Hypermethylation of yeast telomerase RNA by the snRNA and snoRNA methyltransferase Tgs1

Telomeres are the protective ends of linear chromosomes. They are necessary to prevent end-to-end fusions between chromosomes and to shield the chromosome ends from degradation and inappropriate recombination (Smogorzewska and de Lange, 2004). Telomeres are specialized protein-DNA structures defined by simple repetitive sequence motifs (e.g. TG_1-3 in Saccharomyces cerevisiae). Inherently, the conventional DNA replication machinery is unable to fully replicate chromosome ends, thereby causing the progressive shortening of telomeres. Without a mechanism that elongates telomeres after every replication cycle, telomere function would be impaired, resulting in chromosomal instability, loss of genetic information, senescence and finally cell death (Gilson and Geli, 2007; Smogorzewska and de Lange, 2004). Telomerase is an evolutionarily conserved ribonucleoprotein complex consisting of three protein subunits, Est1, Est2 and Est3, and an RNA moiety, which is termed TLC1 in S. cerevisiae. This complex solves the end replication problem by elongating the G-rich DNA strand (Gilson and Geli, 2007; Nugent and Lundblad, 1998). Telomerase is a reverse transcriptase that uses the telomerase RNA as a template and in S. cerevisiae adds TG_1-3 repeats onto the 3′ end of the telomeric DNA. Telomere elongation by telomerase occurs in the late S phase of the cell cycle, and telomerase-dependent single-stranded overhangs are predominantly detectable in this specific phase (Dionne and Wellinger, 1996; Wellinger et al., 1993). However, under rare circumstances, recombination-mediated pathways maintain telomeres in cells in the absence of telomerase (Lundblad and Blackburn, 1993).

Telomerase action is coupled with conventional DNA replication to form double-stranded telomeric DNA. To this end, telomerase cooperates with the DNA polymerase α-primase complex to perform lagging strand synthesis, and this concerted interaction also regulates telomerase activity (Diede and Gottschling, 1999; Grossi et al., 2004). Disturbance of this balanced cooperation, for instance by certain mutations in the telomerase or in DNA polymerase (Polαa), can result in the accumulation of 3′ overhangs and an increase in telomeric repeat length. The single-strand DNA-binding protein Cdc13 has the ability to regulate telomere replication both positively and negatively. Cdc13 recruits telomerase to the chromosome ends, but also interacts with Pol1, the catalytic subunit of Polαa, to promote C-strand synthesis (Qi and Zakian, 2000). The regulatory protein Stn1 competes with telomerase for binding of Cdc13, thus inhibiting the recruitment of telomerase by Cdc13 (Grandin et al., 1997). Stn1 also interacts with Pol12, the B subunit of Polαa, and links telomerase action with the completion of lagging strand synthesis (Grossi et al., 2004).

Telomere length regulation is a sophisticated and complex process that is not completely understood. It comprises a multitude of factors, which together ensure that telomere length is held at a steady-state level. Rap1 is a key component of a process that `measures' telomere length in yeast. A high number of Rap1 molecules bound to telomeres inhibits telomerase activity, whereas a low number of Rap1 molecules facilitates telomerase action (Marcand et al., 1997). This negative regulation requires the proteins Rif1 and Rif2, which interact with the Rap1 C-terminus (Levy and Blackburn, 2004; Wotton and Shore, 1997). Rap1 also recruits the SIR protein complex to telomeres, thus resulting in the formation of heterochromatin-like structures and causing repression in regions adjacent to telomeres (Moretti et al., 1994). Sir2, the histone deacetylase component of the SIR complex (Imai et al., 2000), is also involved in several other cellular functions such as the maintenance of genome stability via the repair of double-stranded DNA breaks by NHEJ (Tsukamoto et al., 1997) and the modulation of cellular aging (Kaeberlein et al., 1999). The double-stranded DNA repeats at the telomere are also bound by the Ku70/Ku80 heterodimer. Ku protects telomeres from degradation (Boulton and Jackson, 1996a), determines the subnuclear localization of telomeres (Laroche et al., 1998) and regulates telomeric silencing (Boulton and Jackson, 1996b) and the timing of telomere replication (Cosgrove et al., 2002). Ku also participates in the recruitment of Sir proteins to the telomeres (Martin et al., 1999).

