Altered RNA processing and export lead to retention of mRNAs near transcription sites and nuclear pore complexes or within the nucleolus

In a screen of >1000 essential gene mutants in Saccharomyces cerevisiae, 26 mutants are found that directly or indirectly affect mRNA processing and/or mRNA export. Single-molecule FISH data show that the majority of these mutants retain mRNAs at discrete locations within the nucleus, which include the nucleolus.

Abstract

Many protein factors are required for mRNA biogenesis and nuclear export, which are central to the eukaryotic gene expression program. It is unclear, however, whether all factors have been identified. Here we report on a screen of >1000 essential gene mutants in Saccharomyces cerevisiae for defects in mRNA processing and export, identifying 26 mutants with defects in this process. Single-molecule FISH data showed that the majority of these mutants accumulated mRNA within specific regions of the nucleus, which included 1) mRNAs within the nucleolus when nucleocytoplasmic transport, rRNA biogenesis, or RNA processing and surveillance was disrupted, 2) the buildup of mRNAs near transcription sites in 3′-end processing and chromosome segregation mutants, and 3) transcripts being enriched near nuclear pore complexes when components of the mRNA export machinery were mutated. These data show that alterations to various nuclear processes lead to the retention of mRNAs at discrete locations within the nucleus.

INTRODUCTION

Nascent RNA transcripts are processed and mature within the nucleus, including folding, cleavage, modification, nuclear export, or decay. Processing is driven by specific RNA–protein and RNA–RNA interactions that occur in the context of a ribonucleoprotein (RNP) particle, with the protein composition of the RNP being largely responsible for the processing path that is followed (Mitchell and Parker, 2014; Oeffinger and Montpetit, 2015; Singh et al., 2015). In the case of mRNA biogenesis in Saccharomyces cerevisiae, cotranscriptional recruitment of capping enzymes, splicing machinery, and various RNA adaptor proteins mediate maturation of the pre-mRNA, which includes cleavage and polyadenylation during 3′-end formation, loading of export receptors (e.g., Mex67p), and release of the messenger ribonucleoprotein (mRNP) from the transcription site (Oeffinger and Zenklusen, 2012; Niño et al., 2013). After release, mRNPs are often exported through nuclear pore complexes (NPCs), which are assembled from ∼30 nucleoporins (Nups) to form a transport channel spanning the nuclear envelope (Wente and Rout, 2010). mRNP docking and translocation through an NPC are directed by interactions between Mex67p and other mRNP components with Nups of the nuclear basket and central transport channel (Bonnet and Palancade, 2014). On reaching the cytoplasmic side of the NPC, export is terminated by the action of the DEAD-box ATPase Dbp5p, which is locally activated by the cytoplasmic nucleoporins Nup159p and Gle1p with the small molecule InsP₆ (Oeffinger and Zenklusen, 2012; Björk and Wieslander, 2014). This ultimately results in removal of Mex67p, release of mRNP into the cytoplasm, and directionality in mRNA export.

Inevitably, errors occur during nuclear RNA processing, which can result in aberrant transcripts being targeted for nuclear degradation via quality control mechanisms (Houseley and Tollervey, 2009; Müller-McNicoll and Neugebauer, 2013; Eberle and Visa, 2014; Porrua and Libri, 2015). A key player in this process is the exosome, which functions in the nucleus and cytoplasm as a nuclease to facilitate and surveil RNA biogenesis from all three nuclear RNA polymerases (Chlebowski et al., 2013; Porrua and Libri, 2013; Schneider and Tollervey, 2013). The nuclear exosome differs in composition from the cytoplasmic form through the addition of a second catalytic subunit, Rrp6p (Briggs et al., 1998), which, together with Dis3p, harbors the nuclease activities associated with the exosome. Nine other protein subunits are found within the exosome (Liu et al., 2006), which function to recognize and feed RNA substrates Dis3p and Rrp6p (Bonneau et al., 2009; Malet et al., 2010), mediate interactions with multiple protein complexes that include the TRAMP, NNS, and SKI complexes (Jacobs et al., 1998; Steinmetz et al., 2001; Kadaba et al., 2004; Ursic et al., 2004; LaCava et al., 2005; Vanácová et al., 2005; Wyers et al., 2005; Vasiljeva and Buratowski, 2006), and regulate the overall activity of Dis3p and Rrp6p (Mitchell et al., 1997; Allmang et al., 1999; Wasmuth and Lima, 2012).

Disruptions to RNA biogenesis, export, and surveillance result in the accumulation of aberrant RNP complexes and RNA processing by-products in the nucleus of the affected cell (Amberg et al., 1992; Kadowaki et al., 1992, 1994a; Doye et al., 1994; Fabre et al., 1994; Gorsch et al., 1995; Segref et al., 1997). For example, when exosome-dependent RNA processing and surveillance are perturbed, the biogenesis of small nucleolar RNA (snoRNA), rRNA, tRNA, mRNA, and other noncoding transcripts is altered (van Hoof et al., 2000; Kuai et al., 2004; LaCava et al., 2005; Vanácová et al., 2005; Wyers et al., 2005; David et al., 2006; Davis and Ares, 2006; Houalla et al., 2006; Carneiro et al., 2007; Gudipati et al., 2012; Schneider et al., 2012; Castelnuovo et al., 2013). Such disruptions to RNA biogenesis and export can often be observed as the accumulation of poly(A)-RNA species within the nucleus of the affected cell (Cole et al., 2002), which has been used to identify many mutants involved in RNA biogenesis and export, including screening of the ∼5000 nonessential genes in S. cerevisiae for mRNA export defects (Hieronymus et al., 2004). The construction of mutant libraries that span essential genes (Ben-Aroya et al., 2008; Breslow et al., 2008; Li et al., 2011) provides an opportunity to conduct comprehensive screens of essential genes for mRNA processing and export defects, as recently performed for tRNA (Wu et al., 2015). Of importance, the screening of both essential (this work) and nonessential (Hieronymus et al., 2004; Thomsen et al., 2008) genes within S. cerevisiae for mRNA-processing defects provides a component list that is necessary for building complete models of mRNA biogenesis and export.

Here we report on a screen of largely essential gene mutants for nuclear poly(A)-RNA accumulation and the characterization of these mutants using single-molecule fluorescent in situ hybridization (smFISH) directed against specific mRNAs. This resulted in the identification of 15 genes that were not previously linked and/or demonstrated to alter RNA processing and mRNA export. In addition, disruption of multiple nuclear processes was found to cause distinct phenotypes that included the accumulation of mRNAs near transcription sites or the nuclear periphery and NPCs or within the nucleolus. These data suggest that alterations to RNA processing and overall nuclear homeostasis cause RNAs to stall or be retained at similar restriction points. This may reflect common failures in mRNA biogenesis and export, as well as active mechanisms to protect the cell during cellular stress and dysfunction.

