β-catenin regulates Pax3 and Cdx2 for caudal neural tube closure and elongation

Mammalian neural tubes exhibit an initial closure at the future hindbrain and cervical boundary, subsequent closures at several different brain regions, and a final closure at the posterior neuropore (PNP) (Bassuk and Kibar, 2009; Copp and Greene, 2010). Failure of closure at these anatomically distinct zones may lead to a spectrum of neural tube defects (NTDs). The cause and prevention of NTDs remain poorly understood (Copp et al., 2013; Wallingford et al., 2013). NTDs are common birth defects with inheritable risks, but only a small number of gene mutations have thus far been linked with human NTDs (De Marco et al., 2011; Kibar et al., 2007). From mutant mouse studies, more than 200 genes have been linked with NTDs (Harris and Juriloff, 2007; Harris and Juriloff, 2010), reflecting the complex genetic basis of neural tube closure.

β-catenin-independent non-canonical Wnt/planar cell polarity (PCP) signaling regulates cytoskeleton dynamics and is a core signaling program for oriented tissue movements such as convergent extension, which is a potential driving force directing neural tube closure and body axis elongation (Copp et al., 2003; Wallingford, 2006; Ybot-Gonzalez et al., 2007). Craniorachischisis, a severe type of NTD associated with defective convergent extension, has been observed in mutant mice of the Celsr, Dvl, Fzd and Vangl PCP signaling gene families, suggesting that mutations in these genes might also cause NTDs in humans (De Marco et al., 2011; Juriloff and Harris, 2012). However, several of these PCP signaling molecules, such as Fzds and Dvls, are also essential components of the canonical Wnt/β-catenin signaling pathway (MacDonald et al., 2009). In addition, Lrp6 is a key co-receptor in the canonical Wnt pathway and is required for a wide range of organogenetic events, including neural tube closure in mice (Carter et al., 2005; Kokubu et al., 2004; Mao et al., 2001; Pinson et al., 2000; Song et al., 2009; Song et al., 2010; Tamai et al., 2000; Wehrli et al., 2000; Zhou et al., 2008; Zhou et al., 2004; Zhou et al., 2010). However, Lrp6 is also implicated in convergent extension during Xenopus gastrulation (Tahinci et al., 2007) and may mediate non-canonical Wnt signaling for neural tube closure (Gray et al., 2013). Thus, the role of canonical Wnt signaling, and especially of β-catenin, in neural tube closure and NTDs remains unknown.

The canonical Wnt/β-catenin pathway has vital roles in development and disease (Clevers and Nusse, 2012). When Wnt proteins bind to the Fzd and Lrp receptors, the degradation function of the Axin1-APC-Gsk3 complex is repressed, resulting in elevated levels of free β-catenin (Li et al., 2012). The accumulated cytoplasmic β-catenin translocates to the nucleus where it binds the Tcf/Lef transcription complex to regulate the promoter activity of downstream target genes for various biological functions (Cadigan, 2012). We hypothesize that canonical Wnt/β-catenin signaling is required for neural tube closure through transcriptional regulation of crucial downstream target genes. β-catenin-null mouse embryos arrest during gastrulation (Haegel et al., 1995; Huelsken et al., 2000). β-catenin conditional gene-targeting analyses have been performed widely in various developmental processes, although not specifically with respect to neural tube closure (Grigoryan et al., 2008). This study focuses on the signaling function of β-catenin, and presents evidence that β-catenin is required for transcriptional activation of the paired-box gene Pax3 and the caudal-type homeobox gene Cdx2, which are crucial effectors for caudal neural tube closure and/or elongation processes (Epstein et al., 1991; Savory et al., 2011; Young et al., 2009). We further demonstrate that ectopic activation of a Pax3 transgene can rescue spina bifida in β-catenin mutant mice and that Pax3 is required, but insufficient, for Cdx2 activation in the dorsal PNP. These results reveal a novel cascade of β-catenin/Pax3/Cdx2 signaling that is required for PNP closure and/or elongation processes during neurulation.

