This Review is part of a thematic series on Genetics of Cardiovascular Development, which includes the followingarticles:

Transcriptional Regulation of Vertebrate Cardiac MorphogenesisCardiac Septation: A Late Contribution of the Embryonic Primary Myocardium to Heart Morphogenesis

Early Signals in Cardiac Development

Development of Specialized Cells Within the HeartLeft/Right PatterningDevelopment Gone Awry: Congentital Heart Disease Christine E. Seidman, Guest Editor

Early Signals in Cardiac DevelopmentStéphane Zaffran, Manfred Frasch

Abstract—The heart is the first organ to form during embryogenesis and its circulatory function is critical from early onfor the viability of the mammalian embryo. Developmental abnormalities of the heart have also been widely recognizedas the underlying cause of many congenital heart malformations. Hence, the developmental mechanisms that orchestratethe formation and morphogenesis of this organ have received much attention among classical and molecularembryologists. Due to the evolutionary conservation of many of these processes, major insights have been gained fromthe studies of a number of vertebrate and invertebrate models, including mouse, chick, amphibians, zebrafish, andDrosophila. In all of these systems, the heart precursors are generated within bilateral fields in the lateral mesoderm andthen converge toward the midline to form a beating linear heart tube. The specification of heart precursors is a resultof multiple tissue and cell-cell interactions that involve temporally and spatially integrated programs of inductivesignaling events. In the present review, we focus on the molecular and developmental functions of signaling processesduring early cardiogenesis that have been defined in both vertebrate and invertebrate models. We discuss the currentknowledge on the mechanisms through which signals induce the expression of cardiogenic transcription factors and therelationships between signaling pathways and transcriptional regulators that cooperate to control cardiac induction andthe formation of a linear heart tube. (Circ Res. 2002;91:457-469.)

Key Words: cardiac induction � growth factors � cardiogenic transcription factors

Our understanding of vertebrate heart development hasbeen propelled to a large part through studies in a

number of model organisms, most notably the chick, amphib-ians, zebrafish, and mouse. In all of these systems, the heartarises from cells in the anterior lateral plate mesoderm of theearly embryo, where they are arranged in bilateral fields oneither side of the prechordal plate and rostral notochord.These fields include the precursors of both myocardial andendocardial cells, although there is apparently no commonpool of bipotential precursors for these two heart cell lineag-es.1,2 The cells of the cardiogenic mesoderm are brought tothese positions by gastrulation movements that occur in close

association with ingressing cells of the rostral endoderm. Inmammalians and birds, the bilateral fields of the cardiogenicmesoderm merge at their anterior margins to form theso-called “cardiac crescent” (Figures 1A and 2A).3 Morerecent studies have identified a second type of heart field thatis located more medially in the splanchnic mesoderm, directlyadjacent to the cardiac crescent. The cells from this bilateralfield, termed anterior heart-forming field, are fated to gener-ate anterior heart structures of the outflow tracts (see review4)(Figures 1A and 2B).

Results from experiments with mesodermal tissue explantsfrom quail and chick embryos indicate that the specification

Original received March 11, 2002; revision received August 9, 2002; accepted August 9, 2002.From Mount Sinai School of Medicine, Brookdale Department of Molecular, Cell and Developmental Biology, New York, NY. Present address for

S.Z. is Departement de Biologie Moleculaire, Institut Pasteur, CNRS URA 1947, Paris, France.Correspondence to Manfred Frasch, PhD, Mount Sinai School of Medicine, Brookdale Department of Molecular, Cell and Developmental Biology, One

Gustave L. Levy Place, New York, NY 10029. E-mail manfred.frasch@mssm.edu© 2002 American Heart Association, Inc.

Circulation Research is available at http://www.circresaha.org DOI: 10.1161/01.RES.0000034152.74523.A8

457

Review

of cardiomyocytes occurs just before or during the formationof the cardiac crescent,5,6 and differentiation markers becomeexpressed shortly thereafter.7–9 The earliest steps of assemblyof the heart tube are initiated by the convergence and fusionof the bilateral heart primordia along the midline. The cells ofthe anterior heart-forming field, after having migrated ante-riorly, are added to the anterior end of the linear tube during

and after this period. The resulting beating tubular heart iscomposed of an external myocardial and an internal endocar-dial layer and also possesses a polarity along the anteropos-terior axis, in which the prospective tissues of the aortic sac,outflow tract (conotruncus), right ventricle, left ventricle, andatria are present in an anterior to posterior order along thetube (Figure 1A) (see references4,10,11).

Figure 1. Stages of cardiogenesis. A, Mouse cardiogenesis at E7.5, E8, and E8.5. Major heart forming field is shown in black and theanterior (secondary) heart forming field (AHF) in gray. AS indicates atria, sinus venosa; CT, conotruncus; RV, right ventricle; LV, left ven-tricle; and A, atria. B, Drosophila cardiogenesis at 6, 12, and 16 hours after fertilization. Heart progenitors (HP) are shown in black andother dorsal mesodermal cell types (DM) in gray. AO indicates aorta; HT, heart.

Figure 2. Early markers for the cardiac crescent (A)and anterior heart forming field (B) in the mouse.

458 Circulation Research September 20, 2002

In all vertebrates, the tubular heart undergoes a processknown as rightward looping. The morphogenetic steps re-quired to achieve looping are guided by molecular asymme-tries that are established in and around the heart by theembryonic left/right axial pathway.12 Furthermore, in highervertebrates, septal division of the chambers and formation ofthe valves, which involves endothelial cells, are essentialsteps leading to the formation of an integrated 4-chamberedheart with separate venous (or inflow) and arterial (oroutflow) poles. During the growth process of the cardiacepithelium another distinct cell lineage, the migrating cardiacneural crest cells, populate the heart through the outflowchannel and contribute to the formation of the great vesselsand outflow septum.

The spatial and temporal orchestration of these processesimplies a complex program of genetic control. As we nowknow, this program is exerted in large part through preciselycontrolled processes of cell-cell signaling and regulators ofgene expression. Many of these processes were initiallydiscovered through embryological studies, which uncoveredinductive tissue interactions and candidates for cytokines thattransmit these signals. More recently, genetic screens in thezebrafish system have provided an alternative and highlysuccessful route toward the identification of regulatory com-ponents in cardiogenesis. Unexpectedly, genetic and molec-ular studies of mesodermal tissue development in an inver-tebrate, the fruit fly Drosophila, have also been instrumentalin the identification of specific genes and processes incardiogenesis that appear to be conserved in all higheranimals. As we will discuss, the insights gained in Drosoph-ila have opened new avenues in the study of vertebratecardiogenesis and led to significant advances in our under-standing of early heart development.

The equivalent of the heart in Drosophila, termed dorsalvessel, is a simple contractile tube that is reminiscent of thetubular heart found in early vertebrate embryos (see re-view13). Additional similarities include the direction of blood

(or hemolymph) flow from posterior to anterior and ultra-structural similarities of the cardiomyocytes. Like the verte-brate heart tube, the Drosophila dorsal vessel also has ananteroposterior polarity, such that the anterior portion, termed“aorta”, is more narrow than the posterior portion, which istermed “heart” (Figure 1B). Furthermore, the posterior por-tion includes specialized cardiomyocytes that form inflowtracts, which are called “ostia.” Segmentation, a hallmark ofinsects, also applies to the dorsal vessel, which features asegmentally repeated pattern of cell identities along the lengthof the tube. In addition to the cardiomyocytes (also called“cardioblasts”), which form 2 regular rows of cells that enclosethe lumen, the dorsal vessel is made up of a second cell type,called “pericardial cells,” which form a more irregular layeraround the outside of the dorsal vessel (Figure 3C).

Developmentally, early cardiogenesis in vertebrates andinsects also shares common features. In particular, heartspecification in both systems requires inductive tissue inter-actions across germ layers, which generate bilateral heartprimordia in the lateral mesoderm. The respective linear tubesof the heart and dorsal vessel are then formed upon fusion ofthese bilateral fields along the midline (Figure 1). As will bediscussed, these similarities extend to, and are likely a resultof, similarities in the molecular and genetic mechanisms thatcontrol early heart development in vertebrate and insectembryos.

In the present review, we summarize our current knowl-edge of the major signaling mechanisms controlling earlyheart development, with a particular focus on the specifica-tion of myocardial cells and heart patterning. We highlightparticular signaling processes using the most informativefindings in specific model organisms as examples and discussthe general insights that can be drawn from different systems.Hence, we will restrict our focus to the stages between midgastrulation and the formation of the linear heart tube becauseit appears that parallels between vertebrate and insect systemscan be best drawn during this period.

Figure 3. Induction and diversification of dorsal vessel cells in Drosophila. A and B, Left side of the early bilateral heart fields visualizedby Eve expression at 6 hours of embryonic development. Cardiac progenitors are induced in areas where Wg-dependent Sloppy Paired(Slp) and Dpp-activated Smad1 (P-Mad) overlap. C, Segmental diversification of dorsal vessel cells as visualized by the expression ofEve, Svp, and Tin at 16 hours of development.

