Global Analysis of Truncated RNA Ends Reveals New … · LARGE-SCALE BIOLOGY ARTICLE Global...

LARGE-SCALE BIOLOGY ARTICLE

Global Analysis of Truncated RNA Ends Reveals NewInsights into Ribosome Stalling in PlantsOPEN

Cheng-Yu Hou,a Wen-Chi Lee,a Hsiao-Chun Chou,a,b Ai-Ping Chen,c Shu-Jen Chou,c and Ho-Ming Chena,1

a Agricultural Biotechnology Research Center, Academia Sinica, Taipei 11529, Taiwanb Institute of Plant Biology, National Taiwan University, Taipei 10617, Taiwanc Institute of Plant and Microbial Biology, Academia Sinica, Taipei 11529, Taiwan

ORCID IDs: 0000-0003-4195-9165 (C.-Y.H.); 0000-0001-5979-0121 (H.-M.C.)

High-throughput approaches for profiling the 59 ends of RNA degradation intermediates on a genome-wide scale arefrequently applied to analyze and validate cleavage sites guided by microRNAs (miRNAs). However, the complexity of the RNAdegradome other than miRNA targets is currently largely uncharacterized, and this limits the application of RNA degradomestudies. We conducted a global analysis of 59-truncated mRNA ends that mapped to coding sequences (CDSs) of Arabidopsisthaliana, rice (Oryza sativa), and soybean (Glycine max). Based on this analysis, we provide multiple lines of evidence to showthat the plant RNA degradome contains in vivo ribosome-protected mRNA fragments. We observed a 3-nucleotide periodicityin the position of free 59 RNA ends and a bias toward the translational frame. By examining conserved peptide upstream openreading frames (uORFs) of Arabidopsis and rice, we found a predominance of 59 termini of RNA degradation intermediatesthat were separated by a length equal to a ribosome-protected mRNA fragment. Through the analysis of RNA degradomedata, we discovered uORFs and CDS regions potentially associated with stacked ribosomes in Arabidopsis. Furthermore, ouranalysis of RNA degradome data suggested that the binding of Arabidopsis ARGONAUTE7 to a noncleavable target site ofmiR390 might directly hinder ribosome movement. This work demonstrates an alternative use of RNA degradome data in thestudy of ribosome stalling.

INTRODUCTION

Steady state levels of RNA are controlled by relative rates oftranscription and RNA degradation. Most mRNAs in eukaryotespossess a 7-methylguanosine cap at the 59 terminus and a poly(A)tail at the 39 terminus, which are crucial for translation and RNAstability. The lossof the59capor the39poly(A) tail abolishesmRNAtranslation and promotes mRNA degradation (Gallie, 1991). Un-capped 59 ends of mRNAs are degraded by 59-39 exoribonu-cleases (XRNs), whereas deadenylated mRNAs are degraded bythe exosome from the 39 end (Lebreton and Séraphin, 2008;Houseley and Tollervey, 2009). Alternatively, deadenylation canalso triggerdecappingofmRNA, followedbydegradation fromthe59 end (Muhlrad et al., 1994).

Translation plays a crucial role in controllingmRNA stability andis required to eliminate aberrant mRNAs through several distinctmechanisms (Shoemaker and Green, 2012). Nonsense-mediatedmRNAdecay (NMD) is often initiatedwhena ribosomeencountersa premature termination codon (PTC) upstream of an exonjunction complex. The interaction between NMD factors on ter-mination factors and exon junction complexes promotes thedegradation of PTC-containing transcripts. The movement of

ribosomes on mRNA molecules can be stopped externally bya stable RNAstructure or internally by the particular peptide that isencoded by the mRNA. The transcripts associated with stalledribosomes are degraded by a specialized RNA surveillancepathway called no-go decay, whichmay result in endonucleolyticcleavage upstream of the stalled ribosomes (Doma and Parker,2006). Nonstop decay also targets transcripts with stalled ribo-somes, but the stalling is due to the lack of an in-frame stop codon(Frischmeyer et al., 2002; van Hoof et al., 2002). Many eukaryoticmRNAs possess short open reading frames (ORFs) in the 59untranslated region (UTR),which canbe translated andpotentiallyregulate mRNA stability and translation. The stop codon ofa translated upstream open reading frame (uORF) might be rec-ognized as a PTC and thus initiate NMD. Some uORFs encodingpeptides conserved across species have been shown to stall ri-bosomes at uORF stop codons, resulting in the repression ofdownstream main ORF translation and acceleration of RNA deg-radation ina fewcases (Gabaetal., 2005;Uchiyama-Kadokuraetal.,2014).In eukaryotes, small RNAsof 20 to30nucleotidesplay akey role

in regulating gene expression through the RNA interferencepathway. Most animal microRNAs (miRNAs) have a seed region,spanning the second to the seventh or the eighth nucleotide,which can base pair perfectly with the 39 UTR of a target mRNA(Bartel, 2009). The targeting of animal miRNAs is often associatedwith translation repression, deadenylation, and mRNA decayusing exoribonucleases. By contrast, plant miRNAs are highlycomplementary to their targets, and cleavage in the middle oftarget sites is the major mode of plant miRNA action (Rhoades

1Address correspondence to homing@gate.sinica.edu.tw.The author responsible for distribution of materials integral to the findingspresented in this article in accordance with the policy described in theInstructions for Authors (www.plantcell.org) is: Ho-Ming Chen (homing@gate.sinica.edu.tw).OPENArticles can be viewed without a subscription.www.plantcell.org/cgi/doi/10.1105/tpc.16.00295

et al., 2002). However, growing evidence indicates that plantmiRNAs can also inhibit the translation of their targets (Brodersenet al., 2008; IwakawaandTomari, 2013; Li et al., 2013b, 2013c; Liuet al., 2013). In vitro assays of RNA-induced silencing complexescontaining Arabidopsis thaliana ARGONAUTE1 (AGO1) demon-strated that plant miRNAs can repress translation initiation un-coupled with deadenylation or mRNA decay (Iwakawa andTomari, 2013). Furthermore, miRNA binding sites in ORFs canhinder themovementof ribosomes (IwakawaandTomari, 2013). InArabidopsis, miRNA-mediated translation repression occurs inthe endoplasmic reticulum and requires the endoplasmic re-ticulum protein ALTERED MERISTEM PROGRAM1 (Li et al.,2013c).

High-throughput approaches for genome-wide profiling of RNAdegradation intermediates that possess a free monophosphate atthe 59 terminus have been developed by several groups and arevariously named parallel analysis of RNA ends (PARE) (Germanet al., 2008), degradome sequencing (Addo-Quaye et al., 2008),genome-wide mapping of uncapped transcripts (Gregory et al.,2008), and 59 P sequencing (5Pseq) (Pelechano et al., 2015). Be-cause intact mRNAs generally possess a 59 cap that blocks theirligation to RNA adaptors, truncated 59 RNA ends with a freemonophosphate can be selectively sequenced by directly ligatingpoly(A)RNAwithRNAadaptors.PAREanddegradomesequencinghave been widely applied in the identification of small RNA-guidedcleavage sites in various plant species (Addo-Quaye et al., 2008;German et al., 2008; Zhou et al., 2010; Shamimuzzaman andVodkin, 2012; Zhao et al., 2012; Li et al., 2013a). Specific PAREhas been developed for the study of plant miRNA processing byspecifically amplifying miRNA processing intermediates (Bolognaet al., 2013). Although these approaches havebeenused to profilemRNA degradation intermediates in mutants impaired in XRNs orproteins with endonucleolytic activity (German et al., 2008;Harigaya and Parker, 2012; Schmidt et al., 2015), the inter-pretations of some results remain challenging because thecomplexity of the RNA degradome is currently still largely un-characterized.

Previously, we showed that plant RNA degradome data po-tentially contain footprints of RNA binding proteins in the 39 UTR(Hou et al., 2014). Pelechano et al. (2015) also demonstrated thatyeast5Pseqdatacontain invivo ribosome footprints, theproductsof cotranslation mRNA decay. The change in these ribosome-protected termini captured by 5Pseq could reflect ribosome dy-namics without the problems caused by translational inhibitorssometimes used in the generation of in vitro ribosome footprints.Codons associated with paused ribosomes in response to oxi-dative stresswere identified from theanalysis of yeast 5Pseqdata.Cotranslation mRNA decay was also demonstrated to be medi-ated by XRN4 and involved in the reprogramming of gene ex-pression under heat stress in Arabidopsis (Merret et al., 2013,2015). Here, we further demonstrate that footprints of ribosomesare widespread in the RNA degradome of various plant species.Our genome-wide analysis of RNA degradation fragments re-vealeduORFs, codingsequence (CDS) regions, andnoncleavablemiRNA target sites that are potentially associated with stackedribosomes. Our findings thus expand the application of plant RNAdegradome data for the elucidation of posttranscriptional generegulation beyond small RNA-guided cleavage.

RESULTS

Signatures of Ribosome Footprints Were Observed in theRNA Degradome

Our previous analysis of plant RNA degradome data revealedpositional enrichmentof 59-truncatedRNAends in theproximityofmotifs recognizedbyRNAbindingproteins.Thissuggests thatRNAbindingproteins attached toRNAmayhinderRNAdegradation andresult in protected RNA fragments (Hou et al., 2014). Therefore, wesuspected that thebindingof ribosomes tomRNAmayalsoprotectmRNA frominvivodegradationand leave ribosome footprints in theRNA degradome. To explore this possibility, we first used PAREdata toplot thepositionaldistributionof59-truncatedmRNAends inthe junctionsof theCDSsandUTRs.Wepredicted that, ifPAREcancapture ribosome footprints in the same way as ribosome profiling(Ribo-Seq), which delineates ribosome positions by generatingribosome-protected mRNA fragments through in vitro nucleasedigestion (Ingolia, 2010), we would observe a 3-nucleotide peri-odicity in theCDSs.A 3-nucleotide periodicity reflects the stepwisemovement of ribosomes during active translation and has beenreported in the analyses of Ribo-Seq data derived from multiplespecies (Ingolia et al., 2009, 2011; Guo et al., 2010; Liu et al., 2013;Bazzini et al., 2014; Juntawong et al., 2014; Vasquez et al., 2014).Consistent with our prediction, 59 ends of truncated mRNA (PAREdata) generated from Arabidopsis seedlings and inflorescencesshowstrong3-nucleotidephasinginthe39 terminusoftheCDSsbutnot in the proximal region of the 39 UTR (Figure 1A). A 3-nucleotideperiodicity in the CDS region was also observed in the CDSs forPARE reads generated from rice (Oryza sativa) inflorescences andsoybean (Glycine max) seeds (Figure 1A). Besides this phasingpattern, the three species all show preferential accumulation ofPARE reads in the translational frame (frame 1) of annotated CDSs(Figure 1B). Although the enrichment in the translational frame isrelatively small in PARE data comparedwith that in yeast Ribo-Seqdata reported previously (Ingolia et al., 2009), the proportion ofPARE reads falling in the translational frame is significantly higherthan that in the other two frames in the three replicates of Arabi-dopsis inflorescencePAREdata (Supplemental Figure 1). Similar tothepreviousfinding in theanalysisofRibo-Seqdata (Liuet al., 2013;Bazzini et al., 2014; Juntawong et al., 2014), PARE data of thesethree plant species also showed an evident increase in the numberof reads at positions 16 and 17 nucleotides upstream of stop co-dons, a pattern consistent with the deceleration of ribosomemovement during translational termination (Figure 1A). Thesecommon features shared between PARE data and Ribo-Seq datastrongly suggest both the presence of in vivo ribosome-protectedmRNA fragments in theplantRNAdegradomeand theoccurrenceof cotranslational RNA decay in plants.

