Developmental Biology 399 (2015) 164–176

Contents lists available at ScienceDirect

Developmental Biology

journal homepage: www.elsevier.com/locate/developmentalbiology

The requirement of histone modiﬁcation by PRDM12 and Kdm4a for the development of pre-placodal ectoderm and neural crest in Xenopus

Shinya Matsukawa a, Kyoko Miwata b, Makoto Asashima b, Tatsuo Michiue a,n a Department of Sciences (Biology), Graduate School of Arts and Sciences, University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan b Research Center for Stem Cell Engineering National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba City, Ibaraki, Japan article info abstract

Article history: In vertebrates, pre-placodal ectoderm and neural crest development requires morphogen gradients and Received 6 September 2014 several transcriptional factors, while the involvement of histone modification remains unclear. Here, we Received in revised form report that histone-modifying factors play crucial roles in the development of pre-placodal ectoderm 21 November 2014 and neural crest in Xenopus. During the early neurula stage, PRDM12 was expressed in the lateral pre- Accepted 23 December 2014 placodal ectoderm and repressed the expression of neural crest specifier genes via methylation of Available online 6 January 2015 histone H3K9. ChIP-qPCR analyses indicated that PRDM12 promoted the occupancy of the trimethylated Keywords: histone H3K9 (H3K9me3) on the Foxd3, Slug, and Sox8 promoters. Injection of the PRDM12 MO inhibited fi Histone modi cation the expression of presumptive trigeminal placode markers and decreased the occupancy of H3K9me3 on PRDM12 the Foxd3 promoter. Histone demethylase Kdm4a also inhibited the expression of presumptive Kdm4a trigeminal placode markers in a similar manner to PRDM12 MO and could compensate for the effects Pre-placodal ectoderm Neural crest of PRDM12. ChIP-qPCR analyses revealed that promotion of the occupancy of H3K9me3 on the Foxd3, Xenopus Slug, and Sox8 promoters was inhibited by Kdm4a overexpression. Taken together, these data indicate that histone modification was essential for pre-placodal ectoderm and neural crest development. & 2015 Elsevier Inc. All rights reserved.

