Vertebrate heart growth is regulated by functional antagonism between Gridlock and Gata5

Abstract

Embryonic organs attain their final dimensions through the generation of proper cell number and size, but the control mechanisms remain obscure. Here, we establish Gridlock (Grl), a Hairy-related basic helix–loop–helix (bHLH) transcription factor, as a negative regulator of cardiomyocyte proliferative growth in zebrafish embryos. Mutations in grl cause an increase in expression of a group of immediate-early growth genes, myocardial genes, and development of hyperplastic hearts. Conversely, cardiomyocytes with augmented Grl activity have diminished cell volume and fail to divide, resulting in a marked reduction in heart size. Both bHLH domain and carboxyl region are required for Grl negative control of myocardial proliferative growth. These Grl-induced cardiac effects are counterbalanced by the transcriptional activator Gata5 but not Gata4, which promotes cardiomyocyte expansion in the embryo. Biochemical analyses show that Grl forms a complex with Gata5 through the carboxyl region and can repress Gata5-mediated transcription via the bHLH domain. Hence, our studies suggest that Grl regulates embryonic heart growth via opposing Gata5, at least in part through their protein interactions in modulating gene expression.

The embryonic heart grows through a combination of cardiomyocyte proliferation and an increase in cell mass due to the generation of myofibrillar arrays (1). Cardiac proliferation diminishes progressively after birth, and postmitotic hypertrophy in the adult heart provides most of the adaptive responses necessary to supply increased cardiac output (2). Despite recent progress that has been made in the regulation of postnatal cardiac hypertrophy, the mechanisms and pathways that control embryonic heart growth are poorly delineated. It is not known what molecular signals restrain cardiomyocyte proliferative growth in the embryonic heart. The GATA zinc-finger transcription factors promote myocardial differentiation and expansion. Among the three gata genes (gata4, gata5, and gata6) in zebrafish, gata5 plays the most prominent role in heart growth and development (3). Mutations in gata5 in zebrafish cause a reduction in expression of early and late myocardial genes and a decrease in cardiac progenitor cells and proliferative cardiomyocytes, resulting in small hearts. Forced gata5 expression in the zebrafish embryo increases heart size and occasionally produces ectopically beating myocardial tissue (4). The phenotype of zebrafish gata5 mutants closely resembles the cardiac phenotypes in gata4 mutant mice. Myocardium-restricted deletion of murine gata4 or gata4/gata6 double heterozygote causes a marked reduction in cardiomyocyte proliferation and results in hypoplastic hearts (5–7). Although it seems to be clear that certain levels of GATA activity are required to drive myocardial proliferative growth, opposing signals might also be necessary to constrain the excessive cardiac growth during development.

grl encodes a hairy-related basic helix–loop–helix (bHLH) transcriptional repressor and belongs to the murine hey/hrt gene family that contains hey1, grl/hey2, and heyL (8–12). In zebrafish, grl/hey2 is the only hey gene that is expressed in the heart and aorta (8, 13). hey1 is expressed in the presomitic mesoderm, whereas heyL shows expression in the ventral side of the neural tube (13). These data suggest that grl may play critical functions in both cardiac and vascular systems. The zebrafish grl^m145 mutant was originally isolated from a mutagenesis screen and classified as a vascular mutant (14). The cardiovascular lesion in the grl^m145 mutant was identified in the aortic bifurcation, where the lateral aortae fail to assemble, leading to a lack of blood flow to the trunk, which resembles coarctation of the aorta in humans (15). The mutant grl^m145 gene causes a point mutation that changes a stop codon to Gly, thereby extending the protein by 44 aa (8). Knockdown grl activity, using antisense morpholino oligonucleotides (MO), phenocopies the grl^m145 mutant and affects arterial differentiation and development (16). The aortic defective phenotype caused by grl^m145 mutation can be rescued by VEGF and two structurally related small molecules (17). Our studies demonstrate that grl promotes arterial differentiation and development in part via the Notch signaling pathway (18). The roles of grl in cardiac development and growth, however, have not yet been examined in zebrafish. Although targeted inactivation of grl/hey2 in mice results in a wide spectrum of cardiovascular malformations (19–22), the mechanisms and pathways that underlie these morphological alterations and its roles in myocardial growth remain unclear.

