RNA 5-Methylcytosine Facilitates the Maternal-to-Zygotic Transition by Preventing Maternal mRNA Decay.

Article RNA 5-Methylcytosine Fac ilitates the Maternal-to-

Zygotic Transition by Preventing Maternal
mRNA Decay

Graphical Abstract

d RNA-BisSeq revealed a dynamic RNA m5C landscape during zebrafish embryogenesis d Ybx1 preferentially recognizes m5C-modified mRNAs d Ybx1 deficiency leads to early gastrulation defects in zebrafish embryos d Ybx1 and Pabpc1a coordinately regulate m5C-modified maternal mRNA stability Yang et al., 2019, Molecular Cell 75, 1–15 September 19, 2019 ª 2019 Elsevier Inc. https://doi.org/10.1016/j.molcel.2019.06.033 Authors Ying Yang, Lu Wang, Xiao Han, ..., Jinbiao Ma, Feng Liu, Yun-Gui Yang Correspondence majb@fudan.edu.cn (J.M.), liuf@ioz.ac.cn (F.L.), ygyang@big.ac.cn (Y.-G.Y.) In Brief RNA modifications exert important effects in many critical physiological processes. Using RNA-BisSeq, Yang et al. provide a comprehensive view of the RNA m5C landscape in zebrafish early embryos and show that m5C-modified maternal mRNAs are stabilized by Ybx1 and Pabpc1a during zebrafish MZT.

Molecular Cell

RNA 5-Methylcytosine Facilitates the Maternal-to-Zygotic Transition by Preventing Maternal mRNA Decay Ying Yang,1,3,4,5,12 Lu Wang,2,11,12 Xiao Han,1,4,12 Wen-Lan Yang,1,4,12 Mengmeng Zhang,6,12 Hai-Li Ma,1,3 Bao-Fa Sun,1,3,4,5 Ang Li,1,3 Jun Xia,2,3 Jing Chen,1,3 Jian Heng,2,3 Baixing Wu,6 Yu-Sheng Chen,1,3 Jia-Wei Xu,7 Xin Yang,1,3 Huan Yao,1,3 Jiawei Sun,8 Cong Lyu,3,9 Hai-Lin Wang,3,4,9 Ying Huang,10 Ying-Pu Sun,7 Yong-Liang Zhao,1,3,4,5 Anming Meng,8 Jinbiao Ma,6,* Feng Liu,2,3,4,* and Yun-Gui Yang1,3,4,5,13,* 1CAS Key Laboratory of Genomic and Precision Medicine, Collaborative Innovation Center of Genetics and Development, College of Future Technology, Beijing Institute of Genomics, Chinese Academy of Sciences, Beijing 100101, China 2State Key Laboratory of Membrane Biology, Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China 3University of Chinese Academy of Sciences, Beijing 100049, China 4Sino-Danish College, University of Chinese Academy of Sciences, Beijing 101408, China 5Institute of Stem Cell and Regeneration, Chinese Academy of Sciences, Beijing 100101, China 6State Key Laboratory of Genetic Engineering, Collaborative Innovation Center of Genetics and Development, Multiscale Research Institute for Complex Systems, Department of Biochemistry, School of Life Sciences, Fudan University, Shanghai 200438, China 7Center for Reproductive Medicine, Henan Key Laboratory of Reproduction and Genetics, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan, 450052, China 8Laboratory of Molecular Developmental Biology, State Key Laboratory of Membrane Biology, Tsinghua-Peking Center for Life Sciences, School of Life Sciences, Tsinghua University, Beijing 100084, China 9State Key Laboratory of Environmental Chemistry and Ecotoxicology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China 10State Key Laboratory of Molecular Biology, Shanghai Key Laboratory of Molecular Andrology, Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences, University of Chinese Academy of Sciences, Shanghai 201204, China 11State Key Laboratory of Experimental Hematology, Institute of Hematology and Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin 300020, China 12These authors contributed equally 13Lead Contact *Correspondence: majb@fudan.edu.cn (J.M.), liuf@ioz.ac.cn (F.L.), ygyang@big.ac.cn (Y.-G.Y.) https://doi.org/10.1016/j.molcel.2019.06.033

The maternal-to-zygotic transition (MZT) is a conserved and fundamental process during which the maternal environment is converted to an environment of embryonic-driven development through dramatic reprogramming. However, how maternally supplied transcripts are dynamically regulated during MZT remains largely unknown. Herein, through genome-wide profiling of RNA 5-methylcytosine (m5C) modification in zebrafish early embryos, we found that m5C-modified maternal mRNAs display higher stability than non-m5C-modified mRNAs during MZT. We discovered that Y-box binding protein

1 (Ybx1) preferentially recognizes m5C-modified
mRNAs through p-p interactions with a key residue, Trp45, in Ybx1’s cold shock domain (CSD), which plays essential roles in maternal mRNA stability and early embryogenesis of zebrafish. Together with the mRNA stabilizer Pabpc1a, Ybx1 promotes the stability of its target mRNAs in an m5C-dependent manner. Our study demonstrates an unexpected mechanism of RNA m5C-regulated maternal mRNA stabilization during zebrafish MZT, highlighting the critical role of m5C mRNA modification in early development.

The maternal-to-zygotic transition (MZT), which involves maternal RNA and protein depletion and zygotic genome activation (ZGA), is one of the most important events during early embryogenesis (Abrams and Mullins, 2009; Tadros and Lipshitz, 2009; Yartseva and Giraldez, 2015). To ensure normal developmental reprogramming during MZT, early embryogenesis initiated bymaternally originatedmRNAs during the period of zygotic genome quiescence after fertilization is followed by a gradual switch to zygotic genetic control, a process accompanied by clearance of maternal RNAs and proteins (Tadros and Lipshitz, 2009; Walser and Lipshitz, 2011; Yartseva and Giraldez, 2015). Due to the transcriptional silencing of the zygotic genome during the initial developmental stages, the maintenance of maternal messages appears particularly critical. Recent studies have illustrated a series of epigenetic events during MZT, including global DNAmethylation, chromatin remodeling, genome reorganization, and downstream transcriptional changes (Eckersley-Maslin et al., 2018). As the early stages of animal embryogenesis are mainly Molecular Cell 75, 1–15, September 19, 2019 ª 2019 Elsevier Inc. 1 (legend on next page) 2 Molecular Cell 75, 1–15, September 19, 2019 driven by maternally deposited yet rapidly cleared mRNAs (Tadros and Lipshitz, 2009; Yartseva and Giraldez, 2015), posttranscriptional regulation, particular RNA epigenetic regulation, is essential for the fate of maternally supplied mRNAs in early development. In zebrafish, several factors have been indicated to be essential in regulating maternal mRNA decay throughmaternal and zygotic pathways, including the zygotically transcribed microRNA miR-430 (Bazzini et al., 2012; Giraldez et al., 2006), suboptimal codon usage (Bazzini et al., 2016; Mishima and Tomari, 2016), N6-methyladenosine (m6A) (Zhao et al., 2017), and uridylation (Chang et al., 2018). In addition, maternal factors, including Nanog, Pou5f1, and SoxB1, have been shown to regulate the zygotic gene activation process by initiating the zygotic developmental program and activating miR-430 expression while facilitating maternal mRNA degradation (Lee et al., 2013). During MZT, maternal mRNAs undergo degradation during three successive stages: fertilization (36%), ZGA (42%), and afterward (23%) (Mathavan et al., 2005; Walser and Lipshitz, 2011); these stages imply a developmentally controlled degradation process. However, the known factors mentioned above have been shown to accelerate the decay of only several hundreds of maternal mRNAs, suggesting that additional factors and pathways may exist to coordinately regulate the decay process of the majority of maternal mRNAs. A recent report showed that abundant and reversible mRNA m6A can promote the decay of a subset of maternal mRNAs dependent on the reader Ythdf2 (Zhao et al., 2017), suggesting a possible role of posttranscriptional RNA modification in the global regulation of maternal mRNA metabolism. 5-Methylcytosine (m5C) is another mRNA modification prevalently distributed not only on noncoding RNAs but also on mRNAs, as revealed by high-throughput sequencing approaches (Amort et al., 2017; Blanco et al., 2016; Blanco et al., 2014; David et al., 2017; Squires et al., 2012; Yang et al., 2017). m5C on mRNAs is catalyzed by the methyltransferase NOP2/Sun RNA methyltransferase family member 2 (NSUN2) in humans and by tRNA-specific methyltransferase 4B (TRM4B) in Arabidopsis (Cui et al., 2017; David et al., 2017; Hussain et al., 2013a, 2013b; Khoddami and Cairns, 2013; Squires et al., 2012; Yang et al., 2017). The modifications are distributed either in locations near to the Argonaute-binding regions within the 30 UTR (Squires et al., 2012) or in the vicinity of the translational start site of mRNA (Amort et al., 2017; Yang et al., 2017). The nuclear protein Aly/REF export factor (ALYREF) is an m5C reader and mediates nuclear export of methylated RNAs (Pfaff et al., 2018; Yang et al., 2017). Additionally, m5C also plays important roles in mouse embryonic stem cells, the mouse brain, and plant tissues (Amort et al., 2017; David et al., 2017). These findings provide reasonable evidence that m5C is probably involved in the complex regulation of mRNA metabolism and early developmental processes. In this study, we found that m5C regulates the stability of thousands ofmaternal mRNAs in a previously unidentifiedmechanism during early zebrafishMZT. Y-box binding protein 1 (Ybx1) is a specific cytoplasmic binding protein of the m5C that recruits poly(A) binding protein cytoplasmic 1a (Pabpc1a) to methylated transcripts to prevent their decay. Our integrated studies reveal a mechanism of m5C modification in gene expression control during early development.

