Human assembloids recapitulate periportal liver tissue in vitro

Human assembloids recapitulate periportal liver tissue in vitro | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Biological Sciences - Article Human assembloids recapitulate periportal liver tissue in vitro Meritxell Huch, Lei Yuan, Anke Liebert, Sagarika Dawka, Robert Arnes-Benito, and 16 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5314788/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 17 Dec, 2025 Read the published version in Nature → Version 1 posted You are reading this latest preprint version Abstract The development of complex multicellular human in vitro systems holds great promise for modelling disease and progressing drug discovery and tissue engineering 1 . In the liver, despite the identification of key signalling pathways involved in hepatic regeneration 2,3 , in vitro expansion of human hepatocytes directly from fresh patient tissue has not yet been achieved, limiting the possibility of modelling liver composite structures in vitro . Here, we first developed human hepatocyte organoids (h-HepOrgs) from 28 different patients. Patient-derived hepatocyte organoids sustain long-term expansion of hepatocytes in vitro and maintain patient-specific gene expression, and bile canaliculi features and function of the in vivo tissue. After transplantation, expanded human hepatocyte organoids rescue the phenotype of a mouse model of liver disease. By combining human hepatocyte organoids with portal mesenchyme and our previously published cholangiocyte organoids 4-6 , we generated patient-specific periportal liver assembloids that retain the histological arrangement, gene expression and cell interactions of the periportal liver tissue, with cholangiocytes and mesenchyme embedded in the hepatocyte parenchyma. We leveraged this platform to model aspects of biliary fibrosis. Our human periportal liver assembloid system represents a novel in vitro platform to investigate human liver pathophysiology, accelerate drug development, enable early diagnosis and progress personalized medicine. Biological sciences/Developmental biology/Self-renewal Biological sciences/Developmental biology/Self-renewal Biological sciences/Stem cells/Regeneration Biological sciences/Stem cells/Regeneration Biological sciences/Developmental biology/Disease model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Main Chronic and end-stage liver diseases account for over 2 million human deaths world-wide 7 . Rodent models have advanced our understanding of liver biology. However, species-specific differences (e.g., metabolism, toxicity) impact our understanding of what are universal concepts and which are species-specific, making the translation of potential therapeutic targets to effective human therapies a significant challenge 8,9 . Human liver single cell and spatial transcriptomics have unveiled human cellular heterogeneity 10-16 . However, their static nature fails to inform us about the highly dynamic processes occurring in disease initiation and progression. Primary hepatocytes fail to expand in culture 17 , and while cancer cell lines have been informative, they suffer from genetic drifts. Reprogrammed hepatocytes (ProliHH) are proliferative, can repopulate the tissue, but present bi-phenotypic and progenitor features 18 . Additionally, none of these models recapitulate the 3D bile canaliculi structures (thin and elongated lumina) observed in tissue 19,20 , making it difficult to model complex disease states and to recapitulate patient-specific traits, which are essential characteristics to enable precision medicine approaches for early diagnosis and treatment. Organoids have emerged as promising models to grow human cells with the potential to better predict effective therapeutic outcomes 1 . Human intestinal organoids effectively model human tissue structure and function 21-23 . However, recapitulating in vitro the architecture and cellular interactions of complex tissues like the human liver remains an unmet challenge. We described the first liver organoid models from adult mouse and human tissue 4-6 (now known as cholangiocyte organoids) where cholangiocyte/ductal cells can be expanded in culture long-term, thereby generating millions of cells ex vivo . We 24,25 and others 26,27 demonstrated that these allow the study of mouse liver regeneration in vitro . Small modifications to this system allowed the generation of branching organoids 28 , akin to the morphogenesis of the developing tissue 29-31 or could be transplanted to reconstruct the bile duct in vivo 32 . Mouse adult hepatocyte organoids have been developed 33,34 . Also, mouse 35 and human 33,36 liver hepatoblast organoids were successfully generated from foetal tissue. However, expanding human adult hepatocytes from patient tissue has remained a challenge 37 . Regrettably, all these models consist only of epithelial cells and lack the ability to fully replicate the cellular interactions and architecture of in vivo adult human liver tissue. Similarly, liver organoids derived from human pluripotent stem cells, although contain stromal and epithelial populations do not replicate native adult liver periportal cell interactions or architecture 38-40 . By co-culturing mouse cholangiocyte organoids with mouse liver portal mesenchyme, we obtained cholangiocyte-portal mesenchyme organoids that retain the binary cell-cell interactions present in the mouse liver 25,41 . Chimeric epithelial co-cultures between mouse cholangiocyte and 2D-human hepatocyte-like cells have been reported 42 . However, a complex 3D multi-cellular model that captures human liver portal cellular interactions does not exist for adult human patient liver tissue yet. Here, we developed an adult human hepatocyte organoid model (h-HepOrgs) that allows the long-term serial expansion (> 3months passaging at 1:2) of human adult hepatocytes directly from fresh patient liver tissue. h-HepOrgs retained gene expression and function of in vivo human adult hepatocytes in a patient-specific manner and formed bile canaliculi structures akin to the ones in human tissue. As we expand and cryopreserve organoids from fresh tissue, we have been able to generate a living-biobank of hepatocyte organoids from 28 donors/patients. We combined these novel patient-derived hepatocyte organoids with primary human portal mesenchyme and human cholangiocyte organoids (h-CholOrg) from the same patient to generate human periportal assembloids that recapitulate functional cell interactions and architecture of the in vivo tissue. Finally, we exploited the potential of this system to model aspects of human biliary fibrosis. YAP and WNT activation facilitate the expansion of human hepatocyte organoids To recapitulate the epithelial-stromal interaction and tissue architecture of human liver tissue, we first sought to obtain an expandable source of adult hepatocytes, cholangiocytes and mesenchyme from the same individual. A pre-requisite was to first identify methods to expand human adult hepatocytes. Hence, we obtained human hepatocytes from patient tissue by perfusion 43,44 (methods) and cultured the cells in our previously published, mouse hepatoblast organoid culture medium (MM) 35 . However, the cultures were rapidly filled with cholangiocyte organoids, preventing any further analysis (Extended Data Fig. 1a, No MACS). Then, we adapted the isolation to include a step of magnetic activated cell sorting (MACS) for EpCAM-positive cholangiocyte cells on the perfused tissue. This allowed obtaining viable hepatocytes from the negative fraction, while generating cholangiocyte organoids from the same patient by culturing the EpCAM-positive cholangiocyte fraction in our previously published Huch et al ., 2015 4 medium (Fig. 1a, Extended Data Fig. 1a-c, Methods and Supplementary Table 1). However, although we detected viable hepatocytes and minimal growth for 7-14 days in MM medium (Fig. 1c, MM), hepatocytes rapidly deteriorated and died thereafter. Next, we sought to identify culture conditions for the long-term expansion of human adult hepatocytes as hepatocyte organoids. We hypothesized that signaling pathways involved in cancer progression or tissue regeneration could drive the exit from quiescence in hepatocytes and activate their proliferative state of in culture. To explore this, we analyzed expression profiles from our human liver cancer organoids 45 and publicly available datasets from mouse partial hepatectomy 33 and compared it to human healthy and cancer tissue (n=4 different datasets, Supplementary Dataset 1 and methods). Ingenuity Pathway Analysis (IPA) revealed several pathways were consistently up- or down-regulated such as AMPK, EGF, mTOR, and IGF-1 across at least two datasets (Extended Data Fig. 1d, Supplementary Dataset 1). WNT, MAPK, and FGFR2 were consistently active, while IL-6, HIPPO, and NOTCH appeared inactive (Fig. 1b). Among upstream regulators, components of the WNT pathway (e.g., CTNNB1, LGR5) were upregulated, whereas LATS1 (a negative regulator of YAP/TAZ in the HIPPO pathway) was downregulated, suggesting YAP activation (Extended Data Fig. 1e). Both WNT and YAP are established drivers of liver regeneration 3,46,47 and cancer 47-49 . Therefore, to promote long-term hepatocyte expansion, we activated WNT and YAP signalling by supplementing our published mouse hepatoblast medium 35 with a WNT surrogate (WntS) 50 and a LATS1/2 inhibitor (TRULI or TDI-011536) 51 . Combining both enabled serial passaging (5–6 passages) of human hepatocyte organoids as solid structures with no lumina (Fig. 1c–f, Extended Data Fig. 1g). TRULI-treated cultures showed superior morphology, so we continued with the MM+WntS+TRULI combination, termed h-HepOrg-EM1 from hereon. The other tested pathways had no consistent or quantifiable organoid growth (Extended Data Fig. 1f). We further optimized the medium (MM) by testing each component’s necessity. Notably, removing nicotinamide improved organoid formation efficiency nearly 10-fold and enabled long-term culture for over 3 months (>10 passages at 1:2 split/week) (Extended Data Fig. 1i, Supplementary Table 2). These results were in line with our IPA analysis showing NAD signaling inactivity (Fig. 1b) and previous reports of nicotinamide hepatotoxicity in humans 52 . Using these optimized conditions (EM1 without nicotinamide, hereafter called h-HepOrg-EM2), we successfully generated expandable human hepatocyte organoids (h-HepOrgs) from 28 patients (11–85 years old, 30% female) with 100% efficiency (Supplementary Table 2). No other tested conditions—including those from human foetal or mouse hepatocyte organoids 33,34 —supported robust expansion (Extended Data Fig. 1j, Supplementary Table 3 and Source Data). h-HepOrgs maintained stable chromosome numbers over time and could be frozen/thawed without loss of expansion capacity, enabling the creation of a living biobank from a total of 28 different donors (Extended Data Fig. 1k-l). Together, these results demonstrate that combination of WNT and YAP activation allows the long-term expansion of adult human hepatocyte organoids. Adult h-HepOrgs retain in vivo structural, transcriptional and functional features To characterize the expanded h-HepOrgs, we first performed RNAseq analysis on early (P1-P3) and late (P10) passage cultures and compared their expression pattern with freshly isolated primary human hepatocytes (PHHs, primary) and human cholangiocyte organoids (h-CholOrgs) from the same donors (when possible). Expanded h-HepOrgs closely correlated with freshly isolated hepatocytes, while h-CholOrgs from the same donors clustered separately (Extended Data Fig. 2a-b). Gene expression and gene set enrichment analysis revealed that the h-HepOrgs exhibited a proliferative signature that was maintained until late passages (>P10) and resembled regenerating tissue after hepatectomy (Extended Data Fig. 2c-e). These results were in agreement with the positive Ki-67 (Fig. 2a, top) and negligible cleaved-caspase 3-staining (Extended Data Fig. 2f). h-HepOrgs exhibited elevated expression of WNT and YAP target genes, consistent with pathway activation following WNT and LATS1/2 inhibitor treatment (Extended Data Fig. 2g-h). Immunofluorescence confirmed the nuclear localization of YAP in TRULI-treated h-HepOrgs (Fig. 2b, middle panel and Extended Fig. 2i), but not in non-treated cultures (Fig. 2b, bottom panel) or in homeostatic tissue (Fig. 2b, top panel). qPCR confirmed these results (Extended Data Fig. 2j). However, we cannot exclude that off-target effects may also contribute to the growth effects, as TRULI can inhibit kinases beyond LATS1/2. Marker gene expression analysis showed that the expanded h-HepOrgs expressed hepatocyte markers such as HNF4A, ALB, several apolipoproteins ( APOC2, APOA4 ) and cytochromes ( CYP3A4, CYP3A7 ), albeit at lower levels compared to freshly isolated hepatocytes (Fig. 2d, Extended Data Fig. 2c, and Supplementary Dataset_2). Cholangiocyte markers such as SOX9, KRT19 or KRT7 were markedly reduced while the expression of the embryonic liver marker AFP suggested incomplete maturation (Extended Data Fig. 2k and Supplementary Dataset_2). qPCR and immunofluorescence analysis confirmed that HNF4A and the apical and polarity marker CD13 ( ANPEP ) were both highly expressed (Fig. 2c, Extended Data Fig. 2k and Supplementary Dataset_2). However, detailed analysis of the distribution of CD13 expression showed the presence of wide, disconnected, round lumina, which does not reflect the morphology of the bile canalicular network formed by hepatocytes in vivo 53-55 (Fig. 2c, compare CD13 in h-HepOrgs to the tissue panel in Fig. 3b). Taken together, these results suggested that the expanding h-HepOrgs in EM2 medium presented an immature hepatocyte state. Therefore, we sought to define a differentiation medium. LATS1/2 inhibition was recently shown to promote cholangiocyte growth 56 , while it is well established that YAP activation drives hepatocyte de-differentiation and its inactivation facilitates re-differentiation 57 . Therefore, we reasoned that reducing YAP activation would facilitate the maturation of h-HepOrgs. Following several iterations, we developed a hepatocyte differentiation medium (h-HepOrg-DM from hereon) in which we removed YAP and FGFR2 activation, maintained Wnt signaling and added Dexamethasone (Extended Data Fig. 3a and methods). Under differentiation medium, the cellular morphology improved: hepatocytes (HNF4A+) reduced proliferation, acquired a significantly higher cytoplasm to nuclei ratio and improved bile canaliculi (CD13+), which presented thinner and more elongated morphology (Fig. 2a, c, compare EM2 vs DM and Extended Data Fig. 3b). Combined these features suggested enhanced hepatocyte maturation. To assess the extent of the maturation, we performed RNAseq analysis. Principal Component Analysis (PCA) revealed that differentiated h-HepOrgs move closer to fresh isolated hepatocytes and farther away from cholangiocyte organoids, when compared to h-HepOrgs in expansion medium (Extended Data Fig. 3c). The cells increased the expression of many mature markers, some to similar levels as freshly isolated human hepatocytes, including ALB , several APO lipoproteins ( APOE, APOA1), bile acid transporters ( ABCC2 (MRP2)) and cholesterol and bile acid metabolic genes ( SCARB1, ABCG8 and CYP27A1). Additionally, several detoxifying enzymes such as CYP2C9, CYP3A5, CYP3A4 and MAOA , some of them pericentrally zonated 13,14,58,59 , were also upregulated (Fig. 2d and Extended Data Fig.3f-i). Consistent with these results, we found significant positive enrichment for many gene sets related to mature hepatocyte functions including cholesterol, fatty acid and drug metabolism, Phase II conjugation, clot formation and bile secretion, amongst others. Conversely, signatures related to cell cycle and proliferation were negatively enriched (Fig. 2e and Extended Data Fig. 3e). Similarly, the expression of the embryonic marker AFP and the cholangiocyte makers KRT7 and KRT19 were reduced (Fig. 2d and Extended Data Fig. 3i). Importantly, some pericentrally zonated genes, such as CYP2E1 and GLUL (Glutamine Synthase, GS), as well as some periportally zonated genes, such as ALDOB or SCD, were highly upregulated (Fig. 2d and Extended Data Fig. 3h). Immunofluorescence analysis for pericentral (GS) or periportal (HAL) markers indicated that some cells within the organoids presented a gradient of expression of zonated genes, with some cells positive and others negative for the markers (Fig. 2f). Dual immunofluorescence staining for CYP2E1 (pericentral marker) and ECAD (enriched in periportal region) highlighted the heterogeneity and spatial distribution of hepatocyte function within the same h-HepOrg, at least for those genes tested (Extended Data Fig.3j). Strikingly, under differentiation conditions, we observed that h-HepOrgs recapitulated the complex cell polarity of in vivo hepatocytes 17 , with the tight-junction and apical polarity marker ZO-1 localized to the apical surface of adjacent hepatocytes (Fig. 3a, bottom panel), resembling the morphology of bile canaliculi (BC) in human liver tissue (Fig. 3a, top panel, Radixin). Immunofluorescence staining with the apical marker CD13 followed by image analysis and reconstruction revealed that the differentiated h-HepOrgs displayed longer and more branched BC networks within each organoid (Fig. 3b, middle panel), when compared to the same organoids in expansion medium (Fig. 3b, top panel) and resembled the in vivo tissue (Fig. 3b, bottom panel). Additionally, the BC network connectivity was also significantly improved, coming closer to that of the tissue (Fig. 3c). Noteworthy, we observed that different patients present fine-detailed differences in BC morphology (Extended Data Fig. 4a-c), with some patients presenting thin and homogenous BC, some wider and inhomogeneous BC, and others full of branchlets (Extended Data Fig. 4b). We found a similar variation in the BC architecture across our different organoid cultures, suggesting that our model could capture different types of BC structures that are observed in patient tissue cohorts (Extended Data Fig. 4d). Given that our h-HepOrgs are derived directly from patient tissue, we next assessed whether they retain patient-to-patient variability in culture, thus enabling patient-specific modeling of hepatocyte-related liver diseases. For that, we analysed the transcriptome of primary hepatocytes at time of isolation and their matching h-HepOrgs under differentiation medium to identify the specific gene signatures of each patient. We found a strong correlation (R 2 =0.7-0.9) between the organoids and the original primary hepatocytes they derived from (Extended Fig. 4e). Interestingly, many of the patient-specific genes we found expressed in organoids and their source cells had been associated with either susceptibility to hepatitis virus infection ( IL1RL1 or ERAP2 ) or liver cancer ( GPC3 ) or to cholestasis during pregnancy ( GABRP ). More interestingly, some genes were involved in metabolic pathways like the glutathione-related genes GSTM3 and GSTM1 , the lactate dehydrogenase LDHC and the lipid metabolic related genes APOA4 , the fatty acyl-CoA reductase FAR2 or the Acyl-CoA synthetase ACSM1, amongst others (Fig. 3d and Supplementary Table 4). These results indicated that h-HepOrgs could preserve patient-to-patient specific signatures, with significant implications for modelling human liver diseases. Next, we compared the functional performance of differentiated h-HepOrgs to primary human hepatocytes (PHHs). HepOrg differentiated in DM exhibited mature hepatic functions, including robust albumin secretion and, moderate cytochrome P450 activity, comparable to 7-day PHHs (Fig. 3e–f). Specifically, differentiated h-HepOrgs displayed CYP2C9 activity equivalent to that of 7-day PHHs, and modestly reduced CYP3A4 activity, while 1-day PHHs demonstrated superior activity for both enzymes. Notably, mass spectrometry analysis revealed that differentiated h-HepOrgs significantly outperformed 1-day PHHs in converting the antiarrhythmic and antihypertensive drug Verapamil into its primary metabolite, Norverapamil (Fig. 3g). This superior overall metabolic performance toward Norverapamil suggests a more robust or sustained expression and coordination among multiple CYP enzymes relevant to Verapamil metabolism including the metabolizing enzymes CYP 2 C8, CYP3A4, and CYP3A5 , all responsible for Verapamil N-demethylation and highly expressed in h-HepOrgs in differentiation medium (Extended Data Fig. 3g, i). Furthermore, we observed inter-donor variability in Verapamil metabolism among h-HepOrgs lines (Fig. 3g), reflecting patient-specific metabolic phenotypes and underscoring the potential of this platform for personalized drug metabolism studies. Remarkably, both expanded and differentiated hepatocyte organoids readily engrafted and maintained their hepatic function in vivo , following xenotransplantation in the Tyrosinemia type I liver disease mouse model ( Fah -/- Rag2 -/- Il2rg -/ mouse) 60 , with grafts distributed throughout the liver parenchyma. Importantly, the engrafted cells were able to rescue the otherwise lethal phenotype of the mice (Fig. 3h and Extended Data Fig. 4f). In summary, we have developed a novel h-HepOrgs model that enables the growth of functional adult human hepatocytes directly from patient tissue, preserving hepatocyte polarity and generating a bile canaliculi network that resembles adult liver tissue in vivo , while retaining some aspects of patient-to-patient variability. Human liver assembloids model periportal tissue We next aimed to reconstruct the periportal region of the liver lobule by reproducing the cellular interactions between hepatocytes, cholangiocytes and portal mesenchyme, specifically portal fibroblasts. To identify conditions to isolate and culture portal mesenchyme over multiple passages we used PDGFRA + liver cells, as PDGFRA is exclusively expressed in liver mesenchyme and absent in other stromal cells, according to different human liver cell atlases 10,13,14,16,61 (Extended Data Fig. 5a-f and methods). Next, we examined publicly available datasets 10,61 to further enrich for healthy portal fibroblasts. We found that the cell surface receptor CD90 ( THY1 ) is expressed exclusively in human portal fibroblasts from healthy individuals (Extended Data Fig. 5g, 6a, b). Immunofluorescence analysis confirmed the restricted expression of CD90 to the periportal region (Extended Data Fig. 6c). Thus, we combined CD90 with PDGFRA to enrich for healthy human portal fibroblasts under defined culture conditions (Extended Data Fig. 6d). RNAseq and qPCR analysis revealed that CD90+/PDGFRA+ cells expressed portal fibroblasts markers (e.g. DCN, THY1 and ASPN ) and several growth factors ( HGF , WNT5A among others), which are all highly enriched in portal mesenchyme in vivo . Conversely, hepatic stellate cell (HSC, RGS5 , LRAT and RELN ) and vascular smooth muscle cell (VSMC, MYH11 ) markers were not detected (Extended Data Fig. 6e-h). Immunofluorescence analysis for vimentin (mesenchyme) and CD90 (portal mesenchyme) confirmed that the majority of the expanded cells were portal fibroblasts (Extended Data Fig. 6i). Next, we aimed to generate human periportal assembloids. To ensure that the structures contained the different cells and to facilitate their visualisation, we tagged human cholangiocyte organoids and portal mesenchymal cells with nuclear fluorescent proteins (Extended Data Fig. 6j), while leaving hepatocyte organoids unlabelled. Then, to determine the proportions of cells to assemble into composite structures we first quantified the physiological proportion of cholangiocytes portal mesenchyme and hepatocytes in vivo in human tissue, which resulted in an average of 15% Chol: 8% MSC: 77% Hep (Extended Data Fig. 7a). To induce the self-assembly of the three cell populations into a single structure that would recapitulate periportal spatial organization, we tested several approaches to aggregate h-HepOrgs with dissociated portal liver mesenchyme and cholangiocytes from h-CholOrgs from the same donor, when possible. Of all the approaches tested, mixing one h-HepOrg structure with a defined number of portal mesenchymal and ductal cells (from h-CholOrgs) in 96-well low adhesion U-plates readily generated structures where the 3 cell types were together with Cholangiocyte and MSC cells embedded inside the HepOrg structure. We called these structures periportal assembloids (Fig. 4a, b). The ratio 1 HepOrg: 25 MSC and 100 Chol better captured the tissue cell ratios at day 6 post-assembloid formation, and was selected for further experiments (Extended Data Fig. 7b, c). To increase the number of assembloids generated, we used Aggrewell TM plates (Fig. 4b and Extended Data Fig. 7d). Notably, both methods generated assembloids with high efficiency (~80% efficiency) (Fig. 4c) and closely maintained the cellular composition and proportions of the tissue (Fig. 4e). Therefore, we only used Aggrewell from hereon. Remarkably, from day 3 onwards, the periportal assembloids recapitulated key architectural features of the in vivo tissue, with ductal cells (KRT19 + and nGFP + ) forming bile duct-like structures containing open lumina, with portal mesenchymal cells (nuclear-RFP) in close proximity, and embedded within the hepatocyte (HNF4A⁺) parenchyma. This architectural organization, where ductal cells are forming an apical lumen, basally contacted by mesenchymal cells and embedded in the hepatocyte structure, was observed in the majority (80%) of the assembloids and across donors. These results were independent of the patient source for the healthy portal liver mesenchyme, indicating minimal impact of patient-origin under healthy conditions (Fig. 4d-f and Extended Data Fig. 7d-h).Vimentin staining confirmed that MSC consistently extended long cellular processes towards the basal side of cholangiocytes, leading to physical contacts and reminiscent of the interactions observed in human tissue, although not completely wrapping the cholangiocytes as in the portal tracts in vivo (Fig. 4f and Extended Data Fig. 7h). Under these conditions, the assembloids could be maintained for at least two weeks in culture, with no evidence of increased cell death or proliferation over time (Extended Data Fig. 7i-l). Next, we employed single cell RNAseq (scRNAseq) analysis to benchmark our model to the in vivo human liver tissue. Clustering, PCA and correlation analysis indicated that the assembloid cells mostly overlap with the corresponding populations from publicly available human liver cell atlases 10,13,14,16,61 (Fig. 4g and Extended Data Fig. 8a). Hepatocytes, cholangiocytes and mesenchymal cells from assembloids expressed classical markers of their in vivo counterparts (Hep: ALB, HNF4A, TTR ; Chol: KRT7, KRT19, EPCAM ; MSC: COL1A1, VIM, THY1 ) (Fig. 4h), indicating that they retain their identity in human assembloids. GSEA confirmed that mesenchymal cells were highly enriched for signatures of ECM organization and cell adhesion, cholangiocytes for cytoskeleton and cell-cell communication, and hepatocytes for fatty acid metabolism, complement and drug metabolism, similar to in vivo human liver tissue (Extended Data Fig. 8c, d). Interestingly, we observed heterogeneous expression of classical zonated hepatocyte markers, with a fraction of hepatocytes expressing periportal ( SAA1 and SAA2, APOA1 ) and others expressing pericentral ( CYP2E1) markers (Fig. 4h and Extended Data Fig. 8b). To investigate whether the periportal assembloid microenvironment and the interaction with portal ductal and mesenchymal populations could promote a more portalized hepatocyte identity, we compared the gene expression profiles of hepatocytes from HepOrgs cultured in DM with those from assembloids (also cultured in DM). Notably, hepatocytes within assembloids exhibited higher expression of periportal markers, including SAA1, SAA2, HAMP, and APOA1 , while showing reduced expression of pericentral genes such as CYP2E1, CYP3A4, and GLUL , compared to HepOrgs cultured alone. (Fig. 4i). Staining for the periportal hepatocyte markers SAA1 and SAA2 confirmed the spatially heterogenous expression of these portal markers, with the positive cells overlapping with regions of ECAD⁺ high cells (Extended Data Fig. 9a), and in agreement with our scRNAseq results (Fig. 4 h). Notably, periportal assembloids exhibited enhanced functional specialization characteristic of periportal hepatocytes. They outperformed h-HepOrgs, cultured in the same conditions, in both urea production and gluconeogenesis (both portal functions), while the drug-metabolizing capacity associated with pericentral hepatocytes was less pronounced compared to hepatocyte organoids, in line with their more portal-like nature. As expected, periportal assembloids retained core hepatocyte functions, with albumin secretion increasing over time to levels matching hepatocyte organoids and exceeding 2D primary hepatocyte cultures. (Fig.4j-l and Extended Data Fig. 9b). These findings suggested that the periportal microenvironment within the assembloids could promote the acquisition of a more portal-like hepatocyte identity. In line with this hypothesis, we noted that some hepatocyte membranes joined the lumen of the bile ducts, similar to what we observed in in vivo tissue and suggestive of physiological cell-cell contact between both cell types (Extended Data Fig. 9c-d). Combined, we have generated a human liver periportal assembloid model that faithfully captures the gene expression, cell interactions and aspects of the tissue architecture of the human liver periportal region in vitro . Human assembloids model aspects of biliary fibrosis Portal mesenchyme often contributes to myofibroblast populations in human fibrosis 62,63 . Hence, we next investigated whether we could utilize our human assembloid model containing portal fibroblasts to recapitulate aspects of human liver disease in vitro, specifically biliary fibrosis. Interestingly, increasing (20-fold) the total number of initial mesenchymal cells, while keeping the other cell numbers constant (even from the same source tissue), altered the cell composition of the assembloids. We found that the number of cholangiocytes (GFP + , KRT19 + ) increased, while the total number of hepatocytes (HNF4A + ) decreased (Fig. 5a-d). Ki-67 staining indicated that c holangiocytes exhibited early proliferative responses to fibrotic cues , while cleaved-caspase 3 staining revealed that the reduction in hepatocyte numbers was associated with increased hepatocyte death, at least in part, through apoptosis (Extended Data Fig. 10a-c). This finding was consistent with our observations in mouse assembloids 64 , suggesting a conserved mechanism across species. scRNAseq clustering and correlation analysis revealed that the hepatocytes, cholangiocytes and mesenchyme from assembloids with excess mesenchyme, but not from homeostatic mesenchyme numbers, recapitulated the gene expression of their corresponding cells from publicly available datasets of diseased livers 10,16 (Fig. 5f). The top markers identifying the three cell populations in the corresponding patient datasets were also highly expressed in the corresponding assembloids cells (Fig. 5g), while GSEA revealed that mesenchyme and cholangiocytes from fibrotic, but not homeostatic assembloids, increased expression of collagen and matrix deposition processes (Fig. 5h, Extended Data Fig. 10f, g and Extended Dataset_4). Similarly, cholangiocytes, but not mesenchyme, presented signatures of proliferation (Fig. 5h and Extended Data Fig. 10f), in agreement with the increased number of GFP+ cholangiocytes detected (Fig. 5c-d). These gene signatures (increased matrix and cholangiocyte numbers) are reminiscent to the fibrotic tissue from human patients with biliary fibrosis and primary sclerosis cholangitis (PSC) 16,62,63 . Hence, from hereon we named the assembloids with excess mesenchyme “fibrotic-like”, to distinguish them from the “homeostatic-like” with homeostatic numbers of mesenchymal cells. Strikingly, we found gene sets for inflammatory reactions including TNF signalling, several interleukins (IL-4, IL-6), NFKB, JAK-STAT and Toll-like receptor cascade among the most positively enriched gene sets in hepatocytes (Fig. 5h and Extended Data Fig. 10d, e) from fibrotic-like assembloids compared to homeostatic ones. Conversely, cell cycle signatures and hepatocyte functions such as bile secretion, lipid or drug metabolism were negatively enriched (Fig. 5h). Both, hepatocytes and cholangiocytes from fibrotic assembloids were also highly enriched in TGF-B signalling signatures (Fig. 5h Extended Data Fig. 10d-f), mirroring the transcriptional changes in patients with biliary fibrosis 16 . Morphologically, we observed that the fibrotic-like assembloids, but not matching homeostatic assembloids, exhibited a cystic-like phenotype reminiscent of cholangiocyte organoids (Extended Data Fig. 10h-i). This observation was in line with the immunofluorescence analysis, which indicated that in fibrotic-like assembloids some hepatocytes (HNF4a + , GFP - ) were positive for the cholangiocyte marker KRT19, and opened lumina, resembling the polarity of simple ductal epithelium, which is suggestive of potential hepatocyte-to-duct trans-differentiation (Fig. 5e and Extended Data Fig. 10j, magenta arrow and asterisk). Interestingly, all these phenotypes: (i) increased signatures in TNF-A, IL-4, IL-6, (ii) increased hepatocyte apoptosis and (iii) increased expression of cholangiocyte markers have been reported in fibrotic patients as well as in recent liver cell atlases of primary sclerosis cholangiatis (PSC) and primary biliary cirrhosis (PBC) patients 16,65 . These results combined suggest that our assembloid model with excess mesenchyme mimics some aspects of human biliary fibrosis as seen in cholangiopathies, including PSC and PBC. Discussion Failure in maintaining the intricate cellular organization and multidirectional interactions of the cells within the liver lobule leads to chronic liver disease, often presenting with cholestasis and fibrosis, which can lead to cirrhosis and cancer 66,67 . While reductionist by nature, ex-vivo systems, offer powerful tools to dissect disease mechanisms, particularly, the relative contributions of the distinct intrinsic cellular programs and microenvironmental cues—including cell-cell interactions—to disease initiation and progression. We recently showed that mouse periportal assembloids retain key architectural features, such as the reconstruction of the bile canaliculi-bile duct connection and can serve as a tractable and modular in vitro model to investigate universal principles of bile canaliculi formation, bile flow, cholestatic injury and biliary fibrogenesis 64 . However, species-specific differences in drug metabolism, toxicity profiles or liver pathophysiology, necessitate the development of complementary human models that capture patient-specific features to better understand diseases mechanisms, identify therapeutic strategies, or screen for therapeutic compounds. Recent advances in human liver models underscore the ongoing efforts and the broad interest in developing physiologically relevant in vitro systems. These include: iPSC-derived hepatocyte organoids exhibiting dual zonation 68 , functional hepatocyte organoids derived from cryopreserved hepatocytes 69 , mass-generation of hepatobiliary organoids 70 , co-cultures of dermal fibroblasts with hepatocyte spheroids 71 or mouse fibroblasts aggregated with hepatocyte spheroids and cholangiocyte organoids 72 . However, a model capable of recapitulating the multicellular periportal liver tissue organization and cellular interactions ex vivo—while enabling inter-individual comparative studies and investigation of patient-specific disease traits—has not yet been developed. Here we overcome this challenge, by establishing long-term expandable human hepatocyte organoids (h-HepOrgs) from adult patient liver tissue and combining them with human cholangiocyte organoids and human portal mesenchyme to form complex periportal liver assembloids. These assembloids recapitulate essential structural and functional features of the native human periportal region and, upon manipulation, model aspects of human biliary fibrosis. Our h-HepOrg model enables long-term expansion while preserving functional drug-metabolizing capabilities and capturing patient-to-patient variability, including differences in metabolic enzymes and disease-associated genes. At both cellular and mesoscale levels, h-HepOrgs mimic fine architectural features such as bile canaliculi morphology and display heterogeneous expression of zonated hepatocyte genes. Although we observed variability in bile canaliculi (BC) morphology among organoids derived from different donors, whether this reflects true patient-to-patient differences will require further investigation. When combining human hepatocyte organoids with human portal mesenchyme and human cholangiocyte organoids, the resulting human liver periportal assembloids recapitulate key architectural features of the native periportal region in vitro , with mesenchymal cells closely associated with cholangiocytes and forming basal contacts, both embedded within a hepatocyte parenchyma. Interestingly, assembloids exhibited increased portal-region functional features. Whether direct interactions between