Characterization and In Vivo Functional Analysis of the Schizosaccharomyces pombe ICLN Gene

Abstract

During the early steps of snRNP biogenesis, the survival motor neuron (SMN) complex acts together with the methylosome, an entity formed by the pICln protein, WD45, and the PRMT5 methyltransferase. To expand our understanding of the functional relationship between pICln and SMN in vivo, we performed a genetic analysis of an uncharacterized Schizosaccharomyces pombe pICln homolog. Although not essential, the S. pombe ICln (SpICln) protein is important for optimal yeast cell growth. The human ICLN gene complements the Δicln slow-growth phenotype, demonstrating that the identified SpICln sequence is the bona fide human homolog. Consistent with the role of human pICln inferred from in vitro experiments, we found that the SpICln protein is required for optimal production of the spliceosomal snRNPs and for efficient splicing in vivo. Genetic interaction approaches further demonstrate that modulation of ICln activity is unable to compensate for growth defects of SMN-deficient cells. Using a genome-wide approach and reverse transcription (RT)-PCR validation tests, we also show that splicing is differentially altered in Δicln cells. Our data are consistent with the notion that splice site selection and spliceosome kinetics are highly dependent on the concentration of core spliceosomal components.

INTRODUCTION

In eukaryotes, an essential step in the production of functional mRNAs is the spliceosome-mediated removal of introns from pre-mRNAs. This machinery is composed of 5 spliceosomal snRNPs and additional non-snRNP-associated factors (1, 2). The biogenesis of these snRNPs is an ordered multistep process. After transcription, the m⁷G-capped snRNAs are exported to the cytoplasm, where they bind to the seven Sm proteins SmB/B′, SmD1, SmD2, SmD3, SmE, SmF, and SmG. Accurate assembly of the Sm core domain is required for subsequent m₃G cap formation, which is followed by the active transport of snRNPs to the nucleus (3).

Although formation of the snRNP core can occur spontaneously in vitro, the process is highly regulated in vivo, and the survival motor neuron (SMN) protein, encoded by the survival motor neuron (SMN1) gene, is a major player in these preliminary assembly steps (4–6). Mutations in SMN1 cause the autosomal recessive disease spinal muscular atrophy (SMA) (7). The SMN protein forms a stable complex with a group of proteins called gemins and is found in the cytoplasm, as well as the nuclei, of cells, where it is enriched within discrete bodies called Cajal bodies (8, 9). During the cytoplasmic step of snRNP biogenesis, the SMN complex interacts with the methylosome, a complex formed by the pICln and WD45 proteins and the PRMT5 methyltransferase (10–12). The methylosome recruits Sm proteins via the pICln subunit, and PRMT5 allows the symmetric dimethylation of arginines within the C tails of SmB, SmD1, and SmD3 (13, 14). The SMN complex further facilitates the loading of Sm proteins onto the snRNA, resulting in the formation of a basic snRNP particle (15, 16). In this process, pICln acts as an assembly chaperone and SMN acts as a catalyst, allowing ring closure of the Sm core protein complex on the snRNA (17). The recent crystal structures of pICln-Sm intermediates show that pICln acts as an Sm protein mimic, allowing Sm proteins to interact with each other even in the absence of RNA (18).

These studies indicate that pICln and SMN are critical regulators of snRNP assembly. However, several important questions regarding the relationship between SMN and pICln still remain unanswered. For example, dosage suppression tests should be performed to check whether overexpression of wild-type pICln can rescue SMN mutants. In order to study the functional relationship between pICln and SMN in vivo, we used Schizosaccharomyces pombe, which is a good model to study snRNP biogenesis and splicing, since its splicing machinery is closely related to mammalian splicing complexes both in composition and in sequence similarity (19, 20).

In this report, we describe the in vivo functional analysis of the uncharacterized fission yeast homolog of human pICln. While the S. pombe ICLN gene is not essential, it is critical for optimal growth and for efficient snRNP production and splicing. Using a genome-wide approach, we found that splicing is altered in Δicln cells, affecting specifically a subset of introns in which the polypyrimidine tract is located further upstream of the branch point and whose A/U content is decreased compared to unaffected introns. Genetic interaction tests also show that snRNP assembly and growth defects occurring in cells with an SMN mutant allele cannot be rescued by modulating the activity of ICln. Finally, we discuss a model in which reduced levels of snRNPs in fission yeast generate a block during the early stages of spliceosome assembly for a subset of pre-mRNAs.

MATERIALS AND METHODS

Yeast strains, media, and genetic methods.

The diploid S. pombe strain heterozygous for the null allele of SpICLN/SPAC1610.01 (h⁺/h⁺ ade6-M210/ade6-M216 ura4-D18/ura4-D18 leu1-32/leu1-32 SPAC1610.01/SPAC1610.01::KanMX4) (Bioneer Corporation, South Korea) was transformed with the sporulation-inducing plasmid pON177 (21) and plated on EMM2 medium containing appropriate supplements. Spores were dissected and germinated at 25°C on yeast extract-supplemented (YES) plates. Cells containing the icln-disrupted null allele (Δicln) were identified by replica plating to YES medium containing 100 μg/ml Geneticin (G-418). Correct homologous recombination of the disrupted SPAC1610.01 allele was checked by PCR amplification of genomic DNA. Standard methods were used for growth and genetic manipulation of S. pombe (22).

Strain FYAB12 carrying an N-terminally tagged green fluorescent protein (GFP)-SpICln allele was constructed by PCR-based one-step homologous recombination (23). The integrating cassette was amplified by PCR from plasmid pUra41nmt1-GFP, using a forward oligonucleotide carrying sequences corresponding to the 100 nucleotides upstream of the start codon of the SpICln gene and a reverse oligonucleotide carrying sequences corresponding to the 100 nucleotides of the N-terminal codons of the fission yeast ICLN gene (including the start codon).

