QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by d'Adda di Fagagna, F. Articles by Jackson, S. P. Search for Related Content PubMed PubMed Citation Articles by d'Adda di Fagagna, F. Articles by Jackson, S. P. Related Content DNA Replication, Recombination, and Repair Social Bookmarking                   What's this?

REVIEW

Functional links between telomeres and proteins of the DNA-damage response

¹ IFOM Foundation—The FIRC Institute of Molecular Oncology Foundation, 20139 Milan, Italy; ² The Wellcome Trust and Cancer Research UK Gurdon Institute, and Department of Zoology, University of Cambridge, Cambridge CB2 1QR, UK

	Abstract

Top Abstract Telomere structure and biology The DNA-damage response DNA-damage checkpoint proteins... DNA-repair proteins and... DDR proteins at functional... Future directions Acknowledgments References

 
In response to DNA damage, cells engage a complex set of events that together comprise the DNA-damage response (DDR). These events bring about the repair of the damage and also slow down or halt cell cycle progression until the damage has been removed. In stark contrast, the ends of linear chromosomes, telomeres, are generally not perceived as DNA damage by the cell even though they terminate the DNA double-helix. Nevertheless, it has become clear over the past few years that many proteins involved in the DDR, particularly those involved in responding to DNA double-strand breaks, also play key roles in telomere maintenance. In this review, we discuss the current knowledge of both the telomere and the DDR, and then propose an integrated model for the events associated with the metabolism of DNA ends in these two distinct physiological contexts.

[Keywords: Telomere; DNA-damage response; checkpoint; senescence; DNA repair]

	Telomere structure and biology

Top Abstract Telomere structure and biology The DNA-damage response DNA-damage checkpoint proteins... DNA-repair proteins and... DDR proteins at functional... Future directions Acknowledgments References

 
The ends of linear chromosomes contain long stretches of DNA tandem repeats (TTAGGG in vertebrates) and terminate in a 3' protruding single-stranded DNA overhang. Due to the inability of the standard lagging-strand DNA replication machinery to copy the most distal telomere sequences (i.e., those at the very end of the chromosome) and to the additional exonucleolytic processing needed to generate protruding overhangs at both ends, telomeric DNA progressively decreases in length as cells go through successive division cycles. Hence, in the absence of specialized telomere homeostatic mechanisms this would ultimately lead to the loss of all telomeric sequences and subsequently to the loss of more internal essential genetic information and ensuing cell death. To circumvent this, many cells maintain their telomeres by the action of telomerase, a specialized reverse transcriptase that uses its associated RNA component as a template to elongate the TG-rich telomeric DNA strand. Although in vitro telomerase activity is dependent on the activity of the reverse transcriptase catalytic subunit (Est2p in the budding yeast Saccharomyces cerevisiae; TERT in mammals) and the telomerase RNA template (Tlc1 in S. cerevisiae and hTR in humans), other factors are clearly needed for telomerase action in vivo (see Table 1). For instance, effective telomerase function in S. cerevisiae requires Est1p and Est3p, and the loss of either of these two proteins—like the loss of Tlc1 or Est2p— leads to progressive telomere shortening (for review, see Blackburn 2000). Furthermore, and as explained below, effective telomerase action in vivo also requires several proteins associated with the DDR.

View larger version (38K): [in this window] [in a new window]   Figure 1. Schematic representation of telomere factors in different organisms.

Another feature of telomeres in some eukaryotic cells is that at various cell cycle stages they appear to cluster and position preferentially at the nuclear periphery (Scherthan 2001). Although this is mostly associated with chromosomal separation during meiosis, at least in S. cerevisiae and Plasmodium falciparum it also occurs during interphase of the mitotic cell cycle (Gotta et al. 1996; Figueiredo et al. 2002). In such situations, the telomeres form clusters in perinuclear chromatin domains that constitute areas of transcriptional repression and modulate recombination between internal tracts of yeast telomeric DNA (Stavenhagen and Zakian 1998; Figueiredo et al. 2002). In S. cerevisiae, the telomeres appear to be tethered to such locations in part via their interaction with the DNA repair protein Ku (see below). To date, however, there is no firm evidence for analogous mechanisms operating in other eukaryotes.

	The DNA-damage response

Top Abstract Telomere structure and biology The DNA-damage response DNA-damage checkpoint proteins... DNA-repair proteins and... DDR proteins at functional... Future directions Acknowledgments References

 
The DDR has evolved to optimize cell survival following DNA damage and control the proliferation of a damaged cell. Probably the best characterized—and most highly evolutionarily conserved—features of the DDR are the recruitment of DNA-repair proteins to sites of DNA damage and the "checkpoint" events that slow down or arrest cell-cycle progression, thus delaying key cell-cycle transitions until the damage has been removed (Zhou and Elledge 2000; Khanna and Jackson 2001). Once the DNA damage has been repaired, the blocks to cell-cycle progression are relieved and cell proliferation can resume. In multi-cellular organisms an inability to repair DNA damage and/or prolonged checkpoint activation can also lead to programmed cell death (apoptosis; Rich et al. 2000), or cause the cell to enter into permanent cell cycle arrest—a state known as senescence (Schmitt 2003). Other aspects of the DDR include changes in chromatin structure at sites of DNA damage (Fernandez-Capetillo and Nussenzweig 2004) and the transcriptional induction and posttranslational modification of DNA-repair and checkpoint proteins as well as other proteins that indirectly influence DNA repair, for example by modulating deoxyribonucleotide availability (Zhou and Elledge 2000; Rouse and Jackson 2002a).

Although it is often useful to study specific aspects of the DDR in isolation, recent findings have suggested that these distinctions are somewhat arbitrary. For example, in some situations "DNA-repair" factors are needed to process initial DNA lesions into structures that can trigger checkpoint activation, and "checkpoint proteins" can control the activity of DNA-repair factors and their recruitment to sites of DNA damage (Lydall and Weinert 1995; Rouse and Jackson 2002a). Consequently, it is probably best to consider the DDR as an integrated and highly coordinated set of events. These issues should therefore be borne in mind in the sections below where, for the sake of simplicity, we summarize the key features of DNA repair and DNA-damage checkpoint events separately and then discuss how each set of factors impacts on telomere biology.

