Fam46a regulates BMP-dependent pre-placodal ectoderm differentiation in Xenopus

Cranial sensory nerve-related organs in the vertebrate are largely differentiated from a specified ectodermal region called the pre-placodal ectoderm (PPE). The PPE is a thick and U-shaped primordium around the neural plate. It is derived from the neural plate border (NPB) at the neurula stage (Couly and Le Douarin, 1985; Baker and Bronner-Fraser, 1997; Holland and Holland, 1999; Baker and Bronner-Fraser, 2001). The PPE differentiates into the anterior pituitary gland and sensory organs, such as the olfactory epithelium, the lens, the trigeminal nerves, the epibranchial nerves, the otic vesicles and the lateral lines (Streit, 2004; Schlosser, 2005, 2006). The neural crest (NC), which is also derived from the NPB, differentiates into sensory nerves and other organs, such as pigment cells, cartilage and secretory cells, mainly in the trunk region (Le Douarin and Kalcheim, 1999; Baker and Bronner-Fraser, 1997; Baker and Bronner-Fraser, 2001). One of the key signaling pathways that organizes the differentiation of PPE and NC cells is the Bone morphogenetic protein (BMP)-mediated pathway.

BMP is a morphogen that is secreted from the ventral side of the embryo and plays a pivotal role in ectodermal patterning. BMP ligands bind to the receptor, and the intracellular domain of the receptor directly phosphorylates R-Smad (Smad1, Smad5 and Smad8), inducing the formation of their dimer. The R-Smad dimers bind to the signal transducer co-Smad (Smad4) in the cytoplasm, and the complex translocates into the nucleus and binds to the promoter region of the target genes of BMP signaling (Massagué et al., 2005; Kitisin et al., 2007; Schmierer and Hill, 2007). The inhibition of BMP signaling induces the neuroectoderm, whereas the activation of BMP signaling induces the non-neural ectoderm during gastrulation (Weinstein and Hemmati-Brivanlou, 1997; De Robertis and Kuroda, 2004). The NPB is induced between the neuroectoderm and the non-neural ectoderm. It has been reported that NPB differentiation requires at least a transient activation of BMP signaling at gastrula stage (Kwon et al., 2010; Grocott et al., 2012; Saint-Jeannet and Moody, 2014). For PPE and/or NC differentiation from the NPB, moderate or complete inhibition of BMP signaling is considered to be essential after that stage (Brugmann et al., 2004; Glavic et al., 2004; Hong and Saint-Jeannet, 2007; Kwon et al., 2010; Watanabe et al., 2015). The PPE and the NC are thought to complement each other (Leung et al., 2013; Nordin and LaBonne, 2014). However, the differentiation mechanism of these tissues, especially the precise regulation of BMP activity in the PPE, is not fully understood.

To explore novel genes that govern the differentiation of PPE, we previously established a method by which the PPE-like cells were induced from Xenopus ectodermal explants, the animal cap, using BMP antagonist Chordin (Chd; Chrd) and Fibroblast growth factor (FGF) inhibitor SU5402 (Watanabe et al., 2015); a weak activation level of FGF signaling is required for PPE induction in addition to an intermediate level of BMP signaling (Ahrens and Schlosser, 2005; Litsiou et al., 2005). Using the PPE-like cells, we performed DNA microarray analysis and successfully identified a novel PPE gene, Fam46a (also known as Tent5a). Fam46a has an NTP transferase-7 domain (Lagali et al., 2002; Kuchta et al., 2009, 2016) and is a causative gene for retinitis pigmentosa and skeletal dysplasia (Diener et al., 2016; Barragán et al., 2008), but its function during early embryogenesis has not been reported. In this study, we examined the role of Fam46a for PPE formation. We show that Fam46a upregulates PPE specification but inhibits NC formation. Also, Fam46a activates BMP signaling by its interaction with Smad1 and Smad4. These results suggest that Fam46a defines the PPE region through precise regulation of BMP signaling during early development.

In Xenopus laevis, which is an allotetraploid species, the amino acid sequences of two homeologs of Fam46a (Fam46a.L and Fam46a.S) were highly similar (Fig. S1A,B). The Fam46a gene exists in a broad range of vertebrates, and the amino acid sequences are well conserved among them. In particular, the NTP transferase-7 domain has more than 90% identity with the human and mouse homologs (Fig. S1C,D). The Fam46 family consists of four paralogs, and they also show a high degree of similarity among them (Fig. S1E,F).

