When embryonic stem cells are allowed to aggregate, the outer layer of the aggregated spheres (referred to as embryoid bodies) differentiates into primitive endoderm. This initial specification of cell lineage facilitates further differentiation of the inner mass of the embryoid bodies. These processes are considered to recapitulate early embryonic development from the blastocyst stage to the egg-cylinder stage. Formation of the primitive endoderm layer in the embryoid bodies was induced solely by aggregation of embryonic stem cells, in the presence of leukemia inhibitory factor/STAT3 and serum/BMP4, which were considered to be sufficient for embryonic stem cell self-renewal. Interestingly, cell aggregation by itself induced Nanog repression at the outer layer, which was essential for aggregation-induced primitive endoderm formation. These data illustrate aggregation-based cell-fate specification during early embryonic development, when downregulation of Nanog plays a crucial role.

Murine embryonic stem (ES) cells are derived from the inner cell mass of blastocyst [day 3.5 (E3.5) preimplantation embryo] and maintained as undifferentiated stem cells on a feeder layer of mouse embryonic fibroblasts (MEFs). Undifferentiated ES cells can also be maintained without a feeder layer, if they are cultured on gelatin-coated dishes in a serum-containing medium supplemented with leukemia inhibitory factor (LIF) (Smith et al., 1988; Williams et al., 1988). When LIF is removed from the medium in vitro, these ES cells start to differentiate into all three germ layers (endoderm, mesoderm and ectoderm) (Evans and Kaufman, 1981; Martin, 1981). In a typical experiment, ES cells are suspended in the medium without LIF on non-adherent culture plates or in hanging drops. The suspended ES cells spontaneously aggregate to form spheres, which are called embryoid bodies. Following the aggregation, the outer layer of the embryoid bodies is specified as extraembryonic primitive endoderm and further differentiates into visceral endoderm (Doetschman et al., 1985). Extracellular matrix, which is secreted from the primitive endoderm layer, then forms a thick basement membrane, separating the primitive endoderm from the inner mass of the embryoid bodies. Along the inside of the basement membrane, a primitive ectoderm layer is formed and cavitation occurs in the core of the embryoid bodies. This series of processes is considered to be an in vitro model recapitulating early embryonic development from the blastocyst stage to the egg cylinder stage (E3.5-E6.5) (Coucouvanis and Martin, 1995).

Although LIF is a key factor in maintaining ES cells in an undifferentiated state, the removal of LIF is not essential for initial specification of the primitive endoderm layer in embryonic development. During gestation, LIF is secreted from the oviduct and upregulated during the blastocyst stage (E3.5). The extraembryonic endoderm layer is formed during this stage and in the presence of LIF (Shen and Leder, 1992). In agreement with this, previous studies have demonstrated that aggregated ES cells can give rise to primitive endoderm even in the presence of LIF (Murray and Edgar, 2001; Shen and Leder, 1992). However, the molecular mechanisms underlying this phenomenon are not fully understood.

Molecular mechanisms for self-renewal of pluripotent ES cells have been extensively studied using monolayer culture of murine ES cells. LIF was identified as a factor secreted by MEFs, and sufficient to maintain self-renewal of murine ES cells without the feeder layer in the presence of fetal calf serum (Smith et al., 1988; Williams et al., 1988). STAT3 activation was then shown to be the responsible downstream pathway of the LIF/gp130 signaling for ES-cell self-renewal (Niwa et al., 1998; Matsuda et al., 1999). More recently, supplementation with BMP4 inhibited differentiation of ES cells in a cooperative manner with LIF, and substituted for fetal calf serum (Ying et al., 2003). In addition to these external factors, a novel homeodomain transcription factor, Nanog, was recently discovered and shown to be an essential factor in maintaining pluripotency of ES cells (Chambers et al., 2003; Mitsui et al., 2003). Interestingly, knockout disruption of the Nanog gene resulted in differentiation of ES cells, primarily into the primitive endoderm lineage, suggesting that Nanog is a crucial factor that prevents ES cells from differentiating into primitive endoderm. With these as background, we investigated molecular events occurring during aggregation-mediated cell-fate determination of ES cells into primitive endoderm.

ES-cell culture

Murine ES cells were maintained in an undifferentiated state on gelatin-coated dishes in Knock-out Dulbecco's modified Eagle's medium (DMEM; Gibco BRL, Grand Island, NY) containing 10% knockout serum replacement (Gibco BRL), 1% fetal bovine serum (Atlanta Biologicals, Norcross, GA), 2 mM L-glutamine, 100 units ml–1 penicillin, 100 μg ml–1 streptomycin, 25 mM HEPES (Gibco BRL), 300 μM monothioglycerol (Sigma, St Louis, MO) and 1000 unit ml–1 recombinant mouse LIF (ESGRO; Chemicon, Temecula, CA). Aggregation of ES cells was induced by making hanging drops on the lids of Petri dishes (2000 cells per drop) using the same medium described above with or without supplementation with LIF (Hamazaki et al., 2001). To activate STAT3ER (fusion protein of STAT3 and estrogen receptor), 1 μM 4-hydroxytamoxifen (4HT) (Sigma) was added to the medium (Oka et al., 2002). In some experiments, 10 ng ml–1 BMP4 (Sigma) was added to the medium.

