|Home | About | Journals | Submit | Contact Us | Français|
Disheveled (Dvl) is a key regulator of both the canonical Wnt and the planar cell polarity (PCP) pathway. Previous genetic studies in mice indicated that outflow tract (OFT) formation requires Dvl1 and 2, but it was unclear which pathway was involved and whether Dvl1/2-mediated signaling was required in the second heart field (SHF) or the cardiac neural crest (CNC) lineage, both of which are critical for OFT development. In this study, we used Dvl1/2 null mice and a set of Dvl2 BAC transgenes that function in a pathway-specific fashion to demonstrate that Dvl1/2-mediated PCP signaling is essential for OFT formation. Lineage-specific gene ablation further indicated that Dvl1/2 function is dispensable in the CNC, but required in the SHF for OFT lengthening to promote cardiac looping. Mutating the core PCP gene Vangl2 and non-canonical Wnt gene Wnt5a recapitulated the OFT morphogenesis defects observed in Dvl1/2 mutants. Consistent with genetic interaction studies suggesting that Wnt5a signals through the PCP pathway, Dvl1/2 and Wnt5a mutants display aberrant cell packing and defective actin polymerization and filopodia formation specifically in SHF cells in the caudal splanchnic mesoderm (SpM), where Wnt5a and Dvl2 are co-expressed specifically. Our results reveal a critical role of PCP signaling in the SHF during early OFT lengthening and cardiac looping and suggest that a Wnt5a→ Dvl PCP signaling cascade may regulate actin polymerization and protrusive cell behavior in the caudal SpM to promote SHF deployment, OFT lengthening and cardiac looping.
Outflow tract (OFT) malformation is one of the most common congenital heart defects in humans (Hoffman and Kaplan, 2002; Samanek, 2000). The OFT, initially a single vascular conduit linking the right ventricle and aortic sac, is septated later into the aorta and pulmonary artery that connect with the left and right ventricles, respectively. Defects in OFT septation lead to persistent truncus arteriosus (PTA), while its misalignment with the ventricles causes double outlet right ventricle (DORV), overriding aorta or transposition of the great arteries. Lineage studies show that the OFT does not arise from expansion of the initial heart tube (also known as the first heart field (FHF), but is added later from the second heart field (SHF) cells in the pharyngeal and splanchnic mesoderm (SpM) (Cai et al., 2003; Dyer and Kirby, 2009a; Kelly et al., 2001; Li et al., 2010; Ma et al., 2008; Mjaatvedt et al., 2001; Verzi et al., 2005; Waldo et al., 2001). Recent studies further suggest that the FHF and SHF originate from a contiguous population of mesodermal progenitors, but differ in the timing of their contribution to the heart (Dyer and Kirby, 2009a; Ma et al., 2008). The FHF reflects the first wave of mesodermal cells that differentiate to form the initial heart tube, whereas the SHF cells remain as rapidly proliferating progenitors that undergo gradual differentiation and deployment to the heart (Dyer and Kirby, 2009a).
A critical balance between proliferation, differentiation and deployment has to be maintained in the SHF to ensure that a threshold number of descendents can be added to the OFT to drive its rapid elongation during cardiac looping (Black, 2007; Dyer and Kirby, 2009a; Li et al., 2010). Sufficient OFT lengthening is essential for proper cardiac looping to reposition the OFT above the interventricular septum so that upon the induction of OFT septation by cardiac neural crest (CNC) cells, the aorta can be connected to the left ventricle (Dyer and Kirby, 2009a; Li et al., 2010). Events that compromise the contribution of the SHF cells to the OFT may not only cause alignment defects, but may also perturb the septation process (Black, 2007; Dyer and Kirby, 2009a; Kirby, 2008; Li et al., 2010; McCulley et al., 2008; Theveniau-Ruissy et al., 2008). Studies from many groups have delineated elegantly how proliferation and differentiation in the SHF can be coordinately regulated by a transcriptional network (Black, 2007; Zhang et al., 2006) that integrates signaling input from pathways including Fgf, Tgfβ/Bmp, Shh, Wnt, Notch and Retinoic acid (Cohen et al., 2007; Dyer and Kirby, 2009b; High et al., 2009; Ilagan et al., 2006; Kwon et al., 2009; Li et al., 2010; Lin et al., 2007; McCulley et al., 2008; Park et al., 2008; Rochais et al., 2009a). In contrast, what governs the deployment of SHF cells to the OFT remains largely unknown.
