Biology of telomeres: lessons from budding yeast

Authors

  • Martin Kupiec

    Corresponding author
    1. Department of Molecular Microbiology and Biotechnology, Tel Aviv University, Ramat Aviv, Israel
    • Correspondence: Martin Kupiec, Department of Molecular Microbiology and Biotechnology, Tel Aviv University, Ramat Aviv 69978, Israel. Tel.: +972 3 640 9031; fax: +972 3 640 9407; e-mail: martin@post.tau.ac.il

    Search for more papers by this author

Abstract

Telomeres are nucleoprotein structures that cap the ends of the linear eukaryotic chromosomes and thereby protect their stability and integrity. Telomeres play central roles in maintaining the genome's integrity, distinguishing between the natural chromosomal ends and unwanted double-stranded breaks. In addition, telomeres are replicated by a special reverse transcriptase called telomerase, in a complex mechanism that is coordinated with the genome's replication. Telomeres also play an important role in tethering the chromosomes to the nuclear envelope, thus helping in positioning the chromosomes within the nucleus. The special chromatin configuration of telomeres affects the expression of nearby genes; nonetheless, telomeres are transcribed, creating noncoding RNA molecules that hybridize to the chromosomal ends and seem to play regulatory roles. The yeast Saccharomyces cerevisiae, with its sophisticated genetics and molecular biology, has provided many fundamental concepts in telomere biology, which were later found to be conserved in all organisms. Here, we present an overview of all the aspects of telomere biology investigated in yeast, which continues to provide new insights into this complex and important subject, which has significant medical implications, especially in the fields of aging and cancer.

Introduction

The genome of most eukaryotic organisms is divided into linear chromosomes. Each chromosomal end is protected by a special nucleoprotein structure called telomere. Telomeres play central roles in maintaining the stability of the genome: they differentiate the natural chromosomal ends, which should not be repaired, from double-stranded DNA breaks (DSBs), which occur often by accident in the cells and need to be repaired urgently to prevent loss of genomic information (Dewar & Lydall, 2012). Protection of the chromosomal ends is conferred by the special folding of telomeres, as well as by specific telomeric proteins. In addition, telomeres provide a solution to the end-replication problem: the regular DNA replication machinery is unable to fully replicate the chromosomal ends (Olovnikov, 1971; Watson, 1972); as a consequence, information is lost with each cell division, eventually resulting in senescence and cell death (Hayflick, 1979; Lundblad & Szostak, 1989; Harley et al., 1990).

Highly proliferative cells, such as mammalian embryonic cells and unicellular organisms, solve this problem by expressing the specialized reverse transcriptase telomerase (Greider & Blackburn, 1987; de Lange, 2009), which is able to extend the telomeres by copying telomeric sequences from an internal RNA template. Indeed, it is enough to express active telomerase to overcome cellular senescence in somatic cells (Bodnar et al., 1998). Cancer cells also require functional telomeres: in about 80% of tumors, the telomerase gene is expressed (DeMasters et al., 1997); in the rest, an alternative mechanism, based on homologous recombination (HR), allows telomere length extension (ALT; reviewed in Conomos et al., 2013). Moreover, experiments have shown that replenishing telomeres is one of the few essential and earliest steps that a normal mammalian fibroblast must take to become cancerous (Hahn et al., 1999). Mutations that affect telomere function result in human diseases, such as dyskeratosis Congenita, idiopathic pulmonary fibrosis, and others (Calado & Young, 2009; Armanios, 2012; Gramatges & Bertuch, 2013). Thus, our understanding of the biology of telomeres has significant medical implications and is especially relevant to the fields of aging and cancer.

Although some differences exist between the organization of telomeres in yeast and mammals, many basic rules are universal. In 2009, Elizabeth Blackburn, Carol Greider, and Jack Szostak received the Nobel Prize in Medicine for their work on telomeres and telomerase. Much of this work was carried out in model organisms, including the yeast Saccharomyces cerevisiae.

Several excellent reviews on various aspects of telomere biology have been published in the last years (Palm & de Lange, 2008; Lydall, 2009; Shore & Bianchi, 2009; Artandi & DePinho, 2010; Giraud-Panis et al., 2010; Dewar & Lydall, 2012; Smekalova et al., 2012; Stewart et al., 2012a,b; Wellinger & Zakian, 2012; Churikov et al., 2013; Conomos et al., 2013; Gramatges & Bertuch, 2013; Lu et al., 2013; Nandakumar & Cech, 2013; Teixeira, 2013). Here, we will concentrate on what we have learnt from the yeast S. cerevisiae, an organism that, by virtue of its fast growth, excellent biochemistry, and superb genetics, has become extensively used for the molecular genetic dissection of many universal cellular processes.

Throughout this review, we will refer to four interconnected aspects of telomere biology: (1) telomere ‘capping’, which prevents the recognition of the natural ends as if they were DSBs that require repair; (2) telomere replication, which is necessary to solve the end-replication problem and must be coordinated with regular DNA polymerases; (3) telomere localization at the periphery of the nucleus and its role in silencing of genes located nearby; finally, (4) telomere length regulation: a typical ‘wild-type’ length is achieved by a complex homeostasis, which is exquisitely regulated. As expected, all these aspects of telomere biology are interconnected, interrelated, and interdependent, and it is not always possible to separate one from another.