Introduction the competence to form pre-placodal ectoderm (Pieper et al., 2012). Indeed, Dlx3 overexpression reduced the expressions of In the vertebrate embryo, the neural plate border is thought to neural plate genes, neural plate border genes, and neural crest be formed by the intermediate activation of bone morphogenetic genes, whereas the injection of Dlx3 morpholino oligomer (MO) protein (BMP) signaling to induce several neural plate border genes expanded their expression (Pieper et al., 2012). (Dincer et al., 2013; Reichert et al., 2013; Tribulo et al., 2003). The On the other hand, a previous report showed that a neural plate border region then further separates into the pre-placodal ectoderm border gene, Msx1, was important for neural crest specification and neural crest, with the former segregating into individual cranial (Monsoro-Burq et al., 2005). A gain-of-function experiment using placodes, such as the adenohypophyseal, olfactory, lens, trigeminal, the Msx1-GR construct showed that Msx1 could induce the lateral line, otic, and epibranchial placodes (reviewed in Baker and expression of several neural crest specifier genes (Tribulo et al., Bronner-Fraser, 2001; Saint-Jeannet and Moody, 2014; Schlosser, 2003). Another study suggested that some neural crest cells could 2010). In contrast, the neural crest differentiates into the peripheral be induced by co-expression of Msx1 and Pax3 in Xenopus animal nervous system, adrenal medulla, melanocytes, facial cartilage, and cap cells in the presence of Wnt and fibroblast growth factor (FGF) tooth dentine (reviewed in Grenier et al., 2009; Minoux and Rijli, signaling (Monsoro-Burq et al., 2005). In the animal cap cells 2010; Steventon et al., 2014; Theveneau and Mayor, 2012). injected with Noggin, Msx1 overexpression could also activate the Pre-placodal ectoderm development requires the expression of expression of Pax3 and Zic1, which play important roles in the a neural plate border gene, Dlx3 (Hans et al., 2004; Kaji and induction and formation of functional neural crest cells (Milet Artinger, 2004; Woda et al., 2003). Dlx3 is regulated by BMP et al., 2013; Monsoro-Burq et al., 2005; Sato et al., 2005). signaling and promotes the expression of pre-placodal ectoderm A boundary formation between the pre-placodal ectoderm and specifier genes such as Six1 and Eya1, known as pan-placodal neural crest may rely on the expression levels of Dlx and Msx, markers (Woda et al., 2003). A recent study also indicated that whose proteins can bind to each other and form an inhibitory Dlx3 expression in nonneural ectoderm was required for acquiring complex (Zhang et al., 1997). Pre-placodal ectoderm genes could be induced by the high transcriptional activity of Dlx (Pieper et al., 2012), whereas neural crest genes were induced by that of Msx n Corresponding author. (Monsoro-Burq et al., 2005). These mutual inhibitions may con- E-mail address: [email protected] (T. Michiue). tribute to the boundary formation between early pre-placodal http://dx.doi.org/10.1016/j.ydbio.2014.12.028 0012-1606/& 2015 Elsevier Inc. All rights reserved. S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176 165 ectoderm and neural crest. However, a recent study showed that 53%, n¼49; compare Fig. 2B with C). As melanocytes are derived neural crest region did not expand to the pre-placodal ectoderm from neural crest cells, whether the injection of PRDM12 also region in the embryo injected with Dlx3 MO (Pieper et al., 2012), inhibited early neural crest differentiation was tested. At the early and thus another mechanism likely determines the boundary. neurula stage, PRDM12 overexpression significantly reduced the A recent study implicated the epigenetic factor Kdm4a, which expression of several neural crest markers such as Foxd3, Slug, and acts as a histone demethylase, was expressed in the neural crest Sox9 (repression in 73%, n¼26, 75%, n¼36, 67%, n¼24, respec- region of chick embryos (Strobl-Mazzulla et al., 2010). We hypothe- tively Fig. 2D–I), whereas expression patterns of the pre-placodal sized that histone modification for transcriptional regulation was markers Islet1 and Six1, and neural marker Sox2, were not affected required for pre-placodal ectoderm and neural crest patterning, and (no repression in 86%, n¼22, 94%, n¼33, 95%, n¼21, respectively focused on histone-modifying enzymes expressed in pre-placodal Fig. 2J–O). A recent study revealed that neural crest cells were ectoderm. To investigate which histone methyltransferases could be induced by both antagonizing BMP and promoting late Wnt involved with pre-placodal ectoderm development, we performed a signaling (Sato et al., 2005). Thus, neural crest cell induction can β-catenin-dependent microarray analysis (Tanibe et al., 2008)and be mimicked by co-injection of Chd and Wnt8 mRNAs in animal identified PRDM12, a member of the PRDM family of histone cap cells (Sato et al., 2005). RT-PCR indicated that PRDM12 over- methyltransferase factors. PRDM12 was expressed in the parietal expression in Chd- and Wnt8-injected animal cap cells inhibited region of the central nervous system and cranial placode in the expression of several neural crest markers in a dose- zebrafish and mouse embryos (Kinameri et al., 2008; Sun et al., dependent manner (Fig. 2P, lanes 4–6). PRDM12 contains a PR/ 2008), and regulated cell proliferation in P19 cells (Yang and SET domain that is probably related to histone methylation and Shinkai, 2013). However, the importance of PRDM12 in embryonic multiple zinc-finger domains (Fig. 2A). Generally, histone methyl- development remains largely unknown. In this study, we demon- transferase acts either as a positive regulator of transcription by strate the role of PRDM12 during early Xenopus development and methylating histone H3 at lysine 4 (H3K4) or as a negative the requirement of histone modification by PRDM12 and Kdm4a for regulator via methylation of histone H3 at lysine 9 (H3K9) or pre-placodal ectoderm and neural crest development. lysine 27 (H3K27) (reviewed in Zentner and Henikoff, 2013). To test whether PRDM12 acts as a positive or negative regulator, Western blotting analysis using each anti-trimethylated lysine Results antibody was performed, which indicated that PRDM12 induced the trimethylation of H3K9 (H3K9me3) in a dose-dependent PRDM12 is expressed in the lateral pre-placodal ectoderm after the manner, but not H3K4 and H3K27 (Fig. 2Q). These results demon- gastrula stage and is regulated by BMP and late Wnt signaling strated that PRDM12 played an inhibitory role in neural crest gene expression and promoted the H3K9me3. Following a previous First, the expression of Xenopus PRDM12 was investigated by study (de Souza et al., 1999), two fusion constructs, VP16-PRDM12 reverse transcription-polymerase chain reaction (RT-PCR) and and EnR-PRDM12, were made (Supplementary Fig. S2A). Micro- whole-mount in situ hybridization (WISH) analyses. PRDM12 was injection with VP16-PRDM12 increased the number of melanocytes zygotically expressed after the late gastrula stage (Fig. 1A) and was and induced several neural crest markers in animal cap cells detected at the lateral pre-placodal ectoderm in the early neurula (Supplementary Fig. S2B, C and F). In contrast, EnR-PRDM12 embryo (Fig. 1B). In the mid-neurula stage, PRDM12 expression decreased pigmentation and repressed several neural crest mar- was also observed in the two stripes of neural plate (Fig. 1C) and kers in Chd- and Wnt-injected animal cap cells, as observed in the was partially co-localized with the lateral expression regions of embryos or animal cap cells injected with PRDM12-WT (Supple- Six1, Pax3, Islet1, and Pax6 (Fig. 1D and E; Supplementary Fig. S1A– mentary Fig. S2D, E and G). Because the effects of EnR-PRDM12 C), but not Foxd3 and Six3 (Fig. 1F and G; Supplementary Fig. S1D). were similar to those of PRDM12-WT, PRDM12 could act as a Wnt and BMP signaling alter the expression of pre-placodal negative regulator for the target gene transcription. These findings ectoderm and neural crest genes (Matsuo-Takasaki et al., 2005; showed that the zinc-finger domains of PRDM12 were required for Sato et al., 2005). To test whether these signals also regulated its repression activity. PRDM12 expression, PRDM12 expression in the embryos regulating the signals was investigated. In embryos injected with Chordin, PRDM12 promotes the occupancy of H3K9me3 on the Foxd3, Slug, PRDM12 was ectopically expressed in the ventral epidermal region and Sox8 promoters (ectopic expression in 79%, n¼29; Fig. 1H, J and L). In contrast, BMP4 overexpression resulted in increased PRDM12 expression Based on the high levels of conservation of the PRDM12 zinc- (increase in 96%, n¼28; Fig. 1N). In the posteriorized embryo finger domains (Supplementary Fig. S3), we hypothesized that a injected with pCS2-Wnt3a, the expression of PRDM12 was target sequence of PRDM12 would also be conserved. Using the repressed in the anterior region (repression in 96%, n¼28; VISTA browser, a conserved noncoding sequence (CNS) around Fig. 1P), whereas anteriorized embryos injected with Dkk1 mRNA several neural crest genes was sought by comparing 7 kb of each showed increased PRDM12 expression toward the posterior genomic region among Xenopus and other species. In the upstream (increase in 93%, n¼27; Fig. 1R). These effects were similar to region of Foxd3, a 48-bp CNS (Fig. 3A) was identified. To test those of Pax3 (Fig. 1I–S), a neural plate border specifier gene whether this sequence actually functioned, luciferase reporter (Plouhinec et al., 2013). Our data indicated that PRDM12 expres- assays were performed using the Foxd3 promoter fragment includ- sion was regulated by both BMP and Wnt signaling in Xenopus ing the CNS region. When this pFoxd3-2000-luc construct was embryos. injected dorsally, high luciferase activity was observed (Fig. 3B). To investigate whether the inhibition of Foxd3expression by PRDM12 PRDM12 overexpression inhibits neural crest formation via the overexpression was dependent on the CNS in the Foxd3 promoter, trimethylation of histone H3K9 a further luciferase assay was performed using each deletion construct. When pFoxd3-2000-luc or pFoxd3-1647-luc was co- Gain-of-function experiments were performed to clarify the injected with PRDM12, inhibition of luciferase activity occurred. role of PRDM12 in early Xenopus development. Injection of Conversely, in the case of co-injection of the other constructs and PRDM12 into the animal pole of two-cell-stage embryos caused PRDM12, luciferase activity was recovered (Fig. 3C), suggesting small head formation and defective melanocytes (suppression of that the inhibition of Foxd3 transcription by PRDM12 was 166 S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176

Fig. 1. Histone methyltransferase PRDM12 is expressed in the lateral pre-placodal ectoderm and regulated by Wnt and BMP signaling. (A) Stage RT-PCR analysis showing that PRDM12 mRNA is zygotically expressed after stage 13. Chordin and ODC were used as internal controls. (B–H) WISH analysis of PRDM12 (B, C and H), Six1 (D), and Foxd3 (F) expression at the indicated embryo stages. (E and G) Double in situ hybridization revealed partial co-localization of the PRDM12 transcripts (dark blue) with Six1 transcripts (light blue; E), but not with Foxd3 transcripts (light blue; G). The PRDM12 expression region is indicated by a red dotted outline in E and G. (H–S) Effects of BMP and Wnt signaling on PRDM12 and Pax3 expression at the neurula stage were analyzed by injecting Chordin mRNA (100 pg/embryo), BMP4 mRNA (20 pg/embryo), pCS2-Wnt3a (50 pg/embryo), and Dkk1 mRNA (10 pg/embryo) into the animal pole at the two-cell stage. (H and I) Dorsal view of the control embryo. (J–M) PRDM12 and Pax3 transcripts were expressed ectopically on the ventral side in embryos injected with Chordin mRNA (M, 76%, n¼33). (N and O) BMP4 mRNA reduced PRDM12 and Pax3 expression (O, 100%, n¼24). (P and Q) Posteriorized embryos caused the defective expression in the anterior part (Q, 91%, n¼33). (R and S) Anteriorized embryos shifted the expression to the posterior region (S, 68%, n¼31). YP; Yolk Plug. Scale bars represent 0.5 mm. S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176 167

