Wnt9 directs zebrafish heart tube assembly via a combination of canonical and non-canonical pathway signaling

During the formation of the nascent zebrafish heart tube, myocardial progenitor cells migrate from bilateral positions within the anterior lateral plate mesoderm towards the embryonic midline. This requires that these cells first migrate as coherent sheets of epithelia towards the embryonic midline. Then myocardial progenitor cells move in angular directions towards endocardial progenitor cells that have already arrived at the embryonic midline. Next, myocardial progenitor cells fuse at posterior positions, followed by a fusion at anterior positions. This generates a disc-shaped structure referred to as the ‘cardiac cone’ (Glickman Holtzman et al., 2007) (Fig. 1A). After remaining at the embryonic midline for several hours, cardiac progenitor cells undergo intricate morphological processes referred to as ‘cardiac tilting and jogging’. This transforms the flat mono-layered heart cone into the nascent heart tube, which extends towards the anterior and left (Bakkers, 2011; Rohr et al., 2008; Smith et al., 2008) (Fig. 1A; Movie 1). The angular movements of myocardial progenitor cells during heart cone formation are impaired in npas4l mutants that lack endocardial cells (Glickman Holtzman et al., 2007). This means that endocardial cells have some role in guiding their angular movements. Yet, it is still unclear which molecular mechanisms underlie this process and how the heart cone transforms into the nascent heart tube.

The growth of an embryo requires an intricate linkage between the positions of cells in tissues and their developmental activities. This is often regulated by morphogens, which are secreted proteins. Morphogens can move over significant distances within tissues to induce distinct responses among target cells in a concentration-dependent manner (Vincent and Briscoe, 2001). This occurs through an activation of signaling cascades and the activation or repression of transcription factors that determine the fate, morphology or behavior of cells. Morphogens can influence positional information and shape cells or tissues by regulating the contractile actomyosin system, adherens junctions or cell polarity (Gilmour et al., 2017). In many cases it is unclear how they shape tissues and organs, and whether cell fate decisions are involved.

Several secreted factors that act as morphogens in cardiac development have been identified, including the Wingless and Int-1 (Wnt) family of proteins (Akieda et al., 2019). The size of the nascent heart is sensitive to Wnt activity, as the overexpression of Wnt8 from a heat-shock transgene at pregastrula stages increases the size of the primary heart field in zebrafish (Ueno et al., 2007). Conversely, at later stages, Wnt8 inhibits the specification of cardiac progenitor cells. This gives Wnt8 biphasic roles, first as an inductive cue in early myocardial development before gastrulation and as a repressive signal afterwards. Overexpression studies during somitogenesis have established Wnt8 as being sufficient for promoting atrial cardiomyocyte differentiation while repressing ventricular cardiomyocyte proliferation (Dohn and Waxman, 2012; Ueno et al., 2007). Which endogenous Wnt ligands orchestrate morphogenetic events during early cardiac development has not been clarified. In vitro, the differentiation of cardiomyocytes from human pluripotent stem cells requires only a few signaling molecules, which include WNTs (Denning et al., 2016). This recapitulates the way cardiogenesis occurs during vertebrate development.

Downstream signaling initiated by Wnt ligands triggers the canonical β-catenin-dependent pathway or two non-canonical pathways called the planar cell polarity (PCP) or Ca²⁺-dependent pathways (Komiya and Habas, 2008). There are many known Wnt ligands, and these can activate exclusively only one of the pathways or trigger several (van Amerongen and Nusse, 2009), in a range of developmental contexts. Wnt11 activates PCP signaling during zebrafish cardiogenesis and a loss of either non-canonical pathways (PCP and calcium-dependent signaling) disrupts the migration of anterior endodermal progenitor cells towards the embryonic midline. Migrations of cardiac progenitor cells are also affected and result in cardia bifida, which is characterized by a bilateral localization of heart primordia (Matsui et al., 2005). PCP signaling positively regulates, via c-Jun N-terminal kinase (Jnk), the basic helix-loop-helix transcription factor Hand2, which plays an essential role in cardiac morphogenesis (Santos-Ledo et al., 2020). wnt11f2 (previously known as wnt11/silberblick) and wnt5b trigger PCP signaling during cardiac looping, which promotes the remodeling of the linear heart tube by polarizing actomyosin contractile networks (Merks et al., 2018).

Wnt9 proteins are ligands that have been implicated in both canonical and non-canonical signaling (Carroll et al., 2005; Goddard et al., 2017; Karner et al., 2009; Montcouquiol et al., 2006). During early stages of cardiac valvulogenesis in mice and zebrafish, Wnt9b/wnt9b expression is activated by the zinc-finger transcription factor Krüppel-like factor 2 (KLF2/Klf2), which has been induced by hemodynamic forces because of blood flow (Goddard et al., 2017). In mice, Wnt9b acts via the canonical pathway to block the proliferation of mesenchymal cells and mediate their condensation, which is required for the final steps of valve maturation (Goddard et al., 2017). A similar circuit is found in zebrafish, where blood flow also affects heart development via a Wnt9-dependent mechanism. Within the nascent zebrafish heart tube, the onset of blood flow induces the expression of klf2a/b, which in turn triggers the expression of two wnt9a/b paralogous genes within the endocardium. This Klf2-Wnt9 axis of mechanosensitive signaling is required for zebrafish cardiac valvulogenesis (Paolini et al., 2021).

