RNA m6A Methyltransferase Mettl3 Regulates Spatial Neural Patterning in Xenopus laevis

ABSTRACT

N⁶-Methyladenosine (m6A) is the most prevalent internal RNA modification and has a widespread impact on mRNA stability and translation. Methyltransferase-like 3 (Mettl3) is a methyltransferase responsible for RNA m6A modification, and it is essential for early embryogenesis before or during gastrulation in mice and zebrafish. However, due to the early embryonic lethality, loss-of-function phenotypes of Mettl3 beyond gastrulation, especially during neurulation stages when spatial neural patterning takes place, remain elusive. Here, we address multiple roles of Mettl3 during Xenopus neurulation in anteroposterior neural patterning, neural crest specification, and neuronal cell differentiation. Knockdown of Mettl3 causes anteriorization of neurulae and tailbud embryos along with the loss of neural crest and neuronal cells. Knockdown of the m6A reader Ythdf1 and mRNA degradation factors, such as 3′ to 5′ exonuclease complex component Lsm1 or mRNA uridylation enzyme Tut7, also show similar neural patterning defects, suggesting that m6A-dependent mRNA destabilization regulates spatial neural patterning in Xenopus. We also address that canonical WNT signaling is inhibited in Mettl3 morphants, which may underlie the neural patterning defects of the morphants. Altogether, this study reveals functions of Mettl3 during spatial neural patterning in Xenopus.

INTRODUCTION

Posttranscriptional gene regulation fine-tunes protein output in dynamic biological systems, such as developing embryos and functioning neurons (1, 2). During embryogenesis, a variety of posttranscriptional mechanisms, together with transcriptional networks, orchestrate cell proliferation, differentiation, and migration (i.e., tissue morphogenesis) in a spatiotemporally dynamic fashion (3). N⁶-Methyladenosine (m6A) modification, among various types of chemical modifications found in mRNAs, is the most prevalent internal mRNA modification and regulates mRNA stability, splicing, and translation (4–6). Methyltransferase-like 3 (METTL3) is the responsible methyltransferase in the large m6A writer complex comprising METTL3, METTL14, and WT1-associated protein (WTAP) (4, 7). Fat mass and obesity-associated protein (FTO) and alpha-ketoglutarate-dependent dioxygenase AlkB homolog 5 (ALKBH5) are the two demethylases for m6A, whose function makes the system reversible and dynamic (8, 9). YTH domain-containing proteins, such as YTH domain-containing family proteins 1 and 2 (YTHDF1 and YTHDF2), recognize m6A marks on the mRNAs and regulate their stability and translation (5, 10).

The m6A writer complex is essential for the differentiation of germ cells, embryonic stem cells, neurons, and blood cells (11–15). However, due to the early embryonic lethality induced by the conventional Mettl3 knockout in mice (16), roles of the RNA m6A methyltransferase beyond gastrulation stages were able to be investigated only by tissue-specific conditional knockouts (17, 18). In particular, roles of m6A methyltransferase in the spatial neural patterning along the anteroposterior (A-P) axis, starting around embryonic day 7 (E7), were not able to be investigated thoroughly, and only the roles of m6A writer in cortical neurogenesis were studied at later stages, around E12, by neuronal ablation of Mettl14 (14).

In this study, we uncovered multiple postgastrula phenotypes of Mettl3 knockdown in Xenopus embryos. We found that loss of Mettl3 anteriorizes neurulae and inhibits neural crest and neuronal cell specification. Knockdown of mRNA destabilization factors or m6A reader, such as Lsm1, terminal uridylyl transferase 7 (Tut7), or Ythdf1, also causes anteriorization of the neural plate along with the loss of neural crest and neuronal cells, suggesting that the functions of Mettl3 in the spatial neural patterning are via the m6A-dependent mRNA destabilization pathway. On the other hand, knockdown of Dicer1 shows distinct ventralization phenotypes. We further address that WNT signaling, one of the major posteriorization signals during embryogenesis (19), is inhibited both in Xenopus embryos and in human cells upon Mettl3 knockdown, implying that the inhibition of the WNT signaling pathway underlies the phenotypes in Mettl3 morphants. Overall, our results unveil the role of Mettl3 in spatial neural patterning of Xenopus embryos.

RESULTS

Mettl3 is required for the proper spatial neural patterning in Xenopus embryos.

