Novel pre- and post-gastrulation expression of Gata4 within cells of the inner cell mass and migratory neural crest cells


  • Nicolas Pilon,

    1. Department of Veterinary Biomedicine, Faculty of Veterinary Medicine, University of Montreal, St-Hyacinthe, QC, Canada
    2. Département des sciences biologiques, Faculté des sciences, Université du Québec à Montréal, Montréal, QC, Canada
    Search for more papers by this author
  • Diana Raiwet,

    1. Department of Veterinary Biomedicine, Faculty of Veterinary Medicine, University of Montreal, St-Hyacinthe, QC, Canada
    Search for more papers by this author
  • Robert S. Viger,

    1. Ontogeny-Reproduction Research Unit, CHUQ Research Centre and Centre de Recherche en Biologie de la Reproduction (CRBR), Department of Obstetrics and Gynecology, Faculty of Medicine, Université Laval, Quebec City, QC, Canada
    Search for more papers by this author
  • David W. Silversides

    Corresponding author
    1. Department of Veterinary Biomedicine, Faculty of Veterinary Medicine, University of Montreal, St-Hyacinthe, QC, Canada
    • Department of Veterinary Biomedicine, Faculty of Veterinary Medicine, University of Montreal, St-Hyacinthe, QC, Canada, J2S 7C6
    Search for more papers by this author


GATA4 is a transcription factor known to be important for the development of many organs such as the heart, intestine, and gonads. However, information regarding the control of its expression is only now beginning to emerge. To further understand the regulation of Gata4 expression during mouse embryonic development, we have generated a novel knockin allele allowing expression of the Cre recombinase under the control of Gata4 regulatory sequences. When these Gata4Cre/+ mice were crossed with the Cre reporter mouse R26R-YFP, we surprisingly found widespread mosaic YFP expression in e10.0 embryos. This particular expression pattern was traced back to the e5.5 stage via a cell lineage study, suggesting activation of transcription at the Gata4 locus around the blastocyst stage. In accordance with this hypothesis, we found that Gata4 is expressed in cultured embryonic stem (ES) cells and within the inner cell mass (ICM) of e4.5 blastocysts. Interestingly, such early Gata4 transcription can be recapitulated in transgenic reporter studies using 5 kb of the proximal rat Gata4 promoter. During mouse development, these 5-kb regulatory sequences were previously reported to direct reporter gene expression to Sertoli cells of the testes [Mazaud Guittot et al. (2007) Biol Reprod 76:85–95]. We now show that these regulatory sequences can also drive robust fluorescent reporter gene expression in migratory neural crest cells. Comparisons to Wnt1-Cre-mediated YFP labelling of neural crest cells suggest that most of the migratory neural crest cells are labelled in e9.5 to e11.5 Gata4p[5kb]-RFP or -GFP embryos. Analysis of GFP transcription via whole-mount in situ hybridization in e10.5 and e11.5 embryos demonstrated that the 5-kb Gata4 promoter is preferentially active in cells of the boundary caps at the dorsal root entry zone and motor exit points flanking the neural tube. RT-PCR gene expression analysis of FACS-purified GFP-positive cells from e9.5 Gata4p[5kb]-GFP embryos revealed co-expression of Gata4 with many neural crest stem cell markers. Together with sphere-forming and differentiation cell culture assays, our results indicate that the Gata4 promoter is active within at least a subset of the neural crest stem cells. Taken altogether, our studies have revealed new Gata4 expression patterns during mouse embryonic development, which are controlled by its 5-kb proximal 5′ flanking sequences. Developmental Dynamics 237:1133–1143, 2008. © 2008 Wiley-Liss, Inc.


The zinc finger containing GATA transcription factors plays key roles in gene activation in multiple cell lineages. Based on sequence homology and expression patterns, the six mammalian members of the evolutionary conserved GATA family are subdivided in two subgroups: GATA1/2/3 and GATA4/5/6. GATA1/2/3 factors are mostly but not exclusively involved in specification of hematopoietic lineages whereas GATA4/5/6 factors are involved in the development of mesodermal- and endodermal-derived tissues, such as heart, gut, liver, and gonads (Molkentin,2000). In spite of the fact that most Gata functional analyses have focused on mesodermal and endodermal cell lineages, accumulating data also suggest important Gata functions in embryonic neurectodermal cells. Using gene targeting approaches, Gata2 and Gata3 have been shown to be involved in the development of some motor neurons of the central nervous system (Pata et al.,1999; Zhou et al.,2000). Gata2 and Gata3 are also required for proper development of neural crest–derived structures such as the sympathetic branch of the autonomic nervous system and the cranio-facial skeleton (Lim et al.,2000; Ruest et al.,2004; Tsarovina et al.,2004). GATA6 protein is also detected in the forebrain, the motor columns of the developing neural tube, as well as in neural crest–derived adrenal medulla cells and pre-cartilaginous primordia of the first branchial arch and nasal process, but its role within these cell populations is currently not known (Brewer et al.,2002; Nemer and Nemer,2003).

