In Saccharomyces cerevisiae, genes near telomeres are transcriptionally repressed, a phenomenon termed telomere position effect (TPE). Yeast telomeres cluster near the nuclear periphery, as do foci of proteins essential for TPE: Rap1p, Sir2-4p, and yKu70p/yKu80p. However, it is not clear if localization of telomeres to the periphery actually contributes to TPE. We examined the localization patterns of two telomeres with different levels of TPE: truncated VII-L and native VI-R. For both telomeres, localization to the nuclear periphery or to the silencing foci was neither necessary nor sufficient for TPE. Moreover, there was no correlation between TPE levels and the extent of localization. Tethering the truncated VII-L telomere to the nuclear periphery resulted in a modest increase in TPE. However, tethering did not bypass the roles of yKu70p, Sir4p, or Esc1p in TPE. Using mutations in RIF genes that bypass the role of Ku in TPE, a correlation between the level of silencing and the number of Rap1p foci present in the nucleus was observed, suggesting that Sir protein levels at telomeres determine both the level of TPE and the number of foci.
SEVERAL lines of evidence suggest that the nuclear periphery is a region conducive to transcriptional silencing. The classic example is the human inactive X chromosome, which localizes to the nuclear periphery (Bourgeois et al. 1985). More recent work in mammalian cells shows that activated genes localize to the nuclear interior whereas repressed genes preferentially reside at the periphery (Kosak and Groudine 2004; Zink et al. 2004). Perhaps the most compelling experiment arguing that this peripheral localization has functional significance for gene repression comes from yeast, where tethering a weakened HMR silencer to the nuclear periphery increases silencing (Andrulis et al. 1998). This view of the nuclear periphery as a potential silencing subcompartment is complicated by recent evidence that the periphery also promotes gene expression (Ishii et al. 2002; Brickner and Walter 2004; Casolari et al. 2004; Menon et al. 2005; Schmid et al. 2006; Taddei et al. 2006).
Some or all telomeres are localized to the nuclear periphery in yeasts, flies, humans, and in the pathogenic protozoa Trypanosoma brucei and Plasmodium falciparum. In each of these organisms, genes near telomeres are transcriptionally silenced, a phenomenon termed telomere position effect (TPE; reviewed in Mondoux and Zakian 2005). In Saccharomyces cerevisiae, many of the proteins required for TPE, such as Rap1p, Sir2p, Sir3p, Sir4p, and the heterodimeric Ku complex colocalize in three to six foci at the nuclear periphery (Klein et al. 1992; Palladino et al. 1993; Gotta et al. 1996; Laroche et al. 1998). Deleting any one of the Sir proteins, the Sir4p-binding Ku proteins, or the C-terminal Sir-interaction domain of Rap1p eliminates TPE and disperses these foci (Aparicio et al. 1991; Hecht et al. 1995; Gotta et al. 1996; Boulton and Jackson 1998; Laroche et al. 1998).
Although Rap1p, Ku, and Sir2-4p are required for TPE at all telomeres, different telomeres have different TPE phenotypes. Most native telomeres have very low or no detectable silencing, but genes near the native VI-R telomere are silenced in almost all cells (Mondoux and Zakian 2007, accompanying article, this issue). TPE has been best studied at truncated telomeres, where the reporter gene is placed immediately adjacent to the telomeric tract with the concomitant deletion of the subtelomeric middle repetitive elements (subtelomeric elements, STE). Although all truncated telomeres show TPE, the level of TPE at different truncated ends also varies. For example, truncated telomere VII-L has a TPE level ∼10-fold higher than truncated telomere V-R, even though both lack STEs (Gottschling et al. 1990). Although STEs contribute to the TPE phenotype of individual telomeres, even when the subtelomeric structure of the VII-L telomere is identical to that of the native VI-R telomere, the two telomeres have different TPE phenotypes (Mondoux and Zakian 2007).
