The FGF-BMP signaling axis regulates outflow tract valve primordium formation by promoting cushion neural crest cell differentiation

Abstract

Rationale

Heart valves develop from precursor structures called cardiac cushions, an endothelial-lined cardiac jelly that resides in the inner side of the heart tube. The cushions are then invaded by cells from different sources, undergo a series of complicated and poorly understood remodeling processes, and give rise to valves. Disruption of the fibroblast growth factor (FGF) signaling axis impairs morphogenesis of the outflow tract (OFT). Yet, whether FGF signaling regulates OFT valve formation is unknown.

Objective

To study how OFT valve formation is regulated and how aberrant cell signaling causes valve defects.

Methods and results

By employing mouse genetic manipulation, cell lineage tracing, ex vivo heart culture, and molecular biology approaches, we demonstrated that FGF signaling in the OFT myocardium upregulated Bmp4 expression, which then enhanced smooth muscle differentiation of neural crest cells (NCCs) in the cushion. FGF signaling also promoted OFT myocardial cell invasion to the cushion. Disrupting FGF signaling interrupted cushion remodeling with reduced NCCs differentiation into smooth muscle and less cardiomyocyte invasion, and resulted in malformed OFT valves.

Conclusions

The results demonstrate a novel mechanism by which the FGF-BMP signaling axis regulates formation of OFT valve primordia by controlling smooth muscle differentiation of cushion NCCs.

Congenital valve defects occur in 2–3% of the population. Cardiac valves developed from cardiac cushions through a series of complicated and poorly defined processes. At early stages, the cardiac cushions are composed of extracellular matrix lined by endothelial cells. The cushions are invaded by endothelial-derived mesenchymal cells after completing the endothelial-mesenchymal-transition (EMT) and neural crest cells (NCCs) from pharyngeal arches ¹. The cushions subsequently undergo a series of remodeling processes. A portion of the proximal cushion undergoes remodeling and forms valve primordia and their supporting apparatus. The valve primordia are then gradually transformed from primary bulbous swollen cushions into thinly tapered valve leaflets ¹. Although cushion formation has been extensively studied, how valve primordia are defined and how valve remodeling is regulated remain unclear. It has been proposed that proximal outflow tract (OFT) myocardial cells receive inductive signals and migrate into the cushion. This process, termed as myocardialization ², subsequently induces surrounding mesenchymal cells to undergo muscle differentiation. These newly differentiated myocardial cells further induce their adjacent mesenchymal cells to differentiate until the entire mesenchymal structure is muscularized. Although required for OFT septation ², whether this process is required for valve formation is unclear.

The fibroblast growth factor (FGF) signaling axis has been shown to play important roles in OFT cushion formation. Yet, its role in OFT valve formation remains to be characterized. The FGF family consists of 18 receptor-binding members that regulate a broad spectrum of cellular activities ³, and play important roles in heart development ⁴. Among them, FGF4 has been shown to regulate valve precursor cell proliferation and differentiation ^5–7, and FGF8 contributes to OFT valve morphogenesis ^{8, 9}. FGF elicits its regulatory activities via activating FGF receptor (FGFR) tyrosine kinases encoded by four highly homologous genes. FGFR substrate 2α (FRS2α) is a broadly expressed adaptor protein that is required for the FGF to activate MAP and PI3 kinase pathways, the two major pathways in the FGF signaling cascade ^10–12. Frs2α null embryos die between embryonic (E) 7.0–7.5 days ¹³. Ablation of Frs2α, or double ablation of Fgfr1/Fgfr2, in heart progenitor cells disrupts endocardial EMT and NCC deployment to the OFT, resulting in OFT alignment and septation defects ¹⁴. Ablation of Fgf8 also leads to OFT alignment and septation defects ⁹. Gain of function and dominant negative mutations of SHP2 that directly bind to FRS2α and mediate FGF signaling to the MAPK pathway causes enlarged valves in Noonan Syndrome and LEOPARD syndrome patients, respectively ^{15, 16}.

