Tbx5 specifies the left/right ventricles and ventricular septum position during cardiogenesis

T-box (Tbx) genes are involved in various aspects of pattern formation in both vertebrate and invertebrate embryos. They encode transcription factors that are characterized by a highly conserved DNA-binding domain (T-box) and an unusual mode of DNA recognition (Kispert and Herrmann, 1993; Muller and Herrmann, 1997; Smith,1999). The biological functions of Tbx genes have been investigated by misexpression by in ovo electroporation in chick and gene targeting in mice (Takeuchi et al.,1999; Rodriguez-Esteban et al., 1999; Koshiba-Takeuchi et al., 2000; Wilkinson et al.,1990; Chapman and Papaioannou,1998; Jerome and Papaioannou,2001; Lindsay et al.,2001; Merscher et al.,2001; Bruneau et al.,2001; Russ et al.,2000).

One of most well-characterized Tbx genes is Tbx5. This gene is expressed in developing forelimb buds and on the dorsal side of the retina. We and another group have reported that Tbx5 is a crucial determinant of wing (forelimb) (Takeuchi et al.,1999; Rodriguez-Esteban et al., 1999). In addition, Tbx5 regulates pattern formation of the eye and also the retinotectum projection along its dorsoventral axis(Koshiba-Takeuchi et al.,2000). The roles played by Tbx5 during heart development,however, remain unclear, although mutations of human TBX5 have been found in individuals with Holt-Oram syndrome (OMIM 142900)(Basson et al., 1997; Li et al., 1997; Basson et al., 1999). In such individuals, characteristic defects of the upper limb and heart are observed(Holt and Oram, 1960). Precise analysis suggests the haploinsufficiency of TBX5 in Holt-Oram syndrome. Similar haploinsufficiency was reported for other T-box genes, such as human TBX3 and TBX1(Jerome and Papaioannou, 2001; Lindsay et al., 2001; Merscher et al., 2001; Bamshad et al., 1997; Epstein, 2001), indicating that the levels of T-box proteins are crucial for normal functioning(Bruneau et al., 2001; Hatcher and Basson, 2001).

Recently, mouse Tbx5 was knocked out to generate heterozygous and homozygous mice (Bruneau et al.,2001). Homozygous Tbx5^del/del mice do not survive past embryonic day 10.5 (E10.5) because of the arrest in heart development at E9.5. By contrast, heterozygous Tbx5^del/+mice show several morphological alterations in both the heart and the forelimb. In such deformed hearts, large atrial septum defects (ASDs) and ventricular septum defects (VSDs) are observed as a variety of complex cardiac malformations. In addition, abnormalities of the cardiac conduction system are found. These lines of evidence highlight the multiple roles of Tbx5in heart development and the haploinsufficiency of Tbx5, providing a valuable model of congenital heart diseases(Bruneau et al., 2001).

Expression patterns of Tbx5 were reported previously(Bruneau et al., 1999; Yamada et al., 2000). In both chick and mouse, Tbx5 is expressed in the precardiac mesoderm. This expression then becomes restricted to the posterior part of the looping heart tube. Later, Tbx5 expression is restricted to the atria and the left ventricle, and a ventricular septum is formed at the boundary of Tbx5-expressing and non-expressing domains. Hence, this gene would provide a novel and valuable marker to explore the mechanism of ventricular specification. Recently, interesting heart phenotypes of transgenic mice have been reported. For example, the Tbx5 gene was overexpressed ubiquitously in the primitive heart tube under the control of a β-myosin heavy chain promoter. Persistent expression resulted in heart looping defects,abnormalities of early chamber development and loss of ventricular-specific gene expression, indicating an essential role for Tbx5 in early heart development (Horb and Thomsen,1999; Hatcher et al.,2001; Liberatore et al.,2000). Nonetheless, the premature death of such transgenic mice prohibited precise analysis of cardiac development.

Vertebrates exhibit different heart morphologies: fish have one ventricle/one atrium, whereas birds and mammals have two ventricles/two atria. Considering the left ventricle-specific expression of Tbx5, this gene could be useful for exploring the evolution of vertebrate hearts and could provide valuable insights on the ventricle specification, onset of congenital heart diseases and evolution of vertebrate heart morphology. For this purpose,we modified our in ovo electroporation techniques to optimize efficient expression of transgenes in the developing heart and analyzed the functional roles of the Tbx5 gene during both chick and mouse cardiac development in detail. In addition, we report that chick Tbx20 is expressed in the right ventricle, showing a mutually exclusive pattern to Tbx5. Our data provide important insights on a combinatorial expression patterns of Tbx genes in the developing heart and their putative interaction.