Plasmid constructions.

A fragment encoding the fission yeast ICLN gene was amplified from the pTN-RC5 cDNA library (a gift from T. Nakamura, YGRC, Osaka, Japan). The human ICLN gene was amplified from plasmid pTZ-hICLN (a gift from U. Fischer, Würzburg, Germany). PCR products were cut with XmaI and BamHI and cloned into the S. pombe XmaI-BamHI-digested plasmid pREP41 (24) and XmaI-BamHI-cut pBluescript using standard methods. The resulting plasmids were named pSpICln, phuICln, and pBS-SpICln, respectively. The S. pombe ICLN gene was also amplified using forward and reverse oligonucleotides carrying restriction sites for SalI and BamHI, respectively. After digestion, the DNA fragment was purified and cloned between the SalI and BamHI sites of the pREP42-N-TAP vector (25), generating the pTAP-SpICln plasmid.

The various pGST-Sm plasmids containing the Sm coding sequences in frame with glutathione S-transferase (GST) were constructed using the Gateway system and a pGEX2T-based vector according to the manufacturer's instructions (Invitrogen). The DNA fragments were obtained by PCR amplification using adequate attB primers and the pTN-RC5 cDNA library. The S. pombe ICLN gene was inserted into the pEntry vector previously cut with SalI and NotI by using a SalI-NotI fragment obtained by digestion of pBS-SpICLN to generate the pDonor-SpICln vector. The plasmid carrying a His-tagged fission yeast ICLN gene was constructed by the Gateway system using the pET15-HIS₆ vector (Novagen). All pDonor constructs were sequenced prior to proceeding to LR recombination to obtain the GST-Sm fusion and the pHIS₆-SpICln clones. All constructs were confirmed by DNA sequence analysis.

To construct the pGFP-SmB and pGFP-SmD1 plasmids, the corresponding fission yeast Sm coding sequences were PCR amplified from the pGST-Sm plasmids using a forward oligonucleotide carrying a SalI restriction site and reverse oligonucleotides with BglII (for SmB) or BamHI (for SmD1) sites. After separation on agarose gels, DNA fragments were purified using the GeneClean procedure and transferred into the SalI-BamHI-cut pREP41-GFP-N vector (24). The cloning junctions and coding sequences were verified by sequencing.

Production of recombinant proteins and in vitro binding assays.

pGEX-2T and pGST-Sm constructs, as well as the pHIS₆-SpICln plasmid, were transformed in Escherichia coli and grown to an optical density at 600 nm (OD₆₀₀) of 0.5. Expression was induced by the addition of 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and the cells were grown for 4 h at 37°C. Recombinant GST and GST fusion proteins were purified by using glutathione-Sepharose beads (GE Healthcare) essentially as described previously (26). Soluble His₆-SpICln was batch purified under native conditions and eluted from the Ni²⁺ resin with 250 mM imidazole as described previously (27). The imidazole was removed, and the protein was concentrated using Centriprep 10 concentrators (Amicon).

For in vitro protein-protein interaction studies, 1 μg of purified GST fusion proteins or GST alone as a negative control was incubated with 1 μg of His₆-SpICln protein in 200 μl binding buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM MgCl₂, 1% Triton X-100, 100 μg of bovine serum albumin/ml). After 2 h of incubation at 4°C with constant rotation, the beads were pelleted and washed five times with 1 ml of binding buffer. The proteins were fractionated on SDS-polyacrylamide gels and transferred onto nitrocellulose membranes, and the GST fusion proteins were revealed by Ponceau staining. The bound His₆-SpICln fusion protein was revealed by Western analysis using an anti-His₆ antibody.

Preparation of extracts and native gel, Northern blotting, and primer extension experiments.

For native gel analysis of snRNPs and the sucrose gradient, extracts were prepared from cells (OD₆₀₀ = 0.6 to 0.8), washed with water, and resuspended to 1 g/ml in AGK400 buffer (10 mM HEPES-KOH, pH 7.9, 400 mM KCl, 1.5 mM MgCl₂, 0.5 mM dithiothreitol [DTT], 1× Complete protease inhibitors, and 10% glycerol) (28). After freezing in liquid nitrogen, the cells were ground to fine powder using a Freezer Mill 6770 grinder (Spex). After thawing on ice, the cells were centrifuged at 14,000 rpm for 10 min at 4°C, and the supernatant was recovered and spun at 55,000 rpm for 30 min at 4°C in a TLA-100.3 rotor (Beckman). The extract was then dialyzed for 2 h against buffer D (20 mM HEPES-KOH, pH 7.9, 0.2 mM EDTA, 100 mM KCl, 0.5 mM DTT, 1 mM phenylmethylsulfonyl fluoride [PMSF], 20% glycerol), and aliquots were stored at −80°C. For native gels, 30 μg of extracts was loaded on 4% acrylamide gels (80:1) and run in 25 mM Tris, 25 mM boric acid, 1 mM EDTA at 13 V for 16 h until bromophenol blue reached the bottom.

Total yeast RNA was purified from exponentially growing cells with Tri-Reagent (Sigma) according to the manufacturer's procedure. Primer extension and Northern blot analyses were performed as described previously (29, 30).

Coimmunoprecipitation experiments and sucrose gradients.