To define the role of β-catenin signaling in neural tube closure, we employed the Cre-loxP conditional gene-targeting approach using β-catenin^ex(2-6)flox mice (Brault et al., 2001) crossed with Pax3^Cre knock-in mice (Lang et al., 2005). A genetic fate-mapping experiment with the Cre reporter Rosa26-lacZ demonstrates high activities of Cre recombinase driven by endogenous Pax3 promoters in the dorsal neural tube, particularly around the closure sites of the PNP regions of E9.5 mouse embryos (Fig. 1A). Wholemount in situ hybridization and real-time PCR experiments demonstrate that β-catenin mRNA expression is effectively ablated in the dorsal neural tube/fold, whereas it is unaffected in other tissues including the paraxial mesoderm around the PNP region, during the closure process of E9.5 β-catenin^{(ex2-6)flox/flox};Pax3^Cre/+ conditional knockout (abbreviated as β-catenin cKO) mutant embryos (Fig. 1B,C).

Because β-catenin is a key molecule in the canonical Wnt pathway, we analyzed Wnt signaling alterations in the β-catenin cKO embryos. The expression of Axin2, a general Wnt/β-catenin signaling downstream target and negative-feedback regulator in the canonical Wnt/β-catenin signaling pathway (Jho et al., 2002), is repressed in the dorsal neural tube, as shown at E9.5 (Fig. 2A,B). The activity of the Wnt/β-catenin signaling reporter BATgal (Maretto et al., 2003) is also dramatically diminished in the dorsal neural tube of β-catenin cKO embryos at E9.5 (Fig. 2C). Whereas the control neural tube closed completely by E10.5, all β-catenin cKO embryos exhibited persistently open PNPs at this age, which consequently developed as spina bifida aperta, as shown at E14.5 (Fig. 2D). The mutant embryos also exhibit dorsally bent and shorter PNPs at E10.5 and kinked and truncated tails at E14.5. These results demonstrate an essential role of β-catenin signaling in PNP closure and elongation processes in mice.

To address the β-catenin-regulated signaling mechanism in the PNP, we examined a panel of locally expressed and functionally important transcription factors and signaling molecules in the mutant embryos. The transcription factor Pax3 is expressed in the dorsal neural folds/tubes during the closure process, and Pax3-null embryos display spina bifida and/or exencephaly, two common types of NTDs (Epstein et al., 1991; Goulding et al., 1991). In addition, Pax3 has been suggested to mediate Fgf8 and/or Wnt signaling for neural crest development (Degenhardt et al., 2010; Monsoro-Burq et al., 2005). Therefore, we hypothesize that Pax3 is a downstream effector of Wnt/β-catenin signaling for the caudal neural tube closure process.

Wholemount in situ hybridization demonstrates that Pax3 expression is diminished markedly in the dorsal PNP of β-catenin cKOs at E9.5 (Fig. 3A). Pax3 expression is diminished more dramatically in the early-born somites, and moderately in the late-born somites of the E9.5 mutants (Fig. 3A). The caudal-type homeobox genes Cdx2 and Cdx4 are also specifically diminished in the dorsal PNP of β-catenin cKOs at E9.5 (Fig. 3B,C). Cdx2 is a survival factor for early embryos around implantation and is also required for caudal body axis elongation during neurulation (Young et al., 2009). Cdx4 is dispensable in mice (van Nes et al., 2006). In addition, the homeobox gene Msx1 is diminished throughout the dorsal neural tube, especially in the dorsal PNPs, and in the upper spinal and cranial regions of the β-catenin mutants (Fig. 3D). Although Msx1 does not play any substantial role in neural tube closure, it is a potential regulator of Pax3 (Monsoro-Burq et al., 2005) and is also a known β-catenin downstream effector (Foerst-Potts and Sadler, 1997; Song et al., 2009). These results suggest that β-catenin is the common upstream regulator of crucial transcription factors required for PNP development.

