developmental biology 247, 307–326 (2002) … · 2007. 5. 3. · experimental studies on the...

Developmental Biology 247, 307–326 (2002)doi:10.1006/dbio.2002.0706

Experimental Studies on the SpatiotemporalExpression of WT1 and RALDH2 in the EmbryonicAvian Heart: A Model for the Regulationof Myocardial and Valvuloseptal Developmentby Epicardially Derived Cells (EPDCs)

J. M. Pérez-Pomares,* A. Phelps,* M. Sedmerova,* R. Carmona,†M. González-Iriarte,† R. Muñoz-Chápuli,† and A. Wessels* ,1

*Department of Cell Biology and Anatomy, Cardiovascular Developmental Biology Center,Medical University of South Carolina, Charleston, South Carolina 29425; and †Departmentof Animal Biology, Faculty of Science, University of Málaga, 29071 Málaga, Spain

Epicardially derived cells (EPDCs) delaminate from the primitive epicardium through an epithelial-to-mesenchymaltransformation (EMT). After this transformation, a subpopulation of cells progressively invades myocardial and valvulosep-tal tissues. The first aim of the study was to determine the tissue-specific distribution of two molecules that are thought toplay a crucial function in the interaction between EPDCs and other cardiac tissues, namely the Wilms’ Tumor transcriptionfactor (WT1) and retinaldehyde-dehydrogenase2 (RALDH2). This study was performed in normal avian and in quail-to-chickchimeric embryos. It was found that EPDCs that maintain the expression of WT1 and RALDH2 initially populate thesubepicardial space and subsequently invade the ventricular myocardium. As EPDCs differentiate into the smooth muscleand endothelial cell lineage of the coronary vessels, the expression of WT1 and RALDH2 becomes downregulated. Thisprocess is accompanied by the upregulation of lineage-specific markers. We also observed EPDCs that continued to expressWT1 (but very little RALDH2) which did not contribute to the formation of the coronary system. A subset of these cellseventually migrates into the atrioventricular (AV) cushions, at which point they no longer express WT1. The WT1/RALDH2-negative EPDCs in the AV cushions do, however, express the smooth muscle cell marker caldesmon. The secondaim of this study was to determine the impact of abnormal epicardial growth on cardiac development. Experimental delayof epicardial growth distorted normal epicardial development, reduced the number of invasive WT1/RALDH2-positiveEPDCs, and provoked anomalies in the coronary vessels, the ventricular myocardium, and the AV cushions. We suggest thatthe proper development of ventricular myocardium is dependent on the invasion of undifferentiated, WT1-positive, retinoicacid-synthesizing EPDCs. Furthermore, we propose that an interaction between EPDCs and endocardial (derived) cells isimperative for correct development of the AV cushions. © 2002 Elsevier Science (USA)

Key Words: cardiac development; EPDCs; epicardium; RALDH2; WT1.

INTRODUCTION

The primitive cardiac tube initially consists of two con-centric layers of cells: an inner endothelial layer, betterknown as the endocardium, and an outer layer formed byprimitive myocardial cells (reviewed in Wessels and Mark-

1

0012-1606/02 $35.00© 2002 Elsevier Science (USA)All rights reserved.

wald, 2000). Both layers are separated by an extracellularmatrix-rich, acellular space, commonly referred to as thecardiac jelly. After completion of cardiac looping, a thirdlayer of cells appears on the myocardial surface of the heart.This epithelial layer is known as the epicardium (Viraghand Challice, 1973). Whereas both the endocardium andmyocardium differentiate from the precardiac mesoderm ofthe original heart fields (Tam and Schoenwolf, 1999; Loughand Sugi, 2000; Wessels and Markwald, 2000), the epicardial
To whom correspondence should be addressed. Fax: (843) 792-
0664. E-mail: [email protected].
cells derive from an extracardiac source of coelomic origin,
307

the proepicardium (Viragh and Challice 1981; Viragh et al.,1993; Männer et al., 2001).

After the formation of the epicardium (Ho and Shimada,1978; Komiyama et al., 1987; Hiruma and Hirakow, 1989;Hirakow, 1992), an epithelial-to-mesenchymal (EMT) trans-formation of a subset of epicardial cells (epicardial EMT)results in the generation of epicardially derived cells (EP-DCs), which initially migrate into the subepicardial space.In vitro epicardial EMT reportedly can be induced byseveral growth factors, including epidermal growth factor(EGF) and basic fibroblast growth factor (bFGF) (Dettman etal., 1998; Morabito et al., 2001). Local accumulation ofEPDCs in the subepicardial space of junctional areas of theheart occurs; these regions include the interventricularsulcus, the atrioventricular (AV) junction, and conoven-tricular (CV) grooves.

The developmental fate of EPDCs can be studied by usingseveral techniques. In particular, the quail-to-chick chi-mera approach (transplantation of a quail proepicardiumonto a chick host heart) has been proven extremely infor-mative. This technique allows the immunohistochemicalidentification of virtually all (pro)epicardially derived quaildonor cells with a quail-specific antibody (QCPN), while atthe same time characterizing their state of differentiationby using cell type-specific (e.g., endothelial and smoothmuscle cell) antibodies. The above-mentioned studies havedemonstrated that, after their formation, a subpopulation ofEPDCs migrates into myocardial tissues (specifically theAV and ventricular myocardium) and differentiates in situinto interstitial fibroblasts, coronary smooth muscle cells,and coronary endothelium (Pérez-Pomares et al., 1997,1998; Dettman et al., 1998; Gittenberger-de Groot et al.,1998; Männer, 1999). The quail-to-chick chimera studieshave also demonstrated that a subset of EPDCs has theability to migrate through the myocardial wall of thedeveloping heart to finally populate the mesenchymal tis-sue of the AV endocardial cushions. Interestingly, it appearsthat EPDCs do not migrate into the endocardial cushions ofthe outflow tract (Männer, 1999).

In recent years, a series of markers have been describedthat might provide more insight into the cell biology of theepicardial and epicardially derived cell. The novel celladhesion molecule bves is expressed in the proepicardium,epicardium, and EPDCs (Reese and Bader, 1999; Wada et al.,2001) and seems to play an important role in epicardialcell–cell interactions. Other factors preferentially ex-pressed in epicardial cells and of crucial importance fornormal cardiac development are FOG2 (Tevosian et al.,2000), alpha-4 integrin (Yang et al., 1995; Pinco et al., 2001),epicardin/capsulin (Hidai et al., 1998; Robb et al., 1998),and Wilms’ tumor transcription factor (WT1; Moore et al.,1999).

In this paper, we specifically focus on the tissue distri-bution of two epicardially expressed molecules that mayplay a crucial role in the interaction between EPDCs andother cardiac tissues, namely WT1 and retinaldehyde-dehydrogenase2 (RALDH2). In a previous study, we dem-

onstrated strong cardiac expression of WT1 restricted tothe late proepicardial stage (H/H18), the developing epi-cardium, and presumptive EPDCs in avian embryos (Car-mona et al., 2001). WT1 expression has also been de-scribed in the developing epicardium in the mouse(Moore et al., 1998). The WT1 gene encodes for a zinc-finger transcription factor reported to be related withnormal cell division and differentiation (Armstrong et al.,1993). WT1 putative targets seem to be multiple, includ-ing the genes of some receptors for retinoic acid (RA) andinsulin-like growth factor1 (IGF1), as well as othergrowth factor genes like the IGF2 or the platelet-derivedgrowth factor-A (PDGF-A) (Little et al., 1999). WT1knockout (k.o.) mice are characterized by underdevelop-ment of the urogenital system (Kreidberg et al., 1993) andby an abnormally developed epicardium with greatlyreduced numbers of EPDCs combined with perturbationof coronary system (Moore et al., 1999). Although WT1 isonly expressed in the epicardium and related cells, thek.o. embryos eventually die because of severe impair-ment of myocardial development resulting in pericardialhemorrhage (Moore et al., 1999).

RALDH2 is a key enzyme in the synthesis of retinoic acid(McCaffery and Dräger, 1995) and is expressed in primitiveavian (Xavier-Neto et al., 2000) and murine (Moss et al.,1998) epicardium but not in myocardial or endocardialcells. The role of retinoic acid (RA), its receptors, andrelated molecules in cardiac development is complex andhas not been elucidated completely. In the case ofRALDH2, it has been postulated that its RA syntheticactivity is related to muscular differentiation as RALDH2-producing mesenchyme invades myogenic populations(Berggren et al., 2001). Cardiac developmental defects canbe induced by depriving the heart of RA as well as byexposing the heart to increased levels of RA. Cardiac defectsare also found in the retinoic acid X receptor (RXR)-alphak.o. mouse (Gruber et al., 1996). One of the features of theRXR-alpha k.o. mouse is a characteristic hypoplastic com-pact (“thin”) myocardial layer (Dyson et al., 1995) thatclosely resembles the myocardial phenotype of the WT1-defective mouse.

All the observations related to the experimental or ge-netic disruption of epicardial development strongly suggestthat the epicardium and the EPDCs play a very importantrole in cardiac development. Indeed, a series of studies showthat experimental and genetic perturbation of epicardialdevelopment leads to severe anomalies in myocardial andcoronary development. (Männer, 1993; Kwee et al., 1995;Yang et al., 1995; Moore et al., 1999; Gittenberger-de Grootet al., 2000). Very little is known, however, about thecellular and molecular mechanisms involved in the inter-action between EPDCs and other tissues in the heart duringcardiogenesis.

In this paper, we describe the origin and fate of EPDCs innormal avian and quail-to-chick chimeric embryos usingRALDH2, WT1, and a panel of “tissue-specific”/differentiation markers. These markers included cytokera-

308 Pérez-Pomares et al.

© 2002 Elsevier Science (USA). All rights reserved.

tin (CK), a good marker for early epicardial development(Vrancken Peeters et al., 1995; Pérez-Pomares et al., 1997),QCPN, a quail-specific antibody, QH1 specific for quail(hem)angioblasts (Pardanaud et al., 1987), and caldesmon/smooth muscle alpha-actin as markers for smooth musclecells and related cell types (Mikawa and Gourdie, 1996). Wethen moved to experimental in ovo models to disruptepicardial development (proepicardial ablations and proepi-cardial block) to analyze in detail how perturbation ofEPDC development interferes with cardiac development.