Coimmunoprecipitations were performed on extracts prepared from Δicln cells carrying pTAP-SpICln and pGFP-SmD1 or empty pGFP plasmids. Cells grown in 100 ml of EMM2 medium without Leu and Ura (EMM2 −Leu −Ura) to an OD₆₀₀ of 0.6 to 0.8 were centrifuged, and the pellet was washed with water. After resuspension in IPP150 calmodulin binding buffer (CBB150) (25) to 1 g/ml, the cells were disrupted with glass beads. Two-hundred-microliter soluble extracts were incubated with 20 μl of calmodulin binding protein resin (Stratagene) for 2 h at 4°C. After incubation, the beads were washed 5 times with CBB150 and resuspended in 30 μl of IPP150 calmodulin elution buffer (CEB150) (25). The eluted proteins were analyzed by SDS-PAGE and Western blotting using anti-PAP (P-1291; Sigma) and anti-GFP (Molecular Probes) antibodies.

Extracts were layered on 11 ml 5 to 20% (wt/vol) sucrose gradients in buffer A (50 mM Tris-HCl, pH 7.4, 25 mM NaCl, 5 mM MgCl₂). After sedimentation at 35,000 rpm for 14 h in an SW41 rotor at 4°C, 500-μl fractions were recovered and 35 μl of odd-numbered fractions was separated by SDS-PAGE and analyzed by Western blotting. The remainder of the fractions were pooled as indicated (see Fig. 2D), incubated with 20 μl calmodulin binding protein resin (Stratagene) or glutathione-Sepharose beads (GE Healthcare) as a control in CBB150 for 2 h at 4°C, and treated as described above.

FIG 2.

SpICln associates with Sm proteins in vitro and in vivo. (A) Glutathione-Sepharose beads and the indicated recombinant GST-Sm fusion proteins or GST alone was incubated with 1 μg of recombinant His₆-tagged SpICln protein. The beads were washed, and bound proteins were separated by SDS-PAGE and transferred onto nitrocellulose membranes. After S-Ponceau staining to view the GST fusions, the membrane was probed with an anti-His antibody to visualize the His-SpICln protein (bottom). (B) Coimmunoprecipitation of TAP-SpICln and GFP-SmD1 proteins. Extracts prepared from Δicln cells carrying the indicated plasmids were incubated with calmodulin binding protein resin. Bound proteins were separated by SDS-PAGE, transferred to nitrocellulose membranes, and blotted with anti-PAP and anti-GFP antibodies. Inputs (2%) are shown on the left. (C) Whole-cell extract prepared from Δicln cells carrying TAP-SpICln and GFP-SmD1 plasmids was separated on a 5 to 20% sucrose gradient. The direction of sedimentation (Sed) is top to bottom, as indicated by the arrow. After collection of the fractions, the odd-numbered fractions were separated on SDS-PAGE and immunoblotted to detect TAP-SpICln and GFP-SmD1 fusion proteins. (D) Fractions 3 to 5 from the sucrose gradient were pooled and incubated with calmodulin binding protein (CBP) resin or glutathione-Sepharose (GS) beads (as a control [Ctl]). The bound proteins were separated by SDS-PAGE and immunoblotted with anti-PAP and anti-GFP antibodies. Input represents 2% of the pooled fractions used in the binding assay.

Microarrays and reverse transcription (RT)-PCR analyses.

Total RNAs from two independent cultures of the Δicln and wild-type strains were hybridized to the tiling arrays. Probe labeling was performed with a GeneChip WT double-stranded cDNA synthesis kit (Affymetrix). Labeled probes were hybridized to GeneChip S. pombe Tiling 1.0FR arrays (Affymetrix) using the manufacturer's protocol. Scanning and data collection were done with a GeneChip Scanner 3000 7G (Affymetrix) and GeneChip operating software. Raw tiling array data were analyzed using the Tiling Analysis Software (TAS) package (Affymetrix) as described previously (31). Briefly, data from duplicate arrays were combined in control and treatment groups, quantile normalized, and analyzed by the two-sample normalization process. The TAS normalization process results in a binary archive (BAR) file containing the differences between control and treatment-normalized signal values (expressed in log₂ scale). In order to analyze differences between exon and intron intensities, a specific R software algorithm was developed to calculate average signal intensities in all S. pombe exon and intron areas from TAS-normalized result files. First, all the positions of genes containing at least one intron were extracted from the S. pombe database of the Affymetrix DAS server to generate a sublist of 4,611 discrete chromosomal areas. This step allowed us to define the peak positions usually identified by statistical tests and P value analyses. These positions were then used to calculate the difference between the mean of signal intensities for each S. pombe intron and that of the two adjacent exons.

For RT-PCR analyses, purified RNA was treated with RQ1 RNase-free DNase (Promega), and the first-strand cDNAs were synthesized using 5 μg of total RNA and pd(N)₆ random oligonucleotides with the First-Strand cDNA synthesis kit (GE Healthcare). For PCR analyses, 1/10 of the reaction mixture was amplified with GoTaq polymerase (Promega), and the cycle number was kept to a minimum to maintain linearity. Primer sequences and PCR regimens are available upon request. PCR products were separated on 1.5 to 2% agarose gels containing ethidium bromide and visualized under UV light.

Statistical analysis.

For statistical analyses, we used a specific function called pValueVariation that we previously developed with the R statistical language (31). This function returns the Wilcoxon test P value, allowing the comparison of nonparametric distributions between two groups defined by the function argument. The first group is defined from the beginning of the list to the argument (a ranked position from the entire intron list sorted by decreasing retention indexes). The second is set from the next argument position to the end of the list. The different P values were obtained by using the pValueVariation function with all the arguments chosen between the early beginning (start position 30) to the close end (30 before the ending) of the ranked list with an increment of 1. The P values obtained were plotted against the argument to draw the pValueVariation curves. The percentages of A/U content in the defined groups of introns were calculated after excluding the 5′ and 3′ splice sites and using introns at least 30 nucleotides in length.

