Differentiation and Enrichment of Hepatocyte-Like Cells from Human Embryonic Stem Cells In Vitro and In Vivo

Authors


Abstract

Human embryonic stem cells (hESC) may provide a cell source for functional hepatocytes. The aim of this study is to establish a viable human hepatocyte-like cell line from hESC that can be used for cell-based therapies. The differentiated hESC were enriched by transducing with a lentivirus vector containing the green fluorescent protein (GFP) gene driven by the α1-antitrypsin promoter; the GFP gene is expressed in committed hepatocyte progenitors and hepatocytes. GFP+ hESC were purified by laser microdissection and pressure catapulting. In addition, differentiated hESC that were transduced with a lentivirus triple-fusion vector were transplanted into NOD-SCID mice, and the luciferase-induced bioluminescence in the livers was evaluated by a charge-coupled device camera. GFP+ hESC expressed a large series of liver-specific genes, and expression levels of these genes were significantly improved by purifying GFP+ hESC; our results demonstrated that purified differentiated hESC express nearly physiological levels of liver-specific genes and have liver-specific functions that are comparable to those of primary human hepatocytes. The differentiated hESC survived and engrafted in mouse livers, and human liver-specific mRNA and protein species were detected in the transplanted mouse liver and serum at 3 weeks after transplantation. This is the first time that human albumin generated by hESC-derived hepatocytes was detected in the serum of an animal model. This also represents the first successful transplantation of differentiated hESC in an animal liver and the first bioluminescence imaging of hESC in the liver. This study is an initial step in establishing a viable hepatocyte-like cell line from hESC.

Disclosure of potential conflicts of interest is found at the end of this article.

Introduction

Liver dysfunction is a major health problem in the world. Unfortunately, human donor livers are in short supply, as are livers that would be unsuitable for transplantation yet suitable for use in a bioartificial liver. Therefore, it would be greatly beneficial if an unlimited supply of functional hepatocytes from other sources was generated. ESC [1], bone marrow stem cells [2], liver stem cells/oval cells [3], cord blood cells [4], and fetal hepatocytes [5] are cell types that display the potential to develop into viable hepatocytes. However, conditions for directing human stem cells to differentiate into a specific lineage, such as hepatocytes, are not yet fully defined.

It is postulated by many that the use of ESC may be the most effective strategy to develop lines that may be valuable in regenerative medicine. The focus of this study is to direct human ESC (hESC) along a hepatocyte lineage, to characterize the cells, and to use them in in vivo systems. The ESC is capable of self-renewal and has the capacity to differentiate into cell types from all three germ layers [6, [7]–8]. Previous reports have demonstrated that mouse and human ESC can generate cells with hepatocyte characteristics [1, 9, [10], [11], [12], [13]–14], and although these studies have shown expressions of albumin and other liver genes in the cells, in most reports, the investigators did not evaluate the level of these liver-associated genes by quantitative methods, suggesting nonphysiological levels. Our hepatocyte-like cells that are derived from hESC differ from those in the other reports in that they express liver-specific genes over a prolonged period while continuing to replicate in culture [11, [12], [13]–14]. Our intent in this study was to enrich the population so that they represented a high percentage of hepatocyte-like cells expressing physiological levels of liver-specific genes. However, the generation of enriched populations of hESC-derived hepatocyte-like cells remains a challenge. Laser microdissection and pressure catapulting (LMPC) is an established technology for high-precision isolation of homogeneous cell types from heterogeneous populations in live or fixed tissue [15, [16]–17]. By using a liver-specific lentivirus vector in conjunction with LMPC, we performed the enrichment of our hepatocyte-like cells from the differentiated hESC populations.

A goal of the present study was to establish whether hESC-derived cells that have been differentiated along a hepatocyte lineage will be potentially useful for therapeutic applications in vivo. This effort will require demonstration of whether cells can engraft, survive, and proliferate in intact immunodeficient mice. Imaging reporter gene expression in animal models appears to be a promising approach for repeatedly and noninvasively monitoring transplanted cells [18, 19]. The use of a new imaging modality was a major component of this study, and we used a bioluminescence optical charge-coupled device (CCD) camera system to evaluate luciferase expression in our differentiated hESC. Thus, the ability of the cells to engraft and survive after transplantation is assessed by a number of approaches, including immunohistochemistry, reverse transcription (RT) polymerase chain reaction (PCR), enzyme-linked immunosorbent assay (ELISA), and imaging with a CCD camera.

Materials and Methods

Lentivirus Vectors

The third-generation lentivirus vector used requires four constructs to produce virus [20]. To achieve liver-specific transgene expression, the human α1-antitrypsin (α1-AT) promoter was used as an internal promoter to drive the green fluorescent protein (GFP) gene (supplemental online Fig. 1A). The triple fusion vector, a second-generation lentivirus vector, requires three constructs to produce virus [18] (supplemental online Fig. 1B). The generation, concentration, and titration of the lentiviral vectors were undertaken as described [20]. The liver specificity of the lentivirus vector containing the α1-AT promoter was evaluated by fluorescence microscopy and quantitative PCR through transductions of human hepatoma cell lines and nonhepatoma cell lines. Primers and the probe sets used to detect human glyceraldehyde-3-phosphate dehydrogenasea (GAPDHa) [21], GAPDHb, GFP [22], and HIV-1 p24 amplicon [21] are listed in supplemental online Table 1.

