Fibroblast growth factor 10 gene regulation in the second heart field by Tbx1, Nkx2-5, and Islet1 reveals a genetic switch for down-regulation in the myocardium

Abstract

During cardiogenesis, Fibroblast Growth Factor (Fgf10) is expressed in the anterior second heart field. Together with Fibroblast growth factor 8 (Fgf8), Fgf10 promotes the proliferation of these cardiac progenitor cells that form the arterial pole of the heart. We have identified a 1.7-kb region in the first intron of Fgf10 that is necessary and sufficient to direct transgene expression in this cardiac context. The 1.7-kb sequence is directly controlled by T-box transcription factor 1 (Tbx1) in anterior second heart field cells that contribute to the outflow tract. It also responds to both NK2 transcription factor related, locus 5 (Nkx2-5) and ISL1 transcription factor, LIM/homeodomain (Islet1), acting through overlapping sites. Mutation of these sites reduces transgene expression in the anterior second heart field where the Fgf10 regulatory element is activated by Islet1 via direct binding in vivo. Analysis of the response to Nkx2-5 loss- and Isl1 gain-of-function genetic backgrounds indicates that the observed up-regulation of its activity in Nkx2-5 mutant hearts, reflecting that of Fgf10, is due to the absence of Nkx2-5 repression and to up-regulation of Isl1, normally repressed in the myocardium by Nkx2-5. ChIP experiments show strong binding of Nkx2-5 in differentiated myocardium. Molecular and genetic analysis of the Fgf10 cardiac element therefore reveals how key cardiac transcription factors orchestrate gene expression in the anterior second heart field and how genes, such as Fgf10, normally expressed in the progenitor cell population, are repressed when these cells enter the heart and differentiate into myocardium. Our findings provide a paradigm for transcriptional mechanisms that underlie the changes in regulatory networks during the transition from progenitor state to that of the differentiated tissue.

The heart is the first organ to form during development, and it performs a vital function throughout the life of the organism. The primitive heart tube begins to pump oxygenated blood and nutrients from approximately embryonic day (E) 8 in the mouse embryo. Fusion at the midline of the left and right parts of the cardiac crescent, the so-called first heart field where differentiated myocardial cells are already present, leads to the formation of the early heart tube. Subsequently, cardiac progenitor cells, first located in pharyngeal mesoderm medial to the crescent and then dorsal and anterior to the heart tube in the second heart field (SHF), continue to add cells to the poles of the heart. The characterization of two heart fields (1–3) was complemented by the demonstration of two myocardial cell lineages (4). The anterior region of the SHF contributes to the right ventricle and, notably, to the myocardium of the outflow tract, which is entirely derived from this source (1, 5). The mesodermal core of the first two pharyngeal arches shares a clonal relationship with myocardium at the arterial pole of the heart (6). As the heart continues to grow with the addition of SHF cells and their differentiation into cardiomyocytes, it also undergoes complex morphogenesis, including looping of the tube and expansion of chamber myocardium. Cardiogenesis requires highly regulated control of these processes to avoid potentially lethal malformations. Any major defect will result in the death of the embryo; however, the frequency of cardiac malformations in newborn humans (∼0.8%) is indicative of the complexity and genetic susceptibility of heart development. Of these malformations, 30% are at the arterial pole (7), underlining the importance of regulation of the SHF contribution during development.

Since the first identification of the mammalian SHF 11 y ago (1), the expression of many genes in these cardiac progenitor cells has been described and their mutant phenotypes studied. Some genes, encoding transcriptional regulators, like ISL1 transcription factor, LIM/homeodomain (Islet1), are expressed throughout the SHF, and their mutation results in defects at both poles of the heart (8). Other genes, like T-box transcription factor 1 (Tbx1), are predominantly expressed in the anterior SHF (9), and outflow tract defects occur in the Tbx1 mutant (10) where the myocardium at the base of the pulmonary trunk is most affected (11). These regulatory proteins are confined to cardiac progenitor cells, whereas others, like the transcription factor NK2 transcription factor related, locus 5 (Nkx2-5), are present both in the SHF and in the differentiating myocardial cells of the heart tube. In Nkx2-5 mutants, the formation of the heart tube is severely affected, owing in part to perturbation of regulatory circuits involving this key transcription factor in cardiac progenitor cells, leading to repression of their proliferation (12). However, there are also direct effects on the myocardium whereby genes normally expressed in the SHF and down-regulated upon differentiation into myocardium, including Isl1, fail to undergo this down-regulation. This leads to persistence of progenitor gene expression in myocardium, and therefore a situation in which gene regulatory networks specific for cardiac progenitor cells or differentiated myocardium cannot be cleanly separated.

Many of the well-characterized intercellular signaling pathways are also active in the SHF and influence progenitor cell contributions to the heart. Both Fibroblast growth factor 8 (Fgf8) and Fibroblast growth factor 10 (Fgf10), for example, are expressed in the anterior SHF and are important for the formation of the arterial pole (13–15). As implied above, the characterization of mutant phenotypes affecting SHF behavior and the effect of such 7mutations on expression of other SHF genes has revealed a complex regulatory network (14). It is clearly very important to develop a comprehensive map of this network to understand how genetic and epigenetic perturbations impact cardiogenesis. At present we know very little about heart networks, particularly in the context of complex cell behaviors and morphogenesis over time.