Gene ontology and pathway analysis.

The lists of genes carrying retained introns were imported into PANTHER (http://www.pantherdb.org/), and the number of genes in each functional classification category was compared to the number of genes from the reference S. pombe genome in that category. The binomial test with Bonferroni correction was used to statistically determine overrepresentation of PANTHER classification categories (32).

Microarray data accession number.

The microarray data are accessible in the GEO database (accession number GSE41403).

RESULTS

Identification of an S. pombe ICln ortholog.

A fission yeast polypeptide sequence that shows 30% identity and 52% similarity to the amino acid sequence of the human pICln gene product over the entire protein sequence was identified from the S. pombe database. The SpICln gene (SPAC1610.01; expected value [E] = 3 × 10⁻⁵) contains no intron and encodes a protein containing 217 amino acids, with a predicted molecular mass of 24.9 kDa. Comparison of the S. pombe ICln protein sequence with the sequences of other eukaryotic homologs indicated a high degree of similarity (24 to 31% identical; 40 to 83% similar). As shown in Fig. 1A, the seven antiparallel β-strands (β1 to β7) and the α-helix forming the pleckstrin homology domain (18, 33) are found in ICln proteins from all organisms. Several strictly or highly conserved amino acid residues, described recently in the crystal structures of two pICln assembly intermediates (18), are also found in the S. pombe ICln polypeptide. For example, positions 50, 72, 82, 83, 156, and 160 (following S. pombe numbering) are invariably occupied by glycine, glycine, leucine, histidine, glutamine, and proline, respectively. The frequency of conserved hydrophilic amino acids (positions 56, 63, and 80) is comparable to that of conserved acidic residues (positions 134, 161, and 171). Remarkably, 4 out of 5 conserved residues of the β5 strand located at the pICln-SmD1 interface (18) are found in the SpICln protein. However, other highly and strictly conserved residues, such as positions 63, 74, 88, and 96, are not found in the S. pombe ICln sequence (Fig. 1A), suggesting structural accommodation or coevolution with interacting partners.

FIG 1.

The SpICln protein is a functional human ortholog and is not essential for viability. (A) Sequence alignment of pICln homologs, with gene identification (gi) numbers. The alignment was created using the PRALINE server (62). Identical amino acids among pICln proteins are shaded in black, similar residues are shaded in gray, and acidic domains are labeled. The secondary structures predicted by PSIPRED (position-specific iterated prediction of protein secondary structure) through the Ali2D server (University of Tübingen, Tübingen, Germany) are indicated below the sequences. The α-helix and the β₁- and β₇-strands (represented by arrows) correspond to the canonical pleckstrin homology domains (18, 33). Proposed highly (+) and strictly (*) conserved residues (18) are also shown. H. sapiens, Homo sapiens; M. musculus, Mus musculus; C. elegans, Caenorhabditis elegans; D. melanogaster, Drosophila melanogaster. (B) Tetrad analysis of diploid S. pombe cells with one deleted copy of ICLN (SPAC1610.01/SPAC1610.01::KanMX4). After sporulation, separated spores were grown on YES plates at 25°C for 5 days (the spores for each tetrad are arranged vertically). (C) Δicln cells have a growth defect and are temperature sensitive. Wild type (wt) and Δicln cells were grown in YES medium, and serial dilutions were spotted on YES plates, which were incubated at the indicated temperatures. (D) The growth defect of Δicln cells is complemented by the human ICln protein. Full-length cDNAs for the S. pombe and human ICLN genes were placed under the control of the nmt1+ promoter in a pREP41 plasmid. Transformants carrying the plasmids indicated on the right were grown in −Leu medium, and an equivalent number of cells were serially diluted, plated on −Leu plates, and incubated at 25°C for 5 days. (E) The indicated cells were transformed with the empty pREP41 vector, a plasmid encoding fission yeast ICLN, or a plasmid carrying the S. pombe SMN gene, and cultures of comparable density were serially diluted, spotted on EMM-Leu plates, and incubated at 25°C for 5 days. (F) Synthetic lethality between ICLN and SMN alleles. The Δicln::KanMX4 strain was crossed with the tdSMN::KanMX4 strain, and the resulting diploid was sporulated. Tetrads were dissected and incubated on YES medium at 25°C for 5 days. The segregation patterns (TT, tetratype; PD, parental ditype; NPD, nonparental ditype) of representative tetrads are shown. Nonviable spores are circled and are all Δicln tdSMN double mutant.

As a first step to analyze the function of the S. pombe ICln protein, the correct deletion of the SPAC1610.01 gene by the KanMX4 cassette was confirmed by PCR of genomic DNA from a heterozygous diploid (data not shown). Following sporulation, the resulting tetrads were dissected, and as shown in Fig. 1B, all four spores produced colonies at 25°C, demonstrating that haploid segregants containing the disrupted ICLN gene were viable and indicating that the S. pombe ICLN gene is not essential for viability. However, the haploid Δicln segregants displayed a noticeable growth defect, since they form small colonies with a growth rate more than 2-fold lower than that of the wild type when grown at 25°C, and they are temperature sensitive at 37°C (Fig. 1C). The growth defect could be complemented by reintroducing a plasmid containing the wild-type S. pombe ICLN gene, as well as by a plasmid containing the human ICLN gene (Fig. 1D), demonstrating that the identified fission yeast ICln protein is a bona fide homolog of the human ICln.

