The LIM homeodomain gene Islet-1 (ISL1) encodes a transcription factor that has been associated with the multipotency of human cardiac progenitors, and in mice enables the correct deployment of second heart field (SHF) cells to become the myocardium of atria, right ventricle and outflow tract. Other markers have been identified that characterize subdomains of the SHF, such as the fibroblast growth factor Fgf10 in its anterior region. While functional evidence of its essential contribution has been demonstrated in many vertebrate species, SHF expression of Isl1 has been shown in only some models. We examined the relationship between human ISL1 and FGF10 within the embryonic time window during which the linear heart tube remodels into four chambers. ISL1 transcription demarcated an anatomical region supporting the conserved existence of a SHF in humans, and transcription factors of the GATA family were co-expressed therein. In conjunction, we identified a novel enhancer containing a highly conserved ISL1 consensus binding site within the FGF10 first intron. ChIP and EMSA demonstrated its direct occupation by ISL1. Transcription mediated by ISL1 from this FGF10 intronic element was enhanced by the presence of GATA4 and TBX20 cardiac transcription factors. Finally, transgenic mice confirmed that endogenous factors bound the human FGF10 intronic enhancer to drive reporter expression in the developing cardiac outflow tract. These findings highlight the interest of examining developmental regulatory networks directly in human tissues, when possible, to assess candidate non-coding regions that may be responsible for congenital malformations.
Citation: Golzio C, Havis E, Daubas P, Nuel G, Babarit C, Munnich A, et al. (2012) ISL1 Directly Regulates FGF10 Transcription during Human Cardiac Outflow Formation. PLoS ONE 7(1): e30677. https://doi.org/10.1371/journal.pone.0030677
Editor: Michael Schubert, Ecole Normale Supérieure de Lyon, France
Received: November 8, 2011; Accepted: December 20, 2011; Published: January 27, 2012
Copyright: © 2012 Golzio et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the Fondation pour la Recherche Médicale (grant number DEQ20071210511 and a fellowship to C.G.); the Agence Nationale pour la Recherche (ANR2007-CRANIRARE); the Association Française contre les Myopathies (MNH-Decrypt-6) and the Institut Pasteur-Necker Transverse Research Program. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Congenital heart malformations occur in approximately 3 per 1000 births, more than half of which are potentially lethal malformations of the outflow tract (OFT) . Extensive studies have been undertaken to identify factors driving the differentiation of cell populations that participate in OFT formation in mice and other species, with the expectation that functional data about evolutionarily conserved molecules can be extrapolated to human development.
Two spatially distinct groups of myocardial progenitors, located in the first and the second heart fields, contribute to the definitive heart pump , . The chambers proper are derived from the former, while the outflow segment of the right ventricle and great arteries and the inflow portion of the atria come from the latter. Initially identified in mouse and chick embryos, there appears to be equivalent spatial segregation between progenitor lineages in lower vertebrates without four-chambered hearts, recently identified in frog  and fish .
Coordination between these separate but adjacent mesodermal primordia is orchestrated by signaling events that converge on a common palette of transcription factors necessary for the site-appropriate differentiation of the multiple cell types present in a mature heart. The LIM homeodomain transcription factor Islet-1 (Isl1) is one of these. Isl1 is necessary for multipotent cardiovascular progenitors within the second heart field to proliferate, survive, and migrate into the forming heart. Isl1 is highly conserved over chordate evolution in this role , . Isl1-null mice die at mid-gestation from gross cardiac malformations, notably the lack of the OFT and right ventricle myocardium . Isl1 is also known to be critical for formation and specification of motoneurons  and of the pancreas , acting in combination with other transcription factors to attain specific and context-dependent effects on differentiation .
In the developing heart, these combinatorial partners include members of the tinman (Nkx), GATA-binding and T-box (Tbx) families –, which may derepress and add permissive marks to chromatin . Such associations indeed appear to be stabilized by the preparatory activity of Swi/Snf-like BAF chromatin remodelling complexes expressed precisely within heart precursor primordia, such as Smarcd3 (Baf60c) .
For example, murine Isl1 directly controls the expression of the early mesodermal transcription factors Mef2c and Nkx2-5 during cardiac development via elements in their promoters that also contain nearby, active GATA-binding sites , . In return, human NKX2-5 itself can bind the GATA4 promoter to positively control its transcription during fetal cardiomyocyte differentiation , while forced co-expression of Smarcd3, Gata4 and Tbx5 can induce Isl1 and Nkx2.5 expression in murine mesoderm not normally fated to integrate the heart, leading to cardiac transdifferentiation .
No ISL1 coding mutations have been identified in humans, probably because of an embryonic lethal phenotype for complete inactivation and no gross effect of haploinsufficiency, as seen for murine Isl1 . Heterozygous ISL1 mutations have not directly been reported to cause conotruncal cardiopathies either, although a block of single nucleotide polymorphisms around and within ISL1 have indeed been found to be in linkage disequilibrium with a risk for complex congenital heart phenotypes involving “developmental structures aberrantly formed as derivatives of the secondary [sic] heart field.” .
In Isl1 homozygous knockout mice, the residual hearts no longer express certain bone morphogenetic protein (Bmp) or Wnt family members, Fgf8 or Fgf10, and are missing the OFT entirely . Fgf10, a secreted member of the fibroblast growth factor family, also characterizes the splanchnic mesoderm of the anterior majority of the murine second heart field . In the mouse, its genetic ablation leads to absence of pulmonary arteries and veins, malposition of the heart apex and thin-walled myocardium , ; the absence of the cognate specific receptor isoform for Fgf10, Fgfr2-IIIb, leads in knockout mice to pulmonary vessel aplasia and to OFT malformations such as double outlet right ventricle or ventricular septal defects with overriding aorta . Despite its strong and specific expression in the murine OFT, the function of cardiac Fgf10 has been difficult to ascertain, and its direct transcriptional regulation by Isl1 suggested but not demonstrated in this tissue. Only Tbx1 and Tbx5 have so far been shown to directly bind to and positively regulate Fgf10 expression in the OFT through a 5′ enhancer element , . However, Isl1 and Fgf10 also play early roles in the specification and outgrowth of vertebrate hindlimbs –, while a consensus Isl1-binding site was identified in silico within a 0.4 kb Fgf10 promoter element that is highly conserved among amniotes and capable of directing expression to the otic anlage .
