Advertisement

Dissecting mechanisms of nuclear mRNA surveillance in THO/sub2 complex mutants

Mathieu Rougemaille, Rajani Kanth Gudipati, Jens Raabjerg Olesen, Rune Thomsen, Bertrand Seraphin, Domenico Libri, Torben Heick Jensen

Author Affiliations

  1. Mathieu Rougemaille1,
  2. Rajani Kanth Gudipati1,2,
  3. Jens Raabjerg Olesen2,,
  4. Rune Thomsen2,
  5. Bertrand Seraphin1,
  6. Domenico Libri*,1,2 and
  7. Torben Heick Jensen*,2
  1. 1 Centre National de la Recherche Scientifique, Centre de Genetique Moleculaire, Gif sur Yvette, France
  2. 2 Department of Molecular Biology, Centre for mRNP Biogenesis and Metabolism, Aarhus University, Arhus C, Denmark
  1. *Corresponding authors: Department of Molecular Biology, Centre for mRNP Biogenesis and Metabolism, University of Aarhus, CF Møllers Alle, Bldg 130, Aarhus, Aarhus C 8000, Denmark. Tel.: +45 8942 2609; Fax: +45 8619 6500; E-mail: thj{at}mb.au.dk Centre National de la Recherche Scientifique, Centre de Genetique Moleculaire, Gif sur Yvette, France. Tel.: +33 1 698 23809; Fax: +33 1 698 23877; E-mail: libri{at}cgm.cnrs-gif.fr
  • Present address: Hoiberg A/S, Store Kongesgade 59A, 1264 Copenhagen K, Denmark.

Abstract

The nuclear exosome is involved in numerous RNA metabolic processes. Exosome degradation of rRNA, snoRNA, snRNA and tRNA in Saccharomyces cerevisiae is activated by TRAMP complexes, containing either the Trf4p or Trf5p poly(A) polymerase. These enzymes are presumed to facilitate exosome access by appending oligo(A)‐tails onto structured substrates. Another role of the nuclear exosome is that of mRNA surveillance. In strains harboring a mutated THO/Sub2p system, involved in messenger ribonucleoprotein particle biogenesis and nuclear export, the exosome‐associated 3′ → 5′ exonuclease Rrp6p is required for both retention and degradation of nuclear restricted mRNAs. We show here that Trf4p, in the context of TRAMP, is an mRNA surveillance factor. However, unlike Rrp6p, Trf4p only partakes in RNA degradation and not in transcript retention. Surprisingly, a polyadenylation‐defective Trf4p protein is fully active, suggesting polyadenylation‐independent mRNA degradation. Transcription pulse–chase experiments show that HSP104 molecules undergoing quality control in THO/sub2 mutant strains fall into two distinct populations: One that is quickly degraded after transcription induction and another that escapes rapid decay and accumulates in foci associated with the HSP104 transcription site.

Introduction

Processing of RNA polymerase II (RNAPII) transcripts and their assembly with proteins into export‐competent messenger ribonucleoprotein particles (mRNPs) are error‐prone processes challenged by nuclear quality control systems. Such control occurs early and sometimes even before mRNP release from the gene (Saguez et al, 2005). This is needed because most factors involved in mRNA processing, packaging and export act cotranscriptionally (for reviews see Jensen et al, 2003; Vinciguerra and Stutz, 2004; Buratowski, 2005; Fasken and Corbett, 2005). One example is provided by the Saccharomyces cerevisiae THO complex composed of four components (Hpr1p, Mft1p, Tho2p and Thp1p). The THO complex affects mRNP assembly and export via its interaction with mRNA export factors Sub2p and Yra1p in the TREX (TRanscription/EXport) complex and is required for efficient recruitment of Sub2p to nascent RNA (Strasser et al, 2002; Zenklusen et al, 2002). Deletion or mutation of any of the THO/TREX components results in rapid nuclear accumulation of polyadenylated mRNA and sequestration of heat‐shock (hs)‐RNAs, HSP104 and SSA4, in transcription site‐associated foci (dots) at elevated temperatures (Jensen et al, 2001a; Libri et al, 2002; Strasser et al, 2002; Zenklusen et al, 2002; Thomsen et al, 2003; Vinciguerra et al, 2005). This hs‐RNP retention requires the 3′ → 5′ exonuclease Rrp6p, an auxiliary component of the nuclear exosome (Jensen et al, 2001a; Libri et al, 2002; Vinciguerra et al, 2005). Steady‐state levels of full‐length hs‐RNAs are also strongly diminished in THO deletion and sub2 mutant strains, a defect which is restored upon RRP6 deletion (Libri et al, 2002). This suggests that in addition to its effect on retention of aberrant hs‐RNP, Rrp6p is also involved in degrading hs‐RNA when export is compromised or inefficient. The mechanistic relationship between these two functions of Rrp6p has not yet been elucidated.

