SIRT1 contributes to telomere maintenance and augments global homologous recombination

Telomeres are specialized nucleoprotein structures that protect the ends of eukaryotic chromosomes from unscheduled DNA repair reactions and degradation (Chan and Blackburn, 2002; de Lange, 2005). In vertebrates, telomeres consist of TTAGGG repeats bound by a specialized multiprotein complex known as shelterin, which has fundamental roles in the regulation of telomere length and protection (Liu et al., 2004; de Lange, 2005). Because of the intrinsic inability of the DNA replication machinery to copy the end of linear molecules, and to endogenous DNA end–degrading activities, telomeres become progressively shorter after every cell division cycle (Harley et al., 1990; Blasco, 2005).

Telomere repeats are generated by a ribonucleoprotein reverse transcriptase known as telomerase (Greider and Blackburn, 1985); however, its abundance and activity in adult tissues is not sufficient to compensate for the progressive telomere attrition that occurs with aging (Collins and Mitchell, 2002). In humans, several studies have described an inverse correlation between telomere length and age in a variety of tissues and between telomere length and certain age-associated diseases (Cawthon et al., 2003; Panossian et al., 2003; Ogami et al., 2004; Canela et al., 2007). Also, factors that may decrease longevity, such as psychological stress or obesity, are proposed to negatively impact on telomerase activity and telomere length (Epel et al., 2004; Valdes et al., 2005). In the absence of telomerase, telomeres can be maintained by a recombination-dependent mechanism (Dunham et al., 2000) known as alternative lengthening of telomeres (ALT; McEachern et al., 2000), which involves DNA repair proteins such as the Mre11–Rad50–Nbs1 and SMC5–SMC6–MMS21 complexes (Jiang et al., 2005; Potts and Yu, 2007).

Silent information regulator 2 (Sir2) family of proteins is a group of NAD⁺-dependent deacetylases/ADP-ribosyltransferases (type III histone deacetylase) initially discovered in yeast and are identified as key regulators of health span and lifespan in yeast and other organisms (Haigis and Guarente, 2006). Yeast Sir2 functions encompass (a) repression of gene expression at the silent mating type loci HML and HMR (Klar et al., 1979; Rine et al., 1979) (b) to suppress recombination at the ribosomal DNA locus, thus preventing generation of toxic ribosomal DNA circles (Sinclair and Guarente, 1997), and (c) to maintain the heterochromatic state of telomeres (Bedalov et al., 2001; Xu et al., 2007).

Mammals have seven (SIRT1–7) known sirtuins, of which SIRT1 is the closest and best-characterized mammalian orthologue of yeast Sir2 (Brachmann et al., 1995; Frye, 1999). SIRT1 deacetylase activity has been involved in chromatin remodeling, gene silencing, and the DNA damage response (Kim and Um, 2008; Cantó and Auwerx, 2009; Milner, 2009). Furthermore, mounting evidence has connected SIRT1 to stress responses in mammals (Abdelmohsen et al., 2007; Wang et al., 2007; Westerheide et al., 2009). In particular, SIRT1 is recruited to the chromatin upon different DNA damage insults, where it favors efficient repair of double-strand breaks (DSBs) by homologous recombination (Oberdoerffer et al., 2008). In this regard, SIRT1 physically interacts and deacetylates the WRN helicase (Vaitiekunaite et al., 2007; Kahyo et al., 2008; Li et al., 2008; Law et al., 2009), in this manner modulating homologous recombination–dependent DSB DNA repair (Uhl et al., 2010). SIRT1 also targets Nbs1, a regulatory subunit of the Mre11–Rad50–Nbs1 complex (Yuan and Seto, 2007; Yuan et al., 2007). SIRT1-mediated Nbs1 deacetylation specifically inhibits Nbs1 phosphorylation and modulates its activity in intra–S phase checkpoint induction (Kastan and Lim, 2000; Lim et al., 2000). Together, these findings suggest a potential role of SIRT1 in the regulation of DNA repair pathways.

Much attention has recently been given to the role of SIRT1 in metabolic tissues, such as the liver, skeletal muscle, and adipose tissues, where it deacetylates a range of substrates, including key metabolic regulators PGC1α, UCP2, NFκB, and Foxo proteins, which in turn have pronounced effects on glucose homeostasis, insulin secretion, and lipid homeostasis (Liang et al., 2009). Among other metabolic effects, SIRT1 is known to participate in the activation of gluconeogenic genes and to increase both hepatic glucose output during calorie restriction (CR) and reverse cholesterol transport and fat mobilization in white adipose tissue (Brooks and Gu, 2009). We have previously generated mice that contain additional copies of the SIRT1 gene under the control of its natural regulatory elements and that express threefold SIRT1 in a homogenous and systemic manner across all tissues, known as SIRT1^super mice (Pfluger et al., 2008). These mice are protected from physiological damage produced by exposure to a high fat diet. In addition, SIRT1^super mice are protected from DNA damage and liver carcinogenesis and show decreased signs of aging, including decreased expression of the aging-associated gene p16Ink4a (Herranz et al., 2010).

