Karim Mekhail, Department of Laboratory Medicine and Pathobiology, Faculty of Medicine, University of Toronto, 1 King's College Circle, Toronto, Ontario, Canada M5S 1A8. Tel. 416-946-8132; fax 416-978-5959; e-mail: firstname.lastname@example.org
Telomere dysfunction is linked with genome instability and premature aging. Roles for sirtuin proteins at telomeres are thought to promote lifespan in yeast and mammals. However, replicative lifespan of the budding yeast Saccharomyces cerevisiae shortens upon deletion of Rif1, a protein that limits the recruitment of the sirtuin histone deacetylase Sir2 to telomeres. Here we show that Rif1 maintains replicative lifespan by ultimately stabilizing another age-related chromosomal domain harboring the ribosomal DNA (rDNA) repeats. Deletion of Rif1 increases Sir2 localization to telomeres and the silent mating-type loci, while releasing a pool of the histone deacetylase from the intergenic spacer 1 (IGS1) of rDNA. This is accompanied by a disruption of IGS1 silent chromatin assembly and increases in aberrant recombination within rDNA repeats. Lifespan defects linked with Rif1 deletion are abolished if rDNA repeats are forcibly stabilized via deletion of the replication fork-blocking protein Fob1. In addition, Sir2 overexpression prevents Rif1 deletion from disrupting Sir2 at IGS1 and shortening lifespan. Moreover, subjecting cells lacking Rif1 to caloric restriction increases IGS1 histone deacetylation and lifespan, while uncovering novel genetic interactions between RIF1 and SIR2. Our data indicate that Rif1 maintains lifespan-sustaining levels of Sir2 at rDNA by preventing excessive recruitment of the histone deacetylase to telomeric and silent mating-type loci. As sirtuin histone deacetylases, such as Sir2 or mammalian SIRT6, each operate at multiple age-related loci, we propose that factors limiting the localization of sirtuins to certain age-related loci can promote lifespan-sustaining roles of these sirtuins elsewhere in the genome.
Gene expression patterns and cellular fate are thought to be intimately connected to the establishment and maintenance of silent chromatin or heterochromatin domains (Kurdistani & Grunstein, 2003). Silencing within these regions is mediated by several enzymes including histone deacetylases (Kurdistani & Grunstein, 2003). However, the level of a given histone deacetylase, which may act simultaneously at multiple silent chromatin loci, is often limiting (Moazed, 2001). In addition, certain factors can directly limit the recruitment of histone deacetylases to specific silent chromatin domains (Moazed, 2001). How such factors affect the distribution of histone deacetylases between all the major silent chromatin domains at which the enzyme typically operates and how this affects the cell overall is unclear (Buck & Shore, 1995; Smith et al., 1998; Michel et al., 2005). The budding yeast Saccharomyces cerevisiae has three major silent chromatin domains, namely the rDNA repeats, telomeres, and mating-type loci (Moazed, 2001). All three domains, which are nonrandomly arranged within the nucleus, are controlled by the conserved histone deacetylase Sir2 (Rine & Herskowitz, 1987; Aparicio et al., 1991; Imai et al., 2000; Moazed, 2001; Mekhail & Moazed, 2010). Sir2 is required for the maintenance of replicative lifespan, a measure of cellular longevity that reflects the number of times that a yeast mother cell replicates its DNA and yields progeny before reaching senescence (Kaeberlein, 2010).