In this report, we demonstrated that ablation of Frs2α, or double ablation of Fgfr1/Fgfr2, in heart progenitor cells led to enlarged OFT valves and bicuspid aortic valves (BAV). FGF regulated Bmp4 expression in the myocardium via AP1 transcription factor binding sites located upstream of the Bmp4 coding sequence. Disrupting FGF signaling diminished Bmp4 expression in the OFT myocardium and reduced smooth muscle (SM) differentiation of cushion NCCs, thus, leaving excessive undifferentiated NCC-derived mesenchymal cells within valve primordia. Treating in vitro cultured heart explants with BMP4 partially rescued the defects. The results demonstrate a novel role of NCC differentiation orchestrated by the FGF-BMP signaling axis during OFT valve formation.

Methods

All animals were housed in the Program for Animal Resources of the Institute of Biosciences and Technology, and were handled in accordance with the principles and procedures of the Guide for the Care and Use of Laboratory Animals. All experimental procedures were approved by the Institutional Animal Care and Use Committee.

Results

Disruption of FGF signaling leads to enlarged OFT valves

Disruption of FGF signaling in mouse second heart field (SHF) progenitors compromises OFT cushion formation and results in OFT alignment and septation defects ^{9, 14}. Since OFT cushions also give rise to both pulmonary and aortic valves, we then investigated whether ablating FGF signaling also affected OFT valve development. H&E staining of E14.5 embryo sections revealed that Fgfr1 and Fgfr2 double conditional ablations in cardiac progenitor cells with Nkx2.5^Cre (Fgfr1/r2^cn/Nkx) led to enlarged OFT valves (Fig. 1A). Individual ablation of Fgfr1 or Fgfr2 failed to cause such defects (data not shown), suggesting redundant activities of the two FGFRs in regulating OFT valve development. Since FRS2α is the major adaptor protein linking the FGFR to MAP kinase and PI3K/AKT pathways, we then ablated Frs2α alleles in the same domain with Nkx2.5^Cre (Frs2α^cn/Nkx) or in the SHF with Mef2C^Cre (Frs2α^cn/Mef). Similar to Fgfr1/r2^cn/Nkx, both Frs2α^cn/Nkx and Frs2α^cn/Mef embryos exhibited enlarged OFT valves (Fig. 1A). The phenotype remained obvious in neonatal hearts (Online Figure I). The average diameters of pulmonary and aortic valves were 70±10 μm and 62±11 μm in E14.5 control embryos and 103±26 μm and 99±29 μm in Frs2α^cn/Nkx embryos, respectively (Fig. 1B). About 20% of Frs2α^cn/Nkx embryos had BAVs (Fig. 1C). All valve defects were associated with OFT septation defects, indicating that OFT cushion remodeling was affected. No apparent defects were found in mitral valves and tricuspid valves (Fig. 1D). Cell fate mapping experiments showed that atrioventricular valves were only composed of endothelial lineage cells, while OFT valves had both endothelium- and NCC-derived cells (Online Figure. II), suggesting that the defects were likely associated with NCC-derived mesenchymal cells.

Figure 1. Disruption of the FGF signaling axis leads to OFT valve hyperplasia.

A, Transverse sections of E14.5 embryos were H&E stained to demonstrate enlarged OFT valves. Pulmonary and aortic valves are highlighted with dotted lines. Inserts are enlarged pictures of the valve. B, The embryos were serially sectioned, and the valve dimensions were measured in every five sections. Averages of three largest values were calculated from each embryo. Data represent an average dimension of at least 10 valves and are expressed as means ± standard deviation. C, Coronal sections of embryos were H&E stained demonstrating bicuspid aortic valves in E14.5 Frs2α^cn/Nkx embryos. D, Transverse sections of E14.5 embryos were H&E stained demonstrating atrioventricular valves. E, BrdU incorporation demonstrating compromised cell proliferation in Frs2α^cn/Nkx OFTs. Percentages of positive cells in the OFT myocardium and cushions from four embryos were calculated and expressed as means ± standard deviations. Fgfr1/r2^f/f, Fgfr1 and Fgfr2 double floxed embryos; Fgfr1/r2^cnNkx, Fgfr1/Fgfr2 double conditional ablation with Nkx2.5^Cre; Frs2α^flox, Frs2α floxed, Frs2α^cn/Nkx, Frs2α conditional ablation with Nkx2.5^Cre; Frs2α^cn/Mef, Frs2α conditional ablation with Mef2C^Cre; PV, pulmonary valve; AV, aortic valve; MV, mitral valve; TV, tricuspid valve; C, cushion; M, myocardium.

