Abstract
In lymphocytes, the expression of early immune response genes is regulated by NF-AT transcription factors1,2 which translocate to the nucleus after dephosphorylation by the Ca2+-dependent phosphatase, calcineurin3. We report here that mice bearing a disruption in the NF-ATc gene fail to develop normal cardiac valves and septa and die of circulatory failure before day 14.5 of development. NF-ATc is first expressed in the heart at day 7.5, and is restricted to the endocardium, a specialized endothelium that gives rise to the valves and septum. Within the endocardium, specific inductive events appear to activate NF-ATc: it is localized to the nucleus only in endocardial cells that are adjacent to the interface with the cardiac jelly and myocardium, which are thought to give the inductive stimulus to the valve primordia4. Treatment of wild-type embryos with FK506, a specific calcineurin inhibitor5, prevents nuclear localization of NF-ATc. These data indicate that the Ca2+/calcineurin/NF-ATc signalling pathway is essential for normal cardiac valve and septum morphogenesis; hence, NF-ATc and its regulatory pathways are candidates for genetic defects underlying congenital human heart disease.
Phosphorylation of amino-terminal serine residues controls transcriptional activity of NF-AT family members by influencing their nuclear localization6. In NF-ATc these amino acids are encoded by exon 3, and we deleted this by homologous recombination in embryonic stem (ES) cells (Fig. 1a, b). No homozygous mutant offspring was identified among 91 pups from heterozygous intercrosses. At E12.5, 50% of the NF-ATc3 mutant embryos were anaemic, oedematous and showed signs of abdominal necrosis (Fig. 1c), and died by E14.5.
Figure 1: Targeting of the NF-ATc gene.
a, Top: structure of the genomic DNA encoding exon 3 of NF-ATc. Arrows depict PCR primers used to identify the wild-type allele. Middle: targeting vector designed to replace exon 3 by a neo cassette. Bottom: targeted locus. Arrows depict PCR primers used to identify the mutant allele. A, ApaI; E, EcoRI; H, HindIII; P, PstI; X, XbaI. Bar, 5' flanking probe. b,Southern blot of EcoRI-digested genomic DNA from wild-type (+/+) and heterozygous (+/-) ES cell lines hybridized with a 5' flanking probe. The wild-type (wt) and mutant (mt) alleles bands are indicated by arrows. c, E12.5 wild-type (left) and NF-ATc3 mutant (right) embryos. The arrowhead points to the necrotic abdominal region in the mutant embryo. Scale bar, 500 
m. d, Northern blot of RNA isolated from NF-ATc+/- and NF-ATc3 E11.5 embryos, hybridized with a 3' NF-ATc probe. e, Western blot performed with the MAb 7A6 (ref. 20) on E11.5 NF-ATc+/-, NF-ATc3 mutant embryos and mature T cells (T).
Mutant embryos do not express any NF-ATc transcript (Fig. 1d) and lack the epitope encoded by exon 3 (Fig. 1e). During embryogenesis, NF-ATc messenger RNA was restricted to the heart (Fig. 2), and was initially found in the endocardial tubes of the E7.5 embryo (Fig. 2a), and in the endocardium and vitelline vein by E8.5 (Fig. 2b). Between E9 and E11.5, expression was observed in the endocardium of the ventricles, atrium, outflow tract (Fig. 2c–f, h) and the apex of the ventricular septum (Fig. 2h). During E11.5–E13.5, NF-ATc was found in the lining of the pulmonary (Fig. 2i), aortic (Fig. 2j) and atrioventricular (AV) valves (Fig. 2k) and cushion (Fig.2l). Sections of E11.5 NF-ATc3 mutant embryos did not show any detectable signal (Fig. 2m), indicating that the NF-ATc3 is a null mutation.
