sign in sign up
<<
Home , GBX2, Primitive streak

Article b-Catenin Regulates Primitive Streak Induction through Collaborative Interactions with SMAD2/ SMAD3 and OCT4

Graphical Abstract Authors Nina S. Funa, Karen A. Schachter, ..., Klaus Hansen, Henrik Semb

Correspondence [email protected]

In Brief In this article, Funa, Schachter, and colleagues et al. provide mechanistic insights into how Wnt signaling controls early lineage decisions. b-catenin occupies regulatory regions in numerous PS and , and direct interactions between b-catenin and SMAD2/SMAD3 in proximity to OCT4 binding are required to specifically activate PS genes.

Highlights Accession Numbers d Wnt signaling induces primitive streak (PS) expression GSE58476 in hESCs d b-catenin binds upstream regulatory regions in most PS and neural crest genes d Direct interactions between b-catenin and SMAD2/SMAD3 mediate PS gene activation d OCT4 binds in proximity to b-catenin and SMAD2/SMAD3 sites to induce PS genes

Funa et al., 2015, Cell Stem Cell 16, 639–652 June 4, 2015 ª2015 Elsevier Inc. http://dx.doi.org/10.1016/j.stem.2015.03.008 Cell Stem Cell Article b-Catenin Regulates Primitive Streak Induction through Collaborative Interactions with SMAD2/SMAD3 and OCT4

Nina S. Funa,1,2,5 Karen A. Schachter,1,5 Mads Lerdrup,3 Jenny Ekberg,2 Katja Hess,1,2 Nikolaj Dietrich,3 Christian Honore´ ,4 Klaus Hansen,3 and Henrik Semb1,2,* 1The Danish Stem Cell Centre (DanStem), University of Copenhagen, 2200 Copenhagen N, Denmark 2Lund Stem Cell Center, Faculty of Medicine, Lund University, 22100 Lund, Sweden 3Biotech Research and Innovation Centre (BRIC) and Centre for Epigenetics, University of Copenhagen, Copenhagen Biocenter, 2200 Copenhagen N, Denmark 4Department of Stem Cell Biology, Novo Nordisk A/S, 2760 Maaloev, Denmark 5Co-first author *Correspondence: [email protected] http://dx.doi.org/10.1016/j.stem.2015.03.008

SUMMARY various signaling pathways, including Wnt, Nodal, and BMP (bone morphogenetic ). It is well known that the trans- Canonical Wnt and Nodal signaling are both required forming growth factor (TGF)-b family member Nodal promotes for induction of the primitive streak (PS), which anterior-posterior axis patterning in the gastrulating mouse guides organization of the early . The Wnt embryo; high levels promote anterior fates, whereas a graded effector b-catenin is thought to function in these response to lower levels allows regional specification toward early lineage specification decisions via transcrip- different mesodermal derivatives (Dunn et al., 2004; Kinder tional activation of Nodal signaling. Here, we demon- et al., 1999). Wnt3 helps maintain high levels of Nodal expression through direct regulation of the proximal in the strate a broader role for b-catenin in PS formation by Nodal locus (Norris and Robertson, 1999; Vincent et al., 2003). analyzing its genome-wide binding in a human em- Nodal mutant in which the proximal epiblast enhancer b bryonic stem cell model of PS induction. -catenin element has been deleted lack anterior structures as a result of a occupies regulatory regions in numerous PS and failure in specification of definitive (DE), which also neural crest genes, and direct interactions between works as a signaling modulator for the anterior b-catenin and the Nodal effectors SMAD2/SMAD3 (Norris and Robertson, 1999; Vincent et al., 2003). In addition, are required at these regions for PS gene activation. mice deficient in activators of the , such Furthermore, OCT4 binding in proximity to these as b-catenin, the Wnt co-receptors Lrp5/Lrp6, and the processor sites is likewise required for PS induction, suggesting of the Wnt molecule Porcupine fail to form PS and a collaborative interaction between b-catenin and (Biechele et al., 2011; Fu et al., 2009; Hsieh et al., 2003; Huelsken OCT4. Induction of neural crest genes by b-catenin et al., 2000; Kelly et al., 2004; Liu et al., 1999). Re-examination of an epiblast-specific Wnt3 mutant mouse shows that PS is is repressed by SMAD2/SMAD3, ensuring proper induced, but not maintained, in these animals, highlighting the lineage specification. This study provides mecha- need of sustained Wnt signaling in the epiblast for nistic insight into how Wnt signaling controls early (Tortelote et al., 2013). Thus, canonical Wnt and Nodal signaling cell lineage decisions. are required for PS formation. However, whether b-catenin is simply required for regulating Nodal expression or targets several key transcriptional regulators of PS is unclear. INTRODUCTION Whereas Wnt and Nodal/TGFb signaling work together in the regulation of PS formation, they play opposite roles during neural During gastrulation, induction of the three germ layers is regu- induction. High levels of Wnt signaling promote the expression of lated by signaling cues from the emerging primitive streak (PS). Gbx2, Pax3, and Tfap2a, which are essential for specification of PS formation is spatially initiated at the prospective posterior posterior neural fates such as neural crest progenitor (NCp) cells end of the mouse embryo and is characterized by regionalized (Kiecker and Niehrs, 2001; Li et al., 2009). In contrast, TGFb expression of lineage-specific markers (Tam et al., 2006). signaling inhibits neural induction (Hemmati-Brivanlou and Through an epithelial-mesenchymal transition, epiblast cells Melton, 1994; Surmacz et al., 2012). In fact, dual blockage of that will give rise to endoderm and mesoderm lineages invagi- BMP and NODAL signaling has been used to direct differentia- nate through the streak and migrate anteriorly, while cells re- tion of human embryonic stem cells (hESCs) into neuroectoderm maining in the epiblast give rise to surface and neural and neural crest progenitors in vitro (Menendez et al., 2011). ectoderm (Tam and Behringer, 1997). Here, we study how b-catenin regulates differentiation of Early stages of are specified both hESCs. Initial activation of Wnt signaling induces expression of temporally and spatially through the coordinated activities of direct targets of b-catenin, among them markers of NCp.

Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. 639 (legend on next page)

