Endoglin is dispensable for angiogenesis, but required for endocardial ...

Developmental Biology 335 (2009) 66–77

Contents lists available at ScienceDirect

Developmental Biology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / d e v e l o p m e n t a l b i o l o g y

Endoglin is dispensable for angiogenesis, but required for endocardial cushion formation in the midgestation mouse embryo Aya Nomura-Kitabayashi a,1, Gregory A. Anderson a,b,1, Gillian Sleep a, Jenny Mena a, Amna Karabegovic a, Sharon Karamath a, Michelle Letarte b,c, Mira C. Puri a,b,⁎ a b c

Sunnybrook Research Institute, Sunnybrook Health Sciences Centre, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N-3M5 Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada The Hospital for Sick Children and The Heart and Stroke Foundation Richard Lewar Centre of Excellence, 555 University Ave., Toronto, Ontario, Canada M5G-1X8

a r t i c l e

i n f o

Article history: Received for publication 21 May 2009 Revised 27 July 2009 Accepted 15 August 2009 Available online 21 August 2009 Keywords: Angiogenesis Endocardial-to-mesenchymal transition Atrioventricular canal Endoglin Hereditary Hemorrhagic Telangiectasia

a b s t r a c t Vascular patterning depends on precisely coordinated timing of endothelial cell differentiation and onset of cardiac function. Endoglin is a transmembrane receptor for members of the TGF-β superfamily that is expressed on endothelial cells from early embryonic gestation to adult life. Heterozygous loss of function mutations in human ENDOGLIN cause Hereditary Hemorrhagic Telangiectasia Type 1, a vascular disorder characterized by arteriovenous malformations that lead to hemorrhage and stroke. Endoglin null mice die in embryogenesis with numerous lesions in the cardiovascular tree including incomplete yolk sac vessel branching and remodeling, vessel dilation, hemorrhage and abnormal cardiac morphogenesis. Since defects in multiple cardiovascular tissues confound interpretations of these observations, we performed in vivo chimeric rescue analysis using Endoglin null embryonic stem cells. We demonstrate that Endoglin is required cell autonomously for endocardial to mesenchymal transition during formation of the endocardial cushions. Endoglin null cells contribute widely to endothelium in chimeric embryos rescued from cardiac development defects, indicating that Endoglin is dispensable for angiogenesis and vascular remodeling in the midgestation embryo, but is required for early patterning of the heart. © 2009 Elsevier Inc. All rights reserved.

Introduction Development of the mammalian embryonic vasculature begins during gastrulation with the emergence of VEGFR2 (Flk-1) expressing mesodermal progenitors that give rise to hematopoietic and endothelial cells (ECs), as well as cardiac myocyte and mural cell progenitors (Huber et al., 2004; Kattman et al., 2007). In addition, formation of the vascular network is tightly coupled to the differentiation and morphogenesis of the heart to achieve unidirectional blood flow. Recent studies have highlighted the dependence of angiogenesis on hemodynamic flow in avian (le Noble et al., 2004) and mouse embryos (Lucitti et al., 2007), indicating that vascular patterning depends critically on the coordinated timing of endothelial/hematopoietic cell differentiation and onset of cardiac function. Endoglin is a transmembrane accessory receptor for several members of the TGF-β superfamily including TGFβ1, TGFβ3, activin, BMP2, BMP7 and BMP9 (Barbara et al., 1999; David et al., 2007). It is

⁎ Corresponding author. Sunnybrook Health Sciences Centre S-234, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N-3M5. Fax: +1 416 480 5703. E-mail address: [email protected] (M.C. Puri). 1 Equal contribution. 0012-1606/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ydbio.2009.08.016

expressed in the progenitors to cardiovascular lineages in the embryo, becoming restricted to ECs from early gestation to adult life (Allinson et al., 2007; Bourdeau et al., 2000; Ema et al., 2006; Hirashima et al., 2004). The critical role of Endoglin in the vascular system is highlighted by the identification of human ENDOGLIN (ENG) as the gene mutated in Hereditary Hemorrhagic Telangiectasia type 1 (HHT1). Heterozygosity for mutant alleles of Endoglin is associated with arteriovenous malformations (AVMs) that can lead to stroke and severe internal hemorrhage(Abdalla and Letarte, 2006; Bourdeau et al., 2001). Due to the clinical importance of HHT, previous investigations of Endoglin function have concentrated on its role in blood vessel endothelium, where it regulates proliferation, migration, and capillary tube formation (Jerkic et al., 2006) and also modulates vascular tone in response to hemodynamic stress (Jerkic et al., 2004; Toporsian et al., 2005). The essential role of Endoglin in cardiovascular development was previously documented (Arthur et al., 2000; Bourdeau et al., 1999; Carvalho et al., 2004; Li et al., 1999; Sorensen et al., 2003). Mice homozygous for a null allele of Endoglin (Eng) mice do not survive embryogenesis and succumb to circulatory arrest and hemorrhage by embryonic day (E)10.5 of gestation. These embryos display numerous lesions in the cardiovascular tree, including a vasculature lacking integrity, with perturbed mural cell association and/or differentiation as well as improper segregation of arterial

A. Nomura-Kitabayashi et al. / Developmental Biology 335 (2009) 66–77

and venous cells leading to shunting between the major vessels. In addition, the yolk sac vasculature showed incomplete vessel branching, severe dilation and hemorrhage. Importantly, Endoglin-deficient embryo hearts do not undergo normal morphogenetic looping, and display reduced trabeculation of the ventricles, and pericardial edema indicative of circulatory arrest. These defects coincide with abrogated endocardial cushion formation in the atrioventricular canal (AVC) (Bourdeau et al., 1999). In this process, reciprocal signaling between the endocardial and myocardial cell layers within the AVC induces transformation of endocardial cells into mesenchymal cells that migrate and invade the localized swelling of the extracellular matrix (Armstrong and Bischoff, 2004). The atrioventricular (AV) mesenchyme subsequently differentiates into the fibrous tissues of the valves (Person et al., 2005). In support of a role in heart morphogenesis, Endoglin is expressed in endocardium and AVC mesenchyme during development (Arthur et al., 2000; Bourdeau et al., 1999; Ema et al., 2006). Analysis of Eng-null embryos was instrumental in determining the overall importance of this receptor during development and in the identification of the developmental stage and cell types primarily affected by the mutation. However, embryonic lethality with complex abnormalities in interdependent cell lineages has hindered determination of the specific stage at which this receptor plays a role in patterning the cardiovascular tree. To address this problem we used Eng-deficient and control embryonic stem (ES) cells in vivo in chimeric mice to investigate the developmental potential and cellular characteristics of Eng-deficient cardiovascular lineage cell types. Our studies have uncovered a cell autonomous requirement for Endoglin in the formation of endocardial cushions by endocardial to mesenchymal transition (EMT), and patterning of the myocardium of the AVC. We also show that Endoglin is not required cell-autonomously for angiogenesis in the midgestation embryo, and that vessel defects in the absence of Endoglin are attributed to abnormal cardiac morphogenesis. Materials and methods Derivation of ES cells Eng+/− ES cells were previously generated (Bourdeau et al., 1999) by gene targeting the parental wild-type 129/Ola-derived E14 ES cell line, replacing 609 base pairs (bp), including exon 1 of Eng and its initiation codon, with the Escherichia coli LacZ gene. Eng−/− ES cells were generated from one Eng+/− clone by selection with high concentration G418 (5.0 mg/mL) for 10 days (Nagy et al., 2003). PCR was used to screen resistant colonies for wild-type and targeted alleles (Bourdeau et al., 1999) (Supplementary Fig. 1). Two independently derived Eng−/− clones and two independent Eng+/− clones (unrecombined after identical selection) were used for ES cell differentiation experiments and chimera aggregations. Aggregations and chimeric embryo analysis Chimeric mice were produced by aggregation of wild-type host embryos (8-cell morulae) with Eng+/− and Eng−/− ES cells (embryo ← → ES cell aggregation) as previously described (Wood et al., 1993). All chimeric embryos were stained for X-gal as previously described (Puri and Bernstein, 2003) to detect donor derived βgalactosidase expressing cells. Embryos were fixed and embedded in paraffin, sectioned and counterstained with Nuclear Fast Red. Extent of chimerism for all individual chimeras was determined using quantitative genomic PCR (Table 1) and glucose phosphate isomerase (GPI) as previously described (Puri and Bernstein, 2003) (Supplementary Fig. 2).

