|Home | About | Journals | Submit | Contact Us | Français|
Efficient coupling between mRNA synthesis and export is essential for gene expression. Sus1/ENY2, a component of the SAGA and TREX-2 complexes, is involved in both transcription and mRNA export. While most yeast genes lack introns, we previously reported that yeast SUS1 bears two. Here we show that this feature is evolutionarily conserved and critical for Sus1 function. We determine that while SUS1 splicing is inefficient, it responds to cellular conditions, and intronic mutations either promoting or blocking splicing lead to defects in mRNA export and cell growth. Consistent with this, we find that an intron-less SUS1 only partially rescues sus1Δ phenotypes. Remarkably, splicing of each SUS1 intron is also affected by the presence of the other and by SUS1 exonic sequences. Moreover, by following SUS1 RNA and protein levels we establish that nonsense-mediated decay (NMD) pathway and the splicing factor Mud2 both play a role in SUS1 expression. Our data (and those of the accompanying work by Hossain et al.) provide evidence of the involvement of splicing, translation, and decay in the regulation of early events in mRNP biogenesis; and imply the additional requirement for a balance in splicing isoforms from a single gene.
A significant part of mRNP biogenesis takes place co-transcriptionally, and functional links between chromatin modifications, transcriptional elongation, splicing and mRNA export have been shown (1,2). Sus1 (ENY2 in metazoans) is a small, evolutionarily conserved 11-kDa protein that is a component of two molecular assemblies with roles in both transcription and mRNA export (3,4). Sus1 can be found in the SAGA complex, involved in chromatin function, where is part of the histone H2B deubiquitinating (DUB) module together with Ubp8, Sgf11 and Sgf73 (5,6). Interactions between these four proteins are necessary to reconstitute full DUB activity in vitro, as revealed recently by structural studies (7,8). In addition, during transcription Sus1 can be recruited to promoters and along coding regions, where it interacts with RNA Pol II and the mRNA export factors Yra1 and Mex67 (9). At the nuclear pore, Sus1 is a component of the TREX2 complex formed by Sac3, Thp1, Sus1, Cdc31 and Sem1 (10–12). Sus1 plays a role in mRNA export from the nucleus and it is implicated in anchoring active genes at the nuclear periphery through gene gating (12,13). Moreover, in specific circumstances Sus1 can be observed at cytoplasmic structures (14). Thus, Sus1 participates in multiple stages of gene expression, with roles in chromatin function and mRNA biogenesis.
In addition, the SUS1 gene structure is of particular interest. Unlike most yeast genes, which are typically intronless, SUS1 bears two introns; and while most yeast introns display conserved splicing signals (15), both the 5′ splice site and branch site of the first intron of SUS1 are non-canonical (3,16). Interestingly, Saccharomyces cerevisiae introns are not randomly distributed throughout the genome, but most are found in highly expressed genes (17).
Notably, introns are found as well in a number of genes functionally relevant to RNA metabolism, such as YRA1, MTR2, SRC1, NSP1 and HRB1 (18). Interestingly, SRC1 is a rare example of alternative splicing in yeast (19,20), and both YRA1 and SRC1 are functionally and genetically linked to SUS1 (10,20). In fact, extensive work on YRA1 points to the intricate circuitry of an mRNP life, encompassing splicing, export, and degradation (21–25).
Here we assessed a link between Sus1 function and SUS1 unusual genomic structure. We find that SUS1 introns are evolutionarily conserved, and their deletion leads to defects in RNA export and to slow growth. The unconventional splicing signals in SUS1 are also important for SUS1 function, which is affected by non-essential splicing factors such as Mud2 and modulated by the non-sense mediated decay (NMD) pathway. Thus, proper SUS1 expression requires a delicate balance of splicing, translation and decay.
Putative orthologs of S. cerevisiae SUS1 gene were identified by BLASTn (26) and WU-BLAST2 (27) searches on public databases and default search parameters were used. A list of species used in our analysis can be found at Supplementary Table S1. The evolutionary transformation of three structural characters of SUS1 (intron number, 5′ss and BS) were optimized onto the known phylogeny of 25 eukaryotes species (28,29). Ancestral and derived multistate transformations were achieved considering equal weighted unordered characters in MacClade program (30).
