The canonical Wnt/β-catenin signaling has remarkably diverse roles in embryonic development, stem cell self-renewal and cancer progression. Here, we show that stabilized expression of β-catenin perturbed human embryonic stem (hES)-cell self-renewal, such that up to 80% of the hES cells developed into the primitive streak (PS)/mesoderm progenitors, reminiscent of early mammalian embryogenesis. The formation of the PS/mesoderm progenitors essentially depended on the cooperative action of β-catenin together with Activin/Nodal and BMP signaling pathways. Intriguingly, blockade of BMP signaling completely abolished mesoderm generation, and induced a cell fate change towards the anterior PS progenitors. The PI3-kinase/Akt, but not MAPK,signaling pathway had a crucial role in the anterior PS specification, at least in part, by enhancing β-catenin stability. In addition,Activin/Nodal and Wnt/β-catenin signaling synergistically induced the generation and specification of the anterior PS/endoderm. Taken together, our findings clearly demonstrate that the orchestrated balance of Activin/Nodal and BMP signaling defines the cell fate of the nascent PS induced by canonical Wnt/β-catenin signaling in hES cells.
During early embryogenesis, gastrulation results in the formation of three definitive germ layers and establishment of the embryonic body plan(Tam and Loebel, 2007). At this stage, undifferentiated epiblast cells undergo an epithelial to mesenchymal transition (EMT) and migrate through a structure called the primitive streak (PS) to generate the mesoderm and endoderm. Distinct regions of the PS induce different subpopulation of mesoderm and endoderm progenitors(Tam and Loebel, 2007). The anterior PS has the potential to form the definitive endoderm and anterior mesoderm, including hepatic endoderm and cardiac mesoderm. The middle region of PS forms the lateral plate mesoderm; and the most posterior region of PS develops extra-embryonic mesoderm progenitors, which give rise to hematopoietic and vascular cells of blood islands. Mesoderm and endoderm generation depends on successful completion of the EMT and migration away from the PS. The EMT program is the process by which polarized epithelial cells are converted into individually motile cells; it occurs not only during early embryogenesis, but also during tumor progression(Thiery, 2002). The important markers of EMT lose epithelial polarities and adherence junctions following the downregulation of E-cadherin, which is an important cell-adhesion molecule of the apical-basal polarity and intercellular adhesion. Although several signaling pathways involved in the EMT processes have been identified(Thiery, 2002), elucidation of the molecular mechanisms that trigger and promote EMT is important for a better understanding of embryogenesis and tumorigenesis.
The canonical Wnt signaling functions are well established in fundamental biological processes (Moon et al.,2004; Tam and Loebel,2007). In early embryogenesis, Wnt/β-catenin signaling has pivotal roles in the formation of the PS, mesoderm and endoderm(Lickert et al., 2002; Tam and Loebel, 2007), and the stabilization of β-catenin leads to premature EMT in mouse embryo(Kemler et al., 2004). Wnt binds to its receptor Frizzled and co-receptor Lrp5/6, and increases the level of cytoplasmic and nuclear β-catenin, followed by inhibition of the GSK3-mediated degradation pathway. Upon inhibition of GSK3 activity,stabilized β-catenin translocates into the nucleus, where it serves as a co-activator of the Lef/Tcf family of DNA-binding proteins to form active transcriptional complexes for specific target genes(Moon et al., 2004). There is accumulating evidence that the Wnt/β-catenin signaling pathway also has an important role in stem cell maintenance and regulation of the cell fate decision in several stem cell systems, including hematopoietic, epidermal and intestinal stem cells (Moon et al.,2004).
Embryonic stem (ES) cells possess the remarkable property of indefinite self-renewal and pluripotency, the ability to differentiate to all cell types of an organism; they also provide an excellent model system for studying cell fate determination in early mammalian development(Keller, 2005). Multiple signaling pathways, such as those involving growth factors, transcriptional regulators and epigenetic modifiers, have crucial roles in regulating the balance between ES-cell self-renewal and lineage commitment(Keller, 2005). It is very important to elucidate molecular mechanisms of the self-renewal and lineage commitment for efficient production of functional cells required for use of hES cells in transplantation therapy and drug discovery. Like other stem cell systems, canonical Wnt/β-catenin signaling is implicated in mouse ES(mES) cell self-renewal and differentiation, depending on the context(Gadue et al., 2006; Hao et al., 2006; Lindsley et al., 2006), but precise roles of this signaling in human ES (hES) cells remains controversial(Dravid et al., 2005; Sato et al., 2004).
