Research ArticleDevelopmental Biology

Wdpcp promotes epicardial EMT and epicardium-derived cell migration to facilitate coronary artery remodeling

See allHide authors and affiliations

Science Signaling  27 Feb 2018:
Vol. 11, Issue 519, eaah5770
DOI: 10.1126/scisignal.aah5770

Remodeling coronary arteries

During the development of the coronary vasculature, endothelial cells form the primitive coronary plexus, which then remodels by recruiting cells from the epicardial layer that ultimately give rise to smooth muscle cells. Liu et al. found that mice expressing a truncated form of the ciliogenesis gene Wdpcp formed a primitive coronary plexus, albeit at an accelerated rate. However, the remodeling of the coronary arteries in these mice was defective because of the impaired migration of epicardial cells. Understanding how coronary vasculature remodels during development may yield strategies to treat coronary artery disease, a leading global cause of mortality and morbidity.


During coronary vasculature development, endothelial cells enclose the embryonic heart to form the primitive coronary plexus. This structure is remodeled upon recruitment of epicardial cells that may undergo epithelial-mesenchymal transition (EMT) to enable migration and that give rise to smooth muscle cells. In mice expressing a loss-of-function mutant form of Wdpcp, a gene involved in ciliogenesis, the enclosure of the surface of the heart by the subepicardial coronary plexus was accelerated because of enhanced chemotactic responses to Shh. Coronary arteries, but not coronary veins in Wdpcp mutant mice, showed reduced smooth muscle cell coverage. In addition, Wdpcp mutant hearts had reduced expression of EMT and mesenchymal markers and had fewer epicardium-derived cells (EPDCs) that showed impaired migration. Epicardium-specific deletion of Wdpcp recapitulated the coronary artery defect of the Wdpcp mutant. Thus, Wdpcp promotes epithelial EMT and EPDC migration, processes that are required for remodeling of the coronary primitive plexus. The Wdpcp mutant mice will be a useful tool to dissect the molecular mechanisms that govern the remodeling of the primitive plexus during coronary development.


The coronary vasculature is essential to provide oxygen and nutrition to cardiac muscle. Congenital and acquired coronary diseases are major conditions impairing cardiac function and lead to substantial morbidity and mortality. An understanding of coronary development will potentially improve the prevention and treatment of these conditions. Two major components of coronary vessels, endothelial and smooth muscle cells, have different cellular origins in mammals (14). Coronary endothelial cells originating from the sinus venosus enclose embryonic hearts to form the primitive coronary plexus (5, 6). The primitive coronary plexus remodels by recruiting smooth muscle cells derived from the epicardial sheet to form the mature vascular hierarchy (7, 8).

Multiple signaling pathways, such as Wnt signaling (9), retinoic acid signaling (10), and fibroblast growth factor (FGF)/Hedgehog (HH)/vascular endothelial growth factor (VEGF)/angiopoietin 2 (ANG2) signaling (11), regulate coronary development as determined by studying mouse mutants with abnormal coronary development. These mouse mutants all have a primitive coronary plexus defect, which makes it difficult to study the mechanisms critical for the plexus remodeling stage of coronary development. Therefore, it is particularly rare to have animal models that can be used to dissect the regulatory mechanisms involved in primitive plexus remodeling.

Wdpcp, the mouse ortholog of Drosophila Fritz, is a component of the genetic module CPLANE (ciliogenesis and planar polarity effector) that is required for collective cell movement and ciliogenesis during embryonic development (1214). Loss of Wdpcp interrupts the development of cardiac outflow tract and leads to pulmonary atresia (12), but nothing is known about its role in coronary vascular development. Because of the ciliary defect and the role of Sonic hedgehog (Shh) signaling in coronary development (11), we examined the coronary development of Wdpcpm/m embryos [Wdpcpm is an allele with a splicing defect mutation that introduces premature stop codon and has been previously referred to as WdpcpCys40 (12)] and found defective primary ciliogenesis in cardiac cells and coronary artery defect in these mutants. However, instead of a defective coronary vascular plexus, we noted that the formation of primitive coronary plexus was accelerated in the hearts of the mutant mice. On the other hand, epicardial lineage tracing with Wt1CreERT2/Rosa26mTmG revealed a reduced number of epicardium-derived cells (EPDCs) due to defective epithelial-mesenchymal transition (EMT). EPDCs showed impaired infiltration into the myocardial wall in a three-dimensional (3D) gel invasion assay. In addition, epicardial deletion of Wdpcp produced the same coronary artery defect as the Wdpcpm/m mutation.


Wdpcp is required for coronary artery development

Wdpcp is an essential component for primary ciliogenesis (12) and should therefore be expressed in all the cells with primary cilia. To verify this assumption, we searched single-cell RNA sequencing analysis of embryonic heart and found that Wdpcp was expressed in epicardial cells, cardiomyocyte, endothelial cells, and fibroblasts of embryonic days 8.5 to 10.5 (E8.5 to E10.5) embryonic hearts (fig. S1, A and B) (15). Whole-mount RNA in situ staining also indicated that it was widely expressed in E13.5 heart (fig. S1, C and D).