Just as TPE levels vary from telomere to telomere, so do patterns of nuclear localization. By fluorescent in situ hybridization, ∼70% of the subtelomeric Y′ sequences localize to the Rap1p foci (Gotta et al. 1996), suggesting that many, but not all, telomeres localize to these foci. Several individual telomeres have been visualized at the periphery by inserting a lac or tet operator array near a single telomere and fusing their respective repressors to fluorescent proteins (Tham et al. 2001; Hediger et al. 2002; Taddei et al. 2004; Bystricky et al. 2005). This type of analysis has shown that individual telomeres are at the periphery in some but not all cells. Furthermore, the fraction of telomeres localized to the periphery varies between telomeres and with position in the cell cycle. The genetic requirements for localization to the periphery also differ between telomeres. For example, the VI-R telomere requires the Ku complex for localization to the nuclear periphery (Hediger et al. 2002), whereas truncated VII-L (Tham et al. 2001), truncated VI-R (Hediger et al. 2002), and native XIV-L (Taddei et al. 2004) do not.
Telomere positioning at the nuclear periphery is clearly not sufficient for TPE, since yKu70p is required for TPE at truncated VII-L (Boulton and Jackson 1998), but VII-L remains at the periphery in its absence (Tham et al. 2001). Nonetheless, other than late in the cell cycle (Tham et al. 2001; Hediger et al. 2002), there is no known case where telomeres are away from the periphery and silenced. Thus, localization to the nuclear periphery may be necessary for TPE. Another model consistent with the current data is that the association of telomeres with the nuclear periphery promotes TPE by bringing telomeres into close proximity to the Rap1 foci, and that localization to the foci (rather than to the nuclear periphery) is important for TPE.
It is not known why different telomeres display different levels of TPE. Given that different telomeres have different patterns and mechanisms of localization to the nuclear periphery, one possible model is that nuclear localization determines the TPE levels of individual telomeres. Here, we tested whether peripheral localization or localization to the Rap1p foci contributes to TPE at two different telomeres, truncated VII-L and native VI-R. These two telomeres have different levels of TPE: in our strain background and growth conditions, truncated VII-L is silenced in ∼15% of cells, whereas, in the same strain background, native VI-R has a very high level of TPE (∼85% TPE; Mondoux and Zakian 2007). These two telomeres also have different genetic requirements for localization to the nuclear periphery, leading to the hypothesis that they might localize differently to the periphery or peripheral Rap1p foci, and that this differential localization might explain the difference in their TPE levels. However, here we report that both telomeres localized equally well to both the nuclear periphery and the Rap1p foci, and their localization was independent of their TPE status.
Because nuclear localization could not explain different levels of TPE between the two telomeres, we examined the relationship between the nuclear periphery, peripheral Rap1p foci, and TPE at the truncated VII-L telomere in more detail. We focused on the VII-L telomere because its inherently lower TPE level makes it possible to determine if genes or conditions increase (as well as decrease) its silencing behavior. Physically tethering the truncated VII-L telomere to the nuclear periphery increased the fraction of cells in which the telomere was at the periphery and resulted in a modest increase in TPE. Remarkably, tethering did not increase localization of the truncated VII-L telomere to the Rap1p foci. Finally, we examined the requirements for the formation of the foci of silencing proteins. The Ku complex is resident in these foci, and foci are dispersed in ykuΔ cells (Laroche et al. 1998). However, Ku is not required for focus formation in certain genetic backgrounds, as deleting RIF1 in a yku70Δ strain restores focus formation (Hediger et al. 2002). Deleting the RIF genes in yku70Δ cells also restores TPE (Mishra and Shore 1999). We find that the number of Rap1p foci per nucleus correlated with the level of TPE at truncated telomere VII-L, as deleting both RIF1 and RIF2 in a yku70Δ strain resulted in higher TPE levels and more foci per nucleus than rif1Δ yku70Δ alone. We hypothesize, therefore, that the level of Sir protein binding at the telomeres determines both TPE levels and Rap1p focus formation.
