The ends of chromosomes in Saccharomyces cerevisiae initiate a repressive chromatin structure that spreads internally and inhibits the transcription of nearby genes, a phenomenon termed telomeric silencing. To investigate the molecular basis of this process, we carried out a genetic screen to identify genes whose overexpression disrupts telomeric silencing. We thus isolated 10 DOT genes (disruptor of telomeric silencing). Among these were genes encoding chromatin component Sir4p, DNA helicase Dna2p, ribosomal protein L32, and two proteins of unknown function, Asf1p and Ifh1p. The collection also included genes that had not previously been identified: DOT1, DOT4, DOT5, DOT6, and TLC1, which encodes the RNA template component of telomerase. With the exception of TLC1, all these genes, particularly DOT1 and DOT4, also reduced silencing at other repressed loci (HM loci and rDNA) when overexpressed. Moreover, deletion of the latter two genes weakened silencing as well, suggesting that DOT1 and DOT4 normally play important roles in gene repression. DOT1 deletion also affected telomere tract length. The function of Dot1p is not known. The sequence of Dot4p suggests that it is a ubiquitin-processing protease. Taken together, the DOT genes include both components and regulators of silent chromatin.
THE natural ends of linear eukaryotic chromosomes are made up of specialized DNA sequences and additional factors that are associated with them. The resulting macromolecular structures, called telomeres, are important in maintaining the integrity of the genome. Whereas broken chromosome ends, which lack telomeres, are commonly substrates for DNA joining, recombination, and degradation, telomeres are poor substrates for such reactions; hence, telomeres serve as protective “caps” for the DNA ends (Zakian 1996; Pryde and Louis 1997; van Steenselet al. 1998).
Telomeres not only define the physical nature of the DNA termini, but they also affect the nearby sequences that make up the distal regions of the chromosomes. In a phenomenon that is likely related to their role as protectors of the DNA ends, telomeres render these telomere-proximal domains inert, or inaccessible, relative to other regions of the genome. This protection has been observed physically, as a decreased accessibility of telomere-proximal DNA to the activity of DNA modifying enzymes expressed in vivo (Gottschling 1992). It has also been observed genetically because telomeres in Saccharomyces cerevisiae, as well as Drosophila and Trypanosoma, repress, or silence, the expression of nearby genes (Leviset al. 1985; Gottschlinget al. 1990; Zomerdijket al. 1991; Horn and Cross 1995).
The phenomenon of telomere-mediated gene silencing has been used to analyze the molecular basis of the telomere's effects on nearby DNA; the understanding that has emerged from this work is that a repressive chromatin structure initiates from the telomere and extends inward along the chromosome, rendering the enveloped DNA inaccessible to factors such as those of the transcriptional machinery (Renauldet al. 1993). Structural components of silent telomeric chromatin include the telomere sequence DNA-binding protein Rap1p, nucleosomal core histones H3 and H4, and nonhistone chromatin components Sir2p, Sir3p, and Sir4p (Aparicioet al. 1991; Kyrionet al. 1993; Thompsonet al. 1994). Current models of telomeric silencing suggest that the Sir proteins are recruited to the telomeres through their interactions with Rap1p and each other, and then “polymerize” along the unique, telomere-adjacent sequences by binding the N-terminal tails of histones H3 and H4 of the associated nucleosomes (reviewed in Grunstein 1997).
Telomeric silencing in S. cerevisiae is inherited in a semistable manner (Gottschlinget al. 1990); i.e., the repressed transcriptional state is generally present through multiple generations of a growing clonal population, but it is occasionally reversed in a stochastic manner. However, the resulting transcriptionally competent state is itself only heritable in the same limited way. When a color marker gene such as ADE2 is located near a telomere (Gottschlinget al. 1990), this switching of transcriptional states results in red-and-white sectored colonies. The switching between expression levels can be explained in part as the effect of shifts in a competition between silencing components and transcriptionalactivating factors for assembly onto telomere-proximal DNA (Aparicio and Gottschling 1994).
The fact that under normal circumstances the preexisting transcriptional state is most often inherited despite this competition indicates the existence of some mechanism to favor the status quo through the successive cell cycles. In particular, assembly (or reassembly) of the silent chromatin must occur during or shortly after each round of DNA replication. Consistent with this idea, a number of unrelated mutations or drug treatments that lengthen S phase, and presumably affect the kinetics and coordination of molecular events in S phase, are able to suppress defects in silencing (Axelrod and Rine 1991; Lamanet al. 1995). Furthermore, silencing is sensitive to mutations in subunits of chromatin assembly factor I, an activity that has been found in vitro to facilitate assembly of newly replicated DNA into nucleosomes (Enomotoet al. 1997; Kaufmanet al. 1997). Hence, there appears to be an intimate coordination between silent chromatin assembly and DNA replication.
Telomeric silencing is mechanistically similar to silencing of the nontelomeric, cryptic mating type loci HML and HMR (Laurenson and Rine 1992; Loo and Rine 1995). In fact, all the factors described above as components of the telomeric silencing apparatus are required at the HM loci, although additional factors are also needed at HML and HMR, some of which appear to be involved in the recruitment of the Sir proteins to these sites (Loo and Rine 1995; Triolo and Sternglanz 1996; Foxet al. 1997). The ribosomal RNA gene locus (rDNA/RDN1) is another region of the S. cerevisiae genome that can silence genes. However, silencing at this locus is qualitatively different than at telomeres or HM loci. For example, of the known silencing components, only the SIR2 gene product is required for rDNA silencing (Bryket al. 1997; Fritzeet al. 1997; Smith and Boeke 1997).
Given that various silent loci use common factors (such as certain Sir proteins), the pattern of silencing achieved in the cell must reflect the equilibrium reached in the competition between these loci for factor binding (see Lloydet al. 1997). For example, the distance that silent chromatin spreads internally along the chromosome from a telomere is directly related to the amount of Sir3p in the cell (Renauldet al. 1993; Hechtet al. 1996), indicating that this component of silencing in S. cerevisiae is normally in limited supply (also see Mailletet al. 1996). If a nontelomeric silencing locus developed a relative advantage in Sir3p binding, it would follow that the spread of telomeric silencing would decrease below its normal levels. Such a competitive interaction between sites would constitute a level of cellular regulation. Supporting this notion is the finding that certain genetic and physiological changes, such as aging, cause a shift in the relative abundance of silencing components between different loci (Buck and Shore 1995; Kennedyet al. 1997). Thus, to understand how silent telomeric chromatin is established or maintained, it is necessary to identify the silencing factors and to appreciate how they compete at a given locus with transcription activation components, how the assembly of silent chromatin is coordinated with DNA replication, and how limited silencing factors are distributed between different silent loci.
In a number of genetic systems, increased dosage or inappropriate expression of gene products in mutant or wild-type forms have been used in the analysis of complex biological assembly processes (reviewed in Herskowitz 1987). We have adopted these approaches to investigate telomeric silencing in S. cerevisiae. Based on the assumption that telomeric silencing is the result of a multimeric complex of factors that is assembled in a coordinated fashion, and that the assembly process might be easily disrupted by a stoichiometric imbalance of its components, we screened for gene products whose increased dosage disrupted telomeric silencing. Here we describe the genes identified in this screen.
MATERIALS AND METHODS
Yeast strains and media: S. cerevisiae strains used in this study are shown in Table 1. Strain UCC3511 was constructed in several steps. The SIR2 gene was disrupted in YPH250 by transformation with pJR531 (gift of J. Rine), followed by selection for His+ cells, thus producing UCC2666. HMRa was then disrupted by transformation with pVZ+HMRa::URA3 digested with BamHI and Sal I, followed by selection for Ura+ transformants, to produce UCC2670. UCC2670 was transformed with pJH423 (gift of R. Esposito), a SIR2-containing plasmid (YEp13, LEU2), and crossed with YPH102 (Sikorski and Hieter 1989), producing a diploid strain that was then sporulated to give UCC2675. Finally, UCC2675 was crossed with YPH499 and sporulated to give UCC3511.
