Foxc2 is required for proper cardiac neural crest cell migration, outflow tract septation, and ventricle expansion

Abstract

Background:

Proper development of the great vessels of the heart, and septation of the cardiac outflow tract requires cardiac neural crest cells. These cells give rise to the parasympathetic cardiac ganglia, the smooth muscle layer of the great vessels, some cardiomyocytes and the conotruncal cushions and aorticopulmonary septum of the outflow tract. Ablation of cardiac neural crest cells results in defective patterning of each of these structures. Previous studies have shown that targeted deletion of the forkhead transcription factor C2 (Foxc2), results in cardiac phenotypes similar to that derived from cardiac neural crest cell ablation.

Results:

We report that Foxc2^−/− embryos on the 129s6/SvEv inbred genetic background display persistent truncus arteriosus and hypoplastic ventricles prior to embryonic lethality. Foxc2 loss-of-function resulted in perturbed cardiac neural crest cell migration and their reduced contribution to the outflow tract as evidenced by lineage tracing analyses together with perturbed expression of the neural crest cell markers Sox10 and Crabp1. Foxc2 loss-of-function also resulted in alterations in PlexinD1, Twist1, PECAM1 and Hand1/2 expression in association with vascular and ventricular defects.

Conclusions:

Our data indicate Foxc2 is required for proper migration of cardiac neural crest cells, septation of the outflow tract, and development of the ventricles.

Introduction

Congenital cardiovascular defects, which occur in 1% of live births and up to 10% of stillbirths, are characterized primarily by abnormal development of the cardiac chambers and malformations of the large blood vessels that constitute the inflow and outflow tract (1, 2). Among the most common cardiac malformations is abnormal patterning and septation of the outflow tract, which culminates in inappropriate mixing of oxygenated and deoxygenated blood (2, 3). These phenotypes are often associated with defects in cardiac neural crest cell migration and/or differentiation, which can affect their contribution to aortic arch artery development and to the formation of the aorticopulmonary septum (4, 5).

Cardiac neural crest cells migrate through the post-otic mesenchyme, and pharyngeal arches 3–6, before entering the outflow tract where they populate the mesenchyme lining the outflow tract endothelium. Subsequently, cells from the endothelium undergo an endothelial to mesenchymal transformation, augmenting the neural crest cell derived mesenchyme, which together form the conotruncal cushions. Appropriate numbers of neural crest cells (6–8) together with cues from the endothelium (9–11) coordinately interact interact to separate the single outflow tract into the aorta and pulmonary trunk through formation of the aorticopulmonary septum.

Cardiac neural crest cells ultimately differentiate into smooth muscle cells, building the walls of these large vessels, as well as the semilunar valves, that prevent backflow of blood into the cardiac chambers (5, 12, 13). In a similar manner to the heart, cardiac neural crest cells and endothelium also coordinately pattern the pharyngeal arch arteries. Portions of the third through sixth pharyngeal arch arteries contribute to the aortic arch as well as the brachiocephalic trunk, common carotid, and subclavian arteries as they emerge from the aortic arch. Failure to properly maintain or degenerate the pharyngeal arch arteries leads to defects in the aortic arch (14, 15).

The forkhead box transcription factor C2 (Foxc2) is expressed during embryonic development in the presomitic and paraxial mesoderm (16, 17), in the endothelium of the pharyngeal arch arteries, outflow tract and ventricles, and in cardiac neural crest cells that colonize the pharyngeal arches and outflow tract (14, 18, 19). More recently, our lineage tracing analyses of Foxc2 expressing cardiac neural crest cells revealed that these cells additionally contribute to the aorta, pulmonary trunk, valves and endocardial cushions (20). Furthermore, studies in mice have indicated that Foxc2 is expressed in, and plays a critical role in proper development of lymphatic vessels and valves (19, 21–23), and consistent with these observations, heterozygous mutations in FOXC2 are associated with congenital lymphedema distichiasis in humans (24).

Two independent laboratories have created targeted deletions of Foxc2 in mice (14, 16), and in both cases, homozygous mutant mouse embryos displayed severe cardiac phenotypes, primarily Type B interruption of the aortic arch – a defect associated with aberrant patterning of the fourth pharyngeal arch arteries. However, the overall phenotypes were quite variable, likely due to the presence of modifying genetic factors on the different genetic backgrounds of mice used. On a C57BL/6 background, approximately 50% of Foxc2^−/− embryos die at embryonic day (E)12.5 while the remainder die perinatally (14). In contrast, on a mixed 129 X Black Swiss background, 95% of Foxc2^−/− embryos die between E11.5-E15.5, with the remaining small portion dying perinatally (16, 18).