In this study, we provide evidence that the planar cell polarity (PCP) pathway is required in the SHF lineage for early OFT morphogenesis and cardiac looping, and may play a key role in the deployment of SHF cells from the SpM to the OFT. A branch of the β-catenin independent non-canonical Wnt pathway, the PCP pathway is required for coordinating cellular polarity in the plane of the epithelium in flies (Zallen, 2007) and polarized cell intercalation or migration during convergent extension (CE) tissue morphogenesis in Xenopus and zebrafish (Heisenberg et al., 2000; Keller, 2002; Wallingford et al., 2000; Yin et al., 2008). This pathway mediates its effects through several components shared with the canonical Wnt pathway, including the Frizzled (Fz) receptor and cytoplasmic protein Disheveled (Dvl), along with a distinct set of core PCP proteins including Van Gogh (Vangl in mammals). During CE, PCP signaling regulates cell morphology and protrusive activity by modulating cytoskeletal organization and dynamics (Keller, 2002; Khadka et al., 2009; Wallingford et al., 2000). Two non-canonical Wnts, Wnt5a and Wnt11, are required to activate PCP signaling during CE in frogs and zebrafish (Heisenberg et al., 2000; Kilian et al., 2003). In mice, we and others have found previously that mammalian core PCP proteins, including Dvl1/2/3, Fz3/6 and Vangl1/2, are essential for tissue morphogenesis in the neural plate, inner ear, skin and limb (Devenport and Fuchs, 2008; Etheridge et al., 2008; Torban et al., 2008; Wang et al., 2011; Wang et al., 2006a; Wang et al., 2005; Wang et al., 2006b; Ybot-Gonzalez et al., 2007). Genetic studies suggest that Wnt5a may regulate PCP signaling during neural plate and limb morphogenesis in mice (Qian et al., 2007; Wang et al., 2011).
During mouse OFT formation, Wnt5a and Vangl2 have been suggested to function in the CNC and OFT cardiomyocytes, respectively (Phillips et al., 2005; Schleiffarth et al., 2007). Their roles in the SHF, however, have been unclear. Previously, we found that mouse Dvl genes are also critical for OFT development (Etheridge et al., 2008; Hamblet et al., 2002). In light of the central role of Dvl in both the canonical Wnt and the PCP pathway, we performed pathway- and tissue-specific mutagenesis of Dvl2 in this study and uncovered a key role of Dvl2-mediated PCP signaling in the SHF. In conjunction with additional genetic, morphometric and histological analysis in Dvl1/2, Vangl2 and Wnt5a mutants, we propose a novel model in which Wnt5a-activated PCP signaling induces a mesenchymal to epithelial conversion in the caudal SpM to promote SHF deployment, OFT morphogenesis and cardiac looping.
Wnt5a and Vangl2 mutant mice were obtained from the Jackson Laboratory and genotyped as described (Murdoch et al., 2001; Yamaguchi et al., 1999). Genotyping of the other mouse strains used in the study has been described previously: Dvl1 (Lijam et al., 1997) and Dvl2 (Hamblet et al., 2002); Dvl2 BAC transgenes (Wang et al., 2006a); Islet1-Cre (Cai et al., 2003) and Wnt1-Cre (Jiang et al., 2000).
Animal care and use was in accordance with NIH guidelines and was approved by the Animal Care and Use Committee of the University of Alabama at Birmingham.
Embryos derived from appropriate crosses were dissected between embryonic days (E) 9.5 to E18.5 and yolk sac was retained for PCR genotyping. Embryos were fixed in 4% paraformaldehyde at 4°C overnight and stored in 70% ethanol.
To quantify the length of the OFT, fixed E9.5 embryos between 24-26 somites were imaged with their right side facing up using a Leica MZ16FA stereoscope equipped with a DFC490 CCD camera. To determine the OFT length, the LAS Interactive Measurement Software Module was utilized to draw and quantify the length of a continuous line along the inner curvature of the OFT from the distal end to the proximal border with the right ventricle (as shown by a red line in Fig. 3A). Statistical analyses were performed using two-tailed Student’s t-test.
Hearts from E11.5 and E18.5 embryos were dissected out and imaged prior to embedding and histological section.
OFT and SpM were micro-dissected from 3 E9.5 wild-type embryos. Samples were pooled and lysed with Trizol reagent (Invitrogen) to collect total RNA. Quantitative real-time PCR was performed as described previously(Wang et al., 2011). GAPDH was used to normalize gene expression. cDNA samples from 3 sets of embryos were tested, and each cDNA sample was tested in replicate.
Primers used were: GAPDH (TGAAGGTCGGAGTCAACGGATTTGGT; AAATGAGCCCCAGCCTTCTCCATG); Dvl2 (AGCAGTGCCTCCCGCCTCCTCA; CCCATCACCACGCTCGTTACTTTG).
Fixed embryos and heart tissue were properly oriented in Histogel (Thermo/Fisher, Pittsburgh, PA) and processed for paraffin embedding, or were processed through sucrose gradients and embedded in OCT (Tissue-Tek/Sakura, Torrance, CA) for cryo-sectioning. Paraffin–embedded samples were sectioned at 7um and stained with Hematoxylin & Eosin (H&E) to examine histology. OCT embedded samples were cryo-sectioned at 25um, air-dried at room temperature, fixed with 4%PFA, permeabilized with PbTX (0.1% Tween in PBS) and stained with Phalloidin-TRITC (Sigma, St.Louis, MO). Fluorescence immuno-histochemistry was performed on paraffin sections as described (Wang et al., 2011). Primary antibodies used were anti-CleavedCaspase-3 (Asp175) (Cell Signaling, Danvers, MA) and anti pHH3 (Ser10) (Millipore, Billerica, MA). Alexa Fluor 488-conjugated goat anti-rabbit IgG (Invitrogen, Carlsbad, CA) was used as secondary antibody. A tyramide signal amplification kit (PerkinElmer, Covina, CA, #NEL741E001KT) was used for anti-Cleaved Caspase-3 detection. All fluorescent and confocal images were acquired with an Olympus FV1000 Laser Confocal Scanning microscope and were subsequently analyzed using the FV10-ASW software. Whole-mount in situ hybridization was carried out with a standard protocol (Wilkinson and Nieto, 1993).
