Search
Close this search box.

Mapping of mitogen and metabolic sensitivity in organoids defines requirements for human hepatocyte growth – Nature Communications

Time-resolved transcriptomic signatures of hepatocyte organoid establishment from human fetal liver

To temporally characterize the transcriptomic responses upon hepatocyte organoid outgrowth from human fetal liver, we processed tissues from 2 donors and seeded the resulting liver cell suspension into domes of basement membrane extract (BME) overlaid with hepatocyte-specific expansion medium. We collected tissue fragments and the cell suspension generated thereof, as well as the cultured cells and emerging fetal hepatocyte (FH) organoids at different time points post seeding. We also collected organoids 1 day after passaging of the established FH organoid lines (Fig. 1a). Within 3 days post seeding, small organoids appeared, which continued to expand into typical FH organoid lines23,25 (Fig. 1b). Immunofluorescence characterization of these lines confirmed their fetal hepatocyte identity (AFP+, ALB+), while cholangiocyte (bile duct) markers CK7 and CK19 were absent (Fig. 1c). The organoids were broadly Ki-67+, displayed occasional polyploid cells, and possessed MRP2+ bile canaliculi (Fig. 1d).

Fig. 1: Temporal transcriptomic characterization of human fetal hepatocyte organoid growth from tissue.
figure 1

a Experimental strategy to temporally address the transcriptomic changes associated with FH organoid growth from tissue. b Representative brightfield images of FH organoids outgrowing from tissue at day 7 post seeding (left) and an established organoid line. Scale bar = 300 μm. c Representative images of immunofluorescence staining for AFP, ALB, CK7 and CK19 in FH organoids. Scale bar = 150 μm. d Representative image of immunofluorescence staining for Ki-67 and F-actin (left) and MRP2 (right) in FH organoids. Asterisks indicate binucleated hepatocytes. Scale bar = 50 μm. e PCA plot visualizing the temporal transcriptomic changes underlying FH organoid growth from tissue across n = 2 donors. f Heatmap displaying the temporal expression patterns of genes significantly differently expressed at least at one time point versus 0 h (|log2FC| > 0.5, p-adj < 0.05) based on responses of n = 2 donors. The expression patterns are visualized as row Z-scores. The n indicates the number of genes belonging to each cluster. g Temporal Z-score expression of the genes identified in the different fetal tissue gene clusters across n = 2 donors. Mean ± SD is plotted, clusters 1–7: n = 3086, 2509, 3083, 4260, 2872, 2111, and 2330 genes, respectively. h Biological processes/cell types associated with the different temporal gene clusters, based on GO-term enrichment analysis and manual inspection. bd Representative of characterization of n = 4 expanding FH organoid cultures. Source data are provided as a Source data file.

We further surveyed the identity of the cells present in an established FH organoid culture from one donor through single-cell RNA sequencing, recovering 1421 cells (Fig. 2a, Supplementary Fig. 1a, b). These analyses confirmed their uniform hepatocyte identity, underscored by e.g. broad ALB, AFP, TTR, SERPINA1, ASGR1 expression, as well as various other fetal/mature hepatocyte markers (Fig. 2b, Supplementary Fig. 1c). Within these, we noted hepatocytes presenting a more glucose/lipid metabolic profile (e.g. ALDOBhigh, FASNhigh), a cluster abundantly expressing drug metabolism genes (e.g. CYP2C9/19high, ABCC2high), as well as a cluster of proliferating hepatocytes (e.g. MKI67+, CDK1+). Most markers of fetal and mature cholangiocytes (e.g. KRT7, MUC5B, FAM178B) were not expressed (Fig. 2b, Supplementary Fig. 1c). These analyses strengthened the observation that hepatocyte organoids exclusively emerge upon outgrowth from fetal liver tissue in the specific culture conditions used.

