RESULTS

Prp40p contains two regions with homology to the double-stranded RNA-binding domain (dsRBD; St. Johnstonet al. 1992). In other proteins, these motifs bind double-stranded RNA (dsRNA) in vitro in a sequence-independent manner (Bass 1995). However, all attempts to show RNA binding by Prp40p have proven unsuccessful (data not shown). Therefore, we call these sequences in Prp40p region 1 and region 2 (Figure 1A). To investigate the functions of these regions, we made in-frame deletions of each. Precise deletion of region 1 caused lethality (data not shown) while deletion of region 2 resulted in a slight growth defect at all temperatures (Figure 1B and data not shown). This slow-growth phenotype indicates that region 2 is important but not essential for some aspect of Prp40p function.

A synthetic lethal screen for genes that interact with PRP40: To investigate the function of region 2, we sought to identify genes whose products interact with this region. To do so, a synthetic lethal screen was performed using the region 2 deletion (prp40Δ2) as a starting mutation. Synthetic lethal screens detect the synergistic defects that result when a cell contains mild mutations in two functionally related genes (Franket al. 1992; Liaoet al. 1993). Thus, a synthetic lethal interaction between two genes is predictive of important functional interactions between those genes.

We started with the strain 40SS, in which the chromosomal copy of PRP40 was disrupted and complemented by wild-type PRP40 carried on a URA3 plasmid. The strain also contained a second plasmid carrying TRP1 and prp40Δ2. This strain was mutagenized with UV light and the surviving cells were allowed to form colonies. These colonies were then replica plated to 5-FOA-containing media to select cells that had lost the PRP40 URA3 plasmid. These cells depended on prp40Δ2 as their sole source of Prp40p. If any of these cells contained a UV-induced genomic mutation that was synthetically lethal with prp40Δ2, they did not survive the 5-FOA treatment. We identified these candidates and recovered them from the pre-5-FOA replica plate.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

—Domains of Prp40p. (A) Domains of Prp40p are indicated with their positions in the amino acid sequence given. WW represents the WWP domain, ARM represents the arginine-rich motif. (B) The phenotypes of prp40 mutations are given. Region 2 K4-A4 changes four lysine residues to alanines. ARM deletion deletes the sequence 348-RLRNYTRDRIAR-359. (C) Comparison of Prp40p and Ynl187wp NES sequences to the NES consensus sequence (Bogerdet al. 1996). Mutant sequences show amino acid substitutions used in this study.

To ensure that the UV mutations caused specific, synergistic defects with prp40Δ2, only UV-induced mutations that had no growth defect in the presence of wild-type PRP40 were pursued (Table 2). By studying only those UV-induced mutations that caused no detectable phenotype on their own, we avoided false positives caused by the additive phenotypes of unrelated mutations. For this test, strains from the pre-5-FOA plates were evaluated. These strains still had a wild-type copy of PRP40 and displayed only the defect caused by the UV-induced mutation. The rate of colony formation of these synthetic lethal strains was compared to that of a wild-type strain at five different temperatures: 18°, 25°, 30°, 33°, and 37°. Nine strains had UV-induced mutations that caused no phenotype in the presence of wild-type PRP40 but died in the absence of region 2. These were designated slf mutants 1–9 (Table 2).

A trans test was used to prove that the UV-induced mutations were in chromosomal genes rather than in the prp40Δ2 plasmid. The prp40Δ2 plasmid was lost from candidate synthetic lethal strains and a fresh copy of the same plasmid was reintroduced. As expected, all candidates still exhibited synthetic lethality after 5-FOA treatment (data not shown). In addition, a test for 5-FOA hypersensitivity was used to demonstrate that the synthetic lethality was not due to a mutation that made the cells more susceptible to this drug; transformation of a PRP40 TRP1 plasmid allowed all candidates to survive on 5-FOA.

