Chromosome ends in Saccharomyces cerevisiae are positioned in clusters at the nuclear rim. We report that Ctf18, Ctf8, and Dcc1, the subunits of a Replication Factor C (RFC)‐like complex, are essential for the perinuclear positioning of telomeres. In both yeast and mammalian cells, peripheral nuclear positioning of chromatin during G1 phase correlates with late DNA replication. We find that the mislocalized telomeres of ctf18 cells still replicate late, showing that late DNA replication does not require peripheral positioning during G1. The Ku and Sir complexes have been shown to act through separate pathways to position telomeres, but in the absence of Ctf18 neither pathway can act fully to maintain telomere position. Surprisingly CTF18 is not required for Ku or Sir4‐mediated peripheral tethering of a nontelomeric chromosome locus. Our results suggest that the Ctf18 RFC‐like complex modifies telomeric chromatin to make it competent for normal localization to the nuclear periphery.
The physical organization of DNA within the nucleus is related to chromatin function. Chromosomes of higher eukaryotes occupy specific nuclear ‘territories’, and the spatial territory of a chromosome frequently reflects its gene‐density, with chromosomes containing a high proportion of nontranscribed sequence located close to the edge of the nucleus (Tanabe et al, 2002). Although it is clear that chromatin is organized and actively positioned within the nuclear space, the mechanisms determining physical organization of chromosomes within nuclei are not understood.
All eukaryotic cells replicate their DNA according to a reproducible temporal program, and spatial organization of the DNA is correlated with replication timing (reviewed in Taddei et al, 2004b). Replication foci are typically spread throughout the nuclear interior during early S phase, while peripheral and perinucleolar DNA replicates in mid to late S phase. Nonexpressed heterochromatic DNA usually replicates late in the S phase (Gilbert et al, 2004; Woodfine et al, 2004). In general, the spatial organization, transcriptional activity and replication timing of chromatin are correlated, but causative relationships between these three properties are unclear.
The organization of the telomeres of Saccharomyces cerevisiae offers a useful model system for studying chromosome positioning, transcriptional activity, and replication timing. The 32 telomeres of haploid yeast cells associate in 3–6 clusters at the nuclear periphery (Gotta et al, 1996). S. cerevisiae subtelomeric sequences are subject to silencing of polymerase II‐mediated transcription (Gottschling et al, 1990) and telomeres are replicated late during S phase (Ferguson and Fangman, 1992; Raghuraman et al, 2001).
Two partially redundant pathways have been identified that mediate tethering of telomeres to the nuclear rim. The first depends on the yeast Ku (yKu) protein complex (which consists of Yku70 and Yku80 proteins), and the second on the Sir4 and Esc1 proteins (Hediger et al, 2002; Taddei et al, 2004a). Different telomeres may differ somewhat in their dependence on these two pathways—for example, during G1, telomere VI‐right (VIR) positioning depends primarily on the Ku pathway and telomere VI‐left (VIL) primarily on the Sir4/Esc1 pathway (Bystricky et al, 2005). Some evidence suggests that the Ku‐dependent pathway tends to dominate during G1, while the Sir4/Esc1‐dependent tethering pathway is dominant in the S phase (Hediger et al, 2002). The telomeres lose their peripheral localization during G2 as cells prepare to enter mitosis, and perinuclear positioning is re‐established in early G1 phase. However, the telomere positioning mechanism is not fully understood, and in particular the molecular components that act with Ku to mediate telomere positioning are not known. Identification of additional positioning components is complicated by the fact that Ku also plays a key role in other telomere‐specific functions, including subtelomeric transcriptional silencing (Gravel et al, 1998; Laroche et al, 1998), telomerase recruitment (Stellwagen et al, 2003), and specification of late replication timing (Cosgrove et al, 2002).
