Zebrafish IRX1b in the embryonic cardiac ventricle


  • Elaine M. Joseph

    Corresponding author
    1. Department of Medicine, Harvard Medical School/Massachusetts General Hospital, Charlestown, Massachusetts
    • Department of Medicine, Harvard Medical School/Massachusetts General Hospital, 149 13th Street, 4th Floor, Charlestown, MA 02129
    Search for more papers by this author

  • The IRX1b gene presented here has been renamed IRX1a by Feijoo CG, Manzanares M, de la Calle-Mustienes E, Gomez-Skarmeta JL, Allende ML. 2004. The Irx gene family in zebrafish: genomic structure, evolution and initial characterization of irx5b. Dev Genes Evol 214:277–284.


The synchronous contraction of the vertebrate heart requires a conduction system. While coordinated contraction of the cardiac chambers is observed in zebrafish larvae, no histological evidence yet has been found for the existence of a cardiac conduction system in this tractable teleost. The homeodomain transcription factor gene IRX1 has been shown in the mouse embryo to be a marker of cells that give rise to the distinctive cardiac ventricular conduction system. Here, I demonstrate that zebrafish IRX1b is expressed in a restricted subset of ventricular myocytes within the embryonic zebrafish heart. IRX1b expression occurs as the electrical maturation of the heart is taking place, in a location analogous to the initial expression domain of mouse IRX1. The gene expression pattern of IRX1b is altered in silent heart genetic mutant embryos and in embryos treated with the endothelin receptor antagonist bosentan. Furthermore, injection of a morpholino oligonucleotide targeted to block IRX1b translation slows the heart rate. Developmental Dynamics 231:720–726, 2004. © 2004 Wiley-Liss, Inc.


Development of the cardiac conduction system has been studied extensively in the mouse and chicken (reviewed in Moorman et al., 1998). Spontaneous cardiac depolarization is initially peristaltic, passing in a single wave from the sinus venosus to the outflow tract. As the heart develops, this electrical phenotype matures and becomes characterized by the appearance of electrical discontinuity between the atrium and ventricle, with the emergence of a delayed atrioventricular conduction, and the eventual synchronous contraction of the ventricular chamber.

This mature electrical phenotype is associated with a morphologically distinct atrioventricular bundle and bundle branches in the mouse and chicken; yet, remarkably little is known of the molecular correlates of this developmental process. The specialized conduction system of the vertebrate ventricle has been demonstrated to arise from nonspecialized myocytes in a process involving induction by endothelin acting in concert with a locally expressed endothelin converting enzyme (Gourdie et al., 1998; Hyer et al., 1999; Hall et al., 2004). Dysfunction of the specialized cardiac conduction system or the normal patterns of intraventricular depolarization by a wide range of inherited and acquired pathological conditions can result in ventricular arrhythmias and sudden cardiac death (Wenink, 1979).

Recently, several markers of the developing conduction system have been identified (reviewed in Moorman et al., 1998; Gourdie et al., 1999). These markers include the following: connexins, which comprise gap junctions needed for the depolarizing action potential to travel through the heart; a phosphorylated form of the cytoskeletal protein desmin; a polysialated form of a neural cell adhesion molecule (PS-NCAM); and an antibody that recognizes HNK-1. In general, these markers have enriched expression within cells of the cardiac conduction system compared with other myocardial cells, rather than having exclusive expression in conduction cells. These markers tend to have a high degree of variability in expression patterns and expression intensities between different organisms and between embryos and adults of the same organism. Further examination of markers with early embryonic expression in ventricular conduction system components should provide better understanding of how the ventricular chamber achieves its coordinated beat. One such marker is the transcription factor gene, IRX1.

In the mouse, IRX1 is first expressed in the heart on embryonic day (E) 10.5 in the ventricular trabeculae where the ventricular septum will form (Christoffels et al., 2000; Gomez-Skarmeta and Modolell, 2002). Later, mouse IRX1 expression becomes restricted to the atrioventricular bundle and bundle branches of the conduction system (E17.5), which develop along the septum (Christoffels et al., 2000; Franco and Icardo, 2001).