Characterization of gene cis-regulatory elements is one approach to understanding network logic. A specific enhancer that controls anterior SHF expression of the Myocyte enhancer factor 2C (Mef2C) gene—which encodes a key transcription factor for cardiogenesis expressed in both progenitor cells and differentiating muscle—provides a classic example of an anterior SHF regulatory element (16). This Mef2c-anterior SHF enhancer, present in the first intron of the gene, has been shown to be regulated by both Islet1 and GATA binding factor 4 (Gata4). It does not seem to be directly regulated by Nkx2-5, although this may occur via the known direct interaction between Nkx2-5 and Gata4 (17). A separate SHF enhancer of Mef2c, also located in the first intron of the gene, has been shown to be regulated by Forkhead box H1 (Foxh1) and Nkx2-5, as well as SMAD family member (Smad)-mediated Transforming Growth Factor-β (TGF-β) signaling (18). An enhancer of Nkx2-5 that directs its expression in the SHF also depends on a Smad binding site and has been shown to be Gata dependent (19, 20). A regulatory element upstream of the Fgf8 gene that is essential for expression in the anterior SHF responds to Tbx1 (21), and in Tbx1 mutants SHF expression of Fgf8 is down-regulated, as is its close relative Fgf10 (22).

Fgf10 expression marks the anterior SHF. It was the characterization of a fortuitous insertion of an nLacZ transgene into the Fgf10 locus (generating the Mlc1v-nlacZ-24 transgenic mouse line) that led to the identification of the SHF as a source of cardiac progenitors for right ventricular and outflow tract myocardium, because in this line transgene transcription reflects that of Fgf10 in the SHF, but perduration of the β-galactosidase reporter marks SHF derivatives at the arterial pole of the heart (1). Similar to Isl1 and other heart progenitor genes, expression of Fgf10 as well as the lacZ transgene of the Mlc1v-nlacZ-24 line is up-regulated in the cardiac crescent region of Nkx2-5 null mutant embryos, persisting in the differentiating heart tube, where they would normally be down-regulated (12).

In this article we address the SHF cis-regulation of the Fgf10 gene to gain further insight into SHF gene regulatory networks and the Nkx2-5–dependent separation of cardiac progenitor and differentiated states. We show that SHF expression of Fgf10 depends on a 1.7-kb regulatory region in the first intron that is both necessary and sufficient to direct expression of a transgene to cardiac sites of Fgf10 transcription. Molecular and genetic analysis reveals how key cardiac transcription factors, notably Islet1 and Tbx1, acting on this element, orchestrate Fgf10 expression in the anterior SHF and how Nkx2-5 mediates repression of this SHF gene on myocardial differentiation.

Results

Definition of the Regulatory Region That Directs Fgf10 Expression to the SHF.

The Fgf10 gene, for which part of the locus is shown in Fig.1A, is transcribed in the pharyngeal mesoderm of the anterior part of the SHF (Fig. 1B), including the mesodermal core of the pharyngeal arches at E9.5, but not in the cardiac outflow tract, although some transcripts are detectable within the right ventricle. The Mlc1v-nLacZ-24 (Mlc1v24) transgene, integrated at −117 kb from Fgf10 (Fig. 1A), shows similar SHF transcription, with strong X-gal staining also in the outflow tract and right ventricle, owing to perdurance of β-galactosidase in SHF-derived cells that contribute to this myocardium (Fig. 1C) (1). To identify the cis-regulatory element of Fgf10 that directs expression of the endogenous gene and the Mlc1v24 transgene in the anterior SHF, we first tested a highly conserved sequence adjacent to the site of integration and a −6-kb sequence including the Fgf10 promoter region, by transient transgenic analysis. No reporter gene expression was seen in the outflow tract or SHF of five or six F0 embryos at E9.5 for the conserved sequence or Fgf10 promoter region, respectively. We therefore carried out BAC transgenic analyses. Transgenes in which the nLacZ reporter had been inserted at the translation start site of Fgf10 with extensive 5′ (−127/+40 kb, RP24-157C5-nLacZ) or 3′ (−62/+138 kb, RP23-375J2-nLacZ) flanking sequences (Fig. 1A) showed the characteristic SHF expression pattern only when more than +40 kb of 3′ sequence was present (Fig. 1 D and E). Interestingly, expression in the mesodermal core of the pharyngeal arches was maintained in the absence of this more 3′ region (Fig. 1D), suggesting that this aspect of Fgf10 expression is controlled by a different regulatory circuit. Subsequent transgenic analyses of deletions in BAC RP23-375J2-nLacZ (Fig. 1A) showed that the +40/+78-kb region of Fgf10 is necessary and sufficient for transgene expression in the anterior SHF (Fig. 1 F and G). Surprisingly, therefore, the Mlc1v-24 transgene has come under the control of a SHF regulatory region at more than 157 kb 3′ to its site of insertion.

Fig. 1.

BAC transgenic analysis revealed that a +40 to +78-kb region of the Fgf10 locus is essential for transcriptional activity in the anterior SHF. (A) The Fgf10 locus, showing the site at −117 kb where two copies of the Mlc1v-nLacZ-24 transgene were inserted and the Fgf10-BAC constructs with the nLacZ reporter inserted at the translation start site of Fgf10. Blue arrowheads represent the nLacZ sequence, and dotted lines represent deleted regions 5′ (−) or 3′ (+). Distances are indicated in kilobases (kb) from the transcriptional start site of Fgf10. Exons are shown as black boxes. (B–G) Whole-mount views of the heart and adjacent structures of E9.5 mouse embryos. (B) Endogenous Fgf10 transcripts were detected by in situ hybridization in the pharyngeal mesoderm, including the anterior SHF (bracket) and weakly in the right ventricle (RV), but not in the outflow tract (OFT), both indicated by arrows. LV, left ventricle; A, right atrium, OV, otic vesicle. First and second pharyngeal arches are numbered. (C) X-gal staining of the Mlc1v-nLacZ-24 transgenic line recapitulated Fgf10 expression, with perduration of β-galactosidase in the outflow tract and right ventricle, resulting from the contribution of labeled cells from the anterior SHF (bracket). (D–G) X-gal staining of transgenic lines of Fgf10-BAC constructs directing the nLacZ reporter (A), with the anterior SHF indicated by a bracket. Numbers of embryos, at E9.5 with nLacZ expression in the anterior SHF (aSHF) compared with the total number of β-galactosidase–positive transgenic embryos is shown in the column on the right of A. A BAC containing −127 to +40 kb (RP24-157C5-nLacZ) did not show labeling in the anterior SHF (D), whereas a BAC containing −62 to +138 kb (R23-375J2-nLacZ) did (E). When the +40 to +78-kb region was deleted from the BAC containing +62 to +138 kb, the anterior SHF and outflow tract was not labeled [X-gal staining of the otic vesicle (OV) provided a positive control] (F), whereas when the +78 to +138-kb region was deleted from the BAC, β-galactosidase–positive cells were retained in the anterior SHF and outflow tract (G).

