The T-Box Transcription Factor Eomesodermin Acts Upstream of Mesp1 to Specify Cardiac Mesoderm During Mouse Gastrulation

LETTERS

The T-box transcription factor Eomesodermin acts upstream of Mesp1 to specify cardiac mesoderm during mouse gastrulation

Ita Costello1, Inga-Marie Pimeisl2, Sarah Dräger2, Elizabeth K. Bikoff1, Elizabeth J. Robertson1,3 and Sebastian J. Arnold2,3

Instructive programmes guiding cell-fate decisions in the region of the primitive streak (APS), in close proximity to the developing mouse embryo are controlled by a few so-termed cardiovascular progenitors10,11. master regulators. Genetic studies demonstrate that the T-box Mesoderm formation and patterning along the proximodistal axis transcription factor Eomesodermin (Eomes) is essential for of the primitive streak is known to be regulated by dose-dependent epithelial-to-mesenchymal transition, mesoderm migration and Nodal/Smad2/3 activities7. The highest level of Nodal/Smad2/3 specification of definitive endoderm during gastrulation1. Here signalling is required to specify APS derivatives, namely the definitive we report that Eomes expression within the primitive streak endoderm, node and notochord5,6,8. The transcription factor Smad4, marks the earliest cardiac mesoderm and promotes formation and its DNA-binding partner the forkhead transcription factor Foxh1, of cardiovascular progenitors by directly activating the bHLH also play essential roles in APS specification6,12,13. Furthermore, (basic-helix-loop-helix) transcription factor gene Mesp1 the T-box transcription factor Eomes acts cooperatively with upstream of the core cardiac transcriptional machinery2–4. In the Nodal/Smad2/3 pathway to promote delamination of nascent marked contrast to Eomes/Nodal signalling interactions that mesoderm and specification of APS fates1. cooperatively regulate anterior–posterior axis patterning and Eomes expression is initiated in the prospective posterior aspect allocation of the definitive endoderm cell lineage1,5–8, formation of the epiblast at embryonic day 5.75 (E5.75; ref. 14). During of cardiac progenitors requires only low levels of Nodal activity gastrulation expression is maintained in the distal primitive streak14,15, accomplished through a Foxh1/Smad4-independent encompassing the same region where cranial, cardiac and paraxial mechanism. Collectively, our experiments demonstrate that mesodermal subcell populations are generated10. Inactivation of Eomes governs discrete context-dependent transcriptional Eomes in the epiblast results in a severe block in EMT and arrests programmes that sequentially specify cardiac and definitive development at gastrulation1. To further characterize Eomes functional endoderm progenitors during gastrulation. contributions within the mesodermal cell lineages we generated an EomesCre reporter allele. Cre messenger RNA expression recapitulates Much has been learned about the coordinated activities of key endogenous expression (Supplementary Fig. S1a), enabling derivation regulatory networks of transcription factors and growth-factor of a fate map of Eomes-expressing cells in later-stage embryos (Fig. 1). signalling pathways governing cell-fate decisions during gastrulation9. EomesCre/+ males were mated to females carrying the ROSA26 R Nascent mesoderm is induced as epiblast cells ingress through the reporter allele16 and the resulting embryos stained for LacZ activity primitive streak and undergo epithelial-to-mesenchymal transition (Fig. 1d,e and Supplementary Fig. S1b). Surprisingly, we found in (EMT). Distinct mesodermal subpopulations become allocated E8.5 and E9.5 EomesCre ; ROSA26 R embryos that LacZ-expressing according to the timing and order of these cell movements. Thus cells are mostly absent from the somites, intermediate and lateral extra-embryonic mesoderm arises from the posterior primitive plate mesoderm and largely restricted to the head mesenchyme, streak, whereas cardiac, paraxial and lateral plate precursors emerge cardiac mesoderm and vasculature (Fig. 1d,e). As expected1, the sequentially as the primitive streak elongates towards the distal definitive endoderm and gut tube exclusively consist of LacZ-marked tip of the embryo. Fate-mapping experiments demonstrate that EomesCre -positive descendants. At later stages endodermal but not definitive endoderm progenitors are specified in the most anterior mesodermal components of developing organs derived from the gut

1Sir William Dunn School of Pathology, University of Oxford, South Parks Road, Oxford OX1 3RE, UK. 2University Medical Centre, Renal Department, Centre for Clinical Research, Breisacher Strasse 66, 79106 Freiburg, Germany. 3Correspondence should be addressed to E.J.R. or S.J.A. (e-mail: [email protected] or [email protected])

Received 26 January 2011; accepted 23 June 2011; published online 7 August 2011; DOI: 10.1038/ncb2304

LETTERS

a 15.2 kb b Wild type 15.2 kb Eomes RV H SE HRV locus (wild type) Targeted 8.0 kb Targeting construct PGK-neo TK Cre RV Targeted RV H H RV neo allele PGK- Cre c 8.0 kb Wild type 15.2 kb RV Reporter RV H H RV Targeted 8.0 kb allele Reporter Cre 6.2 kb 6.2 kb

E8.5 He Hm Gt

Gt He Gt

E9.5 Eomes Cre × ROSA26R Eomes Cre × ROSA26R Figure 1 Fate mapping of Eomes Cre -expressing cells reveals selective the phosphoglycerate kinase–neomycin (PGK-neo) selection cassette and contributions to deﬁnitive endoderm and cardiovascular cell lineages. generate the reporter allele, as shown by Southern blot. (d,e) Fate-mapping (a) Targeting strategy used to generate the Eomes Cre reporter allele. Cre experiments demonstrate that descendants of Eomes Cre -expressing cells recombinase coding sequences were inserted into exon 1 of the Eomes contribute to the myocardium and endocardium of the heart (He), the locus. RV, EcoRV; H, HpaI; S, SphI; E, EagI; Cre, Cre recombinase; head mesenchyme (Hm), vasculature and the endoderm of the primary TK, thymidine kinase. (b) Embryonic stem cell clones were screened gut tube (Gt), but rarely colonize other mesoderm tissues formed from by Southern blot on EcoRV-digested DNA using a 30 probe (red line paraxial and lateral plate mesoderm. Sections were counterstained with in a) to detect wild-type and targeted alleles. (c) Correctly targeted eosin to highlight non-labelled cells. The black lines indicate the plane embryonic stem cells were transiently transfected with Cre to excise of section. tube are LacZ positive (Supplementary Fig. S1c). Eomes-expressing Brachyury, Fgf8 and Snail (ref. 1). However, in marked contrast, cells thus give rise to two discrete progenitor cell populations during expression of the early cardiac marker genes Myl7 (also known as gastrulation, namely the prospective cranial and cardiac mesoderm Mlc2a), Wnt2 and Nkx2.5 was absent (Fig. 2a). Moreover, we observe that emerge from the primitive streak at early stages, and APS severely reduced expression of early vascular marker genes such as derivatives giving rise to the definitive endoderm, node and notochord. Agtrl1, Rasgrp3 and Klhl6 (Fig. 2a). In marked contrast, Eomes+ cells are excluded from the majority To test whether loss of cardiac-gene expression reflects a cell- of mesodermal tissues derived from the paraxial and lateral plate autonomous Eomes requirement we examined the developmental mesoderm. These observations indicate that a discrete subpopulation potential of Eomes-null embryonic stem cells in the context of chimaeric of cells within the pregastrulation epiblast preferentially ingress and embryos. Eomes-null embryonic stem cells1 were injected into wild-type migrate anteriorly as a cohort to form the cardiac crescent and blastocysts carrying the ROSA26LacZ allele17 and the resulting embryos prospective head mesoderm. analysed by LacZ staining (Fig. 2b). As expected1, at E8.5 and E9.5 To directly examine whether Eomes function is required for Eomes-null cells efficiently contribute to the majority of mesodermal specification of cardiovascular progenitors, we analysed E7.5 embryos tissues but are entirely excluded from the forming gut tube. Notably, in carrying an epiblast-specific Eomes deletion (EomesCA/N ;Sox2.Cre; all cases examined (n = 10), the myocardium and endocardium of the ref. 1) by whole-mount in situ hybridization. Embryos lacking developing heart also exclusively consisted of wild-type LacZ-positive Eomes function strongly express mesodermal marker genes, including cells (Fig. 2c,d). Thus we conclude that Eomes plays an essential