To determine the molecular mechanisms by which β-catenin signaling regulates the key downstream factors involved in caudal neural tube formation, we searched the upstream promoter region of the mouse Pax3 gene, which is specifically activated in the dorsal neural tube (Natoli et al., 1997), and found three conserved putative Tcf/Lef1 binding activation sites (termed AS1-3, Fig. 4A; supplementary material Fig. S1). Among these binding sites, AS1 is crucial for Wnt signaling responsiveness, as demonstrated in luciferase reporter assays (Fig. 4B-D). Chromatin immunoprecipitation assays demonstrate the specific binding of the β-catenin-Tcf complex to AS1 in the Pax3 promoter in tissue samples isolated from the caudal part of E9.5 mouse embryos (Fig. 4E). We found a conserved Tcf/Lef1 binding site in the putative Cdx2 promoter region, which also demonstrates binding and responsiveness to the β-catenin transcriptional complex (Fig. 4E-G). These results suggest that both Pax3 and Cdx2 are direct transcriptional targets of β-catenin signaling in the dorsal PNP.

The tail truncation in the β-catenin cKO embryo suggests a defective tail bud. The tail bud is the most posterior signaling center, expressing various signaling molecules including T-box genes, Fgfs and Wnts (Gofflot et al., 1997), and is functionally important in posterior body axis development. Wholemount in situ hybridization demonstrates that the expression patterns of brachyury (T) (Wilkinson et al., 1990), Tbx6 (Chapman and Papaioannou, 1998), Wnt5a (Yamaguchi et al., 1999a), Fgf8 (Crossley and Martin, 1995) and Fgf18 are apparently unaltered in the mutant tail buds at E9.5 (Fig. 5A-E). The expression patterns of Mesp2 in the rostral presomite mesoderm (Takahashi et al., 2000) and of Uncx4.1 (also known as Uncx) in the newly formed somites (Mansouri et al., 1997) are also not significantly altered in the caudal region of E9.5 β-catenin cKOs (Fig. 5F,G). In addition, the expression patterns of the roof plate marker genes Lmx1a and Lmx1b (Mishima et al., 2009) are intact in the mutant embryos at E9.5 (Fig. 5H,I). Moreover, neither representative PCP signaling gene expression patterns nor cell proliferation or apoptosis are significantly affected in the mutant PNPs at E9.5 (supplementary material Figs S2, S3).

Among the crucial tail bud patterning genes, T, Tbx6 and Fgf8 have been identified as the downstream targets of Wnt/β-catenin signaling in the tail bud or in the anterior neural ridge (Szeto and Kimelman, 2004; Wang et al., 2011; Yamaguchi et al., 1999b). Because these genes are not altered at E9.5 in the tail buds of the mutants, we then examined these genes at a later stage when neural tube closure is completed in wild-type or double-heterozygous embryos. Indeed, the in situ mRNA signals of these genes are markedly diminished in the mutant tail buds at E10.5 (Fig. 6) to varied degrees, with dramatically diminished T expression in the tail bud but relatively unaffected T expression in the primitive streak (Fig. 6A-B′), and nearly absent Tbx6 (Fig. 6C-D′) and residual Fgf8 (Fig. 6E-F′) tail bud expression. Uncx4.1 in situ results demonstrate conserved somite numbers, with disrupted somite organization and/or formation in the caudal axis of the E10.5 mutants. These results suggest that a delayed disruption of the tail bud patterning genes contributes to the defective posterior axis in the β-catenin cKO mutants at later ages.

