N6-methyladenosine (m6A) is the most prevalent internal (non-cap) modification present in the messenger RNA of all higher eukaryotes1,2. Although essential to cell viability and development3,4,5, the exact role of m6A modification remains to be determined. The recent discovery of two m6A demethylases in mammalian cells highlighted the importance of m6A in basic biological functions and disease6,7,8. Here we show that m6A is selectively recognized by the human YTH domain family 2 (YTHDF2) ‘reader’ protein to regulate mRNA degradation. We identified over 3,000 cellular RNA targets of YTHDF2, most of which are mRNAs, but which also include non-coding RNAs, with a conserved core motif of G(m6A)C. We further establish the role of YTHDF2 in RNA metabolism, showing that binding of YTHDF2 results in the localization of bound mRNA from the translatable pool to mRNA decay sites, such as processing bodies9. The carboxy-terminal domain of YTHDF2 selectively binds to m6A-containing mRNA, whereas the amino-terminal domain is responsible for the localization of the YTHDF2–mRNA complex to cellular RNA decay sites. Our results indicate that the dynamic m6A modification is recognized by selectively binding proteins to affect the translation status and lifetime of mRNA.
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Gene Expression Omnibus
Gene Expression Omnibus
RNA sequencing data were deposited in the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) under accession number GSE49339 and the processed results were presented as Supplementary Table 1. Processed files were deposited in the Gene Expression Omnibus under accession no. GSE46705.
Tuck, M. T. The formation of internal 6-methyladenine residues in eucaryotic messenger RNA. Int. J. Biochem. 24, 379–386 (1992)
Jia, G., Fu, Y. & He, C. Reversible RNA adenosine methylation in biological regulation. Trends Genet. 29, 108–115 (2013)
Clancy, M. J., Shambaugh, M. E., Timpte, C. S. & Bokar, J. A. Induction of sporulation in Saccharomyces cerevisiae leads to the formation of N6-methyladenosine in mRNA: a potential mechanism for the activity of the IME4 gene. Nucleic Acids Res. 30, 4509–4518 (2002)
Zhong, S. et al. MTA is an Arabidopsis messenger RNA adenosine methylase and interacts with a homolog of a sex-specific splicing factor. Plant Cell 20, 1278–1288 (2008)
Hongay, C. F. & Orr-Weaver, T. L. Drosophila Inducer of MEiosis 4 (IME4) is required for Notch signaling during oogenesis. Proc. Natl Acad. Sci. USA 108, 14855–14860 (2011)
Frayling, T. M. et al. A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science 316, 889–894 (2007)
Jia, G. et al. N6-methyladenosine in nuclear RNA is a major substrate of the obesity-associated FTO. Nature Chem. Biol. 7, 885–887 (2011)
Zheng, G. et al. ALKBH5 is a mammalian RNA demethylase that impacts RNA metabolism and mouse fertility. Mol. Cell 49, 18–29 (2013)
Sheth, U. & Parker, R. Decapping and decay of messenger RNA occur in cytoplasmic processing bodies. Science 300, 805–808 (2003)
Dandekar, T. & Bengert, P. RNA Motifs and Regulatory Elements 2nd edn, 1–11, (Springer, 2002)
He, C. Grand challenge commentary: RNA epigenetics? Nature Chem. Biol. 6, 863–865 (2010)
Wei, C. M. & Moss, B. Nucleotide sequences at the N6-methyladenosine sites of HeLa cell messenger ribonucleic acid. Biochemistry 16, 1672–1676 (1977)
Bokar, J. A., Shambaugh, M. E., Polayes, D., Matera, A. G. & Rottman, F. M. Purification and cDNA cloning of the AdoMet-binding subunit of the human mRNA (N6-adenosine)-methyltransferase. RNA 3, 1233–1247 (1997)
Dominissini, D. et al. Topology of the human and mouse m6A RNA methylomes revealed by m6A-seq. Nature 485, 201–206 (2012)
Meyer, K. D. et al. Comprehensive analysis of mRNA methylation reveals enrichment in 3′ UTRs and near stop codons. Cell 149, 1635–1646 (2012)
Stoilov, P., Rafalska, I. & Stamm, S. YTH: a new domain in nuclear proteins. Trends Biochem. Sci. 27, 495–497 (2002)