MATERIALS AND METHODS
Yeast strains and plasmids for TPE and visualization:
Both the strain with the truncated VII-L telomere (Tham et al. 2001) and the strain with “native” VI-R telomere used for visualization and TPE experiments were constructed in the YPH background (ura3-52 lys2-801 ade2-101 trp1-Δ63 his3-Δ200 leu2-Δ1; Sikorski and Hieter 1989). To construct the strain with the native VI-R telomere, the plasmid pAFS52-LacOp-ARS609 (Hediger et al. 2002) containing an 8-kb lac operator array (Robinett et al. 1996) was integrated at ARS609, placing the lac operator array ∼15 kb from the VI-R telomere. A 2-μm plasmid expressing the C terminus of Sir4p (pCTC23; Chien et al. 1991) was introduced into the strain to reduce TPE. The URA3 TPE reporter was introduced at the VI-R telomere in a manner analogous to the creation of native TPE reporter strains described in (Pryde and Louis 1999). To construct the VI-R TPE reporter, URA3 was amplified from ADH4UCAIV, the same plasmid used to create the truncated VII-L telomere reporter (Gottschling et al. 1990), with primers 6RURAF5 (5′ atatagtatgctcacattttcttattgctgaatagttcttttttacgtttagctgggattcggtaatctccgagcagaag 3′) and 6RURAR4 (5′ atatagtatgctcacattttcttattgctgaatagttcttttttacgtttagctggggtgttgaagaaacatgaaattgcc 3′). Transformants were screened by Southern blotting and confirmed using pulsed-field gel electrophoresis. The resulting strain containing the native VI-R telomere was then restreaked several times on plates with rich media (YEPD) to lose the pCTC23 plasmid. LacI-GFP was introduced into the strain with plasmid pMAM6, which contains LacI-GFP under the control of the His promoter and was constructed from pWHTLacI (Tham et al. 2001) and pRS305 (Sikorski and Hieter 1989).
The truncated VII-L strain used in the tethering experiments contains upstream activation sequence (UAS) sites and a URA3 reporter gene adjacent to the truncated VII-L telomere in a W303 background strain (strain YDS 634; Chien et al. 1993) and has been modified to contain the LacO/LacI-GFP visualization system (Tham et al. 2001) and GalBD–Yif1p tethering system (Andrulis et al. 1998). All experiments using the GalBD–Yif1p tethering system were performed on at least two independent plasmid transformants.
SIR4, YKU70, RIF1, and RIF2 were deleted in the tethering strain background using a PCR-mediated knockout that eliminated the complete open reading frame, replacing it with either a kanamycin- (Wach et al. 1994), hygromycin-, or nourseothricin-resistance cassette (Goldstein and McCusker 1999). SAS2 and ESC1 were deleted in these strain backgrounds by backcrossing several times to the mutants from the deletion strain collection (Research Genetics/Invitrogen, Carlsbad, CA). All deletion strains were verified by PCR and TPE phenotypes. The MEC1 and SIR1 mutant strains have been described previously (Goudsouzian et al. 2006; Mondoux and Zakian 2007).
Immunofluorescence and microscopy:
The protocol for immunofluorescence was adapted from Tham et al. 2001. For visualization of a telomere and the nuclear envelope, the primary antibodies were mouse MAb414 (anti-p62; Covance, Berkeley, CA) and rabbit anti-GFP (Chemicon/Millipore, Temecula, CA). For visualization of a telomere and the Rap1p foci, the primary antibodies were rabbit anti-Rap1p (made by W.H. Tham using methods described in Conrad et al. 1990) and mouse anti-GFP (Chemicon/Millipore, Temecula, CA). Secondary antibodies were conjugated to Alexa 488 and Alexa 546 (1:100 in PBS/BSA; Molecular Probes, Eugene, OR).
Slides were imaged using a DeltaVision platform and a Nikon microscope with a 100× objective (Applied Precision, Issaquah, WA). Cells were optically sectioned into 18 0.15-μm slices and deconvolved using the DeltaVision restoration system. A mercury arc lamp was filtered at 546 nm and 488 nm to detect the AlexFluor fluorophores. For the Rap1p colocalization experiments, a telomere was scored as colocalized to a focus if there was any overlap between the two spots, i.e., yellow staining (combination of the green telomere spot and red Rap1p spot). For nuclear envelope distance measurements, the diameter of the cell was measured at least twice using the DeltaVision software, and the larger diameter used to calculate the radius, which was then used to divide the area of the nuclear section into three equal zones (as in Hediger et al. 2002). The distance from the center of the telomere spot to the closest point of anti-p62 staining was then measured and the telomere's position assigned to zone I, II, or III. Calculation of telomere localization was based on at least three independent experiments with at least 50 nuclei counted per condition, per experiment. For all experiments, standard deviations were calculated and the significance of the data was assessed using the Student's t-test. When counting the number of foci per cell, at least 50 cells were counted for each mutant in two independent experiments.
