|Home | About | Journals | Submit | Contact Us | Français|
Telomere structure allows cells to distinguish the natural chromosome ends from double-strand breaks (DSBs). However, DNA damage response proteins are intimately involved in telomere metabolism, suggesting that functional telomeres may be recognized as DNA damage during a time window. Here we show by two different systems that short telomeres are recognized as DSBs during the time of their replication, because they induce a transient MRX-dependent DNA damage checkpoint response during their prolonged elongation. The MRX complex, which is recruited at telomeres under these conditions, dissociates from telomeres concomitantly with checkpoint switch off when telomeres reach a new equilibrium length. We also show that MRX recruitment to telomeres is sufficient to activate the checkpoint independently of telomere elongation. We propose that MRX can signal checkpoint activation by binding to short telomeres only when they become competent for elongation. Because full-length telomeres are refractory to MRX binding and the shortest telomeres are elongated of only a few base pairs per generation, this limitation may prevent unscheduled checkpoint activation during an unperturbed S phase.
Telomeres, the physical ends of linear eukaryotic chromosomes, are composed of G/C-rich repeated sequences and the G-rich strand terminates with a 3′ single-stranded DNA (ssDNA) overhang known as G-tail. Similarly, interruptions in duplex DNA molecules are resected by 5′-3′ exonucleases to generate 3′-ended ssDNA tails, which elicit a DNA damage response, resulting in checkpoint-mediated cell cycle delay and DNA repair (reviewed in Longhese et al., 2006 ). Despite their resemblance to double-strand breaks (DSBs), telomeres are not subjected to homologous recombination, end-to-end fusions or other events that normally promote repair of DNA breaks (reviewed in Viscardi et al., 2005 ). Moreover, they do not activate the DNA damage checkpoint, whose key players belong to a protein kinase family, including mammalian ATM (Ataxia Telangiectasia Mutated) and ATR (Ataxia Telangiectasia and Rad3-related) and Saccharomyces cerevisiae Tel1 and Mec1 (reviewed in Longhese et al., 2006 ; Shiloh, 2006 ). This implies that cells must be able to differentiate natural chromosome ends from intrachromosomal DSBs, the latter serving as potent stimuli of the DNA damage response and resulting in checkpoint-mediated cell cycle arrest until repaired. This differentiation is thought to be the consequence of a unique organization of chromosomal ends into specialized nucleoprotein complexes (reviewed in Chakhparonian and Wellinger, 2003 ; Smogorzewska and de Lange, 2004 ). In humans, attenuation of checkpoint signaling at telomeres may be achieved through the action of TRF2 that binds directly to telomere repeat sequences and inhibits the ATM protein kinase (Karlseder et al., 2004 ). Conversely, budding yeast telomeres possess a yet unidentified telomeric “anticheckpoint” activity that might inhibit Mec1 signaling (Michelson et al., 2005 ). In any case, telomeric double-stranded DNA is bound by specialized DNA-binding proteins that protect the chromosomal ends from being sensed as DSBs (reviewed in d'Adda di Fagagna et al., 2004 ; de Lange, 2005 ; Viscardi et al., 2005 ) and ensure their complete replication via a specialized reverse transcriptase called telomerase (Hug and Lingner, 2006 ). Telomeric-capping function is dependent on proper DNA end structure as well as on telomere-binding proteins. In S. cerevisiae, the G-tail binding protein Cdc13 and its interacting factors Stn1 and Ten1 are essential components of the telomeric cap (Garvik et al., 1995 ; Grandin et al., 1997 , 2001 ). Moreover, the double-stranded telomeric repeat binding protein Rap1 is required to avoid telomere fusions (Pardo and Marcand, 2005 ).
Telomere length alterations by loss of telomere-associated proteins or by inhibition of telomere replication signal a DNA damage response. Extensive telomere shortening in yeast cells lacking telomerase triggers a Mec1-dependent checkpoint (Ijpma and Greider, 2003 ), as does inactivation of the duplex TG-repeat–binding protein Rap1 (Pardo and Marcand, 2005 ). Moreover, degradation of the C-rich strand caused by inactivation of the telomere end-binding protein Cdc13 leads to accumulation of ssDNA and subsequent DNA damage checkpoint activation (Garvik et al., 1995 ; Lydall and Weinert, 1995 ).