Understanding the morphogenetic roles of specific Wnt signals during cardiac development has been difficult because multiple ligands and receptors are expressed at these stages (Ruiz-Villalba et al., 2016). In previous work, we demonstrated that one ligand, Wnt9a, plays a role in zebrafish cardiac valvulogenesis (Paolini et al., 2021). That study also uncovered potentially redundant and crucial roles of Wnt9a and its paralog, Wnt9b, during cardiac tube formation that were not further characterized.

Here, we demonstrate that Wnt9a/b play multiple roles during the assembly of the nascent heart cone and its transition towards the elongating heart tube. Our genetic studies show that the loss of both Wnt9 paralogs abolishes the completion of heart cone formation and leftward jogging of the heart tube. The simultaneous pharmacological inhibition of both canonical and non-canonical Wnt signaling pathways reproduces the wnt9a/b loss-of-function cardiac phenotype. This suggests that Wnt9a/b controls cardiac morphogenesis via divergent downstream signaling pathways. We also perform a broad misexpression experiment based on a heat-shock-inducible wnt9b transgene, which becomes expressed in all cell types. This reveals that Wnt9 is permissive but not instructive in guiding these early cardiac developmental processes. Finally, we provide evidence that Wnt9a/b prevent the premature differentiation of cardiac progenitor cells. These findings demonstrate that Wnt9 morphogens direct cardiogenesis through complex, interlinked signaling mechanisms.

To characterize the roles of Wnt9a and Wnt9b during cardiac tube formation, we generated wnt9a^sd49 (Grainger et al., 2016) and wnt9b^sa20083 (Sanger Institute Zebrafish Mutation Project Mutant Data Submission) double mutant embryos (hereafter referred to as wnt9^DKO mutants) combined with a transgenic line harboring the myocardium-specific reporter Tg(myl7:EGFP)^twu34 (Huang et al., 2003). Strikingly, at 48 h postfertilization (hpf), wnt9^DKO mutant hearts exhibited a collapsed form that was positioned at the embryonic midline. These hearts had apparently failed to undergo cardiac tilting and jogging, in contrast to wild-type hearts, and become arrested during early stages of heart morphogenesis (Fig. 1B,C; Movie 2). In all wnt9^DKO mutant hearts, cardiac progenitor cells had failed to fuse in anterior positions (Fig. 1C, arrowhead; Movie 3) (n=10 mutants analyzed). This phenotype was similar to that seen in npas4l mutants, which lack the endocardium (Glickman Holtzman et al., 2007). The same phenotype occurred in wnt9b^{sa20083/sa20083} homozygous; wnt9a^sd49/+ heterozygous mutants (Fig. 1D) (n=24/24 embryos analyzed). But wnt9a^sd49/sd49 homozygous; wnt9b^sa20083/+ heterozygous mutants had normally elongated hearts (Fig. 1E) (n=23/23 embryos analyzed). This genetic analysis showed that one wild-type allele of wnt9b is sufficient to rescue the cardiac phenotype, whereas only one wild-type allele of wnt9a does not. In contrast, none of the single mutants exhibited an early cardiac defect (Fig. S1A,B) (n=10/10 wnt9a^sd49/sd49 mutants analyzed; n=11/11 wnt9b^{sa20083/sa20083} mutants analyzed).

To further verify these findings, we used an antisense oligonucleotide morpholino (MO) knockdown approach against Wnt9a/b. This produced a cardiac phenotype identical to that seen in wnt9^DKO mutants at 48 hpf (Fig. S2). At 20 hpf, wnt9a/b double morphant heart cones failed to fuse anteriorly (n=15/15 embryos analyzed); consequently, they did not undergo cardiac jogging at 24 hpf (n=12/12 embryos analyzed). We applied counter-labeling with an antibody against the atrial cardiomyocyte-specific marker Myh6 (Yelon et al., 1999) at 48 hpf. This showed that in wnt9a/b double mutants, the collapsed myocardial ventricle was surrounded by atrial tissue in a cardiac cone-like arrangement (Movie 3). This demonstrated that the two wnt9a/b paralogous genes have essential and partially redundant roles during cardiac cone formation and leftward jogging and that Wnt9b plays a particularly important role in these processes.

Next, we characterized the morphologies of myocardial and endocardial progenitor cells at earlier stages (20-24 hpf). We used a MO knockdown against Wnt9a/b in different genetic backgrounds. First, we used the reporter transgene Tg(nkx2.5:EGFP)^el83, which marks myocardial progenitor cells within the anterior lateral plate mesoderm and during their migration towards the embryonic midline (Witzel et al., 2012), which made it possible to visualize these cells at such an early stage. Another experiment involved the endothelium/endocardium-specific reporter transgene Tg(etv2:GFP)^ci1 (Proulx et al., 2010). We also performed whole-mount in situ hybridizations of wild-type and wnt9^DKO mutants at 20, 24 and 36 hpf with probes against the chamber-specific genes ventricular myosin heavy chain (vmhc; also known as myh7) and atrial myosin heavy chain (amhc; myh6) (Yelon et al., 1999) (Fig. S3). At 20 hpf, wild-type myocardial and endocardial progenitors had fused into a heart cone at the embryonic midline (Fig. 1F,J,N; Fig. S3A,C). At this stage, both wnt9a/b double morphants (Fig. 1G; Fig. S2B,D) and wnt9^DKO mutants exhibited myocardial progenitor cells arranged in two bilateral wings, marked by either EGFP or vmhc expression, that had not fused anteriorly (Fig. 1O) (n=32/32 morphants analyzed; n=7/7 mutants analyzed). This did not affect etv2:EGFP-positive endocardial progenitor cells, which originate in positions of the anterior lateral plate mesoderm and move towards the embryonic midline before the arrival of myocardial progenitor cells (Bussmann et al., 2007). These had already reached the embryonic midline in wnt9a/b double morphants, as seen in wild-type embryos (Fig. 1K; n=6/6 morphants analyzed).