Localization to the nuclear periphery is not correlated with silencing status:
The truncated VII-L and native VI-R telomeres have different levels of TPE (Gottschling et al. 1990; Mondoux and Zakian 2007) and different mechanisms for perinuclear localization (Tham et al. 2001; Hediger et al. 2002). To determine whether their different levels of TPE correlated with different patterns of nuclear localization, and whether localization to the nuclear periphery was necessary for TPE, we monitored nuclear localization and silencing at each telomere.
A strain for monitoring TPE (via a URA3 reporter) and visualization (via the lac operator/LacI-GFP system) of the truncated VII-L telomere has been described previously (Tham et al. 2001; Figure 1). At telomere VI-R, URA3 was integrated into the X element in a manner that largely retains its structure (Mondoux and Zakian 2007; Figure 1). The lac operator array was integrated ∼15 kb from the VI-R telomere (Hediger et al. 2002; Figure 1), and LacI-GFP was introduced into the strain. Previous data (Tham et al. 2001) demonstrated that under all growth conditions, the fraction of cells in which the truncated VII-L telomere localizes to the nuclear periphery does not correlate with the fraction of cells in the population that exhibits TPE. Therefore, TPE levels at truncated VII-L do not correlate with nuclear position. However, it is possible that nuclear position correlates with TPE levels between telomeres. This model predicts that telomeres with the highest levels of TPE, such as the VI-R telomere, would be at the periphery more often than telomeres that have lower levels of TPE, like the truncated VII-L telomere.
To test this hypothesis, we examined the nuclear localization of the truncated VII-L and native VI-R telomeres. Cells were grown in medium lacking uracil, in which the reporter gene is expressed in essentially all cells (0% TPE) or in medium containing 5-fluoroorotic acid (5-FOA; Boeke et al. 1987), in which the URA3 reporter gene is silenced in essentially all cells (∼100% TPE; Figure 2A). We also grew cells in complete medium (YC + Ura) in which the telomeric URA3 gene was repressed in only a subset of cells (∼15% VII-L; ∼85% VI-R). The truncated VII-L and native VI-R telomere reporter strains thus provide a system for examining a complete range of silencing states, from 0 to 100% silencing, at two different telomere ends with different subtelomeric structures and different mechanisms for nuclear localization.
Telomere position relative to the nuclear periphery was determined in formaldehyde-fixed cells using an antibody to a nuclear pore protein as a marker for the nuclear periphery and an antibody to GFP as a marker for the telomere. For each cell, both the distance between the telomere spot and the nuclear periphery and the diameter of the nucleus were measured. The area of the nuclear section was then calculated and divided into equal thirds, and the telomere was scored as localizing to zone I, II, or III, with zone I being defined as the peripheral zone (as in Hediger et al. 2002). Localization to the periphery did not correlate with subtelomeric structure, as the VII-L and VI-R localization patterns were identical. Localization of both telomeres to the peripheral zone I occurred in ∼75% of cells, regardless of silencing status (Figure 2C). For example, in the FOA-grown cells, ∼25% of the telomeres were silent but away from the periphery, and in the minus uracil-grown cells ∼75% of the telomeres were at the periphery but expressed.
Localization to Rap1p foci does not correlate with TPE levels:
Given that peripheral localization of telomeres did not correlate with TPE levels (Figure 2C), we reasoned that a subnuclear compartment, rather than the periphery itself, might promote silencing. The Rap1p foci, which contain the silencing proteins, are an obvious candidate for a subcompartment to which silent telomeres might be preferentially localized. In this model, a telomere would have a certain probability of being associated with the periphery regardless of transcriptional state, but when transcriptionally repressed and at the periphery, it would localize to a Rap1p focus (and vice versa). When the subtelomeric gene is expressed, the telomere could be at the periphery or not, but, if at the periphery, it would be less likely to colocalize with a Rap1p focus than when it is repressed.
