Article

Vaibhav Chikn

0022-2836/© 2017 Elsevi

The Canonical Poly (A) Polymerase PAP1Polyadenylates Non-Coding RNAs and IsEssential for snoRNA Biogenesis inTrypanosoma brucei

e1, †, Sachin Kumar Gupta1

, †, Tirza Doniger 1, K. Shanmugha Rajan1,Smadar Cohen-Chalamish1, Hiba Waldman Ben-Asher1, Liat Kolet 1,Nasreen Hag Yahia1, Ron Unger1, Elisabetta Ullu2, Nikolay G. Kolev2,Christian Tschudi3, 4 and Shulamit Michaeli 1

1 - The Mina and Everard Goodman Faculty of Life Sciences and Advanced Materials and Nanotechnology Institute,Bar-Ilan University, Ramat-Gan 5290002, Israel2 - Department of Epidemiology and Microbial Diseases, Yale School of Public Health, New Haven, CT 06536, USA3 - Department of Internal Medicine, Yale University Medical School, 295 Congress Avenue, New Haven, CT 06536-0812, USA4 - Cell Biology, Yale University Medical School, 295 Congress Avenue, New Haven, CT 06536-0812, USA

Correspondence to Shulamit Michaeli: The Mina and Everard Goodman Faculty of Life Sciences, Bar-Ilan University,Ramat-Gan 5290002, Israel. shulamit.michaeli@biu.ac.ilhttp://dx.doi.org/10.1016/j.jmb.2017.04.015Edited by Lori A Passmore

Abstract

The parasite Trypanosoma brucei is the causative agent of African sleeping sickness and is known for itsunique RNA processing mechanisms that are common to all the kinetoplastidea including Leishmania andTrypanosoma cruzi. Trypanosomes possess two canonical RNA poly (A) polymerases (PAPs) termed PAP1and PAP2. PAP1 is encoded by one of the only two genes harboring cis-spliced introns in this organism, andits function is currently unknown. In trypanosomes, all mRNAs, and non-coding RNAs such as small nucleolarRNAs (snoRNAs) and long non-coding RNAs (lncRNAs), undergo trans-splicing and polyadenylation. Here,we show that the function of PAP1, which is located in the nucleus, is to polyadenylate non-coding RNAs,which undergo trans-splicing and polyadenylation. Major substrates of PAP1 are the snoRNAs and lncRNAs.Under the silencing of either PAP1 or PAP2, the level of snoRNAs is reduced. The dual polyadenylation ofsnoRNA intermediates is carried out by both PAP2 and PAP1 and requires the factors essential for thepolyadenylation of mRNAs. The dual polyadenylation of the precursor snoRNAs by PAPs may function torecruit the machinery essential for snoRNA processing.

© 2017 Elsevier Ltd. All rights reserved.

Introduction

In trypanosomes, all mRNAs are trans-splicedand polyadenylated. In trans-splicing, a small exon,the spliced leader (SL), derived from the small SLRNA, is added to all mRNAs. Trans-splicing andpolyadenylation are linked, and ~140 nt separatethe poly (A) addition site of the upstreamgene from thetrans-splicing addition site of the downstream gene(3′ AG splice site). These coupled processes functionto dissect the polycistronic mRNAs to monocistronicmRNAs [1,2]. Mutations in the polypyrimidine tractaffect the polyadenylation of the upstream gene [3].Recent study revealed that the knockdown of the

er Ltd. All rights reserved.

canonical poly (A) polymerase (PAP) affects the 3′end formation of mRNAs and trans-splicing [4]. Usingtandem-affinity purification with tagged cleavage andpolyadenylation specificity factor (CPSF) 160, apolyadenylation factor [5], and mass spectrometryrevealed 10 associated components of the trypano-some polyadenylation machinery. It included homo-logs to all CPSF subunits, cleavage-stimulating factor(CstF) 50/64, and symplekin present in mammals [5],as well as two hypothetical proteins whose function iscurrently unknown [4]. RNA interference (RNAi)-mediated knockdown revealed that most of thesefactors are essential for growth and required for bothpolyadenylation and trans-splicing, strongly supporting

J Mol Biol (2017) 429, 3301–3318

3302 PAP PAP1 Polyadenylates Non-Coding RNAs

the coupling of these two processes [4]. However, thefactors that coordinate the coupling betweentrans-splicing and polyadenylation are unknown. Aprotein from the splicing machinery U1A but no otherU1 small nuclear ribonucleoprotein particle(snRNP)-associated proteins are present in complexespurified with the Leishmania polyadenylation factorCPSF73. U1Awas shown to affect the polyadenylationof mRNAs, suggesting that this splicing factor is agenuine component of the polyadenylation machineryin these parasites [6].Trypanosomes possess large repertoire of small

nucleolarRNAs (snoRNAs).Our recent studies predictthe presence of 83 H/ACA snoRNAs [7], which guidepseudouridylation and 79 C/D snoRNAs, whichguide 2′-O-methylations [8]. Studies in Leishmaniaidentified 81 H/ACA and 80 C/D snoRNAs [9].Genome-wide mapping of pseudouridines in the twolife stages of Trypanosoma brucei indicated that thismodification, which is elevated in specific positionsin ribosomal RNA (rRNA), may assist in coping withthe temperature shift while cycling between the twohosts [7].In eukaryotes, snoRNAs exhibit very diversemodes

of genomic organization. In animals, most snoRNAsare located within introns of host genes that usuallyencode proteins related to ribosome biogenesis [10].In yeast, the majority of snoRNAs are encoded bymonocistronic genes transcribed from their ownpolymerase II promoter [11]. In plants, most snoRNAsare found in genomic clusters, which are independentand transcribed from their own promoter or locatedin introns of protein-coding genes [12]. snoRNAprocessing is carried out by different mechanisms. Inanimals, the major pathway for processing intronicsnoRNAs depends on intron debranching followed byexonucleolytic trimming of the 5′ and 3′ free ends [13].Processing of polycistronic pre-snoRNA has beenwell described in yeast [14]. A key factor in thisprocess is Rnt1p, an RNase III endonuclease thatcleaves the pre-snoRNA and liberates the individualsnoRNAs with 5′ and 3′ extensions. These extensionsareeliminatedbyRat1porXrn1p5′ → 3′exonucleasesand the nuclear exosome, which has 3′ → 5′ exonu-clease activity [15]. The mature snoRNA ends areprotected by the assembly of snoRNP core proteins[16]. In plants, most snoRNAs are processed frompolycistronic precursors and this implicates an endo-nucleolytic cut to release the snoRNAs [17]. Never-theless, the endonuclease responsible for thiscleavage has not been identified. AtRTL2, the closesthomolog of Rnt1p in Arabidopsis, could be implicated,but no effect on pre-snoRNA accumulation wasdetected in atrtl2 mutants [18].In trypanosomes, the snoRNAs are organized in

clusters, which carry interspersed H/ACA and C/Dgenes [8,9,19,20]. Most of the clusters are tran-scribed polycistronically by RNA polymerase II, butsince conventional pol II promoters do not exist

upstream to protein coding genes, no promoterwas reported upstream to snoRNA clusters [21].However, studies in Leptomonas collosma, a mono-genetic trypanosomatid, demonstrated that 700 bppresent upstream to a snoRNA cluster enhanced theexpression of a tagged snoRNAgene in an orientation-dependent manner [22]. The trypanosome pre-snoRNA encodes for several snoRNAs ranging from1 to9.Among the50snoRNAgenes, 21genesencodefor a single snoRNA,whichwe termedsolitary snoRNA[8]. All snoRNAs either solitary or clustered areliberated from transcripts that are trans-spliced andpolyadenylated [2]. The majority of the clusters arealso repeated in the genome several times. The trans-splicing and polyadenylation stabilize the pre-snoRNAbefore these are processed to liberate the individualRNAs [8,23]. Similar to plants, the enzyme thatliberates the snoRNA from the precursor is currentlyunknown. It is also unknown how the machinery thatpolyadenylates mRNA differs from the one thatpolyadenylates snoRNA transcripts, if at all.All eukaryotic genomes code for one or several

canonical PAPswith similar, catalytic, central andRNAbinding domains. Yeast has a single PAP [24], and thegenomes of higher eukaryotes have two or three PAPgenes. In mammals, two nuclear PAPs, PAP α [25,26]and PAP γ [27–29], were identified. PAP β (TPAP) isinvolved in the polyadenylation of testis-specificmRNAs [30]. The heterogeneity in PAP functionswas recently reviewed [31].The other groups of PAPsare non-canonical PAPs, which include the GLD-2,TRF4/5, and CID1-type PAPs or poly (U) polymer-ases, and 2′-5′-oligo (A) synthetases. These enzymesshare the catalytic domain with canonical PAPs butcontain a different nucleotide base-recognition motif.These groups of polymerases are mainly involvedin RNA turnover and quality control acting on rRNA,tRNAs, snoRNAs, and cryptic RNA polymerase IItranscripts [32–35].The canonical PAP adds long poly (A) tails to the 3′

end of the mRNAs, and in contrast, the Trfp proteinswere shown to add short poly (A) tails to their substrateRNAs,whichare assumed to trigger the efficient decayof theRNAsby the recruitment of the nuclear exosomecomplex [32–34]. Moreover, Trf4p-mediated polyade-nylation is involved in the degradation of crypticunstable transcripts generated by RNA polymerase II[34]. In Drosophila, two ncPAPs TRF4–1 and 2 wereidentified; the first was shown to polyadenylate smallnuclear RNAs (snRNAs) [36]. However, more recentstudies identified this protein in the cytoplasm anddemonstrated its function in mRNA degradation [37].Oligoadenylation of snoRNA precursors and its role insnoRNAprocessingwere also reported in human cells[38].The non-canonical PAPs are working in a complex

knownas theTrf4/Air2/Mtr4p polyadenylation complex(TRAMP) complex, which consists of either one of thetwo non-canonical PAPs (ncPAPs), Trf4p and Trf5p,