UCC3532 was made by transforming YPH499 (Sikorski and Hieter 1989) with pHR10-6 (Singer and Gottschling 1994). UCC4566 was created by transforming UCC3532 with pAK4 (Huanget al. 1997) that was digested with SalI and NotI and selecting for Ura+ transformants. PPR1 was disrupted in UCC4566 using pΔPPR1::LYS2 (Renauldet al. 1993) to create UCC4567.
UCC3500 was made by transforming UCC111 (Aparicio and Gottschling 1994) with pHR10-6 (Singer and Gottschling 1994). UCC3503 was made by successively transforming YPH102 (Sikorski and Hieter 1989) with pVII-L URA3-TEL (Gottschlinget al. 1990) and pHR10-6 (Singer and Gottschling 1994). UCC3500 and UCC3503 were then crossed to create the diploid UCC3519. Finally, DOT1 was disrupted by transforming UCC3519 with pVZ28::LEU2, which was digested with SphI and XbaI, and selecting for Leu+ transformants, thus creating UCC4551. UCC4551 was then sporulated, yielding the haploid segregants UCC4554, UCC4555, UCC4560, UCC4561, UCC4562, and UCC4563.
UCC3503 was transformed with pΔPPR1::LYS2 (Renauldet al. 1993) to create UCC3504. Strains UCC3503 and UCC3505 (Singer and Gottschling 1994) were crossed to each other, and the resulting diploid was sporulated to yield UCC3537. UCC3537 and UCC3504 were crossed to create UCC3542. DOT4 was then disrupted by transforming UCC3542 with plasmid pdot4::HIS3(–), which was digested with SphI and BamHI, and selecting for His+ transformants, thus creating two independent transformants, UCC4572 and UCC4573. UCC4572 was sporulated to yield the haploid segregant UCC4602. UCC4573 was sporulated to yield the haploid segregants UCC4591, UCC4594, and UCC4595.
Strains UCC4586 and UCC4574 were made by transforming UCC3511 and UCC3515, respectively, with pdot1::HIS3(+), which was digested with SphI and XbaI. Similarly, UCC4577 and UCC4578 were created by transforming UCC3511 and UCC3515, respectively, with pdot4::HIS3(–), which was digested with SphI and BamHI. In all these transformations, His+ transformants were selected.
UCC4564 and UCC4565 were made by deleting PPR1 in UCC3511 and UCC3515, respectively, using plasmid pΔPPR1:: LYS2 (Renauldet al. 1993). Strains UCC4579 and UCC4580 were then made by transforming UCC4564 and UCC4565, respectively, with pdot4::HIS3(–), which was digested with SphI and BamHI, and selecting for His+ transformants.
To make UCC4571, UCC4567 was transformed with pdot1::HIS3(+), which was digested with SphI and XbaI. To make UCC4576, UCC4567 was transformed with pdot4:: HIS3(–), which was digested with SphI and BamHI. To make UCC3617, UCC4567 was transformed with pasf1::HIS3 digested with NotI and SalI. To make UCC6541, UCC4567 was transformed with pRS4.2 (Kimmerly and Rine 1987), which was digested with PvuII. In all cases, His+ transformants were selected.
UCC6008, UCC4583, UCC3611, and UCC6542 were made by fragment-mediated transformation of UCC3504 using DNA from plasmids pdot1::HIS3(+), pdot4::HIS3(–), pasf1::HIS3, and pRS4.2, respectively. The plasmids were digested as described above. UCC6550 was created by transforming UCC3504 with plasmid pBlu49::HIS3#1 digested with XhoI and EcoRI. UCC6552 was created by transforming UCC3504 with plasmid pBlu23::HIS3#1 digested with XhoI and NotI. In all cases, His+ transformants were selected.
UCC6555 was made by transforming BY4705 (Brachmannet al. 1998) with pVII-L URA3-TEL (Gottschlinget al. 1990), creating UCC1091, which was subsequently transformed with pΔPPR1::LYS2 (Renauldet al. 1993). To make UCC6562, UCC6555 was transformed with pasf1::HIS3 digested with NotI and Sal I, and His+ transformants were selected.
UCC3615 and UCC3612 were made by deleting ASF1 in UCC4564 and UCC3515, respectively, using pasf1::HIS3 as described above.
UCC6605, UCC6606, UCC6607, UCC6608, and UCC6609 were made by fragment-mediated transformation of JS125 (Smith and Boeke 1997) using DNA from plasmids pdot1::HIS3(+), pdot4::HIS3(–), pBlu49::HIS3#1, pBlu23:: HIS3#1, and pasf1::HIS3, respectively. The plasmids were digested as described above. To make UCC6616, JS128 (Smith and Boeke 1997) was transformed with pRS4.2 (Kimmerly and Rine 1987) digested with PvuII.
TLC1 disruptions were made in UCC3503, UCC3504, UCC3511, UCC4564, UCC3515, UCC4565, and JS125 by fragment-mediated transformation using DNA from pSD166 cut with NotI and SalI. His+ colonies were simultaneously streaked onto fresh plates with media lacking histidine, and were subjected to colony PCR to check for proper integration. Colonies from the restreak were then used in serial dilution assays and in overnight cultures for isolation of genomic DNA before telomeric shortening could cause senescence.
S. cerevisiae cultures were grown at 30° and liquid cultures were agitated at ∼200 rpm. YEPD (rich) growth medium contains 10 g yeast extract, 20 g Bacto-peptone, and 20 g glucose/liter. The synthetic (HC) media has been previously described (Adamset al. 1998). For silencing assays on galactose-containing media, colonies were pregrown on 3% galactose medium for 4 days and then resuspended in water. Tenfold serial dilutions were then plated onto 3% galactose medium lacking or containing uracil, and the cells were again incubated for 4 days before the colonies were counted. Transformations were carried out according to a standard lithium acetate procedure (Ausubelet al. 1995). All 6-azauracil (6-AU)-containing media were made from a 2 g/liter filter-sterilized stock added after the media had been autoclaved and cooled to ∼60°.
Transformation of the pTRP library into UCC3505: UCC3505 cells were pregrown in rich (2% glucose) medium and transformed with the pTRP library DNA (Singer and Gottschling 1994). Transformants were plated onto synthetic medium containing 2% glucose and lacking tryptophan to select for cells that had been transformed with the TRP1-bearing plasmids. After 6 days of growth, the colonies were replica plated onto synthetic medium containing 3% galactose (to induce strong transcription from the GAL1 promoter of the pTRP vector) and lacking tryptophan (HC-trp 3% galactose). After 3 days of growth on HC-trp (3% galactose) medium, the colonies were replica plated onto synthetic medium containing 3% galactose and lacking tryptophan and uracil (to select for strains in which there had been a derepression of the telomeric URA3 gene). The Ura+ colonies were then restreaked onto the same medium and the color of individual colonies was inspected. White Ura+ colonies were then checked for their phenotypes on medium containing 2% glucose. Only those colonies that were Ura– and red/white sectoring when grown on glucose medium were retained. Plasmid DNA was isolated from each of these transformants (Hoffman and Winston 1987) and reintroduced into UCC3505 to confirm that the galactose-dependent loss of silencing was indeed plasmid linked.
6-AU assay: pTRP plasmid-bearing strains to be tested for 6-AU sensitivity were pregrown on solid HC-trp + 3% galactose medium for 4 days at 30°, and then serial dilutions were plated onto HC-trp-ura + 3% galactose medium containing 6-AU at concentrations ranging from 1 to 5 μg/ml. Strains to be tested that were not carrying plasmids were pregrown on YEPD (rich) medium for 3 days at 30°, and then serial dilutions were plated on HC-ura plates that contained 10, 20, or 30 μg/ml 6-AU.