TLC1, the RNA subunit of S. cerevisiae telomerase, was first identified in a screen for high-copy disruptors of telomeric silencing (Singer and Gottschling, 1994). A 48-nucleotide stem-loop structure of TLC1 binds to Ku70 (Peterson et al., 2001). TLC1 and snRNAs have several structural and functional features in common. Both RNAs are transcribed by PolII, have a high uridine content and are able to bind Sm proteins. Furthermore, TLC1 bears a methyl-2,2,7-guanosine (m₃G or TMG) cap at its 5′ end (Seto et al., 1999). The cap structure of small nuclear RNAs (snRNAs) and small nucleolar RNAs (snoRNAs) is formed by the hypermethylation of a 7-monomethylated guanosine by Tgs1, an S-adenosyl-L-methionine (S-AdoMet)-dependent methyltransferase (Mouaikel et al., 2002). The deletion of TGS1 causes a cold-sensitive growth phenotype. In the absence of Tgs1, the conversion of the m⁷G cap to the m₃G structure is missing, resulting in a splicing defect at the restrictive temperature that correlates with the retention of U1 snRNA in the nucleolus. Furthermore, the nucleolar morphology is disturbed in the absence of Tgs1. Also, ribosome biogenesis is impaired in tgs1Δ at the reduced temperature, and to a certain extent at the permissive temperature. Intriguingly, changes in ribosome biogenesis and nucleolar morphology are not dependent on the methyltransferase activity of Tgs1, because the catalytically inactive but cold-sensitive mutants tgs1-W178A and tgs1-D103A do not show these effects (Colau et al., 2004; Mouaikel et al., 2003).

Here, we demonstrate that the m₃G cap formation of the telomerase RNA TLC1 depends on Tgs1 and that this TLC1 modification influences telomere length, structure and function. Tgs1 loss increased telomeric silencing and decreased the rate of artificially triggered telomeric recombination events. Furthermore, we observed that cells lacking Tgs1 activity displayed premature aging. We conclude that the sophisticated functional interaction of telomerase and DNA polymerase in the modulation of telomere elongation is disrupted in the absence of TLC1 hypermethylation by Tgs1.

Tgs1, an S-AdoMet dependent RNA methyltransferase resident in the nucleolus, generates the m₃G cap of U1, U2, U4 and U5 snRNA and a subset of snoRNAs (Mouaikel et al., 2002). Since TLC1 shares several features with snRNA (Seto et al., 1999), we determined whether TLC1 hypermethylation depended on Tgs1. We performed immunoprecipitations using an antibody specifically recognizing the m₃G structure and analyzed the precipitates for the presence of TLC1 RNA by northern blotting. In this assay, TLC1 was readily immunoprecipitated from wild-type cells. However, no precipitation of TLC1 was detected in tgs1Δ, although TLC1 was present in the supernatant of the immunoprecipitates (Fig. 1A). This showed that hypermethylation of TLC1 was dependent on Tgs1.

We next asked how the m₃G structure of the telomerase RNA influences the telomeres. We first determined whether the absence of Tgs1 affected the average telomere length in yeast cells. Telomeric restriction fragment analysis (TRF) of DNA from isogenic wild-type and tgs1Δ strains showed that the telomeres were longer in the absence of Tgs1 (Fig. 1B). Quantification indicated that the telomeres of the wild-type strain were on average 1084 bp long, which was consistent with previous measurements of telomere length in this strain (Austriaco and Guarente, 1997; Grandin et al., 2001). In the tgs1Δ strain, the telomeres were on average 27 bp longer than in the wild-type strain (Table 1). Furthermore, the telomere elongation (1) was stable when cells were passaged for eight generations; (2) was not further increased at the non-permissive temperature; and (3) could be reproduced using different subclones (data not shown). Notably, the catalytically inactive mutant tgs1-W178A, which shows normal nucleolar morphology and efficient pre-rRNA processing (Colau et al., 2004; Mouaikel et al., 2003), caused the same elongated telomere phenotype as tgs1Δ (Fig. 1C). This suggested that the tgs1Δ effect on telomere length was not due to a defect in nucleolar morphology, but depended on the catalytic activity of Tgs1. Furthermore, because mutants defective in pre-mRNA splicing (prp17Δ, prp18Δ and snu17Δ) showed no defect in telomere length regulation (data not shown) and splicing defects of tgs1Δ were only detected at restrictive temperature, this suggested that the tgs1Δ effect on telomere length was independent of splicing defects. In summary, the above observations indicate that the loss of TLC1 cap hypermethylation causes defects in telomere length regulation.