Fig. 2: Rewiring of lipid metabolic programs upon human fetal hepatocyte organoid growth.
figure 2

a Single-cell profiling of human FH organoids. b UMAP plots of the indicated markers. c Experimental strategy to temporally address the transcriptomic changes associated with human organoid growth from single FHs. d Cumulative plots of the amount of differentially expressed genes per time point post seeding, all versus 0 h (|log2FC| > 0.5, p-adj < 0.05) based on responses of n = 2 donors. e GO-term enrichment analysis on the upregulated and downregulated DEGs identified at 24 h versus 0 h based on responses of n = 2 donors. f Temporal Z-score expression of the genes identified in fetal tissue clusters 2 and 7 upon organoid growth from single FHs (single cells) across n = 2 donors. The organoid growth responses from fetal liver tissue are plotted for comparison. Mean ± SD is plotted, clusters 2 and 7: n = 2509 and 2330 genes, respectively. g, h Temporal mRNA expression profiles of selected genes involved in the cell cycle, DNA replication, and Wnt signaling (g) and selected classes of lipid metabolism genes (h). Mean fold change expression of n = 2 donors relative to 0 h is shown. i Representative brightfield images (top) and lipid staining overlaid with Phalloidin (bottom) of organoid growth from single FHs over time. Representative of n = 2 outgrowth experiments. Scale bar = 15 μm. j Quantification of the amount of steatosis positive cells during organoid growth (yellow) and the appearance of multicellular organoids (red, defined as n ≥ 2 cells/organoid) at different time points post seeding of single FHs. Mean ± SD is plotted, n = 3 fields analyzed per time point with a total of n = 137 cells (0 h), n = 135 cells (24 h), n = 289 cells (48 h), and n = 258 cells (72 h). k Schematic illustrating the transient steatosis occurring upon FH organoid growth from single fetal hepatocytes. Source data are provided as a Source data file.

We then performed bulk mRNA sequencing on all collected samples from the temporal FH organoid outgrowth experiment from tissue. Principle component analysis (PCA) revealed significant transcriptomic changes occurring upon plating the liver cells, as well as temporal changes during the following time points (Fig. 1e). We next interrogated the broad transcriptomic changes during organoid outgrowth (Fig. 1f). This revealed 7 distinct hierarchical gene clusters, each displaying unique temporal dynamics (Fig. 1g, h, Supplementary Fig. 2a). Cluster 1 comprised genes whose expression steadily increased with culturing time. GO-term analysis revealed many genes related to translation and respiratory electron transport chain, perhaps reflecting the energy requirements and steady growth of the organoids. Genes in cluster 2 displayed a rapid increase in expression during the early time points, but their expression gradually decreased as their growth advanced into organoids that were ready to be passaged (day 7), yet this response was re-activated one day post passaging (ps+1d). Many of the genes in cluster 2 related to gene expression and DNA replication, suggestive to reflect a dynamic proliferative response. Clusters 3 and 4 comprised genes that were typically associated with markers of non-hepatocyte cells, including hematopoietic cells (KLF1)27, Kupffer cells (MAFB)7, stellate cells (CYGB)8, liver sinusoidal endothelial cells (CLEC4G)7, and fetal cholangiocytes (FAM178B)10, while the mature cholangiocyte marker KRT7 was not detected. Expression of these genes rapidly (cluster 4) or more gradually (cluster 3) faded with time, since these non-hepatocyte cells are not retained in epithelial organoid culture. Cluster 5 was characterized by a single rapid surge in expression of genes related to cytokine signaling, likely originating from initially-present immune cells. Expression profiles of genes in cluster 6 largely comprised genes involved in cytosolic transport and in the unfolded protein response. Cluster 7 comprised genes mainly related to lipid metabolic processes, including cholesterol and triglyceride biosynthesis, as well as genes related to the complement cascade. Expression profiles of cluster 7 genes were dynamic, with a rapid decrease during the early time points of organoid growth, yet a gradual gain in expression until the organoids were ready to be passaged (day 7), while expression decreased again one day post passaging (ps+1d).

