Hey2 regulates the size of the cardiac progenitor pool during vertebrate heart development

Cardiac development is regulated by the activity of concerted signaling, transcriptional and morphogenetic events. Subtle perturbations in these processes, either genetic or environmental, can lead to congenital heart defects (CHDs), the most common class of congenital anomalies. It is now evident that the vertebrate heart is built from two populations of cardiac progenitor cells (CPCs), termed the first heart field (FHF) and second heart field (SHF), which contribute to the heart in two successive windows of differentiation. Cells of the FHF differentiate in an initial wave of cardiogenesis, resulting in formation of the linear heart tube. Over a well-defined developmental window, multi-potent, late-differentiating progenitors of the SHF migrate into the poles of the heart tube to extensively remodel and add structure to the heart (Cai et al., 2003; van den Berg et al., 2009; Hutson et al., 2010; Kelly, 2012). It remains under debate, however, whether the FHF and SHF represent truly distinct populations with unique molecular signatures, or whether they exist as one population with a gradient in timing for deployment to the heart (Abu-Issa et al., 2004; Moorman et al., 2007; Ivanovitch et al., 2017).

As a key driver of cardiac morphogenesis, the balance of proliferation, maintenance and cardiac differentiation events in the CPC compartment must be tightly regulated. However, although cardiac cell fate diversification has been elegantly studied via clonal analysis (Meilhac et al., 2003; Lescroart et al., 2010; Devine et al., 2014), how the magnitude of contribution to the developing heart from the CPC pool is regulated remains poorly understood. This process has perhaps been most examined in SHF progenitors, where several signaling pathways have been implicated in regulating the balance of CPC maintenance/proliferation and differentiation (Prall et al., 2007; Ryckebusch et al., 2008; Sirbu et al., 2008; Tirosh-Finkel et al., 2010; Zhao et al., 2014; Li et al., 2016; Mandal et al., 2017). Notable among these are fibroblast growth factor (FGF) and retinoic acid (RA) signaling, the balance of which fine-tunes the specification and differentiation of cardiac progenitors (Ilagan et al., 2006; Park et al., 2008; Rochais et al., 2009; Sorrell and Waxman, 2011; Witzel et al., 2012). Transcription factors including Islet1 (Isl1), Nkx2.5 and Fosl2 are essential for maintaining the SHF progenitor pool (Cai et al., 2003; Prall et al., 2007; de Pater et al., 2009), yet the full make-up of the transcriptional network required to precisely regulate the behavior of the CPC population remains unclear.

Hey2 is a member of the Hairy/Enhancer of split basic helix-loop-helix (bHLH) subfamily, which act as transcriptional repressors during embryonic development (Davis and Turner, 2001). Zebrafish hey genes exhibit more restricted expression patterns compared with mammals, with hey2 being the only family member detectably expressed in the heart (Winkler et al., 2003). Based on computational approaches, Hey2 has recently been predicted to be a key regulator of human cardiac development (Gerrard et al., 2016). Studies in mice have shown that disruption of hey2 can lead to ventricular septal defects (VSD), as well as other CHD and cardiomyopathy (Donovan et al., 2002; Sakata et al., 2002). Mutation in zebrafish hey2 leads to a localized defect of the aorta resembling human aortic coarctation (Weinstein et al., 1995; Zhong et al., 2000), with Hey2 having a key role in specifying arterial versus venous cell fates (Zhong et al., 2000, 2001; Hermkens et al., 2015). Of note, Hey2 has been suggested to regulate growth of the heart via restraining cardiomyocyte (CM) proliferation (Jia et al., 2007), as well as functioning downstream of Nkx2.5 to prevent ectopic atrial identity within ventricular myocardium (Xin et al., 2007; Targoff et al., 2013). More recent work in human pluripotent stem cells has suggested a key function for HEY2 as an effector of NKX2.5 during cardiomyocyte differentiation (Anderson et al., 2018).

Given the evidence linking Hey2 function with CHD in various model systems, and the association between CHD and SHF progenitors (Cai et al., 2003; Ward et al., 2005), we hypothesized that hey2 may play an earlier than appreciated role in regulating the cardiac progenitor pool. Analysis of a novel null hey2 mutant allele revealed increased myocardial cell number as early as the linear heart tube stage, with a subsequent more robust and extended late addition of cardiac progenitors to the heart. hey2 expression localized to regions overlapping and adjacent to the nkx2.5-positive cardiac mesoderm, suggestive of a function for Hey2 in cardiac progenitors, prior to cardiac differentiation. Through the generation of an inducible lineage-tracing system, we found that hey2-expressing cells progressively give rise to multiple portions of the developing heart. Temporal analysis demonstrated that in the absence of Hey2 function cardiac progenitor cells underwent increased proliferation prior to, but not following, addition to the heart, resulting in an increased cardiac progenitor pool, with hey2 likely acting in a cell-autonomous manner in this context. Taken together, these results suggest that hey2 acts as both a key brake on the proliferative capacity of early cardiac progenitors, affecting the extent of cardiomyocyte addition to the heart.

Previous reports have demonstrated the consequences of knocking down hey2 gene function using either antisense morpholinos (MOs) or with existing hey2 mutants: an ENU mutagenesis allele (grl^m145, (Weinstein et al., 1995; Zhong et al., 2000) and in mosaic TALEN-injected embryos (Hermkens et al., 2015). However, overexpression experiments had suggested that the grl^m145 allele, which encodes a Hey2 protein with a 44 amino acid C-terminal extension, retains some function (Zhong et al., 2000; Jia et al., 2007). With both this and recent controversy regarding use of antisense morpholinos to assess gene function in zebrafish (Kok et al., 2015) in mind, we generated a novel null mutation in hey2 using CRISPR/Cas9-mediated genome editing. By targeting the gene within exon2, we generated a mutant with an 8 bp deletion that produces a predicted premature stop codon at amino acid 53, resulting in deletion of the essential bHLH domain (Fig. 1A). The hey2^hsc25 mutant phenotype became readily apparent by 72 h post-fertilization (hpf, Fig. 1B,C), with mutants exhibiting pericardial edemas accompanied by an enlarged atrium and truncated ventricle (Fig. 1D,E). For experiments, hey2 heterozygotes were chosen as the control group, given their preponderance in the crosses used, and the observation that no significant differences were observed between heterozygous and homozygous wild-type embryos. As in grl^m145 mutants (Zhong et al., 2001), hey2 mutants demonstrated a blockage at the aortic bifurcation, preventing blood flow to the trunk, evident by the accumulation of gata1⁺ red blood cells in the trunk vasculature (Fig. 1F,G, arrowhead). Hey2 mutant embryos displayed a non-looped heart and a reduction in heart rate when compared with controls (Fig. 1H-J; 118.5±0.86 wild-type control; 115.0±2.54 hey2 het; 87±3.87 hey2 mutant, mean±s.e.m.). Immunofluorescent analysis using the well-characterized antibodies MF20 and S46 revealed morphological changes, with apparent denser packing of cardiomyocytes (CMs) in both the atrial and ventricular chambers at 48 hpf in hey2 mutant embryos (Fig. 1K-L), agreeing with a previously described thickening of the ventricular wall in grl^m145 mutants (Jia et al., 2007). This phenotype was quantified through quantitative real-time PCR (qRT-PCR) at 48 hpf, revealing a four- and fivefold increase in both amhc and vmhc expression in hey2 mutants, when compared to heterozygotes (Fig. 1M).

