Efficient generation of hepatoblasts from human ES cells and iPS cells by transient overexpression of homeobox gene HEX.

Human embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) have the potential to differentiate into all cell lineages, including hepatocytes, in vitro. Induced hepatocytes have a wide range of potential application in biomedical research, drug discovery, and the treatment of liver disease. However, the existing protocols for hepatic differentiation of PSCs are not very efficient. In this study, we developed an efficient method to induce hepatoblasts, which are progenitors of hepatocytes, from human ESCs and iPSCs by overexpression of the HEX gene, which is a homeotic gene and also essential for hepatic differentiation, using a HEX-expressing adenovirus (Ad) vector under serum/feeder cell-free chemically defined conditions. Ad-HEX-transduced cells expressed α-fetoprotein (AFP) at day 9 and then expressed albumin (ALB) at day 12. Furthermore, the Ad-HEX-transduced cells derived from human iPSCs also produced several cytochrome P450 (CYP) isozymes, and these P450 isozymes were capable of converting the substrates to metabolites and responding to the chemical stimulation. Our differentiation protocol using Ad vector-mediated transient HEX transduction under chemically defined conditions efficiently generates hepatoblasts from human ESCs and iPSCs. Thus, our methods would be useful for not only drug screening but also therapeutic applications.

Introduction

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,[1,2,3,4] and thereby have the potential to provide an unlimited source of cells for a variety of applications.[5] Hepatocytes are useful cells for biomedical research, regenerative medicine, and drug discovery. They are particularly applicable to drug screenings, such as for the determination of metabolic and toxicological properties of drug compounds in in vitro models, because the liver is the main detoxification organ in the body.[6] For these applications, it is necessary to prepare a large number of functional hepatocytes from human ESCs and iPSCs. Many of the existing methods for cell differentiation of human ESCs and iPSCs into hepatocytes employ undefined, serum-containing medium and feeder cells.[7,8,9] Preparation of human ESC- and iPSC-derived hepatocytes for therapeutic applications and drug toxicity testing in humans should be done in nonxenogenic culture systems to avoid potential contamination with pathogens. Furthermore, the efficiency of the differentiation of the human ESCs and iPSCs into hepatocytes is not particularly high using these methods.[9,10,11,12,13,14]

In vertebrate development, the liver is derived from the primitive gut tube, which is formed by a flat sheet of cells called the definitive endoderm.[5,15] Shortly afterwards, the definitive endoderm is separated into endoderm derivatives containing the liver bud, the cells of which are referred to as hepatoblasts. The hepatoblasts have the potential to proliferate and differentiate into both hepatocytes and cholangiocytes. In the process of hepatic differentiation, the maturation is characterized by the expression of liver- and stage-specific genes. For example, α-fetoprotein (AFP) is an early hepatic marker, which is expressed in hepatoblasts in the liver bud until birth, and its expression is dramatically reduced after birth.[16] In contrast, albumin (ALB), which is the most abundant protein synthesized by hepatocytes, is initially expressed at lower levels in early fetal hepatocytes, but its expression level is increased as the hepatocytes mature, reaching a maximum in adult hepatocytes.[17] Furthermore, isoforms of cytochrome P450 (CYP) proteins also exhibit differential expression levels according to the developmental stages of the liver. Although most CYPs (including CYP3A4, CYP7A1, and CYP2D6) are only slightly expressed or not detected in the fetal liver tissue, the expression levels are dramatically increased after birth.[18] For the development of hepatoblasts, numerous transcription factors are required, such as hematopoietically expressed homeobox (HEX), GATA-binding protein 6, prospero homeobox 1, and hepatocyte nuclear factor 4A.[15,19] Among them, HEX is suggested to function at the earliest stage of hepatic linage.[20] HEX is first expressed in the definitive endoderm and becomes restricted to the future hepatoblasts. Targeted deletion of the HEX gene in the mouse results in embryonic lethality and a dramatic loss of the fetal liver parenchyma.[19,21,22] The hepatic genes, including ALB, prospero homeobox1, and hepatocyte nuclear factor 4A, are transiently expressed in the definitive endoderm of HEX-null embryos, and further morphogenesis of the hepatoblasts does not occur.[23] In general, then, HEX is essential for the definitive endoderm to adopt a hepatic cell fate.