Plasmid constructions: pVZ+HMRa::URA3 was constructed in multiple steps. pFATRS303 was constructed by cloning an XbaI fragment from pFAT10 (Runge and Zakian 1989) into the AatII site of pRS303 (Sikorski and Hieter 1989). pHMRalacZ (a gift from M. Hochstrasser) was digested with BglII and religated to remove the lacZ gene and produce pHMRa. A PstI-EcoRI fragment (4 kbp) from pHMRa was cloned into pFATRS303 that was digested with SmaI to give pFATRS303HMRa2. A SalI-BamHI fragment (4 kbp) containing HMRa was then ligated to pVZ1 (Henikoff and Eghtedarzadeh 1987) that was digested with SalI and BamHI. Finally, a BamHI fragment (1.1 kbp) containing URA3 from pM20 (a gift from R. Schiestl) was cloned into the BglII site of pVZ1+HMRa to produce pVZ+HMRa::URA3.
pVZ28 was made by partially digesting pTRP28 with XhoI and ligating the 1.9-kb fragment containing DOT1 into the Sal I site of pVZ1. pVZ28::LEU2 was made by blunt-end ligation of a BamHI fragment from YDp-L (Berbenet al. 1991) into pVZ28 digested with Afl II and XhoI. pdot1::HIS3(+) was made by blunt-end ligation of a BamHI fragment from YDp-H (Berbenet al. 1991) into pVZ28 digested with AflII and XhoI such that the HIS3 and DOT1 genes had the same transcriptional orientation.
pVZDOT4 was made by blunt-end ligation of an XhoI fragment containing DOT4 from pTRP50 into pVZ1(-H3) (the HindIII site was destroyed by digestion with HindIII, followed by T4 polymerase treatment and blunt-end ligation) digested with PstI/HincII. pdot4::HIS3(–) was made by blunt-end ligation of a BamHI fragment from YDp-H (Berbenet al. 1991) into pVZDOT4 that was digested with HindIII and NcoI such that the HIS3 and DOT4 genes had the same transcriptional orientation.
ASF1 was cloned as a 1.5-kb PCR product from an amplification from YPH499 (Sikorski and Hieter 1989) genomic DNA. The primers used in the reaction were ASF#1: 5′-CGG GATCCTTGGCGAGAATTTCGATTTTCAGG-3′ and ASF#2: 5′-GACTAGTGTGTTTTATGAACTTTTAGGATGACGTATT G-3′. The PCR reaction used 28 pmol of each primer in a 100-μl reaction, which also included 1× Taq buffer (Promega, Madison, WI), 0.2 mm dNTPs, 2 mm MgCl2, Taq enzyme (Promega), and YPH499 (Sikorski and Hieter 1989) genomic DNA from ∼107 cells. The PCR product was digested with BamHI and SpeI, and ligated to pBluescript II KS– (Stratagene, La Jolla, CA) that was digested with BamHI and XbaI, thus constructing pBlueASF1. To construct pasf1::HIS3, a 1.2-kb HIS3-containing BamHI fragment from YDp-H (Berbenet al. 1991) was used to replace a 0.9-kb SnaBI-NdeI fragment of pBlueASF1 through a blunt-end ligation.
pBlu49 was constructed by ligating the 0.8-kb XhoI fragment containing DOT5 from pTRP49 into the SalI site of pBluescript II KS–. To make pBlu49::HIS3#1, the 1.2-kb BamHI fragment containing HIS3 from YDP-H (Berbenet al. 1991) was used to replace the 0.5-kb PflMI-SnaBI internal DOT5 fragment in pBlu49.
pBlu23 was constructed by ligating the 2.1-kb XhoI fragment containing DOT6 from pTRP23 into the SalI site of pBluescript II KS– (Stratagene). To make pBlu23::HIS3#1, the 1.2-kb BamHI fragment containing HIS3 from YDP-H (Berbenet al. 1991) was used to replace the 0.8-kb MluI-AflII DOT6 fragment in pBlu23.
pRS313/Y′RsaI was constructed by ligating the 350-bp RsaI Y′ sequence-containing fragment from pY′ARS into the SmaI site of pRS313 (Sikorski and Hieter 1989). pY′ARS (made by Jeff Stevenson) was constructed by cloning the 1-kb XhoI/SphI fragment from pYP1-L2 (gift from E. Louis) into vector pVZ1, which had been digested with SalI and SphI.
YTCA-1 differs from YTCA-2 (Gottschlinget al. 1990) only in that the 125-bp HaeIII-MnlI fragment is present in reversed orientation.
pSD166 was constructed in multiple steps: A 3.9-kb EcoRI fragment containing TLC1 was excised from pAZ1 (Beeleret al. 1994) and inserted into the EcoRI site of pRS424 (Christiansonet al. 1992) to create p424/TLC1g. From p424/TLC1g, a 4-kbp NotI/SalI fragment was cloned into NotI/SalIdigested pRS425 (Christiansonet al. 1992), creating pSD141. pSD141 was then cut with BglII and NdeI and transformed into UCC3586, and the gap-repaired plasmid was recovered, creating pSD143, a plasmid containing the tlc1::HIS3 disruption. A NotI/SalI fragment from pSD143 was next cloned into NotI/SalI-digested pVZ1, creating pSD166.
Analysis of nucleic acids: Methods for DNA preparation and analysis have been described previously (Hoffman and Winston 1987; Gottschlinget al. 1990). DNA sequencing was carried out using the Taq DyeDeoxy Terminator Cycle Sequencing Kit from Applied Biosystems (Foster City, CA) according to the manufacturer's instructions. To sequence the ends of the DOT cDNA clones, GAL1 and CYC1 primers were used (GAL1, 5′-CCTCTATACTTTAACGTCAAGGAG; CYC1, 5′-GAAAAGGGGCCTGTTTACTCA CAG).
DNA blot hybridization analyses and probe synthesis were carried out using the Genius system from Boehringer Mannheim (Indianapolis, IN) following the manufacturer's instructions. Probes for the Southern analysis were synthesized by PCR: 1 μl miniprep template DNA, 20 pmol of each primer, 1.5 μl 3 m KCl, 10 μl 25 mm MgCl2, 1 μl 1 m Tris, pH 8.5, 10 μl 10× dig PCR mix (2 mm dGTP, 2 mm dATP, 2 mm dCTP, 1mm dTTP, and 0.5 mm digoxigenin-11-dUTP), and 1 μl Taq enzyme were combined with water to bring the final volume to 100 μl. The reaction was then exposed to the following program of conditions: (1) 94° 5 min, (2) 94° 30 sec, (3) 50° 1 min, (4) 70° 2 min, and (5) 34 more repetitions of steps 2–4. To make the Y′ probe, T3 (5′-AGCGCGCAATTAACCCT CACTAAAG-3′) and T7 (5′-CGTAATACGACTCACTATAG GG-3′) primers were used in conjunction with plasmid pRS313/Y′RsaI template DNA. To make the TG1-3 probe, M13 forward (5′-TGTAAAACGACGGCCAGT-3′) and reverse (5′-CAGGAAACAGCTATGACC-3′) primers were used in conjunction with YTCA-1 template DNA.
All gene disruptions were confirmed either by DNA blot hybridization analyses or by colony PCR (Adamset al. 1998).
A screen for overexpressed cDNAs that disrupt telomeric silencing: A S. cerevisiae strain was constructed (UCC3505) that provided an easy yet stringent assay for loss of telomeric silencing. UCC3505 has the ADE2 gene located adjacent to the right telomere of chromosome V (V-R) and URA3 next to the left telomere of chromosome VII (VII-L). Both of these genes are sensitive to telomeric silencing, and the combination provides a two-level filter for screening perturbations of telomeric silencing. This two-level screen inherently excludes gene-specific alterations, such as induction of URA3 or ADE2, as well as single-telomere events, such as chromosomal rearrangements that move the marker gene (URA3 or ADE2) away from the chromosome end [e.g., spontaneous insertion of Y′ DNA elements between the marker gene and the telomere (Singer 1997)].