Regular and Conserved RNA Degradation Patterns WereFound in Conserved Peptide uORFs

Several uORFs in fungi, plants, and animals that encode con-served peptides are able to block ribosomes at stop codons(Wang and Sachs, 1997; Raney et al., 2000; Gaba et al.,2001; Hayden and Jorgensen, 2007; Hood et al., 2007; Uchiyama-Kadokura et al., 2014). Among them, the conserved peptideuORFs (CPuORFs) of yeast CPA1 and an Arabidopsis gene

Ribosome-Protected Fragments in the RNA Degradome 2399

producingS-ADENOSYLMETHIONINEDECARBOXYLASE(SAMDC/AdoMetDC1) have been demonstrated to induce mRNA decaythrough the NMD pathway (Gaba et al., 2005; Uchiyama-Kadokuraetal.,2014).Toprovideadditionalevidencethat theRNAdegradomecontains ribosome footprints, we examined the positional distri-bution of 59-truncated mRNA ends (PARE reads) derived fromseveral CPuORFs of Arabidopsis. In SAMDC, the position 16 nu-cleotides upstream of the uORF stop codon shows a predomi-nant accumulation of PARE reads derived from seedlings but onlya weak enrichment of PARE reads derived from inflorescences(Supplemental Figure 2). BesidesSAMDC, we also examinedPAREreads that mapped to the CPuORFs in a small group of bZIP genesthat regulate the translation of downstream ORFs in response tosucrose concentration (Wiese et al., 2004). Interestingly, these bZIPCPuORFs possess a ladder of PARE peaks at intervals of ;30nucleotides, which is the size of a ribosome-protected fragment inArabidopsis (Liu et al., 2013; Juntawong et al., 2014) (Figure 2A).Counting from the 39 end of bZIP CPuORFs, the first and secondPARE peaks are positioned ;16 and 46 nucleotides upstream of

uORF stop codons, with a few reads present in the 30-nucleotidewindow between these two peaks. A third PARE peak at position276thatwasanadditional30nucleotidesupstreamwasobservedinbZIP2andbZIP11CPuORFs (Figure2A). The30-nucleotidephasingof PARE peaks in the 39 end of CPuORFs provides strong evidencetosupport thenotion thatPAREcaptures thedegradation fragmentsprotected by an array of stacked ribosomes. These three PAREpeaks likely delineate the 59 ends of degradation fragments pro-tectedbyone, two,or threeadjacent ribosomesstalledat aCPuORFstop codon, respectively. Notably, the 30-nucleotide phasing wasnotevident in thesebZIPgeneswhenweanalyzed invitro ribosome-protected mRNA fragments of two Ribo-Seq data sets that weregenerated by independent groups (Liu et al., 2013; Juntawong et al.,2014) (Figure 2A; Supplemental Figure 3).Because bZIP CPuORFs were reported to repress the trans-

lation of the downstream ORF under high sucrose concentration(Wiese et al., 2004), we usedmodified RNA ligase-mediated rapidamplification of cDNA ends (RLM 59 RACE) to test whethertreatment with 6% sucrose would affect the accumulation of

Figure 1. 59-Truncated RNA Ends Show a 3-Nucleotide Periodicity and Frame Bias in the CDS.

(A) The positional distribution of 59 ends of truncated RNAmapped to the regions surrounding the start codon and the stop codon of Arabidopsis, rice andsoybean CDS. PARE data of Arabidopsis seedlings and inflorescences were generated by this study and PARE data of rice inflorescences and soybeanseedswere published by Zhou et al. (2010) and Song et al. (2011). Blue bars indicate positions falling in the translational frame (frame 1) of annotated CDS if59-truncated ends represent the 59 edge of a ribosome and the distance from the 59 edge of a ribosome to the first base of the A site is 17 nucleotides. Redarrowheads beneath thegraphs represent the first nucleotides in the start codon (left side) or the stopcodon (right side). The illustration at thebottomshowsthesizeofanmRNAfragmentprotectedbyaplant ribosomeand thepositionof ribosomesdecodingstart andstopcodon.CDS,darkblue;UTR, lightblue;E,the exit site; P, the peptidyl site; A, the aminoacyl site.(B) The proportion of 59-truncated RNA endsmapped to complete CDS in all three frames. frame 1, the translational frame of TAIR annotated CDS; frames2 and 3, the frames offset +1 and +2 from frame 1.

2400 The Plant Cell

Figure 2. bZIP uORFs Accumulate a Ladder of 59-Truncated RNA Ends at 30-Nucleotide Intervals.

(A) The positional distribution of 59 ends of truncated RNA generated by PARE and ribosome-protected mRNA fragments generated by Ribo-Seq inArabidopsis bZIP2, bZIP11, and bZIP53 59 UTRs. The PARE data of Arabidopsis inflorescences and the Ribo-Seq data of light-treated seedlings wereretrieved from the data sets published byGerman et al. (2008) and Liu et al. (2013) respectively. The PARE data of Arabidopsis seedlingswere generated bythis study. Regions of CPuORFs are shown as dark blue lines under the graphs. The PARE peaks at positions 16, 46, and 76 nucleotides upstream of theuORF stop codon are highlighted in red. TP10M, tags per 10 million; TP50M, tags per 50 million.

degradation fragments truncated at these bZIP CPuORFs. In-deed, a larger number of 59 ends of degradation intermediatescorresponding to the first or second peak in bZIP2 and bZIP11CPuORFs were detected in Arabidopsis seedlings treated with6% sucrose compared with the untreated control plants (Figure2B). This implies either an increase in ribosomes arrested at thesetwo uORFs or the enhancement of RNA degradation. Taken to-gether, these results indicate that degradation fragments appearto reflect the dynamics of ribosomes on uORFs.

Although 64 CPuORFs (Hayden and Jorgensen, 2007) havebeen identified in Arabidopsis by sequence comparison andannotated in The Arabidopsis Information Resource database(TAIR10annotation),most of themhavenotbeendemonstrated toarrest ribosomes at specific positions. To further explore whether

ribosome stalling at stop codons is a common mechanism un-derlying the regulation mediated by Arabidopsis CPuORFs, weglobally analyzed the distribution of PARE readsmapped to the 39end of the 64 Arabidopsis CPuORFs. Some ArabidopsisCPuORFs predominantly accumulate PARE reads at position16 nucleotides upstream of stop codons and some show anadditional peak at position245,246, or247 (Figure 3A). BesidesbZIP genes, CPuORFs in genes encoding several basic helix-loop-helix-type transcription factors, a trehalose-6-phosphatephosphatase, two methyltransferases, and an unknown proteinwere found to possess PARE peaks at these two specific sites(Supplemental Figure 4), implying the stalling of ribosomes atthese uORF stop codons. The analysis of 35 rice CPuORFsshowed the enrichment of PARE reads at the same sites (positions

Figure 2. (continued).

(B)ModifiedRLM59RACEassaysofbZIP2andbZIP11 transcripts inArabidopsis seedlings treatedwith orwithout 6%sucrose.MYB65, a target ofmiR159,wasused asapositive control for themodifiedRLM59RACEassay. Thebrackets indicate thePCRproducts excised andcloned forSanger sequencing (leftpanels). In total, 36and16cloneswere sequenced forbZIP2andbZIP11, respectively. Thepositional distributionof 59-truncatedRNAends increasedunder6%sucrose treatments is shown for thebZIP2 andbZIP11 59UTRs (right panels). RegionsofCPuORFs are shownasdark blue lines under the graphs. TSS,transcriptional start site.

Figure 3. Site-Specific Enrichment of 59-Truncated RNA Ends Is Evident in CPuORFs.

(A)Clusteredheatmapsof 59-truncatedRNAendsmapped in a55-nucleotide region upstreamof the stopcodonofCPuORFsandpredicted uORFs inwild-type Arabidopsis inflorescences and rice seedlings. Predicted uORFs overlapping CPuORFs are not included in the heat maps of predicted uORFs.Arabidopsis and rice PARE data used in this analysis were published by German et al. (2008) and Li et al. (2010), respectively.(B)Clusteredheatmapsof59RNAendsprotectedby ribosomes ina55-nucleotides regionupstreamof thestopcodonofCPuORFsandpredicteduORFs inwild-type Arabidopsis. The Ribo-Seq data sets of light-treated and normoxic (in air) seedlings used in this analysis were published by Liu et al. (2013) andJuntawong et al. (2014), respectively.In (A) and (B), the first nucleotidesof the stop codon is assignedposition 0, and the color of data points represents thePeak_Index value,which is calculatedby dividing the number of PARE or Ribo-Seq reads starting at the position indicated by the number of total reads in a 31-nucleotide flanking region. Thenumbers of annotated CPuORFs and predicted uORFs (indicated in parentheses) and uORFs included in heat maps are shown above the heat maps.

2402 The Plant Cell

216 and246) (Figure 3A), suggesting that ribosome stalling at thestop codon of some CPuORFs is a conserved mechanism acrossspecies. To know whether ribosome profiling could capture thesignature of ribosome stacking in CPuORFs as PARE, we alsoanalyzed in vitro ribosome-protected mRNA fragments of twoRibo-Seq data sets with the number of total reads greatly ex-ceeding that of PARE data sets we used (Liu et al., 2013;Juntawong et al., 2014). Intriguingly, although position 16 or17 nucleotides upstream of the uORF stop codon shows an en-richment of in vitro ribosome-protected mRNA fragments, theenrichment at these positions is less prominent and does not ac-company the enrichment at the positions near 30 nucleotidesupstream (Figure 3B). The same analysis on genome-wide pre-dicted uORFs having lengths greater than 60 nucleotides showedno preferential accumulation of PARE reads for Arabidopsis or riceand Arabidopsis Ribo-Seq reads if predicted uORFs overlappingCPuORFs were excluded (Figures 3A and 3B). Based on theanalysis of PARE data that harbor ribosome footprints, most Ara-bidopsis and rice uORFs may not cause ribosome stalling at stopcodons in the same way as many CPuORFs, at least under theconditions and in the tissues that the PARE data were generated.

Regulatory uORFs Were Identified Using the Patterns ofRNA Degradation Fragments

We suspected that a few regulatory uORFsmay not be conservedbetweenArabidopsis and riceor that theymight havebeenmissedin the previous search because of low sequence homology(Hayden and Jorgensen, 2007). Therefore, analysis of the RNAdegradome might provide an alternative approach for the iden-tification of regulatory uORFs that have the potential to stall ri-bosomes.We thus reverse searched for uORFs usingPARE readspeaking at the regions 16 to 17 and 45 to 47 nucleotides upstreamof the stop codon of predicted uORFs. In addition to fourCPuORFs reported previously (Hayden and Jorgensen, 2007), weidentified four Arabidopsis uORFs with two predominant PAREpeaks representing two tandem ribosomes stacking at stop co-dons (Table 1; Supplemental Figure 5). Further analyseswere thenperformed on these three novel uORFs to investigate sequenceconservation and their regulatory functions.

