Generation of endoderm-derived human induced pluripotent stem cells from primary hepatocytes


  • Potential conflict of interest: Nothing to report.


Recent advances in induced pluripotent stem (iPS) cell research have significantly changed our perspective on regenerative medicine. Patient-specific iPS cells have been derived not only for disease modeling but also as sources for cell replacement therapy. However, there have been insufficient data to prove that iPS cells are functionally equivalent to human embryonic stem (hES) cells or are safer than hES cells. There are several important issues that need to be addressed, and foremost are the safety and efficacy of human iPS cells of different origins. Human iPS cells have been derived mostly from cells originating from mesoderm and in a few cases from ectoderm. So far, there has been no report of endoderm–derived human iPS cells, and this has prevented comprehensive comparative investigations of the quality of human iPS cells of different origins. Here we show for the first time reprogramming of human endoderm-derived cells (i.e., primary hepatocytes) to pluripotency. Hepatocyte-derived iPS cells appear indistinguishable from hES cells with respect to colony morphology, growth properties, expression of pluripotency-associated transcription factors and surface markers, and differentiation potential in embryoid body formation and teratoma assays. In addition, these cells are able to directly differentiate into definitive endoderm, hepatic progenitors, and mature hepatocytes. Conclusion: The technology to develop endoderm–derived human iPS cell lines, together with other established cell lines, will provide a foundation for elucidating the mechanisms of cellular reprogramming and for studying the safety and efficacy of differentially originated human iPS cells for cell therapy. For the study of liver disease pathogenesis, this technology also provides a potentially more amenable system for generating liver disease-specific iPS cells. (HEPATOLOGY 2010;51:1810–1819)

Recent advances in induced pluripotent stem (iPS) cell research have shown great potential for these somatic cell–derived stem cells as sources for cell replacement therapy and for establishing disease models.1-14 Similar to human embryonic stem (hES) cells, human iPS cells have been shown to be pluripotent in in vitro differentiation and in vivo teratoma assays.9-14 Disease-specific iPS cell lines have been generated from fibroblasts and blood cells, and some of the disease features have been recapitulated in tissue culture after directed differentiation of the iPS cells; this has demonstrated the power of this technology in disease modeling.13, 15 However, several key issues have to be addressed in order for iPS cells to be used for clinical purposes. First, although pluripotency has been demonstrated, it is premature to claim that iPS cells are functionally equivalent to hES cells. In fact, one study has suggested that iPS cells have protein-coding and micro-RNA gene expression signatures distinct from those of embryonic stem (ES) cells.1 These differences cannot be completely explained by the reactivation of transgenes used in the reprogramming process because human iPS cells generated without viral or transgene integration have also displayed a different transcriptional signature in comparison with hES cells.2 Second, it has been demonstrated that human iPS cells retain certain gene expressions of the parent cells, and this suggests that iPS cells of different origins may possess different capacities to differentiate.2 This issue is important not only for the generation of functional cell types for therapy but also for safety implications. A comprehensive study using various mouse iPS cells has demonstrated that the origin of the iPS cells has a profound influence on the tumor-forming propensities in a cell transplantation therapy model.3 Mouse tail-tip fibroblast iPS cells (mesoderm origin) have shown the highest tumorigenic propensity, whereas gastric epithelial cells and hepatocyte iPS cells (both are endoderm) have shown lower propensities.3 It is therefore extremely important to establish human iPS cell lines of multiple origins and thoroughly examine the source impact on both the safety issues and their differentiation potentials. In addition, the ability to reprogram human hepatocytes is crucial for developing liver disease models with iPS cells, especially for certain liver diseases involving acquired somatic mutations that occur only in hepatocytes of patients and not in other cell types.16-20

In the mouse, iPS cells have been generated from derivatives of all three embryonic germ layers, including mesodermal fibroblasts,6 epithelial cells of endodermal origin,7 and ectodermal keratinocytes,8 whereas human iPS cells have been produced mostly from mesoderm (fibroblasts and blood cells) or ectoderm (keratinocytes and neural stem cells).9-13, 21, 22 Here we show the reprogramming of human primary hepatocytes (endoderm) to pluripotency. Hepatocyte-derived iPS cells appear indistinguishable from hES cells with respect to colony morphology, growth properties, expression of pluripotency-associated transcription factors and surface markers, and differentiation potential in embryoid body (EB) formation as well as teratoma assays. In addition, these cells are able to directly differentiate into definitive endoderm, hepatic progenitors, and mature hepatocytes.