Replicative lifespan is highly dependent on the stability of the ribosomal DNA (rDNA) repeats, which consist of ~190 rRNA genes that are tandemly arranged on chromosome XII and reside within the nucleolus, a major nuclear compartment that is also the site of ribosomal biogenesis (Mekhail & Moazed, 2010). RNA polymerase (Pol) I and III transcribe rRNA genes, while transcription from promoters within intergenic spacers (IGS1 and IGS2) flanking rRNA genes is mediated by RNA Pol II (Mekhail & Moazed, 2010). Expansion or contraction of the rDNA repeats via recombination can be beneficial under stress (Kobayashi et al., 1998). However, hyperactive or aberrant recombination can lead to chromosomal instability and a shorter replicative lifespan (Gottlieb & Esposito, 1989; Kaeberlein et al., 1999). Ribosomal DNA instability can also lead to the formation of extrachromosomal rDNA circles, which can promote but are not required for lifespan shortening (Sinclair & Guarente, 1997; Ganley et al., 2009). Thus, recombination within the rDNA repeats is tightly regulated. Recombination within the repeats is promoted by an open chromatin structure and is dependent on binding of the fork block 1 (Fob1) protein to IGS1 sequences (Kobayashi & Horiuchi, 1996; Bryk et al., 1997; Fritze et al., 1997; Smith & Boeke, 1997; Huang & Moazed, 2003). Sir2 operates at rDNA as part of a complex called RENT, which also contains Net1/Cfi1 and Cdc14 (Shou et al., 1999; Straight et al., 1999; Visintin et al., 1999). Through a process termed rDNA silencing, compaction of IGS1 chromatin by Sir2 limits access to RNA Pol II and to the recombination machinery preserving chromosome stability (Gottlieb & Esposito, 1989; Bryk et al., 1997; Smith & Boeke, 1997). Other nucleolar factors implicated in rDNA silencing include the Cohibin complex, which is composed of Lrs4 and Csm1 proteins and operates at least partly independently of Sir2 (Huang et al., 2006; Mekhail et al., 2008; Chan et al., 2011; Poon & Mekhail, 2011). Thus, disruption of nucleolar Sir2/Cohibin or Fob1, respectively, decreases and increases replicative lifespan (Defossez et al., 1999; Kaeberlein et al., 1999; Chan et al., 2011).
The function of telomeric Sir2 is also required for the maintenance of replicative lifespan (Gottschling et al., 1990; Dang et al., 2009; Chan et al., 2011). Telomeric Sir2 is the enzymatic subunit of the SIR complex (silent information regulator), which also contains Sir3 and Sir4 (Moazed, 2001). Through a process termed telomere position effect or telomeric silencing, SIR is recruited to telomeres and spreads into subtelomeric regions while promoting silent chromatin assembly and silencing RNA Pol II-transcribed subtelomeric genes (Gottschling et al., 1990; Moazed, 2001). In addition, perinuclear telomere clustering, which relies in part on telomeric Cohibin complexes, increases the local concentration of SIR proteins ensuring efficient telomeric silencing (Mekhail & Moazed, 2010; Chan et al., 2011; Poon & Mekhail, 2012). Disruption of this Sir2-dependent telomeric silencing is linked with subtelomeric DNA instability and a shorter lifespan (Kaeberlein et al., 1999; Dang et al., 2009; Chan et al., 2011). Sir2 also localizes to and represses gene expression at the silent mating-type loci, HMR and HML, which are commonly referred to as HM loci (Rine & Herskowitz, 1987). How this impacts lifespan is unclear (Kaeberlein et al., 1999). Thus, Sir2 promotes lifespan at least via rDNA- and telomere-related processes.
Regulation of telomeric silencing is also partly mediated by the protein Rap1 (repressor activator protein 1), which binds to sites within the terminal C1–3A telomeric repeats (Longtine et al., 1989). Rap1 can recruit Sir4, which then recruits the rest of the SIR complex promoting its spreading into subtelomeric regions (Moretti et al., 1994; Buck & Shore, 1995; Hoppe et al., 2002). In addition, different domains of Rap1 can recruit the telomeric Rif1 and Rif2 proteins forming a complex that is functionally analogous to shelterin with regard to its ability to limit access to telomerase and prevent excessive telomere lengthening (Hardy et al., 1992; Wotton & Shore, 1997; Martinez & Blasco, 2010; Gallardo et al., 2011). However, Sir4 and Rif1, but not Rif2, compete to bind the same domain of Rap1 and thus Rif1 can directly limit the Sir4-dependent recruitment of Sir2 to telomeres (Buck & Shore, 1995; Wotton & Shore, 1997). While the physiological relevance of this antagonistic role of Rif1 with respect to telomeric silencing is unclear, deletion of Rif1, but not Rif2, has been linked with a shortened lifespan suggesting that hyper-recruitment of Sir2 to chromosome ends underlies the shortened lifespan of rif1∆ cells (Austriaco & Guarente, 1997; Managbanag et al., 2008). However, as the recruitment of Sir2 to telomeres is positively linked with lifespan, we tested whether the hyper-recruitment of Sir2 to chromosome ends in rif1∆ cells occurs at the expense of other Sir2-regulated chromosomal domains that are also critical to lifespan maintenance.