RESULTS

Identification of mutants that accumulate nuclear poly(A)-RNA

To potentially identify genes involved in mRNA biogenesis and export, we screened two temperature-sensitive (Ts) S. cerevisiae mutant collections (Ben-Aroya et al., 2008; Li et al., 2011) for the accumulation of poly(A)-RNA in the nucleus using an oligo-dT FISH assay. Together these two Ts collections cover ∼68% (785 of 1156) of essential genes. When a Ts allele was not available, a DAmP allele was used that harbors a disrupted 3′ untranslated region often leading to reduced gene expression (Breslow et al., 2008), which increased coverage of essential genes to ∼91% (1047 of 1156) in our screen (Supplemental Data). For consistency throughout the screening process, regardless of the type of mutant used, all strains were grown into log phase at 25°C, shifted to 37°C for 3 h, and fixed. A 3-h temperature shift was used to balance the time needed to induce the Ts mutant phenotype(s) while minimizing induction of secondary phenotypes caused by the loss of essential cellular activities. In the case of the DAmP alleles, it was reasoned that the temperature shift might act as a stress and exacerbate mutant phenotypes, although DAmP alleles are not necessarily Ts mutants. After fixation, we performed in situ assays using a fluorescently labeled oligo-dT probe to detect poly(A)-RNA.

By comparing the distribution of oligo-dT to the 4′,6-diamidino-2-phenylindole (DAPI) signal, we identified 29 of 1047 mutants that accumulated poly(A)-RNA in the nucleus (Supplemental Figure S1 and Table 1). Of the genes identified, only half (14 of 29) were previously reported to display nuclear accumulation of poly(A)-RNA when disrupted (Table 1). To verify that the poly(A)-RNA accumulation phenotype was linked to the purported mutant being screened, we verified all strains by PCR, and in the case of the 15 newly reported genes, the mutation was rescued by introducing a wild-type allele and/or recapitulated by moving the mutation to a different strain background (Table 1). Note that not all genes previously reported to accumulate poly(A)-RNA in the nucleus when disrupted were identified by our screen, which may be due to the specific allele present in the mutant collection, the length of the temperature shift, or the requirement of a poly(A) tail for detection in the initial in situ screen. Within the set of mutants identified, the distribution of poly(A)-RNA within the nucleus was distinct and included bright foci (brl1-3231), a diffuse nuclear signal (rsp5-3), a diffuse nuclear signal with one or more foci (dbp5-1), or poly(A)-RNA being adjacent to the DAPI-stained DNA mass (dis3-1; Figure 1A). The pattern of poly(A)-RNA accumulation was similar for genes with related biological functions (e.g., RNA processing and surveillance, with an accumulation of poly(A)-RNA next to DAPI; Table 1), which supports the notion that these mutants affect RNA biogenesis in distinct ways and the localization pattern can be informative for gene function.

TABLE 1:

Description and phenotypes associated with mutants that display poly(A)-RNA accumulation. Continued

Gene	Biological process	Previous report/verificationa	Distribution of nuclear poly(A)-RNA	Cells with nuclear poly(A)-RNA accumulation ± SD (%)	GFA1 mRNAs per cell ± SD	Nuclear GFA1 mRNAsb	ITS1 and Nop56-GFP
URA10	Pyrimidine biosynthesis	–	Diffuse	<1	5.1 ± 0.7	0.9 ± 0.9 (1)	–
ALR1	Plasma membrane Mg2+ transporter	Rescued	Foci	36 ± 2	11.1 ± 1.7	1.8 ± 1.6* (1)	–
BRL1	NPC and nuclear envelope biogenesis	Saitoh et al. (2005)	Diffuse signal/foci	34 ± 11	6.0 ± 0.6	1.3 ± 1.1* (1)	–
BRR6	NPC and nuclear envelope biogenesis	de Bruyn Kops and Guthrie (2001)	Diffuse signal/foci	49 ± 4	4.5 ± 1.1	1.1 ± 1.0* (1)	–
CBF2	Chromosome segregation	Rescued	Diffuse signal/foci	27 ± 11	5.2 ± 1.2	0.9 ± 0.9 (1)	Both absent in cells with nuclear poly(A)-RNA
CEP3	Chromosome segregation	Rescued	Diffuse signal/foci	16 ± 4	6.7 ± 0.9	1.0 ± 1.3 (1)	Both absent in cells with nuclear poly(A)-RNA
CLP1	Cleavage and polyadenylation of RNA	Rescued	Foci	32 ± 10	9.6 ± 1.3	1.6 ± 1.3* (1)	Nop56-GFP foci
CSL4	RNA processing and degradation	Remade	Diffuse signal next to DAPI	∼100	6.0 ± 1.9	2.3 ± 1.9* (2)	Enlarged ITS1 and Nop56-GFP area
DBP5	RNA export	Snay-Hodge et al. (1998), Tseng et al. (1998)	Diffuse signal/foci	∼100	4.2 ± 0.3	2.2 ± 1.5* (2)	ITS1 and Nop56-GFP foci
DIS3	RNA processing/degradation	Kadowaki et al. (1994a)	Diffuse signal next to DAPI	∼100	5.5 ± 1.0	2.1 ± 1.5* (2)	Enlarged ITS1 and Nop56-GFP area
ENP1	RNA processing and ribosomal subunit synthesis	Rescued	Diffuse signal next to DAPI	∼100	3.4 ± 0.5	2.0 ± 1.5* (2)	ITS1 decreased or absent
GLE1	RNA export	Del Priore et al. (1996), Murphy and Wente (1996)	Diffuse signal/foci	∼100	2.2 ± 0.3	2.1 ± 1.3* (2)	ITS1 and Nop56-GFP foci
IPL1	Chromosome segregation/cell cycle	Cole et al. (2002)	Diffuse signal/foci	30 ± 12	3.4 ± 0.5	0.8 ± 0.9 (1)	Both absent in cells with nuclear poly(A)-RNA
LDB19	Ubiquitin-dependent endocytosis	Remade	Diffuse signal	17 ± 3	6.2 ± 0.5	0.9 ± 1.0 (1)	–
MEX67	RNA export	Segref et al. (1997)	Diffuse signal/foci	∼100	2.4 ± 0.6	2.1 ± 1.2* (2)	ITS1 and Nop56-GFP foci
MPS1	Spindle pole body/cell cycle	Rescued	Diffuse signal/foci	28 ± 4	6.5 ± 1.2	1.1 ± 1.1 (1)	Both absent in cells with nuclear poly(A)-RNA
MTR4	RNA processing and surveillance	Kadowaki et al. (1994a)	Diffuse signal next to DAPI	∼100	4.9 ± 0.7	2.3 ± 1.5* (2)	ITS1 decreased or absent
NUP133	Nucleocytoplasmic transport	Doye et al. (1994)	Diffuse signal/foci	59 ± 5	5.7 ± 0.8	1.3 ± 1.0* (1)	–
NUP145	Nucleocytoplasmic transport	Fabre et al. (1994), Wente and Blobel (1994)	Diffuse signal/foci	43 ± 2	4.6 ± 0.6	1.2 ± 1.0* (1)	–
NUP159	Nucleocytoplasmic transport	Gorsch et al. (1995)	Diffuse signal/foci	∼100	1.7 ± 0.2	1.6 ± 1.0* (1)	ITS1 and Nop56-GFP foci
PRP2	Pre-mRNA splicing	Rescued	Diffuse signal	34 ± 12	5.9 ± 0.5	1.1 ± 1.1 (1)	–
PTA1	Cleavage and polyadenylation of RNA	Hammell et al. (2002)	Diffuse signal/foci	35 ± 8	6.1 ± 0.9	1.2 ± 1.0* (1)	–
RRP43	RNA processing and degradation	Rescued	Diffuse signal next to DAPI	∼100	5.3 ± 0.3	1.8 ± 1.7* (1)	Enlarged ITS1 and Nop56-GFP area
RSP5	E3 ubiquitin ligase; multiple processes	Neumann et al. (2003), Rodriguez et al. (2003)	Diffuse signal	∼100	5.7 ± 2.0	1.2 ± 0.9* (1)	–
SLI15	Chromosome segregation/cell cycle	Remade	Diffuse signal/foci	30 ± 4	4.9 ± 1.1	1.0 ± 1.2 (1)	Both absent in cells with nuclear poly(A)-RNA
SMC1	Chromosome segregation	Rescued	Diffuse signal/foci	14 ± 2	4.3 ± 0.5	1.0 ± 1.2 (1)	–
SMC3	Chromosome segregation	Rescued	Diffuse signal/foci	51 ± 13	5.7 ± 0.3	1.3 ± 1.2* (1)	–
SMC4	Chromosome organization	Rescued	Diffuse signal/foci	29 ± 7	9.0 ± 2.5	1.1 ± 1.3 (1)	Both absent in cells with nuclear poly(A)-RNA
SPC24	Chromosome segregation	Remade	Diffuse signal/foci	33 ± 6	4.3 ± 1.1	0.9 ± 1.1 (1)	Both absent in cells with nuclear poly(A)-RNA
SRM1/PRP20	Nucleocytoplasmic transport	Amberg et al. (1993), Kadowaki et al. (1994a)	Diffuse signal next to DAPI	∼100	2.1 ± 0.2	0.6 ± 0.7* (0)	ITS1 decreased or absent