In yeast, Sir2, together with Sir3 and Sir4, is recruited to telomeres through Rap1, and this complex spreads along subtelomeric sequences enforcing transcriptional silencing (Perrod and Gasser, 2003; Bühler and Gasser, 2009). A Rap1-like protein is conserved in both human and fission yeast (Konig et al., 1996; Li et al., 2000; Park et al., 2002). However, human Rap1 lacks the DNA-interacting Myb domain and does not bind telomeric repeats. Other Myb box–related-containing proteins in mammals such as TRF2 bind directly to telomeric DNA repeats (Broccoli et al., 1997) and mediate the association of human Rap1 to telomeres (de Lange, 2009; Takai et al., 2010). A potential role of SIRT1 at mammalian telomeres is, therefore, less clear. More recently, SIRT1 depletion has been shown to cause telomere dysfunction (El Ramy et al., 2009). Interestingly, activation of SIRT1 with resveratrol produces an increase in extrachromosomal telomeric DNA and in colocalization of telomeric TRF1 with WRN helicase and BRCA1 in cells that elongate telomeres using ALT (Rusin et al., 2009).

In this study, we set to understand a potential role of mammalian SIRT1 in telomere biology by studying both loss of function (SIRT1-deficient mice) and gain of function (SIRT1^super mice) mouse models (Cheng et al., 2003; Pfluger et al., 2008). Our results indicate that mammalian SIRT1 interacts with the mouse telomeric repeats and acts as a positive regulator of telomere length in vivo by significantly attenuating telomere shortening associated with mouse aging. In addition, we find an important role of SIRT1 in promoting recombination at different chromosomal regions, including telomeres, centromeres, and chromosome arms. These findings link SIRT1 to telomeres and provide new mechanistic explanations for the known roles of augmented SIRT1 expression in protection from DNA damage and in prevention from some age-associated pathologies.

To address a potential role of SIRT1 in telomere length maintenance, we performed both telomere restriction fragment (TRF) analysis and metaphase telomere quantitative FISH (Q-FISH) on MEFs derived from either SIRT1-deficient or SIRT1^super mice and their corresponding wild-type controls. We have previously shown that SIRT1^super MEFs have a threefold increase in SIRT1 expression compared with wild-type controls (Pfluger et al., 2008; Herranz et al., 2010). SIRT1^super MEFs showed a modest but significant increase in mean telomere length compared with wild-type MEFs both by TRF and Q-FISH analyses (Fig. 1 A; and Fig. S1, A and B). Q-FISH analysis further indicated that increased telomere length in SIRT1^super MEFs is accompanied by a significant decrease of signal-free ends (SFEs) or chromosome ends with undetectable telomere signals and by a decrease in the proportion of short telomeres (<30 kb) and an increase in the proportion of long telomeres (>80 kb; Fig. 1 B). These results indicate that elevated SIRT1 expression leads to improved telomere length maintenance in cultured primary MEFs. In turn, SIRT1-deficient MEFs showed significantly shorter telomeres compared with wild-type MEFs as determined both by Q-FISH and TRF analyses (Fig. 1 C; and Fig. S1, A and B). Shorter mean telomere length in SIRT1^−/− MEFs was accompanied by a significant increase in the proportion of short telomeres and a decrease in the proportion of long telomeres (Fig. 1 D). To confirm whether the observed effects were specifically associated with the SIRT1 protein, wild-type and SIRT1^−/− MEFs were transfected with a SIRT1-coding plasmid (for SIRT1 expression levels see Fig. S2, A and B) and evaluated its effect on telomere length by Q-FISH analysis. Both SIRT1^+/+ and SIRT1^−/− MEFs transfected with the SIRT1 plasmid showed an increase in telomere length compared with cells transfected with the control plasmid (Fig. 1 E). In all cases, increased telomere length was accompanied by a significant decrease in the proportion of short telomeres (<30 kb) and an increase in the proportion of long telomeres (>80 kb); an almost significant decrease in SFEs was also observed in the SIRT1-tranfected SIRT1^−/− MEFs (Fig. 1 F). Together, the analyses of gain of function and loss of function MEFs for SIRT1 strongly suggest a role for the mammalian homologue of yeast Sir2 in telomere length maintenance in vivo.

Given the presence of longer telomeres in SIRT1^super MEFs, we wondered whether increased SIRT1 expression also improved telomere length maintenance in the context of adult mouse tissues. We have recently shown that SIRT1 overexpression in mice has significant beneficial effects on liver function (Pfluger et al., 2008; Herranz et al., 2010). Thus, we set to measure telomere length in liver sections from age- and sex-matched SIRT1^super and SIRT^+/+ littermates using telomere Q-FISH. Telomere length analysis of the hepatic parenchyma of 2-yr-old SIRT1^super mice showed a significant increase in mean telomere length compared with wild-type controls (Fig. 2 A). Longer telomeres in the liver of old SIRT1^super mice were accompanied by a significant increase in the proportion of very long telomeres (>80 arbitrary units of fluorescence [auf]) and a significant reduction in the proportion of short telomeres (<60 auf; Fig. 2 B). Furthermore, supporting a role for SIRT1 in telomere maintenance, liver sections from 2-yr-old SIRT1^−/− mice showed significantly shorter telomeres than those of age-matched SIRT1^+/+ controls in the hepatic parenchyma (Fig. 2 D). In concordance with the results in MEFs, shorter mean telomere length was accompanied by a significant increase in the proportion of very short telomeres (<100 auf) and by a significant decrease in the proportion of very long telomeres (>120 auf; Fig. 2 E). The same analysis was conducted in kidney sections (both in renal glomeruli and tubules) from 2-mo- and 2-yr-old mice. Mean telomere length decreased both in the glomeruli and tubules from 2-yr-old mice compared with 2-mo-old wild-type mice, which is in agreement with our previous observations showing telomere shortening associated with old age in many mouse tissues (Flores et al., 2008). Notably, this trend is not observed in SIRT1^super mice, where 2-yr-old mice showed longer telomeres compared with 2-mo-old mice (Fig. 2 G). Longer telomeres in the kidney of SIRT1^super mice were accompanied by an increase in the proportion of cells with long telomeres and a reduction in the proportion of cells with short telomeres in SIRT1^super mice in both age groups (Fig. 2, G and H). In particular, although wild-type kidneys showed an increase in the percentage of cells with short telomeres in 2-yr-old mice compared with 2-mo-old mice, which was concomitant with a decrease in the percentage of cells with very long telomeres (Fig. 2 H), these changes were not observed in kidneys from SIRT1^super mice (Fig. 2 H). Together, these results suggest that a moderate increase in SIRT1 expression (threefold) has beneficial effects on telomere length maintenance during aging. This finding is in agreement with the fact that SIRT1^super mice are protected from some age-related diseases such as hepatic steatosis, insulin resistance, liver cancer associated with metabolic syndrome, and osteoporosis (Pfluger et al., 2008; Herranz et al., 2010).

