratin 18 (CK18) at cell–cell junctions, and some binucleated cells, were also present (Fig. S1c, right, arrowhead). Media supernatants from organoids cultured under both conditions were analyzed for the presence of human albumin, a protein produced in abundance by human hepatocytes. Human albumin secretion was consistently very low across wells grown in fetal hepatocyte conditions. While organoids grown in hepatoblast conditions had higher albumin production, the albumin levels varied widely from well–to–well, with many wells containing no measurable albumin (Fig. S1d). Thus, while the presence of occasional organoids with aspects of hepatocyte phenotype in such conditions was encouraging, the phenotypic and functional consistency across cultures remained too low to meet our design criteria. Thus, we next sought to achieve adult human hepatocyte organoid cultures with high purity, functionality, and well–to–well consistency across cultures. To do this, we further screened the addition or modulation of factors implicated in liver development and regeneration, including growth factors HGF, FGF1, FGF2, FGF7, FGF10, TGFα, and small molecule TGFβ inhibitor, A83–01 (Fig. S1e), while also stimulating Wnt and epidermal growth factor (EGF) pathways and inhibiting rho–associated kinase (ROCK) [36,40]. After one week in such study, brightfield imaging identified two conditions that appeared to robustly support gross organoid growth and morphology (Fig. S1f and g). Organoids cultured under these two conditions also had significantly higher human albumin secretion compared to all other conditions screened, including hepatoblast media (condition 1), and this albumin production was consistent across wells (Fig. S1h). Notably, compared to all others tested, both of these conditions had a low concentration of TGFβ inhibitor A83–01. Of the two conditions with low A83–01, one (condition 9) was supplemented with HGF and FGF10 and generated organoids that demonstrated intermediate or mixed phenotypes, with the majority of organoids exhibiting dense morphology but heterogenous CK19 expression (Fig. S1i, left), consistent with previous findings using similar media conditions [27,41]. Conversely, organoids from the second condition (condition 7, with no HGF or FGF10) demonstrated a more consistent morphology in which organoids were predominantly dense and contained CK18+/CK19– cells (Fig. S1i, right). Due to this phenotypic conservation paired with improved organoid formation and high function, condition 7, which we will hereafter refer to as “AHHO culture conditions”, was used to culture AHHO for subsequent phenotypic, functional and application studies. To characterize the growth and phenotype of human hepatocytes cultured under AHHO culture conditions, we next grew human hepatocytes from a 19–year–old donor as organoids and measured the cross–sectional area of cellular or multicellular objects in cultures over time. On the first day after cell seeding, average object area was 907 μm2 (~34 μm diameter, assuming sphericity) with a normal distribution, reflecting a relatively homogenous population of single hepatocytes or small clumps of a few cells (Fig. 1b–c, and S2a). Over time, the cell population separated into a bimodal distribution, with a lower mode representing inert single cells and an upper mode representing growing organoids. At day 14, ~1/3 of objects formed organoids (Fig. S2a–c) ranging from ~50 μm to over 150 μm in diameter. To further investigate the phenotypic identity of the cells comprising the organoids, we performed immunostaining for various structural and functional hepatic proteins. Immunostaining for CK18 demonstrated that cells self–organized into densely packed spheroids with large cells and a low nuclear to cytoplasmic ratio, similar to primary hepatocytes (Fig. 1d, left). Cells in AHHO were largely negative for the biliary/progenitor cell marker CK19, though occasional cells were found that were CK19+, but generally maintained hepatic as opposed to biliary morphology (Fig. 1d, left). Most cells expressed albumin (Alb), a marker of functional hepatocytes, and arginase–1 (Arg1), an enzyme involved in nitrogen metabolism (Fig. 1d, center and right). AHHO also expressed the cell–cell junction protein E–cadherin (E–cad), which was primarily localized between cells (Fig. 1d, right). Multidrug resistance–associated protein 2 (MRP2), an efflux transporter located in the bile canaliculi on the apical domain of polarized hepatocytes, was expressed on the interior of dense organoids (Fig. 1d, center). Samples were then immunostained for both albumin and Ki67, a protein associated with the cell cycle and thus used as a marker for proliferating cells. Ki67+ nuclei made up 34% and 1% of all albumin + nuclei on days 5 and 15, respectively (Fig. S2d–e). Rare Ki67+ mitotic figures were also identified, suggesting that some of the hepatocytes in organoid culture were actively undergoing cell division (Fig. S2f). We thus sought to further assess population expansion. First, we used CellTiter Glo since it is commonly used in organoid field to examine cell number after expansion. Using this assay, we found that the amount of ATP in organoid cultures increased 30 fold after 2 weeks of culture (Fig. S2g). However, we realized this result could be confounded since hepatocytes are highly metabolic, and metabolism per cell can also undergo changes over time. Thus, we also used a second method to more precisely quantify the number of cells per organoid across each entire well. To do this, we labeled nuclei of all organoids across the wells, volumetrically imaged the entire well (entire Matrigel dome), and then used image analysis tools to computationally count the number of cells within each well. Using this metric for analysis, we found that the number of cells in the population expanded ~1.1 – 1.5 fold over two weeks of culture (Fig. S2h). Thus, hepatocyte organoids cultured in this manner predominantly underwent an increase in metabolic activity over time in culture (i.e., the amount of ATP per cell increases over time), along with a very modest increase in absolute cell number. Alongside AHHO, approximately 1–2 distinct cystic ductal organoids formed in each well that were CK19, CK18, and E–cad positive, and MRP2, Alb, and Arg1 negative (Fig. S2i–S2m). These structures accounted for only ~0.16% of cellular objects (Fig. S2i) and likely arose from contaminating cholangiocytes, which can form organoids with cystic ductal morphologies and have been well characterized previously [23,37–39]. Overall, imaging and immunostaining demonstrated that adult human hepatocytes formed organoids that grew over time in culture, and that these cells maintained robust adult primary hepatocyte morphology and expression of hepatocyte markers. To better assess the 3D morphology and phenotype of cells within organoids, we immunostained, cleared, 3D imaged and computationally reconstructed AHHO. These studies further demonstrated conservation of hepatocyte morphology and protein localization, with cytoplasmic Arg1 staining (Fig. 1e), nuclear expression of hepatocyte transcription factor HNF4α (Fig. 1f), and strong expression of E–cad at cell–cell junctions throughout the organoid depth and at various sizes (Fig. 1e–f). AHHO form from a diverse set of post–pubescent adult donor hepatocytes A major issue confounding development of adult human hepatocyte model systems and therapies has been variability in the viability and the functional quality of