m5C Maintains mRNA Stability during Zebrafish MZT To obtain the global profile of RNA m5C during early zebrafish embryogenesis, we performed RNA sequencing (RNA-seq) and RNA bisulfite sequencing (RNA-BisSeq) on wild-type zebrafish embryos at different stages of early embryogenesis (Figure 1A; Table S1). The RNA-BisSeq method has a dependable conversion rate of C-to-T mutations (Figure S1A) and good repeatability (Figures S1B–S1D) (see details in STARMethods). A total of 2,902–7,521 m5C sites within 1,300–3,741 mRNAs (712–2,324maternal mRNAs) were identified at each stage during embryo development (Figure 1B; Table S1). In addition, the results of ultra-high-performance liquid chromatography-triple quadrupole mass spectrometry coupled with multiple-reaction monitoring (UHPLC-MRM-MS/MS) analysis showed a consistent trend of mRNA m5C modifications (Figure S1E) as the global landscape changes derived from RNA-BisSeq (Figure 1B). Interestingly, in zebrafish embryos, mRNA m5C sites were not remarkably enriched in CG-rich regions (36.07% on average) (Figures S1F and S1G), whereas they displayed distinct distribution, with the majority of m5C sites being enriched and homogeneously distributed in CDS regions (Figures S1H and S1I); these features are apparently unlike those reported for other vertebrates (Amort et al., 2017; Yang et al., 2017) (Figure S1J) and suggest a species-specific m5C distribution. The global mRNA m5C site distributive features in zebrafish embryos prompted us to examine the potential role of m5C in regulating maternal mRNA stability. Upon comparing RNA abundance changes between m5C-modified and unmodified maternal mRNAs during the stages from 2 to 4 h post-fertilization (hpf) and from 4 to 6 hpf, we found that the unmodified mRNAs were more downregulated than the m5C-modified mRNAs over both time spans (p % 7.733 10 12), indicating that m5C modification may stabilize maternal mRNAs during MZT (i.e., 3 hpf) (Figure 1C; Table S1). To further verify the association of m5C with RNA stabilization in zebrafish, we performedRNA-seq on embryos treatedwith the RNA polymerase II inhibitor a-amanitin and calculated the RNA half-lives (see STAR Methods). The results showed that m5Cmodified mRNAs displayed longer half-lives than unmodified mRNAs in embryos at 2–6 hpf (p < 2.20 3 10 16, Figure 1D). Consistently, among the randomly selected representative mRNAs, we observed longer half-lives for m5C-modified mRNAs than for unmodified ones (Figure 1E). Moreover, m5C RNAimmunoprecipitation (RIP)-qPCR verified the enrichment of m5C on methylated rather than unmethylated mRNAs (Figure S1K). We further found that mRNAs with methylation levels that tend to be increased from 2 to 4 hpf or from 4 to 6 hpf exhibited significantly smaller changes in RNA abundance than those with stable methylation levels (Figure 1F) (see STAR Methods). In addition, mRNAs with higher methylation levels and more methylated sites displayed more stable expression levels during the above two stages, and this association was independent of coverage changes (Figures 1G and S1L), suggesting that RNA m5C could maintain mRNA stability in a site- and level-dependent manner during MZT. Upon analyzing the functions of m5C-modified mRNAs, we found that mRNAs involved inmRNAmetabolism and cell-cycle-related processes remained stable during MZT (Figure 1H). Moreover, there was a significant correlation between the RiboMinus RNA-seq gene expression levels and the mRNA-seq data from embryos at 4 hpf (p < 2.20 3 10 16; Figure S1M), excluding possible biases introduced by the poly(A)-tail-lengthbased oligo(dT) enrichment used to generate the mRNA-seq libraries. Additionally, combinatorial analysis of the published m5C transcriptome and mRNA half-life data for humans and mice (Amort et al., 2017; Schwanh€ausser et al., 2011; Squires et al., 2012; Wang et al., 2014; Yang et al., 2017) showed that m5C positively correlates with mRNA half-life in mammals (Figure S1N), which is consistent with our findings in zebrafish in this study and suggests that m5C-mediated mRNA stabilization is a general phenomenon across vertebrates. The zygotically transcribed microRNA miR-430 participates in the clearance of maternal RNAs during zebrafish embryogenesis (Bazzini et al., 2012; Giraldez et al., 2006). To explore whether miR-430 affects m5C-dependent stabilization, we overlapped the m5C-modified mRNAs during MZT with miR-430-targeted mRNAs and found that the majority of m5C-modified mRNAs (2,911/2,925) were not targeted by miR-430 (Figure S1O), suggesting distinct mechanisms for maternal mRNA stabilization by m5C modification and miR-430. Ybx1 Is an m5C-Binding Protein in Zebrafish Since ALYREF is the only m5C-binding protein identified thus far to participate in nuclear export of m5C-modified mRNA in mam- mals (Yang et al., 2017), other binding proteins probably exist to mediate m5C-regulated mRNA stability. To search for other m5C-binding proteins in zebrafish, we carried out RNA affinity chromatography and mass spectrometry (MS) analyses using biotin-labeled oligonucleotides with or without m5C from zebrafish embryo lysates (Table S2) and identified several enriched proteins with m5C-containing oligonucleotides (Figures S2A and S2B). As human Ybx1 has been reported to be involved in mRNA stabilization (Evdokimova et al., 2001), we focused on this specific protein for further analyses. Both in vivo and in vitro pull-down assays showed that the Ybx1 protein can be efficiently pulled down by m5C-modified oligos compared with the unmodified control oligos (Figures S2C and S2D), a finding that was further validated by an electrophoretic mobility shift assay (EMSA) (Figure 2A). Consistently, an m5C-containing 6-mer RNA oligo was shown to bind to the Ybx1 protein cold shock domain (CSD), which recognizes RNA substrates in a sequence-dependent manner and contains a CAUC high-affinity binding motif derived from systematic evolution of ligands by exponential enrichment (SELEX) (Manival et al., 2001; Wei et al., 2012); the methylated 6-mer bound with a 3-fold higher binding affinity (KD) than the unmethylated RNA oligo, as measured by isothermal titration calorimetry (ITC) (Figure 2B), suggesting that Ybx1 has a much higher binding affinity for m5C than for cytosine (C). To assess the molecular basis of the enhanced binding of Ybx1 to m5C-modified RNAs, we solved the crystal structure of the Ybx1 CSD in complex with a 6-mer RNA oligo containing an m5C modification (Tables S3 and S4). Five nucleotides (50-C2A3U4m 5C5U6-3 0) are clearly visible in the crystal structure, of which four nucleotides (50-C2A3U4m 5C5-3 0) directly bind to the surface of the Ybx1 CSD; U6 is stacked over the base of m5C (Figure 2C). The bases C2 and A3 interact with the side chains of the conserved residues His67 and Phe65, respectively, via p-p stacking (Figures S2E and S2F). N3 and N4 of C2 base are bound by Glu97 and Glu101 side chains through a zinc ion and a coordinated water molecule (Figure S2F). N1 and N6 of A3 base form hydrogen bonds with carboxyl groups of main chains of Ala100 and Gly99, respectively. The side chain of Lys98 interacts with the nonbridging oxygen atom of the A3 phosphate group. U4 is stacked over the aromatic side chains of conserved residues of Phe54 (Figures 2C and 2D). N3 and O2 of U4 base form hydrogen bonds with side chains of Asp63 and Trp45 (W45), respectively, O2 of m5C base interacts with the side chain of Asn47, and the 20-OH of the m5C ribose forms a hydrogen bond with the side chain of Asn50 (Figure 2D). The large aromatic side chain of W45 that stacks over the m5C base with a p-p interaction may account for the 3-fold higher binding affinity of the CSD for m5C than for unmodified C. To verify this hypothesis, we mutated W45 to Ala (A) and Phe (F) (Figure S2G) and then measured the binding affinities of the mutants with m5C-modified or unmodified RNA oligos using ITC. TheW45Amutant had no binding affinity for either m5C-modified or unmodified oligos (Figure S2H). Interestingly, the W45F mutant not only had a significantly reduced binding affinity for the m5C-modified RNA oligo compared to W45 but also exhibited an attenuated ability to discriminate between m5C-modified and unmodified RNA oligos (Figure S2I), probably due to a 4 Molecular Cell 75, 1–15, September 19, 2019 reduced p-p interaction between the aromatic side chain and base of m5C with F compared to W, further confirming that W45 is the key residue for m5C recognition. Moreover, we performed Ybx1 RIP-seq (RIP combined with high-throughput sequencing) and individual-nucleotide-resolution UV cross-linking and immunoprecipitation sequencing (iCLIP-seq) to further identify the mRNAs bound by Ybx1, and 3,044 genes were ultimately identified (Figure 2E). We found that most of the Ybx1-binding sites were highly enriched in CDS regions (81.40% on average; Figures S2J and S2K), similar to the m5C distribution pattern (Figures S1I and S1J). In addition, 54.27% (1,652) of the Ybx1-binding mRNAs were detected to be modified by m5C (Figure 2E). We then performed Ybx1 RIP-BisSeq to confirm the preferential interaction and direct association between Ybx1 and m5C. Approximately 77.9% of the m5C sites were localized within the Ybx1 peaks (7,084 and 6,947 m5C sites at 0 and 3 hpf, respectively) (Figure S2L), and the methylation levels of the m5C sites on Ybx1-binding mRNAs were Molecular Cell 75, 1–15, September 19, 2019 5 significantly higher than those on inputmRNAs (p < 2.203 10 16, Figure 2F), supporting the preferential binding of Ybx1 to m5Cmodified mRNAs. Collectively, these findings revealed that Ybx1 is a specific mRNA m5C-binding protein in zebrafish.

Ybx1 Deficiency Leads to Early Gastrulation Arrest during Zebrafish Embryogenesis
To further explore the function of Ybx1-mediated m5C modification during early embryogenesis, we first analyzed the expression pattern of Ybx1 in zebrafish early embryos. Both wholemount in situ hybridization and RNA-seq analyses demonstrated a persistent expression of maternal ybx1mRNA across different developmental stages (Figures S3A and S3B). Consistently, we also observed persistent protein expression of Ybx1 during early embryogenesis (Figure 3A). Next, we generated ybx1mutant using CRISPR/Cas9 to examine the consequences of ybx1 loss of function at early stages of zebrafish embryogenesis (Figure S3C). Both ybx1 mRNA and endogenous Ybx1 protein in the zygotic ybx1 homozygous mutants and in maternal ybx1 / embryos were absent, which suggests that this is a null allele (Figures S3D–S3F). We observed severe developmental arrest at the shield stage (6 hpf) in maternal ybx1 / embryos (Figure 3B), leading to lethality of ybx1 homozygous mutants at 8 hpf; these findings indicated developmental defects owing to the loss of ybx1. To further confirm the results obtained with the mutants, we blocked Ybx1 translation by injecting embryos with two different morpholinos (MOs) and found that bothMOs decreased endogenous Ybx1 expression (Figures S3G and S3H) and induced early gastrulation arrest at the sphere stage (Figures S3I and S3J), similar to the observed effects in maternal ybx1 / mutants. To delineate the underlying molecular mechanism, we conducted RNA-seq in ybx1morphants and control embryos (Table S1). We found that upon ybx1 knockdown, the number of significantly downregulated genes was much more than the upregulated ones, with 1,842 versus 1,072 at 4 hpf and 2,739 versus 1,224 at 6 hpf (Figure S4A). In addition, the downregulated genes were highly enriched in processes related to localization establishment, cellular component organization, and cell-cell recognition (Figure S4B). Furthermore, we observed that the RNA abundance changes (from 4 to 6 hpf) in the ybx1morphants were markedly greater than those in the wild-type embryos (p < 2.20 3 10 16), suggesting that Ybx1 deficiency led to accelerated mRNA decay (Figure 3C). Furthermore, the number of accelerated-decay genes (6,339) was over 6-fold more than the number of slow-decay genes (965) at 4 hpf (Figure S4C). Additionally, we identified 397 m5C-modified maternal mRNAs showing significant downregulation from 3 to 6 hpf in ybx1 morphants compared with control embryos (p < 3.89 3 10 5) (Figure S4D). We next screened the Ybx1-binding mRNAs with stringent criteria of downregulation in ybx1 morphants and simultaneous m5C modification. The regulation of these mRNAs by Ybx1 was further verified by qRT-PCR, which demonstrated that the levels of these mRNAs were decreased significantly upon ybx1 deficiency (Figure S4E). Consistently, overexpression of wild-type ybx1 mRNA (without ybx1 UTR MO- or ATG MObinding sites), but not W45A or W45F mutant mRNA, could reverse the developmental arrest in ybx1-deficient zebrafish em- bryos at the blastula and gastrula stages (Figures S2G, 3D, and 3E). Taken together, these results convincingly demonstrate that Ybx1 plays essential roles in early embryogenesis that are dependent on its ability to recognize m5C and subsequently stabilize mRNA. Ybx1 Recruits Pabpc1a to Maintain the Stability of m5CModified Maternal mRNAs To investigate the detailed mechanism of Ybx1-mediated mRNA stabilization, we next sought to identify the interacting partners of Ybx1 that participate in Ybx1-regulated maternal mRNA stabilization by immunoprecipitation combined with MS analysis (IP/ MS) of zebrafish embryo lysates (Table S2), and we identified Pabpc1a as a potential Ybx1 partner (Figure 4A). PABP family proteins have been shown to recruit distinct proteins to facilitate mRNAstabilization or degradation throughbinding to poly(A) tails (Bernstein et al., 1989; Bernstein and Ross, 1989; Mangus et al., 2003; Nicholson and Pasquinelli, 2019; Smith et al., 2014). We further demonstrated a direct interaction between Pabpc1a and Ybx1 in zebrafish through coimmunoprecipitation (co-IP) (Figure 4B) and in vitro glutathione S-transferase (GST) pulldown assays (Figure 4C), consistent with findings in mammals (Higashi et al., 2011; Skabkina et al., 2003). Moreover, we identified high levels ofpabpc1a coexpressionwith ybx1 in early zebrafish embryos (R= 0.80, p = 5.403 10 3; Figures 4D, 4E, andS4F). Through RNA-seq, 1,432 and 1,205 maternal genes were identified to be downregulated in pabpc1a morphants compared to control embryosat 4 and6hpf, respectively (FigureS4G).Consistent with the results of Ybx1 deficiency, Pabpc1a depletion led to acceleratedmRNAdegradation (p < 2.203 10 16; FigureS4H). In addition, among the genes downregulated upon Ybx1 and Pabpc1a deficiency, 646 and 596 maternal genes overlapped at 4 and 6 hpf, respectively, suggesting that the stability of these genes is commonly regulated by Ybx1 and Pabpc1a (Figures 4F and 4G). Among all downregulated genes at these two stages after depletion of Ybx1 or Pabpc1a, 56.52% were coregulated genes, including development-associated genes such as cap1, tex2, and tpp2 (Figures 4H, S4I, and S4J). To further test whether the RNA-binding affinity of Pabpc1a is regulated by Ybx1, we performed photoactivatable ribonucleoside-enhancedcrosslinkingand immunoprecipitation (PAR-CLIP) and in vitro RNA end biotin-labeling assays on control and Ybx1deficient ZF4 zebrafish embryonic fibroblast cells. TheRNA-binding affinity of Pabpc1a was substantially decreased upon Ybx1 deficiency (Figure 4I); this finding was further validated by Pabpc1a RIP-seq in both control embryos and ybx1 morphants to examine the global changes in Pabpc1a-binding RNAs. Comparedwith themRNAsboundbyPabpc1a alone, themRNAs commonly bound by Ybx1 and Pabpc1a showed significantly decreased Pabpc1a binding enrichment upon Ybx1 deficiency (p < 2.20 3 10 16, Figure 4J), suggesting that Pabpc1a binds to mRNAs through association with Ybx1. Importantly, we observed similar developmental arrest in both pabpc1a morphants and embryos coinjected with pabpc1a and ybx1 MOs upon validating the knockdown efficiency of the pabpc1a ATG MO (Figures 5A and S5A). qRT-PCR analysis showed that the mRNA stability of representative Ybx1-binding transcripts was decreased in morphants compared to control 6 Molecular Cell 75, 1–15, September 19, 2019 embryos (Figure 5B). Insufficient recognition of methylated maternal transcripts due to depletion of functional Ybx1 or Pabpc1a may lead to developmental arrest during early zebrafish development. To validate the phenotype of pabpc1amorphants injectedwith the ATG MO, we used a second splice MO and observed developmental arrest similar to that in the ATG MO-treated embryos (Figures S5A and S5B). Additionally, pabpc1a morphants (ATG Molecular Cell 75, 1–15, September 19, 2019 7 (A) Scatterplot showing the proteins bound to endogenous zebrafish Ybx1 (red). The plot is based on the average peptide numbers of proteins detected in both replicates. Enriched Pabpc1a is highlighted in green. (B) A Co-IP assay revealed the interaction between Ybx1 and Pabpc1a. IP was performed using anti-Myc (top) or anti-FLAG (bottom) beads as indicated. The IP and pull-down efficiency was detected using anti-FLAG or anti-Myc antibodies. RNase A was used to digest endogenous RNAs. (C) Recombinant FLAG-DrPabpc1a protein was mixed with either GST or GST-DrYbx1 protein as indicated, pulled down with GST beads, and subjected to immunoblotting with the indicated antibodies (anti-FLAG, anti-GST and anti-Ybx1). (D) Scatterplots showing the normalized RPKMof ybx1 and pabpc1a from two replicates. The Pearson correlation coefficient (R) and p values are shown in the top left corner. (legend continued on next page) 8 Molecular Cell 75, 1–15, September 19, 2019 MO-treated embryos) with reintroduction of pabpc1a mRNA containing mismatched pabpc1a ATGMO-binding sites showed partial attenuation of the defective phenotypes (Figures S5C and S5D). These data collectively demonstrated that the developmental defects were specifically induced by pabpc1a deficiency. To determine the direct involvement of m5C modification in mRNA stability maintenance, we performed a minigene reporter assay in which we tested the mRNA stability of a cap1 or tpp2 minigene with (wild type [WT]) or without an m5C site (mutant [MUT]) fused to a GFP tag (Figure S5E). The in vitro transcribed reporter mRNAs were injected into control embryos or ybx1 morphants at the one-cell stage andmonitored for their decay kinetics at 4 hpf by qRT-PCR. The results showed that the WT mRNAs of cap1 and tpp2 were degraded significantly faster in ybx1 morphants than in control embryos during MZT, whereas the stability of the MUT mRNAs of cap1 or tpp2 genes was not affected by the status of ybx1 (Figures 5C and S5F). Consistently, fluorescence density of WTmRNAs, but not MUTmRNAs, was significantly lower in ybx1 morphants than in control embryos during the sphere stage (Figures 5D, 5E, S5G, and S5H). We then examined the stability of reporter mRNAs over a time course and found that cap1-WT mRNA was degraded significantly faster in ybx1 morphants than in control embryos from 0 to 6 hpf, whereas the stability of cap1-MUT mRNA was not affected by ybx1 deficiency (Figures 5F and S5I). To provide further direct evidence to support the role of m5C modification in the stabilization of mRNAs, we used m5C-modified cap1 mRNA as a reference to construct another reporter through ligation of a 59-nt oligo 1 with or without m5C and a 25-nt oligo 2. Both oligos were synthesized in vitro, and a 15-nt GFP sequence was added to the 50 end of oligo 1 to facilitate subsequent qRT-PCR analysis. Through this strategy, we obtained two 84-nt reporter RNAs: cap1-C (with unmodified C) and cap1-m5C (Figure S5J). These reporter RNAswere then individually injected into embryos at the one-cell stage, and samples were collected at different time points (0, 2, 4, and 6 hpf) for RNA stability analysis. We found that the cap1-m5C reporter RNAwas significantly more stable than the cap1-C reporter RNA in control embryos (Figure 5G). As expected, there were no obvious differences in the stability of the cap1-m5C or cap1-C reporter RNA from 0 to 6 hpf in ybx1 morphants (Figure S5K) or pabpc1a morphants (Figure S5L). To examine the involvement of the poly(A) tail in the regulation of RNA stability, we next generated reporter RNAs using synthesized oligos containing modified m5C or unmodified C together with poly(A) tails of 30 residues (cap1-m5C-poly(A) and cap1C-poly(A); Figure S5J) and tested the in vivo stability of the re- porters in control embryos. Similar to the cap1-m5C RNA, the cap1-m5C-poly(A) RNA was much more stable than the unmodified control (cap1-C-poly(A)) (Figure S5M). In addition, cap1m5C-poly(A) showed a slightly higher stability than non-poly(A) cap1-m5C RNA (Figures 5G and S5M), suggesting that m5C cooperates with the poly(A) structure to enhance RNA stability. We further analyzed the in vivo binding of Ybx1 or Pabpc1a to the reporter RNAs in control embryos at 4 hpf and found that both Ybx1 and Pabpc1a could specifically bind to m5Cmodified RNA compared to unmodified RNA (Figure S5N). Taken together, these results strongly indicate that m5C promotes mRNA stability in a manner dependent on Ybx1 and Pabpc1a in zebrafish early embryogenesis (Figure 5H). Collectively, our findings demonstrate, for the first time, that the m5C modification and the m5C-binding protein Ybx1 control MZT during zebrafish early development by regulating maternal mRNA stability.

