Both ciliary and non-ciliary functions of Foxj1a confer Wnt/β-catenin signaling in zebrafish left-right patterning

ABSTRACT The Wnt/β-catenin pathway is implicated in left-right (LR) axis determination; however, the underlying mechanism remains elusive. Prompted by our recent discovery that Wnt signaling regulates ciliogenesis in the zebrafish Kupffer's vesicle (KV) via Foxj1a, a ciliogenic transcription factor, we decided to elucidate functions of Foxj1a in Wnt-regulated LR pattern formation. We showed that targeted injection of wnt8a mRNA into a single cell at the 128-cell stage is sufficient to induce ectopic foxj1a expression and ectopic cilia. By interrogating the transcription circuit of foxj1a regulation, we found that both Lef1 and Tcf7 bind to a consensus element in the foxj1a promoter region. Depletion of Lef1 and Tcf7 inhibits foxj1a transcription in the dorsal forerunner cells, downregulates cilia length and number in KV, and randomizes LR asymmetry. Targeted overexpression of a constitutively active form of Lef1 also induced an ectopic protrusion that contains ectopic transcripts for sox17, foxj1a, and charon, and ectopic monocilia. Further genetic studies using this ectopic expression platform revealed two distinct functions of Foxj1a; mediating Wnt-governed monocilia length elongation as well as charon transcription. The novel Foxj1a-charon regulation is conserved in KV, and importantly, it is independent of the canonical role of Foxj1a in the biosynthesis of motile cilia. Together with the known function of motile cilia movement in generating asymmetric expression of charon, our data put forward a hypothesis that Foxj1a confers both ciliary and non-ciliary functions of Wnt signaling, which converge on charon to regulate LR pattern formation.


INTRODUCTION
Vertebrates display a distinct left-right (LR) asymmetry in the disposition of their internal organs (Raya and Belmonte, 2006). The specification of LR axis can be divided into at least two stages: (1) the generation of asymmetric signals around the mouse embryonic node or its derivatives, such as the zebrafish Kupffer's vesicle (KV), and (2) transferring of the asymmetric cues to the lateral plate mesoderm (LPM) and internal organs (Nonaka et al., 2002;Essner et al., 2005;Kramer-Zucker et al., 2005;Hirokawa et al., 2006). Motile cilia have been shown to play a crucial role in establishing LR asymmetry at the first stage. Motile cilia occur as a single cilium extending from cells in the node or KV, where coordinated cilia beating generates a leftward fluid flow that is essential for generating asymmetric signals around the node, such as left-side downregulation of Cerberus/Dan family members Coco (Xenopus), Cerl2 (mouse), and charon (medaka and zebrafish) (Hojo et al., 2007;Oki et al., 2009;Schneider et al., 2010;Schweickert et al., 2010;Nakamura et al., 2012). The asymmetric expression of these Nodal antagonists promotes Nodal (Spaw in zebrafish) activity on the left side of the node, which is then transferred and propagated to the left LPM (Kawasumi et al., 2011).
The Wnt/β-catenin pathway has been shown to play a role in regulating LR pattern formation. Wnt activation by KV-specific overexpression of stabilized β-catenin or KV-specific depletion of Axin, a Wnt/β-catenin antagonist, results in randomized sidespecific gene expression (Schneider et al., 2008), whereas global Wnt activation at levels not causing severe embryo malformation affects the competence of heart field and gives rise to no-looping heart without appreciably altering asymmetric gene expression in LPM (Carl et al., 2007;Lin and Xu, 2009). In contrast, loss of function of Wnt leads to randomized side-specific gene expression and randomized organ laterality as noted in mouse Wnt3a mutant, as well as zebrafish wnt3a, wnt8a, β-catenin and fzd10 morphants (Nakaya et al., 2005;Lin and Xu, 2009;Caron et al., 2012;Zhang et al., 2012).