^aTo verify mutants that had not previously been reported to accumulate poly(A)-RNA, the phenotype was rescued using a wild-type allele or recapitulated by making the mutant in a different strain background as indicated.

^bAverage number of nuclear mRNAs ± SD with the median for the data shown in parentheses.

*Distribution of nuclear mRNAs tested using the Wilcoxon-rank sum test and found to be significant at p < 0.001.

Figure 1:

Poly(A)-RNA localization patterns and nucleolar status. (A) Representative images showing poly(A)-RNA (gray) localization in control (ura10Δ) and select mutant strains compared with DAPI (blue) after 3 h at 37°C. (B) Representative images showing poly(A)-RNA (gray) localization in control (ura10Δ) and select mutant strains compared with DAPI (blue), Nop56p-GFP (green), and ITS1 (red) after 3 h at 37°C. Scale bars, 1 μm.

Nucleolar disruption is linked to poly(A)-RNA accumulation

Studies in Schizosaccharomyces pombe have shown that mutants affecting chromosome biology are associated with the accumulation of nuclear poly(A)-RNA and nucleolar disruption (Kalam Azad et al., 2003; Ideue et al., 2004). Of the genes identified in our screen, approximately one-third have functions that include kinetochore–microtubule attachment, chromosome organization, and cell cycle checkpoint control (Table 1). Given these facts and the reported defects in rDNA segregation within mutants identified here (e.g., IPL1 and SMC genes; Freeman, 2000; D’Amours et al., 2004; Sullivan et al., 2004; Machín et al., 2005), we assayed nucleolar status in all 29 mutants based on the rRNA-processing intermediate (ITS1) and nucleolar protein (Nop56p–green fluorescent protein [GFP]; Gautier et al., 1997; Woolford and Baserga, 2013). By monitoring each nucleolar marker and poly(A)-RNA in the same cell, we found that 21 of the 29 mutants displayed alterations in ITS1 and/or Nop56p-GFP localization and abundance when poly(A)-RNA accumulated (Table 1), which included fragmented nucleoli in mRNA export mutants, as previously reported (Kadowaki et al., 1994b; Dockendorff et al., 1997; Segref et al., 1997; Thomsen et al., 2008). Nucleolar disruption was prominent in seven of the nine mutants that affect chromosome biology, with these mutants often lacking nucleolar ITS1 and Nop56p-GFP in cells with poly(A)-RNA accumulation (e.g., sli15-1 and smc4-1 vs. dis3-1; Figure 1B). Although SMC1 and SMC3 mutants did not show obvious nucleolar defects, recent reports provide a direct role for cohesins (e.g., SMC3) in nucleolar function (Bose et al., 2012; Harris et al., 2014). Together these findings further support a link between poly(A)-RNA accumulation and alterations to the nucleolus, which may often be induced by errors in chromosome segregation.

Identification of mRNA biogenesis and export mutants

Mutations within RNA processing and surveillance pathways have been shown to accumulate poly(A)-RNA species, including rRNA, mRNA, snRNA, snoRNA, and tRNA (van Hoof et al., 2000; Kuai et al., 2004; LaCava et al., 2005; Vanácová et al., 2005; Wyers et al., 2005; Carneiro et al., 2007; Rougemaille et al., 2007; Gudipati et al., 2012; Schneider et al., 2012; Castelnuovo et al., 2013). Consequently the mutants identified here using an oligo-d(T)–based in situ approach may accumulate poly(A)-RNA due to disruptions in RNA biogenesis that are independent of mRNA. To identify those mutants that alter mRNA processing and export, we performed smFISH assays using probes against GFA1, ACT1, or CCW12 transcripts. The mRNAs were selected based on relative expression levels (GFA1 = low and ACT1/CCW12 = high) and the presence of an intron in ACT1 (Ng and Abelson, 1980) that may lead to this mRNA being affected differently than nonspliced mRNAs (i.e., CCW12 and GFA1). To ensure that a block in mRNA export could be observed, we used a mex67-5 strain, which, when shifted to the nonpermissive temperature, rapidly accumulated poly(A)-RNA in the nucleus (Segref et al., 1997). smFISH assays using the gene-specific probes displayed an obvious block in mRNA export at the nonpermissive temperature in mex67-5, and, of importance, a mex67-5/ccw12Δ strain showed no detectable signal with the CCW12 probe set (Figure 2A). This established that these mRNA probes can be used to detect export defects, and in the case of CCW12, the probes were specific for the transcript being targeted.

