Strategies for Analyzing Cardiac Phenotypes in the Zebrafish Embryo

Abstract

The molecular mechanisms underlying cardiogenesis are of critical biomedical importance due to the high prevalence of cardiac birth defects. Over the past two decades, the zebrafish has served as a powerful model organism for investigating heart development, facilitated by its powerful combination of optical access to the embryonic heart and plentiful opportunities for genetic analysis. Work in zebrafish has identified numerous factors that are required for various aspects of heart formation, including the specification and differentiation of cardiac progenitor cells, the morphogenesis of the heart tube, cardiac chambers, and atrioventricular canal, and the establishment of proper cardiac function. However, our current roster of regulators of cardiogenesis is by no means complete. It is therefore valuable for ongoing studies to continue pursuit of additional genes and pathways that control the size, shape, and function of the zebrafish heart. An extensive arsenal of techniques is available to distinguish whether particular mutations, morpholinos, or small molecules disrupt specific processes during heart development. In this chapter, we provide a guide to the experimental strategies that are especially effective for the characterization of cardiac phenotypes in the zebrafish embryo.

I. Introduction

Cardiogenesis is an essential aspect of vertebrate embryogenesis. The heart is the first organ to form and function in the early embryo, and organismal growth and survival will depend on the heart’s central role within the circulatory system. Successful heart development involves the specification and differentiation of cardiac lineages, the proper arrangement of cardiac cells into a particular three-dimensional configuration, the functional specialization of the cardiac tissue, and the proper coordination of all of these processes. Unfortunately, errors in any of these steps can cause cardiac birth defects, and such defects are highly prevalent, occurring in as many as 1 in 100 live births and 1 in 10 still births in Western nations (Payne, Johnson, Grant, & Strauss, 1995; Hoffman & Kaplan, 2002; Ransom & Srivastava, 2007; Bruneau, 2008). Therefore, there is high demand for effective experimental strategies that can decipher the molecular mechanisms that control the form and function of the developing heart.

A number of investigators have been drawn to the zebrafish as a model organism for the analysis of mechanisms regulating embryonic heart development (Staudt & Stainier, 2012). Much of the appeal of working with zebrafish embryos derives from their optical qualities: zebrafish embryos are externally fertilized and transparent, allowing noninvasive assessment of cardiac morphology and function. The zebrafish heart is composed of two major chambers, a ventricle and an atrium, both of which contain an inner layer of vascular endocardium and an outer layer of muscular myocardium. Each of these tissues can be easily visualized with cellular resolution, facilitating distinctions between normal and aberrant phenotypes. Furthermore, in contrast to many other organisms, the zebrafish embryo does not require a functional cardiovascular system for its survival (Pelster & Burggren, 1996), allowing the examination of abnormal cardiac development throughout embryogenesis. Importantly, this feature enables the identification of developmental processes that are influenced by cardiac function, such as aspects of morphogenesis that are dependent upon biomechanical forces generated by blood flow (Freund, Goetz, Hill, & Vermot, 2012; Boselli, Freund, & Vermot, 2015).

Thanks to these advantageous characteristics, hundreds of studies have succeeded in utilizing zebrafish embryos to identify specific genes that play key roles during cardiogenesis. Classical forward genetic screens, using chemical mutagens to induce point mutations throughout the genome, have unearthed a large, though still expanding, collection of players that orchestrate the development of the zebrafish heart (Chen et al, 1996; Stainier et al, 1996; Alexander, Stainier, & Yelon, 1998; Warren, Wu, Pinet, & Fishman, 2000; Beis et al, 2005; Chi, Shaw, Jungblut, et al, 2008). As a complementary approach, reverse genetic strategies have interrogated particular candidate genes in order to evaluate their roles during zebrafish heart formation. Numerous studies have relied upon injection of morpholinos to interfere with the translation or splicing of specific genes and thereby test hypotheses regarding their function (Eisen & Smith, 2008). More recently, the advent of multiple methods for genome editing in zebrafish has made it possible to investigate gene function through the targeted introduction of mutations into chosen genes, facilitating new opportunities for large-scale reverse genetic analysis (Kettleborough et al, 2013; Auer & Del Bene, 2014; Gonzales & Yeh, 2014; Varshney, Sood, & Burgess, 2015). Genome editing methods also provide valuable opportunities to validate morpholino-generated phenotypes through examination of mutants, alleviating concerns about morpholino specificity (Kok et al, 2015; Rossi et al, 2015; Stainier, Kontarakis, & Rossi, 2015). Finally, small molecule screens in zebrafish have also contributed to the identification of novel pathways relevant to heart development (Kaufman, White, & Zon, 2009; Peal, Peterson, & Milan, 2010). In addition, compounds known to interfere with specific signaling pathways have been valuable tools for analyzing when and how such pathways influence the embryonic heart.

How can investigators determine whether their mutations, morpholinos, or small molecules cause significant disruptions in cardiogenesis? Often, the first suggestion of a potential cardiac defect comes from detection of pericardial edema or faulty blood circulation by 48–72 hours post fertilization (hpf), if not earlier (Chen et al, 1996; Stainier et al, 1996). This common phenotype can be the consequence of errors in any of several key steps in embryonic heart development (Fig. 1), including the specification or differentiation of cardiac progenitors, the assembly of the initial heart tube, atrioventricular canal, or cardiac chambers, or defects in establishing cardiac contractility or conduction. Numerous experimental strategies have been devised to help distinguish between these possibilities. In a previous primer on this topic (Miura & Yelon, 2011), we highlighted some of the most useful approaches for distinguishing between different causes of anomalies in heart size, shape, and function. Here, we update this guide, by incorporating state-of-the-art techniques as well as recent advances in our understanding of the mechanisms that underlie cardiac growth, morphogenesis, and functional maturation.

Figure 1.

Key steps of embryonic heart development in zebrafish. In the late blastula (5 hpf), myocardial progenitor cells are found in bilateral zones near the embryonic margin. Fate maps indicate that ventricular (red) and atrial (green) progenitor cells are already spatially organized at this stage (Keegan et al., 2004), even though they are not yet distinguished by molecular markers. After gastrulation is complete (12 hpf), the myocardial progenitors (ventricular progenitors, red; atrial progenitors, green) reside in bilateral heart fields within the anterior lateral plate mesoderm (ALPM). Ventricular and atrial progenitor cells retain their spatial organization as the gastrula forms (Schoenebeck et al., 2007), and myocardial differentiation begins within the ALPM. Next, the process of cardiac fusion brings cardiomyocytes to the embryonic midline, where they form a cardiac cone. By this stage (19 hpf), ventricular and atrial cardiomyocytes can be distinguished by the expression of characteristic molecular markers (ventricular myosin heavy chain (vmhc), red; atrial myosin heavy chain (amhc), green) (Berdougo et al., 2003). Continued migration of cardiomyocytes elongates the cardiac cone to assemble the heart tube (shown here at 30 hpf). By 48 hpf, further growth and morphogenesis of the heart tube results in the formation of a morphologically distinct ventricle (red) and atrium (green) that are separated by a constriction at the atrioventricular canal. Adapted from Miura and Yelon (2011).

II. Regulation of Heart Size

Characterization of a cardiac phenotype in zebrafish often begins with an assessment of heart size, since it is straightforward to use a dissecting microscope to visualize the heart in a living embryo (Chen et al, 1996; Stainier et al, 1996). This simple strategy can reveal whether the heart appears to be enlarged, shrunken, or even absent. Such anomalies in heart size could be the consequence of errors in the production of an appropriate number of cardiomyocytes. However, perceptible defects in heart size could also result from aberrant morphology or arrangement of individual cells, rather than problems with the total number of cells. Therefore, when investigating defects in heart size, it is important to start by determining whether such phenotypes reflect defects in cardiomyocyte production. In this section, we focus on techniques that can differentiate between the possible causes of an inappropriate number of cardiomyocytes.