An inverse transcriptomic relationship between proliferation and lipid metabolism underlies hepatocyte organoid growth

The opposite gene expression trends in tissue clusters 2 and 7 suggested that an inverse relationship between metabolic programs (cluster 7) and proliferation (cluster 2) associates with organoid growth (Fig. 1f, g). Since this tissue dataset did not capture a pure hepatocyte-only response, we designed an experiment to map the early transcriptomic responses upon organoid regrowth from single FHs obtained through single cell dissociation of FH organoid lines (2 donors) (Fig. 2c). We observed widespread transcriptomic changes, peaking at the first 24 h post seeding, with >3000 differentially-expressed genes (DEGs) (Fig. 2d, Supplementary Fig. 3a, b). GO-term enrichment analysis on DEGs identified at 24 h post seeding revealed significant changes in cell cycle and DNA replication genes (upregulated), while genes related to metabolism, including lipid-related processes and cholesterol biosynthesis, were downregulated (Fig. 2e). These observations mirrored the early organoid outgrowth responses from tissue. Visualization of the expression trends of genes contained in tissue clusters 2 and 7 in the current single FH dataset further underscored their similarity, also at later time points (Fig. 2f, Supplementary Fig. 3c). We inspected trends of selected single genes, related to these early reciprocal transcriptomic changes (Fig. 2g, h, Supplementary Fig. 3d). In addition to the characteristic surge in expression of important cell cycle genes (e.g. CCNE1, CDK1) and DNA replication genes (e.g. PCNA, MCM4), we observed induction of Wnt target genes, (LGR5, AXIN2, MYC), highlighting the importance of active Wnt signaling during organoid growth (Fig. 2g). An inverse trend for genes involved in various metabolic processes was observed. Genes involved in de novo lipogenesis (e.g. SREBF1, FASN) and cholesterol biosynthesis (e.g. SREBF2, LSS) were rapidly repressed upon organoid outgrowth. We also noted repressed expression of genes related to lipid export and digestion (e.g. APOB, PNPLA3) (Fig. 2h), suggesting complex alterations at multiple levels of lipid homeostasis.

To assess putative phenotypic consequences, we assessed the presence of lipid droplets during organoid growth from single cells. Dissociated single FHs displayed minimal lipid droplets, yet many cells had accumulated lipids 24 h post seeding. These cells were hypertrophic, yet had not duplicated (Fig. 2i, j, Supplementary Fig. 4a, b). At 48 h, most cells had completed the first round of division and lipid phenotypes rapidly disappeared. From day 3 onwards, the majority of cells continued to form lipid-free multi-cellular organoids (Fig. 2j, k, Supplementary Fig. 4a, b). These observations mirror transient lipid phenotypes and the extensive metabolic-proliferative transcriptomic rewiring observed upon partial hepatectomy in rodents28,29,30,31,32,33. The origin and relevance of the transient steatosis remains unclear, but a lipid-related epigenetic control on cell cycle genes has been postulated34.