Each candidate was then backcrossed to the unmutagenized parent and tetrads were dissected to ensure that the synthetic lethal phenotype resulted from a single mutation. The diploid strain was also used to determine if the synthetic lethal allele was recessive to its wild-type allele (Table 2). Finally, we placed the synthetic lethal mutants into complementation groups by mating them to each other and testing for synthetic lethality in the diploid strain on 5-FOA. Seven candidates, encompassing five different complementation groups, passed all tests (Table 2). Two complementation groups were dominant while the other three were recessive (Table 2).

Characterization and cloning of the slf mutants: The slf mutants were evaluated for their phenotypes in combination with other alleles of prp40 (Table 2). Such allele specificity tests can reveal how the slf genes interact with PRP40. For example, an slf mutation that affects the stability or synthesis of Prp40p might be synthetically lethal with many prp40 alleles. In contrast, if the slf protein interacts specifically with one region of Prp40p, it would be expected to cause synthetic lethality only with prp40 alleles that affect that region. Each slf mutant was tested with the prp40-K4-A4 and prp40ΔARM alleles. prp40-K4-A4 has four lysines in region 2 changed to alanines, while prp40ΔARM has a 12-residue deletion of a region containing five arginines (Figure 1B). These tests defined at least four different allele specificity profiles among the seven slf strains (Table 2). Importantly, no slf mutant displayed synthetic lethality with all alleles of prp40 tested, indicative of specific interactions.

Suppressors of the recessive synthetic lethal mutations were cloned by complementation of the synthetic lethal phenotype by a CEN-based genomic library. Complementing plasmids were recovered, sequenced, and the individual open reading frames present were individually subcloned to determine which ones conferred complementation. In this way, we identified SME1, CRM1, and YNL187w as suppressors of the synthetic phenotypes caused by the three recessive synthetic lethal complementation groups. SME1 suppressed slf7, CRM1 suppressed slf4 and slf8, and YNL187w suppressed slf2 and slf9 (Table 2). We then sequenced the genomic copies of CRM1 from the slf4 and slf8 strains and YNL187w from the slf2 and slf9 strains to determine if they contained the synthetic mutations. ynl187w contained an early frameshift mutation at codon 55 (data not shown). Furthermore, deletion of ynl187w, in an otherwise wild-type background, conferred synthetic lethality with prp40Δ2, suggesting that slf2 and slf9 are ynl187w (data not shown). Interestingly, CRM1 contained no mutations in its coding sequence or in its promoter. Therefore, we conclude that mild overexpression of CRM1 confers dosage suppression of the slf4 and slf8 mutations. We did not sequence SME1 from slf7.

Sme1p is an essential splicing factor and a component of the yeast U1, U2, U4, and U5 snRNPs (Bordonne and Tarassov 1996). Crm1p is an essential nuclear export factor that recognizes leucine-rich NES sequences (Fornerod et al. 1997a,b; Nevilleet al. 1997; Stadeet al. 1997). Interestingly, YNL187w is a nonessential gene encoding a consensus leucine-rich NES sequence (Figures 1C and 2C; Giaeveret al. 2002). Therefore, two genes that have a link to nuclear export display genetic interactions with region 2 of Prp40p. In metazoan cells, using an adapter protein, Crm1p exports U snRNAs from the nucleus to the cytoplasm during snRNP biogenesis (Fornerodet al. 1997a; Ohnoet al. 2000). Although we did not expect compartmentalized snRNP biogenesis to occur in yeast, we investigated whether region 2 of Prp40p was required for some nuclear export event.

Prp40p contains NES sequences: Region 2 of Prp40p contains a leucine-rich sequence that fits the NES consensus sequence (Figure 1C). This sequence (274-L k e L r e y L n g I-284), called NES1, differs from the consensus (L x_(2-3) L x_(2-3) L x L/I; Bogerdet al. 1996) only in the spacing of the last two hydrophobic residues. Intriguingly, a second sequence (340-L q n k L n e L r L-349; NES2) that fits the NES consensus lies 58 residues downstream of region 2 in Prp40p (Figure 1C).

