Background

Transcriptional silencing in the budding yeast Saccharomyces cerevisiae occurs at the silent mating-type loci HMR and HML, telomeres, and at the rDNA locus. At all of these silenced regions, DNA binding proteins recognize specific motifs and recruit a silencing protein complex (reviewed in [1]). HMR and HML are flanked by E and I "silencers." Each silencer has binding sites for ORC, and Rap1 or Abf1. The potent HMR-E silencer has a binding site for all three proteins. At telomeres, Rap1 also contributes a critical DNA-binding function, binding to the TG1–3 repeats.

At both the silent mating loci and at telomeres, the DNA-binding proteins recruit a Sir protein complex that can spread to silence genes at a distance (reviewed in [2]). At HMR-E, for example, this is achieved by ORC recruitment of Sir1 via a Sir1-Orc1 interaction [3], and Rap1 and Abf1 binding to Sir4 and Sir3 [4, 5]. Sir4 and Sir3 multimerize, both with themselves and each other [4, 6]. Sir4 also binds Sir2, and Sir2 plays a crucial role in the spreading of a Sir2, Sir3, Sir4 complex on chromatin by deacetylating histone H4 lysine 16. The deacetylation produces a novel product, 2' O-acetyl-ADP-ribose [7–9]. This compound has recently been shown to produce a conformational change in Sir3 that is likely to promote spreading of the Sir complex. Sir2, Sir3 and Sir4 are essential for silencing at the HM loci and at telomeres, while Sir1 plays a prominent role in silencing at the mating type loci but not at telomeres. The Yku70/Yku80 heterodimer that binds to DNA ends plays an important role in silencing at telomeres, while being dispensable for silencing at HM loci [10].

Deletion of the HMR-E silencer leads to loss of silencing of genes at the HMR locus [11, 12]. However, if heterologous DNA binding sites (such as Gal4 sites) are integrated in place of HMR-E and the strain is transformed with certain Gal4 DNA binding domain (GBD) – silencing protein hybrids, silencing can be restored due to targeting of Sir proteins or proteins that bind to Sir proteins to HMR-E ([13]; see also Figure 1). This so-called "targeted silencing" has been a useful tool for investigating the process of Sir protein recruitment to silenced loci [3, 13–17]. We previously described a screen for proteins capable of targeted silencing at HMR [16]. In this screen many known proteins were identified, as well as several proteins not characterized at the time, which we called Esc proteins because they e stablish s ilent c hromatin. One of these was Esc4 whose characterization we describe in this report. RTT107/ESC4 was also identified in screens for mutants with increased Ty transposon mobility or DNA repair defects [18–20]. For simplicity we will only use the name ESC4 in the remainder of this paper. Esc4 has been shown to be phosphorylated by Mec1 kinase on SQ/TQ motifs in response to DNA damage during S phase [21]. The phosphorylation by Mec1 has quite recently been proposed to be regulated by Slx4, which was also shown to form a complex containing Esc4 [22].

Figure 1
figure 1

Esc4 was identified in an HMR targeted silencing screen. (A) In the presence of both wild-type silencers, HMR-E and HMR-I, the URA3 reporter gene at HMR is completely silenced in a SIR-dependent manner leading to growth in the presence of the drug 5-FOA. (B) When the HMR-E silencer is deleted (and here replaced with a binding site for the Gal4 protein (G)), the HMR locus is no longer silenced and the URA3 reporter gene is expressed. This loss of silencing at hmr::URA3 leads to no growth on 5-FOA media. (C) When the strain in (B) is transformed with a GBD plasmid (GBD-X) capable of causing targeted silencing, this leads to restoration of silencing at hmr::URA3 and growth on 5-FOA. In this way, a GBD hybrid protein library was screened for factors capable of targeted silencing. A GBD-Esc4(1–1070) hybrid was identified.

Full size image

In this report we show that Esc4 has 6 BRCT motifs and that they are important for its function. Many proteins that function in repair, and even a silencing protein, Rap1, contain BRCT (BR CA1 C-terminus) motifs. This motif was first identified by database searching using the C-terminus of the human breast cancer susceptibility protein, BRCA1 [23]. Since then, BRCT motifs from the human DNA repair protein XRCC1 and the two tandem BRCT motifs of BRCA1 protein have been crystallized [24, 25] and crystal and/or solution structures have also been solved of BRCTs from 53BP1, DNA ligase III, and an NAD-dependent DNA ligase [26–29]. More recently, it was discovered that BRCT motifs could specifically bind to phosphoserine-containing proteins [30, 31] and structures of such complexes were subsequently also determined [32–34]. Overall, BRCT motifs are thought to mediate a diverse array of protein-protein interactions, binding to proteins with different structures, as well as to other BRCTs, both inter- and intra-molecularly (reviewed in [35]).