Growth assays to test the intron functionality were done by growing cells at 30°C on synthetic selective medium [SC: glucose 2%, ammonium sulfate 0.5%, yeast nitrogen base 0.17% and supplements (Dropout)] lacking Tryptophan (Trp) to 0.3–0.4 OD600. Subsequently, 10-fold serial dilutions of an equal number of cells were made and drops spotted onto SC-trp plates. Growth was recorded after 2–3 days of incubation at 30 and 37°C. Yeast cell transformations were done by the LiAc/SS carrier DNA/PEG method (31). For wild-type (WT), mud2Δ, upf1Δ and mud2Δupf1Δ splicing pattern comparison, cells were grown at 30°C in YPD (2% glucose, 1% yeast extract and 2% peptone) and collected when 0.6 OD600. To test different stress conditions influencing SUS1 splicing pattern, cells were grown in YPD at 30°C until 0.4 OD600 and then divided in three flask, in order to incubate the cell cultures under different treatments: (i) cells were grown 2h more at 30°C in YPD; (ii) cells were incubated 20min at 42°C; or (iii) carbon source was shifted to galactose and the cells were grown 2h at 30°C. After each treatment, cells were collected, frozen in liquid nitrogen and stored at −80°C until RNA extraction. For copper assays, cultures at 30°C were grown to 0.4 OD600 in SC medium lacking leucine and equal volumes were dropped onto SC-Leu plates containing CuSO4 ranging from 0 to 1.0mM (32). Plates were photographed after 3 days at 30°C.
Total RNA was harvested from yeast cultures by the Hot/Acid-phenol method (33). RNA was quantified using Nanodrop and quality was checked by agarose gels with ethidium bromide (etBr). Northern blot analyses were carried out as in (34). SUS1 and U1 snRNA detection was done using T7-transcribed riboprobes.
RT–PCR and qRT–PCR analysis were performed using 1μg of total RNA. After DNase-I treatment (Promega), RNAs were purified by Phenol/Chloroform extraction. Reverse transcription was performed using standard procedures, with random hexamers and M-MLV reverse transcriptase (Invitrogen). Specific pairs of primers were used to amplify SCR1 or SUS1 transcripts containing exons 1 and 3; using 3μl of cDNA as template (previously diluted 1/20). Amplified products were run in a 2% agarose-EtBr gel and visualised. Specific primers for each SUS1 transcript or SCR1 ncRNAs were used to amplify qRT–PCR products from 3μl of cDNA (previously diluted 1/10) and using SYBR® GreenERTM qPCR SuperMix (Invitrogen) in a 10μl final volume. Each sample was run in duplicate with the standard curve. Real-Time PCR was performed using a LightCycler® (Roche). An activation step of 10min at 95°C followed by 45 cycles of 10s at 94°C, 15s at 50°C and 20s at 72°C were used for mRNA, intron 1 (I1) and intron 2 (I2) primer pairs. For I1 and I2 transcripts, the 72°C step was extended to 30s. For SCR1 RNA, 38 cycles of 10s at 94°C, 15s at 60°C and 20s at 72°C were used. See the ‘Materials and Methods’ in Supplementary Data for a list of primers. The amount of each SUS1 transcripts was represented in arbitrary units (AU). One arbitrary unit is equivalent to 10−2fmol/mg of total RNA.
For crude protein extracts, 1.5 OD600 from cultures on exponential growth at 30°C were collected. Cells were lysed in 150μl of NaOH, 2M plus 7.5% (v/v) β-mercaptoethanol and proteins were precipitated for 10min on ice. After centrifugation, the pellet was resuspended in 50μl of denaturing loading buffer. Samples were heated at 95°C before loading and separation in 16% SDS–PAGE gels. Western blot analysis was performed using anti-LexA and anti-GFP according to standard procedures.
For GFP-Sus1 immunoprecipitations 50ml of WT and mud2Δ upf1Δ strains containing plasmids bearing GFP or GFP-SUS1g were grown on selective medium to OD600 0.5. Cells were harvested, washed, and resuspended in 200μl of ice-cold lysis buffer (50mM HEPES-KOH at pH 7.5, 140mM NaCl, 1mM EDTA, 10% glycerol, 0.5% NP-40, 1mM PMSF and protease inhibitors). The same volume of glass beads was added and cells were broken by four pulses of vortexing during 1min at 4°C. The supernatant were immunoprecipitated during 1h at 4°C in a turning wheel and using anti-GFP antibody (Roche) coupled to dynabeads (Invitrogen). The immunoprecipitates were washed three times for 10min with lysis buffer and subsequently resuspended in 50μl of SDS–PAGE sample buffer. Western blot analysis of samples was made using anti-GFP and anti-Sus1 antibodies.