First, to assess the number of cardiomyocytes in a zebrafish embryo, it is valuable to employ transgenes that express nuclear-localized fluorescent reporters in differentiated myocardial cells. Several transgenes of this type have been used to count the number of fluorescent myocardial nuclei, including Tg(myl7:DsRed2-nuc) (Mably, Mohideen, Burns, Chen, & Fishman, 2003) (Fig. 2A,B), Tg(myl7:H2A-mCherry) (Schumacher, Bloomekatz, Garavito-Aguilar, & Yelon, 2013), and Tg(myl7:NLS-KikGR) (Lazic & Scott, 2011). All three of these reporters are effective for analysis at 48 hpf, but only Tg(myl7:H2A-mCherry) and Tg(myl7:NLS-KikGR) are useful at earlier stages, since the DsRed protein requires approximately 24 hours to properly fold and localize after initial myl7 promoter activation (Lepilina et al, 2006). In addition to evaluating the total number of cardiomyocytes, it can be helpful to resolve the relative proportions of atrial and ventricular cells, since chamber-specific perturbations may reflect errors in execution of atrial or ventricular differentiation programs. To quantitate the number of cells in each chamber (Schoenebeck, Keegan, & Yelon, 2007), any of the myl7-driven transgenes mentioned above can be used in combination with an antibody that recognizes an atrial myosin heavy chain (S46; anti-Amhc (Berdougo, Coleman, Lee, Stainier, & Yelon, 2003)).

Figure 2.

RA signaling restricts the production of atrial and ventricular cells through distinct mechanisms. (A,B) Frontal views of hearts at 48 hpf expressing Tg(myl7:DsRed2-nuc) (red). Atria are labeled with the anti-Amhc antibody S46 (green). Treatment of embryos with the RA receptor antagonist BMS189453 (B) results in enlarged cardiac chambers, relative to the chambers of a wild-type heart (A). Counting of cardiomyocytes demonstrated a ~2-fold increase in the numbers of both atrial and ventricular cells in BMS-treated embryos (Waxman et al., 2008). (C,D) Schematics summarize the results of fate maps constructed at the 40% epiboly stage in WT (C) and BMS-treated (D) embryos. The relative positions of regions containing atrial progenitor cells (green) and ventricular progenitor cells (red) are unchanged when RA signaling is reduced. Percentages indicate the frequency of encountering atrial or ventricular progenitors in each region. (E) Table indicates that reduction of RA signaling increased the frequency of encountering ventricular progenitors, but not atrial progenitors, in fate maps constructed at 40% epiboly. Asterisk indicates a statistically significant difference relative to WT (p<0.005). (F) Table indicates that RA signaling increased the average number of labeled progeny produced by atrial progenitors, but not by ventricular progenitors. Asterisk indicates a statistically significant difference relative to WT (p<0.05). Adapted from Waxman et al. (2008).

What could account for the production of an abnormal number of cardiomyocytes in a zebrafish embryo? One possibility is an early defect in the initial specification of cardiac progenitor cells. Fate maps of the late blastula have demonstrated that cardiac progenitors arise from bilateral multipotential zones within the embryonic margin (Fig. 1) (Keegan, Meyer, & Yelon, 2004). During gastrulation, the cardiac progenitors migrate to reside in bilateral fields within the anterior lateral plate mesoderm (ALPM) (Fig. 1) (Keegan et al, 2004; Schoenebeck et al, 2007). Gastrula fate maps have shown that the dimensions of these heart fields correspond well with the expression pattern of hand2 in the ALPM (Schoenebeck et al, 2007). Myocardial differentiation begins around the 14 somite stage, as indicated by the expression of myocardial markers such as myosin light chain 7 (myl7, also known as cmlc2) (Yelon, Horne, & Stainier, 1999). Ventricular and atrial progenitor cells are spatially organized both within the lateral margin and the ALPM (Fig. 1) (Keegan et al, 2004; Schoenebeck et al, 2007). Once they differentiate, ventricular and atrial cardiomyocytes can be distinguished by the expression of characteristic molecular markers, such as ventricular myosin heavy chain (vmhc) and atrial myosin heavy chain (amhc) (Fig. 1) (Yelon et al, 1999; Berdougo et al, 2003). Thus, heightened or diminished numbers of differentiated cardiomyocytes could reflect altered size of the progenitor population, altered fate decisions made within the progenitor pool, a change in the proliferation rate of cardiac progenitors, or defects in the differentiation of the progenitors into cardiomyocytes.

Fate mapping techniques can be employed to distinguish between these possibilities. Construction of a fate map in mutant, morphant, or drug-treated embryos can reveal alterations in the size of the cardiac progenitor pool, the organization of the progenitors, and their productivity in terms of the number of cardiomyocytes generated (Keegan et al, 2004; Keegan, Feldman, Begemann, Ingham, & Yelon, 2005; Schoenebeck et al, 2007; Thomas, Koudijs, van Eeden, Joyner, & Yelon, 2008; Waxman, Keegan, Roberts, Poss, & Yelon, 2008; Hami, Grimes, Tsai, & Kirby, 2011; Rydeen & Waxman, 2014). In these experiments, embryos are injected at the one-cell stage with a photoactivatable lineage tracer that is subsequently triggered in selected cells (Keegan et al, 2004; Schoenebeck et al, 2007). The locations of the labeled cells are noted, with the help of morphological landmarks. Common landmarks include the shield and notochord during epiboly and somitogenesis stages, respectively (Keegan et al, 2004; Schoenebeck et al, 2007). Additionally, the Tg(gsc:gfp) transgene is useful for identifying the dorsal margin prior to shield stage (Keegan et al, 2004). Embryos are then observed at later stages to determine the contributions of labeled cells to cardiac tissues. In addition to determining whether or not the progeny of selected cells become ventricular and/or atrial cardiomyocytes, it is possible to count the number of cardiomyocytes derived from the labeled cells, particularly when using immunohistochemistry to detect the activated lineage tracer. This sensitive technique can resolve differences from the wild-type fate map and thereby indicate the origin of a defect in heart size: cardiac progenitors might be missing from their usual locations or might be found in atypical locations, ventricular and atrial progenitors might be disorganized, or individual progenitors might give rise to fewer or more cardiomyocytes than usual.

Through a combination of cell counting and fate mapping techniques, several studies have identified genetic pathways that regulate pivotal early steps in establishing heart size in the zebrafish embryo (Schoenebeck et al, 2007; Thomas et al, 2008; Waxman et al, 2008; Rydeen & Waxman, 2014). In one example, these experimental strategies were used to define the differential effect of retinoic acid (RA) signaling on ventricular and atrial progenitor cell populations (Waxman et al, 2008). Following up on prior work that demonstrated an essential role for RA in restricting the production of myocardial progenitor cells (Keegan et al, 2005), this study investigated whether RA has comparable effects on both the atrial and ventricular lineages (Waxman et al, 2008). Embryos treated with compounds that inhibit RA signaling, such as the RA receptor antagonist BMS189453 (BMS; (Schulze et al, 2001)), exhibit dysmorphic cardiac chambers, with both the ventricle and atrium appearing enlarged (Fig. 2A,B). Cell counting experiments demonstrated that the aberrant morphology of BMS-treated hearts is a consequence of a significant increase in the number of cells in both the ventricle and the atrium (Waxman et al, 2008). These results suggested that RA signaling could modulate chamber size through either increased specification or enhanced proliferation of cardiac progenitors. Fate mapping in BMS-treated embryos revealed that reduced RA signaling does not significantly alter the relative spatial organization of atrial and ventricular progenitors (Fig. 2C,D). Importantly, BMS-treated embryos do not harbor ventricular progenitors in the region where normally only atrial progenitors are found, or vice versa, suggesting that neither increase in chamber cell number occurs at the expense of the other lineage. Instead, comparisons of the wild-type and BMS-treated fate maps revealed a change in the frequency of encountering ventricular progenitors when RA signaling is reduced: ventricular progenitors were found almost twice as often within the ventricular territory of BMS-treated embryos as they were found in the same territory of wild-type embryos (Fig. 2C–E). In contrast, the frequency of encountering atrial progenitors within the atrial territory did not change in the BMS-treated fate map (Fig. 2C–E). However, BMS significantly increased the number of labeled cardiomyocytes produced by an atrial progenitor, even though the number of cardiomyocytes produced by each ventricular progenitor was unaffected (Fig. 2F). Together, these data, in combination with other results, led to the conclusion that RA signaling restricts the numbers of ventricular and atrial cardiomyocytes through independent mechanisms: RA signaling represses the specification of ventricular progenitors (but not atrial progenitors), and RA signaling limits the productivity of individual atrial progenitors (but not ventricular progenitors) (Waxman et al, 2008).