The phenotype of Fgf10-null mice demonstrates the irreplaceable role of Fgf10 in epithelial-mesenchymal interactions needed for the development of many organ systems, including but not restricted to endodermal organs and glands of the head and neck , , . However, there appears to be partial functional redundancy with other Fgf family members, including Fgf3 and Fgf8, in the heart and great vessels –, and different Fgfs in other organ systems such as the inner ear, pituitary and limb buds , . Human heterozygous mutations of FGF10 lead to isolated or syndromic aplasia of the lacrimal and salivary glands and ducts , , not clearly involving the heart, hindgut, ear, pancreas or limbs, that were severely affected in homozygous knockout mice but less so or not at all in heterozygotes. The effect on the lungs is subtle and cumulative in haploinsufficient patients, leading to chronic obstructive pulmonary disease . Like for ISL1, no biallelic inactivation of FGF10 has been found to date in human disease , but the more subtle effects of Fgf10+/− phenotypes have only been described progressively over the years since the first murine knockout models.
The spatiotemporal expression of human ISL1 has recently been demonstrated to be compatible with the existence of a subset of embryonic progenitors that would contribute specifically to the inflow and outflow tracts, as in animal models , or that maintain developmental plasticity at later fetal stages . In this work, we demonstrate not only that ISL1 is co-expressed with other transcription factors in the cardiac primordium, but that in vivo it directly binds and positively regulates the transcription of FGF10. ISL1 exerts this effect through an enhancer within the FGF10 first intron that is evolutionarily conserved among mammals, becomes additionally responsive to ISL1 in vitro in the presence of GATA and TBX factors, and is capable of responding to endogenous cardiac OFT transcription factors in a transgenic mouse reporter.
ISL1 binds a novel intronic element of the FGF10 gene in the human heart but not hindlimb
Recent results from our and other groups have demonstrated the expression of both FGF10 and ISL1 in a region probably corresponding to a second heart field in human embryos at appropriate and similar stages of morphogenesis , . A non-exhaustive bioinformatics analysis of the FGF10 locus to search for putative highly conserved ISL1 consensus binding sites with the sequence YTAATGR, using rVista 2.0 (http://rvista.dcode.org)  and the ECR browser (http://ecrbrowser.dcode.org) , identified two candidate regions conserved among therian mammals (Fig. 1A). One had been previously predicted within the FGF10 promoter  and was also common to birds and amphibians, which we termed FGF10-Pr2; another, within the first intron of FGF10, was termed FGF10-Int1. A third promoter region, without an ISL1 consensus binding site, was designated as FGF10-Pr1. A non-canonical (i.e. 5′-TGATTA-3′) potential binding site for GATA-type transcription factors  was observed 52 nucleotides 5′ to the ISL1 cognate sequence in FGF10-Int1 and these sites were nearly identical in nucleotide composition and distance from one another between mice and humans (Fig. 1B). This attracted our attention to three additional potential sites for homeobox-containing transcription factors and another GATA site, as well as a putative, but less conserved, canonical T-box (Fig. 1C), making all of FGF10-Int1 a candidate cis-regulatory module . All other sites identified, within evolutionarily conserved modules, were 100% identical between species.
A: Alignment of genomic regions around and within the human [hg18] FGF10 locus to those of frog [xenTro2], chicken [galGal3], opossum [monDom4], mouse [mm9], dog [canFam2] and rhesus macaque [rheMac2] with colored regions >90% identical and the vertical scale ranging from 50% (bottom) to 100% (top). Color code for genomic features at http://ecrbrowser.dcode.org/ecrInstructions/ecrInstructions.html. The FGF10-Pr1, FGF10-Pr2 and FGF10-Int1 regions examined in this study are boxed. B: A non-canonical predicted site for GATA-type transcription factors is 52 nucleotides 5′ to the ISL1 cognate sequence in FGF10-Int1 in the direction of transcription on the – strand in humans, mice and (not shown) macaque and opossum. C: Nucleotide sequence of the FGF10-Int1 enhancer module and position of conserved putative transcription factor binding sites as predicted by rVista (http://rvista.dcode.org). All indicated human sites are identical to those of the macaque and mouse except for the SMAD prediction, only found in mouse; the ISL1, GATA and HOXA7 sites are also identical to the opossum, and the ISL1, NKX2-5 and TBX sites are also identical to the dog.
Using chromatin immunoprecipitation (ChIP) of microdissected embryonic human hearts, we demonstrated that at Carnegie stages 14–15 (33–36 dpf), ISL1 bound to and enriched a 327 bp FGF10-Int1 fragment (Fig. 2A). In contrast, ISL1 did not occupy FGF10-Pr1 or FGF10-Pr2. Acetylated histone H4 did bind both the ISL1 and FGF10 promoters at CS14-15, confirming that the chromatin around these two promoters is transcriptionally active in the human heart at these stages (Fig. 2B) .
A: Results of end-point PCR after ChIP using anti-ISL1 or non-specific IgG (or no antibody at all) on chromatin derived from human embryonic hearts at Carnegie stages (CS)14-15. B: Analogous results using anti-acetylated histone H4 compared to a non-specific IgG or no antibody at all, and end-point PCR of regions in the 5′ promoter to human ISL1 and FGF10, demonstrating active availability for transcription. C: FGF10 and ISL1 (and ACTB) were co-expressed at foot plate stages (Carnegie stages [CS]16-17, i.e. 37–43 days of gestation) in human hindlimbs as seen by RT-PCR, while only FGF10 and ACTB were transcribed in forelimbs. D: ChIP using anti-ISL1 on chromatin derived from the C16–17 hindlimb demonstrates no enrichment of the FGF10-Int1 amplicon as compared to the negative control, although this fragment is amplifiable from the total input chromatin.
We also examined whether ISL1 could bind to the FGF10-Int1 element in developing human hindlimb buds, since FGF10 and ISL1 are co-transcribed at foot plate stages at CS16-17 (37–43 dpf; Fig. 2C). While FGF10-Int1 was occupied by ISL1 in the CS14-15 heart, ChIP performed on CS16-17 hindlimbs demonstrated no equivalent binding of ISL1 to FGF10-Int1 (Fig. 2D).