At the zebrafish LR organ KV, we and others showed that inhibition of Wnt signaling results in shorter and fewer cilia, disordered fluid flow, downregulation of charon, and reduced cell proliferation (Lin and Xu, 2009;Caron et al., 2012;Zhang et al., 2012). Further investigation suggests that Wnt/β-catenin signaling regulates ciliogenesis via transcriptional control of foxj1a (Caron et al., 2012), a forkhead domain-containing transcription factor that is necessary for ciliogenesis in multiciliated cells of the mouse airway epithelial cells and monocilia biosynthesis in the zebrafish KV and Xenopus gastrocoel roof plate (GRP, frog equivalent of mouse node) (Chen et al., 1998;Brody et al., 2000;Stubbs et al., 2008;Yu et al., 2008). Consistent with Wnt-foxj1 regulation, a recent study in Xenopus reported expansion of foxj1 expression domain in the GRP by ectopic expression of β-catenin (Walentek et al., 2012). However, the Wnt-Foxj1-ciliogenesis-LR asymmetry hypothesis is not completely compatible with observations in the mouse. It has been shown that Wnt3a deficiency is associated with lack of coexpression of mechanosensing proteins PC1 and PC2 in the cilium without affecting cilium structure and motility in the node (Nakaya et al., 2005). While Foxj1 is expressed in the mouse node and deletion of the gene results in randomized LR asymmetry as Foxj1a does in zebrafish, nodal cilia are present in the Foxj1 knockout mice (Chen et al., 1998;Brody et al., 2000;Stubbs et al., 2008;Yu et al., 2008). Together, these inconsistencies suggest other, unrecognized functions of Foxj1 in LR pattern formation, prompting the present study to further interrogate functions of the Wnt-Foxj1 signaling axis in LR patterning.
Here, we present biochemical and genetic evidence to indicate that Wnt signaling directly regulates foxj1a transcription in KV through cooperative action of Lef1 and Tcf7. Using a targeted overexpression platform, i.e. injection of mRNAs into a single cell at the 128-cell stage (Agathon et al., 2003), we showed that Wnt activation induces ectopic foxj1a expression and ectopic cilia formation, possibly secondary to ectopic KV development. We revealed two distinct roles of Foxj1a in conferring Wnt-governed LR patterning. While Wnt controls cilia outgrowth via the canonical role of Foxj1a in ciliogenesis, it regulates charon expression via a novel non-ciliary function of Foxj1a.

RESULTS
Wnt activation promotes foxj1a transcription and induces ectopic foxj1a and ectopic cilia Given that Wnt/β-catenin signaling is required for foxj1a expression and ciliogenesis (Caron et al., 2012), we set out to test the effect of gain-of-Wnt-function. Our previous studies showed a transient activation of foxj1a in the zebrafish dorsal forerunner cells (DFCs) by inducible expression of β-catenin1, although steady-state expression of foxj1a was not altered by overexpression of Wnt3a, Wnt8a, and β-catenin1 (Caron et al., 2012). To validate the transient activation, we used an inducible transgenic Tg(hsp:wnt3a-GFP) strain. The foxj1a transcript level was increased at 1 h after Wnt3a induction (Fig. 1A,B), but returned to a level comparable to that of wild-type embryos at 4 h after heat shock (Fig. 1C,D). In contrast to causing transient upregulation of foxj1a in DFCs, Wnt3a and β-catenin1 activation was found to continuously increase foxj1a expression in the developing pronephros, another tissue in which Wnt is required for foxj1a expression (Fig. S1A-D; data not shown) (Caron et al., 2012).
To seek additional evidence to support a positive regulation of Wnt signaling on foxj1a transcription, we adopted a targeted injection strategy, i.e. injection of mRNAs into a single cell at the 128-cell stage, which allows important developmental pathways to be strongly activated without serious disruption of general development ( Fig. 1E) (Agathon et al., 2003). Consistent with our previous report, a tailbud-like protrusion was induced in 20.3% (48/241) of wnt8a mRNA-injected embryos (Fig. 1F) (Lin et al., 2007). Importantly, all protrusions that were examined expressed ectopic foxj1a (Fig. 1G). In addition, they contained ectopic cilialike structure as revealed by acetylated α-tubulin immunostaining (Fig. 1H). Together, these data suggest that Wnt activation induces foxj1a expression and ectopic cilia formation.
