ABSTRACT
In situ hybridization studies, promoter analyses and antisense RNA experiments have implicated transcription factor GATA-4 in the regulation of cardiomyocyte differentiation. In this study, we utilized Gata4−/− embryonic stem (ES) cells to determine whether this transcription factor is essential for cardiomyocyte lineage commitment. First, we assessed the ability of Gata4−/− ES cells form cardiomyocytes during in vitro differentiation of embryoid bodies. Contracting cardiomyocytes were seen in both wild-type and Gata4−/− embryoid bodies, although cardiomyocytes were observed more often in wild type than in mutant embryoid bodies. Electron microscopy of cardiomyocytes in the Gata4−/− embryoid bodies revealed the presence of sarcomeres and junctional complexes, while immunofluorescence confirmed the presence of cardiac myosin. To assess the capacity of Gata4−/− ES cells to differentiate into cardiomyocytes in vivo, we prepared and analyzed chimeric mice. Gata4−/− ES cells were injected into 8-cell-stage embryos derived from ROSA26 mice, a transgenic line that expresses β-galactosidase in all cell types. Chimeric embryos were stained with X-gal to discriminate ES cell- and host-derived tissue. Gata4−/− ES cells contributed to endocardium, myocardium and epicardium. In situ hybridization showed that myocardium derived from Gata4−/− ES cells expressed several cardiac-specific transcripts, including cardiac α-myosin heavy chain, troponin C, myosin light chain-2v, Nkx-2.5/Csx, dHAND, eHAND and GATA-6. Taken together these results indicate that GATA-4 is not essential for terminal differentiation of cardiomyocytes and suggest that additional GATA-binding proteins known to be in cardiac tissue, such as GATA-5 or GATA-6, may compensate for a lack of GATA-4.
INTRODUCTION
The first morphogenetic event in mouse cardiac development is the formation of a primordial heart tube by cells that segregate from splanchnic mesoderm at 7 days post-coitum (p.c.) (Viragh and Challice, 1981; Kaufman and Navaratnum, 1981; DeRuiter et al., 1992). Two distinct epithelial layers form within this tube, a central endocardium and a peripheral myocardium, which are separated by an extracellular matrix known as cardiac jelly (Kaufman and Navaratnum, 1981; DeRuiter et al., 1992). Between 8 and 9.5 days p.c. the primitive heart tube begins to exhibit rhythmic contractions and undergoes asymmetrical elongation, creating an S-shaped loop (Colvin, 1990). Over the ensuing days of embryogenesis, the growing heart progresses through several additional morphological stages, including formation of endocardial cushion tissue, septation, trabeculation, compaction (expansion of the ventricular free wall) and envelopment by the epicardial mantle (Viragh and Challice, 1981; Colvin, 1990; Mikawa and Fischman, 1996).
Recent studies have delineated a number of transcription factors that are critical for specific stages of vertebrate cardiac morphogenesis, including the homeobox protein Nkx-2.5/Csx (Komuro and Izumo, 1993; Lints et al., 1993; Tonissen et al., 1994; Kern et al., 1995; Lyons et al., 1995), the basic helix-loop-helix proteins dHAND and eHAND (Cserjesi et al., 1995; Srivastava et al., 1995) and a member of the extended Sry family, Sox-4 (Schilham et al., 1996). Based on analogies to skeletal myogenesis, it is presumed that cardiomyocyte lineage committment also is directed by a group of tissue-specific transcription factors (Sartorelli et al., 1993; Chien et al., 1993; Buckingham, 1994; Olson et al., 1995). However, the factors essential for cardiomyocyte differentiation in vertebrates remain poorly characterized (Olson and Srivastava, 1996).
One of the transcription factors implicated in the regulation of cardiac gene expression and differentiation is GATA-4, a member of a family of zinc finger transcription factors recognizing the consensus motif (A/T)GATA(A/G) (Orkin, 1992). GATA-4 mRNA is expressed in the heart of a variety of vertebrate species including frog (Jiang and Evans, 1996), chicken (Laverriere et al., 1994), mouse (Arceci et al., 1993) and man (Huang et al., 1995). GATA-4 expression in cardiac tissue begins early in development and continues throughout the life of the organism (Arceci et al., 1993; Tamura et al., 1993; Heikinheimo et al., 1994; Ip et al., 1994; Laverriere et al., 1994; Jiang and Evans, 1996). In vitro GATA-4 binds and trans-activates the promoters or enhancers of the genes for cardiac α-myosin heavy chain (αMHC) (Molkentin et al., 1994; Huang et al., 1995), cardiac troponin C (cTnC) (Ip et al., 1994) and brain type natriuretic peptide (BNP) (Thuerauf et al., 1994; Grépin et al., 1994). The promoters of other genes expressed selectively in cardiomyocytes, such as atrial natriuretic peptide (ANP), cardiac β-myosin heavy chain (βMHC), and cardiac actin, also contain potential binding sites for GATA-4 (Molkentin et al., 1994; Ip et al., 1994). Stable expression of GATA-4 antisense RNA in P19 embryonal carcinoma cells disrupts DMSO-induced differentiation into cardiomyocytes and expression of the cardiac markers BNP, cTnC, αMHC, and βMHC (Grépin et al., 1995). Ectopic expression of GATA-4 in frog embryos transiently activates expression of cardiac actin and αMHC, but is insufficient to respecify non-cardiac lineages (Jiang and Evans, 1996).