To investigate the functions of RNA m6A modification during Xenopus embryogenesis, we first analyzed the expression patterns of m6A writer mettl3 and m6A readers ythdf1 and ythdf2. Not surprisingly, mettl3, ythdf1, and ythdf2 are ubiquitously expressed throughout early developmental stages with relatively high expressions in neural tissues (neural plate at neurula stage; brain and spinal cord at tailbud/tadpole stage) (Fig. 1), implicating their widespread roles during neural development. We then examined loss-of-function phenotypes of Xenopus Mettl3 using antisense morpholino oligonucleotide (Mettl3 MO). Injection of Mettl3 MO into the dorsal blastomeres or into all 4 marginal blastomeres at the 4-cell stage produced tadpole embryos with a dorsally kinked axis and pigment loss (Fig. 2A), which suggests slight dorsalization and neural crest defects. However, injection of Mettl3 MO into the ventral blastomeres had no gross morphological defects (Fig. 2A). To characterize the phenotypes in terms of spatial neural patterning, we next injected Mettl3 MO unilaterally and performed in situ hybridization for various neural patterning markers at the neurula stage. Mettl3 morphants showed laterally an expanded sox2-positive neural plate and displayed lateral and/or posterior expansion of forebrain markers rx2a and otx2 and posterior shift of mid-hind brain markers en2 and krox20 (Fig. 2B). These changes in marker gene expression suggest overall anteriorization of the neural plate, which is consistent with dorsalization phenotypes at the tadpole stage (Fig. 2A). Also, neural crest cell marker twist and neuronal cell markers delta-1 and n-tubulin were reduced on the Mettl3 MO-injected side (Fig. 2B). These changes in the cell fate markers were confirmed by reverse transcription-quantitative PCR (RT-qPCR), showing increased RNA levels of cement gland (xag) and forebrain marker genes (bf1, otx2 and pax6) while decreased RNA levels of mid-hind brain (en2, krox20 and pax2), neuron (n-tubulin) and neural crest (twist and foxd1) marker genes (Fig. 2C). The specificity of Mettl3 MO phenotypes was confirmed by rescue experiments with the MO-insensitive form of mettl3 mRNA for the expressions of n-tubulin (Fig. 2D) and twist (Fig. 2E). Collectively, our results demonstrate that Xenopus Mettl3 is required for proper anteroposterior neural patterning and neural crest and neuronal differentiation.

FIG 1.

Xenopus mettl3, ythdf1, and ythdf2 are expressed ubiquitously throughout Xenopus embryogenesis. Shown are results of in situ hybridization of Xenopus embryos of various stages, depicting ubiquitous expressions of mettl3, ythdf1, and ythdf2. Note that the expressions of these genes in the neural tissues (neural plate, brain, and spinal cord) are relatively higher than in the other tissues. st, stage.

FIG 2.

Xenopus Mettl3 is required for the anteroposterior neural patterning and neural crest and neuronal specification. (A) Morphological phenotypes of Xenopus tadpole at stage 39 after control morpholino (Con MO) or Mettl3 morpholino (Mettl3 MO) injection to different blastomeres at the 4-cell stage. Dashed lines indicate kinked dorsal axis demonstrating dorsoanteriorization. (B) In situ hybridizations of stage 17 neurulae unilaterally injected with control morpholino (Con MO) or Mettl3 morpholino (Mettl3 MO) along with lineage tracer β-galactosidase mRNA for various neural patterning markers. Red-gal staining indicates injected side (Inj.). Uninjected side (Uninj.) served as an internal control. (C) RNA levels of various neural patterning genes were analyzed with RT-qPCR from stage 15 embryos injected with control or Mettl3 morpholino (n = 3). H4 RNA level was used for normalization. *, P < 0.05; **, P < 0.01 (one-sided t test). (D) Rescue of Mettl3 morphants with MO-insensitive mettl3 mRNA for n-tubulin expression. Shown at the right is the fraction of embryos showing normal or reduced (or absent) n-tubulin expression. (E) Rescue of Mettl3 morphants with MO-insensitive mettl3 mRNA for twist expression. Shown at the right is the fraction of embryos showing normal or reduced (or absent) twist expression.

Ythdf1/2 morphants phenocopy Mettl3 morphants.