In the mouse, initiation of Gata4 gene expression has been described in mesodermal and endodermal cells of the e7.0 gastrulating embryo and is later maintained in several endomesodermal-derived tissues (Molkentin,2000; Nemer and Nemer,2003). Besides these well-recognized expression patterns, Gata4 has also been reported to be expressed in discrete neurectodermal derivatives such as in primitive neural stem cells, in hypothalamic gonadotropin-releasing hormone (GnRH) neurons, and in some neural crest–derived heart stem cells (Lawson and Mellon,1998; Hitoshi et al.,2004; Tomita et al.,2005). Loss-of-function analyses have demonstrated early vital roles for Gata4 in different steps of cardiogenesis (Kuo et al.,1997; Pu et al.,2004; Watt et al.,2004). These different early lethal phenotypes complicate functional analyses in other tissues. However, we and others have also demonstrated important roles for Gata4 in testis development and function (Tevosian et al.,2002; LaVoie,2003; Viger et al.,2004,2005; Manuylov et al.,2007).

Despite its importance in embryogenesis and although many GATA4 target genes have been identified, surprisingly little is known about the transcriptional mechanisms responsible for Gata4 expression itself. Given its expression in different germ layers, it can be hypothesized that different transcriptional enhancers direct Gata4 expression to specific cell lineages. In support of this hypothesis, a distal enhancer mediating Gata4 expression within the lateral mesoderm was identified in a region located 40 kb upstream of the transcriptional start site (Rojas et al.,2005). We have also shown that, in the gonads, the 5-kb proximal promoter can drive reporter gene expression specifically in Sertoli cells and not in Leydig cells or peritubular testicular cells, nor in pregranulosa, granulosa, or theca ovarian cells (Mazaud Guittot et al.,2007).

To further understand the regulation of Gata4 expression during embryonic development, we have generated a new knockin allele allowing expression of the Cre recombinase under the control of endogenous Gata4 regulatory sequences. Surprisingly, these Gata4Cre/+ mice revealed activation of embryonic Gata4 transcription as early as at the blastocyst stage. Moreover, we found that this novel expression pattern is recapitulated in vivo by the previously described 5-kb Gata4 proximal 5′ flanking sequences (Mazaud Guittot et al.,2007). We now also report Gata4 expression in a subset of the neural crest cells (NCC) and that Gata4p[5kb]-GFP and -RFP reporter mouse lines direct robust fluorescent reporter expression in the migratory NCC and the developing peripheral nervous system.


A New Gata4Cre Allele Reveals a Novel Gata4 Expression Pattern in Early Implanted Embryos

To gain further insights into the regulation of Gata4 gene expression during mouse development, we derived mice containing a novel Gata4 knockin reporter allele (Gata4Cre). Via homologous recombination in embryonic stem (ES) cells, we replaced Gata4 exon 2, containing the ATG start codon, with a bicistronic DsRed2-IRES-Cre expression cassette to report endogenous Gata4 expression (Fig. 1A). Chimeras generated from targeted ES cells gave germline transmission and F1 animals were used to evaluate reporter gene expression. Expression of the Red Fluorescent Protein (RFP) from the DsRed2 gene, via direct fluorescence detection, was never detected in embryonic tissues from these animals (data not shown). To evaluate expression of the Cre recombinase, Gata4Cre/+ animals were bred to the Cre reporter mouse line R26R-EYFP. In the R26R-EYFP background, expression of the Yellow Fluorescent Protein (YFP) under the control of the ubiquitous ROSA26 promoter is obtained upon tissue-specific Cre-mediated excision of an interfering cassette (Srinivas et al.,2001). Therefore, not only cells expressing Cre but also their progeny are YFP-labelled and, as such, this model has been widely used for cell lineage studies (Srinivas et al.,2001; Lo et al.,2005; Matsuoka et al.,2005; Laiosa et al.,2006; Samokhvalov et al.,2007). Analysis of e10.0 embryos from crosses of Gata4Cre/+ males and R26R-EYFP females revealed a widespread but mosaic pattern of YFP expression as detected by direct yellow fluorescence (Fig. 1C). This suggested an early zygotic Cre-mediated recombination event, and via similar trials we were able to trace back the initiation of YFP expression in Gata4Cre/+::R26R-EYFP embryos to the e5.5 stage (Fig. 1E, E').

Figure 1.

Generation of a new Gata4 knockin allele. A: Schematic representation of the Gata4 locus and targeting vector. The targeting construct was designed to allow replacement of Gata4 exon2 (E2) by a DsRed2-IRES-Cre bicistronic expression cassette. A neomycin cassette (Neo) was also included to allow G418-mediated selection in ES cells. The internal probe (black box) was used to screen for targeted ES cells and confirm the predicted targeting event. B: Southern blot analysis of genomic DNA from wild-type and heterozygous (+/−) targeted offspring using the internal probe and the indicated restriction endonucleases (B, BamHI; E, EcoRI; S, SacI). C,D,D',E,E': The ESpNP26.23 mouse line (Gata4Cre/+) was mated to a Cre reporter line (R26R-YFP). C,D',E': Cre-mediated activation of YFP expression allows cell lineage tracing and reveals widespread mosaic expression in e10.0, e7.5, and e5.5 embryos. D,E: Same embryo as in D' and E' but without a YFP filter. Images were taken at 20× (C) and 100× (D,D',E,E') magnification.