To further demonstrate that Pax3 is a key effector of β-catenin signaling in the caudal neural tube closure process, we conducted a genetic rescue experiment using Cre-mediated gain-of-function (GOF) of Pax3 (Rosa26-loxP-stop-loxP-Pax3-cDNA) (Wu et al., 2008) in the β-catenin cKO mutants. We found that six out of eight (75%) β-catenin cKO;R26-Pax3-GOF compound embryos [obtained by crossing β-catenin^{(ex2-6)flox/flox};Rosa26-loxP-stop-loxP-Pax3-cDNA^+/- with β-catenin^{(ex2-6)flox/+};Pax3^Cre/+] had closed caudal neural tubes with restored Pax3 mRNA and protein in the dorsal PNP (Fig. 7A-D). Although in situ hybridization and immunolabeling for Pax3 demonstrate that the expression level of the Pax3 cDNA single allele activated by Cre in the β-catenin cKO is lower than the normal level of endogenous Pax3 expression in control embryos (Fig. 7C,D and Fig. 3A), the ectopically activated Pax3 is able to direct PNP closure in the absence of β-catenin in the compound mutants. Interestingly, an abnormally thick roof plate is seen in the closed PNP of the β-catenin cKO;R26-Pax3-GOF embryos at E12.5 (Fig. 7D), which is likely to be related to the ectopic expression of Pax3 cDNA in the mutant mice although the mechanism is as yet undetermined.

A recent study reports that Cdx2 is also involved in neural tube closure in addition to its established role in caudal body axis extension (Savory et al., 2011). We examined whether Cdx2 interacts with Pax3 in PNP closure. Taking advantage of the Pax3^Cre/+ knock-in mouse, we examined Cdx2 expression in Pax3-null (Pax3^Cre/Cre) embryos that exhibit the known closure defects in the PNP. Significantly, Cdx2 mRNA is lost in the dorsal PNP of Pax3-null embryos at E9.5 (Fig. 8A,B), indicating that Cdx2 expression might also be regulated by Pax3. To address this possibility, we examined Cdx2 expression in β-catenin cKO;R26-Pax3-GOF compound embryos. We detected no or little Cdx2 mRNA signal in the dorsal PNP of the β-catenin cKO;R26-Pax3-GOF compound mutant (Fig. 8C), whereas we detected a mild expansion of the Cdx2 expression domain in the dorsal PNP of the Pax3^Cre;R26-Pax3-GOF mutant (Fig. 8D) at E9.5. These results suggest that the ectopically activated Pax3 is unable to induce Cdx2 expression in the absence of β-catenin in the dorsal PNP, and that the ectopically activated Pax3 can rescue the closure defects without Cdx2 activation in the β-catenin cKOs.

Taken together, this study suggests that β-catenin is required for caudal neural tube closure and elongation processes and may act through the transcriptional activation of key downstream target genes in the dorsal PNP as well as in the tail bud (Fig. 8E).

With mouse genetic and molecular biological approaches, this study has revealed a novel function of β-catenin in neurulation. Conditional ablation of β-catenin in the Pax3-expressing dorsal neural tube/fold causes defective PNP closure and elongation and leads to spina bifida and tail truncation. These defects are associated with the diminished expression of Pax3 and Cdx genes in the dorsal PNPs of β-catenin cKO embryos. Pax3 is expressed in the dorsal neuroepithelial cells along the entire anterior-posterior axis during the closure process (Goulding et al., 1991; Solloway and Robertson, 1999). A recent study suggests the association of PAX3 mutations with spina bifida in humans (Agopian et al., 2013), which was not found in early studies (Greene et al., 2009; Lu et al., 2007). Pax3 homozygous mutant mouse embryos exhibit both spina bifida and exencephaly (Engleka et al., 2005; Epstein et al., 1991), whereas the latter is not seen in the β-catenin cKO mutant. This difference might be related to the different mouse genetic approaches employed or to region-specific mechanisms. From the conditional gene-targeting experiment performed in this study, Pax3 is expressed in the closing neural tube of the mutants prior to β-catenin ablation by Pax3^Cre/+. Hence, we observed a dramatic reduction, but not complete absence, of Pax3 mRNAs in the dorsal PNPs of β-catenin cKOs at E9.5. The remaining Pax3 (expressed prior to β-catenin ablation) in the mutants is insufficient for PNP closure, but might be sufficient for cranial neural tube closure. Notably, when Pax3 protein is markedly reduced in heterozygous Pax3^Cre/+ embryos (Engleka et al., 2005), the single Pax3^Cre/+ or the double β-catenin^flox/+;Pax3^Cre/+ heterozygotes do not exhibit NTDs, and are viable and fertile. Moreover, a single copy of the Pax3 cDNA transgene induced by the same Cre is able to rescue the spina bifida in β-catenin^flox/flox;Pax3^Cre/+ mutants. These results demonstrate that Pax3 is a key mediator of β-catenin signaling in the PNP closure process.