Since METTL3 is the RNA methyltransferase for m6A, we next investigated whether the phenotypes produced by Mettl3 MO (Fig. 2) are m6A dependent. For this, we performed additional knockdown experiments for YTH domain-containing proteins Ythdf1 and Ythdf2, which recognize m6A modified mRNAs and regulate their stability and translation (5, 10). Knockdown of Ythdf1 inhibited the specification of neural crest and neuronal differentiation and anteriorized the neural plate, as evident by the reduction of n-tubulin and twist, expansion of rx2a, and the posterior shift of en2 on the MO-injected side (Fig. 3, upper images). On the other hand, knockdown of Ythdf2 only inhibited neuronal differentiation (n-tubulin, white arrowhead), while it had little effects on the neural crest specification and anteroposterior neural patterning (Fig. 3, lower images). These results suggest that the neural patterning defects of Mettl3 morphants are possibly via Ythdf1- and m6A-dependent mRNA regulation.

FIG 3.

Loss of m6A reader, Ythdf1, or Ythdf2 phenocopy neural patterning defects of Mettl3 morphants. Shown are results of in situ hybridizations of stage 17 neurula-stage embryos unilaterally injected with Ythdf1 MO or Ythdf2 MO along with lineage tracer β-galactosidase mRNA for various neural patterning markers. Loss of Ythdf1 inhibited neuronal and neural crest cell specification as well as anteriorization of neural plate (posterior expansion of rx2a and posterior shift of en2). However, loss of Ythdf2 only inhibited neuronal differentiation and had little effect on neural crest cell specification (twist, no change) and anteroposterior neural patterning. Dashed lines indicate midline of the embryos.

Lsm1 and Tut7 morphants show phenotypes similar to those of Mettl3 morphants in the spatial neural patterning.

RNA m6A methylation, recognized by YTHDF1/2, regulates the stability of the mRNAs by recruiting CCR4-NOT1 deadenylase complex to induce poly(A) tail deadenylation of the m6A modified transcripts (20). Once deadenylated, the mRNAs are either readenylated (and thus be reused for translation) or further uridylyl modified by the uridylating enzymes TUT4/7 and degraded by the action of 3′ to 5′ (exosome complex) or 5′ to 3′ (XRN1) exonucleases (21). To gain insights on whether the Mettl3- and m6A-dependent spatial neural patterning (Fig. 2 and 3) requires RNA degradation steps, we performed additional knockdown experiments for Lsm1 and Tut7, a component of the exosome complex and a uridylating enzyme, respectively. Loss of Lsm1 produced phenotypes similar to those of Mettl3 morphants, such as anteriorization of the neural plate (expansion of rx2a and posterior shift or reduction of en2) and inhibition of neural crest (twist reduction) and neuronal (n-tubulin reduction) differentiation (Fig. 4A, first and second columns). However, interestingly, knockdown of Tut7 showed a complete reduction of both the anterior and posterior neural markers (rx2a and en2) and a reduction in pan-neural marker sox2. Thus, the impaired neural crest and neuronal differentiation in Tut7 morphants was likely due to the reduced neural induction (Fig. 4A, third column). Notably, Tut7 knockdown often produced embryos with blastopore closure defects, a typical gastrulation defect, as previously described (22). Thus, we speculated that the pan-neural inhibitions after Tut7 knockdown were secondarily caused by the preceding defects during gastrulation. To address this hypothesis, we unilaterally injected Tut7 MO and selected only those embryos showing no gastrulation defects (i.e., embryos with successful blastopore closure) and raised them until mid-neurula stage (stage 17). Knockdown of Tut7 in neurulae without preceding gastrulation defects showed anteriorization of the neural plate along with inhibition of neural crest and neuronal differentiation (Fig. 4B), demonstrating the reiterated role of Tut7 during maternal-zygotic transition, gastrulation (22), and neurulation (this study). These results overall demonstrate that mRNA destabilization, initiated by the m6A modification and executed by the uridylation and exonuclease-mediated degradation, is required for the proper spatial neural patterning in Xenopus.

FIG 4.

Loss of Lsm1 or Tut7 phenocopy patterning defects of Mettl3 morphants. (A) In situ hybridizations of stage 17 neurulae unilaterally injected with Mettl3 MO, Lsm1 MO, or Tut7 MO for various neural patterning markers. β-galactosidase mRNA was coinjected as a lineage tracer. Loss of Lsm1 successfully phenocopied Mettl3 morphants. Loss of Tut7 caused pan-neural inhibition, while clearly showing open blastopore, suggesting that the pan-neural defects of Tut7 morphants were secondary from gastrulation defects (i.e., blastopore closure defects). (B) In situ hybridizations of stage 17 neurulae unilaterally injected with either control or Tut7 morpholino, with selection of only those showing successful blastopore closure (i.e., no gastrulation defects) for various neural patterning markers. Unilateral Tut7 morphants without gastrulation defects phenocopied Mettl3 morphants in anteroposterior neural patterning and neural crest and neuronal specification.