The Gata4 5-kb Proximal Promoter Recapitulates Gata4 Expression in Early Implanted Embryos

Given the widespread YFP fluorescence detected in e5.5 Gata4Cre/+::R26R-EYFP embryos and as this model needs the sequential expression of two proteins, this suggested that embryonic Gata4 transcription initiates around the blastocyst stage. To verify this hypothesis, we first performed a semi-quantitative RT-PCR analysis on wildtype and Gata4Cre/+ ES cells (Fig. 2A). This analysis revealed that Gata4 is expressed in these ES cell cultures but at low levels. Indeed, Gata4 transcripts were weakly detectable after 25 PCR cycles and clearly detected after 30 cycles, whereas expression of the ES cell marker Oct4 was clearly detected after only 20 cycles. Similar outcomes were also obtained with the female VP10 ES cell line (data not shown). In Gata4Cre/+ ES cells, this analysis also revealed that Cre transcripts were slightly more abundant than Gata4 transcripts, this being most likely due to a higher stability of the Cre mRNA. To exclude the possibility that detection of Gata4 expression was not due to the presence of already differentiated primitive endoderm cells in the cultures (Fujikura et al.,2002), we took advantage of the Gata4Cre/+ ES cells and used an in vitro transient transfection approach reminiscent of the cell lineage assay described above. To this end, we made use of a R26R-GFP plasmid, which allows expression of the Green Fluorescent Protein (GFP) under the control of the ROSA26 800-bp proximal promoter after Cre-mediated excision of a Lox-STOP-Lox interfering cassette. GFP-labelled cells were specifically identified in Gata4Cre/+ ES cell cultures, as soon as 18h post-transfection, and later formed GFP-labelled clones in the form of tightly compacted colonies (Fig. 2E). In contrast to undifferentiated ES cells, which grow as tightly compacted colonies, it has been shown that differentiated primitive endoderm cells grow as dispersed refractive cells with stellate morphology in culture (Jetten et al.,1979; Fujikura et al.,2002). Therefore, according to these morphological criteria, we conclude that most of the GFP-positive cells were undifferentiated-like cells. Only very few flattened GFP-positive differentiated-like cells were identified.

Figure 2.

Gata4 is expressed in undifferentiated ES cells. A: RT-PCR analysis showing semi-quantitative expression of Gata4 within wildtype and Gata4Cre/+ ES cells. Expression of Gata4 after 20, 25, 30, and 35 PCR cycles is compared with expression of the ES cell marker Oct4. Expression of Cre is also evaluated in the Gata4Cre/+ ES cell line. B–E: Evaluation of Cre recombinase activity in ES cells via transfection of the R26R-GFP plasmid. Thirty-six hours after transfection, direct GFP fluorescence is specifically detected within Gata4Cre/+ (D,E) but not within wildtype ES cells (B,C). E: Note that GFP fluorescence is mainly detected within round cells of undifferentiated-like ES cell clones (arrows) in addition to very few isolated and elongated differentiated-like cells (arrowhead).

We then used whole-mount in situ hybridization (ISH) to evaluate Gata4 expression in blastocysts. This analysis revealed Gata4 expression in the ICM of e4.5 embryos (Fig. 3B). This particular Gata4 expression pattern, as detected by whole-mount ISH, appears to be transient as it was undetectable at e3.5, barely detectable at e5.5, and undetectable again at e6.5 (data not shown). Interestingly, we found that this early expression pattern is recapitulated by a GFP reporter transgene driven by 5 kb of rat Gata4 proximal flanking sequences (Gata4p[5kb]-GFP) (Mazaud Guittot et al.,2007). At e3.5, low levels of GFP fluorescence are specifically detected in cells of the ICM of Gata4p[5kb]-GFP blastocysts (Fig. 3C,D). At e5.5, these Gata4 proximal flanking sequences can drive reporter gene expression in the whole embryo in addition to the previously described extra-embryonic expression in the visceral endoderm (Fig. 3E, F) (Heikinheimo et al.,1994; Morrisey et al.,1998). In accordance with the in situ data described above, GFP fluorescence in the embryo becomes barely detectable at e6.5 and undetectable at e7.5 (data not shown); the one-day delay between ISH and GFP reporter data is most likely attributable to the stability of the GFP protein, which is described to have a half-life of about 26 hr (Li et al.,1998; Corish and Tyler-Smith,1999). Taken together, our data demonstrate that Gata4 is expressed in ES cells in vitro and in vivo.

Figure 3.