640 Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. b-catenin’s activation also promotes a sustained increase in and S1B). To validate the functionality of the CHIR-induced PS, NODAL levels and SMAD2/SMAD3 activity, resulting in downre- we followed the conversion of PS into DE by adding ACTIVIN A gulation of NCp and induction of PS genes. (AA) for 2 days after the 24-hr CHIR treatment (Figure 1A). One- Genome-wide analysis of b-catenin binding sites revealed direct day treatment with AA after CHIR removal was sufficient for effi- binding of b-catenin to regulatory regions in both PS and NCp cient induction of DE markers, such as SOX17, FOXA2, CXCR4, genes. Transcriptional regulation of PS genes requires binding and (CER)(Figure S1F). Approximately 95% of the of b-catenin, OCT4, and SMAD2/SMAD3 to these regulatory cells expressed SOX17 at the end of the CHIR-AA 3-day protocol regions; removal of OCT4 or suppression of NODAL signaling (Figure S1C). Notably, the expression levels of SOX17 and prevents b-catenin-mediated PS induction. Subsequent differ- CXCR4 on day 3 were much higher in cells initially treated with entiation into DE is also blocked, while expression of NCp CHIR, compared to high WNT3A or WNT3A+AA (Figures S1D markers is restored. We also show that it is the SMAD2/ and S1E), suggesting a relationship between high initial b-catenin SMAD3 activation levels that determine cell fate choices beyond activity and high DE induction. These results were reproduced in the PS state. Our findings provide mechanistic information on an independent hESC culture system (Figures S2A–S2C). Finally, how hESCs coordinate inputs from the Wnt and Nodal pathways as a sign of the authenticity of the putative DE cells, we show that to regulate early lineage-specific differentiation. they are capable of differentiating into multipotent pancreatic progenitors (75% PDX1+/NKX6.1+ cells; Figure S1G) (Hald RESULTS et al., 2008, Jennings et al., 2013). Binding of NODAL to the ALK4/ALK7 receptors leads to phos- Canonical Wnt Signaling Regulates NODAL Expression phorylation, activation, and nuclear translocation of SMAD2/ and PS Specification SMAD3—a prerequisite for transcriptional regulation of its target To study how activation of Wnt signaling influences PS induction genes. As expected, CHIR-mediated increase in NODAL expres- in hESCs, we used CHIR99021 (CHIR), an inhibitor of glycogen sion led to increased -mediated phosphorylation as well synthase kinase (GSK-3), for 24 hr (day 1) (Figure 1A). Canonical as nuclear localization of SMAD2/SMAD3 (Figures 1G and 1H). Wnt signaling requires the stabilization and activation of b-cate- Addition of the ALK4/ALK5/ALK7 inhibitor SB431542 (SB) nin in order to induce target gene expression. The stability of completely blocked endogenous NODAL signaling as early as b-catenin is regulated through phosphorylation by glycogen syn- 3 hr after treatment as well as CHIR-induced NODAL expression, thase kinase 3 (GSK-3). Upon engagement of Wnt signaling, SMAD2/SMAD3 activation, and PS gene expression (Figures 1D– GSK-3 is inactivated, allowing unphosphorylated b-catenin to 1H). It has been reported that GSK-3b regulates SMAD3 protein accumulate, translocate to the nucleus, and activate transcrip- stability through phosphorylation-induced proteosomal degra- tion by associating with the /lymphoid dation (Guo et al., 2008). CHIR-induced NODAL signaling may enhancer-binding factor (TCF/LEF) transcription factors (Aberle therefore result from direct modulation of SMAD2 activity rather et al., 1997; Novak and Dedhar, 1999). Although GSK-3 pos- than through b-catenin activation. However, we observed no sesses multiple substrates (Forde and Dale, 2007), inhibition change in SMAD2 protein levels in CHIR-treated cells (Figure 1H). with CHIR has consistently been used as a surrogate for Wnt In support of a direct role of b-catenin downstream of CHIR, we signaling to promote b-catenin activation in the context of observed that siRNA-mediated knockdown of the expression of hESC differentiation (Kunisada et al., 2012; Naujok et al., 2014; either b-catenin (siR b-cat) or SMAD2 (siR SMAD2) dramatically Pagliuca et al., 2014; Rezania et al., 2014; Tan et al., 2013). We reduced CHIR-induced expression of NODAL, MIXL1, EOMES, corroborated these findings by showing that CHIR induced lucif- and T (Figure 1I). Neither b-catenin nor SMAD2 knockdown erase expression from a reporter containing TCF binding sites affected each other’s expression levels, or that of the pluripo- (TOPFlash) as well as expression of the direct b-catenin targets tency markers OCT4, NANOG, and (Figure S2E), indicating LEF1 and AXIN2 (Figures 1B–1C). that the cells remained pluripotent before CHIR treatment. Since Interestingly, time-course studies showed that CHIR treatment b-catenin knockdown decreased CHIR-induced SMAD2 nuclear induced expression of NODAL, rapidly followed by induction of localization without affecting SMAD2 expression (Figures S2D– its downstream target and PS markers (T), S2F), our data altogether has uncovered an unanticipated indis- EOMES, MIXL1, and GOOSECOID (GSC)(Figures 1D–1F, S1A, pensable role of b-catenin in the regulation of PS induction.

Figure 1. WNT and NODAL Signaling Regulate PS Formation (A) Diagram describing the differentiation protocol from hESC to DE. (B) Cells transduced with a FopFlash- or TopFlash-luciferase reporter were treated as indicated. Experiments were performed in triplicate, and average values are presented. (C) Expression of b-catenin targets AXIN2 and LEF1 was analyzed at different time points after CHIR treatment and normalized to undifferentiated cells (hES). (D and E) Cells treated for 24 hr as indicated. Gene expression for NODAL, LEFTY (D), T, MIXL1, and EOMES (E) was analyzed at different time points by qPCR and normalized to undifferentiated cells (hES). (F and G) Cells were treated for 24 hr as indicated and immunostained for T (F) and SMAD2 (G). Scale bar, 25 mm. (H) Cells were treated as indicated, and lysates were analyzed by immunoblot. Data are representative of three independent experiments. (I) Cells transfected with siRNA pools for b-catenin, SMAD2, or a scrambled control (scrb) were treated with CHIR for 24 hr. Gene expression was analyzed by qPCR. Data were normalized to CHIR-treated scrb-transfected cells. (J) Cells transfected with siRNA pools for b-catenin or a scrambled control (scrb) were treated with RPMI or 100 ng/ml AA for 3 days and analyzed as described above. Data in this figure are presented as mean ± SEM of at least three independent experiments. *p < 0.05, **p < 0.005, and ***p < 0.0005. See also Figures S1 and S2.

Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. 641 Previous studies in mice and pluripotent stem cells suggest CHIR+SB (Figure 2B; Table S1), and analysis that the main role of b-catenin in PS induction is to regulate Nodal revealed that both CHIR and CHIR+SB induced binding of b-cat- expression (Ben-Haim et al., 2006; Dahle et al., 2010; Liu et al., enin in the proximity of a common set of genes involved in 1999; Singh et al., 2012). If b-catenin’s role is limited to the induc- biological processes such as pattern specification, mesoderm tion of Nodal signaling, then replacing CHIR with AA during the development, and gastrulation (Figure 2C; Table S2). first 24 hr should induce PS gene expression equally efficiently. To validate binding of b-catenin to NODAL, EOMES, and Although expression of the direct AA target LEFTY was readily MIXL1, ChIP-qPCR analysis was carried out. Specifically, anal- induced after 24 hr of AA treatment, we detected only a modest ysis of selected regulatory regions, as determined in our ChIP- induction of NODAL, T, EOMES, and MIXL1 expression in com- seq (from transcription start site [TSS]: NODAL, À13 kb; EOMES, parison to CHIR (Figures 1D and 1E). Time-course analysis of AA À6 kb; and MIXL1, À13 kb), demonstrated binding of b-catenin treatment showed a more substantial increase in the expression to all three regions upon 19 hr treatment with CHIR (Figure 2D). of the PS markers MIXL1, EOMES, and GSC starting on day 2, In confirmation of the results obtained in the ChIP-seq, b-catenin while expression of SOX17 was detected from day 3 (Figure 1J). binding was maintained in cells treated for 19 hr with CHIR+SB Interestingly, expression of siR-b-cat blocked AA-induced ex- (Figure 2D). As NODAL signaling and SMAD2/SMAD3 activity pression of these markers without affecting expression of LEFTY are completely abrogated under these conditions, the data indi- (Figure 1J), indicating that b-catenin is necessary for AA-induced cate that the binding of b-catenin occurs directly in response to differentiation. Wnt/b-catenin signaling also regulates expres- CHIR treatment and is independent of NODAL/SMAD2/SMAD3 sion of the NODAL co-receptor CRIPTO (Morkel et al., 2003), signaling. which is essential for transduction of Nodal signaling to SMAD2/SMAD3 (Gritsman et al., 1999; Yeo and Whitman, b-Catenin and SMAD2/SMAD3 Bind to the Same 2001). However, induction of CRIPTO expression was not de- Regulatory Regions in PS Genes tected in cells treated with CHIR or AA (Figure S2G). These Based on the observations that b-catenin-mediated induction of data further substantiate the idea that b-catenin is required for PS gene expression depends on binding of b-catenin to PS PS gene induction, and, essentially, they indicate that b-cate- genes and on SMAD2/SMAD3 activity, we hypothesized that nin’s role goes beyond the mere induction of NODAL expression b-catenin, in addition to increasing SMAD2/SMAD3 activation and signaling. Consistent with this new concept was the effect of through NODAL expression, more directly collaborates with reducing expression of NODAL with siRNAs. Even though siR- SMAD2/SMAD3 at the chromatin level to regulate PS gene acti- NODAL compromised CHIR-mediated PS and DE induction vation. Indeed, analysis of SMAD2/SMAD3 occupancy at the (Figure S2H), the response was limited compared to siR-b-cat. regulatory regions bound by b-catenin showed a CHIR-depen- Altogether, our results not only support the previously demon- dent induction in the binding of SMAD2/SMAD3 to these sites strated role of the canonical Wnt pathway as a key regulator of (Figure 3A). As expected, binding was reduced in cells treated NODAL activity (Barrow et al., 2007; Brennan et al., 2001; Morkel with CHIR+SB. Interestingly, binding of SMAD2/SMAD3 to the et al., 2003) but more importantly highlight an unanticipated cell- b-catenin binding regions in NODAL, EOMES, and MIXL1 in un- autonomous role of canonical Wnt signaling in the regulation of differentiated hESCs was previously demonstrated by SMAD2/ PS induction. SMAD3 ChIP-seq (Kim et al., 2011). It was also shown that SMAD2/SMAD3 is recruited to different SMAD binding elements b-Catenin Binds Directly to Regulatory Regions in PS (SBEs) upon AA-induced differentiation. We analyzed whether Genes CHIR-induced NODAL signaling promoted binding of SMAD2/ To obtain a global view of b-catenin’s transcriptional targets dur- SMAD3 to the reported AA-responsive SBEs in NODAL, ing early cell fate specification, chromatin immunoprecipitation MIXL1, and GSC (from TSS: NODAL, +0.9 kb; MIXL1, TSS; followed by high-throughput sequencing (ChIP-seq) was per- GSC, À5.5 kb) (Kim et al., 2011). Indeed, we show that CHIR formed in differentiating hESCs. To catch the early changes in treatment induced SMAD2/SMAD3 binding to these SBEs and genome occupancy by b-catenin, ChIP-seq was performed that addition of SB blocked these binding events (Figure 3B). 6 hr after treatment with CHIR or CHIR+SB and compared to Importantly, knocking down b-catenin expression dramatically mock-treated cells. As expected, CHIR treatment resulted in reduced CHIR-induced binding of SMAD2/SMAD3 to both the an increase in the number of b-catenin’s binding sites compared b-catenin- and SMAD-binding elements (Figures 3A and 3B). to control cells (Figure 2A; Table S1). Global analysis of the new Co-immunoprecipitation analysis showed that b-catenin and binding events showed that most of them occurred in upstream SMAD2/SMAD3 form a complex, which is further stimulated and intergenic regions. Importantly, analysis of the ChIP-seq upon CHIR treatment (Figure 3C). To determine whether this data suggests that b-catenin might have a wider impact in regu- interaction translates into a simultaneous occupancy of binding lating PS induction than previously anticipated. Earlier studies sites, ChIP for b-catenin was performed followed by re-ChIP have implicated b-catenin in the regulation of Nodal and T for SMAD2/SMAD3. Indeed, we detected co-occupancy of expression through direct binding to their regulatory regions (Ar- b-catenin and SMAD2/SMAD3 at the b-catenin binding region nold et al., 2000; Ben-Haim et al., 2006; Yamaguchi et al., 1999). on MIXL1 upon CHIR treatment (Figure 3D). As expected, in cells Here, we confirm these results in hESCs and show that b-catenin treated with CHIR+SB the co-occupancy is lost because binds to upstream regulatory regions in most previously identi- SMAD2/SMAD3 binding is prevented. These results support a fied PS genes, including NODAL, T, EOMES, MIXL1, GSC, new concept for how b-catenin controls PS initiation: the forma- WNT3, , and GATA4 (Table S1). Notably, b-catenin binding tion of a complex with SMAD2/SMAD3, and the binding of this in PS gene proximal regions is maintained in cells treated with complex to specific regulatory regions on PS genes allows

642 Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. Figure 2. b-Catenin Binds Directly to Regulatory Regions in the PS Genes Cells treated with CHIR or CHIR+SB or mock treated (RPMI) for 6 hr were subjected to ChIP-seq analysis for b-catenin. (A) Genomic distribution of binding sites as a percentage of the total peaks obtained in each condition. (B) University of California, Santa Cruz (UCSC) tracks were generated for EOMES, MIXL1, and NODAL from the b-catenin ChIP-seq. (C) Gene Ontology analysis for processes in which b-catenin-bound genes are enriched in cells treated with CHIR and CHIR+SB. (D) Cells treated as indicated for 19 hr were subjected to ChIP with a b-catenin and isotype IgG control antibody. Enrichment at the indicated b-catenin binding sites in NODAL, EOMES, and MIXL1 was analyzed by qPCR and normalized to input. Data are presented as fold enrichment relative to an unbound control region.

Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. 643 Figure 3. b-Catenin-Driven Expression of PS Genes Is Dependent on SMAD2/SMAD3 (A and B) Cells treated as indicated for 19 hr were subjected to ChIP with a SMAD2/SMAD3 and isotype IgG control antibody. Cells transfected with siRNA pools for b-catenin were treated with CHIR 48 hr after transfection and processed as the rest. Enrichment at the indicated b-catenin binding sites in NODAL, EOMES, and MIXL1 (A) and at the indicated SBE sites (B) was analyzed by qPCR and normalized to input. Data are presented as fold enrichment relative to an unbound control region. (C) Cells were treated as indicated. Cellular lysates were immunoprecipitated with a SMAD2 or IgG control antibody. The presence of b-catenin, pSMAD2, and total SMAD2 in the precipitates was analyzed by western blot. Input levels are shown as controls on the right side. The data are representative of three inde- pendent experiments. (D) Cells were treated for 19 hr as indicated. b-catenin ChIP was performed followed by ChIP with SMAD2/SMAD3 or isotype IgG control antibody. Enrichment at the MIXL1 À13 kb locus was determined by qPCR and normalized to input. Data are presented as fold enrichment relative to an unbound control region. Data in this figure are presented as mean ± SEM of at least three independent experiments. *p < 0.05, **p < 0.005, and ***p < 0.0005. b-catenin to reinforce NODAL expression and promote PS Menendez et al., 2011; Patani et al., 2009). Our ChIP-seq data commitment. showed CHIR-induced binding of b-catenin to regulatory regions in GBX2, TFAP2A, MSX1, and PAX3, which are expressed in b-Catenin Promotes Expression of Neural Crest cells at the border between neural and nonneural ectoderm as Progenitor Genes in the Absence of SMAD2/SMAD3 well as in early NCp (Figure 4A; Table S1)(Li et al., 2009; Mon- Activation soro-Burq et al., 2005; Plouhinec et al., 2014). In fact, it has The Gene Ontology analysis of the ChIP-seq data also revealed been well documented that Wnt signaling plays a critical role in binding of b-catenin to genes involved in neural development neural crest induction (reviewed in Bae and Saint-Jeannet, and neural crest cell differentiation and that these binding events 2014). Consistent with this observation we found that CHIR treat- were enriched in response to blockage of SMAD2/SMAD3 ment induced early expression (3 hr) of TFAP2A, MSX1, GBX2, signaling (CHIR+SB; Figure 2C). This finding is in line with previ- ZIC1, and PAX3. However, the expression of these genes dimin- ous observations that CHIR+SB treatment induces differentia- ished with time (Figure 4B). The decline in expression of these tion of hESCs toward neural crest cell fates (Chng et al., 2010; NCp genes coincided with the induction of PS gene expression,