In ovo electroporation was carried out as described previously(Takeuchi et al., 1999; Momose et al., 1999). We modified our in ovo electroporation techniques to obtain efficient expression of transgenes in the heart. Briefly, a CUY-21 electroporator (Gene System,Osaka, Japan) was used. Fertilized eggs were purchased from Takeuchi and Yamagishi poultry farms (Nara, Japan). A small window was opened for access. PBS(–) was poured over the embryo to obtain appropriate resistance (1.0 kΩ). Two platinum electrodes (Gene System, Osaka, Japan) were fixed in parallel or in a cross to obtain ubiquitous or restricted expression of the transgene, respectively. An anode was inserted beneath the embryonic endoderm. A cathode was placed on the surface of ectoderm, and the DNA solution was injected by a sharp glass pipette into the space between the embryonic ectoderm and the precardiac mesoderm. The positions of electroporation were selected according to the cardiac fate map(Montgomery et al., 1994; Redkar et al., 2001; Fishman and Chien, 1997). Electric pulses were applied (5 V, 40 mseconds, three times) during injection of the DNA solution into the space between the endoderm and the mesoderm. After the eggshells were sealed, the embryos were allowed to develop in humidified incubators. Expression plasmids were constructed in the RCAS retroviral vector as described previously(Motgan and Fekete, 1996).

In situ hybridization was performed as described previously(Wilkinson, 1993). Probes for ANF, BMP2, Tll1 and VEGF were amplified by RT-PCR using primers derived from the published sequences (GenBank Accession Number X57702,X75914, U75331 and S79680, respectively). Zebrafish probes for Tbx5 and Tbx20 were also amplified by RT-PCR based on the published sequences (GenBank Accession Number AF152607 and AF253325, respectively).

Transgenic mice were generated by pronuclear injection of plasmids into fertilized eggs as described previously(Saijoh et al., 1999). The injected embryos were transferred into pseudopregnant recipients and allowed to develop in utero. Embryos were examined for the presence of the transgene by PCR. Four primers were used for PCR analysis:5′-GAGTTCCCCAAGTGAATGAAA-3′ and 5′-GCAGACATTCAGTGGACT-3′ for the β-MHC construct(Liberatore et al., 2000), and 5′-TGGTGAACCGCATCGAGCTGAAG-3′ and 5′-CGTCCTCGATGTTGTGGCGGATC-3′ for the MLC-2ν construct(Ross et al., 1996).

Zebrafish Tbx20 full-length sequence was amplified by RT-PCR, then subcloned in pCAGGS expression vector(Niwa et al., 1991). Human ANF promoter was amplified by PCR using primers from the published sequence (1539 bp upstream fragment from the initiation codon of ANF). This region was been reported to contain three Tbx5-binding sites (Bruneau et al., 2002). Transfection was performed based on the Polyethylenimine (PEI)-mediated gene delivery method(Boussif et al., 1995). Luciferase and β-gal assays were carried out as described previously(Ogura and Evans, 1995).

In the chick, precardiac precursor cells are formed in two crescent areas at the rostral part of the embryo. These cells form the precardiac mesoderm,which enlarges and fuses to form a heart tube. After a looping event, both atrial and ventricular chambers develop, with two septa separating blood flow(Montgomery et al., 1994; Redkar et al., 2001). As reported previously (Bruneau et al.,1999; Yamada et al.,2000), Tbx5 expression is evident from the very early stages. First, Tbx5 is expressed in a gradient fashion, with the strongest signal at the caudal part of the heart tube. This expression then becomes restricted to the left ventricle and atria at stage 30(Fig. 1A). Tbx5 is expressed in the left side of the ventricular septum, leaving a Tbx5-negative half in the right(Bruneau et al., 1999). This expression pattern strongly suggests that Tbx5 specifies the identity of the left and right ventricles and that the ventricular septum is formed at the boundary of Tbx5-positive and -negative regions (black arrowhead in Fig. 1A) to separate the aortic and pulmonary circulation.

In search of putative Tbx gene(s) expressed in the Tbx5-negative right ventricle, we have found that chick Tbx20 shows a complementary pattern of expression in the heart ventricle (Fig. 1B). Hence its expression is confined to the right ventricle and abuts that of Tbx5at the position of the ventricular septum indicated by a black arrowhead in Fig. 1B. These observations strongly suggest that the right and left ventricles are specified by the distinct Tbx genes expressed in an asymmetric fashion.

Recently, physical interactions between several heart-specific transcription factors have been reported to play pivotal roles during cardiac development (Bruneau, 2002; Nemer and Nemer, 2001). Tbx5 associates directly with Nkx2.5 to promote cardiomyocyte differentiation(Bruneau et al., 2001; Hiroi et al., 2001). Nkx2.5 interacts with GATA4 to regulate several cardiac genes(Durocher et al., 1997; Lee et al., 1998). In addition, physical interactions among Tbx2, Tbx5 and Nkx2.5 are crucial for the cardiac development (Habets et al.,2002). These lines of evidence indicate that the tight crosstalks among heart-specific factors control the orchestrated cardiac development. As expected, two key factors, Nkx2.5 and GATA4 genes, are expressed ubiquitously in the developing heart ventricles(Fig. 1C,D).