3303PAP PAP1 Polyadenylates Non-Coding RNAs

formingacomplexwith theRNA-binding proteinsAir1por Air2p and the RNA helicase Mtr4p [32–34]. TheTRAMP complex has a dual function of poly (A)-tailaddition and recruitment of the exosome complex [39].However, the TRAMPcomplex iswell established onlyfor the two yeast enzymes mentioned above. MostncPAPs do not seem to reside in this type of complex.However, evidence exists for the existence ofTRAMP-like complex in human, and the enzymeassociated with the complex is known as PAPD5 [40].In trypanosomes, two nuclear non-canonical PAPs,

TbncPAP1 and TbncPAP2 [41], and the putativeTbMTR4 [42] were identified and characterized. TheTbncPAP1 is a homolog of yeast TRF4 [41]. Proteo-mic analysis of the TbncPAP1 complex revealed itsassociation with the trypanosome MTR4 and putativeAir1p ortholog [41]. TbMTR4 localizes to the nucleusand is required for normal 5.8S rRNA maturation.Depletion of TbMtr4 causes an increase in rRNApolyadenylation [42].In yeast, snoRNA processing requires the non-

canonical PAP Trf4p, which recruits the exosome[15,43–45]. Some studies demonstrated the role ofnuclear poly (A)-binding protein 2 in the processingof snoRNAs by promoting poly (A) tail trimming frompre-snoRNAs by recruiting the nuclear exosome[46]. Most recently, the nuclear poly (A)-bindingprotein PABPN1 has been implicated in the decay ofnon-coding RNAs in mammals. PABPN1 promoteshyperadenylation by stimulating PAP (PAPα/γ), andthis hyperadenylation is needed for the degradationof transcripts [47].In this study, the role of the two canonical PAPs

PAP1 and PAP2 was investigated to elucidatetheir role in snoRNA processing. Small RNA librariesfrom PAP1- and PAP2-silenced cells indicatedmajor effect on the steady-state level of snoRNAs.Interestingly, the transcriptome analysis by RNA-seq(RNAsequencing) ofPAP1-silenced cells revealed nochanges in the level ofmRNAs but detected effects onnon-coding RNAs; the accumulation of pre-snoRNAsuggests that the recruitment of PAP1 is essential forthe processing of snoRNAs from their cognateprecursors.

Results

T. brucei PAPs PAP1 and PAP2

Trypanosome PAP1 (Tb927.3.3160) was identi-fied via homology to other eukaryotic PAPs [48]. Thisis one of the two genes to undergo cis-splicing in T.brucei. The second canonical polymerase is PAP2(Tb927.7.3780). It was recently shown that PAP2 isinvolved in the polyadenylation of mRNAs in T. brucei[4]. To examine if PAP1 belongs to canonical PAPs,we performed CLUSTALW alignment between ca-

nonical PAP1 and PAP2. PAP1 and PAP2 alignmentscores show 45% similarity (and 28% identity for 47%coverage, with e-value b 6*10−31).The sequence alignment (Fig. 1a) and the three-

dimensional (3D) predicted structure of the two PAPs(Fig. 1b) showed high resemblance between thesetwo enzymes. Both enzymes possess the domainsthat specify canonical PAP: a catalytic domain, acentral domain, nucleotide base-recognition motif1, RNA binding domain, and C-terminal domain. Thetwo PAPs differ mostly in the size of the C-terminal domain, as PAP2 carries a longer C-terminaldomain.The PAP1 is conserved when compared to other

homologs; it shares 41% similarity with the bovinePAP protein (coverage of 99%), and 43% similaritywith the human PAP protein (coverage of 91%).The PAP2 is conserved when compared to otherhomologs; it shares 50% similarity with the bovinePAP protein (coverage of 73%), and 50% similaritywith the human PAP protein (coverage of 64%).When performing a multiple sequence alignmentfor the PAP1 and PAP2 catalytic domain, a highconservation was observed, that is, the conservationof the active site residues between PAP1 andPAP2 is 49% similarity (coverage of 68%; Supple-mentary Fig. S1). In addition, the 3D structures of T.brucei PAP1 and PAP2 proteins' were comparedto all Protein Data Bank using the DALI server[49]. 3D structural conservation was observedfor PAP proteins homologs in bovine, human,yeast, and Caenorhabditis elegans (z-score N 17,rmsd b 3.4).

To examine the role of both PAPs in trypanosomeRNA polyadenylation, we silenced the expression ofthe genes by RNAi using stem-loop constructs [50].Northern analysis showed ~90% reduction in mRNAupon silencing (Supplementary Fig. S2A and B), andgrowth curves of the PAP1- and PAP2-silenced cellsdemonstrated reduction in cell growth (Fig. 2a-i).However,PAP2-silenced cells showedmuch strongergrowth inhibition than PAP1. Next, independentclones were isolated upon transfection, and growthwas monitored continuously. As observed (Fig. 2a-ii),the growth of PAP1-silenced cells was compromisedcompared to uninduced cells but is not as robust asthe effect of PAP2 silencing on growth. In thegenome-wide screen for essential genes in procyclicand bloodstream forms of trypanosomes [51], nosevere phenotype observed for either PAP2 or PAP1,suggesting that the RNA interference (RNAi) targetsequencing (RIT-seq) does not always agree withsilencing of individual genes. A recent study moni-tored the effect of PAP1 silencing on T. bruceigrowth but could not detect growth perturbation [4].Here, we used the stem-loop silencing method,which is stable and selected clones after transfectionand did observe mild growth perturbation as aresult of PAP1 silencing. In this study, two different

Fig. 1. a) CLUSTALW alignment of PAP2 with PAP1. Different domains are shown in different colors. (b) 3D predictedstructure. The model was prepared using the SWISS-Model. Color code for domains: blue, central domain (CD); red,nucleotide base-recognition motif type 1 (NRM); green, RNA binding domain (RBD); and yellow, C-terminal amino acids[70–73].

3304 PAP PAP1 Polyadenylates Non-Coding RNAs

3305PAP PAP1 Polyadenylates Non-Coding RNAs

ways (culture dilution and continuous growth) wereused to monitor growth, and in both cases, PAPssilencing affected the growth of the parasites, but theeffect of PAP2 silencing on growth was more severeand profound.Next, the localization of the proteins was deter-

mined. To this end, the genes were in situ taggedwith the PTP (for ProtC-Tobacco Etch Virus (TEV)-ProtA) construct [52]. The immunofluorescenceresults demonstrated that both PAP1 and PAP2enzymes are confined to the nucleus (Fig. 2b-i). Aprevious study suggested cytoplasmic localization forPAP1 [4]. To further evaluate the protein localization,we tagged the protein at the N terminus using mNeo-green GFP tagging based on Ref. [53]. The datapresented in (Fig. 2b-ii) support the nuclear localizationof this protein. Further support for the localization wasobtained fromadatabase that was generated based onRef. [53]. PAP1 was tagged among the 2000 proteinsexamined in this study, and the results presented in thesite show that the protein is indeed localized to thenucleus‡. The nuclear localization observed in thisstudy is in agreement with the function we observed forthe protein (see below).Since it was already established that PAP2 is the

robust polyadenylation enzyme for mRNAs, wefocused on PAP1 in this study. We first examinedthe effect of PAP1on the level ofmRNAby performingnorthern analysis. The results (Fig. 2c) demonstrateda clear effect by PAP2 but not by PAP1 silencing onthe level of selected mRNAs. Since certain PAPsespecially in humans and plants seem to affect only asubset of mRNAs [31], we performed a genome-wideanalysis by RNA-seq upon the silencing of PAP1.Libraries were prepared as described [2] by randompriming and amplification with SL forward primer (alsotermed 5′ directed library). The normalization proce-dure is based on the premise that most genes are notdifferentially expressed. If the condition had a uniformeffect across all mRNAs, this would be impossible todetect. Thus, we calibrated the amount of RNA in thetwo libraries based on rRNA. There was ~1.3-foldmore rRNA in the control compared with the silencedcells, which was very similar to the difference inlibrary size (~1.2-fold) between the uninduced andtetracycline-induced PAP1-silenced libraries. Theresults presented in Supplementary Table S2 showedno significant effect on mRNAs. However, whenwe found (see below) that PAP1 silencing affectsthe mature snoRNA level, we questioned whetherrRNA could be used to normalize these libraries,as snoRNAs are involved in rRNA processing. Wetherefore spiked the next set of poly (A)-selectedlibraries with bloodstream form (BSF) RNA (seeMaterials and Methods). These libraries were pre-pared after oligo (dT) selection. The amount of readsderived from the BSF spike in the control and thelibrary from the silenced cells was almost the sameand correlated with the size of the library. Despite

