Potential conflict of interest: Nothing to report.
The differentiation capacity of human embryonic stem cells (hESCs) holds great promise for therapeutic applications. We report a novel three-stage method to efficiently direct the differentiation of human embryonic stem cells into hepatic cells in serum-free medium. Human ESCs were first differentiated into definitive endoderm cells by 3 days of Activin A treatment. Next, the presence of fibroblast growth factor-4 and bone morphogenetic protein-2 in the culture medium for 5 days induced efficient hepatic differentiation from definitive endoderm cells. Approximately 70% of the cells expressed the hepatic marker albumin. After 10 days of further in vitro maturation, these cells expressed the adult liver cell markers tyrosine aminotransferase, tryptophan oxygenase 2, phosphoenolpyruvate carboxykinase (PEPCK), Cyp7A1, Cyp3A4 and Cyp2B6. Furthermore, these cells exhibited functions associated with mature hepatocytes including albumin secretion, glycogen storage, indocyanine green, and low-density lipoprotein uptake, and inducible cytochrome P450 activity. When transplanted into CCl4 injured severe combined immunodeficiency mice, these cells integrated into the mouse liver and expressed human alpha-1 antitrypsin for at least 2 months. In addition, we found that the hESC-derived hepatic cells were readily infected by human immunodeficiency virus-hepatitis C virus pseudotype viruses. Conclusion: We have developed an efficient way to direct the differentiation of human embryonic stem cells into cells that exhibit characteristics of mature hepatocytes. Our studies should facilitate searching the molecular mechanisms underlying human liver development, and form the basis for hepatocyte transplantation and drug tests. (HEPATOLOGY 2007;45:1229–1239.)
Orthotopic liver transplantation (OLT) has been successfully used to treat a variety of end-stage liver diseases.1 Hepatocyte transplantation is a potential way to replace orthotopic liver transplantation and is used to transfer patients from whole-organ transplantation, to decrease mortality in acute liver failure, and for treatment of metabolic liver disease.2 However, the major limitation of cell-based therapies for liver disease is the availability of human hepatocytes.2 Human embryonic stem cells (hESCs) proliferate infinitely in vitro while maintaining their potential to differentiate into almost all cell types,3 and thus provide a potential source for obtaining hepatocytes. Finally, because the liver is the main detoxification organ in the body, the hESC derived hepatocytes might also be useful for in vitro drug testing.4
Several studies have demonstrated the capacity of hESCs to differentiate into hepatocyte or hepatocyte-like cells.5–9 However, their differentiation efficiency is low, and most reports performed only limited phenotypic and functional tests on the differentiated cells. In particular, the differentiation strategies previously employed do not exclude the extraembryonic endoderm differentiation of embryonic stem cells (ESCs), which makes the hepatic identities of the differentiated cells controversial. This is especially important for hESC differentiation, as hESCs tend to differentiate toward extraembryonic endoderm cells,10 which also express most of the hepatocyte markers.11
Here we report the development of a novel three-stage method for inducing the differentiation of hESCs into hepatic cells using serum-free medium. The sequential addition of activin A and fibroblast growth factor-4 (FGF4) plus bone morphogenetic protein-2 (BMP2) produced cultures consisting of 70% albumin-positive cells (early hepatic cells) from definitive endoderm cells. After being further maturated in vitro, these cells (maturated hepatic cells) expressed several adult liver cell markers and possessed liver cell functions, such as albumin (Alb) secretion, glycogen storage, indocyanine green (ICG)- and low density lipoprotein (LDL)-uptake, and inducible cytochrome P450 activity. When transplanted into CCl4 injured severe combined immunodeficiency (SCID) mice, these cells were able to integrate into the mouse liver and express human alpha-1 antitrypsin (AAT). In addition, we found that both early hepatic cells and maturated hepatic cells could be infected by HIV-HCV pseudotype viruses.