Fig. 2. Injection of PRDM12 mRNA inhibits neural crest formation via trimethylation of histone H3K9. (A) The PRDM12 protein structure showing a PR/SET domain (black box) and multiple zinc-ﬁnger domains (gray boxes). (B) A control embryo at stage 38. (C) An embryo injected with PRDM12 mRNA (1000 pg/embryos) into the animal pole at the two-cell stage showing the small head and pigmentation defects. (D–O) WISH analysis of expressed transcripts at the mid-neurula stage, with anterior views of each embryo. (D–I) The expression of neural crest markers (Foxd3, Slug, and Sox9) was diminished in embryos injected with PRDM12 mRNA (1000 pg/embryo) at the two-cell stage. In contrast, (J–O) the expression of the presumptive trigeminal placode (Islet1), early pan-placode (Six1), and neural marker (Sox2) remained largely unchanged. (P) RT- PCR analysis showed that expression of the neural crest markers was decreased in a dose-dependent manner by injection of PRDM12 mRNA into Chd and Wnt8 mRNA- injected animal caps. (Q) Western blot analysis with antibodies against trimethylated histones H3K4, H3K9, H3K27, and H3 on samples of Chd and Wnt8 mRNA-injected animal cap cells, with and without PRDM12 mRNA. The scale bars for B and C, and D–O, represent 1 mm and 0.5 mm, respectively. 168 S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176

Fig. 3. PRDM12 promotes trimethylation of H3K9 on Foxd3 promoter regions through binding the CNS. (A) VISTA view of the conserved sequence domains in the genomic region containing the Foxd3 gene. Colored peaks indicate regions of at least 100 bp and with 60% similarity [dark blue, exons; light blue, untranslated regions (UTRs); pink, noncoding regions]. One conserved noncoding sequence (CNS) is shown. (B and C) Luciferase assays using the indicated Foxd3 promoter region. Expression levels were normalized to Renilla luciferase. Error bars represent 7S.D. (n¼3). (B) Luciferase assays on stage 20 embryos that were co-injected at the four-cell stage with pGL3-Control or pFoxd3-2000-luc and pGL-TK (Renilla luciferase vector) into the dorsal or ventral side. (C) The indicated luciferase reporter vector and PRDM12 mRNA were co-injected into the dorsal side of four-cell-stage embryos, which were harvested after injection and subjected to relative luciferase activity measurement. (D–H) ChIP-qPCR analysis using the Chd and Wnt8 mRNA-injected animal cap cells co-injected with or without myc-PRDM12 mRNA. The vertical axes represent the percentage input (ChIP enriched/ input) and the horizontal axes represent the approximate distance from the transcriptional start site (TSS) of each gene. ChIP-qPCR analyses were performed with an anti- myc antibody (D) or anti-H3K9me3 antibody (E–H). The Six1 promoter region (0.5 kbp from the TSS) was used as a negative control. Error bars indicate 7S.E. (n¼3). Differences were considered to be statistically signiﬁcant at pr0.05(*) or pr0.01(**) by Student’s t-test.

dependent on the Foxd3 CNS. To investigate whether PRDM12 S4B, lane 1). Further research investigated whether PRDM12 actually bound to the CNS and methylated histone H3K9 around promoted the occupancy of H3K9me3 at another neural crest the Foxd3 promoter, chromatin immunoprecipitation-quantitative specifier gene promoter. ChIP-qPCR analyses using an anti- PCR (ChIP-qPCR) analyses were performed with anti-myc and anti- H3K9me3 antibody showed that injection of PRDM12 increased H3K9me3 antibodies. Injection of myc-PRDM12 followed by ChIP the occupancy of H3K9me3 at Slug, Sox8, Sox9, and Sox10, whereas analysis with anti-myc antibody showed high occupancy of myc- those of Twist and Six1 were not affected (Fig. 3H; Supplementary PRDM12 in the CNS of the Foxd3 promoter, whereas the occupancy Fig. S5). In particular, the occupancy of H3K9me3 around the Slug of myc-PRDM12 at the 1.0 kbp, 0.5 kbp, and 0 kbp regions and Sox8 promoters was significantly changed by PRDM12 over- from the Foxd3 transcriptional start site did not increase (Fig. 3D). expression (Fig. 3F and G). These results confirmed that PRDM12 ChIP analyses with an anti-H3K9me3 antibody showed increase promoted the occupancy of H3K9me3 on the neural crest specifier occupancy of H3K9me3 in the Foxd3 promoter by PRDM12 over- gene promoters. expression (Fig. 3E). These results suggested that PRDM12 bound to the CNS of Foxd3 and promoted the trimethylation of H3K9 Knockdown of PRDM12 inhibits the expression of presumptive around the Foxd3 promoter region. Moreover, an electrophoretic trigeminal placode markers and expands the neural crest region mobility shift assay (EMSA) to investigate the binding activity of PRDM12 to the CNS region showed a band shift of the CNS probe To assess the requirement for PRDM12 function during early in those extracts containing myc-PRDM12 (Supplementary Fig. development, an antisense MO was designed to knockdown S4A and B), but not in those from embryos injected with PRDM12 PRDM12 (Supplementary Fig. S6). Embryos injected with the depleted zinc-finger domains (PRDM12ΔZNF) (Supplementary Fig. PRDM12 MO into one side of each blastomere at the four-cell

Fig. 4. PRDM12 MO represses the trigeminal placode markers via trimethylation of histone H3K9 on the Foxd3 promoter. (A–E) WISH analysis using the indicated probes. Anterior views of mid-neurula embryos injected with Control MO (20 ng/embryo) or PRDM12 MO (20 ng/embryo) and lacZ mRNA (0.1 ng/embryo) with or without misPRDM12 mRNA (1 ng/embryo) into one side of each blastomere at the four-cell stage. (A–C) Expression of the presumptive trigeminal placode markers (Ath3, EBF3, and Islet1) decreased in the PRDM12 MO-injected embryo, whereas the reduction of these markers was rescued by injection of morpholino-resistant PRDM12 (misPRDM12). In contrast, (D and E) expression of the early pan-placodal marker (Six1) and the neural marker (Sox2) were unchanged. (F) Quantitation of the phenotype percentage in the embryos injected with MOs and mRNA. The injected embryos separated into two phenotypes (no effect or severe defect). Quantification of PRDM12 MO- and misPRDM12 mRNA-injected embryos in Ath3 (n¼24), EBF3 (n¼28), Islet1 (n¼25) compared with PRDM12 MO-injected embryos at neurula stage. po0.0005 by the chi-square test. (G and H) Control MO or PRDM12 MO and lacZ mRNA (0.1 ng/embryo) were sequentially injected into one side of the blastomere from the four-cell to the eight-cell stage. The Slug expression region was expanded anteriorly on the injected side (P, no effect in 100%, n¼27). Scale bars represent 0.5 mm. (I–L) Reverse transcription–quantitative polymerase chain reaction (RT-qPCR) analysis of Ath3 (I), Foxd3 (J), Slug (K), and Sox2 (L) expression was performed on late neurula-stage embryos injected with PRDM12 MO into the animal blastomeres at the four-cell stage. (M and N) ChIP-qPCR analyses of the Foxd3 and Six1 promoter sites were performed using an anti-H3K9me3 antibody. Embryos injected with Control MO or PRDM12 MO were cultured until stage 15, and the pre-placodal ectoderm region was dissected. Error bars indicate 7S.E. (n¼3). S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176 169 stage showed that the expression levels of presumptive trigeminal the expression levels of Six1 (Fig. 4D) and Sox2 (Fig. 4E) were not placode markers such as Ath3 (Fig. 3A), EBF3 (Fig. 4B), and Islet1 affected. A comparison between the effects of PRDM12 MO (Fig. 4C) were markedly diminished at the neurula stage, whereas injection and those of PRDM12 MO and misPRDM12 co-injection 170 S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176 showed that the reduction of presumptive trigeminal placode analyses showed that the high occupancy of H3K9me3 on the marker expression by PRDM12 MO was rescued by co-injection Foxd3, Slug, and Sox8 promoters, but not the Six1 promoter, was of misPRDM12 (Fig. 4F). These results suggested that PRDM12 was suppressed by Kdm4a co-injection (Fig. 6D–G). The dual PHD and required for presumptive trigeminal placode development. To Tudor domains of Kdm4a are known to be selective for binding investigate whether knockdown of PRDM12 also affected neural trimethylated H3K4 (H3K4me3) and H4K20 (Huang et al., 2006; crest formation, WISH analysis was performed on embryos Lee et al., 2008). Thus, we investigated the endogenous methyla- injected with PRDM12 MO and lacZ into one side. A comparison tion levels by ChIP-qPCR using anti-H3K4me3 and anti-H3K9me3 of the PRDM12 MO-injected side with the uninjected side revealed antibodies. In the neural plate border dissected from gastrula an anterior expansion of the Slug expression region (expansion embryos, occupancy of H3K4me3 on the Foxd3 promoter was in 61%, n¼28; Fig. 4G and H). RT-qPCR analysis showed that observed (Fig. 7A), whereas none was detected on H3K9me3 expression of Ath3decreased in a dose-dependent manner (Fig. 4I), (Fig. 7B). Moreover, the occupancy of H3K9me3 on this promoter whereas those of Foxd3 and Slug increased (Fig. 4J and K) and increased in the dissected pre-placodal ectoderm at the neurula Sox2 expression remained unaffected (Fig. 4L). ChIP-qPCR analysis stage (Fig. 7C), but not in the dissected neural crest (Fig. 7D). These using an anti-H3K9me3 antibody revealed that the occupancy of findings suggested that modification of histone H3K9 on the Foxd3 H3K9me3 on the Foxd3 promoter, but not on the Six1 promoter, promoter, which is a common target for PRDM12 and Kdm4a, was decreased in embryos injected with PRDM12 MO (Fig. 4M required for pre-placodal ectoderm and neural crest development. and N). These results suggested that PRDM12 contributed to the regionalization of the pre-placodal ectoderm by inhibiting the expression of several neural crest genes. Discussion

