In Vitro Generation of Functional Liver Organoid-Like Structures Using Adult Human Cells

In this study we used differentiated adult human upcyte® cells for the in vitro generation of liver organoids. Upcyte® cells are genetically engineered cell strains derived from primary human cells by lenti-viral transduction of genes or gene combinations inducing transient proliferation capacity (upcyte® process). Proliferating upcyte® cells undergo a finite number of cell divisions, i.e., 20 to 40 population doublings, but upon withdrawal of proliferation stimulating factors, they regain most of the cell specific characteristics of primary cells. When a defined mixture of differentiated human upcyte® cells (hepatocytes, liver sinusoidal endothelial cells (LSECs) and mesenchymal stem cells (MSCs)) was cultured in vitro on a thick layer of Matrigel™, they self-organized to form liver organoid-like structures within 24 hours. When further cultured for 10 days in a bioreactor, these liver organoids show typical functional characteristics of liver parenchyma including activity of cytochromes P450, CYP3A4, CYP2B6 and CYP2C9 as well as mRNA expression of several marker genes and other enzymes. In summary, we hereby describe that 3D functional hepatic structures composed of primary human cell strains can be generated in vitro. They can be cultured for a prolonged period of time and are potentially useful ex vivo models to study liver functions.


Introduction
The idea to construct in vitro 3D tissue-like structures to be used as model system for the respective organ is an appealing experimental approach. The main focus hereby is to exploit the in vivo physiological mechanism that occurs during organ development or healing (regeneration) and to implement similar mechanisms to develop a functional tissue in vitro. Such 3D liver-like structures would for example meet the needs of the pharmacological and toxicological industry for drug screening [1]. The main techniques to generate 3D cellular constructs are either the formation of spheroids or building of tissue-like structures by placing sheets of cells and extracellular matrix components on top of each other (reviewed in [2]). The disadvantage of spheroids is that the cells are distributed randomly without formation of spatial organization i.e., liver spheroids neither possess typical hepatic cord-like alignment of polarized hepatocytes nor sinusoids lined with endothelial cells reflecting the in vivo situation. Similarly, 3D liver models generated using sandwich cultures can never fully recapitulate the true in vivo architecture of the organ. Such ex vivo formation of tissue for most complex organs such as heart, kidney or brain would be very challenging. However, the liver is exceptional in its ability to regenerate. It is well established that fully differentiated adult liver is capable of regeneration as long as a sufficient amount of intact liver remains after damage (reviewed in [3] and [4]). Therefore, in principle any differentiated adult liver cell should harbor the potential to proliferate and regenerate to a complex and functional organ under suitable conditions. Indeed, Takebe and colleagues described recently that induced-pluripotent stem cells (iPSC) stimulated to become hepatic endoderm-like cells (iPSC-HE) together with mesenchymal stem cells (MSC) and human umbilical vein endothelial cells (HUVEC), self-organized in vitro into macroscopically visible 3D cell clusters by an intrinsic organizing capacity [5]. When these structures were transplanted into mice, they became vascularized, engrafted into the recipient's tissue and produced hepatic factors like albumin. Possible applications of such an organoid structure include replacement therapy but also the possibility to study hepatotoxic effects of new compounds. They could as well be used as simplified model system to investigate processes like liver regeneration, fibrogenesis or malignant transformation. Due to the fact that these organoids are formed out of different cell types, which are in 3D contact to each other, they can be expected to represent a system which is much closer to the depicted in vivo situation than conventional approaches.
In the present work we analyzed if liver organoids could also be generated from adult, differentiated cells and if these organoids can be cultured for long-term to study liver functions. Instead of stem cells, upcyte 1 hepatocytes were used to reflect the parenchymal cells of the liver. In order to depict the native physiological condition of liver, we further used upcyte 1 liver sinusoidal endothelial cells (LSECs) instead of conventional endothelial cells like HUVECs. LSECs are a specialized type of scavenger endothelial cells that are able to endocytose an array of physiological and foreign macromolecules and colloids from the blood [6]. The upcyte 1 process involves stable transduction of primary cells with lentiviral constructs carrying sequences which code for certain proliferation-inducing factors. These cells are cultured in medium containing a defined mixture of growth factors, allowing tighter control over proliferation (up to 40 population doublings) [7], [8]. Employing this process, almost unlimited numbers of cells from one donor can be obtained. The performance of upcyte 1 hepatocytes compared to primary cells has been extensively studied and it was demonstrated that these cells are highly similar to primary cells [9], [10]. Our results show that by using upcyte 1 cells, liver organoids can be generated and these organoids after culturing them for a period of 10 days, express several marker proteins, genes and enzymes to a degree that is comparable to adult human liver. Furthermore, the architecture of these liver organoids to some degree resembles typical hepatic structures. Potential future applications of such in vitro generated hepatic organoids include pre-clinical drug screening or investigation of liver injury processes like fibrogenesis, cirrhosis or inflammation prior to animal studies, or even as a replacement for animal studies.