Previous analyses of the role of Foxc2 in cardiovascular development concentrated on the late gestation and perinatal phenotypes. In this study, we therefore describe and analyze the predominant mid-gestation cardiac phenotype resulting from targeted deletion of Foxc2 (16) on a 129s6/SvEv genetic background. We consistently observed a phenotype in which Foxc2^−/− embryos died between E12.5 – E13.0 with persistent truncus arteriosus (common arterial trunk) and ventricle hypoplasia. Subsequent fate mapping and gene expression analyses indicate that Foxc2 plays important roles in regulating proper cardiac neural crest cell migration, outflow tract septation, and ventricle development.

Results and Discussion

Expression of Foxc2 during cardiac neural crest cell migration

Foxc2 is expressed in the cranial and paraxial mesoderm, and in the endothelium of the pharyngeal arch arteries, aorta, and pulmonary trunk (14, 18, 19). Foxc2 expression has also been reported in the neural crest derived mesenchyme of the pharyngeal arches at E10.5 (18). Consistent with these spatiotemporal domains of activity, Foxc2 targeted deletion mutants exhibit pharyngeal arch artery abnormalities typically associated with ablation of cardiac neural crest cells (14, 18). Consequently, we characterized the domains of Foxc2 expression in E8.5–10.5 wild-type embryos during the migration of cardiac neural crest cells and examined the impact this has on cardiac development and function.

At E8.5, Foxc2 is expressed throughout the cranial mesenchyme, in a region adjacent to the posterior hindbrain as well as in the somites and pre-somitic mesoderm (Figure 1A). From E9.0 to E9.5, Foxc2 continues to be expressed in the cranial, somitic and pre-somitic mesoderm, and begins to be expressed in the proximal mesenchyme of pharyngeal arches two and three as well as in the nascent pharyngeal arch 4/6 region (Figure 1B, 1C). At E10.5, Foxc2 expression can be detected in all pharyngeal arches, in the head and tail mesenchyme, in the left and right dorsal aorta, in the common atrial chamber, and in the outflow tract (Figure 1D–1E). These expression domains are associated with mesenchymal cells derived from the neural crest and mesoderm, prompting us to analyze the activity of Foxc2 in the neural crest population more closely. In E10.5 Wnt1-Cre;YFP embryos, which indelibly label neural crest cells, we observed Foxc2⁺ neural crest cells within the aortic arch arteries (Figure 1F), in the aortic sac (Figure 1J) and in the outflow tract (Figure 1N). Taken together with data from previous studies, our results indicate that Foxc2 is expressed in both the neural crest derived cells of pharyngeal arches 2–6 and the endothelium of the pharyngeal arch arteries. Furthermore, our recent lineage tracing analyses demonstrate that these Foxc2 positive progenitor cells provide a considerable source of mesenchymal and endothelial cells to the aorta, pulmonary trunk, valves and endocardial cushions during cardiac morphogenesis (20).

Figure 1.

Spatiotemporal expression of Foxc2. Foxc2 is expressed during mesenchyme development and colocalizes within the neural crest in the pharyngeal arch arteries and cardiac outflow tract. (A) At E8.5 Foxc2 is expressed in the cranial mesenchyme and developing somites. (B, C) From E9.0 to E9.5 Foxc2 is expressed in the mesenchyme of pharyngeal arches 2 to 6 and in the somites. (D, E) At E10.5 Foxc2 is expressed in the dorsal aorta, the pharyngeal arches, and the cardiac outflow tract. (F) Whole mount co-immunolocalization of Foxc2 in Wnt1-Cre;YFP embryos at E10.5 shows neural crest cells colonizing the aortic arch arteries and cardiac outflow tract (red arrow in F). Foxc2 positive neural crest cells can be observed invading the aortic sac (red arrow in J) and reaching the outflow tract (red arrow in N). Scale bars represents 400μm (A), 50μm (B, C) 80μm in (D, E) and 70μm (F-Q).

Ablation of cardiac neural crest cells is associated with abnormal patterning and arrangement of the great arteries of the aortic arch and disruption of aorticopulmonary septation (7, 8). However, these same defects may also be caused by deficiencies within the endothelium of the pharyngeal arch arteries and the developing outflow tract (9, 10). Therefore, the observed defects in Foxc2 mutant embryos may be a result of a requirement for Foxc2 signaling in either cell population.