By examining how Rif1 affects the localization and function of Sir2 at major silent chromatin domains, our work reveals that cells utilize factors such as Rif1 to enforce lifespan-sustaining distributions of key chromatin-modulatory enzymes between their principal sites of action. We also identify various genetic and environmental approaches that can prevent or counteract chromosomal and cellular defects typically observed following the mislocalization of Sir2 across the genome in cells lacking Rif1. Overall, our work reveals lifespan-sustaining relationships between histone deacetylases, their regulatory factors, nonrandom spatial genome organization, and the cellular environment.
Rif1 affects Sir2 localization and histone acetylation at all major silent chromatin domains
To gain insight into the role of Rif1 in lifespan maintenance, we first examined the genetic relationship between RIF1 and SIR2. The loss of Rif1 significantly shortened replicative lifespan, albeit not to the same extent as deleting Sir2, as expected (Fig. 1A and Table S1) (Austriaco & Guarente, 1997; Kaeberlein et al., 1999). More importantly, rif1∆ sir2∆ cells exhibited a lifespan similar to that of sir2∆ cells (Fig. 1A and Table S1). Controls for lifespan analyses were lrs4∆ and hxk2∆ cells, which exhibited a respectively shortened and extended lifespan relative to wild-type cells, as expected (Fig. 1A; Table S1) (Lin et al., 2002; Chan et al., 2011). These findings suggest that Rif1 may operate via Sir2 to maintain replicative lifespan.
The levels of chromatin silencing proteins such as Sir2 can be limiting (Maillet et al., 1996; Smith et al., 1998). Sir2 is implicated in silent chromatin assembly at all of the three major silent chromatin domains of S. cerevisiae, namely the rDNA repeats, subtelomeres, and silent mating-type loci (Rine & Herskowitz, 1987; Gottschling et al., 1990; Aparicio et al., 1991; Bryk et al., 1997; Fritze et al., 1997; Smith & Boeke, 1997). Thus, we examined if the loss of Rif1 impacts the localization and function of the Sir2 histone deacetylase at all three major silent chromatin domains. We conducted chromatin cross-linking and immunoprecipitation (ChIP) analyses employing antibodies specific to Sir2 or, as a Sir2 functional readout, to histone H3 acetylated on lysines 9 and 14 (diAcH3K9/K14) (Braunstein et al., 1993; Imai et al., 2000; Suka et al., 2001; Armstrong et al., 2002; Huang & Moazed, 2003; Mekhail et al., 2008; Chan et al., 2011). Loss of Rif1 increased the level of Sir2 proteins associated with subtelomeres while decreasing histone H3 acetylation near chromosome ends (Fig. 1B,E). Surprisingly, we also observed a weak increase in Sir2 localization and histone deacetylation at the silent mating-type loci, HMR and HML, of rif1∆ cells (Fig. 1C,E). In contrast, histone acetylation and Sir2 localization were not significantly altered at IGS2 of rDNA (Fig. 1D,E). However, increased Sir2 localization to subtelomeres and HM loci was accompanied by the release of a pool of Sir2 from the rDNA IGS1 region, which exhibited increased histone acetylation (Fig. 1D,E). Differences between the effect of deleting Rif1 on the intergenic spacers of rDNA repeats may be linked with the fact that more Sir2 is normally localized to IGS1 as compared to IGS2 in wild-type cells (Huang & Moazed, 2003). Importantly, pull down of TAP-tagged Rif1 revealed enrichments for DNA sequences at telomeric and mating-type loci but not rDNA repeats, whereas Sir2-TAP localized to telomeres, mating-type loci, as well as the rDNA repeats (Fig. 1H–J). These data suggest that Rif1 proteins at telomeres and mating-type loci may ultimately modulate the distribution of Sir2 between the three major silent chromatin domains.
