Mec1p is a cell cycle checkpoint protein related to the ATM protein kinase family. Certain mec1 mutations or overexpression of Mec1p lead to shortened telomeres and loss of telomeric silencing. We conducted a multicopy suppressor screen for genes that suppress the loss of silencing in strains overexpressing Mec1p. We identified SCS2 (suppressor of choline sensitivity), a gene previously isolated as a suppressor of defects in inositol synthesis. Deletion of SCS2 resulted in decreased telomeric silencing, and the scs2 mutation increased the rate of cellular senescence observed for mec1-21 tel1 double mutant cells. Genetic analysis revealed that Scs2p probably acts through a different telomeric silencing pathway from that affected by Mec1p.
IN the yeast Saccharomyces cerevisiae, chromosomes terminate in a simple repetitive sequence [poly(G1-3T)] that is ∼350-500 bp in length (Greider 1996). The telomeric repeats are packaged into a non-nucleosomal type of chromatin (Wrightet al. 1992). Telomere chromatin structure prevents the transcription of reporter genes at the telomere, a phenomenon called telomere position effect (TPE) or telomeric silencing (Gottschlinget al. 1990).
Telomeric silencing requires a number of proteins that bind at the telomere. For example, the Rap1p (repressor and activator protein) binds directly to telomeric DNA (Gilsonet al. 1993). Rap1p then recruits the Sir3 and Sir4 (silent information regulator) proteins (Morettiet al. 1994; Hecht et al. 1995, 1996). The Rif1 protein (Rap1p-interacting factor; Hardyet al. 1992) competes with Sir3p for binding to Rap1p and acts as a negative regulator of telomeric silencing (Kyrionet al. 1993; Morettiet al. 1994). Additional proteins involved in regulating telomeric silencing include the H3 and H4 histones, proteins involved in regulating posttranslation modifications of histones, and the DNA end-binding Ku proteins (reviewed by Lustig 1998). Many of these proteins also regulate silencing of the silent mating-type loci (Aparicioet al. 1991).
The Mec1p (mitotic entry checkpoint) directs the cellular response to DNA damage and S-phase arrest (Allenet al. 1994; Weinertet al. 1994) and regulates telomere length (Ritchieet al. 1999). The mec1-21 allele or overexpression of the wild-type MEC1 results in loss of telomeric silencing (Craven and Petes 2000); some mutant alleles of MEC1 result in shortened telomeres without loss of telomeric silencing (Longheseet al. 2000). The mec1-21 silencing defect can be suppressed by a mutation in the SML1 (suppressor of mec1 lethality) gene (Craven and Petes 2000); mutations in SML1 result in elevated nucleotide pools (Zhaoet al. 1998). Mec1p directs a signaling cascade that includes the Dun1p kinase (Zhou and Elledge 1993), and dun1 cells also exhibit shortened telomeres and decreased telomeric silencing (Craven and Petes 2000; Longheseet al. 2000).
Strains with mutations in both MEC1 and the related TEL1 gene (Lustig and Petes 1986; Greenwellet al. 1995) have very short telomeres and undergo cellular senescence (Ritchieet al. 1999). Following ∼50 generations of attenuated growth, “survivor” colonies appear by a recombination-dependent mechanism (Ritchieet al. 1999). In summary, the related Mec1p and Tel1p are required for telomere length regulation; Mec1p, but not Tel1p, also has a role in telomeric silencing. Similar observations have also been made in Schizosaccharomyces pombe. Strains with mutations in both rad3+ (the gene equivalent to MEC1) and tel1+ undergo complete loss of telomeres (Naitoet al. 1998), and strains with single mutations in the rad3+ gene lose telomeric silencing (Dahlenet al. 1998; Matsuuraet al. 1999). Thus, the functions of MEC1 at the telomere are widely conserved through evolution.
Since mutations within the kinase domain of Mec1p affect telomere length (Mallory and Petes 2000), it is likely that the effects of the mec1 mutation on telomere length and telomeric silencing are a consequence of lack of phosphorylation of downstream targets. Although Mec1p-dependent phosphorylation of a number of proteins involved in the repair of DNA damage has been observed (reviewed by Lowndes and Murguia 2000), the targets of Mec1p relevant to its effects at the telomere are unknown. One way to search for downstream targets of a protein is to screen for genes that, when overexpressed, suppress mutant phenotypes. Such multicopy suppressors may be downstream targets of the signaling protein, genes that activate competing pathways, or genes that inactivate inhibitory pathways (Guthrie and Fink 1991). Below, we describe a screen for genes that, when overexpressed, suppress the telomeric silencing defect caused by overexpression of MEC1.
