RNase III-dependent Regulation of Yeast Telomerase*

In bakers’ yeast, in vivo telomerase activity requires a ribonucleoprotein (RNP) complex with at least four associated proteins (Est2p, Est1p, Est3p, and Cdc13p) and one RNA species (Tlc1). The function of telomerase in maintaining chromosome ends, called telomeres, is tightly regulated and linked to the cell cycle. However, the mechanisms that regulate the expression of individual components of telomerase are poorly understood. Here we report that yeast RNase III (Rnt1p), a double-stranded RNA-specific endoribonuclease, regulates the expression of telomerase subunits and is required for maintaining normal telomere length. Deletion or inactivation of RNT1 induced the expression of Est1, Est2, Est3, and Tlc1 RNAs and increased telomerase activity, leading to elongation of telomeric repeat tracts. In silico analysis of the different RNAs coding for the telomerase subunits revealed a canonical Rnt1p cleavage site near the 3′ end of Est1 mRNA. This predicted structure was cleaved by Rnt1p and its disruption abolished cleavage in vitro. Mutation of the Rnt1p cleavage signal in vivo impaired the cell cycle-dependent degradation of Est1 mRNA without affecting its steady-state level. These results reveal a new mechanism that influences telomeres length by controlling the expression of the telomerase subunits.

The ends of eukaryotic chromosomes are capped with special structures made of tandem DNA repeats and associated proteins, called telomeres. These structures protect chromosomes from end-to-end fusion, recombination, and nucleolytic degradation (1,2). However, because the conventional DNA replication machinery cannot fully duplicate the ends of linear chromosomes, telomeric DNA will shorten with each round of replication, leading to a replicative senescence (3). To solve this end replication problem, most eukaryotes use the activity of an enzyme called telomerase to ensure the maintenance of telomeric DNA (4). Telomerase is a ribonucleoprotein reverse transcriptase that can extend the 3Ј end of chromosomes using its RNA subunit as a template for the addition of telomeric repeats.
In vitro, telomerase activity requires a reverse transcriptase (Tert, Est2p in yeast) and an RNA template (Terc, Tlc1 in yeast) (5,6). In vivo, yeast telomerase requires additional factors for function, including Est1p, Est3p, and Cdc13p (3,7). Deleting any one of the EST1, EST2, EST3, or TLC1 genes or a particular allele of the CDC13 gene (cdc13-2) causes progressive telomere shortening leading to cellular senescence. Increased expression of individual telomerase genes may also lead to telomeric phenotypes. For example, overexpression of Tlc1 causes telomere shortening by sequestering the Tlc1-binding factor Ku (8,9). Similarly, overexpression of Est1p may cause a slight telomere lengthening in some strains (10). On the other hand, induction of Est1, Est2, and Est3 mRNA expression in cells with defective nonsense-mediated mRNA decay decreases telomere length (11). These observations suggest that a variation in the expression levels of the different components of telomerase holoenzyme influences telomerase function. However, the cellular mechanisms that regulate and balance the expression of the different components of the telomerase ribonucleoprotein remain unknown.
Telomere elongation occurs in late S or early G 2 /M phases of the cell cycle (12,13). There is evidence that the catalytic core of telomerase (Est2p and Tlc1) is present at yeast telomeres throughout the cell cycle but requires an association of Est1p, which accumulates in the S phase of the cell cycle (14,15). Telomerase activity on the telomeres is negatively regulated by Pif1p, a helicase that is thought to remove telomerase from the chromosome ends (16). In addition, the Rif proteins also exert a negative role on telomere length maintenance (17), and several other proteins involved in the DNA damage checkpoint and repair pathways were shown to influence steady-state telomere length in vivo (18). At least some of the factors regulating telomere length influence the activity of the telomerase subunits, yet very little is known about the mechanisms regulating the expression of the telomerase subunits.
Here we show that the expression level of several telomerase subunits is regulated post-transcriptionally by yeast RNase III (Rnt1p). Rnt1p is a double-stranded RNA-specific endoribonuclease involved in the maturation of many RNAs, including small nuclear RNAs, snoRNAs, 3 and pre-rRNAs (19 -21).