Adenovirus (Ad) vectors are one of the most efficient gene delivery vehicles and have been widely used in both experimental studies and clinical trials.[24] Ad vectors are attractive vehicles for gene transfer because they are easily constructed, can be prepared in high titers, and provide high transduction efficiency in both dividing and nondividing cells. We have developed efficient methods for Ad vector-mediated transient transduction into mouse ESCs and iPSCs.[25,26] We have also showed that the differentiations of mouse ESCs and iPSCs into adipocytes and osteoblasts were dramatically promoted by Ad vector-mediated peroxisome proliferator activated receptor γ and runt related transcription factor 2 transduction, respectively.[25,26]

In this study, we hypothesized that transient HEX transduction could efficiently induce hepatoblasts from human ESCs and iPSCs. A previous study demonstrated that HEX regulates the differentiation of hemangioblasts and endothelial cells from mouse ESCs,[27] whereas the role of HEX in the differentiation of hepatoblasts from human ESCs and iPSCs remains unknown. We found that differentiation of hepatoblasts from the human ESC- and iPSC-derived definitive endoderms, but not from undifferentiated human ESCs and iPSCs, could be facilitated by Ad vector-mediated transient transduction of a HEX gene. Furthermore, the Ad-HEX-transduced cells that were derived from human iPSCs were able to differentiate into functional hepatocytes in vitro. All the processes for cellular differentiation were performed under serum/feeder cell- free chemically defined conditions. Our culture systems and differentiation method based on Ad vector-mediated transient transduction under chemically defined conditions would provide a platform for drug screening as well as safe therapies.

Results

Differentiation of human ESC- and iPSC-derived definitive endoderms

Our three-step differentiation protocol is illustrated in . After treatment with 50 ng/ml of Activin A (high-dose) and basic fibroblast growth factor (bFGF) for 5 days on a laminin-coated plate, morphologically, the human ESCs and iPSCs were gradually transformed from typical, defined, tight human ESC, and iPSC colonies (day 0) into less dense, flatter cells containing prominent nuclei (day 5), even though the majority of the cells had a morphology resembling that of undifferentiated cells (). FACS analysis showed that ~46% of human iPSC-derived differentiated cells expressed CXCR4 (expressed in the definitive endoderm but not the primitive endoderm) (Supplementary Figure S1a). Human ESC- and iPSC-derived differentiated cells were immunostained with the definitive endoderm marker, FOXA2 (). However, the majority of the cells expressed the pluripotent marker NANOG, indicating that undifferentiated cells remain in the induced cultures at day 5. After the cells were passaged with trypsin-EDTA and seeded on a laminin-coated plate a second time, the resultant cells were found to be more homogeneous and flatter at day 6 (). Semiquantitative analysis by counting immunopositive cells revealed that the number of FOXA2-positive cells was increased and, in turn, the number of NANOG-positive cells was decreased at day 6 after passaging (). Real-time reverse transcriptase (RT)-PCR analysis showed that the definitive endoderm markers FOXA2 and SOX17 mRNA were upregulated, whereas the pluripotent marker NANOG mRNA was downregulated at day 6 (). These results were consistent with the immunofluorescence results (). The expression levels of the mesoderm marker FLK1 mRNA and ectoderm marker PAX6 mRNA were downregulated or unchanged at day 6 (Supplementary Figure S1b–e). Importantly, the expression of SOX7 mRNA (expressed in the extra-embryonic endoderm but not the definitive endoderm) was downregulated (). These results indicate that the definitive endoderm is induced or selected from human ESCs and iPSCs after passaging. We obtained the same results using another human iPSC line (Supplementary Figure S2a–d).

HEX induces hepatoblasts from the human ESC- and iPSC-derived definitive endoderms

To investigate whether forced expression of transcription factors could promote hepatic differentiation, the human ESC- and iPSC-derived definitive endoderms were transduced with Ad vectors. We used a fiber-modified Ad vector containing the elongation factor-1α promoter and a stretch of lysine residue (K7) peptides in the C-terminal region of the fiber knob to examine the transduction efficiency in the human ESC- and iPSC-derived definitive endoderms. The elongation factor-1α promoter was found to be highly active in human ESCs.[28] The K7 peptide targets heparan sulfates on the cellular surface, and the fiber-modified Ad vector containing K7 peptides was shown to be efficient for transduction into many kinds of cells.[29,30] The human ESC- and iPSC-derived definitive endoderms were transduced with a LacZ-expressing Ad vector (Ad-LacZ) at 3,000 vector particle/cell. X-Gal staining showed that the Ad-LacZ-transduced human ESC- and iPSC-derived definitive endoderms successfully expressed LacZ (). Nearly 100% of the cells transduced with Ad-LacZ were strongly X-gal positive. The transduction efficiency in the human ESC- and iPSC-derived definitive endoderms transduced with the conventional Ad vector containing the wild-type capsid at 3,000 vector particle/cell was ~80% and X-gal staining was much weaker than that in the cells transduced with fiber-modified Ad vectors (Supplementary Figure S6). Next, the human ESC- and iPSC-derived definitive endoderms were transduced with a HEX-expressing fiber-modified Ad vector (Ad-HEX). Although HEX is known to be a transcription factor that is essential for liver development, it remains unclear what the effect of transient HEX overexpression is on differentiation from human ESCs and iPSCs or their derivatives in vitro. We confirmed the overexpression of HEX in the human ESC- and iPSC-derived definitive endoderms transduced with Ad-HEX (Supplementary Figure S3a–f). Gene expression analysis revealed the upregulation of AFP mRNA, which was expressed by hepatoblasts or early hepatocytes, in Ad-HEX-transduced cells as compared with nontransduced cells or Ad-LacZ-transduced cells (). Expression of ALB mRNA, which is the most abundant protein in liver, was also observed in Ad-HEX-transduced cells ().