In the current study, we establish Grl as a negative transcriptional regulator that restricts embryonic heart growth by opposing Gata5 activity in zebrafish. We reveal that the grl mutant heart increases expression of immediate-early growth genes and myocardial genes and contains more cardiomyocytes with increased cell size. We show that forced grl expression in WT embryos causes a marked reduction in heart size, due to a decrease in both cardiomyocyte number and cell volume. Significantly, the hypoplastic heart phenotype induced by grl can be rescued by increasing mRNA of gata5, but not gata4, implying that these two gata factors are not functionally equivalent during heart growth. Our biochemical studies demonstrate a physical association of Grl with Gata5 via the carboxyl region. This association is required for inhibiting Gata5-mediated transcription and appears to mediate Grl-induced repressive effects on myocardial proliferative growth.

Results

grl Mutants Display Increased Embryonic Heart Growth.

In zebrafish embryos, grl is first expressed in the anterior lateral plate mesoderm (ALPM) at the 3-somite stage [supporting information (SI) Appendix, section I]. Thereafter, grl expression becomes predominantly ventricular myocardial with some transcripts also detectable in the atrioventricular boundary and the atrium (SI Appendix, section I). To study grl function during heart development, we examined myocardial gene expression in the cardiac primordia. Before the heart tube assembly, nkx2.5 and gata4 are expressed at comparable levels at the ALPM in grl^m145 mutant embryos, compared with WT siblings (Fig. 1 A–D), suggesting that the precardiac mesoderm and myocardial progenitor cells are normally established in the ALPM. In grl^m145 embryos, these progenitor cells differentiate into cardiomyocytes expressing the normal levels of cardiac myosin light chain (cmlc2) (Fig. 1 E and F) and migrate to the midline to properly form the heart cone (Fig. 1 G and H). Quantitative RT-PCR analyses indicated that the levels of cmlc2 transcripts were not altered in grl^m145 mutants at the onset of myocardial differentiation [17 hours postfertilization (hpf)] compared with WT embryos (Fig. 1U). Together, these data suggest that mutations in grl do not affect myocardial cell commitment and differentiation. It is not until 48 hpf, when the myocardium undergoes concentric growth, that the expression of the sarcomere components cmlc2, atrial myosin heavy chain (amhc), ventricular myosin heavy chain (vmhc), and atrial natriuretic factor (anf) are increased in grl^m145 mutant hearts, when compared with WT counterparts (Fig. 1 I–P).

Fig. 1.

grl mutations cause an increased expression of myocardial growth genes. Expression patterns of nkx2.5 (A and B), gata4 (C and D), cmlc2 (E–H), amhc (I and J), vmhc (K and L), anf (M–P), junb (Q and R), and egr-1 (S and T) in WT, grl^m145, and grl^vu59 embryos were revealed by in situ hybridization at the 5-somite stage (A–D), the 16-somite stage (E and F), the 25-somite stage (G and H), 48 hpf (I–N), and 72 hpf (O–T). wt, wild type. (U) Real-time RT-PCR analyses showing the increased expression of cmlc2, vmhc, amhc, anf, junB, v-fos, and egr-1 in grl^m145 mutants at 48 hpf. Standard deviations were obtained and presented as standard error bars, using Excel. Asterisks indicate statistically significant difference between grl and WT data (P < 0.001). (A–H) Dorsal views with anterior to the top. (I–P) Ventral views with anterior to the top. (Q–T) Lateral views with anterior to the left. Arrow, atrium; arrowhead, ventricle; blue arrow, heart; wt, wild type.

We hypothesized that changes in the expression of anf and sarcomeric components would be accompanied by changes in the expression of growth regulatory genes. We compared transcriptional profiles of grl^m145 hearts with WT hearts at 48 hpf, using microarray analyses. Remarkably, expression of junb, v-fos, early growth response-1 (egr-1), and other myocardial genes are significantly increased in mutant hearts (SI Appendix, section II). egr-1, a zinc-finger transcriptional factor, belonging to a group of immediate-early growth genes such as fos and jun, plays an important role in cell growth and proliferation (1, 23). Quantitative RT-PCR validated the up-regulation of junb, v-fos, and egr-1 in grl^m145 hearts (Fig. 1U). We next examined expression of these genes in grl^vu59 mutants, a recently identified nonsense mutation (SI Appendix, section III), and revealed the increased expression of junb and egr-1 in grl^vu59 mutant hearts (Fig. 1 Q–T). Thus, grl negatively regulates expression of immediate-early growth genes and myocardial genes, which may in turn control heart growth during development.