Telomeric silencing in UCC3505 was monitored using simple phenotypic assays of ADE2 and URA3 expression. Normally, colonies expressing ADE2 are white, while those not expressing it are red (Roman 1956). Because of the epigenetic nature of telomeric silencing, strains with ADE2 near a telomere give rise to genetically identical but phenotypically distinct clonal populations that are visible as red and white sectors within a single colony (Gottschlinget al. 1990). The URA3 gene located at a telomere also normally switches between transcriptional states (Gottschlinget al. 1990). However, URA3 gene expression in UCC3505 was weakened by deleting its transcriptional activator gene, PPR1, which caused the telomere-adjacent URA3 gene to be completely silenced; the cells were thus unable to grow in the absence of uracil (Ura–; Aparicio and Gottschling 1994).
To identify genes or gene fragments whose overexpression interferes with telomeric silencing, UCC3505 was transformed with a high-expression S. cerevisiae cDNA library. The expression of cDNA inserts in this library was controlled by the GAL1 promoter, which is strongly induced by the presence of galactose in the medium (Johnston and Davis 1984; Elledgeet al. 1991; Rameret al. 1992). Of the 330,000 yeast transformants obtained, 48 displayed a plasmid- and galactose-dependent decrease in telomeric silencing. That is, when grown on medium containing galactose, the cells were able to grow in the absence of uracil (Ura+) and gave rise to predominantly white colonies (Ade+; Figure 1).
Identification and sequence analysis of the DOT cDNAs: On the basis of Southern analysis and DNA sequencing, we determined that these 48 clones represented 10 independent genes, which we refer to as the DOT (disruptor of telomeric silencing) genes. Of these genes, 5 had been identified previously: SIR4, ASF1, DNA2, RPL32, and IFH1. Four of the remaining genes are referred to as DOT1, DOT4, DOT5, and DOT6. We compared the sequences of our isolates of these genes to the genomic sequences that were generated by the S. cerevisiae sequencing project to assess the completeness of each of the clones. The final gene, which we named telomerase component 1 (TLC1), has been described elsewhere (Singer and Gottschling 1994).
Out of this collection, overexpression of two of the genes, SIR4 and ASF1, was previously known to interfere with telomeric silencing (Marshallet al. 1987; Renauldet al. 1993; Cockellet al. 1995; Leet al. 1997). The fact that these two genes were also isolated in our screen reassured us of its efficacy.
Sir4p, a component of silent chromatin, is required for telomeric silencing (Aparicioet al. 1991). Eleven SIR4-containing plasmids were isolated, representing at least six independent clones. Only the C-terminal portion of the gene was present in these clones, consistent with earlier findings that overexpression of this region of Sir4p strongly interferes with silencing (Table 2; Cockellet al. 1995). Surprisingly, two of the plasmids (pTRP4 and pTRP58) had reversed inserts, suggesting that the GAL1 promoter on the vector directed transcription of antisense RNA.
The role of ASF1 in silencing is less clear. Nevertheless, 13 ASF1-containing plasmids were isolated in our screen, representing at least seven independent clones (Table 2). All contained the entire open reading frame (ORF) of the gene. One of the clones, pTRP30, was a fusion of RNA sequences from ASF1 and ∼75 nt in the 3′ region of the SUM1 RNA, including the last 7 nt of the SUM1 ORF. Coincidentally, SUM1 encodes a nuclear protein of unknown function that has been implicated in silencing (Klaret al. 1985; Liviet al. 1990; Laurenson and Rine 1991; Chi and Shore 1996).
Gene fragments encoding the N-terminal third of Dna2p were isolated twice in our screen (Table 2). DNA2 is an essential gene that encodes a 3′–5′ DNA helicase required during DNA replication (Budd and Campbell 1995; Buddet al. 1995). Dna2p is 1522 amino acids in length, and its helicase motifs are all in the C-terminal half.
L32 is an essential ribosomal protein (Dabeva and Warner 1987). In our screen, its gene was identified in two plasmids, both of which included the entire open reading frame of RPL32, without its genome-encoded intron (Table 2). In one clone, pTRP54, the RPL32 ORF was followed by the coding region of another ribosomal protein, S24.
The IFH1 gene was originally isolated as a high-copy suppressor of a null allele of FHL1, a gene required for rRNA processing (Hermann-Le Denmatet al. 1994; Cherel and Thuriaux 1995). The function of IFH1, which has a predicted ORF of 1085 amino acids, is unknown. In our study, IFH1-containing clones were isolated six times, with only two of the clones being identical. Each clone encoded only the N-terminal portion of Ifh1p, terminating at residues 212, 213, 216, or 218 (Table 2).
The sole DOT1 cDNA isolated was 1882 bp (pTRP28, Table 2) and found to contain an ORF encoding the entire 582-amino-acid-predicted protein. The predicted sequence of this protein suggests that it is hydrophilic and basic (pI = 9.03; charge at pH 7 = +17.29).
DOT4 was isolated once in our screen and the gene was predicted to encode a ubiquitin-specific hydrolase (reviewed in Hochstrasser 1996). The pTRP50 cDNA insert encoded a 789-amino-acid protein containing nearly all (residues 16–784) of the predicted 792-aminoacid Dot4 protein (Table 2). However, several minor differences exist between the GenBank Dot4p sequence and the predicted sequence of the cDNA-encoded protein (see Table 2).
DOT5 was isolated as the clone pTRP49, whose 807bp insert included the entire ORF encoding a predicted protein of 215 amino acids.
The DOT6 gene sequence predicts a 670-amino-acid protein with a single Myb-related motif between residues 78 and 116. In different proteins from a wide variety of eukaryotes, the Myb domain is involved in sequencespecific DNA binding (Lipsick 1996). Of the two DOT6 cDNAs isolated, one clone (pTRP29) included a 5′ untranslated region and the first 286 amino acids of the Dot6p predicted protein. The second clone (pTRP23) lacked the extreme N terminus, but encoded the C-terminal 634 amino acids of the protein.
Effects of overexpressing the DOT cDNAs on telomeric silencing: To characterize phenotypes associated with overexpression, one clone was chosen as a representative from each of the 10 genes. Each selected clone had the strongest effect of disrupting telomeric silencing within its gene group.
The effect that overexpression of the clones had on telomeric silencing was quantitatively evaluated. Each representative plasmid was retransformed into UCC 3505. The transformants were pregrown on selective medium containing galactose to induce transcription of the cDNAs, and then assayed for derepression of the telomeric URA3 and ADE2 genes. Expression of URA3 was measured as the viability of the transformed strain on medium lacking uracil (Figure 1A). Transcription of the ADE2 gene was assessed qualitatively in terms of colony color, with white sectors reflecting ADE2 expression, and red sectors representing ADE2 repression (Figure 1B). The results from these two assays of telomeric silencing were consistent in all cases. Overexpression of clones containing SIR4, DOT1, TLC1, ASF1, DNA2, and DOT4 each had a strong effect of disrupting telomeric silencing. There was a ≥1000-fold increase in the ability of plasmid-bearing strains to grow in the absence of uracil compared to a strain carrying vector without a cDNA insert. Similarly, colonies of strains that contained these overexpressed cDNAs had more prominent white sectors than those seen with the strain carrying vector alone. Overexpression of DOT5, DOT6, IFH1, and RPL32 had a weaker but still significant ability to interfere with telomeric silencing; smaller fractions of the colonies were white, and there was only a 20–400-fold increase in the ability of the plasmid-bearing strain to grow on medium lacking uracil.
