Generation of a conditional loxP allele of the Pax3 transcription factor that enables selective deletion of the homeodomain

Pax3 is one of nine members of the paired-box-containing transcription factors (Stuart et al., 1994) and is widely expressed in many tissues and organs, including neural tube/neural crest and somites. Pax3 is located on human 2q35 and mouse 1(44.0cM) chromosomes and contains two structurally independent DNA-binding domains, the pairedand homeo-domains. Biochemical/ mutagenesis studies have shown that both domains are functionally interdependent, and that the paired-domain can regulate DNA-binding specificity and dimerization potential of the homeo-domain (Fortin et al., 1998). Loss-of-function mutations in both mouse (Splotch; Auerbach, 1954) and man (Waardenburg syndrome; Baldwin et al., 1995) can result in a wide variety of developmental abnormalities, as does the gain-of-function mutation resulting in the PAX3-forkhead fusion that is associated with alveolar rhabdomyosarcomas (Mansouri, 1998). There are six classical and radiation-induced alleles of Splotch, which all have different pax3 mutations, phenotypes, and times of lethality (reviewed by Chalepakis et al., 1993; Machado et al., 2001). Significantly, using transgenic mice overexpressing pax3 in only the Sp mutant neural tube and neural crest (but not in somites) can rescue neural tube closure and cardiac/neural crestrelated defects and suggests a cell autonomous role for pax3 (Li et al., 1999). In contrast, a lacZ knock-in into pax3-generating pax3-deficient embryonic stem (ES) cells has shown that pax3 does not act cell autonomously in Sp chimeric embryos (Mansouri et al., 2001), as a wild-type environment could rescue pax3deficient neural crest morphogenesis. Additionally, both Sp and Sp homozygotes have neural crest-associated heart defects (Li et al., 1999; Conway et al., 2000) but exhibit differences that suggest that the pathogenesis of the defects may be different as a result of the type of pax3 mutation. Sp mutants have a deletion within the paired-domain and is thought to be a null allele, whereas Sp mutants have a normal paired-domain but lack a functional homeo-domain and may function as a hypomorphic allele (Chalepakis et al., 1993). Whereas Sp hearts have an abnormally thin myocardium and overexpress cyclin-dependent kinase inhibitor p57Kip2 (Kochilas et al., 1999), the Sp mutants have a normal myocardium and have a calcium-handling defect (Conway et al., 1997a, 1997b) and do not overexpress p57Kip2 (not shown). Thus, despite the existence of all these different mouse models, the precise function of pax3 remains unclear. Given widespread expression of pax3, contradictory results, and the fact that Sp allele may act as a dominant negative, we generated floxed pax3 mice (Fig. 1) to recreate a similar mutant pax3 molecule as that present within Sp mutants, but in specific subsets of tissues and at different times during development. Using Cremediated recombination to remove the floxed pax3 homeo-domain, polymerase chain reaction (PCR) screening (Fig. 2) revealed that none of the pax3 / homozygous knockout mutants were born, and that the pax3 heterozygous mutant mice exhibited white spots on their forehead and belly, similar to those seen in Splotch heterozygotes (Auerbach, 1954; Conway et al., 1997b). Additionally, when fetuses were harvested at 14.0 dpc, only dead pax3 / homozygotes were present, indicating that the presence of at least one copy of the homeodomain is required for normal development and in utero survival. This floxed pax3 mouse will now enable us to test whether the neural crest defects can be separated from the accompanying defects within the somites and neural tube by using neural tube/neural crest-specific Cre mice and whether the targeted conditional mutation of pax3 within the myocardium using a heart-specific Cre mouse is sufficient to account for the observed cardiac defects.