ISL1 and GATA4/5/6 are transcribed in the same temporal window as FGF10
In light of the presence of putative conserved GATA-binding sites in FGF10-Int1, we examined the expression of potential cardiac GATA partners and compared it to that of ISL1 at a range of stages covering the morphogenetic changes from directional S-shaped looping of the primitive cardiac tube to the appearance of four distinct chambers  (Figure S1). RT-PCR of mRNAs extracted from microdissected, staged human heart primordia demonstrated that ISL1, GATA4, GATA5, GATA6, and FGF10, were all expressed at CS13-15 (28–36 dpf). In contrast, these genes were no longer transcribed at CS16 (37–40 dpf), despite continued expression of the ubiquitous ACTB (Fig. 3 inset).
A–H: ISL1 in situ at Carnegie stages (CS)12 (26–28 days post fertilization [dpf]), CS13 (28–31 dpf), CS14 (32–33 dpf) and CS15 (34–36 dpf) respectively. E–H are magnifications of A–D respectively. I–K show GATA4 expression in adjacent sections to B–D. A: ISL1 is expressed at CS12 in foregut endoderm, splanchnic mesoderm, and early motoneurons. B, F: At CS13, ISL1 is transcribed by mesenchyme around the cardiac OFT and pharyngeal arches. ISL1 expression continues in the splanchnic mesoderm between the trachea and OFT, and is visible in dorsal root ganglia, at CS14 (C, G) and CS15 (D, H). I–K: GATA4 is expressed in the endocardium and myocardium of the arterial pole at CS13, CS14 and CS15 (I, J, K respectively). Inset: RT-PCR of ISL1, GATA4, GATA5, GATA6, FGF10 and positive control ACTB mRNAs in embryonic human hearts at stages CS13-16 (to 40 dpf). Abbreviations: drg, dorsal root ganglia; es, esophagus; fb, forebrain; fg, foregut; ph, pharynx; nt, neural tube; oft, OFT; ra, right atrium; t, trachea. Arrows, motoneurons. Bar: 110 µm (A–D, I) and 55 µm (E–H, J, K).
On sections through the heart at CS12, no GATA4 expression was observed in the outflow tract region, while ISL1 hybridization was only visible in the endoderm of the ventral foregut (Fig. 3A). In agreement with the RT-PCR data, at CS13-15 both ISL1 and GATA4 were transcribed within the cardiac mesenchyme surrounding the aorticopulmonary trunk (see magnifications Fig. 3F–H and I–K, respectively). ISL1 also maintained expression at CS15 within the splanchic mesenchyme between the trachea and the heart, while GATA4 appeared restricted to the endocardium and myocardium at CS14-15.
ISL1 and GATA4 each can bind the FGF10-Int1 element in vitro
To investigate the specificity of ISL1 binding to its consensus site within FGF10-Int1, we performed an electrophoretic mobility shift assay (EMSA, Fig. 4). ISL1 bound robustly to its FGF10-ISL1 site, as well as to a previously identified positive control site , termed Insulin I-ISL1 (Fig. 4, lanes 2 and 7 respectively). The FGF10-ISL1 binding was specific, since it could be partially competed off by excess unlabeled probe (Fig. 4, lane 3) but not by a hundredfold excess of unlabeled mutated probe (Fig. 4, lane 4). In addition, ISL1 did not bind to a labeled, scrambled FGF10-ISL1 sequence (Fig. 4, lane 5).
Lane 1, 9, 10: WT FGF10-ISL1 site probe alone, or in conjunction with ISL1 (lane 2) and with unlabeled competitor, which reduces the amount of shifted probe (lane 3), or with ISL1 and unlabeled competitor carrying a mutation in the ISL1 binding site (lane 4). Mutated ISL1 does not shift this probe (lane 5). Lane 6: A validated tandem set of ISL1 binding sites from the insulin promoter shows no gel shift unless ISL1 is added (lane 7) and this shift is reduced in the presence of unlabelled probe competitor (lane 8). Lane 11, 14: WT FGF10-GATA site probe alone, or in conjunction with GATA4 (lane 12) and with unlabeled competitor, which completely abrogates the shift of the probe (lane 13).
In order to verify the affinity of the nearby, non-canonical GATA site in FGF10-Int1 for GATA4, we performed another EMSA, confirming that GATA4 was able to occupy this sequence (Fig. 4, lane 12). Binding to the 5′-TGATTA-3′ site was completely abrogated by the addition of unlabeled FGF10-GATA4 probe (Fig. 4, lane 13).
ISL1 and GATA4 cooperate with TBX20 to activate FGF10 via its intronic enhancer
The transcriptional response of murine Nkx2.5 to the combination of Isl1 and Gata4 in vitro can be potentiated by Tbx20, a member of a large family of genes whose products share a common DNA-binding domain, similar to the T (brachyury) transcription factor . We first determined the ability of ISL1 and/or GATA4 to promote luciferase activity using a reporter with a minimal promoter containing the human cardiac-responsive FGF10-Int1 fragment located 3′ to the luciferase sequence, mimicking the endogenous location of this regulatory element relative to the initiation site for FGF10 transcription. Co-transfection into mesenchymal 10T1/2 cells of a GATA4 or ISL1 expression construct, together with the FGF10-Int1-luciferase reporter, indeed resulted in robust activation of luciferase activity (Fig. 5).
LUC-FGF10-Int1, which construct placed the luciferase gene under the control of the FGF10-Int1 element, was transfected alone or together with ISL1, GATA4 and TBX20 expression vectors into 10T1/2 cells. Each factor alone potentiated luciferase expression and these effects were additive in combination.
We then tested whether this human FGF10-Int1 element could drive expression of a luciferase reporter gene in the presence of ISL1, GATA4, and TBX20 proteins separately as well as in combination. Despite the presence of only a single, non-palindromic T-box binding core motif  within the intronic response element (Fig. 1), transfection of TBX20 in addition to GATA4 and ISL1 expression constructs resulted in additive activation of FGF10-Int1-luc (Fig. 5).
Transgenic mouse embryos express FGF10-Int1-driven reporter in cardiac OFT
The strict sequence conservation between humans and mice, and the ability of transfected murine cells to demonstrate ISL1- and cofactor-driven activation of a reporter gene containing the FGF10-Int1 enhancer in vitro, led us to then test the ability of the element to drive reporter expression when introduced in vivo. FGF10-Int1 was therefore subcloned into the pTK-nlacZ reporter plasmid  and introduced into mouse blastocysts. 43 embryos out of 66 injected were recovered at E8.5, 22 of 53 at E9.5, 46 of 94 at E10.5, and 37 of 59 at E11.5. Of these, nine animals had integrated the transgene, confirmed by PCR, and expressed beta-galactosidase activity: n = 2 at E8.5, n = 3 at E9.5, n = 1 at E10.5 and n = 3 at E11.5.
