Canonical Wnt signaling is a positive regulator of mammalian cardiac progenitors

β-Catenin Regulation of SHF Progenitors.

To determine whether canonical Wnt signaling might be active in the heart-forming region, we examined β-catenin protein expression in mouse embryos. At embryonic day (E)8.5, the myocardial layer was enriched in β-catenin, suggesting preferential stabilization of the protein in this region (Fig. 1A). β-Catenin protein was abundant in all chambers of the heart and in the SHF-derived outflow tract at E10.5. Of the canonical Wnt molecules expressed in the developing heart (Wnt2, Wnt7a, and Wnt8a) (20–22), Wnt8a was most specifically expressed in heart precursors at E8.5 (Fig. 1 C and D). Wnt8a was expressed initially throughout the developing heart tube. By E10.5, protein was expressed in all chambers and in the outflow tract. Wnt8a persisted in the ventricular chambers at E12.5, including the compact and trabecular layers. β-Catenin and Wnt8a expression persisted at later stages of development (Fig. 1 B, E, and F).

To examine the cell-autonomous role of canonical Wnt signaling in precardiac mesodermal progenitors or cardiomyocytes, we disrupted or stabilized β-catenin expression in cardiac mesodermal progenitors with ctnnb1^tm2Kem (23) or β-catenin/loxP(ex3) (24) mice, respectively. In ctnnb1^tm2Kem mice, Cre recombinase-mediated excision of exons 2–6 produced a β-catenin-null allele; in β-catenin/loxP(ex3) mice, cre expression activated Wnt/β-catenin signaling by generating stable β-catenin [supporting information (SI) Fig. 5A]. We deleted β-catenin in SHF progenitors before cardiac differentiation by crossing ctnnb1^tm2Kem mice with Islet1-cre mice, in which Cre is expressed at E7.0–7.5 in undifferentiated mesodermal progenitors that give rise to the right ventricle (RV) and outflow tract (3). In Islet1-cre ctnnb1^tm2Kem homozygous embryos, the SHF failed to form the RV, but the left ventricle (LV) was relatively normal in size (Fig. 2 A–C and G–I). In situ hybridization with a probe recognizing the LV-specific transcript Hand1 (25, 26), also expressed in the outflow tract, confirmed the identity of the residual ventricular chamber (Fig. 2 C and I). Consistent with the activity of Islet1-Cre in pharyngeal arch mesoderm, deletion of β-catenin led to hypoplasia of these structures (Fig. 2 B and H). It should be noted that this Isl1-Cre is not very active in the pharyngeal endoderm despite Isl1 mRNA expression in the endoderm. Thus, SHF mesoderm progenitors may require cell-autonomous Wnt/β-catenin signaling for specification, migration, or expansion.

To assess whether β-catenin was required for specification or maintenance of the initial Isl1-positive cells, we examined Isl1 expression in the Islet1-Cre/β-catenin mutant mice. Isl1 was expressed at similar levels and in a comparable domain in mutants and wild-type embryos (Fig. 2 E and K), suggesting that the initial formation and maintenance of SHF progenitors was intact. Migration also appeared intact as Fgf10, marking SHF cells that migrate into the cardiac outflow tract (3, 5, 27), was also expressed appropriately in the β-catenin mutants (Fig. 2 F and L). Finally, we observed no evidence of SHF progenitor cell death (not shown), indicating that the primary defect may lie in failure of SHF cells to differentiate and/or expand in number after specification.

To directly test this idea, we used the Islet1-Cre and β-catenin/loxP(ex3) mice to stabilize β-catenin in the Isl1-positive SHF progenitors. Gain-of-function of β-catenin in these progenitors resulted in a greatly enlarged RV segment from E8.5–9.5 compared with the LV, which was marked by Hand1 expression (Fig. 2 M–O and S–U). In some, the RV segment was so large at an early stage that protrusion of the heart tube was apparent with cardiac dysfunction resulting in developmental delay (Fig. 2 Y, Z, and A′). Most strikingly, the SHF progenitor pool dorsal to the developing heart tube was greatly expanded (Fig. 2 P, Q, V, and W) and contained an increased percentage of phosphohistone H3 (PH3)-positive cells (Fig. 2 R and X), indicating that β-catenin stabilization resulted in marked proliferation of SHF progenitors.