67

Table 1 Summary of chimera analysis. Aggregation no.

Total Eng+/− Chimeric Embryos

1 2 3 Total

8 24 14 46

Total Eng−/− Chimeric Embryos Type 1

Type 2

Type 3

11 14 20 45

1 2 4 7

5 3 37 45

Determination of donor-derived ES cell contribution to chimeric mice by real-time quantitative genomic PCR PCR primers amplify a region of the Eng mutant allele (donor cells) and the Tie2 allele for baseline genome (all cells). The contribution of donor cells was quantified relative to this. Chimera embryo yolk sacs were digested in PCR lysis buffer and proteinase K overnight at 55 °C. 100 ng of nucleic acid, quantified by Nanodrop (Thermo Scientific) was used for quantitative real-time PCR for each sample. The amount of donor cell DNA was determined by comparing the amount of Endoglin-LacZ amplicons in each sample to a standard curve created from a dilution series of pooled DNA from digested yolk sacs of Eng−/− embryos. The amount of host cell DNA was determined by comparing the amount of Tie2 amplicons in each sample to a standard curve created from the same dilution series of pooled Eng−/− DNA. The number of Eng-LacZ amplicons relative to the number of Tie2 amplicons in each sample determined the percent donor cell contribution. The primer sequences were as follows: LacZF (5′TATCTCTGGATACCGGATAAG3′); LacZR (5′TGTAAAACGACGGGATCATCG3′); Tie2F (5′AAGAGCGAGTGGACCATGCGA3′), Tie2R (5′AGGAGCAAGCTGACTCCACAG3′). Quantitative chimera contribution data are summarized in Table 1 and shown in detail in Supplementary Table 1. Embryoid body formation and in vitro angiogenesis assay ES cells were cultured on irradiated murine embryonic fibroblasts (MEFs) in ES cell medium and 2000 U/mL recombinant leukemia inhibitory factor (LIF; Chemicon International). Differentiation of ES cells was induced by withdrawal of LIF and by aggregating the cells into embryoid bodies (EBs) by culturing 1200 cells/drop on the lid of a non-adherent 96-well tissue culture dish for 4 days. Individual EBs were plated into one well of an eight-well glass culture slide (Becton Dickinson (BD) Falcon) for 6 days. Flattened EBs were stained with a monoclonal rat anti-mouse PECAM-1 antibody (CD31, BD Pharmingen), overnight at 4 °C, followed by incubation with biotinylated goat anti-rat IgG (Vector Laboratories). Atrioventricular cushion explant culture Timed matings between Eng+/− mice were established with the morning of a vaginal plug defined as E0.5. All embryos used were 21–25 somites. AVC explant cultures were established as described previously (Camenisch et al., 2002). The AVC and adjacent myocardium from the embryos were dissected and explanted onto type I collagen (Sigma-Aldrich) and allowed to attach for 12 h. Explant cultures were scored for mesenchymal cell invasion at 24 and 48 h and photographed using an inverted Leica DM IL microscope. Migrated mesenchymal cells were identified and counted below the gel surface (Runyan and Markwald, 1983) before the genotype of the respective embryo was determined. Quantitative real-time gene expression analysis Heart tissue from E9.5 embryos collected from Eng +/− heterozygote intercrosses were pooled according to genotype

68


(Eng+/+, Eng+/− and Eng−/−), and total RNA isolated using Trizol (Invitrogen) and cDNAs were generated using the Superscript III Reverse Transcriptase kit (Invitrogen). The quantity of cDNA in each sample was quantified using Nanodrop (Thermo Scientific). Realtime PCR was performed on an ABI Prism 7000 (Applied Biosystems) with primers listed in Supplementary Table 2 and normalized to βactin. Each qPCR was performed in triplicate, and each test was performed at least five times with pools of embryonic tissue from separate littermate collections. A two-tailed Student's t-test (for parametric data with equal variances) was used for statistical analysis. Whole mount RNA in situ hybridization Analysis of gene expression in whole embryos was performed as previously reported (Saga et al., 1996). In brief, digoxigenin (DIG)labelled RNA probe was used to detect mRNA, and probe/mRNA hybrid was detected by alkaline-phosphatase (AP)-labelled anti-DIG antibody (Roche) with BCIP/NBT chromogen (DAKO). Three to six somite-number matched mutant and wild-type embryos were used for each probe. Proliferation and apoptosis analysis Proliferating and apoptotic cells were detected with antiphosphohistone H3 (PHH3) antibody (Cell Signalling, 1:100 dilution) and in situ cell death detection kit, TMR red (Roche), respectively. Alexa 488 conjugated anti-rabbit antibody (Invitrogen/Molecular Probes, 1:250 dilution) and Topro3 (Invitrogen, 1:1000 dilution) were used to detect PHH3 and nuclei, respectively. Embryos were embedded in paraffin, sectioned (7 μm), and imaged on a Zeiss LSM510 confocal microscope. Three wild-type and three mutant embryos were sectioned and two or three sections including AVC were compared from each embryo. Results Endoglin-deficient cells contribute to angiogenesis in chimeric embryos Previous analysis of Eng-deficient embryos demonstrated defects in angiogenesis both in the yolk sac and in the embryo proper (Arthur et al., 2000; Bourdeau et al., 1999; Li et al., 1999) suggesting that Eng-null ECs are compromised in their ability to expand and mature beyond the early differentiation stage of angiogenesis, leading to hemorrhage, growth delay and circulatory arrest. To test whether there is a cell autonomous requirement for Endoglin in angiogenic sprout formation, we compared the ability of Engdeficient vs. control embryonic stem (ES) cells to form EC sprouts in chimeric embryos. Generating chimeric (or mosaic) animals composed of mixtures of wild-type and mutant cells provides one means to analyze genes that are required in different tissues and at multiple stages of development. Chimeric mice were produced by aggregation of wildtype host embryos and either Eng+/− or Eng−/− ES cells, where the host cells rescue the cellular deficiencies conferred by the mutant donor population, thus allowing the analysis of the behaviour of mutant cells in a competitive environment. Since the Engnull allele was produced by insertion of LacZ gene into the first translated exon (Bourdeau et al., 1999), mutant and heterozygous donor cells that normally express Eng are stably labeled with βgalactosidase, allowing their identification in chimeric animals following X-gal staining. Eng−/− embryos die by E9.5–10.0, therefore we analyzed Eng-null and control chimeric embryos at E10.5–11.5, to follow the fate of mutant vs. control cells in embryos that had been rescued of the deleterious effects of Eng loss of function (Table 1; Fig. 1). In addition

