Mecp2 regulates neural cell differentiation by suppressing the Id1 to Her2 axis in zebrafish

DNA methylation in vertebrates silences gene expression by direct interference with the DNA binding of certain transcription factors, or through the recruitment of inhibitory transcription complexes that bind methylated CpG (Kriaucionis and Bird, 2003; Suzuki and Bird, 2008). Methyl-CpG-binding protein 2 (MeCP2), a member of the methyl-CpG-binding domain family, is a global repressor of transcription (Jones et al., 1998; Nan et al., 1998). Human MeCP2 is an X-linked protein that encodes two splice variants denoted MeCP2A and MeCP2B. MeCP2 contains two evolutionarily conserved domains, the methyl-CpG-binding domain and the transcriptional repression domain. Upon binding to CpG islands, MeCP2 forms an inhibitory transcription complex through interactions of the transcriptional repression domain with cofactors such as Sin3A and histone deacetylase 1. Besides its well-documented repressor activity, MeCP2 also acts as a transcription activator to regulate gene expression through modulating either long-range chromatin remodeling or by regulating RNA splicing (Chahrour et al., 2008; Skene et al., 2010; Young et al., 2005).

Genetic mutations of MeCP2 cause a spectrum of autism disorders in children, and MeCP2 is mutated in >95% of patients diagnosed with Rett syndrome (RTT) (Amir et al., 1999). RTT patients have evident abnormalities during their first 6 to 18 months of life, so RTT is considered to be a disease that involves in disruption of brain development. During early childhood, RTT patients display a progressive regression of acquired skills, followed by mental retardation and loss of psychomotor skills. A fraction of patients also have seizures, as well as abnormal cardiac and respiratory cycles (Chahrour and Zoghbi, 2007). To study the molecular pathogenesis of RTT, many mouse models ranging from null MeCP2 mutations to specific point mutations have been created (Guy et al., 2011). These models closely recapitulate several of the motor and cognitive dysfunctions described in RTT patients, revealing a crucial role for MeCP2 in both neurons and glia. Taken together, these findings in human RTT patients and mouse models suggest that MeCP2 has an essential function in brain development. However, it remains largely unclear how MeCP2 acts in neuronal development and differentiation during the embryonic and postnatal stages.

During the development of the central nervous system (CNS), neural precursors give rise to neurons and astrocytes at specific times, at the right locations and in the correct numbers (Anderson, 2001). Neurons differentiate at early embryonic stages, whereas astrocytes emerge later, and each of these steps is spatially and temporally orchestrated. The zebrafish has become a powerful vertebrate model for studying CNS development (Schmidt et al., 2013). To test the hypothesis that the predisposition to neuronal defects in MeCP2 mutant embryos has an impact on the later motor and cognitive dysfunction described in RTT patients, we designed experiments to investigate mecp2 function during neuronal development in zebrafish.

To explore mecp2 function in embryonic neuronal development in zebrafish, we isolated mecp2 cDNA from embryos at 48 hours post-fertilization (hpf) based on the annotated mecp2 sequence in the Zebrafish Genome Browser at the University of California-Santa Cruz. Zebrafish mecp2 contains three exons encoding 524 amino acids, and shares 43% identity with human and mouse MeCP2 (data not shown). BLASTP searches showed that the zebrafish genome contains only one copy of the mecp2 gene. Synteny analysis revealed that the block of DNA on zebrafish chromosome 8 containing mecp2, irak1 and hcfc1b is conserved with a block of DNA containing the homologs of these genes on human and mouse chromosome X. To investigate the role of mecp2 during zebrafish neuronal development, we developed a Mecp2 antibody that specifically recognized the epitope containing amino acids 170–223 of Mecp2 in zebrafish. This Mecp2 antibody also recognized rat MeCP2 protein (data not shown). In zebrafish, mecp2 was broadly expressed from the fertilized egg to the 24 hpf embryo (Fig. 1A–C), but was enriched in the developing brain at 48 hpf (Fig. 1D), suggesting that mecp2 is required for brain development. By immunostaining the adult zebrafish brain, we found that Mecp2 protein was mainly localized in the nuclei of neurons in which Huc, a neuronal cell marker, was highly expressed (Fig. 1E), suggesting that Mecp2 is located in the majority of differentiated neurons.

