Tbx1 affects asymmetric cardiac morphogenesis by regulating Pitx2 in the secondary heart field

TBX1, a member of the T-box family of transcription factors, has been identified as a strong candidate gene for the etiology of 22q11.2 deletion syndrome (22q11DS; also known as DiGeorge/velo-cardio-facial syndrome; MIM 188400/192430). This syndrome occurs with a frequency of 1:4000 live births (Burn and Goodship,1996). Individuals with 22q11DS display mild craniofacial anomalies, immune disorders and cardiovascular defects(DiGeorge, 1965; Shprintzen et al., 1978). The heart anomalies include aortic arch malformations and outflow tract defects(interrupted aortic arch type B, abnormal origin of the right subclavian artery, right sided aortic arch, tetralogy of Fallot), and ventricular and atrial septal defects (VSD, ASD) (Yamagashi, 2002). The Tbx1 gene has been inactivated in the mouse (Jerome and Papaioannou, 2001; Lindsay et al., 2001; Merscher et al.,2001). Although Tbx1 heterozygous null mutant mice survived at normal Mendelian ratios and had only mild cardiovascular defects,homozygous mice died at birth with cleft palate, thymus gland aplasia, a single outflow tract and VSD (Jerome and Papaioannou, 2001; Lindsay et al., 2001; Liao et al.,2004). The Tbx1^-/- mice have, thus, been shown to be a good model for this syndrome. However, the mechanisms of action of Tbx1 are still largely unexplored.

Expression analysis of Tbx1 at stages E8.0-E11.5 in mouse embryogenesis, showed strong expression in the pharyngeal apparatus. It is expressed in the core mesenchyme of the pharyngeal arches, in the pharyngeal endoderm and in the splanchnic and pharyngeal mesoderm(Chapman et al., 1996) in areas of the secondary heart field (SHF) (Xu et al., 2004). The SHF is a recently discovered cell population of cardiac progenitor cells in the pharyngeal/splanchnic mesoderm, which have been shown to be necessary for the proper development of the heart(Kelly, 2005).

The heart develops from a bilaterally symmetric cardiac crescent, the primary heart field located in the lateral mesoderm. Cells of the cardiac crescent form a linear heart tube at E8.0, with the inflow tract located caudally and the outflow tract cranially. Recent studies in the mouse and chick have shown that, additionally, cells from the pharyngeal mesoderm contribute to the distal outflow tract for the further growth of the heart(Mjaatvedt et al., 2001; Waldo et al., 2001; Kelly et al., 2001), thus defining a new secondary cardiogenic region. Further studies have revealed that cells of the SHF not only contribute to the arterial pole, but also to the venous pole of the heart (Cai et al.,2003, Meilhac et al.,2004, Kelly,2005). Perturbation of these cardiac precursor cells resulted in severe cardiac defects (Hu et al.,2004; Xu et al.,2004; Kelly, 2005; Ward et al., 2005), thereby proving their importance for proper heart development.

Recently, it has been implicated that Tbx1 plays a major role in the SHF. Tbx1-null mutants display outflow tract hypoplasia,suggesting that Tbx1 might regulate cell proliferation necessary for growth of the developing outflow tract. It has been hypothesized that this regulation also involves the fibroblast growth factors Fgf10 and Fgf8 (Vitelli et al.,2002; Hu et al.,2004; Xu et al.,2004). All three genes are co-expressed in the pharyngeal mesoderm and endoderm. Moreover, in the Tbx1 null mutant, Fgf8 and Fgf10 are downregulated in the pharyngeal endoderm and the pharyngeal mesoderm, respectively (Vitelli et al.,2002; Xu et al.,2004) (S.N. and V. Aggarwal, unpublished).

