|Home | About | Journals | Submit | Contact Us | Français|
Ribosome synthesis depends on nutrient availability, sensed by the target of rapamycin (TOR) signaling pathway in eukaryotes. TOR inactivation affects ribosome biogenesis at the level of rRNA gene transcription, expression of ribosomal proteins (r-proteins) and biogenesis factors, preribosome processing, and transport. Here, we demonstrate that upon TOR inactivation, levels of newly synthesized ribosomal subunits drop drastically before the integrity of the RNA polymerase I apparatus is severely impaired but in good correlation with a sharp decrease in r-protein production. Inhibition of translation by cycloheximide mimics the rRNA maturation defect observed immediately after TOR inactivation. Both cycloheximide addition and the depletion of individual r-proteins also reproduce TOR-dependent nucleolar entrapment of specific ribosomal precursor complexes. We suggest that shortage of newly synthesized r-proteins after short-term TOR inactivation is sufficient to explain most of the observed effects on ribosome production.
Survival of a cell critically relies on its ability to respond to environmental signals. Consequently, living organisms sense and react to the availability of nutrients. In eukaryotes, nutrient-dependent cell growth is governed by the target of rapamycin (TOR) signal transduction pathway which is highly conserved from yeast to humans (reviewed in reference 86). Rapamycin (RAP) is a macrocyclic lactone, which inhibits the activity of TOR-kinase-containing protein complex 1 (TORC1). TOR inactivation impairs various anabolic processes, including general translation, transcription of many genes, and ribosome biogenesis. It triggers catabolic processes like protein and RNA degradation, as well as autophagy. Clearly, TOR signaling determines the fate of the eukaryotic cell, and it is important to understand the molecular details of this central control mechanism. For consistency, we will focus in this introduction on findings in the model eukaryote Saccharomyces cerevisiae (here, called yeast), unless stated otherwise.
One important target of TOR control is ribosome biogenesis (reviewed in reference 50). Ribosome production relies on the concerted action of all three DNA-dependent RNA polymerases (Pols I, II, and III) to produce equimolar amounts of the structural components of the ribosome, the four ribosomal RNAs (rRNAs) and the more than 70 ribosomal proteins (r-proteins). Furthermore, more than 150 auxiliary factors participate in ribosome maturation (reviewed in references 18, 29, and 72). TOR signaling has been shown to affect this complex process on several levels: (i) transcriptional regulation of all three polymerases, (ii) translation initiation, (iii) RNA processing, and (iv) internuclear and nucleo-cytoplasmic transport processes. These effects are described below.
(i) Early analyses have demonstrated that amino acid depletion and impaired TOR signaling downregulate RNA Pol II-dependent transcription of r-protein genes (58, 82). Genome-wide transcription profiling using microarrays and bioinformatics confirmed that r-protein genes define a coregulated cluster, termed the RP regulon (8, 21, 27, 33-35, 78). These studies also demonstrated that the transcriptional response of the RP regulon to rapamycin treatment and various other environmental changes was similar to that of a large group of other genes involved in ribosome biogenesis, termed the Ribi regulon. Subsequent work identified a subset of transcription factors specifically binding to DNA elements in the promoter regions of RP- and Ribi-regulon-controlled genes in a TOR-dependent manner (35, 48, 49, 61, 62, 79). Interestingly, the Ribi regulon included genes belonging to the RNA Pol I and RNA Pol III transcription machineries (21, 34, 78), thus providing a link between the different nuclear RNA polymerase systems.
The results of in vivo [C3H3]methionine pulse-labeling experiments suggested that RNA Pol I transcription and rRNA processing are strongly impaired shortly after TOR inactivation (58). Since then, various studies have explored the molecular mechanism underlying this observation. It could be shown that the amount of initiation-competent complex of RNA Pol I with the essential transcription factor Rrn3 is reduced in rapamycin-treated cells (11, 56). This observation correlated well with a reduced RNA Pol I association with the rRNA gene locus under these conditions (11, 56). Further evidence has been provided that the RNA Pol I-Rrn3 complex is a primary target of TOR signaling. Thus, it has been shown that the expression of a nondissociable fusion protein composed of the RNA Pol I subunit Rpa43 and Rrn3 attenuates the transcriptional inactivation of the three different RNA Pols upon TOR inactivation (41). In mammals, TOR-dependent changes in the phosphorylation pattern of the Rrn3 homologue transcription initiation factor 1A (TIF-IA) correlate with impaired rRNA gene transcription (51).
Recently, the HMG-box protein Hmo1 has been shown to bind to the RNA Pol I-transcribed region of the ribosomal DNA (rDNA), as well as to a subset of the r-protein gene promoter regions (4, 26, 37, 52). These observations made Hmo1 a potential candidate mediating the cross talk between RNA Pol I and RNA Pol II. Additionally, it has been reported that rapamycin treatment leads to Hmo1 dissociation from its genomic target sites (4).
Another factor which may be directly involved in downregulation of RNA Pol I and RNA Pol III transcription following rapamycin treatment is TORC1 itself. A nuclear fraction of TORC1 associates with the rRNA gene promoter and the 5S rRNA gene locus under normal growth conditions but leaves the nucleus in the presence of rapamycin or upon nutrient deprivation (42). Nuclear localization of TORC1 seems to be important for phosphorylation and, thus, inactivation of the RNA Pol III transcriptional repressor Maf1 (84). A target for TORC1-mediated phosphorylation within the RNA Pol I transcription machinery, however, has yet to be identified. Noticeably, nuclear-cytoplasmic shuttling has also been observed for mammalian TOR (38).