Peripheral localization of DNA within the yeast nucleus has been shown to reinforce transcriptional silencing in a number of cases. For example, artificial localization to the periphery enhances transcriptional repression at a compromised silencer (Andrulis et al, 1998). Consistently, both Ku and SIR4 are required for maximum silencing of subtelomeric genes (Palladino et al, 1993; Gravel et al, 1998; Laroche et al, 1998). However, under some circumstances, positioning and silencing can be separated. Repression can be maintained at an intact silencer that is released from the nuclear peripheral zone (Gartenberg et al, 2004), and at a modified version of telomere VIIL there was no correlation between proportion of peripherally positioned telomeres and the efficiency of silencing (Tham et al, 2001).
The relationship between peripheral localization and replication timing has been less investigated. The telomeres and ribosomal DNA are localized close to the nuclear envelope and replicate in the second half of S phase (Raghuraman et al, 2001; A Cosgrove and A Donaldson, unpublished), suggesting that peripheral localization may favor late replication. Moreover, removal of Ku function leads simultaneously to delocalization and abnormally early replication of telomeres (Cosgrove et al, 2002). Localization of the DNA during G1 phase has been proposed to be particularly crucial for correct timing control (Gilbert, 2002), since the S. cerevisiae telomere late replication program is pre‐established during G1 (Raghuraman et al, 1997). The replication program in mammalian cells is also established during G1 coincident with re‐positioning of DNA within the nucleus (Dimitrova and Gilbert, 1999).
Replication factor C (RFC) is a five subunit ‘clamp‐loading’ complex consisting of the essential gene products Rfc1‐5, all of which belong to the AAA+ ATPase superfamily (Bowman et al, 2004). RFC loads the ring‐shaped PCNA polymerase clamp component of replication forks. Three RFC‐like complexes have been identified in which the largest subunit (Rfc1) is replaced by either Rad24, Ctf18, or Elg1; RAD24, CTF18, and ELG1 are nonessential genes with sequence similarity to RFC1 (reviewed in Kim and MacNeill, 2003). The Elg1 and Rad24 RFC‐like complexes (Elg1‐RLC and Rad24‐RLC) are important for genome stability and checkpoint responses, and Rad24‐RLC has been shown to load the ring‐shaped 9‐1‐1 complex onto damaged DNA. The Ctf18‐RLC is a heptameric complex containing two extra subunits, Ctf8 and Dcc1, in addition to Rfc2‐5 and Ctf18 itself. The function of Ctf18‐RLC remains mysterious. Disruption of either CTF18, CTF8, or DCC1 causes a sister chromatid cohesion defect, but no cohesin loading defect was detected in a ctf8 mutant (Hanna et al, 2001; Mayer et al, 2001; Kenna and Skibbens, 2003). By analogy to RFC and the Rad24‐RLC, the Ctf18‐RLC is believed to act on a ring‐shaped complex. Human Ctf18‐RLC can load PCNA in vitro, although with reduced efficiency when compared to RFC itself (Ohta et al, 2002; Bermudez et al, 2003; Merkle et al, 2003). It has recently been demonstrated that yeast Ctf18‐RLC efficiently unloads PCNA from DNA in vitro (Bylund and Burgers, 2005).
Here, we show that Ctf18‐RLC mediates correct positioning of yeast telomeres at the nuclear periphery. Despite the disruption of telomere peripheral positioning, the telomeres of ctf18 cells replicate late in the S phase, showing that peripheral positioning during G1 is not required for late replication of DNA. We propose that the Ctf18‐RLC may act, through unloading of PCNA and/or exchange of PCNA‐like ring‐shaped complexes, to establish a chromatin structure that is required for telomere positioning.
