Author contributions: F.P.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and financial support; H.Y.: collection and/or assembly of data and data analysis and interpretation; A.J.: collection and/or assembly of data; W.I.: collection and/or assembly of data; F.G.: financial support; M.R.: conception and design, data analysis and interpretation, manuscript writing, and financial support.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS January 10, 2013.
Activation of the canonical Wnt signaling pathway synergizes with leukemia inhibitory factor (LIF) to maintain pluripotency of mouse embryonic stem cells (mESCs). However, in the absence of LIF, Wnt signaling is unable to maintain ESCs in the undifferentiated state. To investigate the role of canonical Wnt signaling in pluripotency and lineage specification, we expressed Wnt3a in mESCs and characterized them in growth and differentiation. We found that activated canonical Wnt signaling induced the formation of a reversible metastable primitive endoderm state in mESC. Upon subsequent differentiation, Wnt3a-stimulated mESCs gave rise to large quantities of visceral endoderm. Furthermore, we determined that the ability of canonical Wnt signaling to induce a metastable primitive endoderm state was mediated by Tbx3. Our data demonstrates a specific role for canonical Wnt signaling in promoting pluripotency while at the same time priming cells for subsequent differentiation into the primitive endoderm lineage. STEM CELLS2013;31:752–764
Embryonic stem cells (ESCs) are pluripotent stem cells derived from the inner cell mass (ICM) of a preimplantation blastocyst and thus exhibit characteristic gene expression profiles and functional properties . ESCs retain the capacity to repopulate an embryo by contributing to all somatic cell lineages, including the germline . For these reasons, ESCs are characterized as naive pluripotent cells.
Recent evidence implies ESCs are heterogeneous and consist of subpopulations with distinct functional and epigenetic states. These metastable subpopulations share similarities to three different transient cell populations in the mouse embryo: the ICM [3–5], the epiblast cells of postimplantation embryos , and the primitive endoderm . Moreover, a particular metastable state is maintained in a dynamic equilibrium allowing for the interconversion between the ground state and a lineage primed state. The molecular mechanisms that direct the generation of specific ESC metastable states have not been defined.
Leukemia inhibitory factor (LIF)/Stat-3 signaling is the predominant pathway involved in the maintenance of the pluripotent ground state in mouse ESCs (mESCs) . Currently, the signaling pathways that regulate the epiblast and primitive endoderm metastable states remain largely unknown. The Wnt family of secreted glycoproteins plays diverse and critical roles in the proper specification and formation of organs and tissues [9, 10]. Previous experiments implicate Wnt signaling in the maintenance of potency for both embryonic and adult stem cells [9, 11–13]. Furthermore, numerous studies demonstrate that activating canonical Wnt signaling promotes pluripotency in both mESCs and human ESCs [14, 15].
Under self-renewing conditions, Wnts synergize with LIF to maintain pluripotency [8, 16]. In addition, core pluripotency factors (Oct4, Sox2, and Nanog) are known to be regulated by canonical Wnt signaling [17–19]. In fact Tcf3 (Tcf7L1), a key downstream effector of β-catenin-mediated canonical Wnt signaling, binds to the promoters of Oct4, Sox2, and Nanog and limits their expression [20, 21]. Consistent with the ability of Tcf3 to repress pluripotency, Tcf3−/− mouse ESCs are delayed in their differentiation as embryoid bodies (EBs) and are able to self-renew even upon LIF withdrawal .
The canonical Wnt pathway is regulated by the expression level and subcellular localization of β-catenin. Intracellular β-catenin is phosphorylated by GSK3-β when bound in a complex with Axin and adenomatous polyposis coli (APC). Phosphorylated β-catenin is subsequently degraded by the proteasome . Upon binding of Wnt to Frizzled and Lrp5/6 cell surface receptors, an intracellular cascade in which Dishevelled (Dvl) inhibits the phosphorylation of β-catenin by GSK3-β is activated [23, 24]. Cytoplasmic β-catenin is stabilized and translocates into the nucleus to serve as a coregulator for Tcf/Lef transcription factors (TFs), which in turn bind to target genes to modulate their expression . Numerous experiments have identified Wnt3a as a canonical Wnt and demonstrate that it can functionally affect pluripotency [26–28].