hepatocytes, cholangiocytes, and portal mesenchyme are sufficient to instruct portal-specific hepatocyte identity remains an open question. Likewise, whether hepatocyte subpopulations at the onset of culture influence differential responses to microenvironmental cues cannot be excluded. Our modular assembloid platform provides a unique system to systematically manipulate individual cellular components and begin to dissect, in a controlled human in vitro setting, how specific microenvironmental signals contribute to human hepatocyte identity and zonation or to epithelial-portal mesenchyme interactions. Of note, by increasing the number of portal mesenchymal cells we generated fibrotic-like assembloids that recapitulated aspects of human cholestatic liver disease and biliary fibrosis, including increased cholangiocyte numbers, reminiscent of 'ductular responses' observed in patients with chronic liver diseases 73 . One caveat of the model, though, is that it still lacks the other stromal components, mainly the other mesenchymal cells (HSC and VSCM) as well as the endothelium and immune cells. The lack of portal vasculature (portal vein and hepatic artery) limits the formation of a true periportal triad, as endothelial networks are essential for oxygen delivery and spatial patterning. Incorporating these to generate more complex assembloids models will be crucial to reproduce all aspects of liver disease. In summary, the patient-derived hepatocyte organoids and periportal assembloid models we present here hold, in our view, the potential to initiate a new era in diverse areas of liver research, including diagnostics, toxicology, personalized drug efficacy screenings and cellular transplantation therapy, opening up new avenues to find therapeutic approaches for cholestatic injury and biliary fibrosis diseases. Declarations Acknowledgements M.H. is supported by the Max Planck Gessellschaft and is recipient of an "Allen Distinguished Investigator Award, a Paul G. Allen Frontiers Group advised grant of the Paul G. Allen Family Foundation, which supports A.L. and R.R.C. This project was partially supported by the European Research Council under the European Union’s Horizon Europe research and innovation programme (grant agreement No 101088869) awarded to M.H. Views and opinion expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them. Part of this work was also funded by the LiSyM grant from the Bundesministerium für Bildung und Forschung (BMBF Federal Ministry of Education and Research) awarded to M.H. (031L0258C, and 031L0315B) and to G.D. and D.S. (031L0258E and 031L0315D). This project was also partially supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation, 514150034) and the DFG under Germany´s Excellence Strategy – EXC-2068 – 390729961- Cluster of Excellence Physics of Life of TU Dresden. D.E.S. was supported by a German Cancer Aid Max Eder Grant (#70113745). We thank Ms Florida Ahmed for help with chromosome counting. We thank Dr. Julia Jarrells and Ms. Jessica Hernandez (MPI-CBG) for assistance with fluorescence-activated cell sorting (FACS), the light microscopy facility for imaging troubleshooting and training (Dr Jan Peychl and Dr Riccardo Maraspini), the Technology Development Studio facility for the high-throughput imaging and image analysis (Dr Rico Barsacchi and Dr Martin Stöter) and the Dresden Concept Genome Center for the RNAseq and scRNAseq library (Susanne Reinhardt and Juliane Bläsche at the DcGC Dresden-concept Genome Center - a core facility of the CMCB and Technology Platform of the TUD (Dresden University of Technology). We thank Ms Jessie Pöche for assistance with hepatocyte isolation and the surgical research laboratory and especially MS Stefanie Hübner as well as the operative team of the Dept. of Visceral, Thoracic and Vascular Surgery, University Hospital Dresden, for the assistance in liver tissue processing. We thank the whole team of the Department of Hepatobiliary Surgery and Visceral Transplantation, Leipzig University Medical Center for their support in the patient acquisition and the liver tissue logistics. We thank Gerda Schicht for her assistance in the hepatocyte isolations at Leipzig University Medical Center. We thank Dr Michele Marass for insightful comments and discussions on the manuscript. Author contributions M.H. designed the study. L.Y., S.D., Y.K., A.L. and R.A-B. performed most of the experiments and, together with M.H., interpreted the results. F.R. performed the scRNAseq analysis. F.R. and D.L.H.T. performed the bulk RNAseq analysis. R.R.C performed the chromosome analysis. A.Sch., A.She, And.She. applied direct infusion mass spectrometry to characterize drug metabolism capacity. F.B., D.E.S. and C.G. assisted with tissue processing and hepatocyte isolation. D.S., G.D. and D.E.S. obtained the patient consent of the tissue samples used in the study. S.D. performed image analysis and bile canaliculi reconstructions with the help from A.S. A.M.D. and A.S. contributed on the first phases of the project. Y.K. and D.C conducted the xenotransplantation experiments with the assistance from S.L.K. L.Y., S.D., Y.K., A.L. and M.H. wrote the manuscript. All authors read and commented on the manuscript. Competing Interests M.H. is an inventor in several patents on organoid technology. Y.K., S.D. and M.H. are inventors in a patent on human hepatocyte organoids. M.H. L.Y, S.D., A.M.D and A.S. are co-inventors in a patent on human assembloids. The remaining authors declare no competing interests. References Lancaster, M. A. & Huch, M. Disease modelling in human organoids. Dis Model Mech 12 (2019). https://doi.org:10.1242/dmm.039347 Campana, L., Esser, H., Huch, M. & Forbes, S. Liver regeneration and inflammation: from fundamental science to clinical applications. Nat Rev Mol Cell Biol 22 , 608-624 (2021). https://doi.org:10.1038/s41580-021-00373-7 Michalopoulos, G. K. & Bhushan, B. Liver regeneration: biological and pathological mechanisms and implications. Nat Rev Gastroenterol Hepatol 18 , 40-55 (2021). https://doi.org:10.1038/s41575-020-0342-4 Huch, M. et al. Long-term culture of genome-stable bipotent stem cells from adult human liver. Cell 160 , 299-312 (2015). https://doi.org:10.1016/j.cell.2014.11.050 Broutier, L. et al. Culture and establishment of self-renewing human and mouse adult liver and pancreas 3D organoids and their genetic manipulation. Nat Protoc 11 , 1724-1743 (2016). https://doi.org:10.1038/nprot.2016.097 Huch, M. et al. In vitro expansion of single Lgr5+ liver stem cells induced by Wnt-driven regeneration. Nature 494 , 247-250 (2013). https://doi.org:10.1038/nature11826 Devarbhavi, H. et al. Global burden of liver disease: 2023 update. J Hepatol 79 , 516-537 (2023). https://doi.org:10.1016/j.jhep.2023.03.017 Ramachandran, P. & Henderson, N. C. Antifibrotics in chronic liver disease: tractable targets and translational challenges. Lancet Gastroenterol Hepatol 1 , 328-340 (2016). https://doi.org:10.1016/S2468-1253(16)30110-8 Friedman, S. L., Neuschwander-Tetri, B. A., Rinella, M. & Sanyal, A. J. Mechanisms of NAFLD development and therapeutic strategies. Nat Med 24 , 908-922 (2018). https://doi.org:10.1038/s41591-018-0104-9 Ramachandran, P. et al. Resolving the fibrotic niche of human liver cirrhosis at single-cell level. Nature 575 , 512-518 (2019). https://doi.org:10.1038/s41586-019-1631-3 Xu, J. et al. A spatiotemporal atlas of mouse liver homeostasis and regeneration. Nat Genet 56 , 953-969 (2024). https://doi.org:10.1038/s41588-024-01709-7 Aizarani, N. et al. A human liver cell atlas reveals heterogeneity and epithelial progenitors. Nature 572 , 199-204 (2019). https://doi.org:10.1038/s41586-019-1373-2 MacParland, S. A. et al. Single cell RNA sequencing of human liver reveals distinct intrahepatic macrophage populations. Nat Commun 9 , 4383 (2018). https://doi.org:10.1038/s41467-018-06318-7 Brazovskaja, A. et al. Cell atlas of the regenerating human liver after portal vein embolization. Nat Commun 15 , 5827 (2024). https://doi.org:10.1038/s41467-024-49236-7 Andrews, T. S. et al. Single-Cell, Single-Nucleus, and Spatial RNA Sequencing of the Human Liver Identifies Cholangiocyte and Mesenchymal Heterogeneity. Hepatol Commun 6 , 821-840 (2022). https://doi.org:10.1002/hep4.1854 Andrews, T. S. et al. Single-cell, single-nucleus, and spatial transcriptomics characterization of the immunological landscape in the healthy and PSC human liver. J Hepatol 80 , 730-743 (2024). https://doi.org:10.1016/j.jhep.2023.12.023 Gissen, P. & Arias, I. M. Structural and functional hepatocyte polarity and liver disease. J Hepatol 63 , 1023-1037 (2015). https://doi.org:10.1016/j.jhep.2015.06.015 Zhang, K. et al. In Vitro Expansion of Primary Human Hepatocytes with Efficient Liver Repopulation Capacity. Cell Stem Cell 23 , 806-819 e804 (2018). https://doi.org:10.1016/j.stem.2018.10.018 Blidisel, A. et al. Experimental Models of Hepatocellular Carcinoma-A Preclinical Perspective. Cancers (Basel) 13 (2021). https://doi.org:10.3390/cancers13153651 Kimlin, L. C., Casagrande, G. & Virador, V. M. In vitro three-dimensional (3D) models in cancer research: an update. Mol Carcinog 52 , 167-182 (2013). https://doi.org:10.1002/mc.21844 Fujii, M. et al. Human Intestinal Organoids Maintain Self-Renewal Capacity and Cellular Diversity in Niche-Inspired Culture Condition. Cell Stem Cell 23 , 787-793 e786 (2018). https://doi.org:10.1016/j.stem.2018.11.016 Spence, J. R. et al. Directed differentiation of human pluripotent stem cells into intestinal tissue in vitro. Nature 470 , 105-109 (2011). https://doi.org:10.1038/nature09691 Mitrofanova, O. et al. Bioengineered human colon organoids with in vivo-like cellular complexity and function. Cell Stem Cell 31 , 1175-1186 e1177 (2024). https://doi.org:10.1016/j.stem.2024.05.007 Aloia, L. et al. Epigenetic remodelling licences adult cholangiocytes for organoid formation and liver regeneration. Nat Cell Biol 21 , 1321-1333 (2019). https://doi.org:10.1038/s41556-019-0402-6 Cordero-Espinoza, L. et al. Dynamic cell contacts between periportal mesenchyme and ductal epithelium act as a rheostat for liver cell proliferation. Cell Stem Cell 28 , 1907-1921 e1908 (2021). https://doi.org:10.1016/j.stem.2021.07.002 Planas-Paz, L. et al. The RSPO-LGR4/5-ZNRF3/RNF43 module controls liver zonation and size. Nat Cell Biol 18 , 467-479 (2016). https://doi.org:10.1038/ncb3337 Pepe-Mooney, B. J. et al. Single-Cell Analysis of the Liver Epithelium Reveals Dynamic Heterogeneity and an Essential Role for YAP in Homeostasis and Regeneration. Cell Stem Cell 25 , 23-38 e28 (2019). https://doi.org:10.1016/j.stem.2019.04.004 Roos, F. J. M. et al. Human branching cholangiocyte organoids recapitulate functional bile duct formation. Cell Stem Cell 29 , 776-794 e713 (2022). https://doi.org:10.1016/j.stem.2022.04.011 Ober, E. A. & Lemaigre, F. P. Development of the liver: Insights into organ and tissue morphogenesis. J Hepatol 68 , 1049-1062 (2018). https://doi.org:10.1016/j.jhep.2018.01.005 Gordillo, M., Evans, T. & Gouon-Evans, V. Orchestrating liver development. Development 142 , 2094-2108 (2015). https://doi.org:10.1242/dev.114215 Zorn, A. M. in StemBook (2008). Sampaziotis, F. et al. Cholangiocyte organoids can repair bile ducts after transplantation in the human liver. Science 371 , 839-846 (2021). https://doi.org:10.1126/science.aaz6964 Hu, H. et al. Long-Term Expansion of Functional Mouse and Human Hepatocytes as 3D Organoids. Cell 175 , 1591-1606 e1519 (2018). https://doi.org:10.1016/j.cell.2018.11.013 Peng, W. C. et al. Inflammatory Cytokine TNFalpha Promotes the Long-Term Expansion of Primary Hepatocytes in 3D Culture. Cell 175 , 1607-1619 e1615 (2018). https://doi.org:10.1016/j.cell.2018.11.012 Prior, N. et al. Lgr5(+) stem and progenitor cells reside at the apex of a heterogeneous embryonic hepatoblast pool. Development 146 (2019). https://doi.org:10.1242/dev.174557 Wesley, B. T. et al. Single-cell atlas of human liver development reveals pathways directing hepatic cell fates. Nat Cell Biol 24 , 1487-1498 (2022). https://doi.org:10.1038/s41556-022-00989-7 Hendriks, D. et al. Mapping of mitogen and metabolic sensitivity in organoids defines requirements for human hepatocyte growth. Nat Commun 15 , 4034 (2024). https://doi.org:10.1038/s41467-024-48550-4 Nuciforo, S. & Heim, M. H. Organoids to model liver disease. JHEP Rep 3 , 100198 (2021). https://doi.org:10.1016/j.jhepr.2020.100198 Marsee, A. et al. Building consensus on definition and nomenclature of hepatic, pancreatic, and biliary organoids. Cell Stem Cell 28 , 816-832 (2021). https://doi.org:10.1016/j.stem.2021.04.005 Thompson, W. L. & Takebe, T. Human liver model systems in a dish. Dev Growth Differ 63 , 47-58 (2021). https://doi.org:10.1111/dgd.12708 Dowbaj, A. M., Kohler, T. N., Cordero-Espinoza, L., Hollfelder, F. & Huch, M. Generation of liver mesenchyme and ductal cell organoid co-culture using cell self-aggregation and droplet microfluidics. STAR Protoc 4 , 102333 (2023). https://doi.org:10.1016/j.xpro.2023.102333 Tanimizu, N. et al. Generation of functional liver organoids on combining hepatocytes and cholangiocytes with hepatobiliary connections ex vivo. Nat Commun 12 , 3390 (2021). https://doi.org:10.1038/s41467-021-23575-1 Damm, G. et al. Effect of glucose and insulin supplementation on the isolation of primary human hepatocytes. EXCLI J 18 , 1071-1091 (2019). https://doi.org:10.17179/excli2019-1782 Kegel, V. et al. Protocol for Isolation of Primary Human Hepatocytes and Corresponding Major Populations of Non-parenchymal Liver Cells. J Vis Exp , e53069 (2016). https://doi.org:10.3791/53069 Broutier, L. et al. Human primary liver cancer-derived organoid cultures for disease modeling and drug screening. Nat Med 23 , 1424-1435 (2017). https://doi.org:10.1038/nm.4438 Belenguer, G. et al. RNF43/ZNRF3 loss predisposes to hepatocellular-carcinoma by impairing liver regeneration and altering the liver lipid metabolic ground-state. Nat Commun 13 , 334 (2022). https://doi.org:10.1038/s41467-021-27923-z Hu, M. et al. Wnt/beta-catenin signaling in murine hepatic transit amplifying progenitor cells. Gastroenterology 133 , 1579-1591 (2007). https://doi.org:10.1053/j.gastro.2007.08.036 Russell, J. O. & Monga, S. P. Wnt/beta-Catenin Signaling in Liver Development, Homeostasis, and Pathobiology. Annu Rev Pathol 13 , 351-378 (2018). https://doi.org:10.1146/annurev-pathol-020117-044010 Zucman-Rossi, J., Villanueva, A., Nault, J. C. & Llovet, J. M. Genetic Landscape and Biomarkers of Hepatocellular Carcinoma. Gastroenterology 149 , 1226-1239 e1224 (2015). https://doi.org:10.1053/j.gastro.2015.05.061 Janda, C. Y. et al. Surrogate Wnt agonists that phenocopy canonical Wnt and beta-catenin signalling. Nature 545 , 234-237 (2017). https://doi.org:10.1038/nature22306 Kastan, N. et al. Small-molecule inhibition of Lats kinases may promote Yap-dependent proliferation in postmitotic mammalian tissues. Nat Commun 12 , 3100 (2021). https://doi.org:10.1038/s41467-021-23395-3 Hwang, E. S. & Song, S. B. Possible Adverse Effects of High-Dose Nicotinamide: Mechanisms and Safety Assessment. Biomolecules 10 (2020). https://doi.org:10.3390/biom10050687 Bebelman, M. P. et al. Hepatocyte apical bulkheads provide a mechanical means to oppose bile pressure. J Cell Biol 222 (2023). https://doi.org:10.1083/jcb.202208002 Segovia-Miranda, F. et al. Three-dimensional spatially resolved geometrical and functional models of human liver tissue reveal new aspects of NAFLD progression. Nat Med 25 , 1885-1893 (2019). https://doi.org:10.1038/s41591-019-0660-7 Morales-Navarrete, H. et al. A versatile pipeline for the multi-scale digital reconstruction and quantitative analysis of 3D tissue architecture. Elife 4 (2015). https://doi.org:10.7554/eLife.11214 Namoto, K. et al. NIBR-LTSi is a selective LATS kinase inhibitor activating YAP signaling and expanding tissue stem cells in vitro and in vivo. Cell Stem Cell 31 , 554-569 e517 (2024). https://doi.org:10.1016/j.stem.2024.03.003 Yimlamai, D. et al. Hippo pathway activity influences liver cell fate. Cell 157 , 1324-1338 (2014). https://doi.org:10.1016/j.cell.2014.03.060 Watson, B. R. et al. Spatial transcriptomics of healthy and fibrotic human liver at single-cell resolution. Nat Commun 16 , 319 (2025). https://doi.org:10.1038/s41467-024-55325-4 Yakubovsky, O. et al. A spatial transcriptomics atlas of live donors reveals unique zonation patterns in the healthy human liver. bioRxiv , 2025.2002.2022.639181 (2025). https://doi.org:10.1101/2025.02.22.639181 Azuma, H. et al. Robust expansion of human hepatocytes in Fah-/-/Rag2-/-/Il2rg-/- mice. Nat Biotechnol 25 , 903-910 (2007). https://doi.org:10.1038/nbt1326 Guilliams, M. et al. Spatial proteogenomics reveals distinct and evolutionarily conserved hepatic macrophage niches. Cell 185 , 379-396 e338 (2022). https://doi.org:10.1016/j.cell.2021.12.018 Wells, R. G. The portal fibroblast: not just a poor man's stellate cell. Gastroenterology 147 , 41-47 (2014). https://doi.org:10.1053/j.gastro.2014.05.001 Lei, L. et al. Portal fibroblasts with mesenchymal stem cell features form a reservoir of proliferative myofibroblasts in liver fibrosis. Hepatology 76 , 1360-1375 (2022). https://doi.org:10.1002/hep.32456 Dowbaj, A. M. et al. Mouse liver assembloids model periportal architecture and biliary fibrosis. Nature (2025). https://doi.org:10.1038/s41586-025-09183-9 Jin, C. et al. Single-cell RNA sequencing reveals the pro-inflammatory roles of liver-resident Th1-like cells in primary biliary cholangitis. Nat Commun 15 , 8690 (2024). https://doi.org:10.1038/s41467-024-53104-9 Hammerich, L. & Tacke, F. Hepatic inflammatory responses in liver fibrosis. Nat Rev Gastroenterol Hepatol 20 , 633-646 (2023). https://doi.org:10.1038/s41575-023-00807-x Kisseleva, T. & Brenner, D. Molecular and cellular mechanisms of liver fibrosis and its regression. Nat Rev Gastroenterol Hepatol 18 , 151-166 (2021). https://doi.org:10.1038/s41575-020-00372-7 Reza, H.A. et al. Multi-zonal liver organoids from human pluripotent stem cells. Nature 641 , 1258-1267 (2025). https://doi.org/10.1038/s41586-025-08850-1 Igarashi, R. et al. Generation of human adult hepatocyte organoids with metabolic functions. Nature 641 , 1248-1257 (2025). https://doi.org:10.1038/s41586-025-08861-y Marsee, A. et al. Mass Generation and Long-term Expansion of Hepatobiliary Organoids from Adult Primary Human Hepatocytes. bioRxiv , 2024.2006.2010.598262 (2024). https://doi.org:10.1101/2024.06.10.598262 Mallanna, S. K. et al. Expandable, Functional Hepatocytes Derived from Primary Cells Enable Liver Therapeutics. bioRxiv , 2024.2012.2028.630269 (2024). https://doi.org:10.1101/2024.12.28.630269 Westerfield, A. D. et al. A 3D in vitro model of the human hepatobiliary junction. bioRxiv , 2025.2007.2011.664464 (2025). https://doi.org:10.1101/2025.07.11.664464 Roskams, T. & Desmet, V. Ductular reaction and its diagnostic significance. Semin Diagn Pathol 15 , 259-269 (1998). Material and Methods Human specimens All human liver tissues used in this study were obtained after informed consent was obtained from patients undergoing operations at either the Department of Visceral, Thoracic and Vascular Surgery (VTG), University Hospital Carl Gustav Carus Dresden (UKD) or at Leipzig University Medical Center. Informed consent was obtained from all participants. The use of the human samples for this study was approved by the corresponding institutional review boards of either the University Hospital Carl Gustav Carus Dresden (ethical vote BO-EK-57022020, ratified on 2020/03/10) or the Leipzig University Hospital (Ethical vote: registration number 322/17-ek, date 2020/06/10 ratified on 2021/11/30 and registration number 450/21-ek, date 2021/11/21 ratified on 2024/10/04). Four samples (F-PHH1-PHH5) were obtained from Lonza Pharma&Biotech - Bioscience Solutions. All information regarding the human samples (sex, age) is provided in Supplementary Table 1 and 2. Isolation of primary human hepatocytes and cholangiocytes Primary human hepatocytes (PHHs) were isolated using a two-step collagenase perfusion method. The human liver tissue received from UKD was perfused with Solution A (composed of 10 mM HEPES, and 2.5 mM EGTA in HBSS) at 39ºC for at least 20 minutes, with a ratio of 15 mL/20 seconds. Subsequently, the perfusion solution was switched to Solution B (containing 100 mM HEPES, 4.8 mM CaCl 2 , and 1 g/L Collagenase P, in HBSS) and perfused at 37ºC for 5-15 min, also at a ratio of 15 mL/20 sec. The digestion process was halted by adding cold Williams' Medium E supplemented with 1% HEPES, 1% GlutaMax, and 1% Penicillin/Streptomycin. The primary human hepatocytes were detached from the tissue by shaking using forceps and combing the cells out of the tissue. Afterwards they were filtered through a 100 µm nylon cell strainer. Cells were then spun at 50G for 5 minutes, and the resulting pellet was resuspended in cold-Williams’ Medium E supplemented with 1% HEPES, 1% GlutaMax and 1% Penicillin/Streptomycin. The cell suspension was kept cold and centrifuged again at 50G for 5 minutes. For samples obtained from Leipzig University Hospital the perfusion procedure differed slightly: Solution A [(composed of 10 mM HEPES (Carl Roth), 143 mM NaCl, 6.7 mM KCl, 2.4 mM EGTA, 5mM N-Acetyl-l-cysteine, 11 mM D-Glucose (all provided by Sigma Aldrich) and 32 U/L human Insulin (Eli Lilly) in ddH 2 O (pH 7.4)) at 39ºC with a ratio of 25 mL/min for at least 20 min. Then the perfusion solution was switched to Solution B (composed of 67 mM NaCl, 6.7 mM KCl, 10 mM HEPES, 0.5% BSA, 4.8 mM CaCl 2 x 2H 2 O (all provided by Sigma Aldrich), and 1 g/L Collagenase P (Roche) in ddH 2 O (pH 7.6), diluted 1:2 in Stop Solution (composed of DPBS with Ca 2+ , Mg 2+ (Gibco), supplemented with 16.7% FBS (Merck)) and perfused at 39 °C for 5-15 min at a ratio of 25ml/min. The digestion was stopped by adding cold Stop Solution. Hepatocytes were filtered through a funnel with gauze (Hartmann) and centrifuged at 51G for 5 min. Cell pellets were washed in DPBS with Ca 2+ , Mg 2+ , centrifugated at 51G for 5 min, and resuspended in William’s Medium E (supplemented with 10% FBS (Merck), 15mM HEPES, 1mM sodium pyruvate, 1% Penicillin/Streptomycin, 1% MEM NEAA (all provided by Gibco), 1 µg/mL Dexamethasone (Jenapharm), and 32 U/L human insulin (Eli Lilly)). The isolated primary human hepatocytes (PHHs) were shipped overnight in ChillProtec plus® medium (Biochrom). Frozen hepatocytes (F-PHH1-f-PHH5, Supplementary Table 2) commercially available from Lonza were defrosted using Human Hepatocyte Thawing Medium (Lonza) following manufacturer’s instructions. The isolated PHHs preparations (either from fresh tissue from Dresden or Leipzig Hospitals or commercially available frozen hepatocytes) were enriched for both EpCAM-negative (hepatocytes) and EpCAM-positive (cholangiocytes) by Magnetic-activated cell sorting (MACS) using an anti-human CD326 antibody (Biolegend) and Anti-Biotin Microbeads (Ultra Pure, Miltenyi) following manufacturer’s instructions. The EpCAM-negative fraction with a viability >50% (Supplementary Table 1) was used to generate hepatocyte organoids as described below (section Hepatocyte organoid culture). The EpCAM-positive fraction, formed by human cholangiocytes, was used to generate human cholangiocyte organoids (h-CholOrg) as described in 4,5 and in section Cholangiocyte organoid culture. A digestion method without perfusion, as the one detailed in Huch et al., 2015 4 , only generated h-CholOrg. HepOrg were not formed under non-perfused protocols. The complete list of patients used and the comparative between digestion and perfusion are provided in Supplementary Table 1-2. Flow cytometric validation of PHH purity following MACS enrichment Freshly isolated PHHs and MACS-enriched EpCAM-negative PHHs (as described above) were centrifuged at 80g for 5 min. Pellets were resuspended in HBSS containing 2% FBS and incubated on ice for 10 min (blocking). After centrifugation (80g, 5 min), cells were resuspended in HBSS with 1% FBS, stained with EpCAM-Alexa 488 (5 μL/test, Biolegend, Cat. 53-8326-42), and incubated for 45 min on ice. Cells were then washed twice with HBSS containing 1% FBS, centrifuged, and resuspended in 200 μL HBSS with 1% FBS, DAPI (1:1000), and DNase I (1:1000) for flow cytometric analysis. Cholangiocyte organoid culture EpCAM-positive cholangiocytes were mixed with Matrigel Growth Factor Reduced (Matrigel, Corning) or Cultrex Basement Membrane Extract 2 (BME2) (Cultrex-RGF Basement Membrane Extract Type 2- BME2 (AMSBIO) at a 50,000 cells per 50μL/well of 24 well plate and cultured at 37°C and 5% CO 2 in h-CholOrg-EM medium as described in: [AdDMEM/F12 medium containing 1% HEPES, 1% Penicillin/Streptomycin, Glutamax, 1x B27 and 1.25 mM N-acetylcysteine (Sigma) supplemented with 10 nM gastrin (Merck/Sigma), 50 ng/mL hEGF (Peprotech, Peprotech), 10% RSPO1 conditioned medium (homemade), 100 ng/mL FGF10 (Peprotech), 10 mM nicotinamide (Merck/Sigma) and 25 ng/mL HGF (Peprotech)], 5 uM A8301 (Tocris) and 10 uM Forskolin (Tocris #1099). For the first 3-5 days in culture this medium was supplemented with 30% WNT3a conditioned medium (Wnt-CM) (homemade), 25 ng/mL Noggin (Peprotech) and 10 μM ROCK inhibitor (Ri) (Y-27632, Merck/Sigma). The grown cholangiocyte organoids were passaged at a 1:3 ratio once a week as described in 5 . Hepatocyte Organoid culture For hepatocyte organoid cultures, the isolated PHHs (EpCAM-negative fraction) were mixed with Matrigel (Corning) or BME2 (AMSBIO), and 12,500-50,000 cells were seeded in 50μL Matrigel or BME2 per well in 24 well plates and incubated at 37°C and 5% CO 2 . After Matrigel solidification, culture medium was added. The culture medium was based on AdDMEM/F12 (Invitrogen) supplemented with 1% HEPES, 1% GlutaMax (ThermoFisher), 1% Penicillin/Streptomycin (ThermoFisher), 1X B27 without retinoic acid (Gibco), 1.25 mM N-Acetylcysteine (Sigma), 10 nM Gastrin (Sigma), and the following growth factors: 50 ng/mL hEGF (Peprotech), 15% RSPO1 conditioned media (home-made), 100 ng/mL FGF10 (Peprotech), 100 ng/mL FGF7 (Peprotech), 50 ng/mL HGF(Peprotech), 10 mM Nicotinamide (Sigma, for EM 1 medium only), 2 µM A83-01 (Tocris), 3 µM CHIR99021 (Tocris), 10 µM Y-27632 (Tocris), 0.5 nM Wnt Surrogate FC Fusin Protein as in Janda et al. 50 (IPA, N001), and 10 µM TRULI (Axon) or TDI-011536 (Selleckchem). After one week to ten days, the organoids were removed from the Matrigel, mechanically dissociated into small fragments using TryplE TM Express (Gibco), and transferred to fresh Matrigel. Passaging was performed once per week at a 1:2 split ratio for at least three months. For preparation of frozen stocks, the organoid cultures were dissociated, mixed with Recovery cell culture freezing medium (Gibco), and frozen following standard procedures. For the optimization of culture conditions, media component screening experiments were performed where each of the components Amphiregulin (AREG, 100 ng/mL, R&D systems; Dexamethasone (1.6 µM, Sigma); G-CSF (50 ng/mL, R&D systems), IL-6 (2 ng/mL, R&D systems), M-3m3FBS (Phospholipase C activator, 25 µM, Tocris), TGFa (100 ng/mL), TRULI (Axon) was added to our previously published mouse hepatoblast medium (MM 35 ) with minor modification: AdDMEM/F12 (Invitrogen) supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, 1X B27 without retinoic acid, 1.25 mM N-Acetylcysteine, 10 nM Gastrin, 50 ng/mL hEGF, 15% RSPO1 conditioned media, 100 ng/mL FGF10, 100 ng/mL FGF7, 50 ng/mL HGF, 10 mM Nicotinamide, 2 µM A83-01, 3 µM CHIR99021, 10 µM Y-27632, and 0.5 nM Wnt Surrogate FC Fusion Protein. Note that the addition of TRULI alone resulted in a significant increase in organoid formation efficiency (Fig.1c,e). However, after 1-2 splits, the cultures rapidly deteriorated and could not be expanded further (Fig.1f). For h-HepOrgs hepatic differentiation, hHepOrg were expanded in EM2 medium, split, seeded and cultured for 2-5 days under EM1 culture medium after which medium was changed to hepatic differentiation medium (DM) composed of: AdDMEM/F12 supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, 1X B27 without retinoic acid, 1.25 mM N-Acetylcysteine, 50 ng/mL hEGF, 15% RSPO1 conditioned media, 50 ng/mL HGF, 2 µM A83-01, 3 µM CHIR99021, 10 µM Y-27632, 0.5 nM Wnt Surrogate FC Fusion Protein, 100 ng/mL FGF19 (R&D systems), and 1.6 µM Dexamethasone (Sigma). Hepatic differentiation medium (DM) was changed every 2-3 days for 7 days. For organoid formation efficiency, primary hepatocytes were isolated and cultured in different media as described above. To prevent organoids from fusing, 25,000 (for EM2 medium) or 50,000 (all other media) viable hepatocytes (viability >80%) were plated in 50μL Matrigel or BME2 and cultured as described above. After 12-14 days, organoid numbers were counted and results expressed as a percentage relative to the initial seeding cell numbers. Isolation of human liver portal fibroblasts Human liver portal fibroblasts (hPFs) were isolated from human liver tissues by collagenase digestion. Briefly, human liver tissue was minced and rinsed with cold-DMEM (Gibco) supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 1% FBS. Minced tissues were incubated with a collagenase solution consisting of 2.5 mg/mL Collagenase D (Roche), 0.1 mg/mL DNase I (Sigma), and 10 µM Y-27632 in DMEM supplemented with 1% HEPES, 1% GlutaMax, and 1% Penicillin/Streptomycin. The incubation was carried out for 30-60 minutes at 37ºC on a shaker set at 120 rpm. The digestion was halted by adding cold-DMEM supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 1% FBS. The suspension was then filtered through a 70 µm cell strainer and centrifuged for 5 minutes at 300G. After removing the supernatant, the cell pellet was resuspended in cold-DMEM supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 1% FBS. The suspension was centrifuged again for 5 minutes at 300G, and the resulting pellet was resuspended in cold-DMEM supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 20% FBS. For sorting, hPFs were stained with 1 μg/test Anti-human CD90 (THY1)-APC, 20 μL/test Anti-human CD140a (PDGFRa)-PE, Anti-CD11b/CD31/CD45-PECy7, and EpCAM-Alexa 488 for 30 minutes on ice and washed twice. THY1-positive hPFs were sorted using a BD FACSAria Fusion and cultured in DMEM supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 20% FBS at 37°C and 5% CO 2 until used for assembloid formation or freeze for biobanking. Virus infection For portal fibroblast infections, cultures (passage 0-1) grown in DMEM +++ with 20% foetal bovine serum (FBS) (Merck/Sigma, #F7524) were washed with PBS and dissociated to single cells by incubating with TrypLE 1x for 6min at 37°C. And the cell concentration was determined by manual counting in haemocytometer, 10,000 cells were plated into 48-well plates and the medium mixed with nRFP-lenti-virus or nGFP-lenti-virus (LVP360-R and LVP360-G, GenTarget Inc) was replaced after 12h and the solution was changed after 72h. For cholangiocyte organoid infection, duct cells (phase 0-1) were extracted from the Matrigel and digested with TrypLE to prepare single-cell suspensions as described in Broutier et al. , 2016 5 , which were then manually counted using a haemocytomer counter to determine cell concentration. In a 48-well plate, 150 ul of cells and 50 ul of virus suspension were added to achieve an MOI= 10-30, mixed thoroughly, centrifuged at 600g for 60 minutes at 32°C, and incubated for 6 hours at 37°C, 5% CO 2 . Cells were collected in 1. 5 ml tubes, centrifuged at 600g for 5 minutes, infected medium was discarded, and cells were resuspended in 25 ul of Matrigel, followed by the addition of EM medium (supplemented with 30% WntCM, 25ng/ml noggin and 10 µM Y-27632 for the first 3 days). Periportal assembloids To generate liver periportal assembloids comprising hepatocytes, cholangiocytes, and portal fibroblasts, we first prepared the cellular components as follows: nGFP-labelled cholangiocyte organoids (passage 5-11), grown in cholangiocyte expansion medium (h-CholOrg-EM) as detailed above, were collected from Matrigel using cold AdDMEM/F12 (ThermoFisher, #15630-056) containing 1% HEPES (ThermoFisher, #15140-122), 1% penicillin/streptomycin (ThermoFisher, #15140-122), and 1% glutamine (ThermoFisher, #350-068). Matrigel was removed and organoids were dissociated to single cell using pre-warmed TrypLE 1x (ThermoFisher, #12605010) for 7-12 minutes at 37°C. nRFP-labelled portal fibroblasts cultures (passage 5-12) grown in DMEM +++ with 20% foetal bovine serum (FBS) (Merck/Sigma, #F7524) were washed with PBS and dissociated to single cells by incubating with TrypLE 1x for 6min at 37°C. Both single-cell suspensions were spun at 200 RCF for 5 minutes, resuspended in DM medium described above but without A8301, and then manually counted with a hemocytometer to determine cell concentration. Cultured HepOrg from EM2 were split and transferred to EM1 for 2d and then to DM medium for 3d. Hepatocyte organoids were then collected and washed using cold AdDMEM/F12 supplemented with 1% HEPES, 1% Pen/Strep and 1% Glutamax and incubated for 10min on ice using cold Cell Recovery Solution (Corning, #354253) to remove the ECM. HepOrg were then resuspended using DM without A8301 and placed into low-attachment 6 well plate and differentiated organoids (with bubbly morphology) were selected and hand-picked under a stereomicroscope. To define an approach for human periportal liver assembloid formation several iterations were performed. First, we sought to identify a medium that would support assembloid formation. i.e., the culture of all three cell types: hepatocytes, cholangiocytes/ductal cells and human portal mesenchyme without overgrowth of any of them, we tested several media and found that a minor adaptation of the differentiation medium used for h-HepOrgs DM without A8301 (assembloids medium) supported the culture of the three cell types while preventing their overgrowth. To determine the optimal quantities of the three cell types required for periportal assembloid formation, we first investigated the proportions of portal fibroblast and ductal cells in healthy human periportal liver tissue. We observed that the ratio varies donor-to-donor from 1:1 to 4:1 ductal cells per fibroblasts. Therefore, we tested these range of ratios in vitro by varying the proportions of mesenchyme and ductal cells that were mixed with 1 single HepOrg (~200 um diameter). In short, in 96-well low-attachment U-plates (cat n Corning, cat. n: #7007), we assembled (as described below) 1 hepatocyte organoid (h-HepOrgs) with 25 portal fibroblasts and 25 or 50 or 100 or 200 cholangiocytes, or with 100 cholangiocytes and 50 or 100 portal fibroblast cells. We selected the proportion 25 portal fibroblasts: 100 cholangiocytes/ductal cells. In AggreWell TM plates (Aggrewell 800, Stem Cell Technologies #34811), we scaled up proportionally taking into account that Agrewell 800 has 300 microwells in each well and used 7,500 portal fibroblasts, 30,000 cholangiocytes, and 100 hepatocyte organoids (proportion 1HepOrg :75 PFs: 300Chol). For non-healthy/non-physiological ratios, use 500 portal fibroblasts, 100 cholangiocytes, and 1 hepatocyte organoid for 96-well low-attachment U-plates, and 15,0000 portal fibroblasts, 30,000 cholangiocytes, and 50 hepatocyte organoids for Aggrewell plates. For the assembly in MW96 we mixed fibroblasts and cholangiocytes in 96-well low adhesion U-plates using 150 μL of DM (without A8301) medium with 2.4mg/ml methylcellulose (MeC, Sigma, #M6385) and spun at 50 RCF for 5min. Individual HepOrgs were then added to the well and the mixture was incubated for 18-24h at 37°C and 5% CO 2 . For the assembly in AggreWell TM , plates were first pre-treated as recommended by the manufacturer. Then, ductal and mesenchymal cells and HepOrgs were mixed in 1.5 mL of DM (without A8301) medium with 2.4mg/ml methylcellulose, spun down 5min at 50 RCF and incubated for 18-24h at 37°C and 5% CO 2 . After 18-24h in suspension in the 96well/aggrewell plate, the cell suspension was collected with a 1 mL pipette and transferred to a low-attachment 6-well plate. The structures were handpicked under a stereomicroscope and seeded in 25 μL matrigel dome in pre-warmed 48-well plates. The Matrigel was allowed to solidify for 30min at 37°C 5% CO 2 and the wells were overlayed with further 300 μL of DM (without A8301) medium. Medium was changed every 3-4 days. Under these conditions, 70% of the initial Cholangiocytes formed a lumen. Raw data were incorporated into the quantification of periportal-like spatial organization in assembloids (Source data file from Extended Data Fig. 7e). Immunostaining of organoids and assembloids For immunofluorescence staining, organoids and assembloids were first extracted from Matrigel with ice-cold Cell Recovery solution and then fixed for 30 min with 4% paraformaldehyde (PFA) at 4°C. Fixed organoids were washed and transferred to µ-Slide 8 Well Chamber Slide (glass bottom, Ibidi). Blocking and permeabilization was performed for 1 hour at RT in PBS containing 2% BSA and either 0.1%, 0.2%, 0.5 % or 1% Triton X-100 depending on antigen (see Supplementary dataset_5). The samples were incubated with primary antibodies overnight at 4°C in blocking solution. After that, the antibody was washed with 3 washes with PBS and the samples were incubated overnight at 4°C or for 8h at RT with secondary antibodies diluted in blocking solution and, if required, also Phalloidin and DAPI were added to the secondary antibody mix. The samples were washed 3 times with PBS and subsequently cleared using Fructose-Glycerol clearing solution (25mL Glycerol, 5.3mL dH2O and 22.5g Fructose – 60% Glycerol and 2.5M Fructose). The samples were stored in PBS until cleared for imaging as described above. The antibodies and dilutions used are listed in Supplementary Dataset_5. For H&E staining, organoids were collected in cold DPBS (Gibco) and fixed with 4% PFA for 30 min and dehydrated and embedded in paraffin using standard methods. Paraffin sections (8 μm) were cut and stained for H&E using standard protocols. Immunostaining of thin and thick tissue sections For thin tissue sections (8-12 μm) and staining, human liver tissues were fixed in 10% formalin overnight whilst rolling at 4°C. After fixation, fixed tissues are washed with PBS and incubated with 10% sucrose for 1-2h, then transferred to 30% sucrose in PBS for 24h and subsequently embedded in OCT compound (VWR, #361603E) to generate OCT-cryopreserved tissue blocks. Tissue blocks were cryo-sectioned on ThermoScientific CryoStar NX70 cryostat. Sections were blocked in PBS with 10% Donkey Serum (DS) and 0.1% Triton X-100 for 2h at RT, incubated with primary antibodies diluted in PBS with 3% Donkey Serum and 0.1% Triton X-100 overnight at 4°C and subsequently washed and incubated with secondary antibodies diluted in 0.05% PBS- BSA and DAPI for 2h at RT. Sections were mounted in Vectashield. The list of used antibodies is available in Supplementary Dataset_5. For thick tissue sections and staining, the protocol from (Fabián Segovia-Miranda et al . 