Early in vertebrate development, the cardiac lineage isamong the first specialized cell types to be established inthe embryo. Expression of nkx-2.5 (Csx-1), a homologue ofthe Drosophila gene tinman (tin), is an early indicator ofcardiogenesis (Lints et al., 1993; Komuro and Izumo, 1993;Harvey, 1996; Schultheiss et al., 1995). In Drosophila em-bryos, tin function is required for cardiac lineage develop-ment (Azpiazu and Frasch, 1993; Bodmer, 1993). In mice,Nkx-2.5 has an important role in heart development be-cause mutation of nkx-2.5 results in cardiac defects andembryonic lethality (Lyons et al., 1995; Tanaka et al., 1999).Similarly, mutations in human NKX-2.5 have been associ-ated with certain forms of congenital heart disease (Schottet al., 1998; Benson et al., 1999). In late gastrulation stageavian embryos, nkx-2.5 expression in the anterior lateralmesoderm demarcates the cardiogenic region (Schultheiss
1 To whom correspondence should be addressed. Fax: (513) 636-
et al., 1995; Ehrman and Yutzey, 1999; reviewed in Yutzeyand Kirby, 2002). Later in development, nkx-2.5 is ex-pressed in other organs, including the stomach, thyroid, andspleen. The complex temporal and spatial regulation ofnkx-2.5 is indicative of multiple regulatory mechanismscontrolling its gene expression.
Signals important for nkx-2.5 gene activation and car-diomyogenic lineage induction are active in the anteriorlateral endoderm of gastrulation stage avian embryos(Schultheiss et al., 1995, 1997; reviewed in Lough and Sugi,2000). Bone morphogenetic proteins (BMPs) are present inthe anterior lateral heart forming region of gastrulatingchicken embryos and have been implicated in early cardiaclineage induction (Schultheiss et al., 1997; Andree et al.,1998; Ehrman and Yutzey, 1999; Schlange et al., 2000).BMP-2 is expressed in the anterior lateral endoderm and issufficient to activate nkx-2.5 gene expression in the ante-rior cardiogenic mesoderm (Schultheiss et al., 1997; Andreeet al., 1998). Nkx-2.5 gene expression is also regulated byBMP signaling in the viscera. In the developing avian
stomach and gizzard, treatment with BMP-4 or ectopic5958. E-mail: [email protected].
0012-1606/02 $35.00
243
expression of a constitutively active type I BMP receptorexpands the domain of nkx-2.5 gene expression (Smith andTabin, 1999; Smith et al., 2000). Therefore, in both theheart and gut of avian embryos, nkx-2.5 gene induction issensitive to BMP signaling. Likewise, in Drosophila em-bryos, the BMP-related factor Decapentaplegic (Dpp) isrequired for tin gene expression in the cardiac and visceralmesoderm (Frasch, 1995).
Transcriptional regulation in response to BMP signalingcan be mediated by Smad proteins (reviewed in Kawabata etal., 1998; Massague and Chen, 2000). These proteins wereinitially identified in Drosophila and include Mad and
Medea, which mediate Dpp signaling. A Mad/Medea bind-ing consensus has been characterized for the Drosophila tinDpp-responsive element induced in cardiac and visceralmesoderm precursors (Xu et al., 1998). In vertebrates,Smads 1 and 5 are pathway-specific activators of BMPsignaling and bind a similar GC-rich consensus sequence(Yingling et al., 1996; Ishida et al., 2000). Other CAGASmad consensus binding sequences have also been identi-fied for Smads 3 and 4 (Dennler et al., 1998; Zawal et al.,1998). In addition to positive BMP regulatory mechanisms,there are multiple negative regulators of BMP signaling(Massague and Chen, 2000). Smad6, an inhibitory Smad, is
FIG. 1. Clustered Smad consensus sequences are present in the mouse nkx-2.5 early regulatory element. (A) SmadA contains a GC-richconsensus sequence reported for tin, vestigial, dmef2, and smad6 regulatory sequences (Xu et al., 1998; Kim et al., 1997; Nguyen and Xu,1998; Ishida et al., 2000). SmadB and SmadC contain CAGA sequences associated with Smad3/4 DNA binding for human c-jun,plasminogen activator inhibitor, cardiac ankyrin repeat protein, and �2(I)collagen TGF-� responsive regulatory sequences (Dennler et al.,1998; Wong et al., 1999; Chen et al., 2000; Zhang et al., 2000; Kanai et al., 2001). (B) Three Smad consensus sequences are present between�3059 and �3012, relative to the reported nkx-2.5 transcriptional start site (Lyons et al., 1995). This Smad consensus region was deletedto generate �Smadnkx-2.5/lacZ.
expressed in the developing heart and may repress BMPsignaling after the initiation of cardiac lineage induction(Yamada et al., 1999). Noggin and chordin also antagonizeBMP signaling and inhibit cardiogenesis and nkx-2.5 geneinduction (Schultheiss et al., 1997; Schlange et al., 2000).Thus, complex negative and positive effectors of BMPsignaling coordinate gene regulation in response to thesefactors throughout embryogenesis.
Several distinct regulatory sequences controlling mousenkx-2.5 gene expression during development have beenidentified (Searcy et al., 1998; Lien et al., 1999; Reecy et al.,1999; Tanaka et al., 1999; Schwartz and Olson, 1999). Atleast six positive and three negative regulatory regions havebeen uncovered that, together, still do not encompass theentire nkx-2.5 expression domain. From these studies, it isapparent that multiple complex mechanisms control nkx-2.5 gene expression during development. Previously, weidentified mouse nkx-2.5 regulatory sequences (�3059/�2554) sufficient for gene expression in the heart forming
region when nkx-2.5 is first expressed (Searcy et al., 1998).Expression from this regulatory element is dependent onGATA consensus binding sites (Searcy et al., 1998). How-ever it is unlikely that GATA factors alone control nkx-2.5gene expression since GATA4/5/6 expression in the heartforming region is activated at the same time or soon afternkx-2.5 (K.E.Y., unpublished observations; Schultheiss etal., 1997). Further sequence analysis revealed a cluster ofthree Smad binding consensus sequences in the nkx-2.5(�3059/�2554) regulatory element. Strikingly, one of theseis a GC-rich consensus sequence similar to the Dpp-responsive Mad binding site of Drosophila tin (Xu et al.,1998). These sequences represent a potential direct target ofBMP-mediated gene activation by Smads in the cardiogenicregion.
The importance of the Smad consensus region in nkx-2.5gene regulation was examined in transgenic mice and incultured transgenic mouse embryos. Deletion of this regionresults in early loss of nkx-2.5/lacZ expression in the
FIG. 2. The Smad consensus region is necessary for early nkx-2.5/lacZ expression in the cardiac crescent and linear heart tube. (A–C)E7.25–E8.5 �Smadnkx-2.5/lacZ transgenic embryos. (D–F) Comparable stages of �3059nkx-2.5/lacZ embryos. Arrowheads indicate thecardiac crescent (A, B, D, E) and anterior linear heart tube (C, F). Arrows in (C), (E), and (F) indicate the left lateral plate mesoderm containingstomach and spleen progenitors.
cardiac crescent, outflow tract myocardium, and visceralmesoderm. Targeted mutation of the three Smad consensussites greatly reduces or completely eliminates nkx-2.5/lacZgene activity in the cardiac crescent. In cultured transgenicembryos, expression of the �3059nkx-2.5/lacZ transgene aswell as endogenous nkx-2.5 is induced by BMP-2 treatment.However, removal of the Smad consensus region results indelayed nkx-2.5/lacZ gene activation in cultured embryos.A negative regulatory role for the Smad consensus regionwas observed in the four-chambered heart. Thus, the Smadconsensus region appears to have both positive and negativeregulatory roles in nkx-2.5 gene expression during develop-ment.
