Author contributions: M.R., M.P.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; A.M.: conception and design, collection and assembly of data, data analysis and interpretation, and final approval of manuscript; L.H., S.S., Z.W., H.Z., L.D.R., E.C., S.F.D, M.L.M., H.v.B., C.M., and G.M.: collection and assembly of data, final approval of manuscript; D.A.: conception and design, financial support, collection and assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript. Equal contribution to this article for M.R. and A.M., and for M.P. and D.A.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS September 2, 2011.
p63, a member of the p53 family, is essential for skin morphogenesis and epithelial stem cell maintenance. Here, we report an unexpected role of TAp63 in cardiogenesis. p63 null mice exhibit severe defects in embryonic cardiac development, including dilation of both ventricles, a defect in trabeculation and abnormal septation. This was accompanied by myofibrillar disarray, mitochondrial disorganization, and reduction in spontaneous calcium spikes. By the use of embryonic stem cells (ESCs), we show that TAp63 deficiency prevents expression of pivotal cardiac genes and production of cardiomyocytes. TAp63 is expressed by endodermal cells. Coculture of p63-knockdown ESCs with wild-type ESCs, supplementation with Activin A, or overexpression of GATA-6 rescue cardiogenesis. Therefore, TAp63 acts in a non-cell-autonomous manner by modulating expression of endodermal factors. Our findings uncover a critical role for p63 in cardiogenesis that could be related to human heart disease. STEM CELLS 2011;29:1672–1683
The transcription factor p63, a homolog of the tumor suppressor gene p53  is expressed in embryonic ectoderm and in adult, somatic basal cells of stratified epithelia, in myoepithelial cells of the breast, salivary glands, and in the proliferative compartment of gastric mucosa [2, 3]. It is characterized by six different isoforms under the control of two distinct promoters and three different alternative splicings leading to the production of DN- or TA-p63, and α, β, or γ isoforms, respectively. p63 is a major epithelial transcription factor that acts as a key regulator of pleiotropic cellular functions, including cell adhesion, cell proliferation, apoptosis, and senescence [4–6]. All known functions for p63 so far are related to development and to physiopathology of stratified epithelium [5, 7–11]. In loss-of-function models, mice lacking p63 have been shown to be severely compromised in their ability to generate an epidermis and many other stratified epithelia as well as the limbs [2, 3]. During mouse embryonic epithelial development, ΔNp63 is the main transcript detected in the developing surface ectoderm as well as in the apical ectodermal ridge of developing limb buds while TAp63 expression appears later at stratification . It has been demonstrated that ΔNp63 is needed for epidermal commitment [7, 9], self-renewal [10, 11], cell survival, and cell adhesion  while TAp63 is required for stratification/differentiation and apoptosis . As ΔNp63 is specifically expressed at sites of epithelial–mesenchymal interaction, it has been suggested that ΔNp63 promotes an epithelial phenotype and suppresses mesenchymal specification in squamous cells [13, 14].
Even though p63 is a master regulator of stratified epithelial homeostasis, a novel putative function of p63 in heart development is emerging from data collected by our recent studies. First, Chip-seq analysis experiments made on epidermal cells identified more than 60 genes involved in cardiac development that are in close proximity to p63-responsive element featuring a p63 binding site and known to be involved in heart development . Second, a specific activity of a p63 enhancer was reported in the embryonic heart area of transgenic mice . Finally, we found few genes related to cardiogenesis in a list of p63 target genes from a large scale chip-on-chip analysis made by the Mantovani's group . Although p63 KO mice have been widely studied for more than 10 years, a cardiac phenotype has not previously been recognized. We therefore evaluated whether p63 could be involved in heart development in vivo and in murine cardiogenesis in vitro. Our data indicate that p63 plays a crucial role in cardiac cell determination through a cell non-autonomous manner and in turn, in heart development. Our findings thus suggest that p63 features both a broader pattern of expression and repertoire of functions than so far reported.
Mice and Cell Lines
Heterozygous p63 knockout mice are Brdm2 mice obtained from the Jackson Laboratory (Bar Harbor, ME, www.jax.org). Numbers of mice and embryos used in this study are described in the figure legends. The mouse CGR8 ESC line used in this study, the ES sh-p63 clones and the culture condition used have been described previously .