We report here that the activation of canonical Wnt/β-catenin signaling in hES cells by conditional activation of stabilized β-catenin disrupted hES-cell self-renewal. Rather, the canonical Wnt/β-catenin and BMP signaling pathway in hES cells has significant roles in establishing the posterior PS/mesoderm progenitors, whereas attenuation of BMP signaling changes the cell fate to the anterior PS/endoderm progenitors. In addition,Activin and Wnt/β-catenin signaling pathways synergistically function in inducing undifferentiated hES cells to differentiate into the anterior PS/endoderm progenitors. This is the first in vitro model system that consistently recapitulates the human early embryogenesis and that enables us to analyze molecular events during the process of early embryogenesis from the epiblast to the PS formation, followed by lineage specification into the mesoderm and endoderm. More importantly, our findings will also be relevant to directed differentiation of specific tissue and cells from hES cells.
MATERIALS AND METHODS
Activation of β-catenin signaling in hES cells
The ΔNβ-cateninER construct, in which the N-terminal 90 amino acids were deleted, was generated by in-frame insertion into the expression vector containing the hormone-binding domain of a mutant estrogen receptor(Littlewood et al., 1995; Sumi et al., 2007). Cell lines expressing ΔNβ-cateninER were obtained by transfection of hES cell lines KhES-1 and KhES-3 with ΔNβ-cateninER expression plasmids,followed by puromysin selection as described previously(Sumi et al., 2007). To activate ΔNβ-cateninER, hES cells stably expressingΔNβ-cateninER were treated with 4OHT (100 nM), as indicated.
Maintenance and differentiation of hES cells
The hES cell line HES-3 was purchased from ES Cell International. The hES cell lines KhES-1, KhES-3 and HES-3 were maintained as described previously(Suemori et al., 2006). For differentiation of hES cells, the cells were dissociated into small clumps and cultured on plates coated with matrigel (BD Biosciences, San Jose, CA) in DMEM/F12 with N2 and B27 supplements (Invitrogen, Carlsbad, CA). The next day,the medium was changed to N2B27 medium with or without 4OHT (100 nM, Sigma, St Louis, MO) in the presence or absence of 250 ng/ml Noggin-Fc chimera (R&D Systems Minneapolis, MN) except as otherwise indicated, and renewed daily thereafter. For Activin-induced differentiation, hES cells were cultured in RPMI supplemented with 2% FBS in the presence or absence of 100 ng/ml Activin A (R&D Systems), 100 ng/ml DKK1 (R&D Systems) or 100 nM 4OHT for 3 days, and then analyzed as described below. SB431542 (Sigma), U0126 and LY294002 (Promega, Madison, WI) were used at a final concentration of 10μM. GSK-3 inhibitor-IX and -X (BIO and BIO-Acetoxime, Calbiochem, La Jolla,CA) were used at a final concentration of 2 to 10 μM. For endothelial cell differentiation, cells cultured with 4OHT for 3 days were trypsinized and plated on collagen I-coated dishes, and then cultured in a StemPro34 serum free medium (Invitrogen) with 20 ng/ml VEGF165 (R&D Systems) for an additional 6 days. For endoderm and cardiac differentiation,β-catenin-activated cells cultured with or without Noggin (250 ng/ml) for 4 days were trypsinized and plated on collagen I-coated dishes, and then cultured in a N2B27 medium without 4OHT in the presence of 10 ng/ml BMP4(R&D Systems) and 5 ng/ml FGF2 for an additional 4 days. All data shown are representative results obtained from at least two independent clones of two different human ES cell lines.
Quantitative and semi-quantitative PCR analysis
Total RNA was isolated from cells using RNeasy Micro Kit (QIAGEN, Valencia,VA) and reverse-transcribed using Omniscript RT Kit (QIAGEN) according to the manufacturer's protocol. For semi-quantitative PCR, PCR reactions were optimized to allow for semi-quantitative comparisons within the log phase of amplification. Real-time RT-PCR analysis was performed on an ABI Prism 7500 Real-time PCR system using the PowerSYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). The expression value of each gene was normalized against the amount of GAPDH and calculated by the ΔΔCt method. The expression level of each gene in the control sample (vehicle or undifferentiated ES cells) was defined as 1.0. The normalized expression values for all control and treated samples were averaged, and average fold-change was determined. Details of the primers used for the semi-quantitative and quantitative PCR can be provided on request.