Having established a hey2 genetic null model, we next evaluated CM number over the course of heart development. As previously reported (Jia et al., 2007), there was a significant increase in myl7:nlsDsRedExpress-positive CM nuclei at 48 hpf in the absence of hey2 when compared with controls (238.7±2.4 in hey2 mutants; 183±1.5 in heterozygous controls, mean±s.e.m., Fig. 2A-C). At 26 hpf, hey2 mutants had an irregularly shaped heart that had failed to elongate, with an expansion in expression of CM differentiation markers myl7 and tnnt2 at the poles of the heart (Fig. 2D-G, arrowheads), as well as demonstrating an increase in transcript levels following qPCR (Fig. 2H,I). Given this observation, we used antibody staining to count CM number in hey2 mutant Tg(myl7nls:DsRedExpress) embryos at 24 hpf. Whereas heterozygous hey2^+/− control embryos contained 143.6±5 (mean±s.e.m.) CMs at 24 hpf, hey2 mutant embryos contained a significantly greater number (182.4±3.9, mean±s.e.m.; Fig. 2J-L). These results highlight that, in the absence of hey2, an elevated number of CMs is detectable as early as 24 hpf, arguing that proliferation of differentiated CMs between 24-48 hpf alone cannot explain the expanded CM population present at 48 hpf.

Previous work has characterized the differentiation and addition of SHF progenitors to the zebrafish heart tube between 24 and 48 hpf (de Pater et al., 2009; Hami et al., 2011; Lazic and Scott, 2011). The expanded ectopic expression of myl7 at 24 hpf in hey2 mutants (Fig. 2E, compare with D) led us to speculate that the SHF population may be expanded. We therefore employed transgenic myl7:nlsKikGR embryos to photoconvert CMs at 24 hpf and monitor subsequent novel (unconverted at 24 hpf) myocardial addition to the heart (Fig. 3A). Given the difficulty in discerning mutants from wild-type embryos at 48 hpf, hey2 morpholino (MO) injection was used for these experiments. Crucially, hey2 morphants phenocopied hey2 mutants with respect to multiple parameters (Fig. S1A-M). We observed a substantial increase in the number of late-differentiating CMs being added to the arterial pole in hey2 morphants (95±7.0) compared with wild-type controls (23.4±1.6), as shown by green-only cells at the arterial pole of the 48 hpf heart (Fig. 3B-D, brackets; mean±s.e.m.). The normal timing of termination for myocardial addition has been estimated to be between 36 and 48 hpf (Lazic and Scott, 2011; Jahangiri et al., 2016). Owing to the increase in cell addition observed between 24 and 48 hpf, we wondered whether the window of myocardial accretion was extended temporally in the absence of Hey2. By photoconverting embryos at 48 hpf, with subsequent imaging at 60 hpf, we were further able to more readily evaluate cardiac progenitor addition in a mutant background (Fig. 3E). We noted SHF-mediated accretion in control embryos occurred at a low level, consistent with previous findings (Fig. 3F,J, 11.9±0.7 cells added, mean±s.e.m.; de Pater et al., 2009; Jahangiri et al., 2016). In contrast, a significant increase in cell addition was evident beyond 48 hpf in both hey2 morphants (Fig. 3G,J; 11.9±0.79 cells added for controls and 20.3±1.37 for morphants, mean±s.e.m.) and mutants (Fig. 3H,I,K; 12.63±0.8 in hey2 heterozygotes versus 29.25±0.9 in mutants, mean±s.e.m.). A subpopulation of SHF progenitors gives rise to the outflow tract (OFT), a structure containing both myocardium and smooth muscle at the myocardial-arterial junction (Waldo et al., 2005; Hami et al., 2011; Zhou et al., 2011; Choi et al., 2013; Zeng and Yelon, 2014). As the OFT appeared lengthened in hey2 morphants and mutants (Fig. 3F-I, brackets), we incubated control and hey2 mutant embryos in DAF-2DA, a compound that specifically labels smooth muscle of the OFT (Grimes et al., 2006). We found that loss of hey2 resulted in a significantly longer OFT (Fig. 3L,M,T; 148.42±11.32 versus 223.35±4.03 arbitrary units, mean±s.e.m., controls versus mutants, respectively). This was associated with an increase in OFT cell number (Fig. 3N-S,U; 86.83±5.63 versus 150.67±3.57, mean±s.e.m., control versus mutants, respectively), ruling out potential trivial explanations (altered morphology) for the change in OFT size. Together, these results demonstrated that in the absence of Hey2, there is an extended window of cardiac accretion, which ultimately leads to an expansion in SHF-derived structures.

Our temporal analyses of CM number and late myocardial addition to the heart strongly suggested that CM proliferation alone is not responsible for the cardiac phenotypes observed following loss of hey2 function. Given previous reports of hey2 expression prior to 24-48 hpf (Winkler et al., 2003), we next carried out a detailed analysis of hey2 expression during key stages of cardiogenesis, spanning early cardiac specification to the formation of the linear heart tube. By using double whole-mount RNA in situ hybridization (WISH), we found hey2 transcripts localized anteromedially to those of nkx2.5, mef2cb and myl7 in the anterior lateral plate mesoderm (ALPM) at 16.5 hpf (Fig. 4A-C). Fluorescent RNA in situ hybridization (FISH) analysis allowed a more detailed account of hey2 transcript localization, and clearly showed a unique anterior domain of hey2 expression (red) adjacent to a region of cells co-expressing hey2 and myl7 (yellow cells, white bracket), with myl7⁺/hey2⁻ cardiac mesoderm (green) found more posteriorly (Fig. 4D,D′). At 20 hpf, when the primitive heart is organized into a cone of differentiating cardiac cells (Yelon et al., 1999), hey2 expression was again evident anteromedially to that of myl7 and mef2cb (Fig. 4E-F). We further observed a domain of hey2 expression lateral to the heart cone, in the region of the pharyngeal mesoderm (Fig. 4E,F, white arrowheads). Analysis using FISH further revealed the presence of hey2⁺ cells in the anterior half of the nkx2.5⁺ heart cone (Fig. 4H, red). A 3D reconstruction (Fig. 4H′) revealed a sequential pattern of hey2⁺/nkx2.5⁻ (red) cells lying anterior to a hey2⁺/nkx2.5⁺ population in the anterior heart cone, with the posterior half of the heart cone occupied by hey2⁻/nkx2.5⁺ cells (green). Following formation of the linear heart tube at 24 hpf, hey2 transcripts were detectable both within and extending from the distal portion of the myl7⁺ ventricle, a region occupied by mef2cb-positive cells of the presumptive SHF (Fig. 4I,J,L,L′; (Lazic and Scott, 2011).