In embryonic development, the earliest activation of Wnt/β-catenin targets occurs in the visceral endoderm as evidenced by Tcf-based reporters . Many components of the Wnt signaling pathway are expressed in the blastocyst and in the uterine wall prior to implantation [30–32]. Although the AxinLacZ reporter is activated in the epiblast and primitive endoderm of the developing blastocyst [28, 33], opposing views persist regarding whether the Wnt/β-catenin pathway plays a functional role within the preimplantation blastocyst [34–36].
The ability of Wnts to maintain pluripotency in mESCs is dependent on the LIF pathway as constitutively active β-catenin alone is unable to maintain self-renewal . This implies that activated canonical Wnt signaling is insufficient to maintain the pluripotent state of ESCs. This further raises the issue of whether canonical Wnt signaling has other functions in undifferentiated mESCs.
We investigated how activated canonical Wnt signaling regulates mESC pluripotency and cell fate commitment. We determined that activation of canonical Wnt signaling reinforces the undifferentiated state by promoting pluripotency genes while at the same time activating a subset of primitive endoderm genes. Notably, canonical Wnt signaling induces a metastable ESC state in which primitive endoderm genes are upregulated. This subtype of primitive endoderm ESC can interconvert between a Nanog-green fluorescent protein (GFP) high population as well as a Nanog-GFP low population. The induction of primitive endoderm genes in the undifferentiated state comes at the expense of the neuroectodermal and mesendodermal potential. Wnt3a-expressing ESCs display a predisposition for the primitive endoderm lineage and as a result display pronounced visceral endoderm potential following differentiation. We further found that induction of the primitive endoderm metastable state, following activation of canonical Wnt signaling, is mediated by the T-box TF Tbx3. Together, our experiments reveal important insights into the early regulation of ESC pluripotency and commitment by canonical Wnt signaling.
MATERIALS AND METHODS
Cell Lines and Cell Culture
The following mouse ESC lines D3 Gl, J1, R1, and Millitrace Nanog GFP were maintained on a layer of irradiated Mouse Embryonic Fibroblast (MEFs) (DR4) in Glasgow Minimal Essential Medium high glucose medium (Wisent, Saint-Jean-Baptiste de Rouville, QC, Canada, http://www.wisent.ca) supplemented with 15% fetal bovine serum (FBS) (Hyclone, Logan, UT, http://www.hyclone.com), 1× nonessential amino acids, 1× sodium pyruvate, 1× Glutamax, β-mercaptoethanol, 1× Pen/Strep (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), and 1,000 U ESGRO (Chemicon, Temecula, CA, http://www.chemicon.com) at 37°C and 5% CO2. Wnt3a retroviral cell lines used throughout the differentiation time course were made from each of the three ESC lines (D3, J1, and R1). Prior to differentiation, mESCs were cultured for two passages on 0.2% gelatin. For EB formation, mESCs were seeded in 20 μl hanging drops at a density of 4 × 104 cells per ml in ESC medium without LIF. EBs were transferred to petri dishes on Day 3 of differentiation and cultured in suspension until Day 5 at which time they were transferred to gelatin-coated culture dishes for the remainder of differentiation. Plat-E cells, a derivative of HEK293T cells were used to generate retrovirus. Plat-E cells were maintained in Dulbecco's modified Eagle's medium (DMEM) high glucose medium supplemented with 10% FBS (Hyclone), 1× Pen/Strep, 1 μg/ml Puromycin, and 10 μg/ml Blastocydin at 37°C and 5% CO2. 80% confluent Plat-E cells were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer's directions. Cells were fed with fresh DMEM 6 hours post-transfection. After 48 hours of transfection retroviral supernatants were collected, centrifuged to remove cells, and filtered (0.45 μM) for purification. The proliferation assay was conducted by plating 200,000 cells per well of a six-well dish. Following each passage, the cells were counted and replated at a density of 200,000 and the fold increase was multiplied to give rise to a total cell number amount. Recombinant Wnt3a was obtained from RnD systems (1324-WN). (2′Z,3′E)-6-Bromoindirubin-3′-oxime reconstituted in Dimethyl Sulfoxide at a concentration of 5 mM (Sigma B1686).