Our study lays the groundwork necessary for elucidating the mechanisms of cellular reprogramming and for studying the safety and efficacy of differentially originated human iPS cells in cell therapy.


AAT, alpha 1-antitrypsin; AFP, alpha-fetoprotein; ALB, albumin; CXCR4, chemokine (C-X-C motif) receptor 4; CYP1A2, cytochrome P450 1A2; CYP3A4, cytochrome P450 3A4; CYP450, cytochrome P450; EB, embryoid body; EGF, epidermal growth factor; ES, embryonic stem; FACS, fluorescence-activated cell sorting; FBS, fetal bovine serum; FGF, fibroblast growth factor; hES, human embryonic stem; HGF, hepatocyte growth factor; hHiPS, human hepatocyte–derived induced pluripotent stem; iMSC, mesenchymal stem cell–derived induced pluripotent stem; iPS, induced pluripotent stem; KLF4, Kruppel-like factor 4; KOSR, knockout serum replacement; MEF, mouse embryonic fibroblast; MSC, mesenchymal stem cell; OCT4, octamer 4; SMA, smooth muscle actin; SOX2, SRY (sex determining region Y)-box 2; WEM, Williams' medium E.

Materials and Methods

Cell Culture.

Primary human hepatocytes, obtained from Lonza, were plated on collagen 1 and Matrigel-coated dishes and cultured in serum containing Williams' medium E (WEM), gentamicin, 0.1 uM dexamethasone, 5% fetal bovine serum (FBS), L-glutamine, 15 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid, 4 ug/mL insulin with 50 ng/mL hepatocyte growth factor (HGF), and epidermal growth factor (EGF). The medium for culturing hES cells and iPS cells was knockout Dulbecco's modified Eagle's medium supplemented with 20% knockout serum replacement (KOSR), nonessential amino acids, 2-mercaptoethanol, GlutaMAX, and 6 ng/mL basic fibroblast growth factor (FGF; all from Invitrogen). The hES cell lines WA09 (H9) and WA01 (H1; WiCell) were cultured on irradiated mouse embryonic fibroblast (MEF) feeder layers in ES medium. This study was performed in accordance with the regulations of the Johns Hopkins Embryonic Stem Cell Research Oversight Committee and with a protocol approved by the Johns Hopkins Institutional Review Board.

Retroviral Production and Reprogramming of Hepatocytes.

Retroviruses for the four factors were independently produced after cotransfection of the 293T cell line in 150-mm dishes with pMX retroviral vectors expressing octamer 4 (Oct4), SRY (sex determining region Y)-box 2 (Sox2), Kruppel-like factor 4 (Klf4), or c-Myc (Addgene) and helper plasmids as previously described.13 The viral supernatant was concentrated approximately 100-fold with Amicon Ultra centrifugal filters (Millipore). Primary hepatocytes were seeded in six-well tissue culture plates at a density of 105 cells/well and were cultured for 2 days before infection. A 1:1:1:1 mix of retroviruses containing Oct4, Sox2, Klf4, and c-Myc was filtered through a 0.45-μm cellulose acetate filter before being added to hepatocytes (passage 1) at a multiplicity of infection of 10 in the presence of 6 ng/mL polybrene.

After incubation for 3 days in WEM, the medium was replaced with hES medium. After transformed colonies were observed in the reprogramming plates, a reliable pluripotent stem cell marker,23 TRA-1-60 antibody (1:200 dilution; Millipore), and Alexa 555-conjugated anti-mouse immunoglobulin M antibody (1:500 dilution; Invitrogen) were added to a live cell culture (without fixation) and incubated for 1 hour at 37°C to distinguish the iPS colonies from non-iPS colonies. TRA-1-60-positive colonies appeared in about 6 to 9 days after retroviral transduction, and individual TRA-1-60-positive colonies were placed onto MEF-coated plates.

Immunofluorescence and Fluorescence-Activated Cell Sorting (FACS) Analysis.