To analyze the change of Fam46a expression during embryogenesis, we performed reverse transcription polymerase chain reaction (RT-PCR) analysis and compared Fam46a with other ectodermal genes (Fig. 1A). Fam46a expression was observed at stage 10 (early gastrula stage). The expression increased further at stage 25 (tailbud stage) and continued to stage 40 (early tadpole stage). Six1 and Eya1 (PPE genes) were expressed from stage 12 until stage 40. Zic1 (an NPB gene), Vent1 and Vent2 (BMP target genes) were also expressed from stage 10. These results indicate that Fam46a is expressed from approximately the same stage as the expressions of NPB genes and BMP-signaling target genes begin.

Next, we carried out whole-mount in situ hybridization from the neurula stage to early tadpole stages (Fig. 1, Fig. S2). At the neurula stage, Fam46a was expressed in the PPE region where PPE marker gene Six1 is expressed (Fig. 1B-C′). At the tailbud stage, Fam46a was expressed in the optic vesicles and the otic vesicles (Fig. 1D-D′′). Interestingly, Fam46a is not expressed in the lens, which is consistent with the expression pattern of Six1 and Eya1 (Pandur and Moody, 2000; David et al., 2001; Ghanbari et al., 2001). At the tadpole stage, Fam46a expression was observed in the branchial arch mesenchyme and the retinal pigment epithelium (Fig. 1E,E′). These expression patterns were similar to those of other PPE genes, including Six1 and Eya1 (Pandur and Moody, 2000; David et al., 2001). We also examined the subcellular localization of GFP-Fam46a using HeLa cells. At 24 h after transfection, GFP-Fam46a was localized in both the nucleus and the cytoplasm (Fig. 1F). This feature is consistent with that of poly(A) polymerase, one of the NTP transferases (Schmidt and Norbury, 2010).

To further confirm the expression of Fam46a in the PPE, we applied the PPE-like cells for quantitative RT-PCR (RT-qPCR) experiments (Watanabe et al., 2015). As expected, Fam46a was significantly expressed in the PPE-like cells (Chd+SU5402), compared with the uninjected cells, which are known to be an epidermal property (Fig. 2A). Similar to the change of Fam46a expression, the PPE genes Six1 and Eya1 are highly expressed in the PPE-like cells (Fig. 2B,C), whereas the expression of the epidermal marker gene XK81 (Krt12.4) severely decreased to around one-third (Fig. 2D). In addition, expression of the NP genes Sox2 and NCAM (Ncam1) in the PPE-like cells was slightly reduced compared with those injected with only Chd (Fig. 2E,F). Expression of the NC genes FoxD3 and Snail (Snai1) were not significantly increased in the PPE-like cells (Fig. 2G,H). These results suggest that Fam46a is highly expressed in the PPE.

To analyze the role of Fam46a during early development, we overexpressed or knocked down Fam46a. Overexpression of Fam46a changed the body color to white, decreased the eye area and shortened the body length (Fig. 3A-B′). The eye area was ∼43% (Fig. S3A, S3C) and the body length was ∼86% (Fig. S3B, S3C) of those seen in the normal embryo. The ratio of the white-colored embryos increased in a dose-dependent manner (Fig. 3E). To knockdown Fam46a, a morpholino oligo (MO) was designed to inhibit the splicing at the 5′ end of exon 2 in both Fam46a.L and Fam46a.S genes (Fig. 3F). The Fam46a MO was evaluated using RT-PCR. A longer mRNA was observed in Fam46a MO-injected embryos (Fig. 3F), indicating that the inhibition of the splicing occurred in Fam46a-knockdown embryos, as expected. The Fam46a MO-injected embryos appear to be darker than the standard MO-injected embryo, and the retina area was grossly reduced at the early tadpole stage (Fig. 3C-D′), whereas the body length was not obviously altered. The ratio of individuals exhibiting the retinal disorder phenotypes increased in a dose-dependent manner (Fig. 3G). To further confirm the specificity of the MO, Fam46a MO and Fam46a mRNA were co-injected, and the number of abnormal tadpoles decreased (Fig. 3G). This result indicates that the phenotypes we observed were dependent on the inhibition of Fam46a. As pigment cells are known to be derived from the NC region cells, the body color phenotype in Fam46a-modulated embryos suggests that Fam46a plays a role not only in PPE but also in NC formation.

