Targeted disruption of semaphorin 3C leads to persistent truncus arteriosus and aortic arch interruption

Classic studies performed almost 20 years ago demonstrated that perturbations of migrating neural crest cells induce cardiovascular defects(Kirby et al., 1983EF21; reviewed by Creazzo et al., 1998EF10). Using chick embryos as a model system, Kirby and coworkers ablated regions of the dorsal neural tube before the emergence of migrating neural crest cells and observed subsequent defects in the septation of the cardiac outflow tract(persistent truncus arteriosus or PTA), double outlet right ventricle and interruption of the aortic arch. Using quail-chick chimeras and other labeling techniques to characterize neural crest cell fate during normal development,crest cells were seen to populate the branchial arches and also to help form the septum that divides the truncus arteriosus into the aorta and pulmonary artery (Waldo et al.,1998EF42).

More recently a number of mutations have been engineered in mice that produce similar cardiovascular defects (see Discussion). By analogy to the avian model, these genes are likely to affect cardiac neural crest development or function. In humans, outflow tract and other crest related anomalies are present in patients with DiGeorge syndrome and are often associated with deletions on chromosome 22q11 (Driscoll,1994EF11). Significant progress has been made in producing mouse models of DiGeorge syndrome by deleting homologous regions of mouse chromosome 16 (Lindsay,1999EF27). More recent and ongoing studies are identifying candidate genes within this region and characterizing their functional roles in neural crest development (Epstein and Buck,2000EF12; Merscher et al.,2001EF31; Lindsay et al.,2001EF28; Jerome and Papaioannou,2001EF19; Guris et al.,2001EF15).

The identification of essential genes is the first step in explaining how cardiac neural crest cells contribute to cardiovascular development. These genes will in turn define complex genetic and developmental pathways that control crest cell specification, migration, differentiation and survival. They will also identify signaling pathways by which crest cells interact with neighboring tissues within the pharyngeal arches and the outflow tract of the heart.

We present evidence that morphogenesis of the outflow tract and great vessels is dependent upon semaphorin 3C (Sema3C), a signaling molecule previously hypothesized to guide extending axons in the developing nervous system. The semaphorins comprise a large family of phylogenetically conserved secreted and transmembrane signaling proteins, some of which function as axon guidance cues (Raper, 2000; Yu and Kolodkin, 1999). Class 3 semaphorins are secreted glycoproteins that contain an approximately 500 amino acid N-terminal semaphorin domain, a C2 type immunoglobulin domain, and a highly basic C-terminal tail. Four of them, including Sema3C, are located on mouse chromosome 5 in a region syntenic to human chromosome 7q21-q31 (Tarantino et al.,2000). The first identified vertebrate semaphorin, semaphorin 3A, is a secreted protein purified on the basis of its ability to repel dorsal root ganglion axons in culture (Luo et al., 1993). semaphorin 3A knockout mice display a complex phenotype in which many different tissues are affected (Behar et al.,1996; Taniguchi et al.,1997). Notably, the peripheral nervous system of semaphorin 3A mutants is severely defasciculated although most axons ultimately appear to connect with their appropriate targets (Catalano et al.,1998). In knockout animals,axons fail to avoid territories that would normally express semaphorin 3A(Taniguchi et al., 1997). These mutant animals also experience postnatal hypertrophy of the right ventricle (Behar et al., 1996). The cause of this hypertrophy has not yet been identified.

The functions of a related semaphorin, Sema3C, have been characterized in vitro. Sema3C can act as either a repellent or an attractant for axons growing in culture. For example, the growth cones of sympathetic neurons are repelled by Sema3C (Koppel et al.,1997EF24), whereas growth cones of rat cortical axons are attracted towards a source of Sema3C (Bagnard et al.,1998EF1).

We have generated mice deficient in Sema3C. Mutants are cyanotic and die just after birth. Their death is attributable to aortic arch malformations and septation defects in the outflow tract of the heart. Our results suggest that Sema3C expression in the proximal cardiac outflow tract facilitates the entry of migrating neural crest cells that are essential for normal septation.