DNA repair requires many DNA-modifying enzymes such as nucleases, ligases, topoisomerases, polymerases and helicases. Saccharomyces cerevisiae SGS1 is a member of the RecQ family of genes encoding DNA helicases. In addition to Escherichia coli recQ, this helicase gene family includes human BLM, WRN, RECQL, and Schizosaccharomyces pombe rqh1+. Members of the RecQ family of helicases have been implicated in genomic stability, aging and cancer. Yeast SGS1 has been shown by several labs to suppress DNA recombination, gross chromosomal rearrangements, and Ty1 transposition, and to exhibit 3' to 5' helicase activity [20, 36–38]. Six SLX genes have been isolated in a synthetic lethal screen using an sgs1 Δ mutant [39]. Two of these genes, SLX2 and SLX3, encode the Mus81/Mms4 nuclease that acts on branched DNA structures. SLX1 is also predicted to be a nuclease, based on its sequence [40]. Slx1 and Slx4 coimmunoprecipitate and slx1 and slx4 mutants display similar phenotypes and have been proposed to function together [39]. In addition, a genome-wide genetic interaction (synthetic genetic array, SGA) screen isolated a total of 24 genes that show a synthetic growth interaction with an sgs1 Δ mutation [41]. This screen identified 4 of the 6 SLX genes previously identified in the sgs1 Δ synthetic lethal screen described above [39], as well as three other genes known to show synthetic interactions [39, 42–45]. The screen also identified 16 other genes that caused synthetic lethality or sickness when mutated with sgs1 Δ. One of these genes was ESC4. Subsequently, there have been other reports confirming this genetic interaction between SGS1 and ESC4 [46, 47].

Here we demonstrate that the C-terminal two BRCT motifs of Esc4 bind to Sir3 and are sufficient for SIR-dependent targeted silencing at HMR. Furthermore, the N-terminal four BRCT motifs in Esc4 bind to Slx4, thus linking this DNA repair protein to silent chromatin.

Results

Esc4 establishes targeted silencing when targeted to HMR

In a screen described previously, we identified Esc4 as a protein that could restore silencing when targeted to an HMR locus harboring a deletion of the HMR-E silencer [16]. Targeting of proteins to HMR was mediated by the binding of a Gal4 DNA binding domain (GBD)-hybrid protein to a Gal4 DNA binding site (G) that replaced the HMR-E silencer (Figure 1). Silencing was assessed using a URA3 reporter gene integrated at the HMR locus.

Targeted silencing by Esc4 is SIR-dependent

In order to understand how Esc4 promoted silencing when targeted to the HMR locus, we tested silencing by GBD-Esc4 in various strains. First, targeted silencing by GBD-Esc4 was compared to GBD alone or to the potent targeted silencing factor GBD-Esc1 in the strain in which the screen was performed, which has the entire HMR-E silencer deleted and replaced with a Gal4 DNA binding site (HMR-E aeb::G) [17]. The fraction of cells silenced by GBD-Esc4 was not as great as by GBD-Esc1 but, nevertheless, significant targeted silencing at HMR was observed despite the complete absence of the HMR-E silencer (Figure 2, HMR-E aeb::G). When targeted silencing by GBD-Esc4 was assessed in strains harboring deletions of just the E and B sites (HMR-E Aeb::G) or the A and E sites (HMR-E aeB::G) of the HMR-E silencer, silencing was increased roughly 50 to 100-fold, as expected for strains with at least one element of the natural silencer (Figure 2).

Figure 2
figure 2

Targeted silencing by Esc4 is SIR -dependent. Targeted silencing by GBD-Esc4 (1–1070) was assessed in hmr::URA3 silencing reporter strains containing different HMR-E elements replaced by a Gal4 DNA binding site (G), similar to the situation shown in Figure 1B. A TRP1-marked plasmid expressing GBD alone served as a negative control and targeted silencing by the potent targeted silencing factor GBD-Esc1(1124–1658) served as a positive control [17]. Targeted silencing by GBD-Esc4 in strains with wild-type SIR genes (SIR+) or sir2 Δ, sir3 Δ, or sir4 Δ mutations was examined.