Structure and sequence of telomeres

The telomeres of most organisms are composed of simple tandem repeats, and, although telomeres vary in length (from c. 350 bp in yeast to several kb in mammals), their general structure and functions are conserved (Fig. 1). The yeast telomeric sequence is not regular and can be described as T(G1–3) (Shampay et al., 1984; McEachern & Blackburn, 1994). These sequences are copied by the catalytic subunit of telomerase from the telomerase RNA template (TLC1 in yeast), whose template sequence is CACACACCCACACCAC (Lin et al., 2004). Thus, the RNA template is partially used, and only very short stretches are copied in each round of telomerase activity from different regions of the template (Forstemann & Lingner, 2001). Although the heterogeneity of the telomeric sequence in yeast makes it difficult to know the exact sequence present at any chromosomal end, the same heterogeneity has been exploited to monitor recently added telomeric DNA: Individual clones differ in the sequences added to the pre-existing telomere; this has been used to ask what telomeres are chosen for elongation and how this mechanism is controlled (Teixeira et al., 2004; Arneric & Lingner, 2007; Chang et al., 2007).

Figure 1.

Structure of the yeast telomere. (a) Schematic representation of a yeast telomere, showing the X and Y′ sequences and the internal and terminal TG overhangs. (b) ‘Fold-back’ structure of the yeast telomere, with representative proteins. Rap1 binds the telomeric repeats; and Rif1, Rif2, and the SIR proteins bind to Rap1. The Ku heterodimer binds to telomeric dsDNA, and the CST complex binds the terminal ssDNA end. (c) Telomerase is recruited to telomeres present in an ‘extensible’ configuration.

The yeast telomeres, as their mammalian counterparts, are not blunt, but exhibit a 3′ extension of the G-rich strand (also called a ‘G-tail’ or ‘G-overhang’). This tail varies in length during the cell cycle, remaining very short for most of the cycle (about 12 nt; Larrivee et al., 2004), but elongating at late S by a mechanism that involves both elongation of the G-strand by telomerase and degradation of the C-strand and is highly coordinated with genomic DNA replication (Dionne & Wellinger, 1998; Frank et al., 2006).

In addition to the telomeric sequences, yeast chromosomal ends, as most other eukaryotes, have subtelomeric repeats. Two repeat families occur exclusively at subtelomeric regions: the X and Y′ elements. X units are present in almost all telomeres, although they may slightly vary in size and in sequence. In about half of the telomeres, and distally to the X units (i.e. toward the chromosome's end), Y′ elements are present, in 1–4 tandem repeats, bordered by the telomeric repeats on the distal end (Fig. 1a). TG repeats are sometimes found between the X and the Y′ elements, as well as between Y′ and Y′ (when more than one Y′ repeat is present in tandem; Walmsley et al., 1984). These are potential sources of genomic instability, as they can recombine with telomeric sequences (Aksenova et al., 2013; Gazy & Kupiec, 2013). Potential origins of replication (ARSs) are present within these elements (Chan & Tye, 1983). Interestingly, subtelomeric repeats appear to be extremely variable from strain to strain: for example, the identity of the Y′-less chromosomal arms differs among related strains (Horowitz et al., 1984; Zakian et al., 1986); this is correlated with a high frequency of recombination among subtelomeric regions (Louis & Haber, 1990).

Linear chromosomes lacking telomeric TG repeats have extremely low stability and tend to be lost (Szostak & Blackburn, 1982; Shampay et al., 1984; Lundblad & Szostak, 1989). However, normal chromosomal arms naturally exist lacking Y′ elements, and chromosomes engineered to lack both X and Y′ elements are very stable (Sandell & Zakian, 1993). Despite this apparent dispensability, subtelomeric regions, in particular Y′, are used to maintain stable chromosomes in cells defective for telomerase activity (Lundblad & Blackburn, 1993; Maxwell et al., 2004).

Telomeric proteins

A number of proteins bind the subtelomeric and telomeric DNA and serve several roles in end protection, DNA replication, and chromatin establishment and maintenance. Most proteins participate in more than one aspect of telomere biology (Fig. 1b and c).

Rap1

Rap1 is an abundant essential protein (c. 4000 molecules per haploid cell; Buchman et al., 1988) that binds double-stranded telomeric DNA via its two tandem myb domains (Konig et al., 1996). In addition, Rap1 works as a general cellular transcriptional activator that binds to upstream promoter regions at a large number of genes and interacts with various coactivator proteins (Tornow et al., 1993; Lieb et al., 2001; Zhao et al., 2006).

It has been calculated that about 20 Rap1 molecules bind each individual telomere in wt cells (Wright & Zakian, 1995). Rap1 binds telomeres with high affinity (Conrad et al., 1990; Lustig et al., 1990) although in a noncooperative manner (Gilson et al., 1993; Williams et al., 2010) and plays a central role in determining telomere length: indeed, it has been proposed that a ‘counting mechanism’ is able to monitor (and respond to) the number of Rap1 molecules (and its partners Rif1 and Rif2, see below) bound to each individual telomere (Krauskopf & Blackburn, 1996, 1998; Marcand et al., 1997; Levy & Blackburn, 2004; Poschke et al., 2012). Rap1 is an essential protein; however, N-terminal deletions that ablate its BRCT domain (which usually interacts with phosphorylated proteins) are not lethal (Moretti et al., 1994; Graham et al., 1999). The telomeric functions of the protein are concentrated in its C-terminus, which interacts with the Rif1/Rif2 proteins, as well as with the gene silencing Sir3/Sir4 complex (Kyrion et al., 1992; Wotton & Shore, 1997; Graham et al., 1999). Rap1 plays several interrelated roles at the telomere: it prevents telomere–telomere fusions (Pardo & Marcand, 2005; Marcand et al., 2008), determines its localization to the nuclear periphery (Gotta & Gasser, 1996; Laroche et al., 1998), affects silencing (Hardy et al., 1992ab; Kyrion et al., 1993; Palladino et al., 1993), and protects the chromosomal ends (Negrini et al., 2007; Vodenicharov et al., 2010). Moreover, its dual role as a telomere component and a general transcription regulator allows Rap1 to serve as an effector of stress-specific expression programs. Rap1 gets relocalized from the telomeres to additional genomic sites upon DNA damage (Tomar et al., 2008), glucose starvation (Buck & Lieb, 2006), and interestingly, senescence initiated by telomere shortening (Platt et al., 2013). Among hundreds of genes affected by these relocatization events, it is possible to find the core histone genes, which are repressed by Rap1 upon senescence (Platt et al., 2013).