Cells were fixed with 4% paraformaldehyde. The following antibodies were used: TRA-1-60 (Millipore; 1:100); SSEA4 (Cell Signaling; 1:200), SSEA3 (Millipore; 1:200); Neuron-specific class III beta-tubulin (Tuj1) (Covance; 1:500), alpha-fetoprotein (AFP; Dako; 1:200), smooth muscle actin (SMA; Dako; 1:100), OCT4 (Millipore; 1:100), NANOG (BD; 1:200), anti-SSEA3 488 (eBiosciences); chemokine (C-X-C motif) receptor 4 (CXCR4; BioLegend; 1:100), albumin (ALB; Dako; 1:200), alpha 1-antitrypsin (AAT; Thermo; 1:200), and cytochrome P450 3A4 (CyP3A4; Enzo; 1:200). The secondary antibodies all belonged to the Alexa Fluor series from Invitrogen.

EB Formation and Spontaneous Differentiation into Three Germ Layer Cells.

HES cells or iPS cells were dissociated by collagenase IV digestion and plated on ultralow attachment plates (Corning) at a density of approximately 1 × 106 cells per well in the presence of a differentiation medium (Dulbecco's modified Eagle's medium supplemented with 20% FBS, L-glutamine, β-mercaptoethanol, and nonessential amino acids). Fifty percent of the medium was replaced with fresh medium every 2 days. After 7 days, the EBs were transferred to 0.1% gelatin-coated culture dishes and cultured for another 2 to 3 days before fixation and staining. Antibodies against Tuj1 (Covance; 1:500), AFP (Dako; 1:200), or SMA (Dako; 1:100) were used to detect the spontaneously differentiated cells from EBs.

Teratoma Formation.

Ten-week-old male nonobese diabetic/severe combined immunodeficient/Il2γC−/− mice (Jackson Laboratories) were anesthetized, and approximately 1 million human hepatocyte–derived induced pluripotent stem (hHiPS) cells, resuspended in 20 to 40 μL of 50% Matrigel, were injected subcutaneously. The mice were euthanized 12 weeks after cell injection, and the tumors were analyzed according to a hematoxylin and eosin protocol. All animal experiments were conducted according to experimental protocols previously approved by the Johns Hopkins Institutional Animal Care and Use Committee.

Differentiation of hES/iPS Cells to Definitive Endoderm.

Sixty percent-confluent cultures were placed in Roswell Park Memorial Institute 1640 medium (supplemented with GlutaMAX and 0.5% defined FBS or 1% KOSR and 100 ng/mL activin A; R&D Systems) for 5 days. The medium was replaced every other day.

Differentiation of Definitive Endoderm to Hepatic Progenitor Cells.

Day 5 endoderm cultures were passaged with 0.05% trypsin/0.53 mM ethylene diamine tetraacetic acid and plated onto collagen I-coated dishes in Roswell Park Memorial Institute 1640 medium supplemented with GlutaMAX and 2% KOSR, 10 ng/mL FGF4, and 10 ng/mL HGF. Two days later, the cells were switched to minimal Madin-Darby bovine kidney maintenance medium (Sigma) supplemented with GlutaMAX and 0.5 mg/mL bovine serum albumin, 10 ng/mL FGF4, and 10 ng/mL HGF. Day 10 hepatic progenitors were used for experiments.

Differentiation of Hepatic Progenitor Cells to Mature Hepatocytes.

Hepatic progenitor cells were switched to complete hepatocyte culture medium supplemented with SingleQuots (Lonza) and containing 10 ng/mL FGF4, 10 ng/mL HGF, 10 ng/mL oncostatin M (R&D Systems), and 10−7 M dexamethasone (Sigma-Aldrich). Differentiation was continued for another 10 days.

Periodic Acid-Schiff Assay for Glycogen.

Cells were fixed with 4% paraformaldehyde and stained with a periodic acid-Schiff staining system (Dako). Cells were counterstained with Hematoxylin-QS and mounted with Vectamount AQ (all from Vector Laboratories).

Cytochrome P450 (CYP450) Assay.

