Cell Res. 2017 N6-甲基腺苷驱动circRNA广泛翻译.pdf

Cell Research(2017)27:626- ARTICLEExtensive translation of circular RNAs driven by N6-methyladenosineYun Yang1,2,3,4,*,Xiaojuan Fan2,*,Miaowei Mao4,5,*,Xiaowei Song2,4,Ping Wu6,7,Yang Zhang8,Yongfeng Jin1,Yi Yang5,Ling-Ling Chen8,Yang Wang9,Catherine CL Wong6,7,Xinshu Xiao3,Zefeng Wang2,41Institute of Biochemistry,College of Life Sciences,Zhejiang University at Zijingang,Zhejiang,Hangzhou,Zhejiang 310058,Chi-na;2CAS Key Lab for Computational Biology,CAS Center for Excellence in Molecular Cell Science,CAS-MPG Partner Institute for Computational Biology,Shanghai Institute for Biological Sciences,Chinese Academy of Sciences,Shanghai 200031,China;3Department of Integrative Biology and Physiology and the Molecular Biology Institute,UCLA,Los Angeles,CA 90095,USA;4Department of Pharmacology,University of North Carolina at Chapel Hill,Chapel Hill,NC 27599,USA;5Synthetic Biology and Biotechnology Laboratory,State Key Laboratory of Bioreactor Engineering,School of Pharmacy,East China University of Sci-ence and Technology,Shanghai,China;6National Center for Protein Science,Institute of Biochemistry and Cell Biology,Shanghai Institutes for Biological Sciences,Chinese Academy of Sciences,Shanghai 200031,China;7Shanghai Science Research Center,Chinese Academy of Sciences,Shanghai 201204,China;8Institute of Biochemistry and Cell Biology,Shanghai Institute for Biolog-ical Sciences,Chinese Academy of Sciences,Shanghai 200031,China;9Institute of Cancer Stem Cell,Dalian Medical University,Dalian,Liaoning 116044,ChinaExtensive pre-mRNA back-splicing generates numerous circular RNAs(circRNAs)in human transcriptome.However,the biological functions of these circRNAs remain largely unclear.Here we report that N6-methyladenosine(m6A),the most abundant base modification of RNA,promotes efficient initiation of protein translation from cir-cRNAs in human cells.We discover that consensus m6A motifs are enriched in circRNAs and a single m6A site is suf-ficient to drive translation initiation.This m6A-driven translation requires initiation factor eIF4G2 and m6A reader YTHDF3,and is enhanced by methyltransferase METTL3/14,inhibited by demethylase FTO,and upregulated upon heat shock.Further analyses through polysome profiling,computational prediction and mass spectrometry reveal that m6A-driven translation of circRNAs is widespread,with hundreds of endogenous circRNAs having translation potential.Our study expands the coding landscape of human transcriptome,and suggests a role of circRNA-derived proteins in cellular responses to environmental stress.Keywords:N6-methyladenosine;circular RNA;cap-independent translation;eIF4G2Cell Research(2017)27:626-641.doi:10.1038/cr.2017.31;published online 10 March 2017*These three authors contributed equally to this work.Correspondence:Zefeng WangE-mail:Received 13 January 2017;revised 3 February 2017;accepted 19 February 2017;published online 10 March 2017IntroductionAlthough circular RNAs(circRNAs)in higher eu-karyotes were first discovered more than 20 years ago 1,2,they attracted little attention until recently when a large number of circRNAs were identified by parallel sequencing 3-7.The majority of circRNAs are gener-ated through back-splicing of internal exons,a non-ca-nonical splicing process promoted by dsRNA structures across circularizing exons 3,8-11.Although there are reports that some circRNAs can function as“decoys”to neutralize miRNA(i.e.,as miRNA sponges)4,7 or to bind and sequester other RNA binding proteins 12,the biological function of most circRNAs is still un-determined.An intriguing possibility is that circRNAs could be translated to produce proteins,because most of circRNAs originate from exons and are localized in the cytoplasm.Indeed,artificial circRNAs with an in-ternal ribosomal entry site(IRES)can be translated in vitro 13 or in vivo 11.However the coding potential of circRNAs remains an open question because of ear-ly reports that most circRNAs are not associated with Yun Yang et al.627www.cell-|Cell Research|SPRINGER NATUREpolysomes 3,14,15.N6-methyladenosine(m6A)is the most abundant in-ternal modification of RNAs in eukaryotes 16,17.The modification preferably occurs in the consensus motif“RRm6ACH”(R=G or A;H=A,C or U)18,19,and is found in over 7 000 mRNAs and 300 non-coding RNAs in human and mouse using m6A-specific immu-noprecipitation(MeRIP-Seq)20,21.The methylation of adenosine is catalyzed by a methyltransferase com-plex consisting of methyltransferase-like 3(METTL3),METTL14 and Wilms tumor 1-associating protein 22-25,and the m6A is demethylated by fat mass and obe-sity-associated protein(FTO)and alkylated DNA repair protein alkB homolog 5(AKLBH5)26,27,which serve as the writers and erasers of m6A.Inside cells,m6A is specially recognized by the YTH domain family protein YTHDF1,2 and 3 that bind m6A and function as m6A readers.The m6A modification can affect multiple stages of RNA metabolism,including mRNA localiza-tion,splicing,translation