To further dissect the +40/+78-kb region, we constructed transgenes in which different genomic fragments covering this region directed the expression of a LacZ reporter preceded by the heat shock 68 gene promoter (hsp) (Fig. 2A). Only the +32 to +48-kb region directed transgene expression to the anterior SHF (Fig. 2B), whereas transgenes with the +48 to +63-kb region (Fig. 2C) or the +63 to +78-kb region (Fig. 2D) showed pharyngeal arch expression or some ectopic expression in the heart, respectively, but not in the anterior SHF (Fig. 2A). These results, together with those of the BAC transgenic analyses (Fig. 1), indicate that the SHF regulatory region of Fgf10 is present between +40 and +48 kb. A sequence comparison of this region between different species indicated notable sequence conservation in a 1.7-kb sequence at +44/+46 kb, in the first intron of the Fgf10 gene (Fig. 2A). This 1.7-kb sequence was therefore placed in front of an hsp-LacZ reporter (Fig. 2A) and tested in transgenic embryos. At E9.5 (Fig. 2E) and at E7.5 (Fig. 2F) this sequence directed expression of the transgene in the anterior SHF and in the mesodermal core of the pharyngeal arches as seen for BAC RP23-375J2 (Fig. 1E and Fig. S1). When the 1.7-kb regulatory sequence was deleted from the −62/+138-kb BAC (RP23-375J2Δ1.7kb-nLacZ) this expression was lost, as shown at E9.5 (Fig. 2G). We therefore conclude that the 1.7-kb region is necessary and sufficient to direct Fgf10 transcription in the anterior SHF. Dissection of the 1.7-kb region into smaller fragments did not give robust expression in transgenic analysis (Fig. S2).

Fig. 2.

Transgenic analysis establishes that a 1.7-kb region at +45 kb in the Fgf10 gene is necessary and sufficient for expression in the anterior SHF. (A) Transgenes in which sequences within the first two introns and second and third exons (exons shown as black boxes), spanning +32 to +78 kb, of the Fgf10 gene were placed 5′ to the LacZ reporter (blue arrowhead), preceded by the hsp68 promoter (green box) are shown, together with the BAC RP23-375J2-nLacZ transgene (Fig. 1) in which the +44 to +46-kb regulatory region has been deleted (RP23-375J2Δ1.7-kb-nLacZ; dotted line). This covers the highly conserved 1.7-kb region shown in a cross-species sequence comparison, which was also placed in front of hsp-LacZ for transgenic analysis. Figures on the right indicate the number of transgenic embryos with expression in the anterior SHF (aSHF) compared with the total number of β-galactosidase–positive transgenic embryos at E9.5. (B–D) Examples of transgenic embryos stained with X-gal. At E9.5, labeling was seen in the anterior SHF (bracket), outflow tract (OFT, arrow), right ventricle (RV), and not in the left ventricle (LV) with +32 to +48 kb (1, 2, first and second pharyngeal arches) (B), was only detected in the arches with +48 to +63 kb (C), and with the +63 to +78-kb fragment expression was absent in the anterior SHF (bracket) (D). The 1.7-kb transgene directed expression in the anterior SHF (bracket), with β-galactosidase activity in its derivatives (outflow tract, right ventricle) in the heart at E9.5 (E) and in the SHF in the region lying medial to the cardiac crescent (arrows) at E7.5 (F). When the 1.7-kb sequence was deleted from the +62 to +138-kb BAC transgene (RP23-375J2Δ1.7-kb-nLacZ), no anterior SHF expression was seen (bracket), as shown at E9.5 (G).

Transcriptional Regulation of the 1.7-kb Sequence in the Anterior SHF.

Because the 1.7-kb sequence directs transgene expression in the anterior SHF, we examined whether it is regulated by transcription factors known to play a role in this population of cardiac progenitor cells.

Nkx2-5.