Labelled cells were observed in the cardiac outflow tract in two of the reporter embryos, at E9.5 and E10.5 respectively, demonstrating the conserved ability of this enhancer to drive gene transcription in both mouse and human hearts. Expression in both cases concerned a few dozen cells, which were not observed in other heart compartments (Fig. 6A–B). Among the positive embryos, a restricted set of additional tissues were also labelled, varying in combinations from one embryo to another in an age-appropriate manner (Table 1). These included the forebrain, the lens, the three first pharyngeal arches, the pancreatic primordia (dorsal and ventral; Fig. 6C), a subset of dorsal root ganglia cells, and motoneurons (Fig. 6D–E). Scattered cells were also positive in the rostral presomitic mesoderm in both E8.5 embryos. Although neither cardiac nor pharyngeal arch expression were visible in the three E11.5 embryos, the tunica media of the internal carotid arteries were positive in one, and the trigeminal and acoustic ganglia were labelled in another. Overall, the sites of transgenic labelling are compatible with activation by Isl1, given what is known about its expression pattern in all of these sites at these stages of development , , , , and thus with its positive regulation of FGF10 transcription in both the human and murine cardiac OFT.
A 1047 bp enhancer region within the first intron of human FGF10, containing multiple transcription factor binding sites including sites validated for ISL1 and GATA4, was placed ahead of a lacZ reporter gene under a thymidine kinase-driven promoter. A: Transgenic mouse at embryonic day (E)10.5, in which expression was activated in dispersed cells of the posterior outflow tract (magnified, insets), in a distal/lateral subdomain of the first two pharyngeal arches, in cells within the trigeminal, acoustic and dorsal root ganglia, and in the lens (right side). B: Same embryo; frontal view. C: Transgenic mouse at E11.5, dorsal and ventral pancreatic primordia. No expression was observed in the limb buds in any injected embryos. In a different transgenic mouse at E11.5, D: motoneuron columns from inner surface of the lumbar spinal cord, and E: cross-section of spinal cord with a labelled subpopulation of cells in the dorsal root ganglia.
We have found that within the first intron of the FGF10 gene there exist highly evolutionarily conserved consensus binding sites for equally conserved transcription factors of the LIM homeodomain, GATA and T box families. These sites are arranged in such a way as to represent a functional cis-regulatory module, with physical spacing between the binding sites that is itself also conserved across species, in particular that between the ISL1 and GATA cognate sites. We have demonstrated that in the human embryonic heart, this module is physically occupied by ISL1 during the period corresponding to the establishment of the cardiac chambers but before septation of the OFT . Binding of ISL1 to the intronic element of FGF10 then ceases in the cardiac OFT, but is never observed in the human or mouse hindlimb bud, for example, where both Isl1 and Fgf10 are expressed shortly thereafter. This observation shows tissue specificity in the function of this binding site and is consistent with the ISL1 expression pattern that we and others  have observed in the human embryonic OFT as well as in the splanchnic mesoderm between CS13-15, as reported in mouse at equivalent morphological stages . Despite a great deal of study of tissue-specific enhancers engaged by Isl1 ,  and the control of Fgf10 expression by transcription factors in the limb  and inner ear , , this is the first report of cis-regulation of FGF10 expression through an intronic element during cardiac development.
In situ hybridization to GATA4 transcripts in adjacent sections demonstrated that at CS12, unlike the morphologically equivalent stage in the mouse , no GATA4 expression was observed in the OFT region. Other subtle differences exist as well between the mouse and human patterns, notably the lack of ISL1 expression outside of the pharyngeal endoderm at CS12, when in the mouse, it is also strongly expressed in the underlying ventral splanchnic mesoderm from an earlier stage . During murine OFT maturation, Isl1- and Gata4-expressing cardiac mesenchyme is also colonized by neural crest cells. However, in the mouse, Isl1 is never expressed in migrating neural crest cells , and Gata4 is rapidly downregulated in both mesectodermal and cardiac neural crest cells . The subpopulation of human ISL1-positive cells in the OFT, that apparently also co-expresses GATA4, is thus likely to be mesodermal in origin. This localization is compatible with the regulation of FGF10.
This conclusion is supported by transgenic mice in which the human FGF10 response element was introduced to drive transcription of a reporter gene, yielding labeled cells in the OFT at stages that morphologically precede cardiac chamber formation. Our complementary in vitro experiments further demonstrated that the single binding site for ISL1 in the 1047 bp FGF10 response element enriched by ChIP is sufficient to drive a three-fold increase in luciferase activity in response to the presence of ISL1 alone. This represents significant and strong activation, since the reporter construct did not contain tandem ISL1 recognition sites but rather preserved the in vivo arrangement of multiple predicted binding sites for conserved transcription factors. Despite the absence of a palindromic T-box consensus site within the intronic response element of FGF10, we obtained transactivation of the reporter, which is in accordance with previous studies showing the response of murine Nkx2.5 to Tbx20 even in the absence of a cognate T-box element . Together with the capacity of GATA4 to transactivate the same reporter in an additive fashion, these results are consistent with a combinatorial action of transcription factors on FGF10 non-coding elements to confer a state of either permission or transcriptional activation to otherwise refractory chromatin.
Among the many dozens of genes highly conserved through evolution and identified as key effectors of animal cardiogenesis, only a handful of them, including a disproportional number of transcription factors (GATA4, NKX2.5, ZIC3, TBX1,TBX20 and CHD7 –), but also intracellular effectors (TAB2 , MID1 ) and ligands (BMP4 ) or membrane-bound proteins (STRA6 , , NOTCH1 , and CFC1 ), have so far been directly linked to congenital heart malformations of the OFT in humans. Mutations in these genes can lead, infrequently and often in association with other developmental anomalies, to persistent truncus arteriosus, double outlet right ventricle, interruption or severe hypoplasia of the aortic arch, tetralogy of Fallot, and valvulopathies. However, there is only partial correspondence between murine and human gene inactivation phenotypes, with many excellent candidate genes through their function in animal model cardiac development not having been found to be mutated in their human counterpart coding sequences.