To determine whether Wdpcp was essential for coronary vascular morphogenesis, we introduced the smooth muscle–specific SM22αLacZ reporter allele into Wdpcpm/m mutant mice to enable visualization of coronary vasculature with β-galactosidase–expressing smooth muscle cells by 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal) staining (16). In mouse, the coronary veins lay right under epicardial layer and were detectable after X-gal staining (arrowhead in Fig. 1A), whereas coronary arteries ran intramyocardially and were obscured by the surrounding myocardium (arrow in Fig. 1A). Wdpcpm/m mutants had various anomalies in coronary artery patterning. Coronary arteries, such as the left coronary artery (LCA), showed reduced X-gal staining in mutants (blue arrows in Fig. 1, B to D) and was discontinuous along the LCA path (red arrows in Fig. 1, B and C). Quantification of blue staining coverage over coronary vessels further confirmed significantly reduced smooth muscle cell coverage of mutant coronary arteries (Fig. 1E). The number of LCA branches was also significantly decreased (Fig. 1F), but we noted an additional short coronary artery on either side of the aorta, which originated either close to or far away from the main coronary artery (yellow arrows in Fig. 1, B to D). To examine the septal branch of the coronary artery (arrow in Fig. 1G), the heart was made translucent with benzyl benzoate and methyl salicylate treatment, and as expected, no discernible artery could be found in the ventricular septum (Fig. 1H). Coronary veins were much less affected in mutants (arrowheads in Fig. 1, B to D), and quantification indicated that there was no change in smooth muscle cell coverage between wild-type (WT) and mutant mice (Fig. 1E). Overall, these observations suggest that Wdpcp plays a critical role in the formation of coronary arteries.

Fig. 1 Coronary vascular defects in E18.5 Wdpcpm/m;SM22αLacZ/+ embryos.

(A to D) Coronary vessels with smooth muscle cells were visualized by X-gal staining in SM22αLacZ/+ hearts (A) and SM22αLacZ/+;Wdpcpm/m hearts (B to D). The blue arrows indicate the left coronary arteries (LCAs), which has reduced blue staining and few branches in (B) and (C). Red arrows point to the disrupted continuity of blue staining in the LCA. Yellow arrows indicate additional short coronary arteries originating from arterial trunk. Blue and yellow arrows in (D) indicate two coronary arteries of similar size. Black arrowheads indicate left coronary veins (LCVs). (E and F) Quantification of smooth muscle coverage in coronary arteries and veins (E) and LCA branch number (F). Ctrl, SM22αLacZ/+; Mutant, Wdpcpm/m;SM22αLacZ/+. (G and H) Transparentized hearts for visualization of the septal artery (SA). Data are presented as means ± SD (n = 3 embryos per genotype) and two-tailed Student’s t test on log-transformed data. **P < 0.01. Scale bar, 500 μm.

Wdpcp mutants have defective ciliogenesis but normal transcriptional responses to Shh signaling

We have previously reported that Wdpcp is required for primary ciliogenesis in the neural tube and kidney and thus is essential for Shh signaling (12). Hence, we performed immunostaining for acetylated α-tubulin and γ-tubulin to visualize primary cilia in E13.5 heart. As expected, coronary endothelial cells and cardiomyocytes from Wdpcpm/m mutants had short or no cilia (fig. S2, A to G). Because Shh signaling in cardiomyocytes transcriptionally activates the expression of mRNAs encoding different Vegf proteins, which subsequently promote the development of primitive coronary plexus (11), we examined the expression of Shh signaling components and targets, which were similar between mutant and WT hearts (fig. S2, H and I).

The formation of the subepicardial coronary plexus proceeds more quickly in the hearts of Wdpcpm/m mice

The coronary vasculature development during embryogenesis can be divided into two stages: the initial endothelial plexus formation, which is followed by remodeling of the preliminary plexus into mature coronary arteries. In mouse, the coronary endothelial cells originate from the sinus venosus (5). They spread around the atrioventricular groove and proceed from the dorsal to ventral side and caudally to wrap the whole ventricle (5, 6). Although the subepicardial endothelial plexus expands on the myocardial surface, it also invades the underlying myocardium to form the intramyocardial plexus (5, 6, 17). To gain insight into the potential mechanisms by which Wdpcp regulates the development of coronary vasculature, we first examined the formation of coronary plexus at E13.5 and E14.5 through whole-mount immunostaining for platelet endothelial cell adhesion molecule (PECAM). Instead of being impaired as we expected, the coronary plexus expanded faster in mutants. At E13.5, the ventral surface of ventricles was just beginning to be covered by coronary plexus (arrowhead in Fig. 2A), and the cardiac apex had yet to be covered by coronary plexus on the dorsal surface (arrow in Fig. 2B) in WT hearts, whereas 25% of the ventral surface and the entire dorsal side were already covered in mutant hearts (Fig. 2, C and D). By E14.5, the coronary plexus fully covered the dorsal surface of ventricles and expanded from both sides to cover 30% of the ventral surface in WT hearts (Fig. 2, E and F), whereas the coronary plexus covered more than two-thirds of the ventral surface and the entire dorsal surface in mutant hearts (Fig. 2, G and H). Furthermore, quantification of the percentage of the dorsal surface area positive for PECAM staining showed a slight increase in plexus density in both E13.5 (Fig. 2I) and E14.5 (Fig. 2J) mutant hearts.

Fig. 2 The subepicardial coronary plexus expands faster in Wdpcp mutants.