Full size image

To test whether targeted silencing by GBD-Esc4 was SIR-dependent, it was tested in strains deleted for the SIR2, SIR3 or SIR4 genes. As seen in Figure 2, silencing by Esc4 required each of these SIR genes. This was expected since targeted silencing by other proteins has been shown to be SIR-dependent in every case examined previously [14, 16]. It thus seemed likely that Esc4 caused silencing by recruiting the Sir protein complex.

Saccharomyces Esc4 proteins contain six BRCT motifs and are homologous to S. pombe Brc1

Esc4 contains several copies of the BRCT motif, originally identified in the human breast cancer susceptibility gene, BRCA1 [23]. The presence of these motifs in Esc4 (encoded by yeast ORF YHR154w) has been observed previously by database searches [48, 49], although there has been disagreement as to how many BRCTs exist in this protein, with some reports suggesting six BRCTs [49, 50] and others only four [21, 22, 48]. Our own analyses including use of Pfam [51, 52] confirms the first five putative BRCTs in Esc4, and it seems likely that residues 935–1049, which are BRCT-like with a conserved W or Y residue near the C-terminus, also form a domain that folds into a BRCT type structure as previously reported [49] (Figure 3). Furthermore, analysis of a homolog of Esc4 from S. pombe, Brc1, using Pfam shows six BRCTs in this homologous protein, with the last two again being the most alternative (E-values of 4.1 × 10-5 and 5.8 × 10-2 for the fifth and sixth, respectively, compared to values ranging from 2.8 × 10-6 to 1.7 × 10-14 for the first four BRCTs), further supporting a conserved total of six motifs in these homologs (Figure 3).

Figure 3
figure 3

Saccharomyces cerevisiae Esc4 has six BRCT motifs that are conserved in both sequence and domain architecture among yeast homologs. The six BRCT motifs in Esc4 are indicated with brackets and yellow color and were described previously [49]. The motifs in Esc4 are shown as they align with a subset of Esc4 homologs: three from other budding yeasts Ashbya gossipii, Kluyveromyces lactis and Candida glabrata and one from the evolutionarily distant fission yeast Schizosaccharomyces pombe Brc1. The highly conserved aromatic residue, a hallmark of BRCT motifs, is marked above the alignment with a red asterisk. Blue color indicates similarity with the predominant residue at the position in the alignment, with darkness correlated with greater similarity as calculated by ClustalW software and displayed with Jalview alignment editor's BLOSUM62 score viewing option [67, 68].

Full size image

The six BRCT motifs of Esc4 protein exist as a set of four tandem motifs at the N-terminus of the protein and two more at the C-terminus (Figure 3, yellow lines). These two sets of BRCT motifs are separated by a 375 amino acid linker region. Although there are proteins from various budding yeasts with compelling sequence similarity to S. cerevisiae Esc4 (e.g. Ashbya gossipii, Kluyveromyces lactis, and Candida glabrata, as shown in Figure 3), currently the only obvious non-budding yeast homolog candidate is Brc1 from the evolutionarily distant fission yeast Schizosaccharomyces pombe (Figure 3, bottom line). S. pombe Brc1 is 878 amino acids in length and also contains four BRCTs at the N-terminus and two at the C-terminus, separated by a linker region. Esc4 and Brc1 are 52% similar and 21% identical overall. Human and mouse PTIP proteins have been reported to be quite similar in sequence [21, 53], but do not contain the classically-conserved C-terminal aromatic residue present in all of the six BRCTs highlighted with a red asterisk in the alignment shown in Figure 3. Therefore, it is not yet clear if these proteins are bona fide Esc4/Brc1 homologs.

The C-terminal two BRCT motifs of Esc4 are sufficient for targeted silencing

The Esc4 hybrid protein isolated in our targeted silencing screen was full-length [16] and therefore contained all six predicted BRCT motifs. To determine which part of Esc4 and which BRCT motifs were responsible for the SIR-dependent targeted silencing, we constructed three GBD hybrids: one to the N-terminal four BRCT motifs (GBD-Esc4N, aa 1–480), one to the linker between the N- and C-terminal sets of motifs (GBD-Esc4L, aa 480–836), and one to the C-terminal two BRCT motifs (GBD-Esc4C, aa 818–1070). These constructs were tested for targeted silencing in a strain harboring deletions of the E and B sites (HMR-E Aeb::G) and in a sir2 Δ derivative of that strain (Figure 4; see Table 1 for strain information). Significant targeted silencing was observed by GBD-Esc4C, although it was not as much as with full-length Esc4. The observed silencing by GBD-Esc4C was SIR-dependent, as observed for the full-length protein (Figure 2). No significant silencing was seen with Esc4N or with the linker region, Esc4L (Figure 4).

