Adult and iPS-derived non-parenchymal cells regulate liver organoid development through differential modulation of Wnt and TGF-β
- 44 Downloads
Liver organoid technology holds great promises to be used in large-scale population-based drug screening and in future regenerative medicine strategies. Recently, some studies reported robust protocols for generating isogenic liver organoids using liver parenchymal and non-parenchymal cells derived from induced pluripotent stem cells (iPS) or using isogenic adult primary non-parenchymal cells. However, the use of whole iPS-derived cells could represent great challenges for a translational perspective.
Here, we evaluated the influence of isogenic versus heterogenic non-parenchymal cells, using iPS-derived or adult primary cell lines, in the liver organoid development. We tested four groups comprised of all different combinations of non-parenchymal cells for the liver functionality in vitro. Gene expression and protein secretion of important hepatic function markers were evaluated. Additionally, liver development-associated signaling pathways were tested. Finally, organoid label-free proteomic analysis and non-parenchymal cell secretome were performed in all groups at day 12.
We show that liver organoids generated using primary mesenchymal stromal cells and iPS-derived endothelial cells expressed and produced significantly more albumin and showed increased expression of CYP1A1, CYP1A2, and TDO2 while presented reduced TGF-β and Wnt signaling activity. Proteomics analysis revealed that major shifts in protein expression induced by this specific combination of non-parenchymal cells are related to integrin profile and TGF-β/Wnt signaling activity.
Aiming the translation of this technology bench-to-bedside, this work highlights the role of important developmental pathways that are modulated by non-parenchymal cells enhancing the liver organoid maturation.
KeywordsOrganoid Liver iPS Hepatocyte 3D culture
Alpha 1 antitrypsin
Bone morphogenetic protein
C-X-X motif receptor 4
Dental pulp-derived MSC
Endothelial growth media
Enzyme-linked immunosorbent assay
Forkhead box A2
GATA binding protein 4
Glutathione S-transferase alpha 1
Human aortic endothelial cells
Hepatocyte growth factor
Hypoxia induced-factor 1A
Hepatocyte nuclear factor 4 alpha
Human natural killer-1
Human umbilical vein endothelial cells
Insulin-like growth factor
Organoid comprised by iPS-derived NPC
Organoid comprised by iPS-derived endothelial cells and dpMSC
iPS-derived neural crest cells
Organoid comprised by iPS-derived MSC and HAEC
Organoid comprised by adult primary NPC
Induced pluripotent stem cells
Mesenchymal stem cells
Nerve growth factor receptor
Periodic acid of Schiff
Reverse transcriptase quantitative polymerase chain reaction
Standard error of mean
Transforming growth factor beta
Tumor necrosis factor alpha
UDP glucuronosyltransferase family 1 member A1
Wnt inhibitory factor 1
Liver organogenesis can be in part recapitulated in part by using organoid technology . The combination of defined ratio of parenchymal progenitor cells (i.e., hepatoblast) and non-parenchymal cells (NPC) (i.e., endothelial cells and mesenchymal cells) recreates a cellular microenvironment akin to the early stages of liver bud development and allows for spontaneous tissue formation . The first attempts to bioengineer complex liver organoids (LOs) used hepatoblasts derived from human pluripotent stem cells (iPS) in conjunction with primary human NPC, such as human umbilical cord-derived endothelial cells (HUVEC) and adipose tissue-derived mesenchymal stem cells (MSCs), all derived from different donors .
NPC contribute to liver development and homeostasis by secreting growth factors (e.g., TNF-α, IL-6, HGF, TGF-β, and BMP2, 4 and 6) that regulate hepatocyte proliferation, DNA synthesis, and hepatic cord formation [3, 4, 5]. Asai and collaborators  showed the distinct contributions of primary lineages of endothelial cells (ECs) and MSC secretome in LO development in vitro. More recently, some other groups reported a series of combined protocols for generating isogenic LOs obtained from whole iPS-derived cells, obtained from the same donor, or by using primary NPCs from the same donor [7, 8, 9]. Takebe and collaborators  successfully generated LOs from human donors that could potentially be applied for high-throughput personalized screening of liver toxicity.