The MZT, which is composed of maternal message decay and ZGA, is one of themost important eventsduring early embryogenesis across vertebrates (Abrams and Mullins, 2009; Yartseva and Giraldez, 2015). RNAmodifications (m6A and uridylation), suboptimal codon usage, and miR-430-mediated zygotic mechanisms have been shown to promote the decay ofmaternalmRNAs (Bazzini et al., 2012, 2016; Chang et al., 2018; Giraldez et al., 2006; Laue et al., 2019; Lim et al., 2014; Mishima and Tomari, 2016; Zhao et al., 2017). However, maternal mRNAs decay differentially and dynamically during MZT, and how this process is regulated remains elusive. Our study demonstrates RNA m5C modification asa keymolecular eventmodulatingmaternalmRNAstability during zebrafishMZT.Approximately54.77%–63.41%ofm5C-modifiedmRNAswerematernalmRNAsatdifferent stagesofMZT, and these m5C-modified maternal mRNAs tended to be more stable and were differentially degraded at the appropriate times compared to the unmodified RNAs. We further found that upon fertilization, amaternal Ybx1 complex containing the poly(A) binding protein Pabpc1a preferentially recognizes m5C-modified maternal mRNAs and maintains their stability during MZT. Functional Gene Ontology (GO) analysis illustrated that these m5Cmodified mRNAs are mainly involved in cell-cycle processes and intracellular protein transport, suggesting a fundamental role of m5C-modified maternal mRNAs in maintaining early embryonic development. Overall, our findings demonstrate that m5C maintains the stability, rather than promoting the decay (as m6A does), of maternal mRNAs during MZT. (E) Double fluorescence in situ hybridization (dFISH) showing the colocalization of ybx1 and pabpc1a mRNAs. Scale bar, 200 mm. (F) Scatterplots showing the fold changes (log2) in maternal genes coregulated by Ybx1 and Pabpc1a. The Pearson correlation coefficient (R) and p values are shown on the top. (G) GO biological processes enriched for the significantly downregulated maternal genes at 4 and 6 hpf that were coregulated by Ybx1 and Pabpc1a. (H) Bar plot showing the fold changes (log2) in the expression of seven target genes in ybx1 and pabpc1a morphants. (I) PAR-CLIP assay showing decreased RNA pulled down by FLAG-Pabpc1a upon Ybx1 knockdown. RNA labeled with biotin at the 30 end (End Biotinylation Kit, Thermo Fisher) was visualized with a chemiluminescent nucleic acid detection module. (J) Cumulative distribution displaying the fold changes in Pabpc1a binding enrichment of mRNAs bound by Pabpc1a only or commonly bound by Ybx1 and Pabpc1a. p values were calculated using two-sided Wilcoxon and Mann-Whitney tests. See also Figure S4. Molecular Cell 75, 1–15, September 19, 2019 9 (legend continued on next page) 10 Molecular Cell 75, 1–15, September 19, 2019 RNA modifications have been recently demonstrated to be important posttranscriptional regulators (Gilbert et al., 2016; Roundtree et al., 2017). Notably, advances in sensitive detection techniques have made it possible to accurately identify these modifications on genome-wide scales (Boccaletto et al., 2018). m5C is one of the first RNAmodifications identified to be present in diverse RNA species, mainly tRNAs and rRNAs, with functions in translational regulation processes, such as protein synthesis and ribosome biogenesis (Blanco et al., 2016; Blanco et al., 2014; Schaefer et al., 2010; Schosserer et al., 2015; Sharma et al., 2013). Transcriptomic profiling has shown the differential distributive features of mRNA m5C in mammals and plants (Amort et al., 2017; David et al., 2017; Yang et al., 2017). In this study, we chose a well-established RNA-BisSeq method with an optimized library construction approach using ACT random hexamers that prevents annealing to inefficiently deaminated RNA templates and enables faithful identification of m5C sites (Yang et al., 2017). In conjunction with a proper control ensuring over a 99% conversion ratio in each sample, we also employed the specialized methylated RNA analysis ToolKit (meRanTK; version 1.2.0) (Rieder et al., 2016) with stringent parameters to align and calculated the methylation levels of m5C (see STAR Methods). Using this well-validated approach, we found that more than half of maternal mRNAs contained m5C modifications and that themajority ofm5C sites displayed species-specific distribution patterns, with enrichment and relatively homogeneous distribution in CDS regions. In particular, the reproducible results from simultaneous RNA-BisSeq of mixed samples of human HeLa cells and zebrafish embryos further support the credibility of our analysis approach. Ybx1 has been shown to bind to RNA through its CSD and stabilize RNA in humans (Evdokimova et al., 2001). We provide the first evidence that Ybx1 is an m5C-binding protein. The recognition of m5C-modified RNA by Ybx1 in which the m5C base lays on the surface of Ybx1, is similar to the structure of DNA 5mC that forms a methyl-Arg-G triad through van der Waals interactions (Ren et al., 2018). In contrast to m6A recognition by YTH family proteins, in which m6A inserts into a hydrophobic pocket and the methyl group faces toward a W residue to form methyl-p interactions (Xu et al., 2014; Zhu et al., 2014), m5C recognition occurs through an apparently different process in which the methyl-modified base stacks over the side chain of a large aromatic W45 residue via a parallel p-p interaction. Intriguingly, upon m5C recognition by Ybx1, the methyl group in m5C notably enhances the p-p interactions. We observed over 3-fold greater Ybx1 binding affinity for m5C-modified RNA than for unmodified control RNA, which is similar to the modest binding preference of MBD4 for DNA 5mC (Walavalkar et al., 2014). We further demonstrated that mutation of W45 to F, which has a smaller aromatic side chain with the potential to decrease the p-p interaction, markedly reduced both the binding affinity of Ybx1 for m5C-modified oligos and the ability of Ybx1 to discriminate between m5C-modified and unmodified RNA oligos, suggesting that the large aromatic side chain of W45 on Ybx1 is critical in mediating the ability of Ybx1 to bind to m5Cmodified RNA. Ybx1 has been reported to exhibit multiple biological functions, including pre-mRNA splicing, mRNA stabilization, and translation, through direct DNA and RNA binding or interaction with various other proteins (Eliseeva et al., 2011; Kumari et al., 2013; Lyabin et al., 2014; Zaucker et al., 2018). Specifically, the association of Ybx1 with tRNA-derived fragments has been implicated in stress responses and breast cancer progression (Goodarzi et al., 2015; Ivanov et al., 2011). In addition, one recent study using a zebrafish ybx1 mutant revealed the importance of Ybx1-regulated translation and protein loading in oocyte maturation, egg activation, and early embryogenesis (Sun et al., 2018). Thus, multiple mechanisms likely exist to coordinately mediate Ybx1’s functions. In our study, we found that the expression levels of many genes were altered upon ybx1 knockdown; among these genes, we focused only on those genes that were modified by m5C and further recognized by Ybx1. Through several biochemical and biological assays, we demonstrated that Ybx1 preferentially binds to m5C-modified mRNAs and regulates maternal mRNA stability during MZT through an m5C-dependent mechanism. In particular, the defective phenotypes in zebrafish development upon ybx1 knockdown could be attenuated only by wild-type ybx1 and not by ybx1 mRNA with a mutated m5C-binding site. Therefore, our findings reveal an additional layer of the mechanism of Ybx1 regulation in RNA metabolism and zebrafish embryogenesis. Through crystal structure analysis and in vitro RNA-protein interaction assays, we identified cytoplasmic Ybx1 as an m5Cbinding protein in zebrafish. Ybx1 specifically recognizes m5Cmodified mRNAs through p-p interactions between m5C and the indole ring of a key residue, W45, in its CSD and plays an essential role in maintaining mRNA stability. This mechanism is different from that of ALYREF. m5C recognition by ALYREF occurs mainly through a conserved positively charged residue, Lys171, and enables an mRNA-export-promoting function (Yang et al., 2017). These findings indicate that different m5C readers are involved in distinct RNA metabolic processes. Pabpc1a has been reported to be a poly(A)-binding protein that regulates RNA stability or degradation under physiological conditions (Bernstein et al., 1989; Bernstein and Ross, 1989; (G) qRT-PCR showing the expression of cap1- m5C (cyan) and cap1-C (black) reporter RNAs in control embryos at the indicated time points during MZT. RNA abundance was determined by qRT-PCR and normalized to 0 hpf values. b-Actin served as an internal RNA control. Error bars represent mean ± SD; n = 3. **p < 0.01, ***p < 0.001, two-tailed Student’s t test. (H) Schematic model of the dynamic regulation of maternal mRNA decay by m5C during zebrafishMZT. Maternal mRNAs without m5Cmodification are degraded rapidly through deadenylation during MZT, whereas a subset of maternal mRNAs with m5C modifications can be protected from degradation through m5C site recognition by the Ybx1 complex, which leads to the stabilization of mRNAs and subsequently enables the performance of their associated functions during early embryogenesis. The p values in (B), (C), and (E)–(G) were all determined using two-tailed Student’s t tests. See also Figure S5. Molecular Cell 75, 1–15, September 19, 2019 11 Mangus et al., 2003; Nicholson and Pasquinelli, 2019; Smith et al., 2014; Wang et al., 1999). In support of these previous findings, we demonstrated that Pabpc1a serves as an interacting partner of Ybx1 to promote the stability of m5C-modified mRNAs. Our RNA reporter assay further showed that the addition of 30 nt poly(A) tail to RNA oligos tends to enhance the stability of m5C-modified reporter RNA, but not the unmodified control RNA (Figures 5G andS5M), further suggesting the central role of Ybx1-recognition of m5C in the promotion of RNA stability. In conclusion, we provide strong evidence that m5C-modified maternal mRNA is specifically recognized by the Ybx1 protein and that this association controls the dynamic stability ofmaternal mRNAs. This dynamic stability temporally and spatially coordinates zebrafish early development and probably coordinates the developmental processes of other vertebrate systems aswell. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d KEY RESOURCES TABLE d LEAD CONTACT AND MATERIALS AVAILABILITY d EXPERIMENTAL MODEL AND SUBJECT DETAILS B Cell Lines B Zebrafish Strains and Embryo Collection d METHOD DETAILS B Plasmid Construction B RNA Preparation and mRNA Purification B Construction of RNA-Seq and RNA-BisSeq Libraries B m5C RIP-qPCR B UHPLC-MRM-MS/MS Analysis B In Vivo RNA Pulldown Assay and MS Analysis B Protein Purification in Mammalian Cells B Western Blotting B Electrophoretic Mobility Shift Assay (EMSA) B Protein Purification in E. coli Cells B ITC Measurements B Crystallization, Data Collection and Structure Determi- nation B RIP-Seq B iCLIP-Seq B Morpholinos, mRNA Synthesis and Microinjection B Generation ybx1 Mutant by CRISPR/Cas9 B WISH and dFISH B Immunofluorescence B Immunoprecipitation and Mass Spectrometry B Coimmunoprecipitation (Co-IP) B In Vitro GST Pulldown Assay B PAR-CLIP and In Vitro RNA End Biotin Labeling B mRNA Stability Analyses B Quantitative Reverse-Transcription PCR B Reporter mRNA Stability Assay B High-throughput Sequencing Data Processing B Stability Regulation Analysis d QUANTIFICATION AND STATISTICAL ANALYSIS d DATA AND CODE AVAILABILITY SUPPLEMENTAL INFORMATION Supplemental Information can be found online at https://doi.org/10.1016/j. molcel.2019.06.033.