Kdm4a overexpression produces a similar effect to injection of In this study, we showed that histone modification in the PRDM12 MO that can be rescued by PRDM12 neural crest gene promoters contributed to pre-placodal ectoderm and neural crest patterning. Specifically, we found that PRDM12, A recent study revealed that demethylase Kdm4a/Jmjd2a was expressed in the lateral pre-placodal ectoderm, inhibited neural important for neural crest specification in chick development crest formation by promoting histone H3K9me3 on the Foxd3, Slug, (Strobl-Mazzulla et al., 2010). Kdm4a contains a jumonji domain and Sox8 promoters. This H3K9me3 is known to promote chro- required for methyl lysine-specific demethylation activity on matin remodeling to form a silent heterochromatin with hetero- histones H3K9 and H3K36. Thus, we speculated that pre- chromatin protein 1 (HP1), an important factor for the recruitment placodal ectoderm and neural crest patterning also required the of other remodeling factors (Bannister et al., 2001; Jacobs and expression of Kdm4a. RT-PCR and WISH analysis showed that Khorasanizadeh, 2002; Lachner et al., 2001). Thus, aggregated Kdm4a was maternally expressed in the animal hemisphere at the structures of heterochromatin induced by the H3K9me3 on these early blastula stage (Supplementary Fig. S7A and B). At the neurula gene promoters would inhibit the transcriptional activity of these stage, Kdm4a expression was observed at the neural plate, neural genes in the pre-placodal ectoderm (Fig. 7E, left). Conversely, plate border, and neural crest, including the trunk neural crest Kdm4a likely recognized, bound to histone H3K4me3 on the Foxd3 region (Supplementary Fig. S7C and D). These expression patterns promoter, and reduced methylation levels of H3K9 to promote the are similar to that of chick Kdm4a (Strobl-Mazzulla et al., 2010). expression of neural crest specifier genes in the neural crest cells When Kdm4a was injected into the animal pole of two-cell-stage (Fig. 7E, right). In short, modification of the methylation levels of embryos, Ath3 and Islet1 expression was not detected in the histone H3K9 on the neural crest gene promoter by PRDM12 and presumptive trigeminal placode region, as observed in the Kdm4a determined a boundary between the pre-placodal ecto- embryos injected with PRDM12 MO (suppression in 73%, n¼33, derm and the neural crest region. Moreover, we demonstrated that 69%, n¼32, respectively Fig. 5A, B, E and F). Notably, the expres- PRDM12 knockdown inhibited the presumptive trigeminal pla- sion of Ath3 and Islet1 was rescued by co-injection of PRDM12 (no code formation in a similar manner to Kdm4a overexpression. In suppression in 78%, n¼23, 83%, n¼18, respectively Fig. 5B–D and chick development, Kdm4a knockdown inhibits the expression of F–H). Furthermore, the expression of Foxd3 and Slug did not several neural crest genes (Strobl-Mazzulla et al., 2010)ina decrease in the embryos injected with Kdm4a (no suppression in manner similar to the PRDM12 overexpression seen in our study. 84%, n¼32, 94%, n¼33, respectively Fig. 5I, J, M and N), whereas PRDM12 and Kdm4a exhibit a complementary effect in histone the inhibition of neural crest markers by PRDM12 overexpression modification of the neural crest gene promoter. was rescued by co-injection with Kdm4a (no suppression in 80%, Dimethylation of H3K9 can recruit HP1 (Bannister et al., 2001; n¼20, 95%, n¼19, respectively; Fig. 5I–P). In contrast, Sox2 Lachner et al., 2001). Furthermore, proteins containing a SET expression was largely unaffected (96%, n¼28, 92%, n¼24, 96%, domain can regulate multiple types of methylation. For example, n¼25, respectively Fig. 5Q–T). Because the embryos injected with Dim-5 regulates mono-, di-, and trimethylation of histone H3K9 PRDM12 MO and Kdm4a showed similar phenotypical effects, we (Collins et al., 2009). If the PR/SET domain of PRDM12 were assumed that Kdm4a overexpression could cause anterior expan- capable of another methylation activity such as Dim-5, PRDM12 sion of the neural crest region, as observed in the PRDM12 MO- could also have modification activities to promote mono- and injected embryos. To test this hypothesis, WISH analyses were dimethylation of histone H3K9. To evaluate these activities, Wes- performed using embryos injected with Kdm4a and lacZ mRNAs tern blotting and ChIP-qPCR analysis with antibodies against into one side. Comparison of the Kdm4a-injected side with the H3K9me1 or H3K9me2 should be performed. uninjected side revealed an anterior expansion of Foxd3 and Slug We showed that injection of PRDM12 MO or Kdm4a caused expression (expansion in 64%, n¼28, 59%, n¼27, respectively inhibition of the presumptive trigeminal placode gene expression Fig. 5U–X). Furthermore, the inhibition of several neural crest at the early neurula stage and decreased the occupancy of markers by microinjection of PRDM12 was rescued by co-injection H3K9me3 on the neural crest gene promoter. These data indicate with Kdm4a in the animal cap cells (Fig. 6A). In the luciferase assay that PRDM12 and Kdm4a can indirectly regulate pre-placodal using pFoxd3-2000-luc, luciferase activity was weakly enhanced ectoderm development by regulating the expression of neural in the embryos injected with Kdm4a (Fig. 6B). Meanwhile, the crest genes. A previous study showed that Foxd3 overexpression repression of luciferase activity by the PRDM12 injection was could inhibit the posterior placode formation including the pre- partially rescued by Kdm4a (Fig. 6C). Moreover, ChIP-qPCR sumptive trigeminal placode at the early neurula stage (Brugmann S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176 171