The three mouse Dvl genes, Dvl1, 2 and 3, display a broad and overlapping expression pattern. We previously reported that mice carrying null mutations in Dvl1 (Dvl1−/−) are viable (Lijam et al., 1997), but Dvl2 null mutants (Dvl2−/−) display partially penetrant neonatal lethality and OFT defects, primarily in the form of DORV and PTA (Hamblet et al., 2002). Deleting both Dvl1 & 2 in mice (Dvl1−/−; Dvl2−/−) increases the penetrance of DORV and PTA to 100% (Fig. 2C), suggesting that Dvl1 and 2 have redundant function during OFT formation. Dvl1−/−; Dvl2−/− mutants also develop three additional defects not observed in each single mutant: i) randomized stereocilia orientation in the inner ear sensory hair cells; ii) shortened cochlea; and iii) failure to close the entire neural tube(Hamblet et al., 2002; Wang et al., 2006a). Our previous studies concluded that the last three defects are due to disruption of Dvl-mediated PCP signaling (Wang et al., 2006a; Wang et al., 2005), but the cause of the OFT defects remained unclear.
To determine the cause of the OFT defects in Dvl1−/−; Dvl2−/− mutants, we first determined the Dvl-mediated pathway(s) disrupted in the mutants. Dvl has three conserved domains, DIX, PDZ and DEP (Fig. 1A) and studies in various model organisms have indicated that the DIX domain is only required for the canonical Wnt pathway while the DEP domain is only required for the PCP pathway (Axelrod et al., 1998; Boutros et al., 1998; Wallingford et al., 2000). Therefore, we could determine the signaling pathway underlying the OFT defect in Dvl1/2 mutants using a set of Dvl2 domain mutations established previously by BAC (bacterial artificial chromosome) recombineering and transgenesis. Because of their large size and low copy number when inserted in to the genome, BAC transgenes are more likely to recapitulate endogenous gene expression patterns and levels (Lee et al., 2001). The series of Dvl2 transgenes were constructed from a BAC clone that contains 170 kb genomic sequence encopassing mouse Dvl2 and its flanking region (Wang et al., 2006a).
As illustrated in Fig. 1A, the Dvl2 BAC allelic series consists of a wild-type BAC (Dvl2-EGFP2), two domain-deletion mutants (ΔDIX-EGFP and ΔDEP-EGFP) and a point mutation (dsh1-EGFP). In the wild-type BAC Dvl2-EGFP2, the Dvl2 coding region is intact, but an EGFP cDNA is inserted in-frame following the last codon of Dvl2, generating a fully functional Dvl2-EGFP fusion protein that allowed us to study the localization of Dvl2 in the neural tube and inner ear previously (Wang et al., 2006a; Wang et al., 2005). In addition, this Dvl2-EGFP2 BAC allele contained two LoxP sites (one in intron2 and one behind exon15, blue triangles in Fig. 1A), resulting in a floxed (LoxP flanked), conditional BAC allele. When crossed into Dvl1−/−; Dvl2−/− mutants, Dvl2-EGFP2 fully complemented the cochlea and neural tube defects and rescued them to fertile adults(Wang et al., 2006a). Not surprisingly, Dvl2-EGFP2 can also rescue the OFT defects in E18.5 Dvl1−/−; Dvl2−/− mutants (Fig. 1B, n=8, also see Fig. 2H for section), indicating that the wild-type Dvl2-EGFP2 BAC transgene can fully substitute for all endogenous Dvl2 function.
We then used the ΔDEP-EGFP BAC transgene to test whether the DEP domain, important only for PCP signaling, is required for rescuing the OFT defect in Dvl1−/−; Dvl2−/− mutants. ΔDEP-EGFP deletes the entire DEP domain (aa442-736) and an identical mutant in Xenopus cannot function in CE, but still activates canonical Wnt signaling (Wallingford et al., 2000). Our previous studies also indicate that ΔDEP-EGFP abolishes PCP signaling during neural tube closure in mice. When we crossed ΔDEP-EGFP into Dvl1−/−; Dvl2−/− mutants and examined the hearts of Dvl1−/−; Dvl2−/−; ΔDEP-EGFP embryos at E18.5, we recovered OFT defects in the form of DORV or PTA, similar to those in Dvl1−/−; Dvl2−/− mutants (compare Fig. 1D to Fig. 2C, n=6). This result indicates that the DEP domain, and therefore Dvl-mediated PCP signaling, is required for rescuing the OFT defects in Dvl1−/−; Dvl2−/− mutants.
To further confirm this finding, we used the dsh1-EGFP BAC transgene that contains a point mutation leading to a K to M substitution at aa446 (Fig. 1A). In flies, this mutation specifically abolishes PCP signaling but leaves Wnt signaling intact (Axelrod et al., 1998; Boutros et al., 1998) and we demonstrated previously that in mice, this mutation also abolished PCP signaling in the inner ear and during neurulation (Wang et al., 2006a). When we bred dsh1-EGFP into Dvl1−/−; Dvl2−/− mutants, we found that it also failed to rescue the OFT defects (Fig. 1E, n=5), confirming that the OFT defect in Dvl1/2 mutant is due, at least in part, to disruption of PCP signaling.
