In Saccharomyces cerevisiae, efficient silencer function requires telomere proximity, i.e. compartments of the nucleoplasm enriched in silencing factors. Accordingly, silencers located far from telomeres function inefficiently. We show here that cells lacking yKu balance between two mitotically stable states of silencing competence. In one, a partial delocalization of telomeres and silencing factors throughout the nucleoplasm correlates with enhanced silencing at a non‐telomeric locus, while in the other, telomeres retain their focal pattern of distribution and there is no repression at the non‐telomeric locus, as observed in wild‐type cells. The two states also differ in their level of residual telomeric silencing. These findings indicate the existence of a yKu‐independent pathway of telomere clustering and Sir localization. Interestingly, this pathway appears to be under epigenetic control.
The gene repression at specific loci plays a critical role in various cell programs. In the yeast Saccharomyces cerevisiae, silencing at telomeres and at the HML and HMR silent mating‐type loci depends upon a proper intranuclear distribution of silencing factors (reviewed in Gotta and Gasser, 1996). At the silent mating‐type loci, silencing is directed by cis‐acting elements, termed the E and I silencers, while for telomeric position effect (TPE), multiple Rap1p molecules bind the telomeric TG1–3 repeats to nucleate silencing. Genetic studies have identified a number of trans‐acting factors that are essential to silencing, including three silent information regulator (Sir) proteins (Sir2, 3 and 4) and the histones H3 and H4. Silencers and telomeres promote the formation of nucleoprotein complexes that are believed to nucleate the spreading of silent chromatin outward. The distance covered is limited by the dosage of certain trans‐acting silencing factors and by cis‐acting insulators (Strahl‐Bolsinger et al., 1997; Donze et al., 1999; Fourel et al., 1999; Pryde and Louis, 1999).
Rap1p, Sir3p and Sir4p are enriched in a limited number of foci located near the nuclear envelope that colocalize with clustered telomeres (Gotta et al., 1996). The unequal distribution of these factors within the nucleoplasm delimits subnuclear compartments which are competent for silencing. Indeed, the focal pattern of Rap1 and Sir proteins is perturbed in a range of mutants that impair silencing (Palladino et al., 1993; Cockell et al., 1995; Hecht et al., 1995). Furthermore, the capacity of a silencer to repress transcription is dependent upon its proximity to telomeres and to the nuclear periphery (Maillet et al., 1996; Marcand et al., 1996; Andrulis et al., 1998). The weakness of silencers in a non‐telomeric, ectopic context is best explained by a sequestration of most of the available Sir3 and Sir4 proteins at telomeric clusters. Interestingly, aging (Kennedy et al., 1997) and the response to double‐strand breaks (Martin et al., 1999; McAinsh et al., 1999; Mills et al., 1999) induce a redistribution of telomeric factors within the nucleus. The formation of a silencing compartment at telomeres depends upon both the capacity of silencing factors to bind telomeric components and an appropriate subnuclear localization of telomeres. Interestingly, the yKu complex appears to be involved in both these processes.
The yKu heterodimer binds to chromosome ends, and deletions of either YKU70 or YKU80, which encode the two yKu subunits, affect the configuration of telomeric DNA by decreasing its size and increasing its single‐strand character (Porter et al., 1996; Boulton and Jackson, 1998; Gravel et al., 1998). Furthermore, in cells lacking yKu, subtelomeric silencing is severely compromised (Gravel et al., 1998; Laroche et al., 1998; Nugent et al., 1998; Fourel et al., 1999; Mishra and Shore, 1999; Pryde and Louis, 1999; Galy et al., 2000). Interestingly, in yKu‐deficient cells, the perinuclear localization of telomeric DNA, Sir proteins and Rap1p is lost, such that all have a more dispersed pattern throughout the nucleoplasm (Laroche et al., 1998). The delocalization of telomeres cannot be interpreted solely as a silencing defect, since the deletion of SIR3 or SIR4 does not greatly modify the focal pattern of telomeric DNA (Gotta et al., 1996). Rather, it appears that yKu helps tether telomeric DNA to the nuclear envelope (Laroche et al., 1998; Galy et al., 2000).
Since the lack of ectopic silencing in wild‐type cells is likely to reflect competition between telomeres and internal silencers for the pool of Sir proteins, we reasoned that the release of Rap1p and Sir from telomeres in yKu‐deficient mutants should lead to efficient ectopic silencing. Our results indeed confirm the existence of a subpopulation of yKu− cells that is competent for ectopic silencing. Surprisingly, these cells can switch to another state which is incompetent for ectopic silencing and has a more wild‐type subnuclear distribution of telomeric components.