To validate the role of Lef1 in mediating Wnt in ciliogenesis, we overexpressed Lef1 in embryos depleted of Fzd10, a proven receptor for Wnt/β-catenin signaling (Caron et al., 2012). Because Lef1 lacks transactivation activity, a constitutively active form of Lef1 (CALef1) was generated. Injection of CAlef1 mRNA resulted in expanded expression of known Wnt targets, such as chordin and ntla ( Fig. S4A-D), proving the functionality of the fusion protein (Lin et al., 2007;Martin and Kimelman, 2008). DFC-targeted injection of CAlef1 mRNA rescued foxj1a levels, significantly restored cilia length (from 2.86±0.6 μm to 4.16±0.7 μm, P=1.02×10 -5 ), and less significantly increased cilia number (from 22±8 to 30±12, P=0.02) in DFC fzd10 MO embryos ( Fig. S4E-G,I-K, M,N). CALef1 overexpression also restored left-sided spaw expression from 45% in DFC fzd10 MO embryos to 72%, a level comparable to that of CAlef1 mRNA injection alone (Fig. S4O). Consistent with other methods to activate Wnt signaling, injection of CAlef1 mRNA alone into DFCs did not significantly affect steady state foxj1a expression and ciliogenesis (Fig. S4H,L). These results justified the use of CALef1overexpression for our further gain-of-Wnt-function studies. a 0.6 kb foxj1a regulatory sequence. This sequence is located approximately −5.2 kb to −4.6 kb upstream of the ATG start codon. D1, D2, and D3 represent putative Lef1/Tcf binding sites. (B) Lef1 interacts with the foxj1a enhancer by electrophoretic mobility shift assay. Lef1-GST fusion protein (100 ng) or glutathione S-transferase (GST) alone (100 ng) was incubated with a 32 P-labeled 0.6-kb foxj1a regulatory sequence in the presence or absence of 50-fold excess of cold probe. The resulting products were loaded onto a 4% acrylamide non-denaturing gel. (C) Lef1 predominantly binds to the D2 site. A Lef1-GST fusion protein (100 ng) was incubated with 32 P-labeled oligonucleotides corresponding to D1, D2, D3, and mutated D2 (D2m) sites in the presence or absence of 100-fold excess of unlabeled oligonucleotide. The reaction mixtures were loaded into a 6% acrylamide nondenaturing gel. TCR indicates T-cell antigen receptor enhancer AAGTTTC motif, serving as a positive control for Lef1 binding.
Targeted overexpression of Lef1 induces ectopic DFC lineage, KV-like structure, foxj1a expression, and cilia formation Injection of CAlef1 mRNA at the 1-cell stage resulted in ectopic foxj1a expression in a region near DFCs (18 of 146 injected embryos; Fig. 4B), while injection of CAlef1 mRNA into a single cell at the 128-cell stage induced more profound tailbud-like protrusions in 25% (72/288) of injected embryos (Figs 4D and 5A). Compared with Wnt8a-induced ectopic structures, these protrusions are more visually identifiable, facilitating later experiments. Consistent with Wnt8a overexpression, CALef1-induced protrusions expressed ectopic foxj1a (Fig. 4D), and contained ectopic monocilia as assessed by acetylated α-tubulin antibody staining (Fig. 4E,F) and costaining with antibodies against basal body marker γ-tubulin (Fig. 4G) or another cilium marker, Arl13B (Fig. S5A). Transmission electron microscopy analysis further revealed tubular structure of a cilium (Fig. S5B). The ectopic cilia have an average length of 4.42±0.31 µm, which is similar to that of KV cilia and much longer than that of primary cilia (∼1 μm) (Yuan et al., 2012). In addition, genes exclusive to motile cilia, such as dnah9 (dynein axonemal heavy polypeptide 9), efhc1 (EF handcontaining protein 1), and wdr78 (WD repeat domain 78) were detected in the protrusions ( Fig. 4H; data not shown) (Hellman et al., 2010), suggesting that ectopic cilia induced by CALef1 are motile. Together, these data confirmed that Wnt activation is sufficient for foxj1a induction and ectopic cilia formation.
