Frs2α-deficiency in cardiac progenitors disrupts a subset of FGF signals required for outflow tract morphogenesis

Congenital heart disease (CHD) represents the most common birth defect in humans, affecting nearly 1% of newborns. Since more than 30% of CHD patients have outflow tract (OFT) defects (Rosamond et al., 2007), understanding how OFT development is controlled at the molecular level is of great interest. The OFT initially develops as an unseptated myocardial tube lined with endothelial cells, the endocardium. The OFT myocardium comprises cells deployed from a population of mesodermal cells called the second heart field (SHF) that resides in the pharyngeal and splanchnic mesoderm (SM), after formation of the primary heart tube(Buckingham et al., 2005; Srivastava, 2006). The OFT then undergoes dramatic remodeling and is divided into the right ventricular/pulmonary and left ventricular/aortic outflows at the arterial pole by the fusion and remodeling of the OFT cushions(Lamers and Moorman, 2002). This process is crucial for separating the pulmonary and systemic circulations postnatally. First, OFT myocardial cells secrete extracellular matrix molecules to form the OFT cushions. Subsequently, the matrix is invaded by cells from two sources: OFT endocardial cells that undergo an endothelial-to-mesenchymal transition (EMT) and neural crest cells (NCCs) that migrate from pharyngeal arches 3, 4 and 6. Disruption of the endocardial EMT or the contribution of NCCs to the OFT cushions can cause OFT defects(Armstrong and Bischoff, 2004; Hutson and Kirby, 2003; Moon et al., 2006).

The FGF family of regulatory polypeptides controls a broad spectrum of cellular processes during development and through adulthood(Eswarakumar et al., 2005; McKeehan et al., 1998). Emerging evidence demonstrates that several FGF ligands are involved in OFT development. Fgf8 and Fgf10 have been shown to be expressed in a temporally and spatially specific manner in the SHF progenitor cells residing in the SM and pharyngeal mesoderm core. Fgf8 is also expressed in the pharyngeal ectoderm and endoderm. Fgf15 is expressed in the pharyngeal endoderm (PE) and pharyngeal mesenchyme. Tissue-specific ablation of Fgf8 disrupts OFT remodeling, resulting in alignment and septation defects (Ilagan et al.,2006; Park et al.,2006). Fgf15 loss-of-function also causes OFT alignment defects (Vincentz et al.,2005). Furthermore, the FGF signaling axis genetically interacts with Tbx1 pathways that regulate the development of OFT and pharyngeal arch derivatives in a tissue-specific manner(Aggarwal et al., 2006; Vitelli et al., 2006). Disruption of Tbx1 function is linked to DiGeorge and other syndromes associated with the microdeletion of chromosome 22q11.2(Lindsay et al., 2001).

The FGFs exert their regulatory activity by activating FGF receptor (FGFR)tyrosine kinases, which are encoded by four highly homologous genes, in partnership with peri-cellular matrix heparan sulfate proteoglycans(McKeehan et al., 1998; Ornitz, 2000). Thus far, only a few intracellular signaling molecules have been shown to bind to FGFRs directly. These include phospholipase Cγ(Mohammadi et al., 1992; Peters et al., 1992), CRK(Larsson et al., 1999), CRKL(Moon et al., 2006),FRS2α (FRS2; SNT1) and FRS2β (FRS3; SNT2)(Kouhara et al., 1997; Ong et al., 1996). FRS2αis a proximal-interactive adaptor protein that has six tyrosine phosphorylation sites that are phosphorylated by FGFRs upon activation by FGF ligands. Among them, phosphorylated Y196, Y306, Y349 and Y392 are GRB2 binding sites that link FGFR kinases to the PI3 kinase pathway; phosphorylated Y436 and Y471 are SHP2 (PTPN11 - Mouse Genome Informatics) binding sites that link FGFR kinases to the MAP kinase pathway(Kouhara et al., 1997; Ong et al., 2000; Zhang et al., 2008). Deleting the FRS2α-binding VT (valine and threonine) dipeptide motif in the intracellular juxtamembrane domain of FGFR1 in mice leads to defects in multiple organs (Hoch and Soriano,2006). In addition to FGFRs, several other receptor tyrosine kinases have been reported to phosphorylate FRS2α, although the roles of FRS2α in these signaling pathways remain poorly defined(Avery et al., 2007; Ong et al., 2000).