The Ctf18‐RLC complex is required for perinuclear positioning of Rap1
To elucidate molecular mechanisms responsible for positioning chromatin within the nucleus, we screened for new gene products involved in localizing S. cerevisiae telomeres. The screen will be described in detail elsewhere; briefly, it is based on examining the subnuclear localization of the telomeric heterochromatin component Rap1. We transformed a series of disruption mutants in nonessential genes with a plasmid encoding GFP‐Rap1, and screened for those mutants in which GFP‐Rap1 localization appeared abnormal. Expression of GFP‐Rap1 in wild‐type (WT) cells (Hayashi et al, 1998) reveals several discrete dots corresponding to the telomere clusters (Figure 1). In unbudded and small‐budded cells, these dots are predominantly localized at the nuclear rim, as expected since telomeres are localized to the nuclear periphery during the early part of the cell cycle. The Ku complex is required for correct localization of telomeres. As a control for the effect of telomere localization on Rap1 positioning, we confirmed that the Rap1 foci were largely dispersed in a yku70 strain (Figure 1). On examination of a ctf18 strain, we found that GFP‐Rap1 foci were almost completely disrupted, with the Rap1‐GFP signal dispersed throughout the nuclear interior (Figure 1). As described above, the Ctf18‐RLC is a seven subunit RFC‐like complex that includes the gene products Ctf8 and Dcc1. We found that Rap1 foci were also dispersed in ctf8 and dcc1 mutants (Figure 1). The fact that the ctf18, ctf8, and dcc1 mutants all share the same Rap1 localization defect suggests that the Ctf18‐RLC is essential for proper Rap1 localization to the nuclear periphery, rather than the effect being due to one of the gene products alone.
Rad24 and Elg1 are the largest subunits of the two other RLC complexes. Neither rad24 nor elg1 mutant showed a GFP‐Rap1 localization defect (Figure 1), suggesting that the role in Rap1 localization within the nucleus is specific to Ctf18‐RLC.
The Ctf18‐RLC is required for telomere positioning
One possible interpretation of the results in Figure 1 is that the Ctf18‐RLC is required for telomere positioning. To address this possibility, we tested the effects of deleting CTF18, CTF8, or DCC1 on telomeres that were fluorescently tagged (Straight et al, 1996). We used a strain in which a single telomere is marked by GFP fused to the lac repressor and the nuclear envelope is simultaneously visualized by GFP tagging of a nuclear pore component, so that the telomere is visible as a bright dot within a circle corresponding to the nuclear envelope (Figure 2A). In the majority of interphase WT cells, the telomere dot appears to touch the nuclear envelope (corresponding to a distance of <230 nm). Telomeres VIR, VIIIL, and XIVL were localized at the nuclear rim in a reduced proportion of ctf18, ctf8, and dcc1 cells, with levels of localization similar to those of a yku70 mutant (Figure 2B). The CTF18, CTF8, and DCC1 gene products are therefore required for correct positioning of S. cerevisiae telomeres at the nuclear periphery.
To test whether another peripherally localized sequence is disrupted in ctf18 cells, we examined the distribution of the ribosomal DNA that is normally packaged at the edge of the nucleus. Observation of the rDNA using a GFP‐tagged Net1 protein (which binds the ribosomal DNA repeats) revealed no apparent mislocalization of the rDNA (data not shown). Nuclear structure therefore does not appear to be grossly disrupted in the ctf18 mutant.
CTF18 is required for telomere localization in the G1 and S phase
To assess the cell cycle stages at which Ctf18 is important for telomere localization, we examined telomere position in cells scored for cell cycle position according to bud size. We quantified the position of telomere XIVL by dividing the nucleus into three concentric zones of equal area (Figure 3A) as described (Taddei et al, 2004a). In this assay, random telomere positioning would be represented by 33% of telomeres scored in each zone. In WT nuclei, telomere XIVL preferentially localizes to the outermost zone in both G1 and S phases (Figure 3B and Table I). In ctf18 G1 phase nuclei, the same telomere was almost randomly positioned (39% of telomeres in Zone 1: Figure 3B and Table I). Once ctf18 cells entered S phase, the telomere remained slightly delocalized from the periphery when compared to WT, although the effect was not as severe as in G1. χ2 analysis confirmed that telomere position during S phase in ctf18 is significantly different from that of WT and from random distribution. Very similar cell cycle effects were observed for telomere VIIIL (localization to Zone 1 in 67% of WT G1 cells, 32% of ctf18 G1 cells, 61% of WT S cells, and 49% of ctf18 S cells). P‐values assessing the statistical significance of these results are given in Table I. We conclude that telomere positioning is affected by CTF18 deletion primarily in G1 phase cells, with the absence of Ctf18 being slightly deleterious in the S phase. Overall, the telomere delocalization phenotype of ctf18 is reminiscent of that described for yku70, which has been reported to affect the positioning of some telomeres primarily during G1 (Hediger et al, 2002).