(A to H) Whole-mount PECAM immunostaining. (A to D) Embryonic day 13.5 (E13.5) heart. Arrowhead indicates the initial coronary plexus expanding from the dorsal side. (E to H) E14.5 heart. Arrow indicates area devoid of coronary plexus. (A, C, E, and G) Ventral view. (B, D, F, and H) Dorsal view. Dashed line in (B), (C), (E), and (G) indicates the front of the expanding subepicardial plexus. Scale bars, 200 μm. (I and J) Quantification of vascular coverage on the dorsal side of the ventricle. (I) E13.5 heart. (J) E14.5 heart. Data are presented as means ± SD (n = 3 embryos per genotype) and two-tailed Student’s t test on log-transformed data. *P < 0.05. WT, wild type; Mutant, Wdpcpm/m. (K) Knockdown efficiency of WDPCP short hairpin RNA–1 (shRNA-1) and shRNA-2 in human umbilical vein endothelial cells (HUVECs). (L) Quantification of cell migration through Transwell membranes. Data are presented as means ± SD (n = 3 biological replicates for each group) and two-tailed Student’s t test on log-transformed data. *P < 0.05 and ***P < 0.005. (M to R) Representative images for the Transwell migration assay. Scale bar, 200 μm. (M and P) HUVECs transfected with scrambled RNA, treated or not with SAG. (N and Q) HUVECs transfected with WDPCP shRNA-1, treated or not with SAG. (O and R) HUVECs transfected with WDPCP shRNA-2, treated or not with SAG.

Because coronary arteries arise from the remodeling of intramyocardial coronary plexus (18) and the coronary artery defect is more severe in the Wdpcpm/m mutant, we analyzed the intramyocardial plexus by staining for PECAM in transverse cryosections of E13.5 and E14.5 hearts. Again, we did not detect obvious defects in the formation of intramyocardial plexus of Wdpcpm/m mutant at either developmental stage (fig. S3, A to H). We quantified the number of endothelial cells in the myocardium, which were similar between WT and mutant hearts (fig. S3I). The growth of compact myocardium is an important developmental process that is interweaved with coronary artery development (19). Consistent with normal coronary plexus development, the compact myocardium of Wdpcpm/m and WT mutant mice had similar thicknesses (fig. S3J).

Faster formation of the subepicardial coronary plexus in the Wdpcpm/m mutant is due to accelerated endothelial cell migration

We first asked whether the faster progression of subepicardial coronary plexus formation was due to an early egression of coronary endothelial cells from sinus venosus. We collected embryonic hearts at E11.5, a stage when coronary endothelial cells start to sprout out from sinus venosus. PECAM immunostaining showed that at this stage, a thread of endothelial cells had sprouted from the sinus venosus and was migrating along the atrioventricular canal myocardium in both WT and Wdpcpm/m mutant hearts (fig. S4, A to D). In addition, we examined the proliferative and apoptotic indexes of coronary endothelial cells to address whether different endothelial cell turnover led to faster coronary plexus expansion. There were no obvious differences in endothelial cell proliferation or signs of apoptotic endothelial cells between WT and E13.5 mutant hearts (fig. S5, A to C).

These data indicate that neither early egression nor cell turnover caused the faster progression of subepicardial coronary plexus formation in the mutant hearts. Hence, we wondered whether Wdpcp deficiency accelerated endothelial cell migration. However, knockdown of WDPCP with two different lentivirally expressed short hairpin RNAs (shRNAs) (Fig. 2K) did not alter the migration of human umbilical vein endothelial cells (HUVECs) in vitro (Fig. 2, L to O). We were unable to verify WDPCP knockdown efficiency by immunoblotting using two different antibodies. Because epicardium-derived Shh is a critical signal regulating coronary plexus development (11), we treated HUVECs with Smoothened agonist (SAG), an agonist of the Shh receptor Smoothened (Smo), which accelerated the migration of WDPCP knockdown cells (Fig. 2, L and P to R). These results suggest that the faster progression of subepicardial coronary plexus in Wdpcpm/m mutants may be due to enhanced chemotaxis of endothelial cells in response to Shh.

Myocardium-specific deletion of Wdpcp does not affect coronary development

Although cardiomyocytes in Wdpcpm/m mutants have ciliary defects and Shh signaling in cardiomyocytes is essential for coronary plexus formation (11), Wdpcpm/m hearts, nonetheless, have normal transcriptional responses to Shh signaling and unimpaired formation of coronary plexus. However, it is unclear whether the ciliary defect in cardiomyocytes affects the remodeling stage of coronary development. To address this, we specifically deleted Wdpcp in the myocardium with cTnt-Cre (20), and the cTnt-Cre;SM22αLacZ/+;Wdpcpm/flox (T-Cko) mutants did not have defects in coronary development (fig. S6, A to D), leading us to investigate the role of Wdpcp in tissues besides the myocardium.