Although telomeres are apparently shielded from being recognized as DSBs, many DNA damage response proteins are associated with chromosomal termini and contribute to several aspects of telomere metabolism. For instance, budding yeast Tel1 and Mec1 exhibit cell cycle–dependent association to telomeres (Takata et al., 2004 ; Verdun et al., 2005 ). Paradoxically, chromatin immunoprecipitation experiments suggest that Tel1 and Mre11 may be at telomeres in different stages of the cell cycle (Takata et al., 2004 ). This contrasts with recent data showing that ATM and Nbs1 both colocalize to telomeres during G2 in human cells (Verdun et al., 2005 ). Tel1/ATM appears to be particularly important for telomere length maintenance in yeast, Drosophila, and mammalian cells, and its inactivation results in shortened telomeres and increased chromosomal end-to-end fusions by nonhomologous end joining (Metcalfe et al., 1996 ; Ritchie et al., 1999 ; Bi et al., 2004 ; Silva et al., 2004 ). Epistasis analysis shows that Tel1 acts at telomeres together with the MRX (Mre11-Rad50-Xrs2)/MRN (Mre11-Rad50-Nbs1) complex (Ritchie and Petes, 2000 ; Tsukamoto et al., 2001 ), which is required for the generation of 3′ ssDNA tails in a telomere formation assay (Diede and Gottschling, 2001 ) and for the proper establishment of the constitutive G-tail end structure (Larrivee et al., 2004 ). Moreover, MRX/MRN appears to protect telomeres from nuclease and DNA repair activities in both yeast and Drosophila (Bi et al., 2004 ; Ciapponi et al., 2004 ; Foster et al., 2006 ). MRX and Tel1 are thought to contribute to yeast telomere length maintenance by promoting telomerase association to telomeres. In fact, recruitment of both the telomerase catalytic subunit Est2 and the telomerase accessory protein Est1 is severely reduced in Mre11 or Tel1 lacking cells (Goudsouzian et al., 2006 ).
In any case, the finding that several factors known to be directly involved in DNA repair and DNA damage checkpoints localize at telomeres and contribute to telomere metabolism suggests that there may be a time window where functional telomeres are recognized as DNA breaks. Consistent with this hypothesis, we have previously shown that rapid telomere expansion caused by turning on Tel1 overexpression in cells with short but otherwise stable telomeres causes the activation of a Rad53-dependent checkpoint. This suggests that telomeres can be recognized as DSBs in a window of time during their replication (Viscardi et al., 2003 ). In this study, we try to assess the signals activating the checkpoint during telomere elongation by analyzing the checkpoint response during prolonged expansion of either a single or multiple shortened telomeres.
All yeast strains were derivatives of W303 (MATa or MATα, ade2–1, can1–100, his3–11,15, leu2–3,112, trp1–1, ura3) and are listed in Table 1. Strain GAL-TEL1, carrying one copy of the GAL1-TEL1 fusion disrupting the TEL1 chromosomal gene, was obtained as previously described (Viscardi et al., 2003 ). Strains expressing the fully functional MRE11-MYC18–tagged allele at the MRE11 chromosomal locus were constructed as previously described (Clerici et al., 2006 ). The GAL-TEL1 strain was used to disrupt the SAE2 and the MRE11 genes, to replace RAD50 with the rad50s-K81I allele (Alani et al., 1990 ) or to integrate the GAL-SAE2 fusion at the LEU2 locus. To generate rad50s mutants, GAL-TEL1 cells were transformed with MscI-digested plasmid pML533, carrying the first 666 base pairs (bp) of the RAD50 coding region with the K81I mutation. Strains Lev187, Lev189, and Lev220 were kindly provided by S. Marcand (Centre National de la Recherche Scientifique [CNRS], Fontenay-aux-Roses, France). Lev220 was used to disrupt the RAD50, MEC1, SML1, or TEL1 genes. GAL1-TEL1 strains carrying either the 2μ vector or 2μ RIF2 plasmid were constructed by transforming the GAL-TEL1 strain with plasmids YEplac195 (2μ URA3) and pML415 (2μ RIF2 URA3), respectively. cdc15–2 strains expressing the methionine-regulatable MET-CDC6 fusion at the CDC6 locus were kindly provided by S. Piatti (University of Milano-Bicocca, Milan, Italy) and were crossed to GAL-TEL1 MRE11-MYC18 strain to obtain strains DMP4702/2C, DMP4701/26C, DMP4699/26B, and DMP4700/34D.