FGF10 is one of these latter genes, whose cardiac knockout phenotype in the mouse is itself subtle. Based on the murine phenotypes of Fgf10 and Fgfr2-IIIb knockouts and their expression patterns , , we had previously found very similar expression during normal human embryonic development; however, sequencing of both FGF10 and FGFR2-IIIb in human fetuses exhibiting great vessel defects that resembled those in knockout mice, among other symptoms, did not demonstrate coding mutations . The responsible gene turned out to encode a protein, STRA6, necessary to bring vitamin A into cells, a first step in transcriptional regulation through retinoic acid receptor binding , . Retinoic acid, a vitamin A metabolite, normally favors Gata4 transcription and limits the spatial expansion of Isl1, Fgf8 and Fgf10 expression in the SHF –, while it promotes Fgf10 transcription in the burgeoning lungs . Coding mutations in FGF10 lead to phenotypic defects only in the submandibular and lachrymal glands and lungs , , despite being as present as Stra6  in many other organ systems. Similarly, heterozygous missense coding mutations in human FGF8 have been shown to be associated with non-syndromic cleft lip and palate , cause pleiotropic defects in forebrain and pituitary formation , and a recent case of recessive holoprosencephaly with asymptomatic, consanguineous parents has been attributed to hypomorphic alleles of FGF8 ; none of these patients presented cardiac malformations. These observations emphasize the danger of extrapolating findings about the detailed mechanisms of action of highly conserved genes across species, and demonstrate the limits of animal models in understanding human organogenesis.
There is increasing evidence that mutations in non-coding, cis-regulatory elements, controlling transcript availability at a given point time or a given tissue, represent an alternative mechanism leading to human congenital malformations. Such mutations can take the forms of those found for coding sequences, involving single nucleotides  or small or large chromosomal rearrangements . We have discovered an evolutionarily conserved cis-regulatory module in the FGF10 gene that is functional during human cardiac development and that could represent an example of the types of non-coding sites in which mutations may be responsible for morphological aberrations. Taken together, our data reveal unexpected complexity in the transcriptional landscape controlling human cardiogenesis, highlight evolutionary conservation as well as species-specific aspects of cardiac signalling networks, and contribute a strategy to identify additional candidate genomic regions for study in congenital malformations of the OFT.
Materials and Methods
Human embryos were obtained from electively terminated pregnancies, anonymously donated to research after informed written consent from donors in concordance with French legislation (94–654 and 08–400) and with prior approval of the protocol (to M.V.) from the Necker ethical review committee. All mice used in this study were housed under specific pathogen-free conditions at the mouse genetics engineering center (C.I.G.M.) of the Pasteur Institute, Paris, under authorization number A75-15-09 from the Paris Departmental Directorate for the Protection of Populations and handled in accordance with French and European directives.
ChIP was carried out as previously described, starting from nuclear isolation , using eleven microdissected and flash-frozen cardiac tubes from human embryos at Carnegie stages (CS) 14–15 . An anti-ISL1 (10 µL, Santa Cruz Sc-23590X) or an anti-GFP antibody as negative control (10 µL, Abcam ab1218), were used per 10 µg of sonicated chromatin. Immunoprecipitated DNA was analysed by end-point PCR (primers, Supplementary Table S1).
ISL1 and GATA4 in situ hybridizations were performed using transverse sections of normal human embryos from CS12 to 15. Tissue fixation, sectioning, and in situ hybridization were carried out as previously described . Total RNA was extracted from pooled whole hearts at individual stages from CS13 to CS16 and RT-PCR was carried out using the GeneAmp kit (Roche), with 500 ng total RNA input for first strand synthesis (primers, Supplementary Table S1).
Expression constructs and electrophoretic mobility shift assays (EMSA)
Human TBX20 and ISL1 expression vectors were generated. Full-length TBX20 cDNA and a fragment of ISL1 cDNA with the N-terminal 142 amino acids removed  were inserted into the multiple cloning site of pcDNA3.1C (Invitrogen). Full-length human GATA4 cDNA was purchased from GenScript (GN026113). HeLa cells were transfected with these constructs, and nuclear protein extracts were made using standard protocols. The LightShift Chemiluminescent EMSA Kit (Pierce) was used as specified. Primers are listed in Supplementary Table S1.
Transactivation assays and reporter constructs
For the FGF10 reporter construct (LUC-FGF10-Int1), 1047 bp of the FGF10 first intron (NCBI36/hg18 chromosome 5:44421556–44422602) were subcloned into the BamHI site 3′ to luc+ in pGL3 (Promega). Mouse 10T1/2 cells  in DMEM/10% fetal calf serum were transfected with FuGene HD (Roche). Cells were harvested and lysed 24 h after transfection. Firefly and Renilla luciferase activities were measured on a Berthold Centro LB960 using the Dual-Luciferase Reporter assay system (Promega). Firefly luciferase activity was normalized to the Renilla luciferase internal control, pRL-CMV (Promega). Experiments were repeated in triplicate in three independent assays.
The same 1047 bp FGF10-Int1 fragment as in the transactivation assays was subcloned into the BamHI site of the pSKT-TK-nLacZ plasmid  and orientation verified by capillary sequencing with a standard T3 primer. The plasmid was linearized with SalI for injection at 2 ng/mL into mouse blastocysts. β-galactosidase-containing cells that had transcribed the reporter plasmid were stained in whole mount by the catalysis of the X-gal (5-bromo-4-chloro-3-indolyl β-D-galactopyranoside) substrate.
Composite image of embryonic hearts at stages ranging from the beginning of the fourth to the ninth week of human gestation (upper left to lower right, Carnegie stages 10–23). Rostral to top. Congenital heart and great vessel malformations arise during this time window when molecular signaling between cardiac progenitors and their environment is impaired.
The authors thank Dr. M. Téboul and the Service d'Orthogénie of the Broussais Hospital in Paris for the human tissues and Dr. D. Montarras for the 10T1/2 cells used in this work. Dr F. Langa Vives provided invaluable assistance in the generation of transgenic mice at the Centre d'Ingénierie Génétique Murine, Institut Pasteur, Paris. Drs. D. Bonnet, M. Buckingham, C. Fournier-Thibault, and R. Kelly provided invaluable discussion.
Conceived and designed the experiments: CG EH GN SZ HCE. Performed the experiments: CG EH PD GN CB HCE. Analyzed the data: CG PD GN SZ HCE. Contributed reagents/materials/analysis tools: EH PD GN AM MV SL SZ. Wrote the paper: CG EH SL SZ HCE.
- 1. Hoffman JI, Kaplan S (2002) The incidence of congenital heart disease. J Am Coll Card 39: 1890–1900.