Frs2α is expressed in mouse embryos during early embryogenesis and almost ubiquitously in all fetal and adult tissues(McDougall et al., 2001). Complete disruption of Frs2α function abrogates FGF-induced activation of MAP and PI3 kinases, chemotactic responses and cell proliferation, and also causes lethality at embryonic day (E) 7-7.5(Hadari et al., 2001). Mice carrying mutations in the two SHP2 binding sites exhibit a variety of developmental defects in many organs, whereas mice carrying mutations in the four GRB2 binding sites generally have less severe phenotypes(Yamamoto et al., 2005).

In order to circumvent the early embryonic lethality resulting from Frs2α ablation and to examine the roles of FRS2α-mediated signals in heart morphogenesis, we employed the Cre-loxP recombination system to tissue-specifically inactivate Frs2α alleles in heart progenitor cells. Ablation of Frs2α in the SHF caused OFT misalignment, including overriding aorta (OA) and double-outlet right ventricle (DORV), by compromising SHF progenitor cell proliferation and thereby reducing the contribution of the SHF-derived mesodermal cells to the OFT myocardium. In addition, ablation of Frs2α in the OFT myocardium increased VEGF expression and disrupted endocardial EMT. Deletion of Frs2α in the SHF and PE reduced Bmp4 expression and disrupted NCC invasion of the OFT cushions, resulting in persistent truncus arteriosus (PTA), a severe OFT septation defect. Furthermore, double ablation of Fgfr1 and Fgfr2 phenocopied ablation of Frs2α. The results suggest that FRS2α-mediated signals in OFT myocardial cells promote endocardial EMT by downregulating VEGF expression, whereas in SM and PE, the FRS2α-mediated signals upregulate BMP4 expression and promote the NCC contribution to the OFT. Together, the results delineate a molecular mechanism of how FGF elicits tissue-specific signals to regulate OFT development.

All animals were housed in the Program for Animal Resources of the Institute of Biosciences and Technology, Texas A&M Health Science Center,and all experimental procedures were approved by the Institutional Animal Care and Use Committee. The mice carrying Frs2α^flox,Fgfr1^flox, Fgfr2^flox, Nkx2.5^Creknock-in alleles and Mef2c^Cre, Wnt1^Cre and Tie2^Cre (Tie2 is also known as Tek - Mouse Genome Informatics) transgenic alleles were bred and genotyped as described (Danielian et al., 1998; Kisanuki et al.,2001; Lin, Y. et al.,2007; Moses et al.,2001; Trokovic et al.,2003; Verzi et al.,2005; Yu et al.,2003). All mice were on a mixed background (including C57/BL6,SV129 and ICR) and all embryos arising from each cross were outbred. The hearts were excised at the stages indicated in the text, fixed with 4%paraformaldehyde (PFA) in PBS for 4 hours and paraffin embedded. The sections were rehydrated and stained with Hematoxylin and Eosin for histological analysis.

Immunostaining was performed on 5-μm sections mounted on Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA). The antigens were retrieved by incubation in citrate buffer (10 mM) for 20 minutes at 100°C or as suggested by manufacturers of the antibodies. The source and concentration of primary antibodies are: mouse anti-AP2α (1:10 dilution) and anti-ISL1(1:500) from Development Studies Hybridoma Bank; anti-phosphorylated ERK(1:100), anti-phosphorylated AKT (1:100) and anti-phosphorylated SMAD1/5/8(1:500) from Cell Signaling (Danvers, MA); anti-FRS2α (1:100),anti-NFATC1 (1:100), anti-phosphorylated histone H3 (1:100), anti-VEGFA(1:100) and anti-PECAM (1:100) from Santa Cruz (Santa Cruz, CA). The specifically bound antibodies were detected with HRP-conjugated secondary antibody (Bio-Rad, Hercules, CA) and visualized using TSA Plus Fluorescence Systems from Perkin Elmer (Boston, MA) on a Zeiss LSM 510 confocal microscope. For TUNEL assays, tissues were fixed and sectioned, and the apoptotic cells were detected with the ApopTag Peroxidase In Situ Kit from Chemicon (Temecula,CA).

