Essential roles of the winged helix transcription factor MFH-1 in aortic arch patterning and skeletogenesis

The cardiovascular system undergoes a complicated series of morphogenetic events to form a heart and an aorta in fetuses (Olson and Srivastava, 1996). Formation of the heart and aorta requires migration, differentiation and precise interactions among multiple cells from several embryonic origins. The tubular heart undergoes rightward looping and then proceeds to the formation of atrial and ventricular chambers. The aortic arch is formed through an extensive remodeling of arch arteries and dorsal aortas. Congenital cardiovascular defects represent the most common group of human birth defects and of spontaneously aborted fetuses. Although anatomical and physiological descriptions of aortic arch anomalies in newborns have existed for half a century (Celoria and Patton, 1959; Moller and Edwards, 1965; van Mierop and Kutsche, 1984), the molecular bases underlying the anomalies remain poorly understood.

The vertebrate skeleton is composed of cartilage and bone, and is generated as a consequence of secondary induction between the mesenchymal primordia and inductive signals emanating from the adjacent tissues. The vertebrate skull is functionally and evolutionarily divided into the neurocranium, dermatocranium and visceral skeleton. These skeletal elements are derived from the cephalic neural crest, cephalic mesoderm and somitic mesoderm in the higher vertebrates (Noden, 1983; Couly et al., 1993). In axial skeleton development, after the de-epithelialization of the somite, sclerotome cells, which accumulate around the notochord, constitute the vertebral body and the intervertebral disc anlage in the ventral region of the vertebrae (reviewed by Christ and Wilting, 1992). Other sclerotome cells migrate dorsally and proceed along the cartilaginous condensation, resulting in the formation of the laminae and spinous processes of the neural arches.

Winged helix proteins are a family of transcription factors that share an evolutionarily conserved DNA-binding domain (Lai et al., 1993; Kaufmann and Knochel, 1996). The roles of winged helix genes in patterning and morphogenesis during development were initially appreciated by mutations in Drosophila (Weigel et al., 1989) and then by targeted disruption of the HNF-3β (Ang and Rossant, 1994; Weinstein, 1994), BF-1 (Xuan et al., 1995) and BF-2 (Hatini et al., 1996) genes, as well as the spontaneous mutation of the whn gene (Nehls et al., 1994) in the nude mouse. For example, embryos homozygous for a null mutation in HNF-3β die at the neurula stage and display defects in node and notochord formation (Ang and Rossant, 1994; Weistein et al., 1994). Mice deficient in BF-1 and BF-2 genes displayed defective formation in the cerebral hemispheres and kidneys, respectively (Xuan et al., 1995; Hatini et al., 1996).

We previously isolated the Mesenchyme Fork Head-1 (MFH1) gene (Miura et al., 1993), which is expressed in developing cartilaginous tissues, kidneys and dorsal aortas (Miura et al., 1993; Kaestner et al., 1996). Genomic and protein structures of MFH-1 are conserved between human and mouse, and the MFH-1 proteins act as a transactivator (Miura et al., 1997). In this study, we investigated the developmental roles of MFH-1 by generating mutant mice with a targeted disruption of the gene. We found that mice lacking MFH-1 had interrupted aortic arches, which is the same phenomenon observed in human anomalies. Furthermore, MFH-1-deficient mice manifested defects in the skeletogenesis of the neurocranium and vertebrae. These results suggest that MFH-1 has indispensable roles in aortic arch formation and skeletogenesis.

