Foxh1 Is Essential for Development of the Anterior Heart Field

15 Pages·2004·0.9 MB·English

DevelopmentalCell,Vol.7,331–345,September,2004,Copyright2004byCellPress
Foxh1 Is Essential for Development
of the Anterior Heart Field
IngovonBoth,1CristoforoSilvestri,2,7 which is the most common form of birth defect in hu-
TubaErdemir,1,7HeikoLickert,1 mans(Harvey,2002a;OlsonandSchneider,2003).Initia-
JohnathonR.Walls,3R.MarkHenkelman,3 tionofheartdevelopmentoccursinthecardiaccrescent,
JanetRossant,1,4RichardP.Harvey,6 or primary heart field, when cells adopt a cardiac fate
LilianaAttisano,5andJeffreyL.Wrana1,4,* inresponsetoextracellularcuessuchasBoneMorpho-
1SamuelLunenfeldResearchInstitute genetic Proteins (BMPs), Fibroblast Growth Factors
MountSinaiHospital (FGFs),andWinglessrelatedfactors(WNTs).Cellsfrom
600UniversityAvenue thisprimaryheartfieldthencontributetothedeveloping
Toronto,OntarioM5G1X5 linearhearttube,whichinmammalsiscomposedlargely
Canada ofprospectiveleftventricle(LV)(Caietal.,2003).Atthe
2InstituteofMedicalScience loopingstage,thereisalargeexpansionatthearterial
UniversityofToronto poleofthedevelopingheartintheRVandOFTregion
3DepartmentofMedicalBiophysics thatoccursconcomitantlywithrightwardloopingofthe
4DepartmentofMedicalGeneticsandMicrobiology linearhearttube(KellyandBuckingham,2002).Atthis
5DepartmentofBiochemistry earlystage,regionalspecificationoftheheartchambers
MedicalSciencesBuilding isalreadyevident.Forexample,Hand2(formerlydHand)
1King’sCollegeCircle and Mef2c, which encode bHLH and MADS box tran-
UniversityofToronto scriptionfactors,respectively,displayrestrictedexpres-
Toronto,OntarioM5S1A8 sion to the RV and OFT, and deletion of these genes
Canada leadstodefectsinRVandOFTmorphogenesis(Linet
6VictorChangCardiacResearchInstitute al.,1997;Srivastavaetal.,1997).
FacultiesofLifeScienceandMedicine Earlyanalysisofheartdevelopmentinthemouseand
UniversityofNewSouthWales fatemappingstudiesinthechicksuggestthattheOFT
384VictoriaStreet doesnotarisefromtheprimaryheartfieldbutisadded
Kensington2051,NewSouthWales laterfromadistinctpopulationofcells(Kellyetal.,2001;
Australia Kelly and Buckingham, 2002; Mjaatvedt et al., 2001;
Waldoet al.,2001). Thissecondary, oranterior, heart-
forming field (AHF) arises from a group of splanchnic
mesodermal cells that lie medial to the primary heart
Summary field cells and extend anterior and dorsal to the early
linearhearttube.Furthermore,inthemouse,fatemap-
The anterior heart field (AHF) mediates formation of pingtogetherwithanalysisoftheFgf10genesuggests
theoutflowtract(OFT)andrightventricle(RV)during that as the arterial pole of the heart undergoes rapid
loopingmorphogenesisoftheheart.Foxh1isafork- expansion,agroupofFgf10-expressingcellsinthepha-
head DNA binding transcription factor in the TGF(cid:1)- ryngealmesodermcontributetosplanchnicmesoderm
Smad pathway. Here we demonstrate that Foxh1(cid:1)/(cid:1) andthencetothegrowinghearttube(Kellyetal.,2001).
mutant mouse embryos form a primitive heart tube, TheAHFisthusthoughttobeacriticalprecursorpopu-
butfailtoformOFTandRVanddisplaylossofouter lationforformationoftheRVandOFT.
curvaturemarkersofthefutureworkingmyocardium, Littleisknownofthesignalsthatregulateformation
similar to the phenotype of Mef2c(cid:1)/(cid:1) mutant hearts. oftheAHF.However,BMPsareexpressedintheAHF
Further,weshowthatMef2cisadirecttargetofFoxh1, (Waldoetal.,2001)andlikelyfunctiontoinduceexpres-
whichphysicallyandfunctionallyinteractswithNkx2-5 sionof thepancardiac homeoboxgene Nkx2-5,which
to mediate strong Smad-dependent activation of a is essential for myocardial differentiation in vitro and
TGF(cid:1)responseelementintheMef2cgene.Thisele- cardiac morphogenesis and chamber differentiation
mentdirectstransgeneexpressiontothepresumptive in vivo(Cripps and Olson,2002; Harvey,2002b; Olson
AHF,aswellastheRVandOFT,apatternthatclosely and Schneider, 2003). Expression of Fgf8 and Fgf10
parallels endogenous Mef2c expressionin the heart. also marks the AHF and may regulate cardiomyocyte
Thus,Foxh1andNkx2-5functionallyinteractandare differentiation or the migratory behavior of precursor
essential fordevelopment of theAHF andits deriva- cellsdestinedtomoveinto thearterialpole(Kellyand
tives, the RV and OFT, in response to TGF(cid:1)-like Buckingham, 2002). The medial splanchnic mesoderm
signals. thatcontributestotheAHFalsoexpressestheLIMho-
meodomainproteinIsl1,andinmouseembryoslacking
Isl1,expressionoflocalBMPandFGFgenesisdown-
Introduction regulated(Caietal.,2003).AlthoughIsl1-positivecells
maycontributecardiomyocytestoboththeanteriorand
Theheartarisesshortlyaftergastrulation,andproblems posteriorpolesoftheheart,themostpronouncedde-
inheartmorphogenesisleadtocongenitalheartdisease, fects in the mutants are in anterior regions (Cai et al.,
2003).ThissuggeststhatIsl1isimportantforexpression
*Correspondence:[email protected] of morphogens that control cardiac development and
7Theseauthorscontributedequallytothiswork. thespecificationofsplanchnicmesodermthatgivesrise
DevelopmentalCell
332
Figure1. DefectsinHeartMorphologyinFoxh1(cid:1)/(cid:1)MutantEmbryos
(A)Scanningelectronmicroscopyanalysisofstagematchedwild-type((cid:2)/(cid:2))andFoxh1nullmutant((cid:1)/(cid:1))embryosatembryonicdaysE8.5
andE9.0.Frontalviewsofeachembryoareshownwiththepericardiumremoved.MostFoxh1(cid:1)/(cid:1)mutantembryohearts(iii,iv)arrestbyLS-
IIwithdisorganizedprimitiveventricles.Scalebar200(cid:3)m.