Dicer1 morphants show ventralization and pan-neural inhibition.

Of note, knockdown of Dicer1, an essential enzyme for the biogenesis of microRNAs (miRNAs) (23), caused pan-neural inhibition without showing apparent gastrulation defects (Fig. 5A). Indeed, knockdown of Dicer1 ventralized the embryo as evident by the increased sizzled (szl) and decreased otx2 expression in the tailbud (Fig. 5B). RNA levels of sizzled were also increased after Dicer1 knockdown in mid-gastrula embryos (Fig. 5C), while the levels of sizzled were decreased after Mettl3 knockdown (Fig. 5D). Therefore, we conclude that the mRNA regulation by the RNA m6A modification regulates proper anteroposterior neural patterning, while miRNA-dependent mRNA regulation is required for the initial neural induction.

FIG 5.

Loss of Dicer1 causes ventralization and defective neural induction. (A) In situ hybridizations of stage 17 neurulae unilaterally injected with Dicer1 MO for various neural patterning markers. β-galactosidase mRNA was coinjected as a lineage tracer. Loss of Dicer1 inhibited all the neural markers, implying ventralization or inhibition of neural induction. (B) In situ hybridizations of tailbud embryos (stage 30) injected with either control or Dicer1 morpholino to various blastomeres at the 4-cell stage for the forebrain marker otx2 and ventral marker sizzled (szl). (C) RNA levels of markers of pan-mesoderm (xbra), dorsal (chordin), or ventral mesoderm (sizzled) analyzed from stage 11 gastrulae marginally injected with either control or Dicer1 MO at the 4-cell stage. H4 was used for normalization. *, P < 0.05; **, P < 0.01 (one-sided t test). (D) RNA levels of mesodermal markers from stage 11 gastrulae marginally injected with either control or Mettl3 MO at the 4-cell stage. H4 was used for normalization. *, P < 0.05; **, P < 0.01 (one-sided t test).

Mettl3 is required for WNT signaling in Xenopus and in human cells.

WNT signaling, along with FGF and retinoic acid (RA) signaling, regulates anteroposterior neural patterning (19) and is also required for neural crest and neuronal differentiation (24, 25). To address whether WNT signaling is the responsible pathway underlying Mettl3-dependent spatial neural patterning, we measured RNA levels of WNT signaling components. Mettl3 morphants at the gastrula stage (stage 11) and neurula stage (stage 15) showed an overall downregulation of WNT target genes (dkk1, axin2, fz1, fz2, fz3, fz4, and lef1) while WNT nontarget genes (ctnnb1, dvl2, and gsk3b) were largely unaffected (Fig. 6A and B). These results suggest that the inhibition of WNT signaling in Mettl3 morphants might be responsible, at least partially, for the anteriorization of neural plate and the impaired specification of neural crest and neuronal differentiation.

FIG 6.

Mettl3 and m6A pathway components are required for WNT signaling in Xenopus and in human cells. (A and B) RNA levels of WNT signaling components from stage 11 gastrulae marginally injected (A) or from stage 15 neurulae dorsally injected (B) with either control or Mettl3 morpholino at the 4-cell stage (n = 3). WNT target genes were overall downregulated, while nontarget genes were not. (C and D) Relative activities TOP-Flash WNT reporter (C) or control reporter harboring mutations in T cell factor (TCF) binding sites (FOP-Flash) (D) were measured in HeLa cells stably expressing various shRNAs and treated with either mock control or WR (WNT3a and RSPO-1) (see Materials and Methods for details). Fold changes were calculated by normalizing WR to the control in each shRNA-expressing cell line. (E) Western blot of β-catenin and METTL3 from cell lysates of control or METTL3-silenced HeLa cells, treated with the control (-) or WR (WNT3a and RSPO-1). GAPDH was used as a loading control. (F) Relative WNT reporter (TOP-Flash) activities were measured in 293T cells stably expressing the indicated shRNAs and treated with either the control or WR (WNT3a and RSPO-1). Fold changes were calculated by normalizing WR to the control in each shRNA-expressing cell line. (G) Relative WNT reporter (TOP-Flash) activities were measured in HeLa cells stably expressing tetracycline-inducible shYTHDF2 (Tet-shYTHDF2), treated or not with doxycycline (Dox) and with control or WR. Fold changes were calculated by normalizing WR to the control under each condition (with or without Dox). (H) Western blot of β-catenin and YTHDF2 from HeLa cells expressing Tet-shYTHDF2, treated with doxycycline (Dox) and/or WR, as indicated. GAPDH was used as a loading control. (I) In situ hybridization of stage 11 gastrula embryos dorsally injected with either control morpholino (Con MO) or Mettl3 MO for the organizer marker chordin. Upper panels are the vegetal view with the dorsal side up, showing chordin signal in the dorsal marginal zone (i.e., organizer region). Lower panels are the cross-section view with the dorsal side on the right. The cross-sectional planes are marked with dashed lines in the insets. (J) Relative WNT reporter (TOP-Flash) activities were measured in Xenopus animal cap explants injected with TOP-Flash and CMV-Renilla reporter plasmid DNAs along with Xwnt8 mRNA, Con MO, or Mettl3 MO as indicated, excised at stage 10 and cultured until stage 12 before harvest. *, P < 0.05; **, P < 0.01; no symbol P > 0.05 (one-sided t test).