To further dissect the expression of hey2 with respect to CPCs, we pursued isolation of key hey2 regulatory regions. Epigenetic analysis of early zebrafish cardiogenesis identified an enhancer, we have termed epicon21, situated 24 kb upstream of hey2 that is conserved from zebrafish to humans, and that overlaps with a recently identified enhancer bound by NKX2.5 in human pluripotent stem cells (Fig. S2A; Anderson et al., 2018; Yuan et al., 2018). Following WISH analysis of Tg(epicon21:EGFP) and myl7 expression at 16.5, 20 and 24 hpf, it was evident that the transgenic closely matched the endogenous hey2 gene expression pattern (Fig. 4C,G,K). Higher resolution analysis revealed Tg(epicon21:EGFP) expression at 20 hpf was restricted to the anteromedial region of the myl7⁻ and Tg(nkx2.5:ZsYellow)⁺ heart cone (Fig. 4G,M-M″), with Tg(epicon21:EGFP) co-expression with Tg(nkx2.5:ZsYellow) in the pharyngeal mesoderm also evident (Fig. 4M″, asterisk; Paffett-Lugassy et al., 2013). A further domain of Tg(epicon21:EGFP) expression was also observed immediately anterior to the heart cone, with these cells having no detectable Tg(nkx2.5:ZsYellow) expression (Fig. 4M″; arrowhead), matching the endogenous position of hey2⁺ progenitors shown by FISH (Fig. 4M″, compare with H′), as well as the position of recently described isl2b-positive SHF cells (Witzel et al., 2017). By 24 hpf, Tg(epicon21:EGFP) expression was evident in the myl7⁻ arterial pole (Fig. 4N-N″; arrowheads and arrows indicate ventricular and atrial regions of the heart tube). Analysis of the expression of Tg(epicon21:EGFP) against that of Tg(myl7:mCherry-RAS) from 48 until 94 hpf demonstrated specificity to the ventricular myocardium and outflow tract (OFT), with an absence of detectable expression in the atrium and atrio-ventricular canal (Fig. S2B-G).

To examine Hey2 protein localization, we further used CRISPR/Cas9 genome editing to place an internal V5 epitope tag into the hey2 locus (Fig. S2H; Burg et al., 2016). Tg(hey2-V5) embryos were viable, demonstrating that this allele was functional. Antibody staining versus V5 at 30 hpf showed V5-Hey2 localization to the posterior ventricular region of the heart tube (Fig. S2I; arrowhead, ventricle; arrow, atrium), with a region of cardiomyocytes (CMs) expressing both myl7:mCherry-RAS and V5-Hey2 (Fig. S2I; asterisk). These expression data replicate both endogenous expression of hey2 (Fig. 4I,L) as well as enhancer expression of hey2 within the heart tube (Fig. 4K,N″). Taken together, our expression analysis revealed subsets of hey2-positive cells between 15 and 20 hpf that: (1) only express hey2; (2) co-express both hey2 with nkx2.5 and/or myl7 in the cardiac cone; and (3) co-express hey2 and nkx2.5 in presumptive cardiac progenitors within the pharyngeal mesoderm. These results, as summarized in Fig. 4O, suggested that hey2 is a prospective early marker of a subset of CPCs, perhaps including the late-differentiating progenitor population, as it is expressed, in part, in a manner consistent with regions shown to contain SHF progenitors (Hami et al., 2011; Guner-Ataman et al., 2013). This suggested that Hey2 may have a previously unappreciated function in early cardiac progenitors.

Successive waves of myocardial differentiation during heart development have been defined in zebrafish, which give rise to discrete regions of the heart (de Pater et al., 2009; Lazic and Scott, 2011). Given the distinct expression of hey2 in a subset of CPCs prior to 24 hpf, we next aimed to map the regions of the heart made from hey2⁺ CPCs. To achieve this, we developed an inducible CreERT2 construct under the control of the hey2 enhancer element (validated above) to allow for temporal control over Cre/Lox-mediated recombination to lineage trace hey2⁺ cells and their descendants within CMs over a range of developmental stages (Fig. 5A). To determine the myocardial fate of hey2⁺ cells, we used the reporter strain myl7:loxp-STOP-loxp-GFP, whereby removal of the stop cassette by Cre switches on GFP expression within myl7⁺ cells. To precisely quantify clones of hey2⁺ cells, we first used a mosaic transgene injection approach. Following injection of hey2 epicon21:CreERT2 plasmid into one-cell stage myl7:loxp-stop-loxp-GFP embryos, 4-hydroxy-tamoxifen (4-HT) was administered during multiple stages of heart development: cardiac specification (16.5 hpf), early and late cone stage (19 and 22 hpf), and finally heart tube stage (24 hpf, Fig. 5B). CMs (GFP⁺) derived from hey2-expressing cells were distributed throughout the heart in varying proportions (Fig. 5C-H). Following tamoxifen addition at 16.5 hpf, hey2⁺ cells contributed to myocardial cells of the atrium, venous pole, inner curvature (IC) and outer curvature (OC) of the heart (Fig. 5C). The contribution to atrial and venous pole CMs from the hey2⁺ population decreased with later addition of tamoxifen at 24 hpf, resulting in labeling of myocardium predominantly in the ventricle proper, OC and OFT (Fig. 5F). A stable Tg(epicon21:CreERT2)^hsc104 line was subsequently established to validate mosaic plasmid injection experiments. When crossed to a Tg(actb2:RSG) constitutive Cre reporter line (Kikuchi et al., 2010), mosaic and inconsistent recombination was evident in the ventricle at 72 hpf following 4-HT administration at 16.5, 19, 22 and 24 hpf (Fig. 5I-L, respectively). In contrast, no atrial CM recombination was evident in these crosses. As the ventricular population (groups 3, 4 and 5 in Fig. 5H) represented the large majority of recombination events across all stages of plasmid-based lineage-tracing experiments, this may represent a bias in hey2 expression in ventricular progenitors at 16.5-24 hpf. Unfortunately, the limited and highly mosaic recombination evident in the stable transgenic line, which may reflect positional effects of transgene site integration, precluded a more meaningful examination. Based on mosaic lineage-tracing results, in conjunction with hey2 expression analysis, the shift in contribution of hey2⁺ cells to CMs of the atrium early, and OC/OFT later, suggested that hey2 expression may dynamically mark subsets of cardiac progenitors over the course of early heart development.