A bacterial artificial chromosome containing the Sema3C locus was purchased from Genome Systems. This BAC was digested with EcoRI,BamHI or XhoI and subcloned into pBluescript. The genomic fragment containing a critical exon was isolated and mapped with restriction enzymes. The majority of this exon was replaced by a PacI site, which was used to insert a cassette consisting of an internal ribosome entry site(IRES) followed by a tau-lacZ gene and a neomycin resistance cassette flanked by loxP sites.

The targeting vector was linearized with XhoI. Electroporation and cell culture of E14 cells (Hooper et al.,1987) were carried out as previously described (Mombaerts et al.,1996). Genomic DNA from G418-resistant ES cell colonies was digested with EcoRI and analyzed by Southern blot hybridization with external probes 5′ and 3′ to the targeting vector. Mice were backcrossed 4 generations into the CD1 background.

The neo-selectable marker was removed from the targeted mutation by crossing Sema3C heterozygotes (neoⁱⁿ) to CD1 mice from a cre pedigree (bcre-23) that consisted of 6 kb of the 5′flanking region of the Brn4/Pou3f4 promoter driving the expression of the Cre recombinase gene (gift from K. Ahn and E. B. Crenshaw III).

Three to five mm of mouse tail was placed into 100 μl of buffer containing 0.5 mg/ml proteinase K, 50 mM KCl, 10 mM Tris/HCl pH 8.3, 2 mM MgCl₂, 0.1 mg/ml gelatin, 0.45% NP40, 0.45% Tween 20. After 2-10 hours digestion at 55°C, samples were heat inactivated for 10 minutes at 95°C. 1 μl of this sample was added to a standard 15 μl PCR reaction with the following primer sets to amplify the wild-type and mutant alleles respectively in Sema3C(neoⁱⁿ) mice;5′-ttcccagtggggtagaacctagagc-3′,5′-gctatcaggacatagcgttggctac-3′, and 5′-gagtctgtttctacagaaatgcatgggtt-3′,5′-ttcccagtggggtagaacctagagc-3′. InSema3C(neo^out) mice the following primer sets were used to amplify the wild-type and mutant alleles respectively 5′-ttcccagtggggtagaacctagagc-3′,5′-gcgattaccgttgatgttgagtggc-3′, and 5′-gagtctgtttctacagaaatgcatgggtt-3′,5′-ttcccagtggggtagaacctagagc-3′. The conditions for the thermocycler were: 94°C for 1 minute, 29 seconds followed by 94°C for 31 seconds/60.9°C for 1 minute 10 seconds/72°C for 1 minute 30 seconds(35 times), 72°C for 5 minutes.

RNA was isolated from E12.5 wild-type and Sema3C mutant embryos by CsCl₂ gradient centrifugation (Chirgwin et al.,1979). The Superscript Preamplification System Kit (GibcoBRL) was used to generate first strand cDNA using 2 μg of total RNA from each mouse sample. PCR was performed using one of 3 primer sets. Primer set #1 5′-aaatggctggcaaaggatcct-3′,5′-cggtccacagcaatctttgt-3′, primer set #3 5′-cggggcaccgtgcaaaggtc-3′,5′-cagtggggtagaacctagagc-3′, primer set #2 5′-ctgactgcgagctgatgattt-3′,5′-attggggctagtatagaca-3′. 30 cycles of PCR (same conditions as above) were performed and samples were run on a 1.5% agarose gel.

To examine the structure of the vascular system, corrosion casting(Polysciences # 07349) was performed. Briefly, P0 wild-type andSema3C mutant mice were anesthetized with ketamine and perfused with phosphate-buffered saline through the left ventricle after a small hole was introduced into the right atrium. A 1 ml tuberculin syringe was used to inject polymer into the left ventricle under gentle pressure. After perfusion with the polymer, the resin was left overnight to harden. After the mass had fully cured, the tissue surrounding the cast was corroded away by using Maceration Solution (Polysciences) at 50°C.

Digoxigenin-labeled antisense and sense riboprobes were synthesized by in vitro run-off transcription of linearized plasmids, using SP6, T7 or T3 RNA polymerase. Sense control probes gave no signal. Whole-mount in situ hybridization was performed as described by Borycki et al. (Borycki et al.,1999EF4). Radioactive in situ hybridization was performed as described previously (Wawersik and Epstein,2000EF43).

