Melanoblasts replace a subpopulation of smooth muscle cells in ctnnb1Δex3 mice
mice were generated by producing an activated form of β-catenin in cells of the Tyr::Cre lineage. LoxP
sequences had previously been inserted into introns 2 and 3 to flank the third exon of the gene encoding β-catenin (ctnnb1
), for which the allele is also known as “f3” 
. Exon 3 encodes serine and threonine residues involved in phosphorylation and degradation of the protein. The resulting modified β-catenin is reportedly more stable 
. We verified in vitro
that the deletion of exon 3 substantially increases protein activity (by five-fold) in B16 melanoma cell lines (), as has been demonstrated in vivo
for recombined cells of the intestinal epithelium 
/° mice were then crossed with heterozygous or homozygous floxed (
/+ or ctnnb1Δex3
mice to produce Tyr::Cre
/°; +/+ (WT) and Tyr::Cre
/+ (ctnnb1Δex3) siblings (Table
). These WT and ctnnb1Δex3 mice were initially backcrossed on a Dct::LacZ
background, to visualize Mb in the DA. Dct
is a member of the tyrosinase gene family and encodes the enzyme dopachrome tautomerase (also known as tyrosinase-related protein 2 or Trp2) involved in eumelanin synthesis. In Dct::LacZ
/° reporter mice, the transgene LacZ
, under the control of the Dct
promoter, is expressed in Mb, retinal pigmented epithelium cells, and in the forebrain, but not in vascular smooth muscle cells. Tyr::Cre/
°; +/+; Dct::LacZ
WT-Dct) and Tyr::Cre
ctnnb1Δex3-Dct) hearts were isolated, fixed and stained with X-gal on embryonic day (E)18.5. Whole-mount observation revealed that there were visibly more Mb in the mutant than the WT DA (). Transverse sections confirmed that Mb were scarce in the WT-Dct DA, consistent with a previous study 
, but were numerous (60±23 Mb per section) in ctnnb1Δex3-Dct DA ().
Melanoblasts are numerous in ctnnb1Δex3 DA.
Transverse sections of WT and mutant ductus arteriosus.
Antibodies directed against α-smooth muscle actin (SMA) and β-galactosidase were used for immunofluorescence analysis of E18.5 ctnnb1Δex3-Dct and WT-Dct DA (). Expression of SMA and β-galactosidase was mutually exclusive: all β-galactosidase-positive cells were SMA-negative, and SMA-positive cells were β-galactosidase negative (). This suggested that at E18.5, the cells forming the DA were fully committed either to SMC (SMA-expressing) or to Mb (β-galactosidase-expressing due to the Dct::LacZ transgene).
SMA-negative cells of the DA of ctnnb1Δex3-Dct mice are β-galactosidase-positive melanoblasts
To trace the progeny of recombined cells, we used the Tyr::Cre
system for recombination and Rosa26R reporter mice for continuous β-galactosidase expression. Tyr::Cre
/+ (ctnnb1Δex3) or Tyr::Cre
/°; +/+ (WT) mice were crossed with Rosa26R/+ mice to generate ctnnb1Δex3-Rosa (Tyr::Cre
/+; Rosa26R/+) and WT-Rosa (Tyr::Cre/
°; +/+; Rosa26R
/+) offspring. Immunostaining of the hearts and DA at E18.5 revealed the existence of three types of cells: (i) SMA-positive and β-galactosidase-negative cells (
SMC1), (ii) SMA-positive and β-galactosidase-positive cells (
SMC2) and (iii) SMA-negative and β-galactosidase-positive cells (
Mb). The SMA-positive cells correspond to SMC, and the β-galactosidase-positive cells to recombined cells that had expressed the Tyr::Cre
transgene. The numbers of SMC (
SMC1+SMC2) were determined in transverse sections of WT-Rosa and ctnnb1Δex3-Rosa DA (). The total counts of SMC were higher in WT-Rosa (382±49) than in ctnnb1Δex3-Rosa (301±53) DA sections. However, the numbers of SMC1 were similar (300±65) in WT-Rosa and ctnnb1Δex3-Rosa, indicating that the number of SMC1 is not altered in ctnnb1Δex3-Rosa animals, but that SMC2 are lacking.
Melanoblasts replace a proportion of the smooth muscle cells in the ctnnb1Δex3 DA.