adult human hepatocytes cryopreserved from different human donors [42–44]. While some functional differences are expected due to natural variation in human biology [42,45,46], hepatocytes from many donors are non–adherent or non–viable in 2D culture, or cannot engraft after implant [31,43,47]. New strategies to culture adult hepatocytes across donors could recover previously unusable cells and enable longer–term hepatic studies on samples with broader genetic and ancestral variation [48,49]. We thus next sought to determine whether AHHO culture conditions could reliably generate organoids from seven post–pubescent adult donors with varying age (19–49 years old), sex, and racial/ethnic demographics (Suppl Table 1). We found that adult human hepatocytes formed organoids under AHHO culture with compact hepatic morphology across all lots, as observed by brightfield morphology (Fig. 2a, S3a). Immunostaining demonstrated that organoids from all lots expressed the epithelial marker CK18 and that individual cells had adult hepatocyte morphology (Fig. 2b). Similar to our earlier studies, cystic ductal organoids expressing CK19 were rare (<0.2%) and found in cultures for all human donor lots (Fig. 2b). Thus, adult human hepatocytes from all seven donors successfully formed organoids when grown in AHHO culture conditions. Conversely, AHHO had little to no AFP expression, similar to adult mature hepatocyte and human liver controls (Fig. 2c). We next sought to interrogate gene expression of two enzymes responsible for the metabolism of ~40% of drugs in humans, cytochrome P450 genes CYP1A2 and CYP3A4 [50]. The majority of CYP enzymes, aside from fetal isoforms such as CYP3A7, are expressed at low levels in fetal developmental stages and increase with hepatocyte maturation [51–55]. Consistent with embryonic and fetal liver stages, human hepatoblasts had low to no CYP1A2 or CYP3A4 expression. In contrast and importantly, AHHO exhibited high gene expression in all lots, and at similar levels as adult controls (Fig. 2c). This maintenance of mature gene expression and dearth of fetal genes demonstrates successful preservation of adult liver transcription across all lots and differentiates AHHO culture from models with an immature phenotype. AHHO from seven separate human donors form reliably and demonstrate robust hepatic functions. (a) Representative brightfield images across all seven lots screened with donor age listed above image. Images were taken in the center of Matrigel droplets on day 15. Scale bars = 100 μm. (b) Immunofluorescence images across all seven donor lots screened stained for CK18 (cyan), CK19 (magenta), and Hoechst (yellow). Scale bars = 100 μm. (c) Gene expression analysis by qRT–PCR across all donors lots, hepatoblast organoids (HBO), freshly thawed primary human hepatocytes (PHH), and whole liver RNA (HuLiv) relative to GAPDH for each sample. Data represented as mean ± SEM of 2 or 3 replicates, each measured in technical triplicate. Levels of ALB and TF in AHHO were not statistically significantly different from HBO. However, AFP levels of HBO were significantly greater than all lots of AHHO and PHH. CYP1A2 levels of HBO were significalty lower than lot 1, 2, and 6 of AHHO, and CYP3A4 levels of HBO were significantly lower than lot 3, 4, 6, and 7 of AHHO. a: p < 0.05 compared to HBO, b: p < 0.01 compared to HBO. (d) Albumin secretion across four donor lots plated in tissue culture plates (PHH) or as AHHO. Data represented as mean ± SEM. 6 replicates at day 3 (black) and day 15 (white). Two–way ANOVA, Sidak's multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (e) Urea secretion across four donor lots plated as PHH or as AHHO. Data represented as mean ± SEM. 6 replicates on day 3 (black) and day 15 (white). Two–way ANOVA, Sidak's multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001. (f) Analysis of CYP3A4 activity after rifampin (left) or dexamethasone (right) induction in four donor lots plated as PHH or as AHHO by P450–Glo assay. Data normalized to cell numbers is represented as mean ± SEM with 6 replicates each of vehicle and treatment. Two–way ANOVA, Sidak's multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (g) Confocal maximum intensity projection of CDFDA (green) accumulation in organoids, overlaid with brightfield image for donor lot 7 (top right). CDFDA maximum intensity projection for donor lots 3, 4, and 6 on bottom. Organoids outlined with white dashed line. Scale bars = 50 μm. (h) Top: Glycogen storage in organoids, visualized with PAS stain (pink) and Hematoxylin counterstain (blue) (top, scale bar = 20 μm). Bottom: Low density lipoprotein (LDL) accumulation in organoids, visualized with Dil–LDL staining (red) and Hoechst (blue) (bottom, scale bar = 50 μm). Next, we explored a panel of functional assays to assess the extent to which hepatocytes maintain important liver functions after organoid culture. Human albumin protein was measured over time to explore the secretory function of AHHO along with freshly thawed PHH as controls. All lots increased albumin production as organoids grew, with a range of 13 to 31–fold increase in AHHO in contrast to 0.3 to 4 fold increase in PHH from day 3 to day 14 (Fig. 2d). The lots with the highest albumin secretion at day 14 were also found to have the highest percentage of hepatocytes expressing Ki67 at day 5 (correlation coefficient R2 = 0.8831), suggesting that the substantive increase in albumin over time may at least partly be due to cell growth and organoid expansion (Fig. S3b). Synthesis of urea, a byproduct of ammonia detoxification, was also present in cultures derived from all tested lots plated as in AHHO as well as in PHH, with differential patterns of urea secretion across lots (Fig. 2e). Urea levels showed more variability, such that some lots exhibited greater levels in AHHO compared to PHH whereas others showed opposite trends (Fig. 2e). To assess hepatic xenobiotic metabolism, we next explored the activity of CYP3A4, the enzyme responsible for 30–60% of drug metabolism in the human liver [56]. AHHO and PHH from four donors were exposed to known CYP3A4 inducers rifampin and dexamethasone, and CYP3A4 activity was measured. AHHO exhibited induction by both drugs, with inter–donor variability in induction levels, as would be expected for mature adult human hepatocytes in which function has been maintained (Fig. 2f) [42,57–60]. AHHO showed similar or better induction of CYP3A4 activity induced by rifampin and dexamethasone compared to PHH at D15 (Fig. 2f). Finally, we examined CYP2C9 activities across different lots in AHHO and found that Lot 4 and 7 exhibited significant CYP2C9 induction with Rifampin, again with lot to lot variability (Fig. S3c). After modifications by CYP enzymes and other biotransformations, many xenobiotics are excreted from hepatocytes by efflux transporters in the multidrug resistance protein (MRP) family [61,62]. MRP2, a transporter that usually exports substrates from within hepatocytes out to bile canaliculi [61], was assessed through incubation of AHHO with fluorescent MRP2 substrate, carboxy dichlorofluorescein diacetate (CDFDA). Fluorescent signal localized to networks or puncta on the interior of the organoids, indicating active MRP2–driven efflux and validating the morphological polarity observed with MRP2 immunostaining (Figs. 2g and 1d). Finally, glycogen accumulation and low–density lipoprotein (LDL) uptake were observed with periodic acid Schiff (PAS) staining and Dil–LDL incubation, respectively, as