Abstract
In yeast the Sir proteins and Rap1p are key regulators of transcriptional silencing at telomeres and the silent mating-type loci. Rap1 and Sir4 also possess anchoring activity; the rotation of plasmids bound by Sir4 or Rap1 is constrained in vivo, and Rap1 or Sir4 binding can also correct the segregation bias of plasmids lacking centromeres. To investigate the mechanistic link between DNA anchoring and regulation of transcription, we examined the ability of a third defined anchor in yeast, the 2μ circle REP3 segregation element, to mediate transcriptional silencing. We find that placement of the REP3 sequence adjacent to the HML locus in a strain deleted for natural silencer sequences confers transcriptional repression on HML. This repression requires the Sir proteins and is decreased in strains lacking the REP3-binding factors Rep1 and Rep2. The yeast cohesin complex associates with REP3; we show that REP3 silencing is also decreased in strains bearing a mutated allele of the MCD1/SCC1 cohesin gene. Conventional silencing is increased in some strains lacking the 2μ circle and decreased in strains overexpressing the Rep1 and Rep2 proteins, suggesting that the Rep proteins antagonize conventional silencing.
YEAST employs a transcriptional silencing mechanism to control the expression of genes specifying cell type (for reviews see Huang 2002; Ruscheet al. 2003). A mechanistically related but weaker silencing effect is observed when genes are placed adjacent to yeast telomeres. Silencing in yeast requires several trans-acting factors that interact directly with cis-acting “silencer” sequences, including Rap1p, Abf1p, and Orc. One function of these factors is to recruit a complex of Sir proteins; the Sir complex includes Sir2, a histone deacetylase (Imaiet al. 2000; Landryet al. 2000; Smithet al. 2000), and the Sir3 and Sir4 proteins, which can bind the N terminus of histones H3 and H4 (Hechtet al. 1995; Carmenet al. 2002). These activities of the Sir proteins likely establish an altered chromatin structure that restricts transcription.
Silencing factors have additional properties that have not been directly linked to their role in repressing transcription. First, the Rap1 and Sir4 proteins have “anchoring” capability in vivo, defined as the ability to constrain the rotation of plasmid molecules to which these proteins are bound (Ansari and Gartenberg 1997; Mirabella and Gartenberg 1997). Second, plasmids containing yeast replication origins (ARSs) typically require centromeres to be segregated efficiently during mitosis; however, these plasmids can be stabilized by the inclusion of silencer sequences (Kimmerlyet al. 1988), by telomere repeats, which bind the Rap1 protein (Longtineet al. 1992), or by tethering the Sir4 protein using a heterologous DNA-binding domain (Ansari and Gartenberg 1997).
Finally, several studies have shown that Rap1p, Sir3p, and Sir4p are localized to discrete foci found at the nuclear periphery (Palladinoet al. 1993; Cockellet al. 1995). These foci can also include Sir2 (Gottaet al. 1997) and are coincident with the location of telomere sequences (Gottaet al. 1996). A weakly silenced HMR locus becomes more strongly repressed when artificially tethered to the nuclear envelope (Andruliset al. 1998), suggesting that localization directly influences transcriptional repression. However, the cause-and-effect relationship of localization to the nuclear periphery and Sir protein-dependent silencing is not yet clear (Thamet al. 2001; Hedigeret al. 2002b).
To examine the contribution of anchoring to silencing, we tested the ability of the REP3 anchor to mediate transcriptional repression. REP3 is a cis-acting element found on the 2μ circle, an endogenous yeast plasmid. Stable mitotic segregation of the 2μ circle relies on REP3 and two plasmid-encoded proteins, Rep1 and Rep2. Like Sir4 and Rap1, REP3 is also known to have plasmid anchoring ability, as defined above (Gartenberg and Wang 1993). Here we demonstrate that the REP3 sequence also has silencing activity; our characterization of REP3-mediated silencing suggests that anchoring and gene repression are mechanistically related functions.