(ii) Rapamycin treatment for only 15 min reduces cellular protein production to less than 50% (2). One downstream consequence of rapamycin treatment is the activation of Gcn2 kinase which, in turn, phosphorylates the α subunit of eukaryotic initiation factor 2 (eIF2α), thus inhibiting translation initiation (9). TOR inactivation further affects translation initiation by the degradation of elF4G, an essential protein required for mRNA translation via the 5′ cap-dependent pathway in yeast (5). However, the exact degradation pathway remains unknown. Interestingly, in mammals TOR controls translation initiation at a similar step by phosphorylation of elF4E-binding protein under normal growth conditions. Upon rapamycin treatment elF4E-binding proteins are dephosphorylated and bind to elF4E, which in turn can no longer act in translation initiation (3, 31, 43, 77).
(iii) As mentioned above, rRNA processing is severely impaired in cells treated with rapamycin (58). This could be due to direct TOR-dependent inactivation of ribosome biogenesis factors and/or to the transcriptional downregulation of genes controlled by the Ribi regulon, leading to rapid depletion of protein factors involved in rRNA maturation. Examples that endogenous protein levels of ribosome biogenesis factors are quickly reduced in the course of rapamycin treatment are Nog1 and Nop7, the amounts of which drop in good correlation with the decrease in the respective mRNA level (32). Nevertheless, rRNA maturation defects are observed significantly before these factors become limiting.
Impaired rRNA processing is not the only example for TOR-mediated effects on RNA metabolism. TOR inactivation also destabilizes a subset of mRNAs, due to either reduced 3′ polyadenylation or accelerated deadenylation and to 5′ decapping (1). It is not understood, however, why only certain mRNAs are targeted and by which of the different mechanisms they are eventually destabilized.
(iv) Finally, recent work has provided evidence that TOR signaling is involved in the control of preribosomal transport processes (32, 76). Nucleolar entrapment of ribosome biogenesis factors early after rapamycin treatment has been observed and has been correlated with the cessation of late rRNA maturation steps or the inhibition of nucleo-cytoplasmic translocation of preribosomes.
Taken together, these observations indicate that TOR inhibition appears to affect ribosome biogenesis on many different levels. However, it has become increasingly difficult to distinguish between primary and secondary effects since the multiple processes leading to mature ribosomes appear to be intimately linked. Here, we performed an in-depth analysis of yeast cellular phenotypes after 15 min of rapamycin treatment. We found that the strong decrease in rRNA production is not closely coupled to the shutdown of RNA Pol I transcription. A significant amount of RNA Pol I remains associated with the rRNA gene upon TOR inactivation, is mobile, and produces transcripts. Furthermore, rDNA interaction with the RNA Pol I transcription factor Hmo1 and other components of the RNA Pol I transcription machinery is unaltered after short times of rapamycin action. Instead, we observe that r-protein production is severely and specifically impaired under this condition. Additionally, short incubation of yeast cells in the presence of cycloheximide (CHX), an inhibitor of eukaryotic protein synthesis, leads to rRNA production defects similar to the rapamycin-mediated maturation phenotype. Both cycloheximide treatment and, in particular, depletion of single r-proteins also provoke nucleolar entrapment of ribosome biogenesis factors. Thus, we conclude that rapid depletion of the endogenous pool of free r-proteins is sufficient to explain various prominent phenotypes observed in the quick response to TOR inactivation.
Yeast strains used in this study are summarized in Table Table1.1. Unless stated otherwise, yeast cells were grown at 30°C in yeast-peptone-dextrose (YPD) medium. For strain construction standard methods using PCR-assisted transformation and homologous recombination were employed (39).
Chromatin endogenous cleavage (ChEC) and ChEC-psoralen analyses were performed as previously described (52).
Chromatin immunoprecipitation (ChIP) was essentially performed as described elsewhere (22). Primer pairs used for amplification were 969 (5′-TCA TGG AGT ACA AGT GTG AGG A-3′) and 970 (5′-TAA CGA ACG ACA AGC CTA CTC-3′), 712 (5′-GAG TCC TTG TGG CTC TTG GC-3′) and 713 (5′-AAT ACT GAT GCC CCC GAC C-3′), 710 (5′-TGG AGC AAA GAA ATC ACC GC-3′) and 711 (5′-CCG CTG GAT TAT GGC TGA AC-3′), and 920 (5′-GCC ATA TCT ACC AGA AAG CAC C-3′) and 921 (5′-GAT TGC AGC ACC TGA GTT TCG-3′). Data were evaluated using experimentally determined efficiency values for each pair of oligonucleotides. Almost identical results were obtained by analyzing the data with the comparative quantitation module of the RotorGene analysis software. Retention of specific DNA fragments was calculated as the fraction of the total input DNA. The mean values and error bars are derived from three independent ChIP experiments analyzed in triplicate quantitative PCRs.
Blot analysis and quantitation were performed as described elsewhere (52, 56). For Northern blot analysis membranes were hybridized with a 32P-labeled 25S oligonucleotide probe (probe 212, 5′-CTC CGC TTA TTG ATA TGC-3′). For the indirect end-labeling analysis shown in Fig. 2A and D, probe RDNp (52) was used. A probe derived from a 3.4-kb NcoI rDNA restriction fragment spanning nucleotides 1044 to 4491 of the 35S rRNA coding sequence was used in the Southern blot analysis depicted in Fig. 2B.
For pulse and pulse-chase experiments, 1 ml of yeast culture with an optical density at 600 nm (OD600) of 1 was sedimented at room temperature. Cells were resuspended in 100 μl of recovery buffer [2% (wt/vol) glucose, 1% (wt/vol) Bacto peptone, 0.6% (wt/vol) malt extract, 0.01% (wt/vol) yeast extract, 12% (wt/vol) mannitol, and 1.3 mM Mg(OAc)2] preheated to 30°C. Pulse-labeling was performed with 20 μCi of [3H]uracil (30 to 50 Ci/mmol; Perkin Elmer) for 15 min. For pulse-chase experiments 60 μCi of [3H]uracil was added to the cell suspension, and incubation was continued for 5 min at 30°C. Then unlabeled uracil was added to a final concentration of 1 mg/ml, and incubation was continued at 30°C for 4, 8, and 16 min. Total RNA was isolated and analyzed as described elsewhere (19).