To explore the roles played by Tbx5, we exploited our in ovo electroporation technique to misexpress this gene in the developing chick heart. With our system, the configurations of the electrodes and DNA injection enabled us to obtain rapid and targeted expression of transgenes in the heart(Takeuchi et al., 1999; Koshiba-Takeuchi et al., 2000; Ogura, 2002). Using this approach, we misexpressed a Tbx5-GFP fusion gene inserted in the RCAS retroviral vector. As reported previously, we did not detect any functional difference between this fusion gene and the wild-type Tbx5(Takeuchi et al., 1999; Koshiba-Takeuchi et al.,2000). In addition, we repeated electroporation of EGFPgene alone in chick heart, but we did not detect any morphological alteration. When the Tbx5-GFP was electroporated into the precardiac mesoderm of embryos at stage 5, we observed robust GFP signals in the entire heart at E5(Fig. 1G). At this stage, the position of ventricular septum is already evident as a small indentation on the surface of the normal heart (arrowhead in Fig. 1E). By contrast, when Tbx5 was misexpressed ubiquitously, as confirmed by the GFP signals(Fig. 1G), such an indentation was not formed, making the contour of the heart round and smooth(Fig. 1F,H). At E8, we obtained robust GFP signals uniformly in the entire ventricle, whereas the normal heart did not show any GFP fluorescence (Fig. 1I), indicating that the Tbx5 gene was misexpressed in the entire heart. As the indentation that is formed on the surface of a normal heart (arrowhead in Fig. 1H)corresponds to the position of the ventricular septum, this morphological change indicates that septum formation was disrupted in the transgene-electroporated heart (Fig. 1H).

To analyze further, we checked the expression of the chick atrial ANF gene in both the normal and electroporated hearts(Fig. 1J). In the normal heart,the ANF gene is expressed in the left ventricle without any signal in the right as observed in mouse (Zeller et al., 1987). The position of the arrowhead in Fig. 1J indicates the boundary between the left and right ventricles. Contrary to the normal heart, ANF was induced strongly in the entire ventricle when Tbx5was misexpressed in the developing precardiac field. This strongly suggests that the formation of the right ventricle was disturbed by the extensive misexpression of Tbx5 (n=8/11). As reported previously(Bruneau et al., 2001), ANF gene is one of direct targets of Tbx5. As ANF is expressed in the entire ventricle, this also suggests that Tbx5 was successfully misexpressed in the developing heart ventricle. In mammals, ANF gene is induced by cardiac stress as first reported by Burnett et al. (Burnett et al., 1986),ubiquitous induction of ANF gene in Tbx5-misexpressed heart might be due to cardiac stress. Nonetheless, we did not observe cardiac overload in these chick hearts. Rather, hemodynamic observation suggests that circulation is severely disturbed and slow, thereby cardiac overload is not a primary cause of ANF induction, although we do not exclude that possibility that subsequent hypoxic stress might partially contribute to this induction.

To check the effects of Tbx5 misexpression on the right ventricle-specific marker gene, we examined the expression of Tbx20 in the Tbx5-misexpressed heart. As shown in Fig. 1B, in the normal heart, Tbx20 is expressed in the right ventricle with its left limit located at the small indentation (arrowhead in Fig. 1K). By contrast, when Tbx5 was misexpressed, Tbx20 expression disappeared from the whole ventricle, which shows again the round and smooth contour(Fig. 1K) (n=4/11). These results strongly suggest that misexpression of Tbx5 in the entire ventricle induces the ANF expression, and represses Tbx20 in the right ventricle, thereby converting the ANF-off/Tbx20-on profile of the right ventricle to the ANF-on/Tbx20-off profile.

To analyze further the morphological changes induced by misexpression of Tbx5, we made a series of continuous sections of the normal and electroporated hearts at E8.0. In the normal heart, the ventricular septum formed at the boundary of the left and right ventricles(Fig. 2A-E) (n=6). At this stage, extensive trabecular formation was also evident in both the right and left ventricles that are connected to the pulmonary artery and aorta,respectively (Fig. 2A,B). Both left and right atrio-ventricular canals formed normally(Fig. 2B-D).

In contrast, several anomalous alterations were found in the electroporated hearts (Type 1 in Fig. 2F-J)(n=8/37). First, the formation of the ventricular septum was completely inhibited, resulting in a single ventricle(Fig. 2G-J). Second, a large dilatation of the atria was also observed(Fig. 2I,J). Third, the ventricular wall was thinner than that of the normal heart, and the trabecular formation was coarse and rough, suggesting that differentiation of cardiac muscle cells was also disturbed (Fig. 2G-J). Fourth, the aorta and pulmonary artery were fused at the base and connected to the single anomalous ventricle, resulting in a double outlet left ventricle (DOLV). Although the truncal septum was formed, the conal septation and/or the conal rotation seemed to be abnormal(Fig. 2F). These morphological changes suggest that misexpression of Tbx5 induces aberrant differentiation of cardiac muscle cells(Hatcher et al., 2001; Liberatore et al., 2000),suppression of ventricular septum formation and malformation of the conotruncal septum, hence resulting in a complex of cardiac defects, as illustrated in Fig. 2P (Type 1).