Based on the expression patterns of hey2 and the lineage contributions of hey2⁺ progenitors (Figs 4 and 5), we wanted to further explore a potential link between Hey2 and the late-differentiating SHF progenitor population. We reasoned that hey2 expression should be affected by modulation of signaling pathways that have been previously implicated in regulating the late-differentiating population. Of these pathways, FGF signaling has a pivotal role during CM addition to the arterial pole of the heart tube as well as specifying the cardiac fields during early development (Ilagan et al., 2006; Park et al., 2008; Zeng and Yelon, 2014). Previous reports also identified hey2 as a target of FGF signaling (Feng et al., 2010; Sorrell and Waxman, 2011). Following inhibition of FGF activity between 16.5 hpf and 20 hpf using a well characterized inhibitor, SU5402, we noted the expected reduction in ventricular CM differentiation but normal atrial differentiation, as shown by vmhc and amhc expression, respectively (Fig. S3A,B; Pradhan et al., 2017). Administration of SU5402 during the same timeframe resulted in reduced cardiac hey2 expression within the heart cone (Fig. S3E,F), with a coincident loss of detectable expression of the SHF progenitor marker mef2cb (Fig. S3C,D). Later addition of SU5402, between 19 and 24 hpf, resulted in a loss of cardiac hey2 expression, yet no overt change to cardiac myl7 or neural hey2 expression (Fig. S3G-H; arrowhead, cardiac; asterisk, neural). Moreover, in contrast to FGF signaling, inhibition of BMP and Notch signaling pathways had no discernable effect on hey2 expression (Fig. S3I-J), indicating that between 16.5 and 24 hpf the expression of hey2 is dependent on the activity of FGF signaling, consistent with (at least in part) a SHF progenitor localization.

Given the increased CM number observed in hey2-deficient embryos from as early as 24 hpf, and the expression of hey2 in CPCs, we explored the hypothesis that this may be due to an increase in the cardiac progenitor pool. Using WISH and qRT-PCR, we observed a significant increase in the expression of the FHF and SHF markers nkx2.5, mef2cb and ltbp3 in hey2 mutants when compared with heterozygous controls at 24 hpf (Fig. 6A-I), an observation that was repeated in hey2 morphants (Fig. S4A-J). At 48 hpf, we noted a higher number of Mef2⁺ cells at the arterial pole and in both chambers of hey2 mutant hearts following double staining of myl7:GFP embryos with Mef2 and GFP antibodies (Fig. 6J-M; arterial pole: 18.3±1.25 versus 70±1.08 (control versus hey2 mutants in L); atrium: 40±4.36 versus 63±1.15 (control versus hey2 mutants in M); ventricle: 90.67±2.03 versus 146.67±8.65 (control versus hey2 mutants in M), mean±s.e.m.), suggesting an expansion of Mef2cb⁺ SHF cells (Lazic and Scott, 2011) in the distal region of the ventricle; however, the antibody used is not specific to Mef2cb. As the effect of loss of hey2 on CM number was evident by 24 hpf, we next analyzed embryos at 16.5 hpf for gene expression changes within the ALPM. Interestingly, we observed increased expression of nkx2.5, hey2 and mef2cb in both hey2 mutant (Fig. 6N-V) and morphant (Fig. S4K-Q) embryos. This expansion in gene expression within the early cardiac mesoderm was evident from as early as 12 hpf, as shown by nkx2.5 expression in hey2 MO injected embryos (Fig. S4R,S). To examine the earliest events of cardiac specification, nkx2.5 expression was assayed at 11 hpf. To our surprise, although nkx2.5 expression was undetectable by WISH in control (hey2 heterozygous) embryos, faint nkx2.5 expression was observed in the ALPM of hey2 mutant embryos (Fig. S4T,U, arrowheads). Overall, these results suggest that hey2 plays an early, previously unappreciated role in restricting the size of the early cardiac progenitor population.

The expansion of CM number in hey2 mutants, coupled with the increased expression of CPC-associated genes, led us to examine whether there was increased proliferation of cardiac progenitors prior to heart formation. We therefore first assessed proliferation via use of a 5-ethynyl-2′-deoxyuridine (EdU) incorporation assay (Fig. 7A). Using Tg(nkx2.5:ZsYellow) embryos to label CMs of both early and late progenitor populations (Guner-Ataman et al., 2013), we found a significant increase in the proportion of labeled CMs in hey2 morphant hearts at 35 hpf following an EdU pulse at 16.5 hpf (arrowheads, Fig. 7B-F; 9.83±1.92 controls versus 27.65±3.04 hey2 MO; mean±s.e.m.). In contrast, pulsing with EdU at 24 hpf did not result in a significant difference in CM labeling between hey2 morphants and wild-type embryos at 35 hpf (Fig. 7D-F; 4.7±2.47% EdU⁺ controls versus 10.6±2.12 hey2 MO; mean±s.e.m.). In order to specifically analyze possible functions for hey2 in FHF versus SHF progenitor populations, we repeated our EdU experiments, but measured the proliferation index at 48 hpf in order to allow a delineation of the localization of CMs in the heart and infer their FHF/SHF origin. In both hey2 morphants (Fig. 7G-I) and mutants (Fig. 7J-L), increased labeling of CMs was evident in both atrial and ventricular chambers, as well as in (SHF-derived) OFT myocardium (Fig. 7I, 21.51±12.55% EdU⁺ versus 59.81±2.67% EdU⁺ (atrial); 27.12±10.11% EdU⁺ versus 63.22±3.23% EdU⁺ (ventricular); 7.84±5.84% EdU⁺ versus 46.33±5.36 % EdU⁺ (OFT); mean±s.e.m., controls versus hey2 MO, respectively; and Fig. 7L, 20.6±5.49% EdU⁺ versus 55.82±5.51% EdU⁺ (atrial); 28.44±2.22% EdU⁺ versus 69.05±1.09% EdU⁺ (ventricular); 24.85±3.35% EdU⁺ versus 67.81±2.71% EdU⁺ (OFT); mean±s.e.m., hey2 heterozygotes versus mutants, respectively). These results suggest an early role for Hey2 in establishing the appropriate number of both FHF and SHF cardiac progenitors that will be added to the heart prior to heart tube formation, at least in part by limiting the extent of their proliferation.