EBs at 5 days of differentiation were briefly rinsed in Phosphate Buffered Saline (PBS) and subsequently frozen in a 2:1 ratio of Tissue-Tek OCT to 30% sucrose in a dry ice/ethanol bath. Serial cryosections 12 μm in thickness were cut using a cryostat (Leicia CM 1850). Sections were fixed in 4% Paraformaldehyde for 10 minutes at room temperature followed by incubation in 100 mM glycine for 10 minutes at room temperature. Fixed cells were permeabilized and blocked in 5% FBS, 0.2% Triton X-100 (Sigma), and 2% bovine serum albumin (BSA) (Fisher Scientific, Hampton, NH, http://www.fisherscientific.com) for 1 hour at room temperature. Cells were incubated with primary antibodies (View Supporting Information Table SI) for 1 hour at room temperature or alternatively 4°C overnight. Secondary antibodies were diluted 1:1,000 in PBS and incubated with fixed cells for 1 hour at room temperature. Nuclei were counterstained with Hoechst 33342 (Sigma Oakville, Ontario, http://www.sigmaaldrich.com). Images were taken on a Carl Zeiss Axioplan 2 Microscope with an AxioCam HRm b/w (Zeiss Toronto, ON, http://microscopy.zeiss.com) camera using Axiovision v3.2 (Zeiss) acquisition software. Photoshop was used to adjust the baseline fluorescence.
Total protein was harvested in RIPA lysis buffer, subjected to SDS-PAGE, and electroblotted onto Immobilon-P membranes (Millipore, Billerica, MA, http://www.millipore.com). Membranes were probed with primary antibody and Horseradish Peroxidase (HRP) conjugated secondary antibody in blocking solution. Electrochemiluminescence (GE Healthcare Baie d'Urfe, QC, http://www.gelifesciences.com) with Biomax XAR film (Kodak) was used to detect target proteins. Primary antibodies used for detection are found in Supporting Information Table SI. The secondary antibody consisted of an HRP-conjugated anti-mouse antibody (Bio-Rad, Hercules, CA, http://www.bio-rad.com).
siRNA Knockdown of Tbx3
short interfering RNA (siRNA) for Tbx3 was obtained from Ambion (Austin, TX, http://www.ambion.com; s74777-4390771) and transfected into Nanog-GFP mESCs using lipofectamine RNAiMax (Invitrogen) according to the manufacturer's instructions. Fresh medium was added 6 hours post-transfection, and the cells were cultured for an additional 48 hours prior to either RNA collection or fluorescent activated cell sorting (FACS) analysis.
Prior to FACS, Nanog-GFP reporter ESCs were trypsinized, pelleted, and filtered through 30 μM filters (Miltenyi Biotech, Bergisch Gladbach, Germany, http://www.miltenyibiotec.com). Cells were separated on a MoFlo cytometer (DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com) equipped with three lasers. Sorting gates were strictly defined based on both side scatter versus GFP as well as phycoerythrin (PE) versus GFP plots. Wild-type J1 ESCs were used as a negative control for GFP.
Total RNA was isolated and subjected to on column DNase digestion using an RNeasy mini Kit as per the manufacturer's instructions (Qiagen, Hilden, Germany, http://www1.qiagen.com). 1 μg of RNA was reverse transcribed using iScript RT kit (Biorad) and subsequently diluted 1:5, with 1 μl of template used per quantitative polymerase chain reaction (qPCR) reaction. SYBR Green real-time PCR reactions (Biorad) were performed in duplicate using an MX3000p PCR machine (Stratagene, LA Jolla, CA, http://www.stratagene.com) and 40 cycles of amplification (95°C 10 seconds; 60°C 30 seconds; 72°C 30 seconds). PCR primers for real-time PCR were designed using the online Primer3 software (http://frodo.wi.mit.edu/primer3/) with gene-specific information obtained from Ensembl. All primer sequences are provided in Supporting Information Table SII. Relative fold change in expression was normalized to GAPDH and calculated using the ΔΔCT method (CT values <30). Primer specificity was validated by denaturation curve analysis (55°C−94°C) resulting in a single peak. Amplification-curve plotting and calculation of cycle threshold (Ct) values were performed using the MX3000p software (v3; Stratagene), with further calculations performed using Microsoft Excel. Each experiment was performed independently on at least three occasions.