To test this model, we determined the frequency of localization of both the truncated VII-L and native VI-R telomeres to the silencing foci, using a Rap1p antibody to visualize the foci. Both the VII-L telomere and the VI-R telomere localized to a Rap1p focus in ∼60% of the cells (Figure 3B). This localization pattern was identical for both telomeres in all silencing states, indicating that localization to the Rap1p foci did not correlate with subtelomeric structure or TPE levels. Thus, localization of telomeres to the foci is neither necessary nor sufficient for TPE.
Tethering a telomere to the nuclear envelope modestly improves TPE:
If localization of telomeres to the nuclear periphery promotes TPE, physically tethering the truncated VII-L telomere to the nuclear envelope should improve its TPE phenotype. To tether the VII-L telomere to the nuclear periphery, four GAL-UAS sites were inserted between the telomere and the URA3 reporter gene (Chien et al. 1993). A Gal DNA-binding domain–Yip1 interacting factor (Yif1p) fusion protein was expressed in these cells (Andrulis et al. 1998). When a weakened HMR locus is tethered to the periphery using this system, silencing was increased 10-to 100-fold compared to the control Gal binding domain alone (from silencing in <0.01% of cells to 0.1–1%; Andrulis et al. 1998).
To demonstrate that the truncated VII-L telomere was physically tethered to the nuclear envelope, we measured the distance between the telomere and the nuclear envelope in fixed cells in strains expressing the GalBD–Yif1p fusion protein and in strains expressing only the Gal DNA binding domain. The tethered strain had a significantly greater proportion of VII-L telomeres that were colocalized to the nuclear envelope (75%; Figure 4C) compared to the GalBD strain (53%; P < 0.04). Thus, the truncated VII-L telomere can be physically tethered to the nuclear envelope via Yif1p.
Tethering increased TPE significantly, ∼3.5-fold over the strain that expressed only the Gal binding domain (P < 0.0005; Figure 4B). However, despite this increase in TPE, the tethered VII-L telomere still had lower TPE than the VI-R telomere (∼58% TPE vs. ∼85% TPE). We tested whether the increased silencing at the tethered VII-L telomere was a result of increased localization to the Rap1p foci. The VII-L telomere colocalized with the Rap1p foci in ∼50% of the cells in both the tethered strain and the empty vector strain in all silencing states (Figure 4D). Thus, the increase in TPE observed when the truncated VII-L telomere was tethered to the nuclear periphery was not due to a measurable increase in localization to the Rap1p foci.
Tethering does not result in longer telomeres or bypass the requirement for yKu70p, Sir4p, or Esc1p in TPE:
Longer telomeres confer increased TPE because they have more binding sites for Rap1p, which can then recruit more Sir proteins to the telomere (Kyrion et al. 1993). Thus, another possibility for the TPE increase between the tethered and untethered VII-L telomeres is that localization to the nuclear periphery results in telomere lengthening. We tested this possibility by examining the length of the tethered VII-L telomere by Southern blotting. Surprisingly, the tethered VII-L telomere was actually slightly shorter in the tethered strain (∼75 bp; Figure 5), compared to GalBD alone (Figure 5) or a no-plasmid control (data not shown). This telomere shortening was specific to the tethered telomere, as bulk telomere length was unchanged. The length of the tethered VII-L telomere was stable over ∼75 generations, in contrast to the shortening phenotype observed when cells lack components of the telomerase holoenzyme (for example, est3Δ; Figure 5). These data demonstrate that increased telomere length does not explain the increase in TPE observed at the tethered telomere.
Since TPE was increased at the tethered VII-L telomere despite its shorter length, it seemed possible that this increase was the result of a bypass of factors normally required for TPE. Both the Sir proteins and Ku complex are brought to the telomere at least in part by interactions with Rap1p (Moretti et al. 1994; Tsukamoto et al. 1997; Moretti and Shore 2001). Deletion of either the Ku complex (Boulton and Jackson 1996, 1998) or Sir4p (Aparicio et al. 1991) abolish TPE at the truncated VII-L telomere. Sir4p also binds to Esc1p, an inner nuclear membrane protein (Andrulis et al. 2002). Together, the Ku complex, Sir4p, and Esc1p function in redundant peripheral localization pathways (Hediger et al. 2002; Taddei et al. 2004). We tested whether tethering the truncated VII-L telomere to the nuclear periphery bypassed the role of these proteins in TPE by comparing TPE levels in the mutant and wild-type versions of the tethered and empty vector strains. In the absence of YKU70 or SIR4, TPE was abolished at both the tethered telomere and the telomere bound to GalBD alone (Figure 6). Therefore, tethering a telomere to the nuclear periphery does not increase TPE by bypassing the requirement for either the Ku or Sir proteins in TPE. The increase in HM silencing observed by tethering is also Sir4p dependent (Andrulis et al. 1998).
