FEBS Letters 587 (2013) 555–561

journal homepage: www.FEBSLetters .org

Hypothesis

Nested introns in an intron: Evidence of multi-step splicing in a largeintron of the human dystrophin pre-mRNA

0014-5793/$36.00 � 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.febslet.2013.01.057

Abbreviations: RT, reverse transcription; PCR, polymerase chain reaction; PAGE,polyacrylamide gel electrophoresis; EJC, exon junction complex⇑ Corresponding authors at: Center for Nano Materials and Technology, Japan

Advanced Institute of Science and Technology, Nomi, Ishikawa 923-1292, Japan.Fax: +81 761 51 1455 (H. Suzuki); Division of Gene Expression Mechanism, Institutefor Comprehensive Medical Science, Fujita Health University, Toyoake 470-1192,Aichi Japan. Fax: +81 562 93 8834 (A. Mayeda).

E-mail addresses: suzuki-h@jaist.ac.jp (H. Suzuki), mayeda@fujita-hu.ac.jp(A. Mayeda).

1 Present address: Graduate School of Medicine, Nagoya University, Nagoya 466-8550, Japan.

Hitoshi Suzuki a,b,⇑, Toshiki Kameyama c, Kenji Ohe c,1, Toshifumi Tsukahara a, Akila Mayeda c,⇑a School of Materials Science, Japan Advanced Institute of Science and Technology, Nomi, Ishikawa 923-1292, Japanb Center for Nano Materials and Technology, Japan Advanced Institute of Science and Technology, Nomi, Ishikawa 923-1292, Japanc Division of Gene Expression Mechanism, Institute for Comprehensive Medical Science, Fujita Health University, Toyoake 470-1192, Aichi, Japan

a r t i c l e i n f o

Article history:Received 8 December 2012Revised 18 January 2013Accepted 23 January 2013Available online 5 February 2013

Edited by Gianni Cesareni

Keywords:SplicingNested splicingLarge intronNested intronDystrophin geneRNase R

a b s t r a c t

The mechanisms by which huge human introns are spliced out precisely are poorly understood. Weanalyzed large intron 7 (110 199 nucleotides) generated from the human dystrophin (DMD) pre-mRNA by RT-PCR. We identified branching between the authentic 50 splice site and the branch point;however, the sequences far from the branch site were not detectable. This RT-PCR product was resis-tant to exoribonuclease (RNase R) digestion, suggesting that the detected lariat intron has a closedloop structure but contains gaps in its sequence. Transient and concomitant generation of at leasttwo branched fragments from nested introns within large intron 7 suggests internal nested splicingevents before the ultimate splicing at the authentic 50 and 30 splice sites. Nested splicing events,which bring the authentic 50 and 30 splice sites into close proximity, could be one of the splicingmechanisms for the extremely large introns.� 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction

There is great variation in the sizes of human genes, which ispredominantly attributable to the variation in intron sizes [1].Most internal exons are between 50 and 200 bp (mean, 163 bp),whereas intron sizes are much more variable, ranging from<50 bp to >1 Mbp (mean, 5849 bp) [2]. We recently identified ex-treme cases of splicing in the human ESRP2 and FHIT genes; i.e.,alternatively spliced 43-nucleotide (nt) intron and cancer-specificmulti-exon skipping with 1189164-nt span, respectively [3,4].

The mechanism of pre-mRNA splicing has been studied withmodel pre-mRNAs containing small (100–250 nt) single introns,which are spliced efficiently in vivo and in vitro [reviewed in 5,6].To study the co-transcriptional splicing mechanisms of huge introns

is demanding, so the proposed mechanism is currently controver-sial. For instance, splicing of a discontinuous pre-mRNA, or a cleavedintron, has been proposed [7] (M.J. Dye, personal communication),while a continuous pre-mRNA has been shown to be necessary forsplicing [8]. The multi-step sequential splicing may have importantimplications for understanding splicing of pre-mRNA containing ex-tremely large introns. One known example is recursive splicing inseveral genes of the fruit fly (Drosophila melanogaster); i.e., the step-wise removal of introns by sequential re-splicing at composite 30/50

splice sites (0-nt length exons) [9,10] (Fig. 1). This recursive splicingwas discovered in several genes of D. melanogaster, but has not yetbeen found in vertebrate genes. On the other hand, the measuredrates of the transcription and splicing of pre-mRNA containing hugeintrons suggest an authentic one-step process [11] (R.A. Padgett,personal communication), but we do not yet know why the splicingmachinery, or spliceosome, ignores so many splice site-like se-quences in a huge intron.