Embryos and newborn mice were fixed in 4% paraformaldehyde in PBS for 24-48 hours. The skin was removed from newborn mice before fixation. After dehydration and embedding in paraffin wax, 8 μm sections were placed on glass slides. After dewaxing in xylenes and rehydration, sections were either stained with Hematoxylin and Eosin, or processed for in situ hybridization.

To elucidate the role of Sema3C in mouse development, the Sema3Clocus was disrupted by homologous recombination in ES cells. Work in our laboratory has demonstrated that class 3 semaphorins have small and widely scattered exons (unpublished observations of L. F. and M. J. R.). It was important, therefore, to identify and replace an exon expected to be crucial for Sema3C function. Structure-function studies have demonstrated the importance of the semaphorin domain for the functions of class 3 semaphorins(Koppel et al., 1997). A highly conserved, cysteine rich region of the semaphorin domain including the residues CCLARDPYCAWD is present in every class 3 semaphorin (Behar et al.,1999). When these amino acids are deleted from a semaphorin 3A expression plasmid, no functional protein is secreted from transfected 293T cells. Cell extracts from transfected cells are devoid of collapsing activity. Thus, deletion of these residues destroys the growth cone collapsing activity of semaphorin 3A (data not shown). Furthermore, class 3 semaphorins must dimerize to have collapsing activity(Koppel and Raper, 1998;Klostermann et al., 1998). Removal of the CCLARDPYCAWD-encoding exon alters the reading frame and thereby causes premature termination of the protein before the cysteine residue that is known to promote dimerization of Sema3A. Deletion of the CCLARDPYCAWD-containing exon and the concomitant premature termination of Sema3C is therefore likely to produce a null allele.

A diagram of the targeting vector we used to mutate Sema3C is shown in Fig. 1A. The majority of the CCLARDPYCAWD-containing exon is replaced by a cassette that includes an IRES followed by a Tau-lacZ fusion protein and a neomycin resistance cassette flanked by LoxP sites (Mombaerts et al.,1996). This targeting vector was electroporated into E14 ES cells (Hooper et al.,1987). Ten of 96 clones that developed resistance to G418 had undergone homologous recombination. Three of these ES cell lines were expanded and injected into blastocysts to derive chimeric mice. Chimeras from all 3 lines were capable of germline transmission of the mutation. The targeting of ES cells and germline transmission were confirmed by Southern blot hybridization(Fig. 1B). The neomycin resistance cassette was removed by crossing mice carrying the mutantSema3C locus to mice that express Cre recombinase in the germ cell lineage (gift from K. Ahn and E. B. Crenshaw III). Proper recombination was verified by PCR with a primer set flanking the excised cassette (data not shown).

Owing to the absence of specific antibodies for the mouse semaphorins, we analyzed Sema3C mRNA in Sema3C mutant mice and their wild-type littermates to determine the effectiveness of the Sema3Cmutation. Three primer sets were used in RT-PCR reactions to detect different regions of the Sema3C mRNA (Fig. 1D). Primer set 1 (blue) amplifies a region of the mRNA 5′of the targeted region, primer set 2 (red) a region 3′ of the targeted region, and primer set 3 (green) a region internal to the targeted region. All three primer sets amplify appropriately sized PCR products in wild-type animals (Fig. 1E). InSema3C mutant mice, only the primer sets to either side of the deleted exon are capable of detecting Sema3C mRNA. Thus, whileSema3C mRNA is produced in Sema3C mutant animals, this message does not contain the exon modified by the targeting construct(Fig. 1E, green lanes).

Sema3C mRNA is barely detectable by in situ hybridization in E12.5 mutant embryos (data not shown). Thus, the mutant form of Sema3C mRNA lacks a critical functional domain and is only expressed at a fraction of the wild-type level. The targeted Sema3C allele contains an IRES followed by a sequence encoding a Tau-lacZ fusion protein. The purpose of this addition is to harness the endogenous Sema3C promoter to drive the expression of tau-lacZ in cells that normally expressSema3C. Perhaps because of the low levels of Sema3C mRNA in mutant mice, no β-galactosidase is detectable by X-gal histochemistry or immunocytochemistry in tissues that normally express Sema3C mRNA(data not shown).