to X-gal staining, we determined the extent of chimerism using a quantitative PCR assay to determine the relative presence of the mutant allele (Supplementary Table 1), or glucose phosphate isomerase (GPI) Assay to distinguish donor vs. host cells carrying GPI-AA vs. GPI-BB isoforms, respectively (Supplementary Fig. 2). Previous studies (Bourdeau et al., 1999) and Supplementary Fig. 3 indicate that LacZ is expressed equally robustly in Eng+/− and Eng−/− embryos, validating that β-galactosidase marks cells in which the Endoglin promoter is active and that the reporter gene expression is not affected by loss of Endoglin function. E10.5 chimeric embryos derived from Eng+/− donor cells were all developmentally normal and displayed a range of donor cell contribution (n = 46) (Figs. 1A, E, I). In contrast, E10.5 embryos derived from aggregation with Eng−/− cells displayed one of three distinct patterns: (Type 1) mosaicism of b20% donor cells (Figs. 1B, F, J) (n = 45); (Type 2) mosaicism of 20–50% donor cells (Figs. 1C, G, K) (n = 7) and (Type 3) mosaicism of N50% donor cells (Figs. 1D, H, L) (n = 45). Although Type 1 and Type 2 embryos were developmentally normal, all Type 3 embryos recapitulated the Eng-null phenotype, displaying growth delay, incomplete vessel development, and pericardial edema (Figs. 1D, H, L). Types 1 and 2 embryos show that Eng−/− cells have the ability to contribute to angiogenic sprouting in many tissues. Comparison of X-gal staining between Eng+/− and Eng−/− Type 2 (rescued) chimeras demonstrated that Eng-null cells contribute to both intersomitic sprouts from the dorsal aorta, and branching vessels from the cardinal vein in the cranial region (Figs. 1C, G, K). Furthermore, Eng−/− ECs can migrate and sprout into the neural ectodermal tissue of the developing brain to an equivalent extent as heterozygous donor cells. (Figs. 2A, B). Also, sections of X-gal-stained yolk sac tissue from mutant and heterozygous chimeric embryos demonstrate that Eng-null cells also contributed to the endothelium of the yolk sac blood islands (Figs. 2C, D). All Type 1 and Type 2 wild-type← → Eng−/− chimeras such as those shown in Figs. 1B, C and 2B, D were rescued of the deleterious phenotypes seen in Eng-null embryos or in Type 3 chimeric embryos with a high contribution of donor cells (Fig. 1D). Earlier studies implicated Endoglin in modulating TGFβ signaling in the yolk sac vasculature as a primary defect in Eng-null mice (Carvalho et al., 2007; Carvalho et al., 2004). Furthermore, loss or gain of function of the TGFβ type II receptor (TβRII) signaling in the yolk sac in vivo demonstrated that the level of TGFβ signaling in the extra-embryonic mesoderm is critical for development of the blood islands, both for proper morphogenesis of the primary vascular plexus and for differentiation of the primitive hematopoietic system (Goumans et al., 1999). In rescued chimeric embryos we did not see a difference in the contribution of Eng-null cells vs. control heterozygous cells, nor did we observe the characteristic block in remodeling of the vessels in the presence of a large proportion of Eng−/− cells (Figs. 2C, D). Endoglin is not required for EC differentiation in vitro To further explore the role of Endoglin in angiogenic sprout formation independent of the requirement for intact circulation, we compared the ability of Eng-deficient ES cells to form EC sprouts after differentiation into embryoid bodies (EBs). We cultured Eng−/− and control (Eng+/− and Eng+/+) ES cell lines in the absence of LIF, using the hanging drop method to form EBs containing mesodermal progenitors. As shown in Figs. 3A–C, mutant and control PECAM-1 stained ECs derived from embryoid bodies, formed sheet-like vascular structures as well as interconnected angiogenic sprouts. PECAM-1 stained, flattened cultures were scored into one of three categories: those displaying no angiogenic sprout formation from a vasculogenic region and those showing fewer or greater than 10


69

Fig. 1. Endoglin heterozygous and mutant cells contribute to angiogenic sprouting in the chimeric embryos at E10.5. Whole mount analysis for donor cell content of ECs from control Eng+/− (A, E, I) and Eng−/− (B–D, F–H, J–L) ES cells at E10.5 after staining with X-gal. Chimeric mice from aggregation of wild-type (WT) and Eng+/− ES cells (WT ← → Eng+/−) develop normally (A) and display donor cell contribution to angiogenic sprouting in the head (E), and intersomitic region (I). Chimeric mice from aggregation of wild-type and Eng−/− ES cells (WT← → Eng−/−) display one of three phenotypes. Type 1 embryos contained less than 20% mutant donor cell contribution developed normally (B), but display low donor cell contribution to vessels in the head (F) or intersomitic region (J). Type 2 embryos contained 20–50% mutant donor cell contribution and were developmentally normal (C), and displayed mutant donor cell contribution to vessels in the head (G), intersomitic region (K). High mutant donor cell contribution produces a deleterious knockout phenotype in Type 3 chimeras (D, H, L) with pericardial edema, growth delay and incomplete vessel remodelling in the head. Arrows indicate donor cell contribution to the indicated tissues (scale bar: 1 mm).

angiogenic sprouts per vasculogenic region, respectively. As shown in Fig. 3D, all genotypes of ES cells produced each type of flattened culture to an equivalent extent. These findings are consistent with the ability of Eng-null ES cells to form vascular network structures in vivo in chimeras and support the view that Endoglin is not required for the early steps of EC differentiation, angiogenic sprouting and network development in the midgestation embryo. Endoglin null cells do not contribute to endocardial cushion mesenchyme Previous studies demonstrated that embryos lacking Endoglin display a defect in cardiac morphogenesis, including incomplete or absent myocardial trabeculation and AV cushion development (Arthur et al., 2000; Bourdeau et al., 1999). This finding is consistent with expression of Endoglin in endocardial cells and in the mesenchymal derivatives that form following endocardial to mesenchymal transition (EMT) in the atrioventricular and outflow tract regions (Bourdeau et al., 1999; Jonker and Arthur, 2002; Qu et al., 1998). Due to widespread defects in the vascular tree, however, it was not clear to what extent cardiac morphogenesis was affected by other circulatory defects leading to overall delay in morphogenesis including that of the heart. Since analyses of EBs in vitro, and chimeras in vivo suggested that angiogenesis is normal

in the absence of Endoglin, we reasoned that heart morphogenesis may be the primary impediment to development in Eng−/− embryos. To address this hypothesis we examined the X-gal staining pattern of mutant and control donor cells in the endocardium and AV cushions of chimeric embryos. As shown in Figs. 4A, C, E, widespread Eng+/− cell contribution is observed for endocardial cells lining the atrial and ventricular myocardium, and in the AVC mesenchymal cell cushion. In contrast, chimeras from Eng−/− cells displayed a range of patterns of contribution. All Type 3 embryos in which overall donor cell contribution was high (N50%) displayed growth retardation and abrogation of cardiac looping, as previously described for Eng knockout embryos (Figs. 1D, L and 4B, D). Furthermore, sections show that when high donor endocardial cell content is evident throughout the atrium and ventricle, the endocardial cushions in the AVC region are severely reduced (Fig. 4F). In contrast, Type 1 embryos with low donor cell contribution (20%) showed few donor cells in the endocardial compartment or cushion mesenchyme (Figs. 4G, I), despite evidence of contribution to other vascular structures in the embryo, albeit low (Fig. 4G, arrowheads). Type 2 embryos, which are developmentally rescued, show that while donor derived β-galactosidase positive cells are present in the endocardium overlying both the AVC and OFT, almost none are

70


Fig. 2. Eng-null ECs display normal sprouting angiogenesis in the neural tube and yolk sac. Sections of X-gal-stained chimeric embryos demonstrate sprouting into the neural tube by both mutant (B) and control (A) ECs (arrows), and contribution to yolk sac vasculature to a similar extent by both mutant (D) and control (C) cells. Yolk sac blood islands composed almost entirely of Eng−/− cells show normal diameter, and cellular organization of endothelium, endoderm and mesothelial cells (scale bar: 100 μm).

present in the mesenchyme (Figs. 4H, J), suggesting that only wildtype endocardial cells could efficiently migrate, invade and transition to the mesenchymal lineage in this cellular context. Therefore, the exclusion of Eng−/− cells from this structure in all E10.5 chimeric embryos that were visibly rescued of the deleterious features of the null phenotype suggests a stringent requirement for Endoglin specifically in the formation of the AV cushion mesenchyme. The presence of donor derived Eng−/− endocardial cells, yet virtual absence from the cardiac cushion mesenchyme

suggests a delay in initiation of the EMT process, or in migration of transformed endocardial cells. Taken together with the donor–host contribution ratios, these observations suggest that if there are only a small number of Eng−/− donor cells contributing to the heart, wild-type host cells can overtake the deficient mutant cells, leading to proper formation of the endocardial cushions. However, in a Type 3 chimera where mutant donor cells exceed a critical number in the endocardium, the endocardial cushions fail to expand properly (Figs. 4K–N).