Activin A, acidic FGF (aFGF), basic FGF (bFGF), FGF4, hepatocyte growth factor (HGF), and bone morphogenetic protein-2 (BMP2) were purchased from Peprotech (Rocky Hill, NJ), Oncostatin M (OSM), human BMP4 and recombinant mouse Noggin were purchased from R&D System (Minneapolis, MN). Dexamethasone (Dex) and 7-pentoxyresorufin were obtained from Sigma-Aldrich (St Louis, MO). Su5402 was obtained from Chemicon (Pittsburgh, PA).
An antibody that recognizes human Sox17 was purchased from R&D system. An antibody against human FoxA2 (HNF3β) was purchased from Upstate Biotechnology (Lake Placid, NY). Antibodies against human cytokeratin-7 (CK-7), CK-18, CK-19, and AAT were purchased from Invitrogen (Grand Island, NY). Antibody against human α-fetoprotein (AFP) and Alb were purchased from DAKO (Glostrup, Denmark). Antibody against human CK8 was obtained from Progen Biotechnik GmbH (Heidelberg, Germany). Antibodies against human nuclei was obtained from Chemicon. FITC- and TRITC-conjugated secondary antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). The mouse IgG1/IgG2a isotype control, antibodies against human Ki67 and CD81 were obtained from BD Biosciences (Flanders, NJ).
Human ESC lines H1 and H9 were obtained from WiCell research institute (Madison, WI). The passage number of the H1 cells and H9 cells used here ranged from 53 to 80 and from 65 to 85, respectively. Human ESCs were maintained on irradiated mouse embryonic fibroblasts, in hESC medium: DMEM/F12 medium supplemented with 20% knockout serum replacement, 1 mM L-glutamine, 1% nonessential amino acids, 0.1 mM β-mercaptoethanol (all from Invitrogen/Gibco), and 4 ng/ml bFGF, under a standard gas atmosphere of humidified air/5% CO2. Karyotype analysis indicated that they had a normal karyotype.
For hepatic differentiation, hESCs were cultured in 1640 medium (Hyclone, Logan, UT) supplemented with 0.5 mg/ml albumin fraction V (Sigma-Aldrich) and 100 ng/ml Activin A for 1 day. On the following 2 days, 0.1% and 1% insulin-transferrin-selenium (ITS) (Sigma-Aldrich) was added to this medium. After 3 days of Activin A treatment, the differentiated cells were cultured in hepatocyte culture medium (HCM) (Cambrex, Baltimore, MD) containing 30 ng/ml FGF4 and 20 ng/ml BMP2 for 5 days. Then the differentiated cells were further maturated in HCM containing 20 ng/ml HGF for 5 days, and 10 ng/ml OSM plus 0.1 μM Dex from then on. The medium was changed evμery day. To test the effects of FGF4 and BMP2 on hepatic differentiation from definitive endoderm cells, the concentrations of the added factors were: FGF4, 30 ng/ml; BMP2, 20 ng/ml; Su5402, 10 μM; Noggin, 800 ng/ml. The concentrations of aFGF, bFGF, and BMP4 used for hepatic differentiation from definitive endoderm cells were 80, 10, and 20 ng/ml, respectively. Epidermal growth factor was omitted from HCM to exclude its possible influence. The same medium without growth factors was used as the negative control during the differentiation.
Cells or tissue sections were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) at room temperature for 20 minutes and blocked with PBS containing 0.1% Triton X-100, 10% normal horse serum, and 1% bovine serum albumin at room temperature for 1 hour. Cells were incubated with primary antibodies at 4°C overnight. Isotype mouse IgG or normal rabbit serum was used as negative controls, and no fluorescence was observed in the negative controls. The primary antibodies against human SOX17, CK7, CK8, CK18, CK19 and Ki67 were diluted at 1:200, antibodies against human nuclei and AAT were diluted at 1:150, antibody against human Alb was diluted at 1:500 and antibody against human AFP was diluted at 1:250. After 5 washes with PBS, FITC-conjugated or TRITC-conjugated secondary antibody diluted at 1:150 was added and incubated at 37°C for 1 hour. Then 1 μg/ml DAPI (Roche, Germany) was used to stain the cell nucleus. Between each step, cells and sections were washed with PBS containing 0.1% bovine serum albumin.
