Frizzled-7 is required for Xenopus heart development

During embryogenesis, the heart is one of the first organs to form. Development of the heart includes specification of cardiac progenitors and formation of the linear heart tube by cell migration and morphogenetic movements (Mohun et al., 2000). In Xenopus, the heart begins to form during early gastrula stages when the cardiac progenitors arise in the dorsolateral mesoderm. Cell movements during gastrulation result in the dorso-anterior translocation of these regions and subsequent ventral migration during neurulation. The heart progenitors, which comprise cells fated to become primary or secondary heart field, form a linear heart tube at the ventral midline before looping and remodelling to form the beating heart (Kriegmair et al., 2013). Understanding the processes underlying heart development and morphogenesis are important for understanding congenital heart disease.

Heart formation is controlled by many signalling pathways including wnt signalling. Wnt6, wnt11, and wnt11-R have all been implicated in Xenopus heart development (Garriock et al., 2005; Gessert et al., 2008; Lavery et al., 2008a; Pandur et al., 2002). Wnt antagonists such as Dickkopf-1, Crescent and Sfrp1 have also been reported to control early heart formation (David et al., 2008; Foley and Mercola, 2005; Gibb et al., 2013; Marvin et al., 2001; Schneider and Mercola, 2001). Little is known however about which Frizzled proteins mediate these signals. Frizzled-7 (fzd7) has been well characterised in Xenopus laevis and other species. It has been shown to be involved in numerous developmental processes as well as being shown to be active in several forms of cancer (Huang and Klein, 2004; Liu et al., 2016; Schiffgens et al., 2016; Xu et al., 2016). Fzd7 has been demonstrated to interact with several wnts including Wnt5a (animal cap elongation assays), Wnt6 (in somite development), Wnt8 (co-immunoprecipitation assays, Xenopus axis duplication) and Wnt11 (gastrulation movements, neural crest development) (Hsieh et al., 1999; Linker et al., 2005; Medina et al., 2000; Medina and Steinbeisser, 2000; Umbhauer et al., 2000; Witzel et al., 2006). It has also been shown to genetically interact with the co-receptors ror2 and ryk (Hikasa et al., 2002; Kim et al., 2008). Xenopus fzd7 has been implicated in gastrulation movements, tissue separation, and neural crest induction (Abu-Elmagd et al., 2006; Djiane et al., 2000; Wheeler et al., 2000; Winklbauer et al., 2001). We have previously shown fzd7 to be expressed in the cardiac region throughout development (Wheeler and Hoppler, 1999). It has also been shown that specific depletion of fzd7 function in Xenopus foregut leads to impaired cardiac morphogenesis, but has no effect on heart specification (Zhang et al., 2013). Here, we further characterise its expression relative to known heart markers, and then use whole-embryo experiments to show that fzd7 is required for heart formation during early embryonic development.

Expression pattern analysis shows Xenopus fzd7 is expressed in the heart-forming regions throughout development (Wheeler and Hoppler, 1999). At stage 10.5 fzd7 is expressed in the dorsal mesoderm from which cardiac tissue originates (Wheeler and Hoppler, 1999) (Fig. 1A). As development progresses, fzd7 expression at stage 25 is maintained in the presumptive cardiac mesoderm as it migrates dorso-laterally to the ventral midline (Fig. 1C-Cii). By stage 29, fzd7 is expressed throughout the cardiac crescent in the cardiac mesoderm (Fig. 1E,Ei). fzd7 expression correlates with that of wnt11 (Fig. 1B, stage 10.5) where expression of both genes seem to be complementary in the presumptive heart region in the dorsal side of the embryo. fzd7 expression also correlates to that of wnt11-R (Fig. 1D-Dii,F,Fi, stages 25 and 29, respectively) where it is expressed in the anterior endoderm at stage 25 when fzd7 is expressed in the heart field. By stage 29, the expression of fzd7 and wnt11-R overlaps (Fig. 1E-Fi). As the heart continues to form, fzd7 is strongly expressed in the lateral plates of mesoderm, cardiac mesoderm, myocardium, and over time, is restricted to the pericardium (Wheeler and Hoppler, 1999) (Fig. 2A-Aii, Bii, Cii, Dii). Using double in situ hybridisation, we analysed fzd7 expression in correlation to that of early heart markers including nkx2-5, troponin-ic (tnnic) and gata6, which are all known to be required for Xenopus cardiogenesis (Afouda and Hoppler, 2011; Afouda et al., 2008; Drysdale et al., 1994; Flaherty and Dawn, 2008; Fu et al., 1998; Garriock et al., 2005; Jiang and Evans, 1996; Martin et al., 2010). fzd7 expression overlaps with that of nkx2-5 (Fig. 2B-Bii), tnnic (Fig. 2C-Cii) and gata6 (Fig. 2D-Dii) in the forming heart. Interestingly, none of these markers are seen in the pericardium except for fzd7 (Fig. 2Aii,Bii,Cii,Dii).