MATERIALS AND METHODS

The animals used in this study were handled in compliance withthe Guide for the Care and Use of Laboratory animals published bythe U.S. National Institutes of Health (NIH). Chick and quail eggswere kept in an incubator at 37°C. The avian embryos were stagedaccording to Hamburger and Hamilton (1951).

Proepicardial Quail/Chick Chimeras

Quail-to-chick proepicardial chimeras were essentially preparedas previously described (Männer, 1999) with the following modifi-cations: Host chick embryos were incubated until stages H/H16–

H/H17. At this point, the eggs were carefully windowed. Usingsharpened tungsten needles, small openings were made through thevitelline and chorionic membranes to expose the pericardial cavity.For each embryo, a small piece of the eggshell membrane was cutwith iridectomy scissors and made to fit exactly between thesinoatrial sulcus and the caudal vitelline veins. Then, stageH/H16–H/H17 quail embryo donors were excised and perfusedwith EBSS (GIBCO). The heart was removed by cutting it throughthe outflow tract and the sinoatrial sulcus. The previously preparedeggshell membrane was introduced through the omphalomesen-teric vein of the quail embryo and pushed until it reached thecardiac lumen, so that the sinus venosus formed a cuff around themembrane, holding the donor proepicardium on its surface (see Fig.2B). The membrane carrying the quail (donor) proepicardium wasinserted facing the ventricular heart surface. After the operation,the eggs were sealed with Scotch tape and reincubated to obtainH/H25–H/H26 (2 chimeras, 96–108 h of incubation), H/H28–H/H29 (4 chimeras, 132–144 h of incubation), or H/H32 (2 chimeras,180 h of incubation). Four chimeras were fixed in modified Am-sterdam’s Fixative (methanol:acetone:water � 2:2:1), dehydrated ina graded series of ethanol, cleared in toluene, carefully oriented,and embedded in paraffin (Paraplast Plus, OXFORD Labware,St. Louis). Finally, 5-�m serial sections were mounted on micro-scope slides (Superfrost/Plus, Fisherbrand, Fisher Scientific, Pitts-burgh, PA). For cryostat sections, four chimeras were processed asdescribed for WT1 immunohistochemistry.

FIG. 1. Expression pattern of CK, RALDH2, and WT1 in the developing avian heart. (A, D, G, J) The specific WT1 expression in epicardialand EPDCs throughout stages H/H19–H/H30. (A) Adhesion of proepicardial villi (arrow) and formation of the epicardial layer (arrowheads)in a H/H19 chick embryo. Note that all the cells express WT1. The coelomic epithelium (double arrowhead) acts as positive control. Bar,30 �m. (D) A detail of the atrioventricular groove of a H/H26 chick embryo. WT1 expression shows a gradient from the outer areas (arrow,strong expression) to the inner areas (asterisk, weak expression) of the groove. Note the presence of WT1-positive EPDCs in the ventricularmyocardium (arrowheads) but not in the atrial myocardium. Bar, 30 �m. (G) The first WT1-positive cells appearing in the ventricular wallof an H/H26 chick embryo are pointed with arrowheads. Pericardium is also WT1-positive. Bar, 30 �m. (J) The ventricular compactmyocardium of an H/H30 chick embryo is presented. The arrowheads indicate isolated WT1-positive cells and arrows point to groups ofcells. Note the absence of WT1-positive cells in the ventricular trabeculated myocardium. Bar, 30 �m. (B, E, H, K) The RALDH2 expressionin epicardial and EPDCs throughout stages H/H19–H/H29. (B) First stages of epicardial development (H/H19 chick embryo). Theproepicardial mass is still prominent and strongly expresses RALDH2 (arrow). Attachment of proepicardial cells has generated an epicardialmonolayer around the AV region and some close areas in the atrium and ventricle (arrowheads). Subepicardial mesenchymal cells arealmost absent from the heart at this point. Bar, 80 �m. (E) The AV region of a stage H/H26 chick embryo is shown. RALDH2 expressionis strong in the EPDCs closer to the epicardium (arrow), while cells in contact with the myocardium have a lower level of WT1 expression(asterisk). Bar, 20 �m. (H) A magnification of the compact ventricular myocardium of a stage H/H26 chick embryo shows the presence ofRALDH2-positive cells into the subepicardial space (arrows) and myocardial layers (arrowheads). Bar, 40 �m. (K) A detail of the ventricularmyocardium of a stage H/H29 chick embryo. Epicardial epithelium expresses RALDH2. The expression in the subepicardial mesenchymeis heterogeneous; some cells strongly express RALDH2 but others do not. Some RALDH2-positive cells can be found infiltrated in themyocardial layers (arrowheads). Bar, 40 �m. (C, F, I, L) The CK expression in epicardial and EPDCs throughout H/H19–H/H29 stages ofavian development. (C) The formation of epicardium (EP) around the AV myocardium is shown as it develops from proepicardial villi (PV)in a stage H/H19 chick embryo. Both epicardium and proepicardial tissue are CK-positive. The few mesenchymal cells that start to populatethe subepicardial matrix are also CK-positive (arrowheads). Bar, 70 �m. In (F), the AV groove of a stage H/H26 chick embryo is presented.Epicardium (EP) and most subepicardial cells (arrowheads) are CK-positive all through the subepicardium. Some CK-expressing cells can beseen in the outer layers of the atrioventricular myocardium of a stage H/H26 chick embryo (arrow). Bar, 13 �m. (I) The CK expression inthe embryonic epicardium of a stage H/H26 chick embryo. Note that some CK-positive cells appear in the ventricular compact myocardiallayer (arrowheads). Bar, 20 �m. In (L), an image of the ventricular wall of a stage H/H29 chick embryo is shown. All the epicardialmesothelium is CK-positive. Some of the subepicardial mesenchymal cells do express CK, whereas this expression is weak or absent inother subepicardial cells. No CK-positive cells are seen within the ventricular myocardium. Bar, 20 �m. A, atrium; AM, atrial myocardium;AVC, atrioventricular canal; AVG, atrioventricular groove; AVM, atrioventricular myocardium; CVM, compact ventricular myocardium;EP, epicardium; EN, endocardium; LI, liver; PE, pericardium; PV, proepicardial villi; SE, subepicardium; SV, sinus venosus; VM, ventricularmyocardium; TVM, trabeculated ventricular myocardium.

309WT1 and RALDH2 in the Embryonic Avian Heart


FIG. 2. Quail-to-chick chimeras. Cartoons in (A–C) depict the basic steps of the quail-to-chick proepicardial transplantation. As shown,the entire quail proepicardium from stage H/H16–H/H17 is transplanted together with a piece of the sinus venosus (all in green) by usinga piece of eggshell membrane as a carrier (in blue). In (B), the specific location of the explant is shown. (C) The chimeric results. All thecartoons are illustrated with laser confocal microscopy pictures. Quail nuclei (QCPN-positive) appear as green dots. (A�) A QCPN-stainedH/H18 quail proepicardium; all the nuclei, including those of the mesothelial cells (arrows), and associated mesenchyme (arrowheads)appear labeled as green dots. Bar, 12 �m. In (B�), the myocardially attached explant (E) is shown in an H/H29 chimera. Note the EPDCs inthe epicardium (EP), subepicardium (arrows), myocardium (double arrowheads), and endocardium of the trabeculae (arrowheads). Bar, 170�m. The detail in (B�) presents the tissue bridge that joins the explant with the myocardial surface in the same specimen (arrowheads). Thearrows indicate the presence of QCPN-positive cells in the subepicardial space. Bar, 35 �m. (C�) The formation of epicardial andsubepicardial tissue from the donor (H/H29 chimera). Vascular structures are indicated with asterisks. Bar, 20 �m. In (C�), EPDCs in thedeveloping AV cushions of an H/H36 chimera are shown. EPDCs are pointed with arrowheads. Bar, 20 �m. A, atrium; CM, cushionmesenchyme; E, explant; EN, endocardium; EP, epicardium; MYO, myocardium; O, outflow tract; SE, subepicardium; and V, ventricle.



ImmunohistochemistryThe normal expression pattern of WT1 was studied in a series of

Japanese quail embryos (Coturnix coturnix japonica) ranging fromstages H/H18 to H/H30 and in chick embryos (Gallus gallus) ofstages H/H18–H/H31. For WT1 immunohistochemistry, the em-bryos were excised and cryoprotected in 10, 20, and 30% sucrosesolutions in TPBS (Tris phosphate-buffered saline), where theywere kept at 4°C until they sunk. Then, the embryos wereembedded in OCT-embedding media (TissueTek) and snap-frozenin liquid nitrogen-cooled isopentane. Sections (10–14 �m) wereobtained in a cryostat, collected on poly-L-lysine-coated slides, andfixed for 10 min in 1:1 methanol/acetone solution at �20°C.Sections were rehydrated in TPBS and nonspecific binding siteswere saturated for 30 min with 4% normal goat serum, 1% bovineserum albumin, and 0.1% Triton X-100 in PBS (GBT). The slideswere incubated overnight at 4°C by using a polyclonal anti-humanWT1 (Santa Cruz, C-19) antibody diluted 1:50 in PBS (0.04 mgIgG/ml) followed by an incubation (1 h, room temperature) in asecondary biotin-conjugated anti-rabbit goat IgG (Sigma) diluted1:100 in GBT. Then, the sections were incubated for 1 h inextravidin–peroxidase complex (Sigma) diluted 1:150 in TPBS,washed in TPBS, and the peroxidase activity developed with SigmaFast 3,3�-diaminobenzidine tablets (DAB) according to the suppli-er’s instructions. Cytokeratin (Dako Z0622; diluted 1:100 in GBT)or RALDH2 (gift from Drs. P. McCaffery and U. Dräger; diluted1:5000 in GBT) expressions were studied in 10-�m paraffin sectionsfollowing the same protocol. Caldesmon (gift from Dr. Gourdie)and alpha-smooth muscle cell actin (Sigma A-2547) stainings innormal control (nonchimeric) or chimeric embryos were performedin both cryostat or paraffin sections following the immunofluores-cence protocol described below.