Shortly after the discovery of GATA-4, two additional members of the vertebrate GATA-binding family, GATA-5 and GATA-6, were identified (Kelley et al., 1993; Laverriere et al., 1994; Jiang and Evans, 1996; Narita et al., 1996). GATA-5 and GATA-6 bear close structural and functional similarity to GATA-4 (Laverriere et al., 1994; Kelley et al., 1993; Jiang and Evans, 1996; Morrisey et al., 1996). Like GATA-4, transcription factors GATA-5 and GATA-6 are expressed in cardiac tissue throughout development and are capable of trans-activating genes expressed selectively in cardiomyocytes (Laverriere et al., 1994; Kelley et al., 1993; Jiang and Evans, 1996; Morrisey et al., 1996). That the distributions of GATA-4, GATA-5 and GATA-6 overlap raises the possibility of functional redundancy or interplay among these cardiac transcription factors, reminiscent of the relationship between the myogenic factors MyoD and myf-5 in differentiating skeletal muscle (Rudnicki et al., 1992; Braun et al., 1992; Edmondson and Olson, 1993; Braun and Arnold, 1994).
Recently we prepared and characterized ES cells homozygous deficient in GATA-4 (Soudais et al., 1995). When differentiated in vitro into embryoid bodies, these Gata4−/− ES cells display a defect in the differentiation of visceral yolk sac endoderm, a cell type known to express GATA-4. Many other aspects of differentiation, including skeletal myogenesis, appear unperturbed in Gata4−/− embryoid bodies (Soudais et al., 1995). In this study, we have utilized these Gata4−/− ES cells to assess whether transcription factor GATA-4 is essential for cardiomyocyte gene expression and differentiation in vitro and in vivo.
MATERIALS AND METHODS
ES cell culture and differentiation
Wild-type and Gata4−/− CCE ES cells were prepared and maintained in culture as described previously (Soudais et al., 1995). ES cells were differentiated into embryoid bodies using the suspension culture method of Doetschman et al. (1985). Alternatively, ES cells were differentiated by a two-stage hanging drop method, modified from procedures described elsewhere (Rohwedel et al., 1994; Metzger et al., 1994; Bautch et al., 1996). Briefly, nearly confluent ES cells were trypsinized, resuspended in ES cell medium consisting of Dulbecco’s modified essential medium supplemented with 15% fetal calf serum, 0.1 mM 2-mercaptoethanol, non-essential amino acids, L-glutamine and penicillin/streptomycin. ES cells were allowed to aggregate in hanging drops (800 cells/50 μl) for varying lengths of time and then plated onto gelatin-coated 24-well tissue culture dishes and covered with 0.5 ml of ES cell medium. The differentiating embryoid bodies were examined daily under phase microscopy for the presence of contracting cardiomyocytes.
Electron microscopy
Embryoid bodies or mouse embryos were washed with PBS, fixed overnight in Karnovsky’s solution, postfixed in OsO4 and embedded in Spurr epoxy resin. Ultrathin sections (70 nm) were stained with lead citrate and uranyl acetate and viewed on a JEOL 1200 EX electron microscope.
Generation and analysis of chimeric mice
ROSA26 mice (Friedrich and Soriano, 1992), bearing a ubiquitously expressed β-galactosidase transgene on a strain 129 background, were obtained from Jackson Labs. To generate the chimeric embryos, male mice homozygous for the ROSA26 transgene were bred with supraovulated C57BL/6J females. 8-cell-stage embryos were harvested at 2.5 days p.c. and injected (Hogan et al., 1994) with 1-4 Gata4−/− ES cells. The resultant embryos were then implanted into the oviducts of pseudopregnant Swiss-Webster female mice (Papaioannou and Johnson, 1993). At the desired gestational ages, the embryos were removed and subjected to further analysis. Precise staging of dissected embryos was performed using The Atlas of Mouse Development (Kaufman, 1992).
X-gal staining of whole-mount embryos and tissue sections
For embryos younger than 10.5 days p.c., whole-mount X-gal staining was employed (Papaioannou and Johnson, 1993). To visualize β-galactosidase expression in late gestation animals, frozen tissue sections were subjected to X-gal staining (Papaioannou and Johnson, 1993). These frozen sections (10 μm) were prepared by embedding animals in OCT cryopreservation solution (Tissue-Tek) after first removing the tails for GPI isoenzyme analysis (Papaioannou and Johnson, 1993). For either whole-mount embryos or frozen sections, the specimens were fixed with 0.2% glutaraldehyde for 5-15 minutes, permeabilized with 100 mM potassium phosphate pH 7.4, 0.02% NP-40 and 0.01% sodium deoxycholate for 5-15 minutes, and then incubated in 0.5 mg/ml X-gal (Promega) with 10 mM K3[Fe(CN)6], 10 mM K4[Fe(CN)6], 100 mM potassium phosphate pH 7.4, 0.02% NP-40 and 0.01% sodium deoxycholate at 37°C overnight. In some instances, stained whole-mount embryos were postfixed in 4% paraformaldehyde, embedded in resin, sectioned and counterstained with eosin to better visualize X-gal staining of internal structures.