In order to evaluate the proportion of Alb-positive cells in day-8 cultures, we randomly picked 30 pictures from 3 independent cultures. Then, the Alb-positive cells and whole cells counterstained by DAPI were counted by Image-Pro Plus software (Media Cybernetics).
Reverse Transcription PCR (RT-PCR) and Real-Time RT-PCR.
Total RNA was isolated from cells using TRIzol Reagent (Invitrogen) and reverse-transcribed using a reverse transcription system (Promega, Madison, WI) according to the manufacturer's protocol. PCR amplification of different genes was performed using EXTaq polymerase (Takara, Japan), with a program of 94°C for 5 minutes, 35 cycles of 94°C for 30 seconds, 50°C–57°C for 30 seconds, 72°C for 30 seconds, and extension at 72°C for 10 minutes. The primers used are shown in Table 1.
Table 1. Primers and Conditions Used for RT-PCR
Product length (bp)
Annealing temperature (°C)
Real-time RT-PCR analysis was performed on an ABI Prism 7300 Sequence Detection System using the SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). The PCR reaction consisted of 12.5 μl of SYBR Green PCR Master Mix, 0.8 μl 10 mM of forward and reverse primers, 10.4 μl water, and 0.5 μl template cDNA in a total volume of 25 μl. Cycling was performed using the default conditions of the ABI 7300 SDS Software 1.3.1: 2 minutes at 50°C, 10 minutes at 95°C, followed by 40 rounds of 15 seconds at 95°C and 1 minutes at 60°C. The relative expression of each gene was normalized against β-actin. The primers used for the quantitative RT-PCR were: albumin, GCACAGAATCCTTGGTGAACAG and ATGGAAGGTGAATGTTTCAGCA; β-actin, CTGGAACGGTGAAGGTGACA and AAGGGACTTCCTGTAACAATGCA.
To determine the total cell protein content, the differentiated cells were lysed with buffer containing 50 mM Tris-Cl (pH 7.5), 150 mM NaCl, 2 mM EDTA and 1% SDS. The total cell protein content was determined by Micro BCA Protein Assay Kit (Pierce, Rockford, IL). The human albumin content in the supernatant was determined by Human Albumin ELISA Quantitation kit (Bethyl Laboratory, Montgomery, TX) under the manufacturer's instructions. The albumin secretion was normalized to the total cell protein.
Uptake of LDL.
The Dil-Ac-LDL staining kit was purchased from Biomedical Technologies (Stoughton, MA) and the assay was performed according to the manufacturer's instructions.
For the co-localization of LDL uptake cells with hepatic marker-expressing cells, the cells in culture after LDL uptake test were fixed in 4% paraformaldehyde and then were further assayed by immunofluorescence as described above.
Periodic Acid-Schiff (PAS) Stain for Glycogen.
The PAS staining system was purchased from Sigma-Aldrich. Culture dishes containing cells were fixed in 4% paraformaldehyde. The further assay was under the manufacturer's instruction.
The ICG uptake test of the differentiated cells was performed as described.12 In the ICG uptake cells, ICG was eliminated from the cells 6 hours later.
Pentoxyresorufin-O-dealkylase (PROD) was measured in the presence or absence of 1 mM phenobarbital sodium as previously described.13
Transplantation of Maturated Cells.
Female SCID Beige mice (10 weeks old) were obtained from Vital River Inc. (Beijing, China). The Institutional Animal Care and Use Committees of Peking University approved all animal procedures. One day before transplantation, each mouse was administered 10 μl of CCl4 (Sigma-Aldrich) diluted at 1:10 in sterilized mineral oil (Sigma-Aldrich). The day-18 induced cells were trypsinized at 37°C for 8 minutes with 0.25% trypsin and then resuspended in DMEM without fetal bovine serum (FBS). Approximately 1 × 106 cells in 0.1 ml of suspension was transplanted into the spleen of female SCID mice (n = 3). Three mice injected with 0.1 ml DMEM were used as negative controls. Eight weeks after cell transplantation, the livers were embedded in OCT compound (Tissue-TEK, Naperville, IL), and then frozen in liquid nitrogen. Sections were cut into 7-μm-thick slices. The hESC-derived cells were detected by anti-human nuclei monoclonal antibody (1:100) and anti-human AAT polyclonal antibody (1:200).
