The splicing of tiny introns of Paramecium is controlled by MAGO

Accepted Manuscript The splicing of tiny introns of Paramecium is controlled by MAGO

Julia Contreras, Victoria Begley, Laura Marsella, Eduardo Villalobo PII: DOI: Reference:

S0378-1119(18)30368-8 doi:10.1016/j.gene.2018.04.007 GENE 42731

To appear in:

Gene

Received date: Revised date: Accepted date:

21 November 2017 27 March 2018 4 April 2018

Please cite this article as: Julia Contreras, Victoria Begley, Laura Marsella, Eduardo Villalobo , The splicing of tiny introns of Paramecium is controlled by MAGO. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Gene(2017), doi:10.1016/j.gene.2018.04.007

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT THE SPLICING OF TINY INTRONS OF Paramecium IS CONTROLLED BY MAGO Julia Contreras1,2§, Victoria Begley1,2§, Laura Marsella1,3 and Eduardo Villalobo1*. 1

Departamento de Microbiología, Universidad de Sevilla, Seville, Spain.

(L.M.), [email protected] (E.V.)

SC RI PT

E-mail: [email protected] (J.C.), [email protected] (V.B.), [email protected]

ORICD: orcid.org/0000-0002-7502-1450 (J.C.), orcid.org/0000-0001-9000-7051 (V.B.), 0000-0001-8740-1832 (L.M.), 0000-0002-0331-115X (E.V.) These authors contributed equally to this work

NU

§

*

MA

Correspondence and reprints

Present address: 2Instituto de Investigaciones Biomédicas, Universidad de Sevilla, Seville,

AC

CE

PT

ED

Spain. 3University of Portsmouth, Portsmouth, Hampshire, United Kingdom.

ACCEPTED MANUSCRIPT Abstract The exon junction complex (EJC) is a key element of the splicing machinery. The EJC core is composed of eIF4A3, MAGO, Y14 and MLN51. Few accessory proteins, such as CWC22 or UPF3, bind transiently to the EJC. The EJC has been implicated in the control of the splicing

SC RI PT

of long introns. To ascertain whether the EJC controls the splicing of short introns, we used Paramecium tetraurelia as a model organism, since it has thousands of very tiny introns. To elucidate whether EJC affects intron splicing in P. tetraurelia, we searched for EJC proteincoding genes, and silenced those genes coding for eIF4A3, MAGO and CWC22. We found that P. tetraurelia likely assembles an active EJC with only three of the core proteins, since

NU

MLN51 is lacking. Silencing of eIF4A3 or CWC22 genes, but not that of MAGO, caused

MA

lethality. Silencing of the MAGO gene caused either an increase, decrease, or no change in the relative levels of some intron-containing mRNAs used as reporters. We suggest that a fine-tuning expression of EJC genes is required for steady intron removal in P. tetraurelia.

ED

Taking into consideration our results and those published by others, we conclude that the EJC

PT

controls splicing independently of the intron size.

AC

CE

Keywords: exon junction complex; intervening sequence; ciliate.

ACCEPTED MANUSCRIPT 1. Introduction The EJC (reviewed in Boehm and Gehring, 2016; Le Hir, Saulière and Wang, 2016; Woodward et al., 2017) consists of a set of four core proteins: eIF4A3, MAGO, Y14, and MLN51 (also known as CASC3 or Barentsz). The EJC deposits onto mRNAs upstream of

SC RI PT

exon-exon junctions (Boehm and Gehring, 2016; Le Hir, Saulière and Wang, 2016; Woodward et al., 2017). The role of the EJC is not only restricted to serve as a simple hallmark distinguishing spliced mRNAs. On the contrary, the EJC also tightly regulates degradation, translation, and splicing of different mRNAs species (Le Hir, Saulière and Wang, 2016). Few accessory proteins transiently interact with the EJC to broaden or modulate

NU

its functions. Examples of these accessory proteins are CWC22 and UPF3. CWC22 links EJC

MA

assembly and mRNA splicing (Alexandrovet al., 2012; Steckelberg et al., 2012). UPF3 links the EJC and the NMD (Nonsense-Mediated Decay, see Hug, Longman and Cáceres, 2016 for a recent review). The NMD is a multiprotein complex that recognises mRNAs containing

ED

Premature Termination Codons (PTC). NMD-recognised mRNAs are rapidly degraded by the

PT

5’-3’ exonuclease called XRN1 (Hug, Longman and Cáceres, 2016). The origin of PTCcontaining mRNAs is diverse but many are obtained after defective splicing or miss-splicing,

CE

such as intron retention.

AC

In Drosophila, the EJC controls the splicing of a particular subset of intron-containing mRNAs. This subset consists generally of genes with long introns (Ashton-Beaucage et al., 2010; Ashton-Beaucage and Therrien, 2011; Roignant and Treisman, 2010). Short and long introns are spliced following different mechanisms (see Ast, 2004 for a review). The fact that the EJC controls splicing of long introns, and that long and short introns are spliced in different ways, raises the question about the impact that the EJC could have on splicing in organisms whose genomes completely lack long introns. To address this question, the ciliate Paramecium tetraurelia arises as a good model, since it has roughly 90,000 tiny introns with

ACCEPTED MANUSCRIPT an average length of 25 bases (Jaillon et al., 2008). In addition, P. tetraurelia has a constitutive or basal degree of intron retention in a known set of genes (Contreras et al., 2014; Jaillon et al., 2008). These features should make the study on the impact of EJC in splicing in P. tetraurelia straightforward.

SC RI PT

The aim of our work was to elucidate if the EJC affects intron splicing in P. tetraurelia cells (henceforth paramecia). To this end, we primarily searched for genes coding for EJC proteins. Of the EJC core proteins, paramecia only lack MLN51; eIF4A3, MAGO and Y14 were successfully identified in silico. In addition, the accessory protein CWC22 was also identified. Secondly, we silenced (knocked-down or down-regulated) the expression of

NU

eIF4A3, MAGO, and CWC22 genes in paramecia by RNAi. Strikingly, silencing of the

MA

eIF4A3 or CWC22 genes caused lethality, while the silencing of the MAGO gene did not. To evaluate the possible contribution of MAGO down-regulation to splicing, we also determined the change in the relative mRNA levels of several reporter genes, most of which contained

ED

introns. We observed that silencing of the MAGO gene in paramecia caused changes

PT

(increase or decrease) in intron retention levels in a subset of intron-containing reporter genes.