We examined whether the absence of Tgs1 could suppress or exacerbate the effect of other mutations affecting telomere length. Rap1 is a key component of telomere length regulation and controls telomere extension by a `protein-counting mechanism' (Marcand et al., 1997). The rap1-12 mutant possesses longer and highly heterogeneous telomeres (Sussel and Shore, 1991) because the negative telomerase regulation mechanism is disturbed (Buck and Shore, 1995). When TGS1 was also deleted, telomeres in rap1-12 cells were shorter and less heterogeneous than in the rap1-12 single mutant but longer than in a tgs1Δ and tgs1-W178A strain (Fig. 2A; Table 1). This suggested that the negative control of telomerase by Rap1 is restored in the absence of Tgs1.

The C-terminus of Rap1 interacts with Rif1 and Rif2 to inhibit telomerase, probably by forming a higher-order telomere structure that regulates telomere length (Levy and Blackburn, 2004; Wotton and Shore, 1997). Single deletions of RIF1 and RIF2 cause elongated telomeres. An additional deletion of TGS1 partially suppressed the elongated telomere phenotype of the single rif2Δ mutant (Fig. 2B; Table 1), which was similar to its effect on rap1-12.

We next investigated the influence of TLC1 hypermethylation on Ku70 function in telomere lengthening. In contrast to rap1-12 and rif2Δ, the ku70Δ strain exhibits shorter telomeres (Boulton and Jackson, 1996a). An additional tgs1 deletion partially suppressed the shortened-telomere effect of the ku70 deletion mutant (Fig. 2C; Table 1), showing that tgs1Δ acts in a Ku70-independent pathway in telomere-length regulation.

Sir4 as a part of the SIR complex forms heterochromatin-like structures at the HM loci and at telomeres in S. cerevisiae (Rusche et al., 2003). The length of telomeres is moderately shortened in sir4Δ cells (Palladino et al., 1993). Telomere length of the sir4Δ tgs1Δ mutant was longer than that in sir4Δ but shorter than in tgs1Δ (Fig. 2D; Table 1), showing that tgs1Δ partially suppressed the telomere-shortening effect of sir4Δ and suggesting that Sir4 and Tgs1 act in distinct pathways of telomere regulation.

Telomere lengthening is often associated with increased telomeric silencing (Buck and Shore, 1995; Palladino et al., 1993). We therefore asked whether tgs1Δ caused more telomeric repression. In wild-type cells, the ADE2 gene inserted at telomere VII becomes partially repressed by telomeric silencing, which is visible as red and white sectors within a single colony (Gottschling et al., 1990). The ratio between red and white sectors changes when the silencing machinery is disturbed. We observed that tgs1Δ or the tgs1-W178A mutant caused a higher amount of red or red-sectoring colonies compared with wild-type cells, showing that silencing was improved in the absence of Tgs1 (Fig. 3A,B). We also analyzed telomeric silencing in the rap1-12 tgs1Δ mutant. The rap1-12 mutation alone causes an increase in telomeric silencing, i.e. stronger repression of ADE2 (Buck and Shore, 1995). We observed that the red color of the rap1-12 strain was more intense than in tgs1Δ, indicating that telomeric repression in rap1-12 was stronger than in tgs1Δ. The rap1-12 tgs1Δ strain displayed an intermediate level of telomeric silencing. Notably, the amount of silencing in these strains correlated with the length of their telomeres, in that strains with longer telomeres showed more silencing, which is in agreement with previous observations of a link between telomere lengthening and stronger silencing (Buck and Shore, 1995).

