6 6 Background: N -methyladenosine (m A) modification in mRNAs was recently shown to be dynamically regulated, indicating a pivotal role in multiple developmental processes. Most recently, it was shown that the Mettl3-Mettl14 writer complex of this mark is required for the temporal control of cortical neurogenesis. The m A reader protein Ythdf2 promotes mRNA degradation by recognizing m A and recruiting the mRNA decay machinery. Results: We show that the conditional depletion of the m A reader protein Ythdf2 in mice causes lethality at late embryonic developmental stages, with embryos characterized by compromised neural development. We demonstrate that neural stem/progenitor cell (NSPC) self-renewal and spatiotemporal generation of neurons and other cell types are severely impacted by the loss of Ythdf2 in embryonic neocortex. Combining in vivo and in vitro assays, we show that −/− −/− the proliferation and differentiation capabilities of NSPCs decrease significantly in Ythdf2 embryos. The Ythdf2 neurons are unable to produce normally functioning neurites, leading to failure in recovery upon reactive oxygen species stimulation. Consistently, expression of genes enriched in neural development pathways is significantly 6 −/− disturbed. Detailed analysis of the m A-methylomes of Ythdf2 NSPCs identifies that the JAK-STAT cascade inhibitory genes contribute to neuroprotection and neurite outgrowths show increased expression and m A enrichment. In agreement with the function of Ythdf2, delayed degradation of neuron differentiation-related 6 −/− m A-containing mRNAs is seen in Ythdf2 NSPCs. Conclusions: We show that the m A reader protein Ythdf2 modulates neural development by promoting m A-dependent degradation of neural development-related mRNA targets. 6 6 Keywords: Ythdf2, N -methyladenosine (m A), Neural development, Neurogenesis, mRNA clearance Background by the methyltransferase complex (Mettl3, Mettl14, Wtap, Over the past decade, more than 100 post-transcriptionally and Kiaa1429) , erased by demethylases (Fto and modified ribonucleotides have been identified in various Alkbh5) [3, 6], and read by the binding proteins (Ythdf1–3, types of RNA . Much more recently, epitranscriptomic Ythdc1–2, and Hnrnp family proteins) [7–10].  regulation at the RNA level via reversible RNA methyla- The reversible/dynamic nature of m A in mRNA and tion has been revealed, beginning from 2011 with the dis- the ability to map this modification transcriptome-wide covery of the reversible potential of N -methyl-adenosine have led to a tremendous increase in the interest and (m A) in mRNA . As a post-transcriptional epitranscrip- understanding of the multiple biological roles of the 6 6 tomic modification, m A is one of the most abundant dynamic m A modification [10, 11]. One of the modifications in mRNA in eukaryotes . It can be written evolutionarily conserved roles of the m A modification is the regulation of meiosis and fertility. This was shown early for the writers of m Ain model organisms  and also for the mammalian m AeraserAlkbh5 * Correspondence: Xu.firstname.lastname@example.org; email@example.com; and the m A reader protein Ythdf2 . The depletion firstname.lastname@example.org Miaomiao Li and Xu Zhao contributed equally to this work. of the m A eraser Fto in mammalian cells causes defects in Department of Microbiology, Oslo University Hospital, Rikshospitalet, NO-0027 Oslo, Norway Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Li et al. Genome Biology (2018) 19:69 Page 2 of 16 energy homeostasis and adipocyte differentiation . It is vitro. However, the properties of differentiated neurons worth mentioning that a loss-of-function mutation in the were influenced, seen as less neurite outgrowth and Fto gene causes growth retardation and multiple malforma- shorter neurites. Removal of Ythdf2 increased the tions in humans . The writer Mettl3 is crucial for main- sensitivity of neurons to reactive oxygen species (ROS) taining mouse stem cell pluripotency, regulating the stress and decreased their recovery capability. RNA-seq 6 6 reprogramming of somatic cells and the circadian rhythm, combined with m A-seq uncovered that the m A- and targeting of the gene in mouse causes early embryonic modified mRNAs involved in negative regulation of lethality [16–20]. The most recent studies in hematopoietic neural development were up-regulated in Ythdf2-defi- stem/progenitor cells have uncovered the crucial role of cient NSPCs, in agreement with the function of Ythdf2. Mettl3 in determining cell fates during vertebrate embryo- The m A-modified mRNA targets, recognized by the genesis [21, 22]. The m A reader proteins Ythdf1–3share a Ythdf2 protein in the wild type, were characterized by set of common mRNA targets and spatiotemporal interplay delayed degradation in Ythdf2 knockout embryos. with each other cooperatively control translation and decay Taken together, our findings reveal the critical 6 6 of these common targets in the cytosol . The m A functions of m A modification and its binding protein readers Ythdc1 and Hnrnpa2b1 regulate splicing and Ythdf2 in neural development. processing of their mRNA targets [8, 24], while Ythdc2 affects translation efficiency as well as stability of target Results and discussion −/− mRNAs . Ythdf2 targeted mice are embryonic lethal 6 6 Recently, mutant models of the mammalian m A In order to study the biological function of the m A readers reveal interesting phenotypes, which again include reader Ythdf2, we generated conditional C57BL/6 Ythdf2 spermatogenesis  and oocyte competence . targeted mice with LoxP sites flanking the 5′ UTR and Moreover, Ythdf2-dependent, m A-modified mRNA exon 1 of the endogenous Ythdf2 locus using CRISPR-Cas9 +/loxp clearance was shown to impact the highly regulated technology (Fig. 1a). The Ythdf2 mice were crossed maternal-to-zygotic transition (MZT) in zebrafish [7, 13]. with mice ubiquitously expressing Cre-recombinase to 6 +/− −/− In Drosophila,m A writer (Ime4, dMettl14) and reader generate the Ythdf2 mice. Then to get Ythdf2 mice, +/− (Yt521-b) mutants exhibit flight defects and poor we intercrossed heterozygous Ythdf2 mice. Interestingly, −/− locomotion due to impaired neuronal functions . no viable Ythdf2 newborn mice were identified in this Most recently, the m A writer Mettl14 was shown to be particular knockout strain. The ratio of wild-type, hetero-, required for the temporal control of mammalian cortical and homozygous knockout mice was not consistent with neurogenesis . These findings strongly suggest the the expected 1:2:1 Mendelian ratio. Noteworthy, the num- 6 +/− potential role of m A modification during nervous system ber of postnatal Ythdf2 mice indicated semi-lethality +/− development, which might be conserved across species. for these mice (Fig. 1b). Furthermore, 34% of Ythdf2 Many histone and DNA encoded epigenetic mechanisms surviving mice have malfunctioning eyes, with eyelids are uncovered to be conserved in this process. Thus, remaining closed (Additional file 1: Figure S1b). Many addressing the role of m A methylation in mRNA will be factors might contribute to this [29, 30], such as an exciting new field to explore and will shed new light on dysfunction of hypothalamic nerve control, but this was neural development. not studied further here. The m A reader Ythdf2 is essential for oocyte To assess the stage of developmental failure, we col- competence and mutation of it causes female lected embryos at E12.5 and E14.5 from heterozygote in- infertility . Here we describe the early brain tercrosses and genotyped them by both PCR with developmental failure of mice lacking Ythdf2 due to primers flanking