Tbx1 has been shown to regulate the proliferation of late-differentiating cardiac progenitors, which reside within pharyngeal mesoderm, an area demarcated by tbx1 expression (Chapman et al., 1996; Buckingham et al., 2005; Hami et al., 2011; Nevis et al., 2013). As our initial analysis suggested that hey2(epicon21:EGFP^hsc28) is co-expressed with Tg(nkx2.5:ZsYellow) within the pharyngeal mesoderm (Fig. 4M″, asterisk), we examined tbx1 expression in hey2 morphant embryos through WISH and found an increase in tbx1 transcript levels at 16.5 hpf (Fig. 7M,N). This increase in tbx1 expression was further evident following qRT-PCR in both hey2 morphant and mutant embryos (Fig. 7O). Analysis at 20 hpf revealed a partial overlap of hey2 and tbx1 expression domains within the pharyngeal mesoderm, with no tbx1 expression detectable within the cardiac cone, consistent with a previous report (Fig. 7P, arrow) (Nevis et al., 2013). Strikingly, in the absence of hey2 function, both hey2 and tbx1 transcripts were significantly upregulated within the pharyngeal mesoderm, along with the expected increase of hey2 within the heart cone itself (Fig. 7Q,R, compare with P). These results suggested a role for Hey2 in restraining CPC proliferation and the size of the progenitor pool adjacent to the heart prior to 24 hpf.

Having established a role for Hey2 in the CPC pool, we finally aimed to determine whether Hey2 acts in CPCs proper, or in the surrounding environment, to affect early stages of cardiac development. Given the technical difficulty of measuring proliferation of transplanted cells, we used SHF contribution to the developing heart as a proxy for CPC behavior. To accomplish this, Tg(myl7:nlsKikGR) donor embryos, either wild type or injected with hey2 MO, were used. Donor cells at 4 hpf (50% epiboly) were transplanted to the margin of wild-type host embryos (Fig. 8A), an approach that has been shown to result in myocardial contribution of donor cells (Stainier et al., 1993; Scott et al., 2007). Transplant embryos were photoconverted at either 24 or 48 hpf and imaged at either 48 or 60 hpf (Fig. 8B,F, respectively). We observed that hey2 morphant donor Tg(myl7:nlsKikGR) cells displayed a significant increase in SHF late contribution (green-only CMs) when compared with early myocardial addition (green+red, or yellow CMs), between 24 and 48 hpf (Fig. 8C-E; percentage of donor CMs that are green only: 22.4±3.99 versus 67.1±6.66, control versus hey2 MO, respectively, mean±s.e.m.). The same result was observed between 48 and 60 hpf, with hey2-deficient myl7:nlsKikGR donors contributing significantly more green-only CMs than controls (Fig. 8F-I; percentage green-only: 15.70±2.33 versus 37.4±2.53, control versus hey2 MO, respectively, mean±s.e.m.). Although the ratio of late versus early differentiated CMs per heart from hey2 donor cells was consistently increased at both 24-48 and 48-60 hpf, an analysis of cell numbers for each category revealed a bias in progenitor populations. From 24-48 hpf, the increased late versus early CM addition ratio observed in hey2 morphants was due to a decreased propensity for donor cells to contribute early to the heart, as evident by a significant decrease in total number of yellow CMs in morphants compared with wild-type embryos (Fig. 8J; 17.07±3.02 versus 3.86±1.26, control versus hey2 MO, respectively, mean±s.e.m.). This was accompanied by an apparent, albeit non-significant, decrease in total CM cell number from hey2 morphant donor embryos when compared with controls during the 24-48 hpf differentiation window (Fig. 8J; 12.42±3.46 total MO cells, 21.85±3.61 total WT cells, mean±s.e.m.). However, when comparing total number of late-differentiating (green-only) CMs contributed, no statistical significance was observed (Fig. 8J; 4.93±0.94 versus 8.57±2.4, control versus hey2 MO, respectively, mean±s.e.m.). In contrast, the higher late- versus early-addition ratio observed between 48-60 hpf was due to a significantly higher amount of late (post-48 hpf) CM addition, with a relatively equivalent amount of early (pre 48-hpf) CM addition, as noted by no significant change in early CM number between control and hey2 morphant embryos (Fig. 8K; 3.75±0.62 versus 11±1.05 green; 23.38±4.98 versus 19.86±3.52 yellow; control versus hey2 MO, respectively, mean±s.e.m.). However, hey2 morphants produced a modestly (significant) larger number of overall CMs between 48 and 60 hpf when compared with controls (Fig. 8K; 31.87±3.49 total MO cells compared with 27.12±5.35 wild type total cells, mean±s.e.m.), further supporting a role for hey2 in regulating cardiomyocyte number. These results suggest, at least in part, a cell-autonomous function for Hey2, in presumptive cardiac progenitors, that affects the size of the cardiac progenitor pool and subsequent addition of CMs to the heart.

This work has uncovered an unappreciated role for the bHLH factor Hey2 in regulating the size of the cardiac progenitor pool, ultimately affecting the extent of the contribution of late-differentiating cardiac progenitors to the zebrafish heart. As shown by fate-mapping (Mjaatvedt et al., 2001; Camp et al., 2012) and lineage-tracing (Cai et al., 2003; Meilhac et al., 2003) approaches, the vertebrate heart is made via subsequent addition of at least two populations of cardiac progenitors. Whether the cardiac progenitor pools that are added early and late to the heart represent molecularly distinct populations remains an unresolved issue. While FHF- and SHF-restricted cells can be identified at early gastrulation stages in mouse by clonal analysis (Lescroart et al., 2010; Devine et al., 2014), this may reflect the result of distinct migratory paths and signaling milieus experienced by cells during gastrulation. A recent study using live imaging of cell-lineage tracing and differentiation status suggests that, in mouse, a discrete temporal lag can be observed between the first and second waves of differentiation that form the heart (Ivanovitch et al., 2017). It is therefore crucial that we gain a greater understanding of how the relative size of cardiac progenitor pool(s), the timing of their differentiation and the extent to which they are added to various components of the heart all affect cardiac development.