However, large-scale differentiation of iPS into multiple cell lineages is challenging in terms of cost and efficiency as opposed to primary cell lineages. As a caveat, the use of standard commercial non-parenchymal cell lines will yield human LOs that are chimeric in nature. Here, we propose to evaluate the effects of applying liver NPCs derived from iPS-derived fetal-like cells versus adult primary NPC cell lines to LO development and functionality.
iPS generation and culture and primary adult cell culture
Induced pluripotent stem cells (iPSs) were generated from three healthy human donors (F9048 = male, 26; F8799 = female, 28; F7405 = male, 23), as previously described . The reprogramming and cell culture protocol were described in Additional file 1: methods. Differentiation protocols and human primary adult cell culture methods were described in Additional file 1: methods.
Prior to cell seeding, Matrigel was diluted 1:1 on ice with cold EGM-2 and dispensed at 380 μL/well in a 24-well plate. Gelling was achieved by incubation in 37°C for at least 30 min. A mixture of iPS-derived cells (1 × 106 hepatoblast, 8 × 105 ECs, and 2 × 105 MSCs, as per Takebe et al. ) was centrifuged for 5 min at 300×g and resuspended in 2 mL of LO culture media (composed of 1:1 EGM-2/hepatocyte differentiation media, see Additional file 1: methods). The cell mixture was seeded on top of the Matrigel bed. Media was changed every other day. In order to assess the rate of mesenchymal condensation, pictures of the wells were taken every 12 h. The confluent cell layer and the total covered area progressive condensation over time was evaluated using ImageJ software.
Proteomic sample processing and analysis followed a previously published protocol . For detailed information, see Additional file 1: methods section. Pathway annotation of protein IDs was performed using the comprehensive EnrichR gene set enrichment analysis web server [12, 13], applying Reactome  and Panther  categorization with the significance threshold set at p < 0.05. Interactome analysis was performed using String  with k-means clustering in three groups.
Statistical analyses to assess LO functional analysis and development quality (Figs. 2 and 4) were performed using one-way ANOVA with Tukey’s post-test. For all other statistical analyses, Student’s two-tailed t tests were used for pairwise comparisons. Data are presented as means ± SEM, or mean of at least three independent experiments, with at least two technical replicates. For the proteomics analyses, statistical tests were performed using Students t test, using Perseus software, and pathway enrichment analysis using EnrichR. Values of p < 0.05 were considered significant. GraphPad Prism software was used to perform all other statistical analyses.
iPS cell differentiation
The hepatic differentiation potential was evaluated and characterized in vitro. Figure 1b shows representative images of flow cytometric analyses. After 3 days of differentiation, 58.5 ± 4.7% (n = 3) of cells were CXCR4+/FOXA2+ (definitive endoderm). Despite starting out with a heterogeneous population, at day 9 of differentiation, a majority of cells (78 ± 5.8%, n = 3) expressed hepatic progenitor markers, such as HNF4A and AFP. After terminal hepatocyte differentiation, 74.3 ± 7.1% (n = 3) of the cells expressed the hepatic markers ALB and UGT1A1 (Fig. 1b). Phase-contrast image showed homogenous hepatocyte morphology in a monolayer culture. Representative images of immunofluorescence (IF) staining for each step of hepatic differentiation are shown in Fig. 1c. After 3 days, 62.8 ± 4.8% (n = 3) of cells were double positive for FOXA2 and CXCR4. After 9 days, 77.4 ± 5.9% (n = 3) were positive for HNF4A and AFP, and at day 18, 88.6 ± 6.9% (n = 3) of cell population were positive for ALB and UGT1A1. Also, at day 18, 90.2 ± 3.1% (n = 3) cells stained positive for Periodic acid-Schiff (PAS) (Fig. 1c).
For assessing endothelial differentiation, the iPS-derived cells were evaluated stepwise throughout the differentiation protocol. Figure 1d shows representative images of the flow cytometric analyses. The first step of the differentiation protocol (day 2) resulted in homogenous mesodermal differentiation, as inferred from the nearly ubiquitous co-expression of Brachyury T and GATA4 (92.4 ± 3.7, n = 3). However, at the end of endothelial differentiation (day 7), only 13.1 ± 2.7% of the cell population was double positive for endothelial markers CD34 and CD31. After magnetic sorting of the CD31+ cells and seeding a 60% confluent cell culture, the great majority of cells were positive for CD31 and a varying percentage of them were also positive for VECAD (48.6 ± 22.8%, n = 3). At day 7 of endothelial differentiation, 12.8 ± 2.2% of cells were positive for CD34. After cell sorting, 90.1% ± 6.7% of cells were double positive for CD31 and VECAD (Fig. 1e). IF staining was performed in 90% confluent culture, which could explain the difference observed in flow cytometry analysis. ECs took up acetylated LDL (95.3 ± 3.7%, n = 3) and were able to generate capillary-like tubular structures in the Matrigel angiogenesis assay (Fig. 1d, e).