MATERIALS AND METHODS
Yeast strains: All strains were isogenic with W303a (leu2-3,112 his3-11,15 ura3-1 ade2-1 trp1-1 can1-100 rad5-535; Thomas and Rothstein 1989), except for alterations introduced by transformation. The genotypes of strains used in our study are shown in Table 1. The names and sequences of oligonucleotides used for strain constructions or strain diagnosis are given in Table 2.
Most W303a-derived strains contain the rad5-535 mutation (Fanet al. 1996). We isolated RAD5 derivatives of such strains by crossing them to the isogenic RAD5 strain W1588-4C (from R. Rothstein). The presence of the rad5-535 mutation was scored by PCR amplification of genomic DNA with the primers RAD5-L and RAD5-R and treatment of the resulting DNA fragment with the MnlI restriction enzyme. The rad5-535 substitution introduces an MnlI site into the RAD5 coding sequence. In direct comparisons of RAD5 and rad5-535 strains, we found no differences in telomeric silencing.
A number of strains with deletions were constructed using the PCR method described by Wach et al. (1994). The SCS2 (suppressor of choline sensitivity) gene was replaced by HIS3 using a PCR fragment (primers, SCS2-KOF and SCS2-KOR; template, pRS303) in a one-step transplacement. The INO1 gene was replaced by HIS3 using a similar approach (primers, INO1-KOF and INO1-KOR; template, pRS303). In three strains, genes were replaced with the kanMX gene (Wachet al. 1994), which confers resistance to geneticin, by the same procedure. These genes and the primers used to generate the PCR fragment for the one-step transplacements were: RAD9 (RAD9-KOF and RAD9-KOR), TEL1 (TEL1-KOF and TEL1-KOR), and YBL091C-A (YBL-KOF and YBL-KOR); the template for the PCR reactions was pFA6-kanMX (Wachet al. 1994).
We also used PCR methods to epitope-tag Sir3p and Scs2p. The primers SIR3-F and SIR3-R for Sir3p and SCS2-F and SCS2-R for Scs2p (sequences in Table 2) were used to amplify the plasmid pFA6a-3HA-kanMX6 (Longtineet al. 1998). The resulting DNA fragments were used to transform W1588-4c to geneticin resistance. In one of the resulting strains (RCY309), the Scs2p contains two hemagglutinin (HA) epitopes inserted immediately upstream of the termination codon and there is an insertion of the kanMX cassette downstream of SCS2. In the second strain (RCY310), the Sir3p has the same 2XHA tag immediately upstream of the termination codon with the same kanMX insertion downstream of SIR3.
We assayed telomeric silencing using a construction in which the URA3 gene was inserted near the end of chromosome XVL (Gottschlinget al. 1990; Craven and Petes 2000). This construction was introduced into various genetic backgrounds by crosses. The resulting diploids (haploid strains shown in parentheses) were: RCY165 (RCY109-15d × W303aU-fr), RCY207 (RCY201 × W303α), RCY211 (RCY207-3a × RCY109-1c), RCY243 (RCY242 × RCY109-25c), RCY269 (RCY268 × RCY243-7a), RCY278 (RCY273 × RCY243-7d), RCY280 (RCY28 × RCY243-1a), RCY282 (RCY269-7c × LPY253), RCY300 (Y286 × RCY269-4a), RCY305 (RCY269-4a × RCY278-1a), RCY307 (MD89 × RCY300-6a), and RCY346 (RCY106-1d × RCY211-2b).