Figure 2: NF-ATc expression is restricted to the endocardium, outflow tract and cardiac valves.
a–f, Whole-mount in situ hybridization. a, Frontal view of E7.5 wild-type embryo. NF-ATc signal (purple) is restricted to the endocardial tubes (e). b,E8.5 embryo. NF-ATc is in the endocardium (e) and vitelline vein (v). c, d, E9.0 embryo. NF-ATc is in the endocardium of the ventricle (arrow in d) and outflow tract (arrowhead). e, f, E10.5 embryo. NF-ATc mRNA detected in the ventricle (arrow in f) and atrium (arrowhead). Embryos were hybridized with a NF-ATc exon 3 antisense probe. g, Cartoon showing the approximate plane of section indicated by Roman numbers in the bottom right of h–m. Radioactive in situ hybridization showing NF-ATc expression in E11.5 and E13.5 wild-type (h–l) and in E11.5 NF-ATc3 mutant embryos (m) in dark field views. h, E11.5. NF-ATc is in the outflow tract (arrow) and lining the apex of the ventricular septum (arrowhead). i, j, E13.5. NF-ATc localizes to the developing pulmonary and aortic valves (arrowheads). k, E11.5. NF-ATc expression in the AV valves endocardium (arrowhead). l, NF-ATc expression in the endocardial lining of the cushion (arrowhead). m, E11.5. NF-ATc3 mutant embryo. No NF-ATc signal is detected. Sections were hybridized with a NF-ATc 3' antisense probe. The bright cells in the atrial and ventricular cavities are unlabelled refracting red blood cells. Bar, 400 m in a, 500 m in b, 600 m in c, 200 m in d, 1.3 mm in e, 300 m in f, 250 m in h–m.
To assess whether NF-ATc was transcriptionally active in endocardial cells, its subcellular localization was investigated by immunofluorescence. At E10.5, nuclear NF-ATc delineated the endocardium of the heart, that was also stained by the endothelial marker PECAM-1 (CD31)7 (Fig. 3a–c). Nuclear NF-ATc is found primarily in cells lining the endocardial cushions, the valve and septal primordia, and along the lining of the bulbus cordis (data not shown and Fig. 3d, e). Cells exhibiting nlear versus cytoplasmic staining often lay adjacent to one another (Fig. 3e), suggesting that short-range signals may induce NF-ATc activation in specific endocardial cells. No protein is detected in mesenchymal cells within the cushion, indicating that cells undergoing the epithelial to mesenchymal transition leading to cushion formation4 downregulate NF-ATc expression. By E11.5, nuclear NF-ATc becomes restricted to the lining of the outflow tract (Fig. 3f, g) and developing valves (Fig. 3h). At E12.5, only cells which line the immediate valve surfaces (Fig. 3i) and the ends of the expanding septa (not shown) exhibited nuclear NF-ATc staining. No staining was found in sections through adult and newborn heart (data not shown), indicating that NF-ATc activity is selectively required during embryonic cardiac development.
Figure 3: Active NF-ATc is expressed in a subset of endothelial cells lining the endocardial cushions and valves.
Wild-type E10.5-E12.5 sections stained with anti-NF-ATc and anti-PECAM-1 (a–c), or anti-NF-ATc alone (d–i). a, b, f, DAPI counterstaining. a, General view of the atrioventricular (AV) canal region at E10.5. Note the complementary nuclear NF-ATc (red fluorescence, arrowhead) and cytoplasmic PECAM-1 stainings (arrow) in the AV canal (b) and lumen of the atrium (c). PECAM-1 staining appears yellow because of the overlap of the green and red fluorescence of minoritary cytoplasmic NF-ATc staining. E10.5. Section through the base of the bulbus cordis (d), showing NF-ATc nuclear signal (e). C, cytoplasmic; n, nuclear. f, g, Nuclear localization of NF-ATc in the outflow tract (g, arrowhead) and semilunar valve (h, arrowhead) of E11.5 embryos. i, E12.5 embryo showing nuclear NF-ATc in the AV canal (large arrowhead) and in the forming mitral valve (arrowhead). Scale bar, 63 m in a, 25 m in b, d, e–h, 16 m in c
NF-ATc3 mutant embryos showed cardiac abnormalities by E12.5. At this time, the outflow tract in the wild-type embryo is divided into the aortic sac and pulmonary trunk, and the ventricular chambers show a complex pattern of myocardial trabeculation (Fig.4a). In 40% of the NF-ATc3 mutant embryos, the ventricular walls were hypertrophied and the chambers were small (Fig. 4b). In these embryos, the proximal region of the outflow tract (bulbus) was severely narrowed or occluded (Fig. 4b, arrowhead), and the aortic sac had thick walls and displayed a narrow lumen that disappeared as the sac approached the conus arteriosus (data not shown). By E13.5, the NF-ATc3 mutant embryos displayed abnormalities of valve structure (Fig. 4d–f). Although septation into right ventricular (pulmonary) and left ventricular (aortic) outflow tract was normal, both semilunar valves were underdeveloped, presenting the poorly organized appearance of earlier stages (Fig.4d, f). The tricuspid and mitral valves were also primitive in comparison to wild type and displayed no well defined leaflets (compare Fig. 4g with h). In addition, an interventricular foramen was still present in most NF-ATc3 mutant embryos (Fig. 4i), indicating defective septum formation. Mice generated from the two independently targeted ES cells showed similar phenotypes.