and degradation,which in turn regulates important biological processes such as stem cell differentiation 28,29.In particular,m6A is re-ported to have multifaceted affects on translation:m6A in 3 UTRs was found to increase translation efficiency through binding of YTHDF1 30,whereas m6A in 5 UTRs was reported to promote cap-independent trans-lation upon heat shock stress through a YTHDF2-pro-tection mechanism 31,32.In addition,a recent study has reported that m6A reader YTHDF3 promotes protein synthesis in synergy with YTHDF1 33,which is fur-ther supported by the finding that cytoplasmic YTHDF3 interacts with ribosomal proteins to promote mRNA translation 34.However the general effect of m6A on protein translation is still an incomplete story with the detailed mechanisms being elusive.Here we report that circRNAs in human cells can be efficiently translated using short sequences containing m6A site as IRESs.Consistently,the translation from cir-cRNA is reduced by m6A demethylase FTO,promoted by adenosine methyltransferase METTL3/14,and requires eukaryotic translation initiation factor eIF4G2 and m6A reader YTHDF3.We found that a large number of cir-cRNAs are methylated,suggesting that such translation could be common to many circRNAs.By sequencing RNase R-resistant RNAs associated with polysomes,we identified hundreds of endogenous translatable cir-cRNAs,many of which contain m6A sites.Our finding suggests a general function of circRNA and has import-ant implications in the translation landscape of human genome.ResultscircRNAs containing m6A motifs are translated inside cellsPreviously we developed a minigene reporter contain-ing split GFP to demonstrate that a circRNA can be trans-lated using a viral IRES 11(Figure 1A).The translation from this circRNA system was rigorously validated by multiple controls,including introducing mutations that disrupt intron pairing,treating the samples with RNase R and cleaving the expression plasmids into linear DNA with restriction enzymes to eliminate potential artifact of exon concatemers 11.To further explore the molecular mechanisms governing this phenomenon,we analyzed whether the circRNAs containing several other endoge-nous IRES or control sequences in human genome can also be translated(Supplementary information,Table S1).Surprisingly,all the inserted sequences,including the three“negative controls”ranging from 38 to 253 nt,efficiently initiated GFP translation as judged by western blots(Figure 1B)and fluorescence microscopy(Supple-mentary information,Figure S1A).Translation from the circRNA was eliminated only when the sequence of re-striction sites was left between the stop and start codons of GFP in the circRNA reporter(Supplementary infor-mation,Figure S1B).This unexpected result suggested that IRESs serving to facilitate translation initiation in circRNAs are more prevalent than previously expected.To understand how the“negative control”sequences induce circRNA translation,we examined their sequenc-es near the translation start site.Interestingly,all“nega-tive control”sequences contain a RRACH fragment(R=G or A;H=A,C or U)near the start codon(Figure 1C,top),resembling the consensus motif of m6A modi-fication(i.e.,RRACH motif),the most abundant internal modification of RNAs 16,17.Moreover,by clustering the randomly sampled 10-nt sequence windows in these three sequences,we also found a motif that resembles the consensus site for m6A modification(Figure 1C,bottom,see Materials and Methods section).Compared to all coding mRNAs,the putative m6A motifs are significantly enriched in known circRNAs(Figure 1D).This observed enrichment is reliable,because in a control analysis,m6A motifs are more enriched in snoRNAs yet depleted in snRNAs 35,and are more enriched in exons compared to introns as expected(Supplementary information,Fig-ure S1C).Furthermore,compared to randomly selected 1 000 sets of control hexamers,the consensus m6A hexam-ers(i.e.,HRRACH)are significantly enriched in known circRNAs(Supplementary information,Figure S1D).Consistently,we found that previously identified m6A peaks from transcriptome-wide mapping of m6A sites us-ing MeRIP-Seq 20,21 have a higher average density in circRNA regions compared to all mRNAs regardless of 628Translation of circular mRNASPRINGER NATURE|Cell Research|Vol 27 No 5|May 2017Figure 1 N6-methyladenosine promotes circRNA translation.(A)Schematic diagram of a circRNA translation reporter con-sisting of a single exon and two introns with complement sequences(marked by heart and crown).The exon can be back-spliced to generate circRNAs that drive GFP translation from the IRES.Green arrows indicate PCR primers used in detecting circRNA.(B)Translation of circRNA can be driven by different endogenous human IRESs(from IGF2,Hsp70 and XIAP)or three control sequences(short fragments of intron,coding region and 5 UTR from beta-Actin gene,see Supplementary in-formation,Table S1).Each reporter was transfected into 293 cells,and protein production was detected by western blot 48 h after transfection.CircRNA was detected by semi-quantitative PCR using circRNA specific primers.