Mice heterozygous for the Sema3C mutation are grossly indistinguishable from their wild-type littermates. Some progeny of heterozygous mice die within the first 24 hours after birth. These sick neonates are cyanotic and when genotyped are found to be homozygous for theSema3C mutation. Interestingly, the penetrance of this phenotype is highly dependent on the strain of mouse. Postnatal mortality ofSema3C mutant mice is lower than 50% on both the 129 and C57BL/6 backgrounds. As the mutation was bred onto the CD1 background, however, the penetrance increased with each successive backcross. Crossing heterozygotes from the fourth backcross generation yields progeny in which the mortality rate for Sema3C homozygote mutants is 96% (22 of the 23 mutants in the first 8 litters examined). The powerful effect of strain upon the penetrance of this phenotype may be useful for identifying modifier genes in the future.

Analysis of 8 CD1 litters at birth demonstrates that wild-type,heterozygous and homozygous animals are present in the appropriate Mendelian ratios expected from a heterozygous cross(Table 1). There are occasionalSema3C mutant mice that do not die shortly after birth (e.g. 1 of 23 mutants from these 8 CD1 litters). These animals are viable, grossly indistinguishable from wild-type or heterozygous littermates, and fertile. The peripheral nervous system of mutant animals was examined for abnormalities. The cranial nerves of E10.5 and E11.5 Sema3C mutant mice were visualized with antibodies that stain neurofilaments and the sympathetic nervous system was examined with antibodies to tyrosine hydroxylase. No misprojections were detected, suggesting that Sema3C by itself is not required for grossly normal peripheral axon pathfinding. A detailed examination of the CNS of Sema3C mutant mice is now in progress.

Pathological evaluation of homozygous mice that die shortly after birth reveals cardiac outflow tract and aortic arch abnormalities. The aortic arch is interrupted in 100% of mice that die postnatally. The interruptions occur either between the left common carotid and left subclavian arteries (type B interruption; Fig. 2B) or between the brachiocephalic and left common carotid arteries (type C interruption; Fig. 2C). Type B interruption of the aortic arch could result from inappropriate regression of the left fourth branchial arch artery. Duplication of the left common carotid is infrequently observed (Fig. 2D).

Sema3C mutant mice frequently exhibit defects in septation of the conotruncus known as persistent truncus arteriosus (PTA). Injection of methyl methacrylate resin into the left ventricle is a sensitive method for detecting communication between the aorta and the pulmonary artery(Fig. 3A,C). Seventy-five percent of P0 Sema3C mutant mice evaluated by this technique have incomplete septation of the conotruncus(Table 1). In animals with PTA,the common valve of the outflow tract has four cusps(Fig. 3C inset). A ventriculoseptal defect is present in these animals just below the common valve. Histological analysis of the heart of a Sema3C mutant mouse with PTA (Fig. 3D) identifies a defect in the membranous portion of the ventricular septum with an intact muscular septum (Fig. 3E). The histology of the ventricular wall (Fig. 3F) and the structure of the mitral and tricuspid valves appear normal in Sema3C mutants (data not shown).

In utero, blood is oxygenated in the placenta rather than the lungs and is diverted from the pulmonary circulation into the descending aorta through a shunt called the ductus arteriosus. The pulmonary and systemic circulations are separated when this shunt closes just after birth. The systemic circulation is maintained in a Sema3C mutant embryo before birth because blood can bypass the interruption of the aortic arch by flowing from the heart into the descending aorta through the ductus arteriosis. After birth, however, blood cannot enter the descending aorta once the ductus arteriosus closes. Sema3C mutant mice therefore die from cardiovascular defects that are incompatible with postnatal life.

To better understand the etiology of this phenotype, we studied the expression of Sema3C in the cardiac outflow tract during development.Sema3C mRNA is expressed in the conotruncus as early as E10.5 and persists through E12.5 (Fig. 4A-C). Thus, expression of Sema3C is contemporaneous with the migration of cardiac neural crest cells into the outflow tract (Waldo et al., 1998). More detailed analysis of the expression pattern in sections at E10.5 demonstrates thatSema3C is expressed in the mesenchyme surrounding the branchial arch arteries (Fig. 4D). At this stage, Sema3C is restricted to the left side of the conotruncus and appears to be expressed in the myocardial cuff.