Zhang, Z. et al. The YTH domain is a novel RNA binding domain. J. Biol. Chem. 285, 14701–14710 (2010)
Cardelli, M. et al. A polymorphism of the YTHDF2 gene (1p35) located in an Alu-rich genomic domain is associated with human longevity. J. Gerontol. A 61, 547–556 (2006)
Hafner, M. et al. Transcriptome-wide identification of RNA-binding protein and microRNA target sites by PAR-CLIP. Cell 141 129–141 (2010)
Peritz, T. et al. Immunoprecipitation of mRNA-protein complexes. Nature Protocols 1, 577–580 (2006)
Ingolia, N. T., Ghaemmaghami, S., Newman, J. R. & Weissman, J. S. Genome-wide analysis in vivo of translation with nucleotide resolution using ribosome profiling. Science 324, 218–223 (2009)
Bazzini, A. A., Lee, M. T. & Giraldez, A. J. Ribosome profiling shows that miR-430 reduces translation before causing mRNA decay in zebrafish. Science 336, 233–237 (2012)
Mukherjee, N. et al. Integrative regulatory mapping indicates that the RNA-binding protein HuR couples pre-mRNA processing and mRNA stability. Mol. Cell 43, 327–339 (2011)
Huang, R., Brown, C. Y. & Morris, D. R. In mRNA Formation and Function (ed. Richter, J. D. ) Ch. 16 (Academic Press, 1997)
Shenton, D. et al. Global translational responses to oxidative stress impact upon multiple levels of protein synthesis. J. Biol. Chem. 281, 29011–29021 (2006)
Kedersha, N. & Anderson, P. Mammalian stress granules and processing bodies. Methods Enzymol. 431, 61–81 (2007)
Reijns, M. A., Alexander, R. D., Spiller, M. P. & Beggs, J. D. A role for Q/N-rich aggregation-prone regions in P-body localization. J. Cell Sci. 121, 2463–2472 (2008)
Kato, M. et al. Cell-free formation of RNA granules: low complexity sequence domains form dynamic fibers within hydrogels. Cell 149, 753–767 (2012)
Gehring, N. H., Neu-Yilik, G., Schell, T., Hentze, M. W. & Kulozik, A. E. Y14 and hUpf3b form an NMD-activating complex. Mol. Cell 11, 939–949 (2003)
Behm-Ansmant, I. et al. mRNA degradation by miRNAs and GW182 requires both CCR4:NOT deadenylase and DCP1:DCP2 decapping complexes. Genes Dev. 20, 1885–1898 (2006)
Pillai, R. S., Artus, C. G. & Filipowicz, W. Tethering of human Ago proteins to mRNA mimics the miRNA-mediated repression of protein synthesis. RNA 10, 1518–1525 (2004)
Hafner, M. et al. PAR-CliP - a method to identify transcriptome-wide the binding sites of RNA binding proteins. J. Vis. Exp. 41, e2034 (2010)
Kishore, S. et al. A quantitative analysis of CLIP methods for identifying binding sites of RNA-binding proteins. Nature Methods 8, 559–564 (2011)
Pearson, W. R., Wood, T., Zhang, Z. & Miller, W. Comparison of DNA sequences with protein sequences. Genomics 46, 24–36 (1997)
Trapnell, C., Pachter, L. & Salzberg, S. L. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25, 1105–1111 (2009)
Anders, S. & Huber, W. Differential expression analysis for sequence count data. Genome Biol. 11, R106 (2010)
Trapnell, C. et al. Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nature Biotechnol. 28, 511–515 (2010)
Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008)
Corcoran, D. L. et al. PARalyzer: definition of RNA binding sites from PAR-CLIP short-read sequence data. Genome Biol. 12, R79 (2011)
Bolte, S. & Cordelieres, F. P. A guided tour into subcellular colocalization analysis in light microscopy. J. Microsc. 224, 213–232 (2006)
Clement, S. L. & Lykke-Andersen, J. A tethering approach to study proteins that activate mRNA turnover in human cells. Methods Mol. Biol. 419, 121–133 (2008)
This work is supported by National Institutes of Health GM071440 (C.H.) and EUREKA GM088599 (T.P. and C.H.). The Mass Spectrometry Facility of the University of Chicago is funded by National Science Foundation (CHE-1048528). We thank A. E. Kulozik, W. Filipowicz and J. A. Steitz for providing the sequence and plasmids of the tether reporter. We thank Dr. G. Zheng and W. Clark for help in polysome profiling. We also thank S. F. Reichard for editing the manuscript.