To address whether this regulation is conserved in human cells, we utilized the TOP-Flash WNT reporter system. Knockdown of METTL3 or YTHDF2 by the constitutively expressed short hairpin RNAs (shRNAs) in HeLa cells suppressed WNT reporter activity (Fig. 6C and D) and β-catenin accumulation in response to WNT3a stimulation (Fig. 6E). Knockdown of METTL3 or YTHDF2 in 293T also caused WNT inhibition (Fig. 6F). Inducible knockdown using Tet-inducible shRNA cell lines also demonstrated that YTHDF2 is required for WNT reporter activation (Fig. 6G) and β-catenin accumulation (Fig. 6H) upon WNT3a stimulation. Overall, our results show that METTL3-dependent m6A RNA modification is required for WNT signaling in Xenopus and in human cells.

DISCUSSION

In this study, we have investigated functions of Mettl3 beyond gastrulation stages. We found that Mettl3 is required for the proper spatial neural patterning along the anteroposterior (A-P) axis of the Xenopus embryo. Mettl3 is also required for the specification of neural crest and neuronal differentiation. We further addressed that WNT signaling may be the underlying pathway of spatial neural patterning regulated by METTL3-dependent mRNA m6A modification. In all, this study proposes that METTL3-dependent m6A RNA modification regulates proper spatial neural patterning during Xenopus embryogenesis.

WNT signaling functions as a morphogen in the vertebrate A-P axis patterning (19). The activity gradient of canonical WNT along the A-P axis of the neural plate specifies and patterns the central nervous system into forebrain, midbrain, hindbrain, and spinal cord. During the initial stages of neural patterning, the posterior-most neural tissues (i.e., spinal cord) are specified in the WNT-highest region, the anterior-most neural tissues (i.e., forebrain) in the WNT-lowest region, and the mid- and hindbrain tissues in between, according to the activity gradient of the WNT signaling along the A-P axis (26). Multiple WNT ligands and WNT antagonists are expressed and secreted to establish the extracellular WNT gradient (19, 27–30). As expected, overexpression of WNT antagonists produce anteriorized embryos, from which the gene name dkk (dickkopf, which means big head in German) or bighead was originated (27, 28). Moreover, intracellular WNT signaling regulators also help to shape the A-P activity gradient of WNT, including kinases (e.g., casein kinase 1 and its subunit DDX3), receptors (e.g., LRP6), transcription factors, signaling adaptors (e.g., DP1) and vesicle trafficking regulators (e.g., TFG) (31–33). In this study, we extended the repertoire of intracellular regulators of A-P patterning by uncovering the role of RNA m6A methyltransferase Mettl3 in Xenopus A-P neural patterning and in WNT signaling.

It is also well established that WNT signaling has a critical role in dorsoventral axis specification (34). For example, maternal WNT11 signaling specifies dorsal axis by inducing organizer genes, such as siamois, goosecoid, and chordin (35). On the other hand, zygotic WNT8 signaling is activated on the ventral mesoderm and is required for the ventral axis patterning (36). Despite these critical roles of WNT signaling in primary dorsoventral axis specification, we found no significant dorsoventral axis defects in Mettl3 morphants but a slight dorsalization (Fig. 2A and data not shown). Indeed, we found that the domain of chordin expression (i.e., organizer) was almost unchanged while the intensity of chordin gene expression was slightly reduced (Fig. 6I), suggesting a partial inhibition of maternal WNT signaling by Mettl3 MO. This is in line with the results showing that WNT signaling is only partially inhibited by METTL3 knockdown in human cells (Fig. 6C to H). Moreover, in Xenopus animal cap explants, the WNT reporter activity induced by the ectopic Xwnt8 mRNA was also partially inhibited by the Mettl3 MO (Fig. 6J). Therefore, we propose that WNT signaling may be one of the underlying pathways of Mettl3 MO-induced neural patterning defects, but we cannot exclude the possibility that other Mettl3 targets may contribute to the Xenopus spatial neural patterning as well.

