The homeobox gene Hex regulates hepatocyte differentiation from embryonic stem cell–derived endoderm


  • Potential conflict of interest: Dr. Snodgrass owns stock in VistaGen Therapeutics.


We investigated the role of the hematopoietically expressed homeobox (Hex) in the differentiation and development of hepatocytes within embryonic stem cell (ESC)–derived embryoid bodies (EBs). Analyses of hepatic endoderm derived from Hex−/− EBs revealed a dramatic reduction in the levels of albumin (Alb) and alpha-fetoprotein (Afp) expression. In contrast, stage-specific forced expression of Hex in EBs from wild-type ESCs led to the up-regulation of Alb and Afp expression and secretion of Alb and transferrin. These inductive effects were restricted to c-kit+ endoderm-enriched EB-derived populations, suggesting that Hex functions at the level of hepatic specification of endoderm in this model. Microarray analysis revealed that Hex regulated the expression of a broad spectrum of hepatocyte-related genes, including fibrinogens, apolipoproteins, and cytochromes. When added to the endoderm-induced EBs, bone morphogenetic protein 4 acted synergistically with Hex in the induction of expression of Alb, Afp, carbamoyl phosphate synthetase, transcription factor 1, and CCAAT/enhancer binding protein α. These findings indicate that Hex plays a pivotal role during induction of liver development from endoderm in this in vitro model and suggest that this strategy may provide important insight into the generation of functional hepatocytes from ESCs. (HEPATOLOGY 2010.)

In the mouse embryo, the liver is first detected as an outgrowth bud of proliferating endodermal cells in the ventral foregut on day 8 of gestation.1–3 The liver develops in close proximity to the cardiac mesoderm, which produces fibroblast growth factor 1 and 2, which in turn are required for the outgrowth of the ventral foregut endoderm2, 4 and the induction of several liver-specific genes, including albumin (Alb) and α-fetoprotein (Afp).5 In addition to fibroblast growth factors, bone morphogenetic protein 4 (BMP-4) expressed in the septum transversum mesenchyme6 has been shown to be essential for early liver development. In the absence of BMP-4, the foregut endoderm does not thicken, and consequently a distinct liver bud does not form. In spite of the lack of liver bud formation in BMP-4–null embryos, Alb expression is induced, suggesting that this factor may play a role in the proper movement of hepatoblasts into the developing liver. Beyond the induction stage, numerous other transcription factors are required for endoderm patterning and organ development. Among these, the hematopoietically expressed homeobox gene Hex (also known as Prh)7–9 is of particular interest, because it has been shown to play a pivotal role in hepatic development.

Hex is expressed at multiple sites in the developing embryo, including the yolk sac and the region of gut endoderm that gives rise to the liver and thyroid bud.10–12 Analysis of Hex-null embryos demonstrated that formation of the liver bud initiates in the absence of a functional protein and that expression of liver-specific genes including Alb, Afp, and Ttr is up-regulated in this endodermal population.13, 14 Whereas the early stages of morphogenesis to a columnar structure can be detected in these mutant embryos, development beyond day 9.5 of gestation does not proceed and expression of Alb is no longer evident,13–15 suggesting that Hex is required to promote growth and differentiation of the hepatoblast stage of development. Taken together, these studies demonstrate that Hex is essential for the development of liver from gut endoderm and that it functions downstream of the signaling pathways that regulate the specification of the hepatic lineage.

Defining the pathways and transcription factors that regulate lineage commitment in the early embryo is essential for our basic understanding of developmental biology as well as for establishing strategies for the directed lineage-specific differentiation of ESCs in culture. By translating findings from the embryo to this in vitro model, it has been possible to develop approaches for the efficient and reproducible induction of endoderm and early hepatic and pancreatic cell fates from both mouse and human ESCs.16–18 Although the above studies have established the principal signaling pathways regulating hepatic specification, none has investigated the role of the key transcription factors in this process. In the present report, we have used the ES/EB model to study the role of Hex in hepatocyte development in vitro and demonstrate that as in the early embryo, this transcription factor is essential for the establishment of hepatocyte lineage.


