Myocardial Bmp2 gain causes ectopic EMT and promotes cardiomyocyte proliferation and immaturity

During mammalian heart development, restricted myocardial Bmp2 expression is a key patterning signal for atrioventricular canal specification and the epithelial–mesenchyme transition that gives rise to the valves. Using a mouse transgenic line conditionally expressing Bmp2, we show that widespread Bmp2 expression in the myocardium leads to valve and chamber dysmorphogenesis and embryonic death by E15.5. Transgenic embryos show thickened valves, ventricular septal defect, enlarged trabeculae and dilated ventricles, with an endocardium able to undergo EMT both in vivo and in vitro. Gene profiling and marker analysis indicate that cellular proliferation is increased in transgenic embryos, whereas chamber maturation and patterning are impaired. Similarly, forced Bmp2 expression stimulates proliferation and blocks cardiomyocyte differentiation of embryoid bodies. These data show that widespread myocardial Bmp2 expression directs ectopic valve primordium formation and maintains ventricular myocardium and cardiac progenitors in a primitive, proliferative state, identifying the potential of Bmp2 in the expansion of immature cardiomyocytes.


Introduction
Formation of the primitive cardiac valves begins at E9.5 in mice, when signals from the atrioventricular canal (AVC) myocardium stimulate the adjacent endocardial cells to undergo an epithelial-mesenchyme transition (EMT) and form the valves primordia. This process is patterned and only AVC endocardial cells are competent to respond to these signals and initiate EMT 1 . Studies in mice have revealed the genetic network controlling AVC myocardium patterning, including the T-box transcription factors Tbx2 and Tbx3, which repress chamberspecific gene expression in AVC 2,3 and Tbx20, which restricts Tbx2 expression to the AVC in a Smaddependent manner 4 .
Bmp2 (bone morphogenetic protein 2) a transforming growth factor beta (Tgfβ) superfamily member, expressed in AVC myocardium, is sufficient for AVC specification and EMT induction [5][6][7][8] . Bmp2 controls AVC myocardial patterning via Tbx2 activation 9 , and attenuates AVC myocardial proliferation via n-Myc repression 10 . Temporal control of BMP2 signalling is crucial for cardiomyocyte differentiation from mouse ES cells (mESC) in vitro. Thus, inhibition of BMP2 signalling before embryoid body (EB) formation, or in mesodermcommitted (Brachyury-T positive) EBs, induces cardiomyogenesis 11 . We asked whether Bmp2 is able to specify a prospective ventricle as AVC, and what is the effect of Bmp2 on chamber cardiomyocytes. We have generated a transgenic line conditionally expressing Bmp2 and examined the consequences of ectopic Bmp2 expression in heart development. Nkx2.5 Cre -driven Bmp2 myocardial overexpression leads to embryonic death at E15.5, and rescues the AVC specification defect of Bmp2-null embryos. E14.5 Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos show enlarged valves and trabeculae, dilated ventricles and ventricular septal defect. Remarkably, transgenic ventricular endocardium is EMTcompetent both in vivo and in vitro. Gene profile and marker analysis of Nkx2.5 Cre/+ ;Bmp2 tg/+ hearts indicated that cardiac cellular proliferation is increased, and while chamber myocardium gene expression is maintained, its maturation is blocked. We obtained similar results using a second myocardial driver (cTnT Cre ), but not with an endothelial-specific driver (Tie2 Cre ), suggesting that Bmp2 needs to reach a certain threshold to drive EMT and prevent cardiomyocyte maturation. Accordingly, forced Bmp2 expression in vitro stimulated EBs proliferation and blocked their progression into cardiomyogenesis, an effect partially rescued by Noggin. These data demonstrate that Bmp2 is an instructive signal for valve formation and that persistent Bmp2 expression maintains cardiomyocytes in a primitive, proliferative state, which may be relevant for the in vitro expansion of cardiac progenitors for regenerative purposes.