It is important to note that the function of hey2 in zebrafish cardiogenesis has been previously addressed in an elegant study (Jia et al., 2007). However, although the overall ‘large heart’ phenotype observed is shared between our hey2^hsc25 and the grl^m145 mutants, our data suggest a role for hey2 prior to 24 hpf, in cardiac progenitors that subsequently impacts cardiac development. In contrast, Jia and colleagues have reported that grl has minimal effect during this time, with grl^m145 mutants having comparable cardiomyocyte number to controls at 24 hpf. This discrepancy may reflect the nature of the hey2 alleles used, with the hey2^hsc25 allele being, we believe, a true null. As the SHF and late myocardial addition in zebrafish had not been described at the time of the previous study, this would also have affected the interpretation of the results. This highlights the fact that Hey2 likely acts at multiple steps of heart development. However, as the morpholino phenotype recapitulates multiple aspects of the CPC phenotype seen in hey2 mutants, we do not believe that the differences in our observations reflect genetic background effects.

Our work has clearly demonstrated a role for Hey2 in restraining proliferation of cardiac progenitors prior to their addition to the heart. Both our mosaic lineage-tracing results (Fig. 5), as well as our analysis of CM number and proliferation (Figs 2 and 7, respectively) support an early role for hey2 in preventing both ventricular and atrial progenitor populations from expanding. Subsequent to this, we observed both increased and extended addition of new CMs to the developing heart tube. The earliest currently available marker of CPCs, nkx2.5, was expanded in its expression in hey2 mutants, and indeed appeared to be induced earlier than in wild-type embryos (Fig. S4T,U). Therefore, the most straightforward explanation for the phenotypes observed in hey2 mutants is that an early increase in CPC specification and proliferation results in a later increase in CM addition to the heart. This is consistent with the early expression of hey2 in the ALPM as early as 10.5 hpf (Winkler et al., 2003). In this model, the expanded CM numbers seen at 24 and 48-72 hpf would be due to an initially greater number of both FHF and SHF progenitors. The extended window of myocardial addition seen in hey2 mutant hearts post-48 hpf would therefore reflect a larger pool of SHF progenitors being available. Subsequent to this, as hey2 expression becomes restricted to the ventricular chamber, its described role in ventricular CM development would be established. Our expression analyses show that at 16.5-20 hpf, hey2 colocalizes with previously identified FHF- and SHF-associated genes, and is in agreement with fate-mapping work that has shown SHF progenitors to reside in an anteromedial position within the ALPM (Hami et al., 2011). Importantly, we and others have shown that hey2 expression within the myocardium is regulated not by a canonical Notch signaling pathway, but through FGF signaling (Targoff et al., 2013; Miao et al., 2018), which is a key regulator of SHF development. This is consistent with the Notch-independent, FGF-mediated expression of Hey2 in other developmental contexts (Doetzlhofer et al., 2009). However, although this result showed coincident loss of SHF-associated progenitors and hey2 expression, we could not preclude that the absence of hey2 simply reflected its association with an FGF-dependent ventricular fate (Targoff et al., 2013).

In disagreement with a model where Hey2 acts primarily in early CPCs, our transplantation data appears to suggest that loss of hey2 does not simply result in a higher likelihood for mutant cells to form CMs at 24 hpf, as overall CM output per transplant was not significantly higher than wild type, and indeed it may have been lower (Fig. 8). In contrast, CM addition post-48 hpf was clearly increased in hey2 morphant transplants. This may suggest roles for hey2 as a cardiomyocyte-limiting factor for both FHF and SHF populations, consistent with functions described for Isl1 in a human iPSC cardiac model (Quaranta et al., 2018). These transplant assays suggest that cardiac contribution is at the very least delayed with the loss of hey2. The observed phenotypes may reflect: (1) a shift in the balance between FHF and SHF progenitor proliferation, favoring the SHF pool; (2) alteration in the timing of cardiac progenitor differentiation; (3) changes in the number of progenitors allocated to the FHF and SHF pools; or (4) a FHF/SHF-agnostic role for Hey2 in cardiac progenitor proliferation and differentiation. It is difficult to reconcile a cell-autonomous function for Hey2 in CPCs with the increased CM number seen at 24 hpf and the stable hey2 enhancer Cre recombination results. In contrast to mosaic lineage tracing based on plasmid injection, the stable transgenic results suggest that hey2 is primarily active in ventricular progenitors. One possibility is that hey2 may have additional, non-autonomous functions, perhaps in the tbx1⁺/nkx2.5⁻ pharyngeal mesoderm, to influence cardiac progenitors. These functions would not have been evident in our current transplant experiments. More elegant single cell transplant or tissue-specific loss-of-function experiments will be required to address this point directly.

Previous reports have highlighted that the cardiac malformations found in animals lacking Hey2 function resemble common human congenital heart defects, including ventricular septal defects, tetralogy of fallot and tricuspid atresia (Donovan et al., 2002). Our data suggest multiple roles for Hey2 in cardiogenesis, first acting to restrain cardiac progenitor proliferation (and potentially affecting diversification of FHF and SHF progenitors) and later affecting the timing of SHF progenitor addition to the developing heart. Hey2 may therefore be an intrinsic regulator of the extent and differentiation of the cardiac progenitor pool, possibly acting as a readout of extrinsic (RA and FGF) niche signals. In this context, it is interesting to note that Hey2 has recently been shown to be a key NKX2.5 target required for CM differentiation of human pluripotent stem cells (Anderson et al., 2018), consistent with the loss of hey2 expression observed in nkx2.5/2.7 mutant zebrafish (Targoff et al., 2013).

In conclusion, hey2 has previously unappreciated roles within cardiogenesis by regulating cardiac progenitor numbers of potentially both heart fields. It remains to be determined whether Hey2 plays a separate role in orchestrating the timing of SHF deployment and subsequent contribution to the development of the vertebrate heart, which will require a conditional mutagenesis approach. Given the apparent cell-autonomous function of Hey2 in CPCs, identifying its transcriptional targets will also be of great interest. Previous studies have reported the downregulation of hey2 expression by RA (Feng et al., 2010), which we have also observed (N.G. and I.C.S., unpublished). Given our findings showing a role for Hey2 in restricting cardiac progenitor proliferation, alongside the known role of epicardial RA signaling in zebrafish heart regeneration (Kikuchi et al., 2011), it will be of great interest to determine whether Hey2 regulates the process of adult cardiac regeneration. Dissecting how Hey2 regulates cardiac development will help address key unanswered questions with respect to the regulatory mechanisms that coordinate the size and differentiation timing of cardiac progenitors to allow for proper heart development to proceed.