Fig. 3. Eng-deficient ECs form vascular networks in vitro. (A) Eng+/+, (B) Eng+/−, and (C) Eng−/− embryoid bodies (EBs) stained with PECAM-1 display angiogenic sprouts and network formation (arrows) from vasculogenic regions (⁎) to similar extent in vitro. (D) EBs were enumerated according to the criteria: no angiogenic sprout formation from vasculogenic region; fewer than 10 angiogenic sprouts forming from vasculogenic region; greater than 10 angiogenic sprouts forming from vasculogenic region.


71

Fig. 4. Eng-null cells do not contribute to endocardial cushions. (A) Developmentally normal chimeric WT← → Eng+/− embryo at E10.5 and (C) looping heart. (B, D) Type 3 WT← → Eng−/− embryo with growth restriction, abnormal cardiac looping and pericardial edema. White lines in (A, B) indicate cutting angle for sections shown in (E, F). (E) Transverse section of WT← → Eng+/− embryo AVC shows donor cell contribution to endocardial cushion mesenchyme (black arrow). (F) Hypoplastic endocardial cushions (red arrow) in Type 3 WT← → Eng−/− embryo. Red arrowheads show reduced trabeculation in Type 3 chimera compared to control (E, red arrowheads). (G, H) Sagittal sections of hearts from rescued Type 1 and Type 2 WT← → Eng−/− chimeric embryos with normal endocardial cushion formation. Mutant donor cells are present in endocardium and capillary ECs (arrowheads) but not in AVC or outflow tract cushion mesenchyme. (I, J) high power view of AVC from embryos in G, H (box region), demonstrating predominantly host wild-type cell contribution to mesenchyme in Type1 and Type2 chimeras (red arrows), with donor cells in endocardium (black arrows). All embryos stained with X-gal. (K–N) Interpretation of endocardial cushion formation in chimeric embryos. (K) A subset of endocardial cells (e) in AVC populates the cardiac jelly (cj) to form cushion mesenchyme. (L) In WT← → Eng+/− chimeras, both donor and host cells can contribute to endocardial cushion. (M) In WT← → Eng−/− chimeras with low donor contribution, Type2, only WT host cells populate the endocardial cushion, whereas (N) in WT← → Eng−/− chimeras with high donor contribution, Type3, EMT is blocked since WT cells are not sufficiently available. Darker blue signifies Endoglinnull cells and star shaped cells are mesenchymal cells after EMT (scale bar: A–D 1 mm; E, F 200 μm; G, H 400 μm; I, J 100 μm ). lv: left ventricle; la: left atrium; v: ventricle; a: atrium; da: dorsal aorta, fg: foregut; ec: endocardial cushion; o: outflow tract; ma: mandible; e: endocardium; m: myocardium; cj: cardiac jelly.

Endocardial-to-mesenchymal transition (EMT) is delayed in Eng-deficient embryos To further address the requirement for Endoglin in EMT, we compared the ability of Eng-deficient vs. control embryo AVC tissues to undergo EMT ex vivo. AVC tissue was dissected from embryos derived from Eng +/− intercrosses and cultured on

hydrated collagen I gels. We scored the ability of endocardial cells from Eng+/+, Eng+/− and Eng−/− embryos to migrate into the gel and undergo mesenchymal transition by phase contrast microscopy at 24 and 48 h following plating (Figs. 5A–F), and enumerated according to whether more or less than 50 cells had migrated at each time point. As shown in Fig 5G, after 24 h approximately 50% of Eng+/+ explants showed that greater than 50 mesenchymal cells

72


Fig. 5. Endothelial to mesenchymal transition is delayed in Eng-null embryos. Mesenchymal transformation and migration in AVC explants from Eng+/+ (A, D), Eng+/− (B, E) and Eng−/− (C, F) at 24 h (A–C) and 48 h (D–F) after plating onto hydrated collagen I gel. Invaded cells were counted using phase microscopy for cells below the explant. (G) Individual explants were scored according to whether greater or less than 50 cells were found migrated into the gel at 24 and 48 h. (H) Quantitative RT-PCR for endocardium (VE-Cadherin, Nfatc1) and endocardial cushion (Sox9, Snai1, Snai2) transcripts, and VEGF, normalized to β-actin expression, using RNA from dissected hearts of stage and size matched embryos at E9.5. Error bars represent the standard error of the mean of each sample. Data are representative of assays performed in triplicate, each using at least four litters of embryos. ⁎ Indicates statistical significance P b 0.05; VEGF P = 0.002, Snai1 P = 0.003).


73

had invaded into the gel, whereas none of the Eng−/− cultures showed extensive migration. After 48 h, mesenchymal cells from Eng−/− explants were observed in the gel, albeit fewer cultures showed greater than 50 transformed cells when compared to control Eng+/+ and Eng+/− (Fig. 5G). The difference in invading cell number between Eng mutant and control explants was more apparent at 24 h, suggesting that while EMT is not completely abrogated in vitro in Eng−/− explants, either initiation of migration or onset of transformation is delayed, consistent with findings using chick explant cultures (Mercado-Pimentel et al., 2007). In addition, we observed that while Eng+/+ and Eng+/− explants did not attach to the collagen I matrix in approximately 6% and 8% of cultures, respectively, Eng−/− explants did not attach in almost 30% of cases (data not shown). This observation suggests that Eng-deficient endocardium may be compromised in its ability to interact with extracellular matrix components of the cardiac jelly. Furthermore, Eng-null cells that had invaded the collagen gel by 48 h displayed a rounded morphology in contrast to the elongated cell shape of Eng+/+ and Eng+/− explant derived invading cells. Taken together with the finding that Eng-null cells do not contribute efficiently to endocardial cushions in chimeric embryos, these data indicate that Eng-null endocardial cells have a reduced ability to respond to myocardial cues for invasion of the cardiac jelly, and transformation to cushion tissue. To further examine the defect in EMT in Eng-deficient hearts we performed quantitative real-time PCR analysis of Eng-null heart tissue (Fig. 5H). Eng-null hearts showed significantly diminished expression of Snai1 (P = 0.003), a transcription factor required for EMT downstream of TGFβ family signaling (Niessen et al., 2008). In contrast, Snai2, involved in EMT, and Nfatc1 required for cushion development (Chang et al., 2004) were not significantly affected in Eng-null embryo heart. The expression of VE-Cadherin (Fig. 5H) which modulates endothelial cell–cell contact (Taddei et al., 2008), and is repressed by Snai1 and Snai2 (Niessen et al., 2008) was not different between Eng-null and control embryos, suggesting that expression of Snai2 modulates normal levels of VE-cadherin in Engnull embryos despite the reduced expression of Snai1. VEGF expression was found to be significantly upregulated in Eng-null hearts, relative to control littermates. Previous genetic and explant studies have demonstrated that VEGF is a potent negative regulator of EMT (Dor et al., 2001) and is significantly upregulated in embryos exposed to hypoxia (Ream et al., 2008). Abnormal patterning of the AVC in Eng-null embryos at E9.5 In addition to providing a signal for EMT, the myocardium of the AV region is phenotypically and functionally distinct from chamber myocardium (Christoffels et al., 2000). Disruption of the boundary between the ventricle and atrium leads to defects in cushion formation and embryonic lethality (Harrelson et al., 2004; Kokubo et al., 2007). Comparison of E9.5 Eng−/− and Eng+/+ embryos, matched for somite number, revealed a narrower constriction in the AVC region as well as size variations of both atrial and ventricular chambers in mutants, prompting us to examine gene expression in this region in more detail. BMP2, which is expressed

Fig. 6. Abnormal AV constriction morphology and gene expression in Eng−/− embryos at E9.5. Whole mount in situ hybridizations for the indicated genes in Eng+/+ (A, C, E, G) and Eng−/− (B, D, F, H) hearts. Note that BMP2 expression is reduced (A, B), and Tbx2 expression is absent in mutants (C, D), and chamber-specific gene expression of Nppa (atrium and ventricle) and Hesr2 (ventricle) domains expand into AVC (E, F and G, H, respectively). Brackets indicate gene expression domains and arrowheads show the border of gene expression between ventricle and AVC. (I, J) Tunel and phosphohistone H3 (PHH3) immunofluorescence staining of Eng+/+ (I) and Eng−/− (J) embryos. Arrows show the endocardial layer in AVC. (K, L) Alcian Blue staining for acidic glycosaminoglycans demonstrates the presence of cardiac jelly in Eng−/− and control embryos (scale bar: H, L 200μm).

