Notch signaling is implicated in many developmental processes. In our current study, we have employed a transgenic strategy to investigate the role of Notch signaling during cardiac development in the mouse. Cre recombinase-mediated Notch1 (NICD1) activation in the mesodermal cell lineage leads to abnormal heart morphogenesis, which is characterized by deformities of the ventricles and atrioventricular (AV) canal. The major defects observed include impaired ventricular myocardial differentiation, the ectopic appearance of cell masses in the AV cushion, the right-shifted interventricular septum (IVS) and impaired myocardium of the AV canal. However, the fates of the endocardium and myocardium were not disrupted in NICD1-activated hearts. One of the Notch target genes, Hesr1, was found to be strongly induced in both the ventricle and the AV canal of NICD1-activated hearts. However, a knockout of the Hesr1 gene from NICD-activated hearts rescues only the abnormality of the AV myocardium. We searched for additional possible targets of NICD1 activation by GeneChip analysis and found that Wnt2, Bmp6, jagged 1 and Tnni2 are strongly upregulated in NICD1-activated hearts, and that the activation of these genes was also observed in the absence of Hesr1. Our present study thus indicates that the Notch1 signaling pathway plays a suppressive role both in AV myocardial differentiation and the maturation of the ventricular myocardium.
INTRODUCTION
The heart is the first functional organ to be formed during mouse development. Cardiac development is a complex process that requires myogenesis and morphogenesis to occur simultaneously with contractility. In addition,distinct cell populations have to be integrated in a temporally and spatially precise fashion. Cardiac mesoderm is specified in the anterior lateral mesoderm, which is a primary heart field, and then converges to form a linear heart tube along the ventral midline of the embryo. The heart tube grows rapidly in length through the addition of cells from the second heart field(Buckingham et al., 2005), and balloons to develop its atrial and ventricular compartments during looping. At the end of loop formation, the atria and ventricles are aligned and connected with each chamber through the atrioventricular (AV) canal. It has been suggested that the formation of this connection might involve rightward expansion of the AV canal (de la Cruz and Miller, 1968), leftward remodeling of the primary ventricular septum (Wenink, 1981) or reorganization of the AV myocardium (Kim et al., 2001a). The AV canal forms the AV cushions that are the primordia of the valves and membranous septae. They first become evident as localized swellings in the cardiac jelly, which are subsequently invaded by endothelial cells. The properties of both the AV endocardium and AV myocardium are distinctive in the AV canal. Recently, Notch signaling has been implicated in many aspects of heart development.
Notch is a transmembrane receptor and consists of several functional domains, a series of EGF and Notch/Lin12 repeats in the extracellular region,a transmembrane domain, a RAM domain, CDC10/ankyrin repeats and a PEST domain in the intracellular region (Reaume et al., 1992). Binding of Notch ligands, including delta-like (Dll)and jagged (Jag) proteins in mammals, to the corresponding Notch receptors leads to the stepwise cleavage of the receptor by specific proteases. As a result of this processing, the Notch intracellular domain (NICD) is released and transferred to the nucleus. The NICD then interacts with RBPjκ [also known as Su(H), CBF-1 and Lag-1; Rbpsuh - Mouse Genome Informatics], and regulates the transcription of the bHLH genes hairy/enhancer of split(Hes) and its related protein Hesr (also known as Hey, HRT,CHF, HERP), which then function as transcriptional suppressors of downstream targets (Artavanis-Tsakonas et al.,1999; Iso et al.,2003). In the mouse, Notch1 begins to be expressed in the notochord and mesodermal tissues, including the posterior mesoderm, splanchnic mesoderm and extra-embryonic mesoderm at the primitive streak stage, and the Notch1 expression profile in the heart is restricted to the endocardium by E8.0 (Williams et al.,1995). Notch4 and one of its ligands, Dll4, are also expressed in the endocardium(Shirayoshi et al., 1997; Timmerman et al., 2004). Notch2 expression in the heart cannot be detected by in situ hybridization during the early stages of development(Hamada et al., 1999), but the Notch2 protein is present in the atrial and ventricular myocardium at E13.5(McCright et al., 2002). Although Notch3 is expressed in the cardiogenic plate at the early headfold stage, it is no longer expressed at E8.0(Williams et al., 1995).