β-Catenin Promotes Proliferation and Differentiation of Cardiomyocytes in Vivo.

To determine whether Wnt/β-catenin signaling could also affect proliferation of cardiac cells even after they have begun to differentiate, we used a transgenic mouse line in which cre expression is under control of a ventricular-specific enhancer of the cardiac regulatory gene, Nkx2.5. Unlike the endogenous Nkx2.5 gene, which is expressed in the SHF, this Nkx2.5 enhancer is not active in undifferentiated SHF cells (28). The Cre-mediated excision event begins at E8.0–8.5 in the cells that have already arrived into the heart and have begun differentiation (28); excision of β-catenin was confirmed by western analysis as nearly complete (SI Fig. 5B).

Nkx2.5-cre, ctnnb1^tm2Kem homozygous embryos died around E12.5 from cardiac dysfunction (Fig. 3 A and G), with significant reductions in ventricular size, particularly the RV (Fig. 3 B–F and H–L) and a thinner compact layer in the ventricular wall (Fig. 3 F and L). To identify the source of the myocardial alterations, we analyzed hearts by immunostaining with anti-PH3 antibody to detect alterations in cell-cycle control or by TUNEL assay to detect apoptosis. The percentage of PH3-positive cells was lower (P < 0.01) in Nkx2.5-cre, ctnnb1^tm2Kem hearts and greater (P < 0.01) in Nkx2.5-cre, β-catenin/loxP(ex3) hearts than in wild-type hearts (Fig. 3M). Similarly, the percentage of cyclin D2-positive cells, a direct target of the β-catenin/TCF complex that promotes cell cycling (29), was increased in Nkx2.5-cre, β-catenin/loxP(ex3) hearts and decreased in Nkx2.5-cre, ctnnb1^tm2Kem hearts (Fig. 3N Upper, P < 0.01). Total protein levels of cyclin D2 were correspondingly regulated in heart lysates (Fig. 3N Lower). Apoptosis was unaffected (not shown). Expression of numerous regulators of cardiogenesis, including BMP4, Hand2, Hrt2, Mef2c, and Sall4, was down-regulated in Nkx2.5-cre, ctnnb1^tm2Kem hearts, and BMP4 and Mef2c expression was up-regulated in Nkx2.5-cre, β-catenin/loxP(ex3) hearts (Fig. 3O). Thus, Wnt/β-catenin activity within cardiac mesoderm appears to positively regulate proliferation and full differentiation of ventricular cardiomyocytes even after early commitment steps have occurred.

Canonical Wnt Signaling Promotes Expansion and Differentiation of Mesodermal Progenitors into Cardiomyocytes in the ESC System.

To assess the earlier function of Wnt signals in cells already committed to the mesodermal but not yet to the cardiac fate and to more precisely control the timing of Wnt stimulation or disruption, we turned to the ESC system. ESCs grown in hanging drops form clusters of cells called embryoid bodies (EBs) that randomly differentiate into derivatives of the three germ layers and can form beating cardiomyocytes, although with low frequency. The frequency is even lower when EBs are grown in suspension, which is technically much easier and would allow higher throughput generation of cardiomyocytes from ESCs.

To target canonical Wnt signaling only after mesoderm commitment, we first profiled mesoderm and cardiac gene expression during ESC differentiation in suspension, given that sensitivity to Wnt effects occurs in narrow windows of time. The early mesoderm marker Brachyury (Bry) was strongly induced on day 3 of EB differentiation (EB3) but was down-regulated afterward, suggesting mesoderm commitment by EB3 (SI Fig. 6). Other early cardiac mesoderm markers, such as the transcription factors Nkx2.5, Tbx5, Gata4, and Mesp1, were highly induced at EB4–6. Late cardiac markers, such as the sarcomeric genes Myh6, Myh7, Mlc2a, and Mlc2v, were induced by EB7–9 (SI Fig. 7).