The Wnt/β-catenin pathway plays pleiotropic roles during embryonic development, including cell fate specification, cell type differentiation, stem cell maintenance, cilia regulation, and tissue morphogenesis (van Amerongen and Nusse, 2009;Caron et al., 2012;Clevers and Nusse, 2012;Zhang et al., 2012). The presence of monocilia similar to KV cilia in length prompted us to test the hypothesis that CALef1 is sufficient to induce the formation of KV-like cells in ectopic protrusions. Indeed, we detected the expression of sox17, a DFC lineage marker, and charon, a Nodal antagonist that is expressed only in the cells lining the KV in teleosts and is required for LR patterning, in CALef1-induced protrusions (Fig. 5C,D) (Alexander et al., 1999;Hashimoto et al., 2004). Additionally, we detected tight junction marker ZO-1 staining in the cilia-forming area of the protrusions (Fig. 5B), suggesting the presence of differentiated epithelial cells. Ectopic expression of sox17, charon, and ZO-1 was also observed after wnt8a mRNA injection (data not shown). In contrast, expression of markers for other motile cilia-containing tissues, such as pax2a and cdh17 for pronephros and nkx2.2b and ntn1b for floor plate, was not observed in the protrusions (data not shown) (Norton et al., 2005;Wingert et al., 2007). Thus, targeted Wnt activation is capable of specifying progenitor cells to DFC lineage and differentiating them into foxj1aexpressing, cilia-forming, and charon-expressing KV-like cells.

Foxj1a confers Wnt-governed cilia formation and charon expression
The robust CALef1-induced protrusion provides an experimental platform to dissect which aspects of Wnt functions are conferred by Foxj1a. Injection of foxj1a mRNA into a single cell at the 128-cell stage failed to induce ectopic protrusions, sox17 expression, and charon expression despite its ability to elicit cilia formation ( Fig. S6; data not shown) (Yu et al., 2008), suggesting that not all functions of Wnt/β-catenin signaling are conferred by Foxj1a. In embryos coinjected with CAlef1 mRNA and foxj1a MO, knockdown of Foxj1a had no effect on the CALef1-induced ectopic protrusion (Fig. 5F), ectopic sox17 expression (Fig. 5H), and ectopic ZO-1 expression (Fig. 5G). We also checked tailbud markers, such as ntla, that can be observed in Wnt-induced protrusions (Lin et al., 2007). Consistently, CALef1 activated ectopic ntla expression (Fig. 5E), which was not affected by Foxj1a knockdown (Fig. 5J). In contrast, we found that ectopic cilia were missing or shorter (1.37±0.26 µm, Fig. 5G) and ectopic charon expression was abolished on Foxj1a knockdown (Fig. 5I).
The abolished charon expression led us to speculate that Foxj1a might confer functions of the Wnt/β-catenin pathway in regulating charon expression in KV (Lin and Xu, 2009;Zhang et al., 2012). During embryogenesis, charon transcription is initiated at the 6-somite stage when foxj1a transcript levels in KV cells are already downregulated and barely detectable (Hashimoto et al., 2004;Caron et al., 2012). The expression profile also supports that Foxj1a might function upstream of charon. Consistently, injection of foxj1a MO at the 1-cell stage downregulated charon transcript levels in KV (Fig. 6B). To evaluate the role of Foxj1a in Wnt-charon regulation, we performed foxj1a rescue experiments. While injection of foxj1a mRNA alone did not significantly alter charon transcription (Fig. 6H), it partially rescued charon levels in DFC fzd10 MO embryos (Fig. 6C,D) and Dkk1-expressing embryos (Fig. 6F,G). The epistatic analysis strongly suggests that Foxj1a mediates Wntregulated charon expression.