In addition to mRNA quality control, Rrp6p and the nuclear exosome also participate in a wide range of RNA‐degradative reactions, including processing and complete degradation of stable RNAs (rRNA, snoRNA, snRNA and tRNA) as well as degradation of so‐called cryptic unstable transcripts (CUTs) (van Hoof and Parker, 1999; Mitchell and Tollervey, 2000; Butler, 2002; Kadaba et al, 2004; Wyers et al, 2005; Davis and Ares, 2006). Purified exosome only shows weak exonucleolytic activity in vitro, a feature that was recently used to identify an exosome activating complex coined TRAMP (Trf4p/Air2p/Mtr4p polyadenylation) (LaCava et al, 2005). The TRAMP complex harbors a poly(A) polymerase (Trf4p or its relative Trf5p) that belongs to the same polymerase family as the conventional Pap1p enzyme (LaCava et al, 2005; Vanacova et al, 2005; Wyers et al, 2005). However, Trf4p lacks the RNA‐binding domain of Pap1p, an activity presumably provided by Air2p, or its relative, Air1p. Adenylation‐triggered RNA decay is widespread in prokaryotes, and thus TRAMP is believed to ease exosome access onto difficult substrates (e.g. highly structured ones) by adding an unstructured tag (Dreyfus and Regnier, 2002; LaCava et al, 2005; Vanacova et al, 2005; Wyers et al, 2005). TRAMP‐assisted exosomal decay might subsequently occur through multiple rounds of adenylation/degradation. Finally, it has been shown that Trf4p/Trf5p‐dependent RNA degradation in eukaryotes is not only restricted to complete decay but also takes place in RNA processing pathways (Egecioglu et al, 2006).

Recently, Trf4p has been linked to nuclear processes involving authentic mRNAs. In yeast, removal of Trf4p increased levels of NRD1 RNA, which in a wild‐type (wt) context is kept low by autoregulated premature transcription termination of the NRD1 gene (Arigo et al, 2006). Furthermore, after depletion of human Rrp6 (PM/Scl100) from HeLa cells, accumulation of mRNA degradation intermediates harboring adenylated 3′ends was observed reminiscent of TRAMP‐mediated exosome decay (West et al, 2006).

In this paper, we present direct evidence that TRAMP complexes harboring Trf4p are involved in nuclear mRNA quality control in S. cerevisiae. However, in contrast to Rrp6p, Trf4p is not required for retention in nuclear foci of either the inducible HSP104 RNA or the constitutively expressed PDR5 RNA. Transcription pulse–chase experiments of the HSP104 gene reveal a partitioning of molecules into a pool undergoing rapid decay and a pool of stable RNAs retained in transcription site foci. On the basis of our results, a model for nuclear mRNA surveillance is discussed.

Results

Polyadenylation‐independent stimulation of HSP104 RNA degradation by Trf4p

To assess directly the involvement of Trf4p in nuclear mRNA surveillance, we deleted the TRF4 gene in cells lacking either one of the two THO components Mft1p or Hpr1p, and in cells carrying the previously described temperature‐sensitive (ts) sub2‐201 allele (Jensen et al, 2001a). Compared with a wt strain, HSP104 RNA levels are severely decreased in mft1Δ, hpr1Δ and sub2‐201 single mutants, after a brief 15 min transcription induction at 37°C, as a result of Rrp6p‐dependent degradation from the 3′end (Libri et al, 2002). We therefore used these conditions and conducted HSP104 RNA Northern blotting and quantitative RT–PCR analysis utilizing a Northern probe and PCR primer pairs specifically targeting the HSP104 3′end (Figure 1A). Deletion of the TRF4 gene caused a significant increase in HSP104 transcript levels in all three mft1Δ, hpr1Δ and sub2‐201 mutant backgrounds (Figure 1B, compare lanes 3 and 4; and D, left panel, compare lanes 2 and 4, and right panel, compare lanes 10 and 11). This was even more evident when polyadenylated 3′‐ends were enzymatically removed by oligo(dT)/RNaseH treatment (Figure 1B and D, panels labeled ‘dT treated’). Real‐time RT–PCR assays confirmed the Northern blot results (Figure 1C). This effect of TRF4 deletion was nearly as prominent as that of deleting RRP6 (Supplementary Figure 1) and strongly implies that the function of the nuclear exosome in aberrant mRNA decay is stimulated by Trf4p.

Figure 1.

Trf4p participates in HSP104 RNA nuclear surveillance. (A) Schematic representation of the assayed HSP104 transcript showing the approximate positions of the utilized RNaseH cleaving DNA oligoes (DL163 and dT), the RT–real‐time PCR primers and the Northern probe. (B) HSP104 Northern analysis of RNA harvested from the indicated yeast strains after a 15 min temperature shift from 25 to 37°C. RNAs visualized in lanes 1–8 and 9–16 (denoted ‘dT‐treated’) were RNaseH‐cleaved by DNA oligonucleotides DL163 and DL163/dT, respectively, before gel loading. U6 snRNA was probed as a control for sample loading. The migration of HSP104 molecules harboring wt‐length poly(A) tails (A+), as well as hyperadenylated poly(A) tails (A++), is denoted to the left of the image. The strains mft1Δ/trf4‐236#1 and mft1Δ/trf4‐236#2 represent two independent spores arising from the cross between mft1Δ and trf4‐236 single mutant strains. (C) Quantitative RT–real‐time PCR analysis of HSP104 RNA 3′ends from some of the RNA samples described in (B). HSP104 RNA levels were normalized to ACT1 RNA, which was unaffected by the relevant mutations. The HSP104/ACT1 RNA ratio from wt samples was set to 1. HSP104 PCR primers are depicted in (A). Averages and standard deviations are calculated from three experiments. (D) HSP104 Northern analysis of RNA harvested from the indicated yeast strains after a 15 min heat pulse at 37°C. Sample treatments and notations of RNAs as described in (B). Lanes 9–14 are all taken from the same PhosphorImage scan. (E) HSP104 Northern analysis of RNA harvested from the indicated yeast strains after a 15 min heat pulse at 37°C. Samples denoted ‘dT selected’ were passed over oligo(dT) columns before loading. Sample treatments and notations of RNAs are as in (B). The two bands denoted by asterisks are of unknown origin. All lanes are taken from the same PhosphorImage scan.