Afp, alpha-fetoprotein; Alb, albumin; BMP-4, bone morphogenetic protein 4; Cps1, carbamoyl phosphate synthetase; Dlk1, Delta-like 1; Dox, doxycycline; EB, embryoid body; ECD, E-cadherin; ESC, embryonic stem cell; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; Hex, hematopoietically expressed homeobox; mRNA, messenger RNA; RT-PCR, reverse-transcription polymerase chain reaction; Tcf1, transcription factor 1.

Materials and Methods

Growth and Differentiation of ESCs.

The development and characterization of Hex+/+, Hex+/−, and Hex−/− ESC lines,15 the GFP-Bry ESC line,19 and tet-Hex ESCs20 have been described. Bry-Ainv cells were generated by targeting green fluorescent protein to the brachyury locus in the Ainv 18 ESC line19, 21 (unpublished data). The Hex-plox targeting plasmid was electroporated into the Bry-Ainv cells, yielding tet-Hex Bry-Ainv ESCs. ESCs were maintained on irradiated mouse embryo fibroblast feeder cells as described.22

To assess the function of Hex in developmental progression of hepatocytes during ESC differentiation, tet-Hex ESCs, in which Hex expression can be induced by exposure to doxycycline (Dox) at specific time points, were cultured as previously described for ectoderm with some modification.20 For differentiation of endoderm, activin induction was performed using a two-step protocol as described.22 To induce Hex expression, Dox (1–30 μg/mL in Iscove's modified Dulbecco's medium with 15% SR and 2 mM glutamine) was added to the cultures at different stages and for varying periods of time. After a total of 10 days of differentiation, EBs were replated on Matrigel-coated 6-well dishes in Iscove's modified Dulbecco's medium supplemented with 15% fetal bovine serum (Vitromex, Geilenkirchen, Germany), 2 mM glutamine, and 10−6 M dexamethasone. Cells from these replated cultures were harvested at the indicated times (total differentiation time) for RNA isolation and immunostaining. Alternatively, serum-induced hepatic differentiation was performed as described.22

Gene Expression Analysis.

For reverse-transcription polymerase chain reaction (RT-PCR), total RNA was extracted using RNeasy mini-kits and then treated with RNase free DNase (Qiagen, Valencia, CA). One microgram of total RNA was then reverse-transcribed to complementary DNA using a Superscript RT kit (Invitrogen, Carlsbad, CA) with random hexamers. PCR was performed using Taq polymerase (Takara Bio, Shiga, Japan) in PCR buffer containing 2.5 mM MgCl2 and 0.2 μM dNTPs. Oligonucleotide primers were performed using the primer pairs shown in Supporting Table 1.

For real-time PCR, commercially available assay mixes for Alb, Afp, carbamoyl phosphate synthetase (Cps1), transcription factor 1 (Tcf1), Hex, BMP-4, Delta-like 1 (Dlk1), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were used to quantify messenger RNA (mRNA) levels, and PCR was performed using a Prism 7700 Sequence Detector (Applied Biosystems, Foster City, CA). mRNA levels were normalized to GAPDH mRNA levels in the same samples.

Fluorescence-Activated Cell Sorting Analysis and Cell Sorting.

EBs-derived cells were stained with a PE-conjugated anti-c-kit antibody (BD PharMingen, San Diego, CA), after which the cells were analyzed using a FACSan (Becton Dickenson, San Jose, CA) or sorted on a FACS Aria cell sorter (Becton Dickenson).


Foxa2 staining of brachyury+ and c-kit+ cells was performed in microtiter wells as described.22 Briefly, the cells were incubated with an anti-Foxa2 primary antibody (goat polyclonal P-19; Santa Cruz Biotechnololgy, Santa Cruz, CA) and visualized using Cy3-conjugated anti-goat secondary immunoglobulin G antibody (Jackson Immunoresearch, West Grove, PA). For Alb staining, day 14 EBs were scraped, embedded in Tissue Tek O.C.T. compound (Sakura, Torrance, CA), and frozen in liquid nitrogen, after which 4-μm-thick sections were cut on a cryostat and placed on polylysine-coated glass microscope slides. After the fixation and permeabilization, the cells were incubated for 1 hour with anti-Alb primary antibody (Biogenesis, Kingston, NH) and visualized using a Cy3-conjugated anti-rabbit immunoglobulin G secondary antibody (Jackson Immunoresearch).