Genomic DNA containing the MFH-1 gene was isolated (Miura et al., 1997) from a library of mouse strain 129 DNA (Stratagene). The XhoI-PstI fragment containing a whole MFH-1 gene was replaced by the PGK-neo gene and the herpes simplex thymidine kinase gene was fused at the 3′ end (Fig. 1A). The targeting vector was introduced in E14.1 cells by electroporation and homologous recombinants were enriched by selection with G418 and gancyclovir and identified by Southern blot analysis. Five independently mutated embryonic stem (ES) clones were used to produce mutant mice by blastocyst injection as described (Tanaka et al., 1995). Chimeras were mated with C57BL/6 females and those derived from five ES clones transmitted the mutation to their offspring. Mice were bred to generate siblings on a C57BL/6 genetic background. In Southern blot analysis (Fig. 1B), DNA prepared from ES cells or embryos was digested with EcoRI, transferred to a nylon membrane and then hybridized with the 0.5 kbp probe A that flanked the 5′ homology region. In genotyping of embryos (Fig. 1C), PCR was performed according to standard protocols to discriminate wild and mutant alleles in the DNA from the yolk sac of embryos. Primer sequences were as follows: 1 (MFH-1 antisense), 5′-CCAGTTCTTAGTCCCCCCAC-3′; 2 (MFH-1 sense), 5′-CCAGTTGGTAACCTGGACTG-3′; 3 (PGK antisense), 5′-GGATGTGGAATGTGTGCGAG-3′.

Total RNA was reverse-transcribed with a first-strand cDNA synthesis kit (Life Science, FA) and amplified with oligonucleotides 5′GATCTCAGCGAGTCCCTCTA-3′ and 5′-CACCTTATCCGGAGAGACCT-3′ as primers for MFH-1 mRNA, oligonucleotides 5′-CCACAGTGCTTCGTTTCCAC-3′ and 5′-CCATGTCCTCTACTGGACCA-3′ as primers for fkh-6 mRNA and oligonucleotides 5′-GTCCTTGAGATGAGCATCGG-3′ and 5′-CGTAGGAGAGACACCTTTCC-3′ as primers for surfactant protein C mRNA, respectively. Without addition of reverse transcriptase, no PCR products were detected.

Embryos were fixed in 4% paraformaldehyde overnight at 4°C, dehydrated, embedded in Paraplast and sectioned at 5 μm. Sections were stained with hematoxylin and eosin for histological analysis. In situ hybridization on tissue sections (Kessel and Gruss, 1991) and in situ hybridization with whole embryos (Wilkinson, 1992) were performed using cRNA transcribed from the NotI-EcoRI fragment of the MFH1 cDNA (Miura et al., 1993) as a probe.

Skeletal preparations were made from newborn mice and cleared skeletons were analyzed under the stereomicroscope as described previously (Kessel and Gruss, 1991).

To examine MFH-1 function during development, the MFH-1 gene was disrupted by homologous recombination in ES cells. Mapping of the MFH-1 gene showed that the major transcript is encoded in a single exon (Miura et al., 1997). A targeting vector was designed in which the whole gene was replaced with a neomycin-resistance cassette (Fig. 1A). Five lines of targeted ES cells were established and chimeric mice derived from the five clones transmitted the mutation to their offspring. Mice derived from each clone had identical phenotypes. The targeted locus was confirmed by Southern blot analysis (Fig. 1B and data not shown) and identified by PCR (Fig. 1C). By RT-PCR analysis, we confirmed that MFH-1 mRNA was not detected in homozygotes but fkh-6 mRNA (the gene is located 8 kb downstream from the MFH-1 gene) was expressed almost equally in homozygotes and wild-type embryos (Fig. 1D). Heterozygous mice were indistinguishable from wild-type animals. Among neonates obtained by crossing heterozygotes, those who died just after birth turned out to be homozygotes. The number of homozygotes was almost half of that expected (Table 1). The homozygotes were indistinguishable from the wild-type macroscopically, but demonstrated respiratory compromise as evidenced by gasping motions and cyanosis, and died within 10 minutes of birth. Dissection of homozygous mice revealed that major anomalies in the aortic arch and the collapsed lung might be causal for the neonatal death of MFH-1 mutants.

