Hippo signaling determines the number of venous pole cells that originate from the anterior lateral plate mesoderm in zebrafish

The differentiation of the lateral plate mesoderm cells into heart field cells constitutes a critical step in the development of cardiac tissue and the genesis of functional cardiomyocytes. Hippo signaling controls cardiomyocyte proliferation, but the role of Hippo signaling during early cardiogenesis remains unclear. Here, we show that Hippo signaling regulates atrial cell number by specifying the developmental potential of cells within the anterior lateral plate mesoderm (ALPM), which are incorporated into the venous pole of the heart tube and ultimately into the atrium of the heart. We demonstrate that Hippo signaling acts through large tumor suppressor kinase 1/2 to modulate BMP signaling and the expression of hand2, a key transcription factor that is involved in the differentiation of atrial cardiomyocytes. Collectively, these results demonstrate that Hippo signaling defines venous pole cardiomyocyte number by modulating both the number and the identity of the ALPM cells that will populate the atrium of the heart.


Introduction
The human heart typically has about 2 billion cardiomyocytes (CMs) (Adler and Costabel, 1975;Laflamme and Murry, 2011), which together form the muscle layer of the heart responsible for contraction. The determination of the final number of CMs in the different parts of the heart involves highly coordinated processes of cell fate specification and proliferation during development. Understanding the relative contributions of these processes during the different stages of cardiac morphogenesis, as well as the mechanisms behind them, is one of the long-standing goals of the cardiac development field.
Mammalian heart morphogenesis is best studied in the mouse. Early in mouse development, a bilateral group of cells in the splanchnic mesoderm specifies into cardiac precursor cells (CPCs) (Saga et al., 1999) and forms the first heart field (FHF). Cells of the FHF then extend toward the midline to form a crescent-shaped epithelium, known as the cardiac crescent. Through a series of morphogenetic steps, the cardiac crescent gives rise to a structure called the heart tube (Kelly et al., 2014;Vincent and Buckingham, 2010). CPCs from the secondary heart field (SHF), which are derived from pharyngeal mesoderm, are added to the arterial and venous poles of the heart tube (Cai et al., 2003;Kelly et al., 2001;Waldo et al., 2001). The FHF is believed to give rise mostly to the left ventricle and parts of the atria, whereas the SHF is believed to give rise mostly to the right ventricle, the outflow tract (OFT) and most of the atria (Cai et al., 2003;Galli et al., 2008;Waldo et al., 2001;Zaffran et al., 2004).
The zebrafish has a simpler heart than that of mouse and humans, containing only two chambers. Nevertheless, the successive phases of CM differentiation during development, as well as their associated genetic pathways, are well conserved between zebrafish and other vertebrates (Staudt and Stainier, 2012). Because of the optical clarity and external development of zebrafish embryos, as well as the amenability of zebrafish to genetic manipulation, the zebrafish is an excellent model for the study of cardiac development. In the early stages of zebrafish development, a bilateral group of cells in the anterior lateral plate mesoderm (ALPM) specifies into CPCs, and the region where they reside is termed the heart field (HF) (Fishman and Chien, 1997). In zebrafish, both the FHF and the SHF derive from the ALPM (Mosimann et al., 2015). As in the mouse, the zebrafish FHF forms the initial heart tube while the SHF elongates the heart tube by adding to its arterial and venous poles. LIM domain transcription factor Islet1 (Isl1) marks a subset of SHF cells (de Pater et al., 2009;Witzel et al., 2012) that eventually give rise to the inflow tract (IFT) of the atrium of the mature heart. A second set of SHF cells is positive for Islet family member Islet2b (Isl2b) and for latent TGFb binding protein 3 (Ltbp3). These cells become the CMs that populate the arterial pole of the heart tube and eventually contribute to the OFT of the ventricle in the mature heart (Zhou et al., 2011;Witzel et al., 2017).
Given that CMs derive from the FHF and SHF cells of the mesoderm, the final CM number in the mature heart depends in part on the specification of mesodermal cells to form CPCs. Nkx2.5, Mef2c, and Hand2 are known to promote CM differentiation in zebrafish (Guner-Ataman et al., 2013;Hinits et al., 2012;Lazic and Scott, 2011;Schindler et al., 2014). Several signaling pathways have been found to act upstream of some of these transcription factors and to restrict the HF at the rostral and caudal boundaries of the ALPM. At the rostral border of the zebrafish ALPM, Tal1 and Etv2, two transcription factors required for vascular and hematopoietic lineage specification, respectively, repress cardiac specification, thereby reducing the number of CMs in the mature heart (Schoenebeck et al., 2007). At the border between the zebrafish ALPM and posterior LPM (PLPM), retinoic acid (RA) signaling from the adjacent forelimb field determines the HF size by restricting the potential differentiation of ALPM cells into heart precursor cells, and thus limits the number of atrial, but not ventricular, cells in the mature heart (Waxman et al., 2008).