Figure 2:

Identification of mRNA export mutants using smFISH. (A) Representative images showing ACT1, GFA1, or CCW12 mRNA (red) localization in control (ura10Δ), mex67-5, or mex67-5/ccw12Δ strains compared with poly(A)-RNA (gray) and DAPI (blue) after 3 h at 37°C. (B) Representative images showing ACT1 or CCW12 mRNA (red) localization in the prp2-1 strain compared with poly(A)-RNA (gray) and DAPI (blue) after 3 h at 37°C. (C) Representative images showing ACT1 mRNA (red) localization in control (ura10Δ), brl1-3231, and brr6-ph strains compared with poly(A)-RNA (gray) and DAPI (blue) after 6 h at 37°C. Scale bars, 1 μm.

Using smFISH data for GFA1, we determined the number of transcripts and distribution of these transcripts between the nucleus and cytoplasm. In a haploid control strain (ura10Δ), GFA1 was found to be present at approximately five copies per cell, with ∼18% of these transcripts being in the nucleus (based on DAPI and ITS1 signals), whereas the mRNA export mutant mex67-5 contained ∼88% of transcripts in the nucleus (Table 1 and Figure 2A). Most mutants showed less than a twofold change in GFA1 levels, with nuclear pools of the mRNA that varied between 16 and 95% (Table 1). In the case of prp2-1, there was no effect on GFA1 localization or transcript number, but ACT1 export was altered (Figure 2B), which is consistent with the role of Prp2p in splicing (Lustig et al., 1986). For brl1-3231 and brr6-ph, these mutants showed no mRNA export defect after a 3-h temperature shift, but given the role of the gene products in nuclear envelope maintenance and NPC biogenesis (de Bruyn Kops and Guthrie, 2001; Saitoh et al., 2005; Hodge et al., 2010; Lone et al., 2015), we expected to observe a defect. We performed a 6-h temperature shift, providing additional time for Ts phenotypes to develop, and under these conditions, we observed mRNA export defects for both mutants (Figure 2C).

mRNAs localize to distinct subdomains of the nucleus in mutants

smFISH data showed that some mutants had a large increase in the fraction of nuclear mRNAs (e.g., mex67-5 and gle1-4), whereas others did not (e.g., rsp5-3 and sli15-1). However, in almost all instances, the localization of mRNAs appeared distinct in the mutants tested (Supplemental Data). For example, strains carrying mutations in genes directly linked to the mRNA export process (e.g., DBP5, GLE1, MEX67, and NUP159) often had mRNAs near the periphery of the DAPI-stained DNA mass (see mex67-5 in Figure 2). This suggests that within these mutants, mRNAs accumulated at or near nuclear pore complexes, as reported for mutants of MEX67 (Hurt et al., 2000; Smith et al., 2015) and DBP5 (Hodge et al., 2011). The prp2-1 mutant also showed ACT1 transcripts near the nuclear periphery (Figure 2B), which may be related to quality control mechanisms that block export of pre-mRNAs (Galy et al., 2004; Palancade et al., 2005; Iglesias et al., 2010; Hackmann et al., 2014). Using Ndc1p-GFP as a marker of NPCs and the nuclear periphery, we quantified the percentage of transcripts within ∼250 nm of the NPC signal in individual cells (n = 50) of select mutants. In control cells, 38 ± 15% of GFA1 mRNAs were found within this distance, which increased in mRNA export mutants to 68 ± 35% (mex67-5) and 61 ± 27% (dbp5-1) (Figure 3A). The prp2-1 mutant did not show an increase in peripheral localization of GFA1 (39 ± 18%), whereas ACT1 increased from 25 ± 12% in control to 49 ± 17% in prp2-1 (Figure 3B). These data are consistent with the accumulation or retention of mRNAs near NPCs when splicing or late steps in the mRNA export pathway are disrupted.

Figure 3:

Accumulation of mRNA near the nuclear periphery and NPCs. (A) Representative images showing GFA1 mRNA (red) localization in control (ura10Δ), dbp5-1, and mex67-5 strains compared with NPCs (green, Ndc1p-GFP) and DAPI (blue) after 3 h at 37°C. (B) Representative images showing GFA1 or ACT1 mRNA (red) localization in the prp2-1 strain compared with NPCs (green, Ndc1p-GFP) and DAPI (blue) after 3 h at 37°C. Scale bars, 1 μm.

A second distinct localization pattern was the accumulation of mRNAs within a nuclear focus. Rarely present in the control strain, 15 mutants showed >5-fold increase in the frequency of nuclear foci with an intensity ≥10-fold that of single transcripts (Supplemental Data), including rsp5-3 (Figure 4A). In quantifying the GFA1 smFISH data, we counted such foci as single mRNAs (see Material and Methods, Gene-specific FISH), which likely underestimates the number of nuclear transcripts in these mutants and leads to a lower level of nuclear accumulation reported in Table 1. Rsp5p is a ubiquitin-ligase that functions in both the cytoplasm and nucleus (Belgareh-Touzé et al., 2008; Kaliszewski and Zoładek, 2008) and has a known role in mRNA biogenesis via modification of the THO/TREX complex (Neumann et al., 2003; Rodriguez et al., 2003; Gwizdek et al., 2005). Given the functions of Rsp5p and other genes with this phenotype, the finding that various mRNA probes show the same defect (i.e., not related to splicing), and the intensity of the smFISH signal, we speculated that these foci represent gene transcription sites.

Figure 4:

Accumulation of ACT1 mRNA near transcription sites. (A) Representative images showing GFA1 mRNA (red) localization in control (ura10Δ) and rsp5-3 strains compared with poly(A)-RNA (gray) and DAPI (blue) after 3 h at 37°C. (B, C) Representative images showing ACT1 or CCW12 mRNA (red) localization in control (ura10Δ), rsp5-3, ldb19Δ, spc24-10, and sli15-1 strains compared with the ACT1 gene locus (green, marked by lacO-array/GFP-LacI) and DAPI (blue) after 3 h at 37°C. Inset, zoomed-in view of boxed region in the merged image. Scale bars, 1 μm.