In most eukaryotes, telomeres are maintained by telomerase, a reverse transcription that adds telomeric repeats de novo after every cell division cycle, in this manner counteracting for the incomplete DNA replication of telomeres caused by the end replication problem (Collins and Mitchell, 2002; Chan and Blackburn, 2004). To address the mechanisms by which cells and tissues from SIRT1-overexpressing mice show improved telomere length maintenance, we first set to measure telomerase activity by using the in vitro TRAP assay (see Materials and methods). As shown in Fig. 3 (A and B), in vitro telomerase activity was similar in wild-type and SIRT1^super MEFs, indicating that increased SIRT1 expression does not significantly affect telomerase in vitro activity.

Next, we addressed whether longer telomeres in SIRT1^super MEFs were dependent on the presence of a proficient telomerase complex. To this end, we generated SIRT1^super mice simultaneously deficient for the essential telomerase RNA component (Terc) gene and measured telomere length in first-generation SIRT1^super;Terc^+/+, SIRT1^super;Terc^−/−, SIRT1^+/+;Terc^+/+, and SIRT1^+/+;Terc^−/− MEFs. Southern blot–based TRF analysis confirmed longer telomeres in SIRT1^super MEFs compared with wild-type controls in a telomerase-proficient Terc^+/+ background (Fig. 3, C and D). Terc deficiency resulted in significant telomere shortening in both the SIRT1^+/+ and SIRT1^super backgrounds, indicating similar rates of telomere erosion in the absence of telomerase activity. Further analysis using the more-sensitive Q-FISH technique confirmed a significant increase in SFEs and in the frequency of short telomeres (<20 kb) both in SIRT1^+/+;Terc^−/− and SIRT1^super;Terc^−/− MEFs (Fig. 3, E and F), again indicating that telomerase deficiency causes similar rates of telomere erosion independently of SIRT1 amounts. Q-FISH analysis also indicated that SIRT1^super;Terc^−/− MEFs showed a similar mean telomere length to that of SIRT1^+/+;Terc^−/− controls, again suggesting that longer telomeres in SIRT1^super MEFs are dependent on the presence of telomerase activity. Interestingly, the percentage of SFEs and cells with short telomeres was significantly increased in SIRT1^super;Terc^−/− MEFs compared with SIRT1^+/+;Terc^−/− controls, suggesting that augmented SIRT1 expression may accelerate telomere loss when in the absence of telomerase activity. In line with this, survival curves of a small cohort of SIRT1^super;Terc^+/+, SIRT1^+/+;Terc^+/+, SIRT1^super;Terc^−/−, and SIRT1^+/+;Terc^−/− mice show that SIRT1^super;Terc^−/− mice have a significant decreased survival when compared with SIRT1^+/+;Terc^−/− mice (Fig. S5), which further suggests a negative effect of increased SIRT1 expression in the context of telomerase deficiency. Together, these results suggest that maintenance of longer telomeres in SIRT1^super MEFs requires the presence of a proficient telomerase complex.

Telomere maintenance by ALT relies on recombination events between telomeric sequences. To address whether SIRT1 influences telomere recombination frequencies, we performed chromosome orientation FISH (CO-FISH) on SIRT1^super and SIRT1^−/− MEFs and their respective wild-type controls, which measures the frequency of sister chromatid exchanges (SCEs) specifically at the telomeric repeats (T-SCE; Bailey et al., 1996). Strikingly, MEFs overexpressing SIRT1 showed a significant increase in the frequency of SCEs at telomeres compared with their wild-type controls (Fig. 4, A and B), suggesting that increased SIRT1 expression favors telomeric recombination. Interestingly, SIRT1 abrogation in MEFs did not significantly affect T-SCEs frequencies compared with wild-type controls. Together, these results suggest that SIRT1 gain of function significantly augments homologous recombination between telomeric sequences but that SIRT1 activity is not essential to mediate these events. Notably, increased telomere recombination may be responsible for accelerated telomere loss of SIRT1^super MEFs compared with controls when a telomerase-deficient background is present (Fig. 3 E). In particular, homologous recombination at telomeric repeats can occur at multiple points, leading to an unequal exchange of telomeric repeats, thus generating chromosomes with unequal telomere lengths including short and long telomeres (Bailey et al., 2004; Blagoev and Goodwin, 2008).