and inside the deleted genomic region failure to regulate neural stem/progenitor cell (NSPC) (Fig. 1a, c) and western blotting with Ythdf2 antibody proliferation and differentiation. During embryonic (Fig. 1d). PCR and western blot analysis confirmed that development, apical progenitor cells in the ventricular the expression of Ythdf2 is completely depleted in −/− zone (VZ) serve as primitive neural stem cells that Ythdf2 embryos. The Mendelian distribution of wild +/− −/− give rise to both the neuronal and glial lineages type, Ythdf2 ,and Ythdf2 was 1:2:1 when genotyped directly or produce secondary progenitors, termed the at embryonic stages E12.5–14.5 (Fig. 1b), suggesting the basal progenitor, in the subventricular zone (SVZ) in stage of embryonic lethality after E14.5. Therefore, we a precisely regulated spatiotemporal order . In this isolated embryos at E18.5 for further analysis. At this −/− study, Ythdf2 knockout embryos displayed delayed cortical stage, 3 out of 41 embryos were genotyped as Ythdf2 neurogenesis. In vivo and in vitro experiments proved that (data not shown). Despite the genotype ratio being Ythdf2-deficient NSPCs display decreased proliferation normal at E12.5 and E14.5, the average number of +/− rates. Furthermore, Ythdf2-deficient NSPCs could embryos per litter was significantly less in Ythdf2 naturally differentiate to neurons but not glial cells in intercrosses compared with wild-type intercrosses, Li et al. Genome Biology (2018) 19:69 Page 3 of 16 −/− Fig. 1 Ythdf2 mice are embryonic lethal. a The gene-targeting strategy to disrupt the Ythdf2 gene in mouse. Conditional Ythdf2 gene-targeted mouse contains LoxP sites flanking the 5′ UTR and exon 1 of the endogenous Ythdf2 locus. WT_F wild-type forward primer, WT_R wild- −/− −/− type reverse primer, KO_F Ythdf2 forward primer, WT_R Ythdf2 reverse primer, Ex exon. b Numbers of offspring from heterozygous +/− Ythdf2 intercrosses. The number and genotype of embryos at E12.5/E14.5 and postnatal are indicated. c PCR analysis of embryo tail DNA showing a 271-bp wild-type band (WT) and a 550-bp targeted band (KO) with primers displayed in a. d Western blot analysis of the −/− Ythdf2 expression in wild-type and Ythdf2 embryos. Two samples for each genotype. Actin was used as loading control. e Numbers of embryos per litter at E12.5/E14.5 and E18.5 from wild-type or heterozygous intercrosses. Error bars represent mean ± standard deviation, n = 7 litters. *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test especially at the late embryonic stage E18.5 (Fig. 1e). E14.5. According to our data, the major lethality of −/− It was reported that removal of Ythdf2 in zebrafish Ythdf2 embryos occurred between E14.5 and E18.5. leads to 31.3% cell arrest and lethality at the one-cell Therefore, disruption of the Ythdf2 gene results in +/− stage by Ythdf2 intercross matings , consistent embryonic lethality during the late developmental stages of with our finding of 30% less embryos at E12.5 and embryogenesis. Li et al. Genome Biology (2018) 19:69 Page 4 of 16 −/− +/− Ythdf2 mice display abnormal cortical development The Ythdf2 mice are semi-lethal. Thus, we also analyzed +/− To determine how depletion of Ythdf2 affects embry- acohort of Ythdf2 mice and found a mean 29 μm onic development, we dissected embryos at E12.5, decrease in the cortical layer at E12.5 and a mean 24 μm −/− E14.5, and E18.5. Although Ythdf2 embryos at E12. decrease in the cortical layer at E14.5 (Fig. 2a, b). We 5 and E14.5 were alive and appeared normal, sagittal suspected that the delayed cortical development derived sectionings of the whole embryos and H&E staining from a defect in the early stages of neurogenesis. In order uncovered dramatically decreased overall cortical to determine whether Ythdf2 expression is temporally −/− thickness of Ythdf2 embryonic fore brains (Fig. 2a). associated with brain development, we analyzed the Compared with their wild-type littermates, there was a expression of Ythdf2 in brain samples by quantitative RT- general 56 μm decrease in the cortical layer at E12.5 and PCR at E12.5, E13.5, E17.5, and E18.5. Ythdf2 was highly 40 μm decrease in the cortical layer at E14.5, yet the cor- expressed during the early stage of neural development texes of both genotypes grew from E12.5 to E14.5 (Fig. 2b). (Additional file 1:Figure S1a). ab cd Fig. 2 Ythdf2 is required for normal embryonic cortical development. a Sagittal brain sections of E12.5 and E14.5 were stained with H&E. An enlarged +/− −/− view of the forebrain cortex is shown. Scale bar indicates 20 μm. b Thickness of the cortical layer in Ythdf2 , Ythdf2 , and their wild-type littermates at E12.5 and E14.5. Error bars represent mean ± standard deviation, n = 3 embryos and 3 technical replicates. c Immunostaining of E12.5 and E14.5 brain +/− −/− sagittal sections for Dcx in wild-type, Ythdf2 ,and Ythdf2 littermates. Nuclei were counterstained with DAPI. VZ ventricular zone, SVZ subventricular +/− zone, IZ intermediate zone, CP cortical plate. d Ratioofthethickness ofDcx-immunolabeled neuronal layers over cortical layers in Ythdf2 −/− and Ythdf2 compared with wild type. Error bars represent mean ± standard deviation, n = 3 embryos and 3 technical replicates. *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test Li et al. Genome Biology (2018) 19:69 Page 5 of 16 +/− To further define the neuronal developmental failure compared with wild type at E12.5 and E14.5. In Ythdf2 associated with Ythdf2 deficiencies, embryonic brain cortex, the number of Phh3-positive cells was significantly slices at different developmental stages were stained with reduced compared to the wild type cortex, yet was higher −/− the immature neuron marker doublecortin (Dcx). At than in Ythdf2 cortex (Fig. 3c, d). Additionally, apical −/− +/− E12.5, the neuronal layer of Ythdf2 and Ythdf2 progenitor cells could also maintain the population by embryos was significantly thinner than that of the wild several rounds of symmetric division in the VZ layer . type, here shown as the ratio of the thickness of Dcx- As there were no obvious changes in the number of apical immunolabeled neuronal layers over cortical layers (Fig. progenitor cells (Sox2 ) in the VZ layer, we concluded 2c, d). Taken together, the in vivo evidence indicates a that Ythdf2-dependent defective neurogenesis was caused striking phenotype of retarded cortical development, by the decreased generation of basal progenitors from resulting from decreased neurogenesis at the early stages apical progenitors. of embryonic brain development. −/− Ythdf2 NSPCs exhibit decreased proliferation in vitro −/− Basal progenitor cells are decreased in Ythdf2 embryos To further understand how Ythdf2 regulates neurogen- Neural stem/progenitor cells (NSPCs) and immature esis, we cultured neurospheres consisting of NSPCs de- −/− neurons are the major cortical components at E12.5 and rived from E14.5 wild-type and Ythdf2 embryonic −/− E14.5 in mice. NSPCs give rise to neurons. Given the fore brain. The Ythdf2 neurospheres were smaller profound effects of Ythdf2 targeting on embryonic brain than the wild-type spheres (Additional file 1: Figure S2a, development, we examined the proliferation and differ- b). We first monitored the influence of Ythdf2 on NSPC entiation capability of NSPCs during development. The proliferation. NSPCs dissociated from the primary neu- T-box transcription factor Eomes (Tbr2) is specifically rospheres were seeded for proliferation testing and the expressed in basal progenitor cells, predominantly in the cell growth was determined at 0, 24, 72, and 120 h. −/− SVZ, which primarily differentiate into superficial layer Compared