Human ESC Line

Human ESC line HSF6 (University of California San Francisco; NIH code UC06) was propagated and maintained as described by the provider and Bodnar et al. [23].

Transduction of Differentiated hESC with Lentivirus Vectors and Fluorescence-Activated Cell Sorting

The differentiation of hESC was initiated by the formation of EBs under the modified culture protocol [11]. Briefly, undifferentiated hESC were treated with collagenase type IV and then seeded on a noncoated, 60-mm tissue culture dish with an ultralow-attachment surface (Fisher Scientific International, Houston, TX, http://www.fisherscientific.com) without mouse embryonic fibroblast feeder layers in Iscove's modified Dulbecco's medium (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) with 20% fetal bovine serum, 2 mM l-glutamine, 0.3 mM monothioglycerol (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), 1% antibiotic-antimycotic, 0.126 U/ml human insulin (Hospira, Inc., Lake Forest, IL, http://hospira.com), and 100 nM dexamethasone (Sigma-Aldrich). Nine days later, EBs were seeded for expansion on collagen type I precoated plates for differentiation under the same culture conditions. The first day of EB formation was designated the first day of differentiation.

Differentiated hESC were transduced with the lentivirus vector containing the α1-AT promoter driving the GFP gene at 10–14 days after EBs were plated (optimal final concentration of virus vector, 1.0 × 108 transduction unit/ml) in the presence of polybrene (8 μg/ml) for 12–16 hours [24, 25]. Fluorescence-activated cell sorting (FACS) analysis was performed 1 week after transduction on a MoFlo cell sorter (Dako, Fort Collins, CO, http://www.dako.com) for detection of GFP+ cells.

Immunochemistry of GFP+ hESC and Analysis by Flow Cytometry

Differentiated hESC were transduced with the liver-specific lentivirus at different time points after differentiation and then fixed in 4% paraformaldehyde for immunochemistry or trypsinized for comparison of the percentage for GFP+ cells or albumin (ALB)+ cells by flow cytometry. For immunochemistry, the cells were incubated at 4°C overnight with primary goat anti-human ALB and α1-AT antibodies (1:1,000) (Bethyl Inc., Montgomery, TX, http://www.bethyl.com) and primary monoclonal antibodies against human α-fetoprotein (AFP) (1:300) and cytokeratin 18 (CK18) (1:800) (Sigma-Aldrich). The next day, cells were incubated in Cy3-conjugated rabbit anti-goat IgG or Cy3-conjugatd goat anti-mouse IgG (1:1,000) (Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com). For flow cytometry, the transduced cells were fixed with Leucoperm reagent A (AbD Serotec, Raleigh, NC, http://www.ab-direct.com) and then were incubated with primary goat anti-human ALB antibody (1:50 to 1:100) and donkey anti-goat IgG-phycoerythrin (10 μl for 1–5 × 105 cells) (R&D Systems Inc., Minneapolis, http://www.rndsystems.com) in Leucoperm permeabilization reagent B. Then, two-color analysis was performed using a FACScan instrument (Becton, Dickinson and Company, San Jose, CA, http://www.bd.com) to detect the percentages of GFP+ cells and ALB+ cells from each sample.

Purification of GFP+ hESC by LMPC and Reverse Transcription for LMPC Samples

GFP+ hESC produced by transduction with the liver-specific lentivirus were isolated to increase the purity of differentiated cells by LMPC at different time points after differentiation. Prior to microdissection, the cells were fixed with 70% cold alcohol for 2 minutes and then immediately dehydrated with 100% cold ethanol. The slide was placed in the holder of the RoboStage (Carl Zeiss MicroImaging Inc., Thornwood, NY, http://www.palm-microlaser.com), and the GFP+ hESC were selected and cut under a fluorescent microscope and catapulted into the medium-filled cap positioned directly over the target area. The average cutting sizes were from 65,000 μm2 to 200,000 μm2 based on the area of GFP+ hESC, and each sample contained 1–3 pieces of microdissected GFP+ hESC. Total RNA from each LMPC sample was extracted by using a 6700 automated nucleic acid workstation (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) according to the manufacturer's instructions. The first-strand cDNA from all total RNA of each LMPC sample was synthesized using the SuperScript III system (Invitrogen) following the manufacturer's instructions.