Foxh1IsRequiredforFormationoftheAHF
333
totheAHF.Ofnote,theforkheadDNAbindingprotein pathway specifies the AHF through a Foxh1-Nkx2-5
Foxh1,previouslynamedFAST2(Labbeetal.,1998),is complexthatregulatesMef2cexpression.
stronglyexpressedintheheart-formingregionsduring
thelooping stage(Weisberg etal.,1998). SinceFoxh1 Results
isanucleartargetoftheTransformingGrowthFactor(cid:4)
(TGF(cid:4))-Smadsignalingpathway(Attisanoetal.,2001), HeartDefectsinFoxh1NullEmbryos
thissuggeststhatTGF(cid:4)-likesignalsmayalsofunction Foxh1isexpressedthroughoutthegastrulatingembryo,
toregulateheartmorphogenesis. includingtheregioncorresponding totheAHF,andat
TGF(cid:4)signalsthroughheteromericcomplexesoftrans- later stages of development it becomes increasingly
membraneSer/Thrkinasereceptorsthatphosphorylate restrictedtotheheart(SupplementalFigureS1[http://
receptor-regulated (R)-Smad2 and R-Smad3. This in- www.developmentalcell.com/cgi/content/full/7/3/331/
ducesassemblyofaR-Smad-Smad4complexandnu- DC1]andWeisbergetal.,1998).Inthecourseofanalyz-
clearaccumulationoftheheteromericSmadcomplex.In ing Foxh1(cid:1)/(cid:1) mutant embryos (Hoodless et al., 2001),
the nucleus, R-Smad2 and R-Smad3 interact with DNA weobservedpericardialedema(datanotshown),sug-
bindingproteins,suchasFoxh1,whileSmad4stabilizes gestingadefectinheartmorphogenesis.Therefore,we
DNAbindingbytheternarycomplexandrecruitscoacti- investigatedthegrossmorphologyofFoxh1(cid:1)/(cid:1)mutant
vators.SinceFoxh1onlybindsR-Smad2andR-Smad3, heartsbyscanningelectronmicroscopy.Theinflowre-
itfunctionsspecificallytotransduceTGF(cid:4)-likesignals gionofthemutanthearts,includingthedevelopingleft
intotranscriptionalresponses(Labbeetal.,1998).Ge- andrightatrialappendages,appearedsimilartothatof
netic analyses in zebrafish and mouse have shown a wild-type embryos (Figure 1A). However, the ventricle
key role for Foxh1 in mediating anterior-posterior pat- and OFT of the Foxh1(cid:1)/(cid:1) mutant hearts were disorga-
terningbynodal-likesignalsduringgastrulation(Whit- nized,particularlyatlaterdevelopmentalstagesaround
man, 2001). In zebrafish, Foxh1 is the product of the E9.0.Analysisofloopingstage(LS)developmentinType
schmalspur gene and mutations in the region of this IFoxh1(cid:1)/(cid:1)mutants(Hoodlessetal.,2001)showedthat
geneencodingtheforkheadDNAbindingdomainlead heartsin46%ofmutantsdevelopedtoLS-I/II,8%devel-
tolossofthegastrulaorganizerandsubsequentlossof opedtoLS-IIIwithloopingalwaysoccurringtotheright,
axialmidlinestructures(Whitman,2001).Inthemouse, and the remaining 46% arrested at LS-0 (Figure 1B).
deletion of Foxh1 leads to constrictions at the embry- Next, we analyzed expression of the gene encoding
onic-extraembryonic boundary with lack of anterior cardiacmyosinbindingproteinC(Mybpc3),whichisa
specification, primitive streak elongation, and subse- general cardiomyocyte marker (Bruneau et al., 2001).