Preparation of Pseudotyped Viruses and Infection Assays.
For pseudotyped virus preparation, 293T cells were transfected with 10 μg plasmids expressing the viral glycoproteins or pcDNA3.1 (+) vector, and 10 μg envelope-defective pNL4.3.Luc.R−E−pro14 viral genome using calcium phosphate transfection method.15 The medium was refreshed 12 hours after transfection. Supernatants were harvested 36 hours later and centrifuged at 3000g for 10 minutes to remove cell debris, and were then used in the infection assays. HIV p24 antigen content was assessed by a commercially available EIA (Beckman Coulter, Miami, FL).
Target cells in 24-well plates were incubated with supernatant containing 2 ng pseudotyped virus (p24) and 4 μg/ml polybrene per well for 4 hours. The supernatants were removed and the cells were incubated in fresh medium at 37°C for another 72 hours, washed in PBS once, and lysed with 100 μl lysis buffer (Promega). Twenty microliters of lysate was tested for luciferase activity by the addition of 50 μl of luciferase substrate, and was measured for 10 seconds in a Wallac Multilabel 1450 Counter (PerkinElmer, Wellesley, MA). For neutralization assays, mAbs for CD81 at concentration of 5 μg/ml were incubated with cells for 30 minutes at 4°C, and the mixtures were tested for infectivity as described above.
Differentiation of hESCs into Definitive Endoderm by Activin A in Serum-Free Medium.
Our 3-stage differentiation strategy is illustrated in Fig. 1A. In the first stage, hESCs were induced to definitive endoderm cells by 3 days of Activin A treatment. In the second stage, the definitive endoderm cells were differentiated into early hepatic cells by the combination of FGF4 and BMP2 treatment for 5 days. In the third stage, the early hepatic cells were further matured by treatment of the cells with HGF for 5 days, followed by OSM plus Dex treatment for another 5 days. The main experiments were carried out in the H1 cell line, and similar results were also obtained by the differentiation of the H9 cell line.
It has been reported that Activin A can efficiently induce definitive endoderm differentiation of hESCs.16 In this study, we used Activin A to induce definitive endoderm differentiation in the first stage of the differentiation process. Previous investigations used culture medium containing FBS.16 FBS, which contains unknown factors, may hamper therapeutic use. Therefore, we used ITS to replace FBS in the differentiation medium for endoderm differentiation. After treatment with 100 ng/ml Activin A in the serum-free medium for 3 days, more than 80% of the differentiated cells expressed definitive endoderm marker genes such as SOX17 and FoxA2 (HNF3β) (Fig. 1B,C). The Activin A–induced cells did not express AFP (Fig. 1D), indicating that these cells were not visceral endoderm cells. These results indicated that Activin A induced efficient definitive endoderm differentiation in the serum-free medium.
Initiation of Hepatic Differentiation from Definitive Endoderm by FGF4 and BMP2.