To further examine the role of Fam46a in PPE and NC specification, we next analyzed the expression pattern of ectodermal marker genes in embryos injected with mRNA or MO of Fam46a. When Fam46a mRNA was injected, the area of expression of Six1, a PPE gene, was expanded (Fig. 4A, upper), whereas Fam46a MO injection caused the reduction of the Six1-expressed region (Fig. 4A, lower). Fam46a overexpression reduced the expressions of Slug (Snai2) and FoxD3 (NC marker genes), whereas Fam46a knockdown induced a remarkable expansion of NC genes' expression (Fig. 4B). Interestingly, when Fam46a MO was injected, the NC gene expression region expanded into the area where the expression of the PPE gene was reduced. We also examined the expression pattern of other ectodermal genes. Fam46a overexpression slightly reduced the expression of XK81 and Sox3 (a neural plate marker gene) (Fig. 4C, upper), whereas Fam46a knockdown slightly increased the expression of these genes (Fig. 4C, lower). It should be noted that the overexpression or knockdown of Fam46a did not show the described phenotype at a high frequency. This may be because mRNA or MO is generally spread to a limited area in Xenopus embryos. From these results, we conclude that Fam46a induces PPE formation, but inhibits specification of the NC, epidermis and neural plate (Fig. 4D).

At the late neurula stage, the PPE region subdivides into the anterior, the lateral and the posterior PPE (Schlosser, 2006; Saint-Jeannet and Moody, 2014). We next examined the expression of Six3 (an anterior PPE gene), Ath-3 (Neurod4; a lateral PPE gene) and Pax8 (a posterior PPE gene). Fam46a knockdown reduced the expression of these genes (Fig. S4A), but expanded that of a posterior NPB gene (Pax3), an eye-marker gene (Pax6) and a cement gland gene (Muc2) (Fig. S4B). These results suggest that Fam46a is necessary for each placode formation and other ectodermal region specification.

To confirm the role of Fam46a in PPE and NC specification, we applied the PPE-like cells and the NC-like cells (Watanabe et al., 2015) and analyzed the expression of ectodermal genes. In the PPE-like cells, Zic1 (an NPB gene), Six1 (a PPE gene), Six6 (an anterior PPE gene) and Pax2 (a posterior PPE gene) were expressed (Fig. 4E and Fig. S5A). However, the expression was severely reduced by the knockdown of Fam46a. In the NC-like cells, NPB genes (Pax3 and Zic1) and NC genes (Snail and Slug) were expressed (Fig. 4F and Fig. S5B). When Fam46a mRNA was injected, the expression of Pax3, Snail and Slug was reduced, but there was no effect on the expression of Zic1 (Fig. 4F and Fig. S5B). These results confirm that Fam46a upregulates the expression of both NPB and PPE genes, whereas it strongly inhibits NC marker gene expression, consistent with our in vivo analyses. In addition, considering the role of Zic1 in PPE formation (Jaurena et al., 2015), Fam46a possibly regulates PPE formation through the maintenance of Zic1.

To further investigate how Fam46a regulates PPE specification, we analyzed an intracellular signaling pathway, BMP signaling. BMP signaling plays a major role in ectodermal patterning, and it has also been shown that Fam46a is one of the Smad1-associating proteins in humans (Colland et al., 2004). These reports suggest that Fam46a regulates PPE formation by modulating BMP signaling with Smad1. To examine this, we applied the type II receptor of BMP (BMPRII; also known as bmpr2) and a truncated type I BMP receptor (tBR) (Suzuki et al., 1994). As already reported, the overexpression of BMPRII induces the expression of the BMP target genes Vent1 and Vent2, whereas the expression of tBR reduces these expressions (Fig. 5A). Similar to the overexpression of BMPRII, the overexpression of Fam46a increased the expression of Vent1 and Vent2, but the knockdown did not (Fig. 5B). These results indicate that Fam46a enhances the expression of BMP target genes. To confirm these results, we also performed RT-qPCR using the animal cap. When Fam46a was overexpressed, the expression of both Vent1 and Vent2 was increased (Fig. 5C).

