Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/6379906

FunctionalBMPreceptorinendocardialcellsisrequiredinatrioventricularcushionmesenchymalcellformationinchick

ARTICLEinDEVELOPMENTALBIOLOGY·JULY2007

ImpactFactor:3.55·DOI:10.1016/j.ydbio.2007.03.015·Source:PubMed

CITATIONS

23

READS

30

3AUTHORS,INCLUDING:

RogerRMarkwald

MedicalUniversityofSouthCarolina

260PUBLICATIONS10,528CITATIONS

SEEPROFILE

YukikoSugi

MedicalUniversityofSouthCarolina

35PUBLICATIONS1,794CITATIONS

SEEPROFILE

Availablefrom:YukikoSugi

Retrievedon:05February2016

06 (2007) 179–192www.elsevier.com/locate/ydbio

Developmental Biology 3

Functional BMP receptor in endocardial cells is required in atrioventricularcushion mesenchymal cell formation in chick

Hiroto Okagawa 1, Roger R. Markwald, Yukiko Sugi ⁎

Department of Cell Biology and Anatomy and Cardiovascular Developmental Biology Center, Medical University of South Carolina,171 Ashley Avenue, Charleston, SC 29425, USA

Received for publication 9 January 2007; revised 9 March 2007; accepted 9 March 2007Available online 16 March 2007

Abstract

Transformation of atrioventricular (AV) canal endocardium into invasive mesenchyme correlates spatially and temporally with the expressionof bone morphogenetic protein (BMP)-2 in the AV myocardium. We revealed the presence of mRNA of Type I BMP receptors, BMPR-1A(ALK3), BMPR-1B (ALK6) and ALK2 in chick AV endocardium at stage-14−, the onset of epithelial to mesenchymal transformation (EMT), byRT–PCR and localized BMPR-1B mRNA in the endocardium by in situ hybridization. To circumvent the functional redundancies among the TypeI BMP receptors, we applied dominant-negative (dn) BMPR-1B-viruses to chick AV explants and whole-chick embryo cultures to specificallyblock BMP signaling in AV endocardium during EMT. dnBMPR-1B-virus infection of AV endocardial cells abolished BMP-2-supported AVendocardial EMT. Conversely, caBMPR-1B-virus infection promoted AV endocardial EMT in the absence of AV myocardium. Moreover,dnBMPR-1B-virus treatments significantly reduced myocardially supported EMT in AV endocardial–myocardial co-culture. AV cushionmesenchymal cell markers, α-smooth muscle actin (SMA), and TGFβ3 in the endocardial cells were promoted by caBMPR-1B and reduced bydnBMPR-1B infection. Microinjection of the virus into the cardiac jelly in the AV canal at stage-13 in vivo (ovo) revealed that the dnBMPR-1B-virus-infected cells remained in the endocardial epithelium, whereas caBMPR-1B-infected cells invaded deep into the cushions. These resultsprovide evidence that BMP signaling through the AV endocardium is required for the EMT and the activation of the BMP receptor in theendocardium can promote AV EMT in the chick.© 2007 Elsevier Inc. All rights reserved.

Keywords: Bone morphogenetic protein receptor; Heart; Atrioventricular cushion; Retroviral gene transfer; Epithelial–mesenchymal transformation

Introduction

Valvuloseptal morphogenesis of the primitive heart tube intoa four-chambered organ requires the formation of theendocardial cushion tissue in two segments of the primaryheart tube, the atrioventricular (AV) canal and the conus/truncusoutlet (OT) (reviews: Eisenberg and Markwald, 1995; Mark-wald et al., 1996; Nakajima et al., 2000; Person et al., 2005). Inthe AV canal, transformation of endocardial endothelial cellsinto invasive mesenchymal cells plays a critical role in cushiontissue formation and this transformation correlates spatially and

⁎ Corresponding author. Fax: +1 843 792 0664.E-mail address: sugiy@musc.edu (Y. Sugi).

1 Present address: Department of Pediatrics, Shakaihoken Shiga Hospital,Fujimidai 16-1, Otsu-city, Shiga 520-0846, Japan.

0012-1606/$ - see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.ydbio.2007.03.015

temporally with the expression of bone morphogenetic protein-2 (BMP-2) in the AV myocardium (Lyons et al., 1990; Jones etal., 1991; Yamada et al., 1999; Yamagishi et al., 1999; Somi etal., 2004; Sugi et al., 2004). BMPs, a subgroup of thetransforming growth factor (TGF)-β super-family, are wellimplicated in embryonic development (Francis-West et al.,1994; Wall and Hogan, 1995; Hogan, 1996). However, earlylethality of conventional BMP-2- (Zhang and Bradley, 1996),BMPR-1A- (Mishina et al., 1995) or BMPR-II- (Beppu et al.,2000) deficient mice significantly hampered addressing the roleof BMPs in AV EMT. Using in vitro culture assays, we wereable to show that BMP-2 was sufficient and BMP-signaling wasrequired for AV EMT in mice (Sugi et al., 2004). More recently,conditional deletion of BMP-2 using the NKx2.5-cre systemconfirmed that myocardially derived BMP-2 was essential forAV endocardial transformation into cushion mesenchyme in

180 H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

mice (Ma et al., 2005; Rivera-Feliciano and Tabin, 2006). BMP-2 was also implicated in AV canal morphogenesis in chick(Yamagishi et al., 1999; Romano and Runyan, 2000). However,expression of BMP receptors in AV endocardium and thefunctional roles of these receptors in AV EMT remained to beelucidated.

BMPs exert their biological function by interacting withcell surface receptors. BMP receptors consist of Type-1 andType-2 receptors and both types of receptors containintracellular serine/threonine kinase domains, which arerequired for signal transduction. Type-2 receptors transpho-sphorylate the glycine–serine-rich domain (GS domain) ofType-1 receptors, which induce activation of the Type-1receptor and transduce signals (Hogan, 1996; Yamashita et al.,1996; de Caestecker, 2004). BMP type 1 receptors areclassified into subtypes, activin-receptor-like kinase-1,(ALK)-1/TTSR-I, activin Type I receptor, (ACTRI)/ALK2,ALK3/BMPR-1A, and ALK6/BMPR-1B (Hogan, 1996; Yama-shita et al., 1996; Yi et al., 2000). Most of the BMP receptorshave ligand binding affinity with other TGFβ super familyproteins, i.e. TGFβs and activin (Yamashita et al., 1996; deCaestecker, 2004). However, BMPR1-A (ALK3) and BMPR1-B (ALK6) bind specifically with BMP-2 and BMP-4, and bindwith BMP-7 at low affinity but do not bind with TGFβs oractivin (Dijke et al., 1994; Raftery and Sutherland, 1999; deCaestecker, 2004).

Fig. 1. Expression of type 1 BMP receptors by chick embryo heart. (A) Messenger Rwere reverse-transcribed and cDNA samples were subjected to PCR to amplify cDNA(lanes 3, 5, and 7) transcripts were detected at stage-14− (lanes 2 and 3) and -18 (lanesfrom a reaction performed on RNA from the stage-18 chick embryo heart. The PCRand 1-kbp ladder size markers, respectively. (B–D) Localization of BMPR-1BmRNAtreatment as a negative control. (C) Antisense probe treatment revealed BMPR-1B earches. (D) Endocardial expression of BMPR1-B was detected in a higher magnificafrom stage-14− endocardium. Arrows in panels B and C, hearts. Arrows in panel D

In this work, we detected transcripts of BMPR-1A (ALK 3),BMPR-1B (ALK 6), and ALK2 in AV endocardial cells atstage-14−, at the onset of AV EMT by RT–PCR and localizedBMPR-1B in endocardium in stage-14−/15 chick embryos.Because multiple BMP receptors are expressed in theendocardium, we sought to address the role of BMP signalingin the AV endocardium using a dominant-negative approach tocircumvent functional redundancies. We over expressed dn- orca-BMPR-1B using RCAS (avian leukemia virus long terminalrepeat with splice acceptor)-based viral constructs to specifi-cally perturb or actively transduce the BMP signaling in chickAV explants and whole-embryo cultures. A significance of thevirally supported gene transfer technique is that we can alter thesignaling in a spatio-temporally specific manner coincident withAV canal morphogenesis without affecting the growth of otherparts of the embryos. Moreover, RCAS virus-infected cells,both in culture and in vivo (ovo), could be visually detectedusing antibodies to RCAS viral protein to make directcomparison of infected- and non-infected cells in the samebiological condition.

