GATA transcription factors integrate Wnt signalling during heart development

The GATA family of transcription factors plays a pivotal role near the top of the cascade of events that lead to heart muscle differentiation(Brewer and Pizzey, 2006; Peterkin et al., 2005; Peterkin et al., 2007). Vertebrate GATA4, GATA5 and GATA6 are expressed in early cardiogenic tissue(Molkentin, 2000; Patient and McGhee, 2002), and regulate cardiomyogenesis, in a partially redundant manner(Holtzinger and Evans, 2007; Kuo et al., 1997; Molkentin et al., 1997; Peterkin et al., 2007). However, the precise roles and molecular targets of GATA transcription factors in vertebrate heart formation are just beginning to be discovered.

Extracellular signals that regulate cardiogenesis have been primarily identified by their ability to induce cardiac differentiation in non-cardiac mesoderm. Among these are Wnt signals, such as Wnt11(Pandur et al., 2002), but also secreted Wnt antagonists, such as Dickkopf-1 and Crescent(David et al., 2008; Marvin et al., 2001; Schneider and Mercola, 2001). Wnt antagonists and Wnt signals were both found to promote heart development because cardiogenesis requires both repression of canonical Wnt/β-catenin signalling, which would otherwise inhibit cardiogenesis(Marvin et al., 2001; Schneider and Mercola, 2001),and activation of non-canonical Wnt11/JNK signalling, which promotes cardiogenesis (Eisenberg and Eisenberg,1999; Pandur et al.,2002; Terami et al.,2004). These findings raise the important issue of how these two Wnt signalling pathways interact to control cardiogenesis.

Because of their prominent roles in cardiogenesis, we investigated the relationship between GATA transcription factors and the two Wnt signalling pathways.

GATA4GR and GATA6GR, activin (Afouda et al., 2005) and Dickkopf-1(Semenov et al., 2001) mRNA expression constructs have been previously described. β-cateninGR is a hormone-inducible β-catenin. Wnt11(Pandur et al., 2002), GATA4(Afouda et al., 2005) and GATA6(Peterkin et al., 2003)morpholinos have been previously characterised.

For animal cap assays, Xenopus embryos were injected at the one-cell stage into the animal pole. Embryos were injected at the four-cell stage into ventral blastomeres for the ventral marginal zone (VMZ) explant experiments or into dorsal blastomeres for the dorsal marginal zone (DMZ)explant experiments and the whole embryo analysis. mRNA and morpholino (MO)were injected in sterile autoclaved water (10 nl). The total amount of MO injected was 5 ng (Wnt11 MO), 10 ng (GATA6 MO) and 20 ng (GATA4 MO) per embryo. Animal cap explants were dissected at stage 8 and cultured in 0.6×MMR as described previously(Afouda et al., 2005). Cycloheximide treatment was carried out as previously described(Tada et al., 1997). DMZ or VMZ explants were dissected at stage 10, when the prospective dorsal side is clearly identifiable, and cultured in 0.6×MMR. Explants were cultured until the appropriate control stage before processing for analysis of RNA expression. Sets of experiments were repeated at least twice in order to confirm results presented here. Cardiomyocyte differentiation of animal cap and DMZ explants was analysed at phenotypical level at control stage 45 and expressed as a percentage of rhythmically beating explants compared with non-beating, but otherwise healthy, explants.

Total RNA extraction and the reverse transcription (RT) reaction were performed as previously described (Afouda et al., 2005). PCR reactions on cDNA were calibrated to be strictly in the linear range (with a linearity control series for each figure)and compared with ODC expression control. Primer sequences and PCR conditions are available from the authors upon request.

The cardiogenesis-specific functions of Wnt signalling and GATA transcription factors are difficult to study in whole embryos because both Wnt signalling and GATA transcription factors are also important regulators at other developmental stages and in other embryonic tissues(Afouda et al., 2005; Liu et al., 2005; Tada and Smith, 2000). We have therefore adapted reliable Xenopus animal cap and cardiac mesoderm explant assays (Ariizumi et al.,2003; Latinkic et al.,2003; Schneider and Mercola,2001), which allow us to examine in isolation the regulatory mechanisms by which Wnt/β-catenin and Wnt11/JNK signalling, and GATA transcription factors, interact to regulate cardiogenesis.

Activin mRNA-injected Xenopus animal cap explants initiate cardiogenesis, as monitored by cardiogenic marker gene expression (GATA4,GATA6, Wnt11, Nkx2-5, MLC2 and TnIc; Fig. 1A) and by differentiation into functional heart muscle(Ariizumi et al., 2003). In order to confirm that Wnt/β-catenin signalling can inhibit development of cardiogenic mesoderm, we activated the pathway in a temporally controlled manner by co-injecting β-cateninGR mRNA (hormone-inducible version ofβ-catenin) and activating it with dexamethasone. Strikingly, the activin-induced expression of GATA transcription factors and other cardiogenic markers was abolished when β-cateninGR was activated at relatively early stages, when cardiogenic mesoderm is being induced (stage 8), but also later during subsequent heart development (stage 18)(Fig. 1A).