The low expression of tBR is known to expand the expression of Six1, FoxD3 and Sox2, and to inhibit that of XK81 (Fig. 5D and E, upper) (Sato et al., 2005). Interestingly, when Fam46a mRNA was co-injected with the low dose of tBR, the expression region of Six1 was drastically expanded (Fig. 5D and E, lower). However, Sox2 expression was not significantly reduced, probably because Sox2 is weakly expressed in the PPE. The expression of FoxD3 and XK81 was almost lost, or reduced. It is possible that Fam46a no longer rescues XK81 expression because tBR greatly reduces BMP signaling. These results suggest that Fam46a plays an important role in PPE differentiation by adjusting the level of BMP signaling.

To investigate how Fam46a is involved in BMP signaling, we next examined the change of Fam46a localization in response to BMP addition. We transfected GFP-Fam46a into HeLa cells and observed the fluorescent signal before and after the addition of BMP4. GFP-Fam46a started to accumulate in the nucleus, from the cytoplasm, after BMP4 addition, and this distinct nuclear localization was still present after 60 min (Fig. S6A-C). These results indicate that Fam46a protein translocates into the nucleus in response to BMP signal activation.

We next examined the direct interaction between Fam46a and BMP signaling components by immunoprecipitation experiments. Western blot analysis indicated that Fam46a physically binds to Smad1 (Fig. 6A). The signal intensity of the precipitates of Smad1 indicates that the binding of Fam46a-Smad1 was weaker than that of Smad1-Smad4 (Fig. 6B). To examine whether Fam46a and Smad1 bind to each other when BMP signaling is not activated, we conducted the same experiment in the presence of LDN193189, an inhibitor of the BMPR type I receptor. The binding of Fam46a and Smad1 was observed even under the inhibition of BMP signaling (Fig. 6C). Next, to identify the binding domain of Smad1 with Fam46a, we performed immunoprecipitation experiments using the deletion constructs of Smad1 (Fig. 6D). Smad1 has two domains: the MH1 domain has a DNA-binding motif, and the MH2 domain includes a SSXS motif, which is phosphorylated by BMPRII. The two domains are separated by a linker sequence. Using the three types of constructs, we found that Fam46a binds to full-length Smad1 (Smad1-WT) and Smad1-MH1, but not to Smad1-MH2. This indicates that Fam46a interacts with MH1 and/or a linker domain of Smad1. Fam46a appeared to interact with Smad1-MH1 more strongly than with Smad1-WT (Fig. 6E). We also examined the interaction between Fam46a and Smad4. There was a strong interaction between Fam46a and Smad4, similar to the binding between Smad2 and Smad4 (Fig. 6F,G).

Because the subcellular localization of Smad1 is known to be important for BMP signaling, we next examined the localization of mCherry-Smad1 under Fam46a-expressed conditions. Canonical Wnt signaling is known to induce organizer-triggering invagination. Therefore, to maintain BMP signaling in the blastula ectoderm, we used β-catenin (ctnnb1) MO for the inhibition of invagination (Inomata et al., 2013). In β-catenin MO and mCherry-Smad1 mRNA-injected embryos, the mCherry-Smad1 nuclear localization level was high (Fig. S7A, 1st line). In Chd mRNA-injected embryos, the ratio of mCherry-Smad1 localization in the nucleus decreased (Fig. S7A, 2nd line). When Chd and Fam46a mRNA were co-injected, mCherry-Smad1 was localized again in the nucleus (Fig. S7A, 3rd line). The nucleus-to-cytoplasm ratio of Smad1 localization was quantitatively evaluated using ImageJ (Fig. S7B). From these results, we concluded that Fam46a promotes the nuclear localization of Smad1.

