The role of Lrp6-mediated Wnt/β-catenin signaling in the development and intervention of spinal neural tube defects in mice

The neural tube is the precursor structure of the brain and spinal cord. Failure of proper neural tube closure at different anatomical regions along the anterior–posterior body axis results in various types of severe neural tube defects (NTDs), such as spina bifida (spinal NTDs) and exencephaly (cranial NTDs), the former of which is the most common type of NTD in humans (Greene and Copp, 2014). Although NTDs affect more than 300,000 newborns worldwide annually, of which ∼3000 cases occur in the United States (Centers for Disease and Prevention (CDC), 2004; Zaganjor et al., 2016), the etiology of NTDs remains poorly understood due to its complexity (Copp et al., 2013; Wallingford et al., 2013). Folate supplementation may prevent a considerable portion of NTDs, but the majority of NTDs that occurred in developed countries with mandatory folic acid fortification are considered unpreventable by folate intake alone (Williams et al., 2015). Therefore, it is imperative to study fundamental mechanisms of normal and defective neural tube closure, which could lead to the development of novel prevention strategies for NTDs.

Mutant mice have been widely used for NTD mechanistic and preventative studies (Greene and Copp, 2005; Zohn, 2020). More than 200 genes have been linked with NTDs in mice (Harris and Juriloff, 2007, 2010), suggesting that the genetic basis of neural tube closure is highly complex. However, only a small number of genes, including several involved in Wnt signaling, have been associated with human NTDs (Allache et al., 2014; Au et al., 2010; De Marco et al., 2014).

Morphogenetic Wnt signaling plays vital roles in early embryonic development, including gastrulation and neurulation. Wnt ligands bind to various types of receptors and co-receptors to transduce signals via β-catenin-dependent (canonical) and β-catenin-independent (noncanonical) pathways. The noncanonical Wnt/planar cell polarity (PCP) signaling pathway regulates cytoskeleton dynamics and collective tissue movements, such as convergent extension, which may drive neural tube closure (Copp et al., 2003; Wallingford, 2006; Wang et al., 2019; Ybot-Gonzalez et al., 2007). The rare but severest type of NTDs, craniorachischisis, is characterized by an entirely open brain and spinal cord (Tobin et al., 2019), and results from failure of the initial neural tube closure at the boundary between the future hindbrain and spinal cord. Craniorachischisis is associated with defective convergent extension and has been mainly found in mice with mutant PCP signaling genes, such as Ptk7 and the Vangl, Celsr, Dvl and Fzd families (De Marco et al., 2011; Juriloff and Harris, 2012). Several of these PCP components, such as Fzds and Dvls, also play essential roles in the canonical Wnt/β-catenin signaling pathway (MacDonald et al., 2009).

The Wnt co-receptor Lrp6 acts upstream of β-catenin in the canonical pathway and is required for a wide range of processes pertaining to embryogenesis and organogenesis, including neural tube closure (Alrefaei and Abu-Elmagd, 2022; Carter et al., 2005; He et al., 2004; Kokubu et al., 2004; Mao et al., 2001; Pinson et al., 2000; Song et al., 2009, 2010; Tamai et al., 2000; Wang et al., 2016; Wehrli et al., 2000; Zhou et al., 2008, 2004, 2010). Multiple studies have identified mutations in human LRP6 gene from patients with NTDs (Allache et al., 2014; Lei et al., 2015; Shi et al., 2018). In mice, both hypermorphic and hypomorphic mutations in Lrp6 gene can cause NTDs. The cranial NTDs in the spontaneous mutant crooked tail mice (Carter et al., 2005) and spinal NTDs in novel N-ethyl-N-nitrosourea (ENU)-induced Skax26 mice (in combination with heterozygous Vangl2^Lp) (Allache et al., 2014) are caused by hypermorphic Lrp6. In contrast, spinal NTDs in the spontaneous ringelschwanz mice are caused by hypomorphic Lrp6 (Kokubu et al., 2004), and Lrp6-null mutants exhibit partially penetrant cranial NTDs and fully penetrant spinal NTDs (Pinson et al., 2000; Zhou et al., 2010). It has been reported that altered noncanonical Wnt signaling may contribute to NTDs in hypermorphic Lrp6 mutants (Allache et al., 2014; Gray et al., 2013). However, the causal mechanisms of NTDs in Lrp6-deficient mutants and the explicit role of Lrp6-mediated canonical Wnt/β-catenin signaling in neural tube closure are thus far obscure.

