To identify factors that interact with TREX-2, a transcription-export complex, which functions at the inner side of the NPC, we performed a synthetic lethality (sl) screen with the sac3Δ strain. Among the candidates, we identified a mutant allele of the SWT1/YOR166 gene. The genetic interaction was verified by absence of growth of a haploid sac3Δ swt1Δ strain, generated by plasmid loss (Figure S1). Other TREX-2 components Thp1 and Sus1 were either sl (thp1Δ) or showed strong synergistic interaction (sus1Δ) with swt1Δ, demonstrating an intimate functional interaction between Swt1 and TREX-2 complex (Table 1 and Figure S1).

To extend these studies, we tested whether SWT1 functionally overlaps with other factors involved in mRNP biogenesis and turnover. Strong genetic interactions were observed between swt1Δ and THO/TREX components (tho2Δ, hpr1Δ, mft1Δ, thp2Δ, and sub2–85) consistent with previous findings [18]. In contrast, no genetic interactions were detected between swt1Δ and the mRNA export receptor Mex67 or adaptor Yra1 [18]. In addition, we found an sl interaction between swt1Δ and NPL3, which encodes an mRNA binding protein with multiple roles in mRNP biogenesis [25,26] (Table 1 and Figure S1).

These studies also uncovered genetic interactions between SWT1 and MLP1, NUP60, and ESC1 (Table 1 and Figure S1). Mlp1 and Nup60 are NPC-associated factors that are thought to constitute the nuclear basket of NPC and participate in mRNP quality control before nuclear export [14]. Esc1 is a protein of the inner side of the nuclear membrane recently shown to affects the association of the Mlp1-Nup60 complex with the NPC [17].

Several factors implicated in various steps of mRNP production, export and surveillance did not interact genetically with swt1Δ, including subunits of the nuclear exosome (Rrp6 and Rrp44/Dis3), components of the TRAMP complex (Trf4 and Mtr4), the Rat1 5′-exonuclease, and factors involved in cytoplasmic mRNA turnover (see Table 1 for factors tested). These genetic analyses indicate that Swt1 interacts specifically with factors required for cotranscriptional assembly and perinuclear quality control of nuclear mRNPs.

The PIN domain of Swt1, which lies in the central region (residues 130–262), is evolutionarily conserved and contains all four conserved acidic residues implicated in catalysis (Figure S2). We therefore tested whether Swt1 purified from yeast has in vitro nuclease activity. We generated FLAG-tagged wild-type Swt1 and an Swt1 mutant expected to be catalytically inactive (Swt1-D135N; Asp 135 is predicted to be a residue of the catalytic center). These were expressed in yeast and purified by affinity chromatography (Figure 1A). The ability of these proteins to degrade single-stranded RNA in vitro was assessed using 5′- or 3′-radiolabeled oligo(A)₃₀ substrates. Wild-type Swt1 efficiently degraded both 5′- and 3′-labeled oligo(A)₃₀ RNA. In contrast, the Swt1-D135N mutant protein exhibited no activity on the 3′-labeled RNA substrate and only very weak activity on the 5′-labeled probe (Figure 1B). The residual activity on the 5′-labeled substrate could be due to a contaminating 3′-5′ exonuclease that is frequently observed in exonuclease assays (C. Schneider, unpublished data). The lack of degradation of the 3′-labeled substrate by this activity probably reflects protection by the presence of a 3′-phosphate on the [³²P]pCp-labeled RNA. Time course experiments further showed that wild-type Swt1 degrades both 5′- and 3′-labeled substrates with very similar kinetics (Figure 1C), consistent with endoribonuclease activity.

To assess whether Swt1 functions as an endonuclease, a structured 52-nucleotide (nt) RNA substrate was used, which contains a stable stem (from the mouse 5.8S rRNA) that closes the 5'- and 3'-ends of the RNA (Figure 1D). The remaining part of the RNA (nt 11–43) is predicted to be at least partially single stranded. Wild-type Swt1, but not the Swt1-D135N mutant, efficiently cleaved the structured 52-nt RNA endonucleolytically, with a clear preference for single-stranded regions of the RNA substrate (Figure 1D). Taken together, the data show that Swt1 exhibits in vitro endonuclease activity without clear sequence specificity but with a preference for single-stranded RNA.