Neosynthesis of proteins was determined essentially as described previously (64). Cells were grown at 30°C in synthetic complete dextrose (SCD) medium lacking methionine and cysteine (SCD-Met-Cys). Two OD600 units of cells were sedimented at room temperature. Cells were resuspended in 200 μl of SCD-Met-Cys medium, 15 μCi of [35S]Met-Cys (1,000 Ci/mmol; Hartmann Analytic) was added, and incubation was continued for 5 min at 30°C (pulse). Cells were sedimented at 4°C. The supernatant was discarded, and cells were stored at −20°C. For extraction, cells were resuspended in 1 ml of cold H2O, mixed with 150 μl of pretreatment solution (1.85 M NaOH, 1 M β-mercaptoethanol), and incubated for 15 min at 4°C. Proteins were precipitated with 150 μl of 55% (wt/vol) trichloroacetic acid (TCA) for 15 min at 4°C and sedimented by centrifugation at 4°C. The supernatant was discarded, and the pellet was resuspended in 25 μl of HU buffer (0.2 M Tris-Cl, pH 6.8, 5% [wt/vol] SDS, 1 mM EDTA, pH 8.0, 0.21 M β-mercaptoethanol, 8 M urea, and 0.2% [wt/vol] bromophenol blue). The sample was neutralized with ammonia vapor, if necessary. Protein samples were incubated for 10 min at 65°C. A total of 2.5 μl of the samples was subjected to scintillation counting to determine the 35S incorporation into total protein. Equal protein amounts were separated in acrylamide-urea gels (a lower separating gel consisting of 16% acrylamide, 375 mM Tris-HCl, pH 8.8, 0.1% [wt/vol] SDS and 4.5 M urea; an upper stacking gel consisting of 4% acrylamide, 125 mM Tris-HCl, pH 6.8, 0.1% [wt/vol] SDS, and 4.5 M urea). After Coomassie staining, the gel was dried and subjected to autoradiography.
Protein bands of interest were excised from Coomassie-stained gels and digested in gel with modified sequencing-grade trypsin (65, 66). Briefly, excised gel slices were cut into small cubes and washed with 50 mM NH4-CO3, 50 mM NH4-CO3-25% (vol/vol) acetonitrile, 25% (vol/vol) acetonitrile, and 50% (vol/vol) acetonitrile, followed by lyophilization. The freeze-dried gel cubes were rehydrated with an equal volume of trypsin in 50 mM NH4-CO3 (final concentration, 4 μg trypsin per 100 μl of gel) for 30 min at room temperature. After addition of another volume of 50 mM NH4-CO3 and incubation for 16 h at 37°C, tryptic peptides were recovered from the gel by diffusion upon shaking the samples twice for 1 h in two volumes of 100 mM NH4-CO3 and for 1 h in two volumes of 100 mM NH4-CO3-35% acetonitrile at 37°C. The supernatants were pooled, and the solvent was removed by lyophilization. Peptides were solubilized in 5 μl of matrix solution (2 mg/ml α-cyano-4-hydroxycinnamic acid [CHCA], 50% [vol/vol] acetonitrile, 0.1% [vol/vol] trifluoroacetic acid [TFA]) and manually spotted on matrix-assisted laser desorption ionization (MALDI) target plates.
Peptide mass fingerprinting (PMF) and tandem MS (MS/MS) analyses were performed with a 4800 Proteomics Analyzer MALDI-TOF/TOF (two-stage time of flight) instrument (ABI) operated in positive-ion reflector mode. The data were evaluated by searching the NCBI nonredundant (nr) protein sequence database using the Mascot module implemented in the GPS Explorer, version 3.5, software (ABI).
Cells from 0.5 ml of exponentially growing yeast culture were sedimented by centrifugation at room temperature. Cells were washed with 100 μl of SCD medium (for strains grown in YPD medium) or 100 μl of synthetic complete galactose (SCG) medium (for cells grown in yeast-peptone-galactose [YPG] medium). Cells were suspended in 20 μl of SCD or SCG medium, respectively, and immediately analyzed by fluorescence microscopy. Images were captured with an AxioCam MR charge-coupled device (CCD) camera mounted on an Axiovert 200 M microscope and processed with Axiovision, version 22.214.171.124.
Yeast RNA Pol I transcribes the 35S precursor rRNA (pre-rRNA), which is rapidly processed to the prominent 20S and 27S pre-rRNAs and subsequently to the mature 18S and 25S rRNAs, respectively (Fig. (Fig.1A).1A). We investigated incorporation of [3H]uracil into the large rRNAs before and after rapamycin (RAP) treatment in a time course experiment (Fig. (Fig.1B).1B). As observed in pulse-labeling experiments performed with [C3H3]methionine (58), synthesis of the 35S rRNA precursor was only moderately downregulated after 15 min of rapamycin treatment (Fig. (Fig.1B,1B, compare lanes 1 and 2; quantification is shown in D). Importantly, cellular [3H]uracil uptake was also only decreased to about 80% of the original value after 15 min of rapamycin treatment (Fig. (Fig.1C;1C; see also Fig. 3B). After 60 min of incubation in the presence of the drug, [3H]uracil uptake was reduced to about 20% of the uptake observed in cells cultured in the absence of rapamycin (Fig. (Fig.1C).1C). Similarly, [3H]uracil incorporation in 35S pre-rRNA was about 20% of [3H]uracil incorporation in 35S pre-rRNA before rapamycin treatment (Fig. (Fig.1B,1B, lane 4). This indicates that the drop in 35S pre-rRNA synthesis at this time point could be due to reduced uptake of [3H]uracil rather than to inhibition of transcription.