The cellular and molecular mechanisms of Pax3 function in neural tube closure remain largely unknown. Elevated apoptosis is found in the defective neural tube of Pax3-deficient Splotch mice (Phelan et al., 1997). Inhibition of p53-dependent apoptosis can reduce the NTD occurrence rate in these mutants (Pani et al., 2002), suggesting an anti-apoptotic role of Pax3 during neural tube closure. On the other hand, apoptosis occurs during normal neurulation and many other embryogenesis processes. Inhibition of caspase activity in chick embryos can prevent neural tube closure (Weil et al., 1997). By contrast, genetic and chemical inhibition of the caspase pathway in mouse embryos demonstrates that apoptosis is not required for the neural tube closure process (Massa et al., 2009). We did not detect any obvious change in apoptosis in association with the diminished Pax3 expression in the mutant PNPs. However, future studies might determine whether inhibition of apoptosis can prevent NTDs in β-catenin cKO embryos. A chimeric mouse study has revealed that Pax3 acts cell-autonomously in the neural tube and may influence cell surface properties (which are as yet undefined) (Mansouri et al., 2001). An in vitro study has shown that Pax3 can induce cell aggregation and regulate phenotypic mesenchymal-epithelial interconversion (Wiggan et al., 2002). Pax3 is also expressed in the secondary neural tube (Shum et al., 2010) that forms by secondary neurulation in the caudal axis at ∼E10.5 (′31-somite stage) in mouse or at corresponding stages in chick embryos, just after primary neural tube closure has been completed (Cambray and Wilson, 2002; Gofflot et al., 1997; Shum and Copp, 1996). It is unclear whether β-catenin/Pax3 signaling is required for secondary neurulation and how this might contribute to spina bifida in the β-catenin cKO or related mutants. The cellular actions and underlying molecular mechanisms of β-catenin/Pax3 signaling in neural tube closure warrant further study.

The current study shows that Cdx2 expression is diminished in the dorsal PNPs of β-catenin cKOs or Pax3-null embryos. Significantly, ectopic activation of the Pax3 transgene can rescue NTDs, but not tail truncations, in the β-catenin cKOs without Cdx2 restoration in the dorsal PNPs. These results suggest that Cdx2 might be activated by the presence of both β-catenin and Pax3 at the closure site of the dorsal PNP, and that Cdx2 might play a less dominant role in PNP closure. Consistent with this, Cdx2^+/-;Cdx4^-/- mice exhibit truncated caudal axial skeletons, but no spina bifida (Young et al., 2009). Among Cdx genes, only Cdx1 and Cdx4 are known Wnt/β-catenin signaling targets (Béland et al., 2004; Lickert et al., 2000; Pilon et al., 2006). However, Cdx1-null and Cdx4-null mice are viable and fertile (Subramanian et al., 1995; van Nes et al., 2006). Intriguingly, Cdx2-null mice die at E3.5 (Chawengsaksophak et al., 1997), whereas Cdx2;Cdx1 compound mutants exhibit the severe NTD craniorachischisis with reduced Ptk7 expression (Savory et al., 2011). These results suggest that Cdx2 has functions distinct from those of other Cdx genes and is crucial for initiating neural tube closure and also for PNP elongation. Our results show that Cdx2 expression is regulated by β-catenin and possibly also by Pax3 during PNP development. Interestingly, a recent study reports that Pax3 expression may be regulated by Cdx and Wnt during neural tube development (Sanchez-Ferras et al., 2012). These opposing observations suggest that Pax3 and Cdx2 might be interactively regulated by each other under the transcriptional control of β-catenin for the PNP closure and elongation processes (Fig. 8E). Future studies might determine whether genetic activation of Cdx2 can rescue spina bifida in either Pax3 or β-catenin mutants. Our results also demonstrate a dorsally bent and shorter PNP or posterior axis in the β-catenin mutant at E10.5, which may be caused by defective Pax3 and Cdx in the dorsal PNP, with relatively conserved signaling gene expression in the ventral axis and tail bud at E9.5. However, it remains unclear whether, or how, β-catenin/Pax3/Cdx signaling integrates the dorsal PNP closure and elongation processes. We also observed that tail bud signaling is relatively unaffected at E9.5 but diminished at E10.5 in the β-catenin mutant embryos, which furthers posterior axis disruptions.