Primary neurogenesis in zebrafish ends at ∼48 hpf (Mueller and Wullimann, 2003). At that point, neurons and glia are well differentiated with characteristic and distinct molecular markers (Trevarrow et al., 1990). Given that mecp2 is enriched in the CNS at 48 hpf, we investigated whether it plays a role in neuronal differentiation. An antisense morpholino inhibiting the translation of mecp2 (mecp2-MO) was designed and used for knockdown of mecp2 in zebrafish embryos. A mecp2-mismatch MO was also used as a control. Morpholino-mediated knockdown of mecp2 was confirmed by immunostaining and western blotting with anti-Mecp2 antibody (Fig. 1F,G). Most embryos tolerated and developed normally with 6 ng of mecp2-MO, and higher doses caused apoptosis of the embryonic brain (data not shown). The brain developed normally in mecp2 morphants, without apparent morphological abnormalities in the patterning of the forebrain, midbrain and hindbrain at 48 hpf. We then turned to systematic analysis of the expression of molecular markers for neural precursors, neurons and glia. Nestin, an intermediate-filament protein, is the best-known neural precursor marker in vivo. Nestin-expressing precursors give rise to neuronal or glial lineages by the asymmetric regulation of gene expression (Mahler and Driever, 2007). Nestin-positive cells were localized in the pretectum, subpallium, the posterior area of the midbrain–hindbrain boundary, and in the cluster of cells bilaterally adjacent to the midline of the hindbrain. Nestin expression was also found in the octaval ganglion, the lateral line ganglia, the facial ganglia and the ganglia of the vagus. Interestingly, mecp2 knockdown in the developing zebrafish brain led to an enlarged pool of nestin-positive progenitors. In particular, nestin increased in the pretectum, subpallium and hindbrain in mecp2 morphants (Fig. 2A,A′; supplementary material Fig. S2G–I). To determine the effect of mecp2 knockdown on neural precursor proliferation, we first injected mecp2-MO into Tg(ef1α:mAG-zGem) transgenic embryos in which cells in S, G2 and M phase are marked with green fluorescence (Sugiyama et al., 2009). The number of green fluorescent cells in the brain of mecp2-MO-injected embryos was increased compared with that in wild-type embryos at 48 hpf (Fig. 2B–C′). Second, the number of mitotic cells positive for phosphorylated histone H3 (pH3) increased in mecp2-MO-injected embryos, and these cells extensively overlapped with Tg(ef1α:mAG-zGem) S, G2 and M phase (Fig. 2D–E′). Taken together, these data suggested that mecp2 plays a suppressive role in the proliferation of neural precursors in the developing zebrafish brain.