Figure 4: Cardiac valves and ventricular septum defects in NF-ATc3 mutant embryos.
H+ E stainings of transversal sections from wild-type (a, c, e, g) and NF-ATc3 mutant (b, d, f, h) embryos. Roman numbers (bottom right) indicate the approximate plane of section according to Fig. 2g. Section through the outflow tract (a, b) showing the stenosis in E12.5 mutant embryos (arrowhead in b) and the reduced lumen of the aortic sac (as). E13.5 embryos sectioned through the pulmonary (c, d, arrowheads), aortic (e, f, arrowheads) and atrioventricular (AV) (g, h, arrowheads) valves. i, E13.5 NF-ATc3 mutant. A septum foramen is present (arrowhead). j–l, E11.5 NF-ATc3 mutant embryos. Immunostaining showing PECAM-1 expression lining the endocardium (j, arrowhead). Radioactive in situ hybridization showing c-act expression (k) in the myocardium (m) and Sox-4 expression (l) in endocardial cushion (ec). Abbreviations: la, left atrium; lv, left ventricule; ra, right atrium; rv, right ventricule; ro, right ventricular outflow; vs, ventricular septum. Scale bar, 250 m.
Vascular endothelium, myocardium and extracellular matrix appeared unaffected in NF-ATc3 mutant embryos, as indicated by the normal expression of PECAM-1 (Fig. 4j), cardiac-actin (Fig. 4k) and fibronectin (data not shown). Although the neural crest participates in the formation of the pulmonary and aortic valves8, NF-ATc is exclusively expressed in the endocardium; thus it is unlikely that defective neural crest differentiation is involved in the abnormal valve development of NF-ATc3 mutant embryos.
Death of NF-ATc3 mutant mice is probably caused by severe defects in forward blood flow resulting from valvular incompetence, stenosis and conotruncal abnormalities. The peripheral oedema and ventricular hypertrophy may be secondary to congestive heart failure and valvular malfunction.
Morphogenesis of the endocardial cushion is induced by sustained signals from the AV myocardium that cause cell migration and differentiation into cushion tissue mesenchyme4. These events can be mimicked by calcium ionophores and activators of protein kinase C9. To determine if the nuclear localization of NF-ATc in the embryonic endocardium is due to Ca2+/calcineurin signalling, we cultured embryos in the presence or absence of FK506. Control embryos showed NF-ATc nuclear localization selectively in those cells that lie adjacent to the myocardium and cardiac jelly (Fig. 5a). In contrast, embryos treated with FK506 showed cytoplasmic localization of NF-ATc in all endocardial cells (Fig. 5b). These results suggest that the nuclear localization of NF-ATc in the endocardium is related to the local activation of calcineurin, and implicate the Ca2+/calcineurin signalling pathway in cardiac valve and septum development. In this pathway, the upstream activator of NF-ATc is almost certainly calcineurin. However, the ligand and receptor that lead to NF-ATc nuclear localization, and its downstream targets, are unknown. The expression of Sox-4, a member of the Sry family of transcription factors10 that is required for cardiac valve development and lymphoid activation11 was normal in NF-ATc3 mutant embryos (Fig. 4l). This indicates that Sox-4 is not likely to be a target for NF-ATc, although both genes may have converging roles in heart and lymphocyte development.