Twelve newborn homozygotes were pathologically examined. No remarkable abnormalities were revealed in the brain, trachea, thymus, thyroid, parathyroid, heart, lung, liver, spleen, kidney and placenta, except collapse of the lung. Interestingly, aortic arch anomalies were found in all neonates (Fig. 2). The most frequent (5 out of 12) anomaly was type B interruption of the aortic arch where part of the aortic arch between the left common carotid artery and the left subclavian artery did not exist (Celoria and Patton, 1959) (Fig. 2B,H). Coarctation of the aortic arch (Fig. 2E) was found in four individuals and atresia of the aortic arch (Fig. 2D) occurred in one neonate. These anomalies are considered to be in the category of type B interruption of the aortic arch, except that the proximal and distal ends are connected with fibrous structures. Type C interruption of the aortic arch, where part of the aortic arch between the brachiocephalic artery and the left common carotid artery did not exist (Celoria and Patton, 1959) (Fig. 2C,I), was also found in one neonate. A similar atypical anomaly was found in one individual as tubular hypoplasia of the proximal part of the aorta (Fig. 2F). The right and left subclavian arteries were found to originate from the proper positions in all neonates but one, where the right subclavian artery originated from a more distal position (Fig. 2E). Aortopulmonary septation was formed normally in all individuals. In summary, 83% of the anomaly was located distal to the left common carotid artery (LCCA) and 17% was proximal to the LCCA.

Since human patients with type B interruption of the aortic arch also have an associated ventricular septal defect (Celoria and Patton, 1959; Moller and Edwards, 1965; van Mierop and Kutsche, 1984), we further examined the intracardiac defects of hearts from newborn homozygotes. Frequently, a tiny ventricular septal defect was observed (Fig. 3C-F). These results indicate that the anomalies in MFH-1^{− /−} mice are the same as those in humans.

The aortic arch is formed through a complex series of remodeling of vessels, as shown in Fig. 4. At the beginning, the aortic sac communicates symmetrically with dorsal aortas via the first and second arch arteries at 9.5 days postcoitum (dpc) (Fig. 4A). At 10.5 dpc, the first and second arch arteries disappear and the third, fourth and sixth arch arteries are formed, symmetrically linking the aortic sac and dorsal aortas (Fig. 4B). Then the left fourth and sixth arch arteries become enlarged compared to the corresponding right arteries, forming an asymmetry (Fig. 4C,D). By 15.5 dpc, the aortic sac, the left fourth arch artery and the left dorsal aorta form the aortic arch. The proximal part of the right subclavian artery originates from the right fourth arch artery. The third arch arteries become the common carotid arteries on each side. The left sixth arch artery is reformed into the pulmonary trunk and ductus arteriosus Botalli, while the distal part of the right sixth arch artery and right dorsal aorta regress (Fig. 4E).

Next, we examined embryos to investigate the pathogenetic process of the anomaly in the aortic arch in MFH-1^{− /−} mice. The remodeling of the dorsal aorta and arch arteries to generate the aortic arch occurs between 10.5 dpc and 14.5 dpc. Genotyping analysis of embryos between 10.5 dpc and 14.5 dpc showed that homozygotes were present following the Mendelian rule. Although all embryos younger than 11.5 dpc were alive, dead embryos whose hearts did not beat were found in about half of the embryos older than 13.5 dpc (Table 1). When we examined embryos at the critical stage around 12.5 dpc (Fig. 5), we found that at 11.5 dpc, the right and left third, fourth and sixth arch arteries communicated with dorsal aortas in both the wild-type and the mutant fetuses; no differences were observed (Fig. 5A,B). At 12.5 dpc, the left fourth arch artery develops to form the isthmus of the aortic arch together with the dorsal aorta. The left third arch artery gives rise to the left common carotid artery. In MFH-1^{− /−} embryos, the left fourth arch artery disappeared (Fig. 5D) or was diminished (Fig. 5E) in most cases. We confirmed by serial histological sections that the arch arteries were symmetrically formed at 10.5 dpc and then that the part corresponding to the left fourth arch artery was not detected at 12.5 dpc (data not shown). We also found that the proximal segment from the left third arch artery was diminished (Fig. 5F) in a small number of cases. The right third and fourth arch arteries were not affected (data not shown). The left sixth arch artery in mutant embryos was not different from the wild-type (Fig. 5 and data not shown). Our results indicate that, in mice lacking MFH-1, the left fourth arch artery might catastrophically regress during remodeling of the aortic arch around 12.5 dpc.