Whole-mount in situ hybridization was performed as previously described(Edmondson et al., 1994). Briefly, after fixation in 4% PFA in PBS overnight, embryos were treated with 10 μg/ml protease K for 5-15 minutes at room temperature, and post-fixed with 4% PFA in PBS for 20 minutes. After prehybridization at 70°C for 2 hours, the hybridization was carried out by overnight incubation at 70°C. Following the hybridization, embryos were washed with TBST [0.25 M Tris-HCl-buffered saline (pH 7.5), 0.01% Triton X-100, 2 mM levamisole],blocked with 10% sheep serum in TBST, and then rocked overnight at 4°C in a 1:4000 dilution of alkaline phosphatase-conjugated anti-digoxigenin antibody(Roche, Indianapolis, IN) in blocking buffer. After eight washes with TBST,specifically bound antibodies were visualized by alkaline phosphatase staining. The embryos were fixed post-hybridization with 4% PFA in PBS. At least three mutants and three control embryos were analyzed for each probe.

Short-term mouse embryo culture was performed as described previously(Grego-Bessa et al., 2007). Briefly, E8.5 mouse embryos were dissected in PBS and cultured in 12-well plates containing 1% agarose to avoid embryo attachment; 1 ml DMEM medium containing 50% FBS and antibiotics (penicillin and streptomycin, 100 U/ml each) was added on top of the agarose. Inhibitors for phosphorylated ERK and PI3K (LY294002, Calbiochem, Darmstadt, Germany) were added to the media to a final concentration of 10 μM. Embryos were cultured in a 5% CO₂incubator at 37°C for 16 hours. The embryos were then dissected in PBS and then fixed in 4% PFA in PBS for further analyses.

Expression of Frs2α in the cardiac mesoderm could be detected as early as the 0-somite stage (ss) by immunostaining(Fig. 1A). To investigate FRS2α function in heart development, we ablated Frs2α in cardiac progenitors by crossing mice bearing the Frs2αconditional null (Frs2α^flox) allele(Lin, Y. et al., 2007) with mice carrying the Nkx2.5^Cre knock-in allele(Moses et al., 2001) or the Mef2c^Cre transgene(Verzi et al., 2005). Nkx2.5^Cre is expressed at the 0 ss and directs Cre activity in the first heart field (FHF), SHF and the PE. Mef2c^Cre is active at the 0-1 ss in the specified SHF, but not in the FHF and PE.

At E9.5, Frs2α was highly expressed in the PE and SM(Fig. 1Ba). In Nkx2.5^Cre;Frs2α^f/f embryos(designated Frs2α^cn/Nkx), Frs2α expression was efficiently disrupted in the OFT myocardium and endocardium, SM and PE, as well as in the atrial and ventricular myocardium of E9.5 embryos(Fig. 1Bb). In Mef2c^Cre;Frs2α^f/fembryos (designated Frs2α^cn/Mef),disruption of Frs2α expression was complete in the SM, OFT myocardium and right ventricle myocardium(Fig. 1Bc), and mosaically in the OFT endocardium. Detailed timecourse analyses showed that the FRS2αprotein level was significantly reduced by the 8-9 ss in Frs2α^cn/Nkxembryos, and by the 11-12 ss in Frs2α^cn/Mefembryos (see Fig. S1 in the supplementary material), which is consistent with the relatively delayed activity of the Mef2c^Cre driver(Verzi et al., 2005). Together, these data indicate that the Frs2α^cn/Nkxembryos had an efficient and widespread Frs2α deletion, whereas the Frs2α^cn/Mefembryos had a more restricted and delayed Frs2α deletion. These two conditional mutants provided an opportunity to dissect the tissue-specific roles of FRS2α in the OFT.

The majority of Frs2α^cn/Nkx or Frs2α^cn/Mef embryos died neonatally with severe OFT malformations (Tables 1, 2, 3). The few surviving neonates died within 3 weeks after birth. We examined hearts from E14.5 fetuses and found that both Frs2α^cn/Nkx and Frs2α^cn/Mef hearts exhibited OFT alignment defects, including OA (Fig. 2A-C) and DORV (Fig. 2D-F). Frequently, the Frs2α^cn/Nkx OFT was completely unseptated or partially septated into aorta and pulmonary artery(Fig. 2H), which is classified as PTA. By contrast, none of the Frs2α^cn/Mef fetuses exhibited septation defects, suggesting that FRS2α-mediated signaling in the OFT myocardium and SM within the Mef2c^Cre lineage is not obligatory for OFT septation.