Wdpcp deficiency impairs epicardial EMT and EPDC migration

The remodeling of coronary plexus into mature coronary vessels involves the recruitment of coronary mural cells into the endothelial network. Smooth muscle cells are the main coronary mural component and are derived from epicardium through EMT (2, 3, 21). Because a key coronary vessel defect in Wdpcpm/m mutant is the reduced number of smooth muscle cells in coronary arteries, we postulated that defective epicardial EMT might be the mechanism underlying the coronary artery defects in Wdpcpm/m hearts. To address this notion, we used epicardium-specific Wt1CreERT2 and Rosa26mTmG reporters to trace EPDCs in the heart (10). Epithelial cells were labeled by peritoneal tamoxifen injection at E10.5. Flow cytometric analysis of dissociated E14.5 heart ventricles showed that the percentage of green fluorescent protein (GFP)–positive populations in mutant heart was significantly reduced compared with WT heart (Fig. 3, A to C), indicating a decreased number of EPDCs. Furthermore, there were fewer GFP+ cells in the myocardial wall of mutant hearts than in WT hearts at E15.5 (Fig. 3, D to F). Consistent with this finding, the expression of EMT markers (Snail2 and Twist1) and mesenchymal markers (Vinculin and Vimentin) were significantly reduced in E13.5 Wdpcpm/m mutant (Fig. 3G), indicating reduced EMT. In contrast, the expression of epithelial markers (β-catenin and E-cadherin) was not changed (Fig. 3G). Although the expression of the epithelial marker ZO-1 was also significantly decreased (Fig. 3G), it was due to reduced expression in cardiomyocytes (Fig. 3, H and I). We more frequently observed round cells in the epicardium of mutant heart (Fig. 3I), suggesting compromised cell migration after the first stage of EMT morphological transformation.

Fig. 3 Decreased EPDC numbers and compromised EMT in Wdpcpm/m hearts.

Epicardium-derived cells (EPDCs) express enhanced green fluorescent protein (EGFP) after CreERT2-mediated recombination in Rosa26mTmG locus upon tamoxifen injection at E10.5. (A and B) Representative fluorescence-activated cell sorting analysis of GFP+ cells from E14.5 Wt1CreERT2/+;Rosa26mTmG/+ and Wdpcpm/m;Wt1CreERT2/+;Rosa26mTmG/+ ventricles. SSC, side scatter. (C) Quantification of GFP+ cells. Data are presented as means ± SD (n = 3 embryos per genotype) and two-tailed Student’s t test on log-transformed data. **P < 0.01. (D and E) Coronary cryosections of E15.5 Wt1CreERT2/+;Rosa26mTmG/+ and Wdpcpm/m;Wt1CreERT2/+;Rosa26mTmG/+ hearts. White arrows indicate regions lacking EPDCs. Insets are magnified boxed areas showing EPDCs. RV, right ventricle; LV, left ventricle; IVS, intraventricular septum. Scale bar, 200 μm. (F) Percentage of ventricular myocardium covered by GFP+ cells. Data are presented as means ± SD (n = 3 embryos per genotype) and two-tailed Student’s t test on log-transformed data. *P < 0.05. (G) Quantitative polymerase chain reaction analysis of epithelial-mesenchymal transition (EMT), mesenchymal and epithelial markers. Data are presented as means ± SD (n = 5 embryos per genotype for Snail2 and Twist1 and n = 3 embryos per genotype for the other markers) and two-tailed Student’s t test on log-transformed data. *P < 0.05 and ***P < 0.005. (H and I) Representative images of ZO-1 immunofluorescence staining on E13.5 embryos (n = 2 embryos per genotype). White arrows point out similar ZO-1 immunostaining in epicardial cells. White arrowheads point out decreased ZO-1 immunostaining in cardiomyocytes. Yellow arrows point out different epicardial cell shapes between WT and mutant hearts. DAPI, 4′,6-diamidino-2-phenylindole.

In WT ventricle, EPDCs occupied nearly the whole compact layer of the myocardial wall, and although EPDCs distributed in a similar pattern in mutant ventricle, they were less infiltrated into the myocardium in some areas (arrows in Fig. 3E), further suggesting compromised EPDC migration. A 3D collagen gel invasion assay using E11.5 heart ventricles showed that epicardial outgrowths from mutant hearts covered less area on the x/y plane (Fig. 4, A and B) and occupied less depth along the z axis (Fig. 4, C to E) compared to those from WT hearts. Similarly, infection with two different lentivirally delivered shRNAs to knock down Wdpcp impaired the capability of EPDCs to infiltrate collagen gels (fig. S7, A to E).

Fig. 4 Compromised invasion by Wdpcp mutant-derived EPDCs.

E11.5 epicardial cells were explanted on a three-dimensional collagen gel matrix, and outgrowths were stained with Alexa Fluor 488–conjugated phalloidin and DAPI (blue fluorescence) to enable visualization of epicardial cell invasion by confocal microscopy. (A and B) x/y plane of representative epicardial outgrowths from WT explants (A) and Wdpcp mutant explants (B). (C and D) Confocal imaging of epicardial outgrowth along the z axis. (E) Depth measurement of epicardial cell invasion along the z axis. Cell infiltration depth along the z axis was determined by the distance between the top and bottom planes within which there were 30 cells. Data are presented as means ± SD (n = 5 embryos per genotype) and two-tailed Student’s t test on log-transformed data. ***P < 0.005. Scale bar, 100 μm.