In addition to molecular pathways regulating the fate decisions of CPCs, the final number of CMs is also determined by signalling pathways that regulate cell proliferation at each stage of cardiac morphogenesis (de Pater et al., 2009;Jopling et al., 2010;Rochais et al., 2009;Vincent and Buckingham, 2010). One well-characterized signaling pathway is the Hippo signaling pathway, which helps to define the number of cells in a variety of tissues and organs . In mammalian cells, the active elements of the Hippo pathway include Ste20-like serine/threonine kinase 1 and 2 (Mst1/2, mammalian orthologs of the fruit fly, Hippo), which phosphorylate Large tumour suppressor kinase 1 and 2 (Lats1/2). Phosphorylated Lats1/2 induce the nuclear export of the transcription factor Yes-associated protein 1 (Yap1) and its paralog WW-domain-containing transcription regulator 1 (Wwtr1), also known as Taz. Lats1/2 inhibits the formation of a complex involving Yap1/Wwtr1 and the TEA domain (TEAD) transcription factors by promoting the nuclear export of Yap1/Wwtr1, thereby repressing the expression of downstream target genes (Zhao et al., 2008).
In CMs of the mouse heart, Hippo signaling has been implicated in cardiac regeneration after myocardial injury (Lin et al., 2014;von Gise et al., 2012;Xin et al., 2013). While Yap1 and Wwtr1 double-null mutations in mice are embryonically lethal before the blastula stage (Nishioka et al., 2009), it has been shown that Nuclear Yap1 induces CM proliferation in the adult and fetal mouse. Furthermore, mice that are depleted of Lats2, Salvador (Salv), or Mst1/2 using CM-specific Cre drivers exhibit a hypertrophic growth due to an increase in CM proliferation . Together, these results suggest that Hippo signaling plays a key role in cardiac proliferation in the mouse. However, it is unclear whether Yap1/Wwtr1 are involved in CPC proliferation within the FHF and SHF before the formation of the heart tube. In addition, although Hippo signaling also regulates the expression of genes that are essential for cell specification and differentiation (Zhao et al., 2008;Nishioka et al., 2009), we still do not know whether Hippo signalling plays a role in cardiac cell fate specification.
In the work described here, we sought to examine the role of Hippo signaling in controlling heart cell number beyond its known roles in CM proliferation. Using zebrafish as a model, we examined the role of Hippo signaling at various stages of embryonic development: at the stage when embryos are specifying the HF, at the stage when the heart tube is formed, and in older embryos when heart morphogenesis is largely completed. We demonstrate that Lats1/2-Yap1/Wwtr1-regulated Hippo signaling determines the number of SHF cells in the venous pole that originate from the caudal part of the ALPM. At the molecular level, we show that Yap1/Wwtr1 promote bone morphogenetic protein-2b (bmp2b) expression and induce the phosphorylation of Smad1/5/9 in both hand2-and Isl1positive cells. Consistently, the absence of lats1/2 leads to increased hand2 expression at the boundary between the ALPM and the PLPM and to an increased number of SHF cells in the venous pole. Together, these findings demonstrate that Hippo signaling restricts the number of CPCs located in the venous pole by suppressing Yap1/Wwtr1-dependent Bmp2b expression and hand2 expression.
We assessed the effect of Lats1/2 depletion on heart development by counting CM number in the atrium and the ventricle of lats1/2 mutant larvae which also contained Tg(myosin heavy chain 6 [myh6]:Nuclear localization signal [Nls]-tagged tdEosFP);Tg(myosin light polypeptide 7 [myl7]:Nls-mCherry). These larvae expressed Nls-tdEosFP in the atrial CMs and Nls-mCherry in the whole CMs ( Figure 1A). We found that the number of atrial CMs, but not ventricular CMs, was significantly increased in the lats1 wt(wild type)/ ncv107 lats2 ncv108 embryos and in the lats1 ncv107 lats2 ncv108 embryos at 74 hr post-fertilization (hpf) ( Figure 1B,C and Figure 1-source data 1).
To confirm that Hippo signaling is involved in determining the number of CMs, we assessed whether Yap1/Wwtr1-dependent transcription is activated in developing CMs by analyzing two Tead-reporter transgenic lines: first, we used a general Tead reporter line, the Tg(eef1a1l1:galdb-hTEAD2DN-2A-mCherry), which expresses human TEAD2 lacking the amino-terminus (1-113 aa) fused with a GAL4 DNA-binding domain under the control of an eukaryotic translation elongation factor 1 alpha 1, like 1 (eef1a1l1) promoter (Fukui et al., 2014); second, we generated a CM-specific Tead reporter line, the Tg(myl7:galdb-hTEAD2DN-2A-mCherry), which expresses the same construct under the control of the myl7 promoter. By crossing these fish with Tg(uas:GFP) lines, cells with nuclear-translocated Yap1 or Wwtr1 can be identified through their expression of GFP (Fukui et al., 2014).
We first showed that the lats1/2 double knock-out (DKO) affected Hippo signaling using the general Tead reporter. We found that the Yap1/Wwtr1 reporter was active in the lats1/2 DKO embryos and in the lats1/2 morphants (Figure 1-figure supplement 1C). We next analyzed the activity of the CM reporter at 74 hpf and found that Yap1/Wwtr1-dependent transcription was present in IFT atrial CMs using the CM Tead reporter line (Figure 1-figure supplement 2A).
To identify the stage at which the Tead reporter starts to be active in CMs more precisely, we analyzed the activity of the reporter at an earlier stage. Previous studies have shown that IFT atrial CMs originate from the venous pole of the heart tube (de Pater et al., 2009). By examining the progeny of the Tead reporter line crossed with the Tg(myl7:Nls-mCherry) line at 24 hpf, we found that the Tead reporter was active in CMs located at the venous pole of the heart tube (Figure 1figure supplement 2B). Importantly, the eef1a1l1 and 2A peptide-driven mCherry expression used for screening purposes is very weak compared to mCherry expression driven by the myl7 promoter and does not affect the intensity analysis. Together, these data suggest that Lats1/2 restrict Yap1/ wwtr1-Tead activation during early CM determination.