It has been suggested that the polyadenylation activity of Trf4p stimulates substrate degradation by the exosome (Vanacova et al, 2005). To test the involvement of polyadenylation by Trf4p in HSP104 degradation, we constructed the mft1Δ/trf4‐236 and hpr1Δ/trf4‐236 mutant strains where deletion of the HPR1 or MFT1 genes is associated with a polyadenylation‐defective, catalytic site mutant allele of TRF4 (Wyers et al, 2005). These strains were analyzed for their HSP104 RNA content after a 15 min heat shock at 37°C. Surprisingly, the Trf4‐236p protein substitutes fully for wt Trf4p in the degradation of HSP104 RNA in mft1Δ and hpr1Δ cells (Figure 1B, lanes 6–8 (and 14–16); C and D compare lanes 2 and 3 (and 6 and 7)). Therefore, the polyadenylation activity of Trf4p is not required for the degradation of HSP104 transcripts.

Adenylated forms of HSP104 are produced by Pap1p

Consistent with previously published data, RNaseH/oligo(dT)‐directed removal of poly(A) sequences reveals that virtually all detectable complete HSP104 RNAs in mft1Δ, hpr1Δ and sub2‐201 strains are polyadenylated (Figure 1B and D; Libri et al, 2002). Interestingly, a fraction of these RNAs harbors poly(A) tails that are slightly longer than those of the wt control (denoted A++ in Figure 1B and D). TRF4 deletion does not lead to removal of these HSP104 species, arguing that this poly(A) polymerase is not responsible for the adenylation (Figure 1B and D). To analyze poly(A) addition requirements in the mft1Δ background in more detail, we constructed mft1Δ/pap1‐1 and mft1Δ/trf4Δ/pap1‐1 mutant strains and subjected them to HSP104 RNA biochemical analysis. Introduction of the pap1‐1 mutation clearly decreased the extent of polyadenylated HSP104 RNAs in both mft1Δ and mft1Δ/trf4Δ mutant contexts (Figure 1E, upper panel). Consistently, very little HSP104 RNA was purified after oligo(dT) selection of total RNA from these strains (Figure 1E, lower panel). We conclude that polyadenylation in the mft1Δ strain background of HSP104 RNAs with both normal and hyperadenylated poly(A) tails, is carried out by Pap1p.

Trf4p is not required for mRNA retention in nuclear foci

HSP104 RNAs accumulate in Rrp6p‐dependent transcription site‐associated foci in THO/TREX mutant strains (Jensen et al, 2001a; Libri et al, 2002; Thomsen et al, 2003; Vinciguerra et al, 2005). To evaluate the effect of TRF4 deletion on HSP104 RNA retention, the mft1Δ/trf4Δ, hpr1Δ/trf4Δ and sub2‐201/trf4Δ strains were subjected to HSP104 RNA fluorescent in situ hybridization (FISH) analysis. As controls, single mutants mft1Δ, hpr1Δ and sub2‐201 as well as double mutants mft1Δ/rrp6Δ, hpr1Δ/rrp6Δ and sub2‐201/rrp6Δ were examined simultaneously. As previously reported, all three mft1Δ, hpr1Δ and sub2‐201 single mutants retain HSP104 RNA in nuclear foci after a 15 min shift to the non‐permissive temperature of 37°C, whereas codeletion of RRP6 in all three contexts abolishes HSP104 RNA retention (Figure 2A; Jensen et al, 2001a; Libri et al, 2002). Strikingly, TRF4 deletion has no effect on HSP104 RNA localization in neither of the three mutant backgrounds.

Figure 2.

Nuclear retention of mRNA in THO/sub2 mutants in the absence of TRF4. (A) HSP104 RNA‐FISH on fixed samples of the indicated yeast wt and mutant strains. Cells were grown at 25°C followed by a temperature shift to 37°C for 15 min before fixation. HSP104 RNA was detected using a mixture of three Cy3‐labeled oligonucleotide probes directed against the 3′end of the transcript. DNA was stained with DAPI. Retained HSP104 RNA was detected in >95% of the mft1Δ, mft1Δ/trf4Δ, hpr1Δ, hpr1Δ/trf4Δ, sub2‐201 and sub2‐201/trf4Δ cells in a given field of view. (B) Left panel: PDR5 RNA‐FISH on fixed samples of the indicated yeast wt and mutant strains after a temperature shift to 37°C for 15 min. PDR5 RNA was detected by a mixture of three Cy3‐labeled oligonucleotide probes directed against the transcript. DNA was stained with DAPI. The fraction of cells with detectable PDR5 RNA dot signals are shown to the right of the images. Right panel: quantitative real‐time RT–PCR analysis of PDR5 RNA 3′ends from RNA samples harvested from cultures shown in the left panel. PDR5 RNA levels were normalized to ACT1 RNA and the PDR5/ACT1 RNA ratio from wt samples was set to 1. Averages and standard deviations are calculated from three experiments. (C) Poly(A)+ RNA‐FISH on the indicated fixed samples of yeast cells after a temperature shift to 37°C for 15 min. Poly(A)+ RNA was detected using an LNA‐modified dT20 probe (Thomsen et al, 2005). DNA was stained with DAPI.

To learn more about the generality of mRNA surveillance in THO/sub2 mutants, we examined RNA expressed from the PDR5 gene, which we have identified as a THO complex target in a genome‐wide study (Rougemaille et al, in preparation). PDR5‐FISH probes detected a weak dot signal in up to 20% of wt cells (Figure 2B, left panel), which presumably stems from nascent RNA transcribed off the 4.5 kb long PDR5 gene. In hpr1Δ cells, signal intensity increased and 95% of analyzed cells contained the PDR5 RNA dot. As observed for hsRNAs, co‐deletion of TRF4 had no dramatic effect on dot intensity, whereas RRP6 deletion resulted in a decrease back to wt levels (Figure 2B, left panel). At the level of RNA decay, the PDR5 RNA also phenocopied that of hsRNAs: low PDR5 3′end levels in hpr1Δ are restored both by TRF4 and RRP6 deletion (Figure 2B, right panel). We conclude that nuclear mRNA surveillance in THO/sub2 mutants is not restricted to hsRNAs.