Cytochrome P450 1A2 (CYP1A2) and CYP3A4 activity was assessed with the pGlo kit (catalog numbers V8771 and V8901, Promega) according to the manufacturer's instructions for nonlytic CYP450 activity estimation. hHiPS cell-derived mature hepatocytes were incubated with hepatocyte culture medium supplemented with CYP3A4 or CYP1A2 pGlo substrates. Four hours after treatment, 50 μL of the culture medium was removed and read on a GloMax model 9101-002 luminometer. The CYP450 activity was expressed as relative light units per milliliter of medium and was normalized against the percentage of ALB-expressing hepatocyte-like cells.


We have previously demonstrated the successful generation of human iPS cells from blood cells by retroviral transduction of transcription factors.13 For the generation of iPS cells of human endoderm origin, human primary hepatocytes were seeded in WEM with human HGF and EGF (Fig. 1A). These hepatocytes were more than 99% ALB-positive before reprogramming (Fig. 1B). Under these conditions, the primary human hepatocytes survived only for a short term (about 7-10 days) and did not proliferate even with HGF and EGF. After hepatocyte retroviral transduction with Oct4, Sox2, Klf4, and c-Myc (Fig. 1C), several hundred (i.e., 100-500) fast-growing small colonies per 105 hepatocytes were observed. By days 6 to 9 post-infection, at least 30% of these colonies displayed typical hES cell-like morphology and stained positively for TRA-1-60 (Fig. 1D,E). TRA-1-60 antibody live staining was performed, and TRA-1–60–positive colonies were picked and seeded onto a layer of irradiated MEFs in hES cell medium.

Figure 1.

hHiPS cell colony formation and characterization. (A) Phase contrast image and (B) ALB expression of primary human hepatocytes before iPS reprogramming (×200; scale bar = 50 μm). (C) Diagram of the hHiPS generation protocol. (D) Typical example of a small ES cell–like TRA-1-60–positive colony (red) adjacent to a TRA-1-60–negative non-iPS colony (arrow). (E) Typical ES cell–like TRA-1-60–positive iPS cell colony before harvest (×100; scale bar = 100 μm). (F) Representative immunofluorescence analysis of one hHiPS cell line (hHiPS10) growing on Matrigel. Clear expression of ES cell surface antigens SSEA4 and TRA-1-60 and nuclear transcription factors OCT4 and NANOG can be observed (×200; scale bar = 50 μm).

Some of these colonies showed a mosaic pattern of TRA-1-60+ and TRA-1-60 cells (Fig. 1D; the arrow indicates a TRA-1-60–negative non-iPS colony adjacent to a TRA-1-60–positive iPS colony). These TRA-1-60–negative colonies were usually darker and thicker and proliferated faster than TRA-1-60–positive colonies during the initial days (up to the first 14 days). However, the majority of these non-iPS clones were not able to survive for a long time (they disappeared within a few passages; the data are not shown), and in most cases, these non-iPS colonies were easily distinguishable by morphology from hES cell-like iPS colonies.

Seventeen TRA-1-60–positive single colonies were picked and passaged onto fresh feeder cells. Most of these subcultures (16 of 17) expanded and gave rise to stable cell lines with an hES cell-like morphology. All the lines could be maintained by standard hES cell culture procedures on both feeder cells and feeder-free Matrigel-coated plates (examples are shown in Supporting Fig. 1A). When fingerprinting was performed to amplify the microsatellite sequences by polymerase chain reaction, all these iPS cells displayed a pattern identical to that of the original primary hepatocytes (Supporting Fig. 1B). Four hHiPS cell lines (hHiPS6, hHiPS10, hHiPS11, and hHiPS14) were selected for further analyses. These lines were maintained in a continuous culture for over 8 months without signs of replicative or karyotypic crisis (Supporting Fig. 1C). All hHiPS cells expressed transcriptional regulators and cell surface markers characteristic of hES cells, including NANOG, OCT4, SSEA4, and TRA-1-60 (Fig. 1F for hHiPS10 and Supporting Fig. 2 for hHiPS6, hHiPS11, and hHiPS14). These cell lines also showed a complete loss of hepatocyte-specific markers such as ALB (data not shown). Overall, the expression of stem cell markers in hHiPS cells was indistinguishable from that of hES cell lines maintained under the same conditions.