DNA-damage checkpoint pathways

To the first approximation, DNA-damage checkpoint events can be likened to a classical intracellular signal-transduction pathway. Thus, the "stimulus" (DNA damage) is detected by a "sensor" (DNA-damage-binding protein) that then triggers the activation of a "transduction" system composed of upstream (proximal) and downstream (distal) protein kinases, together with a series of adaptor proteins (Fig. 2). This kinase cascade amplifies the initial DNA-damage signal and triggers a diverse set of outputs through targeting a range of "effector" proteins. Central to the DDR in all organisms studied are two large and highly conserved protein kinases of the PIKK (phosphatidyl inositol 3-kinase-like kinase) family. In humans these "checkpoint PIKK" proteins are termed ATM and ATR (ATM and RAD3-related), whereas in S. cerevisiae and S. pombe they are known as Tel1p and Mec1p, respectively, and Tel1p and Rad3p, respectively (see Table 2). The available evidence indicates that the two kinases have distinct but partially overlapping functions. Thus, mammalian ATM is involved primarily in sensing and responding to DNA double-strand breaks (DSBs) generated by agents such as ionizing radiation, although in the absence of ATM some of these functions are partly assumed by ATR (Shiloh 2003). By contrast, ATR responds to a wider range of lesions, probably after they have been processed to a common single-stranded DNA intermediate, and is particularly important in responding to DNA damage during S phase (Zou and Elledge 2003). Once activated, the checkpoint PIKK proteins phosphorylate a range of factors including the distal checkpoint kinases CHK1 and CHK2 (Chk1p and Rad53p in S. cerevisiae; Chk1p and Cds1p in S. pombe) that then target various effector proteins involved in modulating DNA repair, transcription, and cell-cycle progression (Bartek and Lukas 2003).

View larger version (40K): [in this window] [in a new window]   Figure 2. Schematic hierarchical representation of the factors activated following the generation of DNA damage.

²Notably, however, the phosphorylation of other checkpoint PIKK targets also requires additional factors, at least some of which may be classified as DNA-damage sensors as they are recruited to sites of DNA damage independently of the PIKK-containing complexes. In humans, these additional factors include the replication factor C (RF-C)-like complex containing hRAD17 in association with the small RF-C subunits, and the proliferating cell nuclear antigen (PCNA)-like hRAD9–hRAD1–hHUS1 (9–1–1) complex (Shiomi et al. 2002; for review, see Karnitz 2004; see Table 2 for yeast orthologs). Although other models for their actions exist, the PCNA- and RF-C-like checkpoint complexes might promote the DDR by enhancing the activity of the checkpoint PIKK proteins and/or by recruiting checkpoint PIKK substrates to the vicinity of DNA damage, thus facilitating their phosphorylation. Finally, efficient checkpoint activation also requires the recently characterized "mediator" proteins, which include mammalian BRCA1, 53BP1, MDC1/NFBD1, and Claspin, together with yeast counterparts such as S. cerevisiae Rad9p (Shiloh 2003; for review, see Stucki and Jackson 2004). One function of these proteins appears to be to facilitate the focal accumulation of checkpoint and DNA-repair factors in damaged regions, thus promoting their phosphorylation and leading to more efficient checkpoint activation and DNA repair (e.g., Gilbert et al. 2001; Goldberg et al. 2003).

DNA-repair pathways

Different DNA-damaging agents tend to yield chemically distinct classes of lesions and, generally speaking, each class of lesions is repaired by one or more distinct DNA-repair pathways (Friedberg et al. 1995). Of particular importance in regard to telomere functions are the two principal pathways of DNA DSB repair: homologous recombination (HR) and nonhomologous end-joining (NHEJ). Both of these systems have been highly conserved throughout eukaryotic evolution but, whereas NHEJ is a major pathway for DNA DSB repair in higher eukaryotes, single-celled organisms such as yeast rely most heavily on HR (Lieber et al. 2003; Sung et al. 2003). HR requires the RAD52 epistasis group of genes and involves the damaged DNA entering into synapsis with an undamaged homologous partner. An early event in HR is the resection of the DNA DSB in the 5'-to-3' direction by a nuclease, whose activity appears to be modulated by the MRN complex. The resulting 3' single-stranded DNA tails are then bound by Rad51p (a process that is facilitated by a range of other HR factors), which catalyzes a strand-exchange reaction with a homologous undamaged DNA molecule. Subsequently, the 3' terminus of the damaged molecule is extended by DNA polymerase, ligation takes place and the DNA crossovers (Holliday junctions) are resolved to yield two intact DNA molecules. By contrast, NHEJ does not require an undamaged partner molecule and essentially any two exposed double-stranded DNA ends can be re-ligated. In all eukaryotic species examined, NHEJ involves the heterodimeric DNA end-binding protein Ku together with DNA ligase IV in association with a regulatory subunit (XRCC4 in mammals). In vertebrates, efficient NHEJ also requires the DNA-dependent protein kinase catalytic subunit (DNA-PKcs; a member of the PIKK family), which is targeted to DNA DSBs by Ku (Smith and Jackson 1999; Downs and Jackson 2004). In many cases, NHEJ also involves additional proteins that help the processing of DNA ends prior to their ligation (Lieber et al. 2003). As discussed further below, proteins associated with certain other DNA-repair pathways have also been implicated in telomeric functions.