CE

Overall, we suggest here that paramecia have an active EJC, probably built with only three of the core proteins. We also suggest that steady intron splicing in paramecia involves a

AC

fine-tuning expression of EJC genes. Consequently, our observation widens the knowledge on the EJC and its relationship to splicing control. Taking into consideration previously published results (Ashton-Beaucage et al., 2010; Ashton-Beaucage and Therrien, 2011; Roignant and Treisman, 2010; Wang, Murigneux and Le Hir, 2014) and the results presented here, we conclude that the EJC controls splicing independently of intron size.

ACCEPTED MANUSCRIPT 2. Material and methods 2.1. Microbial cultures Paramecium tetraurelia strain d4-2 was routinely grown at 27°C in Cerophyl, which is a buffered (123 mM Tris, 105 mM Na2HPO4, and 38 mM NaH2PO4, pH 7.2) wheat grass (5

SC RI PT

g/l) medium supplemented with β-sitosterol (0.4 mg/l). The Cerophyl was inoculated with Escherichia coli HT115 bacteria as food.

E. coli strain HT115 was routinely grown at 37°C in LB broth (10 g/l tryptone, 5 g/l yeast extract, and 10 g/l NaCl) or agar (2% bactoagar in LB broth).

NU

Whenever required, ampicillin (100 µg/ml) and IPTG (0.4 mM) were added to the LB

MA

or Cerophyl. 2.2. Silencing of P. tetraurelia

ED

Silenced paramecia cultures were obtained following the so-called feeding method

PT

(Galvani and Sperling, 2002). Briefly, paramecia were grown for three days in Cerophyl inoculated with HT115 recombinant bacteria as food. Each HT115 recombinant strain bore a

CE

fragment of the gene to be silenced, PteIF4A3, PtMAGO-1 or PtCWC22-1 (see fragment sizes in Table 1) inserted into plasmid pL4440. PteIF4A3, PtMAGO-1 and PtCWC22-1 DNA

AC

fragments were obtained by PCR (see Table 1 for used primer sequences). The abovementioned DNA fragments were ligated at the EcoRI or ApaI/SacI site of pL4440. To prepare the HT115 recombinant strains, PteIF4A3::pL4440, PtMAGO-1::pL4440 or PtCWC221::pL4440 DNA was transformed in HT115 following the Inoue heat shock method (Inoue, Nojima and Okayama, 1990) and using the corresponding ligation reactions (see below). For preparing the food, each HT115 recombinant strain was treated separately as follows. Bacteria were grown in LB medium supplemented with ampicillin until reaching a final optical density

ACCEPTED MANUSCRIPT of 0.3 at 600 nm. Afterwards, the medium was supplemented with IPTG and bacteria were further grown for four hours, except bacterial controls (non-silenced cultures) that were grown for the same time but without supplementing its medium with IPTG. Finally, HT115 bacteria were sedimented by centrifugation, washed and re-suspended twice in fresh Cerophyl, which was added as food for paramecia. Silencing experiments were repeated twice

SC RI PT

in order to obtain biological replicas.

To determine paramecia number, nine cells were collected from cultures, transferred to depression slides (three cells in each depression), and grown at 27°C in the appropriate conditions (silencing or non-silencing medium). After 24 h, cells were counted under a

NU

dissecting microscope.

MA

2.3. Nucleic acids preparations and manipulations

To obtain genomic DNA from paramecia, a conventional extraction/purification method

ED

was used (Sambrook and Russell, 2001). Purified DNA was suspended in EDTA-containing

PT

buffer (10 mM Tris pH 8 and 0.1 mM EDTA pH 8). To obtain total RNA from paramecia, a combination between a conventional guanidine isothiocyanate method (Sambrook and

CE

Russell, 2001) and a conventional spin column method was used. Briefly, paramecia were sedimented in a centrifuge at low speed, lysed by adding 1 ml of RNAzol-RT per 400 µL of

AC

cells, mixed and incubated at room temperature during 15 min; then 1/10 volume of 1-bromo3-chloropropane was added. An RNA-containing aqueous phase was obtained by centrifuging at 12,000g in a table-top centrifuge for 15 minutes at 4°C. RNA was further purified submitting the whole aqueous phase volume to silica columns according to supplier’s directions (Direct-Zol RNA Miniprep, ZymoResearch). To avoid the presence of any contaminant DNA, the optional in-column DNase-treatment step was carried out for 30 minutes at 37°C. Total RNA was eluted in RNase-free water. To obtain cDNA, a

ACCEPTED MANUSCRIPT conventional 20-µl reverse transcription reaction was carried out according to supplier’s directions (Transcriptor First Strand cDNA Synthesis Kit, Roche), using 500 ng of purified DNA-free total RNA. To obtain plasmid DNA from bacteria, a conventional spin column method was used

SC RI PT

according to supplier’s directions (Speedtools Plasmid DNA Purification Kit, Biotools). To obtain DNA from agarose slices, a conventional spin column method was used according to supplier’s directions (Speedtools PCR Clean Up Kit, Biotools).

PCR reactions were routinely performed in conventional 25-µl volumes, using DNA or

NU

cDNA as template, 200 µM dNTPs, 25 pmol each of the reverse and the forward primer (see Table 1 for sequences and location across genes), and 1 U of Taq DNA polymerase (Biotools)

MA

in Mg+2-containing reaction buffer (75 mM Tris-HCl pH 9.0, 50 mM KCl, 20 mM (NH4)2SO4 and 2 mM MgCl2). Forty cycles of denaturation (94°C for 30 seconds), annealing (see

ED

temperatures in Table 1), and polymerisation (72°C for 30 seconds) were regularly done. For semi-quantitative PCR, one to five microlitres of the reverse transcription reaction (diluted

PT

five-fold) were used depending on the gene target to be amplified. In the case of semi-

CE

quantitative PCR, targets were genes chosen according to the presence/absence of introns, in other words, they were intron-less and intron-containing genes, respectively (as denoted in

AC

Table 1). In the latter case, one of the primers (the forward or the reverse) was designed to amplify specifically the intron-containing sequence, unless otherwise stated. When no amplicon was obtained after PCR, a second PCR using intron-independent primers was performed to ensure that the absence of amplification was not due to a false negative result. Additionally, intron-containing genes were chosen according to the presence or absence of a PTC (PTC-containing and PTC-less genes, respectively, as denoted in Table 1). PTC stands for premature termination codon.