We tested the pluripotency of hHiPS cell lines in assays of EB formation in vitro and teratoma induction in vivo. All tested cell lines readily differentiated in vitro into EBs (Fig. 2A,B), which were followed by endoderm, mesoderm, and ectoderm derivatives that stained positively for AFP, SMA, and TuJ1 immunoreactivity, respectively (Fig. 2C-H). About 3 months after injection into immunocompromised mice, independent hHiPS cell lines generated complex teratomas (Fig. 2I-N). Structures resembling all three germ layer tissues, such as glandular epithelia (endoderm; Fig. 2I,J), immature cartilages (mesoderm; Fig. 2K,L), and neural rosettes (ectoderm; Fig. 2M,N), were observed, and this indicated the pluripotency of these hHiPS cell lines.

Figure 2.

hHiPS cells can differentiate into all three primary germ layers in vitro and in vivo. (A,B) EBs derived from the hHiPS10 and hHiPS11 lines, respectively. (C-H) In vitro differentiation of (C,E,G) hHiPS10 and (D,F,H) hHiPS11 cells into all three primary germ cell layers. After the generation of EBs, hHiPS cells spontaneously differentiated into (C,D) endoderm (AFP-positive; red), (E,F) mesoderm (SMA-positive; red), and (G,H) ectoderm (TuJ1-positive neuronal cells; green). The blue nuclear staining was 4′,6-diamidino-2-phenylindole (×100; scale bar = 100 μm). (I-N) Spontaneous differentiation into all three germ layers was evident in teratomas: (I,J) endoderm, (K,L) mesoderm, and (M,N) ectoderm derived from hHiPS10 and hHiPS14 (×200; scale bar = 50 μm).

Taken together, these results demonstrate that human endoderm cells (i.e., hepatocytes), like other germ layer–derived cells (i.e., fibroblasts or blood cells), can be reprogrammed to pluripotency. Morphologically, nascent hHiPS cell colonies could be identified as early as 6 days post-infection. In contrast, colonies of iPS cells generated by retroviral transduction of human mesenchymal stem cells (MSCs) under similar experimental conditions appeared after day 10 post-infection (Supporting Fig. 3), and this was similar to the situation for blood cells.13 We observed about 50 to 100 hHiPS cell colonies from approximately 50,000 viable infected hepatocytes on the basis of morphological criteria and TRA-1-60 staining, and this represented an overall reprogramming efficiency close to 0.1% to 0.2% (n = 3). No significant differences of overall efficiency were observed in the generation of iPS cells from multiple independent human primary hepatocyte cultures.

Several recent reports have shown that human fibroblast (mesoderm origin)-derived iPS cells can be directed to hepatocytes just like ES cells.14, 24 We tested whether the endodermal-origin iPS cells could also be efficiently directed into hepatic cells. We used a hepatic differentiation protocol established for hES cells25 with a slight modification. Our hepatic differentiation protocol was composed of three stages: definitive endoderm induction for 5 days (day 5), hepatic progenitor induction and expansion for another 5 days (day 10), and hepatic maturation for another 10 days (day 20). Cells from these three stages (days 5, 10, and 20) were chosen because we observed that these time frames were most distinct in terms of their marker expression profile (i.e., based on CXCR4, AFP, or ALB positivity) previously for ES cells (data not shown). With this protocol, hES cells, hHiPS cells, and mesenchymal stem cell–derived induced pluripotent stem (iMSC) cells were able to differentiate into hepatic cells of all three stages (Figs. 3 and 4 and Supporting Fig. 5).

Figure 3.

Differentiation of hHiPS cells into Definitive Endoderm and hepatic progenitors: efficient endoderm induction of human iPS cells and ES cells in the presence of 100 ng/mL activin A. (A-E) FACS analysis showed that at least 90% of the induced cells expressed the definitive endoderm marker CXCR4: (A) H9, (B) H1, and (C-E) hHiPS6, hHiPS10, and hHiPS11 5 days after the initial activin A treatment. (F-K) Immunofluorescence analysis of the expression of AFP in differentiating cells 10 days after the initial activin A treatment: H1, H9, hHiPS6, hHiPS10, hHiPS11, and hHiPS14, respectively (×100; scale bar = 100 μm).

Figure 4.

Differentiation of hHiPS cells into mature hepatocytes. (A-C) The differentiated cells possessed proteins indicative of functional hepatocytes, including (A) ALB (green), (B) ALB (green) and AAT (red), and (C) CYP3A4 (pink), on day 20 after differentiation initiation (×200; scale bar = 100 μm). (D) Glycogen storage on day 20. A periodic acid-Schiff assay was performed on differentiating cells 20 days after the initial activin A treatment. Nuclei were counterstained with hematoxylin (blue). Glycogen storage was indicated by pink or dark red cytoplasm (×100; scale bar = 100 μm).