Fig. 5. Overexpression of Kdm4a causes a defect of the presumptive trigeminal placode and rescues the repression of neural crest formation by PRDM12.(A–T) WISH analyses using the indicated transcripts. Anterior view of the embryos. (A, E, I, M and Q) Control embryos. (B, F, J, N and R) Embryos injected with Kdm4a mRNA (200 pg/embryo) or (C, G, K, O and S) PRDM12 mRNA (1000 pg/embryo) or (D, H, L, P and T) Kdm4a and PRDM12 mRNAs into the animal poles at the two-cell stage were harvested until stage 15. (A–H) Yellow arrowheads indicate the expression of the trigeminal placode markers Ath3 and Islet1. The expression was reduced by overexpression of Kdm4a mRNA but not by PRDM12 mRNA (C, 93%, n¼28; G, 86%, n¼35). The reduction of Ath3 and Islet1 was rescued by co-injection with PRDM12 mRNA. (I–P) The expression of the neural crest markers Foxd3 and Slug was not repressed by injection of Kdm4a mRNA, whereas these markers were suppressed by injection of PRDM12 mRNA (K, 53%, n¼45; O, 55%, n¼42). The repression of neural crest markers by PRDM12 mRNA was also rescued by Kdm4a overexpression. (Q–T) The expression of the neural plate marker Sox2 was largely unaffected. (U–X) WISH analyses using embryos injected with lacZ mRNA (100 pg/embryo) with or without Kdm4a mRNA into one side at the four-cell stage showing the expression of Slug and Foxd3.(UandW)Control embryos(noexpansionin92%,n¼25, 96%, n¼28, respectively). (V and X) Expansion of Slug and Foxd3 expression in the injected side was observed. Scale bars represent 0.5 mm. 172 S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176

Fig. 6. Kdm4a overexpression compensated for the effects of PRDM12 via demethylation of H3K9 on the neural crest genes promoter. (A) RT-PCR analysis of neural crest marker expression in the animal cap cells injected with Wnt8 and Chordin mRNAs (each at 50 pg/embryo), with or without PRDM12 and Kdm4a mRNA. (B, C) Luciferase assays on embryos injected with pFoxd3-2000-luc (30 pg/embryo), pGL-TK (10 pg/embryo), and Kdm4a mRNA (0, 50, 100, 200 pg/Sembryo) into the dorsal side at the four- cell stage. Protein was prepared from the injected embryos at stage 18. Error bars indicate 7S.D. (n¼3). (D–G) ChIP-qPCR using an anti-H3K9me3 antibody was performed with the Chd and Wnt8 mRNA-injected animal caps with or without injection of PRDM12 mRNA or PRDM12 and Kdm4a mRNAs. The Six1 promoter was used as a negative control. Error bars indicate 7S.E. (n¼3).

et al., 2004). We also demonstrated that expression of the neural Our results showed that microinjection of PRDM12 caused a crest markers expanded to a section of the pre-placodal ectoderm reduction in neural crest gene expression, whereas an expansion region in embryos injected with PRDM12 MO or Kdm4a mRNA; of the pre-placodal ectoderm and neural plate genes was not thus, the knockdown of PRDM12 and the ectopic expression of observed. However, several studies related to ectodermal pattern- Kdm4a could allow Foxd3 expression at the presumptive trigem- ing indicated that repression of one domain caused the expansion inal placode region, and the expression of the presumptive of the adjacent domain: inhibition of neural crest formation trigeminal placode genes would be repressed by Foxd3. However, caused the expansion of pre-placodal ectoderm or neural plate the expansion of the neural crest markers by Kdm4a overexpres- regions to the neural crest region (Bellmeyer et al., 2003; sion was weaker than that by PRDM12 knockdown, which might Brugmann et al., 2004). For specific inhibition of neural crest indicate that PRDM12 and Kdm4a utilize different mechanisms to genes, other histone-modifying enzymes could be required. regulate histone modification. For demethylation on the neural PRDM12 can bind histone methyltransferase G9a/EHMT2 through crest gene promoter, Kdm4a had to recognize H3K4me3 to bind to its zinc-finger domains (Yang and Shinkai, 2013). G9a, which is the gene promoter. In the pre-placodal ectoderm cells, the occu- widely expressed in many tissues, plays an important role in the pancy of H3K4me3 could be decreased by another histone- recruitment of other histone-modifying factors (Ling et al., 2012; modifying enzyme. Ectopic Kdm4a would fail to bind to these Milner and Campbell, 1993). It can bind DNA methyltransferases gene promoters and would not demethylate histone H3K9. Dnmt3a and G9a, with the former cooperatively regulating neu- Another histone-modifying enzyme could be related to the devel- ronal differentiation, including the peripheral nervous system (Rai opmental patterning of the pre-placodal ectoderm and et al., 2010). Dnmt3a is known to act as a molecular switch for neural crest. neural crest formation by inhibiting the expression of Sox2 and Luciferase analyses using the Foxd3 promoter fragment showed Sox3 in the neural plate border cells (Hu et al., 2012), but whether that PRDM12 overexpression could repress luciferase activity and G9a and Dnmt3a are expressed in the early Xenopus embryo is not that this was dependent on the CNS of the Foxd3 promoter. yet known. However, if G9a and Dnmt3a are expressed at the late However, whether histone proteins directly bind to exogenous gastrula stage, PRDM12 could form an inhibitory complex with DNA to form a proper histone–DNA complex such as a nucleosome G9a and Dnmt3a, and that complex could repress the expression remains unclear. A recent study revealed that histone demethylase of neural crest and neural plate genes. A previous study showed LSD1 could repress luciferase expression in a demethylation that nonneural and neural competence in ectodermal cells was activity-dependent fashion (Shi et al., 2004), suggesting that restricted by some transcriptional factors such as Dlx3 and Sox3 PRDM12 also methylates histone proteins bound to the injected until the late gastrula stage (Pieper et al., 2012). After loss of the DNA and promotes the aggregation of these complexes as hetero- nonneural competence in the neural crest region, Six1 expression chromatin in vivo. might not be able to expand into the neural crest region. S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176 173

Fig. 7. The occupancy of H3K4me3 and H3K9me3 in the indicated tissues and a model of pre-placodal ectoderm and neural crest development. (A–D) ChIP-qPCR analyses using mock antibody, anti-H3K4me3, or anti-H3K9me3 on the indicated tissues. Each tissue was obtained using 100 dissected embryos at each stage. (E) The model of pre- placodal ectoderm and neural crest patterning based on our ﬁndings. Occupancy of H3K9me3 on the Foxd3 promoter is not observed in the neural plate border, whereas that of H3K4me3 is apparent. In the pre-placodal ectoderm, PRDM12 is expressed and speciﬁcally promotes the trimethylation of H3K9 on the Foxd3 promoter to bind at the CNS. A chromatin remodeling factor, HP1, recognizes the trimethylation of H3K9 and recruits other factors to convert euchromatin to heterochromatin. In contrast, in the neural crest region, Kdm4a is expressed to maintain the demethylation of H3K9 on the Foxd3 promoter and bind to the trimethylated H3K4.