74


specifically in the AVC myocardium, and is required for formation of endocardial cushions in the mouse (Ma et al., 2005; RiveraFeliciano and Tabin, 2006; Sugi et al., 2004), appeared at a reduced level in Eng−/− embryos, but was still restricted to the interchamber region (n = 6) (Figs. 6A, B). In addition, expression of Tbx2, a downstream target of BMP2 and transcriptional repressor of chamber-specific genes in the AVC, was absent in Eng−/− embryos (n = 3) (Fig. 6D). Consistent with Tbx2 downregulation, the domains of expression of the chamber myocardial specific genes Nppa (Anf) (n = 3) and Hesr2 (n = 3) were expanded into the AVC region in Eng knockout embryos (Figs. 6F, H) (Harrelson et al., 2004; Kokubo et al., 2007). These results indicate that although early specification of the AVC myocardium occurs in the absence of Endoglin, further expansion is not sustained, and/or patterning is aberrant. As a result, chamber-specific markers are expressed inappropriately, blocking subsequent morphogenesis of the AVC region. Importantly we did not observe differences in cell proliferation or apoptosis in the endocardium between mutant and control embryos (Figs. 6I, J). Since extracellular matrix deposition within the AVC is required for cushion formation, we examined the extent of cardiac jelly deposition in Eng−/− vs. control hearts by staining acidic glycosaminoglycans with Alcian Blue. As shown in Figs. 6K, L, the cardiac jelly appeared as extensive in mutant hearts as in controls, suggesting that insufficient extracellular matrix deposition in the cushion forming region does not explain the inability of Eng-null endocardial cells to undergo EMT. Type 3 chimeras and Eng−/− embryos additionally displayed less trabeculation of the ventricular myocardium (Figs. 4E, F and 6I, J), a phenotype that has been previously demonstrated to reflect growth restriction and tissue hypoxia (Krishnan et al., 2008; Ream et al., 2008). Furthermore, VEGF, a target of the hypoxia responsive pathway mediated by HIF1α was significantly increased in Eng−/− hearts relative to control (Fig. 5H). To address the hypothesis that Eng-null embryos are undergoing tissue hypoxia, we measured the expression of several targets of the hypoxia response by quantitative real time gene expression. Glut-1 and Pfkfb3, genes expressed during anaerobic cellular respiration (Ream et al., 2008), were significantly upregulated in the bodies (Fig. 7)

Fig. 7. Quantitative real time PCR of Eng-null vs. control E9.5 embryos for hypoxia response transcripts. Relative gene expression normalized to β-actin for Bnip3, Glut-1, Pfkfb3, and VEGF. All genes are targets of the hypoxia response and are significantly upregulated in Eng−/− embryos relative to control embryos (P b 0.005).

and hearts (data not shown) of Eng−/− embryos at E9.5. VEGF was also significantly upregulated in Eng−/− bodies, consistent with gene expression analysis in isolated heart tissue. Bnip3, a gene previously shown to be upregulated prior to cellular apoptosis in response to hypoxia (Ream et al., 2008) was also significantly upregulated in Eng−/− embryo bodies (Fig. 7) and hearts (data not shown). However, Tunel staining of cardiac tissue indicated no significant increase in apoptotic cells in Eng−/− heart, suggesting that E9.5 represents a window in which the Eng-null embryo response to cardiac insufficiency is occurring, but viability is not yet compromised. Discussion Previous studies demonstrated that loss of Endoglin affects angiogenesis in both the yolk sac and embryo, as well as heart morphogenesis during early development. The purpose of our study was to determine whether Endoglin is required for both early angiogenesis and cardiac development or if defects in either of these processes are a secondary consequence of the other. Therefore, we first characterized vascular development in Eng-null embryoid bodies and in chimeric embryos. The cultured EBs, when grown in a system that bypasses the requirement for intact circulation, are able to form vascular structures independent of the presence of Endoglin (Fig. 3), consistent with previous findings (Perlingeiro, 2007). Similarly, even under the stringent competitive situation with wild-type host cells that exists in Type 1 chimeras, we observed the ability of Eng-null donor ES cells to differentiate into ECs that contribute to all tissues and organ primordia (Figs. 1B, F, J), indicating that Endoglin is not required for EC sprouting angiogenesis in the midgestation embryo. These results strongly support our in vitro EB culture results and suggest that Eng-deficient ES cells and endothelial cells derived from them are normal with respect to their angiogenic and remodelling potential in the early embryo. However, our findings do not formally exclude the possibility that Endoglin is required for vascular remodelling by non-cell autonomously affecting the behaviour of perivascular support cells. In addition, our experiments do not address the consequences of widespread and complete loss of Endoglin function in mature vascular endothelium during later gestation and in the adult. Several studies have demonstrated that angiogenesis and vascular patterning are dependent on normal cardiac function and the timely establishment of hemodynamic forces in both mammalian and avian embryos (Lucitti et al., 2007; May et al., 2004; Wakimoto et al., 2000). Previous studies suggested that Eng mutant embryos display cardiac looping defects and abrogation of endocardial cushion formation (Arthur et al., 2000; Bourdeau et al., 1999). We thus hypothesized that angiogenesis defects in Eng-null embryos could be explained by insufficient cardiac output, as a consequence of perturbed heart morphogenesis. Therefore, in our second set of experiments, we carefully characterized early heart development, EMT and hypoxia in Eng-null vs. control embryos, as well as the behaviour of Eng-deficient cells in the hearts of chimeric embryos rescued of circulatory failure. We showed that AVC and cardiac chamber restricted gene expressions are perturbed in Eng-null embryos (Fig. 6) and that genes responsive to hypoxia are activated significantly in Eng-null embryos (Fig. 7). We also showed that while Eng-deficient cells contributed readily to blood vessels, they did not contribute to the mesenchymal population in the AV cushions in developmentally rescued Type2 chimeric embryos (Figs. 1C, G, K, 2B, D, and 4H, J). This finding is consistent with a delay in invasion and migration of AVC derived cells in ex vivo cultures of Eng-null explants (Figs. 5A–G). Taken together, our data indicate that the absence of Endoglin leads to a disruption in the patterning of the AVC, causing inappropriate boundary formation between the cardiac chambers leading to a delay in the onset of cardiac function, or its insufficiency. This in turn