changes in the type of libraries [5′ directed SL libraryor poly (A)-selected library], and the RNA used forcalibration, both libraries were highly enriched formature mRNA as seen in Supplementary Fig. S3Aand the bar diagram in Supplementary Fig. S3B).The results on the global effect on the transcriptomewere remarkably similar, r2 is 0.813 (p-value b0.001)for comparing the reads per kilobase of transcriptper million mapped reads (RPKM) in the silencedcells of the two different libraries (Supplementary Fig.S3C). No mRNAs were significantly reduced, and only93 (1%) of the 8487 mRNAs were up-regulated by1.5-fold in the two libraries (Supplementary Table S2).Moreover, 47 of these 93 genes encode for hypothet-ical proteins, 18 ofwhich are annotated as “unlikely”; 11genes are described as “unspecified product” inGeneDB, and 30 as putative. Thus, only five definitivecoding genes (e.g., RBP33) showed an effect underPAP1 silencing. This result prompted us to investigatethe role of PAP1 in the processing of other RNAspecies.Interestingly, snoRNAs precursors were elevated

upon PAP1 silencing. Overall, 9 out of 17 snoRNAclusters were found to be accumulated by 1.5-fold inboth the libraries, and 7 of these 17 pre-snoRNARNAs were elevated at least 2-fold in two libraries.Surprisingly, 63 of the 103 long non-coding RNAs(lncRNAs) identified [2] were elevated at least 1.5-foldupon PAP1 silencing, and no lncRNAs were down-regulated (Supplementary Table S2). Statisticalanalysis using Kernel density plots illustrated thatonly the lncRNAs and pre-snoRNAs are differentiallyaffected in comparison to mRNA (SupplementaryFig. S3D). This asymmetrical distribution to the rightside shows excess of lncRNAs and pre-snoRNAswith increased expression upon PAP1 silencing(Supplementary Fig. S3D). Note that these librarieswill not capture small RNA including maturesnoRNAs.We further validated the results by examining the

level of pre-snoRNAs by RT-PCR. The scheme of thepre-snoRNAs examined is given. The results in(Fig. 3a-i) showed that under PAP1 silencing, thelevel of pre-snoRNAs was elevated, whereas underPAP2 silencing, the level of pre-snoRNAs wasreduced. The expression level is based on thequantification of the amount of PCR product. The pre-snoRNA elevation can also be visualized from theRNA-seqdata (Fig. 3a-ii) prepared fromPAP1-silencedcells. However, in PAP2-silenced cells, these snoRNAprecursors can hardly be detected. The northernanalysis andRNA-seq results (Fig. 3b-i and -ii) showedthat lncRNAs were also elevated as a result of PAP1silencing. Thus, themajor effect seen as result ofPAP1silencing was the up-regulation of snoRNA precursorsand lncRNAs.The data therefore suggested that PAP1 is involved

in the polyadenylation of certain non-coding RNAs.The only way to explain this observation is that PAP1

3306 PAP PAP1 Polyadenylates Non-Coding RNAs

is not the only polymerase that polyadenylates thesetranscripts. Both PAPs can polyadenylate these non-coding RNA transcripts. These results also raise the

Fig. 2 (legend o

possibility that polyadenylation by PAP1 helps recruitthe processing machinery; thus, in the absence ofPAP1, snoRNA processing does not take place,

n next page)

3307PAP PAP1 Polyadenylates Non-Coding RNAs

leading to the accumulation of the pre-snoRNAprecursors (see below).

Effect of PAP1 and PAP2 silencing on the levelof snoRNAs

The effect of PAP1 silencing on the level onpre-snoRNA suggested that polyadenylation byPAP1 might be essential for snoRNA processing.To be able to directly observe change in the level ofmature snoRNAs, we determined the repertoire ofsmall non-coding RNAs in cells before and afterPAP1 or PAP2 silencing. We recently providedevidence that post-ribosomal supernatants (PRSs)are enriched with mature snoRNAs [7], and there-fore, small RNA libraries derived from PRS wereprepared from cells silenced for PAP1 or PAP2.Small RNA libraries were prepared as previouslydescribed [8]. The pie diagram (Fig. 4a) presentschanges in the repertoire of enriched small RNAs asa result of PAPs silencing. Major reduction in thelevel of mature snoRNAs was observed under thedepletion of either PAP1 or PAP2 (SupplementaryTable S3). Note that upon silencing, the relative levelof tRNA was changed, but this change mostprobably reflects the changes in the repertoire ofsmall RNAs present in the PRS post-silencing (seebelow).To verify the results obtained in the RNA-seq data

(Fig. 4a), we examined the small RNA level by eitherprimer extension or northern analysis. The amount ofRNA before and after silencing was normalizedbased on the level of 12S mitochondrial rRNA, whichshould not be affected in these cells. The level oftRNA (tRNAsec and tRNAser) was examined bynorthern analysis, and the results (Fig. 4b) indicatedno change in the level of the tRNAs tested. Thus, thechanges observed in the libraries as seen in the piediagram (Fig. 4a) does not reflect the effect on thelevel of tRNA. Since mature snoRNA level wasreduced, it enabled the enrichment of tRNAs in thelibrary. As no effect was observed for the level oftRNAs, their level could be used as a marker for the

Fig. 2. (A) Silencing of PAP2 and PAP1 affects the growth oprocyclic cells was compared with cells induced for RNAi aftuninduced and induced cultures were diluted daily to 1 × 105monitored before and after tetracycline induction ofPAP1 or PAPplotted in blue and those of induced cultures in red. Data aretriplicate (n = 3). (B) PAP1 and PA2 are localized in the nucleusProtC-TEV-ProtA)-tagged PAP1 and PAP2 proteins were fixed wantibodies as described in Materials and Methods section, andFluorescence of the tagged protein; (b) nuclei stained with DAPcontrast (DIC) merge with panel. (ii) Procyclic cells stably expreswith DAPI. The PAP1-tagged protein localized in nucleus. (a) Flu(c) Merge of panels (a and b); (d) DIC merge with all panels. (C)Procyclic cells carrying the silencing constructs for PAP1 and PAwas extracted from the cells and (20 μg) and was separated onblotted and hybridized with a randomly labeled probe specific fotwice (biological replicates). The level of 7SL RNA (used as a c

RNA present in the samples (loading control). Next,the effect of PAPs silencing on U snRNAs level wasexamined by northern analysis. The results (Fig. 4c)showed no significant effect on the level of U2, U3 U4,and U6 snRNAs, suggesting that these snRNAs arenot regulated by polyadenylation via PAP1 or PAP2 intrypanosomes.After finding that PAP silencing does not affect

tRNA and U snRNA (Fig.4b and c), we verified thechanges observed in snoRNAs using tRNA ascontrol for the amount of RNA in the samples. Asignificant reduction in the level of snoRNAs in bothPAP1- and PAP2-silenced cells was observed byprimer extension (Fig.4d), and indeed, this resultwas also observed when examining the RNA-seqdata (Supplementary Table S3). Furthermore, 73snoRNAs were reduced by at least 1.5-fold in 2 ofthe 3 biological replicates of the libraries; 52 of the 73mature snoRNAs were reduced by 2-fold or greater.We noticed a differential level of reduction in maturesnoRNA even if these originate from the samecluster, suggesting that the reduction depends onthe individual properties of the snoRNA gene(flanking sequences and sequence composition).However, in all cases, we could observe markedreduction in the level of mature snoRNAs in thesilencing of both PAPs. To verify the primer extensiondata and to assure that the reduction in the levelreflects only the level of the mature snoRNA, we usednorthern analysis tomonitor the changes in the level ofthe snoRNA (Supplementary Fig. S4A). The resultsclearly support the reduction in the level of maturesnoRNAs upon PAP1 or PAP2 silencing.

SnoRNA polyadenylation requires thepolyadenylation machinery

In trypanosomes, all mRNAs are trans-spliced. Intrans-splicing, an exon of 39 nt (SL) is added to allmRNAs via a mechanism related to cis-splicing [3].The SL is derived from a small RNA, the SL RNA.During the trans-splicing reaction, a Y-structureintermediate is formed, which is analogous to the

f Trypanosoma brucei. (i) Growth of uninduced cells (−Tet)er tetracycline addition (+Tet) for the time indicated. Bothcells per ml. (ii) Continuous cumulative cell growth was

2 silencing. The numbers of cells in the uninduced culture arerepresented as mean ± s.e.m. Experiments were done in. (i) Transgenic procyclic T. brucei cells expressing PTP (forith 8% paraformaldehyde for 20 min, incubated with anti-IgGdetected by a Alexa488-conjugated secondary antibody. (a)I. (c) Merge of panels (a and b); (d) Differential interferencesing mNeogreen-GFP tagged PAP1 protein were co-stainedorescence of the tagged protein; (b) nuclei stained with DAPI.Silencing of PAP2 but not PAP1 affects the level of mRNAs.P2 uninduced (0 day) or silenced for the time indicated. RNA1.2% agarose and 2.2 M formaldehyde gel. The RNA wasr the gene. The northern blotting experiment was repeatedontrol for equal loading) was determined.

3308 PAP PAP1 Polyadenylates Non-Coding RNAs

lariat in cis-splicing [3]. Inhibition in the first step oftrans-splicing results in the accumulation of SL RNAand the reduction of the Y-structure intermediate.