Adult AB/TL mixed strain zebrafish (Danio rerio) were maintained as per Canadian Council on Animal Care (CCAC) and The Hospital for Sick Children Animal Services (LAS) guidelines. Zebrafish embryos were grown at 28.5°C in embryo medium as previously described (Westerfield, 1993). The following transgenic lines were used: Tg(myl7:EGFP)^twu34 (Huang et al., 2003), Tg(nkx2.5:ZsYellow)^fb7 (Zhou et al., 2011), Tg(myl7:nlsKikGR)^hsc6 (Lazic and Scott, 2011), Tg(myl7:nlsDsRedExpress)^hsc4 (Lou et al., 2011), Tg(gata1:DsRed)^sd2 (Traver et al., 2003) ,Tg(myl7:mCherry-RAS)^sd21 (Yoruk et al., 2012) and Tg(-9.8bactin2:loxP-DsRed-loxP-EGFP)^s928 (bact2:RSG, (Kikuchi et al., 2010). A myl7:lox-stop-lox-GFP^hsc102 transgenic line was made using standard approaches for this study. The hey2^hsc25 mutant line was generated using CRISPR/Cas9 genome editing technology, as previously described (Jao et al., 2013). Primers TAGGCCAGAAAGAAGCGGAGAG and AAACCTCTCCGCTTCTTTCTGG were annealed together and ligated into the pT7-gRNA vector digested with BsmBI to create sgRNA for hey2. An 8 bp deletion allele (starting at nucleotide 151 within exon2) resulting in a premature stop codon at amino acid 53 was isolated (Fig. 1A). An additional allele harboring a 5 bp deletion (starting at nucleotide 152 with a premature stop codon at amino acid 55) was isolated showing an equivalent phenotype (data not shown). For experiments, hey2 heterozygotes were chosen as the control group, given their preponderance in the crosses used, and the observation that no significant differences were observed between heterozygous and homozygous wild-type embryos.

The hey2 enhancer transgenic epicon21:EGFP^hsc28, which contains an epigenetically conserved open chromatin region (epiCon21, also referred to as aCNE21), was identified from comparative epigenetic analysis of open chromatin enriched in zebrafish mesoderm (Yuan et al., 2018). To generate Tg(epicon21:EGFP)^hsc28 animals (see ‘Plasmid constructs’ section for construct generation), 25 ng of E1b-Tol2-GFP-gateway plasmid carrying the epiCon21 enhancer was injected into wild-type embryos at the one-cell stage with 150 ng Tol2 mRNA. Germline carriers were identified and those siblings showing GFP expression within the ventricles were raised and used for hey2 expression analysis. The Tg(epicon21:CreERT2)^hsc104 transgenic was similarly generated via Tol2-mediated transgenesis. Stable founders were identified via crosses with the bact2:RSG Cre reporter fish. Generation of Tg(hey2V5)^hsc27 was performed as previously described (Burg et al., 2016). nCas9n mRNA was injected at a concentration of 150 pg together with 30 pg of hey2gRNA into the yolks of one-cell embryos. V5 tagging oligo (5′-AGTCATGGCCAGAAAGAAGCGGCAAGCCTATCCCAAACCCTCTGCTGGGCCTGGACTCCACAGGAGAGGGGTAATTCATATT-3′) was diluted to 25 μg/μl and 1 nl was injected into the yolks immediately after the RNA injections.

An Epicon21:EGFP construct was generated by amplifying a 626 bp enhancer sequence, located 24 kb upstream of the hey2 locus (Fig. S2A) from zebrafish genomic DNA using forward primer 5′-CAAATCCCCTGACCTCTGCTTTGAG-3′ and reverse primer 5′-GACACACAGTGACATGTCCTATTGCG-3′. The resulting PCR product was subcloned into the enhancer detection vector E1b-Tol2-GFP-gateway (Addgene 37846; Li et al., 2010). For generation of Epicon21:CreERT2, the GFP sequence in the E1b-Tol2-GFP-gateway vector was replaced with CreERT2.

Mutant genotypes were identified by isolating a 300 bp sequence amplified from genomic DNA of either embryos or adult fin-clips using forward primer 5′-GGGCACCCCTGATCTAAAGT-3′ and reverse primer 5′-AAAAAAGAAGAAGGGACCGGA-3′. PCR products were subsequently digested using restriction enzyme AciI. The subsequent digested DNA products would contain two fragments of 200 bp and 100 bp in wild-type DNA. Heterozygous DNA would contain three fragments consisting of a 300 bp product from the uncut allele as well as a 200 bp and 100 bp product from the wild-type allele, while mutant DNA would contain a single 300 bp uncut fragment.

Morpholino oligos were purchased from Genetools. A morpholino targeting the translation start site (underlined) of hey2 (ATG hey2: 5′-TGCTGTCCTCACAGGGCCGCTTCAT-3′) was used throughout this study. The hey2 morpholino (5′-CGCGCAGGTACAGACACCAAAAACT-3′) previously described (Jia et al., 2007) was used to test specificity, as both morpholinos share no sequence overlap. Injection of 1 ng of ATG hey2 morpholino at the one-cell stage yielded a consistent heart phenotype, which phenocopied hey2 mutants in all assays.

Standard RNA whole-mount in situ hybridization (WISH) was performed as previously described (Thisse and Thisse, 2008). The complete coding sequence of hey2 (ZDB-GENE-000526-1) was PCR amplified (sense, 5′-ATGAAGCGGCCCTGTGAGGACAGC; antisense, 5′-TTAAAACGCTCCCACTTCAGTTCC) and used as a riboprobe template. GFP riboprobe sequence was cloned into pGEM-Teasy and transcribed according to standard techniques. Previously described riboprobes were additionally used: myl7 (ZDB-GENE-991019-3), nkx2.5 (ZDB-GENE-980526-321), mef2cb (ZDB-GENE-040901-7), amhc (ZDB-GENE-031112-1), vmhc (ZDB-GENE-991123-5), tbx1 (ZDB-GENE-030805-5) and ltbp3 (ZDBGENE-060526-130) (Chen and Fishman, 1996; Yelon et al., 1999; Hami et al., 2011; Lazic and Scott, 2011). DIG and fluorescein-labeled probes were made using a RNA Labeling Kit (Roche). hey2 mutants were identified either phenotypically (for stages after 48 hpf) or via genotyping following WISH.