Lats1/2 determine the number of IFT CMs derived from Isl1-positive SHF cells Following on from the observation that the Tead reporter is active at 24 hpf in the progeny of SHF cells, we next assessed whether Hippo signaling affects CM number at this stage. Isl1 is a SHF marker because it plays an essential role in the development of CPCs in the SHF, and because its expression delimitates the SHF as early as 24 hpf in zebrafish (Caputo et al., 2015). We generated transgenic fish expressing GFP under the control of an isl1 BAC promoter; the TgBAC(isl1:GFP). The isl1 transgene recapitulated the endogenous isl1 expression patterns (  Confocal images (at 26 hpf) of the TgBAC(isl1:GFP);Tg(myl7:Nls-mCherry) embryos with the lats1 wt/wt lats2 wt/wt (upper panels) or lats1 wt/ncv107 lats2 ncv108 alleles (bottom panels). The boxed regions are enlarged in the panels in the fourth (right) column. Yellow arrows indicate both isl1-and myl7-promoter activities in cells in the venous pole. White arrowheads indicate cells with isl1-promoter activity that are in contact with cells in which there is myl7promoter activity. Confocal 3D-stack images (left two panels) and single-scan images (right two panels). Dorsal view, anterior to the top. (D, E) Quantitative analyses of the number of the isl1-promoter-active SHF cells in the venous pole of the lats1 wt/wt lats2 wt/wt embryos, the lats1 wt/ncv107 lats2 ncv108 embryos, and the lats1/2 DKO embryos (D), and in embryos shown in Figure 2-figure supplement 1D (E). Both cells with isl1-and myl7-promoter activity and with isl1-promoter activity that were in contact cells with myl7-promoter activity were counted as SHF cells. The confocal 3D-stack images and single-scan (2 mm) images are a set of representative images of at least four independent experiments. ** p < 0.01. DOI: https://doi.org/10.7554/eLife.29106.006 The following source data and figure supplements are available for figure 2: Source data 1. The number of isl1-and myl7-promoter-active cells in IFT and OFT cells at 96 hpf ( Figure 2B), the numbers of isl1-promoter-active SHF cells in lats1/2 mutants ( Figure 2D) and the number of isl1-promoter-active SHF cells of embryos at 26 hpf injected with morpholino (MO) and mRNA ( Figure 2E  Lats1/2 determine the number of CMs derived from the hand2promoter-active CMs To identify the mechanism of action of Hippo signaling, we sought to identify the gene targets of Yap1/Wwtr1 in early CPC differentiation. We examined whether Yap1/Wwtr1 regulate the expression of transcription factors nkx2.5, hand2, and gata4, all of which are essential for early CPC differentiation (Schoenebeck et al., 2007). qPCR revealed that hand2 mRNA expression was significantly upregulated ( Figure 3A and Figure 3-source data 1) and that nkx2.5 and gata4 mRNAs expression was unaffected in the lats1/2 morphants ( Figure 3A and Figure 3-source data 1). Wholemount in situ hybridization (WISH) analyses revealed that the hand2 expression domain, corresponding to the region that gives rise to the heart, was expanded in the lats1 wt/ ncv107 lats2 ncv108 embryos, the lats1/2 DKO embryos, and the lats1/2 morphants at 22 hpf ( Figure 3B and Figure 3-figure supplement 1A). These data suggest that Lats1/2 determine the number of atrial CMs by inhibiting Yap1/Wwtr1-dependent hand2 expression.
Overexpression of Hand2 increases the number of SHF-derived CMs in zebrafish (Schindler et al., 2014), so we hypothesized that the increased number of CM in lats1/2 mutants is due to an increase in the number of CPCs in the SHF. We first investigated endogenous hand2 expression during CM development by analyzing TgBAC(hand2:GFP) (Yin et al., 2010), which labels cells in which the hand2 promoter is activated with GFP, and Tg(myl7:Nls-mCherry), which labels CMs and CM progenitors with nuclear-localized mCherry. At 26 hpf, we found that hand2-promoter-active CMs are localized on the anterior side of the growing cardiac tube, corresponding to the venous pole ( Figure 3C). In the lats1/2 DKO embryos, the number of hand2-promoter-active CMs was significantly increased in the venous pole ( Figure 3C,D and Figure 3-source data 1). Similarly, the number of hand2-promoter-active CMs was increased in the venous pole in the lats1/2 morphants (Figure 3-figure supplement 1B). Furthermore, we found that the population of Isl1-positive SHF cells overlaps with the hand2-promoter-active CMs in the very left-rostral end of the cardiac tube ( Figure 3E, brackets). As expected, the population of Isl1-positive SHF cells was increased in the lats1/2 DKO embryos and in the lats1/2 morphants ( Figure 3E and To confirm the involvement of the Hippo pathway in modulating the hand2 expression domain, we next sought to examine the number of hand2-promoter-active CMs in the DKO mutants of yap1 (Figure 3-figure supplement 2A) and wwtr1 (Nakajima et al., 2017). As expected, we found that Together, these results strongly suggest that the increased CPC number in the lats1/2 mutants is due to an expansion of the hand2 expression domain in the SHF. Quantitative-PCR analyses of the expression of nkx2.5, hand2, and gata4 mRNAs in the whole embryos at 24 hpf showing the effects of MO injection (n = 4). Relative expression of mRNA in the MO-injected morphants to that of