	DNA-damage checkpoint proteins and telomeres

Top Abstract Telomere structure and biology The DNA-damage response DNA-damage checkpoint proteins... DNA-repair proteins and... DDR proteins at functional... Future directions Acknowledgments References

 
One of the most compelling indications for a central role of DNA-damage checkpoint factors at telomeres is the observation that inactivation of checkpoint PIKKs leads to major defects in telomere length control and telomeric stability in all organisms examined. For example, inactivation of Tel1p in S. cerevisiae, Rad3p in S. pombe or ATM in human cells causes telomere shortening (Lustig and Petes 1986; Greenwell et al. 1995; Metcalfe et al. 1996; Dahlen et al. 1998; Naito et al. 1998; Matsuura et al. 1999; Hande et al. 2001). In the above yeast mutants, the telomeres initially shorten rapidly but then stabilize at a new, shorter length. By contrast, the compound inactivation of both checkpoint PIKKs in S. cerevisiae or S. pombe leads to a total inability to maintain telomeric tracts by telomerase-dependent mechanisms, thus causing dramatic and progressive chromosome erosion and ensuing loss of proliferative capacity (Naito et al. 1998; Nakamura et al. 1998, 2002; Ritchie et al. 1999). An analogous analysis of mammalian cells defective in both ATM and ATR has not been possible because ATR is essential for cell viability (Brown and Baltimore 2000). Notably, it has recently been observed that Tel1p and Mec1p are alternatively associated with the telomere during the S. cerevisiae cell cycle—Mec1p peaking in S phase and Tel1p in the other phases—and that Mec1p kinase activity governs this association (Takata et al. 2004). Furthermore, in the absence of Tel1p, Mec1p associates with the telomere throughout the cell cycle. Taken together, these observations reveal a crucial role for the yeast checkpoint PIKKs in telomere maintenance, and suggest that these two kinases act in two distinct telomere maintenance pathways that can partially compensate for one another.

Mounting evidence suggests that the checkpoint PIKKs act in analogous ways at the telomere and in the DDR. For example, in all cases examined both roles require the integrity of the PIKK kinase catalytic domain (Greenwell et al. 1995; Mallory and Petes 2000). Furthermore, in line with functional interactions between S. cerevisiae Tel1p and the MRX complex in the DDR (D'Amours and Jackson 2001; Usui et al. 2001), Tel1p and MRX also work in the same pathway of telomere length maintenance (Boulton and Jackson 1998; Nugent et al. 1998; Ritchie and Petes 2000; Gallego and White 2001; Ranganathan et al. 2001). That is, the loss of any one of these proteins causes telomere shortening to a new, stable, length, but no further shortening is observed with compound mutants, at least as detectable with the available techniques. Moreover, as with inactivation of TEL1, disruption of RAD50 in a mec1 mutant background leads to dramatic telomere shortening and ensuing growth arrest (Ritchie and Petes 2000). In S. pombe, inactivation of either RAD3 or RAD26—which encodes the regulatory subunit of Rad3p in the DDR—causes similar telomere shortening (Naito et al. 1998; Nakamura et al. 2002). Furthermore, despite the loss of S. pombe Tel1p or orthologs of the MRX complex having modest, if any, effect on telomere length (Wilson et al. 1999; Hartsuiker et al. 2001; Manolis et al. 2001; Nakamura et al. 2002; Ueno et al. 2003), combining their loss with inactivation of RAD3 leads, in each case, to an inability to maintain telomeres by telomerase-dependent mechanisms (Ritchie et al. 1999; Nakamura et al. 2002; Chahwan et al. 2003). This phenotype is also observed in S. pombe strains deleted for TEL1 and RAD26 (Nakamura et al. 2002). Finally in this regard, RPA—which facilitates the recruitment of mammalian ATR–ATRIP and S. cerevisiae Mec1p–Ldc1p/Ddc2p to single-stranded DNA (Zou and Elledge 2003)—has been implicated in telomere length control (Smith et al. 2000; Mallory et al. 2003) and in controlling the access of Est1p to the telomere (Schramke et al. 2004). Taken together, the available data therefore strongly suggest that triggering checkpoint PIKK activity is necessary for normal telomere homeostasis, and suggest that the mechanism by which this occurs is closely related to the events leading to PIKK activation in the DDR.

Other upstream components of the DDR, particularly potential sensors of DNA lesions, also impinge on telomere length regulation. Perhaps the most compelling evidence for this is the observation that C. elegans strains lacking MRT2—a functional ortholog of human RAD1 that forms part of the 9–1–1 complex—display progressive telomere shortening and loss of germ-line immortality (Ahmed and Hodgkin 2000). However, the deletion of components of the analogous complex in S. cerevisiae causes only mild telomere length changes, and some effects appear to be laboratory or strain specific (Corda et al. 1999; Longhese et al. 2000; Grandin et al. 2001a). There have also been contrasting reports on the potential role of the analogous S. pombe complex in telomere length regulation, although a recent extensive analysis concluded that these factors and Rad17p—a component of the RF-C-like checkpoint complex—do control telomere length and are associated with telomeric DNA in vivo (Nakamura et al. 2002 and references therein). Although the mechanism(s) by which these factors influence telomere length regulation is still unclear, one possibility is that they facilitate the phosphorylation of certain checkpoint PIKK targets involved in telomere maintenance. Alternatively, or in addition, the effects of these factors on telomere length might reflect them altering telomeric chromatin structure (Corda et al. 1999) or the maturation of telomeric lagging-strand DNA replication intermediates. It is noteworthy that S. cerevisiae cells lacking an alternative RF-C-like checkpoint complex containing Elg1p have long telomeres (Kanellis et al. 2003; Smolikov et al. 2004).