Finally, we also investigated the effect of deleting or mutating TRF4 in the mft1Δ background on the localization and levels of general poly(A)+ RNA using an oligo(dT20) LNA‐modified FISH probe (Figure 2C; Thomsen et al, 2005). In the mft1Δ strain, a robust granular poly(A)+ RNA signal, overlapping the DAPI‐stained chromatin region, was detected, which neither deletion of TRF4 nor addition of the trf4‐236 allele changed significantly (Figure 2C, right panel). Thus, transcripts detected by the oligo(dT20) probe behave similar to PDR5‐ and HSP104‐RNAs. Consistent with previous observations, this probe detects a low level of poly(A)+ RNA in wt cells (Figure 2C, left panel; Thomsen et al, 2005). This signal does not overlap that of the DAPI stain and is believed to stem from polyadenylation of stable nucleolar RNAs or from random occurrence of A stretches. In contrast to TRF4 removal, introduction of the pap1‐1 allele reduced the intensity of the chromatin‐associated poly(A)+ signal considerably (Figure 2C, right panel, lower row). Thus, also at the level of RNA poly(A)+‐FISH, it is clear that Pap1p provides the major adenylation activity to chromatin‐restricted mRNAs.

The entire TRAMP complex operates in mRNA surveillance

Besides Trf4p, the TRAMP complex also harbors one of the two RNA‐binding proteins Air1p or Air2p (LaCava et al, 2005; Vanacova et al, 2005; Wyers et al, 2005). A third TRAMP component is the RNA helicase Mtr4p, which also shares functions with the nuclear exosome (de la Cruz et al, 1998). To evaluate the context requirements for Trf4p function in mRNA surveillance, we constructed the mft1Δ/air1Δ/air2Δ triple mutant and subjected it to HSP104 RNA‐FISH and Northern analysis. Deletion of AIR1 and AIR2 in the mft1Δ context largely phenocopied the deletion of TRF4: there was only a slight decrease in the number of HSP104 RNA dot‐positive cells compared with the mft1Δ single mutant (Figure 3A), and HSP104 RNA 3′‐end levels were restored (Figure 3B, compare lane 7 with 8 and 9). We also tested the impact of removing the Rrp6p‐interacting protein, Rrp47p (Mitchell et al, 2003), in the mft1Δ context, and found that while HSP104 RNA 3′‐end levels were restored (Figure 3B, compare lane 2 with lanes 3 and 4), the number of HSP104 dot‐positive cells were 40%, with a concomitant more diffuse FISH signal (Figure 3A). Thus, although not as penetrant, the effect of Rrp47p removal mimics that of an RRP6 deletion.

Figure 3.

HSP104 RNA surveillance characteristics of TRAMP and nuclear exosome components. (A) HSP104 RNA‐FISH on fixed samples of the indicated yeast strains after a temperature shift to 37°C for 15 min. HSP104 RNA was detected as described in legend to Figure 2A. DNA was stained with DAPI. The fraction of cells with detectable HSP104 RNA dot signals are shown to the right of the images. For experiments involving the pTET‐repressible MTR4 construct, cultures were either treated with 10 μg/ml of doxycycline, or not, for 6 h before the temperature shift. (B) HSP104 Northern (left) or quantitative RT–real‐time PCR analysis (right) of RNA harvested from the indicated yeast strains after a 37°C temperature shift for 15 min. Experimental details are as described in legends to Figure 1B and C. The strains mft1Δ/rrp47Δ#1 and mft1Δ/rrp47Δ#2 as well as mft1Δ/air1Δ/air2Δ♯1 and mft1Δ/air1Δ/air2Δ#2 represent two independent spores arising from the relevant crosses.

Finally, we analyzed the effect of decreasing cellular Mtr4p levels on HSP104 RNA surveillance. This was achieved by driving expression of the genomic MTR4 gene by a tetracycline repressible promoter (Tet‐MTR4) in a wt or an mft1Δ background. After 6 h of incubation under repressive‐ (+doxycycline) or permissive (−doxycycline) conditions, cultures were shifted to 37°C for 15 min and subsequently harvested and processed for HSP104‐RNA‐FISH or real‐time RT–PCR analysis. HSP104 3′‐end levels in the mft1Δ/Tet‐MTR4 strain were not decreased as prominently compared with the wt control as in mft1Δ and wt strains carrying the endogenous MTR4 gene (Figure 3B, right panel; compare with Figure 1C). This is presumably due to altered MTR4 expression from the Tet promoter (data not shown). In any case, repressing Mtr4p production restored HSP104 RNA 3′ends back to wt levels. Moreover, lowering Mtr4p levels in the mft1Δ context also resulted in a significant decrease in HSP104 dot containing cells (Figure 3A, right panel), suggesting that these conditions impact nuclear exosome activity.

Taken together, these data strongly indicate that TRF4p functions in the context of TRAMP.