LETTERS

a E7.5 E7.5 E7.5 E7.5 E7.5 E7.75 b c Hm

Wild type Gt Gt Eomes –/– embryonic stem cells E7.5 E7.5 E7.5 E7.5 E7.5 E7.75 He

Eomes mutant d ROSA26 LacZ blastocyst / wild type Hm E7.75 E7.75 E7.75 E9.5 E9.5 E8.5 Gt

Wild type He

Myl7 Wnt2 Nkx2.5 Agtrl1 Rasgrp3 Klhl6

Figure 2 Eomes functional loss disrupts speciﬁcation of cardiovascular for generation of chimaeric embryos. Eomes-null embryonic stem progenitors. (a) Whole-mount in situ hybridization analysis of cells were introduced into wild-type ROSA26LacZ blastocysts. cardiac mesoderm (Myl7, Wnt2, Nkx2.5) and vascular (Agtrl1, (c,d) Histological sections of two independent LacZ -stained E9.5 Rasgrp3, Klhl6) markers in control and Eomes N/CA;Sox2.Cre-mutant chimaeric embryos were counterstained with eosin to identify embryos. Eomes mutants entirely lack expression of cardiac Eomes-mutant cell populations (pink). The myocardium and endocardium marker genes and show signiﬁcantly reduced expression of vascular of the heart (He) and endoderm of the gut tube (Gt) exclusively consist of markers. In contrast, in wild-type embryos vascular markers LacZ -positive wild-type cells. Relatively few Eomes-null cells colonize the broadly delineate the embryonic and extra-embryonic vasculature head mesenchyme (Hm), whereas the majority of other tissues consist of at E8.5 and E9.5. (b) Schematic representation of the protocol mixed populations of Eomes-null and wild-type cells. Scale bar, 500 µm. cell-autonomous role during allocation of both the definitive endoderm levels of Eomes expression, early cardiac markers Mesp1 and Myl7 and cardiovascular progenitors. are upregulated beginning at around day 4. In Eomes-null embryoid To further evaluate Eomes contributions we exploited culture bodies the pan-mesodermal marker Brachyury is robustly induced protocols that promote embryonic stem cell differentiation towards (Fig. 3c), but in marked contrast expression of cardiac-specific genes either definitive endoderm or cardiac fates. To elicit definitive including Mesp1, Myl7, Myl2, Myocardin, Nkx2.5 and Mef2c was endoderm formation, embryonic stem cells harbouring an EomesGFP entirely absent. As judged by microscopic observation and staining reporter allele18 were cultured in the presence of high doses of ActivinA for the cardiac protein troponin I (TnI), close to 100% of wild-type (50 ng ml−1, Supplementary Fig. S2a). After 4 days, as judged by green embryoid bodies contain clusters of contractile cardiomyocytes after fluorescent protein (GFP) expression or staining with an Eomes day 7 of differentiation (Fig. 3d,e and Supplementary Movie SM1). antibody, more than 95% of cells were strongly positive (Supplementary Eomes-null embryoid bodies fail to differentiate into cardiomyocytes by Fig. S2b and data not shown). Next we compared wild-type (CCE), these criteria (Fig. 3d,e and Supplementary Movie SM2). Collectively, Eomes-null (6A6; ref. 1) and Smad2-null (KT15; ref. 7) embryonic these results suggest that Eomes acts upstream of the transcriptional stem cells and monitored definitive endoderm marker-gene expression hierarchy that specifies cardiac fates during gastrulation. by PCR with reverse transcription (RT–PCR). Eomes expression was The bHLH transcription factor Mesp1 has been described as a master detectable 24 h after ActivinA treatment in wild-type cells, severely regulator of multipotent cardiovascular progenitor specification21–23. reduced in Smad2-deficient cells and absent from Eomes-null cells Genetic fate-mapping experiments demonstrate that Mesp1 expression (Fig. 3a). In chimaera studies Smad2-deficient cells robustly contribute marks cardiac progenitors that give rise to the myocardial and to all mesodermal lineages but are excluded from definitive endoderm endocardial derivatives22. Moreover, Mesp1 activity has also been derivatives19 and as shown here only inefficiently upregulate expression implicated in the early steps of cardiac lineage specification in vitro2–4. of Eomes or definitive endoderm markers. Differentiating Eomes-null Mesp1 is strongly expressed at the onset of gastrulation (E6.5) along the embryonic stem cells strongly express primitive streak and mesodermal forming primitive streak, marking the prospective cardiac mesoderm, marker genes, including Brachyury and Mixl1, but expression of and is then rapidly downregulated starting around E7.5 (ref. 22). definitive endoderm marker genes including Sox17 and Foxa2 was Mesp1 mutants form cardiac mesoderm but heart morphogenesis is markedly reduced. Similar conclusions were reached in quantitative highly compromised, leading to cardia bifida21,22. Mesp1 and Mesp2 RT–PCR (qRT–PCR) experiments examining a panel of key definitive- are functionally interchangeable in vivo21. Thus premature upregulated endoderm-associated genes including Gsc, Gata6 and Lhx1 (Fig. 3b). expression of the closely related family member Mesp2 in the Next we examined the development of cardiac progenitors through loss-of-function mutants is sufficient to generate cardiac progenitors embryoid body differentiation in hanging drops with high serum and partially rescue heart morphogenesis23. Mesp1;Mesp2 double- stimulation20 (Supplementary Fig. S2c,d). In wild-type cultures Eomes mutant embryos block at gastrulation, and show an EMT phenotype transcripts are detectable by RT–PCR on day 2, peak around day 4 that closely resembles the Eomes loss-of-function phenotype1,23. At and are downregulated by day 6 (Fig. 3c). Coincident with highest E7.0 Mesp1 is completely absent from Eomes mutants (Fig. 4a). At

LETTERS

a Wild type Eomes –/– Smad2–/– b Gene ID Definitive endoderm Definitive endoderm qPCR fold change qPCR log (fold change) Eomes 2 Eomes >1,000 13.6

Bra Foxa1 25 4.7

Foxa2 51 5.7 Mixl1 Gsc >1,000 14.2

Foxa2 Sox17 >1,000 12.5 Gata6 Sox17 >1,000 10.9 Lhx1 247 7.9 Hprt

d0 d2 d3 d4 d5 d0 d2 d3 d4 d5 d0 d2 d3 d4 d5 d –Act+Activin –Act +Activin –Act +Activin 100 E8.5 RNA 80 c Wild type Eomes –/– 60 Eomes

colonies 40 Mesp1 20

Mesp2 Percentage of beating d6 d7 d8 d6 d7 d8 Bra Wild type Eomes –/– Tbx6

Gata6 Nkx2.5 e

Mef2c

Myl7 Wild type

Myl2

Myocardin

β-actin Eomes –/–

d0 d2 d4 d6 d8 d0 d2 d4 d6 d8 TnI TnI+DAPI Pos. control Neg. control Figure 3 Eomes-null embryonic stem cells fail to give rise to definitive Mef2c and Myocardin), as well as structural proteins (Myl7 and Myl2) endoderm and cardiomyocytes. (a) Wild-type, Eomes −/− and Smad 2 −/− are significantly downregulated in Eomes-null embryoid bodies whereas embryonic stem cells were cultured in the presence of high doses of expression of the pan-mesodermal marker Brachyury is unaffected. ActivinA. Semi-quantitative RT–PCR analysis shows that Eomes −/− and (d) Clusters of beating cardiomyocytes are readily detectable in wild-type Smad 2 −/− cultures strongly express mesoderm marker genes such as embryoid body outgrowths but absent from Eomes-null cultures at day Brachyury and Mixl1, but lack expression of the definitive endoderm 7. Error bars represent the standard error of the mean (s.e.m.) of three marker genes Foxa2 and Sox17.(b) qRT–PCR analysis confirms the independent experiments. (e) At day 8 TnI-positive cardiomyocytes markedly reduced definitive endoderm marker transcript levels in are detectable in wild-type outgrowths but are entirely absent from Eomes −/− at day 4 of ActivinA-induced differentiation. (c) Wild-type and Eomes-mutant cultures. Higher magnification reveals characteristic Eomes-mutant embryoid bodies were induced to form cardiomyocytes in cross-striation of myofibrils. Scale bar, 100 µm for the overview and 10 µm hanging-drop cultures. Semi-quantitative RT–PCR analysis reveals that for the higher-magnification image. DAPI, 4,6-diamidino-2-phenylindole. cardiac-specific transcription factors (Mesp1, Mesp2, Gata6, Nkx2.5, Uncropped images of blots are shown in Supplementary Fig. S7. slightly later stages (E7.25) a few exceptional Eomes mutants show separated by approximately 17 kilobases (kb) (ref. 25). The cis-acting weak Mesp1 staining. At E7.25 Mesp2 expression is undetectable in regulatory elements responsible for controlling temporally and whole-mount in situ hybridization experiments24. However, qRT–PCR spatially restricted expression patterns are well characterized25–27. An analysis demonstrates that Mesp1 and Mesp2 expression are both evolutionarily conserved early mesoderm enhancer (EME, Fig. 4c and markedly downregulated (Fig. 4b). As shown above, differentiating Supplementary Fig. S3b,d) recapitulates Mesp1 expression in nascent Eomes-null embryonic stem cells lack the ability to upregulate Mesp1 mesoderm25, whereas regulatory sequences adjacent to the Mesp2 and Mesp2 (Fig. 3c). Collectively, these experiments indicate that Eomes TSS direct mesodermal expression in the primitive streak, presomitic acts upstream of Mesp1/2 to regulate EMT and control allocation of mesoderm and developing somites26. Both elements contain T-box cardiac progenitors. consensus binding sites25–27 (Supplementary Fig. S3). Furthermore, Mesp1 and Mesp2 are closely linked on chromosome 7, and arranged we have identified a further conserved T-box binding element in close in opposite orientations with the two transcriptional start sites (TSSs) proximity to the Mesp1 TSS (Fig. 4c and Supplementary Fig. S3b,c).