We next determined cardiomyocyte number and cell size in grl mutants, using whole-mount confocal microscopy imaging. Confocal imaging of both WT and mutant hearts that express EGFP revealed that the grl^m145 heart had an apparent looping defect (Fig. 2 A and B). In addition, the size of mutant atrium is enlarged, whereas the mutant ventricle is relatively compact and small in comparison with WT cardiac chambers (Fig. 2 A and B). Histological analyses indicated that increased numbers of cells within the ventricular wall resulted in multiple cell layers in a compact ventricle, whereas the mutant atrium contains more cells in a single layer that normally leads to an enlarged atrium (Fig. 2 C and D). We quantified total cardiomyocyte number in grl^m145 hearts versus WT counterparts in a series of confocal sections. When the heart tube forms at 24 hpf, the cardiomyocyte number in grl^m145 embryos (142 ± 4) is comparable with WT siblings (148 ± 5). At 48 hpf, a large increase in the number of cardiomyocytes in grl^m145 mutants is evident (328 ± 7) compared with WT embryos (281 ± 7). The numbers of mutant ventricular myocytes and atrial myocytes increase proportionally, in comparison with WT counterparts (Fig. 2K). Increased mitotic cells were observed by using antibodies recognizing a mitotic marker phosphohistone 3 (PH3) in grl^m145 mutant hearts (4.1 ± 0.3), compared with WT hearts (2.9 ± 0.4) (Fig. 2 E–H). Moreover, mutant myocytes grow into larger size than WT cells from 48 hpf to 72 hpf (Fig. 2L). Electron microscopy revealed a normal and characteristic sarcomere structure in both mutant and WT cardiomyocytes (Fig. 2 I and J). We also examined cell number and size in grl^vu59 hearts and revealed a greater proportional increase in both parameters compared with WT hearts (Fig. 2 K and L). During early development stages, few myocytes undergo apoptosis in WT and mutant hearts (SI Appendix, section IV). Together, these data suggest that increased cell size and number in grl mutant hearts are due to defects in cell growth and division, a tightly coupled cellular event during heart development.

Fig. 2.

grl mutant embryos display cardiac hyperplasia and looping defects. (A and B) Whole-mount confocal microscopy revealing a normal looped heart (Tg(cmlc2:EGFP)/+) containing a properly formed ventricle and atrium (A) and a grl mutant heart (grl^m145/grl^m145, Tg(cmlc2:EGFP)/+) with an apparent looping defect, an enlarged atrium, and a compact ventricle (B). (C and D) Histological section showing the multiple cell layers in grl^m145 ventricles. (E and F) Immunostaining analyses revealing more PH3-positive cells in grl^m145 hearts than WT hearts. (G and H) Double exposures of fluorescent images showing PH3-positive cells in cmcl2-EGFP transgenic hearts in WT and grl^m145 embryos. (I and J) Transmission electron microscopy showing intact myofibrillar arrays in both WT and grl mutant hearts. (K) Cell number in the atrium and ventricle are increased in both grl^m145 and grl^vu59 hearts compared with WT hearts at 48 hpf. (L) Myocyte size in both grl^m145 and grl^vu59 hearts are increased at 72 hpf but not 48 hpf compared with WT myocytes. Error bars indicate standard deviation, and asterisks indicate statistical significance between grl and WT data (P < 0.01). Red arrow, ventricle; blue arrow, atrium; yellow arrow, atrioventricular boundary; wt, wild type.

Forced Expression of grl in WT Embryos Causes a Marked Reduction in Heart Size.