The authors declare no competing financial interests.
Extended data figures and tables
a, Western blot showing YTHDF1 and YTHDF3 pulled down with an m6A-containing RNA probe. *Thiol-substituted phosphodiester bonds were used to prevent enzymatic cleavage. b, LC-MS/MS showing that m6A was enriched in GST–YTHDF1- or GST–YTHDF3-bound mRNA while depleted in the flow-through portion. c, d, Gel-shift assay measuring the dissociation constant (Kd, nM, indicated at the upper left corner of the gel) of GST-tagged YTH domain family proteins (c, YTHDF2; d, YTHDF1 and YTHDF3) with methylated and unmethylated RNA probes. 4 nmol RNA probe was labelled with 32P and the protein concentration ranged from 20 nM to 5 μM.
a, Left, PAR_CLIP gel image showing 32P-labelled RNA–YTHDF2 complex; right, western blotting of HeLa cell lysate with overexpression of Flag-tagged YTHDF2 (10 μg per lane). Upper band was detected by anti-Flag antibody; lower band was detected by anti-GAPDH antibody. b, Overlap of transcripts identified by PAR-CLIP and RIP-seq of YTHDF2. c, d, YTHDF2 binding motif identified by MEME with top 1,000 scored PAR-CLIP peaks under different motif searching parameters. c, With motif length restricted to 5–10 bp, P = 1.1 × 10−43, 183 sites were found under this motif. d, The motif length was restricted to 5–12 bp. The motif with lowest P value was shown in main text as Fig. 1c, this motif showed the second lowest P value, P = 5.1 × 10−14, 104 sites were found. e, With 7–12 bp, P = 7.5 × 10−42, 231 sites were found under this motif. f, Distribution of PAR-CLIP peaks across the length of mRNA. Each region of 5′ UTR, CDS, and 3′ UTR were binned into 50 segments, and the percentage of PAR-CLIP peaks that fall within each bin was determined. g, Overlap of YTHDF2 PAR-CLIP peaks with m6A peaks in different sub-transcript regions. Over 70% PAR-CLIP peaks in 5′ UTR, CDS, stop codon, and 3′ UTR regions overlap with m6A peaks (at least 1-bp overlap). In contrast, only 20%∼30% of PAR-CLIP peaks in transcription starting sites (TSS) and intergenic regions coincide with m6A peaks. h, Enrichment of YTHDF2 PAR-CLIP peaks in long exons. The length distribution of exons that contain YTHDF2 PAR-CLIP peaks (red) shifts to larger size compared with the length distribution of all exons in the human genome (black).
