β-Catenin promotes cell proliferation and migration during endocardial cushion formation

We crossed floxed β-catenin mice with Tie2^Cre mice to generate the endothelial-specific β-catenin knockout mice (β-catenin^f/f:Tie2^Cre, referred as β-cat^eKO hereafter). Immunostaining confirmed that β-catenin was expressed in both endocardium and myocardium of E9.5 control (β-catenin^f/f or β-catenin^f/+) embryos, but its expression in endocardium was selectively disrupted in their littermate β-cat^eKO embryos (Fig 1A). β-cat^eKO embryos were underdeveloped compared with their littermate controls at E10.5, although they were grossly normal and indistinguishable from controls at E9.5 (Fig S1). HE staining showed that AVC cushions of E9.5 control embryos were being populated by mesenchymal cells, indicating proper formation of endocardial cushions (Fig 1B). In contrast, AVC cushions of β-cat^eKO embryos were hypocellular with fewer mesenchymal cells (Fig 1B and C). Tie2^Cre also deletes genes at the OFT region where EndoMT starts around E9.5, slightly later than the beginning time of EndoMT in the AVC region (Kisanuki et al, 2001). Indeed, we confirmed that there were very few, if any, mesenchymal cells within OFT cushions of either control or β-cat^eKO embryos at E9.5 (Fig S2). EdU labeling showed significantly reduced cell proliferation in endocardium, mesenchyme, and myocardium of β-cat^eKO embryos compared with controls (Fig 1D and E). By comparing the histology of AVC cushions between control and β-cat^eKO embryos, we found that the mesenchymal cells in cushions of control embryos migrated far away from the endocardium and invaded into matrix-rich cushions (Fig 1B). In contrast, mesenchymal cells in the cushion of β-cat^eKO embryos failed to migrate away from the endocardium; instead, they were clustered underneath the endocardium (Fig 1B). Quantification of the cell distribution showed that 40% of mesenchymal cells in control embryos were located within the sub-endocardium region, although this number was greatly increased to 80% in β-cat^eKO embryos (Fig 1F), indicating a migratory defect.

We then characterized this migratory defect by performing an EndoMT assay on collagen gel using AVC tissue explants isolated from hearts of E9.5 embryos. Consistent with in vivo results, EndoMT assay showed that fewer cells migrated out from the explants of β-cat^eKO embryos when compared with that of control explants (Fig 2A and B). Live-cell tracking showed that mesenchymal cells of control explants migrated far away from the explants whereas those cells of β-cat^eKO explants stayed around the explants (Fig 2C). In addition, mesenchymal cells of β-cat^eKO explants exhibited significantly reduced migration velocity and mean square displacement (Fig 2D and E), further suggesting a migratory defect. In contrast, the migration directionality of mesenchymal cells was comparable between control and β-cat^eKO explants (Fig 2F). These findings support that β-catenin is required for cell proliferation and migration during endocardial cushion formation.

β-Catenin is required for filopodia formation

β-Catenin forms a complex with adenomatous polyposis coli at membrane protrusions and modulate mammary tumor cell migration and mesenchymal phenotype (Odenwald et al, 2013). N-terminally phosphorylated β-catenin (phospho-β-catenin) was shown to accumulate at the leading edge of migrating cells (Faux et al, 2010). Consistently, we found that phospho-β-catenin (Thr41/Ser45) was highly expressed in the protrusions (filopodia/lamellipodia) of transforming endocardial cells and transformed mesenchymal cells in control embryos, although such expression was dramatically disrupted in β-cat^eKO embryos (Fig 3A and B). The expression pattern of β-catenin suggests that it may be involved in regulation of filopodia formation. To test this hypothesis, we performed whole-mount staining of isolectin B4 (IB4) and αSMA. The results showed that endocardial and mesenchymal cells at the AVC cushions of E9.5 control embryos exhibited prominent protrusions toward the myocardium (Fig 3C and Video 1). In contrast, very few cells in AVC of β-cat^eKO embryos presented such protrusions (Fig 3C and Video 2). Consistently, staining of F-actin with phalloidin showed that endocardial and mesenchymal cells in AVC of E9.5 control embryos formed microspikes (filopodia), whereas very few cells in AVC of β-cat^eKO embryos had this feature (Fig 3D). Together, these findings indicate that β-catenin is essential for filopodia formation during endocardial cushion formation.