Multicopy suppressor screen: The strain RCY138, containing a TEL-XVL-URA3 telomere and the MEC1-containing plasmid pRC5, is sensitive to 5-fluoro-orotate (5-FOA) because overexpression of Mec1p results in loss of telomeric silencing (Craven and Petes 2000). We transformed this strain with a YEp13-borne genomic library (DeMariniet al. 1997), looking for transformants that had restored silencing. Transformants were selected on plates lacking both histidine (to maintain selection of pRC5) and leucine. Following 3 days of growth, colonies were replicated to plates lacking histidine and leucine but containing 1 mg/ml 5-FOA. Of ∼12,000 His+ Leu+ transformants examined, only 40 were resistant to 5-FOA. Further analysis showed that only 6 of these transformants suppressed the silencing defect caused by Mec1p overexpression in a plasmid-dependent manner. Plasmids were rescued from each of the 6 transformants into Escherichia coli; these plasmids were called pMOS2 (Mec1p-overexpression suppression 2), pMOS7, pMOS13, pMOS21, pMOS24, and pMOS35.
Plasmids: The plasmid pMOS2 (described above) had two open reading frames. The open reading frame (ORF) representing the SCS2 gene was subcloned as a 1.6-kb HindIII-BglII fragment into the BamHI and HindIII sites of the LEU2-containing vector YEplac181 (Gietz and Sugino 1988), resulting in the plasmid pRC12. The high-copy-number LEU2-containing pRC11 plasmid contains an insertion of the RNR1 gene (Craven and Petes 2000). The plasmid pRC5 is a high-copy-number HIS3-containing plasmid with MEC1 (Craven and Petes 2000), and pRC4 (identical to the previously described pRS4; Craven and Petes 2000) is a CEN- and HIS3-containing plasmid with the MEC1 gene. The plasmid pRS423 (Christiansonet al. 1992) is a high-copy-number HIS3-containing vector that was used as a control in some experiments. The SIR3 overexpression plasmid pLP304 contains a 4.5-kb fragment of SIR3 inserted into the LEU2-marked 2-μm vector YEp351 (Stone and Pillus 1996). The plasmid pJH318 (Hirsch and Henry 1986) contains the INO1 gene inserted into YEp351. The plasmid pBAD45 (provided by S. Elledge) has an insertion of MEC1 on a CEN-URA3-containing vector.
Genetic methods, assays for silencing, and measuring sensitivity to DNA damaging agents: Standard methods were used for transformation, media preparation, and tetrad analysis (Guthrie and Fink 1991). Because some mutant phenotypes associated with mec1 or tel1 mutations exhibit a substantial phenotype lag, strains with these mutations were subcloned for ∼100 cell generations before monitoring any phenotypes.
Telomeric silencing assays were performed as described previously (Craven and Petes 2000). Strains were grown overnight in rich growth medium (for plasmid-free strains) or appropriate synthetic media lacking specific amino acids. Cells were suspended in water and diluted 1:5 in serial increments, and 5 μl of the diluted suspensions was spotted on rich growth medium (YPD) or plates containing 1 mg/ml 5-FOA. For some strains, synthetic media lacking histidine and/or leucine were used to force retention of HIS3- and/or LEU2-containing plasmids. To test silencing of the silent mating-type locus, we used strains that contained an insertion of TRP1 integrated at the HML locus. In wild-type strains, silencing results in a Trp- phenotype (Nislowet al. 1997).
Sensitivity to inhibition of growth by hydroxyurea was examined using medium containing 50-200 mm hydroxyurea. The concentration of the DNA-damaging agent methyl methanesulfonate in the medium was 0.05%. To assay the ability of strains to grow in the absence of inositol or choline, we used vitamin-defined synthetic medium as defined by Griac et al. (1996).
Chromatin immunoprecipitation: Yeast strains (RCY309 with HA-tagged Scs2p and RCY310 with HA-tagged Sir3p) were grown in rich growth medium to an OD600 of 1-1.5. The cells were treated with 1% formaldehyde for 2 hr. Crosslinking was stopped with 1 m glycine, and cell extracts were prepared as described by Meluh and Koshland (1997). Immunoprecipitation, deproteinization, and PCR were performed as described by Strahl-Bolsinger et al. (1997). The antibody used for both immunoprecipitations and Western analysis (Mallory and Petes 2000) was HA.11 (Babco, Richmond, CA). Telomeric sequences were detected by PCR using primers homologous to the VL telomere (Millset al. 1999).