By 24 hpf, wild-type myocardial and endocardial progenitors had formed an elongated heart tube that was jogging leftwards and towards the anterior (Fig. 1H,L,P; Fig. S3E,G). This starkly contrasted with the situation in wnt9a/b double morphants (Fig. 1I; Fig. S3F,H) and wnt9^DKO mutants, where myocardial progenitor cells failed to complete the closure of the heart cone and remained at the embryonic midline (Fig. 1Q) (n=28/28 morphants analyzed; n=7/7 mutants analyzed). At this stage in wnt9a/b double morphants, etv2:EGFP-positive endocardial progenitor cells had initiated some leftward movements but failed to complete normal leftward jogging (Fig. 1M, white arrowhead; n=7/7 morphants analyzed). At 36 hpf, vmhc-expressing cells remained at the embryonic midline in these morphants, whereas in wild-type embryos the heart tube had extended towards the left (Fig. S3I,J). amhc had a similar expression: In wnt9a/b double morphants at 36 hpf, amhc expression had expanded symmetrically around the embryonic midline without the leftward- or anteriorly-directed displacement characteristic of wild-type embryos (Fig. S3K,L) (n=19/19 morphants analyzed).

wnt9^DKO mutants exhibited impairments in heart tube formation and cardiac jogging, but the cardiac phenotypes differed strikingly from those observed in zebrafish mutants with defective left/right asymmetry signaling. In mutants lacking the nodal ligand Southpaw, the heart tube elongates along the embryonic midline in the anterior direction (Long et al., 2003). Next, we assessed whether left/right asymmetry signaling was affected in wnt9^DKO mutants. We performed whole-mount in situ hybridizations at 20 hpf against the Nodal pathway gene lefty1, which is normally expressed at the embryonic midline and in a band to the left side of the heart field (Long et al., 2003). The expression of lefty1 was similar in wild-type and wnt9^DKO mutants (Fig. S4; n=20 embryos analyzed). This indicates that the paralogous genes wnt9a and wnt9b orchestrate early myocardial morphogenesis through effects on endocardial and cardiomyocyte progenitor cells that are independent of the signaling pathways that regulate left/right asymmetry (Fig. 1R).

Next, we characterized the morphogenetic dynamics of cardiac cone formation in wnt9a/b double morphants. Using the cardiac progenitor cell reporter transgene Tg(nkx2.5:EGFP)^el83, we carried out longitudinal light-sheet microscopy observations in vivo at 17-22 hpf. In the wild-type, cardiomyocyte progenitors migrated from positions in the lateral plate mesoderm towards the embryonic midline where the cardiac cone fused. The cardiac cone first fused in posterior positions before fusing anteriorly (Fig. 2A; Movie 4). The cone also fused at the posterior in wnt9a/b double morphants but failed to complete anterior fusion (Fig. 2B; Movie 5). For a more detailed analysis of the migratory behavior of cardiomyocytes, we used the longitudinal movie files to reconstruct individual cell tracks (Fig. 2A,B; Fig. S5; Movies 4, 5). Here we assessed both cell velocities and the straightness index, which is used to measure the direction of migration paths (whereby 1 equals a straight line). Our analysis revealed that the loss of Wnt9a/b reduced the velocity of cells and disrupted the straightness of their paths (Fig. 2C,D; Fig. S5; Movies 4,5; n=5 each of wnt9a/b double morphant and wild-type hearts analyzed). This demonstrates that Wnt9a/b play an important role in the migrations of cardiac progenitor cells during cardiac cone formation and jogging.

These results prompted us to check the expression pattern of wnt9b in the wild-type. We did so using whole-mount in situ hybridizations at 16 hpf. At this stage, cardiac progenitor cells in bilateral positions of the anterior lateral plate mesoderm initiate migrations towards the embryonic midline (Bakkers, 2011; Glickman Holtzman et al., 2007). We found that the anterior lateral plate mesoderm was devoid of wnt9b expression at this stage (Fig. 2E,E′, see arrows; Movie 6; n=10 embryos analyzed). It was expressed, however, in the anterior and ventral sides of the neural tube (Fig. 2E,E′; Movie 6). At this stage, the anterior neural tube is made up mostly of neural progenitors, as indicated by a weak expression of the post-mitotic neuronal marker transgene Tg(elavl3:EGFP)^knu3 (Park et al., 2000). wnt9b was never expressed in cells that have developed to the point that they express this reporter transgene (Fig. 2F,F′; Movies 7,8; n=10 embryos analyzed). It is not produced in the population of postmitotic neurons located at the anterior/ventral side of the neural tube. Instead, wnt9b is expressed by neural progenitors at the ventral midline of the neural tube at the time of cardiac cone formation.

The finding that wnt9b is not expressed within cardiac progenitor cells, but rather in neighboring cells at the ventral midline of the neural tube, suggested that the signal plays an instructive role during cardiac cone formation. To test this possibility, we generated the transgenic line Tg(hsp70l:wnt9b_IRES_EGFP)^pbb48, which causes an embryo-wide overexpression of wnt9b (Fig. S6A) and causes an activation of the canonical Wnt signaling reporter Tg(7xTCF-Xla.Siam:nlsmCherry)^ia5 throughout the heart (Fig. S6B,C). When heat-shocked at 3.5 hpf, Tg(hsp70l:wnt9b_IRES_EGFP)^pbb48 embryos developed with a dorsalized phenotype (Fig. S6D,E). If Wnt9b played an instructive role, misexpression ought to disrupt the unidirectional migration patterns of myocardial progenitor cells. The application of a heat shock at 14 hpf (30 mins at 37°C) led to an increase in the size of the cardiac cone in wnt9b-overexpressing embryos at 20 hpf compared with the wild-type (Fig. 2G,H; n=22 embryos analyzed). The increase did not, however, impair either heart cone formation or the leftward-directed jogging of the nascent heart tube at 24 hpf. Myocardial progenitor cells had moved to the left side from the midline, but they were more widely dispersed than in the wild-type (Fig. 2I,J; n=13 Wnt9b-overexpressing embryos analyzed). As a result, Wnt9b is not instructive but is rather permissive during heart cone formation and cardiac jogging.