The role of Foxj1a in charon expression is independent of its ciliogenic function
Members of Cerberus/Dan family Nodal antagonists, including charon in teleosts, are the first molecules exhibiting asymmetric expression around the node. The interplay between these Nodal inhibitors and Nodal has been shown to provide the signals that lead to the establishment of laterality (Hojo et al., 2007;Oki et al., 2009;Schneider et al., 2010;Schweickert et al., 2010;Kawasumi et al., 2011). While their asymmetric expression is thought to be generated by directional cilia-driven fluid flow and further enhanced by asymmetric posttranscriptional decay of mRNA, the transcription of charon seems not dependent on cilia motility (Gourronc et al., 2007;Nakamura et al., 2012). To clarify the relationship among Foxj1a, cilia motility, and charon transcription, we inspected the following mutants with defective motile cilia: oval (ift88 tz288b/tz288b ) whose KV cilia are shorter or missing and cilia beating is barely detectable, lok (ccdc40 to237b/to237b ) that have reduced cilia length and motility, and ntla b195/b195 that harbor severely shorter or missing cilia in KV (Kramer-Zucker et al., 2005;Amack et al., 2007;Sullivan-Brown et al., 2008;Becker-Heck et al., 2011). oval and lok mutants exhibited normal foxj1a expression and, concordantly, normal charon transcript level (Fig. S7B,D,F,H). In contrast, ntla mutants exhibited loss of foxj1a and, concordantly, loss of charon (Fig. S7J,N). The lack of foxj1a and charon expression was not due to absence of DFCs because sox17 was still expressed (Fig. S7L). Importantly, ectopic expression of Foxj1a rescued charon levels in ntla mutants (Fig. S7O). Taken together, these correlated data support that Foxj1a determines charon transcription while cilia beating initiates its asymmetric expression around KV.

DISCUSSION
As described in this article, we revealed the transcription circuit of foxj1a regulation and validated our previous finding that Foxj1a is a direct transcriptional target of Wnt (Caron et al., 2012). We further uncovered two critical roles that the Wnt-Foxj1a axis plays during LR pattern formation. First, Wnt regulates cilia growth via a ciliary function of Foxj1a. This is a known function of Foxj1a that is dependent on its ciliogenic function. Second, Wnt determines charon transcription via a non-ciliary function of Foxj1a. This is a novel function of Foxj1a that is independent of its canonical role as a ciliogenic transcription factor. Because cilia defects will affect directional nodal flow and subsequent asymmetric expression of charon, we propose a model that ciliary and non-ciliary functions of Wnt-Foxj1a signaling converge at charon to regulate LR axis determination (Fig. 8). Our novel discovery of non-ciliary function of Foxj1a underlying Wnt-implicated laterality defect adds new knowledge to the field of LR asymmetry and might be applicable to other signaling pathways that regulate Foxj1a expression.

Molecular nature of Wnt/β-catenin signaling in foxj1a activation
We presented both biochemical and genetic evidence to suggest that Lef1 and Tcf7 are transcription factors that cooperatively confer Wnt-regulated foxj1a transcription during KV development. Unlike severe disruption by inhibition of Wnt signaling (Lin and Xu, 2009;Caron et al., 2012), lef1 and/or tcf7 mutants exhibit moderate defects in foxj1a expression, ciliogenesis, and presumably, LR asymmetry. This phenomenon is likely attributable to maternal expression of Lef1 and/or Tcf7, which compensates for loss of zygotic Lef1 and/or Tcf7 in the mutants (Dorsky et al., 1999;Veien et al., 2005). Precedent for this scenario comes from the observation that cilia can still form in zygotic oval mutants but are absent in maternal-zygotic oval mutants during early embryogenesis (Huang and Schier, 2009). Implication of other Tcfs in KV ciliogenesis is less likely because we were unable to detect expression of other tcfs in DFCs or KV.