Identification of SCS2 as a multicopy suppressor of mec1 TPE defects: Cells that overexpress MEC1 lack the ability to silence a telomeric URA3 gene and, therefore, fail to grow on plates containing 5-FOA (Craven and Petes 2000). We conducted a screen for genes that, when overexpressed, suppress the MEC1 overexpression-silencing defect (details in materials and methods). In a screen of ∼12,000 transformants, we identified six different pMOS plasmids that were capable of suppressing the silencing defect. The identities of yeast genomic DNA within five of these plasmids were determined by DNA sequencing each junction of the insertion and by comparing the sequences with the Saccharomyces Genome Database. The chromosomal coordinates for each insertion were as follows (Roman numerals indicating the chromosome): pMOS2 (V, 26753-30712), pMOS7 (XV, 291737-295525), pMOS13 (VII, 77187-83915), pMOS21 (IX, 404000-410409), and pMOS24 (V, 509064-513891). The strongest suppression of the silencing defect was observed for the pMOS2 plasmid, and our subsequent analysis was restricted to this plasmid.
The plasmid pMOS2 contained two open reading frames. We constructed a plasmid (pRC12) that had one of these ORFs (YER120W, SCS2) and showed that this plasmid suppressed the silencing defect caused by MEC1 overexpression (Figure 1a). SCS2 is a protein of unknown function that has genetic interactions with proteins involved in inositol/lipid biosynthesis (Kagiwadaet al. 1998).
In addition to suppressing the telomeric silencing defect resulting from MEC1 overexpression, SCS2 overexpression suppressed the telomeric silencing defect of mec1-21 and dun1-Δ100 strains. While the mec1-21 and dun1Δ strains RCY109-1c and RCY144-4a harboring a control vector grew poorly on medium with 5-FOA, indicating a silencing defect (Figure 1b), the same strains silenced at wild-type levels upon SCS2 overexpression (Figure 1b). In contrast, SCS2 overexpression did not suppress the telomeric silencing defect of cells lacking the yKU70/HDF1 gene (Figure 1b), which encodes a DNA end-binding protein required for silencing (Boulton and Jackson 1998). Thus, the restoration of silencing by SCS2 overexpression is not generalizable to all telomeric silencing mutants.
Strains with mec1 or dun1 mutations fail to form colonies in media containing hydroxyurea (HU, an inhibitor of ribonucleotide reductase; Zhou and Elledge 1993; Allenet al. 1994) and null mutants of MEC1 are inviable (Zhaoet al. 1998). The inviability of mec1 null mutants, but not the inability to form colonies on HU-containing media, is suppressed by overexpression of RNR1 (Desanyet al. 1998), a gene encoding one of the subunits of ribonucleotide reductase. Overexpression of SCS2 did not rescue the ability of mec1-21 (Figure 2a) or dun1 (data not shown) strains to form colonies on HU-containing media. SCS2 overexpression also did not suppress the essential function of MEC1. A strain (Y602) with a mec1Δ deletion and the plasmid pBAD45 (CEN-containing plasmid with URA3 and MEC1) was transformed with a high-copy-number control plasmid (YEp lac181), a high-copy-number SCS2-containing plasmid (pRC12), or a high-copy-number RNR1-containing plasmid (pRC11). The ability of these strains to lose the MEC1-containing plasmid was monitored using medium containing 5-FOA. Only the strain with plasmid pRC11 was able to lose the MEC1-containing plasmid (Figure 2b).
The scs2Δ mutation causes loss of telomeric silencing: Deletion of the SCS2 open reading frame caused a loss of telomeric silencing, similar to that observed for the mec1-21 (Figure 3) and dun1Δ (data not shown) mutants. We measured telomeric silencing in five independent cultures of isogenic wild-type, scs2, mec1-21, and scs2 mec1-21 strains. The percentages of cells in each culture that were 5-FOAR (range of values shown in parentheses) were: 10% (4.4-14%) for wild type, 0.8% (0.5-1%) for scs2, 1.9% (1.5-2.7%) for mec1-21, and 0.06% (0.03-0.1%) for scs2 mec1-21. Telomere length was unaffected by deletion of SCS2, and scs2 mutants were not more sensitive than wild-type strains to ultraviolet light, hydroxyurea, or methyl methane-sulfonate (data not shown).