Epicardium-specific deletion of Wdpcp leads to abnormal formation of coronary artery

We wondered whether epicardium-specific deletion of Wdpcp could recapitulate the coronary defects of the Wdpcpm/m mutant. WT1CreERT2 and SM22αLacZ alleles were bred into Wdpcpm/flox embryos for epicardial deletion and visualization of coronary vessel, respectively. In the three E18.5 Wdpcpm/flox;WT1CreERT2/+;SM22αLacZ/+ (W-Cko) mutants we collected, all of them had normal outflow tract septation but recapitulated, with various severities, the coronary artery defects of Wdpcpm/m mutants (Fig. 5, A to D). In contrast to the LCA in control mice (Fig. 5A), the LCA in W-Cko mutants did not have an ordered hierarchical structure (Fig. 5B), had reduced branching (Fig. 5C), or was severely truncated (Fig. 5D). Similar to Wdpcpm/m mutants, the W-Cko mutant also had reduced smooth muscle coverage of coronary arteries but not of coronary veins (Fig. 5, B to D). Unlike Wdpcpm/m mutants, only one W-Cko mutant had an additional short coronary artery with an independent origin adjacent to the LCA (Fig. 5C). The low frequency of coronary stem defects may be due to normal outflow tract development in W-Cko mutants. Examination of histological sections showed that the coronary arteries of W-Cko mutants were patent and connected to the aorta (Fig. 5, E to L), which suggests that the remodeling defect of coronary arteries is not caused by lack of blood flow.

Fig. 5 Epicardium-specific deletion of Wdpcp leads to coronary artery defects.

(A to D) Left side view of E18.5 hearts after X-gal staining in Wdpcpm/+;SM22αLacZ/+ heart (A) and Wdpcpm/flox;WT1CreERT2/+;SM22αLacZ/+ mutant hearts with tamoxifen injected at E10.5 (B to D). Black arrows indicate the LCA. Black arrowheads indicate the LCV. Yellow arrow indicates an additional coronary artery originating from the aorta. Red arrow indicates discontinuity of blue staining in LCA. Scale bar, 500 μm. (E to L) Representative coronal sections of X-gal–stained Wdpcpm/flox;WT1CreERT2/+;SM22αLacZ/+ hearts (n = 2 mice). (E to H) Serial sections from dorsal to ventral side show patent LCA originating from the aorta (AO). (I to L) Serial sections from ventral to dorsal side show patent right coronary artery originating from the aorta. PA, pulmonary trunk. Scale bar, 200 μm.


In the mouse, endothelial cells of primitive coronary plexus mainly originate from sinus venosus (5, 6). The forming subepicardial endothelial plexus encloses ventricles from the dorsal atrioventricular groove and concurrently invades myocardial free wall to form intramyocardial plexus (5, 6). The formation of subepicardial plexus is dependent on Shh signaling in cardiomyocytes to produce proangiogenic cytokines (11). The planar cell polarity (PCP) effector gene Wdpcp is essential for ciliogenesis and Shh signaling transduction (12, 13). Many Wdpcpm/m phenotypes, including pulmonary atresia, polydactyly, and tracheoesophageal fistula, reflect the disruption of Shh signaling (12). Given the importance of ciliary regulation in Shh signaling (22), we examined the formation of coronary vascular network in Wdpcpm/m mutants. We observed structural defects in major coronary arteries. However, the initial formation of the primitive coronary plexus was normal, and the formation of the subepicardial coronary plexus was accelerated. Consistently, we did not find changes in the expression of mRNAs encoding Shh signaling components and targets at the same stage.

The working model for Shh signaling transduction is that the binding of Shh to Ptch1 relieves its repression on Smo, increases Smo localization to the primary cilia, and then initiates Gli-dependent transcriptional response through cilia (23, 24). The defective primary ciliogenesis in Wdpcpm/m heart that we observed conflicts with this mutant having normal Shh signaling transcriptional responses and lacking primitive coronary plexus defects typical of Shh mutants. To address this issue, we propose that transcriptional response of Shh signaling in cardiomyocyte is not or less dependent on primary cilia. Primary cilium–independent Shh signaling exists in human mammary epithelial cells (25), and primary cilium–independent regulation of Gli protein through Sufu is a conserved Shh signaling mechanism (26, 27). Hence, it is possible that embryonic cardiomyocytes develop a specialized Shh signaling mechanism that is independent of primary cilia, the details of which remain to be resolved.

The other interesting question is why the formation of subepicardial plexus proceeds more quickly in mutant than WT hearts. A noncanonical Shh signaling of chemotactic response is required for multiple motility processes, such as the migration of fibroblasts, neuroblasts and cholangiocarcinoma cells (2831), neuronal axon extension (3234), and neurite projection (35). In particular, it can promote in vitro tubulogenesis of endothelial cells and may play a role in angiogenesis during liver cirrhosis (36, 37). The noncanonical Shh signaling pathway modulates actin cytoskeleton reorganization through diverse Smo-dependent but Gli-independent mechanisms to regulate cell migration (38). The requirement of Smo in both Shh canonical and noncanonical pathway indicates a mutually competitive inhibition. Cells with defective primary cilium–locating Smo or cells with defective ciliogenesis show decreased canonical transcriptional signaling and enhanced noncanonical chemotactic responsiveness (39, 40). Therefore, it is possible that the primary ciliogenesis defect in coronary endothelial cells of Wdpcpm/m mutants causes increased distribution of Smo into subcellular sites, where chemotactic signaling can be transduced, and increased chemotactic response, thereby leading to faster progression of subepicardial plexus in Wdpcp mutant. Wdpcp knockdown alone did not affect the migration of HUVECs but accelerated migration under Smo stimulation. These data support the notion that the faster proceeding of coronary plexus is caused by an enhanced chemotactic effect of noncanonical Shh signaling due to defective primary ciliogenesis.