Gata4is expressed in blastocysts and pre-gastrulation embryos. A,B: Whole-mount in situ hybridization using Gata4 sense (A) and antisense (B) riboprobes demonstrating Gata4 expression in the ICM of e4.5 blastocysts. C–F: The Gata4 5-kb proximal promoter can drive expression of a GFP reporter gene (Gata4p[5kb]-GFP) within the ICM of e3.5 blastocysts (C,D) as well as in the embryonic (e) and extra-embryonic (Ex-e) compartments of e5.5 embryos (E,F). D,F: Same images as in C and E, respectively, but with a GFP filter. Images were taken at 80× (A,B), 100× (E,F), or 200× (C,D) magnification.

The Gata4 5-kb Proximal Promoter Directs Robust Reporter Gene Expression in Neural Crest Cells

In addition to the early Gata4 expression pattern described above, we have previously shown that the Gata4 5-kb flanking sequences drive reporter gene expression in Sertoli cells of the developing testes (Mazaud Guittot et al.,2007). We now report that, in five different mouse lines (Gata4p[5kb]-RFP lines 2 and 5 as well as Gata4p[5kb]-GFP lines 1, 2 and 6a), these Gata4 5-kb flanking sequences also direct strong reporter gene expression within migratory NCC (Fig. 4). Comparisons to the Wnt1-Cre::R26R- EYFP model, in which pre-migratory and migratory NCC are fluorescently-labelled (Danielian et al.,1998; Srinivas et al.,2001), suggested that a large part of the migratory NCC are RFP-labelled in e11.5 Gata4p[5kb]-RFP embryos (Fig. 4A,B). To validate these observations, we generated Wnt1-Cre::R26R-EYFP::Gata4p[5kb]- RFP triple transgenic embryos and evaluated YFP and RFP co-localization in e11.5 embryos. Analysis of whole-mount e11.5 Wnt1-Cre::R26R-EYFP::Gata4p[5kb]-RFP embryos showed that the vagus nerve and dorsal root ganglia are co-labelled by YFP and RFP fluorescence (Fig. 4C,E,G). Note that co-localization within the other cranial nerves is masked by the strong YFP expression in the branchial arches. Examination of vibratome sections of the trunk of Wnt1-Cre::R26R-EYFP::Gata4p[5kb]-RFP embryos via confocal microscopy clearly revealed extensive overlap of YFP and RFP fluorescence in NCC of the dorsal root ganglia, spinal nerves, and sympathetic ganglia (Fig. 4D, F, H). These observations demonstrate that the Gata4 5-kb 5′-flanking sequences drive reporter gene expression within most of the migratory NCC.

Figure 4.

The Gata4 proximal promoter directs reporter gene expression in neural crest cells. A,B: Comparison of the fluorescence labelling of the neural crest cells (NCC) in Gata4p[5kb]-RFP (A) and Wnt1-Cre::R26R-YFP double transgenic (B) e11.5 embryos. C–H: Overlapping fluorescent protein expression patterns in the whole cervical region (C,E,G) or in a 150 -μm transversal vibratome section of the trunk (D,F,H) of a e11.5 Gata4p[5kb]-RFP::Wnt1-Cre::R26R-YFP triple transgenic embryo. C,D: Detection of fluorescence with the RFP filter. E,F: Detection of fluorescence with the YFP filter. G,H: Co-detection of RFP and YFP fluorescence. Note the overlap in the vagus nerve (X), the dorsal root ganglia (DRG), the sympathetic ganglia (SG), and the spinal nerves (arrowheads).

We then carefully assessed the pattern of reporter gene expression in e9.5 to e11.5 Gata4p[5kb]-GFP embryos via direct fluorescence detection. This analysis revealed that the migratory NCC from all antero-posterior sub-domains (cranial, cardiac, vagal, and trunk) are GFP-labelled as soon as they leave the dorsal neural tube (Fig. 5A,B,B'). At e10.5 and e11.5, this GFP labelling becomes restricted to the developing peripheral nervous system and, in contrast to the Wnt1-Cre::R26R-EYFP model, is now barely detectable in the head mesectoderm (Fig. 5C and E; compare with Fig. 4B). This is compatible with the half-life of GFP (Corish and Tyler-Smith,1999) and with its use as a temporally restricted (24–48 hr) lineage marker. Starting around e11.5, the melanocytes are also GFP-labelled in Gata4p[5kb]-GFP embryos and the GFP labelling of both melanocytes and PNS persists into adulthood (data not shown). We also analyzed the transcription of the Gata4p[5kb]-GFP transgene via whole-mount ISH. At e9.5, no overt differences are seen between GFP transcription and direct GFP fluorescence detection (Fig. 5A,B). However, at later stages a more refined expression pattern becomes evident in the trunk on either side of the neural tube (Fig. 5D and F; compare with C and E). In e10.5 and e11.5 embryos, in contrast to fluorescence detection, GFP transcripts are not detected in cells of the dorsal root ganglia (DRG) but are rather strongly detected in the boundary caps (BC) formed at the dorsal root entry zone and motor exit points between the neural tube and the DRG (Fig. 5D' and F'; compare with Fig. 4D). Interestingly, the BC have been described as a niche for multipotent neural crest stem cells (NCSC) (Hjerling-Leffler et al.,2005). We also performed whole-mount ISH to evaluate endogenous Gata4 expression in NCC by this method. These analyses gave inconsistent results suggesting that expression levels are below detectable levels for ISH and/or transcription is more restricted in comparison to what is seen for the Gata4p[5kb]-GFP transgene (data not shown and see below).