Cell culture and liver organoid formation
The upcyte 1 process was performed according to Burkard et al., 2012 [8] and Levy et al. [9].
Cells were cultured at 37°C, 5% CO 2 and 95% humidity. The cells were passaged at 70-80% confluence and living cells were counted using Trypan Blue exclusion. For the generation of liver organoids, required culture plates, bioreactors and pipette tips were pre-cooled at 4°C before use and Matrigel™ was thawed overnight at 4°C. Matrigel™ was diluted 1:1 with upcyte 1 LSEC Growth Medium. For generation of one liver organoid in a 24-well format or in the bioreactors, 380 μl of diluted Matrigel™ (~200 μl/cm 2 ) was added to the plates/chambers and incubated at 37°C for 30 to 45 minutes to allow for polymerization. Trypsinized cells were resuspended in liver organoid growth medium (upcyte 1 Hepatocyte Growth Medium and upcyte 1 LSEC Growth Medium in 1:1 ratio). 1.0 x10 6 upcyte 1 hepatocytes (from the Donor No. 422A-03), 1.0 x10 6 upcyte 1 LSECs and 0.2 x10 6 upcyte 1 MSCs were mixed in 1 ml of liver organoid growth medium, added to the Matrigel™-coated plates/chambers and further incubated at 37°C, 5% CO 2 and 95% humidity for the formation of liver organoids. We tested two setups of bio-reactors, the Quasi-vivo 1 System from Kirkstall and the Live-box 1 system, developed by the group of Prof. Arti Ahluwalia, Faculty of Engineering University of Pisa, Italy (IVTech srl, Lucca, Italy).

In situ hybridization
Detection of mRNA transcripts in paraffin sections was performed as described previously [11], [12]. The protocol is based on the preparation of double-stranded DNA in vitro transcription templates using polymerase chain reaction (PCR), with primers that include RNA polymerase promoter sequences and size-based purification of PCR fragments containing the target gene-specific cDNA and promoter elements for T7 and SP6 RNA polymerase. The primer sequences used are listed in Table 1.

CYP assay
For measurement of CYP activities (CYP2B6, 2C9 and 3A4), the liver organoids cultured for 10 days were incubated with each of the substrate separately. The substrates used were bupropion-HCl, tolbutamide and testosterone to assess CYP2B6, 2C9 and 3A4 activities respectively. The samples were transferred to a fresh 96-well plate and processed for HPLC analysis according to Burkhard et al. [8]. The metabolites were analysed using UV-HPLC. For tolbutamide, bupropion and their metabolites, 0.1 mg/ml chlorpropamide was used as internal standard. For testosterone and its metabolite, 0.5 mM cortexolone was used as internal standard. The mobile phases used for tolbutamide, bupropion and its metabolites were (A)

RNA isolation and real-time PCR
Total RNA was isolated from liver organoids using "high pure RNA isolation kits" from Roche including on column genomic DNA digestion with RNase free DNase (Mannheim, Germany). 1 μg total RNA was reverse transcribed to cDNA using the "Transcriptor First Strand cDNA Synthesis Kit from Roche. For quantitative real-time PCR (using SyberGreen) for the detection of expressions of drug metabolizing enzymes "Prime PCR assays and controls" from BioRad