with the wild type, Ythdf2 NSPCs showed −/− neurons. In E12.5 and E14.5 Ythdf2 embryos and, to a a slightly decreased proliferation rate after 24 h and a +/− lesser extent, Ythdf2 embryos, there was a dramatic more pronounced reduction after 72 h culturing (Fig. 4a). loss of basal progenitor cells, displayed by the obviously This result is in agreement with the decreased mitotic thinner Tbr2 layer, compared to wild type littermate capability of stem/progenitor cells observed in vivo. embryos (Fig. 3a). The sex determining region Y-box2 (Sox2) is a marker for apical progenitor cells located in Ythdf2-deficient NSPCs show impaired neural the VZ, which can produce deep layer neurons and basal differentiation progenitor cells . The ratio of Tbr2-positive cells to In differentiation assays, NSPCs dissociated from neuro- + + total progenitors (Tbr2 /Sox2 ) was decreased markedly spheres produce both neurons and glial cells after 5 days −/− +/− at E12.5 and E14.5 in Ythdf2 and Ythdf2 embryos culturing. We first assessed the mRNA expression pro- compared with the wild types (Fig. 3b), suggesting the file of Ythdf2 in wild-type neurospheres during differen- decrease in neurons (Dcx ) associates with a reduction tiation by RT-qPCR. The expression of Ythdf2 was up- in the basal progenitor population in SVZ. However, regulated from Day 0 (D0) to D3 during differentiation there was no obvious difference in Sox2-positive apical and remained high till D5, suggesting the involvement progenitor cells in VZ layer (Fig. 3a). of Ythdf2 in regulating differentiation (Fig. 4b). Neur- onal and glial cell lineages can be identified by stain- Mitotic capability of apical progenitor cells is impaired in ing with antibody against microtubule associated −/− Ythdf2 embryos protein 2 (Map2) or glial fibrillary acidic protein The non-self-renewing basal progenitors only experience (Gfap), respectively. We quantified the percentages of −/− one or two mitotic cycles, and the majority of basal pro- Gfap-positive cells for Ythdf2 and wild-type at D5 genitors are established by asymmetric division of apical and D7. Dramatic reduction of glial cells, with abnor- progenitor cells during early cortical development [32, mal branches (Gfap ), was observed in differentiated −/− 33]. We propose that the decrease in basal progenitors Ythdf2 neurospheres (Fig. 4c, d). However, we did + −/− (Tbr2 ) might be caused by the reduced mitotic not observe a significant different ratio of Ythdf2 + −/− capability of the Ythdf2-depleted apical progenitor cells. neurons (Map2 ) at D5, while the ratio of Ythdf2 The E12.5 and E14.5 sagittal sections of wild type, neurons declined significantly more than the wild type at +/− −/− Ythdf2 , and Ythdf2 embryos were co-stained with D7. These results were further substantiated by neuron the mitotic phase marker phospho-histone H3 (Phh3) progenitor antibody neuron-specific class III beta-tubulin and Sox2 to quantify the mitotic capability of the apical (Tuj1) and glial progenitor antibody S100 calcium-binding progenitor cells. The number of Phh3-positive cells de- protein B (S100-β) staining at D3 and D5 (Additional file 1: −/− creased more than two-fold in Ythdf2 cortex Figure S2c, d). At D3, the number of glial lineage Li et al. Genome Biology (2018) 19:69 Page 6 of 16 ab cd Fig. 3 The number of basal progenitors and mitotic capability of apical progenitors depends on Ythdf2. a Immunostaining of E12.5 and E14.5 +/− −/− sagittal sections with Tbr2 (green) and Sox2 (red) antibodies in wild type, Ythdf2 , and Ythdf2 embryos. VZ ventricular zone, SVZ subventricular + + + zone, IZ intermediate zone, CP cortical plate. Nuclei were counterstained with DAPI. b Percentage of Tbr2 cells over Tbr2 /Sox2 at E12.5 and E14.5. Error bars represent mean ± standard deviation, n = 3 biological and 3 technical replicates. Scale bars, 20 μm. c Immunostaining of E12.5 +/− −/− and E14.5 sagittal sections with Phh3 (green) and Sox2 (red) antibodies in wild type, Ythdf2 , and Ythdf2 embryos. Nuclei were counterstained with DAPI. d Number of Phh3 cells per 400 μm of the cortical wall at E12.5/E14.5 from c. Error bars represent mean ± standard deviation, n =3 biological and 3 technical replicates. *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test. Scale bars, 20 μm −/− progenitors had already declined in Ythdf2 cells, morphological analysis of Map2-positive cells showed −/− while no difference was observed for neuronal lineage that Ythdf2 differentiated neurons had less and progenitor cells (Additional file 1: Figure S2c, d). The shorter primary neurites (axons and dendrites). The −/− TUNEL assay showed significantly more dead Ythdf2 mean number of branching neurites per neuron in −/− cells, which might result from impaired differentiation Ythdf2 differentiated cells is less than in the wild type (Additional file 1:FigureS2e,f). (Fig. 4e), and the mean length of the longest neurite in −/− Ythdf2 differentiated cells is shorter than in the wild Ythdf2-deficient neurons display abnormal neurite type (Fig. 4f). The neurite outgrowth is pivotal in neuronal outgrowth and increased sensitivity to arsenite development and maturation, synaptic formation, neuronal Whereas neuronal lineage differentiation (Map2 or function, and functional recovery in diseases . The + −/− Tuj1 ) was not affected at D5 in Ythdf2 cells, the severe effect on neurite branching and extension of Li et al. Genome Biology (2018) 19:69 Page 7 of 16 ac bd ef −/− Fig. 4 Ythdf2 NSPCs exhibit decreased proliferation and defects in natural differentiation in vitro. a Number of viable NSPCs at 0, 24, 72, and 120 h monitored by signal intensity of Presto Blue reagent. Proliferation rate was calculated by normalizing to wild type at 0 h. b mRNA expression levels of Ythdf2 during NSPC differentiation. Cells were collected at differentiation Day 0 (D0), 3, and 5. Isolated total RNAs were + + applied for RT-qPCR analysis. Actin was used as normalization control. c Immunostaining of Map2 and Gfap cells differentiated from E14.5 neurospheres at D5 and D7. Nuclei were counterstained with DAPI. Scale bar indicates 20 μm. d Percentage of Map2 or Gfap positive cells. Error bars represent mean ± standard deviation, n = 3 biological repeats and 3 technical replicates. e Mean number of primary neurites per neuron + + (Map2 ). n = 20 neurons for each biological repeat. f Mean length of the longest neurite of neurons (Map2 ). n = 20 neurons for each biological repeat. *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test −/− Ythdf2 neurons might also contribute to the defective mammalian brain [36, 37]. It is reported that after arse- neurogenesis during neural development. nite treatment, Ythdf2 can co-localize with P body to Besides, we found that differentiated neurons in vitro regulate mRNA decay . We treated differentiated were more sensitive to arsenite treatment. Arsenite was neurons with 5 μM arsenite for 24 h in vitro, followed demonstrated to induce oxidative stress by generating by recovery in fresh medium for 24 h (Additional file 1: ROS and depleting antioxidants in cell lines and Figure S3a). For wild-type neurons, the mean length of Li et al. Genome Biology (2018) 19:69 Page 8 of 16 6 6 neurites was shortened and the number of neurites m A sites from 8201 genes and 17,734 common m A reduced after 24-h arsenite exposure (Additional file 1: sites from 8585 genes in three biological replicates of −/− Figure S3b, c). However, after 24-h culture in fresh wild-type and Ythdf2 neurospheres, respectively medium, the remaining neurites recovered to the original (Additional file 1:FigureS5a). Thehighly over- length and the neurite number