We thank the staff from BL17U at Shanghai Synchrotron Radiation Facility (SSRF) for their assistance with data collection. We thank Xiang Ding and Mengmeng Zhang at the laboratory of Proteomics, core facility in the Institute of Biophysics, Chinese Academy of Sciences (CAS) for their technical support of liquid chromatography-mass spectrometry (LC-MS) analysis and the BIG CAS genomic platform for sequencing. We thank Professor Qiang Wang from the Institute of Zoology, CAS for donating the ZF4 cells. This work was supported by grants from the National Natural Science Fund for Distinguished Young Scholars (31625016); the Strategic Priority Research Program of the Chinese Academy of Sciences, China (XDA16000000); the National Natural Science Foundation of China (31830061 and 31230041); CAS Key Research Projects of the Frontier Science (QYZDY-SSW-SMC027); the National Key R&D Program of China, Stem Cell and Translational Research (2018YFA0109700); the Youth Innovation Promotion Association, CAS (2018133 and 2016097); the K. C. Wong Education Foundation, Shanghai Municipal Science and Technology Major Project (2017SHZDZX01); and the NSFC (consulting grant 91753000).

Y.Y. and W.-L.Y. performed most of the experiments with assistance from H.-L.M., J.C., A.L., and Y.-L.Z.; X.H. performed bioinformatics analysis with help from B.-F.S. and Y.-S.C.; L.W. performed experiments in zebrafish with help from J.X., J.S., and A.M.; M.Z. performed the crystal structure analyses and ITC assaywith help fromB.W.; Y.H. provided advice on structural analysis; J.H. contributed to collecting embryos; X.Y., H.Y., J.-W.X., and Y.-P.S. assisted with the high-throughput sequencing; C.L. and H.-L.W. assisted with the UHPLC-MRM-MS/MS analysis; and Y.-G.Y., F.L., and J.M. conceived this project, supervised the study, interpreted the data, and wrote and provided final approval of the manuscript.

The authors declare no competing interests. Received: November 29, 2018 Revised: April 15, 2019 Accepted: June 24, 2019 Published: August 6, 2019

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Mouse monoclonal 5-methylcytosine antibody [33D3] Abcam Cat#: ab10805; RRID: AB_442823 Rabbit polyclonal anti-Ybx1 This study, prepared by AbMax Biotechnology Co., Ltd Cat#: JYZ042-1 Rabbit anti-Flag-HRP Sigma-Aldrich Cat#: F7425; RRID: AB_439687 Rabbit anti-Myc-HRP Sigma-Aldrich Cat#: C3956; RRID: AB_439680 Mouse anti-GST monoclonal antibody GenScript Cat#: A00865; RRID: AB_914654 Rabbit anti-Actin Cell Signaling Cat#: 4967S; RRID: AB_330288 Rabbit polyclonal anti-Pabp1 Cell Signaling Cat#: 4992; RRID: AB_10693595 Anti-Digoxigenin (DIG)-AP, Fab fragments Roche Cat#: 11093274910; RRID: AB_514497 Anti-DIG-POD, Fab fragments Roche Cat#: 11633716001; RRID: AB_514499 Anti-Fluorescein(FLU)-POD, Fab fragments Roche Cat#: 11426346910; RRID: AB_840257 Alexa Fluor 488 Goat Anti-Rabbit IgG (H+L) Antibody Invitrogen Cat#: A-11034; RRID: AB_2576217 Bacterial and Virus Strains E. coli BL21(DE3) TransGen Biotech Cat#: CD601-03 E. coli strain C41(DE3) Lucigen Cat#: 60442-1 Chemicals, Peptides, and Recombinant Proteins Dulbecco’s MEM (DMEM) high glucose Thermo Fisher Cat#: C11995500BT DMEM/F12(1:1) (1 3 ) GIBCO Cat#:11320033 Penicillin-Streptomycin solution (100 3 ) GIBCO Cat#: 15140-122 Fetal bovine serum Excell Bio Cat#: FSS500 Fetal bovine serum GIBCO Cat#:10091148 TRIzol reagent Ambion Cat#: 15596018 TURBO DNase Ambion Cat#: AM2238 10 3 RNA fragmentation reagent Ambion Cat#: AM8740 40% Sodium bisulfite Sigma-Aldrich Cat#: 243973-500G Hydroquinone Sigma-Aldrich Cat#: 123-31-9 3K Omega 500/pk columns PALL Cat#: OD003C35 RNase-free Tris-HCl buffer (1 M, pH 9.0) TIANDZ Cat#: 80933 Protease inhibitor cocktail Sigma-Aldrich Cat#: P8340 Anti-Flag M2 Affinity Gel Sigma-Aldrich Cat#: A2220 3 3 Flag peptide Sigma-Aldrich Cat#: F4799 PierceTM anti-c-Myc magnetic beads Thermo Fisher Cat#: 88843 5 3 Hi-Density TBE sample buffer Invitrogen Cat#: LC6678 6% TBE gel Invitrogen Cat#: EC6865BOX RNase inhibitor Thermo Fisher Cat#: EO0381 NuPAGETM 4-12% Bis-Tris Gel Invitrogen Cat#: NP0335BOX NuPAGE LDS sample buffer (4 3 ) Invitrogen Cat#: NP0007 Superscrit II Invitrogen Cat#: 18064-014 NuPAGETM MES SDS running buffer (20 3 ) Invitrogen Cat#: NP0002 DIG RNA labeling mix Roche Cat#: 11277073910 (Continued on next page) e1 Molecular Cell 75, 1–15.e1–e11, September 19, 2019

REAGENT or RESOURCE SOURCE IDENTIFIER FLU RNA labeling mix Roche Cat#: 11685619910 TSA Plus Cyanine 3/Fluorescein PerkinElmer Cat#: NEL753001KT BM purple Roche Cat#: 11442074001 mMessage mMachine SP6 kit Thermo Fisher Cat#: AM1340 Moloney Murine Leukemia Virus (M-MLV) Reverse Transcriptase Promega Cat#: M1705 Critical Commercial Assays Dynabeads mRNA purification kit Ambion Cat#: 61006 Dynabeads protein A Thermo Fisher Cat#: 10001D RNA Clean kit Tiangen Biotech Cat#: DP412 RevertAidTM first strand cDNA synthesis kit Thermo Fisher Cat#: K1622 SYBR Permix Ex Taq Takara Cat#: RR420A KAPA Stranded mRNA-Seq Kit KAPA Cat#: K8401 Ribo-Zero rRNA Removal Kit (H/M/R) Illumina Cat#: MRZG12324 NEB Next Ultra RNA library Prep Kit NEB Cat#: E7530 Deposited Data RNA-Seq data for early stage zebrafish embryos This study GEO: GSE120643 GSA: CRA001080 (project PRJCA001013) RNA-BisSeq data for early stage zebrafish embryos This study GEO: GSE120645 GSA: CRA001080 (project PRJCA001013) Ybx1 RIP-Seq data for zebrafish embryos This study GEO: GSE120646 GSA: CRA001080 (project PRJCA001013) Ybx1 iCLIP-Seq data for zebrafish embryos This study GEO: GSE120646 GSA: CRA001080 (project PRJCA001013) RNA-BisSeq data for HeLa cells Yang et al., 2017 GEO: GSE93751 RNA-BisSeq data for mouse brain Yang et al., 2017 GEO: GSE93751 RNA-BisSeq data for HeLa cells Squires et al., 2012 SRA: SRA027832.1 RNA-BisSeq data for embryonic stem cells (ESCs) of mouse Amort et al., 2017 GEO: GSE83432 mRNA half-life data for HeLa cells Wang et al., 2014 GEO: GSE49339 mRNA half-life data for NIH 3T3 cells Schwanh€ausser et al., 2011 SRA: SRA030871 RNA-Seq data for wild-type and miR-430 MO-injected zebrafish embryos Mishima and Tomari, 2016 GEO: GSE71609 Crystal structure data for zebrafish Ybx1 This study PDB: 6A6J Mass spectrometry data of oligo pulldown and Ybx1-IP This study PeptideAtlas: PASS01262 and PASS01263 Original data and source dataset This study https://doi.org/10.17632/27gkytj6kx.1 Experimental Models: Cell Lines HeLa (human) ATCC Cat#:CCL-2; RRID: CVCL_0030 293T (human) ATCC Cat#: CRL-3216; RRID: CVCL_0063 Experimental Models: Organisms/Strains Zebrafish: AB This study N/A Oligonucleotides Oligonucleotides for RNA affinity assay. See Table S2 In vitro synthesized by the Chemical Synthesis Center of the National Institute of Biological Sciences, Beijing N/A Oligonucleotides for Structural analysis, see Table S3 Laboratory of Jinbiao Ma N/A (Continued on next page) Molecular Cell 75, 1–15.e1–e11, September 19, 2019 e2

Further information and requests for resources and reagents may be directed to and will be fulfilled by the Lead Contact, Yun-Gui Yang (ygyang@big.ac.cn).