assessments of hepatocyte functions (Fig. 2h). In sum, AHHO from multiple adult donors demonstrated maintenance of a wide variety of hepatic functions, illustrating their potential as a model of functional human hepatocytes and as a tool for basic research applications and drug screening. Single–cell RNA–sequencing reveals transcriptional maturity and functional heterogeneity in AHHO To further explore the distribution of hepatic functions in AHHO on a cell–to–cell basis [63], we performed single–cell RNA sequencing (scRNAseq) on our organoid population using the 10X Genomics Chromium platform. After analysis by principal component analysis (PCA) and clustering with uniform manifold approximation and projection (UMAP), cells were segregated into six distinct populations (Fig. 3a and S4a). We characterized each cluster by examining differential gene expression and identifying known cell type–specific markers. Clusters 1, 2, and 5 represented hepatocytes with high expression of hepatic markers such as ALB, ALDH1A1, APOC1, and ASGR1. Clusters 0 and 3 were composed of cholangiocytes, with low expression of the above hepatic markers and high expression of biliary markers KRT7, EPCAM, and TFF1–3. Cluster 4 was composed of a stromal population expressing COL3A1, IGFBP7, and DCN (Fig. S4b and S4c). Single–cell RNA sequencing demonstrates transcriptomic maturity and functionality of AHHO. (a) UMAP clustering of single cells from AHHO at day 5 and day 15 in culture. Green, pink, and yellow clusters represent hepatocytes. (b) Hepatocyte clusters from AHHO on day 5 (light orange) and day 15 (dark orange) were subset from initial UMAP (Fig. S4f) and then reclustered for additional analysis with fetal hepatoblast organoids (light grey) and primary adult hepatocytes (dark grey) from seven human liver samples. Insets of AHHO for clearer visualization. (c–f) UMAP plots of AHHO from (a), emphasizing expression of individual genes of interest, and violin plots from (b) showing distribution of expression compared to controls. Genes are divided by hepatic functional axis: (c) plasma protein synthesis, (d) lipid metabolism, (e) amino acid metabolism, and (f) drug metabolism. Violin plot groups color–coded as in (b) with Day 5 (d5) and 15 (d15) AHHO, Fetal = Fetal hepatoblast organoids, and Adult = Adult human liver. (g) Violin plots from (b) showing distribution of fetal hepatic gene expression in AHHO compared to controls. Groups color coded as in (b) with Day 5 (d5) and 15 (d15) AHHO, Fetal = Fetal hepatoblast organoids, and Adult = Adult human liver. Having classified the cells in AHHO based on known hepatocyte markers, we next sought to compare the transcriptional profiles of AHHO to hepatocytes from adult liver. We first compiled and integrated seven published scRNAseq libraries of adult human liver [64–66]. We then clustered this adult human liver data with our AHHO samples as well as with a fetal hepatoblast organoid library as a comparison for maturity [40] (Fig. S4d). Cells clustered into 11 distinct populations, which we identified as hepatocytes, cholangiocytes, stromal/endothelial cells, and immune cells based on cell–type–specific gene expression [67] (Fig. S4e). Cells from our AHHO population clustered with adult liver hepatocytes, cholangiocytes, and stromal cells, validating that each population retains transcriptional similarity to its primary cell of origin (Fig. S4d). Hepatocytes from AHHO were distributed across the hepatocyte clusters from all adult datasets, despite significant variability in the individual adult human liver libraries (Fig. S4d). To further explore the distribution of functional gene expression specifically in the hepatocyte populations, the three hepatocyte clusters (0, 1, and 6) were then subset from the complete dataset (Fig. S4f). The reason for this was subsetting was because while cholangiocyte organoids represented a small minority of organoids in culture (0.16%, Fig. S2g), they were more readily digested to single cells for scRNAseq studies compared to the more densely packed hepatocyte organoids (AHHO). Thus, in our scRNAseq studies, the number of cells digested to single cells and thus analyzed was artificially skewed towards the biliary compartment. We therefore excluded the biliary–like cells for remaining analyses, which focused on hepatocyte phenotyping. In the resulting UMAP projection, AHHO were again found to distribute evenly with hepatocytes from adult liver datasets (Fig. 3b). Genes involved in important hepatic processes were then compared between AHHO, fetal hepatoblast organoids, and hepatocytes from the integrated adult liver datasets by visualizing the distribution of expression of genes representing various axes of hepatocyte metabolic function. Secreted plasma protein genes ALB, TTR, and RBP4 represent key hepatocyte functions and serve as early indicators of endodermal commitment toward hepatic cell fate in development [68]. All three genes had strong expression across the majority of cells in the hepatocyte clusters and exhibited similar gene expression to adult liver and fetal controls (Fig. 3c). Similarly, genes involved in lipid and fatty acid metabolism that are expressed in the fetal liver and throughout adult life, APOA2, APOC1, and FABP1, were strongly expressed across the entire hepatocyte population and at similar levels to adult liver and fetal controls (Fig. 3d) [40, 69]. Genes involved in glutathione and amino acid metabolism, GAMT, MGST1, and HPD, all had high expression in AHHO with similar expression to hepatocytes from adult liver. Interestingly, HPD, an enzyme involved in tyrosine metabolism, had distinctly lower expression in fetal hepatoblast organoids compared to all other conditions including AHHO, which again highlighted the mature phenotype of AHHO (Fig. 3e). We thus set out to query functional genes more carefully with known expression changes throughout hepatocyte development and maturation. As the liver develops and matures pre– and postnatally, many functional genes upregulate or downregulate to adjust to the hepatic requirements for the corresponding stage of development. Regulation of cytochrome enzymes and other genes involved in drug metabolism is especially dynamic as the liver matures. For example, gene expression of enzymes CYP2E1 and CES1 does not rise in the liver until after birth, while CYP3A7 is expressed highly in fetal liver and downregulates postnatally [40]. We found that both CYP2E1 and CES1 were similarly expressed by day 15 AHHO and adult human liver, while fetal controls had little to no expression (Fig. 3f). In contrast, expression of fetal drug–metabolizing enzyme CYP3A7, as well as other fetal hepatic genes AFP, H19, GPC3, DLK1, and SPINK1, was low or absent in AHHO hepatocytes and control hepatocytes from adult liver (Fig. 3g). All fetal hepatic genes had high expression in fetal controls (Fig. 3g). Taken together, these results validate that AHHO exhibit differentiated and mature transcription across multiple axes of liver function, and that day 15 organoids represent a heterogeneous population in which hepatocytes coexist with cells with biliary– and stroma–associated transcription signatures, rather than a uniform loss of hepatocytic identity. In particular, the expression of developmentally–dynamic drug–metabolizing genes emphasizes the value of AHHO for pharmaceutical applications such as drug screening. AHHO repopulate livers of immunodeficient Fah−/− mice Over the past several decades, methods have