An ADE2 expression color assay for ectopic silencing
In order to monitor silencing at non‐telomeric or ectopic sites, we took advantage of the fact that the ADE2 gene expression can be visualized at the clonal level by a colony color assay (Gottschling et al., 1990). When an ADE2 gene bracketed by two silencers is integrated at the non‐telomeric LYS2 locus (Figure 1A), the resulting colonies are white, indicative of full ADE2 expression (Figure 1B, lines a–c). Overexpression of Sir3p, or Sir3p and Sir4p, induces ectopic silencing, as visualized by red or pink colony color (Figure 1B, lines d and e). The overexpression of Sir4p alone also results in an increase in ectopic silencing but to a lesser extent than Sir3p (Figure 1B, line f). These results confirm that Sir‐dependent silencing is inoperative when silencers are integrated far from a telomere, but can be revealed by increasing the dosage of Sir proteins (Maillet et al., 1996). Using the same ADE2/silencer construct, it was shown that the release of telomeric Sir proteins upon DNA damage also partially restores ectopic silencing (Martin et al., 1999).
Inactivation of yKu results in a variegated ectopic silencing
Seven out of 17 yku70::kanMX4 (hereafter yku70) primary transformants of a W303‐1B derivative strain carrying the ADE2/silencer construct (LM11) formed pink colonies, while 10 other transformants produced white colonies, indistinguishable from the parental strain (data not shown). These were obtained in three independent experiments and were confirmed as genomic null alleles by Southern blot. The ectopic silencing of ADE2, as revealed by the pink color of some yku70 transformants of LM11, was lost upon overexpression of the C‐terminal domain of Sir4p (data not shown), consistent with a negative trans‐dominant effect on Sir‐dependent silencing (Cockell et al., 1995), or upon deletion of SIR4 (F. Hediger, personal communication). These data confirm that ADE2 repression in pink yku70 cells reflects a genuine Sir‐mediated silencing. Importantly, the ability to produce two color phenotypes in a YKU70 disrupt is not restricted to a W303‐1B genetic background: when the ADE2/silencers construct is inserted at LYS2 in a S150‐2B derivative strain (EG204), 7 out of 9 primary yku70 transformants analyzed are pink (data not shown).
The ectopic silencing detected in pink yku70 cells is not due to a general improvement of silencer‐mediated repression. Indeed, a HML‐I silencer deleted for the Abf1‐binding site, known to confer an attenuated silencing in wild‐type cells (Boscheron et al., 1996), displays a further decreased silencing capacity in yku70 cells (data not shown). Therefore, the effects of yKu deficiency on silencer‐mediated repression appear to be the opposite at HML and at LYS2: a leaky silencer becomes weaker at HML, while wild‐type silencer function is stronger at LYS2. Since yku70 disruption leads to the partial delocalization of telomeric DNA and silencing factors throughout the nucleoplasm (Laroche et al., 1998), and since HML is sequestered at the nuclear periphery (Laroche et al., 2000), this difference might be explained by a decrease in the concentration of silencing factors at HML concomitant with an elevated dosage of these factors at non‐telomeric locations.
Several arguments indicate that the pink/white phenotypes arise from a loss‐of‐function of yKu. (i) The repression at LYS2 is dependent upon the absence of YKU70, since the introduction of a complementing YKU70 gene reverts the pink phenotype to white (Figure 1B, lines g and h). (ii) The pink or white phenotypes can be observed when YKU80 instead of YKU70 is disrupted (data not shown). (iii) The yku70yku80 double disruption also gives rise to both pink and white colonies (data not shown). (iv) Back‐crosses with either white or pink clones produced white diploid colonies, and their subsequent sporulation leads to both pink and white yku70 colonies with a majority of white (Figure 2A). The fact that yKu− spores exhibit both phenotypes shows that the pink/white phenotype is unstable through meiosis, and suggests that there is no secondary mutation, unlinked to the YKU70 locus, that is responsible for promoting ectopic silencing.
Upon serial replica plating steps, the patches corresponding to pink and white primary transformants tend to remain pink or white. We consider these cells as the first generation after yku70 disruption (G1). The colonies isolated after re‐streaking the G1 patches define the second generation (G2). Importantly, white or sectored G2 colonies can be recovered from pink G1 cells and pink or sectored G2 colonies can be recovered from white G1 cells (Figure 2B). Starting from G2 cells, G3 colonies with a switched color can also be recovered (Figure 2B). Therefore, although the color phenotype of yku70 cells seems stable in populations of cells cultivated in patches, switching events from white to pink or from pink to white can be observed upon re‐streaking the original transformants. The switch rate in LM79 was determined by using the half‐sector assay (Hieter et al., 1985), leading to values of ∼0.6% per cell division from pink to white and ∼0.4% from white to pink.