In addition to the OFT myocardium and SHF, Nkx2.5^Crealso mediated Frs2α ablation in the OFT endocardium(Fig. 1Bb2) and the PE(Fig. 1Bb3), suggesting that FRS2α-mediated signals in the PE and OFT endocardium are required for OFT septation. However, ablation of Frs2α in the OFT endocardium with Tie2^Cre(Kisanuki et al., 2001) did not disrupt OFT morphogenesis (see Fig. S2 in the supplementary material). Therefore, the results suggest that FRS2α-mediated events in the PE, but not the endocardium, are required for OFT septation. Since Nkx2.5^Cre is expressed slightly earlier than Mef2c^Cre, the different OFT phenotypes in the Frs2α^cn/Nkx and Frs2α^cn/Mef mutants also suggest that the need for FRS2α-mediated signals in OFT septation occurs early in the SHF. Although Nkx2.5^Cre knock-in mice have no OFT defect(Moses et al., 2001), the data do not rule out the possibility that the phenotype in Frs2α^cn/Nkx OFT is a synergistic effect of FRS2α deficiency and Nkx2.5 haploid insufficiency.

During early OFT development, cells are deployed from the SHF to the OFT starting at ∼E8.25; ablation of the SHF in chicken embryos has been shown to cause OFT misalignment (Ward et al.,2005). The SHF is the common domain of Nkx2.5^Cre- and Mef2c^Cre-mediated ablation, and the common OFT misalignment phenotypes in these two mutant classes suggest that FRS2α-mediated signals in the SHF promote the progenitor cells residing in the SHF to contribute to the OFT myocardium. To test this possibility,immunostaining was employed to assess the number of cells expressing ISL1, a transcription factor that is expressed in the SHF and PE and which is crucial for the ability of these progenitor cells to proliferate and ultimately contribute to the OFT (Cai et al.,2003; Park et al.,2006).

Both Frs2α^cn/Nkx and Frs2α^cn/Mef embryos had a reduction in ISL1⁺ cells in the distal OFT myocardium at E9.5; Frs2α^cn/Nkx embryos also had fewer ISL1⁺ cells in the OFT endocardium (see Fig. S3 in the supplementary material). To test whether the decrease in ISL1⁺cells was due to a proliferation defect, double staining for phospho-histone H3 (pHH3) and ISL1 was carried out at the 11-12 ss (E8.5). The results revealed that the numbers of total ISL1⁺ cells and proliferating ISL1⁺ cells were both significantly reduced in the OFT and SM/SHF of Frs2α^cn/Nkx and Frs2α^cn/Mef mutants(Fig. 3A). Consistently, both Frs2α^cn/Nkx and Frs2α^cn/Mef OFTs were shorter than those of the controls (Fig. 3Ba-d). Lineage tracing with the R26R-lacZ reporter revealed that theβ-galactosidase-stained tissue in Frs2α^cn/Mef hearts was smaller than that in the heterozygous littermates (Fig. 3Be,f), suggesting that the contribution of the Frs2α-deficient SHF to the OFT and right ventricle was compromised. No clear difference in apoptosis was observed in the SM/SHF of Frs2α^cn/Nkx or Frs2α^cn/Mef embryos, although we detected increased apoptosis in the PE of Frs2α^cn/Nkx embryos (see Fig. S4 in the supplementary material). Together, the results suggest that the shared alignment defects in these two classes of mutants result from proliferation defects in the SHF.

Since FRS2α has multiple tyrosine phosphorylation sites that link the MAP kinase and PI3K/AKT pathways to FGFR kinases(Kouhara et al., 1997; Ong et al., 2000), we used immunostaining to investigate which FRS2α-mediated downstream signaling pathway was interrupted in the mutant mice. Less phosphorylated ERK (MAPK) was detected in the OFT, SM and PE of Frs2α^cn/Nkx embryos at the 11-12 ss, and in the OFT and SM of Frs2α^cn/Mef embryos at the same stage (Fig. 3Ca-c). No significant difference was observed in anti-phosphorylated AKT staining at the same stage (Fig. 3Cd-f). Consistent with these findings, treating cultured E8.5 embryos with an ERK1/2(MAPK3/1 - Mouse Genome Informatics), but not an PI3K/AKT, inhibitor significantly reduced the number of ISL1⁺ cells in the SM, PE and OFT (Fig. 3Da-c), although both inhibitors effectively and specifically inhibited ERK1/2 and AKT phosphorylation, respectively (Fig. 3Dd). These data indicate that the MAP kinase, but not the PI3K/AKT, pathway is essential for the correct behavior of SHF cells. The data are consistent with a previous report that the number of ISL1⁺cells in the SHF is rapidly reduced after Fgf8 ablation in the SHF(Park et al., 2006). PITX2 is a homeobox transcription factor essential for OFT rotation and alignment(Ai et al., 2006). Whole-mount in situ hybridization showed that expression of Pitx2 was unchanged in Frs2α^cn/Nkx and Frs2α^cn/Mef hearts (see Fig. S5 in the supplementary material), suggesting that the alignment defects in these mutants are not the result of altered Pitx2 expression.