Maturation of human fetal hepatocyte organoids based on growth-informed signals

We wondered whether altering growth-associated signals could be exploited to achieve further maturation of FH organoids. We considered Wnt-promoting signals as the most crucial mitogenic factor in the hepatocyte expansion medium promoting proliferation, given the earlier observed Wnt target gene response during organoid outgrowth (Fig. 2g). We assessed the effect of withdrawal of Wnt signals (RSPO1-conditioned medium and the GSK3β inhibitor CHIR-99021 (CHIR)). Removal of RSPO1 did not cause noticeable differences, while CHIR removal led to rapid organoid death (Supplementary Fig. 5a). We then evaluated additional factors that could be needed to induce viable maturation. We focused on HNF4α and CREB, given their central roles in controlling hepatic metabolism processes35,36,37, and based on their predicted involvement in organoid growth (Supplementary Fig. 3e). In addition to CHIR removal, we included forskolin (FSK) (increasing cAMP levels) and dexamethasone (DEX) (reported to induce HNF4α in hepatocytes38) (Fig. 3a). When we switched FH organoids to this medium (termed maturation medium), organoid morphology rapidly changed into structures with a thick wall (Fig. 3b). Of note, while FSK promotes growth of cholangiocyte organoids21, FSK addition to the expansion medium of hepatocyte organoids slows down their growth and induces some traits of maturity39 (Supplementary Fig. 5b). These distinct responses may find their origin in the hepatic cell type-dependent roles of cAMP-PKA signaling40,41. In maturation medium, organoids seized to proliferate, yet remained viable for at least 3 weeks. Cells adopted the typical polygonal shape of mature hepatocytes and became visibly larger (Fig. 3c). By quantifying different morphological features of hepatocyte maturity, we found that cell area and nucleus-to-cytoplasm ratio increased, and binucleated cells were more abundant (Fig. 3d–f). Evaluation of the ALB-AFP expression ratio further suggested acquisition of maturity, with a notable reduction in AFP mRNA expression (Fig. 3g). On protein level, typical ALB granules more prominently appeared, while AFP was absent in most cells (Fig. 3h). Evaluation of mRNA expression of some functional hepatocyte markers revealed increased expression of multiple CYPs as well as the bile acid transporter SLC10A1 (NTCP) in maturation medium (Fig. 3i). Taken together, further maturation of FH organoids can be achieved by transcriptome-informed interference with growth-related signals.

Fig. 3: Modulating culture conditions matures human fetal hepatocyte organoids.
figure 3

a Experimental strategy to mature FH organoids. Altered factors in the maturation medium are indicated. b, c Representative brightfield images (b) and phalloidin staining (c) of FH organoids in expansion medium and maturation medium (14 days post switch). Asterisks indicate binucleated hepatocytes. Scale bar = 400 μm (b), 150 μm (low mag) and 30 μm (high mag) (c). df Quantification of cellular features of FH organoids in expansion and maturation medium based on phalloidin staining, including cell area (d), nucleus-to-cytoplasm ratio (e), and the percentage of binucleated cells (f). Mean ± SD is plotted with n = 20 cells per condition (d, e) and n = 28 quantified areas per condition (f). Maturation versus expansion: ***p < 0.0001, two-tailed unpaired t-test (d, e), **p = 0.0083, Mann–Whitney U test (f). g, h mRNA expression of AFP and ALB (g) and immunofluorescence staining for AFP and ALB (h) in FH organoids in maturation medium relative to expansion medium. Mean ± SD is plotted, n = 4 matured organoid cultures (g). Scale bar = 25 μm (h). i mRNA expression of different functional hepatocyte markers in FH organoids in maturation relative to expansion medium. Mean ± SD is plotted, n = 4 matured organoid cultures. n.d. not detected. b, c, h, Representative of characterization of n = 2 expanding and matured FH organoid cultures. Source data are provided as a Source data file.

Identification of factors influencing human fetal hepatocyte organoid growth

The culture medium used to grow FH organoids has been designed to replace essential non-hepatocyte derived factors, but may be further optimized. Our transcriptomic tissue dataset allowed identifying putative non-hepatocyte derived factors during the early time points of organoid outgrowth when, in addition to the hepatocytes, various other liver cells are still present. Gene cluster 5 was characterized by a transient inflammatory-like response, marked by various interleukins (Figs. 1f,  4a, Supplementary Fig. 2a). Amongst these, IL6 is a well-studied mitogen in mouse liver regeneration42,43. Several growth factors also displayed similar mRNA induction patterns. To predict if such factors were hepatocyte-autonomous or non-hepatocyte derived, we compared gene expression trends across our two transcriptomic datasets (Fig. 4b, Supplementary Fig. 6a). The surge in interleukin expression was prominent in the tissue dataset, while expression was low or near-absent in the single FH dataset, arguing for their non-hepatocyte origin. Instead, expression trends of e.g. EGF family ligands were similar in both datasets, suggesting these to be autonomously produced by hepatocytes (Fig. 4b, Supplementary Fig. 6a). We performed cytokine/growth factor challenges to functionally address their importance in organoid regrowth dynamics from single FHs (Fig. 4c). FHs appeared very responsive to all tested factors (Fig. 4d, Supplementary Fig. 6b), and measurement of the diameter of the outgrowing organoids confirmed these observations (Fig. 4e). Supplementation of NRG1 most significantly boosted FH organoid outgrowth, followed by IL6 and IL11 addition. Surprisingly, IL1β, associated with various hepatic diseases44, likewise boosted FH organoid growth.