The human dystrophin (DMD) gene is one of the largest anno-tated genes; it spans more than 2.5 Mbp and generates a transcriptof �14 kb containing 79 coding exons, which encodes the essential427 kDa dystrophin protein in skeletal and cardiac muscles (http://www.dmd.nl/). The mutations in this DMD gene often result in theDuchenne and Becker muscular dystrophies [reviewed in 12,13].More than 99% of the DMD gene sequence is composed of introns,whose lengths vary widely from 107 nt (intron 14) to 248 401 nt

Fig. 1. Two types of multi-step splicing in large introns, compared with the recently discovered re-splicing of mature mRNA. ‘Recursive splicing’ and ‘re-splicing of maturemRNA’ were first identified in D. melanogaster Ubx pre-mRNA and human TSG101 pre-mRNA, respectively [4,9,10]. The proposed nested splicing pathways (middle column)are mechanistically homologous to the pathway of mRNA re-splicing (right column). Each multi-step splicing pathway is represented here with a minimal number of exons/introns. Black and gray 50/30 indicate the active and inactive 50/30 splice sites in the each process, respectively.

556 H. Suzuki et al. / FEBS Letters 587 (2013) 555–561

(intron 44). When we examined large intron 7 (110 199 nt) of theDMD pre-mRNA, experimental evidence suggested that multiplenested splicing events occur before the ultimate splicing betweenthe authentic 50 and 30 splice sites (Fig. 1), which was theoreticallypredicted and referred to as ‘intrasplicing’ previously [14]. In thisprocess, many splice site-like sequences within the large intronrather play roles in splicing events of the nested introns, and as aresult, they bring the distant authentic splice sites into close prox-imity to facilitate the final splicing, removing the huge intron even-tually. Based on the supportive experimental evidence, here wepropose this ‘nested splicing hypothesis’ as a possible splicingmechanism of human huge introns.

2. Materials and methods

The sequences of all the DNA primers used (Operon Biotechnol-ogies) are listed (Supplementary Table S1).

2.1. Confirmation of the human DMD gene structure

To confirm the exon/intron structure of the human DMD gene,the longest cDNA sequence (accession number: NM_000109) wasmapped onto the appropriate region of the human X chromosomeusing the human BLAT search (http://genome.ucsc.edu/cgi-bin/hgBlat). Human Splicing Finder (HSF; http://www.umd.be/HSF/)was used to search for sequences homologous to the consensus50 and 30 splice sites of DMD intron 7.

2.2. RT-PCR analysis of DMD lariat introns

The methods to detect lariat introns using reverse transcription-polymerase chain reaction (RT-PCR) across branch sites and thepreparation of RNase R-digested RNA sample have been describedpreviously [15]. Human skeletal muscle total RNA (1 lg; Clontech)or RNase R-digested RNA from the same source (1 lg) were usedas the templates for the RT reactions with random hexamer primers.The resulting cDNAs were used as templates for the first PCRs withspecific primers (Supplementary Table S1) and the products werepurified on Sephacryl S300 columns (GE Healthcare), before they

were used as the templates for the second nested PCRs. The secondPCRs with the inner primer sets (Supplementary Table S1) were per-formed as described for the first PCRs. The first and second PCRswere performed with 20 and 10 cycles, respectively (mRNA inFig. 2C), 30 and 19 cycles, respectively (introns 7 and 8 in Fig. 2C),or 35 and 35 cycles, respectively (in Figs. 2A, B and 3). The amplifiedPCR products were analyzed with 6% polyacrylamide gel electropho-resis (PAGE; stained with ethidium bromide). The isolated DNA frag-ments were subcloned into the pCR2.1 TOPO cloning vector(Invitrogen) and the sequences were verified with the M13 re-verse/forward primers.