Silencing of the HML locus requires many of the same proteins necessary for telomeric silencing (Aparicioet al. 1991). Silencing at HML can be conveniently monitored using a strain in which the wild-type TRP1 gene has been inserted at HML (Stone and Pillus 1996). Silencing results in a tryptophan-requiring phenotype. As shown in Figure 4, the scs2 mutation does not reduce silencing at the HML locus, although the rap1-17 mutation, as expected (Kyrionet al. 1993), does result in a silencing defect. We previously observed that mec1-21 also reduced telomeric silencing without affecting silencing at HML (Craven and Petes 2000).
SCS2 shares 48% identity with an uncharacterized open reading frame YBL091C-A. This ORF lacks an ATG start site, but is transcribed (Velculescuet al. 1997). The strain RCY269-2b, which has a deletion of YBL091C-A, was viable and had wild-type levels of telomeric silencing (Figure 3). This deletion also has no effect on telomere length or sensitivity to HU (data not shown). Furthermore, a strain with a deletion of the YBL091C-A ORF and an scs2 mutation has approximately the same telomeric silencing defect as the single scs2Δ mutant (Figure 3). We conclude that SCS2 contributes to telomeric silencing, but that the related open reading frame YBL091C-A does not.
Double mutants of mec1-21 and tel1Δ undergo loss of telomeric sequences and cellular senescence, followed by the emergence of a small number of surviving cells (Ritchieet al. 1999). A triple mutant mec1-21 tel1Δ scs2Δ strain underwent senescence at an accelerated rate compared to mec1-21 tel1Δ mutants (Figure 5, left side). In addition, the survivors derived from the triple mutant strain were less abundant and grew more slowly than mec1-21 tel1Δ survivors. We analyzed seven tetrads containing pairs of mec1-21 tel1Δ and mec1-21 tel1Δ scs2Δ spores. For each pair, the triple mutant senesced at an earlier stage of subculturing than the double mutant. At early stages of subculturing, telomere lengths in mec1-21 tel1Δ strains were the same as those in mec1-21 tel1Δ scs2Δ strains (data not shown), suggesting that the earlier senescence in the mec1-21 tel1Δ scs2Δ strains is not likely to reflect an effect on telomere length. The mec1-21 scs2Δ or tel1Δ scs2Δ mutants were viable and did not senesce even after extended subculturing (Figure 5, right side).
The scs2Δ mutation is not suppressed by overexpression of RNR1 or INO1: The telomeric silencing defects of mec1-21 and dun1Δ are suppressed by overexpression of the RNR1 gene and by the sml1 mutation (Craven and Petes 2000); both of these alterations are likely to lead to elevated nucleotide pools (Zhaoet al. 1998). Neither overexpression of RNR1 (Figure 6a) nor the sml1 mutation (data not shown) reversed the telomeric silencing defect of scs2. These results suggest that SCS2 and MEC1 may affect different pathways required for telomeric silencing.
One model for the effect of the scs2 mutation on telomeric silencing is that scs2 cells have elevated levels of damage. In the presence of DNA damage, telomeric silencing proteins are recruited to the sites of the damage, resulting in loss of silencing; this recruitment requires the Rad9p (Millset al. 1999). Consequently, if the silencing defect in scs2 strains reflects increased levels of DNA damage, strains with mutations in both scs2 and rad9 would have increased telomeric silencing. We examined telomeric silencing in isogenic RAD5 TEL-XVL::URA3 strains with the following genotypes: scs2 (RCY307-3c), rad9 (RCY307-4a), and scs2 rad9 (RCY307-2c). The double mutant strains had the same silencing defect as the scs2 single mutant strain (data not shown), ruling out the simplest forms of this model.
SCS2 was originally identified as a suppressor of the inositol auxotrophy of CSE1 (choline sensitive, a dominant mutation) and ire15 (inositol requiring) mutants (Kagiwadaet al. 1998). Both of these mutants lack the ability to express the INO1 gene, which encodes the enzyme inositol-1-phosphate synthase (Dean and Henry 1989). The Ino1p catalyzes the conversion of glucose-6-phosphate to inositol-1-phosphate, the first committed step in inositol phosphate synthesis. The scs2Δ mutants are leaky inositol auxotrophs at elevated temperatures, and this auxotrophy is suppressed by overexpression of the INO1 gene (Kagiwadaet al. 1998). Overexpression of INO1, however, did not suppress the telomeric silencing defects of scs2Δ or mec1-21 cells (Figure 6b). In addition, deletion of the INO1 gene did not affect telomeric silencing. We conclude that the effects of the scs2 mutation on silencing are not mediated through INO1.