A 99-nucleotide uORF in the 59 UTR of CBL-INTERACTINGPROTEIN KINASE6 (CIPK6) shows two PARE peaks at positions

16 and 46 nucleotides upstream of the stop codon like CPuORFsin bZIP genes (Figures 2 and 4A). We then compared the distri-bution of reads obtained through Ribo-Seq and PARE on CIPK6.The Ribo-Seq data of light-treated seedlings show a corre-sponding peak at position 16 nucleotides but not 46 nucleotidesupstream of the stop codon (Figure 4A). Moreover, position 217accumulates more reads than position 216 in Ribo-Seq data,whereas position 216 has the highest accumulation of PAREreads. Predominant accumulation of ribosome-protected 59 endsat positions 16 and 46 upstream of CIPK6 stop codon was notobserved in the Ribo-Seq data of normoxia (in air) seedlings(Supplemental Figure 6). Although the CIPK6 uORF was not an-notated in the TAIRdatabase, it encodes a conserved peptide andhas been identified previously based on sequence homology bytwo groups (Takahashi et al., 2012; Vaughn et al., 2012). Similarly,the conserved uORF in soybean CIPK6 also possesses PAREpeaks in 30-nucleotide increments at the same positions relativeto the uORF stop codon (Figure 4B), implying that ribosomestalling might be a conserved mechanism for CIPK6 uORF reg-ulation.The function of Arabidopsis CIPK6 uORF in repressing down-

stream ORF expression was validated via transient expressionassaysbyEbinaetal. (2015).Weconfirmed the regulatory functionof ArabidopsisCIPK6 uORF by generating stable transgenic linesthat harbored a reporter gene encoding GUS driven by the Ara-bidopsis CIPK6 promoter containing a wild-type 59 UTR ora uORF-deleted (DuORF) 59 UTR in which the start codon wasconverted into a stop codon (Figure 4C). Overall, the DuORFtransgenic lines showed a higher level of expression of thereporter gene in most tissues (Figure 4C; Supplemental Figure7), indicating that the regulation mediated by CIPK6 uORF iswidespread even under normal growth conditions. Because anapproximate 6-fold difference (P = 0.02, two-tailed Student’s ttest) was detected in the comparison of GUS activity, but thechange of GUS mRNA level was less than 2-fold and not sta-tistically significant (Figures 4D and 4E), CIPK6 uORF likelycontrols downstream ORF expression mainly at the translationallevel.The other two candidates of ribosome stalling uORFs we

identified are located inMYB34 andMYB51, which belong to thesame clade ofMYB transcription factors involved in the regulationof glucosinolate biosynthesis (Celenza et al., 2005; Gigolashvili

Table 1. Arabidopsis uORFs with Two Predominant PARE Peaks Representing Two tandem Ribosomes Stacking at Stop Codons

Gene Model Name Description Coordinates of uORF Coordinates of PARE Peak PARE Library

CPuORFs in TAIR10AT1G75390.1 bZIP44 Basic leucine-zipper 44 258–380 336, 365 INFAT2G18160.1 bZIP2 Basic leucine-zipper 2 201–323 279, 308 SDAT5G01710.1 Methyltransferase 142–252 207, 237 INFAT5G49450.1 bZIP1 Basic leucine-zipper 1 229–303 258, 288 INFNovel uORFsAT1G18570.1 MYB51 MYB domain protein 51 156–221 176, 206 SD, INFAT2G01930.2 BPC1 Basic pentacysteine1 5–310 265, 295 SDAT4G30960.1 CIPK6 CBL-interacting protein kinase 6 132–227 182, 212 SD, INFAT5G60890.1 MYB34 MYB domain protein 34 52–150 106, 135 SD

SD, seedlings; INF, inflorescences.

Figure 4. CIPK6 CPuORF Possesses Footprints of Stacked Ribosomes and Represses Downstream ORF Expression in Various Tissues.

(A) The positional distribution of 59 ends of truncated RNA generated by PARE and ribosome-protected mRNA fragments generated by Ribo-Seq inArabidopsisCIPK6 59UTR. The PARE data of Arabidopsis inflorescences and the Ribo-Seq data of light-treated seedlings plotted were retrieved from thedata sets published by German et al. (2008) and Liu et al. (2013), respectively. TP10M, tags per 10 million.(B) Thepositional distribution of 59-truncatedRNAends generated byPARE in soybeanCIPK6 59UTR. The soybeanPAREdata plottedwere retrieved fromthe data set published by Song et al. (2011). In (A) and (B), regions of CPuORFs are shown as dark-blue lines under the graphs and the first nucleotide of thestop codon is assignedposition 0. ThePAREorRibo-Seqpeaks at positions 16, 46, and76nucleotides upstreamof the uORFstop codon are highlighted inred.(C)Histochemical staining of randomly selectedArabidopsis T1 transgenic plants carrying aGUS reporter genedriven byArabidopsisCIPK6promoterwithwild-type [CIPK6pro:UTR(WT):GUS] and deleted uORF [CIPK6pro:UTR(DuORF):GUS]. The red arrowhead indicates the site mutated. Bar = 1 cm.

2404 The Plant Cell

et al., 2007; Frerigmann and Gigolashvili, 2014). AlthoughMYB34uORF is longer than MYB51 uORF, they both possess two pre-dominant PARE peaks at positions 16 and 46 nucleotides up-stream of uORF stop codons (Figure 5A). The translation of thesetwo MYB uORFs is supported by the higher density of Ribo-Seqreads in the predicted ORFs compared with that in the flankingregions (Liu et al., 2013) (Figure 5A). However, the analysis ofin vitro ribosome-protected mRNA fragments on these two MYBuORFs with two Ribo-Seq data sets (Liu et al., 2013; Juntawonget al., 2014) showed no preferential accumulation at these twosites (Figure 5A; Supplemental Figure 6). The peptide sequencesencoded by these two uORFs are highly conserved at the Cterminus within the mustard (Brassicaceae) family (Figure 5B).Glucosinolates are sulfur- and nitrogen-containing secondarymetabolites that are found mainly in plant species in the order ofBrassicales (Grubb and Abel, 2006), explaining the absence ofconserved uORFs in many other plant species. The negativeimpact of MYB34 CPuORF on the mRNA level of MYB34 wasdemonstratedpreviously inastudyofamutantcontainingaPTC inthis uORF (Bender and Fink, 1998). We confirmed the negativeregulation ofMYB34 uORF on the expression of the downstreamreporter gene by converting the start codon into a stop codon intransient expression assays (Figure 5C). The abolition of MYB51uORF slightly increased the expression of the reporter gene, butthe change was not statistically significant under the conditionswe used for transient expression assays (Figure 5C). The identi-fication of MYB34 and MYB51 uORFs validates the use of RNAdegradome data in discovering lineage or species specific reg-ulatory uORFs.

Ribosome Stacking Was Predicted in CDSs Using the RNADegradome Data

BesidesCPuORFs, the nascent peptide encodedbyaCDS regionin the first exon of ArabidopsisCGS1, known as theMTO1 region,has been reported to block ribosome elongation and induce RNAdecay in response to S-adenosyl-L-methionine (AdoMet)(Onouchi et al., 2005). Similar to the degradation fragments weobserved in bZIP CPuORFs (Figure 2), Haraguchi et al. (2008)detected a ladder of truncated 59 termini separated in length by;30-nucleotide increments in the MTO1 region together with the59 upstream region after treatment with AdoMet. Moreover, theprevious study demonstrated that the truncated mRNA ends aredefinedby the59edgesof stalled ribosomes inanarray. Therefore,we assumed that the RNA degradome could also be used in theidentificationof ribosomestackingoccurring in theCDSregion. Tovalidate this idea, we first compared the PARE peaks around theMTO1 region with the 59 ends of degradation intermediates re-ported previously (Yamashita et al., 2014). Although the plants weused for PARE library construction were grown in soil without

AdoMet treatment, themajor PARE peaks were in close proximityto the 59 termini of degradation intermediates reported previously(Supplemental Figure 8).Next, we performed phasing analysis on PARE peaks in the

CDSs for intervals from 20 to 40 nucleotides. In the analysis ofArabidopsis seedling PARE data, the number of phased regionsdecreased when the length of intervals increased (SupplementalFigure 9A). However, the numbers of phased regions identified atintervals of 32 to 40 nucleotides and 28 nucleotides were sig-nificantly lower than that at the interval of 30 nucleotides in theinflorescence PARE data (Supplemental Figure 9B). To eliminatethe phased regions occurring by chance, we discarded can-didates identified only in a single inflorescence sample. Thisresulted in four phased regions at the intervals of 29 and30 nucleotides and one to three phased regions at the otherintervals (Supplemental Figure 9C).Basedon this result,we thusdiscovered four Arabidopsis protein-coding genes possessinga region potentially associated with protection signatures ofstacked ribosomes (Figure 6). PARE peaks with prominentphasing were identified in genes encoding plastidic type I signalpeptidase 2B (PLSP2B), a pentatricopeptide repeat protein,a RING/U-box superfamily protein, and an unknown protein(Figure 6). However, Ribo-Seq peaks separated in a 30-nucleotideinterval were not detected in these regions, although the po-sition of the most 39 phased PARE peak identified in PLSP2Bshowed predominant accumulation of Ribo-Seq reads (Liuet al., 2013; Juntawong et al., 2014) (Figure 6; SupplementalFigure 10). Notably, the A sites of stalled ribosomes putativelyassociated with these four regions were all predicted to fall inframe 2 but not in the translational frame annotated in TAIR(frame 1) if we assumed that the distance from the ribosome-protected 59 end to the presumed A site was 17 nucleotidesaccording to the result shown in Figure 1. Stalling in the non-translational frame may suggest that the last pausing ribosome inthe four CDS regions we identified is arrested during the step oftranslocation but not decoding.

Analysis of 59-Truncated mRNA Ends Upstream ofmiRNA-Guided Cleavage Sites

Plant miRNAs can guide cleavage in the middle of target sites,resulting in truncated mRNA fragments (Rhoades et al., 2002).However, growing evidence suggests that plant miRNA can re-press target translation (Brodersen et al., 2008; Iwakawa andTomari, 2013; Li et al., 2013b, 2013c; Liu et al., 2013), although theunderlying mechanism is not well characterized. Because ouranalysis of the RNA degradome revealed the footprints of stalledribosomes in uORFs and the CDSs, we predicted that a similaranalysis of miRNA target genes would be useful for elucidatingwhether the binding of plant miRNAs can directly block the

(D)ComparisonofGUSactivity betweenwild-typeandDuORF transgenicplants. Theamountof total proteinwasused for thenormalizationofGUSactivity.(E)Comparison ofGUSmRNA level betweenwild-type andDuORF transgenic plants byqRT-PCR. ThemRNA level ofUBQ5wasused for the normalizationofGUSmRNA level. In (D) and (E), each bar represents the mean of measurements derived from five independent T1 transgenic plants6 SE relative to themeasurement of the wild type. *P < 0.05; ns, no significant difference (two-tailed Student’s t test, n = 5).

movement of ribosomes in planta. Therefore, we investigated thedistribution of PARE reads in a 55-nucleotide region upstream ofputative miRNA-guided cleavage sites in Arabidopsis and rice.Unlike the predominant accumulation of PARE reads at positions

16 and 46 nucleotides upstream of uORF stop codons (Figure 2),no position-specific enrichmentwas found in this region except atthe putative miRNA-guided cleavage sites (Figure 7). This resultthus suggests that the major mechanism by which plant miRNAs

Figure 5. Lineage-Specific CPuORFs in MYB34 and MYB51 Generate 59-Truncated RNA Ends Separated in 30 Nucleotides and Negatively RegulateDownstream ORF Expression.

(A) The positional distribution of 59 ends of truncated RNA generated by PARE and ribosome-protectedmRNA fragments generated by Ribo-Seq in the 59UTR of MYB34 and MYB51. The PARE data of seedlings were generated by this study and the Ribo-Seq data plotted were retrieved from the data setpublished by Liu et al. (2013). Regions of CPuORFs are shown as dark-blue lines under the graphs and the first nucleotide of the stop codon is assignedposition 0. The PARE or Ribo-Seq peaks at positions 16 and 46 upstream of the uORF stop codon are highlighted in red. TP50M, tags per 50 million.(B)Alignment ofpeptidesencodedbyMYB34andMYB51uORFs in theBrassicaceae family. Ath,Arabidopsis thaliana; Aly,Arabidopsis lyrata; Bra,Brassicarapa; Bst, Boechera stricta; Cru, Capsella rubella; Esa, Eutrema salsugineum. The alignment is colored according to residue conservation: red, identicalresidues; orange, conserved residues; pink, block of similar residues.(C) Transient expression assays ofMYB34 andMYB51 uORF regulatory function using LUC reporter constructs in protoplasts. The reporter constructs ofthewild typeanddeleteduORF (DuORF ) are illustrated. The redarrowhead indicates the sitemutated. TheLUC reporter constructswere cotransfectedwithacontrol of aGUSgenedrivenbya35Spromoter. TheLUCactivitywasfirst normalized toGUSactivity and then to thevalueofwild-typeconstruct.Eachbarrepresents themeanofmeasurements derived fromsix independent protoplast transfections6 SE relative to themeasurement of thewild type. *P<0.05; ns,no significant difference (two-tailed Student’s t test, n = 6).