We observed another type of morphological alteration in electroporated hearts, in which the Tbx5-positive domain was expanded rightward by electroporation (Type 2 in Fig. 2K-O) (n=17/23). In this case, the left ventricle expanded, and the right ventricle shrank(Fig. 2K-O), with a distinct VSD (arrowhead in Fig. 2N). The relative sizes of the two ventricles suggested that the ventricular septum was shifted to the right. Conal septation and rotation defects were also observed(Fig. 2K-M). The atrial septum was formed, but it was membranous and thin(Fig. 2N). Trabecular formation and the thickness of the ventricular wall were not affected. Observed cardiac defects are illustrated in Fig. 2P (Type 2).

We also misexpressed the Tbx5 gene in a restricted domain of developing hearts. In this case, we placed two electrodes in a cross configuration (Fig. 3A) to target electric pulses to a restricted part of the precardiac field. In addition, we used RCAS virus-incompetent chick embryos to prevent expansion of transgene expression. Even in such hosts, stable and long-lasting expression of Tbx5-GFP was obtained. When an RCAS-Tbx5-GFP construct was injected at stage 5, discrete GFP fluorescence signals were observed in a limited part of the prospective right ventricle at E4.0(Fig. 3B,C). At E7.0, an abnormal indentation (red arrowhead in Fig. 3D) was observed on the right ventricle, corresponding to the domain of the restricted GFP signals (red arrowhead in Fig. 3E). Experimental design is illustrated in Fig. 3F. When Tbx5-GFP was misexpressed in a restricted region, an ectopic boundary of Tbx5-positive and -negative regions was formed in the developing right ventricle, as illustrated in Fig. 3F. In this case, endogenous Tbx5 expression was stronger in the caudal part of the precardiac mesoderm at stage 5, whereas the electroporated Tbx5 was misexpressed in the restricted rostral area that gives rise to the right ventricle. At stage 10, endogenous caudal expression is maintained in the developing heart tube, while the expression of the introduced Tbx5 transgene became evident at its rostral end. At stage 25, three domains were formed from the left side of heart to the right:(1) an endogenous Tbx5-positive domain (the prospective left ventricle), (2) a Tbx5-negative area (the prospective right ventricle) and (3) an ectopic Tbx5-positive area(Fig. 3F). Hence, this restricted expression results in the extra-boundary formation of Tbx5-positive and -negative regions in the prospective right ventricle.

As expected, the ANF gene was induced in the right ventricle(between two red arrowheads in Fig. 3G) (n=9/14). The boundary of the endogenous cANF expression is indicated by a yellow arrowhead in Fig. 3G. In serial sections of these misexpressed hearts, extension of trabeculae was observed with weak expression of BMP2 at their tips(Fig. 3H). This accelerated growth of trabeculae suggests that an ectopic septum was induced at the new boundary of Tbx5 expression, as ANF gene was expressed in a complementary manner to the extension of trabeculae (the ectopic and endogenous cANF expression is indicated by red and black arrowheads in Fig. 3I, respectively).

To analyze the morphological changes in detail, we made serial sections of electroporated hearts at E6 and carried out in situ hybridization using several ventricular septum markers. Contrary to the normal heart, an ectopic septum-like structure (IVS* in Fig. 3J,K) was formed in the right ventricle (IVS in Fig. 3J,K) (n=8/43). To confirm that this ectopic structure is indeed the ventricular septum, we performed in situ hybridization using several septum markers: BMP2,Tll1 (tolloid-like 1) and VEGF (vascular endothelial growth factor) (Lyons et al.,1990b; Lyons et al.,1995; Clark et al.,1999; Tomanek et al.,1999; Miquerol et al.,2000). As expected, all of these markers were expressed(Fig. 3L-N, respectively). These lines of evidence strongly suggest that restricted expression of Tbx5 in the prospective right ventricle induces an ectopic ventricular septum at the new border of Tbx5 expression in the developing right ventricle.

Misexpression experiments in chick hearts indicate that Tbx5specifies the left ventricle, and that the ventricular septum is formed at the boundary of Tbx5-positve and -negative domains. To confirm this hypothesis, we checked the expression of several markers known to be expressed asymmetrically in the heart. As shown in Fig. 1J, chick ANF was induced in the developing right ventricle when Tbx5 was misexpressed. By contrast, Tbx20, which is expressed in the right ventricle, was repressed (Fig. 1K). Nonetheless, in the chick heart, other markers such as the dHAND and eHAND genes are expressed uniformly in both ventricles (data not shown). This prompted us to carry out misexpression studies in mouse hearts.