The earliest events of cardiogenesis are not easily accessible in the mouse because of the placental nature of the embryos. Therefore, the zebrafish has become a popular model organism for the study of cardiogenesis. Zebrafish have a simple multichambered heart consisting of a sinus venosus, atrium, ventricle, and a bulbus arteriosus (Fishman and Chien, 1997). Zebrafish lack a ventricular septum. Nevertheless, the zebrafish model organism offers a unique opportunity to examine which factors are involved in driving cardiac maturation because embryonic zebrafish are fertilized externally, develop quickly, and can survive without cardiac function for up to 5 days.

In zebrafish, a primary heart tube forms and begins peristaltic waves of contraction by 24 hr postfertilization (hpf; Fishman and Chien, 1997). At 33 hpf, the heart loops. By 36 hpf, a more mature heart beat is present with the atrium and ventricle beating in concert. By 72 hpf, the zebrafish atrium and ventricle are distinct with different sarcomeric protein expression profiles and mature electrophysiologic characteristics. However, no histological evidence has been found for the existence of a cardiac conduction system in the embryonic or adult zebrafish heart. Recently, optical mapping techniques have provided indirect evidence of a functioning cardiac conduction system in the adult zebrafish heart (Sedmera et al., 2003). Isochronal maps were used to demonstrate activation of the ventricle from apex to base and trabecular myocardium was suggested to be functioning as a surrogate for specialized conduction tissue (Sedmera et al., 2003).

Given that IRX genes may specify a subset of ventricular cells with conduction properties, I studied the expression of zebrafish IRX1b in wild-type and genetic mutant ventricles. I find expression of this IRX1-like gene occurs in the zebrafish ventricular trabeculae and is diminished in silent heart genetic mutant zebrafish embryos. In addition, morpholino oligonucleotide targeted “knockdown” of this gene slows the heart rate of injected embryos. Thus, it is possible that zebrafish IRX1b expression marks the cells that act as a simple conduction system in the zebrafish embryo and that the expression of this gene relies upon the activity of the heart itself.


Expression Analysis

The Iroquois (IRX) family of homeodomain genes is a subclass of the three amino acid loop extension (TALE) superfamily (Burglin, 1997; Bosse, 2000). IRX proteins act as either transcriptional repressors or activators; this difference might be regulated by alternative splicing (in this study, two splice variants of the same gene, IRX1b, were isolated). Zebrafish IRX1b is more similar to mouse and chicken IRX1 than is the other zebrafish IRX1-like gene, ziro1 (Cheng et al., 2001; Wang et al., 2001). The IRX1b homeobox region is 98.4% identical to the homeobox region of the mouse IRX1 protein, whereas the homeobox region of ziro1 (Wang et al., 2001) is only 90.4% identical (GenBank accession no. AY017308). Zebrafish–hamster somatic cell radiation hybrid analysis was used to map IRX1b to LG16 between the genetic simple-sequence length polymorphism markers Z1837 and Z7231 (data not shown).

Whole-mount in situ hybridization analysis was used to examine the expression pattern of IRX1b (Fig. 1). IRX1b expression begins at 10 hpf in the neuroectoderm; expression is not detected in the heart field at 24 hpf or earlier (Cheng et al., 2001; Fig. 1A,B). IRX1b cardiac expression begins, after formation of the linear heart tube, at around 30 hpf (Fig. 1B) and continues at all stages examined through 130 hpf (Fig. 1B). These expression data correlate with wild-type heart rates (Fig. 1C). At 24 hpf, the heart rate is low and IRX1b cardiac expression is not detectable. IRX1b expression comes on at a time when the heart rate is increasing and stays on as cardiac depolarization patterns mature (Fig. 1C).

Figure 1.