Microinjection into Xenopus embryo dorsal blastomeres at the 4- or 8-cell stage targets prospective mesoderm including cardiac tissue. In order to test the role of fzd7 in heart development, we inhibited its function by injecting either fzd7 morpholino (fzd7 MO) or its dominant-negative form expressing only the extracellular domain (cysteine rich domain, fzd7 CRD), which would disrupt fzd7 mediated signalling (Abu-Elmagd et al., 2006).

Microinjection of fzd7 MO into the dorsal blastomeres of 4- or 8-cell embryos leads to a reduction of both early cardiac marker nkx2-5 (Fig. 3B-Bii) and later cardiac marker tnnic expression (Fig. 3E-Eii). Adding increasing amounts of fzd7 MO leads to a progressively more severe phenotype with a greater number of embryos affected (Fig. 3C). In situ hybridisation for nkx2-5 and tnnic show embryos with mild convergent extension phenotypes (Fig. 3B,E), but a severe decrease in cardiac gene expression (Fig. 3Bi,Ei) while control morpholino (CMO) show normal heart (Fig. 3A,Ai,D,Di). Some embryos also showed anterior defects (data not shown). Sections through the cardiac region showed not only a decrease of nkx2-5 and tnnic expression, but an absence of recognisable heart structures (Fig. 3Bii,Eii) compared to CMO (Fig. 3Aii,Dii). The number of embryos injected with fzd7 MO which showed heart and/or convergent extension and anterior defects are shown in Table S1.

Overexpression of fzd7 full-length (fzd7 FL) results in severe convergent extension defects, but no cardiac phenotype (Fig. S1A,B). Knockdown with fzd7 MO can also cause a mild convergent extension phenotype and anterior defects (Abu-Elmagd et al., 2006). In order to test whether this cardiac effect is specific to fzd7, we rescued the fzd7 MO cardiac phenotype with fzd7 FL that has been mutated to not bind the fzd7 MO (fzd7 SDM, as described in Abu-Elmagd et al., 2006). Titrating increasing amounts of fzd7 SDM capped RNA from 250 pg to 1 ng results in a modest rescue of the cardiac phenotype (Fig. 3F,Fi; Table S2), thus showing that fzd7 is required for normal heart development.

Interestingly, injecting fzd7 FL at 8-cell stage embryos shows detectable expression of tnnic and nkx2-5, despite some of these embryos showing severe convergent extension movements phenotype (head arrows in Fig. S1A,B). This leads to the suggestion that heart phenotypes are not necessarily due to convergent extension secondary effects.

To further look at the effect of inhibiting fzd7 function, we took a dominant-negative approach using fzd7 CRD. Surprisingly, this did not give a similar result to the MO knockdown. Instead, increasing amounts of fzd7 CRD results in a dose-dependent increase in frequency and severity of cardia bifida. This was very similar to the phenotype seen for wnt-11R knockdown (Garriock et al., 2005). Embryos with very mild convergent extension movement defects displayed a severe cardia bifida phenotype as shown by tnnic (Fig. 4B-Bii,C) and nkx2-5 (Fig. 4G,Gi) expression. Control embryos showed normal expression of tnnic (Fig. 4A-Aii) and nkx2-5 (Fig. 4F,Fi). These results suggest that the cardia bifida phenotype is not a secondary effect of the convergent extension defect. Overexpression of fzd7 FL gives a severe convergent extension phenotype but no cardiac phenotype (Fig. S1A,B). Embryos with cardia bifida were unable to recover and form a normal heart when incubated up to stage 41 (n=23, data not shown). Embryos injected with a dominant-negative form of fzd3 (fzd3 CRD) into the dorsal blastomeres at 4-cell stage did not show cardia bifida (n= 27, Fig. 4E,Ei) indicating that the cardia bifida phenotype is specific to fzd7 CRD. Furthermore, this phenotypic specificity to fzd7 CRD was confirmed by rescuing the cardia bifida with fzd7 FL-capped RNA (Fig. 5A-D,F).