Fig. 4: Identification of autocrine and paracrine signals that boost human fetal hepatocyte organoid growth.
figure 4

a Heatmap displaying gene expression patterns of cytokines, chemokines, and growth factors upon FH organoid growth from tissue. The mean expression of n = 2 donors is visualized as row Z-scores. b Comparison between the temporal mRNA expression profiles of selected cytokines and growth factors upon hepatocyte organoid growth from tissue (black) or single FHs (green). The mean expression of n = 2 donors per dataset is plotted (normalized transcripts). c Experimental strategy to evaluate the effect of cytokines and growth factors on human FH organoid growth. d Representative brightfield images of organoid-derived single FHs at day 0 and the outgrowing organoids challenged with indicated cytokines and growth factors at day 7. Representative of n = 2 challenge experiments. Scale bar = 100 μm. e Quantification of the organoid diameter 7 days post organoid outgrowth from single FHs under the different challenges (n = 40 organoids per condition). Mean ± SD is plotted. EREG: *p = 0.0143; AREG: ***p = 0.0005; NRG1, IL1β, IL6, IL10, IL11: ***p < 0.0001, all versus control, one-way ANOVA with Dunnett’s post hoc test. Source data are provided as a Source data file.

Initial growth of human adult hepatocyte organoids mimics the fetal response yet stalls early

We next focused on organoid outgrowth mechanisms of human adult hepatocytes, which are more difficult to grow as organoids23. We applied a similar experimental approach (using the same expansion medium as used for FH organoids) to evaluate their transcriptomic responses utilizing commercial primary human hepatocyte (PHH) sources from n = 2 donors (Fig. 5a). Small organoids grew out over the course of 7 days (Fig. 5b), but these structures (ALB+) seized to grow (Ki-67) and failed to regrow after passaging (Fig. 5c). We then assessed the transcriptomic responses upon PHH organoid growth. PCA revealed major temporal changes during the first 7 days (Fig. 5d, Supplementary Fig. 7a, b). Of note, both PHH donors displayed largely similar transcriptomic trends (and growth arrest) despite their age differences (0.3 vs. 28 years), suggesting that hepatocyte growth differences between development and adulthood are already largely determined rapidly after birth. To assess whether PHH organoid growth associated with the same inverse proliferation-lipid metabolism transcriptomic relationship as observed in the FH organoid datasets, we evaluated expression of genes belonging to fetal tissue clusters 2 and 7. Notably, we observed different dynamics. The early adult responses resembled the fetal response, i.e. an early surge in proliferative signals concomitant with repression of metabolic signals. Yet, the re-initiation of these inverse programs early after organoid passaging (ps+1d) was lost and, instead, transcriptomic responses flattened out (Fig. 5e). Visualization of the temporal transcriptomic trends of various proliferation and lipid metabolism genes further corroborated this observation (Fig. 5f), and differed from the fetal dynamics (compare with Supplementary Fig. 3d). Various lipid metabolism-associated genes displayed different temporal dynamics, including a gradual increase in expression of fatty acid metabolism-related genes (e.g. SCD, FADS1, FADS2) and cholesterol-related genes (e.g. SQLE, HMGCR) (Fig. 5f). CDKN1A (p21), a hallmark senescence marker45, was gradually induced during culture, suggesting its involvement in the growth arrest. The absence of an inverse proliferation-lipid metabolism response post organoid passaging coincided with their stalled growth.

