Introduction

The Snail family member snail (also known as snail1) encodes a zinc finger-containing transcriptional factor. Disruption of snail in mice causes death at E7.5 before cardiac development1. Mouse embryos with conditional disruption of snail after E8, die at E9.5, partially due to severe cardiovascular defects2. In addition, Snail has been shown to play a direct role in endocardial cushion formation in a mouse model with a conditional snail knockdown3. The data indicate that embryonic Snail deficiency is lethal in mice and Snail displays important function during the cardiac development. However, the mouse models are difficult to be examined and further to visualize how Snail controls heart development during embryonic development. Zebrafish, on the other hand, can survive without cardiovascular function during the first 7 days of development, which makes it a very good model for studying essential genes during heart development4. Therefore, we used zebrafish embryos to address the mechanisms how Snail modulates cardiac development during embryonic development.

There are two duplicated snail1 genes in zebrafish, namely snai1a and snai1b5. They exist in the paraxial and axial mesoderm6,7 and have been implicated in the anterior migration of axial mesendoderm5,8. Here, we show that snai1b plays a key role in the migration of cardiac precursors by modulating the extracellular assembly of fibronectin (Fn) via the expression of α5 integrin. Our results provide the molecular mechanism that how Snail controls heart formation during zebrafish embryonic development.

Results

Knockdown of snai1b induces cardiac defects

In order to identify the location of snai1b expression in zebrafish embryos, we detected snai1b, myl7 and fn expression by whole-mount ISH. We found that snai1b was expressed in the anterior lateral plate mesoderm (LPM), where locates myocardial precursors9, next to the neural crest and with myl7 expression (Supplementary Fig. S1A–D). These results suggest that Snai1b played a role during heart development.

We then injected snai1b antisense morpholino (MO) into embryos at the one-cell stage to determine the Snai1b functions during heart development. The results showed the same phenotypes during zebrafish embryonic development as described. In addition, we observed that more than 70% embryos injected with snai1b MO displayed a delay in cardiac fusion at 24 hours post fertilization (hpf). At 48 hpf, over 55% of embryos displayed gross cardiac defects, including large pericardial edemas, loss of cardiac looping (Figure 1A and B) and weakened heart throb. In order to confirm the snai1b MO specificity, we used a construct that expresses a Snai1b-GFP fusion protein contained snai1b 5′-UTR sequence to test the efficiency of snai1b MO. As expected, the snai1b MO efficiently depleted the expression of the Snai1b-GFP protein (Supplementary Fig. S2A–F), indicating the snai1b MO specifically disrupts the function of Snail in zebrafish embryos. Then, we injected snai1b MO at different concentrations into zebrafish embryos and identified that higher dose of snai1b MO resulted in more abnormal embryos at 48 hpf (Supplementary Fig. S2G). Injection of snai1b mRNA without 5′-UTR sequence, which the snail1b MO binds to, reduced the defects to 41.6% at 24 hpf and 17.8% at 48 hpf (Figures 1G and 2G–J). In addition, snail1a mRNA was unable to rescue the phenotype induced by snai1b knockdown. Together, the results indicate that snail1b displays essential function during the heart morphogenesis in zebrafish embryos.

Figure 1
figure 1

Knockdown of Snai1b induces cardiac morphology defects.

(A, B): Phenotypes of snai1b knockdown at 48 hpf, lateral view. Black arrows: heart valve. (C–F): Changes in heart shape in myl7::GFP embryos injected with control (Ctl) and snai1b MO at 48 hpf and 72 hpf. A, atrium; V, ventricle. (G): Quantification of phenotypes produced by snai1b morphants at 48 hpf, with or without snai1b mRNA at different concentrations. (H, I): Heart endocardium in flk::GFP embryos injected with Ctl and snai1b MO at 72 hpf. White arrows: heart. Scale bars: 100 μm.

Full size image
Figure 2
figure 2

Defects in cardiac development detected by myocardial precursor.