Ectopic myocardial Bmp2 expression disrupts heart morphogenesis
To study the developmental consequences of Bmp2 expression outside of the valve forming field, we generated a transgenic line in which Bmp2 expression is activated upon Cre-mediated removal of a β-Geo-stop cassette (Suppl. Figure S1A and Materials and methods).
We assayed in explants the ability of myocardial Bmp2 to induce ectopic EMT. We first explanted AVC tissue from WT and transgenic hearts as described 1,15 and measured the two-dimensional (2D) and threedimensional (3D) transformation index (TI), which define the migratory and invasive capacity of the explants, respectively 19 . E9.5 WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ AVC explants generated mesenchymal cells invading the collagen gel (Fig. 3a). The 2D and 3D TI values were similar for both genotypes (Fig. 3a, b). Then, we carried out explants with the left ventricle. Cultures from WT ventricles generated a coherent endocardial monolayer surrounding the myocardium (Fig. 3c). In contrast, Nkx2.5 Cre/+ ;Bmp2 tg/+ ventricular explants gave rise to mesenchymal cells that invaded the collagen matrix, similarly to control AVC explants (Fig. 3c, d). These ventricular endocardium-transformed mesenchymal cells had a significantly higher 2D TI and 3D TI than WT explants (Fig. 3c, d). We stained with anti-Cdh5 and α-SMA antibodies, to label the various cell types in the explants 15 . Figure 3e shows that transformed mesenchymal cells in both WT and transgenic AVC explants stain with α-SMA, but not with anti-Cdh5. Endocardial cells of WT ventricular explants, that do not transform (Fig. 3c, d), express Cdh5 and do not express α-SMA (Fig. 3f, left). In contrast, transformed mesenchymal cells of Nkx2.5 Cre/ + ;Bmp2 tg/+ ventricular explants express α-SMA, but not Cdh5 (Fig. 3f, right).