Plasmid construction and transfection

Transgenic ES cells were created that contained a fusion gene between the promoter of the gene encoding α-fetoprotein (Afp) and the gene encoding green fluorescent protein (GFP) (Afp-GFP). A vector containing the fusion gene (pAfp-GFP) (Hamazaki and Terada, 2003) was cloned into pMP8NEBΔLacZ vector using NotI-MluI site. The linearized vector was introduced into hypoxanthine guanine phosphoribosyl transferase (HPRT) deficient ES R1 cells (Misra et al., 2001) by electroporation. Hypoxanthine-aminopterin-thymidine (HAT)-resistant colonies were screened for cells with successful homologous recombination.

Inducible Nanog-overexpressing ES cells (TRE-Nanog ES cells) were created as follows. The open reading frame of Nanog cDNA was generated by polymerase chain reaction (PCR) using LA-Taq polymerase (TAKARA Bio, Otsu, Japan) from cDNA prepared from R1 ES cells. Primers used were; 5′-AAACGCGGATCCATGAGTGTGGGTCTTCCTGGT, and 5′-ACGCGTCGACTCATATTTCACCTGGTGGAGTCA. The PCR fragment was ligated into the BamHI-SalI restriction site of the pTRE2hyg expression vector (Clontech, Palo Alto, CA), which contained a tetracycline-responsive element. This vector was introduced into tetracycline-regulated ES cells (Era and Witte, 2000) by Fugene 6 (Roche, Indianapolis, IN) and selected with hygromycin B (200 μg ml–1) (Invitrogen, Carlsbad, CA). Tetracycline-dependent expression of Nanog was confirmed by the ability to maintain ES-cell morphology in the absence of LIF and to differentiate by addition of 1 μg ml–1 doxycycline (Sigma).

RT-PCR

For reverse-transcription PCR (RT-PCR), total RNA was extracted by using RNA aqueous kit (Ambion, Austin, TX). cDNA was synthesized by using SuperScript II first-strand synthesis system with oligo(dT) (Gibco BRL). PCR was performed by using Taq DNA polymerase (Eppendorf, Westbury, NY). For each gene, the DNA primers were originated from different exons to ensure that the PCR product represents the specific mRNA species and not genomic DNA.

Immunofluorescence staining

ES cell aggregates were washed once with PBS, and fixed in 3.7% formamide/PBS for 15 minutes at room temperature. Fixed cells were embedded in Tissue-Tek OCT compound (Tissue-Tek; Sakura, Torrance, CA) and frozen in a dry-ice/isopentane bath. Cryostat section (7 μm) were treated with 0.5% Triton-X100 in PBS for 5 minutes and blocked with 5% bovine serum albumin in PBS for 1 hour at room temperature. Cells were further incubated with mouse monoclonal anti-mouse-Oct4 antibody (C-10) (Santa Cruz Biotechnology, Santa Cruz, CA), goat polyclonal anti-mouse-osteopontin antibody (R&D Systems, Minneapolis, MN) or goat polyclonal anti-mouse-HNF4α antibody (C-19) (Santa Cruz Biotechnology) for 2 hours at room temperature. After four washes with PBS, cells were incubated with anti-mouse-IgG or anti-goat-IgG antibodies conjugated to fluorescein isothiocyanate (Jackson Immunoresearch Laboratories, West Grove, PA). After four washes with PBS, cells were mounted by Vectashield containing 4,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA).

Cell sorting

A single cell suspension was prepared from aggregates of ES cells by treating them with 0.05% trypsin (Gibco BRL). Dissociated cells were separated by fluorescence-activated cell sorting (FACS) using a Vantage machine (Becton Dickinson, Franklin Lakes, NJ). Data analysis was performed by using CellQuest Acquisition software (Becton Dickinson).

Whole-mount X-Gal staining

ES cell aggregates were fixed in solution containing 2% formaldehyde, 0.2% glutaraldehyde, 5 mM EGTA, 2 mM MgCl2 and 0.02% NP-40 for 10 minutes at room temperature. After washing twice with PBS containing 0.1% Tween-20 (PBST), the samples were stained in 5-bromo-4-chloro-3-indolyl-D-galactoside (X-Gal) mix [5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6, 2 mM MgCl2, 10% dimethyl sulfoxide, 0.01% sodium deoxycholate, 0.02% NP-40, 1 mg ml–1 X-gal] for 8-12 hours at 37°C and then rinsed in PBST. Post-stain fix was performed in 3.7% formaldehyde at 4°C overnight and followed by paraffin embedding. Sections (7 μm) were counterstained with Eosin Y.