The accuracy of all gene replacements and integrations was verified by Southern blot or PCR analysis. Standard yeast genetic techniques and media were according to Rose et al. (1990) . Cells were grown in YEP medium (1% yeast extract, 2% bactopeptone, 50 mg/l adenine) supplemented with 2% glucose (YEPD) or 2% raffinose (YEP+raf) or 2% raffinose and 1% galactose (YEP+raf+gal). Transformants carrying the KANMX4 cassette were selected on YEPD plates containing 400 μg/ml G418 (US Biological, Swampscott, MA).
Chromatin immunoprecipitation (ChIP) analysis was performed as in http://www.molbio.princeton.edu/labs/zakian/2004_methods.htm. Multiplex PCR reactions for Mre11 immunoprecipitates were carried out for 30 cycles. Gel quantitation was determined by using the NIH image program. The relative fold enrichments of telomere-bound Mre11 were calculated as follows: [TELIP/AROIP]/[TELinput/AROinput], where IP and input represent the amount of PCR product in the immunoprecipitates and input samples before immunoprecipitation, respectively. Sequences for all primers are available upon request.
Yeast DNA was prepared according to standard methods and digested with XhoI or StuI enzyme. The resulting DNA fragments were separated by gel electrophoresis in 0.8% agarose gel and transferred to a GeneScreen nylon membrane (New England Nuclear, Boston, MA), followed by hybridization with a 32P-labeled poly(GT) probe or a 32P-labeled probe corresponding of an 840-bp HindIII-StuI ADH4 fragment and exposure to x-ray–sensitive films. Standard hybridization conditions were used.
Visualization of the single-strand overhang at telomeres was done as described in Dionne and Wellinger (1996) . Total protein extract preparation and Western blot analysis were performed as described in Clerici et al. (2001) . Rad53 was detected using anti-Rad53 polyclonal antibodies kindly provided by J. Diffley (Clare Hall Laboratories, South Mimms, United Kingdom) and C. Santocanale (Nerviano, Milano, Italy). Secondary antibodies were purchased from Amersham (Piscataway, NJ), and proteins were visualized by an enhanced chemiluminescence system according to the manufacturer.
We previously showed that prolonged telomere elongation caused by turning on Tel1 overexpression in cells with short telomeres (due to the lack of Tel1 or Ku) triggers transient activation of a Rad53-dependent checkpoint that inhibits the G2/M transition (Viscardi et al., 2003 ). Prolonged telomere elongation may disrupt specialized capping functions, thus generating signals, such as blunt ends and/or 3′ single-stranded overhangs, which persist long enough to be detected as DNA lesions by the checkpoint machinery. The MRX complex is implicated together with Mec1 and Tel1 in triggering checkpoint activation after generation of intrachromosomal DSBs (Ira et al., 2004 ; Nakada et al., 2004 ; Clerici et al., 2006 ) and associates at telomeres during S phase (Zhu et al., 2000 ; Takata et al., 2005 ). We therefore evaluated the possible MRX involvement in the above telomere elongation and checkpoint activation. It is worth mentioning that a galactose-inducible GAL-TEL1 construct is the only Tel1 source in all the GAL-TEL1 strains used in this work (tel1::GAL-TEL1; see Table 1), which therefore have stable short telomeres when grown in the absence of galactose (Viscardi et al., 2003 ). G1-arrested cell cultures in raffinose-containing medium were released from α-factor in the presence of galactose. As shown in Figure 1, GAL-TEL1 cells transiently arrested, as expected, with 2C DNA content (Figure 1A) and underwent phosphorylation of the effector checkpoint kinase Rad53 (Figure 1C), while gradually elongating telomeres to a new equilibrium length (Figure 1B). Conversely, galactose-induced GAL-TEL1 cells lacking the Mre11 subunit of MRX did not undergo either telomere elongation (Figure 1B) or cell cycle arrest and Rad53 phosphorylation (Figure 1, A and C). Thus, the MRX complex is necessary for telomere elongation and checkpoint activation caused by ectopic Tel1 expression.