To test this possibility, we integrated a lacO array ∼400 base pairs upstream of the ACT1 gene, which could be used in com­bination with LacI-GFP and FISH probes to localize both the ACT1 gene and mRNA focus. In rsp5-3, bright ACT1 mRNA foci were in close proximity to the ACT1 gene locus, separated by an average distance of 0.15 ± 0.05 μm, whereas the CCW12 mRNA focus was distinct from the ACT1 gene locus at an average distance of 0.81 ± 0.27 μm (n = 50; Figure 4B). These data are consistent with these bright mRNA foci being at or near transcription sites. We also observed close association between the ACT1 mRNA and gene loci in mutants linked to chromosome segregation (spc24-10 and sli15-3), as well as ldb19Δ (Figure 4C). Ldb19p is a regulator of Rsp5p that functions in ubiquitin-dependent receptor endocytosis (Lin et al., 2008) but is not known to have a nuclear role. These data match previous reports of mRNAs being retained near transcription sites in THO/TREX complex and mRNA export pathway mutants after heat shock (Hilleren et al., 2001; Jensen et al., 2001; Libri et al., 2002). Moreover, these results demonstrate that the retention of mRNAs at or near transcription sites is a phenotype shared by a set of mutants with diverse cellular functions.

The third localization pattern was an accumulation of mRNAs next to the DAPI-stained DNA mass with or near ITS1 (Table 1). This included mutants in components of the exosome (e.g., DIS3, RRP43, and CSL4) and TRAMP complex (MTR4), which have been reported to accumulate poly(A)-RNA in the nucleolus, as well as the heat shock–induced transcript SSA4 and the localized mRNA ASH1 (Brodsky and Silver, 2000; Thomsen, 2003; Carneiro et al., 2007; Rougemaille et al., 2007; Du et al., 2008). To quantify nucleolar localization (defined by ITS1 staining), we compared FISH data for GFA1 between the mRNA export factor (dbp5-1) and exosome component (dis3-1), which both accumulated GFA1 transcripts in the nucleus after temperature shift (Figure 5A). Using these data, we quantified where nuclear mRNA localized with respect to the ITS1 signal and found 0.4 ± 0.6 and 0.8 ± 0.9 GFA1 transcripts in the nucleolus of the control (ura10Δ) and dbp5-1 strains, respectively. This corresponds to 46% (ura10Δ) and 36% (dbp5-1) of total nuclear transcripts in these strains (Table 1). In contrast, 1.7 ± 1.2 GFA1 mRNAs were found in the nucleolus of a dis3-1 strain, which was 80% of the total transcripts in the nucleus of this mutant. This suggests that disruption of exosome function leads to the accumulation of mRNAs within the nucleolus, which is not a general result of mRNAs being retained in the nucleus.

Figure 5:

Nucleolar localization of mRNA. (A–C) Representative images showing GFA1, IMD2, or LEU1 mRNA (green) localization in control (ura10Δ), dbp5-1, dis3-1, and rsp5-3 strains compared with ITS1 (red) and DAPI (blue) after 3 h at 37°C. (D) Representative images showing localization of NAB2-GFP transcripts using GFP in situ probes (green) in control (ura10Δ) and dis3-1 strains compared with ITS1 (red) and DAPI (blue) after 3 h at 37°C. Scale bars, 1 μm.

mRNAs can be classified based on protein-binding profiles, which have been used to define 10 general mRNP classes (Tuck and Tollervey, 2013). These range from mRNAs most likely to be processed and exported to the cytoplasm for translation (class X) to those that have protein-binding patterns similar to CUTs (class I), with CUTs being targets of nuclear RNA surveillance (Wyers et al., 2005). The three transcripts we observed in the nucleolus (GFA1, ACT1, and CCW12) belong to class X, and so to extend our observations to other mRNA classes, we used gene-specific FISH probes to assay the localization of IMD2 (class I) and LEU1 (class II) mRNAs. In both cases, we observed that these mRNAs were retained in the nucleus of a dbp5-1 mutant, localized to the nucleolus in dis3-1, and appeared as a bright nuclear focus in rsp5-3 (Figure 5, B and C). By using strains carrying NAB2-GFP and GFP FISH probes, we also observed mRNAs within the nucleolus of a dis3-1 strain (Figure 5D). This implies that mRNAs in the nucleolus are near full length and not short transcripts resulting from early transcription termination, since the GFP probes are directed against the 3′ end of the transcript. These data demonstrate that various mRNAs localize to distinct subdomains of the nucleus when RNA processing and surveillance pathways are disrupted.

mRNP-associated factors are sequestered in the nucleolus with mRNA

After the observation that mRNAs accumulated in the nucleolus, we characterized the subcellular localization of three proteins involved in mRNA processing. Specifically, in control (ura10Δ), dbp5-1, and dis3-1 strains, we assayed localization of Nab2p-GFP (polyadenosine RNA-binding adaptor protein for Mex67p), Prp19p-GFP (splicing factor), and Hrp1p-GFP (subunit of cleavage factor I; required for the cleavage and polyadenylation of mRNA 3′ ends). In control and dbp5-1 strains, these proteins colocalized with the DAPI-stained DNA mass and were largely absent from the nucleolus as marked by ITS1 (Figure 6, A–C). In contrast, within the dis3-1 strain, these factors were found throughout the nuclear volume within both the DAPI- and ITS1-stained regions, suggesting that factors involved in mRNA biogenesis redistribute to the nucleolus in this mutant, similar to mRNAs.

Figure 6:

Nucleolar enrichment of mRNP-associated proteins in exosome mutants. (A–C, E) Representative images showing localization of Nab2p-GFP, Prp19p-GFP, or Hrp1p-GFP (green) in control (ura10Δ), dbp5-1, dis3-1, rrp6Δ, or enp1-1 strains compared with poly(A)-RNA (gray), ITS1 (red), and DAPI (blue) after 3 h at 37°C. A white line has been added to one cell in each image to denote the border between DAPI and ITS1 signals and aid in comparisons. (D) Representative images showing GFA1 mRNA (red) localization in control (ura10Δ), enp1-1, and srm1-ts strains compared with poly(A)-RNA (gray) and DAPI (blue) after 3 h at 37°C. Scale bars, 1 μm.

We further tested whether loss of Rrp6p activity would lead to the relocalization of these same factors. Rrp6p is a nonessential catalytic subunit of the exosome that, when mutated, results in poly(A)-RNA accumulation in a discrete domain within the nucleolus (Hieronymus et al., 2004; Carneiro et al., 2007; Rougemaille et al., 2007). In an rrp6Δ strain. we observed approximately five GFA1 transcripts per cell, with 18% being nuclear, which is comparable to the control strain (Table 1), but 71% of these nuclear transcripts colocalized with ITS1 as compared with 46% in control. Nab2p-GFP, Prp19p-GFP, and Hrp1p-GFP were also enriched within the nucleolus (Figure 6, A–C), demonstrating that both mRNA- and mRNP-associated factors are redistributed within the nucleus of an rrp6Δ strain.