To monitor visceral endoderm differentiation during ES-cell differentiation, we previously constructed a vector that allows GFP production under the control of the Afp promoter (–259 to +24, Afp-GFP) (Hamazaki and Terada, 2003). In order to avoid the unexpected gene silencing that occurs occasionally with randomly integrated transgenes during ES-cell differentiation, we here knocked-in a single copy of the Afp-GFP transgene into the HPRT locus of wild-type R1 ES cells (Afp-GFP ES cell). The HPRT gene locus has been demonstrated to be excluded from the silencing effect of transgenes (Misra et al., 2001). Afp-GFP ES cells were maintained as a monolayer on gelatin-coated dishes in the presence of LIF and did not produce GFP as expected (Fig. 1A). Cell aggregation was induced by culturing ES cells in hanging drops in the same ES-maintenance medium with or without LIF. GFP expression was observed in the outer layer of aggregates within 2 days regardless of LIF addition.

Fig. 1.

(A) Visualization of visceral endoderm differentiation in ES cell aggregates by using GFP expression under the control of the promoter of the Afp gene (which encodes α-fetoprotein). GFP expression was not detected when Afp-GFP ES cells were cultured as monolayer in the presence of LIF (ES, left). GFP-positive cells appeared on the outer layer of aggregates within 4 days regardless of the presence of LIF in the medium (LIF– or LIF+, middle and right). (top) Phase-contrast image under a light microscope; (bottom) pictures under a fluorescence microscope. Bars, 200 μm. (B) Expression of ES-cell-specific genes and differentiation markers upon ES-cell aggregation. Total RNA was isolated from undifferentiated ES cells (ES) and ES-cell aggregates (Day2, Day4). RNA was subjected to RT-PCR analysis and the PCR products were separated on a 2% agarose gel and visualized by ethidium-bromide staining. TTR, transthyretin. (C) The LIF/STAT3 pathway and BMP4 stimulation did not inhibit visceral endoderm differentiation upon ES-cell aggregation. Total RNA was isolated from STAT3ER ES cells, which were maintained as monolayer cells in the medium containing either LIF or 4HT (ES) and the cells, which were cultured as aggregates for 2 days in the same medium with or without BMP4 as indicated.

Fig. 1.

(A) Visualization of visceral endoderm differentiation in ES cell aggregates by using GFP expression under the control of the promoter of the Afp gene (which encodes α-fetoprotein). GFP expression was not detected when Afp-GFP ES cells were cultured as monolayer in the presence of LIF (ES, left). GFP-positive cells appeared on the outer layer of aggregates within 4 days regardless of the presence of LIF in the medium (LIF– or LIF+, middle and right). (top) Phase-contrast image under a light microscope; (bottom) pictures under a fluorescence microscope. Bars, 200 μm. (B) Expression of ES-cell-specific genes and differentiation markers upon ES-cell aggregation. Total RNA was isolated from undifferentiated ES cells (ES) and ES-cell aggregates (Day2, Day4). RNA was subjected to RT-PCR analysis and the PCR products were separated on a 2% agarose gel and visualized by ethidium-bromide staining. TTR, transthyretin. (C) The LIF/STAT3 pathway and BMP4 stimulation did not inhibit visceral endoderm differentiation upon ES-cell aggregation. Total RNA was isolated from STAT3ER ES cells, which were maintained as monolayer cells in the medium containing either LIF or 4HT (ES) and the cells, which were cultured as aggregates for 2 days in the same medium with or without BMP4 as indicated.

Gene expression was then examined in embryoid bodies for genes related to the undifferentiated phenotype of ES cells and differentiation into various lineages (Fig. 1B). Genes indicating visceral endoderm differentiation, such as hepatocyte nuclear factor (HNF) families and transthyretin were upregulated even in the presence of LIF, consistent with the results shown in Fig. 1A. Differentiation into other lineages such as primitive ectoderm and mesoderm was also examined by expression of FGF5 and Wnt3, respectively (Haub and Goldfarb, 1991; Liu et al., 1999). In contrast to endoderm markers, FGF5 and Wnt3 expression was induced in embryoid bodies only when LIF was withdrawn. Rex1, BMP4 and Nanog, which are expressed in undifferentiated ES cells, were rapidly downregulated upon withdrawal of LIF but sustained in the presence of LIF.