2.3. Plasmid construction for in vitro splicing

PCRs were performed with Pfu Turbo DNA polymerase (Agilent),according to the manufacturer’s instructions, with the indicatedDNA primers (Supplementary Table S1). Using extension PCR onthe human b-globin minigene plasmid pSP64-HbD6 [16] with theprimers HbGE1AS (containing an ApaI site) and HbGE2S (contain-ing a BglII site), we generated the b-globin cDNA plasmid pSP64-HbE1/E2, in which intron 1 with its adjacent 50 and 30 splice siteswas replaced with two unique restriction sites (ApaI and BglII).Nested intron fragments A and B with internal deletions were gen-erated with PCR amplification of human genomic DNA (Promega)using primer sets A50ApaIS, A50XhoIAS, A30XhoIS and A30BglIIASfor the A fragment, and primer sets B50ApaIS, B50XhoIAS, B30XhoISand B30BglIIAS for the B fragment. The amplified products were di-gested with ApaI/XhoI or XhoI/BglII and subcloned into the ApaI/BglII sites of the pSP64-HbE1/E2 plasmid to generate plasmidspSP64-HbE1/E2-DMDniA and pSP64-HbE1/E2-DMDniB.

2.4. In vitro splicing assay of nested introns

The pSP64-HbE1/E2-DMDniA and pSP64-HbE1/E2-DMDniBplasmids were linearized with BamHI and used as the templatesfor in vitro transcription with SP6 RNA polymerase as describedpreviously [17]. The in vitro splicing assays of the nested intronswere performed with HeLa cell nuclear extract as described previ-ously [17], unless otherwise specified below. The splicing products

Fig. 2. RT-PCR detected the whole DMD small lariat intron 8 (1113 nt) but detected only part of DMD large lariat intron 7 (110 199 nt) in muscle cell RNA. (A) Structures ofthe DMD pre-mRNA (exon 8 to exon 9) and lariat intron 8 with the primers (arrows) used for the PCR assays are indicated (all the primer sets, except 50S/30A, weredesigned not to amplify product from linear pre-mRNAs). The PCR products on the PAGE gel are shown. Sequence alignment of the PCR product (amplified with primer set30S/50A) and the DMD gene revealed a 20–50 branched connection between the end of the authentic 50 splice site (arrow) and the branch point, which is located upstreamfrom the authentic 30 splice site (arrow). The other products (from primer sets 30S/30A and 50S/50A) also had consistent sequences. (B) Structures of the DMD pre-mRNA(exon 7 to exon 8) and lariat intron 7 with the primers (arrows), and the PAGE gel are shown as in (A). Sequence alignment of the PCR product (from the primer set SI/AI)and the DMD gene reveals canonical branching between the authentic 50 splice site and the branch point as described in (A). The observed ‘t’ instead of the branching ‘a’could be attributable to a nucleotide mis-incorporated during RT as described previously [28]. The other products (from primer sets SII/AI and SI/AII) also had consistentsequences. (C) RT-PCR detection on the denaturing PAGE gel of the indicated RNAs (mRNA, intron 8, and intron 7) using human skeletal muscle total RNA with (+) orwithout (�) RNase R-digestion. RNase R fully degrades linear RNAs, such as rRNA and mRNA, but circular RNAs and the loop portions of lariat RNAs remain intact (seelanes of total RNA) [15].

H. Suzuki et al. / FEBS Letters 587 (2013) 555–561 557

were directly analyzed with denaturing 5.5% PAGE and autoradiog-raphy. The spliced products were also detected by RT-PCR with aspecific RT primer (HbGBAS) and PCR primers (HbGS and HbGBAS),followed by 6% PAGE.

2.5. Analysis of splicing products during muscle cell differentiation

Human skeletal muscle cells (SkMC cells; Cambrex) weremaintained in SkGM medium supplemented with Single Quotadditives according to the manufacturer’s instructions. To pro-mote the cells’ terminal differentiation, the SkGM medium wasreplaced with Dulbecco’s modified Eagle’s medium (D-MEM/F-12; Life Technologies) supplemented with 2% horse serum. Thetotal RNAs were extracted using TRIzol reagent (Life Technolo-gies) at each indicated time point after the medium was replaced.In the semi-quantitative RT-PCR assays to detect specific DMDsplicing products, the optimal cycle numbers of the first and sec-ond PCRs were selected to reflect the proportional changes in thePCR products [15]; i.e., 20 cycles (with E7-S/E8-A primers) and 16cycles (with E7-S0/E8-A0 primers), respectively, to detect themRNA; 25 cycles (with AI/SI primers) and 17 cycles (with AI0/SI0

primers), respectively, to detect the lariat intron 7; 25 cycles(with +10 k/+25 k primers) and 19 cycles (with +10 k0/+25 k0

primers), respectively, to detect the nested lariat intron A; and25 cycles (with +19 k/+24 k primers) and 21 cycles (with +19 k0/