During liver development, both hepatocytes and cholangiocytes were differentiated from the hepatoblasts. We examined the protein expression of AFP, ALB, and the cholangiocyte marker cytokeratin 7 (CK7) in Ad-HEX-transduced cells by immunostaining (). The AFP-positive populations were detected in Ad-HEX-transduced cells (). ALB-positive cells were also detected, although the detection efficiency was very low (). CK7-positive cells were observed among the Ad-HEX-transduced cells, and all CK7-positive cells were found near the AFP- and ALB-positive cells, suggesting that hepatoblasts are generated by the transient overexpression of a HEX gene. Semiquantitative RT-PCR analysis showed that the expression levels of the liver-enriched transcription factors hepatocyte nuclear factor 1A, hepatocyte nuclear factor 1B, hepatocyte nuclear factor 4A, and hepatocyte nuclear factor 6 mRNA were upregulated in Ad-HEX-transduced cells (Supplementary Figure S4a,b). The expressions of CCAAT/enhancer binding protein α and prospero homeobox 1 mRNA, two transcription factors known to play a pivotal role in the establishment of the hepatoblasts, were also induced in Ad-HEX-transduced cells (Supplementary Figure S4a,b). Taken together, these findings indicate that HEX enhances the specification of hepatoblasts from the human ESC- and iPSC-derived definitive endoderms. Similar results were obtained with another human iPSC line (Supplementary Figure S2e–g).

Time course of differentiation of the definitive endoderm to hepatoblasts

Next, we examined the time course of AFP and CK7 expression during differentiation of human iPSCs to hepatoblasts in Ad-HEX-transduced cells and nontransduced cells. At day 7 (the day after transduction), the expression of AFP was not detectable in Ad-HEX-transduced or nontransduced cells (Supplementary Figure S5a,d). At day 8–9, morphological changes to hepatocyte-like cells were observed in Ad-HEX-transduced cells (Supplementary Figure S5h,i). We also observed homogeneous AFP-positive cells at day 9 (Supplementary Figure S5e). At day 10, CK7-positive cells appeared, indicating that hepatoblasts started to differentiate into hepatocytes and cholangiocytes at day 9–10 (Supplementary Figure S5f). At day 12, ALB-positive cells appeared, indicating that hepatocytes were differentiated from Ad-HEX-transduced cells (). These results showed that HEX induces the hepatoblasts from the definitive endoderm, and the Ad-HEX-transduced cells could differentiate into both hepatocytes and cholangiocytes.

Directed hepatic differentiation from hepatoblasts

With the protocol described above, heterogeneous populations containing CK7-positive cholangiocytes were observed at day 12 (). To promote the differentiation of hepatoblasts to hepatocytes, the human iPSC-derived differentiated cells at day 9 (Supplementary Figure S5e) were dislodged with trypsin-EDTA and plated on collagen I-coated dishes as previously reported.[11] After 8–11 days in culture with medium containing FGF4, HGF, OSM, and DEX, the Ad-HEX-transduced cells became more flattened (Supplementary Figure S5m), whereas the nontransduced cells became fibroblast-like cells (Supplementary Figure S5i). Gene expression analysis showed the upregulation of ALB mRNA in Ad-HEX-transduced cells under this culture condition, whereas the expression of ALB mRNA was reduced in the nontransduced cells at day 18 (). Immunostaining showed that only a small percentage of Ad-HEX-transduced cells expressed ALB at day 12 (), whereas most of the Ad-HEX-transduced cells were ALB-positive at day 18 (). Most of the Ad-HEX-transduced cells also expressed CYP3A4 at day 18 (). More importantly, in the Ad-HEX-transduced cells, CYP7A1 and cytokeratin 18 were detected and these proteins are known to be detected in hepatocytes but not in extra-embryonic cells[31,32] (). Quantitative analysis showed that ~84, 80, 88, and 92% of Ad-HEX-transduced cells expressed ALB, CYP3A4, CYP7A1, and cytokeratin 18, respectively. These results indicate that Ad-HEX-transduced cells could differentiate to hepatic cells. However, the expression level of ALB mRNA in Ad-HEX-transduced cells was lower than that in fetal liver tissue and in turn, the expression of AFP mRNA was maintained (). Therefore, Ad-HEX-transduced cells are committed to the hepatic linage, but are not yet mature hepatocytes.