To examine if snoRNA polyadenylation requires themachinery involved in polyadenylation of mRNAs, wesilenced several factors involved in the process byRNAi, includingCPSF160 (Tb927.11.14560),CPSF73(Tb927.4.1340), CPSF100 (Tb927.11.230), andCstF50 (Tb927.6.1830). Northern analysis showed

Fig. 3 (legend o

that all these factors were efficiently silenced (Supple-mentary Fig. S2C–F) and that all were essential forgrowth (Fig. 5a). Since polyadenylation is linked totrans-splicing, defects in polyadenylation should affecttrans-splicing [3]. Primer extension was performed onRNA from uninduced and silenced cells using a primercomplementary to the 3′ end of SL RNA. The primerextension detects the mature SL RNA and theY-structure intermediate (see illustration in Fig. 5b).

n next page)

3309PAP PAP1 Polyadenylates Non-Coding RNAs

Note that multiple extension products are due to thepresence of hypermodified cap at the 5′ end of the SLRNA [54]. The results demonstrate that as a result ofsilencing of the polyadenylation factors, a clear trans-splicing defect at the first step of splicingwas observed,leading to the accumulation of the SL RNA (Fig. 5b-i)and a marked reduction in the Y-structure intermediate(Fig. 5b-ii). Most relevant to this study is that as a resultof silencing of the polyadenylation factors, the level ofmature snoRNAs was reduced (Fig. 5c-i and -ii). Itsuggests that snoRNA polyadenylation requires thesamemachinery that conductsmRNApolyadenylation.However, it is not clear at this point if only PAP2interacts with the polyadenylation machinery, aspreviously demonstrated [4], or if also PAP1 directlyassociates with these factors. U3 snoRNA level wasunaffected in these experiments (Fig. 5).The results (Fig. 5) demonstrate a specific effect

only on mature snoRNAs, which are processed fromtrans-spliced and polyadenylated transcripts, since noeffect was found on U3 snoRNA that is transcribed bypolymerase III [55].

snoRNA processing requires active MTR4 andnot the non-canonical poly (A) polymerase TRF4present in the TRAMP complex

We know very little how the trypanosome snoRNAsare processed from the polycistronic transcripts togenerate mature snoRNA [8,23]. However, in othereukaryotes, snoRNA processing involves endonucleo-lytic cleavage in the intergenic region, which is followedby the action of 5′ exonuclease(s) and the exosome,leading to release of mature snoRNAs from theircognate precursors [23]. In higher eukaryotes, pro-cessing is mediated by splicing followed by debranch-ing of the lariat, which is the substrate for the exosomeand often also requires a 5′ exonuclease for processing[38]. In yeast, the MTR4 helicase is part of the TRAMPcomplex, which functions in snoRNAprocessing [43]. It

Fig. 3. (A) The effect of PAP1 and PAP2 silencing on thsnoRNAs. The scheme depicting the SL addition site (yellow)(blue), and position of the primers used for the RT-PCR expinduction (−Tet) and after 2.5 days of silencing using randomprimer from the snoRNA coding sequence and SL forward primon 12S mitochondrial RNA. The precursor snoRNA expresnormalized to 12S rRNA. Data are presented as mean ± s.et-test); *P b 0.05; **P b 0.01. Left panel: Precursor snoRNA TBThe read distribution of pre-snoRNA based on RNA-seq upoalong two snoRNA coding sequences and the intergenic rePAP1-and PAP2 silenced cells. Shown is the distributionPAP2-silenced cells (red) and control (blue). Left panel: PrecurTB6Cs2H1. (B).The effect of PAP1 silencing on lncRNAs. (i)uninduced procyclic cells (− Tet) and cells carrying silencinginduction (+Tet). The RNA was separated on 1.2% agarosehybridized with a randomly labeled probe specific for the gene.was determined. (ii) The read distribution of lncRNA based on Rof the lncRNA is presented based on the poly (A)-selected librPAP1-silenced cells (red) and control (blue). TB10.NT.68 and

was therefore of interest to examine if MTR4 is requiredfor snoRNA biogenesis in trypanosomes. MTR4functions together with the non-canonical PAP,TbTRF4, present in the TRAMP complex [41], andthus, the role of these two factors in snoRNAprocessing was investigated. After observing theefficient silencing of MTR4 and TbTRF4 mRNAs(Supplementary Fig. S2G and H) and the growth arrestas a result of silencing MTR4 (Fig. 6a-i) and TbTRF4(Fig. 6a-ii), the effect on mature snoRNA levels wasexamined by primer extension. The effect on the levelof mature snoRNA under MTR4 and TRF4 silencingwas also verified by northern analysis (SupplementaryFig. S4B). The results indicated that MTR4 is essentialfor snoRNA biogenesis (Fig. 6b-i), but not TRF4(Fig. 6b-ii) unlike in other eukaryotes [43].

A model for involvement of dual polyadenylationin snoRNA processing

Based on the finding that PAP1 silencing resultsin the accumulation of pre-snoRNA transcripts(Fig. 3), we propose that the dual polyadenylationis needed to mark these transcripts for furtherprocessing. We still do not know how PAP1 recog-nizes these non-coding transcripts. This recognitionmay require additional RNA-binding protein (s). Ourfinding is summarized in a model illustrated in Fig. 7.We raise the intriguing possibility that RNA-bindingprotein(s), which binds to the 3′ flank, may functionto recruit the exosome/processing machinery (seeDiscussion), and we are currently searching for suchprotein(s).

Discussion

The presence of two or more PAPs in a singleorganism is mostly typical of higher eukaryotes. Thefunction of PAP1 [48] was puzzling after PAP2 was

e level of precursor snoRNAs. (i) The level of precursor, mature snoRNA (green), intergenic precursor sequenceeriment. cDNA was prepared from procyclic cells beforehexamers. Pre-snoRNAs were amplified using reverse

er. The level of RNA in the samples was determined basedsion is relative to uninduced (−Tet) procyclic cells and.m. Experiments were done in triplicate (n = 3; Student's10Cs-7H1. Right panel: Precursor snoRNA TB6Cs2H1. (ii)n PAP1 and PAP2 silencing. The read distribution profilegions are presented based on the 5’ directed library foralong two snoRNA precursor genes for PAP1- and

sor snoRNA TB10Cs7-H1. Right panel: Precursor snoRNAThe level of lncRNAs. Total RNA (20 μg) prepared fromconstruct for PAP1 silencing after 2 and 3 days of RNAiand 2.2 M formaldehyde gel. The RNA was blotted andThe level of 7SL RNA (used as a control for equal loading)NA-seq upon PAP1 silencing. The read distribution profileary. Shown is the distribution along two lncRNA genes forTB11.NT.146 are the lncRNAs.

3310 PAP PAP1 Polyadenylates Non-Coding RNAs

shown to be the robust enzyme polyadenylatingmRNAs [4]. This study demonstrates that PAP1polyadenylates non-coding RNAs that undergo trans-splicing and polyadenylation. Specifically, PAP1 poly-adenylation is essential for snoRNA processing sinceunder its depletion, the level of snoRNA precursorswas elevated, whereas the level of mature snoRNAswas reduced. PAP1 also functions in the biogenesis oflncRNA, with yet to be identified possible biologicalrole in Trypanosomes.The reliance of snoRNA processing on PAP1

observed in this study is specific to small non-codingRNAs, which are transcribed by RNA polymerase II

Fig. 4 (legend o

and undergo trans-splicing. In trypanosomes, the UsnRNAs are transcribed by polymerase III, and theirgenes contain runs of Us as termination signals[55]. On the contrary U1, U2, U4, and U5 snRNAsare transcribed in all other eukaryotes by RNApolymerase II [56]. snRNAs were shown to undergopolyadenylation as part of their processing. Indeed,in Drosophila, TRF4–1 was shown to be involved insnRNA polyadenylation, and studies suggested thatDmFTRF4–1 and DmRrp6 are involved in thepolyadenylation-mediated degradation of snRNAs invivo [36]. As mentioned in the introduction, recentstudies demonstrate that the same enzyme is localized

n next page)

3311PAP PAP1 Polyadenylates Non-Coding RNAs

in the cytoplasm and functions in mRNA degradation[37]. Note that the study on the effect of DmFTRF4–1on snRNA was obtained by overexpressing theenzyme and not under normal conditions, questioningits validity in the nuclear processing of snRNAs. How-ever, in yeast, exosome mutants were shown toaccumulate 3′ extended polyadenylated forms of U4snRNA [43]. Trf4p and Trf5p were shown to poly-adenylate U4 snRNA, and this polyadenylation wasshown to be essential for recruiting the nuclearexosome via Rrp6p [45]. Thus, trypanosomes seemto be different in their snRNA processing, possiblybecause in trypanosomes, the snRNAs are transcribedby polymerase III and not polymerase II as in othereukaryotes [55].Our study suggests that snoRNA precursors under-

go dual polyadenylation and that PAP1 polyadenyla-tion is essential for snoRNA processing. In yeast,snoRNA transcripts were also shown to undergo dualpolyadenylation but by two different polymerases[i.e., Trf4 (non-conventional) from the TRAMP com-plex and the conventional PAP Pap1p]. In eukaryotes,termination of RNA polymerase II is coupled totranscript 3′ end formation. In yeast, the synthesis ofindependently transcribed snoRNAs involves thedefault polyadenylation of two classes of precursorsderived from termination at Nrd1/Nba3-dependentsites or a “fail-safe” mRNA-like signal. Based onthis study, it was suggested that the role of dualpolyadenylation in pre-snoRNA in yeast is to linktranscription termination and 3′ end processing[57]. However, the dual polyadenylation observedin this study seems to have a different biological role.In trypanosomes, not much is known about theregulation of termination of polymerase II except thepresence of tRNA genes at the end of the transcription