been developed to engraft and expand PHH in damaged or diseased mouse livers to generate mice that serve as humanized model systems for investigating new therapies to treat liver disease (Fig. S5a). One such model are mice lacking fumarylacetoacetate hydrolase (FAH), which experience chronic liver injury (tyrosinemia) unless treated with nitisinone [70]. When human hepatocytes transplanted in immunodeficient Fah−/− mice possess a growth advantage due to their functional human FAH gene, they can engraft and expand, ultimately replacing mouse hepatocytes (Fig. S5a). The humanization capacity of transplanted hepatocytes correlates with their maturity level [12,13,36], and thus the extent to which a given hepatic cell population can engraft and repopulate immunodeficient Fah−/− mouse livers has become a surrogate way to assess phenotypic and functional quality of hepatocytes [36]. Previous studies have shown that at least 10% chimerism with human cells is necessary for most humanized mouse model applications, such as HCV infection and drug metabolism studies [71,72]. Thus, we next evaluated the extent to which human hepatocyte organoids could repopulate the livers of Fah−/− NOD Rag1−/− Il2rg null (FNRG) mice [73]. We cultured AHHO for two weeks then digested the organoids into a single cell injectate, which was transplanted into FNRG mice by intrasplenic injection. We also injected freshly thawed cryopreserved hepatocytes from the same donor as a positive control. Human albumin (hAlb) present in mouse blood was then measured over time, as this closely correlates with the level of liver humanization [12,71,72, 74]. Although the initial hAlb levels were lower for organoids compared to freshly thawed cells, possibly due to a different number of initially engrafted nodules, the hAlb levels continued to rise even when those of the freshly thawed cells plateaued (Fig. S5b). By day 130, serum hAlb of AHHO had risen 1900–fold to an average of more than 8 mg/ml. Remarkably, we found that AHHO achieved hAlb levels that correspond with approximately 90% chimerism (that is, 90% of the mouse liver is “humanized”). Consistent with hAlb levels, histological staining confirmed the majority of the liver parenchyma to be human (Fig. S5c). In summary, AHHO can successfully repopulate injured mouse livers, ultimately achieving hAlb levels of >8 mg/ml and repopulation levels of ~90% humanization. AHHO in engineered liver tissues rapidly adopt liver morphology and functional maturity Encouraged by these transplantation results, we finally sought to examine the utility of AHHO for improving the ectopic engraftment of human engineered tissues. To do this, we examined the utility of AHHO in transplanted engineered liver tissues, which we and others are developing as a bridge or alternative to organ transplantation for patients suffering from diverse forms of liver disease [19,75]. AHHO were grown from a 34–year–old donor and encapsulated within fibrin hydrogels to create engineered liver tissues. These tissues were implanted in the gonadal fat pad of FRGN mice [74] to assess their engraftment potential in pre–clinical transplantation studies [75–77]. Human albumin increased over 4 weeks in vivo, suggesting both a functional increase over time as well as a connection to host vasculature (Fig. S6a). Excised grafts displayed large areas in which most cells stained positively for antigens expressed by hepatocytes, CK18 and Arg1 (Fig. S6b–S6c). Occasional CK19+ cells with hepatocyte morphology were also present in grafts, suggesting that some engrafted cells adopt a progenitor phenotype [78]. The mouse red blood cell marker Ter119 stained small, localized areas throughout the graft between Arg1+ hepatocytes, indicating that implanted cells were being supplied with host blood (Fig. S6b). The large size of grafts in this initial study were unexpected in the context of past work [19,75,76], (Fig. S6), given that we had not included any cell populations beyond AHHO themselves within the engineered tissues. Previous studies had shown that optimal engraftment of engineered tissues built from other hepatocyte sources, such as human hepatocyte aggregates [18,75] or iPS–derived hepatocytes [19] benefited from the addition of endothelial and stromal cells (e.g., fibroblasts, mesenchymal stromal cells) prior to tissue transplantation. The ability to remove these cell populations from cell production pipelines could simplify clinical development and possibly also reduce safety concerns of tumorigenicity and stromal cell overgrowth [18,75,79]. We thus set out to directly compare engraftment efficiency of engineered tissues with and without additional endothelial cells and fibroblasts (EC/FB). We also derived tissues using two different primary adult human hepatocyte culture methods – AHHO developed here, or “aggregates”, which we have previously shown have equal to, and in some cases even superior engraftment, compared to freshly thawed human hepatocytes and thus represent the most robust control we were aware of at the time of this study [18,75,76] (Fig. 4a and S7a). Engineered liver tissues generated with AHHO demonstrate superior function and engraftment than those generated with hepatocyte aggregates. (a) Schematic illustrating the in vitro culture of adult human hepatocytes as either AHHO or aggregates and the subsequent fabrication and implantation of engineered liver tissues with either aggregates or organoids, with or without endothelial cells and fibroblasts (EC/FB). (b) Secreted hAlb measured in mouse blood by ELISA. Data represented as mean ± SEM of each mouse at week 1 (black) and week 2 (white). Two–way ANOVA, Sidak's multiple comparisons test, ****p < 0.0001. (c) Secreted human A1AT measured in mouse blood by human A1AT ELISA. Data represented as mean ± SEM of each mouse at week 2. One–way ANOVA, Tukey's multiple comparisons test, ****p < 0.0001. (d) Average graft area calculated from immunofluorescence images. Hepatic area measured as CK18+/CK19– area (cyan). Biliary area measured as CK18+/CK19+ area (magenta). Each data point represents the average graft area of all grafts found in a single mouse. Data represented as mean ± SEM of each mouse at week 2. Two–way ANOVA, Sidak's multiple comparisons test, **p < 0.01, ****p < 0.0001. (e) Average hepatic area and biliary area as a ratio of total epithelial area (CK18+) in each group. Hepatic area measured as CK18+/CK19– area (cyan). Biliary area measured as CK18+/CK19+ area (magenta). (f) Representative fluorescence images used for quantification in (d) and (e). CK18 (cyan), CK19 (magenta), and Hoechst (yellow). Scale bars = 200 μm for low magnification, 50 μm for inset images. Example masks used for quantification in upper left corner of each image. (g) Confocal images of serial sections of organoid-only graft stained for CK18 (cyan) and CK19 (magenta) (left), Alb (magenta) and HNF4α (cyan) (middle), or Arg1 (magenta) and E–cad (cyan) (right). Nuclei stained for Hoechst (yellow) in all images. Scale bars = 100 μm. Inset shows serial sections of neighboring biliary area at the same scale. Both AHHO and aggregates were generated from the same 19–year-old human hepatocyte donor. Notably, this was a different donor than was used in the previous in vivo experiment (Fig. S6), thus together these studies enabled assessment of engraftment of two different adult donors. Again, blood was drawn from implanted mice to assess hAlb production as a metric for hepatic functionality of implants. After one