The predominant perinatal phenotype resulting from the targeted deletion of Foxc2 is that of Type B interruption of the aortic arch (14, 18). However, this type of aortic arch artery defect cannot account for the mid-gestation lethality also reported in these studies. On the 129S6/SvEv genetic background used in this study, we were unable to recover live Foxc2^−/− embryos beyond E13.5. Interestingly, Foxc2^−/− mutant embryos were viable at E12.5 (Figure 2A, E), had steady blood flow through the heart and umbilical cord, without any evidence for hemorrhaging. In contrast, E13.5 Foxc2^−/− mutant embryos exhibited hearts that were not beating, and an absence of pulsing blood flow in the vasculature of the yolk sac compared to controls (Figure 2B-D, F-H). This indicated that on the 129S6/SvEv background, Foxc2^−/− embryos die between E12.5–13.5.

Figure 2.

Wild type and Foxc2^−/− embryo phenotype. E12.5 wild type (A) and Foxc2^−/− mutant embryos (E) appear grossly morphologically normal, however by E13.5 blood does not flow through the yolk sac (F) or umbilical vasculature (arrows in G and H) in Foxc2^−/− mutants, compared to controls (B, arrows in C and D). Scale bars in: A, E=2mm, B, C, D, F, G, H=2.5mm.

To better characterize the pathogenesis of the Foxc2^−/− phenotype on the 129S6/SvEv genetic background, we initially examined the gross morphological appearance of E9.5–10.5 embryos. In E9.5 wild type embryos, the pharyngeal arches appear as robust, distinct structures separated by clear pharyngeal clefts (Figure 3A). In contrast, in Foxc2^−/− mutant embryos, the second and third pharyngeal arches were smaller compared to wild type controls (Figure 3A, D). This phenotypic difference became more pronounced by E10.5 (Figure 3B, C, E, F) whereby Foxc2 mutant embryos displayed clearly hypoplastic pharyngeal arches, with pharyngeal arches 3 through 6 most severely affected (Figure 3E, F). Furthermore, the second and third pharyngeal clefts were markedly smaller and misshapen resulting in less distinct separation between these posterior pharyngeal arches in Foxc2 mutant embryos compared to wild type controls (Figure 3C, F).

Figure 3.

Hypoplastic pharyngeal arch development in Foxc2 mutant embryos. Images of wild type (A-C) and Foxc2^−/− (D-F) embryos at E9.5 and E10.5. As early as E9.5, pharyngeal arch 2 in Foxc2^−/− embryos is smaller than controls (arrows in A and D). Low (B, E) and high (C, F) magnification views of E10.5 wild type and Foxc2^−/− embryos reveal hypoplastic pharyngeal arches 2 through 6 (black arrows) and pharyngeal pouches 2 and 3 in the mutant. Scale bars in A, C, D, F=150 μm; B, E=500 μm.

Aorticopulmonary septum formation and ventricular development are impaired in absence of Foxc2

Altered pharyngeal arch morphology may lead to defects in development of the pharyngeal arch arteries or the outflow tract. Therefore, we examined E12.5 embryos for the presence of gross morphological cardiac anomalies. Wild type embryos exhibited an aorta and pulmonary trunk that had separated from within the outflow tract (Figure 4A, C, E, G). At this stage, the right and left ventricles were clearly separated by the interventricular septum, and semilunar valve formation had also occurred (arrows in Figure 4E). In contrast, each of the Foxc2^−/− embryos examined at E12.5, exhibited an outflow tract that persisted as a single truncus arteriosus (Figure 4B, D, F, H). Furthermore, the muscular portion of the interventricular septum was not distinct, and the mutant ventricles were composed of wispy trabeculae. In addition, the ventricles were hypoplastic compared to wild type controls. Histological sections revealed that the wall of the persistent truncus arteriosus, and the ventricular myocardium in Foxc2^−/− embryos were thin (Figure 4F, H) compared to the same structures in controls (Figure 4E, G,).

Figure 4.

Persistent truncus arteriosus and hypoplastic ventricles in the absence of Foxc2. Wild type hearts (A, C) display septation of the outflow tract into the aorta (Ao) and pulmonary trunk (Pt). Mutant hearts (B, D) present with a persistent truncus arteriosus (PTA) and hypoplastic ventricles. In hematoxylin and eosin stained coronal sections wild type hearts exhibit a clear aorta, pulmonary trunk and semilunar valves (black arrows in E, G). The interventricular septum (*) is well-developed and separates the right and left ventricles. Coronal and transversal sections of mutant hearts (F, H) reveal persistent truncus arteriosus of the outflow tract with an indistinct interventricular septum (*). Black bars in E and F indicate the thickness of the ventricular myocardium. Abbreviations: LV, left ventricle; RV, right ventricle. All scale bars represent 200μm.