While an excessive silencing at telomeres is not thought to alter replicative lifespan, loss of silent chromatin assembly at the rDNA repeats leads to an increase in aberrant recombination events such as unequal sister chromatid exchange (USCE) and shortens lifespan (Gottlieb & Esposito, 1989; Kaeberlein et al., 1999; Chan et al., 2011). Thus, we assessed USCE by measuring the rate of loss of an ADE2 marker gene inserted within the rDNA repeats (Kaeberlein et al., 1999; Mekhail et al., 2008; Chan et al., 2011). Interestingly, Rif1 deletion significantly increased the rate of USCE, although to a lower extent than the deletion of Sir2 (Fig. 1F,G; Table S2). In addition, USCE rates were not statistically different in sir2∆ and sir2∆ rif1∆ cells (Fig. 1G and Table S2). In contrast, USCE rates can be further increased in sir2∆ cells upon deletion of other factors such as Lrs4 and Heh1, which are thought to suppress rDNA recombination via processes that are partly independent of Sir2 (Mekhail et al., 2008). Furthermore, consistent with our ChIP and recombination data, Rif1 ensures efficient silencing of a foreign reporter gene inserted within the rDNA repeats but not silent mating-type or subtelomeric regions (Hardy et al., 1992; Smith et al., 1999; Chan et al., 2011; Park et al., 2011). Taken together, our findings suggest that Rif1 suppresses Sir2 recruitment primarily to telomeres and, to a lesser extent, the mating-type loci to maintain Sir2 levels at the rDNA recombination enhancer sequence-containing IGS1 and thereby ensure efficient histone deacetylation and DNA stability within the rDNA repeats. However, while the loss of rDNA repeat stability may well underlie the lifespan defect observed in rif1∆ cells, our data so far do not fully support this notion. In addition, we still cannot completely rule out the possibility that excessive silent chromatin assembly at the telomeres and HM loci of rif1∆ cells may at least partly contribute to lifespan shortening.
Rif1 is dispensable for lifespan maintenance in a setting where rDNA repeats are forcibly stabilized via Fob1 deletion
We next set out to uncouple rDNA repeat stability from Rif1 deletion. Binding of the nucleolar protein Fob1 to recombination enhancer elements within IGS1 regions of rDNA repeats is required for recombination within the repeats (Kobayashi & Horiuchi, 1996). However, Fob1 is actually required for the recruitment of chromatin silencing factors including Sir2 and Lrs4 to rDNA repeats (Huang & Moazed, 2003). Thus, although Fob1 deletion abrogates rDNA silencing, it still leads to increased rDNA stability and replicative lifespan (Defossez et al., 1999; Huang et al., 2006). Therefore, we investigated the effects of deleting Rif1 in fob1∆ cells. ChIP analyses revealed that the loss of Rif1 in a fob1∆ setting still increased Sir2 localization and histone deacetylation at both the subtelomeric and HM loci (Fig. 2A,B, and D). Interestingly, such rif1∆-dependent increases in Sir2 localization and histone deacetylation were higher in the absence of Fob1 than in its presence (compare Figs. 2D to 1E). This is consistent with the release of a significant fraction of the Sir2 protein population from rDNA repeats in fob1∆ cells (Huang & Moazed, 2003). This also suggests that Rif1 deletion in wild-type cells only leads to respectively moderate and weak increases in Sir2 localization to telomeres and HM loci because of a limiting population of Sir2 proteins, a major fraction of which is normally located at rDNA repeats. In addition, the levels of acetylated histones associated with the rDNA repeats, either at IGS1 or IGS2, were similar in fob1∆ and fob1∆ rif1∆ cells (Fig. 2C,D). Together, these data indicate that Rif1 deletion in fob1∆ cells is ineffectual on chromatin status at rDNA repeats but still leads to excessive silent chromatin assembly at telomeres and HM loci.
Thus, if Rif1 deletion shortens the lifespan of wild-type cells by primarily promoting excessive silencing at telomeres and HM loci, then Rif1 deletion would be expected to significantly shorten the lifespan of fob1∆ cells. However, we found that fob1∆ and fob1∆ rif1∆ cells exhibited a similar extended, replicative lifespan (Fig. 2E and Table S1). In addition, Rif1 deletion in fob1∆ cells failed to increase the rate of USCE within rDNA repeats (Fig. 2F and Table S2). In contrast, deletion of Lrs4, which typically maintains rDNA repeat stability but also ensures telomeric silencing, increased the rate of rDNA USCE and decreased the lifespan of Fob1-deficient cells (Fig. 2E,F; Tables S1 and S2) (Chan et al., 2011). ChIP and lifespan experiments conducted in different strain backgrounds yielded similar results (Fig. 2; Figure S1). Taken together, our results suggest that increased rDNA recombination, as a consequence of decreased Sir2-IGS1 interactions, underlies the rif1∆-associated lifespan defect, which can be prevented if the rDNA repeats are forcibly stabilized via Fob1 deletion.