Fig. 5: Temporal transcriptomic characterization of organoid growth from primary human hepatocytes.
figure 5

a Experimental strategy to temporally address the transcriptomic changes associated with organoid growth from PHHs. b Representative brightfield images of outgrowing PHH organoids at day 7 post seeding. Scale bar = 300 μm. c Representative images of immunofluorescence staining for ALB, Ki-67, and β-catenin protein in PHH organoids. Scale bar = 25 μm. d PCA plot visualizing the temporal transcriptomic changes underlying PHH organoid growth across n = 2 donors. e Temporal Z-score expression of the genes identified in the fetal tissue clusters 2 and 7 during PHH organoid growth. The FH organoid growth responses from fetal liver tissue (see Fig. 1g) are plotted for comparison. Mean ± SD is plotted, clusters 2 and 7: n = 2509 and 2330 genes, respectively. f Heatmaps displaying gene expression patterns of proliferation-related genes and lipid metabolism-related genes during PHH organoid growth. The mean expression trends of n = 2 donors are visualized as row Z-scores. b, c Representative of characterization of n = 2 outgrowing PHH organoid cultures. Source data are provided as a Source data file.

Improved human adult hepatocyte organoid growth through rebalancing the cellular metabolic state

We sought to improve PHH organoid expansion. We first asked whether we could exploit the identified growth stimulating factors identified for FH organoids (Figs. 4d, e6a). Out of all factors tested, only IL6 supplementation notably boosted PHH outgrowth, resulting in increased organoid diameters (Fig. 6b, c). Yet, these organoids gradually displayed spontaneous accumulation of lipids (Fig. 6d, Supplementary Fig. 8a), and almost no cell divisions were detected upon live imaging (Fig. 6e, Supplementary Movie 1). We asked whether “tweaking” the cellular metabolic state could reverse this apparent senescence. We focused on FXR given its central role in liver metabolism. Strikingly, PHHs supplemented with IL6+FXRa robustly grew out, and accordingly organoid diameters increased (Fig. 6a–c). FXRa supplementation alone influenced organoid outgrowth to a much more limited extent (Fig. 6c). Only under IL6+FXRa supplementation, PHH organoids could regrow after passaging, appearing healthy and without signs of lipid accumulation (Fig. 6b). Accordingly, we detected multiple cell divisions upon live imaging (Fig. 6f, Supplementary Fig. 8b, c, Supplementary Movie 1). Prompted by the observation that the senescence marker p21 was gradually induced in culture (Fig. 5f), we additionally included the BMP antagonist Noggin in this PHH expansion medium for the first few days after passaging, which was beneficial for organoid regrowth. Organoids from young PHHs from 2 donors (0.3 and 1.7 years) cultured under these conditions could be expanded for at least 4 months (Fig. 6g, Supplementary Fig. 8d).

Fig. 6: Synergy between IL6 and FXR activation boosts primary human hepatocyte organoid growth.
figure 6

a Experimental strategy to test different organoid growth conditions for PHHs. b Representative brightfield images of outgrowing PHH organoids under fetal culture conditions, and upon supplementation of IL6 and IL6+FXRa at day 7 post seeding. Scale bar = 300 μm. c Quantification of the diameter of outgrowing PHH organoids at day 7 and 14 post seeding under the different conditions. Mean ± SD is plotted, n = 30 organoids per condition. FXRa versus fetal condition, day 7: *p = 0.0441; IL6 versus fetal condition and IL6+FXRa versus fetal condition, day 7: ***p < 0.0001; IL6+FXRa vs IL6, day 7: ***p = 0.0003; IL6 versus fetal condition, day 14: **p = 0.0013, IL6+FXRa versus fetal condition and IL6+FXRa versus IL6, day 14: ***p < 0.0001, one-way ANOVA with Tukey post hoc test. n.s. not significant. d Representative brightfield image and lipid staining overlaid with phalloidin of an IL6-cultured organoid. Arrows point at extensive lipid accumulation. Scale bar = 20 μm. e, f Representative time-lapse images visualizing the divergent growth potencies of an IL6-cultured organoid (e) and an IL6+FXRa cultured organoid (f). White outlines highlight cell division. Scale bar = 20 μm (e, f). g Growth characteristics of PHH organoids when cultured under the different conditions. White dots indicate passaging. Grey crosses indicate terminated cultures (failure to regrow after passaging). df Representative of characterization of n = 2 expanding PHH organoid cultures. Source data are provided as a Source data file.