OFT cushion formation and remodeling is a major process in OFT septation. Hematoxylin and Eosin (H&E) staining revealed that cellularity was reduced in both the proximal and distal segments of Frs2α^cn/Nkx, but not Frs2α^cn/Mef, OFT cushions at E10.5(Fig. 4A,B). At E11.5, the OFT cushions were hypoplastic and failed to fuse at the distal part(Fig. 4C). The results suggest that both EMT and NCC invasion were affected in Frs2α^cn/Nkx embryos and that these collectively contribute to the OFT septation defect.

PECAM (PECAM1) and NFATC1 are expressed in the endocardium. Downregulation of PECAM in the endocardium is a prerequisite for the EMT(Enciso et al., 2003). Immunostaining with anti-PECAM and anti-NFATC1 revealed that both PECAM and NFATC1 were present in the endocardium of control and Frs2α^cn/Mef OFT, and were diminished after the cells underwent EMT and invaded the cushion. By contrast, expression of PECAM and NFATC1 was sustained in Frs2α^cn/Nkx endocardial cells even after the cells had invaded the OFT cushion (Fig. 4D). This indicates that although some cells were successfully activated and could invade the matrix, they failed to complete EMT. Ex vivo culture experiments with E9.5 OFTs further demonstrated the EMT defects in the Frs2α^cn/Nkx myocardium(Fig. 4E). Interestingly,ablation of Frs2α alleles in the OFT endocardium with Tie2^Cre did not cause EMT and OFT defects (see Fig. S2 in the supplementary material), suggesting that FRS2α-mediated signals in the OFT endocardium are not essential for the EMT and that FRS2α-mediated signals regulate OFT endocardial EMT indirectly.

Given that OFT endocardial EMT is critically dependent on signaling molecules secreted by the OFT myocardium(Armstrong and Bischoff, 2004),and that the quantity of FRS2α protein is diminished in the SHF and in the forming OFT myocardium at the 8-9 ss in Frs2α^cn/Nkx, but not Frs2α^cn/Mef, mutants (see Fig. S1 in the supplementary material), it is likely that FRS2α-mediated signals in the myocardial precursors are required upstream of the myocardial signaling cascade that initiates and supports endocardial EMT before the 8-9 ss.

We found increased immunostaining for VEGFA in the Frs2α^cn/Nkx, but not the Frs2α^cn/Mef, OFT myocardium(Fig. 4F), which is consistent with a report that overexpression of VEGFA prevents nascent cushion endothelial cells from undergoing the EMT(Armstrong and Bischoff, 2004). It has been shown that myocardial NFATC2/3/4 promote EMT by suppressing VEGFA expression (Chang et al.,2004). An in vitro assay of mouse embryonic fibroblasts revealed that the transcriptional activity of NFAT is regulated by FGF in an FRS2α-dependent manner (Fig. 4G). These results suggest that FRS2α-mediated signals in the OFT myocardium repress VEGFA expression and promote the OFT endocardial EMT, probably through regulating the transcriptional activity of NFAT. No defects were found in the atrioventricular (AV) cushions of either Frs2α^cn/Nkx or Frs2α^cn/Mef hearts (see Fig. S6 in the supplementary material). It is possible that FRS2α-mediated signals do not regulate AV cushion formation, or that other pathways redundantly regulate the process, as the AV canal also consists of cells from other lineages(Galli et al., 2008; Hutson et al., 2006; Liao et al., 2004).