Unlike the deletion of YKU70 or SIR4, deletion of ESC1 decreases, but does not abolish, TPE at the truncated VII-L telomere (Andrulis et al. 2002). If tethering bypassed the requirement for Esc1p in TPE, we would expect TPE levels to be the same in the tethered wild-type strain and the tethered esc1Δ strain. In agreement with previous results, TPE at the truncated VII-L telomere was decreased in the absence of Esc1p (Figure 6). In addition, although TPE increased when the telomere was tethered, the tethered wild-type strain had higher levels of TPE than the tethered esc1Δ strain (Figure 6), indicating that tethering also does not bypass the role of Esc1p in TPE.
The presence of the Rap1p foci correlates with the potential for TPE:
We found no evidence for a correlation between a telomere's localization to the Rap1p foci and its level of TPE (Figures 3 and 4). However, the model that telomere clustering in the Rap1p foci creates a high local concentration of silencing factors that facilitates TPE is still attractive as, to date, no mutation or condition has been reported that supports TPE in the absence of the foci. We therefore tested whether eliminating other proteins that affect telomeres would disrupt the Rap1p foci yet maintain TPE. Sir1p, Mec1p, and Sas2p are required for wild-type levels of TPE at some telomeres and under some conditions (Reifsnyder et al. 1996; Pryde and Louis 1999; Craven and Petes 2000; Mondoux and Zakian 2007). Using a Rap1p antibody to visualize the foci (as in Figure 3), we observed that Rap1p foci formed in sir1Δ, mec1Δ, and sas2Δ cells (data not shown). The Rap1p foci are known to disperse upon DNA damage (McAinsh et al. 1999). Therefore, we also examined whether foci would form in xrs2Δ and tel1Δ cells, both of which maintain wild-type or slightly reduced TPE (Runge and Zakian 1996; Boulton and Jackson 1998). Rap1p foci were present in both xrs2Δ and tel1Δ strains (data not shown).
The number of silencing foci correlates with TPE levels:
Unlike the Sir4p-binding proteins Sir1p and Esc1p, the Sir4p-binding Ku complex is absolutely required for TPE in wild-type cells (Boulton and Jackson 1998), and its deletion disperses the Rap1p foci (Laroche et al. 1998). Like the Sir proteins, Rif1p and Rif2p bind to the carboxyl terminus of Rap1p, and this binding is mutually exclusive with binding of the Sir complex (Moretti et al. 1994; Wotton and Shore 1997). rif1Δ rif2Δ strains have very long telomeres (Moretti et al. 1994; Wotton and Shore 1997), and this lengthening is telomerase dependent (Teng et al. 2000). Thus, when Rif proteins are absent, more Sir proteins can bind to telomeres both because telomeres are longer and because the Sir complex no longer competes with Rifs for binding to telomere-bound Rap1p. Due to higher Sir binding, rif cells have increased levels of TPE (Kyrion et al. 1993). Silencing can be restored to wild-type levels at the truncated VII-L telomere even in a yku70Δ mutant by deleting both the RIF1 and RIF2 genes. Moreover, a low level of TPE (∼0.1%) is seen when RIF1 alone is deleted in a yku70Δ strain (Mishra and Shore 1999).