It has been previously reported that a Jun N-terminal kinases (Jun) inhibitor phenocopies the wnt11-R cardiac phenotype of effects on cardiac morphogenesis and heart tube fusion, suggesting signalling through the non-canonical pathway (Garriock et al., 2005; Gessert et al., 2008). We therefore determined to rescue the fzd7 CRD phenotype with dishevelled1-Delta-N (dvl1ΔN)-capped RNA. Dvl1ΔN-capped RNA can rescue fzd7 CRD (Fig. 5E,Ei,G,Gi; Table S3), suggesting that fzd7 is required for non-canonical wnt signalling during heart development.

Wnt signalling through the canonical and non-canonical pathways has been implicated in many aspects of heart development (Gessert and Kuhl, 2010; Ruiz-Villalba et al., 2016). How the wnt signals that arise from both non-cardiogenic and cardiogenic tissue are integrated into heart development is less well understood. Frizzled proteins are only a part of the increasingly complicated wnt-receptor complex found at the cell membrane, which can also include Lrp5/6, Ror2, Ryk and Kremen (Bryja et al., 2009; Korol et al., 2008; Mazzotta et al., 2016; van Wijk et al., 2009); however, Frizzled proteins are critical components of the Wnt receptor complex and so understanding their role in heart development is necessary to fully understand the signalling involved. We have previously shown that fzd7 is expressed throughout heart development, and in this study, we show that it is functionally required in both early and late heart development.

Morpholino knockdown of fzd7 leads to effects on heart development, including in some cases a complete loss of heart (Fig. 3). Overexpression of fzd7 gives rise to convergent extension defects as previously reported (Abu-Elmagd et al., 2006; Sumanas and Ekker, 2001; Winklbauer et al., 2001), but does not affect heart development. We can rescue the fzd7 MO phenotype by co-injecting site-directed mutagenized full-length fzd7 (Fig. 3). These results suggest that fzd7 is required for initial heart development, though we cannot exclude the possibility that it may also be playing a more general role in dorsoventral mesoderm patterning. Fzd7 could be interacting with Wnt11 (Kim et al., 2008; Tao et al., 2005; Witzel et al., 2006) , or another wnt ligand such as Wnt3a (Mazzotta et al., 2016), Wnt6 (Gibb et al., 2013; Lavery et al., 2008a, b) or Wnt8c (Ruiz-Villalba et al., 2016; Schneider and Mercola, 2001) during these stages of development.

As suggested, it is possible that the fzd7 morphant cardiac phenotype is a secondary effect of failures in mesoderm specification, patterning, gastrulation, axis formation and tissue separation. We have made efforts to inject embryos at the 4- and 8-cell stages to give as small a convergent extension phenotype as possible to generate normal-looking embryos but with clear heart phenotypes. The results suggest that the effect of fzd7 during early heart development is not secondary to convergent extension defects or mesoderm development, however, this cannot be ruled out completely (Fig. 3).

An interesting feature of the loss-of-function analysis using fzd7 Morpholino and a dominant-negative fzd7 (fzd7 CRD), is that they give different cardiac phenotypes. fzd7 morphants have anterior defects, convergent extension defects and reduction in nkx2-5 expression; whereas fzd7 CRD-capped RNA injections result in embryos with convergent extension defects and cardia bifida, but no head defects or loss of cardiac markers. Interestingly, it has been shown that the only way to replicate the anterior defect phenotype with a fzd7 CRD construct is to inject the capped RNA into oocytes (Medina et al., 2000). This could be because the relevant signalling event has been completed by time the product of mRNA injected at the 4- or 8-cell stage has been generated. It is possible that if we injected oocytes with fzd7 CRD then we might find embryos showing loss of the heart. Another possibility is that the Morpholino is able to disrupt all Wnt signalling through fzd7 by preventing translation of Fzd7 protein, but fzd7 CRD only disrupts non-canonical signalling in this context. The requirement for co-receptors in canonical signalling may allow the CRD to interact with endogenous fzd7 and any Lrps present allowing the receptor complex aggregates to form. In addition to this, it has been shown to be possible to activate canonical Wnt signalling using CRD constructs (Carron et al., 2003). Perhaps canonical Wnt signalling mediated by fzd7 early on during development is allowed to proceed by the Fzd7 CRD, but then when fzd7 switches to mediate non-canonical signalling, the CRD starts to behave as a dominant-negative. Other possibilities are that the Morpholino may have a broader specificity than thought or that the injected RNA of the fzd7 CRD construct may not be very stable, and thus only provide a short term effect compared to the Morpholino. These possibilities remain to be tested further.