Canonical Wnt signaling is required for general mesoderm formation in vivo (30, 31) and in EBs (32). We therefore initially targeted canonical Wnt signaling after mesoderm commitment (EB3) but before induction of early cardiac genes (EB4–5). Wnt3a, the soluble canonical Wnt ligand available, and Dickopff1 (Dkk1), an inhibitor of canonical Wnts, were used to promote or disrupt Wnt/β-catenin signaling, respectively. Bry expression was unaltered by addition of these factors at EB4–6 (Fig. 4A), suggesting that stimulating or inhibiting canonical Wnt signaling after EB3 did not alter the number of mesodermal progenitors. In contrast, stimulation of Wnt/β-catenin between EB4–6 with soluble Wnt3a increased the number of beating EBs at EB12 from 10% to >50% when cells were grown in suspension (Fig. 4B Left, P < 0.01). Inhibiting canonical Wnt signaling with Dkk-1 between EB4–6 resulted in a complete absence of beating EBs (Fig. 4B Left), suggesting that canonical Wnt signaling during this window not only promotes but also is required for cardiomyocyte formation in the ESC differentiation system.

The presence of beating EBs is useful for gross analyses but does not address the number of progenitors or differentiated cardiomyocytes. To more rigorously assess this aspect, we used an ESC line containing GFP under control of the endogenous Nkx2.5 regulatory elements to mark and sort early cardiac mesoderm (B.R.C. and E.C.H., unpublished results). We found that stimulation or inhibition of Wnt signaling between EB4 and EB6 did not affect the initial number of progenitors in day 6 EBs. However, flow cytometry revealed a progressive expansion of the Nkx2.5-positive pool upon Wnt3a stimulation and a decrease in the progenitor pool upon Wnt inhibition as further differentiation proceeded in EBs (Fig. 4C). The numbers of cells positive for Tropomyosin, a late cardiac sarcomeric marker, were similarly increased by the Wnt agonist by EB12 (Fig. 4D). Notably, no Tropomyosin+ cells were observed in EBs treated with the Wnt antagonist after EB4, consistent with the absence of beating EBs (Fig. 4B Left).

In agreement with a positive role for Wnt in the expansion and possibly early differentiation of the cardiac lineage, expression levels of the early cardiac genes Nkx2.5 and Tbx5 was up-regulated by exposure to Wnt3a from EB4–6 and down-regulated by similar exposure to Dkk-1 (Fig. 4E Upper). The SHF marker, Islet1, and the RV marker, Hand2 (26), showed similar responses to Wnt3a and Dkk-1, as did cardiac sarcomeric genes (Fig. 4E Upper).

To assess effects of canonical Wnt signaling after induction of early cardiac mesoderm in ESCs, we treated EBs with Wnt3a or Dkk-1 between EB 7 and EB9. Nearly 50% of EBs treated with Wnt3a and grown in suspension were beating by EB12 (Fig. 4B Right, P < 0.01), and most early and late cardiac markers were up-regulated (Fig. 4E Lower), suggesting that Wnt had a positive effect on cardiomyocyte development even at this relatively late stage. This timing of Wnt manipulation is similar to the ventricular-restricted Nkx2.5-cre transgenic line used in the in vivo studies. Approximately 10% of EBs treated with Dkk-1 were beating (vs. 13% of control EBs; Fig. 4B Right, P > 0.2), and cardiac gene expression was not changed upon Wnt inhibition (Fig. 4E Lower). Thus, after cardiac mesoderm commitment, canonical Wnt signals appear to be dispensable for full cardiac differentiation but may still recruit mesodermal precursors or expand cardiac progenitors. This result is consistent with our in vivo studies with the Nkx2.5-cre mouse described above.