LETTERS

Mesp1 abE7.0 E7.25 Mesp2 cMesp1 Mesp2 0 >4 kb –10 9 kb 22kb –20 Wild type Ex2 Ex1 Ex1 Ex2 –30 Mesp1/2 Mesp2 EME –40 PSME Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp_Con

Fold change PCR –50 products

–60 ....TTCACACTTTT .... Eomes mutant .....AGGTGTGA ..... Site B Site G Site D –70

...CGAGGGGTCAGAATCCACACCTC...... CACCTCACACCT ... Mesp1 ...GTTTTGACACCT ... de 1.2 Eomes ChIP4.0 Eomes ChIP 1.0 3.5 3.0 0.8 2.5 0.6 2.0 0.4 1.5 1.0

Percentage of input 0.2 Percentage of input 0.5 0 0 d0 d4 d0 d4 d0 d4 d0 d4 d0 d4 Con TamCon Tam Con Tam Con Tam Con Tam Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp2_Con Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp2_Con

PolII ChIP0.4 PolII ChIP 0.4 0.2 0.2 0 0 d0 d4 d0 d4 d0 d4 d0 d4 d0 d4 Con Tam ConTam Con Tam Con Tam Con Tam Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp2_Con Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp2_Con Percentage of input Percentage of input

IgG ChIP IgG ChIP 0.2 0.2 0 0 d0 d4 d0 d4 d0 d4 d0 d4 d0 d4 Con Tam ConTam Con Tam Con Tam Con Tam Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp2_Con Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp2_Con Percentage of input Percentage of input

Figure 4 Eomes directly binds conserved T-box sites within the Mesp1 Ex1, exon 1; Ex2, exon 2; PSME, presomitic mesoderm enhancer; Tbx, locus to activate expression. (a) Mesp1, robustly expressed in wild-type T-box site. (d) ChIP analysis of P19Cl6 cells treated for 4 days with DMSO embryos at E7.0, is absent from Eomes N/CA;Sox2Cre mutants. At slightly using antibodies specific for Eomes, RNA polymerase II (PolII) or an IgG later stages (E7.25) Eomes N/CA;Sox2Cre embryos occasionally show weak control. Specific enrichment for genomic loci containing T-box sites (Mesp1_ expression that probably reflects activity of Tbx6, known to be expressed TSS, Mesp1_ Tbx, Mesp2_ Tbx) was observed using the Eomes-specific in E7.5 Eomes mutants1.(b) As judged by qRT–PCR, Mesp1 and Mesp2 antibody. The PolII antibody gave specific enrichment of the Mesp1_ transcripts are markedly reduced in E7.25 Eomes-mutant embryos. TSS, but not other tested regions of the Mesp1/2 locus. Specific binding Error bars represent the standard deviation (s.d.), n = 5 per genotype. to T-box sites (indicated by asterisks) in the Mesp1/2 EME, the Mesp2 (c) Schematic representation of cis-regulatory elements in the Mesp1/2 PSME and the Mesp1 TSS was detectable after 4 days, but not at day locus. The positions of previously identified T-box sites within the Mesp1/2 0. d0, day 0; d4, day 4 of DMSO differentiation of P19Cl6 cells. (e) EME and Mesp2 PSME (nomenclature according to refs 26,27) and a Cells expressing a tamoxifen-inducible Eomes fusion construct (P19EoER putative T-box site identified near the Mesp1 TSS are indicated. The Mesp2 cells) also show Eomes binding to these T-box sites at day 4 of tamoxifen PSME contains three binding elements (sites B, G, D) that contain T-box treatment. Con, non-tamoxifen-induced control P19EoER cells; Tam, day binding motifs. T-box consensus sequences are indicated in red. Red bars 4 tamoxifen-treated P19EoER cells. The most representative plots of three indicate areas amplified by qPCR after ChIP using different antibodies. independent experiments are shown.

To further investigate Mesp1/2 activation during cardiovascular the presence of tamoxifen (EomesER, Supplementary Fig. S4b) also lineage commitment, we used the P19Cl6 embryonal carcinoma cell efficiently induces strong Mesp1 expression within 24 h as assayed by line that efficiently differentiates into beating cardiomyocytes in the qPCR (Supplementary Fig. S4g). To directly evaluate Eomes occupancy presence of 1% dimethylsulphoxide (DMSO) (ref. 28). Transient at the Mesp1/2 locus we carried out chromatin immunoprecipitation Eomes expression seen initially at day 2 is downregulated by day 6, (ChIP) analysis with day 4 DMSO-treated P19Cl6 cells as well as associated with increased Mesp1 expression levels (Supplementary tamoxifen-treated P19Cl6EomesER (P19EoER) cells. Eomes binding Fig. S4a). Activation of Eomes oestrogen-receptor fusion protein in to all three conserved T-box-site-containing regions within the

LETTERS

a E6.5 E7.5 E6.5 E7.5

Foxh1–/– Wild type mutant

E6.5 E7.5 E6.5 E7.5

Lhx1–/– Smad4 Δ mutant mutant

Mesp1 Mesp1

b c E6.25 Eomes E7.5–E8.5 Tbx6

Mesp1/2 Lefty2 Low

expression Mesp1/2

Cardiac Nodal/Smad2/3 levels mesoderm Nodal Eomes Presomitic mesoderm

E6.5-E7.0 Eomes

Smad4/Foxh1

Foxa2 Sox17 expression Gsc expression High

Eomes Definitive

endoderm Tbx6

Figure 5 Eomes and dose-dependent Nodal/Smad2/3 signalling levels Nodal signalling. Eomes-dependent activation of Mesp1/2 marks the control cardiac mesoderm and definitive endoderm specification earliest cardiac progenitors induced in the forming primitive streak. during gastrulation. (a) Smad4 and Foxh1 are critical Nodal pathway Mesp1/2 expression leads to activation of the Nodal antagonist Lefty2 components for transducing high levels of signalling6,12,13. Mesp1 (ref. 23) and direct repression of definitive endoderm genes2. As cells is expressed normally in E6.5 and E7.5 Foxh1-null embryos and begin to migrate away from the primitive streak, Mesp1/2 expression is in embryos lacking Smad4 in the epiblast only (Smad 41). Mesp1 downregulated through a negative-feedback loop2,23,25. In contrast, the expression is also efficiently induced in Lhx1-mutant embryos, which Eomes expression domain extends distally and overlaps with increased show definitive endoderm and midline mesoderm defects. However, the Nodal signalling levels as the primitive streak elongates. Eomes, acting failure of anterior–posterior axis rotation results in induction of Mesp1 cooperatively with Nodal/Smad4/Foxh1-dependent signals in the APS, throughout the proximal epiblast. (b) Eomes activity regulates formation induces definitive endoderm. (c) At later stages from E7.5 onwards, of both cardiac mesoderm and definitive endoderm progenitors during Tbx6 expression in the presomitic mesoderm activates a second wave of gastrulation. Eomes + epiblast cells confined to the posterior side of Mesp1/2 expression in the presomitic mesoderm through occupancy of the embryo before overt streak formation are exposed to low levels of the conserved T-box regulatory elements.