Given the intrinsic relationship of ICln and SMN proteins in mammals, we performed dosage suppression experiments to test whether overexpression of fission yeast ICLN compensates for the growth defect observed in cells carrying a temperature degron allele of SMN (tdSMN) (31). As shown in Fig. 1E (top), tdSMN cells do not grow better with higher levels of SpICln. Reciprocally, overexpression of the SpSMN gene does not complement the growth defect of Δicln cells (Fig. 1E, bottom). We further analyzed genetic interactions between ICLN and SMN by crossing the Δicln strain with the tdSMN strain and checked the viability of the resulting haploid segregants after sporulation. As shown in Fig. 1F, tetratypes resulted in three viable colonies, nonparental ditypes gave rise to two viable colonies, and four viable colonies were observed for parental ditypes. No double mutants with both Δicln and tdSMN alleles could be obtained, indicating that these genes have a synthetic negative relationship, thus confirming their functional connection.

Properties of the S. pombe ICln protein.

The human pICln protein interacts with the spliceosomal Sm proteins (10, 11, 34, 35). If the S. pombe ICln protein is a true functional ortholog, it would be expected to display equivalent interactions. To test this, in vitro protein binding assays under stringent conditions were performed with recombinant His₆-SpICln and GST-Sm fusion proteins. His-SpICLn bound specifically and strongly to immobilized GST-SmD1 and GST-SmD3 proteins (Fig. 2A, lanes 3 and 5) and extremely weakly with GST-SmB (lane 2), while no in vitro binding was detected with GST alone and with GST-SmD2, GST-SmE, GST-SmF, and GST-SmG (lanes 1, 4, 6, 7, and 8, respectively). The observed strong binding of ICln with SmD1 and SmD3 is consistent with previous studies showing that pICln interacts with the Sm domains of the human SmD1 and SmD3 proteins (10, 12, 18, 34).

To determine if Sm proteins associate with SpICln in vivo, immunoprecipitation experiments were performed using whole-cell extract prepared from Δicln cells carrying a TAP-SpICln construct (as the sole source of SpICln) and a GFP-SmD1 plasmid. The GFP-SmD1 fusion could be immunoprecipitated with the TAP-SpICln fusion protein on calmodulin beads, indicating that SpICln associates with Sm proteins in vivo (Fig. 2B). Because pICln has been previously shown to be part of the methylosome implicated in methylation of Sm proteins in human cells (5, 10, 12), sucrose gradient centrifugation experiments were performed using whole-cell extract prepared from Δicln cells carrying the TAP-SpICln and GFP-SmD1 plasmids. After centrifugation, fractions were collected, resolved by SDS-PAGE, and analyzed by Western blotting with antibodies against TAP and GFP tags. As shown in Fig. 2C, TAP-SpICln could be detected in a peak of approximately 6S comigrating with the GFP-SmD1 protein. To determine if the GFP-SmD1 is associated with TAP-SpICln in the 6S fractions, fractions 3 to 5 were pooled and incubated with either calmodulin beads or glutathione-Sepharose beads as a control (Fig. 2D). Western blotting with anti-PAP and anti-GFP antibodies revealed that the GFP-SmD1 fusion is associated with TAP-SpICln in the 6S peak. However, the larger complex corresponding to the human 20S complex containing pICln and the PRMT5 methyltransferase could not be detected by sucrose gradient sedimentation (Fig. 2C), and this could be due to its small amount or to its transient formation in fission yeast.

To characterize the subcellular localization of the SpICln protein, we constructed a strain carrying an endogenous GFP-SpICln fusion gene. The strain carrying this fusion gene as the sole source of ICln grows like a wild-type strain, implying that the GFP-SpICln fusion protein associates correctly with its partners during its function. As shown in Fig. S1A in the supplemental material, GFP-SpICln is found in the cytoplasm but is more concentrated in the nucleus. Such localization is similar to the subcellular localization of the human ICln, which is also predominantly nuclear (36). Altogether, these experiments demonstrate that the interaction properties and the subcellular localization of the ICln proteins are evolutionarily conserved from fission yeast to humans.

Many studies in higher eukaryotes have shown that nuclear import of the snRNPs is dependent on a bipartite nuclear localization signal composed of the Sm core complex and the snRNA m₃G cap structure (37, 38). The nuclear localization signal carried by the complex of Sm proteins appears to be represented by an evolutionarily conserved basic protuberance formed by the C-terminal extensions of the SmB, SmD1, and SmD3 proteins (39, 40). To test the effect of the S. pombe ICLN deletion on the localization of Sm proteins, we analyzed the subcellular localization of GFP-SmB and GFP-SmD1 fusion proteins. As shown in Fig. S1B in the supplemental material, expression of the constructs in wild-type and Δicln cells resulted in the synthesis of fusion proteins of predicted lengths, with the GFP-SmB fusion found at slightly higher levels in the wild-type background. Microscopic imaging of GFP fluorescence indicates that both GFP-SmB and GFP-SmD1 proteins are localized in the nucleus with a diffuse signal in the cytoplasm in wild-type and Δicln cells (see Fig. S1C). Quantification analysis showed that the cytoplasmic/nuclear signal ratios are equivalent in Δicln cells compared to the wild type, indicating that the deletion of the ICLN gene in S. pombe does not significantly alter the subcellular localization of the Sm fusion proteins.

The ICln protein is required for efficient snRNP production and splicing.

Based on the known role of ICln in the assembly of the Sm core complex, we tested whether ICln is essential for snRNP production in S. pombe in vivo. This was performed using Northern blot analysis of RNA extracted from wild-type and Δicln cells grown at 25°C. As shown in Fig. 3A, the levels of the U1, U2, U4, and U5 snRNAs clearly decrease in the Δicln cells, while the level of the S. pombe U3 snoRNA is unchanged and the amount of U6 is diminished only slightly. Quantitation of the gel after normalization using U3 snoRNA as a standard indicated that levels of snRNA present in the Δicln cells represented approximately 32% (for U5) to 90% (for U6) of wild-type levels (Fig. 3B).