+24 k0 primers), respectively, to detect the nested lariat intronB. The total RNA from the SkMC cells (250 ng) and the PCR prod-ucts were analyzed as described in Section 2.2, and the visualizedbands were quantitated with an AlphaImager 2000 instrument(Alpha Innotech). The sequences were verified as described inSection 2.2.

3. Results

3.1. DMD conventional lariat intron 8 was fully detected

We chose two representative introns in the DMD pre-mRNA:very large intron 7 (110 199 nt) and the downstream conventionalintron 8 (1113 nt). We used RT-PCR amplification across the branchsite to detect trace amounts of the endogenous lariat introns fromthe human skeletal muscle total RNA as described previously [15].We first amplified the control intron 8 (1113 nt) from the endoge-nous total RNA. Human skeletal muscle cDNA was used as the tem-plate for a PCR with the 30S/50A primers, followed by a second nestedPCR with the inner primers, which hybridized across the branchpoint in lariat intron 8 (Fig. 2A; see maps). We successfully detectedan RT-PCR product of the expected size (see middle PAGE panel) andthe sequence was verified (see aligned sequences). We also detecteda product of the right size with two different primer sets (30S/30A and50S/50A) in opposite directions, which included the whole loop,

Fig. 3. RT-PCR detected the canonical branched structures of two nested introns in DMD intron 7. (A and B) Structures of the DMD pre-mRNA (exon 7 to exon 8) and thenested lariat introns A and B are indicated schematically with the PCR primers (arrows). The PCR products are shown on the PAGE gel. Sequence alignment of the PCRproducts (amplified with primer sets +25 k/+10 k and +24 k/+19 k) and the DMD gene reveals a branching event between the end of the internal 50 splice site (arrow andunderlined) and the branch point, which is located upstream from the internal 30 splice site (arrow and underlined). The missing nucleotides at the branch point (hyphens)could have resulted from skipping during RT as described previously [28]. The strengths of 50 and 30 splice sites (by the HSF web tool) of the nested intron A are 77.11 (734thof 3466) and 78.90 (1798th of 6559), respectively, and those of the nested intron B are 74.42 (1031th of 3466) and 82.63 (880th of 6559), respectively.

558 H. Suzuki et al. / FEBS Letters 587 (2013) 555–561

suggesting that the intron lariat was intact with a closed loop struc-ture (Fig. 2A; see maps).

3.2. DMD large lariat intron 7 was partially detected

The same RT-PCR approach was applied to large intron 7 (110199 nt). We observed specific PCR products with primer sets (SI/AI, SI/AII, and SII/AI) that hybridized near the authentic branch sitein the presumed lariat intron 7 (Fig. 2B; see middle PAGE panel).The annealed position of the primer sets SI/AII and SII/AI imply thatthe branched product contained at least 156 nt from the 50 splicesite and at least 126 nt from the branch site. Sequencing of thesePCR products clearly demonstrated a branch formation betweenthe cleaved authentic 50 splice site and the branch point that is justupstream from the authentic 30 splice site (Fig. 2B; see aligned se-quences). However, we did not observe the expected PCR productswith the primer sets SI/AIII (�336 nt) and SIII/AI (�226 nt) (seemiddle PAGE panel) under the PCR conditions that amplified theentire 1113-nt lariat intron 8 (Fig. 2A). We detected no more PCRsignals with more distant primer sets (SI/A0.5, � � �, SI/A4.0 andS0.5/AI, � � �, S4.0/AI; Fig. 2B). The RT-PCR undetected regions of in-tron 7, could be detected by the genomic PCR with the correspond-ing primers, which provides quality control data to rule outpossible problems in either the primers or PCR conditions usedin the RT-PCR (Supplementary Fig. S1). These results suggest thatthe expected lariat intron 7 was cleaved to form a Y-shapedbranched fragment or a closed lariat containing gaps in the se-quence (Fig. 2B; black and gray lines), because the sequences at

which the primers should hybridize were missing (gray line;>270 nt from the 50 splice site and >175 nt from the branch site).