The numbers of SMC2+Mb per section were very similar in the DA of WT-Rosa and ctnnb1Δex3-Rosa (around 70) (). However, the numbers of SMC2 alone were 62±29 versus 5±3 per section in WT-Rosa and ctnnb1Δex3-Rosa, respectively, while the numbers of Mb were 5±5 versus 5,800±3,400 per 100 sections in WT-Rosa and ctnnb1Δex3-Rosa, respectively. These various observations show that on a ctnnb1Δex3 background, Mb replaced most of the SMC2 in the DA. In other words, a subset of VNCC is bipotent for SMC2 and Mc, and responds to signaling through β-catenin.
ctnnb1Δex3 mice have a greatly dilated left atrium, which develops after birth
As expected and according to specific crosses, Tyr::Cre
mice were produced in a Mendelian ratio (72 °/°; ctnnb1Δex3/+
versus 67 Tyr::Cre
live births). However, all ctnnb1Δex3 mice died between 4 and 18 weeks of age (). A distinct Tyr::Cre
mouse line, Tyr::CreB
, was used to see if the site of integration of the transgene or other cis-regulatory differences would affect the outcome 
. Nonetheless, all Tyr::CreB
/+ mice died within a similar age range (between 3 and 12 postnatal weeks; ), indicating that the death of these animals was independent of the Tyr::Cre
mouse line used.
ctnnb1Δex3 mice die of heart failure between the second and fourth postnatal months.
The main clinical sign presented by ctnnb1Δex3 mice was prostration 24 hours prior to death. Such prostrated, heterozygous ctnnb1Δex3 mice and their wild-type siblings were sacrificed and autopsied. While never observed in the WT controls (), a major dilation of the left atrium was observed macroscopically in all mutant mice () and a lesser dilation of the left ventricle was observed in about half of the mice (), Micro-computed tomography (CT) analysis after Fenestra® injection was performed on WT and ctnnb1Δex3 mice (
and movies S1, S2). This revealed a rightward shift of the whole heart in live mutant animals, with a clear dilation of the left atrium ().
Dilatation of the ctnnb1Δex3 left atrium at postnatal day 28.
WT and ctnnb1Δex3 hearts were examined at P1, P10 and P28. At P1, WT and ctnnb1Δex3 hearts were comparable, indicating that the enlargement of the left atrium was not associated with a congenital malformation due to a developmental defect (). At P10, the ctnnb1Δex3 left atrium was visibly dilated (). By P28, the expansion of the left atrium was substantial ( and ). Some dilation of the left ventricle, not hypertrophy, was also visible in a limited number of mutants by P28 ( and ).
Progressive dilatation of the ctnnb1Δex3 left atrium during the first weeks of life.
Echocardiographic examination also showed substantial enlargement of the left atrium, associated with thrombi of various sizes (). Thrombus formation was subsequent to chamber dilation and aggravated over time (). In one case, atrial myocardial rupture and pericardial blood effusion was observed on ultrasound analysis, and led to the death of the mouse. Atrial dilation was unlikely to be due to mitral valve dysfunction, because the Doppler mitral inflow pattern was normal (data not shown). Similarly, left-ventricular failure was not the cause, because the fractional shortening was normal. These data led to the hypothesis that death was due to the accumulation of large thrombi in the left atrium and/or its rupture.
Thrombosis develops in mutant mice during the second postnatal month.
The ductus arteriosus is not fully closed in ctnnb1Δex3 mice
When the DA does not fully close after birth, a part of the systolic left-ventricular stroke volume goes directly into the pulmonary artery (left-to-right shunting), leading to a progressive overload of the pulmonary circulation by an increase in pulmonary pressure. Simultaneously, this volume overload triggers the progressive dilation of the left cardiac cavities. Surprisingly, in ctnnb1Δex3 mutant mice, this enlargement affected essentially the left atrium, the left ventricle being more modestly and not systematically dilated. Ultrasound analysis demonstrated that the DA remained open in postnatal ctnnb1Δex3 mice, which is never the case in WT mice (). Ultrasound () and color Doppler analyses () showed blood flow back through the patent foramen ovale () from the right to the left atrium of all ctnnb1Δex3 mice, but not WT mice.
Abnormal circulation of the blood in ctnnb1Δex3 adult heart.
The death of the ctnnb1Δex3 mice thus seemed to result from the failure of DA closure and increased pulmonary pressure, leading ultimately to retrograde blood flow through the foramen ovale from the right to the left, driving the progressive dilation of the left atrium and thrombus formation.