In the second stage of our differentiation method, albumin gene expression determined by real-time RT-PCR was used to evaluate the differentiation efficiency. It has been reported that during mouse liver development, FGFs and BMPs are essential for hepatic specification from foregut endoderm,17, 18 suggesting that these 2 signals may be important for the hepatic differentiation of endoderm cells in vitro. Therefore, we tested the effects of FGF4 and BMP2 on the induction of hepatic differentiation from definitive endoderm cells. The addition of FGF4 or BMP2 alone had little effect on albumin gene expression (Fig. 2A). However, the combination of these 2 factors led to a dramatic increase of albumin gene expression, indicating that both FGF and BMP are needed for effective hepatic differentiation. To further confirm the synergistic effects of these 2 factors, we added the FGFR1 inhibitor Su5402 or the BMP inhibitor Noggin along with FGF4 and BMP2. The addition of Su5402 or Noggin abolished the cooperation effects of FGF4 and BMP2 (Fig. 2A). Interestingly, in the absence of Activin A pretreatment, the addition of FGF4 and BMP2 yielded a low level of albumin gene expression (Fig. 2A), indicating that the effects of FGF4 and BMP2 on hepatic differentiation are specific to the definitive endoderm cells. These results indicated that the sequential addition of Activin A and the combination of FGF4 and BMP2 were able to efficiently induce hepatic differentiation from hESCs. Besides FGF4 and BMP2, we also tested the combination of aFGF and BMP4, bFGF and BMP4, or FGF4 and BMP4. These combinations exhibited similar effects (data not shown).
After 5 days of FGF4 and BMP2 treatment, the morphology of the definitive endoderm cells resembled the cuboidal shapes typical of hepatocytes (Fig. 2B). The spontaneously differentiated hESCs, in contrast, exhibited a mixed morphology consisting mostly of squamous shapes (Fig. 2B). About 30% of the induced differentiated cells expressed Ki67 (Fig. 2C), suggesting that these induced cells are proliferative. In the induced cultures, about 70% of the differentiated cells expressed early fetal liver cell markers AFP, Alb, CK8 and CK18 (Fig. 2D), as detected by immunofluorescence staining. In the control group, the spontaneous differentiation of hESCs produced only less than 10% Alb+ cells (data not shown). The expression of these hepatic markers was further confirmed by RT-PCR. The induced cells expressed AFP, Alb, CK8, CK18, and G6P (Fig. 2E), and the expression of AFP and Alb in the induced cells was higher than that of the spontaneously differentiated hESCs, whereas G6P was not detected in the spontaneously differentiated hESCs. At this stage, among the induced cells, only a few expressed AAT as detected by immunocytochemistry (data not shown), and they did not secrete albumin (Fig. 4A), indicating that most of these cells are early hepatic cells.
During liver development, the hepatocyte and cholangiocyte have been suggested to be derived from a common progenitor, that is, the hepatoblast.19, 20 We then tested the expression of cholangiocyte marker CK7 expression in the induced culture. At this stage, CK7-positive cells were observed among the differentiated cells and some CK7-positive cells formed a bile duct-like structure in the induced culture (Fig. 2F, lower panel). All the CK7+ cells were found near the AFP+ or Alb+ cells (Fig. 2F). Besides, CK19 and Alb double-positive cells were also generated among the induced cells (Fig. 2G), These results suggest that hepatoblasts had been generated during the differentiation process.
In the third stage, we used HGF, OSM and Dex to promote the maturation of early hepatic cells. HGF, OSM, and Dex were reported to be involved in the differentiation of fetal liver cells into mature cells.21–23 After 5 days of HGF treatment followed by a further 5 days of OSM/Dex treatment, the cells became more flattened (Fig. 3A), and about 10% of these cells expressed Ki67 (data not shown). We next checked the expression of hepatic markers in the induced cultures by immunofluorescence staining. A majority of the cells expressed hepatic markers that have previously been shown to be expressed at the end of the second stage of differentiation, including AFP, Alb, CK8 and CK18 (data not shown). Furthermore, most of the maturated hepatic cells showed positive AAT staining (Fig. 3B). We also checked the expression of adult liver cell markers in these maturated cells by RT-PCR. These maturated cells expressed adult liver markers phosphoenolpyruvate carboxykinase (PEPCK), tyrosine aminotransferase (TAT), tryptophan 2,3-dioxygenase (TDO2), hepatocyte nuclear factor 4α (HNF4α), the cytochrome enzymes Cyp2B6 and Cyp3A4, as well as Cyp7A1 (Fig. 3C). Importantly, Cyp7A1 is not expressed in extraembryonic cells.24 In contrast, the expression levels of these genes in spontaneous differentiated hESCs were significantly lower or undetectable (Fig. 3C).