To assess whether transient telomere elongation and checkpoint activation in GAL-TEL1 cells involved MRX telomere association, we performed ChIP analysis in GAL-TEL1 cells carrying a fully functional MRE11-MYC–tagged allele. Sheared chromatin prepared at different time points after galactose addition from formaldehyde cross-linked cell samples was immunoprecipitated with anti-myc antibodies. Quantitative multiplex PCR was then used to monitor coimmunoprecipitation of the nontelomeric ARO1 fragment of chromosome IV, of the subtelomeric ADH4 fragment located 5.3 kilobase pairs (kbp) from telomere VII-L and of the unique TEL fragment, located 52 bp away from telomere VI-R (Figure 1D). We also monitored the efficiency of Mre11-myc immunoprecipitation (ppt in Figure 1D) to ensure that differences in telomere association were not due to different protein accessibility. As shown in Figure 1, D and E, Mre11 association to telomeres, which was enriched about threefold over background in G1-arrested uninduced GAL-TEL1 cells (αf), strongly increased (35-fold) in these cells within 2 h after release into the cell cycle in the presence of galactose, and it remained high during the subsequent 10 h. Mre11 telomere association under these conditions paralleled with telomere elongation, cell cycle arrest, and checkpoint activation. In fact, the amount of telomere-bound Mre11 gradually declined (Figure 1, D and E) at the same time telomere length reached a new equilibrium (Figure 1B) and the checkpoint was switched off (Figure 1, A and C).
MRX is required for the generation of 3′ ssDNA tails in a telomere formation assay (Diede and Gottschling, 2001 ) and for the proper establishment of the constitutive G-tail end structure (Larrivee et al., 2004 ). We therefore asked whether MRX prolonged binding at elongating telomeres may lead to accumulation of long single-stranded 3′ overhangs that are strong signals for checkpoint activation (Zou and Elledge, 2003 ). However, nondenaturing in-gel hybridization with a C-rich radiolabeled oligonucleotide (Dionne and Wellinger, 1996 ) of genomic DNA derived from either galactose-induced GAL-TEL1 or wild-type cells yielded faint signals compared with those detected in cells lacking Ku70, which are known to accumulate telomeric single-stranded G-tails (Figure 2; Gravel et al., 1998 ). This indicates that accumulation of ssDNA at telomeres is not likely responsible for checkpoint activation in galactose-induced GAL-TEL1 cells. This is in agreement with recent data showing that a DSB adjacent to the telomeric sequence causes checkpoint activation before ssDNA generation (Michelson et al., 2005 ).
If Mre11 telomere association were the signal for checkpoint activation in GAL-TEL1 cells, its increased persistence should lead to prolonged checkpoint activation. To evaluate this possibility, we used GAL-TEL1 cells lacking or overexpressing the Sae2 protein, which works in concert with MRX in DSB processing and in subsets of recombination pathways (Rattray et al., 2001 ; Lobachev et al., 2002 ; Clerici et al., 2005 ). Sae2 appears to regulate MRX persistence at DSBs. In fact, both the lack of Sae2 and the rad50s mutation, which is thought to alter MRX ability to interact with Sae2, increase MRX persistence at DNA breaks and prevent DNA damage checkpoint switch off in the presence of a single irreparable DSB (Lisby et al., 2004 ; Clerici et al., 2006 ). Furthermore, SAE2 overexpression impairs both MRX foci formation and checkpoint activation in response to unrepaired DSBs (Clerici et al., 2006 ). When G1-arrested cell cultures in raffinose containing medium were released from α-factor in the presence of galactose, Rad53 phosphorylation (Figure 1C), Mre11 association to TEL (Figure 1, D and E) and cell cycle arrest (Figure 1A) persisted longer in both GAL-TEL1 sae2Δ and GAL-TEL1 rad50s cells than in GAL-TEL1 cells. On the contrary, when G1-arrested GAL-TEL1 GAL-SAE2 cell cultures in raffinose-containing medium were released from G1 arrest in the presence of galactose, both TEL association of Mre11 (Figure 3, A and B) and Rad53 phosphorylation (Figure 3C, bottom) were strongly reduced compared with GAL-TEL1 cells under the same conditions. Thus, the lack of Sae2 and the rad50s mutation prevent both checkpoint switch off and Mre11 dissociation from telomeres in galactose-induced GAL-TEL1 cells. On the contrary, Sae2 high levels impair both Tel1-induced checkpoint activation and Mre11 telomere association under the same conditions. These data strongly suggest that checkpoint activation under these conditions depends on MRX recruitment at telomeres.