In addition to mutations in exosome or TRAMP complex components, mutations in ENP1 and SRM1 caused accumulation of poly(A)-RNA next to the DAPI-stained DNA mass (Figure 6D and Table 1). Enp1p functions in pre-rRNA processing and 40S subunit synthesis, and Srm1p facilitates nucleocytoplasmic trafficking (Tachibana et al., 1994; Koepp et al., 1996; Chen et al., 2003). GFA1 mRNAs could be readily observed with the poly(A)-RNA signal that is adjacent to DAPI in both enp1-1 and srm1-ts strains (Figure 6D), but ITS1 expression was severely reduced in both mutants, preventing us from quantifying GFA1 localization. Nab2p-GFP, Prp19p-GFP, and Hrp1p-GFP were also localized to the nucleolus of the enp1-1 strain (Figure 6E); however, srm1-ts could not be characterized due to defects in nuclear protein import. These findings demonstrate that localization of mRNAs and mRNP-associated proteins to the nucleolus does not occur only when RNA surveillance and quality control machinery is mutated and can occur as a result of disruption to other nuclear processes.

DISCUSSION

mRNA biogenesis involves the interaction of numerous protein factors with each mRNA in a spatially and temporally regulated manner within the nucleus to link the processes of transcription, capping, splicing, 3′-end formation, RNA surveillance, and NPC-mediated export (Oeffinger and Zenklusen, 2012; Niño et al., 2013). Nuclear events also act to define the fate of each mRNP, which can include marking mRNPs in the nucleus for translation, storage, transport, or decay within the cytoplasm (Müller-McNicoll and Neugebauer, 2013; Singh et al., 2015). Consequently the proper and efficient coordination of these nuclear events is central to the fidelity of the gene expression program, with previous work having identified factors involved in these processes through analysis of individual mutants, genetic screens and comprehensive screening of nonessential genes (Shiokawa and Pogo, 1974; Amberg et al., 1992; Kadowaki et al., 1992, 1994a; Doye et al., 1994; Fabre et al., 1994; Gorsch et al., 1995; Hieronymus et al., 2004). However, the critical nature of these events dictates that essential genes would likely be involved in this process, which prompted us to perform a screen of essential genes for function(s) related to mRNA biogenesis in S. cerevisiae. This led to the identification of 29 genes with a nuclear poly(A)-RNA accumulation phenotype (Table 1), of which half (15 of 29) had not previously been reported to have such a defect. Of these, 26 genes were subsequently found to alter mRNA biogenesis and export when mutated, based on the observed accumulation of mRNA in the nucleus and/or altered localization of nuclear mRNAs.

Classes of mRNA biogenesis and export mutants

Gene ontology places the majority of the 29 genes into three biological processes: RNA export from nucleus, nuclear RNA catabolic processes, and chromosome segregation. The identification of mRNA export factors is expected and validates our screen. Links between mRNA export and RNA decay are also well established (Porrua and Libri, 2013; Eberle and Visa, 2014), with various components of the decay machinery having originally been identified in screens for mRNA transport (MTR) mutants (Kadowaki et al., 1992, 1994a). Beyond the related biological functions of the gene products, 26 of the 29 mutants could be further classified into three distinct groups (referred to as class A, B, or C) based on shared phenotypes with respect to the localization of poly(A)-RNA and mRNA within the nucleus (Supplemental Data).

Class A mutants are represented by genes directly involved in mRNA export and NPC function (e.g., DBP5 and NUP159) and were characterized by a diffuse nuclear poly(A)-RNA signal (with or without discrete foci) overlapping the DAPI stain, often accompanied by disrupted nucleoli (Figure 1 and Table 1). These mutants further displayed an increased proportion of mRNAs near the nuclear periphery and NPCs (Figure 3). This suggests that within class A mutants, mRNPs accumulated at or near nuclear pore complexes due to failures in NPC-mediated export, which is consistent with the known function of these factors (Table 1). In support of the smFISH observations and data interpretation, delays in export at the nuclear periphery have been observed using live-cell imaging techniques for mutants of DBP5, MEX67, NAB2, and the Nups MLP1/2 (Hodge et al., 2011; Saroufim et al., 2015; Smith et al., 2015). Of the class A mutants, prp2-1 showed a transcript-specific block of mRNA export with the ACT1 mRNA localized near the nuclear periphery but not another mRNA that lacked an intron (Figures 2 and 3). This observation is consistent with the description of NPC-dependent quality control mechanisms that prevent the export of immature mRNAs, which include pre-mRNAs (Galy et al., 2004; Palancade et al., 2005; Iglesias et al., 2010; Hackmann et al., 2014).

Class B mutants, similar to class A, showed a diffuse poly(A)-RNA distribution (with or without discrete foci) overlapping the DAPI stain (see rsp5-3, Figure 1) but differed in that mRNAs were retained at or near gene transcription sites (Figure 4 and Supplemental Data). Class B mutants included RSP5, PTA1, and CLP1, which are known to be involved in 3′-end processing (Minvielle-Sebastia et al., 1997; Preker et al., 1997; Zhao et al., 1999; Gwizdek et al., 2005) and could be expected to retain mRNAs at transcription sites due to defects in mRNA biogenesis similar to other mutants affecting cotranscriptional processing (Hilleren et al., 2001; Jensen et al., 2001; Libri et al., 2002; Zenklusen et al., 2002; Thomsen et al., 2008). Ldb19p is a known regulator of Rsp5p in endocytosis (Lin et al., 2008) and was also found to accumulate mRNAs near transcription sites when mutated, like rsp5-3 (Figure 4). The functional relationship between Ldb19p and Rsp5p with the shared mRNA-processing defects suggests that Ldb19p also regulates the nuclear function of Rsp5p. In support of a nuclear role for Ldb19p, we were able to detect GFP-Ldbp19p in the nucleus when overexpressed from a GAL promoter in an xpo1-1 mutant that disrupts nucleocytoplasmic transport (Supplemental Figure S2).

Unexpectedly, the majority of class B mutants isolated in our screen (9 of 15) were linked to chromosome segregation and cell division. A previous report noted that mutants affecting chromosome segregation in S. cerevisiae displayed poly(A)-RNA accumulation, which was dependent on ongoing cell division (Cole et al., 2002). Given the requirement for ongoing cell division, the large number of chromosome segregation mutants with a class B phenotype, and the fact that these mutants show defects in <25% of cells after 3 h at nonpermissive temperature (Table 1), it is likely that these defects are related to improper chromosome segregation. This could occur through the random loss of genetic material that encodes any one of the many factors involved in mRNA biogenesis, which would be expected to give rise to variable phenotypes, depending on the processing event affected. However, when poly(A)-RNA accumulated in these mutants it was associated with mRNAs appearing at transcription sites and disruption of the nucleolus (Figures 1 and 4 and Supplemental Data). A possible explanation for this more constant phenotype is that specific chromosomes or regions of DNA are repeatedly lost in these mutants and this gives rise to the class B defect due to the specific gene(s) being affected. Alternatively, it has been reported that aneuploid states involving different chromosome imbalances in S. cerevisiae result in a set of common cellular characteristics (Torres et al., 2007; Sheltzer et al., 2012), which class B mRNA export phenotypes could exemplify. Although the molecular details are unknown, a recent screen for mutants that alter tRNA processing (Wu et al., 2015) did not report a defect for any chromosome segregation mutants and had only two genes (NUP133 and RSP5) in common with the class B mutants identified here. This lack of overlap hints at a more direct relationship between mRNA export and chromosome segregation; as it would be expected that both mRNA and tRNA processing would be equally susceptible to disruption by random chromosome segregation errors.