Western blot and immunofluorescence analysis
Cell lysates were prepared and subjected to sodium dodecyl sulfate-polyacrylamide gel elctrophoresis (SDS-PAGE), followed by western blotting as described previously (Sumi et al., 2007). For immunofluorescence analysis, cells plated on the culture slides were fixed with 3.7% formaldehyde and permeabilized with 0.25%Triton X-100. After blocking, the cells were incubated with primary antibodies, followed by incubation with secondary antibodies. Alexa Fluor 488-or 555-conjugated secondary antibodies were purchased from Invitrogen. Cells were mounted onto glass slides with Vectashield (Vector Laboratories,Burlingame, CA), and then analyzed using a BX61 fluorescence microscope(Olympus, Center Valley, PA). Antibodies against the following proteins were used: VEGF R2/KDR (clone#89115), SOX17 and CXCR4 (clone#44717) (R&D Systems); E-cadherin, Smad2/3, GSK3β and β-catenin (BD Biosciences);N-cadherin (GC-4, Sigma); ZO-1 (Invitrogen); Nanog (ReproCELL); HNF3β,Oct3/4, Brachyury and ERα (Santa Cruz Biotechnology, Santa Cruz, CA);SSEA3 and SSEA4 (Developmental Hybridoma Bank); TRA-1-60 and TRA-1-81(Chemicon); Phospho-Smad1 (Ser463/465), Smad1, phospho-Smad2 (Ser465/467),phospho-MAPK (Thr202/Tyr204), MAPK, phospho-Akt (Ser473), Akt and phospho-GSK3α/β (Ser21/9) (Cell Signaling, Danvers, MA).
Single cell suspensions were fixed with 1% formaldehyde for 30 minutes at 4°C, and incubated first with primary antibody and then with Alexa Fluor 488-, 555-conjugated secondary antibody (Invitrogen). FACS analysis was performed using a FACSCalibur Flow Cytometer (Becton Dickinson).
Karyotype analysis of hES cells
Chromosome spreads were prepared as described elsewhere(Suemori et al., 2006). Briefly, hES cells were incubated in ES medium with KaryoMAX Colcemid Solution(Invitrogen; 0.1 μg/ml of colcemid) for 2 hours, trypsinized, incubated in 0.075 M KCl for 10 minutes and fixed in Carnoy's fixative.
Temporal emergence of the primitive streak and mesoderm induced by stabilized β-catenin
To examine the role of canonical Wnt/β-catenin signaling in hES cell growth and differentiation, we generated stable hES cell lines constitutively expressing a fusion protein of a stabilized β-catenin, a mutant ligand-binding domain of the estrogen receptor (ER)(Littlewood et al., 1995). The stabilized β-catenin was produced by deletion of the N-terminal 90 amino acids, including GSK3 phosphorylation sites(Barth et al., 1997). This fusion protein (ΔNβ-cateninER) can be conditionally activated in response to the ER agonist 4-hydroxy-tamoxifen (4OHT), thus enabling consistent and reversible activation of β-catenin. We obtained several independent hES cell clones with stable expression of theΔNβ-cateninER fusion proteins, following the transfection of this expression plasmid and puromycin selection. Results presented here were obtained using one or two clones, but similar results were obtained with further independent clones from two different hES cell lines: KhES-1 and KhES-3. These transgenic clones had normal karyotype and expressed cell-surface markers for hES cells, including SSEA3, SSEA4, TRA-1-60, TRA-1-81 and pluripotent markers POU5F1, SOX2 and NANOGcomparable with the parental hES cells (see Fig. S1A-D in the supplementary material). No obvious effect of 4OHT on the undifferentiated state of parental hES cells was observed (see Fig. S1B,D in the supplementary material). WhenΔNβ-cateninER cells were cultured in chemically defined medium(Yao et al., 2006) without 4OHT, they formed tightly contacted compact colonies with invisible cell-cell boundaries (Fig. 1A). Thus,ΔNβ-cateninER cells maintained an undifferentiated state in culture without 4OHT. In the presence of 4OHT, however, the morphology of the cells began to differentiate with dissociation of the cell-cell junctions and induction of cell scattering within 1 to 2 days, resembling the EMT(Thiery, 2002). By day 3,β-catenin-activated cells exhibited a similar morphology(Fig. 1A), and proliferated well during the entire process, but some cells underwent apoptotic-like cell death between days 4 and 5, and detached from the substrata (data not shown). The undifferentiated hES cells (vehicle) displayed epithelial-like apical-basal polarity with a component of tight junctions, ZO-1, and formed adherence junctions composed of E-cadherin(Fig. 1B,C). By contrast,mesenchymal marker N-cadherin was expressed only at the periphery of ES cell colonies, suggesting spontaneous differentiation of a subpopulation of ES cells (Ullmann et al., 2007). After 3 days of β-catenin activation, cells had disorganized ZO-1 localization and markedly decreased expression of E-cadherin, and displayed a mesenchymal phenotype, including dissociated adherence junctions and a gradual change from E-cadherin to N-cadherin expression(Fig. 1B,C). These results indicate that conditional β-catenin activation in hES cells does not support their self-renewal, but results in enhanced differentiation via EMT induction within a few days.