ES sh-p63–green fluorescent protein (GFP) clones were generated by transfection of ES sh-p63 cells with pEGFP plasmid (Clontech, St. Germain en Laye, France, www.clontech.com) with Lipofectamine 2000 (Invitrogen, Life Technology SAS, Villebon sur Yvette, France, www.invitrogen.com). Stable fluorescent cells were sorted using FACS Aria (Becton Dickinson, Le Pont de Claix, France, www.bd.com/fr) and further cultivated in the presence of neomycin (G418). For Activin A treatment, ESCs were treated from day 1 before differentiation and during the entire differentiation process with 5 ng/ml Activin A (R&D System, Abingdon, UK, www.rndsystems.com). For the chimeric embryoid body (EB) differentiation, sh-ctrl and GFP-sh-p63 ESCs were cocultured at a 1:1 ratio. The endodermal END2 cell line  was maintained in culture in Dulbecco's modified Eagle's medium supplemented with 7.5% serum. GATA-6 pcDNA3 plasmid was given by M. Nemer (Ottawa, Canada).
Antibodies, Immunofluorescence, FACS, and Immunoblotting Analysis
Immunoblotting, FACS, and immunofluorescence analysis have been already described . The primary antibodies used for immunofluorescence were: mouse anti-p63 (4A4, Sc-8431, Santa Cruz Biotechnology, Tebu International [Le Perray en Yvelines, France, www.scbt.com] at 1/100 dilution), mouse anti-sarcomeric α-actinin (EA-53, A7732, Sigma Aldrich [Lyon, France, www.sigmaaldrich.com] at 1/100 dilution), mouse anti-Islet1 (clone 39.4D5, Developmental studies hybridoma bank [dshb.biology.uiowa.edu] at 1/100 dilution), mouse anti-Mlc2v (clone F109.3E1, Biocytex Alexis Corporation [Marseille, France, www.biocytex.fr] at 1/100 dilution), mouse anti-TroponinT2 (clone Ab-1, MS-295-PO, LabVision [Francheville, France, www.labvision.com] at 1/100 dilution), mouse anti-Nkx2.5 (MAB2444 clone 259416, R&D Systems at 1/100 dilution), rabbit anti-Nkx2.5 (SC14033, Santa Cruz Biotechnology at 1/100 dilution), and rabbit anti-Sox17 (09-038, Millipore SAS [Molsheim, France, www.millipore.com] at 1/100 dilution). Secondary Alexa-488 and Alexa-594 donkey anti-mouse and anti-rabbit antibodies were purchased from Molecular Probes (Invitrogen at 1/1000 dilution). For FACS analysis, ESCs were stained with mouse primary anti-troponinT (clone Ab-1), mouse anti-β-tubulin (clone Tub 2-1, T4026, Sigma Aldrich). Mouse isotype control mAb (PharMingen, Invitrogen) was used to set the background level of fluorescence. FITC-coupled rabbit anti-mouse (Dako France SAS, Trappes, France, www.dako.com) or tri-color-coupled goat anti-mouse (M35006, Caltag Laboratories, Invitrogen) secondary antibodies were added. Samples were analyzed with CellQuest or DIVA BD's softwares on a FACScan, FACScalibur, and FACScanto cytometers (Becton Dickinson). For immunoblotting, mouse anti-p63 (4A4) and anti-β-tubulin (clone Tub 2-1, T4026, Sigma Aldrich) monoclonal primary antibodies were used. Mice embryos used for immunostaining were collected at E11.5 or E12.5 and frozen in “Shandon cryomatrix” medium (Shandon, ref: 6769006). Cryosections of 6 mm thickness were cut using a “Shandon” cryotome (Thermo Shandon, Cergy Pontoise, France, www.thermo.fr, Model A5620SME). Sections (8 μm) of E12.5 and E14.5 embryos embeded in paraffin, stained with hematoxylin/eosin, and observed with a Eclipse Ti Nikon microscope (with the NIS Elements AR Imaging [Nikon France Instruments, Champigny sur Marne, www.nikoninstruments.eu] software) were used to determine the thickness of the ventricular wall.