week of engraftment, the group of animals receiving engineered tissues with only AHHO had on average >2–fold more hAlb than AHHO + EC/FB group or either aggregate group (Fig. 4b). By week two the differences between groups were even more pronounced with AHHO–only group producing ~6x more hAlb than AHHO + EC/FB group, as well as significantly more than both aggregate ± EC/FB control groups, at up to ~25x more hAlb (Fig. 4b). Similarly, human alpha–1 antitrypsin (A1AT) was also identified in the mouse blood of AHHO–only group at ~6x more than AHHO + EC/FB, and greater than both aggregate ± EC/FB groups (up to ~12x difference; Fig. 4c). To further explore the engraftment efficiency of human cells within tissue grafts from each group, implanted tissues were excised and immunostained. Human CK18, a marker of both hepatocytes and cholangiocytes, was identified in grafted areas in all mice (Fig. 4d–f). Using CK19 as a marker to differentiate between hepatocyte and biliary areas, graft size and phenotype were quantified in all groups. Importantly, and in corroboration with secreted protein data (Fig. 4b and c), hepatic graft area (CK18+/CK19–) was found to be significantly larger in organoidonly grafts compared to all other groups (Fig. 4d). Significant biliary area (CK18+/CK19+) was also found in both organoid groups, comprising 13.5% of organoid + EC/FB grafts and 7.6% percent of organoid–only grafts (Fig. 4d and e, and S7b). While the presence of secreted human proteins albumin and A1AT in mouse blood suggested that the CK18+/CK19– cells, which constituted >86% of the human cell area in organoid grafts, were indeed human hepatocytes in engrafted tissues, we sought to further establish their hepatocyte identity via additional immunostaining (Fig. 4f and g). We found that these cells also expressed hepatocyte functional proteins Alb and Arg–1, transcription factor HNF4α, and epithelial cell–cell junction protein E–cad, and were generally negative for CK19, all consistent with a human hepatocyte phenotype (Fig. 4g). Furthermore, cells that stained positively for CK18, hAlb, Arg1, HNF4α, and E–cad also exhibited hepatocyte morphology, occasional binucleation, and tightly packed cells that collectively self–assembled to form elongated structures reminiscent of hepatocyte cords in the mature liver (Fig. 4g). Conversely, CK19+ cells were negative for hepatocyte markers but positive for E–cad (Fig. 4g, insets) and had small, densely packed cuboidal cell morphologies, all consistent with cholangiocytes. In explanted tissue grafts, these cells were generally found in self-assembled tubular structures resembling bile ducts, with diameters ranging from ~15 to 800 μm. Notably, the percent of biliary area in organoid–only grafts was significantly lower than in AHHO + EC/FB group (Fig. 4e and S7b), suggesting that non–parenchymal cells could play a role in regulating the ratio of human hepatocyte versus cholangiocyte populations in engrafted tissues. Taken together, these findings demonstrate that 1) “priming” adult human hepatocytes in vitro as organoids provides a highly functional cell source in engineered liver tissues that robustly engrafts and spatially reorganizes to create structures that morphologically, phenotypically, and functionally resemble both human hepatocyte and cholangiocytes populations and, 2) addition of exogenous non–parenchymal cell populations are not needed for tissue engraftment or assembly. Discussion Here, we describe a novel AHHO culture method that maintains the phenotype, morphology, transcriptome, and functional characteristics of mature human hepatocytes while also facilitating in vitro manipulation and mobilization. AHHO improved the engraftment of human hepatocytes within engineered tissues in preclinical transplantation studies [75]. In the liver field, human organoid culture methods have been developed for hepatoblasts [40] and fetal and pediatric hepatocytes, and more recently adult hepatocytes [26,28,29,36]. Most recently, adult human hepatocytes from several post–pubescent donors could be propagated in vitro using organoid cultures [28,30]. In most of these studies, the focus has been on generating, propagating in vitro, and phenotyping the resultant organoids. Often these cells are then engrafted into the liver, typically via direct injection into the spleen [26,28–30,36]. Here, we generate adult human hepatocyte organoid cultures and instead demonstrate how such cultures can be deployed for implantation in bioengineered tissues [28]. Notably our organoid culture method is also distinct from prior reports in that it focuses on efficiently creating highly functional, mature cells for engraftment in engineered tissues in vivo, as opposed to being optimized for long–term organoid propagation in vitro. Moving forward, it would be interesting to work to dovetail methods for longer term culture of adult hepatocyte organoids [26] together with our methods here for bioengineered tissue engraftment in vivo. Our findings demonstrate that transient AHHO culture of hepatocytes results in superior engraftment outcomes in engineered tissues compared to culture of hepatocyte aggregates. We note that previous studies have shown hepatocyte aggregates, in turn have equivalent or superior engraftment outcomes compared to freshly thawed primary human hepatocytes [17,18]. This work thus has important translational implications since enhancing hepatic engraftment efficiency without the need for other exogenous cell populations could save time and cost in engineering tissues [12,13]. The molecular mechanisms facilitating the improved engraftment observed in these studies are not yet known. One potential mechanism whereby organoid culture may prime hepatocytes for superior engraftment is that AHHO culture recovers hepatocytes that have been damaged by cryopreservation or anoikis [80–83], where ice crystal formation or downregulation of important adhesion molecules such as E–cadherin [31,81] lead to poor cryo–recovery and plating efficiency after thaw. Recovery from such damage could potentially leadto their improved metabolic activity (Fig. S2g). AHHO culture could also serve as a purification step to enrich healthy, expandable hepatocytes and dilute those cells that could not recover from cryopreservation. Interestingly, markers suggest a mixture of periportal (ARG1, HPD) and pericentral (CYP2E1, xenobiotic metabolism) features for the AHHO. This mixed zonation characteristic of AHHO may provide another benefit for the enhancement of survival for the cells in the graft in vivo. Finally, it is possible that non–parenchymal cells that grow up within AHHO alongside hepatocytes support these cells better than the additional exogenous cells. Notably, AHHO also demonstrate an improvement in transcriptional maturity throughout the culture timeline, which has been suggested to improve cell transplantability and engraftment. AHHO could broadly enable studies in areas such as therapeutic regenerative medicine, pharmaceutical screening, and generating patient–specific organoids enabling generation of implants for precision medicine. Despite the progress here, it will be important to take additional steps before these translational applications are realized. For example, prior to use in pharmaceutical screening, it will be important to perform comprehensive analysis of phenotypic characteristics needed for drug metabolism and bile formation by examining expression of additional Cytochrome P450 enzymes (e.g., CYP1A2, CYP2E1) and