Zaffran and Frasch Early Signals in Cardiac Development 459

Early Inductive Processes in HeartSpecification and Determination

Transplantation experiments in chick and mouse embryosdemonstrate that the cardiogenic regions contain inductiveactivities that promote cardiac differentiation.1,14 When cellsfrom regions that are normally not cardiogenic, such as theposterior primitive streak in the chick embryo, are trans-planted into the cardiogenic region, they are induced to formheart instead of blood cells.14 Moreover, coculture experi-ments suggest that the heart-inducing activity originates fromthe anterior endoderm and not from the cardiogenic meso-derm itself.15,16 Removal of early endoderm in newt and frogembryos blocks heart formation showing that, at least inamphibians, endoderm is not only sufficient but also requiredfor cardiac induction.17,18 However, genetic ablation of theendoderm in zebrafish and surgical ablation in the chick doesnot block early cardiogenesis.19,20 Although this result im-plies that differentiated endoderm is not necessary in allvertebrates, signaling from unspecified endodermal precur-sors may still be required for heart induction. From these andother observations, it can be concluded that cells of theanterior endoderm and/or their precursors have a major rolein inducing cardiogenesis. Additional cardiogenic signals arederived from the organizer,21 which act either indirectly bypatterning the early adjacent endoderm or directly by induc-ing cardiogenesis in combination with anterior endodermalsignals. Ectodermal influences on cardiac induction have alsobeen described and are thought to serve mainly in counter-acting negative influences from the neural plate.22

Of note, observations in amphibians and the chick showthat not the entire precardiac mesoderm (the “heart field,”being the mesoderm that differentiates into cardiac muscle onexplantation into tissue culture) gives rise to heart in vivo.23,24

In particular, cardiogenesis is restricted to lateral areas of theprecardiac mesoderm on either side of the prechordal plate.Medial areas fail to form heart tissue in vivo, although medialand lateral endoderm do not differ intrinsically in theirheart-inducing activity.15,25 It appears that this difference incardiac determination of lateral versus medial areas of theheart field is largely due to negative signals from the anteriorneural plate.17,26,27 In addition, ablation studies in the ze-brafish demonstrated that negative signals from the notochordplay a role in limiting heart formation to areas anterior to thenotochord.28 Altogether, it has become apparent that positiveand negative signals act successively and perhaps combina-torially to induce early cardiogenesis within a defined area ofthe anterior lateral plate mesoderm of vertebrate embryos.

Whereas the small size of the Drosophila embryo hasprevented embryological studies on heart induction, experi-ments of this type have been successfully performed in alarger dipteran, the lacewing fly.29 These cell ablation studiesshowed that induction across germ layers is also required forthe specification of the fly dorsal vessel. Unlike in verte-brates, where the heart-inducing activity is predominantly ofendodermal origin, in the fly embryo the essential inductiveactivity resides in the dorsal ectoderm on either side of thegerm band.29

Cardiogenic Transcription FactorsResponding to Inductive Signals in

Myocardial DevelopmentAlthough classical studies have mainly used morphologicalcriteria and myocardial differentiation genes as markers ofcardiac induction, more recent studies have shown that theearliest responses involve the induction of regulatory genesthat encode specific transcription factors. Genes encodingfactors of the NK homeodomain, GATA, T-box, and otherfamilies were found to exert the functions of inductive signalsduring specification, patterning, and differentiation of theheart. In addition, it appears that most, if not all, develop-mental signaling pathways are required to act in combinationwith tissue-specific transcriptional cofactors to elicit induc-tive responses. Studies of the expression and function of earlycardiogenic transcription factors have therefore significantlyadvanced our understanding of inductive signaling processesduring cardiogenesis and the major representatives will bedescribed in the following sections.

NK Homeodomain ProteinsThe prototype of cardiac NK homeodomain proteins is theproduct of the Drosophila tinman gene. tinman was the firstregulatory gene in any species known to be expressed in theprecardiac mesoderm and to function in specifying cardiacprecursors. Hence, the discovery of tinman and homologousNK-homeobox genes in vertebrates has provided a majorimpulse in the field. The activity of Drosophila tinman isabsolutely required for the formation of the dorsal vessel andfor the specification of all of its progenitor cells.30,31 Theexpression of tinman has 3 distinct phases, with the firststarting before gastrulation in the entire mesoderm, except forthe prospective hemocytes. The second phase occurs aftermesoderm migration, when tinman expression becomes re-stricted to the dorsal (ie, lateral) mesoderm,30–32 whichincludes not only the cardiogenic mesoderm but also meso-dermal cells fated to form visceral and dorsal body wallmuscles (Figure 4A). The function of tinman is required,presumably during this stage, for the specification of all 3 ofthese tissues. In the third phase, tinman expression becomesfurther restricted to precursor cells of the dorsal vessel, bothcardioblasts and pericardial cells. This expression pattern ismaintained throughout development, but includes only asubset of dorsal vessel cells in each segment (Figure3C).13,33,34 During this third expression phase, tinman isthought to function in the diversification and differentiationof dorsal vessel cells.

Unlike Drosophila, vertebrates contain several members ofthis subgroup of homeobox genes, which are named Nkx2-3through Nkx2-10 (see review35; not all members are present inevery species and Nkx2-1, 2-2, and 2-4 belong to a different,noncardiac subgroup). In mouse, chick, frog, and zebrafish,the main representative, Nkx2-5, is expressed in the lateralplate mesoderm within the heart field. In the chick and frog,the onset and pattern of early Nkx2-5 expression roughlycoincide with the timing and area of cardiac specification, asdefined by the previously discussed explant studies, thussuggesting that these genes respond to and perhaps exert earlyinductive signals.15,36 As tinman in Drosophila, vertebrate

460 Circulation Research September 20, 2002

Nkx2-5 genes continue to be expressed throughout develop-ment in the heart.37–39

Mouse embryos lacking Nkx2-5 gene activity show earlyheart defects. Although loss-of-function of Nkx2-5 does notblock heart tube formation as in Drosophila tinman mutants,cardiogenesis arrests before looping morphogenesis.40 Hu-man NKX2-5 is haploinsufficient because dominant putativeloss-of-function mutations have been identified, which causecongenital cardiac disease by disrupting cardiac morphogen-esis and, in particular, septation.41

The less severe heart phenotypes of murine Nkx2-5 mutantembryos as compared with tinman mutant fly embryos raisethe question of whether there is functional redundancy amongmembers of this Nkx2 subgroup. At least in the mouse, theother known Nkx2 members are expressed only in subareas oroutside of the Nkx2-5 expression domain.42–44 Accordingly,the heart defects in mouse Nkx2-5/Nkx2-6 double mutantembryos are only slightly more severe than those observed inNkx2-5 single mutants.45 These data appear to suggest there isonly limited redundancy among Nkx2 genes in early cardio-genic tissue. A detailed reexamination of the Nkx2-5 expres-sion profile in the chick has revealed that its expression in thisand perhaps other species does not occur in the entirecardiogenic mesoderm, but rather is largely restricted to theanterior, presumptive ventricular portions.11 Although it ispremature to generalize, most of the available data argueagainst a strict requirement of cardiac Nkx2 genes in thegeneral specification of myocardial fates in vertebrates.

The complete block of cardiogenesis on overexpression ofdominant-negative forms of Nkx2-3 and Nkx2-5 in Xenopushas been used in support of the argument of a redundant

activity of these genes in the specification of cardiacfates.46,47 However, there is increasing evidence of the im-portance of protein-protein interactions between Nkx2 pro-teins and other cardiogenic transcription factors (see next twosections), and thus it is possible that the loss of cardiogenesisin these experiments is due to the interference with thecardiogenic activity of other regulators. For the normalsituation, this interpretation would imply that different typesof transcription factors have overlapping and partially redun-dant functions in cardiac specification.

GATA Factors and CardiogenesisMembers of the cardiac GATA subfamily of factors bind tothe WGATAR motif in the promoter regions of manycardiac- and gut-specific genes (see review48). Gata4 and 2related genes, Gata5 and 6, are expressed in the precardiacmesoderm and developing heart of different vertebrate spe-cies almost simultaneously and spatially overlapping withNkx2-5, although their expression also includes the associatedendodermal layer.49–54 Functional knockout experiments withGata4 in mouse, Gata5 in zebrafish, and simultaneousknockdown experiments with all 3 family members in thechick show early heart defects, including the lack of fusion ofthe bilateral heart primordia (cardia bifida), decrease ofNkx2-5 expression, and a reduction in the number of cardio-myocytes expressing myocardial differentiation genes.55–58

With genetically chimeric embryos in the mouse, it wasshown that Gata4 function is required in the endoderm andnot within the presumptive heart cells for cardiac fusion andmyocardial differentiation.55,59 However, in zebrafish, there isstrong genetic evidence for a mesoderm-autonomous contri-

Figure 4. Inductive events during cardiogenesis in Drosophila (A) and chick (B) embryos. d indicates dorsal; v, ventral; and np, neuralplate.