To clarify this issue, we analyzed RNase R-digested total skel-etal RNA with RT-PCR. We have previously demonstrated thatRNase R (a 30 to 50 exoribonuclease) thoroughly degrades linearRNAs and Y-shaped branched RNAs, while preserving the loopportions of lariat RNAs [15]. Therefore, we observed that theRT-PCR product derived from the DMD mRNA was abolished byRNase R digestion, whereas the product detected from the con-ventional lariat intron 8 remained after RNase R digestion(Fig. 2C). Remarkably, the product detected from lariat intron 7was also resistant to RNase R digestion (Fig. 2C). These resultsindicate that the observed branched intron 7 was not a cleavedlariat (Y-structure), but a closed lariat. Taken together, the partialRT-PCR detection of lariat intron 7 can be attributed to presumedsequence gaps in the closed lariat.

3.3. Possible nested lariat introns were detected in intron 7

The observed gapped segments in closed lariat intron 7prompted us to look for nested splicing events, which occur we as-sumed, before the final splicing that removes the whole intron 7(Fig. 1). We first searched for potential internal 50 and 30 splice sitesin intron 7 using the HSF web tool. We found 3466 and 6559 se-quences that were homologous to the 50 and 30 splice sites, respec-tively. Among these, 127 and 279 sequences had higher scores thanthose of the authentic 50 and 30 splice sites, respectively. We spec-ulated that these potentially strong 50 and 30 splice site sequencesmight be utilized for nested splicing events within large intron 7.

H. Suzuki et al. / FEBS Letters 587 (2013) 555–561 559

An RT-PCR search for the spliced nested intron was successfulwhen we targeted one of the potential 50 splice site sequences witha specific primer (Fig. 3A, +10 k) and scanned the available down-stream branch site with six primers with a constant distance (+15,+20, � � �, +40 k). We detected a PCR product from the specific down-stream primer (see middle panel, +25 k). Sequencing of this PCRproduct revealed a presumed nested lariat intron (14 367 nt;‘nested lariat intron A’) that could be generated by splicing viathe internal 50 and 30 splice sites in intron 7 (see aligned se-quences). Using the same approach with a different set of primers,we found evidence of another nested lariat intron (6380 nt; ‘nestedlariat intron B’) and verified its sequence (Fig. 3B). Interestingly,the nested-intron B was nested in the nested-intron A. Our resultssuggest that at least two nested splicing events occur before the fi-nal splicing, removing large intron 7 eventually. We assume thatthere are more nested introns and that their splicing creates theobserved sequence gaps in large lariat intron 7.

3.4. Nested intron fragments are active in splicing

We next tested whether the fragments of detected nested intronsA and B themselves function in splicing. We performed an in vitrosplicing assay with heterologous pre-mRNAs containing each nestedintron fragment (Fig. 4). Because b-globin exonic RNA is very stablein splicing reactions [18], the original intron 1 of the b-globin

Fig. 4. The nested intron fragments are competent for splicing. Shortened nested intronexon 2) sequence of b-globin (37 nt from the 30 end of exon 1 and 11 nt from the 50 ennuclear extract (the incubation times are indicated). The splicing products were analyzedto an RT-PCR assay (RT-PCR). The positions of the unspliced pre-mRNAs and spliced mRNby sequencing.

minigene, with its flanking splice sites (including extra exonic se-quences), was replaced with each shortened nested intron includingthe flanking splice sites (Fig. 4, see maps). We found that insertednested intron A was efficiently spliced in vitro, and the splice productwas verified by RT-PCR and sequencing (Fig. 4, see upper panels).Nested intron B was also spliced in vitro, albeit less efficiently (seelower panels). These assays confirmed that nested introns A and Bare themselves splicing competent.