Description of yeast strains
MATERIALS AND METHODS
A 712-bp DNA fragment containing the genetically defined 2μ circle REP3 sequence (Kikuchi 1983; Jayaramet al. 1985) was amplified from plasmid YEp24 by polymerase chain reaction (PCR) using primers SP69 (GATAGGATCCGCACTTC TACAATGGCTG) and SP70 (GATAGGATCCATAGAGCGC ACAAAGGAG). This fragment includes a core element consisting of five repeats of a 62-bp sequence (Kikuchi 1983; Jayaramet al. 1985). This fragment was subcloned into plasmid pCR2.1 (Invitrogen, San Diego) to form pLXB1. The REP3 sequences were then removed from pLXB1 as a BamHI fragment and subcloned into the BglII site of pDM11 (Mahoney and Broach 1989) to form pLXB2. Plasmid pAW2, used to delete the SIR2 gene, is based on pRS416 (Sikorski and Hieter 1989) and contains ∼300 bp upstream and downstream of the SIR2 open reading frame, separated by the URA3 gene. Plasmid stability was determined by fluctuation analysis; loss rates shown are the means of three or more independent experiments.
All yeast strains used to assess REP3 silencing are derived from DMY1 (Mahoney and Broach 1989; Table 1). YSH392, in which the HML-E silencer is replaced by the REP3 sequence, was constructed via a two-step gene replacement procedure using DMY19 as the starting strain (Mahoney and Broach 1989). In strain DMY19, the HML-E silencer is replaced by the SUP4 gene. In step 1, pLXB2 was cut with XhoI and transformed into DMY19, producing uracil prototrophs by integration at the HML locus. In step 2, ura– recombinants were selected for by plating this strain on 5-fluoroorotic acid media; ura– isolates that lost SUP4-suppressing activity were then tested for the presence of the REP3 sequence at HML-E by Southern blot. Strain YSH409 is an isolate of strain YLS404 (Chi and Shore 1996) that has lost the 2μ circle. Strains lacking the 2μ circle (cir0 strains) were identified by selecting for isolates that survived overexpression of the 2μ FLP1 gene (Volkert and Broach 1986). These isolates were tested for the presence of 2μ by Southern blot or a PCR assay using 2μ-specific primers. Deletions or disruptions of the SIR2, SIR3, and SIR4 genes were made using pAW2, pAR78 (Braunsteinet al. 1993), and pCTC77 (Chienet al. 1993), respectively. Strains containing galactose-inducible REP1 and REP2 genes were made by transforming YSH409 with KpnI-digested pKA610 DNA (containing GAL10-REP1) and/or EcoRI-digested pKA620 plasmid DNA (containing GAL10-REP2; Somet al. 1988). The YKU70, YKU80, HST3, MIG1, MLP1, MLP2, and ESC1 open reading frames were removed from strain YSH392 by PCR-mediated deletion (Wachet al. 1994). The wild-type MCD1/SCC1 gene was replaced with the scc1-73 allele (Michaeliset al. 1997) by one-step gene replacement. A DNA fragment containing both scc1-73 and the URA3MX gene with ends homologous to sequences 5′ and 3′ of SCC1 was made by hybrid PCR (Hortonet al. 1989) and transformed into YSH392. Isolates that were uracil prototrophs and temperature sensitive for growth were screened by PCR for incorporation of the URA3MX gene adjacent to the SCC1 locus.
RESULTS
REP3-mediated silencing: The Rap1 and Sir4 silencing factors can function as anchors in yeast, mediating stable segregation of episomes and restraining the rotation of plasmids bound to these factors (Ansari and Gartenberg 1997; Mirabella and Gartenberg 1997). To examine if silencing is a general property conferred by noncentromeric anchors, we determined the ability of a third anchor, the 2μ circle REP3 sequence, to mediate silencing. The 2μ circle is a naturally occurring, high-copy plasmid present in almost all strains of Saccharomyces cerevisiae (see Broach and Volkert 1991 for review). Efficient segregation of the 2μ circle likely occurs by anchoring the plasmid to a nuclear structure, which is then equally partitioned during cell division (Wuet al. 1987; Scott-Drew and Murray 1998; Velmuruganet al. 2000; Mehtaet al. 2002; Scott-Drewet al. 2002). This partitioning requires two plasmid-encoded trans-acting factors, Rep1p and Rep2p, which bind a cis-acting sequence, REP3 (Velmuruganet al. 1998). Plasmids containing REP3 are topologically constrained in vivo in a Rep1- and Rep2-dependent way (Gartenberg and Wang 1993), an observation consistent with the proposal that Rep1p and Rep2p fix REP3 to a nuclear structure. If anchoring helps mediate silencing, we reasoned then that the Rep system might substitute for some or all functions of the silencing factors. To test this possibility, we replaced the silencer sequences at HML with the REP3 sequence, creating strain YSH392. This strain is missing all functional HML-I silencer sequences, as well as the Rap1- and Orc-binding sites at HML-E (Mahoney and Broach 1989). A defect in silencing causes a reduction in mating efficiency; therefore, we compared the mating ability of strain YSH392 to an otherwise identical strain lacking the REP3 sequences. The results of these experiments are shown in Figure 1. Our results indicate that the REP3 sequence significantly improves the mating ability of this strain.