quent failure of gastrulation (Type II and III mutants) Comparisontowild-typeembryosrevealedthatoverall
(Hoodlessetal.,2001;Yamamotoetal.,2001).Thisoc- expression inmutant embryoswas notgrossly abnor-
curswithvariablepenetrance,andinlesssevereTypeI mal,indicatingnomajordefectincardiomyocytediffer-
mutants,defectsinspecificationoftheanteriorprimitive entiation(Figure1C).However,serialsectionsrevealed
streak thatare analogousto thezebrafish schmalspur a disorganized myocardium, poor trabeculation in the
mutantsareobserved,withsubsequentlossofmidline ventricularregion,andaverypoorlyformedOFT(Figure
structures(Hoodlessetal.,2001;Yamamotoetal.,2001). 1C, vi–viii). Further, posterior development appeared
Themolecularmechanismsthatcontrolformationof normal,asclearlydefinedrightandleftatriawereevi-
the AHF in the mouse are unknown. Here we demon- dentinthemutants(Figure1C,vi).Analysisofexpres-
stratethat Foxh1is requiredfor formationofthe AHF. sionofthepan-cardiomyocytemarker,(cid:5)cardiacactin
Accordingly, Foxh1(cid:1)/(cid:1) mutant embryos exhibit loss of (Actc1), showed similar results (data not shown). To
Fgf8-andFgf10-positiveAHFcellsandseveredefectsin furtherexplorethemorphologyofthemutanthearts,we
RVandOFTformationthatcloselyresemblethedefects employed optical projection tomography (OPT), which
seen in Mef2c(cid:1)/(cid:1) mutant hearts. Indeed, we demon- allows for three-dimensional imaging of fluorescently
stratethatMef2cexpressionintheanteriorregionofthe stainedembryos(Sharpeetal.,2002). Instrikingcontrast
heartisdependentonFoxh1andidentifyacomposite to wild-type embryos, Foxh1 mutants costained for
Foxh1-Nkx2-5 binding site within the Mef2c gene that Mybpc3andActctohighlightallofthecardiomyocytes
is regulated by TGF(cid:4) signaling in a Smad-dependent displayed a poorly formed OFT that ended in a blind
manner. Furthermore, analysis of a lacZ-transgene sac(SupplementalMovieS1).Next,weanalyzedmyosin
drivenbyanintronicenhancerfragmentofMef2char- lightchain2v(Myl2),anearlymarkeroftheventricular
boringthiselementshowsFoxh1-dependentexpression region (Christoffels et al., 2000). Myl2 showed robust
inthepresumptiveAHFanditsderivatives,theRVand expressioninthefutureventricleregionoftheabnormal
OFT,whereendogenousMef2cisexpressed.Thesere- mutant hearts (Figure 1D), indicating that they had at-
sultsthusdemonstratethataTGF(cid:4)-likeSmadsignaling tainedventricularidentity.However,asalsohighlighted
(B)Schematic(top)ofearlystagesofcardiacloopingmorphogenesisinmouseembryosandsummary(bottom)ofthegrossmorphologyand
observedLSofheartsinTypeIFoxh1(cid:1)/(cid:1)mutants(percentof140TypeIFoxh1(cid:1)/(cid:1)mutantsanalyzed)(Hoodlessetal.,2001;Yamamotoet
al.,2001).
(CandD)Whole-mountinsitustainingforthepancardiomyocytemarker,Mybpc3(C)orthepanventriclemarker,Myl2(D)inE8.5–E8.75wild-
type((cid:2)/(cid:2),i–iv)andFoxh1(cid:1)/(cid:1)mutant((cid:1)/(cid:1),v–viii)embryos.Correspondingsectionsfromeachembryoareshownprogressivelyfromposterior
(iiandvi)toanterior(ivandviii),asindicatedinthewhole-mountpanel(iandv).Theblackarrowheadmarkstheleft/rightventricularboundary
inthesectionsofthewild-typeembryos(ii)andthewhite/blackarrowheadtheputativeleftventricularsegmentinthemutantembryos(v).
Abbreviations:LS,loopingstage;SV,sinusvenosus;A,atria;RA,rightatrium;LA,leftatrium;AVC,atrioventricularcanal;LV,leftventricle;
RV,rightventricle;V,ventricle;OFT,outflowtract;r,right;l,left;a,anterior,p,posterior;d,dorsal;v,ventral.
DevelopmentalCell
334
by Mybpc3 expression, the ventricles of the mutants Foxh1mutants(Figures2Cand2D).Moreover,inchime-
were disorganized and there was an absence of well- ric embryos comprised of high contributions of Foxh1
formed aortic arch arteries in the region of the aortic mutantEScellsinawild-typebackground,heartsdis-
sac(Figures1Cand1D,viii).WealsoexaminedHey2, played a rudimentary phenotype, despite the strong
whichisalsoexpressedintheventricleregion,andob- contributionofwild-typecellstotheendoderm(Hood-
servedexpressionthroughoutthehearttube(datanot lessetal.,2001).Finally,weexaminedEScellsthatwere
shown).TypeICmutants,whicharrestatLS-0anddis- induced to form beating foci of cardiomyocytes after
play severe anterior truncations, revealed similar ex- differentiationintoembryoidbodies(EBs)(Supplemen-
pression of these markers (data not shown). Thus, talFigureS2D)(Boheleretal.,2002).AsshowninSup-
Foxh1(cid:1)/(cid:1)mutantshaveacommonventricleandforma plemental Figure S2E, EBs comprised of Foxh1(cid:1)/(cid:1) ES
rudimentaryhearttube. cells(FKO3)werecompletelydefectiveintheirabilityto
formbeatingfoci,andre-expressionofaFoxh1trans-
RVandOFTDefectsinFoxh1(cid:1)/(cid:1)Mutants generestoredthiscapabilityinadose-dependentman-
TounderstandthemolecularbasisfortheFoxh1(cid:1)/(cid:1)mu- ner.Incontrast,Foxa2(cid:1)/(cid:1)ESlinesformedbeatingfoci
tantheartphenotype,weexaminedmarkersofchamber atafrequencycomparabletowild-typeEScells.Alto-
specification. GATA binding protein 4 and 6 (Gata4, gether,thesedataindicate thatthepatterningdefects
observedinFoxh1mutantheartsarenotsecondaryto
Gata6),which areexpressed inendoderm andcardio-
loss of anterior mesendoderm structures, particularly
myocytesintheposterioraspectoftheheart(Parmacek
defectsindefinitiveendoderm.