Gene-Specific Preamplification and TaqMan PCR of LMPC Samples

For quantitative analysis of LMPC samples, the first-strand cDNA was preamplified using a gene-specific protocol. Briefly, 1/10 volume of the first-strand cDNA was preamplified using a 1:120 dilution of Assay-on-Demand (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) in a multiplex PCR for 20 PCR cycles. The number of PCR cycles for the preamplification was determined in a validation experiment and adjusted to prevent any preamplified gene from entering the PCR plateau phase to prevent PCR competition and alteration of amplification efficiencies. Amplification efficiencies of the multiplex preamplification reaction were assessed by running parallel reactions of the preamplification for different numbers of preamplification cycles. Subsequently, the preamplified materials were diluted 1:100 and analyzed by TaqMan PCR [26], and standard curves were plotted. From the slope of the standard curves generated on these parallel reactions, the amplification efficiencies were calculated. Primers/probes used in both preamplification and TaqMan PCR for determining expression of GAPDHc, ALB, α1-AT, transferrin (TF), and tyrosine aminotransferase (TAT) are listed in supplemental online Table 1.

PCR of Liver-Associated Genes and Transcriptional Factors

cDNAs from preamplifications of LMPC samples were diluted at 1:100, and standard PCR was performed using SuperMix Platinum (Invitrogen). The primers used to detect GAPDHc, CYP2B6, CYP2C9, CYP2E1, CYP3A4 [27], CYP1A1, CYP1B1, arginase (ARG), glucose-6-phosphatase (G-6-P) in both preamplification and PCR are listed in supplemental online Tables 1 and 2.

Extraction of RNA and generation of cDNA from total differentiated hESC at day 23 were performed as previously described [26]. Standard PCR was as described above, and the primers used to detect GAPDHa, hepatocyte nuclear factor (HNF/3β) [28], HNF4 [29], CCAAT/ehancer binding protein (C/EBPα, C/EBP)β [12], bone morphogenetic protein 2 (BMP2), BMP4 [29], and GATA-4 are listed in supplemental online Tables 1 and 2.

Hepatocyte Function Assay in Differentiated hESC

Differentiated hESC were cultured 2 weeks after EBs were plated, and the periodic acid-Schiff's (PAS) staining kit (Polysciences Inc., Warrington, PA, http://www.polysciences.com) was used to detect glycogen storage in the hepatocyte-like cells. Mouse fibroblasts were also stained at the same time to see whether there was any storage of glycogen accumulation under the same culture conditions. The staining was performed per the manufacturer's instructions.

Twenty-five milligrams of Indocyanine Green (ICG) (Akorn, Inc., Buffalo Grove, IL, http://www.akorn.com) was dissolved in 5 ml of solvent in a sterile vial and then added to 20 ml of medium to a final concentration of 1 mg/ml. The ICG solution was added to the differentiated hESC at day 35 and undifferentiated hESC, and incubated at 37°C for 30 minutes. After the dish was washed three times with phosphate-buffered saline, the cellular uptake of ICG was examined by microscopy.

Primary human hepatocytes cultured for 2 weeks [30], differentiated hESC (at day 28), and human prostatic adenocarcinoma cell line PC-3 cells were assessed for CYP1A2 activity as previously described [12, 30]. Briefly, cells were incubated in the presence of methylchloranthrene for 24 hours, and then CYP1A2 activity was determined by the ethoxyresorufin O-deethylase (EROD) assay in the presence or absence of ethoxyresorufin and measured as picomoles of resorufin formed per minute per million cells. The amount of resorufin formed was calculated by extrapolating the data against a standard curve of purified resourfin.

Every 48 hours, the medium was replaced with 4 ml of fresh medium for collecting the supernatant after EBs were seeded on coated six-well plates. The human ALB values secreted in the supernatant was determined by Human Albumin ELISA Quantitation kit (Bethyl) following the manufacturer's instructions.

Preparation of cDNA from Adult Human Primary Hepatocytes

Adult human primary hepatocytes isolated from donor livers were provided by Dr. Stephen Strom (University of Pittsburg). These fresh cells were used to prepare cDNA as previously described [26]. These cDNAs were used in real-time quantitative RT-PCR as calibrators.

Transplantation of hESC and Monitoring

Differentiated hESC were transduced with the triple fusion lentivirus vector as described above at 14 days after EBs were plated. One week later, the cells were detached by cell scraper, homogenized by needles, and then injected under the liver capsule and into parenchyma at approximately 5 × 105 cells per site in two sites, respectively, in each NOD-SCID mice. The triple fusion vector contained the luciferase gene, which produces bioluminescence. The addition of the substrate Beetle Luciferin (Promega, Madison, WI, http://www.promega.com) and subsequent camera imaging occurred from 24 hours to 1 week after the transplantation [31]. Control mice without cell transplantation but with substrate injection were also evaluated. Surgical procedures for transplantation and monitoring and subsequent collection of liver tissue were approved by the Animal Care and Use Administrative Advisory Committee of the University of California Davis.