We showed that injection of PRDM12 MO or Kdm4a mRNA modification. Development of the neurogenic placode might be caused specific inhibition in the expression of the presumptive dependent on the regulation of other histone-modifying enzymes trigeminal placode markers but not the pan-placode marker. as well as PRDM12. Differentiation of neural cells is known to require dynamic histone Analyzing the relationship between PRDM12 and other modification to obtain individual neural cell fates (reviewed in histone-modifying enzymes would be of interest in future ana- Hirabayashi and Gotoh, 2010). Thus, we hypothesize that the lyses. For example, PRDM2, which is a member of the PRDM expression of the neurogenic placode, including the trigeminal family, can regulate the methylation of histone H3K9 for transcrip- placode but not pan-placode, is also regulated by histone tional repression of several genes and act as a transcriptional 174 S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176 activator to bind p300, known as an acetyltransferase (Carling et was performed on a StepOnePlus Real-Time PCR System (Applied al., 2004; Kim et al., 2003). PRDM12 may therefore upregulate Biosystems, Foster City, CA, USA) using a KAPA SYBR Fast qPCR Kit placode specifier genes together with a histone-modifying enzyme (NIPPON Genetics). The expression of each gene was normalized to to promote placode formation. PRDM12 can also repress cell that of ODC. The conditions and the primer sequences for the RT- proliferation by modulating the cell cycle in P19 nerve cells and PCR and the RT-qPCR are described in Table S2. acts as a putative tumor suppressor for chronic myeloid leukemia (Reid and Nacheva, 2004; Yang and Shinkai, 2013). Kdm4a also WISH, double in situ hybridization, and lacZ staining regulates several genes related to cell proliferation and contributes to breast tumor formation (Berry et al., 2012; Kim et al., 2012). To prepare the probes of each transcript, pCS2p-Kdm4a, pCS2p- Based on these findings, the mutual relationship between PRDM12 Sox9, pCS2p-Foxd3, pCS2p-Islet1, pCS2-Pax8, pBluescript-SK- and Kdm4a may contribute to other biological events. PRDM12, pBluescript-SK-Six1, pBluescript-SK-Six3, pBluescript- SK-EBF3, pBluescript-SK-Ath3, pBluescript-SK-Pax3, pGEM-Pax6, pGEM-Slug, and pGEM-Sox2 were linearized with each enzyme Materials and methods to be used as templates, and digoxigenin (DIG)-labeled probes were synthesized with T7 or T3 RNA polymerase (Roche, Basel, DNA construction Switzerland). WISH was carried out according to Harland (1991) with minor modifications. For the double in situ hybridization, a PRDM12 cDNA was inserted into an EcoRV-digested pBlue- DIG-labeled probe was stained with Nitro blue tetrazolium chlor- script-SK or EcoRV-digested pCS2-myc vector (Danno et al., ide (NBT; Roche) and a fluorescein-labeled probe was stained with 2008). A long PCR was then carried out to generate morpholino- 5-bromo-4-chloro-3-indolyl phosphate (BCIP; Roche). Before resistant PRDM12 constructs by replacing the sequence for binding WISH analysis, the embryos injected with LacZ mRNA were stained the antisense morpholino from GAATTAATGATGGGCTCGGTGCTGC with Red-gal (SAFC Biosciences Inc., Lenexa, KS, USA). to GAGCTCATGATGGGAAGTGTATTA (Fig. S6A). A deletion mutant- PRDM12 was generated by a long PCR. Foxd3, Islet1, and Kdm4a Western blotting cDNAs were inserted into a StuI-digested pCS2p vector. For WISH, PRDM12, Sox9, and Pax3 cDNAs were inserted into an EcoRV- Cell lysates were prepared in RIPA buffer [1% NP-40, 1% sodium digested pBluescript-SK vector, and pBluescript-SK-Six1 was deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 150 mM NaCl, digested with EcoRI, blunted using T4 polymerase, and then self- 25 mM Tris–HCl, pH 7.4] with animal caps or whole embryos, and ligated. For the luciferase assays, a Foxd3 promoter fragment measured using a NanoDrop 1000 (Thermo Fisher Scientific, Wal- fi ampli ed by PCR in Xenopus laevis genomic DNA was inserted tham, MA, USA) with a BCA™ protein assay kit (Pierce Biotechnol- into a pGL3 vector (Promega, Madison, WI, USA). To generate ogy Inc., Rockford, IL, USA). Proteins were separated using 10% or several Foxd3 promoter reporter constructs, pFoxd3-1647-luc, 15% SDS–polyacrylamide gel electrophoresis (SDS–PAGE) gels. Wes- pFoxd3-1612-luc, pFoxd3-1577-luc, and pFoxd3-1000-luc were tern blot analysis was performed using anti-histone H3 (Millipore, obtained by PCR, and pFoxd3-1504-luc and pFoxd3-1195-luc were Billerica, MA, USA; 06-755), anti-histone H3K4me3 (Abcam, Cam- digested with PstI and EcoRV, blunted with T4 polymerase, and bridge, UK; ab8580), anti-histone H3K9me3 (Millipore, 07-449), then self-ligated. The primer sequences for all the DNA construc- anti-histone H3K27me3 (Abcam; ab6002), anti-myc (Sigma-Aldrich, tion are listed in Table S1. St. Louis, MO, USA; M4439-100UL), and anti-α-tubulin antibodies (Sigma-Aldrich; T9026-100UL). Embryo manipulation ChIP-qPCR The stages of development referred to in this study are based on the stages of normal Xenopus developmental stages For the ChIP assays, animal cap explants dissected from 100 (Nieuwkoop and Faber, 1967). X. laevis embryos were obtained embryos injected with each mRNA or each MO were harvested at by artificial fertilization and dejellied with 4.6% L-cysteine hydro- stage 15. The explants and embryos were cross-linked in 0.7 chloride at pH 7.8. Synthetic mRNAs were prepared using the phosphate-buffered saline (PBS) with 1% formaldehyde and soni- mMESSAGE mMACHINE SP6 transcription kit (Ambion, Life Tech- cated to digest the genomic DNA into sizes of 400–600 bp. nologies, Carlsbad, CA, USA) for in vitro transcription. pCS2- Samples were incubated for binding with mock control rabbit Chordin (Sasai et al., 1994), pCS2-BMP4 (Nishimatsu et al., 1992), antibody (DakoCytomation, Carpinteria, CA, USA; X0936) or target pCS2-Dkk1, pCS2-myc-PRDM12, pCS2-myc-PRDM12ΔZNF, pCS2- antibodies (anti-myc, anti-H3K4me3, anti-H3K9me3) and Protein myc-misPRDM12, pCS2p-Kdm4a, and pCS2-lacZ (Takahashi et al., A Sepharose Fast Flow (GE Healthcare, Buckinghamshire, UK), and 2000) were linearized with NotI or Asp718 and transcribed with rotated