For mesenchymal differentiation, flow cytometric analysis indicated that the majority of iNCC cells expressed HNK1 and P75 (79.0 ± 3.1%, n = 3). Following mesenchymal terminal differentiation, essentially all the cells showed a typical MSC morphology, were positive for the majority of cell population which expressed CD73 and CD90 (95.1 ± 1.8%, n = 3) and CD105 (Fig. 1f). Also, iNCC staining revealed that the majority (80.3 ± 7.9%) of the cells were double positive for P75 and Vimentin. Similarly, most (85.2 ± 8.0%) iNCC-derived MSCs were double positive for CD73 and Vimentin (Fig. 1g). Additionally, 90.2 ± 5.9% of these MSCs were positive for CD105 (Fig. 1g). Finally, we tested the ability of the iNCC-derived MSC to differentiate into osteogenic, chondrogenic, and adipogenic lineages. Figure 1h shows representative, low-magnification bright-field images and photographs of cell culture wells, and negative controls, stained for Alizarin Red, Alcian Blue, and Oil Red, respectively, after 28-day exposure to the various MSC differentiation induction protocols.
Liver organoid functional analysis
LO developmental pathway analysis
A major aim of the present study was to elucidate the role and efficacy of NPCs, derived either from iPS or from primary cell culture, in the development and functionality of LOs. Previous reports showed that it is possible to generate isogenic LOs using either primary or iPS-derived liver NPCs [8, 9]. However, the impact of using such distinct NPC sources in LO maturation has never been addressed.
Here, we show that all human iPS cell lines used in the present study were characterized by flow cytometry and gene expression (Additional file 1: Figure S1A and B). We successfully differentiated all three iPS cell lines towards hepatoblasts, arterial ECs, and iNCC-derived MSCs (Fig. 1 and Additional file 1: Figure S1C). The differential contribution, if any, of arterial, venous, or lymphatic ECs in the development of LO remains to be elucidated. Thus, we used iPS-derived arterial ECs, with commercially available HAECs as the correspondent adult cell line. For mesenchymal cells, we used iNCC-derived MSC, with dpMSC as the primary adult cell counterpart. Although there are many available protocols to differentiate iPS towards MSC, the vast majority of them lack proper comparison to the specific adult tissue MSC and are reported as a general embryonic or mesodermal MSC differentiation [17, 18, 19]. Since MSCs have tissue-specificity functionality, aiming to properly compare the use of primary versus iPS-derived MSC in LO maturation, we needed to compare matching tissue-derived primary MSC to iPS-differentiated tissue analog. Additionally, it is well known that iPS-derived MSCs obtained from different intermediated germ lines have different properties . Our group recently reported a protocol to generate cranial facial MSC (obtained from iPS-derived neural crest cells) that closely resembles MSC located at the dental pulp . Here, we were able to compare matching tissue specificity of MSCs (i.e., primary vs iPS-derived) in the LO maturation process.
We succeeded in deriving functional hepatoblasts, as assessed by the expression of HNF4A and AFP. The potential of our hepatoblasts to differentiate into hepatocyte was inferred from the expression pattern of mature hepatic markers such as UGT1A1 and CK18 and by PAS staining (Fig. 1b, c).
Arterial endothelial cell commitment was inferred from the expression of general endothelial markers, such as CD31 and VECAD, together with that of specific arterial EC markers such as NOTCH4 and the very low levels of expression of PDPN and EPHB4, which are markers of lymphatic and venous phenotype, respectively (Additional file 1: Figure S1B). The functionality of our iPS-derived ECs was demonstrated by their ability to take up acetylated LDL and to generate capillaries in Matrigel (Fig. 1d, e).
Neural crest differentiation was confirmed by the expression of HNK1 and CD75 (Fig. 1f). iNCC-derived MSC expressed stromal-mesenchymal markers such as CD105, CD90, and CD73 (Fig. 1f, g). Following in vitro induction, the mesenchymal differentiation was confirmed by the osteogenic staining with Alizarin Red (calcium deposits), chondrogenic staining with Alcian Blue (glycosaminoglycans), and adipogenic staining with Oil Red (lipid droplets) (Fig. 1h). This result indicates that we have successfully derived competent multipotent mesenchymal cells, and not fibroblasts.