Among the aforementioned Notch expression patterns, the restricted Notch1 and Notch4 expression in the endocardium during early heart development indicates that there is a crucial role for Notch signaling in endocardial development. Moreover, both Notch1 and Rbpsuh-null mice have revealed in a previous study that Notch1/Rbpsuh signaling is essential for endocardial development and for EMT in the AV cushions (Timmerman et al.,2004). Recently, studies in human have shown that Notch1mutations cause defects in aortic valve formation(Garg et al., 2005). Moreover,the possible downstream target genes of Notch signaling, Hesr1 and Hesr2, are crucial factors during cardiac development. These downstream genes show a complementary expression pattern in the heart: Hesr1 is expressed in the atrium, outflow tract (OFT) and endocardium, and Hesr2 is expressed in the ventricle(Leimeister et al., 1999; Nakagawa et al., 1999). Hesr2-null mice show defects in AV valve formation, and atrial and ventricular septal formation in the heart(Donovan et al., 2002; Gessler et al., 2002; Kokubo et al., 2004; Sakata et al., 2002). Intriguingly, Hesr1/Hesr2 double knockout mice show a severe phenotype of impaired trabeculation, EMT and septation of the heart(Kokubo et al., 2005),although Hesr1-null mice do not show any detectable defects(Fischer et al., 2004; Kokubo et al., 2005).
The suppressive roles of Notch signaling have been reported during myocardial development in a number of different species. In Xenopusembryos, Notch signaling suppresses the expression of myocardial genes, as a result of which the heart precursor cells do not contribute to the myocardium(Rones et al., 2000). Similarly, in the Drosophila heart it has been shown that Notch activity, which is mediated by Su(H), prevents myocardial cell fate determination (Han and Bodmer,2003; Park et al.,1998). Consistent with this finding, Rbpsuh-null ES cells show increased cardiomyogenic differentiation, which is likely to be due to the lack of Notch signaling (Schroeder et al., 2003). However, it is not yet clear whether Notch1 activity is involved in the determination of the cardiac fate in the mouse, as it is in both Xenopus and Drosophila.
A knockout strategy is both a straightforward and powerful method for elucidating gene function but transgenic strategies that use ectopic expression also yield valuable information. To elucidate the significance of the restricted Notch signaling in the endocardium, we investigated the effects of the forced expression of Notch1 in the myocardium, where the Notch signals are normally inactive. We speculated that this may provide an insight into the predominant expression of Notch1 in the endocardial cell lineage. For this purpose, we introduced NICD1 into the cardiac lineage of the mouse using the Cre-loxP system in Mesp1-Cre mice (in which a Cre recombinase gene is knocked into the Mesp1 locus). Mesp1 is a bHLH transcription factor that is initially expressed in the invaginating mesoderm at the onset of gastrulation. Lineage analysis using Mesp1-Cre revealed that Mesp1-expressing cells mainly contribute to the endocardium and myocardium of the heart and to the endothelial cells of the embryonic and extra-embryonic blood vessels(Saga et al., 2000; Saga et al., 1999). The exception is observed in mesenchymal cells in the OFT derived from neural crest cells and some of the cells in the peripheral cardiac conduction system(Kitajima et al., 2006). In our present study, the expression of NICD1 in the entire cardiac lineage of the mouse has allowed us to determine the outcome of the forced expression of Notch1 in the myocardium lineage. The fates of the endocardium and myocardium were found not to be disrupted in NICD1-activated hearts but the forced activation of Notch signaling in myocardium results in the suppression of both the AV myocardial differentiation and the maturation of the ventricular myocardium.
MATERIALS AND METHODS
Generation of transgenic mice
The CAG-CAT-NICD1 construct was generated by substituting NICD1 (Takahashi et al.,2000) for β-galactosidase in the CAG-CAT-Z vector(Yamauchi et al., 1999). This construct was injected into fertilized eggs and permanent transgenic lines were established. The generation of Mesp1-Cre mice and Hesr1-null mice has been described previously(Kokubo et al., 2005; Saga et al., 1999). NICD1-activation of the cardiac lineage in mice was achieved by crossing CAG-CAT-NICD1 mice with Mesp1-Cre mice. A TOPGal transgenic mouse line was established by injection of a vector containing E. coli β-galactosidase with the TOPflash promoter-enhancer(Upstate).
Histological analyses
For histological observations, Hematoxylin and Eosin staining was conducted on paraffin sections and ultrastructures were then observed by transmission electron microscopy (Miyagawa-Tomita et al., 1996). In situ hybridization analyses were performed using cRNA probes for Notch1, Jag1, Bmp6, Hesr1, Hesr2, troponin 1 fast-twitch skeletal muscle isoform (Tnni2), Wnt2, Anf,chisel (Smpx - Mouse Genome Informatics), Bmp2, Tbx2, Cited1,Hand1 and Bmp10. The InsituPro system (M&S Instruments) was used for whole-mount in situ hybridizations according to the manufacture's instructions. Section in situ hybridizations were performed using 20 μm frozen sections. Activated-Notch1 was detected using an anti-cleaved Notch1 antibody (Val1744) (#2421, Cell Signaling Technology) with 6 μm paraffin sections. Other immunohistochemical detections were performed using anti-Myosin (Skeletal, Slow) (#M8421, Sigma) which is highly specific for the slow myosin heavy chain (Mhc), anti-α Smooth Muscle Actin (αSma)(#A2547, Sigma) and anti-CD31 (Pecam-1) (#557355, BD Biosciences) antibodies with 8 μm frozen sections.