Measurement of Alb and Transferrin Secretion from Day 14 EBs.

After culturing EBs for 14 days under various conditions, the medium was changed to serum-free Iscove's modified Dulbecco's medium containing 2 mM glutamine. The EBs were then incubated for an additional 24 hours, and the conditioned medium was collected for assay. Alb and transferrin concentrations in the conditioned medium were measured using solid-phase sandwich enzyme-linked immunosorbent assays (Bethyl, Montgomery, TX) according the manufacturer's instructions.

Microarray Analysis.

For microarray analysis, total RNA was extracted using RNeasy mini kits (Qiagen), after which 10 μg of fragmented target total RNA was used for hybridization of each UniSet Mouse I Expression Bioarray chip (Amersham Life Sciences, Buckinghamshire, UK), which contained 10,012 probes. Once the microarrays were hybridized and washed, biotin-containing transcripts were directly detected using a Streptavidin-Alexa647 conjugate. GeneSpring 6.2 (Silicon Genetics, Inc., Redwood City, CA) was then used to evaluate the data obtained using CodeLink Expression Scanning Software.

Target Preparation and CodeLink™ Microarray Analysis.

Total RNA was isolated from cell samples using the RNeasy mini kit (Qiagen). One microgram of purified, DNase-treated total RNA was used to synthesize biotinylated cRNA target preparation using the CodeLink™ Expression Assay Reagent Kit (Amersham) according to manufacturer's instructions. Ten micrograms of fragmented cRNA was applied to the Uniset Mouse I Expression Bioarray, hybridized for 18 hours, and the arrays processed according to manufacturer's instructions using Streptavidin-Alexa647 detection reagents. Slides were scanned using a GenePix 4000B scanner (Axon) and the images were analyzed with CodeLink™ Expression Analysis Software.


Hex Regulates Expression of Hepatocyte-Related Genes and Secretion of Their Products.

To investigate the role played by Hex in hepatocyte lineage commitment during EB differentiation, we induced endoderm formation from wild-type (Hex+/+), heterozygous (Hex+/−), or Hex-deficient (Hex−/−) ESCs15 using the serum induction protocol that we have described.22 Consistent with our earlier findings, the hepatocyte genes Alb and Afp were both expressed in EBs generated from the Hex+/+ and Hex+/− ESCs. The levels of expression of both genes were markedly reduced in the Hex−/− EBs (Fig. 1A). These findings are in line with those from studies on the early embryo, demonstrating that Hex is required for development of the liver in vivo.13, 15

Figure 1.

Hepatocyte differentiation in the presence or absence of Hex during endoderm induction. (A) Hepatic differentiation of Hex-deficient ESCs. RT-PCR analysis was conducted for EBs generated from Hex+/+ (+/+), Hex+/− (+/−), and Hex−/− (−/−) ESCs. Cells were cultured in the serum-induced protocol for hepatocytes for 14 days. (B,C) Time course of Alb (B) and Afp (C) mRNA expression during hepatic EB differentiation. Differentiation was induced by culture in the following protocols: (1) wild-type ESCs were cultured in serum induction protocol for hepatocytes (□); and (2) tet-Hex ESCs were cultured in the presence of activin for days 2–6 and then cultured without Dox (●) or with Dox on days 6–10 (○) or on days 6–22 (▪). Alb and Afp mRNA levels were quantified by way of real-time PCR and normalized to GAPDH mRNA levels. Alb and Afp mRNA levels are indicated as a ratio compared with those in day 14 fetal liver. (D) Immunostaining of Alb in day 14 EBs cultured with activin with Dox (days 6–10).

