Migration of cardiac neural crest cells in Splotch embryos

In birds and mammals the complex process of cardiac morphogenesis requires interactions between multiple cell types intrinsic and extrinsic to the primitive heart tube. A specific population of neural crest cells, defined by ablation studies in the chick, migrates to the rostral pole of the heart tube and mediates septation of the single great vessel emerging from the primitive heart (Kirby et al., 1983). This results in division of the truncus arteriosus into the aorta and the pulmonary artery and is associated with rotation of the outflow tract, resulting in juxtaposition of the aorta with the left ventricle. Studies in chick embryos indicate that disruption of neural crest migration can result in cardiac outflow tract defects. Thus, ablation of premigratory cardiac neural crest cells emerging between the mid-otic placode and the caudal boundary of somite 3 results in the failure of cardiac outflow tract septation (Kirby et al., 1983). Such studies indicate that errors in this process may underlie relatively common forms of congenital heart disease in humans which are often associated with other neural crest related abnormalities and syndromes (Driscoll, 1994; Jones, 1990).

Molecular pathways regulating cardiac neural crest function and interaction with cardiac tissues remain largely unknown. Normally, neural crest cells make up only a small proportion of the final aorticopulmonary septum and outflow endocardial cushions, as demonstrated by quail-chick chimera studies in which quail neural crest cells are transplanted into chick embryos (Kirby et al., 1983; Waldo et al., 1999). These observations are confirmed by immunohistochemical analysis using HNK-1, an antibody specific for chick neural crest cells, and by diI labeling studies to fate map neural crest cells in the chick (Kirby et al., 1993). In humans and mice, however, there is no equivalent marker for tracking cardiac neural crest cells and thus the contribution of neural crest cells to cardiac formation has been inferred by analogy to chick and by mutational analysis (Epstein, 1996).

The Splotch mutant mouse has served as an important model for congenital heart disease involving the outflow tract of the heart, as homozygous mutant Splotch embryos display persistent truncus arteriosus in addition to other developmental defects (Auerbach, 1954; Franz, 1989). Several alleles of Splotch have been identified and all result from mutation or deletion of the Pax3 gene (Epstein et al., 1991, 1993). Pax3 encodes a transcription factor expressed during embryogenesis in the dorsal neural tube from which neural crest cells emerge (Goulding et al., 1991). Phenotypic resemblance to chick neural crest ablation models has suggested that Splotch represents a mammalian animal model of failed neural crest migration. Some experimental studies have supported this supposition (Conway et al., 1997; Moase and Trasler, 1990), though the lack of appropriate murine crest-specific markers has made investigation of the fate of cardiac neural crest cells in Splotch (and wild-type) embryos difficult.

The cellular function of Pax3 in cardiac neural crest has remained unknown. Recent studies have suggested roles for Pax3 in neuronal differentiation (Koblar et al., 1999) and myelination of axons in the peripheral nervous system (Kioussi et al., 1995). By analogy to other cell types where Pax3 is expressed, it may also play a role in modulating neural crest migration, survival, proliferation and/or interaction with other tissues. For instance, in myogenic progenitors of hypaxial muscles, Pax3 is required for cell migration via pathways that include activation of the c-Met receptor (Bladt et al., 1995; Bober et al., 1994; Daston et al., 1996; Epstein et al., 1996; Goulding et al., 1994). Pax3 may also play important roles in myogenic differentiation (Maroto et al., 1997; Tajbakhsh et al., 1997), survival (Borycki et al., 1999) and proliferation (Barr et al., 1993; Galili et al., 1993; Shapiro et al., 1993).

In this report we have reinvestigated the question of whether Pax3 is required for neural crest migration to the heart. Using a neural crest-specific promoter element from the Pax3 gene to direct Cre-recombinase expression and appropriate Cre-reporter mice (Soriano, 1999), we have been able to track Pax3-expressing neural crest progenitors to the heart, even after Pax3 expression has abated. These results were independently confirmed using a neural crest specific transgene to fate map cardiac neural crest derivatives in the outflow tract of both wild-type and Splotch embryos. Such studies showed the presence of crest cells in the outflow tract of Pax3-deficient embryos. Further analysis of explanted cardiac neural crest cells from Pax3-deficient embryos revealed intact postotic neural crest cell migration, but at an altered rate. These studies suggest that Pax3 is not required for cardiac neural crest migration but may play a role in fine tuning the efficient targeting of crest cells to the outflow tract.