3.5. Nested lariat introns were generated during muscle celldifferentiation

It was critical to examine whether these detected nested intronsare bona fide splicing byproducts from the endogenous DMD gene.DMD gene expression is markedly elevated during the differentia-tion of myoblasts into myotubes [19]. Therefore, we used SkMCcells for the assay, as their terminal differentiation into myotubescan be induced. We then performed semi-quantitative RT-PCR toassay the levels of spliced DMD mRNAs, of intron 7 products, andof nested intron products A and B (Fig. 5A). A minimal level of ma-ture DMD mRNA was detected in uninduced SkMC cells (0 h), butthis increased rapidly for 12 h after induction of differentiation,and accumulated gradually thereafter up to 48 h (Fig. 5B). The lev-els of excised intron 7 increased for the first 12 h, as observed forthe generation of mRNA, but decreased dramatically thereafter to96 h (Fig. 5B), which was apparently attributable to the rapid

fragments A and B (the lengths are indicated) were inserted into an exonic (exon 1/d of exon 2 were deleted). An in vitro splicing assay was performed with HeLa celldirectly by denaturing 5.5% PAGE (dPAGE) and the reaction mixtures were subjectedAs are indicated with their schematic structures. The spliced products were verified

Fig. 5. Nested introns are generated during the DMD splicing process. (A) Analyses of the DMD splicing products during the differentiation of SkMC cells to myotubes. SkMCcells were induced to differentiate into myotubes and the total RNA was prepared during a time course (0–96 h). RT-PCR was used to detect the spliced mRNA, intron 7,nested intron A, and nested intron B products. The 18S and 28S rRNAs are shown as the loading controls for RNA normalization. (B) The RT-PCR signals were quantitated bydensitometry. The relative amounts of the PCR products were normalized to those amplified from RNA extracted from uninduced SkMC cells (0 h). We induced thedifferentiation of SkMC cells twice and obtained consistent RT-PCR results.

560 H. Suzuki et al. / FEBS Letters 587 (2013) 555–561

turnover of the removed intron 7. Remarkably, the generation/deg-radation patterns of nested introns A and B were similar to those ofintron 7, but clearly distinct from those of the mature mRNA(Fig. 5B). The transient generation of nested lariat introns togetherwith de novo DMD gene expression suggests that these nested lar-iat introns are true byproducts of intron 7 splicing.

4. Discussion

Here we report experimental data that support a novel splicinghypothesis, designated ‘nested splicing’, which involves the preced-ing splicing to eliminate nested introns, followed by the ultimatesplicing event between the authentic 50 and 30 splice sites (Fig. 1).We have shown that the lariat RNA excised from DMD large intron7 has a closed lariat structure, but appears to contain gaps in its se-quence. We postulate that preceding nested splicing events gener-ate these gapped segments. This assumption is supported by thesuccessful detection of possible lariat RNAs that were generatedfrom two nested introns through the activation of internal splicesites in large intron 7. A time-course RT-PCR analysis showed thatthe appearance of the nested lariat introns coincided with theDMD splicing process, which argues against their arbitrary genera-tion via the mis-splicing, or abortive splicing, in intron 7. To proveour hypothesis, however, it is critical to identify exact nested lariatintrons that fill the missing segments of lariat intron 7.

The architectures of nested introns within outer intron, termed‘twintrons’, were previously reported in chloroplast and D. melano-gaster genes. The former are group II and/or group III twintrons

that undergo splicing in a sequential manner [reviewed in 20],while the latter are U2 and U12 snRNPs catalyzed twintrons thatare not spliced in a sequential manner but are spliced alternativelyin a developmentally regulated manner [21,22]. These twintronsstructurally similar, but not mechanistically link, to our case ofnested introns in large intron 7 of human DMD gene. Remarkably,our recent discovery of the re-splicing of human mature mRNA im-plies that the spliced product generated via the inner splice sitescan act as the substrate for subsequent splicing via the outer splicesites [4], which mechanistically supports the inner nested splicingfollowed by the outer authentic splicing (Fig. 1). Furthermore, pre-requisite internal splicing followed by external splicing has beenfound to produce specific alternative isoforms from the mamma-lian 4.1R (EPB41) and 4.1B (EPB41L3) pre-mRNAs [23,24]. Theseactual cases of multi-step re-splicing might mechanistically under-pin the proposed model of nested intron spicing, and this possibil-ity motivated us to publish this hypothesis article.