In contrast to 35S pre-rRNA synthesis, the amount of labeled 20S and 27S pre-rRNA dropped drastically after 15 min of rapamycin treatment, demonstrating that pre-rRNA maturation is strongly impaired (Fig. (Fig.1B,1B, compare lane 1 with lane 2; quantification is shown in D). This maturation defect can also be monitored as an increase in the ratio of 35S pre-rRNA to 27S pre-rRNA before and after different times of rapamycin treatment (Fig. (Fig.1E).1E). Again, this largely agreed with results of earlier experiments measuring synthesis of methylated pre-rRNA in yeast cells treated with rapamycin (58).
Taken together, these observations indicate that TOR inactivation has a strong and immediate effect on rRNA maturation, predominantly affecting production of the 20S and 27S rRNA precursors and the mature 18S and 25S rRNAs. 35S pre-rRNA levels, instead, are comparable to those observed under normal conditions after short times of TOR inactivation. However, a straightforward interpretation of the pulse-labeling data as a measure for RNA Pol I transcription is difficult since labeled 35S pre-rRNA levels are the result of a dynamic equilibrium between pre-rRNA neosynthesis, pre-rRNA processing, and pre-rRNA degradation.
Previous experiments showed that association of components of the RNA Pol I initiation complex, upstream activating factor (11), and core factor (56) with the 35S rRNA gene promoter were not affected after prolonged TOR inactivation. Under the same conditions, association of Rrn3 and RNA Pol I with 35S rRNA genes was reduced to about 30% to 40% of the original levels (11, 56).
Here, we wanted to investigate RNA Pol I association with the 35S rRNA gene shortly after TOR inactivation. To this end, chromatin immunoprecipitation (ChIP) experiments were carried out before and after various time points of rapamycin addition using a yeast strain expressing Rrn3 with a tandem affinity purification (TAP) tag (Rrn3-TAP) and Rpa43 with a hemagglutinin (HA) epitope tag (Rpa43-HA). We examined RNA Pol I cross-linking to either the 35S rRNA gene promoter or two regions coding for the 18S rRNA or the 25S rRNA. The 5S rRNA coding region was also included as a control (the cartoon at the bottom of Fig. Fig.1G1G shows the location of the regions amplified by quantitative PCR). In untreated cells the level of 35S rRNA gene promoter fragments and fragments of the RNA Pol I transcribed region coprecipitating with RNA Pol I remained the same over the entire time course (Fig. (Fig.1F,1F, graph). Incubation with rapamycin for 30 and 60 min decreased the amount of these rRNA gene fragments coprecipitating with the RNA Pol I subunit Rpa43 eventually to less than half of the level before drug application (Fig. (Fig.1G,1G, graph), consistent with previous studies (11).
Since cellular Rrn3 levels are affected upon TOR inactivation (56), we also examined the amount of Rrn3-TAP in extracts from cells used for the ChIP experiment by Western blot analysis. The endogenous Rrn3 level remained constant at all time points when cells were grown in YPD medium (Fig. (Fig.1F,1F, right; the Rrn3/Rpa43 ratio appears below the panel). Instead, Rrn3 amounts decreased after 30 and 60 min of rapamycin treatment (Fig. (Fig.1G,1G, right; the Rrn3/Rpa43 ratio appears below the panel) (56) in good correlation with impaired RNA Pol I association with the rRNA gene loci. However, rapamycin treatment for 15 min neither led to a significant reduction of Rrn3 levels nor to a decreased coprecipitation of 35S rRNA gene fragments with Rpa43-HA (Fig. (Fig.1G).1G). We even observed a slight but reproducible increase in 35S rRNA gene fragments coprecipitating with Rpa43-HA under these conditions. Thus, according to ChIP experiments, interaction of RNA Pol I with the 35S rDNA was not affected at time points of rapamycin treatment, where we already detected a drastic reduction in ribosomal subunit production.
We next analyzed the integrity of the RNA Pol I transcription machinery at the 35S rRNA genes after 15 min of rapamycin treatment by an independent approach. We performed chromatin endogenous cleavage (ChEC) experiments (63), which we have already successfully applied to analyze chromatin structure at the rRNA gene locus (52). Using this method, we can precisely map association of proteins with DNA within large genomic regions. We used yeast strains expressing the RNA Pol I subunit Rpa43 and Hmo1 from their genomic loci as fusion proteins with a C-terminal micrococcal nuclease (MNase). MNase fusion proteins are first cross-linked to the genomic DNA by formaldehyde treatment of yeast cells. After isolation of crude nuclei, MNase is activated upon incubation in calcium-containing buffer. Specific cleavage events mediated by the MNase fusion proteins can then be detected by Southern blotting and indirect end-labeling analysis.
Hmo1 was shown to be associated with actively transcribed rRNA genes (52) and to dissociate from the 35S rRNA gene as a consequence of a 60-min rapamycin treatment (4). We compared the association of RNA Pol I- and Hmo1-MNase fusion proteins with 35S rRNA genes before and 15 min after TOR inactivation. As described previously (52), we observed a characteristic cleavage pattern of the Rpa43-MNase fusion protein, with strong cleavage events at the 35S rRNA gene promoter and throughout the RNA Pol I-transcribed sequence (Fig. (Fig.2A,2A, lanes 3 to 5 and 8 to 10). Hmo1-MNase produced a very similar cleavage pattern (Fig. (Fig.2A,2A, lanes 13 to 15 and 18 to 20), consistent with earlier results (22, 52). Consistent with the result of Rpa43-HA ChIP experiments, we did not observe any qualitative or quantitative changes in cleavage by the MNase fusion proteins upon rapamycin treatment for 15 min (Fig. (Fig.2A,2A, compare lanes 3 to 5 with lanes 8 to 10 and lanes 13 to 15 with lanes 18 to 20).