We characterized the PHH organoids under IL6+FXRa supplementation. Some organoids were of dense appearance, while most were thicker-walled with larger lumens (Fig. 7a, Supplementary Movie 2). Individual cells displayed typical polygonal morphology (Fig. 7a, b), and Ki-67 positivity was observed in multiple cells (Fig. 7c). Hepatocyte marker expression was apparent by staining for ALB, while the immature marker AFP was not expressed (Fig. 7d), the latter contrasting the FH organoids (Fig. 1c). Expression of the bile duct markers CK19 and CK7 was not detected (Fig. 7d). The organoids were further broadly positive for A1AT, HNF4α, and CYP3A4, and the tight junction marker ZO1 demarked their complex polarity (Fig. 7e, Supplementary Fig. 8e).

Fig. 7: Characterization of primary human hepatocyte organoid cultures.
figure 7

a Representative low- and high-magnification brightfield images of expanding PHH organoids (passage 4) in the IL6+FXRa condition. Scale bar = 400 μm (low mag) and 100 μm (high mag). b Representative image of H&E staining of PHH organoids. Scale bar = 75 μm. c Representative immunofluorescence staining for Ki-67 of PHH organoids. Scale bar = 100 μm. d, e Representative immunofluorescence staining of PHH organoids for (d) ALB, AFP, CK19, and CK7 and (e) A1AT and HNF4A, CYP3A4, and ZO1. Scale bar = 100 μm (d, e). f Single-cell profiling of PHH organoids. g UMAP plots of the indicated markers. h Single-cell profiles of the FH and PHH organoids (originating from a single library, with the two cultures distinguished based on genotype demultiplexing). i UMAP plots of the indicated fetal and adult hepatocyte markers in the combined single cell datasets. j Schematic illustrating the dynamics underlying human hepatocyte growth and differences between fetal and adult. ae Representative of characterization of n = 2 expanding PHH organoid cultures.

We then surveyed the cellular identity of an established PHH organoid culture from one donor by single-cell RNA sequencing, recovering 3413 cells (Fig. 7f, Supplementary Fig. 9a, b). These analyses revealed broad expression of typical hepatocyte markers (e.g. ALB, TTR, RBP4, TF, HP) (Fig. 7g, Supplementary Fig. 9c). Mature hepatocytes (AHSGhigh, CES1high) were abundant and expressed cytochrome P450s (e.g. CYP2D6, CYP2E1). Within these, a coagulation-high population was apparent marked by abundant expression of F9, F10, F12, and SERPINC1. A smaller progenitor-like population (PROM1+) was also identified. Proliferating hepatocytes were marked by e.g. MKI67 and CENPF positivity. Expression of the mature cholangiocyte markers KRT7, AQP1 and CFTR was mostly absent or confined to some cells in the progenitor-like population (Fig. 7g, Supplementary Fig. 9c). Abundant expression of IL6 target genes (e.g. APP, SAA2, SERPINA3) and the absence of expression of CYP7A1 and CYP17A1 (genes repressed by FXR signaling) further corroborated active IL6+FXRa signaling (Supplementary Fig. 9c). Finally, direct comparison of the single-cell profiles of the PHH organoids with those of the FH organoids (Fig. 2a) revealed their distinct cellular identities (Fig. 7h). We evaluated hepatocyte marker genes reported to change in expression pattern from development to adulthood10. This revealed selective expression of typical fetal (e.g. AFP, GPC3, MT1G) or adult (e.g. APCS, NNMT, CYP2E1) hepatocyte markers across the respective organoid datasets, corroborating preservation of age identity in culture (Fig. 7i, Supplementary Fig. 9d). Altogether, these comparative analyses and functional experiments identified different growth requirements of fetal and adult human hepatocytes in organoid culture (Fig. 7j).