Zaffran and Frasch Early Signals in Cardiac Development 461

bution of Gata genes in Nkx2-5 regulation and cardiomyocytedifferentiation.58 The identification of functionally importantGATA-binding sites or Gata-responsive enhancers upstreamof mouse and frog Nkx2-5 as well as many myocardialdifferentiation genes provides additional support for a role ofthese genes in early myocardial differentiation and perhapsspecification.60–64 Cross-regulation between Gata and Nkxgenes, the importance of protein-protein interactions betweenGATA and Nkx factors, and functional significance ofcombinatorial GATA and Nkx binding sites in cardiac generegulation have also been demonstrated.65–69

Clear evidence for a requirement for Gata genes incardiomyocyte specification has been obtained in the Dro-sophila system.70 Embryos mutant for the Gata gene pannierlack all cardiomyocytes, but have supernumerary pericardialcells. pannier is coexpressed with tinman during the secondtinman expression phase, although within a narrower dorsaldomain, and the two genes appear to act synergistically incardiomyocyte specification. Tinman and Pannier can het-erodimerize like their corresponding vertebrate factors. Thetwo genes are also part of a cross-regulatory loop, since theearly expression of pannier is directly controlled by tinmanwhile the later cardioblast expression of tinman (perhapsindirectly) requires pannier.71 These data suggest that manyaspects of Gata and Nkx genes in cardiogenesis have beenwidely conserved during evolution.

T-Box Factors and the Homeobox Gene Irx4in Cardiac Patterning

The involvement of the T-box genes in heart development isaccentuated by the heart defects of mice and humans carryingmutations in TBX1 and TBX5, which have been associatedwith the human DiGeorge and Holt-Oram syndromes, respec-tively.72–76 Like Nkx2-5 and the cardiac Gata genes, Tbx5genes are expressed in the bilateral cardiac primordia ofmouse, Xenopus, chick, and fish embryos, although theirexpression becomes rapidly restricted to posterior areas of theprospective atria and sinus venosa, as well as the leftventricle.77–80 In agreement with the expression patterns,dominant-negative approaches in frog embryos and homozy-gous null mutations in the mouse cause strong disruptions ofheart development, which particularly affect sinoatrial struc-tures and are accompanied by reductions in Nkx2-5 andGata4 expression.73,79 By contrast, forced Tbx5 expression inthe anterior region of the cardiac crescent results in aberrantventricular morphogenesis and reduced mlc2v expression.81

Altogether, these data suggest a role of Tbx5 in the earlydiversification of atrial versus ventricular cell identities alongthe anteroposterior extent of the heart primordia. The dem-onstration of physical interactions between Tbx5 and Nkx2-5and their coexpression within parts of the heart field suggestsfunctional cooperation of these two cardiac factors during thisprocess.73,82

The particular roles of other T-box factors, including Tbx1and zebrafish hrT, during early cardiogenesis have been lesswell defined. hrT is expressed within the heart field beforeTbx5 in a pattern that is very similar to Gata5.83 In Drosoph-ila, potential homologs of Tbx5 and hrT are also expressed in

the developing dorsal vessel, but their functions are not yetknown due to a lack of genetic data.83,84

Irx4, a member of the iroquois subgroup of homeoboxgenes, is specifically expressed in the prospective ventricularsubarea of the early linear heart tube and expression persistsin the ventricular chambers.85–87 The expression of Irx4 isdownstream of Nkx2-5.88 Functional knockout of Irx4 in themouse results in expansion of atrial and suppression ofventricular differentiation markers, whereas forced Irx4 ex-pression in the chick heart has the opposite effect.85,87 Thesedata provide strong evidence that Irx4 promotes ventricularand suppresses atrial identities within the heart tube.

Signaling Pathways During the Induction ofCardiogenic Mesoderm

BMP/Dpp SignalingEvidence for a direct involvement of BMP signaling in earlycardiogenesis was initially obtained in the Drosophila systemthrough studies of Dpp, a member of the BMP family ofTGF-� proteins. During gastrulation and mesoderm migra-tion, Dpp is expressed in a broad band of cells within thedorsal ectoderm, which in another fly species was shown topossess heart-inducing activity29 (see above). Significantly,Drosophila mutant embryos that lack the activity of Dppform neither a dorsal vessel nor any of its progenitor cells,indicating that Dpp is one of the essential signals conferringthe heart-inducing activity of the dorsal ectoderm.89 A crucialmolecular target of Dpp in this pathway is the tinman gene,whose expression in the dorsal mesoderm is induced by Dpp(Figure 4A). Embryos that are mutant for dpp or othercomponents of the Dpp pathway lack dorsal mesodermalexpression of tinman.89 Ectopic activation of the Dpp path-way ultimately leads to ectopic formation of dorsal vesselcells in the ventral-most areas of the mesoderm.90 However,in spite of their ectopic expression of tinman, cells in lateralareas of the mesoderm cannot be transformed into cardiactissue by ectopic Dpp, presumably because of a requirementfor additional signals that are restricted to the dorsal- andventral-most areas of the germ band.

The molecular mechanisms of tinman induction by Dpphave largely been clarified. In the mesoderm, Dpp signals aremediated through the activation of the effectors Mad(�Smad1) and Medea (�Smad4) (Figure 3A), which canbind to several specific Smad-binding sites within a Dpp-responsive enhancer element that is located downstream ofthe tinman gene.91 Smad binding to these sites is however notsufficient because the synergistic activities of Smad andTinman itself, which bind to adjacent sites, are required toinduce tinman expression. The source of the autocatalyticTinman activity is derived from the transient early activationof the tinman gene by Twist in the entire mesoderm.92 Inessence, Twist-activated Tinman provides the mesoderm withthe competence to respond to Dpp by inducing tinman in adorsally restricted domain during its second phase of expres-sion and, ultimately, to induce cardiac progenitors.

Loss- and gain-of-function experiments show that althoughtinman is required, it is not sufficient to promote cardiogen-esis.30,31,90 One explanation for this observation is that tinman

462 Circulation Research September 20, 2002

may be functional only in conjunction with Dpp signaling inthe early mesoderm. Indeed, there is increasing evidence that,similar to tinman itself, there are additional genes that areinduced by the synergistic effects of Dpp�Tinman within thecardiogenic mesoderm and control heart development. Amore recently studied example is the homeobox gene even-skipped, which is induced during early cardiogenesis in asubset of pericardial cell progenitors and is required for theirnormal differentiation (Figure 3).93,94 The induction of even-skipped involves a specific enhancer element that also con-tains a combination of essential Smad and Tinman bindingsites.95,96 The Gata gene pannier may provide yet anotherexample of an important cardiac regulatory gene that isinduced by synergistic Dpp�Tinman activities.71

Given the overt similarities of the expression and functionof tinman and vertebrate Nkx2-5 genes, it has been examinedwhether the similarities extend to the regulation of thesegenes, and of cardiogenesis, by Dpp-related signals. In chickand Xenopus, solid evidence for a direct role of BMPs incardiac induction has been obtained. In the chick, the expres-sion of at least three dpp-like genes, Bmp2, 4, and 7, includesthe anterior lateral region of the embryo, which overlaps withthe precardiac region that expresses Nkx2-5 and Gata4.97,98

At this stage, Bmp2 and, to a lesser degree Bmp7, areexpressed in the adjacent endoderm, which is known topossess the main inductive activity (see above), whereasBmp4 is expressed within the mesoderm itself and, togetherwith Bmp7, also in the ectoderm. Ectopic application ofBMP-2 or BMP-4–releasing implants in vivo cause ectopicinduction of Nkx2-5 and Gata4, although not of terminaldifferentiation markers, and the exposure of tissue explants tosoluble BMP-2 or -4 induces both the early cardiac regulatorsand also terminal differentiation of cardiac tissue.97–99 Ofnote, these effects are only obtained with BMP-releasingimplants into anterior mesoderm, such as anterior/medialareas that normally develop into head mesenchyme, whereasposterior mesoderm is not competent to elicit any cardiogenicresponses in these assays. Because posterior mesoderm isable to generate heart in response to anterior endoderm, theseobservations strongly suggest that the heart-inducing activityconsists of BMPs in combination with a second endodermalsignal.

The heart-inducing activity of BMP signaling has beenfurther confirmed by studies using inhibitors of BMP signal-ing, including the BMP inhibitor noggin, truncated versionsof type I (tALK3) or type II (tBMPRII) BMP receptors, andinhibitory SMAD6.97,100–102 These studies show that BMPsignal transduction is indeed required within the cardiogenicmesoderm, and not only within the anterior endoderm, topromote cardiac differentiation (Figure 4B). A second impor-tant conclusion from these data are that BMP signaling isrequired for the maintenance of Nkx2-5 and Gata geneexpression at stages during and after the fusion of the bilateralheart primordia, but apparently not for the initial activation ofthese genes in earlier stage embryos. This finding is reminis-cent of the biphasic regulation of Drosophila tinman (seeprevious page). As previously demonstrated for the Dpp-responsive enhancer of Drosophila tinman, mouse Nkx2-5 is

also controlled by enhancer sequences that contain function-ally important SMAD binding sites.103,104

In addition to Nkx2-5, BMP signals in vertebrates are alsolikely to provide direct inputs in controlling several differentcardiac regulatory genes, which may include Gatagenes,97,102,105 T-box genes,106 and the bHLH-factor encodingHand genes.107 Moreover, BMP signals may need the pres-ence of specific cardiogenic transcription factors such asNkx2-5 to be able to induce myocardial differentiation in asynergistic fashion,108 as was shown molecularly in theDrosophila system.