To further investigate the role of Tbx1 during heart development,we performed in situ hybridization to identify potential downstream genes. One gene of particular interest to us was Pitx2, a bicoid-like homeobox gene. It is expressed, like Tbx1, in the pharyngeal mesoderm and Pitx2-null mutants have heart defects reminiscent of individuals with 22q11DS. Hemizygous loss-of-function mutations in human PITX2are responsible for the etiology of Rieger syndrome (RGS1; MIM 180500), which is associated with craniofacial, ocular defects, umbilical abnormalities(Semina et al., 1996) and,occasionally, cardiac defects (Mammi et al., 1998).

Inactivation of Pitx2 in the mouse results in mid-gestational death because of multiple defects, including failure in body wall closure, eye defects, craniofacial abnormalities, right lung isomerism and severe cardiac defects (Gage et al., 1999; Lin et al., 1999; Lu et al., 1999; Kitamura et al., 1999). Moreover, studies in the mouse, chick, frog and zebrafish revealed that Pitx2 mediates signaling for left-right body asymmetry. It is the only gene known so far that is left-asymmetrically expressed during organ morphogenesis of the heart, gut and lung(Meno et al., 1998; Piedra et al., 1998; Ryan et al., 1998; Yoshioka et al., 1998; Campione et al., 1999; Lu et al., 1999). In the mouse, there are three different Pitx2 isoforms: Pitx2a,Pitx2b and Pitx2c (Kitamura et al., 1999). Only Pitx2c is left-asymmetrically expressed and has been shown to be indispensable for proper asymmetric cardiac morphogenesis (Kitamura et al.,1999; Schweickert et al.,2000). Null mutants of Pitx2c died perinatally with severe heart defects including VSD, ASD and aortic arch patterning defects(Liu et al., 2002). In this study, we present a detailed analysis on Tbx1 and Pitx2co-expression in wild-type embryos. We found that Pitx2 and Tbx1 are co-expressed in the left SHF cells of the pharyngeal mesoderm from E8.0 onwards, as well as in some areas of the venous and arterial pole of the heart. Surprisingly, we also found that Tbx1 is asymmetrically expressed. This expression is transient, at E9.0, and is in the left caudal domain of Tbx1 expression in the pharyngeal mesoderm within the Pitx2-positive region. Asymmetric Tbx1 expression was not downregulated in the Pitx2-null mice. By contrast, we detected downregulation of Pitx2 expression in the cardiac region of Tbx1-null mutant, pointing to a possible genetic interaction of these two genes. Indeed, mice double heterozygous for Tbx1 and Pitx2 died perinatally of severe cardiovascular defects occurring with varying penetrance, thus proving our hypothesis.

Sequence analysis of the Pitx2 enhancer (Pitx2-ASE) for left-sided expression revealed a putative T-half site close to an Nkx2.5-binding site. Subsequent molecular studies revealed synergistic activation of a Pitx2 reporter construct by Tbx1 and Nkx2.5. In conclusion, in this study we present for the first time in vitro and in vivo evidence for a novel Tbx1-Pitx2 pathway in the left SHF.

Tbx1^+/- mice(Merscher et al., 2001) were intercrossed with Pitx2^+/- mice(Gage et al., 1999). Genotyping was performed with the following PCR primer pairs: Tbx1, as previously described (Liao et al., 2004);and Pitx2 wild-type and mutant alleles (Pitx2-F, GTG TCT GTA AAA CAC GCG CAT G; Pitx2-R, GTC TCC AGT GAA GCC AAG CCT).

Mouse embryos were treated for histological analysis as described previously (Liao et al.,2004). E18.5 and newborns were fixed in 10% neutral buffered formalin solution for 3 days.

Complementary RNA probes to mouse full-length Tbx1, full-length Pitx2 (Campione et al.,1999), Nkx2.5 (Lyons et al., 1995), Tbx3(Hoogaars et al., 2004), Tbx2 (Habets et al.,2002), Isl1 (Cai et al., 2003), Nppa(Zeller et al., 1987) and myosin heavy chain (Mhc) ATP-binding site, which recognizes all Mhc isoforms (Franco et al.,2002) were generated using standard protocols. Embryos and E10.5 hearts were isolated in PBS, fixed in 4% PFA and gradually dehydrated into 75%methanol/PBS, 0.1% Tween-20. Hybridization protocols for whole-mount in situ hybridization followed standard procedures with slight modifications. In situ hybridization on sections was performed on serial 12 μm paraffin wax embedded sections from E8-E10.5 wild-type embryos. Hybridization was performed as previously described (Franco et al.,2001).