Significantly, combining the deletion of TEL1 with deletion of components of the PCNA-like checkpoint complexes in S. cerevisiae, and the PCNA- and RF-C-like checkpoint complexes in S. pombe, do not result in the senescent phenotypes observed with deletion of MEC1 and TEL1 or RAD3 and TEL1, respectively (Nakamura et al. 2002; Mieczkowski et al. 2003). However, telomere-to-telomere fusions do occur with increased frequency in S. cerevisiae ddc1 tel1 and mec3 tel1 mutants, a phenotype that is similar to that of mec1 tel1 and mec1 mre11 mutants (Mieczkowski et al. 2003). Similarly, components of the analogous S. pombe PCNA- and RF-C-like complexes influence telomeres via the RAD3/RAD26 pathway but their loss does not fully recapitulate the phenotypes of RAD3- or RAD26-deficient strains (Nakamura et al. 2002). Some other less well-characterized DDR factors also regulate telomere functions. For example, S. cerevisiae Tel2p works in the same telomere maintenance pathway as Tel1p (Runge and Zakian 1996) and seems to bind to telomeric DNA (Kota and Runge 1998). Although a role of Tel2p in the DDR has not yet been described, its human counterpart controls sensitivity to DNA damaging agents whereas its C. elegans ortholog influences telomere length, the S-phase checkpoint, and controls life span and biological rhythms (Ahmed et al. 2001; Benard et al. 2001; Lim et al. 2001; Jiang et al. 2003).

It is interesting to note that the components of the DDR that tend to have most impact at the telomere are those that function in the upstream parts of the DDR signalling cascade. Thus, while the checkpoint PIKKs and factors involved in their regulation/activation have major roles in telomere homeostasis, proteins that play important but more downstream functions in the DDR—such S. cerevisiae Rad9p, Rad53p, Dun1p and Chk1p—do not. Furthermore, in instances where such downstream factors influence the telomere, this has generally been ascribed to an indirect effect. For example, the impact of RAD53 or DUN1 deletion on telomere length seems to at least in part reflect defective regulation of deoxyribonucleotide levels (Longhese et al. 2000; Mallory et al. 2003). Where analyzed, downstream components of the DDR in mammals, such as p53 and H2AX, have also not been found to have a major impact on telomere length regulation (Chin et al. 1999; Fernandez-Capetillo et al. 2003). Taken together, the available data are therefore consistent with a model in which telomere homeostasis involves (certain) sensor and upstream kinase components of the DDR that influence telomere structure and telomerase action by mechanisms that do not require the actions of more downstream transducers or effectors of the DDR.

Based on the above, it seems probable that checkpoint PIKKs and their regulatory factors respond to a specific DNA structure(s) arising at telomeres. One situation where such structures may occur is during S phase, when telomeres are replicated and their specialized functions might be temporarily disrupted. In this regard, it is noteworthy that Tel1p and the MRX complex function together in responding to DSBs during S phase (D'Amours and Jackson 2001; Grenon et al. 2001; Usui et al. 2001) and that replication of telomeres may transiently produce similar structures. At the telomere, leading- and lagging-strand DNA replication are expected to produce a blunt end and a recessed end with a 3' overhang, respectively (Chakhparonian and Wellinger 2003). Although replication products bearing a 3' overhang might directly serve as a template for telomerase with little or no processing needed, blunt-ended products would presumably require extensive processing to generate the 3' overhang needed for the binding of telomeric single-stranded DNA-binding proteins such as Cdc13p. In line with this, differential processing of the two products has been revealed by studies with S. cerevisiae strains lacking the Rad27p nuclease, which functions in DNA-base excision repair and in the processing of Okazaki-fragment DNA-replication intermediates (Parenteau and Wellinger 2002).

The use of a de novo telomere addition assay employing a telomeric DNA substrate bearing a HO-endonuclease-induced 5' overhang has revealed an involvement of MRX for telomerase action and Cdc13p binding to the de novo substrate (Diede and Gottschling 2001). Based on these results, it was proposed that the MRX complex helps to prepare telomeric DNA for the loading of Cdc13p, which then protects the chromosome from further degradation and recruits telomerase and other DNA replication components to synthesize telomeric DNA. However, in apparent opposition to this model, the association of Cdc13p with natural yeast telomeres was found to occur efficiently in the absence of Tel1p or MRX and moreover, mutations in the exonuclease domain of Mre11p did not affect telomere length (Moreau et al. 1999; Tsukamoto et al. 2001). The recent finding that the MRX complex does play a modest but detectable role in the generation of telomere overhangs outside S phase could reconcile the above observations (Larrivée et al. 2004). In addition, it is also possible that the MRX complex acts in a partially redundant manner with other proteins at the telomere; one such protein might be the conserved exonuclease Exo1p, which in S. cerevisiae regulates single-stranded telomeric DNA degradation in the absence of Ku (Maringele and Lydall 2002). Further evidence that the yeast MRX complex is involved in recruiting telomerase activity to telomeres is provided by the observation that robust telomere lengthening takes place in mec1 mrx and mec1 tel1 mutant cells in situations where telomerase is targeted to telomeres by way of a protein fusion (Tsukamoto et al. 2001). Such a role may also exist in mammals, as NBS1 associates with telomeres during S phase when telomeres are elongated (Zhu et al. 2000), and is required for effective telomere elongation by telomerase (Ranganathan et al. 2001).

Although there are many ways in which the checkpoint PIKKs and associated components may influence telomere homeostasis, these can be reconciled with a model in which such factors regulate telomerase activity or telomerase access to the telomeric template. One possibility, discussed above, is that such factors are needed for the efficient processing of nascent telomeres into structures compatible with telomerase action. In addition, several lines of evidence suggest that they might also influence telomerase activity more directly. For example, ionizing radiation can influence hTERT nuclear localization (Wong et al. 2002), and telomerase activity was found to increase in extracts derived from rodent cells that had been treated with ionizing radiation or ultra-violet light (Hande et al. 1997, 1998). Conversely, DNA-damage-induced phosphorylation of hTERT by c-Abl (a protein implicated in DNA-PK- and ATM-dependent DDR events) has been found to inactivate telomerase activity (Kharbanda et al. 2000). Nevertheless, it seems unlikely that the checkpoint PIKK proteins control telomere length primarily by influencing intrinsic telomerase catalytic activity, since in vitro telomerase activity is largely unaffected by their loss and in S. cerevisiae the targeting of telomerase to telomeres by way of a protein fusion rescues the senescent phenotype of mec1 tel1 mutant cells (Chan et al. 2001). Therefore, it seems most likely that checkpoint PIKK proteins such as Tel1p and Mec1p mainly control telomere length by regulating the access of telomerase to telomeres by targeting additional telomere-bound factors.