partially increased as the represented m A RRACH (R = G/A, H=U/A) motif growth of new neurites needs longer time (Additional identified using the HOMER algorithm in both wild-type −/− −/− file 1: Figure S3b, c). In contrast, Ythdf2 neurons (P = 1e-471) and Ythdf2 (P = 1e-475) neurospheres showed increased sensitivity to arsenite exposure proved the successful enrichment of m A-modified compared to wild-type neurons. After 24-h recovery, mRNA (Fig. 5b and Additional file 1: Figure S5b). The −/− 6 Ythdf2 neurites could not outgrow to the original m A sites were significantly enriched at start codons, length and no new neurites projected. stop codons, and 3′ UTRs. The m A profile is thus in very good agreement with those reported previously (Fig. 5c Negative regulation of neural development pathways and Additional file 1: Figure S6a). −/− enriched in Ythdf2 neurospheres Based on the statistics from three biological replicates, 6 6 To address the molecular mechanism of modulating 3095 m A sites from 2464 genes and 4109 m A sites NSPC proliferation and differentiation, we performed from 2619 genes were identified to have lower or higher −/− 6 −/− mRNA sequencing in the wild-type and Ythdf2 m A levels in three biological replicates of Ythdf2 neurospheres with three biological replicates. We neurospheres (Fig. 5d and Additional file 1: Figure S6b). identified 2144 up-regulated differentially expressed genes m A sites with significantly higher enrichment (fold −/− (DEGs) and 1756 down-regulated DEGs (Additional file 1: change > 1.5) in all three Ythdf2 replicates were Figure S4a). With more stringent criteria (fold change analyzed further. Based on this stringent criterion, 78 >1.5, P < 0.05 in three replicates), 151 significantly m A sites from 69 genes were markedly up-regulated. up-regulated and 316 significantly down-regulated These genes were enriched for functional clusters like −/− genes were identified in Ythdf2 neurospheres. transcription regulation, phosphorylation, and neuron Interestingly, the up-regulated genes were significantly projection development (Additional file 1: Figure S7a). associated with axon guidance, synapse assembly, On the other hand, 102 m A sites from 99 genes were neuron differentiation, and apoptosis. All these bio- down-regulated. These genes were enriched for func- logical processes are subordinate to nerve develop- tional clusters like transcription regulation, transport, ment (Additional file 1: Figure S4b). The JAK-STAT rhythmic process, and apoptosis (Additional file 1: signaling pathway is up-regulated in neurons and glial Figure S7b). Among these 168 genes, 115 genes had con- cells, which contributes to the neuroprotection and served m A sites across samples, while 65 and 54 genes neurite outgrowth [38, 39]. The genes, highly had newly occurring or absent m A sites, respectively, in all −/− enriched for Gene Ontology (GO) term “negative three Ythdf2 neurospheres (Additional file 1:FigureS7c). regulation of JAK-STAT cascade”, inhibit this cascade, such as Flrt2, Flrt3, Ptprd,and Lrrtm1 and 4.On the Ythdf2 is required for degradation of genes related to contrary, clustered terms, such as “positive regulation neuron differentiation of cell differentiation”, “positive regulation of tran- It is well established that Ythdf2 specifically binds mRNAs scription”, “positive regulation of GTPase activity”, containing m A and promotes mRNA decay [1, 7]. In and “negative regulation of neuron apoptotic process”, Ythdf2-depleted zebrafish embryos, Ythdf2-targeted mRNAs were dominant in down-regulated genes. had extended lifetimes as seen by increased mRNA levels . Hence, we focused on verifying candidate genes with 6 −/− 6 m A-methylomes in wild-type and Ythdf2 neurospheres increased mRNA transcripts and enrichment of m Asites. In order to gain more insight about the role of Among these genes, Nrp2 and Nrxn3 were involved in −/− 6 6 Ythdf2 in m A mRNA decay, we compared the m A nerve development and cell differentiation; Flrt2 and Ptprd −/− methylome of wild-type and Ythdf2 neurospheres. were enriched in negative regulation of JAK-STAT cascade, Initially, we quantified the m A/A ratio of the total regulation of synapse assembly and axon guidance, and −/− mRNAs purified from the wild-type and Ythdf2 neuron differentiation; Ddr2 was related to fibroblast prolif- −/− neurospheres by LC-MS/MS. In Ythdf2 neurospheres, eration; Hlf was involved in rhythmic process; and Nrp2 6 6 the m A abundance was increased by around 10% on and other genes showed enrichment of representative m A −/− average compared with the wild type (Fig. 5a). This is peaks in Ythdf2 neurospheres (Fig. 5e and Additional consistent with the m A-dependent RNA decay function file 1: Figure S8). To further substantiate these findings, we of Ythdf2  and correlates very well with a study in performed m A immunoprecipitation (IP) combined with zebrafish on the role of Ythdf2 in the maternal-to- RT-qPCR. Consistent with our initial findings, m AIP zygotic transition . We identified 16,626 common showed that m A levels increased significantly, while Li et al. Genome Biology (2018) 19:69 Page 9 of 16 ac 6 −/− 6 −/− Fig. 5 Overview of m A methylomes in wild-type and Ythdf2 neurospheres. a The m A contents of mRNAs isolated from wild type and Ythdf2 were 6 −/− 6 quantified by LC-MS/MS. b Sequencing motif in m A peaks verified in wild type and Ythdf2 with HOMER database. c Distribution of m Apeaks along −/− 6 6 transcripts in wild type and Ythdf2 . d Scatter plot showing m A peaks with increased (red) or decreased (green) levels. e Representative m A distribution along Nrp2 transcript. Enrichment coverage of m A and input are displayed as red and blue, respectively. Grey lines define coding sequence (CDS)borders −/− non-methylated actin was used as negative control (Fig. 6a). than in Ythdf2 neurospheres (Fig. 6d and Additional RT-qPCR showed that target genes were markedly enriched file 1: Figure S10). Thus, the increased levels of m A- −/− in Ythdf2 neurospheres compared with the wild type modified mRNA transcripts in the absence of Ythdf2 were (Fig. 6b). Next, we analyzed whether these m A-enriched caused by delayed mRNA clearance, which might genes are real Ythdf2 targets by RNA IP (RIP) analysis. We contribute to the defects in neurogenesis. confirmed that the Ythdf2 antibody was applicable to IP −/− (Additional file 1: Figure S9). Compared with Ythdf2 , Conclusions Nrp2 mRNA and other candidates were enriched by Ythdf2 is essential for oocyte maturation and early zygotic Ythdf2 protein in the wild type, which was verified by development in zebrafish and mouse [13, 25]. Ythdf2 was qPCR (Fig. 6c). To examine whether increased gene recently reported to be required for oocyte competence expression was due to loss of Ythdf2-mediated RNA through the post-transcriptional regulation of the mater- −/− decay, we measured the mRNA life time of these candi- nal transcriptome and homozygous Ythdf2 mice were date genes by inhibition of transcription with actinomy- reported to be partially permissive at weaning, with −/− −/− cin D in wild-type and Ythdf2 neurospheres. After approximately 80% loss of homozygous Ythdf2 mice in actinomycin D treatment, mRNA levels of Nrp2 and the inbred C57BL/6 mice . The targeting of the Ythdf2 other candidate genes in wild type declined more rapidly locus described here caused a complete loss of Li et al. Genome Biology (2018) 19:69 Page 10 of 16 6 6 Fig. 6 Ythdf2 is required for regulating mRNA decay of m A-modified neuron-related gene targets. a m A enrichment of target sites in gene 6 6 candidates, verified by m A IP combined with RT-qPCR. Non-m A-modified gene Actin was used