Splotch mice were obtained from the Jackson Labs and maintained on a C57/Bl6 inbred background. Embryos and adult mice were genotyped by PCR analysis of DNA derived from extra-embryonic membranes and tail biopsy, respectively, as described by Tajbakhsh et al. (1997). Cx43-lacZ mice have been described by Lo et al. (1997). P3pro-Cre consisted of the 1.6 kb proximal Pax3 regulatory region upstream of Cre recombinase. The transgene was created by digestion of P3pro-Pax3 (Li et al., 1999) with BstEII and NotI followed by ligation of a 1.4 kb PCR fragment encoding Cre recombinase and a bovine growth hormone polyadenylation site amplified with primers containing engineered BstEII (5′) or NotI (3′) sites. P3pro-Cre was linearized with NarI and NotI and gel purified to remove vector sequences, purified over an Elutip column and used for oocyte injection at the Institute for Human Gene Therapy Transgenic Core Facility at the University of Pennsylvania. Founders were screened for ability to pass the transgene to F₁ offspring, express Cre recombinase in an expected pattern as screened by in situ hybridization and express sufficient levels of Cre activity to result in lacZ-expressing cell populations after crossing with floxed lacZ-reporter mice (Soriano, 1999).

Embryos to be analyzed for lacZ expression were washed in PBS and fixed for 2 hours in 2% PFA in PBS, rinsed, and stained according to standard techniques. Whole-mount in situ hybridization was carried out essentially as described by Riddle et al. (1993) with modifications (Chen et al., 1999). Radioactive in situ hybridization was performed using ³⁵S-UTP and embryos fixed overnight in 4% PFA, dehydrated and embedded in paraffin and sectioned at 8 μm. Prehybridization, hybridization and wash conditions were as described by Lutz et al. (1994). Slides were exposed to photographic emulsion for 5 days prior to developing and counterstaining with Hoechst 33258 nuclear dye. Embryos and sections were photographed digitally and images were processed using Adobe Photoshop.

Sp⁻/⁻ embryos, fixed in 4% paraformaldehyde, were processed in a graded series of ethanol, cleared in toluene and embedded in Polyfin (Polysciences Inc., Warrington, PA). The specimens were serially sectioned (5 μm) and mounted on aminoalkylsilane-coated slides (Sigma Diagnostics, St. Louis, MO). For the detection of smooth muscle actin (SMA, Sigma, A2547) monoclonal antibody and the atrial Myosin Light Chain polyclonal antibody (aMLC; gift from Dr Kubalak) binding on mouse tissue sections, the indirect conjugated method was applied. Paraffin wax was removed in xylene, the sections were rehydrated through a series of ethanols, and then pretreated for 30 minutes with hydrogen peroxide (3% v/v in phosphate-buffered saline (PBS), pH 7.4) to reduce endogenous peroxidase activity, and preincubated with TENG-T (10 mM Tris, 5 mM EDTA, 150 mM NaCl, 0.25% gelatin, 0.05% Tween-20, pH 8.0) for an additional 30 minutes to minimize nonspecific binding of antibodies. The SMA antibody was diluted in PBS/BSA/NRS (PBS with 1% bovine serum albumin (BSA), and 2.5% normal rabbit serum (NRS)), and the polyclonal antibody was diluted in PBS/BSA. After incubation of the pretreated sections with the primary antibodies overnight, antibody binding was detected with rabbit anti-mouse peroxidase (RAM-Po; Sigma, A-9044) and goat anti-rabbit peroxidase respectively. All incubations were followed by three washes with PBS for 5 minutes. The immunocomplex formed was visualized using a Metal Enhanced Diaminobenzidine (DAB) Substrate Kit (Pierce, Rockford, IL; product number 34065). Sections were mounted in Accu-Mount 60 (Stephens Scientific, Riverdale, NJ). Digital images were collected using a Polaroid Digital Microscope Camera mounted on an Olympus BX40 microscope.

Sp⁻/⁻ embryos, stained as whole mounts for the presence of transgenic lacZ expression, were processed in a graded series of ethanol, cleared in iso-propanol, carefully oriented, embedded in Polyfin (Polysciences Inc., Warrington, PA), and serially sectioned. After removal of the paraffin wax and rehydration, individual sections on selected slides were pretreated with TENG-T for 30 minutes to minimize nonspecific binding of antibodies. Smooth muscle actin (SMA) was detected using anti-smooth muscle actin (Sigma, A2547). The binding of this monoclonal antibody was visualized with goat anti-mouse FITC. Fluorescent labeling was visualized using a Biorad MRC 1024 scanning laser confocal microscope.