a, The YTHDF2 knockdown efficiency is about 80% as detected by RT–PCR (error bars, mean ± s.d., n = 3, biological replicates) and RNA-seq. Although at current stage we could not identify a reliable antibody for YTHDF2, ribosome-profiling of YTHDF2 did indicate that the translation level of YTHDF2 decreased by 80% after siRNA knockdown. RT–PCR results were normalized to that of GAPDH as an internal control. RNA-seq and ribosome profiling results were calculated by actual RPKM. b, YTHDF2 knockdown led to decreased translation efficiency of its targets due to the accumulation of non-translating mRNA. Translation efficiency is calculated as the ratio of ribosome-protected fragments and mRNA input. P value was calculated by using Mann–Whitney U-test (two-tailed, significance level = 0.05). c, Multiple pairwise comparisons (Kruskal–Wallis test) by using the Steel–Dwass–Critchlow–Fligner procedure (two-tailed, significance level = 0.05). d, The regional effect of the YTHDF2-binding site is not significant. Cumulative distribution showing mRNA lifetime log2-fold changes (Δ) between si-YTHDF2 and si-control for non-targets and CLIP-IP common targets with major CLIP peak at 5′ UTR, CDS, 3′ UTR, intron, and non-coding RNA. Except for intron, other regions show similar fold changes (also see Extended Data Fig. 3c). e, The m6A methyltransferase (MT-A70) and demethylase (FTO) remain unchanged with YTHDF2 knockdown.
a–d, Examples of transcripts harbouring m6A peaks and YTHDF2 PAR-CLIP peaks: SON (CDS, a), CREBBP (3′ UTR, b), LDLR (3′ UTR, c), PLAC2 (non-coding RNA, d). Coverage of m6A immunoprecipitation and input fragments are indicated in red and blue, respectively. YTHDF2 PAR-CLIP peaks are highlighted in green. Black lines signify CDS borders. e–n, relative RNA level quantified by gene-specific RT–PCR, and error bars shown in these figure panels are mean ± s.d., n = 6 (two biological replicates × three technical replicates). e, Enrichment fold of SON, CREBBP mRNA, and PLAC2 RNA in YTHDF2-RNA coimmunoprecipitation versus RNA–protein input control, and in m6A in vitro immunoprecipitation versus mRNA input control. f, Relative changes of SON, CREBBP mRNA, and PLAC2 RNA in siYTHDF2 sample versus siControl, and overexpression of YTHDF2 versus overexpression of C-YTHDF2. g–k, Lifetimes of SON, CREBBP mRNA and PLAC2 RNA under siYTHDF2 versus siControl. l–n, YTHDF2 knockdown altered the cytoplasmic distribution of its mRNA targets. The SON (l) and CREBBP (m) mRNA levels decreased in the non-ribosome mRNP portion but increased in the 40S–80S portion under siYTHDF2 compared to siControl. However, they showed different changes in the polysome portion. RPL30 (n) is not a target of YTHDF2 and did not show an increase in the 40S–80S portion.
Extended Data Figure 5 Knockdown of METTL3 (MT-A70) led to decreased binding of YTHDF2 to its targets and increased stability of its target RNAs similar to that of YTHDF2 knockdown.
a, Western blotting showing that the knockdown efficiency of siMETTL3 at 48 h was ∼80%. b–g, Relative RNA level quantified by gene-specific RT–PCR, and error bars shown in these figure panels are mean ± s.d., n = 6 (two biological replicates × three technical replicates). b, Percentages of YTHDF2 targets (SON, CREBBP, LDLR) in YTHDF2-bound portion versus unbound portion decreased significantly after METTL3 knockdown for 48 h. After 24 h transfection of METTL3 siRNA, HeLa cells were transfected with Flag-tagged YTHDF2, and cells were collected after another 24 h. Anti-Flag beads were used to separate YTHDF2-bound portion (IP) from unbound portion (flow-through). Each transcript was quantified by RT–PCR. c, Relative changes of SON, CREBBP and LDLR mRNA in siMETTL3 sample versus siControl. d–g, Lifetimes of SON, CREBBP, and LDLR mRNA under siMETTL3 versus siControl.