RT-PCR analysis
Total RNA was isolated from the atria and ventricles of E10.5 mouse hearts using an RNeasy mini kit (Qiagen). cDNA was generated using SuperScript II reverse transcriptase (Invitrogen). PCR was performed using primers for Hesr1 (5′-ACGACATCGTCCCAGGTTTTG-3′ and 5′-GGTGATCCACAGTCATCTGCAAG-3′), Hesr2(5′-GCTACAAGCTCAGTGATGAGG-3′ and 5′-GCCTGGAGCATCTTCAAATGATCC-3′), Hes1, Hes5(Kaneta et al., 2000), Tnni2 (5′-CCAGCACTGCTGCACAGCA-3′ and 5′-AGACATGGAGCCTGGGATG-3′), Wnt2(Monkley et al., 1996), Bmp6 (5′-AGCAACTAGCAATCTGTGGG-3′ and 5′-CGTTGTAGTCTGAAGAACCG-3′) and Jag1(Timmerman et al., 2004). The number of PCR cycles was optimized for each reaction.
GeneChip analysis
Ventricles with an AV canal were isolated at developmental stage E10.5 and stabilized in RNAlater RNA Stabilization Reagent (Ambion), prior to total RNA preparation. Total RNA isolates were purified using the RNeasy mini kit (Qiagen), according to the manufacturer's instructions. First-strand cDNAs were synthesized by incubating 5 μg of total RNA with 200 U SuperScript II reverse transcriptase (Invitrogen) and 100 pmol T7-(dT)24 primer[5′-GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-(dT)24-3′]. After second-strand synthesis, the double-stranded cDNAs were purified using a GeneChip Sample Cleanup Module (Affymetrix), according to the manufacturer's instructions. Labeling of the double-stranded cDNAs was achieved by in vitro transcription using a BioArray HighYield RNA transcript labeling kit (Enzo Diagnostics, Farmingdale, NY). The labeled cRNA was then purified using a GeneChip Sample Cleanup Module (Affymetrix) and treated with 1×fragmentation buffer (40 mM acetate, 100 mM KOAc, 30 mM MgOAc) at 94°C for 35 minutes. For hybridization to a GeneChip Mouse Genome 430 2.0 Array(Affymetrix), 15 μg of fragmented cRNA probe was incubated with 50 pM control oligonucleotide B2, 1× eukaryotic hybridization control (1.5 pM BioB, 5 pM BioC, 25 pM BioD and 100 pM Cre), 0.1 mg/ml herring sperm DNA, 0.5 mg/ml acetylated BSA and 1× manufacturer-recommended hybridization buffer in a 45°C rotisserie oven for 16 hours. Washing and staining were performed with GeneChip Fluidic Station (Affymetrix) using the appropriate antibody amplification, washing and staining protocol. The phycoerythrin-stained arrays were scanned as digital image files and scanned data were analyzed with GeneChip Operating Software (Affymetrix). The GeneChip analyses were performed using two different RNA samples prepared from the control and NICD1-activated hearts, respectively. All data are available online(http://www.nihs.go.jp/tox/TtgSubmitted.htm)in the National Institute of Health Sciences.
RESULTS
Notch1 signal activation in the cardiac cell lineage of the mouse induces heart abnormalities
To investigate the possible role of Notch1 signaling during cardiac development in the mouse, we generated a transgenic line expressing NICD1 cDNA, which comes under the control of the CAG promoter after excision of the floxed CAT gene by Cre recombinase (designated as CAG-CAT-NICD1 mice). We used Mesp1-Cre mice to induce NICD expression in the cardiac cell lineage in transgenic progeny. By crossing with Mesp1-Cre mice, Cre/lox specific CAT excision occurs in Mesp1-expressing cells and results in the induction of NICD1-expression in the cardiac lineages (designated as NICD1-activated mice). Whole-mount in situ hybridization experiments subsequently revealed a appreciable level of Notch1 induction in the hearts of NICD1-activated mice (NICD1-activated hearts), which exhibited a deformed shape with a rough surface morphology(Fig. 1C,D). The transgenic mice did not develop beyond embryonic day 10.5 (E10.5) and died before E11.5. These heart abnormalities are clearly distinguishable from wild-type mice(Fig. 1A,B).
To confirm Notch1 activation, we performed immunohistochemical staining using antibodies against processed-NICD1. In wild-type hearts, Notch1 activation was observed only in the endocardium(Fig. 1E-H), but neither in myocardium or mesenchymal cells derived from the endocardium of the AV canal. By contrast, activated Notch1 was observed in the entire heart, including the endocardium, the myocardium of both the atria and ventricles, and the AV cushions, in NICD1-activated embryos (Fig. 1I-L). We were therefore able to conclude that Notch activation was successfully induced in the entire cardiac lineage in our double transgenic mice.