To further characterize the properties of β-catenin-activated cells,we examined the expression profiles of genes related to pluripotency and differentiation. Cells activated by β-catenin had a rapidly reduced expression of pluripotent markers NANOG, SOX2 and POU5F1within 1 day of activation, which progressively decreased to undetectable levels by day 3 (Fig. 1D). By contrast, the expression of the PS/nascent mesoderm markers T(Brachyury), GSC and MIXL1(Tam and Loebel, 2007)transiently peaked at day 1 of activation, and gradually decreased by day 5. Expression of an early marker of ventral mesoderm (KDR) and pre-cardiac mesoderm (NKX2-5), and of a mesoderm/endoderm marker(CXCR4) (D'Amour et al.,2005; Yasunaga et al.,2005) increased in an orderly manner following the induction of T, MIXL1 and GSC. The BMP, Nodal and FGF signaling pathways are important for the emergence of the PS, mesoderm and endoderm in mouse(Tam and Loebel, 2007). Gene expression analysis showed the progressive induction of BMP4 and FGF8, and a temporal induction of NODAL by β-catenin activation, suggesting the involvement of these factors in the development of the PS/mesoderm in β-catenin-induced hES cell differentiation(Fig. 1D). By contrast, the trophectoderm marker CGA was not expressed(Fig. 3A). The ectoderm marker PAX6 was dominated by β-catenin activation(Fig. 3A), as the canonical Wnt signaling suppresses neural differentiation(Aubert et al., 2002). Immunoblot and immunofluorescence analysis confirmed downregulation of Oct4 and Nanog, and induction of Brachyury and KDR protein(Fig. 1B,E). Thus, these temporal gene expression patterns in differentiating hES cells recapitulate the emergence of the PS, and mark the period of the transition towards mesoderm in early mammalian development.
To evaluate the relative proportion of mesoderm progenitors induced byβ-catenin activation, we analyzed the expression of CXCR4 and KDR as mesoderm markers by fluorescence-activated cell sorting (FACS) analysis. Weak levels of KDR expression were detected in ∼50% of undifferentiated ES cells, while CXCR4 was not expressed (Fig. 2A). A representative 5-day kinetic experiment indicated that the KDR+/CXCR4+ cells were immediately detected after 2 days of β-catenin activation. After 3 days of activation, more than 80% of cells became KDR+/CXCR4+ double positive, and then this proportion declined between 4 and 5 days. To examine whether the KDR+/CXCR4+ cells have the potential to differentiate towards mesoderm derivatives, β-catenin-activated cells at day 3 were cultured under conditions that induce the endothelial cell lineage with VEGF. These cells had endothelial cell-like morphology and prominent induction of endothelial cell markers, including CD34, CDH5 (VE-cadherin), PECAM, TEK (Tie-2) and VWF (von Willebrand factor)(Fig. 2B,C). When these endothelial-like cells were cultured on Matrigel with VEGF, they formed a meshwork of cells, resembling the capillary-like structures formed by mature endothelial cells (Fig. 2D). These results demonstrate that hES-derived mesoderm progenitors induced byβ-catenin have a potential to differentiate into an endothelial cell lineage.
Antagonism of BMP signaling changes the mesoderm progenitor fate towards the anterior PS
To examine whether BMP and Activin/Nodal signaling are involved in theβ-catenin-mediated formation of PS/mesoderm progenitors, the BMP and Activin/Nodal signaling pathways were attenuated by Noggin or SB431542 (SB),which inhibit BMP or Activin/Nodal receptors ALK4/5/7, respectively(Fig. 3A). In the presence of Noggin, β-catenin-induced expression of the mesoderm markers KDR,FOXF1 and VENTX, and of the PS marker MIXL1, which is also expressed in the posterior PS/mesoderm(Robb et al., 2000), was markedly diminished, whereas Noggin consistently enhanced the expression of T and GSC (Fig. 3A; see Fig. S2 in the supplementary material). By contrast,exposure of cells to SB in the presence of 4OHT prevented the induction of differentiation markers, except for T and trophectoderm marker CGA. These findings demonstrate the requirement of both BMP and Activin/Nodal signaling pathway for the β-catenin-induced mesoderm formation in differentiating hES cells. Intriguingly, in contrast to the abolished mesoderm induction by Noggin, BMP signaling blockade induced the expression of the anterior PS/endoderm markers FOXA1, FOXA2, CER1,SHH and SOX17 (Fig. 3A) (D'Amour et al.,2005; Kanai-Azuma et al.,2002; Kubo et al.,2004; Yasunaga et al.,2005). The visceral endoderm marker SOX7, ectoderm marker PAX6 and trophectoderm marker CGA were not induced in Noggin-treated cells, indicating that they did not differentiate into the visceral endoderm, ectoderm and trophectoderm lineages. Immunofluorescence analysis indicated that more than 70% of FOXA2-positive cells were observed only following combined Noggin and β-catenin activation, and they were co-expressed with Brachyury and SOX17 (Fig. 3B).