54 was used. Immediately after surgical resection, the liver tissue samples were cut into smaller pieces and fixed in 4% PFA for 24 h on a rotator at 4°C, washed 3 times with PBS, followed by quenching with 50mM ammonium chloride solution (NH 4 Cl) for 24h and again washed 3 times with PBS. For storage, liver pieces were kept in PBS at 4°C. For sectioning, livers were embedded in moulds with 4% low-melting agarose (Bio-Rad #1613111) in PBS and cut into 50 or 100 μm-thick sections on a vibratome (Leica VT1200S). For deep tissue imaging, if antigen retrieval was required, tissue sections were placed in Eppendorf tubes with pre-warmed 1X citrate buffer (Sigma-Aldrich, #C9999), pH=6, at 80C for 30 min in a shaking heating block, and then washed 3 times with PBS. Tissue sections were permeabilized with 0.5% Triton X-100 in PBS for 1h at RT. The primary antibodies were diluted in Tx buffer (0.2% gelatin, 300 mM NaCl, and 0.3% Triton X-100 in PBS) and incubated for 48 h at RT. After washing 3×15 min with 0.3% Triton X-100 in PBS, the sections were incubated with secondary antibodies, DAPI (1 mg/mL; 1:1000) and phalloidin for another 48 h. After washing 3×15 min with 0.3% TritonX-100 in PBS and 3×1 min with PBS, the optical clearing started by incubating the slices in 25% fructose for 4 h, continued in 50% fructose for 4 h, 70% fructose overnight, 100% fructose (100% wt/vol fructose, 0.5% 1-thioglycerol, and 0.1 M phosphate buffer, pH 7.5) overnight, followed by a final overnight incubation in SeeDB solution (80.2% wt/wt fructose, 0.5% 1-thioglycerol, and 0.1 M phosphate buffer) 80 . The samples were mounted in SeeDB. The list of used antibodies and dyes is available in Supplementary Dataset_5. For immunohistochemistry of xenotransplant mice tissue sections, the mice liver tissue samples were cut into smaller pieces and fixed in 10% formalin overnight. Sections (4 μm) were subjected to immunohistochemical staining, which was performed using a Dako REAL EnVision Detection System (Dako, #K5007). Anti-hGAPDH antibody (Abcam, #ab128915) was used as the primary antibody and nuclei were counterstained with hematoxylin. Stained tissues were viewed under a Virtual Slide System (Leica, ScanScope CS2). The immunohistochemistry analysis for PDGFRA, DCN and ASPN in human healthy liver tissue was obtained from the publicly available image dataset from Human Protein Atlas (HPA) 81 (version#24proteinatlas.org). The corresponding URL is indicated in the figure legend. Imaging of organoids, assembloids and tissues Bright field images of organoids were obtained with a Leica DMIL LED inverted microscope and Leica DFC 450C camera or with a Leica M80 stereoscope and MC170HD camera and Leica LAS software. H&E staining of organoids were obtained with a Leica DM4B microscope and DMC5400 camera and Leica LAX software. Confocal images of organoids and thick tissue sections were acquired on an inverted single photon point scanning confocal microscope (Zeiss Cell Discoverer 7 with LSM 900 and Airyscan 2) using a Zeiss APOCHROMAT 20x/0.95 Autocorr air objective, with a tube lens of 0.5x or 1x, and a voxel size of 0.4 x 0.4 x 0.5 μm or 0.5 x 0.5 x 0.5 μm for organoids and 0.3 x 0.3 x 0.3 μm for thick tissue sections. Laser lines 405, 488, 561 and 640 were used for excitation of fluorophores and GaAsP-PMT detectors were used for detection. High-resolution Airyscan images were acquired using this system for imaging polarity in detail for the tissue sections with a a voxel size of 0.0823 x 0.0823 x 0.3 μm. Image processing was done using ZEN software or ImageJ/Fiji. Imaging of assembloids and thin tissue sections was performed using an inverted multiphoton laser-scanning microscope (Zeiss LSM 780 NLO). In order to improve the resolution, image denoising was performed with deconvolution using HuygensPro. Raw image stacks were imported into the software, and a point spread function (PSF) was either estimated based on the imaging conditions (numerical aperture, wavelength, and refractive index) or obtained from PSF calibration images. The HuygensPro Classic Maximum Likelihood Estimation (CMLE) algorithm was applied for deconvolution, with an iteration stop criterion based on optimal signal-to-noise ratio and minimal change in successive iterations. Image analysis The quantification of the percentage of YAP-positive nuclei and YAP-negative nuclei was performed using a custom-made pipeline using the Arivis 4D Pro, software (Version 4.2.0). The steps included were background correction, denoising, nuclear segmentation based on DAPI, and quantification of the fluorescence intensity of YAP immunofluorescent staining within the nuclei. The total number of nuclei and the number of YAP-positive nuclei were quantified, and subsequently, the number of YAP-negative nuclei was calculated by subtracting the number of YAP-positive nuclei from the total number of nuclei. Finally, the percentage of YAP-positive and YAP-negative nuclei were calculated. For the quantification of cytoplasmic to nuclear area, a custom-made pipeline was developed using the Arivis 4D Pro, software (Version 4.2.0). For this, a representative 2D z-slice was taken from each organoid. Pre-processing steps included background correction on the Phalloidin channel (marking cell borders) and normalization and denoising on the DAPI channel (marking nuclei). To obtain the nuclear area, nuclear segmentation was done based on DAPI, followed by quantification of the total nuclear area. For the cytoplasmic area, segmentation was done based on Phalloidin to obtain the outline of the area occupied by the cytoplasm. Finally, the ratio of cytoplasmic area to nuclear area was calculated. For 3D visualization of bile canaliculi, high-resolution images obtained as described above. Segmentation was performed on CD13 (for bile canaliculi) and F-actin (cell borders) staining with phalloidin. The analysis of bile canaliculi morphology and bile canaliculi network properties was performed using a custom-made Fiji script publicly available at https://github.com/JulienDelpierre/BileCanaliculiSegmentation. The script description can be found in Dowbaj, Sljukic et al . 64 . Briefly, immunofluorescence images from several conditions were used in this analysis: EM2, DM and liver tissue, from hereon, they will be referred as “structure”. We will refer to individual BC network as “network”. We determined the connectivity of the network by analysing the total number of branching points (number of triple junctions) per structure. We determined the length of the network per structure by analysing the total length of all branches in the structure. To compare between structures of different conditions we plotted these values as dot plot where each dot is one structure. In the case of tissue each dot is one field of view. The extracted features from Fiji were exported as .csv files and plotted using Prism. For assembloids, to visualize the structure from different angles, immunofluorescent images were visualized in 3D using MotionTracking (http://motiontracking.mpi-cbg.de) 55 . For this, Gaussian blurring was applied to the channels of interest and then visualized in 3D. For quantification of cholangiocyte and portal fibroblasts in assembloids, custom-made pipelines in Arivis 4D software (Zeiss) was used. Nuclei were segmented based on diameter, probability threshold, and split sensitivity to align with the expected morphology in the fluorescence images. When segmentation was incomplete due to weak fluorescence signals, missing nuclei were manually added. This approach was utilized to determine the number of nuclei per cell and the number of cells per organoid. All segmentation results were manually reviewed and corrected as necessary. Isolation of mRNA and RT-qPCR analysis RNA was extracted from organoid cultures or freshly isolated tissue using the RNeasy Mini RNA Extraction Kit (Qiagen) with DNAse treatment and reverse-transcribed using reverse-transcribed using Moloney Murine Leukemia Virus reverse transcriptase (Promega). All targets were amplified (40 cycles) using gene-specific primers (Key Resource Table) and PowerUp TM SYBR Green master mix (ThermoFisher) or MiIQ syber green (Bio-Rad) and run on a qPCR instrument Thermo Fisher QuantStudio 7 Pro or GeneAmp PCR System 9700; Applied Biosystems respectively. Data were analyzed using Design & Analysis 2.7.0 software (ThermoFisher). Karyotyping Mitotic metaphases for karyotyping were obtained by subculturing hepatocyte organoids in the active growth phase. The following day, cells were exposed to 0.2 μg/mL colcemid (Gibco) for 60 minutes at 37°C to arrest them in metaphase. Organoids were dissociated into single cells using TryplE TM Express (Gibco). After centrifugation and removal of the supernatant, cells were subjected to hypotonic treatment with a solution of 0.075 M KCl for 30 minutes at 37°C, followed by fixation in a 3:1 methanol-acetic acid solution. The preparation was washed three times with the fixative before slide preparation. Chromosomes were stained with Giemsa (Merck) diluted in Gurr buffer (pH 6.8; Gibco). Images were taken with a Zeiss Axio Imager.Z2 upright motorized stand with an ApoTome.2 for improved z-contrast. Functional assays For functional assays, hepatocyte organoids were cultured in expansion and differentiation medium as described above. As negative controls we used cholangiocyte organoids grown as described above. As positive controls we used freshly isolated primary human hepatocytes cultured in standard 2D-hepatocyte monolayer culture or in sandwich culture 82 . Briefly, fresh isolated PHH were plated onto collagen (1.8 mg/mL, RatCol TM collagen, Advanced Biomatrix) coated 24-well plates at 500,000 or 250,000 cells/well in Williams E medium (PAN Biotech), supplemented with 10% FBS, penicillin/streptomycin and 100 nM Dexamethasone for 3 hours for attachment. For the monolayer culture (1d-PHH monolayer control), the cells were cultured on Williams’ E medium supplemented with 1% HEPES + 1% GlutaMax + 1% Penicillin/Streptomycin and 100nM Dexamethasone for 18h (or 24h, for Albumin assay) and then processed for the functional assays. For sandwich culture, fresh isolated PHH were plated onto collagen as above and overlayed with second collagen layer (1.2 mg/mL, RatCol TM collagen, Advanced Biomatrix) and cultured for 7 days in Williams’ E medium supplemented with CM4000 cell maintenance supplement (Thermo Fisher Scientific). To determine albumin secretion, supernatant from 24 hours was collected and the amount of albumin was determined using a human specific Albumin ELISA kit (Assay Pro) following manufacturer’s instructions on an ELISA plate reader (Tecan Spark 20M). To measure Cytochrome P450 activity, on the day of the experiment cholangiocyte and hepatocyte organoids in EM2 or DM were removed from Matrigel using Cell Recovery solution (Corning). Then organoids, 2D-hepatocyte monolayer, or 2D-sandwich cultures were all cultured in Williams’ E medium supplemented with 1% HEPES + 1% GlutaMax + 1% Penicillin/Streptomycin supplemented with the Luciferin-H substrate (100 µM) or Luciferin-IPA (3 µM) for 6 hours. Cytochrome activity was measured using the P450-Glo Assay Kit (Promega) according to manufacturer’s instructions on a plate reader (PerkinElmer Envision). Results were normalized to total viable cell counts per well. Urea synthesis assay To determine the urea secretion, supernatants of cell culture were collected from 48-well plate after 12-hour culture. The concentration of secreted urea was measured by urea assay kit (Abnova, KA1652) according to the manufacturer’s instructions. Measurement of Gluconeogenesis Gluconeogenesis was assessed using a Glucose-Glo™ Assay (Promega, J6021). Organoids/ assembloids were first washed twice with PBS to remove residual glucose and then incubated for 24 hours in glucose-free medium (Gibco, A2494301) to deplete intracellular glucose stores. Subsequently, the organoids were stimulated for 24 hours in gluconeogenesis-inducing medium (glucose-free medium supplemented with 10 mM lactate) to suppress glycolysis and promote hepatic glucose production. After incubation, 25 µL of supernatants from each well was transferred to a 96-well assay plate and mixed with an equal volume of Glucose Detection Reagent. Following a 60-minute incubation at 37°C, luminescence was measured using a luminometer. Cell Counting Hepatocyte organoids were dissociated into single cells using 10X TrypLE (Gibco, A12177-01) after 10 and 15 days of culture in specified media. Cell counts were determined using a Countess® II FL Automated Cell Counter (Thermo Fisher Scientific). Quantification of xenobiotic metabolism by mass spectrometry Hepatocyte organoids were cultured in differentiation medium as previously described. Assembloids were maintained under the same conditions for 6 days. Freshly isolated primary human hepatocytes (PHHs) were cultured in a monolayer for 24 hours, also as described above. Following culture, all cells were washed twice with PBS. The medium was then replaced with 100 μL of Williams’ E medium supplemented with 1% HEPES, 1% GlutaMAX, 1% Penicillin/Streptomycin, and verapamil (Merck; V-002-1ML) at the final concentration of 4 µM. Cells were incubated for 6 hours, after which the supernatant was collected and analyzed by mass spectrometry. Organoids and assembloids were dissociated into single cells using 10× TrypLE and manually counted using a hemocytometer. The resulting cells were washed twice with PBS and stored at –20°C. Metabolites were separately extracted from the supernatant and from the cells by isopropanol: methanol : chlorophorm mixture (4 : 2: 1, v/v/v) containing 7.5 mM ammonium formate (termed MS mix). Supernatant aliquot of 100 μl was 20-fold (v /v) diluted with MS mix, vortexed, centrifuged for 7 min at 13g and the pellet was discarded. Cells suspended in 100 μL PBS were first lysed using ca 25 stainless steel beads of 0.5 mm size (Next Advance, USA, Cat N 152034) in the Qiagen Ratsch Tissue Lyser at 30 Hz for 8 min and metabolites wereextracted as above. Each sample was prepared in three biological replicates and analyzed by mass spectrometry immediately after extraction. Mass spectrometric analysis was performed on a Q Exactive hybrid quadrupole Orbitrap tandem mass spectrometer (ThermoFischerScientific, USA) in positive ion mode by the direct infusion of total extracts. Prior analyses, the internal standard verapamil- 13 C3 hydrochloride (Merk, Germany, V-079-1ML) was dissolved in methanol and spiked into samples to the final concentration of 200 nM. 40 μL aliquots of each sample were then placed on twin.tech PGR Plate 96 (Eppendorf, Germany, Cat N 0030128.648) and infused into the mass spectrometer via TriVersa NanoMate robotic ion source (Advion Interchim Scientific, USA) using nanoflow chips with the nozzle diameter of 4.1 μm. The ion source was controlled by Chipsoft 8.1.0 software. Spraying voltage and gas backpressure were set to 1.25 kV and 0.95 psi, respectively. Ion transfer capillary temperature was set to 200 °C and S-lens RF level to 50%. Target mass resolution R m/z =200 was set to 140 000 (full width at half maximum, FWHM) for both FT MS and FT MS/MS spectra. For acquiring FT MS spectra automated gain control (AGC) was set at the value of 3 × 10 6 ; maximum injection time 500 ms; acquired mass range m/z 50-700; lock masses m/z 445.12003 and m/z 338.34174. The acquisition cycle consisted of recording MS1 spectra for 1.2 min followed by two MS2 spectra for 1.8 min from the precursors with m/z 455.291 (for verapamil; [M+H] + ) and m/z 441.275 (for norverapamil [M+H] + ); precursor m/z isolation width was 3 Da. Spectra were averaged in Xcalibur Qual Browser v.3.0 (ThermoFischerScientific, USA) over 30 sec time range corresponding to stable spray; peaks of metabolites and standard extracted with 5 ppm mass accuracy. Absolute amount of norverapamil was calculated from its molecular ion intensity normalized to the intensity of the standard. For calibrating, aliquots of Williams E medium containing verapamil (Merk, Germany, Cat N V-002-1ML) with the concentration ranging from 2 μM to 8 nM were diluted 20-fold with MS mix, spiked with the internal standard and analyzed as described above. The determined abundance of norverapamil in supernatant and in cellular pellet were summed up, normalized to 10 4 cells and its production rate was expressed in pmols/ h. Xenotransplantation in Fah-/- / Rag2-/- / Il2rg-/- (FRG) Male and female Fah-/-/Rag2-/-/Il2rg-/- (FRG) mice were obtained from Jackson Laboratory. Mice were housed and maintained under specific pathogen-free conditions in accordance with the Principles of Laboratory Animal Care and the Guide set by the HYU Industry-University Cooperation Foundation. For their maintenance, mice were administered ad libitum NTBC (2-(2-nitro-4-trifluoromethylbenzoyl)-1,3-cyclohexanedione) in drinking water. Mice aged 8-16 weeks old from both sexes were kept on NTBC (2-(2-nitro-4-trifluoromethylbenzoyl)-1,3-cyclohexanedione) in drinking water until 3 days prior to the experiment, when NTBC was withdrawn. Human hepatocyte organoids expanded in h-HepOrgs-EM2 and differentiated in h-HepOrgs-DM medium were dissociated into single cells and prepared for injection. For transplantation experiments commercially available frozen PHHs were used (F-PHH2, Supplementary Table 2). Organoids cultured under h-HepOrgs-EM2 medium as well as isolated hepatocytes (PHHs) from the same donors were used as controls. Following dissociation, 500,000 dissociated organoid cells or 800,000 primary human hepatocytes (PHHs) were resuspended in 100 μl of AdDMEM/F-12 medium and injected into the spleen. The non-injected negative control group received 100 μl of PBS instead of cells. Mice were cycled in and out of NTBC treatment for 3 days every time their body weight dropped below 80% of the initial weight. Ingenuity Pathway Analysis We performed Ingenuity Pathway Analysis (IPA, QIAGEN) to identify potential candidate signalling pathways. For that, we first generated three differentially expressed gene (DEG)lists as DEG between liver cancer organoids and liver cancer tissue or healthy tissue (Supplementary Dataset 1_S1 List_ 1 and 2) and DEG list between partial hepatectomy and healthy tissue (List_3). Gene lists were generated as follows: List_1 and 2: gene expression matrices from hepatocellular carcinoma (HCC)-derived organoids, HCC liver tissue, and liver tissue from healthy donors were obtained from the Gene Expression Omnibus (GEO) under accession number GSE84073 45 . Differentially expressed genes (DEGs) were identified using DESeq2 2 , applying a threshold of |log2 fold change| > 1 and an adjusted p-value (adj-p) < 0.1 (Supplementary Dataset 1_S1). List_3: DEGs comparing partial hepatectomy and undamaged liver hepatocytes in mouse were sourced from the supplementary tables in Hu et al. 33 . Additionally, a list of genes mutated in both HCC-derived cell lines was derived from the whole-exome sequencing (WES) results in Broutier et al. 5 . (List_4). The full list of DEG from Lists_1-4 are provided in Supplementary Dataset 1_S1. The three DEG lists and the mutated-gene list (Lists_1-4) were analyzed using Ingenuity Pathway Analysis, employing the Canonical Pathway Analysis and Upstream Regulator Prediction functions 4 . Detailed IPA methodology is provided elsewhere 5 . Briefly, the significance of the association between the dataset and canonical pathways was determined using a right-tailed Fisher's exact test, followed by Benjamini-Hochberg (BH) correction for multiple testing adjustment. For analyses where log fold changes were available, an activity z-score was computed to predict the activation or inhibition likelihood of specific pathways base. The Upstream Regulator Analysis utilized a computational algorithm to identify upstream regulators potentially responsible for the observed gene expression changes. From the IPA Canonical Pathway Analysis, pathways were filtered based on adj-p < 0.05 and the presence of keyword "signalling" in the pathway name (Supplementary Dataset 1_S2). Selected pathways of interest with the mean adj-p value and frequency of pathway significance across comparisons were plotted in Extended Data Fig.1c (Supplementary Dataset 1_S3). Activity z-scores from the selected pathways were individually plotted as well as their corresponding mean values in Figure 1b (Supplementary Dataset 1_S4,5). Next, results from the Upstream Regulator Analysis were filtered for 1) <0.1 adj-p as upstream regulator, and 2) the molecules from the 2 selected signalling pathways (Supplementary Dataset 1_S6). Key components of the signalling pathways and their adj-p value in upstream regulator analysis were plotted in Extended Data Fig.1d (Supplementary dataset 1_S7). Bulk RNA-Seq Library preparation mRNA was isolated from on average 270 ng total RNA by poly-dT enrichment using the NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB) according to the manufacturer’s instructions. Samples were then directly subjected to the workflow for strand-specific RNA-Seq library preparation (Ultra II Directional RNA Library Prep, NEB). For ligation NEB Next Adapter for Illumina of the NEB Next Multiplex Oligos for Illumina Kit were used. After ligation, adapters were depleted by an XP bead purification (Beckman Coulter) adding the beads solution in a ratio of 0.9:1 to the samples. Unique dual indexing was done during the following PCR enrichment (12 cycles) using amplification primers carrying the same sequence for i7 and i5 index (i5: AAT GAT ACG GCG ACC ACC GAG ATC TAC AC NNNNNNNN ACA TCT TTC CCT ACA CGA CGC TCT TCC GAT CT, i7: CAA GCA GAA GAC GGC ATA CGA GAT NNNNNNNN GTG ACT GGA GTT CAG ACG TGT GCT CTT CCG ATC T). After two more XP bead purifications (0.9:1), libraries were quantified using the Fragment Analyzer (Agilent). Libraries were sequenced on an Illumina NovaSeq 6000 in 100 bp paired-end mode to a depth 40 million read pairs per library. RNA-sequencing data processing Raw bulk RNA-seq data was processed using nf-core/rnaseq v3.18.0 (doi: 10.5281/zenodo.1400710) of the nf-core collection of workflows 83 ,utilising reproducible software environments from the Bioconda 84 and Biocontainers 85 projects. The pipeline was executed with Nextflow v24.10.5 86 . The reference genome used was Homo sapiens GRCh38 (Ensembl release 111). The pipeline was run with custom parameters for trimming (extra_trimgalore_args: "--nextseq 20 --length 15"), alignment (extra_star_align_args: "--outFilterMismatchNmax 999 --outFilterMismatchNoverLmax 0.1 --alignMatesGapMax 200000 --chimSegmentMin 20 --twopassMode Basic --alignIntronMin 20 --alignIntronMax 200000"), and quantification (extra_salmon_quant_args: "--seqBias --gcBias --posBias"). The resulting MultiQC report was inspected to ensure overall sequencing quality and pipeline performance. Transcript-level abundance estimates were imported using the tximeta package 87 to generate a gene-level count matrix. Next, variance stabilizing transformation (VST) from DESeq2 88,89 to normalize the data. Euclidean distance matrices, principal component analysis (PCA), and heatmap visualizations were computed on the VST-transformed values. On some heatmaps, a min-max scaling was applied. In Extended Data Fig. 2a, b, batch correction was performed on the VST-transformed values using limma ’s removeBatchEffect, with sample material type (tissue vs. organoid) treated as the batch variable 90 . For differential expression analysis, DESeq2 was used. For the comparison between MM+WtnS+TRULI and Primary (fresh isolated PHHs), the design formula ~ donor + condition_l3 was applied (Extended Data Fig. 2). Log-fold changes were shrunken using lfcShrink with the ashr method (type="ashr"), applying a fold-change threshold of 1.5 and a significance threshold of α = 0.05 91 . For the comparison between DM and EM2 (Fig. 2e), the design formula ~ ~ batch + donor + condition_l1 was applied. Log-fold changes were shrunken using lfcShrink with the ashr method (type="ashr"), applying a fold-change threshold of 1.5 and a significance threshold of α = 0.05. For the comparison between h-Hep and PF (Extended Data Fig. 6h), the design formula ~ sex + cell_type was applied. Log-fold changes were shrunken using lfcShrink with the ashr method (type="ashr"), applying a fold-change threshold of 4 and a significance threshold of α = 0.05. Gene set enrichment analysis (GSEA) was conducted using the clusterProfiler package, leveraging gseKEGG, gseGO, and gsePathway for pathway enrichment analysis 92 . The zonated gene list (Extended Data Fig. 3h) was obtained by manually curating the genes that are confirmed to be portally or centrally zonated from human spatial transcriptomic datasets from 13,14,58,59 (full list is provided in Supplementary Dataset 2_S6). We then intersected this refined zonated gene lists with our list of differentially expressed genes in the DM vs. EM2 comparison. Donor-specific genes were identified separately for batches Y1/Y2 and S1/S2 using a likelihood ratio test (LRT) with the full model ~ donor and the reduced model ~ 1. Genes with an adjusted P-value (padj) < 0.05 were retained, and the resulting gene lists from the two batches were merged. Pairwise correlations between organoids and primary cells were computed using the donor-specific genes. For the heatmap shown in Extended Data Fig. 4e, sex-specific genes were excluded. The complete software stack for downstream analysis is available as a Docker container (rnaseq-notebook:2025-04-21) archived at https://quay.io/repository/fbnrst/rnaseq-notebook. Single-cell Transcriptomics with 10x Genomics For single cell RNAseq analysis assembloids were generated by assembling h-HepOrgs, cholangiocytes/ductal cells derived from cholangiocytes organoids (nuclear-GFP) and portal fibroblasts (PFs, nuclear-RFP) at a ratio 1 h-HepOrgs: 25 MSC : 100 Cholangiocytes. At 5-6 days after aggregation, assembloids were collected as follows: periportal assembloids were dissociated to single cells using TrypLE 10x for 5min at 37°C. The cells were resuspended in DM and 10μg/mL DNAse in BSA-coated tubes and filtered through a 100 μm strainer. Cell suspensions (30000 – 50000 cells) were concentrated by centrifugation (50 rcf, 5 min, 4 °C) and the volume reduced to ~55 µl. Cells were carefully resuspended and visually inspected under a light microscope to determine cell concentration and quality. The concentrations of the single-cell suspensions were adjusted to 138-912 cells per microliter and carefully mixed with the reverse transcription mix before loading cells on the 10X Genomics Chromium system 93 in a Chromium Single Cell G Chip targeting 3000-10,000 cells per reaction. Following the guidelines of the 10x Genomics Chromium Single Cell Kit v3.1 user manual, the droplets were directly subjected to reverse transcription, the emulsion was broken and cDNA was purified using Dynabeads MyOne Silane (10X Genomics). cDNA was first amplified with 12 cycles, and then purified with 0.6x SPRIselect beads (Beckman Coulter) to enrich cDNA fragments (>400 bp). A quality and quantity control of cDNA on the Fragment Analyzer (using the DNF-473 NGS Fragment Kit, Agilent) was eventually performed to obtain its concentration. The 10X Genomics single cell RNA-seq library preparation - involving fragmentation, dA-Tailing, adapter ligation and 11 or 12 cycles indexing PCR – was performed based on the manufacturer’s protocol. After quantification, the libraries were sequenced on an Illumina Novaseq6000 in paired-end mode (R1/R2: 100 cycles; I1/I2: 10 cycles), generating 230-370 million fragment pairs. The raw sequencing data was then processed with the ‘count’ command of the Cell Ranger software (v8.0.1) provided by 10X Genomics with the option ‘--expect-cells’ set to 10,000 (all other options were used as per default). To build the reference for Cell Ranger, human genome (GRChg38) as well as gene annotation (Ensembl 104) were downloaded from Ensembl. Genome and annotation were processed following the build steps provided by 10x (https://support.10xgenomics.com/single-cell-gene-expression/software/release-notes/build#mm10_2020A) to build the appropriate Cellranger reference. scRNAseq Data Analysis The raw scRNAseq data was processed using nf-core/scrnaseq v3.0.0 (doi: 10.5281/zenodo.3568187) of the nf-core collection of workflows 83 , utilising reproducible software environments from the Bioconda 84 and Biocontainers 85 projects. The pipeline was executed with Nextflow v24.10.5 86 STARSOLO was used as the aligner. The reference genome was set to Homo sapiens GRCh38 (Ensembl release 111) with custom additions for red fluorescent protein (RFP) and green fluorescent protein (GFP) transgenes, obtained from SnapGene (DsRed1 and EGFP, respectively). Outputs were inspected for quality control (QC), and one sample with poor QC was excluded from further analysis. Within nf-core/scrnaseq, technical artefacts were eliminated using CellBender 94 . CellBender output was used for data visualization. Doublet detected per sample was performed using scrublet 95 . Further analysis was perfomed using scanpy 96 . QC was applied with the following thresholds: minimum total counts of 5000, minimum detected genes of 2000, a maximum percentage of counts in the top 50 genes set at 50%, maximum percentage of mitochondrial counts at 15%, and a maximum doublet score of 0.15. Gene filtering was performed to retain genes expressed in at least 10 cells. After filtering, the data underwent normalization, log transformation, and identification of top 3000 highly variable genes. PCA was performed, and batch correction was implemented through harmony integration 97 . UMAP visualization and Leiden clustering were used to identify the three expected cell types 98,99 . To compare homeostatic-like and fibrotic-like organoids, pseudobulk aggregation was performed using decoupler for each cell type 100 . Pseudobulk data was generated by summing raw counts for each sample and cell type, with a minimum requirement of 10 cells per group and 1000 total counts. Differential expression analysis was conducted using pyDESeq2 101 . For each cell type, DESeq2 datasets were created with design factors that included ‘donor’ and ‘condition’, using the ‘homeostatic-like’ condition as the reference. Differentially expressed genes between the homeostatic-like and fibrotic-like conditions were ranked based on the test statistic. Subsequently, gene set enrichment analysis (GSEA) was performed on the ranked lists using clusterProfiler , focusing on KEGG, Reactome, and GO terms. The complete software stack for downstream analysis is available as a Docker container (singlecell-notebook:2025-04-21) archived at https://quay.io/repository/fbnrst/singlecell-notebook. Comparison to Public Datasets Data from Andrews et al. (2022, 2024) and Guilliams et al. (2022) were downloaded in h5ad format from https://cellxgene.cziscience.com/. Additionally, data from Brazovskaja (2024) was obtained from https://data.mendeley.com/datasets/yp3txzw64c/1, and the dataset from Ramachandran et al., 2019 10 was downloaded from https://datashare.ed.ac.uk/bitstream/handle/10283/3433/tissue.rdata?sequence=3&isAllowed=y and converted to h5ad format using the sceasy package. These public datasets were merged with the raw count matrix of our QC-filtered organoid data. Subsequently, the combined dataset underwent normalization, followed by log transformation and detection of the top 4000 highly variable genes. We performed principal component analysis (PCA) and integrated the dataset using Harmony, specifying the concatenation of the paper and donor as batch variables, with a maximum of 20 iterations and a theta value of 1.5. Selected genes were visualized in a dot plot (Fig. 4h). Pseudobulk analyses were then conducted using the decoupler package to summarize gene expression by cell type. This involved generating a pseudobulk dataset where raw counts were summed by sample and cell type, ensuring a minimum of 30 cells per group. Following pseudobulk aggregation, the data was normalized and log-transformed, with the top highly variable genes identified based on mean expression and dispersion. Additionally, the ‘paper’ variable was regressed out to mitigate batch effects. Next, PCA was performed on the pseudobulk data, using 50 principal components for subsequent analyses. Hierarchical clustering was executed using the Pearson correlation metric, and Pearson correlation matrices were plotted, as shown in Fig. 4g, 5f. Marker genes for the 3 major cell types were computed separately for our organoid data and the merged public data. using scanpy’s rank_genes_groups function. For each dataset, the top 300 marker genes for each cell type were selected. Subsequently, gene set enrichment analysis was performed using the gseapy package, leveraging the Enrichr method 102 . The analysis focused on the KEGG 2021 Human and Reactome 2022 gene sets, with a p-value cutoff of 0.05. Shared enriched pathways between the organoid and tissue datasets were identified, and the combined enrichment scores for selected terms were plotted (Extended Data Fig. 8c, d). Data statistical analysis All values are represented as mean ± standard error of the mean (S.E.M.) as specified in legend. Either Two-tailed Mann-Whitney non-parametric or two-tailed ANOVA tests were used. p<0.05 was considered statistically significant. In all cases data from at least 3 independent experiments was used. Calculations were performed using the Prism 9 software package. All p -values are given in the corresponding figure legends. Dispersion and precision measures (e.g., mean, median, SD, SEM) are specified in the figure legends. All the scRNAseq statistics are described above in the corresponding section. Data availability The scRNAseq and bulk RNAseq datasets generated during this study is available in the European Genome-phenome Archive (EGA) (EGAS50000000994). The full lists of all differentially expressed genes, GSEA terms and marker genes are available in Supplementary Datasets 2-4. The raw data used for mass spectrometry is available at the following repository: https://doi.org/10.17617/3.Z9GMJE Code availability The source code used for bile canaliculi segmentation is available at the following repository: https://git.mpi-cbg.de/huch_lab/assembloid-paper The source code for bulkRNAseq and scRNAseq data analysis are available upon request and will be made public upon publication (https://git.mpi-cbg.de/huch_lab/fr_human_assembloid). Additional references 74 Cordero-Espinoza, L. & Huch, M. The balancing act of the liver: tissue regeneration versus fibrosis. J Clin Invest 128 , 85-96 (2018). https://doi.org:10.1172/JCI93562 75 Mederacke, I. et al. Fate tracing reveals hepatic stellate cells as dominant contributors to liver fibrosis independent of its aetiology. Nat Commun 4 , 2823 (2013). https://doi.org:10.1038/ncomms3823 76 Pinzani, M. & Luong, T. V. Pathogenesis of biliary fibrosis. Biochim Biophys Acta Mol Basis Dis 1864 , 1279-1283 (2018). https://doi.org:10.1016/j.bbadis.2017.07.026 77 Desmet, V. J. & Roskams, T. Cirrhosis reversal: a duel between dogma and myth. J Hepatol 40 , 860-867 (2004). https://doi.org:10.1016/j.jhep.2004.03.007 78 Lee, K. P. et al. The Hippo-Salvador pathway restrains hepatic oval cell proliferation, liver size, and liver tumorigenesis. Proc Natl Acad Sci U S A 107 , 8248-8253 (2010). https://doi.org:10.1073/pnas.0912203107 79 Lu, L. et al. Hippo signaling is a potent in vivo growth and tumor suppressor pathway in the mammalian liver. Proc Natl Acad Sci U S A 107 , 1437-1442 (2010). https://doi.org:10.1073/pnas.0911427107 80 Ke, M. T., Fujimoto, S. & Imai, T. SeeDB: a simple and morphology-preserving optical clearing agent for neuronal circuit reconstruction. Nat Neurosci 16 , 1154-1161 (2013). https://doi.org:10.1038/nn.3447 81 Uhlen, M. et al. Proteomics. Tissue-based map of the human proteome. Science 347 , 1260419 (2015). https://doi.org:10.1126/science.1260419 82 Zeigerer, A. et al. Rab5 is necessary for the biogenesis of the endolysosomal system in vivo. Nature 485 , 465-470 (2012). https://doi.org:10.1038/nature11133 83 Ewels, P. A. et al. The nf-core framework for community-curated bioinformatics pipelines. Nat Biotechnol 38 , 276-278 (2020). https://doi.org:10.1038/s41587-020-0439-x 84 Gruning, B. et al. Bioconda: sustainable and comprehensive software distribution for the life sciences. Nat Methods 15 , 475-476 (2018). https://doi.org:10.1038/s41592-018-0046-7 85 da Veiga Leprevost, F. et al. BioContainers: an open-source and community-driven framework for software standardization. Bioinformatics 33 , 2580-2582 (2017). https://doi.org:10.1093/bioinformatics/btx192 86 Di Tommaso, P. et al. Nextflow enables reproducible computational workflows. Nat Biotechnol 35 , 316-319 (2017). https://doi.org:10.1038/nbt.3820 87 Love, M. I. et al. Tximeta: Reference sequence checksums for provenance identification in RNA-seq. PLoS Comput Biol 16 , e1007664 (2020). https://doi.org:10.1371/journal.pcbi.1007664 88 Huber, W., von Heydebreck, A., Sultmann, H., Poustka, A. & Vingron, M. Variance stabilization applied to microarray data calibration and to the quantification of differential expression. Bioinformatics 18 Suppl 1 , S96-104 (2002). https://doi.org:10.1093/bioinformatics/18.suppl_1.s96 89 Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15 , 550 (2014). https://doi.org:10.1186/s13059-014-0550-8 90 Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43 , e47 (2015). https://doi.org:10.1093/nar/gkv007 91 Stephens, M. False discovery rates: a new deal. Biostatistics 18 , 275-294 (2017). https://doi.org:10.1093/biostatistics/kxw041 92 Yu, G., Wang, L. G., Han, Y. & He, Q. Y. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 16 , 284-287 (2012). https://doi.org:10.1089/omi.2011.0118 93 Zheng, G. X. et al. Massively parallel digital transcriptional profiling of single cells. Nat Commun 8 , 14049 (2017). https://doi.org:10.1038/ncomms14049 94 CellBender removes technical artifacts from single-cell RNA sequencing data. Nat Methods 20 , 1285-1286 (2023). https://doi.org:10.1038/s41592-023-01946-4 95 Wolock, S. L., Lopez, R. & Klein, A. M. Scrublet: Computational Identification of Cell Doublets in Single-Cell Transcriptomic Data. Cell Syst 8 , 281-291 e289 (2019). https://doi.org:10.1016/j.cels.2018.11.005 96 Wolf, F. A., Angerer, P. & Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. Genome Biol 19 , 15 (2018). https://doi.org:10.1186/s13059-017-1382-0 97 Korsunsky, I. et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods 16 , 1289-1296 (2019). https://doi.org:10.1038/s41592-019-0619-0 98 Becht, E. et al. Dimensionality reduction for visualizing single-cell data using UMAP. Nat Biotechnol (2018). https://doi.org:10.1038/nbt.4314 99 Traag, V. A., Waltman, L. & van Eck, N. J. From Louvain to Leiden: guaranteeing well-connected communities. Sci Rep 9 , 5233 (2019). https://doi.org:10.1038/s41598-019-41695-z 100 Badia, I. M. P. et al. decoupleR: ensemble of computational methods to infer biological activities from omics data. Bioinform Adv 2 , vbac016 (2022). https://doi.org:10.1093/bioadv/vbac016 101 Muzellec, B., Telenczuk, M., Cabeli, V. & Andreux, M. PyDESeq2: a python package for bulk RNA-seq differential expression analysis. Bioinformatics 39 (2023). https://doi.org:10.1093/bioinformatics/btad547 102 Fang, Z., Liu, X. & Peltz, G. GSEApy: a comprehensive package for performing gene set enrichment analysis in Python. Bioinformatics 39 (2023). https://doi.org:10.1093/bioinformatics/btac757 103 Tirosh, I. et al. Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq. Science 352 , 189-196 (2016). https://doi.org:10.1126/science.aad0501 104 Wang, Y. et al. Comprehensive Molecular Characterization of the Hippo Signaling Pathway in Cancer. Cell Rep 25 , 1304-1317 e1305 (2018). https://doi.org:10.1016/j.celrep.2018.10.001 105 Boj, S. F. et al. Diabetes risk gene and Wnt effector Tcf7l2/TCF4 controls hepatic response to perinatal and adult metabolic demand. Cell 151 , 1595-1607 (2012). https://doi.org:10.1016/j.cell.2012.10.053 Additional Declarations Yes there is potential Competing Interest. M.H. is an inventor in several patents on organoid technology. Y.K., S.D. and M.H. are inventors in a patent on human hepatocyte organoids. M.H. L.Y, S.D., A.M.D and A.S. are co-inventors in a patent on human assembloids. The remaining authors declare no competing interests. Supplementary Files 7SupplementaryInformationGuideEditedLei.docx Supplementary Info Guide SupplementaryDataset3.xlsx Supplementary DAtaset3 SupplementaryTable1isolationrevisioncompleted.xlsx Supplementary Table 1 SupplementaryDataset5reagents.xlsx Supplementary Dataset 5 SupplementaryDataset1v3DT.xlsx Supplementary Dataset 1 Supplementarytable4patienttopatientcopy.xlsx Supplementary Table 4 ExtFigure2rev2.pdf Extended Data Fig 2 SupplementaryTable2donorscomplete.xlsx Supplementary Table 2 ExtFigure10rev2.pdf Extended Data Fig 10 ExtFigure6rev2.pdf Extended Data Fig 6 ExtFigure4rev2.pdf Extended Data Fig 4 SupplementaryTable3Comparisonwithpreviousresearchofgenerationofhepatocytesorganoidsvsourwork.docx Supplementary Table 3 ExtFigure7rev2.pdf Extended Data Fig 7 ExtFigure9rev2.pdf Extended Data Fig 9 ExtFigure1rev2.pdf Extended Data Fig 1 SupplementaryDataset2.xlsx Supplementary Dataset 2 SupplementaryDataset4fibrosisMH2.xlsx Supplementarty Dataset 4 ExtFigure3rev2.pdf Extended Data Fig 3 ExtFigure8rev2.pdf Extended Data Fig 8 ExtFigure5rev2.pdf Extended Data Fig 5 ExtendedDataFigureLegends.docx Cite Share Download PDF Status: Published Journal Publication published 17 Dec, 2025 Read the published version in Nature → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5314788","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Biological Sciences - Article","associatedPublications":[],"authors":[{"id":513899139,"identity":"5b6e1cc2-4946-45d1-960f-3abd0f176cce","order_by":0,"name":"Meritxell Huch","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAsUlEQVRIiWNgGAWjYBACAwYGNoYPDAw8YARkE6eFcQbJWph5wExitZhLJD97bJtzR4aB/ezhFwxlNoS1WM5IMzfO3faMh4EnL82C4VwaEQ67kcMmnbvtMI/9DR4zA8a2w0RqsQRqYZAAa/lPpBZGiBbjB4xtB4jQcuaZmWQv1C8MCeeSidByPPmZxM9td+xBIfbhQ5kdYS1QAHYPm0QC0RqgWpg/kKBjFIyCUTAKRhAAAHZ5NEOGWP5PAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-1545-5265","institution":"Max Planck of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":true,"prefix":"","firstName":"Meritxell","middleName":"","lastName":"Huch","suffix":""},{"id":513899140,"identity":"168a70cf-610f-4897-abb1-ab0652ea309c","order_by":1,"name":"Lei Yuan","email":"","orcid":"https://orcid.org/0009-0006-1378-7000","institution":"Max Planck Institute of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Yuan","suffix":""},{"id":513899141,"identity":"9ff8a8fa-df85-4718-98fa-bfc706b4f659","order_by":2,"name":"Anke Liebert","email":"","orcid":"https://orcid.org/0000-0002-5849-6147","institution":"Max Planck of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":false,"prefix":"","firstName":"Anke","middleName":"","lastName":"Liebert","suffix":""},{"id":513899142,"identity":"1980811b-49b2-4d4b-a727-3e73e0043bca","order_by":3,"name":"Sagarika Dawka","email":"","orcid":"","institution":"Max Planck of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":false,"prefix":"","firstName":"Sagarika","middleName":"","lastName":"Dawka","suffix":""},{"id":513899143,"identity":"2c454ce4-5164-4aad-9a9f-4722e5070bea","order_by":4,"name":"Robert Arnes-Benito","email":"","orcid":"","institution":"Max Planck of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":false,"prefix":"","firstName":"Robert","middleName":"","lastName":"Arnes-Benito","suffix":""},{"id":513899144,"identity":"c491ce5d-9231-4bce-b255-a8341a7f9706","order_by":5,"name":"Fabian Rost","email":"","orcid":"https://orcid.org/0000-0001-6466-2589","institution":"Max Planck Institute for Cell Biology and Genetics","correspondingAuthor":false,"prefix":"","firstName":"Fabian","middleName":"","lastName":"Rost","suffix":""},{"id":513899145,"identity":"03a84d0c-b099-4e84-a459-79885f46416d","order_by":6,"name":"David Tsang","email":"","orcid":"https://orcid.org/0000-0002-1594-2987","institution":"Max Planck of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Tsang","suffix":""},{"id":513899146,"identity":"617506e3-19d0-4231-a641-136699699a30","order_by":7,"name":"Roberta Rezende de Castro","email":"","orcid":"","institution":"Max Planck of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":false,"prefix":"","firstName":"Roberta","middleName":"Rezende","lastName":"de Castro","suffix":""},{"id":513899147,"identity":"41d4a6f9-d175-4ade-8a56-0d9c1d70ea91","order_by":8,"name":"Daniel Stange","email":"","orcid":"https://orcid.org/0000-0003-4246-2230","institution":"University Hospital Carl Gustav Carus, Technische Universität Dresden","correspondingAuthor":false,"prefix":"","firstName":"Daniel","middleName":"","lastName":"Stange","suffix":""},{"id":513899148,"identity":"f348b61e-2446-4f17-8ae0-18af520b6c98","order_by":9,"name":"Franziska Baenke","email":"","orcid":"","institution":"University Hospital Carl Gustav Carus, Technische Universität Dresden, Fetscherstraße 74, Dresden, 01307, Germany.","correspondingAuthor":false,"prefix":"","firstName":"Franziska","middleName":"","lastName":"Baenke","suffix":""},{"id":513899149,"identity":"ff33eeb1-79b9-4795-aa57-0fc1fababf1e","order_by":10,"name":"Yohan Kim","email":"","orcid":"https://orcid.org/0000-0002-4009-1694","institution":"Max Planck of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":false,"prefix":"","firstName":"Yohan","middleName":"","lastName":"Kim","suffix":""},{"id":513899150,"identity":"c485dc40-a1e0-4471-9b9d-b6a8ed00854e","order_by":11,"name":"Andrea Knaust","email":"","orcid":"","institution":"MPI of Molecular Cell Biology and Genetis","correspondingAuthor":false,"prefix":"","firstName":"Andrea","middleName":"","lastName":"Knaust","suffix":""},{"id":513899151,"identity":"31e45566-e2f9-4aad-b5c0-b75359f7dab4","order_by":12,"name":"Anna Shevchenko","email":"","orcid":"https://orcid.org/0009-0008-6157-4442","institution":"Max Planck Institute of Molecular Cell Biology and Genetics","correspondingAuthor":false,"prefix":"","firstName":"Anna","middleName":"","lastName":"Shevchenko","suffix":""},{"id":513899152,"identity":"c78c7b25-2968-4157-b51c-5c11bbfd0c32","order_by":13,"name":"Andrej Shevchenko","email":"","orcid":"https://orcid.org/0000-0002-5079-1109","institution":"Max Planck Institute of Molecular Cell Biology and Genetics","correspondingAuthor":false,"prefix":"","firstName":"Andrej","middleName":"","lastName":"Shevchenko","suffix":""},{"id":513899153,"identity":"c8d9236c-faf3-4b07-b9dc-9f53623806d1","order_by":14,"name":"Aleksandra Sljukic","email":"","orcid":"","institution":"Max Planck of Molecular Cell Biology and Genetics (MPI-CBG)","correspondingAuthor":false,"prefix":"","firstName":"Aleksandra","middleName":"","lastName":"Sljukic","suffix":""},{"id":513899154,"identity":"9346d118-eab4-4238-8c57-45e31861f54c","order_by":15,"name":"Anna Dowbaj","email":"","orcid":"https://orcid.org/0000-0002-6278-8011","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Anna","middleName":"","lastName":"Dowbaj","suffix":""},{"id":513899155,"identity":"148e99f8-5595-4a74-bc03-a84ccf4f9de9","order_by":16,"name":"Christina Goetz","email":"","orcid":"https://orcid.org/0000-0002-8060-1599","institution":"Department of Hepatobiliary Surgery and Visceral Transplantation, Clinic for Visceral, Transplant, Thoracic and Vascular Surgery, Leipzig University Medical Center,","correspondingAuthor":false,"prefix":"","firstName":"Christina","middleName":"","lastName":"Goetz","suffix":""},{"id":513899156,"identity":"7beb7093-4e1c-4189-90a7-9a0d9d56faee","order_by":17,"name":"Georg Damm","email":"","orcid":"https://orcid.org/0000-0002-2104-8076","institution":"Leipzig University","correspondingAuthor":false,"prefix":"","firstName":"Georg","middleName":"","lastName":"Damm","suffix":""},{"id":513899157,"identity":"efb9a67e-3790-44c0-bb19-74f3c75914ce","order_by":18,"name":"Daniel Seehofer","email":"","orcid":"","institution":"University of Leipzig","correspondingAuthor":false,"prefix":"","firstName":"Daniel","middleName":"","lastName":"Seehofer","suffix":""},{"id":513899158,"identity":"5a5aa5d8-e877-4c74-8659-b8e9f29311b9","order_by":19,"name":"Seunghee Kim","email":"","orcid":"","institution":"Department of Surgery, Hanyang University College of Medicine,","correspondingAuthor":false,"prefix":"","firstName":"Seunghee","middleName":"","lastName":"Kim","suffix":""},{"id":513899159,"identity":"f2ab2b38-cbb5-4207-a091-8a1fcf0d9e9b","order_by":20,"name":"Dongho Choi","email":"","orcid":"","institution":"Hanyang University","correspondingAuthor":false,"prefix":"","firstName":"Dongho","middleName":"","lastName":"Choi","suffix":""}],"badges":[],"createdAt":"2024-10-23 01:25:48","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5314788/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5314788/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41586-025-09884-1","type":"published","date":"2025-12-17T05:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":91526559,"identity":"6102edfe-7239-4f1e-9ae3-cfc1f2fe7848","added_by":"auto","created_at":"2025-09-17 11:09:58","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":998171,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePrimary human hepatocytes expand long-term when grown as hepatocyte organoids\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea-f\u003c/strong\u003e, Liver tissues were obtained from patients undergoing surgery and processed for cell isolation as described in methods. Isolated primary human hepatocytes (PHHs) were used to generate human hepatocyte organoids (h-HepOrgs) that would self-renew \u003cem\u003ein vitro\u003c/em\u003e and expand long-term.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e, Schematics depicting the protocol for generating human hepatocyte organoids (h-HepOrgs). See methods for details.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e, Ingenuity Pathway Analysis (IPA) of several publicly available datasets (Lists 1-3) used to identify signalling pathways involved in hepatocyte proliferation. Bar plots show IPA pathway activity z-score for each selected pathway. Dots represent the different datasets. See Supplementary Dataset 1_S5 for details.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRepresentative brightfield images of primary h-HepOrgs cultured in the indicated media (see methods) at day 10 of culture (P0). Scale bar, 500 μm (top); magnification, 100 μm (bottom).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed\u003c/strong\u003e, Representative brightfield images of patient-derived primary h-HepOrgs serially expanded and cultured long-term in h-HepOrgs-EM2. Scale bar, 100 μm (left), 1 mm (right). P, passage; d, day.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee\u003c/strong\u003e, Organoid formation efficiency of h-HepOrgs cultured with the indicated media. Graph represents ± SEM from n = 4-5 donors (biological replicates). Dot colour, same donor. Two-way ANOVA with Tukey Test for multiple comparisons.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef\u003c/strong\u003e, Serial expansion (1:2) of h-HepOrgs from indicated donors. Graph indicates the expansion potential of h-HepOrgs at the indicated media. Dot, passage. Note that for these donors, we check their expandability beyond 10 passages. As detailed in the graph and in Supplementary Table 2, under EM1 the cultures exhibit lower expansion potential, with none of them reaching beyond passage 10. Also, note that for the donors expanded in EM2 that reached passage 10, we stopped culturing them at time of submission (PHH29, DSD40, PHH27).\u003c/p\u003e","description":"","filename":"fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/1b18edc14aeb47e0e645721f.png"},{"id":91526556,"identity":"da893294-097d-4f3e-a819-93cc0b0ec0ac","added_by":"auto","created_at":"2025-09-17 11:09:58","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1091166,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGene expression of human hepatocyte organoids resembles \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003ein vivo\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e hepatocytes.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea-c\u003c/strong\u003e, Human hepatocyte organoids (h-HepOrg) expanded in EM2 or differentiated in DM or liver tissue were analysed for the expression of the indicated markers. Representative images from at least n=2 independent donors from n=3 independent experiments are shown.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eImmunofluorescent staining for the proliferation marker Ki-67 (magenta) and nuclei (DAPI, cyan) in h-HepOrgs-EM2 (top) and h-HepOrgs-DM (bottom). Scale bar, 50 μm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eImmunofluorescent staining for nuclear YAP (magenta), nuclei (DAPI, cyan) and cell borders (F-actin, grey) in liver tissue (top), h-HepOrgs-EM2 (with TRULI) (middle) and h-HepOrgs in no TRULI (bottom). Scale bars, 50 μm. White arrowheads indicate cells positive for nuclear YAP staining while yellow arrowheads indicate nuclear YAP-negative cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec\u003c/strong\u003e, Immunofluorescent staining for the hepatocyte marker HNF4A (magenta), bile canaliculi (BC) marker CD13 (green) and F-actin (grey) in organoids in EM2 (left) or DM (right). Note that in DM h-HepOrgs exhibit better BC architecture; thinner and more interconnected. Scale bars, 50 μm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed-e\u003c/strong\u003e, RNAseq analysis of h-HepOrgs cultured in EM2 or DM (h-HepOrgs-DM), cholangiocyte organoids (h-CholOrg) and primary human hepatocytes (PHHs) either freshly isolated (primary, PHHs) or cultured for 1-day (1d-PHH monolayer).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed\u003c/strong\u003e, Heatmap representing the scaled expression of the indicated genes for known hepatocyte and cholangiocyte (Chol.) markers. Their region-specific expression (pericentral or periportal) or function is indicated in different colours.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee\u003c/strong\u003e, GSEA of h-HepOrgs grown in DM compared to EM2. The full list is presented in Supplementary Dataset_2. The results are presented as a dot plot where the dot colour denotes the adjusted p-value. NES, normalized enrichment score.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef\u003c/strong\u003e, Human liver tissue (left) and h-HepOrgs in DM medium (right) stained for pericentrally (glutamine synthetase, GS, magenta, top) and periportally (histidine ammonia-lyase, HAL, yellow, bottom) zonated liver markers. Nuclei were stained with DAPI (cyan). Fluorescence intensities for GS and HAL are additionally indicated in Fire LUT to better distinguish regional differences (first column in left and right panels). CV, central vein. PV, portal vein (PV). Scale bars, 100 μm (tissue) and 50 μm (organoid). Representative images from n= 3 independent experiments.\u003c/p\u003e","description":"","filename":"fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/143f48e03023a1c6139755f2.png"},{"id":91526557,"identity":"8e38e499-3847-4caa-83ae-a79bd9923dbd","added_by":"auto","created_at":"2025-09-17 11:09:58","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":763556,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHuman hepatocyte organoids mimic \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003ein vivo\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e bile canaliculi structure, capture patient-specific features and retain drug metabolizing function\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea-c\u003c/strong\u003e, h-HepOrgs cultured in DM were analyzed for the presence of apico-basal polarity (\u003cstrong\u003ea\u003c/strong\u003e) and bile canaliculi (BC) network (\u003cstrong\u003eb-c\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e, Left, illustration representing apico-basal polarity and their markers in h-HepOrgs. Right, images showing immunofluorescent staining for the basolateral marker E-cadherin (magenta) and the apical markers Radixin (green, liver tissue, top) and ZO-1 (green, h-HepOrgs-DM, bottom). Nuclei were stained with DAPI (cyan). Scale bar, 25 μm (left); magnification, 10 μm (right). Representative images are shown.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eImmunofluorescence staining for bile canaliculi (CD13, green) and cell borders (F-actin, grey) in h-HepOrgs-EM2 (top), h-HepOrgs-DM (middle) and human liver tissue (bottom). Nuclei were stained with DAPI (cyan). Scale bar, 50 μm. Right panel shows segmented bile canaliculi (BC) in 3D, depicting the local thickness of BC in Fire LUT (blue – thinner, red – thicker). Representative images from at least n=3 independent donors from n= 3 independent experiments are shown.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec\u003c/strong\u003e, Graph showing the total number of triple junctions as a proxy for connectivity within the bile canaliculi (BC) network. For tissue, dot represents one field of view and colour a different donor (n=3). For organoids, dot represents one structure (organoid) in the indicated donors. EM2, dark purple. DM, light purple. Tissue, grey. Bars, mean ± SEM. Statistical comparison was performed between tissue and DM and DM \u003cem\u003evs \u003c/em\u003eEM2. ns, not significant. One-tailed Mann-Whitney test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed\u003c/strong\u003e, Hepatocyte organoids retain donor-specific transcriptional signatures that recapitulate inter-donor variability observed in primary hepatocytes. Heatmap showing scaled expression of donor-specific genes computed across primary human hepatocyte samples (PHH, purple) and hepatocyte organoids in DM (h-HepOrg–DM, yellow). Columns, individual donors. Row, donor-specific gene identified by differential expression analysis. Hierarchical clustering was performed on both samples and genes. Specific genes and their functions are annotated with colour. Details are found in Supplementary Table 4.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee-f\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eHuman hepatocyte organoids expanded in EM2 (dark purple) and differentiated in DM (light purple) were tested for cytochrome activity (\u003cstrong\u003ee\u003c/strong\u003e) and albumin secretion (\u003cstrong\u003ef\u003c/strong\u003e) and compared to human liver cholangiocyte organoids (h-CholOrg, green) and fresh isolated primary human hepatocytes (PHH) cultured as standard 2D for 1 (light pink) or 7 (orange) days. Graphs represent mean +/- SEM for n=4-7 donors from n= 3 independent experiments. Results are expressed as RLU (\u003cstrong\u003ee\u003c/strong\u003e) or ng/ml (\u003cstrong\u003ef\u003c/strong\u003e) normalized by the total cell count. Two-way ANOVA with Tukey test for multiple comparisons.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg.\u003c/strong\u003e The production rate of norverapamil reflects the xenobiotic metabolism capacity of Hepatocytes in HepOrgs. The rate (in pmol/h per 10^3 cells) was determined by mass spectrometry for h-HepOrgs and primary human hepatocytes (PHHs) from 3 donors. In h-CholOrg (negative control), the production rate was below the detection limit. Bars represent mean values +/-SD (n = 3). Unpaired two-tailed Student’s t-test with Welch’s correction, (n = 3 donors per group).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eh\u003c/strong\u003e, \u003cem\u003eFah-/-/Rag2-/-/Il2rg-/-\u003c/em\u003e (FRG) mice were injected with 500,000 cells intrasplenically and cycled on NTBC treatment post transplantation as described in methods. Kaplan Meier survival curve shows that undifferentiated and differentiated h-HepOrgs rescue the survival of Fah mutant mice after NTBC withdrawal. Log-rank test \u003cem\u003ep \u003c/em\u003e= 0.0127.\u003c/p\u003e","description":"","filename":"fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/1cdadc4ee49a660c541a8647.png"},{"id":91524634,"identity":"2af6df7c-eff1-4c33-858a-83de3398f944","added_by":"auto","created_at":"2025-09-17 11:01:58","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":698471,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHuman\u003c/strong\u003e \u003cstrong\u003eperiportal assembloids recapitulate \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003ein vivo\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e liver periportal tissue.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea-f,\u003c/strong\u003e Human cholangiocytes (Chol), portal mesenchyme (PF) and primary hepatocytes (PHHs) were isolated and cultured as described in methods. Cholangiocytes and PHHs were cultured to generate h-CholOrg and h-HepOrgs. PFs and h-CholOrg were lentivirally transduced to generate fluorescently tagged cells to facilitate the identification after assembly. The generated assembloids were collected after 24h post-assembly (day 0), or at 3 or 6 days in culture and processed for further analysis (\u003cstrong\u003ec-l\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e, Experimental approach.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e, Representative bright field images of day 6 periportal assembloids. Scale bar, 100 µm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec\u003c/strong\u003e, Aggregation efficiency after 24 hours post-assembly. Graph represents mean ± SEM of n=3 biological replicates from n= 3 independent experiments. Assembloid indicates Cholangiocytes and Mesenchyme embedded in an HepOrg structure.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed\u003c/strong\u003e, Representative immunofluorescence images of assembloids produced using Aggrewell method (bottom) compared to human liver tissue periportal region (top). PFs (magenta) are visualized using the nRFP (assembloids) and CD90 (tissue) while Cholangiocytes (green) using nGFP (assembloids) and PanCK (tissue). Nuclei (blue) and membranes (grey) are stained with DAPI and Phalloidin, respectively. Scale bars, 50 µm. PV, portal vein.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee\u003c/strong\u003e, Cellular composition of day 6-periportal assembloids. Graph represents mean ±SEM from at least n=3 independent experiments. Dot, percentage of hepatocyte (Hep), cholangiocyte (Chol) and portal mesenchyme (PFs) per structure.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef\u003c/strong\u003e, Representative confocal images of day 3-periportal assembloids formed with cholangiocytes labelled with nGFP and PFs with n-RFP. Assembloids were stained for hepatocyte (HNF4A, yellow, middle panel), cholangiocyte (KRT19, white, left panel) and PFs (vimentin, white, right panel) markers. Nuclei are stained with DAPI (blue), membranes with phalloidin (green, left, grey, middle). Scale bars, 50 µm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg-h, \u003c/strong\u003escRNAseq analysis of assembloids.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg\u003c/strong\u003e, Correlation analysis between mesenchyme (this study, pink), hepatocytes (this study, brown) and cholangiocytes (this study, green) from assembloids and publicly available data from human liver tissue cell atlases.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eh\u003c/strong\u003e, Dot plot shows the expression of hepatocyte, cholangiocyte and mesenchyme markers in assembloids and human liver tissue datasets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ei\u003c/strong\u003e. Liver zonation genes heatmap for h-HepOrgs-DM and hepatocytes from assembloids (processed as \u003cem\u003epseudo\u003c/em\u003ebulk, see methods). Assembloids show enrichment of periportal markers while h-HepOrgs-DM are enriched in pericentral markers.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ej. \u003c/strong\u003eUrea synthesis analysis on day 5-assembloids (brown), h-HepOrgs in DM (purple) and PHHs (1-day culture, pink). CholOrg (green) were used as negative control. Graph, mean ± SD (n=3 donors). One-way ANOVA with Tukey test for multiple comparisons.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ek. \u003c/strong\u003eGluconeogenesis assay on day6-assembloids (brown), hHepOrg in DM (purple). Graph, mean ± SD (n=4 independent donors), two-tailed unpaired t-tests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003el\u003c/strong\u003e, Human albumin secretion in 1, 6, or 10 days-assembloids (brown), h-HepOrgs in DM (light purple) and EM2 (dark purple)). Data from h-HepOrgs is reproduced from Fig. 3f and shown for comparison. Graph, mean +/-SEM from n = 4–7 independent donors; Two-way ANOVA with Tukey test for multiple comparisons. \u003cstrong\u003ej-l\u003c/strong\u003e, Amounts are normalized to total cell number.\u003c/p\u003e","description":"","filename":"fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/f306f2228c515d8bcf197889.png"},{"id":91526560,"identity":"6f1c8fda-7a5a-4b8f-ac8e-722b704607e5","added_by":"auto","created_at":"2025-09-17 11:09:58","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":342196,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePeriportal assembloids mimic aspects of \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003ein vivo\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e human biliary fibrosis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea-h,\u003c/strong\u003e Assembloids were generated by assembling h-HepOrgs, cholangiocytes/ductal cells derived from cholangiocytes organoids (nuclear-GFP) and portal fibroblasts (PFs, nuclear-RFP) at a ratio 1 h-HepOrgs: 25 MSC : 100 Cholangiocytes (homeostatic-like) or 20-times more mesenchyme 1 h-HepOrgs: 500 MSC : 100 Cholangiocytes (fibrotic-like) and 24h (\u003cstrong\u003eb\u003c/strong\u003e) or 7 days later (\u003cstrong\u003ec-h\u003c/strong\u003e) the cultures were collected and processed for immunofluorescence (\u003cstrong\u003eb-e\u003c/strong\u003e) or RNAseq (\u003cstrong\u003ef-h\u003c/strong\u003e) analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea, \u003c/strong\u003eExperimental design.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e, Representative bright field images of 24h-assembloids in Aggrewell. Scale bars, 100 µm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec\u003c/strong\u003e, Representative brightfield images of 7-day homeostatic-like assembloids (left) and matching 20-fold excess mesenchyme assembloids (fibrotic-like, right). Scale bars, 100 µm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed\u003c/strong\u003e, Cell composition of homeostatic-like and fibrotic-like assembloids; Dot, percentage of hepatocyte, cholangiocyte or PFs per structure. Graph, mean ±SEM assembloids from n=3 independent experiments.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee\u003c/strong\u003e, Representative immunofluorescence images of periportal assembloids with homeostatic-like (top) or fibrotic-like (bottom) ratio of mesenchyme (nuclear-RFP, magenta) stained for hepatocyte (HNF4a, yellow, magenta arrow) and cholangiocyte (KRT19, white and nuclear-GFP, white arrow) markers. Low magnification also shows Phalloidin (blue, membrane) and DAPI (blue, nuclei) channels. Scale bars, 50 µm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef\u003c/strong\u003e, Correlation analysis between the three different populations of the fibrotic assembloids: mesenchyme (this study, pink), hepatocytes (this study, brown) and cholangiocytes (this study, green) and publicly available data from diseased human liver cell atlases.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg\u003c/strong\u003e, Dot plot shows hepatocyte, cholangiocyte and mesenchyme markers in fibrotic-like assembloids (this study) and liver tissue datasets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eh\u003c/strong\u003e, The expression pattern of hepatocytes, cholangiocytes and mesenchyme was compared between fibrotic and homeostatic assembloids from the same experiment. GSEA presents the enrichment of terms in fibrotic \u003cem\u003evs\u003c/em\u003e homeostatic assembloids. Selected terms are presented. The full list can be found in Supplementary Dataset 4. NES, normalized enrichment score.\u003c/p\u003e","description":"","filename":"fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/3ada81791f66b144ffaf12cc.png"},{"id":98492063,"identity":"6853bd73-61af-40a7-b10d-d517cd94ff3c","added_by":"auto","created_at":"2025-12-18 08:10:42","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5549664,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/36a1ca18-d667-4694-b4e9-ed937de05a57.pdf"},{"id":91524629,"identity":"bcf8bee3-e260-4805-aa81-f4bae88d3ded","added_by":"auto","created_at":"2025-09-17 11:01:58","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":29749,"visible":true,"origin":"","legend":"Supplementary Info Guide","description":"","filename":"7SupplementaryInformationGuideEditedLei.docx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/a71615c496aabb34b351a8e1.docx"},{"id":91524635,"identity":"071961f4-2754-40de-b498-45cd6de30915","added_by":"auto","created_at":"2025-09-17 11:01:58","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":607975,"visible":true,"origin":"","legend":"Supplementary DAtaset3","description":"","filename":"SupplementaryDataset3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/abe86a462f0e8bac1bd340b3.xlsx"},{"id":91524631,"identity":"57119d49-8509-4b44-b558-2db985518145","added_by":"auto","created_at":"2025-09-17 11:01:58","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":13730,"visible":true,"origin":"","legend":"Supplementary Table 1","description":"","filename":"SupplementaryTable1isolationrevisioncompleted.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/bff5994369777aa96f6c3d46.xlsx"},{"id":91527469,"identity":"5e2cefd1-900c-456d-9e6d-d65b50fd6058","added_by":"auto","created_at":"2025-09-17 11:17:58","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":18930,"visible":true,"origin":"","legend":"Supplementary Dataset 5","description":"","filename":"SupplementaryDataset5reagents.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/5b8331f2303a4db08ab71302.xlsx"},{"id":91526562,"identity":"031b4e3c-15ea-40cd-a217-4b000b5b37a5","added_by":"auto","created_at":"2025-09-17 11:09:59","extension":"xlsx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":951112,"visible":true,"origin":"","legend":"Supplementary Dataset 1","description":"","filename":"SupplementaryDataset1v3DT.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/7b3511364c2afbd8c0dfa0c8.xlsx"},{"id":91527471,"identity":"a93a04a8-a10e-4ead-be70-80814c7ff3f9","added_by":"auto","created_at":"2025-09-17 11:17:59","extension":"xlsx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":21709,"visible":true,"origin":"","legend":"Supplementary Table 4","description":"","filename":"Supplementarytable4patienttopatientcopy.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/ab1861da9ceb21480f5fd356.xlsx"},{"id":91524638,"identity":"16709fc0-6df5-4097-b6fc-f96b171237d7","added_by":"auto","created_at":"2025-09-17 11:01:58","extension":"pdf","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":417045,"visible":true,"origin":"","legend":"Extended Data Fig 2","description":"","filename":"ExtFigure2rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/ebfa2063699183ff5b21da08.pdf"},{"id":91528471,"identity":"3a49ae50-abd7-4fd5-ada9-2680f0ec27fb","added_by":"auto","created_at":"2025-09-17 11:25:59","extension":"xlsx","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":17637,"visible":true,"origin":"","legend":"Supplementary Table 2","description":"","filename":"SupplementaryTable2donorscomplete.