In addition to early errors in progenitor specification or proliferation, heart size defects could also be the consequence of later errors in the recruitment of late-differentiating cardiomyocytes. It is therefore important for analysis of large or small hearts to include careful examination of the stage at which the number of cardiomyocytes becomes aberrant. Several studies have shown that the zebrafish myocardium forms during two distinct phases of differentiation (de Pater et al, 2009; Hami et al, 2011; Lazic & Scott, 2011; Zhou et al, 2011; Mosimann et al, 2015), either of which could be responsible for heart size phenotypes. The first of these phases creates the initial heart tube, beginning with the differentiation of the ventricle and progressing to the atrium; this phase is complete by 1 day post fertilization (dpf) (de Pater et al, 2009; Lazic & Scott, 2011). Later, between 1 and 2 dpf, a separate population of late-differentiating cardiomyocytes appends to the arterial pole of the heart tube, creating the cardiac outflow tract (OFT) as well as a distal portion of the ventricle (de Pater et al, 2009; Lazic & Scott, 2011; Zhou et al, 2011). Thus, phenotypes in which cardiomyocyte number is abnormal at 2 dpf, but normal at 1 dpf, could reflect errors that occur during the second phase of myocardial differentiation, especially if the cell number defects reside in the ventricle or OFT.

Developmental timing assays can provide a direct assessment of the pace of myocardial differentiation in the zebrafish embryo. One such assay employs a pair of independent reporter transgenes (de Pater et al, 2009): a transgene that expresses GFP in differentiated cardiomyocytes under the control of the myl7 promoter, Tg(myl7:egfp) (Huang, Tu, Hsiao, Hsieh, & Tsai, 2003), and a transgene that expresses DsRed under the control of the same promoter, either Tg(myl7:DsRed2-nuc) (Mably et al, 2003) or Tg(myl7:dsred) (Kikuchi et al, 2010). This assay takes advantage of the difference in GFP and DsRed protein folding kinetics to distinguish early-differentiating and late-differentiating populations of cells (Lepilina et al, 2006; de Pater et al, 2009). Since DsRed requires more time to mature and fluoresce than does GFP, early-differentiating cardiomyocytes express both GFP and DsRed at timepoints when the late-differentiating cells express only GFP. Thus, DAPI-stained double transgenic embryos can facilitate counting of both early-differentiating and late-differentiating cardiomyocyte nuclei (de Pater et al, 2009). A second, complementary strategy uses a reporter transgene, such as Tg(myl7:kaede) (de Pater et al, 2009) or Tg(myl7:NLS-KikGR) (Lazic & Scott, 2011), that expresses a green-to-red photoconvertible fluorescent protein under the control of the myl7 promoter. Prior to photoconversion, differentiated cardiomyocytes in these transgenic embryos display green fluorescence. Upon exposure to UV light, the photoconvertible protein converts from its green form to its red form, such that differentiated cells will then display red fluorescence. At subsequent timepoints, any cells that initiated differentiation after the time of photoconversion will exhibit green, but not red, fluorescence. Thus, this assay can distinguish early-differentiating and late-differentiating cardiomyocytes and thereby reveal defects in the accretion of late-differentiating cells at the arterial pole.

Defects at the arterial pole could reflect ineffective execution of the second phase of myocardial differentiation, or they could reflect problems with the progenitor population that gives rise to the late-differentiating cardiomyocytes. This progenitor population is often referred to as the “second heart field” (SHF), whereas the term “first heart field” is used in reference to the progenitors that give rise to the initial heart tube (de Pater et al, 2009; Hami et al, 2011; Lazic & Scott, 2011; Zhou et al, 2011). A few markers have been used to examine the status of the SHF progenitors as they reside adjacent to the arterial pole of the heart tube: some of these, like the genes encoding the TGF-β binding protein Ltpb3 and the transcription factor Mef2cb, are expressed in progenitor cells but become extinguished as differentiation proceeds (Lazic & Scott, 2011; Zhou et al, 2011) (Fig. 3D), whereas the expression of others, like the gene encoding the transcription factor Nkx2.5, persists in the differentiated myocardium (Zhou et al, 2011; Guner-Ataman et al, 2013) (Fig. 3G). Use of these markers to analyze the numbers and locations of SHF progenitor cells can provide a helpful indication of whether a deficiency or surplus of progenitors could underlie an aberrant number of late-differentiating cardiomyocytes.

Figure 3.

The adhesion molecule Cadm4 limits OFT size by restricting SHF progenitor production. (A–C) Hearts at 48 hpf, stained with MF20 (red) and S46 (green) antibodies (Alexander et al., 1998) to visualize the OFT (red, bracket), ventricle (red), and atrium (yellow). Compared to controls (A) with normal OFT length, cadm4 morphants (B) exhibit a significantly elongated OFT, and embryos injected with cadm4 mRNA (C) have a significantly reduced OFT. (D–F) In situ hybridization shows expression of mef2cb (green) in SHF progenitor cells adjacent to the differentiated myocardium at the arterial pole of the heart tube (MF20, red); dorsal views at 30 hpf. In comparison with controls (D), the SHF progenitor population (arrowheads) is expanded in cadm4 morphants (E) and reduced in embryos overexpressing cadm4 (F). (G–I) Dorsal views show Tg(nkx2.5:ZsYellow) (Zhou et al., 2011) expression at 24 hpf; MF20 (red) marks differentiated cardiomyocytes. Controls (G) exhibit a normal number of SHF progenitors (arrows) in the region proximal to the arterial pole. cadm4 morphants (H) exhibit a significant surplus of progenitors in this region, and embryos overexpressing cadm4 (I) exhibit a significant reduction of progenitors. Adapted from Zeng and Yelon (2014).

Several studies have combined the use of developmental timing assays and SHF progenitor markers in order to analyze the origins of OFT size defects (de Pater et al, 2009; Lazic & Scott, 2011; Zhou et al, 2011; Guner-Ataman et al, 2013; Nevis et al, 2013; Zeng & Yelon, 2014; Mosimann et al, 2015). For example, these strategies helped to illuminate the role of the adhesion molecule Cadm4 in regulating the formation of SHF progenitor cells (Zeng & Yelon, 2014). Alteration of cadm4 gene function indicated its potent repressive influence on the size of the OFT: morpholino (MO)-mediated knockdown of cadm4 causes dramatic OFT expansion, whereas overexpression of cadm4 results in a greatly diminished OFT (Fig. 3A–C). Developmental timing assays revealed that the OFT size defects caused by cadm4 loss-of-function and gain-of-function correspond to increased and decreased numbers of late-differentiating cells, respectively (Zeng & Yelon, 2014). Furthermore, evaluation of SHF progenitor markers, including mef2cb and the reporter transgene Tg(nkx2.5:ZsYellow) (Zhou et al, 2011), demonstrated that the SHF progenitor population is expanded in cadm4 morphants and reduced in embryos overexpressing cadm4 (Fig. 3D–I). The proportional effects of cadm4 loss-of-function and gain-of-function on both the progenitor population and the late-differentiating cardiomyocytes, together with other complementary data, supported a model in which Cadm4 limits OFT size by restricting SHF progenitor cell production (Zeng & Yelon, 2014).