Canonical Wnt signaling in the pre-cardiac mesoderm is necessary and sufficient to inhibit cardiac progenitor cell differentiation (Dohn and Waxman, 2012; Ueno et al., 2007). We confirmed this when observing an increase in cardiomyocyte numbers in wnt9^DKO mutants at 20 hpf compared with the wild-type (Fig. 3A-C; n=9 wild-type and n=11 wnt9^DKO mutant embryos analyzed). However, the severe morphogenetic changes observed in wnt9^DKO mutants could not be explained only by a lack of canonical Wnt activity. This can be seen from embryos that overexpress the Wnt antagonist Dkk1; upon a loss of canonical Wnt signaling after gastrulation, zebrafish still undergo heart cone formation and leftward cardiac jogging (Dohn and Waxman, 2012; Ueno et al., 2007). In wnt11f2 (Matsui et al., 2005) or wnt11f1 (previously known as wnt11r) (Choudhry and Trede, 2013) mutants, which inhibit non-canonical Wnt signaling, zebrafish still form a leftward extending heart tube. This suggested that Wnt9a/b signaling triggers the activation of a combination of canonical and non-canonical Wnt pathways during zebrafish cardiac morphogenesis.

To test this hypothesis, we treated zebrafish embryos with IWR-1, an inhibitor of the canonical Wnt pathway (Lu et al., 2009) in combination with TNP-470, which inhibits the non-canonical Wnt pathway (Zhang et al., 2006). IWR-1 abolishes the turnover of the Axin protein; this strongly decreases levels of β-catenin, a central component of canonical Wnt signaling (Chen et al., 2009). TNP-470 targets Methionine aminopeptidase 2, which acts downstream of Frizzled receptors and upstream of the non-canonical Wnt pathway components Dishevelled, JNK and CaMKII (Zhang et al., 2006). Treatment with both small compound inhibitors from 14 to 20 hpf prevented the anterior fusion of the cardiac cone (Fig. 3D,E; n=21 embryos analyzed). This resembled the phenotype of wnt9^DKO mutants (Fig. 1O). In comparison, separate treatments with single inhibitors did not produce an early cardiac cone phenotype (Fig. 3F,G; n=11 embryos analyzed). These findings suggest that signaling via Wnt9a/b leads to the activation of both canonical and non-canonical Wnt signaling pathways during cardiac cone formation and jogging, through a complex regulation of several downstream Wnt components.

Apicobasal polarity is crucial to heart morphogenesis. By the 15-somite stage, cardiomyocyte progenitor cells have formed polarized epithelial sheets that migrate towards the embryonic midline (Rohr et al., 2006). Loss of either Protein kinase C iota (Prkci) or MAGUK p55 subfamily member 5 (Mpp5; Pals1a), proteins which establish apicobasal polarity, disrupts epithelial organization, and cardiac morphogenesis stalls at the heart cone stage (Horne-Badovinac et al., 2001; Rohr et al., 2006). This produces defects in cardiac cone formation and leftward jogging that strongly resemble wnt9^DKO mutant cardiac phenotypes (Fig. 1O) (Horne-Badovinac et al., 2001; Rohr et al., 2006). Conceivably this situation in wnt9^DKO mutants might be due to a disruption of the apico-basal polarity of cardiomyocyte progenitors. To determine whether this was the case, we analyzed the expression of the protein Zonula occludens-1 (ZO-1; Tjp1a), which is a marker of apical tight junctions (Fig. 4A-B′; n=9 wild-type and n=12 mutant embryos analyzed). The subcellular distribution of ZO-1 was similar in wild-type and wnt9^DKO mutants (Fig. 4A″,B″, see arrowheads). This suggested that the loss of the two wnt9a/b paralogous genes did not affect the apico-basal polarity of cardiomyocytes.

Genetic studies have shown that the migration of myocardial progenitor cells toward the midline and the formation of the cardiac cone depends on the presence of endoderm. In zebrafish mutants lacking endoderm, cardiac progenitor cells fail to migrate toward the embryonic midline, which results in cardia bifida (Dickmeis et al., 2001; Fukui et al., 2014). This prompted us to assess the organization of endodermal progenitor cells in wnt9a/b double morphants. We did this using the transgenic endodermal reporter line Tg(sox17:GFP)^s870 (Field et al., 2003), and counter-labeled myocardial tissue with an antibody against Activated leukocyte cell adhesion molecule (Alcam) (Beis et al., 2005) (Fig. 4C-D″). This revealed that at 20 hpf, endodermal tissue at the embryonic midline was similar in the double morphants and wild-type embryos (Fig. 4C-D″; n=15/15 morphants analyzed).