LO generation was compared between all tested groups. No differences were observed in the mesenchymal condensation rate and in the morphology between all tested groups (Fig. 2a, b). Even though we noted some intrinsic variations between the tested cell lines, our RT-qPCR data at day 12 (Fig. 2c) revealed that important genes related to hepatic xenobiotic metabolism of phase I (i.e., CYP1A2 and CYP1A1) and II (i.e., GSTA1) were overexpressed in IIP. Even though CYP3A4 RT-qPCR data showed no significant differences, enzymatic activity levels were significantly higher in group IIP. These data suggest that the hepatic metabolic rate was higher in the presence of dpMSC associated with iPS-derived ECs (i.e., group IIP) and reduced in the presence of adult arterial ECs. ALB and TDO2 gene expression were also significantly elevated in group IIP, which suggests increased hepatic maturation. In addition, IIP produced more albumin at day 12 (Fig. 2c), with reduced AFP gene expression and secretion, as compared to IPI and IPP, but not to III. A concomitant increase in albumin and reduction of AFP secretion is one of the most important hallmarks of hepatocyte maturation [20, 21, 22]. The secretion of AA1T and LDH was not altered between the groups and at the time points tested (Additional file 1: Figure F).
To evaluate the influence of NPC in key signaling pathways relevant to LO formation, we performed a series of western blots (Fig. 4a) [23, 24, 25, 26, 27, 28, 29]. The protein activity analysis revealed that the III and IIP groups exhibited significantly lower activity of TGF-β (Fig. 4b). Also, the IIP group showed significantly reduced Wnt activity (Fig. 4d), while the III group exhibited increased ERK1/2 activity (Fig. 4c). Activated ERK1/2 inhibits GSK3B through c-Met or IGF receptor signaling , which could explain high β-catenin in III. No differences were observed in the signaling of Notch and BMP4 (Fig. 4e, f). Figure 4h compiles the information obtained from our western blot analyses. TGF-β inhibition increased LO albumin production in vitro by inducing hepatoblast differentiation towards hepatocytes, thereby suppressing cholangiocyte differentiation . Also, TGF-β is positively correlated with lower O2(g) levels and activation of HIF1A during liver organogenesis . In addition, Wnt signaling inhibition is known to control hepatocyte differentiation in 3D culture . The combined inhibition of Wnt and TGF-β significantly increases the expression of albumin (more pronounced by TGF-β inhibition), as opposed of what was observed, when these two pathways were activated (Fig. 3g).
In order to confirm and evaluate the impact of previous western blotting analysis, we performed a proteomic profiling of the various LOs and of the NPC culture secretome. We observed intrinsic and differential protein expression patterns assigned by differential contribution of NPC to LO development (Fig. 4a–e). While sharing most of protein IDs (Additional file 1: Figure S1A), the tested NPC composition significantly influenced LO developmental pathways (Additional file 1: Figure S1B).
Most differentially enriched pathways were related to integrin signaling (Fig. 4c–e). The fibronectin receptor ITGAV (integrin receptor αV) was one of the major hits identified in our pathway enrichment analysis, as well as the integrin alpha subunit 5 (ITGA5). ITGAV expression is induced by TGF-β and acts promoting epithelial-mesenchymal transition  and fibrosis . Integrin subunits α5 and β1 are necessary for bile duct epithelial tract formation during liver development . Also, integrin β1 is important for sustaining hepatocyte viability in native ECM and has been implicated in liver regeneration [36, 37]. Importantly, specific integrin subunit combinations during liver organogenesis, such as α5β1, help to generate the different hepatic structures and are influenced by surrounding sinusoids, vascular development, and local ECM . In our secretome analyses of the NPC cultures, Decorin, a well-known endothelial-produced repressor of liver fibrosis and local inhibitor of TGF-β  and c-Met , was significantly increased in group IIP (Fig. 4h). The reduced secretion of IGFBP5 by dpMSC (Fig. 4h), a MAPK signaling activator overexpressed during fibrosis , could explain the high ERK1/2 in III and reduced β-catenin in IIP. dpMSC produces more ECM, except for collagen type IV, but their role in LO maturation remains unclear.