In situ hybridization
Frozen sections (10 μm) were fixed with 4% paraformaldehyde in PBS and subjected to in situ hybridization as described (Wilkinson, 1992). Tissue sections were incubated with 1×10cts/minute of P-labelled antisense riboprobe in a total volume of 80 μl. Probes were synthesized from DNA templates using [P]UTP (1000-3000 Ci/mmol, Amersham) using the following plasmid templates, linearizing restriction endonucleases and polymerases: GATA-4, pBlue-script SK phagemid λG14a (Arceci et al., 1993), BamHI, T7; GATA-6, pCRII subclone, EcoRV, Sp6 (Narita et al., 1996); αMHC, pBluescript II KS subclone of an EcoRI-HindIII fragment encompassing exons 1-3 of the gene, EcoRI, T3 provided by the laboratory of J. Robbins; MLC-2v, pBluescript II KS cDNA provided by the laboratory of K. Chien (Miller-Hance et al., 1993), BamHI, T7; cTnC, pBluescript II KS plasmid containing a 0.5 kb EcoRI to BamHI sublclone of the cDNA, BamHI, T3; dHAND, pBluescript II SK plasmid containing the cDNA provided by Drs D. Shrivastiva and E. Olson, EcoRI, T7; eHAND, pBluescript II SK plasmid containing the cDNA provided by Drs D. Shrivastiva and E. Olson, NotI, T7; Nkx2.5/Csx, pBluescript II SK plasmid containing the cDNA provided by Dr S. Izumo (Komuro and Izumo, 1993), SacI, T7.
Immunofluorescent microscopy
Wild-type or Gata4−/− ES cells were differentiated for 5 days in hanging drop culture and then plated onto a gelatin-coated chamber slide. After 4 additional days of culture, the cells were fixed with 4% paraformaldehyde in PBS and subjected to indirect immunofluorescence (Harlow and Lane, 1988) using a monoclonal antibody against cardiac myosin (Lee et al., 1995), generously provided by Drs Paul Allen and Stacy Smith of Washington University, followed by a rhodamine-congugated rabbit anti-mouse antibody (Boehringer-Mannheim). The stained sections were visualized on a Molecular Dynamics confocal microscope.
To visualize GATA-4 antigen in chimeric heart, frozen tissue sections (10 μm) were fixed with 4% paraformaldehyde in PBS and stained with affinity-purified rabbit anti-GST-GATA-4 antibody (Arceci et al., 1993; Heikinheimo et al., 1994), followed by a rhodamine-congugated goat anti-rabbit antibody (Boehringer-Mannheim). Staining was visualized on a Molecular Dynamics confocal microscope. Nuclei in these same tissue sections were stained with bisBenzimide (Sigma) and photographed on a Zeiss Axioplot fluorescence microscope.
RESULTS
In vitro differentiation of Gata4−/− ES cells into cardiomyocytes
When differentiated in vitro into embryoid bodies, Gata4−/− ES cells display a defect in the differentiation of visceral yolk sac endoderm (Soudais et al., 1995). Expression of visceral endoderm markers, such as α-fetoprotein and hepatocyte nuclear factor-4, is markedly reduced in these mutant embryoid bodies, whereas a variety of other differentiation markers are normally induced (Soudais et al., 1995). To see whether GATA-4 is essential for cardiac lineage committment in vitro, we first examined suspension cultures of Gata4−/− embryoid bodies for evidence of cardiomyocyte differentiation. After 10-16 days in culture, a small percentage (<5%) of the Gata4−/− embryoid bodies exhibited rhythmic beating, consistent with the formation of cardiomyocytes. These beating embryoid bodies were isolated and subjected to electron microscopy to confirm the presence of cardiomyocytes. In agreement with our earlier report (Soudais et al., 1995), the surface of these embryoid bodies lacked the morphological features of visceral endoderm (microvilli, endocytic vacuoles, etc.), verifying that these embryoid bodies were derived from GATA-4-deficient cells (Fig. 1A). Within these same visceral endoderm-deficient embryoid bodies, we observed cells with the hallmark morphological features of car diomyocytes, including sarcomeres, abundant glycogen and junctional complexes (Fig. 1B). These preliminary results suggested that Gata4−/− ES cells retain the ability to differentiate into cardiomyocytes in vitro.
Electron microscopy of a Gata4−/− embryoid body demonstrating the presence of cardiomyocytes. Gata4−/− ES cells were differentiated for 12 days in suspension culture and a beating embryoid body was isolated and subjected to electron microscopy. (A) The surface cells of this embryoid body lack the morphological features of visceral endoderm (microvilli, abundant endocytic vacuoles, etc.), verifying that this embryoid body is GATA-4 deficient. (B) Within the embryoid body are cardiomyocytes containing sarcomeres and glycogen deposits. Adjacent cardiomyocytes are joined by a junctional complex (arrow). Bar, 0.5 nm.
Electron microscopy of a Gata4−/− embryoid body demonstrating the presence of cardiomyocytes. Gata4−/− ES cells were differentiated for 12 days in suspension culture and a beating embryoid body was isolated and subjected to electron microscopy. (A) The surface cells of this embryoid body lack the morphological features of visceral endoderm (microvilli, abundant endocytic vacuoles, etc.), verifying that this embryoid body is GATA-4 deficient. (B) Within the embryoid body are cardiomyocytes containing sarcomeres and glycogen deposits. Adjacent cardiomyocytes are joined by a junctional complex (arrow). Bar, 0.5 nm.