Some potential telomeric targets for the checkpoint PIKKs in yeast have arisen through the work of D. Shore and collaborators, who demonstrated that telomere elongation by telomerase is progressively inhibited in cis by telomere-bound Rap1p. In this elegant model of telomere length homeostasis (the so-called Rap1 counting mechanism; Marcand et al. 1997), progressive telomere shortening causes the gradual loss of telomere-bound Rap1p and, therefore, a progressive relief of its inhibitory function on telomerase activity, ultimately resulting in telomerase-mediated telomere elongation. Significantly, this Rap1 counting mechanism does not function in the absence of Tel1p and, furthermore, the deletion of the Rap1p-binding factors, Rif1p and Rif2p, leads to telomerase-dependent telomere elongation in wild-type but not in tel1 mutant cells (Craven and Petes 1999; Ray and Runge 1999). Taken together, these results suggest a model in which Tel1p and the Rap1p/Rif1p/Rif2p complex promote telomere elongation by acting in the same genetic pathway. Notably, the human homolog of Rap1p, hRAP1, does not appear to bind DNA directly but instead acts together with the telomere-specific DNA-binding protein TRF2 to negatively regulate telomere length in a telomerase-dependent fashion (Li and de Lange 2003). The recent discovery of mammalian orthologues of Rif1p (Adams and McLaren 2004) and the surprising finding that human Rif1 plays important roles in the DDR but seemingly not in telomere homeostasis (Silverman et al. 2004) adds further potential layers of complexity to their functions.

	DNA-repair proteins and telomeres

Top Abstract Telomere structure and biology The DNA-damage response DNA-damage checkpoint proteins... DNA-repair proteins and... DDR proteins at functional... Future directions Acknowledgments References

 
One of the first indications for an involvement of DNA-repair factors in normal telomeric functions was the discovery in 1996 that inactivation of either subunit of the NHEJ protein Ku leads to telomere shortening in S. cerevisiae (Boulton and Jackson 1996; Porter et al. 1996). It was subsequently shown that inactivation of Ku also triggers the rapid loss of telomeric repeats from chromosome termini in S. pombe (Baumann and Cech 2000). However, contrary to when telomerase components are deleted, these telomeres stabilize at a new, shorter, length and there is no progressive further telomere attrition leading to loss of cell proliferation (Boulton and Jackson 1998; Nugent et al. 1998; Polotnianka et al. 1998; Baumann and Cech 2000). Notably, whereas telomere shortening is also caused by the loss of S. cerevisiae Mre11p, Rad50p, or Xrs2p (which also function in NHEJ), this is not the case when S. cerevisiae DNA-ligase 4 or Lif1p (the XRCC4 homolog) are inactivated (e.g., Teo and Jackson 1997; Boulton and Jackson 1998; Herrmann et al. 1998; D'Amours and Jackson 2002). Consistent with these findings, the role of Ku at telomeres appears to be distinct from its roles in NHEJ, as Ku mutants have been identified that affect one function but not the other (Driller et al. 2000; Bertuch and Lundblad 2003; Stellwagen et al. 2003; Roy et al. 2004).

Chromatin immunoprecipitation and immunolocalization studies have shown that Ku is physically associated with telomeres in both S. cerevisiae and S. pombe (Gravel et al. 1998; Laroche et al. 1998; Nakamura et al. 2002), although it is not yet clear whether this reflects direct binding of Ku to telomeric DNA or it being tethered by protein–protein interactions, or both. One mechanism by which Ku functions at the telomere has been revealed by work showing that S. cerevisiae Ku regulates telomere length by interacting directly with TLC1 (Peterson et al. 2001; Stellwagen et al. 2003). Indeed, overexpression of a conserved stem loop of TLC1 leads to Ku-dependent telomere shortening, deletion of this stem loop causes telomere shortening, and a YKU80 mutation that renders Ku unable to bind TLC1 results in short telomeres (Stellwagen et al. 2003). Taken together, these data suggest that the binding of Ku to the telomerase RNA and perhaps other telomere-specific proteins plays a key role in ensuring that telomerase is targeted appropriately to chromosomal ends (Fig. 1). Significantly, the deletion of Ku also impairs the synthesis and/or stability of chromosomal termini in S. cerevisiae. Thus, whereas the telomeric 3' overhang is detectable only during S phase in wild-type cells, in Ku mutants these overhangs are observed in all cell-cycle phases (Gravel et al. 1998). This has lead to the suggestion that the lack of Ku leads to a defect in lagging-strand DNA replication of the telomere (Gravel and Wellinger 2002) and a lack of protection towards Exo1p and other exonucleases, resulting in the generation of the observed constitutive overhang (Maringele and Lydall 2002). In S. cerevisiae, Ku is also required for transcriptional silencing at telomeres (Tsukamoto et al. 1997; Boulton and Jackson 1998)—a function that may in part reflect interactions between Ku and SIR proteins (Tsukamoto et al. 1997; Roy et al. 2004)—and for tethering telomeres to the nuclear periphery (Laroche et al. 1998). Such tethering may limit HR between telomeres (Polotnianka et al. 1998) and ensure that telomeres are replicated in late S phase (Cosgrove et al. 2002).