To further evaluate the role of Hex in hepatic specification in the ESC/EB model, we used an ESC line (AINV18) that enables the regulated expression of a given gene under the control of a tet-inducible promoter. Using this system, we generated ESCs in which Hex expression was induced by the addition of the tetracycline analogue Dox (tet-Hex ESCs). Hex expression was induced in the cells by the addition of Dox (1 μg/mL) to the EB cultures either from days 6–10 or from days 6–22 of differentiation. Quantitative PCR analyses revealed that induction of Hex between days 6 and 10 of culture resulted in a significant up-regulation of Afp and Alb expression compared with the uninduced cultures (Fig. 1B,C). These levels of expression at day 14 represent 2.6% and 2.5% of the expression found in the fetal liver, respectively. By contrast, when Hex was continuously expressed from day 6 to day 22, levels of Alb and Afp mRNA were diminished on day 22 (Fig. 1B,C), suggesting that prolonged Hex expression may disrupt hepatic differentiation or shift the tissue into another fate. Immunostaining showed the presence of clusters of Alb+ cells within the Hex-induced day 14 EBs (Fig. 1D). Hex induction also resulted in enhanced secretion of both Alb and transferrin by the EB-derived cells at day 14 of culture (Fig. 2A,B). These levels of secretion were 7.6% and 5.0% of that of day 14 fetal liver cells, respectively.23 Taken together, these findings indicate that enforced expression of Hex at appropriate stages of differentiation and for a specific period enhances the specification of hepatocyte-like cells from definitive endoderm.

Figure 2.

Secretion of albumin and transferrin from hepatic EBs. (A,B) Differentiation was induced by culture in the following protocols: (1) wild-type ESCs were cultured in the serum induction protocol for hepatocytes (serum); and (2) tet-Hex ESCs were cultured in the presence of activin for days 2–6 and then cultured without Dox (Dox−) or with Dox on days 6–10 (Dox+). At day 14, EBs were cultured in serum-free media for an additional 24 hours. At this point, the media was harvested and the levels of Alb (A) and transferrin (B) were measured using a specific enzyme-linked immunosorbent assay. (C,D) Secretion of Alb (C) and transferrin (D) from activin-induced EBs cultured for 14 days. Cells were treated with Dox at the indicated concentrations for the indicated durations. At day 14, EBs were cultured in serum-free media for an additional 24 hours. At this point, the media was harvested and the levels of Alb and transferrin were measured using specific enzyme-linked immunosorbent assays.

Microarray Analysis of Genes Downstream of Hex.

For a more in-depth analysis of the impact of Hex expression on lineage development, we performed a microarray analysis to identify genes activated downstream of Hex. For these studies, we compared the following populations: (1) cells from day 14 hepatocyte cultures to cells from day 6 EB cells that were induced to form mesoderm by a continuous exposure to serum; (2) cells from day 14 Hex+/+ hepatoycte cultures to cells from day 14 Hex−/− hepatocyte cultures; and (3) and cells from 14-day hepatocyte cultures derived from Hex-induced EBs to a comparable population derived from noninduced EBs. For the third analyses, endoderm was induced with activin and Hex was induced by the addition of Dox from days 6 to 10 of differentiation.

The findings from these different analyses are summarized in Supporting Table 2. These microarray data are provided online. Of the 10,012 genes analyzed, the outcome of the first comparison revealed that the expression levels of 1,155 were significantly up-regulated in endoderm/hepatocyte conditions, in both mesodermal EBs and undifferentiated ESCs (P < 0.05) (comparison). Of these 1,155 genes, 240 were expressed at significantly higher levels (P < 0.05) in the day 14 Hex+/+ hepatocyte cultures compared with the day 14 Hex−/− hepatocyte cultures (comparison 2). Thirty-four of the 240 genes were also up-regulated following Dox induction (comparison 3, tet-Hex Dox(+)/Dox (−) (Supporting Table 2). The genes shown to be regulated by Hex expression could be categorized into five functional groups: (1) serum protein genes such as alb1; (2) coagulation-related genes such as fibrinogens; (3) lipid-related genes such as apolipoproteins; (4) growth factor related genes such as insulin-like growth factor binding protein; and (5) others. The fact that most of these proteins are produced in the liver adds further support to the interpretation that forced expression of Hex at appropriate stages of development efficiently induces liver specification and maturation from the ESC-derived endoderm.

Treatment with a High Concentration of Dox at the Appropriate Time Induces Maturation of Hepatocytes During ESC Differentiation.