Our data showed that Fam46a physically interacts with Smad1 and Smad4 and promotes BMP target gene expression. We further investigated the mechanisms of transcriptional regulation of BMP target genes by Fam46a. In this experiment, we used a Vent2-TCFm-Luc reporter construct containing a Vent2 promoter, which has mutations in the TCF-binding site and only consists of BMP responsive elements (Hikasa et al., 2010). The Vent2-TCFm-Luc reporter was stimulated by BMP4 expression, and also by Fam46a expression (Fig. 7A). However, co-expression of BMP4 and Fam46a did not enhance the luminescence intensity (Fig. 7A). This result suggests that BMP4 or Fam46a expression caused the strong activation and no stronger activation was seen by co-expression. On the other hand, when Fam46a was knocked down, Vent2-TCFm-Luc reporter activation was severely reduced, even in the presence of BMP4 (Fig. 7B). These results indicate that Fam46a is required for the activation of BMP target gene transcription via a Vent2 reporter.

As we described above, we found that Fam46a interacts with MH1/linker domain of Smad1 (Fig. 6E). Smad1 protein is known to be constantly degraded in the cytosol through its phosphorylation on the linker region by Glycogen synthase kinase 3 (GSK-3β) or Mitogen-activated protein kinase (MAPK) (Wnt/FGF signaling components), followed by polyubiquitylation (Fuentealba et al., 2007; Ueberham and Arendt, 2013). So, we examined the amount of Smad proteins present when Fam46a was expressed, to evaluate whether Fam46a is involved in this process. In Myc-Smad1 mRNA-injected embryos, we observed two Smad1 forms, the phosphorylated form and the unphosphorylated form (Fig. 7C). When GFP-Fam46a was expressed, the amount of phosphorylated Smad1 was not significantly altered, whereas the amount of unphosphorylated Smad1 increased (Fig. 7C,E). We also examined the amount of Smad4 protein in Fam46a-expressing cells. When GFP-Fam46a was expressed, the amount of Smad4 remained unaltered (Fig. 7D). Together, these results suggest that Fam46a also facilitates BMP signaling via the stabilization of Smad1, not Smad4.

In this study, we identify a novel PPE gene, Fam46a, and reveal the crucial role of Fam46a in PPE formation. This provides a new regulatory mechanism of BMP-signal activity with an intracellular factor in the PPE.

BMP signaling is required for PPE induction at the early gastrula stage, but the level of the signaling after that stage is still controversial: complete inhibition at late gastrula stage (Ahrens and Schlosser, 2005; Litsiou et al., 2005; Kwon et al., 2010) or intermediate level at early neurula stage (Brugmann et al., 2004; Glavic et al., 2004; Hong and Saint-Jeannet, 2007; Watanabe et al., 2015). Our study shows that Fam46a overexpression induces PPE formation through BMP-signal activation. This suggests that the complete inhibition of BMP signaling is not necessarily required for PPE induction.

PPE and NC arise from the common precursor NPB at late gastrula stage. To distinguish the PPE from the NC, the proper level of FGF signaling is required for PPE formation, while the proper level of Wnt signaling is necessary for NC formation (Sato et al., 2005; Groves and LaBonne, 2014; Saint-Jeannet and Moody, 2014). In addition to these differences, Sox5, which acts as an inhibitor of BMP signaling, upregulates NC specification and inhibits PPE formation through the modulation of BMP signaling (Nordin and LaBonne, 2014). Our study also reveals that Fam46a activates BMP signaling and induces the formation of PPE, not that of NC. These studies suggest that, in addition to the different requirement on FGF and Wnt signaling, BMP activity is slightly different between the PPE and the NC. Thus, the discrete activities of BMP signaling are possibly important for the selective differentiation of the PPE and the NC. The difference in the intracellular components, including Fam46a, could make clear boundary of the signaling activity, which is not easily accomplished by the gradient of growth factors.

It has been shown that human Fam46a is associated with Smad1 by the yeast two-hybrid analysis (Colland et al., 2004), but how Fam46a is involved in BMP signaling has not been clarified. Our study shows that Fam46a binds to both Smad1 and Smad4 and activates the transcription of BMP target genes (Fig. 7). We also found that Fam46a enhances the stabilization of Smad1 protein, but does not increase the amount of the phosphorylated protein (Fig. 7C,E). Smad1 protein is known to be constantly degraded in the cytosol through phosphorylation in its linker region by GSK-3β or MAPK (Wnt/FGF signaling components), followed by polyubiquitylation (Fuentealba et al., 2007; Ueberham and Arendt, 2013). It is possible that the binding of Fam46a with Smad1-MH1 and linker region in the cytosol causes the inhibition of Smad1 phosphorylation and subsequent degradation, and that the stabilization of Smad1 promotes the activation of BMP signaling.