andLeiden,1999),werestronglyexpressedinthemu-
tants (Figure 2A), as was Myh6 (cardiac myosin heavy
chain(cid:5)),whichisnormallyexpressedintheinflowtract DefectiveAnteriorHeartFieldinFoxh1(cid:1)/(cid:1)Mutants
TheAHF isa keycontributor ofcardiogenic precursor
and atria at this stage (Figure 2A). Next, we examined
cells to the RV-OFT of mammalian hearts (Kelly and
the predominantly LV marker Hand1 (formerly eHand)
Buckingham, 2002). Fgf8 and Fgf10 gene transcripts
(Brand, 2003; Olson and Schneider, 2003), which dis-
marktheAHFandarealsoexpressedinthepharyngeal
played robust expression in the prospective LV of the
mesodermthatiscontiguouswiththisstructureinante-
mutants(Figure2B).Instrikingcontrast,expressionof
rior regions (Kelly et al., 2001). Therefore, to explore
Smpx(Chisel)andNppa(AtrialNatriureticFactor),which
whethertheFoxh1(cid:1)/(cid:1)mutantphenotypereflectsdefects
are normally expressed in the outer curvature of the
in the AHF, we examined Fgf8 and Fgf10 expression
left and right ventricles, as well as the forming atrial
in Foxh1(cid:1)/(cid:1) mutant embryos. At E8.0, Fgf8 and Fgf10
appendages(Christoffelsetal.,2000),werecompletely
expression appears to mark a distinct population of
absent in the ventricular region of mutant hearts, al-
splanchnicmesodermalcellsthatliesmedialtothepri-
thoughstillpresentinthefutureatrialregion(Figure2C).
maryheartfieldandwhichprogressivelycomestooc-
Nppa expressionis regulated byTbx5 (Bruneauet al.,
cupyapositiondorsalandanteriortotheprimaryheart
2001).Therefore,weexaminedTbx5expression,which
wasfoundtobenormalintheinflowtractofFoxh1(cid:1)/(cid:1) field,fromwhereitconvergesatthemidlineandenters
thearterialpoleofthedevelopingheart.Inmutantsat
mutant hearts, but was undetectable in the LV (Figure
thisstage,posteriorexpressionofbothFgf8andFgf10
2C). Finally,we examinedexpression ofHand2, which
wasnormal;however,therewasamarkedreductionin
afterloopingbecomesrestrictedtotheRV-OFT(Thomas
expressionintheanteriorregion(Figure3A),whichwas
etal.,1998)andregulatesNppadirectly(Thattaliyathet
mostevidentatE8.25whenanteriorexpressionofFgf8
al.,2002),aswellasWnt11,whichmarkstheOFT(Cai
was absent in the mutants, despite normal posterior
et al., 2003). NeitherHand2 nor Wnt11 was expressed expression.Furthermore,atlaterstagesFoxh1(cid:1)/(cid:1)mu-
at significant levels in stage-matched mutants (Figure
tants showed little detectable expression of Fgf8 and
2D). Together, these data indicate that while posterior
Fgf10 in the heart field, whereas expression of these
markersofheartpatterningarenotseverelyaffectedby
markersinthesurroundingpharyngealmesodermwas
loss of Foxh1, markers of the RV-OFT are absent and
readily detectable (Figure 3A). Next we examined Isl1,
there are defects in patterning the outer curvature of
whichisrequiredfordevelopmentofthemedialsplanch-
the LV. Of note, the RV-OFT defects coupled with the
nic mesoderm from which FGF-expressing AHF cells
lossofNppaandHand2expressionarehighlyreminis- arise. Consistent with an early role in specifying the
centofthenullphenotypeofMef2c(Linetal.,1997). secondary heart field, Isl1-expressing cells have been
Mutation of Foxh1 causes defects in specifying the foundtocontributetobothposteriorandanteriorseg-
anteriorprimitivestreakduringgastrulation,whichleads mentsofthedevelopingheart,andinIsl1mutantsBMP
to defects in anterior mesendoderm formation (Hood- expression is downregulated in addition to FGFs. In
lessetal.,2001).Similardefectsalsooccurinembryonic Foxh1mutants,Isl1expressionwassimilartowild-type
mutants of the forkhead protein, Foxa2 (Dufort et al., embryos at both E8.0 and E8.5 (Figure 3B) and BMP2
1998),andindeedFoxa2expressionintheanteriorprimi- andBMP7expressionwasunaffected(datanotshown).
tivestreakisdependentonFoxh1(Hoodlessetal.,2001). Together, these results suggest that Foxh1 is not re-
Therefore,toexplorewhetherthepatterningdefectsin quired for development of the Isl1-positive medial
Foxh1mutantheartsmightbecausedbyafailureinmes- splanchnicmesodermperse,butisrequiredforforma-
endodermstructures,weexaminedpatterninginFoxa2 tion of the AHF, which arises from this population of
embryonic mutant hearts. Although the anterior mor- cells.
phologyofFoxa2embryonicmutantembryoswasdis-
turbed,expressionofSmpx,Nppa,andWnt11wasde- Mef2cExpressionIsDependentonFoxh1
tectedintheheartatintensitiescomparabletowild-type Chamber specification in the vertebrate heart is con-
embryos (Supplemental Figure S2A–S2C), contrasting trolledbyahierarchicaltranscriptionfactornetworkthat
Foxh1IsRequiredforFormationoftheAHF
335
Figure2. ChamberSpecificationinFoxh1(cid:1)/(cid:1)MutantEmbryos
EmbryosbetweenE8.5andE9.0werehybridizedwiththeindicateddigoxigenin-labeledriboprobes.Onlytheheartregion,withpericardium
removed,isshownforeachstage-matchedembryo.TheanteriorandposteriorboundariesoftheprospectiveleftventricleintheFoxh1(cid:1)/(cid:1)
mutantheartsaremarked(arrowheads).
(A)TheinflowtractmarkersGata4,Gata6,andMyh6areexpressedinFoxh1(cid:1)/(cid:1)mutanthearts.
(B)Hand1,whichmarkstheprospectiveLVregionisexpressedintheventricularregionofFoxh1(cid:1)/(cid:1)mutanthearts.
