Kctd15 inhibits neural crest formation by attenuating Wnt/β-catenin signaling output

Specification, maintenance and differentiation of neural crest (NC) cells depend on multiple signaling pathways. In Xenopus and zebrafish, low levels of Bmp and high levels of Wnt signaling cooperate in NC induction (Saint-Jeannet et al., 1997; Wilson et al., 1997; Dorsky et al., 1998; Lewis et al., 2004). As Wnt signaling regulates anterior-posterior (A-P) patterning of the neural plate (Kim et al., 2000; McGrew et al., 1997; McGrew et al., 1995), Wnt signals might induce NC through posteriorization. However, neural A-P patterning and NC induction are separable events (Wu et al., 2005). A recent study indicates that Wnt-mediated posteriorization of the neural plate border (NPB) rather than the neural plate is crucial in NC induction (Li et al., 2009). Signaling pathways active in NC specification are modulated by activators and inhibitors to regulate their strength and spatial distribution (Hong and Saint-Jeannet, 2007; Sauka-Spengler and Bronner-Fraser, 2008; Zhao et al., 2008). Here we introduce a factor, Potassium channel tetramerization domain containing 15 (Kctd15), that has a profound influence on NC formation in zebrafish and Xenopus embryos. KCTD15 was identified in humans (Hotta et al., 2009; Willer et al., 2009) as a BTB domain-containing protein of unknown function. We show that zebrafish kctd15a and kctd15b are expressed at the NPB at the end of gastrulation. Ectopic expression of Kctd15 inhibits NC specification, whereas knockdown leads to expansion of NC markers. Simultaneous attenuation of Wnt and Kctd15 expression rescues NC specification in zebrafish embryos. We suggest that Kctd15 restricts the NC domain by interfering with the functioning or output of the Wnt/β-catenin signaling pathway.

Zebrafish (Danio rerio) embryos from AB/TL and TOP-dGFP (Dorsky et al., 2002) lines were obtained (Westerfield, 2000), and stages are indicated as hours post fertilization (hpf) (Kimmel et al., 1995). Xenopus laevis were staged according to (Nieuwkoop and Faber, 1967).

The open reading frames (ORFs) of zebrafish kctd15a (BC083478), zebrafish kctd15b (BC078294) and X. laevis kctd15 (BC077862) were subcloned into pCS2⁺. The zebrafish β-cat* construct has four point mutations; S33A, S37A, T41A and S45A.

Morpholino oligonucleotides (MO) were targeted to 5′UTR regions of zebrafish kctd15a (5′-TCCTTCCCTCCTTGGAAGACATAGC-3′) and zebrafish kctd15b (5′-AGCTCTCCTTCCCCCTCTTGATCTT-3′) (Gene Tools). Embryos at 1-2 cells were injected with 1.5 ng Kctd15a plus 0.5 ng Kctd15b MO; 2 ng Wnt8.1a MO (Lewis et al., 2004); or 2-4 ng standard control MO (5′-CCTCTTACCCTCAGTTACAATTTATA-3′) per embryo. 5′-capped mRNAs were prepared using the mMESSAGE mMACHINE Kit (Ambion). Each zebrafish embryo was injected with 50 pg kctd15a, kctd15b, ΔNkctd15a, ΔCkctd15a, kctd10, kctd6 or kctd13 mRNA, or 5 pg β-cat* mRNA. One blastomere of 2-cell Xenopus embryos was injected with 250 pg of Xkctd15 mRNA. For rescue experiments, 1-10 pg of mRNA was injected into fish embryos.

Zebrafish embryos were hybridized with one or two probes (Hauptmann and Gerster, 1994) for kctd15a, kctd15b, pax3 (Seo et al., 1998), foxd3 (Kelsh et al., 2000), sox2 (Okuda et al., 2006), sox10 (Dutton et al., 2001), dlx2a (Akimenko et al., 1994), msxb (Phillips et al., 2006), dlx3b (Ekker et al., 1992), mitfa (Lister et al., 1999), gfp (Dorsky et al., 2002), foxi1 (Solomon et al., 2003), lim3 (Glasgow et al., 1997), prl (Herzog et al., 2003), cldna (Kollmar et al., 2001), eya1 (Sahly et al., 1999), six4.1 (Kobayashi et al., 2000) and gata1 (Detrich et al., 1995). INT-BCIP (red, Roche) and BM purple (blue, Roche) were used. Alcian Blue staining was as described (Barrallo-Gimeno et al., 2004).