Figure 4
figure 4

Targeted silencing activity is achieved by the C-terminus of Esc4. Three different regions of Esc4, containing either the first four BRCT motifs (1–480), linker region (480–836), or last two BRCTs (818–1070) were tested for targeted silencing in a strain with an HMR-E Aeb::G silencer and URA3 reporter gene at HMR. GBD-Esc1 served as a potent positive control and GBD as a negative control. SIR2-dependence of targeted silencing was also determined.

Full size image
Table 1 Yeast strains used in this study
Full size table

The C-terminal BRCT motifs of Esc4 interact with Sir3

Because the C-terminal two BRCT motifs of Esc4 gave targeted silencing, we suspected that this region of the protein was binding to a silencing protein to recruit the Sir complex. Using a LexA-Esc4C hybrid and the two-hybrid reporter strain L40 [54], we tested for two-hybrid interactions with several Gal4 activation domain (GAD)-silencing protein constructs, including Sir1, Sir2, Sir3, Sir4 and Rap1 [3, 17, 55]. A strong interaction (i.e., the lacZ reporter gene generated blue color from X-gal visible in 15 minutes) with GAD-Sir3 (aa 252–978) was identified, as well as a weaker interaction with GAD-Sir4 (aa 839–1358) (Table 2). None was detected with Sir1, Sir2 or Rap1. Because the GAD-Sir4 (aa 839–1358) hybrid contained the region known to bind to Sir3 [7], we hypothesized that LexA-Esc4 was binding to GAD-Sir4 via a bridge of endogenous Sir3. To test this, we used a derivative of strain L40 harboring a sir3 Δ mutation and examined the LexA-Esc4 interaction with GAD-Sir4. In this case, the interaction with GAD-Sir4 was no longer observed, whereas the interaction with GAD-Sir3 and an unrelated two-hybrid control interaction were unaffected (Table 2). When a sir4 Δ derivative of L40 was used, no change in the LexA-Esc4 interactions with GAD-Sir3 or GAD-Sir4 was observed, further supporting the idea that Esc4 requires Sir3 to interact with Sir4, and not vice versa. Taken together, the targeted silencing data strongly suggest an interaction (probably a direct one) between the C-terminal BRCT motifs and the silencing protein Sir3.

Table 2 Esc4 binds to Sir3 in the two-hybrid system
Full size table

The N-terminal four BRCT motifs of Esc4 bind to Slx4

The above results indicated that Esc4 caused SIR-dependent targeted silencing primarily by binding to Sir3 through its C-terminal two BRCT motifs. In addition, Esc4 also contains four BRCT motifs at its N-terminus and these BRCT motifs are more similar to those found in various proteins from diverse eukaryotes (Figure 3, and ref. [49]). In order to identify proteins that bind to these N-terminal BRCT motifs, a two-hybrid screen was performed using a LexA-Esc4N (aa 1–480) fusion protein. Strikingly, from screening ~2 × 107 library plasmids expressing GAD protein hybrids, thirteen clones containing in-frame fusions of GAD to Slx4 were identified (Table 3). In-frame GAD fusions to the six different positions in Slx4 were isolated (see Table 3) and all contained at least the C-terminal half of the protein (residues 383–748). The Esc4N-Slx4C two-hybrid interaction was a very strong one and was specific. Also, Slx4 did not show an interaction with Esc4C. Thus, the N-terminal BRCT motifs of Esc4 are not only necessary for binding Slx4, as reported while this manuscript was in preparation [22], but are indeed sufficient for this interaction in vivo. Furthermore, our two-hybrid data show that the region of Slx4 sufficient for Esc4 binding resides in its C-terminus.