MATERIALS AND METHODS
Generation and analysis of �Smadnkx-2.5/lacZ andmuSmadnkx-2.5/lacZ transgenic mice. A PCR-based strategywas used to delete bases �3059 to �3012 from the distal end of�3059nkx-2.5/lacZ (Searcy et al., 1998). The resulting construct,�Smadnkx-2.5/lacZ, does not contain the three Smad consensussequences located between �3059 and �3012 (Dennler et al., 1998;Xu et al., 1998; Ishida et al., 2000). The oligonucleotide CCCTCT-GTTTGCTTTCTC (�3011 to �2094) and a primer directed againstthe proximal nkx-2.5 sequence were used to amplify nkx-2.5promoter sequences minus the Smad consensus region. The se-quence of the entire amplified promoter was verified and linked tothe lacZ reporter gene with SV40 polyadenylation sequences(Searcy et al., 1998). The �Smadnkx-2.5/lacZ transgene was re-leased with SpeI and NotI and used to generate transgenic mice(Searcy et al., 1998). Transgenic mice and embryos were identifiedby PCR analysis of genomic DNA with PCR primers directedagainst the lacZ transgene (Colbert et al., 1996). Staged embryoswere subjected to whole-mount X-gal staining at E7–E14 (Searcy etal., 1998). Five independent �Smadnkx-2.5/lacZ transgenic lineswere established. Three transgenic lines (Lines 21, 69, and 92)exhibited comparable �-galactosidase (�-gal) expression patternsthroughout development. �3059nkx-2.5/lacZ embryos (Line 81)were generated previously (Searcy et al., 1998).
The three Smad consensus sequences located between �3059and �3012 of �3059nkx-2.5/lacZ were mutated by using theQuikChange Site Directed Mutagenesis Kit (Stratagene). First, theoligonucleotide 5�-TCC ATC AGC TAC ATG AAG AGT ACATTC GCG CTC TA-3� was used to mutate the two CAGA Smadconsensus sequences located at �(3035–3031) and �(3024–3020),resulting in muCAGAnkx-2.5/lacZ. Mutagenesis was performedusing 5–10 ng of �3059nkx-2.5/lacZ within pBluescript with 18PCR cycles of 95°C, 30 s; 55°C, 60 s; 68°C, 18 min. The GC-richSmad consensus sequence located between �3059 and �3050 wasthen mutated by using 10–75 ng of muCAGAnkx-2.5/lacZ with 18PCR cycles of 95°C, 30 s; 55°C, 60 s; 68°C, 22 min. The oligonu-cleotide used to convert the GC-rich Smad site to a SwaI restrictionsite was 5�-CAC CGC GGT GGC ATT TAA ATT GCT CAT CCATC-3� All mutated nucleotides are shown in bold. The resultingconstruct muSmadnkx-2.5/lacZ, with all three Smad consensussequences mutated, was verified by sequencing, and the transgenewas released by SwaI/NotI digest. F0 transgenic mouse embryoswere generated and harvested on embryonic days 7.5 (n � 9transgenic) or 10.5 (n � 6). All F0 embryos were subjected towhole-mount X-gal staining, and transgenics were subsequently
identified by PCR analysis of genomic DNA with primers directedagainst the lacZ transgene.
�-Gal enzymatic assays of transgenic mouse embryos. Proteinextracts were prepared from E7–E8 embryos for analysis of �-galenzymatic activity. Embryos dissected ex utero were incubated in30–50 �l Tropix lysis buffer containing 0.5 mM DTT on ice for15–20 min. Cultured embryos were incubated for 2 days and thenremoved from the chamber slides in 25 �l lysis buffer and incu-bated for 1 h on ice. Embryos were pelleted by centrifugation, andcell lysates were analyzed for �-gal activity in duplicate by usingthe Galactostar reagent (Tropix; PE Biosystems), according to themanufacturer’s instructions. After removal of cell lysates, genomicDNA was prepared from the embryo pellets for PCR genotyping byusing oligonucleotide primers directed against the lacZ transgene(Colbert et al., 1996). Chemiluminescence of �-gal reaction prod-ucts was quantified in duplicate by using a Monolight 3010luminometer (PharMingen). Statistical significance of observeddifferences in �-gal activity was determined by Student’s t testswith P � 0.05.
Mouse embryo cultures. Gastrulation-stage FVBN mouse em-bryos (E7.25) were isolated after timed matings and placed inculture. Cultured embryos were derived from matings of homozy-gous �3059nkx-2.5/lacZ transgenic males or heterozygous�Smadnkx-2.5/lacZ males with nontransgenic FVBN females.Mice were left together overnight and the early morning of thecopulation plug was considered E0.25. After 7 days, the pregnantfemales were sacrificed by CO2 asphyxiation and embryos weredissected into warm culture medium [high glucose DMEM plus10% fetal bovine serum, 2 mM L-glutamine, and penicillin (100U/ml)/streptomycin (100 �g/ml) (pen/strep)]. Embryos were stagedaccording to Downs and Davies (1993), based on morphologicallandmarks including the primitive streak, amniotic folds, node,head process, and allantoic bud. With this timing, E7.25 embryosare at late streak (LS) to early allantoic bud (EB) stages. Embryoswere partially bisected along the lateral aspect of the cup-shapedembryo by using tungsten needles. This dissection enables thewhole embryo to lie flat on the culture matrix. The anterior andposterior regions remain adjoined by the intact primitive streak.
Dissected embryos were transferred to culture medium incollagen-coated eight-well permanox chamber slides (Nunc) andincubated for 24–48 h in high humidity at 37°C with 5% CO2. Insome cases, culture medium was supplemented with 200 ng/mlBMP-2 (R&D Systems). After 1–2 days, cultured embryos werefixed in 2% paraformaldehyde and stained with X-gal to visualize�-gal protein expression (Mikawa et al., 1991). Alternatively,embryo extracts were prepared for �-gal enzymatic assays asdescribed above. For �Smadnkx-2.5/lacZ cultures, genomic DNAwas prepared from X-gal-stained embryos and the presence of thelacZ transgene was ascertained by PCR genotyping (Colbert et al.,1996). For in situ analysis, embryos were fixed overnight in 4%paraformaldehyde/PBS, and then dehydrated through a methanol/PBS series. For immunohistochemistry, embryos were fixed for 30min in 4% paraformaldehyde/PBS and transferred to 70% ethanol.
In situ hybridization and immunohistochemistry. Whole-mount in situ hybridizations with a mouse nkx-2.5 antisense RNAprobe were performed essentially as described previously (Searcy etal., 1998; Wilkinson, 1993). Cultured embryos were hybridized insitu while affixed to chamber slides, and Proteinase K digests wereperformed for 5 min prior to prehybridization. Staining reactionswere stopped after approximately 1 h, and hybridized embryos werepostfixed in 4% paraformaldehyde prior to photography. Culturedembryos affixed to chamber slides were immunostained with
MF-20 hybridoma supernatant (Developmental Studies HybridomaBank, University of Iowa; Bader et al., 1982) and anti-�-gal poly-clonal antisera (Eppendorf - 5 Prime) as described previously(Yutzey et al., 1995). Corresponding secondary antibodies wereTexas Red-conjugated goat anti-mouse IgG and fluorescein (FITC)-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Labora-tories, Inc.).