REP3 silencing requires Rep1p and Rep2p: REP3 silencing could be due to an interaction with Rep1p and Rep2p, as predicted by the premise of our approach, or could be due to an independent mechanism. To determine if REP3 silencing requires 2μ-encoded gene products, we eliminated the 2μ circle in our test strain and repeated the mating assays. As shown in Figure 2, we found that silencing is markedly reduced in this strain, indicating that REP3 silencing does depend on the 2μ circle. However, we note that mating efficiency in the REP3 cir0 strain still exceeds that in the control strain lacking REP3 adjacent to HML. This could reflect a function of the REP3 sequence that is independent of the Rep1 and Rep2 proteins or could be a more indirect effect of losing the 2μ circle (see below).
REP3 silencing requires the SIR gene products: Our results are consistent with the hypothesis that anchoring DNA sequences has a repressive effect on gene expression. There are two general models for this effect. First, anchoring per se could be repressive, perhaps by sequestering sequences from the transcriptional machinery. Alternatively, anchoring could aid the association of transcriptional repressors, such as the Sir proteins. To discriminate between these models, we determined whether REP3 silencing depends on the presence of the Sir proteins. We made deletions or disruptions of the SIR2, SIR3, or SIR4 genes in a strain exhibiting REP3 silencing and measured the effects on mating ability. As shown in Figure 3, elimination of any one of the SIR genes eliminates REP3 silencing. We also found that the silencing retained in YSH393, the cir0 isolate of YSH392, also depends on Sir proteins (not shown). Therefore, our results suggest that REP3 either recruits Sir proteins or relocalizes HML to a location of high Sir protein concentration.
—REP3-mediated silencing. A partial genotype is shown, as well as the results of qualitative and quantitative mating assays. Mating assays were performed as described (Shei and Broach 1995; Dula and Holmes 2000).
An alternative explanation for these results is that the Sir proteins are required for REP3 anchoring function. In this model the role of Sir proteins is indirect, and anchoring remains the dominant event in REP3 silencing. To examine the influence of Sir proteins on REP3 function, we determined the stability of plasmid YEp51 in a set of congenic Sir+ and Sir– strains. YEp51 is a 2μ-derived vector that depends on the REP3 sequence for efficient segregation; plasmids using REP3 for segregation function are unstable in strains lacking the 2μcircle (Kikuchi 1983; Jayaramet al. 1985). We predicted that if REP3 anchoring requires Sir proteins, then this plasmid should also be unstable in a strain lacking Sir protein function. However, we found that the loss rate of YEp51 in wild-type cells was essentially identical to the loss rate in strains lacking the SIR genes (Table 2). Therefore, we conclude that Sir proteins are not required for REP3's anchoring ability.
—REP3 silencing requires the 2μ circle. The mating ability of [cir+] and [cir0] strains is shown.