Increasing cellular Sir2 levels prevents rif1∆-associated defects in rDNA silencing/stability and replicative lifespan
We next asked how the effects of deleting Rif1 may be altered in cells already expressing elevated levels of Sir2. We used a strain in which a second copy of SIR2 is inserted at the LEU2 locus (designated SIR2 over-expression or SIR2-OE) (Kaeberlein et al., 2004). Reverse transcription PCR (RT-PCR) and western blot analysis confirmed that SIR2-OE cells exhibited increased Sir2 mRNA and protein levels (Fig. 3A,B). Importantly, Rif1 deletion was ineffectual on Sir2 mRNA and protein levels both within the wild-type as well as the SIR2 overexpression settings (Fig. 3A,B).
ChIP analysis revealed that SIR2 overexpression does not alter Sir2 levels or histone acetylation at IGS1 within an otherwise wild-type setting (Fig. 3C). However, relative to rif1∆ cells, SIR2-OE rif1∆ cells exhibited both higher IGS1 Sir2 localization and histone deacetylation levels that were similar to those observed in wild-type cells (Fig. 3C). Thus, within our experimental conditions, IGS1 Sir2-binding sites are likely at or near saturation in the wild-type setting but Rif1 deletion causes the release of a pool of Sir2 from these binding sites only when the overall levels of Sir2 are limiting in the cell. In contrast, SIR2-OE increases telomeric Sir2 localization and histone deacetylation in the wild-type setting but not in the rif1∆ setting, in which Sir2-telomere interaction levels are already higher than those observed in wild-type cells (Fig. 3C). The mating-type loci exhibited changes similar to those seen at telomeres while IGS2 was unaltered by Sir2 overexpression or Rif1 deletion (Fig. 3C and not shown; also see Figs. 1C–E and 2A–D). More importantly, our ChIP experiments suggest that increasing the levels of Sir2 at IGS1 regions, which contain the rDNA recombination enhancer sequences, could eliminate the rDNA repeat instability and shorter lifespan typically linked with the loss of Rif1. Indeed, we did find that SIR2 overexpression prevented Rif1 deletion from reducing rDNA repeat stability and excessively shortening cellular lifespan (Fig. 3D,E; Tables S2 and S1). SIR2 overexpression within the wild-type setting did not alter rDNA repeat stability but increased replicative lifespan likely due to increased telomeric silencing, as expected (Fig. 3C–E). We note that we did observe a statistically insignificant yet qualitative difference between the lifespan of SIR2-OE cells and SIR2-OE rif1∆ cells and so this could reflect effects on Sir2 or other processes in the cell within these experimental settings (Fig. 3E). All in all, our data so far suggest that the release of a pool of Sir2 from rDNA IGS1 regions underlies the rDNA repeat instability and, at least in part, the reduced replicative lifespan linked with the loss of Rif1.
Relationships between Rif1, Sir2, and calorie restriction in rDNA and lifespan maintenance
Calorie restriction (CR) is an environmental intervention that restricts the concentration of glucose in the cellular milieu (Kaeberlein, 2010). CR is thought to extend replicative lifespan via processes that are both dependent and independent of Sir2 (Lin et al., 2000; Kaeberlein et al., 2005). In an effort to reveal putative relationships between Rif1, Sir2, and dietary restriction, we set out to test the effect of CR on cells lacking Rif1 and/or Sir2.