Defects in OFT cushion cell recruitment, proliferation, and apoptosis are not causal factors for enlarged valves in Frs2α^cn OFT

Both Frs2α^cn/Nkx and Frs2α^cn/Mef embryos had enlarged OFT valves (Fig. 1A). Yet, only Frs2α^cn/Nkx embryos show reduced endothelial and NCC contributions to OFT cushions and have small OFT cushions ¹⁴. Thus, the valve defect likely did not result from cushion insufficient recruitment of endothelium- and NCC-derived cells into OFT cushions. Since the enlarged Frs2α^cn OFT valves had more mesenchymal cells at E14.5 day, BrdU labeling was used to determine whether mutant cushions had increased cell proliferation activities. Both mesenchymal and myocardial cells in Frs2α^cn/Nkx OFT cushions actually had reduced proliferation at E10.5 and E12.5 (Fig. 1E), which was consistent with our previous report ¹⁴. This suggests that enlarged valves are not caused by proliferation defects.

To test whether the valve enlargement was caused by apoptosis defects, TUNEL assays were employed to assess cushion cell apoptosis at E13.5. Compared with control Frs2α^f/f OFT cushions, apoptotic cell numbers in Frs2α^cn/Nkx OFT cushions were significantly reduced (Fig. 2A, B). Consistently, Fgfr1/r2^cn/Nkx OFT cushions also exhibited reduced apoptosis at E13.5 (Fig. 2C), suggesting that the FGF might be involved in regulating OFT cushion cell apoptosis. Surprisingly, most apoptotic cells were not located at the valve primordia. Instead, they were located beneath the valve primordia, suggesting that enlarged valves were not directly caused by reduced apoptosis. Moreover, Frs2α^cn/Mef OFT that did not exhibit increased apoptosis in the cushion (Fig. 2D), nevertheless, still had enlarged OFT valves (Fig. 1A), further indicating that OFT valve hyperplasia was not caused by defective apoptosis. Consistent with previous findings in chicken models ¹⁷, cell lineage tracing revealed that only NCC-, but not endocardium-, derived cushion mesenchymal cells underwent apoptosis (Online Figure III).

Figure 2. Disruption of FGF signaling reduces OFT cushion cell apoptosis during remodeling.

A&B, Apoptosis in E13.5 OFT cushions. Representative sections from anterior to posterior sides were shown in A. Averages of total apoptotic cell numbers in all sections from three individuals were calculated and shown in B as means ± standard deviations. C. Apoptotic cells in Fgfr1/r2^Nkx, Frs2α^cn/Mef, and control embryos at E13.5.

Ablation of Frs2α causes defects in cushion myocardialization and smooth muscle differentiation, leaving excessive undifferentiated mesenchyme within the OFT valve primordia

Unlike wildtype cells that were round, Frs2α^cn/Nkx OFT valve cells were spindle-like (Fig. 1A, inserts), suggesting differentiation defects. We then investigated whether aberrant cushion cell differentiation caused the valve remodeling defects. During this stage, proximal OFT myocardial cells, which express SM α-actin (SMA) and sarcomeric myosin, invade the cushion and induce adjacent mesenchymal cells to undergo SM differentiation ^18–20. Immunostaining with anti-SMA antibodies revealed that at E12.5, the SMA positive myocardial cells protruded into proximal OFT cushions of wildtype embryos; some cells in the cushion also expressed low level SMA (Fig. 3Aa). In Frs2α^cn/Nkx embryos, however, the myocardial cells failed to extend into the cushion, and SMA expressing cushion mesenchymal cells was reduced (Fig. 3Ab), indicating defects in both myocardial cell invasion and mesenchymal cell differentiation. At E13.5, increased SMA positive cells appeared in control OFT cushions, which separated OFT cushions into two parts and, thus, defined the valve primordia (Fig. 3Ac). However, the SMA positive cells in Frs2α^cn/Nkx OFT cushions were reduced and failed to separate valve primordia from the rest of the cushion, leaving excessive mesenchymal cells within the valve (Fig. 3Ad). At E14.5, myocardialization and SM differentiation of the proximal outlet septum were almost completed in control cushions; the valve was sculpted, and the aorta was separated from the right ventricle by muscular tissues (Fig. 3Ae). In contrast, the outlet septum in Frs2α^cn/Nkx remained unmuscularized, and the valve still remained bulbous and swollen (Fig. 3Af). Consistently, similar defects were found in Fgfr1/r2^cn/Nkx and Frs2α^cn/Mef embryos (Fig. 3B&C), further confirming the role of FGF signaling in this process.