Effects of overexpressing the DOT cDNAs on HM silencing: To determine whether the action of each gene was limited to telomeric silencing, the effect of overexpression on nontelomeric silenced loci HML and HMR was assayed. The 10 cDNA overexpression plasmids were transformed into yeast strains that had the URA3 gene inserted into the HML or HMR locus. (URA3 is silenced much better at HML than at HMR because of the difference in the way the gene was inserted within the two HM loci. Thus, silencing at the HMR locus was more sensitive to perturbation than silencing at HML.) These transformed strains were pregrown on galactose, and derepression of URA3 was measured as the viability of the transformed strains on medium lacking uracil (Figure 2, A and B). Consistent with previously published data, SIR4 and ASF1 overexpression derepressed both these silenced loci, causing the strains to have ∼100% viability on media lacking uracil (Renauldet al. 1993; Cockellet al. 1995; Leet al. 1997). Similarly, overexpression of clones containing DOT1, DOT4, and IFH1 caused a dramatic decrease in silencing at the HML and HMR loci. RPL32 overexpression had a weak effect at both loci. DNA2, DOT5, and DOT6 had a weak effect at HML and a stronger effect at HMR. Finally, TLC1 had no effect at HML and a weak effect at HMR. [This weak effect was highly variable between transformants (see Figure 2B).] Thus, TLC1 had a primarily telomere-specific effect, while the rest of the genes affected silencing both at telomeres and the HM loci, though they varied widely in the potency of their effects.
Effects of overexpressing the DOT cDNAs on silencing within the rDNA locus: Silencing within the tandemly duplicated repeats of rDNA is qualitatively different than at telomeres and the HM loci (see Introduction). Therefore, the effects of DOT cDNA overexpression were also examined at the rDNA, using strains having a single URA3 gene inserted within the RDN1 locus. Changes in URA3 expression were assessed by examining both the strains' abilities to grow on media lacking uracil and their growth on medium containing 5-fluoro-orotic acid (5-FOA), which is converted to a toxic compound by the URA3 gene product (Boekeet al. 1987; Gottschlinget al. 1990). Analyzing URA3 expression through the combination of both these assays provided a greater range of sensitivity for the degree of rDNA silencing.
The effects of overexpressing the DOT cDNAs on rDNA silencing can be divided into four classes (Figure 2C). Overexpression of DNA2, ASF1, DOT4, DOT6, and IFH1 modestly reduced rDNA silencing, causing increases in sensitivity to 5-FOA and a commensurate increase in growth on media lacking uracil. Overexpression of DOT1 also caused a loss of rDNA silencing, as indicated by a significant increase in 5-FOA sensitivity. Curiously, there was no corresponding increase in growth on media lacking uracil. This may indicate that overexpression of DOT1 causes a higher rate of switching between repressed and active states, with the average fraction of active cells in the population remaining constant. TLC1, DOT5, and RPL32 had essentially no effect on rDNA silencing. In contrast to its effect at telomeres and the HM loci, the data suggest that SIR4 overexpression, if anything, caused a subtle increase in rDNA silencing (note the lower level of growth in the second from the left spot on the plate labeled –Uracil in Figure 2C).
Effects of overexpressing the DOT cDNAs on a nonsilenced gene: For those cases in which overexpression of a DOT cDNA disrupted silencing at all four loci, it was possible that the apparent increased expression was not caused by a defect in silencing, but by an unrelated mechanism, such as active induction of the marker genes or stabilization of their protein products. To determine whether the effects of the DOT cDNAs were restricted to derepressing genes in silenced loci, we examined their action on a nonsilenced URA3 gene. We used an assay employing 6-AU, a competitive inhibitor of the URA3-encoded enzyme orotidine 5′-phosphate decarboxylase (Loisonet al. 1980). 6-AU is readily taken up by yeast from the medium. Hence, the ability of a strain to grow in the absence of exogenously provided uracil and in the presence of 6-AU reflects the level of in vivo activity of the URA3-encoded enzyme. That is, the greater the cellular levels of active URA3 gene product, the greater the resistance of the strain to 6-AU. We used this assay because it provides greater sensitivity to changes in URA3 expression than measuring message levels by Northern analysis (Losson and Lacroute 1981; Aparicio and Gottschling 1994).
The assay was carried out in a strain with the URA3 gene inserted into ADH4, a nonsilenced locus on chromosome VII. Under conditions in which expression of the DOT cDNA was induced (galactose-containing medium), the ability of the strain to grow in the presence of 6-AU was tested (Figure 3). Two positive controls were included in the analysis, the overexpression of URA5 and PPR1. Cells overexpressing the URA5 gene product, which creates the normal substrate for orotidine 5′-phosphate decarboxylase (Ura3p), had improved growth on this medium compared to cells with vector alone (Figure 3). Also, cells overexpressing Ppr1p, which induces URA3 transcription, were resistant to the effects of the 6-AU (Figure 3; Losson and Lacroute 1981). In contrast, none of the overexpressed DOT cDNAs, with the possible exception of DOT6, caused a significant improvement in growth of the transformed strain on 6-AU medium, compared to the strain carrying empty vector. Thus, the improvement of marker gene (URA3) expression caused by the high levels of each DOT cDNA, except perhaps DOT6, appeared to occur through a defect in silencing.
Effect of DOT cDNA overexpression on telomeric DNA tract length: An effect on telomeric silencing may well be accompanied by effects on other aspects of telomere structure or metabolism. For example, overexpression of TLC1, the telomerase RNA gene, causes the telomere DNA tract at the end of the chromosome to shorten (Singer and Gottschling 1994). To see if this was true for the other DOT cDNAs, each representative clone was overexpressed and telomere DNA length was measured. Because changes in telomere length can take many generations to manifest themselves (Lustig and Petes 1986), plasmid-bearing transformants of UCC 3505 were cultured on galactose medium for ∼100 generations before genomic DNA was collected for Southern analysis of telomere length (Figure 4). As expected, the typical heterogeneity in telomere length of a population of cells was observed, even when examining a unique chromosome end. Only TLC1 and SIR4 overexpression had significant and reproducible effects of shortening telomere length, consistent with earlier reports for TLC1 (Singer and Gottschling 1994). In contrast, DOT5 overexpression caused a modest increase in the average telomere length. Thus, overexpression of only three DOT genes, TLC1, SIR4, and DOT5, affected telomere length regulation and telomeric silencing.
Effects of deleting the DOT genes on telomeric silencing: When overexpression of a gene or gene fragment interferes with a biological process or structure, such as the one described here for telomeric silencing, it may be because the wild-type gene product normally participates in the process or is a component of the structure. The overexpressed gene product may act at an inappropriate time or place and, thus, interact with its partner protein(s) to create a futile complex that interferes with the normal cellular process. For example, both deletion and overexpression of SIR4 result in the same phenotype, loss of silencing (Ivyet al. 1986; Marshallet al. 1987; Aparicioet al. 1991). To ascertain whether any of the other DOT genes are important for telomeric silencing, the genomic copy of DOT1, DOT4, DOT5, DOT6, TLC1, or ASF1 was deleted and the effect on silencing was examined. This analysis could not be done, however, for RPL32, IFH1,or DNA2, which are essential for viability (Dabeva and Warner 1987; Budd and Campbell 1995; Cherel and Thuriaux 1995; data not shown).
The gene deletions were made in strains in which telomeres V-R and VII-L were labeled with ADE2 and URA3, respectively. In addition, two versions of these strains were made, one that was wild type for PPR1, the transcriptional activator that is responsible for URA3's inducible transcription, and one that was mutant (ppr1). We analyzed telomeric URA3 expression, as reflected by growth on media lacking uracil and resistance to 5-FOA, in both PPR1 and ppr1 cells to provide a greater range of phenotypic sensitivity to differences between wild-type and dot strains.