The scs2Δ telomeric silencing defect is suppressed by overexpression of SIR3 or by the rif1 mutation: One important component of telomeric silencing appears to be the level of Sir3p bound at and near the telomere. Sir3p binds to the carboxy terminus of the telomere-binding protein Rap1p in competition with Rif1p (Morettiet al. 1994; Hechtet al. 1996). Telomeric silencing is decreased by sir3 mutations (Aparicioet al. 1991) and elevated by overexpression of Sir3p (Renauldet al. 1993) or mutations of RIF1 (Kyrionet al. 1993). The scs2Δ telomeric silencing defect was completely suppressed by multiple copies of the SIR3 gene (Figure 7a) and by the rif1 mutation (Figure 7b). Overexpression of Sir3p also suppressed the telomeric silencing defect of mec1-21 (Figure 7a).
One interpretation of the observation that the scs2 telomeric silencing defect is suppressed by the rif1 mutation is that Scs2p negatively regulates the function of Rif1p. As described above, Scs2p overexpression suppresses the inositol auxotrophy associated with mutations in the INO1 pathway. To find out whether the rif1 mutation might interact with mutations in the INO1 pathway, we examined the ability of isogenic spores (derived from the diploid RCY346) of the wild-type, ino1, rif1, and ino1 rif1 genotypes to grow on medium lacking inositol. Wild-type and rif1 strains grew normally, whereas ino1 and ino1 rif1 strains grew very slowly (although at the same rates). Thus, Rif1p does not appear to affect the INO1 pathway.
Both telomeric heterochromatin (Gottaet al. 1996) and Scs2p (Kagiwadaet al. 1998) localize to the perinuclear region of the cell, raising the possibility that Scs2p might bind directly or indirectly to telomeres. To test this possibility, we tagged the Scs2p with an HA epitope (RCY309); we also constructed a strain (RCY310) containing HA-tagged Sir3p, a known telomere-binding protein (Hechtet al. 1996). The HA-tagged Scs2p protein was proficient for telomere silencing and could be readily detected by Western blot. Using formaldehyde crosslinking and chromatin immunoprecipitation (ChIP) analysis (details in materials and methods), we failed to detect Scs2p bound to the telomere, although we could readily detect the binding of telomeric sequences to an HA-tagged version of the Sir3p control (data not shown). Thus, it is unlikely that Scs2p affects telomeric silencing through a stable direct interaction with telomeric heterochromatin. We cannot exclude the possibility of an unstable association of Scs2p with the telomere.
The major conclusions of this study are: (1) telomeric silencing defects caused by overexpression of Mec1p or by the mec1-21 mutation are suppressed by overexpression of SCS2; (2) deletion of SCS2 causes a partial loss of telomeric silencing and accelerates senescence in mec1-21 tel1 cells; and (3) loss of silencing in scs2Δ cells is suppressed by multiple copies of SIR3 and loss of RIF1, but not by multiple copies of RNR1 or INO1. Multicopy suppressors function through one of three mechanisms: activation or increase in levels of a downstream target in the same pathway as the mutated protein, inactivation of an inhibitory pathway of the mutated protein, or activation of a parallel pathway of the mutated protein. We discuss our results in the context of these possibilities.
Previous studies identified RAD53, DUN1, and RNR1 as multicopy suppressors of the essential function of Mec1p (Sanchezet al. 1996; Desanyet al. 1998). These proteins are thought to function as downstream effectors in the same DNA repair checkpoint pathway as Mec1p (reviewed by Lowndes and Murguia 2000). Mec1p is a protein kinase (Mallory and Petes 2000; Paciottiet al. 2000) and several of the proteins downstream of Mec1p, such as Rad53p (Sanchezet al. 1996), are phosphorylated in a Mec1p-dependent fashion in vivo.