Fluorescent in situ hybridization (FISH) against poly(A)+RNA was done growing yeast cells in 100ml of YPD medium at 30°C to an 0.5 OD600. Then, cultures were divided and a half of each culture were rapidly shifted to 37°C incubator for 3h, and the other half was incubated at 30°C during that time. Cells were immediately fixed adding 10% of formaldehyde, 1h at room temperature. The fixative was removed by two rounds of centrifugation and washed with 0.1M potassium phospate (pH 6.4). Cells were resuspended in ice-cold washing buffer (1.2M sorbitol and 0.1M potassium phosphate, pH 6.4), and subsequently, cell wall was digested with 0.5mg/ml of Zymolyase 100T and samples were applied on poly-l-lysine-coated slide wells. Non-adhering cells were removed by aspiration, cells were rehydrated with 2× SSC (0.15M NaCl and 0.015M sodium citrate) and hybridized overnight at 37°C in 20µl of prehybridization buffer (formamide 50%, dextran sulfate 10%, 125µg/ml of Escherichia coli tRNA, 4× SSC, 1× Denhardt’s solution and 500µg/ml herring sperm DNA) with 0.8pmol of Cy3-end-labeled oligo(dT) in a humid chamber. After hybridization, slides were washed with 1× SSC at room temperature, air-dried and mounted using VECTASHIELD® Mounting Medium with DAPI. Detection of Cy3-oligo(dT) was performed using a Leica DM600B fluorescence microscope.
Two features of SUS1 genomic structure are of particular interest. As we previously showed, SUS1 (ENY2 in metazoans) bears two introns, a rare feature in the S. cerevisiae genome (3). Only 5% of the genes in S. cerevisiae have introns and of those just ten carry two (18,35–37). In addition, while most yeast introns display conserved splicing signals, both the 5′ splice site and branch site of the first intron of SUS1 are non-canonical. Interestingly, other yeast introns with regulated splicing also have non-canonical splice sites (21). Thus, we decided to explore the degree of evolutionary conservation of these features, which may indicate their biological relevance.
We identified SUS1 homologs in different species (Supplementary Tables S1 and S2), and performed an evolutionary analysis. Our results (Figure 1) reveal the following: first, SUS1 is likely to have evolved from an ancestral gene containing four introns (Figure 1A). While in fungi genes containing one, three or four introns are equally probable; after the branching to C. neoformans (lane 3 in Figure 1) a single-intron gene (SUS1 I1) is unequivocally the most parsimonious ancestral reconstruction. Subsequent evolution leads to the addition of a second intron in most of the Saccharomycotina group of species. Second, having two introns in fungi correlates with carrying a non-canonical 5′ss in the first intron (Figure 1B and Table 1). Third, non-consensus BS in I1 is only present in S. bayanus, S. mikatae, S. paradoxus and S. cerevisiae (Figure 1C). Strikingly, in fungi, SUS1 genes having I1 with the non-consensus BS UACUGAC correlates with the presence of a highly conserved 3′UTR of ~300nt (Supplementary Figure S1). Altogether, these data suggest a biological relevance for SUS1 introns in its expression and we decided to further address this question.
We used quantitative and semi-quantitative RT–PCR to determine the relative splicing efficiencies of SUS1 introns. Bands corresponding to the fully spliced (mRNA) and SUS1 transcripts retaining I1, were consistently detected in WT cells (Figure 2A). This is in agreement with both data from Hossain et al. (16) and genome-wide studies (38,39), reporting detectable amounts of SUS1 transcripts retaining I1. We hypothesize that splicing of SUS1 I1 is either inefficient or subject to regulation, leading to cellular accumulation of unprocessed SUS1 transcripts.