Table 3 The amino-terminal BRCT motifs of Esc4 bind to Slx4
Full size table

Genetic and phenotypic analysis of ESC4

A heterozygous diploid strain with a complete deletion of ESC4 was constructed and dissected to generate a null mutant haploid. This esc4 Δ mutant grew normally and also mated with normal efficiency (suggesting no gross defect in HM silencing). Furthermore, when an esc4 Δ mutation was introduced into a strain with a telomere reporter gene, no telomeric silencing defect was seen (data not shown). Thus, although Esc4 binds to Sir3, Esc4 does not appear to be a protein required for Sir protein-mediated silencing.

The BRCT motif was originally identified in the human BRCA1 tumor suppressor protein [23]. BRCA1 functions in DNA repair and DNA damage-sensing in cell cycle checkpoints (reviewed in [56]). As shown in Figure 5, strains deleted for ESC4 grew significantly less well than wild type on medium containing either MMS or HU. This result confirms reports that have since been published [18, 21, 22, 46], as does our observation that esc4 Δ mutants are not sensitive to ultraviolet radiation [18, 21] (data not shown).

Figure 5
figure 5

Genetic evidence that ESC4 and SLX4 function together in a parallel pathway to SGS1 for repair of damaged DNA. Strains were grown in YPD medium to saturation and then ten-fold serial dilutions were spotted onto YPD, YPD + HU (100 mM), or YPD + MMS (0.032% in part A and 0.014% in part B) media. Cells were incubated at 30°C for 2 days (SC and SC + MMS) or 3 days (SC + HU) before being photographed. WT, wild type. An asf1 Δ mutant was used as a positive control for MMS and HU sensitivity.

Full size image

As shown in Figure 5, like esc4 Δ mutants, slx4 Δ mutants were sensitive to 0.032% MMS. Furthermore, the esc4 Δ slx4 Δ double mutant did not exhibit a greater MMS sensitivity than either single mutant (Figure 5A and data not shown), suggesting that they cooperate in providing resistance to MMS. Another group reported that an esc4 Δ slx4 Δ double mutant was more sensitive than either single mutant but the difference was very slight [22]. In contrast, an esc4 Δ, but not a slx4 Δ mutant, was significantly HU-sensitive, and the double mutant was no more sensitive than the esc4 Δ strain. Thus, Esc4 appears to act independently of Slx4 in providing resistance to HU.

Because Esc4 bound to Slx4 and because the mutant was sensitive to MMS, this suggested that Esc4 might function in the same pathway as Slx4. SLX4 was first identified in a screen for genes required for viability of yeast cells deleted for SGS1 [39]. Therefore, we tested if esc4 was also synthetically lethal with sgs1. To do this, an esc4 Δ mutant was crossed with an sgs1 Δ mutant, the diploid was sporulated and dissected and meiotic progeny were analyzed. Haploid esc4 Δ sgs1 Δ cells were viable, but were noticeably slower-growing that either single mutant (e.g., see Figure 5, YPD plate). While this work was in progress, this genetic interaction was also observed in genome-wide studies [41, 46].

An sgs1 Δ mutant showed sensitivity to both MMS and HU (Figure 5), as expected based on previously published results. An asf1 Δ mutant was used as a control and displayed sensitivity to both DNA damaging chemicals, as expected [57, 58]. Interestingly, an esc4 Δ sgs1 Δ mutant displayed MMS and HU sensitivity that was much more pronounced than that of either single mutant (Figure 5A and 5B). The enhanced sensitivity of this double mutant (10,000-fold on 0.014% MMS plates) seemed to be due to a synergistic repair defect and not entirely due to the growth defect (10-fold difference in colony number on YPD control plates) that was also observed in the esc4 Δ sgs1 Δ strain.

Discussion

By screening a library of factors that could function in place of the HMR-E silencer when targeted to DNA, we identified Esc4 for its ability to e stablish s ilent c hromatin [16]. Protein sequence analysis showed that Esc4 protein contains six BRCT motifs; four are found in tandem at the amino-terminus and two more are at the carboxy-terminus. The entire Esc4 protein was present in the hybrid identified in the targeted silencing screen. Since targeted silencing by Esc4 at HMR was found to be SIR-dependent, it seemed likely that some region within Esc4 was attracting a silencing protein complex to DNA. We tested subsets of the BRCT motifs, as well as the linker between them, for targeted silencing at HMR. These experiments demonstrated that the C-terminal two BRCTs caused targeted silencing that was nearly as strong as with full-length Esc4. Because silencing by this pair of BRCT motifs of Esc4 was also SIR-dependent, it seemed very likely that this region was recruiting a Sir protein when tethered to DNA. Therefore, we tested the C-terminal BRCT motifs for interactions with known silencing proteins by two-hybrid analysis. We identified a specific interaction with Sir3 (aa 252–978). We conclude that binding of Sir3 by Esc4 is likely to be responsible for the SIR-dependent targeted silencing activity.