Remodeling of primitive coronary plexus into mature vasculature hierarchy involves the transformation of epicardial cells through EMT and the migration of generated EPDCs. Multiple signaling pathways, such as Wnt (9), FGF (41), and HH (42) signaling, had been previously found to be involved in the regulation of coronary development through the epicardium but later were found to be dispensable in the epicardium for regulating coronary development in another study (43). Therefore, much need to be understood for the regulation of this process. Having excluded a primitive coronary plexus defect in the Wdpcpm/m mutant, we examined whether the remodeling of primitive plexus was affected. Conditional deletion revealed that Wdpcp deficiency in epicardium caused primitive plexus remodeling defects due to impaired EMT and EPDC migration, which is consistent with the role of Wdpcp in modulating the actin cytoskeleton and regulating directional cell migration (12, 13, 44, 45). Obviously, both cellular defects would have a substantially greater impact on the number of intramyocardial EPDCs than on the number of subepicardial EPDCs, which may explain why the formation of subepicardial vein was less affected than that of the intramyocardial artery. The relatively wide phenotypic variation of the W-Cko mutant is likely due to different tamoxifen induction efficiency (10, 46). Notably, the coronary artery defect of one W-Cko mutant is even more severe than that of the Wdpcpm/m mutant, suggesting that an acute deletion of Wdpcp may have a more profound effect on EMT and EPDC migration. It is consistent with the impaired collagen gel filtration by EPDCs derived from explants with Wdpcp knockdown.

Proper and timely formation of the coronary stem is essential for coronary development, especially for blood flow–triggered remodeling of the primitive coronary plexus (47, 48). In W-Cko mutants, we noted that all the coronary arteries were correctly connected and patent except one case that had an additional short coronary artery from one side of the aorta. Nonetheless, all mutants had coronary artery defect similar to Wdpcpm/m mutants. These results suggest that the coronary artery remodeling defect in Wdpcpm/m mutants did not result from insufficient coronary blood flow.

In summary, we found an in vivo developmental process that is subject to the regulation of noncanonical Shh signaling and present a mouse model that has coronary plexus remodeling defect but has no defect in the formation of primitive plexus. It will be an excellent model to study the molecular mechanisms that specifically regulate the primitive plexus remodeling process during coronary development.


Experimental animals

Animal care and use in all the experiments were in accordance with the guidelines approved by the Institutional Animal Care and Use Committee of Shanghai Children’s Medical Center. The mouse lines used in the current study have been described previously, such as Wdpcpm/m (12), Wdpcpflox (12), SM22αLacZ/+ (16), Wt1CreERT2/+ (10), Rosa26mTmG/+ (49), and cTnt-Cre (20). All mice were maintained in C57BL6/J background since obtained. To induce nuclear translocation of CreERT2, pregnant female mice at day 10.5 postcoitum were given one dose of tamoxifen (75 mg/kg body weight) (T5648, Sigma-Aldrich) that was dissolved in corn oil through oral gavage.

X-gal staining

Hearts were dissected out in cold phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde (PFA) at 4°C for 30 min. After washing in PBS, tissues were equilibrated in PBS containing 0.1% NP-40 and 0.1% sodium deoxycholate for 10 min, followed by incubation with X-gal (1 mg/ml) at 37°C overnight. The next day, tissues were rinsed in prewarmed PBS and fixed in 4% PFA. Samples were photographed under stereomicroscope (M205 FA, Leica) and stored in 70% ethanol. Next, the hearts were paraffin-embedded, sectioned, and counterstained by nuclear fast red. The branch number of coronary arteries was counted directly. Smooth muscle cell coverage of coronary artery was analyzed by Image-Pro Plus 6. The contour of coronary vessel was outlined by using “trace-Auto” tool or manual drawing (for weakly stained or unstained portions). The “measure-count/size” tool was used to calculate the percentage of blue pixel coverage along coronary vessel, and the result was defined as smooth muscle cell coverage.


For whole-mount PECAM staining, samples were fixed in methanol/dimethyl sulfoxide (DMSO) (4:1) overnight at 4°C and then bleached in methanol/DMSO/30% hydrogen peroxide (4:1:1) for 4 hours at room temperature. After rehydration through 50% methanol, 15% methanol, and PBT.3 (PBS containing 0.3% Triton X-100), samples were incubated with rat anti-mouse PECAM-1 (550274, BD Biosciences) at 1:100 dilution in PBT.3+ (PBS containing 0.3% Triton X-100 and 2% blocking reagent) at 4°C overnight. Next, samples were washed twice in PBT.3 and twice in PBT.5 (PBS containing 0.5% Triton X-100) for 30 min each. After washing, horseradish peroxidase (HRP)–conjugated goat anti-rat immunoglobulin G (IgG) at 1:100 dilution was incubated with samples, followed by color reactions with 3,3-diaminobenzidine (DAB) kit (08G01A22, Boster). Images were taken using a stereomicroscope (M205 FA, Leica). For each sample, five different regions with the same area size on the dorsal side were randomly selected to assess the coverage of coronary plexus. The measure-count/size tool of Image-Pro Plus 6 was used to calculate the percentage of dark brown pixel coverage.