We performed time‐lapse analysis to examine whether delocalized telomeres in ctf18 nuclei can still visit the nuclear periphery. Tracings of movies showing the typical behavior of telomere XIVL in WT and ctf18 cells are shown in Figure 3C (Movies in Supplementary data). In WT G1 phase cells (Sup_3.mpg), telomere XIVL remained confined within 0.2 μm of the periphery for most of the analysis. Although the telomere did occasionally leave the nuclear envelope, it returned to the periphery after a short time and usually remained there. In the S phase, telomere movement became still more confined, suggesting even more stable localization (Sup_4.mpg). These data are consistent with previous observations in WT cells (Heun et al, 2001b). In ctf18 cells in contrast, telomere XIVL was not confined to nuclear periphery but spent more time in the interior, particularly during G1 phase (Sup_5.mpg). The telomere was not excluded from the edge of the nucleus and did pay occasional visits to the periphery, but failed to become stably localized during those visits. Once ctf18 cells entered S phase, brief periods of telomere positioning at the periphery were observed, although these were still not of the duration or stability seen in WT S phase cells (Sup_6.mpg).
We measured the duration of localization events (Figure 3D). In unbudded and small‐budded ctf18 mutant cells, periods of internal localization and brief visits to the periphery were increased at the expense of stable peripheral localization periods, to the extent that long‐term (greater than 1 min) localization periods were almost never observed. We conclude that Ctf18‐RLC is important for telomere positioning in both G1 and S phase, and that its primary role is to permit the establishment of stable localization at the nuclear periphery.
Telomeres replicate at the normal time in a ctf18 mutant strain
S. cerevisiae telomeres are normally late‐replicating and are localized at the nuclear periphery. Our discovery of a new effector of telomere localization enabled us to investigate whether peripheral localization of telomeres during G1 is a prerequisite for their late replication. We examined the replication program of a ctf18 strain using the dense isotope transfer technique (Donaldson et al, 1998). Figure 4A shows the replication programs of WT, ctf18 and yku70 mutant cells analyzed by using this method. Markers for early and late replication in the S phase are provided by the early replication origin ARS305 and a late‐replicating sequence on chromosome XIV that lies far from either telomere (chr XIV‐int). We examined the replication time of telomere VIIIL, whose perinuclear localization depends on Ctf18 as shown in Figure 2B. We found that telomere VIIIL replicated late in the ctf18 mutant as in WT cells (Figure 4A, left and centre panels). This result contrasts with the situation in the yku70 mutant in which telomere VIIIL replicated much earlier in the S phase (Figure 4A, right panel). To measure an ‘average’ telomere replication time, we examined the replication time of the Y′ sequences. Y′ is one of the repeated sequence elements found at more than half of yeast telomeres, so that examining Y′ replication time gives a good view of overall telomere replication time. Y′ sequences replicated late in the ctf18 mutant as in WT cells. In the yku70 mutant, Y′ sequences replicated much earlier in the S phase as shown previously (Cosgrove et al, 2002). A replication time can be assigned for a sequence as the time at which half the final level of replication has occurred, and the interval between the replication of the early and late marker sequences can be taken as a measure of S phase length. In the ctf18 strain, this interval was 16.4 min for the experiment shown in Figure 4A, compared with 23.8 min in the WT and 19.7 min in the yku70 mutant strain. The S phase program may therefore be slightly compressed in the ctf18 strain. ‘Replication index’ (RI) values can be calculated to compare the replication programs in different strains and adjust for differences in the speed at which cultures release from synchrony. RI values express the time of replication of each sequence as a proportion of elapsed S phase. Figure 4B shows the replication programs of WT, ctf18, and yku70 strains plotted according to RI. In this format, it is clear that there is no significant change in the relative replication time of Y′ sequences in the ctf18 strain when compared to WT. Analysis of two additional loci (sequences close to the left ends of chromosomes III and VI) also confirmed that the RI values in the ctf18 S phase were very similar to WT, any slight differences observed lying within experimental error.