Knockdown of factors in the m6A-dependent mRNA degradation pathway, such as Ythdf1, Ythdf2, Lsm1, or Tut7, phenocopied the neural patterning defects of Mettl3 morphants (Fig. 3 and 4). On the other hand, knockdown of microRNA biogenesis factor Dicer1 showed pan-neural inhibition phenotypes (Fig. 5A) that are distinct from Mettl3 morphant phenotypes of anterior neural expansion and posterior neural inhibition (Fig. 2). Although there are certain similarities between Mettl3 MO and Dicer1 MO in neural patterning, such as the reduction in the posterior neural tissues (e.g., hindbrain marker en2), neuron (n-tubulin), and neural crest (twist), we speculate that the reductions of these markers in Mettl3 morphants were caused by the anteriorization of the neural plate (Fig. 2B and Fig. 4A), while the same changes in Dicer1 morphants were due to the impaired neural induction (Fig. 5A). Therefore, these results suggest that m6A-dependent mRNA regulation is important for the A-P patterning while microRNA-dependent mRNA regulation is important for the neural induction, highlighting the specificity of mRNA m6A pathway in the spatial neural patterning of the Xenopus embryo.

MATERIALS AND METHODS

Plasmids, oligonucleotides, antibodies, and reagents.

The coding region of Xenopus mettl3.L gene was PCR amplified, cloned into the pEN_TTmcs (Addgene; 25755), and recombined into the pCS2-6myc gateway destination vector using GATEWAY LR Clonase II enzyme mix. Short hairpin RNA (shRNA) oligonucleotides against human YTHDF2 were synthesized (Cosmogenetech) as follows: shYTHDF2 F (for pSLIK cloning), 5′-AGCGATGGCTATTGGGAACGTCCTTTAGTGAAGCCACAGATGTAAAGGACGTTCCCAATAGCCAA-3′, and shYTHDF2 R (for pSLIK cloning), 5′-GGCATTGGCTATTGGGAACGTCCTTTACATCTGTGGCTTCACTAAAGGACGTTCCCAATAGCCAT-3′.

The oligonucleotides were annealed and cloned first into pEN_TTmiRc2 (Addgene; 25752) and then recombined to pSLIK-zeo (Addgene; 25736) or pSLIK-neo (Addgene; 25735) vector using GATEWAY LR Clonase II enzyme mix. pEN_TTmcs, pEN_TTmiRc2, pSLIK-zeo, and pSLIK-neo were from Iain D. C. Fraser (NIH, USA); pCS2-6myc gateway destination vector was from Edward M. De Robertis (UCLA, USA).

Control shRNA in the pLKO.puro backbone was from David Sabatini (MIT, USA; Addgene; 1864). shRNA oligonucleotides against human METTL3 and YTHDF2 were synthesized (Cosmogenetech) and ligated to pLKO1.puro (Addgene; 8453). Sequences of shRNA oligonucleotides are as follows: shMETTL3#1 F, 5′-CCGGCGTCAGTATCTTGGGCAAGTTCTCGAGAACTTGCCCAAGATACTGACGTTTTTG-3′; shMETTL3#1 R, 5′-AATTCAAAAACGTCAGTATCTTGGGCAAGTTCTCGAGAACTTGCCCAAGATACTGACG-3′; shMETTL3#2 F, 5′-CCGGGCCAAGGAACAATCCATTGTTCTCGAGAACAATGGATTGTTCCTTGGCTTTTTG-3′; shMETTL3#2 R, 5′-AATTCAAAAAGCCAAGGAACAATCCATTGTTCTCGAGAACAATGGATTGTTCCTTGGC-3′; shYTHDF2#1 F, 5′-CCGGGCTACTCTGAGGACGATATTCCTCGAGGAATATCGTCCTCAGAGTAGCTTTTTG-3′; and shYTHDF2#1 R, 5′-AATTCAAAAAGCTACTCTGAGGACGATATTCCTCGAGGAATATCGTCCTCAGAGTAGC-3′.