To obtain embryos for neural crest outgrowth studies, mice were mated and the day the vaginal plug was found was considered E0.5. Embryos were collected at E8.5, and the yolk sac from each embryo was harvested for genotyping by PCR analysis. The method used to establish the neural crest outgrowth cultures has been described previously (Huang et al., 1998). Briefly, each embryo was treated with 0.5 mg/ml of collagenase/dispase (Sigma Co., St. Louis, MO.), then bisected longitudinally to expose the hindbrain neural folds. Each half was further transected to isolate the postotic region of the hindbrain, after which the dorsal ridge of the hindbrain neural fold was surgically removed from the surrounding tissues and cultured on fibronectin-coated 35 mm Petri dishes containing Dulbecco’s modified Eagle’s medium (DMEM) with high glucose and 10% fetal bovine serum. Digital images of the outgrowth cultures were captured at 24 and 48 hours by videomicroscopy, and then analyzed using the NIH Image software. We measured the area of the outgrowth and subtracted from it the area of the dense central neural tube mass, To control for differences in migration area resulting from random variations in the shape of the explant (as neural crest cells emerge only from the edges of the explant), the outgrowth area (mm²) was divided by the perimeter (mm) of the explant to arrive at the migration index. For monitoring cell proliferation, 48-hour explant cultures were incubated with BrdU (10 μm/ml) for 2 hours, then fixed, and processed for immunodetection using an anti-BrdU antibody in conjunction with the True Blue™ peroxidase substrate (Kierkegaard and Perry Laboratories, Gaithersburg, MD). To determine the total cell number in the outgrowth after BrdU immunostaining, the cultures were further stained with hematoxylin. The proliferation rate was then calculated as the percentage of BrdU-labeled cells over the total number of cells in the outgrowth. Labeled cells were counted with the aid of NIH Image. For time lapse videomicroscopy, images of the explants were captured every 10 minutes over a 20-hour interval using IPLab software. For these studies, the explant cultures were maintained at 37°C on a heated microscope stage in phosphate-buffered L-15 medium containing 10% fetal bovine serum. The captured images were then converted to a Quicktime movie for viewing and motion analysis. The motion analysis was carried out using the Dynamic Image Analysis Software (Solltech, Inc.).

In Splotch mice, a mutation in Pax3 at the splice acceptor site for exon four results in aberrant RNA splicing and a non-functional transcript (Epstein et al., 1993). Nevertheless, mutant RNA transcripts can be detected by in situ hybridization and cells that would normally be expressing Pax3 can therefore be detected. Pax3 is expressed by neural crest progenitors within the dorsal neural tube and mutations in Pax3 result in neural crest defects. A prior report has suggested that Pax3 is expressed by cardiac neural crest cells after migrating from the neural tube and within the outflow tract of the heart (Conway et al., 1997). We attempted to detect the expression of the non-functional Pax3 transcript in Splotch embryos to determine the fate of cardiac neural crest cells in the absence of functional Pax3. In parallel, we re-examined the normal expression pattern of Pax3 transcripts in cardiac neural crest and in the hearts of wild-type embryos. As shown in Fig. 1A (arrow), a stream of Pax3-expressing cells can be seen emerging from the dorsal region of the embryo at E10.5 and migrating ventrally towards the developing heart. This expression pattern is missing in Splotch littermates (Fig. 1B) as previously reported (Conway et al., 1997). However, a detailed examination of such whole-mount in situ hybridized embryos revealed that this stream of cells does not represent neural crest cells destined to populate the heart. Rather, these are hypoglossal myogenic progenitors populating the hypoglossal cords that will form the muscles of the neck (Mackenzie et al., 1998) (Fig. 1C). A detailed analysis by whole-mount in situ hybridization of embryos between E9.5 and E12.5 failed to detect Pax3-expressing cells within the heart, either in wild-type or Splotch embryos (Fig. 1 and data not shown).

These findings were confirmed by radioactive in situ hybridization studies. Analysis of consecutive sections through the cardiac and aortic arch regions of E9.5 to E12.5 wild-type embryos failed to reveal any specific Pax3 signal in the heart, outflow tract or great vessels (data not shown). Likewise, immunohistochemistry using a recently characterized Pax3 antibody (Li et al., 1999) failed to detect Pax3 protein expression in this region, though Pax3-positive cells in the neural tube, dorsal root and sympathetic ganglia and the limb buds were present (Fig. 3A and data not shown). Hence, we conclude that Pax3 is not expressed at detectable levels in the embryonic heart at these time points and that Pax3 expression significantly or completely abates prior to cardiac neural crest cell arrival in the outflow tract of the heart. These results raise the question of whether Pax3 expression is indeed associated with the cardiac neural crest cell progenitors. They also indicate the need for an alternate strategy for detecting cardiac neural crest cells in Splotch and wild-type embryos. Therefore, we have taken advantage of a Cre-lox approach in combination with the Pax3 promoter/enhancer to determine if Pax3 is expressed in cardiac neural crest progenitors.