In addition to defects in the specification and differentiation of FHF or SHF progenitor cells, alterations in the numbers of cardiomyocytes could also be a consequence of changes in identity that occur subsequent to differentiation. For example, recent work has shown that mutation of the transcription factor genes nkx2.5 and nkx2.7 results in an expansion of atrial cardiomyocytes at the expense of their ventricular counterparts (Targoff et al, 2013). Cell counting assays demonstrated that embryos lacking nkx gene function have relatively normal numbers of cardiomyocytes at 1 dpf, whereas they seem to lose ventricular cells and gain atrial cells between 1 and 2 dpf (Targoff et al, 2013). The correspondence between the numbers of ventricular cells lost and atrial cells gained suggested the possibility of cells changing their chamber assignment during the 1 to 2 dpf timeframe. Consistent with this, cell labeling experiments suggested that ventricular cardiomyocytes can transform into atrial cardiomyocytes in nkx-deficient embryos: after localized photoconversion of Tg(myl7:kaede)-expressing cells in the ventricle at 1 dpf, red fluorescent cells were detected in the atrium at 2 dpf. Together, these experiments suggested a pivotal role for Nkx transcription factors in maintaining ventricular chamber identity (Targoff et al, 2013).

Altogether, a substantial array of available techniques can facilitate characterization of a variety of types of heart size defects, allowing insight into factors influencing specification, differentiation, and cardiac chamber identity. Ongoing efforts in the field are devising a broader toolbox for genetically-induced fate mapping of specific cardiac lineages. In addition to the development of transgenes that can be selectively expressed in either the FHF or SHF populations (Zhou et al, 2011; Guner-Ataman et al, 2013; Mosimann et al, 2015), new tools are available that can track the contributions of neural crest cells to the embryonic heart (Cavanaugh, Huang, & Chen, 2015). Separate sets of new transgenes are facilitating novel methods for following heart growth. One type of tool employs fluorescent cell cycle indicators to monitor cardiomyocyte proliferation in live embryos (Choi et al, 2013), and another strategy uses recombination-based multicolored labeling for high-resolution clonal analysis that can reveal the dynamics of proliferation over time (Gupta & Poss, 2012). As our techniques for tracking cell destiny and growth become increasingly sophisticated, our understanding of the regulation of heart size will continue to expand.

III. Regulation of Cardiac Morphology

The specific shape of the mature zebrafish heart is critical for its ability to drive effective circulation. This final structure is formed through a series of morphogenetic steps, beginning with heart tube assembly and followed by cardiac chamber emergence (Fig. 1), as well as definition of the atrioventricular canal and growth of the ventricular trabeculae. Production of an appropriate number of cardiomyocytes is an important prerequisite for establishing normal cardiac morphology, since a shortage or surplus of cells can lead to gross malformations of the heart (e.g. Fig. 2B; (Waxman et al, 2008)). However, normal cell number is not sufficient to insure proper morphogenesis. Defects in heart shape can have a variety of origins that are unrelated to specification and differentiation, including aberrant cardiomyocyte movements, failure to execute normal cell shape changes, ineffective patterning of the atrioventricular canal, or inadequate progression of chamber maturation. In this section, we cover a series of experimental strategies for determining the possible causes of a misshapen heart in a zebrafish embryo.

The assembly of the heart tube begins with the movement of bilateral populations of cardiomyocytes toward the embryonic midline, where they meet and merge through a process called cardiac fusion (Fig. 1; (Glickman Holtzman, Schoenebeck, Tsai, & Yelon, 2007; Bakkers, Verhoeven, & Abdelilah-Seyfried, 2009)). The differentiating cardiomyocytes move collectively, as cohesive epithelial sheets, and the initial contact between these sheets connects posterior subsets of contralateral cells, followed by interactions between anterior subsets of cells (Fig. 4A; (Trinh & Stainier, 2004; Glickman Holtzman et al, 2007)). Together, these connections create a ring of cardiomyocytes that provides a topological foundation for the specific dimensions of the heart tube (Fig. 4B).

Figure 4.

Loss of angular movement in clo mutant embryos. Selected images from time-lapses of cardiac fusion in wild-type (A,B) and clo mutant (C,D) embryos expressing Tg(myl7:egfp) (green). Dorsal views, anterior up. Arrows indicate paths traveled by individual cardiomyocytes from their starting positions (A,C) until their ending positions (B,D). Red arrows indicate medial movement, and yellow arrows indicate angular movement; methods for quantitative analysis of direction of movement are described by Glickman Holtzman et al. (2007). (A,B) In wild-type embryos, cardiac fusion involves critical angular movements by anterior and posterior subsets of cardiomyocytes. (C,D) In clo mutant embryos, anterior and posterior cardiomyocytes exhibit medial movement rather than angular movement, resulting in a dysmorphic ring of cardiomyocytes. Adapted from Glickman Holtzman et al. (2007).

Disruption of cardiac fusion can lead to cardia bifida, a dramatic condition in which an embryo exhibits a pair of separate hearts in bilateral positions, instead of a single heart at the midline. In zebrafish mutants with cardia bifida, cardiac fusion fails because cardiomyocytes do not reach the midline; this phenotype is typically evident in the aberrant expression patterns of myocardial markers, such as myl7, between 18 and 22 hpf (e.g. (Kikuchi et al, 2000; Kupperman, An, Osborne, Waldron, & Stainier, 2000; D’Amico, Scott, Jungblut, & Stainier, 2007; Osborne et al, 2008)). Mutations that cause cardia bifida have revealed that the extracellular environment has a profound influence on cardiac fusion. For instance, mutations disrupting endoderm specification or integrity cause cardia bifida, indicating that interactions between the myocardium and the adjacent endoderm are crucial for promoting cardiomyocyte movement (e.g. (Kikuchi et al, 2000; Kupperman et al, 2000; Kikuchi et al, 2001; Osborne et al, 2008; Kawahara et al, 2009; Ye & Lin, 2013; Ye, Xie, Hu, & Lin, 2015)). In addition to the endoderm, the composition of the extracellular matrix (ECM) is important for the execution of cardiac fusion, since mutations causing either diminished or excessive ECM deposition can hinder cardiomyocyte motility (Trinh & Stainier, 2004; Trinh, Yelon, & Stainier, 2005; Arrington & Yost, 2009; Garavito-Aguilar, Riley, & Yelon, 2010). Therefore, when investigating new cardia bifida phenotypes, it is important to examine the specification and morphogenesis of the anterior endoderm, using appropriate markers (e.g. Tg(-0.7her5:egfp), axial, sox17 (Kupperman et al, 2000; Osborne et al, 2008; Kawahara et al, 2009; Ye & Lin, 2013; Ye et al, 2015)), as well as the deposition and composition of the ECM, using immunofluorescent detection of relevant components (e.g. Fibronectin and Laminin (Trinh & Stainier, 2004; Arrington & Yost, 2009; Garavito-Aguilar et al, 2010)).

Subtle errors in cardiac fusion do not necessarily result in a cardia bifida phenotype, but can still create a misshapen heart tube of abnormal length or width. To analyze the precise nature of a cardiac fusion defect, it is valuable to conduct time-lapse analysis, using confocal microscopy to track individual cardiomyocytes expressing the transgene Tg(myl7:egfp) (Huang et al, 2003). Using this approach, it is feasible to evaluate multiple parameters of cell movement, including direction of cell displacement, rate of cell displacement, and straightness of the migratory path (Glickman Holtzman et al, 2007; Fish et al, 2011). Thus, comparison of cardiomyocyte trajectories in wild-type and mutant embryos can distinguish whether a defect in cardiac fusion reflects inappropriate direction, inadequate speed, or other types of aberrant movement. For example, time-lapse analysis in cloche (clo) mutant embryos suggested that the abnormally short and wide clo mutant heart tube is a result of aberrant cell movements during cardiac fusion. During wild-type cardiac fusion, angular movements of the most anterior and posterior cardiomyocytes play a key role in creating the stereotypical shape of the myocardial ring at the midline (Fig. 4A,B; (Glickman Holtzman et al, 2007)). In contrast, cardiomyocytes in clo mutants only move medially, instead of at an angle, and thereby create a dysmorphic ring that is atypically oblong, with an abnormally small inner circumference (Fig. 4C,D; (Glickman Holtzman et al, 2007)). Interestingly, clo mutants lack the endocardium, the specialized vascular tissue that creates an endothelial lining within the myocardial heart tube (Stainier, Weinstein, Detrich, Zon, & Fishman, 1995). In wild-type embryos, the endocardial precursors can be found at the midline, where they become surrounded by the myocardial ring (Glickman Holtzman et al, 2007; Fish et al, 2011). Thus, the aberrant cardiomyocyte movements that occur in the absence of the endocardium in clo mutants suggest that myocardial-endocardial interactions play an important role in directing cardiac fusion (Glickman Holtzman et al, 2007).