Myocardial Bmp2 gain-of-function stimulates cardiomyocyte proliferation and prevents chamber maturation
The enlarged trabeculae and valves in E14.5, Nkx2.5 Cre/ + ;Bmp2 tg/+ embryos (Fig. 1b) prompted us to measure the surface of the myocardium and AVC, and their total number of cells. The compact myocardium area was similar in WT and transgenic embryos, whereas the trabecular myocardium and AVC regions were larger and contained more cells in E14.5 Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos (Fig. 4a).
(see figure on previous page) Fig. 1 Ectopic myocardial Bmp2 expression disrupts cardiogenesis. a Confocal images of GFP expression plus whole-mount and sectioned in situ hybridisation (ISH) of Bmp2 in E9.5 WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ hearts. White arrowheads mark normal Bmp2 expression in AVC myocardium; black arrowheads mark ectopic Bmp2 expression in chamber myocardium of the transgenic heart. b Top two rows, hematoxylin and eosin (H&E) staining of general and detailed views (insets) of transverse and parasagittal sections of E10.5 WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ hearts, showing mesenchymal cells in the AVC (black arrowheads) and also in the left ventricles of transgenic hearts (white arrowheads). Third row, whole-mount images of E14.5 WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos showing oedema in the dorsal region of the transgenic embryo (white arrowhead). H&E stained transverse sections show general and detailed views of ventricular and valve dysmorphology and a ventricular septal defect in the E14.5 Nkx2.5 Cre/+ ;Bmp2 tg/+ heart (asterisk). Fourth row, details of the atrioventricular (AV) valves (arrowheads) and left ventricle in the WT and transgenic heart; trabeculae (white arrowheads) are larger in the transgenic heart. The arrows mark the WT coronary vessels and the dysmorphic ones in the transgenic heart. c ISH showing Bmp2 mRNA in the AV valves in both E14.5 genotypes (arrowheads) and ectopically throughout the myocardium of Nkx2.5 Cre/+ ;Bmp2 tg/+ ventricles. d Staining for pSmad 1/5 (red), MF 20 (green), isolectin B4 (IB4, white) and DAPI (blue) in E14.5 hearts. General views of transverse heart sections. Note the ventricular septal defect (asterisk) in the transgenic heart. Detailed views show the right ventricle and AV valves of both genotypes, with discrete pSmad 1/5 staining in WT trabecular endocardium (white arrowheads), capillaries in the compact myocardium (yellow arrowheads) and AV valves mesenchyme (arrows, mesenchyme marked with a yellow asterisk). The Nkx2.5 Cre/+ ;Bmp2 tg/+ heart shows widespread pSmad1/5 expression both in ventricles (including cardiomyocytes, arrows) and in AV valves mesenchyme. e pSmad1/5 activation index in the trabecular and compact myocardium (Tm, Cm) of E14.5 WT and transgenic hearts. a atrium, AVC atrioventricular canal, IVS interventricular septum, la left atrium, lv left ventricle, mv mitral valve, ra right atrium, rv right ventricle, tv tricuspid valve, v ventricle; scale bars 200 μm Fig. 2 Expanded expression of the EMT drivers Twist1, Snail, Slug and Sox9, and reduction of Cdh5 and Irx5 expression in the ventricles of Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos. a ISH showing Twist1, Snail, Slug, Cdh5 and Irx5 mRNA in E9.5 WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ hearts. Black and yellow arrowheads mark expression or lack of expression in AVC mesenchyme cells, respectively; white arrowheads indicate expression in chamber endocardium. Cdh5 and Irx5 are transcribed in WT chamber endocardium, while are strongly reduced in the transgenic ventricle (black arrows). Scale bars 200 μm. b Confocal images of E9.5 WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ hearts stained with Sox9 (green), IB4 (white), α-SMA (red) and DAPI (blue). Mesenchyme (1, arrowheads) and endocardial cells (1, arrow) in the WT AVC express Sox9, while IB4-positive endocardial cells of the right ventricle do not (2, yellow arrows). Nkx2.5 Cre/+ ;Bmp2 tg/+ hearts show Sox9-positive mesenchyme (3, arrowheads) and endocardial cells (3, arrows) in AVC, but also in the right ventricle (4, arrowheads and arrows). Scale bars 100 μm. Abbreviations as in Fig. 1 We examined cell proliferation by measuring 5bromodeoxyuridine (BrdU) incorporation in WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ E12.5 hearts. We observed a significantly increased proliferation in AVC valves mesenchyme, but not in aortic or pulmonary valves of transgenic hearts (Suppl. Figure S3C and Fig. 4b). Proliferation was also increased in compact and trabecular myocardium of the right ventricle, and throughout ventricular endocardium (Fig. 4b, c). This elevated proliferation was sustained but less pronounced at E14.5 (Suppl. Figure S3B, C). Thus, Bmp2 overexpression in Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos leads to a proliferative expansion of valves and trabeculae.
The compact myocardium markers Hey2 20 and n-Myc 21 were expanded to trabecular myocardium at E14.5 (Fig. 5a), suggesting impaired chamber maturation in transgenic embryos. Likewise, Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos showed increased Tbx20 expression in AVC valves and expansion to the trabeculae (Fig. 5a). In contrast, Tbx2 was normally expressed in AVC valves (Fig. 5a). These results are in accordance with the reported repression of Tbx2 by Hey2 and Tbx20 in chamber myocardium, resulting in AVC-restricted Tbx2 expression 10,22 . Tbx20 interacts with Bmp/Smad signalling to confine Tbx2 expression to the AVC 23 .
(see figure on previous page) Fig. 3 The AVC and ventricles of Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos undergo EMT in vitro. a Confocal images of AVC explants from E9.5 WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ hearts cultured on collagen gels (top and lateral projections). All explants were stained with phalloidin-FITC (green), anti-α-SMA-Cy3 (red) and DAPI (blue). AVC mesenchymal cells (α-SMA-positive) invade the collagen gel (arrowheads). b Quantification of migrating (2D, endocardial) and invading (3D, mesenchyme) cells in WT and transgenic AVC explants. c Confocal images of stained left ventricular explants (LV). WT LV endocardial cells do not undergo EMT and form a monolayer on the surface of the gel (arrowhead). Transgenic LV endocardial cells transform, migrate and invade the collagen gel (arrowheads in top views and lateral projections). d Quantification of migrating and invading cells in LV explants shows a significantly higher 2D and 3D transformation indexes (TI) in transgenic explants than in WT. TI is the number of migrating (2D) or invading (3D) cells divided by the total number of cells in each explant. m myocardium. t test ** P < 0.01, ***P < 0.005. e, f Confocal images of explants. e The insets show general views of WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ AVC explants stained with anti-Cdh5 (white), anti-α-SMA (red) and DAPI (blue). The large images are a magnification of the area marked in the insets and show mesenchymal cells stained with anti-α-SMA and DAPI, but not with anti-Cdh5 (arrowheads). f The insets show general views of WT and Nkx2.5 Cre/+ ;Bmp2 tg/+ ventricular explants stained with the same antibodies than above (in the WT ventricular explant the myocardium (m) has been removed). The large images are a magnification of the area marked in the insets and show that WT ventricular endocardial cells express Cdh5 (F, white arrows), while transformed mesenchymal cells in transgenic ventricular explants express α-SMA (F, arrowheads). Scale bars 100 μm in insets; 50 μm in large images  Bmp2 gain-of-function in myocardium, but not endothelium, disrupts ventricular development We used the myocardial-specific cTnT-Cre line, active from E8.0 onwards 35 , that caused a phenotype similar to Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos. E10.5 cTnT Cre/+ ;Bmp2 tg/+ mice showed enlarged AVC, and mesenchymal cells in the right ventricle (Suppl. Figure S5A). E14.5 cTnT Cre/+ ; Bmp2 tg/+ embryos had ventricular septal defect, dilated ventricles and coronaries, and thickened valves and trabeculae (Suppl. Figure S5B), and did not progress beyond E17.5 (Supplemental Table S3). cTnT Cre/+ ;Bmp2 tg/+ hearts expressed Bmp2 throughout the myocardium (Suppl. Figure S5B). These embryos showed upregulated Tbx20 expression (Suppl. Figure S5C), expansion of Hey2 to the trabeculae (Suppl. Figure S5C) and reduced Cx40 and Bmp10 expression (Suppl. Figure S5D), indicating impaired ventricular chamber maturation. Coronary vessels were also affected, as the loss of Cx40 expression indicated (Suppl. Figure S5D). Tbx2 and Cx43/Gja1 expression was unaltered (Suppl. Figure S5D).