We have previously demonstrated that conditional ablation of β-catenin in the dorsal neural folds diminishes expression of the paired-box gene Pax3, which encodes a transcription factor critical to neural tube closure, resulting in an open spinal NTD phenotype (Zhao et al., 2014). The β-catenin-Tcf/Lef1 complex modulates activity of the Pax3 promoter, and ectopic activation of Pax3 transgene can rescue spinal NTDs in β-catenin mutants. Moreover, β-catenin and Pax3 cooperate to regulate the caudal type of homeobox gene Cdx2 for caudal body axis elongation (Zhao et al., 2014). In the current study, we demonstrate that conditional ablation of Lrp6 in dorsal neural folds causes fully penetrant spinal NTDs, as a potential consequence of diminished activities of canonical Wnt/β-catenin signaling and its downstream target gene Pax3, which are consistent with those occurring in conditional β-catenin mutants. Moreover, our genetic and pharmacological rescue approaches demonstrate the crucial roles of Lrp6-mediated Wnt/β-catenin signaling in the development of spinal NTDs.

To determine the cell lineage-specific role of Lrp6 in neural tube closure, we carried out conditional gene-targeting analyses by breeding Lrp6^flox mice (Zhou et al., 2010) with the Pax3^Cre knock-in mice (Lang et al., 2005). Genetic fate mapping of Pax3^Cre mice crossed with the Cre reporter Rosa26-lacZ revealed restrictive activities of the Cre recombinase in the dorsal neural folds, including in the recently closed dorsal midline region and the dorsal edge cells along the pending closure sites of the PNP of embryonic day (E)8.5 mouse embryos (Fig. 1A). Genetic fate mapping of Pax3^Cre at E9.5 PNP has been detailed in our previous study (Zhao et al., 2014). The neural tube closes completely from E10.5 onwards in the non-Cre homozygous Lrp6^flox/flox or in the double-heterozygous Pax3^Cre/+;Lrp6^flox/+ littermate control mice (Fig. 1B). However, the conditional knockout Pax3^Cre/+;Lrp6^flox/flox (abbreviated as Pax3Cre;Lrp6-cKO or Lrp6-cKO) mice exhibit fully penetrant tail truncations and spinal bifida, as shown by the consistently open PNP at E12.5 and open caudal spinal cord at E18.5 (Fig. 1C,D). These results indicate that Lrp6 is required in the Pax3-expressing dorsal edge cells for PNP closure, which is consistent with the role of β-catenin in the same lineage for PNP closure, as we previously demonstrated (Zhao et al., 2014).

To validate whether Lrp6-mediated Wnt/β-catenin signaling is disrupted in the mutant PNP, we incorporated two representative Wnt/β-catenin signaling reporter mouse lines, BATgal (Maretto et al., 2003) and TOPgal (DasGupta and Fuchs, 1999), to the conditional gene-targeting approaches. X-gal staining for the signaling reporter lacZ of either BATgal or TOPgal indicated a clear reduction of Wnt/β-catenin signaling activities in the dorsal PNP of Lrp6-cKO embryos at E9.5, compared to that of their triple heterozygous littermate controls (Fig. 1E-H). The signaling reporter shows intensive Wnt/β-catenin signaling activities in the notochord (Ukita et al., 2009), which was not altered in the mutants as Pax3^Cre is not expressed in the notochord. Because these signaling reporters may not visualize some less robust but functional Wnt/β-catenin signaling in vivo, we further examined the expression of Axin2, a general downstream target gene and negative feedback regulator of the canonical Wnt/β-catenin signaling pathway (Jho et al., 2002). Wholemount in situ hybridization showed a substantial reduction in Axin2 mRNA at E9.5 in the dorsal PNP of Lrp6-cKO embryos compared to that in the littermate control embryos (Fig. 1I,J). These results demonstrate diminished Wnt/β-catenin signaling in Lrp6-deficient dorsal neural folds.