After cardiac fusion is complete, heart tube assembly continues, as the ring of cardiomyocytes undergoes an additional collective migration directed toward the left side of the embryo (Fig. 5A; (Baker, Holtzman, & Burdine, 2008; de Campos-Baptista, Holtzman, Yelon, & Schier, 2008; Smith et al, 2008; Lenhart, Holtzman, Williams, & Burdine, 2013; Veerkamp et al, 2013)). Together with asymmetric involution of the right side of the myocardial ring (Rohr, Otten, & Abdelilah-Seyfried, 2008), these cell movements create an elongated heart tube with its atrial end pointed leftward (Fig. 1). Tube elongation relies upon the differential migration of cardiomyocytes from particular regions of the myocardial ring: cells in the posterior region exhibit a greater overall leftward displacement compared to cells in the anterior region, and cells on the left side of the ring move at a greater velocity than cells on the right side (Fig. 5A; (Baker et al, 2008; de Campos-Baptista et al, 2008; Smith et al, 2008; Lenhart et al, 2013; Veerkamp et al, 2013)). Thus, heart tube morphology and position depend upon the coordination of carefully choreographed patterns of cell movement.

Figure 5.

Trajectories of cardiomyocyte movement during heart tube elongation. Images depict the initial frame of time-lapses tracking heart tube elongation in embryos expressing Tg(myl7:egfp) (green). Dorsal views, anterior up. Arrows indicate paths traveled by individual cardiomyocytes as they leave these starting positions and proceed with heart tube elongation. (A) In wild-type embryos, patterns of cell movement exhibit left-right asymmetry: although all tracked cells move toward the anterior and left, the cells on the left (yellow tracks) move considerably faster than the cells in the center (white tracks) or the cells on the right (red tracks). (B) In ntl morphants, the left-right asymmetry of cell behavior is lost: all cells move toward the anterior, but not toward the left, and the velocity of the cells on the left (yellow tracks) is similar to that of the cells on the right (red tracks). (C) Similarly, in spaw morphants, cells move toward the anterior, rather than toward the left. However, cells move more slowly in spaw morphants, compared to the velocities observed in wild-type embryos or ntl morphants. Adapted from Lenhart et al. (2013).

To distinguish whether an inappropriately assembled heart tube is the result of problems during cardiac fusion or tube elongation, it is helpful to use myocardial markers, such as myl7, vmhc, and amhc (Yelon et al, 1999; Berdougo et al, 2003), at a series of timepoints to determine whether the first signs of abnormal morphology occur before or after the formation of the myocardial ring. Additional information about the cellular mechanisms underlying tube elongation defects can come from examination of markers that exhibit apicobasal polarity in cardiomyocytes, such as aPKC, β-catenin, and ZO-1 (Trinh & Stainier, 2004; Rohr, Bit-Avragim, & Abdelilah-Seyfried, 2006). Apicobasal polarity is disrupted by mutations in heart and soul (prcki) and nagie oko (mpp5) (Horne-Badovinac et al, 2001; Peterson, Mably, Chen, & Fishman, 2001; Rohr et al, 2006), both of which block heart tube elongation. Finally, to determine the specific type of cell movement defects disrupting heart tube elongation, it is valuable to use time-lapse microscopy to track the cells as they migrate. As is the case for cardiac fusion, the most popular transgene for time-lapse imaging of tube elongation has been Tg(myl7:egfp) (Huang et al, 2003).

Using a time-lapse strategy, a number of studies have identified molecular directors of asymmetric cell movement during heart tube elongation (Baker et al, 2008; de Campos-Baptista et al, 2008; Smith et al, 2008; Lenhart et al, 2013; Veerkamp et al, 2013). In one example, time-lapse analysis has illustrated the way in which left-sided activation of Nodal signaling is crucial for controlling both the direction and rate of cardiomyocyte movement (Lenhart et al, 2013). In wild-type embryos, left-sided expression of southpaw (spaw), which encodes a Nodal ligand (Long, Ahmad, & Rebagliati, 2003), seems to drive the leftward movement of cardiomyocytes, with cells on the left side moving faster than cells on the right (Fig. 5A; (Lenhart et al, 2013)). If spaw is expressed symmetrically on both the left and right sides of the embryo, as in ntl morphants, then the differential movement of cardiomyocytes toward the left is lost (Fig. 5B; (Lenhart et al, 2013)). In these embryos, cells on the left and right sides behave similarly, with both sets of cells moving toward the anterior, instead of leftward, at comparable velocities. Ultimately, this creates a heart tube that extends anteriorly, rather than toward the left side of the embryo. A leftward bias of cardiomyocyte movement is also missing in embryos lacking spaw function, yet, in this case, all cells move more slowly (Fig. 5C; (Lenhart et al, 2013)). These data point toward a role of spaw in stimulating the rate of cardiomyocyte movement: the Spaw signal induces faster cardiomyocyte movements, and asymmetric localization of Nodal signaling can thereby direct the leftward elongation of the heart tube (Lenhart et al, 2013). Further studies have shown complex interactions between Nodal and Bmp signaling in regulating asymmetric aspects of cardiomyocyte movement (Smith, Noel, et al, 2011; Lenhart et al, 2013; Veerkamp et al, 2013). The downstream molecular mechanisms through which these signaling pathways regulate the directional migration of cardiomyocytes are exciting topics for ongoing research.

Although many types of heart shape defects originate during heart tube assembly, dysmorphic phenotypes can also appear during the stages of cardiac chamber emergence. Between 24 and 48 hpf, the remodeling of the heart tube into a chambered heart involves the formation of characteristic chamber curvatures (Fig. 1): a bulging curvature designated as the outer curvature (OC) and a recessed curvature called the inner curvature (IC) (Auman et al, 2007). Regionally restricted changes in cardiomyocyte size and shape are thought to underlie the morphological differences between the chamber curvatures. In the linear heart tube, ventricular cardiomyocytes all appear relatively small and round. However, during chamber emergence, cells in the ventricular OC become enlarged and elongated, thereby increasing the surface area of the chamber wall and causing it to bulge outwards, while cells in the ventricular IC retain a rounded morphology and exhibit a smaller increase in size (Fig. 6A,B; (Auman et al, 2007)). Errors in the execution of these cell shape changes can result in abnormal ventricular shape: for example, if cells fail to expand properly, the ventricle can become abnormally compact and narrow, whereas excessive expansion can create a dilated and round ventricle (Fig. 6C,D; (Auman et al, 2007)).

Figure 6.

Regional patterns of cell shape change underlie the emergence of chamber curvatures and are regulated by cardiac function. Confocal projections of live Tg(myl7:egfp)-expressing hearts that also exhibit mosaic expression of Tg(myl7:dsredt4) (red) at 52 hpf. Arrows point to representative cells expressing both dsredt4 and egfp. Ventricular cells in the OC (A) are flattened and elongated, whereas cells in the IC (B) are relatively cuboidal. When blood flow is reduced, as in wea (amhc) mutants (C), OC cells are generally smaller and less elongated. When ventricular contractility is eliminated, as in haf (vmhc) mutants (D), OC cells are considerably enlarged and elongated. Adapted from Auman et al. (2007).