For cryosection immunohistochemistry, sections were briefly fixed in cold acetone for 3 min. After washing in PBS, sections were incubated in 3% hydrogen peroxide to quench endogenous peroxidase. After blocking in 10% goat serum/PBS for 1 hour, sections were incubated with the same PECAM antibody at 1:100 dilution in blocking solution at 4°C overnight. The next day, sections were incubated with secondary antibody HRP-conjugated goat anti-rat IgG at 1:100 dilution. After DAB staining, sections were counterstained with hematoxylin. To quantify the thickness of the compact layer, we selected sections at similar positions from each heart. For each section, we measured the thickness at four similar locations where the trabecular myocardium segregated from compact layer clearly. For each location, the thickness was obtained by averaging the measurements from three neighboring positions. The same sections were used to quantify the number of endothelial cells within the compact layer by direct counting, and the area of myocardial compact layer was measured by the measure-count/size tool of Image-Pro Plus 6.

For immunofluorescence staining, sections were postfixed in 4% PFA, permeabilized in PBS with 0.25% Triton-100 (PBST), and then blocked in 1% bovine serum albumin/PBST for 1 hour. Sections were incubated with the following antibodies in tandem: primary antibodies: anti-PECAM (1:100; 28364, Abcam), anti–vascular endothelial (VE)–cadherin (1:1000; AF1002, R&D Systems), anti-Ki67 (1:200; 15580, Abcam), anti–γ-tubulin (1:500; T6557, Sigma-Aldrich), anti–acetylated tubulin (1:500; T7451, Sigma-Aldrich), and anti–ZO-1(1:100; 40-2300, Thermo Fisher Scientific); secondary antibodies: goat anti-rabbit IgG Cy3 (1:200; 6939, Abcam), goat anti-rabbit IgG 633 (1:500; A-21070, Thermo Fisher Scientific), goat anti-mouse IgG1 488 (1:500; A-21121, Thermo Fisher Scientific), goat anti-mouse IgG2b 594 (1:500; A-21145, Thermo Fisher Scientific), donkey anti-goat IgG (1:500; A-11055, Thermo Fisher Scientific), goat anti-rabbit IgG (1:500; A-21429, Thermo Fisher Scientific), and donkey anti-mouse IgG 488 (1:500; A-21202, Thermo Fisher Scientific). Terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining was performed with Roche kit 7791-13-1 according to the manufacturer’s instructions. The number of ciliated endothelial cells was obtained by counting PECAM+acetylated α-tubulin+γ-tubulin+ cells. The number of ciliated cardiomyocytes was obtained by counting cTnT+acetylated α-tubulin+γ-tubulin+ cells. The proliferative index was calculated as the ratio of Ki67+VE-cadherin+ cells to total VE-cadherin+ cells. The ratio of TUNEL+VE-cadherin+ to total VE-cadherin+ cells was scored as the percentage of apoptotic endothelial cells. Five different sections at similar positions from each heart were used for the quantitative analyses of cilium number, proliferative and apoptotic cells.

Lentiviral shRNA knockdown

To knock down WDPCP (GenBank no. NM_015910) in HUVECs and Wdpcp (GenBank no. NM_145425) in mouse cardiac explants, we purchased two different lentiviral shRNA constructs for each gene from Shanghai Genechem Co. Ltd. For human WDPCP shRNA-1 and shRNA-2, the shRNA targeting sequences are 5′-TGGAGTCTTCTGATGTAAA-3′ and 5′-TATTCATTACCTTGCACTA-3′, respectively. For mouse Wdpcp shRNA-1 and shRNA-2, the shRNA targeting sequences are 5′-CTTCTTCCTTCCATTGGTT-3′ and 5′-GAATCACGAGATTATCCTT-3′, respectively. The sequence of scrambled shRNA is 5′-TTCTCCGAACGTGTCACGT-3′. Human WDPCP shRNA and mouse Wdpcp shRNA fragments were subcloned into GV248 (hU6-MCS-Ubiquitin-EGFP-IRES-puromycin) and GV113 (hU6-MCS-CMV-RFP) plasmids, respectively. Lentiviral particles were produced by transfecting human embryonic kidney (HEK) 293FT cells with shRNA, psPAX2 (#12260, Addgene), and pMD2.G (#12259, Addgene) plasmids. Knockdown efficiency was determined by quantitative polymerase chain reaction (qPCR).

Transwell migration assay

HUVECs were transduced with prefiltered lentiviral shRNA and treated with polybrene (8 μg/ml; 107689, Sigma-Aldrich). After 24 hours, HUVECs were digested and subject to fluorescence-activated cell sorting (FACS) to isolate enhanced GFP–positive (EGFP+) cells. A total of 5 × 103 cells were placed on top of the filter membrane of a Transwell insert with 8-μm pores (CLS3464, Costar). The 0.6-ml chemoattractant (M10 plus 100-nm SAG) was carefully added into the bottom of the lower chamber in a 24-well plate. After 12 hours, the Transwell insert was taken out. Nonmigrated cells and media on the upper side of the filter were removed with a Q-tip without damaging the membrane and cells on the lower side. Membranes were fixed with 70% ethanol for 10 min and stained with 0.2% crystal violet for 10 min at room temperature. Images were captured using Leica DM6000 B microscope, and the number of cells on the lower surface was analyzed with the measure-count/size tool of Image-Pro Plus 6.