Cell Lines
293T cells used in in vitro assays were originally purchased from ATCC (RRID: CVCL_0063), authenticated by short tandem repeat (STR) analysis, and routinely confirmed to be free of mycoplasma. All cell lines were maintained in standard DMEM (Gibo) supplemented with 10% fetal bovine serum (Shanghai ExCell Biology Inc.) and 1 3 penicillin/streptomycin (Invitrogen) in standard humidified 5% CO2, 37 C cell culture incubators. Zebrafish ZF4 cells were cultured in DMEM/F12 (1:1) (GIBCO) supplemented

REAGENT or RESOURCE SOURCE IDENTIFIER Recombinant DNA pcDNA3-Flag-ybx1 This study N/A pcDNA3-Flag-ybx1-W45A This study N/A pcDNA3-Flag-ybx1-W45F This study N/A pcDNA3-Flag-pabpc1a This study N/A pCS2+-Myc-ybx1 This study N/A pCS2+-Myc-ybx1-W45A This study N/A pCS2+-Myc-ybx1-W45F This study N/A pCMV-myc-ybx1 This study N/A pCMV-myc-ybx1-W45A This study N/A pCMV-myc-ybx1-W45F This study N/A pGEX-5X-2-ybx1 This study N/A Software and Algorithms Trimmomatic (version 0.33) Bolger et al., 2014 http://www.usadellab.org/cms/index.php?page= trimmomatic; RRID: SCR_011848 Fastx Toolkit (version 0.0.13) http://hannonlab.cshl.edu/ fastx_toolkit http://hannonlab.cshl.edu/fastx_toolkit/download. html; RRID: SCR_005534 MeRanTK (version 1.2.0) Rieder et al., 2016 https://icbi.i-med.ac.at/software/meRanTK/ TopHat (version 2.1.1) Trapnell et al., 2009 http://ccb.jhu.edu/software/tophat/manual.shtml; RRID: SCR_013035 BWA (version 0.7.10) Li and Durbin, 2009 http://bio-bwa.sourceforge.net/bwa.shtml; RRID: SCR_010910 HTSeq (version 0.6.0) Anders et al., 2015 https://www.huber.embl.de/HTSeq; RRID: SCR_005514 MACS2 (version 2.1.1) Feng et al., 2012; Zhang et al., 2008 https://github.com/taoliu/MACS; RRID: SCR_013291 CTK (version 1.0.3) Shah et al., 2017 https://zhanglab.c2b2.columbia.edu/index.php/ CTK_Documentation BEDTools (version 2.26.0) Quinlan and Hall, 2010 http://code.google.com/p/bedtools; RRID: SCR_006646 Cluster (version 3.0) de Hoon et al., 2004 http://bonsai.hgc.jp/ mdehoon/software/cluster/ software.htm WebLogo Crooks et al., 2004 http://weblogo.threeplusone.com/create.cgi; RRID: SCR_010236 DEGseq Wang et al., 2010 http://bioinfo.au.tsinghua.edu.cn/software/degseq; RRID: SCR_008480 Cytoscape (version 3.6.0) Shannon et al., 2003 https://www.cytoscape.org; RRID: SCR_003032 DAVID (version 6.8) Dennis et al., 2003 https://david.ncifcrf.gov; RRID: SCR_001881 Metascape Zhou et al., 2019 http://metascape.org; RRID: SCR_016620 e3 Molecular Cell 75, 1–15.e1–e11, September 19, 2019 with 20% fetal bovine serum (GIBCO) and 13 penicillin/streptomycin (Invitrogen) in standard humidified 5% CO2, 28 C cell culture incubators.

Zebrafish Strains and Embryo Collection
Zebrafish embryos were obtained by the natural mating of AB strain and grown in embryo medium at 28.5 C. All zebrafish work was approved by the Ethical Review Committee of Institute of Zoology, Chinese Academy of Sciences, China. Zebrafish embryos were staged according to standard morphological criteria (Kimmel et al., 1995). Control or morpholino-injected zebrafish embryos at each indicated stage were collected for RNA-Seq, RNA-BisSeq and other experimental analyses.


Plasmid Construction
Total RNA extracted from zebrafish embryo using TrizolTM Reagent (Invitrogen) was used for cDNA synthesis by reverse-transcription PCRwith RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher) and oligo(dT) primer. The obtained cDNA was used as template to amplify genes of interest in the course of subsequent cloning. Zebrafish ybx1 gene (GenBank: NM_001126457.1) encoding full length Ybx1 was amplified by PCR and subcloned into pcDNA3-Flag, pCMV-Myc and pGEX-5X-2 and pCS2+ vectors, respectively. Zebrafish pabpc1a gene (GenBank: NM_001031676.1) encoding full length Pabpc1a was amplified by PCR and subcloned into pcDNA3-Flag and pCS2+ vectors, respectively. All the primers used for cloning are enlisted in the Table S4.

RNA Preparation and mRNA Purification
Total RNA was isolated from zebrafish embryos harvested at different stages using TRIzolTM Reagent (Invitrogen). mRNA was extracted with using Dynabeads mRNA Purification Kit (Ambion) and subjected to TURBO DNase (Invitrogen) treatment at 37 C for 30min and ethanol precipitation. After centrifugation and extensivewashingwith 75%ethanol, the RNAwas dissolved and quantified using Qubit 3.0 (Thermo Fisher).

Construction of RNA-Seq and RNA-BisSeq Libraries
RNA-Seq libraries were directly constructed using the KAPA Stranded mRNA-Seq Kit (KAPA) following the manufacturer’s instructions. The preparation of RNA-BisSeq libraries were performed according to a previous study with some modifications (Yang et al., 2017). Briefly, around 200 ngmRNA premixed with the in vitro transcribedmouseDhfrmRNA at a ratio of 300:1 (DhfrmRNA serves as methylation conversion control) was fragmented to 100 nt fragments for 1min at 90 C in 103RNA fragmentation reagent (Ambion), then stopped by 10 3 RNA stop solution (Ambion), and precipitated with 100% ethanol. The RNA pellet was resuspended in 100 ml bisulfite solution (pH 5.1), which is a 100:1 mixture of 40% sodium bisulfite (Sigma-Aldrich) and 600 mM hydroquinone (SigmaAldrich), and subjected to heat incubation at 75 C for 4.5 h. The reaction mixture was desalted by passing through Nanosep with 3K Omega 500/pk columns (PALL Corporation) with centrifugation. After washed with nuclease-free water and centrifuged for five times, the RNA was finally dissolved in 100 mL nuclease-free water and then desulfonated by incubation with an equal volume of 1 M Tris-HCl (pH 9.0) at 75 C for 1 h. After ethanol precipitation, the RNA was resuspended in 11 ml of RNase-free water and subjected to library construction. Reverse transcription was carried out with superscript II Reverse Transcriptase (Invitrogen) and ACT random hexamers. The following procedures were performed with the KAPA Stranded mRNA-Seq Kit (KAPA) according to the manufacturer’s instructions. Sequencing was performed on Illumina HiSeq X-Ten sequencing system with paired end 150 base pair (bp) read length, or HiSeq 2500 sequencing system with paired end 125 bp read length. m5C RIP-qPCR The procedure of m5C immunoprecipitation was modified from the previously reported m6A RIP (Zhang et al., 2017). Briefly, mRNAs from wild-type zebrafish embryos at 4 hpf were first treated with TURBO DNase and then fragmented into 200–300 nt fragments by incubation at 90 C for 30 s in fragmentation buffer (Ambion). The fragmented RNA was precipitated with ethanol and collected for following reaction. 5 mg of anti-m5C antibody (Abcam; RRID: AB_442823) was incubated with 40 mL Dynabeads Protein A (Thermo Fisher) in IPP buffer (150 mMNaCl, 0.1%NP-40, 10 mM Tris-HCl, pH 7.4, 0.4 U/ml RNase inhibitor) for 1 h at room temperature. Then the mRNAs (2 mg) were incubated with the prepared antibody-beads mixture for 4 h at 4 C. The flow-through was obtained through separating with themagnetic stand and subjected to RNA extraction with phenol-chloroform extraction and ethanol preparation. The beads pellet was then washed three times with IPP buffer and the bound RNAs were recovered by proteinase K digestion, phenolchloroform extraction and ethanol preparation. One-tenth of the fragmented RNA was saved as input control. Equal amount of input RNA, flow-through RNA and IPed RNA was subjected to reverse transcription and the subsequent qPCR. The relative enrichment of m5C in each sample was calculated by normalizing to input. The primers used were listed in Table S4.

UHPLC-MRM-MS/MS analysis was performed following the previous reported method (Yang et al., 2017). Briefly, 200 ng mRNA purified from zebrafish embryos at indicated time points was digested overnight with 0.1 U Nuclease P1 (Sigma-Aldrich) and 1.0 U calf intestinal phosphatase (NEB) in 50 mL reaction volume at 37 C. After filtered by ultra-filtration tubes (MWcutoff: 3 kDa, Molecular Cell 75, 1–15.e1–e11, September 19, 2019 e4 Pall, Port Washington, NewYork), the mixture was subjected to UHPLC-MRM-MS/MS analysis to detect m5C, rC, rU, rG, and rA. An Agilent 1290 UHPLC system coupled with a 6495 triple quadrupole mass spectrometer (Agilent Technologies) was used for the analysis. The UHPLC separation of mononucleosides was performed with a Zorbax Eclipse Plus C18 column (100 mm 3 2.1 mm I.D., 1.8 mm particle size, Agilent Technologies) and the mass spectrometer was operated in the positive ion mode. The adopted MRM mode was: m/z 258/126 for m5C, m/z 244/112 for rC, m/z 245/113 for U, m/z 284/152 for rG and m/z 268/136 for rA. Each sample was analyzed for at least three times with the injection volume of 5 ml. The amounts of m5C and rC was calibrated based on standard curves. Nitrogen was used as a nebulizing and desolvation gas of MS detection. The nebulization gas: 40 psi; the flowrate of desolvation gas: 9 L/min; the source temperature: 300 C; capillary voltage: 3,500 V; collision gas: high purity nitrogen (99.999%). The ribonucleotide standard for m5C was purchased from TCI, China. In Vivo RNA Pulldown Assay and MS Analysis In vivo RNA pull down assay was performed following the previous report (Yang et al., 2017) with some modifications. The same biotin-labeled RNA oligonucleotides with or without m5C analyzed in Yang et al. were used in this study. Zebrafish embryos in early stages were lysed with lysis buffer (50 mM Tris-HCl pH 7.5, 350 mM NaCl, 0.4 mM EDTA, 1% NP-40, 1 mM DTT, 0.4 U/ml RNase inhibitor (Thermo Fisher)). After centrifuging to remove cell debris, the zebrafish embryo lysate was pre-cleared for 1 h at 4 C by incubation with pre-washed streptavidin-conjugated magnetic beads (NEB). The streptavidin-conjugated magnetic beads were also pre-cleared by incubation with 0.2 mg/ml tRNA (Sigma-Aldrich) and 0.2 mg/ml BSA (Amresco) for 1 h at 4 C under gentle rotation. Then the pre-cleared embryo lysate and the streptavidin-conjugated magnetic beads were incubated together with biotinlabeled RNA oligonucleotides at 4 C for 2 h under gentle rotation. Afterward, the beads-oligo-protein mixture was washed with wash buffer for five times and heated in NuPAGETM LDS Sample Buffer (43 ) (Invitrogen) at 95 C for 10min. Samples were separated on NuPAGETM 4%–12% Bis-Tris Gel (Invitrogen) and stained with Coomassie brilliant blue. The protein-containing gel slices were sent to Institute of Biophysics, CAS for MS analysis. The protein levels of Ybx1 pulled down by Oligo-m5C were detected with anti-Ybx1 polyclonal antibody (AbMax) through western blotting.