ACCEPTED MANUSCRIPT DNAs were cut at 37°C for 4 hours in 30-µl volumes, containing 1 U of the chosen restriction enzyme(s) and the reaction buffer recommended in each case by the supplier (Roche). DNA inserts were ligated to plasmid pL4440 (kindly provided by Andrew Fire, Carnegie Institution of Washington, Baltimore, Maryland, USA) at room temperature overnight in 10-µl volumes, containing 3 U of T4 Ligase (Promega) and the reaction buffer

SC RI PT

provided by the supplier (30 mM Tris-HCl pH 7.8, 10 mM MgCl2, 10 mM DTT, 1 mM ATP and 5% PEG).

DNAs were electrophoresed onto TBE-agarose gels using TBE running buffer (89 mM Tris-Borate and 2 mM EDTA, pH 8.3). For routine electrophoresis, 1% low EEO agarose was

NU

employed. For DNA quantifications, 3% low melting agarose was employed and equal loads

MA

for each well applied. 2.4. Bioinformatics

ED

Gene searches were done by BLAST using human sequences as bait against

PT

ParameciumDB, unless otherwise stated. Sequences were retrieved from public databases: GenBank (http://www.ncbi.nlm.nih.gov/) and ParameciumDB (http://paramecium.cgm.cnrs-

were

done

CE

gif.fr/). See text and Table 1 for sequence accession numbers. Pairwise sequence alignments using

either

the

BLAST

(https://blast.ncbi.nlm.nih.gov/Blast.cgi or

AC

http://paramecium.cgm.cnrs-gif.fr/cgi/tool/blast)

the

and

Needle

(http://www.ebi.ac.uk/Tools/psa/emboss_needle/) tools. Multiple sequence alignment was done with MUSCLE tool (https://www.ebi.ac.uk/Tools/msa/muscle/). Protein domains were searched

using

sequence

similarity

by

Hidden

Markov

Models

at

Pfam

(http://pfam.sanger.ac.uk/search). Amplicon quantities (mRNA relative abundance) were estimated through the Quantity One 1-D Analysis software (BioRad) using images digitally taken from agarose gels after

ACCEPTED MANUSCRIPT semi-quantitative PCR. For accurate quantifications, images were taken avoiding pixel saturation. Quantifications were done subtracting the image background, considering the amplicon width, and taking the trace quantity value, using the above mentioned BioRad’s software. For each amplicon under study and sample, the trace quantity value was normalised against the trace quantity value of PtND7 amplicon (see Table 1) in the corresponding

SC RI PT

sample. For determining the change in the relative mRNA level, ratios were calculated by dividing the trace quantity values of silenced amplicons and non-silenced amplicons. In those cases where no amplification at the assayed conditions was obtained in the non-silenced sample, the presence/absence argument was assigned to the ratio instead of a numerical value.

NU

3. Results

MA

3.1. EJC orthologues in P. tetraurelia

Human eIF4A3 (HseIF4A3; GenBank accession number NP_055555.1) hit twelve

-3e-52). All these sequences harbour the same architectural domain as HseIF4A3 protein

PT

159

ED

sequences in ParameciumDB with a high alignment score (557-203) and low E-value (e-

namely, an N-terminal DEAD/DEAH box helicase domain (pfam 00270), and a C-terminal

CE

helicase domain (pfam 00271). However, only GSPATG00029794001 (henceforth PteIF4A3) was kept for silencing experiments, since it was the only sequence that contains five short

AC

conserved sequences that defines eIF4A3 and not eIF4A1 or eIF4A2 (see Fig. 1A and Li et al., 1999). PteIF4A3 has no ohnologues in the database. An ohnologue is a paralogue originating after a whole genome duplication. Further evidence to support the orthology found between PteIF4A3 and human eIF4A3 was provided by reciprocal BLAST analysis. Human MAGO (also known as MAGOH; GenBank accession number NP_002361.1) hit four sequences (GSPATG00037128001, GSPATG00014236001, GSPATG00035625001, and GSPATG00029413001) in ParameciumDB with a high alignment score (233-185) and

ACCEPTED MANUSCRIPT low E-value (5e-62-1e-47). All these putative MAGO sequences are ohnologues to each other, according to ParameciumDB. Only GSPATG00029413001 (henceforth PtMAGO-1) was kept for silencing experiments (see supplementary data for a multiple alignment of MAGO ohnologues). Human and Paramecium MAGO deduced amino acid sequences were aligned to each other (Fig. 1B) in order to identify those MAGO amino acid residues that, according to

SC RI PT

X-ray crystal structure analyses (Shi and Xu, 2003), interact with Y14. As shown in Fig. 1B, 11/15 of the interacting amino acid residues in human are identical in PtMAGO-1. Further evidence to support the orthology found between PtMAGO-1 and MAGOH was provided by reciprocal BLAST analysis.

NU

Human Y14 sequence (GenBank accession number NP_005096.1) hit four sequences in

MA

ParameciumDB with a high alignment score (120-106) and low E-value (6e-28-9e-24). These four sequences (GSPATG00005114001, GSPATG00027102001, GSPATG00017571001, and GSPATG00001994001) are ohnologues to each other, according to ParameciumDB. Human

ED

and Paramecium Y14 deduced amino acid sequences were aligned to each other (Fig. 1C) in

PT

order to identify those Y14 amino residues that, according to X-ray crystal structure analyses (Shi and Xu, 2003), interact with MAGO. As shown in Fig. 1C, 15/19 of the interacting

CE

amino acid residues in human are identical in Y14 ohnologues in Paramecium. Further

AC

evidence to support the orthology found between each identified PtY14 and human Y14 was provided by reciprocal BLAST analysis. Human CWC22 (GenBank accession number NP_065994.1) hit two sequences (GSPATG00003845001 and GSPATP00009855001) in ParameciumDB with a high alignment score (489-488) and low E-value (e-138-e-137). These two putative CWC22 sequences

are

ohnologues

to

each

other,

according

to

ParameciumDB.

Only

GSPATG00003845001 (henceforth PtCWC22-1) was kept for silencing experiments. Human and Paramecium CWC22 deduced amino acid sequences were aligned to each other (Fig. 1D)

ACCEPTED MANUSCRIPT in order to identify those CWC22 amino acid residues that, according to X-ray crystal structure analyses (Buchwald et al., 2013), interact with eIF4A3. As shown in Fig. 1D, 9/12 of the interacting amino acid residues in human are identical in CWC22 ohnologues in Paramecium. Further evidence to support the orthology found between each identified

SC RI PT

PtCWC22 and human CWC22 was provided by reciprocal BLAST analysis. Finally, human MLN51 sequence hit no sequences in ParameciumDB. Lack of MLN51 was also confirmed by the absence of hits when MLN51 from other organisms were used as baits for BLAST searches.