Mesp1/2 locus was clearly observed (Fig. 4d,e and Supplementary expression. Specification of early heart progenitors proceeds normally Fig. S5). The EME, which controls early Mesp1 expression in nascent in T (Brachyury) mutant embryos. Therefore, a strong argument mesoderm25, gave the strongest signal. Occupancy at other genomic can be made that the earliest Mesp1 expression domain marking the regions, or in uninduced cells, was undetectable. The T-box sites cardiac progenitors is directly activated by Eomes occupancy at these adjacent to the Mesp2 TSS are known to be occupied by Tbx6 in T-box sites. presomitic mesoderm27. This cis-acting regulatory element, and the Genetic studies analysing double-heterozygous-mutant embryos minimal 220-bp (base pair) Mesp1 EME T-box element, regulate demonstrate that Eomes and Nodal function cooperatively in Mesp2 and Mesp1 expression respectively at later stages during anterior–posterior axis patterning and formation of APS derivatives1. somitogenesis26. Further T-box family members are also known to be We wondered whether dose-dependent Nodal signals could potentially expressed in the early-gastrulation-stage embryo14,15,29–31. However, regulate Eomes-dependent Mesp1 activation. To test this possibility only Brachyury and Eomes are exclusively present in the posterior first we examined Smad4 (ref. 6)- and Foxh1 (ref. 12)-mutant embryos epiblast and nascent mesoderm overlapping with sites of Mesp1 at E6.5 and E7.5. Interestingly, Mesp1 expression is unaffected by

LETTERS loss of either Smad4 or Foxh1 (Fig. 5a). The lim-homeodomain gives rise to the cardiac crescent. Within this early subset, Mesp1 transcription factor Lhx1 is also required for specification of definitive directly represses genes required for formation of definitive endoderm, endoderm and anterior axial midline tissues32,33. Similarly, we including Foxa2, Gsc and Sox17 (ref. 2). Mesp1/2 expression also observe that compromised definitive endoderm development has no activates expression of the Nodal antagonist Lefty2 (ref. 23) to further noticeable impact on specification of Mesp1+ cardiac progenitors. insulate cardiac progenitors as they migrate away from the source Thus Lhx1 loss-of-function embryos show strong expression of of Nodal signalling (Supplementary Fig. S6e). Consistent with this both Mesp1 and Eomes (Fig. 5a and data not shown). Conversely, idea, upregulated Nodal signalling in Tgif1/2 double-mutant embryos it is known that expression of the APS markers Lhx1 and Gsc inhibits Mesp1 expression, whereas decreased levels of Nodal activity is unperturbed in Mesp1/2 double-mutant embryos23. Lowering rescue Mesp1 expression36. Nodal5,34 or Smad2/3 (ref. 7) levels during gastrulation also selectively Eomes activation of Mesp1/2 in cardiac progenitors is only transient disturbs definitive endoderm specification. Next, we manipulated owing to a Mesp1/2 autoregulatory negative-feedback loop2,23,25. At ActivinA concentrations in differentiating embryonic stem cells late streak stages expression of Mesp1/2 is reactivated in presomitic and confirmed that low levels (5 ng ml−1) are sufficient for robust mesoderm by Tbx6 occupancy of the conserved T-box sites27 Mesp1 expression, whereas conversely maintaining cultures in high (Fig. 5c). In contrast, during streak elongation Eomes expression is ActivinA concentrations (50 ng ml−1) leads to induction of Sox17 maintained by high levels of Nodal/Smad2 signalling downstream of a (Supplementary Fig. S6a,b). Collectively, these results indicate that Wnt3/Lef1 feedforward loop8. Acting together with Nodal-dependent specification of both cardiac and definitive endoderm progenitors Smad2/3/4/Foxh1 transcription complexes5–7,12,34, Eomes promotes requires Eomes, but these lineages arise independently and are formation of APS progenitors that give rise to the definitive endoderm, dependent on local Nodal/Smad/Foxh1 signalling levels. node and notochord. It will be interesting to learn more about the Multipotent mesodermal progenitor cells that give rise to diverse developmental regulation of Eomes transcriptional partnerships and tissues of the emerging body plan become progressively allocated the cell-type-specific changes in chromatin architecture, which govern as epiblast cells transit the primitive streak. Fate-mapping and T-box site occupancy at the Mesp1/2 locus and selection of target genes grafting studies have shown that cranial and cardiac mesoderm derive in the definitive endoderm cell lineage. from the earliest wave of cells that exit at the mid-streak stage, METHODS whereas presomitic/paraxial and lateral plate mesoderm emerge at more posterior levels at late-streak stages10,11. Notably, the present Methods and any associated references are available in the online experiments identify a subset of proximal posterior epiblast cells version of the paper at http://www.nature.com/naturecellbiology already committed to adopt a cardiac fate many hours before Note: Supplementary Information is available on the Nature Cell Biology website mesoderm induction and overt primitive streak formation. The early ACKNOWLEDGEMENTS Eomes expression domain marks cardiac progenitors programmed We thank N. Hortin, A. Salman and M. Pavlovic for technical assistance, C. Böhlke to activate Mesp1, previously identified as the master regulator that and A. Hofherr for help with imaging techniques, S. Stefanovic for qPCR primer acts instructively to specify cardiovascular cell fates. Recent studies optimization and H. Niwa and Y. Saga for plasmids. This work was supported by demonstrate that Mesp1 initiates global changes in gene expression by the Emmy Noether Programme and SFB850 of the German Research Council to S.J.A. and a Programme Grant from the Wellcome Trust to E.J.R. directly binding to regulatory sequences at the promoters of many key genes in the core cardiac transcriptional machinery. Mesp1 upregulates AUTHOR CONTRIBUTIONS expression of Hand2, Myocardin, Nxk2.5 and Gata4, and represses genes I.C., E.K.B., E.J.R. and S.J.A. designed experiments, I.C., I-M.P., S.D., E.J.R. and S.J.A. carried out research, I.C., E.J.R. and S.J.A. analysed data and I.C., E.K.B., E.J.R. and 2 governing pluripotency and mesodermal and endodermal cell fates . S.J.A. wrote and edited the manuscript. Furthermore, Mesp1 promotes EMT through selective upregulation COMPETING FINANCIAL INTERESTS of the zinc-finger repressor Snail4, enabling the nascent cardiac The authors declare no competing financial interests. progenitors to migrate anteriorly to underlie the developing headfolds, where they coalesce to form the cardiac crescent. Acting a few hours Published online at http://www.nature.com/naturecellbiology Reprints and permissions information is available online at http://www.nature.com/ later during primitive streak elongation, Eomes/Nodal signalling results reprints in specification of definitive endoderm1 that gives rise to the entire gut 1. Arnold, S. J., Hofmann, U. K., Bikoff, E. K. & Robertson, E. J. Pivotal roles for endoderm lineage. Eomes was recently shown to be a key player in the eomesodermin during axis formation, epithelium-to-mesenchyme transition and transcriptional network upstream of definitive endoderm specification endoderm speciﬁcation in the mouse. Development 135, 501–511 (2008). 35 2. Bondue, A. et al. Mesp1 acts as a master regulator of multipotent cardiovascular during human embryonic stem cell differentiation . progenitor speciﬁcation. Cell Stem Cell 3, 69–84 (2008). How does a single transcription factor, Eomes, govern allocation 3. David, R. et al. MesP1 drives vertebrate cardiovascular differentiation through of two independent, non-overlapping, multipotent progenitor cell Dkk-1-mediated blockade of Wnt-signalling. Nat. Cell Biol. 10, 338–345 (2008). 4. Lindsley, R. C. et al. Mesp1 coordinately regulates cardiovascular fate restriction populations as epiblast cells sequentially transit the primitive streak? and epithelial–mesenchymal transition in differentiating ESCs. Cell Stem Cell 3, We suggest that the key parameter controlling cardiac versus definitive 55–68 (2008). 5. Vincent, S. D., Dunn, N. R., Hayashi, S., Norris, D. P. & Robertson, E. J. Cell fate endoderm cell fate is the timing and duration of exposure to Nodal decisions within the mouse organizer are governed by graded Nodal signals. Genes signalling (Fig. 5b). Low levels of Nodal activity in the posterior epiblast Dev. 17, 1646–1662 (2003). 6. Chu, G. C., Dunn, N. R., Anderson, D. C., Oxburgh, L. & Robertson, E. J. Differential are sufficient to activate Eomes and induce cardiac mesoderm formation requirements for Smad4 in TGFβ-dependent patterning of the early mouse embryo. at early postimplantation stages (Supplementary Fig. S6d). Eomes Development 131, 3501–3512 (2004). 7. Dunn, N. R., Vincent, S. D., Oxburgh, L., Robertson, E. J. & Bikoff, E. K. expression is necessary and sufficient to activate Mesp1/2 and promote Combinatorial activities of Smad2 and Smad3 regulate mesoderm formation and EMT and migration of a discrete mesodermal subpopulation that patterning in the mouse embryo. Development 131, 1717–1728 (2004).