To analyze whether a dysmorphic cardiac chamber is the result of defects in cellular morphologies, it is valuable to quantify cell size and shape using any of a variety of tools for labeling cardiomyocyte boundaries. Immunofluorescence with phalloidin or an anti-Dm-grasp antibody can outline cardiomyocytes in fixed tissue, and transgenes that highlight cardiomyocyte membranes, such as Tg(myl7:egfp-hsras) or Tg(myl7:mkate-caax), are effective in live embryos (Auman et al, 2007; Chi, Shaw, Jungblut, et al, 2008; Deacon et al, 2010; Lin, Swinburne, & Yelon, 2012). Size is typically expressed in terms of cardiomyocyte surface area, and shape is typically evaluated using a circularity index that quantifies deviation from a perfectly circular morphology. For example, quantitative morphometric analysis in connexin46 mutants has shown that their misshapen ventricle contains misshapen cardiomyocytes, with OC cells that are relatively round, having failed to undergo the typical elongation seen in wild-type embryos (Chi et al, 2010). Together with additional data, these results point to an important role of cardiac conduction, mediated by Connexin-containing gap junctions, in regulating cardiac chamber emergence (Chi et al, 2010).

When evaluating the origins of chamber emergence defects, it is also important to look beyond the cell shape changes in the myocardium, since altered myocardial morphology may be a secondary response to a primary defect in another tissue. Notably, recent work has indicated that increased endocardial proliferation, together with patterned changes in endocardial cell shape, also accompanies the emergence of ventricular curvatures (Dietrich, Lombardo, Veerkamp, Priller, & Abdelilah-Seyfried, 2014), suggesting that endocardial growth could drive the enlargement and elongation of the myocardial OC. Methods for counting and outlining endocardial cells, employing transgenes and antibodies parallel to those used for the myocardium (Dietrich et al, 2014), are therefore helpful to interrogate whether an endocardial deficiency could be responsible for a misshapen cardiac chamber. Moreover, both the endocardium and the myocardium have been shown to be exquisitely responsive to cardiac function during the process of chamber emergence (Auman et al, 2007; Lin et al, 2012; Dietrich et al, 2014). Blood flow promotes endocardial proliferation (Dietrich et al, 2014) and also encourages cardiomyocytes to enlarge and elongate (Fig. 6C; (Auman et al, 2007; Lin et al, 2012)). At the same time, cardiomyocyte contractility seems to limit the degree of myocardial cell shape change (Fig. 6D; (Auman et al, 2007)), suggesting that the acquisition of normal cardiomyocyte morphology requires a balance between external physical forces, such as blood flow, and internal physical forces, such as contractility. Thus, it is worthwhile to consider whether an observed dysmorphic chamber could originate with a primary defect in cardiac function. (The following section will highlight a number of methods for analysis of embryonic heart function.)

The expansive ballooning of the cardiac chambers between 24 and 48 hpf contrasts sharply with the morphological constriction that occurs at the junction between chambers, the atrioventricular canal (AVC), during the same timeframe. Proper differentiation of the AVC is critical as it is the future site of the atrioventricular valve, as well as being essential for the conduction delay that leads to sequential contractions of the atrium and ventricle (Beis et al, 2005; Milan, Giokas, Serluca, Peterson, & MacRae, 2006; Chi, Shaw, Jungblut, et al, 2008; Peal, Lynch, & Milan, 2011). During AVC differentiation, myocardial cells at this junction contract their apical surfaces, causing the region to pinch inward, toward the lumen of the tube (Beis et al, 2005). In addition, AVC cells secrete more ECM than their neighbors, increasing the distance between the myocardium and endocardium (Moorman & Christoffels, 2003; Armstrong & Bischoff, 2004; Lagendijk, Goumans, Burkhard, & Bakkers, 2011; Patra et al, 2011). Finally, endocardial cells transition from a squamous to a more columnar morphology, which further constrains the lumen of the heart tube at the AVC (Beis et al, 2005; Scherz, Huisken, Sahai-Hernandez, & Stainier, 2008). By 48 hpf, the combination of these three actions yields visible thickenings of endocardium within the AVC, referred to as cardiac cushions. The cushions then remodel into valve leaflets that emerge by 72 hpf, and retrograde flow of erythrocytes from the ventricle back into the atrium is completely blocked by the primitive leaflets by 96 hpf (Scherz et al, 2008).

Errors in AVC formation can result in several morphological abnormalities that are readily detectable with a dissecting microscope at 48 hpf (Chen et al, 1996; Stainier et al, 1996). One common consequence of aberrant AVC differentiation is the absence of a notable constriction between the atrium and the ventricle. In addition, since the cardiac cushions act as a partial barrier against retrograde blood flow, AVC formation errors often lead to obvious regurgitation or ‘toggling’ of blood from the ventricle back into the atrium. While these morphological cues point to an abnormal AVC, more in-depth analysis is necessary to reveal how AVC patterning has gone awry.

There are several molecular markers that can provide insight into the status of AVC differentiation in the developing zebrafish heart, and these are often employed to distinguish whether phenotypes represent failed AVC development or the ectopic presence of AVC characteristics. For example, two commonly used markers, bmp4 and tbx2b, are normally expressed at high levels within the AVC myocardium (Fig. 7B) and are absent from this junction in embryos that lack AVC specification (Chi, Shaw, De Val, et al, 2008; Verhoeven, Haase, Christoffels, Weidinger, & Bakkers, 2011). Examples of such a phenotype include the slipjig mutation, which disrupts the gene encoding the transcription factor Foxn4 (Chi, Shaw, De Val, et al, 2008), or the elimination of Wnt pathway activity through overexpression of the negative regulator axin1 (Verhoeven et al, 2011). These results have led to a model in which Foxn4 and the canonical Wnt pathway act as upstream inducers of AVC differentiation. AVC endocardium is marked by the elevated expression of notch1b (Fig. 7A), which, along with bmp4 and tbx2b, is expressed ectopically within the cardiac chambers when AVC differentiation is not properly confined. For example, mutation of tmem2, which encodes a transmembrane protein of unknown function, leads to abnormal expansion of AVC differentiation markers (Fig. 7; (Smith, Lagendijk, et al, 2011; Totong et al, 2011)), as does mutation of apc, which causes constitutive Wnt pathway activity (Hurlstone et al, 2003; Verhoeven et al, 2011). These and other complementary analyses of the expression patterns of AVC differentiation markers have revealed that a variety of factors collaborate in an elaborate patterning network that restricts cardiac cushion formation to the AVC.

Figure 7.

Mutation of tmem2 causes ectopic AVC differentiation. (A–C, E–G) Whole-mount in situ hybridization indicates expression of AVC differentiation markers at 48 hpf. White dots outline the atrium (A) and ventricle (V). (A–C) Frontal views of wild-type embryos depict the restricted expression of the AVC endocardial marker notch1b (A) and the AVC myocardial markers bmp4 (B) and versican (C). (E–G) In contrast, tmem2 mutant embryos exhibit expanded expression of all three markers. (D,H) Immunofluorescence reveals localization of the endocardial cushion marker Dm-grasp at 48 hpf. (D) In wild-type embryos, Dm-grasp is found throughout the myocardium (outer layer), but is only found in the endocardium at the AVC (arrowheads). (H) In tmem2 mutants, Dm-grasp is found at the AVC (arrowheads), as well as throughout the ventricular endocardium. Adapted from Totong et al. (2011).

One caveat of using AVC differentiation markers is that some of these genes, like bmp4 and notch1b, are initially expressed throughout the ventricle before becoming restricted to the AVC around 36 hpf (Walsh & Stainier, 2001). This can lead to confusion, since a phenotype featuring ectopic bmp4 and notch1b expression could indicate either ectopic AVC differentiation or a delay in cardiac chamber maturation. In this regard, the cell adhesion molecule Dm-grasp, which is found throughout the myocardium but is only present within the endocardium in the cardiac cushions (Fig. 7D; (Beis et al, 2005)), is a particularly useful marker for analysis of the AVC. Since Dm-grasp does not appear in the AVC endocardium until the cushions begin to form (Beis et al, 2005), ectopic Dm-grasp localization (Fig. 7H) is unlikely to be an artifact of developmental delay or to represent failed chamber maturation. Ongoing studies of AVC development will benefit from the incorporation of additional markers that represent key stages of AVC differentiation. Transgenic indicators of signaling pathway activity, such as Tg(7xTCF-Xla.Sia:GFP) (Moro et al, 2012)and Tg(Bre: GFP) (C. Alexander et al, 2011), will be especially valuable, as they can quantitatively report on the location and magnitude of Wnt and Bmp signaling, respectively, within the AVC.