Wnt/Wingless SignalingNeither in Drosophila nor in vertebrates are the spatialdomains of BMP signaling sufficient to define cardiogenicmesoderm. In both the insect and vertebrate systems, it hasbeen shown that Wnt-signaling has a major role in furtherrestricting the domains in which BMPs can elicit cardiogenicresponse. However, as discussed below, the particular mech-anisms through which this outcome is achieved differsbetween the systems.

The activity of the major Wnt signaling protein in Dro-sophila, Wingless, is essential for all aspects of cardiogen-esis.109 The cardiogenic activity of Wingless is requiredduring and probably shortly after the period when Dpp isexpressed in the dorsal ectoderm adjacent to the cardiogenicmesoderm.109 During this period, Wingless is expressed inperiodic transverse stripes in the ectoderm, which intersectdorsally with the longitudinal Dpp domains. The progenitorsof the dorsal vessel are induced precisely within the areas ofthe mesoderm where the Wingless domains intersect with thedorsal Dpp domain. By contrast, the areas that are exposed toDpp but not Wingless form visceral muscle progenitors.Combined with existing genetic data, a simple model hasbeen proposed that the combination of Dpp and Winglesssignals is required to elicit a cardiogenic response in theadjacent mesoderm while Dpp alone induces visceralmesoderm.13

Additional studies have addressed the genetic and molec-ular mechanisms of how combined Dpp and Wingless sig-naling promotes cardiogenesis. The Wingless signals aretransduced in the mesoderm via the canonical Wingless/Wntpathway110 and participate in the induction of target genesthat are required for heart specification. One of the two directmesodermal targets of the Wingless signaling cascade that areknown to date is the fork head domain encoding gene sloppypaired (slp).111 Hence, sloppy paired becomes expressed in apattern of transverse stripes directly beneath the ectodermalWingless stripes (Figure 3B). As for wingless itself, loss ofslp activity results in a complete absence of cardiogenicprecursors, showing that slp is an essential mediator ofWingless signaling in this developmental pathway. It ispresently not firmly established whether slp serves to activatespecific cardiac downstream targets in a direct manner orwhether it acts indirectly by inhibiting the inappropriateactivation of visceral mesoderm genes that would blockcardiogenesis.

The second known example of a direct Wingless target inthe mesoderm is even-skipped. It has been shown that the

Zaffran and Frasch Early Signals in Cardiac Development 463

pericardial enhancer element of even-skipped (see previouspage) contains several binding sites for the Wingless effectordTCF, which are required in addition to the Smad andTinman binding sites for its normal activity in pericardialprogenitors.95,96 These findings identify even-skipped as aparadigm for a cardiac target on which the activities ofTinman as well as the Dpp and Wingless signaling cascadesconverge. Interestingly, whereas partial inactivation of thedTCF binding sites within this even-skipped enhancer leads toa reduction of its activity,95 the inactivation of all sites resultsin ectopic activity within the dorsal mesoderm, which occurseven in wingless mutant embryos.96 These and other obser-vations (M. Frasch, unpublished data, 2001) indicate thatWingless signaling is mainly required to relieve the repres-sive activity of the HMG protein dTCF to allow induction ofeven-skipped and perhaps other cardiac genes by Dpp andother signals (Figure 5B).

In contrast to Drosophila, where Wingless is essential forcardiac induction, studies in vertebrates have uncovered animportant role of Wnt signaling in blocking cardiogenesis.Specifically, overexpression of Wnt3A and Wnt8 in injectedfrog embryos or exposure of mesoderm from the heart field ofchick embryos to Wnt3A and Wnt-1 blocks expression ofNkx2-5, Tbx5, and cardiac differentiation.112,113 In the chick,this treatment results in a transformation from cardiogenicmesoderm into blood. Conversely, injection of an inhibitorycomponent of the canonical Wnt signaling cascade, GSK3�,causes ectopic cardiac differentiation in cells of the ventralmarginal zone from frog embryos that are normally destinedto form blood cells.113

Strong candidates for endogenous Wnt antagonists thatfunction to derepress cardiac induction have been identifiedin both chick and frog. These are the secreted factors Crescentand Dkk-1, which can inhibit specific Wnt ligands, includingWnt3A and Wnt-8. In posterior lateral plate mesoderm and

ventral marginal zone explants from chick and frog embryos,respectively, Crescent and Dkk-1 are potent inducers of theexpression of Nkx2-5, Tbx5, and of cardiac differentia-tion.112,113 Moreover, both of these Wnt antagonists areexpressed in the organizer and organizer-derived anteriorendoderm that is known to possess heart-inducing activity,whereas Wnt-3A and -8 are present predominantly in theposterior lateral plate and paraxial mesoderm during thisstage. In addition, Wnt-3A and Wnt-1 are expressed in theanterior neural tube and serve as inhibitors of cardiogenesis inthe adjacent anterior paraxial mesoderm.114

Taken together, these findings are able to explain many ofthe embryological data on cardiac induction and provide amodel for the signaling processes that restrict cardiogenesisto the anterior lateral mesoderm. In this model, Wnt antago-nists act as signals from the organizer/anterior endoderm toinitiate cardiogenesis in the adjacent mesoderm by establish-ing a zone of reduced Wnt-3a/Wnt-8 activity and expression.Elevated levels of Wnt-1 and Wnt-3a from the anterior neuraltube restrict this zone to lateral portions of the anteriormesoderm (Figure 4B). Reduced Wnt signaling then allows,either by default of through the activity of yet undefinedsignals from the endoderm, expression of Nkx2-5 and othercardiac regulatory genes in the anterior lateral mesoderm andprovides it with the competence to respond to BMP signalsfrom the endoderm and lateral mesoderm. In turn, BMPsignaling acts to maintain Nkx2-5 expression and may continueto act synergistically with Nkx2-5 and other transcriptionalregulators to promote cardiac differentiation (Figure 5A).

In light of the numerous similarities between Drosophilaand vertebrate cardiogenesis, it is puzzling that Wnt signalingappears to have opposite effects on cardiac induction, namelya positive one in Drosophila and a negative one in verte-brates. It is conceivable that there are other Wnt familymembers in vertebrates that act positively in cardiac induc-

Figure 5. Pathways of cardiac induction invertebrates (A) and Drosophila (B). TheSP1-encoding gene btd represses tin inthe blood mesoderm.92

464 Circulation Research September 20, 2002

tion and cannot be inhibited by Dkk-1 and Crescent (see NoteAdded in Proof). A role of negatively-acting Wnt familymembers in Drosophila can also not be excluded, althoughgenes encoding soluble Wnt antagonists are not detectable inthe genome. We speculate that, with the advent of solubleWnt antagonists in ancestors of the vertebrate lineage, nega-tive regulation of cardiogenesis by Wnts and its release byantagonists may have become the dominant mode of verte-brate cardiac induction.

Fibroblast Growth Factor SignalingThere is substantial evidence from both Drosophila andvertebrates that fibroblast growth factor (FGF) signaling ismaking a direct contribution to the specification of heartprogenitors. However, progress toward the understanding ofthe specific role of FGF signals in this pathway and itsrelationship with other pathway components has lagged dueto the indirect effects of these signals on heart development asa result of their involvement in early mesoderm induction andcell migration during gastrulation (see reviews115,116).

In Drosophila, one of the two FGF receptors from thisspecies, named Heartless (Htl), is expressed specifically inthe mesoderm, starting from early gastrulation until differen-tiation (Figure 4). The observed absence of the dorsal vesselin heartless mutant embryos is, at least in part, an indirectconsequence of an early requirement for FGF signaling inmesoderm migration (see review116). Hence, the majority ofmesodermal cells fail to reach the dorsal ectoderm in htlmutants, which in turn prevents cardiac induction via Dppand other signals. In addition, a more direct role of FGFsignaling in Drosophila cardiomyocytes and pericardial cellspecification has been revealed in experiments in which theFGF signaling pathway is inhibited only at a time aftermesoderm migration is completed.117,118 Transmission of theFGF signal within the mesoderm involves the Ras pathwaybecause forced expression of an activated version of Ras canpartially rescue the loss of pericardial cell fates in FGF-receptor mutant embryos.118 The activities of a mesoderm-specific signal transducer, Dof (aka, Heartbroken), upstreamof Ras and of Tinman downstream are also required.118–120

However, the identity and distribution of the ligand of theFGF receptor Htl are not yet known.

In the chick system, it has been shown by in vitro culturesthat FGF-2 and -4 can induce cardiogenesis in non-precardiacmesoderm, although induction of Nkx2-5, Gata4, and cardiacdifferentiation occurs efficiently only if BMP-2 or -4 is alsoprovided.121,122 FGF is only required transiently, whereasBMPs are required continuously for cardiogenic induction inthis system. Although FGFs are expressed in the early chickembryo at the correct time and place to be able to fulfill ananalogous function in cardiac specification in vivo, theavailable data cannot distinguish between direct and possibleindirect effects of FGF signals, as for example in theregulation of cell migration.