Restricting glucose concentrations decreased IGS1 histone acetylation, increased rDNA stability, and extended the lifespan of wild-type cells (Fig. 4A–C; Tables S1 and S2). Interestingly, CR reduced histone acetylation at the rDNA IGS1 of rif1∆ cells, which now exhibited a level of rDNA stability and replicative lifespan similar to that of wild-type cells grown under normal glucose conditions (Fig. 4A–C; Tables S1 and S2). We still observed a qualitatively very small, albeit statistically insignificant, difference between the lifespans of calorie restricted rif1∆ cells and nonrestricted wild-type cells, which could indicate that CR may be rescuing the lifespan of rif1∆ cells via processes that do not, at least in part, typically implicate Rif1 and therefore Sir2 as well (Fig. 4C). Consistent with these findings and with previous reports, we found that CR partly restores rDNA IGS1 histone deacetylation and rDNA repeat stability in sir2∆ cells (Fig. 4D–E) (Riesen & Morgan, 2009; Smith et al., 2009). We also found that CR fails to alter the lifespan of sir2∆ cells, possibly reflecting the fact that rDNA IGS1 histone acetylation levels as well as rDNA instability in such CR-treated sir2∆ cells are still significantly higher than the levels observed in wild-type cells cultured under normal glucose concentrations (Fig. 4D–F; also compare Fig. 4B–E; Table S2). This may also reflect the overall ability of sir2∆ to significantly limit or even abolish the lifespan-extending effects of various environmental and genetic interventions (Lin et al., 2000; Delaney et al., 2011). However, the fact that CR partly reduces the levels of acetylated histones at rDNA IGS1 in sir2∆ cells, also suggests that CR is able to activate or direct one or several other histone deacetylases to the rDNA repeats, at least in the absence of Sir2. Alternatively, CR may be reducing the function of histone acetyl transferases at the rDNA repeats of sir2∆ cells. Interestingly, CR reduces rDNA IGS1 histone acetylation as well as rDNA repeat instability in rif1∆ cells and sir2∆ cells, but not rif1∆ sir2∆ cells (Fig. 4D–E; Table S2). This suggests that the loss of Rif1 may somehow inhibit or counteract, at least in part, the CR-triggered processes normally responsible for the reduction in acetylated histone levels at rDNA IGS1 in sir2∆ cells. CR also failed to alter the lifespan of rif1∆ sir2∆ cells (Fig. 4F). This could be due to the fact that CR fails to alter histone acetylation levels or rDNA repeat stability in these cells (Fig. 4D–E; Table S2). Alternatively, this could again be reflective of the inability of cells lacking Sir2 to respond to a myriad of typically lifespan-extending interventions. Overall Sir2 protein levels in either wild-type or rif1∆ cells were similar irrespective of glucose concentration (Fig. 4G). We note that we did notice a second lower-migrating and faint band in anti-Sir2 immunoblotting of calorie restricted cells (Fig. 4G). While this may reflect changes to the post-translational modification status of a pool of Sir2 proteins, Sir2-IGS1 interactions were unaltered under CR (Fig. 4A).
These data indicate that calorie restricting rif1∆ cells renders their chromatin assembly at rDNA IGS1 as well as their cellular lifespan similar to the phenotypes observed for wild-type cells grown under normal glucose conditions. In addition, our CR experiments reveal possible intersections between Sir2, Rif1, and CR in the maintenance of rDNA repeats.
Chromosomal domains are nonrandomly arranged relative to each other and to various nuclear landmarks. In S. cerevisiae, this is the case for the three major silent chromatin domains – the nucleolar rDNA repeats, telomeres arranged in 4–8 perinuclear clusters, and HM loci, which tend to be restricted to somewhat peripheral regions of the nucleus (Mekhail & Moazed, 2010). Our results reveal that Rif1 exploits this nonrandom organization to maintain replicative lifespan by enforcing an optimal rDNA-stabilizing and lifespan-sustaining distribution of Sir2 between its major sites of action (Fig. 5A,B). Forced stabilization of rDNA repeats via Fob1 deletion renders the loss of Rif1 ineffectual on lifespan even in the presence of increased Sir2 localization to telomeres and mating-type loci (Fig. 5C). This indicates that the shortened lifespan of rif1∆ cells is linked with the release of Sir2 from the recombination modulating IGS1 and not the increased localization of this histone deacetylase to telomeres or mating-type loci. Consistent with these findings, over-expressing Sir2 prevents Rif1 deletion from increasing rDNA IGS1 histone acetylation, compromising rDNA repeat stability, or shortening replicative lifespan relative to wild-type cells (Fig. 5D). Our findings also suggest that the increased lifespan of SIR2-OE cells relative to wild-type cells is more likely to be linked with increased Sir2 localization and histone deacetylation at nonrDNA sites such as telomeres. However, the increased lifespan of SIR2-OE rif1∆ cells relative to rif1∆ cells is more likely to be linked with the restored localization and function of Sir2 at rDNA IGS1. Moreover, exposing rif1∆ cells to environmental CR decreases rDNA IGS1 histone acetylation, possibly through processes that are dependent and independent of Sir2, and restores rDNA repeat stability while extending replicative lifespan (Fig. 5E). Interestingly, unlike rif1∆ cells or sir2∆ cells, the inability of rif1∆ sir2∆ cells to decrease rDNA IGS1 histone acetylation under caloric restriction suggests that Rif1 deletion inhibits or counteracts CR-triggered processes that can lower the levels of rDNA IGS1 histone acetylation in the absence Sir2. Such processes could modulate Sir2 homologues such as Hst2, other histone modifying enzymes, or possibly trigger histone modulatory mechanisms linked with the target of rapamycin 1 (Tor1) (Kaeberlein et al., 2005; Lamming et al., 2005). It is important to note that previous studies have revealed that CR relying on different glucose concentrations can yield at least partly different results and so our cells may be similarly affected (Kaeberlein, 2010). In addition, changes to histone acetylation may vary at different locations within the rDNA intergenic spacers. Overall, our findings indicate that Rif1 limits the local recruitment of Sir2 to telomeres and the silent mating-type loci. This reinforces the association of this histone deacetylase with rDNA repeats, thereby maintaining chromosome stability and replicative lifespan. Furthermore, rif1∆-related defects in rDNA repeat stability and replicative lifespan can be prevented or counteracted via fob1∆-dependent rDNA repeat stabilization, Sir2 overexpression, as well as Sir2-dependent/independent processes triggered by caloric restriction.