Figure 3. Ablation of the FGF signaling axis leads to defective myocardialization and SM differentiation in OFT cushions.

A-C, Transverse sections of embryos were immunostained with anti-SMA antibody demonstrating defects in myocardialization and SM differentiation. The specifically bound antibodies were visualized either with peroxidase activities (A) or fluorescent dyes (B&C). Arrows indicate migrating myocardial cells (My), and arrow heads indicate cushion mesenchymal cells (Cu) undergoing SM differentiation. Dotted lines outline the valve primordia. To-Pro3 (TP3) was used for nuclear counterstaining. D, Immunostaining with MF20 antibody revealed that the myocardial cells in control OFT were elongated and had long lamellipodia extending into the cushion mesenchyme as indicated by arrows (a); reduced lamellipodia were observed in Frs2α^cn/Nkx OFT myocardial cells (b). E, Myocardial cells isolated from E13.5 OFT migrating through the membranes were identified by immunostaining with MF20 antibody (c–d). The ratios of migrated MF-20 positive cells over total MF-20 positive cells were calculated from 3 individual samples and were expressed as means ± standard deviations (e). F, Real-time RT-PCR analyses of E12.5 OFTs. Data are normalized to GAPDH and expressed as folds of changes over wide type samples.

Immunostaining with anti-sarcomeric myosin MF20 antibody was carried out to further dissect whether the unmuscularized OFT cushions were caused by defective myocardialization or SM differentiation. In control OFT, the myocardial cells next to cushions were elongated and had lamellipodia extending into the cushion mesenchyme (Fig. 3Da). In contrast, Frs2α^cn/Nkx OFT myocardial cells were round and almost had no lamellipodia extending into the cushion (Fig. 3Db), suggesting compromised inward migration of OFT myocardial cells. To confirm this, cells from E13.5 proximal OFTs were isolated for transwell migration assays, and the myocardial cells that migrated through the membrane were identified by sarcomeric myosin expression. It was apparent that Frs2α^cn/Nkx OFTs had fewer cells migrating through the membrane than controls (Fig. 3E), indicating that FRS2α-mediated signals promoted myocardial cell migration in a cell autonomous manner. Focal adhesion kinase (FAK) and Vangl2 mediated noncanonical Wnt signaling pathways are involved in regulating OFT myocardium migration ^{18, 21}. However, expressions of FAK, Rock1, Vangl2, and Wnt11 were not changed in E12.5 Frs2α^cn/Nkx OFT (Fig. 3F), suggesting that the FGF regulated OFT myocardium migration through other pathways.

FRS2α-mediated signals are required for regulating SM differentiation of cushion mesenchymal cells in the OFT