Myocardial defects in the NICD1-activated heart
Histological observations by H&E staining, followed by marker analyses,revealed four major defects in NICD1-activated hearts: (1) impaired ventricular myocardial differentiation; (2) the ectopic appearance of cell masses in the AV cushion; (3) right-shifted IVS; and (4) impaired myocardium of the AV canal. Aberrant myocardial trabeculation was the first anomaly to be observed (Fig. 2A,B). In the wild-type mouse heart, the trabeculae extend to the inner space of the ventricle from the compact layer that has a thickness of two to three cells(Fig. 2A). By contrast, the trabeculae were found to have accumulated in the ventricular wall in NICD1-activated hearts (Fig. 2B). The abnormal accumulation of trabecular cells near the compact layer is likely to be the cause of the rough external appearance of the ventricle (Fig. 1D). To investigate possible alterations in the properties of the trabecular cells in the NICD1-activated heart, we examined the expression of Bmp10, which is a gene that is known to be specifically expressed in the trabecular cells and to be important for the growth of trabecular myocardium(Fig. 2C)(Neuhaus et al., 1999). Bmp10 expression was detected in the NICD1-activated heart,indicating that the trabecular cells were not strongly affected by Notch1 activation (Fig. 2D). However,this expression was not detected in the trabeculae of the AV myocardium(asterisk in Fig. 2D) (see below).
To examine the possibility that Notch1 signaling influences the fate decisions of the myocardium in mouse, as is the case in Drosophilaand Xenopus (Han and Bodmer,2003; Rones et al.,2000), we analyzed the expression of cell-type specific markers by RT-PCR and immunohistochemical staining. Semi-quantitative RT-PCR analysis showed no detectable changes in the expression of the myocardial genes, myosin light chain (Mlc) 2a (Myl7 - Mouse Genome Informatics), Mlc2v (Myl2 - Mouse Genome Informatics), Mhca(Myh6 - Mouse Genome Informatics), Nkx2.5, Mef2c and Gata4 in the ventricle (see Fig. S1 in the supplementary material). Although Mhcb (Myh7 - Mouse Genome Informatics) appears to be induced in the atrium of NICD1-activated hearts, this might be due to contamination of the expanded left ventricle. We also examined protein markers such as Mhc, αSma (for myocardium) and Pecam (for endocardium) using specific antibodies. Although we found no significant changes in the expression patterns of Mhc and αSma, except for the AV canal(Fig. 2F, H, see below), Pecam expression was significantly reduced in the endocardium surrounding the trabecular cells in NICD1-activated hearts(Fig. 2J). Furthermore,Pecam-positive cells were often observed in the interventricular septum (IVS,see arrowheads in Fig. S2 in the supplementary material),which may indicate anomalies in the myocardium of the IVS. To further investigate these transgenic embryos for possible defects in the trabecular cells, we examined their fine structures by transmission electron microscopy. At this stage,myofibrils are found to be well developed in the wild-type mouse heart and the sarcomere structures with Z bands were well organized in the wild-type trabecular myocardial cells (Fig. 2K). In NICD1-activated hearts, however, the myofibrils were poorly formed with an unclear sarcomere structure(Fig. 2L), indicating that myocardial maturation is inhibited in NICD1-activated hearts. These results indicate that Notch1 signaling does not influence the fate of myocardial cells in the mouse, but it acts as an inhibitor of myocardial differentiation, which is associated with abnormal trabeculation in the ventricle.
Appearance of ectopic cell masses in the AV cushion in the NICD1-activated heart
The second principal anomaly of the NICD1-activated heart that we observed occurs in the AV cushion. For the formation of the AV cushion, a subset of endocardial cells that overlie the AV canal undergoes endothelial-mesenchymal transformation (EMT), followed by invasion of the cardiac jelly. In NICD1-activated hearts, mesenchymal cells in the AV cushion were found,indicating that EMT events seemed to occur normally. However, ectopic cell masses were detected in the cushion tissue(Fig. 2B, arrowheads). Immunohistochemical analyses identified that these cell masses were Mhc- andαSma-positive (arrowheads in Fig. 2F,H), but Pecam negative (arrowhead in Fig. 2J), suggesting that these cells possess myocardial properties. It is noteworthy that these cells were never observed in wild-type cushion tissues at E10.5(Fig. 2E,G). To determine how such ectopic cells develop in the NICD1-activated heart, we observed then at E9.5, when it is known that EMT starts to occur in the AV cushion. As is clearly shown in Fig. 2N, we found ectopic cell masses, which were Mhc- and αSma-positive(Fig. 2O,P), in the NICD1-activated heart but not in the wild-type samples(Fig. 2M). Furthermore, serial Hematoxylin and Eosin staining of sections along the AP axis indicated that these cells were derived from the myocardial cells located at the AV canal,where trabeculation does not generally occur at this stage (see Fig. S3 in the supplementary material).