Primary antibodies used were anti-β-catenin (Santa Cruz; sc-7963), anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH; Santa Cruz; sc-32233 and sc-25778), anti-METTL3 (Abcam; ab195352), and anti-YTHDF2 (Abcam; ab170118).

Recombinant murine WNT3a (PeproTech; 315-20) and human R-spondin 1 (R&D Systems; 4645-RS-025) were purchased and used. Doxycycline hyclate (Sigma; D9891) was dissolved in water and used at a concentration of 1 μg/ml to induce shRNA expression in HeLa cells.

Xenopus embryo manipulation.

Xenopus laevis was obtained from the Korean Xenopus Resource Center for Research and from Nasco (WI). Embryo collection, microinjection, and in situ hybridization were performed as described previously (37). Developmental stages were determined as previously described (38). Sequences of morpholinos are as follows or described previously (22, 39): control MO, 5′-CCTCTTACCTCAGTTACAATTTATA-3′; Mettl3 MO, 5′-TATCTGCTGCTTCTGTGCAATTACC-3′; Ythdf1 MO, 5′-GCAGACATCCTTCATGAAGAGCGGG-3′; Ythdf2 MO, 5′-CTCCAGAAGACTACTGGCGGACATG-3′; Lsm1 MO, 5′-TGGCGGTCCCGGGCATATAATTCAT-3′; Tut7.L MO, 5′-CTTGTTTGGCAGCATCCCCCATTTT-3′; Tut7.S MO, 5′-TGGCTTGTTTGGCAGCATCCCCCAT-3′; and Dicer1 MO, 5′-CTGCAAAATGCAGGGCTTTCATAAA-3′.

Xenopus maintenance and embryo experiments were carried out with protocols approved by Seoul National University Institutional Animal Care and Use committees (Seoul, Republic of Korea).

In situ hybridization.

Whole-mount in situ hybridization was performed as described (39). Briefly, Xenopus neurulae or tailbud embryos were fixed in MEMFA (0.1 M morpholinepropanesulfonic acid [MOPS; pH 7.4], 2 mM EGTA, 1 mM MgSO₄, and 4% formaldehyde) for 20 min before LacZ lineage tracing staining, or for 4 h. Fixed embryos or explants were dehydrated in ethanol and stored at −20°C until used. After rehydration with PBST (phosphate-buffered saline [PBS] containing 0.1% Tween 20), samples were treated with 1 μg/ml of proteinase K, postfixed with 4% paraformaldehyde/0.1% glutaraldehyde, and prehybridized with hybridization buffer (50% deionized formamide, 5× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 1 mg/ml of yeast tRNA, 100 μg/ml of heparin, 0.1% CHAPS, 10 mM EDTA, and 0.1% Tween 20) at 65°C for 4 to 6 h before hybridization. Digoxigenin (DIG)-labeled antisense RNA probes were synthesized in vitro and used for hybridization overnight at 65°C. The hybridizations were detected by using alkaline phosphatase-conjugated anti-DIG antibody (Roche), stained with BM purple (Roche) as a substrate, and photographed under a Leica S8 APO stereomicroscope (Leica Microsystems). Constructs for in vitro probes synthesis of Xenopus sox2, rx2a, otx2, en2, krox20, twist, delta-1, n-tubulin, and szl were gifts from Edward M. De Robertis (UCLA, USA) and Jin-Kwan Han (POSTECH, Republic of Korea). Antisense probes for Xenopus mettl3, ythdf1, and ythdf2 were synthesized using PCR-amplified cDNAs as templates.

RT-qPCR.

Total RNAs were extracted with TRIzol (Invitrogen) and treated with DNase I (TaKaRa). Total RNAs (200 to 500 ng) were reverse transcribed using Primescript RT master mix (TaKaRa). Quantitative PCR was performed using Power SYBR green PCR master mix (Thermo Fisher Scientific) under a StepOnePlus real-time PCR system (Thermo Fisher Scientific). qPCR primer sequences are listed in Table 1.