644 Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. Figure 4. Suppression of NODAL Signaling Allows WNT-Induced NCp Fate (A) UCSC tracks were generated for GBX2, TFAP2A, and MSX1 from the b-catenin ChIP-seq. (B) Cells were treated as indicated, and gene expression was analyzed at different time points by qPCR. Data were normalized to undifferentiated cells (hES). (C and D) Cells transfected with siRNA pools for b-catenin, SMAD2, NODAL, or a scrambled control (scrb) were treated with CHIR for 24 hr. Gene expression was analyzed by qPCR. Data were normalized to CHIR-treated scrb-transfected cells. (E) Cells were treated as indicated. Gene expression was analyzed at different time points by qPCR. Data were normalized to undifferentiated cells (hES). (F) Cells were treated as indicated for 24 hr and immunostained for T and TFAP2A. Scale bar, 50 mm. suggesting that b-catenin-induced Nodal/SMAD2/SMAD3 sig- (18 to 24 hr after CHIR treatment) (Figure 1E) underscores the naling may block neural crest specification. Indeed, while NCp importance of b-catenin-mediated induction of Nodal signaling gene expression declined upon reduction in b-catenin expres- as a key inhibitory signal for NCp induction. sion levels (Figure 4C, see inset for TFAP2A), knocking down expression of SMAD2 or NODAL promoted NCp gene expres- OCT4 Co-Occupies b-Catenin Binding Regions to sion in response to CHIR treatment (Figures 4C and 4D). Similar Regulate PS Gene Expression results were obtained by pharmacological blockage of SMAD2/ Although b-catenin-induced transcription is mainly mediated SMAD3 activation (CHIR+SB) (Figures 4E and 4F). The fact that through its interaction with the TCF/LEF , there are re- blockage of SMAD2/SMAD3 activity results in increased expres- ports suggesting that b-catenin associates with a variety of sion of NCp genes only at the time when Nodal signaling is active different transcription factors (reviewed in Valenta et al., 2012).

Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. 645 Figure 5. OCT4 Promotes PS Gene Induction (A) Putative TCF/LEF and OCT4 binding sequences identified by motif enrichment analysis in the b-catenin peaks. (B) Histograms showing the distribution of distances from each b-catenin peak (left histogram) or randomized control region (right histogram) to the nearest peak in previously published OCT4 ChIP-seq data (GSM539547). (C) Cells were treated with CHIR for 19 hr and subjected to ChIP with an OCT4 or isotype IgG control antibody. Enrichment at the indicated b-catenin binding sites in NODAL, EOMES, and MIXL1 was analyzed by qPCR and normalized to input. Data are presented as fold enrichment relative to an unbound control region. (D) Cells were treated for 19 hr as indicated. b-catenin ChIP was performed followed by ChIP with OCT4 or isotype IgG control antibody. Enrichment at the indicated loci was determined by qPCR and normalized to input. Data are presented as fold enrichment relative to an unbound control region. (E) Cells were transfected with siRNA pools for OCT4 or a scrambled control (scrb). Twelve hours after transfection, cells were treated with RPMI or CHIR for 24 hr, and lysates were analyzed by immunoblot. (F) Cells were transfected as in (E) and treated for 24 hr as indicated. Expression of markers for PS was analyzed by qPCR and normalized to scrb-transfected control cells. Data in this figure are presented as mean ± SEM of at least three independent experiments. *p < 0.05, **p < 0.005, and ***p < 0.0005.

To identify b-catenin’s binding partners during transcriptional OCT4 in PS induction has also been documented (DeVeale regulation of PS genes, we performed de novo binding motif et al., 2013; Thomson et al., 2011; Trott and Martinez Arias, analysis on the b-catenin ChIP-seq peaks. We found that similar 2013; Wang et al., 2012); however, the mechanism by which motifs were enriched under both CHIR and CHIR+SB conditions, OCT4 regulates PS gene expression is still not fully understood. including a putative TCF/LEF motif (Figure 5A). Surprisingly, a To elucidate whether b-catenin and OCT4 could co-regulate motif resembling a previously described OCT4 binding motif early hESC differentiation, we compared their genome-wide oc- (Mullen et al., 2011) was also enriched in both conditions (Fig- cupancy during early lineage specification. Because no ChIP- ure 5A), suggesting that b-catenin and OCT4 may transcription- seq data are available for OCT4 in differentiating hESCs, we ally coordinate early developmental programs in hESCs. A direct compared the b-catenin ChIP-seq data from CHIR-treated interaction between b-catenin and OCT4 has previously been re- hESCs (6 hr) with OCT4 ChIP-seq data obtained from undifferen- ported in mouse embryonic stem cells (mESCs) under pluripo- tiated hESCs (Mullen et al., 2011). We found that for 35.9% of the tency conditions (Faunes et al., 2013; Kelly et al., 2011; Takao b-catenin peaks (378 out of 1,053), an OCT4 binding region was et al., 2007). Although this interaction promotes OCT4 transcrip- located within a distance of 1 kb of the b-catenin binding site tional activity, it occurs in a TCF/LEF-independent manner and (Figure 5B). To test that this was not an indirect consequence does not require b-catenin’s transcriptional activity (Faunes of both factors randomly occurring in the proximity of genes, et al., 2013; Kelly et al., 2011). Binding of OCT4 to a TCF/LEF- we generated a set of randomized control regions that on a pop- OCT4 composite site has been shown to regulate Mesp1 expres- ulation level matched the b-catenin peaks in respect to TSS dis- sion during differentiation of mESCs (Li et al., 2013). A role for tance and orientation. The vast majority of these control regions