Next, we set to determine whether the effect of increased SIRT1 expression on homologous recombination was specific of telomeric regions or reflected a global role of SIRT1 in favoring homologous recombination throughout the genome. To this end, we measured SCE events both at centromeric repeats (C-SCE; Jaco et al., 2008) and chromosome arms (SCE; see Materials and methods). As shown in Fig. 4 (C and D), SIRT1^super MEFs showed increased frequencies of C-SCE, whereas SIRT1^−/− MEFs showed normal frequencies compared with wild-type MEFs. SCE frequencies at chromosome arms were also significantly elevated in SIRT1^super MEFs compared with wild-type controls or SIRT1^−/− MEFs, confirming an impact of increased SIRT1 expression on global recombination frequencies (Fig. 4, E and F).

As SIRT1 deficiency in MEFs results in shorter mean telomere length compared with wild-type MEFs and SIRT1 overexpression significantly increases SCE frequencies, we wondered whether SIRT1 abrogation or SIRT1 overexpression could lead to increased damage at telomeres. γ-H2AX foci have been previously shown to indicate the presence of DNA DSBs, including those associated with critically short and dysfunctional telomeres, also known as TIFs (d’Adda di Fagagna et al., 2003; Takai et al., 2003). SIRT1 overexpression in SIRT1^super MEFs did not significantly increase the occurrence of TIFs when compared with their respective wild-type controls (Fig. 5, A and B). In contrast, SIRT1-deficient MEFs showed a significant increase in the proportion of cells with more than three TIFs compared with the wild-type controls (Fig. 5, A and B). These results suggest that although increased SIRT1 expression has no deleterious effects on telomere function, SIRT1 deficiency results in significant telomere dysfunction.

To further explore a putative role of SIRT1 in telomere protection, we performed metaphase Q-FISH analysis with a telomeric probe on SIRT1^super and SIRT1^−/− MEFs and their respective wild-type controls. SIRT1^−/− MEFs showed a significantly increased proportion of chromosome ends with multiple telomeric signals (MTS) (Fig. 6, A, C, and D), a type of aberration recently related to increased telomere fragility (Muñoz et al., 2005; Blanco et al., 2007; Martínez et al., 2009; Sfeir et al., 2009). In particular, MTS have been proposed to result from replication fork stalling and increased breakage at telomeres, as they are increased by aphidicolin treatment, which is known to inhibit DNA synthesis and induce DNA breaks at chromosome-fragile sites (Durkin and Glover, 2007). In concordance with this notion, MTS were increased in wild-type MEFs treated with aphidicolin and further increased in SIRT1-deficient MEFs (Fig. 6, A, C, and D), suggesting that SIRT1-deficient cells are more prone to replication fork stalling and breakage. However, chromosome end-to-end fusions were not increased by SIRT1 deficiency, indicating that SIRT1 is not essential for telomere capping (Fig. 6 B). Importantly, SIRT1 overexpression in SIRT1^super MEFs did not increase MTS or any other type of telomere aberration (Fig. 6, A–D), again supporting the notion that increased SIRT1 expression has no deleterious effects on telomere function and integrity. Absence of TIFs and telomere aberrations (fusions and MTS) associated with SIRT1-increased expression is in agreement with normal telomere capping in these cells. Notably, decreased SFE associated with SIRT1 overexpression (Figs. 1 and 3) may reflect longer telomeres in these cells, as not all SFEs (telomeres under the Q-FISH detection level) correspond to TIFs.

Telomeric chromatin is composed of TTAGGG repeats bound by shelterin. Mammalian telomeres also contain arrays of nucleosomes enriched in histone heterochromatic marks (Blasco, 2007), and they are associated with long, noncoding telomeric RNAs or TERRA (Azzalin et al., 2007; Schoeftner and Blasco, 2008). Telomeric and subtelomeric chromatin are characterized by a high density of H3K9me3 and H4K20me3 histone trimethylation marks and by HP1 binding (García-Cao et al., 2004; Gonzalo et al., 2005). In addition, subtelomeric DNA repeats are heavily methylated (Blasco, 2007).

Given that SIRT1 has histone deacetylase activity, we first set to determine the abundance of different histone marks at telomeric and pericentromeric chromatin in SIRT1^super MEFs and their corresponding wild-type controls by using chromatin immunoprecipitation (ChIP) followed by detection of telomeric repeats (see Materials and methods). Interestingly, we observed a significant reduction in the density of H3K9Ac at telomeres when SIRT1 deacetylase was overexpressed (Fig. 7, A and B). These results are in line with a role of SIRT1 in regulation of acetylation levels at mammalian telomeres. Interestingly, when a similar analysis was performed on pericentromeric chromatin, we also observed decreased H3K9Ac, which almost reached significance, and a significant increase in H4K20me3 trimethylation (Fig. 7, C and D), which is suggestive of induction of a more heterochromatic state. In line with this, when we performed a similar ChIP analysis on SIRT1^−/− MEFs, we observed that absence of SIRT1 results in a significant decrease of the heterochromatic marks H3K9me3 and HP1-γ at telomeric repeats (Fig. 7, E and F), a trend that did not reach significance in the case of pericentomeric repeats (Fig. 7, G and H). ChIP values were corrected both by the respective telomere and centromere inputs and by H3 and H4 abundance at these regions (Fig. 7). We did not detect significant changes in the density of the TRF1 shelterin protein at telomeres (Fig. 7, A, B, E, and F). As negative control, TRF1 was not bound to pericentromeric chromatin (Fig. 7, C, D, G, and H). Together, these results suggest that SIRT1 levels negatively modulate histone acetylation at telomeric chromatin, concomitantly increasing heterochromatic marks such as H3K9me3 and H4K20me3.