The anterior PS/endoderm populations develop definitive endoderm and anterior mesoderm derivatives, including hepatic, pancreatic and cardiac lineage (Tam and Loebel,2007). To examine whether β-catenin-activated cells treated with Noggin have a similar differentiation potential with cells in the anterior PS, β-catenin-activated cells with or without Noggin at day 3 were harvested and re-cultured with FGF2 and BMP4 for 4 days. These factors have a crucial role in the differentiation of definitive endoderm and mesoderm progenitors (Gouon-Evans et al.,2006; Mima et al.,1995; Zhang and Bradley,1996). The expression of early hepatic and pancreatic endoderm markers (AFP and PDX1), cardiac markers [NKX2-5 andα myosin heavy chain (MYH6)] was prominently induced only in the Noggin-treated cells (Fig. 3C). These results indicate that β-catenin-activated cells treated with Noggin possess the characteristic of anterior PS/endoderm progenitors similar to that of mouse ES cells and embryo(Gadue et al., 2006).
To determine the optimal requirement of β-catenin and Noggin for generation of mesoderm and the anterior PS progenitors, cells were cultured with these factors for various time periods and then analyzed for gene expression. Induction of T, GSC and FOXA2, and inhibition of mesoderm (KDR) was dependent on the Noggin dose (see Fig. S2A in the supplementary material), and not observed in the absence of 4OHT, indicating the necessity for both β-catenin activation and BMP signaling inhibition in the cell fate change toward the anterior PS progenitors. The expression of T and GSC and mesoderm marker FOXF1 was upregulated within 1 day of β-catenin activation, and reached peak levels by continuous 3 days of activation. The addition of Noggin repressed expression of the FOXF1 gene, whereas the expression of the T, GSC and FOXA2 genes was induced to maximal levels 3 days after β-catenin activation (see Fig. S2B in the supplementary material). Thus, transient activation of β-catenin in hES cells was sufficient to induce mesoderm progenitors, but continuous activation was required for maximal induction of mesoderm markers. In contrast to the requirement for β-catenin, the last 2 days of Noggin treatment were sufficient to inhibit FOXF1 and to induce FOXA2, although continuous treatment of Noggin needed for maximal induction of T and GSC (see Fig. S2C in the supplementary material).
Involvement of PI3-kinase and MAPK signaling pathways inβ-catenin-induced mesoderm and endoderm differentiation
To understand the molecular mechanisms of the PS specification viaβ-catenin and/or BMP antagonism, we investigated the downstream targets of Activin/Nodal and BMP signaling that might affect cell fate specification. SMAD transcriptional factors have a central role in the TGFβ signaling pathway, and graded Nodal/SMAD signaling governs cell fate decisions in the PS(Vincent et al., 2003). Immunoblot analysis revealed that the expression of Oct4 protein was dramatically reduced in both 4OHT- and Noggin-treated cells, while Brachyury expression was induced in those cells (Fig. 4A). Neither Activin nor BMP4 alone induced the expression of Brachyury protein in these culture conditions, indicating the requirement of cooperative actions with β-catenin signaling. Consistent with a previous report that Activin/Nodal signaling supports hES-cell self-renewal(James et al., 2005), an active (phosphorylated) form of SMAD2 protein (P-SMAD2), but not SMAD1, was observed in the undifferentiated hES cells (vehicle) as well as in Activin A-treated cells. Activated SMAD2 in β-catenin activated cells with or without Noggin, however, was reduced compared with that in a vehicle control. Although Activin/Nodal signaling is essential for generation of the PS,mesoderm and endoderm, persistent activation of SMAD2 results in maintenance of the pluripotent state (James et al.,2005). Thus, the reduced SMAD2 signaling activity might be necessary to form the PS, mesoderm and endoderm progenitors in hES cell differentiation. By contrast, SMAD1 (P-SMAD1) was activated in 4OHT-treated cells as well as in BMP4-treated cells, whereas it was completely blocked by Noggin (Fig. 4A). To examine the role of Activin/Nodal signaling in the differentiation of the anterior PS progenitors induced by β-catenin and BMP signaling blockade, ES cells were treated with SB during hES cell differentiation. Inhibition of Activin/Nodal signaling completely suppressed FOXA2 protein expression induced by β-catenin and Noggin (Fig. 4B). These findings, together with the data shown in Fig. 3A, indicate that Activin/Nodal signaling is essential for the formation and specification of the nascent PS in β-catenin-mediated hES cell differentiation.