Rif1 and Rif2

Using the yeast two-hybrid method, two factors were isolated, Rif1 and Rif2, that bind the C-terminal region of Rap1 (Hardy et al., 1992ab; Wotton & Shore, 1997). Cells defective for each of these factors exhibit long telomeres, indicating that the function of these proteins is to negatively regulate the elongation of telomeres (Hardy et al., 1992ab; Wotton & Shore, 1997). Rif1 and Rif2 binding confers to Rap1-bound telomeric DNA, a higher order structure by interconnecting different Rap1 units. This structure is functionally important, although its function remains enigmatic (Shi et al., 2013). Despite these common structural function, Rif1 and Rif2 seem to play several roles independently of each other. As explained above, rif1 and rif2 mutants show elongated telomeres. The double mutant, however, exhibits much longer and unregulated telomeres, indicating that the two proteins participate in alternative regulatory mechanisms (Wotton & Shore, 1997; Romano et al., 2013).

Although telomeric DNA resembles one half of a broken chromosome, one of the main functions of telomeres is to ‘cap’ this end to prevent its recognition by the DNA repair machinery of the cell. Rif1 and Rif2 play an important role in this process. They both play nonoverlapping roles in masking a DSB flanked by a very short array of telomeric repeats (Ribeyre & Shore, 2012). Surprisingly, however, Rap1 and Rif2, but not Rif1, inhibit the access of nucleases and the nonhomologous end-joining (NHEJ) machinery (Marcand et al., 2008; Bonetti et al.,2010ab; Cornacchia et al., 2012); Rif2 also prevents the association of the Tel1/MRX complex (the yeast version of ATM/MRN) with telomeres (Hirano et al., 2009; Chapman et al., 2013) and plays a role in recruiting the histone deacetylase Rpd3L (Poschke et al., 2012). In contrast, Rif1, but not Rif2, is essential for cell viability when the CST activity fails (CST is a protein complex with a role in telomere capping, see below; Addinall et al., 2011; Anbalagan et al., 2011; Di Virgilio et al., 2013). Rif1 seems to play an independent role in transducing environmental signals to the telomere-maintaining machinery (Harari et al., 2013; Romano et al., 2013), a role not shared with Rif2. Surprisingly, Rif1 seems also to carry out checkpoint-regulating functions at the telomeres independently of Rap1 (Feldheim et al., 2011; Harari et al., 2011; Xue et al., 2011; Escribano-Diaz et al., 2013; Zimmermann et al., 2013). This finding suggests that Rif1 may be able to bind DNA sequences by itself. Consistently, recent work has uncovered functions carried out by Rif1 that are independent of telomeres, but are related to DNA transactions. Both fission yeast and mammalian Rif1 control the replication-timing program, determining which regions should replicate at any given time (Hardy et al., 1992ab; Cornacchia et al., 2012; Hayano et al., 2012; Yamazaki et al., 2012). The mammalian ortholog of Rif1 has also recently been found to play a central role in determining whether a DSB will be processed by the HR or the NHEJ pathways (Chapman et al., 2013; Di Virgilio et al., 2013; Escribano-Diaz et al., 2013; Zimmermann et al., 2013). The exact mechanism of these Rif1-regulated events is the current subject of much investigation.

Yku70 and Yku80

The conserved Ku complex, composed of two proteins of c. 70 and 85 kDa (Yku70 and Yku80 in yeast), plays central roles in NHEJ in all eukaryotes studied to date (Gilson et al., 1993; Palladino et al., 1993; Hirano & Sugimoto, 2007; Vodenicharov & Wellinger, 2007; Bonetti et al.,2010ab). Yeast cells, however, lack the DNA-PK activity associated with Ku in mammals (Collis et al., 2005). As NHEJ must be avoided at telomeres, it is surprising that Ku is also a natural component of telomeres. However, Ku plays an essential role in telomere maintenance (Porter et al., 1996; reviewed in Bertuch & Lundblad, 2003; Dewar & Lydall, 2012). In yeast, the Ku complex seems to be recruited to the telomeres in a number of ways: (1) via the interactions between Yku80 and Sir4 [Sir4 is a member of the heterochromatin-specific complex silent information regulator (SIR), see below] (Martin et al., 1999; Bonetti et al.,2010ab). (2) In a Sir4-independent fashion, to subtelomeric X sequences (Boulton & Jackson, 1992ab). (3) The Ku complex is associated with telomerase RNA (TLC1) and participates in the import of TLC1 to the nucleus (Rathmell & Chu, 1994) and possibly in the recruitment of telomerase (Taccioli et al., 1994; Gravel et al., 1998; Roy et al., 2004). (4) The Ku proteins play a role in anchoring the telomeres to the perinuclear space (Laroche et al., 1998) by a still mysterious mechanism that involves the small protein modifier SUMO (Marvin et al.,2009ab). (5) Finally, Ku activity has been shown to prevent exonucleolytic activity at broken chromosomes and at telomeres (Bonetti et al.,2010ab; Mimitou & Symington, 2010). Thus, Ku affects almost all aspects of telomere biology. Interestingly, specific mutations have been found, which separate the roles that Ku plays in NHEJ and in telomere biology (Ribes-Zamora et al., 2007; Lopez et al., 2011).