For this purpose, we used two expression constructs(Fig. 4A) to target transgene expression in developing mouse hearts. One is a mouse Tbx5 expression construct in which the β-MHC (myosin heavy chain) promoter was used to misexpress this gene uniformly in the ventricle(Liberatore et al., 2000; Lyons et al., 1990a). In another construct, we used the MLC-2ν (myosin light chain)promoter, which was reported to drive transgene expression in the right ventricle (Ross et al., 1996). In this construct, the chick Tbx5-EGFP fusion gene was used to monitor expression. With these two constructs, we made several transgenic mice in which the Tbx5 genes were misexpressed transiently. We did not establish a stable transgenic line, as we speculated that misexpression of the Tbx5 gene in the ventricle itself would induce severe abnormalities of heart morphology, and hence cause premature death of the embryos.

When Tbx5 was misexpressed uniformly with the β-MHCpromoter, several morphological changes were observed(Fig. 4C). In the normal heart,the developing right ventricle is already formed and evident as a bulge in the right side of heart tube at E10 (black arrowhead in Fig. 4B). By contrast, this bulge was not formed in the hearts in which Tbx5 was misexpressed by the β-MHC promoter (red arrowhead in Fig. 4C). In such hearts, the prospective left ventricle seemed to be connected to the conus cordis with a small junction. In addition, this abnormal junction was shifted rostrally compared with the normal heart (red arrowhead in Fig. 4C). This indicates that uniform expression of Tbx5 perturbs the normal development of both the right ventricle and the atrioventricular connection. We confirmed the uniform expression of the transgene by in situ hybridization (data not shown).

Next, we analyzed the morphological changes induced by the MLC-2ν/chick Tbx5-EGFP construct. When we checked transgene expression by GFP fluorescence signals, GFP expression was evident on the right side of the heart tube as expected (yellow arrowheads in Fig. 4D,F), indicating that the chick Tbx5 gene was misexpressed in the prospective right ventricle. Contrary to the electroporation experiments in the chick, the Tbx5-EGFP gene was expressed in a gradient fashion in this transgenic mouse, as confirmed by both GFP signals(Fig. 4D,F) and whole-mount in situ hybridization using chick Tbx5 as a probe(Fig. 4H). Although Tbx5-EGFP was strongly expressed at the right-most end of the prospective right ventricle, this expression became faint at the middle of the ventricle. We did not detect GFP fluorescence at the position of the ventricular septum (Fig. 4D,H). Interestingly, such transgenic hearts exhibit a consistent morphological alteration, namely, swelling of the prospective right ventricle (red arrowheads in Fig. 4E,G). Later, a small protrusion was formed on the surface of the induced swelling in the prospective right ventricle at E14.5 (blue arrowhead in Fig. 4I).

To confirm the nature of these morphological alterations, we checked the expression of several markers by in situ hybridization(Fig. 5). As reported previously, mouse eHAND is expressed in the prospective left(Fig. 5A,J). By contrast, dHAND genes is predominantly expressed in the prospective right,albeit expanding to the bulbus cordis, the part of the prospective left ventricles and the ventricular septum (Fig. 5D,M) (Srivastava et al.,1997; Firulli et al.,1998; Srivastava et al.,1995). In addition, the mouse ANF gene is expressed in the left ventricle (Fig. 5G,R)(Zeller et al., 1987), as is found in the chick. Hence these three genes provide good markers to confirm the identity of the left and right ventricles. When we analyzed transgenic hearts in which Tbx5 was misexpressed uniformly with theβ -MHC promoter, we found that the domains of eHAND and mouse Anf expression were expanded (n=3/3, n=4/4,respectively), although the gross size of the heart decreased(Fig. 5B,H). By contrast,expression of dHAND was repressed, leaving the entire ventricle dHAND-negative (Fig. 5E) (n=5/5).

We carried out the same analysis on hearts in which the Tbx5 gene was misexpressed with the MLC2v promoter in a gradient manner, with robust expression at the right-most end. In such hearts, the eHANDand mouse Anf genes were strongly induced in the swelling formed by the expression of this transgene (red arrowheads in Fig. 5C,I) (n=4/4, n=4/4, respectively), whereas the expression of the dHANDgene was suppressed (Fig. 5F)(n=3/3).

The same results were obtained when analyzed in tissue sections. This analysis also enabled us to examine the induced morphological changes in detail. As observed with whole-mount in situ hybridization, the eHAND, mouse Anf and dHAND genes are expressed in similar fashions (Fig. 5J,M,R). In addition, we found that the ventricular septum is developing at this stage(E10.5). In the deformed hearts dissected from the β-MHC-Tbx5transgenic mice, both eHAND and mouse Anf genes were induced almost to the right end of the developing ventricle (black arrowhead in Fig. 5K,Q), whereas expression of the dHAND gene was found to be repressed completely(Fig. 5N). Furthermore, septum formation was completely suppressed in such hearts(Fig. 5K,N,Q). Instead, a tiny bulge of ventricular wall was formed in a small eHAND/mouse Anf-negative region (black arrowheads in Fig. 5K,N,Q).