Our above results suggested that the absence of TLC1 cap hypermethylation resulted in an alteration of telomere length regulation. Some mutants affecting telomere length also cause a disturbed equilibrium of telomerase action and lagging-strand synthesis, which results in a severely increased amount of single-stranded telomeric DNA of the G-strand (Adams Martin et al., 2000; Gravel et al., 1998; Grossi et al., 2004). We therefore asked whether tgs1Δ affected the level of telomeric single-stranded 3′ overhangs. Although single-stranded telomeric DNA was not detectable in wild-type cells, we detected a marked accumulation of single-stranded telomeric DNA in the absence of Tgs1 by a non-denaturing Southern blot analysis using a CA-rich DNA fragment as a probe (Fig. 4A,B). This observation suggested that the coupling of G- and C-strand synthesis at telomeres is disturbed in the absence of TLC1 hypermethylation by Tgs1.

We next investigated double mutants of tgs1Δ and genes involved in the regulation of the coupled G- and C-strand synthesis to elucidate the role of TLC1 hypermethylation in this process. Cdc13 recruits telomerase to the telomeric 3′ overhangs, thereby positively regulating telomere extension. However, Cdc13 also has a negative role in regulation of telomere replication (Chandra et al., 2001). The loss of this second regulatory activity in a cdc13-5 mutant results in elongated telomeres and an extensive elongation of the G-strand, which is accompanied by a reduced ability to coordinate synthesis of the C-strand. We examined the consequence of tgs1Δ in the cdc13-5 mutant, which on its own displays long and heterogeneous telomeres. The additional deletion of TGS1 exacerbated this effect: telomeres were longer in a cdc13-5 tgs1Δ mutant than in the cdc13-5 mutant (Fig. 5A; Table 1).

Stn1 also functions in the negative regulation of the lagging strand replication by blocking the telomerase binding site at Cdc13 (Grandin et al., 2000). Similarly to cdc13-5, the stn1-13 mutant exhibits elongated and very heterogeneous telomeres, indicating an impaired telomerase-regulating activity and a disturbed cooperation of telomerase and double-strand synthesis (Grandin et al., 1997). The additional tgs1 deletion or mutation caused a shift of telomeric signals to an even higher molecular fragment size than in the stn1-13 strain (Fig. 5A; Table 1). These additive effects suggested that Tgs1, similarly to Cdc13 and Stn1, is involved in the cooperation of telomerase and C-strand synthesis but acts through a distinct pathway.

To further test this notion, we examined the effect of the tgs1 deletion in mutants of the Polα complex. We generated double mutants of pol1-17 and pol12-216 with tgs1Δ. The temperature-sensitive pol1-17 mutation in the catalytic subunit of Polα causes slightly longer telomeres at the permissive temperature and displays drastically elongated telomeres after growth at the semi-permissive temperature (Carson and Hartwell, 1985). The pol12-216 mutant of the putative regulatory B subunit of Polα also displays extended telomeres (Grossi et al., 2004). After growth at the semi-permissive temperature, we observed in the pol1-17 mutant a partial suppression of the elongated telomere phenotype by tgs1Δ (Fig. 5B; Table 1), suggesting that Pol1 acted in a pathway distinct from that of Tgs1. By contrast, tgs1Δ was epistatic to pol12-216. The additional deletion of TGS1 caused no marked change in telomere length in pol12-216 cells (Fig. 5B; Table 1), but single-stranded overhangs in pol12-216 tgs1Δ accumulated at the same level as those in tgs1Δ (data not shown). Thus, we suggest that Tgs1 acts downstream of Pol12. Since Pol12 is involved in the coordination between G- and C-strand synthesis (Grossi et al., 2004), this suggested that Tgs1, and thus TLC1 hypermethylation, was also involved in this pathway.