Nkx2-5 is expressed both in the SHF and in the myocardium of the heart. In Nkx2-5 mutant embryos, Fgf10 expression is induced throughout the myocardium (12), implying that Nkx2-5 directly or indirectly suppresses Fgf10 expression in the heart. We crossed the 1.7kb-hsp-LacZ transgenic line onto the Nkx2-5^GFP/+ mouse line and examined transgene expression in Nkx2-5 heterozygote (Fig. 3 A and B) and mutant (Fig. 3 D and E) embryos at E9.5. With a single allele of Nkx2-5, β-galactosidase activity in the anterior SHF, outflow tract, and right ventricle was normal, whereas in the absence of Nkx2-5, β-galactosidase was still detected in the SHF and truncated outflow region, but there was now strong activity in the single ventricle of the mutant heart. To confirm the ventricular identity of this region of mutant hearts, the Nkx2-5^GFP/+ mouse line was crossed onto a transgenic line in which an nLacZ reporter is controlled by the Mef2c enhancer, which is specifically active in the anterior SHF (Fig. 3C) (16). As expected, in the presence of one allele of Nkx2-5, β-galactosidase activity was seen in the outflow tract and right ventricle that derive from these progenitor cells. In the absence of Nkx2-5, the remaining highly truncated outflow region was labeled, but the single ventricle was negative, demonstrating its left ventricular identity (Fig. 3F) consistent with our previous findings (12). Therefore, as for the native Fgf10 gene and Mlc1v24 transgene, the 1.7-kb regulatory sequence responds to the absence of Nkx2-5 by abnormal transcriptional activity in ventricular myocardium, indicating that it is a target of the repression exerted by Nkx2-5 on the Fgf10 gene.

Fig. 3.

Nkx2-5 binds to the Fgf10 1.7-kb regulatory region and controls its transcriptional activity in the heart. (A–F) X-gal staining of E9.5 transgenic embryos, with the 1.7-kb-hsp-LacZ (A, B, D, and E) or Mef2c-aSHF-nLacZ (C and F) transgenes in a Nkx2-5 heterozygote (A–C) or homozygote mutant background (D–F). LacZ expression under the control of the 1.7-kb regulatory region was retained in the anterior SHF but ectopically induced in the single ventricle of Nkx2-5 mutants (D and E), whereas the Mef2c-aSHF enhancer did not change its expression pattern, which was limited to the remaining outflow region of the mutant heart (C and F). 1, 2, first and second pharyngeal arches; OFT, outflow tract; RV, right ventricle; LV, left ventricle; V, single ventricle in the mutant. (G) Localization of putative Nkx2-5 binding sites in the 1.7-kb regulatory region (+44/+48 kb) and the oligo DNA sequences (sites 1–8), which were used for EMSA. (H) The 1.7-kb regulatory region with mutations in all putative Nkx2.5 binding sites did not show ectopic β-galactosidase activity in the heart, and the activity decreased in the SHF (bracket). Two embryos in a total of seven β-galactosidase-positive embryos, at E9.5, had anterior (a) SHF expression. (I) EMSA with reticulocyte lysates containing Nkx2-5 protein (+) showed substantial Nkx2-5 binding (horizontal arrow) to sites 2 and 4 and binding to sites 3, 5, and 7 in the 1.7-kb region, compared with controls with lysate alone (−). In the two right-hand lanes an oligonucleotide with a consensus Nkx2-5 binding site from the atrial natriuretic factor (Anf/Nppa) gene (39) was used as a positive control.

The 1.7-kb sequence contains eight putative Nkx2-5 binding sites (Fig. 3G), and Electrophoretic Mobility Shift Assay (EMSA) showed that sites 2 and 4 bind the transcription factor, with weaker binding to sites 3, 5, and 7 (Fig. 3I and Fig. S3). Sites 2 and 7 correspond to the previously described Nkx2-5 consensus, 5′-T(C/T)AAGTG-3′ (23). Unexpectedly, the 1.7-kb sequence in which all of the Nkx2-5 sites had been mutated showed no up-regulation of transgene expression in the left ventricle in mice wild type for Nkx2-5. β-Galatosidase activity was present in the right ventricle, lower in the outflow tract and, at E9.5, was clearly attenuated in the SHF (Fig. 3H), compared with that of a transgenic embryo with the wild-type sequence (Fig. 2E) . This therefore suggests that the activation of the 1.7-kb sequence in the anterior SHF is influenced directly through Nkx2-5–like binding sites and that ectopic activation in the heart in the Nkx2-5 mutant involves these same sites, and probably the action of another transcription factor, possibly Islet1.

Islet1.