Our findings provide evidence that BMP signaling throughAV endocardial endothelial cells is required for AV EMT andthat the activation of the BMP receptor in endocardial cellspromotes AV cushion mesenchymal cell formation in the chick.This work was partially reported in an abstract form (Okagawaet al., 2000).

NAwas extracted from stage-14− and -18 chick embryo hearts. Extracted RNAsof BMPR-1A and BMPR-1B. Both BMPR-1A (lanes 2, 4, and 6) and BMPR-1B4 and 5). For the PCR controls (lanes 6 and 7), reverse transcriptase was omittedproducts were separated on a 1.1% agarose gel. Lanes 1 and 8 contained 100-bpby whole-mount in situ hybridization of stage-14 chick embryo. (B) Sense probexpression in the heart. BMPR-1B was also detected in somites and pharyngealtion picture. (E) Transcripts of BMP-R1A, BMPR-1B and ALK-2 were detected, endocardium.

181H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

Materials and methods

Chick embryos

Viral-free fertilized eggs of White Leghorn (Gallus gallus domesticus)chicken were purchased from SPAFAS (Preston, CN) and incubated in a humidatmosphere at 37.5 °C. Stages of embryonic development were determined byusing the criteria of Hamburger and Hamilton (1951).

Detection of Type I BMP receptor expression from chick heart byRT–PCR

RT–PCR analysis of BMPR-1A, BMPR-1B, and ALK2 was performedbased on the methods described previously (Sugi and Markwald, 2003).

Fig. 2. AVendocardial monolayer cultured with M199 alone (A and B), BMP-2 (C and2+noggin (I and J) on hydrated collagen gels. (A, C, E, G and I) surface view. (B, Dmonolayer was explanted from stage-14− chick embryos and cultured for 48 h. Whenformation was not detected (A and B). In contrast, BMP-2-treated explants formed i(I and J), invasive mesenchymal cell formation was largely inhibited by adding dnBcaBMPR-1B exhibited larger number of invasive mesenchymal cells within the coll

Briefly, total RNA was isolated and purified from chick hearts fromHamburger and Hamilton (1951) stage-14− and -18 chick embryos and stage-14− endocardium using the RNAzol method (Chomczynski and Sacchi, 1987;TEL-TEST, Inc.). AV endocardium was prepared by culturing stage-14− AVcanal, facing endocardial-side down to the fibronectin (20 μg/ml, Sigma)-coated culture wells and by removing myocardium in 4 h. ComplementaryDNA was prepared by using oligo-dT (Promega) primed M-MLV reversetranscriptase (Promega). The cDNA was amplified using Thermus aquaticus(Taq) DNA polymerase (Promega). Primers (Invitrogen) for whole heartswere designed for BMPR-1A (forward, 5′-CAAGCACTATTGTAAGT-CAATG-3′, reverse, 5′-TCAAATCTTTACATCTTGTGAT-3′), BMPR-1B(forward, 5′-TGCTACTTCAGGTATAAGCGGC-3′, reverse 5′-TTAGAGCT-TAATGTCCTGCGAC-3′). Primers for stage-14− AV endocardium weredesigned for BMPR-1A (forward, 5′-CTGAGAGTTGAGCGATTG-3′,reverse, 5′-CAGCCAGAAGCAAGTGTTGG-3′), BMPR-1B (forward, 5′-

D), BMP-2+dnBMPR-1B (E and F), BMP-2+caBMPR-1B (G and H) or BMP-, F, H and J) deep view (−80 μm from the surface of the gel). AV endocardialendocardial monolayer was cultured in M199 alone, invasive mesenchymal cellnvasive mesenchymal cells (arrows in panel D). Similar to the effect of nogginMPR-1B to BMP-2 (E and F). Conversely, AV monolayer cultured in BMP-2+agen gel (arrows in panel H). Scale bar=100 μm.

Fig. 3. Quantitative evaluation of invasive mesenchymal cell formation from AVendocardial monolayer treated with M199 alone, BMP-2, BMP-2+dnBMPR-1B, BMP-2+caBMPR-1B or BMP-2+noggin. BMP-2 supported mesenchymalcell formation was significantly inhibited in noggin- (*p<0.001 by Student'st-test) or dnBMPR-1B- (*p<0.001 by Student's t-test) treated explants andwas enhanced in caBMPR-1B- (**p<0.03 by Student's t-test) treated explants.Number of explants evaluated for each treatment: 24 for M199 alone, 16 forBMP-2, 20 for BMP-2+dnBMPR-1B, 21 for BMP-2+caBMPR-1B and 8 fornoggin. N=number of areas counted in each explant.

182 H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

GGAGAGCAGAAAAGAAGATAGTGAGG-3′, reverse, 5′-TGGTGTGGAA-TAGGAGTGTCC-3′) and ALK2 (forward, 5′-CTGTGTTGGGGTCACTG-3′,reverse, 5′-TGGAAGCAGCCTTTCTGG-3′). Sequence information was from

Fig. 4. AV endocardial monolayer treated with BMP-2+dnBMPR-1B, BMP-2+caBMAMV3c2 (A, D, and G) and a mesenchymal marker, anti-α-SMA (B, E, and H) or antthe explants. Panels C, F, and I depict different areas of the same explants. These pictuas α-SMA-negative TGFβ3-negative endocardial epithelial cells (A–C), whereas, caBmesenchymal cells (D–F). Panels G–I are explants not infected with the virus but cultand anti-TGFβ3. Scale bar=100 μm.

GenBank and accession numbers are L49204 for BMPR-1A, D13432 forBMPR-1B, and U38622 for ALK2, respectively. That the predicted PCRproducts were not amplified from genomic DNA was verified by treatingsamples with RNase-free DNase-1 (Promega) before RT. As a negative control,the RT step was omitted. The PCR products were verified by thermal cyclesequencing using Taq DNA polymerase and fluorescent dye-labeled termina-tions (Medical University of South Carolina, Biotechnology ResourceLaboratory).

In situ hybridization

Digoxigenin-labeled, single-stranded RNA was generated using the DIGRNA labeling kit (Roche) according to manufacturer's instructions. A chickBMPR1-B cDNA cloned into pBluescript II SK (+) (kindly provided by Dr. T.Nohno, Okayama, Japan) was linearized using EcoRI for the sense probe orBamHI for the antisense probe. T3 RNA polymerase or T7 RNA polymerasewas used to construct the sense or antisense probe, respectively.

For whole-mount preparation, chick embryos from stages 8 to 18 werecollected and fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS)and processed for whole mount in situ hybridization. After the treatment ofproteinase K (Roche) and rinsing in PBS–Tween 20, samples were treated withhybridization buffer for prehybridization for 2–4 h then hybridized with adigoxigenin-labeled probe (0.2 μg/ml) in the hybridization buffer at 65 °Covernight. After rinsing, samples were treated for immunohistochemicaldetection of digoxigenin by using alkaline phosphatase-conjugated sheep anti-digoxigenin Fab fragment (Roche) and signals were detected by nitrobluetetrazolium chloride (NBT)/5-bromo-4-chloro-3-indolyl phosphate, toluidinesalt (BCIP) (Roche).

In situ hybridization of tissue sections was performed as previouslydescribed (Somi et al., 2004) with the following modifications. Paraformalde-hyde-fixed stage-10 to -18 chick embryos were embedded in paraffin and 10-μmsections were cut and processed through 20 μg/ml proteinase K (Ambion)/PBS

PR-1B or M199 alone were double immunostained with a RCAS viral marker,i-TGFβ3 (C, F, and I). Panels A and B, D and E, G and H depict the same area ofres indicate that dnBMPR-1B-infected cells tend to stay on the surface of the gelMPR-1B-infected cells tend to become invasive SMA-positive, TGFβ3-positiveured with M199 alone, which shows negative staining for AMV3c2, anti-α-SMA

183H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

for 5–7 min at room temperature. Hybridization was performed using 0.2 μg/mlDig-labeled riboprobe at 70 °C overnight. Color reactions using NBT/BCIPwere allowed to develop for 48–72 h at room temperature.

Generation and propagation of retrovirus

Two mutant constructs of chick BMPR-1B cloned into the replication-competent avian retroviral vector RCAS-(A) (Hughes et al., 1987) were kindlyprovided by Dr. L. Niswander (Howard HughesMedical Institute and Universityof Colorado Health Sciences Center–Fitzsimmons Campus Denver, CO). Thedominant-negative form (dnBMPR-1B) blocks BMP signaling pathways by asingle amino acid substitution (Lys-231 to Arg) within the adenosinetriphosphate binding site in the kinase domain. Because dnBMPR-1B lacks theactivity of an intracellular kinase domain, dnBMPR-1B expressed at the cellsurface can bind BMPs but does not transmit signals. The constitutively activeform (caBMPR-1B) has a single amino acid substitution (Gln-203 to Asp) withinthe GS activation domain. caBMPR-1B is activated constitutively and transmitssignals without BMP binding. dnBMPR-1B and caBMPR-1B viruses werepreviously constructed and blocked and activated BMP signaling, respectively, inavian embryonic tissue cultures (Zou and Niswander, 1996; Zou et al., 1997).