Since pol12-216 was originally identified as a telomere-stabilizing mutant (Grossi et al., 2004), we next asked whether the absence of m₃G capping of TLC1 affected telomeric recombination in the same way. To do so, we measured the telomeric instability in a system with relaxed telomere length regulation. The strain used in this assay carries 32 misoriented Rap1-binding sites, which cause an increased telomeric recombination rate. In this assay, telomeric instability can be detected by the increased number of red or red-sectoring colonies caused by the loss of the telomeric ADE2 marker. We found that tgs1Δ as well as tgs1-W178A had fewer red-sectoring colonies than the corresponding wild-type strain (Fig. 6), indicating that the telomeric recombination was suppressed in the absence of TLC1 cap hypermethylation. Thus, the tgs1 mutants had a similar effect in this assay as the pol12-216 mutant.

Telomere length and stability is connected to cellular aging in yeast as well as in higher organisms. Therefore, we next asked how the absence of Tgs1 influenced the replicative lifespan of yeast cells. Interestingly, we observed that tgs1Δ caused premature aging (Fig. 7A). Because the lifespan analysis was performed at a permissive temperature, we suggest that the premature aging of tgs1Δ cells was not due to disturbed splicing caused by the missing hypermethylation of snRNAs, but rather was due to the effect of the missing hypermethylation of TLC1 and the consequent effects on telomere length regulation and structure. Furthermore, the catalytically inactive tgs1-W178A mutant displayed the same premature aging phenotype as tgs1Δ (Fig. 7B), thus arguing that the aging defect was not due to the nucleolar defect, but was rather a result of the missing hypermethylation of TLC1. The pol12-216 mutant showed a very similar premature aging effect as tgs1-W178A and tgs1Δ (Fig. 7B), which was in agreement with the observed epistasis of tgs1Δ and pol12-216 in telomere length regulation.

Telomerase exists in a wide range of organisms and ensures the proper replication of linear chromosome ends. A common feature of all telomerase enzymes is the presence of an RNA that serves as a template for reverse transcription (Smogorzewska and de Lange, 2004). In S. cerevisiae, telomerase comprises the 1200 nucleotide TLC1 RNA, which has structural similarities to snRNA (Seto et al., 1999). Here, we identified TLC1 as a new target for hypermethylation by the S-AdoMet-dependent methyltransferase Tgs1. Tgs1 was originally found to convert the 5′ end of several snRNAs and snoRNAs to m₃G structures (Mouaikel et al., 2002), and we show here that hypermethylation of the telomerase RNA also depended on Tgs1. The presence of the m₃G cap on TLC1 affected telomere length regulation, because telomeres were elongated and single-stranded 3′ overhangs accumulated in the absence of Tgs1. Thus, Tgs1 is identified as a new factor involved in telomere regulation in yeast.

Although several genetic screens have been performed to identify factors involved in telomere length homeostasis (Lundblad and Szostak, 1989; Singer et al., 1998), Tgs1 so far escaped identification, probably because the effect of tgs1Δ alone is modest and may easily be overlooked, especially in large-scale approaches (Askree et al., 2004; Gatbonton et al., 2006). However, additional mutations of several genes involved in telomere length regulation, namely rap1-12, rif2Δ, sir4Δ and ku70Δ, caused marked effects of tgs1Δ on telomere length and supported the notion that hypermethylation of TLC1 by Tgs1 affected telomere length regulation. In line with this finding, telomeric repression was improved in tgs1Δ cells, probably by the increased number of Rap1 molecules bound to the telomeric repeats and subsequent improved recruitment of the Sir proteins.

Since several snRNAs and snoRNAs lack hypermethylation in tgs1Δ cells (Mouaikel et al., 2002), we could hypothesize that the effects observed on telomere structure and function are caused indirectly by changes in snRNA and snoRNA functions, for instance by reduced splicing activity or defects in ribosome biogenesis. However, splicing defects in tgs1Δ mutants are only seen at reduced temperature (Mouaikel et al., 2002), whereas all effects on telomere function described here were observed at the permissive temperature. Also, we observed that telomere length in mutants defective in pre-mRNA splicing, was not changed, supporting our notion that TLC1 hypermethylation, rather than splicing defects, caused changes in telomere length regulation.