Figure 5.

Expression of the Gata4p[5kb]-GFP transgene during early development of the peripheral nervous system. A,C,E: Direct GFP fluorescence detection in e9.5 to e11.5 Gata4p[5kb]-GFP embryos showing labelling of the migratory NCC along the whole rostro-caudal axis at e9.5 (A) and the developing peripheral nervous system at e10.5 (C) and e11.5 (E). c, cranial NCC; t, trunk NCC; ca, cardiac NCC; v, vagal NCC; V, trigeminal nerve; VII, facial nerve; VIII, vestibulocochlear nerve; IX, glossopharyngeal nerve; X, vagus nerve; DRG, dorsal root ganglia; SG, sympathetic ganglia; H, heart. Note that the Gata4 5-kb promoter directs strong reporter gene expression in the heart. Also, specifically in the Gata4p[5kb]-GFP line 6a (shown here), an unspecified labelling is detected in the floor and roof plates of prosomere 1 (asterisk in C). B,D,F:GFP whole-mount in situ hybridization showing transcription from the Gata4p[5kb]-GFP transgene. The dotted lines indicate the level at which the 150-μm vibratome sections in B', D', and F' were made. B',D',F': Note that GFP transcripts are detected in migratory NCC on either side of the neural tube (NT), in enteric neural crest cells (eNCC) invading the foregut, and in the sympathetic ganglia (arrowhead) flanking the dorsal aorta. D,D',F,F': In e10.5 and e11.5 embryos, flanking NT expression becomes restricted to cells of the boundary caps at the dorsal root entry zone and motor exit points (BC in D and F, arrows in D' and F').

Gata4 Is Coexpressed With NCSC Markers in Sorted Gata4p[5kb]-GFP Expressing Cells

To verify the hypothesis that Gata4 is expressed in migratory NCC but below detectable levels for ISH and/or in a more restricted fashion in comparison to the Gata4p[5kb]-GFP transgene, we used an RT-PCR approach on “purified” NCC. Cranial and trunk migratory NCC were obtained by FACS-based isolation of GFP-positive cells from dissected heads and trunks of Gata4p[5kb]-GFP embryos. To this end, e9.5 embryos were chosen because, in contrast to later stages, GFP transcription closely matched direct GFP fluorescence detection at this stage (Fig. 5). Via RT-PCR, we found that Gata4 is co-expressed with the NCSC marker Sox10 (Kim et al.,2003; Kleber et al.,2005) in a large subset of the sorted cranial NCC (Fig. 6B, left). We also assessed expression of the other members of the GATA4/5/6 family as well as other multipotent NCC markers in the sorted cranial and trunk migratory NCC. This experiment revealed that, under these experimental conditions, Gata6 but not Gata5 is co-expressed with Gata4 and a number of other NCSC markers including Nestin, Slug/Snai2, RhoB, p75NTR, and Sox9 (Fig. 6B, right) (Mujtaba et al.,1998; Sefton et al.,1998; Morrison et al.,1999; Cheung et al.,2005; Lee et al.,2007). However, it should be noted that very low levels of Gata5 expression were also detected by nested RT-PCR (data not shown). Taken together these results demonstrate that significant levels of Gata4 and Gata6 transcripts are found within at least a subset of the migratory NCC.

Figure 6.

In vitro characterization of GFP-labelled neural crest cells from e9.5 Gata4p[5kb]-GFP embryos. A: The heads and trunks from a pool of e9.5 Gata4p[5kb]-GFP were dissected, cells were dissociated, and both cranial (C) and trunk (T) GFP+ cells were obtained by FACS. B: Left: RT-PCR analysis of the sorted cells showing co-expression of Gata4 with Sox10 in cranial GFP+ cells. Sox10 is a marker of neural crest stem cells. Right: RT-PCR analysis of both cranial and trunk sorted NCC showing co-expression of Gata4 with Gata6 as well as with other neural crest stem cell markers (Nestin, Slug/Snai2, p75NTR, RhoB). C,C': These sorted GFP+ cells can grow as proliferating spheres, which can be passaged when cultured in non-adherent conditions in a serum-free medium containing EGF and bFGF. D,D': When cultured on fibronectin/polyD-Lysine-coated dishes in the presence of TGFβ1 for 2 weeks, these GFP-positive cells can differentiate into smooth muscle cells. D: DAPI staining of the nucleus of three sorted cells. D': Detection of α-smooth muscle actin (αSMA) via immunofluorescence.