(A–J): myl7 expression marking myocardial precursor cells in zebrafish embryos. (A, C, E, H): embryos injected with Ctl MO develop normally at 15, 18, 24 and 48 hpf. (B, D, F, I): embryos injected with snai1b MO display cardiac fusion delays (B, D, F) and band-like heart shape at 48 hpf (I). Myl7 expression (black arrows) is reduced at 15 and 18 hpf (B, D). (G, J): embryos injected with both snai1b MO and mRNA display partial rescue of heart defects. (K–M): rescue of heart defects (arrowhead) in embryo injected with snai1b MO and mRNA at 48 hpf.

Full size image

Next, we injected snai1b MO into the myl7::GFP transgenic zebrafish embryos, which expresses GFP in the developing heart, to visualize the heart development. The results showed that depletion of snai1b induced abnormal cardiac looping in the myl7::GFP embryos. At 48 hpf, most snai1b morphants developed band-like hearts with the atrium on the left or at the midline and had a small ventricle (Figure 1C and D). These phenotypes were even more pronounced at 72 hpf (Figure 1E and F). In contrast, the flk1::GFP transgenic zebrafish embryos exhibited normal endocardium formation in snai1b morphants (Figure 1H and I). In addition, most band-like hearts formed in the morphants were able to contract and drive blood circulation, but not as efficiently as controls, indicated by the flow of the blood cells. These results further indicate that snai1b is required for embryonic cardiac development in zebrafish embryos.

Snai1b controls the migration of cardiac precursors

The morphological defects in the hearts of snai1b disrupted animals led us to determine which stages during heart development were controlled by Snai1b. Cells that contribute to the formation and development of the heart in the embryo undergo several phases of migration. The heart precursor cells migrate towards the anterior-lateral plate mesoderm after involuting during the early stages of gastrulation, subsequently, they fuse at the midline, where they form the linear heart tube10. The whole-mount ISH showed that the myocardial precursors with myl7 expression were located bilaterally in control embryos at 15 hpf (Figure 2A). In contrast, the myocardial precursors were not localized in bilateral groups and were farther away from the midline in the snai1b morphants (n = 19/25; Figure 2B). At 18 hpf, the bilateral groups of myocardial precursors began to fuse together at the midline in controls, while they remained bilateral distribution in the snai1b morphants (n = 20/27; Figure 2C and D). By 24 hpf, the precursors merged at the midline to form a single heart tube in control animals and started to migrate to the midline in the snai1b morphants (n = 12/17; Figure 2E and F). These results indicate that snai1b modulates the migration of cardiac precursors during heart embryonic development.

In addition, snai1b morphants had decreased myl7 expression detected by ISH (Figure 1 and 2A–D), suggesting the defect of myocardial precursors on cell fate in the morphants. To examine the fate of myocardial precursors in the snai1b morphants, we used TUNEL staining to detect apoptosis. We found that snai1b morphants had high levels of apoptosis. Knocking down p53 reduced apoptosis levels in the heart region (Supplementary Fig. S3A–C), but did not alleviate the other snai1b MO phenotypes, as judged by myl7 expression at 15 hpf (Supplementary Fig. S3D–F). Then myl7 expression by qRT-PCR showed that its expression was not altered in snai1b morphants (Supplementary Fig. S3G), indicating that total cells with myl7 expression remained in snai1b morphants. The results demonstrate that reduction of myl7 expression detected by ISH results from the spreading distribution of myl7 expressing cells on the mesodermal plate in snai1b morphants, not the reduction of myl7 expressing cells and further confirm that snai1b controls the migration of cardiac precursors during heart embryonic development.

The examination of endoderm formation by foxa2 expression in snai1b morphants at 26 hpf showed that there was not different in both control and morphant embryos (Supplementary Fig. S3H and I). Since endoderm formation appeared to be normal in the snai1b morphants, we excluded the possibility that defects in endoderm formation were the reason for the delay or inhibition of myocardial fusion in these mutants. Taken together, the results indicate that Snai1b is not necessary for the specification of myocardial precursors and the formation of endoderm during embryonic heart development.