- 2. Waldo KL, Kumiski DH, Wallis KT, Stadt Ha, Hutson MR, et al. (2001) Conotruncal myocardium arises from a secondary heart field. Development 128: 3179–3188.
- 3. Kelly RG, Brown NA, Buckingham ME (2001) The arterial pole of the mouse heart forms from Fgf10-expressing cells in pharyngeal mesoderm. Dev Cell 1: 435–440.
- 4. Brade T, Gessert S, Kühl M, Pandur P (2007) The amphibian second heart field: Xenopus islet-1 is required for cardiovascular development. Dev Biol 311: 297–310.
- 5. Zhou Y, Cashman TJ, Nevis KR, Obregon P, Carney Sa, et al. (2011) Latent TGF-β binding protein 3 identifies a second heart field in zebrafish. Nature 474: 645–648.
- 6. Stolfi A, Gainous TB, Young JJ, Mori A, Levine M, et al. (2010) Early chordate origins of the vertebrate second heart field. Science 329: 565–568.
- 7. Cai C-L, Liang X, Shi Y, Chu P-H, Pfaff SL, et al. (2003) Isl1 identifies a cardiac progenitor population that proliferates prior to differentiation and contributes a majority of cells to the heart. Dev Cell 5: 877–889.
- 8. Lee SK, Pfaff SL (2003) Synchronization of neurogenesis and motor neuron specification by direct coupling of bHLH and homeodomain transcription factors. Neuron 38: 731–745.
- 9. Ahlgren U, Pfaff SL, Jessell TM, Edlund T, Edlund H (1997) Independent requirement for ISL1 in formation of pancreatic mesenchyme and islet cells. Nature 385: 257–260.
- 10. Dodou E, Verzi MP, Anderson JP, Xu S-M, Black BL (2004) Mef2c is a direct transcriptional target of ISL1 and GATA factors in the anterior heart field during mouse embryonic development. Development 131: 3931–3942.
- 11. Sizarov A, Ya J, de Boer Ba, Lamers WH, Christoffels VM, et al. (2011) Formation of the building plan of the human heart: morphogenesis, growth, and differentiation. Circulation 123: 1125–1135.
- 12. Stennard FA, Costa MW, Elliott DA, Rankin S, Haast SJP, et al. (2003) Cardiac T-box factor Tbx20 directly interacts with Nkx2-5, GATA4, and GATA5 in regulation of gene expression in the developing heart. Dev Biol 262: 206–224.
- 13. Miller SA, Huang AC, Miazgowicz MM, Brassil MM, Weinmann AS (2008) Coordinated but physically separable interaction with H3K27-demethylase and H3K4-methyltransferase activities are required for T-box protein-mediated activation of developmental gene expression. Genes Dev 22: 2980–2993.
- 14. Lickert H, Takeuchi JK, Von Both I, Walls JR, McAuliffe F, et al. (2004) Baf60c is essential for function of BAF chromatin remodelling complexes in heart development. Nature 432: 107–112.
- 15. Takeuchi JK, Mileikovskaia M, Koshiba-Takeuchi K, Heidt AB, Mori AD, et al. (2005) Tbx20 dose-dependently regulates transcription factor networks required for mouse heart and motoneuron development. Development 132: 2463–2474.
- 16. Riazi AM, Takeuchi JK, Hornberger LK, Zaidi SH, Amini F, et al. (2009) NKX2-5 regulates the expression of beta-catenin and GATA4 in ventricular myocytes. PLoS One 4: e5698.
- 17. Takeuchi JK, Bruneau BG (2009) Directed transdifferentiation of mouse mesoderm to heart tissue by defined factors. Nature 459: 708–711.
- 18. Stevens KN, Hakonarson H, Kim CE, Doevendans PA, Koeleman BPC, et al. (2010) Common variation in ISL1 confers genetic susceptibility for human congenital heart disease. PLoS ONE 5: e10855.
- 19. Marguerie A, Bajolle F, Zaffran S, Brown NA, Dickson C, et al. (2006) Congenital heart defects in Fgfr2-IIIb and Fgf10 mutant mice. Cardiovasc Res 71: 50–60.
- 20. Vega-Hernández M, Kovacs A, De Langhe S, Ornitz DM (2011) FGF10/FGFR2b signaling is essential for cardiac fibroblast development and growth of the myocardium. Development 138: 3331–3340.
- 21. Xu H, Morishima M, Wylie JN, Schwartz RJ, Bruneau BG, et al. (2004) Tbx1 has a dual role in the morphogenesis of the cardiac outflow tract. Development 131: 3217–3227.
- 22. Agarwal P, Wylie JN, Galceran J, Arkhitko O, Li C, et al. (2003) Tbx5 is essential for forelimb bud initiation following patterning of the limb field in the mouse embryo. Development 130: 623–633.
- 23. Min H, Danilenko DM, Scully SA, Bolon B, Ring BD, et al. (1998) Fgf-10 is required for both limb and lung development and exhibits striking functional similarity to Drosophila branchless. Genes Dev 12: 3156–3161.
- 24. Sekine K, Ohuchi H, Fujiwara M, Yamasaki M, Yoshizawa T, et al. (1999) Fgf10 is essential for limb and lung formation. Nat Genet 21: 138–141.
- 25. Yang L, Cai C-L, Lin L, Qyang Y, Chung C, et al. (2006) Isl1Cre reveals a common Bmp pathway in heart and limb development. Development 133: 1575–1585.
- 26. Ohuchi H, Yasue A, Ono K, Sasaoka S, Tomonari S, et al. (2005) Identification of cis-element regulating expression of the mouse Fgf10 gene during inner ear development. Dev Dyn 233: 177–187.
- 27. Ohuchi H, Hori Y, Yamasaki M, Harada H, Sekine K, et al. (2000) FGF10 acts as a major ligand for FGF receptor 2 IIIb in mouse multi-organ development. Biochem Biophys Res Comm 277: 643–649.
- 28. Fairbanks T (2004) Fibroblast growth factor 10 (Fgf10) invalidation results in anorectal malformation in mice. J Ped Surg 39: 360–365.
- 29. Urness LD, Bleyl SB, Wright TJ, Moon AM, Mansour SL (2011) Redundant and dosage sensitive requirements for Fgf3 and Fgf10 in cardiovascular development. Dev Biol 356: 383–397.