For double antigen colocalization, cryostat (QCPN/WT1;QCPN/RALDH2) or paraffin sections (QCPN/cytokeratin; QCPN/RALDH2; QCPN/caldesmon; QH1/caldesmon; vimentin/RALDH2;vimentin/cytokeratin: MF20/cytokeratin) of the proepicardial chi-meras were used. Cryostat sections were obtained and processed asdescribed for WT1. The first labeling step was an overnightincubation at room temperature of the sections in the monoclonalantibodies (QCPN undiluted, QH1 diluted 1:200 in GBT, anti-vimentin 1:75 in GBT, and MF20 1:20 in PBS; all obtained from theDevelopmental Studies Hybridoma Bank). After three washes inPBS, monoclonal antibody labeling was detected by using sheep-anti-mouse biotinylated antibody (Amersham RPN-1001; 1:100 inPBS, 2 h, room temperature), washed again (3 � 10 min, PBS), andincubated in extravidin–FITC (Amersham RPN-1232). After wash-ing in PBS (3 � 10 min), the sections were then incubated overnightat room temperature in polyclonal antibodies (anti-WT1 1:50 inPBS, anti-cytokeratin 1:100 in PBS, anti-caldesmon 1:100 in PBS,and anti RALDH2 1:5000 in PBS). Then, 3 � 10 min PBS washeswere followed by an incubation (2 h, room temperature) with agoat-anti-rabbit antibody (AlexaFluor-568, Molecular Probes). Afterextensive washes, the sections were mounted in 1:1 PBS/glycerolsolution and analyzed by using a BioRad MRC 1024 laser scanningconfocal microscope. In some of the figures, the caldesmon expres-sion (goat-anti-rabbit Alexa Red 568 detected with the 605-nmwave-length filter on the confocal microscope) is shown in purpleto improve signal/noise ratio.

Proepicardial Blocks and Proepicardial Ablations

Proepicardial block. To block attachment of the proepicar-dium to the heart, the eggshell membrane technique (Männer,

1993) was used. Chick embryos were incubated until stagesH/H16–H/H17. Openings in the vitelline and chorionic mem-branes were created by using tungsten needles. A small piece ofeggshell membrane was inserted between the sinoatrial sulcus andthe caudal vitellin veins, thus temporarily preventing the heart tobe covered by epicardium.

Proepicardial ablation. The ablation was performed by excis-ing the proepicardium of chick embryos at stages H/H16–H/H17using fine dissecting tungsten needles.

After the surgical procedure, blocked or ablated embryos werereincubated for 96–120 h (resulting in H/H25–H/H27 embryos),132–144 h (resulting in H/H28–H/H29 embryos), or 180–216 h(resulting in H/H32–H/H35 embryos). The embryos were fixed inmodified Amsterdam’s fixative, embedded in Paraplast Plus, sec-tioned on a microtome (5 �m), and stained with hematoxylin–eosin or processed for immunohistochemical staining. In addition,some of the blocked or ablated embryos used for WT1 character-ization were frozen in liquid nitrogen-cooled isopentane, sectionedin a cryostat, and immunostained for WT1.

RESULTS

Spatiotemporal Expression Patterns of WT1,Cytokeratin, and RALDH2 (Fig. 1)

Wilms’ Tumor Expression

In the stages that precede epicardial development, WT1expression is restricted to the proepicardium and to thecoelomic epithelium that constitutes the developing pari-etal pericardium (data not shown; see also Carmona et al.,2001). At the end of stage H/H17, the proepicardium at-taches to the heart and spreads over the myocardium,forming the epicardial layer which strongly expresses WT1.Around stage H/H18, the first mesenchymal cells can beseen in the subepicardial space of the areas covered byepicardium. These cells first appear in the AV groove and atthe end of stage H/H19 (Fig. 1A) in the CV groove andventricular region. Virtually all early subepicardial mesen-chymal cells (EPDCs) are WT1-positive. As epicardial de-velopment proceeds, WT1 expression is found in the epi-cardium that covers the cardiac chambers. In stagesH/H20–H/H26, the number of mesenchymal cells increasesin the subepicardial space of the AV groove, CV groove, andthe subepicardium of the ventricles (Figs. 1D and 1G). Inthese junctional regions, a gradient in WT1 expression isevident. Within the AV subepicardium, the EPDCs closer tothe epicardium express more WT1 than those located closerto the myocardium (Fig. 1D). At stages H/H25–H/H26, thefirst WT1-positive cells can be seen invading the ventricu-lar myocardium. No WT1-positive mesenchymal cells canbe detected in the atrium (Fig. 1D). WT1 continues to beexpressed in the epicardium and subepicardial cells duringlater stages of development (H/H27–H/H33). At thesestages, the number of WT1-positive intramyocardial cellswithin the ventricular myocardium increases significantly.Most of the WT1-positive cells are found in the outer(compact) myocardial layer, whereas only a few WT1-



positive cells can be found in the inner trabecular layers ofthe myocardium (Fig. 1J).

RALDH2 and CK Expression

The expression patterns obtained with the anti-CK andthe anti-RALDH2 antibodies are similar. During earlystages of epicardial development (H/H17–H/H19), the pro-epicardium and the developing epicardial layer stronglyexpress RALDH2 and CK (Figs. 1B and 1C). The subepicar-dial mesenchyme of the ventricles is scarce and seems to beorganized as a single cell layer. Subepicardial mesenchymalcells are absent from the atrium. During stages H/H20–H/H26, both RALDH2 and CK are expressed in the subepicar-dial mesenchymal population. RALDH2 and CK display thesame characteristic gradient of expression described forWT1. The difference in expression between the outer andinner subepicardial layers is higher for RALDH2 than forCK (Figs. 1E and 1F). At the end of stage H/H26, a numberof RALDH2/CK-positive cells can be seen in the compactmyocardial layer of the ventricles (Fig. 1H and 1I), but not inthe atrial or outflow tract myocardium. At stages H/H27–H/H33, the RALDH2 and CK expression in the epicardiumis strong, but in most subepicardial cells and mesenchymalcells infiltrating the myocardial layers, the level of expres-sion is weak or undetectable (Figs. 1K and 1L).

Quail-Chick Chimeras (Fig. 2)

Quail-chick chimeras were prepared by transplantingproepicardium from quail origin (Figs. 2A and 2A�) into achick host by using a piece of eggshell membrane (Fig. 2B).This procedure not only allows the quail epicardium topopulate the chick heart, but it also prevents the attach-ment of the chick proepicardium to the heart (Fig. 2B).

The epicardium formed on the myocardial surface of thechick hosts in the quail-chick chimeras was QCPN-positive, indicating a donor (quail) origin (Figs. 2B� and 2B�).The subepicardial mesenchymal cells were also QCPN-positive (Figs. 2B�, 2B�, and 2C�). A subset of these QCPN-positive cells contribute to the formation of subepicardial

coronary vessels (Fig. 2C�). Around stages H/H25–H/H26,QCPN-positive cells start to migrate into the ventricularwall. From stages H/H28–H/H29 onward, QCPN-positivecells are found throughout the ventricular wall, some quailcells even being observed in the endocardial lining of thedeveloping ventricular trabeculae (Fig. 2B�). In the AVjunction, QCPN-positive cells start to invade the myocar-dium at stages H/H29–H/H30, at which stage sporadicQCPN-positive cells are also observed in the endocardialcushions. A considerable number of QCPN-positive cellsare detected in the cushions around H/H31–H/H32 (Fig.2C�). It is important to note that QCPN-expressing immi-grating cells were never observed in the atrial myocardium(Fig. 2B�) or the outflow tract (data not shown).

Molecular Phenotype of the Epicardial-DerivedCells in the Quail-Chick Chimera (Figs. 3 and 4)

QCPN-positive (i.e., quail-derived) epicardial and subepi-cardial cells show the same molecular phenotype as de-scribed for the normal epicardium and its derivatives (seeFig. 1). Thus, CK (Figs. 3A and 3B), RALDH2 (Figs. 3C and3D), and WT1 (Figs. 3E and 3F) are expressed in theepicardium and virtually all subepicardial cells. CK andRALDH2 are also expressed in the cells invading theamyocardium, but this expression is lost progressively asthe cells continue to migrate. The WT1 expression can stillbe observed in QCPN-positive cells that have migrated deepinto the compact myocardium and the trabeculae (Figs. 3Eand 3F). However, WT1/QCPN-positive cells are not foundin the endocardial cushions (not shown).

A subset of epicardial and subepicardial QCPN-positivecells in the chimeras also express caldesmon. These cellsare specifically found in the AV and CV grooves related tothe developing media of the subepicardial coronary arteries.(Figs. 4A and 4B). Interestingly, some of the QCPN-positivecells found in the AV endocardial cushions express caldes-mon as well (Figs. 4D–4F). The majority of the caldesmon-expressing cells are located in the subendothelial layer ofthe cushions, although some weakly caldesmon-expressingQCPN-positive cells are found in the core of the cushions.

FIG. 3. Proepicardial cell tracing in quail-to-chick chimeras. (A) The AV groove of a stage H/H32 chimera. EPDCs are labeled with thequail-specific nuclear marker QCPN (green dots) and an anti-cytokeratin antibody (red). Bar, 65 �m. (B) A detail of a cluster of EPDCs (arrowin A). Some EPDCs in the myocardium retain their cytokeratin immunoreactivity in the cytoplasm (arrowheads). Bar, 33 �m. In (C), theRALDH2 expression (red) in donor-derived tissues (green nuclei) of a stage H/H29 chimera is shown. The epicardial epithelium, derivedfrom the donor, is completely RALDH2-positive. In some areas of the ventricle, EPDCs (QCPN-positive nuclei) that express RALDH2(cytoplasmic expression) can be seen within the myocardial layers (arrowheads). Bar, 65 �m. (D) The AV sulcus of a stage H/H29 chimera.The RALDH2 signal is partially lost and the majority of the infiltrated EPDCs are RALDH2-negative (arrowheads). However, some cells arestill faintly stained by the RALDH2 antibody (arrows). Bar, 65 �m. (E, F) Details of WT1 expression in a stage H/H32 chimera. Epicardialand subepicardial populations coexpress QCPN (green) and WT1 (red). Both are nuclear markers, so nuclei of the EPDCs appear as yellowdots in the epicardium, subepicardium, and myocardium. Part of the invasive EPDC population invades the myocardium as rows ofQCPN/WT1-positive cells (arrow in F). The pericardial tissue, which is strongly expressing WT1, is QCPN-negative and not derived fromquail donor tissue, (red dots). Bars, 33 �m. AM, atrial myocardium; AVM, atrioventricular myocardium; EN, endocardium; EP, epicardium;LI, liver; MYO, myocardium; PE, pericardium; SE, subepicardium; VM, ventricular myocardium.