as negative control. b Gene expression of gene candidates, verified by RT-qPCR with input RNA. Non-changed gene Actin was used as negative control. c Ythdf2 binding levels of gene candidates, verified by Ythdf2 RIP combined with RT-qPCR. Non-m A-modified gene Actin wasused asnegativecontrol. d Representative mRNA profile of Nrp2 at −/− 0-, 2-, and 4-h time points after actinomycin D (5 μg/ml) treatment (h.p.t.) in wild type and Ythdf2 . Error bars represent mean ± standard deviation, n = 2 biological replicates. *P < 0.05, **P <0.01, ***P < 0.001, Student’s t-test Li et al. Genome Biology (2018) 19:69 Page 11 of 16 −/− homozygous Ythdf2 mice in our inbred C57BL/6 and Additional file 1: Figure S2c). However, morphological −/− background and the majority of Ythdf2 embryos died at analysis demonstrated abnormal neurite outgrowth of −/− late embryonic stages (Fig. 1). Of note, intercrossing Ythdf2 neurons that are more vulnerable to stress and +/− Ythdf2 mice results in constantly smaller litter size than fail to recover from neurite degeneration (Fig. 4e, f and from wild type matings (Fig. 1e), indicating an essential Additional file 1: Figure S3). Proper neurite outgrowth role of Ythdf2 in early embryo development. Here we and branching is pivotal for establishing neuronal reveal a crucial role of m A in mRNA and its binding circuits which facilitate nervous system function . protein Ythdf2 in neural development at embryonic Interestingly, RNA-seq analysis shows that differentially developmental stages. The mammalian nervous system expressed genes (DEGs) relate to functions such as arises from the ectoderm, with both neurons and glial axon regulation, synapse assembly, and neuron differ- cells (astrocytes and oligodendrocytes) generated from entiation (Additional file 1: Figure S4). Among them, NSCs in a precisely regulated spatiotemporal order . genes such as Ddr2, Rnf135, Flrt2, Hlf, Nrp2, Nrxn3, We propose that erroneous recognition and degradation and Ptprd have both up-regulated mRNA and m A of m A-containing mRNA at this stage leads to the levels (Fig. 6a, b). Ythdf2-mediated mRNA decay affects dysregulation of neural development. the translation efficiency and lifetime of m A-modified The m A level in mRNAs is higher in brain than in mRNA targets . By recruiting the Ccr4-not deadeny- other studied mouse organs, indicating a crucial role lase complex, Ythdf2 initiates the degradation of its during normal brain development . Recent studies mRNA targets at specialized decay sites . The RIP found m A-modifying enzymes Mettl3, Alkbh5, and Fto combined RT-qPCR and mRNA life-time assays display to be involved in regulating progression of glioblastoma, that mRNA levels of these genes are stabilized due to 6 −/− indicating that m A epitranscriptomic regulation plays the complete absence of Ythdf2 in Ythdf2 NSPCs roles in the nervous system. Very recently, Yoon et al.  (Fig. 6c, d and Additional file 1: Figure S10). Delayed used a methyltransferase Mettl3-Mettl14 complex knock- mRNA degradation causes the retention of m A- 6 −/− out to demonstrate that m A depletion extends cortical modified transcripts in Ythdf2 neurospheres, leading neurogenesis by protracting cell cycle progression of NPCs. to increased m Aenrichment. −/− In this study, we demonstrate the severe impact of Ythdf2 Last but not least, while homozygous Ythdf2 is +/− deletion on corticogenesis, neurogenesis, and gliogenesis. embryonic lethal, heterozygous Ythdf2 is unexpectedly +/− During early neural development, the decreased thickness only partially lethal, with 30% of the surviving Ythdf2 of cortex is attributed to the dramatically thinner CP and mice having eye defects (Additional file 1: Figure S1b), SVZ layers, composed of neurons (Dcx )and basal which may reflect haploid insufficiency of Ythdf2 and a + 6 progenitor cells (Tbr2 ), respectively (Figs. 2c and 3a). malfunctioning nervous system. Furthermore, m Ais Multiple factors are supposed to contribute to this. First, highly enriched in mouse brain, and the level is consistent with the documented function of m Ain dramatically increased with postnatal aging . Taken proliferation of NPCs , our in vivo and in vitro evidence together, we propose that m A and Ythdf2 have a pivotal reveals that the proliferation capability of the NSPCs is function in brain not only during embryonic neural −/− severely compromised in Ythdf2 embryonic cortex development but also in postnatal life. Thus, functions of NSPCs. Further, apical progenitors symmetrically divide m A and Ythdf2 on postnatal nervous system development into more apical progenitors to expand the stem cell/ merits further investigations. progenitor VZ pool . No significant change in the During revision of this manuscript, two studies relating −/− 6 thickness of the VZ layer in Ythdf2 embryos suggests to the role of m A in the adult mammalian nervous that the symmetric division of apical progenitors is not system were reported. One study found that either the 6 6 disturbed. However, apical mitosis is significantly m A methyltransferase Mettl14 or the m A-binding −/− decreased in Ythdf2 embryos. Apical progenitor cells protein Ythdf1 regulate functional axon regeneration in can give rise to basal progenitor cells and neurons by the peripheral nervous system in vivo by modulating asymmetric division [40, 41]. The switch between injury-induced protein translation . In another study it −/− 6 symmetric and asymmetric cell division of Ythdf2 was discovered that m A in mRNA regulates histone neural progenitor cells, which determines self-renewal modification in part by destabilizing transcripts that or differentiation, may be disturbed. It is worth men- encode histone-modifying enzymes, which might be a pre- tioningthatitisnotconfirmedthatthe neural defects viously unknown mechanism of gene regulation in observed contribute to embryonic lethality. So it will be mammalian cells . interesting to address this relationship by generating In summary, our study demonstrates a pivotal neural-specific Ythdf2 knockout mice. Second, the function of Ythdf2-mediated m A epitranscriptomic −/− NSPCs derived from Ythdf2 embryo brains generated regulation in cortical neurogenesis during embryonic similar numbers of neurons as wild type in vitro (Fig. 4c neural development, via regulating RNA degradation of Li et al. Genome Biology (2018) 19:69 Page 12 of 16 m A-tagged genes associated with neural development Primers and differentiation. Ddr2 Forward, TTGGCCACCCAAACAATCCA Reverse, AGACCCCTCTGGTCACCAAC Methods Mob3b Forward, GAAAGCGATCCTGACTTCCAG Reverse, GCTAGCAGCACTTAGAGGGT Generation of conditionally Ythdf2 gene-targeted mice Rnf135 Forward, ACTGGGAAGTGGACACTAGG The Ythdf2 conditional knockout mouse model (mYthdf2- Reverse, CCAGGAGTCCATAGTCCTTCC CKO) was generated as described in Additional file 1: Speg Forward, CTAGTGGTGCGGGCAAATCT Figure S1a by Applied Systemcell Inc. (CA, USA) using Reverse, CCTGGTTAGCGGGAATTGGT CRISPR-Cas9 technology. A cocktail of active guide RNA Flrt2 Forward, GACTGCCACATCCCCAACAA molecules (gRNAs), two single-stranded oligo donor Reverse, CACCTTTCTAACGCTGGACCT nucleotides (ssODNs) and qualified Cas-9 mRNA was Hlf Forward, CTGAAGGAGAACCAGATCGCA microinjected into the cytoplasm of C57BL/6 embryos. Reverse, TTCTTGCATTTGCCCAGCTC Two LoxP sites were inserted, flanking the upstream of 5′ Nrp2 Forward, CCCTTTGGAAACTGAATGCCA UTR and intron 1 regions, resulting in loss and changes in Reverse, GATCCCCTTCACAGCTGCAT fl/fl size of PCR products. Ythdf2 mice were genotyped and Nrxn3 Forward, ACGTATGGGCTCCATTTCCT further sequenced for the LoxP cassettes at the designated Reverse, TTCTTGAGGCTTCCCGTGAG locations. Potential