Xenopus 2-cell embryos were injected with mRNA(s) for Xwnt3a (300 pg/embryo), Xchordin (300 pg/embryo), Xkctd15 (500 pg/embryo) or β-cat* (50 pg/embryo). Animal caps were dissected at stage 8-9, cultured to stage 18 and RNA was analyzed by RT-PCR (Zhao et al., 2008).

HEK293T cells in DMEM with 10% fetal calf serum were transfected using FuGENE (Roche). We used Xkctd15 and zβ-cat* in pCS2⁺ in addition to TOPflash luciferase and Renilla luciferase pRL-CMV (Promega) constructs. Assays were performed as described (Zhao et al., 2008).

Cells in the NPB express kctd15a and kctd15b at the 1-somite (1s) stage (Fig. 1A,B). kctd15a colocalized with the preplacodal marker dlx3b (Toro and Varga, 2007) (Fig. 1C), but not with premigratory NC cells expressing foxd3 (Monsoro-Burq et al., 2005; Steventon et al., 2005) (Fig. 1D). Subsequently, pharyngeal arches express kctd15a; additional expression domains include pronephric ducts, cranial placode precursors and the brain (Fig. 1E; see Fig. S1C,D,G in the supplementary material). kctd15b transcripts occur maternally (see Fig. S1A,B in the supplementary material), and subsequently are seen in the olfactory placode, cranial NC, lateral line primordium, pharyngeal arches, fin buds and optic tectum (see Fig. S1E,F in the supplementary material).

Embryos injected with kctd15a or b mRNA showed a loss of pigmentation, short tail, bent axis, and loss of yolk extension (Fig. 1F,G; see Fig. S2A-C in the supplementary material). The non-NC-derived retinal pigmented epithelium was unaffected, suggesting that Kctd15 blocks the development of NC-derived melanophores. The effect is specific as mRNA injection of the related kctd6, kctd10 and kctd13 genes had no effect (see Fig. S2 in the supplementary material).

Embryos co-injected with MOs against kctd15a and kctd15b (Kctd15a^MO/15b^MO) exhibited small heads, abnormal somites and increased cell death (Fig. 1J,K), an apparent increase in pigmentation in the dorsal hindbrain region (Fig. 1L,M) and abnormal pharyngeal arches at 72 hpf (Fig. 1M, inset; see Fig. S3 in the supplementary material). These phenotypes were rescued by injection of kctd15a mRNA (see Fig. S4 in the supplementary material). Thus, suppressing Kctd15 had a complementary effect on pigmentation as overexpression, whereas pharyngeal arch development was impaired in both cases (Fig. 1M; see Fig. S3 in the supplementary material), suggesting early and late Kctd15 functions.

Kctd15 contains a BTB (POZ) domain (residues 32-119), a protein-protein interaction module (Stogios et al., 2005). Constructs encoding the N-terminal BTB domain (ΔCkctd15a) or the C-terminal region (ΔNkctd15a) were injected, showing that ΔCkctd15a caused a similar phenotype as full-length protein (Fig. 1G,I), whereas ΔNkctd15a had no effect (Fig. 1F,H). Thus, the BTB domain of Kctd15 is essential for its function in the embryo.

The above experiments suggest a role for Kctd15 in melanophore and pharyngeal arch development, both involving NC cells. Therefore, we analyzed the NC-specifier pax3 and the pan-NC markers foxd3 and sox10 (Monsoro-Burq et al., 2005; Steventon et al., 2005) in Kctd15 morphants and overexpressing embryos. Expression of markers was reduced or lost in overexpressing embryos (Fig. 2A,C,D,F,G,I); similar results were obtained for msxb and sox9b (data not shown). Consistently, Kctd15a^MO/15b^MO-injected embryos showed upregulation and expansion of these markers (Fig. 2B,E,H). These effects on foxd3 expression are consistent at different stages (see Fig. S5 in the supplementary material). The pan-neural marker sox2 is expressed at normal levels in Kctd15-overexpressing or morphant embryos (Fig. 2J-L) but the sox2 domain is smaller in the morphants owing to a narrowing of the neural tube that might lead to the small head phenotype at later stages (Fig. 1K).