Protein Purification in Mammalian Cells
293T cells were transiently transfected with pcDNA3-Flag-Drybx1 or pcDNA3-Flag-Drpabpc1a plasmids using PEI transfection reagent. 48 h later, cells were lysed with lysis buffer (50 mM Tris-HCl, pH 7.4, 500 mMNaCl, 1%NP-40, 13 Protease inhibitor cocktail) and then sonicated (10%output, 10 s pulse-on, 20 s pulse-off) for 1min by a Sonic Dismembrator (Thermo Fisher). After removing the cell debris through centrifugation at 13,300 rpm for 20 min, the lysates were incubated with anti-Flag M2 Affinity Gel (Sigma-Aldrich) for 4 h at 4 C. After washing with lysis buffer for five times and TBS buffer (20mMTris-HCl pH 7.4, 150mMNaCl) for twice, the beadsbound proteins were eluted with 1 mg/ml 3 3 Flag peptide (Sigma-Aldrich) for 1 h at 4 C. The elute containing purified protein was condensed using VIVASPIN 500 (Sartorius Stedim Biotech) and quantified by Coomassie brilliant blue staining and western blotting. In Vitro RNA Pulldown Assay In vitroRNA pulldown assaywas performed according to the previously reportedmethod (Yang et al., 2017) with somemodifications. Generally, 10 pmol of purified Flag-Ybx1 protein and 10 pmol of biotin-labeledOligo-m5C or Oligo-Cwere incubatedwith 15 mL streptavidin-conjugated magnetic beads (NEB) in binding buffer (50mMTris-HCl pH 7.5, 250mMNaCl, 0.4mMEDTA, 0.1%NP-40, 1mM DTT, 0.4 U/ml RNase inhibitor) for 1 h at 4 C. After washing with binding buffer for three times, the beads-bound proteins were heated in NuPAGETM LDS Sample Buffer (4 3 ) (Invitrogen) and then separated on the NuPAGETM 4%–12% Bis-Tris Gel (Invitrogen), and subjected to western blotting analysis with anti-Flag antibody (Sigma-Aldrich; RRID: AB_439687).

Western Blotting
Western blotting was performed as previously reported (Zhang et al., 2017). Samples were separated on SDS-PAGE and transferred onto PVDF membrane. After blocking with 5% non-fat dried milk in 1 3 TBST buffer (150 mM NaCl, 20 mM Tris-HCl, pH7.4, 0.1% Tween-20) for 1 h, the membrane was then incubated for 1 h at 4 C with indicated antibodies. Protein levels were visualized using ECL Western Blotting Detection Kit (GE Healthcare). The following antibodies were used: anti-Ybx1 antibody (Abmax), anti-b-Actin antibody (Cell Signaling Technology; RRID: AB_330288), anti-Pabpc1 antibody (Cell Signaling Technology; RRID: AB_10693595), anti-Flag antibody (Sigma-Aldrich; RRID: AB_439687), anti-Myc antibody (Sigma-Aldrich; RRID: AB_439680), and anti-GST antibody (GenScript, RRID: AB_914654). Quantification of each band was carried out by using Quantity One software (Bio-Rad).

Electrophoretic Mobility Shift Assay (EMSA)
Purified wild-type Flag-DrYbx1 was diluted to series of concentrations of 0 mM, 0.4 mM, 0.8 mM, 1.6 mM, 3.2 mM in binding buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 0.4 mM EDTA, 0.1% NP-40, and 40 U/ml RNase inhibitor, 1 mM DTT, 50% glycerol, 5 ng/ml BSA) 1 mL synthesized RNA probe with or without m5C (6.25 nM final concentration) and 1 mL purified protein (0 nM, 4 nM, 80 nM, 160 nM, 320 nM final concentration, respectively) were mixed and incubated at room temperature for 30 min. Then, 1 mL glutaraldehyde (0.2% final concentration) was added into the mixture and incubated at room temperature for 15 min. The total 11 mL RNA-protein mixture was mixed with 5 mL of 5 3 Hi-Density TBE Sample buffer and separated on 6% TBE gel on ice at 80 V for 30 min, followed by membrane transferring. The nucleic acids were detected by using the chemiluminescent nucleic acid detection module (Thermo Fisher) according to the manufacturer’s instructions. Quantification was carried out using Quantity One e5 Molecular Cell 75, 1–15.e1–e11, September 19, 2019 software (Bio-Rad). The RNA binding ratio at each protein concentration was determined by (RNA-protein)/(free RNA) (RNA-protein)). Error bars indicate ± SEM of three replicates. p values were determined using two-tailed Student’s t test. Protein Purification in E. coli Cells DrYbx1 (30-122) construct was PCR amplified from pENTER-Drybx1 plasmid containing full length ybx1. PCR products were double digested with restriction endonucleases BamHI and XhoI, and then cloned into a modified pSUMO vector encoding 63 His tag at N terminus following a SUMO fusion-tag and the Ulp1 cleavage site. Mutations were produced using the overlap PCR (Table S3). The plasmid containing the cDNA of DrYbx1 (30-122 aa) was transformed into E. coli BL21(DE3) to express recombinant proteins. Escherichia coli (E. coli) cells were cultured in LBmedium at 37 Cwith 50 mg/ml kanamycin until the OD600 reached 0.6-0.8, then the cells were cooled at 18 C for around 30 min before 0.2 mM IPTG were added to induce expression overnight (about 14-16 h). Cells were collected by centrifugation at 6,000 rpm for 15min. Cell pellets were re-suspended in lysis buffer containing 20mM Tris-HCl pH 8.0, 500 mM NaCl, 25 mM imidazole pH 8.0 and lysed using a high-pressure cell disrupter. Cell lysates were centrifuged at 18,000 rpm for 1 h at 4 C. Supernatants were purified with nickel-chelating affinity column HisTrap (GE Healthcare), then washed with lysis buffer and eluted with buffer containing 20 mM Tris-HCl pH 8.0, 500 mMNaCl and 500 mM imidazole pH 8.0. Target fusion protein was dialyzed for 3 h with buffer containing 20mM Tris-HCl pH 8.0, 500mMNaCl and Ulp1 protease was added to remove the N-terminal tag. The mixture was applied to another Ni-NTA column to remove the Ulp1 protease and 6 3 His-SUMO tag. The flowthrough was diluted five-fold with 20 mM Tris-HCl pH8.0, to yield a solution with the low-salt buffer (20 mM Tris-HCl pH 8.0, 100 mM NaCl), which was then loaded onto a HiTrap Q-FF column to remove extra nucleic acid. The elution was concentrated by centrifugal ultrafiltration, loaded onto aHiLoad 16/60 Superdex75 column (GEHealthcare) equilibratedwith buffer containing 10mMTris-HCl pH 8.0, 150 mM NaCl. The resulting product was eluted as a monomer with high purity. Purified proteins were concentrated to around 20 mg/ml and stored in a 80 C freezer. For the purification of recombinant GST-Ybx1 protein, the plasmid containing full length ybx1, pGEX-5X-2-Ybx1, was transformed intoE. coli strain C41(DE3). After enlarged culture, cells harboring the plasmidwere harvested and lysed by high-pressure cell cracker in lysis buffer containing 20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 2 mM DTT, 1 mM PMSF. After centrifugation, the supernatant was incubated with Glutathione Sepharose 4 Fast Flow beads (GE Healthcare). The target proteins were eluted with 20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 2 mM DTT, 10 mM glutathione and then loaded on a HiLoad Superdex 200 16/60 column (GE Healthcare) which was developed with gel-filtration buffer (20 mM Tris-HCl, pH 7.4, 200 mM NaCl).

ITC Measurements
Calorimetric experiments were carried out on a MicroCal iTC200 calorimeter (GE Healthcare) at 25 C. The reaction buffer used for testing the binding affinities between wild-type and mutant proteins of DrYbx1 (30-122 aa) and unmodified or m5C-modified 6-mer RNA oligomers was 10 mM Tris-HCl pH 8.0, 150 mM NaCl. The unmodified (6-mer-C: 50-U1C2A3U4C5U6-30) and the m 5C-modified (6-mer-m5C: 50-U1C2A3U4m 5C5U6-3 0) RNA oligonucleotides were ordered from Dharmacon (GE Healthcare Dharmacon Inc.) and were diluted in the reaction buffer. The sample cell was loaded with 250 mL of RNA at 8 mM and the syringe with 80 mL of protein at 100 mM. For mutant protein W45F, the sample cell was loaded with 250 mL of RNA at 20 mM and the syringe with 80 mL of protein at 200 mM. Titration was performed by injecting protein into RNA samples and involved 25 injections while stirring at 750 rpm. Calorimetric titration data were fitted with the Origin 7.0 (MicroCal) under a single binding site model. DG0 of protein-RNA binding was computed as RTln (1/KD), where R, T, and KD are the gas constant, temperature and dissociation constant, respectively.

Crystallization, Data Collection and Structure Determination
Crystallization was carried out using the hanging-drop, vapor-diffusion method at 18 C by mixing equal volumes of protein and well solution. DrYbx1 protein (30-122 aa) in buffer 10 mM Tris-HCl pH 8.0, 150 mM NaCl was concentrated to 10 mg/ml. The complex of DrYbx1 and RNA was prepared by mixing proteins and 6-mer RNA (50-U1C2A3U4m 5C5U6-3 0) at the molar ration of 1:1.2. Crystals of the complex were grown under the condition of 0.05 M Zinc acetate, 22% PEG 3,350. Crystals were cryoprotected by adding the crystallization solution supplemented with 25% glycerol and flash frozen in liquid nitrogen. X-ray diffraction data of DrYbx1 (30-122 aa) in complex with RNA was collected from a single crystal at 100 K at beamline BL17U1 of the Shanghai Synchrotron Radiation Facility (SSRF) (Table S3). Datasets were integrated and scaled using HKL2000. The phase was determined by molecular replacement withHsYBX1 (PDB code: 1H95) as an initial model using Phaser (McCoy et al., 2007), and the RNAmolecules were built manually using Coot (Emsley and Cowtan, 2004). The model was refined using PHENIX (Adams et al., 2010), and the final refinement statistics are summarized in Table S3. Primers used for vector construction and mutation are listed in Table S4. All molecular representations were prepared using PyMOL (The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC).

RIP was performed as described previously (Wang et al., 2014) with some modifications. Briefly, 800 zebrafish embryos at different time points (0, 2, 4 and 6 hpf) were resuspendedwith 2mL lysis buffer (150mMKCl, 10mMHEPES pH 7.6, 2mMEDTA, 0.5%NP-40, 0.5 mM DTT, 1:100 protease inhibitor cocktail, 0.4 U/ml RNase inhibitor) by rotating at 4 C for 30 min and centrifuged at 15,000 g for 15 min. The supernatant was collected and pre-cleared with Dynabeads protein A (Thermo Fisher) by rotation at 4 C for 1 h. 50 mL pre-cleared embryo lysate was saved as Input and the left were incubated with affinity purified rabbit polyclonal anti-Ybx1 antibody Molecular Cell 75, 1–15.e1–e11, September 19, 2019 e6 (AbMax) and 30 mL Dynabeads Protein A at 4 C for 4 h or overnight. After washing with 1 mL ice-cold NT2 buffer (200 mM NaCl, 50 mM HEPES pH 7.6, 2 mM EDTA, 0.05% NP-40, 0.5 mM DTT, 0.4 U/ml RNase inhibitor) for eight times and once with 1 mL icecold 1 3 PK buffer (100 mM Tris-HCl pH 7.4, 50 mM NaCl, 10 mM EDTA, 0.2% SDS), the beads were treated in 200 ml PK buffer containing 20 ml proteinase K (Roche) for 1 h at 55 C. The solution was collected and subjected to RNA extraction with Acid Phenol: ChCl3 (pH4.3–4.7) and ethanol precipitation. The Input RNA was extracted by using TRIzol TM Reagent. Both the Input and IPed RNA were treated by TURBO DNase (Invitrogen) and subjected to RNA-BisSeq and RNA-Seq library construction. For RIP-qPCR analysis of the reporter RNAs, embryos injected with cap1-m5C or cap1-C RNA at the one-cell stage were collected at 4 hpf and subjected to RIP according to the above procedures. For Pabpc1a RIP, in vitro transcribed Flag-pabpc1a mRNA was co-injected with reporter RNAs. Equal amount of Input and IPed RNA were subjected to reverse transcription using cDNA synthesis using RevertAidTM First Strand cDNA Synthesis Kit (Thermo Fisher) and same amount of cDNA was used as template in the downstream PCR analysis with Takara SYBR Premix Ex Taq (Takara) on a CFX96 Real-Time PCR System (Bio-Rad). The relative binding enrichment of Ybx1 or Pabpc1a was calculated by normalizing to the Input. The p values were determined using two-tailed unpaired Student’s t tests. *, p < 0.05; **, p < 0.01; ***, p < 0.001, ****, p <0.0001. The sequences of primers used are listed in Table S4. iCLIP-Seq iCLIP was carried out as previously described (König et al., 2010). 1,000 zebrafish embryos at 4 hpf were irradiated twice with 0.15 J/cm2 (Stratalinker 2400, Stratagene), lysed and subjected to mild fragmentation. Crosslinked RNA-protein complexes were immunoprecipitated using polyclonal Ybx1 antibody (Abmax) and Dynabeads protein A. RNA was extracted and subjected to library construction using Smarter smRNA-Seq kit (Clontech Laboratories Inc).