2406 The Plant Cell

repress target translationmaynotbebyacting asphysical barriersto hinder the movement of ribosomes.

Discovery of Potential Footprints of Ribosomes Hindered byAGO7 in Arabidopsis TAS3

Surprisingly, threeArabidopsisTAS3genessharedahighlysimilarbut unusual pattern of truncated 59 RNA ends upstream of the

noncleavable target site of miR390 (Figure 8A). A PARE peak waslocated immediately or four nucleotides upstreamof the first baseof noncleavable target sites of miR390. In the proximal regionupstream of this peak, there were three additional PARE peaksarranged at regular intervals of 28 to 30 nucleotides. This regulardegradation pattern in TAS3 highly resembled that detected inbZIP CPuORFs (Figure 2). Therefore, we assumed that the peakadjacent to the noncleavable miRNA site might be the footprint of

Figure 6. Analysis of PARE Data Reveals CDS Regions Potentially Associated with Stacked Ribosomes.

The positional distribution of 59 ends of truncated RNA generated by PARE and ribosome-protected mRNA fragments generated by Ribo-Seq in CDSregions potentially associated with stacked ribosomes. PARE data of Arabidopsis inflorescences plotted were generated by this study while the Ribo-Seqdata plotted were retrieved from the data set published by Liu et al. (2013). The 30-nucleotide phased PARE peaks and the corresponding Ribo-Seq peaksare highlighted in red with coordinates indicated above. TP50M, tags per 50 million.

AGO7, which loads miR390, whereas the upstream peaks likelyalso represented ribosome footprints. Intriguingly, TAS3 wasannotated tobenoncodingRNAbecauseof lackofa longORFandtheproductionof conserved trans-actingsiRNAs (tasiRNAs)whentargeted by miR390 (Allen et al., 2005).

To determine whether TAS3 might serve as a template fortranslation, we first predicted ORFs in three Arabidopsis TAS3genes. All three TAS3 genes have an ORF that terminates 7 to11 nucleotides upstream of the noncleavable site of miR390(Figure8A). Thecodingnatureof thepredictedORFs in threeTAS3geneswasevaluatedwithRibo-Seqdatapublishedpreviously (Liuet al., 2013; Juntawonget al., 2014). ThepredictedORFs inTAS3aand TAS3b have relatively dense ribosome-protected fragmentscomparedwith other regions, supporting the coding ability (Figure8A; Supplemental Figure 11). Moreover, themajor peaks revealedbyRibo-Seqwere close to the predominant PAREpeaks inTAS3aand TAS3b. This result suggested that translated ORFs werepresent upstream of the noncleavable target site of miR390 and59-truncated RNA ends mapped to this region were likely ribo-some footprints. If these PARE peaks represent the footprints ofribosomes hindered by miR390-AGO7 complex but not thecleavage remnants directed by tasiRNAs, they should disappearin the ago7 but be sustained in the mutant of RNA-DEPENDENTRNAPOLYMERASE6 (RDR6), which acts downstreamof AGO7 inthe tasiRNA biogenesis pathway (Mallory and Vaucheret, 2010).As predicted, the result of modified RLM 59 RACE clearly showed

that degradation fragments with 59 ends mapped to these posi-tions were barely detected in ago7 (Figure 8B). On the other hand,the amounts of these degradation fragments were comparablebetween wild-type and rdr6 plants, although both ago7 and rdr6showed increased amounts of TAS3a transcripts compared withthe wild type and no tasiRNA production (Figures 8B to 8D). ThePAREdataofwild-type, rdr6, andago7 inflorescencesshowed thesameresult as themodifiedRLM59RACEassay.AllPAREpeaks inthe ORFs of three TAS3 genes vanished in ago7 except the one atposition 258 in TAS3b (Supplemental Figure 12). The result ofmutant analysis therefore indicates that the formation of these59-truncatedRNAendsupstreamof thenoncleavable target siteofmiR390 depends on AGO7 but is independent of tasiRNA. Takentogether, these results imply that noncleavable targeting of AGO7may arrest ribosomes as a road block in planta. Because the lastmajor PARE peaks in the ORFs in TAS3a, TAS3b, and TAS3c arelocated 19, 14, and16nucleotides upstreamof the first nucleotideof the stop codons respectively, the miR390-AGO7 complexappears to arrest a ribosome in the step of elongation on TAS3abut in the step of termination on TAS3b and TAS3c.

DISCUSSION

The RNA degradome is composed of degradation products ofendoribonucleases and XRNs acting through diverse pathways.Here, we not only demonstrate that ribosome-protected mRNA

Figure 7. No Site-Specific Enrichment of 59-Truncated RNA Ends Is Detected in the Proximal Region Upstream of miRNA-Guided Cleavage Sites.

Clustered heat maps of 59-truncated RNA ends mapped to a 55-nucleotide region upstream of miRNA-guided cleavage sites in seedlings and in-florescences ofArabidopsis and rice. TheArabidopsisPAREdataplottedweregeneratedby this study,while the ricePAREdataplottedwere retrieved fromthe data sets published by Li et al. (2010) and Zhou et al. (2010). The presumedmiRNA-guided cleavage site is assigned position 0 and the color representsthe value of Peak_Index, which is calculated by dividing the reads starting at the position indicated by the total reads in a 31-nucleotide flanking region. Thenumbers of knownmiRNA target sites (indicated in parentheses) and target sites possessing PARE reads and included in heatmaps are shown above heatmaps.

2408 The Plant Cell

Figure 8. Arabidopsis TAS3 Genes Accumulate AGO7-Dependent but RDR6-Independent Phased 59-Truncated RNA Ends Upstream of NoncleavablemiR390 Target Sites.

(A)Thedistribution of 59 endsof truncatedRNAgenerated byPAREand ribosome-protectedmRNA fragments generated byRibo-Seq in three ArabidopsisTAS3 genes. The PARE data of seedlings plottedwere generated by this study and the Ribo-Seq data of light-treated seedlings plottedwere retrieved fromthe data set published by Liu et al. (2013). The predominant PARE and Ribo-Seq peaks are highlighted in red and marked with their distances to the firstnucleotides of noncleavable miR390 target sites. TP50M, tags per 50 million.

ends are pervasive in the RNA degradome, but also demonstrateexciting new applications of RNA degradome data in the study ofposttranscriptionalgene regulation.Predominantaccumulationof59-truncated mRNA ends in a 30-nucleotide interval likely rep-resents an array of stacked ribosomes in the uORF, CDS, or theupstream region of noncleavable binding sites of AGO7 (Figure 9).Genome-wide analysis of 59-truncated mRNA ends mapped touORFs suggests that many CPuORFs may repress downstreamORF expression by stalling ribosomes at CPuORF stop codons,whereas themajorityof predicteduORFs lack this ability (Figure3).In addition, the analysis of 59-truncated mRNA ends occurringupstream of miRNA target sites uncovered the signature of ri-bosomestacking in the noncleavable target site ofmiR390but notother cleavable target sites (Figures 7 and 8). Novel regulatoryuORFs andCDS regions with potential to cause ribosome stallingwere identified through the analysis of the RNA degradome(Figures 4 to 6, Table 1).

Comparison of Ribosome Footprints in RNA DegradomeData and Ribo-Seq Data

Accumulation of degradation intermediates starting at the 59 edgeor the A site of stalled ribosomes has been reported in bacteria,yeast, and Arabidopsis (Hayes and Sauer, 2003; Sunohara et al.,2004; Doma and Parker, 2006; Haraguchi et al., 2008), and it hasbeen proposed that endoribonucleases are involved in the pro-duction of these degradation fragments. However, a recent ge-nome-wide survey of monophosphorylated 59 RNA ends in yeastshowed that XRN1 is also involved in the accumulation of 59-truncated RNA ends protected by ribosomes during RNA decay(Pelechano et al., 2015). Using the plant degradome data, weidentified an array of 59-truncated RNA ends separated at30-nucleotide intervals in uORFs and CDSs, implying an array ofstacked ribosomes (Figures 2 and 4 to 6). However, the30-nucleotide phasing pattern was not observed in Ribo-Seq datafor the regions we identified from RNA degradome analysis. Thediscrepancy between Ribo-Seq data and RNA degradome data islikelydue to thedifferentproceduresof librarypreparation, althoughboth methods could capture ribosome-protected mRNA frag-ments. A typical Ribo-Seq protocol includes a step to selectfragments protectedby a single ribosomebut not stackeddisomesor trisomes (Ingolia, 2010). Therefore, a region associated withstacked ribosomes may rather show a low density of protectedfragments in Ribo-Seq data. On the other hand, degradome se-quencing, PARE, genome-wide mapping of uncapped transcripts,

or 5Pseq profiles truncated RNA termini without monosome se-lection and can thus capture fragments protected by stacked ri-bosomes. In addition, in the absence of monosome selection andtranslation inhibitor treatment to block ribosomes onmRNA, thesemethodsmayalsoallowthe identificationof fragmentsprotectedbyribosomes on distinct translation states.A previous global study of yeast ribosome footprints demon-

strated that ribosome-protected fragments fall into two majorgroups (28 to30nucleotides longand20to22nucleotides long) thatare stabilized by different translation inhibitors (Lareau et al., 2014).The size of the protected fragments reflects the configuration andtranslation state of the ribosomes. In the three Arabidopsis TAS3genes, we detected PARE peaks arranged at 24- to 30-nucleotideintervals immediately upstream of the first base of noncleavablemiR390 target sites (Figure 8A). The shorter intervals between twoPARE peaks may imply a ribosome pausing at the translocationstage instead of the decoding stage, which should lead toa 30-nucleotide protected fragment in Arabidopsis. In addition totechnical differences, the interplay between ribosome stalling andRNA decay likely enhances the signal of stalled ribosomes in RNAdegradomedata. By contrast, the signal of stalled ribosomesmightbe embedded in the footprints of active ribosomes in Ribo-Seqdata. Insummary, theuseofdegradomedata for the identificationofregionsassociatedwithstackedribosomesmayoutperformtheuseof Ribo-Seq data in some cases.