To test whether dsh1-EGFP can rescue the partially penetrant OFT defects in Dvl2−/− single mutants, we crossed Dvl2+/−; dsh1-EGFP to Dvl2−/− mice. In the progeny, we found that 73% of Dvl2−/− mutants survived to weaning (66 expected; 48 recovered); but surprisingly, only 18% of the expected Dvl2−/−; dsh1-EGFP mutants survived to weaning (66 expected; 12 recovered). Heart dissection and sectioning at E18.5 revealed DORV or PTA in 21% of Dvl2−/− mutants (3 out of 14 embryos, data not shown) and in 70% of Dvl2−/−; dsh1-EGFP mutants (5 out of 7 embryos, data not shown), indicating that dsh1 exerted a dominant negative effect during OFT development. We reason that the mutant dsh1 protein may be able to interact with most of the molecular partners of Dvl, but the complexes containing dsh1 cannot function properly in the PCP pathway, thus antagonizing the activity of remaining Dvl1 or 3. Collectively, these results indicate an essential role of Dvl-mediated PCP signaling in OFT development.
Finally, we used the ΔDIX-EGFP BAC transgene to assess whether perturbation of canonical Wnt signaling may also contribute to the OFT defects in Dvl1−/−; Dvl2−/− mutants. ΔDIX-EGFP deletes part of the DIX domain (aa67-159) essential for Wnt signaling (Capelluto et al., 2002), but retains the ability to mediate PCP signaling during mouse neural tube closure (Wang et al., 2006a). When crossed into Dvl1−/−; Dvl2−/− mutants, ΔDIX-EGFP was able rescue the OFT defects at E18.5 (Fig. 1C, n=8) and the lethality at perinatal stage (6 Dvl1−/−; Dvl2−/−; ΔDIX-EGFP mutants recovered at weaning, 6 expected), indicating that restoring only PCP signaling activity is sufficient to rescue the OFT defect in Dvl1−/−; Dvl2−/− mutants. Together, these data indicate that the OFT defect in Dvl1−/−; Dvl2−/− mutants is solely due to disruption of the PCP pathway.
Both the SHF (Cai et al., 2003; Kelly et al., 2001; Li et al., 2010; Verzi et al., 2005; Waldo et al., 2001) and the CNC (Snider et al., 2007; Waldo et al., 2005a) lineage are essential for OFT development. To determine which lineage requires Dvl1/2 function, we performed tissue specific gene-ablation using the Dvl2-EGFP2 BAC transgene. Because Dvl2-EGFP2 contains two LoxP sites flanking exons 3 and 15, we predicted that Cre mediated recombination could inactivate the transgene by deleting all exons from 3 to 15, including the entire 3′ UTR (Fig. 2A). Using a ubiquitously expressed EIIa-Cre transgene(Lakso et al., 1996) and 3 primers surrounding the two LoxP sites, we confirmed that Dvl2-EGFP2 could indeed be efficiently recombined as predicted (Fig. 2B).
To inactivate both Dvl1 and 2 only in the CNC, we crossed Dvl1−/−; Dvl2−/−; Dvl2-EGFP2 mice with Dvl1−/−; Dvl2+/−; Wnt1-Cre mice that expressed Cre in all CNC progenitors in the dorsal neural tube(Jiang et al., 2000). We found normal OFT formation in the hearts from E18.5 Dvl1−/−; Dvl2−/−; Dvl2-EGFP2; Wnt1-Cre embryos (n=5, Fig. 2E and 2I). Furthermore, over 95% of Dvl1−/−; Dvl2−/−; Dvl2-EGFP2; Wnt1-Cre mice survived beyond weaning (26 expected, 25 recovered, Table 1). Therefore, deleting both Dvl1 and 2 only in the CNC lineage is not sufficient to recapitulate the OFT defect observed in Dvl1−/−; Dvl2−/− null mutants.
In contrast, in parallel crosses where we used Islet1-Cre (Isl1-Cre) that expressed Cre in SHF progenitors in the pharyngeal and splanchnic mesoderm (Cai et al., 2003; Ma et al., 2008; Sun et al., 2007), we recovered OFT defects in E18.5 Dvl1−/−; Dvl2−/−; Dvl2-EGFP2; Islet1-Cre embryos (8 mutants recovered, 4 with DORV and 4 with PTA, representative mutant heart with PTA is shown in Fig. 2F and 2J). Also consistent with OFT defects as the cause of lethality, only 8% of Dvl1−/−; Dvl2−/−; Dvl2-EGFP2; Islet1-Cre mutants survived to weaning (25 expected, 2 recovered, Table 1), and these survivors were likely due to incomplete inactivation of Dvl2-EGFP2 by Isl1-Cre (Ma et al., 2008). Collectively, the tissue specific gene-ablation experiments indicate that Dvl1/2 function is required in the SHF, but not the CNC, lineage for proper OFT development.