the control is shown. (B) WISH analyses at 22 hpf of lats1 wt/wt lats2 wt/wt (n = 7), lats1 wt/ncv107 lats2 wt/ncv108 (n = 22), lats1 wt/ncv107 lats2 ncv108 (n = 14) and lats1 ncv107 lats2 ncv108 (n = 6) embryos using an antisense probe for hand2. (C) Confocal 3D-stack images (at 26 hpf) of TgBAC(hand2:GFP);Tg(myl7:Nls-mCherry)-labeled embryos carrying the lats1 wt/wt lats2 wt/wt (upper panels) or lats1 ncv107 lats2 ncv108 allele (bottom panels). GFP images (left), merged GFP and mCherry images (center), and enlarged images of the boxed regions in the center panels (right) are shown. (D) Quantitative analysis of the number of cells in which both hand2 and myl7 promoters are activated at 26 hpf. (E) Confocal 3D-stack images (at 26 hpf) of the TgBAC(hand2:GFP);Tg(myl7:Nls-mCherry)-labeled embryos containing the lats1 wt/wt lats2 wt/wt (upper panels, n = 9) or the lats1 ncv107 lats2 ncv108 allele (bottom panels, n = 5) immunostained with the anti-Isl1 antibody (anti-Isl1 Ab). Square brackets denote the SHF cells that are Isl1-positive, both hand2-and myl7-promoter-active cells that are Isl1-positive, and hand2-promoter-active cells that are in contact with myl7-promoter-active cells. pp indicates the pharyngeal pouch, which expresses the hand2-promoter-activated GFP signal. The first, second, third and fourth columns show Isl1 immunostaining, merged images of GFP and Isl1 immunostaining, merged images of Isl1 immunostaining and mCherry labeling, and merged images of all the three (GFP, mCherry, and Isl1 immunostaining), respectively. All of the images are of the dorsal view, anterior to the top. The confocal 3D-stack images and the WISH images are a set of representative images from at least four independent experiments. ** p < 0.01. DOI: https://doi.org/10.7554/eLife.29106.010 The following source data and figure supplements are available for figure 3: Source data 1. Quantification of the relative mRNAs expression levels ( Figure 3A) and the numbers of both hand2-and myl7-promoter-active cells ( Figure 3D hand2-promoter-active cells at the caudal end of the ALPM migrate toward the venous pole of the cardiac tube In zebrafish, the origin of venous pole CMs is unknown. In amniotes, the venous pole progenitors are located in the most caudal domain of the ALPM (Abu-Issa and Kirby, 2008; Galli et al., 2008). Considering that hand2 is expressed in the zebrafish LPM (Schoenebeck et al., 2007), we hypothesized that hand2-promoter-active cells originate from the caudal end of the ALPM. We performed time-lapse imaging from 14 hpf to 26 hpf to investigate whether hand2-promoter-active cells of the LPM contribute to the venous pole cells. Cell-tracking analysis revealed that the caudal cells of the ALPM migrate toward the venous pole ( Figure 4A,B, and Video 1). ALPM cells move toward the posterior of the cardiac disc by 20 hpf, and subsequently move anterior-laterally toward the venous pole of the cardiac tube by 26 hpf (Figure 4A-C). These results indicate that hand2-promoter-active cells in the venous pole differentiate from the caudal ALPM. To confirm that the caudal ALPM cells are incorporated into the IFT atrial CMs, we sought to perform cell-tracking of hand2-promoteractive cells in the caudal region of ALPM cells following photoconversion. To do so, we injected embryos with a plasmid that expresses tdEosFP under the control of hand2 BAC promoter and photoconverted the cells in the caudal region of both sides of ALPM in these embryos. We found that the photoconverted cells were incorporated into the IFT of the atrium and the OFT of the ventricle ( Figure 4D). This indicates that the cells of the caudal region of both sides of ALPM can become IFT CMs of the atrium.
We next analyzed whether the Tead reporter is active in the hand2 expression domain of the ALPM. We crossed TgBAC(hand2:GFP) lines with general Tead mRFP1 reporter fish. At the 12 somite stage (ss), the Tead-reporter-active cells were found in the entire region of the ALPM and overlapped with hand2-promoter-active cells in ALPM ( Figure 4E). We further noticed that the Tead reporter was inactive in the rostral region of the PLPM at 12 ss ( Figure 4-figure supplement 1A). These data suggest that Hippo signaling acts upstream of hand2 expression in the ALPM and may play a role in determining CM fate in the ALPM.
Hippo signaling regulates the number of SHF cells from the caudal end of the ALPM Tead reporter activation in the cells of ALPM prompted us to ask whether Lats1/2-Yap1/Wwtr1 signaling is involved in the proliferation and/or specification of those cells. We examined the proliferation of isl1-promoter-active cells using the EdU incorporation assay and found that the number of isl1-promoter-active and EdU-positive CMs in the lats1/2 morphants was comparable to that of controls ( Figure 5A,B). In addition, there was no difference in the number of EdU-positive blood cells and endocardial cells among the two groups (data not shown). Importantly, the timing of EdU incorporation did not affect the results of the proliferation analyses ( Figure 5B), suggesting that the increase in the number of isl1-promoter-active SHF cells that resulted from the depletion of Lats1/2 is not caused by cell proliferation after the differentiation of SHF cells from the ALPM.