results in aborted angiogenesis and vascular remodeling, as well as decreased oxygen perfusion, hypoxia and growth delay. When cardiac insufficiency is bypassed, as in Type 2 chimeras where growth delay, cardiac looping defects, endocardial cushion formation and angiogenesis defects are rescued (Figs. 1C, G, K, 2B, D, and 4H, J), Eng−/− endocardial cells were at a competitive disadvantage in the process of cushion formation, since these tissues were found to be composed almost exclusively of wild-type donor tissue (Figs. 4H, J). Therefore, our data suggest that in the early embryo, Endoglin is first required for patterning the AVC and is subsequently required in endocardial cells for EMT. Furthermore, in the Eng-null heart, high VEGF production, likely as a consequence of hypoxia in the cardiac chambers (Fig. 5H), may also contribute to the suppression of EMT (Dor et al., 2001), compounding the effects of incorrect cellular patterning in the AVC. Endoglin's role as a component of the TGFβ1 signaling complex has been extensively characterized in endothelial cells in vitro (Lastres et al., 1996; Lebrin et al., 2004; Pece-Barbara et al., 2005), and a role for TGFβ1/β3 ligands during EMT in the developing heart was suggested from avian studies using the collagen explant system (Potts and Runyan, 1989; Proetzel et al., 1995). However, in vivo mouse genetic analyses do not support a central role for TGFβ1/β3 signalling during EMT, although this pathway is likely required for later steps of valve maturation (Bartram et al., 2001; Dickson et al., 1995; Dunker and Krieglstein, 2002; Kaartinen et al., 1995; Proetzel et al., 1995; Sanford et al., 1997; Sridurongrit et al., 2008). In particular, endothelial and myocardial specific knockouts of TβRII (Jiao et al., 2006; Park et al., 2008), the only Type II receptor available for TGFβ1/β3 signaling, strongly indicate that these growth factors are largely dispensable for early cardiac morphogenesis and in particular for formation of the AV cushion mesenchyme. This argues that some compensatory stimulus induces EMT in the absence of TGFβ signaling and that Endoglin likely interacts with this pathway to modulate EMT. In this regard, Endoglin can interact with multiple TGFβ superfamily members in addition to TGFβ1 and TGFβ3, including ActivinA, BMP2, BMP7 and BMP9 in the presence of the appropriate ligand binding receptor (Barbara et al., 1999; David et al., 2007). Both BMP2 and BMP7 have been implicated in AV cushion development. Endoglin may mediate the effects of BMP2 specifically in the AVC endocardium due to BMP2 restricted expression in the AVC myocardium (Fig. 6A) (Lyons et al., 1995). Moreover, BMP2 can potently induce EMT in mouse cardiac explants (Sugi et al., 2004). Conditional mutagenesis of BMP2 and its downstream signalling receptors, Bmpr2, and Alk3(Bmpr1a) interfere with early cushion formation (Gaussin et al., 2002; Ma et al., 2005; Park et al., 2006; Rivera-Feliciano and Tabin, 2006; Song et al., 2007). The defects described for BMP2 deletion in cardiac tissue (Ma et al., 2005; Rivera-Feliciano and Tabin, 2006), are consistent with our findings in Eng−/− embryos. In both cases, EMT does not occur, and AVC myocardium expresses genes characteristic of differentiated chamber myocardium, consistent with the view that Endoglin modulates BMP2 signaling in the heart. On the other hand, BMP2-deficient hearts displayed a defect in cardiac jelly deposition (Ma et al., 2005; Rivera-Feliciano and Tabin, 2006), whereas this was not observed in Eng−/− embryos, suggesting that endocardial cushion swelling mediated by BMP2 does not involve Endoglin. We showed that BMP2 expression in the AVC myocardium was reduced in Eng-null embryos and that EMT is delayed in these embryos in vitro. In Type 2 rescued chimeras, where cushion formation is restored (Figs. 4H, J), Eng-null cells cannot respond to EMT induction signals to the same extent as wild-type cells. This indicates that, while reduced BMP2 expression contributes to the block in EMT in Eng−/− embryos, Endoglin is subsequently also required cell autonomously in endocardium to respond to the EMT induction cues, because chimera analysis showed that only wild-type, Endoglin expressing donor cells populate the endocardial cushions in embryos rescued of

75

defects. Thus, it is interesting to consider that Endoglin may modulate several responses required in the AVC endocardium: first for patterning the AVC myocardium leading to the induction of BMP2 and its downstream modulators, and secondly for induction of cellular events leading to EMT. In this regard, BMP7, another ligand for Endoglin is also expressed throughout the mouse heart tube from E8.5 (Lyons et al., 1995), and may play a role in these processes. Further experiments to determine Endoglin's functional role in BMP 2, 7, and 9 signalling, in the endocardium are required to pinpoint the molecular mechanism of its role in heart development. Our study shows that Endoglin expression and function in endocardial cells is required for proper patterning of the myocardium at E9.5. This indicates that signals leading to induction of BMP2 and Tbx2 in the AVC myocardium are derived from the endocardium. The importance of endocardial–myocardial signalling has been documented in EMT and trabeculation of the ventricles (Wagner and Siddiqui, 2007). Our results suggest that endocardial-myocardial signalling is also requisite for the process of patterning the atrioventricular boundary. Live-cell imaging studies in zebrafish demonstrated the requirement for the earliest specified endocardial cells to direct the movement of cardiomyocytes in formation of a functioning heart tube (Holtzman et al., 2007). Since Endoglin is expressed in endocardial cells from their first appearance in the cardiac crescent (Arthur et al., 2000), it will be important to define the mechanisms by which Endoglin patterns the AVC myocardium, and how it directs the transcriptional networks that are essential for cardiac morphogenesis. Recently, endothelial-to-mesenchymal transition, a process analogous to embryonic EMT has been implicated in cardiac fibrosis, leading to cardiac hypertrophy, a condition associated with myocardial dysfunction and morbidity. In this regard, TGFβ1 and BMP7, both Endoglin ligands, were found to be mutually antagonistic in this process (Zeisberg et al., 2007). It will be interesting to determine whether Endoglin is also involved in EMT leading to cardiac fibrosis, and whether it plays a role in this disease. Acknowledgments The authors thank Drs. Janet Rossant and Hiroki Kokubo for in situ probes, Petia Stefanova for histology, and Drs. Aly Karsan and Jorge Filmus for comments on the manuscript. Funding for this work was provided by the Heart and Stroke Foundation of Canada (grants NA-5427 and T5598). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ydbio.2009.08.016. References Abdalla, S.A., Letarte, M., 2006. Hereditary haemorrhagic telangiectasia: current views on genetics and mechanisms of disease. J. Med. Genet. 43, 97–110. Allinson, K.R., Carvalho, R.L., van den Brink, S., Mummery, C.L., Arthur, H.M., 2007. Generation of a floxed allele of the mouse endoglin gene. Genesis 45, 391–395. Armstrong, E.J., Bischoff, J., 2004. Heart valve development: endothelial cell signaling and differentiation. Circ. Res. 95, 459–470. Arthur, H.M., Ure, J., Smith, A.J., Renforth, G., Wilson, D.I., Torsney, E., Charlton, R., Parums, D.V., Jowett, T., Marchuk, D.A., Burn, J., Diamond, A.G., 2000. Endoglin, an ancillary TGFbeta receptor, is required for extraembryonic angiogenesis and plays a key role in heart development. Dev. Biol. 217, 42–53. Barbara, N.P., Wrana, J.L., Letarte, M., 1999. Endoglin is an accessory protein that interacts with the signaling receptor complex of multiple members of the transforming growth factor-beta superfamily. J. Biol. Chem. 274, 584–594. Bartram, U., Molin, D.G., Wisse, L.J., Mohamad, A., Sanford, L.P., Doetschman, T., Speer, C.P., Poelmann, R.E., Gittenberger-de Groot, A.C., 2001. Double-outlet right ventricle and overriding tricuspid valve reflect disturbances of looping, myocardialization, endocardial cushion differentiation, and apoptosis in TGF-beta(2)knockout mice. Circulation 103, 2745–2752. Bourdeau, A., Dumont, D.J., Letarte, M., 1999. A murine model of hereditary hemorrhagic telangiectasia. J. Clin. Invest. 104, 1343–1351.