Lack of RRP6 still leads to release of aberrant mRNA in the absence of TRF4

Previously, transcription site retention of HSP104 RNA in THO/sub2 mutants always correlated with HSP104 transcript instability. Accordingly, transcripts are both released and stabilized upon RRP6 deletion. Thus, the lack of correlation between these two events upon deletion of TRAMP components in a THO/sub2 mutant was surprising, especially in light of the multiple physical and functional connections of the TRAMP complex with the exosome. The mechanism by which deletion of RRP6 promotes transcript release in, for example, an hpr1Δ strain is unknown and might involve, directly or indirectly, Trf4p. It is possible for instance that the large amounts of Trf4p‐dependent polyadenylated rRNAs and sno/snRNAs, which accumulate in rrp6Δ cells interfere with the retention process (Kuai et al, 2004; Wyers et al, 2005; Davis and Ares, 2006; Egecioglu et al, 2006). Perhaps this ‘excess’ nuclear polyadenylated species titrates away factor(s) required for transcript retention in THO/sub2 mutants, thus provoking mRNP release in for example, an hpr1Δ/rrp6Δ mutant background. Transcript release in this context would then be an indirect effect of the RRP6 deletion and should require Trf4p.

To evaluate this possibility, we constructed the triple mutant hpr1Δ/rrp6Δ/trf4Δ and subjected it to RNA‐FISH analysis. However, we first verified the presence of a prominent oligo (dT20) FISH signal in rrp6Δ single mutant cells (Figure 4A). In line with the reported polyadenylation of stable nucleolar RNA species (Kuai et al, 2004; Wyers et al, 2005; Davis and Ares, 2006; Egecioglu et al, 2006), this signal was clearly separated spatially from the DAPI stain (Figure 4A, ‘DAPI overlay’), and partly overlapped that of the nucleolar antigens Nop1p and Nsr1p (Figure 4B, ‘PolyA/NOP1’ and data not shown). Consistently, a similar poly(A)+ RNA localization was observed in a strain lacking the RRP47 gene (Supplementary Figure 2 and data not shown). Moreover, as expected for a Trf4p‐dependent polyadenylation process, deletion of TRF4 (but not inactivation of Pap1p) significantly decreased the nucleolar dT signal observed in rrp6Δ cells (Figure 4A). The low residual poly(A)+ RNA level in the nucleolus of rrp6Δ/trf4Δ cells is most likely produced by a Trf5p containing TRAMP complex, which was recently reported to adenylate rRNAs (Houseley and Tollervey, 2006).

Figure 4.

Deletion of TRF4 does not restore mRNA retention in Δhpr1/rrp6Δ cells. (A) Poly(A)+ RNA‐FISH on the indicated fixed samples of rrp6Δ, rrp6Δ/trf4Δ and rrp6Δ/pap1‐1 cells after a temperature shift to 37°C for 30 min. Poly(A)+ RNA and DNA was visualized as described in legend to Figure 2B. (B) Dual poly(A)+ RNA‐FISH and Nop1p immunolocalization analysis on fixed rrp6Δ cells temperature shifted to 37°C for 30 min. Nop1p was detected using a monoclonal anti‐Nop1p antibody followed by an FITC‐conjugated secondary antibody. (C) HSP104 RNA‐FISH on fixed samples of Δhpr1Δ, hpr1Δ/trf4Δ, hpr1Δ/rrp6Δ and hpr1Δ/rrp6Δ/trf4Δ cells as indicated. Strains were temperature shifted to 37°C for 15 min before fixation. HSP104 RNA and DNA was stained as described in legend to Figure 2A. Retained HSP104 RNA ‘dots’ were detected in >90% of the hpr1Δ and hpr1Δ/trf4Δ cells and <10% of the hpr1Δ/rrp6Δ and hpr1Δ/rrp6Δ/trf4Δ cells in a given field of view.

Decreasing the nucleolar poly(A)+ signal by TRF4 removal does not ‘re‐install’ the HSP104 dot in hpr1Δ/rrp6Δ/trf4Δ triple mutant cells (Figure 4C). Thus, by these criteria, release of mRNA from chromatin upon RRP6 removal is not a consequence of increased nucleolar poly(A)+ RNA load. These experiments also more generally indicate that Trf4p is not required for mRNP release in the hpr1Δ/rrp6Δ strain.

Persistence of nuclear HSP104 RNAs in the sub2‐201 mutant in the absence of transcription

So far, it is unclear whether the HSP104 RNAs present in THO/sub2 mutant cells after a 15 min heat pulse represent species that are in the process of being degraded or molecules that have escaped nuclear degradation. Some or all of these molecules might be in transcription site foci, although, it is not clear whether degradation is actually taking place at these sites. To investigate this issue, we analyzed localization and turnover of HSP104 RNAs in wt and sub2‐201 cells after a 15 min transcription pulse at 42°C, as well as at different time points in a subsequent chase period. Transcription was shut off by rapidly decreasing the temperature of heat‐shocked cells to 25°C and simultaneously adding the transcription inhibitor thiolutin. HSP104 RNA was analyzed by Northern blotting, real‐time RT–PCR and FISH analysis. The sub2‐201 mutant strain and the 42°C temperature were chosen because of the more robust build‐up of HSP104 RNA in transcription site foci under these conditions (data not shown).

As shown by both Northern and RT–PCR analyses, HSP104 RNA was gradually turned over in a wt strain in the absence of transcription (Figure 5A and B). In marked contrast, the lower amount of HSP104 RNA observed in sub2‐201 cells after a 15 min heat shock remained constant for at least an hour after addition of thiolutin. This was paralleled by the persistence of HSP104 RNA‐FISH dots in sub2‐201 cells after transcription shut off (Figure 5C). These experiments suggest that both total and transcription site‐restricted HSP104 RNAs in sub2‐201 cells are remarkably stable in the period following a 15 min heat shock. Such a result is intriguing because the low level of HSP104 RNAs in THO/sub2 mutants, at the 15 min time point, is restored to wt levels upon deletion of RRP6, suggesting that the majority of HSP104 RNAs are degraded extremely rapidly in single THO/sub2 mutants (Libri et al, 2002).