Primary fibroblasts cells were collected from day-10 chick embryos andtransfected with retroviral constructs with FuGENE 6-transfection reagent(Roche). Supernatant from the transfected cultures was collected andconcentrated by ultracentrifugation. Concentrated virus was resuspended inM199 (Invitrogen) and stored at −80 °C. The viral titer was measured withimmunohistochemical detection of viral protein p19 by monoclonal antibody,

Fig. 5. AVendocardial/myocardial co-culture treated with dnBMPR-1B, caBMPR-1Bfrom the surface) within the gel. AV explants from stage-14− embryos were culturedcaBMPR-1B (E and F) or noggin (500 ng/ml) (G and H) for 48 h. Explants treated wiexplants cultured in M199 alone (A and B). Conversely, similar to the effect of nocollagen gel (C and D). m, myocardium. arrows, invasive mesenchymal cells. Scale

AMV3c2 (Developmental Studies Hybridoma Bank) using virus-infectedfibroblast culture. A viral titer (2–5×109–10 active virion/ml) was constantlyachieved after ultracentrifugation of the viral supernatant.

AV explant culture

The culturing procedure was described previously (Bernanke andMarkwald,1982; Runyan and Markwald, 1983; Ramsdell and Markwald, 1997). Briefly,AV explants from stage-14− chick embryo hearts were placed on the hydratedcollagen gels (rat tail tendon collagen, 1.0 mg/ml, Collaborative Research),equilibrated in medium (M 199, Invitrogen), supplemented with 1% chick serum(Sigma) and ITS (5 μg/ml insulin, 5 μg/ml transferrin, and 5 ng/ml selenium,Collaborative Research), and antibiotics (100 units/ml penicillin and 100 μg/mlstreptomycin, Invitrogen). For AVendocardial endothelial culture, myocardiumwas removed 4 h after placement of the explants on the collagen gel.Recombinant human noggin (kindly provided by Regeneron Pharmaceuticals,Inc., NY; final concentration, 500 ng/ml), BMP-2 (kindly provided by GeneticsInstitute, MA; final concentration, 200 ng/ml), and dnBMPR-1B or caBMPR-1B-RCAS virus (final concentration, 1.5×108 virion/ml) were added alone or incombination with BMP-2 to the culture 4 h after the placement of the explantsonto the gel. For some experiments, endocardium was co-cultured withassociated myocardium and was infected with dnBMPR-1B or caBMPR-1B, ortreated with noggin. Cultures were observed daily under an inverted microscope(Olympus, IMT-2). For the index of mesenchymal cell formation, cells on thesurface and at −80-μm deep from the surface of the collagen gel were countedafter 48 h of culture.

, or noggin. (A, C, E, and G) surface view. (B, D, F and H) deep view (−80 μmin medium 199 (M199) alone (A and B), M199 with dnBMPR-1B (C and D),th caBMPR-1B formed more invasive mesenchymal cells (E and F) than controlggin (G and H), when cultured with dnBMPR-1B, fewer cells invaded into thebar=100 μm.

Fig. 6. Quantitative evaluation of invasive mesenchymal cell formation from AVendocardial/myocardial co-culture treated with dnBMPR-1B, caBMPR-1B, ornoggin. Mesenchymal cell formation was significantly inhibited in noggin- ordnBMPR-1B-treated explants and enhanced in caBMPR-1B-treated explants(*p<0.005 by Student's t-test). Number of explants tested for each treatment: 15for M199-treated, 19 for dnBMPR-1B-treated, 22 for caBMPR-1B-treated and11 for Noggin-treated. N=number of areas counted in each explant.

184 H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

Immunostaining of cultured explants

Cultured explants were double-immunostained with a combination ofAMV3c2 and anti-α-SMA (Sigma), anti-TGFβ3 (Santa Cruz) and anti-α-SMA,or AMV3c2 and anti-TGFβ3 depending on the experimental necessity.AMV3c2 recognizes RCAS viral protein, and α-SMA has been used as amarker for AV cushion mesenchymal cell formation in the chick (Nakajima etal., 1997). The specificity of the anti-TGFβ3 was previously tested byimmunopreabsorption using chick AV cushion tissues (Ramsdell andMarkwald,1997). After a 48-h culture, explants were fixed with 4% paraformaldehyde andsubsequently rinsed with 1.5% Tween/PBS (PBST), blocked with 10% normalgoat serum (ICN) and incubated with primary antibody, AMV3c2. Immunecomplexes were detected with FITC-conjugated goat anti-mouse IgG (ICN). Fordouble-immunostaining with anti-TGFβ3 and anti-α-SMA, explants were fixedwith 100% methanol at −20 °C and treated for immunostaining as describedabove. For double-immunostaining with anti-TGFβ3 and AMV3c2, sampleswere processed for AMV3c2 immunostaining as described above and stainedwith anti-TGFβ3. Immunostained samples were observed under a confocalmicroscope, Bio-Rad MRC-1000 Laser Scanning Confocal Imaging Systemequipped with a Zeiss Axioskop.

TGFβ3 immunointensity in cells

Detection of immunointensity was described previously (Sugi et al., 2004).Briefly, original photographs were taken under a laser confocal microscope (Bio-Rad MCR-1000) equipped with a Zeiss Axioskop. For each sample, the samecondenser iris opening (4.0) and 30% laser was applied. TGFβ3 immunointen-sity was evaluated by measuring the intensity of TGFβ3 immunofluorescence ofthe cells in the photograph using the computer software, Adobe Photoshop 5.5(Adobe System, Inc.). AMV3c2 positive endocardial epithelial cells ormesenchymal cells were analyzed to obtain the average intensity of TGFβ3immunofluorescence for each viral treatment. The average fluorescent intensityof endocardial epithelial cells cultured with M199 alone was used as a control forTGFβ3 immunointensity.

Microinjection

Fertilized chicken eggs were incubated in a humid atmosphere at 37.5 °C,windowed, and carefully staged. A volume (1–5 nl) of dnBMPR-1B orcaBMPR-1B-virus solution (viral titer, 2–5×109 active virion/ml) containingPolybrene (100 μg/ml) was pressure injected into cardiac jelly near theendocardial side of AV canal in stage-13 chick embryos in ovo with amicroinjector apparatus, the Picospritzer II pressure system (Parker, NJ), whichwas equipped with three-dimensional manipulators, MMN-1 and MMO-220ND (Narishige, Japan), and one-dimensional micromanipulator, MMO-220(Narishige, Japan). Eggs were resealed and incubated for an additional twodays. Embryos were fixed with 4% paraformaldehyde and processed forAMV3c2-immunostaining followed by FITC-conjugated secondary antibodydetection. Virally infected cells in the AV canal were detected under a confocalmicroscope (Bio-Rad MRC-1000, Laser Scanning Confocal Imaging Systemequipped with a Zeiss Axioskop). Some embryos were embedded in paraffin(Paraplast X-TRA) and processed for sectioning at 6 μm.

Results

Expression of Type I BMP receptors by chick AV endocardium

RT–PCR analysis revealed mRNA-expression of BMPR-1Aand BMPR-1B from stage-14− and -18 chick embryonic hearts(Fig. 1A). BMPR-1A, BMPR-1B, and ALK2 were detected fromcultured stage-14− AV endocardium (Fig. 1E).

BMPR-1B mRNA was localized by whole-mount in situhybridization in stage-14 chick embryos. BMPR-1BmRNAwasexpressed in endocardium (arrows in Figs. 1C, D) and somites(Fig. 1C). BMPR-1B was also expressed in branchial arches and

the neuroepithelial wall of the ventricle in the brain (not shown).Although endocardial expression was low, its expression wasconsistent with the RT–PCR data shown above. Restriction ofBMPR-1B to the AV or OT region was not clearly detected(Figs. 1C, D). Section in situ hybridization also revealedBMPR-1B expression in the endocardium. However, there wasno clear restriction of the BMPR-1B to the cushion formingsegments of in the heart (not shown).