The CST complex

A third conserved complex is composed of the Cdc13, Stn1, and Ten1 proteins. This complex is structurally similar to replication protein A (RPA), which binds ssDNA during cellular DNA replication and DNA repair (Nugent et al., 1996; Grandin et al., 2000, 2001ab; Qi & Zakian, 2000; Pennock et al., 2001; Petreaca et al., 2006; Churikov et al., 2013). Indeed, domains can be swapped between the two complexes, without losing functionality (Gao et al., 2007; Gelinas et al., 2009).

The CST binds single-stranded telomeric repeats through oligosaccharide/oligonucleotide/oligopeptide bind-ing (OB) folds, a common motif in ssDNA and RNA binding proteins (Pennock et al., 2001; Sun et al., 2009, 2011). It has been proposed that the CST outcompetes and replaces RPA at telomeres; however, RPA can also be detected at telomeres and is probably functional during DNA replication (Schramke et al., 2004; Luciano et al., 2012; Grandin & Charbonneau, 2013). Thus, both complexes are able to bind the telomeric repeats, and the division of work between them may be intricately linked to the mechanism of replication at telomeres.

Conversely, despite high affinity of Cdc13 for single-stranded TG repeats (Lin & Zakian, 1996; Nugent et al., 1996), Cdc13 can in principle be recruited to broken chromosomes (DSBs) to promote telomere addition at nontelomeric ssDNA sequences (Mandell et al., 2011). This process is tightly monitored by phosphorylation and de-phosphorylation of Cdc13 at position S306 by the checkpoint kinases and phosphatases (Zhang & Durocher, 2010). The two proteins associated with Cdc13, Stn1, and Ten1, were isolated as genetic and physical interactors of Cdc13 (Grandin et al., 1997, 2001ab). Mutations that inactivate Cdc13, such as the temperature-sensitive cdc13-1 allele, result in telomere uncapping and cell death (Grandin et al., 2001ab; Maringele & Lydall, 2004ab). Complete lack of Cdc13 activity leads to telomeric DNA resection, generating ssDNA that stimulates a checkpoint-mediated cell cycle arrest (Garvik et al., 1995; Maringele & Lydall, 2004ab; Vodenicharov & Wellinger, 2006). At the permissive temperature, many cdc13 strains (such as those carrying the cdc13-1 allele), as well as stn1 and ten1 mutants, elongate their telomeres (Garvik et al., 1995; Lin & Zakian, 1996; Grandin et al., 1997, 2000, 2001ab; Evans & Lundblad, 1999; Meier et al., 2001; Petreaca et al., 2006), indicating that their normal activity prevents telomere elongation. However, the interactions between the three proteins are not completely understood. Stn1 and Ten1 appear to regulate the activity of Cdc13 (Churikov et al., 2013) possibly by modulating the interactions with subunits of polymerase alpha (see below). On the other hand, mutations in STN1 (Grandin et al., 1997) or overexpression of both Stn1 and Ten1 (Petreaca et al., 2006; Sun et al., 2009) suppresses the lethality of cdc13-1 mutants.

The SIR complex

In many organisms, genes located close to telomeres undergo silencing (also called telomere position effect or TPE). This phenomenon is due to the heterochromatic nature of subtelomeric regions, which represses promoter activity independently of the specific promoter sequence. The area silenced varies among strains and chromosomal ends, but can be as long as 10–15 kb from the telomere ends (Pryde & Louis, 1999).

The SIR complex consists of three proteins, Sir2, Sir3, and Sir4. The SIR complex interacts with histones to form the silencing machinery in S. cerevisiae (Rusche et al., 2003; Liou et al., 2005; Cubizolles et al., 2006). The complex is recruited to telomeres by interactions with Rap1. Interestingly, although Rap1 binds to the TG repeats at the telomere ends, bound Rap1 can also be found by chromatin immunoprecipitation (ChIP) at a distance from the telomere end, several kb away (Strahl-Bolsinger et al., 1997; Poschke et al., 2012). The SIR complex is present in this subtelomeric region too, and it has been suggested that the yeast chromosome folds back, allowing contact between terminal and subtelomeric regions (Fourel et al., 1999, 2001; Ferrari et al., 2004; Poschke et al., 2012), in a way that may protect the DNA ends, akin to the mammalian T-loop (Griffith et al., 1999) (Fig. 1b). Transcription factors that bind to the X regions participate in the nucleation of SIR-dependent repression, as well as the Ku complex (Enomoto et al., 1994; Laroche et al., 1998; Fourel et al., 1999, 2001; Mishra & Shore, 1999; Ferrari et al., 2004; Radman-Livaja et al., 2011).

From these nucleation sites, the SIR complex spreads along the chromatin fiber (Hecht et al., 1996; Strahl-Bolsinger et al., 1997). This spreading is dependent on the deacetylation activity of Sir2 (Tanny et al., 1999; Imai et al., 2000; Smith et al., 2000), which, interestingly, is stimulated by its interactions with Sir4 (Ghidelli et al., 2001; Tanny et al., 2004; Hsu et al., 2013) and generates high-affinity nucleosomal binding sites for Sir3. The Sir3–Sir4 dimer constitutes the structural backbone of silent chromatin (reviewed in Moazed et al., 2004). The SIR complex plays a still enigmatic role in tethering telomeres to the nuclear envelope (see below). Among other interactions, Sir4 interacts with the Mps3 nuclear envelope protein (Bupp et al., 2007), and this interaction may contribute to organizing chromosomes within the nucleus.