Results
Spontaneous self-assembly of liver organoids using upcyte 1 cells In order to generate liver organoids we used three different upcyte 1 cell types (all human): hepatocytes, liver sinusoidal endothelial cells (LSECs) and mesenchymal stem cells (MSCs). Fig  1 shows the typical morphology of these cells in monolayer culture and the presence of celltype specific marker proteins. When plated on Matrigel™-coated 24-well plates, defined mixtures of these three cell types self-assembled into a compact organoid structure (Fig 2A). In order to test if this process can be downscaled to formats commonly used in industrial settings, i.e. 48 and 96 well plates, cell number, medium and Matrigel™ volumes were reduced according to the surface area, which still resulted in formation of organoids (S1 Fig). To avoid any nutrient or oxygen-deficiency, liver organoids generated under static conditions (in the bioreactors) were connected to a dynamic system after 24 h and constantly perfused with medium for 10 days. We tested two setups of bio-reactors, the Quasi-vivo 1 System from Kirkstall and the Livebox 1 system, developed by the group of Prof. Arti Ahluwalia, Faculty of Engineering University of Pisa, Italy (IVTech srl, Lucca, Italy) (Fig 2B and 2C). In both systems macroscopically similar liver organoids formed within a comparable time. H&E staining of samples from both systems were compared and in both no strong necrosis was observed (in contrast to static conditions for more than 72h, data not shown). For all further experiments the Kirkstall system was used.

Spatial distribution of the diverse cell types within the organoids
In order to determine the individual localization of the three different cell types at early and late stages of organoid formation, immunohistochemical stainings using antibodies against cell-type specific marker proteins were performed (Fig 3). The results demonstrate that the different cell types are not simply randomly distributed within the organoid but that they rapidly start to form structures with reproducible spatial distribution of hepatocytes and LSECs. Upcyte 1 hepatocytes arranged in closely adherent clusters and moved towards the periphery of Hepatocytes inside the organoid express mRNA for typical markers as well as for many hepatic enzymes and they show typical liver morphology In an attempt to further characterize the functional status of the hepatocytes within the organoids, we performed in situ hybridizations on serial paraffin sections of organoids from early and late stages. The results demonstrate that hepatocytes within the organoid retain a constitutive expression level of typical markers (e.g. cytokeratin 8 and 18) as well as important functional hepatic genes like albumin or enzymes of the glutamine and glucose-metabolism (Fig 4).
To further investigate the functionality and polarity of the hepatocytes within the organoids immunofluorescent double-staining of the epithelial marker protein E-Cadherin and albumin was performed in 10-day liver organoids and was directly compared to adult human liver ( Fig  5A). The presence and membraneous localization of E-Cadherin was further confirmed by immunohistochemistry (Fig 5B). On the mRNA level expressions of E-Cadherin, albumin and the cell-cell contact protein zona occludens 1 (ZO-1) were analyzed by real-time PCR and compared to primary human hepatocytes or whole liver from a total of four different donors ( Fig 5C). In adult, differentiated liver cell proliferation is a rare event (see arrows in Fig 5B, lower right picture). In 10-day liver organoids in contrast we found strong positive staining for the proliferation marker Ki67 (Fig 5B, lower left picture) indicating that such organoids did not yet reach a fully differentiated, quiescent status. In summary these data indicate that at least some portion of the hepatocytes within the liver organoids express polarity markers (E-Cadherin and ZO-1) and synthesize albumin to a degree which is comparable to adult human liver or primary human hepatocytes.

Hepatocytes inside the organoids express drug metabolizing enzymes
To further investigate if hepatocytes show functional properties of liver parenchymal cells, liver organoids were analyzed for basal activity of three prototypic CYPs, CYP-3A4, -2B6 and 2C9. CYP3A4 is an enzyme that metabolizes a large number of clinically important substrates [14]. Drugs metabolized mainly by CYP2B6 include artemisinin, bupropion, cyclophosphamide, efavirenz, ketamine, and methadone [15] while CYP2C9 is the principal isoform of the CYP2C subfamily in the human liver and is involved in the oxidation of several endogenous and xenobiotic compounds, including many therapeutic drugs [16]. The results show that hepatocytes within the liver organoids produce all these 3 enzymes. The basal activity levels were very comparable to that of the primary human hepatocytes which were initially used to generate the  upcyte 1 hepatocyte strain (Fig 6A). On the mRNA level expressions of five additional drug metabolizing enzymes (CYPs 2a6, 2d6, 2e1 as well as Sult1a1 and Ugt1a3) were determined by real-time PCR and were compared to primary human hepatocytes or whole liver from 4 different donors (Fig 6B). While there was an expectable variability in the expression levels between the four donors the levels of expression in the 10-day organoids was mostly within an average range. This further supports the conclusion that hepatocytes within the organoids are functional cells allowing the use of these structures to analyze e.g. hepatic toxicity of certain substances or drugs. Comparing the overall architecture of the parenchymal areas within the LO's with adult human liver we noticed that the cell shape of individual hepatocytes frequently resembled that of primary cells in the liver (see rectangular frames in Fig 6C) and hepatocytes often formed circular arrangements of cells, also resembling similar structures of adult liver. However, the average seize of the nuclei and the relation of nucleus to cytoplasm of the LOhepatocytes is different from the adult liver with nuclei being frequently bigger and showing less cytoplasmic areas (Fig 6C).