RESULTS
Clustered Smad consensus binding sequences arepresent in mouse nkx-2.5 regulatory sequences. A targetconsensus binding sequence (GCCGnCGc) for DrosophilaMAD and Medea has been reported based on Dpp-responsive elements in tinman, dmef2, and vestigial genes(Kim et al., 1997; Xu et al., 1998; Nguyen and Xu, 1998). Inmice, a GCCGnCGC-like motif present in the smad6promoter is responsive to BMP signals mediated bySmad1/5 and binds Smad5 and Smad4 (Ishida et al., 2000).This 7-bp Smad1/5-induced sequence is present with nomismatches in the mouse nkx-2.5 early cardiac regulatoryelement (Fig. 1A). Additional Smad-responsive regulatoryelements containing the consensus CAGA are present inhuman plasminogen activator inhibitor type 1, c-jun,PDGF-B, CARP, and �2procollagen genes (Dennler et al.,1998; Zawal et al., 1998; Wong et al., 1999; Taylor andKhachigian, 2000; Chen et al., 2000; Kanai et al., 2001).Two CAGA consensus sequences are present in addition tothe distal GC-rich site between �3059 and �3012 of themouse early cardiac regulatory element. The presence ofthree potential Smad-responsive sequences within a shortstretch of DNA is characteristic of genetic elements regu-lated by Smad-dependent signaling mechanisms (Stroscheinet al., 1999; Johnson et al., 1999). The mouse nkx-2.5 Smadconsensus region sequence is highly conserved in humannkx-2.5 genomic DNA with 48/50 identical nucleotides(Turbay et al., 1996; GenBank Accession No. AC008412).Therefore, the Smad consensus region represents a poten-tial direct target for BMP-mediated induction of mousenkx-2.5 gene expression.
The Smad consensus region is required for early induc-tion of nkx-2.5 gene expression. The importance of theSmad consensus region in embryonic gene expression wasexamined in transgenic mice. The distal 50 bp containingthe Smad consensus region were deleted from �3059nkx-2.5/lacZ to produce �Smadnkx-2.5/lacZ (Fig. 1B). TheGATA sites required for gene expression from the �3059 to�2554 element remain intact (Searcy et al., 1998). Stablelines of transgenic mice were generated with �Smadnkx-2.5/lacZ in order to examine transgene expression through-out embryogenesis. �Smadnkx-2.5/lacZ expression wasmonitored in parallel with �3059nkx-2.5/lacZ by X-galstaining of transgenic embryos (Fig. 2). Expression of the�3059nkx-2.5/lacZ transgene containing the Smad consen-sus region is initiated in the cardiac crescent at the latehead fold stage (Fig. 2D). This timing corresponds to theinitiation of endogenous nkx-2.5 expression (Searcy et al.,
1998). In contrast, no �-gal expression was detected for�Smadnkx-2.5/lacZ in transgenic embryos at cardiac cres-cent stages (Figs. 2A and 2B; E7.25–E7.5). In two indepen-dent transgenic lines, expression of �Smadnkx-2.5/lacZwas first detected at E8.0 when the heart differentiates (Fig.2C). In the third transgenic line, �Smadnkx-2.5/lacZ ex-pression was not detected until after formation of theprimitive heart tube at E9.0. Thus, initiation of �Smadnkx-2.5/lacZ expression is delayed during the early stages ofheart formation. Expression of �3059nkx-2.5/lacZ is alsoinduced in the left lateral splanchnic plate mesoderm,which contains stomach and spleen progenitors (Figs. 2Eand 2F). No expression of �Smadnkx-2.5/lacZ was observedin the left lateral plate visceral mesoderm at E8–E9. How-ever, �Smadnkx-2.5/lacZ is expressed in the spleen afterE10.5 (Fig. 4A). Thus, �Smadnkx-2.5/lacZ expression in thesplenic region is delayed by approximately 3 days relative to�3059nkx-2.5/lacZ. Together, these analyses demonstratethat deletion of the Smad consensus region results in adelay in nkx-2.5 gene induction in both cardiac and splenicprogenitor populations.
�-Gal enzymatic assays were performed on protein ex-tracts of individual embryos as a more sensitive assay fortransgene expression and to compare levels of gene expres-sion in different transgenic lines (Fig. 3). Litters of�3059nkx-2.5/lacZ and �Smadnkx-2.5/lacZ embryos wereisolated at approximately E7.25 (LS-OB stages; Downs andDavies, 1993) and E8.25 (early heart tube). Protein extractsand genomic DNA were prepared for individual embryos.�-Gal expression was quantified for each embryo by using achemiluminescent reporter assay. E7.25 corresponds to theinitiation of endogenous nkx-2.5 gene expression. At thisstage, transgenic �3059nkx-2.5/lacZ embryos exhibit astatistically significant 2.5-fold increase in �-gal enzymeactivity over nontransgenic embryos. No increase in �-galexpression over nontransgenic levels was observed in threeindependent transgenic �Smadnkx-2.5/lacZ lines. AtE8.25, an approximate 50-fold increase in �-gal expressionwas observed for �3059nkx-2.5/lacZ transgenic embryosrelative to nontransgenic embryos. For all three �Smadnkx-2.5/lacZ transgenic lines, weak but significant induction ofgene expression also was observed at E8.25. These assaysconfirm the delayed activation and necessity of the Smadconsensus region for early embryonic expression from nkx-2.5 regulatory sequences.
The Smad consensus region is required for repression ofnkx-2.5 regulatory sequences in the right ventricle. Afterthe heart differentiates, �Smadnkx-2.5/lacZ expressionwas detected in the outflow tract and right ventricle of theprimitive heart (E10.5) (Fig. 4A). Similar restriction oftransgene expression to the anterior heart also was observedfor �3059nkx-2.5/lacZ (Fig. 4C). However, outflow tractexpression of �Smadnkx-2.5/lacZ appeared to be reducedrelative to �3059nkx-2.5/lacZ. This could be due to a delayin �Smadnkx-2.5/lacZ activation in the secondary heartfield that develops into the conotruncal myocardium(Waldo et al., 2001). At E13.5, expression of �3059nkx-2.5/
lacZ is restricted to the conotruncal region and rightventricular lumen of the heart (Fig. 4D; Searcy et al., 1998).Surprisingly, �Smadnkx-2.5/lacZ is expressed throughoutthe right ventricular myocardium, but expression was notdetected in the left ventricle (Fig. 4B). The expanded expres-sion in the E13.5 right ventricle suggests that regulatoryinteractions with the Smad consensus region can have bothpositive and negative roles in regulating nkx-2.5 gene ex-pression in the embryonic heart.