(C)Consensus m6A mo-tifs are enriched in the“negative control”sequences.Top,RRACH motif near the start codon,with the putative m6A-modified adenosine highlighted in red and the start codon labeled in green.Bottom,enriched motifs discovered by k-mer sampling/clustering.(D)Accumulative distribution of m6A motif in circRNA and mRNA.(E)Density of m6A peaks(from MeRIP-seq)that are mapped to all mRNAs and known circRNAs regions.P-value=3.2e7 by students t-test.(F)m6A motifs directly promote circRNA translation.0-2 copies of m6A motifs(GGACU)and an adenosine-free control sequence(CGGTGCCGGTGC)were inserted into the upstream of the start codon in circRNA reporters,and circRNA and GFP translation were detected similarly as in panel B.(G)Known m6A sites(RSV and RSVns)and their mutations were tested for the activity of driving translation.Experimental procedures are the same as in panel B.In last panel,the total RNA was also treated with RNase R before RT-PCR.Yun Yang et al.629www.cell-|Cell Research|SPRINGER NATUREthe relative positions(Figure 1E).Because m6A was recently found to increase mRNA translation efficiency 30,32,our observations strongly suggested that the m6A containing RRACH sequences may be involved in the translation initiation of circRNAs.To test this hypothesis,we inserted a short fragment(19 nt)containing different copies of consensus m6A motifs before the start codon of circRNA reporter and measured GFP protein production in transfected 293 cells(Fig-ure 1F).As expected,circRNAs containing one or two m6A motifs were efficiently translated into GFP protein,whereas the mutation of both motifs greatly reduced(but did not completely eliminate)the GFP level(Figure 1F).In addition,the circRNA with single m6A site has similar translation efficiency compared to circRNA with two m6A sites,indicating that a single m6A site is sufficient to initiate translation(i.e.,excessive m6A modification may not increase circRNA translation efficiency).The transla-tion of GFP was eliminated when we inserted a sequence without any adenosine residue(Figure 1F).In addition,we tested two other sequences(RSV and RSVns)that were reported to undergo m6A modification 23,and found that both sequences strongly induced protein trans-lation.Importantly,mutants that decrease m6A methyla-tion in these sequences 23 also reduced the translation efficiency(Figure 1G),further supporting the notion that m6A drives protein translation from circRNAs.The pro-duction of circRNAs from all reporters was further val-idated with northern blot using a probe that specifically recognizes full-length GFP(Supplementary information,Figure S2A).As previously reported,the circRNA and its translation products were detected even after the linearization of re-porter plasmids with MluI digestion(Supplementary infor-mation,Figure S2B),strongly supporting translation from circRNA because the pre-mRNA with multiple copies of concatenated GFP fragments cannot be produced in this scenario.The same set of sequences was also able to drive protein translation in HeLa cells(Supplementary informa-tion,Figure S3A and S3B),indicating that m6A-initiated translation is cell type independent.However in HeLa cells,there is still some expression of GFP following mu-tation of both m6A motifs(Supplementary information,Figure S3A).This might be because either m6A can occur at a non-canonical site or some sequences can initiate translation in m6A-independent fashion in HeLa cells.Modulation of m6A level in circRNA affects translation efficiencyTo further evaluate the importance of m6A in trans-lation of circRNAs,we examined whether circRNAs with m6A motifs are indeed methylated using RNA-IP.We found that antibody against m6A specifically pulled down circRNA containing the RSV m6A site and a known m6A-containing mRNA(SON mRNA)20,but not a control mRNA without m6A(GAPDH)(Figure 2A).Cir-cRNA containing mutated m6A site(RSV-mut)was also pulled down by m6A antibody,but with dramatically re-duced efficiency(Figure 2A).This observation suggests that the mutation can reduce but not completely eliminate the m6A modification at RSV site,which is consistent with the small amount of GFP production in RSV-mut reporter(Figure 1G).Alternatively,there may be other minor sites for m6A modification in the circRNA.In addition,co-ex-pression of m6A demethylase FTO 26 significantly re-duced the abundance of immunoprecipitated SON mRNA or RSV-containing circRNA(Figure 2A)and decreased the translation of GFP from the circRNA(Figure 2B),further confirming that the circRNA/mRNA with m6A sites are methylated and circRNA translation is indeed driven by m6A.In agreement with these observations,co-expression of m6A methyltransferase METTL3/14 sig-nificantly increased the RNA-IP signal from the circRNA or mRNA containing m6A but not from the control RNA(Figure 2C),and greatly increased protein transla