Whole hearts from E10.5 embryos were dissected and stored in RNAlater RNA stabilization reagent (QIAGEN) at 4°C. Total RNA was extracted using RNeasy Protect Mini Kit (QIAGEN) and cDNA was synthesized using SuperScript First-Strand Synthesis System (Invitrogen). The PCR reactions were performed with a LightCycler (Roche) and FastStart DNA Master SYBR Green I (Roche). All values were normalized to the level of Gapdh, which was used as an internal control in each sample. Primers specific for Pitx2 were 5′-GTCTCTTCTCCAAAGACTCC-3′ and 5′-CGGCGATTCTTGAACCAAAC-3′. Primers specific for Gapdhwere 5′-TTCACCACCATGGAGAAGGC-3′ and 5′-GGCATGGACTGTGGTCATGA-3′. Six wild-type and 11 Tbx1^-/- embryos were tested and statistical significance was calculated by both unpaired Student's t-test and one-sided Wilcoxon rank-sum test.

Sequences of oligonucleotides containing the T-half site of the Pitx2-ASE (wild type), mutated (M1, M2) or the consensus (Cons)T-half site are as follows: wild type, CAATCAGGTGTAAAGAGGAA; M1,CAATCAGATTTGAAGAGGAA; M2, CAATCAAATTTGAAGAGGAA; Cons, CAATCAGGTGTGAAAAGGAA. A total of 100 pmol of each 5′- and 3′-oligonucleotide were added to 5 μl annealing buffer (100 mM Tris pH 8, 500 mM NaCl, 10 mM EDTA pH 8) and 43 μl H₂O. The reaction was boiled for 15 minutes and slowly annealed by cooling down to room temperature. A total of 20 pmol of annealed oligonucleotides were radiolabeled with [γ-³²P]dATP using T4-Kinase (NEB) and purified using Sephadex G-50 spin columns (Amersham). Full-length cDNA of Tbx1 was cloned into pcDNA3.1 (Invitrogen) and transfected into 293T cells. Cell lysate was used for the DNA-binding assay and incubated for 30 minutes on ice in a reaction containing 5% glycerol, 10 mM HEPES pH 7.5, 25 mM KCl, 1 mM DTT, 1 mM EDTA and 5 mM MgCl₂(DNA-Binding buffer), 1 μg dIdC and 0.1 pmol radiolabeled probe (wild-type,M1, M2 or Cons) was added. When using unlabeled wild-type competitor oligonucleotides in a 100-fold molar excess and Tbx1 antibody for the supershift, samples were preincubated for 20 minutes on ice before adding radiolabeled probe. DNA-protein complexes were resolved on a 6% non-denaturing polyacrylamide gel in 0.5×TBE. Gels were exposed to Kodak film (Biomax MS).

The 900 bp enhancer element of the Pitx2c isoform(Pitx2-ASE) (Shiratori et al.,2001) was cloned into pGl3 SV40 luciferase vector (Promega). Plasmid transfections were performed in six-well plates using Polyfect Reagent(QIAGEN) according to the manufacturer's protocol. Cos7 cells were transfected with 0.15 μg of the Pitx2-ASE luciferase construct and with 0.3μg of the expression constructs Tbx1 in pcDNA3.1,Flag-Nkx2.5 in pCI, respectively. The empty vectors pcDNA3.1 and pCI were used as controls. A total of 0.03 μg of CMV-βGal construct was co-transfected to normalize transfection efficiency. Luciferase andβ-galactosidase activity were assayed 48 hours after transfection using Luciferase assay system (Promega) in a TD -20/20 luminometer andβ-galactosidase reporter gene activity detection Kit (Sigma)respectively. The mutagenized reporter (Pitx2-Mut) was generated using the Quick change site-directed mutagenesis kit (Stratagene). The T-half site in the Pitx2-ASE was mutated to 5′-AAATTTGAAG-3′. Transfections with Pitx2-Mut were carried out as mentioned above.