AV endocardial endothelium was cultured on collagen gels toassess the effect of dnBMPR-1B, caBMPR-1B, or noggin undertreatment of BMP-2

AV endocardial epithelial monolayers were explanted fromstage-14− chick embryos. When AV endocardial monolayerswere cultured in M199 alone in the absence of myocardium,epithelial monolayers extended on the surface of the collagengels (Fig. 2A) but invasive mesenchymal cells were notdetected (Figs. 2B and 3) as previously reported (Krug et al.,1985; Mjaatvedt et al., 1987; Nakajima et al., 1994; Ramsdelland Markwald, 1997). In contrast, when endocardial epithelialmonolayers were cultured with BMP-2 in M199, the explantsformed invasive mesenchymal cells within 48 h (Figs. 2D and3). BMP-2-supported mesenchymal cell formation was largelyinhibited by adding dnBMPR-1B virus to the culture (Figs.2E, F and 3) similar to the effect of noggin, an antagonist toBMP-2/4 (Zimmerman et al., 1996) (Figs. 2I, J and 3).Conversely, when AV endocardial epithelial monolayers weretreated with BMP-2+caBMPR-1B they exhibited a largernumber of invasive mesenchymal cells within the collagen gel(Figs. 2H and 3).

Cultured AV endothelial monolayers were further evaluatedfor the expression of mesenchymal markers. The explants weredouble immunostained with a virus marker, AMV3c2 and

185H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

mesenchymal cell markers, anti-α-SMA and anti-TGFβ3.TGFβ3 has been used as a chick cardiac cushion mesenchymalcell marker (Nakajima et al., 1997). In the AV endocardialmonolayers treated with BMP-2+dnBMPR-1B, most of theendocardial epithelial cells were positive for AMV3c2 (Fig. 4A)but not with α-SMA (Fig. 4B) or TGFβ3 (Fig. 4C). Whereasmany mesenchymal cells stained with AMV3c2 expressed α-SMA and TGFβ-3 in AV endocardial explants treated withBMP-2+caBMPR-1B (Figs. 4D–F). These results indicated thatdnBMPR-1B-infected cells remained on the surface of the gel asα-SMA-negative endocardial epithelial cells, whereas,caBMPR-1B-infected cells became α-SMA-positive, TGFβ3-positive mesenchymal cells. Taken together, our findings

Fig. 7. AVendocardial/myocardial co-culture treated with dnBMPR-1B or caBMPR-1BG) and a mesenchymal marker, anti-α-SMA (B, D, F, and H). dnBMPR-1B-infectedepithelial cells (arrows in panels A and B). Whereas many dnBMPR-1B-non-infeccaBMPR-1B-infected cells tended to become invasive SMA-positive mesenchymal cas α-SMA-negative epithelial cells (arrowheads in panels E and F). Scale bar=100

indicated that dominant-negative BMPR-1B perturbed/inhibitedBMP-2-supported AV EMT, whereas active BMPR-1B pro-moted EMT in the chick AV endocardial monolayer.

AV endocardial epithelium was co-cultured with myocardiumon collagen gels to assess the effect of dnBMPR-1B,caBMPR-1B, or noggin

AV endocardium from stage-14− chick embryos was co-cultured with associated myocardium on the collagen gels.After 48 h in co-culture with M199 alone, endothelialmonolayers extended on the surface of the collagen gels (Fig.5A) and invasive mesenchymal cells were formed (Figs. 5B

were double immunostained with a RCAS viral marker, AMV3c2 (A, C, E, andcells tended to stay on the surface of the gel as α-SMA-negative endocardial

ted cells transformed and expressed α-SMA (arrowheads in panels C and D).ells (arrows in panels G and H), while caBMPR-1B-non-infected cells remainedμm.

Fig. 8. Quantitative evaluation of invasive mesenchymal cell formation in AVendocardial/myocardial co-culture treated with dnBMPR-1B or caBMPR-1B(Fig. 7). An infection ratio (%) was calculated from the AMV3c2-positive cellnumber/total cell number of surface endocardial endothelial cells (Endo) ormesenchymal cells (Mes) in dnBMPR-1B- or caBMPR-1B-treated explants,respectively. For endocardial endothelial cells, dnBMPR-1B-infected cellsoccupied a significantly larger portion than caBMPR-1B-infected cells(*p<0.002 by Student's t-test). Conversely, for mesenchymal cells, caBMPR-1B-infected cells occupied a significantly larger portion than dnBMPR-1B-infected cells (*p<0.002 by Student's t-test). N=number of the areas counted.

Fig. 9. Quantitative evaluation and schematic illustration of mesenchymal cellformation in endocardial/myocardial co-culture treated with dnBMPR-1B orcaBMPR-1B. dnBMPR-1B-treated explants formed significantly fewermesenchymal cells in comparison with caBMPR-1B-treated explants. If virus-infected (AMV3c2-positive) cells were specifically compared, the differencebetween dnBMPR-1B-infected explants and caBMPR-1B-infected explants wasmore prominent in mesenchymal cell formation. This diagram shows thatdnBMPR-1B significantly reduced mesenchymal cell formation from AVendocardial/myocardial co-culture. N=number of explants.

186 H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

and 6) as previously reported (Krug et al., 1985; Nakajima etal., 1994; Mjaatvedt et al., 1987; Ramsdell and Markwald,1997). When explants were treated with dnBMPR-1B, endo-cardial monolayers also grew on the surface of the collagen gels(Fig. 5C) but significantly fewer invasive mesenchymal cellswere formed (Figs. 5D and 6) similar to the effect of noggintreatments (Figs. 5G, H, and 6). In contrast, AVexplants treatedwith caBMPR-1B formed more invasive mesenchymal cells(Figs. 5F and 6) than control explants cultured in M199 alone(Fig. 5B).

To evaluate mesenchymal marker expression in AV endo-cardial explants co-cultured withmyocardium, the explants weredouble immunostained with AMV3c2, a virus marker, and anti-α-SMA, a mesenchymal cell marker. Regardless of which typesof viruses infected the cells, when mesenchymal cells wereformed, α-SMA was detected in all mesenchymal cells (Figs.7D, H) but not in endocardial epithelial cells (Figs. 7B, F). Manyendocardial epithelial cells in explants treated with dnBMPR-1Bwere AMV3c2-positive (Figs. 7A and 8) whereas significantlyfewer mesenchymal cells were AMV3c2-positive, indicatingthat the majority of dnBMPR-1B-infected endocardial cellsremained epithelial and stayed on the surface of the collagen gels(Figs. 7C and 8). Fewer number of endocardial epithelial cells inthe explants treated with caBMPR-1B showed AMV3c2 staining(Figs. 7E and 8), whereas larger numbers of mesenchymal cellsexhibited AMV3c2 staining (Figs. 7G and 8), indicating that amajority of caBMPR-1B virus-infected cells transformed intomesenchymal cells and invaded into collagen gels.

Viruses can infect cells at the surface of the explants directlyexposed to the viral solution but cannot penetrate deeper intothe tissue mass. Thus, dnBMPR-1B and caBMPR-1B viruses donot infect as many endocardial epithelial cells in co-culturedexplants as seen in endocardial monolayer cultures. Therefore,

to assess the differences in the effects of caBMPR-1B anddnBMPR-1B infection in mesenchymal cell formation, theinfection ratio of the invasive mesenchymal cells was comparedin the caBMPR-1B-infected and dnBMPR-1B-infected AVexplants co-cultured with myocardium. As seen in Fig. 9,there was a statistically significant tendency for dnBMPR-1B-infected cells to remain on the surface of the gel as α-SMA-negative endocardial epithelial cells, while caBMPR-1B-infected cells transformed into α-SMA-positive invasivemesenchymal cells (also see scheme in Fig. 15). Althoughviruses can infect surface of the myocardium in co-culturedexplants, it is less likely that the effects seen with dn- andcaBMPR-1B virus-infection were largely affected by viralinfection of the myocardium because we were able toimmunohistochemically identify AMV3c2-positive cells anddirectly compared virus-infected endocardial cells with non-infected cells in the same explant.