Recently it was shown that Rnt1p regulates the expression of glucose-dependent genes, degrades unspliced mRNAs, and cleaves mRNAs with abnormal 3Ј ends (22)(23)(24)(25). The majority of Rnt1p substrates include a short RNA stem capped with an AGNN tetraloop structure (26). These particular structures are cleaved at a fixed distance from the tetraloop (27)(28)(29). Deletion of RNT1 induced the expression of many components of telom-erase, leading to an increase in telomerase activity and elongation of the telomeres. A point mutation that impairs catalytic activity of Rnt1p without affecting its expression or interaction with RNA fails to restore normal expression of telomerase subunits or reduce telomere lengths. In silico analysis revealed a canonical Rnt1p cleavage site within the coding sequence of the Est1 mRNA. In vitro analyses showed that this predicted structure was directly cleaved by Rnt1p in the absence of other factors. Disruption of this Rnt1p cleavage site by silent mutations impaired the cell cycle-dependent regulation of Est1 mRNA, suggesting that Rnt1p triggers RNA cleavage in a cell cycle-dependent manner. These results demonstrate that the expression level of the telomerase subunits influences their function and reveal a new regulatory mechanism that controls the accumulation of telomerase-related RNAs.
In Vitro RNA Cleavage-In vitro cleavage reactions were performed as described previously (29). Cleavage of model RNA substrates was performed using T7-synthesized transcripts and recombinant Rnt1p (0.2 pmol) (39) in a 20-l reaction buffer containing 150 mM KCl and incubated 20 min at 30°C. Cleav-age of total RNA extracted from yeast was performed as described previously (22). Briefly, 50 g of total RNA was incubated with 4 pmol of recombinant Rnt1p for 10 min at 30°C as described for the cleavage of model substrates.
Analysis of mRNA Stability-The stability of the different mRNAs was determined after inhibiting transcription using 10 g/ml of thiolutin as previously described (22). Cells were grown at 26°C to mid-log phase and RNA was extracted at different time points after the addition of the drug. The levels of mRNA were quantified by real-time PCR using RPR1 (pol III transcript) as an internal control. Half-lives were estimated using the single exponential decay formula y ϭ 100 e Ϫbx . Thiolutin was kindly provided by Pfizer (Groton, CT).
Primer Extension and Northern Blot Analysis-Primer extension was performed using a primer specific for EST1 (5Ј-GGG-TTGACGACGACAGGCGTGG-3Ј) as described previously (19). Extended products were separated on a 6% denaturing polyacrylamide gel and visualized by autoradiography. Northern blots were performed as described previously (19) using 15 g of total RNA and a 1% denaturing agarose gel or a 4% polyacrylamide gel. The RNA was visualized by autoradiography using randomly labeled probes corresponding to specific genes. The RNA was quantified using Instant Imager (PerkinElmer Life Sciences).
Real-time PCR Analysis-Synthesis of cDNA was performed using 2 g of total RNA and SuperScript II Reverse Transcriptase (Invitrogen). Gene-specific primer protocol using a "Flap sequence" common to all primers was used to avoid DNA amplification during subsequent PCR amplification. The cDNA was purified with Wizard SV Gel and PCR Clean-Up System (Promega). Q-PCR was performed on an ABI PRISM 7700 (Applied Biosystems) as previously described (40). PCR were done in triplicate using a 25-l reaction volume. TaqMan probes (5Ј,6-FAM, 3Ј-TAMRA) and specific primers (Integrated DNA Technologies, Coralville, IA) were used at concentrations of 250 and 900 nM, respectively, in a reaction buffer containing 50 mM KCl, 5.5 mM MgCl 2 , 10 mM Tris, pH 8.0, with 0.625 units of homemade Taq (41), 0.25 unit of UDG (New England Biolabs, Pickering, ON), 50 nM ROX Reference Dye (Invitrogen), 200 M dATP/dGTP/dCTP, 400 M dUTP (Amersham Biosciences). Data were analyzed by the comparative threshold method. All primers and probes used for realtime PCR are listed in Table 1.
Telomere Length Analysis-Telomere length was analyzed as described earlier (42). Genomic DNA was extracted from cells grown to mid-log phase. 5 g of genomic DNA was digested with XhoI, separated on a 1% agarose gel, and transferred to a nylon membrane (Hybond Nϩ, Amersham Biosciences). DNA was visualized by autoradiography with a randomly labeled telomeric probe.