To study NC precursor differentiation, we tested dlx2a and mitfa. At 24 hpf, we observed a loss of dlx2a in pharyngeal arches and of mitfa in melanophore precursors in kctd15a mRNA-injected embryos (Fig. 2O,R). foxd3 in the notochord and dlx2a in the telencephalon were unaltered (Fig. 2F,O), illustrating the specificity of NC inhibition. In Kctd15a^MO/15b^MO-injected embryos, the expression of dlx2a in pharyngeal arches was maintained (Fig. 2M,N) and expression of mitfa was increased (Fig. 2P,Q). Whereas the increased mitfa expression correlates with the intense pigmentation of MO-injected embryos (Fig. 1L,M), the inhibition of pharyngeal arch formation in these embryos (see Fig. S3B in the supplementary material) points to a late function of Kctd15 in this lineage.

At early stages, kctd15 is expressed in the preplacodal domain but excluded from NC precursors; thus, it might have a role in delineating preplacodal versus NC identity. We find that kctd15a RNA-injected embryos show enhanced expression of the preplacodal markers eya1 and six4.1 in the anterior domain (see Fig. S6A-D in the supplementary material). Similarly, pan-pituitary markers lim3 and prl were expanded (see Fig. S6E-H in the supplementary material). Expression of foxi1, a marker for otic progenitors, was unaffected and the later otic marker cldna was reduced (see Fig. S6I-L in the supplementary material). These results suggest that loss of NC caused by Kctd15 overexpression was compensated by expansion of the anterior preplacodal domain, whereas both NC and placode development were inhibited by Kctd15 in posterior regions.

Wnt signaling is necessary for NC specification and later differentiation (Dorsky et al., 1998; Garcia-Castro et al., 2002; LaBonne and Bronner-Fraser, 1998; Saint-Jeannet et al., 1997). We injected embryos with Wnt8.1 MO or with MOs for Wnt8.1, Kctd15a and Kctd15b and assayed NC specification. As reported (Lewis et al., 2004), foxd3 expression was lost in Wnt8.1 morphants (Fig. 3A,B). Importantly, we observed a dramatic rescue of foxd3 expression in embryos injected with a combination of Wnt8.1 and Kctd15a/b MOs (Fig. 3C,D). Similar results were obtained using sox10 (data not shown). kctd15a expression in Wnt8.1 morphants was unaffected (see Fig. S7A,B in the supplementary material). These observations indicate that Kctd15 interferes with Wnt signal transduction or output, such that the reduction in ligand availability in the Wnt8.1 morphants is compensated for by alleviating the inhibition of pathway output by Kctd15.

To probe the relationship between Kctd15 and Wnt signaling, we treated embryos with LiCl, which inhibits Gsk3β activity leading to β-catenin stabilization (Klein and Melton, 1996; van de Water et al., 2001). Embryos were treated with 0.1 M LiCl (shield to bud) or 0.2 M LiCl (shield to 80% epiboly) and assayed for foxd3 expression. Although 0.2 M LiCl led to ectopic foxd3 expression, Kctd15 overexpression abolished all foxd3 expression (Fig. 3E-J), indicating that Kctd15 acts downstream of Gsk3β in NC induction. By contrast, constitutively active β-catenin (β-cat*) rescued foxd3 expression in kctd15 RNA-injected embryos, although ectopic foxd3 expression was abolished (Fig. 3K-M). These results indicate that Kctd15 acts upstream of or in concert with β-catenin in NC induction.