To assess the relative capacities of wild-type and Gata4−/− ES cells to differentiate into cardiomyocytes, we employed a two-stage, hanging drop culture system similar to those described elsewhere (Rohwedel et al., 1994; Metzger et al., 1994; Bautch et al., 1996). Embryoid bodies were grown for 3-8 days in hanging drop culture and then transferred to individual plastic wells and allowed to adhere. These adherent bodies were maintained in culture for up to 20 days and were monitored daily for the presence of beating cardiomyocytes. In wild-type embryoid bodies, beating cardiomyocytes were first evident at 7-8 days of differentiation. Embryoid bodies containing at least one contracting cardiocyte at any point during the secondary culture were scored as positive. For comparisons of cardiomyocyte differentiation, this hanging drop culture method has advantages over conventional one-stage suspension cultures. Beating cardiomyocytes can be more reliably seen and scored in adherent cultures, and a high percentage of wild-type embryoid bodies grown under these conditions contain cardiomyocytes. As shown in Fig. 2, wild-type embryoid bodies differentiated into cardiomyocytes more frequently than Gata4−/− embryoid bodies. When cultured for 5 days as a hanging drop before secondary plating, 100% of wild-type embryoid bodies contained cardiomyocytes, whereas only 25% of the mutant embryoid bodies had cardiomyocytes (Fig. 2). To confirm that beating cells within the two-stage embryoid body cultures were cardiomyocytes rather than skeletal or smooth muscle cells, we utilized immunofluorescence microscopy with a monoclonal antibody specific for cardiac myosin. Secondary platings of both wild-type and Gata4−/− embryoid bodies contained contracting cells that expressed abundant cardiac myosin (Fig. 3A,B). Whether the reduced propensity of Gata4−/− ES cells to differentiate into cardiomyocytes in vitro reflects a cell autonomous defect in cardiac differentiation or an indirect effect, such as a lack of trophic support by yolk sac endoderm, cannot be distinguished from these experiments. However, in vivo studies detailed below suggest that the effect is not cell autonomous.
In vitro differentiation of wild-type and Gata4−/− ES cells into beating cardiomyocytes. Individual embryoid bodies derived from either wild-type (hatched bars) or Gata4−/− (solid bars) ES cells were grown in hanging drop culture for the indicated lengths of time and then transferred to plastic wells, allowed to adhere and maintained in culture for up to 20 additional days. Adherent embryoid bodies containing beating cardiocytes were scored as positive.
In vitro differentiation of wild-type and Gata4−/− ES cells into beating cardiomyocytes. Individual embryoid bodies derived from either wild-type (hatched bars) or Gata4−/− (solid bars) ES cells were grown in hanging drop culture for the indicated lengths of time and then transferred to plastic wells, allowed to adhere and maintained in culture for up to 20 additional days. Adherent embryoid bodies containing beating cardiocytes were scored as positive.
Expression of cardiac myosin in cardiomyocytes derived from in vitro differentiation of wild-type and Gata4−/− ES cells. Adherent cultures of (A) wild-type or (B) Gata4−/− embryoid bodies were subjected to confocal immunofluorescence microscopy with a monoclonal antibody against cardiac myosin heavy chain. Bars, 10 μm.
Expression of cardiac myosin in cardiomyocytes derived from in vitro differentiation of wild-type and Gata4−/− ES cells. Adherent cultures of (A) wild-type or (B) Gata4−/− embryoid bodies were subjected to confocal immunofluorescence microscopy with a monoclonal antibody against cardiac myosin heavy chain. Bars, 10 μm.
In vivo differentiation of Gata4−/− ES cells into cardiomyocytes
To test whether the Gata4−/− ES cells can contribute to the developing heart in vivo, we analyzed chimeric mice. Gata4−/− ES cells were injected into 8-cell-stage embryos derived from ROSA26 mice, which bear a β-galactosidase transgene that is ubiquitously expressed throughout development (Friedrich and Soriano, 1992). The injected embryos were then implanted into pseudopregnant females. Embryos and attached yolk sacs were harvested at various stages of development and stained for β-galactosidase activity using X-gal. Host-derived cells in the chimeric embryos could be readily distinguished from Gata4−/− ES cell descendants on the basis of β-galactosidase expression, both in whole-mount preparations and in tissue sections (Fig. 4).
Gata4−/− ES cells can differentiate into normal appearing hearts in 9-10 day p.c. highly chimeric mouse embryos. Gata4−/− ES cells were introduced into ROSA26 8-cell-stage embryos and the resultant chimeras were harvested at 9-10 days p.c. Whole-mount embryos or tissue sections were stained with X-gal. (A) A non-chimeric ROSA26 embryo. Bar, 500 μm. (B) High percentage Gata4−/− ES↔ROSA26 chimeric embryo. The heart is ES cell-derived, whereas the yolk sac contains host-derived cells. Bar, 250 μm. (C)Section through the heart of a high percentage Gata4−/− ES↔ROSA26 chimeric embryo similar to that shown in B. The endocardium and myocardium are derived exclusively from Gata4−/− ES cells, while the visceral yolk sac endoderm is derived from host cells. Bar, 50 μm. (D) Electron microscopy of myocardial cells in the heart of the highly chimeric Gata4−/− ES↔ROSA26 embryo shown in B. Note the presence of sarcomeres and glycogen. No electron-dense X-gal precipitates are evident in this cell. Bar, 1 nm. Abbreviations: e, endocardium; ht, heart; m, myocardium; ve, visceral endoderm of the yolk sac; ys, yolk sac.