Investigation of Regulatory uORFs Using theRNA Degradome

Comprehensive identification of regulatory uORFs has beenchallenging as predicted uORFsmay not be translated or regulatedownstream ORF expression through distinct mechanisms(Barbosa et al., 2013). Prior to the development of Ribo-Seq,translated uORFs with regulatory functions were often identifiedbymutant screening or evolutionary conservation (Hill andMorris,1993; Delbecq et al., 1994; Wiese et al., 2004; Imai et al., 2006;HaydenandJorgensen,2007).AlthoughRibo-Seqdatahavebeenused in the identification of translated uORFs (Fritsch et al., 2012;Liu et al., 2013; Ingolia et al., 2014), the application of Ribo-Seqdata to the identification of ribosome stalling uORFs has not beenreported. In this study, we show that some CPuORFs over-accumulate 59-truncated RNA ends with a signature of stackedribosomes (Figure 2A). We further demonstrate that the accu-mulation of 59-truncatedRNAends inbZIPCPuORFs is enhancedin response to a high concentration of sucrose (Figure 2B). The

(B)Comparison of 59-truncated RNAends generated in the region upstreamof the noncleavablemiR390 target site ofTAS3a in wild type, rdr6, and ago7 bythemodifiedRLM59RACE assay. The bracket indicates the PCRproducts excised and cloned for Sanger sequencing (left panel). In total, 12 and 13 clonesweresequenced for thewild typeand rdr6, respectively. Thepositional distributionof 59-truncatedRNAends revealedby themodifiedRLM59RACEassay isplotted relative to the non-cleavable miR390 target site (right panel). The cleavage target of miR159,MYB65, is used as a positive control for the modifiedRLM 59 RACE assay. In (A) and (B), the first nucleotide of the noncleavable miR390 target site is assigned position 0.(C)Comparison of TAS3a transcript levels in the wild type, rdr6, and ago7 by qRT-PCR. ThemRNA level ofUBQ5was used for the normalization of TAS3amRNA levels. Each bar represents themean ofmeasurements derived from four biological replicates6 SE relative to themeasurement of thewild type. *P <0.05; **P < 0.01 (two-tailed Student’s t test, n = 4).(D)Comparison of TAS3a tasiRNA produced in the wild type, rdr6, and ago7 by RNA gel blot with U6 as the loading control. Numbers to the left of the blotshow sizes in nucleotides. Antisense DNA oligonucleotides were used as probes for U6 and a TAS3 tasiRNA.

2410 The Plant Cell

data thus support the previous hypothesis that the conservedpeptides encoded by these bZIPCPuORFs can stall ribosomes atstop codons in response to sucrose concentration (Wiese et al.,2004) and suggest the application of degradome data in the studyof regulatory uORFs. Through the global analysis of 59-truncatedRNA ends occurring in predicted uORFs, we identified novel ri-bosome stalling uORFs in Arabidopsis. Notably, uORFs inMYB34andMYB51 are conserved and specific in theBrassicaceae family(Figure 5B). Since RNA degradome data of diverse plant speciesare available in the public domain, these data sets might be veryuseful for the identification of lineage or species specific ribosomestalling uORFsasevolutionary conservation is not required for thistype of analysis.

Analysis of the RNA Degradome Allows Dissection ofRibosome Pausing

Besides uORFs,CDS regions alsopossess 30-nucleotide phased59-truncated mRNA ends (Figure 6). Some regions likely encodenascent peptides which can stall ribosomes during translationelongation as the MTO1 region in CGS1 or are upstream ofstructured regions that can block the movement of ribosomes(Onouchi et al., 2005). The one with the last peak of 59-truncatedRNA ends located 16 nucleotides upstream of main ORF stopcodons as in the degradation pattern observed in CPuORFs maycause ribosome pausing during translation termination. A pre-vious study showed that the binding of EUKARYOTIC RELEASEFACTOR 1 (eRF1) to three types of stop codons causes a con-formational change in the ribosome and a 2-nucleotide shift to-ward the 39 end of mRNA (Kryuchkova et al., 2013). Because theposition17nucleotidesupstreamof themainORFstopcodon is inframe with the 3-nucleotide periodicity observed in the CDSs(Figure 1A), the specific accumulation of 59-truncatedmRNAendsat this position suggests the pausing of a ribosomewith the A siteat thestopcodon.On theotherhand, the59-truncatedmRNAendspeak 16 nucleotides upstream of the stop codon may imply thestalling of ribosomes after the conformational change of ribosomes

induced by the binding of eRF1. The integration of the RNA de-gradome data, protein sequences, and RNA structures will bringnew insights into the regulation of ribosome pausing.

Exploration of Plant miRNA-Mediated TranslationalRepression with RNA Degradome

Unlike animal miRNAs, which mainly target the 39 UTR, the ma-jority of plant miRNAs bind to their target CDSs through nearlyperfect base pairing. Although cleavage sites in the middle ofplant miRNA target sites have been extensively validated bydegradome data in many species, translational repression me-diated by plant miRNAs has also been demonstrated throughmultiple approaches. Previous analysis of Arabidopsis Ribo-Seqdata showed that miRNA targets have lower translational effi-ciency compared with non-miRNA targets (Liu et al., 2013).However, no preferential accumulation of ribosome footprintswas observed in the region upstream of miRNA target sites inArabidopsis Ribo-Seq data (Liu et al., 2013). Because Ribo-Seqhas some drawbacks in detecting stacked ribosomes or ribo-somes pausing at the translocation stage as we have discussed,we used RNA degradome data to reexamine whether plant miRNAscan directly stall ribosomes. Consistent with the previous observa-tion in Ribo-Seq data (Liu et al., 2013), no site-specific enrichmentof 59-truncated RNA ends was found in the region upstreamof Arabidopsis and rice miRNA target sites (Figure 7). Thisstrengthens the notion that directly blocking ribosomemovementthrough the binding of RNA-induced silencing complex contrib-utes little to plant miRNA-mediated translational repression inplanta. Nevertheless, the binding of AGO7 to a well characterizednoncleavable target site ofmiR390seems tohinder themovementof ribosomes and cause ribosome stacking as the upstream re-gions show signatures of stacked ribosomes in the same way asbZIP CPuORFs (Figures 2 and 8). The same degradation patterncorresponding to ribosomestackingamong the threeArabidopsisTAS3 genes suggests that this unique configuration of ORFs andnoncleavable miR390 target sites might be crucial for tasiRNA

Figure 9. Schematic Representation of RNA Degradation Fragments Protected by Stacked Ribosomes in Distinct Regions.

(A) Degradation signatures caused by stacked ribosomes upstream of uORF stop codons. Position 0 is the first nucleotide of the uORF stop codon.(B) Degradation signatures caused by stacked ribosomes in the CDS. Position 0 is the 39 edge of the most 39 stalled ribosome.(C) Degradation signatures caused by stacked ribosomes upstream of a noncleavable miRNA target site bound by AGO7. Position 0 is the first nucleotideof the miRNA target site.Predominant peaks of 59-truncatedRNAends are indicatedwith vertical lines and their distances to theputative sites causing ribosome stalling (highlightedin red) or to the 39 edge of the most 39 stalled ribosome (highlighted in blue) are marked below. The height of vertical lines represents the abundance of 59-truncated RNA ends. 60S, 60S subunit of ribosome; 40S, 40S subunit of ribosome.

production. Although the function of ORFs upstream of themiR390 noncleavable site has not been tested, positioninga target siteofmiR173, anotherwell-known tasiRNA triggerboundto AGO1 (Cuperus et al., 2010), within 10 nucleotides of a stopcodon of an upstreamORFwas shown to enhance the productionof artificial tasiRNA (Zhang et al., 2012). Moreover, a recentlypublished article also demonstrated that an ORF surrounding themiR173 target site on TAS2 is translated and plays a crucial role intasiRNA production (Yoshikawa et al., 2016). That signatures ofstacked ribosomes exist close to themiR390 noncleavable targetsite revealed in this study strongly suggests that translationcontributes to tasiRNA biogenesis through a conserved mecha-nism regardless of miRNA triggers or AGO involved.

A bottleneck in the development of RNA degradome dataapplications is the complex composition of mRNA degradationintermediates, complicating interpretation. The discovery of ri-bosome footprints in RNA degradome data opens up the pos-sibility of newapplicationsofsuchdata inposttranscriptional generegulations beyond the validation of miRNA-guided cleavage. Itshould be possible to apply the RNA degradome data analysesdemonstrated in this study to many other plant species that haveRNA degradome data available, enabling deeper insights intoribosome stalling and mechanisms of RNA decay.

METHODS

Plant Materials and Growth Conditions

Arabidopsis thaliana (ecotype Col-0) used in this studywas grown in soil oron 0.8%Bacto-agar plates containing half-strengthMurashige andSkoogmedium (pH 5.7) and 1% sucrose under a 16/8-h light/dark cycle with anirradiance of 50 to 90 mmol photons m–2 s–1 at 22°C. For the generation ofPARE data, 11-d-old seedlings and inflorescences of wild-type Arabi-dopsis were used for total RNA isolation. For the analysis of 59-truncatedRNAendsgenerated frombZIP2andbZIP11by themodifiedRLM59RACEassay, 10-d-old wild-type seedlings were transferred into liquid half-strength Murashige and Skoog medium with or without 6% sucrose. Afterincubation in growth chambers with rotary shaking at 40 rpm underconstant light for 24h, the seedlingswerecollected forRNAextraction. Theinflorescences of Arabidopsis wild type, ago7 (GK-824A08-025510), andrdr6 (CS24285) were harvested for the analysis of 59-truncated RNA endsgenerated from TAS3a with a modified RLM 59 RACE assay and for PARElibrary construction.

PARE Library Construction and Sequencing

Total RNA isolated by PureLink Plant RNA Reagent (Thermo Fisher) andMaxTract high-density gel tubes (Qiagen) was used for PARE libraryconstruction following the protocol published previously (Zhai et al., 2014).PARE libraries were constructed with ;80 mg total RNA and then se-quenced on the Illumina HiSeq 2500 platform.

Modified RLM 59 RACE Assay

Modified RLM 59 RACE assay was performed to detect 59-truncated RNAends using GeneRacer Kit (Thermo Fisher). First, 2 to 3 mg of total RNAisolated by PureLink Plant RNA Reagent and MaxTract high-density geltubes was ligated with the 59 RNA adapter and then reversely transcribedwith the oligo(dT) primer. Next, cDNA was used as the template for PCRanalysiswith aGeneRacer 59 primer anda gene-specificprimer. NestedPCRwasperformedwithaGeneRacer59nestedprimerandagene-specificnested

primer if no PCR products were detected in the primary PCR. Amplifiedproductsof expectedsizeweregelpurified,cloned intopJET1.2/bluntcloningvector or pCR4-TOPOTA vector (ThermoFisher), and sequenced. A target ofmiR159,MYB65, was includedas thepositive control for themodifiedRLM59RACE assay. The primers are listed in Supplemental Table 1.

Analysis of 59-Truncated RNA End Distribution

In addition to the in-housePAREdata of Arabidopsis, previously publishedPARE data of Arabidopsis, rice (Oryza sativa), and soybean (Glycine max)downloaded from the Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih.gov/geo/)were also analyzed in this study. Accession numbers aregiven at the end of Methods. Trimmed reads were mapped to the corre-sponding genomes or gene sequences downloaded from TAIR (https://www.arabidopsis.org/; TAIR 10), the MSU Rice Genome AnnotationProject (http://rice.plantbiology.msu.edu/; release 6.1), and Phytozome(https://phytozome.jgi.doe.gov/; Phytozome v11.0: Gmax_275_Wm82.a2.v1) with Bowtie (http://bowtie-bio.sourceforge.net/; v1.0.0). Only per-fectly matched 20-nucleotide sequences were used in the followingmetagene analyses and sequences of low complexity (repeats $ 15 nu-cleotides) orwith$10hits in thegenomewereexcluded. TheabundanceofPARE sequences was assigned to the position corresponding to the firstnucleotide of the sequence. Known Arabidopsis and rice CPuORF andmiRNA target sites were retrieved from previous reports (Hayden andJorgensen, 2007; Zheng et al., 2012).

For metagene analysis of PARE reads in the regions flanking the startcodon and the stop codon of annotated CDSs, the abundance of PAREreads at each position in the defined regions on a transcript was firstnormalized by dividing the value by the sum of PARE reads starting in thedefined region. Then, the relative abundance at each position across thedefined region was calculated as the sum of normalized abundance ofPARE reads starting at the same position for all genes. For heat maps ofPARE readdistribution in regionsupstreamofuORFstopcodonsormiRNAguided cleavage sites, we normalized the abundance of PARE reads ateach position by dividing the value by the sum of PARE reads starting ina 31-nucleotide window flanking the indicated position. The distribution ofnormalized PARE abundance was then clustered using Ward’s methodwith R package (https://www.r-project.org/; version 2.15.2) and displayedas heat maps.