Bmp2 stimulates proliferation and prevents cardiomyogenesis in vitro
The increased proliferation and impaired cardiomyocyte maturation of Nkx2.5 Cre/+ ;Bmp2 tg/+ mice prompted us to test the ability of R26CAGBmp2-eGFP mouse embryonic stem cells (mESC) to form embryoid bodies (EBs) and differentiate into cardiomyocytes 37 . R26CAGBmp2-eGFP mESCs were transfected with a Cre-expressing plasmid, and GFP-positive clones (Bmp2-mESCs) were identified (Suppl. Figure S6B). Non-recombined clones were used as control mESCs. ELISA revealed around 3.5-fold Bmp2 increase in the culture medium of Bmp2-ESCs (Suppl. Figure S6C). Bmp2-expressing embryoid bodies (Bmp2-EBs) expressed GFP, whether cultured in suspension or plated (Suppl. Figure S6D). Bmp2-EBs at day 3 (d3) and d5 of culture were larger than control EBs (Fig. 6a, b). To determine whether this size difference was due to Bmp2 overexpression, we cultured EBs with different concentrations (see Experimental procedures) of the Bmp antagonist Noggin 38 . Overall, 500 ng/ml Noggin had an effect on Bmp2-EBs, being added from 3 days before EBs formation (d-3) to day 17 (d17). Noggin reduced the size of control and Bmp2-EBs at d3 (Fig. 6a,b) and prevented Bmp2-EBs from growing more than control EBs (Fig. 6a,  b). By d5, Noggin was unable to reduce the size of Bmp2-EBs (Fig. 6a, b), but reduced the size of control EBs (Fig. 6a, b). Phospho-Histone3 (PH3) and BrdU analyses revealed markedly increased proliferation in Bmp2-EBs at d3 (Fig. 6c and Suppl. Figure S6E). These data indicate that Bmp2 overexpression in EBs promotes proliferation and that continuous Noggin-mediated Bmp2 blockade limits this effect.
Our in vitro data show that constitutive Bmp2 overexpression in EBs affects cardiac differentiation. To test whether the increased BMP2 concentration affected cardiogenesis after cardiac specification had taken place, we cultured control R26CAG-Bmp2 mESCs in the presence or absence of human BMP2 (20 ng/ml) from d3 to d17. The beating ability of WT and BMP2-treated EBs was similar. qPCR on samples collected before and after BMP2 addition revealed no differences in BraT or Mesp1 expression between BMP2-treated and untreated EBs (Fig. 6f and Suppl. Figure S6G); however, BMP2 addition decreased the expression of early (Fig. 6f) and late cardiac differentiation markers (Suppl. Figure S6G). This result is consistent with our in vivo results and previous in vitro data 11 suggesting that increased BMP2 signalling following cardiac mesoderm specification prevents cardiomyocyte differentiation.

Discussion
By activating Bmp2 expression throughout the embryonic myocardium, we provide genetic evidence indicating that Bmp2 is an instructive myocardial signal, able to induce the formation of cardiac valve primordia from AVC and non-AVC endocardial cells. Bmp2 gain in the embryonic myocardium promotes cardiac cell proliferation, disrupts valve remodelling and chamber cardiomyocyte patterning and maturation. Our data are informative about the mechanisms underlying mammalian cardiac patterning, and suggest that timely Bmp2 activation may be useful in the ex vivo expansion of immature cardiomyocytes.