A fraction of ChEC samples was analyzed by psoralen cross-linking. Psoralen cross-linking makes it possible to distinguish between transcribed and nontranscribed 35S rRNA genes (12, 13). In combination with ChEC, it also reveals the preferential association of MNase fusion proteins with either the actively transcribed or nucleosomal rRNA genes (52). After psoralen cross-linking and EcoRI digestion, restriction fragments of the 35S rRNA coding DNA split in two populations migrating with different mobilities in the gel: (i) a heavily cross-linked slow-migrating band (s-band), which corresponds to actively transcribed (open) rRNA genes, and (ii) a poorly cross-linked fast-migrating band (f-band), representing the nucleosomal, nontranscribed (closed) fraction of rRNA genes (12, 13, 52). The ratio of s-band to f-band did not significantly change upon short-term rapamycin action (Fig. (Fig.2B,2B, compare lane 1 with lane 3 and lane 5 with lane 7). This is consistent with results from electron microscopic inspection of chromatin spreads, demonstrating that the number of actively transcribed rRNA genes is not affected upon TOR inactivation (11). Furthermore, the s-band fragment is completely degraded by both Rpa43-MNase and Hmo1-MNase in all samples which have been subjected to ChEC prior to psoralen cross-linking (Fig. (Fig.2B,2B, compare lane 2 with lane 4 and lane 6 with lane 8). This demonstrates that after short-term inactivation of TOR signaling, RNA Pol I and Hmo1 remain associated with the open rRNA gene population.
ChEC was also employed to investigate the possibility that RNA Pol I is transcriptionally inactive and stalled at 35S rRNA genes after rapamycin treatment. Specifically, we asked whether RNA Pol I can still leave the transcribed region after TOR inactivation and inhibition of RNA Pol I (re)initiation. Therefore, we used a strain carrying a heat-sensitive allele of RRN3 (rrn3-8) (7) to block RNA Pol I initiation by temperature shift. The outline of the experiment is illustrated in Fig. Fig.2C.2C. Two isogenic yeast strains, both expressing Rpa43-MNase and carrying either the temperature-sensitive rrn3-8 allele or the RRN3 wild-type allele were grown at 24°C before rapamycin was added to one half of each culture. Cells were incubated for a further 15 min before a sample was withdrawn and subjected to formaldehyde cross-linking. The remainder of the culture was shifted to 37°C (restrictive for rrn3-8) for another 90 min before formaldehyde treatment.
As observed before (Fig. (Fig.2A),2A), Rpa43-MNase-mediated rDNA degradation was neither qualitatively nor quantitatively affected after 15 min of rapamycin treatment in cells expressing either wild-type RRN3 or mutant rrn3-8 (Fig. (Fig.2D,2D, lanes 1 to 10 and 21 to 30). After a temperature shift of RRN3 wild-type cells to 37°C for 90 min, we observed that the Rpa43-MNase-mediated cleavage events at the 35S rRNA gene promoter were reduced in chromatin from rapamycin-treated cells compared to Rpa43-MNase cleavage in chromatin from untreated cells (Fig. (Fig.2D,2D, compare lanes 11 to 15 with lanes 16 to 20). We interpret these data to show a reduced transcription initiation rate after prolonged TOR inactivation. This is in good correlation with a decrease in endogenous Rrn3 levels (Fig. (Fig.1G)1G) and initiation-competent RNA Pol I-Rrn3 complexes under these conditions (56). This is also in agreement with ChIP results, where prolonged rapamycin treatment leads to a decreased Rpa43 association with the 35S rRNA gene promoter and the RNA Pol I-transcribed sequence (Fig. (Fig.1G)1G) (11). Importantly, even after 90 min of rapamycin treatment, substantial amounts of RNA Pol I are still associated with the 35S rRNA gene compared to RNA Pol I association in the absence of rapamycin (Fig. (Fig.2D2D left panel; compare degradation of the uncut, full-length XcmI-fragment marked by an arrow in lanes 13 to 15 and lanes 18 to 20).
Temperature shift of the strain carrying the rrn3-8 allele led to almost complete loss of Rpa43-MNase-mediated rDNA cleavage in cells grown either in the absence or the presence of rapamycin (Fig. (Fig.2D,2D, right panel, lanes 31 to 40; note the lack of degradation of the uncut, full-length XcmI fragment, marked by an arrow). This demonstrates that TOR inactivation did not immobilize RNA Pol I on the rDNA template. We performed a similar experiment inactivating TOR by starving the same strains for histidine and leucine for which they are auxotrophs. After 2 h of starvation, we inactivated Pol I transcription initiation by shifting the cells to the restrictive temperature. The results were virtually identical to those obtained upon rapamycin treatment (data not shown). Thus, we conclude that RNA Pol I is not stalled on the rDNA template upon TOR inactivation.
The above results suggested that downregulation of 35S rDNA transcription cannot account for the drastic shutdown of ribosome production after short times of rapamycin treatment. Besides rRNA gene transcription, TOR inactivation very quickly affects both translation initiation and pre-rRNA maturation (2, 58). Interestingly, strong pre-rRNA processing defects have also been observed after inhibition of translation by cycloheximide (14, 47, 70, 73, 83). Thus, it is possible that impaired translation after short-term TOR inactivation (2) accounts for the ribosomal maturation defects observed in this situation.