FIG. 4. EPDCs differentiate into a smooth muscle cell(-like) phenotype. In (A) and (B), the formation of coronary vessels from the quailproepicardial explant is shown in a stage H/H36 chimera. In (A), QCPN-labeled EPDCs (green nuclei) are found in the subepicardium andmyocardium (arrowheads), and in endothelium (double arrowhead) and media of the coronary vessels. The smooth muscle cells express thesmooth muscle antigen caldesmon (red staining, arrows). Bar, 33 �m. In (B), another detail from the same chimera is presented.Quail-derived vascular cells have been stained with the QH1 antibody (green), while smooth muscle cells appear in red (caldesmonexpression, arrow). Note the presence of vessels of small caliber that have not developed a muscular wall (QH1 red labeling, arrowheads).Bar, 33 �m. The image in (C) shows smooth muscle cell �-actin expression in the AV cushions of an H/H29 chimera. Arrows indicate



The caldesmon expression pattern coincides with the ex-pression pattern of smooth muscle alpha-actin (Fig. 4C).

Proepicardial Block and Proepicardial Ablation(Figs. 5 and 6)

Proepicardial block experiments result in different phe-notypes as compared with those obtained by a direct pro-epicardial ablation.

The proepicardial block experiments generally do notprevent the formation of the epicardium, but rather cause ageneral delay in the process. This is a result of the fact thatthe proepicardium will often grow around the blockingeggshell membrane to reach the myocardial surface of theheart. Proepicardial ablation prevents epicardial develop-ment for 12–24 h; after that period, a compensatory me-sothelial proliferation often develops in the original proepi-cardial location and initiates the development of anepicardial-like tissue.

The cardiac phenotype observed after blocking experi-ments ranges from almost normal to severely abnormal(Figs. 5A and 5B). The abnormalities, most of them typicallyobserved from stages H/H27–H/H28 onward, include local-ized or severe reduction of the thickness of the compactventricular myocardium (Figs. 5A–5D), small gaps in myo-cardial continuity (Fig. 5E), and occasionally pouch-likemyocardial interruptions (Fig. 5F) where the ventricularlumen is contiguous with the subepicardial space/subepicardial vascular structures (Fig. 5F). These small andlarge disruptions in the myocardial continuity sometimesresult in subepicardial bleeding (hemorrhage) (Figs. 5D–5F).The malformations also include abnormalities of the coro-nary system (see Discussion). Another characteristic fea-ture of the proepicardial blocked embryos was valvuloseptaldysmorphogenesis. This was particularly prominent in theolder embryos examined (H/H35). The AV cushions of theseexperimental embryos were much larger than those of thecontrols (Fig. 5B, compare with Fig. 5A).

Compared with the blocked embryos, the cardiac abnor-malities seen in proepicardial ablated embryos (Figs. 6B and6C) are more severe and are present at earlier stages ofdevelopment. These abnormalities frequently include widemyocardial pouch-like discontinuities, similar to those thatwere occasionally seen in the blocked specimens (see above)and which form endocardial–subepicardial connections(Fig. 6C), and occasionally malformations of the ventricularseptum (Fig. 6B). In addition, the formation of AV endocar-

dial cushion tissue is also abnormal. Compared with nor-mals (Fig. 6A), the AV cushions in ablated embryos extenddeeper into the ventricular segments (Fig. 6B). It is impor-tant to note that most of the ablated embryos died aroundstage H/H31, i.e., the stage that the coronary subepicardialplexus connects to the circulation. A few ablated embryos,however, survived up to stage H/H34–H/H35.

Molecular Characterization of the Phenotype ofProepicardial Ablated (Fig. 7) and BlockedEmbryos (Fig. 8)

As mentioned above, proepicardial ablation is followedby the generation of a compensatory (pro)epicardial-liketissue (Fig. 7A), which eventually migrates over the myo-cardial surface of the heart. This migration is delayed (i.e.,the epicardial-like tissue arrives later) and incomplete (i.e.,some areas do not become covered at all). In addition, in thelimited subepicardial space that is formed in those areasthat are covered, only very few subepicardial cells can befound. The myocardial wall in the ablated embryos that isnot covered by the epicardial-like tissue is considerablythinner than the myocardial wall in the areas covered withthis tissue (Fig. 7B). The walls in these specimens fre-quently show myocardial indentations and/or interruptions(Fig. 7C).

At stage H/H26–H/H27, the pouch-like structures thatbulge out from the ventricular lumen protrude into thesubepicardium for 100–200 �m. The pouches are lined withvimentin-positive endothelial cells. These cells, however,do not express epicardial markers such as CK and RALDH2.These epicardial markers, however, are expressed in thesubepicardial tissues surrounding the pouches (Figs. 7D and7E). In the embryos that survived up to stage H/H34–H/H35, the subepicardial tissues adjacent to the endotheliallining of the pouches undergo compaction to form a multi-layered wall. Within this wall, the cell layers closest to theendothelium express caldesmon (Fig. 7F).

The AV cushions in experimental embryos are mal-formed (see Figs. 5 and 6). The expression of SMA in theendocardial AV cushions of control and proepicardial ab-lated specimen at H/H31 is shown in Figs. 7G and 7H. Nosignificant difference in the level of SMA expression wasobserved.

In the blocking experiments, scarce and discontinuousepicardium was found. The flattened epicardium covered athin subepicardial space in which only a few CK-positive

subendocardial areas of strong expression. Bar, 65 �m. (D–F) The presence of donor-derived EPDCs in the AV cushions of the same chimeraillustrated in (C). EPDCs appear as green dots (QCPN nuclear staining); some of these cells coexpress the smooth muscle lineage-relatedmarker caldesmon (purple). (D) Some of these EPDCs are caldesmon-positive (arrows) while others, located in the core of the cushions, arecaldesmon-negative (arrowheads). Bar, 65 �m. (E) The tendency of the EPDCs to accumulate in the subendothelial layers of the AV cushions(arrowheads), where their caldesmon expression seems to be stronger. Bar, 22 �m. In (F), EPDCs migrating into the cushions from themyocardium are shown. Arrowheads point to caldesmon-negative EPDCs, while the arrow indicates the presence of caldesmon expressingEPDCs. Bar, 33 �m. B, blood; CM, cushion mesenchyme; EN, endocardium; EP, epicardium; MYO, myocardium; SE, subepicardium.



and RALDH2-positive mesenchymal cells were seen. Thisabnormal epicardium/subepicardium usually covered a“thinned” myocardial area (Figs. 8B–8D). The number ofintramyocardial RALDH2- and WT1-positive cells in theexperimental embryos was also reduced when comparedwith controls (Figs. 8A–8D).

DISCUSSION

There is a growing body of evidence that the epicardiumis of great importance for general cardiac development. Asalready indicated, experimental manipulation of epicardialgrowth (i.e., microsurgery) as well as genetic perturbation(i.e., knock-out technology) of molecules involved in epi-cardial cell behavior (e.g., migration, differentiation) havebeen reported to disrupt normal development of the heart,resulting in coronary and myocardial dysmorphogenesis(Dyson et al., 1995; Kwee et al., 1995; Yang et al., 1995;Moore et al., 1999; Wu et al., 1999; Gittenberger-de Grootet al., 2000; Tevosian et al., 2000) and general cardiacdefects involving endocardial cushion tissues and septalstructures (Gittenberger-de Groot et al., 2000; Pérez-Pomares et al., 2001). Most of the knock-out mice die inutero around ED12–ED13 due to a severe pericardial bleed-ing and/or cardiac failure caused by thinning of the myo-cardial wall. Thus, interference with epicardial develop-ment leads to abnormalities in the formation of superficialand internal cardiac structures.

The epicardial epithelium only covers the outer surface ofthe heart. Initially, it was difficult to understand howdefective epicardial development could generate such awide spectrum of cardiac malformations. Recent studies,however, have shown that the epicardium generates apopulation of epicardially derived cells (EPDCs) that in-vades the heart tissues in a well-defined spatiotemporalmanner, reaching the subepicardial, myocardial and valvu-loseptal tissues (Pérez-Pomares et al., 1997, 1998, 2001;Dettman et al., 1998; Gittenberger-de Groot et al., 1998;Männer, 1999). It has been established that EPDCs differ-entiate into coronary smooth muscle cells, cardiac fibro-blasts, and coronary endothelium (Mikawa and Fischman,1992; Pérez-Pomares et al., 1997, 1998; Dettman et al.,1998; Gittenberger-de Groot et al., 1998; Männer, 1999).Very little information was available about the invasivepopulation of EPDCs in terms of their timing of migration,range of distribution, and expression of epicardial markers.In this study, we show the normal distribution of EPDCscombining a cell-tracing system (quail-to-chick chimeras)with immunohistochemical molecular characterization us-ing epicardial and differentiation markers.

EPDCs Invasion of Developing Cardiac Tissues

Our findings show that a population of EPDCs invadesthe subepicardial space as well as the ventricular myocar-dium, the AV myocardium, and the AV endocardial cush-

ions approximately between the stages H/H28–H/H29. Theepicardial epithelium expresses the transcription factorWT1 and the RA-synthesizing enzyme RALDH2. The sub-epicardial mesenchymal EPDCs also express both markers.However, as soon as they migrate into the myocardium, aprocess that starts around H/H26, RALDH2 expression isreduced significantly. The level of expression of WT1 in theEPDCs then gradually decreases as the cells invade deeperinto the myocardial wall. Thus, the EPDCs found in the AVendocardial cushions, many of which express caldesmon,do not express RALDH2 or WT1. In addition, the terminallydifferentiated coronary endothelium of EPDC origin (sub-epicardial as well as intramyocardial), which is character-ized by the expression of the endothelial specific markerQH1 (in quail), generally does not express WT1 orRALDH2. Therefore, the loss of expression of WT1 andRALDH2 is related to the organization of EPDCs in recog-nizable structures and the upregulation of molecular mark-ers of cell lineage-specific differentiation. When we delayepicardial development, a decrease in the overall number ofEPDCs, and hence also the number of WT1/RALDH2-positive cells, is observed. This reduction in the number ofcells correlates with malformations of the structures inwhich they are normally found. These malformations in-clude the “thin myocardial syndrome,” abnormal coronaryvessels, and dysmorphic AV cushions. This spectrum ofmalformations closely resembles that seen in mice defi-cient for genes involved in epicardial development (Dysonet al., 1995; Kwee et al., 1995; Yang et al., 1995; Moore etal., 1999; Wu et al., 1999; Tevosian et al., 2000).