Apart from Drosophila, the clearest data on the in vivofunction of FGF signaling in cardiac induction has beenobtained in the zebrafish system. Zebrafish FGF8 is ex-pressed in the cardiogenic fields of the lateral plate, as well asin specific areas of the neural tube. fgf8 (acerebellar) mutant

embryos display strong heart defects with a particular loss ofventricular structures.123 At earlier stages, strong reductionsof Nkx2-5 and Gata4 are observed from the onset of theirexpression. Importantly, incubation of embryos during earlysomitogenesis (ie, after the induction and migration of themesoderm) with a specific inhibitor of the FGF-receptorFgfr1, SU5402, results in a phenocopy of the acerebellarheart phenotype, including the block of Nkx2-5 expression.123

Together, these data provide strong evidence that FGF signalsnot only act during early mesoderm development but are alsorequired more directly for the induction of cardiogenictranscription factors during subsequent stages (Figure 4B). Itis possible that different members of the FGF family havepartially redundant activities during this process, which mayaccount for the normal expression of Gata6 and the formationof residual heart (particularly atrial) tissue in acerebellarmutants.

As shown in the mouse, the cells of the anterior heart-forming field already express FGF-10 at the cardiac crescentstage (Figure 2B).4 Therefore, it is tempting to speculate thatapart from a possible role in the anterior migration of thesecells, FGF-10 may also be involved, perhaps in combinationwith BMPs, in the induction of Nkx2-5 and other regulatorsthat drive arterial pole development.

Notch SignalingSignaling via Notch receptors plays an important role in earlycardiogenesis of both Drosophila and vertebrates. Notch,which is involved in a wide range of developmental contexts,is activated by its transmembrane ligands Delta (Drosophilaand vertebrates) as well as Serrate (Drosophila) and itsvertebrate homologs Jagged. The glycosyl transferase Fringedifferentially modulates the responsiveness of Notch to Deltaversus Serrate/Jagged ligands, although this specific aspecthas not been examined in early heart development. Inaddition, the transmission of the signal to target genesinvolves the nuclear factor Su(H)/RBP-J (see review124).

During Drosophila cardiogenesis, Notch functions duringtwo distinct developmental events, namely the process oflateral inhibition and the control of lineage decisions duringasymmetric cell divisions of heart progenitors. Lateral inhi-bition involves reciprocal interactions between neighboringcells through Notch and its ligands, which occur withingroups of cells that have equivalent potentials to develop intospecific heart progenitors. As a result of these mutual inhi-bitions, the Notch pathway becomes inactivated within onecell of an equivalence group, which allows it to become aheart progenitor (provided it receives the proper combinationof other signals and cues, such as Dpp, Wingless, Tinman,etc). By contrast, continued activation of the Notch pathwayin the surrounding cells prevents them from being specifiedas heart progenitors despite the presence of cardiogenicregulators and signals. It is thought that Notch activity blocksthe functions of cardiogenic signals in these cells although themolecular mechanisms of this interference are not known.Based on several genetic observations in Drosophila, itappears that this type of reciprocal signaling controls howboth the cardioblast and pericardial cell progenitors aresingled out from larger cell clusters. In particular, mutant

Zaffran and Frasch Early Signals in Cardiac Development 465

embryos in which Notch or Delta is inactive produce stronglyincreased numbers of cardioblasts125,126 and pericardial pro-genitors.127 Although these embryos form disrupted dorsalvessels with highly increased numbers of cardioblasts, theirpericardial progenitors fail to generate any mature pericardialcells due to a subsequent requirement for Notch in thislineage (see next paragraph).

The specification of cardiac progenitor cells in Drosophilais followed by a cell division which in many, if not all, casesis asymmetric and generates two daughter cells with differentidentities. For example, one particular progenitor divides togenerate one pericardial founder and one somatic musclefounder cell, whereas another lineage generates one cardio-blast and another type of pericardial cell from a commonprogenitor.33,94,127 Similar to analogous processes duringneuronal development, differential activity of Notch is re-sponsible for the acquisition of asymmetric cardiac cell fatesin the two daughter cells. Differential Notch signaling in thetwo daughters is achieved by the differential segregation ofan intracellular inhibitor of the Notch pathway during mitosis,which is encoded by the numb gene. In numb mutantembryos, there are equal levels of Notch signaling activity inthe two daughter cells and therefore two identical cellidentities are generated. Through this mechanism, Notchsignaling and asymmetrically segregating Notch pathwayinhibitors play key roles in the diversification of cardiac cellfates.

The most informative studies on the role of Notch signal-ing in vertebrate cardiogenesis have been performed in theXenopus system.128 The available data indicate that thefunction of Notch in vertebrate cardiogenesis may be analo-gous to the Notch-dependent lateral inhibition processes inDrosophila. Xenopus Serrate1 and Notch1 are initially ex-pressed in a pattern overlapping with one another and withNkx2-5, while during subsequent stages (before and duringmyocardial differentiation) expression refines such that Ser-rate1 becomes restricted largely to dorsolateral (presumptivemesocardial and pericardial roof) areas of the heart field.Conditional activation of the Notch pathway using activatedversions of Notch or Su(H) results in upregulation of Serrate1and inhibition of myocardial differentiation, which reflectsthe situation that is normally observed in the dorsolateralareas of the heart field. Conversely, conditional inactivationof the pathway with dominant-negative components causesectopic expression of myocardial markers and an expansionof the heart.128 Based on these observations and the observedlack of an effect on Nkx2-5 and Gata4 expression in theseexperiments, it has been proposed that endogenous Notchsignaling influences the selection between myocardial andmesocardial/pericardial roof cell fates within the Nkx2-5–expressing area of the heart field.

The direct target genes of Notch in heart development arenot known. At least in certain contexts, the Hrt (�Hesr/Hey)genes of the Hairy/Enhancer of Split family were proposed ascandidates,129 although there is presently no evidence that thedifferential expression of these genes along the anteroposte-rior axis of the heart tube is controlled via Notch signaling.Nevertheless, the observations in Xenopus, the heart defectsof hypomorphic Notch2 mouse mutant embryos,130 and the

heart abnormalities associated with Jagged1 haploinsuffi-ciencies in human Alagille syndrome patients (see review131)indicate that Notch signaling is likely to be required duringmultiple processes in early vertebrate cardiogenesis. In thiscontext, it is interesting to note that the Notch-dependentareas in the frog heart field probably correspond to theanterior heart-forming fields of the mouse and chick.4 Indeed,the heart defects in the Alagille syndrome, which focus on theanterior outflow tracts, could indicate that in amniotes, Notchsignaling has an analogous role in the medial-lateral subdi-vision of the heart field as in frogs that, in this case, mayinvolve the delayed specification of myocardium of theoutflow region.

Conclusion and PerspectivesThrough the powerful combination of embryological, molec-ular, and genetic approaches in a number of vertebrate andinvertebrate model systems, we have gained significant in-sight into the molecular programs controlling early events ofcardiogenesis. The properties that have been ascribed toparticular signaling molecules and some of their targetsprovide molecular explanations for many of the classicalembryological findings on cardiac induction.

It has become apparent that, although a complete disrup-tion of any of the major pathways in early heart developmentresults in embryonic lethality, more subtle disruptions byhaploinsufficiencies and hypomorphic or dominant-negativemutations can lead to malfunctioning of the heart at laterstages during the lifespan of an organism.41,132,133 From aclinical perspective, we can therefore anticipate that thegrowing knowledge on early pathways in cardiogenesis andthe essential genes will be increasingly useful for the under-standing and diagnosis of heart disease. Moreover, there ismounting evidence of cardiomyocyte regeneration in themammalian heart, be it through the contribution of residentstem cells, circulating stem cells, or bone marrow cells thatcan be mobilized to the heart (see review134). It is thereforeconceivable that the knowledge gained on the roles of earlysignals in normal cardiogenesis will provide valuable guide-lines in designing future stem cell therapies that wouldrequire expansion, reprogramming, and proper differentiationof cardiogenic stem cell populations.

Note Added in ProofA recently published study (Pandur et al135) has indeeddemonstrated that Wnt-11 is positively required for heartinduction through a noncanonical Wnt signaling pathway.

AcknowledgmentsWe gratefully acknowledge the support of our work by the NationalInstitutes of Health (HD30832 and DK59406). We thank Patrick Lofor Figure 3C and Hanh Nguyen for critical reading of themanuscript.

References1. Tam PP, Parameswaran M, Kinder SJ, Weinberger RP. The allocation of

epiblast cells to the embryonic heart and other mesodermal lineages: therole of ingression and tissue movement during gastrulation. Devel-opment. 1997;124:1631–1642.

466 Circulation Research September 20, 2002

2. Cohen-Gould L, Mikawa T. The fate diversity of mesodermal cellswithin the heart field during chicken early embryogenesis. Dev Biol.1996;177:265–273.

3. DeHaan RL, Ursprung H. Morphogenesis of the vertebrate heart. In:DeHaan RL, Ursprung H, eds. Organogenesis. Austin, Tex: Holt,Rinehart, Winston; 1965:377–419.

4. Kelly R, Buckingham M. The anterior heart-forming field: voyage to thearterial pole of the heart. Trends Genet. 2002;18:210–216.