Immunohistochemistry of Human Liver Gene Expression in Mouse Liver

Mice were sacrificed 3 weeks after transplantation, and pieces of liver tissue with transplanted cells were removed and placed in a Petri dish. The liver tissue was fixed, embedded in Tissue-Tek O.T.C. (Sakura Finetek Europe B.V., Zoeterwoude, The Netherlands, http://www.sakuraeu.com), and processed as described [32]. It was stored at −70°C until it was sectioned. Liver tissue from a control mouse was treated in the same manner. The sectioned slides were fixed, and immunostaining with goat anti-human ALB and α1-AT antibodies was performed as described above.

RT-PCR of Human Liver Gene Expression in Mouse Liver

The total RNAs from both transplanted and nontransplanted liver tissue were extracted, and cDNA was generated as previously described [26]. Before standard PCR, the first-strand cDNAs were preamplified as described above. The primers used to detect human GAPDHc, ALB, α1-AT, TF, AFP, CYP1B1, and mouse β-actin in both preamplification and PCR are listed in supplemental online Tables 1 and 2.

ELISA of Human Albumin in Mouse Serum

The first collection of mouse blood samples were taken by tail incision as baseline before transplantation, and then mouse blood samples were collected weekly after transplantation until the mice were sacrificed. Mouse blood samples were also collected from mice without transplantation as controls. ELISA was performed as described above.

Statistical Analysis

All data were summarized as means ± SEM from at least three independent measurements. An unpaired Student t test was used to analyze the data. p < .05 was considered statistically significant.

Results

Transduction with the Liver-Specific Lentivirus Vector and FACS

After transduction with the lentivirus vector containing the α1-AT promoter, a very high percentage of GFP+ cells was obtained in hepatoma cells (supplemental online Fig. 2A) compared with that in nonhepatoma cells (supplemental online Fig. 2B). The relative GFP expression driven by the α1-AT promoter was significantly higher in hepatoma cells than in nonhepatoma cells (supplemental online Fig. 2C). To determine whether this difference was caused by transduction efficiency, TaqMan PCR showed that the relative copy number of HIV-1 p24 amplicons from proviral DNA integrated into the host genome in hepatoma cells and nonhepatoma cells was not different (supplemental online Fig. 2D). These data demonstrated that high GFP expression in the hepatoma cells was not due to higher transduction efficiency; on the contrary, they showed that transduction with this lentiviral vector containing the GFP gene can be used to demonstrate liver-specific gene expression in transduced cells.

A number of hESC expressed the GFP gene and were positive by fluorescence microscopy after transduction with the liver-specific lentivirus vector (Fig. 1A, 1C). GFP+ cells tended to be in clumps and had similar size and morphology. The results of FACS analysis showed that as many as 14.2% of the transduced, differentiated hESC were GFP-positive (Fig. 1H). The mean fluorescent intensity was determined using cells that had signal intensities higher than the control cells without transduction with virus (Fig. 1F), which avoids the intrinsic background fluorescence of the cells. The mean percentage of transduced cells that were GFP+ was 11.2% ± 1.8%, and the percentage was 0.18% ± 0.003% in untransduced, differentiated hESC. The results suggest that the use of the liver-specific lentivirus vector may well be effective in enhancing the purity of our hepatocyte-like, differentiated hESC.

Figure Figure 1..

GFP expression evaluated by fluorescent microscopy and fluorescence-activated cell sorting (FACS) after transduction with a lentivirus vector. Differentiated human ESC (hESC) were transduced with the liver-specific lentivirus vector containing the α1-antitrypsin promoter driving the GFP gene at 2 weeks after EBs were plated. GFP was visualized 1 week after the transduction by fluorescent microscopy (A, C) (magnification, ×100). The images of these cells from normal optical microscopy were merged with corresponding images from fluorescent microscopy (B, D) (magnification, ×100). FACS analysis was performed after transduction. For detection of GFP+ cells, dead cells and debris were excluded by forward scatter and SSC plots in the histogram analysis plots from untransduced cells (E) and transduced cells (G). Differentiated hESC not transduced with the lentivirus vector were used as control cells (F). The mean fluorescent intensity was determined using cells that had signal intensities higher than that of the control cells (H). Abbreviations: GFP, green fluorescent protein; SSC, side scatter.

Immunochemistry of GFP+ hESC and Analysis by Flow Cytometry

After transduction with the liver-specific lentivirus vector, immunochemistry showed that GFP+ hESC expressed liver-specific proteins AFP, ALB, α1-AT, and CK18 (Fig. 2B, 2F, 2J, 2N), demonstrating a very high concordance with GFP+ cells and immunostaining. Flow cytometry results showed the percentages of ALB+ hESC from day 30–58 after differentiation (Fig. 2U), demonstrating a similarity to the percentage of GFP+ hESC (Fig. 2T, 2U). At an early stage of differentiation (around 22 days), GFP+ hESC strongly expressed AFP by immunochemistry (Fig. 2B). Later, AFP staining became very weak in the GFP+ cells up to day 52 (Fig. 2D). These results were confirmed by quantitative RT-PCR. Compared with the human hepatoma cell line Hep G2, the relative level of AFP in hESC was 2.04 ± 0.02 at 23 days and 0.16 ± 0.004 at 52 days after differentiation (primers/probe used are listed in supplemental online Table 1). During the same time period of differentiation, from day 22 to day 52, immunochemical staining of GFP+ cells revealed high and stable expressions of the more mature liver proteins ALB and α1-AT (Fig. 2F, 2H [ALB]; Fig. 2J, 2L [α1-AT]). These results suggest that more mature liver-specific gene expression is stable in the differentiated cells over time and that the transduction with the liver-specific lentivirus vector is effective. The decrease in AFP expression over time (at the same time that ALB expression is stable) strongly suggests that these cells are of hepatocyte lineage, not primitive endoderm.