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/446c51d892e29b46a1e6d5d8.xlsx"},{"id":91524644,"identity":"0c3addce-4c47-4f07-9b2b-104ada073f2b","added_by":"auto","created_at":"2025-09-17 11:01:59","extension":"pdf","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":925569,"visible":true,"origin":"","legend":"Extended Data Fig 10","description":"","filename":"ExtFigure10rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/a96b963a32981d76440ef504.pdf"},{"id":91524641,"identity":"d216ea79-c2ce-42bc-b0dc-390a54146c07","added_by":"auto","created_at":"2025-09-17 11:01:59","extension":"pdf","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":1162011,"visible":true,"origin":"","legend":"Extended Data Fig 6","description":"","filename":"ExtFigure6rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/d98dc970cdfbe4601cde9e34.pdf"},{"id":91526565,"identity":"ca563038-ceca-4794-acf3-34765e62e357","added_by":"auto","created_at":"2025-09-17 11:09:59","extension":"pdf","order_by":11,"title":"","display":"","copyAsset":false,"role":"supplement","size":1910401,"visible":true,"origin":"","legend":"Extended Data Fig 4","description":"","filename":"ExtFigure4rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/cce561f16298caf2dc4c735c.pdf"},{"id":91527470,"identity":"bdb3ac04-f459-4bf7-a4af-83627a170ce8","added_by":"auto","created_at":"2025-09-17 11:17:58","extension":"docx","order_by":12,"title":"","display":"","copyAsset":false,"role":"supplement","size":21679,"visible":true,"origin":"","legend":"Supplementary Table 3","description":"","filename":"SupplementaryTable3Comparisonwithpreviousresearchofgenerationofhepatocytesorganoidsvsourwork.docx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/d3eaf1c143392d06caee8f5c.docx"},{"id":91526568,"identity":"c8b74cd5-1989-4eeb-aba4-ba2c50036a3d","added_by":"auto","created_at":"2025-09-17 11:09:59","extension":"pdf","order_by":13,"title":"","display":"","copyAsset":false,"role":"supplement","size":2288167,"visible":true,"origin":"","legend":"Extended Data Fig 7","description":"","filename":"ExtFigure7rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/d48ad5cb2eb040afcd206ea7.pdf"},{"id":91524651,"identity":"c9e4ad07-26ab-4c90-b55a-0d2a74c4f7fb","added_by":"auto","created_at":"2025-09-17 11:01:59","extension":"pdf","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":2581925,"visible":true,"origin":"","legend":"Extended Data Fig 9","description":"","filename":"ExtFigure9rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/4a248952c42587177b6a685c.pdf"},{"id":91524653,"identity":"0b522314-2d53-4e18-8ea4-e7d468b0f504","added_by":"auto","created_at":"2025-09-17 11:01:59","extension":"pdf","order_by":15,"title":"","display":"","copyAsset":false,"role":"supplement","size":3422867,"visible":true,"origin":"","legend":"Extended Data Fig 1","description":"","filename":"ExtFigure1rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/d2daf42445daeb6c7f56fc62.pdf"},{"id":91524655,"identity":"4a9603ba-a3ef-456c-a03c-719958454bc6","added_by":"auto","created_at":"2025-09-17 11:02:00","extension":"xlsx","order_by":16,"title":"","display":"","copyAsset":false,"role":"supplement","size":37244721,"visible":true,"origin":"","legend":"Supplementary Dataset 2","description":"","filename":"SupplementaryDataset2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/2a5b281e5b5b4b63a69b5aa4.xlsx"},{"id":91527474,"identity":"e971551b-1d85-4835-a087-83a03aafabc5","added_by":"auto","created_at":"2025-09-17 11:17:59","extension":"xlsx","order_by":17,"title":"","display":"","copyAsset":false,"role":"supplement","size":377545,"visible":true,"origin":"","legend":"Supplementarty Dataset 4","description":"","filename":"SupplementaryDataset4fibrosisMH2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/cf4e0aaeda023dbcae84d1b9.xlsx"},{"id":91527473,"identity":"b488223a-bdfa-43d7-bb74-1a5d9247dd11","added_by":"auto","created_at":"2025-09-17 11:17:59","extension":"pdf","order_by":18,"title":"","display":"","copyAsset":false,"role":"supplement","size":635934,"visible":true,"origin":"","legend":"Extended Data Fig 3","description":"","filename":"ExtFigure3rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/0fbe6b852d4c2c63a2012258.pdf"},{"id":91524652,"identity":"b5522941-1a81-49c8-a341-fc887f3165a2","added_by":"auto","created_at":"2025-09-17 11:01:59","extension":"pdf","order_by":19,"title":"","display":"","copyAsset":false,"role":"supplement","size":688753,"visible":true,"origin":"","legend":"Extended Data Fig 8","description":"","filename":"ExtFigure8rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/9ab7fd1dee135fa9e54b4524.pdf"},{"id":91524649,"identity":"0b5af8fe-8f64-4b52-9690-ecc57fc62dda","added_by":"auto","created_at":"2025-09-17 11:01:59","extension":"pdf","order_by":20,"title":"","display":"","copyAsset":false,"role":"supplement","size":810799,"visible":true,"origin":"","legend":"Extended Data Fig 5","description":"","filename":"ExtFigure5rev2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/99a395143725ffda0b8a3e3b.pdf"},{"id":91524646,"identity":"41dacbb3-c79f-48c2-b9db-cb67bd6c3851","added_by":"auto","created_at":"2025-09-17 11:01:59","extension":"docx","order_by":21,"title":"","display":"","copyAsset":false,"role":"supplement","size":64438,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedDataFigureLegends.docx","url":"https://assets-eu.researchsquare.com/files/rs-5314788/v1/4decebb8d8672fe6586127b3.docx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nM.H. is an inventor in several patents on organoid technology. Y.K., S.D. and M.H. are inventors in a patent on human hepatocyte organoids. M.H. L.Y, S.D., A.M.D and A.S. are co-inventors in a patent on human assembloids. The remaining authors declare no competing interests.","formattedTitle":"Human assembloids recapitulate periportal liver tissue in vitro","fulltext":[{"header":"Main","content":"\u003cp\u003eChronic and end-stage liver diseases account for over 2 million human deaths world-wide\u003csup\u003e7\u003c/sup\u003e. Rodent models have advanced our understanding of liver biology. However, species-specific differences (e.g., metabolism, toxicity) impact our understanding of what are universal concepts and which are species-specific, making the translation of potential therapeutic targets to effective human therapies a significant challenge \u003csup\u003e8,9\u003c/sup\u003e. Human liver single cell and spatial transcriptomics have unveiled human cellular heterogeneity \u003csup\u003e10-16\u003c/sup\u003e. However, their static nature fails to inform us about the highly dynamic processes occurring in disease initiation and progression. Primary hepatocytes fail to expand in culture\u003csup\u003e17\u003c/sup\u003e, and while cancer cell lines have been informative, they suffer from genetic drifts. Reprogrammed hepatocytes (ProliHH) are proliferative, can repopulate the tissue,\u0026nbsp;but present bi-phenotypic and progenitor features\u003csup\u003e18\u003c/sup\u003e. Additionally, none of these models recapitulate the 3D bile canaliculi structures (thin and elongated lumina) observed in tissue\u003csup\u003e19,20\u003c/sup\u003e, making it difficult to model complex disease states and to\u0026nbsp;recapitulate patient-specific traits, which are essential characteristics to enable precision medicine approaches for early diagnosis and treatment.\u003c/p\u003e\n\u003cp\u003eOrganoids have emerged as promising models to grow human cells with the potential to better predict effective therapeutic outcomes\u003csup\u003e1\u003c/sup\u003e.\u0026nbsp;Human intestinal organoids effectively model human tissue structure and function\u003csup\u003e21-23\u003c/sup\u003e. However, recapitulating \u003cem\u003ein vitro\u0026nbsp;\u003c/em\u003ethe architecture and cellular interactions of complex tissues like the human liver remains an unmet challenge. We described the first liver organoid models from adult mouse and human tissue\u003csup\u003e4-6\u003c/sup\u003e (now known as cholangiocyte organoids) where cholangiocyte/ductal cells can be expanded in culture long-term, thereby generating millions of cells \u003cem\u003eex vivo\u003c/em\u003e. We\u003csup\u003e24,25\u003c/sup\u003eand others\u003csup\u003e26,27\u003c/sup\u003e demonstrated that these allow the study of mouse liver regeneration \u003cem\u003ein vitro\u003c/em\u003e. Small modifications to this system allowed the generation of branching organoids \u003csup\u003e28\u003c/sup\u003e, akin to the morphogenesis of the developing tissue\u003csup\u003e29-31\u003c/sup\u003e or could be transplanted to reconstruct the bile duct \u003cem\u003ein vivo\u003c/em\u003e\u003csup\u003e32\u003c/sup\u003e. Mouse adult hepatocyte organoids have been developed\u003csup\u003e33,34\u003c/sup\u003e. Also, mouse \u003csup\u003e35\u003c/sup\u003e and human \u003csup\u003e33,36\u003c/sup\u003e liver hepatoblast organoids were successfully generated from foetal tissue. However, expanding human adult hepatocytes from patient tissue has remained a challenge \u003csup\u003e37\u003c/sup\u003e. Regrettably, all these models consist only of epithelial cells and lack the ability to fully replicate the cellular interactions and architecture of \u003cem\u003ein vivo\u003c/em\u003e adult human liver tissue. Similarly, liver organoids derived from human pluripotent stem cells, although contain stromal and epithelial populations do not replicate native adult liver periportal cell interactions or architecture \u003csup\u003e38-40\u003c/sup\u003e. By co-culturing mouse cholangiocyte organoids with mouse liver portal mesenchyme, we obtained cholangiocyte-portal mesenchyme organoids that retain the binary cell-cell interactions present in the mouse liver\u003csup\u003e25,41\u003c/sup\u003e. Chimeric epithelial co-cultures between mouse cholangiocyte and 2D-human hepatocyte-like cells have been reported\u003csup\u003e42\u003c/sup\u003e. However, a complex 3D multi-cellular model that captures human liver portal cellular interactions does not exist for adult human patient liver tissue yet. \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHere, we developed an adult human hepatocyte organoid model (h-HepOrgs) that allows the long-term serial expansion (\u0026gt; 3months passaging at 1:2) of human adult hepatocytes directly from fresh patient liver tissue. h-HepOrgs retained gene expression and function of \u003cem\u003ein vivo\u003c/em\u003e human adult hepatocytes in a patient-specific manner and formed bile canaliculi structures akin to the ones in human tissue. As we expand and cryopreserve organoids from fresh tissue, we have been able to generate a living-biobank of hepatocyte organoids from 28 donors/patients. We combined these novel patient-derived hepatocyte organoids with primary human portal mesenchyme and human cholangiocyte organoids (h-CholOrg) from the same patient to generate human periportal assembloids that recapitulate functional cell interactions and architecture of the \u003cem\u003ein vivo\u0026nbsp;\u003c/em\u003etissue. Finally, we exploited the potential of this system to model aspects of human biliary fibrosis.\u0026nbsp;\u003c/p\u003e\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n"},{"header":"YAP and WNT activation facilitate the expansion of human hepatocyte organoids","content":"\u003cp\u003eTo recapitulate the epithelial-stromal interaction and tissue architecture of human liver tissue, we first sought to obtain an expandable source of adult hepatocytes, cholangiocytes and mesenchyme from the same individual. A pre-requisite was to first identify methods to expand human adult hepatocytes. Hence, we obtained human hepatocytes from patient tissue by perfusion \u003csup\u003e43,44\u003c/sup\u003e (methods) and cultured the cells in our previously published, mouse hepatoblast organoid culture medium (MM)\u003csup\u003e35\u003c/sup\u003e. However, the cultures were rapidly filled with cholangiocyte organoids, preventing any further analysis (Extended Data Fig. 1a, No MACS). Then, we adapted the isolation to include a step of magnetic activated cell sorting (MACS) for EpCAM-positive cholangiocyte cells on the perfused tissue. This allowed obtaining viable hepatocytes from the negative fraction, while generating cholangiocyte organoids from the same patient by culturing the EpCAM-positive cholangiocyte fraction in our previously published Huch \u003cem\u003eet al\u003c/em\u003e., 2015\u003csup\u003e4\u003c/sup\u003e medium (Fig. 1a, Extended Data Fig. 1a-c, Methods and Supplementary Table 1). However, although we detected viable hepatocytes and minimal growth for 7-14 days in MM medium (Fig. 1c, MM), hepatocytes rapidly deteriorated and died thereafter.\u0026nbsp;\u003c/p\u003e\u003cp\u003eNext, we sought to identify culture conditions for the long-term expansion of human adult hepatocytes as hepatocyte organoids.\u0026nbsp;We hypothesized that signaling pathways involved in cancer progression or tissue regeneration could drive the exit from quiescence in hepatocytes and activate their proliferative state of in culture. To explore this, we analyzed expression profiles from our human liver cancer organoids\u003csup\u003e45\u003c/sup\u003e and publicly available datasets from mouse partial hepatectomy\u003csup\u003e33\u003c/sup\u003e and compared it to human healthy and cancer tissue (n=4 different datasets, Supplementary Dataset 1 and methods). Ingenuity Pathway Analysis (IPA) revealed several pathways were consistently up- or down-regulated such as AMPK, EGF, mTOR, and IGF-1 across at least two datasets (Extended Data Fig. 1d, Supplementary Dataset 1). WNT, MAPK, and FGFR2 were consistently active, while IL-6, HIPPO, and NOTCH appeared inactive (Fig. 1b). Among upstream regulators, components of the WNT pathway (e.g., CTNNB1, LGR5) were upregulated, whereas LATS1 (a negative regulator of YAP/TAZ in the HIPPO pathway) was downregulated, suggesting YAP activation (Extended Data Fig. 1e).\u0026nbsp;\u003c/p\u003e\u003cp\u003eBoth WNT and YAP are established drivers of liver regeneration \u003csup\u003e3,46,47\u003c/sup\u003e and cancer \u003csup\u003e47-49\u003c/sup\u003e. Therefore, to promote long-term hepatocyte expansion, we activated WNT and YAP signalling by supplementing our published mouse hepatoblast medium \u003csup\u003e35\u003c/sup\u003e with a WNT surrogate (WntS) \u003csup\u003e50\u003c/sup\u003e and a LATS1/2 inhibitor (TRULI or TDI-011536) \u003csup\u003e51\u003c/sup\u003e. Combining both enabled serial passaging (5–6 passages) of human hepatocyte organoids as solid structures with no lumina (Fig. 1c–f, Extended Data Fig. 1g). TRULI-treated cultures showed superior morphology, so we continued with the MM+WntS+TRULI combination, termed h-HepOrg-EM1 from hereon. The other tested pathways had no consistent or quantifiable organoid growth (Extended Data Fig. 1f).\u003c/p\u003e\u003cp\u003eWe further optimized the medium (MM) by testing each component’s necessity. Notably, removing nicotinamide improved organoid formation efficiency nearly 10-fold and enabled long-term culture for over 3 months (\u0026gt;10 passages at 1:2 split/week) (Extended Data Fig. 1i, Supplementary Table 2). These results were in line with our IPA analysis showing NAD signaling inactivity (Fig. 1b) and previous reports of nicotinamide hepatotoxicity in humans \u003csup\u003e52\u003c/sup\u003e. Using these optimized conditions (EM1 without nicotinamide, hereafter called h-HepOrg-EM2), we successfully generated expandable human hepatocyte organoids (h-HepOrgs) from 28 patients (11–85 years old, 30% female) with 100% efficiency (Supplementary Table 2). No other tested conditions—including those from human foetal or mouse hepatocyte organoids \u003csup\u003e33,34\u003c/sup\u003e —supported robust expansion (Extended Data Fig. 1j, Supplementary Table 3 and Source Data). h-HepOrgs maintained stable chromosome numbers over time and could be frozen/thawed without loss of expansion capacity, enabling the creation of a living biobank from a total of 28 different donors (Extended Data Fig. 1k-l).\u003c/p\u003e\u003cp\u003eTogether, these results demonstrate that combination of WNT and YAP activation allows the long-term expansion of adult human hepatocyte organoids.\u0026nbsp;\u003c/p\u003e"},{"header":"Adult h-HepOrgs retain in vivo structural, transcriptional and functional features ","content":"\u003cp\u003eTo characterize the expanded h-HepOrgs, we first performed RNAseq analysis on early (P1-P3) and late (P10) passage cultures and compared their expression pattern with freshly isolated primary human hepatocytes (PHHs, primary) and human cholangiocyte organoids (h-CholOrgs) from the same donors (when possible). Expanded h-HepOrgs closely correlated with freshly isolated hepatocytes, while h-CholOrgs from the same donors clustered separately (Extended Data Fig. 2a-b). Gene expression and gene set enrichment analysis revealed that the h-HepOrgs exhibited a proliferative signature that was maintained until late passages (\u0026gt;P10) and resembled regenerating tissue after hepatectomy (Extended Data Fig. 2c-e). These results were in agreement with the positive Ki-67 (Fig. 2a, top) and negligible cleaved-caspase 3-staining (Extended Data Fig. 2f). h-HepOrgs exhibited elevated expression of WNT and YAP target genes, consistent with pathway activation following WNT and LATS1/2 inhibitor treatment (Extended Data Fig. 2g-h). Immunofluorescence confirmed the nuclear localization of YAP in TRULI-treated h-HepOrgs (Fig. 2b, middle panel and Extended Fig. 2i), but not in non-treated cultures (Fig. 2b, bottom panel) or in homeostatic tissue (Fig. 2b, top panel). qPCR confirmed these results (Extended Data Fig. 2j). However, we cannot exclude that off-target effects may also contribute to the growth effects, as TRULI can inhibit kinases beyond LATS1/2. Marker gene expression analysis showed that the expanded h-HepOrgs expressed hepatocyte markers such as \u003cem\u003eHNF4A, ALB,\u003c/em\u003e several apolipoproteins (\u003cem\u003eAPOC2, APOA4\u003c/em\u003e) and cytochromes (\u003cem\u003eCYP3A4, CYP3A7\u003c/em\u003e), albeit at lower levels compared to freshly isolated hepatocytes (Fig. 2d, Extended Data Fig. 2c, and Supplementary Dataset_2). Cholangiocyte markers such as \u003cem\u003eSOX9, KRT19\u003c/em\u003e or \u003cem\u003eKRT7\u003c/em\u003e were markedly reduced while the expression of the embryonic liver marker \u003cem\u003eAFP\u003c/em\u003e suggested incomplete maturation (Extended Data Fig. 2k and Supplementary Dataset_2). qPCR and immunofluorescence analysis confirmed that HNF4A and the apical and polarity marker \u003cem\u003eCD13\u003c/em\u003e (\u003cem\u003eANPEP\u003c/em\u003e) were both highly expressed (Fig. 2c, Extended Data Fig. 2k and Supplementary Dataset_2). However, detailed analysis of the distribution of CD13 expression showed the presence of wide, disconnected, round lumina, which does not reflect the morphology of the bile canalicular network formed by hepatocytes \u003cem\u003ein vivo\u003c/em\u003e\u003cem\u003e\u003csup\u003e53-55\u003c/sup\u003e\u003c/em\u003e (Fig. 2c, compare CD13 in h-HepOrgs to the tissue panel in Fig. 3b). Taken together, these results suggested that the expanding h-HepOrgs in EM2 medium presented an immature hepatocyte state. Therefore, we sought to define a differentiation medium.\u0026nbsp;\u003c/p\u003e\u003cp\u003eLATS1/2 inhibition was recently shown to promote cholangiocyte growth \u003csup\u003e56\u003c/sup\u003e, while it is well established that YAP activation drives hepatocyte de-differentiation and its inactivation facilitates re-differentiation \u003csup\u003e57\u003c/sup\u003e. Therefore, we reasoned that reducing YAP activation would facilitate the maturation of h-HepOrgs. Following several iterations, we developed a hepatocyte differentiation medium (h-HepOrg-DM from hereon) in which we removed YAP and FGFR2 activation, maintained Wnt signaling and added Dexamethasone (Extended Data Fig. 3a and methods). Under differentiation medium, the cellular morphology improved: hepatocytes (HNF4A+) reduced proliferation, acquired a significantly higher cytoplasm to nuclei ratio and improved bile canaliculi (CD13+), which presented thinner and more elongated morphology (Fig. 2a, c, compare EM2 vs DM and Extended Data Fig. 3b). Combined these features suggested enhanced hepatocyte maturation. To assess the extent of the maturation, we performed RNAseq analysis. Principal Component Analysis (PCA) revealed that differentiated h-HepOrgs move closer to fresh isolated hepatocytes and farther away from cholangiocyte organoids, when compared to h-HepOrgs in expansion medium (Extended Data Fig. 3c). The cells increased the expression of many mature markers, some to similar levels as freshly isolated human hepatocytes, including \u003cem\u003eALB\u003c/em\u003e, several APO lipoproteins (\u003cem\u003eAPOE, APOA1),\u003c/em\u003e bile acid transporters (\u003cem\u003eABCC2\u0026nbsp;\u003c/em\u003e(MRP2)) and cholesterol and bile acid metabolic genes (\u003cem\u003eSCARB1, ABCG8\u0026nbsp;\u003c/em\u003eand\u003cem\u003e\u0026nbsp;CYP27A1).\u0026nbsp;\u003c/em\u003eAdditionally, several detoxifying enzymes such as \u003cem\u003eCYP2C9,\u003c/em\u003e \u003cem\u003eCYP3A5, CYP3A4\u0026nbsp;\u003c/em\u003eand\u003cem\u003e\u0026nbsp;MAOA\u003c/em\u003e, some of them pericentrally zonated\u003cem\u003e\u003csup\u003e13,14,58,59\u003c/sup\u003e\u003c/em\u003e\u003cem\u003e,\u0026nbsp;\u003c/em\u003ewere also upregulated (Fig. 2d and Extended Data Fig.3f-i). Consistent with these results, we found significant positive enrichment for many gene sets related to mature hepatocyte functions including cholesterol, fatty acid and drug metabolism, Phase II conjugation, clot formation and bile secretion, amongst others. Conversely, signatures related to cell cycle and proliferation were negatively enriched (Fig. 2e and Extended Data Fig. 3e). Similarly, the expression of the embryonic marker \u003cem\u003eAFP\u003c/em\u003e and the cholangiocyte makers \u003cem\u003eKRT7\u0026nbsp;\u003c/em\u003eand\u003cem\u003e\u0026nbsp;KRT19\u0026nbsp;\u003c/em\u003ewere reduced (Fig. 2d and Extended Data Fig. 3i).\u003c/p\u003e\u003cp\u003eImportantly, some pericentrally zonated genes, such as \u003cem\u003eCYP2E1\u0026nbsp;\u003c/em\u003eand \u003cem\u003eGLUL\u003c/em\u003e (Glutamine Synthase, GS), as well as some periportally zonated genes, such as\u003cem\u003e\u0026nbsp;ALDOB\u0026nbsp;\u003c/em\u003eor\u003cem\u003e\u0026nbsp;SCD,\u003c/em\u003e were highly upregulated (Fig. 2d and Extended Data Fig. 3h). Immunofluorescence analysis for pericentral (GS) or periportal (HAL) markers indicated that some cells within the organoids presented a gradient of expression of zonated genes, with some cells positive and others negative for the markers (Fig. 2f). Dual immunofluorescence staining for CYP2E1 (pericentral marker) and ECAD (enriched in periportal region) highlighted the heterogeneity and spatial distribution of hepatocyte function within the same h-HepOrg, at least for those genes tested (Extended Data Fig.3j).\u0026nbsp;\u003c/p\u003e\u003cp\u003eStrikingly, under differentiation conditions, we observed that h-HepOrgs recapitulated the complex cell polarity of \u003cem\u003ein vivo\u003c/em\u003e hepatocytes\u003csup\u003e17\u003c/sup\u003e, with the tight-junction and apical polarity marker ZO-1 localized to the apical surface of adjacent hepatocytes (Fig. 3a, bottom panel), resembling the morphology of bile canaliculi (BC) in human liver tissue (Fig. 3a, top panel, Radixin). Immunofluorescence staining with the apical marker CD13 followed by image analysis and reconstruction revealed that the differentiated h-HepOrgs displayed longer and more branched BC networks within each organoid (Fig. 3b, middle panel), when compared to the same organoids in expansion medium (Fig. 3b, top panel) and resembled the \u003cem\u003ein vivo\u0026nbsp;\u003c/em\u003etissue (Fig. 3b, bottom panel). Additionally, the BC network connectivity was also significantly improved, coming closer to that of the tissue (Fig. 3c). Noteworthy, we observed that different patients present fine-detailed differences in BC morphology (Extended Data Fig. 4a-c), with some patients presenting thin and homogenous BC, some wider and inhomogeneous BC, and others full of branchlets (Extended Data Fig. 4b). We found a similar variation in the BC architecture across our different organoid cultures, suggesting that our model could capture different types of BC structures that are observed in patient tissue cohorts (Extended Data Fig. 4d).\u0026nbsp;\u003c/p\u003e\u003cp\u003eGiven that our h-HepOrgs are derived directly from patient tissue, we next assessed whether they retain patient-to-patient variability in culture, thus enabling patient-specific modeling of hepatocyte-related liver diseases. For that, we analysed the transcriptome of primary hepatocytes at time of isolation and their matching h-HepOrgs under differentiation medium to identify the specific gene signatures of each patient. We found a strong correlation (R\u003csup\u003e2\u003c/sup\u003e=0.7-0.9) between the organoids and the original primary hepatocytes they derived from (Extended Fig. 4e). Interestingly, many of the patient-specific genes we found expressed in organoids and their source cells had been associated with either susceptibility to hepatitis virus infection (\u003cem\u003eIL1RL1\u003c/em\u003e or \u003cem\u003eERAP2\u003c/em\u003e) or liver cancer (\u003cem\u003eGPC3\u003c/em\u003e) or to cholestasis during pregnancy (\u003cem\u003eGABRP\u003c/em\u003e). More interestingly, some genes were involved in metabolic pathways like the glutathione-related genes \u003cem\u003eGSTM3\u003c/em\u003e and \u003cem\u003eGSTM1\u003c/em\u003e, the lactate dehydrogenase \u003cem\u003eLDHC\u003c/em\u003e and the lipid metabolic related genes \u003cem\u003eAPOA4\u003c/em\u003e, the fatty acyl-CoA reductase \u003cem\u003eFAR2\u003c/em\u003e or the Acyl-CoA synthetase \u003cem\u003eACSM1,\u0026nbsp;\u003c/em\u003eamongst others (Fig. 3d and Supplementary Table 4). These results indicated that h-HepOrgs could preserve patient-to-patient specific signatures, with significant implications for modelling human liver diseases.\u0026nbsp;\u003c/p\u003e\u003cp\u003eNext, we compared the functional performance of differentiated h-HepOrgs to primary human hepatocytes (PHHs). HepOrg differentiated in DM exhibited mature hepatic functions, including robust albumin secretion and, moderate cytochrome P450 activity, comparable to 7-day PHHs (Fig. 3e–f). Specifically, differentiated h-HepOrgs displayed CYP2C9 activity equivalent to that of 7-day PHHs, and modestly reduced CYP3A4 activity, while 1-day PHHs demonstrated superior activity for both enzymes. Notably, mass spectrometry analysis revealed that differentiated h-HepOrgs significantly outperformed 1-day PHHs in converting the antiarrhythmic and antihypertensive drug Verapamil into its primary metabolite, Norverapamil (Fig. 3g). This superior overall metabolic performance toward Norverapamil suggests a more robust or sustained expression and coordination among multiple CYP enzymes relevant to Verapamil metabolism including the metabolizing enzymes \u003cem\u003eCYP\u003c/em\u003e\u003cem\u003e2\u003c/em\u003e\u003cem\u003eC8, CYP3A4, and CYP3A5\u003c/em\u003e, all responsible for Verapamil N-demethylation and highly expressed in h-HepOrgs in differentiation medium (Extended Data Fig. 3g, i). Furthermore, we observed inter-donor variability in Verapamil metabolism among h-HepOrgs lines (Fig. 3g), reflecting patient-specific metabolic phenotypes and underscoring the potential of this platform for personalized drug metabolism studies.\u0026nbsp;\u003c/p\u003e\u003cp\u003eRemarkably, both expanded and differentiated hepatocyte organoids readily engrafted and maintained their hepatic function \u003cem\u003ein vivo\u003c/em\u003e, following xenotransplantation in the Tyrosinemia type I liver disease mouse model (\u003cem\u003eFah\u003csup\u003e-/-\u003c/sup\u003eRag2\u003csup\u003e-/-\u003c/sup\u003eIl2rg\u003csup\u003e-/\u003c/sup\u003e\u003c/em\u003e mouse)\u003csup\u003e60\u003c/sup\u003e, with grafts distributed throughout the liver parenchyma. Importantly, the engrafted cells were able to rescue the otherwise lethal phenotype of the mice (Fig. 3h and Extended Data Fig. 4f).\u0026nbsp;\u003c/p\u003e\u003cp\u003eIn summary, we have developed a novel h-HepOrgs model that enables the growth of functional adult human hepatocytes directly from patient tissue, preserving hepatocyte polarity and generating a bile canaliculi network that resembles adult liver tissue \u003cem\u003ein vivo\u003c/em\u003e, while retaining some aspects of patient-to-patient variability.\u0026nbsp;\u003c/p\u003e"},{"header":"Human liver assembloids model periportal tissue ","content":"\u003cp\u003eWe\u0026nbsp;next aimed to reconstruct the periportal region of the liver lobule by reproducing the cellular interactions between hepatocytes, cholangiocytes and portal mesenchyme, specifically portal fibroblasts. To\u0026nbsp;identify conditions to isolate and culture portal mesenchyme over multiple passages we used PDGFRA\u003cem\u003e+\u0026nbsp;\u003c/em\u003eliver cells, as \u003cem\u003ePDGFRA\u003c/em\u003e is\u003cem\u003e\u0026nbsp;\u003c/em\u003eexclusively expressed in liver mesenchyme and absent in other stromal cells, according to different human liver cell atlases \u003csup\u003e10,13,14,16,61\u003c/sup\u003e\u0026nbsp; (Extended Data Fig. 5a-f and methods).\u003cem\u003e\u0026nbsp;\u003c/em\u003eNext, we examined publicly available datasets\u003csup\u003e10,61\u003c/sup\u003e to further enrich for healthy portal fibroblasts. We found that the cell surface receptor CD90 (\u003cem\u003eTHY1\u003c/em\u003e) is expressed exclusively in human portal fibroblasts from healthy individuals (Extended Data Fig. 5g, 6a, b). Immunofluorescence analysis confirmed the restricted expression of CD90 to the periportal region (Extended Data Fig. 6c). Thus, we combined CD90 with PDGFRA to enrich for healthy human portal fibroblasts under defined culture conditions (Extended Data Fig. 6d). RNAseq and qPCR analysis revealed that CD90+/PDGFRA+ cells expressed portal fibroblasts markers (e.g. \u003cem\u003eDCN, THY1\u003c/em\u003e and \u003cem\u003eASPN\u003c/em\u003e)\u003cem\u003e\u0026nbsp;\u003c/em\u003eand several growth factors (\u003cem\u003eHGF\u003c/em\u003e, \u003cem\u003eWNT5A\u0026nbsp;\u003c/em\u003eamong others), which are all highly enriched in portal mesenchyme \u003cem\u003ein vivo\u003c/em\u003e. Conversely, hepatic stellate cell (HSC, \u003cem\u003eRGS5\u003c/em\u003e, \u003cem\u003eLRAT\u003c/em\u003e and \u003cem\u003eRELN\u003c/em\u003e) and vascular smooth muscle cell (VSMC, \u003cem\u003eMYH11\u003c/em\u003e) markers were not detected (Extended Data Fig. 6e-h). Immunofluorescence analysis for vimentin (mesenchyme) and CD90 (portal mesenchyme) confirmed that the majority of the expanded cells were portal fibroblasts (Extended Data Fig. 6i).\u003c/p\u003e\u003cp\u003eNext, we aimed to generate human periportal assembloids. \u003ca id=\"_anchor_1\" href=\"#_msocom_1\" language=\"JavaScript\" name=\"_msoanchor_1\"\u003e\u003c/a\u003eTo ensure that the structures contained the different cells and to facilitate their visualisation, we tagged human cholangiocyte organoids and portal mesenchymal cells with nuclear fluorescent proteins (Extended Data Fig. 6j), while leaving hepatocyte organoids unlabelled. Then, to determine the proportions of cells to assemble into composite structures we first quantified the physiological proportion of cholangiocytes portal mesenchyme and hepatocytes \u003cem\u003ein vivo\u003c/em\u003e in human tissue, which resulted in an average of 15% Chol: 8% MSC: 77% Hep (Extended Data Fig. 7a). To induce the self-assembly of the three cell populations into a single structure that would recapitulate periportal spatial organization, we tested several approaches to aggregate h-HepOrgs with dissociated portal liver mesenchyme and cholangiocytes from h-CholOrgs from the same donor, when possible. Of all the approaches tested, mixing one h-HepOrg structure with a defined number of portal mesenchymal and ductal cells (from h-CholOrgs) in 96-well low adhesion U-plates readily generated structures where the 3 cell types were together with Cholangiocyte and MSC cells embedded inside the HepOrg structure. We called these structures periportal assembloids (Fig. 4a, b). The ratio 1 HepOrg: 25 MSC and 100 Chol better captured the tissue cell ratios at day 6 post-assembloid formation, and was selected for further experiments (Extended Data Fig. 7b, c). To increase the number of assembloids generated, we used Aggrewell\u003csup\u003eTM\u003c/sup\u003e plates (Fig. 4b and Extended Data Fig. 7d). Notably, both methods generated assembloids with high efficiency (~80% efficiency) (Fig. 4c) and closely maintained the cellular composition and proportions of the tissue (Fig. 4e). Therefore, we only used Aggrewell from hereon. Remarkably, from day 3 onwards, the periportal assembloids recapitulated key architectural features of the \u003cem\u003ein vivo\u003c/em\u003e tissue, with ductal cells (KRT19\u003csup\u003e+\u003c/sup\u003e and nGFP\u003csup\u003e+\u003c/sup\u003e) forming bile duct-like structures containing open lumina, with portal mesenchymal cells (nuclear-RFP) in close proximity, and embedded within the hepatocyte (HNF4A⁺) parenchyma. This architectural organization, where ductal cells are forming an apical lumen, basally contacted by mesenchymal cells and embedded in the hepatocyte structure, was observed in the majority (80%) of the assembloids and across donors. These results were independent of the patient source for the healthy portal liver mesenchyme, indicating minimal impact of patient-origin under healthy conditions (Fig. 4d-f and Extended Data Fig. 7d-h).Vimentin staining confirmed that MSC consistently extended long cellular processes towards the basal side of cholangiocytes, leading to physical contacts and reminiscent of the interactions observed in human tissue, although not completely wrapping the cholangiocytes as in the portal tracts \u003cem\u003ein vivo\u0026nbsp;\u003c/em\u003e(Fig. 4f and Extended Data Fig. 7h). Under these conditions, the assembloids could be maintained for at least two weeks in culture, with no evidence of increased cell death or proliferation over time (Extended Data Fig. 7i-l).\u0026nbsp;\u003c/p\u003e\u003cp\u003eNext, we employed single cell RNAseq (scRNAseq) analysis to benchmark our model to the \u003cem\u003ein vivo\u003c/em\u003e human liver tissue. Clustering, PCA and correlation analysis indicated that the assembloid cells mostly overlap with the corresponding populations from publicly available human liver cell atlases\u003csup\u003e10,13,14,16,61\u003c/sup\u003e (Fig. 4g and Extended Data Fig. 8a). Hepatocytes, cholangiocytes and mesenchymal cells from assembloids expressed classical markers of their \u003cem\u003ein vivo\u003c/em\u003e counterparts (Hep: \u003cem\u003eALB, HNF4A, TTR\u003c/em\u003e; Chol: \u003cem\u003eKRT7, KRT19, EPCAM\u003c/em\u003e; MSC: \u003cem\u003eCOL1A1, VIM, THY1\u003c/em\u003e) (Fig. 4h), indicating that they retain their identity in human assembloids. GSEA confirmed that mesenchymal cells were highly enriched for signatures of ECM organization and cell adhesion, cholangiocytes for cytoskeleton and cell-cell communication, and hepatocytes for fatty acid metabolism, complement and drug metabolism, similar to \u003cem\u003ein vivo\u003c/em\u003e human liver tissue (Extended Data Fig. 8c, d).\u0026nbsp;\u003c/p\u003e\u003cp\u003eInterestingly, we observed heterogeneous expression of classical zonated hepatocyte markers, with a fraction of hepatocytes expressing periportal (\u003cem\u003eSAA1 and SAA2, APOA1\u003c/em\u003e) and others expressing pericentral (\u003cem\u003eCYP2E1)\u003c/em\u003e markers (Fig. 4h and Extended Data Fig. 8b). To investigate whether the periportal assembloid microenvironment and the interaction with portal ductal and mesenchymal populations could promote a more portalized hepatocyte identity, we compared the gene expression profiles of hepatocytes from HepOrgs cultured in DM with those from assembloids (also cultured in DM). Notably, hepatocytes within assembloids exhibited higher expression of periportal markers, including \u003cem\u003eSAA1, SAA2, HAMP, and APOA1\u003c/em\u003e, while showing reduced expression of pericentral genes such as \u003cem\u003eCYP2E1, CYP3A4, and GLUL\u003c/em\u003e, compared to HepOrgs cultured alone. (Fig. 4i). Staining for the periportal hepatocyte markers SAA1 and SAA2 confirmed the\u0026nbsp;spatially heterogenous\u0026nbsp;expression of these portal markers,\u0026nbsp;with the positive cells overlapping with regions of ECAD⁺ high cells (Extended Data Fig. 9a), and in agreement with our scRNAseq results (Fig. 4 h).\u003c/p\u003e\u003cp\u003eNotably, periportal assembloids exhibited enhanced functional specialization characteristic of periportal hepatocytes. They outperformed h-HepOrgs, cultured in the same conditions, in both urea production and gluconeogenesis (both portal functions), while the drug-metabolizing capacity associated with pericentral hepatocytes was less pronounced compared to hepatocyte organoids, in line with their more portal-like nature. As expected, periportal assembloids retained core hepatocyte functions, with albumin secretion increasing over time to levels matching hepatocyte organoids and exceeding 2D primary hepatocyte cultures. (Fig.4j-l and Extended Data Fig. 9b).\u0026nbsp;\u003c/p\u003e\u003cp\u003eThese findings suggested that the periportal microenvironment within the assembloids could promote the acquisition of a more portal-like hepatocyte identity. In line with this hypothesis, we noted that some hepatocyte membranes joined the lumen of the bile ducts, similar to what we observed in \u003cem\u003ein vivo\u003c/em\u003e tissue and suggestive of physiological cell-cell contact between both cell types (Extended Data Fig. 9c-d).\u003c/p\u003e\u003cp\u003eCombined, we have generated a human liver periportal assembloid model that faithfully captures the gene expression, cell interactions and aspects of the tissue architecture of the human liver periportal region \u003cem\u003ein vitro\u003c/em\u003e.\u0026nbsp;\u003c/p\u003e"},{"header":"Human assembloids model aspects of biliary fibrosis ","content":"\u003cp\u003ePortal mesenchyme often contributes to myofibroblast populations in human fibrosis\u003csup\u003e62,63\u003c/sup\u003e. Hence, we next investigated whether we could utilize our human assembloid model containing portal fibroblasts to recapitulate aspects of human liver disease \u003cem\u003ein vitro,\u0026nbsp;\u003c/em\u003especifically biliary fibrosis. Interestingly, increasing (20-fold) the total number of initial mesenchymal cells, while keeping the other cell numbers constant (even from the same source tissue), altered the cell composition of the assembloids. We found that the number of cholangiocytes (GFP\u003csup\u003e+\u003c/sup\u003e, KRT19\u003csup\u003e+\u003c/sup\u003e) increased, while the total number of hepatocytes (HNF4A\u003csup\u003e+\u003c/sup\u003e) decreased (Fig. 5a-d). Ki-67 staining indicated that c\u003cstrong\u003eholangiocytes exhibited early proliferative responses to fibrotic cues\u003c/strong\u003e, while cleaved-caspase 3 staining revealed that the reduction in hepatocyte numbers was associated with increased hepatocyte death, at least in part, through apoptosis (Extended Data Fig. 10a-c). This finding was consistent with our observations in mouse assembloids\u003csup\u003e64\u003c/sup\u003e, suggesting a conserved mechanism across species.