Zebrafish IRX1b gene expression at 1 and 2 days postfertilization (dpf). A: Panel 1: At 22 hours postfertilization (hpf), IRX1b is expressed in the central nervous system. A lateral view of a 26 somite stage embryo (22 hpf) with the anterior at the left and the posterior to the right. Dorsal is at the top and ventral is at the bottom. Expression is particularly strong in the midbrain and hindbrain structures (the diencephalon, tectum, and cerebellum) but is excluded from the forebrain and the midbrain–hindbrain junction. T, telencephalon; D, diencephalon; TC, tectum; MHB, midbrain–hindbrain junction; C, cerebellum; HB, hindbrain. Panel 2: A dorsal view of a 22 hpf embryo shows IRX1b expression in the otic vesicles (OV). Panel 3: At 24 hpf, IRX1b is expressed in the somitic mesoderm (SM). Panel 4: An anterior view of a 24 hpf embryo shows IRX1b expression also occurs in the eye (E). Panel 5: A dorsal view of a 24 hpf embryo shows expression in the spinal cord (SC) and somitic mesoderm. Panel 6: At 24 hpf, expression is also seen in the two nephric ducts (ND, arrow). B:IRX1b is not detected in the newly formed heart tube at 24 hpf (left-most panel). By 30 hpf, weak IRX1b expression is detected in the primary heart tube (H, heart). This expression continues at the later stages examined (48, 58, 130 hpf; all other panels). C: The onset of IRX1b cardiac expression is correlated with an increase in embryonic heart rate. The heart rates of wild-type embryos were counted at 24, 30, 48, 55, 60, 72, and 144 hpf. The average heart rates (beats per minute) were plotted versus age (hpf) with error bars indicating standard deviations. The greatest change in beats per minute per hour of ageing occurs between 24 and 30 hpf.

Sections of whole-mount in situ hybridized embryos show IRX1b expression in the trabeculae (the spongy myocardium) of the heart ventricle and to a lesser extent in the compact working myocardium (Fig. 2). This expression of IRX1b in the ventricle occurs in a “veil-like” pattern (Fig. 2C) with a distinctive ring of stronger expression (Fig. 2D).

Figure 2.

IRX1b expression in the ventricular chamber of the wild-type embryonic heart. A: Whole-mount in situ hybridization analysis of a wild-type embryo at 53 hours postfertilization (hpf) shows IRX1b expression (blue/purple color) in the ventricle. The atrium is stained by the S46 antibody (brown color), which recognizes an atrial specific myosin heavy chain isoform. B: A section of an IRX1b whole-mount in situ hybridized embryo at 53 hpf, with S46 antibody staining of the atrium, shows a “veil-like” IRX1b expression pattern in the ventricle. C: A sectioned embryo shows IRX1b expression predominantly in the trabeculated spongy myocardium of the ventricle at 53 hpf (blue/purple, arrow). The compact myocardial layer is indicated by an arrowhead. D:IRX1b expressing cells in the ventricle are arranged in a ring structure, as shown by the more intense staining in this section of a whole-mount in situ hybridized embryo at 54 hpf.

Expression Analysis of Mutant and Pharmacologically Treated Embryos

Studies have shown that blood flow through the zebrafish heart is required for some aspects of normal cardiac development, such as heart looping (Hove et al., 2003). Trabeculae are first evident around the time of heart looping (Viragh and Challice, 1977). To test whether IRX1b transcription possibly responds to a hemodynamic signal, IRX1b expression was examined in silent heart genetic mutant zebrafish embryos (Fig. 3A). The silent heart mutation in the cardiac sarcomeric troponin T gene prevents the heart from contracting (Sehnert et al., 2002). The IRX1b cardiac gene expression pattern is weakened in these silent heart mutant fish, as determined by whole-mount in situ hybridization analysis (Fig. 3A compared with Fig. 3B).

Figure 3.

IRX1b gene expression in silent heart mutant embryos. A,B: Whole-mount in situ hybridization shows weakened IRX1b cardiac expression in the silent heart genetic mutant (tc300b allele, at 54 hours postfertilization [hpf]; compared with IRX1b expression in a wild-type sibling embryo at 54 hpf (B, heart indicated by arrow).

There is evidence from studies in chickens that hemodynamic forces are required for proper development of the cardiac conduction system (Reckova et al., 2003). Specialized conduction cells are induced from cardiomyocytes located near functioning coronary arteries. The induction of these cells is thought to depend on endothelin signaling (Gourdie et al., 1998; Hyer et al., 1999). To investigate whether zebrafish IRX1b expression might occur in response to endothelin signaling, wild-type zebrafish embryos were treated with the endothelin receptor antagonist bosentan (Fig. 4; Krum et al., 1998). Treatment of zebrafish with low levels of bosentan (200 μM) shifted the IRX1b cardiac expression domain toward the outlet of the ventricle (Fig. 4A). The IRX1b cardiac expression level in the bosentan-treated embryos is equivalent to the expression level of untreated wild-type hearts, as shown by the results of reverse transcription-polymerase chain reactions (RT-PCR; Fig. 4E).

