Endodermal and Hepatic Differentiation from Human Embryonic Stem Cells and Human Induced Pluripotent Stem Cells

1Laboratory of Stem Cell Regulation, National Institute of Biomedical Innovation, Osaka 567-0085, Japan 2Laboratory of Biomedical Innovation, Graduate School of Pharmaceutical Sciences, Osaka University, Osaka 565-0871, Japan 3Laboratory of Biochemistry and Molecular Biology, Graduate School of Pharmaceutical Sciences, Osaka University, Osaka 565-0871, Japan 4The Center for Advanced Medical Engineering and Informatics, Osaka University, Osaka 565-0871, Japan


Introduction
The liver has many functions, including carbohydrate metabolism, glycogen storage, lipid metabolism, urea synthesis, drug detoxification, production of plasma proteins, and destruction of erythrocytes. The liver is composed of several types of cells, including epithelial, endothelial, and hematopoietic cells. Of these cells, hepatocytes play the most important role in major hepatic functions. Hepatocytes are thus useful cells for biomedical research, regenerative medicine, and drug discovery. They are particularly useful for drug screenings, such as for the determination of metabolic and toxicological properties of drug compounds in in vitro models. For these applications, however, it is necessary to prepare a large number of the functional hepatocytes, which can no longer proliferate in in vitro culture. Isolated primary hepatocytes are the current standard in vitro model, because they express large amounts of drug-metabolizing enzymes and transporters [1]. However, isolated hepatocytes lose their differentiated properties, such as some cytochrome P450 activities that are induced by reference compounds, even under the optimized culture conditions [2,3]. Moreover, it can be difficult to set up long-term cultures with primary hepatocytes, because they can no longer proliferate in in vitro culture [4].
Human embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) are able to replicate indefinitely and differentiate into most cell types of the body, and have the potential to provide an unlimited source of cells for a variety of applications [5][6][7][8]. Among the differentiated cells from ESCs and iPSCs, induced hepatocytes have a wide range of potential applications in biomedical research, drug discovery, and the treatment of liver disease. In this review, we provide an up-to-date overview of the wide variety of endodermal and hepatic differentiation protocols. These protocols were designed to reconstruct the in vivo environment in a variety of ways, including by addition of soluble factors into culture medium, transduction of differentiationrelated genes, co-cultivation with other lineage cells, and use of a threedimensional culture system.

Definitive Endoderm Differentiation from ESCs
Gastrulation of the vertebrate embryo starts with the formation of three germ layers: the ectoderm, mesoderm, and endoderm. The endoderm contributes to the digestive and respiratory tracts and their associated organs [9]. The endoderm differentiates into various organs, including the liver, pancreas, lungs, intestine, and stomach. To examine the molecular mechanisms of endoderm specification during early embryogenesis, endoderm differentiation from ESCs has been widely investigated as an in vitro model [10]. It has been reported that mouse ESCs have the ability to differentiate into definitive endoderm (DE) cells [11][12][13]. In recent studies, specific growth factors are used to generate DE cells from ESCs. In DE differentiation, it is well known that nodal signaling plays a crucial role and induces the expression of endoderm-related genes [14]. Activin A, a member of the nodal family, is a ligand of the type II activin receptor and can transmit a downstream signal by using Smad adaptor proteins [15][16][17][18]. Therefore, activin A is widely used to generate DE from ESCs. Although embryoid body (EB) formation is also used in the differentiation of ESCs, activin A could generate DE more efficiently than the EB formation [19]. In addition, using activin A with other factors such as fibroblast growth factor (FGF) 2 or Wnt3a proved to be more effective.

Hepatic Maturation from ESC-derived Hepatoblasts
Hepatoblasts differentiate into two distinct lineages, hepatocytes and cholangiocytes. During the fetal hepatic maturation, the number of hepatoblasts decreases, and in turn, the number of mature hepatocytes increases [42]. In this process, AFP is highly expressed in the fetal liver, and then the number of AFP-positive cells decreases in a later maturation step and almost disappears in the adult liver [43,44] Because drug discovery is one of the most anticipated applications of ESC-derived hepatocyte-like cells, it is important to generate ESCderived hepatocyte-like cells that have the same characteristics as primary human hepatocytes. Even when the various hepatic functions described above are observed in ESC-derived hepatocytes, expression level of hepatocyte-related genes in ESC-derived hepatocytes is often lower than that of human hepatocytes [59]. To generate functional hepatocytes which have characteristics similar to primary human hepatocytes, exogenous transduction of transcription factor genes that can control the expression of hepatocyte-related genes is suitable for efficient differentiation of hepatocyte-like cells from ESCs. Sequential transduction of the SOX17, HEX, and HNF4α genes, which are central regulators of liver development, in ESC-derived hepatoblasts has been shown to successfully induce mature hepatocyte-like cells that have the same features as primary human hepatocytes [60] ( Figure 1). Furthermore, these hepatocyte-like cells could catalyze the toxication of several compounds, suggesting that the ESC-derived hepatocytes have potential for use in drug-screening applications. Overexpression of the Foxa2, Hnf4α, and c/EBPα genes into expandable liver-derived progenitor cells resulted in mature hepatocyte phenotypes [61]. Many other studies have shown the effect of the transduction of differentiation-related genes to promote hepatic differentiation from various origins (summarized in Table 1) [24,25,27,28,38,39,60,61,62-67], demonstrating that transduction of differentiation-related genes into ESCs would be a powerful strategy to generate mature hepatocytelike cells.