Smad consensus sequences are specifically required forthe early cardiac induction of nkx-2.5/lacZ gene expres-sion. Targeted mutations in the three nkx-2.5 Smad con-sensus sequences were made in order to determine therequirement for these sites in the early cardiac induction of�3059nkx-2.5/lacZ. Mutations were introduced in themost distal GC-rich site Smad consensus sequence and inthe two CAGA sequences to eliminate putative Smadrecognition sites in the Smad consensus region (Fig. 5). Theresulting muSmadnkx-2.5/lacZ sequence has no additionalmutations, and the remaining sequence of �3059nkx-2.5/lacZ remained intact. Transient transgenic F0 mouse em-bryos were generated by using muSmadnkx-2.5/lacZ and
were analyzed on E7.5 or E10.5. The embryos were firstsubjected to X-Gal staining for visible transgene expressionand were then genotyped. Of the nine transgenic embryosobtained at E7.5, six exhibited no �-gal expression and threetransgenic embryos had barely discernable staining (Fig. 5).None of the embryos exhibited the significant early cardiacexpression obvious for �3059nkx-2.5/lacZ (Fig. 2E) or en-dogenous nkx-2.5. In addition, the negligible expression ofmuSmadnkx-2.5/lacZ in 3/9 transgenic embryos appearedto be in the anterior endoderm and not in the overlyingcardiogenic mesoderm. The embryo with the strongeststaining in the anterior intestinal portal is represented inFig. 5A. Thus, targeted mutation of the Smad consensussequences inhibits early cardiac nkx-2.5/lacZ expression atE7.5.
Transgenic muSmadnkx-2.5/lacZ F0 mouse embryos alsowere analyzed at E10.5. muSmadnkx-2.5/lacZ transgenicembryos exhibited transgene expression in the outflowtract and right ventricle of the heart as well as in the spleenprimordium (n � 2). The expression of muSmadnkx-2.5/lacZ in the outflow tract appeared diminished in compari-son to �3059nkx-2.5/lacZ (Figs. 4C and 5B). A similar loss
FIG. 3. Quantification of �-gal enzyme activity of transgenic embryos developing in utero. �-Gal enzyme activity was quantified bychemiluminescent reporter gene assays (Galactostar, Tropix) for individual embryos. Embryonic expression at E7.25 and E8.25 wasdetermined for �3059nkx-2.5/lacZ (Line 81) embryos relative to three independent transgenic �Smadnkx-2.5/lacZ lines (L-21, 69, 92).E7.25 embryos were at LS/OB stages. Statistical significance of �-gal expression relative to nontransgenic embryos was determined byStudent’s t test for each experimental group (n � 5–29; P � 0.05). Error bars represent SEM. *, statistically significant differences.
FIG. 4. Deletion of the nkx-2.5 Smad consensus region results in increased gene expression in the four-chambered heart. At E10.5,expression of �Smadnkx-2.5/lacZ expressed �3059nkx-2.5/lacZ in anterior heart (arrowhead) and spleen primordium (small arrow) (A, C).However, expression of �Smadnkx-2.5/lacZ appears to be reduced in the secondary heart field of the conotruncal myocardium. After heartchamber formation (E13.5), �Smadnkx-2.5/lacZ is expressed throughout the right ventricle (B). Expression of �3059nkx-2.5/lacZ isrestricted to the conotruncal region of the E13.5 heart (D).FIG. 5. Targeted mutation of Smad consensus binding sequences delays early induction of nkx-2.5/lacZ gene expression. Mutationsintroduced in �3059nkx-2.5/lacZ to generate muSmadnkx-2.5/lacZ are shown in red (see Fig. 1 for native sequence). muSmadnkx-2.5/lacZF0 transgenic embryos were examined at E7.5 (A) and E10.5 (B). At E7.5, expression was not detected in the cardiac crescent in 9/9 transgenicembryos (arrowheads), but �-gal activity was detected in the underlying anterior intestinal portal in 1/9 transgenic embryos (A). At E10.5,muSmadnkx-2.5/lacZ is expressed in the anterior heart tube and spleen primordium (lower arrow) but is diminished in the conotruncalmyocardium (upper arrow) relative to �3059nkx-2.5/lacZ (B; and Fig. 4C).
249nkx-2.5 Gene Induction
in outflow tract expression was observed for �Smadnkx-2.5/lacZ and may represent a requirement for the Smadconsensus sequences in the secondary induction of nkx-2.5expression in the conotruncal myocardium. Together, thesestudies demonstrate that targeted mutation of the threeSmad consensus sequences located in the 50-bp Smadconsensus region results in a delay in nkx-2.5 cardiacexpression. However, these Smad consensus sites do notappear to be required for later expression in the heart andspleen.
BMP-2 treatment induces nkx-2.5 gene expression incultured mouse embryos. An in vitro mouse embryoculture system was devised in which to examine theregulation of nkx-2.5 gene activation (Fig. 6). Transgenicmice generated with �3059nkx-2.5/lacZ activate �-gal ex-pression in the heart forming region prior to the early headfold stage (Downs and Davies, 1993; Searcy et al., 1998).Expression of �-gal in these embryos therefore serves as aneasily detectable marker for early nkx-2.5 gene induction.�3059nkx-2.5/lacZ embryos isolated after timed matingsare cultured for 2 days corresponding to E7.25–E9. Duringthe culture period, the cardiomyogenic lineage is deter-mined and cardiac-specific gene expression is initiated
(Auda-Boucher et al., 2000). �3059nkx-2.5/lacZ transgenicembryos were isolated at E7.25 [late streak (LS) to noallantoic bud (OB) stages; Downs and Davies, 1993] andplaced in culture. After 24–48 h, �-gal expression is acti-vated and beating cardiomyocytes are present. In manycultured embryos, �3059nkx-2.5/lacZ expression is local-ized to two heart forming regions similar to the endogenousnkx-2.5 expression domain determined by in situ hybrid-ization (Figs. 6A and 6B). Immunohistological analysesdemonstrate that �3059nkx-2.5/lacZ expression is colocal-ized with contractile protein gene expression in the sameembryo (Figs. 6C and 6D). Thus, �-gal expression in cul-tured �3059nkx-2.5/lacZ embryos is representative of earlyinduction of nkx-2.5 gene expression.
The ability of BMP-2 to activate expression of �3059nkx-2.5/lacZ or the endogenous nkx-2.5 gene was examined incultured transgenic embryos (Fig. 7). �3059nkx-2.5/lacZembryos were isolated at the LS stage and treated withBMP-2 (200 ng/ml). In control embryos, �-gal expression isrestricted to the cardiogenic region (Figs. 6A and 6B). For themajority of BMP-2-treated embryos, localization of �-galexpression appeared to be increased relative to untreatedembryos (Figs. 7A and 7B). In 6/10 X-gal-stained cultured
FIG. 6. Expression of �3059nkx-2.5/lacZ in cultured mouse embryos colocalizes with the cardiomyogenic lineage. Embryos were stagedaccording to Downs and Davies (1993) prior to dissection. (A) X-Gal staining of no bud (OB) stage transgenic embryo cultured for 2 days.(B) In situ hybridization of cultured OB embryo with nkx-2.5 antisense RNA probe. (C, D) Antibody colocalization of �-gal (C) and MF-20reactive sarcomeric myosin heavy chain expression (D) in the same early head fold (EHF) cultured embryo. Arrows in each panel indicatethe cardiogenic region.