To address the molecular mechanisms underlying spinal NTDs in Lrp6-cKO mutants, we examined a panel of NTD-associated genes that are specifically expressed in the dorsal neural folds, including the transcription factor Pax3 (Epstein et al., 1991; Goulding et al., 1991) and the caudal-type homeobox genes Cdx2 and Cdx4 (Young et al., 2009; van Nes et al., 2006). Wholemount in situ hybridization showed diminished expression of Pax3, Cdx2 and Cdx4 in the dorsal neural folds of Lrp6-cKO embryos at E9.5 (Fig. 2). We have previously demonstrated that both Pax3 and Cdx2 are downstream target genes of β-catenin signaling, and Cdx2 is additionally regulated by Pax3 during PNP closure (Zhao et al., 2014). The transcription factor Msx1 is a regulator of Pax3 (Monsoro-Burq et al., 2005), and it is also a known β-catenin downstream effector (Foerst-Potts and Sadler, 1997; Song et al., 2009). Msx1 was restrictively expressed in the dorsal midline during PNP closure, but it was absent in Lrp6-cKO PNPs at E9.5 (Fig. 3A,B). These results demonstrate the consistent roles of Lrp6 and β-catenin in regulation of these key downstream target genes during PNP closure.

We next examined whether conditional ablation of Lrp6 in the dorsal neural folds affects additional critical genes, such as T-box, Fgf and Wnt genes in the tailbud signaling center, which are required for caudal body axis formation (Gofflot et al., 1997) and may also play a role in caudal neural tube closure. Wholemount in situ hybridization demonstrated that the expression patterns of brachyury (T) (Wilkinson et al., 1990), Tbx6 (Chapman and Papaioannou, 1998), Wnt1 (Parr et al., 1993) and Wnt5a (Yamaguchi et al., 1999) were not apparently altered in Lrp6-cKO tailbuds, dorsal neural folds or related caudal tissues at E9.5 (Fig. 3C-J). The expression patterns of Fgf8, Fgf17 and Fgf18 (Maruoka et al., 1998) were also not altered in the mutant tailbuds or related tissues (Fig. 4A-F). Mesp2, modulated by Notch and Fgf signaling (Niwa et al., 2011), is restrictively expressed in the presomites (Takahashi et al., 2000) and remained unaffected in E9.5 Lrp6-cKOs (Fig. 4G,H). These results suggest that the tailbud signaling center and new somite formation are not altered in the Lrp6-cKOs and might not contribute to PNP closure.

To examine whether Lrp6 deficiency in the dorsal neural folds affects noncanonical Wnt/PCP signaling, we first examined two representative genes, Vangl2 (Kibar et al., 2001) and Ptk7 (Lu et al., 2004), which are required for PCP signaling and neural tube closure. The expression of these genes was unaltered in the dorsal neural folds of Lrp6-cKO mutants at E9.5 (Fig. 5A-D). Dvl2 is a mediator of canonical and noncanonical Wnt signaling pathways and is required for neural tube closure (Hamblet et al., 2002). Dvl2 phosphorylation does not trigger β-catenin signaling and can be detected by band shifts of immunoblots (Gonzalez-Sancho et al., 2004). We detected no changes in Dvl2 phosphorylation in PNP samples of Lrp6-cKO mutants (Fig. 5E). In addition, proliferation, as shown by bromodeoxyuridine (BrdU) incorporation assays, and apoptosis, as shown by terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assays, were unaffected in the dorsal neural folds of Lrp6-cKO mutants (Fig. 6). These results suggest that conditional inactivation of Lrp6 in dorsal neural folds may not modify PCP signaling, proliferation or apoptosis during caudal neural tube closure.