To test this idea, we analyzed SUS1 splicing under different conditions of temperature and carbon source, where SUS1 function is more determinant (10). Quantitative RT–PCRs of RNA collected from WT cells incubated 20min at 42°C revealed that SUS1 expression is reduced by the temperature shift, with accumulation of unspliced SUS1 transcripts and reduction of the corresponding mRNA (Figure 2B). Notably, SUS1 expression is induced by growth in galactose, while its splicing pattern remains unchanged (Figure 2B). Altogether, our results indicate changes in expression and splicing efficiency dependent on growth conditions and suggest that SUS1 splicing is a regulated event.
To analyze SUS1 splicing we generated constructs based on the CUP1 reporter system, widely used to follow splicing efficiency in vivo (32). cup1Δ cells were transformed with one of the following reporters (Figure 3A): pACT1-CUP1, pSUS1g-CUP1 (with both SUS1 introns), pSUS1-I1Δ-CUP1 (I1 removed) and pSUS1-BS-CUP1 (I1 with UACUGAC mutated to UACUAAC, rendering its BS consensus). Splicing was monitored by both assessing copper tolerance and primer extension analyses. As illustrated in Figure 3A deletion of I1 improves growth compared to the genomic construct (compare lanes 2 and 4), likely due to a better splicing efficiency of SUS1 transcripts lacking I1, which suggests that I1 splicing is limiting for SUS1 expression. Accordingly, the construct SUS1-BS-CUP1 displays a similar copper tolerance to the construct SUS1-I1Δ-CUP1 (Figure 3A, lanes 3 and 4). These results were further confirmed by primer extension (data not shown) and are consistent with those presented in the accompanying paper (Hossain et al. this issue). Thereby we conclude that SUS1 splicing efficiency likely is limited by the suboptimal BS in I1.
We further explored this possibility by assessing the role of Mud2 in SUS1 expression. Mud2, the yeast homolog of U2AF, interacts with BBP, an essential factor that identifies the BS and its required for splicing of a variety of transcripts (40,41). Mud2 is not essential, so we analyzed SUS1 splicing in mud2Δ cells. Northern blot analysis of RNA from mud2Δ cells shows accumulation of SUS1 pre-mRNA species with a concomitant decrease of mature RNA, compared to WT (see below in Figure 4B and Hossain et al). Thus, we decided to further analyze SUS1 splicing in cup1Δ mud2Δ cells, using our CUP1 reporters. Consistent with the northern blot analysis, deletion of MUD2 leads to a marked reduction in copper tolerance for cells carrying SUS1g-CUP1 (Figure 3A, compare lanes 2 and 6). Interestingly, while mutation of the I1 BS to consensus (SUS1-BS-CUP1) dramatically improves copper tolerance in mud2Δ cells, this tolerance is further increased by deleting I1 (Figure 3A, lanes 7 and 8). This is consistent with a requirement for Mud2 in I1 splicing in part but not exclusively due to the non-consensus BS, and the non-consensus 5′ss of I1 is likely to be involved as well (Hossain et al. this issue).
To further assess the dependency of SUS1 splicing on Mud2 in vivo, we have quantified by qRT–PCR the different splicing forms in mud2Δ cells. Absence of Mud2 correlates with a substantial enrichment of pre-mRNA retaining either both introns or I1, rising from 16% of total transcripts to 68% (a 4-fold increase, Figure 3B). However, the largest change in absolute terms is the dramatic decrease in the mature mRNA species (a 10-fold change). In summary, our data support a prominent role for Mud2 in SUS1 expression. It is likely that other factors involved in the early recognition of introns are relevant to SUS1 expression as well (Hossain et al., this issue).
The reduced SUS1 expression in mud2Δ cells (Figure 3B) might be a consequence of the appearance of premature stop codons (PTC) in transcripts containing I1, which would trigger NMD (42). Consistent with this hypothesis, qRT–PCR and northern analyses of RNA from both WT and upf1Δ strains revealed a 1.5-fold increase in accumulation of intron-containing SUS1 transcripts in upf1Δ cells (Figure 4A and B, lane 3).
To test whether products of SUS1 splicing (or mis-splicing) in mud2Δ cells were targeted for degradation by the NMD pathway, we determined the levels of SUS1 transcripts in mud2Δ upf1Δ cells (Figure 3). Northern blot and semi-quantitative RT–PCR analyses (Figure 4B) show accumulation of unspliced SUS1 transcripts in mud2Δ upf1Δ cells, compared to those of WT. qPCR results (Figure 4C) indicate that this accumulation corresponds to a 6-fold increase. Thus we conclude that SUS1 splicing depends on Mud2 and that SUS1 unspliced transcripts are targeted by NMD.