293T cells were transfected with a Tbx1-GFP construct in pEGFP-N1(Clontech) and a Flag-Nkx2.5 construct in pCI (Promega) using Polyfect Reagent (QIAGEN) and lysed after 48 hours in RIPA-buffer (50 mM Tris pH 7.5, 200 mM NaCl, 1% Triton X-100, 0.25% DOC, 1 mM EDTA). The Co-Ip was performed with Anti-Flag M2 affinity gel (Sigma) according to the manufacturer's protocol. Immunoblotting was performed using Anti-Flag M2(Sigma) and Anti-Tbx1 antibodies (Zymed). Blots were exposed to Kodak film(Biomax MS).

Tbx1 and Pitx2 are co-expressed during mouse embryogenesis in the arterial and venous pole of the heart as early as E8 until E9.5 (Fig. 1). They are co-expressed in the left pharyngeal mesoderm, the left horn of the sinus venosus and the outflow tract up to the inner curvature of the common ventricle (Fig. 1A-L). These regions are additionally characterized by the expression of the transcription factor Nkx2.5, an early marker of cardiomyocytes(Lyons et al., 1995), and Isl1, a marker for cells of the SHF(Cai et al., 2003). Pitx2 is left-asymmetrically expressed in this region. Closer investigation of Tbx1 expression at stage E9.0 revealed that expression in the left pharyngeal mesoderm extends more caudally compared with its expression on the right side (Fig. 1M-O,). The left asymmetric expression of Tbx1prompted us to investigate whether this aspect of their regulation could be shared. To achieve this, we analyzed Tbx1 expression in the Pitx2 null mice, where this asymmetrical Tbx1 expression was retained (Fig. 1P). Thus, the Tbx1 transient left asymmetry must be regulated through alternative pathways.

The co-expression of Pitx2 and Tbx1 led us to investigate a potential epistatic relationship between the two genes, by performing expression studies in Tbx1^-/- mutants. In situ hybridization studies in the Tbx1^-/- mice indeed revealed reduced expression of Pitx2. At E8, Pitx2 expression was reduced in the region of the left SHF, the outflow tract and the left splanchnic mesoderm (Fig. 2A-D). At this stage, the most caudal domain of Pitx2 head mesenchyme expression was also reduced, possibly owing to fewer cells in this region in the Tbx1^-/- mice. However, Pitx2expression was unaltered in the rostral part of the head mesenchyme(Fig. 2E,F). Expression of the cardiogenic genes Nkx2.5 and Isl1 was not downregulated in these embryos (Fig. 2G,H and not shown), thus indicating that their cardiac field is correctly specified. To determine whether Pitx2 expression was also altered later in heart development, we examined stages E10.0-E10.5. We found variable downregulation of Pitx2 in the Tbx1-null mutants, ranging from total absence of expression to general downregulation in the left atrium, ventral region of the ventricles, inner curvature and outflow tract(Fig. 2I-N). Pitx2expression was unaltered in other regions of the embryo(Fig. 2O,P). The degree of downregulation in E10.5 hearts was assessed by quantitative RT-PCR analysis. There was a statistically significant difference (P<0.05) in Pitx2 mRNA levels between wild-type and Tbx1^-/-embryos (Fig. 2Q). Thus, we could conclude that the left cardiac morphogenetic gene, Pitx2(Yoshioka et al., 1998; Piedra et al., 1998; Ryan et al., 1998; Campione et al., 1999) may act downstream of Tbx1 in the genetic pathway of heart development.