Hepatic Differentiation from iPSCs
The iPSC technology raises the possibility of generating patientspecific cell types of all lineages [68,69]. Because drug metabolism capacity differs among individuals [70], it is difficult to make a precise  of drug toxicity by using primary human hepatocytes isolated from a single donor. A hepatotoxicity screening utilizing iPSC-derived hepatocyte-like cells would allow the investigation of individual drug metabolism capacity [71][72][73][74][75][76][77]. A study has shown the generation of hepatocyte-like cells from patient-specific human iPSCs [78][79][80]. In the same study, it was demonstrated that patientspecific iPSC-derived hepatocytes are a potential source for modeling diseases whose phenotypes are caused by protein dysregulation within adult cells. A novel drug discovery that reflects the individual genetic information would be possible by using an iPSC library representing different ethnic groups, sexes, and disease phenotypes.

Hepatic Differentiation by Co-culture and Threedimensional Culture
In order to facilitate maturation of the ESC-or iPSC-induced hepatocyte-like cells and to enhance the differentiation efficiency of those cells, development of a differentiation system that more closely mimics progenitor development in vivo will be needed. Such culture system is also relevant to the culture of primary hepatocytes. Normal culture condition of hepatocytes in vitro differs substantially from the environment in vivo. Thus, it is difficult to maintain the physiological function of the hepatocytes. To overcome this difficulty, development of a culture system for highly functional hepatocytes is required. So far, co-culture methods with other lineage cells and three-dimensional culture methods have been used to support these challenges.
Co-culture methods have been attempted with primary hepatocytes and other kinds of cells [81][82][83][84][85], because cell-cell interactions are important in embryogenesis and organogenesis. In particular, heterotypic cell-cell interactions in the liver, such as interactions of parenchymal cells with non-parenchymal cells, play a fundamental role in liver function [86]. It has been reported that small hepatocytes could be induced to differentiate into mature hepatocytes by co-culturing with non-parenchymal cells in vitro [87]. Cell-cell interactions between embryonic cardiac mesoderm and definitive endoderm have been shown to be essential for liver development [88]. Transcription factors that are critical for hepatic development have been identified from these cell-cell interactions [88]. ES cells co-cultured with cardiac mesoderm showed spontaneous differentiation into hepatocytes [89]. These results suggest that the combined differentiation methods, such as addition of soluble factors into culture medium, transduction of differentiation-related genes or co-cultivation with other lineage cells, may further enhance the differentiation and maturation efficiency of hepatocytes.
Recently, numerous three-dimensional (3D) culture methods have been reported. Among these, the spheroid culture methods, which include the hanging-drop method and the float-culture method using culture dishes coated with non-adherent polymer, have been widely used to culture primary hepatocytes in vitro. As various micropatterning technologies have been developed, various micro-patterned substrates, employing both surface engineering and synthetic polymer chemistry for utilizing spheroid culture, have been reported [90,91]. Spheroid culture methods permit the maintenance of liver-function of primary hepatocytes in comparison with the two-dimensional (2D)culture.
The bioreactor method is also used for culturing primary hepatocytes. By studying various optimized conditions, flow conditions [92] and cell densities [93], this system has not only shown advantages in terms of maintaining the functions of primary hepatocytes in vitro in comparison with 2D-culture [94,95], but also has shown effects of spontaneous differentiation from ESCs into hepatocytes [96,97]. It has been reported that 3D culture using a bioreactor induces more functional maturation in hepatocytes differentiated from ESCs than 2D-culture [97]. The 3D culture methods using polymer scaffold systems have also demonstrated effectiveness both in culturing primary hepatocytes [98,99] and in differentiation from ESCs into hepatocytes in vitro [100][101][102]. These data showed that hepatocytes could be induced from ESCs on a polymer scaffold. ALB expression was detected earlier and the mRNA expression level of ALB was higher than in 2D culture. Furthermore, cell-sheet engineering has recently been reported [103,104]. Cell-sheet 3D culture was performed by using a culture dish coated with a temperature-responsive polymer, poly (N-isopropylacrylamide) [105][106][107]. Some groups have adopted culture methods with a combination of 3D culture and co-culture and showed that the liver function of primary hepatocytes could be maintained more strongly and longer than without co-culture conditions [108][109][110]. These combined methods will likely be a more effective differentiation condition to gain mature hepatocytes from ESCs and iPSCs.