We therefore directly compared effects of short-term rapamycin and cycloheximide treatment on the production of rRNAs in yeast. We first examined rRNA production in a pulse-chase experiment before and after the cells were cultured in the presence of either rapamycin or cycloheximide (CHX) for 15 min. Cells were pulsed for 5 min with [3H]uracil and chased with an excess of unlabeled uracil for 4, 8, and 16 min before cellular RNAs were isolated and analyzed. As for rapamycin treatment, we first determined if cycloheximide affects [3H]uracil uptake and found that uptake was also slightly reduced, as observed after rapamycin treatment (Fig. (Fig.3B).3B). Both rapamycin and cycloheximide treatment led to a strong maturation defect, resulting in the relative accumulation of radiolabeled 35S rRNA precursor over the remainder of the other rRNAs (Fig. (Fig.3A,3A, compare lanes 1 to 4 with lanes 5 to 8 and lanes 9 to 12). Whereas residual pre-rRNA processing still occurred in cells cultured for 15 min in the presence of rapamycin (Fig. (Fig.3A,3A, lanes 5 to 8), processing was nearly abolished after 15 min of cycloheximide treatment (Fig. (Fig.3A,3A, lanes 9 to 12). The stringency of the maturation defect caused by rapamycin or cycloheximide treatment can be deduced from the different ratios of 35S rRNA to 27S rRNA (Fig. (Fig.3C).3C). We also measured incorporation of [3H]uracil into the mature 25S rRNA. After a 16-min chase, only 1/10 of the label detected in the untreated strain was found in the strain cultured in the presence of rapamycin (Fig. (Fig.3D).3D). Tritium labeling of 25S rRNA in cells treated with cycloheximide was at background levels (Fig. (Fig.3D).3D). We note that, in agreement with earlier observations (14, 47, 70, 73, 83), inhibition of mRNA translation by cycloheximide provokes strong phenotypes in pre-rRNA processing and nascent ribosomal subunit accumulation, similar to effects observed after short-term inactivation of the TOR pathway by rapamycin.
Since the effect of cycloheximide treatment on RNA Pol I transcription is controversial (10, 23, 36, 67, 69, 70), we investigated RNA Pol I association with the rRNA gene in ChIP experiments. The results obtained in a time course experiment after treatment of cells with cycloheximide were of striking similarity to those observed with rapamycin-treated cells (compare Fig. Fig.3E3E and and1G).1G). Rpa43-HA cross-linking to the rRNA gene promoter region was only marginally affected after 15 min of cycloheximide addition but decreased to about 50% of the value obtained in the untreated control after 60 min (Fig. (Fig.3E,3E, graph). The data for Rpa43 cross-linking to the RNA Pol I-transcribed region underlined that a significant proportion of RNA Pol I remains associated with the rDNA. As with the rapamycin treatment, we observed an increase in the fragments of the 35S rRNA coding region coprecipitating with Rpa43-HA (Fig. (Fig.1G1G and and3E,3E, graphs). We also measured the Rrn3-TAP level in the extracts used for ChIP by Western blot analysis. As observed before (Fig. (Fig.1G,1G, graph), there was only a slight decrease in endogenous Rrn3 levels 15 min after addition of cycloheximide (Fig. (Fig.3E,3E, blot). At later time points Rrn3 levels dropped to values approximating the Rrn3 reduction seen upon rapamycin action (compare blots in Fig. Fig.1G1G and and3E;3E; Rrn3/Rpa43 ratios are indicated below the panels). Thus, neither short-term rapamycin treatment nor inhibition of translation by cycloheximide significantly influences association of RNA Pol I with the rRNA gene sequences while strongly affecting ribosomal subunit production.
The sudden decrease of pre-rRNA production after inhibition of protein synthesis by cycloheximide strongly argues for “an early shortage of some proteins characterized by small pool sizes and rapid turnover” affecting ribosome biogenesis (70). As discussed in the study of Stoyanova and Hadjiolov, r-proteins are likely candidates because the pool size of non-ribosome-bound, “free” ribosomal proteins is limited due to their fast turnover rate and rapid assembly into preribosomal particles (25, 40, 81, 85). Strong ribosome biogenesis defects are observed after conditional shutdown of individual r-proteins (19, 20, 53, 57, 59). Only moderate changes in r-protein levels in the context of r-protein gene haploinsufficiency in human disease (44, 45) or in yeast mutants (15, 17, 46, 55, 68, 87) provoke pre-rRNA processing and growth phenotypes.
We therefore compared neosynthesis of r-proteins upon short-term rapamycin treatment and treatment with different concentrations of cycloheximide. To this end, two different wild-type yeast strains were cultured in either the absence or presence of rapamycin or cycloheximide for 15 min prior to pulse-labeling of proteins with [35S]methionine-cysteine (Fig. (Fig.4A).4A). After extraction, 35S incorporation into total protein was determined. Polypeptides in the extracts were separated in 16% urea gels and analyzed by Coomassie staining and autoradiography. As previously reported (2), total protein production decreased to around 50% to 60% of the value in the untreated sample after 15 min of rapamycin treatment (Fig. (Fig.4B).4B). Downregulation of protein synthesis was virtually identical when cells were treated with 0.1 μg/ml cycloheximide, the lowest concentration used in the experiment (Fig. (Fig.4B).4B). However, when we analyzed 35S incorporation into specific polypeptides after urea gel electrophoresis and autoradiography, we found that neosynthesis of a group of small proteins was specifically affected upon the addition of rapamycin but not in the presence of 0.1 μg/ml cycloheximide (Fig. (Fig.4C,4C, compare lanes 8 and 9 in top panels; the region of interest is marked by a black bar on the left side of the autoradiogram, enlarged in the bottom panels).