Fluorescent RNA in situ hybridization (FISH) was performed as previously described (Targoff et al., 2013), with minor modifications. Embryo permeabilization using proteinase K was omitted from the protocol. Prehybridization, hybridization and SSC washes were performed at 65°C. Hybridization solution included 10% dextran sulphate, as well as 10 mg heparin and 100 mg tRNA. Detection of digoxygenin-labeled probes were performed using either 1:25 Alexa Fluor 555 Tyramide (Invitrogen) or by deposition of 1:50 TSA plus cyanine 3 solution (Perkin Elmer). Detection of fluorescein-labeled probes was performed by deposition of 1:50 TSA Plus Fluorescein solution (Perkin Elmer).

Quantitative real-time PCR (qRT-PCR) was performed using the Roche LightCycler 480 with Platinum SYBR green master mix used as per the manufacturer's instructions (ThermoFisher Scientific, 11733038). Primers used were as follows: hey2 forward, 5′ GTGGCTCACCTACAACGACA 3′; reverse, 5′ CCAACTTGGCAGATCCCTGT 3′; mef2cb forward, 5′ CAGCCCAGAGTCAAAGGACA; 3′, reverse, 5′ AGGGCACAGCACATATCCTC 3′; nkx2.5 forward, 5′ TCTCTCTTCAGCGAAGACCT 3′; reverse, 5′ CTAGGAAGTTCTTCGCGTAA 3′; gfp forward, 5′ CACTACCAGCAGAACACCCC 3′; reverse, 5’ ATGTGATCGCGCTTCTCGTT 3′; vmhc forward, 5′ ACATAGCCCGTCTTCAGGATTTGG 3′; reverse, 5′ GAGAGAAAGGCAAGCAAGTACTGG 3′. Previously described primers were used for quantification of β-actin (Tang et al., 2007), ltbp3 (Zhou et al., 2011), tbx1 (Zhang et al., 2006) and amhc (Jia et al., 2007) transcript levels. For early stages of development, where morphological features could not be used to distinguish hey2 mutants, embryos were individually prepped for both genomic DNA and mRNA, genotyped, and then pooled as appropriate for cDNA synthesis.

The FGF receptor inhibitor SU5402 (Tocris 3300) was used at a concentration of 10 μM from 16.5 to 20 h post-fertilization (hpf) or from 19 to 24 hpf. BMP and Notch signaling inhibitors dorsomorphin (Tocris 3093) and DAPT (Tocris 2634/10), respectively, were used at a concentration of 10 μM and 50 μM between 16.5 and 20 hpf. All compounds were diluted into 1% DMSO in embryo medium. Vehicle controls were treated with 1% DMSO. Incubations were performed at 28°C. CreERT2-mediated recombination was performed using the active metabolite of tamoxifen, 4-hydroxy-tamoxifen (4-HT), diluted to 10 μM working concentration in embryo medium. At 16.5, 19, 22 and 24 hpf, embryos were dechorionated and transferred to petri dishes containing 4-HT medium or DMSO for controls and kept in the dark at 28°C for 4 h. Following treatment, 4-HT was removed by extensive washes with fresh embryo medium and development proceeded at 28°C until desired stage was reached.

Bright-field images were taken using a Zeiss AXIO Zoom V16. RNA in situ hybridization images were captured using a Leica M205FA microscope with the LAS V6 software package. Confocal microscopy was performed using a Nikon A1R laser-scanning confocal microscope and a 40× water immersion objective. Double-fluorescent RNA in situ hybridization (Fig. 4D,H,L) and DAF-2DA cell labeling (Fig. 3L-S) were captured using a Leica TSC SP8 STED confocal at 100× magnification.

Whole-mount immunofluorescent staining was carried out as previously described (Alexander et al., 1998). Primary antibodies used were: α-MYH6 supernatant at 1:10 (DSHB, S46); α-MHC supernatant at 1:10 (DSHB, MF20); α-MEF-2 (C21) at 1:250 (Santa Cruz, sc-313); α-RCFP at 1:400 (Clontech, 632475); α-DsRed at 1:200 (Clontech, 632496); α-V5 at 1:500 (ThermoFisher Scientific, R960-25) and α-GFP at 1:1000 (Torrey Pines Biolabs). Smooth muscle of the bulbus arteriosus was visualized using the NO indicator DAF- 2DA (Sigma, D2813), as previously described (Grimes et al., 2006). CM nuclei of myl7:nlsDsRedExpress transgenic embryos were counted following immunostaining. Embryonic hearts were dissected and flat-mounted prior to confocal imaging.

Photoconversion on myl7:nlsKikGR embryos was carried out as previously described (Lazic and Scott, 2011) using the UV channel on a Zeiss Axio Zoom V16 microscope. Images were captured using a Nikon A1R laser scanning confocal. For mounting, embryos were fixed in 4% PFA for 20 min and washed three times in PBS. Embryos were agitated in 5% saponin/PBS 0.5%Tx-100 followed by dehydration to 75% glycerol/PBS and left overnight at 4°C. Hearts were dissected and flat mounted prior to imaging.

EdU incorporation assays were performed as previously described (Zeng and Yelon, 2014). Embryos were incubated in 10 mM EdU at 16, 22 and 24 hpf on ice for 30 min. A Click-iT imaging kit (Invitrogen) was used to visualize EdU incorporation. Tg(nkx2.5:ZsYellow) expression was detected using α-RCFP (Clontech 632475) with Alexa Fluor 488-conjugated goat anti-rabbit secondary. The Proliferation Index (Fig. 7) was calculated as the percentage of Nkx2.5:ZsYellow⁺ cells (green) that successfully incorporated EdU (red) in a given experiment.

Transplantation was performed as previously described (Scott et al., 2007). At 4 hpf, 20-30 cells from a myl7:nlsKikGR donor, either uninjected or injected with hey2 MO, were placed into two locations along the margin of wild-type host embryos. Host embryos were subjected to UV illumination at 24 hpf, and subsequently scored at 48 hpf for contribution of donor cells to early- (green and red) or late-differentiating (green-only) CMs.

Excel software was used to perform Student's t-test with two-tailed distribution. Graphs display mean±s.e.m. unless otherwise stated. Box plot graphs were prepared using BoxPlotR web-tool (http://www.boxplot.tyerslab.com).

We thank Angela Morley and Allen Ng for expert fish care and maintenance. Paul Paroutis at the Hospital for Sick Children Imaging Facility provided valuable support for microscopy experiments. We thank previous reviewers of this manuscript for thoughtful suggestions and insights. Neil Chi kindly provided the myl7:EGFP, and Caroline and Geoffrey Burns provided the nkx2.5:ZsYellow transgenic lines.