Cell Cycle Synchronization and FACS Analysis-Cells were synchronized in G 1 by the addition of ␣-factor (Bioshop, Burlington, ON) at a final concentration of 0.5 g/ml for 3 h. Cells were harvested and resuspended in fresh media containing 200 g/ml Pronase (Roche Diagnostics) to release the cells from their arrest. Samples were taken at different time points to either extract RNA or monitor the release into the cell cycle by FACS analysis as described previously (34). The RNAs corre-sponding to the different genes was quantified by real-time PCR using Rpr1 as an internal control.
Western Blot and Telomerase Activity-Telomerase assays were performed as previously described (35). The EST2 gene was modified to incorporate an N-terminal Protein A tag as described (35). Immunoprecipitation was performed using IgG-Sepharose beads (Amersham Biosciences) and 2 mg of total protein extracted from either RNT1 (NLYH15) or rnt1⌬ (NLYH12) cells. Extension of a telomeric primer (5Ј-TAGGG-TAGTAGTAGGG-3Ј) was monitored to determine telomerase activity. Quantifications were performed using a Storm Phos-phorImager (Storm 860, Amersham Biosciences). Total protein (30 g) or immunoprecipitated proteins were run on an 8% SDS-PAGE and transferred onto Hybond-C membrane (Amersham Biosciences). Western blots were performed using 1:7500 diluted (total protein) or 1:5000 diluted (immunoprecipitated protein) polyclonal rabbit anti-protein A antibodies (Sigma). Donkey anti-rabbit IgG conjugated with horseradish peroxidase was used as a secondary antibody to allow detection with ECL Plus reagents (Amersham Biosciences). After membrane stripping, blots were incubated with mouse anti-PGK Ig diluted 1:500 (Molecular Probes) as primary and sheep anti-mouse IgG conjugated with horseradish peroxidase diluted 1:5000 (Amersham Biosciences) as secondary antibody. Detection and quantification were done using a Storm Fluorescence Imager (Storm 860, Amersham Biosciences).
Non-linear Regression Analysis and Graph Generation-The graphs shown in Fig. 5C were generated using GraphPad Prism version 4.03 for Windows (GraphPad Software, Inc., San Diego, CA). The data from three independent experiments for EST1-4; bar1⌬ cells and two independent experiments for EST1; bar1⌬ were used for the analysis. To choose the best equation to model the data, we compared the different classical equations and all polynomial order equations available in GraphPad. To select the best equation type, we used an F test with p Ͻ 0.05. Twoby-two comparisons were performed to identify the simplest model that best fit the data. The equation used is a polynomial The obtained values are shown in supplemental Table 1. The use of non-linear regression was required because the mRNA level is affected by many variables: mRNA decay, Rnt1p cleavage, constitutive expression, and cell cycle-dependent expression. The use of a non-linear regression also allows for the two main sources of variation in the measurement (normal experimental variation and variation in the cell synchronization) to be determined.

RESULTS
We previously performed a microarray analysis of RNA extracted from cells lacking RNT1 and compared it with RNA from wild type cells (22). After detailed inspection of the data, we noticed that many telomerase-related mRNAs were induced upon the deletion of RNT1 (Fig. 1A). This induced expression of RNAs coding for telomerase subunits was similar to that of known Rnt1p substrates like Rps22B (23) and Mig2 (22) mRNAs. The impact of RNT1 deletion on the expression of telomerase-related RNAs was verified by Northern blot analysis. Consistent with the microarray data, the mRNAs coding for the main components of the telomerase RNP complex (Est1, Est2, and Est3) and the telomerase RNA Tlc1 accumulated in the absence of Rnt1p (Fig. 1B). No major increase in the RNA amounts coding for other telomere-related proteins like Cdc13, Ten1, and Tel1 were detected by Northern blot (data not shown). We conclude that Rnt1p is required for normal expression of most core telomerase subunits.