To further investigate how the OFT defect arose in Dvl1−/−; Dvl2−/− mutants, we examined their heart morphology at different stages of development. By E9.5, the OFT in Dvl1−/−; Dvl2−/− and Dvl1−/−; Dvl2−/−; Dvl2-EGFP2; Isl1-Cre mutants already lacked the rightward curve apparent in wild-type embryos (green arrows and traced by red lines in Fig. 3A&B and data not shown), suggesting aberrant cardiac looping. Consistent with this idea, in frontal views, the right ventricle was located higher than the left ventricle (red line, Fig. 3F) in Dvl1−/−; Dvl2−/−; Dvl2-EGFP2; Isl1-Cre and Dvl1−/−; Dvl2−/− mutants, while in wild-type embryos the two ventricles were on the same horizontal plane (red line, Fig. 3E). At E11.5, the right ventricle in Dvl1−/−; Dvl2−/− and Dvl1−/−; Dvl2−/−; Dvl2-EGFP2; Isl1-Cre mutants eventually descended to the same level as the left ventricle, but an apparently shorter OFT was aligned with only the right ventricle, while in wild-type embryos the OFT was situated over the inter-ventricular septum (compare Fig. 4A&B and Fig. 4E&F). The mis-alignment of OFT is therefore the most likely cause of DORV during subsequent OFT remodeling and septation in Dvl1/2 mutants.
To further characterize the role of PCP signaling in early OFT morphogenesis, we examined Looptail (Vangl2Lp) mice (Kibar et al., 2001; Murdoch et al., 2001) carrying a mutation in Vangl2, one of the two Vang homologs in mice. Vangl2Lp/Lp homozygotes are also known to display DORV (Henderson et al., 2001). When we examined Vangl2Lp/Lp mutants at E9.5, we found that their OFTs also lacked the rightward curve (Fig. 3C) and displayed aberrant cardiac looping (Fig. 3G), similar to Dvl1/2 mutants.
PCP signaling in frogs and zebrafish requires non-canonical Wnt ligand Wnt5a (Kilian et al., 2003; Wallingford et al., 2001). In mice, null mutation of Wnt5a (Wnt5a−/−) causes PTA or DORV (Schleiffarth et al., 2007). Our examination of E9.5–11.5 Wnt5a−/− mutants revealed cardiac looping defects similar to those in Dvl1/2 and Vangl2 mutants (Fig. 3D and data not shown), suggesting that aberrant cardiac looping may also contribute to the OFT defects in Wnt5a mutants.
A previous study on Vangl2Lp/Lp mutants attributed the abnormal cardiac looping to neural tube closure and axial rotation defects (Henderson et al., 2001). However, in both Dvl1−/−; Dvl2−/−; Dvl2-EGFP2; Isl1-Cre and Wnt5a−/− mutants, neural tube closure and axial rotation are normal, yet aberrant cardiac looping persists (Fig. 3D&F). To explore alternative causes for the looping defects, we assessed OFT length since maximal OFT extension is critical for cardiac looping and proper OFT alignment (Li et al., 2010; Rochais et al., 2009b; Sugishita et al., 2004; Yelbuz et al., 2002). When we measured the OFT length along the inner curvature (from the distal end of the OFT to the border between the OFT and the right ventricle) at E9.5 (24-26 somites), we found significant shortening of the OFT in each mutant, with Wnt5a mutants more severe than Dvl1/2 and Vangl2 mutants (Fig. 3I; wild-type: 1.01 ± 0.04 mm, Dvl1−/−; Dvl2−/−: 0.76 ± 0.07 mm, Vangl2Lp/Lp: 0.77± 0.03 mm, Wnt5a−/−: 0.70± 0.07, p<0.01 between wild-type and each mutant). Therefore, our morphometric analyses indicate that the cardiac looping defect in mouse PCP mutants is correlated with OFT shortening.
While the signaling mechanism of Wnt5a in mammals has been controversial, the similar OFT shortening and looping defects in Wnt5a, Dvl1/2 and Vangl2 mutants suggest that Wnt5a may function through the PCP pathway during early OFT morphogenesis. To test this hypothesis, we studied the genetic interaction between Vangl2 and Wnt5a. We found that reducing the dosage of Wnt5a by 50% significantly enhanced both the cardiac looping and OFT shortening defects in Vangl2Lp/Lp mutants. In E9.5 Vangl2Lp/Lp; Wnt5a+/− mutants, the right ventricle was located almost vertically on top of the left ventricle (compare Fig. 3H to 3G), and the OFT was also shortened significantly when compared to Vangl2Lp/Lp mutants (Fig. 3I, 0.69± 0.04 mm in Vangl2Lp/Lp; Wnt5a+/− vs. 0.77± 0.03 mm in Vangl2Lp/Lp; p=0.02). In E11.5 Vangl2Lp/Lp; Wnt5a+/− mutants, the right ventricle was still higher than the left ventricle and the OFT was aligned solely with the right ventricle (compare Fig. 4C&D and 4G&H). The genetic interaction between Wnt5a and Vangl2 supports the hypothesis that Wnt5a signals through the PCP pathway during early OFT morphogenesis.
To investigate how Wnt5a-initiated PCP signaling might regulate early OFT morphogenesis, we analyzed Wnt5a expression by in situ hybridization. Between E9.0 to 10.5, Wnt5a was highly expressed in the caudal region of the SHF splanchnic mesoderm (SpM) (green arrow, Fig. 5A, also see ref. (Chen et al., 2012; Schleiffarth et al., 2007; Yamaguchi et al., 1999)) and in the anterior region of the first and second pharyngeal arches, but interestingly, not within the OFT or the rest of the heart proper.