Alternative states of telomeric organization in yku70 cells
The above data suggest that a yku70 strain can comprise two populations of cells, as indicated by two alternative states of the ectopic silencing reporter gene. Therefore, we asked whether either the pink or the white phenotype correlates with any of the other well‐characterized phenotypes of yku70 mutants.
Pink and white colonies do not exhibit any difference in their thermosensitivity (our unpublished data). When the mean length of Y′‐associated telomeres or the URA3‐associated VII‐L telomere was measured, no change could be detected between pink and white yku70 cells obtained from either LM11 or EG204. In addition, telomeric DNA of both pink and white yku70 asynchronous populations of cells exhibits about the same increase in single‐stranded character when compared with wild‐type cells (data not shown). Finally, immunoblotting confirms that there is no difference in the steady‐state levels of Sir or Rap1 proteins in white and pink yku70 cells (our unpublished data).
Strikingly, the pink yku70 or yku80 cells exhibit a higher level of residual telomeric silencing when compared with white colonies [Figure 1B, compare the % of fluoro‐orotic acid (FOA)‐resistant colonies of lines g or i with m or o, and Figure 3], although TPE is still largely compromised in both. This result suggests that telomeres in pink and white colonies behave slightly differently, at least with respect to their residual silencing capacity.
A further function proposed for yeast Ku is to anchor telomeres at the nuclear periphery, which helps maintain pools of Rap1p and Sir proteins at these sites (Laroche et al, 1998). Immunofluorescence and fluorescence in situ hybridization (FISH) with a subtelomeric Y′ repeat indicate that this function is also differentially affected in white and pink yku70 cells (Figure 4). As expected, immunostaining for Sir4p and Rap1p clearly shows a delocalized pattern of silencing factors in the pink yku70 populations (Figure 4, compare panels a and b, d and e). Surprisingly, in the yku70 white population, the dispersed signal for Sir4p and Rap1p is weaker and is often dominated by one very large focus of silencing proteins (visible as a bright green spot in Figure 4, panels c and f). This bright spot is often juxtaposed to the nuclear envelope (visualized here by anti‐pore staining in red). The different distribution of silencing factors in the pink and white isolates of LM79 correlates well with the reduced availability of Sir proteins for internal silencing in the yku70 white colonies.
When the position of telomeric DNA is probed in these same strains using a subtelomeric Y′ repeat probe, we see that telomeric foci appear more dispersed in pink as compared with white yku70 cells (Figure 4, panels g–l). This qualitative difference was confirmed by scoring foci in a large number of nuclei, which reveals a decrease in the percentage of nuclei with only peripheral Y′ foci and a lower number of total peripheral Y′ signals in pink as compared with white cells (Table 1). In the white yku70 cells we also occasionally see one focus that is significantly brighter than the others, consistent with the Rap1p staining (our unpublished data). These findings indicate that yKu‐independent pathways for telomere anchoring function in the yku70 white colonies. It is noteworthy that some telomeric foci are perinuclear in the LM79‐pink cells as well (Figure 4, panel h), suggesting that yKu‐independent pathways for anchoring are not completely inoperative in these cells.
Sir3p is limiting for ectopic silencing in yku70 cells
The fact that ectopic silencing is only partially established in yku70 cells, as revealed by the white/pink phenotypes, may reflect an insufficient or nonfunctional release of silencing factors. Therefore, we investigated whether Sir3p and Sir4p are still limiting for silencing at LYS2 in yku70 cells. Overexpression of Sir3p and/or Sir4p in either white or pink yku70 cells provokes an increase in ectopic silencing (Figure 1B, lines i–l and o–r). The colony color does not change after transformation with the control plasmids lacking SIR gene inserts (Figure 1, lanes i and o), indicating that the transformation procedure per se does not modified the switch frequency of LM79. We note that overexpression of Sir3p appears to have more effect than Sir4p; this may reflect either differences in the quantities of the overexpressed proteins, negative trans‐dominant effects on silencer‐mediated silencing, or the fact that Sir4p is not limiting for ectopic silencing in yku70 cells. Interestingly, the reverse is observed for TPE: overexpression of Sir4p, but not Sir3p, rescues TPE in white or in pink yku70 cells (Figure 1B, lines l and r). This suggests that only Sir4p is limiting for TPE in yku70 cells, confirming previous observations (Mishra and Shore, 1999). In summary, TPE and ectopic silencing both appear to be stimulated by the increased dosage of Sir3p and/or Sir4p, but differ in their relative dependence upon these two Sir proteins.