LETTERS

8. Ben-Haim, N. et al. The nodal precursor acting via activin receptors induces 22. Saga, Y. et al. MesP1 is expressed in the heart precursor cells and required for the mesoderm by maintaining a source of its convertases and BMP4. Dev. Cell 11, formation of a single heart tube. Development 126, 3437–3447 (1999). 313–323 (2006). 23. Kitajima, S., Takagi, A., Inoue, T. & Saga, Y. MesP1 and MesP2 are essential for the 9. Arnold, S. J. & Robertson, E. J. Making a commitment: cell lineage allocation development of cardiac mesoderm. Development 127, 3215–3226 (2000). and axis patterning in the early mouse embryo. Nat. Rev. Mol. Cell Biol. 10, 24. Saga, Y., Hata, N., Koseki, H. & Taketo, M. M. Mesp2: a novel mouse gene expressed 91–103 (2009). in the presegmented mesoderm and essential for segmentation initiation. Genes Dev. 10. Lawson, K. A., Meneses, J. J. & Pedersen, R. A. Clonal analysis of epiblast 11, 1827–1839 (1997). fate during germ layer formation in the mouse embryo. Development 113, 25. Haraguchi, S. et al. Transcriptional regulation of Mesp1 and Mesp2 genes: 891–911 (1991). differential usage of enhancers during development. Mech. Dev. 108, 11. Tam, P. P., Parameswaran, M., Kinder, S. J. & Weinberger, R. P. The allocation 59–69 (2001). of epiblast cells to the embryonic heart and other mesodermal lineages: the 26. Oginuma, M., Hirata, T. & Saga, Y. Identification of presomitic mesoderm (PSM)- role of ingression and tissue movement during gastrulation. Development 124, specific Mesp1 enhancer and generation of a PSM-specific Mesp1/Mesp2-null 1631–1642 (1997). mouse using BAC-based rescue technology. Mech. Dev. 125, 432–440 (2008). 12. Hoodless, P. A. et al. FoxH1 (Fast) functions to specify the anterior primitive streak 27. Yasuhiko, Y. et al. Functional importance of evolutionally conserved Tbx6 binding in the mouse. Genes Dev. 15, 1257–1271 (2001). sites in the presomitic mesoderm-specific enhancer of Mesp2. Development 135, 13. Yamamoto, M. et al. The transcription factor FoxH1 (FAST) mediates Nodal signaling 3511–3519 (2008). during anterior–posterior patterning and node formation in the mouse. Genes Dev. 15, 28. Habara-Ohkubo, A. Differentiation of beating cardiac muscle cells from a derivative 1242–1256 (2001). of P19 embryonal carcinoma cells. Cell Struct. Funct. 21, 101–110 (1996). 14. Russ, A. P. et al. Eomesodermin is required for mouse trophoblast development and 29. Chapman, D. L. et al. Expression of the T-box family genes, Tbx1–Tbx5, during early mesoderm formation. Nature 404, 95–99 (2000). mouse development. Dev. Dyn. 206, 379–390 (1996). 15. Ciruna, B. G. & Rossant, J. Expression of the T-box gene Eomesodermin during early 30. Chapman, D. L., Agulnik, I., Hancock, S., Silver, L. M. & Papaioannou, V. E. Tbx6, mouse development. Mech. Dev. 81, 199–203 (1999). a mouse T-Box gene implicated in paraxial mesoderm formation at gastrulation. Dev. 16. Soriano, P. Generalized lacZ expression with the ROSA26 Cre reporter strain. Biol. 180, 534–542 (1996). Nat. Genet. 21, 70–71 (1999). 31. Wilkinson, D. G., Bhatt, S. & Herrmann, B. G. Expression pattern of the mouse T 17. Friedrich, G. & Soriano, P. Promoter traps in embryonic stem cells: a genetic gene and its role in mesoderm formation. Nature 343, 657–659 (1990). screen to identify and mutate developmental genes in mice. Genes Dev. 5, 32. Shawlot, W. & Behringer, R. R. Requirement for Lim1 in head-organizer function. 1513–1523 (1991). Nature 374, 425–430 (1995). 18. Arnold, S. J., Sugnaseelan, J., Groszer, M., Srinivas, S. & Robertson, E. J. Generation 33. Tam, P. P., Khoo, P. L., Wong, N., Tsang, T. E. & Behringer, R. R. Regionalization of and analysis of a mouse line harboring GFP in the Eomes/Tbr2 locus. Genesis 47, cell fates and cell movement in the endoderm of the mouse gastrula and the impact 775–781 (2009). of loss of Lhx1(Lim1) function. Dev. Biol. 274, 171–187 (2004). 19. Tremblay, K. D., Hoodless, P. A., Bikoff, E. K. & Robertson, E. J. Formation of the 34. Norris, D. P., Brennan, J., Bikoff, E. K. & Robertson, E. J. The Foxh1-dependent definitive endoderm in mouse is a Smad2-dependent process. Development 127, autoregulatory enhancer controls the level of Nodal signals in the mouse embryo. 3079–3090 (2000). Development 129, 3455–3468 (2002). 20. Kuzmenkin, A. et al. Functional characterization of cardiomyocytes derived from 35. Teo, A. K. et al. Pluripotency factors regulate definitive endoderm specification murine induced pluripotent stem cells in vitro. FASEB J. 23, 4168–4180 (2009). through eomesodermin. Genes Dev. 25, 238–250 (2011). 21. Saga, Y. Genetic rescue of segmentation defect in MesP2-deficient mice by MesP1 36. Powers, S. E. et al. Tgif1 and Tgif2 regulate Nodal signaling and are required for gene replacement. Mech. Dev. 75, 53–66 (1998). gastrulation. Development 137, 249–259 (2010).