The co-expression with several multipotent NCC markers at e9.5 and the preferential expression of the Gata4p[5kb]-GFP transgene in the BC at e10.5 and e11.5 suggested that the Gata44 promoter is active within at least a subset of the NCSC. If this is true, GFP-positive cells should have the capacity to self-renew and differentiate to specific cell fates in response to instructive growth factors (Kleber et al.,2005). To verify this possibility, we used sphere-forming and differentiation cell culture assays on the Gata4p[5kb]-GFP-labelled NCC. When cultivated on uncoated dishes, in NeuroCult serum-free medium supplemented with EGF (20 ng/ml) and bFGF (40 ng/ml), these sorted fluorescent cells proliferate as spheres (Fig. 6C, C') reminiscent of neurospheres, cardiospheres, and skin-derived precursor cells (Reynolds and Weiss,1992; Toma et al.,2001; Tomita et al.,2005). In addition, when cultivated on fibronectin/polyD-Lysine in the presence of FBS (5%) and TGFβ1 (10 ng/ml), these cells can be differentiated into smooth muscle cells (Fig. 6D,D') (Shah et al.,1996). Therefore, along with the BC expression, these data show that the Gata4 promoter is active within at least a subset of the embryonic NCSC.


Our data demonstrate that Gata4 is more widely expressed in the developing embryo than previously reported, and that initiation of its transcription in the embryo proper occurs earlier than previously thought. Indeed, early Gata4 expression was previously reported in the extra-embryonic visceral endoderm, but was not reported before e7.0 in embryonic tissues (Arceci et al.,1993; Heikinheimo et al.,1994; Morrisey et al.,1998; Nemer and Nemer,2003). We now provide evidence for Gata4 expression in intra-embryonic cells of the blastocyst and, furthermore, that the sequences controlling this novel expression pattern are contained within the Gata4 5-kb proximal promoter. Although we cannot exclude the possibility that at least part of Gata4 expression in ES cell lines, as detected by RT-PCR and Cre reporter assays, might be due to the presence of some cells having initiated primitive endoderm differentiation in the cultures, our Cre reporter assay in mice strongly suggests that at least weak Gata4 transcription must occur in cells of the ICM contributing to the embryo proper. Moreover, our results validate previous analyses of the transcriptome of both mouse and human undifferentiated ES cells, which suggested expression of Gata4 as well as Gata5 and Gata6 via either microarray or massively parallel signatures sequencing (MPSS) approaches (Abeyta et al.,2004; Brandenberger et al.,2004; Boyer et al.,2005; Wei et al.,2005).

Analysis of Gata4Cre/+::R26R-EYFP embryos revealed the presence of widespread but mosaic YFP fluorescence. Such mosaic expression could be explained by a delayed activation of the combinatorial Cre reporter system in some cells of the ICM before they divide. Indeed, due to weak expression from the Gata4 locus, a threshold level of Cre recombinase might not be reached in time for these cells. In support of this hypothesis, varying levels of GFP fluorescence in cells of the ICM are evident in Gata4p[5kb]-GFP blastocysts. Interestingly, varying levels of Gata6 expression within cells of the ICM have also been reported and this mosaic expression pattern was found to be significant (Koutsourakis et al.,1999; Chazaud et al.,2006). Indeed, cells of the ICM weakly expressing Gata6 are also weakly expressing Nanog whereas strong Gata6 or Nanog expression is found in a mutually exclusive manner (Chazaud et al.,2006). It has been proposed that this mutually exclusive strong expression of Gata6 and Nanog in ICM cells is required for primitive endoderm and epiblast fate decisions respectively (Chazaud et al.,2006). As in vitro loss-of-function and overexpression studies have also specifically implicated Gata4 in primitive endoderm specification, our in vivo data strongly support the previously proposed model of cross-regulatory interactions between Gata4 and Gata6 at the base of the transcriptional cascade controlling differentiation of the visceral endoderm (Soudais et al.,1995; Morrisey et al.,1998; Fujikura et al.,2002). According to this model, during ES cell self-renewal, overall GATA expression is repressed at a point below a threshold level necessary to initiate differentiation. Once repression is relieved, GATA6 protein levels increase initially and transactivate Gata4 expression. Then, once the overall GATA protein level is increased over the threshold, extra-embryonic endoderm differentiation is definitely initiated. As this process is expected to occur between e3.5 and e4.5 (Chazaud et al.,2006), this may explain why we were able to detect Gata4 expression only in e4.5 and not in e3.5 blastocysts, via a (not sensitive) in situ approach. Taken together, these observations suggest that GATA4, like GATA6, is specifically involved in primitive endoderm differentiation and, as such, should be kept at low levels in undifferentiated cells of the ICM. However, the analysis of Gata4-null embryos has not revealed any important in vivo function for GATA4 during implantation (Kuo et al.,1997; Molkentin et al.,1997), most likely due to functional redundancy between GATA4 and GATA6 (Holtzinger and Evans,2005; Xin et al.,2006), Thus, concomitant loss-of-function of at least GATA4 and 6 will most likely be required to definitely and fully reveal the GATA function in primitive endoderm specification in vivo. Our data additionally underscore the fact that the presence of Gata4 transcripts by themselves should not be considered a good indicator of endoderm or cardiomyocyte differentiation from ES cells (Capo-Chichi et al.,2005; Yuasa et al.,2005).