NU

3.2. eIF4A3 and CWC22 silencing in P. tetraurelia

When silencing of PteIF4A3 or PtCWC22-1 were carried out, the paramecia were found

ED

and the same results were obtained.

MA

to be dead after the second day of silencing. These experiments were repeated one more time,

3.3. MAGO silencing in P. tetraurelia

PT

The silencing of PtMAGO-1 was done twice in order to have two biological replicas.

CE

After the third day of silencing, cell numbers were lower: there was about 15% less cells in silenced cultures compared to control (non-silenced) cultures.

AC

3.4. Change in relative mRNAs levels after MAGO silencing After the third day of the experiment, relative mRNA levels were obtained by semiquantitative PCR of a set of genes in both silenced and in non-silenced cultures (Table 1). With the obtained relative mRNA levels, changes (ratios between silenced and non-silenced cultures) were calculated. Thereafter, it was estimated that there was about 28% less PtMAGO-1 mRNA in silenced cultures compared to non-silenced ones, indicating that PtMAGO-1 was knocked-down successfully. Since the purpose was to evaluate the impact on

ACCEPTED MANUSCRIPT splicing of MAGO defective paramecia, mRNA targets (amplicon types, as named in Table 1) were chosen among intron-containing and intron-less genes. In the case of intron-containing genes, PCR primers were specifically chosen to amplify an intron sequence. If the EJC was indeed implicated in splicing in paramecia, the same mRNA levels and changes in mRNA levels would be expected in MAGO defective paramecia for intron-less targets and intron-

SC RI PT

containing targets, respectively, when compared to the control. Additionally, introncontaining targets were chosen among PTC-containing and PTC-less genes (as denoted in Table 1). PTC-containing genes are those in which intron retention lead to the appearance of a premature termination codon in its un-spliced mRNA. These un-spliced and PTC-containing

NU

mRNAs are supposed to be the subject of degradation by NMD (reviewed in Hug, Longman and Cáceres, 2016). Conversely, PTC-less genes are those in which intron retention does not

MA

lead to the appearance of a PTC in its un-spliced mRNA, and consequently they are supposed not to be the subject of degradation by NMD. In some organisms NMD and EJC are linked

ED

(reviewed in Hug, Longman and Cáceres, 2016). However, NMD and EJC are not linked in ciliates (Contreras et al., 2014; Tian et al., 2017). This implies that MAGO knock-down will

PT

not necessarily reproduce NMD knock-down. Therefore, we expect the mRNA levels of PTC-

CE

containing targets in MAGO and NMD defective paramecia to not necessarily match one another. PTC-containing targets were included in the analyses to ensure that phenotypes

AC

obtained after MAGO knock-down were not due to NMD indirect knock-down. Results (Fig. 2) were grouped into three categories according to obtained ratios (from now on, ratio values, abbreviated as R, will be expressed as mean±standard deviation): 1) ratio≈1 corresponded to targets with no change in relative mRNA level in silenced paramecia; 2) ratio>1 (or presence/absence) corresponded to targets with increased relative mRNA level in silenced paramecia (up-regulated targets, as denoted in Fig. 2); and 3) ratio<1 corresponded

ACCEPTED MANUSCRIPT to targets with decreased relative mRNA level in silenced paramecia (down-regulated targets, as denoted in Fig. 2). Three genes gave results belonging to category 1 namely, PtMORN (R=1.01±0.02), PtTUBULIN-1 (R=1.05±0.05), and PtFERROCHELATASE (R=1.07±0.08). The two formers

SC RI PT

are intron-less genes, while the latter is an intron- and PTC-containing gene (see Table 1). Despite PtFERROCHELATASE being an intron-containing gene, no change in its mRNA level was observed in MAGO defective paramecia.

Six genes gave results belonging to the category 2 namely, PtUNC (presence/absence),

NU

PtUPF2 (presence/absence), PtSNARE (R=1.51±0.08), PtTUBULIN-2 (R=2.61±0.72), PtCOPINE-1 (presence/absence) and PtCOPINE-2 (R=1.49±0.02). All these are intron-

Note

that

PtUNC,

PtUPF2,

MA

containing genes, of which only the two latter target introns of PTC-less type (see Table 1). and

PtCOPINE-1

gave

presence/absence

results

ED

(electrophoretograms in Fig. 2). PtUNC gave three different amplicons in MAGO defective paramecia. The large-sized amplicon was estimated to be barely above 500 bp, a size similar

PT

to the length of a three-introns retention (514 bp). The medium-sized amplicon was estimated

CE

to be around 500 bp, a size similar to the length of a two-introns retention (491 or 492 bp). The small-sized amplicon was estimated to be barely below 500 bp, a size similar to the

AC

length of a one-intron retention (469 bp). PtUPF2 gave a single amplicon, whose estimated size (175 bp) was similar to the length of a two-introns retention (181 bp). Finally, PtCOPINE-1 gave a single amplicon, whose estimated size (175 bp) was similar to the length of the intron retention (181 bp). The absence of amplifications of PtUNC, PtUPF2, and PtCOPINE-1 on non-silenced samples was not due to false negative results because the second PCR reactions, targeting these three genes by means of intron-independent primers, produced the expected amplicons on both silenced and non-silenced samples (results not shown).

ACCEPTED MANUSCRIPT Three genes, all of them intron-containing genes, gave results belonging to the category 3 namely, PtXRN1-1 (R=0.74±0.001), PtXRN1-2 (R=0.69±0.07 and 0.4±0.06), and PtENY (R=0.79±0.03). Note that the single intron of PtXRN1-1 and the second intron of PtXRN1-2 produce PTC-containing type amplicons (see Table 1).