We next tested whether Lats1/2 affect SHF cell specification in the ALPM. At 10 ss, the ALPM and the PLPM can be characterized by the expression of gata4, nkx2.5, tal1, and hand2. gata4 labels the multipotent myocardial-endothelial-myeloid progenitors of the ALPM; nkx2.5 is a marker for the ventricular HF; tal1 marks the hematopoietic cell progenitors; hand2 marks both the ALPM and the PLPM at 10 ss ( Figure 5C) with a clear gap in between. Interestingly, the gap length of the lats1 wt/ ncv107 lats2 ncv108 embryos, the lats1/2 DKO embryos, and the lats1/2 morphants was significantly shorter than that of wildtype embryos ( Figure 5D . We also found that tal1 expression was decreased in the rostral end of the PLPM in the lats1/2 DKO embryos and the lats1/2 morphants ( Figure 5F and Figure 5-figure supplement 1B). A similar analysis in yap1/wwtr1 DKO embryos revealed that hand2 expression was decreased in both ALPM and PLPM but that the expression of tal1 was unaffected ( Figure 5D,F). Importantly, the expression of gata4 and nkx2.5 was unaffected in lats1/2 DKO embryos and in lats1/2 morphants, suggesting that Hippo signaling mainly affects the hand2 expression domain ( Figure 5G, and Figure 5-figure supplement 1C). These observations were consistent with the results of qRT-PCR ( Figure 3A).
To confirm the specificity of Hippo action, we examined the expression of etv2, which is a marker for blood-vessel progenitors (Schoenebeck et al., 2007), and of hoxb5b, which is a regulatory molecule of RA signaling in the forelimb field (Waxman et al., 2008) in the LPM. The expression levels of both etv2 and hoxb5b were comparable between the control and the lats1/2 morphants ( Figure 5-figure supplement 1C). Collectively, these results suggest that Lats1/2 negatively regulate Yap1/Wwtr1-dependent differentiation of the LPM into the SHF at the boundary between ALPM and PLPM.
Hippo signaling regulates Bmpdependent smad activation that determines the number of SHF cells in the venous pole Signaling mediated by Bone morphogenetic proteins (Bmps) affects various contexts of heart development via Smad phosphorylation-dependent transcriptional activation. Bmp-Smad signaling is known to be essential for SHF formation, FHF-derived CM development, endocardium development and epicardium development (Prall et al., 2007;Schlueter et al., 2006;Tirosh-Finkel et al., 2010;Yang et al., 2006). Yap1 is known to promote Bmp2b expression in zebrafish neocortical astrocyte differentiation (Huang et al., 2016) and Bmp2 in mouse endothelial cells (Neto et al., 2018). In the zebrafish embryo, bmp2b, but not bmp4, is expressed in the LPM (Chung et al., 2008). We hypothesized that Yap1/Wwtr1 are involved in bmp2b-dependent signaling during early cardiogenesis. To investigate whether Bmp-Smad signaling is activated in the ALPM, we examined Bmp-dependent transcription using the Tg(BRE:GFP) fish embryos, in which the Bmp-responsive element (BRE) drives GFP expression (Collery and Link, 2011). At 14 hpf, BRE-positive cells were found in the ALPM (Figure 6-figure supplement 1A). Because Bmps induce the phosphorylation of Smad1/5/9 (Smad9 is also known as Smad8) (Heldin et al., 1997), we examined the phosphorylation of Smad1/5/9 in the embryos at 14 hpf using immunohistochemistry. The phosphorylated Smad1/5/9-positive cells were found in the ALPM and eyes ( Figure 6A). Phosphorylation of Smad1/5/9 was enhanced in the lats1/2 DKO embryos and decreased in the yap1/ wwtr1 DKO embryos ( Figure 6A). At 10 ss, bmp2b expression was increased in the ALPM and eyes of the lats1/2 DKO embryos and decreased in the yap1/wwtr1 DKO embryos ( Figure 6B). Consistently, bmp2b mRNA was increased in the lats1/2 morphants at 10 ss ( Figure 6-figure supplement  1B). Although we could not detect bmp4 in the ALPM in the early ss (data not shown), bmp4 mRNA was increased in the venous pole of the lats1/2 morphants at 26 hpf ( Figure 6-figure supplement  1C). These results suggest that Hippo signaling functions upstream of Bmp-dependent Smad activation in the ALPM during early cardiogenesis.
By analyzing the BRE reporter, we found that the number of Bmp signal-active cells marked by GFP in the venous pole was increased in the lats1 wt/ncv107 lats2 ncv108 embryos and/or the lats1/2 DKO embryos, as well as in the lats1/2 morphants at 24 hpf ( Figure 6C,D, Figure 6-figure supplement 2A and Figure 6-source data 1). Immunohistochemistry revealed that the number of phosphorylated Smad1/5/9-positive and hand2-promoter-active cells was also increased at the venous pole of the lats1/2 morphants at 26 hpf ( Figure 6-figure supplement 2B). These results suggest that Yap1/Wwtr1 promote bmp2b expression and subsequent Bmp signaling and that Lats1/2 restrict Yap1/Wwtr1-dependent Bmp signaling leading to the formation of the proper venous pole.