Because NCCs make crucial contributions to the distal OFT cushions, we examined migrating NCCs at E9.5 by immunostaining with anti-AP2α(TCFAP2α - Mouse Genome Informatics) antibody, which labels migrating NCCs and the ectoderm. The results showed that the number of migrating NCCs(AP2α⁺) in pharyngeal arches 3 and 4/6 and around the aortic sac in Frs2α^cn/Nkx embryos was reduced(Fig. 5A,B). By contrast, no change in the number of migrating NCCs was detected in Frs2α^cn/Mef embryos. Co-immunostaining with anti-AP2α and anti-pHH3 antibodies revealed significantly decreased proliferation in AP2α⁺ cells of Frs2α^cn/Nkx embryos(Fig. 5C,D). We did not detect an increase in apoptosis in NCCs of Frs2α^cn/Nkx embryos at E10.5 (see Fig. S4 in the supplementary material). These data suggest that the reduced number of NCCs in the OFT is a result of defective NCC proliferation, although NCC migration might also be affected. These defects are not attributable to Frs2α loss-of-function in NCCs because the Nkx2.5^Cre expression domain does not include the neural crest (Moses et al., 2001). Furthermore, ablation of Frs2α in NCCs with Wnt1-Cre(Danielian et al., 1998) did not cause OFT defects (data not shown). These findings indicate that FRS2α-mediated signals in NCCs are not required for NCC deployment and that the loss of FRS2α-mediated signals from the SM and/or PE causes secondary defects in the ability of NCCs to contribute to the OFT.

BMPs from the OFT myocardium, SM and PE control NCC patterning(Liu et al., 2004; Ma et al., 2005; Stottmann et al., 2004), and their expression is positively regulated by FGFs(Choi et al., 2005; Gotoh et al., 2005). Whole-mount in situ hybridization demonstrated that Bmp4 expression was reduced in the OFT, SM and PE of Frs2α^cn/Nkx embryos(Fig. 6Ah,k). By contrast,after ablation of Frs2α with Mef2c^Cre, Bmp4 expression was reduced in the OFT and SM, but not in the PE (Fig. 6A). Consistently, phosphorylation of SMAD1/5/8, which are downstream targets of the BMP receptor, was reduced in the same domains(Fig. 6B). These data suggest that in Frs2α^cn/Mef embryos, the net level of BMP signaling in the pharynx and OFT, although decreased, is sufficient to support NCC proliferation and migration into the OFT cushions, whereas in Frs2α^cn/Nkx mutants, BMP signaling falls below a critical threshold.

Among the four FGF receptors, Fgfr1 and Fgfr2 are broadly expressed in the pharyngeal region at E8.5 and Fgfr3 is only expressed in pharyngeal arch 1 (Trokovic et al., 2005; Wright et al.,2003). At E9.5, Fgfr2 expression persists in the PE and SM. At the same stage, Fgfr1 was expressed in the PE, but not in the SM (Fig. 7A and see Fig. S7 in the supplementary material). To test whether FGFR1 and FGFR2 synergistically regulate OFT development, we conditionally inactivated Fgfr1 and Fgfr2, individually or simultaneously, using Nkx2.5^Cre; these were denoted Fgfr1^cn/Nkx,Fgfr2^cn/Nkx and Fgfr1/r2^cn/Nkx,respectively. Histological analyses showed that Fgfr1^cn/Nkx embryos had no obvious OFT defects, whereas both Fgfr2^cn/Nkx and Fgfr1/r2^cn/Nkxmutants often developed OA and DORV (Table 3, Fig. 7B). In addition, Fgfr1/r2^cn/Nkx double mutants also exhibited the PTA phenotype (Fig. 7). Similar to our findings in Frs2α^cn/Nkx mutant OFTs,both the proximal and distal segments of Fgfr1/r2^cn/NkxOFT cushions were hypocellular (Fig. 8A). Immunostaining revealed sustained PECAM expression in cushion cells, increased VEGFA expression in the OFT myocardium, and reduced AP2α⁺ cell proliferation in the aortic sac and pharyngeal arches in Fgfr1/r2^cn/Nkx double mutants(Fig. 8B,C). The finding that Fgfr1/r2^cn/Nkx and Frs2α^cn/Nkx mutants had similar defects in endocardial EMT and NCC contribution suggested that FGFR1 and FGFR2 redundantly regulate OFT remodeling via FRS2α-dependent pathways. In addition, Fgfr1/r2^cn/Nkx embryos also exhibited reduced total and proliferating ISL1⁺ cell numbers and compromised activation of the MAP kinase, but not AKT, pathway(Fig. 8D), indicating that the FGFR1/2-FRS2α-MAP kinase signaling axis in the SM is required for regulating the accrual of SHF cells to the OFT myocardium.