embryos, there was an apparent expansion of �-gal expres-sion relative to untreated embryos. Similarly, increasedexpression of the endogenous nkx-2.5 gene was detected in
nontransgenic embryos treated with BMP-2 in parallelexperiments (Figs. 7C and 7D). In both cases, only a limitedregion of the embryo was capable of activating nkx-2.5 gene
FIG. 7. BMP-2 treatment has a limited ability to activate �3059nkx-2.5/lacZ or endogenous nkx-2.5 expression. Late streak stage�3059nkx-2.5/lacZ transgenic embryos (A, B) or nontransgenic embryos (C, D) were cultured in the absence (A, C) or presence (B, D) ofBMP-2 (200 ng/ml). Embryos were isolated at the LS stage and cultured as described in Fig. 6. For treated embryos, BMP-2 was includedthroughout the culture period. After 2 days, cultured transgenic embryos were stained with X-gal or prepared for embryo extracts. Six often treated embryos exhibited apparent expansion of �3059nkx-2.5/lacZ expression by X-gal staining. (Arrows indicate paired cardiogenicregions). (C, D) In situ hybridizations were performed with digoxigenin-labeled nkx-2.5 riboprobe on nontransgenic embryos (arrowsindicate nkx-2.5 expression). (E) �-Gal enzymatic activity was determined for treated and untreated �3059nkx-2.5/lacZ (L-81) andnontransgenic (ntg) embryos. Statistical significance of �-gal expression of BMP-2-treated relative to untreated control embryos wasdetermined by Student’s t test for each experimental group (n � 4–6; P � 0.05). *, statistically significant difference.
regulatory sequences. Thus, BMP-2 alone is sufficient toactivate �3059nkx-2.5/lacZ or endogenous nkx-2.5 only ina limited population of cells in the murine embryo.
Nkx-2.5/lacZ gene expression in cultured embryos wasquantified by using chemiluminescent reporter assays for�-gal enzyme activity. A statistically significant 60% in-crease in �-gal activity was observed for BMP-2-treatedembryos relative to untreated transgenic controls (Fig. 7E).The increase in �3059nkx-2.5/lacZ expression in responseto BMP-2 likely represents expanded activation of nkx-2.5gene expression within the region of cardiogenic potential.In comparably staged avian embryos, only the anteriormedial mesoderm activates nkx-2.5 expression whentreated with BMP-2 alone (Schultheiss et al., 1997; Andreeet al., 1998; Schlange et al., 2000). Thus, in culturedtransgenic mouse embryos, induction of �3059nkx-2.5/lacZ is similarly restricted in the embryo even in thepresence of ubiquitous BMP-2.
Induction of �Smadnkx-2.5 gene expression is delayedin cultured transgenic mouse embryos. The requirementfor the Smad consensus region in early nkx-2.5 gene induc-tion was further examined in cultured transgenic embryos.After 1 day in culture, �3059nkx-2.5/lacZ is easily detect-able by X-gal staining of transgenic embryo cultures (Fig.8C). In contrast, minimal X-gal staining was observed after1 day for �Smadnkx-2.5/lacZ embryos (Fig. 8A). After 2days, �-gal expression levels in cultured embryos are con-sistent with expression observed in differentiated hearttubes of corresponding �Smadnkx-2.5/lacZ embryos devel-oping in utero. Thus, the reduced early expression of�Smadnkx-2.5/lacZ represents a delay in gene activationrelative to �3059nkx-2.5/lacZ in cultured embryos. Theseanalyses support the importance of the clustered Smadconsensus sequences in the early embryonic induction ofnkx-2.5 gene expression.
DISCUSSION
A potential Smad responsive region was identified inearly cardiac regulatory sequences of mouse nkx-2.5. Thisregion contains three Smad DNA binding consensus se-quences based on homology with confirmed vertebrateSmad binding sites. One of the nkx-2.5 Smad sequences hashigh identity with Mad/Medea binding sites in Dpp-responsive sequences of Drosophila tin (Xu et al., 1998;Dennler et al., 1998; Ishida et al., 2000). In transgenic mice,the nkx-2.5 Smad consensus region is required for theinitial induction of gene expression in the forming heartand visceral mesoderm. �Smadnkx-2.5/lacZ embryos ex-hibit a delay in reporter gene activation relative to�3059nkx-2.5/lacZ embryos in the heart primordia,conotruncal myocardium, and visceral mesoderm. Targetedmutation of the three Smad sites inhibits induction ofnkx-2.5 transgene expression in the cardiac crescent. Laterin development, the Smad consensus sites are not requiredfor nkx-2.5/lacZ gene expression in the heart or spleen
primordium. A potential negative regulatory function of theSmad consensus region was apparent in �Smadnkx-2.5/lacZ expression in the right ventricle of the heart. Experi-ments in cultured embryos demonstrated that BMP-2 issufficient to activate �3059nkx-2.5/lacZ and endogenousnkx-2.5 gene expression in a subset of cells in the embryo.However, deletion of the Smad consensus region results ina delay in �Smadnkx-2.5/lacZ induction in cultured em-bryos. These data are consistent with distinct Smad-dependent and -independent regulatory mechanisms con-trolling nkx-2.5 gene expression during development.
Complex regulatory elements control mouse nkx-2.5gene expression during development. The temporal andspatial regulation of nkx-2.5 gene expression during mousedevelopment involves several distinct cis-acting regulatoryelements (reviewed in Schwartz and Olson, 1999). Twocardiac regulatory elements located approximately 3 and 10kb 5� to the nkx-2.5 transcriptional start site have beenidentified (Searcy et al., 1998; Lien et al., 1999; Reecy et al.,1999; Tanaka et al., 1999; Schwartz and Olson, 1999). Bothof the cardiac regulatory elements are dependent on GATAconsensus binding sites for gene activity. In previous stud-ies, expression of the �3059/�2554 regulatory element wassufficient for the initial induction of nkx-2.5 gene expres-sion in the heart, pharynx, and spleen progenitors (Searcy etal., 1998). This element is highly conserved in human andXenopus genomic sequence of the nkx-2.5 locus (Turbay etal., 1996; Sparrow et al., 2000). In transgenic Xenopusembryos, the homologous nkx-2.5 element is also GATA-dependent and is sufficient for gene expression throughoutthe embryonic heart. However, in transgenic mice, theXenopus element activates gene expression in the sameanterior segments of the heart as the �3059/�2554 mousenkx-2.5 sequence (Searcy et al., 1998; Sparrow et al., 2000).Thus the �3059/�2554 mouse nkx-2.5 regulatory elementand its Xenopus homologue are sufficient for gene activa-tion in early cardiac and visceral mesoderm progenitors andtheir derivatives. The early cardiac and visceral expressionof mouse �3059nkx-2.5/lacZ is comparable to the dorsalmesoderm expression of the Drosophila tin Dpp-responsiveregulatory element (Xu et al., 1998).