Snai1b maintains Fibronectin layer in the heart region

Several lines of evidence support the fact that Fibronectin (Fn) is required during myocardial precursor migration11,12 and that snail modulates Fn deposition13. Thus, we examined Fn deposition in zebrafish embryos at 17 hpf when myocardial precursors begin to fuse. The results showed that fine Fn layers were deposited in the extracellular matrix at the midline between the endoderm and cardiac precursors and surrounded the myocardial precursors later on in control embryos. In contrast, the Fn deposition were dramatically reduced and displayed non-continuous patterns in the region surrounding the myocardial precursors in snai1b morphants, especially in the midline region (Figure 3A–D), indicating that snai1b is required for proper Fn deposition in the embryonic heart region of zebrafish. Since Fn plays important function in morphological segmentation boundary of somites in zebrafish embryos14 and knockdown of snai1b affects the shape of somite5, we also measured the Fn pattern in the abnormal somite boundary regions and identified that Fn assembly was disrupted significantly in snai1b morphants (n = 16/22; Supplementary Fig. S7). The results are consistent with previous observations5 and indicate that the Fn assembly mediated by Snai1b also plays a critical role in morphogenesis of somite.

Figure 3
figure 3

Fn around myocardial precursors is modulated by Snai1b.

Transverse sections of myl7::GFP (green) transgenic embryos immunostained for Fn (red). DAPI (blue) staining indicates nuclei. (A, B): Fn fibrils around myocardial precursors and across the midline in control embryos. (C, D): Fn levels decrease in snai1b MO injected embryos. (E, F): Fn levels are restored in embryos injected with α5β1 integrin protein. (B, D and F): Magnified view of dashed rectangle of A, C and E, respectively. Scale bars: 50 μm.

Full size image

Next, we examined the expression of fn mRNA in snai1b morphants by in situ hybridization and found out that fn expression was not altered after Snai1b knockdown in zebrafish embryos (Supplementary Fig. S4A and B). In fact, qRT-PCR results indicated that the levels of fn mRNA were higher in snai1b morphants at 17 hpf than those in control embryos (Supplementary Fig. S4C). The data indicate that snai1b does not directly regulate fn expression. Consistent with these observations, Fn protein co-injected with snai1b MO did not rescue the defects caused by Snai1b knockdown (Table 1). The results suggest that the defective generation and formation of cardiac precursors are caused by the abnormal Fn deposition.

α5 integrin acts downstream of Snai1b for Fn deposition in the embryonic heart

Since integrins are essential for Fn fibril formation15, we tested the expression of several integrins in zebrafish embryos during heart formation (Figures 4A–C and Supplementary Fig. S5). The results showed that α5 integrin was markedly decreased in snai1b morphants at 15 hpf compared to that in control embryos (n = 11/16; Figure 4A, B). The examination of expressing patterns and snai1b mRNA induction showed that α5 integrin was co-expressed with snai1b mRNA and was up-regulated in the LPM by snai1b mRNA (n = 14/21; Figure 4C, Supplementary Fig. S1E, Supplementary Fig. S5J).

Figure 4
figure 4

α5β1 integrin protein rescues cardiac defects induced by Snai1b knockdown.

In situ hybridization for indicated genes. (A–C): Expression pattern of α5 integrin in different embryos at 15 hpf, dorsal view. (D): different types of phenotypes displayed in injected embryos at 20 hpf. Graph shows phenotypic ratios of injected embryos at the same stage. Percentage of embryos of type 3 between embryos with snai1b MO and those with both snai1b MO and 25 pg α5β1 integrin protein was statistically significant (P = 0.023, t-test). Ctl means Ctl MO, MO means snai1b MO.

Full size image

To determine whether α5 integrin was able to rescue the migration defect of myocardial precursors in snai1b morphants, we co-injected snai1b MO with α5 integrin mRNA at different concentrations. Accordingly, both co-injecting 50 pg and 100 pg α5 integrin mRNA rescued the myocardial defects in snai1b morphants (Supplementary Fig. S6). Furthermore, we injected integrin proteins into the embryos to determine the function of α5 integrin in snai1b morphants. The results showed that α5β1 integrin protein was able to rescue the Fn fibril assembly at 17 hpf (Figure 3E and F) and the cardiac defects at 20 hpf (Figure 4D) in snai1b morphants. Taken together, the results demonstrate that Snai1b controls the expression α5 integrin which in turn controls Fn deposition necessary for myocardial precursor migration during heart morphogenesis in zebrafish embryos.