To define more precisely the time frame during which Hex exerts this effect, Dox was added from days 2–6, days 6–10, or days 10–14. Forced expression of Hex could induce Alb mRNA only when Dox was added between days 6–10, but not when added earlier (days 2–6) or later (days 10–14) than these times (Fig. 3A). Next, Dox was added to the EB cultures for a 24-hour period between days 5 and 9 of differentiation. Alb expression was measured at day 14 of culture using real-time PCR. As shown in Fig. 3A, Alb message was only increased in the population in which Hex was induced at day 6 for 24 hours. Induction at earlier or later time points had little effect on Alb expression. These findings suggest that Hex expression during this narrow stage of differentiation is critical for hepatic specification of the ESC-derived endoderm.

Figure 3.

Dox induction of hepatocyte differentiation is stage-specific and concentration-dependent. (A) Dox was applied at the indicated time points and duration and Alb mRNA levels were measured on day 14. (B,C) Dox was added at the indicated concentrations on days 6–7 (black bars) or from day 6 to day 10 (white bars). Levels of Alb (B) and Afp (C) mRNA were measured on day 14. Alb and Afp mRNA levels were quantified by way of real-time PCR and normalized to GAPDH mRNA levels. Alb and Afp mRNA levels are indicated as the ratio compared with those in day 14 fetal liver (A-C). FL, day 14 fetal liver. (D) RT-PCR analysis of genes associated with liver maturation. Dox was added at indicated concentrations on days 6-7 and expression patterns were analyzed on day 14. AL, adult liver; FL, day 14 fetal liver.

When next evaluated the effects of Dox concentration and the duration of induction on the levels of Alb expression. At concentrations of 1 and 10 μg/mL, Dox induced dose-dependent expression of Alb (Fig. 3B) and Afp (Fig. 3C), regardless of the duration of Dox exposure. Interestingly, exposure to a higher concentration of Dox (30 μg/mL) for 1 day increased Alb and Afp expression dramatically, to levels of 60% and 62% of those found in day 14 fetal liver (Fig. 3B,C). With the high concentrations of Dox, however, we observed a significant decrease in size (up to 75%) of the resulting EBs. The up-regulated expression of Alb and Afp by Hex was dependent on prior induction of endoderm by activin, as no expression was detected serum-induced EB that contained mesoderm and little, if any endoderm (data not shown). A longer exposure (4 days) to 30 μg/mL of Dox disrupted EB differentiation and suppressed expression of Alb and Afp mRNA.

In addition to Alb and Afp, other genes involved in hepatocyte maturation and function, including tyrosine aminotransferase, Cps1, fibrinogen β, apolipoprotin A2 (Apo A2), Apo C2, cytochrome P450 (Cyp3a11 and Cyp7a1) were also induced in the population generated from EBs treated with high concentrations (30 μg/mL) of Dox on day 6 (Fig. 3D). Consistent with the expression data, we observed that secretion of Alb and transferrin increased with increasing concentrations of Dox, reaching levels of 12.6 and 2.9 μg/mg protein/24 hours, respectively, following induction by 30 μg/mL Dox (Fig. 2C,D). These levels of secretion were 84% and 48% of that of day 14 fetal liver cells, respectively.23

Hex Induces Hepatocyte-Related Genes in the c-kit+ Endodermal Cells.

The above studies indicate that Hex induces a hepatic fate in the context of the whole EB population. We have previously demonstrated that endoderm segregates to the c-kithigh/CXCR4+ brachyury-positive (GFP-Bry+) population of activin-induced EBs.18 The studies of Tada et al.24 have shown that activin induced endoderm also express E-cadherin (ECD). As shown in Fig. 4A, activin induced a GFP-Bry+/c-kithigh population in a dose-dependent fashion over a 6-day period. Analyses of ECD expression indicated that the majority of the c-kit+ population induced with high concentrations of activin also expressed ECD (Fig. 4B). Molecular analyses revealed that the c-kit+ population isolated from day 6 activin-induced EBs expressed genes indicative of endoderm induction, including Foxa2, Sox17, Cereberus, and those associated with hepatic (Hex) and pancreatic (Ipf1) specification (Fig. 4C). Immunocytochemical analysis showed that the incidence of Foxa2+ cells within the c-kithigh population was much higher than within the GFP-Bry+/c-kitlow population (Fig. 4D). Together, these analyses confirm that the c-kithigh population is enriched for endoderm. We next sorted the GFP-Bry+/c-kithigh and GFP-Bry+/c-kitlow fractions from day 6 activin-induced EBs and cultured the cells at high density in the presence of Dox (1 μg/mL) for 4 days. Under these conditions the cells will reaggregate and form EB-like structures that support the continued differentiation of the respective populations. Alb message was only induced in the endoderm-enriched c-kithigh population, clearly demonstrating that Hex functioned to specify a hepatic fate directly in definitive endoderm (Fig. 4E).