To test whether the induced cells possess liver cell functions, we measured the albumin secretion of the differentiated cells. The induced cells secreted albumin at a level of 2.07 ng/day per μg of total cell protein (Fig. 4A), which was a similar level with that of Huh-7 hepatoma cell line (2.04 ng/day per μg total cell protein, data not shown). We then assayed for glycogen storage of the differentiated cells by PAS staining. Many induced cells could be stained by PAS (Fig. 4B). However, PAS stained cells were rarely observed in the spontaneously differentiated hESCs. Subsequently, we examined ICG take-up, which is one of the liver-specific functions used for the identification of differentiated hepatocytes in vitro.12 The induced cells could uptake ICG from medium and exclude the absorbed ICG six hours later (Fig. 4C). Among spontaneously differentiated hESCs, ICG-positive cells were rarely observed.
To further prove the maturated cells were functional in vitro, we tested the ability of differentiated cells to uptake LDL. More than 80% of the induced could uptake LDL and most of the LDL-taken cells also expressed the liver cell markers AFP, Alb, AAT and CK18 (Fig. 4D). In the spontaneously differentiated hESCs, most cells were unable to uptake LDL and only a small portion of the LDL-taken cells expressed the liver cell marker AAT (data not shown). We used a PROD assay to evaluate the P450 activity, an indicator of a liver cell's detoxify ability, in the differentiated cells. For the induced cells cultured in the absence of phenobarbital sodium induction, a few cells exhibited weak PROD activity (Fig. 4E). The presence of phenobarbital sodium induction increased the PROD activity, indicating that the induced cells possessed inducible P450 activity. In the spontaneously differentiated hESCs, few cells had PROD activity even in the presence of phenobarbital sodium induction (Fig. 4E).
To investigate whether the differentiated cells were engraftable, we transplanted 1 × 106 induced cells into the spleen of CCl4-treated liver-injured SCID mice. Two months later, in mice transplanted with induced differentiated cells, we could detect approximately 500 human nuclei in 150 different areas (Fig. 4F, upper panel). A similar frequency of AAT-positive cells could also be detected (Fig. 4F, lower panel). However, in the sham-transplanted mice, we could not detect any human nuclei or AAT-positive cells in a total of 300 different areas (data not shown). No teratoma formation was observed in the livers of recipient mice. These results suggest that the hESC-derived hepatic cells could engraft into the mouse liver, and retain the ability to produce liver-specific proteins.
Entry of HIV-HCV Pseudotype Viruses into hESC-Derived Hepatic Cells.
To test whether the differentiated cells were susceptible to HCV, we incubated the early and maturated hepatic cells with supernatant containing HIV-HCV pseudotype viruses. These pseudoparticles were generated by incorporating HCV glycoproteins into HIV-derived core particles, and the infection was then determined by examining the luciferase expression of the infected cells. The hESC-derived early hepatic cells (day 8 of differentiation) and maturated hepatic cells (day 18 of differentiation) had a similar susceptibility to HIV-HCV pseudotype viruses compared with the positive control Huh-7 cells (Fig. 5). Hela cells were used as negative control cells,25, 26 and could not be infected by the HIV-HCV pseudotype viruses but could be infected by HIV-vesicular stomatitis virus protein G (VSV G) pseudotyped viruses (Fig. 5). To ascertain the specificity of HIV-HCV pseudotype virus infection, we tested whether the susceptibility of the induced cells could be neutralized by an antibody specific for CD81, the receptor of HCV.27–29 We found that these infections were effectively neutralized by anti-CD81 monoclonal antibody at 5 μg/ml, but not by the isotype control mouse IgG (Fig. 5). Altogether, these results demonstrated that the hESC derived early and late hepatic cells were susceptible to infection by HIV-HCV pseudotype viruses.