Failure of distal outflow tract septation is associated with failure of semilunar valve formation and function (25, 26). These valves are required to prevent retrograde flow of blood through the circulatory system during embryonic development in the mouse. Retrograde flow of blood due to hypoplastic endocardial cushions, persistent truncus arteriosus, and semilunar valve failure typically precedes embryonic lethality by approximately 24 hours. In turn, this retrograde flow is associated with thinning of the ventricular myocardium, as seen in Foxc2^−/− embryos (Figure 4F, H). Such a mechanism for cardiac failure could explain the embryonic lethality of Foxc2^−/− embryos at E13.5.

Abnormal cardiac neural crest cell migration and contribution to the developing outflow tract

Defects in outflow tract septation may be due to perturbation of the migration and/or differentiation of cardiac neural crest cells, or due to a disruption in development of the outflow tract endothelium with which cardiac neural crest cells interact. Foxc2 is expressed in migrating cardiac neural crest cells, and in the outflow tract endothelium (Figure 1; (20)). Therefore, we initially examined whether Foxc2 loss-of-function disrupted cardiac neural crest cell migration in association with the observed cardiac phenotypes. Whole mount in situ hybridization with markers of migrating neural crest cells such as Sox10, revealed a dense, population of cardiac neural crest cells caudal to the otic vesicle in E9.0 Foxc2^−/− embryos compared to controls (brackets in Figure 5A, B). In addition, the distal portion of neural crest cells within the third pharyngeal arch stream migrated abnormally through the typically neural crest-free zone between the second and third pharyngeal arches, towards the second pharyngeal arch (Figure 5A, B). This suggests that Foxc2 could play a direct role in proper neural crest cell migration and may be associated with the less distinct separation of caudal pharyngeal arches in Foxc2 mutants compared to controls (Figure 3). Despite these abnormal patterns of neural crest cell migration at E9.0, no alterations in the patterning of cranial ganglia were observed at E10.5 as evidenced by a normal segmental pattern of Sox10 expression and staining of cranial nerve ganglia (Figure 5C, D).

Figure 5.

Abnormal neural crest cell migration in Foxc2^−/− embryos. Sox10 whole mount in situ hybridization in E9.0 and E10.5 wild type (A, C) and Foxc2^−/− (B, D) embryos. Mutant neural crest cells enter the space between the pharyngeal arch 2 and 3 streams (red arrowheads in A, B). Cardiac neural crest cells migrating aberrantly in the post-otic region (white brackets in A, B) are more densely distributed in Foxc2^−/− mutant embryos compared to controls. However, no abnormalities were noted in the pattern of Sox10 expression at E10.5 (C, D) which marks glial cells supporting cranial ganglia. (E, F) Crabp1 is expressed in cardiac neural crest cells migrating through pharyngeal arches (PA) 3–6 (demarcated by red and yellow hashed lines). Green arrowheads are positioned at the dorsal edge of staining in PA3–6. White brackets in E and F demarcate the neural crest cell free zone between the second and third pharyngeal arch streams. (E’, F’) Schematic drawings of the Crabp1 positive neural crest cell population (green shading) with anatomical landmarks. Note the reduced neural crest cell-free zone and the separation between the domain of Crabp1 labeled cells and the developing atrium (A). Abbreviations: A, aorta; OFT, outflow tract; OV, otic vesicle; PA1, pharyngeal arch 1. All scale bars = 100μm.

Cardiac neural crest cells express cellular retinoic acid binding protein during their migration through the pharyngeal arches and into the cardiac outflow tract (27, 28). Although Crabp1 expression was detected in Foxc2^−/− embryos at E9.5, some changes were noted compared to control embryos at this stage (Figure 5E, F). Similar to the results of Sox10 staining, Crabp1 positive cells invaded the normally neural crest free region between the second and third arch streams in Foxc2^−/− embryos (brackets Figure 5E, F). Additionally, in Foxc2^−/− embryos, Crabp1 expressing neural crest cells exhibited diminished or delayed migration through pharyngeal arches 3–6, compared to wild type embryos. This is more clearly recognized in the schematic drawings of CrabpI staining (Figure 5E’ and F’). Note the differences in the shape and size of the green shaded regions and their decreased proximity to the developing atria.