Furthermore, the absence of OFT phenotypes in Fgfr1^cn/Nkx and all mutants carrying one conditional null allele indicates that the aforementioned defects of Frs2α^cn/Nkx, Fgfr2^cn/Nkx, or Fgfr1/r2^cn/Nkx mutants are not caused by Nkx2.5 heterozygosity because all these mutants had one Nkx2.5-null allele owing to the Cre knock-in.

Here we report that FRS2α-mediated signaling pathways in OFT myocardial precursor cells that reside in the SHF are required for normal expansion and deployment of these cells to the OFT myocardium. These FRS2α-mediated pathways are also required to indirectly regulate endocardial EMT and the recruitment of NCCs into the OFT cushions. Ablation of Frs2α in the SHF compromised MAP kinase activation and caused OFT alignment and septation defects. Our results are consistent with those presented in a companion study (Park et al., 2008), in which disrupting FGF signaling either by ablation of Fgfr1 and Fgfr2, or by overexpression of the FGF antagonist sprouty 2 (Spry2), in the SHF causes arterial pole defects. Frs2α^cn/Nkx hearts also exhibited defects in the atria and ventricles. This report focuses on the OFT defects.

Among the 22 FGF homologs, Fgf8 and Fgf10 are expressed in the SHF (Ilagan et al.,2006; Kelly et al.,2001; Park et al.,2006). Deficient or excessive FGF8 results in abnormal SHF development and OFT defects in a dosage- and spatiotemporal-specific manner(Hutson et al., 2006; Ilagan et al., 2006; Park et al., 2006). Ablation of Fgf10 results in mispositioning of the heart in the thoracic cavity (Marguerie et al.,2006). Although Fgf15 is expressed in the PE, it is unclear whether this mesodermal domain in the pharyngeal arches includes the SHF (Vincentz et al., 2005);loss of Fgf15 also causes OFT alignment defects. The OFT hypoplasia and misalignment we observed after ablation of Frs2α in the SHF are consistent with the requirements for these ligands during OFT morphogenesis and with the report that ablation of the Fgfr2IIIbisoform causes OFT alignment and ventricular septal defects (VSDs)(Marguerie et al., 2006). Ablation of Fgf8 function in heart precursors while they still reside in the primitive streak prevents formation of the OFT and right ventricle(Park et al., 2006). However,we did not observe this after ablation of Frs2α with Nkx2.5^Cre, suggesting either that the onset of Nkx2.5^Cre is too late to disrupt transduction of the FGF8 signals that regulate the earliest phases of SHF development, or that these early signals are not FRS2α-dependent. In addition, ablation of Fgf8 with Nkx2.5^Cre also results in a significant loss of the Nkx2.5^Cre lineage and in severe OFT and right ventricle truncation by E9.5. Fgf8;Nkx2.5^Cre mutants have significant decreases in cell proliferation and increases in cell death in both the PE and SM (Ilagan et al.,2006). Although Frs2α^cn/Nkxmutants also had a significant decrease in cell proliferation in the PE and SM, no increase in cell death was found associated with Frs2α^cn/Nkx. Thus, the FGF8 signals that promote SHF cell proliferation are likely to be mediated via FRS2α, and those that prevent cell death in these two domains are likely to be elicited via FRS2α-independent pathways.

Here we show that ablation of Frs2α significantly reduced SHF cell proliferation (by 50%), as reflected in the shortened OFT. A similar correlation between compromised proliferation, shorted OFTs and disrupted OFT alignment and rotation has been reported elsewhere(Ilagan et al., 2006; Park et al., 2006). In addition, because Frs2α ablation did not prevent all SHF cell proliferation, it is likely that other signaling pathways are also involved. It has been shown that Wnt signaling can regulate SHF cell proliferation via both FGF-dependent and FGF-independent pathways(Cohen et al., 2007; Lin, L. et al., 2007). Therefore, further efforts are needed to clarify this issue. No heart misposition phenotype was observed in Frs2α^cn embryos, implying that FGF10 regulates heart position independently of FRS2α-mediated signaling,although the data do not rule out the possibility that Nkx2.5^Cre and Mef2c^Cre are not expressed in the cells that regulate heart position.