FIG 3.

The levels of snRNPs are decreased in fission yeast Δicln cells. (A) RNA was extracted from Δicln and wild-type cells grown at 25°C, separated on a 6% polyacrylamide–8 M urea gel, and subjected to Northern analysis. Hybridization was performed using probes specific for the indicated RNAs. (B) Quantitation of the Northern blot shown in panel A by scanning densitometry of the RNAs present in Δicln and wild-type cells. The error bars indicate standard deviations. (C) Analysis of snRNPs in Δicln (Δ) and wild-type cells by native gel electrophoresis. Extracts were prepared from cells grown at 25°C, and aliquots were separated on 4% native gels. The RNA was subjected to Northern blot analysis and hybridized with probes for the indicated RNAs. *, the U2/U5/U6 postsplicing complex; ×, the U4/U6 di-snRNP. (D) Splicing is inhibited in Δicln cells. Total RNA was isolated from Δicln and wild-type cells and used for primer extension with a ³²P-labeled primer specific to the U6 gene. Pre, species corresponding to the U6 precursor that contains an intron; mature, the spliced U6 RNA.

To further determine the status of the snRNPs in Δicln cells under more physiological conditions, we analyzed snRNP complexes by native gel electrophoresis, followed by Northern blot analyses. Compared to the wild-type strain, the levels of U1, U2, U4/U6, and U5 snRNPs were all decreased in the Δicln strain, demonstrating that the SpICln protein is required for efficient formation and/or stability of the spliceosomal snRNPs (Fig. 3C). To determine whether ICln is required for pre-mRNA splicing in vivo, we tested the splicing efficiency of the intron-containing U6 snRNA precursors by primer extension (Fig. 3D). While the wild-type strain did not accumulate detectable pre-U6 snRNA, precursor accumulation was readily visible in Δicln cells, indicating that splicing is inhibited in these cells. Altogether, these studies demonstrate that ICln is required for optimal snRNP production and function in vivo.

Deletion of the IClN gene generates differential pre-mRNA splicing defects.

As mentioned above, we previously showed that limited amounts of spliceosomal snRNPs affect the splicing of a subset of introns in a strain carrying a tdSMN allele (31). In order to test whether differential splicing defects are also observed in Δicln cells, we measured the accumulation of introns in pre-mRNAs genome-wide using tiling microarrays. Total RNA was isolated from independent cultures of Δicln, as well as wild-type cells, and probes prepared from the corresponding RNA were hybridized to Affymetrix tiling arrays. Using the tiling array software, diagrams were generated in which bars represent the differences between the Δicln and wild-type intensities (Fig. 4A). Visual inspection of these tiling array profiles shows that higher intronic signals can be observed for a subpopulation of genes in the Δicln cells compared to the wild type. While no significant changes in the intronic signal occur for the majority of introns, splicing inhibition can be observed in a subset of introns. For example, a decrease of splicing efficiency is observed for both introns of a given gene (SPCC594.01 and SPBC649.04) (Fig. 4A, a and b) or only in some introns when multi-intronic genes are considered (SPBC1703.10) (Fig. 4A, c). The tiling array profiles were confirmed by semiquantitative RT-PCR experiments, showing that, for the displayed genes, the pre-mRNAs accumulate in the Δicln cells at the expense of the fully mature transcripts (Fig. 4A, right). Altogether, these studies show that depletion of the SpICln protein results in the inhibition of splicing in a specific subset of introns. We used bioinformatics analyses to calculate the difference between the mean signal intensities for each intron and the mean signal intensities of the two adjacent introns (see Materials and Methods), allowing the calculation of a retention index for each intron (see Table S1 in the supplemental material). The top 20 most inefficiently spliced introns are listed in Table S2 in the supplemental material.

FIG 4.

Differential splicing defects in Δicln cells revealed by microarray analyses. (A) Tiling array profiles of genes showing changes in intron signals in the Δicln cells compared to wild-type cells. The scale on the y axis represents the log₂ fold change between the treatment (Δicln) and control (wt) group signals. RT-PCR validation tests of the genes shown on the left were performed on total RNA isolated from the wild-type and Δicln cells grown at 25°C. Exon-specific primers were used to amplify the corresponding spliced and unspliced species. The PCR products were separated on 1.5% to 2% agarose gels and visualized by ethidium bromide staining. The gene systematic identifier and the intron numbers are indicated below the gel, and the schematics of spliced and intron-containing mRNAs are shown to the left of the gel. (B) The retained introns contain PPTs located further upstream of the branch point. The histogram depicts the frequencies of PPTs of 5 or more pyrimidines at the indicated positions upstream of the branch point adenosine in the group of 250 introns compared to the remaining 4,361 introns. The circled numbers represent positions having significant different frequencies (see text for more details). The Wilcoxon P value is indicated. (C) The retained introns contain a lower percentage of A/U. The histograms show a comparison of the base compositions (percent A/U) of the group of retained introns (group 250) and the remaining group of introns (group 4361) (t test; P = 2.2 × 10⁻¹⁶).

Inefficiently spliced introns have polypyrimidine tracts located far upstream of the branch point and have a lower A/U content.

