Pitx2c patterns anterior myocardium and aortic arch vessels and is required for local cell movement into atrioventricular cushions

Pitx2 is a paired-related homeobox gene that has been shown to play a central role in the late aspects of left-right asymmetric morphogenesis(Capdevila et al., 2000;Harvey, 1998). Identified as the gene mutated in Rieger syndrome 1, Pitx2 also functions in eye,tooth and abdominal wall development(Alward, 2000;Semina et al., 1996). Importantly, Pitx2 has been shown to be a direct target of the left-right signaling pathways that originate early in development through the function of nodal (Shiratori et al.,2001). Loss-of-function experiments performed in mice have revealed a role for Pitx2 in left-right asymmetry of many organs but its role in heart and vascular development is less clear(Gage et al., 1999;Kitamura et al., 1999;Lin et al., 1999;Lu et al., 1999b).

Although Pitx2-null mice have severe cardiac phenotypes that are similar to those observed in humans with laterality defects, thePitx2-null phenotype suggests that Pitx2 function is important after looping morphogenesis, as Pitx2 mutant hearts loop correctly to the right (Gage et al.,1999; Kitamura et al.,1999; Lin et al.,1999; Liu et al.,2001; Lu et al.,1999b). Analysis of individuals with laterality defects has revealed a spectrum of associated cardiac septation and valve anomalies,including abnormalities in conotruncal and right ventricular development,atrial lateralization and atrioventricular (AV) septation(Brown and Anderson, 1999;Icardo and Sanchez de Vega,1991). These observations suggest that the genetic pathways regulating left-right asymmetry may also directly regulate valve and cushion morphogenesis and that subtle defects in left-right asymmetry may be a common etiologic factor for congenital heart disease.

In common with human patients, Pitx2-null mice display atrial septal defects (ASD), abnormal AV septation (resulting in complete AV canal)and abnormal arterioventricular connections(Kitamura et al., 1999;Liu et al., 2001).Pitx2 mutants also have a hypoplastic right ventricle. AlthoughPitx2 mutant mice have severe cardiac anomalies, the primary function of Pitx2 in heart development remains unclear as the Pitx2mutant heart phenotypes could be secondary to delayed looping morphogenesis or embryonic rotation.

In this work, we have used a combination of gene expression analysis and gene targeting approaches to investigate Pitx2 function in cardiovascular development in more detail. Our data demonstrate that thePitx2c isoform is expressed in a presumptive secondary heart field that invades the heart after looping morphogenesis. Pitx2c was expressed in a subpopulation of left branchial arch and splanchnic mesoderm apposed to forming branchial arch arteries (BAAs) and in left aortic sac mesothelium. An isoform-specific deletion of Pitx2c, generated by gene targeting in embryonic stem cells, revealed that Pitx2cfunctions to regulate asymmetric BAA remodeling and to pattern the outflow tract (OFT). Fate-mapping studies with a Pitx2 cre knock-in allele revealed that Pitx2 daughter cells invade the AV cushions and valves in a Pitx2-dependent fashion, suggesting a role for Pitx2 in local cell movement or survival within the heart. Our results provide insight into Pitx2 function in post-looping cardiac morphogenesis and in BAA remodeling.

To generate the Pitx2 δc neo targeting vector, we replaced the Pitx2 exon 4 that encodes the Pitx2c-specific exon with a LoxP flanked PGKneomycin cassette. The Pitx2 δc neo targeting vector introduced a novel EcoRV site into the mutant Pitx2locus that we used to screen for homologous recombination events by Southern blot using flanking probes.

After homologous recombination, the Pitx2 δc neo allele resulted in deletion of the majority of exon 4, including all coding sequences within this exon. The Pitx2 δc neo targeting vector was electroporated into AK7 ES cells, targeted clones identified by Southern blot,and injected into 3.5 dpc C57BL/6J mouse embryos to generate chimeras. To induce recombination between the two loxP sites and remove PGKneomycin cassette, we crossed Pitx2 δc neo chimeras to CMVCre recombinase deleter strain. Pitx2 δc neo and δcalleles were maintained on a mixed 129/Sv×C57BL/6J genetic background.