In addition to the SMADs pathway, accumulating evidence indicates that SMAD-independent pathways are involved in TGFβ signaling(Derynck and Zhang, 2003). BMP signaling has an antagonistic role in the canonical Wnt/β-catenin signaling pathway through inhibition of the PI3-kinase/Akt signaling pathway by PTEN (He et al., 2004; Kobielak et al., 2007). Consistent with the requirement for Akt signaling to maintain ES cell pluripotency (Watanabe et al.,2006), an active form of Akt (P-Akt) was observed in undifferentiated hES cells (vehicle), and later downregulated inβ-catenin activated cells (Fig. 4C). There was a slight increase in the inactive form of GSK3β (P-GSK3β), which is phosphorylated by Akt, in β-catenin activated cells, regardless of the reduced active form of Akt, suggesting the involvement of an Akt-independent regulatory pathway(Etienne-Manneville and Hall,2003). By contrast, inhibition of BMP signaling enhanced phosphorylation of both Akt and GSK3β. Immunofluorescence analysis showed that total β-catenin was localized at the cell membrane in the absence of 4OHT, whereas ΔNβ-cateninER was diffusively distributed within cells, indicating that the ΔNβ-cateninER protein was kept in an inactive form in the absence of 4OHT (Fig. 4D). When cells were treated with 4OHT, ΔNβ-cateninER was concentrated in the nuclei even in the presence or absence of Noggin(Fig. 4D). By contrast, totalβ-catenin was localized at the cell membrane and slightly in the nucleus by 4OHT treatment in comparison with vehicle control cells. Conversely, in Noggin-treated cells, β-catenin was preferentially accumulated in the cytoplasm and nuclei compared with cell treated with 4OHT alone(Fig. 4D). These data suggest that inhibition of BMP signaling by Noggin might enhance the stability of endogenous β-catenin through the Akt/GSK3β signaling pathway during the PS specification. The cadherin family modulates nucleo-cytoplasmic localization, stability and transactivation of β-catenin(Moon et al., 2004). An increased cytoplasmic pool of β-catenin due to disorganized interactions with cadherin family members might thereby enhance β-catenin-mediated transactivation. Indeed, expression of T transcript and protein,which is a direct downstream target of β-catenin(Yamaguchi et al., 1999), and GSC, a direct target of T(Messenger et al., 2005), was enhanced in Noggin-treated cells compared with cells treated with 4OHT alone(Fig. 3A; see Fig. S2A-C in the supplementary material). By contrast, Erk1/2 phosphorylation was enhanced inβ-catenin-activated cells, but was even more prominent in BMP-antagonized cells (Fig. 4C).
It has been shown that inhibition of BMP and FGF/MAPK signaling pathways potentiates the induction of endoderm in Xenopus and zebrafish(Poulain et al., 2006; Sasai et al., 1996). To explore the possible role of the PI3-kinase and MAPK signaling pathways in cell fate specification induced by β-catenin and BMP antagonism, cells were cultured in the presence of MEK1/2 (U0126) or PI3-kinase (LY294002)inhibitors. Consistent with an earlier finding that Erk2 is essential for mesoderm induction (Yao et al.,2003), inhibition of MEK1/2 completely abolished the induction of mesoderm progenitors (FOXF1), whereas inhibition of PI3-kinase had no effect on mesoderm induction (Fig. 5A). Interestingly, the expression of FOXA2 was slightly,but consistently, upregulated by the inhibition of MEK1/2, compared with cells treated with 4OHT alone, suggesting that uncommitted progenitors change their cell fate toward the anterior PS progenitors following MEK1/2 signaling blockade (Fig. 5A). By contrast, FOXA2 expression induced by the antagonism of BMP signaling was completely blocked by the inhibition of PI3-kinase, but not MEK1/2,signaling. Immunofluorescence analysis of FOXA2 protein expression confirmed the quantitative RT-PCR results (Fig. 5B). Thus, these findings demonstrate that the PI3-kinase, but not MEK1/2, pathway, is essential for changing the differentiation ofβ-catenin-mediated hES cells from mesoderm to the anterior PS progenitors.