Discussion

Previous studies have shown that the combined expression of slug and snail is required for epithelial–mesenchymal transition (EMT) in cardiac cushion morphogenesis3,16 and snail is additionally required for left-right asymmetry determination in the heart2. Here, we show that Snai1b knockdown causes defects in myocardial precursor migration and delays cardiac fusion of these precursors in the zebrafish embryos. In addition, our results provide that the delayed fusion of myocardial precursors can be restored by α5 mRNA or α5β1 integrin protein, indicating that Snail-integrin signaling is essential for cardiac development in zebrafish.

Whether snail genes are mesodermal determinants has been long debated. There is increasing evidence indicating that the activity of Snail is related not only to cell fate, but also to cell migration17. In fact, mice mutant for Snail die at gastrulation, yet they can form mesoderm and express mesodermal markers1. The lethality of Snail mutants makes it difficult to investigate the role of Snail in organ formation in the mouse. Zebrafish, on the other hand, is an ideal model to investigate the role of Snail in organogenesis. Here, we show that Snai1b controls the migration of myocardial precursors, rather than modulating their generation. The data further demonstrate that Snail is required for cell movement, not fate determination.

Snail has been reported to play a role in endocardial development in mice3. However, the results presented here suggest that Snai1b does not involve in endocardial development in zebrafish. One possible reason might be that Snai1a can compensate for Snai1b function and maintains the generation of endocardial tissue. Another possibility is that Snai1b does not regulate the generation of endocardium, which is derived from a distinct region in the anterior LPM where snai1b does not express18.

Our studies indicate that myocardial precursor migration defects are due to the disruption of the Fn layer. Previous studies have shown that Fn is a multi-domain ECM protein that mediates multiple cellular behaviors15 and is expressed early in embryonic development in the mesoderm as well as in between the embryonic germ layers19,20. Fn-deficient murine embryos do not undergo primitive heart tube fusion, but instead, form heart tubes with thickened myocardial tissue lacking cardiac jelly and with abnormal endocardium at E8.021. Fn deficiency is in fact lethal within the first 10 embryonic days due to cardiovascular and vascular defects22. In zebrafish embryos, Fn deposition at the midline is required for the myocardial precursor migration and the formation of adhesion junctions among these cells11,12. Our results are consistent with these previous studies that support the idea that the fibrils of Fn is required for the migration of myocardial precursors and indicate that Fn deposition is indeed a downstream effector of Snai1b during cardiac development. Furthermore, in this study, the snail1b morphants show delayed fusion of the myocardial sheets and seemingly smaller hearts (Figure 2). However, the cardiac defects of natter/fn mutants are described as a lack of myocardial fusion at 24 hpf and single myocardial cells located in the anterior mesoderm12. The difference mainly results from the fact that Fn deposition was reduced, rather than completely inhibited in snail1b morphants. Indeed, cardiac precursors are partially surrounded by Fn deposition in these morphants.

Previous studies have shown that Fn accumulation is dependent on α5 integrin in zebrafish embryos14,23. Overexpression of snail increases the expression of α5β1 integrin24. Our results display here that Snai1b regulates α5 integrin expression and controls Fn deposition, rather than expression, in the heart region of zebrafish embryos. Injection of α5 integrin mRNA or α5β1 integrin protein is able to rescue the defect in Fn layer formation in Snai1b knockdown embryos, as well as to restore the migration of myocardial precursors. The results provide strong evidence that Snai1b regulates the expression of α5 integrin to modulate Fn assembly required for myocardial precursor migration during zebrafish embryonic development.