We further investigated whether Bmp-Smad activation promotes hand2 expression. We made use of Smad7, an inhibitory-Smad that blocks Bmp-Smad signaling by interacting with activated Bmp Video 1. hand2-promoter-active cells in the caudal region of the ALPM move to the venous pole. Timelapse recording of 3D-rendered confocal images of the TgBAC(hand2:GFP) embryo from 14 hpf (10 ss) to 26 hpf. Note the migration of the caudal region of both the left ALPM (magenta) and the right ALPM (cyan) toward the venous pole of the cardiac tube. Changes in the colors reflect the tracking time (blue, 0 hr; red, 12 hr). Dorsal view, anterior to the top. The time-lapse movie is a set of representative data from six independent experiments. Video 1 is related to Figure 4A  type I receptors and preventing the downstream activation of receptor-regulated Smads (Souchelnytskyi et al., 1998). We first showed that overexpression of smad7 mRNA caused dorsalization, demonstrating that Smad7 is essential for proper Bmp-Smad signaling, using 200 pg of mRNA ( Figure 6-figure supplement 3A,B). Interestingly, the injection of lower concentration of mRNA (100 pg) led to a dorsalization phenotype in only 10% of the injected embryos. Therefore, the remaining embryos without dorsalization phenotype were used to assess heart development (Figure 6-figure supplement 3A,B). The non-dorsalized embryos looked healthy. The injection of smad7 mRNA did not cause fragmentation of the cells (data not shown), but the embryos exhibited a decreased number of isl1-promoter-active cells in the TgBAC(isl1:GFP);Tg(myl7:Nls-mCherry) assay ( Figure 6E,F and Figure 6-source data 1). Collectively, these results suggest that Bmp-dependent signaling is required for CPC fate determination.
Finally, to confirm the necessity of Bmp-Smad-regulated signaling during SHF formation, we treated the TgBAC(hand2:GFP);Tg(myl7:Nls-mCherry) embryos and the TgBAC(isl1:GFP);Tg(myl7: Nls-mCherry) embryos with a Bmp inhibitor, DMH1, from 14 hpf to 26 hpf. The efficiency of DMH1 was confirmed by decreased phosphorylation of Smad1/5/9 ( Figure 6-figure supplement 2C). The expression of Isl1 and the promoter activity of hand2 were greatly reduced in the embryos treated with DMH1 ( Figure 6-figure supplement 3C). The number of isl1-promoter-active SHF cells was decreased in the embryos treated with DMH1, whereas the number of isl1-promoter-inactive CMs in the DMH1-treated embryos was comparable to that in the control embryos ( Figure 6G,H, and Figure 6-source data 1). We thus conclude that Lats1/2 restrict Yap1/Wwtr1-promoted Bmp2bdependent signaling, which is required for both hand2-and isl1-promoter activity during SHF formation.

Discussion
Here, we show for the first time that the Hippo signaling pathway is involved in the determination of LPM cell fate by promoting venous pole identity and increases in atrial CM number (Figure 7). We show that Yap1/Wwtr1-promoted signaling increases the size of the SHF domain and that Lats1/2, by inhibiting Yap1/Wwtr1 activity, restrict it. We propose that the increased number of SHF cells in the lats1/2 DKO embryos may result from a change in fate determination of hand2-negative cells, which become hand2-positive cells in the boundary between ALPM and PLPM. Indeed, we found that the expression of the marker of blood-cell progenitors tal1 was repressed in the rostral region of PLPM in lats1/2 DKO embryos. lats1/2 mutants exhibited a subtle increase in the number of Isl1positive atrial SHF cells, with no defect apparent in other organs. Importantly, despite the known role of Hippo signaling in CM proliferation during heart maturation, Hippo signaling was not found to affect cell proliferation during these early stages. Therefore, Hippo signaling contributes specifically to the determination of LPM differentiation.
We speculate that Hippo signaling cooperates with other signaling pathways to determine HF formation. For example, hoxb5b expression in the forelimb field limits the extent of the HF at the posterior border of the ALPM, cells of which differentiate into atrial but not ventricular CMs (Waxman et al., 2008). Furthermore, other signals generated in the pronephric field in the intermediate mesoderm and in the angiogenic field in the rostral region of PLPM are important for the regulation of cell fate at the posterior HF boundary (Kimmel et al., 1990;Mudumana et al., 2008).
Our results also help to clarify the origin of venous pole CMs in zebrafish and its link to the SHF. In the mouse embryo, the anterior and posterior SHF cells differentiate into the OFT/right ventricular myocardium and the IFT/atrial myocardium, respectively (Galli et al., 2008;Verzi et al., 2005). The posterior-SHF in the HF is located caudally (Abu-Issa and Kirby, 2008;Galli et al., 2008). Teadreporter activation occurred in the caudal zone of the ALPM, and these cells were shown to become Source data 1. The numbers of BRE-positive cardiomyocytes in lats1/2 mutants at 24 hpf ( Figure 6D), isl1-promoter-active SHF cells with smad7 mRNA injection ( Figure 6F) and isl1-promoter-active SHF cells and GFP-negative CMs with DMH1 treatment ( Figure 6H Isl1-positive, hand2-promoter-active CMs that are localized to the venous pole of the atrium. Both sides of the HF located caudally contributed to the venous pole of the cardiac tube. Further, mammalian SHF cells have multi-potential to differentiate into endocardial cells and smooth muscle cells in addition to myocardial cells (Chen et al., 2009b). By generating BAC transgenic fish, we found that isl1-promoter-active cells were detected in the atrial myocardium, endocardium, and epicardium, but not in the ventricular myocardium. Together with previous reports, our results suggest that the properties of zebrafish Isl1-positive SHF cells are similar to those of mammalian posterior-SHF cells.