Generation of mesoderm and the anterior PS from genetically unmanipulated ES cells
Our findings suggest that mesoderm and the anterior PS progenitors can be developed from hES cells by a combination of BMP signaling inhibition andβ-catenin activation. The 6-bromoindirubicin-3′-oxime (BIO), a GSK3 inhibitor, sufficiently activates canonical Wnt/β-catenin signaling(Sato et al., 2004). We examined whether mesoderm and the anterior PS progenitors could be derived from genetically unmanipulated hES cells using BIO. At a lower concentration of BIO (2 μM), hES cells formed an undifferentiated compact colony and maintained their undifferentiated state(Fig. 6A,B), as described previously (Sato et al.,2004). By contrast, treatment with a 2.5-fold higher concentration of BIO (5 μM) stimulated the dissociation of cell-cell adhesion and mesoderm induction with a concurrent loss of pluripotent markers(Fig. 6A,B). Immunofluorescence analysis showed that β-catenin was mainly localized at the cell membrane and cytoplasm by the lower concentrations of BIO, whereas the higher concentrations of BIO prominently enhanced nuclear accumulation ofβ-catenin (Fig. 6E). These cells, however, had lower proliferation rates than cells activated byΔNβ-cateninER (data not shown). It might be due to the fact that BIO targets various protein kinases, such as Cdk family and MAP kinases(Meijer et al., 2003), and/or that GSK3β has a role in chromosomal alignment during mitosis(Tighe et al., 2007). Thus,the biphasic effect of the canonical Wnt/β-catenin signaling on the differentiation and maintenance of pluripotency was observed. Moreover, the combination of Noggin and BIO induced expression of FOXA2 and repressed mesoderm markers in hES cells (Fig. 6C,D). Similar data were obtained in another hES cell line, HES-3 and KhES-1 cells (Fig. 6E-G;see Fig. S3A in the supplementary material).
We further examined whether the canonical Wnt/β-catenin is involved in the specification of mesendoderm/endoderm induced by Activin during hES cell differentiation, because it has been demonstrated the synergistic interaction of the canonical Wnt and Nodal/Activin signaling pathway in mesoderm and endoderm specification in mouse embryo and ES cell system(Gadue et al., 2006; Tam and Loebel, 2007). Consistent with a previous report (D'Amour et al., 2005), expression of mesendoderm markers (T and GSC) and definitive endoderm markers (CER1 and FOXA2) was induced in the presence of a high concentration of Activin A (Fig. 7A). By contrast,expression of these mesendoderm/endoderm markers was markedly diminished when Wnt signaling was inhibited by the addition of DKK1(Glinka et al., 1998). Conversely, activation of ΔNβ-cateninER by 4OHT with Activin enhanced expression of mesendoderm/endoderm markers rather than Activin alone(Fig. 7B). Immunoblot analysis showed that phosphorylation of SMAD2 and expression of FOXA2 protein were enhanced in cells by β-catenin activation with Activin(Fig. 7C). Taken together,these data indicate that mesendoderm/endoderm specification of hES cells in the culture was defined by the cooperative interaction of Wnt/β-catenin and Activin signaling pathway.
Genetic evidence from a wide variety of vertebrate species demonstrates that the canonical Wnt/β-catenin signaling has crucial roles in diverse developmental processes during embryonic patterning, such as the PS, mesoderm and axis formation, and is also involved in the formation of definitive endoderm progenitors (Lickert et al.,2002; Tam and Loebel,2007). In the present study, we clearly demonstrated the cooperative interactions of the canonical Wnt/β-catenin, Activin/Nodal and BMP signaling pathways for the induction and specification of the PS,mesoderm and endoderm during hES cell differentiation. The stabilization ofβ-catenin in hES cells is sufficient to induce the posterior PS and mesoderm formation through the induction of Activin/Nodal and BMP signaling. Importantly, Activin/Nodal and Wnt/β-catenin signaling were synergistically required for the generation and specification of the anterior PS/endoderm. Moreover, we found that blockade of BMP or MAPK, but not PI3-kinase, signaling completely abolished mesoderm induction, and instead changed the cell fate toward the anterior PS/endoderm progenitors, indicating that BMP and MAPK signaling have an antagonistic role in the formation of the anterior PS/endoderm progenitors. Taken together, our findings indicate that balance of BMP and Activin/Nodal signaling with the canonical Wnt/β-catenin signaling specifies the cell fate of the nascent PS into the mesoderm or endoderm progenitors (Fig. 7D). Our conclusion is supported by reports using mouse ES cell system (Murry and Keller,2008; Nostro et al.,2008).