After birth, the very few Mc in the DA normally differentiate, produce melanin and remain in the LigA. The number of pigmented cells was substantially higher in ctnnb1Δex3 LigA than WT LigA by four weeks (). Histological analysis revealed that the ctnnb1Δex3 LigA was not fully closed (). The surface areas of the intimal cushion and lumen, when present, were larger in ctnnb1Δex3 than WT mice (). Moreover, blood was observed in the canal and large numbers of pigmented cells were present in the tunica media of the ctnnb1Δex3 LigA (). Only ctnnb1Δex3 mice presented a PDA, as well as disorganized, fibrotic lung alveolar structures (). These observations suggested that abnormally high pulmonary pressure may have been the result of the PDA and cause lung damage in ctnnb1Δex3 mice as compared to WT mice.
Closure of the ligamentum arteriosum in ctnnb1Δex3 adult heart.
Histological analysis of WT and ctnnb1Δex3 lungs at P28.
ctnnb1Δex3 mice can be partially rescued by indomethacin
In humans, indomethacin is widely used to treat PDA, by inhibiting the cyclooxygenases that participate in prostaglandin biosynthesis. As a proof of concept, pregnant Tyr::Cre/Tyr::Cre; +/+; Dct::LacZ/Dct::LacZ females that had been crossed to ctnnb1Δex3/+ males, and thus carrying litters with both WT-Dct and ctnnb1Δex3-Dct embryos, were injected with indomethacin at E18.5 and compared with mock-injected controls. Embryos were removed from three indomethacin-injected females, 4 hours after injection. Five WT-Dct and five ctnnb1Δex3-Dct hearts were isolated and fixed to obtain transverse sections of the DA. Mb were visualized by the activity of β-galactosidase (). Indomethacin induced the closure of the DA as expected, in WT-Dct mice, but also by means of the remaining SMC in ctnnb1Δex3-Dct DA.
Indomethacin treatment and survival of ctnnb1Δex3 mice.
Newborn ctnnb1Δex3 pups were also directly subcutaneously injected with indomethacin at birth, and this significantly improved their survival rate and indeed cured about twenty per cent of animals (n
29; ). Therefore, ctnnb1Δex3 mice could be partially rescued by indomethacin treatment, suggesting that one of the primary causes of death in ctnnb1Δex3 mice was the failure of full DA closure at birth. However, we cannot exclude the possibility that the left atrium is structurally abnormal in ctnnb1Δex3 mutants, possibly contributing to the death of the animals.
β-catenin does not lead to overall Cox-2 induction in the DA
β-catenin can directly induce the expression of Ptgs2
(encoding cyclooxygenase-2 or Cox-2) and stabilizes its mRNA by interacting with AU-rich elements of the 3′-UTR in vitro
. Cox-2 in turn is known to catalyze the formation of PGE2, and elevated levels of PGE2 are associated with an open DA.
One possible explanation for the failure of DA closure in ctnnb1Δex3 mice is that unduly high levels of Cox-2 were induced by the augmented β-catenin signaling. Therefore, we verified that Cre-recombined, truncated ctnnb1Δex3 β-catenin mRNA was detectable in ctnnb1Δex3 DA but not in WT DA, as determined by extracting total RNA from DAs and performing semi-quantitative RT-PCR (). β-catenin was visible in both cytoplasm and nuclei, as assessed by immunofluorescence, in both ctnnb1Δex3-Dct and WT-Dct Mb; as expected, more Mb were visible in the former on sections (). Nonetheless, the amount of Ptgs2 (Cox-2) mRNA in ctnnb1Δex3 DA was similar to that in WT DA (), indicating that increased β-catenin signaling in ctnnb1Δex3 DA did not result in massive induction of Cox-2. Therefore, Cox-2, although a target of β-catenin, is probably not principally involved in the failure of DA closure in mutant mice at birth.
ctnnb1Δex3 is produced in melanoblasts and SMC cells of E18.5 DA.
ctnnb1Δex3-mi mice lacking melanocytes still die of heart failure
Mitf-deficient mice (mi
) have no Mc and a white coat, resulting from a recessive null allele for the microphthalmia-associated transcription factor. In mi mice, Mc are genetically ablated during development at around E11.5, which is after specification of the common SMC2/Mc precursor at about E9.0, but well before birth 
. To evaluate the contribution of Mc to the full closure of the ctnnb1Δex3 DA at birth, ctnnb1Δex3-mi (Tyr::Cre
) mice were produced. As anticipated, the ctnnb1Δex3-mi mice lacked cutaneous Mc and were white, like mi mice (data not shown). Surprisingly, ctnnb1Δex3-mi mice showed clinical signs similar to ctnnb1Δex3 mice and died at similar ages (). Left atrium enlargement was observed in ctnnb1Δex3-mi mice at P28 (). PDA was observed at P2 in ctnnb1Δex3-mi, but not in mi mice ().