From these experiments, we conclude that the telomeric DNA of a ctf18 mutant replicates at its normal, late time in the S phase despite the aberrant subnuclear localization of the chromosome ends during the G1 phase. Our observation of telomere delocalization combined with normal replication timing in the ctf18 strain shows that peripheral positioning of telomeres during G1 is not essential for their late replication.
Ku complex remains bound to a telomere in a ctf18 strain
The telomere positioning phenotype of ctf18 resembles that reported for yku70 and yku80 mutants (Hediger et al, 2002). Normal telomeric replication timing in the ctf18 mutant suggested that Ku binding to telomeres is intact, but we wished to test directly whether Ku is loaded onto telomeres in the ctf18 strain. We examined binding of Myc‐tagged Yku80 protein to two loci in the vicinity of telomere VIR. Yku80‐Myc was bound at the telomeric sequence, but not to the locus 5 kb away from telomere (Figure 5A), consistent with previous studies (Martin et al, 1999; Roy et al, 2004). We observed no significant change in this telomere‐specific binding in the ctf18 mutant, showing that Ctf18 is not required for Ku binding to chromosome VIR. Yku70 was required for Yku80 to bind the telomere, as expected since Ku binds DNA as a heterodimer. Consistent with the observation that Ku still binds telomeres in ctf18, the ctf18 mutant has only a slight defect in telomere length control (data not shown; Askree et al, 2004; Smolikov et al, 2004), while yku70 mutant displays a severe telomere length defect (Boulton and Jackson, 1996; Porter et al, 1996).
We tested whether Ctf18 itself is localized at telomeres. Immunofluorescence and in vivo labeling experiments gave no suggestion that Ctf18 is specifically located at telomere clusters (data not shown). The higher sensitivity technique of chromatin immunoprecipitation also provided no evidence for Ctf18 binding specifically to telomere VIR (Figure 5B). It therefore seems unlikely that Ctf18 is a structural component of a telomere peripheral localization pathway, and we believe that it is more likely to play a regulatory role.
CTF18 is required for both Ku and Sir4‐mediated telomere positioning pathways during G1
Two molecular pathways have been described that mediate localization of telomeres. Ku complex is believed to form a link between telomeres and the nuclear envelope by binding to an unidentified envelope‐bound component. A second pathway involves interaction of the telomere‐bound Sir4 protein with the nuclear envelope‐bound protein Esc1. To clarify whether Ctf18‐RLC affects telomere positioning through the Ku pathway, through the Sir4‐Esc1 pathway, or through a previously unidentified pathway, we studied the localization of telomere XIVL in a set of double mutants. This telomere was chosen for analysis because it requires both Ku‐ and Sir4‐dependent pathways for full positioning (Figure 6A); many other telomeres show more complete disruption of positioning on deletion of either of the known pathways (Hediger et al, 2002; Taddei et al, 2004a), which could obscure additional effects of further mutations.