Diminished cardiac neural crest cell contribution to Foxc2^−/− outflow tract

In order to characterize the long-term fate of cardiac neural crest cells in Foxc2^−/− embryos, and better understand any defects in their contribution that may underpin the pathogenesis of the cardiac phenotype, we generated Foxc2^−/−;Wnt1Cre;R26R mice to enable lineage tracing of neural crest cells (Figure 6A). In E9.0 Foxc2 mutants, the distribution of lacZ positive neural crest cells (Figure 6B) in the post-otic region, resembled that observed via Sox10 and Crabp1 in situ hybridization. To further validate alterations in the distribution of cardiac neural crest cells, we generated Foxc2^−/−;6.5Pax3Cre;26R mice (29). The 6.5Pax3Cre line labels cranial neural crest cells populating the second pharyngeal arch and mosaically labels cardiac neural crest cells populating the more caudal (III-VI) pharyngeal regions (Figure 6C, D). Wild-type 6.5Pax3Cre;R26R embryos exhibit a noticeably larger population of lacZ positive cardiac neural crest cells in pharyngeal arches 3–6 than in Foxc2^−/−;6.5Pax3Cre;R26R embryos. In each of the Foxc2^−/−;6.5Pax3Cre embryos analyzed, pharyngeal arch three displayed the most marked reduction in labeled cardiac neural crest cells compared to controls.

Figure 6.

Lineage tracing of neural crest cells indicates reduced contribution to outflow tract. (A, B) At E9.0, labeled neural crest cells in the post-otic region (white brackets) invade the typically neural crest free mesenchyme between the second and third pharyngeal arch streams (red arrowheads). In the bracketed region, labeled neural crest cells are more densely distributed. (C, D) Mosaic labeling of neural crest cells in pharyngeal arches 2–6 using the 6.5Pax3Cre;R26R mouse line, reveals a notable decrease in labeled neural crest cells in pharyngeal arches 3–6 of E9.5 Foxc2^−/− embryos compared to controls (black arrows). (E – H) In histological sections of E11.5 embryos, labeled cardiac neural crest cells have populated the outflow tract in both wild type and mutant. However, the failure of aorticopulmonary septum formation results in persistent truncus arteriosus (PTA) in Foxc2^−/− hearts (black arrows in F, H). Abbreviations: 3, 4/6 pharyngeal arch; OFT, outflow tract. All scale bars = 100μm.

By E11.5 cardiac neural crest cells populated the full length of the outflow tract in both wild type and mutant embryos (Figure 6E-H). However, there was a considerable overall reduction in lacZ staining in Foxc2^−/− mutants compared to controls which was indicative of a reduced population of cardiac neural crest cells in the outflow tract of Foxc2^−/− embryos. Furthermore, in the outflow tract of E11.5 embryos, the central portions of the endocardial ridges were fused, and the future lumens of the aorta and pulmonary trunk were visible (Figure 6E, G). In contrast, the outflow tract in Foxc2^−/− mutants maintained an endocardial layer between the two distinct endocardial cushion ridges, such that only one lumen was visible, indicating a delay or failure of ridge fusion to create the aorticopulmonary septum (arrows in Figure 6F, H). While our data indicate that cardiac neural crest cells arrive in the outflow tract, their numbers are reduced compared to controls, and aorticopulmonary septation fails to occur.

Our data suggests that in the absence of Foxc2, some cardiac neural crest cells migrate aberrantly through the post-otic mesenchyme and are delayed in their progression through the pharyngeal arch mesenchyme into the outflow tract. In comparison, we found no evidence for significant alterations in proliferation, cell division or apoptosis in E8.5–10.5 Foxc2^−/− embryos compared to controls. Thus, abnormal pharyngeal morphology, coupled with altered migration and reduced contribution of cardiac neural crest cells to the outflow tract, likely accounts for the failure of outflow tract septation. This model is consistent with previous studies indicating that cardiac neural crest cells must be present in sufficient numbers within the outflow tract cushions for normal aorticopulmonary septation to occur (6, 8).

Disruption of cardiac neural crest cell guidance and differentiation signals in the absence of Foxc2

Proper migration of neural crest cells has previously been shown to require both attractive and repulsive guidance cues. For example, in chick embryos, attractive cues in the cardiac outflow tract mediated by Semaphorin 3C, guide cardiac neural crest cells that express PlexinD1(Plxnd1) and Neuropilin1 (Nrp1)(30, 31). In mouse embryos, Plxnd1 is required in vascular endothelial cells for proper patterning of the outflow tract (9, 10). Furthermore Plxnd1^fx/fx;Tie2Cre embryos and human mutations in PLXND1 (32) are associated with the pathogenesis of persistent truncus arteriosus. Therefore, we compared the expression of Plxnd1 between wild type and Foxc2^−/− embryos (Figure 7A-D).

Figure 7.