Ribo-Seq Data Analysis

Previously publishedRibo-Seq data sets of Arabidopsiswere downloadedfrom the GEO database. Accession numbers are given at the end ofMethods. Trimmed reads of length$20 nucleotides weremapped to genesequences downloaded from TAIR with Bowtie. Two-nucleotide mis-matcheswere allowed formappingRibo-Seq reads of normoxia seedlings,whereas perfect matches were used in the mapping of Ribo-Seq reads oflight-treatedseedlings.TheabundanceofRibo-Seqsequenceswasassignedto the position corresponding to the first nucleotide of the sequence.

Identification of uORFs with Footprints of Stacked Ribosomes

RNA ends protected by monosome and disome pausing at uORF stopcodons should result in PARE peaks at positions 16 and 46 nucleotidesupstream of the stop codon. A custom Perl script was used to identifyuORFs possessing this degradation signature with in-house ArabidopsisPARE data (Supplemental Script 1). First, the sequences of Arabidopsis 59UTRs based on TAIR10 annotation were used to predict uORFs by lookingfor anATGcodonpairedwith the nearest in-frame stop codon.Next, PAREreadsweremapped to the predicted uORFs by this Perl script. Candidatesof uORFs with a signature of ribosome stacking at the stop codon wereselected based on the following criteria. To capture the signature of ri-bosome stacking, the distribution of PARE reads was evaluated in two

2412 The Plant Cell

31-nucleotide regions upstream of the stop codon. The first region spanspositions 21 to231 and the second region spans positions232 to262,with thefirst nucleotideof theuORFstopcodon isset to0. In thefirst region,themost abundant peak (thefirstmajor peak)was required tobeat position216or217 andwith the number of raw reads$3. In the second region, themost abundant peak (the secondmajor peak) was required to be located atposition 245,246, or247. Moreover, the abundance of the major peakswas required to be at least 2-fold higher than that of the second mostabundant peak in the same region but outside the possible positions formajor peaks.

Identification of Degradation Signatures Representing StackedRibosomes in the CDSs

To identify degradation signatures representing stacked ribosomes in theCDSs, we first mapped trimmed PARE sequences to Arabidopsis CDSsbased on TAIR10 annotation with Bowtie. Because the protected degra-dation termini caused by three stacked ribosomes are three predominantPARE peaks separated by ;30 nucleotides, we calculated two values,Peak_Abundance andPeak_Index, to evaluate thepredominance of PAREreads accumulated at each position. Peak_Abundance was defined as thesum of PARE reads starting at the indicated position together with thepositions one nucleotide upstream and downstream. Peak_Index wascalculated by dividing the Peak_Abundance by the total PARE readsstarting in a 31-nucleotide window flanking the indicated position. Posi-tionswith Peak_Index$ 0.3were selected as the firstmajor peak andwerereset to 0 in further analysis. The second and third major peaks were re-quired to be at the regions between229 and231, and259 and261 withPeak_Index$ 0.3. In addition, the abundance of three major peaks had tobe the highest in the region for the calculation of Peak_Index and with rawreads$3.Theabundanceof the threemajorpeaksalsoneeded tobe2-foldhigher than that of the second most abundant peak falling in the31-nucleotide window but outside the possible positions for major peaks.To ensure that the degradation signature caused by stacked ribosomeswas prominent among all degradation events within a gene, a region wasselected only if all three major peaks were ranked in the top 1% of allpositionswith regard toPAREabundance. Toeliminate thephased regionsoccurring by chance, the candidates uncovered from only a single samplewere removed. The in-house Perl script for PARE peak analysis in CDSs isprovided as Supplemental Script 2. The same criteria were applied for thephasing analysis of PARE peaks for intervals from 20 to 40 nucleotides.

Generation of Transgenic Lines

A 1.2-kbDNA fragment of ArabidopsisCIPK6 promoter region plus 59UTRwas cloned into gateway vector pHGWFS7 upstream aGUS reporter genethroughGateway LRClonase II Enzymemix (Thermo Fisher). Site-directedmutagenesis was applied to change the start codon of CIPK6 CPuORF toa stop codon. The two constructs with wild-type uORF and DuORF weretransformed into wild-type Arabidopsis through the floral dip method(Zhang et al., 2006). Primers used to clone the 1.2kb CIPK6 promoterfragment are listed in Supplemental Table 1.

GUS Activity Assay and GUS Staining

The extract of ground tissues extracted by GUS extraction buffer (50 mMNaHPO4, pH 7.0, 10 mM 2-mercaptoethanol, 10 mM Na2EDTA, 0.1%sodium lauryl sarcosine, and 0.1% Triton X-100) was mixed with MUGbuffer (4-methylumbellifery b-D-glucuronide) at 37°C for 20 min, and thenthe reaction was terminated by adding stop buffer (0.2 M sodium car-bonate). The fluorescence intensity of 4-methylumbelliferone was mea-sured by fluorometer at 450 nm when excited at 365 nm. The amount of4-methylumbelliferone was calculated with the standard curve and thennormalized to the amount of total protein which was measured using the

Bradford dye bindingmethod with the Bio-Rad Labs protein assay kit. TheGUS transgenicplantswerestained inGUSstainingsolution [0.1MNaPO4,pH 7.0, 10 mM EDTA, 0.1% Triton X-100, 1 mM K3Fe(CN)6, and 2 mMX-Gluc] at 37°C overnight. After staining, samples were washed with 50%ethanol until chlorophyll was removed, which took either overnight orseveral days depending on the tissue. Then, the stained samples wereobserved under a stereomicroscope (SteREO Lumar.V12; Zeiss) andphotographed with a digital camera (AxioCam MRc; Zeiss).

Quantitative RT-PCR

Two micrograms of total RNA was used as a template for reverse tran-scription with ToolsQuant II Fast RT kit (Biotools). The resulting cDNAwasdiluted 20-fold and 5 mL was used for qRT-PCR in a 20 mL reaction withSYBR Green PCR Master Mix (Applied Biosystems) on a 7500 Fast Real-Time PCR System (Applied Biosystems) using the following program: 20 sat 95°C, followed by 40 cycles of 3 s at 95°C, and 30 s at 60°C, with anadditional melt curve stage consisting of 15 s at 95°C, 1 min at 60°C, and15sat95°C.TAS3aandGUSexpression levelswerenormalized to the levelof UBQ5 expression and were averaged from at least three independentbiological samples, followed by normalizing to the corresponding value ofthe wild type. The primers used are listed in Supplemental Table 1.

Protoplast Transient Assay

For construction of reporter plasmids, MYB34 and MYB51 59 UTRs wereamplified fromArabidopsisgenomicDNAandcloned intopJD301betweenthe cauliflower mosaic virus 35S promoter and the firefly luciferase (LUC)coding regionusing theNcoI restriction site. ToabolishMYB34andMYB51uORFs, the start codon of uORFswas converted into a stop codon by site-directedmutagenesis. Primers used in thecloningofMYB34andMYB5159UTRsequences are listed in Supplemental Table 1. Arabidopsismesophyllprotoplasts were isolated from 3- to 4-week-old rosette leaves followingthemethoddescribedpreviously (Wuet al., 2009). Equal amounts (20mg) ofreporter plasmids and internal control plasmids containing the 35S pro-moter drivingaGUSgenewere cotransfected into 105protoplasts in aPEGsolution (40% polyethylene glycol 4000, 0.2 Mmannitol, and 0.1 M CaCl2)at room temperature for 5 to 10 min. The transfected protoplasts wereincubated at 22°C in the dark for 16 h and then lysedwithCell Culture LysisReagent (Promega). LUC activity was measured with a Luciferase AssaySystem (Promega) according to the manufacturer’s instructions andnormalized to GUS activity, which was measured as described above.

RNA Gel Blot of Small RNA

Small RNA gel blot analysis was performed as described previously (Leeet al., 2015). The probes used in the detection of small RNAs derived fromTAS3 and U6 are listed in Supplemental Table 1.

Alignment of Peptides Encoded by MYB34 and MYB51 uORFs

ThesequencesofMYB34andMYB51genes indifferentplant specieswereidentified by a BLASTP search and downloaded from the Phytozomedatabase (https://phytozome.jgi.doe.gov). The 59 UTR sequences of or-thologous genes retrieved for uORF analysis with Serial Cloner (http://serialbasics.free.fr/; version 2.6) are listed in Supplemental Data Set 1.Peptides encoded byMYB34 andMYB51 uORFwere then alignedwith theuse of Vector NTI software (Thermo Fisher).

Accession Numbers

The PARE data generated in this study as well as public PARE data areavailable in theGEOdatabaseunder seriesGSE77549 (Arabidopsis in-housePARE data) and accession numbers GSM280226 (public Arabidopsis

inflorescence PAREdata), GSM647200 (public soybean seedPARE data),GSM434596 (public riceseedlingPAREdata), andGSM476257 (public riceyoung inflorescencePAREdata).TheArabidopsisRibo-Seqdatasetsusedin this study are available in the GEO database under the accessionnumbers GSM1226369 (light-treated seedlings) and GSM1224475 (nor-moxia seedlings). Sequences of individual genes used in PARE dataanalysis or functional assays andmutants used canbe found in the TAIR orPhytozome databases under the following accession numbers: AT3G02470for Ath-SAMDC, AT2G18160 for Ath-bZIP2, AT4G34590 for Ath-bZIP11,AT3G62420 for Ath-bZIP53, AT4G30960 for Ath-CIPK6, Glyma09g14090for Gly-CIPK6, AT5G60890 for Ath-MYB34, AT1G18570 for Ath-MYB51,AT3G01120 for Ath-CGS1, AT3G17185 for Ath-TAS3a, AT5G49615 forAth-TAS3b, AT5G57735 for Ath-TAS3c, AT1G69440 for Ath-ago7, andAT3G49500 for Ath-rdr6.

Supplemental Data

Supplemental Figure 1. 59-Truncated RNA Ends Show a 3-NucleotidePeriodicity and Significant Frame Bias in the CDS.

Supplemental Figure 2. Overaccumulation of 59-Truncated RNA Endsat the 39 End of Arabidopsis SAMDC uORF.

Supplemental Figure 3. The Comparison of PARE and Ribo-SeqRead Distribution in Arabidopsis bZIP uORFs.

Supplemental Figure 4. Site-Specific Enrichment of 59-TruncatedRNA Ends in Arabidopsis CPuORFs.

Supplemental Figure 5. PARE Read Distribution in ArabidopsisuORFs with RNA Degradation Signatures Representing RibosomeStacking at Stop Codons.

Supplemental Figure 6. The Comparison of PARE and Ribo-SeqRead Distribution in Arabidopsis CIPK6, MYB34, and MYB51 uORFs.

Supplemental Figure 7. Negative Regulation of CIPK6 uORF inReporter Gene Expression in Various Tissues.

Supplemental Figure 8. The Comparison of PARE Read Distributionwith the 59 Ends of the Degradation Intermediates Previously Identifiedaround the MTO1 Region of CGS1.

Supplemental Figure 9. Phasing Analysis of PARE Peaks in the CDS.

Supplemental Figure 10. The Comparison of PARE and Ribo-Seq ReadDistribution in CDS Regions with 30-Nucleotide Phased PARE Peaks.

Supplemental Figure 11. The Comparison of PARE and Ribo-SeqRead Distribution in Arabidopsis TAS3 Genes.

Supplemental Figure 12. Three Arabidopsis TAS3 Genes AccumulateAGO7-Dependent but RDR6-Independent Phased 59-Truncated RNAEnds Upstream of Noncleavable miR390 Target Sites.

Supplemental Table 1. Sequences of Primers for Cloning, ModifiedRLM 59 RACE, qRT-PCR, and Probes for RNA Gel Blot Analysis.