To further investigate the in vivo role of Swt1 in the context of transcription-coupled mRNA export and nuclear mRNP surveillance, we performed mRNA export and pre-mRNA leakage assays with the viable double mutants (Figure S1) carrying swt1Δ together with thp2Δ, mft1Δ (THO/TREX), sus1Δ (TREX-2), nup60Δ, mlp1Δ, and esc1Δ (perinuclear mRNP quality control). Deletion of SWT1 significantly enhanced nuclear poly(A)⁺ RNA accumulation relative to the thp2Δ, mft1Δ, sus1Δ, and nup60Δ single mutants (Figures 2A, S3A and unpublished data). Moreover, swt1Δ induced nuclear accumulation of poly(A)⁺ RNA in the mlp1Δ and esc1Δ mutants, which was not detected in the single mutants (Figure 2A; see also [27,28]).

To assess whether the poly(A)⁺ RNA accumulation reflects the accumulation of nuclear mRNA, we tested a specific transcript, the heat-shock SSA1 mRNA (Figure 2A). This mRNA was clearly accumulated in the nucleus of the mpl1Δ swt1Δ double mutant, but not in the corresponding single mutants. In contrast, no nuclear accumulation was observed when the nuclear export of tRNA was assayed (unpublished data). To quantify whether the levels of SSA1 transcript were increased in the mlp1Δ swt1Δ double mutant, we measured the amount of SSA1 mRNA before and after heat shock by real-time reverse-transcription polymerase chain reaction (RT-PCR). The SSA1 levels were mildly increased in the single mutants, but a synergistic increase was observed in the mlp1Δ swt1Δ double mutant in comparison to wild-type strain (Figure S4).

To test whether the double mutants of swt1Δ and mlp1Δ or nup60Δ result in synergistic defects in pre-mRNP surveillance, we monitored cytoplasmic leakage of unspliced pre-mRNA reporters [11,12,14]. No leakage of the reporter with intact splicing signals (pJCR1) was observed in the swt1Δ strain, whereas the nup60Δ strain showed increased leakage of the unspliced pre-mRNA, which was synergistically enhanced in the swt1Δ nup60Δ double mutant (Figure 2B). Next, we tested whether pre-mRNA reporters with a mutation in the 5′ splicing site (mut5′SS) or in the branch-point site (mutBP) exhibit an increased cytoplasmic leakage in swt1Δ cells. Notably, leakage of the mutBP reporter was increased approximately 4-fold in the swt1Δ or mlp1Δ single mutants relative the wild type (Figure 2B; see also [14,16]). Cytoplasmic leakage of both mutant pre-mRNA reporters was again strongly synergistically enhanced in the swt1Δ mlp1Δ double mutant (Figure 2B).

We conclude that loss of Swt1 leads to increased cytoplasmic leakage of splicing-defective pre-mRNAs, which can be synergistically enhanced by mutations in the Mlp1-Nup60 complex. This suggests that Swt1 and Mlp1-Nup60 have distinct but overlapping roles in nuclear mRNP surveillance. At first sight, it appears contradictory that an increased pre-mRNA leakage goes along with nuclear accumulation of poly(A)⁺ RNA. However, these two effects are necessarily not mutually exclusive. When nuclear quality control is severely compromised (e.g., in mlp1Δ swt1Δ or nup60Δ swt1Δ cells), larger amounts of aberrant mRNPs could accumulate in the nucleus and hence have a higher chance to leak out.

In a previous study [18], Swt1 lacking the PIN (endonuclease) domain was shown to complement the lethality of swt1Δ in THO mutant strains. However, Swt1 is a low-abundance protein (see below), and the constructs were overexpressed from the relatively strong NOP1 promoter. To assess whether the endonuclease activity of Swt1 is essential for complementation of swt1Δ phenotypes under physiological expression levels, we analyzed the catalytically inactive Swt1-D135N mutant expressed from the endogenous promoter (Figure 3A). Swt1-D135N and wild-type Swt1 were expressed at similar levels (Figure 3B), but Swt1-D135N failed to complement the growth defects of swt1Δ in any double-mutant combination (Figure 3A and unpublished data). Moreover, Swt1-D135N did not rescue the nuclear mRNA accumulation induced by swt1Δ in the background of the mlp1Δ, nup60Δ, sus1Δ, or thp2Δ mutations (Figure 3C and unpublished data). Finally, the nuclear leakage of the unspliced pre-mRNA reporter (mutBP) in Swt1-D135N–expressing cells (Figure 3D) was indistinguishable from the leakage observed in the swt1Δ strain (Figure 2B). Similar loss-of-function phenotypes were observed for the swt1 truncation mutant lacking the entire PIN domain (Swt1ΔPIN) (Figure S5 and unpublished data).