To validate our hypothesis that Pitx2 acts genetically downstream of Tbx1, we intercrossed heterozygous animals and generated Pitx2^+/-; Tbx1^+/- mice to ascertain whether double heterozygous mice display a more severe phenotype than do single heterozygous animals. As stated above, Tbx1^+/- mice have only mild cardiovascular defects (Jerome and Papaioannou, 2001; Lindsay et al., 2001; Merscher et al.,2001; Liao et al.,2004). None has been found in Pitx2^+/- mice(Gage et al., 1999; Kitamura et al., 1999), nor have we seen any in our study. Analysis of the genotype distribution of 104 mice at P10 resulted in 47 wild-type, 28 Pitx2^+/-, 27 Tbx1^+/-, but only two Pitx2^+/-;Tbx1^+/- mice. This suggested that the mice might die during embryogenesis or the neonatal period. A subsequent closer examination of the newborn mice revealed that the Pitx2^+/-;Tbx1^+/- mice survived embryogenesis; however, they become cyanotic immediately after birth and then died. Histological analyses of E15.5, E18.5 embryos and newborns (n=20) revealed severe cardiac defects (Fig. 3A-H), though with incomplete penetrance (n=12; 60%). In the affected animals,expressivity of the phenotype was variable(Table 1): in some of the double-heterozygous mice, the aorta and the pulmonary artery arose from the right ventricle (double outlet right ventricle, DORV)(Fig. 3A), which was not observed in either the Tbx1 or the Pitx2 single heterozygous mice in the genetic background analyzed(Fig. 3B,C). VSD, ASD(Fig. 3D) and atrio-ventricular valve defects (Fig. 3E) were additionally present in the Pitx2^+/-; Tbx1^+/-mice, whereas in single heterozygous mice, the atrial and ventricular structures were normal (Fig. 3G,H). Stenosis of the pulmonary trunk(Fig. 3A) and abnormal drainage of the pulmonary vein into a common (Fig. 3F) instead of the left atrium(Fig. 3H) could also be found. Moreover, mispositioning of the aorta and malformation of the coronary vessels did occasionally occur (data not shown). The severe phenotype found in the Tbx1^+/- mice provides evidence for genetic interaction between Tbx1 and Pitx2. Many of the defects were commonly observed in Pitx2c homozygous null mutant animals(Liu et al., 2002). Phenotypic variability in the double heterozygous animals may result from either genetic modifiers or stochastic factors.

To determine the developmental onset of these defects, we analyzed Pitx2^+/-; Tbx1^+/- embryos at E10.5. Shape and size differences of the hearts of the double versus the single heterozygotes were already visible at this stage. An enlarged atrio-ventricular canal (AVC),reduced ventricular expansion and abnormal ventricular shape were observed(Fig. 3I-L). The severity of the malformation was also variable between the different double heterozygous mice, consistent with the later variable phenotype. Though malformed, these hearts were properly patterned in the AVC and around the inner curvature, as indicated by the in situ hybridization staining with the molecular markers Tbx3 (Fig. 4A,B) and Tbx2 (Fig. 4C,D)(Habets et al., 2002; Hoogaars et al., 2004). However, reduced expression of the chamber marker Nppa(Fig. 4E,F) could be detected in the Pitx2^+/-; Tbx1^+/- mutants, which might be attributed to the reduced degree of ventricular expansion, as additionally visualized by marker staining for Mhc, in the double heterozygous versus wild-type mice (Fig. 4G,H).