(C)SmpxandNppa,whichmarktheoutercurvatureofthedevelopingventriclesandTbx5intheLV,areseverelydownregulatedinFoxh1(cid:1)/(cid:1)
mutanthearts.NotethatNppa,Smpx,andTbx5expressiondomainsintheinflowtractareundisturbedintheFoxh1(cid:1)/(cid:1)mutants.
(D)Hand2andWnt11,whichmarktheOFT-RVandOFT,respectively,arenotexpressedinFoxh1(cid:1)/(cid:1)mutantheartsatE9.0.
includes Nkx2-5 and Mef2c (Brand, 2003; Cripps and Buckingham, 2002), causes RV-OFT defects that are
Olson,2002).Nkx2-5,whichisimportantformyocardial similar to the Foxh1(cid:1)/(cid:1) heart phenotype. In particular,
differentiationinvitroandheartmorphogenesisinvivo hearts of both mutants fail to express Nppa in their
(OlsonandSchneider,2003),isalsorequiredfordeploy- prospective ventricular regions and Hand2 in the RV-
ment of AHF cells, and its expression there involves OFT(Linetal.,1997)(Figures2Cand2D).Thisprompted
distinctregulatoryelements(SchwartzandOlson,1999). ustoexamineNkx2-5andMef2c,whichareatthetopof
Furthermore,anullmutationinMef2c,whichhasbeen thishierarchicaltranscriptionfactornetwork.Strikingly,
proposedtoreflectdefectsinAHFformation(Kellyand while Nkx2-5 expression in the mutants was normal,
DevelopmentalCell
336
Foxh1IsRequiredforFormationoftheAHF
337
Mef2cexpressionintheventriclesandOFTwasabsent, complex on DNA (Labbe et al., 1998). Therefore, we
althoughinflowtractexpressionofMef2cwasstillevi- overexpressedR-Smad2andSmad4inHepG2cellsand
dent(Figure3C).Toconfirmthis,weperformedquantita- evaluated Fi3A activation. R-Smad2 and Smad4 had
tivereal-timeRT-PCRonRNAextractedfromE8.5em- minimaleffectswhenexpressedwithFoxh1alone,but
bryos using primers for the cardiac-specific transcript whenNkx2-5wascoexpressed,theSmadssignificantly
ofMef2c.InRNAisolatedfromFoxh1(cid:2)/(cid:1)embryos,Mef2c increased activation of the enhancer (Figure 4C). Fur-
expressionwasdecreasedto50%ofwild-type,whereas thermore,expressionofaFoxh1mutant(Foxh1-SIMm),
in Foxh1(cid:1)/(cid:1) mutants the levels were less than 20% of whichisdefectiveinSmadbinding(Germainetal.,2000),
thecontrols(Figure 3D).Wealsoexamined Mef2bex- abolishedallTGF(cid:4)responsiveness,asdidaFoxh1mu-
pression,whichisupregulatedinMef2c(cid:1)/(cid:1)mutants(Lin tant (Foxh1-FKHDm) that is defective in its forkhead
etal.,1997),andfoundthatitwasincreasedby2.5-fold DNAbindingdomain(Figure4C).Wealsoobservedthat
inFoxh1(cid:1)/(cid:1)mutants(Figure3D).Thesedatashowthat Fi3AwasunresponsivetoBMP2stimulation(Figure4C),
Foxh1iscriticalfordevelopmentoftheAHFandforma- consistentwiththespecificfunctionofFoxh1inTGF(cid:4)-
tion of the RV-OFT inpart by regulating expression of likesignalingpathways(Labbeetal.,1998).Therefore,
Mef2c, a key component of the cardiac transcription TGF(cid:4)-dependent activation of Fi3A is dependent on
factorgenenetwork. boththeSmadbindingandforkheadDNAbindingactiv-
ityofFoxh1.
TheFi3AelementcontainstwopotentialFoxh1bind-
Foxh1InteractswithNkx2-5toRegulate
ingsites(Figure4D).Mutationoftheupstreamsite(Fi3A-
aTGF(cid:1)-ResponsiveElement
F1m)hadnoeffectonFi3Aactivation,whereasmutation
intheMef2cGene ofthedownstreamsite(Fi3A-F2m)abolishedallTGF(cid:4)-
To determine whether Foxh1 might directly regulate
responsiveness (Figure 4E). We noted that this Foxh1
Mef2c,wesearchedforFoxh1bindingsitesintheMef2c
site is situated two base pairs upstream of an Nkx2-5
geneandfoundfivecandidates,twoofwhicharefound
consensusbindingsiteandhasadjacentSmadbinding
closetoeachotherwithinFi3A(Figure4A).Wecloned
elements(Figure4D),suggestingthatFoxh1andNkx2-5
each of these regions into a heterologous luciferase
synergizethroughthiselement.Totestthis,wegener-
reporter plasmid carrying the E1b basal promoter and
ated a 132 bp element that incorporated the Foxh1,
testedthemforTGF(cid:4)-andFoxh1-responsivenessusing
Nkx2-5, and Smad binding elements (FNE, for Foxh1
HepG2cells,whichdonotexpressFoxh1(Labbeetal.,
and Nkx2-5 element; Figure 4D). Analysis of the wild-
1998).Ofthefourelementstested,onlyFi3Awasfound typeFNErevealedthatitrespondedtoTGF(cid:4)inaFoxh1-
to be TGF(cid:4)-responsive in a Foxh1-dependent manner
and Nkx2-5-dependent manner (Figure 4F), whereas
(Figure4B),althoughtheoverallactivitywaslow.Mef2c mutation of the Foxh1 and Nkx2-5 sites (FmNE and
expression is also dependent on Nkx2-5 (Cripps and FNmE,respectively;Figure4D)abolishedresponsiveness
Olson, 2002), so we also tested these elements in the (Figure4F).Thus,activationoftheTGF(cid:4)-responsiveFNE
presence of Nkx2-5 expression. Nkx2-5 induced Fi3A within intron 3 of Mef2c is dependent upon adjacent
activity but did not allow for TGF(cid:4)-responsiveness on Foxh1andNkx2-5bindingsites.