Many introns regulate gene expression by a variety of mechanisms, and in yeast there are several instances of introns involved in autoregulatory loops (YRA1 or RPL30) (17). Here we show that SUS1 introns have a critical role on mRNA levels, and in order to assess whether they are necessary to maintain appropriate Sus1 amounts, we determined the quantity of Sus1 protein expressed from several constructs including those carrying modified splicing signals. Plasmids bearing SUS1cDNA or the genomic version of SUS1 (SUS1g), N-terminally fused to LexA or GFP tags, were introduced into a mutant strain lacking endogenous SUS1. Levels of LexA (Figure 5A) or GFP (Supplementary Figure S2) were monitored by western blot analyses. Our results indicate that elimination of SUS1 introns causes accumulation of Sus1 protein when compared to SUS1g (Figure 5A compare upper band in lanes 2 and 3).
To study how the presence and position of each intron contribute to Sus1 protein levels, we cloned different SUS1 constructs in pBTM-LexA (43) under the ADH1 promoter. Expression of the LexA-Sus1 fusion proteins was monitored by western blotting using anti-LexA antibodies (Figure 5A). Notably, the presence of I1 at both locations (2xI1, Figure 5A construct 4) provoked the strongest reduction in Sus1 protein levels (Figure 5A, lane 4). As expected from our splicing analysis, I1 elimination (lane 7), substitution by I2 (lane 5) or mutation of its BS signal to the consensus (lane 9), stimulated Sus1 expression to levels comparable to the SUS1cDNA version. Of note, swapping introns (lane 6) or removing I2 (lane 8) also affects Sus1 protein expression. We conclude that both the presence and the position of both introns are key determinants of Sus1 protein expression levels.
Although production of protein isoforms by alternative splicing in yeast is rare, SUS1 potentially could be alternatively spliced to produce different proteins. N-terminal tagging of SUS1 with both introns enables the detection of alternate forms of Sus1 protein if they exist. Strikingly, a smaller intense band similar in size to the tag alone was observed when using both antibodies (marked band with an asterisk in Figure 5A). This band is consistent with a smaller version of Sus1 protein translated from pre-mRNAs retaining either I1 or both introns. Intron 1 retention introduces a PTC, and the predicted protein translated from this sequence will include exon 1 and 6 additional residues. To assess whether this protein is not a degradation product of full-length Sus1, we hypothesized that its abundance will correlate with an increase in SUS1 transcripts containing I1. Therefore, we determined Sus1 protein relative abundance in two different situations. First, we followed the ratio small/full-length Sus1 protein (pLexA-SUS1g) in mud2Δ upf1Δ cells, where unspliced versions of SUS1 are enriched compared to WT (Figure 4B and C). Consistent with our hypothesis, we observe an accumulation of the small Sus1 in the double mutants (Figure 5B). Second, we cloned into pBTM-LexA a version of the SUS1 gene [SL-9 mut, (10)] that has a mutated BS sequence (UACUGAC mutated to CACUGAC), which blocks splicing of I1, driving accumulation of SUS1 transcripts retaining I1 (Figure 5C, left panel). As expected a prominent band consistent with the size of the small Sus1 protein, is recognized using anti-LexA antibodies, while no full-length Sus1 protein is observed (Figure 5C right panel). Altogether, we conclude that a small version of Sus1 is translated from SUS1 transcripts containing I1.
To further verify that the small band detected in our western analyses contains Sus1 residues, we transfected pGFP-SUS1g in WT and mud2Δ upf1Δ cells and subsequently we immunoprecipitated whole cell extracts using anti-GFP antibodies. Western blotting reveals two GFP forms (Figure 5D), one (~38kDa) consistent with full-length Sus1-GFP; and a second (~28kDa) consistent with the translation of Sus1 exon 1 plus 6 residues, fused to GFP (GFP-Sus1p*). Consistent with our model, this form is enriched in extracts from mud2Δupf1Δ cells bearing pGFP-SUS1g (which leads to accumulation of SUS1 transcripts retaining I1, lane 3). Additional confirmation that the 28kDa band includes a truncated version of Sus1 was obtained by probing the same immunoprecipitated with anti-Sus1 antibodies, producing the same result (Figure 5D, right panel).