By E12.5, expression of Sema3C is restricted to the proximal outflow tract and the great vessels (Fig. 4E). Strong expression is also seen in the mesenchyme of the aorticopulmonary septation complex where neural crest cells form the developing septum (Fig. 4F).Sema3C is expressed in cells adjacent to neural crest migration pathways, suggesting a non cell-autonomous function. It is also expressed within the aorticopulmonary septation complex, possibly even within the neural crest itself. It is not uncommon to observe an axon guidance molecule to be expressed in neurons that are responsive to that cue. One dramatic example is the modulation of ephrin responsiveness by ephrin expression in retinal ganglion cells (Hornberger et al., 1999). It is possible that cardiac neural crest cells may both express and respond to Sema3C.

Sema3C has been shown to influence growth cone motility and guidance in vitro. Cardiac neural crest cells are known to be required for proper formation and septation of the developing cardiac outflow tract. We hypothesized that Sema3C acts as a guidance cue for migrating cardiac crest cells, that its loss prevents crest cells from populating the aortic arch or the proximal outflow tract, and that the resulting absence of crest cells in these structures is sufficient to induce the cardiac abnormalities inSema3C mutant mice.

We compared the migration of cardiac crest cells into the proximal cardiac outflow tract in wild-type and mutant embryos. Cardiac crest cells were visualized with a series of molecular markers including plexinA2. This transmembrane molecule is a member of a family of receptors that is thought to help mediate semaphorin signaling (Takahashi et al.,1999; Tamagnone et al.,1999). In an accompanying paper, transgenic and fate-mapping techniques demonstrate that plexinA2 is expressed by neural crest cells that surround the aortic arches and migrate into the cardiac outflow tract (Brown et al.,2001). The complementary expression patterns of Sema3C and plexinA2 raise the possibility that plexinA2 is a receptor component for Sema3C. PlexinA2 expression(Fig. 5E) is colocalized with other cardiac crest cell markers, including Foxc1(Fig. 5A) and endothelin receptor A (EdnrA; Fig. 5C), in the developing heart of wild-type embryos. Crest cells in wild-type embryos are seen to encase the branchial arches by E10.5 (not shown)and they populate the endocardial cushions of the outflow tract by E12.5Fig. 5A,C,E). Patterning of neural crest cells in the outflow tract of the heart is altered inSema3C mutant littermates. Altered expression of Foxc1(Fig. 5B), endothelin receptor A (Fig. 5D), and plexinA2(Fig. 5F) suggests that neural crest cell invasion of the outflow tract is significantly impaired in mutant embryos.

The most dramatic and fully penetrant cardiac phenotype observed in P0Sema3C mutant pups is interruption of the aortic arch. This anomaly could arise either from a failure in formation of the fourth left branchial arch artery, or from inappropriate regression of this artery segment during the complex process that reconfigures the symmetric embryonic vasculature into the asymmetric adult pattern. This remodeling process is known to require the cardiac neural crest, although the mechanisms by which specific right or left sided segments of branchial arch arteries are maintained or lost remain obscure. Branchial arch remodeling takes place in the mouse between embryonic days 10.5 and 13.5. Examination of sections of E10.5 and E11.5 Sema3Cmutant embryos demonstrates that the branchial arch arteries are present and normal morphologically (Fig. 6A-H). The anomalies seen later must therefore arise from a failure of the arch arteries to remodel correctly.

Consistent with our observation of impaired neural crest migration into the cardiac outflow tract at E12.5, we first detected structural cardiovascular abnormalities at E13.5. Histologic analysis of an E13.5 Sema3C mutant embryo revealed a ventricular septal defect (VSD;Fig. 6J) not seen in wild-type litter mates (Fig. 6I). In the mutant embryo, the aorta was seen to arise from the right ventricle(Fig. 6K) resulting in a double outlet right venticle. This malformation is seen after neural crest ablation in chick embryos (Kirby et al.,1983).