5. Antin PB, Taylor RG, Yatskievych T. Precardiac mesoderm is specifiedduring gastrulation in quail. Dev Dyn. 1994;200:144–154.

6. Montgomery MO, Litvin J, Gonzalez-Sanchez A, Bader D. Staging ofcommitment and differentiation of avian cardiac myocytes. Dev Biol.1994;164:63–71.

7. Bisaha JG, Bader D. Identification and characterization of a ventricu-lar-specific avian myosin heavy chain, VMHC1: expression in differen-tiating cardiac and skeletal muscle. Dev Biol. 1991;148:355–364.

8. Han Y, Dennis JE, Cohen-Gould L, Bader DM, Fischman DA.Expression of sarcomeric myosin in the presumptive myocardium ofchicken embryos occurs within six hours of myocyte commitment. DevDyn. 1992;193:257–265.

9. Edmondson DG, Lyons GE, Martin JF, Olson EN. Mef2 gene expressionmarks the cardiac and skeletal muscle lineages during mouse embryo-genesis. Development. 1994;120:1251–1263.

10. Christoffels VM, Habets PE, Franco D, Campione M, de Jong F, LamersWH, Bao ZZ, Palmer S, Biben C, Harvey RP, Moorman AF. Chamberformation and morphogenesis in the developing mammalian heart. DevBiol. 2000;223:266–278.

11. Redkar A, Montgomery M, Litvin J. Fate map of early avian cardiacprogenitor cells. Development. 2001;128:2269–2279.

12. Harvey RP. Links in the left/right axial pathway. Cell. 1998;94:273–276.

13. Bodmer R, Frasch M. Genetic determination of Drosophila heart devel-opment. In: Harvey RP, Rosenthal N, eds. Heart Development. SanDiego, Calif: Academic Press; 1999:65–90.

14. Inagaki T, Garcia-Martinez V, Schoenwolf GC. Regulative ability of theprospective cardiogenic and vasculogenic areas of the primitive streakduring avian gastrulation. Dev Dyn. 1993;197:57–68.

15. Schultheiss TM, Xydas S, Lassar AB. Induction of avian cardiac myo-genesis by anterior endoderm. Development. 1995;121:4203–4214.

16. Yatskievych T, Ladd A, Antin P. Induction of cardiac myogenesis inavian pregastrula epiblast: the role of the hypoblast and activin. Devel-opment. 1997;124:2561–2570.

17. Jacobson A. Heart determination in the newt. J Exp Zool. 1961;146:139–151.

18. Nascone N, Mercola M. An inductive role for the endoderm in Xenopuscardiogenesis. Development. 1995;121:515–523.

19. Gannon M, Bader D. Initiation of cardiac differentiation occurs in theabsence of anterior endoderm. Development. 1995;121:2439–2450.

20. Alexander J, Rothenberg M, Henry G, Stainier D. casanova plays anearly and essential role in endoderm formation in zebrafish. Dev Biol.1999;215:343–357.

21. Sater A, Jacobson A. The role of the dorsal lip in the induction of heartmesoderm in Xenopus laevis. Development. 1990;108:461–470.

22. Jacobson A. Influences of ectoderm and endoderm on heart differen-tiation in the newt. Dev Biol. 1960;2:138–154.

23. Rosenquist GC, De Haan RL. Migration of precardiac cells in the chickembryo: a radioautographic study. Carnegie Inst Wash ContribEmbryol. 1966;38:111–121.

24. Sater A, Jacobson A. The restriction of the heart morphogenetic field inXenopus laevis. Dev Biol. 1990;140:328–336.

25. Fullilove S. Heart induction: distribution of active factors in newtendoderm. J Exp Zool. 1970;175:323–326.

26. Climent S, Sarasa M, Villar J, Murillo-Ferrol N. Neurogenic cells inhibitthe differentiation of cardiogenic cells. Dev Biol. 1995;171:130–148.

27. Raffin M, Leong LM, Rones MS, Sparrow D, Mohun T, Mercola M.Subdivision of the cardiac Nkx2.5 expression domain into myogenic andnonmyogenic compartments. Dev Biol. 2000;218:326–340.

28. Goldstein A, Fishman M. Notochord regulates cardiac lineage inzebrafish embryo. Dev Biol. 1998;201:247–252.

29. Seidel F, Bock E, Krause G. Die Organisation des Insekteneies. Natur-wissenschaften. 1940;28:434–447.

30. Azpiazu N, Frasch M. tinman and bagpipe: two homeo box genes thatdetermine cell fates in the dorsal mesoderm of. Drosophila Genes Dev.1993;7:1325–1340.

31. Bodmer R. The gene tinman is required for specification of the heart andvisceral muscles in Drosophila. Development. 1993;118:719–729.

32. Bodmer R, Jan LY, Jan YN. A new homeobox-containing gene, msh-2,is transiently expressed early during mesoderm formation of Drosophila.Development. 1990;110:661–669.

33. Ward E, Skeath J. Characterization of a novel subset of cardiac cells andtheir progenitors in the Drosophila embryo. Development. 2000;127:4959–4969.

34. Molina M, Cripps R. Ostia, the inflow tracts of the Drosophila heart,develop from a genetically distinct subset of cardial cells. Mech Dev.2001;109:51–59.

35. Evans S. Vertebrate tinman homologues and cardiac differentiation.Semin Cell Dev Biol. 1999;10:73–83.

36. Tonissen K, Drysdale T, Lints T, Harvey R, Krieg P. XNkx-2.5, aXenopus gene related to Nkx-2.5 and tinman: evidence for a conservedrole in cardiac development. Dev Biol. 1994;162:325–328.

37. Komuro I, Izumo S. Csx: a murine homeobox-containing gene specif-ically expressed in the developing heart. Proc Natl Acad Sci U S A.1993;90:8145–8149.

38. Lints TJ, Parsons LM, Hartley L, Lyons I, Harvey RP. Nkx-2.5: a novelmurine homeobox gene expressed in early heart progenitor cells andtheir myogenic descendants. Development. 1993;119:419–431.

39. Chen J, Fishman M. Zebrafish tinman homolog demarcates the heartfield and initiates myocardial differentiation. Development. 1996;122:3809–3816.

40. Lyons I, Parsons LM, Hartley L, Li R, Andrews JE, Robb L, Harvey RP.Myogenic and morphogenetic defects in the heart tubes of murineembryos lacking the homeo box gene, Nkx2-5. Genes Dev. 1995;9:1654–1666.

41. Schott JJ, Benson DW, Basson CT, Pease W, Silberbach GM, Moak JP,Maron BJ, Seidman CE, Seidman JG. Congenital heart disease causedby mutations in the transcription factor NKX2–5. Science. 1998;281:108–111.

42. Nikolova M, Chen X, Lufkin T. Nkx2.6 expression is transiently andspecifically restricted to the branchial region of pharyngeal-stage mouseembryos. Mech Dev. 1997;69:215–218.

43. Pabst O, Schneider A, Brand T, Arnold H. The mouse Nkx2–3 home-odomain gene is expressed in gut mesenchyme during pre- and postnatalmouse development. Dev Dyn. 1997;209:29–35.

44. Biben C, Hatzistavrou T, Harvey R. Expression of NK-2 classhomeobox gene Nkx2–6 in foregut endoderm and heart. Mech Dev.1998;73:125–127.

45. Tanaka M, Schinke M, Liao HS, Yamasaki N, Izumo S. Nkx2.5 andNkx2.6, homologs of Drosophila tinman, are required for developmentof the pharynx. Mol Cell Biol. 2001;21:4391–4398.

46. Fu Y, Yan W, Mohun T, Evans S. Vertebrate tinman homologuesXNkx2-3 and XNkx2-5 are required for heart formation in a functionallyredundant manner. Development. 1998;125:4439–4449.

47. Grow M, Krieg P. tinman function is essential for vertebrate heartdevelopment: elimination of cardiac differentiation by dominant inhib-itory mutants of the tinman-related genes, XNkx2-3 and XNkx2-5 DevBiol. 1998;204:187–196.

48. Molkentin JD. The zinc finger-containing transcription factors GATA-4,-5, and -6: ubiquitously expressed regulators of tissue-specific geneexpression. J Biol Chem. 2000;275:38949–38952.

49. Serbedzija G, Chen J, Fishman M. Regulation in the heart field ofzebrafish. Development. 1998;1095–1101.

50. Heikinheimo M, Scandrett JM, Wilson DB. Localization of transcriptionfactor GATA-4 to regions of the mouse embryo involved in cardiacdevelopment. Dev Biol. 1994;164:361–373.

51. Laverriere A, MacNeill C, Mueller C, Poelmann R, Burch J, Evans T.GATA-4/5/6, a subfamily of three transcription factors transcribed indeveloping heart and gut. J Biol Chem. 1994;269:23177–23184.

52. Jiang Y, Evans T. The Xenopus GATA-4/5/6 genes are associated withcardiac specification and can regulate cardiac-specific transcriptionduring embryogenesis. Dev Biol. 1996;174:258–270.

53. Morrisey EE, Ip HS, Lu MM, Parmacek MS. GATA-6: a zinc fingertranscription factor that is expressed in multiple cell lineages derivedfrom lateral mesoderm. Dev Biol. 1996;177:309–322.