The PDA of ctnnb1Δex3 mice is not rescued by removing melanocytes.
Additional mice were generated in order to estimate the numbers of Mb, through β-galactosidase activity, and SMC through SMA immunoreactivity, in transverse sections of the DA. Tyr::Cre/°; ctnnb1Δex3/+; mivga9/vga9; Dct::LacZ/° (ctnnb1Δex3-mi-Dct) and Tyr::Cre/°; +/+; mivga9/vga9; Dct::LacZ/° (mi-Dct) mice were compared to ctnnb1Δex3-Dct mice (cf. and ). No Mb were observed in the DA of ctnnb1Δex3-mi-Dct or mi-Dct, in contrast to the ctnnb1Δex3-Dct DA. However, the numbers of SMC were similar in ctnnb1Δex3-Dct and ctnnb1Δex3-mi-Dct DA and reduced relative to the DA of mi-Dct, such that no cellular compensation for SMC was observed in ctnnb1Δex3-mi-Dct DA ().
The general heart morphology appears disrupted in ctnnb1Δex3-mi mice compared to mi mice (), as this is the case for ctnnb1Δex3 mice compared to WT mice (). A conjunction of events may explain these phenomenona, which are potentially linked, but are not yet understood. These abnormalities could be due to the general blood flow defect in the mutant mice, leading to abnormal pressures on the chambers, thus affecting the morphology. However, it could be associated with melanocytes. Indeed, Mc are present in the heart, and not only in the DA 
. A careful analysis allowed the visualization of a general increase in Mc numbers in different parts of ctnnb1Δex3 hearts. It includes the foramen ovale (), and the mitral, tricuspid and aortic valves (not shown). In wild-type and ctnnb1Δex3 hearts, no Mc was found in the pulmonary valves. All ctnnb1Δex3 Mc of the heart would have a different expression pattern compared with WT Mc, leading to molecular and cellular modifications. Besides the consequence on the DA, these modifications do not appear dramatic on ctnnb1Δex3 heart. The valves did not present major defects and the Doppler mitral inflow pattern was normal. However, mi mice do not have Mc in the heart due to the lack of Mitf. Mitf is not only expressed in Mc, it is also expressed in cardiomyocytes and is important to regulate cardiac growth and hypertrophy 
. On C57BL/6 background, mi hearts are not dramatically affected. The conjunction of the lack of Mitf in the heart with the disappearance of an increased number of Mc may explain the different heart morphology and a potential earlier death of ctnnb1Δex3-mi mice, although this hypothesis remains elusive.
In order to confirm that β-catenin does not play a major role in the cardiac Mc lineage after its segregation from SMC2, Dct::Cre
/+ mice were also produced, and found to be normally viable, with no evident enlargement of the left atrium when sacrificed at one-year old. In these mice, the activated β-catenin is produced specifically in Mb after specification. In another approach, Tyr::CreERt2
/° mice 
were crossed with ctnnb1Δex3
/+ mice to produce a tamoxifen-inducible CreERt2 in Tyr
-expressing cells. Tamoxifen was administered at E18.5, well after the segregation of the SMC2 and Mc lineages, and all double heterozygous mice were phenotypically normal: viable, again with no enlargement of the left atrium. We counted Dct/Trp2-positive cells, corresponding to Mc, in DA sections at P2 in both tamoxifen- and mock-induced, unrecombined Tyr::CreERt2
/+ mice. The numbers were very small in both cases, similar to that in WT mice (not shown). Repeating these experiments in both WT and ctnnb1Δex3 backgrounds using mi mice, as expected from the results in , failed to detect any Dct/Trp2-positive cells. The mice were viable and with no cardiac defect.
In conclusion, the DA of ctnnb1Δex3-mi and ctnnb1Δex3 mice did not fully close, demonstrating that the presence or absence of differentiated Mc was irrelevant to the onset of PDA. Rather, the absence of a significant proportion of SMC (SMC2 population) was associated in both cases with the failure of full DA closure and death; indomethacin injection reduced or prevented these manifestations (see ).