LIF is known to activate both the JAK/STAT and Ras/mitogen-activated-protein-kinase (MAPK) signal-transduction pathways in ES cells (Niwa et al., 1998; Ernst et al., 1996). Activation of STAT3 alone was shown to be sufficient to maintain pluripotency in the monolayer culture of ES cells (Matsuda et al., 1999). Ras/MAPK signal was proposed to induce differentiation rather than self-renewal of ES cells (Burdon et al., 1999). Because Ras/MAPK activation through LIF might affect endoderm differentiation in the context of cell aggregates, we tested whether cell-aggregation-induced endoderm differentiation was indeed independent of the STAT3 activity. To activate STAT3 directly, we used ES cells containing a vector expressing a fusion protein of STAT3 and the ligand-binding domain of the estrogen receptor (STAT3ER ES cells) (Matsuda et al., 1999; Oka et al., 2002). In these cells, a synthetic ligand of estrogen, 4-hydroxytamoxifen (4HT) is able to maintain the STAT3ER ES cells in an undifferentiated state in the absence of LIF. By using these ES cells, we examined visceral endoderm gene expression upon aggregation in the presence of 4HT (Fig. 1C). Even when STAT3 was continuously activated by 4HT, the differentiation toward endoderm proceeded upon aggregation.

BMP4 was recently reported to be a cofactor that is required to suppress differentiation of ES cells in the presence of LIF and that can substitute for the requirement for fetal calf serum for ES-cell maintenance (Ying et al., 2003). We tested the effect of additional BMP4 on visceral endoderm differentiation using STAT3ER ES cells. As shown in Fig. 1C, supplementation with BMP4 did not prevent ES-cell differentiation into visceral endoderm. In summary, the soluble factors, which have been known to be essential factors for maintaining ES cells, LIF/STAT3 and BMP4, failed to block initiation of visceral endoderm differentiation upon aggregation.

To define further the molecular events occurring in embryoid bodies, particularly in the presence of LIF, we sorted the outer endoderm layer cells and other inner cells by taking advantage of Afp-GFP expression (Fig. 2A). The cell aggregates were dissociated using trypsin and subjected to FACS. Based on the forward-scatter and side-scatter analysis, the GFP-negative cells were shown to be relatively homogeneous compared with the GFP-positive cells, which exhibited larger cells and coarser nuclear configuration. RT-PCR analysis for those isolated cell populations demonstrated that visceral endoderm marker genes including GATA4/6, HNF families, transthyretin and Afp were exclusively expressed in the GFP-positive population. Furthermore, we could also detect gene expression of BMP2 and Ihh, which are known to be secreted from visceral endoderm and are crucial for mesoderm differentiation of epiblasts (Dyer et al., 2001; Maye et al., 2000). ES-specific genes including Nanog (with exception of Oct-4) were downregulated in this GFP-positive population. Although Oct-4 gene expression was detectable by RT-PCR, the downstream target gene of Oct-4 (which encodes osteopontin) was downregulated in the GFP-positive cells. Therefore, we further investigated protein expression of Oct-4 by immunofluorescence staining and revealed that Oct-4 expression was remarkably reduced in the outer layer of aggregates, implying post-transcriptional suppression (Fig. 2B) (Xu et al., 2004). By contrast, HNF4 was detected primarily in the GFP-positive outer layer of the aggregates.

Fig. 2.

(A) Separation of GFP-positive visceral endoderm cells from ES-cell aggregates. Afp-GFP ES cells were aggregated for 4 days in the presence of LIF and subjected to FACS. The cells in R2 region were collected as GFP-positive cells and those in R3 as GFP-negative cells. The y axis indicates forwards scatter and the x axis indicates GFP expression (left). Total RNA was isolated from both GFP-positive and GFP-negative cells, and the RNA was subjected to RT-PCR analysis. PCR products were separated on a 2% agarose gel and visualized by ethidium-bromide staining (right). OPN, osteopontin; Ihh, Indian hedgehog; TTR, transthyretin; AFP, α-fetoprotein. (B) Oct4 and osteopontin (OPN) expression were downregulated in the outer layer of ES-cell aggregates. After 4 days of aggregation of Afp-GFP ES cells in the presence of LIF, the aggregates were fixed with 3.7% formaldehyde and subjected to frozen section. The sections were incubated initially with anti-Oct4 antibody (1:100 dilution), anti-osteopontin antibody (1:500 dilution) or anti-HNF4α antibody (1:100 dilution), and subsequently with fluorescein-isothiocyanate-conjugated anti-mouse-IgG (1:200 dilution; for Oct4) or anti-goat-IgG (1:200 dilution; for osteopontin and HNF4α) (left, red). Endogenous GFP expression derived from Afp-GFP ES cells was detected in the corresponding section (middle, green). Cells were co-stained with 4,6-diamidino-2-phenylindole to demonstrate nuclei (right, blue). Bars, 50 μm.

Fig. 2.