Telomere XIVL positioning was random during G1 phase in the sir4 yku70 mutant (Figure 6A), showing that during G1 the Ku and Sir positioning mechanisms are the only pathways involved in localizing this telomere. Either the sir4 or yku70 mutations alone resulted in significant peripheral positioning (Figure 6A), demonstrating that each pathway can mediate some positioning independent of the other—that is, in the absence of Sir4, the Ku pathway can position telomere XIVL to some extent and vice versa. However, introducing the ctf18 mutation in the sir4 background (Figure 6A, ctf18 sir4) resulted in completely random telomere positioning, showing that without Ctf18 the Ku pathway can no longer mediate any telomere positioning. Introducing the ctf18 mutation into the yku70 background also resulted in random telomere positioning, showing that Ctf18 is also required for the residual telomere positioning by the Sir pathway in the yku70 mutant. These results suggest that, at least for telomere XIVL, both the Ku and Sir positioning pathways are largely dependent on Ctf18. This interpretation is consistent with the observation that disruption of CTF18 leads to a more severe positioning defect than either the sir4 and yku70 mutations alone (Figure 6A).
We also examined the effects of the various mutations on telomere positioning during S phase. In this case, the ctf18 mutant retained a significantly higher level of telomere XIVL positioning than the sir4 yku70 double mutant (Figure 6A). One interpretation of this result might be that during S phase, the Ku and Sir positioning pathways are less dependent on Ctf18 than they are in G1. However, during S phase, significant telomere positioning remained in the sir4 yku70 double mutant (Figure 6A), suggesting that an additional telomere positioning pathway plays a role at this cell cycle stage. It is intriguing to speculate that this additional pathway could be related to S phase events such as telomere replication or telomerase extension. Because the components of this additional pathway are unidentified, it is not possible to assess from our results whether Ctf18 is required for this novel positioning mechanism, or whether instead the dependence of the Ku and Sir4 pathways on Ctf18 is altered during S phase. Further clarification will require the identification of components of the S phase‐specific Ku/Sir‐independent positioning pathway.
CTF18 is not required for Ku‐mediated and Sir4‐mediated linkage of an internal locus to the nuclear periphery
The double‐mutant analysis showed that, at least during G1, the Ku and Sir‐mediated telomere positioning pathways are largely dependent on Ctf18. Taddei et al (2004a) developed a system that allows artificial tethering of an internally located locus to the nuclear periphery. We wished to examine whether localization of an internal site requires Ctf18, or whether instead it can occur in the absence of Ctf18. We used a strain in which the early replication origin locus ARS607 is flanked by lexA operators (to permit tethering by LexA‐Sir4PAD or by LexA‐yku80‐9) and by a series of lac operator sequences (to enable visualization using LacI‐GFP) (Figure 6B). The ARS607 locus is randomly located in a strain bearing this lacop‐lexAop‐ARS607 construct if no LexA fusion protein is expressed. As described previously, expression of LexA‐yku80‐9 leads to significant localization of ARS607 to the nuclear envelope (Figure 6C). The LexA‐yku80‐9 construct was still able to localize ARS607 when CTF18 was deleted. Similarly, the localization mediated by LexA‐Sir4PAD was not affected in the ctf18 mutant (Figure 6D). χ2 analysis confirmed that the ctf18 mutation has no significant effect (Table II). Function of the Ctf18‐RLC is therefore not required for Yku80 and Sir4 to bring about the peripheral localization of a chromosomal domain. We conclude that tethering of an ectopic locus to the nuclear periphery by either Yku80 or Sir4PAD bypasses the need for Ctf18.
We have found that the Ctf18‐RFC‐like Complex is critical for correct positioning of S. cerevisiae telomeres close to the nuclear periphery, particularly during G1 phase. Our findings represent the discovery of a new molecular effector of chromosome localization, and the identification of a new role for the Ctf18‐RLC in intranuclear organization. In mutants in any of the three of the subunits unique to the Ctf18‐RLC (Ctf18, Dcc1, and Ctf8), we observed two phenotypes that indicate disrupted telomere organization—the dislodgement of individual chromosome ends from the nuclear periphery and the dispersal of Rap1 from its normal localization pattern in foci within the nucleus. The Ctf18‐RLC is unique among the three known alternative RFC complexes in having this function in chromosome positioning.