We next analyzed the impact of both SIRT1 deficiency and SIRT1-increased expression on global DNA methylation measured as the proportion of methylated CpG at the B1 SINE repeats, and we determined DNA methylation of subtelomeric sequences located at chromosomes 1 and 2 by bisulphite sequencing (see Materials and methods). As shown in Fig. S3 and Fig. S4, we did not find significant differences in DNA methylation levels at these regions between SIRT1^super, SIRT1^−/−, and their respective wild-type controls.

To address whether the effects of SIRT1 on telomere length, telomere heterochromatin, and telomere integrity (fragility and recombination) were mediated by binding of SIRT1 to telomeres in vivo, we performed ChIP analysis of telomeric and centromeric repeats with SIRT1 antibodies. We did not detect specific SIRT1 binding to telomeric chromatin in MEFs. To rule out this apparent lack of SIRT1 binding to telomeric repeats as the result of a transient association, we set to perform SIRT1 ChIP analysis in cells with very long telomeres such as those of pluripotent stem cells (Flores et al., 2008; Marion et al., 2009b). Interestingly, stem cells express high levels of SIRT1 compared with differentiated tissues (McBurney et al., 2003; Kuzmichev et al., 2005; Saunders et al., 2010), and SIRT1 has been shown to regulate neural and glial specification of neural precursors (Prozorovski et al., 2008), differentiation of skeletal myoblasts (Fulco et al., 2008), and to inhibit spermatogenesis (Coussens et al., 2008). To this end, we produced iPS from SIRT1^+/+ and SIRT1^−/− MEFs and evaluated SIRT1 binding to telomere repeats in vivo. As shown by ChIP analysis, SIRT1 specifically binds to telomeric repeats but not to pericentric major satellite regions in iPS cells (Fig. 8, A and B). These results suggest that SIRT1 effects on telomere function and in length regulation may be mediated by binding of SIRT1 to telomeric repeats.

The Sir2 family of nutrient-responsive NAD⁺-dependent deacetylases has emerged as key regulator of lifespan in yeast and other organisms (Haigis and Guarente, 2006). CR studies have shown that SIRT1, the closest mammalian homologue of yeast Sir2, also plays an important role in longevity and resistance to age-associated diseases in higher organisms by increasing health span (Chen et al., 2005; Bordone et al., 2007; Boily et al., 2008; Campisi and Yaswen, 2009; Cantó and Auwerx, 2009; Cohen et al., 2009; Saunders and Verdin, 2009; Herranz et al., 2010). However, the molecular mechanisms by which SIRT1 impacts on CR, lifespan, and health span are still largely unknown (Wood et al., 2004).

Telomere shortening is associated with organismal aging and increased telomeric damage, proposed to be causative of aging-associated pathologies in humans (Blasco et al., 1997; Flores et al., 2005; Ornish et al., 2008; Martínez et al., 2009). Generation of the first knockout mice for the telomerase RNA component (Blasco et al., 1997) showed a conserved role for telomerase as the main activity responsible for maintaining telomere length in mammals and uncovered its importance in tissue renewal and lifespan and showed that short telomeres could cause multiple defects associated with decreased adult stem cell functionality. Remarkably, telomerase-deficient mice from the first generation already show a decreased median and maximum life span, indicating that telomere length and telomerase are rate limiting for mouse longevity in spite of mice born with very long telomeres (García-Cao et al., 2004). This is in line with recent findings showing that mouse telomeres suffer a dramatic shortening when comparing old (2-yr old) with young individuals (2-mo old; Flores et al., 2008). Furthermore, decreasing the rate of telomere erosion by forcing telomerase expression in adult tissues significantly delays aging and extends the mean lifespan of cancer-resistant mice by 40% (Tomás-Loba et al., 2008).

In this study, we set out to address whether SIRT1 may have an impact on telomere length maintenance by studying both loss of function (SIRT1^−/− mice) and gain of function mouse models for SIRT1 (SIRT1^super mice). The latter mouse model is particularly interesting for aging studies, as we recently showed that SIRT1^super mice have an increased health span and are protected from metabolic syndrome and liver cancer associated with metabolic syndrome (Pfluger et al., 2008; Herranz et al., 2010). Furthermore, several SIRT1 direct and indirect activators, including resveratrol, have been recently described to have beneficial effects on several age-associated diseases (Baur et al., 2006; Lagouge et al., 2006; Milne et al., 2007; Feige et al., 2008; Pearson et al., 2008; Yamazaki et al., 2009).