Figure Figure 2..

Analysis of GFP+ hESC for expression of human liver-specific genes by immunochemistry and flow cytometry. In (A–P), differentiated hESC transduced with the liver-specific lentivirus vector containing the GFP gene were immunostained for AFP, ALB, α1-AT, and CK18 at 22D (left two columns) and 52D (right two columns) after differentiation. The first and third columns (A, E, I, M; C, G, K, O) were evaluated by green fluorescent microscopy; the second and fourth columns (B, F, J, N; D, H, L, P) were evaluated by red fluorescent microscopy after immunostaining with Cy3-conjugatd detection antibodies for AFP (B, D), ALB (F, H), α1-AT (J, L), CK18 (N, P). (A, B, E, F, I–P): magnification, ×100; (C, D, G, H): magnification, ×200. In (Q–T), two-color analysis by flow cytometry was performed after transduction of hESC with the liver-specific lentivirus vector followed by immunostaining for albumin at day 38. For detection of GFP+ cells and ALB+ cells, dead cells and debris were excluded by FSC and SSC plots in the histogram analysis plots from untransduced cells (Q) and transduced cells (R). Differentiated hESC not transduced with the lentivirus vector were used both as control for GFP detection and isotype control for ALB detection (S). The mean fluorescent intensity was determined using cells that had signal intensities higher than the controls (T). In the lower panel (U), comparison of the percentages of GFP+ cells and ALB+ cells by flow cytometry at different time points of differentiation after transduction of hESC with the liver-specific lentivirus (n ≥ 3 in each group). Abbreviations: α1-AT, α1-antitrypsin; AFP, α-fetoprotein; ALB, albumin; CK18, cytokeratin 18; D, days; FSC, forward scatter; GFP, green fluorescent protein; hES, human embryonic stem; PE, phycoerythrin; SSC, side scatter.

Expression of Liver-Specific Genes in LMPC GFP+ hESC

GFP+ hESC were purified by LMPC after differentiation, and the expression of liver-specific genes in these hESC at days 26, 30, 33, and 37 was determined by quantitative RT-PCR. The results showed that the expression levels of liver-specific genes were significantly improved by purifying GFP+ hESC through LMPC (Fig. 3). During these time points, the average relative levels were 50.25% ± 3.35% for α1-AT (Fig. 3A), 29.75% ± 2.25% for ALB (Fig. 3B), and 85.25% ± 4.17% (Fig. 3C) for TF compared with primary human hepatocytes. The expression level of TAT varied considerably (Fig. 3D).

Figure Figure 3..

Expression of liver-specific genes in purified green fluorescent protein (GFP)+ human ESC (hESC) by laser microdissection and pressure catapulting (LMPC). GFP+ differentiated hESC transduced with a liver-specific lentivirus vector containing the GFP gene were purified by LMPC at different times after differentiation. Quantitative PCR following preamplification was performed to determine expression levels of human liver-specific genes α1-AT (A), albumin (B), transferrin (C), and TAT (D). Relative values of each gene were calibrated with human primary hepatocytes (3 ≤ n ≤ 10 in each group). Abbreviations: α1-AT, α1-antitrypsin; D, days; HPH, human primary hepatocytes; ND, not done; TAT, tyrosine aminotransferase.

In addition, other indicators of more mature hepatocytes, G-6-P, ARG, CYP2B6, CYP2E1, CYP2C9, CYP3A4, CYP1A1, and CYP1B1, were also expressed in LMPC GFP+ hESC as determined by PCR following preamplification at day 30 after differentiation (Fig. 4A).

Figure Figure 4..

Polymerase chain reaction (PCR) for expression of liver genes in green fluorescent protein+ human ESC (hESC) and transcription factors in differentiated hESC. (A): PCR following preamplification was used to determine expression of human liver-associated genes CYP2B6, CYP2C9, CYP2E1, CYP3A4, CYP1A1, CYP1B1, ARG, and G-6-P at day 30 after differentiation. GAPDH was detected at the same time as a housekeeping gene. (B): PCR was used to detect expression of transcription factors HNF3β, HNF4, C/EBPα, C/EBPβ, BMP2, BMP4, and GATA4 in total hESC at day 23 after differentiation. GAPDH was detected at the same time as a housekeeping gene. Abbreviations: ARG, arginase; BMP, bone morphogenetic protein; bp, base pairs; G-6-P, glucose-6-phosphatase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Expressions of Liver-Associated Transcription Factors as Determined by PCR

Liver-associated transcription factors HNF3β, HNF4, GATA4, C/EBPα, C/EBPβ, BMP2, and BMP4 were all expressed, as determined by PCR from total hESC at day 23 after differentiation (Fig. 4B).