The OFT cushion mesenchyme mainly consists of both endocardium- and NCC- derived cells. To determine how FRS2α-mediated signals regulated cushion cell differentiation and the origin of cushion muscular cells, cell fate mapping was carried out with ROSA26 reporter mice ²² and Tie2^cre driver that was expressed in the endocardium ²³ or Ap2α^cre that was expressed in NCCs ²⁴. Immunostaining revealed that most mesenchymal cells in proximal OFT cushions were labeled by Ap2α^cre, but not Tie2^cre, suggestive NCC-origin of the cells. These NCC-derived cells were SMA positive and MF20 negative (Fig. 4Aa&b), whereas endothelial-derived cells were both SMA and MF20 negative (Fig. 4Ac&d), indicating that NCC-, but not endothelial-, derived cells in the OFT cushions underwent SM differentiation. All MF-20 positive cells located at the junction of the OFT myocardium and cushion were β-galactosidase negative in both Tie2^cre-ROSA26 and Ap2α^cre-ROSA26 mice, suggesting that these cells were neither from the endothelium nor from NCCs, but from the myocardium (Fig. 4Ab&d). Double staining of β-gal and SMA of E12.5 Nkx2.5^Cre-Frs2α^flox/W-ROSA26 embryos revealed that all SMA positive OFT cushion mesenchymal cells were β-gal negative (Fig. 4Ba). Since Nkx^2.5Cre is expressed in both OFT myocardial and endocardial cells, the results confirmed that these SMA-positive mesenchymal cells were neither myocardial nor endocardial derived. Furthermore, since most proximal OFT cushion mesenchymal cells were NCC-origin, these β-gal negative and SMA positive cells were most likely derived from NCCs. This population of cells were severely reduced in Nkx2.5^Cre-Frs2α^f/f-ROSA26 OFT cushions (Fig. 4Bb), indicating that NCC differentiation was disrupted in mutant cushions. Furthermore, both β-gal and SMA negative cells in OFT valves were increased from 17.1%±5 in Nkx2.5^Cre-Frs2α^f/w-ROSA26 embryos to 26.7%±10 in Nkx2.5^Cre-Frs2α^f/f-ROSA26 embryos at E14.5 (Fig. 4C), suggesting that more undifferentiated NCCs were left in mutant than in control valve primordia. Since moderate caspase-3 activity is required for stem cell differentiation ^25–27, we subsequently investigated whether caspase-3 activation levels underlie cushion mesenchymal differentiation and apoptosis. Double immunostaining revealed that wildtype OFT cushions had many mesenchymal cells exhibiting moderate cleaved caspase-3 and SMA staining at E13.5 (Fig. 5Aa, green arrows). Frs2α^cn/Nkx OFT cushions had reduced caspase-3 positive (strongly and moderately) and SMA positive cells (Fig. 5A); Frs2α^cn/Mef OFT cushions only had reduced moderate caspase-3 and SMA double positive cells (Fig. 5B), suggesting that apoptotic and differentiated OFT cushion cells are different subpopulations. To confirm that caspase-3 activity played a key role in cushion cell differentiation, ex vivo cultured E12.5 embryonic hearts were treated with H₂O₂ to induce caspase-3 activation. As expected, H₂O₂ treatment induced caspase-3 activation at both high and moderate levels and SMA expression (Fig. 5C). More importantly, increased SMA expression was found in cells exhibiting moderate caspase-3 activities. Real-time RT-PCR analyses also revealed that SMA expression was increased by H₂O₂ treatments (Fig. 5D). Consistently, inhibition of caspase-3 significantly suppressed expression of SMA in NCCs isolated from pharyngeal arches 1 and 2 of E10.5 embryos, as revealed by both immunostaining and real-time RT-PCR analyses (Fig. 5E). The results further demonstrated that appearance of moderate caspase-3 activity induced OFT cushion cells to differentiate.

Figure 4. Ablation of Frs2α in heart progenitor cells compromises NCC-derived cushion cells undergoing SM differentiation.

A, Transverse sections of embryos were immunostained with anti-β galactosidase and anti-SMA or MF20 antibodies as specified. B&C, Double immunostaining of transverse embryo sections revealed that SMA expression in β-gal negative NCCs was reduced in Frs2α^cn/Nkx OFT cushions (B) and valves (C). SMA, SM α-actin. Percent of β-gal negative cells was scored from serial sections of Frs2α^cn/Nkx and control OFT valves and was expressed as means ± standard variations.

Figure 5. Reduced moderate caspase-3 activation in NCC-derived cushion mesenchymal cells in the Frs2α^cn/Nkx OFT.