Telomerase

The genetic screens carried out by Lundblad and Szostak (1989) that identified ‘ever shorter telomere’ (est) mutants defective in components of the telomerase holoenzyme (est1, est2, est3) also found a fourth complementation group that was allelic to CDC13 (Lendvay et al., 1996). However, contrary to the defective capping phenotype of the cdc13-1 allele, which leads to G2/M cell cycle arrest and massive telomeric DNA resection, the est4 allele of CDC13 (re-named cdc13-2; Nugent et al., 1996) showed a senescent behavior, characteristic of cells unable to support telomerase activity. This suggested that Cdc13 could function in both telomere capping and replication. Est1, Est2, Est3, and Tlc1 form the yeast telomerase holoenzyme (Hughes et al., 2000). A fusion between Cdc13 and Est1 was shown to lead to telomere elongation, even in the presence of cdc13-2 mutations or Est1 alleles unable to interact with telomerase (Evans & Lundblad, 1999). Moreover, if Cdc13 was fused directly to the Est2 catalytic subunit, the essential Est1 subunit became dispensable, demonstrating that the Cdc13-Est1 interaction has as its goal the recruitment of telomerase.

Telomere capping

One of the main functions of the telomere is to prevent the cell from repairing its natural chromosomal ends as if they were DSBs. This function, called telomere capping, is extremely important: DSBs are among the most serious types of DNA damage a cell can undergo, and efficient response mechanisms have evolved to cope with the presence of even a single DSB. Below, I summarize the cellular response to a broken chromosome and then compare it to that observed when the telomeric capping function is missing. For a more in-depth comparison, see Dewar and Lydall (2012).

The DNA damage response

A single DSB [created by external insults (radiation, chemical treatment) or internal cellular metabolism (reactive oxygen species, errors during DNA replication)] elicits a robust DNA damage response (DDR, sometimes referred to as the DNA damage checkpoint), which includes cell cycle arrest and attempts to repair the break (Sandell & Zakian, 1993; Aylon & Kupiec, 2003). Depending on the cell cycle phase (Ira et al., 2004; Aylon et al., 2004; Aylon & Kupiec, 2005), the broken arms are then either ligated in a sequence-independent manner (by the nonhomologous end-joining, or NHEJ, mechanism) or processed by nucleases, to generate ssDNA that can then engage in HR with similar sequences at other genomic locations (Aylon et al., 2003).

The decision of whether to repair the break by NHEJ or by HR falls early in the process: once the ends start to be resected, they are committed to repair by HR (Aylon et al., 2003). The first step in the resection process depends on the MRX complex (Mre11–Xrs2–Rad50) and the nuclease Sae2, which together generate a short (50–100 nucleotide) overhang of 3′ ssDNA (Mimitou & Symington, 2008). This first step is followed by a more extensive resection carried out by a combination of the exonuclease Exo1, the helicase Sgs1, and the helicase/nuclease Dna2 (Tsubouchi & Ogawa, 2000; Gravel et al., 2008; Mimitou & Symington, 2008; Zhu et al., 2008; Bonetti et al., 2009). The specific functions of these enzymes and their interactions are still being elucidated. A triple mutant devoid of Sae2, Exo1, and Sgs1 or the double mutants defective for Sae2 and Sgs1 or Sgs1 and Exo1 show no resection, whereas a sae2Δ exo1Δ double mutant still shows resection. This suggested a model in which MRX/Sae2 acts first in combination with Sgs1 and Dna2, and then the Sgs1/Dna2 pair allows Exo1 to extend the resection. Apparently, Exo1 cannot initiate resection by itself; double mutants sae2Δ sgs1Δ are inviable, whereas sae2Δ exo1Δ cells are viable (Tsubouchi & Ogawa, 2000; Gravel et al., 2008; Mimitou & Symington, 2008; Zhu et al., 2008; Bonetti et al., 2009). The activity of Exo1 is inhibited by the Ku complex (Yku70/Yku80): in the absence of Ku, sae2Δ sgs1Δ cells become viable (Bonetti et al.,2010ab; Mimitou & Symington, 2010). Also consistent with this model, the resection reaction can be carried out in vitro by combining MRX with Sgs1 and Dna2 (Cejka et al., 2010; Niu et al., 2010) or MRX/Sae2 and Exo1 (Nicolette et al., 2010; Nimonkar et al., 2011). Chromatin configuration also plays a still enigmatic role in controlling resection: the Sgs1–Dna2-dependent machinery requires a nucleosome-free gap adjacent to the DSB for efficient resection, and Exo1 activity is blocked by regular nucleosomes and may require the incorporation of the H2A.Z for its activity (Adkins et al., 2013).

As a result of the resection activity, ssDNA is created, which gets rapidly covered by the ssDNA binding protein RPA. This leads to the activation of the DDR kinases (Zou & Elledge, 2003) through two main branches: Mec1, the yeast ortholog of ATR, and its partner Ddc2 (ATRIP) are recruited to the ssDNA-RPA (Paciotti et al., 2000; Rouse & Jackson, 2000; Kondo et al., 2001; Zou & Elledge, 2003). In parallel, the 9-1-1 complex (composed in yeast by the Rad17, Ddc1, and Mec3 proteins) is also loaded onto RPA-coated ssDNA by the Rad24 (Rad17 in humans) clamp loader (Kondo et al., 2001; Melo et al., 2001; Majka & Burgers, 2003). Mec1-phosphorylated histone H2A at the site of DNA damage attracts the Rad9 adaptor protein (Downs et al., 2004; Naiki et al., 2004; Toh et al., 2006; Hammet et al., 2007; Usui et al., 2009), which also interacts with methylated H3K79 (Wysocki et al., 2005; Lazzaro et al., 2008). In addition, Dpb11 serves as a bridging partner connecting the Mec1/Ddc2, the 9-1-1, and the Rad9 branches (Mordes et al., 2008; Pfander & Diffley, 2011). Rad9 undergoes phosphorylation by the single yeast cyclin-dependent kinase (CDK1), generating a substrate for Dpb11 binding. By simultaneously binding Mec1 and phosphorylated Rad9, Dpb11 enforces the checkpoint signal transduction and restricts it to the proper cell cycle phase (Pfander & Diffley, 2011).