(A) Separation of GFP-positive visceral endoderm cells from ES-cell aggregates. Afp-GFP ES cells were aggregated for 4 days in the presence of LIF and subjected to FACS. The cells in R2 region were collected as GFP-positive cells and those in R3 as GFP-negative cells. The y axis indicates forwards scatter and the x axis indicates GFP expression (left). Total RNA was isolated from both GFP-positive and GFP-negative cells, and the RNA was subjected to RT-PCR analysis. PCR products were separated on a 2% agarose gel and visualized by ethidium-bromide staining (right). OPN, osteopontin; Ihh, Indian hedgehog; TTR, transthyretin; AFP, α-fetoprotein. (B) Oct4 and osteopontin (OPN) expression were downregulated in the outer layer of ES-cell aggregates. After 4 days of aggregation of Afp-GFP ES cells in the presence of LIF, the aggregates were fixed with 3.7% formaldehyde and subjected to frozen section. The sections were incubated initially with anti-Oct4 antibody (1:100 dilution), anti-osteopontin antibody (1:500 dilution) or anti-HNF4α antibody (1:100 dilution), and subsequently with fluorescein-isothiocyanate-conjugated anti-mouse-IgG (1:200 dilution; for Oct4) or anti-goat-IgG (1:200 dilution; for osteopontin and HNF4α) (left, red). Endogenous GFP expression derived from Afp-GFP ES cells was detected in the corresponding section (middle, green). Cells were co-stained with 4,6-diamidino-2-phenylindole to demonstrate nuclei (right, blue). Bars, 50 μm.

Among several essential molecules for maintaining pluripotency of ES cells, we were particularly interested in Nanog as a potential negative regulator of primitive endoderm differentiation. The gene-targeting study revealed that Nanog-deficient blastocysts failed to develop inner cell mass and predominantly differentiated into extraembryonic endoderm lineage (Chambers et al., 2003; Mitsui et al., 2003). Because the isolated GFP-positive endoderm cells demonstrated significant reduction of Nanog expression, we further confirmed that downregulation of Nanog in the outer layer cells by using Nanog heterozygous-null ES cells, in which Nanog homeodomain locus was targeted by IRES β-geo (internal ribosomal entry site, β-galactosidase/neomycin-resistance gene) cassette (Fig. 3) (Mitsui et al., 2003). In these ES cells, β-galactosidase expression represents endogenous Nanog expression. Consistent with the FACS data above, Nanog expression was downregulated only in the outer layer of the aggregates in the presence of LIF within 2 days. X-Gal staining was sustained in the inner mass of aggregates over 4 days. By contrast, X-gal staining eventually disappeared in whole aggregates within 2 days in the absence of LIF.

Fig. 3.

Downregulation of Nanog expression in ES cells upon aggregation. Whole-mount X-Gal staining of Nanog βgeo ES cells that were aggregated for the indicated number of days in the presence of LIF (LIF+) or the absence of LIF (LIF–). Rosa26 ES cells were stained as a positive control for whole-mount X-gal staining (Rosa). The paraffin sections were counterstained with eosin Y (pink). Bars, 50 μm.

Fig. 3.

Downregulation of Nanog expression in ES cells upon aggregation. Whole-mount X-Gal staining of Nanog βgeo ES cells that were aggregated for the indicated number of days in the presence of LIF (LIF+) or the absence of LIF (LIF–). Rosa26 ES cells were stained as a positive control for whole-mount X-gal staining (Rosa). The paraffin sections were counterstained with eosin Y (pink). Bars, 50 μm.

We then examined whether induced overexpression of Nanog could inhibit the visceral endoderm differentiation upon ES-cell aggregation. We constructed a vector in which Nanog expression was driven by tetracycline-responsive promoter and introduced this into tetracycline-regulated ES cells (tet-off system) (Era and Witte, 2000). We cloned a stable ES-cell line (TRE-Nanog ES cells) that could be maintained in an undifferentiated state in the absence of doxycycline and differentiate in the presence of doxycycline (Fig. 4A). By using the TRE-Nanog ES cells, we examined visceral endoderm differentiation upon aggregation in the presence of LIF. Interestingly, the aggregates remained as smooth spheres lacking a distinguished outer layer when Nanog was constitutively expressed. There were no obvious changes in cell growth or apoptosis in these aggregates (Fig. 4B). When gene expression was examined in these aggregates, overexpression of Nanog (absence of doxycycline) completely inhibited the production of endoderm markers such as GATA4/6, HNF families and transthyretin (Fig. 4C). Nanog overexpression inhibited the differentiation even without LIF. This result demonstrated that Nanog downregulation was indispensable for visceral endoderm differentiation of ES cells upon aggregation.

Fig. 4.

(A) TRE-Nanog ES cells were plated on gelatin-coated plate and cultivated for 3 days in the presence or absence of LIF and doxycycline (DOX, 1 μg ml–1). In the absence of DOX, Nanog transgene was overexpressed in the cells (tet-off system). Nanog-overexpressed ES cells maintained undifferentiated morphology regardless of the presence of LIF. Bars, 500 μm. (B) Morphology of TRE-Nanog ES-cell aggregates. TRE-Nanog ES cells were subjected to aggregation culture in the presence or absence of LIF and DOX. The phase-contrast images were captured under a light microscope after 4 days of aggregation. Bars, 200 μm. (C) Overexpression of Nanog inhibited visceral endoderm gene expression in the aggregates. Isolated total RNA from each condition in B was subjected to RT-PCR analysis. PCR products were separated on a 2% agarose gel and visualized by ethidium-bromide staining.