IVS shift and AV myocardial defect in the NICD1-activated heart
The third anomaly of the NICD1-activated mouse heart is the noticeable difference in the size of the ventricles. The left ventricle in the NICD1-activated mouse appears to be expanded, whereas the right ventricle is reduced in size, compared with wild type. In addition, the position of the IVS that separates the right and left ventricles is shifted to the right side in the NICD1-activated mouse (compare arrows in Fig. 2A with 2B). However, no enhanced cell proliferation was observed in the left ventricles of NICD-activated hearts (data not shown), indicating that the property of each ventricle might be affected by Notch1 activation. The expression of Cited1, which is negative in the IVS, revealed differences in size between the right and left ventricles (Fig. 3A-D). We found cells showing reduced Cited1 expression in the NICD1-activated heart (arrows in Fig. 3D), which may indicate abnormalities in cardiomyocyte differentiation. Furthermore, the expression of Hand1, which is a known marker for the left ventricle(Fig. 3E,G), showed expanded expression in the NICD1-activated heart(Fig. 3F,H).
It was also noted that the AV myocardial wall shows evidence of trabeculation in NICD1-activated mice (Fig. 2B,D,F,H,J,N,O,P; Fig. 3D,H). The properties of the chamber and AVC region were examined by molecular analyses using the chamber markers Anf and Smpx(Chisel) to demarcate the ventricles and atria(Fig. 3I-L), and Bmp2and Tbx2 to characterize the AV canal(Fig. 3M-P). In the NICD1-activated heart, AVC marker expression, Tbx2 in particular, is greatly reduced (Fig. 3P). In addition, the Anf and Smpx expression patterns revealed enlarged left ventricles and ambiguous AV boundaries (arrows in Fig. 3J,L). In the wild-type heart, the AV canal allows the endocardium to develop AV cushion tissue by EMT(Fig. 2A, brackets) and even though the AV myocardium shows weak Mhc and αSma staining in the wild-type heart (Fig. 2E,G),the trabeculae never develop in the AV myocardium at this stage. This indicates that the properties of the AV myocardium are distinct from the chamber myocardium. However, in the NICD1-activated heart, AV cushion tissues were formed and the trabeculae could be observed in the expected AV canal,which may indicate a lack of AV myocardial characteristics(Fig. 2B, asterisk).
Hesr1 is ectopically induced in the NICD1-activated ventricle
Hes and Hesr family members are known to be downstream targets of Notch signaling (Iso et al., 2003). We therefore examined whether these genes are induced in NICD1-activated hearts. RT-PCR and in situ hybridization analyses revealed that the expression of Hesr1 and Hesr2 is restricted to the myocardium of the atria and ventricles of wild-type hearts, respectively(Fig. 4A,B,D), which is consistent with previous reports(Leimeister et al., 1999; Nakagawa et al., 1999). However, in NICD1-activated hearts, Hesr1 was found to be ectopically induced at high levels in the myocardium of the ventricle(Fig. 4A,C), whereas Hesr2 was not shown to be induced in the myocardium of the atrium(Fig. 4A,E). Intriguingly, Hesr1 was not expressed in the endocardium at this stage, despite the activation of Notch1 (Fig. 4B, Fig. 1G). A noticeable change was also observed in the AV myocardium of the transgenic hearts, in which the extended trabeculation appeared to be coincident with the co-expression of Hesr1 and Hesr2. As Hesr2 expression occurs in the wild-type ventricle, ectopic Hesr1 expression might be the cause of the ventricle and AV canal defects in the NICD1-activated mice. In addition, Hes5 expression was induced in both the atria and ventricles in Notch1-activated hearts (Fig. 4A), although it is not expressed in the wild-type heart.
NICD1 activation in a Hesr1-null background rescues the AV myocardium defect
Hesr1 is strongly induced in the NICD1-activated mouse heart and is known to be a transcriptional repressor, which lead us to hypothesize that the major cardiac defects observed in NICD1-activated mice might be caused by the ectopic expression of Hesr1. To examine this possibility, we generated NICD1-activated mice in a Hesr1-null background. Hesr1 knockout mice show no detectable defects in heart morphogenesis(Fig. 5A,C,E)(Kokubo et al., 2005). Hence,if the defects in NICD1-activated hearts were due to the suppression of specific genes by elevated Hesr1, we would expect that they would be rescued by the absence of Hesr1. Unexpectedly, however, most of defects observed in NICD1-activated hearts were also elicited by Notch1 activation,even in the absence of Hesr1 (Fig. 5A-D, Fig. 1A-D). In the NICD1-activated/Hesr1-null hearts, impaired ventricular trabeculation and cell masses in the cushion tissue(Fig. 5F, arrowhead) were again observed, as seen in NICD1-activated hearts(Fig. 2B). However, we observed that there was much less trabeculation of the AV myocardium in the NICD1-activated/Hesr1-null background, suggesting that the property of the AV myocardium is recovered (Fig. 5F, brackets), whereas the IVS was again right-shifted(Fig. 5F, arrow). These data indicate that ectopic Hesr1 expression in the myocardium may have caused the defects observed in the AV myocardium.