Pitx2c is the isoform that is left-asymmetrically expressed. Its expression is regulated by a 900 bp enhancer (Pitx2-ASE) that is located between exons 4 and 5 (Shiratori et al., 2001). T-box-containing genes frequently require co-factors for transcriptional regulation. There is an Nkx2.5-binding site in the Pitx2-ASE, required to maintain asymmetric Pitx2cexpression (Shiratori et al.,2001). We detected a putative T-half site(5′-AGGTGTAAAG-3′) in the enhancer, 26 bp downstream of the Nkx2.5-binding site. Thus, we hypothesized that Pitx2c might be a direct downstream target of Tbx1, and we tested this hypothesis by performing in vitro assays in cell culture. We co-transfected Cos7 cells with a Pitx2-ASE luciferase reporter and either Tbx1 or Nkx2.5 expression vectors. In both cases, this resulted in only weak activation of the Pitx2c enhancer(Fig. 5A). It has previously been shown that Nkx2.5 acts in combination with other T-box proteins, such as Tbx2 during development of the chamber myocardium. Tbx2 and Nkx2.5 form a complex on the atrial natriuretic factor (Anf) promoter to repress its activity (Habets et al.,2002). Therefore, we co-transfected both expression constructs with the Pitx2-ASE reporter, which resulted in a strong activation(12-fold) of the Pitx2-ASE (Fig. 5A). Indeed, transactivation did not occur when the T-half site in the Pitx2-ASE plasmid (Pitx2-Mut) was mutated(Fig. 5B). These results show that Tbx1 and Nkx2.5 can synergistically activate the enhancer and that this requires functional integrity of the Tbx1-binding site.

Specific binding of Tbx1 to the T-half site of the Pitx2-ASE was assayed with electromobility shift experiments(Fig. 5C). Whole-cell extracts of 293T cells transfected with Tbx1 expression vector and oligonucleotides containing either the T-half site of the Pitx2-ASE(Wt-oligo), mutated T-half sites (M1, M2) or a consensus T-half site were used in the DNA-binding assay. Lysates transfected with the Tbx1expression vector resulted in the formation of a Tbx1-DNA complex in the presence of wild-type oligo. To prove the specificity of the binding, 100-fold excess of non-labeled wild-type oligo was added, resulting in loss of the signal. Addition of a Tbx1 antibody, which recognizes a peptide in the C-terminal third of the Tbx1 protein, outside the conserved T-box, resulted in a supershift of the Tbx1-DNA complex, whereas addition of pre-immune serum did not. Part of the DNA-protein complex has not been supershifted, possibly owing to the binding of other T genes present in the cell extract to the same sequence. For example, it has recently been reported that Tbx20 can also activate the Pitx2c enhancer in vitro; it is likely that it binds to the same T-half site (Takeuchi et al.,2005). A DNA-protein complex that could not be supershifted with the Tbx1 antibody was also visible in control experiments in non-Tbx1transfected 293T cells, suggesting that endogenous T-box proteins may be present in these cells (data not shown). Lysates transfected with Tbx1 expression vector resulted in a weak Tbx1-DNA complex in the presence of an oligonucleotide containing the consensus T-half site. Addition of 100-fold excess of the consensus oligonucleotide resulted in loss of the signal (Fig. 5C). Mutated oligos M1 and M2 did not lead to a T-box-DNA complex.

To ascertain whether Tbx1 and Nkx2.5 interact physically with each other,we performed co-immunoprecipitation experiments using cell lysates of 293T cells transfected with Tbx1-GFP and Flag-Nkx2.5. Indeed,immunoprecipitation with antibody directed against the Flag epitope followed by immunoblotting for Tbx1 and Flag revealed interaction between Tbx1 and Nkx2.5 (Fig. 5D).

In this paper, we present for the first time evidence of genetic interaction between the 22q11DS candidate gene, Tbx1 and the left asymmetrical morphogenetic gene Pitx2. We show herein that Pitx2c is a target of Tbx1 in the left SHF, and that Tbx1 can synergistically activate the Pitx2-ASE by interaction with Nkx2.5. This activation is achieved by physical interaction between the two proteins. The relevance of these finding will be discussed below, also in relation to cardiac disease in 22q11DS.