Morpholinos, mRNA Synthesis and Microinjection
The antisense morpholinos (MOs) including ybx1 UTR MO, ybx1 ATG MO, pabpc1a ATG MO and pabpc1a Splice MO were purchased from GeneTools. The sequences of gene-specific MOs are described as following: ybx1 UTR MO (50- GTGGCTCTCTAGTGTGTTTTCCC 30) (This work), ybx1 ATG MO (50- CTCGCTGCTCATGTTGTTTTCTTG 30) (This work), pabpc1a ATG MO (50-TTCATTTTCACGGCTGGAGGGTTTT - 30) (Mishima et al., 2012), pabpc1a Splice MO (50- TTACAGACGCTCATCATTACCTTGC - 30) (This work). Stock solutions at 1 mM in ddH2O were prepared. The mRNAs were synthesized and the rescue experiments were performed as described in previous study (Wang et al., 2013). 4 ng of MOs and 100 ng of mRNAs were injected into the one-cell stage embryos, respectively. Generation ybx1 Mutant by CRISPR/Cas9 The gRNA target site (GGAGAACTATCAGAGCGACC) for ybx1 was designed using the website (https://zlab.bio/guide-designresources), then synthesized by T7 RNA polymerase and purified using mirVanaTM miRNA Isolation Kit (Ambion) in vitro. ybx1 gRNA (60 pg) and Cas9 mRNA (250 pg) were co-injected into the one-cell stage zebrafish embryos. Genomic DNA was extracted from normal developing embryos after injection at 24 hpf. A 350 bp DNA fragment spanning the target site was amplified from the genomic DNA by PCR (P1: 50- GCATTGTCTTCATTCTGTCA-30, P2: 50- GAGATGACAGAGGTATGCTT 30) and sequenced. Founder (F0) embryos were raised to adulthood and outcrossed with wild-type zebrafish to screen for heritable mutations. Siblings of F1 embryos carrying mutations were raised to adulthood to establish mutant fish lines.

The RNAprobes targeting ybx1 and pabpc1awere first labeled by digoxigenin (DIG) with DIGRNA labelingmix (Roche) or Fluorescein (FLU) with FLU RNA labeling mix (Roche), and then transcribed by T7 polymerase. Embryos were fixed in 4% PFA at different stages and dehydrated with methanol at 20 C. Whole mount in situ hybridization (WISH) was performed as previously described (Wang et al., 2011). After wishing and permeabilizing, the embryos were hybridized with DIG-labeled antisense RNA probe at 65 C for 12 h. Then the embryoswere washed and incubated with anti-DIG-AP antibody (Roche, RRID: AB_514497, 1:5,000) at 4 Covernight. After washing with PBST, the embryos were stained with BM-purple. Double fluorescence in situ hybridization (dFISH) was performed as previously described to detect co-expression of ybx1 and pabpc1a (Jin et al., 2009). The embryos were hybridized with DIG-labeled ybx1 and FLU-labeled pabpc1a antisense RNA probes, incubated with anti-DIG-POD antibody (Roche, RRID: AB_514499, 1:100) or anti-FLU-POD antibody (Roche, RRID: AB_840257, 1:100), and stained with TSA-FITC/Cy3 amplification reagent (PerkinElmer, 1:100). Immunofluorescence Wild-type zebrafish embryos at different stages were fixed with 4% paraformaldehyde in phosphate buffered saline (PBS) overnight at 4 C. 1%BSAwith 0.3% Triton X-100 was used to block the unspecific binding of the antibodies. The embryos were incubated with anti-Ybx1 antibody (Abmax) diluted in 1% BSA overnight at 4 C. Then, the embryos were incubated with anti-Rabbit-Ig-fluorescein e7 Molecular Cell 75, 1–15.e1–e11, September 19, 2019 (Invitrogen; RRID: AB_2576217) for 2 h at R.T. The embryos were counterstained with DAPI and images were acquired by Nikon confocal A1 (Nikon, Japan).

Immunoprecipitation and Mass Spectrometry
Embryos at shield stage (6 hpf) were collected (a total of 1,000 embryos/condition) and deyolking buffer. Cells were pelleted by centrifuging 5 min at 500 rpm and washed once with PBS, then pipetted up and down to disrupt the embryos with 1 mL Ringer buffer (116 mM NaCl, 2.9 mM KCl, 5 mM HCPES). Cells were pelleted by centrifuging 5 min at 500 rpm and then washed with PBS once, followed by flash freezing in liquid nitrogen. The cell pellet was resuspended in 4 mL IP-MS lysis buffer (20 mM Tris-HCl pH 7.3, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.1%SDS, 2 mM PMSF, 1 mM Na3VO4, 1:100 protease inhibitor cocktail) and frozen embryos were thawed on ice in lysis buffer and homogenized by biological sample homogenizer (OMNI, T = 30 s, S = 4). Then it was rotated at 4 C for 30 min and centrifuged at 15,000 g for 15 min to clear the lysate. 50 ml Dynabeads Protein A were incubated with 6.5 mg Ybx1 antibody (Abmax) in 500 ml IP-MS lysis buffer at R.T. for 1 h, and then washed twice with IP-MS lysis buffer and added to the embryo lysates. The beads and embryo lysates were rotated at 4 C for 4 h or overnight. The beads were then collected, washed eight to ten times with 1 mL ice-cold washing buffer (50 mM Tris-HCl pH 7.9, 100 mMKCl, 5 mMMgCl2, 0.2 mM EDTA, 20% NP-40, 10% glycerol), and washed twice with 1ml ice-cold high-salt washing buffer (50 mM Tris-HCl pH 7.9, 1 M NaCl, 5 mMMgCl2, 0.2 mM EDTA, 20% NP-40, 10% glycerol). Samples were subjected to NuPAGETM 4%–12% Bis-Tris Gel (Invitrogen) and visualized by Coomassie blue staining. The protein-containing gel was analyzed by usingmass spectrometry in Institute of Biophysics, Chinese Academy of Science.

Coimmunoprecipitation (Co-IP)
293T cells were co-transfected with indicated plasmids and harvested with lysis buffer (500 mM NaCl, 50 mM Tris-HCl pH 7.4, 1% NP-40, 1 mM Na3VO4, 1 mM glycerophosphate, 1 mM NaF and 13 Protease inhibitor cocktail) followed by sonication on ice (10-15 cycles with 10 s pulse-on and 20 s pulse-off, 10% amplitude). After centrifugation at 14,000 rpm for 15min the supernatant was incubated with RNase A (1 mg/ml) (Sigma-Aldrich) by gently rotating for 30 min at room temperature and supernatant was cleared by centrifuging at 14,000 rpm for 15min. The supernatant was incubatedwith anti-Flag or anti-Myc beads for 4 h or overnight in amoving rotor at 4 C, washed three times with lysis buffer and boiled in 2 3 SDS loading buffer. The immunoprecipitates were resolved on SDS-PAGE gel and transferred onto PVDF membrane (EMD Millipore) followed by western blotting with indicated antibodies. In Vitro GST Pulldown Assay The purified Flag-DrPabpc1a protein was incubated with glutathione Sepharose (GE Healthcare) to be pre-cleared in NETN buffer (100 mM NaCl, 1 mM EDTA, 20 mM Tris-HCl pH 7.4, 0.5% NP-40) for 1 h with gentle rotation at 4 C. Then pre-cleared FlagDrPabpc1a wasmixed with GST or GST-DrYbx1 protein and incubated overnight at 4 Cwith equal amount of glutathione Sepharose beads. After washing the beads five times with NETN buffer, proteins bound to the beads were heated in 43 NuPAGE LDS Sample Buffer (Invitrogen) at 95 C for 10 min, and analyzed on 10% SDS-PAGE followed by western blotting analysis with anti-Flag (SigmaAldrich; RRID: AB_439687) and anti-GST (Genscript, RRID: AB_914654) antibodies. PAR-CLIP and In Vitro RNA End Biotin Labeling The ZF4 cells were first transfected with DrYbx1 siRNAs and pcDNA3- Flag-Drpabpc1a through electroporation and then cultured in medium supplemented with 200 mM 4-thiouridine (4-SU) (Sigma-Aldrich) for 14 h. The cells were irradiated once with 400 mJ/cm2 at 365 nm using the CL-1000 Ultraviolet Crosslinker (UVP) for crosslinking and then harvested in lysis buffer (50 mM TrisHCl pH 7.5, 100 mM NaCl, 2 mM EDTA, 0.5% (v/v) NP-40, 1 mM NaF, 1 3 Protease inhibitor cocktail (Sigma-Aldrich), 0.04 U/ml RNase inhibitor (Thermo Fisher)). The following immunoprecipitation was performed as described previously (Yang et al., 2017) with some modifications. The protein-RNA-beads complex was labeled with biotin using the RNA 30 End Biotinylation kit (Thermo Fisher) according to the manufacturer’s instructions. After extensive washing, the beads were resuspended in 10 mL 23 LDS loading buffer (Invitrogen) and 20 mL 13 LDS loading buffer, boiled at 95 C for 10 min. To detect RNA-protein complexes, samples were separated by SDS-PAGE and visualized by the chemiluminescent nucleic acid detection module (Thermo Fisher) according to the manufacturer’s instructions. mRNA Stability Analyses Synchronously developing embryos were treated with 0.2 ng of Pol II inhibition a-amanitin (Sigma-Aldrich) at one-cell stage andwere collected at indicated time points. 100 embryos were collected for each time point in duplicates. The total RNA was extracted by TrizolTM Reagent (Invitrogen) and used for RNA-Seq and qRT-PCR analysis. For RNA-Seq, an equal amount of external RNA control consortium (ERCC) RNA spike-in control (Thermo Fisher) was added to the total RNA samples as internal controls. The RNA was subjected to ribosomal RNA depletion with Ribo-Zero rRNA Removal Kit (Illumina) followed by library construction using NEB Next Ultra RNA library Prep Kit (NEB). RNA stability profiling was generated from two biological replicates. For qRT-PCR, the RNA expression levels of individual mRNA in different samples of embryos at indicated time points were determined and normalized to the 0 hpf values. b-Actin was used as an internal control. The half-life of mRNAs was calculated through an exponential growth equation provided in GraphPad Prism 7.0 after combing the three replicates according to previously reported methods (Huang et al., 2018; Molecular Cell 75, 1–15.e1–e11, September 19, 2019 e8 Wang et al., 2014). The results are shown as the mean ± Standard Deviation of three independent experiments. p values were determined using two-tailed Student’s t tests.

Quantitative Reverse-Transcription PCR
Quantitative reverse-transcription PCR (qRT-PCR) was performed to assess the relative RNA abundance. All RNA templates used for qRT-PCR were pre-treated with TURBOTM DNase digestion in the purification step. qRT-PCR was performed by using cDNA synthesis using RevertAidTM First Strand cDNA Synthesis Kit (Thermo Fisher). The reactions were performed with Takara SYBR Premix Ex Taq (Takara) according to the manufacturer’s instructions and quantified by a CFX96 Real-Time PCR System (Bio-Rad). b-Actin was used as an internal control. The p values were determined using two-tailed unpaired Student’s t test. *, p < 0.05; **, p < 0.01; ***, p < 0.001, ****, p <0.0001. The sequences of primers used in this study are listed in Table S4.