We next investigated whether the increased expression of grl in zebrafish embryos was sufficient to suppress myocardial growth. Notably, we found that injection of embryos with synthetic grl mRNA at low doses (≈45 pg) caused formation of small hearts (Fig. 3 A–J). TUNEL and acridine orange assays indicated that the observed phenotype in affected hearts did not result from cell death (SI Appendix, section IV). Double immunofluoresent staining, using antibodies that recognize differentially expressed ventricular and atrial epitopes, indicated that both ventricular and atrial differentiation occurred in the affected hearts (Fig. 3 A–F). These hearts correctly form myocardial and endocardial layers (Fig. 3 G and H). It appears that the mutant ventricular growth is more impaired than the atrial growth (Fig. 3 I and J). We next measured the number and area of cardiomyocytes in embryos expressing increased levels of grl mRNA. Using confocal microscopic imaging of these EGFP-expressing hearts, we found substantially lower cardiomyocyte number (Fig. 3K) and a marked diminished cell size compared with controls (Fig. 3L). Elevated grl expression appears to be cell type-specific because endothelium and overall embryonic growth were not affected (data not shown). Interestingly, at the onset of myocardial differentiation, the number of cardiomyocytes expressing cmlc2 was not noticeably different in grl-misexpressing embryos when compared with controls (data not shown). Thus, it appears that the reduction in cardiomyocyte number is not due to a reduced production in myocardial precursor cells but rather to a decreased proliferative capacity of differentiating cardiomyocytes.

Fig. 3.

grl misexpression in WT embryos reduces the heart size. (A and B) Immunostaining, using MF20 antibodies (TRITC for red), revealing a control heart (A) and an affected heart in grl-misexpressing embryos (B). (C and D) Immunostaining with F46 antibodies (FITC for green) showing a control atrium (C) and a small atrium in grl-misexpressing embryos (D). (E and F) Double exposures of immunofluorescent images showing a control ventricle and atrium (E) and a small ventricle and atrium in grl-misexpressing embryos (F). (G) Histological analyses showing the normal myocardium (arrowhead) and endocardium (arrow) with blood cells present in cardiac chambers. (H) grl-misexpressed heart develops myocardium (arrowhead) and endocardium (arrow). (I and J) Confocal microscopy showing a control heart (I) and a hypoplastic heart (J). (K and L) grl misexpression causes a reduction in cardiomyocyte number and cell size compared with controls. Error bars indicate standard deviation, and asterisks indicate statistically significant difference between grl-misexpressing and WT data (P < 0.001). Con, control embryos. Red arrow and v, ventricle. Blue arrow and a, atrium. (M) Schematic representation of the grl deletion mutants and their effects on embryonic heart growth. Percentage (%) indicates the number of embryos with cardiac hypoplasia or normal phenotype divided by the total number of injected embryos. n, microinjection times.

The grl gene is predicted to encode a protein containing a DNA binding domain, an HLH domain, an Orange domain, and a YRPW motif near its carboxyl terminus (8). To determine which of these domains are critical for grl to repress heart growth, we constructed a series of grl mutant forms harboring deletions in the basic motif (grl-basic), the HLH domain (grl-hlh), the Orange domain (grl-orange), the YRPW motif (grl-yrpw), and the carboxyl region (grl-trunc) (Fig. 3M). We then microinjected synthetic mRNA encoding each of these mutants into WT embryos and examined their effects on heart size and morphology. Although grl-orange or grl-yrpw misexpression caused the cardiac hypoplasia, embryos misexpressing grl-basic, grl-hlh, or grl-trunc failed to develop hypoplastic hearts (Fig. 3M). These observations indicate that the bHLH domain and the carboxyl region in Grl play critical roles for negatively regulating embryonic heart growth.

grl Regulates Embryonic Heart Growth by Action Within Cardiomyocytes.

To determine whether grl controls heart growth by acting within cardiomyocytes, we directed the expression of Grl-EGFP fusion protein in cardiomyocytes under control of the zebrafish cmlc2 promoter. The cmlc2 promoter has been shown to confer mosaic expression of reporter genes in zebrafish hearts (24). We examined the proliferative status of atrial myocytes with grl ectopic expression, as well as ventricular myocytes with increased grl expression, in the developing heart. WT cardiomyocytes expressing EGFP under the control of cmlc2 promoter formed clusters of cells in the ventricle and atrium at 48 hpf (Fig. 4A; n = 35). These clustered cells are considered to be clones of progeny derived from a single parental cell (24). Some of these cells were observed to be undergoing cytokinesis, and the dividing nuclei were easily detectable by DsRed2-Nuc labeling (Fig. 4B; n = 18). In contrast, ventricular and atrial myocytes expressing grl-EGFP failed to form such cell clusters (Fig. 4C; n = 46). These scattered cardiomyocytes maintaining single nuclei failed to divide (Fig. 4D; n = 46). Moreover, the size of cardiomyocytes expressing grl-DsRed or grl-EGFP was substantially less than those cells expressing EGFP alone (Fig. 4D; Fig. 4C versus Fig. 4A), which is consistent with the observation in grl misexpression experiments (Fig. 3L).