Islet1 is a candidate regulator of the 1.7-kb element, because Fgf10 expression in the SHF is diminished in Isl1 mutant embryos (8), and Isl1 expression is ectopically induced in the ventricle of Nkx2-5 mutant hearts (12). To test a potential role of Islet1 in the ectopic activation of the 1.7-kb regulatory sequence in the ventricle of Nkx2-5 mutants, we generated a mouse line that overexpresses Islet1 in the heart. Normally, Isl1 transcripts are present in the SHF and at other sites, such as the dorsal root ganglia, in the embryo, but are not detected in the myocardium of the cardiac chambers (Fig. 4 A–C). Conditional expression of an Islet1 coding sequence, under the control of the widely expressed CAG promoter, was induced by Cre recombination, after excision of an intervening CAT reporter sequence. This conditional CAG-floxedCAT-Isl1pA transgenic line was crossed with a Mesoderm posterior1^Cre/+ (Mesp1^Cre/+) line, in which the Cre recombinase is expressed in all myocardial progenitor cells (24), leading to Islet1 expression in the heart and head mesoderm, as shown by in situ hybridization for Isl1 transcripts (Fig. 4 D–F) and by immunohistochemistry with an Islet1 antibody. No signal was observed in the control left ventricle (Fig. 4 G and H), whereas many cardiac cells were Islet1 positive in the wall of the left ventricle, as well as other heart regions, after transgene recombination mediated by the Mesp1 controlled Cre (Fig. 4 I and J). The 1.7kb-hsp-LacZ transgenic line was crossed with the conditional Islet1-expressing line and then recrossed onto the Mesp1^Cre/+ line so that Islet1 was ectopically expressed in the myocardium. In these embryos, β-galactosidase activity was seen mosaically in atrial and left ventricular myocardium (Fig. 4 M and N), where the 1.7kb-hsp-LacZ transgene is not normally expressed and where, in contrast to the outflow tract and right ventricle, the presence of β-galactosidase, because of perdurance from earlier SHF expression, was not observed (Fig. 4 K and L). Transgene expression in the SHF was similar to that of control. Because both Islet1 and Nkx2-5 transcription factors contain homeodomains that interact with DNA, we speculated that Islet1 may also bind to one or more of the putative Nkx2-5 binding sites in the 1.7-kb regulatory sequence. Using the same oligonucleotides as in Fig. 3 for EMSA, we observe robust Islet1 binding to sites 1, 3, 4, 5, 7, and 8, overlapping with Nkx2-5 binding at site 4 and potentially also at sites 3, 5, and 7 (Fig. 4O and Fig. S3). The Islet1 consensus binding site 5′-(C/T)TAATG(A/G)-3′ (25) is similar but not identical to these sites. Two consensus binding sites have been described for the related protein Islet2 (26), with the first corresponding to the published Islet1 site, whereas the second, 5′-C(C/T)TAAGTG-3′, resembles sites 1, 3, 4, 5, and 7. EMSA experiments with Islet1 and increasing concentrations of Nkx2-5 showed that the factors compete for these sites (Fig. 5A). Site 2 or sites 1, 3, 5, 7, and 8 were preferentially bound by Nkx2-5 or Islet1, respectively. These results, summarized in Fig. 5B, suggest that ectopically induced Islet1 can activate the 1.7-kb regulatory sequence through these homeodomain binding sites in the myocardium of Nkx2-5 mutants. Mutation of the homeodomain binding sites will compromise Islet1 as well as Nkx2-5 binding in the SHF, resulting in reduced transcriptional activity of the 1.7-kb regulatory sequence in cardiac progenitor cells (Fig. 3H). We carried out ChIP experiments with Nkx2-5 or Islet1 antibodies, using in vivo extracts prepared from the anterior SHF, from right ventricular/outflow tract myocardium, or from left ventricular myocardium of E9.5 embryos. Nkx2-5 binding to the native 1.7-kb regulatory sequence of the endogenous Fgf10 gene was not observed in the SHF, whereas strong binding was observed with extracts from the heart, including the left ventricle. Islet1, on the other hand, is present on the 1.7-kb sequence in the anterior SHF, but expected binding was also seen in the left ventricle (Fig. 5 C and D). Western blots with Islet1 antibody on chromatin extracts from the ChIP experiments confirmed the presence of Islet1 protein (Fig. S4). We cannot exclude minor contamination with outflow tract/right ventricular myocardium; however, the relative band intensities for Islet1 were consistent between different experiments and with different batches of Islet1 antibody (Fig. 5 C and D). On Western blots with extracts of left ventricular or outflow tract/right ventricular myocardium, Islet1 was not clearly detectable, whereas it was present in the SHF extract. With highly sensitive immunohistochemistry on sections, some cells, mainly in the interventricular region, were Islet1 positive (Fig. S4). These may be the source of the ChIP signal, although their number, relative to the SHF, might suggest that more cells have Islet1 binding to chromatin than detected by direct antibody interaction. We conclude that Islet1 occupies homeodomain sites on the Fgf10 1.7-kb sequence in the heart, including in the left ventricle. The absence of activity of the 1.7-kb sequence in the left ventricle would suggest that these sites are not sufficient to drive transcription in wild-type mice in the presence of Islet1 and strong Nkx2-5 binding, and/or that they are restricted to a few cells in the interventricular region where the demarcation of transgene expression is less clear cut (Discussion).

Fig. 4.

Islet1 directly activates the Fgf10 1.7-kb regulatory region in the heart. (A–G) In situ hybridization on E9.5 embryos showing Isl1 transcripts in control hearts (A–C) and transgenic hearts, where Isl1 was ectopically expressed (D–F). In control embryos (A), with enlargement of left lateral (B) and right lateral (C) views, Isl1 was expressed in the SHF (bracket, B and C) but not in the right ventricle (RV), left ventricle (LV), or atria. LA, left atria; OFT, outflow tract; 1, 2, first and second pharyngeal arches. The transgenic mouse that conditionally overexpresses Isl1 (CAGfloxedCAT-Isl1pA) was crossed with a Mesp1^Cre/+ line to induce Islet1 expression throughout the heart (D–F), presented as in A–C. (G–J) Immunohistochemistry for Islet1 protein on sections of control hearts (G and H) and transgenic hearts (I and J). Islet1 protein was not detected in the left ventricle of control embryos at E9.5 (G and H), whereas it was present in the transgenic heart (red labeling in I and J). H and J are enlargements corresponding to the boxed area in G and I, respectively. (K–N) X-gal staining of control 1.7-kb-hsp-nLacZ transgenic embryos at E9.5 (K and L) and of these transgenic embryos on the CAG-floxedCAT-Isl1pA; Mesp1^Cre/+ genetic background, where Islet1 was ectopically expressed in the heart. The outflow tract and right ventricle were β-galactosidase positive in the control heart (K and L), whereas the left ventricle and atria also showed X-gal labeling in the Islet1-positive hearts (M and N). K and M are lateral views of embryos in the cardiac region; L and N show frontal views of isolated hearts. (O) EMSAs with reticulocyte lysates with (+) or without (−) Islet1, with the oligonucleotides containing putative Nkx2-5 binding sites used in Fig. 3G, showed strong Islet1 binding to sites 1, 3, 4, 5, 7, and 8. The sequence in the Mef2c-aSHF enhancer that contains a consensus Islet1 binding site (16) provided a positive control (right lanes).

Fig. 5.