In order to find features in the retained introns that explain their inefficient splicing, we used the nonparametric Wilcoxon test to compare distributions between two groups carrying a defined property. We first tested the distribution of the sequences corresponding to the 5′ splice donor site, to the branch point, or to the acceptor site and found no significant differences between groups of introns (data not shown). In contrast, analysis of the positions of polypyrimidine tracts (PPTs) located upstream of the branch point site allowed the identification of a group of 250 introns significantly different from the remaining group (P = 0.004) (see Fig. S2A in the supplemental material). While the frequencies of PPT sizes are not significantly different in the two groups of introns (see Fig. S2B in the supplemental material), the frequencies of PPTs located upstream of the branch point at positions −5, −7, and −9 are decreased by a factor of 3 in the group of retained introns, and this reduction is concomitant with an increase in the frequency of PPTs located further upstream of the branch point at positions −17 to −19 (Fig. 4B).

To determine other intronic features that may contribute to inefficient splicing, we analyzed the base composition of the two groups of introns, and interestingly, we noticed that the group of retained introns has a lower A/U percentage, with a decrease of 3.9 to 5.5% compared to the other group of introns (Fig. 4C). Altogether, these statistical analyses indicate that the position of the PPT located upstream of the branch point adenosine, as well as the intronic A/U content, may be distinguishing features between the two groups of introns.

Biological functions of the genes containing poorly spliced introns.

We next compared the group of 250 retained introns found in Δicln cells with the group of 455 retained introns that was found to be significantly different in the tdSMN mutant (31). We found that 57 genes are common to both groups, and their systematic identification numbers in PomBase are shown in Table S3 in the supplemental material. To test whether specific categories of genes were overrepresented among these common genes, we performed a statistical analysis with the Panther classification system, using Bonferroni correction for multiple testing, implying that the confidence for their overrepresentation is significant (32, 41). As shown in Table 1, nine genes belonging to the Panther protein class “enzyme modulator” were found to be significantly overrepresented in the subset of common genes compared to the entire reference list of S. pombe genes. Interestingly, a majority of the genes code for proteins with GTPase activity and play a role in intracellular protein transport. These results suggest that introns in these classes of genes might be particularly sensitive to a decrease in the level of functional snRNPs in fission yeast. However, further studies are needed to determine the biological relevance of splicing inhibition of corresponding pre-mRNAs to Δicln and tdSMN growth defects.

TABLE 1.

Names and properties of genes significantly overrepresented in retained introns in Δicln and tdSMN cells^a

Gene ID	Uniprot number	Gene product	Panther family	GO molecular function(s)	GO biological process(es)
SPBC21D10.05c	O74345	UCP3 UBA domain-containing protein 3	Centaurin/ARF related	Small GTPase; regulator activity	G-protein-coupled receptor; protein signaling pathway
SPAC16.01	Q10133	GTP-binding protein Rho2	GTP-binding protein	GTPase activity; protein binding	Intracellular protein transport
SPAC23C4.09c	O13929	Uncharacterized protein C23C4.09c	Programmed cell death protein 5	Protein binding	Induction of apoptosis
SPAC1A6.05c	Q9Y827	Uncharacterized protein C1A6.05c	ADP ribosylation factor related	GTPase activity; protein binding	Intracellular protein transport
SPAC105.02c	Q9P7I0	Ankyrin repeat-containing protein C105.02c	Not named	Protein binding; phosphatase regulator activity	Protein amino acid phosphorylation
SPBC15C4.03	O60112	Uncharacterized Rab geranylgeranyltransferase C15C4.03	RAB-GDP dissociation inhibitor	Acyltransferase activity; protein binding; small GTPase regulator activity	Intracellular protein transport; vesicle-mediated transport; synaptic transmission
SPCC1919.03c	P78789	Uncharacterized protein C1919.03c	5′-AMP-activated protein kinase, β subunit	Protein binding; kinase regulator activity	Protein amino acid phosphorylation; response to stress
SPBC28E12.03	O74360	RGA4 probable rho-type GTPase-activating protein 4	Chimerin-related Rho GTPase-activating protein	Protein binding; small GTPase; regulator activity	G-protein modulator
SPAC22F3.05c	Q09767	ADP-ribosylation factor-like protein Alp41	ADP-ribosylation factor-like 2	GTPase activity; protein binding	Small GTPase; synaptic vesicle exocytosis; intracellular transport

^aOverrepresentation of particular classes of proteins was tested by Panther analysis. Nine genes from the Panther protein class reported as “enzyme modulator” were found to be overrepresented in the subset of 57 genes (listed in Table S3 in the supplemental material) common to the groups of retained introns defined as significantly different in Δicln and tdSMN cells. The binomial test using Bonferroni correction was used to determine statistical overrepresentation in the PANTHER protein class (32). The systematic identification (ID) and Uniprot numbers, as well as the gene standard names, are shown, together with the Panther family annotations and the gene ontology (GO) terms for molecular function and biological process.

DISCUSSION

Numerous in vitro experiments have demonstrated that pICln interacts with the Sm proteins and plays an important role in the early steps of snRNP formation by acting as an assembly chaperone (17, 18). In this study, we used an in vivo approach to investigate the properties of the fission yeast pICln protein and its functional relationship with the SMN protein.

While disruption of the IClN gene causes embryonic lethality in the mouse (42), surprisingly, we found that the ICLN gene is not essential in fission yeast. However, it is required for optimal yeast cell growth, indicating that it participates in critical cellular processes. A homologous protein in Saccharomyces cerevisiae (YKL183w; LOT5) is also not essential, suggesting that in both yeasts, the assembly of the Sm protein core complex can occur in vivo, at least partially, even without pICln chaperone activity. An alternative explanation is that yeasts might contain additional proteins with redundant functions; however, we failed to identify other significant homologs in both the S. cerevisiae and S. pombe databases.