Figure 4.

IRX1b gene expression in wild-type embryos treated with the endothelin receptor antagonist bosentan. A: Wild-type embryos were treated from 23 to 54 hours postfertilization (hpf) with 200 μM of bosentan. In bosentan-treated embryos, the cardiac expression domain of IRX1b is shifted toward the outflow tract (arrow). Bosentan-treated embryos have a heart rate that is essentially the same as wild-type untreated embryos (109 ± 8.2 bpm, n = 129 compared with 110 ± 9.5 bpm, n = 135). B: A wild-type sibling control embryo showing the normal expression pattern of IRX1b in the heart at 54 hpf (arrow). C: Embryos treated with bosentan (at 23 hpf) and then withdrawn from treatment (at 48 hpf) show a recovery of IRX1b heart expression by 76 hpf (arrow). D:IRX1b expression in a control embryo at 76 hpf (arrow). E: Reverse transcriptase-polymerase chain reaction (RT-PCR) indicates IRX1b expression in the hearts of bosentan treated embryos (Bos) is about equal to the level of IRX1b expression in the hearts of untreated siblings (−), assayed at 55–60 hpf. EF1-α is a control for RT-PCR. The markers shown in the first lane (M) are 0.38 kilobases and 0.20 kilobases for the upper IRX1b panel and 0.20 kilobases for the lower EF1-α panel. Whole embryo template without reverse transcriptase is included as a negative control in the last two lanes.

Morpholino Phenotype

To study the null phenotype for the IRX1b gene, a morpholino oligonucleotide was designed to “knockdown” translation of IRX1b (IRX1b-MO) when injected into wild-type embryos at the one-cell stage of development (Nasevicius and Ekker, 2000). The IRX1b morphants exhibit profound dose-dependent bradycardia, with slowing of the heart rate to 65% of the normal control heart rate in 2 dpf embryos when a high dose (1,000 μM) was injected (92 ± 24.9 beats per minute, bpm, compared with 141 ± 12.5 bpm, n = 48; Fig. 5). A medium IRX1b-MO injection dose (350 μM) gave a heart rate (115 ± 6.8 bpm, n = 107) that was 84% of the negative control injected embryo heart rate (136 ± 13.1 bpm, n = 70) and a low dose (200 μM) gave a heart rate of 96% (131 ± 6.2 bpm, n = 11).

Figure 5.

Phenotype of embryos injected with a morpholino designed to block IRX1b translation. A: Injection of an IRX1b morpholino (IRX1b-MO) antisense oligonucleotide prevents proper central nervous system development as shown here in these lateral views of a live 2 days postfertilization (dpf) embryo injected with a high dose of 1,000 μM. B: Dorsal view of the same IRX1b morpholino-injected embryo. C: A wild-type sibling of the morpholino-injected embryo shown in A and B. D: Dorsal view of the same wild-type embryo. E: IRX1b-MO–injected embryos develop with a distinctive ventricular (V) and atrial cardiac chamber (A); these chambers both have an endocardium (en) and a myocardium (my). F: Methylene blue azure II stained sections of IRX1b morpholino-injected embryos show other embryonic structures to be disrupted. The eye of IRX1b-MO–injected embryos has an altered retinal laminar pattern (compare with I). G: Fin outgrowth is reduced (compare with J). H: Semicircular canals do not form in the otic vesicles of IRX1b-MO–injected embryos, but otoliths do form (compare with K). I: Section through the eye of a wild-type sibling shows the normal laminar pattern of retinal neurons. J: Section through a wild-type fin shows the extent of fin outgrowth as well as the usual cell arrangement within the fin. K: Section through a wild-type sibling otic vesicle shows the presence of semicircular canal precursors (arrows).