Most of these polypeptides migrated with the same velocity as r-proteins derived from an affinity-purified 80S ribosome (Fig. (Fig.4C,4C, bottom panels, compare lane 1 with lanes 2 and 8). We focused on the apparent level of neosynthesized proteins in two prominent bands (Fig. (Fig.4C,4C, bottom panels, asterisks), for which we further unambiguously identified r-proteins as major components by mass spectrometry (Table (Table2).2). We found that production of these proteins was only moderately affected in the presence of 0.1 μg/ml cycloheximide, whereas they were no longer detectable in the rapamycin-treated sample (Fig. (Fig.4C,4C, lanes 8 to 10 in the bottom panels). 35S labeling of these proteins was similar, however, in cells treated with concentrations of 1 to 10 μg/ml cycloheximide (Fig. (Fig.4C,4C, bottom panels; compare lanes 9, 11, and 12). Thus, upon rapamycin treatment the production of r-proteins is specifically inhibited, which might be caused by the combination of downregulation of general translation (2) and the strong reduction in r-protein mRNA levels (58, 82).
To directly correlate the reduction in r-protein production with the reduction in ribosome production, we performed pulse-chase experiments with [3H]uracil in the same strains used for the protein analysis. As expected, production of mature rRNAs was severely impaired upon rapamycin treatment (Fig. (Fig.4D,4D, compare lanes 2 and 4; quantification is shown in E). In the presence of 0.1 μg/ml cycloheximide, a concentration where total protein production (but not r-protein production) was reduced to the level observed after rapamycin treatment (Fig. (Fig.4B),4B), robust 25S rRNA synthesis could still be observed (Fig. (Fig.4D,4D, lanes 2 and 6, and E). In contrast, the defect in 25S rRNA production was virtually identical in rapamycin-treated cells and cells incubated at concentrations of 1 to 10 μg/ml cycloheximide (Fig. (Fig.4D,4D, lanes 4, 8 and 10, and E), in which r-protein production was similarly affected (Fig. (Fig.4C,4C, lanes 9, 11, and 12 of the bottom panels). Thus, there is a very good correlation between r-protein production and rRNA synthesis.
We further investigated whether cycloheximide treatment can mimic other cellular phenotypes resulting from TOR inactivation. It has been suggested that TOR controls late stages of ribosome maturation in yeast, including intranuclear transport of preribosomes (32), as well as their nucleo-cytoplasmic translocation (76). In particular, it was found that the nuclear GTP-binding protein Nog1 is tethered to the nucleolus upon nutrient starvation or rapamycin treatment (32). Similarly, the KH-domain protein Dim2p and the HEAT repeat/Armadillo domain and export factor Rrp12 (54) are nucleolar restricted upon rapamycin treatment (76).
We compared the effect of cycloheximide and rapamycin treatment on cellular localization of Rrp12 and Nog1. We studied two isogenic strains expressing either Rrp12 or Nog1 as a fusion protein with a C-terminal green fluorescent protein (GFP) from their respective genomic loci. Exponentially growing cells were split in three and further incubated in the absence or presence of either rapamycin or cycloheximide for 15 and 30 min before live-cell imaging. As described earlier (32, 76), we observed nucleolar entrapment of both Rrp12-GFP and Nog1-GFP after 30 min of rapamycin treatment (Fig. (Fig.5A,5A, compare panel YPD with RAP in columns 4 and 8). The change in subcellular localization was more obvious for Rrp12-GFP since it is distributed all over the cell under normal conditions (Fig. (Fig.5A,5A, compare panels YPD and RAP in column 4), whereas Nog1-GFP is already nuclear in cells cultured in YPD (Fig. (Fig.5A,5A, column 8, compare panels YPD and RAP).
Importantly, we also observed nucleolar entrapment of both GFP fusion proteins upon treatment with cycloheximide (Fig. (Fig.5A,5A, compare panels YPD with CHX in columns 4 and 8). The kinetics of nucleolar concentration of the ribosome biogenesis factors in cells treated with cycloheximide appeared to be even faster than those seen in cells treated with rapamycin (Fig. (Fig.5A,5A, compare panels RAP with CHX in columns 2 and 6). We conclude that inhibition of general mRNA translation is sufficient to provoke nucleolar entrapment of ribosome biogenesis factors, as observed after short-term inactivation of the TOR pathway by rapamycin.
We speculated that limited r-protein production might be sufficient to explain nucleolar entrapment of preribosomal complexes observed after inhibition of translation and TOR signaling. We employed yeast strains with deletions of the chromosomal copies of RPS5 or RPL25 and conditionally expressing the respective wild-type allele under the control of the glucose repressible GAL1 promoter (19, 57). GAL1 promoter-dependent expression of RPS5 and RPL25 in these cells supports growth in galactose-containing medium. After transfer to glucose-containing medium—and thus repression of RPS5 or RPL25 expression—specific steps in nuclear maturation of either small (RPS5)- or large (RPL25)-ribosomal subunit precursors are strongly and selectively affected (19, 20, 57, 74, 75). We genetically modified these strains for constitutive expression of the Rrp12-GFP or the Nog1-GFP fusion protein from the respective genomic location. As a control, we included the corresponding wild-type strains expressing RPS5 and RPL25 from their endogenous loci and, in addition, the respective Rrp12-GFP or Nog1-GFP fusion protein. Exponentially growing cells in galactose-containing medium (YPG) were split into two parts. Half of the cells were transferred into glucose-containing medium (YPD), whereas the other half was again cultured in YPG medium. Incubation was continued for 90 min before live-cell imaging. In RPS5 and RPL25 wild-type strains no significant change in Rrp12-GFP or Nog1-GFP localization could be detected under the two different culture conditions (Fig. (Fig.5B,5B, compare panels YPG and YPD in columns 2 and 6). Wild-type-like localization of Rrp12-GFP or Nog1-GFP was also observed in the strains carrying RPS5 or RPL25 under the control of the conditional promoter cultured in YPG (Fig. (Fig.5B,5B, compare panel YPG in column 2 with column 4 and column 6 with column 8). In contrast, depletion of rpS5 and rpL25 in these strains cultured in YPD medium led to nucleolar entrapment of the respective GFP fusion proteins (Fig. (Fig.5B,5B, compare panels YPG with YPD in columns 4 and 8). Very similar results were obtained by depleting rpS14 in an Rrp12-GFP-expressing strain (data not shown).