Fig. 4. (A) Pie diagrams describing the RNA content of tAnnotation of the reads obtained from RNA-seq of the small Rmolecules among the reads are summarized in the pie chart. Theach RNA class. (i) PRS library for cells before silencing and afbut for PAP2 silencing. (B)The level of tRNAs upon PAP1 anconstructs for PAP1 or PAP2 were silenced for 1, 2, and 3 da(+Tet) cells. Total RNA (10 μg) was separated on 10% denaturoligonucleotides complementary to the indicated tRNA. TheQuantification of the tRNA levels under PAP1 and PAP2 silenccells and normalized to the 12S rRNA. Data are presented as(C) The level of snRNA upon PAP1 and PAP2 silencing. (i) PPAP2 were silenced for 1, 2, and 3 days. RNA was extracted(10 μg) was separated on 10% denaturing gel and subjectecomplementary to the indicated snRNA. The level of 12S rRNsnRNAs level under PAP1 and PAP2 silencing. The exprenormalized to 12S rRNA. Data are presented as mean ± s.e.mof snoRNAs upon PAP1 and PAP2 silencing. (i) Total RNA waswithout induction (−Tet) and after 2.5 days of induction (+Tet).radiolabeled oligonucleotides complementary to snoRNAs andtRNA extension was used to demonstrate that equal amountsnoRNA levels under PAP1 and PAP2 silencing. Fold decreastRNA. Data are presented as mean ± s.e.m. Experiments w**P b 0.01.

units and the role of base J in transcription terminationin Leishmania [58]. The dual polyadenylation intrypanosomes and especially the joining of PAP1and its polyadenylation are essential for snoRNAprocessing. We hypothesized the existence of RNA-binding protein(s), which binds in the 3′ flank of thetranscript [upstream to poly (A) addition site] andmay recruit both PAP1 and the processing machinery(Fig. 7). Interestingly, RNA-binding proteins wereshown to affect poly (A) tail formation [59].The finding that in trypanosomes,MTR4 functions in

snoRNAprocessing, whereas TbTRF4 is not involvedin snoRNA polyadenylation, is intriguing. This isespecially interesting in light of the observation thatthe TRAMP complex enhances RNA degradation bythe nuclear exosome Rrp6 in yeast [60]. However,other mechanisms were also shown to function inrecruiting the exosome to process snoRNAs. In thefission yeast, Pab2, the nuclear poly (A)-bindingprotein, was shown to interact with the exosomeand to promote the processing of snoRNAs [46].More recently, it was demonstrated in mammaliancells that the human poly (A)-binding proteinPABPN1 functions in the decay of nuclearnon-coding RNAs. PABPN1 promotes hyperadenyla-tion by stimulating PAPα/γ. It was demonstratedthat hyperadenylation is required for the degradationof PABPN1 targets and is used to degrade mRNAwith reduced intron number or incomplete splicing.Interestingly, this decay can take place even ontranscripts that carry genetically encoded poly (A)[47]. It will be interesting to investigate if thetrypanosome nuclear PABPN also functions insnoRNA processing.Despite the fact that TbTRF4 is not involved in

snoRNA processing, it may have other essential

he small RNA libraries upon PAP1 and PAP2 silencing.Nome is presented. The percentages of the different RNAe percentage reported is the mean of the two replicates forter 2.5 days of silencing of PAP1; (ii) the same as that of (i)d PAP2 silencing. (i) Procyclic cells carrying the silencingys. RNA was extracted from uninduced (−Tet) or silenceding gel and subjected to northern analysis with radiolabeledlevel of 12S rRNA was used as a loading control. (ii)

ing. The expression levels are relative to uninduced (−Tet)mean ± s.e.m. Experiments were done in triplicate (n = 3).rocyclic cells carrying the silencing constructs for PAP1 orfrom uninduced (−Tet) or silenced (+Tet) cells. Total RNAd to northern analysis with radiolabeled oligonucleotidesA was used as a loading control. (ii) Quantification of thession levels are relative to uninduced (−Tet) cells and. Experiments were done in triplicate (n = 3). (D) The levelprepared from procyclic cells carrying the RNAi constructsTotal RNA (10 μg) was subjected to primer extension withtRNA. The products were separated on a denaturing gel.of RNA used in the experiment. (ii) Quantification of thees are relative to uninduced (−Tet) cells and normalized toere done in triplicate (n = 3; Student's t-test); *P b 0.05;

3312 PAP PAP1 Polyadenylates Non-Coding RNAs

functions in the cell, since its silencing affected growth.First, the TRAMP complex was shown in yeast toregulate the degradation of splice-out introns without

Fig. 5 (legend o

the need to have polyadenylation by Trf4p. It was alsodemonstrated that Trf4 functions in themaintenanceoftelomere length [60]. Future studies should reveal the

n next page)

3313PAP PAP1 Polyadenylates Non-Coding RNAs

RNA substrates or processes that are regulated byTRF4 in trypanosomes.In plants, three different PAPs exist. Each may

govern the polyadenylation of different subset ofgenes. PAPS1 was shown to polyadenylate mostlymRNA-encoding ribosomal proteins, cell divisionfactors, and major carbohydrate-metabolic proteins[61]. Interestingly, it was suggested the PAP genesexpanded from a single ancestral gene by a series ofduplication events and that individual members haveundergone functional specialization, including onethat functions in the cytoplasm and is expressed inpollen [62]. Indeed, PAP1 and PAP2 are highlysimilar except that PAP1 is shorter and carries a cis-spliced intron. It is still puzzling why this gene keptthe intron and why trypanosomes retained only twogenes with cis-spliced introns. It is interesting to notethat the T. brucei serine/arginine-rich proteins (SRproteins) are essential for PAP1 splicing and that SRproteins' function is known to be highly regulated byphosphorylation under different environmental cues[63].Although much progress has been achieved in the

description of the snoRNA repertoire and their function[7,8,64], little is known about how the snoRNAprecursors are dissected to release individual snoR-NAs in trypanosomes.Our attempt to identify RNaseIIIhomologs to Rnt1, which is involved in snoRNAprocessing in yeast, failed. Silencing of all RNase IIIhomologs present in the T. brucei genome failed todetect defects in snoRNAprocessing (our unpublisheddata). We cannot rule out the possibility that like inyeast, the 3′ end processing machinery can introducea cleavage in the precursor that is polyadenylationindependent [44].In summary, this study describes a novel role for

polyadenylation, which is essential for the processing

Fig. 5. (A) Silencing of polyadenylation factors CPSF160,procyclic trypanosome cell lines containing inducible RNAi cowere obtained after RNAi induction with tetracycline or unindudiluted daily to 5 × 104 cells per ml. The growth of uninddouble-stranded RNA production (+Tet). (B) Effect of the silencwas prepared from procyclic cells carrying the RNAi constru(+Tet). RNA (10 μg) was subjected to primer extension with raSL RNA. The products were separated on a 10% denaturing gand their structure is illustrated with the binding site of the primewas used to demonstrate that equal amount of RNA was used itrans-splicing defect. (i) Quantitation of the SL RNA accumulaand normalized to U3 snRNA. Data are presented as mean ± st-test); *P b 0.05; **P b 0.01. (ii) Quantitation of the “Y” structurenormalized to U3 snRNA. Data are presented as mean ± s.et-test); *P b 0.05; **P b 0.01. (C) Effect of silencing polyadenylprocyclic cells carrying the RNAi constructs for polyadenylainduction (+Tet). Total RNA (10 μg) was subjected to primer exto snoRNAs and U3 snoRNA. The products were separateddemonstrate that equal amount of RNA was used in the expsilencing of different polyadenylation machinery factors. Folnormalized to U3 snRNA. Data are presented as mean ± s.et-test); *P b 0.05; **P b 0.01.

of snoRNA. The better understanding of snoRNAs inthis organism is of outmost importance since these aredifferentially regulated in the two life stages of theparasite and are essential for coping with growth atelevated temperatures [7]. Thus, controlling thesnoRNA level and their processing may offer a newdrug target for fighting the devastating diseasescaused by these parasites.

Materials and Methods

Cell growth, construct preparation,and transfection

Procyclic T. brucei strain 29-13, which carriesintegrated genes for T7 polymerase and the tetra-cycline repressor, was grown in SDM-79 mediumsupplemented with 10% fetal calf serum in thepresence of 50 μg/ml hygromycin and 15 μg/mlG418. The silencing constructs were prepared aspreviously described [50] using oligonucleotides listedin Supplementary Table S1. Cells were transfected aspreviously described [65].

Northern analysis

Total RNA was prepared with TRIzol reagent(Sigma), and 20 μg/lane was fractionated on a 1.2%agarose and 2.2 M formaldehyde gel. The specificmRNA was detected using anti-sense RNA probes.For analyzing small RNAs such as tRNA and snRNA,total RNA (10 μg) was fractionated on a 10% poly-acrylamide gel containing 7 M urea. The RNA wastransferred to a nylon membrane (Hybond; AmershamBiosciences) and probed with anti-sense RNA probespecific to the gene.