In AVendocardial monolayer explants TGFβ3 was expressed inmesenchymal cells induced by BMP-2 or BMP-2+caBMPR-1Bbut not expressed in endocardial epithelial cells treated withBMP-2+dnBMPR-1B or M199 alone

When AVexplants were cultured with dnBMPR-1B+BMP-2in the absence of myocardium, an endocardial monolayer grewon the surface of the collagen gels but few invasivemesenchymal cells were formed. Double immunolabeling ofanti-TGFβ3 and anti-α-SMA showed that endocardial epithelialcells expressed little TGFβ3 (Fig. 10A) or α-SMA (Fig. 10B).Whereas, when explant was cultured with BMP-2 alone in theabsence of myocardium, endocardial epithelial cells did notexpress TGFβ3 or α-SMA (asterisks in Figs. 10E, F) but manymesenchymal cells expressed TGFβ3 and α-SMA (arrows in

187H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

Figs. 10E, F). When cultured with caBMPR-1B+BMP-2, manyinvasive mesenchymal cells were formed and expressed bothTGFβ3 (Fig. 10C) and α-SMA (Fig. 10D). When explants werecultured with control medium M199 alone, only endocardialmonolayers were formed and little, if any, expression of TGFβ3(Fig. 10G) and α-SMAwas detected (Fig. 10H).

In AV endocardial/myocardial co-culture, TGFβ3 expressionwas suppressed in dnBMPR-1B-infected endocardial cellsbut enhanced in caBMPR-1B-infected endocardial epithelialcells

To address the relationship between BMP signaling andother known endocardially derived mesenchymal cell inducers,such as TGFβs (Potts and Runyan, 1989; Potts et al., 1991;Nakajima et al., 1994, 1998; Ramsdell and Markwald, 1997) inAV cushion formation, AVexplants co-cultured withmyocardium

Fig. 10. AVendocardial monolayer treated with BMP-2+dnBMPR-1B, BMP-2+caB(A, C, E, and G) and a mesenchymal marker, anti-α-SMA (B, D, F, and H). Endocexpress TGFβ3 (A) or α-SMA (B). Whereas, when explants were cultured withexpressed both TGFβ3 (C) and α-SMA (D), similar to the explants treated with BMnegative, α-SMA-negative endothelial cells, but not invasive mesenchymal cells wcells. Arrows indicate mesenchymal cells which express TGFβ3 and α-SMA. Scal

and infected with dnBMPR-1B or caBMPR-1B were doubleimmunostained with AMV3c2, a virus marker, and anti-TGFβ3.dnBMPR-1B-infected cells tended to stay on the surface of the gelas TGFβ3-negative endocardial epithelial cells (arrowheads inFigs. 11A–C and 12). Conversely, caBMPR-1B-infected cellstended to become TGFβ3-positive endocardial epithelial cells orinvasive TGFβ3-positive mesenchymal cells (arrows in Figs.11D–F and 12). Some of the endocardial cells infected withcaBMPR-1B appeared to express TGFβ3 when they were on thesurface of the gels, before they transformed into invasivemesenchymal cells (arrowheads in Figs. 11D–F). There werestatistically significant differences in TGFβ3 immunointensitybetween dnBMPR-1B-infected and caBMPR-1B-infected endo-cardial endothelial cells (Fig. 12). However, when dnBMPR-1Binfection did not completely block TGFβ3 expression, cellsappeared to transform into mesenchymal cells and expressTGFβ3 (arrows in Figs. 11A–C).

MPR-1B, BMP-2 or M199 alone were double immunostained with anti-TGFβ3ardial epithelial cells in the explants treated with BMP-2+dnBMPR-1B did notcaBMPR-1B+BMP-2, many invasive mesenchymal cells were formed andP-2 alone (E and F). When explants were cultured with M199 alone, TGFβ3-ere formed. Asterisks indicate TGFβ3-negative, α-SMA-negative endotheliale bar=100 μm.

Fig. 11. AVendocardial/myocardial co-culture treated with dnBMPR-1B or caBMPR-1B was double immunostained with a RCAS viral marker, AMV3c2 (A and D)and anti-TGFβ3 (B and E). Panels C and F are merged images of panels A and B, and panels D and E, respectively. The majority of dnBMPR-1B-infected endocardialcells remained as TGFβ3-negative epithelial cells (arrowheads in panels A–C). Only a few dnBMPR-1B-infected endocardial cells became mesenchymal andexpressed TGFβ3 (arrows in panels A–C). In contrast, the majority of caBMPR-1B-infected cells became TGFβ3-positive mesenchymal cells (arrows in panels D–F).Note that a few caBMPR-1B-infected cells remained as epithelial cells but expressed TGFβ3 (arrowheads in panels D–F). Scale bar=100 μm.

188 H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

Microinjection study of dnBMPR-1B and caBMPR-1B into AVcushion tissue

The dnBMPR-1B or caBMPR-1B virus was microinjectedinto cardiac jelly near the endocardial side of the AV canal inchick embryos in ovo at stage-13 when endocardial cells hadnot yet transformed. Eggs were incubated for an additional twodays. Virus-infected cells were detected by a viral marker,AMV3c2 immunohistochemistry, and confocal microscopy.The majority of the cells that had been infected with dnBMPR-1B remained at or near the luminal surface of the AV cushion asepithelial cells (Figs. 13A–C and 14), whereas most of thecaBMPR-1B-infected cells transformed into mesenchymal cellsand invaded deeply into AV cushion (Figs. 13D–F and 14). It isimportant to note that, in some representative cases, whilecushion mesenchyme was formed, all of the AMV3c2-positivednBMPR-1B-infected cells remained as endocardial epithelialcells (Fig. 13).

Discussion

A dominant-negative approach can circumvent functionalredundancies among Type I BMP receptors in chick AV EMT

Our present study indicated that dnBMPR-1B infection ornoggin treatment of AV endocardium significantly reducedEMT whereas caBMPR-1B infection enhanced EMT from AVendocardium in the presence of BMP-2 or associated AVmyocardium in chick AV culture (Figs. 2, 5 and 15). Moreover,our microinjection study revealed that dnBMPR-1B virus-infected cells remained in the endocardial epithelium, whereascaBMPR-1B-infected cells invaded deeper into the cushion as

mesenchyme in vivo (ovo) (Fig. 13). These data providedevidence that BMP signaling through endocardial endothelialcells was required in AV EMT and active BMPR-1B in theendocardial cells could induce AV EMT. However, our data donot exclude the possibilities that other BMP receptors in theendocardium are involved in BMP-supported AV EMT, sincedominantly expressed mutant BMPR-1B can block signaltransduction through any probable BMP receptor. For example,the chick activin type I receptor (ALK2) was implicated in AVmesenchymal cell formation/migration by the study in whichneutralizing antibodies to chALK2 were used in chick AVexplants (Lai et al., 2000). Furthermore, active ALK2 wasreported to induce ventricular endocardial cells to undergo EMTin the chick (Desgrosellier et al., 2005).

More recently, functional redundancies among endocardiallyexpressed type I BMP receptors have been suggested by thestudies in which ALK3 (BMPR-1A) or ALK2 was conditionallydeleted from endothelium in mice (Ma et al., 2005; Wang et al.,2005; Park et al., 2006; Song et al., 2007). ALK2 can bind otherTGFβ superfamily, TGFβs and Activin in addition to BMPs;however ALK3 transmits BMP signaling exclusively (deCaestecker, 2004). Regardless of the differences in ligandspecificities, mutant mice with ALK2 and ALK3 conditionalablation from the endothelial cells displayed defects in AV septaand valves, which appeared to result from a failure ofendocardial EMT. Although ALK3 conditional deletion fromthe endothelium produced a severe deficiency, neither ALK2nor ALK3 conditional deletion showed complete disruption ofEMT, despite the fact that the BMP-2 conditional deletion fromthe myocardium completely disrupted AV EMT (Ma et al.,2005; Rivera-Feliciano and Tabin, 2006). In fact, one-third ofthe ALK3 conditional deletion mutants were reported to form

Fig. 12. Quantitative evaluation of AMV3c2 and anti-TGFβ3 double immuno-stained AV endocardial/myocardial co-culture treated with dnBMPR-1B orcaBMPR-1B (Fig. 11). The data represent the average of TGFβ3 immunostainingintensity from multiple AMV3c2 positive endocardial epithelial cells ormesenchymal cells. The average immunointensity of endocardial epithelialcells in the explants cultured with M199 alone was determined as 100% intensityas control. Immunointensity of dnBMPR-1B infected endocardial epithelial cellswas weaker than control (*p<0.04 by Student's t-test), whereas, immunointen-sity of caBMPR-1B infected endocardial epithelial cells was stronger thancontrol (**p<0.0001 by Student's t-test). There was no significant difference inTGFβ3 immunointensity of mesenchymal cells within explants treated withdnBMPR-1B, caBMPR-1B or M199 alone. Number of the explants evaluated foreach treatment: 15 for dnBMPR-1B, 13 for caBMPR-1B and 5 for M199 alone.N=number of cells evaluated for immunointensity.