We then asked whether the increased number of neural progenitors influences the subsequent generation of neurons and glia. Glial fibrillary acidic protein (GFAP) is a well-known marker for radial glial and neural progenitors (Bernardos and Raymond, 2006; Lyons et al., 2003). gfap was elevated in the forebrain, midbrain and hindbrain in mecp2-MO-treated embryos, compared with that in wild-type embryos (Fig. 2F,F′; supplementary material Fig. S2G′–I′). By contrast, expression of microtubule-associated protein 2 (MAP2), which is highly expressed in differentiated neurons (Coolen et al., 2012), decreased in the brain of mecp2 morphants (Fig. 2G,G′; supplementary material Fig. S2G″–I″). We then investigated neuronal differentiation defects in mecp2 morphants. In zebrafish, the postmitotic neuronal cell marker neurod is expressed in the cortex, cerebellum, olfactory bulb, eye and the developing inner ear (Thomas et al., 2012). Using in situ hybridization, we found that expression of neurod decreased in mecp2 morphants (Fig. 2H,H′; supplementary material Fig. S2G‴–I‴), and accordingly, the number of Tg(neurod:GFP)-positive cells also declined (Fig. 2I,I′) in mecp2-MO injected embryos. The amount of the differentiating neuronal marker huc (Park et al., 2000) was reduced in the forebrain, midbrain and hindbrain in mecp2 morphants as assessed by immunostaining (Fig. 2J,J′). We further analyzed the expression pattern of the motor neuron marker islet1 in mecp2 morphants using the transgenic zebrafish line Tg(islet1:GFP) (Higashijima et al., 2000). Analysis of islet1 expression in the cranial nerves of Tg(islet1:GFP) transgenic embryos revealed that motor nerves V, VII and X were reduced in size in mecp2 morphants, suggesting that motor neurons decreased in the developing brain. However, knockdown of mecp2 did not affect motor neuron migration (Fig. 2K,K′). Quantitative real-time PCR substantiated that nestin and gfap were increased whereas map2, neurod, huc and islet1 were decreased in mecp2 morphants (Fig. 2L). These findings indicate that mecp2 promotes neural progenitor fate versus neuronal fates during late brain development. The ectopic differentiation of gfap-expressing neural or glial progenitors in the CNS of mecp2 morphants is likely due to a blockade of neuronal differentiation from progenitors.

Previous studies have demonstrated that the Notch signaling pathway is crucial for maintaining neural stem cells in an undifferentiated state while it promotes astrocyte differentiation (Gaiano and Fishell, 2002; Louvi and Artavanis-Tsakonas, 2006). Given that glial-cell-specific GFAP is a downstream target of Notch in the brain, we explored whether the changes in the neural cell fates in mecp2 morphants were caused by abnormalities in Notch signaling. We chose to examine the expression of several Notch signaling related genes (notch1a, notch1b, deltaA, deltaB and deltaC) in mecp2 morphants. notch1a was restricted to the ventricular proliferation zones of the mesencephalic tectum, anterior dorsal thalamus, entire cerebellar plate, and rhombic lip region. As expected, we found that notch1a and the other Notch signaling genes were upregulated in mecp2 morphants at 48 hpf (supplementary material Fig. S1A–J′). Moreover, the direct target of the notch intracellular domain (NICD), her4.1, which labels proliferating neural precursors (Jung et al., 2012; Takke et al., 1999), was also activated in mecp2 morphants (supplementary material Fig. S1K–L′), suggesting that there is elevated Notch overall signaling in mecp2 morphants, which is consistent with a recent report on MeCP2 function in regulating adult neurogenesis through the Notch signaling pathway in mice (Li et al., 2014).

Many studies have shown that Her4.1 represses pro-neuronal genes in the developing CNS to maintain neural precursor fates in embryos and neural stem cell fates in later stages (Pasini et al., 2004; Takke et al., 1999). However, co-injection with her4.1-MO did not rescue the phenotype in mecp2 morphants (data not shown), excluding a direct involvement of her4.1 downstream from Mecp2 in the developing zebrafish brain. While searching for candidate her and hey genes downstream from Mecp2, we found that her2 (the zebrafish ortholog of mammalian Hes5), another member of the Hairy/E(spl) family of basic helix-loop-helix proteins, was also induced in the brain, especially in the hindbrain of mecp2 morphants (supplementary material Fig. S1M–N′). Quantitative real-time PCR further confirmed the elevated levels of these Notch signaling genes (Fig. 3A). Importantly, knockdown of her2 suppressed the over-proliferation of neural precursors labeled by Tg(ef1α:mAG-zGem) in mecp2 morphants, leading to comparable levels of neural precursors in both wild-type and mecp2 morphant embryos (Fig. 3B). Accordingly, co-injection of her2-MO suppressed nestin (neural precursor marker) and gfap (glial cell marker) expression to the normal level in mecp2 morphants (Fig. 3C,C″,D,D″), whereas injection of her2-MO downregulated nestin and gfap expression in wild-type embryos (Fig. 3C‴,D‴ and Fig. 4E; supplementary material Fig. S2J,K,J′,K′). Furthermore, co-injection of her2- MO also rescued expression of the neural cell markers map2, neurod, huc, and islet1 in mecp2 morphants (Fig. 3E–E″ and Fig. 4A–A″,B–B″,C–D″), whereas injection of her2MO upregulated map2 and neurod expression in wild-type embryos (Fig. 3E‴ and Fig. 4A‴,B‴,E; supplementary material Fig. S2J″,K″,J‴,K‴). In addition, her2-MO efficiently and specifically inhibited the her2 5′UTR–EGFP reporter (supplementary material Fig. S2D–F). Taken together, these results demonstrate that Mecp2 controls the expression of her2 to maintain neural cell differentiation, which is also supported by a recent report on Her2 function in regulating neurogenesis and gliogenesis in zebrafish (Cheng et al., 2014).