Since NES1 lies primarily within region 2, we tested whether the loss of NES1 was responsible for the growth defect of prp40Δ2. Conservative amino acid changes in the leucine residues of NES1 (Figure 1C) were made and their effect on growth was evaluated. As seen in Figure 2A, the prp40-nes1 mutant strain (prp40-nes1 SLF) displayed a growth defect relative to wild type (PRP40 SLF) at 36°. However, the loss of NES1 did not account for all of the function lost when region 2 was deleted, as prp40-nes1 grew better than prp40Δ2 (data not shown). Nonetheless, since the prp40-nes1 strain exhibited a growth defect at elevated temperatures, we conclude that NES1 is important for Prp40p function. Furthermore, we conclude that region 2 contains amino acids involved in a separate function.

The PRP40 NES sequences interact with SLF4 and YNL187w: We next tested whether the loss of NES1 was responsible for the synthetic lethality observed when prp40Δ2 is combined with slf4. In the presence of wild-type PRP40, slf4 mutants grew at the same rate as wild-type cells at 36° (Figure 2A, compare PRP40 SLF4 and PRP40 slf4). As described above, the prp40-nes1 mutant strain displayed a growth defect at 36° (Figure 2A, prp40-nes1 SLF4). However, when the prp40-nes1 and slf4 mutations were combined in the same strain, a much stronger 36° growth defect resulted (Figure 2A, prp40-nes1 slf4). We note that this phenotype did not completely mimic the lethal phenotype of prp40Δ2 slf4, presumably because prp40Δ2 affects another function of Prp40p. Nonetheless, this synergistic growth defect demonstrates a functional interaction between SLF4 and the NES1 sequence of Prp40p. Furthermore, dosage suppression of the prp40Δ2 slf4 synthetic lethality by CRM1 suggests a functional link between the NES1 sequence and CRM1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

—The NES sequences of PRP40 interact genetically with SLF4, YNL187w, and themselves. (A) Yeast strains with different combinations of PRP40 or prp40-nes1 and SLF4 or slf4 were tested for growth at 36° on YEPD plates. (B) Yeast strains with PRP40, prp40-nes1, prp40-nes2, or prp40-nes1/nes2 were tested for growth at 33° on YEPD (SLF4 was wild type in all cells). (C) Yeast strains with PRP40 or prp40-nes1 were tested for growth at 36° in the presence or absence of YNL187w. The genotypes are indicated around each plate.

We also tested whether NES2 was important for Prp40p function. Conservative amino acid changes in the leucine residues of NES2 were made and their phenotypes were tested (Figure 1C). Surprisingly, strains with the prp40-nes2 mutation displayed no growth defect at any temperature. For example, growth at 33° is shown in Figure 2B. It is conceivable that the lack of a mutant phenotype after mutation of NES2 is due to the difference in severity of mutations between NES1 and NES2 (Figure 1C). However, mutation of two alanines is sufficient to disrupt the interaction of other NESs with Crm1p (Wenet al. 1995). The simplest explanation for this result is that NES1 and NES2 are functionally redundant. We tested this hypothesis by constructing the prp40-nes1/nes2 double mutant. This mutant allowed us to test for cis synthetic lethality between these two regions. When grown at 33°, no growth defect was observed when either NES was mutated singly (Figure 2B, prp40-nes1 and prp40-nes2). However, when mutations in both NESs were combined in cis, a strong growth defect at 33° resulted (Figure 2B, prp40-nes1/nes2). This synergistic defect suggests a partial functional redundancy between the two NES sequences. Although it is unknown whether these growth defects result from gross changes in protein structure, two observations suggest they do not. First, the leucine-to-alanine changes are conservative mutations not likely to disrupt global protein structure. Second, Western analysis showed that prp40p-GFP alleles were expressed at similar levels (data not shown).