SC RI PT

4. Discussion Splicing is the removal of the intervening bases from mRNAs that are transcribed from intron-containing genes. The splicing reaction is mediated by different protein complexes, collectively called the splicing machinery, that bind to the translating and still un-spliced

NU

mRNA. A member of these splicing complexes is the EJC, which deposits onto the mRNA 20-24 bases upstream of exon-exon junctions (Boehm and Gehring, 2016; Le Hir, Saulière

MA

and Wang, 2016; Woodward et al., 2017). Four proteins constitute the EJC core in human: eIF4A3, MAGO, Y14, and MLN51. The complex made of mRNA, eIF4A3, Y14, and MAGO

ED

constitutes the pre-EJC, which is assembled before exon ligation in the nucleus (Gehring et al., 2009). MNL51 binds to this pre-EJC once the mRNA is exported into the cytoplasm

PT

(Gehring et al., 2009). In silico, we have identified in the ciliate P. tetraurelia gene

CE

orthologues coding for each of the EJC proteins, except that coding for MLN51. Likewise, eIF4A3, MAGO, and Y14 but not MLN51 have been identified in the ciliate Tetrahymena

AC

thermophila (Tian et al., 2017). Identified genes in paramecia possess, in silico, those amino acids that are key for their different functions consequently, we took for granted that these protein-coding genes would perform their functions as EJC members. Since data indicate that MLN51 would be dispensable for the EJC function (Gehring et al., 2009), we assume that these two ciliates would assemble an EJC without MLN51. PteIF4A3 knock-down led to paramecia death. Since PteIF4A3 is a single copy gene, its function cannot be replaced by any other gene paralogue or ohnologue. Note that paramecia

ACCEPTED MANUSCRIPT have undergone three successive rounds of whole genome duplications (WGD), though not all gene ohnologues have been retained in its genome after WGD rounds (Aury et al., 2006); PteIF4A3 is an example of a gene whose ohnologues has been lost after WGD. The observed lethal phenotype and the absence of ohnologues indicates that PteIF4A3 is essential for paramecia. Similarly, the orthologue of eIF4A3 in Saccharomyces cerevisiae, named FAL1,

SC RI PT

is an essential gene for this yeast, since a fal1 null mutant produces a lethal phenotype (Kressler et al., 1997). Lethality of the deletion of FAL1 in S. cerevisiae is due to an 18S rRNA biogenesis defect. Whether lethality of eIF4A3 in paramecia is due to defective 18S rRNA biogenesis and/or splicing must be further investigated. However, our data would

NU

indicate that eIF4A3 lethality in paramecia is at least related to splicing defects. Firstly, because CWC22 knock-down also results in lethality and secondly, because CWC22 directs

MA

eIF4A3 towards splicing (Alexandrov et al., 2012; Steckelberg et al., 2012). PtMAGO-1 knock-down did not lead to cell death but to a decrease in paramecia

ED

number. Likewise, knock-out of MAGO orthologue in T. thermophila, named MAG1, did not

PT

lead to cell death (Tian et al., 2017). MAG1 is a single gene in T. thermophila, while PtMAGO-1 belongs to a multicopy gene family in paramecia. The absence of lethality after

CE

PtMAGO-1 silencing in paramecia is likely due to the fact that its function could be replaced

AC

by another ohnologue(s). The reason by which paramecia have kept the MAGO gene redundancy after WGD is unknown. However, it is known that gene dosage constraints have been the most important driving force for maintaining gene ohnology after WGD in paramecia (Aury et al, 2016). Furthermore, it is known that MAGO and Y14 have co-evolved due to their obligate dimerization mode of function (Gong, Zhao and He, 2014). Notably, MAGO and Y14 have maintained the same gene copy number (four ohnologues) in paramecia. Overall, data would indicate that gene dosage could play some important role in

ACCEPTED MANUSCRIPT maintaining MAGO/Y14 ohnology. However, whether gene dosage has played an evolutionary role for maintaining MAGO/Y14 redundancy needs additional analysis. Our results suggest that MAGO dysregulation leads to changes in the normal levels of miss-splicing products associated with many tiny introns in paramecia. This in agreement

SC RI PT

with the observation that EJC dysregulation in Drosophila and human lead to changes in the normal levels of miss-splicing products associated with genes harbouring long introns (Ashton et al., 2010; Roignant and Treisman, 2010; Wang, Murigneux and Le Hir, 2014). The main types of miss-splicing products are: intron retention (IR), exon skipping (ES), alternative 3′ or 5′ splice sites, and mutually exclusive events (see Ast, 2004 for a review).

NU

Due to the mechanistic differences between splicing of long and short introns, the variety of

MA

mRNA products obtained following miss-splicing is greatly constrained by intron length. ES is frequently observed due to miss-splicing of long introns, while IR is frequently observed due to miss-splicing of short introns. Paramecia tiny introns are extremely particular because

ED

IR but no ES has been observed following miss-splicing (Jaillon et al., 2008). The changes in

PT

IR levels after PtMAGO-1 down-regulation followed two different trends, depending on the targeted gene: a few genes increased their normal IR level, while some others decreased their

CE

normal IR level. In the former group of genes, three of them (PtUNC, PtCOPINE-1 and

AC

PtUPF2) gave presence/absence results. Both PtUNC and PtCOPINE-1 have one targeted intron whose length is multiple of 3 bp (the so-called 3n introns). The vast majority of 3n introns are constitutively spliced or rarely retained in paramecia (Jaillon et al., 2008). Intron length along with constitutive splicing would explain that PtUNC and PtCOPINE-1 gave no amplifications in non-silenced cells. PtUPF2 has two introns, but neither is 3n in length. According to primer positioning in PtUPF2, concomitant splicing of these two introns must eventually occur to detect an amplification product. Perhaps, the low occurrence of the

ACCEPTED MANUSCRIPT concomitant splicing of these two introns would explain that PtUPF2 gave no amplification in non-silenced cells. In human, eIF4A3 knock-down led to increases or decreases in IR in more than 100 genes (Wang, Murigneux and Le Hir, 2014). Consequently, the observation made in

SC RI PT

paramecia is akin to that made in humans: EJC affects IR only in a particular set of genes. We cannot draw any correlation between the observed IR trends, independently if they are increased, presence/absence, or decreased, and intron features in paramecia, such as the presence/absence of PTC, the size (the analysed introns are even in size) or their 5'- and 3'splice site composition (the analysed introns turned out to be uninformative for this purpose,

NU

see also Jaillon et al., 2008). The fact that in paramecia not all introns are affected or are

MA

differentially affected (up- or down-regulated) by PtMAGO-1 knock-down indicates that splicing control by EJC may have a random component. Despite this, we suggest that paramecia need the fine-tuning activity of the EJC to avoid abnormal IR, the only event of

ED

miss-splicing detected in these microorganisms. An example of this fine-tuning activity was

PT

observed when analysing the splicing of the three introns present in PtUNC. We reported that the un-silenced paramecia removed these three introns concomitantly, while MAGO defective

CE

paramecia displayed different miss-splicing events namely, full and partial IR.