Gene Expression Analysis (RNA-Seq)
Global gene expression was assessed in retroviral and recombinant Wnt3a cell lines relative to mock infected and BSA treated controls, respectively. Total RNA was isolated and subjected to on column DNase digestion using an RNeasy mini Kit as per the manufacturer's instructions (Qiagen). mRNA-Seq was performed at the Erasmus Center for Biomics (Rotterdam, The Netherlands) as per manufacturer's instructions (Illumina). Raw Solexa reads were mapped to the mouse genome (mm9) by Tophat . Gene expression values were calculated by Cufflinks with default parameters . Differentially expressed genes between Wnt3a treated samples and corresponding mock or BSA controls were called by Cuffdiff [38–40]. Heatmaps were generated by GenePattern (http://www.broadinstitute.org/cancer/software/genepattern). Gene ontology (GO) of expression data was done using the functional annotation module of DAVID 6.7 (http://david.abcc.ncifcrf.gov) (Dennis et al., 2003; Huang da et al., 2009).
Statistical analysis was conducted using the Student's t test method of determining inference based on small samples. Two-tailed paired t tests were conducted between treated and control groups at a given day of differentiation or globally for all treated or control samples over the course of differentiation. * indicates a p value of <.05 while ** indicates a p value of <.01.
Activated Canonical Wnt Signaling Maintains Pluripotency but Primes Cells for Primitive Endodermal Commitment
To investigate the regulation of ESC pluripotency and early lineage specification by canonical Wnt signaling, we generated mESCs lines that constitutively express Wnt3a. Mouse ESCs were infected with a self-inactivating retrovirus containing an Elongation Factor 1 alpha (EF1α) promoter to drive the expression of Wnt3a followed by an IRES-Puro for selection. This approach ensures constant and maintained activation of canonical Wnt signaling in a cell-autonomous manner. Elevated levels of Wnt3a mRNA with respect to endogenous Wnt3a expression in control cell lines illustrate the effective and sustained overexpression of Wnt3a in mESCs (Fig. 1A). As previously described [41, 42], we observed a significant increase in the rate of proliferation, percentage of Ki67+ cells, along with a compacted colony morphology in Wnt3a-expressing mESCs (Fig. 1B, 1C; Supporting Information Fig. S1).
To gain insight into how Wnt3a affects pluripotency, we conducted RNA-seq from total mRNA isolated from cells either overexpressing Wnt3a or treated with 250 ng/ml recombinant Wnt3a protein for three passages. We confirmed RNA-seq read quality and the distribution of reads prior to further gene analysis (Supporting Information Fig. S2). In addition, the transcriptional activation of canonical Wnt targets was confirmed in recombinant and retroviral cell lines (Supporting Information Fig. S3).
To investigate the global transcriptional effects of canonical Wnt activation in mESCs, we conducted an unbiased DAVID analysis of our RNA-seq data throughout a 10-day differentiation time course (Supporting Information Fig. S4). Significant enrichment of GO terms representing Chordate embryonic development and embryonic development ending in birth or egg hatching was observed. To determine specifically which cellular lineages—either blastocyst or germ layer derived—were being affected by Wnt3a, we choose to focus on panels of known markers for each lineage.
TFs act to define cell-type-specific lineage identity. Therefore, we compiled panels of known TFs along with additional gene markers specific for a given cellular lineage [43–45]. We focused on early commitment and the lineages involved in the formation of the blastocyst (trophoblast, primitive endoderm, and epiblast), and the primary germ layers (ectoderm, mesoderm, endoderm). In this manner, we could determine the broad effects of Wnt3a on pluripotency and lineage induction.
mRNA levels of the key pluripotency factor Nanog, a known target of canonical Wnt signaling in mESCs, and Klf4 were increased following Wnt3a induction . By contrast, the majority of the other pluripotency markers remained either unchanged, or in the case of Dppa3 (Stella), were reduced (Fig. 1D, 1E). Our RNA-seq and qPCR results for the trophectoderm panel of genes displayed mixed induction and repression upon Wnt3a overexpression in the undifferentiated state. Notably, expression levels of the core TFs that define the primitive endodermal lineage displayed a striking increase in the undifferentiated state both for RNA-seq and qPCR (Gata4, sixfold; Gata6, fivefold; Sox17, 20 fold; Sox7, fivefold) (Fig. 1D, 1E).