PlexinD1 expression is disrupted in the absence of Foxc2. (A) At E9.0 PlexinD1 is expressed in a distinct pattern in the post-otic vasculature (*), pharyngeal arch 3 mesenchyme (red arrowhead), and heart (black arrow) in wild type embryos. However, this pattern of expression is altered in Foxc2^−/− embryos (red arrowhead and black arrow in B). (C, D) PlexinD1 is expressed in the endothelium of the outflow tract (arrow) and ventricle (V) in wild type embryos but is absent from these tissues in Foxc2^−/− mutants. This disruption in PlexinD1 is not due to a loss of mesenchyme as evidenced by Twist expression, the domain of which appears expanded in Foxc2^−/− mutants (E, F), or vascular endothelium as indicated by normal PECAM1 localization (arrows in G and H) in mutant embryos. All scale bars = 100μm.

In E9.0 wild type embryos Plxnd1 was expressed in the post-otic mesenchyme and pharyngeal arches, especially the neural crest cell derived third pharyngeal arch mesenchyme, as well as the vascular endothelium of intersomitic vessels, dorsal aorta, outflow tract, and primitive ventricle of the heart (Figure 7 A, C). In contrast, in Foxc2^−/− embryos, there was no detectable expression of Plxnd1 in the third pharyngeal arch mesenchyme, outflow tract or ventricular endocardium, and the expression of Plxnd1 in intersomitic vessels and dorsal aortae was also reduced or absent (Figure 7 B, D). Thus, Foxc2 may be required for the proper expression of Plxnd1.

The lack of pharyngeal Plxnd1 expression in Foxc2^−/− embryos was not due to an absence of post-otic or pharyngeal arch mesenchyme or vasculature as evidenced by the expression of Twist1 (33) (Figure 7 E, F) and PECAM1 (Figure 7 G, H). However, the domain of Twist1 labeled cells was expanded ventrally in Foxc2^−/− embryos, consistent with the altered distribution of neural crest cells in the post-otic region. Interestingly, proper migration of cardiac neural cre is not only required for aorticopulmonary septation, but also for elongation of the conotrucus (8, 34, 35). Consistent with this idea, the second pharyngeal arch artery and its distal connection to the proximal extent of the outflow tract appears narrower (Figure 7G, H) and the conotruncus shorter in E12.5 Foxc2^−/− embryos compared to controls (Figure 4).

Although Plxnd1 is expressed in neural crest cells (30), conditional deletion of Plxd1 in neural crest cells is well tolerated in the form of normal outflow tract development. In contrast, endothelial cell specific deletion of Plxd1 with Tie2Cre, recapitulates the congenital heart defects characteristic of Plxd1^−/− mutants including persistent truncus arteriosus. The association of Plxnd1 loss-of-function with persistent truncus arteriosus (32), therefore suggests that altered Plxnd1 activity in endothelial cells in Foxc2^−/− mutant embryos is likely a critical mechanistic component underpinning the pathogenesis of the Foxc2^−/− mutant cardiac phenotype.

In addition to abnormal elongation and septation of the outflow tract in Foxc2^−/− mutant embryos we also observed hypoplasia of the developing ventricles. The basic helix-loop-helix transcription factors Hand1 and Hand2 are well known to be expressed in the ventricles and outflow tract, and also to be required for their proper morphogenesis and differentiation (36, 37). Studies of mice deficient in Hand1 and Hand2, either individually or in combination, revealed gene dosage dependent effects with respect to outflow tract defects and ventricular hypoplasia. (38–43). More specifically, Hand1^cKO/cKO; Hand2^+/− embryos display ventricle hypoplasia with thin myocardial walls, which is similar to the phenotype observed in the heart of Foxc2^−/− mutant embryos.

Therefore, we performed whole mount in situ hybridization with Hand1 and Hand2 to further characterize the molecular basis of the ventricular and outflow tract defects in Foxc2^−/− mutant mice. Hand1 localizes to the left ventricle and outflow tract in E10.5 wild type embryos, while Hand2 is more broadly expressed throughout both ventricles and the outflow tract in a pattern that encompasses Hand1 (Figure 8A, C). In contrast in Foxc2^−/− mutant embryos, Hand1 expression was reduced to a small domain in the anterior region of the left ventricle and was completely absent from the outflow tract (Figure 8B). Similarly, Hand2 expression was absent from the right and left ventricles and was considerably reduced in the outflow tract of Foxc2^−/− mutant embryos (Figure 8D). Our data indicate that Hand1 and Hand2 are expressed only in a small subset of their normal expression domains within the hearts of Foxc2^−/− mutant embryos. The perturbed expression of Hand1 and Hand2 in association with the ventricle and outflow tract defects in Foxc2^−/− mutant embryos is consistent with the phenotypic effects of mutations in Hand1 and Hand2.