76


Bourdeau, A., Faughnan, M.E., Letarte, M., 2000. Endoglin-deficient mice, a unique model to study hereditary hemorrhagic telangiectasia. Trends Cardiovasc. Med. 10, 279–285. Bourdeau, A., Faughnan, M.E., McDonald, M.L., Paterson, A.D., Wanless, I.R., Letarte, M., 2001. Potential role of modifier genes influencing transforming growth factorbeta1 levels in the development of vascular defects in endoglin heterozygous mice with hereditary hemorrhagic telangiectasia. Am. J. Pathol. 158, 2011–2020. Camenisch, T.D., Molin, D.G., Person, A., Runyan, R.B., Gittenberger-de Groot, A.C., McDonald, J.A., Klewer, S.E., 2002. Temporal and distinct TGFbeta ligand requirements during mouse and avian endocardial cushion morphogenesis. Dev. Biol. 248, 170–181. Carvalho, R.L., Jonker, L., Goumans, M.J., Larsson, J., Bouwman, P., Karlsson, S., Dijke, P.T., Arthur, H.M., Mummery, C.L., 2004. Defective paracrine signalling by TGFbeta in yolk sac vasculature of endoglin mutant mice: a paradigm for hereditary haemorrhagic telangiectasia. Development 131, 6237–6247. Carvalho, R.L., Itoh, F., Goumans, M.J., Lebrin, F., Kato, M., Takahashi, S., Ema, M., Itoh, S., van Rooijen, M., Bertolino, P., Ten Dijke, P., Mummery, C.L., 2007. Compensatory signalling induced in the yolk sac vasculature by deletion of TGFbeta receptors in mice. J. Cell Sci. 120, 4269–4277. Chang, C.P., Neilson, J.R., Bayle, J.H., Gestwicki, J.E., Kuo, A., Stankunas, K., Graef, I.A., Crabtree, G.R., 2004. A field of myocardial-endocardial NFAT signaling underlies heart valve morphogenesis. Cell 118, 649–663. Christoffels, V.M., Habets, P.E., Franco, D., Campione, M., de Jong, F., Lamers, W.H., Bao, Z.Z., Palmer, S., Biben, C., Harvey, R.P., Moorman, A.F., 2000. Chamber formation and morphogenesis in the developing mammalian heart. Dev. Biol. 223, 266–278. David, L., Mallet, C., Mazerbourg, S., Feige, J.J., Bailly, S., 2007. Identification of BMP9 and BMP10 as functional activators of the orphan activin receptor-like kinase 1 (ALK1) in endothelial cells. Blood 109, 1953–1961. Dickson, M.C., Martin, J.S., Cousins, F.M., Kulkarni, A.B., Karlsson, S., Akhurst, R.J., 1995. Defective haematopoiesis and vasculogenesis in transforming growth factor-beta 1 knock out mice. Development 121, 1845–1854. Dor, Y., Camenisch, T.D., Itin, A., Fishman, G.I., McDonald, J.A., Carmeliet, P., Keshet, E., 2001. A novel role for VEGF in endocardial cushion formation and its potential contribution to congenital heart defects. Development 128, 1531–1538. Dunker, N., Krieglstein, K., 2002. Tgfbeta2−/− Tgfbeta3−/− double knockout mice display severe midline fusion defects and early embryonic lethality. Anat. Embryol. (Berl) 206, 73–83. Ema, M., Yokomizo, T., Wakamatsu, A., Terunuma, T., Yamamoto, M., Takahashi, S., 2006. Primitive erythropoiesis from mesodermal precursors expressing VEcadherin, PECAM-1, Tie2, endoglin, and CD34 in the mouse embryo. Blood 108, 4018–4024. Gaussin, V., Van de Putte, T., Mishina, Y., Hanks, M.C., Zwijsen, A., Huylebroeck, D., Behringer, R.R., Schneider, M.D., 2002. Endocardial cushion and myocardial defects after cardiac myocyte-specific conditional deletion of the bone morphogenetic protein receptor ALK3. Proc. Natl. Acad. Sci. U. S. A. 99, 2878–2883. Goumans, M.J., Zwijsen, A., van Rooijen, M.A., Huylebroeck, D., Roelen, B.A., Mummery, C.L., 1999. Transforming growth factor-beta signalling in extraembryonic mesoderm is required for yolk sac vasculogenesis in mice. Development 126, 3473–3483. Harrelson, Z., Kelly, R.G., Goldin, S.N., Gibson-Brown, J.J., Bollag, R.J., Silver, L.M., Papaioannou, V.E., 2004. Tbx2 is essential for patterning the atrioventricular canal and for morphogenesis of the outflow tract during heart development. Development 131, 5041–5052. Hirashima, M., Bernstein, A., Stanford, W.L., Rossant, J., 2004. Gene-trap expression screening to identify endothelial-specific genes. Blood 104, 711–718. Holtzman, N.G., Schoenebeck, J.J., Tsai, H.J., Yelon, D., 2007. Endocardium is necessary for cardiomyocyte movement during heart tube assembly. Development 134, 2379–2386. Huber, T.L., Kouskoff, V., Fehling, H.J., Palis, J., Keller, G., 2004. Haemangioblast commitment is initiated in the primitive streak of the mouse embryo. Nature 432, 625–630. Jerkic, M., Rivas-Elena, J.V., Prieto, M., Carron, R., Sanz-Rodriguez, F., Perez-Barriocanal, F., Rodriguez-Barbero, A., Bernabeu, C., Lopez-Novoa, J.M., 2004. Endoglin regulates nitric oxide-dependent vasodilatation. FASEB J. 18, 609–611. Jerkic, M., Rodriguez-Barbero, A., Prieto, M., Toporsian, M., Pericacho, M., Rivas-Elena, J.V., Obreo, J., Wang, A., Perez-Barriocanal, F., Arevalo, M., Bernabeu, C., Letarte, M., Lopez-Novoa, J.M., 2006. Reduced angiogenic responses in adult Endoglin heterozygous mice. Cardiovasc. Res. 69, 845–854. Jiao, K., Langworthy, M., Batts, L., Brown, C.B., Moses, H.L., Baldwin, H.S., 2006. Tgfbeta signaling is required for atrioventricular cushion mesenchyme remodeling during in vivo cardiac development. Development 133, 4585–4593. Jonker, L., Arthur, H.M., 2002. Endoglin expression in early development is associated with vasculogenesis and angiogenesis. Mech. Dev. 110, 193–196. Kaartinen, V., Voncken, J.W., Shuler, C., Warburton, D., Bu, D., Heisterkamp, N., Groffen, J., 1995. Abnormal lung development and cleft palate in mice lacking TGF-beta 3 indicates defects of epithelial-mesenchymal interaction. Nat. Genet. 11, 415–421. Kattman, S.J., Adler, E.D., Keller, G.M., 2007. Specification of multipotential cardiovascular progenitor cells during embryonic stem cell differentiation and embryonic development. Trends Cardiovasc. Med. 17, 240–246. Kokubo, H., Tomita-Miyagawa, S., Hamada, Y., Saga, Y., 2007. Hesr1 and Hesr2 regulate atrioventricular boundary formation in the developing heart through the repression of Tbx2. Development 134, 747–755. Krishnan, J., Ahuja, P., Bodenmann, S., Knapik, D., Perriard, E., Krek, W., Perriard, J.C., 2008. Essential role of developmentally activated hypoxia-inducible factor 1α for cardiac morphogenesis and function. Circ. Res. 103, 1139–1146.