Figure 5.

HSP104 RNAs detected in sub2‐201 mutant cells after a 15 min heat shock are remarkably stable. (A) HSP104 Northern analysis of RNA harvested from the indicated wt or sub2‐201 strains incubated at 25°C (lanes 1 and 2), heat shocked for 15 min at 42°C (lanes 3 and 4) or heat shocked for 15 min at 42°C followed by a ‘chase period’ of the indicated length (lanes 5–12). RNAs were RNaseH‐cleaved using DNA oligonucleotide DL163 before gel loading. U6 snRNA was probed as a control for sample loading. The migration length of HSP104 molecules harboring wt length poly(A) tails (A+), and hyperadenylated poly(A) tails (A++), is denoted to the left of the phosphorImage. (B) Quantitative RT–real‐time PCR analysis of HSP104 RNA 5′ends and 3′ends from RNA samples described in (A). HSP104 RNA levels were normalized to 5S rRNA, which was unaffected by the different conditions. Averages and standard deviations are calculated from two experiments. (C) HSP104 RNA‐FISH on fixed samples of sub2‐201 cells heat shocked for 15 min at 42°C (left panel), or heat shocked for 15 min at 42°C followed by a ‘chase period’ of 30 min. HSP104 RNA was detected as described in legend to Figure 2A. DNA was stained with DAPI. Retained HSP104 RNA was detected in >80% of cells in both conditions. (D) Nuclear run on (NRO) analysis of nuclei harvested from wt or sub2‐201 cells after 15 min heat shock at 42°C. Radioactive RNA samples were hybridized to DNA oligonucleotide NRO probes complementary to the approximate positions of the HSP104 as indicated on top. Hybridization to an 18S rRNA probe was used as an internal control. HSP104 NRO signals were quantitated by normalizing to the 18S rRNA signal and setting the value of probe 2 in the wt strain to 100 (lower right).

The different half‐lives of steady‐state HSP104 RNAs in sub2‐201 and wt cells suggest that these RNAs have different fates and, notably, that in mutant cells, they do not undergo translation and ensuing cytoplasmic decay. Consistently, impairment of the major cytoplasmic decay pathway by deletion of the 5′ → 3′ exonuclease XRN1 strongly affected the pseudo‐pulse–chase kinetics of HSP104 molecules in the wt but not in the sub2‐201 background (Supplementary Figure 3). This indicates that these HSP104 RNAs in sub2‐201 cells are dead‐end products, which do not experience a cytoplasmic phase.

To demonstrate that the low amount of HSP104 RNA present in the sub2‐201 strain is not due to decreased transcription, nuclear run on (NRO) experiments were performed on nuclei isolated in these conditions from either wt or sub2‐201 cells. As shown in Figure 5D, NRO signals were indistinguishable between the two strains at every assayed position of the HSP104 gene, which indicates that RNAPII activity is not affected by mutation of Sub2p. Thus, the simplest interpretation of these experiments is that HSP104 RNAs are rapidly degraded shortly after transcriptional induction. Surprisingly, however, transcripts that evade this decay have dramatically increased half‐lives.

Discussion

Several phenotypes have been associated with mutation or deletion of individual subunits of the THO/TREX complex, including defects in transcription, genomic stability as well as mRNP biogenesis and nuclear export (Chavez et al, 2000; Jensen et al, 2001a; Libri et al, 2002; Strasser et al, 2002; Zenklusen et al, 2002; Huertas and Aguilera, 2003). Although the precise function of THO/TREX is still unresolved, THO/sub2 mutant strains provide good systems to study nuclear mRNA surveillance as these cells phenotypically illustrate the two major hallmarks of this process: (i) retention and (ii) elimination of aberrant mRNAs. In this study, we have taken advantage of THO/sub2 mutant cells to show the involvement of the TRAMP complex in the degradation‐leg of nuclear mRNA surveillance. Furthermore, we have probed the relationship between transcript retention and decay, and reached the surprising conclusion that the HSP104 transcripts remaining in mutants after a 15 min transcription induction, of which a considerable fraction is presumably transcription site associated, are remarkably stable. This is in sharp contrast to the majority of HSP104 molecules, which undergo rapid decay during the first minutes of transcription induction.

Trf4p, but not its poly(A) polymerase activity, is required for efficient degradation of HSP104 RNAs in THO/sub2 mutants

As a pool of newly synthesized HSP104 RNAs in THO/sub2 mutants, and in mutants of other mRNA export factors, are hyperadenylated (Jensen et al, 2001b; Libri et al, 2002), we reasoned that these RNAs might be nuclear exosome degradation intermediates tagged by the TRAMP complex that was recently discovered as an exosome activator for degradation and maturation of nuclear RNAs (Kadaba et al, 2004; LaCava et al, 2005; Vanacova et al, 2005; Wyers et al, 2005; Egecioglu et al, 2006). However, the long‐sized poly(A) tails are unaffected by deletion of TRF4, but rather disappear after inactivation of the canonical poly(A) polymerase Pap1p. Nonetheless, Trf4p is required for efficient degradation of HSP104 RNAs as the amount of these rises to near wt levels in double mft1Δ/trf4Δ, hpr1Δ/trf4Δ and sub2‐201/trf4Δ mutant strains. These observations therefore identify Trf4p (in the context of TRAMP) as a nuclear mRNA surveillance factor.