However, Swt1-D135N, Swt1ΔPIN, and other partial deletion constructs could complement swt1Δ for growth in the tho2Δ background, when overexpressed under the control of the strong ADH1 promoter (Figure S5). In contrast, other swt1Δ phenotypes, including nuclear pre-mRNA leakage and nuclear poly(A)⁺ RNA accumulation in mlp1Δ swt1Δ strains, could not be suppressed by overexpression of Swt1-D135N or Swt1ΔPIN mutants from the ADH1 promoter (Figure 3D and unpublished data). Thus, under physiological expression levels, the endonuclease activity of Swt1 is essential for its function, whereas the requirement for catalytic activity can be partially relieved by overexpression of the inactive Swt1 protein.

To assess possible roles for the endonuclease activity of Swt1 in the degradation of defective mRNPs, we determined whether an “aberrant” transcript is stabilized in swt1-D135N cells relative to the otherwise isogenic wild type. As an abnormal transcript, we used the intron-containing lacZ construct (expressed from mutBP plasmid; see above) that was used in the pre-mRNA leakage assays. This harbors a mutation in the splicing branch-point site that efficiently blocks pre-mRNA splicing [12]. The wild-type intron-containing lacZ transcript encoded by plasmid pJCR51 was used as a control pre-mRNA. Accumulation of the aberrant transcript was substantially elevated in swt1-D135N cells relative to wild-type strain (mut. bars in Figure 3E), whereas the levels of normally processed lacZ mRNA was unchanged between the swt1-D135N and wild-type strains (norm. bars in Figure 3E). These results suggest that the endoribonuclease activity of Swt1 participates in the selective elimination of defective mRNA transcripts.

As Swt1 participates in nuclear mRNP surveillance, its expression and/or targeting may be subject to regulation. Quantitative western blot analysis using anti-Swt1 antibodies indicated that in vivo levels of Swt1 endonuclease are very low (∼200–300 molecules per cell) (unpublished data; see also [18]), which might be important to avoid inappropriate RNA degradation. In a large-scale analysis, Yor166/Swt1 was reported to be toxic following overexpression by the Tet-off system [29]. Consistent with this finding, a dominant-lethal phenotype was induced by overexpression of wild-type Swt1 but not of catalytically inactive mutant Swt1-D135N from the GAL1 promoter (Figure 3F). The majority of Swt1 overexpressing cells showed strong nuclear accumulation of both poly(A)⁺ RNA (Figure 3E) and SSA1 mRNA, whereas tRNA was normally exported (Figure S3B and unpublished data). Notably, when the localization of intron-containing/intronless lacZ reporter constructs was analyzed, the nuclear accumulation of an intron-containing pre-mRNA reporter was stronger than for the intronless mRNA (Figure S3C), consistent with a defect in pre-mRNA surveillance. These analyses demonstrate that overexpression of catalytically active Swt1 is toxic and interferes with mRNA export and/or pre-mRNA surveillance.

Subcellular fractionation previously indicated that TAP-tagged Swt1 is located in the nucleus [18]. The exact location of Swt1 could not be determined, however, because the abundance of green fluorescent protein (GFP)- or immuno-labeled Swt1 expressed from the endogenous promoter was too low to be visualized (unpublished data and [18]). The attachment of multiple GFP tags can allow the detection of low-abundance proteins (N. Daigle, A. Bancaud, and J. Ellenberg, unpublished data). We therefore generated a functional 4xGFP-Swt1 construct expressed from SWT1 endogenous promoter, which complemented the double-deletion strains swt1Δ mlp1Δ, swt1Δ nup60Δ, or swt1Δ tho2Δ (Figure S6 and unpublished data). When analyzed in the swt1Δ strain, 4xGFP-Swt1 was localized in both the nucleus and cytoplasm, with a slight nuclear enrichment in some cells (Figure 4A). When expressed in the nup133Δ nuclear pore-clustering mutant [30], 4xGFP-Swt1 exhibited significant coclustering with NPCs at the nuclear periphery, which was lost upon complementation of the nup133Δ strain by plasmid-derived NUP133 (Figure 4A). When a 4xGFP-Swt1 construct lacking the catalytic PIN domain (4xGFP-Swt1ΔPIN) was expressed in swt1Δ cells, it was concentrated in a few spots. These were always close to the NPCs, as revealed by covisualization of Nup120-mCherry (Figure 4B), and 4xGFP-Swt1ΔPIN also associated with NPCs in nup133Δ cells. Targeting of Swt1ΔPIN to the NPCs did not depend on Mlp1, Nup60, or Esc1. Instead, 4xGFP-Swt1ΔPIN showed more distinct and punctuate nuclear envelope staining in the mlp1Δ, nup60Δ, and esc1Δ strains than in the corresponding isogenic wild-type strains (Figure 4B and unpublished data). The nucleocytoplasmic localization of full-length 4xGFP-Swt1 in the mlp1Δ, nup60Δ, and esc1Δ cells was not altered (unpublished data). Together with the genetic data, these observations indicate that Swt1 interacts with the NPCs.