EPDCs–Myocardial Interaction: A Possible Role forWT1 and RALDH2 in the Formation of theCompact Layer of the Myocardial Wall

The invasion of the ventricular myocardium by EPDCs(stages H/H25–H/H26) coincides with the onset of themyocardial thickening that leads to the generation of thecompact layer of the ventricular wall. As mentioned above,at this point, the EPDCs are expressing WT1 and RALDH2.There is reason to believe that this expression is related toa specific role for EPDCs in myocardial development. WT1encodes a zinc-finger transcription factor (Call et al., 1990;Gessler et al., 1990) which is involved in a large number ofnormal functions for cell proliferation, differentiation, andsurvival (Armstrong et al., 1993). However, the precisefunction of WT1 in normal vertebrate development is notyet established. It is interesting to note that, in the WT1k.o. mouse, all the organs that normally harbor an invasivepopulation of WT1-positive cells (i.e., kidneys, gonads,adrenals, spleen, cardiac ventricle) either are absent or areabnormally developed (Moore et al., 1998, 1999). It isimportant to emphasize that the ventricular phenotype ofthe WT1 k.o. mouse is virtually identical to that of theVCAM1 k.o. and alpha-4-integrin k.o. mouse, mice thatlack a normal epicardium. Embryos of these three k.o.models die due to cardiac failure likely caused by ventric-



FIG. 5. Proepicardial blocks. (A, C) The normal (control) phenotype of a stage H/H35 chick embryo. (B, D–F) An experimental chick embryo at stageH/H35. In (A) and (B), transverse sections at the level of the AV valves are shown Bar, 700 �m. (B) Abnormal ventricular morphology in the ablatedembryo [cf. the size of the left AV (asterisk) and the right AV valves (arrow) in A and B]. The AV valves in the experimental embryo are dysplastic andthe interventricular septum (IVS) looks underdeveloped and poorly compacted. The arrowheads indicate the presence of abnormal trabeculation (cf. A)Bar, 700 �m. In (C), a detail of the compact myocardial layer of the right ventricle in a control embryo is shown. The thickness of the compactmyocardium is indicated by the double-headed arrow. Bar, 25 �m. In (D), the myocardium of an experimental embryo is shown at the samemagnification. The double-headed arrows indicate the reduced thickness of the compact myocardium, around one-fifth of the one in the control. Bar,25 �m. In (E), a higher magnification of the myocardium of a proepicardially blocked embryo is presented. The arrow points to a myocardialdiscontinuity in the thin compact layer. Blood accumulates in the subepicardial area as indicated by arrowheads. Bar, 50 �m. In (F), a pouch-likestructure is shown. The lumen of the pouch is connected with the endocardium (asterisk) and its wall is formed by numerous cells that present a smoothmuscle-like phenotype (arrowheads). Accumulation of blood in the subepicardium is indicated by an arrow. Developing subepicardial vessels are oftenfound in relation with the pouches (double arrowhead). Bar, 100 �m. EP, epicardium; IVS, interventricular septum; LA, left atrium; LV, left ventricle;MYO, myocardium; PE, pericardium; RV, right ventricle; RA, right atrium; SE, subepicardium.



ular myocardial hypoplasia around ED13 (Kwee et al., 1995;Yang et al., 1995; Moore et al., 1999). At this developmentalstage, the coronary system is not yet functional (i.e., notconnected to the general systemic circulation). This impliesthat myocardial hypoplasia can not be a direct result ofreduced blood supply from defective coronary arteries.Taken together, these observations strongly suggest thatthe observed myocardial phenotype is caused by abnormalepicardial development. As discussed above, RALDH2 andWT1 are expressed in mouse and avian epicardial andepicardially derived cells. Reducing/abolishing the expres-sion of WT1 within the primitive myocardial wall, either bygenetic means (WT1 k.o.) or by reducing the number of cellsthat express WT1 in the wall (i.e., experimental perturba-tion of epicardial development), leads to basically the samecardiac phenotype. The most characteristic feature in thisrespect is the thin myocardial wall. Therefore, we concludethat WT1 expressed by the intramyocardial EPDCs plays acrucial role in the formation of the thick compact layer ofthe ventricular myocardial wall. In this respect, it is inter-esting to note that during normal development the myocar-dial layer of the atrium, that remains thin through devel-opment, does not get invaded by EPDCs.

WT1-positive epicardial and epicardially derived cellsalso express the RA-synthesizing enzyme RALDH2. It isinteresting to note that in humans the WT1 transcriptionfactor downregulates the RAR-alpha gene (Goodyer et al.,

1995) and that the phenotype of the WT1 k.o. mouse closelyresembles that of the RA-depleted (Wilson et al., 1953) andRXR-alpha k.o. mice (Sucov et al., 1994). Depletion of RAfrom the diet is known to severely disturb heart develop-ment, causing hypoplasia of the ventricles (“thin myocar-dium”), a phenotype that is also seen in RXR-alpha-defective mice. Furthermore, ventricular expression of themyosin light chain 2a (MLC-2a) gene is abnormally persis-tent in RXR-alpha-deficient embryos. This fact can beconsidered both as retardation in ventricular maturation oras misspecification of the ventricle into an atrial fate(Dyson et al., 1995). However, other authors have suggestedthat, in the absence of proper retinoid signaling, cells mightprecociously reach a state of terminal differentiation (Kast-ner et al., 1994, 1997).

Proepicardial block results in a decreased amount ofWT1-positive, RALDH2-positive intramyocardial EPDCs,and myocardial abnormalities, including areas of thin com-pact myocardium, gaps in the myocardial layers, and dys-morphic AV valvuloseptal tissues. Myocardial defects werecommonly seen in areas that would normally be covered byepicardium during the first hours of epicardial development(i.e., in the inner curvature AV, CV, and ventricular areas).Thus, it is likely that the abnormal myocardial develop-ment is directly related to the reduction of the numbers ofintramyocardial EPDCs in experimental embryos. A pos-sible role for epicardial cells in the regulation of cardiomy-

FIG. 6. Proepicardial ablations. (A) An H&E-stained section of a stage H/H31 normal embryo with well-formed ventricles andinterventricular septum (asterisk). The arrows indicate normally developed ventricular epicardium. (B) A section of a stage H/H31 ablatedembryo at a comparable level. The experimental heart is smaller and the amount of subepicardial mesenchyme is greatly reduced (arrows).The interventricular septum is absent (asterisk) and the valvuloseptal tissue is abnormally extended into the ventricle (arrowheads). Notethat pericardium has been removed. Bars, 700 �m. In (C), a magnification of a pouch-like structure protruding into the subepicardial spaceof the right ventricle in a stage H/H35 ablated embryo is shown. The pouch consists of an endocardially lined lumen (asterisk) surroundedby densely packed layers of cells (arrowheads). Bar, 100 �m. EP, epicardium; LV, left ventricle; MYO, myocardium; PE, pericardium; RV,right ventricle; SE, subepicardium.FIG. 7. Molecular characterization of the phenotype of proepicardial ablated embryos. (A) A cytokeratin staining (red) of a stage H/H27ablated chick embryo. The region shown in this confocal plane corresponds to the original location of the proepicardium (asterisk). Thearrows indicate the accumulation of mesenchymal coelomic tissue around the area; the arrowheads point to the flattened epicardiumformed over the ventricular myocardium. Bar, 133 �m. In (B), the delayed formation of the epicardium in a stage H/H27 ablated chickembryo is shown. The myocardium has been counterstained with the myosin-specific MF20 antibody (green) and the epicardial cytokeratinexpression appears in red. Note the boundary between epicardially covered and bare myocardium (arrow). The uncovered region correspondsto an area of thin myocardium (arrowhead, compare to the epicardially covered area, double arrowhead). Bar, 66 �m. (C–E) Confocal imagesof a pouch-like structure in the ventral right ventricular wall of a stage H/H26 ablated embryo. In (C), the myocardium (MF20-positive)appears in green. RALDH2 expression, in red, appears in the epicardial epithelium (EP) and subepicardial mesenchyme. Bar, 66 �m. In (D),a double RALDH2 (red) and vimentin (green) staining of this pouch is presented. The lumen of the pouch (asterisk) is covered byvimentin-positive endothelium (arrowheads). The endocardium (double arrowheads) is also vimentin-positive. The outer (epicardial) sideof the structure is covered by a thick layer of RALDH2-positive subepicardial mesenchyme (arrows). Bar, 33 �m. (E) A double cytokeratin(red) and vimentin (green) staining of the same embryo and area shown in (C) and (D). The lumen of the pouch is covered byvimentin-positive endothelial tissue and associated mesenchyme (arrowheads). The epicardial side consists of a cytokeratin-expressing beltof tissue (arrows); note that most of the subepicardial tissue coexpresses cytokeratin and vimentin (labeled in yellow). Some of thesedouble-stained mesenchymal cells immigrate into the depths of the pouch (double arrowheads). Bar, 33 �m. In (F), a section of a pouch inan older ablated chick embryo is shown (H/H34). The wall of the pouch is now more compact and defined; its external side is indicated byarrows. Layers of caldesmon-positive (red) cells (arrowheads) are seen in close vicinity of the inner endocardial lining of the pouch (asterisk).Bar, 66 �m. The sections in (G) and (H) show the expression of SMA (arrows) in the endocardial AV cushion of a control (G) andexperimental (H) embryo at stage H/H31. Bars, 90 �m. A, atrium; CM, cushion mesenchyme; EP, epicardium; Li, liver; MYO, myocardium;SE, subepicardium; SV, sinus venosus; V, ventricle; VM, ventricular myocardium.



ocyte differentiation was also indicated in studies whereepicardial cells were cocultured with isolated cardiomyo-cytes (Eid et al., 1992). In these studies, it was shown thatwhen cultured in the presence of epicardial cells, isolatedventricular myocytes express more ventricular myosinheavy chain and form myofibrillar arrays that are moreorganized than those in cells that were not cultured in thepresence of epicardial cells. It is important to realize thatthese observations were made years before it became clearthat the epicardium gives rise to a subpopulation of highlyinvasive cells that populate several structures of the devel-oping heart, including the ventricular myocardial wall(Dettman et al., 1998; Gittenberger-de Groot et al., 1998;Männer, 1999).