Figure 4.

Endoderm potential of c-kithigh cells and Hex-induced hepatocyte differentiation in activin-induced endoderm. Tet-Hex Bry-Ainv cells were cultured for 2 days in SP34, after which EBs were transferred to SR medium containing various concentrations of activin. (A) Dose-dependent induction of c-kithigh cells on day 6. (B) Fluorescence-activated cell sorting profile for c-kit and ECD in day 6 EBs stimulated by activin (100 ng/mL). (C) RT-PCR analyses of endoderm-related genes in the presorted (pre), GFP-Bry+ c-kitlow, and GFP-Bry+c-kithigh day 6 EB populations. (D) Immunostaining of Foxa2 in GFP-Bry+/c-kitlow GFP-Bry+/c-kithigh cells from day 6 activin-induced EBs. (E) Expression of Alb mRNA in EBs on day 14 from the different day 6 EB-derived populations. Cells from the pre-sorted or sorted populations were reaggregated with or without Dox (1 μg/mL) from day 6 to 10, cultured in SR medium, and replated on day 10. Levels of Alb mRNA normalized to those of GAPDH mRNA in day 14 EBs are shown. Alb mRNA levels were indicated as the ratio compared with those in day 14 fetal liver.

Hex and BMP-4 Act in Concert to Induce Alb Expression During EB Differentiation.

We have previously demonstrated that BMP-4 signaling induces a hepatocyte fate in activin-induced endoderm during EB differentiation.18 To further examine the relationship between BMP-4 and Hex, day 6 activin-induced EBs were exposed to BMP-4, to Dox (1 μg/mL) or to both BMP-4 and Dox (1 μg/mL) from days 6 to 10. As previously shown, BMP-4 did induce Alb and Afp mRNA (Fig. 5A). The levels of Alb and Afp mRNA detected in day 14 hepatocyte cultures reached 19.7% and 25.8% of those found in day 14 fetal liver, respectively, and were much higher than those induced by 1 μg/mL of Dox (days 6–10). The addition of basic fibroblast growth factor, hepatocyte growth factor, and vascular endothelial growth factor had no effect on Alb and Afp expression (data not shown). Interestingly, the combination of BMP-4 and Dox further increased Alb and Afp mRNA levels to 40.3% and 43.3% of those found in day 14 fetal liver, respectively (Fig. 5A). Expression of Cps1, a gene that encodes carbamoyl-phosphate synthetase 1 expressed in mature hepatocytes, was also synergistically induced in the presence of BMP-4 and Dox on day 14 (Fig. 5B).

Figure 5.

BMP-4 and Hex act synergistically to enhance expression of liver-related genes during EB differentiation. (A-E) Tet-Hex ESCs induced by activin between day 2 and day 6 of differentiation were stimulated with BMP-4 (50 ng/mL) and/or Hex (Dox; 1 μg/mL, days 6–10) as indicated. (A) Expression of Alb (black bars), Afp (white bars), and (B) Cps1 were evaluated on day 14. (C) Expression of Afp (black bars), Cebpa (white bars), and Tcf1 (striped bars) were evaluated on day 10. (D) CYP7a1 and TAT were evaluated on day 10 and day 14 by way of RT-PCR. (E) Expression of Hex was evaluated on day 10. (F,G) Hex+/+ (+/+) or Hex−/− (−/−) ESCs were cultured in the absence (black bars) or the presence (white bars) of BMP-4 from day 6 to day 10. Levels of Alb (F) and Tcf1 (G) mRNA were measured on day 14 and day 10, respectively. Gene expression levels were quantified using real-time PCR and normalized to those of GAPDH mRNA. Levels of Alb, Afp, Tcf1, and Cebpa mRNA were indicated as the ratio compared with those in day 14 fetal liver (A,D,F,G). Cps1 mRNA levels (B) were indicated as the ratio compared with those in adult liver, because fetal liver does not express this gene.