Notably, we observed that tgs1-W178A, a mutant that lacks methyltransferase activity and displays normal nucleolar morphology and rRNA processing, caused the same effects on telomere length, telomeric silencing and recombination, indicating that the missing hypermethylation by Tgs1 rather than a defect in ribosomal biogenesis or nucleolar function was responsible for the observed changes. Furthermore, we found that the cellular localization of Sir2-GFP was not altered by tgs1Δ (data not shown), thus arguing that changes in telomere function and silencing were not caused by a gross Sir2 redistribution in the cell. In summary, we favor the view that the absence of Tgs1 directly affected telomerase via the missing TLC1 m₃G cap rather than through indirect effects in splicing, ribosome assembly, nucleolar morphology or Sir2 distribution.

How does the missing m₃G cap of TLC1 alter telomerase action? TLC1 is an integral part of telomerase and serves as the template for reverse transcription. TLC1 has a secondary core structure that is conserved among many eukaryotes. Mutations in either structural or template sequences of telomerase RNA cause deficiencies in the enzymatic activity of telomerase. Some structures of TLC1, for instance stem-and-loop structures within the pseudoknot, are directly required for template function (Prescott and Blackburn, 1997; Tzfati et al., 2003) or to prevent synthesis beyond the template boundary (Tzfati et al., 2000). Conversely, other TLC1 features are necessary for protein binding to Est1 (Seto et al., 2002), Est2 (Livengood et al., 2002), the Sm proteins (Seto et al., 1999) and to Ku70/80 (Peterson et al., 2001). It is thus conceivable that the missing m₃G cap resulted in changes in the secondary TLC1 structure. These in turn might lead to changes of telomerase activity either by directly altering template function or by disturbing the interaction with Est or Sm proteins, thus causing reduced telomerase activity. However, mutations affecting Sm protein binding (Seto et al., 1999) and TLC1 dimerization (Gipson et al., 2007), and changes in cellular concentrations (Mozdy and Cech, 2006) showed telomerase defects that were distinct from those of tgs1Δ. It is also possible that the intrinsic enzymatic activity of telomerase is unchanged, but that the missing m₃G cap might alter the positive or negative regulation processes of telomerase action at the telomere. Our genetic data supports the hypothesis that telomerase regulation rather than enzymatic activity was affected. Also, the effect of a missing TLC1 m₃G cap might be comparable with that of the absence of the Pif1 helicase, which acts as an inhibitor of telomerase nucleotide addition processivity in yeast (Boule et al., 2005). By contrast, tgs1Δ has effects that are distinct from those of the reduction of TLC1 length, which causes shortened telomeres, even though the in vitro telomerase activity seems to be higher than for wild-type TLC1 (Zappulla et al., 2005).

Importantly, we observed an accumulation of single-stranded 3′ overhangs in the absence of Tgs1 and genetic links of TGS1 to genes encoding Polα components, strongly supporting the notion that the missing m₃G cap of TLC1 affected the coordination between G- and C-strand synthesis. Our data imply that the genetic interactions between TGS1 and other genes regulating telomere homeostasis are complex. However, tgs1Δ was epistatic to pol12-216, a mutant of the B subunit of Polα, and displayed the same stabilizing phenotype in a telomeric recombination assay (Grossi et al., 2004), as well as a similar premature aging phenotype in the replicative lifespan analysis, suggesting that Tgs1 functioned downstream of Pol12 and affected the coupling between telomerase and DNA polymerase activity. This pathway is distinct from telomerase inhibition by Cdc13 and Stn1, because the defects in mutants of both telomerase regulators were exacerbated by tgs1Δ. However, although similar in some aspects to pol12-216, tgs1Δ also differs in other aspects, because tgs1Δ, but not pol12-216, caused an increase in single-stranded overhangs. Taken together, our observations suggest that in cells lacking the m₃G cap of TLC1, telomerase inhibition by Polα is disturbed. Consequently, telomerase may be less restricted than under wild-type conditions in the addition of TG repeats to the G-strand.