If Rnt1p influences the stability of Est1, Est2, Est3, and Tlc1 RNAs, we expected the deletion of RNT1 to slow the turnover and increase the half-life of these RNAs. To examine this possibility, the half-life of each affected RNA was determined in wild type and rnt1⌬ cells after inhibiting new transcription using thiolutin. Quantitative PCR analysis was used to estimate the mRNA amounts at different time points after transcription inhibition and the half-life was calculated for each RNA species (Table 2). We used two unrelated RNAs (Spt15 and Act1) with different half-lives and that are thought not to be directly affected by Rnt1p as controls (supplemental Fig. 1). Indeed, the half-lives of these control RNAs were only slightly longer in rnt1⌬ cells when compared with that in wild type cells due to a general slow metabolism of rnt1⌬ cells (Table 2 and data not shown). In contrast, these experiments showed a significant increase in the half-lives of Est1, Est2, and Est3 mRNAs ( Table  2). As previously described, the Tlc1 RNA was already very stable in wild type cells (43), and no decay was observed in both wild type and rnt1⌬ cells even after 1 h of transcription inhibition ( Table 2). When normalized to Spt15, the relative half-lives of Est1 and Est2 were about 2 times longer in rnt1⌬ than in wild type cells, whereas that of Est3 was only slightly longer in rnt1⌬ cells (Fig. 1C). If the increase in the half-life of the Est2 mRNA of about 2-fold was significant, it should lead to a corresponding increase in the amount of Est2p. Indeed, deleting RNT1 in cells expressing a protein A-tagged version of the telomerase catalytic subunit Est2p increased the expression of both mRNA and protein expression by 2 to 3 times as compared with that detected in wild type cells (supplemental Fig. 2 and Fig. 1D).
We conclude that deletion of RNT1 slows the decay rate of several telomerase-related mRNAs leading to an increase in the expression of these telomerase subunits.
To evaluate the impact of Rnt1p on telomerase function, we examined telomere length in wild type and rnt1⌬ cells by Southern blotting. DNA was extracted from relevant strains and digested with the restriction enzyme XhoI. Due to a conserved XhoI site in the telomere-proximal YЈ-elements, many terminal restriction fragments appear as somewhat smeary bands of about 1.2 kb size on such blots (44). In the absence of Rnt1p, the size of the terminal restriction fragments increased as compared with that of wild type cells, indicating that the deletion of RNT1 increases telomere length ( Fig. 2A). This increase in telomere length is directly linked to the expression of Rnt1p, because transformation of rnt1⌬ cells with a plasmid harboring RNT1 restores normal telomere length ( Fig. 2A).
We next examined whether this increase in telomere length is associated with an increase in telomerase activity. Relative telomerase activity was assessed in extracts derived from wild type and rnt1⌬ cells expressing a protein A-tagged version of The data were extracted from a previously performed microarray analysis using RNA extracted from wild type or rnt1⌬ cells (22). Mig2 and Rps22B mRNAs are shown as positive controls, whereas Act1 mRNA is shown as a negative control. B, Northern blot analysis of RNAs coding for the telomerase RNP complex. Total RNA was extracted from wild type or rnt1⌬ cells and visualized using gene specific randomly labeled probes. The RNA expression level was quantified using Instant Imager and the relative mRNA amount (RMA) was calculated using Act1 mRNA as reference. The RMA indicated below each gel is an average of three experiments with a S.D. Ϯ 0.2 or less. C, graphical representation of the relative increases in the half-lives of the telomerase mRNAs upon the deletion of RNT1. The relative increase in the half-lives of the telomerase mRNAs was normalized to the half-life of Spt15 mRNA, which serves as a control for possible indirect effects caused by the slow metabolism of rnt1⌬ cells. The different half-life values used to calculate the relative increase are indicated in Table 2. The data were obtained from three independent transcription inhibition experiments. D, Western blot analysis of Est2p expression. Total proteins were extracted from wild type or rnt1⌬ cells expressing proA-EST2. Immunoprecipitation was performed using IgG-Sepharose beads. Proteins from total extract (upper panel) or from immunoprecipitation (lower panel) were separated using SDS-PAGE and Est2p was visualized using antibodies against the protein A. The expression level of Est2p was normalized to the IgG heavy chains or to Pgk1p. The relative protein amount (RPA) is an average of three experiments with a S.D. Ϯ 0.6.