In zebrafish, her2 was expressed in the whole brain at 24 hpf, and was then expressed mainly in the forebrain and midbrain at 48 hpf (supplementary material Fig. S3A,B). Activating Notch signaling by over-expression of NICD slightly increased the her2 expression (supplementary material Fig. S3A′,B′,C). By contrast, inhibiting Notch signaling by applying DAPT, a γ-secretase inhibitor, did not abolish the ectopic her2 expression in mecp2 morphants (supplementary material Fig. S3D,E). her2 was still highly expressed in the hindbrain of mecp2 morphants at 48 hpf, although the overall expression decreased after DAPT treatment (supplementary material Fig. S3D′). We found that DAPT could not rescue the phenotype in mecp2 morphants (data not shown), suggesting that Notch signaling exerts little control over her2 in the developing zebrafish brain.

Next, we turned to look for other signaling pathways that regulate her2 expression in the developing brain. Hes proteins belong to the Hairy/E(spl) family, and repress the expression of their own genes by directly binding to their promoters in mouse embryos (Bessho et al., 2003; Hirata et al., 2002). It has been shown that Id proteins activate Hes expression by releasing its negative autoregulation (Bai et al., 2007). In zebrafish, the id family contains five members, id1, id2a, id2b, id3 and id4. We found that id2b was not detectable during embryogenesis (data not shown), but id1 was increased in the forebrain, midbrain and hindbrain (Fig. 5A′) of mecp2 morphants, whereas id2a, id3 and id4 were decreased (supplementary material Fig. S3F′,G′,H′). Quantitative real-time PCR further confirmed the increased id1 and reduced id2a, id3 and id4 in mecp2 morphants (Fig. 5D; supplementary material Fig. S3I). Conversely, over-expression of mecp2 inhibited id1 expression (Fig. 5A″,D). Western blot analysis revealed that Id1 was upregulated in mecp2-MO-injected embryos, whereas it was downregulated in embryos injected with mecp2 mRNA (Fig. 5F). In addition, chromatin immunoprecipitation (ChIP) analysis of cell lysates from the zebrafish brain at 48 hpf using the anti-Mecp2 antibody demonstrated that Mecp2 bound to the chromatin of putative Mecp2-binding sites in the id1 gene promoter region, which is rich in CpG islands. This result is consistent with the previous report that Id1 proteins are significantly increased in the MeCP2-deficient mouse and the human RTT brain (Peddada et al., 2006). Our data support the hypothesis that Mecp2 directly inhibits the transcriptional activity of id1 by specifically binding to its regulatory sequences. By contrast, Mecp2 did not bind to the promoter region of her2 or her4.1, although they are also rich in CpG islands (Fig. 5G). Overexpression of id1 in wild-type embryos caused high expression of her2 in the whole brain, similar to that in mecp2 morphants (Fig. 5B′,E). Moreover, co-injection with id1-MO rescued the expression of her2 to the normal level in mecp2-MO-injected embryos (Fig. 5C–C″). Consistent with this, injection of id1-MO decreased her2 in wild-type embryos (Fig. 5B″,E). In addition, id1-MO efficiently and specifically inhibited the id1 5′UTR–EGFP reporter (supplementary material Fig. S2A–C). Taken together, we conclude that her2 is directly regulated by the Mecp2–Id1 axis during zebrafish brain development.