Telomere length is intimately linked to the replicative capacity of cells in yeast as well as in higher eukaryotes, although there is not a simple relationship (Austriaco and Guarente, 1997; Blasco, 2005; Campisi, 2005). Interestingly, we observed that yeast cells exhibited premature aging in the absence of Tgs1. The aging assays were performed at the permissive temperature for the tgs1 deletion: conditions under which no splicing defect is observed (Colau et al., 2004; Mouaikel et al., 2002). Furthermore, the catalytically inactive tgs1-W178A mutant displayed a similar premature aging effect as tgs1Δ and pol12-216. Thus, the simplest explanation for the aging defect of tgs1Δ is that the premature aging is the result of the absence of an m₃G cap on TLC1. However, we cannot exclude the possibility that an unknown target of Tgs1 exists that affected aging directly or indirectly at the permissive temperature. Notably, the observation that tgs1Δ caused life shortening in yeast is consistent with previous studies, which found that telomere lengthening (as seen in tgs1Δ) causes a lifespan reduction, whereas shortening of telomeres increases lifespan (Austriaco and Guarente, 1997).

Similarly to snRNA, TLC1 is rich in uridine, is transcribed by RNA polymerase II and binds to the same Sm proteins as found in snRNPs (Seto et al., 1999). As shown here, both RNA types are provided with a m₃G cap by the same methyltransferase. Thus, one can hypothesize that TLC1 and snRNAs also share their biogenesis pathway. In higher eukaryotes, spliceosomal U snRNAs (uridine-rich small nuclear RNAs) are exported into the cytoplasm to bind Sm proteins (Mattaj, 1986). They are then re-imported into the nucleus by Snurportin1 (SNP1), which binds to the m₃G cap and together with importin β (Impβ) triggers the nuclear import of snRNA (Huber et al., 1998). In this case, the m₃G cap is part of the bipartite nuclear localization signal, which facilitates nuclear import of snRNA. However, a yeast ortholog of Snurportin1 has not yet been identified, and it is unclear whether yeast snRNPs follow this pathway. Given that TLC1 can shuttle between cytoplasm and nucleus (Teixeira et al., 2002), that the importin Mtr10 is required for full telomerase activity (Ferrezuelo et al., 2002), and that Tgs1 is located in the nucleolus (Mouaikel et al., 2002), our data suggest the following alternative scenarios: (1) TLC1 is hypermethylated in the nucleolus and then exported into the cytoplasm. (2) TLC1 is hypermethylated in the cytosol. This implies shuttling of Tgs1 between nucleus and cytoplasm, or the presence of a small cytosolic Tgs1 fraction. (3) Hypermethylation occurs after re-import of TLC1. However, our data show that capping of TLC1 was not essential for the TLC1 re-import. Notably, the effect of tgs1Δ on telomere length was distinct from that of the lack of Mtr10 (Ferrezuelo et al., 2002), suggesting that tgs1Δ cells did not have a major TLC1 import defect. Thus, the mechanism of TLC1 re-import apparently differs from that of human snRNAs. However, it is also possible that the missing TLC1 cap affects the kinetics of TLC1 nuclear import or of telomerase assembly.

Tgs1 is evolutionarily conserved, with homologues in fungi, flies and humans. The Drosophila homologue DTL and the human PIMT/TGS1 are larger than S. cerevisiae Tgs1, suggesting that they might have more complex functions than Tgs1. PIMT was originally isolated as a factor that interacts with the nuclear coactivator PRIP (peroxisome proliferator-activated receptor-interacting protein) (Zhu et al., 2001). Thus, one might speculate that its methylation activity is coupled to transcription. Also, because the human telomerase RNA TERC, like TLC1, carries a m₃G cap (Jady et al., 2004), it is possible that PIMT provides TERC hypermethylation in human cells. In agreement with this, PIMT displays RNA-binding activity. Conversely, the Drosophila homolog of Tgs1, DTL, probably lacks telomerase RNA methylation function, because Drosophila telomeres are maintained by a telomerase-independent mechanism. It will be interesting to see whether PIMT provides TERC hypermethylation and how the capping affects telomere length regulation and aging in human cells.