In this study, we find that overexpression of SIRT1 in mice decreases the rate of telomere erosion associated with cell division and tissue aging (i.e., liver and kidney), whereas SIRT1 abrogation results in an increased telomere erosion. The beneficial effects of increased SIRT1 activity on telomere length maintenance occur without any detrimental effects on telomere integrity or telomere capping. Furthermore, we show that improved maintenance of telomeres in SIRT1^super cells is dependent on an active telomerase complex, suggesting that SIRT1-increased expression impacts the telomerase pathway. Interestingly, it has been previously described that silencing of SIRT1 by RNAi results in an increase of telomerase expression in HeLa cells (Narala et al., 2008). Our findings suggest that the increased health span of SIRT1^super mice may be due, at least in part, to a better maintenance of telomere length with increasing age (Pfluger et al., 2008; Herranz et al., 2010). Interestingly, both mean and maximum longevity are not increased in SIRT1^super mice in spite of increased health span (Herranz et al., 2010) and longer telomeres (this study), which goes in line with our recent observations that improved telomere maintenance increases longevity only when in highly cancer-resistant backgrounds (Tomás-Loba et al., 2008). Finally, these findings also open the interesting possibility that SIRT1-dependent effects of CR, or the beneficial effects of SIRT1 activators, could be partially mediated by SIRT1-positive effects on telomere length maintenance, although this remains to be addressed.

In addition, SIRT1^super cells showed augmented frequencies of sister chromatid homologous recombination events at telomeres, centromeres, and chromosome arms, indicating that SIRT1-increased expression significantly impacts homologous recombination. Interestingly, these results also suggest that increased SIRT1 expression may positively impact DNA repair efficiency, thereby also contributing to increased tissue fitness and health span. In support of this, SIRT1 has been implicated in protection from DNA damage–induced cell death and carcinogenesis (Oberdoerffer et al., 2008; Wang et al., 2008; Herranz et al., 2010). Finally, SIRT1 is likely to influence homologous recombination through deacetylation of Nbs1 and WRN helicase, two important players in DNA repair by homologous recombination (Yuan et al., 2007; Uhl et al., 2010).

In contrast with the effects of SIRT1 overexpression on telomere length maintenance and improved homologous recombination, we found that SIRT1 deficiency had a negative impact both on telomere length maintenance and telomere integrity, which is in agreement with recent data (El Ramy et al., 2009). In particular, SIRT1^−/− MEF showed an increase in chromosomes with MTS, which are further increased by aphidicolin treatment, indicative of increased telomere fragility and breakage. Accordingly, cells lacking SIRT1 showed a higher level of telomere damage (TIFs). Together, these findings suggest a role of SIRT1 on preventing telomere damage, which could again be mediated by some of its known deacetylation targets, including Nbs1 and WRN. Additionally, SIRT1-altered expression resulted in changes in the heterochromatic status of telomeres, which go in line with a model where SIRT1 levels negatively modulate histone acetylation at telomeric chromatin, concomitantly increasing heterochromatic marks such as H3K9me3 and H4K20me3. Previous studies have shown that SIRT1 can deacetylate H3K9Ac and H4K16Ac (Vaquero et al., 2004; Vaquero et al., 2007), H3K56Ac (Chen et al., 2008; Das et al., 2009), and H3K26Ac (Oberdoerffer et al., 2008) and as a consequence of decreased acetylation, can lead to higher abundance of heterochromatic marks like H2K9me3 and H4K20me1 (Vaquero et al., 2004). A SIRT1-dependent increase in H4K20me3 has not been previously reported. However, SIRT1 can specifically target the histone methyltransferase suppressor of variegation 3-9 homologue (Suv39h1), leading to increased H2K9me3 (Vaquero et al., 2007), and therefore, H2K9me3 recruits HP1, which is necessary to assemble H4K20me3 at pericentric chromatin (Schotta et al., 2004). Importantly, we show in this study that SIRT1 can interact in vivo with telomeric repeats in the context of iPS cells, arguing that SIRT1 roles in regulation of telomere length, integrity, and recombination and telomere heterochromatin could be explained at least in part by a direct SIRT1 binding to telomeres in vivo.

In summary, the findings described in this study demonstrate that increased expression of SIRT1, the closest mammalian orthologue of yeast Sir2, improves telomere length maintenance in vivo and significantly increases recombination frequencies at telomeres, centromeres, and chromosome arms. These effects of increased SIRT1 expression are potentially beneficial to preserve genome integrity and stability and open new avenues to understand the known effects of increased SIRT1 expression on health span and protection from some age-associated diseases.

SIRT1^super (C57BL/6 background) and SIRT1^−/− (mixed 129/Sv and C57BL/6 background) mice were provided by M. Serrano (Spanish National Cancer Centre, Madrid, Spain; Pfluger et al., 2008) and F. Alt (Harvard University, Cambridge, MA), respectively (Cheng et al., 2003). To generate SIRT1^super Terc^+/+, SIRT1^+/+ Terc^+/+, SIRT1^+/+ Terc^+/+, and SIRT1^+/+ Terc^−/− mice, SIRT1^super mice were crossed with Terc^−/− (C57BL/6 background) animals (Blasco et al., 1997), and the resulting double-heterozygous breeding pairs were used to generate G1 (first generation) of single Terc^−/− SIRT1^super and SIRT^+/+ mice. Mice were housed in a Specific Pathogen Free facility and treated according to the Federations of European Laboratory and Animal Science Association. All animals were maintained in a normal maintenance diet (2018; Harlan Laboratories).

Primary MEFs (passage 2) were obtained from SIRT1^+/+, SIRT1^−/−, SIRT1^super, SIRT1^super Terc^+/+, SIRT1^+/+ Terc^+/+, SIRT1^+/+ Terc^+/+, and SIRT1^+/+ Terc^−/− embryos as described previously (Blasco et al., 1997). To generate iPS cells, MEFs from the indicated genotypes were reprogrammed as described previously (Marión et al., 2009a,b). SIRT1^+/+ and SIRT1^−/− MEFs were transfected with pHA-SIRT1 (#10962; Addgene) or pCruz-HA (sc-5045; Santa Cruz Biotechnology, Inc.) plasmids using the Neon Transfection System (Invitrogen) according to the manufacturer’s conditions, and cells were harvested 4 d after transfection.