To verify that Lrp6 acts through β-catenin to regulate PNP closure, we tested genetic rescue experiments by gain-of-function (GOF) of β-catenin in Lrp6-cKO mutants. Ctnnb1^flox(ex3) mice were used, in which the conditional removal of the floxed exon 3 (that encodes a phosphorylation site for β-catenin protein degradation) will stabilize β-catenin, leading to constitutive activation of canonical Wnt signaling (Harada et al., 1999). The results indicated rescued PNP closure in the compound mutants of Pax3Cre;Lrp6-cKO;β-catenin-GOF mutants (five out five embryos) at E10.5 (Fig. 7). However, the dorsal midline region and roof plate were markedly widened in the mutants, suggesting a side effect of constitutive activation of β-catenin signaling in the dorsal neural folds.

The finding that defective Wnt/β-catenin signaling pathway is essentially related to spinal NTDs provides a significant translational implication for NTD intervention. We therefore attempted to treat spinal NTDs in Lrp6-cKOs through maternal supplementation of lithium ion, a commonly used psychotherapeutic medication and well-known Wnt signaling agonist acting through inhibition of intracellular β-catenin degradation (Cohen and Goedert, 2004; Meijer et al., 2004). After intraperitoneal injections of lithium chloride (LiCl) solution to the pregnant females at E7.5-E9.5, ∼18% (5/28) of the Lrp6-cKO mice had completely closed neural tubes and elongated tails, as determined at E18.5, which were not observed in the control group of Lrp6-cKOs treated with the NaCl placebo (0/30 Lrp6-cKOs) (Fig. 8A-C). Conversely, 20% (6/30) of the Lrp6-cKOs in the control group exhibited severe closure defects up to the lumbar level, which were not observed in the LiCl-treated Lrp6-cKOs (0/28). These results demonstrate effective intervention of spinal NTDs in Lrp6-cKOs by lithium treatment. Lesions of variable lengths at the sacrococcygeal levels were also observed in both the LiCl-treated and placebo groups of Lrp6-cKOs, suggesting that environmental or other factors may affect canonical Wnt signaling activity and NTD severity. After lithium treatment, Pax3 expression in the neural folds of Lrp6-cKOs was restored (Fig. 8D).

Together, these results elucidate the essential roles of the Lrp6-mediated canonical Wnt/β-catenin signaling pathway in murine caudal neural tube closure, which may provide significant clues to both the etiology and prevention of human NTDs.

This study reveals the essential role of Lrp6-mediated Wnt/β-catenin signaling in caudal neural tube closure. Lrp6 is a Wnt co-receptor acting upstream of β-catenin in the canonical pathway. We found fully penetrant spinal NTDs in Pax3Cre;Lrp6-cKO mutants, which is phenotypically consistent with Pax3Cre;β-catenin-cKO mutants, as demonstrated in our previous study (Zhao et al., 2014), suggesting a consistent role of Lrp6 and β-catenin in caudal neural tube closure. At the molecular signaling level, we observed diminished expression of several genes encoding transcription factors, including Pax3, Cdx2, Cdx4 and Msx1, in the dorsal neural folds of Lrp6-cKO mutants, which were identically downregulated in the β-catenin-cKO neural folds, further suggesting that Lrp6 and β-catenin act in the same signaling cascade to regulate critical downstream target genes during caudal neural tube closure. Our genetic rescue experiments revealed that constitutively active β-catenin in the dorsal neural folds can rescue closure defects in Lrp6-cKO embryos, which firmly validates that Lrp6 acts through β-catenin to promote caudal neural tube closure. Moreover, maternal supplementation of an agonist of the Wnt/β-catenin signaling could reduce the frequency and severity of NTDs in Lrp6-cKO mutants, further substantiating the essential roles of Lrp6-mediated Wnt/β-catenin signaling in caudal neural tube closure.