Ad-HEX-transduced cells exhibit hepatic functions

To test the hepatic function in the Ad-HEX-transduced cells, we investigated the liver metabolism, because P450 cytochrome enzymes play a critical role in this function. We examined the expression level of several members of this multigene family, i.e., CYP3A4, CYP7A1, mRNA and CYP2D6 in Ad-HEX-transduced cells by real-time RT-PCR. The real-time RT-PCR analysis showed that the mRNAs for CYP3A4, CYP7A1, and CYP2D6 were expressed in Ad-HEX-transduced cells, whereas none of these mRNAs were expressed in the nontransduced cells (). The expression levels of CYP3A4 in Ad-HEX-transduced cells were similar to those observed in primary human hepatocytes, which were cultured 48 hours after plating the cells, or fetal liver tissues but lower than those in adult liver. The CYP2D6 and CYP7A1 mRNA expressions in Ad-HEX-transduced cells were lower than those in primary hepatocytes or adult tissues. Next, we investigated the metabolism of the P450 3A4 substrates by measuring the activity of P450 isozymes. The metabolites were detected in Ad-HEX-transduced cells, and their activity was 3.4-fold higher than that in the most commonly used human hepatocyte cell line, HepG2 (; DMSO column). This result was consistent with the real-time RT-PCR data (). We further tested the induction of CYP3A4 upon chemical stimulation, because CYP3A4 is the most prevalent P450 isozyme in the liver and is involved in the metabolism of a significant proportion of the currently available commercial drugs. Because CYP3A4 can be induced with rifampicin, both Ad-HEX-transduced cells and HepG2 cells were treated with rifampicin, followed by treatment with CYP3A4 substrate. Ad-HEX-transduced cells produced 5.4-fold higher levels of metabolites in response to rifampicin treatment (; rifampicin column). This result indicates that P450 isozymes are active in Ad-HEX-transduced cells.

Discussion

The object of this study was to develop an efficient method for generating hepatoblasts and hepatocytes from human ESCs and iPSCs for application to drug toxicity screening tests as well as therapeutics such as regenerative medicine. We found that transient HEX transduction in the definitive endoderm together with a culture under chemically defined conditions was useful for this purpose.

It has been reported that a high concentration of Activin A induces differentiation of human ESCs into the definitive endoderm.[8,33,34] On the other hand, undifferentiated human ESCs are maintained by a low concentration of Activin A.[35] Several studies have shown that bFGF promotes the differentiation of ESCs into the definitive endoderm and inhibits the differentiation of ESCs into the extra-embryonic endoderm.[35,36,37,38] bFGF has been reported to inhibit the BMP signaling, which can promote the extra-embryonic lineage differentiation.[39] The extra-embryonic endoderm expresses most of the hepatocyte markers, such as AFP.[40] Contamination of the extra-embryonic endoderm makes it difficult to estimate the hepatic differentiation from human ESCs and iPSCs.[11,14,40] In this study, we showed that both Activin A and bFGF induce definitive endoderm populations, while they repress the extra-embryonic endoderm differentiation (). Interestingly, after the differentiated cells that were cultured on laminin-coated plates with Activin A and bFGF were passaged at day 5, FOXA2-positive cells (definitive endoderm) were enriched in the resultant cells at day 6 (). This may have been because FOXA2-positive cells efficiently adhered to the laminin-coated plate and/or because trypsinized, single undifferentiated ESCs/iPSCs cannot survive. The passaging of differentiated cells might be attributed to the reduction in the number of not only the extra-embryonic endoderm cells but also the undifferentiated cells. However, the efficiency of the definitive endoderm differentiation in this study was not as efficient as that reported by other groups.[8,33,34] Other cell lineages, such as the mesoderm and extra-embryonic endoderm, might remain at day 6 ( and Supplementary Figure S1). Further improvement of the culture conditions will thus be needed in order to enhance the definitive endoderm differentiation.