To gain further insight into the onset of hepatic development in these cultures, we evaluated the expression of Tcf1 and Cebpa, as these transcription factors are known to play a pivotal role in the establishment of the early liver by directly regulating expression of a variety of genes, including albumin, transferrin, and fibrinogen.25 Both BMP-4 and Dox (Hex) induced the expression of Tcf1 and Cebpa on day 10 of culture. As observed with the previous set of genes, the combination of BMP-4 and Dox resulted in a synergistic induction of expression of both (Fig. 5C), although the effect on Tcf1 was significantly greater than that observed on Cebpa. Taken together, these results suggest that BMP-4 and Hex function in a synergistic fashion to establish the liver fate, as defined by the up-regulation of expression of Tcf1, Cebpa, Alb, and Afp.

In contrast to the above set of genes, neither BMP-4 nor Hex alone induced expression of CYP7a1 or TAT, two genes indicative of hepatic maturation, at day 10 of differentiation (Fig. 5D). The combination of BMP-4 signaling and Hex expression did result in low levels of CYP7a1 and TAT expression at this time. By day 14, Hex but not BMP-4 induced CYP7a1 and TAT expression. These findings indicate that maintenance of appropriate levels of Hex is essential for maturation of the hepatic lineage in culture.

The effect of BMP-4 does not appear to be mediated directly through Hex, because Hex expression was not up-regulated following BMP-4 treatment of activin-induced endoderm (Fig. 5E). Although BMP signaling did not induce Hex, a functional Hex gene is required for establishment of the liver fate, because BMP-4 was unable to induce Alb expression in Hex−/− endoderm (Fig. 5F). In contrast, BMP-4 did induce Tcf1 expression in the absence of Hex, although the levels were not as high as those observed in the wild-type population (Fig. 5G). Findings from these analyses suggest that Hex and BMP-4 can regulate Tcf1 expression independently, but optimal levels of expression require both pathways. We also evaluate if Hex can induce BMP-4 mRNA levels. However, Hex did not affect the gene expression levels of BMP-4 (data not shown).

To determine if BMP-4 and Hex have an impact on the hepatoblast stage of development, we analyzed the different populations for expression of Dlk1. Tanimizu et al.26 have shown that Dlk1 is expressed on progenitors with hepatoblast potential as fetal liver cells sorted for this marker displayed both hepatocyte and biliary epithelial potential. Dlk1 message was detected in day 10 EBs cultured in the absence of BMP-4 and Dox induction. Addition of BMP-4, but not the induction of Hex, increased the levels of Dlk1 expression. The relatively high levels of Dlk1 observed in the absence of BMP-4 appear to be a result of activin signaling, because substantially lower levels were detected in cells differentiated in the absence of activin. Induction with BMP-4 doubled the expression levels of Dlk1 in either the absence of presence of activin. Finally, neither factor induced significant levels of Dlk1 in the absence of a functional Hex gene.


The directed differentiation of ESCs in culture is emerging as a powerful model system for studying mammalian development in vitro as well as a renewable source of functional cell types for transplantation for future cell-based therapy and for drug discovery and toxicology testing.27 Of the different cell populations that can be generated from these pluripotent stem cells, hepatocytes are of particular interest because hepatocyte-based therapy has been considered as a new generation and effective treatment mode for liver diseases28 and the liver is a primary target organ of drug toxicity.29 A number of reports have documented the efficient generation of immature hepatocytes from both mouse and human ESCs,16–18 demonstrating that specification of this lineage can be studied in this model system. The most successful approaches to date have translated developmental biology to the culture dish and recreated the key aspects of the normal hepatic developmental program in the differentiation cultures. Although these studies collectively show that it is possible to generate populations with hepatic characteristics, the cells that do develop in the cultures remain immature. Complimentary approaches to cytokine-induced differentiation for the generation of functional cell populations include gain- and loss-of-function of key regulatory genes. Using a gain-of-function strategy in this study, we have shown that enforced expression of Hex at specific developmental step promotes the differentiation of the hepatic lineage from mouse ESCs.