In this study, we demonstrated that hESCs can be induced with high efficiency to differentiate into hepatic cells in serum-free mediums. This differentiation process resembles natural liver development. The matured hepatic cells expressed various adult liver markers including TAT, TDO2, PEPCK, Cyp7A1, Cyp3A4, and Cyp2B6. Furthermore, these cells possessed liver-specific functions, such as albumin secretion, ICG and LDL uptake, glycogen storage and inducible cytochrome P450 activity. The differentiated cells integrated into mouse livers when transplanted into CCl4-injured mice.
We showed that sequential treatment of hESCs with Activin A and FGF4/BMP2 was crucial for highly efficient hepatic differentiation. Activin A is reported to be able to induce definitive endoderm differentiation of embryonic stem cells16, 30, 31 and is nonpermissive for visceral endoderm.31 Previously, we showed that a combination of Activin A and retinoic acid efficiently induced pancreatic cells from mouse ES cells.32 Previous studies on the hepatic differentiation of human5–9 and mouse ESCs33–36 did not exclude the extraembryonic differentiation of ESCs, which could also lead to the expression of many liver cell markers.11 To overcome this problem, we combined Activin A treatment with hepatic linage differentiation conditions, and induced hESCs into definitive endoderm cells by 3 days of Activin A treatment (Fig. 1B-D). After Activin A treatment, we found that the combination of FGF4 and BMP2 efficiently initiated hepatic differentiation from the definitive endoderm. It was reported that FGFs could replace the cardiac mesoderm for liver initiation from foregut endoderm.17 BMPs, working in parallel to FGF, were also found to be critical for the morphogenetic growth of the hepatic endoderm.18 A similar mechanism also exists in avian liver development, in which BMPs together with a cardiogenic generated factor are necessary for hepatic commitment, and the cardiogenic generated factor can be replaced by FGFs.37 In this study, we also found that the combination of FGF4 and BMP2 could efficiently initiate hepatic differentiation from definitive endoderm cells (Fig. 2A). Altogether, these results suggested these 2 signals are conserved during the process of liver formation among different species.
Interestingly, in the absence of Activin A pretreatment, the day-3 differentiated cells expressed FGF and BMP receptors (data not shown), but their response to FGF4 and BMP2 was not hepatogenic (Fig. 2A). This result is consistent with previous observations of mouse embryonic liver development, in which the hepatogenic effect of FGFs was found to be specific for the endoderm.17
It was reported that during liver development in vivo, cholangiocytes and hepatocytes are derived from the common progenitors, called hepatoblasts.19, 20 In our study, after cells were treated with FGF4 and BMP2, we also found a few cholangiocytes in close proximity to early hepatic cells (Fig. 2F). We also found that CK19/Alb double positive cells were generated (Fig. 2G), indicating that the differentiation process might induce the appearance of hepatoblasts. In summary, sequential treatment of hESCs with Activin A and FGF4/BMP2 caused the efficient differentiation of these cells into hepatic cells. Importantly, this process generally mimicked natural embryonic liver development observed in vivo.
In this study, multiple standards were employed to identify the hepatic identities of differentiated cells. First, most of the differentiated cells expressed liver cell marker genes such as Alb, AFP and AAT (Figs. 2D, 3C). Although extraembryonic endoderm cells also express such genes, in our method, Activin A treatment caused the exclusion of the visceral endoderm differentiation of hESCs. Therefore, such markers could represent hepatic cells in our differentiation process. Second, due to the fact that a single functional test may not be uniquely specific to liver cells (for example, LDL could be taken up by liver, adrenal, and spleen cells; ICG could be taken up by liver cells and trophoblast cells; glycogen could be detected in liver and muscle), we combined these complementary functional tests to characterize the differentiated cells. Finally, we combined LDL uptake with the hepatic marker expression. LDL was co-localized with hepatic markers AFP, Alb, AAT, and CK18 in the maturated hepatic cells (Fig. 4D). At the single cell level, the hESC-derived hepatic cells possessed hepatic marker expression pattern as well as liver cell function. In conclusion, with these multiple standards, the induced differentiated cells were similar to the adult hepatocytes in certain extent.