Hepatoblasts and hepatocytes were differentiated from the human ESC- and iPSC-derived definitive endoderms by transient overexpression of the homeobox gene HEX. A fiber-modified Ad vector containing K7 peptides mediated much higher gene expression than conventional Ad vectors in the human ESC- and iPSC-derived definitive endoderms (Supplementary Figure S6). This new hepatic differentiation protocol shows that HEX induces AFP-positive hepatoblasts at day 9 and ALB-positive hepatocytes at day 12 from human ESCs and iPSCs, whereas the previous protocols require a few weeks or months to induce AFP- and ALB-positive hepatocytes from PSCs.[9,10,11] Previous studies suggested that HEX could regulate liver-enriched transcription factors such as hepatocyte nuclear factor 4A and hepatocyte nuclear factor 6.[19,23] Overexpression of the HEX gene under the conditions employed in the present study could activate several transcription factors that are required for hepatic differentiation (Supplementary Figure S4a,b). However, the Ad-HEX-transduced cells showed a low level of expression of ALB and some CYP450 species, as well as a high level of AFP expression, indicating that the cells were still immature. To promote further hepatic differentiation or maturation, it may be effective to culture the hepatic cells in a 3D environment or on feeder cells such as cardiomyocyte- or endothelium-derived cells.[41,42] In addition, the function of our hepatic cells was still limited. Further analysis of the other functions of our hepatic cells, such as glycogen storage, uptake of indocyanine green and organic anion low-density lipoprotein, and transplantation of Ad-HEX-transduced cells into the liver of immunodeficient mice, is clearly needed for the appreciation to drug screening and therapeutic treatment modalities.

During the preparation of this article, Kubo et al. have reported that HEX could promote hepatoblast differentiation from mouse ESCs.[43] Their report is consistent with our data, suggesting that HEX plays a pivotal regulatory role in not only mouse but also human hepatic differentiation. They also showed that the overexpression of HEX at the definitive endoderm stage is critical for hepatic specification of the mouse ESCs. We also confirmed that forced expression of in the undifferentiated human ESCs and iPSCs did not elevate the expression of ALB and CK7 (Supplementary Figure S7), indicating that HEX enhances the hepatic differentiation not from the undifferentiated cells but from the definitive endoderm. However, Kubo et al. used recombinant mouse ESCs (tet-HEX ESCs), in which the tetracycline-regulated HEX expression cassette is integrated into the host cell genome to induce HEX in a stage-specific manner. Their system would not be appropriate for clinical use because the transgene is randomly integrated into the host cell genome and this leads to a risk of mutagenesis.[44] On the other hand, we generated human hepatoblasts by Ad vector-mediated transient HEX transduction, method which avoids the integration of exogenous DNA into the host chromosome.

Touboul et al. reported that human ESCs and iPSCs can differentiate into functional hepatocytes under chemically defined conditions.[34] In the present study, hepatoblasts were generated in a chemically defined serum-free medium, which minimized exposure to animal cells and proteins, and on a defined extracellular matrix, such as laminin or collagen, which do not contain undefined growth factors. To generate hepatocytes, hepatocyte culture medium, which is serum-free but not defined, was used in the stage III. When defined hESF-medium was used in the stage III, the expression levels of ALB and CYP3A4 mRNA were half the levels seen in the cells cultured with hepatocyte culture medium in the preliminary experiment (data not shown). Human ESCs and iPSCs were also grown for maintaining the undifferentiated state on a feeder layer, which contains xenoantigen such as bovine apolipoprotein B-100. Bovine apolipoprotein B-100 is known to be a dominant xenoantigen for cell-based therapies.[45] Human ESC- and iPSC-derived hepatocytes should be generated and cultured under chemically defined conditions not only to avoid potential contamination with pathogens for the safer therapeutic application, but also to obtain reproducible results using the differentiation protocols.[34,46] Development of differentiation protocols using other genes of transcription factors as well as HEX genes based on a chemically defined medium is under way. Overall, our strategy should provide a novel protocol for hepatic differentiation from human ESCs and iPSCs, which could be useful for regenerative medicine and drug screening.

Materials and Methods

. Ad vectors were constructed by an improved in vitro ligation method.[47,48] The human HEX complementary DNA derived from pDNR-LIB-HEX (Invitrogen, Carlsbad, CA) was inserted into pHMEF5,[29] which contains the human elongation factor-1α promoter, resulting in pHMEF-HEX. The pHMEF-HEX was digested with I-CeuI/PI-SceI and ligated into I-CeuI/PI-SceI-digested pAdHM41-K7,[30] resulting in pAd-HEX. Ad-HEX and Ad-LacZ, both of which contain the elongation factor-1α promoter and a stretch of lysine residues (K7) peptides in the C-terminal region of the fiber knob, were generated and purified as described previously.[26,29] The vector particle titer was determined by using a spectrophotometric method.[49]