CPSF100, CPSF73, and CstF-50. (i–iv) Growth curves ofnstructs for CPSF160, CPSF100, CPSF73, and CstF-50ced as control. Both induced and uninduced cultures wereuced cells (−Tet) was compared with cells induced foring of polyadenylation factors on trans-splicing. Total RNActs without induction (−Tet) or after 2.5 days of inductiondiolabeled oligonucleotides situated in the intron region toel. The SL RNA and Y-structure intermediate are marked,r. The cap4 positions are indicated. U3 snoRNA extensionn the experiment. SmD1 silencing was used as a control fortion. Fold increases are relative to uninduced (−Tet) cells.e.m. Experiments were done in triplicate (n = 3; Student's. Fold decreases are relative to uninduced (−Tet) cells and.m. Experiments were done in triplicate (n = 3; Student'sation factors on snoRNAs. (i) Total RNA was prepared fromtion factors without induction (−Tet) or after 2.5 days oftension with radiolabeled oligonucleotides complementaryon a denaturing gel. U3 snoRNA extension was used toeriment. (ii) Quantitation of the snoRNAs level under thed decreases are relative to uninduced (−Tet) cells and.m. Experiments were done in triplicate (n = 3; Student's

Fig. 6. (a) Silencing ofMTR4 and TRF4 affects the growth of the parasites. Growth curves of procyclic cell line containinginducible RNAi constructs for (i) MTR4 and (ii) TRF4 were obtained after RNAi induction with tetracycline or uninduced ascontrol. Both induced and uninduced procyclic cultures were diluted daily to 5 × 104 cells per ml. The growth of uninducedcells (−Tet) was compared with cells induced for double-stranded RNA production (+Tet). (b) Effect of the silencing ofMTR4and TbTRF4 on the level of snoRNA. Total RNAwas prepared fromprocyclic cells carrying theRNAi constructs forMTR4 andTRF4 factors without induction (−Tet) or after 2.5 days of induction (+Tet). RNA (10 μg) was subjected to primer extensionwith radiolabeled oligonucleotides complementary to snoRNAs and U3 snoRNA. The products were separated on adenaturing gel. U3 snoRNA extension was used to demonstrate that equal amount of RNA was used in the experiment.Quantitation of the snoRNAs level under the silencing ofMTR4 and TbTRF4 is shown. Data are presented as mean ± s.e.m.Experiments were done in triplicate (n = 3; Student's t-test); *P b 0.05. (i). MTR4, (ii). TRF4.

3314 PAP PAP1 Polyadenylates Non-Coding RNAs

Primer extension

RNA was prepared from T. brucei procyclic cellsusing TRI-Reagent (Sigma). Primer extension analysiswas performed as described [66,67] using 5′-end-labeled oligonucleotides specific to target RNAs. Theextension products were analyzed on 6% polyacryl-amide7 Mureagel and visualizedbyautoradiography.

RNA-seq of transcriptome

For the enrichment of the transcripts containingpoly (A) + RNA, total RNA was treated with RQ1RNase-free DNase I and subjected to two rounds ofpoly (A) + selection. First-strand cDNA synthesis wasinitiated with random hexadeoxynucleotide primers.After incubation with RNase H and Escherichia coli

Fig. 7. Model illustrating SnoRNA processing requires dual polyadenylation by two canonical poly (A) polymerases.Trypanosome snoRNAs are processed from pre-snoRNA transcripts, which are trans-spliced and polyadenylated. Theendonuclease that liberates snoRNAs from the cluster is unknown. The 5′ exonuclease and the exosomes likely operate togenerate the mature and processed snoRNA. All pre-snoRNA transcripts undergo dual polyadenylation by PAP2 and byPAP1. snoRNA polyadenylation requires the machinery that polyadenylates mRNA. Factors of the polyadenylationmachinery such as CPSF, CstF, and Symplekin are illustrated. It is currently unknown how PAP1 is recruited to conductpolyadenylation. We propose that RNA-binding proteins (s) might be involved in the process. The site for binding is likely tobe located in the 3′ flank situated downstream to the last snoRNA in the cluster and the poly (A) addition site. This proteinmay recruit PAP1 and the exosome/processing machinery. The data presented here indicate that the processing ofsnoRNA requires that the pre-snoRNA be dually polyadenylated by both PAPs.

3315PAP PAP1 Polyadenylates Non-Coding RNAs

DNA polymerase I, double-stranded cDNA wasfragmented with DNase I, and cDNA fragmentscorresponding in size to about 200 bp were size-selected on an agarose gel. The cDNA ends wererepaired, a single dA was added at the 3′ ends, andgenomic adapters (Illumina, Inc. All rights reserved)were added. Libraries were enriched by limited PCRand purified on an agarose gel. As an external control,in order to be able to detect a global effect on thelibraries, we spiked total Procyclic RNA with RNAextracted from BSF trypanosomes (we mixed 107

procyclic cells carrying the PAP1 silencing constructwith 104 BSF cells prior to RNA extraction).For the enrichment of transcripts containingSL, total

RNA was treated with Terminator 5′-monophospha-tedependent exonuclease, followed by DNase I, andfirst-strand cDNA synthesis was initiated with randomprimers. Second-strand cDNA synthesis was primedwith the SL Primer (5′-GCTATTATTAGAACAGTTTCTGTACTATATTG-3′) and platinum Pfx DNA poly-merase. cDNA was further processed as describedabove. The level of rRNA was used to normalize thechange in mRNA transcriptome.

Libraries of small RNAs

PRS was prepared as described [7]. The RNA (10–20 μg) was fragmented using the Ambion RNAfragmentation kit (AM8740). The RNA was dephos-phorylated at the 3′ end using T4 polynucleotide kinase(in the absence of ATP). The 5′ end of the RNA wasrepaired using polynucleotide kinase in the presence oftracer radioactive [γ32P] ATP. The material wasseparated on 15% polyacrylamide denaturing gel,and the radioactive bands at a size of ~25–40 ntwere excised from the gel. RNA was eluted, and the 3′adaptor was ligated (3′-RAppCTGTAGGCACCAT-CAAT/3′DDG) with T4 RNA ligase 2 (New EnglandBiolabs). The reaction was loaded on a 15% polyacryl-amide denaturing gel, the radioactive higher molecularweight bands (40–60 nt) were excised, and the 5′RNAadaptor (5′-ACACGACGCUCUUCCGAUCU-3′) wasligated using T4 RNA ligase. RNA was extracted, andcDNAwas synthesized in the presence of radiolabeleddCTP as tracer. The reaction was loaded on adenaturing 12% polyacrylamide gel and eluted. PCRamplification was performed with RT product using

3316 PAP PAP1 Polyadenylates Non-Coding RNAs

“Platinum” DNA polymerase (Invitrogen) and solexaprimers. The PCR products were sequenced by IonTorrent sequencing platform.

RNA-seq analysis

Reads were mapped to the T. brucei Genome(version 5)§ using smalt v0.7.5¶ with the defaultparameters, allowing non-unique reads to be mappedrandomly to their best match in the genome. For eachgene, raw read counts were obtained using Multicovfrom the Bedtools suite (v 2.17.0) [68]. RPKM wasutilized as the quantification method to obtain ameasure for the expression of each gene. Logbased-2 fold changes were computed from normalizedread counts for genes expressed above a minimumlevel (sum of RPKM N =10) across all the samples.Bedtools and custom scripts were used to performthese calculations. The expression levels obtained forall coding, lncRNA, and pre-snoRNA genes in the 5′directed and poly (A) libraries are reported in Supple-mentary Table S2. The 5′ directed library expressionlevels are presentedasRPKM,and thepoly (A) librariesas a function of the level of BSF-specific genes that hadbeen spiked in. The expression levels obtained for allsnoRNA genes in the PRS libraries are reported inSupplementaryTableS3. IntegrativeGenomicsViewer[69]was used for visualization.

Accession numbers

Sequencing data reported in this paper will bedeposited in the Sequence Read Archive (SRA) afterthe acceptance of the paper.Supplementary data to this article can be found

online at http://dx.doi.org/10.1016/j.jmb.2017.04.015.

Acknowledgments

The authors dedicate this study to the memoryof Elisabetta Ullu, an excellent and inspiring scientist,who initiated this collaborative study and was a leaderof the trypanosome RNA biology field and above all adear friend and the most generous colleague. Thestudy was funded by a grant from the Israel-USBinational Science Foundation (grant 2011254), by agrant from the I-core Program of the Planning andBudgeting Committee, by the Israel Science Founda-tion (grant 1796/12), and by David and Inez MyersChair in RNA silencing of diseases (to ShulamitMichaeli).

Received 2 November 2016;Received in revised form 14 April 2017;

Accepted 23 April 2017Available online 26 April 2017

Keywords:Trypanosomes;

poly (A) polymerase;canonical polyadenylation;

non-coding RNAs;snoRNA maturation

†V.C. and S.K.G. contributed equally to this work.‡http://tryptag.org/?query=Tb927.3.3160

§http://tritrypdb.org/common/downloads/release-5.0/TbruceiTREU927/

¶http://www.sanger.ac.uk/resources/software/smalt/

Abbreviations used:SL, spliced leader; PAP, poly (A) polymerase; CPSF,

cleavage and polyadenylation specificity; CstF, cleavage-stimulating factor; snoRNA, small nucleolar RNA; snRNA,small nuclear RNA; TRAMP, Trf4/Air2/Mtr4p polyadeny-lation complex; rRNA, ribosomal RNA; 3D, three-dimen-sional; BSF, bloodstream form; RPKM, reads per kilobaseof transcript per million mapped reads; lncRNA, long non-coding RNA; PRS, post-ribosomal supernatants; RNAi,RNA interference; ncPAPs, non-canonical PAPs; RNA-

seq, RNA sequencing; snRNP, small nuclear ribonucleo-protein particle.