The Pitx2 δabc^creneo will be described elsewhere. Briefly, to generate this allele, an IRES cre PGKneomycin cassette was introduced into the PvuII and NruI sites of Pitx2exon 5, that encodes part of the homeodomain, generating a null allele ofPitx2. Wholemount in situ with a cre recombinase(cre) probe confirmed that expression of cre recapitulated endogenous Pitx2 expression pattern.

Whole-mount in situ hybridization was performed as described(Lu et al., 1999b). ThePitx2c probe was a 1 kb genomic fragment containing exon 4 that was linearized with XhoI and transcribed with T7 polymerase. The semaphorin 3c probe has been described previously(Brown et al., 2001) and thecre probe was a cDNA fragment that was linearized with EcoRI and transcribed with T7 polymerase.

For histology, embryos were fixed overnight in buffered formalin,dehydrated through graded ethanol and paraffin embedded. Sections were cut at 7-10 μm and H&E stained. lacZ staining was described(Lu et al., 1999a).

Injection of casting dye: 18.5 dpc embryos were harvested and sternum removed. Yellow casting dyes (Connecticut Valley Biological Supply) were injected into right ventricles using a capillary pipette, followed by blue dye into left ventricle. Corrosion casts: 18.5 dpc embryos were isolated and the heart exposed by a thoracic incision. Batson number 17 acrylic (Polysciences)was injected into right and left ventricles until great arteries were filled. After hardening overnight in distilled water at 4°C, tissues were removed with Maceration Solution at 50°C for 24 hours without shaking.

Embryos were dissected and placed in ice cold PBS. Individual embryos were placed in warm PBS to facilitate ventricular contractions. Using a pulled glass pipette, India Ink was injected into ventricles until ink penetrated small vessels. Embryos were post fixed in 10% formalin and cleared in benzyl alcohol:benzyl benzoate (2:1).

Previous studies have shown that the Pitx2c isoform is asymmetrically expressed in the developing embryo while the Pitx2aand Pitx2b isoforms are co-expressed with Pitx2c in symmetrical regions of the embryo(Kitamura et al., 1999;Liu et al., 2001;Schweickert et al., 2000;Yu et al., 2001). For example,Pitx2c is expressed in left lateral plate mesoderm and in the left side of most organ primordial such as developing guts, heart and lungs. The three Pitx2 isoforms in mice are co-expressed in periocular mesenchyme, oral and dental epithelium, as well as anterior body wall. A fourth Pitx2 isoform, Pitx2d, has recently been described in humans (Cox et al., 2002).

Because of the correct dextral looping in Pitx2 null embryos, we hypothesized that Pitx2 functioned in a cell population that contributed to the heart during or after cardiac looping. Experiments performed in chick and mouse embryos have revealed that cells outside the primary heart field contribute to conotruncal development. One cell population originates in the splanchnic and branchial arch mesoderm, and migrates into the OFT and right ventricle of the looped heart(Kelly et al., 2001;Mjaatvedt et al., 2001;Waldo et al., 2001). We found that cells within this presumptive secondary heart field expressPitx2c asymmetrically at 9.5 dpc both as they migrate and after populating the OFT and right ventricle, revealing that this cell population has laterality.

Whole-mount in situ using a Pitx2c-specific probe on 9.5 dpc embryos revealed left-sided Pitx2c expression in splanchnic mesoderm at the level of and just caudal to the OFT(Fig. 1A,B). Serial sectioning showed that Pitx2c expression in left splanchnic mesoderm was continuous with expression in left aortic sac mesothelium and OFT myocardium(Fig. 1C-F). Expression of thePitx2a and Pitx2b isoforms was not detected in the presumptive secondary heart field or developing OFT (not shown).