Liver-Specific Functions in Differentiated hESC

Differentiated hESC before isolation by LMPC showed glycogen accumulation by PAS staining, and these PAS-positive cells tended to have a hepatocyte-like morphology (Fig. 5A). There was no glycogen accumulation in adjacent cell populations (Fig. 5A, arrows), nor was there glycogen storage in mouse fibroblasts (data not shown). The cellular uptake of ICG is a unique characteristic of hepatocytes. ICG uptake was also observed in a percentage of differentiated hESC, the positive cells showing green staining (Fig. 5B); there was no ICG uptake in undifferentiated hESC when stained with ICG (data not shown), indicating that ICG uptake only occurred after differentiation. EROD assay results showed that the CYP1A2 activity of the differentiated hESC was inducible at levels that was comparable to levels in cultured human primary hepatocytes, and no CYP1A2 activity was found in PC-3 cells (Fig. 5C). ELISA showed that the differentiated hESC secreted albumin as early as 1 week after the EBs were seeded, and these cells secreted higher levels at later time points (Fig. 5D).

Figure Figure 5..

Hepatic functional assays of differentiated human ESC (hESC). (A): hESC were detected by periodic acid-Schiff's staining for glycogen accumulation at 23 days after differentiation (magnification, ×100). (B): hESC were stained by Indocyanine Green (ICG) staining for cellular ICG uptake at 35 days after differentiation (magnification, ×100). (C): CYP1A2 activity was measured by the EROD assay at day 28 after differentiation compared with cultured human primary hepatocytes and the adenocarcinoma cell line PC-3 cells (n = 6 in each group). (D): Albumin secretion by the differentiated hESC was determined by enzyme-linked immunosorbent assay test from day 11 to day 48 during differentiation (n ≥ 4 at each time point). Abbreviations: EROD, ethoxyresorufin O-deethylase; hES, human embryonic stem.

Molecular Imaging by CCD Camera of Mice Transplanted with hESC

From 24 hours to 1 week after transplantation of NOD-SCID mice with hESC transduced with the triple fusion lentivirus, the whole mouse was imaged with a CCD camera. Positive signals were obtained in four of six mice (Fig. 6A). Control mice without transplantation with hESC were also imaged at the same time (Fig. 6B). The results indicate that differentiated hESC survived after transplantation in mouse livers.

Figure Figure 6..

Charge-coupled device (CCD) camera image of differentiated human ESC (hESC) after transplantation. (A): Differentiated hESC that were transduced with the lentiviral triple-fusion vector were injected under the liver capsule and into parenchyma in NOD-SCID mice. The addition of the substrate and subsequent imaging by CCD camera occurred from 24 hours to 1 week after the transplantation. (B): A control mouse without cell transplantation but with substrate injection. Abbreviations: Max, maximum; Min, minimum; p, photon; sec, seconds; st, steradian.

Analysis of Human Liver-Associated Gene Expressions by hESC in Mouse Liver and Serum

Liver tissue from mice that were signal-positive by CCD camera after transplantation with hESC was prepared for further analysis at 3 weeks after transplantation. Immunochemistry showed cells positive for human ALB and α1-AT staining (Fig. 7A, 7D, shown by arrow). There was no cross-reactivity with mouse ALB and α1-AT. Most positive cells had similar size and morphology and were positive for both ALB (Fig. 7G) and α1-AT (Fig. 7H) staining. No expression for human ALB or α1-AT was detected in the livers of the control mice (data not shown). Human liver-specific gene expression was evaluated by standard PCR following preamplification. Human ALB, α1-AT, TF, CYP1B1, and GAPDH were expressed in transplanted mouse livers (Fig. 7I). There was no signal for human AFP in these mouse livers. No human gene expression was found in the liver samples of a control mouse that did not Received human cells (data not shown).

Figure Figure 7..

Immunochemistry, polymerase chain reaction (PCR), and enzyme-linked immunosorbent assay (ELISA) of human liver-specific gene expression by human ESC (hESC) in mouse livers and serum. Red fluorescent microscopy of transplanted mouse liver tissues that were immunostained with primary anti-human antibodies for albumin (ALB) (A, G) and for human α1-antitrypsin (α1-AT) (D, H) followed by staining with Cy3-conjugatd detection antibodies. (B, E): Normal optical images of mouse liver tissues. The images of these liver slides when the blue fluorescence of 4,6-diamidino-2-phenylindole was merged with the corresponding red fluorescent immunostaining for ALB (C) or for α1-AT (F). (A–F): magnification, ×100; (G, H): magnification, ×200. (I): PCR was used to detect human liver gene expression from mouse liver tissue following preamplification for the following genes: ALB (lane 2), α1-AT (lane 3), transferrin (lane 4), α-fetoprotein (lane 5), CPY1B1 (lane 6), and glyceraldehyde-3-phosphate dehydrogenase (lane 7); lane 1 shows a marker. Mouse β-actin was evaluated at the same time as a housekeeping gene. (J): ELISA was performed to detect human ALB in the mouse serum of transplanted mice from 1 week to 3 weeks after transplantation. Abbreviation: bp, base pairs.