Finally, class C mutants displayed poly(A)-RNA accumulation next to the DAPI stain with or adjacent to the nucleolus (based on ITS1) and an increased frequency of mRNAs within this compartment (Figures 1, 5, and 6 and Supplemental Data). Many of these mutants are involved in RNA processing and surveillance as components of the exosome and TRAMP complexes, which raises the possibility that the detected mRNAs are aberrant and only become detectable due to a loss of surveillance and decay activities. However, we did not detect an increase in the number of GFA1 transcripts in these mutants (Table 1), which may be expected if these were common products of transcription that are normally degraded. By using probes against the GFP coding sequence at the 3′ end of an mRNA (Figure 5), we also observed nucleolar mRNA accumulation, suggesting that these transcripts are near full length and not the product of early transcript termination events. Other mutants (e.g., enp1-1 and srm1-ts) that are not part of the decay machinery caused similar mRNA localization phenotypes, suggesting that nucleolar mRNAs are not solely detectable in mutants of the RNA surveillance machinery.

In addition to mRNAs, we also observed enrichment of mRNP-associated factors within the nucleolus in class C mutants (Figures 6). The redistribution of mRNP-associated factors may occur as constituents of mRNPs present within the nucleolus, but it is also possible that poly(A)-RNA binding proteins, such as Nab2p (Anderson et al., 1993), localize to the nucleolus due to promiscuous binding of the protein to accumulated poly(A)-RNA. The latter would effectively deplete the activities of poly(A)-RNA binding proteins, which include mRNA export adaptors, and provide a mechanism by which class C mutants with various cellular functions cause the same terminal phenotype. The heat shock–induced transcript SSA4, the localized mRNA ASH1, and PHO84 antisense RNA have also been reported to enter the nucleolus (Brodsky and Silver, 2000; Thomsen, 2003; Carneiro et al., 2007; Du et al., 2008; Castelnuovo et al., 2013), and mRNAs have recently been observed to transit through the nucleolus for export in live cells (Saroufim et al., 2015). Further study is needed to address how often mRNAs enter the nucleolus during normal mRNA biogenesis, the functional significance of this event, and the molecular mechanism(s) facilitating nucleolar localization.

Nuclear homeostasis and mRNA biogenesis

Disruption of different nuclear processes resulted in classes A–C of mRNA biogenesis defects. We envision this occurring by multiple mechanisms: 1) directly as a result of mutation in a factor involved in mRNA biogenesis and export, 2) by activating quality control mechanisms that retain immature/aberrant mRNPs, and 3) due to changes in nuclear structure/function that result in inefficiencies and failures in mRNA biogenesis and export. The first two mechanisms are exemplified by the mex67-5 mutation, as it is well established that Mex67p functions directly in mRNA export at NPCs (Segref et al., 1997; Santos-Rosa et al., 1998; Hurt et al., 2000; Sträßer et al., 2000; Lund and Guthrie, 2005; Smith et al., 2015), and a quality control mechanism was described to protect cells during loss of Mex67p function by retaining mRNAs at transcription sites (Kallehauge et al., 2012). Moreover, both mechanisms are highlighted by our smFISH data through the observation of mRNAs that accumulated near the nuclear periphery (e.g., export failure) and the approximately eightfold increase in transcription site foci (e.g., retention) in the mex67-5 strain (Figure 2 and Supplemental Data). Similar quality control mechanisms have been described in other mutants to cause mRNAs to be retained near transcription sites and at NPCs to prevent export of pre-mRNAs and aberrant mRNPs (Hilleren et al., 2001; Libri et al., 2002; Galy et al., 2004; Palancade et al., 2005; Rougemaille et al., 2008; Saguez et al., 2008; Iglesias et al., 2010; Hackmann et al., 2014), which together may explain many of the observed localization patterns in class A and B mutants.

The third mechanism leading to mRNA biogenesis and export defects likely operates at the systems level and is represented by srm1-ts, enp1-1, and the many mutants that affect chromosome segregation and exosome function. In these cases, altering nucleocytoplasmic transport and various aspects of RNA biogenesis and inducing aneuploidy would alter nuclear homeostasis, and, in turn, disturb mRNA biogenesis and export. For example, the enp1-1 mutation induced strong poly(A)-RNA accumulation in the nucleolus and redistribution of mRNA processing factors to this compartment (Figure 6), which likely affects mRNA processing and export by making these essential factors limiting. This relationship has also been observed to work in reverse, with mRNA export defects causing nucleolar disruption (Kadowaki et al., 1994b; Dockendorff et al., 1997; Segref et al., 1997; Thomsen et al., 2008). In this way, enp1-1 (and other such mutants) function by perturbing overall nuclear homeostasis and would not be considered an mRNA processing or export factor per se.

Within this paradigm of system-level perturbations, how chromosome segregation mutants affect the cell and cause specific mRNA processing and export defects remains to be determined. We find it noteworthy that aneuploidy is known to alter tumorigenesis through mechanisms that include changes in gene expression (Gordon et al., 2012; Holland and Cleveland, 2012; Santaguida and Amon, 2015; Dürrbaum and Storchová, 2016), and our work highlights the fact that many chromosome segregation mutants cause the accumulation of mRNAs at transcription sites, which provides one mechanism by which this may occur. Although these types of relationships could be casually termed as indirect, these system-level perturbations are important to understand because it is likely that disease states arise from mutation(s) due to both the specific process affected by the mutation and the system-level changes this in turn induces.