Telomerase activity was measured with a modified TRAP as previously described (Blasco et al., 1997). Whole cell extracts were incubated with (UUAGGG)₃ or (CCCUAA)₃ RNA oligonucleotides for 5 min on ice and for 5 min at room temperature before beginning the telomerase extension reaction. To degrade spiked RNA oligonucleotides before PCR amplification, samples were treated with 10 pg RNase A (QIAGEN) at room temperature for 20 min.

Cells were harvested by trypsinization, washed once with PBS, lysed with cold RIPA, resuspended in SDS-PAGE loading buffer, and sonicated. Equal amounts of protein (50–100 mg) were analyzed by gel electrophoresis followed by Western blotting. The following antibodies were used for immunoblotting: rabbit anti-SIRT1 (ab12913; Abcam), mouse anti-HA (sc-7392; Santa Cruz Biotechnology, Inc.), and anti–α-tubulin (T6557; Sigma-Aldrich) as a loading control. Antibody binding was detected after incubation with a secondary antibody coupled to horseradish peroxidase using enhanced chemiluminescence.

MEFs growing in coverslips were permeabilized by incubation in PBS with 0.5% BSA and 0.1% Triton X-100 for 15 min. After permeabilization, slides were blocked in 2% BSA for 1 h and incubated with primary antibody. Permeabilized sections were incubated with rabbit anti-TRF1 (1:200) prepared as described previously (Muñoz et al., 2005), mouse monoclonal anti-phospho–histone H2AX antibody (1:500; Millipore), and secondary antibodies Cy3-conjugated goat anti–rabbit (1:400) and Alexa Fluor 555 donkey anti–mouse (1:300; Jackson ImmunoResearch Laboratories, Inc.). Slides were washed in PBS with 0.1% Triton X-100 and mounted in Vectashield with DAPI (Vector Laboratories). Confocal microscopy was performed at room temperature with a laser-scanning microscope (TCS SP5; Leica) using a Plan Apo 63× 1.40 NA oil immersion objective (HCX; Leica). The pictures show the maximal projection of z stacks generated using advanced fluorescence software (LAS; Leica).

Q-FISH was performed directly on liver and kidney sections as previously described (Zijlmans et al., 1997; Muñoz et al., 2005). Confocal microscopy was performed at room temperature with a laser-scanning microscope (TCS SP5) using a Plan Apo 63× 1.40 NA oil immersion objective (HCX). The pictures show the maximal projection of z stacks generated using advanced fluorescence software (LAS). Images were analyzed with Definiens XD software package. The DAPI image was use to define the nuclear area, and the fluorescent PNA telomeric probe screen was used to detect single telomeric signal inside of each nuclei.

We prepared metaphases and performed Q-FISH hybridization as previously described (Samper et al., 2000; Gonzalo et al., 2006). To correct for lamp intensity and alignment, images from fluorescent beads (Invitrogen) were analyzed in parallel using the TFL-Telo program (provided by P. Lansdorp, British Columbia Cancer Research Centre, Vancouver, British Columbia, Canada). Telomere fluorescence values were extrapolated from the telomere fluorescence of lymphoma cell lines LY-R (R cells) and LY-S (S cells) with known telomere lengths of 80 kb and 10 kb, respectively. We captured the images at room temperature using a charge-coupled device camera (FK7512; COHU) on a fluorescence microscope (DMRB; Leica) with a Plan Apo 100× 1.40 NA oil immersion objective (HCX; Leica) using Q-FISH software (Leica) in a linear acquisition mode to prevent the saturation of fluorescence intensity. TFL-Telo software (Zijlmans et al., 1997) was used to quantify the fluorescence intensity of telomeres from at least 10–15 metaphases for each data point. When indicated, primary MEFs were treated with 0.5 µM aphidicolin for 24 h. For analysis of chromosomal aberrations, metaphases were analyzed by superimposing the telomere image on the DAPI image using the TFL-Telo software.

Cells were included in agarose plugs, and TRF analysis was performed as described previously (Blasco et al., 1997). Mean telomere length was calculated by densitometric analysis using ImageJ (National Institutes of Health). After building 2D representation of signal intensity of each lane, mean telomere length was defined as the intersection point of intersection of the x axis and a line that divided the area below the curve in equal parts.

Global DNA methylation levels were determined using the B1-SINE Cobra method as previously described (Benetti et al., 2007a,b). Estimation of the fraction of methylated B1 elements for each genotype was performed using the following formula: ([molarity of 45-bp band]/2)/([molarity of 45-bp band] + [molarity of 100-bp band]).

DNA methylation of subtelomeric genomic regions of chromosomes 1 and 2 (q-arms) was performed by PCR analysis after bisulfite modification as described previously (Benetti et al., 2007a,b). To examine the methylation status of subtelomeric CpG islands, 17–55 colonies were automatically sequenced for each cell line. Bisulfite genomic sequencing primers used against subtelomeric regions in chromosomes 1 and 2 were 5′-CACCTCTAACCACTTAAACCTAACAA-3′ and 5′-GGGAGTGGAAGGAATTAGTAGGTT-3′, which flank positions 197,042,227–197,042,569 of chromosome 1 in the mouse NCBI36 genome assembly, and 5′-TTACCAATACCACCATTCCTCCA-3′ and 5′-GAGAGTAGTTAATTAGATGAGGAATA-3′, which flank positions 181,837,807–181,838,281 at chromosome 2 in the mouse NCBI36 genome assembly. Results were analyzed for DNA methylation analysis using BiQ Analyzer software (Applied Biosystems).