The deletion of a crucial exon and early truncation of the signaling molecule Sema3C leads to mice with congenital cardiovascular defects. These consist of interruption of the aortic arch and improper septation of the cardiac outflow tract. During cardiac development, the single great vessel emerging from the heart is the truncus arteriosus, from which the symmetric branchial arch arteries originate. A series of complex morphogenic events results in the division of the truncus arteriosus into the aorta and the pulmonary artery, and the realignment of the great vessels so that the aorta emerges from the left ventricle (Olson and Srivastava,1996EF34). Failure of proper septation of the conotruncus leads to a pathological condition known as persistent truncus arteriosus. Reconfiguration of the heart and outflow tract arise from remodeling of the branchial arch arteries. The branchial arch arteries begin as symmetric sets of vessels that feed the paired branchial arches. Each branchial arch artery contributes to specific portions of the adult aortic arch or its tributaries (Creazzo et al.,1998EF10). The left fourth arch artery, for example, develops into the portion of the aortic arch between the left common carotid and left subclavian arteries. Thus, the mature configuration of the aortic arch and its tributaries results from the regulated growth and regression of the branchial arch arteries.

Previous work from other laboratories has demonstrated that both remodeling of the branchial arch arteries and septation of the outflow tract require a subset of neural crest cells known as the cardiac crest. Transplantation of quail neural crest cells into developing chicks has shown that cardiac neural crest cells contribute to the tunica media of the aortic arch and to the septum of the outflow tract (Waldo et al.,1998EF42). When the cardiac neural crest is ablated in the chick, the resulting malformations are variable but commonly include persistent truncus arteriosus and absence of various combinations of branchial arch arteries, including those derived from pharyngeal arches 3 and 4 (interrupted aortic arch) and pharyngeal arch 6(absent ductus arteriosus; Kirby et al.,1983EF21; Creazzo et al.,1998EF10).

The similarity between the cardiovascular phenotypes in Sema3Cmutant mice and crest ablated chicks suggests that the mouse phenotype reflects a requirement for Sema3C in some important aspect of cardiac neural crest cell development. Two additional observations are consistent with this hypothesis. First, branchial arch arteries form normally in Sema3Cmutants. Subsequent interruption of the aortic arch must therefore be attributable to a failure of the branchial arches to reorganize correctly, a process known to require the cardiac neural crest. Second, incomplete septation of the outflow tract is evident very early in Sema3C mutant embryos and can be attributed to an impairment in crest cell migration into the proximal portion of the tract.

Several engineered and spontaneous mutant mouse lines have been described with interruption of the aortic arch or improper septation of the conotruncus. The cardiac defects in these lines have been ascribed to abnormalities in cardiac neural crest development. In most of these mutant mice, either septation of the conotruncus or remodeling of the branchial arch arteries is separately disturbed. Pax3, retinoic acid receptors, and connexin 43 are all expressed in cardiac neural crest cells (Franz,1989EF13; Mendelsohn et al.,1994EF30; Reaume et al.,1995EF36). Disruption of these genes leads to defects in septation of the conotruncus. It has been hypothesized that the cardiac defects arise in these knockout mice because the specification of, or communication between, cardiac neural crest cells is perturbed. Interruption of the aortic arch is not observed in mice with these mutations.

Mutations in the endothelin pathway result in a different set of cardiovascular anomalies. Disruption of this intercellular signaling pathway,which mediates communication between the arterial endothelium and the surrounding mesenchyme, leads to interruption of the aortic arch and abnormal persistence of some vessels that normally regress. This has been demonstrated by targeted disruption of endothelin-1, endothelin converting enzyme-1, or the endothelin receptor A (Clouthier et al.,1998; Kurihara et al.,1995; Yanagisawa et al.,1998). Mice null forHoxA3 or the winged helix transcription factors, Foxc1(Mf-1) or Foxc2 (Mfh-1), also die as the result of aortic arch anomalies (Winnier et al.,1999; Chisaka and Cappechi,1991). Although PTA is observed in 6% of EdnrA mutant mice and 10% of endothelin converting enzyme-1 mutant mice, septation of the conotruncus usually occurs normally in these knockout lines.