Valve primordium formation, Bmp2 and endocardial competence
Explant assays with chicken AVC and ventricles showed that only AVC endocardium was able to undergo EMT 1 . Expression, explants and loss-of-function studies in mice confirmed that Bmp2 is crucial for AVC specification and cushion formation [5][6][7][8] . Our results show that myocardial Bmp2 gain directs ectopic EMT, expansion of EMT markers to ventricles and loss of chamber endocardial identity. Our data overturn the notion that only AVC endocardial cells are "competent" to respond to Bmp2 1 , as ventricular endocardial cells also respond to Bmp2, upregulate EMT drivers and undergo full transformation. Thus, AVC endocardium competence to undergo EMT results from the tightly regulated AVC-restricted Bmp2 signalling, and not from the segregation of EMT-competent and noncompetent endocardial cells in the early embryo.
The inability of forced endothelial-endocardial Bmp2 expression to trigger ectopic EMT, and the slight increase in Bmp2 transcription, suggests that Tie2 Cre -driven Bmp2 levels might not reach a threshold required for promoting EMT outside the AVC. Our results show that forced myocardial Bmp2 expression promotes ectopic EMT, but does not alter the timing of valve primordium formation, as indicates the normal EMT drivers expression in the valves of E14.5 Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos.
Bmp2 has been suggested to regulate AVC myocardial patterning through Tbx2 activation 7 . Hey2 and Tbx20 both directly repress Tbx2 in chamber myocardium, and confine its expression to the AVC through interaction with Bmp/Smad signalling 10,22,23 . Thus, the expanded expression of Hey2 and Tbx20 in E14.5 Nkx2.5 Cre/+ ; Bmp2 tg/+ embryos may explain why Tbx2 expression does not extend to the ventricles, despite being positively regulated by Bmp2 42 . In addition, Bmp10 induces Tbx20 promoter activity in vitro through a conserved Smad1 binding site 43 . We hypothesise that the expanded Smad1/ 5 expression in Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos could induce Tbx20 expression in a Smad-dependent manner, and thus repress Tbx2 in the chambers.

Proliferative and differentiation-inhibitory effects of Bmp2 gain in the myocardium
The increased cellular proliferation in Nkx2.5 Cre/+ ; Bmp2 tg/+ hearts could seem paradoxical, since Bmp2 is normally expressed in the AVC, which is less proliferative than chamber myocardium 44 . Thus, one would expect that myocardial Bmp2 overexpression would attenuate cardiomyocyte proliferation. However, myocardial Bmp2 overexpression leads to increased and expanded expression of genes promoting cardiomyocyte proliferation, such as Hey2, n-Myc, Id2 and Tbx20 20,21,32,33 (Fig. 7). Myocardial Bmp2 overexpression causes a phenotype similar to that caused by myocardial Tbx20 overexpression, which through BMP2/pSmad1/5/8 signalling promotes cardiomyocyte proliferation and maintenance of embryonic characteristics in foetal and adult mouse hearts 45,46 . Therefore, the expanded Tbx20 expression in Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos may reflect a positive feedback loop by which the expanded Smad1/5 expression in Nkx2.5 Cre/+ ;Bmp2 tg/+ embryos could induce Tbx20 expression throughout the heart. Smad4 inactivation with cTnT Cre leads to a hypocellular myocardial wall, due to reduced ventricular cardiomyocyte proliferation. Expression of the Smad4 target n-Myc is downregulated in myocardial Smad4 mutants, as well as that of n-Myc target genes Cyclin D1, D2 and Id2 33 . This negative effect of Bmp signalling loss on cardiomyocyte proliferation is compatible with our data showing the positive effect on cardiomyocyte proliferation of myocardial Bmp2 gain.
Structural, marker and gene profiling analyses revealed that increased myocardial proliferation in Nkx2.5 Cre/+ ; Bmp2 tg/+ embryos is accompanied by defective ventricular chamber maturation, with an expansion of the more primitive and proliferative compact myocardium markers and the downregulation of trabecular markers (Fig. 7). These in vivo data were consistent with our in vitro EB differentiation results showing that constitutive Bmp2 expression stimulates EB proliferation, attenuates cardiac mesoderm specification and prevents cardiomyocyte differentiation. Defective maturation of Bmp2-EBs is reflected in the lack of beating ability, and reduced expression of Nkx2.5, Gata4 and Tbx5. Addition of BMP2 to the medium after cardiac specification did not affect EB beating but did impair cardiac differentiation. These observations are consistent with those showing that transient BMP signalling inhibition induces cardiomyocyte differentiation of mESC 11 . Thus, our data show that widespread myocardial Bmp2 expression maintains chamber myocardium and early cardiac progenitors in a primitive, proliferative state and identify Bmp2 as a potential factor for the expansion of cardiomyocytes in vitro. Studies in zebrafish show that bmp2b overexpression stimulates cardiomyocyte dedifferentiation and proliferation and enhance cardiac regeneration 47 . Likewise, Tbx20 overexpression in adult cardiomyocytes promotes their proliferation and improves cardiac function after myocardial infarction through the activation of multiple proproliferation pathways, including BMP signalling 48 .