Fig. 4.

(A) TRE-Nanog ES cells were plated on gelatin-coated plate and cultivated for 3 days in the presence or absence of LIF and doxycycline (DOX, 1 μg ml–1). In the absence of DOX, Nanog transgene was overexpressed in the cells (tet-off system). Nanog-overexpressed ES cells maintained undifferentiated morphology regardless of the presence of LIF. Bars, 500 μm. (B) Morphology of TRE-Nanog ES-cell aggregates. TRE-Nanog ES cells were subjected to aggregation culture in the presence or absence of LIF and DOX. The phase-contrast images were captured under a light microscope after 4 days of aggregation. Bars, 200 μm. (C) Overexpression of Nanog inhibited visceral endoderm gene expression in the aggregates. Isolated total RNA from each condition in B was subjected to RT-PCR analysis. PCR products were separated on a 2% agarose gel and visualized by ethidium-bromide staining.

In accordance with previous reports (Murray and Edgar, 2001; Shen and Leder, 1992), we have demonstrated that primitive endoderm specification occurs in the outer layer of ES-cell aggregates regardless of the presence of LIF (Fig. 5). Furthermore, continuous activation of STAT3, which is a crucial downstream mediator of LIF signaling, could not prevent primitive endoderm formation upon aggregation. By contrast, expression of primitive ectoderm and mesoderm markers were induced in ES-cell aggregates only when LIF was withdrawn (Fig. 1), suggesting that the LIF/STAT3 pathway probably plays a pivotal role in preventing ES cells from differentiating into these lineages.

Fig. 5.

Primitive endoderm specification of mouse ES cells can be induced solely by cellular aggregation regardless of the presence of external factors (LIF and BMP4), which are sufficient to maintain self-renewal of ES cells in monolayer culture (top). ES-cell aggregation induces repression of the Nanog gene in the outer layer independent of the LIF/STAT3 and BMP4/Smads pathways (Cavaleri and Schöler, 2003; Ying et al., 2003). The repression of Nanog subsequently leads to ES cells differentiating into the primitive endoderm (bottom).

Fig. 5.

Primitive endoderm specification of mouse ES cells can be induced solely by cellular aggregation regardless of the presence of external factors (LIF and BMP4), which are sufficient to maintain self-renewal of ES cells in monolayer culture (top). ES-cell aggregation induces repression of the Nanog gene in the outer layer independent of the LIF/STAT3 and BMP4/Smads pathways (Cavaleri and Schöler, 2003; Ying et al., 2003). The repression of Nanog subsequently leads to ES cells differentiating into the primitive endoderm (bottom).

BMP4 has recently been identified as another essential factor for ES-cell self-renewal in monolayer culture (Ying et al., 2003). It works as a cofactor to suppress the differentiation of ES cells, especially into neuroectoderm lineage, by activating the Smads/Id pathway in the presence of LIF. In aggregation culture, the addition of BMP4 to the culture medium did not alter the differentiation preference of ES cells towards the primitive endoderm, regardless of the presence of LIF. Thus, even the combination of both LIF and BMP4 was insufficient to inhibit primitive-endoderm differentiation upon ES-cell aggregation. It should be noted here that BMP4 is known rather to induce visceral endoderm differentiation in the absence of LIF (Coucouvanis and Martin, 1999; Turksen and Troy, 2001).

We further investigated here how Nanog was involved in primitive-endoderm specification upon aggregation. The present study clearly demonstrated that cell aggregation by itself induced Nanog repression at the outer layer. Furthermore, by using the inducible Nanog overexpression system, we proved that this Nanog downregulation was essential for subsequent ES-cell differentiation into primitive endoderm upon aggregation. These cell-aggregation-based changes in vitro should represent the process of cell-lineage specification in the inner cell mass of blastocysts. Importantly, Nanog repression occurred solely by cell aggregation, independent of the LIF/STAT3 or BMP4 pathways. The data strongly suggested that there should be a novel regulatory mechanism by which cell aggregation repressed Nanog expression (Fig. 5).

Aggregation itself would generate polarity that would separate a single outer layer cells and other inner cells. Obviously, the inner cells stayed as pluripotent stem cells in the presence of LIF/STAT3. Indeed, in Fig. 2, when the inner (GFP-negative) cells were re-aggregated, they again formed the GFP-positive outer layer (data not shown). The outer layer cells, by contrast, will have membrane polarization. E-Cadherin has been specifically implicated in the segregation of extraembryonic cells during preimplantation mouse development, at both trophectoderm segregation in compaction and endoderm segregation in the inner cell mass (Damsky et al., 1993). Of interest, the outer layer of the embryoid bodies forms a tight junction, whereas the monolayer culture of ES cells express a very low level of claudins (Kubota et al., 2001) (data not shown). Polar distribution of E-cadherin might lead to signals such as activation of Cdc42/Rac and atypical protein kinases C as well as membrane sequestration of β-catenin (Etienne-Manneville and Hall, 2002).