In the mouse embryo, Hex is required for establishment of the fetal liver from the liver bud,13–15 positioning it at the level of lineage progression and maturation rather than at the earliest specification step. The requirement for Hex in progression of the hepatic lineage was also observed in the ESC differentiation cultures, as the Hex−/− ESC-derived endoderm population expressed significantly lower levels of Alb compared to wild-type cells. This observation adds to the growing body of evidence indicating that lineage specification in the in vitro differentiation model recapitulates that observed in the early embryo.30 In contrast to the loss of function, enforced expression of Hex dramatically enhanced hepatic differentiation, suggesting that this gene plays a pivotal regulatory role in the progression of the hepatic lineage in culture. The effects of Hex expression are striking, because the induced population expressed genes indicative of hepatic maturation (CYP7A1 and TAT) and secreted levels of Alb and transferrin comparable to those secreted by primary rat hepatocytes in culture. These findings highlight the importance of maintaining appropriate levels of Hex expression in promoting maturation of the hepatic lineage in ESC differentiation cultures. Enforced expression of key transcription factors to promote differentiation from ESCs is not a new approach, because it has been effectively used to generate skeletal myocytes31 and hematopoietic cells with stem cell characteristics.21 All three studies used the same inducible system, enabling the regulated expression of the gene of interest. While gain-of-function strategies may not be appropriate for the generation of functional cell types for clinical use, the findings from such studies do provide important insights into the key transcriptional pathways that regulate lineage development.

Studies performed initially in the mouse embryo6 and subsequently in the ESC model18 have shown that BMP-4 is required for hepatic specification of definitive endoderm. Although the precise relationship of BMP-4 and Hex are not known, the findings from our study are consistent with a model in which both BMP-4 and Hex are required for specification of the hepatoblast from activin-induced definitive endoderm (Fig. 6C). Once the hepatoblast is specified, these factors appear to regulate independent sets of genes that interact at some point to promote hepatic differentiation as suggested by the following observations. First, BMP-4 signaling does not induce Hex expression, indicating that the BMP pathway does not directly regulate this transcription factor. Second, enforced expression of Hex and BMP-4 display synergistic effects in the up-regulation of Alb and Afp expression. Third, BMP-4 can induce HNF1/Tcf1, a key regulator of Ttr, Alb, Afp, fibrinogen α, fibrinogen β, and ApoA2 expression in the absence of Hex. HNF1/Tcf1 may represent one point of convergence within this regulatory network as the levels of expression induced by BMP-4 are higher in wild-type cells than in the Hex−/− endoderm and dramatically enhanced by the enforced expression of Hex. In addition to playing a regulatory role at the level of Tcf1 expression, evidence exists that Hex can physically interact with Tcf132 suggesting that these transcription factors may function as coactivators of downstream targets.

Figure 6.

Regulation of Dlk1 expression by activin, Hex, and BMP-4. (A) Expression of Dlk1 in populations induced with BMP-4, Hex, or both. Tet-Hex ESCs induced by activin between day 2 and day 6 of differentiation were stimulated with BMP-4 (50 ng/mL) and/or Hex (Dox; 1 μg/mL, days 6–10) as indicated. Expression of Dlk1 was evaluated on day 10. (B) Effect of activin and BMP-4 on Dlk1 expression in wild-type (+/+) and Hex null (−/−) ESCs. Hex+/+ (+/+) or Hex−/− (−/−) ESCs were cultured in the absence (0 ng/mL) or presence (100 ng/mL) of activin from day 2 to day 6. They were subsequently cultured in the absence (black bars) or presence (white bars) of BMP-4 from day 6 to day 10. Levels of Dlk1 mRNA were measured on day 10. Gene expression levels were quantified using real-time PCR and normalized to those of GAPDH mRNA. AL, adult liver; FL, day 14 fetal liver. (C) Model depicting the role of BMP-4 and Hex in the specification and maturation of the hepatic lineage from ESCs.

In conclusion, the findings presented here document a pivotal role for Hex in the establishment of the hepatic lineage in ESC cultures and in doing so provide further evidence that lineage commitment in this model accurately reflects that observed in the early embryo. Stage-specific enforced expression of Hex promoted hepatic development and maturation, indicating that this strategy may provide an efficient method for the production of relatively mature cell types for studies on lineage commitment, for transplantation in preclinical model of liver disease and for drug discovery and analyses.


We thank Mako Yabunouchi and Fumie Otsuka for their excellent technical assistance.