We report an intriguing characteristic of yKu‐deficient cells which was uncovered in strains carrying an ADE2 reporter for ectopic silencing at LYS2. Roughly half of the primary transformants are able to achieve ectopic silencing, as revealed by a pink color, while the others were white and, thus, do not display ectopic silencing like the parental cells. These two phenotypes of otherwise identical yku70 cells are mitotically stable, although the switches from one to the other can be scored at a low frequency. Since the two phenotypes are observed in primary transformants, and since a cell can switch from one to the other, it is unlikely that additional mutations are responsible for the pink or white phenotype. Therefore, the pink and white phenotypes appear to correspond to two alternative states of genetically identical yku70 cells.
The two alternative states of yku70 cells do not reflect a property restricted to the silencing construct inserted at LYS2, but reflect bimodality in more general characteristics. First, we observed that pink cells exhibit a slight increase in residual TPE as compared with white cells. Secondly, the nuclear distribution of telomeric DNA and silencing factors is more diffuse in pink than in white cells. We propose that yKu− cells balance between epigenetic states characterized, at least in part, by the degree to which silencing factors are released from telomeres. Since it is hard to reconcile an increased release of silencing factors from telomeres with the slight improvement of the residual telomeric silencing observed in pink yku70 cells, we suggest further that the two states of yku70 cells are associated with other, unidentified differences in telomere organization.
That all the telomeres of one cell can establish and maintain a given functional state in a coordinated fashion over long growth periods suggests that long‐range mechanisms exist to homogenize the organization of telomeres. The switch between different nuclear states in yKu− cells is reminiscent of the profound nuclear changes that occur during cellular processes like senescence, differentiation or tumorigenesis in higher eukaryotes (for example see Brown et al., 1999; Bridger et al., 2000). There is no doubt that yeast studies will shed light on the basic principles that underlie these large‐scale remodelings of nuclear architecture and their epigenetic control.
Yeast strains and plasmids.
Genotypes of strains used in this study are listed in Table 2. The plasmid used to complement the yku70 disruption is pGBD‐YKU70, a derivative of pAS2 (Martin et al., 1999). pEADE2I contains an ADE2 gene flanked by silencers HML‐E and HML‐I with ADE2 promoter proximal to the I silencer (our unpublished construction). To integrate silencer‐flanked reporter at the LYS2 locus, a NotI fragment from pEADE2I was integrated into the XhoI of the LYS2 gene DNA (as in Maillet et al., 1996). An ade2 disruption was created in the S150‐2B strain with a linear fragment released from pade2::TRP1 (our unpublished construction).
Null alleles of YKU70 were created using a PCR product (Wach et al., 1994). In this strategy, the resistance kanMX4 selectable marker replaces the entire open reading frame. Null mutations of YKU80 were created with a yku80::HIS3 construct. For overexpression of the C‐terminal part of SIR4, we used pADH‐SIR4C (termed pCSIR4; Cockell et al., 1995). The plasmids used to overexpressed Sir3p and/or Sir4p are the following: p2μ‐ASir3, a derivative of pAAH5 (Maillet et al., 1996); pFP320, a derivative of pRS424 (Maillet et al., 1996).
Silencing and microscopic methods.
For the ADE color assay, cells were plated on rich medium without supplemented adenine (YPD‐ADE) at 30°C for 3 days, and then shifted to 4°C for at least two weeks to allow the red color to develop. Assays to evaluate expression of telomere‐proximal URA3 gene were performed by a spot assay (Gottschling et al., 1990). Immunofluorescence, FISH and confocal microscopy were performed as described (Gotta et al., 1996).
Data mentioned as not shown in the text are available as supplementary data at EMBO reports Online.
We thank J. Broach, D. Gottschling, R. Rothstein, E. Louis and D. Shore for providing various strains and plasmids. L.M. thanks the ‘Fondation pour la Recherche Médicale’ for his fellowship. This work was supported by La Ligue contre le Cancer (E.G.) and the Swiss National Science Foundation and Swiss Cancer League (S.M.G.).
- Copyright © 2001 European Molecular Biology Organization