Besides Lef1 and Tcf7, several observations prompted us to propose that Wnt needs cofactors to fulfill its function on foxj1a expression. First, Wnt signaling has a broader functional domain than foxj1a, which exhibits a more tissue-restricted expression pattern (Aamar and Dawid, 2008;Yu et al., 2008). Second, Wnt activation enhances foxj1a levels transiently in DFCs but continuously in the developing pronephros. It is possible that a negative feedback mechanism is utilized in DFCs while a positive feed-forward mechanism is engaged in the pronephros. Third, Wnt activation induces ectopic foxj1a expression in a small proportion of cells, which might be fate-specified to express certain transcription cofactors. In support of our cofactor hypothesis, it has been reported that members of the Lef/Tcf family bind to target DNA weakly and with moderate specificity, which demands cooperative interactions with other factors to achieve tight and specific control of target gene regulation. For example, SMADs and members of the homeodomain family (Pitx2, Cdx-1, Nrarp) are engaged in cooperative interactions with Lef/Tcfs to regulate transcription of target genes (Labbe et al., 2000;Nishita et al., 2000;Beland et al., 2004;Ishitani et al., 2005;Vadlamudi et al., 2005). Identification of Lef1 and Tcf7 laid the foundation for our future search for Wnt cofactors in controlling foxj1a expression in KV, which will enable us to gain further insight into the molecular nature of this novel function of Wnt.

Ciliary function of Foxj1a in mediating Wnt signaling
Overexpression of Foxj1 was shown to induce node-like monocilia in Xenopus and zebrafish by activating a large number of genes that encode components unique to motile cilia, including dynein arms, central pair, and radial spokes (Stubbs et al., 2008;Yu et al., 2008). Consistently, our data indicate that Foxj1a is responsible for cilia length elongation and cilia component expression. By employing a targeted Wnt activation system, we showed that Wnt is sufficient to induce a distinctive ectopic tissue protrusion, which can be used to conduct gain-of-Wnt-function analysis. Ectopic expression of foxj1a was detected, and knocking down Foxj1a ablated the de novo cilia synthesis, underscoring the ciliary function of Wnt-Foxj1a signaling. This platform overcomes the shortcoming of early embryonic lethality resulting from global manipulation of important developmental pathways, and is ideal for future studies to identify pathways or molecules that are involved in Foxj1 expression and cilia synthesis. In addition to cilia growth, this platform is also useful for testing candidate genes that are implicated in cilia assembly. Given that the ectopic protrusion is morphologically distinguishable, de novo synthesized cilia can be identified for structural analysis via transmission electron microscopy.
Our past and present studies placed Wnt/β-catenin signaling upstream of ciliogenesis via direct transcriptional regulation of foxj1a (Caron et al., 2012). In contrast, functions of cilia upstream of Wnt signaling are contradictorily reported. Some studies suggest that cilia exert an inhibitory role on the Wnt/β-catenin pathway (Gerdes et al., 2007;Corbit et al., 2008) and are required for switching between the Wnt/β-catenin and Wnt/PCP pathways (Simons et al., 2005). However, other investigators argue against a function of cilia in regulating Wnt signaling (Huang and Schier, 2009;Ocbina et al., 2009). Of note, we observed tissue-specific foxj1a activation by Wnt, as evidenced by transient and persistent increase in DFCs and the developing pronephros, respectively. It remains to be determined whether foxj1a levels are differentially regulated by a feedback or a feed-forward mechanism to which cilia-Wnt regulation might contribute. (C-H) Foxj1a overexpression rescues charon levels in embryos with reduced Wnt signaling. charon transcript levels were downregulated by DFC-targeted injection of fzd10 MO (4 ng) (C, 28/29) and by Dkk1 induction (F, 26/30), which were restored by injection of foxj1a mRNA (2 ng) (D, 23/28; G, 25/32). Injection of foxj1a mRNA alone did not significantly affect charon expression (H, 16/16). Tg(hsp:dkk1-GFP) embryos were heat shocked at 70% epiboly, and GFP+ (Dkk1+) embryos were selected under fluorescence microscope. Shown are dorsal views of tailbud region at 10 to 12 somites.