We examined in detail the expression of the MFH-1 gene in the developing arch arteries in order to clarify the relationship between the phenotype and the expression of the gene. Whole-mount in situ hybridization showed that the MFH-1 mRNA was expressed in the third, fourth and sixth arch arteries on the right and left sides, as well as in the dorsal aortas in 10.5 dpc embryos (Fig. 6B,D). Weak signals were also detected in the endocardium of the ventriculus. Furthermore, we performed in situ hybridization of sections of embryos from 9.5 dpc to 12.5 dpc. At 9.5 dpc, the signals were weakly detected in the arch arteries and dorsal aorta in the caudal region (Fig. 7E and data not shown). At 10.5 dpc, the signals were strongly detected in the symmetrical third, fourth and sixth arch arteries on both sides (Fig. 7F). At 11.5 dpc, signals were strongly detected in the asymmetrical arch arteries on both sides (Fig. 7G). Characteristically, the MFH-1 mRNA was expressed in broad areas surrounding arteries (Fig. 7F,G), suggesting that it is expressed in the surrounding mesenchymal cells and endothelium. At 12.5 dpc, the signals were less strongly detected in the arteries (Fig. 7H and data not shown). These results indicate that the area of aortic arch anomalies is more restricted than that of MFH-1 mRNA expression.

Defects of skeletogenesis were seen in the craniofacial bones and vertebral column, while appendicular skeletal systems and the sternum were intact.

All newborn homozygotes manifested a complete cleft of the secondary palate (Fig. 8A,B). Histological examination showed the loss of the soft palate as well as a lateral shift of the maxillary and palatal shelves and pterygoid bones (Fig. 8C,D and data not shown). In the skeletal preparation, several bony components contributing to the formation of the secondary palate and the soft palate, including the palatal processes of the maxillary and palatal bones, and the pterygoid bone were present but deformed and located more laterally in MFH-1-deficient mice (Fig. 8E,F). In the skull base, ossification of the presphenoid bone was missing or delayed (Fig. 8EH). The formation of the optic canal, which is a structure contiguous with the presphenoid was also affected (Fig. 8I,J). The basisphenoid bone was slightly shortened craniocaudally and a tiny defect was reproducibly seen in the mediocaudal region (Fig. 8G,H). The medial region of the alisphenoid bone was also affected, while the ascending lamina was unaffected. The mediocaudal portion of the alisphenoid bone was missing, resulting in a loss of the posterior wall of the foramen ovale. The pterygoquadrate cartilage was also absent (Fig. 8K,L). In the middle ear, the maleus and gonial bones were hypoplastic and the maleus and incus were fused in MFH-1 mutants, whereas the stapes was unaffected (Fig. 8M,N). Ossification of the otic vesicle was also affected in MFH-1 mutants. In newborn mutants, Meckel’s cartilage had a sigmoidal appearance compared with the straight rod of cartilage in wild-type mice (Fig. 8E,F). However, Meckel’s cartilage appeared normal in paraffin sections. The mandible itself and tooth buds were unaffected (data not shown). The hyoid bone and thyroid, cricoid and tracheal cartilages were not deformed in MFH-1deficient mice (data not shown). Therefore the cranial defects in MFH-1 mutant mice were restricted primarily to the neurocranium derived from the cranial neural crest within the pre-mandibular and first branchial arch and, to a lesser extent, to structures that arise from the cephalic mesoderm (Noden, 1983; Couly et al., 1993; Osumi-Yamashita et al., 1994). Because cranial neural crest cells give rise not only to skeletogenic components but also to the cranial nerves and musculature in the head and face, we analyzed cranial musculature and nerves, particularly the trigeminal nerve. Histological analysis showed no significant alterations in the cranial musculature or trigeminal nerve (Fig. 8C,D, and data not shown). Thus, only skeletogenic elements were affected in the cranial region of MFH-1-deficient mice.