If NES1 and NES2 were completely functionally redundant, mutation of NES1 would cause no phenotype as long as NES2 were wild type. However, mutation of NES1 did cause a significant growth defect at 36° (Figure 2A; compare PRP40 SLF4 and prp40-nes1 SLF4). We found no synthetic lethal interactions between prp40-nes2 and slf4 (data not shown). The simplest model to explain these functional interactions predicts that NES1 and NES2 act as independent and redundant export signals; mutation of either NES alone does not prevent export but mutation of both NES sequences does.

Since an allele of ynl187w was also synthetically lethal with prp40Δ2, and because Ynl187wp is predicted to also contain a leucine-rich NES sequence, we tested whether the NES sequences of PRP40 interacted genetically with YNL187w. As expected, deletion of ynl187w caused no growth defect in the presence of wild-type PRP40 (Figure 2C, PRP40 ynl187wΔ; Giaeveret al. 2002). As noted above, prp40-nes1 caused a growth defect at 36° in the presence of YNL187w (Figure 2, A and C). Interestingly, however, deletion of ynl187w in the prp40-nes1 strain exacerbated this phenotype (Figure 2C, prp40-nes1 ynl187wΔ). Similarly, a triple mutant, in which ynl187w was deleted in the prp40-nes1/nes2 strain, displayed a growth defect more severe than any combination of two of these mutations (data not shown). These synergistic defects are suggestive of a functional interaction between ynl187w and the NES sequences of Prp40p. We have not, however, attempted to map the YNL187w region of functional interaction to the putative leucinerich NES.

We next tested for a physical interaction between Prp40p and Crm1p, using both glutathione S-transferase pull downs and a yeast two-hybrid assay. In extensive trials, no specific interaction was observed. Although a weak interaction was detected by two-hybrid assay, it was not dependent on intact NES sequences. Although addition of U1 snRNA and Gsp1p to pull-down reactions had no effect, we believe that other components are likely required for a specific interaction between the NES sequences and Crm1p. Thus, we instead attempted to test the function of the NES sequences of Prp40p.

The Prp40p NES sequences are required for splicing: The presence of NES sequences in the essential splicing factor PRP40 suggests that nuclear export of Prp40p, the U1 snRNP, or other splicing factors may be required for splicing. To test this, we utilized the temperature-sensitive phenotype caused by mutation of NES1 to perform an in vivo splicing assay. Cells were grown at a permissive temperature (30°) to midlog phase and then shifted to restrictive temperature (37°). Total RNA was isolated from the cells prior to and after temperature shift and the level of splicing of two RNAs was measured by primer extension at different intervals after temperature shift. Using a primer that annealed to the second exon of the intron-containing U3 RNA allowed us to distinguish unspliced RNA from spliced RNA (Figure 3A, lane A). The mature U3 RNA is transcribed and processed from two genes, each containing a different size intron that is removed by the spliceosome (Myslinskiet al. 1990). Thus, primer extension reveals two different-sized pre-U3 RNAs but only one mature U3 RNA (Figure 3A, lane A). Temperature shift caused no increase in unspliced RNA levels in a wild-type strain (Figure 3A, lanes A–E) but caused a dramatic increase in unspliced RNA levels in both the prp40-nes1 and the prp40-nes1/nes2 mutant strains (Figure 3A, lanes F–J and K–M, respectively). Such an increase in unspliced RNA is indicative of a reduction in splicing efficiency. Therefore, we quantified the RNA levels of each species by densitometry. The ratio of the amount of the shorter spliced species to the amount of the longer unspliced species yielded a relative measurement of the splicing efficiency for that sample. At permissive temperatures, both prp40-nes1 and prp40-nes1/nes2 mutations displayed similar splicing efficiencies as an isogenic wild-type strain (Figure 3B, open bars). After shift to 37°, however, both prp40-nes1 and prp40-nes1/nes2 mutations displayed significantly reduced splicing efficiencies (Figure 3B, prp40-nes1 and prp40-nes1/nes2, shaded and solid bars) while the splicing of an isogenic wild-type strain remained unchanged (Figure 3B, PRP40, shaded and solid bars). Similar results were observed for the splicing efficiency of RPL30 mRNA (data not shown). Mutation of the NES sequences mimicked the splicing defect observed in a strain expressing prp40-10, an allele containing a nonsense mutation just downstream of NES2 at W379 (Figure 3A, lanes N–P; Figure 3B, prp40-10; Kao and Siliciano 1996). Thus, disruption of the NES1 sequence in Prp40p caused not only a temperature-sensitive growth defect but also a temperature-sensitive splicing defect. These data suggest that prp40-nes mutants die at elevated temperatures because they are defective at splicing their mRNAs. These data also suggest that nuclear export of Prp40p is required for pre-mRNA splicing. It is possible, however, that the leucine residues of the NES sequences perform auxiliary roles in the splicing reaction. The splicing defect resulting from mutation of the NESs therefore may result indirectly from the disruption of Prp40p export, directly from disruption of the splicing reaction, or from disruption of both.