Fig. 4.

Increased grl activity inhibits cardiomyocyte proliferative growth in the embryonic heart. Microinjection of cmlc2:EGFP and cmlc2:grl-EGFP into WT embryos (A and C), and microinjection of cmlc2:DsRed-Nuc and cmlc2:grl-DsRed-Nuc into cmlc2-EGFP transgenic embryos (B and D) at one- to two-cell stages. At 48 hpf, the microinjected embryos were subjected to cardiomyocyte analyses, using Zeiss optics (A and C) and double-channel confocal optics (B and D). A, atrium; V, ventricle.

gata5 Antagonizes the grl-Mediated Cardiac Growth Effects.

Because GATA factors promote myocardial proliferative growth in the heart, we tested whether certain GATA family members can reverse the grl-induced cardiac hypoplasia. We coexpressed gata5 and grl at different molar ratios at one- to two-cell stages in embryos. Remarkably, at a molar ratio of 1:1, forced expression of gata5 completely rescued cardiac hypoplasia in embryos misexpressing grl (Fig. 5C, D, G, and H and SI Appendix, section V), whereas gata5 misexpression alone expanded the cmlc2 domain (Fig. 5 A, B, E, and F and SI Appendix, section V) (4). Furthermore, cardiomyocyte numbers in rescued hearts were restored by coexpression of gata5 to near normal levels, compared with controls at 48 hpf (Fig. 5 E and H and SI Appendix, section V). To elucidate the specific interaction between grl and gata5, we tested whether gata4 can suppress the grl-induced cardiac inhibitory effects. Remarkably, the small hearts induced by grl were not rescued after the coexpression of gata4 at any tested molar ratios. gata4 misexpression alone caused a marginal expansion of the cmlc2-expression domain that delineates the heart tube (SI Appendix, section V). Conversely, we examined whether a reduced gata5 activity suppressed the excessive cardiac growth in grl mutants. We microinjected gata5 antisense morpholinos (gata5-MO) into both mutant and WT embryos. Although knockdown gata5 at low doses in WT embryos predominantly affected myocardial proliferative growth and caused a marked reduction in heart size and cardiomyocyte number (Fig. 5 I and J and SI Appendix, section VI), reduced gata5 activity in grl mutant embryos resulted in only modest effects on heart volume and cell number (Fig. 5 K and L and SI Appendix, section VI). Hence, gata5 knockdown suppresses the excessive heart growth in grl mutants. Together, our results support the notion that a functional antagonism, partly dependent on molar ratios, exists between grl and gata5. This regulatory interplay is critical for the proper heart growth during embryogenesis.

Fig. 5.

grl and gata5 function antagonistically in controlling embryonic heart growth. (A and E) cmlc2 in situ hybridization revealing the normal size of the heart tube (A) and heart (E) in egfp-misexpressing embryos. (B and F) Forced gata5 expression (90 pg) in WT embryos causes a marginal expansion of both the heart tube (B) and heart (F). (C and G) Forced grl expression (45 pg) causes hypoplasia of both the heart tube (C) and heart (G). (D and H) Coexpression of gata5 (90 pg) and grl (45 pg) (molar ratio 1:1) in WT embryos results in a relatively normal size for both the heart tube (D) and heart (H). (I) A normal-sized heart labeled by cmlc2 in situ in WT embryos microinjected with gata5-control morpholinos. (J) Microinjection of gata5-MO (≈13 ng) in WT embryos causes development of a hypoplastic heart. (K) grl mutants display a hyperplastic heart. (L) gata5 morpholino injection (≈13 ng) into grl mutants results in a modest reduction of heart size. (A–D) Lateral views. The red line indicates the length of the heart tube. (E–L) Ventral views.