A recent report has also claimed that forced expression of an active Ras mutant in ES cells caused downregulation of Nanog expression and led to an endoderm phenotype (Yoshida-Koide et al., 2004). Aggregation-based polarity might activate the Ras pathway, leading to repression of the Nanog gene. Of interest, ES cells with a deletion of Grb2, an upstream activator of the Sos1/Ras pathway, are known to fail to generate primitive endoderm (Cheng et al., 1998). Such possibilities are now under investigation. Cell-fate specification following aggregation is a process repeatedly seen during embryogenesis. The aggregation-based in vitro ES-cell differentiation into primitive endoderm would serve as a useful model to study molecular mechanisms underlying this universal process in development.

We thank S. Duncan, T. Yokota, T. Era and P. Oh for providing valuable materials, S. Sugrue, M. Rutenberg, A. Singh and S. Kehoe for critical reading of the manuscript, and M. Jorgenson, A. Meacham and D. Smith for technical assistance. This work was supported by National Institutes of Health Grants, DK59699 and RR17001 (to N.T.).

Burdon, T., Stracey, C., Chambers, I., Nichols, J. and Smith, A. (
1999
). Suppression of SHP-2 and ERK signalling promotes self-renewal of mouse embryonic stem cells.
Dev. Biol.
210
,
30
-43.
Cavaleri, F. and Schöler, H. R. (
2003
). Nanog: a new recruit to the embryonic stem cell orchestra.
Cell
113
,
551
-552.
Chambers, I., Colby, D., Robertson, M., Nichols, J., Lee, S., Tweedie, S. and Smith, A. (
2003
). Functional expression cloning of Nanog, a pluripotency sustaining factor in embryonic stem cells.
Cell
113
,
643
-655.
Cheng, A. M., Saxton, T. M., Sakai, R., Kulkarni, S., Mbamalu, G., Vogel, W., Tortorice, C. G., Cardiff, R. D., Cross, J. C., Muller, W. J. et al. (
1998
). Mammalian Grb2 regulates multiple steps in embryonic development and malignant transformation.
Cell
95
,
793
-803.
Coucouvanis, E. and Martin, G. R. (
1995
). Signals for death and survival: a two-step mechanism for cavitation in the vertebrate embryo.
Cell
83
,
279
-287.
Coucouvanis, E. and Martin, G. R. (
1999
). BMP signaling plays a role in visceral endoderm differentiation and cavitation in the early mouse embryo.
Development
126
,
535
-546.
Damsky, C., Sutherland, A. and Fisher, S. (
1993
). Extracellular matrix 5: adhesive interactions in early mammalian embryogenesis, implantation, and placentation.
FASEB J.
7
,
1320
-1329.
Doetschman, T. C., Eistetter, H., Katz, M., Schmidt, W. and Kemler, R. (
1985
). The in vitro development of blastocyst-derived embryonic stem cell lines: formation of visceral yolk sac, blood islands and myocardium.
J. Embryol. Exp. Morphol.
87
,
27
-45.
Dyer, M. A., Farrington, S. M., Mohn, D., Munday, J. R. and Baron, M. H. (
2001
). Indian hedgehog activates hematopoiesis and vasculogenesis and can respecify prospective neurectodermal cell fate in the mouse embryo.
Development
128
,
1717
-1730.
Era, T. and Witte, O. N. (
2000
). Regulated expression of P210 Bcr-Abl during embryonic stem cell differentiation stimulates multipotential progenitor expansion and myeloid cell fate.
Proc. Natl. Acad. Sci. USA
97
,
1737
-1742.
Ernst, M., Oates, A. and Dunn, A. R. (
1996
). Gp130-mediated signal transduction in embryonic stem cells involves activation of Jak and Ras/mitogen-activated protein kinase pathways.
J. Biol. Chem.
271
,
30136
-30143.
Etienne-Manneville, S. and Hall, A. (
2002
). Rho GTPases in cell biology.
Nature
420
,
629
-635.
Evans, M. J. and Kaufman, M. H. (
1981
). Establishment in culture of pluripotential cells from mouse embryos.
Nature
292
,
154
-156.
Hamazaki, T. and Terada, N. (
2003
). In vitro differentiation of embryonic stem cells into hepatocytes.
Methods Enzymol.
365
,
277
-287.
Hamazaki, T., Iiboshi, Y., Oka, M., Papst, P. J., Meacham, A. M., Zon, L. I. and Terada, N. (
2001
). Hepatic maturation in differentiating embryonic stem cells in vitro.