Discussion
Shortage of liver for organ transplantation to treat end-stage liver disease and lengthy waiting lists of organ recipients explains the importance of developing functional liver ex vivo through tissue engineering [17]. The intention to reduce animal experiments is of high priority and in addition, animal models can never fully reflect the human cell interactions. Conventional in vitro systems often lack the necessary cellular cross talk and cells in these systems have often lost major properties of the corresponding primary cells, which profoundly limits their possible applications.
In this manuscript we demonstrate that organoid structures with some similarities to adult liver tissue can be generated in vitro using differentiated adult human cells. Similar structures were generated earlier by Takebe et al. [5] using endodermal cells derived from induced pluripotent stem (iPS) cells combined with HUVECs and mesenchymal stem cells (MSCs). The authors demonstrated that MSCs play a pivotal role for this three-dimensional structure formation. MSCs are multipotent, self-renewing, adult stem cells, which can differentiate into hepatocytes (reviewed in [18]). Such differentiation can be induced by addition of growth factors like hepatocyte growth factor (HGF) or basic fibroblast growth factor (bFGF) [19] or by co-culturing the cells with hepatocytes [20], [21].
Because MSC-derived hepatocytes acquire the ability to express hepatocyte marker genes like albumin it would be difficult to distinguish between the original hepatocytes and MSCderived hepatocytes in our 10-day organoids, therefore we cannot exclude that at least some of the hepatocytes in the organoids originated from MSCs. However, direct differentiation of MSCs into hepatocytes might not represent the major mechanism how MSCs support survival of liver tissue as it has been shown that intravenous bolus of conditioned medium from MSC can already provide a significant survival benefit in rats undergoing fulminant liver failure [22]. Therefore the possible hepatoprotective effect of MSCs seems to be mainly based on the secretion of factors which on the one side may modulate cells of the immune system (which are not present in our setup) but also exert direct anti-apoptotic effects on hepatocytes [23]. The exact nature of these MSC-derived factors that might be essential for organoid formation are yet to be explored. However, Takebe et al. have already described that application of inhibitors against bone morphogenetic proteins (BMPs) or fibroblast growth factor (FGF) effectively prevented the structure formation in their setting [5].
The data presented here demonstrate that differentiated adult upcyte 1 human cells can be used for the formation of three-dimensional hepatic structures in vitro which are viable and functional up to a period of 10 days in culture (longer culture times are currently under investigation). Such structures resemble the three-dimensional architecture of the in vivo human liver better than conventional monolayer cultures consisting of hepatoma cell lines or isolated primary cells. Also, availability of these primary cells from human sources is rather sporadic and not accessible for everybody. Apart from this, primary cells show a highly unstable phenotype in culture and produce very variable results due to donor differences and/or varying effectiveness of the isolation process itself. Upcyte 1 cell strains which exhibit primary cell phenotypes can be obtained in large amounts from the same donor which meets the demand for comprehensive sets of experiments. Due to the standardized protocol, results obtained with this system are highly reproducible. The use of such organoid liver structures generated in vitro is therefore suitable for a large range of applications such as the testing of subchronic or chronic toxicity caused by slow metabolized compounds and in addition it may help to reduce the number of animal experiments required for therapeutic drug development in the future.
Our results show that hepatocytes within such organoids express epithelial membrane markers like E-Cadherin and ZO-1, implying that compared to monolayer cultures they retain a more polarized morphology. We also show that the cells express albumin and produce several drug metabolizing enzymes to a degree that is comparable to adult human liver. This demonstrates that the liver organoids represent a 3D set up which is most likely superior to monolayer cultures of primary hepatocytes where the cells rapidly lose polarity and enzymatic functionality. Nevertheless, the cells within the organoids showed very high proliferative activity which might point to the conclusion that after 10 days the structures are still developing and that they did not (yet) reach a fully differentiated status.
Ongoing experiments aim at further optimizing the parameters for long term culture and differentiation of liver organoids. Additionally, integration of further liver cell types like Kupffer cells and hepatic stellate cells is planned in order to generate structures which resemble even more closely the in vivo conditions.