We also examined the expression of the Dvl2-EGFP2 BAC transgene capable of fully replacing endogenous Dvl2 (Fig. 1B & 2D). At E9.5, Dvl2-EGFP was expressed highly in the SpM (red arrow in Fig. 5B) and pharyngeal arches, but at much lower level in the OFT. To determine whether endogenous Dvl2 is also differentially expressed in the SpM and OFT, we performed quantitative RT-PCR using micro-dissected SpM and OFT from E9.5 embryos. Our results indicate that endogenous Dvl2 expression in the SpM is also over four fold higher than that in the OFT (Fig. 5C).
The overlapping expression pattern of Wnt5a and Dvl2-EGFP prompted us to carry out extensive analysis to identify the defects that could contribute to early OFT shortening in Wnt5a−/− and Dvl1−/−; Dvl2−/− mutants. Examination of phospho-Histone H3 and cleaved caspase 3 staining at E9.5 revealed no difference in cell proliferation or apoptosis rates in the mutant OFT or the SHF splanchnic and pharyngeal mesoderm, suggesting that the OFT shortening is not caused by cell proliferation or apoptosis defects in the OFT or the SHF (supplemental Fig. 1).
Interestingly, when we examined H&E (hematoxylin and eosin) stained sagittal sections of E9.5 embryos, we found an unusual histological abnormality in the caudal SpM of both Wnt5a and Dvl1−/−; Dvl2−/− mutants. In E9.5 wild-type embryos, the caudal SpM dorsal and anterior to the inflow tract (IFT) consisted of loosely packed mesenchyme adjacent to a cohesive, epithelial-like sheet (black box in Fig. 5D and enlarged view in 5J). Rostrally, the epithelial sheet gradually gained a single-layered, columnar character and became contiguous with the OFT, while the loosely packed mesenchyme became sparse (yellow box in Fig. 5D and enlarged view in Fig. 5G). The rostral SpM of Wnt5a−/− and Dvl1−/−; Dvl2−/− mutants appeared similar to that in the wild-type (compare yellow-boxed area in Fig. 5D to Fig. 5E&F and enlarged view in Fig. 5G to Fig. 5H&I), but in the caudal SpM of both mutants, SHF progenitor cells aggregated into compact clusters instead of forming an epithelial-like sheet as in the wild-type (compare black-boxed area in Fig. 5D to Fig. 5E&F and enlarged view in Fig. 5J to Fig. 5K&L).
The aberrant cell packing in the caudal SpM of Wnt5 and Dvl1/2 mutants coincides with Wnt5a expression in this region (Fig. 5A). To characterize this abnormality further, we performed confocal scanning microscopy on the H&E stained sections, taking advantage of the fact that Eosin can emit strong fluorescence upon laser beam excitation (excitation 525nm; emission 545nm), allowing for high resolution assessment of cellular morphology (de Carvalho and Taboga, 1996; McMahon et al., 2002). Detailed confocal analysis revealed that wild-type SpM cells residing in the caudal region of the epithelial-like sheet (boxed area in Fig. 5J) did not possess typical epithelial morphology. Instead, they retained their mesodermal character and displayed a highly protrusive morphology with numerous filopodia-like extensions (white arrows, Fig. 5M). In both Wnt5a−/− and Dvl1−/−; Dvl2−/− mutants, however, caudal SpM cells were more rounded with smooth surfaces and fewer extensions (Fig. 5N&O). Furthermore, instead of organizing into a cohesive sheet of 1–2 cell layers as in the wild-type, the mutant cells formed multi-layered, compact clusters (compare Fig. 5M to 5N&O).
The aberrant cell morphology and packing in the caudal SpM of Dvl1−/−; Dvl2−/− and Wnt5a−/− mutants prompted us to further examine actin organization, since filopodia formation requires actin polymerization and Dvl-mediated PCP signaling is important for actin polymerization during Xenopus CE (Khadka et al., 2009). To this end, we stained sagittal cryosections of E9.5 wild-type and Wnt5a−/− and Dvl1−/−; Dvl2−/− mutant embryos with phalloidin, a marker for F-actin. Consistent with the H&E staining results, cells in wild-type caudal SpM were organized into a largely single-layered structure with actin filaments aligned along the apical-basal axis (Fig. 6A). Along the basal side of these cells, actin filaments also extended into numerous filopodia (green arrows in Fig. 6A). The loosely packed mesenchymal cells (red asterisk in Fig. 6A and enlarged view in Fig. 6D) adjacent to the epithelial-like sheet also extended multiple F-actin rich filopodia (yellow arrowheads in Fig. 6D), indicative of highly protrusive morphology.
In contrast, caudal SpM cells in both Wnt5a−/− and Dvl1−/−; Dvl2−/− mutants displayed significantly diminished actin polymerization, with only diffuse phalloidin staining present at the borders between adjacent cells (Fig. 6B&C). They also exhibited a more compact, multilayered organization with very few loosely-packed mesenchymal cells, and even these are rounded and lack F-actin rich filopodia (red asterisks in Fig. 6B&C and enlarged views in Fig. 6E&F).