overnight. The complexes were centrifuged, eluted, and SP6 RNA polymerase. Antisense MOs were obtained from Gene reverse cross-linked. The DNA was purified using a QIAquick Tools (Philomath, OR, USA). The control MO used had the follow- purification cleanup kit (Qiagen, Venlo, The Netherlands). The 0 0 ing sequence, 5 -CCTCTTACCTCAGTTACAATTTATA-3 , and the qPCR was performed as described previously and genomic expres- 0 sequence of the PRDM12 MO was 5 -GCAGCACCGAGCCCATCAT- sion was normalized to the ODC fragment. The primer sequences 0 TAATTC-3 . Each MO was suspended in nuclease-free water (Gibco, are listed in Table S2. Life Technologies, Carlsbad, CA, USA). Synthesized mRNAs, MOs, or expression vectors were microinjected using a PLI-100 picoinjector Luciferase reporter assay and EMSA (Harvard Apparatus, Holliston, MA, USA). Luciferase reporter assays were performed using the Dual- RNA isolation and RT-PCR or RT-qPCR Luciferase Reporter Assay System (Promega). To mimic the endogenous conditions, pGL3-control or pFoxd3-luciferase vectors Total RNA was extracted from three embryos or 10 animal caps were linearized by digestion with Asp718. These vectors were using Isogen II (NIPPON Genetics, Tokyo, Japan), and cDNA was co-injected together with a Renilla luciferase plasmid (pGL-TK) synthesized with random primers and SuperScript III (Invitrogen, into Xenopus embryos. The injected embryos were harvested at Carlsbad, CA, USA). ODC and EF1a served as loading controls. qPCR stage 18 and then lysed in Passive Lysis Buffer (Promega) for S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176 175 analysis using a Luminescencer-JNR (ATTO Co., Ltd., Tokyo, Japan). Kim, T.D., Shin, S., Berry, W.L., Oh, S., Janknecht, R., 2012. The JMJD2A demethylase EMSAs were performed using a non-radioimmunoassay (non-RI) regulates apoptosis and proliferation in colon cancer cells. J. Cell. Biochem. 113, 1368–1376. method. Double-stranded probes containing the putative binding Kim, K.C., Geng, L., Huang, S., 2003. Inactivation of histone methyltransferase by site (CNS on the Foxd3 promoter region Fig. S4A) were labeled with mutations in human cancers. Cancer Res. 63, 7619–7623. DIG. Protein extracts were obtained from each mRNA-injected Kinameri, E., Inoue, T., Aruga, J., Imayoshi, I., Kageyama, R., Shimogori, T., Moore, A.W., animal cap and then separated on 4.5% TBE-acrylamide gels. 2008. Prdm proto-oncogene transcription factor family expression and interaction with the notch-hes pathway in mouse neurogenesis. PLoS One 3, e3859. Western blotting was performed using anti-DIG antibody-conju- Lachner, M., O’Carroll, D., Rea, S., Mechtler, K., Jenuwein, T., 2001. Methylation of gated horseradish peroxidase (HRP; DakoCytomation, D5101). histone H3 lysine 9 creates a binding site for HP1 proteins. Nature 410, 116–120. Lee, J., Thompson, J.R., Botuyan, M.V., Mer, G., 2008. Distinct binding modes specify the recognition of methylated histones H3K4 and H4K20 by JMJD2A-tudor. Nat. Struct. Mol. Biol. 15, 109–111. Acknowledgments Ling, B.M., Gopinadhan, S., Kok, W.K., Shankar, S.R., Gopal, P., Bharathy, N., Wang, Y., Taneja, R., 2012. G9a mediates Sharp-1-dependent Inhibition of skeletal muscle differentiation. Mol. Biol. Cell 23, 4778–4785. We thank Dr. N. Sudo for the technical advice and Drs. K. Matsuo-Takasaki, M., Matsumura, M., Sasai, Y., 2005. An essential role of Xenopus Tanegashima and S. Yamashita for their invaluable comments on Foxi1a for ventral specification of the cephalic ectoderm during gastrulation. this manuscript. This work was supported by the Japan Society for Development 132, 3885–3894. Milet, C., Maczkowiak, F., Roche, D.D., Monsoro-Burq, A.H., 2013. Pax3 and Zic1 the Promotion of Science (JSPS; KAKENHI #25127704 and drive induction and differentiation of multipotent, migratory, and functional #26440115). S. M. is also supported by the JSPS for Young neural crest in Xenopus embryos. Proc. Natl. Acad. Sci. U.S.A. 110, 5528–5533. Scientists (#120800000402). Milner, C.M., Campbell, R.D., 1993. The G9a gene in the human major histocompat- ibility complex encodes a novel protein containing ankyrin-like repeats. Biochem. J. 290, 811–818. Minoux, M., Rijli, F.M., 2010. Molecular mechanisms of cranial neural crest cell migration Appendix A. Supporting information and patterning in craniofacial development. Development 137, 2605–2621. Monsoro-Burq, A.H., Wang, E., Harland, R., 2005. Msx1 and Pax3 cooperate to mediate FGF8 and WNT signals during Xenopus neural crest induction. Dev. Supplementary data associated with this article can be found in Cell 8, 167–178. the online version at http://dx.doi.org/10.1016/j.ydbio.2014.12.028. Nieuwkoop, P.D., Faber, J., 1967. Normal Table of Xenopus Embryos. North-Holland, Amsterdam. Nishimatsu, S., Suzuki, A., Shoda, A., Murakami, K., Ueno, N., 1992. Genes for bone References morphogenetic proteins are differentially transcribed in early amphibian embryos. Biochem. Biophys. Res. Commun. 186, 1487–1495. Pieper, M., Ahrens, K., Rink, E., Peter, A., Schlosser, G., 2012. Differential distribution Baker, C.V., Bronner-Fraser, M., 2001. Vertebrate cranial placodes I. Embryonic of competence for panplacodal and neural crest induction to non-neural and – induction. Dev. Biol. 232, 1 61. neural ectoderm. Development 139, 1175–1187. Bannister, A.J., Zegerman, P., Partridge, J.F., Miska, E.A., Thomas, J.O., Allshire, R.C., Plouhinec, J.L., Roche, D.D., Pegoraro, C., Figueiredo, A.L., Maczkowiak, F., Brunet, L.J., Kouzarides, T., 2001. Selective recognition of methylated lysine 9 on histone H3 Milet, C., Vert, J.P., Pollet, N., Harland, R.M., Monsoro-Burq, A.H., 2013. Pax3 and – by the HP1 chromo domain. Nature 410, 120 124. Zic1 trigger the early neural crest gene regulatory network by the direct Bellmeyer, A., Krase, J., Lindgren, J., LaBonne, C., 2003. The protooncogene c-Myc is activation of multiple key neural crest specifiers. Dev. Biol. 15, 461–472. an essential regulator of neural crest formation in Xenopus. Dev. Cell 4, Rai, K., Jafri, I.F., Chidester, S., James, S.R., Karpf, A.R., Cairns, B.R., Jones, D.A., 2010. – 827 839. Dnmt3 and G9a cooperate for tissue-specific development in zebrafish. J. Biol. Berry, W.L., Shin, S., Lightfoot, S.A., Janknecht, R., 2012. Oncogenic features of the Chem. 285, 4110–4121. – JMJD2A histone demethylase in breast cancer. Int. J. Oncol. 