Once all these players are in place at the DNA damage site, Mec1 phosphorylates and activates the downstream effector kinases Rad53 (Chk2) and Chk1 (Sun et al., 1998; Sanchez et al., 1999; Blankley & Lydall, 2004; Usui et al., 2009). The activity of these kinases prevents cell cycle progression: Rad53 phosphorylates the anaphase-promoting complex (Hu et al., 2001; Agarwal et al., 2003), and Chk1 phosphorylates the securin Pds1 (Sanchez et al., 1999), blocking progress through anaphase. In a parallel branch, Rad53 activates the Dun1 kinase, thus upregulating DNA damage-responsive genes and resulting in an increase in the cellular dNTP levels (Zhou & Elledge, 1993; Gardner et al., 1999; Zhao & Rothstein, 2002; Andreson et al., 2010).

Importantly, a feedback mechanism seems to operate, in which Mec1, Rad9, and Rad53 play roles in inhibiting resection (Lydall & Weinert, 1995; Jia et al., 2004; Lazzaro et al., 2008; Morin et al., 2008; Segurado & Diffley, 2008) thus preventing excessive chromosomal DNA degradation.

In human cells, two major, evolutionarily related checkpoint kinases control the response to DNA damage: ATR and ATM. Whereas ATR responds to problems during DNA replication by activating Chk1, the ATM is elicited by DSBs and activates Chk2 (Cimprich & Cortez, 2008; Shiloh & Ziv, 2013). In S. cerevisiae and Schizosaccharomyces pombe, in contrast, the ATR ortholog (Mec1 and Rad3) activates both Chk1 and Chk2/Rad53, whereas Tel1, the ATM ortholog, plays a role mainly at telomeres, or when the Mec1 pathway is dysfunctional (Sabourin & Zakian, 2008). Tel1 is recruited to DSBs by the Mre11/Rad50/Xrs2 (MRX) complex, where it usually plays a secondary role, helping to activate the Mec1 checkpoint pathway (Mantiero et al., 2007). However, a large number of DSBs is able to elicit a Tel1-dependent, Mec1-independent response (Mantiero et al., 2007). The activity of Sae2, which initiates resection, also stimulates the Mre11 nuclease, which helps to liberate MRX from the DSB, committing the cells to the Mec1-dependent response (Langerak et al., 2011; Limbo et al., 2011). In the absence of resection or Mec1, Tel1 is recruited via interactions with Xrs2 (Usui et al., 2001; Nakada et al., 2003). Tel1 is then able to elicit the checkpoint response similarly to Mec1 (Usui et al., 2001; Mantiero et al., 2007; Limbo et al., 2011).

As we have seen, the broken DNA ends are resected in cells with an active CDK1 to generate RPA-covered ssDNA. With the help of several mediator proteins, the RPA is displaced and replaced by the Rad51 protein, the eukaryotic ortholog of bacterial RecA. This Rad51 nucleofilament is the main intermediate in the HR process: it allows strand exchange and pairing between molecules sharing sequences (reviewed in Krejci et al., 2012). Sequence homology can be usually found in sister chromatids, homologs, or just similar sequences ectopically located. A genome-wide search for homology allows the Rad51 filament to find its recombination partner (Barzel & Kupiec, 2008; Agmon et al., 2013). Repair by HR results in the joining of the broken ends, sometimes incorporating information transferred from the ‘donor sequence’ to the originally broken molecule (called ‘gene conversion’). If the two DNA molecules share a long stretch of sequence identity, gene conversion can be associated with a crossing over event that exchanges information reciprocally between partners (Inbar et al., 2000).

Telomere uncapping

As we have seen, telomeres naturally carry a ssDNA end covered by the RPA-like CST complex (Lin & Zakian, 1996; Grandin et al., 1997, 2001ab; Gao et al., 2007). The cdc13-1 allele is temperature sensitive: at the restrictive temperature, telomeres become uncapped (Garvik et al., 1995), resulting in their recognition by the cell as a DSB. A robust DDR ensues, which includes extensive ssDNA resection and cell cycle arrest (Garvik et al., 1995; Jia et al., 2004). A similar response is observed with other temperature-sensitive alleles of the CST partners (Gao et al., 2007) or even with a strain deleted for the CDC13 gene (Vodenicharov & Wellinger, 2006; see below). The genetic control of resection differs from the one observed at nontelomeric DSBs: the MRX complex, which usually participates in the initiation of resection, inhibits resection at telomeres, and mutations in the MRX genes exhibit increased ssDNA levels (Foster et al., 2006). An elegant labeling experiment demonstrated that MRX binds specifically to leading-strand telomeres, where it could generate ssDNA for the CST to bind (Faure et al., 2010).

When telomeres become uncapped, or in the absence of active Tel1 and Mec1 pathways, the cells also repair some of the exposed telomere DNA by NHEJ, creating telomere–telomere fusions (Mieczkowski et al., 2003; Pardo & Marcand, 2005; Marcand et al., 2008), which contribute to the formation of gross chromosomal rearrangements (Myung et al., 2001). The genetic control of fusion formation is complex, involving several alternative pathways (Mieczkowski et al., 2003; Pardo & Marcand, 2005; Marcand et al., 2008).

Telomerase is also found at the leading-strand telomeres; it could play a protective role preventing resection (Vega et al., 2007). In contrast to the relatively small effect of mutations in EXO1 on the resection of DSBs (Mimitou & Symington, 2008), Exo1 is the main nuclease at uncapped telomeres, and exo1Δ mutants show reduced resection levels (Maringele & Lydall, 2002; Zubko et al., 2004). This effect is probably due to the requirement by Exo1 for overhangs to initiate its activity: these naturally occur at telomeres. Sgs1 is also active in the resection of uncapped cdc13-1 telomeres (Ngo & Lydall, 2010), and Dna2 is likely to be involved too, as its human counterpart participates in telomere processing (Lin et al., 2013). Surprisingly, however, extensive resection is observed upon Cdc13 inactivation in the absence of both Exo1 and Sgs1, suggesting the existence of an alternative nuclease (Ngo & Lydall, 2010). The Pif1 helicase plays a role in controlling access of this nuclease, as its inactivation abolishes all resection in an exo1Δ mutant (Dewar & Lydall, 2010). Another important player at Cdc13-inactivated telomeres is the 9-1-1 complex and its loader, Rad24, which contribute to regulate resection. A similar role has been observed in nontelomeric DSBs (Aylon & Kupiec, 2003).