ELISA showed that human albumin was detected in the mouse serum of three transplanted mice from 1 week to 3 weeks after transplantation (Fig. 7J) and that human albumin was not detectable in the serum of these mice before transplantation. No human albumin was detected in the serum of control mice (data not shown).

Discussion

It is evident that improved and safer liver transplantation would be valuable, as would approaches that provide for more transplantations in a timely manner. Developing a hepatocyte-like cell line from hESC would provide a valuable tool for pharmacology and toxicology studies, as well as for use in cell-based therapeutics.

Our initial studies showed the partial differentiation of hESC to hepatocyte-like cells by optimizing the culture condition, and hESC-derived hepatocyte-like cells expressed liver-specific genes and function over a prolonged period [11]. Although a variety of factors and reagents were beneficial, dexamethasone was particularly important in our optimal condition [11]. Dexamethasone can enrich the endodermal population at an early stage of differentiation with a limited exposure to serum [10, 33]. It is thought to promote the expression of the differentiated hepatocyte phenotype through suppression of cell division [34, 35], and it also is active in the induction of enzymes concerned in gluconeogenesis in the liver, such as phosphoenolpyruvate carboxykinase, tyrosine aminotransferase, and others [33, 36]. In the present study, we used a liver-specific lentivirus vector in conjunction with LMPC to enhance the purity of this differentiated cell population. We initiated this study by determining that lentiviral transduction did not affect the pluripotent characteristics of the hESC (data not shown).

There have been other investigators who have attempted to purify differentiated hepatocyte-like cells using reporter systems with mouse ESC [13, 37], but because these transgene vectors were plasmids, the transgene expression was transient and did not pass to the progeny. In our study, we used lentivirus as the transgene vector for stable transgene expression that is passed to the progeny. In addition, we used the human liver-specific α1-AT promoter to drive the transgene. α1-AT is a late gene product during differentiation; its promoter is initiated for transcription at stages of committed hepatocyte progenitors and hepatocytes [38, 39]. Our results suggest that transduction with a lentivirus vector containing this promoter driving the transgene could be used to demonstrate liver-specific gene expression in transduced cells (Fig. 1; supplemental online Fig. 2). This liver specificity was also confirmed by immunohistochemistry and flow cytometry. There was a very high concordance with GFP+ cells and immunostaining with AFP, albumin, and α1-AT (Fig. 2).

Both immunochemistry and quantitative RT-PCR showed that expression levels of AFP decreased in GFP+ hESC over time (Fig. 2B, 2D); during the same time period, immunostaining of GFP+ hESC revealed high and stable expression of the more mature liver proteins ALB and α1-AT (Fig. 2F, 2H [ALB]; Fig. 2J, 2L [α1-AT]). ELISA and quantitative PCR evaluation of mature liver-specific genes (ALB, α1-AT, and TF) revealed stable levels of expression over time (Figs. 3A–3C, 5D). The combination of these findings strongly suggests that these cells are of hepatocyte lineage, not primitive endoderm. Moreover, major pluripotent-associated genes OCT4 and Nanog were not detectable in LMPC GFP+ hESC (supplemental online Table 3), indicating that the transition of hESC from pluripotent to a differentially more restricted state is accompanied by global changes in gene expression.

The GFP+ hESC express a large series of liver-associated genes (ALB, α1-AT, AFP, CK18, TF, TAT, G-6-P, ARG, CYP1A1, CYP2B6, CYP1B1, CYP2E1, CYP2C9, and CYP3A4) as determined by PCR and/or immunochemistry and/or ELISA (Figs. 2, 3, 4A, 5D). ALB, α1-AT, TF, TAT, G-6-P, and ARG are indicators of more mature hepatocytes and play important functions in the liver. CYPs play key roles in the detoxification of drugs or other xenobiotics as well as endogenous substrates in the liver [40, 41]. These CYPs are present in human liver by expression or induction; for example, CYP2B6, CYP2E1, CYP2C9, and CYP3A4 are expressed in human liver microsomes [40, 42, [43], [44], [45]–46], and they are considered relatively hepatocyte-specific [2, 47], especially CYP3A4 [46]. Differentiated hESC also accumulate glycogen as shown by PAS staining, demonstrate ICG uptake, CYP1A2 activity, and secretion of albumin (Fig. 5), as well as synthesize urea [11]. CYP1A2 is expressed primarily in the liver and plays a central role in the metabolism of aromatic amines, estrogen compounds, caffeine, and certain drugs [42, 47, 48]. The CYP1A2 activity of the differentiated hESC reached higher levels when induced (Fig. 5C). These results demonstrate that our differentiated hESC express liver-specific functions that are comparable to those of primary human hepatocytes.