The extracellular matrix glycoprotein Fibronectin (Fn) is deposited between the endoderm and cardiac precursors at the embryonic midline and it is essential for the timely migration of cardiomyocyte progenitors (Trinh and Stainier, 2004). To assess whether the absence of the Wnt9 paralogs alters Fn deposition, we assessed Fn expression at heart cone stages in wnt9^DKO in comparison with wild-type embryos. High-resolution x-z scan projections through different section planes of the heart cone revealed no qualitative differences in Fn deposition between these conditions (Fig. S7; n=7/7 mutants analyzed). Therefore, Fn deposition is not directly affected or causative to the wnt9^DKO cardiac phenotype. These findings suggested that Wnt9a/b control myocardial progenitor cell migrations and heart cone assembly while not disrupting endodermal midline migration and fusion. Furthermore, the loss of these proteins affected myocardial progenitor cell migrations without causing any obvious defects in epithelial apico-basal polarity or Fn deposition.

In endocardial cells, wnt9a/b are induced by blood flow and Klf2, and their expression is required for cardiac valvulogenesis (Goddard et al., 2017; Paolini et al., 2021). This raised the question of whether the activity of Wnt9a/b during heart cone formation requires Klf2. To determine this, we examined whether klf2a/b double morphants affected the development of the heart cone or cardiac jogging. At 20 hpf, the heart cone of klf2a/b double morphants was fused at the embryonic midline (Fig. 5C; n=10/10 embryos analyzed) as in the wild-type (Fig. 5A). By 24 hpf, the heart tube was elongating towards the left (Fig. 5D; n=9/9 embryos analyzed), again resembling the wild-type (Fig. 5B). This confirms the finding that klf2a/b double mutants develop an elongated heart tube (Fontana et al., 2020) and distinguishes them from wnt9^DKO mutants at the heart cone and jogging stages (Fig. 1G,I). At this point, the production of Wnt9a/b does not appear to be contingent upon the activity of Klf2a/b.

Conceivably, Wnt9a/b activity might depend on the endocardium itself, which fails to form in early zebrafish embryos lacking Npas4l (Reischauer et al., 2016) However, the phenotype of npas4l zebrafish is quite different from those lacking Wnt9a/b (Fig. 1G,I) (Reischauer et al., 2016; Glickman Holtzman et al., 2007) and heart cone formation proceeds normally at 20 hpf (Fig. 5E; n=8/8 embryos analyzed), as does cardiac jogging towards the left at 24 hpf (Fig. 5F; n=10/10 embryos analyzed). To further establish whether Wnt9a/b signaling for cardiac progenitor cell migration involved the endocardium, we performed a genetic epistasis experiment by knocking down Wnt9a/b in npas4l^m378 mutants. Although npas4l^m378 mutants had an enlarged myocardial heart tube at 48 hpf (Fig. 5H) compared with the wild-type heart (Fig. 5G), the myocardium of npas4l^m378 mutant; wnt9a/b double morphants was collapsed (Fig. 5J; n=30/30 embryos analyzed). This resembled the phenotype of wnt9a/b double mutants (Fig. 1C) and morphants (Fig. 5I). This shows that the loss of Wnt9a/b is epistatic over the npas4l mutant phenotype, and thus rules out an involvement of the endocardium in wnt9a/b-mediated signaling during cardiac cone formation.

Another way Wnt9 might affect cardiac morphogenesis is by forcing the premature differentiation of myocardial progenitor cells. Canonical Wnt signaling has a role in inhibiting cardiac progenitor differentiation (Dohn and Waxman, 2012; Ueno et al., 2007). This is in line with our observation that wnt9^DKO mutants had more cardiomyocyte progenitor cells (Fig. 3A-C). The differentiation state of these cells could affect their motility and alter the development of the tissue. Next, we determined whether Wnt9b affected the differentiation states of myocardial progenitors. To do this, we examined levels of the expression of developmental marker genes in myocardial cells, comparing Tg(hsp70l:wnt9b_IRES_EGFP)^pbb48 transgenic embryos, which overexpressed Wnt9b, with their wild-type heat-shocked siblings. Upon heat-shock at 14 hpf, the differentiation of myocardial progenitor cells at 20 hpf was impaired (Fig. 5K). This is in line with an established role of canonical Wnt signaling in preventing excessive cardiac differentiation during stages of heart cone formation.

How morphogen signaling controls tissue and organ shape is a major problem of developmental biology. Secreted Wnt family molecules act as morphogens in many developmental and pathological processes. Wnt9a/b contribute to cardiac development when they are induced by blood flow and have essential roles during valvulogenesis (Goddard et al., 2017; Paolini et al., 2021). Loss of wnt9a results in a complete lack of valve leaflets in zebrafish (Paolini et al., 2021) and loss of Wnt9b causes an enlarged and malfunctioning atrioventricular valve in mice (Goddard et al., 2017). Here, we demonstrate that the wnt9a/b paralogous genes play an equally crucial role earlier, during the first stages of heart development in zebrafish.

Our findings demonstrate that Wnt9-mediated signaling is necessary for the formation of the cardiac cone of the nascent zebrafish heart and requires signaling via both canonical and non-canonical pathways (Fig. 5L). We made a closer investigation to determine the extent to which the effects of the two ligands and the pathways can be distinguished. We find that these two branches of Wnt signaling have partially redundant effects on myocardial progenitor cells that were uncovered only through an examination of wnt9^DKO mutants or by pharmacologically inhibiting both pathways. We show that a loss of Wnt9a/b causes the premature differentiation of myocardial progenitor cells and that their migration toward the embryonic midline is disrupted. These two phenotypes might be connected as differentiated cells tend to lose their motile capacity. This is the case during zebrafish cardiac regeneration, when differentiated cardiomyocytes first undergo dedifferentiation, then become motile and migrate to injured regions of the heart (Itou et al., 2012; Kikuchi et al., 2010).