We further tested whether crm1 or ynl187w mutations also conferred splicing defects. However, the deletion of neither ynl187w nor a temperature-sensitive allele of crm1 (xpo1-1) conferred a splicing defect in vivo (data not shown; Stadeet al. 1997).

The Prp40p NES sequences affect subcellular localization: To test if the NES sequences of Prp40p function as true nuclear export signals, we fused the gene encoding green fluorescent protein (GFP) to the carboxy terminus of PRP40. We first expressed this fusion protein from its own promoter on a CEN-based plasmid. Unfortunately, expression was not high enough to visualize any Prp40p-GFP. We then constructed a 2-μm plasmid that expressed PRP40-GFP from the GPD promoter. This high expression construct allowed Prp40p-GFP to be visualized easily. Importantly, both the low-expression and high-expression constructs fully complemented a prp40 disruption (data not shown).

We investigated the localization of Prp40p-GFP by fluorescence microscopy and, as expected of an essential splicing factor, it was almost completely nuclear (Figure 4A). We conclude that any nuclear export of Prp40p by its NES sequences must be limited in some manner. Export may be a brief or rare event, it may involve only a small fraction of the cellular Prp40p pool, or it may lead to Prp40p's immediate degradation. In an attempt to uncouple Prp40p from this limiting factor or event, we deleted amino acids 1–256, leaving only the carboxy-terminal half of Prp40p fused to GFP. This construct, prp40Δ256-GFP, contained both NES sequences and several putative nuclear localization sequences but lacked essential amino-terminal domains. This construct, therefore, could not complement a prp40 disruption but could be expressed in a wild-type background.

By fluorescence microscopy, prp40Δ256p-GFP was localized to the nucleus but unlike Prp40p-GFP, a significant amount of the protein was also detected in the cytoplasm (Figure 4B). This suggests that prp40Δ256p-GFP was shuttling between the nucleus and the cytoplasm. The strong nuclear signal suggests that either import was more active than export or prp40Δ256p-GFP was more stable in the nucleus. Mutation of the NES sequences in prp40Δ256p-GFP caused a significant nuclear accumulation of the fusion protein, indicating that these sequences were required for nuclear export (Figure 4, compare B to C, D, and E). The intensity of GFP fluorescence in the nucleus and in the cytoplasm was calculated for 50–100 cells of each allele of prp40Δ256-GFP. When the ratio of GFP detected in the nucleus to that detected in the cytoplasm was compared between alleles, the prp40Δ256-nes1/nes2 double mutant had a twofold greater ratio than the prp40Δ256 allele (Figure 4F). This difference was statistically significant (P = 3 × 10^–31) as determined by an F-test. Importantly, mutation of either NES1 or NES2 alone also caused a statistically significant nuclear accumulation (P = 4 × 10^–24 and P = 0.005, respectively) with NES1 mutations causing the most dramatic nuclear accumulation. These results clearly demonstrate that the NES sequences in prp40Δ256p-GFP function in subcellular localization. Although it is possible that this shuttling occurs only with the carboxy-terminal half of Prp40p, we expect that the NES sequences also direct the export of full-length Prp40p because a significant amount of Prp40p sequence still surrounds the NESs in prp40Δ256p-GFP.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