FEBS Lett.
497
,
15
-19.
Haub, O. and Goldfarb, M. (
1991
). Expression of the fibroblast growth factor-5 gene in the mouse embryo.
Development
112
,
397
-406.
Kubota, H., Chiba, H., Takakuwa, Y., Osanai, M., Tobioka, H., Kohama, G., Mori, M. and Sawada, N. (
2001
). Retinoid X receptor alpha and retinoic acid receptor gamma mediate expression of genes encoding tight-junction proteins and barrier function in F9 cells during visceral endodermal differentiation.
Exp. Cell Res.
263
,
163
-172.
Liu, P., Wakamiya, M., Shea, M. J., Albrecht, U., Behringer, R. R. and Bradley, A. (
1999
). Requirement for Wnt3 in vertebrate axis formation.
Nat. Genet.
22
,
361
-365.
Martin, G. R. (
1981
). Isolation of a pluripotent cell line from early mouse embryos cultured in medium conditioned by teratocarcinoma stem cells.
Proc. Natl. Acad. Sci. USA
78
,
7634
-7638.
Matsuda, T., Nakamura, T., Nakao, K., Arai, T., Katsuki, M., Heike, T. and Yokota, T. (
1999
). STAT3 activation is sufficient to maintain an undifferentiated state of mouse embryonic stem cells.
EMBO J.
18
,
4261
-4269.
Maye, P., Becker, S., Kasameyer, E., Byrd, N. and Grabel, L. (
2000
). Indian hedgehog signaling in extraembryonic endoderm and ectoderm differentiation in ES embryoid bodies.
Mech. Dev.
94
,
117
-132.
Misra, R. P., Bronson, S. K., Xiao, Q., Garrison, W., Li, J., Zhao, R. and Duncan, S. A. (
2001
). Generation of single-copy transgenic mouse embryos directly from ES cells by tetraploid embryo complementation.
BMC Biotechnol.
1
,
12
-21.
Mitsui, K., Tokuzawa, Y., Itoh, H., Segawa, K., Murakami, M., Takahashi, K., Maruyama, M., Maeda, M. and Yamanaka, S. (
2003
). The homeoprotein Nanog is required for maintenance of pluripotency in mouse epiblast and ES cells.
Cell
113
,
631
-642.
Murray, P. and Edgar, D. (
2001
). The regulation of embryonic stem cell differentiation by leukaemia inhibitory factor (LIF).
Differentiation
68
,
227
-234.
Niwa, H., Burdon, T., Chambers, I. and Smith, A. (
1998
). Self-renewal of pluripotent embryonic stem cells is mediated via activation of STAT3.
Genes Dev.
12
,
2048
-2060.
Oka, M., Tagoku, K., Russell, T. L., Nakano, Y., Hamazaki, T., Meyer, E. M., Yokota, T. and Terada, N. (
2002
). CD9 is associated with leukemia inhibitory factor-mediated maintenance of embryonic stem cells.
Mol. Biol. Cell
13
,
1274
-1281.
Shen, M. M. and Leder, P. (
1992
). Leukemia inhibitory factor is expressed by the preimplantation uterus and selectively blocks primitive ectoderm formation in vitro.
Proc. Natl. Acad. Sci. USA
89
,
8240
-8244.
Smith, A. G., Heath, J. K., Donaldson, D. D., Wong, G. G., Moreau, J., Stahl, M. and Rogers, D. (
1988
). Inhibition of pluripotential embryonic stem cell differentiation by purified polypeptides.
Nature
336
,
688
-690.
Turksen, K. and Troy, T. C. (
2001
). Claudin-6: a novel tight junction molecule is developmentally regulated in mouse embryonic epithelium.
Dev. Dyn.
222
,
292
-300.
Williams, R. L., Hilton, D. J., Pease, S., Willson, T. A., Stewart, C. L., Gearing, D. P., Wagner, E. F., Metcalf, D., Nicola, N. A. and Gough, N. M. (
1988
). Myeloid leukaemia inhibitory factor maintains the developmental potential of embryonic stem cells.
Nature
336
,
684
-687.
Xu, H. M., Liao, B., Zhang, Q. J., Wang, B. B., Li, H., Zhong, X. M., Sheng, H. Z., Zhao, Y. X., Zhao, Y. M. and Jin, Y. (
2004
). Wwp2, an E3 ubiquitin ligase that targets transcription factor Oct-4 for ubiquitination.
J. Biol. Chem.
279
,
23495
-23503.
Ying, Q. L., Nichols, J., Chambers, I. and Smith, A. (
2003
). BMP induction of Id proteins suppresses differentiation and sustains embryonic stem cell self-renewal in collaboration with STAT3.
Cell
115
,
281
-292.
Yoshida-Koide, U., Matsuda, T., Saikawa, K., Nakanuma, Y., Yokota, T., Asashima, M. and Koide, H. (
2004
). Involvement of Ras in extraembryonic endoderm differentiation of embryonic stem cells.
Biochem. Biophys. Res. Commun.
313
,
475
-481.