We have created transgenic mice expressing Cre recombinase driven by the proximal Pax3 regulatory elements (Li et al., 2000). These regulatory elements when used to direct Pax3 expression, are sufficient to rescue cardiac development in Splotch embryos (Li et al., 1999). Hence they will likely target gene expression in the precise cell types in which Pax3 is required for cardiac development, presumably the cardiac neural crest progenitors. Examination of Cre recombinase expression in the Pax3-Cre transgenic mice by in situ hybridization confirmed a pattern of expression which recapitulated a subset of the endogenous Pax3 expression domains that included prominent expression in the dorsal neural tube (data not shown). This pattern of expression is identical to that previously observed for this regulatory element directing lacZ expression (Natoli et al., 1997; Li et al., 1999). When these Cre-expressing transgenic mice were crossed with a floxed-lacZ reporter mouse line (Soriano, 1999), lacZ expression was induced in neural crest cells that are derived from progenitors that normally express Pax3 (Fig. 2). For these studies, embryos at various stages of development were harvested to examine the distribution of β-gal-expressing cells. As the floxed lacZ locus once activated, will continue to provide β-gal expression even after Cre expression has ceased, the pattern of β-gal expression is in fact an accurate fate map of Pax3-expressing neural crest cells.

At E10.5, β-gal-expressing cells were seen invading the cardiac outflow tract, forming two parallel columns lining the truncus arteriosus (see arrow in Fig. 2A). This distribution is consistent with that previously described for cardiac crest cells (Kirby et al., 1983; Waldo et al., 1999). These results confirm that Pax3-expressing cells are fated to populate the outflow tract, even though Pax3 expression itself has abated by the time crest cells have migrated into the heart. By E12.5, the truncus arteriosus has septated into two vessels, the aorta and pulmonary arteries. Labeled cells derived from Pax3-expressing precursors are seen within the aorticopulmonary septum (arrow, Fig. 2C) and within the walls of the great vessels. In addition, labeled cells populate the dorsal root ganglia (arrows, Fig. 2D) which are also neural crest derived.

A striking difference was noted in the abundance of β-gal-expressing cells populating the third, fourth and sixth pharyngeal arches as compared to the first two pharyngeal arches which contained very few labeled cells (Fig. 2B). Note that Pax3 itself is not expressed at significant levels by cells of the pharyngeal arches at this time (Fig. 1A). While we cannot rule out that our Cre-expressing transgene is missing regulatory elements required for expression by Pax3-positive precursors of the first two arches, these results suggest that Pax3-expressing neural crest cells specifically populate arches 3, 4 and 6. This finding is of significance given that only crest cells associated with these three caudal arches participate in cardiac development (Phillips et al., 1987).

Pax3pro-Cre transgenic mice were useful for identifying cardiac neural crest cells in wild-type embryos, as demonstrated above. However, the usefulness of this system for tracking cardiac neural crest in Splotch embryos is unclear since any potential alterations of β-galactosidase expression could be attributed to altered Pax3 promoter activity in the absence of Pax3. Therefore, to determine if cardiac neural crest cells in Splotch embryos migrate to the cardiac outflow tract, we used a transgene composed of a connexin 43 (Cx43) promoter driving lacZ (Lo et al., 1997). This transgene provides β-gal expression in mouse neural crest cells, including the cardiac crest cells (Lo et al., 1997; Waldo et al., 1999). It contains 6.5 kb of the proximal upstream genomic regulatory regions of Cx43. Although the specific regulatory regions responsible for tissue specific expression have not been identified, the pattern of β-galactosidase expression in the developing heart displayed by Cx43-lacZ embryos (Waldo et al., 1999) is strikingly reminiscent of the pattern of β-galactosidase expression that represents Pax3 fate mapped cardiac neural crest cells (compare Fig. 2A with 4A, 2C with 6A). Therefore, we chose to use this transgene to track neural crest cells in Splotch embryos. However, first we had to determine whether absence of Pax3 activity in Splotch embryos might perturb Cx43 promoter activity. This was examined by analyzing Cx43 expression by in situ hybridization of E10.5 wild-type and Splotch embryos. As shown in Fig. 3A and C, Pax3 and Cx43 display similar expression patterns in wild-type embryos. Both are expressed in the dorsal neural tube (arrows) and in neural crest derivatives including dorsal root and sympathetic ganglia. In Splotch embryos, Cx43 expression is grossly unchanged (Fig. 3D) while Pax3 protein is absent (Fig. 3B), demonstrating that Pax3 is not required for tissue-specific Cx43 expression. These results suggested that the Cx43-lacZ transgene can be used as a marker to track neural crest cells in Splotch embryos.