Gata4−/− ES cells can differentiate into normal appearing hearts in 9-10 day p.c. highly chimeric mouse embryos. Gata4−/− ES cells were introduced into ROSA26 8-cell-stage embryos and the resultant chimeras were harvested at 9-10 days p.c. Whole-mount embryos or tissue sections were stained with X-gal. (A) A non-chimeric ROSA26 embryo. Bar, 500 μm. (B) High percentage Gata4−/− ES↔ROSA26 chimeric embryo. The heart is ES cell-derived, whereas the yolk sac contains host-derived cells. Bar, 250 μm. (C)Section through the heart of a high percentage Gata4−/− ES↔ROSA26 chimeric embryo similar to that shown in B. The endocardium and myocardium are derived exclusively from Gata4−/− ES cells, while the visceral yolk sac endoderm is derived from host cells. Bar, 50 μm. (D) Electron microscopy of myocardial cells in the heart of the highly chimeric Gata4−/− ES↔ROSA26 embryo shown in B. Note the presence of sarcomeres and glycogen. No electron-dense X-gal precipitates are evident in this cell. Bar, 1 nm. Abbreviations: e, endocardium; ht, heart; m, myocardium; ve, visceral endoderm of the yolk sac; ys, yolk sac.
In initial experiments, we generated highly chimeric embryos by introducing up to four Gata4−/− ES cells into 8-cell-stage ROSA26 embryos. As is typical of chimeras derived nearly exclusively from ES cells (Nagy et al., 1993), these high percentage Gata4−/− ES↔ROSA26 chimeric embryos did not survive to term. However, at 8-10 days p.c., we did observe intact embryos that were essentially totally ES cell-derived (Fig. 4B). These embyros were slightly growth retarded compared to littermates, but were otherwise unremarkable in their gross morphology (Fig. 4B). The hearts in these embryos beat and were normally looped. Sections through these hearts revealed normal appearing endocardium and myocardium that did not stain with X-gal and were therefore completely derived from Gata4−/− ES cells (Fig. 4C). In contrast, the visceral yolk sac endoderm of these animals was derived exclusively from host cells and stained intensely with X-gal, both in whole-mount preparations (Fig. 4B) and in tissue sections (Fig. 4C). Control tissue sections through non-chimeric ROSA26 embryos revealed uniform, abundant β-galactosidase expression in endocardium and myocardium (data not shown), establishing that the β-galactosidase transgene was not subject to random loss or inactivation during cardiac differentiation. Heart tissue from 9 day p.c. high percentage Gata4−/− ES↔ ROSA26 embryo was also subjected to electron microscopy, which confirmed the presence of cardiomyocytes that contained sarcomeres and glycogen (Fig. 4D). These same car-diomyocytes lacked electron-dense X-gal precipitates (Papaioannou and Johnson, 1993), verifying that these cells were derived from Gata4−/− ES cells rather than host cells.
We next examined the ability of Gata4−/− ES cells to contribute to cardiomyocytes in lower percentage (20-60%) chimeras, which survived to later stages of development. These less chimeric animals were generated by introducing 1-2 Gata4−/− ES cells into 8-cell-stage ROSA26 embryos. Whole-mount X-gal staining could not be used to detect ES cell contribution to animals older than 10.5 days p.c., because of poor substrate penetration into tissue. This technical limitation was circumvented by first preparing frozen tissue sections and then staining these sections with X-gal to identify chimeric heart tissue. To limit the number of embryos sectioned and stained, the tails of potentially chimeric 12-18 day p.c. fetuses were screened for ES cell contribution using GPI isoenzyme analysis. Animals displaying mosaicism in the tail were then cryo-sectioned, and a subset of the tissue sections were stained with X-gal to assess ES cell contribution to the heart. Adjacent unstained sections were reserved for in situ hybridization studies, as detailed below. We found that tail mosaicism of >30% generally predicted mosaicism in the heart.
X-gal staining of sections through the hearts of 12-18.5 day p.c. chimeric animals showed that Gata4−/− ES cells can contribute to endocardium, myocardium, epicardium and valve tissue (Figs 5-8). Large, wedge-shaped patches of β-galactosidase-negative cells were observed in ventricular myocardium (e.g., see Fig. 5C). The distribution of ES cell descendants in these mosaic mouse hearts was reminiscent of earlier retroviral lineage tracing studies on chick heart, which showed that the progeny of individual precardiac mesodermal cells form cone-shaped myocyte colonies that can extend across the full thickness of the ventricular myocardium (Mikawa et al., 1992a,b; Mikawa and Fischman, 1992). Gata4−/− ES cells contributed to myocardial cells in the atrium, the ventricular trabeculae, the interventricular septum and the ventricular free wall (Figs 5-9). Comparing patches of ventricular myocardium derived from ES cells versus host cells, no differences were evident in the ventricular wall thickness or degree of trabeculation. These findings verified that myocardium derived from Gata4−/− ES cells undergoes two of the later stages of cardiac development, trabeculation and expansion of the compact zone.
Expression of GATA-4 mRNA, GATA-6 mRNA and β-galactosidase in adjacent sections through the myocardium of a 14 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. The left side of the Fig. shows low magnification views of in situ hybridizations for (A) GATA-4 or (B) GATA-6. The right side of the figure show high magnification views of (C) X-gal staining and in situ hybridizations for (D) GATA-4 or (E) GATA-6. These high magnification views correspond to the boxed region in A. Note that X-gal staining in the heart correlates with GATA-4 mRNA expression and that GATA-6 message is expressed in regions lacking GATA-4 mRNA. Abbreviations: a, atrial myocardium; v, ventricular myocardium. Bars, 100 μm (A,B), 50 μm (C-E).