Reporter mRNA Stability Assay
To generate the minigene constructs, cap1 mRNA (144 nt) containing one m5C modification site (Cytosine 1267 from start codon AUG) and tpp2 mRNA (150 nt) containing one m5C modification site (Cytosine 830 from start codon AUG) were synthesized from GeneScript Inc. and inserted into the pCS2-EGFP vector named pCS2-EGFP-cap1-WT and pCS2-EGFP-tpp2-WT. In the corresponding mutant plasmids, the Cytosine (C, methylated site) was synonymously mutated to non-Cytosine site without altering the amino acid to generate pCS2-EGFP-cap1-MUT and pCS2-EGFP-tpp2-MUT. The plasmids were linearized by NotI and the mRNAs were generated by in vitro transcription using mMessage mMachine SP6 kit (Thermo Fisher) and Poly(A) tailing kit (Thermo Fisher) according to the manufacturer’s instructions. Products were purified with the RNA Clean kit (Tiangen BioTech) and subjected to injections. 100 pg of WT or MUT mRNAs were injected into embryos at the one-cell stage. Control embryos or ybx1 morphants were collected at sphere stage for RNA expression detection and fluorescence analysis. The RNA expression levels of reporter mRNAs in control embryos or ybx1 morphants were detected by qRT-PCR. The fluorescence intensity signals were analyzed by Image-Pro Plus 6.0. To provide further direct evidence to support the role of m5C modification in stabilizing mRNAs, several reporter RNAs were constructed based on m5C-modified cap1mRNA containing (cap1-m5C RNA) or not containing m5C (cap1-C RNA). 59 nt Oligo 1 with or without m5C and 25 nt Oligo 2were synthesized in vitro (Bioneer). The two oligos were jointed together with DNA bridge Oligo through splint ligation and thus yields the 84 nt reporter RNAs: cap1-m5C and cap1-C. To facilitate the subsequent qRT-PCR analysis for distinguishing the reporter RNA from the endogenous one, a 15 nt sequence of GFP was added at the 50 end of Oligo 1. For the generation of poly(A)-tailed reporter RNAs, Oligo 2-poly(A) containing 30 nt poly(A) tail was synthesized. Embryos at the one-cell stage were injectedwith equal amount ofm5C-modified RNA or control one and collected at indicated time points post injection for the RNA stability analysis. The RNA stability was detected throughmeasuring the RNA abundance changes over the time course by qRT-PCR. The sequences of primers used in this study are listed in Table S4.

High-throughput Sequencing Data Processing
RNA-BisSeq analysis: RNA-BisSeq was carried out on Illumina HiSeq 2500 or X-ten platform with paired-end 125 bp or 150 bp read length. Raw reads were stripped of adaptor sequences and removed low quality bases using Trimmomatic (version 0.33, RRID: SCR_011848) (Bolger et al., 2014). The processed reads with lengths greater than 35 nt were defined as clean reads. Zebrafish and human reference genomes (version zv9 and hg19) were downloaded from UCSC database. The alignment procedure was performed by mapping the clean reads against zv9/hg19 genome by meRanT align (meRanTK version 1.2.0) (Rieder et al., 2016) with stringent parameters: -fmo -mmr 0.01. The unmapped reads were discarded (Table S1). Samples with over 99% conversion ratio were used to call m5C sites by meRanCall (meRanTK, version 1.2.0) (Rieder et al., 2016) with parameters: -mBQ 20 -mr 0 -cr 0.99 -fdr 0.05. The methylation level was calculated by the formula: i/(i + j) (i, number of reads showing methylation (C) at each site; j, the number of reads without methylation (T). Representative bisulfite mapping alignments and methylation level of some representativem5C-modifiedmaternal geneswere shown. To ensure the sufficient conversion efficiency, only sites with coverage depthR 30, methylated cytosine depth R 5 and methylation level R 0.1 were used to do further analysis. The m5C sites were annotated by applying BEDTools’ intersectBed (version 2.26.0, RRID: SCR_006646) (Quinlan and Hall, 2010) (Table S1). The annotated m5C sites were used to do repeatability test. The overlap ratios of m5C sites and mRNAs between biological replicates at each stage were over 60% and 80%, respectively, and the Pearson correlation coefficient of methylation level were 0.88. The distribution of m5C sites among whole transcript (including CDSs, 50UTRs, 30 UTRs and intron) and CG, CHG, and CHH (H = A, C, U) were calculated according to our previous method (Yang et al., 2017). Up- and down-stream 10 nt sequences of m5C sites were extracted with BEDTools to detect the sequence preference for m5C, and logo plots were generated with WebLogo (RRID: SCR_010236) (Crooks et al., 2004). RNA-BisSeq in a mixed sample containing equal amount of RNA from human HeLa cells and zebrafish embryos (2 hpf) were used to validate the reliability of our techniques. RNA-Seq analysis: RNA-BisSeq corresponding transcriptome was also sequenced by Illumina HiSeq 2500 or X-ten platform. The raw reads filtered by RNA-BisSeq data analysis method were aligned to the zv9 genome with TopHat (Trapnell et al., 2009) (version 2.1.1, RRID: SCR_013035, default parameters). Only uniquely mapped reads (qR 20) were kept for the subsequent analysis for each sample (Table S1). The number of reads mapped to each Ensemble gene was counted using the HTSeq (Anders et al., 2015) (version 0.6.0, RRID: SCR_005514, parameters: -m union -s no). RPKM was computed as the number of reads which map per kilobase e9 Molecular Cell 75, 1–15.e1–e11, September 19, 2019 of exon model per million mapped reads for each gene (Table S1). RPKM of overlapped genes with over 0.93 Pearson correlation coefficient (R) between two replicates were considered credible. RIP-Seq analysis: Raw reads were stripped of adaptor sequences by fastx_clipper (FASTX-Toolkit, version 0.0.13, RRID: SCR_005534) and paired-end reads were merged into a single fragment. Three random bases sequence and poly(A) or poly (T) were used as index to remove duplicate sequence. The processed reads with smart sequence and at least 5 nt continuous poly(A) or poly(T) sequence have been cut off (in-house script) and only reads with lengths greater than 18 nt and high quality bases (Trimmomatic, version 0.33, RRID: SCR_011848) (Bolger et al., 2014) were defined as clean reads. Reads were aligned to the zv9 genome with TopHat (Trapnell et al., 2009) (version 2.1.1, RRID: SCR_013035, default parameters). Only uniquely mapped reads (qR 20) were kept for the subsequent analysis for each sample. MACS2 (version 2.1.1, RRID: SCR_013291) (Feng et al., 2012; Zhang et al., 2008) were used for the peak calling with the options ‘‘-c -f BAM --nomodel --gsize=1.4e9 --keep-dup all -n -B -p 1e-3.’’ The peaks were annotated by applying BEDTools’ intersectBed (RRID: SCR_006646) (Quinlan and Hall, 2010) (Table S1) and only the overlapped peaks from two replicates were used for downstream analysis. The distribution of Ybx1 enriched peaks among whole mature transcript (including CDSs, 50UTRs, 30 UTRs and intron) were calculated similar to that of m5C sites. iCLIP-Seq analysis: Raw reads were trimmed according to RIP-Seq analysis methods. Low-quality bases were filtered by fastq_filter.pl, a custom perl script from CLIP Tool Kit (CTK, version 1.0.3) (Shah et al., 2017), and reads shorter than 24 nt were discarded. The remaining reads were mapped to the zv9 genome with BWA (version 0.7.10, RRID: SCR_010910) (Li and Durbin, 2009), allowing % 0.06 error rate (substitutions, insertions, or deletions) per read (bwa aln -n 0.06 -q 20) as shown in the online CTK Documentation (https://zhanglab.c2b2.columbia.edu/index.php/CTK_Documentation). To identify the Ybx1 targets’ locus, the mode of truncation calling was performed. For each truncation position, the coverages of unique tag (k) and truncations (m) were determined by CITS.pl program. Reads were converted into bedgraph format normalized by BEDTools (version 2.26.0, RRID: SCR_006646) (Quinlan and Hall, 2010) (Table S1). The truncation position identified in both replicates were considered as credible Ybx1-binding sites. As the crosslink events in a binding site commonly spread acrossmore than one nucleotide, lots of studies further defined binding regions (clusters or peaks) on RNA after identifying the crosslinked sites (Avolio et al., 2018; Haberman et al., 2017; Olgeiser et al., 2019; Rehfeld et al., 2018; Sutandy et al., 2018). According to the recommendation of the usermanual of CTK (https://zhanglab.c2b2. columbia.edu/index.php/ICLIP_data_analysis_using_CTK), we set the overlapping standard of m5C site location in ± 500 nt around truncation sites

Stability Regulation Analysis
To identify maternal genes, Cluster3.0 (version 3.0) (de Hoon et al., 2004) was used to divide all expressed genes into three clusters according to the previous study (Zhao et al., 2017). All expressed genes were divided into two groups (early MZT, 2-4 hpf, and late MZT, 4-6 hpf) based on with (639-1,508) or without (7,797-8,093) m5C modification at both time points and total methylation level. According to the features of total methylation level, mRNAs were divided into three groups:4 hpf < 2 hpf (270) or 6 hpf < 4 hpf (1,110) (mRNAs with lower level at next stage,Omethylation level < 0.2, red line), 2 hpf = 4 hpf (2,398) or 4 hpf = 6 hpf (1,431) (mRNAs with relative stable level between two stages, 0.2 %Omethylation level % 0.2) and 4 hpf > 2 hpf or 6 hpf > 4 hpf (mRNAs with higher level at next stage,Omethylation level > 0.2, yellow line). Gene expression level changes (log RPKMnext stage/ RPKMlast stage) were calculated in each group. The p valueswere calculated using two-sidedWilcoxon andMann-Whitney test. To ensure the total methylation level of eachmRNAwas not affected by coverage change, m5C-containing mRNAs were used to do correlation analysis of total methylation level change and coverage change. R-statistical tools were used to calculate Pearson correlation coefficient. The mRNA half-lives were calculated based on RiboMinus RNA-Seq data of a-amanitin-treated embryos. RPKMwere normalized by linear-fitting of ERCC (spike-in), according to the previously reported method (Wang et al., 2014). RNA degradation rate and half-life were calculated by the following formulas: Log2ðAt=A0Þ= kt t1=2 = 2 klast stage + knext stage t, A and k represent transcription inhibition time (h), mRNA quantity and linear coefficient, respectively. mRNA half-lives were compared between mRNA sets with and without m5C modification. miR-430 maternal targets were identified from RNA-Seq data of wild-type and miR-430 MO-injected embryos at 6 hpf (GEO: GSE71609) according to the method reported by Mishima et al. (Mishima and Tomari, 2016). Potential Ybx1 targets identified byRIP-Seq and iCLIP-Seqwithm5Cmodification and significant downregulated in ybx1morphants versus control samples (Fold change > 1.2) were high-confident Ybx1-binding mRNAs. The dysregulated genes at 4 and 6 hpf control zebrafish embryos and embryos injectedwith ybx1MOswere selected byDEGseqpackage inR language (RRID: SCR_008480) (Wang et al., 2010). Genes with Fold change < 1.5 or > 1.5 with FDR < 0.05 were defined as significantly down- or up- regulated genes. GO analysis was performed using the DAVID bioinformatics database (version 6.8, RRID: SCR_001881) (Dennis et al., 2003) and Metascape (http://metascape.org, RRID: SCR_016620) (Zhou et al., 2019). GO terms with p value of less than 0.05 were considered as statistically significant. Cytoscape (version 3.6.0, RRID: SCR_003032) (Shannon et al., 2003) was used for visualization. Molecular Cell 75, 1–15.e1–e11, September 19, 2019 e10

All statistical analyses of qPCR and imaging were performed at least three independent biological or experimental replicates. Student’s two-tailed unpaired t test was used for statistical comparisons and data were shown as mean ± SD. The p values were used for significance. All statistical analyses were carried out using R or GraphPad Prism (version 7.0).

The RNA-BisSeq, RNA-Seq, RIP-Seq, and iCLIP-Seq data in this study have been deposited in the Gene Expression Omnibus database with accession numbers GEO: GSE120645, GSE120643 and GSE120646. These data were also deposited in the Genome Sequence Archive (Wang et al., 2017) in BIG Data Center (Big Data Center Members, 2017), Beijing Institute of Genomics (BIG), Chinese Academy of Sciences, with accession number GSA: CRA001080 (project PRJCA001013). The atomic coordinates and structure factors for the crystal structures have been deposited in the Protein Data Bank (PDB) with the accession code PDB: 6A6J. The mass spectrometry data of oligo pulldown and Ybx1-IP have been deposited in the PeptideAtlas with accession numbers PASS01262 and PASS01263, respectively. The original imaging data and source dataset have been deposited in Mendeley Data (https://doi.org/10.17632/27gkytj6kx.1). e11 Molecular Cell 75, 1–15.e1–e11, September 19, 2019

Ying Yang, Lu Wang, Xiao Han, Wen-Lan Yang, Mengmeng Zhang, Hai-Li Ma, Bao-Fa Sun, Ang Li, Jun Xia, Jing Chen, Jian Heng, Baixing Wu, Yu-Sheng Chen, Jia-Wei Xu, Xin Yang, Huan Yao, Jiawei Sun, Cong Lyu, Hai-Lin Wang, Ying Huang, Ying-Pu Sun, Yong-Liang Zhao, Anming Meng, Jinbiao Ma, Feng Liu, Yun-Gui Yang
Molecular cell