A-C, Transverse sections of embryos were double immunostained with anti-SMA and anti-caspase-3 (cleaved) antibodies. Samples in panel C were overnight-cultured E12.5 hearts after being treated with 5 μM H₂O₂ for 30 min. Cells with SMA and moderate levels of caspase-3 (can only be detected with 40X or 63X objectives) are indicated with green arrows, and those with strong caspase-3 activation (can be detected with 10X objectives) were indicated with yellow arrows; Cas3, caspase-3. D, E12.5 hearts were treated with H₂O₂ at indicated concentrations for 30 min and then cultured overnight. Total RNA was isolated from OFTs, and SMA expression was analyzed with real-time RT-PCR. Data were normalized with GAPDH, calculated from three replicate samples, and expressed as folds of increase. E, NCCs isolated from pharyngeal arch 1 and 2 of E10.5 embryos were treated with caspase-3 inhibitor (I) or DMSO solvent (D) for 24 hours. Expression of SMA was analyzed with immunostaining with anti-SMA antibody (a-c), or with real-time RT-PCR. The percent of SMA positive cells was calculated from 5 randomly selected views. Data are means ± standard deviations of three replicate samples.

The BMP4 pathway mediates FGF signals in regulating OFT cushion cell differentiation

Since both Nkx2.5^cre and Mef2c^cre were not expressed in NCCs, FRS2α-mediated signals in regulating NCC-derived cushion mesenchymal differentiation must be non-cell autonomous and mediated by other signaling pathways. Furthermore, ablation of Frs2α in endothelial cells with Tie2^cre did not cause apparent defects ¹⁴, therefore, the mediating signals likely were derived from the OFT myocardium. BMP4 has been shown to control NCC apoptosis ²⁸; ablation of Bmp4 also leads to valve hyperplasia ²⁹. Since Bmp4 expression in the SHF is regulated by FRS2α-mediated signaling pathways ¹⁴, we then tested whether BMP4 was the downstream pathway that mediated FGF signals in regulating OFT cushion remodeling. Wholemount in situ hybridization demonstrated that Bmp4 expression was reduced in the Frs2α^cn/Nkx OFT myocardium at E12.5 (Fig. 6A). Consistently, phosphorylation of Smad1/5/8, downstream targets of the BMP receptor, was dramatically reduced in Frs2α^cn/Nkx cushion mesenchymal cells (Fig. 6A). Ablation of Bmp4 with Mef2c^cre also reduced apoptosis, expression of SMA, and activation of caspase-3 in OFT cushion mesenchymal cells (Fig. 6B). Together, it is suggested that BMP4 is the downstream mediator of FGF for regulating OFT cushion remodeling. To test this possibility, E12.5 hearts were cultured in the presence or absence of BMP4 overnight. Both immunostaining and real-time RT-PCR analyses revealed that BMP4 increased SMA expression in Frs2α^cn/Nkx OFT (Fig. 6C&D) suggesting that BMP4 treatment rescued the defects. Ablation of one Bmp4 allele in Frs2α^cn/Mef embryos increased the penetrance of valve hyperplasia from 38% to 58% (Fig. 6E), further suggesting that BMP4 mediated FGF signaling in regulating OFT cushion remodeling.

Figure 6. FRS2α-mediated signals regulate OFT cushion remodeling via BMP4.

A, Whole-mount in-situ hybridization with E12.5 embryos demonstrates reduced Bmp4 expression in Frs2α^cn/Nkx OFT (panels a and b). Red arrows indicate Bmp4 expression. Panels c and d, transverse sections of E12.5 embryos were immunostained with antibody against phosphorylated Smad1/5/8. B, Apoptotic cells in transverse sections of E13.5 embryos were detected with TUNEL assays (a, b). Panels c and d, double immunostaining of transverse sections of E13.5 embryos with anti-SMA and anti-caspase-3 antibodies. Green arrows indicate cells with moderate caspase-3 activation and SMA expressions, and yellow arrow indicates strong caspase-3 activation. C, Embryonic hearts collected at E12.5 were cultured overnight in the presence or absence of 10 ng/ml of BMP4. Paraffin sections of the hearts were then immunostained with anti-SMA antibodies. D, Real-time RT-PCR analyses of SMA expression in ex vivo cultured OFTs. Data were normalized to GAPDH and expressed as folds of changes over untreated samples. E, Ablation of one Bmp4 allele increases the penetrance of OFT valve defects in Frs2α^cn/Mef embryos.