—Mutation of the NES sequences of Prp40p caused a splicing defect in vivo. (A) Primer extension analysis revealed the levels of pre-U3 RNA and mature U3 RNA isolated from strains expressing wild-type or temperature-sensitive alleles of prp40 grown at 30° (0 hr) or shifted to 37° for 2, 4, 8, or 12 hr. The time points for each lane are denoted above the gel. The two top bands correspond to pre-U3 RNA arising from two independent genes differing in intron size (Myslinskiet al. 1990). A primer for the U5 snRNA was added as an internal control (lanes A–P). Lane Q is a control lane containing only the U3 RNA primer and lane R contains digested, ³²P-labeled pBluescript as a size marker. The asterisk (*) denotes a band arising from nonspecific priming by the U5 snRNA primer as it does not appear in lane Q. (B) The relative splicing efficiency of different strains before and after temperature shift as calculated by dividing the total mature U3 RNA level by the total pre-U3 RNA level in each lane and dividing by 1000. Open bars, 30°; shaded bars, 2 hr at 37°; solid bars, 4 hr at 37°.

prp40Δ256p-GFP subcellular localization is affected by Crm1p: Leptomycin B is a cytotoxin that specifically inhibits Crm1p function in many organisms including Xenopus laevis and S. pombe (Nishiet al. 1994; Wolffet al. 1997). While S. cerevisiae Crm1p is resistant to this drug, a single-point mutation (creating a cysteine at amino acid 539) renders the S. cerevisiae protein leptomycin B sensitive (Kudoet al. 1999; Neville and Rosbash 1999). We visualized the subcellular distribution of prp40Δ256p-GFP in a S. cerevisiae strain sensitive to leptomycin B (Lmb4). This strain displayed wild-type growth on rich media but died in the presence of 20–100 ng/ml leptomycin B (data not shown). Cells were grown to midlog phase prior to the addition of either carrier or 100 ng/ml leptomycin B for 45 min. Fluorescence microscopy revealed that leptomycin B caused a significant nuclear accumulation of prp40Δ256p-GFP as compared to carrier alone (Figure 5, compare A with B). This difference is statistically significant as determined by an F-test (P = 6 × 10^–19; Figure 5C). As described in other instances, nuclear accumulation was observed as early as 15 min after cytotoxin addition (data not shown). Thus, inhibition of Crm1p mimicked the effect of NES mutations in prp40Δ256p-GFP (compare Figure 5B with Figure 4E). As expected, no effect was observed upon the addition of leptomycin B to a wild-type CRM1, S. cerevisiae strain (data not shown). These results demonstrate that Crm1p function is necessary for the proper subcellular localization of prp40Δ256p-GFP.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

—Prp40p NES sequences function as nuclear export signals. (A) The subcellular localization of Prp40p fused to GFP is almost exclusively nuclear as determined by fluorescence microscopy. (B–E) The carboxy-terminal half of prp40 alleles was fused to GFP and overexpressed in wild-type cells. The subcellular localization of each protein was determined by fluorescence microscopy. A significant amount of cytoplasmic fluorescence was observed with constructs containing wild-type NESs (B) while constructs containing mutant NESs (C–E) were almost entirely nuclear. Shown to the right in B–E are DAPI fluorescence pictures of the same fields of view to locate nuclei. (F) The average ratio of the intensity of the nuclear to the cytoplasmic GFP in 50–100 cells. The variances between the NES mutants and the wild-type samples were compared by an F-test and the P values are shown.