Importantly, HSP104 transcript degradation does not depend on the poly(A) polymerase activity of Trf4p, that is, the catalytic site mutant trf4‐236 substitutes fully for wt TRF4. This allows the somewhat surprising conclusion that polyadenylation is not an obligatory event for the TRAMP complex to stimulate the nuclear exosome. Other recent data also suggest that polyadenylation might not be of universal importance for TRAMP activity. For instance, trf4‐236 only partially phenocopies the effect of a TRF4 gene deletion on the degradation of CUTs (Wyers et al, 2005). Moreover, artificial addition of an oligo(A) tail to a synthetic structured RNA substrate in vitro is not sufficient per se to promote TRAMP‐independent degradation by the exosome (LaCava et al, 2005). Consistently, Trf4p, but not its poly(A) polymerase activity, is essential in the absence of its paralogue Trf5p (Wyers et al, 2005). The mechanism by which TRAMP promotes polyadenylation‐independent stimulation of exosome activity is so far unknown, but it may include the helicase activity of Mtr4p, or the RNA‐binding activity of the Air1/2p proteins, to favor exosome recruitment to the substrate.

Although TRAMP is required for efficient degradation of the bulk of HSP104 RNAs in THO/sub2 mutants, lack of TRAMP components Trf4p, Air1p and Air2p does not affect the transcription site‐associated retention of HSP104 RNA in these mutants. This phenotype contrasts that of an RRP6 deletion where HSP104 transcripts in THO/sub2 mutants are both stabilized and released. The finding implies that either transcript retention is not mechanistically coupled to degradation, or that the Rrp6p degradation activity that mediates retention is Trf4p‐independent. As retained transcripts are remarkably stable (see below), it is possible that TRAMP is only required for the exosome to access malformed RNAs that have been released into the nucleoplasm.

A model for nuclear mRNA surveillance in THO/sub2 mutants

HSP104 transcription pulse–chase experiments in the sub2‐201 mutant show only minor decay of the total transcript pool in the chase period. Moreover, these RNAs do not have a cytoplasmic phase as demonstrated by their insensitivity to impairment of the major cytoplasmic decay pathway. The fraction of HSP104 RNAs, retained in transcription site foci, is also stable, as the RNA‐FISH signal persists for at least 30 min after transcription shut off. Thus, these data demonstrate the existence of nuclear mRNAs, in THO/sub2 mutants, with very low turnover rates. However, our data also strongly suggest that the major fraction of HSP104 transcripts produced in sub2‐201 cells before starting the chase are rapidly degraded. This is because after the 15 min transcription pulse, complete HSP104 RNAs in the sub2‐201 strain only constitute roughly 20% of wt levels, despite indistinguishable transcription activities in wt and sub2‐201 cells. Furthermore, deletion of RRP6 or TRF4 leads to restoration of HSP104 RNA levels in the sub2‐201 background, indicating that the ‘missing’ fraction of these transcripts (approximately 80%) was rapidly degraded. Taken together, these observations are best explained by a model in which stable HSP104 RNAs in sub2‐201 cells must have escaped early degradation and now persist in transcription site‐associated foci or elsewhere in the nucleus where destruction is occurring at a lower rate. Perhaps stochastic failure to access nuclear degradation exposes these molecules to protective coating by RNA‐binding proteins or perhaps sequestration in certain nuclear regions makes them inaccessible to the nuclear exosome, or to TRAMP.

What then is the nature of the Rrp6p‐dependent mechanism that retains RNAs? First, Rrp6p, and possibly the exosome, might physically tether malformed mRNPs and RRP6 deletion might break these links, thus allowing release. Although this explanation is consistent with the fact that the exosome can be crosslinked to the HSP104 locus in a transcription‐dependent manner (Andrulis et al, 2002; Hieronymus et al, 2004; our unpublished results), it is at odds with the fact that we were unable to detect an increase in exosome occupancy in THO/sub2 mutants (data not shown). This implies that the exosome is recruited to chromatin independent of mRNP surveillance. Second, deletion of RRP6 could release foci‐retained HSP104 molecules because of indirect effects, for example, accumulation of polyadenylated stable RNAs in Rrp6p‐minus cells could potentially titrate away factors required for transcription site retention. As abolishing a large share of stable RNA polyadenylation by deletion of TRF4 did not restore HSP104 RNA retention in the hpr1Δ/rrp6Δ strain we can rule out this latter possibility. We cannot however exclude the existence of other indirect effects. As an important corollary, Rrp6p and Rrp47p have previously been assigned roles in mRNA export based on the detection of a nuclear poly(A)+ RNA‐FISH signal in rrp6Δ and rrp47Δ cells (Hieronymus et al, 2004). However, data presented here argue against this interpretation and instead suggest that the reported observation reflects the accumulation of poly(A)+ RNA in the nucleolus rather than retention of mRNA in the nucleus.

A third possibility, and the one we favor, is that an export‐antagonizing event, involving Rrp6p, negatively affects HSP104 RNA export and triggers the building up of transcripts detected by FISH (Figure 6). The export‐promoting activity challenged by Rrp6p could be an aspect of the mRNA 3′end formation process where transcripts with exposed 3′ends would be desired targets of the enzyme. Furthermore, major mRNP remodeling steps occur at the 3′ends of genes, which could be slowed by Rrp6p competition. Regardless the precise molecular event, we suggest that in wt conditions, the export‐promoting activity is strong enough to out‐compete Rrp6p. However, in a THO/sub2 mutant context, diminished/slowed transcript release allows Rrp6p to attack kinetically disfavored mRNPs. Most of these are rapidly degraded in the first wave of decay (requiring TRAMP for complete degradation). A minor fraction (approximately 20%) of the molecules, represented by the stable pool of HSP104 RNA in nuclear dots, is engaged in a non‐productive degradation pathway despite the fact that they also failed to productively enter export. Perhaps these mRNPs escaped rapid decay because of local exhaustion of decay factors; however, they also missed the export pathway because of insufficient ‘release signals’ (e.g. THO/Sub2p), or because they simply missed the right kinetic window within which export competence can be achieved. An important feature of this model is that Rrp6p is involved in mRNA dot creation rather than maintenance.