The fact that the first proximally spliced product acts as thesubstrate for subsequent splicing events suggests an unknownmechanism, distinct from that underlying conventional one-stepsplicing. In conventional splicing, splicing does not proceed onthe ligated exons even though many splice site-like sequences ex-ist. A substantial problem that remains to be resolved is to identifythe factor(s) that allows these intermediate RNAs to remain as thesubstrates for subsequent splicing. The exon junction complex(EJC) is loaded onto the splice junction of the mature mRNA in asplicing-dependent manner, and is involved in mRNA export, sub-cellular localization, quality control, and translation [reviewed in

H. Suzuki et al. / FEBS Letters 587 (2013) 555–561 561

25]. Therefore, it is intriguing to ask whether the EJC, which mightbe a potential signal for the completion of splicing, is not loadedonto the intronic splice junction after a nested splicing event. Inter-estingly, it was reported that EJCs are not associated with everyspliced junction and that their assembly is a regulated process[26,27]. Our exploratory study provides a useful model with whichto study the mechanisms of splice site selection, splicing initiation,and splicing termination.

Acknowledgments

We would like to thank Dr. A. Malhotra for providing purifiedrecombinant RNase R; Drs. G.R. Screaton, J. Wang and M.Q. Zhangfor their helps in the initial stages of this study; Drs. M.P. Deut-scher, K.E. Rudd and A.R. Krainer for their valuable suggestionsand encouragement; and T. Venkataraman for technical assistance.A.M. was supported by a Developmental Grant from the MuscularDystrophy Association and a Grant-in-Aid for Challenging Explor-atory Research from Japan Society for the Promotion of Science(JSPS). H.S. was supported in part by a Grant-in-Aid for Young Sci-entists (B) from the JSPS, and in part by an Intramural ResearchGrant (22–5) for Neurological and Psychiatric Disorders of NCNP.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.febslet.2013.01.057.

References

[1] Sakharkar, M.K., Perumal, B.S., Sakharkar, K.R. and Kangueane, P. (2005) Ananalysis on gene architecture in human and mouse genomes. In Silico Biol. 5,347–365.

[2] Zhu, L., Zhang, Y., Zhang, W., Yang, S., Chen, J.Q. and Tian, D. (2009) Patterns ofexon–intron architecture variation of genes in eukaryotic genomes. BMCGenomics 10, 47.

[3] Sasaki-Haraguchi, N., Shimada, M.K., Taniguchi, I., Ohno, M. and Mayeda, A.(2012) Mechanistic insights into human pre-mRNA splicing of human ultra-short introns: potential unusual mechanism identifies G-rich introns.Biochem. Biophys. Res. Commun. 423, 289–294.

[4] Kameyama, T., Suzuki, H. and Mayeda, A. (2012) Re-splicing of mature mRNAin cancer cells promotes activation of distant weak alternative splice sites.Nucleic Acids Res. 40, 7896–7906.

[5] Matlin, A.J. and Moore, M.J. (2007) Spliceosome assembly and composition.Adv. Exp. Med. Biol. 623, 14–35.

[6] Will, C.L. and Lührmann, R. (2011) Spliceosome structure and function. ColdSpring Harb. Perspect. Biol. 3, a003707.

[7] Dye, M.J., Gromak, N. and Proudfoot, N.J. (2006) Exon tethering in transcriptionby RNA polymerase II. Mol. Cell 21, 849–859.

[8] Fong, N., Ohman, M. and Bentley, D.L. (2009) Fast ribozyme cleavage releasestranscripts from RNA polymerase II and aborts co-transcriptional pre-mRNAprocessing. Nat. Struct. Mol. Biol. 16, 916–922.

[9] Hatton, A.R., Subramaniam, V. and Lopez, A.J. (1998) Generation of alternativeUltrabithorax isoforms and stepwise removal of a large intron by resplicing atexon–exon junctions. Mol. Cell 2, 787–796.

[10] Burnette, J.M., Miyamoto-Sato, E., Schaub, M.A., Conklin, J. and Lopez, A.J.(2005) Subdivision of large introns in Drosophila by recursive splicing atnonexonic elements. Genetics 170, 661–674.

[11] Singh, J. and Padgett, R.A. (2009) Rates of in situ transcription and splicing inlarge human genes. Nat. Struct. Mol. Biol. 16, 1128–1133.

[12] Muntoni, F., Torelli, S. and Ferlini, A. (2003) Dystrophin and mutations: onegene, several proteins, multiple phenotypes. Lancet Neurol. 2, 731–740.