Activation of the Wnt signaling pathway in mESC results in the reduced expression of neural ectoderm genes [47–51]. Consistent with these reports, our data show that Pax6, an early marker for neural development, exhibited a 70% reduction in expression in Wnt3a-treated mESC (Fig. 1D, 1E). Furthermore, all 13 ectodermal markers examined by RNA-seq displayed reduced or unchanged expression levels in Wnt3a expressing cells. The majority of endodermal factors were similarly downregulated or unchanged (Fig. 1D, 1E). Mesodermal genes displayed weak induction in undifferentiated cells. In concordance with our RNA-seq results, qPCR validation of RNA levels for core lineage markers followed similar patterns (Fig. 1E).
The effect of recombinant Wnt3a on mESC was similar yet subdued in comparison to retroviral overexpression of Wnt3a. These changes include the induction of core pluripotency and primitive endoderm genes in the undifferentiated state and promotion of visceral endoderm following differentiation (Supporting Information Fig. S5). To further support this claim, we compared the number of genes per lineage panel that undergo an increase in expression following either recombinant or retroviral Wnt3a treatment. This analysis demonstrates that indeed recombinant Wnt3a treatment results in a similar pattern of gene induction during differentiation when compared to retroviral expression of Wnt3a for a given panel of genes (Supporting Information Fig. S5B).Therefore, we conclude that Wnt3a alters the transcriptional program of ESCs to both promote pluripotency and prime cells for entry into the primitive endoderm lineage.
Canonical Wnt Signaling Perturbs the Pluripotent Ground State of mESCs by Inducing a Primitive Endoderm Metastable State
Our gene expression analysis indicated that activated canonical Wnt signaling could induce primitive endoderm genes in mESCs maintained in the undifferentiated state. Recent evidence suggests that pluripotent mESCs can exist in multiple metastable states that are characterized by distinct transcriptional profiles and functional characteristics akin to the epiblast, ICM, and the primitive endoderm [3, 6, 7]. These metastable states are maintained in proliferating mESC cultures in the presence of LIF and can be distinguished based on the expression of Nanog , Rex1 , Stella , Hex , or Pecam1 . The developmental mechanisms that direct mESCs to form a specific metastable state are currently unknown.
Typically, mESCs display high and low Nanog-GFP populations in the undifferentiated state, which can interconvert and are defined as metastable states . Therefore, to investigate whether canonical Wnt signaling induces a metastable mESC state, we analyzed the expression of GFP from a Nanog-GFP reporter cell line following retroviral infection of Wnt3a or empty vectors (Fig. 2A). As noted previously, 93.1% of mESC infected with mock virus displayed high Nanog-GFP expression (Fig. 2B). Importantly, following infection with Wnt3a virus we observed an almost 10-fold expansion in the size of the Nanog-GFP low population (Fig. 2B).
To characterize the identity of the different Nanog-GFP populations (high/low), they were FACS sorted and the expression of lineage markers was analyzed by qPCR. The Nanog-GFP low population displayed markedly upregulated levels of primitive endoderm genes Gata4, Gata6, Sox17, Sox7, Cubn, and Dab2 relative to the Nanog-GFP high population (Fig. 2C). However, these same cells exhibited maintenance of expression of pluripotency gene such as Oct4 and Nanog. Therefore, the observed upregulation in primitive endoderm genes together with the maintenance of pluripotency gene expression suggests that these cells are in a lineage primed state .
Previous reports indicate that the activation of Sox17 in the primitive endoderm commits ESCs to terminal differentiation . As Sox17 is upregulated in the Nanog-GFP low population, we investigated whether these cells were irreversibly committed to differentiation. Sorted Wnt3a Nanog-GFP high or low cells were cultured in the undifferentiated state for 9 days to determine their ability to interconvert. Nanog-GFP high cells maintain GFP expression while rapidly regenerating a GFP low population (Fig. 3). We next tested whether the Wnt3a Nanog-GFP low cells could reactivate high levels of Nanog expression as evidenced by GFP. Two days postsorting of Wnt3a induced Nanog-GFP low cells; we observed the formation of colonies exhibiting heterogeneous GFP expression (Supporting Information Fig. S4). FACS analysis of cells after 3, 6, and 9 days of culture revealed a return to the original distribution of Nanog-GFP expression (Fig. 3; Supporting Information Fig. S4). These results indicate that an equilibrium exists between the primitive endoderm and ICM metastable states and that this equilibrium can be modified but not exclusively restricted by Wnt3a. Taken together, these results strongly support the hypothesis that activated canonical Wnt signaling induces a primitive endoderm metastable state.