- 30. Watanabe Y, Miyagawa-Tomita S, Vincent SD, Kelly RG, Moon AM, et al. (2010) Role of mesodermal FGF8 and FGF10 overlaps in the development of the arterial pole of the heart and pharyngeal arch arteries. Circ Res 106: 495–503.
- 31. Vitelli F, Taddei I, Morishima M, Meyers EN, Lindsay EA, et al. (2002) A genetic link between Tbx1 and fibroblast growth factor signaling. Development 129: 4605–4611.
- 32. Liu W, Levi G, Shanske A, Frenz DA (2008) Retinoic acid-induced inner ear teratogenesis caused by defective Fgf3/Fgf10-dependent Dlx5 signaling. Birth defects Res Part B 83: 134–144.
- 33. Herzog W, Sonntag C, von der Hardt S, Roehl HH, Varga ZM, et al. (2004) Fgf3 signaling from the ventral diencephalon is required for early specification and subsequent survival of the zebrafish adenohypophysis. Development 131: 3681–3692.
- 34. Entesarian M, Matsson H, Klar J, Bergendal B, Olson L, et al. (2005) Mutations in the gene encoding fibroblast growth factor 10 are associated with aplasia of lacrimal and salivary glands. Nat Genet 37: 125–127.
- 35. Rohmann E, Brunner HG, Kayserili H, Uyguner O, Nürnberg G, et al. (2006) Mutations in different components of FGF signaling in LADD syndrome. Nat Genet 38: 414–417.
- 36. Klar J, Blomstrand P, Brunmark C, Badhai J, Håkansson HF, et al. (2011) Fibroblast growth factor 10 haploinsufficiency causes chronic obstructive pulmonary disease. J Med Genet 48: 705–709.
- 37. Martinovic-Bouriel J, Bernabé-Dupont C, Golzio C, Grattagliano-Bessières B, Malan V, et al. (2007) Matthew-Wood syndrome: report of two new cases supporting autosomal recessive inheritance and exclusion of FGF10 and FGFR2. Am J Med Genet Part A 143: 219–228.
- 38. Genead R, Danielsson C, Wärdell E, Kjaeldgaard A, Westgren M, et al. (2010) Early first trimester human embryonic cardiac Islet-1 progenitor cells and cardiomyocytes: Immunohistochemical and electrophysiological characterization. Stem Cell Res 4: 69–76.
- 39. Loots GG, Ovcharenko I (2004) rVISTA 2.0: evolutionary analysis of transcription factor binding sites. Nucleic Acids Res 32: W217–W221.
- 40. Ovcharenko I, Nobrega MA, Loots GG, Stubbs L (2004) ECR Browser: A tool for visualizing and accessing data from comparisons of multiple vertebrate genomes. Nucleic Acids Res 32: W280–W286.
- 41. Merika M, Orkin SH (1993) DNA-Binding Specificity of GATA Family Transcription Factors. Mol Cell Biol 13: 3999–4010.
- 42. Blanchette M, Bataille AR, Chen X, Poitras C, Laganière J, et al. (2006) Genome-wide computational prediction of transcriptional regulatory modules reveals new insights into human gene expression. Genome Res 16: 656–668.
- 43. Vettese-Dadey M, Grant PA, Hebbes TR, Crane- Robinson C, Allis CD, et al. (1996) Acetylation of histone H4 plays a primary role in enhancing transcription factor binding to nucleosomal DNA in vitro. EMBO J 15: 2508–2518.
- 44. Moorman A, Webb S, Brown NA, Lamers W, Anderson RH (2003) Development of the heart: (1) formation of the cardiac chambers and arterial trunks. Heart 89: 806–814.
- 45. Conlon FL, Fairclough L, Price BMJ, Casey ES, Smith JC (2001) Determinants of T box protein specificity. Development 128: 3749–3758.
- 46. Hadchouel J, Carvajal JJ, Daubas P, Bajard L, Chang T, et al. (2003) Analysis of a key regulatory region upstream of the Myf5 gene reveals multiple phases of myogenesis, orchestrated at each site by a combination of elements dispersed throughout the locus. Development 130: 3415–3426.
- 47. Pfaff SL, Mendelsohn M, Stewart CL, Edlund T, Jessell TM (1996) Requirement for LIM homeobox gene Isl1 in motor neuron generation reveals a motor neuron-dependent step in interneuron differentiation. Cell 84: 309–320.
- 48. Yuan S, Schoenwolf GC (2000) Islet-1 marks the early heart rudiments and is asymmetrically expressed during early rotation of the foregut in the chick embryo. AnatRec 260: 204–207.
- 49. Snarr BS, O'Neal JL, Chintalapudi MR, Wirrig EE, Phelps AL, et al. (2007) Isl1 expression at the venous pole identifies a novel role for the second heart field in cardiac development. Circ Res 101: 971–974.
- 50. Kawakami Y, Marti M, Kawakami H, Itou J, Quach T, et al. (2011) Islet1-mediated activation of the beta-catenin pathway is necessary for hindlimb initiation in mice. Development 138: 4465–4473.
- 51. Sasaki H, Yamaoka T, Ohuchi H, Yasue A, Nohno T, et al. (2002) Identification of cis-elements regulating expression of Fgf10 during limb development. The IntJ Dev Biol 46: 963–967.
- 52. Lilleväli K, Haugas M, Matilainen T, Pussinen C, Karis A, et al. (2006) Gata3 is required for early morphogenesis and Fgf10 expression during otic development. Mech Dev 123: 415–429.
- 53. Rojas A, De Val S, Heidt AB, Xu SM, Bristow J, et al. (2005) Gata4 expression in lateral mesoderm is downstream of BMP4 and is activated directly by Forkhead and GATA transcription factors through a distal enhancer element. Development 132: 3405–3417.
- 54. Tomita Y, Matsumura K, Wakamatsu Y, Matsuzaki Y, Shibuya I, et al. (2005) Cardiac neural crest cells contribute to the dormant multipotent stem cell in the mammalian heart. J Cell Biol 170: 1135–1146.
- 55. Goldmuntz E, Geiger E, Benson DW (2001) NKX2.5 mutations in patients with tetralogy of Fallot. Circulation 104: 2565–2568.
- 56. Posch MG, Perrot A, Schmitt K, Mittelhaus S, Esenwein EM, et al. (2008) Mutations in GATA4, NKX2.5, CRELD1, and BMP4 are infrequently found in patients with congenital cardiac septal defects. Am J Med Genet A 146A: 251–253.