Extraction of RNAs and qRT-PCR analysis was performed as previously described . Each gene was amplified using the appropriate specific primers—36B4: 5′-TCCAGGCTTTGGGCATCA-3′, 5′-CTTTATCAGCTGCACATCACTCAGA-3′; ΔNp63: 5′-TGTACCTGGAAAACAATGCCCA-3′, 5′-GACGAGGAGCCGTTCTGAATCT-3′; TAp63: 5′-TGGATGAACCTTCCGAAAATG-3′, 5′-TGCGGATACAATCAATGCTAAT-3′; Brachyury: 5′-GACTTCGTGACGGCTGACAA-3′, 5′-CGAGTCTGGGTGGATGTAG-3′; Islet1: 5′-GCAGCAACCCAACGACAAAACTA-3′, 5′-TATCTGGGAGCTGCGAGGACAT-3′; Nkx2.5: 5′-TTACCCGGGAGCCTACGGTG-3′, 5′-GCTTTCCGTCGCCGCCGTGCGC-3′; Tbx5: 5′-CCAGCTCGGCGAAGGGATGTTT-3′, 5′-CCGACGCCGTGTACCGAGTGAT-3′; TnT2: 5′-CGGAAGAGTGGGAAGAGACAAAGAACG-3′, 5′-TCGCCAGAATCCAGAACAATGC-3′; Mlc2v: 5′-GCCAAGAAGCGGATAGAAGG-3′, 5′-CTGTGGTTCAGGGCTCAGTC-3′; Activin a: 5′-CGGAAGAGTACCTGGCACAT-3′, 5′-CTTCTTCCCATCTCCATCCA-3′; and Sox17: 5′-CGAGCCAAAGCGGAGTCTC-3′, 5′-TGCCAAGGTCAACGCCTTC-3′; GATA6: 5′-CCGCGAGTGCGTGAACT-3′, 5′-CGCTTCTGTGGCTTGATGAG-3′.
Imaging of Cytosolic Calcium Concentration ([Ca2+]C) in Embryonic Hearts
Mouse embryos (E12.5) were loaded with 10 μM fluo-4/AM (Molecular Probes, Eugene, OR) for 15–30 minutes, washed and incubated for 30 minutes, and transferred to the stage of a confocal microscope (LSM510 meta, Carl Zeiss France S.A.S., France, www.micro-shop.zeiss.com) in a culture dish containing ADS buffer added with 2 mM Ca2+ . Cells were illuminated at 488 nm using the Argon laser, and fluorescence were recorded by the photomultipliers after crossing a Haupt. Frab. Teiler 488 and a secondary beamsplitter (NFT) 545 dichroic mirrors and two bandpass filters at 505–530 nm. To calculate the frequency and synchronization of Ca2+ spikes, regions of interest were selected in cells within a beating area and the average pixel intensity was plotted as a function of time. All experiments were performed at 35 ± 2°C and were repeated at least three times.
Embryos were fixed with 2% glutaraldehyde in sodium cacodylate buffer. Fixed tissues were embedded in epoxy resin, processed for thin sectioning, and examined with a Philips CM12 transmission electron microscope equipped with an Olympus France SAS (Rungis, France, www.olympus.fr) SIS CCD camera.
Cardiac Defects in p63-Deficient Mice
We first analyzed the cardiac phenotype of p63-null embryos. During embryonic development, no visible macroscopic defect was observed in p63−/− hearts. However, histological sections at E14.5 revealed dilated cardiomyopathy with deficient trabeculation and a thin ventricular wall (Fig. 1A). Systematic morphometric heart measurement performed on four independent mutant and wild-type (WT) embryos, each derived from four different litters confirmed a decrease in the thickness of ventricular walls (Fig. 1B). This was associated with reduced trabeculation. No cardiac defects were detected in the hearts of p63+/− heterozygous mice. Hearts from p63+/+, p63+/−, and p63−/− newborn mice were analyzed at the ultrastructural level. Normal maturation of cardiomyocytes is accompanied by sarcomere assembly and biogenesis of multiple mitochondria, which are juxtaposed to the myofibrillar apparatus in WT and heterozygote mice (Fig. 2A, upper panels). In contrast, cardiomyocytes from p63−/− mice exhibited pronounced myofibrillar fragmentation and disorganization of the myofibrillar apparatus with a lack in z line. Mitochondria were randomly distributed, large, and swollen (Fig. 2A, lower panel). These defects were equally apparent in right and left ventricular chambers (not shown). Accordingly, we observed significant lethality in utero of homozygous mutant embryos, as their frequency at E18.5 appeared significant lower (18.5%) than the expected Mendelian ratio of 25% over 24 pregnancies (Supporting Information Table 1). To directly evaluate the consequences of the absence of p63 on cardiac function, WT and p63−/− E12.5 embryos were loaded with Fluo-4 to monitor spontaneous Ca2+ spiking in the myocardium, as an index of excitation–contraction coupling. Rhythmic and large Ca2+ oscillations were monitored in WT heart embryos. In contrast, weak Ca2+ spikes were observed in p63-deficient embryos (Fig. 2B).