Rnt1p as a Negative Regulator of Yeast Telomerase
Est2p. In this procedure, telomerase was partially purified using IgG-Sepharose beads and used for a direct extension of a telomeric primer (44). As shown in Fig. 2B, deletion of RNT1 resulted in an increase in relative telomerase activity as compared with the internal control (Fig. 2B). The observed activity is telomerase specific, because it is RNA dependent and was not observed in precipitates from untagged cells. The observed increase in the telomerase activity is Rnt1p-dependent but strain background-independent and was also observed after the inactivation of a temperature-sensitive allele of RNT1 (data not shown). Thus, the accumulation of the RNAs of telomerase core subunits upon the deletion of Rnt1p leads to an increase in both its in vitro activity and telomere length. Deletion of Rnt1p affects many cellular pathways and slows cell metabolism (34). To ensure that the increased levels of RNA of telomerase subunits is not a secondary effect generated by long term changes in cell metabolism, we monitored the kinetics of telomerase induction upon inactivating a temperature-sensitive allele of Rnt1p (rnt1-ts) (22). If the RNAs coding for the telomerase subunits are targets of Rnt1p-mediated RNA metabolism, their steady-state levels should change with the same kinetics at that of known Rnt1p substrates (e.g. snoRNAs). Total RNA was extracted from rnt1-ts cells at different intervals after shift to the restrictive temperature (37°C). The extracted RNA was analyzed using Northern blots or real-time PCR. As expected, Northern blots probed with a snR43-specific probe, a known substrate of Rnt1p, documented a processing defect as early as 2 h after shift to 37°C (Fig. 3A). Real-time PCR analysis of the same RNA samples indicated that similar to snR43, the levels for RNAs coding for the telomerase subunits increased 2 h after shift to the restrictive temperature (Fig. 3B). Est2 mRNA level increased very rapidly after the inactivation of Rnt1p, whereas the Tlc1 RNA accumulated somewhat more slowly. The steady-state amount of Act1 mRNA increases about 2-fold upon the deletion of Rnt1p, but this increase is thought to be due to indirect effects on overall cell metabolism and a reduction in the amount of mature rRNA in such cells. Strikingly, the kinetics of mRNA accumulation for Act1 upon Rnt1p inactivation was very different from that observed for the telomerase-associated RNAs. Even 8 h after the shift to the restrictive temperature, there was no detectable increase in Act1 mRNA (22). Given the similarities in the kinetics of the increases of telomerase subunit RNAs with those of the accumulation of non-processed bona fide Rnt1p substrates, the data FIGURE 2. Rnt1p is required for normal telomerase activity. A, Southern blot analysis of telomere length. Genomic DNA was extracted from wild type or rnt1⌬ cells and digested with XhoI restriction enzyme to release the terminal restriction fragment (TRF). The DNA was separated on a 1% agarose gel, transferred to a nylon membrane, and visualized with a randomly labeled telomeric probe. Note that this experiment was repeated several times using different clones and strain backgrounds and the increase in the telomere length was reproducible for each clone and not due to clonal variation. B, in vitro assay of the telomerase activity. Telomerase was partially purified by immunoprecipitating proA-EST2 using IgG affinity chromatography. RNA-dependent extension of a telomeric primer was monitored using denaturing PAGE. The increase in the relative telomerase activity (RTA) was calculated using the activity of the wild type telomerase as a reference. The data were obtained from three independent experiments using three independent cell extracts with a S.D. Ϯ 0.8.

FIGURE 3. Rnt1p catalytic activity is required for the normal expression of the telomerase subunits. A, Northern blot analysis of H/ACA snoRNA in cells expressing
Rnt1p temperature-sensitive allele (rnt1-I338T). RNA was extracted from rnt1-I338T cells at different intervals after shift to 37°C. A radiolabeled probe against snR43 was used to detect both mature and extended species as indicated on the right. B, graphical representation of the relative increase in mRNA levels of different telomerase components in cells expressing Rnt1p TS allele (rnt1-I338T). RNA was extracted as described in A and quantified using real-time PCR. The levels of the different mRNAs are presented relative to that of Act1. The indicated values represent an average of three independent experiments. The graph curves were obtained using Microsoft Excel. C, comparison between the expression levels of the telomerase-related RNAs in wild type cells, rnt1⌬ cells, and cells expressing a catalytically impaired version of Rnt1p (rnt1-D247R). The RNA was analyzed by Northern blot gene-specific probes. Expression levels were quantified using Instant Imager and normalized to Act1 mRNA as a loading control. The indicated values represent an average of three independent experiments with a S.D. Ϯ 0.4 or less. FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4377
suggest that Est1, Est2, Est3, and Tlc1 RNAs are targets of Rnt1p-controlled RNA metabolism.