Ythdf2-CKO mice were generated by Ptprd Forward, TGAGCCATACAGGGCACTTG fl/fl crossing Ythdf2 mice with Cre_Del_GT_07 mice from Reverse, GCCTCCTAAGTCAGGATTCTTGT the Norwegian Transgenic Center (NTS, Oslo, Norway). Soat1 Forward, GTGCAAGGGTGAGCCTATGT For Ythdf2 genotyping, ear-clip samples were lysed in al- Reverse, GTGTGAGCAACTTGTACGGC kaline lysis reagent (25 mM NaOH, 0.2 mM EDTA, pH 12) Actin Forward, TTCTTTGCAGCTCCTTCGTT Reverse, ATGGAGGGGAATACAGCCC at 95 °C for 30 min, followed by adding neutralization re- agent (40 mM Tris-HCl, pH 5). PCR conditions for wild type and knockouts: 95 °C, 2 min, 1 cycle; 95 °C, 30s; 60 °C, 30s; 72 °C, 1 min; 35 cycles. The PCR products were de- Western blotting scribed in Additional file 1:FigureS1b. Primers forgeno- Total protein lysate was extracted with RIPA buffer typing were as follows: wild-type allele (WT), 5′-TAC (20 mM Tris-HCl, pH 7.4, 20% glycerol, 0.5% NP40, GGGTGAGGTGTCTTTTTCTT-3′,5′-GAAAGAGAGG 1 mM MgCl , 150 mM NaCl, 1 mM EDTA, 1 mM AAACGAGGAAG-3′; targeted allele (KO), 5′-GGCTC EGTA). Protein concentrations were measured using the TCCCTTCCCGAGAT-3′,5′-GCTTTTGTCCCTGACAC Bradford Assay, and 50–100 μg protein extracts were TCG-3′. subjected to SDS-PAGE. Then proteins were transferred to a nitrocellulose membrane, blocked with 5% non-fat Antibodies milk and incubated with first antibodies for 1 h at room The following antibodies were used at the appropriate dilu- temperature. After incubation with secondary antibody tions: mouse anti-Map2 (M4403, Sigma), rabbit anti-Gfap against mouse (1:10,000) or rabbit (1: 10,000) for 1 h at (Z0334, DAKO), mouse anti-Tuj1 (MAB1195, R&D Sys- room temperature, the membrane was visualized with an tems), rabbit anti-s100-β (ab52642, Abcam), rabbit anti-Dcx ECL Western Blotting Detection Kit (32,106, Thermo). (ab18723, Abcam), rabbit anti-Tbr2 (ab23345, Abcam), rabbit anti-phospho-Histone H3 (PHH3; Ser10; 06–570, Immunohistochemistry and immunofluorescence Millipore), mouse anti-Sox2 (ab79351, Abcam), mouse anti- For immunohistochemistry, embryonic brain tissues Nestin (MAB353, Millipore), rabbit anti-Ythdf2 (RN123PW, were dissected in cold PBS and fixed in 4% PFA at 4 °C MBL), mouse anti-anti-β-actin (A1978, Sigma). for 48 h. Slides (4 μm thick) were sectioned by micro- tome (HM355s, Thermo Scientific) and deparaffinized RT-qPCR analysis and cleared in Clear-Rite™ 3 (6901TS, Thermo) followed The total RNA was isolated using TRIzol LS Reagent (Life by rehydration in an EtOH gradient. After antigen re- Technologies, 10,296–010). Normally, 1 μgtotal RNAwas trieval in citrate buffer (pH 6.4), the slides were blocked used for reverse transcription using High-Capacity cDNA with blocking buffer (5% goat gut, 5% BSA, 0.1% tween- Reverse Transcription Kit (ThermoFisher, 4,368,814). The 20, 0.5% Triton X-100) for 1 h, and incubated with pri- quantitative PCR reactions were carried out with Power mary antibodies overnight at 4 °C. Secondary antibodies SYBR Green PCR Master Mix (Life Technologies, were applied at room temperature for 1 h. For immuno- 4,368,708) on a StepOnePlus™ Real-Time PCR System in- fluorescence, cultured cells were fixed with 4% parafor- strument (Applied Biosystems). Primers used in this study maldehyde (PFA), permeabilized with 0.1% Triton X-100, were as follows. and stained with primary antibodies and secondary Li et al. Genome Biology (2018) 19:69 Page 13 of 16 antibodies. Nuclei were visualized with mounting medium Ythdf2 RIP with DAPI (BioNordika, H-1200). Images were taken with Ythdf2 RIP was carried out with a modified procedure a Leica SP8 confocal microscope equipped with a ×40 oil . Briefly, 1 × 10 collected NSPCs were lysed in immersion lens. NETN buffer (20 mM Tris-Cl, pH 8.0; 100 mM NaCl, 1 mM EDTA, 0.5% NP-40, freshly added protease inhibi- H&E staining tor cocktail and RNasin) for 20 min on ice. After centri- Tissue slides were stained in haematoxylin (Richard-Allen fugation, the supernatant containing the RNA–protein Scientific, 12,687,756) and eosin (Nerliens Meszansky, complex was incubated with 5 μg Ythdf2 antibody 161,170) after dehydration and rehydration, followed by (RN123PW, MBL) for 2 h at 4 °C. Then 30 μl Dynabeads differentiation in acetic acid in 100% ethanol at 1:50,000 G beads were added and rotated for 2 h, at 4 °C. Beads dilution for 5 s. Then, the sections were dehydrated in as- were collected with a magnetic stand and washed with cending series of ethanol, treated with xylene, and cover- NETN buffer four times. The RNA–protein complex slipped using Cytoseal XYL xylene-based mounting was eluted by incubating with NETN buffer with 0.1% medium (8312–4, Thermo). Images were taken with a SDS and 30 μg proteinase K at 50 °C for 30 min. RNAs Zeiss AxioPlan 2 microscope system. were further purified with RNA Clean and Concentrator-5 (Zymo). TUNEL assay Cellswere grownoncoverslips, fixedonice with 4% LC-MS/MS PFA for 10 min, and permeabilized with 0.2% Triton Purified mRNA was digested by nuclease P1 (2 U, X-100 in PBS-Tween for 30 min on ice. After incubat- Wako) in 25 μl of buffer containing 10 mM of NH OAc ing in 3% H O in PBS for 10 min, slides were rinsed (pH 5.3) at 42 °C for 2 h, followed by the addition of 2 2 twice with PBST. Slides were incubated with 50 μl NH HCO (1 M, 3 μl, freshly made) and alkaline phos- 4 3 TUNEL reaction mixture for 60 min at 37 °C. Nuclei phatase (0.5 U). After an additional incubation at 37 °C were visualized with mounting medium with DAPI for 2 h, the sample was diluted to 50 μl and filtered (0. (BioNordika, H-1200). Images were taken with a Leica 22 μm pore size, 4 mm diameter, Millipore), and 5 μlof SP8 confocal microscope equipped with a × 40 oil the solution was subjected to LC-MS/MS. Nucleosides immersion lens. were separated by reverse-phase ultra-performance li- quid chromatography on a C18 column with on-line Neurosphere proliferation and differentiation mass spectrometry detection using an Agilent 6410 Neurospheres derived from E14.5 embryonic fore brains QQQ triple-quadrupole LC mass spectrometer in posi- were cultured with DMEM/F12 (GIBCO) supplemented tive electrospray ionization mode. The nucleosides were with 20 ng/ml EGF (R&D Systems, 236-EG-200), 10 ng/ml quantified using the nucleoside to base ion mass transi- bFGF (R&D Systems, 234-FSE 025), N2 supplement (Life, tions of 282 to 150 (m A) and 268 to 136 (A). 17,502–048), and B27 supplement without vitamin A Quantification was performed in comparison with the (Thermo, 12,587,010). Under the proliferating condition, standard curve obtained from pure nucleoside standards cells were grow as free-floating neurospheres. For second- running on thesamebatch ofsamples. Theratio ofm A ary neurosphere formation, cells in primary neurospheres to A was calculated based on the calibrated concentrations. were trypsinized with TrypLE™ Express Enzyme (Gibco, 12,604,021) combined with DNaseI (Thermo, 18,047,019), mRNA isolation and m A-RIP dissociated mechanically by pipetting onto a six-well plate NSPCs (1 × 10 ) dissociated from neurospheres were at 5 × 10 cells per well. For the neurosphere differentiation collected for total RNA isolation with Direct-zol RNA assay, a set of neurospheres were trypsinized to obtain a miniprep plus with TRI Reagent (Zymo research, R2073) suspension of dissociated cells. These cells were then plated and DNase I digestion following the manufacturer’s in- in tissue culture plates pre-coated with poly-L-lysine structions. We applied 1 mg total RNA for further (Sigma, P6516). Cells were cultured in differentiation mRNA purification with a Dynabeads mRNA DIRECT™ medium (minus EGF and bFGF) and collected at different purification kit (Thermo, 