646 Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. were found at distances greater than 10 kb from OCT4 binding with CHIR for 24 hr and then by varying concentrations of AA regions, suggesting that the colocalization of a large part of the on days 2 and 3, ranging from no AA to low AA (10 ng/ml) and b-catenin peaks to OCT4 binding regions was unlikely to occur high AA (50–100 ng/ml). To our surprise, induction of SOX17 by chance (Figure 5B). Similar observations have been made expression was observed in cells exposed to no, low, and high for SMAD3 and OCT4 in undifferentiated hESCs where SMAD3 AA. As expected, SB addition significantly reduced induction and OCT4 co-occupy sites throughout the genome (Mullen of SOX17 (Figures 6A–6D). Remarkably, SOX17 expression et al., 2011). was maintained even after 2 days without AA supplementation. Interestingly, there was considerable overlap in the peaks ob- However, these SOX17+ cells were mainly detected in clusters tained in the b-catenin and OCT4 ChIP-seq for the PS markers in contrast to the homogeneous expression pattern observed EOMES, NODAL, MIXL1, WNT3, and EVX1 (Table S3). These under high-AA exposure (Figure 6D). Similarly, CXCR4 and data suggest that coordinated binding of b-catenin and OCT4 CER were equally induced by low and high doses of AA might regulate expression of PS genes during CHIR-induced dif- (Figure 6A); however, expression was significantly lower in the ferentiation. In fact, ChIP-qPCR experiments demonstrated that absence of exogenous AA. Interestingly, these conditions in cells treated with CHIR, OCT4 binds to the b-catenin regulato- promoted expression of endothelial markers, such as CD31 ry regions in EOMES, NODAL, and MIXL1 (Figure 5C). Sequential (PECAM-1), VE-cad, and RUNX1 (Figure S3A), consistent with ChIP (ChIP for b-catenin followed by re-ChIP for OCT4) showed previous reports demonstrating expression of Sox17 in vascular that b-catenin and OCT4 co-occupy the b-catenin binding re- progenitors in the mouse (Choi et al., 2012; Engert et al., 2013). gions in NODAL and MIXL1 upon CHIR stimulation (Figure 5D). However, markers characteristic for cardiac mesoderm, ecto- Interestingly, in cells treated with CHIR+SB the chromatin co-oc- derm, or primitive endoderm were not significantly upregulated cupancy of OCT4 and b-catenin was reduced, suggesting that under any of these conditions (Figures S3B, S3D, and S3E). efficient binding of these factors to genomic sites might be These results indicate that although endogenous NODAL is suf- dependent on the presence of activated SMAD2/SMAD3. The ficient for differentiation of the PS, increased SMAD2/SMAD3 observation that PS induction requires cooperative binding of activation (sufficient with 10 ng/ml of AA) is necessary for effi- b-catenin, OCT4, and SMAD2/SMAD3 to regulatory regions in cient differentiation of PS into DE. PS genes identifies for the first time a functional link between Although blocking SMAD2/SMAD3 signaling (SB) during days these factors in PS induction. 2 and 3 totally abrogated expression of DE markers, it signifi- To functionally test the role of OCT4 in b-catenin-driven early cantly increased expression of T and NCp genes (Figures 6B, differentiation, OCT4 expression was reduced by expressing 6C, and 6E). The facts that EOMES or MIXL1 expression is OCT4 siRNAs (siR OCT4) (Figures 5E and 5F). Previous work reduced and TBX6 expression is increased (Figures 6B and demonstrated that reduced expression of OCT4 in hESCs results S4C) suggest that the T-expressing cells most likely represent in induction of trophectoderm (TE) markers through BMP4 acti- the transition from PS into a mesodermal fate such as paraxial vation (Wang et al., 2012; Zafarana et al., 2009). However, mesoderm (Tan et al., 2013). In agreement with previous data upon a short 12 hr transfection with OCT4 siRNAs, OCT4 (Menendez et al., 2011), prolonged blockage of SMAD2/ mRNA and protein levels were significantly downregulated, SMAD3 signaling leads to expression of additional NCp markers, without affecting the levels of NANOG and SOX2 (Figure 5E). including SOX9 and SLUG (in addition to TFAP2A, MSX1, and No expression of the TE marker HAND1 could be detected (Fig- PAX3)(Figure 6C). ure 5E), suggesting that the cells remained in a pluripotent state. Altogether, our data show that b-catenin-induced endoge- Reduced expression of OCT4 dramatically inhibited CHIR- nous NODAL signaling is sufficient for its contribution to PS induced NODAL and PS gene expression (Figure 5F), consistent induction but that conversion of PS into DE requires higher con- with the requirement of a b-catenin/OCT4/SMAD2/SMAD3 centrations of NODAL. Furthermore, blocking endogenous complex to promote PS gene expression. Our findings highlight SMAD2/SMAD3 activation redirects PS into mesodermal and the existence of multiple collaborative interactions between neural crest fates. b-catenin and OCT4, which ultimately determine the outcome of b-catenin-mediated transcriptional regulation of early cell DISCUSSION fate specification. To recapitulate the early steps in canonical Wnt-induced cell Fate Determination beyond the PS State Is Regulated by lineage specification in hESCs, we have taken as a reference NODAL Levels point studies performed in vivo where Wnt and Nodal/TGFb/ On the basis of studies in vivo (embryo) and in vitro (ESCs), the BMP cross-regulatory signaling has been described. The results current model for how the PS transitions into DE involves presented here offer new insights into the role of b-catenin in increased SMAD2/SMAD3 activation (Robertson, 2014; Trem- early lineage commitment. Our b-catenin ChIP-seq study is blay et al., 2000). To mimic this step in vitro, most protocols de- unique in that it provides for the first time a genome-wide view signed to induce PS and DE from pluripotent stem cells use very of the extent of b-catenin involvement in early hESC differentia- high concentrations of AA (50–100 ng/ml). Our results indicate tion, revealing a direct regulation of both PS and NCp genes. Our that activation of b-catenin is sufficient to induce PS via study shows that b-catenin’s role in PS induction is not merely to increased NODAL production leading to increased SMAD2/ activate SMAD2/SMAD3 through regulation of NODAL expres- SMAD3 activation. This raises the interesting question whether sion; efficient PS induction relies on the formation of a physical the endogenously produced NODAL is sufficient to differentiate complex with SMAD2/SMAD3 and binding in proximity to PS progenitors into DE. To test this idea, hESCs were treated OCT4 on chromatin at promoter proximal regions of PS genes.

Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. 647 (legend on next page)

648 Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. Figure 7. Mechanism for the Regulation of Wnt-Induced Early Lineage Choices Under pluripotency conditions, OCT4 and SMAD2 co-occupy upstream regulatory regions on PS genes. CHIR treatment promotes stabilization b-catenin, which is recruited to regulatory regions in PS and NCp genes. Initially, activation of Wnt signaling induces expression of NCp. Binding of b-catenin, SMAD2/SMAD3, and OCT4 to chro- matin promotes a sustained increase in NODAL levels and SMAD2/SMAD3 activity, resulting in downregulation of NCp gene expression and in- duction of PS genes. See text for more details.

Our data indicate that b-catenin-mediated enhancement of represses NCp expression through its induction of SMAD2/ NODAL signaling promotes expression of PS genes while block- SMAD3 activity via NODAL expression. ing expression of NCp genes (Figure 7). We also demonstrate that the concentration of exogenously We show for the first time that b-catenin binds to regulatory added AA determines whether the CHIR-induced PS popula- regions in a majority of PS gene loci. However, binding of tion will differentiate into DE (SOX17+), b-catenin alone is not sufficient to initiate transcription and re- (T+), or neural crest (TFAP2A+). The potential of PS cells to quires that a complex with SMAD2/SMAD3 is formed and binds differentiate into multiple lineages may be explained by the ex- together with OCT4 on chromatin. This mode of action prob- istence of various lineage-specific progenitors or multipotent ably explains the delay in PS gene induction observed when progenitors. In support of the former, we noted that CHIR treat- only exogenously added AA is used to initiate differentiation. ment resulted in variable levels of T expression (Figure 1G), Most likely, a gradual accumulation of Wnt ligands in response which is consistent with in vivo studies showing variable T pro- to prolonged AA treatment leads to the required binding of tein levels in individual PS cells of wild-type E7.5 embryos. b-catenin and activated SMAD2/SMAD3 to the regulatory re- Whereas cells with lower T levels tend to differentiate into neu- gions in PS genes. ral , those expressing high levels exit rapidly from the b-catenin also binds to regulatory regions in NCp genes, but streak to form mesoderm (Wilson and Beddington, 1997). expression of these genes can be sustained only if NODAL These results are also consistent with recent studies in epiblast signaling is inhibited. The precise mechanism by which this oc- stem cells (EpiSCs) and mESCs, showing that Wnt signaling in- curs is not completely clear. We failed to detect binding of duces differentiation of a heterogeneous population of T+ PS SMAD2/SMAD3 to the b-catenin regulatory regions of TFAP2A cells with different potential to contribute to endoderm or neu- and MSX1 in response to CHIR (data not shown), suggesting romesoderm lineages (Tsakiridis et al., 2014). Along the same that direct binding of SMAD2/SMAD3 to these regions is unlikely lines, a recent report suggests that human PS progenitors to mediate regulation of their expression. Furthermore, a screen possess distinct identities that restrict their potential to differ- through reported genome-wide SMAD3 binding regions in undif- entiate into mesodermal derivatives (Mendjan et al., 2014). ferentiated hESCs showed that SMAD3 is absent from the Alternatively, CHIR treatment may induce differentiation of mul- regulatory regions on TFAP2A and GBX2 (Mullen et al., 2011). tipotent progenitors. In support of this option, clonal analysis in Similarly, SMAD2 ChIP-seq datasets from undifferentiated cells the mouse embryo demonstrated that although endoderm and also report lack of SMAD2 binding to the b-catenin sites in GBX2, segregate during gastrulation, neural and MSX1 (Brown et al., 2011; Kim et al., 2011), and TFAP2A (Brown mesodermal lineages share a common bipotent progenitor et al., 2011). Thus, it seems unlikely that SMAD2/SMAD3 re- (Tzouanacou et al., 2009). Although our data are consistent presses NCp induction through directly binding to NCp genes. with the idea that CHIR treatment generates PS-like cells that Interestingly, NANOG knockdown increases expression of retain a certain degree of plasticity toward different lineages, GBX2 and PAX3 in hESCs (Wang et al., 2012). Since NANOG future studies will have to elucidate the existence of lineage- expression is regulated by SMAD2/SMAD3 activity (Vallier specific or bipotent progenitors for human PS and neural crest et al., 2009; Xu et al., 2008), the boost in NCp gene expression lineages. detected upon CHIR+SB treatment might reflect the combined actions of a direct inducer (b-catenin) and the removal of a nega- EXPERIMENTAL PROCEDURES tive regulator (NANOG). b In summary, we show that -catenin regulates NCp gene Cell Culture and Differentiation expression through multiple mechanisms. First, b-catenin SA121 hESCs were grown in DEF-CS media (Takara). When indicated, cells directly induces NCp expression. Second, b-catenin indirectly were cultured in Matrigel (BD Biosciences) with mTeSR1 media (