Fig. 13. Microinjection study of dnBMPR-1B and caBMPR-1B into AV cushiontissue. dnBMPR-1B (A–C) or caBMPR-1B (D–F) virus was microinjected intocardiac jelly near the endocardial side of AV canal in chick embryos at stage-13when endocardial cells had not transformed yet in ovo. Eggs were resealed andincubated for an additional two days. Virus-infected cells (green) were detectedby AMV3c2 immunohistochemistry and confocal microscopy (A–E). Panels Band E represent higher power views of panels A and D respectively. Panels Cand F are 5-μm sections of the AV canal of chick embryos and represent a typicalcase after virus microinjection. The majority of dnBMPR-1B-infected cellsstayed at or near the luminal surface of the AV cushion as epithelial cells (arrowsin panels A–C), whereas most of caBMPR-1B-infected cells transformed intomesenchymal cells and invaded deeper into AV cushion (arrows in panels D–F).The green color in the myocardium in panels A and D is autofluorescencecaused by stacking images by confocal microscopy. In panel C (a representativesection of AV canal from a dnBMPR-1B-treated embryo), most of theendocardial epithelial cells are AMV3c2-positive, whereas many AV cushionmesenchymal cells are AMV3c2-negative. Since 100% of virus infection cannotbe achieved, these viral marker-negative mesenchymal cells are presumablyendocardially derived cells that escaped from virus infection or cells which havedifferent origins. a, atrium; v, ventricle; av, atrioventricular canal; o, outflowtract; e, endocardial epithelial cell; m, myocardium. Scale bar in panelA=100 μm for (A and D). Scale bar in panel B=100 μm for (B, E, C, and F).

189H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

rather normal AV cushion mesenchyme (Ma et al., 2005). Thiscould result from variability of cre activity; however, it ispossible that the Type I BMP receptor in the endocardium havefunctional redundancies for AV EMT. This view is consistentwith the findings that BMPR-1B is required for chondrogenesisin the limb (Baur et al., 2000; Yi et al., 2000), where BMPR-1Bis the only Type I BMP receptor that transmits BMP signaling inchondrocytes (Abe, 2006) but not for endocardial EMT in thecardiac AV cushion, where multiple BMP receptors areexpressed in the endocardial cells. In the present study,dnBMPR-1B infection of endocardium disrupted AV EMTboth in the AV explant cultures in vitro and whole-embryocultures in vivo (ovo). The discrepancy between BMPR-1Bknockout data in the mouse and our data from dnBMPR-1B-virus infection in the chick may reflect the functional differencein endocardial BMP receptors between the two species.Nevertheless, the significance of our dominant-negativeapproach is that it does effectively demonstrate unequivocallythe necessity for BMP signaling in AV endocardial cells duringAV cushion tissue formation.

TGFβ expression in AV endocardium can be regulated by BMPsignaling

TGFβs (Potts and Runyan, 1989; Potts et al., 1991;Nakajima et al., 1994, 1998; Ramsdell and Markwald, 1997;Boyer et al., 1999a; Camenisch et al., 2002) and their receptors(Brown et al., 1996, 1999; Boyer et al., 1999b; Lai et al., 2000;

Boyer and Runyan, 2001; Desgrosellier et al., 2005) are well-documented factors implicated in AV cushion mesenchymal cellformation and migration in the chick. AV endocardial epithelialcells in the chick, upon activation by myocardial signaling,appear to express TGFβ3 as part of an in vivo autocrinemechanism for regulating and sustaining the transformationfrom endocardial to mesenchymal phenotype (Nakajima et al.,1994, 1998; Ramsdell and Markwald, 1997). While upregula-tion of TGFβ3 in the endocardial endothelial cells is critical for

Fig. 14. Quantitative evaluation of mesenchymal cell formation in the AV canalmicroinjected with dnBMPR-1B or caBMPR-1B virus (Fig. 13). Embryos werescored positive in mesenchymal cell formation when AMV3c2-positive cellswere found in cushion mesenchyme but were scored negative when AMV3c2-positive cells were found only in endocardial epithelium. These results indicate atendency of dnBMPR-1B-infected cells to remain as endocardial epithelial cellswhereas caBMPR-1B-infected cells transform into mesenchymal cells.N=number of embryos evaluated.

190 H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

AV EMT in the chick, the factor from the AV myocardium thatregulates TGFβ3 expression in AV endocardial cells has notbeen determined.

Therefore, in the present study, we sought to determine theinteractions between BMP signaling and endocardiallyexpressed TGFβ3 in AV EMT. Data from our in vitro cultureassays indicated that dnBMPR-1B infection significantlyreduced TGFβ3 expression in AV endocardial endothelialcells, whereas caBMPR-1B-infection enhanced TGFβ3-expres-sion even if caBMPR-1B-infected endocardial cells remainedon the surface of the collagen gels in culture in the presence of

Fig. 15. Inhibition and enhancement of mesenchymal cell formation by dnBMPR-1100% in co-cultured explants, non-infected cells can transform into mesenchymal ceformation is affected by the types of the virus (dn- vs. ca-BMPR-1B) infected the en

AV myocardium or BMP-2 (Figs. 11 and 12). Thus, activationof BMPR appeared to correlate with an upregulation of TGFβ3expression in endocardial endothelial cells, which supports theidea that BMP ligands secreted by the myocardium regulateTGFβ3-expression in the endocardial cells via BMP pathwaysin AVendocardial cells during EMT. Secretion of TGFβ3 in theendocardial cells, in turn, would enhance AV endocardial EMTand migration of the endocardial cells as described (Nakajima etal., 1994, 1998; Ramsdell and Markwald, 1997). Our previousreport (Sugi et al., 2004) also indicated that BMP-2 inducedTGFβ2 expression in AVendocardial cells in mouse AVexplantculture, in which TGFβ2 appears to be an endocardialactivation marker as reported by Camenisch et al. (2002). Ourresults are consistent with the recent mouse data obtained by theconditional deletion of BMP-2 using the Nkx2.5-cre systemwhich displays defects in AV endocardial EMT along withdownregulation of TGFβ2 in the AV endocardium (Ma et al.,2005). Although BMP signaling in the myocardium can alsoregulate myocardial TGFβ2 expression as evident in the Alk3conditional knockout (Gaussin et al., 2002, 2005), it is clear thatinteractions of BMP signaling and TGFβ transcriptionalregulation in the endocardial cells during AV EMT appear tobe critical for completion of EMT and remain to be elucidated.

BMP signaling through AV endocardial cells is essential foractivation of AV endocardial cells in the chick

Although some cushion mesenchyme is formed in TGFβ2knockout mice (Sanford et al., 1997; Bartram et al., 2001),TGFβ2 is highly expressed in AV myocardium at the onset ofAV EMT in mice and TGFβ2, but not TGFβ3, appears to berequired for AV EMT in mouse culture assays (Camenisch et al.,2002). However, chick TGFβ2 expression is not restricted to

B or caBMPR-1B was schematically illustrated. Since viral infection cannot bells regardless of viral treatment. However, the total amount of mesenchymal celldocardial cells.

191H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

AV myocardium but expressed in both AV endocardium andmyocardium before and during EMT and both TGFβ2 andTGFβ3 appear to be necessary for the full morphogeneticprograms of EMT in the chick (Boyer et al., 1999a; Camenischet al., 2002; Molin et al., 2003). These data suggest substantialinterspecies differences for TGFβ signaling between the chickand the mouse in the distinct temporal events during AV EMT.Regarding BMP signaling, BMP-2 expression is restricted tothe AV canal myocardium coincident with AV EMT in both thechick (Yamagishi et al., 1999; Yamada et al., 1999; Somi et al.,2004) and the mouse (Lyons et al., 1990; Sugi et al., 2004; Ma etal., 2005; Rivera-Feliciano and Tabin, 2006). In our assaysusing chick AV explants, recombinant human noggin ordnBMPR-1B disrupted cell–cell separation, an initial step ofEMT when the AV endocardial epithelium was induced byBMP-2 or associated myocardium (Figs. 2 and 5). Furthermore,our data revealed that BMP-2 added to the AV endocardialmonolayer upregulated TGFβ3 and α-SMA-expression, whichare activation markers of AV EMT in the chick (Figs. 10 and11). Therefore, consistent with our previous proposal in mice(Sugi et al., 2004), BMP signaling in endocardial cells likelyregulates activation of endocardial cells to transform them intocushion mesenchymal cells before TGFβ-signaling takes placein AV EMT in the chick. This appears to be consistent with therecent data from TGFβR2/Tie2cre mice (Jiao et al., 2006). Ourcurrent work using RCAS-viral gene transfer to activate orinactivate BMP receptor function in the endocardium in vitroand in vivo (ovo) has provided evidence that BMP signalingthrough the endocardium was essential for AV EMT in thechick. Our data further support the proposal that myocardialBMP-2 directly affects AV endocardium through a paracrinepathway that is essential for EMT and that activation of BMPreceptors in endocardial cells induces AV EMT.