bile salt export pump (BSEP). Prior to clinical translation for therapeutic applications, it will be important to develop cost–effective and GMP compatible culture conditions, such as identifying alternative reagents that can be substituted for Matrigel and conditioned media currently used for AHHO culture. It will also be important to assess both passage and expansion efficiency over extended periods of culture to determine the feasibility of this approach for autologous versus allogeneic cell therapies, the latter of which would require the ability for massive expansion from a single donor. For allogenic therapy, it would also be important to address immunogenicity, for example by genetic editing and immune ‘cloaking’ to generate hypoimmunogenic allogenic grafts [84,85]. Finally, it will be important for future studies to establish the potential efficacy of these approaches in animal injury animals, such as by therapeutic rescue of animals with acute liver injury induced by acetaminophen administration [86]. A second potential limitation is that while this culture method did sustain adult human hepatocytes for several weeks in culture, some other non–parenchymal cell populations also expanded over the course of culture, such as cells with biliary and stroma–associated transcriptional signatures. Thus, studies that require perfectly pure cultures of only hepatocytes may need to adjust the culture conditions and/or leverage purification strategies. That said, we note that this could also be considered an advantage of our system, which maintains adult human hepatocytes in organoids in the presence of some non–parenchymal support cells, as in situ in human liver. The additional presence of nonparenchymal cells could lead to more biologically relevant and functional human liver models and/or therapeutic tissues and organs. Taken together, our results suggest that AHHO could broadly enable a variety of new basic science and clinical studies. Experimental model and subject details Mice All animal procedures were conducted according to protocols approved by the University of Washington (4388–01), Weill Cornell Medicine (2015–0031) and Rockefeller University (21056–H) Institutional Animal Care and Use Committees. Mice for implant surgeries were 12– to 20–week–old Fah−/−, Rag2−/−, Il2rgnull on the NOD strain background (FRGN) from Yecuris (UW) and for intrasplenic injection 6– to 8–week old Fah−/− NOD Rag1−/− Il2rgnull (FNRG) mice (WCM and RU). Cell lines HA–R–spondin1–Fc 293T cells (Cultrex, R&D 3710–001–01) and L-Wnt3a cells (ATCC CRL–2647) were used to generate conditioned medium of R–spondin1 and Wnt3a (described below). Human umbilical vein endothelial cells (HUVEC, Lonza C2519A) were grown in EGM–2 media (Lonza) and neonatal normal human dermal fibroblasts (NHDF, Lonza CC–2509) were grown in DMEM + 10% FBS + 1% penicillin-streptomycin. HUVECs and NHDFs were used in engineered liver tissues between passages 4–8. Primary cells Cryopreserved primary human hepatocytes were purchased through Thermo Fisher (Cat# HMCPIS (lots Hu8366, Hu8373, Hu8300, Hu8287, and Hu1880) HMCPMS (lot Hu8360), and HMCPTS (lots Hu8339 and Hu 8375)) and cultured as organoids, as described below. Methods Conditioned media Rspo1 conditioned media was generated according to manufacturer's instructions. Briefly, Rspo1 expressing 293T cells (Sigma Aldrich, SCC111) were grown in dishes until confluent, given 50 ml of conditioning medium (DMEM/F12 and 1% GlutaMAX) and incubated at 37°C for 7–10 days. Wnt3a conditioned media was generated by growing Wnt3a expressing cells (L Wnt–3A, ATCC CRL–2647), to confluence, replacing their media with conditioning medium (DMEM + GlutaMAX, 10% FBS, 1% penicillin–streptomycin, 1% non–essential amino acids, 1% sodium pyruvate, and 1% HEPES) and incubating at 37 ° C for 4 days. Conditioned media was tested using the Super TopFlash assay with HEK 293 STF cells (ATCC CRL–3249), according to manufacturer's instructions. AHHO and PHH culture of hepatocytes Cryopreserved hepatocytes were thawed into 37°C medium and quickly spun down at 70 × g. In PHH, cells were plated in tissue culture plastic plates coated with type I collagen and cultured in ITS media (high–glucose DMEM (Corning) containing 10% (vol/vol) FBS (Bio-West), 1% (vol/vol) ITS supplement (insulin, transferrin, sodium selenite; BD Biosciences), 0.49 pg/mL glucagon, 0.08 ng/mL dexamethasone, 0.018 M Hepes, and 1% (vol/vol) penicillin–streptomycin (pen–strep; Invitrogen), as previously reported [17]. In AHHO, cells were mixed with 45% organoid media and 55% Matrigel (Corning, 356231) and seeded in 20 μl droplets in each well of 48–well plates. After Matrigel had solidified, 200 μl of organoid medium was added to each well. Fetal hepatoblast organoid medium [40], fetal hepatocyte organoid medium [36], and AHHO medium composition all shared the same basal medium (DMEM/F12, 1x GlutaMAX, 1x HEPES, 1x penicillin–streptomycin, 2% B27, 10 mM nicotinamide, 1.25 mM N–acetyl cysteine, 10 nM gastrin) and further medium supplements are summarized below. Hepatoblast organoid (HBO) medium: 40% Basal medium, 50% Wnt3a conditioned medium, 10% Rspo1 conditioned medium, 50 ng/ml EGF, 50 μM A83–01, and 10 μM y–27632. Fetal hepatocyte (FH) medium: 85% basal media, 15% Rspo1 conditioned medium, 10 μM y–27632, 3 μM CHIR99021, 2 μM A83–01, 50 ng/ml EGF, 50 ng/ml HGF, 100 ng/ml FGF7, 100 ng/ml FGF10, and 20 ng/ml TGFα. AHHO medium: 40% Basal medium, 50% Wnt3a conditioned medium, 10% Rspo1 conditioned medium, 50 ng/ml EGF, 5 μM A83–01, and 10 μM y–27632. In the initial experiment comparing FH and HBO media, basal media supplemented with A83–01 and all conditioned media were sterile–filtered prior to adding growth factors. In the subsequent experiments, A83–01 was added after filtration to prevent absorption to the filter membrane. Medium was refreshed every 2–3 days with care not to disturb the Matrigel droplet. Organoid size and growth were quantified from minimum intensity projections of brightfield z–stacks taken on a Nikon Eclipse Ti inverted high–resolution widefield microscope. Brightfield microscopy and morphometric analysis Brightfield z–stack images were taken in the center of each organoid well at multiple timepoints on a Nikon Eclipse Ti inverted high-resolution widefield microscope. Z–stacks were flattened into a single image using a minimum intensity projection (MinIP) with FIJI or Nikon NIS–Elements software. Organoid size and growth were quantified from MinIP images in ImageJ software and graphed with GraphPad Prism software. Organoid histology, immunostaining, and microscopy Organoids were harvested from Matrigel and fixed in 4% PFA at room temperature for 30 min, then embedded in HistoGel (VWR). For 2D histology, HistoGel pellets were dehydrated in ethanol, embedded in paraffin, and sectioned using a microtome (5–6 μm sections). Immunofluorescence staining of organoids was performed by deparaffinizing sections, performing antigen retrieval, blocking with normal donkey serum for 1 h at room temperature, then incubating with primary antibody overnight at 4° C. The following day, sections were washed with PBS–T and then incubated with secondary antibody + Hoechst 33342 (Thermo H3570) for 1 h at room temperature, washed, and mounted with Fluoromount–G (Invitrogen). For 3D imaging, organoids in HistoGel were blocked in bovine serum albumin and normal donkey serum overnight at room temperature then incubated with primary antibodies for 24 h at 37° C. After washing