Generation of R26CAGBmp2 transgenic line
See the supplemental experimental procedures.

Additional mouse lines
The following mouse strains were used: R26CAGBmp2 tg (this report), Nkx2.5 Cre12 , cTnT Cre35 , Tie2 Cre36 and BMP2 flox49 . For simplicity, R26CAGBmp2 tg/+ is abbreviated in the text and figures as Bmp2 tg/+ . Details of Fig. 7 Myocardial Bmp2 gain promotes ectopic EMT, stimulates cardiac proliferation and disrupts ventricular patterning and cardiomyocyte maturation. a Left, E9.5 WT heart, chamber (ventricles and atria, green and light green) and non-chamber myocardium (AVC, blue) are specified. Bmp2 expression in AVC myocardium drives EMT via Twist1, Snail and Slug activation in AVC endocardium (purple). The Irx5-positive ventricular endocardium is coloured in red and the atrial endocardium in yellow. Right, E14.5 WT heart, the compacting ventricular myocardium is formed by an outer compact myocardium (Hey2-, n-Myc-positive, light green) and an inner trabecular myocardium (Bmp10-, Cx40-, Sema3a-positive, dark green). Tbx20 is expressed in compact and weakly, in trabecular myocardium. The maturing AVC valves (blue) express Bmp2 and Tbx2. B; left, E9.5 Nkx2.5 Cre/+ (or cTnT Cre/+ );Bmp2 tg/+ transgenic heart. Bmp2 is normally expressed in AVC myocardium (blue) and ectopically in ventricular myocardium (light blue), driving Twist1, Snail and Slug expression in the Irx5-negative ventricular endocardium (purple), causing ectopic EMT. Right, E14.5 Nkx2.5 Cre/+ (or cTnT Cre/+ );Bmp2 tg/+ transgenic heart. Bmp2 is expressed in an expanded AVC valve region (blue) and throughout the ventricles (green), leading to increased and expanded Tbx20, Hey2, n-Myc and Id2 expression in this tissue. As a consequence, cardiomyocyte proliferation is increased and chamber patterning/maturation is disrupted. The discontinuous arrow represents a potential positive feedback loop between Tbx20 and Bmp2 and the suggested negative regulation of Tbx2 by Tbx20 and Hey2. The asterisk indicates a ventricular septal defect. Abbreviations as in Fig. 1 genotyping will be provided upon request. Animal studies were approved by the CNIC Animal Experimentation Ethics Committee and by the Community of Madrid (Ref.

Immunohistochemistry
For details about antibodies and protocols see supplemental experimental procedures.

Proliferation analysis and quantification on developing hearts
Cell proliferation in the developing heart was evaluated from BrdU incorporation 50 . For details see supplemental experimental procedures.

AVC and left ventricle explants
E9.5 WT and transgenic AVCs were harvested in sterile PBS. Left ventricles (lv) were carefully dissected, avoiding contamination with AVC tissue. Explants were placed on collagen gels with the endocardium face down 6 . For details see supplemental experimental procedures.

Explant culture quantification
For details see supplemental experimental procedures.

Confocal imaging
Confocal images of E9.5 whole-embryos, stained explants and tissue sections were acquired with a Nikon A1R laser scanning confocal microscope and NIS-Element SD Image Software. Images of stained explants were collected as z-stacks. Z-projections and lateral sections were assembled using ImageJ. Images were processed in Adobe Photoshop Creative Suit 5.1.

RNA-Seq
Hearts of E14.5 WT and Nkx2.5 CRE/+ ; Bmp2 tg/+ embryos (12 per genotype) were isolated on ice-cold PBS and the atria removed. Tissue was homogenised in Trizol (Invitrogen) using a Tissuelyzer (Qiagen). RNA was pooled into three replicates per genotype. For details see supplemental experimental procedures.

Accession number
Data are deposited in the NCBI GEO database under accession number GSE100810.