Altogether, these data indicate that shortage of r-protein expression is sufficient to induce ribosomal maturation defects and nucleolar entrapment of preribosomal complexes observed upon TOR inactivation.
A limited cellular r-protein production provides a very reasonable explanation for the instantaneous shutdown of ribosome biogenesis in situations in which protein synthesis is downregulated. As structural components of the ribosome, r-proteins have to be produced continuously in (at least) stoichiometric amounts with the rRNA to guarantee proper assembly, processing, and maturation of the ribosomal subunits (80). TOR inactivation leads to a rapid downregulation of r-protein production on the transcriptional and the translational levels affecting production of a large subset of r-proteins (Fig. (Fig.4C)4C) (58, 82), most of which are essential for ribosome biogenesis (19, 53, 57, 59). Compared to the effects of depleting individual r-proteins, a drop in the free pools of all ribosomal proteins after TOR inactivation will likely result in significant synthetic effects on maturation of ribosomal subunits. This is supported by the observation that in prokaryotes both in vivo and in vitro ribosome assembly is not a fully cooperative process (16).
A comprehensive analysis of the nuclear function of Tor1 demonstrated that in certain tor1 mutants the rapamycin-mediated rRNA maturation phenotype can be uncoupled from the inhibition of r-protein mRNA synthesis (42). Whether rapamycin still affects mRNA translation (2) in these mutant strains and, thereby, r-protein gene expression on a posttranscriptional level is unknown. In this regard, it would be interesting to correlate rRNA maturation with r-protein production upon TOR inactivation in these genetic backgrounds.
More than 30 years ago Warner and Udem characterized different temperature-sensitive mutant yeast strains, which had been isolated in a genetic screen for strong ribosome biogenesis defects (28, 83). Many of the mutant alleles found in this screen were subsequently shown to code for components of the nuclear pre-mRNA splicing machinery (60). Since r-protein genes constitute almost 50% of the intron-containing genes in yeast, defects in splicing predominantly affected r-protein levels (24). Interestingly, the rRNA maturation phenotype in these mutant strains closely resembled the maturation phenotype seen after yeast treatment with cycloheximide (83). This suggests that decreased r-protein production is sufficient to explain the ribosome biogenesis phenotype observed after shutdown of general protein expression.
RNA Pol I transcription seems not to be the primary target of TOR inactivation, since we did not observe any significant qualitative or quantitative changes in the assembly of the RNA Pol I transcription machinery at the rDNA template after 15 min of rapamycin treatment (Fig. 1F and G and and2).2). Under these conditions RNA Pol I was still mobile, giving rise to nascent 35S pre-rRNA (Fig. (Fig.1B1B and and2D),2D), which does not, however, exclude the possibility that transcription elongation is partly affected (88). Thus, rRNA synthesis likely continues after TOR inactivation while r-protein production is strongly impaired (Fig. (Fig.4C).4C). This imbalance of structural components of the ribosome is then presumably adjusted by rapid degradation of misassembled, excess rRNA precursors. In support of this idea, in vivo depletion of individual r-proteins of the large or small ribosomal subunit leads frequently to a drastic relative accumulation of the corresponding other subunit both in yeast and in mammals (53, 57, 59). This strongly indicates that in eukaryotes misassembled ribosomal subunits are efficiently turned over (see reference 57 for discussion).
Chromatin spreading and electron microscopic inspection of rRNA gene transcription after 10 min of rapamycin treatment suggested that RNA Pol I density dropped to about 60% of the RNA Pol I density in untreated cells (11). Since determination of RNA Pol I density by this method critically relies on counting of both RNA Pol I molecules and nascent rRNAs, such an analysis will be compromised by rapid (most likely cotranscriptional) degradation of misassembled transcripts.
It should be noted that in another yeast strain background rapamycin treatment for only 20 min reduces the coprecipitation of 35S rRNA gene fragments with RNA Pol I to about 40% of the original level in ChIP experiments (41). However, the amounts of rapamycin used by Laferte et al. were more than twice as high as the amounts used in the present work. In the earlier study, RNA Pol I association with rRNA genes upon short-term TOR inactivation was only modestly affected in a yeast mutant expressing an Rrn3-Rpa43 fusion protein as the only source of those factors (41). Furthermore, transcription of 35S rRNA was only slightly impaired under this condition (41). Despite the fact that RNA Pol I transcription is deregulated in this strain, r-protein production measured by [35S]methionine-cysteine incorporation is also severely affected 15 min after TOR inactivation, concomitant with a strong rRNA maturation defect (data not shown).
Consistent with earlier data (11), we observe a reduction in RNA Pol I association with the rRNA gene sequence after prolonged times of drug action. This correlates well with a reduction in the level of initiation competent RNA Pol I-Rrn3 complexes, which is in part the consequence of impaired Rrn3 expression (56). mRNA levels of RRN3 as of many other genes involved in ribosome biogenesis are rapidly downregulated in response to TOR inactivation (8, 21, 27, 33-35, 78). Together with the inhibition of general mRNA translation by TOR inactivation (2), this will eventually lead to a depletion of many factors important for ribosome biogenesis, including components of the three different transcriptional machineries. Thus, in an intermediate and long-term response, yeast cells downregulate all the processes leading to mature ribosomes, a physiological situation encountered when they approach stationary phase or undergo sporulation.
We thank Eduard Hochmuth and Rainer Deutzmann for help with the mass spectrometry analysis and people of the institute for critical discussions.
This work was supported by grants of the Deutsche Forschungsgemeinschaft to P.M., H.T., and J.G.
Published ahead of print on 13 December 2010.