Expression of GATA-4 mRNA, GATA-6 mRNA and β-galactosidase in adjacent sections through the myocardium of a 14 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. The left side of the Fig. shows low magnification views of in situ hybridizations for (A) GATA-4 or (B) GATA-6. The right side of the figure show high magnification views of (C) X-gal staining and in situ hybridizations for (D) GATA-4 or (E) GATA-6. These high magnification views correspond to the boxed region in A. Note that X-gal staining in the heart correlates with GATA-4 mRNA expression and that GATA-6 message is expressed in regions lacking GATA-4 mRNA. Abbreviations: a, atrial myocardium; v, ventricular myocardium. Bars, 100 μm (A,B), 50 μm (C-E).
Previously we demonstrated that differentiated Gata4−/− ES cells do not express detectable GATA-4 mRNA by in situ hybridization or RNase protection (Soudais et al., 1995). To confirm that an absence of β-galactosidase activity correlates with GATA-4 mRNA deficiency, we performed X-gal staining and in situ hybridization for GATA-4 message on adjoining sections through a chimeric heart (Figs 5, 6). As expected, all patches of β-galactosidase-negative myocardium (Figs 5C, 6B) did not express GATA-4 mRNA (Figs 5A,D, 6D), while adjacent β-galactosidase-positive tissue expressed GATA-4 message. As a further test of the system, we prepared Gata4+/+ ES↔cell ROSA26 embryos and verified that GATA-4 mRNA was expressed in β-galactosidase-negative cardiac tissue (data not shown). Thus, the absence of X-gal staining correctly identified ES-cell-derived cardiac tissue in chimeric animals.
Expression of GATA-4 mRNA, GATA-6 mRNA and β-galactosidase in adjacent sections through the myocardium of a 17 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. (A) H&E-stained section. (B) X-gal-stained section. (C,D) In situ hybridizaton for GATA-6 and GATA-4, respectively. Note that X-gal staining in the heart correlates with GATA-4 mRNA expression and that GATA-6 message is expressed in regions lacking GATA-4 mRNA. Abbreviations: a, atrium; av, atrioventricular valve; pa, pulmonary artery; v, ventricular myocardium. Bar, 50 μm.
Expression of GATA-4 mRNA, GATA-6 mRNA and β-galactosidase in adjacent sections through the myocardium of a 17 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. (A) H&E-stained section. (B) X-gal-stained section. (C,D) In situ hybridizaton for GATA-6 and GATA-4, respectively. Note that X-gal staining in the heart correlates with GATA-4 mRNA expression and that GATA-6 message is expressed in regions lacking GATA-4 mRNA. Abbreviations: a, atrium; av, atrioventricular valve; pa, pulmonary artery; v, ventricular myocardium. Bar, 50 μm.
Next, we confirmed that GATA-4 protein was absent from β-galactosidase-negative heart tissue in Gata4−/− ES cell ROSA26 embryos. X-gal staining and indirect immunofluorescence for GATA-4 were performed on adjacent sections through chimeric hearts. GATA-4 antigen was detected in the nuclei of host-derived myocardial cells (Fig. 7A), but not in the nuclei of ES cell-derived myocardial cells, which were visualized with the nuclear dye bisBenzimide (Fig. 7B).
Expression of GATA-4 protein in myocardium from an 18 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. (A) Indirect immunofluorescence with affinity-purified anti-GATA-4 antibody. (B) Counterstaining of the same tissue section with bisBenzimide to visualize all nuclei. Note the absence of GATA-4 protein within the nuclei of myocardial cells in the left-center of this tissue section. Staining of an adjacent section with X-gal showed that β-galactosidase expression correlated with GATA-4 protein (not shown). Bar, 50 μm.
Expression of GATA-4 protein in myocardium from an 18 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. (A) Indirect immunofluorescence with affinity-purified anti-GATA-4 antibody. (B) Counterstaining of the same tissue section with bisBenzimide to visualize all nuclei. Note the absence of GATA-4 protein within the nuclei of myocardial cells in the left-center of this tissue section. Staining of an adjacent section with X-gal showed that β-galactosidase expression correlated with GATA-4 protein (not shown). Bar, 50 μm.
Expression of cardiac differentiation markers in Gata4−/− myocardium
We went on to use in situ hybridization of mosaic heart tissue to examine the expression of a series of differentiation markers by Gata4−/− cardiocytes, including genes thought to be regulated by GATA-4. Chimeric hearts from 14-18 day p.c. embryos were subjected to in situ hybridization. To optimize our ability to detect differences between wild-type and Gata4−/− tissue, all in situ hybridizations were exposed to photographic emulsion for varying lengths of time to ensure that grain accumulation was linear with respect to time.
We assessed the expression of several myofilament genes in the atria and ventricles of chimeric embryos. One gene presumed to be regulated by GATA-4, αMHC (Molkentin et al., 1994; Huang et al., 1995), was abundantly expressed in Gata4−/− myocardium (Fig. 8A,B). Another putative target gene for GATA-4, cTnC (Ip et al., 1994), was expressed equally well by wild-type and Gata4−/− myocardium (Fig. 8C,D). The ventricle-specific transcript, MLC-2v, a relatively late marker of myocyte differentiation and chamber specification (Miller-Hance et al., 1993; Kubalak et al., 1994; Lyons et al., 1995), was also present in Gata4−/− cardiomyocytes (Fig. 8E,F). These findings indicate that GATA-4 is not essential for myofilament gene expression in cardiomyocytes.