Comparing dot1 and DOT1 strains, the DOT1 gene product was found to be important for telomeric silencing (Figure 5A). In DOT1 PPR1 strains, the telomeric URA3 gene was silenced in a large fraction of the cells, as evidenced by the high frequency of growth on medium containing 5-FOA. However, when DOT1 was deleted, the resistance of the strain to 5-FOA declined by ∼105-fold. Similarly, in strains in which telomeric silencing of the URA3 gene was made stronger by the absence of PPR1, the deletion of DOT1 still reversed this repression, resulting in an ∼1000-fold increase in viability of the strain on medium lacking uracil (Figure 5A). The loss of telomeric repression in dot1 strains also occurred for the ADE2 gene located at telomere V-R. Whereas wildtype colonies had prominent red sectors representing cells in the population in which the ADE2 gene was silenced, the colonies of dot1 strains were almost completely white (data not shown).
Disrupting DOT4 also had a strong effect on telomeric silencing. The colonies of dot4 strains were less red than their wild-type counterparts, consistent with a decreased repression of the telomeric ADE2 gene (data not shown). Moreover, there was a 1000-fold increase in the ability of ppr1 strains to grow in the absence of uracil when DOT4 was deleted, suggesting a decrease in silencing of the telomeric URA3 gene (Figure 5B). However, there was no increased 5-FOA sensitivity in dot4 PPR1 strains. At present, it is difficult to interpret the significance of this difference because there are pleiotropic defects in dot4 strains, including slowed growth (note colony size in Figure 5B). It is possible that the dot4 mutation may also affect 5-FOA utilization or uptake.
Examining telomeric silencing in a TLC1 deletion strain presents an unusual circumstance in the analysis. As a result of losing TLC1 function and, consequently, telomerase activity, the (TG1-3) DNA tracts at the ends of the chromosomes shorten with each cell division. Therefore, the level of silencing URA3 at the VII-L telomere was determined in a population of cells while the average length of their terminal (TG1-3) repeats was examined. As can be seen in Figure 5C, a population of cells with an average VII-L telomeric tract that is about half the length of wild-type cells still silences the telomeric gene very efficiently. Thus, TLC1 is not directly required for telomeric silencing. These results also demonstrate that telomeric DNA tracts as short as 180 bp can efficiently silence genes.
Deletion of DOT5 or DOT6 had no detectable effects on telomeric silencing (data not shown). However, disruption of ASF1, which also caused cells to be slow growing, resulted in a modest telomeric silencing defect (Figure 5D).
Taken together, we conclude that the DOT1 and DOT4 gene products are important factors for telomeric silencing while ASF1 may play a minor role. TLC1 does not appear to play a direct role in telomeric silencing. Also, the DOT5 and DOT6 gene products are not required for telomeric silencing.
Effects of deleting the DOT genes on HM silencing: To determine whether the DOT genes were important for silencing at the HML and HMR loci, a test similar to the one described above for telomeric silencing was conducted. Strains in which the URA3 gene was inserted into either HML or HMR were constructed, and the genomic copy of DOT1, DOT4, DOT5, DOT6, TLC1, or ASF1 was deleted. These strains were compared for their ability to grow on media lacking uracil and their resistance to 5-FOA.
Consistent with the effect at telomeres, deletion of DOT1 caused a decrease in silencing at the HML and HMR loci (Figure 6A). In the wild-type strain, silencing of URA3 at the HMR locus resulted in a high level of 5-FOA resistance. When the DOT1 gene was deleted, however, the ability of the strain to grow in the presence of 5-FOA decreased dramatically, indicating increased expression of the URA3 marker gene. In DOT1 strains in which the URA3 gene was located at HML, the repression of URA3 caused poor viability on media lacking uracil. However, when DOT1 was deleted, plating efficiency on media lacking uracil was increased.
DOT4 was also found to be involved in silencing at both HM loci (Figure 6B). As with the strains in which URA3 was located at a telomere, decreased silencing at HML and HMR in the dot4 mutants was observed as increased viability on media lacking uracil compared to wild-type strains.
Even though they had shorter telomeres, strains without TLC1 showed no change in silencing at HML or HMR compared to strains with TLC1 (Figure 6C). As was true for telomeric silencing, deletion of DOT5 and DOT6 had no detectable effect on silencing at HML and HMR (data not shown), and deletion of ASF1 caused a weak derepression at both HML and at HMR (Figure 6D). This weak effect was not reported in earlier work on ASF1 and may reflect a difference in the assays used (Leet al. 1997).
Effects of deleting the DOT genes on rDNA silencing: Deletion of the DOT genes had a somewhat different spectrum of effects on rDNA silencing than on the telomeric or HM loci (Figure 7). While dot4 cells had slightly less rDNA silencing than wild-type cells, as judged by sensitivity to 5-FOA, deletion of DOT1, DOT5, DOT6, and TLC1 had no effect. Deletion of ASF1 and SIR4 resulted in a subtle increase of rDNA silencing, as judged by decreased growth in the absence of uracil. This subtle change in rDNA silencing when SIR4 was deleted is consistent with an earlier study (Smith and Boeke 1997).
Effects of deleting the DOT genes at a nonsilenced locus: To have a clearer understanding of the effects of DOT gene deletions on silencing of URA3 at telomeres, HML, HMR, and within the rDNA cluster, we examined the expression of URA3 at a nonsilenced locus in a set of dot– strains. To assay the expression of this URA3 marker, the ability of the strain to grow in the presence of 6-AU was measured. As mentioned above, cells in which URA3 expression is increased are better able to grow in the presence of 6-AU. Under the conditions chosen for this assay, the parental (ppr1) strain grows poorly on plates containing 6-AU (Figure 8). If URA3 transcription is improved by the presence of the PPR1 gene product, resistance to 6-AU rises sharply. As expected, deletion of the SIR4 gene, a recognized component of silencing chromatin, had no effect on the expression of the nonsilenced URA3 gene. Similarly, deletion of DOT1 had no effect. ASF1, which was observed to have very weak effects at telomeres and HML, also failed to discernibly improve the resistance of the cells to 6-AU. In contrast, deletion of DOT4 caused a significant resistance to the presence of 6-AU. This result may represent improved transcription of the unsilenced URA3 in dot4 strains, or (as with the 5-FOA experiments described above) it may reflect an unrelated mechanism of 6-AU resistance, such as a defect in uptake of the 6-AU compound.
Effects of deleting the DOT genes on telomeric DNA tract length: Finally, to assess whether the DOT genes have a role in maintaining normal telomeric DNA structure, telomere length was measured using a TG1-3 probe that detected all telomeres in the cell in strains deleted for one of the nonessential genes isolated in the screen [DOT1, DOT4, DOT5, DOT6, ASF1, and SIR4 (TLC1 results were published earlier in Singer and Gottschling 1994)]. Deletion of SIR4 caused a modest telomere length decrease (Figure 9), as had been reported earlier (Palladinoet al. 1993). The only other reproducible difference was a result of deleting DOT1; dot1 cell telomeres were somewhat more heterogeneous in length than wild-type cells.
We have identified a group of 10 genes involved in telomeric silencing, based on the ability of either a fulllength or partial cDNA clone of each gene to disrupt telomeric silencing when overexpressed. The DOT genes include two genes that had previously been known to disrupt telomeric silencing when overexpressed: silent chromatin component SIR4 as well as ASF1, whose role in silencing is not known. In an earlier report, we described the defect in telomeric silencing when the telomerase RNA template gene TLC1 is overexpressed (Singer and Gottschling 1994). The remaining 7 genes have not been reported previously as having an effect on telomeric silencing: ribosomal protein gene RPL32, DNA helicase gene DNA2, IFH1, and the newly identified genes DOT1, DOT4, DOT5, and DOT6.