Figure 6. Graded NODAL Signaling Modulates Fate of PS-like Cells (A–C) Cells were treated with CHIR for 24 hr, followed by various concentrations of AA or SB for an additional 48 hr as indicated. Gene expression for markers of DE (A), PS (B), and NCp (C) was analyzed by qPCR and normalized to D1 CHIR-treated cells. (D and E) Cells were treated as indicated and co-immunostained for T/SOX17 (D) and T/TFAP2A (E). Scale bars, 100 mm (left) and 50 mm (right). Data in this figure are presented as mean ± SEM of at least three independent experiments.**p < 0.005 and ***p < 0.0005. See also Figure S3.

Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. 649 Technologies). hESC cultures were treated with 3 mM CHIR99021 (Stemgent) C.H. is an employee and shareholder of Novo Nordisk A/S, and K.H. is a in RPMI-1640. Control cells were left untreated or treated with 100 ng/ml Acti- cofounder of the biotech company EpiTherapeutics ApS. vin A (Peprotech) and/or 200 ng/ml Wnt3a (R&D Systems), unless otherwise indicated. After 24 hr, cells were treated with 100 ng/ml AA in RPMI for Received: August 14, 2014 24 hr, followed by AA+2%B27 for 1 to 2 days. When applicable, 10 mM Revised: February 3, 2015 SB431542 (Sigma) was added. PE differentiation was performed as previously Accepted: March 17, 2015 described (Rezania et al., 2011). For AA-induced differentiation, cells were Published: April 23, 2015 treated with 100 ng/ml AA in RPMI+2%B27 for 3 days. REFERENCES RNA Extraction and qRT-PCR Total RNA was extracted with the RNeasy Plus Mini kit (QIAGEN) and reverse- Aberle, H., Bauer, A., Stappert, J., Kispert, A., and Kemler, R. (1997). beta-cat- transcribed using the SuperScript III First-Strand synthesis kit (Invitrogen). The enin is a target for the ubiquitin-proteasome pathway. EMBO J. 16, 3797– TGFb signaling qPCR array was from SA Biosciences. qRT-PCR experiments 3804. were performed using the StepOnePlus system (Applied Biosystems) and Arnold, S.J., Stappert, J., Bauer, A., Kispert, A., Herrmann, B.G., and Kemler, SYBR-Green Master Mix (Invitrogen). Primers utilized in this study are available R. (2000). Brachyury is a target gene of the Wnt/beta-catenin signaling in the Supplemental Information. pathway. Mech. Dev. 91, 249–258. Bae, C.-J., and Saint-Jeannet, J.-P. (2014). Induction and specification of neu- siRNA Silencing ral crest cells: Extracellular signals and transcriptional switches. In Neural hESCs were transfected with siRNA-pools (SMAD2 and b-catenin, ON- Crest Cells: Evolution, Development, and Disease, P. Trainor, ed. (Academic TARGET plus SMARTpool, Dharmacon; NODAL, Dharmacon/Ambion; and Press), pp. 27–49. OCT4, Ambion) or scrambled siRNA control (Ambion) using Lipofectamine RNAiMAX (Invitrogen) in complete media. Final concentration was 25– Barrow, J.R., Howell, W.D., Rule, M., Hayashi, S., Thomas, K.R., Capecchi, 50 nM. Cells were treated 48 hr after transfection. With siR-OCT4, experiments M.R., and McMahon, A.P. (2007). Wnt3 signaling in the epiblast is required were initiated 12 hr after transfection. for proper orientation of the anteroposterior axis. Dev. Biol. 312, 312–320. Ben-Haim, N., Lu, C., Guzman-Ayala, M., Pescatore, L., Mesnard, D., ChIP Sequencing and Data Processing Bischofberger, M., Naef, F., Robertson, E.J., and Constam, D.B. (2006). The ChIP was performed as described previously (Dietrich et al., 2012) with slight precursor acting via activin receptors induces mesoderm by maintaining modifications. Antibodies used were b-catenin (R&D Systems, AF1329) and a source of its convertases and BMP4. Dev. Cell 11, 313–323. Goat IgG (ab37373, Abcam). DNA from three parallel ChIPs was pooled, and Biechele, S., Cox, B.J., and Rossant, J. (2011). Porcupine homolog is required 10 ng was used for making ChIP-seq libraries. All peak finding, colocalization, for canonical Wnt signaling and gastrulation in mouse embryos. Dev. Biol. 355, and annotation were done in the program EaSeq (M.L. and K.H., unpublished 275–285. data). ChIP-seq data are available in the GEO database under the accession Brennan, J., Lu, C.C., Norris, D.P., Rodriguez, T.A., Beddington, R.S., and number GEO: GSE58476. Additional information can be found in the Supple- Robertson, E.J. (2001). Nodal signalling in the epiblast patterns the early mental Experimental Procedures. mouse embryo. Nature 411, 965–969. Brown, S., Teo, A., Pauklin, S., Hannan, N., Cho, C.H., Lim, B., Vardy, L., Dunn, ACCESSION NUMBERS N.R., Trotter, M., Pedersen, R., and Vallier, L. (2011). Activin/Nodal signaling controls divergent transcriptional networks in human embryonic stem cells The NCBI GEO accession number for the ChIP-seq data reported in this paper and in endoderm progenitors. Stem Cells 29, 1176–1185. is GEO: GSE58476. Chng, Z., Teo, A., Pedersen, R.A., and Vallier, L. (2010). SIP1 mediates cell-fate decisions between neuroectoderm and mesendoderm in human pluripotent SUPPLEMENTAL INFORMATION stem cells. Cell Stem Cell 6, 59–70.