This severe cardiac phenotype, including the defect in trabeculation of the p63-deficient mice, suggests that the production/migration of early progenitors might have been limited or submitted to enhanced apoptosis. The LIM homeodomain protein Islet-1 (Isl1) is transiently expressed in a population of multipotential mesodermal precursors emerging from a second heart lineage located medially and posteriorly to the cardiac crescent at E8.0. This cell population contributes to two-thirds of the hearts including the right ventricle, part of the atria, and the outflow tract (OFT) [20–22]. We analyzed isl1 expression by immunofluorescence staining at E11.5 in both WT and p63−/− embryos (Fig. 2C). In WT embryos, isl1-positive cells were observed in the primary left ventricle (PLV), wall of the atrium, the sinus venosus, the epicardium and the OFT. In p63−/− embryos, the isl1-positive cells could be found in the epicardium, in the atria the PLV in a much reduced number (Fig. 2C). These results indicate that p63-deficient mice have a decreased pool of isl1-positive progenitor cells. At earlier stage of development, before the limb deficiency phenotype can be observed, genotyping of these p63 KO embryos are problematic. This is due to the particular strategy used to generate the brdm2 p63 null mice, with targeting vectors that used a gap-repair-dependent mechanism. It led to the duplication of a large region of the p63 targeting gene that prevents to genotype by PCR amplification while Southern blot cannot be done on such small embryos that has to be further analyzed by immunostaining . Thus, we have extracted RNA samples from E8.5 embryo litters from four independent heterozygous mice mating and analyzed the relative expression of p63 (for genotyping of null embryos) by qRT-PCR along with islet-1, nkx2.5, and tbx5 genes. A representative litter is illustrated in Figure 2D and shows that genotyping of p63 null embryos can be achieved by the analysis of p63 gene expression. Embryo # six did not express p63 at all and interestingly displayed a lower expression of cardiac-related genes. These data confirm the severe cardiac phenotype in absence of p63 and demonstrate that p63 is required at early stage for proper heart development from both the first and second heart lineages.
P63 Is Necessary During Early ESC Cardiogenesis
To get more insight into the origin of cardiac defects in mouse embryos, we used ESCs. ESCs spontaneously produce beating cardiomyocytes a few days following the formation of EBs and thus provide a model that recapitulates successive steps of early cardiogenesis in vitro and allow for mechanistic studies. Compared to WT ESCs, ESCs stably deficient for p63 (sh-p63-ES)  were not able to produce beating cardiomyocytes (Table 1). No difference was observed between expression of Brachyury gene in WT and sh-p63 ES cells on differentiation (Fig. 3A), indicating that there is no defect in mesodermal induction of p63-deficient ESCs. However, we found that the expression of markers related to cardiac cell fate, including the transcription factors islet1, nkx2.5, and tbx5 and the sarcomeric marker, troponin-T, was largely reduced (Fig. 3A). Mature cardiomyocytes expressing ventricular myosin light chain two (MLC2v) and α-actinin, both organized within sarcomeric striated and aligned structures were observed in WT EBs (Fig. 3C, 3D). In contrast, sh-p63 ESC-derived EBs featured few cardiomyocytes expressing MLC2v or α-actinin as assessed by real-time qPCR at the mRNA level (Fig. 3B) and by immunofluorescence at the protein level (Fig. 3C, 3D). These results suggest that p63 deficiency impairs early cardiogenesis or maturation of cardiomyocytes.