The telomere localization defect in a ctf18 mutant provided the opportunity to test one model for replication timing control. Several studies had suggested a close relationship between late replication and peripheral positioning of the DNA during G1 phase (Dimitrova and Gilbert, 1999; Heun et al, 2001a). In particular, deletion of the Yku70 subunit of the Ku heterodimer dislodges telomeres from the nuclear periphery, and simultaneously causes aberrantly early activation of telomere‐proximal replication origins during S phase (Laroche et al, 1998; Cosgrove et al, 2002). Mutation of the Sir proteins causes a less dramatic but still noticeable disruption of telomere localization, and a slight but significant advancement in telomere replication timing (Stevenson and Gottschling, 1999). The ctf18 mutation clearly abolishes telomere localization to the nuclear periphery during G1, but we found that the telomeres of ctf18 cells replicate at their normal late time in the S phase, quite unlike the aberrant early telomere replication observed in a yku70 mutant. The ctf18 mutant phenotype therefore demonstrates that peripheral localization during G1 is not a prerequisite for the late replication of telomeres, and shows that the mechanisms of replication timing control must be distinct from those controlling G1 telomere intranuclear positioning.
By what mechanism does the Ctf18‐RLC affect telomere positioning? This question is difficult to address while the precise molecular role of the Ctf18‐RLC remains unclear. ctf18 mutants are slightly compromised in mating‐type and telomeric silencing (Suter et al, 2004), but the most prominent previously reported phenotype of ctf18 is premature separation of sister chromatids (Hanna et al, 2001; Mayer et al, 2001; Naiki et al, 2001). We found that cohesin mutants show only a slight defect in telomere XIVL positioning under conditions where the sister chromatid separation defect is clear (data not shown). Moreover, not all mutants that affect sister chromatid cohesion compromise telomere positioning. For example, CHL1 is required for cohesion (Petronczki et al, 2004; Skibbens, 2004), but the chl1 mutation did not compromise telomere peripheral positioning as assessed by Rap1 localization and analysis of a tagged telomere (data not shown). Taking these results together, we have found no convincing evidence that the effect of ctf18 on telomere positioning is a consequence of defective sister chromatid cohesion.
Two independently acting telomere positioning mechanisms have been characterized in budding yeast—the Ku‐dependent and Sir4/Esc1‐dependent pathways. Ctf18 is required for the full activity of both pathways (Figure 6A). Ctf18‐RLC is not required to load the Ku complex (Figure 5A), and chromatin fractionation experiments (not shown) gave no suggestion that the association of Ku with chromatin is altered in a ctf18 mutant strain. ctf18 and dcc1 strains retain significant telomeric silencing (Suter et al, 2004), implying that Sir and Rap1 loading onto telomeres (which is essential for telomeric silencing) is not severely compromised. Since it does not seem to be required to load Ku or Sir complexes, Ctf18‐RLC must presumably play a regulatory role to activate telomere positioning by the Ku‐ and Sir‐dependent pathways. Ku or Sir fragments tethered to an internal locus are capable of re‐positioning that chromosome domain in a ctf18 mutant, showing that isolated fragments can bypass the need for Ctf18‐RLC to activate peripheral positioning (Figure 6). These results are consistent with models in which Ctf18‐RLC ‘unmasks’ the inherent positioning capability of telomeric chromatin. The requirement for Ctf18 for localization of telomeres therefore probably reflects a particular characteristic of telomeric chromatin, such as the need for a regulatory modification to telomeric chromatin to establish its competence for linkage to the periphery. For example, a post‐translational modification of another telomeric protein (such as Rap1) might be required to allow linkage of telomeric heterochromatin to the nuclear periphery by Ku and Sir proteins.
The role of Ctf18 in regulating telomere positioning is doubly mysterious since the Ctf18‐RLC is proposed to act at replication forks during S phase (see below) whereas the ctf18 mutation is most deleterious to telomere positioning during G1 phase. It would be informative to test whether the presence of Ctf18‐RLC during DNA replication is required for telomere positioning in the subsequent G1 and S phase.