ChIP assays were performed as previously described (García-Cao et al., 2004). In brief, after cross-linking and sonication, chromatin from 4 × 10⁶ cells were used per each immunoprecipitation with Protein A/G Plus agarose beads (sc-2003; Santa Cruz Biotechnology, Inc.) and the following antibodies: 6 µg anti–histone H3 (ab1791; Abcam), 6 µg anti–histone H4 (ab10158; Abcam), 6 µg anti-H3K9me3 (#07-442; Millipore), 6 µg anti-H4K20me3 (#07-749; Millipore), 6 µg anti-SIRT1 (ab12193; Abcam), 8 µl rabbit polyclonal antibody to TRF1 (raised in our laboratory against full-length mouse TRF1 protein), and 10 µl monoclonal anti–γ-HP1 (#05-690; Millipore) or preimmune serum. The immunoprecipitated DNA was transferred to a Hybond N+ membrane using a dot blot apparatus. The membrane was then hybridized with either a telomeric probe containing TTAGGG repeats or a probe recognizing major satellite sequences, which is characteristic of pericentric heterochromatin. Quantification of the signal was performed with ImageQuant software (Molecular Dynamics). The amount of telomeric and pericentric DNA after ChIP was normalized for the total telomeric or centromeric DNA signal, respectively, for each genotype and for the H3 and H4 abundance at these regions, thus correcting for differences in the number of telomere repeats or in nucleosome spacing.

Cells were subcultured in the presence of BrdU (Sigma-Aldrich) at a final concentration of 10⁻⁵ M and allowed to replicate their DNA once at 37°C. Colcemid was added in a concentration of 0.2 µg/ml for the last 4 h. Cells were recovered, and metaphases were prepared as previously described (Samper et al., 2000). CO-FISH was performed as previously described (Gonzalo et al., 2006) using first a telomeric (CCCTAA)₇ PNA probe labeled with Cy3, a second telomeric (TTAGGG)₇ PNA probe labeled with fluorescein, or a minor satellite PNA probe labeled with Cy3 (Cy3-oo-TTCCAACGAATGTGTTTT), which hybridizes with the lagging DNA strand, followed by a second hybridization with a minor satellite PNA probe labeled with FITC (Flu-oo-AAAACACATTCGTTGGAA), which hybridizes with the leading DNA strand (Applied Biosystems). Metaphase spreads were captured on a fluorescence microscope (DMRB; Leica).

Genomic SCEs were visualized using an adapted fluorescence plus Giemsa protocol (Perry and Wolff, 1974). MEFs were grown at 37°C for two rounds of DNA replication in the presence of 3 µg/ul BrdU. Colcemid was added in a concentration of 0.2 µg/ml for the last 4 h. Cells were then recovered, and metaphases were prepared as described previously (Samper et al., 2000). Metaphase spreads on slides were immersed in 5 µg/ml Hoechst 333258 solution (Invitrogen) and washed with abundant water. The slides were exposed to UV light for 15 min in the presence of 2× SSC, washed with water, stained with Giemsa solution (Merck) for 2 min, and washed with water. Images were captured using a brightfield microscope (AX70; Olympus) using a Plan Apo 100× 1.4 NA oil immersion objective. Images were captured using a color camera (DP70; Olympus) and ImagePro Plus high-end image acquisition and analysis software (Media Cybernetics). All of the image capture was performed at room temperature. Metaphases were analyzed for harlequin staining. Each color switch was scored as one SCE.

Student’s t test was performed using Excel (97; Microsoft) or Prism software (version 5; GraphPad Software, Inc.). Fisher exact test was performed using SPSS version 15.

Fig. S1 shows TRF analysis of primary MEFs (SIRT1^super, SIRT1^+/+;SIRT1^+/+, and SIRT1^−/−) and quantification of their mean telomere length per genotype. Fig. S2 shows Western immunoblotting against SIRT1 and tubulin in primary MEFs of SIRT1^+/+ and SIRT1^−/− transfected with SIRT1 coding (pHA-SIRT1) and control plasmid. Fig. S3 shows global DNA methylation measured as methylations of B1-SINE repeats in primary MEFs (SIRT1^super, SIRT1^+/+;SIRT1^+/+, and SIRT1^−/−). Fig. S4 shows subtelomeric DNA methylation of chromosomes 1 and 2 in MEFs (SIRT1^super, SIRT1^+/+;SIRT1^+/+, and SIRT1^−/−). Fig. S5 shows survival curves of SIRT1^super x Terc^+/+, SIRT1^super x Terc^−/−, SIRT1^+/+ x Terc^+/+, and SIRT1^+/+ x Terc^−/− mice.

J.A. Palacios is funded by a Juan de la Cierva contract from the Spanish Ministry of Innovation and Science. M.A. Blasco’s laboratory is funded by the Spanish Ministry of Innovation and Science, the European Union (FP7-Genica), the European Research Council (ERC Advance Grants), the Spanish Association Against Cancer (AECC), and the Körber European Science Award (to M.A. Blasco).