Vertebral abnormalities were seen in medial structures. In the ventral region, the vertebral bodies of MFH-1 mutants appeared shorter than the wild type in the craniocaudal direction (Fig. 9A,B). In the cervical region, ossification centers could not be seen in M F H −1-deficient mice, except for the atlas. The transverse foramen, through which the vertebral artery passes, was not formed normally (Winner et al., 1997 and data not shown). Dorsally, the cartilaginous condensation of the lamina of the neural arches in the cervical region, most prominently in the at l a s, were split at the midline, aligned irregulrly and occasionally bifurcated (Fi g. 9E,F). In the thoracolumbar region, ossification centers were small, irregularly aligned and split (Fi g. 9C,D). Th eproximal part of several ribs was occasionally fused (indicate d by an arrow in Fi g. 9D). The spinous process and laminae of the neural arches were apparently lacking, which results in spina bifida occulta. In the cranial region, ossification of the supra occipital bone was totally impaired, this bone being replaced by a cartilaginous mass (Winner et al., 1997 and data not shown). Histological analyses on 14.5 dpc fetuses showed that the musculature and dermis developed normally in the trunk.

As described by Kaestner et al. (1996), the expression of MFH1 in 7.5 and 8.5 dpc embryos was observed in the non-notochord mesoderm of the trunk region and head mesoderm of the prechordal region (data not shown). Because the development of the axial and craniofacial skeletons begins and proceeds between 9.5 dpc and 12.5 dpc, we systematically investigated the MFH-1 expression during this period.

Almost ubiquitous but graded MFH-1 expression was seen in the mesenchymal cells in the head region of 9.5 dpc embryos (Figs 10A-D, 11A,B). Strong expression was seen in the mesenchymal condensation around the optic vesicle, and mesenchyme underlying the midbrain and hindbrain, whereas the expression in the nasal placode and distal region of the mandibular component of the first branchial arch was weak (Figs 10C,D, 11A,B). In 10.5 dpc embryos, mesenchymal cells surrounding the optic cup and underlying diencephalon showed strong MFH-1 expression, while mesenchymal cells comprising facial primordia were completely negative for MFH-1 expression (Figs 10E-H, 11C-E). At this stage, the palatal shelves of the maxilla and palatine were not yet prominent, while MFH-1 expression strongly demarcated the area giving rise to palatal processes (Fig. 11D). More caudally, MFH-1 expression was seen around Rathoke’s pouch but not in the maxillary component of the branchial arch or medial or lateral front nasal mass (Figs 10G,H, 11C,D). At the level of the otic vesicle, strong expression was seen around the vesticular portion of the otic vesicle and the proximal part of the mandibular component of the first branchial arch (Fig. 10I,J). Weak expression was observed in the perichordal region. In 11.5 dpc embryos, MFH-1 was expressed strongly in the maxillary prominence, the proximal part of the precursor for Meckel’s cartilage, the lateral process giving rise to the palatine, and the mesenchymal cells underlying the rhombencephalon and weakly in the tongue primordium (Fig. 10K,L). At the level of the optic cup, strong MFH-1 expression was seen in the mesenchymal cells underlying the diencephalon and surrounding the optic cup, the prospective sclerotic coat of the eyeball and the mesenchymal condensation giving rise to the nasal septum (Fig. 10M,N). In 12.5 dpc fetuses, MFH-1 expression was localized in the mesenchymal cells around Rathoke’s pouch, Meckel’s cartilage mass and the lateral palatine process (Fig. 10O,P). Thus, MFH-1 is initially expressed ubiquitously in the head mesoderm and subsequently is localized in the mesenchymal condensations giving rise to the floor and wall of the neurocranium, palatine and Meckel’s cartilage.