Supplemental Information includes Supplemental Experimental Procedures, Choi, E., Kraus, M.R., Lemaire, L.A., Yoshimoto, M., Vemula, S., Potter, L.A., three figures, and three tables and can be found with this article online at Manduchi, E., Stoeckert, C.J., Jr., Grapin-Botton, A., and Magnuson, M.A. http://dx.doi.org/10.1016/j.stem.2015.03.008. (2012). Dual lineage-specific expression of Sox17 during mouse embryogen- esis. Stem Cells 30, 2297–2308. AUTHOR CONTRIBUTIONS Dahle, Ø., Kumar, A., and Kuehn, M.R. (2010). Nodal signaling recruits the his- tone demethylase Jmjd3 to counteract polycomb-mediated repression at N.S.F., K.A.S., and H.S. conceived and designed the experiments. N.S.F. per- target genes. Sci. Signal. 3, ra48. formed experiments in Figures 1, 4, 5, 6, and S1–S3. K.A.S. performed exper- DeVeale, B., Brokhman, I., Mohseni, P., Babak, T., Yoon, C., Lin, A., Onishi, K., iments in Figures 1, 2, 3, 5, and S2. J.E. and K.H. contributed to cell culture. Tomilin, A., Pevny, L., Zandstra, P.W., et al. (2013). Oct4 is required E7.5 for N.D. prepared the ChIP-seq library, and M.L. performed the bioinformatic proliferation in the primitive streak. PLoS Genet. 9, e1003957. analysis. C.H. generated the data in Figure S1F. N.S.F., K.A.S., and H.S. wrote Dietrich, N., Lerdrup, M., Landt, E., Agrawal-Singh, S., Bak, M., Tommerup, N., the manuscript. Rappsilber, J., So¨ dersten, E., and Hansen, K. (2012). REST-mediated recruit- ment of polycomb repressor complexes in mammalian cells. PLoS Genet. 8, ACKNOWLEDGMENTS e1002494. Dunn, N.R., Vincent, S.D., Oxburgh, L., Robertson, E.J., and Bikoff, E.K. We thank Q. Jin (Lund Stem Cell Center) and J. Larsen (DanStem) for technical (2004). Combinatorial activities of Smad2 and Smad3 regulate mesoderm for- assistance, S. Agrawal-Singh (BRIC) for technical advice regarding the ChIP mation and patterning in the mouse embryo. Development 131, 1717–1728. protocol, and W. Sukparangsi (DanStem) for assistance in graphical design. We are grateful to J. Brickman, P. Serup, A. Martı´nez-Arias, and O. Lieven Engert, S., Burtscher, I., Liao, W.P., Dulev, S., Schotta, G., and Lickert, H. b for comments on the manuscript. This work was supported by Novo Nordisk (2013). Wnt/ -catenin signalling regulates Sox17 expression and is essential A/S; the Novo Nordisk Foundation, Section for Basic Stem Cell Biology, Dan- for organizer and endoderm formation in the mouse. Development 140, Stem, University of Copenhagen, Denmark; the Danish Strategic Research 3128–3138. Council, DanStem, University of Copenhagen, Denmark; the Stem Cell Center, Faunes, F., Hayward, P., Descalzo, S.M., Chatterjee, S.S., Balayo, T., Trott, J., Lund University, Sweden; and the Swedish Foundation for Strategic Research, Christoforou, A., Ferrer-Vaquer, A., Hadjantonakis, A.K., Dasgupta, R., and Sweden. K.H. was supported by the Danish National Research Foundation. Arias, A.M. (2013). A membrane-associated b-catenin/Oct4 complex

650 Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. correlates with ground-state pluripotency in mouse embryonic stem cells. pluripotent stem cells to multipotent neural crest cells. Proc. Natl. Acad. Sci. Development 140, 1171–1183. USA 108, 19240–19245. Forde, J.E., and Dale, T.C. (2007). Glycogen synthase kinase 3: a key regulator Monsoro-Burq, A.H., Wang, E., and Harland, R. (2005). Msx1 and Pax3 coop- of cellular fate. Cell. Mol. Life Sci. 64, 1930–1944. erate to mediate FGF8 and WNT signals during Xenopus neural crest induc- Fu, J., Jiang, M., Mirando, A.J., Yu, H.M., and Hsu, W. (2009). Reciprocal regu- tion. Dev. Cell 8, 167–178. lation of Wnt and Gpr177/mouse Wntless is required for embryonic axis forma- Morkel, M., Huelsken, J., Wakamiya, M., Ding, J., van de Wetering, M., tion. Proc. Natl. Acad. Sci. USA 106, 18598–18603. Clevers, H., Taketo, M.M., Behringer, R.R., Shen, M.M., and Birchmeier, W. Gritsman, K., Zhang, J., Cheng, S., Heckscher, E., Talbot, W.S., and Schier, (2003). Beta-catenin regulates Cripto- and Wnt3-dependent gene expression A.F. (1999). The EGF-CFC protein one-eyed pinhead is essential for nodal programs in mouse axis and mesoderm formation. Development 130, 6283– signaling. Cell 97, 121–132. 6294. Mullen, A.C., Orlando, D.A., Newman, J.J., Loven, J., Kumar, R.M., Bilodeau, Guo, X., Ramirez, A., Waddell, D.S., Li, Z., Liu, X., and Wang, X.F. (2008). Axin S., Reddy, J., Guenther, M.G., DeKoter, R.P., and Young, R.A. (2011). Master and GSK3- control Smad3 protein stability and modulate TGF- signaling. transcription factors determine cell-type-specific responses to TGF-beta Genes Dev. 22, 106–120. signaling. Cell 147, 565–576. Hald, J., Sprinkel, A.E., Ray, M., Serup, P., Wright, C., and Madsen, O.D. Naujok, O., Lentes, J., Diekmann, U., Davenport, C., and Lenzen, S. (2014). (2008). Generation and characterization of Ptf1a antiserum and localization Cytotoxicity and activation of the Wnt/beta-catenin pathway in mouse embry- of Ptf1a in relation to Nkx6.1 and Pdx1 during the earliest stages of mouse onic stem cells treated with four GSK3 inhibitors. BMC Res. Notes 7, 273. pancreas development. J Histochem Cytochem. 56, 587–595. Norris, D.P., and Robertson, E.J. (1999). Asymmetric and node-specific nodal Hemmati-Brivanlou, A., and Melton, D.A. (1994). Inhibition of activin receptor expression patterns are controlled by two distinct cis-acting regulatory ele- signaling promotes neuralization in Xenopus. Cell 77, 273–281. ments. Genes Dev. 13, 1575–1588. Hsieh, J.C., Lee, L., Zhang, L., Wefer, S., Brown, K., DeRossi, C., Wines, M.E., Novak, A., and Dedhar, S. (1999). Signaling through beta-catenin and Lef/Tcf. Rosenquist, T., and Holdener, B.C. (2003). Mesd encodes an LRP5/6 chap- Cell. Mol. Life Sci. 56, 523–537. erone essential for specification of mouse embryonic polarity. Cell 112, 355–367. Pagliuca, F.W., Millman, J.R., Gu¨ rtler, M., Segel, M., Van Dervort, A., Ryu, J.H., Peterson, Q.P., Greiner, D., and Melton, D.A. (2014). Generation of functional Huelsken, J., Vogel, R., Brinkmann, V., Erdmann, B., Birchmeier, C., and human pancreatic b cells in vitro. Cell 159, 428–439. Birchmeier, W. (2000). Requirement for beta-catenin in anterior-posterior axis formation in mice. J. Cell Biol. 148, 567–578. Patani, R., Compston, A., Puddifoot, C.A., Wyllie, D.J., Hardingham, G.E., Allen, N.D., and Chandran, S. (2009). Activin/Nodal inhibition alone accelerates Jennings, R.E., Berry, A.A., Kirkwood-Wilson, R., Roberts, N.A., Hearn, T., highly efficient neural conversion from human embryonic stem cells and im- Salisbury, R.J., Blaylock, J., Piper Hanley, K., and Hanley, N.A. (2013). poses a caudal positional identity. PLoS ONE 4, e7327. Development of the human pancreas from foregut to endocrine commitment. Diabetes 62, 3514–3522. Plouhinec, J.L., Roche, D.D., Pegoraro, C., Figueiredo, A.L., Maczkowiak, F., Brunet, L.J., Milet, C., Vert, J.P., Pollet, N., Harland, R.M., and Monsoro-Burq, Kelly, O.G., Pinson, K.I., and Skarnes, W.C. (2004). The Wnt co-receptors Lrp5 A.H. (2014). Pax3 and Zic1 trigger the early neural crest gene regulatory and Lrp6 are essential for gastrulation in mice. Development 131, 2803–2815. network by the direct activation of multiple key neural crest specifiers. Dev. Kelly, K.F., Ng, D.Y., Jayakumaran, G., Wood, G.A., Koide, H., and Doble, B.W. Biol. 386, 461–472. b (2011). -catenin enhances Oct-4 activity and reinforces pluripotency through Rezania, A., Riedel, M.J., Wideman, R.D., Karanu, F., Ao, Z., Warnock, G.L., a TCF-independent mechanism. Cell Stem Cell 8, 214–227. and Kieffer, T.J. (2011). Production of functional glucagon-secreting a-cells Kiecker, C., and Niehrs, C. (2001). A gradient of Wnt/beta-catenin from human embryonic stem cells. Diabetes 60, 239–247. signalling regulates anteroposterior neural patterning in Xenopus. Rezania, A., Bruin, J.E., Arora, P., Rubin, A., Batushansky, I., Asadi, A., Development 128, 4189–4201. O’Dwyer, S., Quiskamp, N., Mojibian, M., Albrecht, T., et al. (2014). Reversal Kim, S.W., Yoon, S.J., Chuong, E., Oyolu, C., Wills, A.E., Gupta, R., and Baker, of diabetes with insulin-producing cells derived in vitro from human pluripotent J. (2011). Chromatin and transcriptional signatures for Nodal signaling during stem cells. Nat. Biotechnol. 32, 1121–1133. endoderm formation in hESCs. Dev. Biol. 357, 492–504. Robertson, E.J. (2014). Dose-dependent Nodal/Smad signals pattern the early Kinder, S.J., Tsang, T.E., Quinlan, G.A., Hadjantonakis, A.K., Nagy, A., and mouse embryo. Semin. Cell Dev. Biol. 32, 73–79. Tam, P.P. (1999). The orderly allocation of mesodermal cells to the extraem- Singh, A.M., Reynolds, D., Cliff, T., Ohtsuka, S., Mattheyses, A.L., Sun, Y., bryonic structures and the anteroposterior axis during gastrulation of the Menendez, L., Kulik, M., and Dalton, S. (2012). Signaling network crosstalk mouse embryo. Development 126, 4691–4701. in human pluripotent cells: a Smad2/3-regulated switch that controls the bal- Kunisada, Y., Tsubooka-Yamazoe, N., Shoji, M., and Hosoya, M. (2012). Small ance between self-renewal and differentiation. Cell Stem Cell 10, 312–326. molecules induce efficient differentiation into insulin-producing cells from hu- Surmacz, B., Fox, H., Gutteridge, A., Fish, P., Lubitz, S., and Whiting, P. (2012). man induced pluripotent stem cells. Stem Cell Res. (Amst.) 8, 274–284. Directing differentiation of human embryonic stem cells toward anterior neural Li, B., Kuriyama, S., Moreno, M., and Mayor, R. (2009). The posteriorizing gene ectoderm using small molecules. Stem Cells 30, 1875–1884. Gbx2 is a direct target of Wnt signalling and the earliest factor in neural crest Takao, Y., Yokota, T., and Koide, H. (2007). Beta-catenin up-regulates Nanog induction. Development 136, 3267–3278. expression through interaction with Oct-3/4 in embryonic stem cells. Biochem. Li, Y., Yu, W., Cooney, A.J., Schwartz, R.J., and Liu, Y. (2013). Brief report: Biophys. Res. Commun. 353, 699–705. Oct4 and canonical Wnt signaling regulate the cardiac lineage factor Mesp1 Tam, P.P., and Behringer, R.R. (1997). Mouse gastrulation: the formation of a through a Tcf/Lef-Oct4 composite element. Stem Cells 31, 1213–1217. mammalian body plan. Mech. Dev. 68, 3–25. Liu, P., Wakamiya, M., Shea, M.J., Albrecht, U., Behringer, R.R., and Bradley, Tam, P.P., Loebel, D.A., and Tanaka, S.S. (2006). Building the mouse gastrula: A. (1999). Requirement for Wnt3 in vertebrate axis formation. Nat. Genet. 22, signals, asymmetry and lineages. Curr. Opin. Genet. Dev. 16, 419–425. 361–365. Tan, J.Y., Sriram, G., Rufaihah, A.J., Neoh, K.G., and Cao, T. (2013). Efficient Mendjan, S., Mascetti, V.L., Ortmann, D., Ortiz, M., Karjosukarso, D.W., Ng, Y., derivation of lateral plate and paraxial mesoderm subtypes from human em- Moreau, T., and Pedersen, R.A. (2014). NANOG and CDX2 pattern distinct bryonic stem cells through GSKi-mediated differentiation. Stem Cells Dev. subtypes of human mesoderm during exit from pluripotency. Cell Stem Cell 22, 1893–1906. 15, 310–325. Thomson, M., Liu, S.J., Zou, L.N., Smith, Z., Meissner, A., and Ramanathan, S. Menendez, L., Yatskievych, T.A., Antin, P.B., and Dalton, S. (2011). Wnt (2011). Pluripotency factors in embryonic stem cells regulate differentiation signaling and a Smad pathway blockade direct the differentiation of human into germ layers. Cell 145, 875–889.

Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc. 651 Tortelote, G.G., Herna´ ndez-Herna´ ndez, J.M., Quaresma, A.J., Nickerson, J.A., Vincent, S.D., Dunn, N.R., Hayashi, S., Norris, D.P., and Robertson, E.J. Imbalzano, A.N., and Rivera-Pe´ rez, J.A. (2013). Wnt3 function in the epiblast is (2003). Cell fate decisions within the mouse organizer are governed by graded required for the maintenance but not the initiation of gastrulation in mice. Dev. Nodal signals. Genes Dev. 17, 1646–1662. Biol. 374, 164–173. Wang, Z., Oron, E., Nelson, B., Razis, S., and Ivanova, N. (2012). Distinct line- Tremblay, K.D., Hoodless, P.A., Bikoff, E.K., and Robertson, E.J. (2000). age specification roles for NANOG, OCT4, and SOX2 in human embryonic Formation of the definitive endoderm in mouse is a Smad2-dependent pro- stem cells. Cell Stem Cell 10, 440–454. cess. Development 127, 3079–3090. Wilson, V., and Beddington, R. (1997). Expression of T protein in the primitive Trott, J., and Martinez Arias, A. (2013). Single cell lineage analysis of mouse streak is necessary and sufficient for posterior mesoderm movement and so- embryonic stem cells at the exit from pluripotency. Biol. Open 2, 1049–1056. mite differentiation. Dev. Biol. 192, 45–58. Tsakiridis, A., Huang, Y., Blin, G., Skylaki, S., Wymeersch, F., Osorno, R., Xu, R.H., Sampsell-Barron, T.L., Gu, F., Root, S., Peck, R.M., Pan, G., Yu, J., Economou, C., Karagianni, E., Zhao, S., Lowell, S., and Wilson, V. (2014). Antosiewicz-Bourget, J., Tian, S., Stewart, R., and Thomson, J.A. (2008). Distinct Wnt-driven primitive streak-like populations reflect in vivo lineage pre- NANOG is a direct target of TGFbeta/activin-mediated SMAD signaling in hu- cursors. Development 141, 1209–1221. man ESCs. Cell Stem Cell 3, 196–206. Tzouanacou, E., Wegener, A., Wymeersch, F.J., Wilson, V., and Nicolas, J.F. (2009). Redefining the progression of lineage segregations during mammalian Yamaguchi, T.P., Takada, S., Yoshikawa, Y., Wu, N., and McMahon, A.P. embryogenesis by clonal analysis. Dev. Cell 17, 365–376. (1999). T (Brachyury) is a direct target of Wnt3a during paraxial mesoderm specification. Genes Dev. 13, 3185–3190. Valenta, T., Hausmann, G., and Basler, K. (2012). The many faces and func- tions of b-catenin. EMBO J. 31, 2714–2736. Yeo, C., and Whitman, M. (2001). Nodal signals to Smads through Cripto- dependent and Cripto-independent mechanisms. Mol. Cell 7, 949–957. Vallier, L., Mendjan, S., Brown, S., Chng, Z., Teo, A., Smithers, L.E., Trotter, M.W., Cho, C.H., Martinez, A., Rugg-Gunn, P., et al. (2009). Activin/Nodal sig- Zafarana, G., Avery, S.R., Avery, K., Moore, H.D., and Andrews, P.W. (2009). nalling maintains pluripotency by controlling Nanog expression. Development Specific knockdown of OCT4 in human embryonic stem cells by inducible 136, 1339–1349. short hairpin RNA interference. Stem Cells 27, 776–782.

652 Cell Stem Cell 16, 639–652, June 4, 2015 ª2015 Elsevier Inc.