DOI: 10.1038/ncb2304 METHODS

METHODS seeded at 3.7 × 105 cells/6 cm dish in media containing 1% DMSO (Sigma). Generation of the Eomes Cre reporter allele. The EomesCre allele was generated Cells were electroporated with linearized pCAG-EomesER-IRESPuro vector38 and using the same strategy as previously described for the enhanced GFP knock-in selected in 1 µg ml−1 puromycin to generate P19EoER subclones. Expression of the allele18. The targeting vector encompasses a 8.25-kb HpaI fragment of the Eomes EomesER fusion protein was confirmed by Western blot analysis. For EomesER locus. Cre coding sequences were introduced into the exon 1 start site followed activation, 1 µg ml−1 of 4-hydroxytamoxifen (Sigma, H7904) was added to the by a loxP-flanked neomycin-resistance cassette between the SphI (translational culture media. start) and EagI sites, resulting in removal of ∼500 bp of the endogenous 50 0 coding region. The 3 homology arm was flanked by a pMC1.TK negative RNA isolation and one-step and quantitative RT–PCR analysis. RNA was selection cassette. A linearized targeting vector was electroporated into CCE isolated using the RNeasy Kit (Qiagen) according to the manufacturer’s instructions, embryonic stem cells and drug-resistant embryonic stem cell colonies screened using on-column DNase treatment. Complementary DNA was generated using the 0 by Southern blot using a 3 external probe on EcoRV-digested DNA (wild-type SuperScriptIII kit (Invitrogen) with oligo-dT primers. qRT–PCR was carried out allele; 15.2 kb; targeted allele; 8.0 kb). Correctly targeted clones were transfected using the Quantitect SYBRGreen PCR kit and a Rotar-gene Q (Qiagen) and analysed with pMC1.Cre and resulting subclones screened by Southern blot to detect using the 11Ct method, as described previously39. For one-step RT–PCR analysis, the reporter allele (6.2 kb). Two independent excised cell clones were used the one-step RT–PCR kit (Qiagen) was used with gene-specific primers according to to generate germline chimaeras. Offspring were genotyped by PCR using Cre the manufacturer’s instructions. Primer sequences are provided in Supplementary 0 0 specific primers (Cre-fw, 5 -GCATAACCAGTGAAACAGCATTGCTG-3 ; Cre-rev, Table S1. 50-GGACATGTTCAGGGATCGCCAGGCG-30). The strain was maintained on a 129SvEv/C57Bl/6 mixed genetic background. Chromatin immunoprecipitation. Cells were cross-linked for 10 min at room temperature with 1% (v/v) formaldehyde and quenched with 125 mM glycine. Mouse strains, genotyping and generation of chimaeric embryos. All animal Prepared chromatin was sonicated to 200–500 bp and immunoprecipitated with procedures were approved by the Ethical Review Committee of the University of 15 µg of anti-Eomes (Abcam, ab23345), anti-PolII amino terminal (Santa Cruz, Oxford. The ROSA26 R reporter16, ROSA26 gene-trap17, Foxh1 (ref. 12), Smad4CA sc-899x) or normal rabbit IgG (Santa Cruz, sc-2027). Eluted DNA was recovered (ref. 6), EomesCA/N ;Sox2.Cre (ref. 1) and Lhx1 (ref. 32) strains were genotyped as by phenol–chloroform extraction, precipitated with ethanol and resuspended in TE previously described. For generation of chimaeric embryos, blastocysts recovered buffer. ChIP material was analysed using a Rotar-Gene Q (Qiagen) and SYBRGreen from matings of ROSA26 males to CD1 outbred females were injected with 12–14 master mix (Qiagen). The amount of precipitated DNA was compared with the Eomes-null embryonic stem cells1 and transferred into E2.5 pseudopregnant foster starting input material as percentage of input. Each ChIP experiment was carried out females. at least three times on separate biological samples. qPCR was carried out in triplicate. ChIP primer sequences are provided in Supplementary Table S2. Whole-mount in situ hybridization, LacZ staining and histology. Whole- mount in situ hybridization analysis of E6.5–9.5 embryos dissected and fixed with Immunofluorescence. Embryonic stem or P19Cl6 cultures were fixed in 4% 4% paraformaldehyde overnight at 4 ◦C was carried out according to standard paraformaldehyde for 10 min at room temperature and permeabilized with 0.2% protocols37 using probes for Agtrl1, Cre, Eomes, Klhl6, Mesp1, Myl7, Nkx2.5, Rasgrp3 Triton X/PBS for 20 min before blocking for 1 h in 10% FCS, 0.3% BSA, 0.3% Triton and Wnt2. LacZ staining was carried out as described37. For histology, embryos X in PBS. Primary antibodies used include rabbit anti-Eomes (Abcam, ab23345, were post-fixed in 4% paraformaldehyde, dehydrated through an ethanol series, 1:100), rat anti-Eomes (eBioscience, 14-4876, 1:100), goat anti-Gata4 (Santa Cruz, embedded in paraffin, sectioned at 8 µM and eosin counterstained. sc-1237, 1:100), mouse anti-troponin I (Abcam, ab19615, 1:200) and mouse anti-smooth muscle actin (Dako, M0851, 1:100). Primary antibodies were applied in blocking solution overnight at 4 ◦C. Cells were washed with 0.1% Triton X/PBS Definitive endoderm and cardiomyocyte differentiation assays. Wild-type and incubated in blocking solution for 1 h at room temperature with appropriate (CCE), Eomes-null (6A6) and Smad2-null (KT15) feeder-depleted embryonic stem conjugated secondary antibodies: goat anti-rabbit Alexa Fluor 555; goat anti-rat cells were cultured in DMEM (Invitrogen) with 15% FCS, 1% non-essential Alexa Fluor 488; donkey anti-goat Alexa Fluor 488, goat anti-mouse Alexa Fluor −1 amino acids, 0.1 mM β-mercaptoethanol and 1,000 U ml recombinant leukaemia 488 (all Molecular Probes/Invitrogen, 1:500) and donkey anti-mouse Cy3 (Jackson inhibitory factor (Millipore). To induce definitive endoderm differentiation, Immuno Research, 1:2,000). Coverslips were mounted with Vectashield mountant embryonic stem cells were cultured in serum-free ESGRO Complete clonal-grade containing DAPI (Vector Labs, H-1200). Fluorescent images were captured with a medium (Millipore) on 0.1% gelatin-coated dishes for at least two passages before Zeiss epifluorescence microscope or an inverted laser scanning microscope (LSM −1 seeding at low density (18,000 cells ml ) in ESGRO Complete basal medium 510 Meta Duo Live) equipped with a ×25/×63 Plan-Apochromate objective. (Millipore) in bacteriological-grade dishes, to promote embryoid body formation. 50 ng ml−1 ActivinA (R&D Systems) was added after 48 h. For titration of ActivinA Western blot analysis. Cell lysates were prepared using radioimmunoprecipita- effects, the medium was changed after 72 h and new medium added with 5 tion assay (RIPA) buffer, subjected to SDS–polyacrylamide gel electrophoresis and or 50 ng ml−1 ActivinA. For cardiac differentiation, embryonic stem cells were transferred onto polyvinylidene difluoride membranes. Membranes were blocked resuspended at 1 × 104 cells ml−1 in IMDM (Invitrogen) supplemented with 20% with 5% milk powder in Tris-buffered saline with Tween 20, incubated in primary FCS, 1% non-essential amino acids and 0.1 mM β-mercaptoethanol in hanging antibodies overnight including rabbit anti-Eomes (Abcam, ab23345, 1:1,000), rabbit drops (10 µl) plated on the inside lids of bacteriological dishes. After 48 h embryoid anti-phospho-Smad2 (Cell Signaling, 3101, 1:1,000) and mouse anti-γ-tubulin bodies were transferred in 10 ml medium to 10 cm bacteriological dishes. Embryoid (Sigma, T6557, 1:3,000). Secondary antibodies were donkey anti-rabbit horseradish bodies were plated on tissue culture dishes at day 4, allowed to adhere and scored peroxidase (Amersham NA934V, 1:2,000) and goat anti-mouse (Dako, P0447, at d7. They were then plated on fibronectin-coated 15 µm slides 2×9well (Ibidi) 1:2,000). Blots were developed by chemiluminescence using ECL plus (Amersham). and cultured for a further 96 h before imaging and immunostaining. The number and percentage of beating embryoid bodies was counted in three independent experiments. Live-cell images were recorded on a Zeiss Observer Z.1 microscope 37. Nagy, A., Gertsenstein, M., Vintersten, K. & Behringer, R. Manipulating the mouse equipped with an Axiocam MRm camera at frame rates of 10 s−1 and 36 s−1. embryo: a laboratory manual 3rd edn (Cold Spring Harbor Laboratory Press, 2003). 38. Niwa, H. et al. Interaction between Oct3/4 and Cdx2 determines trophectoderm differentiation. Cell 123, 917–929 (2005). P19Cl6 cell culture, differentiation and generation of EomesER-expressing 39. Costello, I., Biondi, C. A., Taylor, J. M., Bikoff, E. K. & Robertson, E. J. Smad4- subclones. P19Cl6 embryonal carcinoma cells were routinely cultured in α-MEM dependent pathways control basement membrane deposition and endodermal cell (Invitrogen) supplemented with 10% FCS. To induce differentiation, cells were migration at early stages of mouse development. BMC Dev. Biol. 9, 54 (2009).