. A human ESC line, khES1, was obtained from Kyoto University (Kyoto, Japan).[50] khES1 was used following the Guidelines for Derivation and Utilization of Human Embryonic Stem Cells of the Ministry of Education, Culture, Sports, Science and Technology of Japan after approval by the review board at Kyoto University. Human ESCs were maintained on a feeder layer of mitomycin-inactivated mouse embryonic fibroblasts (ICR; ReproCELL Incorporated, Tokyo, Japan) with Dulbecco's modified Eagle's medium/F-12 (Sigma, St Louis, MO) supplemented with 0.1 mmol/l 2-mercaptoethanol, 0.1 mmol/l nonessential amino acids, 2 mmol/l -glutamine, 20% GIBCO knockout serum replacement (Invitrogen), and 5 ng/ml bFGF (Sigma) in a humidified atmosphere of 3% CO2 and 97% air at 37 °C. Human ESCs were dissociated with 0.1 mg/ml dispase (Roche Diagnostics, Burgess Hill, UK) into small clumps, and subcultured every 5 or 6 days.

Two human iPS clones derived from the embryonic human lung fibroblast cell line MCR5 were provided from JCRB Cell Bank (Tic, JCRB Number: JCRB1331; and Dotcom, JCRB Number: JCRB1327).[3,4] In the present study, we mainly used the Tic cell line, but similar results were obtained using the Dotcom cell line, and these are shown in the supplementary figures. Human iPSCs were maintained on a feeder layer of mitomycin-inactivated mouse embryonic fibroblasts (Hygro Resistant Strain C57/BL6; Hygro, Millipore, MA) on a gelatin-coated flask in human iPS medium. Human iPS medium consists of knockout Dulbecco's modified Eagle's medium/F12 (Invitrogen), supplemented with 0.1 mmol/l 2-mercaptoethanol, 0.1 mmol/l nonessential amino acids, 2 mmol/l -glutamine, 20% knockout serum replacement, and 10 ng/ml bFGF in a humidified atmosphere of 5% CO2 and 95% air at 37 °C. Human iPSCs were dissociated with 0.1 mg/ml dispase (Roche) into small clumps and subcultured every 7 or 8 days.

In vitro . Before the initiation of cellular differentiation, the medium of human ESCs and iPSCs was exchanged for a defined serum-free medium hESF9 and cultured in a humidified atmosphere of 10% CO2 and 90% air at 37 °C.[46] hESF9 consists of hESF-GRO medium (Cell Science & Technology Institute, Sendai, Japan) supplemented with five factors (10 µg/ml human recombinant insulin, 5 µg/ml human apotransferrin, 10 µmol/l 2-mercaptoethanol, 10 µmol/l ethanolamine, 10 µmol/l sodium selenite), oleic acid conjugated with fatty acid free bovine ALB, 10 ng/ml bFGF, and 100 ng/ml heparin (all from Sigma). For induction of definitive endoderm, human ESCs and iPSCs were dissociated into single cells with Accutase (Invitrogen) and cultured for 5 days on a mouse laminin-coated tissue 12-well plate (6.0 × 104 cells/cm2) in hESF-GRO medium (Cell Science & Technology Institute) supplemented with the five factors, 0.5 mg/ml fatty acid free bovine ALB (BSA) (Sigma), 10 ng/ml bFGF, and 50 ng/ml Activin A (R&D Systems, Minneapolis, MN) in a humidified atmosphere of 10% CO2 and 90% air at 37 °C. The medium was refreshed every day.

For induction of hepatoblasts, the human ESC- and iPSC-derived definitive endoderms (day 5) were dissociated with 0.0125% trypsin- 0.01325 mmol/l EDTA, and then the trypsin was inactivated with 0.1% soybean trypsin inhibitor (Sigma). The cells were seeded at 1.2 × 105 cells/cm2 on a laminin-coated 12-well plate with hESF-DIF (Cell Science & Technology Institute) medium supplemented with the five factors, 0.5 mg/ml BSA, 10 ng/ml bFGF, and 50 ng/ml Activin A in a humidified atmosphere of 10% CO2 and 90% air at 37 °C. The next day, the cells were transduced with 3,000 vector particle/cell of Ad vectors (Ad-HEX and Ad-LacZ) for 1.5 hours in hESF-DIF medium supplemented with the five factors, BSA, 10 ng/ml FGF4 (R&D Systems) and 10 ng/ml BMP4 (R&D Systems).[10] The medium was refreshed every day.

For induction of hepatocytes, human iPSC-derived hepatoblasts in one well (day 9) were passaged onto two wells with 0.0125% trypsin-0.01325mmol/l EDTA and 0.1% trypsin inhibitor, on type I collagen-coated tissue 12-well plate (15 µg/cm2) (Nitta Gelatin, Osaka, Japan). The cells were cultured in hepatocyte culture medium supplemented with SingleQuots (Lonza, Walkersville, MD), 10 ng/ml FGF4, 10 ng/ml HGF (R&D Systems), 10 ng/ml Oncostatin M (R&D Systems), and 0.392 ng/ml dexamethasone (Sigma).[11] The medium was refreshed every 2 days.