Supplemental Data Set 1. Sequences for the Analysis of ConserveduORFs in MYB34 and MYB51.

Supplemental Script 1. In-House Perl Script for PARE Peak Analysisin uORF.

Supplemental Script 2. In-House Perl Script for PARE Peak Analysisin CDSs.

ACKNOWLEDGMENTS

WethankShu-HsingWuandTzyy-JenChiouofAcademiaSinica forhelpfuldiscussions andMing-CheShih for supportingHsiao-ChunChou.Wealsothank the Academia Sinica Agricultural and Biotechnology Research

Center core facilities for help with transgenic plants and protoplast assaysand Miranda Loney for English editing of this article. This work wassupported by Academia Sinica.

AUTHOR CONTRIBUTIONS

H.-M.C. designed the research. C.-Y.H., H.-C.C., and H.-M.C. performedthe sequence data analyses.W.-C.L., H.-C.C., A.-P.C., and S.-J.C. carriedout experiments. C.-Y.H., W.-C.L., and H.-M.C. wrote the article. Allauthors read and approved the final manuscript.

Received April 12, 2016; revised September 14, 2016; accepted October13, 2016; published October 14, 2016.

REFERENCES

Addo-Quaye, C., Eshoo, T.W., Bartel, D.P., and Axtell, M.J. (2008).Endogenous siRNA and miRNA targets identified by sequencing ofthe Arabidopsis degradome. Curr. Biol. 18: 758–762.

Allen, E., Xie, Z., Gustafson, A.M., and Carrington, J.C. (2005).MicroRNA-directed phasing during trans-acting siRNA biogenesisin plants. Cell 121: 207–221.

Barbosa, C., Peixeiro, I., and Romão, L. (2013). Gene expressionregulation by upstream open reading frames and human disease.PLoS Genet. 9: e1003529.

Bartel, D.P. (2009). MicroRNAs: target recognition and regulatoryfunctions. Cell 136: 215–233.

Bazzini, A.A., Johnstone, T.G., Christiano, R., Mackowiak, S.D.,Obermayer, B., Fleming, E.S., Vejnar, C.E., Lee, M.T., Rajewsky,N., Walther, T.C., and Giraldez, A.J. (2014). Identification of smallORFs in vertebrates using ribosome footprinting and evolutionaryconservation. EMBO J. 33: 981–993.

Bender, J., and Fink, G.R. (1998). A Myb homologue, ATR1, activatestryptophan gene expression in Arabidopsis. Proc. Natl. Acad. Sci.USA 95: 5655–5660.

Bologna, N.G., Schapire, A.L., Zhai, J., Chorostecki, U.,Boisbouvier, J., Meyers, B.C., and Palatnik, J.F. (2013). MultipleRNA recognition patterns during microRNA biogenesis in plants.Genome Res. 23: 1675–1689.

Brodersen, P., Sakvarelidze-Achard, L., Bruun-Rasmussen, M.,Dunoyer, P., Yamamoto, Y.Y., Sieburth, L., and Voinnet, O.(2008). Widespread translational inhibition by plant miRNAs andsiRNAs. Science 320: 1185–1190.

Celenza, J.L., Quiel, J.A., Smolen, G.A., Merrikh, H., Silvestro, A.R.,Normanly, J., and Bender, J. (2005). The Arabidopsis ATR1 Mybtranscription factor controls indolic glucosinolate homeostasis.Plant Physiol. 137: 253–262.

Cuperus, J.T., Carbonell, A., Fahlgren, N., Garcia-Ruiz, H., Burke,R.T., Takeda, A., Sullivan, C.M., Gilbert, S.D., Montgomery, T.A.,and Carrington, J.C. (2010). Unique functionality of 22-nt miRNAsin triggering RDR6-dependent siRNA biogenesis from target tran-scripts in Arabidopsis. Nat. Struct. Mol. Biol. 17: 997–1003.

Delbecq, P., Werner, M., Feller, A., Filipkowski, R.K., Messenguy,F., and Piérard, A. (1994). A segment of mRNA encoding the leaderpeptide of the CPA1 gene confers repression by arginine on a het-erologous yeast gene transcript. Mol. Cell. Biol. 14: 2378–2390.

Doma, M.K., and Parker, R. (2006). Endonucleolytic cleavage ofeukaryotic mRNAs with stalls in translation elongation. Nature 440:561–564.

Ebina, I., et al. (2015). Identification of novel Arabidopsis thalianaupstream open reading frames that control expression of the main

2414 The Plant Cell

coding sequences in a peptide sequence-dependent manner. Nu-cleic Acids Res. 43: 1562–1576.

Frerigmann, H., and Gigolashvili, T. (2014). MYB34, MYB51, andMYB122 distinctly regulate indolic glucosinolate biosynthesis inArabidopsis thaliana. Mol. Plant 7: 814–828.

Frischmeyer, P.A., van Hoof, A., O’Donnell, K., Guerrerio, A.L.,Parker, R., and Dietz, H.C. (2002). An mRNA surveillance mecha-nism that eliminates transcripts lacking termination codons. Sci-ence 295: 2258–2261.

Fritsch, C., Herrmann, A., Nothnagel, M., Szafranski, K., Huse, K.,Schumann, F., Schreiber, S., Platzer, M., Krawczak, M., Hampe,J., and Brosch, M. (2012). Genome-wide search for novel humanuORFs and N-terminal protein extensions using ribosomal foot-printing. Genome Res. 22: 2208–2218.

Gaba, A., Jacobson, A., and Sachs, M.S. (2005). Ribosome occu-pancy of the yeast CPA1 upstream open reading frame terminationcodon modulates nonsense-mediated mRNA decay. Mol. Cell 20:449–460.

Gaba, A., Wang, Z., Krishnamoorthy, T., Hinnebusch, A.G., andSachs, M.S. (2001). Physical evidence for distinct mechanisms oftranslational control by upstream open reading frames. EMBO J. 20:6453–6463.

Gallie, D.R. (1991). The cap and poly(A) tail function synergistically toregulate mRNA translational efficiency. Genes Dev. 5: 2108–2116.

German, M.A., et al. (2008). Global identification of microRNA-targetRNA pairs by parallel analysis of RNA ends. Nat. Biotechnol. 26:941–946.

Gigolashvili, T., Berger, B., Mock, H.P., Müller, C., Weisshaar, B.,and Flügge, U.I. (2007). The transcription factor HIG1/MYB51regulates indolic glucosinolate biosynthesis in Arabidopsis thaliana.Plant J. 50: 886–901.

Gregory, B.D., O’Malley, R.C., Lister, R., Urich, M.A., Tonti-Filippini, J., Chen, H., Millar, A.H., and Ecker, J.R. (2008). A linkbetween RNA metabolism and silencing affecting Arabidopsis de-velopment. Dev. Cell 14: 854–866.

Grubb, C.D., and Abel, S. (2006). Glucosinolate metabolism and itscontrol. Trends Plant Sci. 11: 89–100.

Guo, H., Ingolia, N.T., Weissman, J.S., and Bartel, D.P. (2010).Mammalian microRNAs predominantly act to decrease targetmRNA levels. Nature 466: 835–840.

Haraguchi, Y., Kadokura, Y., Nakamoto, M., Onouchi, H., andNaito, S. (2008). Ribosome stacking defines CGS1 mRNA degra-dation sites during nascent peptide-mediated translation arrest.Plant Cell Physiol. 49: 314–323.

Harigaya, Y., and Parker, R. (2012). Global analysis of mRNA decayintermediates in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci.USA 109: 11764–11769.

Hayden, C.A., and Jorgensen, R.A. (2007). Identification of novelconserved peptide uORF homology groups in Arabidopsis and ricereveals ancient eukaryotic origin of select groups and preferentialassociation with transcription factor-encoding genes. BMC Biol. 5:32.

Hayes, C.S., and Sauer, R.T. (2003). Cleavage of the A site mRNAcodon during ribosome pausing provides a mechanism for trans-lational quality control. Mol. Cell 12: 903–911.

Hill, J.R., and Morris, D.R. (1993). Cell-specific translational regula-tion of S-adenosylmethionine decarboxylase mRNA. Dependenceon translation and coding capacity of the cis-acting upstream openreading frame. J. Biol. Chem. 268: 726–731.

Hood, H.M., Spevak, C.C., and Sachs, M.S. (2007). Evolutionarychanges in the fungal carbamoyl-phosphate synthetase smallsubunit gene and its associated upstream open reading frame.Fungal Genet. Biol. 44: 93–104.

Hou, C.Y., Wu, M.T., Lu, S.H., Hsing, Y.I., and Chen, H.M. (2014).Beyond cleaved small RNA targets: unraveling the complexity ofplant RNA degradome data. BMC Genomics 15: 15.

Houseley, J., and Tollervey, D. (2009). The many pathways of RNAdegradation. Cell 136: 763–776.

Imai, A., Hanzawa, Y., Komura, M., Yamamoto, K.T., Komeda, Y.,and Takahashi, T. (2006). The dwarf phenotype of the Arabidopsisacl5 mutant is suppressed by a mutation in an upstream ORF ofa bHLH gene. Development 133: 3575–3585.

Ingolia, N.T. (2010). Genome-wide translational profiling by ribosomefootprinting. Methods Enzymol. 470: 119–142.

Ingolia, N.T., Lareau, L.F., and Weissman, J.S. (2011). Ribosomeprofiling of mouse embryonic stem cells reveals the complexity anddynamics of mammalian proteomes. Cell 147: 789–802.

Ingolia, N.T., Ghaemmaghami, S., Newman, J.R., and Weissman,J.S. (2009). Genome-wide analysis in vivo of translation with nu-cleotide resolution using ribosome profiling. Science 324: 218–223.

Ingolia, N.T., Brar, G.A., Stern-Ginossar, N., Harris, M.S.,Talhouarne, G.J., Jackson, S.E., Wills, M.R., and Weissman,J.S. (2014). Ribosome profiling reveals pervasive translation outsideof annotated protein-coding genes. Cell Reports 8: 1365–1379.

Iwakawa, H.O., and Tomari, Y. (2013). Molecular insights into mi-croRNA-mediated translational repression in plants. Mol. Cell 52:591–601.

Juntawong, P., Girke, T., Bazin, J., and Bailey-Serres, J. (2014).Translational dynamics revealed by genome-wide profiling of ribo-some footprints in Arabidopsis. Proc. Natl. Acad. Sci. USA 111:E203–E212.

Kryuchkova, P., Grishin, A., Eliseev, B., Karyagina, A., Frolova, L.,and Alkalaeva, E. (2013). Two-step model of stop codon recogni-tion by eukaryotic release factor eRF1. Nucleic Acids Res. 41:4573–4586.

Lareau, L.F., Hite, D.H., Hogan, G.J., and Brown, P.O. (2014). Dis-tinct stages of the translation elongation cycle revealed by se-quencing ribosome-protected mRNA fragments. eLife 3: e01257.

Lebreton, A., and Séraphin, B. (2008). Exosome-mediated qualitycontrol: substrate recruitment and molecular activity. Biochim. Bio-phys. Acta 1779: 558–565.

Lee, W.C., Lu, S.H., Lu, M.H., Yang, C.J., Wu, S.H., and Chen, H.M.(2015). Asymmetric bulges and mismatches determine 20-nt mi-croRNA formation in plants. RNA Biol. 12: 1054–1066.

Li, B., Duan, H., Li, J., Deng, X.W., Yin, W., and Xia, X. (2013a).Global identification of miRNAs and targets in Populus euphraticaunder salt stress. Plant Mol. Biol. 81: 525–539.

Li, J.F., Chung, H.S., Niu, Y., Bush, J., McCormack, M., and Sheen,J. (2013b). Comprehensive protein-based artificial microRNAscreens for effective gene silencing in plants. Plant Cell 25: 1507–1522.