\u0026nbsp;\u003c/p\u003e\u003cp\u003escRNAseq clustering and correlation analysis revealed that the hepatocytes, cholangiocytes and mesenchyme from assembloids with excess mesenchyme, but not from homeostatic mesenchyme numbers, recapitulated the gene expression of their corresponding cells from publicly available datasets of diseased livers \u003csup\u003e10,16\u003c/sup\u003e (Fig. 5f). The top markers identifying the three cell populations in the corresponding patient datasets were also highly expressed in the corresponding assembloids cells (Fig. 5g), while GSEA revealed that mesenchyme and cholangiocytes from fibrotic, but not homeostatic assembloids, increased expression of collagen and matrix deposition processes (Fig. 5h, Extended Data Fig. 10f, g and Extended Dataset_4). Similarly, cholangiocytes, but not mesenchyme, presented signatures of proliferation (Fig. 5h and Extended Data Fig. 10f), in agreement with the increased number of GFP+ cholangiocytes detected (Fig. 5c-d). These gene signatures (increased matrix and cholangiocyte numbers) are reminiscent to the fibrotic tissue from human patients with biliary fibrosis and primary sclerosis cholangitis (PSC)\u003csup\u003e16,62,63\u003c/sup\u003e. Hence, from hereon we named the assembloids with excess mesenchyme “fibrotic-like”, to distinguish them from the “homeostatic-like” with homeostatic numbers of mesenchymal cells.\u003c/p\u003e\u003cp\u003eStrikingly, we found gene sets for inflammatory reactions including TNF signalling, several interleukins (IL-4, IL-6), NFKB, JAK-STAT and Toll-like receptor cascade among the most positively enriched gene sets in hepatocytes (Fig. 5h and Extended Data Fig. 10d, e) from fibrotic-like assembloids compared to homeostatic ones. Conversely, cell cycle signatures and hepatocyte functions such as bile secretion, lipid or drug metabolism were negatively enriched (Fig. 5h). Both, hepatocytes and cholangiocytes from fibrotic assembloids were also highly enriched in TGF-B signalling signatures (Fig. 5h Extended Data Fig. 10d-f), mirroring the transcriptional changes in patients with biliary fibrosis\u003csup\u003e16\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eMorphologically, we observed that the fibrotic-like assembloids, but not matching homeostatic assembloids, exhibited a cystic-like phenotype reminiscent of cholangiocyte organoids (Extended Data Fig. 10h-i). This observation was in line with the immunofluorescence analysis, which indicated that in fibrotic-like assembloids some hepatocytes (HNF4a\u003csup\u003e+\u003c/sup\u003e, GFP\u003csup\u003e-\u003c/sup\u003e) were positive for the cholangiocyte marker KRT19, and opened lumina, resembling the polarity of simple ductal epithelium, which is suggestive of potential hepatocyte-to-duct trans-differentiation (Fig. 5e and Extended Data Fig. 10j, magenta arrow and asterisk). Interestingly, all these phenotypes: (i) increased signatures in TNF-A, IL-4, IL-6, (ii) increased hepatocyte apoptosis and (iii) increased expression of cholangiocyte markers have been reported in fibrotic patients as well as in recent liver cell atlases of primary sclerosis cholangiatis (PSC) and primary biliary cirrhosis (PBC) patients\u003csup\u003e16,65\u003c/sup\u003e. These results combined suggest that our assembloid model with excess mesenchyme mimics some aspects of human biliary fibrosis as seen in cholangiopathies, including PSC and PBC.\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eFailure in maintaining the intricate cellular organization and multidirectional interactions of the cells within the liver lobule leads to chronic liver disease, often presenting with cholestasis and fibrosis, which can lead to cirrhosis and cancer \u003csup\u003e66,67\u003c/sup\u003e.\u0026nbsp;While reductionist by nature,\u0026nbsp;ex-vivo systems, offer powerful tools to\u0026nbsp;dissect disease mechanisms, particularly, the relative contributions of the distinct intrinsic cellular programs and microenvironmental cues\u0026mdash;including cell-cell interactions\u0026mdash;to\u0026nbsp;disease initiation and progression. We recently showed that mouse periportal assembloids retain key architectural features, such as the reconstruction of the bile canaliculi-bile duct connection and can serve as a tractable and modular in vitro model to investigate universal principles of bile canaliculi formation, bile flow, cholestatic injury and biliary fibrogenesis\u003csup\u003e64\u003c/sup\u003e . However, species-specific differences in drug metabolism, toxicity profiles or liver pathophysiology, necessitate the development of complementary human models that capture patient-specific features to better understand diseases mechanisms, identify therapeutic strategies, or screen for therapeutic compounds. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eRecent advances in human liver models underscore the ongoing efforts and the broad interest\u003cem\u003e\u0026nbsp;\u003c/em\u003ein developing physiologically relevant in vitro systems. These include: iPSC-derived hepatocyte organoids exhibiting dual zonation \u003csup\u003e68\u003c/sup\u003e, functional hepatocyte organoids derived from cryopreserved hepatocytes \u003csup\u003e69\u003c/sup\u003e, mass-generation of hepatobiliary organoids \u003csup\u003e70\u003c/sup\u003e, co-cultures of dermal fibroblasts with hepatocyte spheroids \u003csup\u003e71\u003c/sup\u003e or mouse fibroblasts aggregated with hepatocyte spheroids and cholangiocyte organoids \u003csup\u003e72\u003c/sup\u003e. However, a model capable of recapitulating the multicellular periportal liver tissue organization and cellular interactions ex vivo\u0026mdash;while enabling inter-individual comparative studies and investigation of patient-specific disease traits\u0026mdash;has not yet been developed.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHere we overcome this challenge, by establishing long-term expandable human hepatocyte organoids (h-HepOrgs) from adult patient liver tissue and combining them with human cholangiocyte organoids and human portal mesenchyme to form complex periportal liver assembloids. These assembloids recapitulate essential structural and functional features of the native human periportal region and, upon manipulation, model aspects of human biliary fibrosis. Our h-HepOrg model enables long-term expansion while preserving functional drug-metabolizing capabilities and capturing patient-to-patient variability, including differences in metabolic enzymes and disease-associated genes. At both cellular and mesoscale levels, h-HepOrgs mimic fine architectural features such as bile canaliculi morphology and display heterogeneous expression of zonated hepatocyte genes. Although we observed variability in bile canaliculi (BC) morphology among organoids derived from different donors, whether this reflects true patient-to-patient differences will require further investigation. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhen combining human hepatocyte organoids with human portal mesenchyme and human cholangiocyte organoids, the resulting human liver periportal assembloids recapitulate key architectural features of the native periportal region \u003cem\u003ein vitro\u003c/em\u003e, with mesenchymal cells closely associated with cholangiocytes and forming basal contacts, both embedded within a hepatocyte parenchyma. Interestingly, assembloids exhibited increased portal-region functional features. Whether direct interactions between hepatocytes, cholangiocytes, and portal mesenchyme are sufficient to instruct portal-specific hepatocyte identity remains an open question. Likewise, whether hepatocyte subpopulations at the onset of culture influence differential responses to microenvironmental cues cannot be excluded. Our modular assembloid platform provides a unique system to systematically manipulate individual cellular components and begin to dissect, in a controlled human \u003cem\u003ein vitro\u003c/em\u003e setting, how specific microenvironmental signals contribute to human hepatocyte identity and zonation or to epithelial-portal mesenchyme interactions.\u003c/p\u003e\n\u003cp\u003eOf note, by increasing the number of portal mesenchymal cells we generated fibrotic-like assembloids that recapitulated aspects of human cholestatic liver disease and biliary fibrosis, including increased cholangiocyte numbers, reminiscent of \u0026apos;ductular responses\u0026apos; observed in patients with chronic liver diseases\u003csup\u003e73\u003c/sup\u003e. One caveat of the model, though, is that it still lacks the other stromal components, mainly the other mesenchymal cells (HSC and VSCM) as well as the endothelium and immune cells. The lack of portal vasculature (portal vein and hepatic artery) limits the formation of a true periportal triad, as endothelial networks are essential for oxygen delivery and spatial patterning.\u0026nbsp;Incorporating these to generate more complex assembloids models will be crucial to reproduce all aspects of liver disease.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn summary, the patient-derived hepatocyte organoids and periportal assembloid models we present here hold, in our view, the potential to initiate a new era\u0026nbsp;in diverse areas of liver research, including diagnostics, toxicology,\u0026nbsp;personalized\u0026nbsp;drug efficacy screenings and cellular transplantation therapy, opening up new avenues to find therapeutic approaches for cholestatic injury and biliary fibrosis diseases.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eM.H. is supported by the Max Planck Gessellschaft and is recipient of an \u0026quot;Allen Distinguished Investigator Award, a Paul G. Allen Frontiers Group advised grant of the Paul G. Allen Family Foundation, which supports A.L. and R.R.C. This project was partially supported by the European Research Council under the European Union\u0026rsquo;s Horizon Europe research and innovation programme (grant agreement No 101088869) awarded to M.H. Views and opinion expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them. Part of this work was also funded by the LiSyM grant from the Bundesministerium f\u0026uuml;r Bildung und Forschung (BMBF Federal Ministry of Education and Research) awarded to M.H. (031L0258C, and 031L0315B) and to G.D. and D.S. (031L0258E and 031L0315D). This project was also partially supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation, 514150034) and the DFG under Germany\u0026acute;s Excellence Strategy \u0026ndash; EXC-2068 \u0026ndash; 390729961- Cluster of Excellence Physics of Life of TU Dresden. D.E.S. was supported by a German Cancer Aid Max Eder Grant (#70113745). We thank Ms Florida Ahmed for help with chromosome counting. We thank Dr. Julia Jarrells and Ms. Jessica Hernandez (MPI-CBG) for assistance with fluorescence-activated cell sorting (FACS), the light microscopy facility for imaging troubleshooting and training (Dr Jan Peychl and Dr Riccardo Maraspini), the Technology Development Studio facility for the high-throughput imaging and image analysis (Dr Rico Barsacchi and Dr Martin St\u0026ouml;ter) and the Dresden Concept Genome Center for the RNAseq and scRNAseq library (Susanne Reinhardt and Juliane Bl\u0026auml;sche at the DcGC Dresden-concept Genome Center - a core facility of the CMCB and Technology Platform of the TUD (Dresden University of Technology). We thank Ms Jessie P\u0026ouml;che for assistance with hepatocyte isolation and the surgical research laboratory and especially MS Stefanie H\u0026uuml;bner as well as the operative team of the Dept. of Visceral, Thoracic and Vascular Surgery, University Hospital Dresden, for the assistance in liver tissue processing. We thank the whole team of the Department of Hepatobiliary Surgery and Visceral Transplantation, Leipzig University Medical Center for their support in the patient acquisition and the liver tissue logistics. We thank Gerda Schicht for her assistance in the hepatocyte isolations at Leipzig University Medical Center. We thank Dr Michele Marass for insightful comments and discussions on the manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eM.H. designed the study. L.Y., S.D., Y.K., A.L. and R.A-B. performed most of the experiments and, together with M.H., interpreted the results. F.R. performed the scRNAseq analysis. F.R. and D.L.H.T. performed the bulk RNAseq analysis. R.R.C performed the chromosome analysis. A.Sch., A.She, And.She. applied direct infusion mass spectrometry to characterize drug metabolism capacity. F.B., D.E.S. and C.G. assisted with tissue processing and hepatocyte isolation. D.S., G.D. and D.E.S. obtained the patient consent of the tissue samples used in the study. S.D. performed image analysis and bile canaliculi reconstructions with the help from A.S. A.M.D. and A.S. contributed on the first phases of the project. Y.K. and D.C conducted the xenotransplantation experiments with the assistance from S.L.K. L.Y., S.D., Y.K., A.L. and M.H. wrote the manuscript. All authors read and commented on the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eM.H. is an inventor in several patents on organoid technology. Y.K., S.D. and M.H. are inventors in a patent on human hepatocyte organoids. M.H. L.Y, S.D., A.M.D and A.S. are co-inventors in a patent on human assembloids. The remaining authors declare no competing interests.\u003cstrong\u003e\u003c/strong\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eLancaster, M. A. \u0026amp; Huch, M. Disease modelling in human organoids. \u003cem\u003eDis Model Mech\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e (2019). https://doi.org:10.1242/dmm.039347\u003c/li\u003e\n\u003cli\u003eCampana, L., Esser, H., Huch, M. \u0026amp; Forbes, S. Liver regeneration and inflammation: from fundamental science to clinical applications. \u003cem\u003eNat Rev Mol Cell Biol\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 608-624 (2021). https://doi.org:10.1038/s41580-021-00373-7\u003c/li\u003e\n\u003cli\u003eMichalopoulos, G. K. \u0026amp; Bhushan, B. Liver regeneration: biological and pathological mechanisms and implications. \u003cem\u003eNat Rev Gastroenterol Hepatol\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 40-55 (2021). https://doi.org:10.1038/s41575-020-0342-4\u003c/li\u003e\n\u003cli\u003eHuch, M.\u003cem\u003e et al.\u003c/em\u003e Long-term culture of genome-stable bipotent stem cells from adult human liver. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e160\u003c/strong\u003e, 299-312 (2015). https://doi.org:10.1016/j.cell.2014.11.050\u003c/li\u003e\n\u003cli\u003eBroutier, L.\u003cem\u003e et al.\u003c/em\u003e Culture and establishment of self-renewing human and mouse adult liver and pancreas 3D organoids and their genetic manipulation. \u003cem\u003eNat Protoc\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 1724-1743 (2016). https://doi.org:10.1038/nprot.2016.097\u003c/li\u003e\n\u003cli\u003eHuch, M.\u003cem\u003e et al.\u003c/em\u003e In vitro expansion of single Lgr5+ liver stem cells induced by Wnt-driven regeneration. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e494\u003c/strong\u003e, 247-250 (2013). https://doi.org:10.1038/nature11826\u003c/li\u003e\n\u003cli\u003eDevarbhavi, H.\u003cem\u003e et al.\u003c/em\u003e Global burden of liver disease: 2023 update. \u003cem\u003eJ Hepatol\u003c/em\u003e \u003cstrong\u003e79\u003c/strong\u003e, 516-537 (2023). https://doi.org:10.1016/j.jhep.2023.03.017\u003c/li\u003e\n\u003cli\u003eRamachandran, P. \u0026amp; Henderson, N. C. Antifibrotics in chronic liver disease: tractable targets and translational challenges. \u003cem\u003eLancet Gastroenterol Hepatol\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 328-340 (2016). https://doi.org:10.1016/S2468-1253(16)30110-8\u003c/li\u003e\n\u003cli\u003eFriedman, S. L., Neuschwander-Tetri, B. A., Rinella, M. \u0026amp; Sanyal, A. J. Mechanisms of NAFLD development and therapeutic strategies. \u003cem\u003eNat Med\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 908-922 (2018). https://doi.org:10.1038/s41591-018-0104-9\u003c/li\u003e\n\u003cli\u003eRamachandran, P.\u003cem\u003e et al.\u003c/em\u003e Resolving the fibrotic niche of human liver cirrhosis at single-cell level. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e575\u003c/strong\u003e, 512-518 (2019). https://doi.org:10.1038/s41586-019-1631-3\u003c/li\u003e\n\u003cli\u003eXu, J.\u003cem\u003e et al.\u003c/em\u003e A spatiotemporal atlas of mouse liver homeostasis and regeneration. \u003cem\u003eNat Genet\u003c/em\u003e \u003cstrong\u003e56\u003c/strong\u003e, 953-969 (2024). https://doi.org:10.1038/s41588-024-01709-7\u003c/li\u003e\n\u003cli\u003eAizarani, N.\u003cem\u003e et al.\u003c/em\u003e A human liver cell atlas reveals heterogeneity and epithelial progenitors. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e572\u003c/strong\u003e, 199-204 (2019). https://doi.org:10.1038/s41586-019-1373-2\u003c/li\u003e\n\u003cli\u003eMacParland, S. A.\u003cem\u003e et al.\u003c/em\u003e Single cell RNA sequencing of human liver reveals distinct intrahepatic macrophage populations. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 4383 (2018). https://doi.org:10.1038/s41467-018-06318-7\u003c/li\u003e\n\u003cli\u003eBrazovskaja, A.\u003cem\u003e et al.\u003c/em\u003e Cell atlas of the regenerating human liver after portal vein embolization. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 5827 (2024). https://doi.org:10.1038/s41467-024-49236-7\u003c/li\u003e\n\u003cli\u003eAndrews, T. S.\u003cem\u003e et al.\u003c/em\u003e Single-Cell, Single-Nucleus, and Spatial RNA Sequencing of the Human Liver Identifies Cholangiocyte and Mesenchymal Heterogeneity. \u003cem\u003eHepatol Commun\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 821-840 (2022). https://doi.org:10.1002/hep4.1854\u003c/li\u003e\n\u003cli\u003eAndrews, T. S.\u003cem\u003e et al.\u003c/em\u003e Single-cell, single-nucleus, and spatial transcriptomics characterization of the immunological landscape in the healthy and PSC human liver. \u003cem\u003eJ Hepatol\u003c/em\u003e \u003cstrong\u003e80\u003c/strong\u003e, 730-743 (2024). https://doi.org:10.1016/j.jhep.2023.12.023\u003c/li\u003e\n\u003cli\u003eGissen, P. \u0026amp; Arias, I. M. Structural and functional hepatocyte polarity and liver disease. \u003cem\u003eJ Hepatol\u003c/em\u003e \u003cstrong\u003e63\u003c/strong\u003e, 1023-1037 (2015). https://doi.org:10.1016/j.jhep.2015.06.015\u003c/li\u003e\n\u003cli\u003eZhang, K.\u003cem\u003e et al.\u003c/em\u003e In Vitro Expansion of Primary Human Hepatocytes with Efficient Liver Repopulation Capacity. \u003cem\u003eCell Stem Cell\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 806-819 e804 (2018). https://doi.org:10.1016/j.stem.2018.10.018\u003c/li\u003e\n\u003cli\u003eBlidisel, A.\u003cem\u003e et al.\u003c/em\u003e Experimental Models of Hepatocellular Carcinoma-A Preclinical Perspective. \u003cem\u003eCancers (Basel)\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e (2021). https://doi.org:10.3390/cancers13153651\u003c/li\u003e\n\u003cli\u003eKimlin, L. C., Casagrande, G. \u0026amp; Virador, V. M. In vitro three-dimensional (3D) models in cancer research: an update. \u003cem\u003eMol Carcinog\u003c/em\u003e \u003cstrong\u003e52\u003c/strong\u003e, 167-182 (2013). https://doi.org:10.1002/mc.21844\u003c/li\u003e\n\u003cli\u003eFujii, M.\u003cem\u003e et al.\u003c/em\u003e Human Intestinal Organoids Maintain Self-Renewal Capacity and Cellular Diversity in Niche-Inspired Culture Condition. \u003cem\u003eCell Stem Cell\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 787-793 e786 (2018). https://doi.org:10.1016/j.stem.2018.11.016\u003c/li\u003e\n\u003cli\u003eSpence, J. R.\u003cem\u003e et al.\u003c/em\u003e Directed differentiation of human pluripotent stem cells into intestinal tissue in vitro. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e470\u003c/strong\u003e, 105-109 (2011). https://doi.org:10.1038/nature09691\u003c/li\u003e\n\u003cli\u003eMitrofanova, O.\u003cem\u003e et al.\u003c/em\u003e Bioengineered human colon organoids with in vivo-like cellular complexity and function. \u003cem\u003eCell Stem Cell\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 1175-1186 e1177 (2024). https://doi.org:10.1016/j.stem.2024.05.007\u003c/li\u003e\n\u003cli\u003eAloia, L.\u003cem\u003e et al.\u003c/em\u003e Epigenetic remodelling licences adult cholangiocytes for organoid formation and liver regeneration. \u003cem\u003eNat Cell Biol\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 1321-1333 (2019). https://doi.org:10.1038/s41556-019-0402-6\u003c/li\u003e\n\u003cli\u003eCordero-Espinoza, L.\u003cem\u003e et al.\u003c/em\u003e Dynamic cell contacts between periportal mesenchyme and ductal epithelium act as a rheostat for liver cell proliferation. \u003cem\u003eCell Stem Cell\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 1907-1921 e1908 (2021). https://doi.org:10.1016/j.stem.2021.07.002\u003c/li\u003e\n\u003cli\u003ePlanas-Paz, L.\u003cem\u003e et al.\u003c/em\u003e The RSPO-LGR4/5-ZNRF3/RNF43 module controls liver zonation and size. \u003cem\u003eNat Cell Biol\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 467-479 (2016). https://doi.org:10.1038/ncb3337\u003c/li\u003e\n\u003cli\u003ePepe-Mooney, B. J.\u003cem\u003e et al.\u003c/em\u003e Single-Cell Analysis of the Liver Epithelium Reveals Dynamic Heterogeneity and an Essential Role for YAP in Homeostasis and Regeneration. \u003cem\u003eCell Stem Cell\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 23-38 e28 (2019). https://doi.org:10.1016/j.stem.2019.04.004\u003c/li\u003e\n\u003cli\u003eRoos, F. J. M.\u003cem\u003e et al.\u003c/em\u003e Human branching cholangiocyte organoids recapitulate functional bile duct formation. \u003cem\u003eCell Stem Cell\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, 776-794 e713 (2022). https://doi.org:10.1016/j.stem.2022.04.011\u003c/li\u003e\n\u003cli\u003eOber, E. A. \u0026amp; Lemaigre, F. P. Development of the liver: Insights into organ and tissue morphogenesis. \u003cem\u003eJ Hepatol\u003c/em\u003e \u003cstrong\u003e68\u003c/strong\u003e, 1049-1062 (2018). https://doi.org:10.1016/j.jhep.2018.01.005\u003c/li\u003e\n\u003cli\u003eGordillo, M., Evans, T. \u0026amp; Gouon-Evans, V. Orchestrating liver development. \u003cem\u003eDevelopment\u003c/em\u003e \u003cstrong\u003e142\u003c/strong\u003e, 2094-2108 (2015). https://doi.org:10.1242/dev.114215\u003c/li\u003e\n\u003cli\u003eZorn, A. M. in \u003cem\u003eStemBook\u003c/em\u003e (2008).\u003c/li\u003e\n\u003cli\u003eSampaziotis, F.\u003cem\u003e et al.\u003c/em\u003e Cholangiocyte organoids can repair bile ducts after transplantation in the human liver. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e371\u003c/strong\u003e, 839-846 (2021). https://doi.org:10.1126/science.aaz6964\u003c/li\u003e\n\u003cli\u003eHu, H.\u003cem\u003e et al.\u003c/em\u003e Long-Term Expansion of Functional Mouse and Human Hepatocytes as 3D Organoids. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e175\u003c/strong\u003e, 1591-1606 e1519 (2018). https://doi.org:10.1016/j.cell.2018.11.013\u003c/li\u003e\n\u003cli\u003ePeng, W. C.\u003cem\u003e et al.\u003c/em\u003e Inflammatory Cytokine TNFalpha Promotes the Long-Term Expansion of Primary Hepatocytes in 3D Culture. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e175\u003c/strong\u003e, 1607-1619 e1615 (2018). https://doi.org:10.1016/j.cell.2018.11.012\u003c/li\u003e\n\u003cli\u003ePrior, N.\u003cem\u003e et al.\u003c/em\u003e Lgr5(+) stem and progenitor cells reside at the apex of a heterogeneous embryonic hepatoblast pool. \u003cem\u003eDevelopment\u003c/em\u003e \u003cstrong\u003e146\u003c/strong\u003e (2019). https://doi.org:10.1242/dev.174557\u003c/li\u003e\n\u003cli\u003eWesley, B. T.\u003cem\u003e et al.\u003c/em\u003e Single-cell atlas of human liver development reveals pathways directing hepatic cell fates. \u003cem\u003eNat Cell Biol\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 1487-1498 (2022). https://doi.org:10.1038/s41556-022-00989-7\u003c/li\u003e\n\u003cli\u003eHendriks, D.\u003cem\u003e et al.\u003c/em\u003e Mapping of mitogen and metabolic sensitivity in organoids defines requirements for human hepatocyte growth. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 4034 (2024). https://doi.org:10.1038/s41467-024-48550-4\u003c/li\u003e\n\u003cli\u003eNuciforo, S. \u0026amp; Heim, M. H. Organoids to model liver disease. \u003cem\u003eJHEP Rep\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 100198 (2021). https://doi.org:10.1016/j.jhepr.2020.100198\u003c/li\u003e\n\u003cli\u003eMarsee, A.\u003cem\u003e et al.\u003c/em\u003e Building consensus on definition and nomenclature of hepatic, pancreatic, and biliary organoids. \u003cem\u003eCell Stem Cell\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 816-832 (2021). https://doi.org:10.1016/j.stem.2021.04.005\u003c/li\u003e\n\u003cli\u003eThompson, W. L. \u0026amp; Takebe, T. Human liver model systems in a dish. \u003cem\u003eDev Growth Differ\u003c/em\u003e \u003cstrong\u003e63\u003c/strong\u003e, 47-58 (2021). https://doi.org:10.1111/dgd.12708\u003c/li\u003e\n\u003cli\u003eDowbaj, A. M., Kohler, T. N., Cordero-Espinoza, L., Hollfelder, F. \u0026amp; Huch, M. Generation of liver mesenchyme and ductal cell organoid co-culture using cell self-aggregation and droplet microfluidics. \u003cem\u003eSTAR Protoc\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 102333 (2023). https://doi.org:10.1016/j.xpro.2023.102333\u003c/li\u003e\n\u003cli\u003eTanimizu, N.\u003cem\u003e et al.\u003c/em\u003e Generation of functional liver organoids on combining hepatocytes and cholangiocytes with hepatobiliary connections ex vivo. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 3390 (2021). https://doi.org:10.1038/s41467-021-23575-1\u003c/li\u003e\n\u003cli\u003eDamm, G.\u003cem\u003e et al.\u003c/em\u003e Effect of glucose and insulin supplementation on the isolation of primary human hepatocytes. \u003cem\u003eEXCLI J\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 1071-1091 (2019). https://doi.org:10.17179/excli2019-1782\u003c/li\u003e\n\u003cli\u003eKegel, V.\u003cem\u003e et al.\u003c/em\u003e Protocol for Isolation of Primary Human Hepatocytes and Corresponding Major Populations of Non-parenchymal Liver Cells. \u003cem\u003eJ Vis Exp\u003c/em\u003e, e53069 (2016). https://doi.org:10.3791/53069\u003c/li\u003e\n\u003cli\u003eBroutier, L.\u003cem\u003e et al.\u003c/em\u003e Human primary liver cancer-derived organoid cultures for disease modeling and drug screening. \u003cem\u003eNat Med\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 1424-1435 (2017). https://doi.org:10.1038/nm.4438\u003c/li\u003e\n\u003cli\u003eBelenguer, G.\u003cem\u003e et al.\u003c/em\u003e RNF43/ZNRF3 loss predisposes to hepatocellular-carcinoma by impairing liver regeneration and altering the liver lipid metabolic ground-state. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 334 (2022). https://doi.org:10.1038/s41467-021-27923-z\u003c/li\u003e\n\u003cli\u003eHu, M.\u003cem\u003e et al.\u003c/em\u003e Wnt/beta-catenin signaling in murine hepatic transit amplifying progenitor cells. \u003cem\u003eGastroenterology\u003c/em\u003e \u003cstrong\u003e133\u003c/strong\u003e, 1579-1591 (2007). https://doi.org:10.1053/j.gastro.2007.08.036\u003c/li\u003e\n\u003cli\u003eRussell, J. O. \u0026amp; Monga, S. P. Wnt/beta-Catenin Signaling in Liver Development, Homeostasis, and Pathobiology. \u003cem\u003eAnnu Rev Pathol\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 351-378 (2018). https://doi.org:10.1146/annurev-pathol-020117-044010\u003c/li\u003e\n\u003cli\u003eZucman-Rossi, J., Villanueva, A., Nault, J. C. \u0026amp; Llovet, J. M. Genetic Landscape and Biomarkers of Hepatocellular Carcinoma. \u003cem\u003eGastroenterology\u003c/em\u003e \u003cstrong\u003e149\u003c/strong\u003e, 1226-1239 e1224 (2015). https://doi.org:10.1053/j.gastro.2015.05.061\u003c/li\u003e\n\u003cli\u003eJanda, C. Y.\u003cem\u003e et al.\u003c/em\u003e Surrogate Wnt agonists that phenocopy canonical Wnt and beta-catenin signalling. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e545\u003c/strong\u003e, 234-237 (2017). https://doi.org:10.1038/nature22306\u003c/li\u003e\n\u003cli\u003eKastan, N.\u003cem\u003e et al.\u003c/em\u003e Small-molecule inhibition of Lats kinases may promote Yap-dependent proliferation in postmitotic mammalian tissues. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 3100 (2021). https://doi.org:10.1038/s41467-021-23395-3\u003c/li\u003e\n\u003cli\u003eHwang, E. S. \u0026amp; Song, S. B. Possible Adverse Effects of High-Dose Nicotinamide: Mechanisms and Safety Assessment. \u003cem\u003eBiomolecules\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e (2020). https://doi.org:10.3390/biom10050687\u003c/li\u003e\n\u003cli\u003eBebelman, M. P.\u003cem\u003e et al.\u003c/em\u003e Hepatocyte apical bulkheads provide a mechanical means to oppose bile pressure. \u003cem\u003eJ Cell Biol\u003c/em\u003e \u003cstrong\u003e222\u003c/strong\u003e (2023). https://doi.org:10.1083/jcb.202208002\u003c/li\u003e\n\u003cli\u003eSegovia-Miranda, F.\u003cem\u003e et al.\u003c/em\u003e Three-dimensional spatially resolved geometrical and functional models of human liver tissue reveal new aspects of NAFLD progression. \u003cem\u003eNat Med\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 1885-1893 (2019). https://doi.org:10.1038/s41591-019-0660-7\u003c/li\u003e\n\u003cli\u003eMorales-Navarrete, H.\u003cem\u003e et al.\u003c/em\u003e A versatile pipeline for the multi-scale digital reconstruction and quantitative analysis of 3D tissue architecture. \u003cem\u003eElife\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e (2015). https://doi.org:10.7554/eLife.11214\u003c/li\u003e\n\u003cli\u003eNamoto, K.\u003cem\u003e et al.\u003c/em\u003e NIBR-LTSi is a selective LATS kinase inhibitor activating YAP signaling and expanding tissue stem cells in vitro and in vivo. \u003cem\u003eCell Stem Cell\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 554-569 e517 (2024). https://doi.org:10.1016/j.stem.2024.03.003\u003c/li\u003e\n\u003cli\u003eYimlamai, D.\u003cem\u003e et al.\u003c/em\u003e Hippo pathway activity influences liver cell fate. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e157\u003c/strong\u003e, 1324-1338 (2014). https://doi.org:10.1016/j.cell.2014.03.060\u003c/li\u003e\n\u003cli\u003eWatson, B. R.\u003cem\u003e et al.\u003c/em\u003e Spatial transcriptomics of healthy and fibrotic human liver at single-cell resolution. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 319 (2025). https://doi.org:10.1038/s41467-024-55325-4\u003c/li\u003e\n\u003cli\u003eYakubovsky, O.\u003cem\u003e et al.\u003c/em\u003e A spatial transcriptomics atlas of live donors reveals unique zonation patterns in the healthy human liver. \u003cem\u003ebioRxiv\u003c/em\u003e, 2025.2002.2022.639181 (2025). https://doi.org:10.1101/2025.02.22.639181\u003c/li\u003e\n\u003cli\u003eAzuma, H.\u003cem\u003e et al.\u003c/em\u003e Robust expansion of human hepatocytes in Fah-/-/Rag2-/-/Il2rg-/- mice. \u003cem\u003eNat Biotechnol\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 903-910 (2007). https://doi.org:10.1038/nbt1326\u003c/li\u003e\n\u003cli\u003eGuilliams, M.\u003cem\u003e et al.\u003c/em\u003e Spatial proteogenomics reveals distinct and evolutionarily conserved hepatic macrophage niches. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e185\u003c/strong\u003e, 379-396 e338 (2022). https://doi.org:10.1016/j.cell.2021.12.018\u003c/li\u003e\n\u003cli\u003eWells, R. G. The portal fibroblast: not just a poor man\u0026apos;s stellate cell. \u003cem\u003eGastroenterology\u003c/em\u003e \u003cstrong\u003e147\u003c/strong\u003e, 41-47 (2014). https://doi.org:10.1053/j.gastro.2014.05.001\u003c/li\u003e\n\u003cli\u003eLei, L.\u003cem\u003e et al.\u003c/em\u003e Portal fibroblasts with mesenchymal stem cell features form a reservoir of proliferative myofibroblasts in liver fibrosis. \u003cem\u003eHepatology\u003c/em\u003e \u003cstrong\u003e76\u003c/strong\u003e, 1360-1375 (2022). https://doi.org:10.1002/hep.32456\u003c/li\u003e\n\u003cli\u003eDowbaj, A. M.\u003cem\u003e et al.\u003c/em\u003e Mouse liver assembloids model periportal architecture and biliary fibrosis. \u003cem\u003eNature\u003c/em\u003e (2025). https://doi.org:10.1038/s41586-025-09183-9\u003c/li\u003e\n\u003cli\u003eJin, C.\u003cem\u003e et al.\u003c/em\u003e Single-cell RNA sequencing reveals the pro-inflammatory roles of liver-resident Th1-like cells in primary biliary cholangitis. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 8690 (2024). https://doi.org:10.1038/s41467-024-53104-9\u003c/li\u003e\n\u003cli\u003eHammerich, L. \u0026amp; Tacke, F. Hepatic inflammatory responses in liver fibrosis. \u003cem\u003eNat Rev Gastroenterol Hepatol\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 633-646 (2023). https://doi.org:10.1038/s41575-023-00807-x\u003c/li\u003e\n\u003cli\u003eKisseleva, T. \u0026amp; Brenner, D. Molecular and cellular mechanisms of liver fibrosis and its regression. \u003cem\u003eNat Rev Gastroenterol Hepatol\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 151-166 (2021). https://doi.org:10.1038/s41575-020-00372-7\u003c/li\u003e\n\u003cli\u003eReza, H.A.\u003cem\u003e et al.\u003c/em\u003e Multi-zonal liver organoids from human pluripotent stem cells. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e641\u003c/strong\u003e, 1258-1267 (2025). https://doi.org/10.1038/s41586-025-08850-1\u003c/li\u003e\n\u003cli\u003eIgarashi, R.\u003cem\u003e et al.\u003c/em\u003e Generation of human adult hepatocyte organoids with metabolic functions. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e641\u003c/strong\u003e, 1248-1257 (2025). https://doi.org:10.1038/s41586-025-08861-y\u003c/li\u003e\n\u003cli\u003eMarsee, A.\u003cem\u003e et al.\u003c/em\u003e Mass Generation and Long-term Expansion of Hepatobiliary Organoids from Adult Primary Human Hepatocytes. \u003cem\u003ebioRxiv\u003c/em\u003e, 2024.2006.2010.598262 (2024). https://doi.org:10.1101/2024.06.10.598262\u003c/li\u003e\n\u003cli\u003eMallanna, S. K.\u003cem\u003e et al.\u003c/em\u003e Expandable, Functional Hepatocytes Derived from Primary Cells Enable Liver Therapeutics. \u003cem\u003ebioRxiv\u003c/em\u003e, 2024.2012.2028.630269 (2024). https://doi.org:10.1101/2024.12.28.630269\u003c/li\u003e\n\u003cli\u003eWesterfield, A. D.\u003cem\u003e et al.\u003c/em\u003e A 3D \u0026lt;em\u0026gt;in vitro\u0026lt;/em\u0026gt; model of the human hepatobiliary junction. \u003cem\u003ebioRxiv\u003c/em\u003e, 2025.2007.2011.664464 (2025). https://doi.org:10.1101/2025.07.11.664464\u003c/li\u003e\n\u003cli\u003eRoskams, T. \u0026amp; Desmet, V. Ductular reaction and its diagnostic significance. \u003cem\u003eSemin Diagn Pathol\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 259-269 (1998). \u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Material and Methods","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHuman specimens\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll human liver tissues used in this study were obtained after informed consent was obtained from patients undergoing operations at either the Department of Visceral, Thoracic and Vascular Surgery (VTG), University Hospital Carl Gustav Carus Dresden (UKD) or at Leipzig University Medical Center. Informed consent was obtained from all participants. The use of the human samples for this study was approved by the corresponding institutional review boards of either the University Hospital Carl Gustav Carus Dresden (ethical vote BO-EK-57022020, ratified on 2020/03/10) or the Leipzig University Hospital (Ethical vote: registration number 322/17-ek, date 2020/06/10 ratified on 2021/11/30 and registration number 450/21-ek, date 2021/11/21 ratified on 2024/10/04). Four samples (F-PHH1-PHH5) were obtained from Lonza Pharma\u0026amp;Biotech - Bioscience Solutions. All information regarding the human samples (sex, age) is provided in Supplementary Table 1 and 2.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eIsolation of primary human hepatocytes and cholangiocytes\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePrimary human hepatocytes (PHHs) were isolated using a two-step collagenase perfusion method. The human liver tissue received from UKD was perfused with Solution A (composed of 10 mM HEPES, and 2.5 mM EGTA in HBSS) at 39\u0026ordm;C for at least 20 minutes, with a ratio of 15 mL/20 seconds. Subsequently, the perfusion solution was switched to Solution B (containing 100 mM HEPES, 4.8 mM CaCl\u003csub\u003e2\u003c/sub\u003e, and 1 g/L Collagenase P, in HBSS) and perfused at 37\u0026ordm;C for 5-15 min, also at a ratio of 15 mL/20 sec. The digestion process was halted by adding cold Williams\u0026apos; Medium E supplemented with 1% HEPES, 1% GlutaMax, and 1% Penicillin/Streptomycin. The primary human hepatocytes were detached from the tissue by shaking using forceps and combing the cells out of the tissue. Afterwards they were filtered through a 100 \u0026micro;m nylon cell strainer. Cells were then spun at 50G for 5 minutes, and the resulting pellet was resuspended in cold-Williams\u0026rsquo; Medium E supplemented with 1% HEPES, 1% GlutaMax and 1% Penicillin/Streptomycin. The cell suspension was kept cold and centrifuged again at 50G for 5 minutes.