We previously reported that the V-shaped complex Cohibin, which is a structural analogue of the conserved Ndc80 subcomplex Spc24-Spc25, clusters telomeres and links them with the nuclear envelope via interactions with conserved SUN and LEM domain-containing inner nuclear membrane proteins (Chan et al., 2011). This Cohibin-dependent clustering of telomeres increases the local concentration of Sir2 ensuring efficient silent chromatin assembly and DNA stability within subtelomeric regions. This telomeric function of Cohibin is required for the maintenance of replicative lifespan (Chan et al., 2011). Thus, together with the findings that we report here, our work reveals that interplay between the functions of Cohibin and Rif1 at telomeres is critical to lifespan-sustaining processes at both telomeres and rDNA repeats. First, Cohibin ensures telomere clustering to promote silencing at telomeres. However, telomeric Rif1 prevents the excessive recruitment of Sir2 to maintain lifespan-sustaining threshold levels of the histone deacetylase at rDNA repeats. We also note that our data indicate that Rif1 limits the recruitment of Sir2 to telomeres and, to a lesser extent, the silent mating-type loci. As Rap1 is implicated in the recruitment of SIR proteins to telomeres as well as the HM loci (Buck & Shore, 1995), our results point to a model in which Rif1 and SIR proteins compete for Rap1 binding at telomeres and to a lesser extent at HM loci. This model is bolstered by our findings that Rif1 localizes to telomeres and mating-type loci but not rDNA. As to why Rif1 deletion seems to favor Sir2 localization to telomeres more than it does to the HM loci remains unclear. The lengthening of telomeres upon Rif1 deletion may contribute, although it is likely insufficient, to promote the shift of a greater fraction of the Sir2 protein pool released from rDNA to telomeres (Austriaco & Guarente, 1997). Another possibility may also be that telomere clustering significantly increases the local concentration of SIR proteins, making it easier for Rap1/Sir4-dependent Sir2 recruitment upon Rif1 deletion.
Our study highlights the importance of modulating the distribution of sirtuin histone deacetylases across the genome to lifespan regulation and suggests that similar processes may exist in other organisms. Humans have seven sirtuin genes, SIRT1-SIRT7, and disruption of the mammalian histone deacetylases SIRT1 or SIRT6 is linked with premature aging phenotypes [Reviewed in (Haigis & Guarente, 2006; Tennen et al., 2011)]. Mammalian SIRT1 regulates rDNA repeats and telomeres while SIRT6 is linked with the regulation of telomeres and the lifespan-modulating NFκB-dependent genes (Murayama et al., 2008; Kawahara et al., 2009; Palacios et al., 2010). Shelterin proteins, which are telomeric factors that are dysfunctional in premature aging patients suffering from Werner or Bloom syndromes, or other telomeric factors, may thus maintain cellular lifespan by controlling the distribution of histone deacetylases such as SIRT1 or SIRT6 between telomeres and other age-related loci (Martinez & Blasco, 2010). In addition, Sir2, SIRT1, SIRT6, and other sirtuins play various roles in regulating metabolic responses to nutrient availability, and it is therefore possible that Rif1 may affect cross talk between chromosomal and metabolic roles of sirtuins (Chalkiadaki & Guarente, 2012).