Liver-associated transcriptional factors also play important roles in hepatocyte differentiation and liver development [49]. HNF3β, HNF4, GATA-4, C/EBPα, C/EBPβ, BMP2, and BMP4 were all expressed during differentiation of our hESC (Fig. 4B). By embryonic day (E) 8.5 in the mouse, definitive endoderm has formed the gut tube and expresses HNF3β, whereas the foregut endoderm is induced toward the hepatocyte lineage by a-fibroblast growth factor (aFGF), bFGF, and FGF8 produced by the adjacent cardiac mesoderm. These factors are required to induce a hepatic fate and not the default pancreatic fate [50, 51]. HNF4 is first detected in the primitive endoderm of the blastocyst at E4.5 [52], and the ability of HNF4 to regulate liver genes, as well as its expression throughout hepatic development, suggests a significant role for this factor in the differentiation of the hepatocyte lineage [52]. BMP2 and BMP4 produced by the transversum mesenchyme are required for hepatocyte differentiation, as they increase levels of the GATA-4 transcription factor and are required for albumin expression, while downregulating pancreatic gene expression [53, 54]. The increase in GATA-4 primes the endoderm to become responsive to FGFs [51]. GATA-4, HNF3β, and BMP4 are required for hepatic specification and are important mediators of hepatocyte differentiation [55, 56]. C/EBP α and C/EBPβ are also involved in both liver development and mature liver functions [49] Thus, the expression of these factors in our hESC is crucial to prime these cells for further hepatocyte differentiation. The cell source of these factors is not clear from our analysis.

The use of the liver-specific lentivirus vector and LMPC appears to be effective in enhancing the purity of hepatocyte-like, differentiated hESC. These results demonstrate that LMPC-treated differentiated hESC have expressed near-physiological levels of liver-specific genes. Heterogeneity is a major obstacle to in vitro differentiation of embryonic and adult stem cells [57, 58], and the ability to differentiate ESC into specific cell types is a significant challenge that must be addressed prior to therapeutic applications. LMPC is a robust technology that offers an innovative method to isolate homogeneous cell types from heterogeneous populations [59]. This technology has been improved with the new PALM Microbeam HT platform (Carl Zeiss MicroImaging); this platform allows for the isolation of cells in a viable state, which in turn permits subsequent reculturing [15, 16, 57]. Thus, not only will we be able to effectively isolate individual or groups of GFP+, hepatocyte-like cells from a mixed population of lentivirus-transduced cells, but in the future, our aim will be to reculture this enriched cell population following the LMPC procedure.

The present study is also an initial attempt to establish the in vivo fate of the hepatocyte-like cells. The novel technology of noninvasive and repetitive molecular imaging of cells within a living subject was also used in this study [19]. Differentiated hESC that were transduced with the lentiviral triple-fusion vector were injected under the liver capsule and into parenchyma in NOD-SCID mice and evaluated 1 week after cell transplantation with the CCD camera (Fig. 6A). This represents the first successful transplantation and engraftment of differentiated hESC in an animal liver and also the first bioluminescent imaging of hESC in the liver. RT-PCR, ELISA, and immunochemistry showed abundant amounts of human liver-specific mRNA and protein species in the mouse liver and serum at 3 weeks after transplantation, when the animals were sacrificed (Fig. 7). To our knowledge, this is the first time that human liver protein generated by hESC-derived hepatocytes was detected in the serum of an animal model. Of note, there was no signal for AFP, suggesting that the cells were well-differentiated and/or become more differentiated in vivo. These results validate the use of a lentivirus containing a reporter gene for bioluminescence imaging as a potential noninvasive approach to monitoring hepatic stem cell engraftment and repopulation in a liver cell transplantation model.

In summary, we exploited a novel system for directing and enriching hepatic differentiation from hESC in vitro and in vivo. The use of the liver-specific lentivirus vector and LMPC appear to be effective in enhancing the purity of hepatocyte-like, differentiated hESC. Our study also demonstrated our ability to transplant cells effectively and safely in NOD-SCID mice and to effectively use the triple fusion lentivirus vector and the CCD camera to track transplanted cells in vivo. These results represent the initial step in establishing a viable hepatocyte-like cell line from hESC that can be used in cell-based therapeutics or used for pharmacology and toxicology studies.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.

Acknowledgements

We gratefully thank Drs. Stephen Strom, Mark Feitelson, Alan Chen, and Ralph W. deVere White for providing cells for this study. This work was supported in part by NIH Grants AA014173 and DK075415 (to M.A.Z.), DK46952 (to S.G.), PAR-04-069, HL078632, and CA082214 (to S.S.G.); by the Alpha-1 Foundation (to M.A.Z.); by the Stem Cell Foundation (to M.A.Z.); and by California Institute of Regenerative Medicine (to M.A.Z.).

Ancillary