\u003c/p\u003e\n\u003cp\u003eFor samples obtained from Leipzig University Hospital the perfusion procedure differed slightly: Solution A [(composed of 10 mM HEPES (Carl Roth), 143 mM NaCl, 6.7 mM KCl, 2.4 mM EGTA, 5mM N-Acetyl-l-cysteine, 11 mM D-Glucose (all provided by Sigma Aldrich) and 32 U/L human Insulin (Eli Lilly) in ddH\u003csub\u003e2\u003c/sub\u003eO (pH 7.4)) at 39\u0026ordm;C with a ratio of 25 mL/min for at least 20 min. Then the perfusion solution was switched to Solution B (composed of 67 mM NaCl, 6.7 mM KCl, 10 mM HEPES, 0.5% BSA, 4.8 mM CaCl\u003csub\u003e2\u003c/sub\u003e x 2H\u003csub\u003e2\u003c/sub\u003eO (all provided by Sigma Aldrich), and 1 g/L Collagenase P (Roche) in ddH\u003csub\u003e2\u003c/sub\u003eO (pH 7.6), diluted 1:2 in Stop Solution (composed of DPBS with Ca\u003csup\u003e2+\u003c/sup\u003e, Mg\u003csup\u003e2+\u003c/sup\u003e (Gibco), supplemented with 16.7% FBS (Merck)) and perfused at 39 \u0026deg;C for 5-15 min at a ratio of 25ml/min. The digestion was stopped by adding cold Stop Solution. Hepatocytes were filtered through a funnel with gauze (Hartmann) and centrifuged at 51G for 5 min. Cell pellets were washed in DPBS with Ca\u003csup\u003e2+\u003c/sup\u003e, Mg\u003csup\u003e2+\u003c/sup\u003e, centrifugated at 51G for 5 min, and resuspended in William\u0026rsquo;s Medium E (supplemented with 10% FBS (Merck), 15mM HEPES, 1mM sodium pyruvate, 1% Penicillin/Streptomycin, 1% MEM NEAA (all provided by Gibco), 1 \u0026micro;g/mL Dexamethasone (Jenapharm), and 32 U/L human insulin (Eli Lilly)). The isolated primary human hepatocytes (PHHs) were shipped overnight in ChillProtec plus\u0026reg; medium (Biochrom).\u003c/p\u003e\n\u003cp\u003eFrozen hepatocytes (F-PHH1-f-PHH5, Supplementary Table 2) commercially available from Lonza were defrosted using Human Hepatocyte Thawing Medium (Lonza) following manufacturer\u0026rsquo;s instructions.\u003c/p\u003e\n\u003cp\u003eThe isolated PHHs preparations (either from fresh tissue from Dresden or Leipzig Hospitals or commercially available frozen hepatocytes) were enriched for both EpCAM-negative (hepatocytes) and EpCAM-positive (cholangiocytes) by Magnetic-activated cell sorting (MACS) using an anti-human CD326 antibody (Biolegend) and Anti-Biotin Microbeads (Ultra Pure, Miltenyi) following manufacturer\u0026rsquo;s instructions. The EpCAM-negative fraction with a viability \u0026gt;50% (Supplementary Table 1) was used to generate hepatocyte organoids as described below (section Hepatocyte organoid culture). The EpCAM-positive fraction, formed by human cholangiocytes, was used to generate human cholangiocyte organoids (h-CholOrg) as described in\u003csup\u003e4,5\u003c/sup\u003e and in section Cholangiocyte organoid culture. A digestion method without perfusion, as the one detailed in Huch et al., 2015\u003csup\u003e4\u003c/sup\u003e , only generated h-CholOrg. HepOrg were not formed under non-perfused protocols.\u003c/p\u003e\n\u003cp\u003eThe complete list of patients used and the comparative between digestion and perfusion are provided in Supplementary Table 1-2.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFlow cytometric validation of PHH purity following MACS enrichment\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFreshly isolated PHHs and MACS-enriched EpCAM-negative PHHs (as described above) were centrifuged at 80g for 5 min. Pellets were resuspended in HBSS containing 2% FBS and incubated on ice for 10 min (blocking). After centrifugation (80g, 5 min), cells were resuspended in HBSS with 1% FBS, stained with EpCAM-Alexa 488 (5 \u0026mu;L/test, Biolegend, Cat. 53-8326-42), and incubated for 45 min on ice. Cells were then washed twice with HBSS containing 1% FBS, centrifuged, and resuspended in 200 \u0026mu;L HBSS with 1% FBS, DAPI (1:1000), and DNase I (1:1000) for flow cytometric analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCholangiocyte organoid culture\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEpCAM-positive cholangiocytes were mixed with Matrigel Growth Factor Reduced (Matrigel, Corning) or Cultrex Basement Membrane Extract 2 (BME2) (Cultrex-RGF Basement Membrane Extract Type 2- BME2\u003csup\u003e\u0026nbsp;\u003c/sup\u003e(AMSBIO) at a 50,000 cells per 50\u0026mu;L/well of 24 well plate and cultured at 37\u0026deg;C and 5% CO\u003csub\u003e2\u003c/sub\u003e in h-CholOrg-EM medium as described in: [AdDMEM/F12 medium containing 1% HEPES, 1% Penicillin/Streptomycin, Glutamax, 1x B27 and 1.25 mM N-acetylcysteine (Sigma) supplemented with 10 nM gastrin (Merck/Sigma), 50 ng/mL hEGF (Peprotech, Peprotech), 10% RSPO1 conditioned medium (homemade), 100 ng/mL FGF10 (Peprotech), 10 mM nicotinamide (Merck/Sigma) and 25 ng/mL HGF (Peprotech)], 5 uM A8301 (Tocris) and 10 uM Forskolin (Tocris #1099). For the first 3-5 days in culture this medium was supplemented with 30% WNT3a conditioned medium (Wnt-CM) (homemade), 25 ng/mL Noggin (Peprotech) and 10 \u0026mu;M ROCK inhibitor (Ri) (Y-27632, Merck/Sigma). The grown cholangiocyte organoids were passaged at a 1:3 ratio once a week as described in\u003csup\u003e5\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHepatocyte Organoid culture\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor hepatocyte organoid cultures, the isolated PHHs (EpCAM-negative fraction) were mixed with Matrigel (Corning) or BME2\u003csup\u003e\u0026nbsp;\u003c/sup\u003e(AMSBIO), and 12,500-50,000 cells were seeded in 50\u0026mu;L Matrigel or BME2 per well in 24 well plates and incubated at 37\u0026deg;C and 5% CO\u003csub\u003e2\u003c/sub\u003e. After Matrigel solidification, culture medium was added. The culture medium was based on AdDMEM/F12 (Invitrogen) supplemented with 1% HEPES, 1% GlutaMax (ThermoFisher), 1% Penicillin/Streptomycin (ThermoFisher), 1X B27 without retinoic acid (Gibco), 1.25 mM N-Acetylcysteine (Sigma), 10 nM Gastrin (Sigma), and the following growth factors: 50 ng/mL hEGF (Peprotech), 15% RSPO1 conditioned media (home-made), 100 ng/mL FGF10 (Peprotech), 100 ng/mL FGF7 (Peprotech), 50 ng/mL HGF(Peprotech), 10 mM Nicotinamide (Sigma, for EM 1 medium only), 2 \u0026micro;M A83-01 (Tocris), 3 \u0026micro;M CHIR99021 (Tocris), 10 \u0026micro;M Y-27632 (Tocris), 0.5 nM Wnt Surrogate FC Fusin Protein as in Janda \u003cem\u003eet al.\u003c/em\u003e\u003csup\u003e50\u003c/sup\u003e(IPA, N001), and 10 \u0026micro;M TRULI (Axon) or TDI-011536 (Selleckchem).\u003c/p\u003e\n\u003cp\u003eAfter one week to ten days, the organoids were removed from the Matrigel, mechanically dissociated into small fragments using TryplE\u003csup\u003eTM\u003c/sup\u003e Express (Gibco), and transferred to fresh Matrigel. Passaging was performed once per week at a 1:2 split ratio for at least three months. For preparation of frozen stocks, the organoid cultures were dissociated, mixed with Recovery cell culture freezing medium (Gibco), and frozen following standard procedures.\u003c/p\u003e\n\u003cp\u003eFor the optimization of culture conditions, media component screening experiments were performed where each of the components Amphiregulin (AREG, 100 ng/mL, R\u0026amp;D systems; Dexamethasone (1.6 \u0026micro;M, Sigma); G-CSF (50 ng/mL, R\u0026amp;D systems), IL-6 (2 ng/mL, R\u0026amp;D systems), M-3m3FBS (Phospholipase C activator, 25 \u0026micro;M, Tocris), TGFa (100 ng/mL), TRULI (Axon) was added to our previously published mouse hepatoblast medium (MM\u003csup\u003e35\u003c/sup\u003e) with minor modification: AdDMEM/F12 (Invitrogen) supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, 1X B27 without retinoic acid, 1.25 mM N-Acetylcysteine, 10 nM Gastrin, 50 ng/mL hEGF, 15% RSPO1 conditioned media, 100 ng/mL FGF10, 100 ng/mL FGF7, 50 ng/mL HGF, 10 mM Nicotinamide, 2 \u0026micro;M A83-01, 3 \u0026micro;M CHIR99021, 10 \u0026micro;M Y-27632, and 0.5 nM Wnt Surrogate FC Fusion Protein. Note that the addition of TRULI alone resulted in a significant increase in organoid formation efficiency (Fig.1c,e). However, after 1-2 splits, the cultures rapidly deteriorated and could not be expanded further (Fig.1f).\u003c/p\u003e\n\u003cp\u003eFor h-HepOrgs hepatic differentiation, hHepOrg were expanded in EM2 medium, split, seeded and cultured for 2-5 days under EM1 culture medium after which medium was changed to hepatic differentiation medium (DM) composed of: AdDMEM/F12 supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, 1X B27 without retinoic acid, 1.25 mM N-Acetylcysteine, 50 ng/mL hEGF, 15% RSPO1 conditioned media, 50 ng/mL HGF, 2 \u0026micro;M A83-01, 3 \u0026micro;M CHIR99021, 10 \u0026micro;M Y-27632, 0.5 nM Wnt Surrogate FC Fusion Protein, 100 ng/mL FGF19 (R\u0026amp;D systems), and 1.6 \u0026micro;M Dexamethasone (Sigma). Hepatic differentiation medium (DM) was changed every 2-3 days for 7 days.\u003c/p\u003e\n\u003cp\u003eFor organoid formation efficiency, primary hepatocytes were isolated and cultured in different media as described above. To prevent organoids from fusing, 25,000 (for EM2 medium) or 50,000 (all other media) viable hepatocytes (viability \u0026gt;80%) were plated in 50\u0026mu;L Matrigel\u003csup\u003e\u0026nbsp;\u003c/sup\u003eor BME2 and cultured as described above. After 12-14 days, organoid numbers were counted and results expressed as a percentage relative to the initial seeding cell numbers.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eIsolation of human liver portal fibroblasts\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHuman liver portal fibroblasts (hPFs) were isolated from human liver tissues by collagenase digestion. Briefly, human liver tissue was minced and rinsed with cold-DMEM (Gibco) supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 1% FBS. Minced tissues were incubated with a collagenase solution consisting of 2.5 mg/mL Collagenase D (Roche), 0.1 mg/mL DNase I (Sigma), and 10 \u0026micro;M Y-27632 in DMEM supplemented with 1% HEPES, 1% GlutaMax, and 1% Penicillin/Streptomycin. The incubation was carried out for 30-60 minutes at 37\u0026ordm;C on a shaker set at 120 rpm. The digestion was halted by adding cold-DMEM supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 1% FBS. The suspension was then filtered through a 70 \u0026micro;m cell strainer and centrifuged for 5 minutes at 300G. After removing the supernatant, the cell pellet was resuspended in cold-DMEM supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 1% FBS. The suspension was centrifuged again for 5 minutes at 300G, and the resulting pellet was resuspended in cold-DMEM supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 20% FBS. For sorting, hPFs were stained with 1 \u0026mu;g/test Anti-human CD90 (THY1)-APC, 20 \u0026mu;L/test Anti-human CD140a (PDGFRa)-PE, Anti-CD11b/CD31/CD45-PECy7, and EpCAM-Alexa 488 for 30 minutes on ice and washed twice. THY1-positive hPFs were sorted using a BD FACSAria Fusion and cultured in DMEM supplemented with 1% HEPES, 1% GlutaMax, 1% Penicillin/Streptomycin, and 20% FBS at 37\u0026deg;C and 5% CO\u003csub\u003e2\u003c/sub\u003e until used for assembloid formation or freeze for biobanking.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eVirus infection\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor portal fibroblast infections, cultures (passage 0-1) grown in DMEM\u003csup\u003e+++\u003c/sup\u003e with 20% foetal bovine serum (FBS) (Merck/Sigma, #F7524) were washed with PBS and dissociated to single cells by incubating with TrypLE 1x for 6min at 37\u0026deg;C. And the cell concentration was determined by manual counting in haemocytometer, 10,000 cells were plated into 48-well plates and the medium mixed with nRFP-lenti-virus or nGFP-lenti-virus (LVP360-R and LVP360-G, GenTarget Inc) was replaced after 12h and the solution was changed after 72h.\u003c/p\u003e\n\u003cp\u003eFor cholangiocyte organoid infection, duct cells (phase 0-1) were extracted from the Matrigel and digested with TrypLE to prepare single-cell suspensions as described in Broutier \u003cem\u003eet al.\u003c/em\u003e, 2016\u003csup\u003e5\u003c/sup\u003e, which were then manually counted using a haemocytomer counter to determine cell concentration. In a 48-well plate, 150 ul of cells and 50 ul of virus suspension were added to achieve an MOI= 10-30, mixed thoroughly, centrifuged at 600g for 60 minutes at 32\u0026deg;C, and incubated for 6 hours at 37\u0026deg;C, 5% CO\u003csub\u003e2\u003c/sub\u003e. Cells were collected in 1. 5 ml tubes, centrifuged at 600g for 5 minutes, infected medium was discarded, and cells were resuspended in 25 ul of Matrigel, followed by the addition of EM medium (supplemented with 30% WntCM, 25ng/ml noggin and 10 \u0026micro;M Y-27632 for the first 3 days).\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003ePeriportal assembloids\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo generate liver periportal assembloids comprising hepatocytes, cholangiocytes, and portal fibroblasts, we first prepared the cellular components as follows: nGFP-labelled cholangiocyte organoids (passage 5-11), grown in cholangiocyte expansion medium (h-CholOrg-EM) as detailed above, were collected from Matrigel using cold AdDMEM/F12 (ThermoFisher, #15630-056) containing 1% HEPES (ThermoFisher, #15140-122), 1% penicillin/streptomycin (ThermoFisher, #15140-122), and 1% glutamine (ThermoFisher, #350-068). Matrigel was removed and organoids were dissociated to single cell using pre-warmed TrypLE 1x (ThermoFisher, #12605010) for 7-12 minutes at 37\u0026deg;C. nRFP-labelled portal fibroblasts cultures (passage 5-12) grown in DMEM\u003csup\u003e+++\u003c/sup\u003e with 20% foetal bovine serum (FBS) (Merck/Sigma, #F7524) were washed with PBS and dissociated to single cells by incubating with TrypLE 1x for 6min at 37\u0026deg;C. Both single-cell suspensions were spun at 200 RCF for 5 minutes, resuspended in DM medium described above but without A8301, and then manually counted with a hemocytometer to determine cell concentration. Cultured HepOrg from EM2 were split and transferred to EM1 for 2d and then to DM medium for 3d. Hepatocyte organoids were then collected and washed using cold AdDMEM/F12 supplemented with 1% HEPES, 1% Pen/Strep and 1% Glutamax and incubated for 10min on ice using cold Cell Recovery Solution (Corning, #354253) to remove the ECM. HepOrg were then resuspended using DM without A8301 and placed into low-attachment 6 well plate and differentiated organoids (with bubbly morphology) were selected and hand-picked under a stereomicroscope.\u003c/p\u003e\n\u003cp\u003eTo define an approach for human periportal liver assembloid formation several iterations were performed. First, we sought to identify a medium that would support assembloid formation. i.e., the culture of all three cell types: hepatocytes, cholangiocytes/ductal cells and human portal mesenchyme without overgrowth of any of them, we tested several media and found that a minor adaptation of the differentiation medium used for h-HepOrgs DM without A8301 (assembloids medium) supported the culture of the three cell types while preventing their overgrowth. To determine the optimal quantities of the three cell types required for periportal assembloid formation, we first investigated the proportions of portal fibroblast and ductal cells in healthy human periportal liver tissue. We observed that the ratio varies donor-to-donor from 1:1 to 4:1 ductal cells per fibroblasts. Therefore, we tested these range of ratios \u003cem\u003ein vitro\u003c/em\u003e by varying the proportions of mesenchyme and ductal cells that were mixed with 1 single HepOrg (~200 um diameter). In short, in 96-well low-attachment U-plates (cat n Corning, cat. n: #7007), we assembled (as described below) 1 hepatocyte organoid (h-HepOrgs) with 25 portal fibroblasts and 25 or 50 or 100 or 200 cholangiocytes, or with 100 cholangiocytes and 50 or 100 portal fibroblast cells. We selected the proportion 25 portal fibroblasts: 100 cholangiocytes/ductal cells. In AggreWell\u003csup\u003eTM\u003c/sup\u003e plates (Aggrewell 800, Stem Cell Technologies #34811), we scaled up proportionally taking into account that Agrewell 800 has 300 microwells in each well and used 7,500 portal fibroblasts, 30,000 cholangiocytes, and 100 hepatocyte organoids (proportion 1HepOrg :75 PFs: 300Chol).\u003c/p\u003e\n\u003cp\u003eFor non-healthy/non-physiological ratios, use 500 portal fibroblasts, 100 cholangiocytes, and 1 hepatocyte organoid for 96-well low-attachment U-plates, and 15,0000 portal fibroblasts, 30,000 cholangiocytes, and 50 hepatocyte organoids for Aggrewell plates.\u003c/p\u003e\n\u003cp\u003eFor the assembly in MW96 we mixed fibroblasts and cholangiocytes in 96-well low adhesion U-plates using 150 \u0026mu;L of DM (without A8301) medium with 2.4mg/ml methylcellulose (MeC, Sigma, #M6385) and spun at 50 RCF for 5min. Individual HepOrgs were then added to the well and the mixture was incubated for 18-24h at 37\u0026deg;C and 5% CO\u003csub\u003e2\u003c/sub\u003e. For the assembly in AggreWell\u003csup\u003eTM\u003c/sup\u003e, plates were first pre-treated as recommended by the manufacturer. Then, ductal and mesenchymal cells and HepOrgs were mixed in 1.5 mL of DM (without A8301) medium with 2.4mg/ml methylcellulose, spun down 5min at 50 RCF and incubated for 18-24h at 37\u0026deg;C and 5% CO\u003csub\u003e2\u003c/sub\u003e. After 18-24h in suspension in the 96well/aggrewell plate, the cell suspension was collected with a 1 mL pipette and transferred to a low-attachment 6-well plate. The structures were handpicked under a stereomicroscope and seeded in 25 \u0026mu;L matrigel dome in pre-warmed 48-well plates. The Matrigel was allowed to solidify for 30min at 37\u0026deg;C 5% CO\u003csub\u003e2\u003c/sub\u003e and the wells were overlayed with further 300 \u0026mu;L of DM (without A8301) medium. Medium was changed every 3-4 days. Under these conditions, 70% of the initial Cholangiocytes formed a lumen. Raw data were incorporated into the quantification of periportal-like spatial organization in assembloids (Source data file from Extended Data Fig. 7e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eImmunostaining of organoids and assembloids\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor immunofluorescence staining, organoids and assembloids were first extracted from Matrigel with ice-cold Cell Recovery solution and then fixed for 30 min with 4% paraformaldehyde (PFA) at 4\u0026deg;C. Fixed organoids were washed and transferred to \u0026micro;-Slide 8 Well Chamber Slide (glass bottom, Ibidi). Blocking and permeabilization was performed for 1 hour at RT in PBS containing 2% BSA and either 0.1%, 0.2%, 0.5 % or 1% Triton X-100 depending on antigen (see Supplementary dataset_5). The samples were incubated with primary antibodies overnight at 4\u0026deg;C in blocking solution. After that, the antibody was washed with 3 washes with PBS and the samples were incubated overnight at 4\u0026deg;C or for 8h at RT with secondary antibodies diluted in blocking solution and, if required, also Phalloidin and DAPI were added to the secondary antibody mix. The samples were washed 3 times with PBS and subsequently cleared using Fructose-Glycerol clearing solution (25mL Glycerol, 5.3mL dH2O and 22.5g Fructose \u0026ndash; 60% Glycerol and 2.5M Fructose). The samples were stored in PBS until cleared for imaging as described above. The antibodies and dilutions used are listed in Supplementary Dataset_5.\u003c/p\u003e\n\u003cp\u003eFor H\u0026amp;E staining, organoids were collected in cold DPBS (Gibco) and fixed with 4% PFA for 30 min and dehydrated and embedded in paraffin using standard methods. Paraffin sections (8 \u0026mu;m) were cut and stained for H\u0026amp;E using standard protocols.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eImmunostaining of thin and thick tissue sections\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor thin tissue sections (8-12 \u0026mu;m) and staining, human liver tissues were fixed in 10% formalin overnight whilst rolling at 4\u0026deg;C. After fixation, fixed tissues are washed with PBS and incubated with 10% sucrose for 1-2h, then transferred to 30% sucrose in PBS for 24h and subsequently embedded in OCT compound (VWR, #361603E) to generate OCT-cryopreserved tissue blocks. Tissue blocks were cryo-sectioned on ThermoScientific CryoStar NX70 cryostat. Sections were blocked in PBS with 10% Donkey Serum (DS) and 0.1% Triton X-100 for 2h at RT, incubated with primary antibodies diluted in PBS with 3% Donkey Serum and 0.1% Triton X-100 overnight at 4\u0026deg;C and subsequently washed and incubated with secondary antibodies diluted in 0.05% PBS- BSA and DAPI for 2h at RT. Sections were mounted in Vectashield. The list of used antibodies is available in Supplementary Dataset_5.\u003c/p\u003e\n\u003cp\u003eFor thick tissue sections and staining, the protocol from (Fabi\u0026aacute;n Segovia-Miranda \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e54\u003c/sup\u003e was used. Immediately after surgical resection, the liver tissue samples were cut into smaller pieces and fixed in 4% PFA for 24 h on a rotator at 4\u0026deg;C, washed 3 times with PBS, followed by quenching with 50mM ammonium chloride solution (NH\u003csub\u003e4\u003c/sub\u003eCl) for 24h and again washed 3 times with PBS. For storage, liver pieces were kept in PBS at 4\u0026deg;C. For sectioning, livers were embedded in moulds with 4% low-melting agarose (Bio-Rad #1613111) in PBS and cut into 50 or 100 \u0026mu;m-thick sections on a vibratome (Leica VT1200S). For deep tissue imaging, if antigen retrieval was required, tissue sections were placed in Eppendorf tubes with pre-warmed 1X citrate buffer (Sigma-Aldrich, #C9999), pH=6, at 80C for 30 min in a shaking heating block, and then washed 3 times with PBS. Tissue sections were permeabilized with 0.5% Triton X-100 in PBS for 1h at RT. The primary antibodies were diluted in Tx buffer (0.2% gelatin, 300 mM NaCl, and 0.3% Triton X-100 in PBS) and incubated for 48 h at RT. After washing 3\u0026times;15 min with 0.3% Triton X-100 in PBS, the sections were incubated with secondary antibodies, DAPI (1 mg/mL; 1:1000) and phalloidin for another 48 h. After washing 3\u0026times;15 min with 0.3% TritonX-100 in PBS and 3\u0026times;1 min with PBS, the optical clearing started by incubating the slices in 25% fructose for 4 h, continued in 50% fructose for 4 h, 70% fructose overnight, 100% fructose (100% wt/vol fructose, 0.5% 1-thioglycerol, and 0.1 M phosphate buffer, pH 7.5) overnight, followed by a final overnight incubation in SeeDB solution (80.2% wt/wt fructose, 0.5% 1-thioglycerol, and 0.1 M phosphate buffer)\u003csup\u003e80\u003c/sup\u003e. The samples were mounted in SeeDB. The list of used antibodies and dyes is available in Supplementary Dataset_5.\u003c/p\u003e\n\u003cp\u003eFor immunohistochemistry of xenotransplant mice tissue sections, the mice liver tissue samples were cut into smaller pieces and fixed in 10% formalin overnight. Sections (4 \u0026mu;m) were subjected to immunohistochemical staining, which was performed using a Dako REAL EnVision Detection System (Dako, #K5007). Anti-hGAPDH antibody (Abcam, #ab128915) was used as the primary antibody and nuclei were counterstained with hematoxylin. Stained tissues were viewed under a Virtual Slide System (Leica, ScanScope CS2).\u003c/p\u003e\n\u003cp\u003eThe immunohistochemistry analysis for PDGFRA, DCN and ASPN in human healthy liver tissue was obtained from the publicly available image dataset from Human Protein Atlas (HPA)\u003csup\u003e81\u003c/sup\u003e (version#24proteinatlas.org). The corresponding URL is indicated in the figure legend.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eImaging of organoids, assembloids and tissues\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBright field images of organoids were obtained with a Leica DMIL LED inverted microscope and Leica DFC 450C camera or with a Leica M80 stereoscope and MC170HD camera and Leica LAS software. H\u0026amp;E staining of organoids were obtained with a Leica DM4B microscope and DMC5400 camera and Leica LAX software.\u003c/p\u003e\n\u003cp\u003eConfocal images of organoids and thick tissue sections were acquired on an inverted single photon point scanning confocal microscope (Zeiss Cell Discoverer 7 with LSM 900 and Airyscan 2) using a Zeiss APOCHROMAT 20x/0.95 Autocorr air objective, with a tube lens of 0.5x or 1x, and a voxel size of 0.4 x 0.4 x 0.5 \u0026mu;m or 0.5 x 0.5 x 0.5 \u0026mu;m for organoids and 0.3 x 0.3 x 0.3 \u0026mu;m for thick tissue sections. Laser lines 405, 488, 561 and 640 were used for excitation of fluorophores and GaAsP-PMT detectors were used for detection. High-resolution Airyscan images were acquired using this system for imaging polarity in detail for the tissue sections with a a voxel size of 0.0823 x 0.0823 x 0.3 \u0026mu;m. Image processing was done using ZEN software or ImageJ/Fiji.\u003c/p\u003e\n\u003cp\u003eImaging of assembloids and thin tissue sections was performed using an inverted multiphoton laser-scanning microscope (Zeiss LSM 780 NLO). In order to improve the resolution, image denoising was performed with deconvolution using HuygensPro. Raw image stacks were imported into the software, and a point spread function (PSF) was either estimated based on the imaging conditions (numerical aperture, wavelength, and refractive index) or obtained from PSF calibration images. The HuygensPro Classic Maximum Likelihood Estimation (CMLE) algorithm was applied for deconvolution, with an iteration stop criterion based on optimal signal-to-noise ratio and minimal change in successive iterations.\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eImage analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe quantification of the percentage of YAP-positive nuclei and YAP-negative nuclei was performed using a custom-made pipeline using the Arivis 4D Pro, software (Version 4.2.0). The steps included were background correction, denoising, nuclear segmentation based on DAPI, and quantification of the fluorescence intensity of YAP immunofluorescent staining within the nuclei. The total number of nuclei and the number of YAP-positive nuclei were quantified, and subsequently, the number of YAP-negative nuclei was calculated by subtracting the number of YAP-positive nuclei from the total number of nuclei. Finally, the percentage of YAP-positive and YAP-negative nuclei were calculated.\u003c/p\u003e\n\u003cp\u003eFor the quantification of cytoplasmic to nuclear area, a custom-made pipeline was developed using the Arivis 4D Pro, software (Version 4.2.0). For this, a representative 2D z-slice was taken from each organoid. Pre-processing steps included background correction on the Phalloidin channel (marking cell borders) and normalization and denoising on the DAPI channel (marking nuclei). To obtain the nuclear area, nuclear segmentation was done based on DAPI, followed by quantification of the total nuclear area. For the cytoplasmic area, segmentation was done based on Phalloidin to obtain the outline of the area occupied by the cytoplasm. Finally, the ratio of cytoplasmic area to nuclear area was calculated.\u003c/p\u003e\n\u003cp\u003eFor 3D visualization of bile canaliculi, high-resolution images obtained as described above. Segmentation was performed on CD13 (for bile canaliculi) and F-actin (cell borders) staining with phalloidin. The analysis of bile canaliculi morphology and bile canaliculi network properties was performed using a custom-made Fiji script publicly available at https://github.com/JulienDelpierre/BileCanaliculiSegmentation. The script description can be found in Dowbaj, Sljukic \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e64\u003c/sup\u003e. Briefly, immunofluorescence images from several conditions were used in this analysis: EM2, DM and liver tissue, from hereon, they will be referred as \u0026ldquo;structure\u0026rdquo;. We will refer to individual BC network as \u0026ldquo;network\u0026rdquo;. We determined the connectivity of the network by analysing the total number of branching points (number of triple junctions) per structure. We determined the length of the network per structure by analysing the total length of all branches in the structure. To compare between structures of different conditions we plotted these values as dot plot where each dot is one structure. In the case of tissue each dot is one field of view. The extracted features from Fiji were exported as \u003cem\u003e.csv\u003c/em\u003e files and plotted using Prism.\u003c/p\u003e\n\u003cp\u003eFor assembloids, to visualize the structure from different angles, immunofluorescent images were visualized in 3D using MotionTracking (http://motiontracking.mpi-cbg.de)\u003csup\u003e55\u003c/sup\u003e. For this, Gaussian blurring was applied to the channels of interest and then visualized in 3D.\u003c/p\u003e\n\u003cp\u003eFor quantification of cholangiocyte and portal fibroblasts in assembloids, custom-made pipelines in Arivis 4D software (Zeiss) was used. Nuclei were segmented based on diameter, probability threshold, and split sensitivity to align with the expected morphology in the fluorescence images. When segmentation was incomplete due to weak fluorescence signals, missing nuclei were manually added. This approach was utilized to determine the number of nuclei per cell and the number of cells per organoid. All segmentation results were manually reviewed and corrected as necessary.\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eIsolation of mRNA and RT-qPCR analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRNA was extracted from organoid cultures or freshly isolated tissue using the RNeasy Mini RNA Extraction Kit (Qiagen) with DNAse treatment and reverse-transcribed using reverse-transcribed using Moloney Murine Leukemia Virus reverse transcriptase (Promega). All targets were amplified (40 cycles) using gene-specific primers (Key Resource Table) and PowerUp\u003csup\u003eTM\u003c/sup\u003e SYBR Green master mix (ThermoFisher) or MiIQ syber green (Bio-Rad) and run on a qPCR instrument Thermo Fisher QuantStudio 7 Pro or GeneAmp PCR System 9700; Applied Biosystems respectively. Data were analyzed using Design \u0026amp; Analysis 2.7.0 software (ThermoFisher).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eKaryotyping\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMitotic metaphases for karyotyping were obtained by subculturing hepatocyte organoids in the active growth phase. The following day, cells were exposed to 0.2 \u0026mu;g/mL colcemid (Gibco) for 60 minutes at 37\u0026deg;C to arrest them in metaphase. Organoids were dissociated into single cells using TryplE\u003csup\u003eTM\u003c/sup\u003e Express (Gibco). After centrifugation and removal of the supernatant, cells were subjected to hypotonic treatment with a solution of 0.075 M KCl for 30 minutes at 37\u0026deg;C, followed by fixation in a 3:1 methanol-acetic acid solution. The preparation was washed three times with the fixative before slide preparation. Chromosomes were stained with Giemsa (Merck) diluted in Gurr buffer (pH 6.8; Gibco). Images were taken with a Zeiss Axio Imager.Z2 upright motorized stand with an ApoTome.2 for improved z-contrast.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFunctional assays\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor functional assays, hepatocyte organoids were cultured in expansion and differentiation medium as described above. As negative controls we used cholangiocyte organoids grown as described above. As positive controls we used freshly isolated primary human hepatocytes cultured in standard 2D-hepatocyte monolayer culture or in sandwich culture \u003csup\u003e82\u003c/sup\u003e. Briefly, fresh isolated PHH were plated onto collagen (1.8 mg/mL, RatCol\u003csup\u003eTM\u003c/sup\u003e collagen, Advanced Biomatrix) coated 24-well plates at 500,000 or 250,000 cells/well in Williams E medium (PAN Biotech), supplemented with 10% FBS, penicillin/streptomycin and 100 nM Dexamethasone for 3 hours for attachment. For the monolayer culture (1d-PHH monolayer control), the cells were cultured on Williams\u0026rsquo; E medium supplemented with 1% HEPES + 1% GlutaMax + 1% Penicillin/Streptomycin and 100nM Dexamethasone for 18h (or 24h, for Albumin assay) and then processed for the functional assays. For sandwich culture, fresh isolated PHH were plated onto collagen as above and overlayed with second collagen layer (1.2 mg/mL, RatCol\u003csup\u003eTM\u003c/sup\u003e collagen, Advanced Biomatrix) and cultured for 7 days in Williams\u0026rsquo; E medium supplemented with CM4000 cell maintenance supplement (Thermo Fisher Scientific).\u003c/p\u003e\n\u003cp\u003eTo determine albumin secretion, supernatant from 24 hours was collected and the amount of albumin was determined using a human specific Albumin ELISA kit (Assay Pro) following manufacturer\u0026rsquo;s instructions on an ELISA plate reader (Tecan Spark 20M). To measure Cytochrome P450 activity, on the day of the experiment cholangiocyte and hepatocyte organoids in EM2 or DM were removed from Matrigel using Cell Recovery solution (Corning). Then organoids, 2D-hepatocyte monolayer, or 2D-sandwich cultures were all cultured in Williams\u0026rsquo; E medium supplemented with 1% HEPES + 1% GlutaMax + 1% Penicillin/Streptomycin supplemented with the Luciferin-H substrate (100 \u0026micro;M) or Luciferin-IPA (3 \u0026micro;M) for 6 hours. Cytochrome activity was measured using the P450-Glo Assay Kit (Promega) according to manufacturer\u0026rsquo;s instructions on a plate reader (PerkinElmer Envision). Results were normalized to total viable cell counts per well.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eUrea synthesis assay\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo determine the urea secretion, supernatants of cell culture were collected from 48-well plate after 12-hour culture. The concentration of secreted urea was measured by urea assay kit (Abnova, KA1652) according to the manufacturer\u0026rsquo;s instructions.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eMeasurement of Gluconeogenesis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGluconeogenesis was assessed using a Glucose-Glo\u0026trade; Assay (Promega, J6021). Organoids/ assembloids were first washed twice with PBS to remove residual glucose and then incubated for 24 hours in glucose-free medium (Gibco, A2494301) to deplete intracellular glucose stores. Subsequently, the organoids were stimulated for 24 hours in gluconeogenesis-inducing medium (glucose-free medium supplemented with 10 mM lactate) to suppress glycolysis and promote hepatic glucose production.\u003c/p\u003e\n\u003cp\u003eAfter incubation, 25 \u0026micro;L of supernatants from each well was transferred to a 96-well assay plate and mixed with an equal volume of Glucose Detection Reagent. Following a 60-minute incubation at 37\u0026deg;C, luminescence was measured using a luminometer.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eCell Counting\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHepatocyte organoids were dissociated into single cells using 10X TrypLE (Gibco, A12177-01) after 10 and 15 days of culture in specified media. Cell counts were determined using a Countess\u0026reg; II FL Automated Cell Counter (Thermo Fisher Scientific).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eQuantification of xenobiotic metabolism by mass spectrometry\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHepatocyte organoids were cultured in differentiation medium as previously described. Assembloids were maintained under the same conditions for 6 days. Freshly isolated primary human hepatocytes (PHHs) were cultured in a monolayer for 24 hours, also as described above. Following culture, all cells were washed twice with PBS. The medium was then replaced with 100 \u0026mu;L of Williams\u0026rsquo; E medium supplemented with 1% HEPES, 1% GlutaMAX, 1% Penicillin/Streptomycin, and verapamil (Merck; V-002-1ML) at the final concentration of 4 \u0026micro;M. Cells were incubated for 6 hours, after which the supernatant was collected and analyzed by mass spectrometry.\u003c/p\u003e\n\u003cp\u003eOrganoids and assembloids were dissociated into single cells using 10\u0026times; TrypLE and manually counted using a hemocytometer. The resulting cells were washed twice with PBS and stored at \u0026ndash;20\u0026deg;C.\u003c/p\u003e\n\u003cp\u003eMetabolites were separately extracted from the supernatant and from the cells by isopropanol: methanol : chlorophorm mixture (4 : 2: 1, v/v/v) containing 7.5 mM ammonium formate (termed MS mix). Supernatant aliquot of 100 \u0026mu;l was 20-fold (v /v) diluted with MS mix, vortexed, centrifuged for 7 min at 13g and the pellet was discarded. Cells suspended in 100 \u0026mu;L PBS were first lysed using \u003cem\u003eca\u003c/em\u003e 25 stainless steel beads of 0.5 mm size (Next Advance, USA, Cat N 152034) in the Qiagen Ratsch Tissue Lyser at 30 Hz for 8 min and metabolites wereextracted as above. Each sample was prepared in three biological replicates and analyzed by mass spectrometry immediately after extraction.