The actin organization in the rostral SpM cells of both Wnt5a−/− and Dvl1−/−; Dvl2−/− mutants (Fig. 6H&I), however, was similar to that in the wild-type (Fig. 6G), with F-actin enriched primarily at apical surface. Moreover, actin polymerization and organization in the myocardial layer of the OFT also appeared normal in both mutants (Fig. 6K&L). Therefore, the aberrant cell morphology and defective actin polymerization in Wnt5a−/− and Dvl1−/−; Dvl2−/− mutants are specific to SHF progenitors in the caudal SpM, coinciding with Wnt5a expression (Fig. 5A).
Dvl genes are evolutionarily conserved, key cytoplasmic regulators of both the canonical Wnt and the PCP pathway. Previous genetic studies in the mouse using β-catenin conditional knockout and over-expression mutants have demonstrated clearly that the canonical Wnt pathway regulates OFT development through controlling progenitor expansion and differentiation in the SHF (Ai et al., 2007; Cohen et al., 2007; Klaus et al., 2007; Kwon et al., 2009; Lin et al., 2007) as well as cell proliferation in the CNC (Kioussi et al., 2002). On the other hand, although the PCP pathway is clearly indispensable for OFT development since mouse PCP mutants display severe OFT defects, few genetic studies had been carried out to define how and in what lineage the PCP pathway regulates OFT development.
In this study, we first used a Dvl2 BAC allelic series, consisting of domain deletions and a point mutation that specifically disrupt either canonical Wnt or PCP signaling, to determine the cause of the OFT defects in mice lacking Dvl1 and 2, two of the three Dvl genes in mammals. Our results clearly demonstrated that in Dvl1−/−; Dvl2−/− mutants, the OFT defects arise solely from disruption of PCP signaling. This conclusion is most strongly supported by the fact that the Dvl2 BAC transgene carrying the dsh1 point mutation, known to disrupt PCP signaling but leave canonical Wnt signaling intact (Axelrod et al., 1998; Boutros et al., 1998; Park et al., 2005), completely failed to rescue the OFT defects in Dvl1−/−; Dvl2−/− mutants. In contrast, the ΔDIX BAC transgene, which lacks residues critical for Wnt but not PCP signaling (Boutros et al., 1998; Capelluto et al., 2002; Rothbacher et al., 2000), was able to rescue the OFT defects in Dvl1−/−; Dvl2−/− mutants. Our result is also consistent with previous studies in which a LEF/TCF Wnt reporter was crossed into Dvl1−/−; Dvl2−/− mutants and no defects in Wnt signaling activity were detected during embryogenesis (Etheridge et al., 2008). Therefore, it appears that the remaining Dvl3 in Dvl1−/−; Dvl2−/− mutants is sufficient to maintain Wnt, but not PCP, signaling activity above the threshold level for normal OFT development.
Using the floxed Dvl2-EGFP2 BAC transgene and tissue specific Cre lines, our genetic studies further indicate that during mouse OFT development, Dvl1/2 are required in the Isl1-Cre positive SHF lineage, but are dispensable in the Wnt1-Cre positive CNC lineage. Therefore, the CNC anomalies described previously in Dvl1−/−; Dvl2−/− mutant (Hamblet et al., 2002; Kioussi et al., 2002) could be either secondary to defects in the SHF, or insufficient to perturb overall OFT development.
Taken together, our genetic analyses indicate that Dvl1/2-mediated PCP signaling is essential in the SHF lineage for OFT development. This conclusion is further supported by the fact that deleting Dvl1/2 in the SHF causes cardiac defects similar to those following mutation of core PCP gene Vangl2, including aberrant cardiac looping and OFT mis-alignment from E9.5. Furthermore, mutation of non-canonical Wnt gene Wnt5a causes similar cardiac looping defects and Wnt5a genetically interacts with Vangl2, suggesting that Wnt5a may act as a PCP ligand. Importantly, our morphometric analyses indicate that aberrant cardiac looping in each PCP mutant is correlated with reduced OFT length, characteristic of compromised recruitment of SHF cells during OFT lengthening (Rochais et al., 2009b).
Interestingly, during OFT lengthening at E9.5, Wnt5a expression was not detected within the OFT by in situ hybridization (Fig. 5A and ref. (Schleiffarth et al., 2007; Yamaguchi et al., 1999)). Instead, Wnt5a is co-expressed with Dvl2 in the caudal SpM, which harbors SHF progenitors that are deployed to give rise to the inferior wall of the OFT (Bertrand et al., 2011; Li et al., 2010; Waldo et al., 2005b). In Dvl1/2 and Wnt5a mutants, SHF cells in the SpM display no defects in either cell proliferation or apoptosis, suggesting that the shortened OFT in Dvl1/2 and Wnt5a mutants is not due to reduced cell proliferation or survival, but may arise from defects in the deployment of SHF cells.