Given that Id1 is a direct target of Mecp2 in zebrafish, we then investigated the functional relationship between mecp2 and id1 during neural cell differentiation. Co-injection of either mecp2 mRNA or id1-MO rescued the phenotype in mecp2 morphants (Fig. 6A,B″–D″; supplementary material Fig. S3J″–M″,J‴–M‴). Compared to wild-type embryos, nestin, gfap, map2, neurod, huc and islet1 returned to normal levels, and motor nerves V, VII, and X recovered to sizes comparable to those in mecp2 morphants co-injected with id1-MO. Consistently, overexpression of mecp2 in wild-type embryos inhibited id1 expression (Fig. 5A′), leading to attenuated neural cell proliferation and astrogenesis, but enhanced primary neurogenesis (Fig. 6A; supplementary material Fig. S3A″–D″; Fig. S4E). Direct knockdown of id1 in wild-type embryos gave the same results (Fig. 6A; supplementary material Figs S2L–M‴ and S4A‴–D‴,F). Moreover, ectopic expression of id1 in wild-type embryos induced excess neural progenitors and glia while suppressing neuronal differentiation, which mimics the phenotype in mecp2 morphants (Fig. 6A; supplementary material Fig. S4A′–D′,G). Taken together, our data suggest that Mecp2–Id1–Her2 signaling plays an essential role during embryonic neural cell differentiation.

Having generated mecp2 genetic mutants (F₀) using the Cas9/gRNA system, by which biallelic mutants were previously generated in F₀ mutants (Chang et al., 2013), we used them to further substantiate the mecp2 morphant phenotype. Similar to mecp2 morphants, we found that nestin and gfap were increased while map2 and neurod were decreased in the Cas9/mecp2-gRNA-induced mutants (Fig. 7A,B–E′). These mecp2 genetic mutants were confirmed by Sanger sequencing (Fig. 7F). Therefore, our data clearly demonstrated that mecp2 is crucial for neural cell differentiation during zebrafish brain development.

In this study, we found that (1) mecp2 was expressed maternally and broadly in embryos before 24 hpf, but was enriched in the developing brain after 48 hpf, where it was expressed in the nuclei of embryonic and adult neurons; (2) either knockdown or genetic depletion of mecp2 over-activated Notch signaling and impaired neuronal differentiation in zebrafish; and (3) mecp2 inhibited the transcription of id1, which, in turn, regulated her2 expression by a delicate interplay, eventually controlling neural cell development and differentiation. Our data provide new evidence showing that, in zebrafish embryos, early embryonic neural defects caused by disrupting the MeCP2–Id1–Her2 signaling contribute to the later behavior abnormalities in mecp2 mutants, which is similar to what is found in RTT patients.