EPDCs, Coronary Vasculature, and Pouch-LikeStructures

Proepicardial block or ablation does not prevent themyocardium from becoming covered by “epicardial-like”tissue. In the proepicardial block, this tissue might derivefrom the original proepicardium as a result of late proepi-cardial attachment to myocardium (usually heterotopicresulting from the experimental manipulations). This tis-sue has been described as a secondary sinu-ventricularmesocardium by Männer (1993). In the ablation experi-ments, the epicardial-like tissue likely derives from coelo-mic epithelium in the vicinity of the original proepicar-dium. All the mesothelium of the region is highlyproliferative as shown by BrdU incorporation (data notshown). Removal of the main mass of proepicardial tissuedoes not seem to prevent the growth of a new proepicardial-like structure. It is important to clarify that the tissue inboth the normal proepicardia and compensatory structuresfound after ablation are of mesothelial nature, i.e., deriva-tives of the coelomic epithelium and not of the sinusvenosus myocardium or of the hepatocytes of the liverprimordium.

The “epicardial-like” tissue in proepicardial block andablation locally produces “EPDCs” through EMT, albeitthat the overall number of these cells in the subepicardiumis considerably lower than that seen in normal hearts. Inaddition, in our experimental embryos, coronary develop-ment was found to be perturbed. These abnormalitiesincluded widening of coronaries and ectopic location oflarger vessels. We did not perform a detailed study ofcoronary development in normal and epicardially perturbedhearts in the context of this paper. However, this aspect isthe focus of a forthcoming study.

The alterations of coronary vessel formation was accom-panied by the formation of pouch-like structures that ap-pear to be ectopic endocardial sinusoids. Our data suggestthat direct contact between endocardium and subepicardialtissues cause massive differentiation of EPDCs into thesmooth muscle cell lineage, which leads to the formation ofa “media-like” cell layer. This is in accordance with therole that the endothelium plays in recruitment and differ-

entiation of perivascular/smooth muscle cells (Hungerfordand Little, 1999), a process that is apparently regulated byPDGF-BB secretion (Carmeliet, 2001).

The pouch-like structures resemble the conotruncal ab-normalities found in the connexin43 k.o. mouse (Reaume etal., 1995; reviewed in Lo, 1999). Interestingly, recent obser-vations suggest that this particular phenotype might (par-tially) be caused by an alteration in epicardial migration inthe conoventricular region of the heart (Li et al., 2002). Inthis respect, it is important to note that there is a growingbody of evidence that outflow tract epicardium might bepartly derived from coelomic epithelium at the anteriorpole of the heart (Pérez-Pomares et al., 2001).

EPDCs in the Endocardial Cushions. A Link to theCoronary Smooth Muscle Cell Lineage

EPDCs in the AV endocardial cushions express thesmooth muscle cell-related markers caldesmon and alphasmooth-muscle actin (�SMA). Both markers are also de-tected in EPDCs from the subepicardium, where EPDCscontribute to the formation of the media of the developingcoronary arteries. Caldesmon expression is not abundant inthe myocardial layers, though it can be found in somescattered cells in close relation to the developing coronaryendothelium. It appears that the expression of caldesmon inEPDCs is related to the proximity of these cells to theendothelium. Although the endocardial cushions areknown to lack smooth muscle cells, a population of valvu-lar interstitial cells, which present an intermediate pheno-type between fibroblasts and vascular muscle cells, hasbeen described in mammals (Filip et al., 1986). The embry-onic origin and role of these cells, which are coupled andinnervated by motor nerve endings, are not clear. Here, wepresent data that at least a subset of these progenitors ofthese valvular interstitial cells could be EPDCs that haveinvaded the endocardial cushions. These valvular EPDCsseem to have (at least partially) gone through the samedifferentiation events as the EPDCs that form the coronarysmooth muscle cells (Mikawa and Gourdie, 1996; Lander-holm et al., 1999; Vrancken Peeters et al., 1999; Lu et al.,2001). Given the abnormalities observed in the experimen-tal embryos, one can speculate that valvular EPDCs form animportant role in the development of these structures.However, when we compared smooth muscle cell markerexpression in valvuloseptal tissue of control and experi-mental specimens (Figs. 7G and 7H), we did not find asignificant difference in expression in comparable regions.In this respect, it is noteworthy that it is well documentedthat SMA is also expressed in cushion mesenchymal cellsafter undergoing endocardial-to-mesenchymal transforma-tion (Nakajima et al., 1997, 1998; Ramsdell and Markwald,1997). Hence, it is likely that the expression of smoothmuscle cell markers in cushion mesenchymal cells is aresult of both differentiation events, and that the SMA/caldesmon-positive cells that we detect immunohisto-chemically in sections can be endocardially as well as



epicardially derived. The fact that we did not observe adifference in the expression of smooth muscle cell markersin the endocardial cushion tissues of control vs proepicar-dial ablated embryos leads to a number of hypotheticalmechanisms. First, the migratory behavior of EPDCs inproepicardial ablations (hence tissue derived from the com-pensatory tissue) could be altered compared with that ofEPDCs (derived from the proepicardium) in control hearts.

In this scenario, it is possible that subsets of EPDCs doarrive in the cushions but not in the same time frameand/or number. It is also possible that the EPDCs in thecushions are involved in controlling endocardial-to-mesenchymal transformation similar to the proposedmechanism for the regulation of developing endotheliumby perivascular cells/pericytes through the tie2/Ang1 sys-tem (Suri et al., 1996; Maisonpierre et al., 1997). If this is

FIG. 8. EPDCs characterization in proepicardial blocked embryos. (A, C) Images of a control chick embryo at stage H/H27. (B, D) A H/H27proepicardially blocked chick embryo. The detail of the right ventricular myocardium in (A) shows WT1 expression (red nuclei) in theepicardium and EPDCs. The image shows a well-developed epicardium and an abundant subepicardium with WT1-positive cells. EPDCsin the myocardium are indicated with arrowheads. Bar, 45 �m. In (B), a picture of the same region at the same magnification in an ablatedembryo is presented. The WT1 expression (red nuclei) is normal in the pericardium (PE, positive control) as compared with the same tissuein (A). However, the amount of WT1-positive cells in the delayed epicardium is low (arrowheads, compare with A) and WT1-reactive EPDCsare almost absent from the myocardial layers. Note that the thickness of the compact myocardium of the area is at least one-half of thatof the normal myocardium (double-headed arrows in A and B). Bar, 45 �m. In (C), a confocal image of a RALDH2-stained H/H27 chickembryo is shown. The red cytoplasmic labeling is found in the epicardium (EP) as well as in some mesenchymal cells in the myocardiallayers (arrowheads). Bar, 90 �m. In (D), a detail of an H/H27-blocked embryo is presented. The RALDH2 expression in the epicardium ismore heterogeneous, with areas of very weak expression. RALDH2-positive cells are not found in the myocardium. Bar, 90 �m. BW, bodywall; EP, epicardium; MYO, myocardium; PE, pericardium; SE, subepicardium.



the case, a change in the number of EPDCs in the cushions,or the temporal window in which the EPDC invade thecushions, would not only alter the EPDC-related SMAexpression, but would at the same time also perturb theendocardially derived SMA-positive population.

CONCLUSION

We conclude that all or at least most of the subepicardialmesenchymal cells are EPDCs as suggested by their expres-sion of WT1 and RALDH2 in an early stage. These markersseem to disappear from most EPDCs as they differentiateinto other cell lineages, such as fibroblasts, vascular smoothmuscle, and endothelial cells (Dettman et al., 1998;Gittenberger-de Groot et al., 1998; Pérez-Pomares et al.,1998). The nondifferentiated cells, which continue to ex-press WT1 and RALDH2, invade the ventricular myocar-dium. It is tempting to speculate that EPDC–RALDH2-mediated synthesis of RA is an effective way to locallyrelease RA within the developing ventricular layers. Wesuggest that the developmental process that leads to ven-tricular myocardium compaction is dependent on the inva-sion by retinoid-producing EPDCs. It is plausible that thefirst wave of RA-producing EPDCs invading the ventricularmyocardium plays a role in the induction of downstreammolecules involved in myocardial maturation. One of thesedownstream targets could be IGF-2 (Vincent et al., 1996), amitogen which is involved in the thickening of the myo-cardium (Liu et al., 1996). The role played by WT1 in thedevelopment of the myocardial wall might be to maintainthe EPDCs in an undifferentiated, RA-producing state. Theinactivation of this gene would cause premature differen-tiation of the EPDCs, resulting in the downregulation of RAproduction. This hypothesis is supported by the “thinmyocardium” syndrome reported for the WT1-k.o. andRXR-alpha-k.o. mice.

EPDCs were never observed in the myocardial walls ofatria and OFT. In itself this is not surprising as the epicar-dial epithelia of atria and OFT do not undergo EMT togenerate a population of subepicardial EPDCs (Pérez-Pomares et al., 1997). In line with the above hypothesisabout the role of EPDCs in myocardial development, themyocardial walls of the atria and OFT do not thicken overtime but instead remain thin. It is possible that the absenceof EPDCs in atria and OFT is a result of differences inintrinsic properties between the atrial/OFT myocardiumand the ventricular myocardium. It is also possible, but lesslikely, that these phenomena are resulting from intrinsicdifferences in the epicardial epithelium covering thesesegments.

The presence of numerous EPDCs in the AV valves,combined with the observation that proepicardial block/ablations impair AV valve development, suggest an impor-tant role for EPDCs in valvuloseptal morphogenesis. Given

the fact that the only other tissue types in the AV valves areendocardial and endocardially derived mesenchyme, onehas to assume that proper interaction between EPDCs andthese cells is imperative for correct valve formation. Thedevelopmental mechanisms involved are yet to be eluci-dated but could involve regulation of endocardial EMT,regulation of cell division and cell death (cf. Lakkis andEpstein, 1998), regulation of cell migration, and regulationof extracellular matrix production.