Nkx2-5 and Islet1 compete for homeodomain sites and bind to the Fgf10 regulatory region in vivo. (A) Competitive EMSA. When Nkx2-5 is added to the oligos 3, 4, 5, and 7 (Fig. 3G) in addition to Islet1, Islet1 binding was weakened, depending on Nkx2-5 dose. (B) Localization of Nkx2-5 and Islet1 binding sites. Positions of primers used for quantitative PCR (qPCR) and real-time qPCR are indicated by arrows and arrowheads, respectively. (C and D) ChIP analysis with E9.5 embryo extracts from the SHF, outflow tract (OFT), and right ventricle (RV) or left ventricle (LV), with Islet1 or Nkx2-5 antibodies. A representative result of qPCR with 5′ and 3′ primers for the 1.7-kb sequence is shown in C. Strong Nkx2-5 binding to the 1.7-kb regulatory region of Fgf10 in the left ventricle, but not in the SHF, was observed, whereas Islet1 was bound to the 1.7-kb sequence both in the SHF and left ventricle. ChIP results are summarized in D. Histograms show significant binding of Islet1 in all extracts and of Nkx2-5 in outflow tract/right ventricle and left ventricle extracts. *P < 0.05; **P < 0.01. Comparison of values on the y axis between extracts or antibodies is not valid because of limiting quantities of material for the former and differences in antibody performance for the latter.

Tbx1.

Another important regulator of the anterior SHF is the T-box transcription factor Tbx1. Studies on Tbx1 mutant embryos suggest that this factor may be a candidate activator of Fgf10 transcription (21, 22, 27–29). We therefore analyzed expression of the 1.7kb-hsp-LacZ transgene on a Tbx1 mutant background. In contrast to control embryos where β-galactosidase activity was seen in the anterior SHF, and in the myocardium of the outflow tract and right ventricle (Fig. 6 A–C), β-galactosidase activity was difficult to detect in the SHF and outflow tract of E9.5 Tbx1 null embryos but remained strong in the right ventricle (Fig. 6 D–F). Six putative T-box binding sites are present in the 1.7-kb sequence (Fig. 6G). When all these sites are mutated, transgene expression in the SHF and outflow tract is severely down-regulated (Fig. 6H). These results indicate that Tbx1 is a direct activator of Fgf10 transcription, acting through the 1.7-kb regulatory sequence in the anterior SHF in progenitor cells that will contribute to the myocardium of the outflow tract but not the right ventricle.

Fig. 6.

Tbx1 regulates the Fgf10 1.7-kb regulatory region in the anterior SHF. (A–F) X-gal staining of 1.7-kb-hsp-LacZ transgenic embryos at E9.5 on control (Tbx1^+/+) (A–C) or Tbx1 mutant (Tbx1^−/−) backgrounds (D–F). In control embryos, β-galactosidase–positive cells were present in the anterior SHF (bracket in A, white arrowheads in B), outflow tract (OFT), and right ventricle (RV) (A–C), whereas labeling in the anterior SHF and outflow tract was strongly reduced in Tbx1 mutant embryos (D–F). A and D are lateral views of the embryo; B and E are ventral views with the heart tube removed to show the SHF in the dorsal pericardial wall; and C and D show frontal views of isolated hearts. (G) There are six putative Tbx binding sites in the 1.7-kb sequence. (H) When these binding sites were mutated in the 1.7-kb-hsp-LacZ transgene, β-galactosidase–positive cells were lost in the anterior SHF (bracket) and outflow tract at E9.5. Some anterior SHF (aSHF) expression was observed in 3 of 12 transgenic embryos, with X-gal staining in the right ventricle.

Discussion

We have identified a 1.7-kb sequence in the first intron of the Fgf10 gene that is conserved between species and directs transcription in the anterior SHF. BAC deletion analysis shows that this regulatory region is necessary, as well as sufficient, for directing expression to these cardiac progenitor cells. This is distinct from a recently described sequence in the first intron of the human FGF10 gene, which was shown to bind Islet1 in extracts of embryonic hearts but did not direct robust cardiac/SHF expression in vivo (30). The element that we describe is required for expression in the pharyngeal mesoderm of the SHF but not in the mesodermal core of the first two pharyngeal arches, which depends on a −127/−62-kb region of the Fgf10 gene, as well as the −62/+138-kb region. Thus, distinct regulatory circuits control Fgf10 expression in different subpopulations of pharyngeal mesoderm. Investigation of the regulation of the 1.7-kb sequence, both on gain and loss of function mutant backgrounds and by mutation of potential target sites for transcription factors, shows that it can respond to Tbx1, Nkx2-5, and Islet1. Its activation in the SHF at E9.5 depends on Tbx1, thus limiting Fgf10 expression to the anterior SHF. This is consistent with observations on Tbx1 mutants where Fgf10 transcripts are down-regulated (22). An additional T-box binding element at the Fgf10 locus has been identified in the promoter region and shown to be responsive to Tbx1 and T-box transcription factor 5 (Tbx5) in COS-7 cells (27), although the in vivo role of this site remains to be evaluated. Our results demonstrate that Fgf10, as well as Fgf8 (21), which play a major signaling role in the anterior SHF (13), are under Tbx1 control. Detection of strong β-galactosidase activity from the Fgf10 transgene in the right ventricle in the absence of Tbx1 is consistent with observations on Tbx1 mutants and genetic tracing experiments, which indicate that Tbx1-positive progenitors principally contribute to myocardium at the base of the pulmonary trunk (11) but do not contribute significantly to right ventricular myocardium, which is also colonized by cells that had expressed Fgf10 (1).