β-Catenin regulates cell proliferation transcriptionally and migration non-transcriptionally

β-Catenin is a dual functional protein–mediating cell–cell adhesion at cell membrane and transcriptional regulation in the nucleus (Valenta et al, 2011). We used a β-catenin^DM allele to disrupt selectively the transcriptional function of β-catenin (Valenta et al, 2011) and generated a compound β-catenin mutant mice (β-catenin^f/DM:Tie2^Cre, referred as β-cat^eKO/DM hereafter) in which endocardial cells and their mesenchymal progenies completely lost the transcriptional function of β-catenin, whereas maintained one copy of β-catenin with the cell adhesive function. We then analyzed the AVC cushion development in control, β-cat^eKO/eKO (disruption of both functions of β-catenin in endocardial cells and their progenies) and β-cat^eKO/DM embryos. HE staining showed that both β-cat^eKO/eKO and β-cat^eKO/DM embryos had hypocellular AVC cushions (Fig 4A and B) and decreased cell proliferation (Fig 4D and E), indicating that transcriptional function of β-catenin regulates cell proliferation. In contrast, many mesenchymal cells in AVC cushions of control and β-cat^eKO/DM embryos migrated far away from the endocardium, whereas those cells of β-cat^eKO/eKO embryos failed to do so (Fig 4A and C). Consistently, whole-mount staining of IB4 showed that mesenchymal cells in control and β-cat^eKO/DM embryos had prominent filopodia, whereas few, if any, cells in β-cat^eKO/eKO embryos acquired this structure (Fig 4F and Video 3, Video 4, and Video 5), confirming that the non-transcriptional function of β-catenin regulates filopodia formation and cell migration. These findings support that transcriptional and non-transcriptional function of β-catenin promotes cell proliferation and migration, respectively, during endocardial cushion formation.

β-Catenin supports cell proliferation by suppressing p21

To determine the molecular mediators through which β-catenin regulates cell proliferation and migration, we examined the expression of candidate genes that are known to be involved in early endocardial cushion formation. The AVC tissues were microdissected from E9.5 control, β-cat^eKO/eKO, and β-cat^eKO/DM embryos and used for gene expression analysis by quantitative PCR (qRT-PCR). The qRT-PCR results showed that the expression of Notch1, Hey1, Hey2, Bmp2, Tgfbr2, Bmpr1a, and Bmpr2, which are involved in the NOTCH and BMP signals essential for EndoMT (Garside et al, 2013; Wang et al, 2021), was not affected in either β-cat^eKO/eKO or β-cat^eKO/DM embryos (Fig S3). We confirmed by immunostaining that active NOTCH1 (N1ICD) and BMP downstream effector pSMAD1/5/9 were not affected in both mutant embryos (Fig S4). The expression of Nrg1, which regulates EndoMT during endocardial cushion formation through the NRG1/ErBB pathway (Armstrong & Bischoff, 2004), was not affected in β-cat^eKO/eKO embryos but significantly up-regulated in β-cat^eKO/DM embryos (Fig S3). The expression of Erbb3 was significantly reduced in both mutant embryos, whereas the expression of Erbb2 and Erbb4 was not changed in either group (Fig S3). In addition, the ID genes (Id1, Id2, and Id3), known for their functions in EndoMT (Fraidenraich et al, 2004; Kowanetz et al, 2004), were not affected in either β-cat^eKO/eKO or β-cat^eKO/DM embryos (Fig S3). As matrix protein dynamics are essential for endocardial cushion formation (Hinton et al, 2006), we also examined the expression of matrix genes (Has2, Acan, and Vcan) and found their expression was not altered in either mutant (Fig S3). In contrast, the qRT-PCR results showed that the expression of Snail, Slug, Twist1, and Msx1, which critically control EndoMT during endocardial cushion formation (Wirrig & Yutzey, 2011), was markedly reduced in both mutant embryos, whereas the expression of Msx2 was not changed in either mutant (Fig S3).

Consistent with the phenotype of reduced proliferation, cell cycle inhibitor p21 was dramatically increased in both β-cat^eKO/eKO and β-cat^eKO/DM embryos (Fig 5A). RNAscope and immunostaining confirmed that the number of p21-expressing endocardial or mesenchymal cells in the AVC region was dramatically increased in both mutant embryos when compared with that in control embryos (Fig 5B–D). In line with the in vivo findings, β-catenin inhibition by XAV939 significantly up-regulated the expression of p21 in HUVECs and pig aortic valve interstitial cells (PAVIC) (Fig 5E and F). Together, these results indicate that β-catenin promote cell proliferation likely by suppressing p21 expression.