In some cases BRCT motifs have been shown to bind to phosphorylated serine residues. Specifically, they have been shown to bind to phosphopeptides with the following consensus: pSxxF [32]. Interestingly, Sir3 has an SxxF sequence (aa 583–586) within the Esc4-interacting region that we describe here (aa 252–978) and, furthermore, Sir3 protein has been shown to be phosphorylated [59, 60], suggesting that an Esc4 BRCT motif or perhaps the combination of the two in the C-terminus may bind to phospho-Sir3. However, not all proteins bound by BRCT motifs have the SxxF motif [61], so the precise BRCT-interacting region of Sir3 could be elsewhere.

In addition to binding Sir3 via C-terminal tandem BRCT motifs, Esc4 also binds to Slx4 via four tandem N-terminal BRCTs, as we have shown here by two-hybrid screening. This two-hybrid result demonstrates that these four BRCTs are sufficient for binding Slx4 and agrees with a recent report showing that the N-terminal BRCT motifs are required for this interaction [22]. It seems quite possible that Esc4 could bind Sir3 and Slx4 concurrently, given that these nuclear proteins' binding sites within Esc4 map to BRCT clusters separated by a long linker. Slx4 has been shown to heterodimerize with the endonuclease Slx1 to cleave DNA containing 5' -flap structures, such as in stalled replication forks, to facilitate their repair [62]. Thus, Esc4 binds the silencing protein Sir3 and also to Slx4, an important DNA repair complex component. Esc4 may play a role in facilitating repair of aberrant DNA structures, perhaps specifically within silent chromatin. Esc4 is a Mec1 kinase target [21] and this phosphorylation is required for its repair function. It is possible that phosphorylation of Esc4 by Mec1, which occurs just N-terminal to the Sir3-binding BRCTs, regulates association with Sir3 or other factors required for its ability to repair particular chromosomal loci in S phase.

We analyzed Esc4 protein alignments (such as that shown in Figure 3) for evidence of conserved regions in the protein other than BRCT motifs. One region of interest was the SQ/TQ motifs between amino acids 743 and 807, which were shown to be important for function in DNA repair [21]. We did not find that these motifs were well-conserved, suggesting that the specific site of phosphorylation is not particularly critical in proteins with otherwise similar overall BRCT domain architecture (i.e., [BRCT]4-linker-[BRCT]2 arrangement). This may be because of some differences in Esc4 functions in diverse yeasts or may suggest that flexibility is tolerated in positioning of the phosphorylation sites, and therefore the exact relative location of kinase target sites has not been constrained during evolution.

Future structural and genome sequencing studies are likely to unveil similarities and differences between multi-BRCT domain containing proteins. Whether these proteins play largely protein-scaffolding roles or also contain intrinsic enzymatic properties will be interesting to discover.

Conclusion

We have shown that Esc4 caused targeted silencing when tethered at a weakened HMR locus. The targeted silencing activity was primarily due to the C-terminal two tandem BRCT motifs in Esc4, which bound to Sir3, probably through a direct interaction. This interaction led to the recruitment of the Sir complex and hence caused targeted silencing. The N-terminal BRCT domains were sufficient for binding to Slx4, which functions with Esc4 in DNA repair. Thus, the nuclear Esc4 protein uses its six BRCT motifs to connect diverse proteins involved in DNA repair and silent chromatin.

Methods

Targeted silencing

Esc4 was identified in a targeted silencing screen that has been described previously [16]. Briefly, a Gal4 DNA binding domain (GBD) library was screened for hybrid proteins capable of establishing silencing of a URA3 reporter gene integrated in place of mating-type genes at an HMR locus that had the HMR-E silencer replaced by a Gal4 DNA binding site (G) (see Figure 1A). A full length GBD-Esc4 (1–1070) clone, aeb15, was identified as being capable of establishing targeted silencing of hmr::URA3, causing resistance to 5-fluororotic acid (5-FOA). This GBD-Esc4 clone was subsequently transformed into strain YEA76 (HMR-E Aeb::G hmr::URA3 gal4::LEU2) and YEA77 (HMR-E aeB::G hmr::URA3 gal4::LEU2) (see Table 1 and Yeast Strain section for details) and tested for targeted silencing. To test SIR-dependence of targeted silencing by Esc4 at HMR, shown in Figure 4, targeted silencing in strain YEA76 (SIR+) was compared with that in sir mutant derivatives YAM7 (sir2 Δ), YEA 118 (sir3 Δ), and YKAB17 (sir4 Δ).