To further support this hypothesis, we asked whether MRX telomere association and checkpoint activation may be limited by high levels of Rif2, a negative regulator of telomere elongation (Hardy et al., 1992 ; Wotton and Shore, 1997 ) that we identified as multicopy suppressor of GAL-TEL1 checkpoint response (Viscardi et al., 2003 ). Indeed, Mre11 telomere association (Figure 3, D and E), Rad53 phosphorylation and telomere elongation (Figure 3F) were severely impaired in galactose-induced GAL-TEL1 cells carrying the RIF2 gene on a 2μ plasmid compared with otherwise isogenic cells containing the empty vector.
Consistent with defective MRX association at telomeres due to Sae2 excess and with impaired MRX functions caused by the sae2Δ or rad50s mutations (Clerici et al., 2005 , 2006 ), telomere elongation was not detectable in galactose-induced GAL-TEL1 sae2Δ, GAL-TEL1 rad50s, and GAL-TEL1 GAL-SAE2 cells (Figures 1B and and3C).3C). On the other hand, GAL-TEL1 sae2Δ and GAL-TEL1 rad50s cells exhibited persistent DNA damage checkpoint activation (Figure 1, A and C). This finding suggests that MRX recruitment at telomeres may signal checkpoint activation independently of telomere elongation. We then asked whether inhibition of telomere elongation without altering MRX recruitment at telomeres could still lead to checkpoint activation after GAL-TEL1 induction. We could not verify this possibility by deleting the telomerase encoding genes in order to inhibit telomere elongation, because extensive telomere shortening and degradation in estΔ cells (Lendvay et al., 1996 ) caused MRX dissociation from telomeres (data not shown).
We therefore used a system based on the observation that defective resection of an intrachromosomal DSB caused by CDK1 inactivation does not impair MRX association at DSB ends (Ira et al., 2004 ). Because generation of 3′-ended single-stranded G telomeric tails, which is necessary to elongate telomeres, is dependent on the passage of replication forks (Dionne and Wellinger, 1998 ; Vodenicharov and Wellinger, 2006 ), inhibition of replication origin firing may impair telomere elongation without altering MRX recruitment at telomeres. We blocked initiation of DNA replication by preventing prereplicative complex (pre-RC) assembly through depletion of the Cdc6 protein (Piatti et al., 1995 ; Cocker et al., 1996 ) in a GAL-TEL1 strain where the only functional Cdc6 was expressed from the methionine-repressible MET3 promoter. A cdc15 mutation in the same strain allowed to uniformly arrest cells in late anaphase in the absence of galactose and methionine. Cells were then released into cell cycle at permissive temperature in the presence of methionine and galactose to fully repress MET-CDC6 and induce GAL-TEL1, respectively. Under these circumstances, cells exited from their late anaphase arrest in the absence of de novo Cdc6 synthesis and, as a consequence, both MET-CDC6 cdc15 and MET-CDC6 GAL-TEL1 cdc15 cells failed to replicate DNA (Figure 4A). Moreover, the appearance of cells with less than 1C DNA content ~210–240 min after the shift to 25°C indicated that they eventually underwent a reductional anaphase in the absence of DNA replication, as previously described (Piatti et al., 1995 ; Figure 4A). Conversely, both cdc15 and GAL-TEL1 cdc15 cells expressing CDC6 from its own promoter entered S phase after the shift to 25°C in the presence of methionine and galactose. GAL-TEL1 cdc15 cells then eventually arrested with 2C DNA content because of the activation of the DNA damage checkpoint (Figure 4A). As shown in Figure 4B, galactose-induced GAL-TEL1 cdc15 cells that could enter S phase underwent telomere elongation. On the contrary, preventing Cdc6 synthesis and replication origin firing in similarly treated MET-CDC6 GAL-TEL1 cdc15 cells did not allow these cells to elongate telomeres (Figure 4B). Strikingly, inhibition of telomere elongation did not prevent either checkpoint activation or MRX association to telomeres. In fact, Rad53 phosphorylation (Figure 4C) and Mre11 association to TEL (Figure 4, D and E) were induced in both GAL-TEL1 cdc15 and MET-CDC6 GAL-TEL1 cdc15 cells after methionine and galactose addition. Conversely, Rad53 phosphorylation was below the detection level in similarly treated cdc15 and MET-CDC6 cdc15 cells (Figure 4C), which did not show any increase in MRX telomere association (Figure 4, D and E) and did not elongate their already full-length telomeres (Figure 4B and data not shown). Altogether, these data further support the notion that DNA damage checkpoint activation induced by ectopic Tel1 expression does not require telomere elongation, but correlates with increased MRX association at short telomeres.