We then performed rescue experiments to determine whether the elevated p21 expression contributed to the cell proliferation defect caused by β-catenin deficiency. β-Catenin knockdown dramatically up-regulated p21 expression and inhibited cell proliferation in HUVEC, and p21 knockdown rescued the cell proliferation defect (Fig 6A and C). Similarly, β-catenin inhibition with XAV939 markedly repressed the proliferation of PAVIC, which was partially restored by p21 inhibition with UC2288 (Fig 6B and D). Taken together, these results demonstrate that p21 suppression by β-catenin is essential for cell proliferation and proper endocardial cushion formation.

β-Catenin is dispensable for endocardial-to-mesenchymal fate change

Endocardial cushions formation begins with an EndoMT process in which some cushion endocardial cells gradually loose the endothelial marker VE-cadherin and gain the mesenchymal markers (αSMA, VIMENTIN, PDGFRβ) (Armstrong & Bischoff, 2004). Down-regulation of VE-cadherin is prerequisite for transformed endocardial cells to delaminate from the endocardial sheet and migrate into the matrix-rich cushions (Timmerman et al, 2004). Immunostaining showed that VE-cadherin protein in AVC endocardium was significantly increased in β-cat^eKO/eKO embryos but not in β-cat^eKO/DM embryos when compared with that in controls, although it in ventricular endocardium was comparable among three groups (Fig S5A). This result indicates that β-catenin negatively regulates the protein level of VE-cadherin via its non-transcriptional function. In support of this notion, RNAscope and qRT-PCR results showed that the mRNA level of VE-cadherin in cushion endocardial cells was comparable among control, β-cat^eKO/eKO, and β-cat^eKO/DM embryos (Fig S5B and C). On the other hand, immunostaining showed that the cushion endocardial cells in control, β-cat^eKO/eKO, and β-cat^eKO/DM embryos expressed a comparable level of mesenchymal markers including αSMA, VIMENTIN, and PDGFRβ (Fig 7A). SNAIL, SLUG, and SOX9 are key transcriptional factors inducing endocardial cells to gain mesenchymal fate. The expression of these markers in cushion endocardium was not affected by β-catenin loss (Fig 7B). Together, these results indicate that β-catenin is dispensable for endocardial cells to gain mesenchymal fate.

Experimental mouse models

Tie2^Cre mice (Braren et al, 2006) were crossed with the floxed β-catenin mice (Brault et al, 2001) to generate conditional β-catenin knockout mice in which β-catenin was selectively deleted in the endothelium including endocardium during embryogenesis. β-catenin^DM mice (Valenta et al, 2011) which express a modified β-catenin protein with a single amino acid mutation (D164A) in the first armadillo repeat and a C-terminal truncation. This mutant β-catenin protein lacks any detectable transcriptional activity (signaling function) but retains the intact cell adhesive or cell–cell contact function at the cell membrane (Valenta et al, 2011). All mouse strains were maintained on the C57B6 background. Mouse experiments were performed according to the guideline of the National Institute of Health and the protocol approved by the Institutional Animal Care and Use Committee of Albert Einstein College of Medicine. Noontime on the day of detecting vaginal plugs was designated as E0.5. The embryos were isolated from the pregnant female mice which were euthanized by inhalation of carbon dioxide gas using a euthanasia chamber. Adult mice and mouse embryos were genotyped by PCR using DNA from tail and yolk sac, respectively.

Histology

Embryos were dissected between E9.5 and E11.5, fixed overnight at 4°C using 4% PFA in PBS, dehydrated through a serial ethanol solution, cleared with xylene, and embedded in paraffin wax. Embryos were oriented for sagittal sections and cut at 6 μm using a Leica microtome. Hematoxylin and eosin (HE) staining was performed for histology by using a standard protocol. The HE-stained tissue sections were examined and photographed using a Zeiss Axio Observer Z1 inverted microscope. Serial sections across the entire AVC cushion region were used for quantification of the number of mesenchymal cells. To analyze the distribution of mesenchymal cells in the endocardial cushions, cells located underneath the endocardium (sub-endocardium) and far away from the endocardium and underneath the myocardium (sub-myocardium) were counted, respectively. The percentage of mesenchymal cell in the sub-endocardium or the sub-myocardium was calculated and the data were presented as mean ± SD. Data from 4–6 embryos for each genotype were used for statistical calculation.