TABLE 1.

qPCR primers used in this study

Target	Forward sequence	Reverse sequence
odc	CAGCTAGCTGTGGTGTGG	CAACATGGAAACTCACACC
H4	GACGCTGTCACCTACACCGAG	CGCCGAAGCCGTAGAGAGTG
xag	GGAGAAAGTATTGAGTGGGTACAG	CAGGTGGTGAATCACCATCAG
bf1	ATCATCACCAGCAGCAGC	TGTCCAGCTCGTCCTCTTC
otx2	ATGGGCTCAGCCTGACTAC	AGTGGTCCTTTCCCTCCTC
pax6	CGACCCTGCGACATTTCTC	GGATCGATCCGGTCTCGTAATATC
en2	AAGAATAAACCACCAGGATGAGC	GGACATTGACTCGGTGGTG
krox20	AATGACAAGCGGTCGCTG	CAATGGAGATCTTGCCCATATAGG
pax2	ACCTGATGTGGTGAGACAGAG	ACACATCCATGGCTGACC
n-tubulin	ATGCTGATCTACGCAAAC	AGATAGCAGCTACTGTGAG
twist	ATGGCCAGCTGCAGTTATGT	ATTGTCTGTGTGTGTGGCCT
foxd1	ACAACCTGTCCCTCAACGAC	CTCCCTAAGCACAAGCTCCG
xbra	GCTGGAAGTATGTGAATGGAG	TTAAGTGCTGTAATCTCTTCA
chordin	GTTGTACATTTGGTGGGAA	ACTCAGATAAGAGCGATCA
sizzled	GCCTTTGACATTGGATTATCCAC	TTTGGCAACCGCATCTCC
dkk1	ATCACCCCTCGGCTCTGTAA	CCTCCCTTGCTTAGTCGCTC
axin2	GTTCACCATGGATTCTCAGCAC	CATCTGGAGAAGCGGGAATC
fz1	GGCAGTATAACGGCGAGAAAG	GGTTGTAGGCGATATCGGTG
fz2	TCTGCACAGACATTGCCTAC	TAGAACTGGTGGACCTCCAG
fz3	GTCGCAAGTTATGTCAGAGAGC	AAAGCGGCTACATTCCATCTC
fz4	TGCACGGAGAAGATTAACATCC	CCAAACTCCTTCAGCACCG
lef1	AGATCTATGCCGAGATCAGCAAC	GGGATAATCTCAGTCTCATTGACC
ctnnb1	AATCAACGCTCTTGCCACCA	TTGCAGCTCCCGAATGATGT
dvl2	AGATCTCAGATGACAATGCCAAG	CAGAATCCGGCTGAGAACC
gsk3b	CCAGGGGCCCGACAGACAGC	GGCCTGGTACACCACACCA

Cell culture.

Lenti-X 293T (Clontech; 632180), HeLa, and 293T cells were cultured in Dulbecco modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS). Plasmid DNAs were transfected using FuGENE HD (Promega; E2312) according to the manufacturer’s protocol. Lentiviral supernatants were produced using Lenti-X Packaging Single Shots (vesicular stomatitis virus G protein [VSV-G]; Clontech; 631275) mixed with pLKO.1 puro-based or pSLIK-based lentiviral constructs transfected to Lenti-X 293T cells. Twenty-four hours after transfection, supernatants were collected and concentrated using Lenti-X concentrator (Clontech; 631231). Stable shRNA-expressing HeLa and 293T cell lines were established by infecting cells with shRNA-containing lentiviral supernatants together with 4 μg/ml of protamine sulfate for 24 h. After 24 to 48 h of culture with fresh medium, cells were selected with medium containing 1 to ∼2 μg/ml of puromycin (AG Scientific, P-1033) for 2 to 3 days or with medium containing 0.2 mg/ml of Zeocin (Invitrogen; R25001) or 0.2 mg/ml of Geneticin (Invitrogen; 10131027) for ∼2 weeks.

Western blotting.

For Western blots, cells were lysed and the lysates were separated on Novex WedgeWell Tris-glycine minigels (Thermo Fisher Scientific) and transferred to a polyvinylidene fluoride membrane (GE Healthcare) or to a nitrocellulose membrane (Amersham). Blots were washed with PBST, blocked with 3% bovine serum albumin (BSA) for 1 h at room temperature, and incubated with primary antibodies overnight at 4°C. Membranes were then washed three times with PBST and incubated with secondary antibodies for 1 h at room temperature. Secondary antibodies were from Li-Cor (IRDye 680 and IRDye 800, 1:10,000) for infrared imaging of proteins using the Li-Cor Odyssey system.

Luciferase assay.

Luciferase activities were measured using a dual-luciferase assay system (Promega; E1980) according to the manufacturer’s protocol. WNT reporter plasmid (TOP-Flash) or control reporter (FOP-Flash) together with cytomegalovirus (CMV)-Renilla plasmid was transiently transfected to shRNA-expressing cells. After 24 h of transfection, cells were treated with recombinant WNT3a and RSPO-1 (WR) for an additional 14 h and harvested for luciferase activity measurement.