All in all, our work highlights the importance of fine tuning the distribution of sirtuin histone deacetylases between various age-related loci in yeast and points to possibly similar processes in other organisms. We expect that future work aiming to identify additional factors that directly or indirectly modulate the localization of sirtuins within nuclear space should reveal processes that fine-tune chromatin structure to sustain the lifespan of many organisms.
Strains and materials
Endogenous genes were deleted as described (Mekhail et al., 2008). Yeast strains are listed in Table S3. Antibodies used were either commercially available or previously generated. The anti-Sir2 antibody was used previously (Mekhail et al., 2008) and the anti-H3 and anti-diAcH3-K9-K14 antibodies were commercially available (Millipore).
Replicative lifespan of yeast strains were determined by micromanipulation as previously described (Kaeberlein et al., 1999). The number of times that a new mother cell buds was measured and the Wilcoxon rank-sum test was conducted for statistical analysis. For each genotype, two independent clones (20 cells for each) were analyzed and data was pooled from two to four separate experiments. Only simultaneously counted isogenic strains were compared to each other on graphs and within our statistical analyses.
Experiments were performed as described with modifications (Huang & Moazed, 2003; Mekhail et al., 2008; Chan et al., 2011). Relative fold enrichments were obtained according to the following calculations: [rDNAIP/CUP1IP]/[rDNAInput/CUP1Input], [TELIP/ACT1IP]/[TELInput/ACT1Input], or [HMIP/ACT1IP]/[HMInput/ACT1Input]. Full experimental procedures are in the Supporting Information with primers listed in Table S4.
Assays were performed essentially as described (Kaeberlein et al., 1999; Mekhail et al., 2008). Cells were grown to OD600 = 0.4–0.8, sonicated briefly, and spread (about 400 cells per plate) on thick plates (5 mg L−1 adenine). Incubation was at 30°C for 3 days, 4°C for 3 days, then RT for 1 day. Rates were obtained by dividing the number of half-sectored colonies by the total number of colonies excluding completely red colonies. For the calorie restriction USCE experiments, cells were grown overnight under CR conditions before re-diluting, further growing to OD600 = 0.4–0.8, and then plating on low adenine media.
Calorie restriction consisted of using media containing 0.05% glucose instead of the standard 2% glucose in YEP media. Triggering CR effects was confirmed by the extended lifespan of wild-type and other cells as well as the reduction of histone acetylation marks at silent chromatin domains.
Genomic DNA preparation and RNA extraction
For genomic DNA preparation, cells were subjected to bead-beating in the presence of Triton X-100 and phenol-chloroform-isoamyl alcohol before DNA precipitation. DNA pellets were washed, dried, resupended in TE and RNAse treated. For RNA extractions, total RNA was prepared via standard hot phenol extraction and EtOH precipitation before re-suspending pellets in 0.1% DEPC and treatment with DNase I. Full experimental procedures for genomic DNA and RNA extractions are included in the Supporting Information section online.
Semi-quantitative RT-PCR and quantitative PCR
Semi-quantitative reverse transcriptase PCR was performed as described with modifications (Xu et al., 2007). RT reactions relied on random nonamers and the M-MLV reverse transcriptase. Quantitative real-time PCR was performed using the CFX Connect Real-Time system (Bio-Rad). Full experimental procedures for RT and quantitative PCR are included in the Supporting Information section online.
We thank Marc Ting, Mark McCormick, and Brian Kennedy for strains; Betty P.K. Poon for discussions and comments. J.S.S. is supported by the Canadian Institutes of Health Research (CIHR) Vanier Doctoral Canada Graduate Scholarship. T.T.L. is supported in part by the Linda and Avrum Gotlieb Excellence Award. This work was supported by a CIHR grant to K.M., who holds the Canada Research Chair in Spatial Genome Organization. The authors do not have any conflict of interest to declare.
Experimental design by J.S.S., J.N.Y.C., and K.M.; writing by J.S.S. and K.M.; text editing by J.S.S., J.N.Y.C., and K.M; strains by J.S.S. and J.N.Y.C.; ChIP experiments by J.N.Y.C. primarily and also by J.S.S.; USCE by J.S.S. and J.D.W.; RT-PCR, qPCR, and Western Blot analyses by J.N.Y.C.; lifespan analyses by J.S.S., C.P., and T.T.L.