Immunofluorescence staining

Tissue sections prepared as described above were boiled for 10 min in sodium citrate (10 mM, pH 6.0) (Vector Laboratories) to retrieve antigens and then blocked with 5% normal horse serum in PBS before being incubated with primary and secondary antibodies (Table S1). The signal was amplified using the TSA Plus Cyanine 3 System (Perkin Elmer) as noted for some antibodies. The stained tissue sections were photographed using a Leica confocal microscope.

EdU staining

Cell proliferation was determined by the EdU assay. 2 h before collecting embryos, pregnant female mice at E9.5 were administrated with EdU (Life Technology) through intraperitoneal injection at a concentration of 100 mg/kg. After the 2 h chasing, embryos were fixed in 4% PFA at 4°C for 1 h, soaked in 15% and 30% sucrose sequentially, and embedded in OCT compounds in the sagittal orientation. The tissues were cut at 8 μm thickness and mounted on the positive charged slides. Tissue sections were then fixed in cold ethanol and acetone (1:1) solution for 5 min and stored at −80°C. Before staining, tissue sections were air dried for 45 min. Serial sections crossing the entire cushion region were first stained with PECAM1 antibody followed by EdU staining using an EdU imaging Kit (Life Technology) and counterstained with DAPI. The stained sections were photographed using a Leica confocal microscope. The positive EdU labeling was quantified for the endocardium, mesenchyme, or myocardium using ImageJ software. At least three embryos were analyzed for each genotype.

RNA extraction and qRT-PCR

The AVC cushion tissues were microdissected from E9.5 embryos and subjected to isolation of total RNAs using TRIzol (Invitrogen). Tissues from 2–3 embryos were pooled as one sample. First-strand cDNA was synthesized using the Superscript II Reverse Transcriptase Kit (Invitrogen). qRT-PCR was carried out using the Power SYBR Green PCR Master Mix (ABI). Gene-specific primers were used for genes of interest (Table S2). The relative level of gene expression was normalized to an internal control (level of Gapdh) and calculated using the 2^−ΔΔCT method. Three samples per genotype were analyzed, and the mean relative expression of each gene between groups was used for statistical analysis.

RNAscope

E9.5 embryos were collected and processed for frozen sections. mRNA expression pattern of genes was detected by using an RNAscope 2.5 HD Reagent Red Kit (ACD BIO) according to the manufacturer’s instructions. The images were collected using a Leica confocal microscope.

In vitro EndoMT assay

EndoMT assay was performed as described previously (Lencinas et al, 2011). AVC tissues were microdissected out from E9.5 embryos and cultured on the rat tail collagen gel in four-well plates. Explants were cultured for 48 h and stained with phalloidin to label the F-actin, whereas nuclei were stained by DAPI. Cells that migrated away from the AVC explants and invaded into the gel were counted. To track the cell migration, time-lapse images were collected and subjected to cell tracking analysis using DiPer as described previously (Gorelik & Gautreau, 2014).

Cell culture

Primary HUVECs and PAVIC were cultured in endocardial cell medium (ScienCell Research Laboratories) and DMEM, respectively. To block the canonical WNT signaling, XAV939 were added to the culture media at a final concentration of 1 μM for HUVEC or 2.5 μM for PAVIC. PAVIC was treated with 2.5 μM of UC2288 to inhibit p21. To knockdown the expression of Ctnnb1 (β-catenin) and Cdkn1A (p21), the cells were transfected with gene-specific siRNA, whereas the scramble siRNA were introduced simultaneously as controls. The knockdown efficiency was evaluated using qRT-PCR. EdU was added into the culture media with a final concentration of 10 μM. 2 h later, the cells were processed for EdU staining as described above. Immunostaining was performed to detect the expression level of p21 protein.

Statistical analysis

Statistical analysis was performed using GraphPad Prism. All data were presented as mean ± SEM. t test was used for comparisons between two groups. One-way ANOVA was used for comparisons among three groups. Probability value < 0.05 was considered as significant.