Figure 8.

Loss of Hand1 and Hand2 expression in the developing heart is associated with a disruption in ventricle expansion. (A, C) Hand1 and (B, D) Hand2 whole mount in situ hybridization in E10.5 wild-type and Foxc2^−/−hearts. In Foxc2^−/− hearts (C), Hand1 is reduced to a small anterior domain of the developing left ventricle compared to controls (A). Hand2, which is normally expressed throughout the outflow tract and heart (B), is found only in the outflow tract and at comparatively reduced levels in Foxc2^−/− mutant embryos (D). All scale bars = 100 μm.

The regulation of Hand1 and Hand2 expression is complex and poorly understood in the developing heart (36), with cardiac and neural crest specific enhancers as well as chamber specific expression controlled by factors such as Gata4 (44). The Endothelin signaling pathway is one well characterized upstream regulator of Hand1 and Hand2 activity and furthermore, Foxc2 has been shown to directly activate Endothelin-converting enzyme-1 (Ece-1) through enhancer binding. Although this might suggest that altered Endothelin signaling could mediate the changes in Hand1/2 expression as part of the Foxc2 mutant phenotype, Endothelin signaling is unaffected in Foxc2 mutants (18, 45, 46). Furthermore, comparisons of individual and compound Foxc2;Ednra mutants reveal that each signaling pathway has a distinct but synergistic role in aorticopulmonary septation (45). Therefore, the alteration of Hand1 and Hand2 expression in Foxc2^−/− mutants which is consistent with the pathogenesis of ventricular hypoplasia, may occur independently of Endothelin signaling. In agreement with this idea, Foxc2 has been shown to bind to the promoter region of the Hand1 gene highlighting the potential for a direct mechanism of action of Foxc2 on Hand1/2 activity (47).

In summary, our data further support a role for Foxc2 in proper development of the heart and the great vessels. Foxc2 is expressed in the neural crest cell derived pharyngeal arch mesenchyme, in the developing vasculature, in the outflow tract, aorta and pulmonary trunk. This complex pattern led us to hypothesize that Foxc2 would be required for outflow tract septation. Our observation of abnormal cardiac neural crest cell migration and their reduced contribution to the outflow tract provide a better understanding of the pathogenesis of persistent truncus arteriosus in Foxc2^−/− embryos. Interestingly, Plxnd1 is a guidance cue known to govern cardiac neural crest cell migration and mutations in Plxnd1/PLEXIND1 are associated with persistent truncus arteriosus (32). This implies that the loss of Plxnd1 may be a major contributing factor to the cardiac phenotype in Foxc2^−/− mutant embryos. In addition, we also found altered domains of Hand1 and Hand2 expression in the hearts of Foxc2^−/− mutant embryos. As Foxc2 is expressed in neural crest cells and in the endothelium lining the outflow tract and ventricles, our data suggest that Foxc2 may play multiple tissue specific roles in cardiac morphogenesis.

Our studies were undertaken in a global Foxc2 null mutant, rendering it not possible to determine the cell autonomous nature or tissue specificity of Foxc2 function. However, future analyses of a Foxc2 conditional knockout (48) specifically in endothelial cells versus neural crest cells will therefore be essential for clarifying the tissue specific requirement for Foxc2 in cardiac development. Interestingly, a recent conditional deletion of Foxc2 in Tie2Cre expressing endothelial cells, did not result in aortic arch or ventricular septal defects (32, 49). This implies that Foxc2 may be specifically required in cardiac neural crest cells for proper heart development, and that conditional deletion in cardiac neural crest cells will recapitulate persistent truncus arteriosus. This hypothesis still needs to be formally tested, but such future studies may also shed new light on the relationship between Foxc2 and its regulation of potential downstream transcription targets such as PlexinD1, Hand1, and Hand2.