Li, S., et al. (2013c). MicroRNAs inhibit the translation of targetmRNAs on the endoplasmic reticulum in Arabidopsis. Cell 153: 562–574.

Li, Y.F., Zheng, Y., Addo-Quaye, C., Zhang, L., Saini, A.,Jagadeeswaran, G., Axtell, M.J., Zhang, W., and Sunkar, R.(2010). Transcriptome-wide identification of microRNA targets inrice. Plant J. 62: 742–759.

Liu, M.J., Wu, S.H., Wu, J.F., Lin, W.D., Wu, Y.C., Tsai, T.Y., Tsai,H.L., and Wu, S.H. (2013). Translational landscape of photomor-phogenic Arabidopsis. Plant Cell 25: 3699–3710.

Mallory, A., and Vaucheret, H. (2010). Form, function, and regulationof ARGONAUTE proteins. Plant Cell 22: 3879–3889.

Merret, R., Descombin, J., Juan, Y.T., Favory, J.J., Carpentier,M.C., Chaparro, C., Charng, Y.Y., Deragon, J.M., and Bousquet-Antonelli, C. (2013). XRN4 and LARP1 are required for a heat-triggered

mRNA decay pathway involved in plant acclimation and survival dur-ing thermal stress. Cell Reports 5: 1279–1293.

Merret, R., Nagarajan, V.K., Carpentier, M.C., Park, S., Favory,J.J., Descombin, J., Picart, C., Charng, Y.Y., Green, P.J.,Deragon, J.M., and Bousquet-Antonelli, C. (2015). Heat-inducedribosome pausing triggers mRNA co-translational decay in Arabi-dopsis thaliana. Nucleic Acids Res. 43: 4121–4132.

Muhlrad, D., Decker, C.J., and Parker, R. (1994). Deadenylation of theunstable mRNA encoded by the yeast MFA2 gene leads to decappingfollowed by 59–>39 digestion of the transcript. Genes Dev. 8: 855–866.

Onouchi, H., Nagami, Y., Haraguchi, Y., Nakamoto, M., Nishimura,Y., Sakurai, R., Nagao, N., Kawasaki, D., Kadokura, Y., andNaito, S. (2005). Nascent peptide-mediated translation elongationarrest coupled with mRNA degradation in the CGS1 gene of Ara-bidopsis. Genes Dev. 19: 1799–1810.

Pelechano, V., Wei, W., and Steinmetz, L.M. (2015). Widespreadco-translational RNA decay reveals ribosome dynamics. Cell 161:1400–1412.

Raney, A., Baron, A.C., Mize, G.J., Law, G.L., and Morris, D.R.(2000). In vitro translation of the upstream open reading frame in themammalian mRNA encoding S-adenosylmethionine decarboxylase.J. Biol. Chem. 275: 24444–24450.

Rhoades, M.W., Reinhart, B.J., Lim, L.P., Burge, C.B., Bartel, B.,and Bartel, D.P. (2002). Prediction of plant microRNA targets. Cell110: 513–520.

Schmidt, S.A., Foley, P.L., Jeong, D.H., Rymarquis, L.A., Doyle, F.,Tenenbaum, S.A., Belasco, J.G., and Green, P.J. (2015). Identifi-cation of SMG6 cleavage sites and a preferred RNA cleavage motifby global analysis of endogenous NMD targets in human cells.Nucleic Acids Res. 43: 309–323.

Shamimuzzaman, M., and Vodkin, L. (2012). Identification of soy-bean seed developmental stage-specific and tissue-specific miRNAtargets by degradome sequencing. BMC Genomics 13: 310.

Shoemaker, C.J., and Green, R. (2012). Translation drives mRNAquality control. Nat. Struct. Mol. Biol. 19: 594–601.

Song, Q.X., Liu, Y.F., Hu, X.Y., Zhang, W.K., Ma, B., Chen, S.Y., andZhang, J.S. (2011). Identification of miRNAs and their target genesin developing soybean seeds by deep sequencing. BMC Plant Biol. 11: 5.

Sunohara, T., Jojima, K., Yamamoto, Y., Inada, T., and Aiba, H.(2004). Nascent-peptide-mediated ribosome stalling at a stop co-don induces mRNA cleavage resulting in nonstop mRNA that isrecognized by tmRNA. RNA 10: 378–386.

Takahashi, H., Takahashi, A., Naito, S., and Onouchi, H. (2012).BAIUCAS: a novel BLAST-based algorithm for the identification ofupstream open reading frames with conserved amino acid se-quences and its application to the Arabidopsis thaliana genome.Bioinformatics 28: 2231–2241.

Uchiyama-Kadokura, N., Murakami, K., Takemoto, M., Koyanagi, N.,Murota, K., Naito, S., and Onouchi, H. (2014). Polyamine-responsiveribosomal arrest at the stop codon of an upstream open reading frame of

the AdoMetDC1 gene triggers nonsense-mediated mRNA decay inArabidopsis thaliana. Plant Cell Physiol. 55: 1556–1567.

van Hoof, A., Frischmeyer, P.A., Dietz, H.C., and Parker, R. (2002).Exosome-mediated recognition and degradation of mRNAs lackinga termination codon. Science 295: 2262–2264.

Vasquez, J.J., Hon, C.C., Vanselow, J.T., Schlosser, A., and Siegel,T.N. (2014). Comparative ribosome profiling reveals extensivetranslational complexity in different Trypanosoma brucei life cyclestages. Nucleic Acids Res. 42: 3623–3637.

Vaughn, J.N., Ellingson, S.R., Mignone, F., and Arnim, Av. (2012).Known and novel post-transcriptional regulatory sequences areconserved across plant families. RNA 18: 368–384.

Wang, Z., and Sachs, M.S. (1997). Ribosome stalling is responsiblefor arginine-specific translational attenuation in Neurospora crassa.Mol. Cell. Biol. 17: 4904–4913.

Wiese, A., Elzinga, N., Wobbes, B., and Smeekens, S. (2004). Aconserved upstream open reading frame mediates sucrose-inducedrepression of translation. Plant Cell 16: 1717–1729.

Wu, F.H., Shen, S.C., Lee, L.Y., Lee, S.H., Chan, M.T., and Lin, C.S.(2009). Tape-Arabidopsis Sandwich - a simpler Arabidopsis pro-toplast isolation method. Plant Methods 5: 16.

Yamashita, Y., Kadokura, Y., Sotta, N., Fujiwara, T., Takigawa, I.,Satake, A., Onouchi, H., and Naito, S. (2014). Ribosomes ina stacked array: elucidation of the step in translation elongation atwhich they are stalled during S-adenosyl-L-methionine-inducedtranslation arrest of CGS1 mRNA. J. Biol. Chem. 289: 12693–12704.

Yoshikawa, M., Iki, T., Numa, H., Miyashita, K., Meshi, T., andIshikawa, M. (2016). A role of a short open reading frame en-compassing the microRNA173 target site of the TAS2 transcript intrans-acting small interfering RNA biogenesis. Plant Physiol. 171:359–368.

Zhai, J., Arikit, S., Simon, S.A., Kingham, B.F., and Meyers, B.C.(2014). Rapid construction of parallel analysis of RNA end (PARE)libraries for Illumina sequencing. Methods 67: 84–90.

Zhang, C., Ng, D.W., Lu, J., and Chen, Z.J. (2012). Roles of targetsite location and sequence complementarity in trans-acting siRNAformation in Arabidopsis. Plant J. 69: 217–226.

Zhang, X., Henriques, R., Lin, S.S., Niu, Q.W., and Chua, N.H.(2006). Agrobacterium-mediated transformation of Arabidopsisthaliana using the floral dip method. Nat. Protoc. 1: 641–646.

Zhao, M., Tai, H., Sun, S., Zhang, F., Xu, Y., and Li, W.X. (2012).Cloning and characterization of maize miRNAs involved in re-sponses to nitrogen deficiency. PLoS One 7: e29669.

Zheng, Y., Li, Y.F., Sunkar, R., and Zhang, W. (2012). SeqTar: aneffective method for identifying microRNA guided cleavage sitesfrom degradome of polyadenylated transcripts in plants. NucleicAcids Res. 40: e28.

Zhou, M., Gu, L., Li, P., Song, X., Wei, L., Chen, Z., and Cao, X.(2010). Degradome sequencing reveals endogenous small RNAtargets in rice (Oryza sativa L. ssp. indica). Front. Biol. 5: 67–90.

2416 The Plant Cell

DOI 10.1105/tpc.16.00295; originally published online October 14, 2016; 2016;28;2398-2416Plant Cell

Ming Chen−Jen Chou and Ho−Ping Chen, Shu−Chun Chou, Ai−Chi Lee, Hsiao−Yu Hou, Wen−ChengGlobal Analysis of Truncated RNA Ends Reveals New Insights into Ribosome Stalling in Plants

This information is current as of June 26, 2018

Supplemental Data /content/suppl/2016/10/17/tpc.16.00295.DC2.html /content/suppl/2016/10/14/tpc.16.00295.DC1.html

References /content/28/10/2398.full.html#ref-list-1

This article cites 77 articles, 28 of which can be accessed free at:

Permissions https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X

eTOCs http://www.plantcell.org/cgi/alerts/ctmain

Sign up for eTOCs at:

CiteTrack Alerts http://www.plantcell.org/cgi/alerts/ctmain

Sign up for CiteTrack Alerts at:

Subscription Information http://www.aspb.org/publications/subscriptions.cfm

is available at:Plant Physiology and The Plant CellSubscription Information for

Global Analysis of Truncated RNA Ends Reveals New … · LARGE-SCALE BIOLOGY ARTICLE Global...

Documents

Ff truncated final 3

Regresi Spline Polynomial Truncated Biprediktor untuk

TRUNCATED OCTAHEDRON

Nonparametric Methods for Doubly Truncated DataNonparametric Methods for Doubly Truncated Data Bradley Efron and Vah e Petrosian Abstract Truncated data plays an important role in

MENGGUNAKAN REGRESI ZERO-TRUNCATED NEGATIVE BINOMIAL

CANopen User Manual (truncated version)

Multidimensional NMR analysis reveals truncated lignin ... · Multidimensional NMR analysis reveals truncated lignin structures in wood decayed by the brown rot basidiomycete Postia

OF TRUNCATED ORTHOTROPIC CONICAL SHELLS WITH GENERALIZED ... · OF TRUNCATED ORTHOTROPIC CONICAL SHELLS WITH GENERALIZED ELASTIC EDGE RESTRAINT ... OF TRUNCATED ORTHOTROPIC CONICAL

Analysis and Design of Optimized Truncated Scarfed Nozzles ... · PDF fileAnalysis and Design of Optimized Truncated Scarfed Nozzles ... AND DESIGN OF OPTIMIZED TRUNCATED SCARFED NOZZLES

Recent progress on truncated Toeplitz operators

Truncated Middle Turbinnate Snp

cient Broadband Truncated-Pyramid-Based Metamaterial

UPSET-FORGING OF SINTERED ALUMINIUM TRUNCATED …

Right-truncated data

Truncated Shabbat Evening Service

Friend Like Me Truncated Arr.cbw Synthesizer

Thermoelastic Analysis of Rotating Thick Truncated Conical ...jsm.iau-arak.ac.ir/article_524264_3a069afa854573459815f0e67d00588c.pdf · Thermoelastic Analysis of Rotating Thick Truncated

Truncated Differentials - COSIC · Truncated differentials Impossible differentials Truncated differentials - Feistel network Consider Feistel network where round function is a bijection

Lecture 9 Models for Censored and Truncated Data – Truncated

Joint Representation and Truncated Inference Learning for Correlation ...openaccess.thecvf.com/content_ECCV_2018/papers/Yingjie_Yao_Joi… · Joint Representation and Truncated Inference