Expression of myofibrillar genes in Gata4−/− myocardium from a 15 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. Adjacent tissue sections were subjected to X-gal staining (A,C,E) or in situ hybridization for cardiac αMHC (B), cardiac TnC (D) or cardiac MLC-2v (F). Note that each of these cardiac transcripts is abundantly expressed in Gata4−/− ventricular myocardium, as indicated by the arrowheads labelled v. Abbreviations: a, atrium; ec, endocardial cushion tissue; v, ventricular myocardium, Gata4−/−. Bar, 100 μm.
Expression of myofibrillar genes in Gata4−/− myocardium from a 15 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. Adjacent tissue sections were subjected to X-gal staining (A,C,E) or in situ hybridization for cardiac αMHC (B), cardiac TnC (D) or cardiac MLC-2v (F). Note that each of these cardiac transcripts is abundantly expressed in Gata4−/− ventricular myocardium, as indicated by the arrowheads labelled v. Abbreviations: a, atrium; ec, endocardial cushion tissue; v, ventricular myocardium, Gata4−/−. Bar, 100 μm.
We also examined expression of a group of transcription factors that are expressed selectively in cardiac tissue and have been implicated in the regulation of cardiac development. The gene encoding the cardiac homeodomain protein Nkx2.5/Csx (Komuro and Izumo, 1993; Tonissen et al., 1994; Lints et al., 1993), which is essential for normal heart formation and expression of MLC-2v (Lyons et al., 1995), was abundantly expressed in GATA-4-deficient myo-cardium (Fig. 9A,B). Genes for two other transcription factors essential for normal cardiogenesis, dHAND and eHAND (Cserjesi et al., 1995), were expressed equally well in wild-type and GATA-4-deficient cardiac tissue (Fig. 9C-F). Thus, GATA-4 is not required for expression of any of these three transcription factor genes.
Expression of transcription factor genes in Gata4−/− myocardium from a 15 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. Adjacent tissue sections were subjected to X-gal staining (A,C,E) or in situ hybridization for Nkx2.5/Csx (B), dHAND (D) or eHAND (F). Note that each of these cardiac transcripts is expressed in Gata4−/− ventricular myocardium, as indicated by the arrowheads labelled v. Abbreviations: a, atrium; ec, endocardial cushion tissue; v, ventricular myocardium, Gata4−/−. Bar, 100 μm.
Expression of transcription factor genes in Gata4−/− myocardium from a 15 day p.c. Gata4−/− ES↔ROSA26 chimeric embryo. Adjacent tissue sections were subjected to X-gal staining (A,C,E) or in situ hybridization for Nkx2.5/Csx (B), dHAND (D) or eHAND (F). Note that each of these cardiac transcripts is expressed in Gata4−/− ventricular myocardium, as indicated by the arrowheads labelled v. Abbreviations: a, atrium; ec, endocardial cushion tissue; v, ventricular myocardium, Gata4−/−. Bar, 100 μm.
These in situ hybridization studies confirm that GATA-4-deficient cells can differentiate into cardiomy-ocytes that express a wide range of cardiac-specific transcripts. Furthermore, these results establish that GATA-4 is not required for in vivo expression of several genes believed to be targets for activation by this transcription factor. In an effort to understand why GATA-4 was not essential for in vivo expression of these differentiation markers, we also examined expression of GATA-6 myocardium throughout development. We found that Gata4−/− myocardium expresses abundant GATA-6 message (Figs 5B,E, 6C), indicating that GATA-6 mRNA expression in the heart does not require transcription factor GATA-4. That GATA-6 mRNA in the chimeric heart tissue. Previous studies have shown that GATA-6 mRNA is present in atrial and ventricular message is present in Gata4−/− myocardium raises the possibility that the presence of any GATA-binding protein (GATA-4, GATA-5 or GATA-6) in heart may be sufficient for expression of cardiac genes.
DISCUSSION
A variety of previous studies implicated transcription factor GATA-4 in the regulation of cardiomyocyte gene expression and differentiation. First, the temporal and spacial patterns of GATA-4 expression suggested a role for this factor in cardiac lineage committment (Heikinheimo et al., 1994). During formation and bending of the heart tube (8 days p.c.), GATA-4 RNA and protein are expressed in endocardium, myocardium and embryonic structures containing precardiac mesoderm such as the septum transversum and intraembryonic coelomic epithelium. Expression of GATA-4 by the myocardium continues through gestation and after birth, supporting a role for this factor in the maintenance of the cardiac phenotype (Heikinheimo et al., 1994). Second, in vitro co-transfection experiments showed that GATA-4 binds and trans-activates the promoters and/or enhancers of several genes expressed selectively or exclusively in cardiac tissue, including αMHC (Molkentin et al., 1994; Huang et al., 1995), cTnC (Ip et al., 1994) and BNP (Thuerauf et al., 1994; Grépin et al., 1994). Third, introduction of GATA-4 mRNA into Xenopus embyros resulted in transient, ectopic expression of cardiac actin and αMHC (Jiang and Evans, 1996).