Using our transgenic mouse models, together with RT-PCR and cell culture assays, we have also found Gata4 expression within migratory NCC. Our results further suggest that Gata4 is expressed within at least a subset of the embryonic NCSC. In support of this, it is noteworthy that primitive neural stem cells have also been reported to express Gata4 (Hitoshi et al.,2004). It is also interesting to note that neural crest–derived Gata4-expressing dormant stem cells have been described in the post-natal heart (Tomita et al.,2005). These dormant stem cells were found to represent only a small subset of the neural crest–derived heart cells. Cardiac NCC originates from the caudal hindbrain and start to colonize the cardiovascular system through the pharyngeal arches around e9.5 (Stoller and Epstein,2005). As these cardiac NCC are included in our cranial GFP-positive sample, expression of Gata4 in only a limited number of neural crest–derived heart cells might, therefore, explain our inability to definitely identify migratory NCC expressing Gata4 using standard in situ approaches. However, we cannot exclude the possibility that Gata4 transcript levels in embryonic NCC might be beneath the level of ISH sensitivity.

In a variety of cell types, GATA4 is known for its involvement in cell fate decisions and it also appears to be involved in the process of cell migration (Molkentin,2000; Viger et al.,2008). In NCC, GATA4 may be involved in either one or both of these processes. Unfortunately, the early lethality of Gata4-null embryos around e8.5–9.5 precludes the analysis of NCC development in this model (Kuo et al.,1997; Molkentin et al.,1997). In addition, given the co-expression of Gata4 and Gata6 in migratory NCC, conditional loss-of-function models for both genes will be most likely required to reveal their role in these cells. In the meantime, the Gata4p[5kb]-GFP and -RFP mouse models described herein represent interesting new models for the study of NCC development.


Gene Targeting and Generation of Gata4Cre/+ Offspring

The targeting vector allowing replacement of Gata4 exon 2 by a Cre reporter gene cassette (Fig. 1A) was generated with a 2.8-kb fragment of Gata4 intron 1 as the 5′ homologous arm and 3.2 kb of Gata4 intron 2 as the 3′ homologous arm. These genomic sequences were obtained by PCR amplification of FVB/n mouse genomic DNA, using the Advantage II DNA polymerase mix (Clontech), and the respective oligonucleotides: 5′ arm forward, 5′- SacII-CCGCGGGAGCCTAAAGGAAAGAC-3′; 5′ arm reverse, 5′-XmaI-AACGAGGAGAGGCCCAACTCGC-3′; 3′ arm forward, 5′-NdeI-GGTGAGTTACCTTAGGGCCTTGAG-3′; 3′ arm reverse, 5′-BclI-GCTTCTAACAATTGCTTCAGGGTCC-3′. These PCR products were cloned into the pGEM-T plasmid (Promega) and integrity of the sequences was confirmed by sequencing. These Gata4 genomic sequences were then subcloned into a generic bicistronic DsRed2-IRES-Cre knockin vector also containing a SV40 promoter/neomycin resistance cassette. This generic knockin vector was generated from pIRES2-EGFP and pDsRed2 plasmids (Clontech). Details regarding plasmid construction are available upon request.

R1 ES cells were cultured on feeder cells under standard conditions as previously described (Pilon et al.,2007). Cells were electroporated with 40 μg of linearized targeting vector and selected with G418 (200 μg/ml) for 7 days. Surviving clones were isolated and homologous recombination assessed by genomic southern blot using BamHI digestion and hybridization with the probe depicted in Figure 1. Positive clones were confirmed for the fidelity of recombination by southern blot analysis using additional restriction enzymes (Fig. 1). Targeted ES cells were injected into C57BL/6 blastocysts according to standard procedures (Nagy et al.,2003) to successfully generate germline chimeras. These chimeras were mated with FVB/n females and heterozygous animals were identified by Southern blot analysis as described above. Subsequent genotyping of established Gata4Cre/+ lines was performed by PCR using the oligonucleotides depicted in Figure 1A: primer 1, 5′-GCCTAGAGGTTTCTGCTTTGATGC-3′; primer 2, 5′-GGCCATGGCCAGGCTTTGGTAC-3′; primer 3, 5′-ATCTCGAACTCGTGGCCGTTCAC-3′. The wild-type allele was identified using primers 1 and 2 (237 bp) whereas the targeted allele was identified using primers 1 and 3 (163 bp).