We examined coronal and parasagittal sections through 9.5 and 10.5 dpc embryos to investigate in more detail the Pitx2c expression pattern during these timepoints prior to remodeling of the BAA. Ventral coronal sections through 9.5 dpc embryos showed left-sided Pitx2c expression in aortic sac mesothelium near the junction of the aortic sac and the BAA(Fig. 1F). More dorsal coronal sections revealed low levels of Pitx2c expression in left branchial arch and splanchnic mesoderm in proximity to the third BAA(Fig. 1G). Serial parasagittal sections through 10.5 dpc embryos showed Pitx2c expression in ventral branchial arch and splanchnic mesoderm that was continuous withPitx2c expression in left branchial arch mesoderm evident on more lateral sections (Fig. 1H,I). The parasagittal sections at this timepoint also revealed the diminished intensity in Pitx2c expression dorsally towards the dorsal aorta(Fig. 1I). Our expression studies also showed Pitx2c expression in the left atrium, primary interatrial septum, left dorsal mesocardium, and right ventricular and interventricular myocardium (Fig. 1J-M). From these studies, we conclude that Pitx2c is asymmetrically expressed in a subpopulation of the presumptive secondary heart field that contributes to the OFT and right ventricular myocardium after cardiac looping, suggesting that Pitx2c provides laterality to the OFT and right ventricle myocardium. Moreover, asymmetric Pitx2cexpression in ventral branchial arch and splanchnic mesoderm, with higher levels ventrally near the junction of the BAA and aortic sac, suggests a role for Pitx2c in formation of the BAAs.

The Pitx2c isoform is encoded by exons 4, 5 and 6, and uses a distinct promoter from that of the Pitx2a and Pitx2bisoforms (Shiratori et al.,2001). The Pitx2 exon 5 and exon 6, which encode the homeodomain, are common to all Pitx2 isoforms in mice(Fig. 2A,B). To directly investigate Pitx2c function using a loss-of-function approach in mice, we constructed a targeting vector that replaced thePitx2c-specific exon 4 with a PGKneomycin LoxP cassette, thePitx2 δc neo targeting vector(Fig. 2B,C). Upon germline transmission, the Pitx2 δc neo allele was crossed to thecmv cre recombinase deletor strain to remove the PGK neomycin cassette and generate the final Pitx2 δc allele(Fig. 2C,D). Both Pitx2δc neo and Pitx2 δc^-/- mutants were obtained at the Mendelian ratio at 18.5 dpc. Mutant neonates were born alive but quickly became cyanotic and died a few minutes after birth. We noted thatPitx2 δc^-/- mutants turned normally, suggesting that the Pitx2a and Pitx2b isoforms have redundant function withPitx2c in turning or body wall closure as Pitx2a, Pitx2bhomozygous mutant embryos also turn normally(Liu et al., 2001)(Fig. 2E).

To determine if Pitx2c had a role in patterning the great vessels of the aortic arch, we performed casting dye and corrosion cast experiments on 18.5 dpc Pitx2 δc^-/- embryos. Among the 21 mutant embryos examined, 57% (12 out of 21) had the wild-type pattern of left aortic arch with right innominate artery while 29% (six out of 21) had right aortic arch with left innominate artery (Fig. 3A-D). In addition, 14% (three of 21) showed double aortic arch without innominate artery (Fig. 3E-J). Of the double aortic arches, two were right dominant and the other left dominant. Thus, of all arches examined, 62% (n=13)were left dominant and the remaining 38% (n=6) were right dominant. In addition, all Pitx2 δc^-/- embryos had double outlet right ventricle (DORV) in which both the aorta and pulmonary artery drain the right ventricle (Fig. 3O,P). Blood exited the left ventricle of the Pitx2δc^-/- embryos through a ventricular septal defect.