To further decipher the molecular mechanisms controlling SUS1 splicing, we have investigated whether exonic sequences flanking SUS1 introns might contribute to their splicing. To do this, we have analyzed the splicing efficiency of SUS1 introns in another exonic context, using the TAF14 gene. TAF14 has one canonical intron (position 10–114nt, from the start codon). We have followed the expression of this construct by western analysis of LexA fusions (Figure 6A). As anticipated given its reduced splicing, replacement of the TAF14 native intron by SUS1 I1 decreases the amount of expressed LexA-Taf14 protein (Figure 6A, lanes 1 and 2). In an attempt to recapitulate SUS1 gene organization, we cloned I2 at the same distance from I1 that it is located in SUS1 ORF, creating a Taf14 exon2* (see construct design in Figure 6A). Surprisingly, incorporation of I2 leads to a further splicing block, causing a strong reduction in Taf14 protein levels (Figure 6A, lanes 2 and 3). This is consistent with the observed importance of both introns in SUS1 expression (Figure 5). Next we decided to explore the role of exon 2 sequences in SUS1 splicing using our LexA-TAF14 construct containing SUS1 introns (TAF14-I1&I2SUS1). Intriguingly, replacing the TAF14 sequence between SUS1 introns (TAF14 exon 2*) by SUS1 exon 2, restores splicing to almost native Taf14 levels (Figure 6A, compare lanes 1 and 4).
To better substantiate this observation, we have monitored splicing efficiency by growth in copper using a CUP1 construct in which the original SUS1 exon 2 was replaced by a sequence corresponding to TAF14 exon 2* (see construct design in Figure 6B). In agreement with our western results, replacement of SUS1 exon 2 by TAF14 sequence results in reduced copper tolerance consistent with a reduced splicing (Figure 6B, lane 3). Thus we conclude that exon 2 sequences on SUS1 gene support proper SUS1 splicing.
To address the biological relevance of SUS1 introns we have examined the phenotype of sus1Δ cells transformed with plasmids containing either SUS1 with its introns (pSUS1g) or without (pSUS1cDNA). sus1Δ cells are sensitive to high temperature (ts) and their growth can be affected by the carbon source [data not shown and (10)]. Notably, while sus1Δ cells transformed with pSUS1g lose the ts phenotype, there is less suppression of the phenotype with the pSUS1cDNA (Figure 7A, see growth at 37°C). This is consistent with the differential expression profile observed for SUS1 transcripts at high temperature (Figure 2B), and we speculate that this profile is relevant to the role of Sus1 at 37°C.
An intriguing explanation could be that SUS1 transcripts containing I1 have functionality at higher temperatures. To assess this hypothesis, we tested the growth at 37°C of sus1Δ cells transformed with pSUS1-SL-9 (with an I1 containing a non-consensus BS, which renders mostly SUS1-I1 retained transcripts, as described in Figure 5C). As expected from its deficient splicing, cells transformed with this construct grow similar to an empty plasmid at 30°C (Figure 7A, compare growth of p-empty and pSUS-SL-9 at 30°C after 24h). Surprisingly, this construct is able to complement the ts phenotype exhibited at 37°C to a larger extent (Figure 7A, compare growth of p-empty and pSUS-SL-9 at 37°C after 48h) than the cDNA, albeit less so than the gDNA. These data indicate that SUS1 introns are required for optimal growth and have a biological relevance.
To further address whether changes in Sus1 protein expression could account for the differences observed in our complementation assay, we followed the levels of Sus1 isoforms at different temperatures when expressed from the genomic version (pSUS1g), the cDNA (pSUS1cDNA), or the SL-9 BS mutant (pSUS1-SL-9) (Figure 7B). As expected from the results of the RNA analyses, a down-regulation of full-length Sus1p is observed at 37°C (compare upper band in lanes 2 and 3). Although a reduction in Sus1p levels is also appreciable in cells containing the cDNA version, Sus1p levels at 37°C are still elevated when compared to those of wt both at 30 or 37°C (lanes 4 and 5 versus 2 and 3). A slight downregulation of Sus1p* also occurs at 37°C (lanes 2 and 3) and this is more apparent when SUS1 is expressed form the SL-9 BS mutant version (lanes 6 and 7). These data support our hypothesis that the changes in the relative levels of Sus1 isoforms account for the differences observed in our complementation assays.