Transplantation of Human ESC-or iPSC-derived Hepatocyte-like Cells
Because of the species differences between humans and other animals, it is difficult to apply biological phenomena of animals to humans in the early phase of drug screening [111]. It is known that chimera mice with human hepatocytes would be a powerful tool to predict drug toxicity and drug metabolism in vivo [112][113][114][115]. In addition, chimera mice are useful to investigate the molecular mechanisms involved in infection with human hepatitis B virus (HBV) and HCV, because there is no suitable small animal model for such study [116][117][118]. However, large amounts of human hepatocytes must  be prepared for these technologies, thus requiring large numbers of chimera mice. If it becomes possible to generate a robust chimera mouse model with hepatocyte-like cells differentiated from human ESCs or iPSCs, then chimera mice with humanized livers could be widely used in pharmaceutical development. To this end, several groups have reported the generation of chimera mice with hepatocytelike cells differentiated from human ESCs and iPSCs. Cai et al. reported that human ESC-derived hepatocyte-like cells were transplanted into the carbon tetrachloride (CCl 4 )-injured liver of severe combined immunodeficiency (SCID) mice and human alpha-1-antitrypsin (AAT) expression was detected in the liver [37]. Touboul et al. [119] showed that human ESC-derived hepatocyte-like cells can engraft and express human ALB and AAT in the liver of urokinase-type plasminogen activator-transgenic Rag2IL-2Rg -/-(uPA-Rag2IL-2Rg -/-) mice. Duan et al. [120] reported that human ESC-derived hepatocytelike cells were transplanted into the liver of NOD.CB17-Prkdc scid / NcrCrl (NOD/SCID) mice and a significant level of human ALB was detected in the recipient mouse serum. Basma et al.
[49] generated chimera mice and rats that secreted higher levels of human ALB than previously reported chimera mice. They sorted human ESC-derived hepatocyte-like cells based on surface asialoglycoprotein-receptor 1 (ASGPR1) expression and injected them into the spleen of uPA-SCID mice. Thereafter, they detected a much higher level of human ALB and human AAT in the mouse serum on day 75 after transplantation. They also performed transplantation into Nagase analbuminemic rats treated with both retrorsine, which can prevent proliferation of rat hepatocytes, and FK506, which can suppress immune response, after partial hepatectomy, demonstrating that large clusters of engrafted cells were observed in these rats and human ALB levels were reached at 20,000 ng/ml [49].
The growth speed of hepatocyte-like cells is slower than that of DE cells and hepatoblasts, both of which are immature stage cells as compared with hepatocyte-like cells [60]. It is likely that immature cells can proliferate better than mature cells in the mouse liver. Therefore, several groups have attempted to transplant DE cells or hepatoblasts. In one such attempt, human ESC-derived DE cells were successfully engrafted into the livers of NOD/SCID mice, which were treated with CCl 4 and retrorsine, and these mice expressed human AAT in the liver [57]. Recently, Liu et al. [121] compared the engraft efficiency of human ESC-derived multi-stage hepatic cells. They transplanted human DE, hepatoblasts and hepatocyte-like cells differentiated from human ESCs into the dimethylnitrosamine-injured liver of NOD/Lt-SCID/IL-2Rg −/− (NSG) mice, demonstrating that at low cell dosages, the engraftment efficiency of DE cells was slightly higher than that of hepatoblasts and hepatocyte-like cells differentiated from human ESCs. These results suggest that DE cells, which have proliferative capability, can regenerate liver better than hepatocyte-like cells, which have lower proliferative capability.
These technologies, which use ESC-derived cells, can be applied to iPSC-derived hepatocyte-like cells. Si-Tayeb et al.
[59] injected human ESC-and iPSC-derived hepatocyte-like cells into the liver of neonatal mice and they detected human ALB expression clusters. Liu et al. [121] also transplanted human ESC-and iPSC-derived hepatocyte-like cells into mice, and achieved similar results. These findings indicate that human iPSC-derived hepatocyte-like cells can engraft into the rodent liver in a manner similar to human ESC-derived hepatocyte-like cells.
Although human ESC-or iPSC-derived hepatocyte-like cells can engraft in the mouse liver, the human ALB levels in chimera mice engrafted with human ESC-or iPSC-derived hepatocyte-like cells are much lower than those in chimera mice engrafted with human primary hepatocytes [49,112,117,121], suggesting that the efficiency of replacement in chimera mice generated with human ESC-or iPSCderived hepatocyte-like cells would be low. Therefore, the chimerism of mice with human ESC or iPSC-derived hepatocyte-like cells should be improved to apply this technology to industrial applications.

Conclusions
In this review, we have described several protocols that could promote the differentiation of human ESCs or iPSCs into endodermal and hepatic cells. These methods are all based on the in vivo developmental process of embryos. In the future, by using a combination of these protocols or through the discovery of molecular findings about liver development, more efficient protocols for hepatic differentiation could be developed for regenerative medicine and drug development.