β-catenin has dual roles in Wnt signaling and cell adhesion (Valenta et al., 2012), and the latter has recently been demonstrated to be important in cell integration in the caudal body axis. Conditional ablation of β-catenin with Cdx1-Cre in all three germ layers of the posterior embryo leads to a progressive disintegration and loss of posterior structures before caudal neural tube closure (Hierholzer and Kemler, 2010). The posterior disintegration in these mutant embryos is linked with the abnormal localization of N-cadherin (Hierholzer and Kemler, 2010). A more severe disintegration phenotype is observed in the N-cadherin-null embryo, which exhibits an abnormally weaving neural tube and undetermined phenotypes in PNP closure (Radice et al., 1997). However, our preliminary observations indicate that conditional ablation of N-cadherin in Pax3^Cre knock-in mice (the same as used in this study) does not cause NTDs. These observations suggest that the cell adhesion function of β-catenin in the Pax3-expressing neuroepithelial cells might not be required for neural tube closure.

β-catenin-independent PCP signaling acts through cytoskeletal dynamics to control cell polarity and oriented tissue movements. Single or combined mutations in mouse PCP signaling genes, such as Celsr1 (Curtin et al., 2003), Dvl1/2/3 (Etheridge et al., 2008; Hamblet et al., 2002; Wang et al., 2006a), Fzd3/6 (Wang et al., 2006b), Ptk7 (Lu et al., 2004), Scrib (Murdoch et al., 2003) and Vangl2 (Kibar et al., 2001; Murdoch et al., 2001), will lead to the failure of initial neural tube closure and result in severe NTDs. The PCP-related Daam1-binding protein MIM is required for neural tube closure in Xenopus (Liu et al., 2011). Among these genes, Fzds and Dvls also act upstream of β-catenin in the canonical Wnt pathway. In addition, Lrp6, a Wnt co-receptor acting upstream of β-catenin and also involved in convergent extension, is required for neural tube closure and caudal axis elongation (Carter et al., 2005; Kokubu et al., 2004; Pinson et al., 2000; Tahinci et al., 2007; Zhou et al., 2010). Although we did not detect significant alterations in representative PCP signaling gene expression patterns in the conditional β-catenin mutants, future studies are needed to examine whether PCP signaling functions are indirectly altered in β-catenin signaling mutants or whether β-catenin signaling is altered and contributes to NTDs in the PCP signaling mutants, including mutants of Fzds and Dvls.

In summary, this study revealed a novel function and underlying mechanism of β-catenin signaling in the transcriptional modulation of Pax3 and Cdx2 in the caudal neural tube closure and elongation processes, which might provide a basis for a better understanding of the cause and prevention of spina bifida and related disorders.