The corrosion casting experiments revealed that Pitx2c had an important role in patterning of the BAAs. To determine if Pitx2c had a role in the initial formation of the BAAs or was important in BAA remodeling, we performed India ink injections at 11.0 and 11.5 dpc at the initiation of BAA remodeling. At these timepoints, all Pitx2c mutant embryos (n=6) formed symmetric BAAs that were indistinguishable from wild type littermates (Fig. 3K-N). From this, we conclude that Pitx2c functions in remodeling of the BAA. The very discrete, asymmetric Pitx2cexpression pattern within the region of the forming BAAs also supports the idea that Pitx2c would have a role in modulating BAA remodeling rather than in the initial endothelial tube assembly.

Great vessel remodeling and patterning of the conotruncal region, both defective in Pitx2c mutants, require normal development of the cardiac neural crest. To determine if cardiac neural crest contributed to the conotruncal region of Pitx2 δc^-/- embryos, we performed a fate-mapping experiment with the wntl cre transgenic line that directed cre expression to the precursors of the cardiac neural crest and the Rosa26 reporter line(Jiang et al., 2000;Soriano, 1999). creexpression will induce recombination at the Rosa26 locus resulting in expression of lacZ in all descendents of Wntl-expressing cells that include the cardiac neural crest. At both 11.5 and 12.5 dpc, we found that cardiac neural crest contributed normally to the conotruncal region and aortic and pulmonic valves of Pitx2 δc^-/-embryos suggesting that Pitx2 function in conotruncal cushion morphogenesis occurred subsequent to neural crest migration into thePitx2 mutant heart (Fig. 4A-F). Analysis of sections through the branchial arch arteries of 10.5 dpc wild type and Pitx2 δc^-/- embryos showed similar amounts of mesenchyme surrounding the arteries further supporting the idea that cardiac neural crest was correctly deployed in Pitx2δc^-/- embryos (Fig. 4G-J).

To determine how loss of Pitx2 affected development of OFT myocardium, we examined the expression of semaphorin 3c (Sema3c), an OFT myocardial marker, in wild type and Pitx2c mutants(Brown et al., 2001;Feiner et al., 2001). Although at 10.5 and 11.5 dpc, Sema3c was expressed normally inPitx2c mutant OFT myocardium (Fig. 5A-D), this expression was downregulated by 12.5 dpc(Fig. 5E,F). These data suggested that OFT myocardium was correctly specified and that migration of OFT myocardial precursors was intact in Pitx2c mutants.

To establish this more firmly, we used a Pitx2 cre knock-in allele(Pitx2 δabc^creneo), an allele of Pitx2 that expresses cre in the endogenous Pitx2 expression domain, to mark cells fated to express Pitx2. The Pitx2δabc^creneo allele has a cre recombinasePGKneomycin cassette introduced into Pitx2 exon 5 to generate a null Pitx2 allele, removing function of all Pitx2isoforms (see Materials and Methods, andFig. 5G,H). Moreover,expression of cre from the Pitx2 δabc^creneoqualitatively recapitulates the endogenous Pitx2 spatiotemporal expression pattern (not shown). At both 10.5 and 12.5 dpc, spatial expression of cre was similar in the OFT of the controlδabc^creneo heterozygotes andδabc^creneo;δabc^null Pitx2-null mutant embryos, supporting the idea that Pitx2 patterns the OFT myocardium after it is established (Fig. 5I,J and not shown). Taken together, these data support the notion that Pitx2 functions in branchial arch and splanchnic mesoderm, a developmental field that is distinct from cardiac neural crest. Moreover,downregulation of Sema3c expression in Pitx2 mutants suggests that Pitx2 has a role in maintenance of gene expression in OFT myocardium.