54. Morrisey EE, Ip HS, Tang Z, Lu MM, Parmacek MS. GATA-5: atranscriptional activator expressed in a novel temporally and spatially-restricted pattern during embryonic development. Dev Biol. 1997;183:21–36.

55. Kuo CT, Morrisey EE, Anandappa R, Sigrist K, Lu MM, Parmacek MS,Soudais C, Leiden JM. GATA4 transcription factor is required for

Zaffran and Frasch Early Signals in Cardiac Development 467

ventral morphogenesis and heart tube formation. Genes Dev. 1997;11:1048–1060.

56. Molkentin JD, Lin Q, Duncan SA, Olson EN. Requirement of thetranscription factor GATA4 for heart tube formation and ventral mor-phogenesis. Genes Dev. 1997;11:1061–1072.

57. Jiang Y, Tarzami S, Burch JB, Evans T. Common role for each of thecGATA-4/5/6 genes in the regulation of cardiac morphogenesis. DevGenet. 1998;22:263–277.

58. Reiter JF, Alexander J, Rodaway A, Yelon D, Patient R, Holder N,Stainier DY. Gata5 is required for the development of the heart andendoderm in zebrafish. Genes Dev. 1999;13:2983–2995.

59. Narita N, Bielinska M, Wilson DB. Cardiomyocyte differentiation byGATA-4-deficient embryonic stem cells. Development. 1997;124:3755–3764.

60. Searcy R, Vincent E, Liberatore C, Yutzey KE. A GATA-dependentnkx-2.5 regulatory element activates early cardiac gene expression intransgenic mice. Development. 1998;125:4461–4470.

61. Lien C, Wu C, Mercer B, Webb R, Richardson J, Olson E. Control ofearly cardiac-specific transcription of Nkx2–5 by a GATA-dependentenhancer. Development. 1999;126:75–84.

62. Reecy J, Li X, Yamada M, DeMayo F, Newman C, Harvey R, SchwartzR. Identification of upstream regulatory regions in the heart-expressedhomeobox gene Nkx2–5. Development. 1999;126:839–849.

63. Sparrow D, Cai C, Kotecha S, Latinkic B, Cooper B, Towers N, EvansS, Mohun T. Regulation of the tinman homologues in Xenopus embryos.Dev Biol. 2000;227:65–79.

64. Charron F, Nemer M. GATA transcription factors and cardiac devel-opment. Semin Cell Dev Biol. 1999;10:85–91.

65. Durocher D, Charron F, Warren R, Schwartz RJ, Nemer M. The cardiactranscription factors Nkx2-5 and GATA-4 are mutual cofactors. EMBOJ. 1997;16:5687–5696.

66. Lee Y, Shioi T, Kasahara H, Jobe S, Wiese R, Markham B, Izumo S.The cardiac tissue-restricted homeobox protein Csx/Nkx2.5 physicallyassociates with the zinc finger protein GATA4 and cooperatively acti-vates atrial natriuretic factor gene expression. Mol Cell Biol. 1998;18:3120–3129.

67. Shiojima I, Komuro I, Oka T, Hiroi Y, Mizuno T, Takimoto E, MonzenK, Aikawa R, Akazawa H, Yamazaki T, Kudoh S, Yazaki Y. Context-dependent transcriptional cooperation mediated by cardiac transcriptionfactors Csx/Nkx-2.5 and GATA-4. J Biol Chem. 1999;274:8231–8239.

68. Davis D, Wessels A, Burch J, An Nkx-dependent enhancer regulatescGATA-6 gene expression during early stages of heart development. DevBiol. 2000;217:301–319.

69. Molkentin J, Antos C, Mercer B, Taigen T, Miano J, Olson E. Directactivation of a GATA6 cardiac enhancer by Nkx2.5: evidence for areinforcing regulatory network of Nkx2.5 and GATA transcriptionfactors in the developing heart. Dev Biol. 2000;217:301–309.

70. Gajewski K, Fossett N, Molkentin J, Schulz RA. The zinc fingerproteins Pannier and GATA4 function as cardiogenic factors in Dro-sophila. Development. 1999;126:5679–5688.

71. Gajewski K, Zhang Q, Choi C, Fossett N, Dang A, Kim Y, Kim Y,Schulz R. Pannier is a transcriptional target and partner of Tinmanduring Drosophila cardiogenesis. Dev Biol. 2001;233:425–436.

72. Basson CT, Bachinsky DR, Lin RC, Levi T, Elkins JA, Soults J, Grayzel D,Kroumpouzou E, Traill TA, Leblanc-Straceski J, Renault B, KucherlapatiR, Seidman JG, Seidman CE. Mutations in human TBX5 cause limb andcardiac malformation in Holt-Oram syndrome. Nat Genet. 1997;15:30–35.

73. Bruneau BG, Nemer G, Schmitt JP, Charron F, Robitaille L, Caron S,Conner DA, Gessler M, Nemer M, Seidman CE, Seidman JG. A murinemodel of Holt-Oram syndrome defines roles of the T-box transcriptionfactor Tbx5 in cardiogenesis and disease. Cell. 2001;106:709–721.

74. Jerome LA, Papaioannou VE. DiGeorge syndrome phenotype in micemutant for the T-box gene, Tbx1. Nat Genet. 2001;27:286–291.

75. Lindsay EA, Vitelli F, Su H, Morishima M, Huynh T, Pramparo T,Jurecic V, Ogunrinu G, Sutherland HF, Scambler PJ, Bradley A, BaldiniA. Tbx1 haploinsufficieny in the DiGeorge syndrome region causesaortic arch defects in mice. Nature. 2001;410:97–101.

76. Merscher S, Funke B, Epstein JA, Heyer J, Puech A, Lu MM, Xavier RJ,Demay MB, Russell RG, Factor S, Tokooya K, Jore BS, Lopez M,Pandita RK, Lia M, Carrion D, Xu H, Schorle H, Kobler JB, ScamblerP, Wynshaw-Boris A, Skoultchi AI, Morrow BE, Kucherlapati R. TBX1is responsible for cardiovascular defects in velo-cardio-facial/DiGeorgesyndrome. Cell. 2001;104:619–629.

77. Chapman DL, Garvey N, Hancock S, Alexiou M, Agulnik SI,Gibson-Brown JJ, Cebra-Thomas J, Bollag RJ, Silver LM, Papaioannou

VE. Expression of the T-box family genes, Tbx1-Tbx5, during earlymouse development. Dev Dyn. 1996;206:379–390.

78. Bruneau BG, Logan M, Davis N, Levi T, Tabin CJ, Seidman JG,Seidman CE. Chamber-specific cardiac expression of Tbx5 and heartdefects in Holt-Oram syndrome. Dev Biol. 1999;211:100–108.

79. Horb M, Thomsen G. Tbx5 is essential for heart development. Devel-opment. 1999;126:1739–1751.

80. Begemann G, Ingham P. Developmental regulation of Tbx5 in zebrafishembryogenesis. Mech Dev. 2000;90:299–304.

81. Liberatore CM, Searcy-Schrick RD, Yutzey KE. Ventricular expressionof tbx5 inhibits normal heart chamber development. Dev Biol. 2000;223:169–180.

82. Hiroi Y, Kudoh S, Monzen K, Ikeda Y, Yazaki Y, Nagai R, Komuro I.Tbx5 associates with Nkx2-5 and synergistically promotes cardiomyo-cyte differentiation. Nat Genet. 2001;28:276–280.

83. Griffin KJ, Stoller J, Gibson M, Chen S, Yelon D, Stainier DY,Kimelman D. A conserved role for H15-related T-box transcriptionfactors in zebrafish and Drosophila heart formation. Dev Biol. 2000;218:235–247.

84. Lo PC, Frasch M. A role for the COUP-TF-related gene seven-up in thediversification of cardioblast identities in the dorsal vessel of Dro-sophila. Mech Devel. 2001;104:49–60.

85. Bao ZZ, Bruneau BG, Seidman JG, Seidman CE, Cepko CL. Regulationof chamber-specific gene expression in the developing heart by Irx4.Science. 1999;283:1161–1164.

86. Christoffels V, Keijser A, Houweling A, Clout D, Moorman A. Pat-terning the embryonic heart: identification of five mouse Iroquoishomeobox genes in the developing heart. Dev Biol. 2000;224:263–274.

87. Bruneau BG, Bao ZZ, Fatkin D, Xavier-Neto J, Georgakopoulos D,Maguire CT, Berul CI, Kass DA, Kuroski-de Bold ML, de Bold AJ,Conner DA, Rosenthal N, Cepko CL, Seidman CE, Seidman JG. Car-diomyopathy in Irx4-deficient mice is preceded by abnormal ventriculargene expression. Mol Cell Biol. 2001;21:1730–1736.

88. Bruneau BG, Bao ZZ, Tanaka M, Schott JJ, Izumo S, Cepko CL,Seidman JG, Seidman CE. Cardiac expression of the ventricle-specifichomeobox gene Irx4 is modulated by Nkx2-5 and dHand. Dev Biol.2000;217:266–277.

89. Frasch M. Induction of visceral and cardiac mesoderm by ectodermalDpp in the early Drosophila embryo. Nature. 1995;374:464–467.