Non-ciliary function of Foxj1a in transducing Wnt signaling
In addition to the canonical role of Foxj1a in ciliogenesis, our data reveal a novel function of Foxj1a in KV development, i.e. it confers Wnt-regulated charon expression. Unlike its function as a ciliogenic transcription factor, this novel function of Foxj1a is likely independent of its DNA-binding ability. Consistent with this nonciliary function of Foxj1a, we and others found that cilia and cilia motility are not required for overall charon expression (Gourronc et al., 2007). In fact, non-ciliary functions of Foxj1 have also been suggested in other tissues, including the differentiation of radial glia into ependymal cells and a subset of astrocytes in the postnatal brain and specification of cellular lineage in the forebrain (Jacquet et al., 2009(Jacquet et al., , 2011. On the basis of sequential expression of foxj1a and charon during KV development, Foxj1a might not directly regulate charon transcription. Instead, we favor the hypothesis that Foxj1a might affect charon expression through regulating KV cell differentiation. Non-ciliary function of the Wnt-Foxj1 axis in asymmetry remains to be investigated in mammals. The LR axis is randomized in Wnt3a and Foxj1 null mice, but cilia at the embryonic node are still present (Chen et al., 1998;Zhang et al., 2004;Nakaya et al., 2005). The presence of nodal cilia might be explained by the Rfx family of X-box-binding transcription factors, which act redundantly with Foxj1 to induce motile ciliogenesis in the neural tube and, presumably, in the mouse node but not in the zebrafish KV (Bonnafe et al., 2004;Yu et al., 2008;Cruz et al., 2010;Bisgrove et al., 2012). It would be informative to determine in the future whether Cerl2, whose deficiency results in a wide range of laterality defects, is downregulated in Wnt3a and Foxj1 null mice (Marques et al., 2004). Strengthening the Wnt-Cerl2 relationship, recent studies suggested that Wnt3 is asymmetrically expressed in the crown cells of the node and enhances asymmetric decay of Cerl2 mRNA (Nakamura et al., 2012;Kitajima et al., 2013). Nonetheless, these mechanistic studies will help us to better interpret related human diseases such as congenital heart diseases associated with LR defects.
To perform targeted overexpression, mRNAs (100 pg), alone or together with MOs (150 pg), were injected into single cells at the animal pole of 128cell staged embryos. gfp mRNA (100 pg) was coinjected to serve as a marker (Agathon et al., 2003;Lin et al., 2007). Embryos that harbored localized green fluorescence protein (GFP)-fluorescence signals or contained ectopic protrusions were collected at 10 to 21 somites for further analysis.

In situ hybridization and immunofluorescence
Single-color, whole-mount in situ hybridization was performed as previously described (Xu et al., 2002). Cilia, basal body, and apical-basal polarization of KV cells were visualized using antibodies against acetylated α-tubulin (Sigma-Aldrich), γ-tubulin (Sigma-Aldrich), and the tight junction protein ZO-1 (Invitrogen) as previously described (Lin and Xu, 2009). Cilia length and number were measured using AxioVision software (Zeiss) and analyzed by the Student t-test. For analysis of the ectopic protrusion, immunostaining was performed in whole-mount embryos, and then the ectopic protrusion was dissected and mounted in VECTASHIELD (Vector Laboratories), or the embryos were embedded in JB-4 plastic resin (Polysciences) and sectioned at 4-μm sections .

Electrophoretic mobility shift assay
Electrophoretic mobility shift assay was performed as previously described (Hellman and Fried, 2007). Full-length zebrafish lef1 cDNA was cloned into a pGEX4T-1 plasmid. Lef1-GST fusion protein was expressed in Escherichia coli and purified using Glutathione-Sepharose 4B (GE Healthcare). Briefly, Lef1-GST fusion protein was incubated with a 32 P-labeled 0.6-kilobase (kb) regulatory sequence of foxj1a in the presence or absence of excess cold probe or with 32 P-labeled oligonucleotides corresponding to putative Lef/Tcf binding sites (Caron et al., 2012). The sequences for the oligonucleotide probes are GGGACTCTGTTTACAGG-GGG (D1), TATTTATCCTTTGTTTAGAT (D2), TATTTATCCATAGT-TTAGAT (D2m, mutated D2), and TACACCCACAGAGACATTTG (D3). and Tcf7 binding to the foxj1a regulatory sequence. The Wnt-Foxj1a axis appears to regulate LR asymmetry via at least two mechanisms: the first involves cilia outgrowth that is dependent on the canonical function of Foxj1a in cilia formation, and the second involves expression of charon that requires nonciliary function of Foxj1a. Yellow highlighting indicates novel discoveries of the study.