Table 1. Loss of cardiomyocyte differentiation of ESCs at day 9 (as measured by the beating area/EB ratio and the % of TroponinT-positive cells) on total sh-p63 and ΔNp63-, TAp63-, or p63-specific knockdown
Next, we monitored expression of both TAp63 and ΔNp63 isoforms within EBs. qRT-PCR analysis showed that, TAp63 and ΔNp63 genes were differentially expressed in the course of ESC differentiation (Supporting Information Fig. 1A). TAp63 transcript was detected by day 1–3 of differentiation and its expression was maintained throughout ESC differentiation. The RNA encoding the ΔNp63 isoform, which controls epidermal commitment , were detected later and reached their highest level of expression at day 15 (Supporting Information Fig. 1A). To investigate whether early activation of TAp63 gene expression was required for the early stages of cardiogenesis, we designed a knockdown strategy. WT ESCs were transiently transfected with different sets of small-interfering RNAs (siRNAs) targeting TAp63, ΔNp63, all isoforms of p63 or control (ctrl) scramble sequences. The specificity and efficiency of the siRNAs was confirmed in epithelial cells, as illustrated in Supporting Information Figure 1B. Cardiogenesis was monitored in WT and and p63-downregulated EBs. Whereas no effect was observed with control siRNA, siRNA-mediated inactivation of TAp63 or p63 gene expression highly reduced the number of beating areas (Table 1), expression of cardiac-related genes, such as islet1, nkx2.5, tbx5, MLC2v, α-actinin, and troponin-T (Supporting Information Fig. 1C) as well as the percentage of troponin-T-positive cells at day 9 within EBs (Table 1). siRNA specifically raised against ΔNp63 did not reduce cardiogenesis but even slightly enhanced it. This feature might reflect the dominant-negative effect of the truncated ΔNp63 isoform toward TAp63 . These results demonstrate the direct involvement of TAp63 during early ESC-derived cardiogenesis.
p63 Is Endodermal and Acts in a Non-Cell Autonomous Manner on Cardiogenesis
Immunofluorescence staining revealed no apparent colocalization of cardiac markers with p63 in WT ESC-derived cardiomyocytes as exemplified in Supporting Information Figure 2A (left) using nkx2.5 and p63 antibodies. Accordingly, p63 was not expressed in the developing heart, neither at the mRNA (not shown) nor at the protein levels (Supporting Information Fig. 2A, right). In addition to epidermal cells, endodermal-derived epithelial cells, such as in the thymus or esophagus, have been shown to express p63 during development and later at the adult stage [10, 23]. Accordingly, p63 transcripts were detected in the endoderm of E8.5 embryos by in situ hybridization (ISH) (Supporting Information Fig. 2B). Early in development, as soon as the setting of the anterior visceral endoderm expressing sox17, the primitive mesoderm and endoderm are in close proximity, allowing extensive cross-talk between these lineages . At this stage, the vertebrate endoderm plays an inductive role in adjacent tissues, including parts of the heart. Expression of the transcription factor sox-17 in the mesendoderm is essential for the specification of cardiac mesoderm in ESCs . We analyzed the expression of p63 and sox-17 by immunostaining during WT and shp63-ESC differentiation. Interestingly, most of the sox-17-positive ESCs were positive for p63 at early stages of ES cardiogenesis (Fig. 4A). As a matter of fact, we detected expression of p63 in the endodermal END2 cell line (Supporting Information Fig. 3A). RT-PCR and Western blot analysis confirmed that the main isoform expressed in endodermal cells is TAp63 γ (Supporting Information Fig. 3B, 3C). Interestingly, this cell line, positive for sox-17 (Supporting Information Fig. 3A), has been used as a cardiogenic inducer of pluripotent cells . These data suggest the endodermal origin of TAp63 effect on cardiogenesis. In the absence of p63, the reduction in the number of sox-17-positive cells is apparently due to both a specific repression of sox-17 gene expression by TAp63, as illustrated by qRT-PCR analysis and a decrease in endodermal commitment, as pdx1 and foxa2 gene expression are reduced in cells deficient for p63 (data not shown).
The production of p63 in endodermal Sox-17-positive cells suggests that its effect on the cardiac fate could occur in a cell-non-autonomous manner. Therefore, we generated chimeric EBs in which p63-deficient ESCs expressing GFP reporter gene (GFP+-sh-p63) were mixed with sh-control ESCs. EBs were then allowed to undergo spontaneous cardiac differentiation. The percentage of TroponinT-positive cells at day 12 significantly increased in GFP+-p63-deficient cells that had been mixed with sh-control ESCs (Fig. 4A). Immunofluorescence staining confirmed the presence of GFP+-sh-p63 ESCs within the large area of TroponinT-positive cells, as well as of α-actinin-positive and MLC2v-positive cells (Fig. 4B, 4C and Supporting Information video). In sh-p63 ESCs, sox17 expression was significantly reduced at day 3 compared to WT but was still expressed at day 9 (Supporting Information Fig. 3D). This was confirmed by immunostaining of EBs generated from sh-p63 ESCs with Sox17 antibody (Fig. 4D), suggesting that the absence of p63 expression in sox-17+-endodermal cells could affect early cardiac cell specification but also specification and/or proliferation of the endodermal cells. Therefore, these results demonstrate that p63 expression promotes cardiac specification mainly through a cell-non-autonomous mechanism. However, a tight proximity between the cells or local high concentration of endodermal secreted factors controlled by p63 might be necessary as neither the conditioned medium of WT ESCs nor coculture of WT and GFP+-shp63 EBs were able to rescue cardiogenesis (data not shown).