AC

Our results also suggest that IR fine-tuning activity by PtMAGO-1 is conclusive. On the one hand, we observed that PtMAGO-1 down-regulation did not affect the relative mRNA level of intron-less genes, as expected. Considering that the EJC binds essentially to introncontaining mRNAs, it would be surprising if MAGO down-regulation affected intron-less mRNA levels. On the other hand, we observed that PtMAGO-1 down-regulation affected basal IR levels. Noticeably, changes in basal IR levels were true for all the analysed introns except that of PtFERROCHELATASE. Note that PtFERROCHELATASE mRNA is degraded by the NMD when its single intron is retained because it causes the appearance of PTC-

ACCEPTED MANUSCRIPT containing mRNA. Indeed, we and others (Contreras et al., 2014; Jaillon et al., 2008) have observed that the retention of the single intron of PtFERROCHELATASE increased after knocking-down different NMD factors, such as UPF1, UPF2, or UPF3. The uneven behaviour of PtFERROCHELATASE intron in EJC (PtMAGO-1 knock-down in this report) and NMD defective paramecia (Contreras et al., 2014; Jaillon et al., 2008) suggests that these two types

SC RI PT

of machinery are not linked in this ciliate. Certainly, if there was an EJC/NMD linkage, the EJC defects would lead to NMD defects, and both would give rise to an even behaviour for PTC-containing mRNAs (for instance, PtFERROCHELATASE intron retention). The observation that PtXRN1-1 and PtXRN1-2 are down-regulated in MAGO defective cells when

NU

their PTC introns are retained is also incompatible with an EJC/NMD linkage. These two data further reinforce the idea that NMD and EJC are not linked in ciliates, as we (Contreras et al.,

MA

2014) and others (Tian et al., 2017) previously claimed.

ED

5. Conclusions

We suggest that a fine-tuning expression of EJC genes is required for the steady

PT

removal of tiny introns. This report widens the knowledge on the EJC and its relationship to

CE

splicing control since, taking also into consideration previously published results (AshtonBeaucage et al., 2010; Ashton-Beaucage and Therrien, 2011; Roignant and Treisman, 2010;

AC

Wang, Murigneux and Le Hir, 2014), we can suggest that the EJC controls splicing independently of the intron size. Funding This work was supported by grants to EV from Ministerio de Ciencia e InnovaciónFEDER (BFU2009-10393), and by scholarships to JC and VB, from Ministerio de Educación (Spain), and to LM from the ERASMUS+ Program (European Community).

ACCEPTED MANUSCRIPT Conflict of interest All the authors declare they do not have any conflict of interest. Contributors Julia Contreras, Victoria Begley and Laura Marsella performed the experiments as well as

SC RI PT

revised the manuscript. Eduardo Villalobo design the study, write and revised the manuscript. All the authors approved the final article. Appendix A. Supplementary Data

NU

Supplementary data to this article can be found online at (TO BE PROVIDED BY THE

AC

CE

PT

ED

MA

JOURNAL)

ACCEPTED MANUSCRIPT References Alexandrov, A., Colognori, D., Shu, M.D., Steitz, J.A., 2012. Human spliceosomal protein CWC22 plays a role in coupling splicing to exon junction complex deposition and nonsense-mediated decay. Proc. Natl. Acad. Sci. USA 26, 21313-21318. doi:

SC RI PT

10.1073/pnas.1219725110. Ashton-Beaucage, D., Therrien, M., 2011. The exon junction complex: a splicing factor for long intron containing transcripts?. Fly 5, 224-233. doi: 10.4161/fly.5.3.15569. Ashton-Beaucage, D., Udell, C.M., Lavoie, H., Baril, C., Lefrançois, M., Chagnon, P.,

NU

Gendron, P., Caron-Lizotte, O., Bonneil, E., Thibault, P., Therrien, M., 2010. The exon junction complex controls the splicing of MAPK and other long intron-containing

MA

transcripts in Drosophila. Cell 143, 251-262. doi: 10.1016/j.cell.2010.09.014.

doi:10.1038/nrg1451.

ED

Ast, G., 2004. How did alternative splicing evolve?. Nature Rev. Genetics 5, 773-782.

PT

Aury, J.M., Jaillon, O., Duret, L., Noel, B., Jubin, C., Porcel, B.M., Ségurens, B., Daubin, V.,

CE

Anthouard, V., Aiach, N., Arnaiz, O., Billaut, A., Beisson, J., Blanc, I., Bouhouche, K., Câmara, F., Duharcourt, S., Guigo, R., Gogendeau, D., Katinka, M., Keller, A.M.,

AC

Kissmehl, R., Klotz, C., Koll, F., Le Mouël, A., Lepère, G., Malinsky, S., Nowacki, M., Nowak. J.K., Plattner, H., Poulain, J., Ruiz, F., Serrano, V., Zagulski, M., Dessen, P., Bétermier, M., Weissenbach, J., Scarpelli, C., Schächter, V., Sperling, L., Meyer, E., Cohen, J., Wincker, P., 2006. Global trends of whole-genome duplications revealed by the ciliate Paramecium tetraurelia. Nature 444, 171-178. doi:10.1038/nature05230. Berget, S.M., 1995. Exon recognition in vertebrate splicing. J. Biol. Chem. 270, 2411-2414. doi: 10.1074/jbc.270.6.2411.

ACCEPTED MANUSCRIPT Boehm, V., Gehring, N.H., 2016. Exon junction complexes: supervising the gene expression assembly line. Trends Genet. 32, 724-735. doi: 10.1016/j.tig.2016.09.003. Buchwald, G., Schuessler, S., Basquin, C., Le Hir, H., Conti, E., 2013. Crystal structure of the human eIF4AIII-CWC22 complex shows how a DEAD-box protein is inhibited by a domain.

Proc.

Natl.

Acad.

Sci.

USA

110,

E4611-8.

SC RI PT

MIF4G

10.1073/pnas.1314684110.

doi:

Contreras, J., Begley, V., Macias, S., Villalobo, E., 2014. An UPF3-based nonsense-mediated decay

in

Paramecium.

Res.

Microbiol.

84184-84186.

doi:

NU

10.1016/j.resmic.2014.10.008.