Rnt1p is an endoribonuclease that could influence the RNA levels of telomerase subunits by directly cleaving the respective RNAs thereby accelerating their degradation. To test this possibility, we monitored the impact on the expression of the telomerase subunits of a point mutation in RNT1 (D247R), which leads to an impaired Rnt1p RNA-cleavage activity without affecting other functions (45,46). Total RNA was extracted from wild type, rnt1⌬, or rnt1-D247R cells and analyzed using Northern blots. As shown in Fig. 3C, RNAs extracted from rnt1⌬ and rnt1-D247R cells exhibit similar increases in the mRNAs of telomerase subunits when compared with wild type cells. We conclude that the RNA-cleavage activity of Rnt1p is required for the regulation of normal expression levels of telomerase subunits.
Most Rnt1p substrates include a short RNA stem capped with a terminal AGNN tetraloop (27)(28)(29). We exploited this feature to search for potential Rnt1p substrates within the RNAs coding for telomerase subunits. This in silico search was performed using a previously established strategy that scores potential substrates based on similarity to known cleavage signals and other overall structural features (47). One stem-loop structure with high similarity to known substrates was predicted within the coding sequence of Est1 mRNA near its 3Ј end (Fig. 4A). A model RNA representing the predicted cleavage signal was synthesized in vitro and tested for cleavage using purified recombinant Rnt1p. As shown in Fig. 4B, the Est1 RNA structure is cleaved by Rnt1p in vitro at the predicted distance from the tetraloop and mutations that disrupt the formation of the stem-loop structure (Est1-4) abolished the cleavage. We also tested the capacity of Rnt1p to cleave the predicted structure in the context of mRNA isolated from cells. Total RNA was extracted from wild type cells or cells expressing a version of Est1 mRNA (EST1-4) carrying silent mutations that disrupt the Rnt1p cleavage signal (Fig. 4A). The RNA was incubated with purified recombinant enzyme and the cleavage was detected by either Northern blot or primer extension. As shown in Fig. 4C, Est1 mRNA is cleaved by Rnt1p, releasing a product of ϳ2000 nucleotides, consistent with the cleavage of the predicted structure, whereas the mRNA for Est1-4 was not cleaved. Primer extension analysis mapped the cleavage sites of Rnt1p at 14 and 16 bp from the stem-loop as predicted (data not shown). Examination of the EST1 sequence from other species revealed the presence of a potential Rnt1p cleavage site in all Saccharomyces species where the enzyme specificity is conserved (data not shown). To ensure that our in silico search did not miss any potential cleavage sites, we performed a similar in vitro cleavage assay on the RNAs coding for the other components of telomerase. This analysis did not reveal any Rnt1pspecific cleavage in vitro (data not shown). These results suggest that Est2, Est3, and Tlc1 RNAs do not contain canonical Rnt1p cleavage sites, but they do not exclude the possibility that these RNAs are cleaved by Rnt1p using an alternative mechanism in vivo. We conclude that the mRNA coding for Est1p is a direct substrate of Rnt1p.
Most known mRNA substrates of Rnt1p are cleaved by this enzyme in response to variation in growth conditions or cellu-lar signals (22,24). Therefore, we presumed that the cleavage of the Est1 RNA in vivo might also be triggered by a specific signal. The function of Est1p is cell cycle regulated and there is evidence that the protein is associated with active telomerase at the telomeres in late S phase (14). A small reduction of Est1 mRNA expression has also been observed in G 1 -arrested cells (48). We therefore hypothesized that Est1 mRNA stability could be regulated in a cell cycle-and Rnt1p-dependent manner. To test this hypothesis, we monitored the expression level of the Est1 mRNA during the cell cycle using synchronized cell cultures of wild type cells (EST1; bar1⌬) and cells expressing the version of Est1 mRNA that is not cleaved by Rnt1p (EST1-4; bar1⌬). EST1; bar1⌬ and EST1-4; bar1⌬ cells were synchronized in G 1 phase of the cell cycle with ␣-factor and after release, cell cycle progression was monitored by FACS analysis (Fig. 5A). As documented by the FACS profiles, a proportion of the cells were delayed in re-entering the cell cycle after release. To ensure that this incomplete release does not interfere with the detection of possible cell cycle effects on mRNA expression, we