MATERIALS AND METHODS

Yeast strains

Strains constructed for this study are listed in the Supplemental Data or were taken directly from mutant collections (Ben-Aroya et al., 2008; Breslow et al., 2008; Li et al., 2011). Strains with genome-encoded, GFP-tagged NOP56, NAB2, PRP19, and HRP1 were generated by amplifying each gene with the GFP::HIS3MX6 cassette plus ∼300 base pairs of flanking sequence directly from the yeast GFP collection (Huh et al., 2003). The resulting PCR product was transformed into yeast to integrate the GFP tag into the yeast genome (Gietz et al., 1992). Deletion strains were made by homologous recombination with a selectable marker made by PCR (Longtine et al., 1998). NDC1-yeGFP and GAL1-GFP-LDB19 strains were generated by transforming a PCR product with homology to the gene locus made from pKT148 (Sheff and Thorn, 2004) and pFA6a-His3MX6-PGAL1-GFP (Longtine et al., 1998), respectively. The NOP56-GFP plasmid (pBM461) was made by cloning NOP56-GFP with ∼400 base pairs of promoter sequence from the yeast GFP collection (Huh et al., 2003) into pRS313 (Sikorski and Hieter, 1989) using XhoI and NotI sites introduced by PCR. Tagging of the ACT1 locus with a lacO array was performed in a wild-type strain as previously described using pSR13 and pAFS78 (Straight et al., 1996; Rohner et al., 2008) into which mutant alleles (i.e., LDB19, SLI15, and SPC24) were introduced by transformation of a PCR product. Rescue plasmids were taken from the Yeast ORF Collection (e.g., CEP3, ENP1, MPS1, and PRP2; Gelperin et al., 2005), generated by cloning the gene (e.g., ALR1, CLP1, and RRP43) ± 500 base pairs of flanking sequence into pRS41N using SacI and HindIII sites introduced by PCR, or obtained from the laboratory of Vivien Measday (e.g., CBF2 and IPL1) or Doug Koshland (SMC1, SMC3, and SMC4). All primer sequences are available upon request.

Screen for poly(A)-RNA accumulation

Yeast strains from mutant collections (Ben-Aroya et al., 2008; Breslow et al., 2008; Li et al., 2011) were grown in yeast extract/peptone/dextrose into log phase and then shifted directly to 37°C for 3 h. After temperature shift, cells were fixed for 15 min using 5% formaldehyde, and poly(A)-RNA was detected by FISH using an fluorescein isothiocyanate–labeled oligo-dT probe (Cole et al., 2002). After hybridization of the probe and washing steps, mounting medium with DAPI was applied to each sample and a coverslip was affixed. Imaging was performed on an Olympus IX81 microscope with 100× oil immersion objective (numerical aperture [NA] 1.4) controlled by MetaMorph software (Molecular Devices, Sunnyvale, CA) using identical exposure settings. During each day of screening, ura10Δ, mex67-5, and dbp5-1 strains were included as controls to ensure consistency in the FISH procedure. Imaging data (>200 cells) were used to visually score nuclear accumulation of poly(A)-RNA, with mutants showing evidence of accumulation being validated in triplicate. The identity of each strain found to accumulate poly(A)-RNA was subsequently verified by PCR.

Gene-specific FISH

mRNA transcripts were detected using a mixture of up to 48 fluorescently labeled gene probes of 20 nucleotides in length (Biosearch Technologies, Petaluma, CA) with a modified oligo-dT FISH procedure (Amberg et al., 1993). Probe sequences used for each transcript are listed in the Supplemental Data. Briefly, cells were fixed with 5% formaldehyde for 15 min at 37°C, washed with buffer A (0.1 M potassium phosphate, pH 6.5, and 0.5 mM MgCl₂), and treated with Zymolyase (250 μg/ml) in buffer B (0.1 M potassium phosphate, pH 6.5, 0.5 mM MgCl₂, and 1.2M sorbitol) for 35 min at 37°C. Spheroplasted cells were spotted on poly-l-lysine–coated slides, incubated for 10 min, washed with buffer A to remove unattached cells, permeabilized in cold methanol for 6 min and acetone for 30 s, and allowed to air dry. Cells were rehydrated with hybridization solution (5× SSC [saline-sodium citrate buffer], 5× Denhardt’s solution, 0.1% Tween 20, 0.01 mg/ml single-stranded DNA, 0.02 mg/ml Escherichia coli tRNA, and 10 mM vanadyl ribonucleoside complex) for 5 min and incubated with fresh hybridization buffer for at least 1 h at 37°C. Hybridization buffer with 1 ng of fluorescein-labeled LNA oligo-dT probe (Exiqon, Woburn, MA), 20 ng of Quasar 570–labeled mRNA probe, and/or 4 ng of Quasar 670–labeled ITS1 probe was added and incubated overnight (∼14 h) at 37°C in a humidity chamber. The next day, cells were washed sequentially with 2× SSC, 1× SSC, 0.5× SSC, and 2× phosphate-buffered saline for 5 min at room temperature, and after the final wash, slides were dipped into 100% ethanol for 10 s and air dried, mounting medium with DAPI was applied to each sample, and a coverslip was affixed. Imaging was performed on a DeltaVision Elite microscope system equipped with a front-illuminated scientific complementary metal-oxide-semiconductor (sCMOS) camera driven by Softworx 6 (GE Healthcare, Piscataway, NJ) using an Olympus 60×/1.42 NA oil objective. Image analysis was performed in FIJI (Schindelin et al., 2012), with each three-dimensional smFISH data set being reduced to a single maximum Z-projection. mRNA FISH signals were identified as single points within FIJI using noise tolerances set for each individual image and/or smFISH experiment to minimize the detection of false spots and compensate for sample variability. Masks of DAPI and ITS1 signals were used to quantify the number of mRNAs within these compartments and determine the number of cells within the image being analyzed. Average transcript number per cell was determined from each image, given the number of transcripts and nuclei counted. In control and most mutant strains, foci varied in intensity within a small range and were therefore counted as single transcripts. Bright nuclear foci were apparent in some mutants, but these were counted as single transcripts. This approach was taken for three reasons: 1) where tested, these foci were near transcription sites, suggesting that they were not mature transcripts or were not released from the site of transcription, 2) the distribution of smFISH probes across the entire length of the mRNA did not allow the differentiation of multiple short transcripts from a full-length mRNA, and 3) these bright foci were generally rare.

Imaging mRNA processing factors

Yeast strains carrying integrated GFP reporters were grown in selective media overnight at room temperature into log phase and shifted directly to 37°C for 3 h. After temperature shift, cells were fixed for 15 min using 4% paraformaldehyde, and poly(A)-RNA was detected using 2 ng of TYE 563–labeled LNA oligo-dT probe (Exiqon) and 4 ng of Quasar 670–labeled ITS1 probe (Biosearch Technologies) by FISH (Cole et al., 2002). After the final wash, slides were dipped in 100% ethanol for 10 s and air dried, mounting medium with DAPI was then applied to each sample, and a coverslip was affixed. Imaging was performed on a DeltaVision Elite microscope system equipped with a front-illuminated sCMOS camera driven by Softworx 6 using an Olympus 60×/1.42 NA oil objective. Image analysis was performed in FIJI (Schindelin et al., 2012).

Abbreviations used:

mRNP

messenger ribonucleoprotein

NPC

nuclear pore complex

Nup

nucleoporin

RNP

ribonucleoprotein

smFISH

single-molecule fluorescent in situ hybridization.