SUPPLEMENTARY INFORMATION

DOI: 10.1038/ncb2304

a E6.5 E6.75 E7.0

E7.25 E7.5 E7.5

EomesCre

b E6.5 E6.75 E7.5

R EomesCre x ROSA26 c Lung Stomach E16.5 E16.5

Small Intestine Pancreas E16.5 E16.5

R EomesCre x ROSA26

Figure S1 The EomesCre reporter allele faithfully recapitulatesSupplementary endogenous Figurereveal 1_Arnold Cre-mediated excision selectively within the extra-embryonic ectoderm Eomes expression during gastrulation and marks endodermal components and the emerging cells of the primitive streak. Dashed white lines indicated of gut derived organs. (a) WISH analysis of Cre mRNA expression at early LacZ positive migrating cardiac progenitors. The black arrow indicates stages of post-implantation development (E6.5-E7.5). At the onset of definitive endoderm cells migrating anteriorly. (c) EomesCre expression gastrulation, Cre transcripts are detectable in the proximal posterior region specifically marks endodermal but not mesodermal tissues within organs within the emerging primitive streak, as well as in the extra-embryonic derived from the gut tube. Animals carrying the EomesCre allele were mated ectoderm. As gastrulation proceeds, Cre is expressed throughout the streak. with ROSA26R LacZ reporter mice. Organs taken from E16.5 were stained Slightly later at E7.5, Cre remains predominantly in the anterior primitive for LacZ expression. Endoderm-derived cells lining the alveolar ducts and streak before expression becomes down-regulated prior to head-fold stages of bronchi, the epithelium of the stomach and small intestine as well as the development. (b) Fate mapping from the EomesCre;ROSA26R LacZ reporter glandular tissues of the pancreas all express LacZ whereas mesenchymal cross. Post-implantation embryos (E6.5-E7.5) stained for LacZ expression cells are unlabelled.

SUPPLEMENTARY INFORMATION

Feeder-free d0: Embryoid body d2: Activin ESC Defined medium formation induction d3-5: Analysis

DE-differentiation protocol

b GFP Eomes locus

D4 D4 D4

+ActA

-ActA

BF Dapi GFP

c Feeder-free d0: Embryoid body/ d2: non-adherent ESC +20% Serum hanging drops dish d4: adherent dish d5-8: Analysis

cardiac-differentiation protocol

d ES wt Eomes -/-

Supplementary Figure 2_Arnold Figure S2 ES cell differentiation protocols for the generation of definitive Activin addition) were fixed and mounted onto coverslips to visualise endoderm (DE) and cardiomyocytes. (a) Schematic of the protocol used eGFP expression. Blue, DAPI; Green, GFP-expression from the EomesGFP to differentiate ES cells to DE under chemically defined conditions in the reporter allele. (The punctate green signal evident in untreated cells is absence of feeder cells. Beginning at day 0 (d0), ES cells were grown at due to autofluorescence). (c) For differentiation to cardiomyocytes cells low density in suspension cultures to form EBs. High concentration (50ng/ were cultured in high serum in hanging drops. At day 4 the resulting EBs ml) ActivinA recombinant protein was added at d2 cells. (b) ES cells were allowed to adhere to tissue culture dishes. (d) Appearance of EBs harbouring an EomesGFP reporter allele18 were used to monitor Eomes after 4 days in suspension culture, and outgrowths formed after 8 days of induction. EBs at day 4 (48 hrs post 50ng/ml ActivinA treatment, or without differentiation. Scale bar, 500mm.

SUPPLEMENTARY INFORMATION

a Brachyury Eomes

b Mesp1_TSS Mesp1/2_EME Mesp1

Human Gene Exon Rhesus UTR Dog CNS Horse

c d Mesp1_TSS Mesp1_EME

e Mesp2_PSME Mesp2

Human Gene Exon Rhesus UTR Dog CNS Horse

f g Mesp2_PSME (Site B) Mesp2_PSME (Site G & D)

Figure S3 Inter-species sequence alignments of the Mesp1/2 genomic locus degree of conservation. The newly identified T-box site near the TS site is identifies evolutionarySupplementary conserved T-box sites. (a) TheFigure consensus 3_Arnoldsequence also highly conserved. The T-box sites are indicated with a red line above the binding site of the T-box protein Brachyury as identified from the JASPAR sequence, while complete sequence conservation is indicated underneath by an transcription factor binding database (http://jaspar.genereg.net/), in comparison asterisk (*).CNS, conserved non-coding sequence; UTR, untranslated region; to the inferred Eomes binding site (derived from data in in Badis et al., TSS, transcription start site; EME, early mesoderm enhancer. (e) Interspecies 2009), reveal an almost identical T-box half site motif. (b) Genomic regions of alignment of the Mesp2 locus. PSME, pre-somitic mesoderm enhancer. (f, g) orthologous Mesp1 gene from numerous vertebrate species (indicated on left) ClustalW alignments of Sites B, G & D, from the PSME identified previously were aligned against the mouse Mesp1 genomic region using the VISTA browser in the Mesp2 mouse locus26, 27, also reveal a high degree of sequence program. (c, d) Alignments of genomic sequences adjacent to the Mesp1 TS conservation. Badis, G et al. Diversity and complexity in DNA recognition by site (c) and within Mesp1 EME (d) using the ClustalW program reveal a high transcription factors. Science 324, 1720-1723 (2009).

SUPPLEMENTARY INFORMATION

a P19Cl6 b DMSO differentiation P19EoER Eomes differentiation Mesp1 Eomes Cxcr4 Mesp1 Foxa2 Cxcr4 Gsc Foxa2 Sox17 Gsc Nkx2.5 Brachyury Gata4 MixL1 Brachyury Tbx3 MixL1 Tbx3 Oct4 Oct4 Hprt

Hprt d0 d2 d2 d4 d4

- Tx DS Tx DS neg. d0 d2 d4 d6 d8 control WT neg. d4 EB control

c f Dapi Eomes Eomes/Gata4 SMA/Dapi TropI/Dapi - Tx d14 DMSO +Tx d Eomes Eomes/Dapi g -Tx

+Tx

e P19EoER Clone1 Clone2 Clone3 P19Cl6 [kDa] - + - + - + - + Tx α-Eomes 100

Figure S4 P19Cl6 in vitroSupplementary differentiation. (a) P19Cl6 Figure cells cultured 4_Arnold in of DMSO treatment P19Cl6 cultures contain numerous cardiomyocytes as the presence of 1% DMSO were analysed for expression of pluripotency judged by smooth muscle actin (SMA) and TroponinI (TnI) staining. (d) The (Oct4), mesoderm (Brachyury, Mixl1) and endoderm (Foxa2, Gsc, Sox17) majority of tamoxifen-treated P19EoER cells are stained with the Eomes marker genes by RT-PCR. Transient Eomes expression peaks at day2 of antibody. (e) The EomesER fusion protein is greatly stabilised by tamoxifen differentiation. Eomes activation precedes robust induction of the cardiac treatment, coincident with nuclear translocation. (f) Eomes activation in mesoderm marker Mesp1, as well as other cardiac markers (Nkx2.5, Gata4). P19EoER cells results in co-expression of the cardiac transcription factor (b) Activation of a tamoxifen-inducible EomesER fusion protein in P19EoER Gata4 at early stages of differentiation. (g) Q-PCR analysis reveals Mesp1 cells similarly initiates cardiac differentiation. RT-PCR analysis at days 2 and expression is increased 24 hours after addition of tamoxifen in P19EoER 4 after tamoxifen (Tx) treatment confirms Eomes directly activates Mesp1 cultures but not DMSO treated control cultures. Tx, Tamoxifen; DS, DMSO. and induction occurs earlier than in DMSO induced cells. (c) After 14 days Scale bar, 50mm.

SUPPLEMENTARY INFORMATION

1.2

0.8 d0_Round1 d4_Round1 0.6 d0_Round2 d4_Round2 % Input d0_Round3 0.4 d4_Round3 Percentage of input

0.2

0 Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp2_Con

3.5

2.5 Con_Round1 Tam_Round1 2 Con_Round2 Tam_Round2 Con_Round3 1.5 Tam_Round3

Percentage of input 1

0.5

0 Mesp1_Ex2 Mesp1_TSS Mesp1_Tbx Mesp2_Tbx Mesp2_Con

Supplementary Figure 5_Arnold

Figure S5 Eomes binds to conserved sites within the Mesp1/2 locus. sites within the Mesp1/2 locus after 4 days of tamoxifen treatment. As above, (a) P19Cl6 cells treated for 4 days with DMSO were analysed by ChIP the ChIP Q-PCR results of three independent experiments (biological replicates) using an Eomes-specific antibody. Q-PCR results from three independent are presented. Binding to each of the T-box sites (Mesp1_TSS, Mesp1_Tbx, ChIP experiments (Round1-3) are shown. d0, day 0; d4, day 4 of DMSO Mesp2_Tbx) is clearly evident. There was no detectable enrichment of Eomes differentiation of P19Cl6 cells.(b) Cells expressing the tamoxifen inducible at the control regions (Mesp1_Ex2, Mesp2_Con). Con, non-tamoxifen induced Eomes fusion construct, P19EoER cells, reveal Eomes binding to the T-box control P19EoER cells; Tam, day 4 tamoxifen-treated P19EoER cells.