PCP signaling regulates polarized cell intercalation and directional cell migration during CE in Xenopus and zebrafish, but how might this pathway promote SHF deployment in the mouse? The expression of Wnt5a in the caudal SpM does not support Wnt5a functioning as a chemoattractant to guide directional cell migration. In contrast, our studies revealed that SHF cells in the caudal SpM of Wnt5a and Dvl1/2 mutants lack the normal protrusive morphology and display defective actin polymerization and filopodia formation, suggesting that PCP signaling may normally promote cell intercalation in this region. A recent study in mice suggests that as the SpM is recruited into the OFT rostrally, it is replenished by SHF progenitors caudally (Li et al., 2010). Our finding that wild-type cells in the caudal SpM display a highly protrusive morphology (Fig. 5M) supports this view. We hypothesize that Wnt5a-activated PCP signaling may induce protrusive activity in the caudal SpM to incorporate surrounding SHF progenitors into a cohesive, epithelial-like sheet. Rapid incorporation of progenitor cells at the caudal end may push the sheet rostrally to become recruited into the inferior wall of the OFT (Fig. 7A). In Wnt5a and Dvl1/2 mutants, cells in the caudal SpM form compact clusters rather than becoming rearranged into a sheet, and may in turn compromise the recruitment of the SpM rostrally into the OFT, leading to OFT shortening defects (Fig. 7B).
Our model has important implications towards our understanding of the biology of SHF development. Careful studies from Kirby’s group characterized the rostral SHF SpM adjacent to the OFT as “a pseudostratified columnar layer of epithelial cells” in the chick (Waldo et al., 2005b). We observed similar cell morphology in the mouse (Fig. 5G). How loosely packed mesenchymal cells in the SpM are converted to an epithelial sheet and how cells in this epithelial sheet are deployed to the OFT remained unknown. Our model in which PCP-mediated cell intercalation promotes a mesenchymal to epithelial conversion in the caudal SpM provides an answer to both questions. Our model is also consistent with chick vital dye labeling experiments which indicates that SpM is recruited into the OFT as a cohesive cohort instead of individually migrating cells (van den Berg et al., 2009).
Previous studies revealed that mutations in Vangl2 or Wnt11 also affect later OFT development during a process known as myocardialization, where cardiomyocytes lose their epithelial context and extend protrusions to invade the cushion mesenchyme and muscularize it(Phillips et al., 2005; Zhou et al., 2007). Therefore, PCP-induced protrusive cell behavior may be required at multiple stages of cardiogenesis.
In Xenopus, PCP signaling modulates actin organiztion and filopodia formation through the formin homology protein Daam1 (Disheveled Associated Activator of Morphogenesis 1), which links Dvl with actin binding protein Profilin in a Rho dependent fashion (Khadka et al., 2009; Tanegashima et al., 2008). Mice carrying a hypomorphic Daam1 mutation have been reported recently to display DORV and defective actin organization in the cardiomyocytes of the ventricles (Li et al., 2011). Whether Daam1 mutants have defects in the SHF SpM similar to those in Wnt5a and Dvl1/2 mutants needs to be determined in the future.
In summary, our studies provide novel and important insight into how SHF deployment to the OFT can be promoted by Wnt5a-activated PCP signaling in the caudal SpM. It is interesting to note that another non-canonical Wnt ligand, Wnt11, is expressed in the rostral pharyngeal region of the SHF and the OFT. Wnt11 expression in the OFT activates Tgfβ2 expression to regulate CNC and endocardial cell development (Zhou et al., 2007), but whether Wnt11 may also have an earlier role in activating PCP signaling in the pharyngeal mesoderm to promote SHF deployment from this rostal region needs to be addressed in the future. Finally, whether and how transcriptional regulation of Wnt5a/11 may place PCP signaling under the control of a global signaling network to coordinate SHF proliferation, differentiation and deployment are important and exciting questions that await future exploration. In this regard, a recent study has demonstrated that Wnt5a expression in the SHF progenitors in the caudal SpM is activated in part by Tbx1, a transcription factor critical for proper OFT formation (Chen et al., 2012). TBX1 mutations are known to cause DiGeorge syndrome in humans. Therefore, reduced SHF deployment due to a failure in activating Wnt5a-initiated PCP signaling may be an important pathogenic mechanism for the OFT malformations in DiGeorge syndrome patients.
(A–C) Anti-pHH3 staining revealed that mitotic nuclei (red signals indicated by yellow arrows in A-C) could be detected in SHF progenitors in the caudal SpM of transversely sectioned E9.5 wild-type (A), Dvl1−/−; Dvl2−/− (B) and Wnt5a−/− (C) mutant embryos. Quantification of the proliferation rate (the ratio of pHH3 positive nuclei to total nuclei) revealed no significant differences between the wild-type and Dvl1/2 and Wnt5a mutants (D). (E–G) Cell death was examined in transverse sections of the caudal SpM by anti-cleaved-caspase 3 staining. No apoptotic cells were observed in the SpM of E9.5 wild-type (E), Dvl1−/−; Dvl2−/− (F) or Wnt5a−/− (G) embryos. Few apoptotic cells were observed in the foregut endoderm in wild-type and Dvl1−/−; Dvl2−/− embryos (green signal indicated by red arrows in E&F) Nuclei were counterstained with DAPI (blue) in all the panels. FG: Foregut endoderm; SpM: splanchnic mesoderm.
We are grateful to Dr. Rosa Serra, Megan Cox and Ching-Fang Chang for their help with cryosectioning. This work was supported by NIH grant R01 HL109130, American Heart Association Grants 0635262N and 11GRNT6980004 and start-up funds from the University of Alabama at Birmingham to JW.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.