We have analyzed the functions of the MFH-1 gene by gene targeting in mice. MFH-1 mutant mice showed defects in the aortic arch and in the skeletal structure. This work clearly indicates that during the aortic arch formation, the MFH-1 gene product is essential for the extensive remodeling of the left fourth arch artery. It is also demonstrated that the MFH-1 protein is required for the normal development of the neurocranium, palatine and axial skeleton.

About half of the homozygous mice died around 12.5 dpc and the other half died just after birth (Table 1). We found that the heart beat stopped in homozygous embryos older than 13.5 dpc and also that large pools of blood were present at several sites, suggesting that embryonal death is likely to be directly or indirectly due to abnormal remodeling of the aortic arch. Surviving homozygotes died soon after birth. In addition to interruption of the aortic arch, the inability to inhale air into the lungs seems to be one cause of neonatal death. In fact, the lungs dissected from newborn homozygotes sank in water, while those from heterozygotes floated on water. However, histologically the lungs of the mutant mice were indistinguishable from those in the wild type before birth (data not shown). Furthermore, cartilage surrounding the trachea was normal and no obstruction was found in the airway in newborn mutant mice. The RT-PCR experiment indicated that the surfactant protein C mRNA was expressed at almost the same levels in the lungs from newborn mutants and from the wild-type animals (data not shown). Further studies are needed to determine the cause for collapse of the lung.

The role of MFH-1 in aortic arch patterning

Aortic arch development can be divided into two functional phases; first, the generation of the dorsal aorta and arch arteries and, second, the remodeling of these primary structures. Ablation of the premigratory cardiac neural crest cells in chick embryos results not only in agenesis or hypogenesis of the thymus, thyroid and parathyroid gland but also in cardiovascular malformations including persistent truncus arteriosus, double-outlet right ventricle, tetralogy of Fallot and interrupted aortic arch (Kirby and Waldo, 1990, 1995). These results imply that accumulation of neural-crest-derived cells in prospective arch artery regions is a critical initial process for the subsequent remodeling of the arch arteries. In the mutant mice, the left fourth arch artery was first formed and then disappeared (Fig. 5). Our results indicate that the MFH1 gene product is not essentially required for the accumulation of neural crest cells and the formation of primary structures, although MFH-1 expression can be seen in the developing arch arteries starting from a very early stage.

However, MFH-1 is required for the extensive remodeling of the dorsal aorta and arch arteries to generate the aortic arch, particularly in the derivation of the left fourth arch artery. The left fourth arch artery undergoes extensive morphological changes to participate in the aortic arch formation, while the other arch arteries do not. They merely contribute to arteries that branch from the aortic arch. Thus, the MFH-1 protein might play a critical role during this morphogenetic change of the left fourth arch artery and may be involved in the extensive enlargement of the artery.

The Type B interruption of the aortic arch is also found in the DiGeorge syndrome in humans (discussed later) and mice deficient in endothelin-1 gene. In endothelin-1-deficient mice (Kurihara et al., 1994, 1995), in addition to the interruption of the aortic arch, the persistence of the first and second arch arteries and the subsequent formation of extra arteries, a ventral septal defect, and an absence of the right subclavian artery are frequently observed. Thus, the phenotypes of MFH-1-deficient and endothelin-1-deficient mice are very similar, with only slight differences. RT-PCR analysis of whole embryo RNA revealed that endothelin-1 mRNA was detected almost equally in 11.5 dpc embryos of MFH-1^{− /−} mice and wild-type mice (data not shown). This result suggests that it is unlikely that the endothelin-1 gene is downstream of the MFH-1 gene in neural-crest-derived cells.

Abnormalities in the aortic arch are seen in Splotch mice (Pax-3 gene mutant) (Franz, 1989) and mice deficient in retinoic acid receptors (Mendelsohn et al., 1994) and neurofibromatosis type-1 gene (Brannan et al., 1994). However, these anomalies are restricted to the conotruncal region, that is, they result in persistent truncus arteriosus but not in interrupted aortic arch. MFH-1 and endothelin-1 genes may play roles in an overlapping subset of neural-crest-derived cells, while Pax-3, retinoic acid receptor and neurofibromatosis type-1 genes might be active in distinct subpopulations of neural-crest-derived cells.