Inhibiting only canonical Wnt pathway signaling is not sufficient to explain the defective cardiac progenitor cell migration in wnt9^DKO mutants. This suggests that non-canonical Wnt signaling must also be involved in orchestrating convergence movements of anterior myocardial progenitor cells towards the embryonic midline. This could be because non-canonical signaling affects the cytoskeletal organization of anterior myocardial progenitor cells during their migration towards the midline, but this will require further investigation. Another possibility is that the defects observed in wnt9^DKO mutants arise because Wnt9 affects the apico-basal polarity of cardiomyocyte progenitor cells in some way that has escaped detection so far.

These observations suggest that canonical and non-canonical Wnt pathways cooperate to direct the convergence of myocardial progenitor cells towards the embryonic midline. In contrast, the migration of endocardial progenitor cells is not affected in wnt9^DKO mutants. This suggests that Wnt9a/b signaling exerts a specific control on myocardial progenitor cell migrations at a crucial stage in heart development.

The defects in migration that we observe in wnt9^DKO mutants may be due to defects in the organization of the actin cytoskeleton. The effectors of this system are controlled by Frizzled/PCP signaling, activated by non-canonical Wnt ligands (Gray et al., 2011). During sprouting angiogenesis, low levels of non-canonical Wnt signaling disrupt collective migrations of endothelial cells. This is caused by a weakening of adherens junctions, which couples the force transmission from neighboring cells or the extracellular matrix towards the actin cytoskeleton (Carvalho et al., 2019). Hence, a disruption of cytoskeletal organization could slow down myocardial progenitor cells during their migration towards the midline. It remains unclear why the lack of the Wnt9 paralogs specifically affects the migration of myocardial progenitors at the anterior of the heart cone. Future work will determine whether cells at the anterior express distinct Wnt9 receptors.

Our genetic studies demonstrate that wnt9b plays a particularly important role during early cardiac morphogenesis. At heart cone stages, its expression is absent from the lateral plate mesoderm, but present within the ventral neural tube. This suggests that during heart cone assembly, wnt9b is produced in the ventral part of the neural tube and spreads towards the heart field, affecting the dynamics of myocardial progenitors in a paracrine manner. In other contexts, Wnt ligands play a role in coordinating the development of neural and vascular tissues. For example, in the developing nervous system, Wnt7a/b coordinate neural tube and retina vascularization (Paredes et al., 2018; Peguera et al., 2021). Neural tissues also secrete Wnt7a/b to activate Wnt signaling in endothelial cells during the vascularization of the spinal cord (Daneman et al., 2009). In vitro studies have shown that the development of cardiomyocytes is influenced by neural tissue, as cardiomyocytes and neurons develop concurrently from embryonic stem cells, suggesting bidirectional communication between the two cell types (Murashov et al., 2005).

Single genes frequently play roles in the development of several traits and contribute to diverse pathologies. Here, we show how wnt9a and, in particular, wnt9b contribute to heart morphogenesis before the onset of blood flow. This complements earlier studies, which demonstrated that Wnt9a and Wnt9b act later, in a flow-dependent manner, to sculpt heart valves (Goddard et al., 2017; Paolini et al., 2021). It is known that pleiotropic congenital heart defects (CHD) may arise from the same gene defect (Paaby and Rockman, 2013). These studies hint that Wnt9a/b and blood flow may contribute to CHDs. Further studies are needed to determine whether mutations in human WNT9A/B paralogous genes are associated with CHDs.

Handling of zebrafish was carried out according to FELASA guidelines and in compliance with German and Brandenburg state law, carefully monitored by the local authority for animal protection (LAGV, Brandenburg, Germany; Animal protocol #2347-18-2015). The following strains of zebrafish were maintained under standard conditions as previously described (Westerfield et al., 1997): wnt9a^sd49 (Grainger et al., 2016), wnt9b^sa20083 (Sanger Institute Zebrafish Mutation Project Mutant Data Submission), npas4l^m378 (Reischauer et al., 2016), Tg(myl7:EGFP)^twu34 (Huang et al., 2003), Tg(nkx2.5:EGFP)^el83 (Witzel et al., 2012), Tg(etv2:GFP)^ci1 (Proulx et al., 2010), Tg(elavl3:EGFP)^knu3 (Park et al., 2000), Tg(hsp70l:wnt9b_IRES_EGFP)^pbb48 (this study), Tg(sox17:GFP)^s870 (Field et al., 2003), Tg(7xTCF-Xla.Siam:nlsmCherry)^ia5 (Moro et al., 2012). The developmental stage of the embryos used is indicated for each experiment in the results and figure legends.

The following MOs were injected into the yolk at the one-cell stage in 1 nl total volume: wnt9a (5′-AAGAATTGTCCTGCCTACCCGAAGT-3′) (1 ng/embryo) (Paolini et al., 2021), wnt9b (5′-ACCTGTAAGCCTAACGAAAACACAA-3′) (1 ng/embryo) (Jackson et al., 2015), klf2a (5′-CTCGCCTATGAAAGAAGAGAGGATT-3′) (1 ng/embryo) (Nicoli et al., 2010), klf2b (5′-AAAGGCAAGGTAAAGCCATGTCCAC-3′) (5 ng/embryo) (Renz et al., 2015), cloche/npas4l (5′-GAGTCTCCGCAGCTCATCTCACA-3′) (1 ng/embryo) (Reischauer et al., 2016). Control embryos (WT) were always injected with 1 nl of water.