Collectively, the expression of integrin β1, but not αV, and reduced TGF-β and Wnt signaling observed in the combination of iPS-derived EC and dpMSC, might explain the observed differences in hepatocyte function in various LOs. Our data suggests that high TGF-β activity induced by HAEC (Fig. 3b) increased expression of ITGAV and induced ECM remodeling that impairs hepatocyte maturation. Additionally, we suggest that Wnt signaling repression in IIP is due to reduced secretion of IGFBP5 by dpMSC.
Our data indicates that reduced activity of TGF-β and Wnt contributes for the increased albumin secretion and hepatic function observed in the combination of dpMSC and iPS-derived ECs as NPCs. These differential growth factor stimuli generate substantial changes in integrin and ECM profiles that regulate liver development. In translational terms, this work provides important insights for assessing future strategies to advance organoid technologies aiming at high-throughput drug screening platforms and regenerative therapy approaches.
We would like to thank the funding agencies FAPESP/CEPID, CAPES, INCT, and CNPq. Ernesto Goulart and Luiz Carlos Caires-Junior are FAPESP grantees (grant numbers 2015/14821-1 and 2017/16283-2, respectively). We acknowledge the Mass Spectrometry Facility at Brazilian Biosciences National Laboratory (LNBio), CNPEM, Campinas, Brazil for their support on mass spectrometry analysis. We thank Valdemir Melechco Carvalho from Fleury Laboratory for the support on the proteomic experiments. Authors have no conflict of interest to disclaim. This work is dedicated to Raúl Marcel González Garcia, in memoriam.
EG designed the study under MZ mentorship and prepared the manuscript. EG, LCCJr, KATS, and BHAS performed the cellular experiments and analyzed the data. EG, LCCJr, KATS, GSK, CMM, AFA, and DO performed functional and proteomics analysis. EG, LCCJr, KATS, EC, JAG, SR, PIL, and MZ discussed the results and commented on the manuscript. All authors read and approved the final manuscript.
This work was supported by FAPESP (2013/08028–1).
Ethics approval and consent to participate
The experimental procedures involving samples from human subjects are described in section “Ethics statement.”
Consent for publication
The authors declare that they have no competing interests.
- 10.Miller E, Kobayashi G, Musso C, Allen M, Ishiy F, de Caires L, Goulart E, Griesi-Oliveira K, Zechi-Ceide R, Richieri-Costa A, et al. EIF4A3 deficient human iPSs and mouse models demonstrate neural crest defects that underlie Richieri-Costa-Pereira syndrome. Hum Mol Genet. 2017;26:2177–91.PubMedPubMedCentralCrossRefGoogle Scholar
- 17.Sheyn D, Ben-David S, Shapiro G, De Mel S, Bez M, Ornelas L, Sahabian A, Sareen D, Da X, Pelled G, Tawackoli W, Liu Z, Gazit D, Gazit Z. Human induced pluripotent stem cells differentiate into functional mesenchymal stem cells and repair bone defects. Stem Cells Transl Med. 2016;5(11):1447–60.PubMedPubMedCentralCrossRefGoogle Scholar
- 24.Clotman F, Jacquemin P, Plumb-Rudewiez N, Pierreux CE, Van der Smissen P, Dietz HC, Courtoy PJ, Rousseau GG, Lemaigre FP. Control of liver cell fate decision by a gradient of TGF beta signaling modulated by Onecut transcription factors. Genes Dev. 2005;19(16):1849–54.PubMedPubMedCentralCrossRefGoogle Scholar
- 30.Ng S, Saeb-Parsy K, Blackford S, Segal J, Serra M, Horcas-Lopez M, No D, Mastoridis S, Jassem W, Frank C, et al. Human iPS derived progenitors bioengineered into liver organoids using an inverted colloidal crystal poly (ethylene glycol) scaffold. Biomaterials. 2018;182:299–311.PubMedPubMedCentralCrossRefGoogle Scholar
- 36.Speicher T, Siegenthaler B, Bogorad RL, Ruppert R, Petzold T, Padrissa-Altes S, Bachofner M, Anderson DG, Koteliansky V, Fässler R, Werner S. Knockdown and knockout of β1-integrin in hepatocytes impairs liver regeneration through inhibition of growth factor signalling. Nat Commun. 2014;21(5):3862.CrossRefGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.