References

[1] T.N. Siegel, D.R. Hekstra, X. Wang, S. Dewell, G.A. Cross,Genome-wide analysis of mRNA abundance in two life-cyclestages of Trypanosoma brucei and identification of splicing andpolyadenylation sites, Nucleic Acids Res. 38 (2010) 4946–4957.

[2] N.G. Kolev, J.B. Franklin, S. Carmi, H. Shi, S. Michaeli, C.Tschudi, The transcriptome of the human pathogen Trypa-nosoma brucei at single-nucleotide resolution, PLoS Pathog.6 (9) (2010), e1001090.

[3] S. Michaeli, Trans-splicing in trypanosomes: machinery andits impact on the parasite transcriptome, Future Microbiol 6(2011) 459–474.

[4] H. Koch, M. Raabe, H. Urlaub, A. Bindereif, C. Preusser, Thepolyadenylation complex of Trypanosoma brucei: character-ization of the functional poly(A) polymerase, RNA Biol. 13(2016) 221–231.

[5] D.C. Di Giammartino, J.L. Manley, New links between mRNApolyadenylation and diverse nuclear pathways, Mol. Cell 37(2014) 644–649.

[6] I.D. Tkacz, S.K. Gupta, V. Volkov, M. Romano, T. Haham, P.Tulinski, et al., Analysis of spliceosomal proteins in Trypa-nosomatids reveals novel functions in mRNA processing, J.Biol. Chem. 285 (2010) 27,982–27,999.

[7] V. Chikne, T. Doniger, K.S. Rajan, O. Bartok, D. Eliaz, S.Cohen-Chalamish, et al., A pseudouridylation switch in rRNAis implicated in ribosome function during the life cycle ofTrypanosoma brucei, Sci. Rep. 6 (2016) 25,296.

[8] S. Michaeli, T. Doniger, S.K. Gupta, O. Wurtzel, M. Romano,D. Visnovezky, et al., RNA-seq analysis of small RNPs inTrypanosoma brucei reveals a rich repertoire of non-codingRNAs, Nucleic Acids Res. 40 (2012) 1282–1298.

[9] D. Eliaz, T. Doniger, I.D. Tkacz, S.K. Gupta, N.G. Kolev, R.Unger, et al., Genome-wide analysis of small nucleolar RNAsof Leishmania major reveals a rich repertoire of RNAs

3317PAP PAP1 Polyadenylates Non-Coding RNAs

involved in modification and processing of rRNA, RNA Biol.12 (11) (2015) 1222–1255.

[10] J.P. Bachellerie, J. Cavaille, A. Huttenhofer, The expandingsnoRNA world, Biochimie 84 (2002) 775–790.

[11] A.K. Hopper, Cellular dynamics of small RNAs, Crit. Rev.Biochem. Mol. Biol. 41 (2006) 3–19.

[12] J. Rodor, I. Letelier, L. Holuigue, M. Echeverria, NucleolarRNPs: from genes to functional snoRNAs in plants, Biochem.Soc. Trans. 38 (2010) 672–676.

[13] T. Kiss, W. Filipowicz, Exonucleolytic processing of smallnucleolar RNAs from pre-mRNA introns, Genes Dev. 9 (1995)1411–1424.

[14] G. Chanfreau, G. Rotondo, P. Legrain, A. Jacquier, Processingof a dicistronic small nucleolar RNA precursor by the RNAendonuclease Rnt1, EMBO J. 17 (1998) 3726–3737.

[15] C. Allmang, J. Kufel, G. Chanfreau, P. Mitchell, E. Petfalski,D. Tollervey, Functions of the exosome in rRNA, snoRNAand snRNA synthesis, EMBO J. 18 (1999) 5399–5410.

[16] E. Caffarelli, A. Fatica, S. Prislei, E. De Gregorio, P.Fragapane, I. Bozzoni, Processing of the intron-encodedU16 and U18 snoRNAs: the conserved C and D boxescontrol both the processing reaction and the stability of themature snoRNA, EMBO J. 15 (1996) 1121–1131.

[17] D.J. Leader, G.P. Clark, J. Watters, A.F. Beven, P.J. Shaw,J.W. Brown, Splicing-independent processing of plant box C/D and box H/ACA small nucleolar RNAs, Plant Mol. Biol. 39(1999) 1091–1100.

[18] P. Comella, F. Pontvianne, S. Lahmy, F. Vignols, N. Barbezier,A. Debures, et al., Characterization of a ribonuclease III-likeprotein required for cleavage of the pre-rRNA in the 3′ETS inArabidopsis, Nucleic Acids Res. 36 (2008) 1163–1175.

[19] X.H. Liang, S. Uliel, A. Hury, S. Barth, T. Doniger, R. Unger,et al., A genome-wide analysis of C/D and H/ACA-like smallnucleolar RNAs inTrypanosomabrucei reveals a trypanosome-specific pattern of rRNAmodification, RNA 11 (2005) 619–645.

[20] X.H. Liang, A. Hury, E. Hoze, S. Uliel, I. Myslyuk, A. Apatoff,et al., Genome-wide analysis of C/D and H/ACA-like smallnucleolar RNAs in Leishmania major indicates conservationamong trypanosomatids in the repertoire and in their rRNAtargets, Eukaryot. Cell 6 (2007) 361–377.

[21] A.C. Ivens, C.S. Peacock, E.A. Worthey, L. Murphy, G.Aggarwal, M. Berriman, et al., The genome of the kinetoplastidparasite, Leishmania major, Science 309 (2005) 436–442.

[22] X.H. Liang, A. Ochaion, Y.X. Xu, Q. Liu, S. Michaeli, Smallnucleolar RNA clusters in trypanosomatid Leptomonascollosoma. Genome organization, expression studies, andthe potential role of sequences present upstream from thefirst repeated cluster, J. Biol. Chem. 279 (2004) 5100–5109.

[23] S. Uliel, X.H. Liang, R. Unger, S. Michaeli, Small nucleolarRNAs that guide modification in trypanosomatids: repertoire,targets, genome organisation, and unique functions, Int. J.Parasitol. 34 (2004) 445–454.

[24] J. Lingner, J. Kellermann, W. Keller, Cloning and expressionof the essential gene for poly(A) polymerase from S.cerevisiae, Nature 354 (1991) 496–498.

[25] T. Raabe, F.J. Bollum, J.L. Manley, Primary structure andexpression of bovine poly(A) polymerase, Nature 353 (1991)229–234.

[26] E. Wahle, G. Martin, E. Schiltz, W. Keller, Isolation andexpression of cDNA clones encoding mammalian poly(A)polymerase, EMBO J. 10 (1991) 4251–4257.

[27] C.B. Kyriakopoulou, H. Nordvarg, A. Virtanen, A novelnuclear human poly(A) polymerase (PAP), PAP gamma, J.Biol. Chem. 276 (2001) 33,504–33,511.

[28] K. Perumal, K. Sinha, D. Henning, R. Reddy, Purification,characterization, and cloning of the cDNA of human signalrecognition particle RNA 3′-adenylating enzyme, J. Biol.Chem. 276 (2001) 21,791–21,796.

[29] S.L. Topalian, S. Kaneko, M.I. Gonzales, G.L. Bond, Y. Ward,J.L.Manley, Identification and functional characterization of neo-poly(A) polymerase, an RNA processing enzyme overex-pressed in human tumors, Mol. Cell. Biol. 21 (2001) 5614–5623.

[30] S. Kashiwabara, T. Zhuang, K. Yamagata, J. Noguchi, A.Fukamizu, T. Baba, Identification of a novel isoform of poly(A)polymerase, TPAP, specifically present in the cytoplasm ofspermatogenic cells, Dev. Biol. 228 (2000) 106–115.

[31] R.S. Laishram, Poly(A) polymerase (PAP) diversity in geneexpression–star-PAP vs canonical PAP, FEBS Lett. 588(2014) 2185–2197.

[32] J. LaCava, J. Houseley, C. Saveanu, E. Petfalski, E.Thompson, A. Jacquier, et al., RNA degradation by theexosome is promoted by a nuclear polyadenylation complex,Cell 121 (2005) 713–724.

[33] S. Vanacova, J. Wolf, G. Martin, D. Blank, S. Dettwiler, A.Friedlein, et al., A new yeast poly(A) polymerase complexinvolved in RNA quality control, PLoS Biol. 3 (2005) e189.

[34] F. Wyers, M. Rougemaille, G. Badis, J.C. Rousselle, M.E.Dufour, J. Boulay, et al., Cryptic pol II transcripts aredegraded by a nuclear quality control pathway involving anew poly(A) polymerase, Cell 121 (2005) 725–737.

[35] S. Vanacova, R. Stefl, The exosome and RNA quality controlin the nucleus, EMBO Rep. 8 (2007) 651–657.