for 6 h at room temperature, organoids were incubated with secondary antibodies overnight at 37 °C. Finally, organoids were washed for 6 h at room temperature and then cleared by bathing in Ce3D solution [87] overnight at room temperature with Hoechst 33342. All antibody information is included in Table 1. Images were obtained using a Nikon Eclipse Ti inverted high–resolution widefield microscope, a Nikon A1R scanning confocal microscope, or a Leica SP8 confocal microscope. Images were processed using Adobe Photoshop or ImageJ software. For morphometric analyses, images were thresholded and pixels were measured using ImageJ software. Table 1Antibody information.AntibodyVendor/Product #SpeciesDilution Human cytokeratin-18Dako M701029-2Mouse1:25Cytokeratin-19Abcam ab52625Rabbit1:100Human albuminBethyl A80-129AGoat1:100Arginase-1Sigma HPA003595Rabbit1:200MRP2Thermo TA812520Mouse1:500HNF4aAbcam ab201460Rabbit1:100Ter119BD Pharmingen 550565Rat1:100EcadR&D AF748Goat1:100Ki67Abcam ab16667Rabbit1:500Goat anti-mouse 555Invitrogen A21127Goat1:500Donkey anti-mouse 555Invitrogen A31570Donkey1:1000Donkey anti-rabbit 594Invitrogen A21207Donkey1:1000Donkey anti-rabbit 647Invitrogen A31573Donkey1:1000Donkey anti-goat 488Invitrogen A11055Donkey1:1000Donkey anti-goat 647Invitrogen A21447Donkey1:1000Donkey anti-rat 488Invitrogen A21208Donkey1:1000 Antibody information. Functional analyses Human albumin and alpha–1–antitrypsin within organoid media or mouse blood were measured with ELISA (Bethyl E80–129 and E88–122). Urea was measured in organoid media using a BUN assay according to manufacturer's instructions (Fisher SB0580250). CYP3A4 or CYP2C9 induction was assessed by adding CYP inducers (25 μM rifampin or 25 μM dexamethasone) or vehicle controls (DMSO or EtOH) to organoid cultures for 3 days after which CYP activity was measured on organoid cultures using the P450–Glo CYP3A4 or CYP2C9 Assay (Promega V9001, V8791). In some experiments, cell viability over culture was measured by quantifying ATP using CellTiter Glo 3D as directed by manufacturer (Promega G9681). Bile canalicular transport was assessed by incubating organoids for 30 min with 2 μg/ml 5–[and–6]–car–boxy–2′,7′–dichloro-fluorescein diacetate (CDFDA, Sigma 21884), a fluorescent MRP2 substrate. After incubation organoids were imaged on a Nikon Eclipse Ti inverted microscope with a Yokogawa W1 spinning disk head. Low-density lipoprotein (LDL) uptake was assessed by incubating organoids with 20 μg/ml Dil–LDL, a fluorescently labeled LDL (Thermo L3482), for 3 h at 37° C with Hoechst 33342 diluted 1:1000. After incubation organoids were imaged on a Leica SP8 confocal microscope. Finally, glycogen accumulation was assessed with periodic acid–Schiff staining (PAS, Sigma, 395B–1 KT) according to manufacturer's instructions. Functional analyses were performed between days 14–19 in culture. RNA isolation and qRT–PCR Organoids were lysed in Qiazol (QIAGEN) or Trizol (Thermo) and RNA was extracted using the phenol–chloroform extraction method. RNA was reverse transcribed into cDNA using random hexamer primers and a Superscript III first–strand synthesis kit (Thermo 18080093) according to manufacturer's instructions. qPCR was performed using SYBR–green (BioRad 1725122). The 2dCT method [88] was used to report fold change in gene expression compared to internal control gene GAPDH or RPLP0. Primer sequences are included in Table 2. Table 2Primer sequences.GeneForward primerReverse primer SERPINA1 GATCAACGATTACGTGGAGAAGGCCTAAACGCTTCATCATAGGCA AFP AGTGAGGACAAACTATTGGCCTACACCAGGGTTTACTGGAGTC ALB GAGACCAGAGGTTGATGTGATGAGTTCCGGGGCATAAAAGTAAG ARG1 GCCAAGTCCAGAACCATAGGAGCAGACCAGCCTTTCTCAA KRT19 TCCGAACCAAGTTTGAGACGCCCTCAGCGTACTGATTTCCT CYP3A4 CAGCCTGGTGCTCCTCTATCACCATCATAAAAGCCCCACA CYP1A2 ACCTTGTGACCAAGCCTGAGAAGGAGGAGTGTCGGAAGGT TF TGATTGCATCAGGGCCATTGGCCAGGTAAGCATCATACACCA Primer sequences. Organoid digestions and barcoding for single cell RNA sequencing Organoids were grown in Matrigel under AHHO culture conditions for 5 or 15 days, as described above. At each timepoint, organoids were pooled and suspended in TrypLE (Thermo) for 20–30 min until they had digested to single cells. The single–cell suspension was mixed with 1% BSA to prevent clumping and sticking to tubes/tips. Immediately before partitioning, cells were filtered through a FlowMi Cell Strainer (Belart) to remove cell clumps and debris. Cells were counted then loaded into a 10X Chromium flow cell using the Single Cell 3′ v3 kit. 10X genomics barcoding and library construction were performed according to manufacturer's recommendations and generated libraries from 2327 cells at day 5 and 3411 cells at day 15 in culture. Single–cell RNA sequencing and analysis Libraries were sequenced using an Illumina NextSeq 2000. Reads were aligned to the human genome GRCh38–2020–A and counted with the 10X Cell Ranger pipeline version 6.1.1 [89]. The day 5 organoid library generated 74,577 mean reads per cell with 1320 median genes per cell. The day 15 organoid library generated 41,264 mean reads per cell with 2296 median genes per cell. Cell Ranger outputs were loaded into R using the Seurat v4.0 package [90] and all bioinformatic analysis was performed following the recommended vignettes for data integration and clustering as described by the Satija lab. First, cell filtering was performed on individual libraries to remove cells with >500 genes and >25% mitochondrial RNA content to exclude fragments and dying cells from analysis; 1849 and 2796 cells remained from days 5 and 15, respectively. Next, global–scaling normalization and highly variable feature selection were performed. Cells were assigned a cell cycle score which was regressed during data scaling to lessen the influence of these genes from clustering and prevent cell cycle genes from masking variations in liver gene signatures between cells. Two integrated datasets were generated: an integrated day 5 and day 15 organoid sample, and an integrated sample of day 5 and day 15 organoids, a publicly available hepatoblast sample (ArrayExpress under accession E–MTAB–7189), and multiple human liver datasets (GEO under accession numbers GSM4808962 and GSM4808963, GSM4808967, GSM5615002, GSM5616004, GSM5616006, GSM4041154, and GSM4041159). Integrated datasets were scaled, principal component analysis was performed, followed by UMAP clustering, and gene expression was visualized to assign cluster identities to known liver cell types. All plots were generated using R Studio and the Seurat package. Human hepatocyte grafting into mice with liver injury via intrasplenic injection The organoid injectate was prepared following 30 min of digestion in Cell Recovery Solution on ice. After organoid digestion from Matrigel, cells were spun down at 70 × g. Organoids were resuspended in TrypLE, incubated for 5 min, and then strained with a 70 μm cell strainer. The cell strainer was rinsed with HBSS, 0.0025 mg/ml DNase I, 2% FBS (HBSS–DF) and again with HBSS, dropwise. Cells were sedimented at 70 × g for 5 min and resuspended in HBSS, followed by mechanical dissociation for hemocytometer quantification. Prior to transplantation 6–8 week–old female FNRG mice were maintained on an ad libitum chow diet with amoxicillin and drinking water containing 16 mg/mL NTBC (Yecuris). Mice were housed under a 12–h light cycle from 7:00 a.m. (ZT0) to 7:00 p.m. (ZT12). Retrorsine (catalog no. R0382; Sigma Aldrich) was dissolved at 20 mg/mL in 100% ethanol at 56 ° C and stored in aliquots at − 20 °C. Retrorsine was