Cell Lineage Studies, Gata4 Transgenic Reporter Animals, and In Situ Hybridization

For cell lineage studies, Gata4Cre/+ mice were bred with the R26R-EYFP Cre-reporter line which was kindly provided by F. Costantini (Srinivas et al.,2001). The Wnt1-Cre mouse model (Danielian et al.,1998) was purchased from Jackson Labolatories and the generation of Gata4 transgenic reporter mice has been described previously (Mazaud Guittot et al.,2007). All animal manipulations were performed in accordance to national standards and institutional protocols. For gene expression analyses, transgenic or wild-type FVB/n mice were mated overnight and noon of the day of vaginal plug was considered as e0.5. Whole-mount in situ hybridizations were performed as previously described (Pilon et al.,2006). The Gata4 probe was synthesized from a RT-PCR product of 458 bp corresponding to the 3′ ORF end using a DIG RNA labelling kit (Roche Diagnostics Canada, Laval, Canada) according to the manufacturer's instructions. The GFP probe was generated from the EGFP cDNA (Clontech) subcloned into pBluescript (Stratagene). Pictures were taken with a Retiga 1300 (Q imaging) or Nikon DXM1200 digital camera mounted on a Nikon Eclipse E800 fluorescence microscope or a Leica MZFLIII stereomicroscope. For confocal microscopy, the Nikon Eclipse E800 microscope was used with epifluorescence and a Nikon C1-confocal configuration.

RT-PCR and Cre Activity Assays in ES Cells

For RT-PCR and Cre activity assays, ESpNP26.23 and parental R1 ES cells were cultivated in standard ES cell medium onto gelatin-coated dishes but without feeder cells. Approximately 1 × 106 cells were used for mRNA extraction. Messenger RNA isolation and reverse transcription were performed as described previously (Cory et al.,2007). With the exception of Cre, primers were designed to encompass an intron, allowing the detection of contaminating amplification of genomic DNA by the presence of a larger band. Specific pairs of oligonucleotides and expected size of the amplification products are listed in Table 1. PCR amplifications, using Advantage 2 polymerase mix (Clontech), consisted of 40 cycles of 35 sec at 96°C, 30 sec at 61°C, and 45 sec at 68°C. Amplified bands were size fractioned on a 1.5% agarose gel.

Table 1. Oligonucleotide Pairs Used for RT-PCR Analyses
GeneForward primerReverse primerSize (bp)

For transfection assays, approximately 10,000 cells were seeded onto a 24-well plate. Cells were transfected with 0.2 μg of the Cre reporter plasmid R26R-GFP using Effectene reagent (Qiagen) according to the manufacturer's instructions. The R26R-GFP construct was generated by introducing the ROSA26 800bp proximal promoter juxtaposed to a sequence encoding transcriptional and translational stop (STOP) flanked by loxP sites into the pEGFP-1 plasmid. ROSA26 promoter sequences were kindly provided by P. Soriano (Zambrowicz et al.,1997) whereas the interfering Lox-STOP-Lox cassette was generated as previously described (Lakso et al.,1992). Excision of the STOP cassette by Cre-mediated recombination leads to expression of GFP.

In Vitro Analyses of the Migratory Gata4p[5kb]GFP-Positive Cells

Heads and trunks were dissected out from a pool of stage-matched e9.5 Gata4p[5kb]-GFP embryos and cells were dissociated in NeuroCult NSC basal medium (StemCell Technologies) supplemented with collagenase (50 U/ml) and dispase (2.4 U/ml) for 45 min at room temperature. For RT-PCR gene expression profiling (see above), 10,000 fluorescent and non-fluorescent cells were purified using FACS isolation on a MoFlo cell sorter (Cory et al.,2007). For cell proliferation culture assays, FACS-purified cells (20,000 cells/cm2) were cultured in serum-free NeuroCult NSC proliferation medium (StemCell technologies) supplemented with EGF (20 ng/ml) and bFGF (40 ng/ml) (Sigma) on uncoated cell culture dishes. For cell differentiation into smooth muscle cells, GFP-positive cells were cultured on fibronectin/polyD-lysine coated 8-well chamber slides (1,000 cells/well) for 2 weeks in NeuroCult basal medium supplemented with 5% FBS and TGFβ1 (10ng/ml) (Sigma). Smooth muscle cells were identified via immunofluorescence using a mouse monoclonal anti- αSMA antibody (Sigma) at a 1:400 dilution and a bovine anti-mouse FITC-coupled secondary antibody (Santa-Cruz).


The authors thank Veronique Paradis (IRIC, Montreal, Canada) for the gift of the female ES cell line VP10 and Dr. Frank Costantini for the gift of the R26R-EYFP mouse line. Safron Dornan is acknowledged for the generation of the R26R-GFP plasmid and Céline Forget is thanked for her technical assistance. Eric Massicotte and Martine Dupuis (IRCM, Montreal, Canada) are thanked for the FACS analysis. This work was supported by CIHR grants to D.W.S. and R.V. and a grant from Univalor to D.W.S.