\u003c/p\u003e\n\u003cp\u003eMass spectrometric analysis was performed on a Q Exactive hybrid quadrupole Orbitrap tandem mass spectrometer (ThermoFischerScientific, USA) in positive ion mode by the direct infusion of total extracts. Prior analyses, the internal standard verapamil-\u003csup\u003e13\u003c/sup\u003eC3 hydrochloride (Merk, Germany, V-079-1ML) was dissolved in methanol and spiked into samples to the final concentration of 200 nM. 40 \u0026mu;L aliquots of each sample were then placed on twin.tech PGR Plate 96 (Eppendorf, Germany, Cat N 0030128.648) and infused into the mass spectrometer via TriVersa NanoMate robotic ion source (Advion Interchim Scientific, USA) using nanoflow chips with the nozzle diameter of 4.1 \u0026mu;m. The ion source was controlled by Chipsoft 8.1.0 software. Spraying voltage and gas backpressure were set to 1.25 kV and 0.95 psi, respectively. Ion transfer capillary temperature was set to 200 \u0026deg;C and S-lens RF level to 50%. Target mass resolution R\u003csub\u003em/z\u003c/sub\u003e \u003csub\u003e=200\u003c/sub\u003e was set to 140 000 (full width at half maximum, FWHM) for both FT MS and FT MS/MS spectra. For acquiring FT MS spectra automated gain control (AGC) was set at the value of 3 \u0026times; 10\u003csup\u003e6\u003c/sup\u003e; maximum injection time 500 ms; acquired mass range \u003cem\u003em/z\u003c/em\u003e 50-700; lock masses \u003cem\u003em/z\u003c/em\u003e 445.12003 and \u003cem\u003em/z\u003c/em\u003e 338.34174. The acquisition cycle consisted of recording MS1 spectra for 1.2 min followed by two MS2 spectra for 1.8 min from the precursors with \u003cem\u003em/z\u003c/em\u003e 455.291 (for verapamil; [M+H]\u003csup\u003e+\u003c/sup\u003e) and \u003cem\u003em/z\u003c/em\u003e 441.275 (for norverapamil [M+H]\u003csup\u003e+\u003c/sup\u003e); precursor \u003cem\u003em/z\u003c/em\u003e isolation width was 3 Da.\u003c/p\u003e\n\u003cp\u003eSpectra were averaged in Xcalibur Qual Browser v.3.0 (ThermoFischerScientific, USA) over 30 sec time range corresponding to stable spray; peaks of metabolites and standard extracted with 5 ppm mass accuracy. Absolute amount of norverapamil was calculated from its molecular ion intensity normalized to the intensity of the standard. For calibrating, aliquots of Williams E medium containing verapamil (Merk, Germany, Cat N V-002-1ML) with the concentration ranging from 2 \u0026mu;M to 8 nM were diluted 20-fold with MS mix, spiked with the internal standard and analyzed as described above. The determined abundance of norverapamil in supernatant and in cellular pellet were summed up, normalized to 10\u003csup\u003e4\u003c/sup\u003e cells and its production rate was expressed in pmols/ h.\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eXenotransplantation in Fah-/- / Rag2-/- / Il2rg-/- (FRG)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMale and female \u003cem\u003eFah-/-/Rag2-/-/Il2rg-/-\u003c/em\u003e (FRG) mice were obtained from Jackson Laboratory. Mice were housed and maintained under specific pathogen-free conditions in accordance with the Principles of Laboratory Animal Care and the Guide set by the HYU Industry-University Cooperation Foundation. For their maintenance, mice were administered \u003cem\u003ead libitum\u003c/em\u003e NTBC (2-(2-nitro-4-trifluoromethylbenzoyl)-1,3-cyclohexanedione) in drinking water.\u003c/p\u003e\n\u003cp\u003eMice aged 8-16 weeks old from both sexes were kept on NTBC (2-(2-nitro-4-trifluoromethylbenzoyl)-1,3-cyclohexanedione) in drinking water until 3 days prior to the experiment, when NTBC was withdrawn. Human hepatocyte organoids expanded in h-HepOrgs-EM2 and differentiated in h-HepOrgs-DM medium were dissociated into single cells and prepared for injection. For transplantation experiments commercially available frozen PHHs were used (F-PHH2, Supplementary Table 2). Organoids cultured under h-HepOrgs-EM2 medium as well as isolated hepatocytes (PHHs) from the same donors were used as controls. Following dissociation, 500,000 dissociated organoid cells or 800,000 primary human hepatocytes (PHHs) were resuspended in 100 \u0026mu;l of AdDMEM/F-12 medium and injected into the spleen. The non-injected negative control group received 100 \u0026mu;l of PBS instead of cells. Mice were cycled in and out of NTBC treatment for 3 days every time their body weight dropped below 80% of the initial weight.\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eIngenuity Pathway Analysis\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe performed Ingenuity Pathway Analysis (IPA, QIAGEN) to identify potential candidate signalling pathways. For that, we first generated three differentially expressed gene (DEG)lists as DEG between liver cancer organoids and liver cancer tissue or healthy tissue (Supplementary Dataset 1_S1 List_ 1 and 2) and DEG list between partial hepatectomy and healthy tissue (List_3). Gene lists were generated as follows: List_1 and 2: gene expression matrices from hepatocellular carcinoma (HCC)-derived organoids, HCC liver tissue, and liver tissue from healthy donors were obtained from the Gene Expression Omnibus (GEO) under accession number GSE84073\u003csup\u003e45\u003c/sup\u003e. Differentially expressed genes (DEGs) were identified using DESeq2\u003csup\u003e2\u003c/sup\u003e, applying a threshold of |log2 fold change| \u0026gt; 1 and an adjusted p-value (adj-p) \u0026lt; 0.1 (Supplementary Dataset 1_S1). List_3: DEGs comparing partial hepatectomy and undamaged liver hepatocytes in mouse were sourced from the supplementary tables in Hu \u003cem\u003eet al.\u003c/em\u003e\u003csup\u003e33\u003c/sup\u003e. Additionally, a list of genes mutated in both HCC-derived cell lines was derived from the whole-exome sequencing (WES) results in Broutier \u003cem\u003eet al.\u003c/em\u003e \u003csup\u003e5\u003c/sup\u003e. (List_4). The full list of DEG from Lists_1-4 are provided in Supplementary Dataset 1_S1.\u003c/p\u003e\n\u003cp\u003eThe three DEG lists and the mutated-gene list (Lists_1-4) were analyzed using Ingenuity Pathway Analysis, employing the Canonical Pathway Analysis and Upstream Regulator Prediction functions\u003csup\u003e4\u003c/sup\u003e. Detailed IPA methodology is provided elsewhere\u003csup\u003e5\u003c/sup\u003e. Briefly, the significance of the association between the dataset and canonical pathways was determined using a right-tailed Fisher\u0026apos;s exact test, followed by Benjamini-Hochberg (BH) correction for multiple testing adjustment. For analyses where log fold changes were available, an activity z-score was computed to predict the activation or inhibition likelihood of specific pathways base. The Upstream Regulator Analysis utilized a computational algorithm to identify upstream regulators potentially responsible for the observed gene expression changes. From the IPA Canonical Pathway Analysis, pathways were filtered based on adj-p \u0026lt; 0.05 and the presence of keyword \u0026quot;signalling\u0026quot; in the pathway name (Supplementary Dataset 1_S2). Selected pathways of interest with the mean adj-p value and frequency of pathway significance across comparisons were plotted in Extended Data Fig.1c (Supplementary Dataset 1_S3). Activity z-scores from the selected pathways were individually plotted as well as their corresponding mean values in Figure 1b (Supplementary Dataset 1_S4,5). Next, results from the Upstream Regulator Analysis were filtered for 1) \u0026lt;0.1 adj-p as upstream regulator, and 2) the molecules from the 2 selected signalling pathways (Supplementary Dataset 1_S6). Key components of the signalling pathways and their adj-p value in upstream regulator analysis were plotted in Extended Data Fig.1d (Supplementary dataset 1_S7).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eBulk RNA-Seq Library preparation\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003emRNA was isolated from on average 270 ng total RNA by poly-dT enrichment using the NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB) according to the manufacturer\u0026rsquo;s instructions. Samples were then directly subjected to the workflow for strand-specific RNA-Seq library preparation (Ultra II Directional RNA Library Prep, NEB). For ligation NEB Next Adapter for Illumina of the NEB Next Multiplex Oligos for Illumina Kit were used. After ligation, adapters were depleted by an XP bead purification (Beckman Coulter) adding the beads solution in a ratio of 0.9:1 to the samples. Unique dual indexing was done during the following PCR enrichment (12 cycles) using amplification primers carrying the same sequence for i7 and i5 index (i5: AAT GAT ACG GCG ACC ACC GAG ATC TAC AC NNNNNNNN ACA TCT TTC CCT ACA CGA CGC TCT TCC GAT CT, i7: CAA GCA GAA GAC GGC ATA CGA GAT NNNNNNNN GTG ACT GGA GTT CAG ACG TGT GCT CTT CCG ATC T). After two more XP bead purifications (0.9:1), libraries were quantified using the Fragment Analyzer (Agilent). Libraries were sequenced on an Illumina NovaSeq 6000 in 100 bp paired-end mode to a depth 40 million read pairs per library.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eRNA-sequencing data processing\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRaw bulk RNA-seq data was processed using nf-core/rnaseq v3.18.0 (doi: 10.5281/zenodo.1400710) of the nf-core collection of workflows\u003csup\u003e83\u003c/sup\u003e ,utilising reproducible software environments from the Bioconda \u003csup\u003e84\u003c/sup\u003eand Biocontainers\u003csup\u003e85\u003c/sup\u003e projects. The pipeline was executed with Nextflow v24.10.5\u003csup\u003e86\u003c/sup\u003e. The reference genome used was Homo sapiens GRCh38 (Ensembl release 111). The pipeline was run with custom parameters for trimming (extra_trimgalore_args: \u0026quot;--nextseq 20 --length 15\u0026quot;), alignment (extra_star_align_args: \u0026quot;--outFilterMismatchNmax 999 --outFilterMismatchNoverLmax 0.1 --alignMatesGapMax 200000 --chimSegmentMin 20 --twopassMode Basic --alignIntronMin 20 --alignIntronMax 200000\u0026quot;), and quantification (extra_salmon_quant_args: \u0026quot;--seqBias --gcBias --posBias\u0026quot;). The resulting MultiQC report was inspected to ensure overall sequencing quality and pipeline performance.\u003c/p\u003e\n\u003cp\u003eTranscript-level abundance estimates were imported using the \u003cem\u003etximeta\u003c/em\u003e package\u003csup\u003e87\u003c/sup\u003e to generate a gene-level count matrix. Next, variance stabilizing transformation (VST) from \u003cem\u003eDESeq2\u003c/em\u003e \u003csup\u003e88,89\u003c/sup\u003eto normalize the data. Euclidean distance matrices, principal component analysis (PCA), and heatmap visualizations were computed on the VST-transformed values. On some heatmaps, a min-max scaling was applied. In Extended Data Fig. 2a, b, batch correction was performed on the VST-transformed values using \u003cem\u003elimma\u003c/em\u003e\u0026rsquo;s removeBatchEffect, with sample material type (tissue vs. organoid) treated as the batch variable\u003csup\u003e90\u003c/sup\u003e. For differential expression analysis, DESeq2 was used. For the comparison between MM+WtnS+TRULI and Primary (fresh isolated PHHs), the design formula ~ donor + condition_l3 was applied (Extended Data Fig. 2). Log-fold changes were shrunken using lfcShrink with the ashr method (type=\u0026quot;ashr\u0026quot;), applying a fold-change threshold of 1.5 and a significance threshold of \u0026alpha; = 0.05\u003csup\u003e91\u003c/sup\u003e. For the comparison between DM and EM2 (Fig. 2e), the design formula ~ ~ batch + donor + condition_l1 was applied. Log-fold changes were shrunken using lfcShrink with the ashr method (type=\u0026quot;ashr\u0026quot;), applying a fold-change threshold of 1.5 and a significance threshold of \u0026alpha; = 0.05. For the comparison between h-Hep and PF (Extended Data Fig. 6h), the design formula ~ sex + cell_type was applied. Log-fold changes were shrunken using lfcShrink with the ashr method (type=\u0026quot;ashr\u0026quot;), applying a fold-change threshold of 4 and a significance threshold of \u0026alpha; = 0.05. Gene set enrichment analysis (GSEA) was conducted using the \u003cem\u003eclusterProfiler\u003c/em\u003e package, leveraging gseKEGG, gseGO, and gsePathway for pathway enrichment analysis\u003csup\u003e92\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe zonated gene list (Extended Data Fig. 3h) was obtained by manually curating the genes that are confirmed to be portally or centrally zonated from human spatial transcriptomic datasets from\u003cem\u003e\u003csup\u003e13,14,58,59\u003c/sup\u003e\u003c/em\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e(full list is provided in Supplementary Dataset 2_S6).\u003cem\u003e\u0026nbsp;\u003c/em\u003eWe then intersected this refined zonated gene lists with our list of differentially expressed genes in the DM vs. EM2 comparison.\u003c/p\u003e\n\u003cp\u003eDonor-specific genes were identified separately for batches Y1/Y2 and S1/S2 using a likelihood ratio test (LRT) with the full model ~ donor and the reduced model ~ 1. Genes with an adjusted P-value (padj) \u0026lt; 0.05 were retained, and the resulting gene lists from the two batches were merged. Pairwise correlations between organoids and primary cells were computed using the donor-specific genes. For the heatmap shown in Extended Data Fig. 4e, sex-specific genes were excluded.\u003c/p\u003e\n\u003cp\u003eThe complete software stack for downstream analysis is available as a Docker container (rnaseq-notebook:2025-04-21) archived at https://quay.io/repository/fbnrst/rnaseq-notebook.\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003e\u003cem\u003eSingle-cell Transcriptomics with 10x Genomics\u003c/em\u003e\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eFor single cell RNAseq analysis assembloids were generated by assembling h-HepOrgs, cholangiocytes/ductal cells derived from cholangiocytes organoids (nuclear-GFP) and portal fibroblasts (PFs, nuclear-RFP) at a ratio 1 h-HepOrgs: 25 MSC : 100 Cholangiocytes. At 5-6 days after aggregation, assembloids were collected as follows: periportal assembloids were dissociated to single cells using TrypLE 10x for 5min at 37\u0026deg;C. The cells were resuspended in DM and 10\u0026mu;g/mL DNAse in BSA-coated tubes and filtered through a 100 \u0026mu;m strainer. Cell suspensions (30000 \u0026ndash; 50000 cells) were concentrated by centrifugation (50 rcf, 5 min, 4 \u0026deg;C) and the volume reduced to ~55 \u0026micro;l. Cells were carefully resuspended and visually inspected under a light microscope to determine cell concentration and quality. The concentrations of the single-cell suspensions were adjusted to 138-912 cells per microliter and carefully mixed with the reverse transcription mix before loading cells on the 10X Genomics Chromium system \u003csup\u003e93\u003c/sup\u003e in a Chromium Single Cell G Chip targeting 3000-10,000 cells per reaction. Following the guidelines of the 10x Genomics Chromium Single Cell Kit v3.1 user manual, the droplets were directly subjected to reverse transcription, the emulsion was broken and cDNA was purified using Dynabeads MyOne Silane (10X Genomics). cDNA was first amplified with 12 cycles, and then purified with 0.6x SPRIselect beads (Beckman Coulter) to enrich cDNA fragments (\u0026gt;400 bp). A quality and quantity control of cDNA on the Fragment Analyzer (using the DNF-473 NGS Fragment Kit, Agilent) was eventually performed to obtain its concentration. The 10X Genomics single cell RNA-seq library preparation - involving fragmentation, dA-Tailing, adapter ligation and 11 or 12 cycles indexing PCR \u0026ndash; was performed based on the manufacturer\u0026rsquo;s protocol. After quantification, the libraries were sequenced on an Illumina Novaseq6000 in paired-end mode (R1/R2: 100 cycles; I1/I2: 10 cycles), generating 230-370 million fragment pairs.\u003c/p\u003e\n\u003cp\u003eThe raw sequencing data was then processed with the \u0026lsquo;count\u0026rsquo; command of the Cell Ranger software (v8.0.1) provided by 10X Genomics with the option \u0026lsquo;--expect-cells\u0026rsquo; set to 10,000 (all other options were used as per default). To build the reference for Cell Ranger, human genome (GRChg38) as well as gene annotation (Ensembl 104) were downloaded from Ensembl. Genome and annotation were processed following the build steps provided by 10x (https://support.10xgenomics.com/single-cell-gene-expression/software/release-notes/build#mm10_2020A) to build the appropriate Cellranger reference.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003escRNAseq Data Analysis\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe raw scRNAseq data was processed using nf-core/scrnaseq v3.0.0 (doi: 10.5281/zenodo.3568187) of the nf-core collection of workflows \u003csup\u003e83\u003c/sup\u003e, utilising reproducible software environments from the Bioconda\u003csup\u003e84\u003c/sup\u003e and Biocontainers \u003csup\u003e85\u003c/sup\u003eprojects. The pipeline was executed with Nextflow v24.10.5 \u003csup\u003e86\u003c/sup\u003eSTARSOLO was used as the aligner. The reference genome was set to \u003cem\u003eHomo sapiens\u003c/em\u003e GRCh38 (Ensembl release 111) with custom additions for red fluorescent protein (RFP) and green fluorescent protein (GFP) transgenes, obtained from SnapGene (DsRed1 and EGFP, respectively). Outputs were inspected for quality control (QC), and one sample with poor QC was excluded from further analysis. Within nf-core/scrnaseq, technical artefacts were eliminated using CellBender\u003csup\u003e94\u003c/sup\u003e. CellBender output was used for data visualization. Doublet detected per sample was performed using scrublet\u003csup\u003e95\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eFurther analysis was perfomed using scanpy\u003csup\u003e96\u003c/sup\u003e. QC was applied with the following thresholds: minimum total counts of 5000, minimum detected genes of 2000, a maximum percentage of counts in the top 50 genes set at 50%, maximum percentage of mitochondrial counts at 15%, and a maximum doublet score of 0.15. Gene filtering was performed to retain genes expressed in at least 10 cells. After filtering, the data underwent normalization, log transformation, and identification of top 3000 highly variable genes. PCA was performed, and batch correction was implemented through harmony integration\u003csup\u003e97\u003c/sup\u003e. UMAP visualization and Leiden clustering were used to identify the three expected cell types\u003csup\u003e98,99\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eTo compare homeostatic-like and fibrotic-like organoids, pseudobulk aggregation was performed using \u003cem\u003edecoupler\u003c/em\u003e for each cell type\u003csup\u003e100\u003c/sup\u003e. Pseudobulk data was generated by summing raw counts for each sample and cell type, with a minimum requirement of 10 cells per group and 1000 total counts. Differential expression analysis was conducted using \u003cem\u003epyDESeq2\u003c/em\u003e\u003csup\u003e101\u003c/sup\u003e. For each cell type, DESeq2 datasets were created with design factors that included \u0026lsquo;donor\u0026rsquo; and \u0026lsquo;condition\u0026rsquo;, using the \u0026lsquo;homeostatic-like\u0026rsquo; condition as the reference. Differentially expressed genes between the homeostatic-like and fibrotic-like conditions were ranked based on the test statistic. Subsequently, gene set enrichment analysis (GSEA) was performed on the ranked lists using \u003cem\u003eclusterProfiler\u003c/em\u003e, focusing on KEGG, Reactome, and GO terms.\u003c/p\u003e\n\u003cp\u003eThe complete software stack for downstream analysis is available as a Docker container (singlecell-notebook:2025-04-21) archived at https://quay.io/repository/fbnrst/singlecell-notebook.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eComparison to Public Datasets\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData from Andrews et al. (2022, 2024) and Guilliams et al. (2022) were downloaded in h5ad format from https://cellxgene.cziscience.com/. Additionally, data from Brazovskaja (2024) was obtained from https://data.mendeley.com/datasets/yp3txzw64c/1, and the dataset from Ramachandran \u003cem\u003eet al.,\u0026nbsp;\u003c/em\u003e2019 \u003csup\u003e10\u003c/sup\u003e was downloaded from https://datashare.ed.ac.uk/bitstream/handle/10283/3433/tissue.rdata?sequence=3\u0026amp;isAllowed=y and converted to h5ad format using the sceasy package.\u003c/p\u003e\n\u003cp\u003eThese public datasets were merged with the raw count matrix of our QC-filtered organoid data. Subsequently, the combined dataset underwent normalization, followed by log transformation and detection of the top 4000 highly variable genes. We performed principal component analysis (PCA) and integrated the dataset using Harmony, specifying the concatenation of the paper and donor as batch variables, with a maximum of 20 iterations and a theta value of 1.5. Selected genes were visualized in a dot plot (Fig. 4h).\u003c/p\u003e\n\u003cp\u003ePseudobulk analyses were then conducted using the decoupler package to summarize gene expression by cell type. This involved generating a pseudobulk dataset where raw counts were summed by sample and cell type, ensuring a minimum of 30 cells per group. Following pseudobulk aggregation, the data was normalized and log-transformed, with the top highly variable genes identified based on mean expression and dispersion. Additionally, the \u0026lsquo;paper\u0026rsquo; variable was regressed out to mitigate batch effects. Next, PCA was performed on the pseudobulk data, using 50 principal components for subsequent analyses. Hierarchical clustering was executed using the Pearson correlation metric, and Pearson correlation matrices were plotted, as shown in Fig. 4g, 5f.\u003c/p\u003e\n\u003cp\u003eMarker genes for the 3 major cell types were computed separately for our organoid data and the merged public data. using scanpy\u0026rsquo;s rank_genes_groups function. For each dataset, the top 300 marker genes for each cell type were selected. Subsequently, gene set enrichment analysis was performed using the \u003cem\u003egseapy\u003c/em\u003e package, leveraging the Enrichr method\u003csup\u003e102\u003c/sup\u003e. The analysis focused on the KEGG 2021 Human and Reactome 2022 gene sets, with a p-value cutoff of 0.05. Shared enriched pathways between the organoid and tissue datasets were identified, and the combined enrichment scores for selected terms were plotted (Extended Data Fig. 8c, d).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData statistical analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll values are represented as mean \u0026plusmn; standard error of the mean (S.E.M.) as specified in legend. Either Two-tailed Mann-Whitney non-parametric or two-tailed ANOVA tests were used. p\u0026lt;0.05 was considered statistically significant. In all cases data from at least 3 independent experiments was used. Calculations were performed using the Prism 9 software package. All \u003cem\u003ep\u003c/em\u003e-values are given in the corresponding figure legends. Dispersion and precision measures (e.g., mean, median, SD, SEM) are specified in the figure legends. All the scRNAseq statistics are described above in the corresponding section.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData availability\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe scRNAseq and bulk RNAseq datasets generated during this study is available in the European Genome-phenome Archive (EGA) (EGAS50000000994). The full lists of all differentially expressed genes, GSEA terms and marker genes are available in Supplementary Datasets 2-4.\u003c/p\u003e\n\u003cp\u003eThe raw data used for mass spectrometry is available at the following repository:\u003c/p\u003e\n\u003cp\u003ehttps://doi.org/10.17617/3.Z9GMJE\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eCode availability\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe source code used for bile canaliculi segmentation is available at the following repository:\u003c/p\u003e\n\u003cp\u003ehttps://git.mpi-cbg.de/huch_lab/assembloid-paper\u003c/p\u003e\n\u003cp\u003eThe source code for bulkRNAseq and scRNAseq data analysis are available upon request and will be made public upon publication (https://git.mpi-cbg.de/huch_lab/fr_human_assembloid).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAdditional references\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e74 Cordero-Espinoza, L. \u0026amp; Huch, M. The balancing act of the liver: tissue regeneration versus fibrosis. \u003cem\u003eJ Clin Invest\u003c/em\u003e \u003cstrong\u003e128\u003c/strong\u003e, 85-96 (2018). https://doi.org:10.1172/JCI93562\u003c/p\u003e\n\u003cp\u003e75 Mederacke, I.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Fate tracing reveals hepatic stellate cells as dominant contributors to liver fibrosis independent of its aetiology. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 2823 (2013). https://doi.org:10.1038/ncomms3823\u003c/p\u003e\n\u003cp\u003e76 Pinzani, M. \u0026amp; Luong, T. V. Pathogenesis of biliary fibrosis. \u003cem\u003eBiochim Biophys Acta Mol Basis Dis\u003c/em\u003e \u003cstrong\u003e1864\u003c/strong\u003e, 1279-1283 (2018). https://doi.org:10.1016/j.bbadis.2017.07.026\u003c/p\u003e\n\u003cp\u003e77 Desmet, V. J. \u0026amp; Roskams, T. Cirrhosis reversal: a duel between dogma and myth. \u003cem\u003eJ Hepatol\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 860-867 (2004). https://doi.org:10.1016/j.jhep.2004.03.007\u003c/p\u003e\n\u003cp\u003e78 Lee, K. P.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The Hippo-Salvador pathway restrains hepatic oval cell proliferation, liver size, and liver tumorigenesis. \u003cem\u003eProc Natl Acad Sci U S A\u003c/em\u003e \u003cstrong\u003e107\u003c/strong\u003e, 8248-8253 (2010). https://doi.org:10.1073/pnas.0912203107\u003c/p\u003e\n\u003cp\u003e79 Lu, L.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Hippo signaling is a potent in vivo growth and tumor suppressor pathway in the mammalian liver. \u003cem\u003eProc Natl Acad Sci U S A\u003c/em\u003e \u003cstrong\u003e107\u003c/strong\u003e, 1437-1442 (2010). https://doi.org:10.1073/pnas.0911427107\u003c/p\u003e\n\u003cp\u003e80 Ke, M. T., Fujimoto, S. \u0026amp; Imai, T. SeeDB: a simple and morphology-preserving optical clearing agent for neuronal circuit reconstruction. \u003cem\u003eNat Neurosci\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 1154-1161 (2013). https://doi.org:10.1038/nn.3447\u003c/p\u003e\n\u003cp\u003e81 Uhlen, M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Proteomics. Tissue-based map of the human proteome. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e347\u003c/strong\u003e, 1260419 (2015). https://doi.org:10.1126/science.1260419\u003c/p\u003e\n\u003cp\u003e82 Zeigerer, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Rab5 is necessary for the biogenesis of the endolysosomal system in vivo. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e485\u003c/strong\u003e, 465-470 (2012). https://doi.org:10.1038/nature11133\u003c/p\u003e\n\u003cp\u003e83 Ewels, P. A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The nf-core framework for community-curated bioinformatics pipelines. \u003cem\u003eNat Biotechnol\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 276-278 (2020). https://doi.org:10.1038/s41587-020-0439-x\u003c/p\u003e\n\u003cp\u003e84 Gruning, B.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Bioconda: sustainable and comprehensive software distribution for the life sciences. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 475-476 (2018). https://doi.org:10.1038/s41592-018-0046-7\u003c/p\u003e\n\u003cp\u003e85 da Veiga Leprevost, F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e BioContainers: an open-source and community-driven framework for software standardization. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 2580-2582 (2017). https://doi.org:10.1093/bioinformatics/btx192\u003c/p\u003e\n\u003cp\u003e86 Di Tommaso, P.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Nextflow enables reproducible computational workflows. \u003cem\u003eNat Biotechnol\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 316-319 (2017). https://doi.org:10.1038/nbt.3820\u003c/p\u003e\n\u003cp\u003e87 Love, M. I.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Tximeta: Reference sequence checksums for provenance identification in RNA-seq. \u003cem\u003ePLoS Comput Biol\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, e1007664 (2020). https://doi.org:10.1371/journal.pcbi.1007664\u003c/p\u003e\n\u003cp\u003e88 Huber, W., von Heydebreck, A., Sultmann, H., Poustka, A. \u0026amp; Vingron, M. Variance stabilization applied to microarray data calibration and to the quantification of differential expression. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e18 Suppl 1\u003c/strong\u003e, S96-104 (2002). https://doi.org:10.1093/bioinformatics/18.suppl_1.s96\u003c/p\u003e\n\u003cp\u003e89 Love, M. I., Huber, W. \u0026amp; Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. \u003cem\u003eGenome Biol\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 550 (2014). https://doi.org:10.1186/s13059-014-0550-8\u003c/p\u003e\n\u003cp\u003e90 Ritchie, M. E.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e limma powers differential expression analyses for RNA-sequencing and microarray studies. \u003cem\u003eNucleic Acids Res\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, e47 (2015). https://doi.org:10.1093/nar/gkv007\u003c/p\u003e\n\u003cp\u003e91 Stephens, M. False discovery rates: a new deal. \u003cem\u003eBiostatistics\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 275-294 (2017). https://doi.org:10.1093/biostatistics/kxw041\u003c/p\u003e\n\u003cp\u003e92 Yu, G., Wang, L. G., Han, Y. \u0026amp; He, Q. Y. clusterProfiler: an R package for comparing biological themes among gene clusters. \u003cem\u003eOMICS\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 284-287 (2012). https://doi.org:10.1089/omi.2011.0118\u003c/p\u003e\n\u003cp\u003e93 Zheng, G. X.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Massively parallel digital transcriptional profiling of single cells. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 14049 (2017). https://doi.org:10.1038/ncomms14049\u003c/p\u003e\n\u003cp\u003e94 CellBender removes technical artifacts from single-cell RNA sequencing data. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 1285-1286 (2023). https://doi.org:10.1038/s41592-023-01946-4\u003c/p\u003e\n\u003cp\u003e95 Wolock, S. L., Lopez, R. \u0026amp; Klein, A. M. Scrublet: Computational Identification of Cell Doublets in Single-Cell Transcriptomic Data. \u003cem\u003eCell Syst\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 281-291 e289 (2019). https://doi.org:10.1016/j.cels.2018.11.005\u003c/p\u003e\n\u003cp\u003e96 Wolf, F. A., Angerer, P. \u0026amp; Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. \u003cem\u003eGenome Biol\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 15 (2018). https://doi.org:10.1186/s13059-017-1382-0\u003c/p\u003e\n\u003cp\u003e97 Korsunsky, I.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Fast, sensitive and accurate integration of single-cell data with Harmony. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 1289-1296 (2019). https://doi.org:10.1038/s41592-019-0619-0\u003c/p\u003e\n\u003cp\u003e98 Becht, E.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Dimensionality reduction for visualizing single-cell data using UMAP. \u003cem\u003eNat Biotechnol\u003c/em\u003e (2018). https://doi.org:10.1038/nbt.4314\u003c/p\u003e\n\u003cp\u003e99 Traag, V. A., Waltman, L. \u0026amp; van Eck, N. J. From Louvain to Leiden: guaranteeing well-connected communities. \u003cem\u003eSci Rep\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 5233 (2019). https://doi.org:10.1038/s41598-019-41695-z\u003c/p\u003e\n\u003cp\u003e100 Badia, I. M. P.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e decoupleR: ensemble of computational methods to infer biological activities from omics data. \u003cem\u003eBioinform Adv\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, vbac016 (2022). https://doi.org:10.1093/bioadv/vbac016\u003c/p\u003e\n\u003cp\u003e101 Muzellec, B., Telenczuk, M., Cabeli, V. \u0026amp; Andreux, M. PyDESeq2: a python package for bulk RNA-seq differential expression analysis. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e (2023). https://doi.org:10.1093/bioinformatics/btad547\u003c/p\u003e\n\u003cp\u003e102 Fang, Z., Liu, X. \u0026amp; Peltz, G. GSEApy: a comprehensive package for performing gene set enrichment analysis in Python. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e (2023). https://doi.org:10.1093/bioinformatics/btac757\u003c/p\u003e\n\u003cp\u003e103 Tirosh, I.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e352\u003c/strong\u003e, 189-196 (2016). https://doi.org:10.1126/science.aad0501\u003c/p\u003e\n\u003cp\u003e104 Wang, Y.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Comprehensive Molecular Characterization of the Hippo Signaling Pathway in Cancer. \u003cem\u003eCell Rep\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 1304-1317 e1305 (2018). https://doi.org:10.1016/j.celrep.2018.10.001\u003c/p\u003e\n\u003cp\u003e105 Boj, S. F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Diabetes risk gene and Wnt effector Tcf7l2/TCF4 controls hepatic response to perinatal and adult metabolic demand. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e151\u003c/strong\u003e, 1595-1607 (2012). https://doi.org:10.1016/j.cell.2012.10.053\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5314788/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5314788/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe development of complex multicellular human \u003cem\u003ein vitro\u003c/em\u003e systems holds great promise for modelling disease and progressing drug discovery and tissue engineering\u003csup\u003e1\u003c/sup\u003e. In the liver, despite the identification of key signalling pathways involved in hepatic regeneration\u003csup\u003e2,3\u003c/sup\u003e, \u003cem\u003ein vitro\u003c/em\u003e expansion of human hepatocytes directly from fresh patient tissue has not yet been achieved, limiting the possibility of modelling liver composite structures \u003cem\u003ein vitro\u003c/em\u003e. Here, we first developed human hepatocyte organoids (h-HepOrgs) from 28 different patients. Patient-derived hepatocyte organoids sustain long-term expansion of hepatocytes \u003cem\u003ein vitro \u003c/em\u003eand\u003cem\u003e \u003c/em\u003emaintain patient-specific gene expression, and bile canaliculi features and function of the \u003cem\u003ein vivo\u003c/em\u003e tissue. After transplantation, expanded human hepatocyte organoids rescue the phenotype of a mouse model of liver disease. By combining human hepatocyte organoids with portal mesenchyme and our previously published cholangiocyte organoids\u003csup\u003e4-6\u003c/sup\u003e, we generated patient-specific periportal liver assembloids that retain the histological arrangement, gene expression and cell interactions of the periportal\u003cem\u003e \u003c/em\u003eliver\u003cem\u003e \u003c/em\u003etissue, with cholangiocytes and mesenchyme embedded in the hepatocyte parenchyma. We leveraged this platform to model aspects of biliary fibrosis. Our human periportal liver assembloid system represents a novel \u003cem\u003ein vitro\u003c/em\u003e platform to investigate human liver pathophysiology, accelerate drug development, enable early diagnosis and progress personalized medicine.\u003c/p\u003e","manuscriptTitle":"Human assembloids recapitulate periportal liver tissue in vitro","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-17 11:01:54","doi":"10.21203/rs.3.rs-5314788/v1","editorialEvents":[],"status":"published","journal":{"display":false,"email":"[email protected]","identity":"nature","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"nature","sideBox":"Learn more about [Nature](http://www.nature.com/nature/)","snPcode":"","submissionUrl":"","title":"Nature","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"d2d7ed1e-d376-4533-bca8-eb18a49cbdbe","owner":[],"postedDate":"September 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":54596906,"name":"Biological sciences/Developmental biology/Self-renewal"},{"id":54596907,"name":"Biological sciences/Developmental biology/Self-renewal"},{"id":54596908,"name":"Biological sciences/Stem cells/Regeneration"},{"id":54596909,"name":"Biological sciences/Stem cells/Regeneration"},{"id":54596910,"name":"Biological sciences/Developmental biology/Disease model"}],"tags":[],"updatedAt":"2025-12-18T08:10:35+00:00","versionOfRecord":{"articleIdentity":"rs-5314788","link":"https://doi.org/10.1038/s41586-025-09884-1","journal":{"identity":"nature","isVorOnly":false,"title":"Nature"},"publishedOn":"2025-12-17 05:00:00","publishedOnDateReadable":"December 17th, 2025"},"versionCreatedAt":"2025-09-17 11:01:54","video":"","vorDoi":"10.1038/s41586-025-09884-1","vorDoiUrl":"https://doi.org/10.1038/s41586-025-09884-1","workflowStages":[]},"version":"v1","identity":"rs-5314788","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5314788","identity":"rs-5314788","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}