In addition to cardiac neural crest, OFT and inner curvature myocardium invades the cardiac cushions (van den Hoff et al., 1999; van den Hoff et al., 2001). To determine if descendents ofPitx2-expressing myocardium populated the cardiac cushions, we used the Pitx2 δabc^creneo allele and the Rosa26reporter allele to follow the fate of Pitx2 daughter cells afterPitx2 expression had been extinguished. At timepoints whenPitx2c is actively expressed in the heart, 9.5 dpc until 12.5 dpc,distinctions between Pitx2 daughter cells and newly labeledPitx2-expressing cells can be made in regions of the heart that never express Pitx2c. For example, at 9.5 and 10.5 dpcPitx2-expressing cells are restricted to the left side of the forming OFT (Fig. 1C-E). By contrast,lacZ-positive cells were detected on both sides of the OFT tract myocardium, suggesting that labeled Pitx2 daughter cells, found on the right side of the OFT, had moved from the left side(Fig. 6A). The distribution oflacZ-positive cells in the OFT tract in Pitx2 null embryos was similar to that of the wild type, suggesting that Pitx2 is not required for movement of the myocardial precursors from branchial arch mesoderm into the OFT (Fig. 6A,B). We noted that the number of lacZ-labeled cells in the OFT myocardium of 10.5 dpc embryos was less than what would be expected from the Pitx2c expression pattern. This may reflect the delay between cre transcription and Cre-mediated excision that requires the accumulation of adequate levels of Cre protein. Moreover, the delay in the readout is also lengthened by the need for transcription and translation oflacZ from the Rosa26 locus(Nagy, 2000).

At 12.5 dpc, when Pitx2c is still expressed in the heart, thePitx2 δabc^creneo allele cannot distinguish between newly labeled lacZ-positive cells and Pitx2c descendents that are no longer expressing Pitx2c. At this timepoint,lacZ-labeled cells were predominantly found in the myocardium overlying the interventricular groove with some cells found in the proximal OFT (Fig. 6C,D). By 14.5 dpc and 16.5 dpc, when Pitx2c expression is extinguished(Fig. 6F,H), there was an increase of lacZ-positive Pitx2c descendents over the medial aspect of the heart, suggesting an outward expansion of Pitx2cdescendents from the right ventricular and inner curvature myocardium(Fig. 6E,G). Sections through 16.5 dpc hearts, after Pitx2c expression had been extinguished,demonstrated that lacZ-positive Pitx2 daughter cells populated the myocardium of the proximal OFT, as well as the remodeled membranous and muscular ventricular septum and atrial septum(Fig. 6I,J).

Analysis of the fate of Pitx2 daughter cells in the Pitx2δabc^creneo; δc mutant embryos, that turned normally and survived longer than Pitx2 null embryos, revealed that fewer lacZ-positive cells were found in the right ventricular and inner curvature myocardium of Pitx2 mutant embryos at both 14.5 dpc(Fig. 6K,L) and 18.5 dpc(Fig. 6M,N). Serial transverse sections through the 18.5 dpc hearts revealed that in Pitx2δabc^creneo heterozygous hearts lacZ-labeled cells were found at the inferior border of the heart near the cardiac apex(Fig. 6O,Q). By contrast, in the Pitx2 δabc^creneo; δc mutantslacZ-labeled cells were not found at the inferior boundary of the heart (Fig. 6P,R). Moreover,sections through 12.5 dpc hearts revealed that lacZ-positive cells were found in the central AV cushion of Pitx2δabc^creneo heterozygous embryos, revealing thatPitx2 descendents contributed to the cushion mesenchyme(Fig. 6S). By contrast, in both 12.5 dpc and 14.5 dpc Pitx2-null mutant embryos,lacZ-positive cells were excluded from the central AV cushion mesenchyme suggesting that Pitx2 function was required for invasion of Pitx2 daughters into the AV cushion(Fig. 6T,U). As Pitx2cexpression is never detected in endocardium, this fate mapping data suggests a myocardial source for the lacZ-labeled cells in the AV cushion.

Defective valve morphogenesis is a common feature in human patients with laterality defects and Pitx2-null embryos(Brown and Anderson, 1999;Icardo and Sanchez de Vega,1991; Liu et al.,2001). At 16.5 dpc, lacZ-positive Pitx2descendents were detected in the AV valve leaflets of Pitx2δabc^creneo heterozygotes but were excluded from the valve leaflets of Pitx2 δabc^creneo; δc mutants(Fig. 6V,W).