Experimental Procedures

Animal husbandry and genotyping:

All mice were housed, and all experiments were conducted in compliance with protocols approved by the Institutional Animal Care and Use Committee at the Stowers Institute for Medical Research. Foxc2 mice were maintained on a 129S6/SvEv background. Wnt1-Cre (stock number 003829 - H2afvTg^{(Wnt1-cre)11Rth} Tg(Wnt1-GAL4)11Rth/J) and R26R (stock number 003474, B6.129S4-Gt(ROSA)26Sor^tm1Sor/J) mice were obtained from the Jackson Laboratory and intercrossed with the Foxc2 line to generate Foxc2^−/−;Wnt1-Cre;R26R mouse embryos. The 6.5Pax3Cre mouse line (29) was also intercrossed with Foxc2^+/− mice to generate Foxc2^−/−;6.5Pax3Cre;R26R embryos. Genotyping of all mouse strains was determined using qPCR with specific probes designed for each strain (Transnetyx, Inc, Cordova, TN, http://www.transnetyx.com). Primer sequences for each assay can be found in Table 1. A minimum of 5 embryos were used in each of the analyses performed.

Table 1.

Genotyping Primers

Primer Name	Forward	Reverse	Reporter
Foxc2WT	CGGCCACACGTTTGCA	GTTGAACATCTCCCGGACGTT	CCAACAGCAAACTTTC
Foxc2Mut	CCATCAGAAGCTGACTCTAGAGGAT	GGAGAAAAGACCCACACACGTTT	CCGAGCTCGAATTCAA
R26RWT	TTCCCTCGTGATCTGCAACTC	CTTTAAGCCTGCCCAGAAGACT	CCGCCCATCTTCTAGAAAG
R26R	TTCCCTCGTGATCTGCAACTC	GGACTACTGCGCCCTACAG	TCTTTCTAGTGGATCCCCC
Cre	TTAATCCATATTGGCAGAACGAAAACG	CAGGCTAAGTGCCTTCTCTACA	CCTGCGGTGCTAACC

Histological Staining and Whole-mount in situ hybridization:

E12.5 embryos were fixed in 4% paraformaldehyde (PFA) and embedded in O.C.T. compound. 10 μm sections were stained with hematoxylin and eosin following standard procedures. Whole mouse embryos (9.0 – 10.5dpc) were fixed overnight in 4% PFA at 4°C, then rinsed in phosphate buffered saline (PBS) with 1% Tween-20 followed by step-wise dehydration to 100% methanol. Anti-sense digoxigenin-labeled (dig-UTP, Roche) riboprobes were synthesized for Crabp1, Foxc2, Hand1, Hand2, and Sox10, Twist, and PlexinD1. Whole mount in situ hybridizations were performed according to standard protocols (50, 51) with minor modifications.

Immunohistochemistry and DAPI staining:

PECAM-1 (550274, BD Biosciences) and Foxc2 (AF6989, R&D Systems) whole mount immunostaining were respectively performed on E9.0 and E10.5 embryos, previously fixed in 4% paraformaldehyde and dehydrated to 100% methanol. Endogenous peroxidase activity was blocked by incubation in Dent’s bleach for 2 hours at room temperature. Embryos were rehydrated to PBST, blocked in 3% BSA in PBS twice for 1 hour at room temperature, and incubated in 1:500 dilution of anti-PECAM-1, anti-FOXC2 or anti-GFP overnight at room temperature. Embryos were rinsed in PBST, and then incubated with a 1:200 dilution of a horseradish peroxidase (HRP) conjugated secondary antibody or 1:1000 fluorescent secondary antibodies. HRP signal was detected through diaminobenzidene (DAB) staining (Sigma, D5905). For gross visualization of whole embryos and hearts, tissues were incubated overnight in DAPI (2μg/ml) to label all nuclei (52). Tissues were then cleared in 50% glycerol and scanned using a Zeiss LSM5 Upright Pascal Confocal microscope. Projected Z-stacks were flattened and exported to Adobe Photoshop.

β-galactosidase staining:

Whole embryos from E9.0–11.5 were collected and fixed from 30 to 60 minutes in 0.2% glutaraldehyde, 5μM EGTA, 100μM MgCl₂ on ice. Embryos were rinsed and stained using Millipore Tissue Rinse Solutions (#BG-6-B, #BG-7-B) and Tissue Stain Base (#BG-8-C) containing 1mg/ml Xgal (5-bromo-4-chloro-3-indolyl β-D-galactopyranoside, Sigma, B4252) according to the manufacturer’s protocol. Embryos were post-fixed in 4% PFA and selected embryos were embedded in paraffin and sectioned (10μm).

Bullet Points:

• Foxc2 is expressed in migrating neural crest cells and cardiac mesenchyme.

• On the 129s6/SvEv genetic background Foxc2 loss-of-function embryos exhibit persistent truncus arteriosus and ventricle hypoplasia, leading to embryonic lethality at E13.5.

• Aberrant cardiac neural crest migration in association with disruption in guidance cues precedes reduced contribution to the outflow tract.

• Ventricle patterning and development is impaired in the absence of Foxc2.