administered at 2–week intervals, typically the first dose at 3 to 4 weeks prior to transplantation and a second dose 2 to 3 weeks prior to transplantation (Fig. S5). FNRG recipients were withdrawn from NTBC 4 to 5 days prior to surgery. Briefly, under sterile conditions and using isoflurane anesthesia, the skin was cleaned with povidone iodine, and a 0.5– to 1–cm incision was created over the left flank, after which the peritoneum was mobilized and opened. The ventral tip of the spleen was mobilized onto the peritoneum, and ~65 μL of cell suspension of 0.5 million cells were injected into the spleen using a 28–gauge insulin syringe. After hemostasis was achieved by applying pressure to the injection site, the peritoneum was closed using 4.0 Vicryl suture (Ethicon) and the skin was closed using metal Mikron Autoclips (BD Biosciences). Mice received two doses of postoperative buprenorphine (Abbot Animal Health) 0.05 mg/kg for analgesia. Metal clips were removed 14 days after surgery. Cell aggregation To generate hepatocyte aggregates, primary human hepatocytes were thawed, counted, and plated into pyramidal microwells as described previously [18,75] at a concentration of 100 hepatocytes per microwell. If cells were aggregated with fibroblasts, NHDFs were tryp-sinized, counted, and added to pyramidal microwells at a concentration of 160 NHDFs per microwell immediately before adding hepatocytes. Cells were incubated at 37°C overnight to form aggregates. Fabrication of engineered liver seeds AHHO were grown for 2 weeks and then digested out of Matrigel (Corning) with Cell Recovery Solution (Corning), washed once in PBS, and resuspended in diluted fibrinogen. If HUVECs or NHDFs were included in engineered liver tissues, they were suspended in diluted thrombin. Diluted thrombin with or without non–parenchymal cells was mixed with an equal volume of AHHO in diluted fibrinogen and 60 μl was pipetted into 6 mm PDMS gaskets. After 30 min of incubation, fibrin/cell gels were released from gaskets and implanted in mice as engineered liver tissues. Final cell concentrations were 1.25 million HUVECs/ml, 3.6 million NHDFs/ml, or 2.25 million hepatocytes/ml resulting in individual 60 μl tissues that contained ~135K hepatocytes. Fibrin was used at a final concentration of 10 mg/ml. Implantation of engineered tissues and induction of liver injury FNRG mice were anesthetized with isofluorane. Three organoid tissues were sutured onto the gonadal fat pads of each mouse, resulting in each mouse being implanted with ~400K hepatocytes. Immediately after surgery, nitisinone (NTBC) was withdrawn from animals’ drinking water for 14 days to induce liver injury. For experiments that extended beyond 14 days, NTBC was then reintroduced to the drinking water to allow for mouse recovery, then removed again after 4 days for the remainder of the experiment. Tissue harvesting and histology Organoid tissues were identified using the surgical suture as a landmark then were excised and fixed in 4% paraformaldehyde for 48 h at 4° C. Excess fat was removed from around the implant. Trimmed grafts were then dehydrated, paraffin–embedded, and sectioned on a microtome. Sections were then immunostained with antibodies listed in Table 2 according to the methods in “Organoid histology, immunostaining and microscopy”. H&E histology of FNRG livers were performed as described [73]. Quantification and statistical analysis Morphometry Collected immunofluorescence images were analyzed using ImageJ software. Images were thresholded and masks were generated for each channel. Epithelial area was measured from CK18+ masks. Hepatic area was measured from CK18+ masks – CK19+ masks. Biliary area was measured from CK18+/CK19+ masks. Statistical analysis Data in graphs are expressed as the mean ± SEM, as denoted in figure legends. Statistical significance was determined with PRISM software using t-test, one-way ANOVA, or two-way ANOVA followed by Dunnetťs, Sidak's, or Tukey's multiple comparison test, as denoted in figure legends. Supplementary Material Supplementary data to this article can be found online at https://doi.org/10.1016/j.biomaterials.2026.124465. Acknowledgements This work was supported by funding from the National Institutes of Health (NIH) (K.R.S., R01DK128551; Y.J. R01AA027327; Y.J. P01HL160472), Wellcome Leap HOPE Program (K.R.S. and L.V.), W.M. Keck Foundation (K.R.S.), the Advanced Research Projects Agency for Health (ARPA–H) under award number D25AC00460-00 (K.R.S.), Allen Distinguished Investigator Award from the Allen Foundation, NIH NIBIB Cardiovascular training grant (S.H.S., T32EB001650), the Environmental Pathology/Toxicology NIEHS training program (T.N.M., T32 ES007032), and the NSF Postdoctoral Fellowship in Biology (T.N.M, 2109919). We thank Dr. Dale Hailey and the UW ISCRM Garvey imaging core for assistance with microscopy, Dr. Mary Regier and the UW ISCRM genomics core for guidance and assistance with single cell RNA sequencing, and the ISCRM histology core for shared equipment. We thank all the animal husbandry and veterinary staff in The Department of Comparative Medicine's vivarium. We thank all human liver tissue donors for their generous contributions to science. All schematics are created with Biorender.com. Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Kelly Stevens reports financial support was provided by National Institutes of Health. Kelly Stevens reports financial support was provided by Wellcome Leap Fund. Kelly Stevens reports financial support was provided by W M Keck Foundation. Kelly Stevens reports financial support was provided by Advanced Research Projects Agency for Health. Kelly Stevens reports financial support was provided by Allen Foundation Inc. Tara McCray reports financial support was provided by National Institute of Environmental Health Sciences. Tara McCray reports financial support was provided by National Science Foundation. Ludovic Vallier reports financial support was provided by Wellcome Leap Fund. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Supplementary data Footnotes CRediT authorship contribution statement Sarah H. Saxton: Conceptualization, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. Inkyung Kang: Formal analysis, Investigation, Visualization, Writing – original draft, Writing – review & editing. Shannon Mitchell: Formal analysis, Investigation, Visualization, Writing – original draft, Writing – review & editing. Jeremy Lotto: Formal analysis, Investigation, Visualization. Aditya Upadhyay: Investigation. Chenhui Zou: Investigation. Corrine Quirk: Investigation. Tara N. McCray: Investigation. Olivia Prado: Investigation. Fredrik Johansson: Investigation. Chelsea L. Fortin: Investigation. Dorice L. Goune: Investigation. Ekaterini D. Zacharis: Investigation. Ludovic Vallier: Funding acquisition, Resources. Ype P. de Jong: Funding acquisition, Resources. Kelly R. Stevens: Conceptualization, Funding acquisition, Resources, Supervision, Writing – original draft, Writing – review & editing. Data availability Data will be made available on request. References We acknowledge that papers authored by women and scholars from historically excluded racial and ethnic groups are systematically under–cited. So that we are not further perpetuating this problem, we have made every attempt to reference relevant papers in a manner that is equitable in terms of gender and racial representation. Associated Data Supplementary Materials Data Availability Statement Data will be made available on request.