In contrast to these other mouse mutants, disruption of the Sema3Clocus causes both aortic arch interruption and persistent truncus arteriosus in the majority of mutant mice (Table 1B). One interpretation of this result is that Sema3C might function upstream of these other genes in a pathway regulating cardiac neural crest development. For example, loss of Sema3C function could affect crest cell specification, block cardiac crest cell migration, or reduce the expression levels of other genes known to be important in crest cell directed remodeling of the cardiac outflow tract. We have not exhaustively assayed the expression levels of genes that are expressed in cardiac crest or those known to affect remodeling of the outflow tract, but our data do show that three of them, Foxc1, the endothelin A receptor, and plexinA2 (see Brown et al.,2001) are all expressed inSema3C mutant mice in the regions of the branchial arches normally populated by neural crest cells. These observations argue that cardiac crest cells are properly specified and begin to migrate normally in Sema3Cmutant mice.

Neural crest cell migration into the proximal cardiac outflow tract,however, appears to be impaired. Since Sema3C is normally expressed in the myocardial cuff of the outflow tract, a plausible and attractive interpretation of these results is that Sema3C acts as an attractant for cardiac crest cells and helps to promote their entry into the outflow tract. This proposed function would be analogous to its chemoattractant activity for cultured cortical axons (Bagnard et al.,1998). Alternatively, Sema3C could promote entry indirectly, perhaps by inducing the expression of a permissive or attractive signal in nearby tissues. While our results support the hypothesis that Sema3C promotes the migration of cardiac crest cells into the proximal outflow tract, we cannot eliminate the possibilities that Sema3C affects cardiac crest cell proliferation, differentiation or survival within the proximal tract.

Cardiac defects including persistent truncus arteriosus and interruption of the aortic arch have been described in neuropilin-1 knockout mice(Kawasaki et al., 1999). These mice die between E12.5 and E13.5 and have multiple defects in several organ systems. Neuropilin-1 is capable of acting as a coreceptor for the angiogenic signals vascular endothelial growth factor (VEGF) and placenta growth factor 2(PIGF) (Migdal et al., 1998;Soker et al., 1998). Cardiovascular defects present in neuropilin-1^-/- embryos have been interpreted to result from impaired angiogenic signaling. However,neuropilin-1 can also serve as an obligatory coreceptor for a subset of class 3 semaphorins including semaphorin 3A and semaphorin 3C (He and Tessier-Lavigne, 1997;Kolodkin et al., 1997). Defects in axon pathfinding in neuropilin-1^-/- embryos can be ascribed to a loss of sensitivity to these guidance cues. The striking similarity between cardiac defects observed in the neuropilin-1knockout and in the Sema3C mutant mice is consistent with the hypothesis that neuropilin-1 functions as a receptor component for Sema3C in cardiac neural crest. Our results strongly argue that the cardiac defects observed in neuropilin-1-deficient embryos are due to the inability of cardiac neural crest cells to respond to Sema3C.

The cardiovascular defects resulting from the disruption of theSema3C locus closely resemble heart defects seen in human infants with congenital heart disease. Some of these arise as a consequence of microdeletions within chromosome 22q11 (Driscoll,1994), however, a majority of infants with outflow tract and aortic arch defects do not have this particular chromosomal abnormality (Goldmuntz et al.,1998). Semaphorins, their receptors, and the signaling pathways they influence now constitute a new class of candidate genes for human congenital cardiac disease.

We thank Dr Trish Labosky for instruction and help in growing ES cells, Dr Jean Richa of the Penn Transgenic Core for ES cell injections, Dr Bryan Crenshaw for his generous gift of bcre-23 mice, and Vickie Cormie, Cynthia Ito-Nam, Lisa Chang and Christian Metallo for their assistance in maintaining the mouse colony. Thanks to all the members of the Raper lab for their valuable input. This work was supported by NIH RO1 NS38548 to J. A. R., NIH HL62974, HL61475 to J. A. E., NIH F32AR08584 to C. B. B. and NIH T32HD07516 to L. F. A. L. W. is supported by the Cancer Research Fund of the Damon Runyon-Walter Winchell Foundation Fellowship, DRG-1530.