P63 Controls Endodermal GATA and Activin A Factors
The GATA transcription factors are essential in endoderm formation and more specifically the endodermal and mesodermal GATA4 and GATA6 factors are required for cardiogenesis . GATA6 and GATA4 are both expressed in primitive endoderm cells . Mice harboring a null mutation in GATA6 lack visceral endoderm, fail to express endodermal genes, and exhibited a specific defect in endoderm differentiation . Accordingly, GATA6 gene expression was reduced in shp63 ESCs at early time point (Fig. 5A). To rescue expression of GATA6, sh-p63 ESCs were transiently transfected with a GATA6 expression construct 24 hours prior to differentiation. ESC-derived EBs were monitored for cardiac fate rescue. According to the endodermal origin of p63 function, cardiac gene expression (Fig. 5B), myofibrillogenesis (Fig. 5C), and beating activity (Fig. 5D, 5E) were efficiently rescued.
Inhibin bA gene encodes for Activin A, an endoderm-specific stimulator of early cardiogenesis, promoting proliferation, survival, and differentiation of cardiac mesodermal cells [29, 30]. Undifferentiated ESCs expressed high levels of activin-A mRNA that reduced on differentiation (Fig. 6A). On the contrary, undifferentiated sh-p63 ESCs expressed low amounts of activin-A and the expression of activin-A was not changed further during the differentiation process. Accordingly, our ChIP-seq analysis  identified in vivo physical binding of p63 onto activin-A regulatory genomic sequences. Addition of exogeneous Activin A to the sh-p63 ESCs, 24 hours before induction of differentiation and during the whole process, efficiently rescued the expression of cardiac-related genes (islet1, nkx2.5, tbx5, MLC2v, and α-actinin; Fig. 6B), cell differentiation into TroponinT (Fig. 6C) and α-actinin-positive cells (Fig. 6D), and the appearance of beating areas (Fig. 6E). Activin A treatment did not significantly enhance cardiogenesis of sh-control ESCs (not shown). The rescue by Activin A of sh-p63 ESCs was not due to reactivation of endogenous p63 gene expression (Supporting Information Fig. 4B). All these data support a regulatory role of p63 in early cardiac specification through an endodermal signaling pathway.
Cardiac Defects in p63-Deficient Mice
It has been generally accepted that mice lacking p63 die soon after birth from dehydration, due to the lack of a skin barrier [2, 3]. Herein, we uncovered a novel cardiogenic function of p63 which has been neglected and that might contribute to lethality of p63-deficient mice. Cardiac defects in p63-deficient mice include both left and right ventricular dilation, thin ventricular walls, a poor trabeculation and at the cell level an impaired myofibrillogenesis. This severe phenotype is accompanied in myocytes by the presence of large, swollen, and disorganized distribution of mitochondria. Altogether these parameters indicate that p63-deficient mice suffer heart failure. Our findings suggest that a serious impairment in heart function at birth contributes to the early death of p63−/− neonates in addition to the apparent defects in skin. The cardiac morphogenetic defects in p63−/− embryos are not as dramatic as in vitro in the shp63-ESC model. This is not surprising as it has been often observed that a cardiac phenotype in the absence or in the presence of a mutation of a gene linked to cardiogenesis is more pronounced in ES-deficient cells than in KO mice .
We next asked the origin of these cardiac defects. The allocation of cardiac progenitors and specification of cardiomyocytes takes place in the epiblast as early as 6.5 days post coitum (dpc). The cardiac fate is acquired during the ingression of epiblast cells through the primitive streak. During this process, the cardiac mesoderm and the emerging definitive endoderm progenitors remain in tight proximity between the most anterior and posterior regions of the streak, within the mesendoderm . Cell determination occurs 1 day later in the late primitive streak E7.5 dpc [32, 33] when cells move from the posterior to the anterior region under the influence of instructive factors secreted by the pharyngeal endoderm. These events lead to the heart forming region within the embryos with cardiac progenitors migrating to form the cardiac crescent. As we did not find any expression of p63 in this heart lineage or in fully developed E18.5 heart but in endodermal cells in the E8.5 embryo (Supporting Information Fig. 2) and in cultured cells (Fig. 4), we surmised that p63 exerted its cardiogenic action through the endoderm.