Because MRX association to intrachromosomal DSBs is responsible for the subsequent checkpoint activation (Ira et al., 2004 ; Nakada et al., 2004 ; Clerici et al., 2006 ), we asked whether MRX association to a single telomere could trigger checkpoint activation also in the presence of physiological Tel1 levels. To address this issue, we took advantage of a system developed by S. Marcand (CNRS) that allows to reduce the size of a single telomere without affecting the length of the other telomeres in the cell and to follow its subsequent elongation until returning to equilibrium (Marcand et al., 1999 ). In this system (Figure 5A), generation of one short telomere can be achieved using the yeast strain Lev187, where the left part of chromosome VII beyond the ADH4 gene was replaced by a short stretch of TG1–3 repeats and a URA3 gene, which is in turn flanked by two recognition target sites (FRT) for the recombinase Flp1. The latter can be expressed from the inducible GAL1 promoter. Because the probability of telomere elongation increases dramatically as distal TG-tract length decreases below 200 bp (Teixeira et al., 2004 ; Negrini et al., 2007 ), we used also strains Lev189 and Lev220, which differ from each other and from Lev187 for the length of the distal TG1–3 telomeric repeats, whose mean size is 265 bp in Lev187, 140 bp in Lev189, and 90 bp in Lev220 (Figure 5A; Marcand et al., 1999 ). This difference is due to the fact that the length regulation mechanism takes into account also the additional 270 bp blocks of TG1–3 telomeric repeats, which are present in single (TEL270) or double copy (2 TEL270) between the URA3 gene and the telomere-proximal Flp1 site in strains Lev189 and Lev220, respectively (Marcand et al., 1997 ; Figure 5A).
As depicted in Figure 5A, galactose addition leads to the deletion of the internal telomeric tract containing URA3 by an in vivo recombination event triggered by GAL1-FLP1 induction, leaving in the three strains a short telomere, which can be detected 5 h after galactose addition by Southern blot analysis of StuI-digested DNA using an ADH4 probe. The circular DNA molecule that is concomitantly generated cannot be sensed by the checkpoint machinery. When Lev187, Lev189, and Lev220 cells, exponentially growing in raffinose, were shifted to galactose, Flp1 induction led to an artificially shortened telomere of different length in the three strains (Figure 5B). In agreement with the observation that the probability of telomere elongation increases dramatically as TG-tract length decreases below 200 bp (Teixera et al., 2004 ; Negrini et al., 2007 ), we found that the Flp1-induced Lev187 telomere, whose length is essentially equivalent to that of a native telomere, did not show a significant increase in length even 33 h after Flp1 induction (Figure 5B). Conversely, generation of 140- and 90-bp tracts of telomeric TG-repeats in Lev189 and Lev220 cells, respectively, resulted in their subsequent elongation, although to different extents, until they reached a new equilibrium length (Figure 5B). Strikingly, the increase in size of the shortened telomeres correlated with DNA damage checkpoint activation. In fact, Rad53 phosphorylated forms were below the detection level in all uninduced cell cultures and in galactose-induced Lev187 cells, while they became detectable 5 h after galactose addition in both Lev189 and Lev220 cells, and gradually disappeared concomitantly with the decline of telomere elongation rate (Figure 5, B and C). Furthermore, the amount and persistence of phosphorylated Rad53 were sufficient to cause transient accumulation of large budded cells with a single nucleus in Lev220 cells (Figure 5D). The lower amount and persistence of phosphorylated Rad53 in Lev189 cells did not seem sufficient to significantly affect cell cycle progression.