Extended Data Figure 6 Co-localization of YTHDF2 with protein markers of P bodies, stress granules, and deadenylation complexes.
a–h, Fluorescence immunostaining of Flag-tagged YTHDF2 (green, anti-Flag, Alexa 488) and other protein markers (DCP1a and GW182 for P bodies and eIF3 for stress granule, DDX6 (also known as RCK/p54) and HuR for both, CNOT7, PAN2, and PARN for deadenylation complex; magenta of Alexa 647 is the colour for the marker, green + magenta = white for the co-localization spot). The scale of the magnified region (while frame) is 1.8 μm × 1.8 μm. i, Co-localization between YTHDF2 and different protein markers were characterized by Pearson’s coefficient, for each pair, n = 5∼7. YTHDF2 seems to have better co-localization with P bodies than stress granules. It also seems to co-localize best with CNOT7 (also known as CAF1 or POP2) which is a subunit of the CCR4-NOT deadenylation complex. j, Western blotting results showing that immunoprecipitation (IP) of Flag-tagged full length YTHDF2 and N-YTHDF2 (N-terminal domain) also pulled down the P-body marker DCP2, but not with mock control or C-YTHDF2 (the C-terminal domain). For IP samples, each lane was loaded with 2 μg IP portion; and the input lane was loaded with 10 μg input portion which corresponded to ∼1% of overall input). k, Comparison of P/Q/N (highlighted) rich regions of YTHDF1-3 with other aggregation-prone proteins. l, C-YTHDF2 is capable of selective binding of m6A-containing RNA. LC-MS/MS showing that m6A-containing RNA was enriched in the His6-tagged C-YTHDF2-bound mRNA while reduced in the flow-through portion. Error bars shown in the figure are mean ± s.d., n = 4 (two biological replicates × two technical replicates).
a, Structural presentation of the two domains of YTHDF2. b, Scheme of the reporter assay: the RNA reporter vector encodes firefly luciferase (F-luc) as the primary reporter and Renilla luciferase (R-luc) on the same plasmids acting as transfection control for normalization. Five Box B RNA elements were inserted at the 3′ UTR of F-luc as positive tether reporter (noted as F-luc-5BoxB); the effecter was a fusion of N-YTHDF2 and λ peptide which recognizes Box B with high affinity. c, The F-luc luciferase activity (protein translation) for N-YTHDF2–λ was reduced by ∼20% compared to that of N-YTHDF2 and λ controls. Error bars shown in the panel are mean values ± s.d. from n = 8 (biological replicates). d, e, The reporter mRNA lifetime was significantly reduced (∼40%) when bound by N-YTHDF2–λ as compared to the controls of N-YTHDF2 and λ. Doxycycline (Dox, 400 ng μl−1) was used to inhibit transcription of the reporter. 18 h post transfection of reporter and effecters, Dox was removed to allow a pulse transcription of F-luc-5BoxB for 4 h. Then Dox was added back and the samples were collected at indicated time point. The amounts of F-luc-5BoxB were determined by RT–PCR, normalized to R-luc, then for each time series, samples at t = 0 h were set as 100%. Error bars shown in the panel are mean ± s.d., n = 6 (two biological replicates × three technical replicates). f, Scheme of poly(A) tail length assay. g, h, Tethering N-YTHDF2 to the reporter mRNA does not significant trigger deadenylation of the reporter. The PCR products of reporter poly(A) tail were visualized in 10% TBE gel stain (g) and no significant difference of the deadenylation rate was observed (h). i–l, Shorter poly(A) tail lengths were observed in the YTHDF2-bound fraction for the N-YHTDF2-tethered reporter RNA (i and j) as well as the native target RNA CREBBP (k and l). Tether reporter F-luc-5BoxB and Flag-tagged YTHDF2-N-λ (i) or full length Flag-tagged YTHDF2 (k) were expressed in HeLa cells, and subjected to immunoprecipitation with anti-Flag beads. RNA recovered from input, IP and flow-through were further processed and the final PCR products for F-luc-5BoxB (i) or CREBBP (k) were visualized in 10% TBE gel. j and l, each lane were re-plotted against base pair, after log fitting of relative gel mobility with base pairs.