Grl Forms a Protein Complex with Gata5 and Represses Gata5-Mediated Transcription.

In zebrafish, both grl and gata5 are expressed in myocardial precursor cells and the embryonic heart (SI Appendix, section I) (4). We hypothesized that the opposing effects between Grl and Gata5 may be mediated through protein–protein interactions. Coimmunoprecipitations of Myc-tagged Gata5 and EGFP-tagged Grl in COS1 cells revealed the physical association between Gata5 and Grl (Fig. 6A). Subsequent experiments done with EGFP-tagged Grl mutants showed that Grl-basic, Grl-hlh, and Grl-yrpw mutants were able to associate with Gata5, whereas Grl-orange had weak interaction with Gata5. In contrast, Grl-trunc failed to interact with Gata5 (Fig. 6A). These results indicate that the carboxyl region containing the Orange domain and its contiguous regions (but not YRPW domain) is necessary for the physical association. In vivo, forced expression of grl-trunc failed to inhibit myocardial growth (Fig. 3M). This may be due to a disruption of Grl association with Gata5, suggesting that protein interactions are required for the inhibitory effects of Grl on heart growth. The Grl mutants lacking a basic or a helix–loop–helix domain were able to bind to Gata5 but did not result in cardiac hypoplasia when misexpressed (Fig. 3M), implying that protein interactions between Grl and Gata5 are necessary but not sufficient to cause grl-induced cardiac hypoplasia.

Fig. 6.

Grl physically binds Gata5 and represses Gata5-mdiated transcription. (A) Interaction of Grl and Gata5 depends on the carboxyl region. COS1 cells were cotransfected with EGFP-tagged Grl, Grl-trunc, Grl-basic, Grl-hlh, Grl-orange, Grl-yrpw constructs, and Myc-tagged Gata5 or empty PcDNA3.4 vector. Output: Protein interactions were detected by immunoblotting, using anti-EGFP antibody after immunoprecipitation, using anti-Myc antibody. Input: Gata5, Grl, and Grl mutant proteins were stably expressed and detected in the transfected lysate, using anti-Myc and anti-EGFP antibody. (B) The repressive activity of Grl is found in the bHLH domain. Fold activation of the anf promoter-driven luciferase reporter in the presence of Gata5-myc with EGFP-tagged Grl, Grl-basic, Grl-hlh, Grl-orange, Grl-yrpw, or Grl-trunc. (C) Western blot analyses showing expression of Grl and Grl-mutant proteins. Fold activation of anf luciferase activity was normalized by dividing the luciferase activity with the relative amount of proteins. Grl and Grl mutants did not transactivate or repress the reporter in the absence of Gata5 (data not shown).

We reasoned that Grl mutant proteins lacking a basic or a helix–loop–helix domain, although not devoid of associative capacity, may lack a transcriptional repressive activity resulting in a failure to inhibit embryonic heart growth. To test this hypothesis, we performed luciferase assays, using an anf promoter that is normally activated by GATA factors in a vitro system. HeLa cells were cotransfected with gata5 and with grl or grl mutant expression constructs. The native Grl protein strongly repressed gata5-transactivation of the anf reporter (Fig. 6B). Grl-orange and Grl-yrpw preserved the transcriptional repressive activity and also resulted in inhibition of heart growth in certain degrees (Figs. 3M and 6B). In contrast, Grl-basic and Grl-hlh bound to Gata5 but demonstrated a loss of repressive activity and failed to develop cardiac hypoplasia (Figs. 3M and 6 A and B). These data revealed the possible existence of a transcriptional repression domain (the bHLH domain) along with a separate protein binding domain (the carboxyl region) within Grl. Surprisingly, deletion of the carboxyl region, the putative Gata5 binding region, did not result in a complete loss of repressive activity. Because Hey2/Hrt2 protein does not affect repression through direct DNA binding (25–27), this result may be explained by the possibility that the bHLH domain might recruit other DNA-binding factors or Gata5-interacted proteins in HeLa cells that restore Grl repressive ability, but this may not occur in vivo and in cultured cardiomyocytes (Fig. 3M and SI Appendix, section VII). The presence of these unidentified cellular factors may provide an additional level to tissue specificity to the regulation of Gata5 activity. Thus, our in vitro observations are largely in agreement with the functions of these mutants in vivo and support the notion that the Grl cardiac growth effect is mediated at least in part through the interaction with Gata5 resulting in inhibition of Gata5-mediated transcription.