It is possible that the bradycardia observed in the hearts of the IRX1b-MO–injected embryos is secondary to heart structural changes beyond an effect on the trabeculae. Yet, the hearts of IRX1b-MO–injected embryos develop with a distinctive atrial and ventricular chamber containing both myocardial and endocardial layers (Fig. 5E). In addition to bradycardia, IRX1b-MO morphant embryos exhibit effects on brain, eye, fin, and ear development (Fig. 5). The eyes of wild-type zebrafish have a retina with a layered pattern of cell types, including an inner ganglion cell layer surrounding the lens and an outer photoreceptor cell layer just inside of a thin outer retinal pigmented epithelium. The phenotype of the morphant embryos indicates a disruption of the normal retinal patterning (Fig. 5F compared with Fig. 5I). The pectoral fins of wild-type embryos at 2 dpf consist of mesenchymal cells within an outer epithelium. A subpopulation of the inner mesenchymal cells at this stage have become precartilage cells and are stacked in an orderly manner. The morphant fins do not appear to have a population of precartilage cells within a population of mesenchymal cells (Fig. 5G,J). Also, the morphant fin is stubby, having grown a distance away from the body equal to only one width of the fin base. The wild-type ear primordium at this stage consists of an otic vesicle with semicircular canals beginning to develop from the dorsal walls. The morphant embryos do not appear to be undergoing this morphogenic event (Fig. 5H compared with Fig. 5K). The developmental effects seen in the morphant embryos are not unexpected, because IRX1b is expressed in these various regions of the embryo.


The zebrafish ortholog of the mammalian conduction system marker IRX1 is a transcription factor gene expressed in the trabeculae of the embryonic zebrafish heart. IRX1b expression occurs at a time when the conduction pattern of the fish heart is maturing. Also, IRX1b expression is regulated by factors that effect pathways known to be critical for induction of specialized conduction cells in other organisms. This study suggests that IRX1b may be an early molecular marker of conduction cells in the zebrafish heart and raises the possibility that this highly tractable organism might be suitable for the study of the ontogeny of the conduction system and its role in normal and abnormal cardiac rhythms.


Cloning and Sequencing of IRX1b Gene

IRX1b was cloned using a PCR performed on zebrafish 3 dpf cDNA with the primers 5′-ACCACCAGCACGCTCAAG-3′ (U) and 5′-TGTGGAGACCTGTGTGAGGG-3′ (D). The resulting amplicon was gel purified, randomly labeled, and used to screen a 3 dpf embryonic heart cDNA lambda ZAP express library (constructed by Ton et al., 2000). A full-length clone was isolated from this library by using an in vivo excision procedure (Stratagene). The sequence of the full-length clone is listed as GenBank accession no. AY043241. In addition, partial clones of a splice variant of this same gene were also isolated. PCR was conducted on cDNA template from 12 hpf zebrafish embryos by using the primers 5′-AGAGAGAAACTCTTGCACTTCTCG-3′ (U) and 5′-GGAAAACACCAACACATAAAGACA-3′ (D) to isolate a full-length clone of this variant. This full-length clone was TA-cloned into the pCR4 vector (Invitrogen). The sequence of this is noted in GenBank accession no. AF359505 (Cheng et al., 2001).

Gene Mapping

IRX1b was mapped by radiation hybridization analysis on the Goodfellow T51 panel using two primer sets designed against the 3′-untranslated region of the gene (Kwok et al., 1998). One primer set was 5′-GGAACCGAAAGGGACTAGGT-3′ and 5′-AGGCCAAAACGGAAGAATTT-3′, and the other primer set was 5′-TGGGCTACCCGCAGTATTTA-3′ and 5′-GCTGGTGTAAGGCAGAAAGG-3′. Both primer pairs were placed in the same bin using the online mapping software at the Max Planck Institute in Tuebingen, Germany (http://wwwmap.tuebingen.mpg.de).

In Situ Hybridization Analysis

Zebrafish (Danio rerio) eggs were fertilized, cultured, and staged according to Westerfield (1995). Culture medium was supplemented with 0.003% 1-phenyl-2-thiourea to prevent melanization for whole-mount in situ hybridization analysis. Whole-mount in situ hybridizations were performed by using BM purple substrate or 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (BCIP/NBT). The probes for in situ hybridization were antisense digoxigenin-labeled mRNA probes made from linear full-length IRX1b-pBKCMV DNA template using in vitro RNA synthesis kits (Ambion). Sectioned whole-mount in situ hybridized embryos were embedded in plastic and sectioned at an 8–10 μm thickness.