The establishment of the left right axis in the vertebrate embryo involves four steps: the determination of the left-right polarity in or near the node,signals that transfer the left-right identity from the node to the lateral plate mesoderm (LPM), expression of signaling molecules Nodal and Lefty1 in the left LPM, and activation of Pitx2, which regulates asymmetric organ morphogenesis(Hamada, 2001). The expression of the last is regulated via an enhancer in the Pitx2 gene that is located in the intron between exons 4 and 5. This enhancer(Pitx2-ASE), which is responsible for the left-sided expression,contains three FAST-binding sites and an Nkx2.5-binding site, which are necessary for asymmetric regulation of Pitx2. First, Pitx2expression is initiated through Nodal signaling from the node through FAST transcription factors. After Nodal expression has ceased, Pitx2 expression is maintained by the transcription factor Nkx2.5(Shiratori et al., 2001).

In this study, we show synergistic activation of asymmetric Pitx2expression through Nkx2.5 and Tbx1. In addition, we found a transient asymmetric Tbx1 expression in the left pharyngeal mesoderm. We also uncovered regions of co-expression of Tbx1, Pitx2 and Nkx2.5in the SHF from E8.0 onwards. Based upon these results, we propose that Tbx1 and Nkx2.5 are both required for maintenance of Pitx2 expression in the SHF, which in turn is necessary for proper asymmetric cardiac remodeling. The importance of Tbx1 in Nkx2.5-expressing regions has recently also been demonstrated by Xu et al.(Xu et al., 2004). They generated a Tbx1 floxed allele (flanked by loxP sites) and inactivated Tbx1 through Nkx2.5-Cre. These mice had severe outflow tract defects reminiscent of those in individuals with 22q11DS. Their studies showed that Tbx1 is required in Nkx2.5-expressing cells in the SHF. Intriguingly, a genetic interaction between the two genes has not been reported (Xu et al.,2004). However, it cannot be ruled out that analysis of a greater number of animals or in a different genetic background might indeed reveal a genetic interaction.

It is tempting to speculate that Tbx1 is not the only protein required with Nkx2.5 for maintenance of Pitx2 expression. It is likely that other T-box proteins can bind to the T-half site in the Pitx2-ASE and regulate Pitx2 expression in other regions of the heart. It has recently been published that Tbx20 can also activate the Pitx2enhancer in vitro. A knockdown of Tbx20 in mice has revealed its important role for the development of the outflow tract and the right ventricle (Takeuchi et al.,2005), both Pitx2-expressing structures. It is possible that Tbx20 regulates Pitx2 expression in these structures.

The heart originates from three different cell populations, the cardiac crescent or primary heart field, located in the anterior lateral mesoderm, the cardiac neural crest and the anterior or secondary heart field (AHF/SHF). The AHF/SHF has recently been discovered in studies in chick and mouse and is located in the pharyngeal mesoderm. At first, cells of this region have been shown to contribute to the distal part of the outflow tract(Waldo et al., 2001; Mjaatvedt et al., 2001; Kelly et al., 2001). However,subsequent studies have revealed that these cells also contribute to the venous pole of the heart (Meilhac et al.,2004). A model to explain the origin of different cell populations that give rise to the heart has recently been published(Meilhac et al., 2004)(reviewed by Kelly, 2005). It proposes that the embryonic heart derives from two cell lineages. The first lineage contributes to the left ventricle, atria, inflow region and partly to the right ventricle. The second lineage, which is characterized by the expression of the LIM-homeodomain protein Isl1(Cai et al., 2003), is thought to contribute mainly to the right ventricle and the outflow tract, but also to the atria and the inflow region.

The AHF/SHF appears to be a subpopulation of this second lineage(Kelly, 2005). In this study,we show that Pitx2 and Tbx1 are co-expressed in Isl1-positive regions in the SHF and that double heterozygous mice for Pitx2 and Tbx1 show defects of the arterial as well as the venous pole of the heart. These findings suggest that the Tbx1-Pitx2 pathway must be a crucial part of the second lineage.