Endodermal Effect of TAp63 on Cardiogenesis
To get a better insight into the genetic pathway underlying the function of p63, we used the powerful model of ESCs. Indeed ESCs deficient in p63 (i.e., shp63-ESC line) did recapitulate the embryonic cardiac phenotype of p63−/− embryos. Differentiating ESCs within EBs expressed less cardiac genes and featured very few mature MLC2v, α-actinin-positive cardiomyocytes. A similar impairment of myofibrillogenesis as well as in spontaneous Ca2+ spiking as in p63 KO embryos was observed (Fig. 2B). Using this model, we found that sox17, coexpressed with p63 in WT cells and required for cardiogenesis both in vitro  and in vivo  was indeed absent in p63-deficient cells. Three other sets of experiments point to a cell-non-autonomous cardiogenic function. First, deficient cardiogenesis of shp63 cell line could be rescued by expression of the endodermal transcription factor GATA6. Second, addition of the endodermal inductive factor activin-A to differentiating EBs also restored cardiogenesis in shp63 cells. Finally, chimeric EBs generated from WT and shp63-ESCs recapitulated a normal cardiogenesis. Altogether, the findings acquired with ESCs allow us to strongly argue that p63 is a key endodermal cardiogenic factor. A lack of p63 and more specifically of TAp63 normally expressed in the visceral endoderm (i.e., ISH embryo in Supporting Information Fig. 2 or END2 cell line in Supporting Information Fig. 3) and in the pharyngeal endoderm (and its derivative like the thymus) is likely to limit the pool of cardiac progenitors of the first heart lineage at the late primitive streak stage and later of the second heart lineage emerging in the developing pharyngeal mesoderm and located anterior to the primary cardiac crescent. Such an hypothesis is supported by the limited number of isl1-positive cells in E11.5 p63−/− embryo, a factor marking a population of proliferative multipotent cardiovascular progenitors , and expressed transiently in cells of the first heart lineage . Several hypotheses might account for the lack of isl1+ progenitors, a premature stop in their proliferation, apoptosis or a limitation of their specification and determination early on. Although we cannot exclude the first two possibilities, the pattern of expression of p63 in the endodermal layers at early stages of embryogenesis together with the in vitro observed nonautonomous cellular effects of p63 rather favor the last hypothesis. Whether p63 controls the proliferation/survival and/or the specification of multipotent cardiovascular progenitors remains to be investigated.
In conclusion, we report a previously unrecognized but essential role of p63 in cardiogenesis, in the early events of cardiac progenitor production, enriching its list of functions during mammalian embryogenesis. The recent reports of one patient affected by both ectodermal dysplasia and arrhythmogenic right ventricular cardiomyopathy  and two other ED patients suffering from an atrial septal defect  suggest that patients affected by p63-linked ED syndromes could have an increased risk for unexpected heart failure. All the mutations found in ED patients are located either in the ΔN-specific or in the common domains of the p63 gene . Except for missense mutations identified in patients with apparently isolated limb defects , no other mutation in TAp63-specific sequences has been described so far. The study described here suggests that such TAp63-specific mutations would be either lethal in utero or responsible for cardiomyopathies for which no gene has yet been assigned. Moreover, duplication and deletion of the region 3q27-ter (where the p63 gene resides) are associated with congenital malformations, including dilated cardiomyopathy [37–39]. Together, these data warrant a systematic analysis of potential genotype/phenotype correlations between p63 mutations and defects in cardiogenesis.
We thank Benjamin Brinon and Pierre Gounon for help in histology and electron microscopy of embryonic hearts and Cédric Blanpain and Antoine Bondue for discussion. S.S. is an AFM postdoctoral fellow. A.M. and H.Z. are Epistem EU predoctoral and postdoctoral fellows, respectively. This work was supported by the Sixth EEC Framework Program within the EPISTEM project (LSHB-CT-2005-019067), the Agence Nationale pour la Recherche (ANR BLANC 06), INSERM, and Israel Science Foundation.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.