Lats1/2-Yap1/Wwtr1-Tead signaling regulates Bmp2b expression, which is necessary for the formation of hand2-promoter-active cells in the ALPM and Isl1-expressing cells. Although previous reports have shown that isl1-positive and mef2-positive cells reside at the venous pole (de Pater et al., 2009;Hinits et al., 2012), the molecular mechanism explaining how ALPM cells give rise to these CPCs at the venous pole has remained unclear. To date, a number of signaling molecules, such as TGFb, FGF, and BMP, have been reported to regulate arterial pole formation in the zebrafish Figure 7. A schematic illustration of Hippo signaling in the ALPM and the border of the ALPM and PLPM in wildtype (WT) and Lats1/2 double knockout (DKO) embryos. In WT embryos, at the caudal part of the ALPM, Hippo signaling is inactive, whereas Tead-dependent transcription co-activated by Yap1/Wwtr1 is active and promotes bmp2b expression and hand2 expression. Cells expressing hand2 become the Isl1-positive SHF cells in the venous pole of the heart tube, and eventually populate the inflow tract. In the rostral region of the PLPM, Hippo signaling is active and hand2 expression is suppressed. Cells from this region do not become cells of the venous pole. In the Lats1/2 DKO, Hippo signaling is absent in the caudal part of the ALPM, and hand2 expression is increased. In the rostral region of the PLPM, Hippo signaling is absent and hand2 is expressed. hand2 expression in these cells promotes SHF specification, and these cells are integrated into the venous pole of the heart tube, and eventually populate the atrium of the heart. DOI: https://doi.org/10.7554/eLife.29106.025 heart tube (de Pater et al., 2009;Hami et al., 2011;Zhou et al., 2011). We found that the Tead activation signal overlaps with the Bmp-reporter-positive signal in the ALPM. By analyzing lats1/2 mutants and yap1/wwtr1 mutants, we show that Hippo signaling controls bmp2b expression in the ALPM. Bmp-Smad inhibition expands the tal1 expression domain to restrict LPM fate (Gupta et al., 2006). Interestingly, hand2 expression is diminished in the alk3 mutant, which is affected in a Bmp type I receptor 1a, at 12 ss (de Pater et al., 2012). In our experiments, Bmp-Smad inhibition results in the suppression of hand2-promoter-activated GFP expression at 15 hpf. Therefore, we conclude that bmp2b expression is positively regulated by Yap1/Wwtr1, which balances the cell fate between the HF and the blood cell field at the boundary between the ALPM and the PLPM.
We believe that the downstream targets of the Hippo-dependent Bmp-mediated signal, especially the transcription-factor-(Smads)-dependent signal, might promote distinct functions that depend on their time of action. It has been shown previously that Bmp signaling is upstream of both hand2 and nkx2.5 in zebrafish, and that the expression of both mRNAs is lost in Bmp signaling mutants (de Pater et al., 2012;Kishimoto et al., 1997;Reiter et al., 2001). However, we found here that yap1/wwtr1 double mutants still have hand2 expression (although this is reduced) and normal nkx2.5 expression, even though Bmp signaling is decreased. Consistent with this possibility, Bmp signal-dependent dorso-ventral axis formation is not dependent on Lats1/2-Yap1/Wwtr1-Bmp signaling, suggesting that additional regulators of Bmp signaling are active in the ALPM. Another possibility is that Hippo signaling acts independently on both bmp2b expression and hand2 expression.

Image acquisition and image processing
To obtain the images of embryos, the pigmentation of the embryos was suppressed by the addition of 1-phenyl-2-thiourea (PTU) (Sigma-Aldrich, St. Louis, MO) into breeding E3 media. Embryos were dechorionated and mounted in 1% low-melting agarose dissolved in E3 medium. Confocal images of 2.0 mm steps were taken with a FV1200 confocal microscope system (Olympus, Tokyo, Japan) equipped with a water immersion 20x lens (XLUMPlanFL, 1.0 NA, Olympus). Images were processed with a FV10-ASW 4.2 viewer (Olympus). The distance between the hand2-positive region of the ALPM and the PLPM was measured using DP2-BSW software (Olympus). Cell-tracking data containing nuclei positions were analyzed using Imaris8.4.1 software (Bitplane, Zurich, Switzerland).

EdU incorporation assay
The TgBAC(isl1:GFP);Tg(myl7:Nls-mCherry) embryos injected with control MO or lats1/2 MOs were incubated with 2 mM of 5-ethynyl-2-deoxyuridine (EdU) from 14 to 26 hpf or from 20 to 36 hpf, and subsequently fixed using 4% PFA at 96 hpf. EdU-incorporated cells were labelled by Click-iT EdU Alexa Fluor 647 Imaging Kits (Thermo Fisher Scientific) following the manufacturer's instructions. Images were taken using the FV1200 confocal microscope system. The number of EdU-positive isl1promoter-active CMs was determined by counting the number of cells with overlapping Alexa Fluor 647-positive signal, isl1-promoter-activated signal and myl7-promoter-activated signal.