Mouse Lines and Genetics.

The Nkx2.5-cre, ctnnb1^tm2Kem or Islet1-cre, ctnnb1^tm2Kem homozygous embryos were obtained by crossing Nkx2.5-cre, ctnnb1^{tm2Kem flox/+} or Islet1-cre, ctnnb1^{tm2Kem flox/+} lines with ctnnb1^{tm2Kem flox/flox} lines, respectively. Nkx2.5-cre, β-catenin/loxP(ex3), or Islet1-cre, β-catenin/loxP(ex3) heterozygous embryos were obtained by crossing Nkx2.5-cre or Islet1-cre lines with β-catenin/loxP(ex3) ^flox/+ lines, respectively. Wild-type and mutant embryos were identified as described (23, 24).

Immunohistochemistry, Western Blotting, and in Situ Hybridization.

Whole or cryosectioned embryos and EBs were stained with the following antibodies: mouse anti-Islet1 (1:200) and anti-Tropomyosin (1:100; Developmental Studies Hybridoma Bank, Iowa City, IA), goat anti-Wnt8a (15 μg/ml, R&D Systems, Minneapolis, MN), goat anti-β-catenin (1:50), and rat anti-cyclin D2 (1:100) (Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anti-phospho-histone H3 (1:100; Upstate Biotechnology, Charlottesville, VA), and horseradish peroxidase-conjugated (with TSA plus Fluorescent Systems; PerkinElmer, Waltham, MA) secondary antibodies (The Jackson Laboratory, Bar Harbor, ME). For Western blotting, lysates from E12.5 ventricles were analyzed by using antibodies against β-catenin, cyclin D2, and GAPDH (1:50, 1:150, and 1:100, respectively; Santa Cruz Biotechnology). mRNA in situ hybridization was performed with designated antisense probes as described (3, 5, 27).

ESC Culture, Flow Cytometry, and Gene Expression Analysis.

Murine ESCs were propagated undifferentiated in maintenance medium of Glasgow MEM (Sigma, St. Louis, MO; G5154) supplemented with 10% FBS (HyClone, Logan, UT; SH30071.03), 1 mM 2-mercaptoethanol (Sigma M7522), 2 mM l-glutamine (Gibco-BRL, Carlsbad, CA; 25030–081), 1 mM sodium pyruvate (GIBCO-BRL 11360-070), 0.1 mM minimum essential medium containing nonessential amino acids, and leukemia inhibitory factor (LIF)-conditioned medium (1:1,000). EBs were formed by culturing ESCs (6 × 10⁵ per well) for 3 days in ultra-low-attachment six-well plates (Corning, Lowell, MA; 07200601) in differentiation medium (DM), which contained the same components as maintenance medium but had 20% FBS and no LIF. For early treatment, the medium was replaced with DM containing Wnt3a (150 ng/ml, R&D Systems) or Dkk1 (500 ng/ml, R&D Systems) at the start of day 4; on day 6, the medium was replaced with fresh DM. For late treatment, DM was replaced with Wnt3a- or Dkk1-containing DM at the start of day 7 and switched to fresh DM on day 9. Respective media were changed for early-, late-, and untreated EBs every 3 days. EBs were monitored for beating from day 8 to day 12 and collected between days 3 and 12 at regular intervals. At respective days of differentiation, EBs were dissociated via trypsin and passed through a nylon cell strainer. A Becton Dickinson (Franklin Lakes, NJ) FACS Diva flow cytometer and cell sorter was used for detecting Nkx2.5-GFP⁺ cells. For gene expression analysis, total RNA was isolated, and real-time PCR was performed with the ABI Prism system (7900HT, Applied Biosystems, Foster City, CA). TaqMan primers used in this study are listed in SI Table 1. All samples were run in triplicate. Real-time PCR data were normalized and standardized with SDS2.2 software (Applied Biosystems).