Several lines of evidence indicate that Ku also functions in telomere maintenance in mammals. For example, it has been reported that human Ku interacts with both TRF1 and TRF2 (Hsu et al. 2000; Song et al. 2000; Peterson et al. 2001), suggesting that it may cooperate with these proteins to regulate telomere length and establish telomere end-protection, respectively. In line with this idea, chromatin immunoprecipitation studies have revealed that Ku physically associates with mammalian telomeres in vivo (Hsu et al. 1999; d'Adda di Fagagna et al. 2001). In addition, inactivation of one allele of the gene for Ku80 in human cells results in telomere shortening (Myung et al. 2004). Furthermore, inactivation of both alleles leads to cell death, although it is not clear whether this is due to further telomere shortening or an inability to cope with endogenous DNA damage (Li et al. 2002). Differently, Ku inactivation is not lethal in mice. Although the reason for this difference between humans and mice is not clear, it has been observed that the Ku80 locus expresses a primate-specific alternative form of the protein, known as KARP-1, that is absent in rodents (Myung et al. 1997). In mice, the analysis of the role of Ku at telomeres has generated some contrasting conclusions. One study showed that cells derived from transgenic mice lacking Ku80, and embryonic stem cells lacking Ku70, have shorter telomeres than their controls, while cells lacking Ligase IV or XRCC4 do not display marked telomere length alterations (d'Adda di Fagagna et al. 2001). This report also showed that Ku inactivation causes elevated chromosomal instability, leading to chromosomal fusions that generally lacked detectable telomeric repeats at the fusion sites. By contrast, a report from another group observed that inactivation of Ku80 did not lead to telomere shortening and that the chromosomal fusions retained telomeric DNA at the fusion points (Samper et al. 2000). Furthermore, an additional report from the same group suggested that Ku is a negative regulator of telomere access by telomerase (Espejel et al. 2002a). Since both groups analyzed mice with the same genetic deletion, the differences reported may originate from variations in the experimental procedures of telomere length measurement, or from differences in mouse or cell maintenance. Importantly, both analyses found that Ku inactivation does not lead to the dramatic changes in telomeric overhangs that are observed in yeast.

Perhaps surprisingly, inactivation of Ku in Arabidopsis thaliana was found to lead to telomerase-dependent telomere lengthening and inefficient C-strand maintenance (Bundock et al. 2002). However, the compound inactivation of Ku and telomerase in A. thaliana causes a faster rate of telomere shortening than telomerase inactivation alone (Riha et al. 2002; Riha and Shippen 2003). Significantly, the deletion of MRE11 also caused telomere elongation in A. thaliana (Bundock and Hooykaas 2002). Although these findings were unexpected, it is noteworthy that while telomerase inactivation restricts life span in most organisms, it extends life span in A. thaliana (Riha et al. 2001). A unifying model for the telomeric functions of Ku in different species is further complicated by the observation that inactivation of Ku in chicken DT40 cells does not seem to affect telomere length (Wei et al. 2002).

In the mouse, inactivation of the Ku-associated NHEJ protein, DNA-PKcs, leads to telomere fusions in the absence of detectable telomere shortening, suggesting that it may be involved in telomere capping (Bailey et al. 2001; Gilley et al. 2001; Espejel et al. 2002b). Consistent with this idea, DNA-PKcs is associated with telomeric DNA in human cells (d'Adda di Fagagna et al. 2001) and inhibition of DNA-PKcs catalytic activity by chemical inhibitors results in telomere fusions in human cells (Bailey et al. 2004). Finally, it has been shown that mice lacking DNA-PKcs and Terc display faster rates of telomere loss than mice lacking Terc alone (Espejel et al. 2002b).

Whether proteins associated with HR also have key functions at normal telomeres is still unclear. Thus, while loss of RAD52 or RAD51 does not affect telomere length in S. cerevisiae, rad52 tlc1, rad51 tlc1, or rad52 est1 double mutant cells senesce at a faster rate than tlc1 or est1 single mutants (Lundblad and Blackburn 1993; Le et al. 1999) and Rad54 knock-out mice have recently been shown to bear shorter telomeres than matched controls (Jaco et al. 2003). Furthermore, Rad51 inactivation in chicken DT40 cells has been reported to increase the presence of the telomeric overhangs (Wei et al. 2002). Most recently, it was established that the RAD51-related protein RAD51D colocalizes with telomeres in human cells and that inactivation of this factor leads to cell death, possibly as a consequence of telomere uncapping (Tarsounas et al. 2004). In light of these findings, it is tempting to speculate that RAD51D, possibly in a complex with certain other HR factors, promotes telomere T-loop formation. In addition, and as discussed below, HR factors can play key roles in maintaining telomere length by telomerase-independent mechanisms.

Other DNA-repair proteins have also been implicated in telomere maintenance. For example, the mammalian DNA repair protein poly(ADP-ribose) polymerase (PARP-1)—which functions in DNA base-excision repair and single-strand break repair (D'Amours et al. 1999)— acts at the telomere. Indeed, a study of PARP-1 knock-out mice provided the first demonstration of a protein of the DDR functioning at the telomere in vertebrates (d'Adda di Fagagna et al. 1999). In this report, PARP-1 inactivation was shown to lead to stable, shortened, telomeres and genomic instability in two different mouse genetic backgrounds and in different tissues. Furthermore, the compound inactivation of PARP-1 and p53 lead to very long and heterogeneous telomeres (Tong et al. 2001), perhaps reflecting the ability of both PARP-1 and p53 to suppress HR (Mekeel et al. 1997; Schultz et al. 2003). However, a different group reported that PARP-1 inactivation does not affect telomere length (Samper et al. 2001). The use of two different genetic deletions in two different mouse strains may help to explain these apparently contradictory results. Finally, XPF/XRCC1— which interacts with ERCC1 to form a structure-specific endonuclease involved in nucleotide excision repair (de Laat et al. 1999)—was recently shown to regulate the stability of the telomeric 3' overhang (Zhu et al. 2003).