To determine whether Bmp4 expression was regulated by the FGF, in vitro cultured hearts were treated with FGF2, and the Bmp4 expression was assessed with real-time RT-PCR. The results demonstrated that treating the heart with FGF2 significantly increased Bmp4 expression (Fig. 7A). As a main downstream pathway of FRS2α, the MAP kinase cascade activates AP1 transcription activity by inducing c-Jun and c-Fos phosphorylation. Immunostaining revealed that c-Jun nuclear localization was reduced in the Frs2α^cn/Nkx OFT myocardium (Fig. 7B), indicating compromised c-Jun activation. The upstream non-coding sequence of the Bmp4 allele has two candidate AP1 binding sites that potentially can be regulated by FGF signals. Thus, the 3.6 Kb upstream non-coding sequence of the Bmp4 allele was cloned into a luciferase reporter plasmid. Luciferase activity analyses revealed that FGF2 promoted expression of the reporter; deletion of the AP1 binding sites diminished the response (Fig. 7C). Furthermore, in vivo ChIP assays demonstrated that interaction of c-Jun and the proximal, but not the distal, AP1 binding site was reduced in mutant OFTs (Fig. 7D). Together, the data suggest that the FGF signaling axis in the OFT myocardium regulates Bmp4 expression via activation of AP1 to control cushion NCC differentiation during OFT cushion remodeling.

Figure 7. The FGF signaling axis regulates Bmp4 expression via activating the AP1 transcription activity.

A, E12.5 heart explants were cultured in the presence or absence of 10 ng/ml FGF2 overnight, and Bmp4 expression was analyzed with real-time RT-PCR. Data were normalized to GAPDH and expressed as folds of increase over untreated samples. B, Transverse sections of E12.5 embryos were immunostained with anti-c-JUN antibody, showing reduced c-JUN nuclear localization in the OFT myocardium of Frs2α^cn/Nkx embryos. Inserts are high-magnification views of the boxed areas. OFT cushions are highlighted with dotted lines. C, AP1-binding sites in the Bmp4 allele were required for FGF2 stimulation. Panel a, luciferase reporter constructs. Panel b, luciferase reporter transfected C2C12 cells were incubated overnight in the presence or absence of 10 ng/ml FGF2, and the luciferase activity was analyzed. Data are means and standard deviation from 3 replicate wells. ΔAP, AP1 site deleted mutant. D, E12.5 hearts were subjected to ChIP analyses with anti-C-Jun antibody. Pulled-down DNA fragments were analyzed with real-time PCR and expressed as folds of changes over untreated samples.

Discussion

In this report, we demonstrated that FRS2α-mediated signaling in the myocardium regulated OFT cushion remodeling and defined valve primordia via BMP4. Ablating this signaling axis disrupted cushion myocardialization and SM differentiation, and caused valve enlargement. The results demonstrate a novel FGF-BMP signaling axis from the OFT myocardium to the cushion mesenchyme regulating OFT cushion NCC differentiation, which plays a key role in OFT valve primordium formation. The findings are rather surprising since previous studies show that ablation of FGF signaling leads to small OFT cushions, which give rise to the OFT valves and septum ^{9, 14}.

We further demonstrated that the FGF promoted NCC differentiation non-cell autonomously via BMP4. The SM differentiation of cushion mesenchymal cells separated them into two populations: one underwent SM differentiation and contributed to OFT septation; the other did not undergo muscle differentiation and contributed to OFT valves (Fig. 8). At E14.5, only a few undifferentiated NCCs remained in WT valves. In mutant mice, however, the disrupted SM differentiation left many undifferentiated NCC-derived mesenchymal cells in the cushions and caused enlarged valves, even though total cushion mesenchymal cells were reduced (Fig. 8). In some severe cases that had OFT septation defects, the aortic and pulmonary valve primordia were not separated, resulting in a large fused valve.

Figure 8. A working model of FGF signaling in OFT valve formation.

Defective cushion NCC differentiation fails to define valve primordia and leaves excessive cells within valve primordia, which results in a large valve. Valve primordia are outlined with red lines.