We crossed Cx43-lacZ transgenic Splotch mice with Splotch heterozygotes and litters were examined at E10.5. In wild-type embryos, Cx43-lacZ-labeled neural crest cells emerge from the neural tube along the entire rostral-caudal axis of the embryo at evenly spaced intervals (Fig. 4B, arrowheads). In Splotch embryos, labeled cells are also seen emerging from the neural tube though the intensity of staining is reduced (Fig. 4D, arrowheads). By E12.5, these β-gal-expressing cells are observed in all of the dorsal root ganglia in wild-type embryos (Fig. 5A), while only a few irregularly spaced and small clusters of labeled cells were evident in Splotch embryos (Fig. 5B). We confirmed that the labeled cells were neuronal in nature by performing immunohistochemistry using a neuralspecific neurofilament monoclonal antibody (not shown). In Splotch embryos, neurofilament staining confirmed a paucity of ganglionic structures and those that were detected were small. The Cx43-lacZ transgene also labeled cranial ganglia in wild-type embryos (Fig. 5C). In addition to abnormalities previously reported for cranial ganglia in Splotch (Tremblay et al., 1995), we noted a marked reduction in the facial ganglia and nerves (compare arrows, Fig. 5C,D). The Cx43-lacZ transgene also labeled all pharyngeal arches in wild-type embryos (Fig. 4B, arches are numbered). We found a significant decrease in the size of pharyngeal arches 3, 4 and 6 in Splotch embryos (Fig. 4D,F), consistent with the observation that Pax3-expressing neural crest cells were specifically fated to populate these arches as demonstrated in Fig. 2B. Significantly, these caudal arches contain crest cells that are destined to migrate to the heart.

Examination of E10.5 embryos revealed β-gal-positive crest cells in the outflow tract of both the wild-type and Splotch mutant embryos. Fig. 4A shows two β-gal-positive neural crest cell streams migrating into the cardiac outflow tract of wild-type E10.5 embryos (arrow). This same pattern was also observed in the outflow tract of Splotch embryos (Fig. 4C,E), but generally the labeled crest cells show less distal progression into the outflow tract as compared to the crest cell streams found in wild-type embryos of the same developmental stage (compare Fig. 4C,E with Fig. 4A). Nevertheless, it is of significance to note that in all cases, some labeled cells were identified in the cardiac outflow tract of the Splotch mutant embryos.

Further examination of wild-type and Splotch Cx43-lacZ hearts at E12.5 and E13.5 confirmed that neural crest cells are capable of long range migration in the absence of Pax3. Between E12.5 and E13.5, the outflow tract of the heart completes septation to give rise to two distinct vessels, the aorta and the pulmonary artery (Fig. 6A,D). At these stages of development, an abundance of β-gal-positive cells are found proximally at the base of the outflow tract (arrows, Fig. 6A,D).

Histologic analyses have previously shown that these correspond to β-gal-positive crest cells situated in the outflow septum (Waldo et al., 1999). In Splotch mutant embryos exhibiting absent or abnormal septation of the outfow tract, β-gal-positive cells can still be seen in the heart (Fig. 6B,C,E), albeit with reduced abundance in mutants with the most severe cardiac abnormalities (see below).

In Splotch mutants exhibiting the absence of outflow septation, that is with persistent truncus arteriosus (PTA), β-gal-positive cells are significantly reduced in abundance in the outflow tract (Fig. 6B,C; compare to Fig. 6A,D). In contrast, in Splotch mutant embryos in which partial septation has taken place, i.e. in embryos exhibiting a double outlet right ventricle (DORV), β-gal-positive cells are found at levels approaching that of wild-type embryos, but with a distinctly different distribution (Fig. 6E, compare to 6D). This may reflect the abnormal positioning of the outflow septum in these Splotch mutants (Fig. 6E) which is associated with a failure of rotation that normally juxtaposes the aorta with the left ventricle. Injection of dye into such DORV hearts (E13.5) indicates the presence of a ventricular septal defect, an abnormality frequently seen in conjunction with cardiac defects arising from cardiac neural crest perturbation (Fig. 6F).