itsown.However,whenNkx2-5wascoexpressedwith To characterize binding ofFoxh1 and Nkx2-5 to the
Foxh1, the two synergistically activated Fi3A and in- FNE,weemployedaDNAimmunoprecipitationapproach
creased TGF(cid:4)-dependent activation of the element to usingCOS1cellsexpressingNkx2-5orFoxh1(seeEx-
80-foldoverthecontrol.Ofnote,Fi9alsoshowedNkx2-5 perimentalProcedures).IneitherFoxh1orNkx2-5immu-
responsivenessbutitneithersynergizedwithFoxh1nor noprecipitates,wereadilydetectedtheboundFNE(Fig-
displayedTGF(cid:4)-dependentregulation,whereastheother ure 5A), and mutation of the Foxh1 forkhead domain
two elements displayed no response to any treatment abolishedFoxh1binding(datanotshown).Wealsocon-
(Figure 4B). The Fi3A construct was therefore chosen firmed by electrophoretic mobility shift assays that
forfurtherexamination. Foxh1andNkx2-5boundtoFNE(datanotshown).To
Foxh1 interacts with R-Smad2-Smad4 complexes demonstratethatFoxh1occupiestheendogenousMef2c
that are in the TGF(cid:4) pathway, to form a Foxh1-Smad FNE during cardiomyocyte differentiation, we employed
Figure3. Foxh1IsRequiredfortheAnteriorHeartFieldandMef2cExpression
(A)DefectsintheanteriorheartfieldofFoxh1(cid:1)/(cid:1)mutants.Wild-type((cid:2)/(cid:2))andFoxh1nullmutant((cid:1)/(cid:1))embryosatE8.0andE8.25–E8.5
(indicated)wereanalyzedbywhole-mountinsituforFgf10(leftpanels)andFgf8(rightpanels)expression.Attheearlierstage,bothFgf10
andFgf8areexpressedinFoxh1(cid:1)/(cid:1)mutantsinthesplanchnicmesodermthatliesmedialtothecardiaccrescent,whereasatlaterstages
anteriorexpressionisdecreasedrelativetocontrols.Thisismostprominentinmutantheartsatthelinearhearttubestage(E8.25).Atthe
earlyloopingstage,expressionintheOFT-RVofmutantheartsisabsent,whereasexpressioninthesurroundingmesodermisevident.
(B)AnalysisofIsl1expression.VentralviewsofE8.0(left)andE8.5(middle)wild-type((cid:2)/(cid:2))andFoxh1nullmutant((cid:1)/(cid:1))embryosanalyzed
forIsl1expressionwitharightlateralviewoftheE8.5embryoalsoshown(right).NotethatIsl1expressioninthemutantsiscomparableto
wild-typepatterns.
(C)AnalysisofNkx2-5andMef2cexpression.Leftlateral(Nkx2-5)andrightlateral(Mef2c)viewsofwild-type((cid:2)/(cid:2))andFoxh1nullmutant
((cid:1)/(cid:1))embryosanalyzedforNkx2-5andMef2cexpressionbywhole-mountinsitu.NotethatwhileNkx2-5isexpressedthroughoutmutant
hearts,Mef2cexpressionisabsentinthepoorlyformedOFTandRV,butnotintheposteriordomain.Arrowheadsmarkanteriorandposterior
boundariesoftheLVinmutantembryos.
(D)AnalysisofMef2familymemberexpression.RNApurifiedfromE8.5wild-type,Foxh1(cid:2)/(cid:1),andFoxh1(cid:1)/(cid:1)embryos((cid:2)/(cid:2),(cid:2)/(cid:1),(cid:1)/(cid:1),respectively)
wasanalyzedbyreal-timePCRusingprimersspecificforMef2borthecardiactranscriptofMef2c.Resultsareplottedrelativetothewild-
typecontrol(mean(cid:6)SD)foreachgene.
DevelopmentalCell
338
Figure4. Nkx2-5andFoxh1SynergizetoRegulateTGF(cid:4)ResponsivenessofanIntronEnhancerintheMef2cGene
(A)SchematicrepresentationoftheMef2cgene.Numberedboxesdepictexonsofuntranslatedandtranslatedregionsinwhiteandblack,
respectively(Wangetal.,2001).Openrightanglearrowsindicatetissue-specifictranscriptionstartsites,andclosedarrowsshowtherelative
positionsofFoxh1sitesintheintrons(Fi3A,Fi3B,Fi7,Fi9).
(B,C,E,andF)Luciferasereporterassays.HepG2cellsweretransientlytransfectedwiththeindicatedMef2c-Lucreporterconstructswith
((cid:2))andwithout((cid:1))Foxh1,Nkx2-5,Smad2,orSmad4expressionconstructs.Luciferaseactivityinlysatesfromcellsincubatedovernightin
the absence (white bars) or presence of 100 pM TGF(cid:4) (black bars) or 1 nM BMP2 (hatched bars) was determined and normalized to
(cid:4)-galactosidaseactivity.Dataarepresentedasfoldactivationoverbasalpromoteractivity(relativeactivity),andareexpressedasmean(cid:6)
SDoftriplicatesfromarepresentativeexperiment.(C)ActivityofMef2c-Fi3A-Luc.Mef2c-Fi3A-Lucactivitywasassayedinthepresenceof
wild-typeFoxh1((cid:2)),theFoxh1SIMmmutant(A),theFoxh1FKHDmmutant(R),andtheindicatedcombinationsofNkx2-5,Smad2,andSmad4.