We employed Pax3^Cre knock-in mice for conditional gene-targeting analyses of both Lrp6 and β-catenin, which generated consistent phenotypic and mechanistic results. Our genetic fate-mapping experiments demonstrated that Pax3^Cre became activated in the border region between the non-neural surface ectoderm and neural plate in the pending closure sites of dorsal neural folds at E8.5. It has been demonstrated that conditional removal of β-catenin using Grhl3^Cre in these border cells also causes spinal NTDs (Kimura-Yoshida et al., 2015). However, it remains unclear whether Pax3^Cre and Grhl3^Cre activities overlap in the same border cells during primary neurulation. We recently demonstrated that Grhl3 regulates dynamic movements of the non-neural surface ectodermal border cells for multicellular rosette formation and convergent cellular protrusions to promote caudal neural tube closure (Zhou et al., 2020). Conditional gene-targeting analyses of Lrp6 using Grhl3^Cre may clarify whether Lrp6-mediated Wnt/β-catenin signaling is also activated in these border cells as a modulator of their molecular properties and cellular dynamics for neural tube closure.

Lrp6 may exert a key role in the signaling interactions between canonical and noncanonical Wnt pathways. Loss-of-function (LOF) of Lrp6-mediated Wnt/β-catenin signaling may affect noncanonical Wnt/PCP signaling directly or indirectly. Our results showed that Lrp6 deficiency did not change the expression patterns of two representative NTD-associated PCP signaling genes, Vangl2 and Ptk7, in the mutant PNP, suggesting that Lrp6-mediated Wnt/β-catenin signaling does not directly regulate transcriptional activation of these PCP signaling genes. We also showed that Lrp6 deficiency did not change the Dvl2 phosphorylation level. Because Dvl2 phosphorylation is independent of Lrp5/Lrp6 co-receptors and does not stabilize β-catenin (Gonzalez-Sancho et al., 2004), it is an excellent indicator of altered noncanonical Wnt signaling. It has been reported that different Wnts may signal through different co-receptors, Lrp6 for canonical signaling and Ror1/Ror2 for noncanonical signaling, and that canonical and noncanonical Wnts may exert reciprocal inhibition by competition for Fzd binding (Grumolato et al., 2010). Interestingly, an early study demonstrated that either LOF or GOF of Lrp6 disrupted the convergent extension that is a hallmark of PCP signaling during Xenopus gastrulation (Tahinci et al., 2007). The authors showed that an intracellular domain of Lrp6 can inhibit Wnt/PCP signaling by unknown mechanisms, while potentiating Wnt/β-catenin signaling. Thus, Lrp6 may serve as a molecular switch from noncanonical to canonical Wnt signaling (Tahinci et al., 2007). Nevertheless, it remains unknown whether or how the noncanonical inhibitory function of this intracellular domain of Lrp6 contributes to neural tube closure. Another study reported that the extracellular domains of Lrp6 (or its homolog Lrp5) can also inhibit noncanonical Wnt signaling in vitro (Bryja et al., 2009). These authors further showed that Wnt5a deficiencies partially rescued heart defects and fully rescued the partially penetrant cranial NTD exencephaly in Lrp6-null mutants, suggesting that GOF of noncanonical Wnt signaling is the cause of these defects in Lrp6-null mutants (Bryja et al., 2009). However, Wnt5a deficiencies had no rescue effects on the fully penetrant spinal NTDs of Lrp6-null mutants, suggesting distinctly different mechanisms underlying cranial and spinal NTDs in Lrp6-deficient mutants. Above all, it will be important to determine whether upregulation of the Wnt/PCP signaling activities can indeed cause NTDs.