Discussion

In this study, we describe a regulatory circuit that involves Gridlock and a well known transcriptional activator Gata5 in controlling myocardial proliferative growth. We establish that Hairy-related transcriptional repressor Grl is a negative regulator of Gata5 responsive heart growth. Mutations in grl cause up-regulation of immediate-early growth genes, myocardial genes, and development of cardiac hyperplasia. This study is the first demonstration of the ability of a bHLH transcriptional repressor to inhibit cardiogenesis in the context of developing embryo. Importantly, we link the opposing effects between Grl and Gata5 to their transcriptional interactions and provide evidence that this effect is domain-specific. The normal role of grl in the heart appears to prevent excessive growth of the myocardium and serve to limit embryonic heart size. The ability of grl misexpression to cause a reduction in heart growth compliments our finding that reduced grl activity causes excessive myocardial proliferative growth. The cell-autonomous inhibitory effects of grl on cell growth and proliferation can be observed in the developing heart and in cultured cardiomyocytes (SI Appendix, section VII). These data suggest that cell growth and division are tightly coupled in the embryonic heart.

The hey2 mutant mouse displays a wide spectrum of cardiac morphologic malformations and occasionally develops cardiomyopathy (19–22). Our present study shows that alterations of cardiomyocyte proliferative growth develop in grl mutant hearts. Additionally, Hairy-related transcription factors are involved in patterning the atrioventricular canal (28, 29). These data suggest that cellular derangements in combination with patterning defects may underlie the morphologic alterations in these malformed hearts and potentially provide a cellular basis for some forms of congenital heart disease. In zebrafish, reduced grl activity up-regulates expression of both atrial and ventricular genes (Fig. 1). However, deletion of hrt2/hey2 in murine cardiomyocytes causes ectopic atrial gene expression in the ventricle (30). This might be due to the difference of expression patterns of hrt2/hey2 and grl in mice versus fish. Murine hey2 is expressed in the ventricle and not the atrium (27), whereas grl is the only hey gene that is expressed in both cardiac chambers (SI Appendix, section I). Thus, murine hey1 and possible heyL may play a compensatory function in the atrium or even the whole heart in the absence of hey2 activity. Nevertheless, these data provide an evolutionary conserved role for grl/hey2 that acts as an important negative regulator of myocardial genes.

We observed that gata5 but not gata4 can rescue the grl-induced cardiac hypoplasia, demonstrating that these two gata factors have distinctive functions in regulating heart growth in zebrafish. Our results highlight the specificity and effectiveness of gata5 in counteracting the effects of a bHLH transcriptional repressor on heart growth. We further link the transcriptional interactions to functional antagonism between Grl and Gata5 in embryonic heart growth. This may provide a mechanistic explanation for Grl-dependent cardiac repressive effects. In this context, Grl interacts with Gata5 via the carboxyl region and results in inhibiting the Gata5-mediated gene transcription via its bHLH domain. This cross-transcriptional interaction prevents excessive heart growth and thus ensures an optimal mass for the embryonic heart. The ability of transcriptional repression and physical association of murine Hey2/Hrt2 with Gata4 has been shown to lie in the same basic domain (25, 26). In our study, we have found that the Grl carboxyl region containing the Orange domain is required for physical association with Gata5, whereas the bHLH domain is critical for inhibiting the Gata5 transcriptional activity. These data separate the protein-binding region from the transcriptional repression domain. The different physical capacities of murine Hey2/Hrt2 and zebrafish Grl may reflect differences in the activities of the two homologues or species-dependent differences on two Gata factors. Because the expression of early growth genes is up-regulated in grl mutant hearts, further elucidation of these factors in this important developmental pathway may lead to a better understanding the mechanisms that control embryonic heart growth.

Materials and Methods

Zebrafish Strains and Molecular Constructs.