61,011) for two rounds. The time points. mRNA quality was checked using a 2100 Bioanalyzer in- strument with an Agilent RNA 6000 Nano kit (5067–1511). Proliferation assay RNA fragmentation (1 μg) was performed by sonic- Dissociated single NSPCs were seeded at a density of ation at 10 ng/μlin100 μl RNase-free water with 1.0 × 10 per well in 96-well plates. The proliferation Bioruptor Pico (Diagenode) with 30 cycles of 30 s on rates were measured at 24, 72, and 120 h with Presto- followed by 30 s off; 5% of the fragmented RNA was Blue Cell Viability reagent (A13262, ThermoFisher saved as input. m A IP was performed with an EpiMark® Scientific) as instructed. N -Methyladenosine Enrichment Kit (NEB, E1610S) Li et al. Genome Biology (2018) 19:69 Page 14 of 16 following the kit manual adapted for the KingFisher™ mRNA life-time assay −/− Duo Prime Purification System. In detail, 1 μl N6- Wild-type and Ythdf2 neurospheres were trypsinized methyladenosine antibody from the kit and 25 μl Protein with TrypLE™ Express Enzyme (Thermo, 12,605,010). G beads (NEB #S1430) were used for each affinity pull The dissociated cells were seeded into plates coated down. After incubating with RNA, the beads were with PDL (Millipore, A-003-E) and laminin (R&D washed with 200 μl low salt reaction buffer twice, and Systems, 3446–005-01). After 12-h culturing, cells were then 200 μl high salt reaction buffer twice. RNA that treated with 5 μg/ml actinomycin D (Sigma, A9415) for was pulled down (IP) was eluted with 50 μl RLT buffer 2 and 4 h, while cells without treatment were used as twice, and recovered by RNA Clean and Concentrator-5 0 h. Cells were collected at designated time points and (Zymo). Both input and IP were subjected to RNA library total RNA was extracted for reverse transcription and preparation with Truseq Stranded mRNA Library Prep Kit qPCR. with the RFP incubation step shorten from 8 min to 20 s. Sequencing was carried out on Illumina HiSeq 4000 ac- Statistical analysis cording to the manufacturer’s instructions. All statistical analyses were performed with GraphPad Prism 5. Student’s t-test was adapted and data are shown as mean ± standard deviation. P value is used Sequencing data analysis for significance. The sequencing data were mapped to mouse genome version mm10 downloaded from UCSC. Data analysis was Additional file carried out as previously described. Briefly, reads were aligned to mm10 using TopHat v2.0.142. For input ana- Additional file 1: Figures S1–S10. This document contains additional supporting evidence for this study presented in the form of supplemental lysis (RNA-seq), RPKM were calculated by Cuffnorm3. 6 figures. (PDF 1160 kb) For m A peak calling, the longest isoform was used if multiple isoforms were detected. Aligned reads were Acknowledgments extended to 100 nucleotides (average fragment size) and We thank the animal facility at Oslo University Hospital for mouse handling. converted from genome-based coordinates to isoform- based coordinates to eliminate interference from introns Funding This work was funded by the Norwegian Cancer Society (to A.K. and M.B.), in peak calling. The longest isoform of each mouse gene the Norwegian Research council, the Health Authority South East, and was scanned using a 100-nucleotide sliding window with National Institute of Health (RM1 HG008935 to C.H.). C.H. is an investigator of 10-nucleotide steps. To reduce bias from potentially in- the Howard Hughes Medical Institute. accurate gene structure annotation and the arbitrary use Availability of data and materials of the longest isoform, windows with read counts less than High-throughput sequencing data have been deposited in the Gene 1/20 of the top window in both m A IP and input samples Expression Omnibus database under accession number GSE104867 . All were excluded. For each gene, the read count in each the other data generated or analyzed during this study are included in the article and additional files. window was normalized by the median count of all windows of that gene. The window was called positive if Authors’ contributions FDR < 1% and log2(enrichment score) ≥ 1. Overlapping AK, XZ, MML and MB conceived the project, designed the experiments, and positive windows were merged. The following four wrote the manuscript. MML and XZ performed the experiments with the help of WW and SPP; HLS and CH provided sequencing and mass spectrometry numbers were calculated to obtain the enrichment score analyses; XZ designed and QFP performed the bioinformatics analysis. AK, XZ of each peak (or window): read count of the IP sample in and MML drafted the manuscript with substantial input from all co-authors. the current peak/window (a); median read count of the IP All authors approved the paper. sample in all 100-nucleotide windows on the current Ethics approval and consent to participate mRNA (b); read count of the input sample in the current All mouse experiments were approved by the Norwegian Animal Research peak/window (c); and median read count of the input Authority by Norwegian Food Safety Authority and done in accordance with institutional guidelines at the Centre for Comparative Medicine at Oslo sample in all 100-nucleotident windows on the current University Hospital. Animal work was conducted in accordance with the mRNA (d). The enrichment score of each window was rules and regulations of the Federation of European Laboratory Animal calculated as (a × d)/(b × c). Common peaks shared in the Science Association’s (FELASA). triplicates of a sample were kept with the peak annotation Competing interests from replicate 1. C. H. is a scientific founder of Accent Therapeutics, Inc. The other authors For motif analysis, consensus motif was determined by declare that they have no competing interests. using HOMER4. For GO analysis, differentially expressed genes or m A-modified genes were uploaded to DAVID Publisher’sNote (http://david.abcc.ncifcrf.gov/). The GO terms were Springer Nature remains neutral with regard to jurisdictional claims in ranked and presented according to −log2(P value). published maps and institutional affiliations. Li et al. Genome Biology (2018) 19:69 Page 15 of 16 Author details 20. Chen T, Hao YJ, Zhang Y, Li MM, Wang M, Han W, et al. m(6)A RNA Department of Microbiology, Oslo University Hospital, Rikshospitalet, methylation is regulated by microRNAs and promotes reprogramming to NO-0027 Oslo, Norway. Department of Molecular Medicine, Institute of Basic pluripotency. Cell Stem Cell. 2015;16:289–301. Medical Sciences, University of Oslo, NO-0317 Oslo, Norway. Department of 21. Zhang C, Chen Y, Sun B, Wang L, Yang Y, Ma D, et al. m6A modulates Clinical and Molecular Medicine, Norwegian University of Science and haematopoietic stem and progenitor cell specification. Nature. 2017;549: Technology (NTNU), Trondheim, Norway. Department of Chemistry, Institute 273–6. for Biophysical Dynamics, The University of Chicago, 929 East 57th Street, 22. Vu LP, Pickering BF, Cheng Y, Zaccara S, Nguyen D, Minuesa G, et al. The Chicago, IL 60637, USA. Howard Hughes Medical Institute, The University of N6-methyladenosine (m6A)-forming enzyme METTL3 controls myeloid Chicago, 929 East 57th Street, Chicago, IL 60637, USA. Department of differentiation of normal hematopoietic and leukemia cells. Nat Med. 2017; Computational Biology, St. Jude Children’s Hospital, Memphis, TN 38105, 23:1369–76. USA. 23. Shi H, Wang X, Lu Z, Zhao BS, Ma H, Hsu PJ, et al. YTHDF3 facilitates translation and decay of N6-methyladenosine-modified RNA. Cell Res. 2017;27:315–28. Received: 3 December 2017 Accepted: 26 April 2018 24. Alarcon CR, Goodarzi H, Lee H, Liu X, Tavazoie S, Tavazoie SF. HNRNPA2B1 Is a Mediator of m(6)A-Dependent Nuclear RNA Processing Events. Cell. 2015; 162:1299–308. 