It has been reported that human Fam46a interacts with Bag6 (previously known as Bat3) (Etokebe et al., 2015). This protein interacts with Smad1 and negatively regulates BMP signaling (Goto et al., 2011). It is also possible that Fam46a competes with Bag6 for association with Smad1, thereby enhancing BMP signaling.

In addition, based on analyses of the amino acid sequence and three-dimensional structure, Fam46a was recently reported as one of the non-canonical poly(A) polymerases (Kuchta et al., 2016; Mroczek et al., 2017). Non-canonical poly(A) polymerases add a poly(A) tail to pre-mRNA 3′-UTR ends after transcription (Richter, 1999). Because of this, this enzyme is considered to increase the stability of mRNA (Richter, 1999). These analyses suggest that it is also possible that Fam46a contributes to the stabilization of Smad mRNAs. Further analysis is required to reveal whether mRNAs are stabilized by Fam46a.

In summary, we have represented two possible modes of Fam46a function in BMP signaling in a schematic (Fig. 7F). In the first mode, when the BMP ligand associates with the receptor, Fam46a binds to both Smad1 and Smad4, and this complex translocates into the nucleus. This complex then binds to the promoter of a BMP target gene and activates transcription. In the second mode, Fam46a is involved in Smad1 stabilization and indirectly activates the BMP signaling pathway. Smad1 is constantly marked by ubiquitylation and is degraded, but Fam46a inhibits this process. In conclusion, Fam46a regulates PPE specification via the activation of BMP signaling in this region. In addition, the efficiency of PPE induction in mammalian stem cells is lower than in other ectodermal tissues. Because the amino acid sequences of Fam46a in Xenopus and human are highly similar, the finding of the role of Fam46a in PPE formation in Xenopus will contribute not only to developmental biology, but also to the stem cell field.

All experiments using Xenopus laevis were approved by The Office for Life Science Research Ethics and Safety, University of Tokyo. Frogs (wild type; female, 90 g-119 g; male, >60 g) were purchased from Watanabe Zoshoku (Hyogo, Japan). The developmental stages of Xenopus embryos were determined according to the standard description (Nieuwkoop and Faber, 1994). Fertilized eggs were obtained by artificial fertilization, and the embryos were de-jellied by 4.6% L-cysteine hydrochloride in 1×Steinberg's solution (SS) (pH 7.8). Embryos were cultured in 1×SS to stage 9 and then placed in 0.1×SS to the appropriate stage.

Xenopus Fam46a was cloned into a vector: pCS2 or pCS2-GFP. In this study, pCS2-Chordin (Chd), pCS2-Wnt8, pCS2-BMP4, pCS2-GFP, pCS2-GFP-Fam46a, pCS2-GFP-Smad4, pCS2-6-Myc-Smad1, pCS2-Smad2-GFP, pCS2-6-Myc-Smad4, pCS2-6-Myc-MH1-Smad1 and pCS2-6-Myc-MH2-Smad1 were used as templates for in vitro transcription. mRNAs were transcribed in vitro using the mMESSAGE mMACHINE SP6 Kit (Invitrogen) and were then microinjected with a picoinjector PLI-100 (Harvard Apparatus) in 5% Ficoll/1×SS. Injected embryos were cultured to stage 9 and were then placed into 0.1×SS.

The morpholino antisense oligonucleotides (MOs) that we used in this study were designed as follows:

Fam46a-MO, 5′-GCCTCCTGCAATGTGAAATATAAGA-3′;

Standard control oligo (Std-MO), 5′-CCTCTTACCTCAGTTACAATTTATA-3′.

Fam46a-MO was synthesized against two Xenopus homoeologous genes, Fam46a.L and Fam46a.S, to block the exon-intron splicing (Fig. 3F). The sequence of the Std-MO was referred to Gene Tools.