The 1.7-kb Fgf10 regulatory element also contains multiple homeodomain binding sites that constitute putative sequences for Nkx2-5 binding. EMSA shows that Nkx2-5 can bind to several of these sites, a result confirmed by in vivo ChIP analysis (Fig. 5B), notably in the embryonic heart, where Nkx2-5 is strongly bound to the 1.7-kb region. To date, Islet1 binding to DNA has not been considered in the context of potential Nkx2-5 binding sites. We now show that these two key cardiac regulatory factors have overlapping interaction with homeodomain binding sites for which they compete. This therefore raises important questions about their regulatory relationship. In the SHF, the level of Nkx2-5 is ∼fivefold lower than in differentiated myocardium (12), and the activity of the 1.7-kb Fgf10 regulatory sequence is not strongly affected by the absence of Nkx2-5. Islet1 is thus likely to play a major role in its activation in these cardiac progenitor cells and indeed in Isl1 mutants, SHF expression of Fgf10 is down-regulated (8). When all of the homeodomain sites are mutated, 1.7kb-hsp-LacZ transgene expression is down-regulated in the SHF at E9.5; however, β-galactosidase activity is still detected in the arterial pole of the heart, notably in the right ventricle. This would suggest that other sites in the 1.7-kb regulatory sequence, targeted by other factors, can drive expression in right ventricular progenitors at earlier stages of SHF development or may reflect an increase in the low-level transcription of Fgf10 observed in right ventricular myocardium (1).

The regulatory circuitry in the myocardium of the heart is particularly intriguing. Fgf10, like other SHF genes, including Isl1, is not normally expressed in the left ventricle. However, in Nkx2-5 mutants, Fgf10 expression (12) and also that of the transgene under the control of the 1.7-kb sequence are up-regulated in the heart, where Isl1 is also abnormally expressed. A direct role for Nkx2-5 in repression of progenitor genes such as Isl1 and Fgf10 in the differentiating myocardium is consistent with the function of Nkx2-5 (tinman) as a repressor in the heart (dorsal vessel) of Drosophila (31). Ectopic expression of an Isl1-expressing transgene in the myocardium results in up-regulation of the 1.7-kb Fgf10-hsp-LacZ transgene, showing that high levels of Islet1 can overcome Nkx2-5 repression. The 1.7-kb sequence also depends on T-box elements for activity. In the myocardium Tbx1 is absent; however, other Tbx factors, such as Tbx5 or Tbx20 present in the heart, may contribute to activation of the 1.7-kb element upon Islet1 misexpression.

ChIP experiments indicate that Nkx2-5 normally occupies binding sites in the 1.7-kb sequence in the myocardium and probably plays a direct role in normal repression of Fgf10 expression. However, mutation of the homeodomain sites in the 1.7-kb sequence does not lead to up-regulation in the heart, indicating that de-repression alone is insufficient for activation, which we conclude depends, in the Nkx2-5 mutant, on the persistent expression of Islet1 acting through one or more of the Nkx2-5/Islet1 sites. Surprisingly in control embryos, where Islet1 is not detectable in ventricular myocardium, Islet1 occupancy of sites on the 1.7-kb sequence is seen in ChIP experiments, not only in extracts from the SHF or outflow tract/right ventricle of E9.5 embryos but also in extracts prepared from left ventricular myocardium. This may reflect contamination, but, as discussed in Results, the constant level observed in different experiments argues against this possibility. Left ventricular myocardium is derived from the first heart field and initially constitutes the primitive heart tube after fusion of the cardiac crescent (2). The phenotype of Isl1 mutants indicates that formation of the poles of the heart, but not the left ventricle, is Islet1 dependent (8). Initial genetic tracing experiments also suggested that Isl1-expressing cells did not contribute to this part of the heart. However, more sensitive detection methods now indicate that Islet1 is also expressed in the first heart field/cardiac crescent (12), and genetic tracing experiments indicate that the Isl1-lineage also contributes to the left ventricle (32, 33). It is possible that Islet1 occupancy of homeodomain sites is maintained through rounds of cell division, after myocardial differentiation, but that Nkx2-5 repression prevents any functional effect. These data also suggest that Islet persists only on select targets on DNA, because Islet1 protein is undetectable throughout the myocardium, as we confirm by Western blotting and immunohistochemistry. However, we do detect a few Islet1-positive cells, particularly in the interventricular region. It is not clear whether these are sufficient to account for the ChIP signal, nor whether they correspond to previously identified, rare, Islet1-positive cells detected in the developing heart (34). Further work will be required to characterize these cells and to determine which cells have Islet1 binding to the 1.7-kb sequence.

We summarize our interpretation in the model presented in Fig. 7. The key finding from our analysis of the Fgf10 1.7-kb sequence is the dynamic interplay between two major cardiac transcription factors, such that the 1.7-kb element acts as a switch, regulating the on/off status of the Fgf10 gene during progressive differentiation of SHF progenitors.

Fig. 7.

Model of Fgf10 regulation by Nkx2-5 and Islet1 in the heart. Nkx2-5 is expressed both in the progenitor cells of the SHF, which give rise to the arterial pole of the heart, and in the myocardium of the left ventricle, derived from the first heart field (FHF) (red) . The expression level of Nkx2-5 is higher in the myocardium. Islet1, on the other hand, is strongly expressed in the SHF. It is also expressed at a low level in the FHF, and it continues to be bound to chromatin in at least some cells in the myocardium, including in the left ventricle (green). The two factors bind homeodomain sites on the Fgf10 regulatory region, with overlapping specificities. Islet1 principally activates, and Nkx2-5 represses, transcription. Furthermore, Nkx2-5 negatively regulates Isl1 expression. This balance of Nkx2-5 and Islet1 determines the expression pattern of Fgf10, which is directly repressed by Nkx2-5 and requires Islet1 for activation. Activation of the 1.7-kb sequence also depends on T-box sites. In the SHF, Tbx1 is the principal activator, and we speculate that other Tbx factors, such as Tbx5, can play this role in the heart when Nkx2-5 repression is lifted.