. For details of these procedures, See Supplementary Materials and Methods, Supplementary Tables S1 and S2.

SUPPLEMENTARY MATERIAL Figure S1. Characterization of the human ESC (khES1)- and iPSC (Tic)-derived definitive endoderms. (a) Quantitative depiction of flow cytometric analysis for CXCR4 expression on the iPSC (Tic)-derived definitive endoderm (day 5). (b-e) Real-time RT-PCR analysis of the levels of the mesoderm marker FLK1 mRNA and ectoderm marker PAX6 mRNA expression at days 0, 5 and 6. Data are presented as the mean ± SD from triplicate experiments. The graphs represent the relative gene expression level when the level in undifferentiated cells (day 0) was taken as 1. The scale bar represents 50 μm. Figure S2. Efficient differentiation of another human iPSC line (Dotcom) into hepatoblasts by overexpression of the HEX gene. (a, b) The immunofluorescent staining of the human iPSC (Dotcom)-derived definitive endoderm before (a; day 5) and after passaging (b; day 6). The cells were immunostained with antibodies against FOXA2 and NANOG. Nuclei were stained with DAPI. (c) The quantitative analysis of the immunofluorescent staining in (a and b). Data are presented as the mean of immunopositive cells counted in 8 independent fields. (d) Real-time RT-PCR analysis of the level of definitive endoderm (FOXA2 and SOX17) at day 5 and 6. Data are presented as the mean ± SD from triplicate experiments. The graphs represent the relative gene expression level when the level in undifferentiated cells (day 0) was taken as 1. (e, f) Immunocytochemistry against the expression of AFP and CK7 in non-transduced (e) and Ad-HEX-transduced (f) cells (day 12), both of which were induced from the human iPSC (Dotcom)-derived definitive endoderm. Nuclei were stained with DAPI. (g) Real-time RT-PCR analysis of the levels of AFP and ALB gene expression in non-transduced cells and Ad-HEX-transduced cells, both of which were induced from the iPSC (Dotcom)-derived definitive endoderm (day 0, 5, 6, and 12). The cells were transduced with Ad-HEX as described in Figure 1a. The data at day 6 was obtained before the transduction with Ad-HEX. The graphs represent the relative gene expression level when the level in the fetal liver was taken as 100. The scale bar represents 50 μm. Abbreviations: AFP, α-fetoprotein; ALB, albumin, CK7, cytokeratin 7; NONE, non-transduced cells; LacZ, Ad-LacZ-transduced cells; and HEX, Ad-HEX-transduced cells. Figure S3. Overexpression of HEX in the human ESC (khES1)- and iPSC (Tic)-derived definitive endoderms. (a-b) Real-time RT-PCR analysis of the HEX expression in non-transduced cells, Ad-LacZ-transduced cells, and Ad-HEX-transduced cells, all of which were induced from the human ESC (khES1)-derived (a) and iPSC (Tic)-derived (b) definitive endoderms. (c-f) The expression of HEX protein in Ad-HEX-transduced cells at day 9 (c) and at day 18 (d), and non-transduced cells at day 9 (e) and at day 18 (f), all of which were induced from the human iPSC (Tic)-derived definitive endoderm. The cells were transduced with Ad-HEX as described in Figure 1a. The data at day 6 was obtained before the transduction with Ad-HEX. The graphs represent the relative gene expression level when the level in the fetal liver was taken as 100. The scale bar represents 50 μm. Abbreviations: NONE, non-transduced cells; LacZ, Ad-LacZ-transduced cells; and HEX, Ad-HEX-transduced cells. Figure S4. Characterization of Ad-HEX-transduced hepatoblasts. Semi-quantitative RT-PCR analysis of the mRNAs of hepatic markers (HNF1A, HNF1B, HNF4A, HNF6, CEBPA, PROX1 and CK7) mRNA. Abbreviations: (Lanes 1-3) Human undifferentiated khES1 (Lane 1), human ESC (khES1)-derived definitive endoderm at day 5 (Lane 2) and at day 6 (Lane 3). (Lanes 4-6) Non-transduced cells (Lane 4), Ad-LacZ-transduced cells (Lane 5) and Ad-HEX-transduced cells (Lane 6), all of which were induced from the human ESC (khES1)-derived definitive endoderm (day 12). Lane 7, HepG2 cells. Lane 8, Primary hepatocytes. Lane 9, Fetal liver. Lane 10, Adult liver. (Lanes 11-14) Human undifferentiated iPSCs (Lane 11), human iPSC (Tic)-derived definitive endoderm at day 5 (Lane 12) and at day 6 (Lane 13). (Lanes 14-19), Non-transduced cells (Lane 14, day9; Lane 17, day 12), Ad-LacZ-transduced cells (Lane 15, day 9; Lane 18, day 12) and Ad-HEX-transduced cells (Lane 16, day 9; Lane 19, day12), all of which were induced from the human iPSC (Tic)-derived definitive endoderm. Figure S5. Progression of differentiation of the definitive endoderm to hepatoblasts. (a-f) Immunocytochemistry of AFP and CK7 expression from day 7 to day 10 in non-transduced cells (a-c) and Ad-HEX-transduced cells (d-f), all of which were induced from the human iPSC (Tic)-derived definitive endoderm. Nuclei were stained with DAPI. (g-n) Sequential morphological changes from day 8 to day 21 in non-transduced cells (g-j) and Ad-HEX-transduced cells (k-n), all of which were induced from the human iPSC (Tic)-derived definitive endoderm. The Scale bar represents 50 μm. Abbreviations: AFP, α-fetoprotein; CK7, cytokeratin 7; NONE, non-transduced cells; LacZ, Ad-LacZ-transduced cells; and HEX, Ad-HEX-transduced cells. Figure S6. X-gal staining of human iPSC (Tic)-derived definitive endoderms transduced with a conventional or a fiber-modified Ad vector containing the EF-1α promoter. Human iPSC (Tic)-derived definitive endoderms were transduced with a LacZ-expressing conventional or fiber-modified Ad vector containing the EF-1α promoter for 1.5 h at 3,000 VP/cell. The next day after transduction, X-gal staining was performed as described in the Materials and Methods. The scale bar represents 50 μm. Abbreviations: NONE, non-transduced cells; LacZ, Ad-LacZ-transduced cells. Figure S7. HEX promotes the differentiation into the hepatic lineage, not from undifferentiated iPSCs (Tic), but from iPSC (Tic)-derived definitive endoderm. Semi-quantitative RT-PCR analysis of ALB and CK7 expression levels. (lane 1-3) Human undifferentiated iPSCs (Tic) were transduced with mock (lane 1), Ad-LacZ (lane 2) or Ad-HEX (lane 3), and cultured for 6 days on a mouse laminin-coated tissue plate in hESF-DIF medium supplemented with the 5 factors (10 μg/ml human recombinant insulin, 5 μg/ml human apotransferrin, 10 μM 2-mercaptoethanol, 10 μM ethanolamine, 10 μM sodium selenite), 0.5 mg/ml fatty acid free bovine albumin (BSA), 10 ng/ml FGF4 and 10 ng/ml BMP4. At day 6, semi-quantitative RT-PCR analysis of the ALB and CK7 expression levels was performed. (lane 4) human iPSC (Tic)-derived definitive endoderms, which were induced for 6 days as described in materials and methods section, were transduced with Ad-HEX, and cultured on a mouse laminin-coated tissue plate with in hESF-DIF medium supplemented with the 5 factors, BSA, 10 ng/ml FGF4 and 10 ng/ml BMP4. At day 12, semi-quantitative RT-PCR analysis of the ALB and CK7 expression levels was performed. (lane 5) HepG2. (lane 6) Fetal liver. Table S1. List of Taqman gene expression assays and primers. Table S2. List of antibodies used. Materials and Methods.