What is the relationship between the regulatory role of Ctf18‐RLC in telomere positioning and the molecular function of RFC‐like complexes? RFC itself loads the ring‐shaped sliding clamp PCNA onto replication forks, while Rad24‐RLC loads the ring‐shaped ‘9‐1‐1 complex’. By analogy, Ctf18‐RLC is believed to load or unload a ring complex. Yeast Ctf18‐RLC can unload PCNA from DNA very efficiently in vitro (Bylund and Burgers, 2005), and PCNA unloading after DNA synthesis is the clearest suggestion for the biochemical function of Ctf18‐RLC (Bylund and Burgers, 2005). However, it is not obvious why compromised PCNA unloading should lead to either telomere depositioning or defective sister chromatid cohesion. It is noteworthy that the Ctf8 and Dcc1 subunits of Ctf18‐RLC are required for telomere positioning (Figure 2) and establishment of cohesion (Hanna et al, 2001; Mayer et al, 2001), but dispensable for in vitro PCNA unloading. Perhaps, Ctf18‐RLC has two activities, and the apparently unrelated in vitro and in vivo observations of its properties reflect different aspects of its function. We suggest that PCNA unloading by Ctf18‐RLC might be coupled in vivo to another chromatin modification that is required to activate the Ku and Sir telomere positioning pathways in the subsequent G1 phase. Unlike the PCNA‐unloading step, the second, coupled step would be expected to require Ctf8 and Dcc1 since these subunits are required for telomere positioning. One possibility is that the second activity of Ctf18‐RLC involves loading of another ring‐shaped complex.
Ctf18‐RLC could conceivably play a related role at other chromosomal loci, for example, to establish cohesin loading sites as competent for sister chromatid attachment. If so, Ctf18‐RLC might be envisaged as having a general involvement in activating particular properties of specialized chromatin sites following DNA replication. It will be of interest to investigate whether the complex is involved in regulating chromosome organization within mammalian nuclei.
Materials and methods
Gene deletion collections were purchased from EUROSCARF. Other strains are described in Supplementary data.
Plasmid pAT4‐yku80‐9 (encoding Yku80‐9 fused to LexA) and pAT4‐Sir4PAD (encoding Sir4PAD fused to LexA) were as described (Taddei et al, 2004a). Additional plasmids are described in Supplementary data.
Cell cycle classification was as follows: unbudded cells=G1 phase; bud size less than 2 μm=S phase; bud larger than 2 μm with round nucleus not at the bud neck=G2; bud larger than 2 μm with elongated nucleus at the bud neck=M phase. Microscopic techniques are described in Supplementary data.
Chromatin immunoprecipitation of Myc‐tagged Yku80 and Ctf18 proteins was performed as described (Strahl‐Bolsinger et al, 1997; Tanaka et al, 1997), using monoclonal anti‐Myc antibody (9E11) (Abcam). Units of DNA in each PCR reaction were calculated relative to amplification of a dilution series of whole‐genomic standard DNA by the same primer pair. Details of primer pairs used are described in Supplementary data.
Analysis of replication timing program
Dense isotope transfer experiments were carried out as described previously (Donaldson et al, 1998), using α‐factor synchronization and release at 30°C in light medium. Probes are described in Supplementary data.
Supplementary data are available at The EMBO Journal Online.
Supplementary Materials and Methods
Supplementary Movies 1
Supplementary Movies 2
Supplementary Movies 3
Supplementary Movies 4
Thanks to Luis Aragon, Aki Hayashi‐Hagihara, Yasushi Hiraoka, Susan Gasser and Virginia Zakian for materials, and to Tomoyuki Tanaka and Kozo Tanaka for technical advice. Berndt Müller and Conrad Nieduszynski provided comments on the manuscript.
- Copyright © 2006 European Molecular Biology Organization