Overexpression of a subset of the DOT genes also altered silencing at HML, HMR, and the RDN1 locus. All of the genes except for TLC1 reduced silencing at the HM loci, and all except TLC1 and RPL32 reduced rDNA silencing. SIR4 overexpression resulted in a very mild increase in silencing at RDN1. These different effects of the DOT genes reflect the qualitative similarities and differences between the four silencing loci.
To model how overexpression of the DOT genes might disrupt telomeric silencing, it is worth reviewing a few aspects of our current understanding of silent telomeric chromatin.
The ability to silence a gene requires the coordinated assembly of a complex set of molecules (e.g., histones, Rap1p, Sir3p, Sir4p, etc.) onto a scaffold of telomereproximal DNA. To achieve silencing, this assembly must occur in the face of a challenge by transcriptional machinery attempting to assemble onto the same DNA scaffold. It must also be reproduced every time the chromosome is duplicated.
Just as assembly of the Sir proteins is a requisite for establishing and maintaining silent chromatin, duplication of silent chromatin likely requires that it be transiently remodeled or taken apart. For instance, silent chromatin components may be modified in coordination with DNA replication to permit passage of the replication fork (Bradbury 1992; Itoet al. 1997), or the replication machinery may have an intrinsic ability to dissociate silent chromatin as it polymerizes new DNA strands along the chromosome (Bonne-Andreaet al. 1990).
Silent chromatin is limited to a subset of loci in the yeast genome, yet some silent chromatin components, such as histones H3 and H4, are present along the entire chromosome, and others, such as Rap1p, are present at a multitude of nonsilenced loci (Shore 1994). The mechanism(s) by which specificity for silencing is imparted upon these nonspecific proteins is not clear. Moreover, there are multiple silencing loci–telomeres, HML, HMR, RDN1–that are distinct physically and structurally. Coordination between these different loci must be achieved to maintain the appropriate level of silencing at each locus.
Given these challenges to its formation, it is not surprising that telomeric silencing is semistable (Gottschlinget al. 1990). Moreover, perturbations of any one of the many conditions required for silent telomeric chromatin formation could easily shift the balance in favor of silencing disruption. This would lead to the strong phenotype of gene expression that we selected in our screen. Perturbations that may arise from overexpressed cDNAs may have a dominant positive or negative effect with respect to their normal gene function. As such, they may disrupt silencing if their gene products normally participate in the process of disassembly or assembly of any silent locus. By the model mentioned earlier, in which a defective or overabundant subunit may poison an entire complex, the DOT genes may themselves encode part of the silencing structure. Alternatively, they may include genes whose products are not intimately associated with silent telomeric chromatin, but, rather, affect its assembly in a more indirect way, such as modulating the synthesis or turnover of silent chromatin components.
Another possibility is that overabundance of a truncated or full-length gene product may cause it to associate with a new set of molecules and, thus, involve it in telomeric silencing even though it normally has no role in this process. These new interactions may result from the production of a gene product that is improperly regulated (because of the production of an incomplete gene product or an abnormally high level of synthesis) or from the sheer excess of the overproduced protein, which increases the frequency of a low-affinity interaction. According to this model, the illegitimate interactions that result from this cross-reaction would preclude normal productive interactions and, thus, disrupt the formation of silencing chromatin at the telomere.
In light of these ideas, we offer speculation about each of the DOT genes and why they were identified in our screen.
SIR4: It had previously been observed that overexpression of the entire SIR4 gene or overproduction of just the C-terminal region of the protein results in a loss of silencing at HM loci and telomeres (Ivyet al. 1986; Marshallet al. 1987; Renauldet al. 1993). Hence, our identification of SIR4 (Table 3) served primarily as a positive control for our screen.
The mechanism by which SIR4 overexpression disrupts silencing has been studied by others. Sir4p interacts via its C-terminal region with Sir3p (Morettiet al. 1994; Strahl-Bolsingeret al. 1997), suggesting that overexpression of SIR4 may titrate Sir3p, a required silencing factor, away from chromatin. This model is supported by the finding that overexpression of SIR3 suppresses the loss of silencing caused by overexpression of SIR4 (Marshallet al. 1987).
A protein interaction model might not apply to all the SIR4 clones we identified. Surprisingly, pTRP4 and pTRP58 were inserted into the expression vector in the reverse orientation with respect to the GAL1-promoter such that they could produce RNA that is antisense to SIR4 sequence. We speculate that such an antisense RNA reduces the level of Sir4p in the cell. It is noteworthy that while antisense technology works well in many organisms, our results represent one of the rare cases in which antisense RNA expression produces a phenotype in S. cerevisiae (Nasret al. 1995; Khoet al. 1997; Machadoet al. 1997).
ASF1: Overexpression of ASF1 strongly derepressed both telomeric and HM loci and had a significant effect on the rDNA locus (Table 3; Leet al. 1997). As has been suggested by others, the presence of acidic stretches within Asf1p and the upregulation of its gene before and during S phase suggest that it is involved in replication or chromatin assembly (or disassembly) (Leet al. 1997). Regardless of how ASF1 may normally function, it is likely that the loss of cell cycle regulation when its cDNA is under GAL1-directed expression plays an important role in the disruption of silencing.
IFH1: IFH1 is an essential gene whose connection to silencing has not been recognized previously (Cherel and Thuriaux 1995). However, clones encoding the N-terminal region of Ifh1p were isolated six times in our screen. Overexpression of this region weakened silencing not only at the telomeres, but also at the HM loci and rDNA (Table 3).
The highly acidic domain in the N terminus of IFH1, like that found in ASF1, may mediate interaction with chromatin proteins. Because IFH1 is proposed to normally interact with FHL1 (Hermann-Le Denmatet al. 1994; Cherel and Thuriaux 1995), a member of the fork head family of proteins of which mammalian histone H5 is also a member (Kaufmann and Knochel 1996), IFH1 may be particularly suited for counteracting the repressive nature of analogous DNA-binding proteins at silenced loci. It is not known at this point whether IFH1 normally plays a role at the silent loci or whether it affected these loci by virtue of its overexpression.
DNA2: DNA2 is an essential gene that encodes a 3′–5′ DNA helicase whose function is required during DNA replication (Budd and Campbell 1995; Buddet al. 1995). Gene fragments encoding the N-terminal region, which does not include the helicase domains, were isolated twice in our screen and found to diminish silencing at telomeres, HM loci, and rDNA (Table 3). The N terminus of Dna2p has no motifs that indicate its function; however, the importance of the region has been underscored by the finding that deletions and point mutations within it are lethal (Budd and Campbell 1995). Genetic and biochemical data suggest that Dna2p acts at the replication fork. Thus, its overexpression may cause a defect in DNA replication that indirectly affects silent chromatin assembly (Lamanet al. 1995), as mentioned for ASF1. Alternatively, DNA2 may be more directly involved in chromatin assembly or disassembly at the replication fork. The helicase may have a dual role of loosening chromatin structure in combination with separating the DNA strands.
RPL32: L32 is an essential ribosomal protein. Overexpression of its cDNA clone caused a subtle but reproducible loss of silencing at telomeres, HML, and HMR, but had no effect on expression of a marker in the rDNA locus (Table 3).
Unlike most yeast genes, RPL32 contains an intron. The L32 protein negatively regulates its own expression by two mechanisms: L32 binds its pre-mRNA and inhibits splicing, and it binds its own spliced transcript and inhibits translation (Dabevaet al. 1986; Eng and Warner 1991; Dabeva and Warner 1993). Because the RPL32 cDNAs we isolated contained no intron, the first form of regulation could not prevent the high level of induction of mRNA synthesis directed from the GAL1 promoter. However, it is not clear how much L32 protein was actually synthesized, given the inhibition of translation that normally comes into play. Therefore, at this point, it is formally possible that it is the mRNA of RPL32 rather than the protein that is causing a disruption of silencing.