Lastres, P., Letamendia, A., Zhang, H., Rius, C., Almendro, N., Raab, U., Lopez, L.A., Langa, C., Fabra, A., Letarte, M., Bernabeu, C., 1996. Endoglin modulates cellular responses to TGF-beta 1. J. Cell Biol. 133, 1109–1121. le Noble, F., Moyon, D., Pardanaud, L., Yuan, L., Djonov, V., Matthijsen, R., Breant, C., Fleury, V., Eichmann, A., 2004. Flow regulates arterial-venous differentiation in the chick embryo yolk sac. Development 131, 361–375. Lebrin, F., Goumans, M.J., Jonker, L., Carvalho, R.L., Valdimarsdottir, G., Thorikay, M., Mummery, C., Arthur, H.M., ten Dijke, P., 2004. Endoglin promotes endothelial cell proliferation and TGF-beta/ALK1 signal transduction. EMBO J. 23, 4018–4028. Li, D.Y., Sorensen, L.K., Brooke, B.S., Urness, L.D., Davis, E.C., Taylor, D.G., Boak, B.B., Wendel, D.P., 1999. Defective angiogenesis in mice lacking endoglin. Science 284, 1534–1537. Lucitti, J.L., Jones, E.A., Huang, C., Chen, J., Fraser, S.E., Dickinson, M.E., 2007. Vascular remodeling of the mouse yolk sac requires hemodynamic force. Development 134, 3317–3326. Lyons, K.M., Hogan, B.L.M., Robertson, E.J., 1995. Colocalization of BMP 7 and BMP 2 RNAs suggests that these factors cooperatively mediate tissue interactions during murine development. Mech. Dev. 50, 71–83. Ma, L., Lu, M.F., Schwartz, R.J., Martin, J.F., 2005. Bmp2 is essential for cardiac cushion epithelial-mesenchymal transition and myocardial patterning. Development 132, 5601–5611. May, S.R., Stewart, N.J., Chang, W., Peterson, A.S., 2004. A Titin mutation defines roles for circulation in endothelial morphogenesis. Dev. Biol. 270, 31–46. Mercado-Pimentel, M.E., Hubbard, A.D., Runyan, R.B., 2007. Endoglin and Alk5 regulate epithelial-mesenchymal transformation during cardiac valve formation. Dev. Biol. 304, 420–432. Nagy, A., Gerstenstein, M., Vintersten, K., Behringer, R., 2003. Manipulating the mouse embryo: A laboratory manual. Niessen, K., Fu, Y., Chang, L., Hoodless, P.A., McFadden, D., Karsan, A., 2008. Slug is a direct Notch target required for initiation of cardiac cushion cellularization. J. Cell Biol. 182, 315–325. Park, C., Lavine, K., Mishina, Y., Deng, C.X., Ornitz, D.M., Choi, K., 2006. Bone morphogenetic protein receptor 1A signaling is dispensable for hematopoietic development but essential for vessel and atrioventricular endocardial cushion formation. Development 133, 3473–3484. Park, S.O., Lee, Y.J., Seki, T., Hong, K.H., Fliess, N., Jiang, Z., Park, A., Wu, X., Kaartinen, V., Roman, B.L., Oh, S.P., 2008. ALK5- and TGFBR2-independent role of ALK1 in the pathogenesis of hereditary hemorrhagic telangiectasia type 2. Blood 111, 633–642. Pece-Barbara, N., Vera, S., Kathirkamathamby, K., Liebner, S., Di Guglielmo, G.M., Dejana, E., Wrana, J.L., Letarte, M., 2005. Endoglin null endothelial cells proliferate faster and are more responsive to transforming growth factor beta1 with higher affinity receptors and an activated Alk1 pathway. J. Biol. Chem. 280, 27800–27808. Perlingeiro, R.C., 2007. Endoglin is required for hemangioblast and early hematopoietic development. Development 134, 3041–3048. Person, A.D., Klewer, S.E., Runyan, R.B., 2005. Cell biology of cardiac cushion development. Int. Rev. Cytol. 243, 287–335. Potts, J.D., Runyan, R.B., 1989. Epithelial-mesenchymal cell transformation in the embryonic heart can be mediated, in part, by transforming growth factor beta. Dev. Biol. 134, 392–401. Proetzel, G., Pawlowski, S.A., Wiles, M.V., Yin, M., Boivin, G.P., Howles, P.N., Ding, J., Ferguson, M.W., Doetschman, T., 1995. Transforming growth factor-beta 3 is required for secondary palate fusion. Nat. Genet. 11, 409–414. Puri, M.C., Bernstein, A., 2003. Requirement for the TIE family of receptor tyrosine kinases in adult but not fetal hematopoiesis. Proc. Natl. Acad. Sci. U. S. A. 100, 12753–12758. Qu, R., Silver, M.M., Letarte, M., 1998. Distribution of endoglin in early human development reveals high levels on endocardial cushion tissue mesenchyme during valve formation. Cell Tissue Res. 292, 333–343. Ream, M., Ray, A.M., Chandra, R., Chikaraishi, D.M., 2008. Early fetal hypoxia leads to growth restriction and myocardial thinning. Am. J. Physiol. Regul. Integr. Comp. Physiol. 295, R583–R595. Rivera-Feliciano, J., Tabin, C.J., 2006. Bmp2 instructs cardiac progenitors to form the heart-valve-inducing field. Dev. Biol. 295, 580–588. Runyan, R.B., Markwald, R.R., 1983. Invasion of mesenchyme into three-dimensional collagen gels: a regional and temporal analysis of interaction in embryonic heart tissue. Dev. Biol. 95, 108–114. Saga, Y., Hata, N., Kobayashi, S., Magnuson, T., Seldin, M.F., Taketo, M.M., 1996. MesP1: a novel basic helix-loop-helix protein expressed in the nascent mesodermal cells during mouse gastrulation. Development 122, 2769–2778. Sanford, L.P., Ormsby, I., Gittenberger-de Groot, A.C., Sariola, H., Friedman, R., Boivin, G.P., Cardell, E.L., Doetschman, T., 1997. TGFbeta2 knockout mice have multiple developmental defects that are non-overlapping with other TGFbeta knockout phenotypes. Development 124, 2659–2670. Song, L., Fassler, R., Mishina, Y., Jiao, K., Baldwin, H.S., 2007. Essential functions of Alk3 during AV cushion morphogenesis in mouse embryonic hearts. Dev. Biol. 301, 276–286. Sorensen, L.K., Brooke, B.S., Li, D.Y., Urness, L.D., 2003. Loss of distinct arterial and venous boundaries in mice lacking endoglin, a vascular-specific TGFbeta coreceptor. Dev. Biol. 261, 235–250. Sridurongrit, S., Larsson, J., Schwartz, R., Ruiz-Lozano, P., Kaartinen, V., 2008. Signaling via the Tgf-beta type I receptor Alk5 in heart development. Dev. Biol. 322, 208–218. Sugi, Y., Yamamura, H., Okagawa, H., Markwald, R.R., 2004. Bone morphogenetic protein-2 can mediate myocardial regulation of atrioventricular cushion mesenchymal cell formation in mice. Dev. Biol. 269, 505–518.

A. Nomura-Kitabayashi et al. / Developmental Biology 335 (2009) 66–77 Taddei, A., Giampietro, C., Conti, A., Orsenigo, F., Breviario, F., Pirazzoli, V., Potente, M., Daly, C., Dimmeler, S., Dejana, E., 2008. Endothelial adherens junctions control tight junctions by VE-cadherin-mediated upregulation of claudin-5. Nat. Cell Biol. 10, 923–934. Toporsian, M., Gros, R., Kabir, M.G., Vera, S., Govindaraju, K., Eidelman, D.H., Husain, M., Letarte, M., 2005. A role for endoglin in coupling eNOS activity and regulating vascular tone revealed in hereditary hemorrhagic telangiectasia. Circ. Res. 96, 684–692. Wagner, M., Siddiqui, M.A., 2007. Signal transduction in early heart development (II): ventricular chamber specification, trabeculation, and heart valve formation. Exp. Biol. Med. 232, 866–880.

77

Wakimoto, K., Kobayashi, K., Kuro-o, M., Yao, A., Iwamoto, T., Yanaka, N., Kita, S., Nishida, A., Azuma, S., Toyoda, Y., Omori, K., Imahie, H., Oka, T., Kudoh, S., Kohmoto, O., Yazaki, Y., Shigekawa, M., Imai, Y., Nabeshima, Y.i., Komuro, I., 2000. Targeted disruption of Na+/Ca2+ exchanger gene leads to cardiomyocyte apoptosis and defects in heartbeat. J. Biol. Chem. 275, 36991–36998. Wood, S.A., Allen, N.D., Rossant, J., Auerbach, A., Nagy, A., 1993. Non-injection methods for the production of embryonic stem cell-embryo chimaeras. Nature 365, 87–89. Zeisberg, E.M., Tarnavski, O., Zeisberg, M., Dorfman, A.L., McMullen, J.R., Gustafsson, E., Chandraker, A., Yuan, X., Pu, W.T., Roberts, A.B., Neilson, E.G., Sayegh, M.H., Izumo, S., Kalluri, R., 2007. Endothelial-to-mesenchymal transition contributes to cardiac fibrosis. Nat. Med. 13, 952–961.