We show that S. pombe ICln interacts strongly only with SmD1 and SmD3 proteins. These results are consistent with the recent crystal structure of a 6S ICln-Sm assembly intermediate, showing that the β5 domain of ICln interacts with the β4 strand of SmD1 to form an antiparallel β pair (18). It is noteworthy that the region encompassing the pICln β5 strand is evolutionarily highly conserved compared to the overall low sequence conservation observed for ICln proteins from different organisms (Fig. 1A) (18). Interestingly, in the proposed structure, the β0 strand of pICln, which associates with the β5 strand of SmG, becomes ordered only upon formation of the 6S intermediate, precluding an association with monomeric SmG protein, and these observations might potentially explain the inability of both proteins to interact in the in vitro binding assays. The strong interaction of SpICln with SmD3 that we detected in vitro is also consistent with previous studies showing that pICln forms a complex with the SmB-SmD3 dimer (10, 17, 34) and further suggests that pICln might interact with the β4 surface of SmD3 by conventional Sm-Sm contacts via its β5 strand domain.

We also found that SpICln and SmD1 comigrate in the 6S region of the sucrose gradient (Fig. 2C), but we were unable to detect a 20S complex corresponding to the mammalian methylosome containing the PRMT5 methyltransferase and pICln and Sm proteins (10, 11). It is possible that the formation of the 20S complex is very transient in fission yeast, and consistent with this hypothesis, we were unable to detect accumulation of GFP-Sm proteins in the cytoplasm in Δicln cells (see Fig. S1C in the supplemental material). An apparent nonessential fission yeast homolog of PRMT5, named Skb1, has been shown to possess catalytic activity in vitro (43), but whether the arginines of the RG repeats in S. pombe SmB, SmD1, and SmD3 represent its physiological substrates and are used to regulate Sm core assembly and snRNP biogenesis remains an open question.

As is the case for tdSMN, the growth defect of Δicln cells probably results from the decrease in the levels of the spliceosomal U1, U2, U4, and U5 snRNPs (Fig. 3). Accordingly, we found that splicing of pre-mRNAs is differentially affected upon ICln depletion, showing that the spliceosome is selectively perturbed when snRNP production is altered. A similar, albeit more pronounced, decrease in snRNP levels and splicing inhibition was also observed in S. pombe tdSMN cells carrying a temperature degron allele of the SMN protein (31). Previous studies in zebrafish demonstrated that knockdown of snRNP assembly factors (i.e., Gemin2, ICln, and SMN) causes the degeneration of motor axons (44, 45), suggesting that depletion of snRNP levels has a similar effect on the function of a gene subset required for axon outgrowth. Our data obtained using S. pombe Δicln and tdSMN cells show that the splicing of a class of genes possessing GTPase activity is significantly altered when snRNP levels are reduced (Table 1). Although highly speculative, it is tempting to propose that this might also be the case when snRNP levels are decreased in other models, linking SMN dysfunctions to defects in GTPase activities. In this regard, it is noteworthy that actin cytoskeleton dynamics are controlled by G proteins/GTPases (46–48) and that defects in neuritogenesis observed upon SMN depletion could be caused by alterations in cytoskeletal integrity (49). These observations clearly warrant further studies to test whether splicing of pre-mRNA coding for G proteins/GTPases in SMN-depleted mammalian cells is also inhibited.

Our bioinformatics analysis shows that a fraction of inefficiently spliced introns with a high retention index have PPTs that are located further upstream of the branch point (Fig. 4B). The same feature was observed for inefficiently spliced introns in the tdSMN cells, confirming that in S. pombe, decreased levels of snRNPs affect splicing of introns carrying suboptimal PPTs, which is in agreement with previous reports showing that PPTs are an important element for efficient splicing in fission yeast (31, 50, 51). Our results also show that the A/U enrichment in introns contributes to efficient splicing (Fig. 4C). This is consistent with a previous study showing that introns, which are spliced correctly even in the absence of U2AF⁵⁹, have higher A/U content (52). Taking into account the results of our study, we propose that, when snRNP levels are sufficient (as in wild-type cells), early recognition of pre-mRNA by U1 snRNP allows the efficient tethering of other factors (for example, U2AF) to the prespliceosome, leading to further steps of spliceosome assembly and efficient splicing. In contrast, due to suboptimal recognition determinants (increased distance between PPT and the branch point and/or decreased A/U content), lower levels of snRNPs hinder efficient formation of early splicing complexes, leading to inhibition of splicing assembly and to intron retention.

Our study is in agreement with the notion that alterations in the concentrations of core spliceosomal components have different effects on splice site selection. Indeed, it has recently been shown that knockdown of SmB/B′ in human cells results in reduced levels of snRNPs and reduction in the inclusion levels of many (but not all) alternative exons (53). Altered splicing patterns and differential intron retention have also been observed in yeasts and mammals upon mutation, deletion, or knockdown of core spliceosomal and spliceosome assembly factors (52, 54–60). Furthermore, inefficient spliceosome formation has also been observed recently in SMN-depleted human cells and in cells treated with leptomycin B (LMB), which inhibits export of nascent snRNAs and indirectly inhibits snRNP assembly (61). Mature snRNPs showed increased mobility upon inhibition of snRNP biogenesis, presumably as a consequence of reduced availability of snRNPs for spliceosome formation (61). A similar mechanism might also explain the defects observed when snRNP production is inhibited in Δicln and tdSMN cells: due to suboptimal PPT and A/U content, snRNPs require longer times to stochastically bind to “weak” introns and are thus hindered in associating stably with pre-mRNAs during the early steps of spliceosome assembly. Future studies should identify suppressors of growth defects in Δicln or tdSMN cells and explain why the splicing efficiency of only a subset of introns is altered when snRNP levels are decreased. They could also potentially reveal splicing components able to compensate for early splicing defects observed in cells carrying low levels of spliceosomal snRNPs.