a, b, The top molecular function of YTHDF2 targets is “Gene Expression and RNA Transcription”, and the top cellular function is “Cell Death and Survival”. Ingenuity Pathway Analysis of function category of YTHDF2 targets and non-targets revealed that the two gene groups are heterogeneous in their functional composition. (*top two functions for YTHDF2 targets and **top two functions for YTHDF2 non-targets.). c, d, Pie charts of molecular types of differentially expressed YTHDF2 targets (c) versus non-targets (d) upon YTHDF2 knockdown. Differentially expressed genes (P value <0.05) caused by YTHDF2 knockdown were grouped to YTHDF2 targets (796 gene) and non-targets (1554) based on their presence or absence in YTHDF2 PAR-CLIP binding sites, and subject to Ingenuity Pathway Analysis (the category “other” was not shown). The results show that the group of YTHDF2 targets is transcription regulators whereas that of non-targets is enzyme, indicating that m6A may significantly affect gene expression via tuning mRNA stabilities of transcription factors through YTHDF2. e, f, YTHDF2 knockdown led to reduced cell viability. The IPA analysis of ribosome profiling data of YTHDF2 knockdown (48 h) versus control predicts decreased cell viability (e). Ribosome profiling data was chosen since it may better reflect the translation status. MTT assay provided experimental evidence of reduced cell viability upon YTHDF2 knockdown. P values that were calculated from Student’s t-test were 0.036, 4.7 × 10−4, and 9.4 × 10−4, at 48 h, 72 h and 96 h respectively (f). Error bars shown in the figure are mean ± s.d., n = 10 (biological replicates).
a, Overlap of three biological replicates (rep1–rep3) for PAR-CLIP. Numbers showing the sum of genes identified in each sample. b, Correlation of enrichment fold as log2(IP/input) between two technical RIP replicates. In rep1 the input mRNA was purified by poly(dT) beads, whereas in rep2 the input RNA was processed by rRNA removal. c–e, Box plot showing consistent results from two biological replicates that were conducted for ribosome profiling and mRNA lifetime profiling, respectively. For mRNA lifetime profiling, rep1 was normalized by spike-in control that was proportional to cell numbers, whereas rep2 was normalized by spike-in that was proportional to total RNA concentrations. Despite the technical variations, YTHDF2 knockdown resulted in significant lifetime increase of its targets. (T, 1,277 CLIP+RIP targets; NT, 3,905 non-targets; box, the first and third quartiles; notch, the median; dot in the box: the data average; whisker, 1.5 × standard deviation; cross, the 1 and 99 percentiles; short line, the maximum and minimum; P values were calculated by Mann–Whitney U-test, two-tailed, significant level = 0.05). f–h, Correlation of RPKM between technical mRNA input samples prepared by poly(A) selection (x axis) and by rRNA removal (y axis), which are comparable to the variations between biological replicates that prepared by the same mRNA selection method.
About this article
Cite this article
Wang, X., Lu, Z., Gomez, A. et al. N6-methyladenosine-dependent regulation of messenger RNA stability. Nature 505, 117–120 (2014). https://doi.org/10.1038/nature12730
Journal of Luminescence (2021)
Cancer Cell International (2020)
Overview of distinct 5-methylcytosine profiles of messenger RNA in human hepatocellular carcinoma and paired adjacent non-tumor tissues
Journal of Translational Medicine (2020)
N6-Methyladenosine Level in Silkworm Midgut/Ovary Cell Line Is Associated With Bombyx mori Nucleopolyhedrovirus Infection
Frontiers in Microbiology (2020)
Cell Reports (2020)