Sus1 is necessary for mRNA export, and in sus1Δ cells nuclear accumulation of poly(A)+ RNA is readily detected at 30 and 37°C [Figure 7C and (10)]. To assess the effect of SUS1 introns on this phenotype, we performed RNA in situ hybridization (FISH) in exponentially growing sus1Δ cells, transformed with pSUS1g, pSUS1cDNA, or pSUS1-I1Δ (Figure 7C). Consistent with the growth complementation results shown in Figure 7B, both SUS1g and SUS1cDNA were able to complement to a large extend the mRNA export defect of sus1Δ cells at 30°C (Figure 7C upper panel). In contrast, FISH signal from sus1Δ cells transformed with pSUS1cDNA and grown at 37°C, indicates that the SUS1 gene lacking both introns is not able to efficiently complement the sus1Δ mRNA export defect. Moreover, this is replicated by a construct lacking I1. Thus we conclude that SUS1 introns are required for the SUS1 role in optimal mRNA export in S. cerevisiae.
The study presented here provides insights into the intricate mechanism that regulates SUS1 expression, a crucial factor linking transcription and mRNA export. We found that (i) SUS1 introns are required for its functionality, (ii) protein levels of Sus1 are controlled by SUS1 introns, (iii) SUS1 expression is regulated by splicing and NMD and (iv) intronic and exonic sequences participate in SUS1 regulation. The accompanying manuscript by Hossain et al. in this issue exposes the relevance of SUS1 expression as a model for understanding the role of core spliceosomal components in alternative splicing.
Discovery of the SUS1 gene in a genetic screening revealed an intriguing genomic organization (10). SUS1 gene (438nt) consists of three exons (of 71, 140, and 77nt) and two introns (80 and 70nt). Notably, their size and position are widely conserved (Supplementary Table S2). Sus1 plays important roles during transcription elongation and mRNA biogenesis, which likely explains its high degree of evolutionary conservation, from yeast to human. Why does this small gene, with key functions at different stages in gene expression, contains two introns in yeast, where most genes contain none? A provocative hypothesis is that this allows the SUS1 gene to be a sensor of these processes, acting in a yet unknown feedback control mechanism. We demonstrate here that splicing and decay of SUS1 transcripts regulate expression of Sus1 protein. This strategy is reminiscent of that of YRA1, another factor involved in the coupling of mRNA export and transcription. Intriguingly, YRA1 shares with SUS1 having both an atypical intron and its expression regulated at the level of splicing, degradation and export of its transcripts, although likely by different mechanisms (22,24,25). Moreover, mutations in SUS1 and YRA1 are synthetic lethal, and Sus1 and Yra1 interact physically. A possible scenario is that these factors are finely tuned to work together to sense or modulate correct mRNP biogenesis, as they are sensitive to alterations in several aspects of this pathway.
Our findings reveal that I1 is the major determinant of SUS1 splicing efficiency. Elimination of I1 or mutation of its BS leads to more efficient SUS1 splicing. These data are consistent with the observed weak splicing efficiency of I1, largely due to its non-consensus BS and imply that I1 and its BS are crucial for SUS1 expression. We also show that splicing of SUS1 largely depends on the BS recognition factor Mud2. However, we see that deletion of I1 has a stronger effect than mutation of the BS on mud2Δ cells. These results imply that other features of I1 are dependent on Mud2, likely the non-consensus 5′ splice site of I1 (GUAUGA). In agreement with this, Hossain et al., in the accompanying study have addressed the relevance of non-canonical splicing signals for I1 and I2, revealing that the 5′ splice site of I1 is in fact an important determinant of its processing.
We find an intriguing co-evolution between the presence of a non-consensus BS in I1 and the existence of a conserved long sequence at the 3′-end of the gene. In this context, transcriptome analyses showed that SUS1 transcripts retaining I1 carry a longer 3′UTR than the fully spliced RNAs (38). We hypothesize that efficient splicing and proper 3′-end formation in SUS1 could be linked, as it has been shown for other transcripts (44). Current work in our lab tries to address how important is this 3′-end sequence in SUS1 regulation.