Rad53 phosphorylation in galactose-induced Lev220 cells requires both Mec1 and Tel1. In fact, it was below the detection level in galactose-induced Lev220 mec1Δ cells, and its amount was reduced in Lev220 tel1Δ cells compared with wild type under the same conditions (Figure 5E).
Similar to what was observed for checkpoint activation in GAL-TEL1 cells, both telomere elongation and checkpoint activation during the return to equilibrium of the Flp1-shortened telomere depend on MRX and correlate with MRX telomere association. In fact, when Lev220 rad50Δ cells exponentially growing in raffinose were shifted to galactose, neither the length of the Flp1-shortened telomere increased, nor Rad53 was phosphorylated (Figure 6, A and B). Moreover, ChIP analysis showed that the association of Mre11 to a unique TEL fragment 80 bp away from telomere VII-L was strongly induced (35-fold) in Lev220 cells 5 h after galactose addition and gradually decreased as telomere elongation rate declined (Figure 6, A, C, and D).
Both Mre11 association to TEL VII-L (Figure 6E) and Rad53 phosphorylation (Figure 6F) were prevented when Flp1 was induced in Lev220 cells that were kept arrested in the G2 phase by nocodazole treatment during galactose addition. Similarly to what was previously observed (Marcand et al., 2000 ) and consistent with defective MRX-telomere association, the Flp1-induced shortened telomere did not elongate in these cells (Figure 6G). This result contrasts with the observation that an HO-induced short telomere can be elongated to some extent in nocodazole-arrested cells (Diede and Gottschling, 2001 ). We believe this discrepancy might be due to the different systems used to generate the short telomere. In agreement with previous findings indicating that MRX binds to telomeres only during S phase (Takata et al., 2005 ), our data suggest that only S phase short telomeres are suitable to recognition by MRX and checkpoint signaling in the presence of physiological Tel1 levels.
Telomeres differ from DSBs because they do not induce a DNA damage response. However, proteins required for the latter are intimately involved in the regulation of telomere function, suggesting a time window where functional telomeres can be recognized as DNA breaks. By using two different systems, here we provide evidence that telomeres elicit an MRX-dependent DNA damage response when they become competent for elongation. Specifically, ectopic Tel1 induction in cells with otherwise stable short telomeres results in telomere elongation and checkpoint activation. Similarly, conversion into a full-length telomere of a single Flp1-induced telomere with short TG tracts parallels DNA damage checkpoint activation. In both cases, checkpoint activation correlates with telomere elongation, strongly suggesting that telomeres are dealt with in a similar manner as DSBs during the time of their replication.
We also show that the MRX complex, which is required in both the systems above for checkpoint activation and telomere elongation, is recruited to short telomeric ends concomitantly with their elongation and checkpoint activation. Furthermore, both the lack of Sae2 and the rad50s mutation increase MRX persistence at short telomeres and prevent checkpoint switch off after Tel1 induction. On the contrary, Sae2 high levels reduce MRX telomere association and impair checkpoint activation in GAL-TEL1 cells. This strongly suggests that checkpoint activation during telomere elongation can be ascribed to MRX recruitment at telomeres.
Interestingly, the MRX complex is not required to activate the checkpoint when telomeres undergo extensive shortening due to the lack of telomerase or when they are exposed to exonuclease degradation leading to ssDNA accumulation due to the lack of Cdc13 (Ijpma and Greider, 2003 ; Foster et al., 2006 ). This indicates that elongating telomeres generate checkpoint signals that are different from those of uncapped telomeres. Consistent with this hypothesis, we found that accumulation of ssDNA at telomeres is not likely responsible for the MRX-dependent checkpoint activation at elongating telomeres.