In contrast, the hypermorphic Lrp6^{crooked tail} homozygous embryos exhibit cranial NTDs with defective RhoA signaling and apical–basal cell polarity, suggesting a defective noncanonical Wnt signaling mechanism (Carter et al., 2005). The compound mutants of another hypermorphic Lrp6^Skax2 homozygous combined with heterozygous Vangl2^Lp exhibit spinal NTDs (Allache et al., 2014), reiterating a defective noncanonical Wnt signaling mechanism in these hypermorphic Lrp6 mutants. Nevertheless, it remains unknown whether or how noncanonical Wnt signaling is altered and may contribute to NTDs in Lrp6-deficient mutants and its contribution to NTD incidence. Mutants deficient in the PCP signaling gene Ptk7 exhibit the severest NTD, craniorachischisis (Lu et al., 2004). Unexpectedly, a later study demonstrated that Ptk7 modulates both canonical and noncanonical Wnt signaling through physical interaction and stabilization of Lrp6 proteins (Bin-Nun et al., 2014). Therefore, Ptk7 or related PCP signaling may actually act upstream of Lrp6-mediated Wnt/β-catenin signaling, suggesting complex signaling crosstalk between the canonical and noncanonical Wnt pathways during neural tube closure.

The current study showed that maternal supplementation of LiCl during a relatively short period (from E7.5 to E9.5) can reduce the frequency and severity of spinal NTDs, with ∼18% of Lrp6-cKO mutants exhibiting completely closed neural tubes and an additional 20% with reduced NTD lesion sizes in the rescue group. Lithium inhibits Gsk3 enzymic activities (Klein and Melton, 1996) to stabilize β-catenin in the canonical Wnt signaling pathway (Liu et al., 2002). Thus, the results of our NTD rescue experiments by lithium support the role of Lrp6-mediated Wnt/β-catenin signaling in PNP closure and may also provide a translational implication for NTD intervention using small-molecule agonists of Wnt/β-catenin signaling (Fig. 8E). There are several other Gsk3 inhibitors that can be used for therapeutic treatments (Cohen and Goedert, 2004; Meijer et al., 2004), including the small-molecule CHIR99021 and 6-bromoindirubin-3′-oxime (BIO), which have been used widely as potent Wnt agonists (Sato et al., 2004; Silva et al., 2008). It is important to note that Gsk3 inhibitors are not specific to Wnt signaling and have broad functions, such as in insulin signaling, NFAT signaling and Hedgehog signaling (Cline et al., 2002; Crabtree and Olson, 2002; Price and Kalderon, 2002; Ring et al., 2003). Regardless of these broad effects, our results showed that lithium supplementation reduced the frequency and severity of NTDs in Lrp6-deficient mutants by restoring the expression of Pax3, which is a key downstream target gene of Wnt/β-catenin signaling. As demonstrated in our previous work, genetic activation of Pax3 can rescue spinal NTDs in the conditional β-catenin mutants (Zhao et al., 2014). Intriguingly, p53 (also known as Trp53) LOF by genetic or pharmacological approaches can rescue NTDs in Pax3 mutants, suggesting a role of apoptosis in the development of NTDs (Pani et al., 2002). We did not detect significantly altered apoptosis in Lrp6-deficient mutants, but it would be important to test whether p53 LOF can rescue NTDs in Lrp6-deficient mutants as well. Conversely, severe midline defects, including cranial and spinal NTDs and midline facial clefts, occurred in mice with triple knockout of the intrinsic apoptotic genes Bax, Bak (also known as Bak1) and Bok (Ke et al., 2018), demonstrating that developmental apoptosis is required for normal neural tube closure and related midline fusion processes. Thus, completely inhibiting apoptosis seems inappropriate for preventing NTDs.

In summary, this study demonstrated that conditional ablation of Lrp6 in dorsal neural folds resulted in spinal NTDs due to diminished canonical Wnt/β-catenin signaling and downstream target genes. This could be rescued either by genetic activation or pharmacological stabilization of β-catenin in vivo (Fig. 8E), indicating LRP6-mediated Wnt/β-catenin signaling as a novel target for intervention for NTDs in humans.