EPDC lineage tracing

Wdpcpm/+;Wt1CreERT2/+ and Wdpcpm/+;Rosa26mTmG mice were crossed to generate Wdpcpm/m;Wt1CreERT2/+;Rosa26mTmG embryos. At day 10.5 postcoitum, pregnant females were given tamoxifen (75 mg/kg body weight) by gavage to induce Cre-mediated recombination. Embryos were collected at E15.5. Hearts were dissected out and briefly fixed in 4% PFA. Samples were then embedded in optimal cutting temperature (OCT) and sectioned into 10-μm section on a Leica Cryostat. Sections were again fixed briefly in 4% PFA and washed in PBS. Next, sections were mounted and image-captured. Quantitative analysis was carried out with sections at similar positions from each heart. GFP+ cells were counted under microscope, and the area of myocardial compact layer was measured with measure-count/size tool of Image-Pro Plus 6.

Flow cytometric analysis

E14.5 Wt1CreERT2/+;Rosa26mTmG and Wdpcpm/m;Wt1CreERT2/+;Rosa26mTmG embryonic heart ventricles were isolated by microdissection in cold Hanks’ balanced salt solution (HBSS). Single-cell suspension of epicardium and EPDCs were prepared as previously described (50). In brief, 2-ml cold digestion solution containing 1% collagenase IV (1463110, Gibco), 0.05% trypsin, and 1% chicken serum was used to digest heart ventricle for 6 to 7 min in 37°C shaker. Solution was pipetted up and down three times, and the tube was allowed to stand for 15 s. Supernatant was transferred into a 50-ml falcon tube and kept cold on ice, and 0.25-ml horse serum was added to neutralize the digestion solution. These steps were repeated for seven to eight times, and supernatants were collected and filtered through a 70-μm cell strainer. Cells were centrifuged at 200g for 5 min at 4°C and resuspended in 0.25-ml HBSS for FACS analysis.

3D collagen gel invasion assay

Collagen gels were prepared as previously reported with minor modification (51). In brief, type I rat tail collagen (354249, BD Biosciences) (final concentration, 1.2%), 10× M199 (Gibco), and 2.2% sodium bicarbonate (final concentration, 0.22%) were mixed with sterile water. The mixture was placed in a four-well dish (144444, Nalge Nunc) and left in a cell culture incubator to solidify for 2 hours. Gels were washed with Dulbecco’s modified Eagle’s medium (DMEM) three times for 15 min each. Then, the gels were conditioned overnight in DMEM with 10% fetal bovine serum (FBS), 1% insulin-transferrin–selenium-X (51500-056, Gibco), and penicillin/streptomycin. E11.5 hearts were collected and coronally cut into halves. Explants were placed on gels with the epicardial face down and cultured on gels with DMEM containing 10% FBS for 2 days and then were removed from gels. Gels were cultured for an additional 2 days. Gels were washed with PBS twice, fixed in 4% PFA, permeabilized with 0.5% Triton X-100, and stained with Alexa Fluor 488–conjugated phalloidin and 4′,6-diamidino-2-phenylindole. Explant outgrowths were photographed under a Leica SP8 confocal microscope. Invasion depth was determined by measuring the distance between the top and bottom planes with 30 cells using Image-Pro Plus 6. The assay for WT explants transfected with lentiviral shRNA was similar, except that initial transfection with lentivirus for 12 hours before putting on gel. To measure knockdown cell invasion depth, only cells transfected with lentivirus (red fluorescence) was analyzed by Image-Pro Plus 6.

Quantitative real-time PCR analysis

Total RNA was isolated from E13.5 embryonic heart and reverse-transcribed into complementary DNA according to the standard procedures. Real-time polymerase chain reaction (PCR) was performed using an Applied Biosystems 7900 real-time PCR instrument with the primers shown in table S1.


Fig. S1. Expression of Wdpcp in embryonic heart.

Fig. S2. Ciliary defects but normal transcriptional response to Shh signaling in Wdpcp mutant hearts.

Fig. S3. Normal formation of the intramyocardial endothelial plexus.

Fig. S4. Normal egression of coronary endothelial cells from the sinus venosus.

Fig. S5. Proliferative and apoptotic status of coronary endothelial cells in E13.5 hearts.

Fig. S6. Normal coronary vessel formation in E18.5 embryos with myocardium-specific deletion of Wdpcp.

Fig. S7. Wdpcp knockdown recapitulates the gel infiltration defect of Wdpcp mutant-derived EPDCs.

Table S1. List of qPCR primers used in the study.


Acknowledgments: We thank A. Baldini (Institute of Genetics and Biophysics, Naples, Italy) for helpful discussion. Funding: The work was supported by grants 2013CB945302 from the Ministry of Science and Technology, 31371465 and 31771612 from the National Natural Science Foundation of China, 20171925 from Shanghai Municipal Education Commission–Gaofeng Clinical Medicine Grant Support, XBR2015 from Shanghai Municipal Commission of Health and Family Planning, and the Program for Professor of Special Appointment (Eastern Scholar) at Shanghai Institutions of Higher Learning. Author contributions: X.L., Y.W., and Z.H. designed and performed experiments and analyzed data. F.L. performed part immunostaining and qPCR experiments. M.Z. analyzed single-cell sequencing data. H.S. provided mouse husbandry support and did mouse genotyping. B.Z. and C.W.L. provided suggestions and mouse lines. Z.Z. designed and analyzed data. S.T. provided suggestions on statistical analysis. X.L., Y.W., and Z.Z. wrote the paper. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. The Wdpcpm/m mutant mice require a material transfer agreement from the National, Heart, Lung, and Blood Institute.

Stay Connected to Science Signaling

Navigate This Article