In the first stage, activin A efficiently induced the endoderm differentiation of hES and iPS cell lines. After 5 days of activin A treatment, approximately 90% of the cells in culture expressed the endoderm marker CXCR4 and lost the ES/iPS cell marker SSEA3 (Fig. 3A-E and Supporting Fig. 5A). To induce early hepatic cells or hepatic progenitor cells, HGF and FGF4 were used. We tested for the expression of AFP, a marker of early hepatic cells, in the differentiated hES cells and hHiPS cells. Approximately 90% of the cells were positively stained for AFP in both ES-derived and iPS-derived cells on day 10 (Fig. 3F and Supporting Fig. 5B). No obvious differences were observed between three types of pluripotent stem cells (i.e., hES, hHiPS, and iMSC cells).

At the end of the final differentiation stage, we tested the expression of hepatic markers in the differentiated ES and iPS cells. A majority of the cells on day 20 expressed mature hepatic markers, including ALB (Fig. 4A and Supporting Fig. 5C), AAT (Fig. 4B and Supporting Fig. 5C), and CYP3A4 (Fig. 4C). The efficiency and pattern of the iPS cell–derived hepatocytes were similar to those of hES cell derivatives.

We further tested the ability of the differentiating cell populations to store glycogen, another characteristic of functional hepatocytes.25 Cells on day 20 were stained for cytoplasmic glycogen with the periodic acid-Schiff staining procedure (Fig. 4D). Similarly to hES cell–derived cells, the majority of the differentiated hHiPS cells (up to 90%) were stained by periodic acid-Schiff (Fig. 4D), and this indicated that they had the capacity to store glycogen. Hence, in a manner consistent with the gain of expression of mature hepatic markers, the differentiating cells from hHiPS cells also exhibited a gain of hepatic functionality.

In order to further validate the functionality, multiple hHiPS cell line–derived hepatocytes were assessed for their metabolic capabilities (Fig. 5). The CYP450 enzymes are critical in drug metabolism, and CYP3A4 and CYP1A2 are the key enzymes.14 All three tested hHiPS cell lines (ihH6, ihH10, and ihH11) exhibited both CYP450 enzyme activities as assessed by the generation of luminescent metabolites (Fig. 5A,B). We observed only slight variations within these lines. These results suggest that human hepatocyte-derived iPS cells are able to differentiate into functional hepatocytes that can be used in drug testing and disease modeling.

Figure 5.

CYP450 metabolism in hHiPS cell-derived mature hepatocytes. Mature hHiPS cell-derived hepatocytes displayed CYP450 metabolism. iPS cell-derived hepatocytes were incubated with hepatocyte culture medium supplemented with CYP3A4 or CYP1A2 pGlo substrates (Promega) according to the manufacturer's instructions. After 4 hours of treatment, 50 μL of the culture medium was removed and read on a GloMax luminometer. The CYP1A2 and CYP3A4 activity is expressed as relative light units per milliliter of culture medium (n = 6). All three hHiPS cell lines (ihH6, ihH10, and ihH11) exhibited both CYP450 enzyme activities.


Here we demonstrate that human endoderm cells (i.e., hepatocytes) can be rapidly reprogrammed to pluripotency. Because iPS cells may retain certain gene expressions and/or epigenetic features of their original cell types1-5 and the origin of iPS cells has a profound influence on their tumor-forming propensity,3 it is important to generate and compare human iPS cells of different origins, including endoderm-derived cells, and systematically compare their safety and differentiation potentials. For the purpose of developing disease models from patient-specific iPS cells to study liver disease mechanisms, it is also crucial to generate hepatocyte-derived iPS cells because certain somatic mutations that play important roles in liver disease progression are acquired only in liver cells and therefore the liver disease features cannot be recapitulated by skin fibroblast-derived iPS cells derived from the same patients.16-20