Yeast strains used in this study are listed in Table S1; plasmid constructs are described in Table S2. Site-directed mutagenesis of the SWT1 gene to create the swt1-D135N mutant was performed by fusion PCR [32]. The synthetic lethality screen with the sac3Δ strain was performed as previously described [33].

For in vitro measurements of Swt1 nuclease activity, the FLAG-Swt1 wild-type and D135N mutant protein were expressed from plasmids pADH111-FLAG-SWT1 or pADH111-FLAG-swt1-D135N (ADH1 promoter) and purified by anti-FLAG affinity chromatography. Briefly, 4 l of yeast culture were harvested at an optical density at 600 nm (OD₆₀₀) of approximately 4.0, resuspended in FLAG buffer (50 mM Tris-HCl [pH 7.5], 200 mM NaCl, 0.5 mM DTT, 0.05% NP-40) to a final volume of 50 ml, lysed by bead-beating (Pulverissete 6; Fritsch) and incubated with 120 μl of anti-FLAG M2 agarose (Sigma) for 120 min at 4 °C. Anti-FLAG slurry was washed by 750 volumes of FLAG buffer, and the bound material was eluted with FLAG peptide according to the manufacturer's instructions (Sigma) followed by concentration and buffer exchange.

The analyses of nuclease activity of the purified proteins on 5′-end– or 3′-end–radiolabeled RNAs were performed at 30 °C in 15 mM Tris-HCl (pH 7.6), 75 mM NaCl, 2 mM DTT, 100 μg/ml BSA, 0.8 U/μl RNasin, 4.5% glycerol, 0.05% NP-40, 0.3 μM Escherichia coli tRNA, and 7.5 mM MnCl₂. Prior to addition of 20–30 fmoles of the labeled RNA substrate, the 20 or 30 μl reactions containing 10 pmoles of protein were pre-incubated for 15 min at 30 °C. Subsequently, 4-μl aliquots were taken at the time points indicated in Figure 1, and the reaction was stopped by the addition of formamide-containing RNA loading buffer. Reaction products were resolved on a denaturing 12% polyacrylamide/8 M urea gel and visualized by autoradiography. 5′-[³²P]– and 3′-[³²P]–labeled RNA substrates (Dharmacon) were labeled and gel-purified as described [34].

At least four independent experiments were performed for each strain containing an indicated lacZ pre-mRNA leakage reporter (pJCR1, pmut5′SS, and pmutBP) or intronless lacZ control (pLGSD5). Transformants were pooled and grown overnight in selective raffinose medium. Cells were diluted, grown to mid-log phase and the expression of reporter was induced by 2% galactose for 150 min. β-Galactosidase assays were performed in triplicate for each sample according to the published protocol [11] with the modifications described in [16].

To induce SSA1 expression, heat shock (42 °C for 15 min) was applied to exponentially grown cultures by the addition of an equal volume of prewarmed medium. The expression of lacZ pre-mRNA reporters (pmutBP or pJCR51) was induced by an addition of galactose (2% final concentration) to early-log phase cultures and their subsequent incubation in minimal galactose-raffinose medium for 6 h. Cells were lysed by glass beads and vortexing, and total RNA was isolated with the RNeasy Mini Kit (QIAGEN). RNA was reverse transcribed using the QuatiTect Reverse Transcription Kit (QIAGEN), and levels of specific RNA transcript were measured by quantitative real-time PCR using the ABI Prism 7000 system (Applied Biosystems) and ABsolute QPCR ROX Mix (ABgene). Each cDNA sample, as well as controls without reverse transcriptase, was amplified in triplicate with the primer and probe sets described in Table S3. The levels of SSA1 and lacZ RNA were normalized to RPB4 and GAL1 mRNA levels, respectively, which were found to be unchanged under the tested conditions.

Fluorescence in situ hybridization (FISH) using oligo(dT), SSA1 transcript-, tRNA-, and lacZ-specific Cy3-labeled probes was performed as described [33,35,14]. The localization of the 4xGFP-Swt1, 4xGFP-Swt1ΔPIN, and Nup120-mCherry fusions was analyzed in a Zeiss Axio Imager.Z1 microscope equipped with AxioVision 4.6.3 software. The expression of Swt1 was analyzed by immunoblotting using rabbit polyclonal anti-Swt1 serum generated against the recombinant Swt1 (PickCell Laboratories) and anti-rabbit HRP-conjugate as a secondary antibody (BIO-RAD). The expression of FLAG-tagged Swt1 protein variants was analyzed by anti-FLAG-HRP conjugate (Sigma).