Corrosion casting and scanning electron microscopy was used to analyze the morphology of the pulmonary veins in Pitx2 mutant embryos. The left superior caval vein (LSCV) normally flows into the coronary sinus, while the right superior caval vein (RSCV) and inferior caval vein (ICV) are connected to the right atrium (RA) by thin strips at the valves. The left and right pulmonary veins join to a common pulmonary vein (PV) that drains into the left atrium (LA) (Fig. 7A,C). In most Pitx2 δc^-/- embryos, morphology of these structures was defective, with all these veins running together into a common medial venous sinus (Fig. 7B,D). Consistent with this phenotype, fate mapping with thePitx2 δabc^creneo allele showed thatlacZ-positive Pitx2 daughter cells were observed bilaterally in the pulmonary veins of Pitx2 δabc^creneoheterozygous embryos but were severely reduced in the pulmonary veins ofPitx2 δabc^creneo; δc mutant embryos(Fig. 7E-H).

In this work, we provide evidence that Pitx2c patterns a presumptive secondary heart field, the branchial arch mesoderm, that invades the heart after looping and contributes to the OFT and right ventricular myocardium. This finding is consistent with the phenotypes of thePitx2-null embryos that have correct dextral looping of the heart tube but severe defects in cardiac morphogenesis(Gage et al., 1999;Kitamura et al., 1999;Lin et al., 1999;Liu et al., 2001;Lu et al., 1999b). We also show that Pitx2c has an important role in asymmetric remodeling of the BAAs. Moreover, our data reveal that Pitx2 daughter cells invade the AV cushions and valves and that this cellular movement into the cushions requires Pitx2 function. The data presented here provide insight into the phenotypes observed in humans with laterality syndromes and demonstrate a direct causal link between the genetic pathways regulating left right asymmetry and complex cardiac morphogenesis.

Recent advances have revealed that the primary heart field receives contributions from a number of secondary fields. Functional studies have implicated the cardiac neural crest in patterning of the aortic arch vessels and conotruncus of the heart. For example, mice with mutations in components of the endothelin signaling pathway(Yanagisawa et al., 1998),forkhead genes (Iida et al.,1997; Kume et al.,2001; Winnier et al.,1999) and splotch mutant mice(Epstein et al., 2000) have defective arterioventricular connections secondary to cardiac neural crest abnormalities. Moreover, inactivation of Sema3c and neuropilin 1 leads to faulty conotruncal cushion formation as a result of aberrant cardiac neural crest migration (Brown et al.,2001; Feiner et al.,2001; Kawasaki et al.,1999). By contrast, our data reveal that Pitx2c has an important role in patterning a separate heart field derived from the branchial arch and splanchnic mesoderm.

We found that Pitx2c is expressed asymmetrically in the left branchial arch and splanchnic mesoderm within cells that will contribute to the OFT myocardium. Moreover, Pitx2c is also expressed in OFT myocardium and right ventricular myocardium, regions of the heart that are populated by cells derived from branchial arch and splanchnic mesoderm, but not by cardiac neural crest (Jiang et al.,2000; Kelly et al.,2001; Mjaatvedt et al.,2001; Waldo et al.,2001). Our fate-mapping studies with the Wnt1 cre andRosa26 reporter mice also show that cardiac neural crest migrates normally in Pitx2 mutants.