In the present study, a potential Smad-responsive region(�3059 to �3011) was identified within the early cardiacand splenic regulatory element of the mouse nkx-2.5 gene(Lyons et al., 1995; Searcy et al., 1998). This sequence isrequired for the initial induction of nkx-2.5/lacZ expressionin the heart and spleen and targeted mutation of the Smadsites results in the loss of early cardiac gene expression. Thesame sequence appears to have a negative regulatory role inthe right ventricle of the heart. The requirement for theSmad consensus region for early, but not later, cardiacexpression is suggestive of different regulatory interactionsacting on the same element over time. Previous studiesdemonstrated that this same regulatory region requirespaired GATA sites for its expression (Searcy et al., 1998).Therefore, the initiation of cardiac and splenic gene expres-sion regulated by the �3059/�2554 sequence requires the
Smad consensus region but maintenance requires GATAinteractions. A Smad consensus sequence that binds Smad4is located adjacent to these GATA sites and is required forexpression early in the cardiac crescent and later in thedifferentiated heart of transgenic embryos (Lien et al.,2002). Distinct regulatory mechanisms controlling earlycardiac and visceral mesoderm vs differentiated cardiocyteexpression have also been demonstrated for Drosophila tin(Xu et al., 1998). Thus, there is evidence in both flies andmice that distinct regulatory programs control tinman/nkx-2.5 expression in the cardiogenic and visceral mesoderm vsdifferentiated cardiac myocytes.
Induction of nkx-2.5 expression is mediated by BMPs inthe heart and gut. In chicken embryos, BMP-2 and -7 areexpressed in the cardiogenic region, and nkx-2.5 gene ex-pression is activated with BMP-2 treatment (Schultheiss et
al., 1997; Andree et al., 1998). In cardiogenic P19CL6 cells,BMP signaling mediated by TAK1 and Smads is required forcardiac differentiation (Monzen et al., 1999, 2001). Recentstudies have identified a second cardiogenic field in thepharynx and outflow tract of avian embryos (Waldo et al.,2001; reviewed in Yutzey and Kirby, 2002). BMP-2 isexpressed in the adjacent pharyngeal endoderm, and noggintreatment inhibits secondary cardiomyogenesis in theconotruncal myocardium (Waldo et al., 2001). In the aviangut, BMP-4 signaling through BMPR1 induces of nkx-2.5gene expression in the pyloric sphincter (Smith and Tabin,1999; Smith et al., 2000). Therefore, BMPs are present in thedeveloping heart and viscera when nkx-2.5 gene expressionis initiated. None of these studies demonstrated a directregulation of nkx-2.5 gene expression by Smads. However,direct binding of Smad homologues to Drosophila tin regu-
FIG. 8. �Smadnkx-2.5/lacZ is expression is delayed relative to �3059nkx-2.5/lacZ in early stage cultured mouse embryos. �Smadnkx-2.5/lacZ LS embryos were cultured for 1 (A) or 2 (B) days and stained with X-gal. �3059nkx-2.5/lacZ LS embryos were cultured for 1 (C)or 2 (D) days. Arrows in each panel indicate �-gal expression consistent with heart forming region. The delay in �Smadnkx-2.5/lacZactivation in cultured embryos is consistent with initiation of transgene expression in the differentiated heart tube of embryos developingin utero (Fig. 2).
latory sequences is required for induction of gene expres-sion by Dpp in cardiac and visceral mesoderm (Xu et al.,1998). In this study, we demonstrate that deletion ormutation of Smad consensus binding sequences in themouse nkx-2.5 gene inhibits gene activation in cardiac andvisceral mesoderm. An additional Smad binding site in the�3059/�2554 regulatory region also is required for nkx-2.5transgene expression in the developing heart (Lien et al.,2002). These observations are initial evidence that thesenkx-2.5 regulatory sequences represent a direct down-stream target of BMP signaling mediated by Smads.
Regulation of gene expression by Smads. Smad1 andSmad5 are associated with BMP-mediated signal transduc-tion in several developmental systems, and both are ex-pressed in the cardiogenic region of early stage avian andmouse embryos (data not shown; and Hata et al., 1998;Chen et al., 1998; Whitman, 1998; Yamada et al., 1999;Chang et al., 1999). Targeted mutation of smad5 in miceresults in embryonic lethality in homozygous mutant em-bryos with defects in cardiac development similar to thoseobserved for BMP-2 mutant embryos (Chang et al., 1999;Zhang and Bradley, 1996). In Xenopus embryos, inhibitionof BMP-mediated induction of cardiogenesis can be over-come by ectopic expression of XSmad1 (Shi et al., 2000).Together, these studies provide evidence for a role of Smads1 and 5 in mediating BMP-2 induction of early cardiac geneexpression. Smads 1 and 5 regulate mouse smad6 geneexpression by binding the sequence GCCGCGCC (Ishida etal., 2000). This sequence is present in the nkx-2.5 Smadconsensus region at �3059 to �3050. The nkx-2.5 Smadconsensus region also contains two CAGA sequences. Simi-larly clustered CAGA and GC-rich Mad/Medea bindingsites are required for dorsal mesodermal expression of aDrosophila even-skipped regulatory element (Knirr andFrasch, 2001). In vertebrates, Smads 3 and 4 have beenshown to bind and activate CAGA sequences in the humanPAI-type 1 and �2(I)-collagen genes in response to TGF-�signaling (Zawal et al., 1998; Dennler et al., 1998; Zhang etal., 2000). Nkx-2.5 gene expression has not previously beenshown to be regulated through TGF-� or Activin signaling.However, the presence of multiple potential Smad targetsequences in nkx-2.5 regulatory sequences and the complexpattern of nkx-2.5 gene expression support regulation bymultiple TGF-� superfamily members.
The nkx-2.5 Smad consensus region between �3059 and�3011 appears to have a negative role in nkx-2.5 generegulation in the right ventricle of the heart. This observa-tion suggests that the nkx-2.5 Smad consensus region canmediate either the induction or repression gene expressionat different times and in different tissues during embryo-genesis. In general, Smads bind DNA weakly and requirecofactors for their gene regulatory activity (reviewed inAttisano and Wrana, 2000). In some cases, Smad interac-tions with the same cis-acting element can activate orinhibit gene expression depending on the Smad partners(Labbe et al., 1998; Wotton et al., 1999). The complexregulation of the nkx-2.5 Smad consensus region in trans-
genic mice is initial evidence for different Smad cofactors inthe cardiogenic region versus the differentiated heart. Thehigh degree of nucleotide identity throughout this regionbetween mouse and human genomic sequences also may beindicative of conserved Smad cofactor interactions. Al-though there are no obvious consensus binding sites forpreviously identified Smad cofactors in the Smad consensusregulatory region, it seems likely that the developmentalregulation of nkx-2.5 gene expression involves complexprotein interactions on this cis-element.
ACKNOWLEDGMENTS
We thank Jeff Molkentin, Bob Schulz, Ilona Skerjanc, MelissaColbert, Rosa Serra, Woody Benson, Brian Keane, Lisa Ehrman, RonWaclaw, and members of the Division of Molecular CardiovascularBiology for critical discussions and technical advice and we thankEric Olson for communication of results prior to publication.Transgenic mice were generated by Karen Yager and the Children’sHospital Transgenic Animal Core. This work was supported by anEstablished Investigator Award from the American Heart Associa-tion (for K.E.Y) and a Grant-in-Aid from the American HeartAssociation-Ohio Valley Affiliate. Support for E.B.V. was providedby NIH HL07752.
REFERENCES
Andree, B., Duprez, D., Vorbusch, B., Arnold H. H., and Brand, T.(1998). BMP-2 induces ectopic expression of cardiac lineagemarkers and interferes with somite formation in chicken em-bryos. Mech. Dev. 70, 119–31.
Attisano, L., and Wrana, J. L. (2000). Smads as transcriptionalco-modulators. Curr. Opin. Cell Biol. 12, 235–243.