Hengstler et al. have critically reviewed the criteria required to evaluate the similarity between stem cell derived hepatic cells and genuine hepatocytes.38 As they pointed out, certain markers such as albumin, are relatively easy to induce. In contrast, the induction of other factors such as Cyp3A is more difficult. They also emphasize the importance of detecting the expression of Cyp3A4.38 Another marker, CYP7A1 has been suggested to be specifically expressed in the liver, but not in yolk sac cells.24 In our studies, the induced differentiated cells could express Cyp7A1, Cyp3A4, and Cyp2B6 (Fig. 3B). Moreover, the activity of CYP2B6 could be up-regulated by pentobarbital (Fig. 4E). These results suggest that the differentiated cells generated in our culture exhibit some characteristics of hepatocytes. To further define the similarity between hESC-derived hepatic cells with adult hepatocytes, it is necessary to quantitatively evaluate the function of differentiated cells in comparison with primary hepatocytes.38 In this study, we quantitatively examined the albumin secretion of the differentiated cells, which was similar to a hepatoma cell line Huh7 (Fig. 4A). However, the albumin secretion by Huh7 cell line is much lower than primary hepatocytes, indicating our differentiated cells were still not functionally matured. Although we performed multiple functional assays to demonstrate the hepatic characteristics of our differentiated cells, we only obtained a few quantitative data and were unable to directly compare our hESC derived hapatocytes to primary hepatocytes. Therefore, future work will focus on more detailed comparisons between the hESC-derived hepatic cells and primary hepatocytes.
In this study, we demonstrated that hESC-derived hepatic cells could be infected by pseudotyped HIV-HCV, and these cells could provide a potential source of hepatocytes for antiviral drug screening in vitro. Previously, it has been reported that primary hepatocytes and several hepatoma cell lines could be infected by pseudotyped HIV-HCV in vitro.39 However, the usage of these cells for antiviral drug development was limited because of the variation in transduction efficiencies observed in primary hepatocytes derived from individual biopsies.39 Moreover, the hepatocarcinoma cells used in these studies might have some characteristics different from normal hepatic cells. In our studies, we also found that hepatic cells at the early differentiation stage (day 8) could also be infected by HIV-HCV pseudotype viruses, suggesting that liver progenitor cells may be a possible cellular target of HCV infection in vivo. HCV-infected liver progenitor cells may play a key role in the pathogenesis of HCV infection. In fact, it has been reported that HCV was able to infect fetal hepatocytes.40 These HCV infected liver progenitor cells could potentially contribute to persistent viral infection,41 fibrosis by altered hepatocyte regeneration42, 43 and genesis of liver tumors.44
Although several previous studies5–9 have reported that hESCs could differentiate into hepatic cells, there is no report about whether these hESC-derived hepatic cells are engraftable. Here we showed that hESC derived hepatocyte could integrate at low efficiency into CCl4 injured mouse liver and retain the ability to express human AAT (Fig. 4F). A major problem for the liver cell engraftment is that very limited space is available for the donor cells,45 thus the transplanted liver cells integrate into the CCl4-treated mouse with only low efficiency.46 It has been reported that with certain mouse models with genetic diseases, such as Alb-uPA47 and FAH−/− mouse models,48 in which the recipient liver was constitutively injured, extensive liver repopulation could be reached. In the future, our studies will focus on constructing suitable mouse models for the hESC-derived hepatic cell transplantation. The cell sources of human primary hepatocytes remains limited and genetically heterogeneous, whereas hESC-derived hepatic cells could be potentially obtained on a large-scale with high reproducibility, and used for establishing mouse models with a humanized liver.
We thank Dr. Zhigang Lu and Matt for critical reading of the manuscript. We also thank Jing Zhang, Shuguang Duo, Dongxin Zhao, Xiaolei Yin, Yizhe Zhang, Chengyan Wang, Aihua Zheng, Fei Yuan, Jun Yong, Wei Jiang, Jie Yang, Jian Li, Pengbo Zhang, and other colleagues in our laboratory for technical assistance and advice in carrying out these experiments.