For the targeted silencing experiments shown in Figures 2 and 4, assays were carried out as follows: strains were transformed with plasmids expressing the appropriate GBD hybrid protein, grown at 30°C for two days in SC -Trp medium (to select for the GBD plasmid), serially diluted ten-fold five times and spotted on SC -Trp + 5-FOA plates (to assay for silencing at HMR) or on SC -Trp control plates.

Plasmids

Plasmid aeb15, expressing GBD-Esc4(1–1070) was isolated in the targeted silencing screen. This plasmid was recovered from a library based on pGBT9.C (ADH1 UAS -G BD , TRP1, CEN4/ARS1; see Acknowledgements). To generate plasmids for use in the two-hybrid system, ESC4 sequences were amplified from genomic DNA and subcloned into plasmid pSTT91, a derivative of pBTM116 that contains the ADE2 gene ([63]; see also [64]). Plasmids pAM2 (LexA-Esc4N (a.a. 1–480) and pAM7 (LexA-Esc4C (a.a. 818–1070) were used for two-hybrid experiments. To test for LexA-Esc4C binding to GAD-Sir3 and GAD-Sir4, plasmid pEDA195, GAD-Sir3(252–978), and pCTC36, GAD-Sir4(839–1358), were used. pCTC36 expresses the same region of SIR4 as plasmid pCTC18 [55] except that the Sir4 hybrid is expressed from pGAD-R. pEDA195 was constructed by cloning a Pst I fragment of the SIR3 gene into vector pGAD424. Plasmid pAM2, expressing LexA-Esc4(1–480), and a clone expressing GAD-Slx4(383–748) isolated in the two-hybrid screen were used as a positive control in two-hybrid experiments summarized in Table 3.

Yeast strain construction

All strains are listed in Table 1. To make slx1 Δ, slx4 Δ, sir2 Δ, sir4 Δ, and sgs1 Δ mutants, PCR primers with 5' homology to sequences flanking these ORFs and 3' homology to sequences in a plasmids harboring selectable marker genes were used for PCR, generating targeting cassettes that were transformed into yeast as has been previously described [65]. esc4 Δ ::his5+ mutants were generated by the same method, but using a different plasmid as a template for PCR (gift of N. Dean, Stony Brook University). Strain YEA76 and its derivatives are derived from strain YSB35 [13].

Two-hybrid screening and direct tests

Screening was performed essentially as described [54, 63]. Plasmid pAM2, which expresses LexA-Esc4(1–480), (ADE2, TRP1, 2 μ) was co-transformed with approximately 1 μg of GAD library (LEU2, 2 μ; [66]) into strain L40 [54], which contains LexA binding sites upstream of both the HIS3 gene and the LacZ gene. The following specificity tests were performed: (1) His+ candidates were assayed for lacZ expression both after curing candidates of either the LexA or the GAD plasmid, (2) GAD hybrids were tested for interactions with nonspecific LexA hybrids (e.g. LexA-lamin) by mating candidates cured of the bait plasmid to LexA-containing AMR70 [54], and (3) the GAD hybrid plasmids which passed the aforementioned tests were subsequently retransformed into L40 along the original bait and tested for LacZ expression.

Two-hybrid tests done using the C-terminal BRCT motifs in Esc4 were performed by co-transforming L40 with plasmid pAM7, expressing LexA-Esc4(818–1070), and various GAD hybrids to silencing proteins, such as Sir1, Sir2, Sir3, Sir4, and Rap1. To test whether Sir3 bridged binding of LexA-Esc4(818–1070) to Sir4, a sir3Δ::kanMX6 derivative of L40 was generated (see yeast strains section for details), strain YRJ3, and the interaction was then retested in this strain. The control sir4 Δ-derivative of L40 used was strain YJL03.

DNA damage sensitivity tests

Cultures were grown in YPD medium ~18 h at 30°C and then serially diluted ten-fold, five times before being spotted onto plates containing YPD medium with MMS, HU, or no chemical. Cells were then incubated at 30°C and, in the case of MMS media, the plates were wrapped in aluminum foil.