Yeast tgs1 deletion strains were produced either by transformation of a PCR generated HIS-MX cassette or by genetic crosses to the required partner strains and subsequent tetrad dissection. Deletions were verified by PCR analysis. For genomic integration of tgs1-W178A, a XbaI-HindIII fragment from pGFP-Tgs1/W178A (Mouaikel et al., 2003) carrying the mutant allele lacking a promoter was cloned into pRS405. pRS405-tgs1-W178A was linearized in TGS1 by NsiI and transformed into wild-type yeast cells. This integration created a strain with tgs1-W178A under the native TGS1 promotor and with a promotor-less wild-type TGS1 allele. Such integrants displayed the cold-sensitive phenotype of the tgs1-W178A allele. The strains used in this study are shown in supplementary material Table S1.

Yeast cell extracts were prepared from 40 OD units of logarithmically growing cells. Cells were harvested and washed once in PBS. Pellets were resuspended in 500 μl lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.05% NP40, 1% Triton X-100, complete protease inhibitor, 1 mM PMSF, 5 U RNasin, 0.5 mM DTT) and vortexed in the presence of glass beads at 4°C for 2 minutes. Cell debris was removed by centrifugation at 12,500 g for 5 minutes and 50 μl of the supernatant were set aside as input. 10 μl of antibody raised against the 2,2,7-trimethylguanosin structure [Oncogene, K121 (Mouaikel et al., 2002)] was added to the remaining supernatant. After 2 hours of binding, the lysate-antibody mix was incubated with Protein-G-Sepharose (Amersham Pharmacia) for 2 hours. Immunoprecipitates were washed three times with lysis buffer. Antibody-RNA complexes were eluted from Protein-G-Sepharose by adding 250 μl of 1% SDS/TE twice to the beads followed by incubation at 65°C for 10 minutes. Supernatants were removed and kept for further analysis. After treatment of input, immunoprecipitates and supernatants with 0.4 mg/ml proteinase K for 30 minutes at 37°C, a phenol-chloroform extraction was performed. RNA was precipitated with 0.3 M sodium acetate pH 5.3 and three volumes of cold 100% ethanol. RNA was separated by MOPS-formaldehyde agarose electrophoresis and analyzed by northern blot as described (Marchfelder et al., 2003). As a probe, a 1300 bp XhoI fragment of pTRP-TLC1 (Singer and Gottschling, 1994) labeled by random primer labeling with [α-³²P]dCTP was used. Blots were exposed to a phosphorimager screen (Molecular Dynamics) for 12 hours.

Genomic DNA was digested with XhoI, separated by agarose gel electrophoresis and transferred onto Zeta probe GT membranes (Bio-Rad) as described (Sambrook et al., 1989). Southern blots were hybridized with a C_1-3A containing 280 bp EcoRI fragment of pCT300 labeled by random primer labeling with [α-³²P]dCTP as a probe (Wellinger et al., 1993). Telomere length was determined using the QuantityOne software package (Bio-Rad). Values are derived from at least two independent determinations.

Analysis of single-stranded overhangs was performed by non-denaturing Southern blot analysis by treating the agarose gels with 10× SSC for 30 minutes after electrophoresis. DNA was blotted onto Zeta probe GT membranes using 10× SSC as a transfer buffer and hybridized as described above.

Analysis of replicative lifespan of wild-type, tgs1Δ, tgs1-W178A and pol12-216 strains was performed as described (Park et al., 2002). Cells were incubated at 30°C for one doubling time between micromanipulation separation steps.

We thank Rémy Bordonné, Dan Gottschling, Jasper Rine, David Shore and Virginia Zakian for generously providing yeast strains and plasmids. Albrecht Bindereif, Wolfgang Goedecke, Enno Hartmann and Harry Scherthan are gratefully acknowledged for critical reading of the manuscript and helpful discussions. We are especially indebted to Uta Marchfelder and Senta Gerber for excellent technical assistance and the members of the Ehrenhofer-Murray lab for scientific support. This work was supported by the Max-Planck-Society, the University of Duisburg-Essen and the Jürgen Manchot foundation (J.G.).