90. Yin Z, Frasch M. Regulation and function of tinman during dorsalmesoderm induction and heart specification in Drosophila. Dev Genet.1998;22:187–200.

91. Xu X, Yin Z, Hudson J, Ferguson E, Frasch M. Smad proteins act incombination with synergistic and antagonistic regulators to target Dppresponses to the Drosophila mesoderm. Genes Dev. 1998;12:2354–2370.

92. Yin Z, Xu X-L, Frasch M. Regulation of the Twist target gene tinmanby modular cis-regulatory elements during early mesoderm devel-opment. Development. 1997;124:4871–4982.

93. Frasch M, Hoey T, Rushlow C, Doyle HJ, Levine M. Characterizationand localization of the even-skipped protein of Drosophila. EMBO J.1987;6:749–759.

94. Su M, Fujioka M, Goto T, Bodmer R. The Drosophila homeobox geneszfh-1 and even-skipped are required for cardiac-specific differentiationof a numb-dependent lineage decision. Development. 1999;126:3241–3251.

95. Halfon M, Carmena A, Gisselbrecht S, Sackerson C, Jimenez F, BayliesM, Michelson A. Ras pathway specificity is determined by the inte-gration of multiple signal-activated and tissue-restricted transcriptionfactors. Cell. 2000;103:63–74.

96. Knirr S, Frasch M. Molecular integration of inductive and mesoderm-intrinsic inputs governs even-skipped enhancer activity in a subset ofpericardial and dorsal muscle progenitors. Dev Biol. 2001;238:13–26.

97. Schultheiss TM, Burch JB, Lassar AB. A role for bone morphogeneticproteins in the induction of cardiac myogenesis. Genes Dev. 1997;11:451–462.

98. Andrée B, Duprez D, Vorbusch B, Arnold H-H, Brand T. BMP-2induces ectopic expression of cardiac lineage markers and interfereswith somite formation in chicken embryos. Mech Dev. 1998;70:119–131.

99. Nakajima Y, Yamagishi T, Ando K, Nakamura H. Significance of bonemorphogenetic protein-4 function in the initial myofibrillogenesis ofchick cardiogenesis. Dev Biol. 2002;245:219–303.

468 Circulation Research September 20, 2002

100. Schlange T, Andree B, Arnold H, Brand T. BMP2 is required for earlyheart development during a distinct time period. Mech Dev. 2000;91:259–270.

101. Shi Y, Katsev S, Cai C, Evans S. BMP signaling is required for heartformation in vertebrates. Dev Biol. 2000;224:226–237.

102. Walters M, Wayman G, Christian J. Bone morphogenetic proteinfunction is required for terminal differentiation of the heart but not forearly expression of cardiac marker genes. Mech Dev. 2001;100:263–273.

103. Liberatore C, Searcy-Schrick R, Vincent E, Yutzey K. Nkx-2.5 geneinduction in mice is mediated by a Smad consensus regulatory region.Dev Biol. 2002;244:243–256.

104. Lien C, McAnally J, Richardson J, Olson E. Cardiac-specific activity ofan Nkx2–5 enhancer requires an evolutionarily conserved Smad bindingsite. Dev Biol. 2002;244:257–266.

105. Reiter J, Verkade H, Stainier D. Bmp2b and Oep promote early myo-cardial differentiation through their regulation of gata5. Dev Biol. 2001;234:330–338.

106. Yamada M, Revelli J, Eichele G, Barron M, Schwartz R. Expression ofchick Tbx-2, Tbx-3, and Tbx-5 genes during early heart development:evidence for BMP2 induction of Tbx2. Dev Biol. 2000;228:95–105.

107. Sparrow D, Kotecha S, Towers N, Mohun T. Xenopus eHAND: a markerfor the developing cardiovascular system of the embryo that is regulatedby bone morphogenetic proteins. Mech Dev. 1998;71:151–163.

108. Jamali M, Karamboulas C, Rogerson P, Skerjanc I. BMP signalingregulates Nkx2–5 activity during cardiomyogenesis. FEBS Lett. 2001;509:126–130.

109. Wu X, Golden K, Bodmer R. Heart development in Drosophila requiresthe segment polarity gene wingless. Dev Biol. 1995;169:619–628.

110. Park M, Wu X, Golden K, Axelrod JD, Bodmer R. The Winglesssignaling pathway is directly involved in Drosophila heart development.Dev Biol. 1996;177:104–116.

111. Lee H, Frasch M. Wingless effects mesoderm patterning and ectodermsegmentation events via induction of its downstream target sloppypaired. Development. 2000;127:5497–5508.

112. Marvin MJ, Di Rocco G, Gardiner A, Bush SM, Lassar AB. Inhibitionof Wnt activity induces heart formation from posterior mesoderm. GenesDev. 2001;15:316–327.

113. Schneider VA, Mercola M. Wnt antagonism initiates cardiogenesis inXenopus laevis. Genes Dev. 2001;15:304–315.

114. Tzahor E, Lassar AB. Wnt signals from the neural tube block ectopiccardiogenesis. Genes Dev. 2001;15:255–260.

115. Rossant J, Ciruna B, Partanen J. FGF signaling in mouse gastrulationand anteroposterior patterning. Cold Spring Harb Symp Quant Biol.1997;62:127–133.

116. Narasimha M, Leptin M. Cell movements during gastrulation: come inand be induced. Trends Cell Biol. 2000;10:169–172.

117. Beiman M, Shilo B, Volk T. Heartless, a Drosophila FGF receptorhomolog, is essential for cell migration and establishment of severalmesodermal lineages. Genes Dev. 1996;10:2993–3002.

118. Michelson AM, Gisselbrecht S, Zhou Y, Baek KH, Buff EM. Dualfunctions of the heartless fibroblast growth factor receptor in developmentof the Drosophila embryonic mesoderm. Dev Genet. 1998;22:212–229.

119. Michelson A, Gisselbrecht S, Buff E, Skeath J. Heartbroken is a specificdownstream mediator of FGF receptor signalling in Drosophila. Devel-opment. 1998;125:4379–4389.

120. Vincent S, Wilson R, Coelho C, Affolter M, Leptin M. The Drosophilaprotein Dof is specifically required for FGF signaling. Mol Cell. 1998;2:515–525.

121. Ladd A, Yatskievych T, Antin P. Regulation of avian cardiac myo-genesis by activin/TGF� and bone morphogenetic proteins. Dev Biol.1998;204:407–419.

122. Barron M, Gao M, Lough J, Requirement for BMP, and FGF signalingduring cardiogenic induction in non-precardiac mesoderm is specific,transient, and cooperative. Dev Dyn. 2000;218:383–393.

123. Reifers F, Walsh E, Leger S, Stainier D, Brand M. Induction anddifferentiation of the zebrafish heart requires fibroblast growth factor 8(fgf8/acerebellar). Development. 2000;127:2549–2561.

124. Mumm J, Kopan R. Notch signaling: from the outside in. Dev Biol.2000;228:151–165.

125. Hartenstein AY, Rugendorff A, Tepass U, Hartenstein V. The functionof the neurogenic genes during epithelial development in the Drosophilaembryo. Development. 1992;116:1203–1220.

126. Zaffran S, Astier M, Gratecos D, Guillen A, Semeriva M. Cellularinteractions during heart morphogenesis in the Drosophila embryo. BiolCell. 1995;84:13–24.

127. Carmena A, Gisselbrecht S, Harrison J, Jimenez F, Michelson A. Combi-natorial signaling codes for the progressive determination of cell fates in theDrosophila embryonic mesoderm. Genes Dev. 1998;15:3910–3922.

128. Rones M, McLaughlin K, Raffin M, Mercola M, Serrate, and Notchspecify cell fates in the heart field by suppressing cardiomyogenesis.Development. 2000;127:3865–3876.

129. Nakagawa O, McFadden D, Nakagawa M, Yanagisawa H, Hu T, SrivastavaD, Olson E. Members of the HRT family of basic helix-loop-helix proteinsact as transcriptional repressors downstream of Notch signaling. Proc NatlAcad Sci U S A. 2000;97:13655–13660.

130. McCright B, Gao X, Shen L, Lozier J, Lan Y, Maguire M, Herzlinger D,Weinmaster G, Jiang R, Gridley T. Defects in development of thekidney, heart and eye vasculature in mice homozygous for a hypo-morphic Notch2 mutation. Development. 2001;128:491–502.

131. Piccoli D, Spinner N. Alagille syndrome and the Jagged1 gene. SeminLiver Dis. 2001;21:525–534.

132. Chien K. Genomic circuits and the integrative biology of cardiacdiseases. Nature. 2000;407:227–232.

133. Srivastava D, Olson EN. A genetic blueprint for cardiac development.Nature. 2000;407:221–226.

134. Anversa P, Nadal-Ginard B. Myocyte renewal and ventricular remod-elling. Nature. 2002;415:240–243.

135. Pandur P, Lasche M, Eisenberg LM, Kuhl M. Wnt-11 activation of anon-canonical Wnt signalling pathway is required for cardiogenesis.Nature. 2002;418:636–641.

Zaffran and Frasch Early Signals in Cardiac Development 469