(D)Schematicrepresentationofthe906bpFi3Aelementanditsderivatives.Smad(SBE,green),Foxh1(F1orF2,red),andNxk2-5(N,blue)
sitesandtheircorrespondingmutants(F2mandNm)areindicated.NumberingisaccordingtoGenBank#NT_039589.
(E)ActivityofFi3Aandderivativescontainingmutationsineitherthefirst(Fi3A-F1m)orsecond(Fi3A-F2m)Foxh1sites.(F)Characterization
ofFNE.AminimalTGF(cid:4)-responsiveelement(FNE,forFoxh1andNkx2-5element)wasidentifiedinFi3AandFoxh1(FmNE)andNkx2-5(FNmE)
sitemutantstestedasindicated.
Foxh1IsRequiredforFormationoftheAHF
339
chromatinimmunoprecipitation(ChIP)usingEScellsdif- nonresponsiveFoxh1mutant,Fi3A-F2m,werelinkedto
ferentiated into cardiomyocytes (Supplemental Figure the hsp68 promoter to drive expression of lacZ. After
S2D) (Boheler et al., 2002). For this, we subjected generatingstabletransfectantsinEScells,wefirstchar-
Foxh1(cid:1)/(cid:1)EScellsreconstitutedwithFlag-taggedFoxh1 acterizedtransgeneactivityinvitrousingEScellsdiffer-
(clonePF13)orthecontrolR1wild-typeEScelllineto entiatedintoEBs(seeSupplementalFigureS2Dforfur-
invitroEBdifferentiation(SupplementalFigureS2E)and therdetails).AfterplatingEBsongelatin-coateddishes,
performed ChIP. As shown in Figure 5B, the region of beatingfociweremarkedandthecellsthenprocessed
theMef2cgenecorrespondingtotheFNEwasreadily to detect lacZ activity. In ES cells harboring the wild-
apparentinFoxh1,butnotincontrolnordifferentiated typeFi3A-hsp68LacZpAtransgene(cloneS1-2),100%
R1 ES cell immunoprecipitates. Furthermore, analysis of beating foci stained for lacZ with minimal staining
of Flag-Foxh1 immunoprecipitates using primers lo- observedinnonbeatingcells(Figure6A).Wealsoana-
cated3.2kbupstreamoftheFNEyieldednoproduct, lyzed a second ES line harboring the wild-type Fi3A-
confirming specific Foxh1 binding to the sheared FNE hsp68LacZpA transgene (clone S1-6) and observed
fragment(datanotshown). TodetermineifFoxh1and similar expression in beating foci (data not shown). In
Nkx2-5 co-occupy the FNE, we next employed a two- contrast, none of the ES cell lines with the Fi3A-F2m-
stepDNAimmunoprecipitationprocedure(Benchabane hsp68LacZpA mutant transgene displayed detectable
and Wrana, 2003). Forthis, protein-DNA complexes in lacZexpression(Figure6Aanddatanotshown).Next,
either Foxh1 or Nkx2-5 immunoprecipitates were iso- we analyzed these lines using tetraploid aggregation,
lated,eluted,andthensubjectedtoasecondimmuno- which leads to 100% contribution of ES cells to the
precipitation using the opposite antibody. In cells ex- embryo (Nagy et al., 2003). Analysis of tetraploids de-
pressing either Foxh1 or Nkx2-5 alone, no FNE was rivedfromcloneS1-2,whichharborsthewild-typeFi3A-
detected in the double immunoprecipitate, whereas hsp68LacZpAtransgene,revealedstaininginthedevel-
PCRproductwasreadilydetectedincellscoexpressing opingmyocardiumthatextendedanterioranddorsalto
Foxh1andNkx2-5(Figure5C).Thus,Foxh1andNkx2-5 the developing heart tube at E8.25 (Figure 6B), similar
canco-occupytheMef2cFNE. toexpressionoftheAHFmarkers,Fgf8andFgf10(Fig-
Co-occupation of the FNE suggested the possibility ures 3A and 6B). At E9.0, transgene expression was
thatFoxh1andNkx2-5physicallyassociate.Toinvesti- observed in the RV and OFT in the same domain as
gatethis,wetestedFoxh1immunoprecipitatesforthe Mef2c and Fgf10 (Figures 6C and 6D) and in the right
presenceofNkx2-5andobservedthatNkx2-5coprecip- sinus venosus (RSV) where Mef2c is also strongly ex-
itated with Foxh1 (Figure 5D). Bacterially expressed pressed. Analysis of sections derived from these em-
Nkx2-5alsointeractedwithFoxh1,indicatingtheinter- bryosfurtherrevealedtransgeneexpressioninapopula-
actionisdirect.MutationoftheFoxh1forkheaddomain
tionofpharyngealandsplanchnicmesodermcellsthat
(FKHDm)abrogatedthisinteraction,whereastheSmad
corresponds tothe prospective AHF,as well asin the
bindingdomainmutants(SIMmand(cid:7)SID)hadnoeffect
myocardiallayeroftheRV,OFT,andRSV(Figure6D).