Rep1 and Rep2 antagonize conventional silencing: As controls for our experiments demonstrating a requirement for the 2μ circle to mediate REP3 silencing, we also isolated cir0 versions of a number of strains lacking chromosomal REP3 sequences. Unexpectedly, the loss of the 2μ circle increased the mating ability of some of these strains. This effect was most pronounced when comparing strains DMY19 and YSH406, which differ only by the absence of the 2μ circle in YSH406. In these strains, the Orc- and Rap1-binding sites of the HML-E silencer are replaced by the SUP4 gene (Mahoney and Broach 1989; see Figure 4). One possible explanation for this increase is that the Rep proteins antagonize conventional silencing; the loss of Rep1 and Rep2 proteins in the cir0 strain would then lead to improved silencing. This model predicts that elevating the expression of the Rep1 and Rep2 proteins might decrease conventional silencing. To test this prediction, we constructed cir0 strains that had integrated copies of the REP1 and REP2 genes, each fused to the inducible GAL10 promoter. These strains also contain the yeast ADE2 gene integrated at the HMR locus, placing its transcription under control of the HMR silencer sequences. Yeast strains that do not express the Ade2 protein accumulate a red pigment when grown on media containing limiting adenine. Therefore, a change in silencing in these strains produces detectable changes in colony color. We compared a control strain with strains overexpressing REP1, REP2, or both REP1 and REP2 (Figure 5). We observed that increasing the levels of either Rep1 or Rep2 proteins had no effect on expression of the ADE2 reporter, while overexpression of both REP1 and REP2 appears to disrupt silencing, allowing increased expression of ADE2. A straightforward model for this antagonism is that the Rep proteins bind and titrate a limiting factor for silencing, such as the Sir proteins. To further investigate the interaction of the Rep and Sir proteins, we examined whether overexpression of SIR3 or SIR4 affects the stability of plasmids that rely on the REP3 sequence (and Rep proteins) for segregation function (Table 2). In this experiment we observed no effects of elevating Sir proteins on REP3 plasmid stability and therefore failed to find evidence for a direct Rep-Sir interaction. However, interactions with a Sir protein complex could depend on normal ratios of individual Sir proteins and not be detected in these experiments.
REP3 plasmid stability assays
—REP3 silencing is Sir dependent. The mating ability of a set of strains deleted for different SIR genes is shown.
Trans-factors influencing REP3 silencing: During the course of our experiments an independent study was published describing a form of transcriptional repression mediated by the 2μ circle replication origin sequences (Grunweller and Ehrenhofer-Murray 2002). This silencing depended on the Sir proteins, as well as on the Mig1 and Hst3 proteins; the 2μ origin sequence mediating silencing contains likely binding sites for the Mig1 protein (Grunweller and Ehrenhofer-Murray 2002). To examine the similarity of this silencing to the one we describe here, we determined the influence of Mig1p and Hst3p on REP3 silencing (Figure 6). We find that deletion of the HST3 gene has no measurable effect on REP3 silencing, while deletion of MIG1 increases REP3 silencing. There are no close matches to the Mig1p-binding site in the REP3 sequence used in our experiments, so the effect of deleting the MIG1 gene is likely to be indirect. We conclude that REP3-mediated silencing is distinct from that caused by the 2μ origin.
Prior studies have defined distinct candidates for proteins that serve as tethering sites involved in silencing and/or plasmid anchoring. The Mlp1 and Mlp2 proteins are associated with the nuclear pore complex and have been reported to influence both telomere localization and telomere silencing (Galyet al. 2000; Feuerbachet al. 2002), although this effect has not been observed in all cases examined (Andruliset al. 2002; Hedigeret al. 2002a). The Ku heterodimer, coded for by the YKU70 and YKU80 genes, interacts with telomeres and influences both telomere localization and silencing (Gravelet al. 1998; Martinet al. 1999; Hedigeret al. 2002b). Finally, Esc1 is a protein located at the nuclear periphery that binds Sir4 and is required for Sir4-mediated plasmid segregation (Andruliset al. 2002). We next tested the influence of these proteins on REP3 silencing and REP3 plasmid stability (Figure 6 and Table 2). REP3 silencing is not affected in a strain missing both MLP1 and MLP2, nor is REP3 silencing decreased in strains lacking YKU70 or YKU80. Instead, deletion of YKU80 improves REP3 silencing. This could be due to release of Sir proteins from the telomere, increasing their concentration at other locations (Martinet al. 1999), although prior studies generally observed identical phenotypes when deleting YKU70 or YKU80. Finally, strains lacking the ESC1 gene show no changes in REP3 silencing. None of these genes had significant effects on REP3 plasmid stability (Table 2).