In agreement with previous results (Mishra and Shore 1999), the deletion of RIF1or both RIF1 and RIF2 restored TPE in a yku70Δ strain (Figure 7A). The TPE level in yku70Δ rif1Δ cells was low (∼0.1%), while TPE in the yku70Δ rif1Δ rif2Δ strain was close to wild type (∼8% in yku70Δ rif1Δ rif2Δ vs. ∼23% in wild type, P < 0.05; Figure 7B). As reported previously, the Rap1p foci were dispersed in yku70Δ cells (Laroche et al. 1998; Figure 7C) and restored in yku70Δ rif1Δ cells (Hediger et al. 2002; Figure 7C). As expected, foci were also restored in a yku70Δ rif1Δ rif2Δ strain (Figure 7C), indicating that Ku is not essential for the formation of Rap1p foci in this background. In addition, the number of Rap1p foci per nucleus correlated with the level of TPE. In wild-type haploid cells, most nuclei contain three to six Rap1p foci (Klein et al. 1992), a result confirmed here (Figure 7D). TPE in the yku70Δ rif1Δ rif2Δ strain was reduced threefold compared to the wild-type strain (Figures 7, B and C). The number of foci per nucleus in this strain was also significantly different from wild type (P < 4.4 × 10−8). Although most of the nuclei had three to six foci, no nuclei were observed with more than six foci (0% vs. 13% in wild type), and more nuclei were observed with fewer than three foci (28% vs. 14% in wild type). TPE in the yku70Δ rif1Δ mutant was reduced >100-fold compared to wild type (Figure 7A), and most of the nuclei in this mutant contained only one to four foci (Figure 7D). Moreover, no yku70Δ rif1Δ nuclei had more than six foci (0% vs. 13%), and almost half had fewer than three foci (46% vs. 14% in wild type). This distribution was significantly different from wild type (P < 3.2 × 10−16).
We investigated whether there are differences in the nuclear localization of the truncated VII-L and native VI-R telomeres that could explain their different TPE phenotypes. However, both telomeres localized equally well to the nuclear periphery in both silent and expressed states (Figure 2C). Likewise, both telomeres localized equally well to the Rap1p foci, and this localization was also independent of transcriptional state (Figure 3B). Even when tethered to the nuclear periphery, the truncated VII-L telomere was not associated more often with Rap1p foci, and its silencing level was still lower than that of the native VI-R telomere (Figure 4). Likewise, silencing of the HMR locus does not require its localization to the nuclear periphery, as a plasmid-borne HMR locus is silent and away from the nuclear envelope in a ykuΔ esc1Δ strain (Gartenberg et al. 2004). However, HM silencing occurs in both ykuΔ and ykuΔ esc1Δ cells in which there are no Rap1p foci (Laroche et al. 1998; Maillet et al. 2001; Gartenberg et al. 2004) while neither strain is competent for TPE (Boulton and Jackson 1998; Gartenberg et al. 2004). We find no correlation between telomere localization and TPE levels even in cells competent for TPE with intact Rap1p foci. Our data seem to rule out models in which telomere placement at the periphery serves as a mechanism to maintain transcriptional repression by bringing telomeres in close proximity to silencing proteins. In addition, our data argue against models in which placement near the nuclear periphery promotes TPE by other mechanisms. For example, our data do not support a model in which association with the nuclear periphery constrains telomere mobility in a manner that makes it more difficult for RNA polymerase to transcribe through a telomere-linked gene.
Although the Rap1p foci did not constitute a nuclear subcompartment that enhanced silencing via colocalization with telomeres, the presence of the Rap1p foci correlated with the potential for silencing. Thus, a possible model is that the integrity of the Rap1p foci is required for TPE. The foci could act as heterochromatin-assembly “factories,” and a telomere's localization to a focus might be necessary but not sufficient to establish, but not to maintain, TPE. Alternatively, localization with a focus might be necessary for maintenance as well, but the amount of time required for the telomere to “visit” the focus might be too brief to distinguish in our assay.
Our favored model for the restoration of the foci of silencing proteins in the yku70Δ rifΔ mutants is that, as is the case for TPE, it is not the presence of the Ku complex or Rap1p at the telomere per se, but the amount of Sir2, Sir3, and Sir4 protein they can recruit to the telomere that is responsible for the formation of the foci. Ku was not required for focus formation in rif strains that could recruit sufficient levels of Sir proteins to telomeres via a Ku-independent pathway (Figure 7C; Hediger et al. 2002). Likewise, Ku is not needed for TPE in these strains (Figure 7A; Mishra and Shore 1999). In the model we propose, the loss of Ku or deletion of the Rap1p C terminus disperses the foci (Hecht et al. 1995; Laroche et al. 1998) because of the concomitant loss of Sir proteins at the telomere. Deleting the Rif proteins in the yku70Δ background restores the foci because of the concomitant increase in Sir proteins at the telomere. Therefore, the presence of the foci themselves need not be required for TPE. Rather, the conditions that create the foci (i.e., the level of Sir-protein binding to telomeres) are the same conditions that permit TPE. This model would predict that no mutations could bypass the requirement for the Sir proteins in either TPE or focus formation, and no such mutations have been identified. Our finding that TPE levels correlated with the number of Rap1p foci per nucleus is also consistent with the idea that more Sir binding at the telomere contributes to both focus formation and TPE levels.