Wnt2, Bmp6, Tnni2 and Jag1 are induced in NICD1-activated hearts
As it was unlikely that Hesr1 would be the sole mediator of Notch1 signaling, and because Hesr1-independent mechanisms must be involved in the formation of the defects in NICD1-activated hearts, we examined the expression of several genes that could potentially be involved in these myocardial abnormalities. However, following RT-PCR experiments, we could not find any critical changes in the expression levels of the genes that we selected for analysis (Mlc2a, Mlc2v, Mhca, Mhcb, Nkx2.5, Mef2c, Gata4, Nrg1, Erbb4,Ccnd1, Jmj, Snail) (Supplementary Fig. S1). Therefore, to identify candidate genes that are differentially regulated in NICD1-activated hearts,we performed comparative GeneChip analysis using RNA isolates from ventricles with an AV canal from transgenic and wild-type hearts at E10.5. These experiments identified several genes showing differences in their expression levels between wild-type and NICD1-activated mice. Since we had observed that the lack of Hesr1, which is known to function as a transcriptional suppressor(Iso et al., 2001b), only rescues the AV myocardium, we focused on genes that were upregulated in NICD1-activated hearts (see Table S1 in the supplementary material). Among the candidates that we analyzed were Wnt2, Bmp6 and troponin 1 fast-twitch skeletal muscle isoform (Tnni2), which we subsequently confirmed by RT-PCR and in situ hybridization(Fig. 6). In wild-type hearts, Wnt2 is expressed strongly in the posterior region, e.g. the sinus venosus (Monkley et al., 1996)and the atrium, but it is difficult to detect its expression in the ventricle of wild-type hearts (Fig. 6B,B′). However, Wnt2 expression was easily detectable in NICD1-activated hearts (Fig. 6C). Transverse sections also revealed stronger ectopic expression of Wnt2 in the left ventricle, including the AV canal(Fig. 6C′).
Similar expression changes was observed for Bmp6(Fig. 6D-E′), which is normally expressed in the OFT and cushion tissue(Kim et al., 2001b). In wild-type hearts, Tnni2 expression was found to be marginally expressed in the atrium (Fig. 6F,F′), but to be strongly induced both in the atria and the ventricles in NICD1-activated hearts (Fig. 6G,G′). Tnni2 is one of the components of troponin 1, and is known to inhibit the interaction between actin and myosin in the absence of Ca2+ (Clark et al.,2002; Gordon et al.,2000). Hence, the elevated expression of Tnni2 might be one of the leading causes of the induction of the myocardial structural defects in NICD1-activated mice. We observed that one of the known Notch ligands, Jag1, is also induced in NICD1-activated hearts(Fig. 6A). Jag1 is expressed in the atria, right ventricle and OFT of wild-type hearts(Fig. 6H,H′), and this expression was found to be expanded into the whole heart in NICD1-activated mice (Fig. 6I,I′). We also examined the distribution of these genes at E8.5 to determine whether their expression profiles are affected from an earlier stage. Although the upregulation of NICD1 expression was not clear in the NICD1-activated hearts at this stage (see Fig. S4A,B in the supplementary material), we observed increased expression of the above genes and also morphological changes (rough surface) in NICD1-activated hearts (see Fig. S4D,F,H,J in the supplementary material).
As all of these expression changes coincide with the Hesr1expression pattern, we speculated that they are in fact the result of secondary effects of Hesr1 induction. To address this possibility, we examined the expression of each of these genes in the NICD1-activated hearts of Hesr1 knockout mice. As shown in Fig. 6J-M, elevated gene expression was observed for Wnt2, Bmp6,Tnni2 and Jag1, even in the absence of Hesr1. These data indicate that the activation of these genes is either mediated by a factor other than Hesr1 or is directly regulated by NICD1.