Auda-Boucher, G., Bernard, B., Fontaine-Perus, J., Rouaud, T.,Mericksay, M., and Gardahaut, M. F. (2000). Staging of thecommitment of murine cardiac cell progenitors. Dev. Biol. 225,214–225.
Azpiazu, N., and Frasch M. (1993). tinman and bagpipe: Twohomeo box genes that determine cell fates in the dorsal meso-derm of Drosophila. Genes Dev. 7, 1325–1340.
Bader, D., Masaki, T., and Fischman, D. A. (1982). Immunochemi-cal analysis of myosin heavy chain during avian myogenesis invivo and in vitro. J. Cell Biol. 95, 763–770.
Benson, D. W., Silberbach, G. M., Kavanaugh-McHugh, A., Cottrill,C., Zhang, Y., Riggs, S., Smalls, O., Johnson, M. C., Watson,M. S., Seidman, J. G., Plowden, J., and Kugler, J. D. (1999).Mutations in the cardiac transcription factor NKX2.5 affectdiverse cardiac developmental pathways. J. Clin. Invest. 104,1567–1573.
Bodmer, R. (1993). The gene tinman is required for specification ofthe heart and visceral muscles in Drosophila. Development 118,719–729.
Chang, H., Huylebroeck, D., Verscheuren, K., Guo, Q., MatzukM. M., and Zwijsen, A. (1999). Smad5 knockout mice die atmid-gestation due to multiple embryonic and extraembryonicdefects. Development 126, 1631–1642.
Chen, S. J., Yuan, W., Lo, S., Trojanowska M., and Varga, J. (2000).Interaction of Smad3 with a proximal Smad-binding element ofthe human alpha2(I) procollagen gene promoter required for
transcriptional activation by TGF-beta. J. Cell Physiol. 183,381–392.
Chen, Y. G., Hata, A., Wotton, D., Shi, Y., Pavletich, N., andMassague, J. (1998). Determinants of specificity in TGF-betasignal transduction. Genes Dev. 12, 2144–2152.
Colbert, M. C., Kirby, M. L., and Robbins, J. (1996). Endogenousretinoic acid signaling colocalizes with advanced expression ofthe adult smooth muscle myosin heavy chain isoform duringdevelopment of the ductus arteriosus. Circ. Res. 78, 790–798.
Dennler, S., Itoh, S., Vivien, D., ten Dijke, P., and Gaulthier, J. M.(1998). Direct binding of Smad3 and Smad4 to critical TGF�-inducible elements in the promoter of human plasminogenactivator inhibitor-type 1 gene. EMBO J. 17, 3091–3100.
Downs, K. M., and Davies, T. (1993). Staging of gastrulating mouseembryos by morphological landmarks in the dissecting micro-scope. Development 118, 1255–1266.
Ehrman, L., and Yutzey, K. E. (1999). Lack of regulation in the heartforming region of avian embryos. Dev. Biol. 207, 163–175.
Frasch, M. (1995). Induction of visceral and cardiac mesoderm byectodermal Dpp in the early Drosophila embryo. Nature 374,464–467.
Harvey, R. P. (1996). NK-2 homeobox genes and heart develop-ment. Dev. Biol. 178, 203–216.
Hata, A., Lagna, G., Massague, J., and Hemmati-Brivanlou, A.(1998). Smad6 inhibits BMP/Smad1 signaling by specificallycompeting with the Smad4 tumor suppressor. Genes Dev. 12,186–197.
Ishida, W., Hamamoto, T., Kusanagi, K., Yagi, K., Kawabata, M.,Takehara, K., Sampath, T. K., Kato, M., and Miyazono, K. (2000).Smad6 is a Smad1/5-induced Smad inhibitor. J. Biol. Chem. 275,6075–6079.
Johnson, K., Kirkpatrick, H., Comer, A., Hoffmann F. M., andLaughton A. (1999). Interaction of Smad complexes with tripar-tite DNA-binding sites. J. Biol. Chem. 274, 20709–20716.
Kanai, H., Tanaka, T., Aihara, Y., Takeda, S., Kawabata, M.,Miyazono, K., Nagai, R., and Kurabayashi, M. (2001). Transform-ing growth factor-�/Smads. signaling induces transcription of thecell type-restricted ankyrin repeat protein CARP gene throughCAGA motif in vascular smooth muscle cells. Circ. Res. 88,30–36.
Kawabata, M., Imamura, T., and Miyazono K. (1998). Signal trans-duction by bone morphogenetic proteins. Cytokine GrowthFactor Rev. 9, 49–61.
Kim, J., Johnson, K., Chen, H. J., Carroll, S., and Laughton, A.(1997). Drosophila Mad binds to DNA a directly mediatesactivation of vestigial by Decapentaplegic. Nature 388, 304–308.
Knirr, S., and Frasch, M. (2001). Molecular integration of inductiveand mesoderm-intrinsic inputs governs even-skipped enhanceractivity in a subset of pericardial and dorsal muscle progenitors.Dev. Biol. 238, 13–26.
Komuro, I., and Izumo, S. (1993). Csx: a murine homeobox-containing gene specifically expressed in the developing heart.Proc. Natl. Acad. Sci. USA 90, 8145–8149.
Labbe, E., Silvestri, C., Hoodless, P. A., Wrana, J. L., and Attisano,L. (1998). Smad2 and Smad3 positively and negatively regulateTGF�-dependent transcription through the forkhead DNA bind-ing protein FAST2. Mol. Cell 2, 109–120.
Lien, C. L., Wu, C., Mercer, B., Webb, R., Richardson J. A., andOlson, E. N. (1999). Control of early cardiac-specific transcrip-tion of Nkx2–5 by a GATA- dependent enhancer. Development126, 75–84.
Lien, C. L., McAnally, J., Richardson, J. A. and Olson, E. N. (2002).Cardiac-specific activity of an Nkx2.5 enhancer requires anevolutionarily conserved binding site. Dev. Biol., in press.
Lints, T. J., Parsons, L. M., Hartley, L., Lyons I., and Harvey, R. P.(1993). Nkx-2.5: A novel murine homeobox gene expressed inearly heart progenitor cells and their myogenic descendants.Development 119, 419–431.
Lough, J., and Sugi, Y. (2000). Endoderm and heart development.Dev. Dyn. 217, 327–342.
Lyons, I., Parsons, L. M., Hartley, L., Li, R., Andrews, J. E., Robb, L.,and Harvey, R. P. (1995). Myogenic and morphogenetic defects inthe heart tubes of murine embryos lacking the homeobox geneNkx2–5. Genes Dev. 9, 1654–1666.
Massague, J., and Chen, Y. G. (2000). Controlling TGF-� signaling.Genes Dev. 14, 627–644.
Mikawa, T., Fischman, D. A., Dougherty, J. P., and Brown, A. M. C.(1991). In vivo analysis of a new lacZ retrovirus vector suitablefor cell lineage marking in avian and other species. Exp. Cell Res.195, 516–523.
Monzen, K., Shiojima, I., Hiroi, Y., Kudoh, S., Oka, T., Takimoto,E., Hayashi, D., Hosoda, T., Habara-Ohkubo, A., Nakaoka, T.,Fujita, T., Yazaki, Y., and Komuro, I. (1999). Bone morphogeneticproteins induce cardiomyocyte differentiation through themitogen-activated protein kinase kinase kinase TAK1 and car-diac transcription factors Csx/Nkx-2.5 and GATA-4. Mol. Cell.Biol. 19, 7096–7105.