Canonical Wnt Signaling Alters the Differentiation Potential of mESCs
To investigate whether the induction of a metastable primitive endoderm state functionally altered the lineage potential of mESCs, we conducted a differentiation time course spanning a mid differentiation time point (Day 5) and a late stage differentiation time point (Day 10) (Fig. 4A). We observed a striking phenotype when Wnt3a-expressing mESCs were differentiated as EBs. These changes include an increase in EB size relative to mock-infected control cells and the development of large cysts after 6 days of differentiation (Fig. 4B, 4C). We further characterized the effect of Wnt3a on the differentiation ability of mESCs by conducting RNA-seq from cells isolated after 5 and 10 days of differentiation.
Pluripotency markers increased following Wnt3a-induced differentiation at Day 5 (8/10) and were subsequently strongly downregulated by Day 10 (8/10). Primitive endoderm markers displayed broad induction throughout the differentiation time course (Day 5–6/10 and Day 10–8/10). Trophectoderm markers demonstrated mixed induction and inhibition after 5 days of differentiation (4/11) and further upregulation by 10 days (8/11) of differentiation.
The vast majority of ectodermal markers 8/13 (Day 5), and 11/13 (Day 10) displayed reduced or unchanged expression levels in Wnt3a expressing cells. Endoderm markers as a whole were all either downregulated or unchanged following differentiation (10/10 Day 5, 9/10 Day 10). Mesodermal genes displayed weak induction during early differentiation (4/10), while on day 10 core mesodermal genes T, Mixl1, and Nodal were all strongly inhibited (>threefold downregulation). In concordance with our RNA-seq results, qPCR validation of RNA levels for core lineage markers followed similar patterns (Fig. 4E).
Taken together, these results support the notion that canonical Wnt signaling in the absence of LIF promotes the formation of cystic EBs and alters the gene expression profile to match differentiation into the primitive endoderm lineage. Furthermore, these results suggest that priming of primitive endoderm genes in the undifferentiated state functionally alters the differentiation potential of mESCs.
Sustained Activation of Canonical Wnt Signaling Directs Visceral Endoderm Commitment
During blastocyst development, the primitive endoderm is localized to a layer of cells that cover the epiblast and separate it from the blastocoel cavity. Following blastocyst implantation, the primitive endoderm will give rise to both the parietal endoderm to line the outer trophectoderm and the visceral endoderm to cover the epiblast . Since activated canonical Wnt signaling induces upregulation of primitive endoderm genes and the formation of cystic EBs, we next tested whether Wnt3a promotes visceral endoderm.
To assess visceral endoderm in our cultures, we examined the expression levels of genes enriched in visceral endoderm at the RNA and protein level. Primary visceral endoderm cell surface proteins Amn, Cubn, and ApoE are all significantly upregulated based on our RNA-seq results (Fig. 5A) as well as by qPCR (Fig. 5B). Dab2 a cell membrane adapter protein along with the TFs Sox7, Hnf4a, Hnf1b, and Foxa2 are all upregulated upon Wnt3a-induced differentiation.
We confirmed the induction of visceral endoderm markers by immunofluorescence in differentiated EBs. Following 5 days of differentiation, we observed an increase in the prevalence of Dab2, Gata4, and Sox17 proteins within EBs following Wnt3a overexpression (Fig. 5C). This increase in the prevalence of EBs that express Gata4, Sox17, and Dab2 was quantified (Fig. 5D) and further validated by Western blot analysis (Fig. 5E). Additional data demonstrating an increase in the prevalence of Cubn positive EBs are shown in Supporting Information Figure S8. Following continued Wnt3a expression, the gene expression profile and protein complement of differentiating mESCs is enriched for primitive and visceral endoderm genes. Taken together these results indicate that upon differentiation the expression of Wnt3a directs commitment toward the visceral endoderm lineage.