Whole-mount in situ hybridization (WISH)
The antisense hand2, isl1, bmp2b, bmp4, gata4, nkx2.5, etv2, tal1, and hoxb5b RNA probes labeled with digoxigenin (DIG) were prepared using an RNA labeling kit (Roche, Basel, Switzerland). WISH was performed as previously described (Fukui et al., 2014). Colorimetric reaction was carried out using BM purple (Roche) as the substrate. To stop the reaction, embryos were washed with PBS-T, fixed with 4% PFA for 20 min at room temperature, and subsequently immersed in glycerol. Images were taken using a SZX-16 Stereo Microscope (Olympus).

Generation of transgenic lines
To monitor atrial CM development, we established a transgenic (Tg) zebrafish line expressing a nuclear localization signal (Nls)-tagged tandem Eos fluorescent protein under the control of the myosin heavy chain 6 (myh6) promoter; Tg(myh6:Nls-tdEosFP). pTol2-myh6 vector was constructed by modifying the pTol2 vector and inserting the myh6 promoter as a driver of the expression of the target molecule. The primers used to amplify the myh6 promoter were 5 0 -AGAGCTAAAGTGGCAGTG TGCCGAT-3' and 5 0 -TCCCGAACTCTGCCATTAAAGCATCAC-3 0 . An oligonucleotide encoding Nls derived from SV40 (PKKKRKV) was inserted into pcDNA-tdEosFP (MoBiTec, Gö ttingen, Germany) to generate the plasmids expressing Nls-tagged tandem Eos fluorescent protein (Nls-tdEosFP). The Nls-tdEosFP cDNA was subcloned into the pTol2-myh6 vector to construct the pTol2-myh6:Nls-tdEosFP plasmids.
To monitor CM development, we developed a transgenic line that expresses EGFP under the control of the myl7 promoter; Tg(myl7:EGFP). The EGFP was subcloned into a pTol2-myl7 vector to construct the pTol2-myl7:EGFP plasmids. All of the cDNAs amplified by PCR using cDNA libraries were confirmed by DNA sequencing.
To monitor SHF development, we established a transgenic line that expressed GFP under the control of isl1 BAC promoter/enhancer; the TgBAC(isl1:GFP). pRedET plasmid (Gene Bridges, Heidelberg, Germany) was introduced into E. coli containing a CH211-219F7 BAC clone encoding the isl1 gene (BacPAC resources) by electroporation (1800V, 25 mF, 200 W) to increase the efficiency of homologous recombination, as previously described (Ando et al., 2016). Tol2 long terminal repeats in opposite directions flanking an ampicillin resistance cassette were amplified by PCR using Tol2_amp as a template, and these sequences were inserted into the BAC vector backbone. The cDNA encoding both GFP and a kanamycin resistance cassette (GFP_KanR) was amplified by PCR using a pCS2-GFP_KanR plasmid as a template, and inserted into the start ATG of the isl1 gene. Primers to amplify the GFP_KanR for isl1 gene were 5 0 -gggccttctgtccggttttaaaagtggacctaacaccgccttactttct-tACCATGGTGAGCAAGGGCGAGGAG-3' and 5 0 -aaataaacaataaagcttaacttacttttcggtggatcccc-catgtctccTCAGAAGAACTCGTCAAGAAGGCG-3' (small letters are the homology arm to the BAC vector, whereas capital letters are the primer binding site to the template plasmid).
Tol2-mediated zebrafish transgenesis was performed by injecting 30 pg of the transgene plasmid together with 50 pg tol2 transposase mRNA, followed by subsequent screening of F 1 founders and establishment of single-insertion transgenic strains through selection in F 3 generations.

Photoconversion
We performed photoconversion experiments by examining the transient hand2-promoter-dependent expression of tdEosFP. Tg(myl7:EGFP) embryos were injected with 30 pg of pTol2-hand2 BAC: tdEosFP plasmid along with 50 pg tol2 transposase mRNA. To trace ALPM cells, the caudal region of the left or right ALPM was photoconverted by a 405 nm laser at 15 hpf (12 ss). The photoconverted cells expressing red fluorescence were traced in the heart region at 52 hpf.
To construct the pTol2-hand2 BAC:tdEosFP, pRedET plasmid was introduced into E. coli containing a CH211-95C16 BAC clone encoding the hand2 gene (BacPAC resources) by electroporation. Tol2 long terminal repeats in opposite directions flanking an ampicillin resistance cassette were amplified by PCR using Tol2_amp as a template, and were inserted into the BAC vector backbone. The cDNA encoding tdEosFP together with a kanamycin resistance cassette (tdEosFP_KanR) was amplified by PCR using the pCS2-tdEosFP_KanR plasmid as a template, and then inserted into the start ATG of the hand2 gene. Primers to amplify the tdEosFP_KanR for the hand2 gene were 5 0 -ccaaagcgtactccgtctgtggttcgccgtagggtatagacaagtctgtcACCATGAAGATCAACCTCCGTATGGAAG-3' and 5 0 -tagccgtcatggtgcatcacagggtggtggggaaaccctccaactaaactTCAGAAGAACTCGTCAA-GAAGGCG-3' (small letters represent the homology arm to BAC vector, and capital letters the primer binding site to the template plasmid).

Data analysis and statistics
Data were analyzed using GraphPad Prism 7 (GraphPad Software, La Jolla, CA). All columns shown in histograms represent a mean ± SEM. The statistical significance of multiple groups was determined by one-way ANOVA with Bonferroni's post hoc test. The numbers of atrial and ventricular CMs at 74 hpf were analyzed by Student's t-test. The statistical significance of two groups was determined by Student's t-test.