Previous data show that the somites still formed but had an abnormal shape in the snai1b morphant5. In consistent with the data, we also identified that morphological segmentation boundary of somite is disturbed and also Fn assembly was disrupted significantly in the snai1b morphant (Supplementary Fig. S7). Since both α5 integrin and Fn play important roles in morphological segmentation boundary of somite in zebrafish embryos14, it is reasonable to propose that snail-integrin signaling exists wider than the heart field to control embryonic morphogenesis in other tissues and organs during embryo development. In present works, we observe that Snai1b is involved in the movements during embryonic gastrulation as described previously5. Their data show that snail1b plays a role in anterior migration of the cell in the axial mesendoderm during gastrulation, via controlling the cell-cell contact mediated by E-cadherin5, suggesting there are other signals controlled by snai1b to mediate anterior migration of the cell during gastrulation and the Snai1b involved signaling is required to be investigated in the future work.

Methods

Zebrafish strains

Zebrafish were raised and maintained following standard procedure25. Wild-type zebrafish belonged to the AB strain. The transgenic lines used were Tg(myl7::GFP)26 and Tg(flk1::GFP)27.

Antisense morpholino and mRNA or protein injections

Antisense morpholino oligonucleotides (MO) were purchased from Gene Tools (Philomath, OR, USA) and used as previously described28. An MO with the sequence 5′-CCTCTTACCTCAGTTACAATTTATA-3′ was used as a control (Ctl MO). The snai1b MO, which blocks snai1b translation by hybridizing to bases -28 to -4, was used as previously described5. The zebrafish snai1b coding region was cloned using PCR and was ligated into the pcDNA3.1(+) vector or pEGFP-N1 vector. The origin PCR primers for snai1b full sequence were as follows:

Sense primer: 5′ GCTGAAGTTTCGAGGGGATATT 3′

Anti-sense primer: 5′ CCACTAGAGCGCCGGACAGC 3′

Sense primer (ligate into pcDNA3.1(+) vector):

5′ CCGGAATTCATGCCACGCTCATTTCTTGT 3′

Anti-sense primer (ligate into pcDNA3.1(+) vector):

5′ CCGCTCGAGGAGCGCCGGACAGCAGCC 3′

Sense primer (ligate into pEGFP-N1 vector):

5′ CCGGAATTCGCTGAAGTTTCGAGGGGATATTTACA 3′

Anti-sense primer (ligate into pEGFP-N1 vector):

5′ CGGGATCCCGGAGCGCCGGACAGCAGCC 3′

The integrin α5 mRNA was as previously described14. The snai1b and integrin α5 mRNA was obtained using the mMessage mMachine T7 Kit (Invitrogen). Embryos were injected at the one-cell stage with 3 ng snai1b MO or 10–200 pg mRNA per embryo. The integrin α5β1 protein was bought from R&D systems (3230-A5-050).

In situ hybridization and histology

Whole-mount in situ hybridization (ISH) or the double ISH with DAB and NBT/BCIP was carried out as previously described29,30. The antisense RNA probe was synthesized from the relative cDNA with a digoxygenin (DIG) or fluorescein (FLU) RNA labeling kit (Roche). In brief, embryos were permeabilized with Proteinase K (10 μg/mL, Promega) and hybridized overnight at 65°C with the DIG-labeled and FLU-labeled antisense probes. After several washes at 65°C and room temperature, DAB staining followed NBT/BCIP (Roche) staining was performed according to the manufacturer's instructions.

Immunohistochemistry

Primary antibodies used in this study included: rabbit polyclonal anti-fibronectin antibody (1:200; Sigma, F3648), mouse anti-GFP antibody (1:500, Novus Biologicals, NBP1-47583). All Alexa-fluor-labeled secondary antibodies were purchased from Invitrogen and used at a 1:800 dilution. Embryos were fixed overnight at 4°C in 4% PFA in PBS and then cut into 10-μm sections. Immunohistochemistry was performed using these sections as previously described12. After washing with PBS, the sections were imaged using a Zeiss fluorescence microscope.

TUNEL assay

Embryos were fixed overnight at 4°C in 4% PFA in PBS, washed 3 times with PBS and then permeabilized with acetone for 5 min. After 3 washes with PBS, terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling (TUNEL) assay was performed using the Millipore In Situ Apoptosis Detection Kit as described31.

The quantitative PCR and Statistical analysis

The quantitative PCR was performed as previously described32. Statistical significance was evaluated with independent samples t-test. Differences were considered significant for p < 0.05.