If the protein is actually overexpressed and is the active component, the interactions that cause it to weaken silencing may be in the context of its function in the ribosome. Overexpressing L32 may cause a translational defect that lowers the level of a critical silencing factor. Another possibility is that L32 has a function apart from the ribosome that is much more closely related to chromatin structure. It has been proposed that many ribosomal proteins originated as proteins with a different function (often involving nucleic acid interaction) and were co-opted for use in the ribosome. In accordance with this model, ribosomal proteins have been found to participate in a variety of cellular functions, including transcription, RNA processing, and DNA repair (Wool 1996). Hence, L32 might have a direct role in silencing gene transcription. Finally, it is possible that the overexpressed L32 binds and suppresses translation of mRNAs other than its own; one of these could be the message for a critical silent chromatin component or regulator.
DOT1: DOT1 is a previously unidentified gene whose overexpression disrupted silencing at telomeres, the HM loci, and rDNA (Table 3), but had no effect on an unsilenced locus. Moreover, deleting DOT1 also reduced silencing at the telomeres and HM loci (Table 3). In addition, the deletion of DOT1 caused increased heterogeneity in telomere length.
DOT1 was identified recently in a mutant screen for genes involved in a meiotic checkpoint and referred to as PCH1 (pachytene checkpoint; S. Roeder, personal communication). By immunostaining, it was found to be associated with chromosomes in meiosis and present within the mitotic nucleus. Taken together with our findings, it seems very likely that DOT1 is a chromatin protein, and, like SIR4, it is normally important for the formation of repressive chromatin.
DOT4: DOT4, another previously unidentified gene, caused a loss of silencing at telomeres and the HM loci and a weak effect at the rDNA locus when overexpressed (Table 3), either as the truncated clone isolated in this work or as a full-length genomic clone (A. Kahana and D.E. Gottschling, unpublished results). Overexpression of DOT4 had no effect on the expression of an unsilenced marker gene, suggesting that DOT4 overexpression specifically reversed the effects of repressive chromatin rather than generally increasing gene transcription.
Sequence analysis suggested that DOT4 encodes a ubiquitin-processing protease (Ubp), 1 of 17 predicted to be in S. cerevisiae (Hochstrasser 1996). These enzymes cleave ubiquitin moieties from proteins. Conjugation of ubiquitin to proteins can target them for degradation by the 26S proteosome; removal of ubiquitin from a protein substrate by a Ubp would, therefore, be expected to result in stabilization of the protein. Conversely, Ubps can also act to enhance protein degradation by increasing the pool of free ubiquitin monomers or by helping to clear the proteosome of proteolytic fragments attached to ubiquitin. Aside from its role in regulating protein stability, ubiquitin conjugation to a protein substrate has also been implicated in macromolecular protein complex assembly (Finleyet al. 1989; Davie and Murphy 1990; Chenet al. 1996; Hicke 1997).
Other components of the ubiquitin-dependent proteolytic pathway have been associated with silencing. Ubp3p was found to bind to a Sir4p affinity column, and deletion of UBP3 results in an increase in telomeric silencing (Moazed and Johnson 1996). Also, deletion of the ubiquitin-conjugating enzyme RAD6 weakens silencing at telomeres, the HM loci, and the RDN1 locus (Bryket al. 1997; Huanget al. 1997).
Although the effects of overexpressing DOT4 suggest a connection between the protein and silencing chromatin, directly testing this connection was complicated because, in addition to causing defects in silencing, deletion of DOT4 caused a growth defect. Thus, at this point, it is difficult to conclude what role DOT4 might normally play in silencing.
DOT5: Overexpression of DOT5 had a relatively strong disruptive effect on telomeric silencing, a more modest effect on HML and HMR silencing, and no effect on rDNA silencing (Table 3). Its overexpression also caused an increase in telomere length. However, deletion of this gene had no effect on any of the silent loci (Table 3) and did not change telomere length.
Dot5p itself may not be required for normal telomere structure, but may interact with some required factor. Overexpression of DOT5 may shift the steady state of that interaction, causing a decrease in the concentration of silencing factor available for telomere binding. Curiously, DOT5 maps immediately adjacent to EST3, a gene required for replication of telomeric DNA (Morris and Lundblad 1997). While it is not known if EST3 affects silencing, it may be that high levels of DOT5 expression affect EST3 expression. A change in EST3p levels may explain why DOT5 overexpression caused a change in telomere length (Figure 4).
DOT6: The DOT6 sequence predicts a protein with a single Myb-related motif. The Myb domain comprises ∼50 amino acids and is involved in sequence-specific DNA binding (Lipsick 1996). Intriguingly, it was recently found that various telomere sequence-binding proteins, including the telomere repeat-binding factors from human cells and Schizosaccharomyces pombe, contain a single repeat of a Myb-related sequence (Bilaudet al. 1996). The S. cerevisiae telomere repeat-binding protein Rap1p has two motifs related to this sequence. It is interesting to note that both cDNAs of DOT6 isolated in our screen contained the motif.
Overexpression of the DOT6 cDNA caused moderate disruption of telomeric and RDN1 silencing, but had only a small effect at HML and HMR (Table 3). Overexpression of the clone had no effect, however, on telomere length. Moreover, deletion of DOT6 had no effect in any of these assays (Table 3). The simplest explanation of these results is that DOT6 normally has no role in silencing, but that overexpression of its Myb-like sequence competed with the related region in Rap1p for DNA binding. This competition would have to be limited, however, because it did not result in an effect on telomere length, which is sensitive to telomeric Rap1p levels (Conradet al. 1990; Lustiget al. 1990; Kyrionet al. 1992; Marcandet al. 1997).
TLC1: As described in our earlier work, overexpression of the telomerase RNA gene TLC1 disrupts telomeric silencing specifically and causes shortening of the telomeric DNA tract (Singer and Gottschling 1994). In the present study, we found that TLC1 was not directly required for silencing; strains without TLC1 and with only half the normal length of telomeric DNA at the end of the chromosome were still silenced (Table 3). From these results, we suggest that the loss of telomeric silencing when TLC1 is overexpressed is not the consequence of telomere DNA shortening, but rather, that TLC1 RNA is interacting with a telomere-specific silencing factor. Furthermore, TLC1 interferes with the telomeric silencing factor when overexpressed, but TLC1 is not normally required for its telomeric silencing function. From this, we propose that this putative factor is not only important in telomeric silencing, but that it also serves as an anchor for telomerase to localize near the end of the chromosome.
The study of telomeric silencing has yielded insights both specific to telomere structure and generalizable to the larger, interacting collection of repressive loci in the genome. The genes affecting telomeric silencing identified in this work include both newly studied genes and previously known genes whose wider roles had not before been recognized. Although the function of several of these genes is still not known, it appears likely that most of these genes affect silencing through very different mechanisms. This finding reinforces the notion that epigenetic regulation in the cell is the result of an intricate and dynamic system that may be affected and regulated at multiple levels.
We thank J. Broach, J. Choy, S. Diede, R. Esposito, M. Hochstrasser, E. Louis, J. Rine, R. Schiestl, and J. Stevenson for plasmids; J. Boeke and J. Smith for strains; S. Elledge for the cDNA library; and S. Diede and M. Dubois for comments on the manuscript. This work was supported by a Medical Scientist National Research Service Award 5T32 GM07281 (A.K. and M.S.S.), a National Defense Science and Engineering Graduate Fellowship (M.S.S.), a Pew Charitable Trust Biomedical Scholars Fellowship, a Cancer Research Foundation Fletcher Scholarship, and National Institutes of Health grant GM43893 (D.E.G.).
Communicating editor: F. Winston
- Received April 20, 1998.
- Accepted July 14, 1998.
- Copyright © 1998 by the Genetics Society of America