The Lrp6^flox mice for conditional gene-targeting analyses have previously been described (Zhou et al., 2010). The Pax3^Cre knock-in mice (Engleka et al., 2005), Rosa26-lacZ mice (Soriano, 1999), and canonical Wnt/β-catenin signaling reporter lines of BATgal (Maretto et al., 2003) and TOPgal (DasGupta and Fuchs, 1999) mice donated by different investigators were obtained through The Jackson Laboratory. Ctnnb1^flox(ex3) mice (MGI:1858008, gift from M. Taketo, Kyoto University, Kyoto, Japan) (Harada et al., 1999) were used for genetic rescue by conditional activation of β-catenin. These mouse strains were maintained on a C57BL/6J or a mixed B6;129 background. All mice were housed in the vivarium at the University of California, Davis (Davis, CA, USA). Pregnant, timed-mated mice were euthanized prior to cesarean section. Noon of the conception day was designated as E0.5. All research procedures using mice were approved by the University of California, Davis Animal Care and Use Committee and conformed to National Institutes of Health guidelines.

Mouse embryos were fixed in 1% paraformaldehyde (PFA) for ∼30 min on ice and processed for X-gal staining as previously described (Song et al., 2009; Wang et al., 2011; Zhao et al., 2014). Embryos fixed in 4% PFA overnight at 4°C were processed for wholemount in situ hybridization using digoxigenin-labeled antisense RNA probes as previously described (Zhao et al., 2014). Antisense RNA probes were synthesized based on sequence information provided by the Allen Brain Atlas (https://portal.brain-map.org/). At least three mutants and three littermate control (Cre-lacking or double heterozygous) embryos were used for each in situ experiment, which showed consistent results.

In vivo stimulation of the Wnt/β-catenin signaling pathway by LiCl was carried out as previously described with minor modifications (Song et al., 2009; Tian et al., 2010). Pregnant females were injected intraperitoneally with 200 mg/kg LiCl or an equivalent dose of a NaCl control solution on E7.5, E8.5 and E9.5. Embryos were collected at E18.5 for phenotype analysis or at E9.5 (1 h after the third-day injection) for real-time reverse transcription quantitative PCR (RT-qPCR) analyses.

Total RNAs were isolated from the caudal neural folds and pooled from five E9.5 embryos of the mutant or control groups. Heterozygous Pax3^Cre/+ embryos were used as the control. After reverse-transcription, real-time PCR was carried as described in our previous publications (Song et al., 2009; Zhao et al., 2014). The mRNA level of Pax3 was normalized to the mRNA level of glyceraldehyde-3-phosphate dehydrogenase (Gapdh) to allow for relative comparisons among different experimental groups using the ΔΔC_t method.

Acute BrdU labeling was performed by intraperitoneal injection of BrdU at 100 mg/kg body weight to the pregnant mice 1 h prior to sampling. Immunohistochemistry was carried out on paraffin or frozen sections using primary antibodies against BrdU (1:100; M0744, Dako) and Alexa Fluor-conjugated secondary antibodies (1:400; Molecular Probes). TUNEL assays were performed using the Dead End Fluorometric TUNEL System (Promega) as detailed by the manufacturer. The total numbers of BrdU- or TUNEL-positive cells were counted in the dorsal PNP above the dorsolateral hinge points on each confocal micrograph (Fig. 6) then divided by the total cell numbers on the same areas to obtain the percentages of the positive cells. In each experiment, at least three sections from three mutants or three littermate controls at the same age were counted and averaged for statistical analyses.

Immunoblotting was carried out according to the standard protocol (Wang et al., 2016). Protein samples were pooled from the caudal part of five embryos of E9.5 control or Pax3Cre;Lrp6-cKO embryos. Primary antibodies against Dvl2 (1:200; sc-13974, Santa Cruz Biotechnology) and β-actin (1:5000; sc-1616, Santa Cruz Biotechnology) were used.

At least three littermate controls and three mutant embryos were used for each statistical evaluation. Significances were assessed by unpaired, two-tailed Student's t-test or pairwise comparison (one-way ANOVA) when appropriate. In all cases, P≤0.05 was considered statistically significant.

We are grateful to the former and current Zhou laboratory members who provided technical assistance or helpful discussions.