Acquisition of the final shape of the zebrafish heart requires additional steps of maturation beyond cardiac chamber emergence and AVC formation. Notably, a series of recent studies have investigated the mechanisms that drive the structural elaboration of the heart through the process of trabeculation (Liu et al, 2010; Peshkovsky, Totong, & Yelon, 2011; Gupta & Poss, 2012; Staudt et al, 2014; Samsa et al, 2015). Trabeculae are muscular projections that protrude into the lumen of the ventricle, thereby increasing muscle mass and altering functional output. To analyze trabecular morphogenesis in zebrafish, investigators have used a variety of myl7-driven fluorescent reporter transgenes, together with either confocal microscopy or selective plane illumination microscopy (SPIM), to facilitate live imaging of the inner surface of the ventricular myocardium (Liu et al, 2010; Peshkovsky et al, 2011; Staudt et al, 2014; Samsa et al, 2015) (Fig. 8). These studies have followed the progression of trabeculation over time, with a particular focus on the initiation of this process after 2 dpf (Liu et al, 2010; Peshkovsky et al, 2011; Staudt et al, 2014). Prior to the onset of trabeculation, the ventricular wall has a generally uniform thickness, and the ventricular lumen has a correspondingly smooth inner contour. To initiate trabeculation, cardiomyocytes delaminate and migrate inward, and these protruding cells then proliferate to propagate finger-like extensions further into the ventricle. Subsequent expansion of these structures ultimately creates an elaborate network of interconnected lumenal ridges (Liu et al, 2010; Peshkovsky et al, 2011). When trabeculation fails to occur, embryos display a progressive reduction of cardiac function, including a notable decrease in ventricular contractility (Liu et al, 2010).

Figure 8.

Neuregulin signaling is required for the initiation of trabeculation. Confocal reconstructions of the ventricular myocardium in wild-type embryos, expressing Tg(myl7:dsredt4) and Tg(myl7:egfp-hshras) and treated either with DMSO (A–C) or AG1478 (D–F), an inhibitor of ErbB receptors, from 27–75 hpf. Volume reconstructions (A,D) and lumenal surface reconstructions (B,E) are shown. The AVC is outlined with a dotted white line (A,D). White asterisk (D) indicates an area where the acquisition of fluorescent signal was blocked by overlying melanocytes. Optical slices show sagittal sections (C, F), transverse sections (C′, F′), and coronal sections (C″, F″) through the region of interest, as shown in maximum intensity projection (C‴, F‴). (A–C) DMSO-treated control embryos exhibit lumenal protrusions and primitive ridges (yellow arrowheads). (D–F) In contrast, AG1478-treated embryos do not initiate trabeculation and instead exhibit a smooth lumenal surface and a uniform thickness of the chamber wall. Adapted from Peshkovsky et al. (2011).

Evaluation of trabeculation defects in zebrafish embryos has revealed a number of factors that control specific aspects of trabeculation, including the timing of its onset, the number and distribution of delaminating cells, and the proliferation of the trabecular myocardium (Liu et al, 2010; Peshkovsky et al, 2011; Staudt et al, 2014; Samsa et al, 2015). For example, live imaging approaches have been valuable for demonstrating the earliest point at which Neuregulin signaling is required during the process of trabeculation (Liu et al, 2010; Peshkovsky et al, 2011; Staudt et al, 2014). Several studies in mouse have shown that Neuregulin signaling is required for normal trabeculation (Gassmann et al, 1995; Lee et al, 1995; Meyer & Birchmeier, 1995; Lai et al, 2010). In zebrafish, as in mouse, Neuregulin signals are produced by the endocardium, and the corresponding ErbB receptors are present in the myocardium (Goishi et al, 2003; Milan et al, 2006). Inhibition of Neuregulin signaling in the zebrafish embryo, through mutation of erbb2 or pharmacological inhibition of ErbB receptors, blocks the displacement of cardiomyocytes into the ventricular lumen (Liu et al, 2010; Peshkovsky et al, 2011; Staudt et al, 2014) (Fig. 8). These findings, in combination with additional data, have led to the conclusion that Neuregulin, delivered from the endocardium to the myocardium, is required to initiate trabeculation at the onset of the process. Further data have indicated important interactions between the Neuregulin and Notch signaling pathways during this process (Samsa et al, 2015); moreover, trabeculation, like chamber emergence and AVC differentiation, also appears to be dependent upon cardiac function, since mutations disrupting blood flow through the ventricle or cardiac contractility inhibit trabecular morphogenesis (Peshkovsky et al, 2011; Staudt et al, 2014; Samsa et al, 2015). It will be interesting for future studies to probe more deeply into how the convergence of these inputs triggers the changes in cardiomyocyte behavior that underlie trabeculation.

IV. Regulation of Cardiac Function

The heart drives circulation with forceful contractions triggered by propagated electrical impulses. Functional deficiencies could originate with inappropriate cardiac morphology, since the specific architecture of the heart is crucial for efficient pumping. Conversely, since blood flow and contractility regulate multiple aspects of cardiac morphogenesis, inadequate cardiac function can also be the cause of a dysmorphic heart. Cardiac functional defects can range from the absence of a heartbeat to a subtle arrhythmia, and these problems can be caused by abnormalities in the contractile apparatus or the conduction system. In this section, we discuss several experimental techniques that are suitable for analyzing defective heart function in the zebrafish embryo.

Cardiac function is easily observed in the developing zebrafish embryo, with basic features of contraction and blood flow being visible even on a dissecting microscope. Sarcomere assembly begins within the primitive heart tube, facilitating the start of contractility by 24 hpf (Huang, Zhang, & Xu, 2009). Mature sarcomeres are organized by the time of chamber emergence, with robust, serial contractions of the atrium and ventricle becoming apparent by 48 hpf. The cardiac conduction system matures over the same timeframe. Within the early heart tube, electrical activity travels unidirectionally and smoothly from the venous pole to the arterial pole (Milan et al, 2006; Chi, Shaw, Jungblut, et al, 2008). Once the cardiac chambers emerge, a conduction delay separates the atrial and ventricular contractions; this delay is a consequence of the formation of specialized conduction tissue at the AVC.

Qualititative assessment of heart function can rapidly discern the presence of severe defects, such as failure of chamber contraction or lack of blood flow through the dorsal aorta. Quantitative methods can be employed to characterize more subtle phenotypes, even when cardiac contraction and blood flow appear superficially normal. Assessment of heart rate is straightforward and can elucidate nuanced alterations in the speed and rhythm of contraction. Additionally, degree of contractility can be quantified using high-speed video microscopy to determine ventricular fractional shortening, a comparison of ventricular dimensions at diastole and systole (Rottbauer et al, 2005; Rottbauer et al, 2006; Fink et al, 2009).

Even subtle defects in cardiac function can cause abnormal blood flow patterns with significant consequences for cardiac morphogenesis. The ability to sense and respond to changes in the magnitude and direction of blood flow is a general property of endothelial cells, including the endocardium (Culver & Dickinson, 2010; Boselli et al, 2015). As development proceeds, different regions of the heart experience specific flow patterns (Vermot et al, 2009; Heckel et al, 2015). In particular, the AVC undergoes pronounced temporal oscillations in flow direction during the cardiac cycle (i.e. enhanced retrograde flow from the ventricle back into the atrium) (Scherz et al, 2008; Vermot et al, 2009; Heckel et al, 2015). These oscillations are largest at early stages, as the cardiac cushions begin to form. Recent work has suggested that the flow oscillations themselves drive specific gene expression programs within the AVC endocardium to regulate cushion and valve morphogenesis (Vermot et al, 2009; Heckel et al, 2015). Thus, analysis of blood flow patterns has the potential to provide mechanistic insight into the origin of morphological defects.