Acknowledgments

The authors thank Dr. L. Niswander (Howard HughesMedical Institute and the University of Colorado HealthSciences Center–Fitzsimmons Campus Denver, CO) for kindlyproviding us with dnBMPR-1B and caBMPR-1B constructsinserted into the RCASA viral vector and Dr. T. Nohno(Kawasaki Medical University, Okayama, Japan) for providingthe cDNA clone of chick BMPR-1B. Recombinant humanBMP-2 protein was a generous gift from Genetics Institute, Inc.(Cambridge, MA). Recombinant human noggin was kindlyprovided by Regeneron Pharmaceuticals, Inc. (Tarrytown, NY).Monoclonal antibody, AMV3c2, was obtained from theDevelopmental Studies Hybridoma Bank maintained by theDepartment of Pharmacology and Molecular Sciences (MD)and the Department of Biological Sciences, University of Iowa(Iowa City, IA) under contract N01-HD-6-2915 from theNICHD. The authors are also grateful to Dr. Thomas Trusk, Mr.Joshua Spruill, and Ms. Sue Tjepkema-Barrows for their helpwith computer-assisted photography. This work was conductedin a facility constructed with support from the NIH (grant C06RR018823) from the Extramural Research Facilities Program ofthe National Center for Research Resources. This work was

supported by AHA-SC (YS) and HL52813 and HL33756(RRM and YS).

References

Abe, E., 2006. Function of BMPs and BMP antagonists in adult bone. Ann.N. Y. Acad. Sci. 1068, 41–53.

Bartram, U., Molin, D.G.M., Wisse, L.J., Mohamad, A., Sanford, L.P.,Doetcshman, T., Speer, C.P., Polemann, R.E., Gittenberger-de Groot, A.C.,2001. Double-outlet right ventricle and overriding tricuspid valve reflectdisturbances of looping, myocardialization, endocardial cushion differentia-tion, and apoptosis in TGFβ2-knockout mice. Circulation 103, 2745–2752.

Baur, S.T., Mai, J.J., Dymecki, S.M., 2000. Combinatorial signaling throughBMP receptor 1B and GDF5: shaping of the distal mouse limb and thegenetics of distal limb diversity. Development 127, 605–619.

Beppu, H., Kawabata, M., Hamamoto, T., Chytil, A., Minowa, O., Noda, T.,Miyazono, K., 2000. BMP type II receptor is required for gastrulation andearly development of mouse embryos. Dev. Biol. 221, 2249–2258.

Bernanke, D.H., Markwald, R.R., 1982. Migratory behavior of cardiac cushiontissue cells in a collagen lattice culture system. Dev. Biol. 91, 235–245.

Boyer, A.S., Runyan, R.B., 2001. TGFβ type III and TGFβ type II receptorshave distinct activities during epithelial–mesenchymal cell transformation inthe embryonic heart. Dev. Dyn. 221, 454–459.

Boyer, A.S., Ayerinskas, I.I., Vincent, E.B., McKinney, L.A., Weeks, D.L.,Runyan, R.B., 1999a. TGFβ2 and TGFβ3 have separate and sequentialactivities during epithelial–mesenchymal cell transformation in theembryonic heart. Dev. Biol. 208, 530–545.

Boyer, A.S., Erickson, C.P., Runyan, R.B., 1999b. Epithelial–mesenchymaltransformation in the embryonic heart is mediated through distinct pertussistoxin-sensitive and TGFβ signal transduction mechanisms. Dev. Dyn. 214,81–91.

Brown, C.B., Boyer, A.S., Runyan, R.B., Barnett, J.V., 1996. Antibodies to thetype II TGFβ receptor block cell activation and migration duringatrioventricular cushion transformation in the heart. Dev. Biol. 174,248–257.

Brown, C.B., Boyer, A.S., Runyan, R.B., Barnett, J.V., 1999. Requirement oftype III TGF-β receptor for endocardial cell transformation in the heart.Science 283, 2080–2082.

Camenisch, T.D., Molin, D.G.M., Person, A., Runyan, R.B., Gittenberger-deGroot, A.C., McDonald, J.A., Klewer, S.E., 2002. Temporal and distinctTGFβ ligand requirements during mouse and avian endocardial cushionmorphogenesis. Dev. Biol. 248, 170–181.

Chomczynski, P., Sacchi, N., 1987. Single step method of RNA isolation by acidguanidinium thiocyanate–phenol–chloroform extraction. Anal. Biochem.162, 156–159.

de Caestecker, M., 2004. The transforming growth factor-β superfamily ofreceptors. Cytokine Growth Factor Rev. 15, 1–11.

Desgrosellier, J.S., Mundell, N.A., Mcdonnell, M.A., Moses, H.L., Barnett, J.V.,2005. Activin receptor kinase 2 and smad6 regulate epithelial–mesenchymaltransformation during cardiac valve formation. Dev. Biol. 280, 201–210.

Dijke, P.T., Yamashita, H., Sampath, T.K., Reddi, A.H., Estevez, M., Riddle,D.L., Ichijo, H., Heldin, C.-H., Miyazono, K., 1994. Identification of typeI receptors for osteogenic protein-1 and bone morphogenetic protein-4.J. Biol. Chem. 269, 16985–16988.

Eisenberg, L.M., Markwald, R.R., 1995. Molecular regulation of atrioventri-cular valvuloseptal morphogenesis. Circ. Res. 77, 1–6.

Francis-West, P.H., Tatla, T., Brickell, P.M., 1994. Expression patterns of thebone morphogenetic protein genes Bmp-4 and Bmp-2 in the developingchick face suggest a role in outgrowth of the primordia. Dev. Dyn. 201,168–178.

Gaussin, V.C., Van de Putte, T., Mishina, Y., Hanks, M.C., Zwijsen, A.,Huylebroeck, D., Behringer, R.R., Schneider, M.D., 2002. Endocardialcushion and myocardial defects after cardiac myocyte-specific conditionaldeletion of the bone morphogenetic protein receptor ALK3. Proc. Natl.Acad. Sci. U. S. A. 99, 2878–2883.

Gaussin, V., Morley, G.E., Cox, L., Zwijsen, A., Vance, K.M., Emile, L., Tian,Y., Liu, J., Hong, C., Myers, D., Conway, S.J., Depre, C., Mishina, Y.,

192 H. Okagawa et al. / Developmental Biology 306 (2007) 179–192

Behringer, R.R., Hanks, M.C., Schneider, M.D., Huylebroeck, D., Fishman,G.I., Burch, J.B., Vatner, S.F., 2005. Alk3/Bmpr1a receptor is required fordevelopment of the atrioventricular canal into valves and annulus fibrosus.Circ. Res. 77, 219–226.

Hamburger, V., Hamilton, H.E., 1951. A series of normal stages in thedevelopment of the chick embryo. J. Morphol. 88, 49–92.

Hogan, B.L.M., 1996. Bone morphogenetic proteins: multifunctional regulatorsof vertebrate development. Genes Dev. 10, 1580–1594.

Hughes, S.H., Greenhouse, J.J., Petropoulos, C.J., Sutrave, P., 1987. Adapterplasmids simplify the insertion of foreign DNA into helper-independentretroviral vectors. J. Virol. 61, 3004–3012.

Jiao, K., Lanworthy, M., Batts, L., Brown, C.B., Moses, H.L., Baldwin, S.,2006. Tgfβ signaling is required for atrioventricular cushion mesenchymeremodeling during in vivo cardiac development. Development 133,4585–4593.

Jones, C.M., Lyons, K.M., Hogan, B.L.M., 1991. Involvement of bonemorphogenetic protein-4 (BMP-4) and Vgr-1 in morphogenesis in themouse. Development 111, 531–542.

Krug, E.L., Runyan, R.B., Markwald, R.R., 1985. Protein extracts from earlyembryonic hearts initiate cardiac endothelial cytodifferentiation. Dev. Biol.112, 414–426.

Lai, Y.T., Beason, K.B., Brames, G.P., Desgrosellier, J.S., Cleggett, M.C.,Shaw, M.V., Brown, C.B., Barnett, J.V., 2000. Activin receptor-like kinase2 can mediate atrioventricular cushion transformation. Development 222,1–11.