The conditional β-catenin^(ex2-6)flox, the β-catenin signaling reporter BATgal and the Pax3^Cre knock-in mice were obtained through the Jackson Laboratory (Brault et al., 2001; Lang et al., 2005; Maretto et al., 2003). According to the Jackson Laboratory, these mouse strains were maintained on C57BL/6J or a mixed B6;129 background. The Cre-inducible Rosa26-loxP-stop-loxP-Pax3-cDNA (Pax3-GOF) mice obtained from the Epstein laboratory at the University of Pennsylvania were described previously (Wu et al., 2008). All research procedures using laboratory mice were approved by the UC Davis Animal Care and Use Committee and conform to NIH guidelines.

Embryos were fixed in 1% paraformaldehyde (PFA) for ∼30 minutes on ice and processed for X-gal staining as described previously (Song et al., 2009; Wang et al., 2011). Embryos fixed in 4% PFA overnight at 4°C were processed for wholemount in situ hybridization using digoxigenin-labeled antisense RNA probes (supplementary material Table S1). Wild-type (β-catenin^flox) or double-heterozygous (β-catenin^flox/+;Pax3^Cre/+) embryos were used for the normal controls, which showed no significant differences in X-gal staining or in situ mRNA signals. At least two mutants and two control embryos were used for each in situ experiment, which showed consistent results.

Total RNAs were isolated from the caudal neural folds and pooled from three to five E9.5 embryos. Heterozygous Pax3^Cre/+ embryos were used as controls. After reverse transcription, real-time PCR was carried out as described (Song et al., 2009). The mRNA levels of β-catenin or Axin2 were normalized to that of Gapdh to allow for comparisons among different experimental groups using the ΔCT method.

Acute BrdU labeling was performed by intraperitoneal injection of BrdU at 100 mg/kg body weight of pregnant mice 1 hour prior to sampling. Immunohistochemistry was carried out on sections using primary antibodies against BrdU (1:100; Dako), active caspase 3 (1:200; Promega) and Pax3 (1:10; Developmental Studies Hybridoma Bank).

The 1205 bp region from -1189 to +16 of the mouse Pax3 gene promoter contains three presumptive Tcf/Lef activation sites (supplementary material Fig. S1). This ‘wild-type’ regulatory region was amplified by PCR and cloned into the pGL2-basic luciferase reporter vector. The mutated promoter constructs were generated by deleting one or three Pax3 activation sites (Fig. 4A). The 685 bp promoter-containing region (from -394 to +291) of the mouse Cdx2 gene has one Tcf/Lef binding site that was also mutated for luciferase reporter assays (Fig. 4F). Similar to previously described studies (Song et al., 2009; Wang et al., 2011), L cells were transiently transfected with the wild-type or mutated promoter-driven luciferase reporters, together with the constitutively active β-catenin and wild-type Lef1 expression constructs. A control expression vector (pcDNA) was introduced into L cells to monitor the baseline luciferase reporter activity. Varying amounts of the dominant-negative Lef1 expression constructs or Wnt3a proteins were used for competitive repression or dose-dependent activation of the wild-type Pax3 promoters. Twenty-four hours after transfection, luciferase activities were measured using the Dual-Luciferase Assay Kit (Promega) as described previously.

Extracts were prepared from caudal tissue of five E9.5 wild-type mouse embryos. Chromatin extraction and immunoprecipitation were performed as previously described (Song et al., 2009; Wang et al., 2011). Anti-β-catenin antibody (Santa Cruz Biotech; sc-7199) was used to pull down the β-catenin-Tcf/Lef DNA-binding complex, and rabbit IgG (Invitrogen; 10500C) was used for the negative control. The activation sites were amplified with specific primers (supplementary material Table S1).

At least three controls and three mutant mice were used for each statistical evaluation. Significances were assessed by Student’s t-test or pairwise (one-way ANOVA) when appropriate. In all cases, P≤0.05 were considered significant.

We are grateful to Andrew J. Copp (University College London) for helpful comments and discussion of an early version of the manuscript and to the journal reviewers for providing constructive comments and suggestions.