165,

De Conti, L., Baralle, M., Buratti, E., 2013. Exon and intron definition in pre-mRNA splicing.

MA

Wiley Interdiscip. Rev. RNA 4, 49-60. doi: 10.1002/wrna.1140.

ED

Galvani, A., Sperling, L., 2002. RNA interference by feeding in Paramecium. Trends Genet. 18, 11-12. doi: 10.1016/S0168-9525(01)02548-3.

PT

Gehring, N.H., Lamprinaki, S., Hentze, M.W., Kulozik, A.E., 2009. The hierarchy of exon-

CE

junction complex assembly by the spliceosome explains key features of mammalian nonsense-mediated

mRNA

decay.

PLoS

Biol.

7,

e1000120.

doi:

AC

10.1371/journal.pbio.1000120. Gong, P., Zhao, M., He, C., 2014. Slow co-evolution of the MAGO and Y14 protein families is required for the maintenance of their obligate heterodimerization mode. PLoS One 9, e84842. doi: 10.1371/journal.pone.0084842. Hug, N., Longman, D., Cáceres, J.F., 2016. Mechanism and regulation of the nonsensemediated decay pathway. Nucleic Acids Res. 44, 1483-1495. doi: 10.1093/nar/gkw010.

ACCEPTED MANUSCRIPT Inoue, H., Nojima, H., Okayama, H., 1990. High efficiency transformation of Escherichia coli with plasmids. Gene 96, 23-28. Jaillon, O., Bouhouche, K., Gout, J.F., Aury, J.M., Noel, B., Saudemont, B., Nowacki, M., Serrano, V., Porcel, B.M., Ségurens, B., Le Mouël A., Lepère, G., Schächter, V.,

SC RI PT

Bétermier, M., Cohen, J., Wincker, P., Sperling, L., Duret, L., Meyer, E., 2008. Translational control of intron splicing in eukaryotes. Nature 451, 359-362. doi: 10.1038/nature06495.

Kressler, D., de la Cruz, J., Rojo, M., Linder P., 1997. Fal1p is an essential DEAD-box

NU

protein involved in 40S-ribosomal-subunit biogenesis in Saccharomyces cerevisiae. Mol. Cell. Biol. 17, 7283-7294.

MA

Le Hir, H., Saulière, J., Wang, Z., 2016. The exon junction complex as a node of post-

ED

transcriptional networks. Nat. Rev. Mol. Cell. Biol. 17, 41-54. doi: 10.1038/nrm.2015.7. Li, Q., Imataka, H., Morino, S., Rogers, G.W., Richter-Cook, N.J., Merrick, W.C.,

PT

Sonenberg, N., 1999. Eukaryotic translation initiation factor 4AIII (eIF4AIII) is

CE

functionally distinct from eIF4AI and eIF4AII. Am. Soc. Microbiol. 19, 7336–7346. Roignant, J.Y., Treisman, J.E., 2010. Exon junction complex subunits are required to splice

AC

Drosophila MAP kinase, a large heterochromatic gene. Cell 143, 238-250. doi: 10.1016/j.cell.2010.09.036. Sambrook, J.F., Russell, D.W., 2001. Molecular Cloning: A Laboratory Manual, 3rd ed. Cold Spring Harbor Laboratory Press. Shi, H., Xu, R.M., 2003. Crystal structure of the Drosophila Mago nashi-Y14 complex. Genes Dev. 17, 971-976. doi: 10.1101/gad.260403.

ACCEPTED MANUSCRIPT Steckelberg, A.L., Boehm, V., Gromadzka, A.M., Gehring, N.H., 2012. CWC22 connects pre-mRNA splicing and exon junction complex assembly. Cell Rep. 27, 454-461. doi: 10.1016/j.celrep.2012.08.017. Tian, M., Yang, W., Zhang, J., Dang, H., Lu, X., Fu, C., Miao, W., 2017. Nonsense-mediated

SC RI PT

mRNA decay in Tetrahymena is EJC independent and requires a protozoa-specific nuclease. Nucleic Acids Res. 45, 6848-6863. doi: 10.1093/nar/gkx256. Wang, Z., Murigneux, V., Le Hir, H., 2014. Transcriptome-wide modulation of splicing by the exon junction complex. Genome Biol. 15, 551. doi: 10.1186/s13059-014-0551-7.

NU

Woodward, L.A., Mabin, J.W., Gangras, P., Singh, G., 2017. The exon junction complex: a lifelong guardian of mRNA fate. Wiley Interdiscip. Rev. RNA. 8. doi:

AC

CE

PT

ED

MA

10.1002/wrna.1411.

ACCEPTED MANUSCRIPT FIGURE LEGENDS Fig. 1. Multiple sequence alignments. Alignments were done by MUSCLE using deduced amino acid sequences corresponding to each gene, as they were retrieved from databases. Genes sequences were retrieved from GenBank (human sequences) or ParameciumDB (Paramecium sequences). Accession numbers are indicated in the text or Table 1. Symbols: a

SC RI PT

star (*) denotes identical amino acids; a dot (.) and a colon (:) denote conservative and semiconservative amino acids, respectively; a dash (-) is a gap introduced to improve the alignment. Ptet indicates Paramecium tetraurelia. (A) eIF4A3 alignment: bold and greyshaded amino acids indicate five short conserved sequences that define eIF4A3 sequences

NU

(see text for reference). (B) MAGO alignment: human sequence was aligned with the four ohnologues found in Paramecium. Bold and grey-shaded amino acids indicate those residues

MA

that interact with Y14 (see text for reference). (C) Y14 partial alignment: human sequence was aligned with the four ohnologues found in Paramecium. Bold and grey-shaded amino

ED

acids indicate those residues that interact with MAGO (see text for reference). (D) CWC22 partial alignment: human sequence was aligned with the two ohnologues found in

PT

Paramecium. Bold and grey-shaded amino acids indicate those residues that interact with

CE

eIF4A3 (see text for reference).

Fig. 2. Change in the relative mRNA abundance after MAGO silencing. Bars represent the

AC

fold change (arbitrary units) between silenced and unsilenced samples (ratio), except for down-regulated targets for which ratios are inversely proportional. The dashed black lines denote no change. Legends in each bar indicate given names of targeted genes, as stated in Table 1. Electrophoretograms represent those genes to which presence/absence results in silenced/unsilenced (leftmost and rightmost lanes, respectively) samples were obtained. Note that the PCR for PtUNC (image for the uppermost row) targets three introns (see Table 1).