Figure 6.

Model of transcription site‐associated mRNA surveillance in THO/sub2 mutants. Upper half: in wt conditions, cotranscriptional mRNA packaging and 3′end formation occur efficiently and the mRNP particle is released for export. Lower half: in a THO/sub2 mutant context, a less efficient assembly/processing pathway leads to Rrp6p/exosome intervention and rapid RNA decay. A minor fraction of molecules escape nuclear degradation but cannot enter a productive export pathway. See text for further details.

The factors that ultimately define export competence as well as the dynamics of factor exchange around the nascent RNA molecule in the late phases of transcription are still elusive. Their definition will be of paramount importance to unveil the precise mechanism underlying quality control of mRNA at the site of transcription.

Materials and methods

Yeast strains and manipulations

Yeast strains used in this study are all derived from W303 and are described in Supplementary Table S1. Crosses were performed using standard laboratory procedures. All temperature shifts were performed by the addition of an equal volume of prewarmed medium to a 25°C culture, followed by incubation at 37 or 42°C for the appropriate times. For experiments involving the transcription inhibitor thiolutin, cells were heat shocked at 42°C for 15 min, rapidly pelleted and resuspended in medium containing 50 μg/ml of thiolutin at 25°C. Aliquots were taken at different time points for RNA analyses. It was verified that addition of thiolutin before heat shock efficiently inhibited HSP104 transcription activation (data not shown).

For experiments with cells containing the MTR4 gene under control of the Tet promoter, cultures were grown for 6 h in the presence of 10 μg/ml doxycyclin to repress MTR4 expression before HSP104 induction.

RNA‐FISH analysis

RNA‐FISH and dual RNA‐FISH/protein immunolocalization analyses were performed as previously described (Jensen et al, 2001b; Thomsen et al, 2003, 2005). HSP104‐ and PDR5‐RNAs were detected using a mixture of Cy3 body‐labeled oligonucleotide probes. HSP104 probes THJ203, THJ204 and THJ206 have previously been described (Jensen et al, 2001b). For PDR5 detection, the following probe mix was used (X denotes an Amino‐C6‐dT modification disposed for Cy3‐labeling): PDR5‐1 (5′‐CAGAGXCCTTGCCAGXTTTTGGATXCGAGCTTCXGTATGCTCAXCGAACC), PDR5‐2 (CCXGGTCTACCXAAAACGACTAGCAAXTCACCTGGGXTTAGGCAACCAXCC) and PDR5‐3 (CCCXATCGACACCCXTGATACGGXTCTGTGGGGXTTTCAACCTCGCXACTG). Poly(A)+ RNA was detected by an LNA‐modified dT20 probe 5′end labeled with Cy3 (Thomsen et al, 2005). Nsr1p, Nop1p and Nsr1p monoclonal antibodies (EnCor) were used for immunostaining following the manufacturer's recommendations.

RNA preparations and analysis

RNA was prepared by the hot phenol method as described previously (Libri et al, 2002). Reverse transcription was performed with MMLV reverse transcriptase (Invitrogen) and cDNA synthesis was primed with random hexamers, oligo dT and specific primers directed against U4 snRNA. The reaction was diluted ten times before real‐time PCR analysis. Amplifications were performed in duplicate using a LightCycler (Roche) with no‐reverse transcriptase controls to estimate the contribution of contaminating DNA. Amplification efficiencies were measured for each primer pairs and every set of amplification reactions. Amplification primers have the following sequences: HSP104‐3′: DL528 (sense): 5′‐GTTCTACCAAATCACGAAGC and DL529 (antisense): 5′‐TCTAGGTCATCATCAATTTCC; HSP104‐5′: DL258 (sense): 5′‐ATATGAACGACCAAACGC and DL259 (antisense): 5′‐AGATCATAGTCGTAACGGC); ACT1: DL377 (sense): 5′‐ATGTTCCCAGGTATTGCCGA and DL378 (antisense): 5′‐ACACTTGTGGTGAACGATAG; PDR5: DL795 (sense): 5′‐ACTGACACCTGTAGTTTCTGTC and DL796 (antisense): 5′‐TTCTCCATCTCTCACTGTAGA. At least three independent experiments were performed to calculate averages and standard deviations. RNaseH/ Northern blot analysis was performed as described previously (Libri et al, 2002).

Transcription analysis

NRO analysis on the HSP104 gene was performed as described previously (Jensen et al, 2004).

Supplementary data

Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).

Supplementary Information

Supplementary Information [emboj7601669-sup-0001.pdf]

Acknowledgements

We thank Dorthe Riishøj and Jocelyne Boulay for excellent technical assistance, F Lacroute for help with yeast crosses and H Klein, A Aguilera and A Lebreton for gift of strains and also thank members of the DL and THJ laboratories for stimulating discussions, and Michael Rosbash, Claire Moore, Bjarne Bonven and Sylvie Camier for critical reading of the manuscript. This work was supported by La Ligue Contre le Cancer (BS, Equipe labilisée 2005), the CNRS (DL), the ANR (programme CUTs) (DL), the foundation pour la Recherche Medicale (DL, Equipe labilisee FRM 2007), the Danish National Research Foundation (Grundforskningsfonden) (DL and THJ) and the Novo Nordisk Foundation (THJ). MR is a recipient of a fellowship from the French Ministry of Research.

References