Bosentan Treatment

Embryos were treated with bosentan, a drug used to treat hypertension. Bosentan (4-tert-butyl-N-[6-(2- ydroxyethoxy)-5-(2-methoxy-phenoxy)- [2,2′]bipyrimidin-4-yl]-benzenesulfonamide) is an endothelin receptor antagonist that acts as an inhibitor of both endothelin A and endothelin B receptors, with a slightly higher affinity for endothelin A receptors (Krum et al., 1998). Embryos were treated with 200 μM of bosentan for various lengths of time. Bosentan was prepared as a 40 mM stock in dimethyl sulfoxide that was diluted to a 10 mM stock by adding water and sodium bicarbonate (0.0175%) as a buffer. This solution was diluted to the final 200 μM working solution with embryo culture medium.

In the experiments presented, embryos were treated from 23 to 54 hpf (n = 149) or treated from 23 to 54 hpf followed by recovery in bosentan-free medium until 74–76 hpf (n = 84). Untreated control embryos were exposed to a 2-min pulse of 200 μM bosentan at 23 hpf and then cultured in bosentan-free medium until 54 hpf (n = 155) or 74–76 hpf (n = 74). After treatment, standard whole-mount in situ hybridization analysis was used to examine IRX1b expression.

RT-PCR Analysis of IRX1b Cardiac Expression

Heart tissue was dissected from bosentan-treated embryos (n = 42) and untreated wild-type sibling embryos (n = 40) at the 55–60 hpf stage. Nonradioactive RT-PCR was conducted on the tissue according to a modified protocol of Wilson and Melton (1994). RNA was isolated using proteinase K lysis. After RQ-DNAse I treatment of the samples, MMLV Reverse Transcriptase was used to produce cDNA from the heart tissue samples. PCR was run with an annealing temperature of 55°C. The IRX1b primer set was as follows: U, 5′-CATTTGCACCCCCAAAAG-3′; and D, 5′-GGACAAACTCCGTCAACTCG-3′ (30 cycles, 232 base pair product). The RT-PCR controls included analysis of elongation factor-1alpha (EF1-α), with primers: U, 5′-TGGGCACTCTACTTAAGGAC-3′; and D, 5′-TGTGGCAACAGGTGCAGTTC-3′ (20 cycles, 169 base pair) and inclusion of whole embryo template and whole embryo template with no RT. Also, a control for the presence of ventricular tissue was conducted with primers for ventricular myosin heavy chain (vmhc): U, 5′-CAAGAATGCAGGAATGCTGA-3′; and D, 5′-TCTTCTGTTCGGCCTCAACT-3′ (30 cycles, 259 base pair) (data not shown).

Morpholino Translational Block Assay

The antisense IRX1b morpholino oligonucleotide sequence was 5′-GGAAAGACATCTCCTCCGCCACGTC-3′ (Gene-tools, LLC). IRX1b morpholino was injected into the yolk of one-cell stage embryos at doses ranging from 8.4 ng to 1.68 ng in a volume of ∼1 nl (1,000 μM to 200 μM; Nasevicius and Ekker, 2000). A standard negative control morpholino (targeting a human β-globin splice site; n = 70) and a positive control morpholino (targeting the translational start site of zebrafish chordin; n = 70) were also injected into sibling embryos.

Heart Rate Analysis

Heart rates of morpholino-injected embryos (n = 107 at 350 μM), including negative (n = 70) and positive controls (n = 70) were counted at 52–58 hpf in an automated manner (according to Milan et al., 2003). Bosentan-treated (n = 129) and bosentan-untreated sibling embryos (n = 135) were counted by the same method. This automated heart rate counting system is useful for comparative purposes, because the procedure involves cooling embryos to room temperature (Milan et al., 2003). Manual heart rate counts were performed in a temperature-controlled room for determining actual heart rates of wild-type embryos at various developmental time points.


I thank Mark C. Fishman for the support he provided. I also thank Cassandra Belair, Calum MacRae, David Milan, Thomas Schultheiss, and Len Zon for providing useful reagents or discussions.