The specific roles of Mec1 and Tel1 in activating the MRX-dependent checkpoint during elongation of multiple short telomeres is difficult to assess, because high Tel1 levels can bypass Mec1 requirement (Clerici et al., 2001 ). On the other hand, we found that both Mec1 and Tel1 contribute to checkpoint activation in response to elongation of a single short telomere. This situation is reminiscent of the checkpoint activated by a single DSB, where MRX mediates the recruitment of Tel1 at the DSB ends (Nakada et al., 2003 ) and is necessary to activate the Mec1-dependent pathway, possibly by allowing generation of RPA-coated ssDNA (Zou and Elledge, 2003 ; Nakada et al., 2004 ; Mantiero et al., 2007 ). Once MRX is recruited at telomeres, checkpoint activation does not require telomere elongation. In fact, the sae2Δ and rad50s alleles, which cause MRX persistence at short telomeres and do not allow DNA damage checkpoint switch off in galactose-induced GAL-TEL1 cells, also prevent telomere elongation in the same cells, possibly by altering MRX nuclease activity (Clerici et al., 2005 , 2006 ). Furthermore, we found that inhibition of replication origin firing impairs telomere elongation in galactose-induced GAL-TEL1 cells without affecting either MRX association at short telomeres or checkpoint activation. These data indicate that telomere-bound MRX is sufficient to activate the checkpoint independently of telomere elongation, suggesting that MRX binding at short telomeres is the signaling event for checkpoint activation. They also imply that only telomeres that become susceptible to be bound by MRX and, therefore, suitable for elongation, can be recognized as DSBs by the checkpoint machinery. Indeed, MRX was shown to be enriched at telomeres during S phase (Zhu et al., 2000 ; Takata et al., 2005 ), and only telomeres with short TG tracts are avidly bound by MRX, as well as by the telomerase enzyme (Negrini et al., 2007 ; this study).
In this context, our results indicate that, under unperturbed conditions, only S phase telomeres are potentially detectable as DSBs by the checkpoint machinery (Figure 7). However, the yeast telomerase enzyme only acts on short telomeres within one cell cycle (Teixera et al., 2004 ), and the rate of telomere elongation appears limited to a few base pairs per generation (Marcand et al., 1999 ). This limitation may prevent unscheduled checkpoint activation during an unperturbed S phase, because the checkpoint signals do not persist long enough to be detected by the checkpoint machinery.
MRX access, telomere elongation, and checkpoint activation are instead prevented at long or full-length telomeres, suggesting that the latter have acquired a structure that may physically hide the telomeric ends from MRX and telomerase. This protective cap can be formed immediately after DNA replication and maintained until the next S phase. Although the precise molecular nature of this capped structure is unknown, we have found that high levels of Rif2 inhibit both telomere elongation and checkpoint activation in galactose-induced GAL-TEL1 cells. Because it has been shown that Rap1 is necessary to suppress both Cdc13 and MRX recruitment at long TG tracts (Negrini et al., 2007 ), this suggests that Rap1, Rif1, and Rif2 might favor the formation of a t-loop structure similar to the one observed in many other organisms (de Lange, 2005 ).
In conclusion, we have shown that telomeres behave similarly to intrachromosomal DSBs when they are suitable for elongation. This biological response appears to be conserved throughout evolution, because functional human telomeres have been shown to undergo a structural change that elicits a DNA damage response during or after DNA replication (Verdun et al., 2005 ; Verdun and Karlseder, 2006 ).
We thank J. Diffley, S. Marcand, and C. Santocanale for providing antibodies and yeast strains; S. Piatti for providing yeast strains and for critical reading of the manuscript; and all the members of our laboratory for useful discussions and criticisms. This work was supported by grants from Associazione Italiana per la Ricerca sul Cancro to M.P.L. and Cofinanziamento 2005 MIUR/Università di Milano-Bicocca to G.L. V.V. and H.C. were supported by the fellowship Ninni Gattuso from Fondazione Italiana per la Ricerca sul Cancro and by an EC Research Training Network Grant (HPRN-CT-2002-00238), respectively. M.P.L. dedicates this work to Giorgio Terragni, who was born on 1 November 2006.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E07-03-0285) on May 30, 2007.