Explants from Xenopus embryos (animal caps) are induced to the NC by expression of Wnt and a Bmp antagonist (Saint-Jeannet et al., 1997). We used this system to further explore the function of Kctd15. Kctd15 function is conserved in frogs, as injection of Xenopus kctd15 mRNA into one blastomere of 2-cell embryos suppressed the NC marker slug (snail2) (see Fig. S8 in the supplementary material). For explant experiments, embryos were injected with wnt3a plus chordin (chrd) mRNAs, with or without kctd15 mRNA, and animal caps were assayed for NC markers. Induced slug, foxd3 and sox9 expression was strongly inhibited by the co-injection of kctd15 mRNA (Fig. 4A). In zebrafish embryos, β-cat*-mediated induction of NC proved largely resistant to Kctd15 inhibition. This is also true in Xenopus animal caps (Fig. 4B). Thus, the explant and whole embryo experiments are consistent, indicating that Kctd15 inhibits NC induction at or above the level of β-catenin.

To determine whether Kctd15 is a general Wnt inhibitor, we tested Kctd15 in the Xenopus axis duplication assay (Smith and Harland, 1991; Sokol et al., 1991). Co-injection of Xkctd15 did not reduce the frequency or completeness of Wnt8-induced axis duplications (data not shown). A recent study showed that an activator of Wnt signaling, Skip (1-341), was inefficient in inducing axis duplications (Wang et al., 2010), suggesting that certain modulators of Wnt signaling are inefficient in early frog embryos. In zebrafish, cells responsive to β-catenin can be monitored in the transgenic TOP-dGFP reporter line (Dorsky et al., 2002). GFP expression was reduced in kctd15a mRNA-injected transgenic embryos compared with controls (see Fig. S9A,B in the supplementary material). More detailed inspection of the NPB region revealed that kctd15 mRNA-injected transgenic embryos showed a loss of GFP expression in the NC domain (see Fig. S9C,D in the supplementary material). Inhibition of the Wnt pathway could also be observed using TOPflash luciferase in HEK293T cells. Kctd15 reduced basal, Wnt3a or LiCl-stimulated TOPflash activity by about 3-fold, but not β-cat*-stimulated activity (Fig. 4C).

In this report, we introduce Kctd15 as a novel regulator of NC formation in zebrafish and Xenopus. We provide evidence that Kctd15 acts as an inhibitor of NC specification by interfering with the output of Wnt signaling. We show that: (1) ectopic expression of Kctd15 leads to a dramatic reduction in the induction of premigratory NC cells in whole embryos and of early NC marker genes in induced animal caps; (2) conversely, the expression domain of NC-specific genes is expanded in Kctd15 morphants; (3) loss of NC induction in zebrafish Wnt8.1 morphants was rescued by attenuating Kctd15 expression; (4) constitutively active β-catenin rescued the suppression of NC induction by Kctd15. These results show that Kctd15 is a potent regulator of NC formation.

Precursors of the NC and the preplacodal region arise from an initially mixed population of cells at the NPB (Streit, 2002) and are separated as a result of Wnt signaling and the action of Gbx2 (Li et al., 2009; Litsiou et al., 2005). Although the induction of Gbx2 in the NC domain is a key feature in its formation (Li et al., 2009), placode-specific inhibitors of NC induction might be required to assist in the separation of the two domains. We suggest that Kctd15 represents such a factor with a role in defining the NC versus the anterior preplacodal domain at the NPB. This suggestion is supported by the enhanced expression of several anterior preplacodal markers in kctd15-overexpressing embryos (see Fig. S6 in the supplementary material).

The BTB domain is found in many functional classes of proteins (Stogios et al., 2005) and various properties and associations have been reported for Kctd proteins (Bayon et al., 2008; Ding et al., 2009; Ding et al., 2008). Here, we have linked the inhibitory role of Kctd15 to Wnt pathway signal transduction or output. Our results show that Kctd15 is a novel, highly effective and specific regulatory component in NC induction with an apparent role in restricting the NC domain in the developing embryo.

We thank Hyunju Ro for the β-catenin mutant and other reagents, Martha Rebbert for technical support, Robert Cornell (University of Iowa) for plasmids, Brian Brooks for support and members of the Dawid laboratory for discussion. This research was supported by the Intramural Research Program of the NICHD, NIH. Deposited in PMC for release after 12 months.