Figure 1

A strategy of differentiation of human embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) to hepatoblasts and hepatocytes. (a) Schematic representation illustrating the procedure for differentiation of human ESCs (khES1) and iPSCs (Tic) to hepatocytes. (b–i) Phase contrast microscopy showing sequential morphological changes (day 0–12) from (b–e) human ESCs (khES1) and (f–i) iPSCs (Tic) to hepatoblasts via the definitive endoderm. Bar = 50 µm. bFGF, basic fibroblast growth factor; BMP4, bone morphogenetic protein 4; DEX, dexamethasone; FGF4, fibroblast growth factor 4; HGF, hepatocyte growth factor; OSM, Oncostatin M; HCM, hepatocytes culture medium; *, hESF-GRO medium that was supplemented with 10 µg/ml human recombinant insulin, 5 µg/ml human apotransferrin, 10 µmol/l 2-mercaptoethanol, 10 µmol/l ethanolamine, 10 µmol/l sodium selenite, 0.5 mg/ml fatty acid free BSA; **, hESF-DIF medium that was supplemented with 10 µg/ml insulin, 5 µg/ml apotransferrin, 10 µmol/l 2-mercaptoethanol, 10 µmol/l ethanolamine, 10 µmol/l sodium selenite, 0.5 mg/ml BSA.