DDR proteins at dysfunctional telomeres

Most human somatic cells do not express sufficient telomerase to cope with the inability of the DNA replication machinery to fully replicate chromosomal termini. Consequently, telomeres progressively shorten upon repeated cell divisions, ultimately becoming so short that their normal functions are perturbed. It is still unclear what is the minimal length below which a telomere triggers a DDR. Recently, it has been shown that a DDR at critically short telomeres is associated with the absence of TRF2, at least as detected by immunofluoresence experiments (Herbig et al. 2004), suggesting that the recruitment of this protein to a telomere could be the limiting factor. In some cell types telomere dysfunction lead to apoptosis whereas in others, such as human fibroblasts, it triggers a permanent growth arrest called senescence. Recent work has established that telomere-initiated senescence shares many features of a cell-cycle arrest induced by DNA-damaging agents that cause DSBs (d'Adda di Fagagna et al. 2003). These include the activation of upstream checkpoint PIKKs, mediators, and downstream kinases of the DDR, and the appearance of senescence-associated DNA damage foci (SDFs) containing DDR factors, as detected by immunofluorescence; one report, however, concluded that the detectability of such markers is only transient (Bakkenist et al. 2004). The appearance of DDR markers in senescent cells is triggered with the direct contribution of eroded telomeres, as revealed by the specific accumulation of -H2AX and other markers of the DDR at chromosome termini in senescent cells. Significantly, interfering with the actions of checkpoint kinases by siRNA or by dominant-negative constructs leads to a significant proportion of senescent cells resuming cell cycle progression into S phase, indicating that DNA-damage checkpoint activation is causally associated with the senescent state (d'Adda di Fagagna et al. 2003; Herbig et al. 2004).

Similarly, progressive telomere shortening caused by inactivation of telomerase in S. cerevisiae leads to the accumulation of cells that are unable to divide further and which display an activated DDR—as determined by the phosphorylation of Rad53p—and a morphology reminiscent of senescent mammalian cells (Enomoto et al. 2002; IJpma and Greider 2003). Moreover, inactivation of checkpoint factors such as Mec3p, Mec1p, Lcd1p/Ddc2p, or Rad24p allows a portion of such cells to bypass this senescence-like state and continue proliferating. Therefore, as in mammalian cells, severe telomere shortening in yeast leads to the activation of the DDR and concomitant cell-cycle arrest. These findings are consistent with biochemical experiments and micro-array expression analyses, which have shown that yeast cells with critically short telomeres have a global gene expression profile that overlaps with that of cells exposed to DNA-damaging agents (Nautiyal et al. 2002). Furthermore, the observation that mice with shortened telomeres are more sensitive to radiation (Goytisolo et al. 2000; Wong et al. 2000) is consistent with a model in which dysfunctional telomeres are perceived as DSBs and therefore cells bearing them are more sensitive to additional DNA damaging agents generating DSBs. Taken together, these results suggest that eroded telomeres and DNA damage trigger very similar responses and ultimately produce similar outcomes.

Telomere shortening is not the only way the protective function of the telomere can be lost. In mammals, removal of TRF2 from the telomere leads to a DDR that results in cell-cycle arrest or apoptosis, depending on the cell type (van Steensel et al. 1998). Moreover, the DDRs in senescent and TRF2-inhibited cells appear to be strikingly similar (d'Adda di Fagagna et al. 2003; Takai et al. 2003). Taken together, these results suggest that the loss of telomeric DNA is not detrimental per se, but it is the loss of telomere-bound factors that results in telomere deprotection and concomitant activation of the DDR. This idea is further supported by the observation that cells senesce with a shorter mean telomere length if TRF2 is overexpressed; presumably the additional TRF2 helps to stabilize short telomeres (Karlseder et al. 2002). Analogously, inactivation of S. cerevisiae CDC13, STN1, or TEN1—which form a complex that binds to and protects the protruding telomeric 3' overhang—leads to dramatic activation of the DDR (Garvik et al. 1995; Grandin et al. 1997, 2001b; Pennock et al. 2001). In addition, a DDR leading to rapid telomere degradation has been observed in S. pombe lacking Pot1p—a telomeric single-stranded protein similar to those found in ciliated protozoa (Baumann and Cech 2001). Whether human Pot1p has a similar protective role, however, is still unclear (Colgin et al. 2003; Loayza and de Lange 2003).

Perhaps unexpectedly, unregulated telomere lengthening can also induce a DDR, as has been observed in S. cerevisiae cells bearing short telomeres and overexpressing Tel1p (Viscardi et al. 2003), and also can cause genome instability and telomere fusions, as observed in S. pombe cells lacking Taz1p (Ferreira and Cooper 2001). Furthermore, in human cells the overexpression of a human ortholog of yeast Est1p—a factor necessary for telomerase mediated telomere elongation (Snow et al. 2003)—leads to telomere uncapping (Reichenbach et al. 2003). Although the mechanisms that trigger the DDR under these circumstances are still unclear, it is possible that the uncoupling of the synthesis of the two strands, caused by an overactive telomerase, might lead to generation of an excess of single-stranded DNA that triggers a DDR. Overall, these observations reveal that a variety of perturbations of telomere structure can trigger a DDR very similar to that caused by exogenous DNA-damaging agents.

Dysfunctional telomeres are not only substrates for the cell-cycle checkpoint machinery but are also targeted by the DNA-repair apparatus. Indeed, in both mammals and yeast, critically short telomeres are substrates for recombination and are prone to telomere–telomere fusions. This leads to frequent chromosomal circularization in S. pombe cells lacking telomerase (Nakamura et al. 1998), and chromosomal aberrations resulting from chromosome end fusions in human fibroblasts approaching replicative senescence and in late generation telomerase-deficient mice (Blasco 2002). Similarly, uncapped telomeres are substrates for endjoining events that involve well-known NHEJ factors (Baumann and Cech 2001; Ferreira and Cooper 2001; Smogorzewska et al. 2002; Mieczkowski et al. 2003). Notably, however, there are suggestions that differences exist between the mechanism of telomere end fusions and NHEJ of DNA DSBs caused by DNA-damaging agents. For example, although S. cerevisiae Nej1p—an essential NHEJ component—does not affect the stability of telomeres in wild-type cells, it suppresses telomere fusions mediated by NHEJ in yeasts maintaining their telomeres via HR (Liti and Louis 2003).