The REP3 sequence has been shown to associate with members of the cohesin protein complex (Mehtaet al. 2002). The specific role of cohesins in REP3's segregation function is not yet clear, but cohesins may mediate an association of the 2μ circle with chromosomes (Mehtaet al. 2002). The genes coding for cohesins are essential; therefore, to examine their influence on REP3 silencing we replaced the wild-type copy of the MCD1/SCC1 cohesin gene with the scc1-73 temperature-sensitive allele (Michaeliset al. 1997) in a strain exhibiting REP3 silencing. We tested the effects of this mutation at 30°, a temperature at which the strain exhibits wild-type growth properties. Figure 7 shows that this mutation markedly reduces REP3 silencing. This is unlikely to be an indirect effect of reducing the copy number of 2μ circle in our strain; all gene deletions described in this work had no effects on 2μ copy number, with the exception of the MIG1 knockout, in which 2μ copy number was reduced (not shown). As a control for the scc1-73 experiment, we isolated a revertant of this strain that had regained wild-type growth at 37°. We observed that this strain also reverted the mating defect (Figure 7).
—Increased mating of a [cir0] strain. The mating ability of congenic [cir+] and [cir0] strains is compared.
DISCUSSION
We have defined a new form of silencing in yeast mediated by the 2μ circle REP3 sequence. This sequence was previously defined as a cis-acting element required for stable mitotic segregation of the yeast 2μ plasmid. The behavior of plasmids bearing the REP3 sequence suggests that by binding the Rep1 and Rep2 proteins REP3 mediates attachment to a nuclear structure. A similar proposal has been put forth to explain the stable segregation and anchoring mediated by the Sir4- and Rap1-silencing proteins (Ansari and Gartenberg 1997; Andruliset al. 2002). Here we have shown that, like sequences bound by Rap1 or Sir4, the REP3 sequence can mediate transcriptional repression, raising the possibility that anchoring ability is linked to the ability to silence gene expression. Despite these similarities, the plasmid-anchoring mechanisms mediated by Sir4, Rap1, and the Rep proteins appear to be distinct. Rap1 and Sir4 are found at the nuclear periphery at discrete locations that include the Sir2, Sir3, and Ku proteins (Gottaet al. 1996; Larocheet al. 1998). The simple model that these foci provide attachment points for plasmids bound by Rap1 and Sir4 is likely not true; instead, attachment appears to be mediated by the Sir4-binding protein Esc1, a nuclear peripheral protein with a localization distinct from the well-characterized Sir/Rap1 foci (Andruliset al. 2002). While the Rep1 and Rep2 proteins are also found in foci that may be preferentially localized to the nuclear periphery (Scott-Drew and Murray 1998), Rep protein clustering is not coincident with Sir protein foci (Scott-Drewet al. 2002). We also show in this study that deletion of YKU70, a mutation that eliminates Sir/Rap1 protein clustering (Larocheet al. 1998), does not diminish REP3 silencing. In addition, while deletion of the ESC1 gene eliminates Sir4-mediated plasmid segregation, it does not affect the ability of REP3 to act as an anchor (Andruliset al. 2002) or mediate segregation of REP3 plasmids (Table 2). Finally, the Mlp and Yku proteins have been shown to contribute to telomere localization and telomere silencing (Larocheet al. 1998; Galyet al. 2000; Feuerbachet al. 2002); here we observed no significant effects of deleting genes for these proteins on REP3 plasmid segregation, and these gene deletions did not diminish REP3 silencing. Therefore, known mediators of Sir4-mediated plasmid segregation and Sir-protein clustering do not significantly affect REP3 silencing.
—Overexpression of the Rep1 and Rep2 proteins decreases silencing. The results of a colony color assay for silencing at HMR are shown. Transcription of the ADE2 gene is placed under control of the HMR silencer sequences. A decrease in the expression of ADE2 leads to an increase in red color. A partial genotype of each strain is listed. hmrΔA strains lack the HMR-E ACS element (the ORC-binding site). Overexpression of REP1 and REP2 had no effect on the expression of ADE2 when at its normal chromosomal location (not shown).