monitored the expression levels of the well established cell cycle-regulated gene CLN2 (49). Total RNA was extracted at each time point from the synchronized cultures and analyzed by Northern blot (Fig. 5B). The relative mRNA levels were determined using real-time PCR (Fig. 5C). As expected, the mRNA of the cell cycle-regulated gene CLN2 displayed similar expression patterns in both EST1; bar1⌬ and EST1-4; bar1⌬ cells and its expression in both cases was tightly linked to cell cycle progression (Fig. 5, A-C). Also consistent with previous results, we found that the expression of Est1 mRNA in EST1; bar1⌬ cells follows the cell cycle-dependent expression pattern of Cln2 mRNA (49). Est1 mRNA was least expressed in the G 1 phase as expected (48) and most expressed in S phase. This result is consistent with the model suggesting that Est1p function is required for telomerase activity at the end of S phase (14). The cell cycle-dependent induction of Est1-4 mRNA, which carries mutations disrupting Rnt1p cleavage site, was similar to that of the Est1 mRNA. However, at the end of the S phase, more Est1-4 mRNA than Est1 mRNA was detected. As observed in Fig. 5D, the highest detected Est1 RNA value was similar in EST1; bar1⌬ and EST1-4; bar1⌬ strains, whereas the lowest Est1 RNA value was significantly higher in EST1-4; bar1⌬ strain compared with the EST1; bar1⌬ strain. As expected, no difference was observed for Cln2 mRNA between the two strains. We conclude that the identified Rnt1p cleavage of the Est1 mRNA contributes to efficient cell cycledependent repression of Est1 mRNA.

DISCUSSION
In this study, we show that yeast RNase III negatively regulates the expression of yeast telomerase subunits and contributes to the cell cycle regulation of Est1 mRNA level. Decreased decay rates and increased steady-state levels of the RNAs lead to an increase in telomerase activity and to telomere elongation ( Figs. 1 and 2). The catalytic endoribonucleolytic activity of Rnt1p was required for controlling the expression of the telomerase RNP complex, suggesting that Rnt1p targets one or more telomerase-related RNAs for direct cleavage (Fig. 3). Indeed, recombinant Rnt1p cleaved Est1 mRNA in vitro and mutations that disrupt Rnt1p cleavage impaired the cell cycle repression of Est1 mRNA without affecting the decay of other cell cycle-regulated RNAs (Figs. 4 and 5). These results reveal a new layer of regulation that controls telomere length by adjusting the expression of telomerase subunits and identify new targets for eukaryotic RNase III.
Efforts to understand how telomere length is regulated revealed a cell cycle controlled machinery that encompasses competing telomere elongation and shortening factors (50,51). However, very little is known about how the expression of telomerase itself is regulated and what impact variations in its expression levels may have on the telomere length. Earlier studies indicated that the expression levels of individual components of the telomerase RNA complex are not interdependent because an overexpression of any single factor appeared not to alter the expression of the others (9,52,53). It was proposed that an increase in the telomerase activity requires increased expression of at least the majority of the components of the telomerase RNP complex (52). However, conditions that change these expression levels accordingly and factors that regulate them were not identified thus far. The present study identifies Rnt1p as such a factor. The increase in telomerase activity upon the deletion of RNT1 is not a generic response to general perturbation in RNA metabolism (Fig. 3). In addition, previous studies have shown that mutations inhibiting nonsense-mediated mRNA decay induces the expression of Est1, Est2, and Est3 mRNAs but, unlike deletion of RNT1, lead to short telomeres (11). The simultaneous induction of the different telomerase subunit RNAs in the absence of RNT1 indicates that either Rnt1p controls the expression of each RNA independently or it induces the expression of a single RNA that in turn induces the  FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 others. We prefer the first possibility because expressing the subunit RNAs individually from a heterologous and repressible promoter did not affect the expression of the other subunits. 5 Consistently, a mutation that only disrupts Rnt1p cleavage of Est1 mRNA does not increase the expression of the other components of the telomerase holoenzyme (data not shown).