In situ hybridization was performed as described previously (Morita et al., 2013). For lineage tracing, lacZ mRNA-injected embryos were stained with Red-Gal (Sigma-Aldrich) or X-Gal (Wako). The digoxigenin (DIG)-labeled RNA probes for Fam46a, Six1, Eya1, Slug, FoxD3, Snail, XK81, Sox3 and Sox2 were transcribed in vitro. The signals with a DIG-labeled RNA probe were detected with NBT-BCIP (Roche).

The animal cap is the animal pole region of Xenopus blastula. The cells were manually dissected at stage 8.5-9 with forceps and tungsten needles, and the specimens were cultured in 1×SS until stage 15.

Total RNA was purified by ISOGEN II (Nippon Gene). Reverse transcription was carried out using SuperScript III Reverse Transcriptase (Invitrogen). PCR was performed with ExTaq DNA polymerase (TakaraBio). RT-qPCR was performed using the KAPA SYBR FAST qPCR Kit (KAPA Biosystems). Primer sets used for PCR are listed in Table S1.

Results shown are representative of at least three biological replicates. Error bar represents S.E. between the triplicate sets. Unpaired two-tailed t-tests were used to determine the statistical significance.

HeLa cells were cultured in Dulbecco's Modified Eagle's Medium (High Glucose) (Wako) containing 10% fetal bovine serum (Cell Culture Bioscience) and 1% Penicillin-Streptomycin (Sigma-Aldrich) under 5% CO₂ at 37°C. Transfections were performed using Lipofectamine 3000 transfection kit (Invitrogen) according to the manufacturer's instructions. Transfected cells were cultured in the medium for 24 h, then treated with or without BMP4 (100 ng/ml). The nuclei were stained by NUCLEAR-ID Red (Enzo), according to the manufacturer's instructions.

HeLa cell images were acquired with a confocal microscope (Olympus) using UPLSAPO 20×objective. After taking pre-images, cells were treated with or without BMP4 (100 ng/ml) and then cultured for 120 min in 37°C, 5% CO₂. Images were captured at the time points of 0 min, 60 min and 120 min after BMP4 treatment.

The GFP-Fam46a fluorescence intensities in the cytoplasm and nuclear were quantified using ImageJ. The nuclear/cytoplasm ratios were calculated for each cell and compared between pre- and 120 min-images.

Embryos were lysed in RIPA buffer [0.1% NP40, 20 mM Tris-HCl (pH 8), 10% glycerol] supplemented with a protease inhibitor cocktail (Roche). Proteins were detected using the following antibodies: anti-Myc (1:1000, 562, Medical and Biological Laboratories), anti-GFP (1:1000, GTX113617, GeneTex), anti-β-tubulin (1:1000, ab6046, Abcam), anti-pSMAD1 (1:6000, #06-702, EMD Millipore) and anti-SMAD1 (1:1000, #9512, Cell Signaling Technology). Secondary antibodies that were conjugated with horseradish peroxidase were applied and detected by chemiluminescence using Chemi-Lumi One L (Nacalai Tesque).

Immunoprecipitation experiments were performed according to the manufacturer's instructions (GE Healthcare Life Science). Briefly, embryos were lysed with Lysis buffer [1% NP-40, 20 mM Tris-HCl (pH 8.0), 137 mM NaCl, 10% glycerol, 2 mM EDTA] supplemented with protease inhibitor cocktail (Roche). The supernatant was incubated with anti-GFP (1:100, sc-9996, Santa Cruz Biotechnology) or anti-Myc (1:100, 562, Medical and Biological Laboratories) for 1 h at 4°C, then incubated with protein G sepharose (GE Healthcare) for 24 h at 4°C, and analyzed by immunoblotting.

Luciferase reporter DNA, Renilla luciferase DNA and the indicated mRNA or MO were co-injected into both cells of two-cell embryos. Embryos were cultured to stage 10.5, three sets of ten embryos were collected and assayed with a Dual-Luciferase Reporter Assay System (Promega) and Fluoroskan Ascent FL plate reader (Thermo Fisher Scientific). Error represents S.E. between the triplicate sets. Unpaired two-tailed t-tests were used to determine the statistical significance.

We thank Drs Y. Ito, Y. Onuma, S. Matsukawa and K. Mizuno for technical advice. We also thank Drs H. Hikasa and S. Sokol for critical discussion and for providing the Vent2-TCFm-Luc reporter construct.