Because of the observed inhibition of endogenous GATA expression byβ-catenin signalling and the prominent cardiogenesis-promoting role of GATA factors (Latinkic et al.,2003; Peterkin et al.,2003; Peterkin et al.,2007), we tested whether GATA genes are the relevant targets of Wnt/β-catenin signalling in cardiogenesis. We reinstated GATA function with hormone-inducible GATA proteins(Afouda et al., 2005) that are activated at the same time as co-injected β-cateninGR was causing reduced expression of the endogenous GATA genes. The inhibition of cardiogenic marker gene expression by β-cateninGR was indeed rescued by activating GATA4GR and GATA6GR (Fig. 1A). This result suggests a regulatory pathway in which Wnt/β-catenin signalling restricts cardiogenesis by inhibiting GATA gene expression.

In order to substantiate our findings, we conducted similar experiments using dorsal marginal zone (DMZ) explants(Fig. 1B) and analysis in whole embryos (Fig. 1C). DMZ explants differentiate into heart tissue in the absence of added factors(Foley and Mercola, 2005; Schneider and Mercola, 2001). Activation of β-cateninGR either at stage 10 or 18 strongly reduced the expression of GATA genes and other heart development markers, yet their expression was restored by reinstated GATA4 or GATA6 activity(Fig. 1B). As expected, in the whole embryo analysis the β-catenin-mediated reduction of cardiogenic gene expression is only clearly detectable in the strictly heart muscle-specific genes MLC2 and TnIc; nevertheless, their expression is restored by activated GATA function (Fig. 1C, see Fig. S1 in the supplementary material). Our results demonstrate that GATA transcription factors are sufficient to overcome the negative regulation of cardiogenesis by Wnt/β-catenin signalling and therefore suggest that GATA transcription factors act downstream of Wnt/β-catenin signalling in cardiomyogenesis.

The requirement for GATA function in cardiogenesis is complicated by extensive redundancy between different GATA genes(Holtzinger and Evans, 2007; Peterkin et al., 2007). In order to test whether GATA function is required for Wnt11 expression and cardiomyogenesis, we used morpholinos to inhibit GATA4 and GATA6 expression. We first used activin-injected animal caps as before to induce cardiogenic marker gene expression (Fig. 2A) but also VMZ explants injected with Dkk-1 mRNA to induce ectopic cardiogenesis (Fig. 2B)(Foley and Mercola, 2005; Marvin et al., 2001; Schneider and Mercola, 2001)and intact embryos (Fig. 2C). In all three assays, we found that inhibition of either GATA4 or GATA6 alone caused some reduction in gene expression of Wnt11 and other cardiogenic markers, but a much stronger reduction when both were inhibited, confirming that Wnt11 gene expression and heart development are dependent on normal GATA function.

The pro-cardiogenic activity of GATA transcription factors in overexpression experiments (Gove et al.,1997; Latinkic et al.,2003; Reiter et al.,1999) is shared with Wnt11(Eisenberg and Eisenberg, 1999; Pandur et al., 2002; Terami et al., 2004), which is co-expressed with GATA factors during early cardiogenesis(Ku and Melton, 1993) (see Fig. S3 in the supplementary material). We therefore investigated whether Wnt11 is a direct target of GATA transcription factors in cardiomyogenesis. We activated GATAGR proteins with dexamethasone in the presence of protein synthesis inhibitors that prevent indirect gene induction, and assayed for Wnt11 and Nkx2-5 expression. Controls without dexamethasone showed no or low Nkx2-5 gene expression. With dexamethasone, Nkx2-5 expression was induced during early stages (Fig. 3A)and at later stages (Fig. 3B),even in the presence of cycloheximide, confirming Nkx2-5 as a direct target of GATA regulation (Brewer et al.,2005; Searcy et al.,1998). Importantly, our experiments show for the first time that Wnt11 gene expression is also directly regulated by GATA transcription factors(Fig. 3A,B), as Wnt11 gene expression was induced by dexamethasone-activated GATA4GR or GATA6GR, even when protein synthesis was inhibited by cycloheximide, unlike the negative control myosin heavy chain (MHCα).