Our data suggest that Pitx2c is not required for the initial migration of branchial arch mesoderm into the outflow tract. Analysis ofSema3c expression in Pitx2c mutants suggests that the defect in Pitx2c mutants occurs relatively late in conotruncal development. Moreover, cre expression in Pitx2 δabc^creneo;δabc^null mutants was similar to that observed in theδabc^creneo heterozygous OFT. One idea to explain these data is that Pitx2 functions to maintain signaling between the outflow tract myocardium and underlying endothelium and forming conotruncal cushions. This model would be similar to what has been proposed forPitx2 in craniofacial development where Pitx2 has a role in epithelial-mesenchymal signaling important for tooth organogenesis(Lin et al., 1999;Lu et al., 1999b). Another possibility, based on the ventricular myocardial defect observed inPitx2 mutant embryos (see below), is that Pitx2 regulates local expansion of OFT myocardium. We favor the first hypothesis as we do not detect differences in the number or localization of lacZ-labeled cells in the OFT of wild-type and Pitx2 mutant embryos. Nonetheless,it is still formally possible that subtle differences in OFT myocardial expansion could be responsible for the conotruncal defect observed inPitx2 mutants. Taken together, our findings support the idea thatPitx2 patterns the branchial arch mesoderm and OFT myocardium to support normal development of the conotruncus.

Development of the branchial arch arteries and subsequent remodeling into the mature aortic arch arteries involves a series of paracrine signaling events (Fishman and Kirby,1998; Hanahan,1997; Yancopoulos et al.,2000). The forming BAA endothelial tubes are located within the branchial arch mesoderm in close proximity to surface ectoderm and the endoderm-derived epithelium of the branchial pouches. Signaling from endothelium to mesenchyme is thought to be important for recruitment of supporting cells, such as smooth muscle precursors and pericytes, which are important for stabilization of the forming endothelial tubes(Hanahan, 1997).

Vascular remodeling involves local disruption of the critical interaction between endothelium and support cells with resulting regression of the endothelium. In one system, endothelial regression occurs by programmed cell death secondary to loss of survival factors(Meeson et al., 1996;Meeson et al., 1999). The mechanisms underlying asymmetric remodeling of the BAAs, resulting in left-sided aortic arch, are poorly understood.

Cell ablation studies in chick embryos and loss-of-function experiments performed in mice have defined a role for cardiac crest in maintaining the integrity of branchial arch arteries (Brown et al., 2001). However, recent fate mapping experiments usingWnt1 cre and Rosa26 reporter mice, while confirming the importance of the cardiac crest in mouse BAA formation, suggest that cardiac neural crest does not provide the signal for asymmetric remodeling of the BAAs(Jiang et al., 2000).

The important role of branchial arch endoderm in BAA development has been illustrated by phenotypes of individuals with DiGeorge syndrome and mouse models of this syndrome that include severe defects in aortic arch artery formation (Lindsay et al.,2001; Merscher et al.,2001). Importantly, defects were observed more commonly in the right fourth BAAs of a haploinsufficent mouse model for DiGeorge Syndrome(Lindsay and Baldini, 2001). The gene implicated in these events, Tbx1, is expressed in branchial arch endoderm, suggesting that endoderm-derived signals may have a role in asymmetric remodeling of BAA.

Pitx2c is expressed asymmetrically in a very discrete population of cells in proximity to the left aortic sac and left BAAs. Despite this restricted expression, there is a strong BAA phenotype in Pitx2cmutants. These observations suggest that Pitx2c may have a role in recruitment or maintenance of supporting cells to the left BAAs and aortic sac. In wild-type embryos, Pitx2c may be important for stabilization of left-sided BAAs, such as the sixth BAA, that will form the left-sided ductus arteriosus. In the absence of Pitx2c function, maintenance of the sixth BAA would be impaired, resulting in formation of a right-sided ductus arteriosus in some embryos. This alteration would initiate a cascade,perhaps resulting from the altered hemodynamics of the persistent right-sided sixth BAA, to alter remodeling of the other BAAs. Although these ideas will need to be verified in future experiments, our data provide new information about the role of Pitx2 in asymmetric remodeling of the BAA.