The Ku complex negatively regulates resection at both DSBs (Mimitou & Symington, 2010) and at telomeres (Maringele & Lydall, 2002; Bonetti et al.,2010ab), where it also plays a capping role. Surprisingly, however, at telomeres lacking Ku, Chk1, rather than Rad53, is in charge of the cell cycle arrest (Teo & Jackson, 2001; Maringele & Lydall, 2002). In addition, in uncapped telomeres that lack Ku70, the 9-1-1 complex seems to play no role in checkpoint activation, which is entirely dependent on Exo1 (Booth et al., 2001; Maringele & Lydall, 2002; Zubko et al., 2004).

Thorough experiments from the Wellinger laboratory have shown (Vodenicharov & Wellinger, 2006, 2007; Vodenicharov et al., 2010) that the Cdc13 complex is only required for capping during late S and G2/M phases, but not in G1. Several explanations are possible for this cell cycle dependency: it may be related to the CDK1 activity, which is required for regulating the resection machinery (Aylon et al., 2004; Ira et al., 2004; Aylon & Kupiec, 2005). Alternatively, this cell cycle dependency may be related to the timing of DNA replication: physical interactions have been observed between CST members and lagging-strand replication components (Nugent et al., 1996; Qi & Zakian, 2000; Grossi et al., 2004). Interestingly, mutations in Ku components promote a change in the pattern of end-processing, allowing MRX-dependent resection in G1 (Clerici et al., 2008; Bonetti et al.,2010ab; Vodenicharov et al., 2010) even in cells having an intact CST.

Similarly to the CST and Ku complexes, the Rap1-Rif1-Rif2 complex plays a role in telomere capping. Rap1 inactivation leads to Exo1-driven resection that, surprisingly, leads to cell cycle arrest at G1 instead of the Mec1-dependent G2 arrest (Vodenicharov et al., 2010). Elimination of the C-terminal region of Rap1 or mutations in Rif2 leads to an MRX-dependent, but Exo-independent accumulation of telomeric ssDNA (Bonetti et al.,2010ab). These results highlight a separation of function in controlling resection, with Ku inhibiting Exo1 and Rap1-Rif2 inhibiting MRX activity (Bonetti et al.,2010ab). Rif1 and Rif2 bind the C-terminus of Rap1, but seem to have opposite effects on telomere capping: defects caused by inactivation of the CST complex or Ku are exacerbated by loss of Rif1 but alleviated by loss of Rif2 (Addinall et al., 2011; Anbalagan et al., 2011). Mutations in Rap1 result in increased levels of both NHEJ and resection at the telomeres, suggesting that Rap1 binding is essential to prevent any telomere processing (Pardo & Marcand, 2005; Marcand et al., 2008). Moreover, it has been shown that Rif2, but not Rif1, prevents the association of MRX/Tel1 to telomeres (Hirano et al., 2009; Bonetti et al.,2010ab).

Finally, it is possible that telomerase may have a capping function, in addition to its DNA synthesis activity. Although no increased end degradation is observed in cells recruiting lower levels of telomerase (e.g. in cells carrying mutations in the TLC1 RNA gene), combining these mutations with cdc13-1 leads to increased temperature sensitivity, suggesting some protective role for telomerase (Vega et al., 2007; Addinall et al., 2011).

Finally, situations exist in which telomere capping is attained by alternative, Cdc13-independent mechanisms. The Tbf1 binding protein, for example, has been shown to cap yeast telomeres carrying human telomeric repeats (Fukunaga et al., 2012; Ribaud et al., 2012; Di Domenico et al., 2013). Strains deleted for the CDC13 gene can be created by inactivating the resection machinery (e.g. Exo1) together with the Rad9 checkpoint or the Pif1 helicase (Zubko & Lydall, 2006; Dewar & Lydall, 2010; Ngo & Lydall, 2010). Similarly, yeast cells without Cdc13 are able to grow if they overexpress a truncated version of Stn1 and Ten1 (Petreaca et al., 2006). The telomeres in these strains are still maintained by the activity of telomerase. In the absence of both telomerase and HR, cells senesce, but can be kept alive by deleting the EXO1 gene. In these strains, large palindromic structures cap the telomeres and preserve viability (Maringele & Lydall, 2004ab).

Comparison to mammals

Similar to the yeast telomeres, mammalian telomeres contain specialized dsDNA and ssDNA binding proteins (‘shelterin’, reviewed in Palm & de Lange, 2008). However, the specific complexes involved are slightly different, as one would expect from cells facing different environments and dissimilar biologic backgrounds. A Rap1 ortholog is present and was originally described as unable of binding DNA, although such ability has been recently shown for the human protein (Arat & Griffith, 2012). Rap1 binds to the TRF2 protein, which, together with TRF1, covers the telomeric dsDNA. The ssDNA overhang is bound by POT1, which is linked to the TRF1-TRF2-RAP1 complex by the TIN2 and TPP1 proteins. In addition, a CST complex is present too; it contains orthologs of the Stn1 and Ten1 yeast proteins, but Cdc13 is replaced by CTC1 (Palm & de Lange, 2008; Giraud-Panis et al., 2010; Sfeir & de Lange, 2012). Although its precise role is still unknown, similar to the yeast CST, it resembles RPA and its removal results in telomere uncapping (Bryan et al., 2013).