SUPPLEMENTARY INFORMATION

a b Eomes

Mesp1 Eomes Sox17

β-actin

ActA [ng/ml] 00 50 50 5 d0 d2 d3 d4 d4 Mesp1 pos. control neg. control

c [kDa] Sox17 75 α-Eomes

50 α-pSmad2 ActA [ng/ml] 00 50 50 5

d0 d2 d3 d4 d4 50 α-γ-tubulin

ActA [ng/ml] 00 50 50 5 d0 d2 d3 d4 d4

Eomes Mesp1

Eomes Mesp1 Lefty2 Foxa2 Fz8

Figure S6 Dynamic and dose-dependent TGFb signalling induces cardiac of high ActivinA levels favours definitive endoderm formation, indicated by vs. endoderm progenitors.Supplementary (a, b) Wildtype ES cells wereFigure differentiated 6_Arnold elevated Sox17 expression. (d) Eomes expression in the epiblast proceeds under serum-free conditions and stimulated after 48hrs (d2) with high- that of Mesp1. The Mesp1 expression domain becomes apparent prior to dose ActivinA (50ng/ml). After 72hrs (d3) of the experiment, medium was primitive streak initiation in the most proximal epiblast and the domain changed and either high-dose ActivinA levels (50ng/ml) maintained, or extends following streak formation reflecting migration of nascent mesoderm ActivinA administration lowered to 5ng/ml. The mRNA expression of Eomes, (red arrow). (e) Stage matched embryos at the early amniotic fold stage Mesp1 (cardiac progenitors) and Sox17 (definitive endoderm progenitors) stained for expression of Eomes, Mesp1, Lefty2, Foxa2 and Frizzled8 (Fz8). was monitored by semi-quantitative (a) and quantitative RT-PCR (b). Error The proximal extent of the epiblast is marked by a green line and the extent bars represent s.e.m. (c) Eomes protein expression levels and phosphorylation of the cardiac mesoderm expressing Mesp1 and Lefty2 in the proximal status of Smad2 (pSmad2) were analysed by Western blot. Overall, reducing epiblast by a red line. Expression of Foxa2 and Fz8, in response to continued TGFb signalling thresholds after the initial high-dose ActivinA induction expression of Eomes in the distal region of the extending streak, shows phase significantly promotes cardiac fates whereas continued maintenance induction of the discrete DE progenitor population.

SUPPLEMENTARY INFORMATION

Fig. 3a HPRT HPRT HPRT

ES wt Eomes -/- Smad2 -/-

Bra Mixl1 Eomes Foxa2 Sox17 ES wt

Bra Mixl1 Eomes Foxa2 Sox17 Eomes -/-

Bra Mixl1 Eomes Foxa2 Sox17 Smad2 -/-

Fig. 3c Eomes Gata6 Myl7

Myocardin Mesp1 Nkx2.5

Mef2c Mesp2 beta-actin

Bra Myl2

Suppl.Fig. 4e

α-Eomes

EomesER-expression in P19Cl6 +/- Tx

Suppl. Fig. 6a Suppl. Fig. 6c Eomes α-Eomes

Mesp1 α-pSmad2

Sox17

beta-actin α-γ-Tubulin

Supplementary Figure 7_Arnold Figure S7 Original full scans corresponding to Fig. 3a, Fig. 3c, Supplementary Fig. S4e and Supplementary Fig. S6a, c. As size marker for RT-PCRs the 1kbPlus ladder was used (Invitrogen).

SUPPLEMENTARY INFORMATION

Table S1 RT-PCR primers. Forward and reverse primer sequences used for semi-quantitative RT-PCR and qRT-PCR analysis of indicated genes.

Table S2 ChIP primers. Primer pair name, location within the Mesp1/2 locus, and forward and reverse primer sequences used for qChIP analysis.

Supplementary Movie S1 Differentiation of wildtype ES cells. Real-time movie showing multiple foci of beating cardiomyocytes within a colony generated from one embryoid body from differentiated wild-type ES cells (CCE) at day 8 of the experiment.

Supplementary Movie S2 Differentiation of Eomes mutant ES cells. Real-time movie showing the absence of beating cardiomyocytes within a colony generated from one embryoid body from differentiated Eomes null ES cells (6A6) at day 8 of the experiment.

Supplementary Table S1: RT-PCR primers

Gene Name Forward Primer Sequence Reverse Primer Sequence Product Brachyury AACTTTCCTCCATGTGCTGAGAC TGACTTCCCAACACAAAAAGCT 533 bp Cxcr4 GGCTGTAGAGCGAGTGTTGC GTAGAGGTTGACAGTGTAGAT 390 bp Eomes TGTTTTCGTGGAAGTGGTTCTGGC AGGTCTGAGTCTTGGAAGGTTCATTC 323 bp Foxa1 CATGAGAGCAACGACTGGAA TGTTGCTGACAGGGACAGAG 72 bp Foxa2 TGGCTGCAGACACTTCCTACT CAACATCAGTACAACCCTCTGGT 487 bp Gata4 GCCTGTATGTAATGCCTGCG CCGAGCAGGAATTTGAAGAGG 500 bp Gata6 GCCGCACCGCTGACTCCTG ACGCGCTTCTGTGGCTTGATGA 278 bp Gsc AAACGCCGAGAAGTGGAACAAG AAGGCAGGGTGTGTGCAAGTAG 177 bp Lhx1 CCCAGCTTTCCCGAATCCT GCGGGACGTAAATAAATAAAATGG 74 bp Mef2c AGGCACCAGCGCAGGGAATG CCACCGGGGTAGCCAATGACT 239 bp Mesp1 TGTACGCAGAAACAGCATCC TTGTCCCCTCCACTCTTCAG 144 bp Mesp2 CGCCTGGCCATCCGCTACAT CACCCCCAGGACACCCCACTACT 203 bp Mixl1 ACTTTCCAGCTCTTTCAAGAGCC ATTGTGTACTCCCCAACTTTCCC 487 bp Myl2 GCCAAGAAGCGGATAGAAGGCGGG CTGTGGTTCAGGGCTCAGTCCTTC 500 bp Myl7 TTCTAATGTCTTCTCAATG GAAGCTGCTTGAACTCTTCC 326 bp Myocardin AGTGGGCCCAGCATTTTCAACATC CCCTCCCCATTTTCCCCACTTC 196 bp Nkx2.5 CAGCAACTTCGTGAACTTTGGC AATCTGAGGGACAGGGCATAGTGG 212 bp Oct4/Pou5f1 CGTTCTCTTTGGAAAGGTGTTC GAACCATACTCGAACCACATCC 319 bp Sox17 TTTGTGTATAAGCCCGAGATGG AAGATTGAGAAAACACGCATGAC 448 bp Tbx3 TTATTTCCAGGTCAGGAGATGGC GGTCGTTTGAACCAAGTCCCTC 397 bp Tbx6 TGGAGAACCAGGAACTGTGGAAGG CCAGAAGAAACAAGTAGCGGGC 141 bp Hprt GCTGGTGAAAAGGACCTCT CACAGGACTAGAACACCTGC 249 bp β-Actin ACCAACTGGGACGACATGGAGAAGA TACGACCAGAGGCATACAGGGACA 214 bp

Supplementary Table S2: ChIP primers

Primer pair Locus Forward Primer Sequence Reverse Primer Sequence name Mesp1_Tbx Mesp1 TGGGTTCCTTTGGGAGCTGCTTGG CTCTCACCTCTGTTCTGATGGGG Mesp1_TSS Mesp1 TTGGGTCCGCTGTAGCTTTATC GCTATGGTTCAAAGGGTTCAGGC Mesp1_Ex2 Mesp1 AAACCCTCCAAATGACACTAGCAC ACCATTCCAACCCTCCTTGG Mesp2_Tbx Mesp2 CGGGATAGACATCCAGGTACCCA GGCTGGTGTGACTCTGGGAAGCT Mesp2_Con Chr7 GGTCTGTTTGCAGCTGATTCTGAA CAGTTCTCACCTTGCTTCCATGT