Conversely, the ventricular septum formed normally in hearts of the MLC2v-Tbx5 transgenic mice (Fig. 5L,O,R), although the right ventricle appeared to be enlarged. In this enlarged right ventricle, the mouse Anf gene was induced strongly (Fig. 5R). Likewise,expression of the eHAND gene was induced (red arrowhead in Fig. 5L), albeit expression disappeared near the septum (black arrowhead in Fig. 5L). In this eHAND-negative domain near the septum, expression of the dHAND gene was detected (red arrowhead in Fig. 5O), although this gene was completely suppressed in the remaining part of the right ventricle(Fig. 5O).

Taken together, these lines of evidence strongly suggest that the forced expression of the Tbx5 gene in the prospective right ventricle converts expression patterns of several right and left ventricular markers with extensive morphological alterations.

Our embryological data indicate that Tbx5 specifies the left ventricle and the ventricular septum is formed at the boundary between Tbx5-positive and Tbx5-negative regions. Interestingly,another T-box gene, chick Tbx20 is expressed in a complementary fashion, hence expressed in the Tbx5-negative right ventricle. As misexpression of Tbx5 in the right ventricle represses Tbx20expression, these two Tbx genes may be mutually exclusive. To understand molecular interaction between Tbx5 and Tbx20, we carried out a set of transfection assays using human ANF promoter-luciferase reporter construct. As reported previously, Tbx5 and Nkx2.5 synergistically activate this promoter (Bruneau, 2002; Nemer and Nemer, 2001; Bruneau et al., 2001; Hiroi et al., 2001). In addition, another heart-specific transcription factor, GATA4 again synergistically activates cardiac α-actin promoter with Nkx2.5(Durocher et al., 1997; Lee et al., 1998). In addition, recently, it has been reported that Tbx2 abrogates the synergistic activation of the ANF promoter by Tbx5 and Nkx2.5(Habets et al., 2002). These lines of evidence suggest that the tight interactions and the crosstalks among heart-specific transcription factors play essential roles for the chamber formation of heart (Bruneau,2002; Nemer and Nemer,2001).

When a human ANF promoter-luciferase construct was transfected to COS7 cells along with several expression plasmids, Tbx5, Nkx2.5 and GATA4 activated this promoter about 4.7-, 3.7- and 3.5-fold, respectively(Fig. 6). As reported previously, when both Tbx5 and Nkx2.5 were co-expressed, synergistic activation was observed (about 11.6 fold). Interestingly, when both Tbx5 and GATA4 were co-expressed, robust synergistic activation was obtained(25.3-fold), indicating that the synergism between Tbx5 and GATA4 is more potent. By contrast, when full-length zebrafish Tbx20 was misexpressed, these synergistic actions between Tbx5/Nkx2.5 and Tbx5/GATA4 were abrogated (7.0-and 3.2-fold, respectively). As ANF gene is expressed in the left ventricle, our data indicate that the left side expression of this gene is regulated in two ways: (1) activation by Tbx5 in the left ventricle and (2)repression by Tbx20 in the right ventricle. These lines of evidence suggest that the specification of chick left/right ventricles and the position of ventricular septum are controlled by the distinct actions of two Tbx genes expressed in the mutually exclusive fashion. When we used chick Tbx20, exactly same data were obtained (data not shown).

Results obtained from our experiments are summarized in Fig. 7. Normally, the Tbx5-positive/Tbx20-negative part develops to become the left ventricle, and the Tbx5-negative/Tbx20-positive area gives rise to the right ventricle. The ventricular septum is formed at the boundary of these two domains, separating the aortic and pulmonary blood flows completely. These observations indicate that the domains of Tbx5/Tbx20 expression demarcate the left and right ventricles.

As shown in the chick and mouse experiments, when Tbx5 was misexpressed ubiquitously in the heart tube, ventricular septum formation was disturbed, resulting in a single ventricle and several malformations in the atria and the tissues derived from the bulbus cordis/truncus arteriosus (Type 1 in Fig. 2P and Fig. 7). These morphological changes were accompanied by the complete repression of Tbx20. When a small Tbx5-negative region was left in the chick heart, the ventricular septum was shifted to the right, with a small right ventricle and an enlarged left ventricle. In such hearts, VSDs and conal septation/rotation defects were frequently observed (Type 2 in Fig. 2P and Fig. 7). When misexpression occurred in a more restricted area in the chick heart, an ectopic septum was formed at the new boundary in the right ventricle(Fig. 4J,K; Type 3 in Fig. 7). These lines of evidence strongly suggest that Tbx5 specifies the identity of left ventricle through the tight interactions with Tbx20 and other heart-specific factors during chick cardiac development. In both the chick and mouse experiments, several malformations were observed outside of the ventricle. First, septation defects were frequently found in the conus cordis(Type 1 in Fig. 7, Fig. 2F-J). Second, ASDs were evident with abnormal dilation of atria(Fig. 2I,J). These observations indicate that Tbx5 has essential roles during development of multiple heart tissues.