Monzen, K., Hiroi, Y., Kudoh, S., Akazawa, H., Oka, T., Takimoto,E., Hayashi, D., Hosoda, T., Kawabata, M., Miyazono, K., Ishii, S.,Yazaki, Y., Nagai, R., and Komuro, I. (2001). Smads, TAK1,. andtheir common target ATF-2 play a critical role in cardiomyocytedifferentiation. J. Cell Biol. 153, 687–698.
Nguyen, H. T., and Xu, X. (1998). Drosophila mef2 expressionduring mesoderm development is controlled by a complex arrayof cis-acting regulatory modules. Dev. Biol. 204, 550–566.
Reecy, J. M., Li, X., Yamada, M., DeMayo, F. J., Newman, C. S.,Harvey, R. P., and Schwartz, R. J. (1999). Identification ofupstream regulatory regions in the heart-expressed homeoboxgene Nkx2–5. Development 126, 839–849.
Schlange, T., Andree, B., Arnold, H.-H., and Brand, T. (2000). BMP2is required for early heart development during a distinct timeperiod. Mech. Dev. 91, 259–270.
Schott, J. J., Benson, D. W., Basson, C. T., Pease, W., Silberbach,G. M., Moak, J. P., Maron, B. J., Seidman, C. E., and Seidman, J. G.(1998). Congenital heart disease caused by mutations in thetranscription factor NKX2–5. Science 281, 108–111.
Schultheiss, T. M., Xydas, S., and Lassar, A. B. (1995). Induction ofavian cardiac myogenesis by anterior endoderm. Development121, 4203–4214.
Schultheiss, T. M., Burch, J. B. E., and Lassar, A. B. (1997). A role forbone morphogenetic proteins in the induction of cardiac myo-genesis. Genes Dev. 11, 451–462.
Schwartz, R. J., and Olson, E. N. (1999). Building the heart piece bypiece: Modularity of cis-elements regulating Nkx2–5 transcrip-tion. Development 126, 4187–4192.
Searcy, R. D., Vincent, E. B., Liberatore, C. M., and Yutzey, K. E.(1998). A GATA-dependent nkx-2.5 regulatory element activatesearly cardiac gene expression in transgenic mice. Development125, 4461–4470.
Shi, Y., Katsev, S., Cai, C., and Evans, S. (2000). BMP signaling isrequired for heart formation in vertebrates. Dev. Biol. 224,226–237.
Smith, D. M., and Tabin, C. J. (1999). BMP signalling specifies thepyloric sphincter. Nature 402, 748–749.
Smith, D. M., Nielson, C., Tabin, C. J., and Roberts, D. J. (2000).Roles of BMP signaling and Nkx2.5 in patterning at the chickmidgut–foregut boundary. Development 127, 3671–3681.
Sparrow, D. B., Cai, C., Kotecha, S., Latinkic, B., Cooper. B.,Towers, N., Evans, S. M., and Mohun, T. J. (2000). Regulation ofthe tinman homologues in Xenopus embryos. Dev. Biol. 227,65–79.
Stroschein, S. L., Wang W., and Luo, K. (1999). Cooperative bindingof Smad proteins to two adjacent DNA elements in the plasmin-ogen activator inhibitor-1 promoter mediates transforminggrowth factor �-induced Smad-dependent transcriptional activa-tion. J. Biol. Chem. 274, 9431–9441.
Tanaka, M., Wechsler, S. B., Lee, I. W., Yamasaki, N., Lawitts, J. A.,and Izumo, S. (1999). Complex modular cis-acting elementsregulate expression of the cardiac specifying homeobox geneCsx/Nkx2.5. Development 126, 1439–1450.
Taylor, L. M., and Khachigian, L. M. (2000). Induction of platelet-derived growth factor B-chain expression involves transactiva-tion by Smads. J. Biol. Chem. 275, 16709–16716.
Turbay, M., Chen, Z., Blanchard, K. M., and Izumo, S. (1996).Molecular cloning, chromosomal mapping and characterizationof the human cardiac-specific homeobox gene hCsx. Mol. Med. 2,86–96.
Waldo, K. L., Kumiski, D. H., Wallis, K. T., Stadt, H. A., Hutson,M. R., Platt, D. H., and Kirby, M. L. (2001). Conotruncalmyocardium arises from a secondary heart field. Development128, 3179–3188.
Whitman, M. (1998). Smads and early developmental signaling bythe Tgf� superfamily. Genes Dev. 12, 2445–2462.
Wilkinson, D. G. (1993). In situ hybridization. In “Essential Devel-opmental Biology” (C. D. Stern and P. W. H. Holland, Eds.) pp.257–274. Oxford Univ. Press, New York.
Wong, C., Rougier-Chapman, E. M., Frederick, J. P., Datto, M. B.,Liberati, N. T., Li, J.-M., and Wang, X.-F. (1999). Smad3-Smad4and AP-1 complexes synergize in transcriptional activation of the
Wotton, D., Lo, R. S., Lee, S., and Massague, J. (1999). A Smadtranscriptional corepressor. Cell 97, 29–39.
Wrana, J. L. (2000). Regulation of Smad activity. Cell 100, 189–192.Xu, X., Yin, Z., Hudson, J. B., Ferguson, E. L., and Frasch, M. (1998).
Smad proteins act in combination with synergistic and antago-nistic regulators to target Dpp responses to the Drosophilamesoderm. Genes Dev. 12, 2354–2370.
Yamada, M., Szendro, P. I., Prokscha, A., Schwartz, R. J., andEichele, G. (1999). Evidence for a role of Smad6 in chick cardiacdevelopment. Dev. Biol. 215, 48–61.
Yingling, J. M., Das, P., Savage, C., Zhang, M., Padgett, R. W., andWang, X.-F. (1996). Mammalian dwarfins are phosphorylated inresponse to transforming growth factor � and are implicated incontrol of cell growth. Proc. Natl. Acad. Sci. USA 93, 8940–8944.
Yutzey, K. E., Gannon, M., and Bader, D. (1995). Diversification ofcardiomyogenic cell lineages in vitro. Dev. Biol. 170, 531–541.
Yutzey, K. E., and Kirby, M. L. (2002). Wherefore heart thou?Embryonic origins of cardiogenic mesoderm. Dev. Dyn., in press.
Zawal, L., Dai, J. L., Buckhaults, P., Zhou, S., Kinzler, K. W.,Vogelstein, B., and Kern, S. E. (1998). Human Smad3 and Smad4are sequence-specific transcription activators. Mol. Cell 1, 611–617.
Zhang, H., and Bradley, A. (1996). Mice deficient for BMP2 arenonviable and have defects in amnion/chorion and cardiac de-velopment. Development 122, 2977–2986.
Zhang, W., Ou, J., Inagaki, Y., Greenwel, P., and Ramirez, F.(2000). Synergistic cooperation between Sp1 and Smad3/Smad4 mediates transforming growth factor �1 stimulation of�2(I)-collagen (COL1A2) transcription. J. Biol. Chem. 275,39237–39245.
Received for publication December 6, 2001Revised January 18, 2002
Accepted January 21, 2002Published online March 11, 2002