Activated Canonical Wnt Signaling Acts Through Tbx3 to Induce a Primitive Endoderm Metastable State
Canonical Wnt signaling typically involves β-catenin binding with a Tcf/Lef family TF on the promoter of Wnt target genes. To identify specific genes that are targeted by Wnt3a in undifferentiated mESCs, we further analyzed our RNA-seq gene profiles obtained following retroviral and recombinant treatment. In this fashion, we would gain mechanistic insight into how Wnt3a activates primitive endoderm genes.
We identified genes that were both upregulated or downregulated in unison by either Wnt3a retroviral overexpression and recombinant Wnt3a treatment. Based on this, we identified three TFs Tbx3, Pou2f3, and Cdx1 that were significantly upregulated upon Wnt3a overexpression or recombinant Wnt3a treatment (Fig. 6A). Upregulation of these genes was subsequently confirmed by qPCR in ESCs overexpressing Wnt3a (Fig. 6B).
Of these three genes, Tbx3 has recently been shown to play a role in pluripotency and is also involved in inhibiting ectoderm and trophectoderm while promoting primitive endoderm formation . Taking into account these findings, we used siRNA directed against Tbx3 to determine whether it is required for Wnt3a induction of primitive endoderm genes and subsequently the Nanog-GFP low population. Indeed, sorted Wnt3a Nanog-GFP low cells that typically express enhanced Tbx3 demonstrate a reduction in Gata6 expression levels following treatment with siTbx3 (Fig. 6C). Furthermore, Nanog-GFP cells transfected with scrambled siRNA displayed a Nanog-GFP low population whereas the mock-infected Nanog-GFP cells did not (Fig. 6D). Upon knockdown of Tbx3, the percentage of Wnt3a Nanog-GFP low cells underwent a 25% ± 2.7% decrease. Conversely, when Tbx3 was knocked down we observed a 56% ± 4.1% increase in the percentage of Nanog-GFP high cells (Fig. 6E). Therefore, Tbx3 plays an important role in mediating Wnt3a induction of the primitive endoderm metastable state.
Embryonic stem cells were initially viewed as a homogeneous pluripotent population that has the capacity to give rise to all cell types of the embryo proper. This simplistic view has recently been challenged following the identification of distinct metastable states that exist in equilibrium within the undifferentiated state. Metastability confers to ESCs the unique ability to interconvert between two distinct cellular states. The first state is fully pluripotent whereas the metastable state is lineage primed but reversible. Cells that reside within a given metastable state display a unique epigenetic signature and are primed for differentiation into a specific lineage . However, the developmental signals and the molecular mechanisms that govern the formation and maintenance of metastable states have not been previously defined.
Here we demonstrate that induction of the canonical Wnt signaling pathway in mESCs induces a pluripotent primitive endodermal metastable state. Moreover, we found that canonical Wnt signaling primes undifferentiated mESC for differentiation into the primitive endoderm lineage. Indeed, this early priming into the primitive endoderm culminates in a profound enrichment of visceral endoderm at the expense of other germ layer lineages in differentiating mESC cultures. Furthermore, we found that induction of a primitive endoderm metastable state following activation of the canonical Wnt signaling pathway is mediated by the T-box TF Tbx3 (Fig. 7). Thus, these results provide a mechanistic framework that explains how activated canonical Wnt signaling can promote visceral endoderm development.
Three metastable states have been defined in mESC. The first is an ICM state characterized by cells that express Stella, Rex1, Nanog, and Pecam1. The second state is similar to the postimplantation epiblast state and is defined by induction of Fgf5 and Brachyury. The third and most recently identified is a primitive endoderm-like metastable state characterized by the expression of the primitive endoderm marker Hex . This state was first identified based on its low expression of Nanog and high expression of the primitive endoderm marker, Gata6 . Further studies using Rex1, Hex, and Sox17 support the idea for a primed primitive endoderm state in mESCs, yet differences exist regarding the ability of this population to reversibly convert back into the naive pluripotent state [5, 7, 53].
Our results demonstrate that activation of canonical Wnt signaling is capable of inducing a primed primitive endodermal metastable state in mESCs. Wnt3a primes mESCs for primitive endoderm specification (Fig. 1D, 1E) and gives rise to a Nanog-GFP low population that upregulates expression of primitive endoderm genes (Fig. 2B, 2C). Importantly, this population is capable of interconverting back to Nanog-GFP high cells following FACS isolation and in vitro culture (Fig. 3A, 3B). These abilities support the characterization of this pluripotent primitive endoderm population as being metastable. Furthermore, lineage priming into the primitive endoderm is apparent as robust differentiation into visceral endoderm occurs in the absence of LIF (Fig. 4B–4E).