Although it has been reported that mature OFT valve cells are largely derived from the endocardium (de Lange et al., 2004; Jiang et al., 2000), the part of OFT cushions that gave rise to valves and the underlying region were mainly comprised of NCC-derived mesenchymal cells at E13.5 (Fig. S2). These NCC lineage cells then either underwent apoptosis or SM differentiation (Figs. 4&5), leaving mainly endothelial lineage cells in OFT valves. The data are consistent with previous findings that no endocardial lineage cells in OFT valves undergo apoptosis and that the endocardial lineage replaces NCCs during late stages of valve formation ³⁰. Frs2α^cn/Nkx OFT valves had significantly increased non-Nkx2.5 lineage cells (Fig. 4C), suggesting that more NCC-lineage contributed to mutant OFT valves. Unlike OFT cushions, AV cushions only had the endocardium lineage mesenchymal cells (Fig. S2). This likely explains why AV valves were not affected by ablation of FGF signaling.

Mutations in SM α-actin are associated with familial BAV defects ³¹, implying that defective SM differentiation of NCCs may contribute to BAV. However, although NCC differentiation was disrupted in both Frs2α^cn/Nkx and Frs2α^cn/Mef OFT cushions, only Frs2α^cn/Nkx embryos exhibited BAV defects, indicating that BAV defects in Frs2α mutants were not caused by disrupted NCC differentiation. Since cushion mesenchymal apoptosis was disrupted in Frs2α^cn/Nkx, but not Frs2α^cn/Mef, OFT cushions, it is possible that reduced apoptosis contributed to failure in separating the valve leaflets from each other.

FRS2α has four GRB2 and two SHP2 binding sites. SHP2 is a ubiquitously expressed protein tyrosine phosphatase that plays a central role in the MAP kinase and other signaling pathways ^{32, 33}. Gain-of-function of SHP2 enhances MAPK activation and promotes the EMT and mesenchymal cell proliferation; dominant negative SHP2 suppresses MAPK activation and fails to arrest the proliferation of mesenchymal cells at later stages ³⁴. Thus, a narrow range of MAPK activity is required for embryonic development. Ablation of Frs2α in heart progenitor cells reduces MAPK activation in the myocardium. Here, we demonstrated that deletion of Frs2α led to reduced AP1 transcription activity, resulting in reduced Bmp4 expression in the myocardium, inhibited cushion NCC differentiation, and caused valve hyperplasia. The results reveal a novel mechanism that FGF signals in the OFT myocardium negatively regulate the valve size through promotion of NCC differentiation.

The regulatory mechanism underlying SM differentiation of OFT cushion NCCs remains poorly understood. It has been shown that transient and moderate caspase-3 activities promote stem cell differentiation ²⁵. Consistently, only those cells with moderate caspase-3 activity in OFT cushions undergo SM differentiation. Ablation of Frs2α compromised BMP4 signaling in OFT cushion cells and reduced the population of cells with SMA expression and moderate caspase-3 activity. Therefore, the results suggest that BMP4 is an upstream regulator for caspase-3 activation in OFT cushion NCCs. It remains unanswered why these cushion mesenchymal cells have different caspase-3 activities. One possibility is that differentiated NCCs are insensitive to BMP4 stimulation, and undifferentiated NCCs are BMP sensitive and will undergo apoptosis when they receive sustained BMP4 stimulation. However, further experimentation is needed to test this hypothesis.

In conclusion, we demonstrated a novel FGF-BMP signaling axis between the OFT myocardium and cushion mesenchyme that regulated OFT cushion myocardialization and SM differentiation during cushion remodeling and that the two processes play an important role in defining OFT valve primordia. The results shed new light on understanding how perturbation of cell signaling and NCC differentiation lead to congenital heart valve defects.

Abbreviations footnote

BMP

bone morphogenetic proteins

EMT

endothelial-mesenchymal transition

FGF

fibroblast growth factor

FGFR

FGF receptor

FRS2α

FGFR substrate 2α

FHF

first heart field

NCCs

neural crest cells

OFT

outflow tract

SHF

second heart field

SM

smooth muscle

SMA

smooth muscle actin