Wild-type and Splotch Cx43-lacZ transgenic hearts were sectioned after staining for β-galactosidase activity to determine the precise distribution of neural crest cells within the outflow tract. Fig. 7A (arrow) shows the normal location of neural crest cells clustered between the aorta and pulmonary artery (see also Waldo et al., 1999). In Fig. 7B-D, serial sections through the outflow tract of a Splotch transgenic embryo with PTA reveal an atypical pattern of neural crest cell distribution. Some β-galactosidase positive cells were located within the myocardial layer instead of within the mesenchymal regions of the endocardial cushions (arrows, Fig. 7B,D). These results suggest disturbances in the targeting of neural crest cells in the outflow tract of Splotch mutants exhibiting PTA.

The Splotch heart with PTA shown in Fig. 6C was sectioned and examined histologically confirming a complete absence of outflow tract septation. Shown in Fig. 8A-D are sections of β-gal-positive cells seen in the outflow endocardial ridges of this mutant Splotch heart (arrows, Fig. 8B). Neural crest origin of these lacZ-positive cells was indicated by immunohistochemical analysis of smooth muscle actin (SMA) expression, a marker found in cardiac neural crest cells but not in mesenchymal cells of the endocardial cushions (Waller III, R., A. Phelps, R. R. Markwald, C. W. L., R. P. Thompson and A .W., unpublished data). LacZ-positive cells expressed SMA (arrows, Fig. 8C,D). We also examined the expression of atrial myosin light chain (aMLC, Fig. 8E) compared to SMA (Fig. 8F) in adjacent sections of Splotch hearts. aMLC is expressed by myocardial cells while SMA is expressed by myocardial cells and neural crest cells. Examination of serial sections through a Splotch mutant heart showed some SMA positive cells populating the compact mesenchyme of the endocardial cushions (arrows, Fig. 8F). Significantly, these SMA-positive cells do not express aMLC (Fig. 8E). These results also confirm the presence of neural crest derivatives in the outflow tract of Splotch mutant embryos.

Since neural crest cells from Splotch embryos appeared to emerge from the neural tube at approximately the correct stage of embryonic development, we sought to analyze the migratory behavior of cardiac neural crest cells using an explant culture system. For these studies, E8.5 neural tube segments caudal to the mid-otic placode and rostral to somite 3 were dissected from wild type, Sp^+/− and Sp^−/− littermates and cultured for 24 or 48 hours to monitor neural crest outgrowth. By measuring the area of the outgrowth, we obtained an estimate of the overall rate of neural crest migration, referred to as the migration index (MI in Table 1, see methods for details). This analysis showed no deficiency in the ability of neural crest cells to emerge and migrate from the Sp^−/− neural tube explants. In fact, we observed a marked elevation in the apparent rate of neural crest migration in Sp^−/− embryos as compared to that of Sp^+/− or wild-type embryos (Table 1). This is indicated by a significant increase in the migration index of the Sp^−/− neural crest outgrowths (Table 1). This was observed both at 24 hours and 48 hours of culture. Analysis of BrdU incorporation demonstrates this increase in the migration index is not likely accounted for by changes in the rate of cell proliferation (Table 1). To further examine whether the rate of neural crest cell emergence may be altered in the Sp^−/− embryos, we carried out time lapse videomicroscopy with images of the outgrowth cultures captured every 10 minutes over a 20-hour interval. Such analysis showed no difference in the rate of crest cell emergence from Sp^−/− explants as compared to that of heterozygous or wild-type explants. Overall, these results are consistent with the hypothesis that Pax3 is not required for the emergence or migration of cardiac neural crest cells. They further suggest that Pax3 is nevertheless important for crest cell migration, possibly by fine tuning the migratory behavior of the cardiac neural crest cells.

These studies provide a series of observations that alter our understanding of Pax3 function in cardiac neural crest development. Pax3 is expressed by neural crest progenitors in the dorsal neural tube and by other neural crest derivatives such as the dorsal root ganglia and various cranial ganglia (Goulding et al., 1991; Tremblay et al., 1995). However, our studies indicated that Pax3 is down-regulated in many neural crest lineages. Of particular significance for cardiac morphogenesis is the fact that in wild-type embryos, Pax3 expression was not detected in large regions of the pharyngeal arches or in the outflow tract of the heart. Instead, we observed the persistent expression of Pax3 by migrating hypoglossal myoblasts that previously were misinterpreted as Pax3-expressing neural crest cells (Conway et al., 1997). This unexpected finding compelled us to reexamine the role of Pax3 in the cardiac malformations of Splotch mutants, and its involvement in the modulation of neural crest cell migration.