The zebrafish strains used in this study were raised according to standard procedures (31). grl^m145 mutants and cmlc2:EGFP line are described in refs. 16 and 32. grl^vu59 mutants were generated by TILLING method (33) (SI Appendix, section III). grl and gata5 deletion constructs were generated by PCR-based subcloning technique (SI Appendix, section VIII). The zebrafish gata4 and gata5 cDNAs were identified and cloned by using Titan One Tube RT-PCR System (Roche Diagnostics, Indianapolis, IN). The zebrafish whole-length gata4 cDNA sequence has been deposited in GenBank (DQ886664) and used in this study.

RNA Purification, Microarray Analysis, and Real-Time PCR.

Microarray analysis was performed with a Zebrafish Genome Array (Affymetrix, Santa Clara, CA) (SI Appendix, section II). Real-time PCR was performed with the SYBR green Kit in the 7900HT Systems (Applied Biosystems, Foster City, CA) (SI Appendix, section IX).

In Situ Hybridization, Immunofluorescent Staining, TUNEL, and Acridine Orange Assay.

RNA in situ hybridization was carried out as described in ref. 34. Immunofluorescent staining was performed by using antibodies MF20, S46, and phosphohistone3 (Santa Cruz Biotechnology, Santa Cruz, CA). TUNEL assay was performed with the Cell Death Detection KIT TMR red (Roche Diagnostics) (35). Live embryos were used for acridine orange staining (Sigma–Aldrich, Piscataway, NJ) (35). For histological analysis, specimens were fixed in 4% paraformaldehyde, dehydrated, embedded in plastic (JB-4), and sectioned at 5 μm. Nomarski and fluorescent photomicroscopy were performed by using Axioplan 2 and Axiocam camera (Carl Zeiss, Thornwood, NY).

Microinjection of mRNA and MO.

Capped mRNAs were synthesized by using an mMESSAGE mMACHINE kit (Ambion, Austin, TX) with gata4, gata5, grl, grl-basic, grl-hlh, grl-orange, grl-yrpw, or grl-trunc expression plasmids and microinjected into one- to two-cell embryos. gata5-MO and control-MO targeting ATG (36) were synthesized from Gene Tools (Philomath, OR). For DNA microinjections, linearized cmlc2-egfp, cmlc2-grl-egfp, cmlc2-dsred-Nuc, and cmlc2-grl-dsred-Nuc at 100 ng/μl were used.

Whole-Mount Confocal Microscopy and Morphometric Analyses.

To examine the embryonic hearts, the embryos were ventrally positioned toward the microwell with a 45° right-sided orientation in a special viewing chamber. Confocal images were captured by using a Zeiss LSM 510 Confocal Microscope System with ×20/0.75 planapochromat objective. The confocal pinhole was adjusted to 1 Airy unit for optimal z resolution with the 0.75 N.A. objective resulting in serial sections 0.9 μm thick. Images were acquired at 0.45-μm intervals. 3D projections were constructed by using the LSM Browser software (Zeiss), and analysis was performed by Image J software that reads .lsm files (http://rsb.info.nih.gov/ij). For clearly visualizing each cell's surface margin, embryos were compressed and imaged with the ×40 planapochromat objective. Thirty cardiomyocytes were chosen from each group of grl mutant, grl-misexpressing, and control embryos, and their cell sizes were measured. Cell numbers in the heart tube, atrium, and ventricle were determined by counting 10 embryos in each group of grl mutant, grl-misexpressing, and control embryos at 24 and 48 hpf. Several (five to two) representative embryos were chosen for counting cell number from groups of normal, medium, and small heart phenotypes, which had been subjected to injection of grl+gata5 or grl+gata4 mRNAs to WT embryos or to injection of gata5-MO to grl mutant and WT embryos. Numbers of cells were counted in each confocal section by using Pickpointer embedded in the Image J. Pickpointer permits a single user-defined mark to appear throughout z stacks of images, allowing the tracking of a single cell in overlapped z sections to avoid double counting.

Cell Transfection, Immunoprecipitation, and Luciferase Assays.

Coimmunoprecipitation assays were performed by using a modified protocol (25). COS1 cells were cotransfected with the indicated expression vectors for Myc- or EGFP-tagged proteins. Luciferase assays were conducted in HeLa cells cotransfected with the appropriate reporter and expression constructs, and cell lysates were assayed as described in ref. 25.

Abbreviations

bHLH

basic helix–loop–helix

hpf

hours postfertilization

MO

morpholino oligonucleotides.