41, 1701 1706. Reichert, S., Randall, R.A., Hill, C.S., 2013. A BMP regulatory network controls Brugmann, S.A., Pandur, P.D., Kenyon, K.L., Pignoni, F., Moody, S.A., 2004. Six1 ectodermal cell fate decisions at the neural plate border. Development 140, promotes a placodal fate within the lateral neurogenic ectoderm by functioning 4435–4444. as both a transcriptional activator and repressor. Development 131, 5871–5881. Reid, A.G., Nacheva, E.P., 2004. A potential role for PRDM12 in the pathogenesis of Carling, T., Kim, K.C., Yang, X.H., Gu, J., Zhang, X.K., Huang, S., 2004. A histone chronic myeloid leukaemia with derivative chromosome 9 deletion. Leukemia methyltransferase is required for maximal response to female sex hormones. 18, 178–180. Mol. Cell. Biol. 24, 7032–7042. Saint-Jeannet, J.P., Moody, S.A., 2014. Establishing the pre-placodal region and Collins, R.E., Tachibana, M., Tamaru, H., Smith, K.M., Jia, D., Zhang, X., Selker, E.U., breaking it into placodes with distinct identities. Dev. Biol. 389, 13–27. Shinkai, Y., Cheng, X., 2009. In Vitro and in Vivo analyses of a Phe/Tyr switch Sasai, Y., Lu, B., Steinbeisser, H., Geissert, D., Gont, L.K., De Robertis, E.M., 1994. controlling product specificity of histone lysine methyltransferases. J. Biol. Xenopus Chordin: a novel dorsalizing factor activated by organizer-specific Chem. 280, 5563–5570. homeobox genes. Cell 79, 779–790. Danno, H., Michiue, T., Hitachi, K., Yukita, A., Ishiura, S., Asashima, M., 2008. Sato, T., Sasai, N., Sasai, Y., 2005. Neural crest determination by co-activation of Pax3 Molecular links among the causative genes for ocular malformation: Otx2 and and Zic1 genes in Xenopus ectoderm. Development 132, 2355–2363. Sox2 coregulate Rax expression. Proc. Natl. Acad. Sci. U.S.A. 105, 5408–5413. Schlosser, G., 2010. Making senses development of vertebrate cranial placodes. Int. de Souza, F.S., Gawantka, V., Gómez, A.P., Delius, H., Ang, S.L., Niehrs, C., 1999. The – zinc finger gene Xblimp1 controls anterior endomesodermal cell fate in Rev. Cell Mol. Biol. 283, 129 234. Spemann’s organizer. EMBO J. 18, 6062–6072. Shi, Y., Lan, F., Matson, C., Mulligan, P., Whetstine, J.R., Cole, P.A., Casero, R.A., Shi, Y., Dincer, Z., Piao, J., Niu, L., Ganat, Y., Kriks, S., Zimmer, B., Shi, S.H., Tabar, V., Studer, 2004. Histone demethylation mediated by the nuclear amine oxidase homolog – L., 2013. Specification of functional cranial placode derivatives from human LSD1. Cell 119, 941 953. pluripotent stem cells. Cell Rep. 5, 1387–1402. Steventon, B., Mayor, R., Streit, A., 2014. Neural crest and placode interaction during – Grenier, J., Teillet, M.A., Grifone, R., Kelly, R.G., Duprez, D., 2009. Relationship the development of the cranial sensory system. Dev. Biol. 389, 28 38. between neural crest cells and cranial mesoderm during head muscle devel- Strobl-Mazzulla, P.H., Sauka-Spengler, T., Bronner-Fraser, M., 2010. Histone fi – opment. PLoS One 4, e4381. demethylase JmjD2A regulates neural crest speci cation. Dev. Cell 19, 460 468. Hans, S., Liu, D., Westerfield, M., 2004. Pax8 and Pax2a function synergistically in Sun, X.J., Xu, P.F., Zhou, T., Hu, M., Fu, C.T., Zhang, Y., Jin, Y., Chen, Y., Chen, S.J., Huang, otic specification, downstream of the Foxi1 and Dlx3b transcription factors. Q.H., Liu, T.Q., Chen, Z., 2008. Genome-wide survey and developmental expres- fi Development 131, 5091–5102. sion mapping of zebra sh SET domain-containing genes. PLoS One 3, e1499. Harland, R.M., 1991. In situ hybridization: an improved whole-mount method for Takahashi, S., Yokota, C., Takano, K., Tanegashima, K., Onuma, Y., Goto, J., Asashima, Xenopus embryos. Methods Cell Biol. 36, 685–695. M., 2000. Two novel nodal-related genes initiate early inductive events in Hirabayashi, Y., Gotoh, Y., 2010. Epigenetic control of neural precursor cell fate Xenopus nieuwkoop center. Development 127, 5319–5329. during development. Nat. Rev. Neurosci. 11, 377–388. Tanibe, M., Michiue, T., Yukita, A., Danno, H., Ikuzawa, M., Ishiura, S., Asashima, M., Hu, N., Strobl-Mazzulla, P., Sauka-Spengler, T., Bronner, M., 2012. DNA methyl- 2008. Retinoic acid metabolizing factor xCyp26c is specifically expressed in transferase3A as a molecular switch mediating the neural tube-to-neural crest neuroectoderm and regulates anterior neural patterning in Xenopus laevis. Int. J. fate transition. Genes Dev. 26, 2380–2385. Dev. Biol. 52, 893–901. Huang, Y., Fang, J., Bedford, M.T., Zhang, Y., Xu, R.M., 2006. Recognition of histone Theveneau, E., Mayor, R., 2012. Neural crest delamination and migration: from H3 lysine-4 methylation by the double tudor domain of JMJD2A. Science 312, epithelium-to-mesenchyme transition to collective cell migration. Dev. Biol. 748–751. 366, 34–54. Jacobs, S.A., Khorasanizadeh, S., 2002. Structure of HP1 chromodomain bound to a Tribulo, C., Aybar, M.J., Nguyen, V.H., Mullins, M.C., Mayor, R., 2003. Regulation of lysine 9-methylated histone H3 tail. Science 295, 2080–2083. Msx genes by a Bmp gradient is essential for neural crest specification. Kaji, T., Artinger, K.B., 2004. dlx3b and dlx4b function in the development of Development 130, 6441–6452. Rohon–Beard sensory neurons and trigeminal placode in the zebrafish neurula. Woda, J.M., Pastagia, J., Mercola, M., Artinger, K.B., 2003. Dlx proteins position the Dev. Biol. 276, 523–540. neural plate border and determine adjacent cell fates. Development 130, 331–342. 176 S. Matsukawa et al. / Developmental Biology 399 (2015) 164–176

Yang, C.M., Shinkai, Y., 2013. Prdm12 is induced by retinoic acid and exhibits anti- Zhang, H., Hu, G., Wang, H., Sciavolino, P., Iler, N., Shen, M.M., Abate-shen, C., 1997. proliferative properties through the cell cycle modulation of P19 embryonic Heterodimerization of Msx and Dlx homeoproteins results in functional carcinoma cells. Cell Struct. Funct. 38, 195–204. antagonism. Mol. Cell. Biol. 17, 2920–2932. Zentner, G., Henikoff, S., 2013. Regulation of nucleosome dynamics by histone modiﬁcations. Nat. Struct. Mol. Biol. 20, 259–266.