In summary, we suggest that EPDCs not only materiallycontribute to the different tissues of the heart (i.e., coronaryendothelium/smooth muscle, interstitial fibroblasts, andcushion mesenchyme), but that they also have a criticalfunction in the differentiation of the ventricular myocar-dium and the AV valves.

ACKNOWLEDGMENTS

We thank Gerardo Atencia and Marı́a José Ávila for theirvaluable technical help, and Tanya Rittmann for creative assis-tance in preparing illustrations. We also thank Drs. Peter McCaf-fery and Ursula Dräger for their kind gift of the RALDH2 antibodyas well as Dr. R. G. Gourdie for the caldesmon antibody. The MF20,vimentin (Amf-17b), QCPN, and QH1 monoclonal antibodies wereobtained from the Developmental Studies Hybridoma Bank main-tained by the Department of Pharmacology and Molecular Sciences(John Hopkins University School of Medicine, Baltimore, MD) andthe Department of Biological Sciences (University of Iowa, IA City,IA), under contract NO1-HD-2-3144 from the NICHD. This workwas supported by funds from the NIH (Grant NHLBI 1PO1HL52813-06) and the AHA (Grant AHA-GIA 005099U) to A.W. andJ.M.P-P. and by Grants PM98–0219 (Spanish Ministry of Health)and FEDER 1FD-0693 (European Community) to R.M-C.

The authors would like to dedicate this paper to the memoryof Szabolcs Viragh (1930 –2001), whose descriptive studies oncardiac morphogenesis, in particular on the development of theepicardium and conduction system, form an integral part of thefoundation on which contemporary cardiac developmental bi-ologists now build their more molecularly oriented projects inthe field. His enthusiasm discussing the ultrastructure andfunction of the developing heart, his sense of humor, hiskindness, and his warm voice are sadly missed by those whoworked closely with him over the years.

REFERENCES

Armstrong, J. F., Pritchard-Jones, K., Bickmore, W. A., Hastie,N. D., and Bard, J. B. (1993). The expression of the Wilms’ tumourgene, WT1, in the developing mammalian embryo. Mech. Dev.40, 85–97.

Berggren, K., Ezerman, E. B., McCaffery, P., and Forehand, C. J.(2001). Expression and regulation of the retinoic acid syntheticenzyme RALDH-2 in the embryonic chicken wing. Dev. Dyn.222, 1–16.

Call, K. M., Glaser, T., Ito, C. Y., Buckler, A. J., Pelletier, J., Haber,D. A., Rose, E. A., Kral, A., Yeger, H., Lewis, W. H., Jones, C., and



Housman, D. E. (1990). Isolation and characterization of a zincfinger polypeptide gene at the human chromosome 11 Wilms’tumor locus. Cell 60, 509–520.

Carmeliet, P. (2001) Cardiovascular biology. Creating unique bloodvessels. Nature 412, 868–869.

Carmona, R., González-Iriarte, M., Pérez-Pomares, J. M., andMuñoz-Chápuli, R. (2001). Localization of the Wilms’ tumourprotein WT1 in avian embryos. Cell Tissue Res. 303, 173–186.

Dettman, R. W., Denetclaw, W., Jr., Ordahl, C. P., and Bristow, J.(1998). Common origin of coronary vascular smooth muscle,perivascular fibroblasts, and intermyocardial fibroblasts in theavian heart. Dev. Biol. 193, 169–181.

Dyson, E., Sucov, H. M., Kubalak, S. W., Schmid-Schönbeim,G. W., DeLano, F. A., Evans, R. M., Ross, J., Jr., and Chien, K.(1995). Atrial-like phenotype is associated with embryonic ven-tricular failure in retinoid X receptor ��/� mice. Proc. Natl. Acad.Sci. USA 92, 7386–7390.

Eid, H., Larson, D. M., Springhorn, J. P., Attawia, M. A., Nayak,R. C., Smith, T. W., and Kelly, R. A. (1992). Role of epicardialmesothelial cells in the modification of phenotype and functionof adult rat ventricular myocytes in primary coculture. Circ. Res.71, 40–50.

Filip, D. A., Radu, A., and Simionescu, M. (1986). Interstitial cellsof the heart valves possess characteristics similar to smoothmuscle cells. Circ. Res. 59, 310–320.

Gessler, M., Poutska, A., Cavenee, W., Neve, R. L., Orkin, S. H.,and Bruns, G. A. (1990). Homozygous deletion in Wilms’ tu-mours of a zinc-finger gene identified by chromosome jumping.Nature 343, 774–778.

Gittenberger-de Groot, A. C., Vrancken Peeters, M. P., Mentink,M. M., Gourdie, R. G., and Poelmann, R. E. (1998).Epicardium-derived cells contribute a novel population to themyocardial wall and the atrioventricular cushions. Circ. Res.82, 1043–1052.

Gittenberger-de Groot, A. C., Vrancken Peeters, M. P., Bergwerff,M., Mentink, M. M., and Poelmann, R. E. (2000). Epicardialoutgrowth inhibition leads to compensatory mesothelial outflowtract collar and abnormal cardiac septation and coronary forma-tion. Circ. Res. 87, 969–971.

Goodyer, P., Dehbi, M., Torban, E., Bruening, W., and Pelletier, J.(1995). Repression of the retinoic acid receptor-alpha gene by theWilms’ tumor suppressor gene product, WT1. Oncogene 10,1125–1129.

Gruber, P. J., Kubalak, S. W., Pexieder, T., Sucov, H. M., Evans,R. M., and Chien, K. R. (1996). RXR alpha deficiency confersgenetic susceptibility for aortic sac, conotruncal, atrioventricularcushion, and ventricular muscle defects in mice. J. Clin. Invest.98, 1332–1343.

Hamburger, V., and Hamilton, H. L. (1951). A series of normalstages in the development of the chick embryo. J. Morphol. 88,49–92.

Hidai, H., Bardales, R., Goodwin, R., Quertermous, T., and Quer-termous, E. E. (1998). Cloning of capsulin, a basic helix–loop–helix factor expressed in progenitor cells of the pericardium andthe coronary arteries. Mech. Dev. 73, 33–43.

Hirakow, R. (1992). Epicardial formation in staged human embryos.Kaibogaku Zasshi 67, 616–622.

Hiruma, T., and Hirakow, R. (1989). Epicardial formation in em-bryonic chick heart. Computer aided reconstruction, scanningand transmission electron microscopic studies. Am. J. Anat. 184,129–138.

Ho, E., and Shimada, Y. (1978). Formation of the epicardiumstudied with the electron microscope. Dev. Biol. 66, 579–585.

Hungerford, J. E., and Little, C. D. (1999). Developmental biology ofthe vascular smooth muscle cell: Building a multilayered vesselwall. J. Vasc. Res. 36, 2–27.

Kastner, P., Grondona, J. M., Mark, M., Gansmuller, A., LeMeur,M., Decimo, D., Vonesch, J. L., Dolle, P., and Chambon, P. (1994).Genetic analysis of RXRalpha developmental function: Conver-gence of RXR and RAR signalling pathways in heart and eyemorphogenesis. Cell 78, 987–1003.

Kastner, P., Messaddeq, N., Mark, M., Wendling, O., Grondona,J. M., Ward, S., Ghyselinck, N., and Chambon, P. (1997). VitaminA deficiency and mutations of RXRalpha, RXRbeta and RARal-pha lead to early differentiation of embryonic ventricular cardi-omyocytes. Development 124, 4749–4758.

Komiyama, M., Ito, K., and Shimada, Y. (1987). Origin and devel-opment of the epicardium in the mouse embryo. Anat. Embryol.176, 183–189.

Kreidberg, J. A., Sariola, H., Loring, J. M., Maeda, M., Pelletier, J.,Housman, D., and Jaenisch, R. (1993). WT-1 is required for earlykidney development. Cell 74, 679–691.

Kwee, L., Baldwin, H. S., Shen, H. M., Stewart, C. L., Buck, C.,Buck, C. A., and Labow, M. A. (1995). Defective development ofthe embryonic and extraembryonic circulatory systems in vas-cular cell adhesion molecule (VCAM-1) deficient mice. Develop-ment 121, 489–503.

Lakkis, M. M., and Epstein, J. A. (1998). Neurofibromin modulationof ras activity is required for normal endocardial–mesenchymaltransformation in the developing heart. Development 125, 4359–4367.

Landerholm, T. E., Dong, X. R., Lu, J., Belaguli, N. S., Schwartz,R. J., and Majesky, M. W. (1999). A role for serum response factorin coronary smooth muscle differentiation from proepicardialcells. Development 126, 2053–2062.

Li, W. E. I., Waldo, K., Linask, K. L., Chen, T., Wessels, A.,Parmacek, M. S., Kirby, M. L., and Lo, C. W. (2002) An essentialrole for connexin43 gap junctions in mouse coronary arterydevelopment. Development 129, 2031–2042.

Little, M., Holmes, G., and Walsh, P. (1999). WT1: What has thelast decade told us? BioEssays 21, 191–202.

Liu, Q., Yan, H., Dawes, N. J., Mottino, G. A., Frank, J. S., and Zhu,H. (1996). Insulin-like growth factor II induces DNA synthesis infetal ventricular myocytes in vitro. Circ. Res. 79, 716–726.

Lo, C. W. (1999). Genes, gene knockouts, and mutations in theanalysis of gap junctions. Dev. Genet. 24, 1–4.

Lough, J, and Sugi Y. (2000). Endoderm and heart development.Dev. Dyn. 217, 327–342.

Lu, J., Landerholm, T. E., Wei, J. S., Dong, X. R., Wu, S. P., Liu, X.,Nagata, K., Inagaki, M., and Majesky, M. W. (2001). Coronarysmooth muscle differentiation from proepicardial cells requiresrhoA-mediated actin reorganization and p160 rho-kinase activ-ity. Dev. Biol. 240, 404–418.