Our findings suggest a direct repressive role for Nkx2-5 on the 1.7-kb Fgf10 element. Future work should address this, as well as how Nkx2-5 represses Isl1 expression in myocardium, issues germane to the changes in the regulatory networks during transition from the progenitor to the differentiated myocardial states.

Materials and Methods

Mouse Strains.

The Mlc1v24-enhancer trap line, Tbx^+/−, Nkx2.5^GFP/+, and Mesp1-Cre mice are as previously described (1, 10, 24, 35). To generate a conditional Isl1-overexpression transgenic line, murine Isl1 cDNA was inserted into a CAG-flox-CAT-flox vector (36), and the transgene introduced into fertilized eggs to establish transgenic lines.

BAC Reporter Constructs.

Two BAC clones containing the Fgf10 locus, RP24-157C5, and RP23-375J2 were purchased from the BACPAC resources center, and BAC recombination was performed according to Lee et al. (37) using DY380 or EL250 bacterial strains. To insert nLacZ cDNA into the first ATG of the Fgf10 gene, an nLacZ and FRT-Kanamycin resistant-FRT cassette was engineered and then inserted into the Fgf10 locus, and then the Kanamycin resistance gene was removed by Flp recombinase. To delete +40k to +78k, +78k to +138k, or 1.7 kb at the +45k region, from the Fgf10 RP23-375J2-nLacZ BAC, the Kanamycin or Ampicillin resistant gene was inserted between two homologous arms, and recombined BACs were selected by kanamycin or ampicillin resistance. Oligonucleotide primers used for BAC recombination are listed in SI Materials and Methods.

Plasmid Constructs.

The +31916 to +48073, +48015 to +63052, +78023 to +93052, and 1.7-kb (+43916 to +45638) regions of the Fgf10 gene were cloned into the Ass-hsp-LacZ-pA vector (38). Mutations in Tbx- or Nkx2.5-binding sites in the 1.7-kb sequence were induced by using the QuikChange Multi Site-Directed Mutagenesis Kit (Stratagene). Oligonucleotides used for mutations are listed in SI Materials and Methods.

X-Gal Staining, Whole-Mount in Situ Hybridization, and Immunohistochemistry.

For X-gal staining, embryos were dissected in PBS and then fixed for 10–30 min (depending on the stage) with 4% (wt/vol) paraformaldehyde (PFA) in PBS. Embryos were washed twice with PBS and then stained with X-gal (Roche) for 4–16 h at 37 °C. After developing the blue color, embryos were rinsed in PBS and postfixed in 4% (wt/vol) PFA. Whole-mount in situ hybridization was performed with a digoxigenin-labeled Isl1 probe. Fluorescent immunohistochemistry on sections was carried out with an anti-Islet1 antibody (Developmental Studies Hybridoma Bank). Images were obtained with Apotome Zeiss and Axiovision software.

Electrophoretic Mobility Shift Assay.

Oligonucleotides, which are listed in Fig. 3G, were labeled with digoxigenin (DIG)-11-ddUTP by terminal transferase (Roche). Proteins were prepared from pcDNA3-Nkx2.5-HA or pCITE2a-Isl1 expression vector using TnT Quick Coupled Transcription/Translation Systems (Promega). The Islet1 protein lacks the N-terminal LIM domain, which inhibits DNA binding (16). Labeled oligonucleotides and proteins were incubated in 75 mM NaCl, 1 mM EDTA, 1 mM DTT, 10 mM Tris·Cl (pH 7.5), 6% (wt/vol) glycerol, 2.5 μg BSA, and 1 μg DTT for 30 min on ice, and electrophoresed in a 6% (wt/vol) DNA retardation gel (Invitrogen). After electrophoresis the gel was transferred to Hybond-N+ membrane (GE Life Sciences) and cross-linked, and then the membrane was incubated with anti-DIG antibody conjugated with alkaline phosphatase. The membrane was reacted with CDP-Star (Roche), and signals were detected by X-ray film or the LAS 4000 system (GE Life Sciences).

Chromatin Immunoprecipitation.

E9.5 embryos were dissected, and the SHF, outflow tract/right ventricle, and left ventricle regions were separately collected in cold PBS. ChIP was performed with a LowCell ChIP kit (Diagenode) according to the manufacturer’s protocol. Antibodies used for ChIP were anti-Nkx2.5 (Santa Cruz Biotechnology) and anti-Islet1 [Developmental Studies Hybridoma Bank (against rat Islet1 C terminus) or Abcam (a 300-aa C-terminal peptide from human Islet1)] antibodies. After ChIP, precipitated chromatin DNA was quantitated by PCR with the following primers; forward, 5′-GACTGGCTGCCTTACTATGCAG-3′ and reverse, 5′-GGAGATTATTCAACAAGGTAC-3′, or by CFX96 real-time PCR (BioRad) with the following primers; forward, 5′-GGCTTTACAGCAACATCTATCCAC-3′ and reverse, 5′-GGCTGTTGTGATTAATGGCTTC-3′, and the percentage of the immunoprecipitation was calculated by dividing the amount of immunoprecipitated Fgf10-regulatory region by the amount of input DNA. Independent preparations of extracts from 45 and 48 embryos were used for the ChIP-PCR and ChIP–real-time PCR experiments, respectively. In each experiment real-time PCR was performed in triplicate.