DISCUSSION
In this study, we have generated NICD1-activated mice and analyzed their cardiac lineage. NICD1-activated hearts showed abnormal morphologies, such as disrupted cardiomyocyte differentiation, the appearance of ectopic cell masses in the AV cushion, a right-shifted IVS and defects in the AV myocardium. Moreover, these NICD1-activated mice died at E10.5. It has been shown that the AV myocardium is important for connection formation between the right atrium and right ventricle, and the left atrium and left ventricle at the end of looping (Kim et al., 2001a). Furthermore, it has been suggested that the AV node develops from that part of the AV canal myocardium. Therefore, we speculate that the transgenic embryo does not form a complete heart with normal cardiac morphology and conduction system, resulting in death at an early stage. Comparisons with NICD1-activated hearts in a Hesr1-null background demonstrate that ectopic Hesr1 expression is responsible for the differentiation of the AV myocardium. In NICD1-activated hearts, the expression of Wnt2, Bmp6,Tnni2 and Jag1 could be induced in addition to Hesr1,and this might well be the cause of these myocardial abnormalities.
The involvement of Notch signaling in cardiac cell fates
In Xenopus, the expression of Serrate1 (Xenopus Jag1) and Notch1 overlaps in the dorsal side of the heart, and the forced expression of the activated form of Su(H) (XenopusRBPjκ) suppresses the expression of myocardial genes(Rones et al., 2000). This indicates that Jag1-Notch1 signaling in Xenopus has an inhibitory effect upon myocardial fate determination. Although the expression levels of myocardial genes were shown previously not to change in either NICD1-activated, Notch1 or Rbpsuh mutant hearts(Timmerman et al., 2004), our current histological data suggests that Notch1 signaling suppresses myocardial differentiation but does not affect myocardial cell fate decisions. In addition, the ectopic expression of Su(H) in Xenopus embryos results in the reduction of the number of myocardial cells that contribute to the heart (Rones et al., 2000). In the mouse, by contrast, abnormal trabeculation of the AV myocardium was observed in NICD1-activated hearts and this defect was rescued by a lack of Hesr1, which indicates that Notch/Hesr1 signaling might have suppressive effects for AV myocardial differentiation in the mouse.
The suppressive effects of Notch1 signaling on cardiomyocyte maturation
We observed the suppression of myocardial differentiation in NICD1-activated hearts and confirmed the presence of myofibrillar disorganization in their trabeculae by TEM analyses. Notch1 signaling in the wild-type mouse heart is restricted to the endocardial cells, as we observe Notch1 activation by antibody staining only in the endocardium. In NICD1-activated hearts, however, Notch1 signaling is also activated in the myocardium. Therefore, it is likely that the ectopic Notch1 activation in myocardium may directly suppresses differentiation, but it is also possible that secreted factor(s) from the endocardium (Bmp6 and Wnt2 are the candidates), that might be activated by additional NICD1, negatively regulate cardiomyocyte differentiation. If this is indeed the case, we speculate that Notch1 signaling negatively regulates cardiomyocyte differentiation. However,the effects of Notch1/Rbpsuh signaling upon myocardial development are not fully clear because heart formation itself is severely retarded in Rbpsuh-null mutants (Timmerman et al., 2004). We also have investigated conditional knockout embryos for the Rbpsuh allele driven by Mesp1-Cre. However, the resulting phenotype was similar to that of the Rbpsuh-null embryo and the heart was underdeveloped (data not shown), possibly owing to the early onset of Mesp1.
The role of Notch1 signaling in the AV myocardium
In our current experiments, Mhc, αSma-positive and Pecam-negative cell masses were observed in the AV cushion of NICD1-activated hearts. From our observations of serial sections of NICD1-activated hearts, we postulate that these cell masses could originate from the AV myocardium, where trabeculation is normally prevented in the wild-type AV canal. Bmp2 is a crucial factor in the determination of AV myocardial identity by regulating Tbx2 and its downstream genes, including Anf and Smpx (Ma et al.,2005). In the NICD1-activated heart, the decreased expression of Bmp2 and Tbx2 in the AV myocardium should cause the loss of an AV identity, which in turn might confer the chamber identity to the AV canal, as evidenced by ambiguous Anf and Smpx expression. This may also induce ectopic trabeculation in the AV myocardium. The other possibility that we have considered for the origins of the cell masses are the endocardial cells. This is based upon the severe defects that can be observed in endocardial development and EMT in both Notch1 and Rbpsuh-null mice (Timmerman et al., 2004). Furthermore, it has been reported that Notch1 activation induces the transformation of either AV canal explants or endothelial cells to αSma-positive cells(Noseda et al., 2004; Timmerman et al., 2004). Although it is not clear whether these transformed cells also express Mhc, the cell masses in the NICD1-activated AV cushion might well be derived from endocardial cells due to excess EMT.
Implications of the upregulated genes identified in NICD1-activated mouse hearts
In vitro experiments have indicated that both Hesr1 and Hesr2 are induced by NICD1 (Iso et al., 2002; Iso et al.,2001a; Nakagawa et al.,2000). However, in NICD1-activated hearts, Hesr1 was strongly induced but not Hesr2, suggesting that the regulatory mechanism is different between these genes. A possible function for the Hes proteins in these events, as they are also induced in NICD1-activated hearts,will need to be addressed in future studies.