BMP signaling negatively regulates endoderm formation in Xenopus(Sasai et al., 1996), but the molecular mechanisms responsible for this inhibition are unclear. There are several lines of evidence for antagonistic interactions between BMP and Wnt/β-catenin signaling. In Xenopus, Brachyury (Xbra), which is a direct target of Wnt signaling, associates with SMAD1 in response to BMP4,and inhibits goosecoid induction and anteriorization of Xbra in the posterior-ventral region of the embryo(Messenger et al., 2005). In mice, BMP signaling suppresses the expression of Lef1, and consequently attenuates the transcriptional activity of β-catenin/Lef1(Jamora et al., 2003). In addition, PTEN decreases the stability of β-catenin under the control of BMP signaling via the inhibition of PI3-kinase/Akt signaling and subsequent activation of GSK3β (He et al.,2004; Kobielak et al.,2007). Consistent with these notions, PI3-kinase signaling was essential for the generation of endoderm progenitors, at least in part,through the Akt-mediated modulation of cytoplasmic free β-catenin levels. As enhanced activation of Wnt/β-catenin signaling is necessary for the specification of anterior endoderm progenitors in mES cells(Zamparini et al., 2006), our data suggest that modulation of cytoplasmic free-β-catenin levels,associated with BMP-induced inhibition of the PI3-kinase/Akt pathway, provides a molecular link between BMP and Wnt signaling pathways for the cell fate specification of the nascent PS in hES cell differentiation.
Previous reports have shown that the canonical Wnt/β-catenin signaling pathway supports self-renewal and pluripotency of both mouse and human ES cells (Hao et al., 2006; Sato et al., 2004). Although these reports seem to contradict our findings, there is some evidence that Wnt signaling plays a role in the lineage specification of ES cells, depending on their context (Dravid et al.,2005; Gadue et al.,2006; Hao et al.,2006; Lindsley et al.,2006). Importantly, pluripotent epiblast stem cells (EpiSCs) have been recently established from mouse post-implantation epiblasts, and appear to have features similar to those of hES cells(Brons et al., 2007; Tesar et al., 2007). These reports suggest that the distinct properties of hES-cell self-renewal and differentiation might be due to their epiblast origin. In agreement with this,in mouse embryo the epiblast differentiates prematurely into mesoderm cells when stabilized β-catenin is constitutively expressed(Kemler et al., 2004). The possibility that the Wnt signaling pathway has different functions at various developmental stages by cooperating with distinct partners and/or regulating distinct downstream targets might affect the interpretation of the effect of Wnt signaling on self-renewal and differentiation of mES and hES cells.
Alternatively, there might be a threshold of Wnt/β-catenin activity involved in the biphasic property of Wnt signaling. Blockade of GSK3βwith higher concentrations of BIO prominently induced nuclear translocation ofβ-catenin and mesoderm differentiation of unmanipulated hES cells,whereas the lower concentrations of BIO seemed to support their self-renewal(Fig. 6; see Fig. S3A in the supplementary material). A similar activity-dependent effect of β-catenin on self-renewal or differentiation was obtained by titrating the 4OHT concentration (see Fig. S3B in the supplementary material). At the lower concentrations of 4OHT, which anticipates modest activation of β-catenin,hES cells were seemingly maintained in a self-renewal state, despite the weak induction of mesoderm markers, whereas at the higher concentration of 4OHT the undifferentiated stem cell state of the hES cells was abolished. These findings are consistent with a model in which small changes in the cellular levels of crucial transcriptional factors, such as Oct3/4 and PU.1, define the lineage commitment of stem cells (Gurdon and Bourillot, 2001; Laslo et al., 2006; Niwa et al.,2000). Thus, we propose that the canonical Wnt/β-catenin signaling in hES cells has biphasic roles in controlling self-renewal and differentiation, depending on a specific threshold of β-catenin activity,although distinct properties of the individual hES cell lines reflect differences in their susceptibility to BIO(Fig. 6; see Fig. S3 in the supplementary material).
In summary, we have demonstrated that the canonical Wnt/β-catenin signaling pathway in differentiating hES cells has significant roles in establishing the PS/mesendoderm and mesoderm progenitors, which recapitulates the global developmental program during the early embryogenesis. More importantly, we show that the nascent PS populations changed their cell fate to the anterior PS/endoderm or posterior PS/mesoderm progenitors following modulation of the Activin/Nodal and BMP signaling pathways. Thus, the reciprocal balance of Activin/Nodal and BMP signaling pathways have crucial roles in the cell fate specification of the naive PS/mesendoderm, which is induced by activation of the canonical Wnt/β-catenin signaling in hES cell differentiation (Fig. 7D). Because precise regulation of cell lineages is indispensable for efficient production of functional cells from hES cells, our findings would be valuable for devising methods for such functional cell production. Future studies, such as genome-wide epigenetic and gene expression analysis, will further enhance our understanding of how lineage specification of hES cells is determined.
We thank Dr Pierre Chambon (IGBMC, France) for the kind gift of the pCreERT2 plasmid. This work was supported by the Grant-in-Aid for Young Scientists (B) and the National BioResource Project of the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), and by the Japan Society for the Promotion of Science.