When telomeres become critically short in the absence of telomerase in S. cerevisiae, rare survivors emerge that maintain their telomeres through RAD52-dependent mechanisms of HR (Lundblad and Blackburn 1993; Le et al. 1999). These survivors employ either RAD50-dependent amplification of TG-repeats (type II recombination) or RAD51-dependent acquisition of subtelomeric elements (and their deletion derivatives) by a large number of telomeres (type I recombination; Lundblad and Blackburn 1993; Teng and Zakian 1999; Teng et al. 2000; Chen et al. 2001; for review, see Lundblad 2002). It is noteworthy that, although such events might occur most commonly on telomeres that either have lost telomerase activity or Ku (McEachern et al. 2000), recombination can also occur on long telomeres that have been uncapped by the loss of Cdc13p, suggesting that these factors protect chromosome ends from such reactions (Booth et al. 2001; Grandin et al. 2001a; DuBois et al. 2002; Tsai et al. 2002; Grandin and Charbonneau 2003). As mentioned previously, the loss of telomerase function in S. pombe leads to chromosomal circularization in surviving cells (Baumann and Cech 2000). However, when Taz1p is also deleted in such backgrounds, the ensuing survivors more frequently use recombinational modes for telomere maintenance (Nakamura et al. 1998). Thus, as in S. cerevisiae, telomere end-protection proteins actively inhibit HR among homologous telomeric sequences in S. pombe.

In mammals, a significant but relatively small portion of tumours (mostly sarcomas), and cell lines transformed by the SV40 virus, show a very heterogeneous telomeric pattern with some very long telomeres (Neumann and Reddel 2002). These cells do not express detectable telomerase and are believed to maintain their telomeres by HR, as demonstrated by their ability to amplify a tagged subtelomeric sequence in trans onto other chromosomal termini (Dunham et al. 2000; Niida et al. 2000; Varley et al. 2002). Significantly, a portion of cells maintaining telomeres by this "ALT" mechanism (for alternative lengthening of telomeres) display evidence of a DDR at some telomeres. In these cells, telomere-specific binding proteins and telomeric DNA—possibly including this in an extra-chromosomal form—colocalize in subnuclear structures known as PML bodies together with proteins usually associated with DNA damage checkpoint signalling and HR such as MRE11, NBS1, RAD50, RAD51, RAD52, RPA, BLM, and WRN (Henson et al. 2002). Although care should be used to interpret these colocalization data, as very long telomeres might render DDR proteins that are normally associated with telomeres more detectable than when telomeres are shorter, it is tempting to speculate that such structures represent sites where telomeres are being maintained by HR-based mechanisms. How cells become able to maintain their telomeres in this manner is still open to conjecture. Cell-fusion experiments suggest that ALT cells generally carry a recessive mutation(s) (Neumann and Reddel 2002). Furthermore, circumstantial evidence suggests that p53 suppresses ALT, as cell lines derived from Li-Fraumeni syndrome patients—which bear inherited p53 mutations—are frequently ALT, as are SV40 transformed cell lines in which p53 activity has been essentially ablated. The observation that p53 negatively regulates HR (Mekeel et al. 1997), possibly by inhibiting RAD51 activity (Linke et al. 2003), lends further support to this idea. Finally, it is possible that changes in telomeric chromatin are associated with the assumption of ALT. For example, a change in telomeric chromatin that made it more open and accessible to HR proteins could render the cell more susceptible to the initiation of ALT. In this regard, it is noteworthy that inactivation of the S. cerevisiae HHO1 gene, which encodes the linker histone Hho1p, makes it more easy for the yeast cell to enter into HR-dependent mechanisms of telomere maintenance (Downs et al. 2003). Perhaps inactivation or deregulation of linker histones, or possibly other chromatin changes, could lead to a similar situation in the mammalian system.

	DDR proteins at functional and dysfunctional telomeres: what's the difference?

Top Abstract Telomere structure and biology The DNA-damage response DNA-damage checkpoint proteins... DNA-repair proteins and... DDR proteins at functional... Future directions Acknowledgments References

 
As discussed above, components of the DDR are necessary both for normal telomere homeostasis and for responding to dysfunctional telomeres. For example, the checkpoint PIKKs are necessary both for telomere homeostasis and to mount a DDR in reaction to the disruption of the telomere protective structure following telomere shortening and/or telomere deprotection. Similarly, Ku can protect normal telomeric ends from resection and ensuing end fusions, and yet the NHEJ apparatus actually mediates chromosomal end-to-end fusions caused by telomere dysfunction. A key challenge for the telomere field is to explain how the DDR apparatus distinguishes between functional and dysfunctional telomeres and produces two very different outcomes.

The most obvious difference between these two situations is the amount and constitution of telomere-associated proteins. It is therefore possible that the protein complexes associated in a sequence-specific manner with telomeric DNA have the ability to limit the DDR. Thus, when too few (or none) of such proteins are at a telomere, the DDR would become unrestrained, leading to chromosomal end fusions, cell-cycle arrest and/or apoptosis (Fig. 3). We envision several, not necessarily mutually-exclusive mechanisms by which telomeric proteins might inhibit a full DDR being elicited from a functional telomere. By direct steric hindrance and/or by facilitating the formation of higher-order telomeric DNA structures, or by confining the telomere to specific subnuclear regions these proteins might physically prevent DDR factors from gaining access to telomeric DNA. Examples of such activities might include single-stranded telomeric DNA-binding proteins with telomere protecting functions, factors promoting T-loop formation, the generally compact and repressive state of telomeric chromatin and the localization to the nuclear periphery of telomeres in some species.

View larger version (26K): [in this window] [in a new window]   Figure 3. Schematic representation of the mechanisms controlling telomere length homeostasis, telomere elongation of a short telomere, and the generation of a DDR at a critically short telomere. At a telomere in equilibrium, telomere factors (TFs) inhibit the activation of upstream DNA damage kinases (PIKKs), preventing them from activating proteins (such as Cdc13p) that can trigger telomere elongation. PIKK activation is also inhibited by proteins (e.g., Cdc13p) that can res