In addition to Hes family genes, we have found that many genes are induced in NICD1-activated hearts. Among these, the upregulation of Tnni2might be responsible for the observed myocardial defects. In the mouse heart,the most abundant troponin 1 is Tnni3 (cardiac Tnni), but Tnni2 is also transiently expressed in embryonic hearts from E9.5 to E16.5 (Zhu et al., 1995). Tnni2 mutant mice have not been reported, but Tnni3 mutant mice have been generated and show shortened sarcomere lengths(Huang et al., 1999). A single Tnni gene has been identified in Drosophila; its mutant, hdp3, exhibits impaired sarcomere structure(Nongthomba et al., 2004). The strong induction of Tnni2 by NICD1 may therefore be one of main causes of the disrupted sarcomere structure in NICD1-activated hearts. Wnt2 is normally expressed in early cardiac mesoderm and thereafter in the sinus venosus and OFT regions(Monkley et al., 1996). However, signals detected using a LEF/TCF reporter (TOPGal) did not differ between wild-type and NICD1-activated hearts (see Fig. S5 in the supplementary material), which suggests that Wnt2 uses a non-canonical Wnt pathway in the mouse heart. Recently, it has been shown that a non-canonical Wnt11 signaling pathway is crucial for cardiogenesis and cardiomyocyte differentiation(Pandur et al., 2002; Terami et al., 2004), and that its expression pattern overlaps in the OFT with Wnt2. These non-canonical Wnt signaling mechanisms must therefore be involved in cardiogenesis.
We also demonstrate in our current experiments that there is a strong induction of Bmp6 by Notch1 signaling. The importance of BMP6 signaling during cardiac development was previously reported by the analysis of Bmp6/Bmp7 double-null mice, which showed retarded OFT cushion development, reduced trabeculation and failure of septation(Kim et al., 2001b), although single mutations of Bmp6 or Bmp7 did not induce any defects in the heart (Dudley et al.,1995; Luo et al.,1995; Solloway et al.,1998). As Bmp signaling inhibits myogenic differentiation synergistically with Notch1 signaling in C2C12 cells(Dahlqvist et al., 2003), this synergistic effect might lead to the suppression of myocardial differentiation in NICD1-activated ventricles.
Although NICD1 is activated in both sides (left and right) of the atrium and ventricle in the NICD1-activated heart, the strong induction of Wnt2 and Bmp6 was restricted to the left ventricle, whereas Tnni2 and Jag1 are induced in both sides of the ventricle. The reason for the restriction of Wnt2 and Bmp6 is currently unknown, but it may suggest the different responsiveness between the left and right ventricles to NICD1. The upregulation of these genes in NICD1-activated hearts was not mediated by Hesr1 and therefore might be regulated by direct binding of Rbpsuh to the enhancer region or by other Hes genes. Multiple putative binding sites for Su (H) (Rbpsuh) (more than 80% homology to consensus sequence) and also Hairy-binding sites (N-box sites) are present within the 10 kb upstream and downstream flanking regions of the Wnt2,Bmp6, Tnni2 and Jag1 transcription start sites (TFSEARCH; http://mbs.cbrc.jp/research/db/TFSEARCH.html). However, further enhancer analyses will be required to determine their functional relevance.
Our present study was initially designed to further understand the function of Notch signaling during cell fate decisions by ectopically expressing activated Notch1 in their precursors. Although we did not observe any cell fate changes during cardiac development, our detailed analyses of transgenic mouse phenotypes and downstream target genes enabled us to uncover several important aspects of Notch signaling in cardiac development. The functional differences between Hesr1 and Hesr2 are now of great interest in these events,and understanding the crosstalk between Notch and other signaling pathways,such as Wnt or Bmp, are obviously crucial for correct heart development. Further studies using gain- or loss-of-function experiments will now be required to fully elucidate the molecular mechanisms underlying cardiac development in the mouse.
Supplementary material
Acknowledgements
We are grateful to Dr Hiroaki Nagao for the electron microscopic analysis,to Dr Satoshi Kitajima for helpful advice and discussions about Microarray analysis, and to the Mesp1-lineage experiments and Dr Tasuku Honjo for generously providing the Rbpsuh mutant mice. We would also like toacknowledge our colleagues who provided us with cDNAs for use as probes,including Drs R. A. Conlon (Notch1), T. A. Mitsiadis (Jag1),I. Satokata (Anf), M. Shirai (Tbx2), M. E. Dickinson(Bmp2) and T. Takeuchi (Hand1). We also thank Ms Yuki Takahashi and Ms Yuka Sato for technical assistance, and Mr Okamura and Mr Oginuma for their support. This work was supported by the Organized Research Combination System and National BioResource Project of the Ministry of Education, Culture, Sports, Science and Technology, Japan.