The open reading frame of zebrafish wnt9b (ENSDARG00000037889) was amplified by PCR with AK3 and AK4 primers (Table 1) and cloned into the Gateway pDONR221 vector via Gateway BP cloning (Thermo Fisher Scientific) to generate pME-Wnt9b. The pDest_Tol2pA_hsp70l:wnt9b_IRES_EGFP vector was created by Gateway LR cloning (Thermo Fisher Scientific) from pDest_Tol2pA, p5E-hsp70l, pME-Wnt9b and p3E-IRES-eGFPpA. We injected 15 pg of plasmid into one-cell-stage embryos with 15 pg of Tol2 transposase mRNA.

wnt9b^sa20083 mutants were genotyped through PCR amplification of the coding region containing the point mutation (with Wnt9bgeno3_FW and Wnt9bgeno3_RV primers, see Table 1) and enzymatic digestion of the amplified fragment (234 bp) was carried out using NaeI (New England Biolabs). The mutated fragment does not contain the restriction site and was not cut.

To overexpress wnt9b during the stages of heart cone fusion and leftward jogging, transgenic fish carrying Tg(hsp70l:wnt9b_IRES_EGFP)^pbb48 were crossed to Tg(myl7:EGFP)^twu34 transgenic fish and the resulting embryos were heat-shocked at 3.5 or 14 hpf (30 min at 37°C) and screened for EGFP expression. To inhibit canonical Wnt signaling, embryos were treated with 20 µM IWR-1 (Selleckchem) between 15 and 20 hpf. To inhibit non-canonical Wnt signaling, embryos were treated with 50 µM TNP-470 (Sigma-Aldrich) between 15 and 20 hpf. Control embryos were always treated with an equivalent amount of DMSO. To inhibit pigmentation, embryos were treated with 1-phenyl-2-thiourea (PTU) (Sigma-Aldrich) after 24 hpf.

For RT-qPCR experiments, 25 heat-shocked EGFP-positive Tg(hsp70l:wnt9b_IRES_EGFP)^pbb48 embryos were pooled. Similarly, 25 heat-shocked EGFP-negative siblings (three biological replicates) were pooled. Total RNA was extracted using Trizol (Thermo Fisher Scientific) and Phase Lock Gel Heavy tubes (1.5 ml, 5 Prime) and the corresponding cDNA was synthesized from total RNA with the RevertAid H Minus First Strand cDNA Synthesis kit (Thermo Fisher Scientific). RT-qPCR experiments were performed as previously described (Renz et al., 2015) using 9 ng cDNA per technical replicate and the KAPA Sybr Fast qPCR kit (Roche) on an Analytic-Jena qTOWER 3 (Analytic Jena, 844-00553-2). Cycle threshold (Ct) values were determined by Analytic Jena qPCRsoft (Analytic Jena, version 4.1.3.0). eif1b was used as a housekeeping gene for normalization. wnt9a/b morphant sample values were normalized to 1, using the comparative threshold cycle method (2-DDCT) (Livak and Schmittgen, 2001). As every single biological replicate represents an independent experiment from an independent clutch of embryos, two-tailed, paired t-tests were performed using GraphPad Prism (version 9). For the experiment in Fig. 5K, the mean values of the fold changes are indicated in Table 2. The mean values of fold changes for Fig. S6A are indicated in Table 5. All primers are listed in Table 1.

Zebrafish embryos were collected at 20 hpf or 24 hpf and fixed with 4% paraformaldehyde overnight at 4°C. The vmhc, amhc, lefty1 and wnt9b antisense in situ probes were generated by PCR amplification from 24 hpf WT cDNA, with the reverse primers containing a T7 promoter overhang sequence (primers listed in Table 1). Antisense RNA was synthesized using the DIG RNA Labeling kit (Roche) and T7 polymerase. Whole-mount in situ hybridization was performed as previously described (Jowett and Lettice, 1994). For fluorescence in situ hybridization experiments, embryos were treated with 2% H₂O₂ in PBS for 10 min to quench endogenous peroxidase. For signal detection, embryos were incubated in 1:300 cyanine 3 amplification reagent (PerkinElmer) diluted in 1× plus amplification diluent (PerkinElmer). Brightfield images were recorded with 20× objectives on an Axioskop (Zeiss) with an EOS 5 D Mark III (Canon) camera and processed using Adobe Camera Raw and Adobe Photoshop (Adobe Creative Cloud). Fluorescence images were recorded on an LSM 880 confocal microscope (Zeiss) and processed with Fiji (Schindelin et al., 2012) in order to minimize the background and highlight the signal.