Maisonpierre, P. C., Suri, C., Jones, P. F., Bartunkova, S.,Wiegand, S. J., Radziejewski, C., Compton, D., McClain, J.,Aldrich, T. H., Papadopoulos, N., Daly, T. J., Davis, S., Sato,T. N., and Yancopoulos, G. D. (1997). Angiopoietin-2, anatural antagonist for Tie2 that disrupts in vivo angiogenesis.Science 277, 55– 60.

Männer, J. (1993). Experimental study on the formation ofthe epicardium in chick embryos. Anat. Embryol. 187, 281–289.



Männer, J. (1999). Does the subepicardial mesenchyme contributemyocardioblasts to the myocardium of the chick embryo heart?A quail-chick chimera study tracing the fate of the epicardialprimordium. Anat. Rec. 255, 212–226.

Männer, J., Pérez-Pomares, J. M., Macı́as, D., and Muñoz-Chápuli, R. (2001). The origin, formation, and developmentalsignificance of the epicardium: A review. Cell Tissues Organs169, 89 –103.

McCaffery, P., and Dräger, U. (1995). Retinoic acid synthesizingenzymes in the embryonic and adult vertebrate. Adv. Exp. Med.Biol. 372, 173–183.

Mikawa, T., and Fischman, D. A. (1992). Retroviral analysis ofcardiac morphogenesis: Discontinuous formation of coronaryvessels. Proc. Natl. Acad. Sci. USA 89, 9504–9508.

Mikawa, T., and Gourdie, R. G. (1996). Pericardial mesodermgenerates a population of coronary smooth muscle cells migrat-ing into the heart along with ingrowth of the epicardial organ.Dev. Biol. 174, 221–232.

Moore, A. W., McInnes, L., Kreidberg, J., Hastie, N. D., and Schedl,A. (1999). YAC complementation shows a requirement for WT1in the development of epicardium, adrenal gland and throughoutnephrogenesis. Development 126, 1845–1857.

Moore, A. W., Schedl, A., McInnes, L., Doyle, M., Hecksher-Sorensen, J., and Hastie, N. D. (1998). YAC transgenic analysisreveals Wilms’ tumour 1 gene activity in the proliferatingcoelomic epithelium, developing diaphragm and limb. Mech.Dev. 79, 169–184.

Morabito, C. J., Dettman, R. W., Kattan, J., Collier, J. M., andBristow, J. (2001). Positive and negative regulation of epicardial–mesenchymal transformation during avian heart development.Dev. Biol. 234, 204–215.

Moss, J. B., Xavier-Neto, J., Shapiro, M. D., Nayeem, S. M.,McCaffery, P., Dräger, U., and Rosenthal, N. (1998). Dynamicpatterns of retinoic acid synthesis and response in the developingmammalian heart. Dev. Biol. 199, 55–71.

Nakajima, Y., Mironov, V., Yamagishi, T., Nakamura, H., andMarkwald, R. R. (1997). Expression of smooth muscle alpha-actinin mesenchymal cells during formation of avian endocardialcushion tissue: A role for transforming growth factor beta3. Dev.Dyn. 209, 296–309.

Nakajima, Y., Yamagishi, T., Yoshimura, K., Nomura, M., andNakamura, H. (1998). Antisense oligodeoxynucleotide comple-mentary to smooth muscle alpha-actin inhibits endothelial-mesenchymal transformation during chick cardiogenesis. Dev.Dyn. 216, 489–498.

Pardanaud, L., Altmann, C., Kitos, P., Dieterlen-Lievre, F., andBuck, C. A. (1987). Vasculogenesis in the early quail blastodisc asstudied with a monoclonal antibody recognizing endothelialcells. Development 100, 339–349.

Pérez-Pomares, J. M., Macı́as, D., Garcı́a-Garrido, L., and Muñoz-Chápuli, R. (1997). Contribution of the primitive epicardium tothe subepicardial mesenchyme in hamster and chick embryos.Dev. Dyn. 210, 96–105.

Pérez-Pomares, J. M., Macı́as, D., Garcı́a-Garrido, L., and Muñoz-Chápuli, R. (1998). The origin of the subepicardial mesenchymein the avian embryo: An immunohistochemical and quail-chimera study. Dev. Biol. 200, 57–68.

Pérez-Pomares, J. M., Macı́as, D., Garcı́a-Garrido, L., and Muñoz-Chápuli, R. (1999). Immunohistochemical evidence of a me-sothelial contribution to the ventral wall of the aorta in avianembryos. Histochem. J. 31, 771–779.

Pérez-Pomares, J. M., Phelps, A., Muñoz-Chápuli, R., and Wes-sels, A. (2001). The contribution of the proepicardium to aviancardiovascular development. Int. J. Dev. Biol. 45(Suppl. 1),155–156.

Pinco, K. A., Liu, S., and Yang, J. T. (2001). alpha4 integrin isexpressed in a subset of cranial neural crest cells and in epicardialprogenitor cells during early mouse development. Mech. Dev.100, 99–103.

Ramsdell, A. F., and Markwald R. R. (1997). Induction of endocar-dial tissue in the avian heart is regulated, in part, by TGFbeta-3-mediated autocrine signaling. Dev. Biol. 188, 64–74.

Reaume, A. G., de Sousa, P. A., Kulkarni, S., Langille, B. L., Zhu, D.,Davies, T. C., Juneja, S. C., Kidder, G. M., and Rossant, J. (1995).Cardiac malformation in neonatal mice lacking connexin43.Science 267, 1831–1834.

Reese, D. E., and Bader D. M. (1999). Cloning and expression ofhbves, a novel and highly conserved mRNA expressed in thedeveloping and adult heart and skeletal muscle in the human.Mamm. Genome 10, 913–915.

Robb, L., Mifsud, L., Hartley, L., Biben, C., Copeland, N. G.,Gilbert, D. J., Jenkins, N. A., and Harvey, R. P.(1998) Epicardin:A novel basic helix—loop–helix transcription factor gene ex-pressed in epicardium, branchial arch myoblasts, and mesen-chyme of developing lung, gut, kidney, and gonads. Dev. Dyn.213, 105–113.

Sucov, H. M., Dyson, E., Gumeringer, C. L., Price, J., Chien, K. R.,and Evans, R. M. (1994). RXRa mutant mice establish a geneticbasis for vitamin A signaling in heart morphogenesis. Genes Dev.8, 1007–1018.

Suri, C., Jones, P. F., Patan, S., Bartunkova, S., Maisonpierre, P. C.,Davis, S., Sato, T. N., and Yancopoulos, G. D. (1996). Requisiterole of angiopoietin-1, a ligand for the TIE2 receptor, duringembryonic angiogenesis. Cell 87, 1171–1180.

Tam, P. P. L., and Schoenwolf, G. C. (1999). Cardiac fate maps:Lineage allocation, morphogenetic movement, and cell commit-ment. In “Heart Development (R. P. Harvey and N. Rosenthal,Eds.) pp. 3–18. Academic Press, San Diego.

Tevosian, S. G., Deconinck, A. E., Tanaka, M., Schinke, M.,Litovsky, S. H., Izumo, S., Fujiwara, Y., and Orkin, S. H. (2000)FOG-2, a cofactor for GATA transcription factors, is essential forheart morphogenesis and development of coronary vessels fromepicardium. Cell 101, 729–739.

Vincent, T. S., Re, G. G., Hazen-Martin, D. J., Tarnowski, B. I.,Willingham, M. C., and Garvin, A. J. (1996). All-trans-retinoicacid-induced growth suppression of blastemal Wilms’ tumor.Pediatr. Pathol. Lab. Med. 16, 777–789.

Viragh, S., and Challice, C. E. (1973). Origin and differentiation ofcardiac muscle cells in the mouse. J. Ultrastruct. Res. 42, 1–24.

Viragh, S., and Challice, C. E. (1981). The origin of the epicardiumand the embryonic myocardial circulation in the mouse. Anat.Rec. 201, 157–168.

Viragh, S., Gittenberger-de Groot, A. C., Poelmann, R. E., andKalman, F. (1993). Early development of quail heart epicardiumand associated vascular and glandular structures. Anat. Embryol.188, 381–393.

Vrancken Peeters, M.-P. F. M., Gittenberger-de Groot, A. C.,Mentink, M. M. T., and Poelmann, R. E. (1999). Smooth musclecells and fibroblasts of the coronary arteries derive fromepithelial-mesenchymal transformation of the epicardium. Anat.Embryol. 199, 367–378.



Vrancken Peeters, M.-P. F. M., Mentink, M. M. T., Poelmann, R. E.,and Gittenberger-de Groot, A. C. (1995). Cytokeratins as amarker for epicardial formation in the quail embryo. Anat.Embryol. 191, 503–508.

Wada, A. M., Reese, D. E., and Bader, D. M. (2001) Bves.prototype of a new class of cell adhesion molecules expressedduring coronary artery development. Development 128, 2085–2093.

Wessels, A., and Markwald R. R. (2000). Cardiac morphogenesisand dysmorphogenesis. I. Normal development. Methods Mol.Biol. 136, 239–259.

Wilson, J. G., Roth, C. B., and Warkany, J. (1953). An analysis of thesyndrome malformations induced by maternal vitamin A defi-ciency. Effects of restoration of vitamin A at various times duringgestation. Am. J. Anat. 92, 189–217.

Wu, H., Lee, S. H., Liu, X., and Iruela-Arispe, M. L. (1999).Inactivation of erythropoietin leads to defects in cardiac morpho-genesis. Development 126, 3597–3605.

Yang, J. T., Rayburn, H., and Hynes, R. O. (1995). Cell adhesionevents mediated by alpha 4 integrins are essential in placentaland cardiac development. Development 121, 549–560.

Xavier-Neto, J., Shapiro, M. D., Houghton, L., and Rosenthal, N.(2000). Sequential programs of retinoic acid synthesis in themyocardial and epicardial layers of the developing avian heart.Dev. Biol. 219, 129–141.

Received for publication November 16, 2001Revised February 26, 2002

Accepted April 19, 2002Published online June 7, 2002



INTRODUCTIONMATERIALS AND METHODSFIG. 1FIG. 2

RESULTSFIG. 3FIG. 4

DISCUSSIONFIG. 5FIG. 6FIG. 7FIG. 8

CONCLUSIONACKNOWLEDGMENTSREFERENCES

developmental biology 247, 307–326 (2002) … · 2007. 5. 3. · experimental studies on the...

Documents