[13] Aartsma-Rus, A., Van Deutekom, J.C., Fokkema, I.F., Van Ommen, G.J. and DenDunnen, J.T. (2006) Entries in the Leiden Duchenne muscular dystrophymutation database: an overview of mutation types and paradoxical cases thatconfirm the reading-frame rule. Muscle Nerve 34, 135–144.

[14] Ott, S., Tamada, Y., Bannai, H., Nakai, K. and Miyano, S. (2003) Intrasplicing-analysis of long intron sequences. Pac. Symp. Biocomput. 339–350.

[15] Suzuki, H., Zuo, Y., Wang, J., Zhang, M.Q., Malhotra, A. and Mayeda, A. (2006)Characterization of RNase R-digested cellular RNA source that consists oflariat and circular RNAs from pre-mRNA splicing. Nucleic Acids Res. 34, e63.

[16] Krainer, A.R., Maniatis, T., Ruskin, B. and Green, M.R. (1984) Normal andmutant human b-globin pre-mRNAs are faithfully and efficiently splicedin vitro. Cell 36, 993–1005.

[17] Mayeda, A. and Krainer, A.R. (2012) In vitro splicing assays in: Alternative pre-mRNA Splicing: Theory and Protocols (Stamm, S., Smith, C. and Lührmann, R.,Eds.), pp. 320–329, Wiley-Blackwell, Weinheim, Germany.

[18] Mayeda, A. and Ohshima, Y. (1990) b-globin transcripts carrying a singleintron with three adjacent nucleotides of 50 exon are efficiently spliced in vitroirrespective of intron position or surrounding exon sequences. Nucleic AcidsRes. 18, 4671–4676.

[19] Chelly, J., Montarras, D., Pinset, C., Berwald-Netter, Y., Kaplan, J.C. and Kahn, A.(1990) Quantitative estimation of minor mRNAs by cDNA-polymerase chainreaction. Application to dystrophin mRNA in cultured myogenic and braincells. Eur. J. Biochem. 187, 691–698.

[20] Copertino, D.W. and Hallick, R.B. (1993) Group II and group III introns oftwintrons: potential relationships with nuclear pre-mRNA introns. TrendsBiochem. Sci. 18, 467–471.

[21] Scamborova, P., Wong, A. and Steitz, J.A. (2004) An intronic enhancer regulatessplicing of the twintron of Drosophila melanogaster prospero pre-mRNA by twodifferent spliceosomes. Mol. Cell. Biol. 24, 1855–1869.

[22] Borah, S., Wong, A.C. and Steitz, J.A. (2009) Drosophila hnRNP A1 homologsHrp36/Hrp38 enhance U2-type versus U12-type splicing to regulatealternative splicing of the prospero twintron. Proc. Natl. Acad. Sci. USA 106,2577–2582.

[23] Parra, M.K., Tan, J.S., Mohandas, N. and Conboy, J.G. (2008) Intrasplicingcoordinates alternative first exons with alternative splicing in the protein 4.1Rgene. EMBO J. 27, 122–131.

[24] Parra, M.K., Gallagher, T.L., Amacher, S.L., Mohandas, N. and Conboy, J.G.(2012) Deep intron elements mediate nested splicing events at consecutiveAG dinucleotides to regulate alternative 30 splice site choice in vertebrate 4.1genes. Mol. Cell. Biol. 32, 2044–2053.

[25] Le Hir, H. and Seraphin, B. (2008) EJCs at the heart of translational control. Cell133, 213–216.

[26] Saulière, J., Haque, N., Harms, S., Barbosa, I., Blanchette, M. and Le Hir, H.(2010) The exon junction complex differentially marks spliced junctions. Nat.Struct. Mol. Biol. 17, 1269–1271.

[27] Saulière, J., Murigneux, V., Wang, Z., Marquenet, E., Barbosa, I., Le Tonqueze, O.,Audic, Y., Paillard, L., Roest Crollius, H. and Le Hir, H. (2012) CLIP-seq ofeIF4AIII reveals transcriptome-wide mapping of the human exon junctioncomplex. Nat. Struct. Mol. Biol. 19, 1124–1131.

[28] Gao, K., Masuda, A., Matsuura, T. and Ohno, K. (2008) Human branch pointconsensus sequence is yUnAy. Nucleic Acids Res. 36, 2257–2267.