In mice, Fgf8 deficiency phenocopies syndromes associated with 22q11 deletions in humans, which are characterized by cardiovascular, thymic,parathyroid and craniofacial defects(Frank et al., 2002; Vitelli et al., 2002), and Fgf8 modifies the vascular phenotypes resulting from Tbx1haploinsufficiency (Vitelli et al.,2006). Furthermore, loss of Crkl function, another gene in the 22q11 deletion that contributes the human phenotypes, disrupts phosphorylation of FRS2α downstream of FGF8/FGFR interactions(Guris et al., 2001; Moon et al., 2006). Our new findings implicate FRS2α-mediated signaling as a molecular mechanism underlying the cardiovascular features in these mouse models and, possibly, in affected humans. Moreover, our data uncover branchpoints in the downstream effector pathways that mediate distinct aspects of FGF signaling.

The process of endocardial EMT generates a subset of the OFT cushion mesenchymal cells and is regulated both by autonomous signals within the endocardium and by paracrine signals from the myocardium(Armstrong and Bischoff, 2004). We have shown that expression of Bmp4, which is required in the myocardium for EMT (J.F.M., unpublished) and NCC invasion(Liu et al., 2004), is decreased in response to Frs2α ablation in the SHF. We further demonstrate that Frs2α function is essential for the suppression of VEGFA expression in the myocardium. Myocardial NFATC2/3/4 promote EMT by suppressing VEGFA expression(Chang et al., 2004). Furthermore, the transcriptional activity of NFATC2/3/4 can be regulated by MAP kinase and PI3K/AKT pathways downstream of the FGF signaling axis(Macian, 2005). Indeed,ablation of Frs2α in mouse embryonic fibroblasts blocked NFAT transcriptional activity in response to FGF stimulation(Fig. 4F). In the accompanying report, Park and colleagues show that soluble factors from wild-type OFT can rescue EMT defects in ex vivo cultured Fgf8 mutant OFTs(Park et al., 2008).

Nkx2.5^Cre is homogenously expressed in the OFT endocardium, whereas Mef2c^Cre is mosaically expressed in this domain. However, we ruled out a requirement for Frs2α in the endocardium during EMT because ablating Frs2α with Tie2^Cre did not disrupt EMT. Together, our data suggest that the FGF signaling axis promotes endocardial EMT by promoting Bmp4 expression and suppressing myocardial VEGFA production via an FRS2α-MAP kinase-NFATC pathway.

Although Nkx2.5^Cre is not expressed in NCCs, ablation of Fgfr1/Fgfr2 or Frs2α with Nkx2.5^Cre reduced the NCC contribution to the OFT cushions. Notably, ablation of Frs2α in NCCs with Wnt1-Cre did not cause OFT defects (data not shown). Park and colleagues also demonstrate that ablation of FGF receptors or overexpression of the FGFR antagonist Spry2 in the SHF, but not in NCCs, also suppresses NCC invasion of the OFT cushions(Park et al., 2008). Interestingly, suppression of the FGFR signaling axis, either by ablation of Fgf8, Fgfr1/2 or Frs2α, or by overexpression of Spry2 (here and in the accompanying report), reduced the BMP expression in the SHF and PE that has been shown to be crucial for NCCs to contribute to the OFT cushion (Liu et al.,2004). Consistent with the expression pattern of Nkx2.5^Cre and Mef2c^Cre, Frs2α^cn/Nkx embryos had reduced Bmp4 expression in both SM and PE, whereas Frs2α^cn/Mef only had reduced Bmp4expression in the SHF (Fig. 6Aj,l). Given the fact that Mef2c^Cre is homogenously expressed in the SHF(Ai et al., 2007), the results suggest that early BMP4 signaling from both PE and SHF is required for NCC contribution to the OFT cushion. Together, these results indicate that the FGF8-FGFR/FRS2α signaling axis in the SM and/or PE indirectly regulates NCC contribution to the OFT cushion via the BMP signaling axis.

In summary, we report that an FRS2α-mediated FGF signaling pathway in the SHF and PE controls OFT extension and alignment by promoting the expansion of OFT precursors in the SHF, and also controls OFT septation by regulating OFT cushion formation through promoting the EMT of the cushion endocardium and NCC recruitment. Our findings provide a mechanism as to how the FGF signaling axis regulates OFT morphogenesis.

We thank Dr David M. Ornitz for the Fgfr2-floxed mice; Dr Juha Partanen for the Fgfr1-floxed mice; Dr Jeffery D. Molkentin for the NFAT reporter; Dr Wallace L. McKeehan, Ms Young Xu and Mary Cole for critical reading of the manuscript; and Kerstin McKeehan for excellent technical support. This work was supported by Public Health Service Grants NIH-CA96824,AHA0655077Y from The American Heart Association and DAMD17-03-0014 from the US Department of Defense.