Lyons, K.M., Pelton, R.W., Hogan, B.L.M., 1990. Organogenesis and patternformation in the mouse: RNA distribution patterns suggest a role for bonemorphogenetic protein-2A (BMP-2A). Development 109, 833–844.

Ma, L., Lu, M.-F., Schwartz, R.J., Martin, J.F., 2005. Bmp is essential for cardiaccushion epithelial–mesenchymal transition and myocardial patterning.Development 132, 5601–5611.

Markwald, R.R., Eisenberg, L., Eisenberg, C., Sugi, Y., 1996. Epithelial–mesenchymal transformations in early avian heart development. Acta Anat.156, 173–186.

Mishina, Y., Suzuki, A., Ueno, N., Boehringer, R.R., 1995. Bmpr encodes a typeI bone morphogenetic protein receptor that is essential for gastrulationduring mouse embryogenesis. Genes Dev. 9, 3027–3037.

Mjaatvedt, C.H., Lepera, R.C., Markwald, R.R., 1987. Myocardial specificityfor initiating endothelial–mesenchymal cell transition in embryonic chickheart correlates with a particulate distribution of fibronectin. Dev. Biol. 119,59–67.

Molin, D.G.M., Bartram, U., Van der Heiden, K., Van Iperen, L., Speer, C.P.,Hierch, B.P., Poelmann, R.E., Gittenberger-de-Groot, A.C., 2003. Expres-sion patterns of tgfβ 1–3 associate with myocardialisation of the outflowtract and the development of the epicardium and the fibrous heart skeleton.Dev. Dyn. 227, 431–444.

Nakajima, Y., Krug, E.L., Markwald, R.R., 1994. Myocardial regulation oftransforming growth factor beta expression by outflow tract endothelium inthe early embryonic chick heart. Dev. Biol. 165, 615–626.

Nakajima, Y., Mironov, V., Yamagishi, T., Nakamura, H., Markwald, R.R.,1997. Expression of smooth muscle alpha-actin in mesenchymal cells duringformation of avian endocardial cushion tissue: a role for transforminggrowth factor beta3. Dev. Dyn. 209, 296–309.

Nakajima, Y., Yamagishi, T., Nakamura, H., Markwald, R.R., Krug, E.L., 1998.An autocrine function for transforming growth factor (TGF)-β3 in thetransformation of atrioventricular canal endocardium into mesenchymeduring chick heart development. Dev. Biol. 194, 99–113.

Nakajima, Y., Yamagishi, T., Hokari, S., Nakamura, H., 2000. Mechanismsinvolved in valvuloseptal endocardial cushion formation in early cardiogen-esis: roles of transforming growth factor (TGF)-β and bone morphogeneticprotein (BMP). Anat. Rec. 258, 119–127.

Okagawa, H., Hutson, D.R., Markwald, R.R., Sugi, Y., 2000. Role of bonemorphogenetic protein receptor 1B in atrioventricular cushion mesenchymalcell formation. FASEB J. 14, 275a.

Park, C., Lavine, K., Mishina, Y., Deng, C.-X., Ornitz, D.M., Choi, K., 2006.Bone morphogenetic protein receptor 1A signaling is dispensable forhematopoietic development but essential for vessel and atrioventricularendocardial cushion formation. Development 133, 3473–3484.

Person, A.D., Klewer, S.E., Runyan, R.B., 2005. Cell biology of cardiac cushiondevelopment. Int. Rev. Cytol. 243, 287–335.

Potts, J.D., Runyan, R.B., 1989. Epithelial–mesenchymal cell transformation inthe embryonic heart can be mediated, in part by transforming growth factorβ. Dev. Biol. 134, 392–401.

Potts, J.D., Dagle, J.M., Walder, J.A., Weeks, D.L., Runyan, R.B., 1991.Epithelial–mesenchymal transformation of embryonic cardiac endothelialcells in inhibited by a modified antisense oligodeoxynucleotide totransforming growth factor β3. Proc. Natl. Acad. Sci. U. S. A. 88,1516–1520.

Raftery, L.A., Sutherland, D.J., 1999. TGF-β family signal transduction inDrosophila development: from Mad to Smads. Dev. Biol. 210, 251–268.

Ramsdell, A.F., Markwald, R.R., 1997. Induction of endocardial cushion tissuein the avian heart is regulated, in part, by TGFbeta-3-mediated autocrinesignaling. Dev. Biol. 188, 64–74.

Rivera-Feliciano, J., Tabin, C.J., 2006. Bpm2 instructs cardiac progenitors toform the heart-valve-inducing field. Dev. Biol. 295, 580–588.

Romano, L.A., Runyan, R.B., 2000. Slug is an essential target of TGFβ2signaling in the developing chicken heart. Dev. Biol. 223, 91–102.

Runyan, R.B., Markwald, R.R., 1983. Invasion of mesenchyme into three-dimensional collagen gels: a regional and temporal analysis of interaction inembryonic heart tissue. Dev. Biol. 95, 108–114.

Sanford, L.P., Ormsby, L., Gittenberger-de Groot, A.C., Sariola, H., Friedman,R., Boivin, G.P., Cardell, E.L., Doetschman, T., 1997. Tgf-beta-2 knockoutmice have multiple developmental defects that are nonoverlapping withother Tgf-beta knockout phenotypes. Development 124, 2659–2670.

Somi, S., Buffing, A.A., Moorman, A.F., Van Den Hoff, M.J., 2004. Dynamicpatterns of expression of BMP isoforms 2, 4, 5, 6, and 7 during chicken heartdevelopment. Anat. Rec., A. Discov. Mol. Cell. Evol. Biol. 279, 636–651.

Song, L., Faessler, R., Mishina, Y., Jiao, K., Baldwin, H.S., 2007. Essentialfunction of Alk3 during AV cushion morphogenesis in mouse embryonichearts. Dev. Biol. 301, 271–286.

Sugi, Y., Markwald, R.R., 2003. Endodermal growth factors promoteendocardial precursor cell formation from precardiac mesoderm. Dev.Biol. 263, 35–49.

Sugi, Y., Yamamura, H., Okagawa, H., Markwald, R., 2004. Bone morphoge-netic protein-2 can mediate myocardial regulation of atrioventricularcushion mesenchymal cell formation in mice. Dev. Biol. 296, 505–518.

Wall, N.A., Hogan, B.L.M., 1995. Expression of bone morphogenetic protein-4(BMP-4), bone morphogenetic protein-7 (BMP-7), fibroblast growth factor-8 (FGF-8) and sonic hedgehog (SHH) during branchial arch development inthe chick. Mech. Dev. 53, 383–392.

Wang, J., Sridurongrit, S., Dudas, M., Thomas, P., Nagy, A., Schneider, M.D.,Epstein, J.A., Kaartinen, V., 2005. Atrioventricular cushion transformation ismediated by ALK2 in the developing mouse heart. Dev. Biol. 286, 209–310.

Yamada, M., Szendro, P.I., Prokscha, A., Schwartz, R.J., Eichele, G., 1999.Evidence for a role of Smad6 in chick cardiac development. Dev. Biol. 215,48–61.

Yamagishi, T., Nakajima, Y., Miyazono, K., Nakamura, H., 1999. Bonemorphogenetic protein-2 acts synergistically with transforming growthfactor-beta 3 during endothelial–mesenchymal transformation in thedeveloping chick heart. J. Cell. Physiol. 180, 35–45.

Yamashita, H., Dijke, P.T., Heldin, C.H., Miyazono, K., 1996. Bonemorphogenetic protein receptors. Bone 19, 569–574.

Yi, S.E., Daluiski, A., Pederson, R., Rosen, V., Lyons, K.M., 2000. The type IBMP receptor BMPRIB is required for chondrogenesis in the mouse limb.Development 127, 621–630.

Zhang, H., Bradley, A., 1996. Mice deficient for BMP2 are nonviable and havedefects in amnion/chorion and cardiac development. Development 122,2977–2986.

Zimmerman, L.B., Jesus-Escobar, J.M.D., Harland, R.M., 1996. The spemannorganizer signal noggin binds and inactivates bone morphogenetic protein 4.Cell 86, 596–606.

Zou, H., Niswander, L., 1996. Requirement for BMP signaling in interdigitalapoptosis and scale formation. Science 272, 738–741.

Zou, H., Wieser, R., Masssague, J., Niswander, L., 1997. Distinct roles of type Ibone morphogenetic protein receptors in the formation and differentiation ofcartilage. Genes Dev. 11, 2192–2203.