Figure 5: FK506 treatment blocks NF-ATc nuclear import in the endocardium.
a, NF-ATc (red) and DAPI (green) stainings in endocardial cells from a wild-type control embryo cultured for 1 h with ethanol carrier. Overlap of NF-ATc and DAPI stainings (orange) demonstrates nuclear NF-ATc localization (large arrowhead), coexisting with cytoplasmic (red) distribution (small arrowhead). b, FK506 treatment prevents nuclear localization of NF-ATc (red), as revealed by red cytoplasmic signal in the periphery of the cells (arrowhead). Scale bar, 6 m.
An intriguing feature of Ca2+ signalling is that although Ca2+ regulation is ubiquitous, both Ca2+ and calcineurin have been implicated in some of the most specific biological responses including axonal guidance12, memory13 and lymphocyte activation by antigens3,14,15. Our results suggest that at least one way by which ubiquitious calcium signals could direct specific biological processes is through the tissue-specific expression of target molecules such as NF-ATc, involved in cardiac valve formation in the embryo and activation of immune response. Furthermore, as NF-AT family members are expressed in different tissues16, specificity of response to a Ca2+ stimulus could be refined by both different tissue distribution and DNA sequence recognition17.
Ventricular septal and valve defects are the most frequent ofhuman congenital cardiac defects, present in nearly 1% of allbirths18. Future analysis will shed light on the potential involvement of NF-ATc and its regulatory pathways1,19 in human cardiac abnormalities.
Methods
Generation of NF-ATc3 mutant mice. A targeting vector was designed to replace a 1.5 kb genomic fragment containing exon 3 with the pMC1neo resistance expression cassette in reverse orientation to NF-ATc transcription. The targeting vector was linearized and electroporated into E14K ES cells. After G418 selection, homologous recombinants were identified by polymerase chain reaction (PCR) and confirmed by Southern blot. PCR, Southern, northern and western blots were performed according to standard procedures. The anti-NF-ATc monoclonal antibody 7A6 (ref. 20) recognizes the epitope deleted by the NF-ATc3 mutation (amino acids 176–380). The mutant phenotype was analysed in C57BL/6 and CD1; no differences were observed.
Histological analysis. Embryos were fixed in 4% paraformaldehyde, dehydrated, embedded in wax, sectioned and processed for H+ E staining according to standard protocols.
Whole-mount in situ hybridization. Embryos were fixed and processed following published protocols21. The NF-ATc probes used were a 1.6 kb 5' NF-ATc cDNA fragment containing exon 3, and a 900 bp 3' NF-ATc cDNA fragment, downstream to the region deleted by the NF-ATc3 mutation.
Radioactive in situ hybridization. Embryos were isolated and processed as for histological analysis. Hybridization was performed following published protocols22. The NF-ATc probes used were the ones cited above; the 
-cardiac actin probe was a 130 bp 5' fragment23; the Sox-4 probe was a 300 bp 5' fragment11.
Immunohistochemistry. Wild-type and mutant embryos were isolated and processed for histology as above. Staining was performed using a rat antibody against PECAM-1 (Pharmingen, 1:200 dilution), a biotinylated secondary antibody against rat IgG, and avidin-conjugated peroxidase (Vector), according to published protocols24.
Immunofluorescence. Embryos were dissected out of the yolk sac and immediately embedded in OCT (Tissue-Tek), frozen and sectioned. The anti-NF-ATc antibody 7A6 was used at 1:500. The signal was enhanced by an amplification step with anti-mouse IgG-biotin (CALTAG) 1:500, and avidin-Rhodamine (Vector) 1:2,000 (Figs 3a–c and 5), or avidin D-FITC (Vector) 1:1,000 (Fig. 3e, g, h, l). The anti-PECAM-1 rat antibody (Pharmingen) was used at 1:200 followed by incubation with an anti-rat IgG-FITC (CALTAG) at 1:400. Nuclei were counterstained with DAPI (Molecular Probes, 2.5 g ml-1) in PBS.
Short-term embryo culture. Embryos were collected at day E8.5 from plugged CD1 females, dissected out of Reichert's membrane, and cultured at 37 °C in DMEM containing 25% fetal bovine serum and 25% rat serum, using a BTC rotating bottle culture unit with continual gassing with 5% CO2 in balanced air. Embryos were cultured for 1 h in the presence or absence of 50 nM FK506, dissected away from the yolk sacs, and embedded in OCT for cryosectioning and staining with 7A6.