Rnt1p as a Negative Regulator of Yeast Telomerase
The results thus raise the question of how Rnt1p exerts this negative regulation on the expression levels of the mRNAs coding for telomerase subunits. One possibility is that Rnt1p simultaneously cleaves Est1, Est2, Est3, and Tlc1 RNAs in the nucleus, before they are exported to the cytoplasm (54). This is supported by the fact that the catalytic activity of Rnt1p is required for the normal expression of all four RNAs and that the response of these RNAs to the inactivation of Rnt1p follows the same kinetics as that of known substrates. Moreover, our data do show that Est1 mRNA can directly be cleaved by Rnt1p in vitro (Fig. 4). Our failure to detect such direct cleavage of Est2, Est3, and Tlc1 RNAs could be explained if those RNAs require additional cellular factors for cleavage. Previous studies have shown that chaperones may mediate the cleavage of noncanonical Rnt1p substrates (55). For example, processing of the snoRNA U18 is Rnt1p-dependent in vivo, but recombinant Rnt1p alone does not cleave pre-U18 snoRNA in vitro (55). Such an in vitro cleavage of pre-U18 requires the presence of Nop1p, a nucleolar protein that associates with U18. It is therefore possible that yet to be identified proteins mediate the cleavage of Est2, Est3, and Tlc1 by Rnt1p in a U18-like fashion. Another possibility is that Rnt1p cleaves another RNA that will affect telomerase mRNA levels. Because the catalytic activity of Rnt1p is required for normal telomerase mRNA levels, that possibility could be envisaged but no such example has been observed yet.
The discovery that Rnt1p directly cleaves Est1 mRNA adds to a growing list of Rnt1p substrates that include mRNAs coding for glucose and iron-related proteins (22,24). To date, yeast is the only eukaryote in which direct cleavage of mRNAs by RNase III enzymes has been documented. In vertebrates, conditional mRNA degradation is normally carried out in the cytoplasm by the machinery of RNA interference (56,57). The fact that Rnt1p is localized in the nucleus suggests that it cleaves its target mRNAs in the nucleoplasm, which raises questions about the function of this cleavage and its contribution to the overall regulation of gene expression (34). Nascent mRNAs are exported rapidly to the cytoplasm and at any given time, the bulk of cellular mRNAs are found in the cytoplasm (54). This explains why deletion of RNT1 or disruption of Rnt1p cleavage activity does not dramatically increase the half-life of its target RNAs. Similarly, a disruption of its cleavage site in the targeted RNA, like in the case of the Est1-4, does not lead to a major increase of the RNA steady-state level. This observation is consistent with those made in previous studies using other Rnt1p mRNA targets (22,24). It is therefore unlikely that the function of Rnt1p cleavage is to fully silence the expression of EST1 or similarly regulated genes. Instead, Rnt1p-dependent cleavage may provide a fail-safe mechanism for programmed transcrip-tion inhibition or act as a fine tuner for gene expression. Indeed, expressing the mRNAs coding for telomerase subunits from heterologous promoters dramatically increases their expression in an Rnt1p-dependent manner. 5 It is known that the abundance of Est1p is cell cycle regulated (14). Consistent with recent results (48) and with previous genome-wide transcriptional profiling (49), we show here that this is accompanied by a cell cycle-dependent regulation of the mRNA coding for Est1p. A significant increase of Est1 mRNA was observed during the transition from G 1 to S phase of the cell cycle. This increase correlates with an increase of the protein level observed in S phase (14). After S phase and when cells enter G 2 , the level of Est1 mRNA rapidly decreased (Fig. 5). It is noteworthy that the normally strictly nucleolar localization of Rnt1p is relaxed exactly at this point in the cell cycle such that the enzyme is found throughout the nucleus in G 2 /M (34). This cell cycle regulated control of Rnt1p localization could thus be responsible for the cell cycle-dependent degradation of the Est1 mRNA. A comparable situation was previously reported for the endoribonuclease MRP: the exit of this enzyme from the nucleolus allows the cleavage of the cell cycle-regulated Clb2 mRNA (58). The G 1 /S phase-dependent increase in Est1 mRNA therefore occurs when Rnt1p is sequestered in the nucleolus and consequently is unchanged, even if Rnt1p cleavage site is disrupted in the Est1-4 mRNA. This implies that transcription of EST1 itself is induced in a cell cycle-dependent manner and corroborates the inclusion of this gene in the list of G 1 /S-induced genes that encompasses several DNA-replication genes (49). The G 2 -dependent decrease of the Est1 mRNA level was not blocked by the deletion of RNT1, suggesting that normal decay of the mRNA combined with reduced transcription contribute to Est1 repression. Therefore, our results indicate that telomerase activity in yeast is regulated by a combination of transcriptional and post-transcriptional events.