Our data suggest that Pitx2 has a greater role in AV cushion morphogenesis when compared with formation of the conotruncal cushions.Pitx2-null embryos have severe defects in the central mesenchymal mass that forms the AV cushions and valves resulting in complete AV canal(Kitamura et al., 1999;Liu et al., 2001). The conotruncal phenotype is a failure of rotation of the truncus arteriosus and conotruncal cushion dysmorphology(Kitamura et al., 1999;Liu et al., 2001). Genetic evidence from mice implicates Bmp-signaling in conotruncal cushion morphogenesis (Kim et al.,2001). Noggin overexpression experiments performed in chick embryos revealed that Bmp signaling in conotruncal cushion formation functioned through a mechanism involving regulation of cardiac neural crest migration (Allen et al., 2001). Less is known about the signaling pathways that regulate AV cushion morphogenesis, although recent experiments suggest that Bmp-signaling has a central role (Gaussin et al.,2002).

Data from zebrafish suggest that composition of matrix is of crucial importance in the initial formation of valves and implicate Wnt and Bmp signaling in these events (Walsh and Stainier, 2001). In vitro studies suggest that the action of matrix metalloproteases on cushion mesenchyme is required for migration of mesenchyme into the forming cushions (Song et al., 2000). This epithelial-mesenchymal transition that leads to cushion deposition requires Tgfβ signaling(Brown et al., 1996;Brown et al., 1999). In addition, Tgfβ2 null mice have multiple defects in valve and septal morphogenesis, implicating this signaling pathway in cushion morphogenesis(Bartram et al., 2001;Sanford et al., 1997). Our data reveal that Pitx2 has a role in regulating cellular movement into the formed AV cushion, a late step in cushion morphogenesis. One idea to explain these data is that Pitx2c is required for the myocardial invasion of AV cushion mesenchyme. Pitx2c is expressed in the inner curvature myocardium that surrounds the AV cushion and these myocardial cells have been shown to invade the AV cushion mesenchyme(van den Hoff et al., 2001). However, another possible source of cells that invade the AV cushion is dorsal mesocardium that also expresses Pitx2c. Further experiments are currently under way to elucidate the exact source of invading Pitx2daughter cells.

The data presented here extend our previous understanding of Pitx2function in development of the venous pole of the heart. Previous studies have demonstrated an important role for Pitx2 in patterning of the atrial appendages and atrial septation (Kitamura et al., 1999; Liu et al.,2001). Analysis of the Pitx2c mutants also reveal a role for Pitx2 in morphogenesis of the pulmonary and caval veins. Fate mapping suggests a direct role for Pitx2 in vein morphogenesis asPitx2 daughters populate pulmonary and caval veins. Moreover,diminished contribution of Pitx2 daughters to the Pitx2mutant pulmonary vein suggests a role for Pitx2 in cell movement or cell sorting that may be similar to Pitx2 function in AV cushion morphogenesis. Alternatively, Pitx2 may function to regulate proliferation or survival of pulmonary vein and AV cushion progenitors.

Pitx2c is expressed in the right ventricular and inner curvature myocardium (Campione et al.,2001; Schweickert et al.,2000). Our fate mapping experiment revealed that Pitx2daughter cells expand to extensively populate both right and left ventricular myocardium. In Pitx2 mutant embryos, fewer Pitx2 daughters are observed contributing to ventricular myocardium. Moreover, analysis ofcre expression in Pitx2 mutants, that marks the right ventricle, suggested that the size of the right ventricle was reduced inPitx2 mutants. One interpretation of these data is thatPitx2 functions in growth of the right ventricular myocardium. Further experiments will be required to distinguish between defective movement of precursors into the right ventricle and failure of the right ventricular myocardium to proliferate.

We thank R. Behringer, A. Bradley and P. Soriano for reagents; J. Epstein for sema3c in situ probe; A. McMahon and D. Rowitch for wnt1 cre transgenic line; and A. Baldini for critical comments and insightful discussions. C. L. was supported in part by Harry S. and Isabel C. Cameron Foundation. Supported in part by a grants from NIDCR (R29 DE12324 and R01DE013509), by grant number 5-FY00-135 from March of Dimes to J. F. M., and by the British Heart Foundation (RG/98004 to N. A. B.).