Forced expression of Gata4 and Gata6 has been previously demonstrated to direct primitive endoderm differentiation in mESCs [56, 57]. Interestingly, we found induction of primitive endodermal genes such as Gata6, Gata4, Sox7, and Sox17 in both the undifferentiated and differentiated states. In addition, we observed a broad increase in primitive endoderm markers following Wnt3a induction of the canonical Wnt signaling pathway. Therefore, although Wnt3a-treated mESCs remain pluripotent; they become primed for entry into the primitive endoderm lineage. This is an important distinction, as our experiments support the notion that Wnt3a promotes and simultaneously maintain two functionally discrete cells types.
In embryonic development and in vitro differentiation of mESCs, canonical Wnts are associated with the induction of brachyury (T) and the mesendodermal lineage [27, 47, 58]. Interestingly, prior to gastrulation, the earliest activation of downstream target genes of canonical Wnt signaling occur in the nascent visceral endoderm . Notably, mESCs expressing mutant activated β-catenin display an enlarged visceral endoderm and preferentially contribute to the visceral endoderm along with other extra-embryonic tissues . Conversely, mice deficient for β-catenin lack appropriate organization of the visceral endoderm resulting in defective anterior–posterior axis formation. In vitro, mESCs with an inactive form of APC (ApcMin/Min), a key regulator of β-catenin degradation, display increased levels of visceral endoderm upon differentiation [15, 59]. Our experiments elucidate these results by demonstrating that activated canonical Wnt signaling induces a primitive endoderm metastable state in undifferentiated mESCs.
The mechanisms involved in the induction and maintenance of metastable states remain poorly understood. Of several conserved genes upregulated upon recombinant Wnt3a treatment or retroviral overexpression, Tbx3 was chosen due to its expression in the undifferentiated state and for its role in specification of primitive endoderm. Recent studies have shown that Tbx3 along with Klf4 are able to promote pluripotency in the absence of LIF . Tbx3 was previously identified as a downstream target of canonical Wnt signaling in murine and human liver tumors suggesting that a direct mechanism exists for canonical Wnts to activate Tbx3 . Tbx3 is required for extra-embryonic endoderm commitment and also functions to suppress trophectoderm and ectoderm differentiation in mESCs . Overexpression of Tbx3 in mESCs maintained in the presence of LIF results in the induction of Gata6, Sox17, and Dab2 suggesting that the primary role of Tbx3 is to induce the primitive endoderm. Thus, our finding that Tbx3 is required for Wnt3a to induce the primitive endoderm state supports the hypothesis that Tbx3 is a canonical-Wnt target in mESCs.
In conclusion, our data reveals that activation of the canonical Wnt pathway by Wnt3a results in the induction of a metastable population of undifferentiated cells with characteristics of the primitive endoderm. Moreover, our results elucidate a mechanism mediated by Tbx3 in which activation of the canonical Wnt signaling pathway in undifferentiated mESC alters the gene expression profile and distribution of metastable states to promote a primitive endoderm metastable state. Indeed, stimulation of canonical Wnt signaling biases differentiation into the visceral endoderm lineage at the expense of ectoderm, mesoderm, and endoderm. Thus, our experiments indicate that induction of metastable states is subject to developmental control pathways acting during embryogenesis. Together, these findings provide important insight into the regulation of lineage specification and commitment during the development of the early embryo.
We thank Dr. James Ellis for the HSC1-eGFP and HSC1-eGFP IRES Puro vectors; David Wilson for help and technical advice regarding tissue sectioning; Paul Oleynik for cell sorting. M.A.R. holds the Canada Research Chair in Molecular Genetics and is an International Research Scholar of the Howard Hughes Medical Institute. This work was funded by the Canadian Institutes of Health Research grant 160266 and Ontario Research Fund grants to M.A.R., and by EuTRACC, a European Commission sixth Framework grant to F.G. F.D.P was supported by scholarships from the Canadian Stem Cell Network, the Muscular Dystrophy Society of Canada, and a Doctoral Research Award from the Canadian Institutes of Health Research.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.