25. Ivanova I, Much C, Di Giacomo M, Azzi C, Morgan M, Moreira PN, et al. The References RNA m6A reader YTHDF2 is essential for the post-transcriptional regulation 1. Roundtree IA, Evans ME, Pan T, He C. Dynamic RNA modifications in gene of the maternal transcriptome and oocyte competence. Mol Cell. 2017;67(6): expression regulation. Cell. 2017;169:1187–200. 1059–67. 2. Saletore Y, Meyer K, Korlach J, Vilfan ID, Jaffrey S, Mason CE. The birth of the 26. Lence T, Akhtar J, Bayer M, Schmid K, Spindler L, Ho CH, et al. m6A Epitranscriptome: deciphering the function of RNA modifications. Genome modulates neuronal functions and sex determination in Drosophila. Nature. Biol. 2012;13:175. 2016;540:242–7. 3. Jia GF, Fu Y, Zhao X, Dai Q, Zheng GQ, Yang Y, et al. N6-Methyladenosine in 27. Yoon KJ, Ringeling FR, Vissers C, Jacob F, Pokrass M, Jimenez-Cyrus D, et al. nuclear RNA is a major substrate of the obesity-associated FTO. Nat Chem Temporal control of mammalian cortical neurogenesis by m6A methylation. Biol. 2011;7:885–7. Cell. 2017;171(4):877–89. 4. Desrosiers RC, Friderici KH, Rottman FM. Characterization of Novikoff 28. Gotz M, Huttner WB. The cell biology of neurogenesis. Nat Rev Mol Cell Biol. hepatoma mRNA methylation and heterogeneity in the methylated 5′ 2005;6:777–88. terminus. Biochemistry. 1975;14:4367–74. 29. Nordstrand LM, Svard J, Larsen E, Nilsen A, Ougland R, Furu K, et al. Mice 5. Schwartz S, Mumbach MR, Jovanovic M, Wang T, Maciag K, Bushkin GG, et lacking Alkbh1 display sex-ratio distortion and unilateral eye defects. PLoS al. Perturbation of m6A writers reveals two distinct classes of mRNA One. 2010;5:e13827. methylation at internal and 5′ sites. Cell Rep. 2014;8:284–96. 30. Castranio T, Mishina Y. Bmp2 is required for cephalic neural tube closure in 6. Zheng G, Dahl JA, Niu Y, Fedorcsak P, Huang CM, Li CJ, et al. ALKBH5 is a the mouse. Dev Dyn. 2009;238:110–22. mammalian RNA demethylase that impacts RNA metabolism and mouse 31. Lizarraga SB, Margossian SP, Harris MH, Campagna DR, Han AP, Blevins S, et fertility. Mol Cell. 2013;49:18–29. al. Cdk5rap2 regulates centrosome function and chromosome segregation 7. Wang X, Lu Z, Gomez A, Hon GC, Yue Y, Han D, et al. N6-methyladenosine- in neuronal progenitors. Development. 2010;137:1907–17. dependent regulation of messenger RNA stability. Nature. 2014;505:117–20. 32. Englund C, Fink A, Lau C, Pham D, Daza RA, Bulfone A, et al. Pax6, Tbr2, and 8. Xiao W, Adhikari S, Dahal U, Chen YS, Hao YJ, Sun BF, et al. Nuclear m(6)A Tbr1 are expressed sequentially by radial glia, intermediate progenitor Reader YTHDC1 Regulates mRNA Splicing. Mol Cell. 2016;61:507–19. cells, and postmitotic neurons in developing neocortex. J Neurosci. 9. Hsu PJ, Zhu Y, Ma H, Guo Y, Shi X, Liu Y, et al. Ythdc2 is an N6- 2005;25:247–51. methyladenosine binding protein that regulates mammalian 33. Kang W, Wong LC, Shi SH, Hebert JM. The transition from radial glial to spermatogenesis. Cell Res. 2017;27:1115–27. intermediate progenitor cell is inhibited by FGF signaling during 10. Dominissini D, Moshitch-Moshkovitz S, Schwartz S, Salmon-Divon M, Ungar corticogenesis. J Neurosci. 2009;29:14571–80. L, Osenberg S, et al. Topology of the human and mouse m6A RNA 34. Taverna E, Gotz M, Huttner WB. The cell biology of neurogenesis: toward an methylomes revealed by m6A-seq. Nature. 2012;485:201–6. understanding of the development and evolution of the neocortex. Annu 11. Meyer KD, Saletore Y, Zumbo P, Elemento O, Mason CE, Jaffrey SR. Rev Cell Dev Biol. 2014;30:465–502. Comprehensive analysis of mRNA methylation reveals enrichment in 3' 35. Chen H, LinW, Zhang Y,Lin L,Chen J,ZengY,etal. IL-10promotes UTRs and near stop codons. Cell. 2012;149:1635–46. neurite outgrowth and synapse formation in cultured cortical neurons 12. Shah JC, Clancy MJ. IME4, a gene that mediates MAT and nutritional after the oxygen-glucose deprivation via JAK1/STAT3 pathway. Sci Rep. control of meiosis in Saccharomyces cerevisiae. Mol Cell Biol. 1992;12: 2016;6:30459. 1078–86. 36. Jomova K, Valko M. Advances in metal-induced oxidative stress and human 13. Zhao BS, Wang X, Beadell AV, Lu Z, Shi H, Kuuspalu A, et al. m6A- disease. Toxicology. 2011;283:65–87. dependent maternal mRNA clearance facilitates zebrafish maternal-to- 37. Mishra D, Flora SJ. Differential oxidative stress and DNA damage in rat brain zygotic transition. Nature. 2017;542:475–8. regions and blood following chronic arsenic exposure. Toxicol Ind Health. 14. Zhao X, Yang Y, Sun BF, Shi Y, Yang X, Xiao W, et al. FTO-dependent 2008;24:247–56. demethylation of N6-methyladenosine regulates mRNA splicing and is 38. Justicia C, Gabriel C, Planas AM. Activation of the JAK/STAT pathway required for adipogenesis. Cell Res. 2014;24:1403–19. following transient focal cerebral ischemia: signaling through Jak1 and Stat3 15. Boissel S, Reish O, Proulx K, Kawagoe-Takaki H, Sedgwick B, Yeo GS, et al. in astrocytes. Glia. 2000;30:253–70. Loss-of-function mutation in the dioxygenase-encoding FTO gene causes 39. Planas AM, Soriano MA, Berruezo M, Justicia C, Estrada A, Pitarch S, et al. severe growth retardation and multiple malformations. Am J Hum Genet. Induction of Stat3, a signal transducer and transcription factor, in reactive 2009;85:106–11. microglia following transient focal cerebral ischaemia. Eur J Neurosci. 1996; 16. Fustin JM, Doi M, Yamaguchi Y, Hida H, Nishimura S, Yoshida M, et al. RNA- 8:2612–8. methylation-dependent RNA processing controls the speed of the circadian 40. Kriegstein A, Noctor S, Martinez-Cerdeno V. Patterns of neural stem and clock. Cell. 2013;155:793–806. progenitor cell division may underlie evolutionary cortical expansion. Nat 17. Wang Y, Li Y, Toth JI, Petroski MD, Zhang Z, Zhao JC. N(6)-methyladenosine Rev Neurosci. 2006;7:883–90. modification destabilizes developmental regulators in embryonic stem cells. 41. Merkle FT, Alvarez-Buylla A. Neural stem cells in mammalian development. Nat Cell Biol. 2014;16:191–8. Curr Opin Cell Biol. 2006;18:704–9. 18. Batista PJ, Molinie B, Wang J, Qu K, Zhang J, Li L, et al. m(6)A RNA modification controls cell fate transition in mammalian embryonic stem 42. Conde C, Caceres A. Microtubule assembly, organization and dynamics in cells. Cell Stem Cell. 2014;15:707–19. axons and dendrites. Nat Rev Neurosci. 2009;10:319–32. 19. Geula S, Moshitch-Moshkovitz S, Dominissini D, Mansour AA, Kol N, Salmon- 43. Du H, Zhao Y, He J, Zhang Y, Xi H, Liu M, et al. YTHDF2 destabilizes m(6)A- Divon M, et al. m6A mRNA methylation facilitates resolution of naive containing RNA through direct recruitment of the CCR4-NOT deadenylase pluripotency toward differentiation. Science. 2015;347(6225):1002–6. complex. Nat Commun. 2016;7:12626. Li et al. Genome Biology (2018) 19:69 Page 16 of 16 44. Weng YL, Wang X, An R, Cassin J, Vissers C, Liu Y, et al. Epitranscriptomic m6A regulation of axon regeneration in the adult mammalian nervous system. Neuron. 2018;97:313–25. 45. Wang Y, Li Y, Yue M, Wang J, Kumar S, Wechsler-Reya RJ, et al. N6- methyladenosine RNA modification regulates embryonic neural stem cell self-renewal through histone modifications. Nat Neurosci. 2018;21:195–206. 46. Peritz T, Zeng F, Kannanayakal TJ, Kilk K, Eiriksdottir E, Langel U, et al. Immunoprecipitation of mRNA-protein complexes. Nat Protoc. 2006;1:577–80. 47. Li M, Zhao X, Wang W, Shi H, Pan Q, Lu Z, et al. Ythdf2-mediated m6A mRNA clearance modulates neural development in mice. Data sets. NCBI GEO. 2018. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE104867. Accessed 1 Jan 2018.
– Springer Journals
Published: May 31, 2018