There are currently several popular methods for measuring cardiac blood flow dynamics in the zebrafish embryo. One approach is simply to track individual erythrocytes as they flow through the heart (Scherz et al, 2008; Vermot et al, 2009). Alternatively, Digital Particle Image Velocimetry (DPIV) can generate spatial flow maps by cross-correlating the positions of groups of circulating particles (usually erythrocytes) between imaging frames (Hove et al, 2003). Both techniques require an extremely rapid imaging rate of at least 150 frames per second, which is attainable with commercially available confocal microscopes as well as SPIM systems (Mickoleit et al, 2014). These techniques also require some method of generating contrast between the erythrocytes and their surrounding plasma. This is accomplished either by using the Tg(gata1:dsred) line (Scherz et al, 2008; Heckel et al, 2015), in which erythrocytes fluoresce red, or by injecting embryos with BODIPY-ceramide so that the plasma fluoresces and the erythrocytes appear as dark spots (Hove et al, 2003; Vermot et al, 2009). Although both erythrocyte tracking and DPIV can provide useful data, neither approach is perfect. The three-dimensional movement of the heart makes it difficult to track individual cells through the entire cardiac cycle (Boselli & Vermot, 2015), and DPIV is limited by the dimensions of the erythrocytes, making it ineffective at calculating shear rates in narrow regions of the heart, such as the AVC (Boselli & Vermot, 2015). A different approach, which circumvents some of these limitations, is to computationally estimate the flow field based on the dynamics of the cardiac chamber walls (Boselli & Vermot, 2015; Heckel et al, 2015). This strategy has been successfully used to quantify changes in blood flow dynamics induced by the loss of erythrocytes upon gata1 or gata2 knockdown and to show correlations between the oscillatory flow magnitude and a specific gene expression program in the AVC endocardium (Heckel et al, 2015).

Aberrant blood flow dynamics often originate with defects in the cardiac contractile apparatus. When analyzing cardiac contractility phenotypes, it is important to examine sarcomere structure between 24 and 48 hpf in order to distinguish between errors in myofibril assembly and maintenance (Yang, Shih, & Xu, 2014). In terms of spatial resolution, transmission electron microscopy is the most powerful technique for pinpointing ultrastructural defects in sarcomeres. For example, in silent heart mutants, which lack the cardiac troponin T gene tnnt2, myofibrils fail to assemble, as the thick filaments are disorganized in the absence of normal thin filaments (Sehnert et al, 2002). In contrast, in pickwick mutants, which lack titin function, nascent myofibrils form normally but higher-order sarcomeric structures are absent, suggesting a key role of Titin in sarcomere organization and maintenance (Xu et al, 2002).

The analysis of cardiac contractility defects is complicated by the dynamics of sarcomere content, which differs between individual cardiomyocytes and changes over time, as chamber emergence is accompanied by myofibril growth and reorganization (Lin et al, 2012; Yang & Xu, 2012; Reischauer, Arnaout, Ramadass, & Stainier, 2014; Yang et al, 2014). Fortunately, several transgenic tools provide opportunities to examine sarcomere formation, remodeling, and maintenance over time in live embryos. One tool, the transgene Tg(myl7:actn3b-egfp), facilitates live imaging of alpha-actinin localization, permitting the visualization of Z-bodies and Z-bands during myofibril maturation (Fig. 9A–C) (Lin et al, 2012). Coupled with markers that outline cardiomyocytes, Tg(myl7:actn3b-egfp) allows real-time monitoring of myofibril content on the level of individual cells. In one example, this strategy revealed an important influence of hemodynamic forces in promoting the accumulation of myofibril content during the emergence of the zebrafish ventricle (Fig. 9D–F; (Lin et al, 2012)). Another transgene, Tg(myl7:lifeact-gfp), uses the actin-labeling peptide Lifeact to enable the observation of cytoskeletal dynamics in the maturing myocardium (Reischauer et al, 2014). By employing this tool to distinguish subcellular differences in cytoskeletal organization, a recent study demonstrated a potent influence of the ErbB2 receptor on myofibrillar remodeling during ventricular maturation (Reischauer et al, 2014). These and other emerging tools will enable future studies to delve deeper into the mechanisms through which the integration of biomechanical inputs with growth factor signaling pathways directs the organization and growth of the contractile apparatus.

Figure 9.

Myofibril maturation is diminished when blood flow is reduced. (A–F) Ventral views, arterial pole at the top, of dissected hearts at 28, 38, and 50 hpf depict localization of Actn3b-egfp (green) in Z-bodies and Z-bands and Dm-grasp (red) at cell boundaries in wild-type (wt) embryos (A–C) and wea (amhc) mutant embryos (D–F). Actn3b-egfp and Dm-grasp localization indicate the dynamic progression of cell size and myofibril content during chamber emergence. Wild-type and wea mutant ventricular cardiomyocytes are similar at 28 hpf (A,D), but cell size expansion and myofibril growth and organization do not progress normally in wea mutants (B,C,E,F). Although organized Z-bands do form in wea, their growth is diminished such that the overall myofibril content at 50 hpf is significantly reduced in wea mutant embryos (F), compared to their wt siblings (E). Adapted from Lin et al. (2012).

For cardiac function phenotypes that are not associated with defects in the contractile apparatus, it is logical to investigate whether cardiac conduction is aberrant. Electrical currents can be assayed in vivo by electrocardiography, and similar patch clamp techniques can be used to stimulate the heart to test its excitability (Rottbauer et al, 2001). For optical mapping of cardiac conduction, calcium flux can be monitored using fluorescent dyes (Ebert et al, 2005; Langenbacher et al, 2005; Milan et al, 2006) or with a fluorescent calcium indicator transgene (Tg(myl7:gCaMP); (Chi, Shaw, Jungblut, et al, 2008)). In addition, transmembrane action potential can be evaluated using voltage-sensitive dyes (Panakova, Werdich, & Macrae, 2010), as well as with a novel dual-function transgenic reporter, CaViar, that can map both calcium and action potential dynamics simultaneously (Hou, Kralj, Douglass, Engert, & Cohen, 2014). This combination of techniques can detect a wide variety of conduction abnormalities, ranging from cell-intrinsic defects in calcium handling to failure to develop specific types of conduction tissue. For example, the arrhythmia observed in tremblor (tre) mutant embryos is caused by mutation of the gene encoding the sodium/calcium exchanger Ncx1h (Ebert et al, 2005; Langenbacher et al, 2005; Shimizu et al, 2015). The consequent defects in calcium handling in tre mutants interfere with the normal rhythm of calcium transients (Fig. 10), leading to irregular calcium signals that result in unsynchronized contractions. In a different case, slipjig (foxn4) mutants fail to specify the AVC and therefore do not develop specialized AVC conduction tissue; as a consequence, they exhibit an absence of atrioventricular conduction delay (Chi, Shaw, Jungblut, et al, 2008). Given the intricate interconnections between conduction, contraction, blood flow, and morphogenesis, it will be exciting for future studies to utilize emerging optogenetic tools that can manipulate the spatial pattern of electrical impulses (Arrenberg, Stainier, Baier, & Huisken, 2010) in order to probe the relationship between cardiac function and cardiac morphology in more depth.

Figure 10.

Optical mapping of cardiac conduction reveals aberrant calcium excitation in tre mutants. Sequence of heat map images depicts calcium activation in a wild-type (WT) heart expressing Tg(myl7:gCaMP) at 36 hpf; isochronal lines represent every 33 milliseconds. In WT embryos, calcium activation initiates in the sinus venosus (bottom) and concludes in the ventricle (top). In contrast, heat map images from a tre mutant heart indicate that only sporadic calcium activation occurs in the absence of Ncx1h function. Adapted from Shimizu et al. (2015).

V. Summary

A wide variety of techniques are available to investigators seeking to determine the origins of defects in heart size, shape, and function in the zebrafish embryo. Whereas some of the applicable techniques require specific reporter transgenes or specialized microscopes, most are readily accessible and should facilitate characterization of cardiac phenotypes in a broad range of laboratories. Since cardiogenesis involves the orchestration of a myriad of molecular processes – ranging from morphogen signaling to tissue mechanics and from gene regulatory networks to ion channel function – this field will continue to attract researchers from a wide range of scientific disciplines. Such interdisciplinary collaborations, together with the continued advancement of new tools for genome editing and live imaging, predict a bright future for the value of investigating heart development in the zebrafish embryo.