[36] R. Nakamura, R. Takeuchi, K. Takata, K. Shimanouchi, Y. Abe,Y. Kanai, et al., TRF4 is involved in polyadenylation of snRNAs inDrosophila melanogaster, Mol. Cell. Biol. 28 (2008) 6620–6631.

[37] C. Harnisch, S. Cuzic-Feltens, J.C. Dohm, M. Gotze, H.Himmelbauer, E. Wahle, Oligoadenylation of 3′ decayintermediates promotes cytoplasmic mRNA degradation inDrosophila cells, RNA 22 (2016) 428–442.

[38] H. Berndt, C. Harnisch, C. Rammelt, N. Stohr, A. Zirkel, J.C.Dohm, et al., Maturation of mammalian H/ACA box snoR-NAs: PAPD5-dependent adenylation and PARN-dependenttrimming, RNA 18 (2012) 958–972.

[39] T.W. Turowski, D. Tollervey, Cotranscriptional events ineukaryotic ribosome synthesis, Wiley Interdiscip. Rev. RNA 6(2014) 129–139.

[40] M. Lubas, M.S. Christensen, M.S. Kristiansen, M. Domanski,L.G. Falkenby, S. Lykke-Andersen, et al., Interaction profilingidentifies the human nuclear exosome targeting complex,Mol. Cell 43 (2011) 624–637.

[41] R.D. Etheridge, D.M. Clemens, P.D. Gershon, R.Aphasizhev, Identification and characterization of nuclearnon-canonical poly(A) polymerases from Trypanosomabrucei, Mol. Biochem. Parasitol. 164 (2009) 66–73.

[42] M. Cristodero, C.E. Clayton, Trypanosome MTR4 is involvedin rRNA processing, Nucleic Acids Res. 35 (20) (2007)7023–7030.

[43] A. van Hoof, P. Lennertz, R. Parker, Yeast exosome mutantsaccumulate 3′-extended polyadenylated forms of U4 smallnuclear RNA and small nucleolar RNAs, Mol. Cell. Biol. 20(2000) 441–452.

[44] A. Fatica,M.Morlando, I. Bozzoni, Yeast snoRNAaccumulationrelies on a cleavage-dependent/polyadenylation-independent3′-processing apparatus, EMBO J. 19 (2000) 6218–6229.

[45] D.E. Egecioglu, A.K. Henras, G.F. Chanfreau, Contributionsof Trf4p- and Trf5p-dependent polyadenylation to theprocessing and degradative functions of the yeast nuclearexosome, RNA 12 (2006) 26–32.

3318 PAP PAP1 Polyadenylates Non-Coding RNAs

[46] J.F. Lemay, A. D'Amours, C. Lemieux, D.H. Lackner, V.G. St-Sauveur, J. Bahler, et al., The nuclear poly(A)-binding proteininteracts with the exosome to promote synthesis of noncod-ing small nucleolar RNAs, Mol. Cell 37 (2010) 34–45.

[47] S.M. Bresson, O.V. Hunter, A.C. Hunter, N.K. Conrad,Canonical poly(A) polymerase activity promotes the decayof a wide variety of mammalian nuclear RNAs, PLoS Genet.11 (2015) e1005610.

[48] G. Mair, H. Shi, H. Li, A. Djikeng, H.O. Aviles, J.R. Bishop,et al., A new twist in trypanosome RNA metabolism: cis-splicing of pre-mRNA, RNA 6 (2000) 163–169.

[49] L. Holm, P. Rosenstrom, Dali server: conservation mappingin 3D, Nucleic Acids Res. 38 (2010) W545–W549.

[50] Z. Wang, J.C. Morris, M.E. Drew, P.T. Englund, Inhibition ofTrypanosoma brucei gene expression by RNA interferenceusing an integratable vector with opposing T7 promoters, J.Biol. Chem. 275 (2000) 40,174–40,179.

[51] S. Alsford, D.J. Turner, S.O. Obado, A. Sanchez-Flores, L.Glover, M. Berriman, et al., High-throughput phenotypingusing parallel sequencing of RNA interference targets in theAfrican trypanosome, Genome Res. 21 (2011) 915–924.

[52] A. Gunzl, B. Schimanski, Tandem affinity purification ofproteins, Curr Protoc Protein Sci. 55 (2009) 19.19.1–19.19.16.

[53] S. Dean, J. Sunter, R.J. Wheeler, I. Hodkinson, E. Gluenz, K.Gull, A toolkit enabling efficient, scalable and reproduciblegene tagging in trypanosomatids, OpenBiol. 5 (2015) 140,197.

[54] M. Mandelboim, C.L. Estrano, C. Tschudi, E. Ullu, S.Michaeli, On the role of exon and intron sequences intrans-splicing utilization and cap 4 modification of thetrypanosomatid Leptomonas collosoma SL RNA, J. Biol.Chem. 277 (2002) 35,210–35,218.

[55] V. Nakaar, A.O. Dare, D. Hong, E. Ullu, C. Tschudi,Upstream tRNA genes are essential for expression of smallnuclear and cytoplasmic RNA genes in trypanosomes, Mol.Cell. Biol. 14 (1994) 6736–6742.

[56] S. Egloff, D. O'Reilly, S. Murphy, Expression of humansnRNA genes from beginning to end, Biochem. Soc. Trans.36 (2008) 590–594.

[57] P. Grzechnik, J. Kufel, Polyadenylation linked to transcriptiontermination directs the processing of snoRNA precursors inyeast, Mol. Cell 32 (2008) 247–258.

[58] H.G. van Luenen, C. Farris, S. Jan, P.A. Genest, P. Tripathi,A. Velds, et al., Glucosylated hydroxymethyluracil, DNA baseJ, prevents transcriptional readthrough in Leishmania, Cell150 (2012) 909–921.

[59] C.R. Eckmann, C. Rammelt, E. Wahle, Control of poly(A) taillength, Wiley Interdiscip. Rev. RNA 2 (2011) 348–361.

[60] S. San Paolo, S. Vanacova, L. Schenk, T. Scherrer, D. Blank,W. Keller, et al., Distinct roles of non-canonical poly(A)polymerases in RNA metabolism, PLoS Genet. 5 (2009)e1000555.

[61] C. Kappel, G. Trost, H. Czesnick, A. Ramming, B. Kolbe, S.L.Vi, et al., Genome-wide analysis of PAPS1-dependentpolyadenylation identifies novel roles for functionally special-ized poly(A) polymerases in Arabidopsis thaliana, PLoSGenet. 11 (2015) e1005474.

[62] L.R. Meeks, B. Addepalli, A.G. Hunt, Characterization ofgenes encoding poly(A) polymerases in plants: evidence forduplication and functional specialization, PLoS One 4 (2009)e8082.

[63] S.K. Gupta, V. Chikne, D. Eliaz, I.D. Tkacz, I. Naboishchikov,S. Carmi, et al., Two splicing factors carrying serine-argininemotifs, TSR1 and TSR1IP, regulate splicing, mRNA stability,and rRNA processing in Trypanosoma brucei, RNA Biol. 11(2014) 715–731.

[64] S.K. Gupta, A. Hury, Y. Ziporen, H. Shi, E. Ullu, S. Michaeli,Small nucleolar RNA interference in Trypanosoma brucei:mechanism and utilization for elucidating the function ofsnoRNAs, Nucleic Acids Res. 38 (2010) 7236–7247.

[65] M. Mandelboim, S. Barth, M. Biton, X.H. Liang, S. Michaeli,Silencing of Sm proteins in Trypanosoma brucei by RNAinterference captured a novel cytoplasmic intermediate inspliced leader RNA biogenesis, J. Biol. Chem. 278 (2003)51,469–51,478.

[66] X.H. Liang, L. Liu, S. Michaeli, Identification of the firsttrypanosome H/ACA RNA that guides pseudouridine forma-tion on rRNA, J. Biol. Chem. 276 (2001) 40,313–40,318.

[67] Y. Xu, L. Liu, C. Lopez-Estrano, S. Michaeli, Expressionstudies on clustered trypanosomatid box C/D small nucleolarRNAs, J. Biol. Chem. 276 (2001) 14,289–14,298.

[68] A.R. Quinlan, I.M. Hall, BEDTools: a flexible suite of utilitiesfor comparing genomic features, Bioinformatics 26 (2010)841–842.

[69] J.T. Robinson, H. Thorvaldsdottir, W. Winckler, M. Guttman,E.S. Lander, G. Getz, et al., Integrative genomics viewer,Nat. Biotechnol. 29 (2011) 24–26.

[70] M. Biasini, S. Bienert, A. Waterhouse, K. Arnold, G. Studer, T.Schmidt, et al., SWISS-MODEL: modelling protein tertiaryand quaternary structure using evolutionary information,Nucleic Acids Res. 42 (2014) W252–W258.

[71] K. Arnold, L. Bordoli, J. Kopp, T. Schwede, The SWISS-MODEL workspace: a web-based environment for proteinstructure homology modelling, Bioinformatics 22 (2006)195–201.

[72] F. Kiefer, K. Arnold, M. Künzli, L. Bordoli, T. Schwede, TheSWISS-MODEL repository and associated resources,Nucleic Acids Res. 37 (2009) D387–D392.

[73] N. Guex, M.C. Peitsch, T. Schwede, Automated comparativeprotein structure modeling with SWISS-MODEL and Swiss-PdbViewer: a historical perspective, Electrophoresis 30(2009) S162–S173.