In the chick and mouse hearts in which Tbx5 was misexpressed ubiquitously, the ventricular wall was thinner than normal, and the trabecular formation was coarse and rough. These phenotypic changes indicate that Tbx5 regulates cardiac muscle differentiation. Recently, it was found that the Tbx5 protein associates and interacts physically with the cardiac homeoprotein Nkx2.5, which is essential in cardiac muscle development(Bruneau et al., 2001; Hiroi et al., 2001). In addition to this, we found that the interaction between Tbx5 and GATA4 synergistically activates the ANF promoter. These lines of evidence strongly suggest that Tbx5 possesses multiple interfaces necessary for the multiple protein-protein interactions. Hence, changes of the level of Tbx5 protein in differentiating cardiomyocytes might affect the transcriptional control of cardiac genes by disturbing the balance of multiple interactions. Both the loss and gain of Tbx5 function disturb the transcriptional control of Tbx5 targets, probably through the abnormal balance between the Tbx5, Tbx20, GATA4 and Nkx2.5 proteins.

Contrary to chick Tbx20, it has been reported that mouse Tbx20 is expressed uniformly in all four heart chambers(Kraus et al., 2001). Similar differences in expression patterns of cardiac genes can be found in dHAND and eHAND. Mouse dHAND (Hand2 – Mouse Genome Informatics) and eHAND (Hand1 – Mouse Genome Informatics) genes are expressed differently in the ventricles, whereas chick HAND genes are expressed uniformly in the developing heart. In search of putative right ventricle-specific markers, we cloned chick desmin,dystrophin and SM22α genes, as mouse counterparts of these genes are expressed in the right ventricle(Kuisk et al., 1996; Kimura et al., 1997; Moessler et al., 1996; Li et al., 1996). However,these genes are expressed uniformly in chick heart (data not shown). These lines of evidence suggest that the mechanism of ventricular specification might be different in species. Because ANF gene is the direct target of Tbx5, it might not be adequate to argue that chick ANF is the left ventricle-specific marker. As described above, several genes are expressed in different manners in mouse and chick hearts. Hence, it is important to isolate novel left or right ventricle-specific markers. For this purpose, we are performing cDNA subtraction and RDA (representational difference analysis) to isolate region-specific markers in developing vertebrate hearts.

Recently, functions of zebrafish T-box gene tbx20 (previously known as hrT) gene have been reported(Szeto et al., 2002). Interestingly, loss of tbx20 function resulted in upregulation of Tbx5. Conversely, misexpression of tbx20 induced downregulation of tbx5. These data indicate that Tbx20 regulates tbx5 expression in zebrafish. Our chick data also indicate that Tbx5 represses Tbx20 when misexpressed(Fig. 1K). These lines of evidence suggest that the tight regulatory interaction between Tbx5 and Tbx20 is crucial in zebrafish and chick, but not in mouse right ventricle. Comparative and comprehensive approaches using various molecular markers should be carried out to uncover the mechanism of development and evolution of different vertebrate hearts.

In addition, Tbx20 represses the synergistic action of Tbx5 and GATA4 on human ANF promoter, indicating that Tbx20 represses ANFexpression in the right ventricle. In mouse heart, ANF expression is also restricted to the left ventricle, although mouse Tbx20 is expressed uniformly. This suggests that different mechanism might operate in the mouse left ventricle to sustain the action of Tbx5 or inhibit the function of Tbx20. As the levels of T-box proteins are crucial for normal development,the levels of Tbx5 and Tbx20 proteins might be important for the development of mouse left ventricle. In addition, we do not exclude the possibility that unknown factor(s) might be involved in this process.

Interestingly, the ANF and connexin 40 (cx40) genes were found to be direct targets of the Tbx5/Nkx2.5 protein complex(Bruneau et al., 2001; Hiroi et al., 2001). Consistent with this, misexpression of Tbx5 in the right ventricle induces robust expression of the ANF gene in both mice and chicks, as described above (Figs 1 and 5). Although we did not examine the expression of cx40 in our system, we found that the beating pattern of electroporated chick hearts was abnormal: simultaneous contraction of atria and ventricles instead of normal serial beating (data not shown). This could be related to the abnormal conduction systems found in both the heterozygous Tbx5^del/+ mice and individuals with Holt-Oram syndrome.

As observed in our gain-of-function approaches, Tbx5 misexpression disturbs the normal differentiation of cardiac muscle. By contrast, multiple anomalies found in both the heterozygous Tbx5^del/+ mice and individuals with Holt-Oram syndrome indicate the haploinsufficiency of Tbx5 in cardiac development. These lines of evidence strongly suggest that the level of Tbx5 expression in the developing cardiac muscle cell is crucial. As reported, the Tbx5 protein interacts with Nkx2.5(Bruneau et al., 2001; Hiroi et al., 2001), and another T-box protein Tbr1 makes a complex with CASK, one of the membrane-associated guanylate kinases (MAGUKs), to regulate transcription of target genes (Hsueh et al.,2000). These physical interactions suggest that the balances between Tbx proteins and other interacting partners are important for the orchestrated processes of pattern formation. As genetic analysis of Drosophila suggests the interaction between optomotor blind(omb), one of the Drosophila T-box genes, and DPP/WG signaling, the physical interactions between Tbx proteins and other factors including signal transduction factors could be a general characteristic(Srivastava and Olson, 2000; Conlon et al., 2001). Solving these putative interactions would be an important key to understanding the roles played by Tbx genes during development.