In Xenopus, a previous study has shown that MeCP2 regulates Hairy2a expression and plays an important role in neuroectoderm differentiation. Knockdown of MeCP2 in Xenopus embryos leads to a very low survival rate after neurulation and severe axial and anterior defects (Stancheva et al., 2003). In contrast, we found minimal developmental defects in mecp2 morphants in zebrafish. Consistent with our findings, others have reported that a nonsense mecp2 mutant appears to develop normally, with no evident morphological defects, but shows clear behavioral changes during early development, such as motor anomalies and defective thigmotaxis (Pietri et al., 2013). Taken together, we reason that the failure of neural progenitors to differentiate into completely mature neurons with full function in mecp2 mutants and morphants might contribute to later behavioral anomalies. Meanwhile, Pietri et al. noted that they could not completely exclude the presence of a second genetically-linked mutation in their mecp2 mutants (Pietri et al., 2013), so we analyzed a second mecp2 mutant allele. We have generated a mecp2 genetic mutant using the CRISPR/Cas9 system (Chang et al., 2013). Importantly, we found a neuronal phenotype in Cas9-induced mutants similar to that in mecp2 morphants (Fig. 7). Taken together, we conclude that mecp2 knockdown impairs neuronal differentiation, leading to abnormal behavior in later stages. In addition, we note that histone deacetylase 1, which interacts with Sin3A and MeCP2 to form an inhibitory transcription complex (Jones et al., 1998; Nan et al., 1998), is required for inhibiting the expression of Notch target genes during zebrafish neurogenesis and for maintaining the production of mature motor neurons (Cunliffe, 2004; Yamaguchi et al., 2005). The phenotypes in both hdac1 morphants and the genetic mutant hdac1^add are similar to those in mecp2 morphants, substantiating the role of mecp2 in neuronal differentiation in non-mammals.

Previous studies have given contradictory results on MeCP2 function in mammalian neuronal development. Some researchers have found that MeCP2 mutation has no effect on the proliferation of neural precursors in vitro. Neural precursors could differentiate into morphologically mature neurons and astrocytes in MeCP2 mutants, suggesting that MeCP2 is mainly involved in the maturation and maintenance of neurons (Kishi and Macklis, 2004). However, others have demonstrated that ectopic expression of MeCP2 in neural precursors not only inhibits astrocyte differentiation but also promotes neuronal differentiation in mice (Kohyama et al., 2008; Tsujimura et al., 2009). MeCP2 binds to the highly-methylated regions of genes such as astrocyte-specific GFAP and S100B and suppresses their expression (Cheng et al., 2011; Namihira et al., 2004). By contrast, loss of MeCP2 accelerates astrogenesis, with elevated expression of the glial markers GFAP and S100B (Forbes-Lorman et al., 2014; Okabe et al., 2012). These results are well aligned with our findings in mecp2 morphants in zebrafish. In addition, recent reports have shown that the soma size is reduced and that there are smaller nuclei in neurons derived from MeCP2-null embryonic stem cells, with active transcription and protein synthesis are reduced (Li et al., 2013; Yazdani et al., 2012). This is consistent with our result that the soma size of motor neurons was reduced in Tg(islet1:GFP) transgenic embryos injected with mecp2-MO (data not shown). Taken together with other work in non-mammals and mammals, we conclude that MeCP2 has a highly conserved function in neuronal development, and that its distinct functions in different species might be due to its temporal and spatial expression pattern in the brain.

Hairy/E(spl) factors (Hes in mouse, Her in zebrafish) are known for their role in controlling the fate of neural precursors at embryonic stages (Kageyama et al., 2007). In mouse embryos, overexpression of Hes1, Hes3 or Hes5 suppresses neuronal differentiation and maintains neural stem cells in the brain (Ishibashi et al., 1994; Ohtsuka et al., 2001). Meanwhile, triple knockout of Hes1, Hes3 and Hes5 results in accelerated cell differentiation; neural stem cells prematurely differentiate into neurons and are prevented from self-renewal (Hatakeyama et al., 2004). In this work, we found that the expression of her2, which is the ortholog of mouse Hes5 (Sieger et al., 2004), was increased in mecp2-MO injected embryos. Co-injection with her2-MO rescued neural cell differentiation in mecp2 morphants. Hes/Her family proteins are known to suppress proneuronal genes, such as Neurogenin1 and Mash1 (Bae et al., 2005; Hans et al., 2004; Ishibashi et al., 1995; Scholpp et al., 2009; Takke et al., 1999). Previous studies have shown that Neurogenin1 inhibits astrocyte differentiation by sequestering the CBP–Smad1 transcription complex away from astrocyte-specific genes, and by inhibiting the activation of STAT3 transcription factors that are necessary for astrogenesis (Sun et al., 2001). Furthermore, the loss of proneuronal genes blocks neurogenesis and accelerates the formation of astrocytes in mice (Nieto et al., 2001; Tomita et al., 2000). Thus, Hes/Her-protein-induced astrocytic versus neuronal fate specification is achieved, in part, through the inhibition of proneuronal genes. Therefore, we suggest that her2 regulates neuronal differentiation in the same way in zebrafish, which is supported by a recent study (Cheng et al., 2014). However, the delicate function of Her2 protein in brain development is unclear, and its downstream targets remain to be identified.