In the cranial region, MFH-1 is first expressed ubiquitously in the cephalic neural-crest and, presumably, in cephalic mesoderm-derived cells. Subsequent strong expression is localized in mesenchymal cells underlying the developing brain, giving rise to the floor and wall of the neurocranium and surrounding the optic capsule. In MFH-1-deficient mice, hypoplasia of the basi- and alisphenoid bones, absence of several ossification centers in the neurocranium, deformities of the middle ear ossicles, and cleft palate were reproducibly observed. Thus, MFH-1 is involved in the skeletogenesis of the neurocranium but not the dermatocranium. Lack of the MFH1 gene product also results in hypoplasia of the vertebrae and insufficient chondrification or ossification of medial structures. Since skeletogenic condensations are initiated by increased mitotic activity or by the aggregation of cells toward centers that allow the accumulation and subsequent differentiation of mesenchymal cells to produce sufficient amounts of extracellular matrix, MFH-1 might be involved in the proliferation, aggregation or differentiation of the neural-crest and cephalic mesoderm-derived mesenchymal cells and sclerotome cells (Hall and Miyake, 1992; Winner et al., 1997).

Interestingly, very similar defects of the neurocranium, palatine and vertebral column are observed in Gli2-deficient mice (Mo et al., 1997). Gli2 gene product is supposed to be a mediator of Sonic hedgehog (SHH) signals. Since grafts of the notochord or SHH-producing cell pellets in the vicinity of the unsegmented paraxial mesoderm induced an additional expression domain of MFH-1, MFH-1 expression might also be dependent upon SHH in the paraxial mesoderm (H. K. and N. M., unpublished observation). Thus, during the cartilaginous condensation of the neurocranium and the axial skeleton, MFH-1 and Gli2 gene products might function under the SHH signal-transducing cascade.

Skeletal phenotypes of MFH-1-deficient mice are fully penetrant, although they are subtle. Presumably, the closely related forkhead gene, MF-1, whose expression shows extensive overlap with that of MFH-1, may partially compensate for the loss of MFH-1 (Sasaki and Hogan, 1993; Miura et al., 1993; Kaestner et al., 1996). Analysis of mice lacking both MFH-1 and MF-1 might be necessary to understand the essential functions of MFH-1 during vertebral column and skull development.

DiGeorge syndrome in humans is typically characterized by aplasia or hypoplasia of the thymus, hypoplasia of the parathyroid glands, outflow tract defects of the heart, cleft palate and mildly dysmorphic facial features (Wilson et al., 1993; Goldmuntz and Emanuel, 1997). Among these criteria, MFH-1-deficient mice exhibited type B interruption of the aortic arch with ventricular septal defect and cleft palate, and possibly craniofacial abnormalities, but not defects of the thymus and parathyroid. Thus, phenotypes of MFH-1-deficient mice overlapped with those in patients with DiGeorge syndrome. However, the human MFH-1 gene is located at chromosome 16 (Kaestner et al., 1996), not at chromosome 22, which is linked to many cases of DiGeorge syndrome (Wilson et al., 1993). Therefore, the MFH-1 gene is unlikely to be the defective gene in DiGeorge syndrome. Rather, an absence or a defect of MFH-1-positive cells in the prospective aortic arch region might be the cause for interruption of the aortic arch in the DiGeorge syndrome. It is also possible that MFH-1 gene might be one of the target genes of the DiGeorge gene in subsets of neural-crest-derived cells.

MFH-1-deficient mice are useful as a mouse model for interruption of the aortic arch. Molecular analysis of the mutant mice at the stage of remodeling of the aortic arch will open a new era in understanding the gene(s) involved in aortic artery formation.

We thank Dr N. Shirafuji and S. Aoki for technical assistance. This work was supported by Grants-in-Aid from the Ministry of Education, Science, Sports and Culture of Japan and from the Agency of Science and Technology.