Utilizing a binary Cre-lox system to activate β-galactosidase expression in Pax3-expressing cells and their descendants, we fate mapped Pax3-expressing cells without the need for tissue transplantation or other experimental manipulations. Such studies showed that Pax3-expressing neural crest cells are fated to populate pharyngeal arches 3, 4 and 6, the three caudal arches known to contribute crest cells involved in outflow tract septation (Phillips et al., 1987). Importantly, we also observed that the Pax3-expressing crest cells were fated to invade the cardiac outflow tract. These studies relied on the use of the proximal Pax3 promoter that we have previously characterized (Li et al., 1999) to direct expression of Cre recombinase. While this promoter was sufficient to direct Pax3 expression in Splotch embryos resulting in correction of gross congenital cardiac defects, it remains possible that regulatory elements not included in this transgene are important for directing expression in cranial pharyngeal arches or in cells that contribute to later cardiac development or function.

We also examined the distribution of neural crest cells using the Cx43-lacZ transgene marker. This lacZ marker was chosen for our studies as it is well characterized with regard to the fidelity with which it labels cardiac neural crest cells (Waldo et al., 1999). Moreover, our in situ hybridization studies showed that Cx43 promoter activity is not altered by the loss of Pax3 function, and thus validated the use of this transgene marker in the Splotch mutants. These studies showed that cardiac neural crest cells do emerge and migrate to the heart outflow tract in Splotch mutant embryos. In addition, we observed a reduction in the size of the three caudal arches, and in the abundance of β-gal-expressing crest cells in the outflow tract. Similar studies are ongoing using our Cre-lox system to examine the distribution of crest cells in Splotch mutants. As this involves a three way cross, a large number of embryos in several generations would have to be screened in order to obtain the desired β-gal-expressing Splotch mutant embryos. Furthermore, the ability of Pax3 to regulate its own promoter will have to be evaluated in order to interpret the results that emerge.

Studies with the Cx43-lacZ marker also showed that trunk neural crest cells in Splotch embryos emerge and participate in the formation of the DRG, albeit the DRG are greatly reduced in numbers and size, and appear to be malpositioned and abnormally organized. We also observed crest cells participating in the formation of the cranial ganglia in Splotch mutants. Notable was the absence of the facial nerve in Splotch embryos. Although this finding has not been previously reported, our results are generally consistent with previous observations obtained by others using three different lacZ transgenic markers in Splotch embryos (Kothary, 1991; Serbedzija and McMahon, 1997; Tremblay et al., 1995). Variations among the results of these studies may in part reflect strain background differences, as has been described for the DRG deficiency seen in the Sp vs. Sp^d mutants (Tremblay et al., 1995).

The cardiac outflow tract defects and other neural crest related abnormalities in Splotch mutants are strikingly similar to defects seen in chick embryos after ablation of premigratory neural crest cells, and a neural crest migration defect secondary to Pax3 deficiency has been postulated (Conway et al., 1997; Franz, 1989). By examining the expression of the Cx43-lacZ transgene in wild-type and Splotch embryos, we demonstrated that Pax3 is not required for cardiac neural crest migration to the heart. However, the distribution of neural crest cells within the cardiac outflow tract was not normal in Splotch. In normal embryos, two parallel columns of neural crest cells invade the truncus arteriosus and form a spiral structure that progresses in a rostral to caudal direction. In Splotch, these two columns of cells are also present, but they are abnormally positioned, and in some embryos, the intensity of β-gal staining is also reduced, suggesting that there is a marked reduction in the abundance of cardiac crest cells. Interestingly, the degree to which the intensity of staining was reduced appeared to be correlated with the severity of the cardiac defect, such that embryos with PTA exhibited fewer cardiac neural crest cells in the outflow tract region than those exhibiting DORV. Whether these differences in crest cell abundance are causally related to the cardiac defect remains to be determined. In this regard, our findings are generally consistent with those of Conway et al. (1997) who used a series of markers in addition to Pax3 itself to demonstrate a reduction in neural crest cells within the outflow tract in severely affected Splotch embryos. Some differences between those results and ours may also be related to the use of different Splotch alleles.

A role for Pax3 in regulation of neural crest migration has been supported by analogy to the role of Pax3 in migratory muscle precursors. In limb muscle progenitors, compelling evidence indicates that Pax3 can regulate migratory processes. Pax3 is required to mediate migration of myoblasts to the developing limb bud (Bladt et al., 1995; Bober et al., 1994; Daston et al., 1996; Epstein et al., 1996; Goulding et al., 1994). This process involves the direct activation of the c-met gene by Pax3, a process that is required for myoblasts to migrate in response to the secreted ligand for c-Met, scatter factor, which is expressed by limb mesenchyme (Brand-Saberi et al., 1996). In the absence of Pax3 (or c-Met), limb myoblasts fail to delaminate from the ventral-lateral lip of the somite and there is a complete absence of myoblasts invading the limb bud (Bladt et al., 1995; Daston et al., 1996; Epstein et al., 1996; Franz et al., 1993).