Tbx1 expression has already been previously shown in the AHF/SHF in the pharyngeal mesoderm as wells as in the pharyngeal endoderm where it is co-expressed with two fibroblast growth factor genes, Fgf8 and Fgf10 (Vitelli et al.,2002; Xu et al.,2004; Hu et al.,2004) (reviewed by Kelly,2005). Both genes are down regulated in the Tbx1-null mutant, suggesting that they are downstream of Tbx1 on both sides of the SHF (Vitelli et al., 2002; Xu et al., 2004) (S.N. and V. Aggarwal, unpublished). In this study, we provide evidence for a novel and complementary Tbx1-Pitx2 pathway in the left SHF(Fig. 6). Tbx1 activates Pitx2 directly, and by interaction with Nkx2.5 it is responsible for maintenance of Pitx2 expression in the left SHF. Target genes of Pitx2 in the secondary heart field have not yet been identified. However, in vivo and in vitro studies support a role for Pitx2 in cell proliferation and migration (Kioussi et al., 2002; Wei and Adelstein,2002), as well as in regulating growth factor activity(Liu et al., 2003).

Several studies have already investigated molecular and genetic relationships between the Tbx and Pitx gene families. In the process of limb patterning, Pitx1 acts by imparting a positional cue in response of which the activation of Tbx4 occurs that directs hindlimb outgrowth(Minguillon et al., 2005). In the pituitary gland, the T box factor, Tpit, cooperates with Pitx1/2 in the activation of the POMC promoter (Lamolet et al., 2001). Whereas in our experiments we have not identified a feedback action of Pitx2 on the Tbx1 expression, we cannot rule out an additional possible cooperative interaction between the two genes. Indeed,preliminary Tbx1 and Pitx2 Co-IP experiments (S.N., unpublished) suggest this possibility. It still has to be determined, however, if the physical interaction between Tbx1 and Pitx2 is relevant within the SHF because candidate downstream target genes are not known.

The 22q11DS is characterized by a multitude of cardiovascular defects occurring in 60% of individuals with 22q11DS. The different malformations occur with variable penetrance. This suggests that genetic modifiers in the form of single nucleotide polymorphisms (SNPs) and/or stochastic factors alter the severity of syndrome and modify the penetrance of heart defects.

Studies in Df1/+ mice possessing a 1 Mb deletion in the region of synteny to the 22q11.2 critical region showed that different genetic backgrounds play a major role in altering the penetrance of cardiovascular defects relevant to the disorder (Taddei et al.,2001). This suggests that genetic variation, or SNPs, can influence the severity of heart defects in 22q11DS. One such candidate has been identified, the vascular endothelial growth factor (VEGF) gene(Stalmans et al., 2003). Inactivation of one isoform of Vegf¹⁶⁴ in the mouse produces similar heart defects as in individuals with 22q11DS. SNP analysis in these individuals revealed a SNP variant in the 5′ untranslated region of VEGF, which could be found more frequently in individuals with cardiac defects compared with those with no detectable heart malformations(Stalmans et al., 2003). It is possible that functional SNPs that change an amino acid or that are present in conserved non-coding regulatory regions of PITX2, NKX2.5, FGF8 or FGF10, might alter the expressivity of the syndrome. If this is the case, they might be considered genetic modifiers of the disorder.

We thank Dr Radma Mahmood and Dawn Lee for technical assistance, Dr Richard P. Harvey, Dr Silvia Evans, Dr Virginia E. Papaioannou, Dr Vincent M. Christoffels and Dr Luciano D'Adamio for probes and constructs, and Ji Yon Bang for help with statistical analysis of the Real time-PCR. This work was supported by grants from the March of Dimes (FY2005-443), NIDCD (R01 DC05186-03) (B.E.M.), from NIH (EY014126) (P.J.G.) and (P01 HL075215)(J.A.E.), and from EU (IP Heart Repair LSHM-CT-2005-018630) (M.C.). Part of this work has been submitted by S.N. as a doctoral thesis (entitled: Genetic interaction of Tbx1 and Pitx2 is required for early heart development) to the University of Karlsruhe, Germany.