ACCEPTED MANUSCRIPT Legends in each row indicate given names of targeted genes, as stated in Table 1. For each

AC

CE

PT

ED

MA

NU

SC RI PT

experiment n=2.

ACCEPTED MANUSCRIPT Table 1 PCR amplified genes. Gene names have been assigned by homology. Detailed PCR cycling conditions are shown in M&M (only annealing temperatures are indicated here). In cartoons, genes are indicated by boxes, introns by black-shadowed boxes, and primers by arrows (not to scale

T P

I R

of actual sizes). Right-hand and left-hand arrowheads correspond respectively to forward and reverse primers. Amplicon type as they are described in the text. Intron sizes from left-most to right-most introns; this information is not applicable (NA) to the first three listed genes. Accession number

PCR primers (forward/reverse)

Annealing

Given name

(5’-3’)

(oC)

Genes used for silencing experiments GSPATG00029794001

GATAAGCAGAATTCATAAGAGAAT

PteIF4A3

GCTCTCTTGAATTCGATAAAATCA

D E

57

GSPATG00029413001 PtMAGO-1

T P E

ATGCAAAATAATTAAACCGATGAGT

C S U

Amplicon type

M

N A

Amplicon/intron sizes

Gene cartoon

(bp)

Locations of PCR primers

NA

351/NA

57

NA

292/NA

53

NA

333/NA

50

NA

100/NA

CATTTTGAATCGTTTTCTCTTTTC

C C

GSPATG00003845001

ATAGGGCCCATCCAGGAGTTATTCCTGAA

PtCWC22-1

TATGAGCTCATAATCAACACTGCTTTAA

A

Genes used for mRNA quantifications GSPATG00029413001 PtMAGO-1

TGGAAATTAAATTAGGAAATGCA ACTCTTAAGCCATCTGGGTCT

ACCEPTED MANUSCRIPT GSPATG00002403001 PtND7

GCAGCCATGGGTATTTCCA TATAAGGATCAAAGTACCCTGA

GSPATG00017017001

CGTTTATGTGTATGATGATTCATG

PtMORN

CATCACTTCATTGTTTGTTATCAAT

51

Intron-less

135/NA

55

Intron-less

166/NA

T P

I R

Table 1. continued GSPATG00020765001

GCTGAAATCACAAACTCAGCCTT

PtTUBULIN-1

TGGGGACAACATCACCTCTGTAA

GSPATG00000033001 PtFERROCHELATASE GSPATG00022169001

GTGATCTCTAACAATATAGGCTAG

61

59

AAATTCCTTGGACCATTGTCCAT ATGTAATAAGTGACTCCTGCA

A M

217/24

Intron-containing

D E

CTTGACGCTTTCAATAAGTATTAC

PT

TTGAATTGAAAACAGTAGGTGC

U N

Intron-containing

112/NA

PTC-containing

55

PtUNC

SC

Intron-less

55

514/22, 23, 24

PTC-containing NA

160/NA

TCCCATGTGTTAGTTGAATTGG GSPATG00017015001 PtUPF2

E C

GTAATATTATAGATAAATAGTAG

AC

CTAAGGCATTATTAAATGAAATG GGCCATGCTTCTGATATCTCTT

Intron-containing 49

60

PTC-containing NA

181/23, 29

100/NA

ACAGATTCAACTAAGCCAATAACTTG GSPATG00012472001

GGAATTAAACAATGTGAAGCT

PtSNARE

CTAGAGTCGAACATCAAAATAC

51

Intron-containing PTC-containing

120/22

ACCEPTED MANUSCRIPT GSPATG00021912001 PtTUBULIN-2 GSPATG00011681001 PtXRN1-1

TGCTGAAATCACAAACTCAGCCC

Intron-containing

TGGGGACAACATCACCTCTGTAG

55

TGTATTATATATTCAAATATTAAG

125/22

PTC-containing Intron-containing

55

TCATATGGACACTGTTAGGAT

169/23

GTAATATTTAAATTCATTGAATGAG

I R

Intron-containing 53

120/25

TGCAGAATCCATATGTATATCT GSPATG00026799001

PTC-containing

C S U

AGGAATCCAAGATTGCTGACC

Intron-containing

PtXRN1-2

57 CTCTAGATCTAAGAATTTTTGGTAC

GSPATG00002949001 PtENY

PtCOPINE-1

TATGACTTCTTTACAGTAATC

50

D E

PT

TGAAGTTATTTCTAATTGGTGG

E C

CATTGGCTGGGCAACAAGAAC TTATGGTTACCTATGCTTCTC GSPATG00026056001 PtCOPINE-2

N A

AGATTCCAGGTAATCAAAATG

GTACAGAATATTTATATATACTAG

AC

GTATAGTAGATTATTATATGCTAG TGAAGTTATTTCTAATTGGTGG

149/25

PTC-less

Table 1. continued GSPATG00026237001

T P

PTC-containing

51

53

51

Intron-containing

M

PTC-less

Intron-containing PTC-less NA Intron-containing PTC-less

109/24

141/24

186/NA

141/24

ACCEPTED MANUSCRIPT

Data Statement

AC

CE

PT

ED

MA

NU

SC RI PT

Our data are fully available

ACCEPTED MANUSCRIPT Abbreviations list EJC: exon junction complex PTC: premature termination codon

SC RI PT

NMD: nonsense-mediated decay IR: intron retention BLAST: basic local alignment search tool

EDTA: ethylenediaminetetraacetic acid

MA

EEO: electroendosmosis

NU

dNTPs: deoxynucleotide triphosphates

ED

ES: exon skipping

PT

LB: Luria-Bertani

MUSCLE: multiple sequence comparison by log-expectation

CE

PCR: polymerase chain reaction

AC

PEG: polyethylene glycol Pfam: protein families TBE: tris-borate-EDTA

WGD: whole genome duplication

ACCEPTED MANUSCRIPT Highlights P. tetraurelia likely assembles an active EJC with only eIF4A3, MAGO and Y14.



Silencing of eIF4A3 or CWC22 causes lethality.



Silencing of MAGO causes changes in the levels of some intron-containing mRNAs.



A fine-tuning expression of EJC genes is required for the removal of tiny introns.



The EJC controls splicing independently of the intron size.

AC

CE

PT

ED

MA

NU

SC RI PT



Figure 1ac

Figure 1D

Figure 2