As an opposite approach, we electroporated a dominant negative form of Tbx5 (EnR-Tbx5: fusion of Engrailed suppressor domain and chick Tbx5), expecting a leftward shift of the ventricular septum. Nonetheless, misexpression of EnR-Tbx5 led premature death of embryos from E1.5 to E3 (n=76/76), probably because this type of misexpression itself disturbs cardiac development at very early stages, such as heart tube looping and differentiation.

As shown in Fig. 2F-O,ventricular myocardium was thin when Tbx5 was misexpressed. This would suggest that misexpression of Tbx5 might induce apoptosis or inhibition of cell growth. Nonetheless, when we performed the TUNEL assay to detect apoptotic cell death in differentiating cells, we could not detect any difference between the normal and the Tbx5-misexpressed hearts (data not shown). Even between E4 and E5, at which the ventricular septum starts to form, apoptotic cell death was not evident, indicating that the loss of ventricular septum is not due to Tbx5-induced apoptosis. Rather, we speculate that retardation of myocardium growth is an indirect effect of abnormal heart development, as septum-less single ventricle hearts exhibited an abnormal beating pattern(data not shown). In addition, both ASD and VSD could cause severe circulation defects, resulting in loss of hemodynamic stimulation to the developing myocardium. We also do not exclude the possibility that overexpressed Tbx5 Inhibits proliferation of myocardiac cells.

Contrary to the phenotypes observed in chick hearts, misexpression of Tbx5 in mouse hearts did not affect growth of myocardium. As shown in Fig. 5L,O,R, misexpression of Tbx5 driven by the MLC2v promoter induced swelling of the right ventricle, leaving the size of the left ventricle normal. In addition,thickness of the ventricle was normal, even when Tbx5 was misexpressed in the entire ventricle by the β-MHC promoter. These lines of evidence suggest that the abnormal phenotypes observed in both chick and mouse were not due to the proliferation/apoptosis defects. Nonetheless, we do not exclude that possibility that over or high doses of Tbx5 misexpression might affect the growth and/or differentiation of cardiomyocytes, as the balances with other transcription factors, such as Tbx20, GATA4 and Nkx2.5 are pivotal for the normal patterning of vertebrates hearts.

As mentioned previously, vertebrates exhibit different heart morphologies:fish possess one ventricle/one atrium; amphibians have one ventricle/two atria; reptiles have two incomplete ventricles/two atria; and birds/mammals have two complete ventricles/two atria. Although we have not yet expanded our analysis to other vertebrates, the Tbx5 and Tbx20 genes could be good markers with which to explore the evolution of heart morphology of various vertebrate animals. To explore further, we checked expression patterns of zebrafish tbx20 and tbx5 in developing hearts(right panel in Fig. 7). Interestingly, zTbx20 is expressed in the Bulbus arteriosus (BA) and the ventricle near the BA, whereas zTbx5 in the atrioventricular junction (AVJ) and the ventricle near the AVJ. Expression of these Tbx genes in the ventricle is complementary at both 48 and 120 hpf, yet showing gradient expression without making a clear boundary observed in chick heart. These observation are compatible with the mutually repressive actions of these Tbx genes (Szeto et al., 2002) and our observations that the ventricular septum is formed at the distinct expression boundary of Tbx5.

Our data provide important insights on cardiac development, onset of human congenital heart diseases, and evolution of vertebrate hearts. Although we are far from a complete understanding, precise molecular analysis of the Tbx5 gene would provide valuable information for the comprehensive understanding of vertebrate pattern formation.

We thank Dr Katherine E. Yutzey for providing the β-MHCpromoter construct, Drs Malcolm P. Logan and Cliff Tabin for the chick and mouse HAND genes and Dr K. Inoue for zebrafish cDNA. We also thank Drs Naoyuki Miura, Antonio Baldini and Benoit G. Bruneau for critical comments on the manuscript, and Drs Malcolm P. Logan and Cliff Tabin for sharing unpublished data before publication. This work was supported by a Grant-in-Aid for Scientific Research on Priority Areas (C) from the Ministry of Education,Science Sports and Culture of Japan (T.O.); Special Coordination Funds for Promoting Science and Technology from the Science and Technology Agency(T.O.); Creative Basic Research from the Ministry of Education, Science Sports and Culture of Japan (T.O.); the Uehara Memorial Foundation (T.O.); the Inamori Foundation and the Toray Science Foundation (T.O.). This work was also funded in part by the Dan Charitable Trust Fund for Research in the Biological Sciences (J.K.T.) and The Mochida Memorial Foundation for Medical and Pharmaceutical Research (K.K.-T.). J.K.T. is supported by JSPS Research Fellowships for Young Scientists.