During embryogenesis, genes in the Hairy/E(spl) family are under the influence of multiple signaling pathways. Besides Notch signaling, the Wnt and SHH pathways also enhance Hes1 expression (Ingram et al., 2008; Shimizu et al., 2008). In this work, we showed that her2 was regulated in an Id1-dependent manner during neuronal differentiation. Mecp2 directly inhibited the transcriptional activity of id1 by binding to its promoter region, which is rich in CpG islands. Knockdown of mecp2 caused ectopic expression of id1, leading to the activation of her2 expression. Moreover, id1 knockdown rescued neuronal differentiation in mecp2 morphants. Id protein, which is also a helix-loop-helix protein, plays an important role in CNS development (Ruzinova and Benezra, 2003). Neither Id1- nor Id3-knockout mice have an evident phenotype in the whole brain (Pan et al., 1999; Yan et al., 1997). However, double Id1- and Id3-null mice display defective proliferation and premature differentiation in the developing brain (Lyden et al., 1999). Ectopic Map2-positive neurons are prematurely generated from normally neuron-free regions in double Id1- and Id3-knockout mouse embryos (Bai et al., 2007). On the one hand, overexpression of Id genes suppresses neuronal differentiation while promoting neural cell proliferation and astrogenesis (Bai et al., 2007; Cai et al., 2000; Jung et al., 2010), because Id proteins not only inhibit negative regulators of the cell cycle such as p16, p21 and p57 (Ohtani et al., 2001; Zheng et al., 2004), but also suppress pro-neuronal genes such as Mash1 and Neurod (Peddada et al., 2006; Vinals et al., 2004). On the other hand, knockdown of id2a enhances Notch signaling activity in zebrafish (Uribe et al., 2012). We found that id2a decreased in the forebrain, midbrain and hindbrain of mecp2 morphants, consistent with the finding that Notch signaling was also activated in mecp2 morphants. It has been shown that loss of Id2 results in decreased numbers of newborn neurons and increased numbers of newborn astrocytes in mice (Havrda et al., 2008). Taken together, these data suggest that id2a also acts downstream from MeCP2 during neural cell differentiation. Therefore, our work on the new Mecp2–Id1–Her2 axis during neural differentiation provides a framework for further deciphering the molecular pathology of human RTT.

The Tg(ef1α:mAG-zGem)^pku322 transgenic reporter zebrafish for the S, G2 and M phases of the cell cycle was generated by using Tol2-based transgenesis. Tg(neurod:EGFP) and Tg(islet1:EGFP)^mp17 fish were kindly provided by Dr Bo Zhang (Peking University, Beijing, China). Wild-type TL, AB, and transgenic zebrafish were raised and handled in accordance with the guidelines of the Peking University Animal Care and Use Committee, accredited by AAALAC.

Zebrafish mecp2 and mouse MeCP2 were amplified and subcloned into the vectors pEASY Blunt and pXT7 for antisense probe and mRNA synthesis. Plasmid for making mRNA of zebrafish NICD, and plasmid clones for making probes of zebrafish notch1a, notch1b, deltaA, deltaB and deltaC were isolated by RT-PCR from early zebrafish embryos based on the annotated genes in the Zebrafish Model Organism Database (www.zfin.org).