An equivalent role for Pax3 in the regulation of neural crest migration has not been reported. Our analysis of cardiac crest cell migration in explant cultures showed no evidence of a defect or delay in the emergence of neural crest cells. This result contrasts with a previous study showing delayed emergence of trunk neural crest cells from neural tube explants derived from Splotch embryos (Moase and Trasler, 1990). This discrepancy may reflect the fact that the previous study involved the analysis of trunk neural crest cells rather than vagal or postotic crest cells that give rise to the cardiac neural crest. However, another in vivo study using a lacZ reporter transgene controlled by a wnt1 promoter element reported normal emergence of trunk neural crest cells in Splotch embryos (Serbedzija and McMahon, 1997). Surprisingly, these authors also reported absence of labeled neural crest cells in the caudal trunk region. This result conflicts with our findings and that of others using different lacZ reporter constructs (Kothary, 1991; Tremblay et al., 1995). The failure to detect wnt1-lacZ-expressing crest cells in the caudal trunk could reflect the down regulation of wnt1 promoter activity in Splotch embryos. Interestingly, in this same study, labeled neural crest cells derived from the caudal trunk level of Splotch embryo neural tube fragments were observed to migrate after transplantation into chick embryos. This was interpreted to indicate that the migration defect in Splotch embryos is non-cell autonomous, perhaps secondary to the absence of Pax3 in non-neural crest tissues (such as the somite) that indirectly affected neural crest migration. However, this interpretation is probably incorrect, as we recently showed that reconstitution of Pax3 expression only in neural crest populations is sufficient to rescue cardiac septation and dorsal root ganglia formation in Splotch embryos (Li et al., 1999). An alternative interpretation of the chimera studies is that in the presence of wild-type Pax3-expressing chick cells, wnt-1 promoter activity may be up regulated via a non-cell autonomous mechanism. The resulting up regulation of β-gal expression would then give rise to the appearance of neural crest cell “rescue” in these embryos.

Our studies also indicate that there is a significant elevation in the apparent rate of crest migration in the absence of Pax3. This is based on the quantitative analysis of the outgrowth area in the neural crest outgrowth cultures. This result contrasts with the finding of fewer crest cells in the outflow tract. These apparently conflicting results suggest that the targeting of neural crest cells to the heart is compromised. For example, the loss of Pax3 function may perturb the directionality or persistence of cardiac crest cell movement, or alter the chemotropic response of crest cells to migratory cues. To address these questions, we are currently examining neural crest migration by time lapse videomicroscopy, and analyzing their migratory response to various growth factors known to be expressed along the cardiac crest migratory route.

It is also possible that Pax3 may have other roles in neural crest cells unrelated to migration, including modulation of cell survival. In limb myoblasts, we have noted increased apoptosis in the absence of Pax3 (Borycki et al., 1999). We also noted clusters of apoptotic cells in the regions usually occupied by DRGs, suggesting programmed cell death of Pax3-deficient neural crest cells. We were unable to detect apoptotic cardiac neural crest cells in the hearts of Splotch embryos, but the normal high levels of apoptosis in mid-gestation endocardial cushions and cardiac mesenchyme makes such studies technically difficult. Given the documented ability of Pax3 to suppress apoptosis in other systems (Bernasconi et al., 1996; Phelan et al., 1997), it is possible that changes in the level of apoptosis may contribute to the altered distribution of neural crest cells in the hearts of Splotch embryos.

In summary, our studies indicate that Pax3 is normally expressed by cardiac neural crest progenitors and is down-regulated prior to cardiac neural crest population of the outflow tract of the heart. Pax3-dependent neural crest cells are also fated to populate pharyngeal arches 3, 4 and 6, and in the absence of Pax3, these arches are hypoplastic. While neural crest defects including cardiac outflow tract disorders characterize Pax3-deficient Splotch embryos, Pax3 is not required for postotic neural crest cell migration in explant cultures or cardiac neural crest cell migration into the heart outflow tract in vivo. Taken together, our results suggest that Pax3 is not required for the emergence or migration of neural crest cells. Rather, the role of Pax3 is likely more subtle, perhaps involving the fine tuning of the migratory behavior of crest cells or in mediating other critical roles in neural crest function.

We thank Aimee Phelps for technical assistance. This work was supported by grants from the NIH (HL62974, HL61475), March of Dimes and the W. W. Smith Charitable Trust to J. A. E., from the NIH (HL52813) to A. W., and from the NIH (HD29573) and N. S. F. (IBN31544) to C. W. L.