- 57. Megarbane A, Salem N, Stephan E, Ashoush R, Lenoir D, et al. (2000) X-linked transposition of the great arteries and incomplete penetrance among males with a nonsense mutation in ZIC3. Eur J Hum Genet 8: 704–708.
- 58. Gong W (2001) Mutation analysis of TBX1 in non-deleted patients with features of DGS/VCFS or isolated cardiovascular defects. J Med Genet 38: 45e–45.
- 59. Kirk EP, Sunde M, Costa MW, Rankin SA, Wolstein O, et al. (2007) Mutations in cardiac T-box factor gene TBX20 are associated with diverse cardiac pathologies, including defects of septation and valvulogenesis and cardiomyopathy. Am J Hum Genet 81: 280–291.
- 60. Vissers LELM, van Ravenswaaij CMA, Admiraal R, Hurst JA, de Vries BBA, et al. (2004) Mutations in a new member of the chromodomain gene family cause CHARGE syndrome. Nat Genet 36: 955–957.
- 61. Thienpont B, Zhang L, Postma AV, Breckpot J, Tranchevent L-C, et al. (2010) Haploinsufficiency of TAB2 causes congenital heart defects in humans. Am J Hum Genet 86: 839–849.
- 62. Pinson L, Auge J, Audollent S, Mattei G, Etchevers H, et al. (2004) Embryonic expression of the human MID1 gene and its mutations in Opitz syndrome. J Med Genet 41: 381–386.
- 63. Pasutto F, Sticht H, Hammersen G, Gillessen-Kaesbach G, Fitzpatrick DR, et al. (2007) Mutations in STRA6 cause a broad spectrum of malformations including anophthalmia, congenital heart defects, diaphragmatic hernia, alveolar capillary dysplasia, lung hypoplasia, and mental retardation. Am J Hum Genet 80: 550–560.
- 64. Golzio C, Martinovic-Bouriel J, Thomas S, Mougou-Zrelli S, Grattagliano-Bessieres B, et al. (2007) Matthew-Wood syndrome is caused by truncating mutations in the retinol-binding protein receptor gene STRA6. Am J Hum Genet 80: 1179–1187.
- 65. McBride KL, Riley MF, Zender Ga, Fitzgerald-Butt SM, Towbin Ja, et al. (2008) NOTCH1 mutations in individuals with left ventricular outflow tract malformations reduce ligand-induced signaling. HumMol Genet 17: 2886–2893.
- 66. Goldmuntz E, Bamford R, Karkera JD, dela Cruz J, Roessler E, et al. (2002) CFC1 mutations in patients with transposition of the great arteries and double-outlet right ventricle. Am J Hum Genet 70: 776–780.
- 67. Ryckebusch L, Wang Z, Bertrand N, Lin S-C, Chi X, et al. (2008) Retinoic acid deficiency alters second heart field formation. Proc Natl Acad Sci U S A 105: 2913–2918.
- 68. Waxman JS, Keegan BR, Roberts RW, Poss KD, Yelon D (2008) Hoxb5b acts downstream of retinoic acid signaling in the forelimb field to restrict heart field potential in zebrafish. Dev Cell 15: 923–934.
- 69. Kostetskii I, Jiang Y, Kostetskaia E, Yuan S, Evans T, et al. (1999) Retinoid signaling required for normal heart development regulates GATA-4 in a pathway distinct from cardiomyocyte differentiation. Dev Biol 206: 206–218.
- 70. Desai TJ, Malpel S, Flentke GR, Smith SM, Cardoso WV (2004) Retinoic acid selectively regulates Fgf10 expression and maintains cell identity in the prospective lung field of the developing foregut. Dev Biol 273: 402–415.
- 71. Bouillet P, Sapin V, Chazaud C, Messaddeq N, Decimo D, et al. (1997) Developmental expression pattern of Stra6, a retinoic acid-responsive gene encoding a new type of membrane protein. Mech Dev 63: 173–186.
- 72. Riley BM, Mansilla MA, Ma J, Daack-Hirsch S, Maher BS, et al. (2007) Impaired FGF signaling contributes to cleft lip and palate. Proc Natl Acad Sci U S A 104: 4512–4517.
- 73. Arauz RF, Solomon BD, Pineda-Alvarez DE, Gropman AL, Parsons JA, Roessler E, Muenke M (2010) A hypomorphic allele in the FGF8 gene contributes to holoprosencephaly and is allelic to gonadotropin-releasing hormone deficiency in humans. Mol Syndromol 1: 59–66.
- 74. McCabe MJ, Gaston-Massuet C, Tziaferi V, Gregory LC, Alatzoglou KS, et al. (2011) Novel FGF8 mutations associated with recessive holoprosencephaly, craniofacial defects, and hypothalamo-pituitary dysfunction. J Clin Endocrinol Metab 96: E1709–18.
- 75. Benko S, Fantes JA, Amiel J, Kleinjan D-J, Thomas S, et al. (2009) Highly conserved non-coding elements on either side of SOX9 associated with Pierre Robin sequence. Nat Genet 41: 359–364.
- 76. VanderMeer JE, Ahituv N (2011) cis-regulatory mutations are a genetic cause of human limb malformations. Dev Dyn 240: 920–930.
- 77. Havis E, Anselme I, Schneider-Maunoury S (2006) Whole embryo chromatin immunoprecipitation protocol for the in vivo study of zebrafish development. Biotechniques 40: 34, 36, 38 passim.
- 78. O'Rahilly R, Müller F (1987) Developmental Stages in Humans: including a revision of Streeter's “Horizons” and a survey of the Carnegie collection. Washington, D.C.: Carnegie Institution of Washington.
- 79. Delous M, Baala L, Salomon R, Laclef C, Vierkotten J, et al. (2007) The ciliary gene RPGRIP1L is mutated in cerebello-oculo-renal syndrome (Joubert syndrome type B) and Meckel syndrome. Nat Genet 39: 875–881.
- 80. Sanchez-Garcia I, Rabbitts TH (1993) Redox regulation of in vitro DNA-binding activity by the homeodomain of the Isl-1 protein. J Mol Biol 231: 945–949.
- 81. Reznikoff CA, Brankow DW, Heidelberger C (1973) Establishment and characterization of a cloned line of C3H mouse embryo cells sensitive to postconfluence inhibition of division. Cancer Res 33: 3231–3238.