Whole-mount immunohistochemistry was performed on 20, 24 and 48 hpf embryos. The following primary antibodies were used: mouse anti-Zn-8/Alcam (1:25, Developmental Studies Hybridoma Bank, zn-8, AB_531904), mouse anti-Myh6 (1:10, Developmental Studies Hybridoma Bank, AB_528376), mouse anti-ZO-1 (1:200, Thermo Fisher Scientific, 33-9100), rabbit anti-Fibronectin (1:100, Sigma-Aldrich, F3648). The secondary antibodies used were Alexa Fluor 633-conjugated goat anti-mouse (1:200, Thermo Fisher Scientific, A-21052), HRP-conjugated goat anti-rabbit (1:1, Thermo Fisher Scientific, from Tyramide SuperBoost kit, B40958). Embryos were fixed with 4% paraformaldehyde overnight. For Zn-8/Alcam staining, embryos were permeabilized with ice-cold acetone for 10-15 min at −20°C. After 1 h of incubation in blocking solution [PBST (PBS with 0.1% Tween20) with 1% DMSO and 5% normal goat serum (NGS)], embryos were incubated with primary antibody diluted in PBST with 1% DMSO and 1% NGS. For Myh6 staining, embryos were incubated for 2 h in blocking solution [PBST with 10% NGS, 2 mg/ml bovine serum albumin (BSA) and 0.2% saponin] and then with primary antibody diluted in PBST with 0.2% saponin. For ZO-1 staining, embryos were incubated for 2 h in blocking solution (PBST with 10% NGS, 2 mg/ml BSA and 0.8% Triton X-100) and then with the primary antibody in PBST and 0.8% Triton X-100. For Fn staining, embryos were incubated with 0.5% Triton X-100 in PBST and then 20 mins in 3% H₂O₂, a step required for signal amplification. After that, embryos were incubated for 2 h in a blocking solution (PBST with 3% BSA and 5% NGS) and then with a primary antibody. For signal amplification, embryos were incubated with HRP-conjugated goat anti-rabbit and then in a staining buffer containing Alexa Fluor 647 (Tyramide SuperBoost Kit, Thermo Fisher Scientific). In Figs 1G-J, 2E-F′ and 3A,B, nuclei were visualized using DAPI (4,6-diamidino-2-phenylindole; Sigma-Aldrich) (1:1000) staining. To image the cardiac cone at 20 hpf and the heart tube during leftward jogging, embryos were manually deyolked, flattened and mounted in SlowFade Gold (Thermo Fisher Scientific) for imaging. All 48 hpf embryonic hearts were manually extracted and mounted in SlowFade Gold. Images were recorded on an LSM 880 confocal microscope (Zeiss) and processed with Fiji in order to minimize the background and highlight the signal. For each experiment, the same imaging settings were used. Fiji software was also used to obtain the x-z orthogonal view of the cardiac cone at 20 hpf as shown in Fig. 4A″,B″.

Live imaging was performed using a light-sheet Z.1 microscope (Carl Zeiss) equipped with a water immersion 20× detection objective lens (W Plan Apochromat, NA 1.0), dual-sided 10× illumination objective lenses (LSMF, NA 0.2), a pco.edge scientific CMOS camera (PCO) and ZEN software. Embryos were manually dechorionated, anesthetized in 0.16 mg/ml Tricaine in egg water (60 µg/ml sea salt) (this solution was also used to fill the light-sheet chamber during imaging), transferred into 1% low-melting agarose (Lonza 50081) containing 0.16 mg/ml Tricaine in egg water and withdrawn into glass capillaries using a metal plunger. Embryos were positioned with the dorsal side facing the outer wall of the agarose cylinder. For the time-lapse datasets of the heart field during leftward jogging (Movies 1, 2), z-stacks encompassing the entire heart field (based on GFP fluorescence), a z-interval of 1 µm, exposure of 30-50 msec, and a time interval of 5 min for 100 cycles for wild-type and 10 min for 100 cycles for wnt9^DKO were used. For the time-lapse datasets of the heart field during cardiomyocyte progenitor cell migrations towards the embryonic midline (Movies 4, 5), z-stacks encompassing the entire heart field (based on GFP fluorescence), a z-interval of 1 µm, exposure of 30-50 msec and a time interval of 2.5 min for 120 cycles were used.

The tracking spot function of Imaris software (Oxford Instruments) was used for cell tracking analyses shown in Fig. 2A-D. The estimated diameter of each object/cell was set at 7.5 µm. Only cardiomyocyte progenitor cells in anterior positions, which were visible for at least five consecutive time points, were considered for analyses. These cells were highlighted by light red dots in wild-type (Fig. 2A; Movie 4) and cyan dots in wnt9a/b double morphants. (Fig. 2B; Movie 5). The cardiomyocyte migration straightness index and velocity were determined by the same software and tracking algorithm. A straightness index equal to 1 indicates a cell that migrates straight over some distance, whereas a value of 0 indicates that the cell trembles and oscillates around the same position. Each dot in Fig. 2C,D represents the average of the straightness index (Fig. 2C) and velocity of cardiomyocytes (Fig. 2D) for each embryo analyzed. The averages of the straightness index and velocity between wild-type and wnt9a/b double morphants were compared and the statistical significance of the resulting difference was determined by a two-tailed, unpaired Student's t-test, using GraphPad Prism (version 9). The average values of cardiomyocyte progenitor cell straightness index of migration and velocity are indicated in Table 3.

Imaris software (Oxford Instruments; counting spots function) was used for counting cardiomyocytes as shown in Fig. 3A-C. To better distinguish cardiomyocytes for automatic counting, cardiomyocyte-specific EGFP was colocalized with DAPI. The automatic counting was done only with nuclei that were EGFP/DAPI double positive. The averages of the cardiomyocyte numbers between wild-type and wnt9^DKO mutants were compared and the statistical significance of the resulting difference was determined by a two-tailed, unpaired Student's t-test, using GraphPad Prism (version 9). The average values of cardiomyocyte numbers are indicated in Table 4.

All statistical analyses were performed with GraphPad Prism (version 9). Data representations and P-value calculations are indicated in the figure legends and method details. All indicated P-values are two-tailed and significance was defined as P<0.05.

We thank Julien Vermot (Imperial College London), for sharing the wnt9b^sa20083 mutant line and genotyping protocol. For critical reading and comments on the manuscript, we are indebted to all members of the Seyfried group and Russ Hodge. Thanks to O. Baumann for his support of confocal microscopy, and A. Hubig and B. Wuntke for technical support. The Zeiss light-sheet Z.1 microscope was funded by Deutsche Forschungsgemeinschaft grant INST 336/104-1 FUGG.