Our observation that overexpression of Scs2p suppresses the telomeric silencing defects of mec1-21 and dun1 is consistent with the possibility that Mec1p and Scs2p act in the same pathway. There are, however, several observations that are difficult to explain by this hypothesis. First, the scs2Δ silencing defect is not suppressed by elevated nucleotide pools, a condition that suppresses the mec1-21 silencing defect. Second, the scs2Δ mutation is suppressed by loss of the RIF1 gene, which does not affect the mec1-21 silencing defect (Craven and Petes 2000). Third, it is unlikely that Scs2p is a direct substrate for the kinase activity of Mec1p because an epitope-tagged version of Scs2p does not exhibit altered expression, mobility, or processing in mec1-21 cells (data not shown). Although none of these arguments are conclusive, the simplest interpretation of the data is that Scs2p does not function in the same pathway affecting telomeric silencing as Mec1p.
An alternative explanation is that the Scs2p affects a pathway that competes with that regulated by Mec1p. If Scs2p were part of a pathway that inhibits Mec1p and if Scs2p overexpression disrupted this pathway, then Scs2p overexpression might restore silencing to mec1-21 mutants. This model, however, does not explain the loss of silencing observed in scs2Δ strains, since loss of Scs2p should result in more efficient Mec1p-mediated telomeric silencing.
Consequently, we favor a model in which Scs2p regulates telomeric silencing in a pathway operating independently of Mec1p. Since the silencing defect of scs2Δ mutants is restored by altering the balance of Sir3p at the telomere by overexpression of Sir3p or loss of the competing Rif1p (Figure 7), one possibility is that Scs2p acts in the Sir3p/Rif1p pathway of silencing. By this model, Scs2p could positively regulate Sir3p proteins or negatively regulate Rif1p. Scs2p is related to the Aplysia californica synaptobrevin/vesicle-associated membrane protein (VAMP)-associated protein VAP-33, which functions in protein secretion (Skehelet al. 1995; Lapierreet al. 1999). Scs2p is associated with the endoplasmic reticulum (ER), although scs2 strains do not have defective protein secretion (Kagiwadaet al. 1998). Kagiwada et al. (1998) suggested that Scs2p might be a membrane-bound transcription factor that is released from the ER to the nucleus in response to certain cellular signals. If the Scs2p is a transcriptional activator of one or more silencing proteins, then overexpression of Scs2p might relieve the telomeric silencing defect caused by overexpression or mutation of the Mec1p. Loss of Scs2p might result in a diminished level of silencing proteins and partial loss of telomeric silencing. Although we observed no effect of the scs2 mutation on the silent mating-type loci, telomeric silencing is often more sensitive to subtle changes in the levels of silencing proteins than silencing at the mating-type loci (Aparicioet al. 1991). It is unlikely that Scs2p acts as a negative regulator of Rif1p, since strains with rif1mutations have elongated telomeres (Hardyet al. 1992) and scs2 strains have wild-type-length telomeres. Thus, for Scs2p to be a negative regulator of Rif1p, the telomere-length regulatory activity of Rif1p would have to be separable from its effects on telomeric silencing.
One alternative intriguing possibility is that Scs2p alters silencing indirectly through the synthesis or processing of phospholipids. These lipids might serve as docking sites for heterochromatin on the nuclear membrane or be part of a signaling cascade that regulates silencing. One argument against this model is that the only known target of Scs2p in the phospholipid pathway, the INO1 gene, has no effect on silencing when deleted or overexpressed.
In summary, SCS2 is involved in regulating telomeric silencing. Although we identified SCS2 in a genetic screen for genes that were multicopy suppressors of a silencing defect associated with Mec1p overexpression, our results suggest that Scs2p regulates telomeric silencing in a different pathway from Mec1p.
We thank K. Ritchie and J. Mallory for helpful discussions and comments on the manuscript; M. Dominska and L. Stefanovic for expert technical assistance; and R. Rothstein, D. DeMarini, J. Pringle, L. Pillus, S. Henry, and S. Elledge for strains and plasmids. This work was supported by a National Institutes of Health grant GM52319 to T.D.P. and a fellowship (PF-4435) from the American Cancer Society to R.J.C.
Communicating editor: J. Rine
- Received November 17, 2000.
- Accepted February 9, 2001.
- Copyright © 2001 by the Genetics Society of America