In this study, Gata4−/− ES cells were used to test whether this transcription factor is necessary for cardiomyocyte lineage commitment and gene expression. Two systems were used to assess the ability of GATA-4-deficient ES cells to differentiate into cardiomyocytes. First, we examined embryoid body differentiation, an we determined the ability of Gata4−/− ES cells to contribute to cardiomyocytes in vivo by analyzing chimeric mice. We found that GATA-4-deficient ES cells retain the capacity to differentiate in vitro into cardiomyocytes, although Gata4−/− embryoid bodies differentiate into cardiomyocytes less often than wild-type embryoid bodies. The reduced propensity of Gata4−/− embryoid bodies to form cardiomyocytes could reflect either a cell autonomous defect in cardiac differentiation or an indirect effect, such as a lack of growth factor production or nutritional support. Gata4−/− embryoid bodies are deficient in visceral yolk sac endoderm (Soudais et al., 1995), a cell layer implicated in nutrient uptake (Gardner, 1983) and trophic support (Coucouvanis and Martin, 1995; Palis et al., 1995). In theory, this lack of visceral endoderm could impair cardiomyocyte differentiation in embryoid bodies by compromising nutrient transport or production of growth/differentiation factors. Such a possibility is not unreasonable, given that disruption of embryonic endoderm has been shown to adversely affect cardiogenesis in established in vitro model of cardiomyogenesis (Doetschman et al., 1985; Robbins et al., 1992; Miller-Hance et al., 1993). Second, frog (Nascone and Mercola, 1995) and chick (Schultheiss et al., 1995; Gannon and Bader, 1995) embryos. Additional evidence for an indirect effect comes from our studies with chimeric mice showing that Gata4−/− ES cells retain the capacity to differentiate in vivo into apparently normal, contracting cardiomyocytes that express a variety of cardiac markers, including cTnC, αMHC, MLC-2v, Nkx-2.5/Csx, GATA-6, dHAND and eHAND. Hence, GATA-4 is not essential for cardiomyocyte differentiation in vitro or in vivo.
Our conclusions on the role of GATA-4 in cardiac differentiation differ somewhat from those made on the basis of antisense inhibition studies with P19 embryonal carcinoma cells (Grépin et al., 1995). Stable expression of GATA-4 antisense RNA in P19 cells completely blocks DMSO-induced differentiation into cardiomyocytes and expression of BNP, cTnC, αMHC, βMHC and MLC-1a transcripts (Grépin et al., 1995). These antisense studies led to the suggestion that GATA-4 is essential for cardiomyocyte gene expression and differentiation (Grépin et al., 1995). Our experiments with null ES cells confirm that GATA-4 deficiency reduces in vitro cardiomyo-genesis, but establish that this factor is not essential for cardiac gene expression or differentiation, either in vitro or in vivo. We assume that the discrepancies between studies with ES and P19 cells reflect differences in these two model systems of cardiomyogenesis. ES cells are totipotential and spontaneously differentiate into cardiomyocytes, whereas P19 embryonal carcinoma cells are less pluripotential and require DMSO to induce cardiac differentiation. That GATA-4 appears essential in some cell models of cardiac differentiation but not others is reminiscent of the situation encountered with MyoD in studies of skeletal myogenesis; in certain experimental cell systems (e.g., 10TG cells) MyoD appeared to be critical for skeletal muscle differentiation, while in null ES cells and knockout animals this transcription factor proved to be non-essential because of compensation by other basic helix-loop-helix family members (Edmondson and Olson, 1993; Rudnicki et al., 1992).
Jiang and Evans (1996) first proposed that GATA-4, GATA5, and GATA-6 may have overlapping functions in the heart based on the observation that injection of Xenopus embryos with mRNA encoding any one of these three factors results in premature expression of cardiac contractile genes. Our observation that GATA-6 message is abundantly expressed in Gata4−/− myocardium raises that possibility that the presence of any one of the three cardiac GATA-binding proteins (GATA-4, GATA-5, or GATA-6) is sufficient for cardiac differentiation and in vivo expression of target genes such as αMHC and cTnC. It is also conceivable that GATA-5 or GATA-6 can be induced to compensate for a deficiency in GATA-4 expression, although we have no direct evidence to support this notion. A precedent exists for functional redundancy among cardiac transcription factors. The basic helix-loop-helix proteins eHAND and dHAND appear to have overlapping roles in the regulation of cardiac morphogenesis; incubation of stage 8 chick embryos with dHAND or eHAND antisense oligonucleotides revealed that either oligonucleotide alone had no effect on embryogenesis, whereas together the oligonucleotides arrested development at the looping heart stage (Srivastava et al., 1995).
By coupling X-gal staining with in situ hybridization, we have shown that null ES ROSA26 chimeras can be used to test whether a particular transcription factor is essential for in vivo gene expression in the heart. At the same time, certain limitations of chimera analysis must be kept in mind. Analysis of Gata4−/− ES ROSA26 chimeras may not detect subtle problems in cardiomyocyte function, development or longevity caused by GATA-4 deficiency. Addressing these possibilities awaits analysis of Gata4 null mice, assuming that GATA-4 deficiency does not result in a lethal phenotype due to yolk sac deficiency.
ACKNOWLEDGEMENTS
We than K. Green and J. Saffitz for performing the electron microscopy. We are grateful to the laboratories of D. Srivastava, E. Olson, K. Chien, S. Izumo and M. Parmacek for providing cDNA probes. We also thank P. Allen and S. Smith for providing the monoclonal antibody against cardiac myosin. The Gata4−/− ES lines were prepared in collaboration with C. Soudais, C. Simon and J. Leiden of the University of Chicago, and we are indebted to the assistance and generosity of these investigators. We thank J. Gordon and members of his laboratory for suggestions regarding the ROSA26 chimeras, Bill Coleman for assistance with the confocal microscope, Kevin Roth for advice on immunofluorescence, and finally J. Gitlin and C. MacArthur for critiquing the manuscript. Supported by NIH Grant HL52134, AHA Grant 94-605, the March of Dimes and the Monsanto/Searle-Washington University Biomedical Agreement. D. B. W. is an Established Investigator of the AHA.