The observation that deletion of ESC1 has only subtle effects on telomeric silencing led to a proposal that Esc1p could represent one of multiple redundant pathways that function to localize silencing complexes (Andruliset al. 2002). Our observation that the REP3 sequence mediates Sir-dependent transcriptional repression raises the possibility that it may utilize one of these proposed alternate pathways. This model could accommodate the observation that Rep protein overexpression antagonizes conventional silencing; if Rep proteins and silencing factors share an anchoring site, alterations in Rep protein concentration could alter accessibility of the site to silencing factors. Experiments demonstrating that recruiting telomere or HMR-linked reporters to the nuclear membrane using a lipid-docking strategy (Andruliset al. 1998, Feuerbachet al. 2002) suggest that fairly nonspecific localization can aid silencing. However, thus far there is no direct evidence that the Rep system anchors plasmids to a peripheral nuclear position, and the mechanism for REP3's ability to mediate plasmid segregation is not clear. Mehta et al. (2002) have proposed a mechanism for segregation of the 2μ circle on the basis of the ability of the REP3 sequence to attract the cohesin complex and to associate with chromosomes. However, the contribution of these properties to REP3's anchoring and silencing abilities is also unclear. The 2μ circle does not appear to associate with the mitotic spindle (Scott-Drewet al. 2002), nor is it associated with cohesin-binding sites on chromosomes (Scott-Drewet al. 2002). Here we observe that a mutation in the MCD1/SCC1 cohesin gene affects REP3 silencing, but does not alter 2μ copy number.
—Trans-factors and REP3 silencing. Qualitative and quantitative mating assays were performed on strains differing only at the indicated loci.
REP3 silencing may occur via a mechanism that is independent of its anchoring and partitioning functions. For instance, REP3 may directly or indirectly recruit the Sir proteins. The REP3 sequence we used does not contain consensus sequences for known silencer-binding factors, which recruit the Sir proteins to their known sites of action. Sir proteins could be attracted via interactions with Rep1, Rep2, or other REP3-associated proteins. We also note that the HMLα promoter sequences contain a Rap1-binding site that has been shown to contribute to silencing in some contexts (Cheng and Gartenberg 2000); thus, a minimal ability to attract Sir proteins may be important for the silencing we have observed, a phenomenon also observed using the lipid-docking approach (Andruliset al. 1998). Similar context effects may explain why the REP3 sequence was not found to mediate significant silencing of a plasmid-based reporter gene (Grunweller and Ehrenhofer-Murray 2002). Genes found on the 2μ circle are subject to complex transcriptional regulation (Somet al. 1988; Veit and Fangman 1988; Grunweller and Ehrenhofer-Murray 2002). In particular, the FLP1, REP1, and RAF genes of the 2μ circle are transcriptionally repressed by a mechanism that depends on the Rep1 and Rep2 proteins (Somet al. 1988; Veit and Fangman 1988). Thus, REP3 may participate in controlling gene expression in its natural context.
—MCD1/SCC1 influences REP3 silencing. The temperature-dependent growth and mating efficiencies of strains differing only at the MCD1/SCC1 locus are shown. The SCC1REV strain (YSH555) is a revertant of strain YSH554 that has growth characteristics indistinguishable from the parent (YSH392) strain. YPD columns show growth of the three strains on nonselective media at the indicated temperatures. The SD column shows the results of a qualitative mating test performed at 30°.
One set of proteins known to associate with REP3 via interactions with Rep1 and Rep2 is the cohesin complex (Mehtaet al. 2002). We show here that mutations in the Mcd1/Scc1 cohesin reduce REP3 silencing. Cohesins have been identified in two prior contexts related to Sir-dependent silencing. First, the Smc1 and Smc3 cohesins were found to be important for establishing a boundary to silencing at the HMR locus (Donzeet al. 1999). Second, depletion of the Mcd1/Scc1 cohesin allowed establishment of silencing to occur earlier in the cell cycle (Lauet al. 2002). Therefore, cohesins may influence the association of Sir proteins with target sequences. Our results are consistent with cohesins positively influencing the association of Sir proteins, leading to increased silencing of REP3-associated sequences.
Acknowledgments
We thank Karen Malatesta, Marc Gartenberg, Kim Nasmyth, and James Broach for providing strains and plasmids; Adrienne Woike, Xiaobin Li, and Marie Knight for assistance in strain construction and plasmid stability assays; and members of the Holmes lab for helpful discussions. This study was supported by grant RPG-98-351-01-MGO from the American Cancer Society and grant MCB-0096561 from the National Science Foundation to S.G.H.
Footnotes
-
Communicating editor: L. Pillus
- Received July 30, 2002.
- Accepted October 3, 2003.
- Copyright © 2004 by the Genetics Society of America