Despite the lack of correlation between the nuclear periphery and silencing status in wild-type strains, tethering the truncated VII-L telomere to the nuclear periphery increased TPE (Figure 4C). However, the tethered VII-L telomere was still silenced in fewer cells (58%) than the untethered native VI-R telomere (85%). Likewise, tethering a weakened HMR locus to the periphery increases its silencing level, but its new level is still much lower than that for a wild type, untethered HMR locus (Andrulis et al. 1998). Improved silencing of the tethered VII-L telomere was not due to bypassing the functions of three proteins, yKu70p, Sir4p, and Esc1p, that affect both TPE and nuclear localization (Figure 6). In addition, this increase in TPE was not due to a measurable increase in localization to Rap1p foci (Figure 4D), a surprising result given that Rap1p foci are also at the nuclear periphery, or to an increase in telomere length (Figure 5).
The data presented here show that localization to the nuclear periphery or to Rap1p foci does not correlate with level of TPE. Furthermore, localization to the nuclear periphery or to the Rap1p foci was neither necessary nor sufficient for silencing of either the VI-R or the VII-L telomere (Figures 2 and 3). Telomeres can be away from the periphery and be silent, suggesting that localization is not important for silencing. However, these data do not rule out a model in which an individual telomere must visit the periphery to establish or maintain TPE. Neither subtelomeric sequence (Mondoux and Zakian 2007) nor, as shown here, nuclear localization can explain the difference in TPE levels between the native VI-R and the truncated VII-L telomeres. These results do not exclude the possibility that trans mechanisms other than nuclear localization, like higher order chromosome structure or dynamics, could play a role in determining the TPE phenotypes of individual telomeres. There could also be cis influences on TPE, for example, proximal sequences that promote or repress TPE, or transcriptional activity from nearby genes. The VI-R telomere has a high level of TPE, which may be due in part to the repression of its most proximal gene, YFR057w (Wyrick et al. 1999; Vega-Palas et al. 2000). Nucleosome spacing could also influence preferential recruitment of the Sir proteins to particular telomeres, as it influences directional silencing at the silent mating-type loci (Zou et al. 2006).
Localization of telomeres to the periphery may be important for a telomere function other than silencing. Peripheral localization of telomeres does not determine their late replication (Hiraga et al. 2006), but it could influence recombination, as the Ku complex regulates telomere–telomere recombination as a maintenance mechanism in the absence of telomerase (Tsai et al. 2002). Furthermore, double-strand breaks in the XI-L subtelomeric region cannot be repaired if its localization to the periphery is eliminated via disruption of the Nup84 complex (Therizols et al. 2006). In mammalian cells, localization away from the periphery correlates with large-scale compaction at the IgH locus (Kosak et al. 2002), and telomere localization could also be linked to changes in higher-order chromosome structure. When the truncated VII-L telomere was tethered to the nuclear envelope, it was shorter than the telomere bound to the Gal DNA binding domain alone (Figure 6). Therefore, the nuclear localization of telomeres may play a role in length regulation either because the telomeres themselves must leave the periphery in order to be lengthened by telomerase or because active telomerase is not available in particular nuclear subcompartments.
We thank Susan Gasser, David Shore, and Rolf Sternglanz for providing strains and plasmids and James Broach, Paul Schedl, and Eugenia Xu for comments on the manuscript. This work was supported by National Institutes of Health grants to V.A.Z. and by a National Science Foundation predoctoral fellowship to M.A.M.
Communicating editor: F. Winston
- Received August 2, 2007.
- Accepted October 5, 2007.
- Copyright © 2007 by the Genetics Society of America