GATA4 can drive full cardiomyogenesis in Xenopus animal pole explants, leading to cardiogenic gene expression and differentiation into beating cardiomyocytes (Latinkic et al.,2003). Furthermore, Wnt11 activation of a JNK-dependent pathway is required for cardiogenesis in Xenopus, as well as in cell culture models (Pandur et al., 2002). As we had found that expression of Wnt11 requires GATA function and that it was even among direct targets of regulation by GATA transcription factors, we investigated the requirement for Wnt11 downstream of GATA function using a Wnt11 morpholino (11MO) (Pandur et al.,2002) together with GATAGR mRNAs.

We found that either GATA4GR or GATA6GR were able to initiate cardiomyogenic gene expression (MLC2, TnIc) in animal cap explants(Fig. 4A), but that only GATA4GR was able to induce beating cardiomyocytes(Fig. 4B,D). We had earlier noticed subtle but consistent differences between GATA4GR and GATA6GR activities (Fig. 1, see Fig. S2 in the supplementary material), indicating that GATA4GR is a slightly stronger inducer of cardiogenic marker gene expression when activated at earlier stages, whereas GATA6GR the relatively stronger inducer when activated later. Our investigation does not address whether this is simply due to technical differences between the GATA4GR and the GATA6GR constructs or whether it reflects functional differences between the endogenous genes(Nemer, 2008; Peterkin et al., 2005; Peterkin et al., 2007; Xin et al., 2006; Zhao et al., 2008), but we think it is related to their differing abilities to induce beating cardiomyocytes (Charron et al.,2001; Latinkic et al.,2003). Nevertheless, morpholino-mediated inhibition of Wnt11 caused clear reduction of cardiomyogenic gene expression(Fig. 4A,C) and reduced differentiation into beating cardiomyocytes(Fig. 4D). These results show that Wnt11 is required to a significant degree for mediating the cardiogenesis-promoting activity of GATA factors.

To confirm that the need for Wnt11 function is not confined to artificially induced cardiogenesis, we tested whether Wnt11 function is also required for cardiomyogenic gene expression and cardiomyocyte differentiation in DMZ explants and in whole embryos. We found indeed that morpholino-mediated inhibition of Wnt11 causes reduced MLC2 and TnIc expression in DMZ explants and in whole embryos (Fig. 4E)and that it reduces the number of beating DMZ explants and embryos with a detectable heart beat (Fig. 4F). Our results therefore argue that Wnt11 is a key effector of cardiogenesis downstream of GATA transcription factors.

Our results suggest a regulatory pathway controlling cardiomyogenesis whereby GATA transcription factors link canonical and non-canonical Wnt signalling. We have confirmed that Wnt/β-catenin signalling inhibits cardiogenesis (Cohen et al.,2008; Eisenberg and Eisenberg,2006; Naito et al.,2006; Tzahor,2007; Ueno et al.,2007), that GATA4/6 promote cardiomyogenesis(Latinkic et al., 2003; Peterkin et al., 2003; Peterkin et al., 2007) and that Wnt11 is required for cardiomyogenesis(Pandur et al., 2002). We also show for the first time that GATA4/6 can overrule Wnt/β-catenin-mediated inhibition of cardiogenesis, that GATA4/6 directly induces Wnt11 expression and that GATA4/6 promotes myocardial differentiation largely via Wnt11. These findings establish a hierarchy of regulation with canonical Wnt/β-catenin signalling restraining GATA gene expression, which otherwise induces non-canonical Wnt11/JNK signalling to promote subsequent aspects of heart muscle differentiation.

Our results do not necessarily argue for a strictly linear pathway. We find that repression of GATA gene expression by activated Wnt/β-catenin signalling is often not complete, which is likely to reflect additional regulation of GATA gene expression, for example by Nkx transcription factors(Molkentin et al., 2000) and by Wnt11 (Pandur et al.,2002), which may represent reinforcing regulatory loops operating together with the regulatory hierarchical pathway we describe. Furthermore,Wnt11 is not the only cardiogenesis-promoting factor induced by activated GATA; Nkx2-5, a key regulator of heart development(Brewer et al., 2005; Cripps and Olson, 2002), is also induced, which may suggest an alternative pathway downstream of GATA4/6. Our results provide further evidence for this notion in the observed incomplete abolition of cardiogenic gene expression and only reduced differentiation into beating cardiomyocytes when Wnt11 is inhibited, which is also consistent with the relatively mild heart phenotypes observed in the zebrafish silberblick (wnt11) mutation(Matsui et al., 2005) and the mouse Wnt11 knockout (Majumdar et al.,2003).

We thank Professor Makoto Asashima (University of Tokyo) for primer sequences and PCR conditions, Dr Danielle Lavery for discussions, and Yvonne Turnbull for technical assistance. This work was funded by The Wellcome Trust(071101/Z/03/Z), British Heart Foundation (PG/07/043) and the Royal Society.