The fzd7 CRD phenotype is very similar to the wnt11-R Morpholino phenotype (Garriock et al., 2005). It has previously been shown that DM-GRASP/alcam expression lies downstream of wnt11-R signalling and that DM-GRASP/alcam can mediate non-canonical wnt signalling effects on morphogenetic movements involved in the developing heart. The DM-GRASP/alcam Morpholino phenotype is also similar to the fzd7 CRD phenotype in that they both lead to a cardia bifida-like phenotype and a thickening of the myocardium. This suggests fzd7 could be mediating the wnt11-R control of DM-GRASP/alcam expression. This needs to be investigated further.

Ruiz-Villalba et al. (2016) suggest a model where periodic switching between proliferation and differentiation within the developing heart is mediated by the periodic and reciprocal activity of the canonical and non-canonical wnt pathways. fzd7 could be playing a crucial role in this process depending upon the Wnts and other receptors expressed at specific times.

In conclusion, we have shown fzd7 to be involved in heart development. Further investigation is required to determine the specific wnt(s) it is interacting with at different stages of heart development.

All experiments were performed in compliance with the relevant laws and institutional guidelines at the University of East Anglia. The research was approved by the local ethical review committee according to UK Home Office regulations. Xenopus laevis embryos were obtained as previously described (Harrison et al., 2004). Staging of the embryos was carried out according to the normal timetable of Nieuwkoop and Faber (Nieuwkoop and Faber, 1994). Embryos at the required stages were fixed in MEMFA, washed in PBS, dehydrated in ascending grades of Methanol/PBS, then stored in 100% MeOH at –20°C until processing for single or double in situ hybridisation.

fzd7 full-length (fzd7 FL) and dominant-negative form fzd7-cysteine rich domain (fzd7 CRD) were sub-cloned into pCS2+ at Cla1–Xho1 restriction sites as described in Wheeler et al. (2000). fzd7 MO titration by RNA in the rescue experiments was avoided by creating a site-directed mutagenesis construct of the full-coding sequence of fzd7 (fzd7 SDM) as described in Abu-Elmagd et al. (2006). fzd3 full-length (fzd3 FL) and fzd3 CRD were kind gifts from Peter Klein (University of Pennsylvania). Dishevelled construct (Dvl1-Delta-N) was a gift from Roberto Mayor (University College, London) (De Calisto et al., 2005).

All capped mRNAs of all genes used for RNA injections were prepared according to the manufacturer's instructions using the SP6 mMessage mMachine Ambion kit (Invitrogen™ AM1340). Anti-sense oligonucleotides, morpholinos (MOs), were obtained and designed by Gene Tools (www.gene-tools.com, Oregon, USA) using the reported sequence for the control morpholino (CMO) (5′-CCTCTTACCTCAgTTACAATTTATA-3′) and fzd7MO (5′-GCGGAGTGAGCAGAAATCGGCTGA-3′) (Sumanas and Ekker, 2001). MOs were diluted, prepared before use according to the manufacturer's instructions and tested using the in vitro translation assay (TNT coupled reticulocyte lysate system, Promega-L4600). For targeting the heart, the DB of the 4- and 8-cell stage embryos were injected as previously described (Lavery et al., 2008a). Capped mRNA and MOs were co-injected with lac-Z for lineage tracing. Each experiment was carried out as an internally controlled group. Each experiment was carried out three times and the number of embryos in each class were pooled.

fzd7 in pBluescript was linearised with XbaI and transcribed by T7; nkx2-5 was linearised with BamH1 and transcribed with T7; troponin-lC (tnnic) was linearised with XhoI and transcribed with T3; gata6 was linearised with XbaI and transcribed with T7. Promega probe synthesis manufacturing instructions were followed with fzd7 probe labelled with Fluorescene-substituted nucleotide (Fl-UTP) and for other heart makers labelled with DIG-substituted nucleotide. Each RNA probe was added to 10 ml hybridisation buffer and stored at –20˚C for in situ hybridisation. Single (Harland, 1991) or double (Knecht et al., 1995) in situ hybridisation was carried out as previously described (Abu-Elmagd et al., 2006). Anti-Fluorescein was detected using Fast Red tablets (Kelloff et al., 2006) while anti-Digoxigenin was detected with NBT/BCIP. Frozen and wax sectioning were carried out as described (Harrison et al., 2004; Hatch et al., 2016). Images were taken using Leica microscope and Axiovision software.

The authors would like to thank Stefan Hoppler, Boni Afouda and Andrea Münsterberg as well as members of the Wheeler and Münsterberg labs for many helpful discussions.