The morpholino (MO) antisense oligonucleotides from Gene Tools LLC (Oregon, USA) were suspended in RNase-free water as a 2 mM stock. The sequences of mecp2-MO, mecp2-mismatch MO, her2-MO, her2-mismatch MO her4.1-MO (Pasini et al., 2004), id1-MO and id1-mismatch MO are shown in supplementary material Table S1. Capped mRNAs were synthesized from linearized pXT7 constructs using the mMessage mMachine kit (Ambion). The Cas9 gene sequence was codon-optimized and ordered from Biomed (Beijing). Cas9 mRNA and mecp2-gRNA were synthesized as described previously (Chang et al., 2013). Cas9 mRNA (500 pg) and mecp2-gRNA (200 pg) were injected into one-cell embryos. In all microinjection experiments, a volume of 2.3 nl was injected into one- to two-cell stage embryos.

To construct the id1 5′UTR–EGFP and her2 5′UTR–EGFP reporters, the EGFP coding sequence was amplified from pEGFP-N1 and cloned into pCS2+ vector, and −112 to +22 bp of id1 and −86 to +22 bp of her2 were then cloned into the modified pCS2+ vector. The corresponding primers are shown in supplementary material Table S1. Capped sense Tol2 transpose, id1 5′UTR–EGFP and her2 5′UTR–EGFP mRNA were synthesized using the mMESSAGE mMACHINE in vitro transcription kit (Ambion, Austin, TX) and were purified by using an RNeasy mini kit (Qiagen, Hilden, Germany). The purified mRNA and/or corresponding morpholino were injected into zebrafish embryos at the one-cell stage.

Antisense probes were synthesized using a DIG RNA labeling kit (Roche). Some probes were transcribed from linearized constructs using T3 or T7 polymerase, and the others (nestin, gfap, neurod, map2, her2, her3, her4.1, her5, her6, her8, her9, helt, her11, her12, her13.1, her13.2, her15, hey1, hey2, id1, id2a, id2b, id3 and id4) were transcribed from PCR fragments containing a T7-polymerase-binding site. These PCR fragments were isolated from embryonic cDNAs and were confirmed by Sanger sequencing by comparing with the annotated genes in the Zebrafish Model Organism Database (www.zfin.org). The probes were detected using an alkaline-phosphatase-coupled Fab fragment antibody with BCIP-NBT staining (Roche).

Total RNA was isolated from embryos using Trizol (Invitrogen). First-strand cDNA was synthesized from 2 μg of total RNA with SuperScript II RT (Life Technologies) according to the manufacturer's protocol. Quantitative real-time PCR was performed using an iQ5 real-time PCR detection system (Bio-Rad) and an SYBR Premix Ex Taq kit (Takara). The primer sequences are listed in supplementary material Table S1.

Zebrafish embryos, adult zebrafish brains and rat brains were lysed in modified RIPA lysis buffer supplemented with protease and phosphatase inhibitors. Protein samples were harvested for western blot or immunoprecipitation reactions as described (Hu et al., 2006). Chromatin immunoprecipitation (ChIP) assays were performed using a Pierce™ agarose ChIP kit according to the manufacturer's protocol. ChIP DNA was amplified using primers designed to span 5′ CpG islands in the promoter of each gene. PCR primer sequences for isolating the three promoters are shown in supplementary material Table S1. Zebrafish embryos and adult zebrafish brains were processed for immunostaining as described previously (Wang et al., 2013). The following antibodies were used: anti-pH3 (06-570) (Millipore), anti-HuC (16A11) (Invitrogen), anti-Id1 (C-20) (Santa Cruz Biotechnology), fluorescence-conjugated secondary antibodies (Invitrogen), and horseradish peroxidase (HRP)-conjugated secondary antibodies (Santa Cruz). Rabbit and zebrafish polyclonal anti-MeCP2 antibodies were produced in our laboratory.

All experiments were repeated at least three times. Student's t-test was used to evaluate the significance of differences between experimental and control groups. P<0.05 was taken as a statistically significant effect.

We appreciate members of Dr Xiong's laboratory providing critical comments on this manuscript, and Dr I. C. Bruce for manuscript editing.