(Figure5E).WenextexaminedNkx2-5mutantsandob-
Forthistransgenicline,wealsoobservedectopicstain-
served reduced binding upon removal of the first 136
inginthevasculature.Incontrast,therewasnodetect-
amino acids of Nkx2-5 and undetectable association
ablestainingintetraploidsderivedfromEScellsharbor-
upon deletion of the homeodomain ((cid:7)NHD and (cid:7)HD).
ing the Foxh1 mutant element (Figure 6C). We also
ThehomeodomainalonewasalsoabletobindFoxh1to
examinedtetraploidembryosderivedfromoursecond
somedegree(Figure5F).Therefore,theamino-terminal-
Fi3A-WT ES line (S1-6), which also displayed staining
homeodomainregionofNkx2-5isimportantforefficient
inbeatingfoci.Althoughoverallstainingwasweakerin
bindingtoFoxh1.Todetermineifthephysicalinteraction
tetraploidembryosderivedfromthisline,similarstaining
between Nkx2-5 and Foxh1 is important for activation
in the AHF and its derivatives was observed (data not
oftheFNE,wefocusedonNkx2-5((cid:7)N)whichdecreased
shown). These results point to a key role for Foxh1 in
Foxh1bindingbutdoesnotinterferewithNkx2-5DNA
specifying the AHF during cardiac morphogenesis in
bindingactivity(datanotshown).Whiletheincreasein
partbysynergisticallyinteractingwithNkx2-5toregu-
FNEactivityinducedbyexpressingNkx2-5onitsown
lateaTGF(cid:4)-andSmad-responsiveelementintheMef2c
wasunaffectedbythisdeletion,synergismwithFoxh1
gene(Figure7).
was significantly decreased (Figure 5G). The residual
co-operativity likely reflects retention of some Foxh1
bindingactivitymediatedbythehomeodomain.Incon- Discussion
trast, deletion of the homeodomain abolished both
Nkx2-5-dependentactivationoftheFNEandsynergism Foxh1IsRequiredforFormationoftheAHF
withFoxh1,consistentwiththenotionthatthehomeo- Cell marking studies and the analysis of a gene trap
domainofNkx2-5isalsocriticalfortheinteraction,as in the mouse Fgf10 gene have shown that the major
demonstrated in vitro (Figure 5G). Collectively, these contribution of cardiogenic cellsto the arterial pole of
dataindicatethatFoxh1andNkx2-5physicallyinteract the heart occurs from the newly described AHF (Kelly
andcooperatetomediatesynergisticTGF(cid:4)-andSmad- andBuckingham,2002).TheAHFarisesfromsplanchnic
dependent activation of an FNE present in the Mef2c mesodermalcellsthatliemedialtothecardiaccrescent
gene. and which later extend dorsally and anteriorly to the
arterial pole of the developing heart. In addition, cells
TheFNEDirectsTransgeneExpression from the pharyngeal mesoderm contribute via the
intheAHFandRV-OFT splanchnicmesodermtoexpansionofthearterialpole
TodetermineiftheFNEisactiveinvivo,wegenerated togiverisetotheRVandOFT(KellyandBuckingham,
two transgenes in which either wild-type Fi3A or the 2002).Thispopulationofsplanchnicmesodermmakes
DevelopmentalCell
340
Figure5. PhysicalandFunctionalInteractionsbetweenFoxh1andNkx2-5ontheFNE
(A)BothNkx2-5andFoxh1occupytheFNE.COS1cellsweretransfectedwithcDNAencodingeitherT7-taggedNkx2-5orFlag-taggedFoxh1
alongwiththepGL3-FNEreporterconstruct.Protein-DNAcomplexesfromtransfectedcellswereimmunoprecipitatedwitheitheranti-T7
((cid:5)-T7)antibodyoranti-Flag((cid:5)-Flag).TotalinputDNAandDNArecoveredfromtheimmunoprecipitateswereanalyzedbyPCR.Expression
ofNkx2-5andFoxh1wasconfirmedbyimmunoblotting(IB)using(cid:5)-T7and(cid:5)-Flagmonoclonalantibodies,respectively.
(B)Foxh1occupiestheendogenousFNE.EmbryoidbodiesobtainedfromFoxh1(cid:1)/(cid:1)EScellsreconstitutedwithFlag-Foxh1(PF13)orwild-
typeEScells(R1)weresubjectedtoinvitrodifferentiationandaChIPassayperformedusingeither(cid:5)-Flagantibodytoimmunoprecipitate
Flag-Foxh1oracontrolantibody.Afterremovalofthecrosslinks,immunoprecipitatedDNAwasamplifiedbyPCRusingFNE-specificprimers.
ExpressionofFlag-Foxh1wasanalyzedbyimmunoblotting(IB)totallysates(lowerpanel).
(C)Foxh1andNkx2-5co-occupytheFNE.COS1cellsweretransfectedwithT7-Nkx2-5and/orFlag-Foxh1togetherwithpGL3-FNE.After
formaldehydecrosslinking,lysatesweresubjectedtoimmunoprecipitationwiththeindicatedantibody(1stIP),elutedwith1%SDSandthen
re-immunoprecipitatedwith(cid:5)-T7or(cid:5)-Flagantibody(2ndIP).TotalinputFNEandFNEpresentinimmunoprecipitateswasanalyzedbyPCR.
TheexpressionofNkx2-5andFoxh1wasconfirmedbyimmunoblotting(IB)using(cid:5)-T7and(cid:5)-Flag,respectively.
(D) Nkx2-5 and Foxh1 interact. Lysates obtained from COS1 cells expressing Flag-Foxh1 and T7-Nkx2-5 either alone or together were
immunoprecipitated(IP)using(cid:5)-Flagantibody.ImmunoprecipitateswereseparatedbySDS-PAGEandimmunoblotted(IB)using(cid:5)-T7.The
expressionofFoxh1andNkx2-5wasconfirmedbyimmunoblottingtotalcelllysatesusing(cid:5)-Flagand(cid:5)-T7antibody.
(E)Foxh1interactswithNkx2-5throughitsforkheaddomain.TomaptheregionofFoxh1requiredforbindingtoNkx2-5,lysatesfromcells
expressingFlag-taggedwild-typeFoxh1(F-Foxh1),theSmad-interactingmotifmutantFoxh1(F-SIMm),theDNAbindingmotifmutantFoxh1