Intron 2 of SUS1 also appears to play an important role in splicing, despite its apparent lack of unusual features. We find that removing I2 can be detrimental for protein expression for LexA-SUS1 (Figure 5A, lane 8), and intriguingly, the sole presence of I2 provokes a splicing block when placed into the TAF14 gene (Figure 6A), which is consistent with the data from Hossain et al. addressing the splicing efficiency of I2. In addition, swapping SUS1 introns also affects Sus1 protein levels, strongly arguing for a co-dependence in splicing of both SUS1 introns. We also have found that splicing of SUS1 introns is also affected by exonic sequences, as evidenced by the low expression of TAF14-SUS1 chimerical constructs (Figure 6). Remarkably this can be suppressed by replacing the TAF14 newly-created middle-exon with that of SUS1. Consistent with this, substituting SUS1 exon 2 by this TAF14 ‘exon 2’ reduces Cup1 expression in SUS1-CUP1 constructs. Thus our data suggest a positive role for exon 2 in SUS1 splicing. Work is currently under way to address whether exonic splicing enhancers are involved.
We have verified the biological relevance of SUS1 introns, by showing that a cDNA version of SUS1 is not able to fully complement sus1Δ phenotypes in mRNA export and growth [Figure 7; we also assess that this is not due to a lack of protein production from an intron-less construct (lanes 2 and 3 versus 4 and 5, Figure 7B)]. Similar behavior was observed for YRA1 cDNA (21). Observations that SUS1 splicing is inefficient under optimal growth (Figure 2A) suggest that SUS1 splicing could be regulated. Consistently, we find that the ratios between the different species of SUS1 RNAs are influenced by growth conditions. In fact, at higher temperatures in which SUS1 function is more critical, pre-mRNA forms retaining the I1 or both introns (I1 and I2) are more abundant (Figure 2B). This change in the transcript ratios could account for the lack of fully complementation by SUS1-cDNA at 37°C (Figure 7B and C). In this context a striking observation is the existence of a small protein containing part of Sus1, likely translated form a transcript containing I1 (Figure 5). Western analyses are consistent with this peptide being produced, in agreement with the functional requirement of transcripts containing I1, suggesting a role of the I1 SUS1 transcripts. Our results are consistent with the notion that the small Sus1 peptide is not a proteolytic product of the full-length protein. First, it accumulates in mud2Δupf1Δ cells, where full-length Sus1 production is reduced (Figure 5); and second, the small Sus1 product is absent in cells bearing an intron-less SUS1, where full-length Sus1 is enhanced (Figure 7). An alternate explanation, invoking a selective degradation of full-length Sus1 when encoded by an intron-containing SUS1 gene, is difficult to conceptualize and not consistent with the abundance of the small peptide in cells bearing the SUS1 SL-9 BS mutant, which cannot be spliced and thus cannot encode full-length Sus1. Notably, the functional requirement of transcripts containing I1 at high temperature is suggestive of a biological role for this unprocessed SUS1 transcript. More work is needed to assign this functionality to the pre-mRNA, its product, or both.
Our results, together with those from the accompanying report of Hossain et al., provide compelling evidence indicating that expression of SUS1 hinges on a balance of several factors including limited splicing and degradation pathways, modulated by intronic and exonic sequences. The question now emerges as to how this complex strategy gives SUS1 its relevant role in multiple aspects of mRNP biogenesis.
Supplementary Data are available at NAR Online.
Funding for open access charge: Spanish Ministry of Science and Innovation (BFU2008-04082-C02-02 and BFU2008-00363); Generalitat Valenciana (ACOMP2010/255 and ACOMP2011/057) and Ramon y Cajal/IP3 contracts (MEC) (to S.R-N. and J.V.); Formación de Personal Investigador (FPI) and Formación de Personal Universitario (FPU) Predoctoral fellowships from MICINN and MEC (to B.C.B. and V.G.M.). respectively.
Conflict of interest statement. None declared.
The authors are grateful to Tracy Johnson and Munshi Azad Hossain for sharing unpublished results and for their fruitful comments and advices on our work. The authors thank N. Gómez, Dr. Martinez-Jimenez and all I-25 lab members for scientific comments and technical support.