Although human hepatocyte reprogramming shares the main features reported for iPS cell generation from other cell types such as fibroblasts, including the acquisition of self-renewal ability and pluripotency (Figs. 1 and 2), the overall pace appears to be faster than that of fibroblast, MSC, or blood cell reprogramming.9-13 Mouse hepatocytes also appear to be more easily reprogrammed than fibroblasts.7 It is conceivable that hepatocytes per se are more amenable to reprogramming, perhaps because, unlike fibroblasts or MSCs, they are not required to undergo a mesenchymal-to-epithelial transition to give rise to iPS cells. However, we cannot rule out the possibility that iPS colonies appear faster because, in the case of hepatocytes, no cell replating during the early stage of reprogramming is necessary as a result of their nonproliferative nature in culture, whereas proliferating MSCs or fibroblasts require culture splitting during early reprogramming (Supporting Fig. 3), and thus it takes longer for colonies to emerge.

It also remains to be determined whether the cells that undergo reprogramming are early or mature hepatocytes. Although our hepatocyte source is homogeneously ALB-positive (see Fig. 1B), we have also observed that 20% to 30% of these cells are also AFP-positive (Supporting Fig. 4). Therefore, it is possible that the reprogrammed cells are from only the AFP+ALB+ early hepatocytes. It is currently difficult to definitively determine the target cell type because this would require viable cell sorting of the primary hepatocytes and AFP and ALB are the cytoplasmic markers (not surface proteins for viable cell sorting). Moreover, primary hepatocytes would unlikely be healthy enough for reprogramming after the cell sorting and replating processes because primary hepatocytes, even without sorting stress, can survive only a week in culture.

We have also shown that hHiPS cells can be directly induced to differentiate into endoderm, hepatic progenitors, and mature hepatocytes with a stepwise differentiation method (Figs. 3 and 4). The hepatic differentiation efficiency of hHiPS cells is comparable to that of the hES cell lines. We have also observed similar hepatic differentiation from bone marrow (mesoderm)-derived iPS cells (i.e., iMSC cells) in terms of marker expression based on immunofluorescence staining (Supporting Fig. 5). We could not detect a significant enhancement in the efficiency of hHiPS cells in either spontaneous (Fig. 2) or directed differentiation processes (Figs. 3 and 4). This is not surprising because the differentiation efficiency is already very high in hES cells (more than 90%). A more stringent test of the differentiation ability would be in vivo functionality of the differentiated cells. Significant future studies should examine whether hepatocytes from hHiPS cells offer any advantages in in vivo assays over differentiated cells from hES cells or from iPS cells of other origins. Comprehensive studies comparing human iPS cells from all three developmentally distinct germ layers are needed to determine whether the cell origin for reprogramming has a critical influence on the functionality or safety of differentiated cells.

An important potential use of human iPS cell–derived hepatic cells is drug development. Most drugs rely on liver CYP450 activity for detoxification, which cannot be tested in animal liver cells because of species differences.26 In this study, we looked for the expression of CYP450 in the differentiated cells and found that the hHiPS cell–derived hepatic cells expressed Cyp3A4, as detected by immune staining (Fig. 4C), in addition to fully differentiated hepatic markers (ALB and AAT expression; Fig. 4A,B). The functionality was also confirmed by the glycogen storage activity (Fig. 4D) and CYP450 metabolic activity (Fig. 5). These data suggest that human iPS cell–derived hepatic cells may be used as a potential cell source for the generation of hepatocytes for drug metabolism analysis.

In conclusion, we have reprogrammed human endoderm-derived cells into iPS cells and shown that they can be directly differentiated into hepatic cells. Our reprogramming of primary hepatocytes to pluripotency should provide a valuable experimental model for investigating the basis of cellular reprogramming of other human endoderm cells. Equally important is the generation of liver disease-specific iPS cells for studying liver disease pathogenesis, including hepatocellular carcinoma and liver cirrhosis, with available liver tissues that can be obtained after partial hepatectomy or diagnostic liver biopsy. Human hepatocyte–derived iPS cells offer advantages for modeling certain liver diseases associated with acquired somatic mutations that are restricted to liver cells.16-20 The generation of endoderm–derived human iPS cells also facilitates comparative studies designed to determine the most suitable iPS cells in terms of safety and efficacy for treating particular diseases. We believe that these studies would lead to a better understanding of pluripotency and reprogramming and would eventually lead to safer and more potent cell therapy.