Experiments in which shelterin proteins were removed in mammalian cells demonstrated that, like in yeast, capping requires the repression of the ATM and ATR branches of the DDR, and the inhibition of both NHEJ and HR. TRF2 seems to play a more active role than TRF1, which is mainly involved in telomere replication and length regulation (Ohki & Ishikawa, 2004; Sfeir et al., 2009). In a recent tour de force, Sfeir and de Lange (2012) removed shelterin completely in mouse cells and analyzed the mechanisms at action in its total absence. Their results suggest six different pathways that impinge on the end-protection problem: the ATM (yeast Tel1) pathway and classical NHEJ repair are repressed by the activity of TRF2, whereas the ATR (yeast Mec1) pathway and HR are inhibited by ssDNA-bound POT1. As a second line of defense, Ku and 53BP1 (yeast Rad9) prevent alternative NHEJ and hyper-resection (as well as HR; Sfeir & de Lange, 2012). The integration of the CST complex into this framework awaits a better characterization of its roles in DNA replication and end-processing (Stewart et al., 2012ab; Wang et al., 2012). Thus, the basic mechanisms that maintain telomere length and structure seem to be universal, despite clear differences in protein composition and cellular regulation.

Telomere DNA replication

A second important role played by telomeres is to solve the end-replication problem: due to the nature of DNA synthesis, which requires an RNA primer, information is lost at the telomeres with each cell division (Olovnikov, 1971; Watson, 1972). However, it should be noted that the main loss of telomere repeats with each replication cycle is due to the leading-strand replication of the resected telomere end (e.g. Fig. 2d). Telomere repeat addition, carried out by the specialized reverse transcriptase telomerase, brings a solution to the problem. However, telomerase activity must be coordinated with the replication of the rest of the genome. Replication origins located close to telomeres replicate very late in the S phase of the cell cycle (Raghuraman et al., 2001); this phenomenon is independent of the replication origin sequence: any origin located near telomeres is fired late (Ferguson & Fangman, 1992). The mechanism by which proximity to telomeres affects origin firing is still unknown, but it is apparently independent of telomeric silencing, as mutations in the SIR complex do not affect replication timing (Stevenson & Gottschling, 1999). Whether this effect is related to perinuclear tethering has not been investigated; mutations in Ku, which also affect telomere tethering, cause earlier firing of origins close to telomeres (Cosgrove et al., 2002; Lian et al., 2011). It has been suggested that this effect is telomere-length dependent and particularly Rif1 dependent (Lian et al., 2011). Indeed, short telomeres fire earlier in S than those of normal size, implying that telomere length (or the amount of telomeric proteins bound) plays a role in dictating replication timing (Bianchi & Shore, 2007ab).

Figure 2.

Replication of telomeres. (a) In G1 cells, telomeres are unavailable for elongation. Est2, the catalytic subunit of telomerase is present at the telomeres, but inactive. Ku and the CST are present too. (b) End resection is carried out by a combination of nucleases and helicases, controlled by the kinases Tel1 and CDK1. Resection creates ssDNA, which may bind RPA. (c) Telomerase is recruited through interactions between Est1 and the CST. It is still unclear whether this depends, or not, on the arrival of the replication fork, and how resection is terminated. The amount of Cdc13 at the telomeres increases (and possibly that of RPA, not shown). (d) Upon activation, telomerase elongates the TG-rich strand. Polα-primase, recruited by the CST, completes lagging-strand replication. It is not clear whether telomerase and Polα-primase activities are concomitant, or even whether they depend on each other. Okazaki fragments at telomeres are eventually ligated to the Okazaki fragments created by the moving replication fork. Note that the leading-strand synthesis leaves a short, blunt-ended telomere that needs to be resected to allow telomerase activity. Note also that the amount of resection will determine the rate of telomere attrition in the absence of telomerase activity (the longer the resection, the shorter the leading telomere). RPA is assumed to be present at ssDNA between Okazaki fragments.

As expected from the obligatory coordination between telomerase activity and genome-wide replication, mutations in DNA polymerases or replication factors affect telomere length (Carson & Hartwell, 1985; Askree et al., 2004; Grossi et al., 2004; Gatbonton et al., 2006). However, this effect is still telomerase dependent (Adams Martin et al., 2000; Grossi et al., 2004) implying that lack of coordination between replication and telomerase, rather than a direct role replacing telomerase, is responsible for the phenotypes observed. Indeed, the CST components interact physically with subunits of DNA polymerase alpha/primase (Qi & Zakian, 2000; Grossi et al., 2004; Sun et al., 2011). As Cdc13 also interacts with telomerase through its Est1 subunit (Qi & Zakian, 2000), the CST is in an excellent position to serve as coordinator. The G-strand overhang is created postreplicationally (Dionne & Wellinger, 1996, 1998) by degradation of the C-strand. Note that after removal of the primer RNAs, the strand synthesized by the leading DNA polymerase should have a 3′ overhang, whereas the other end should be blunt (Fig. 2d). However, both ends undergo C-strand degradation to generate G-rich protruding overhangs (Dionne & Wellinger, 1996; Wellinger et al., 1996). Thus, importantly, it is the extent of resection of the telomeres, rather than the length of the RNA primer, the main factor determining the Hayflick limit (the number of generations a cell lacking telomerase can undergo before senescence). Importantly, the two types of telomeres are treated differently. For example, the MRX complex can be found at the leading telomere, but not at the lagging one (Faure et al., 2010). As telomerase cannot work on blunt-ended DNA, C-strand degradation is essential for its activity. The presence of MRX at telomeres replicated by the leading-strand polymerase (Faure et al., 2010) suggests that these are preferentially elongated by telomerase, as MRX is required to recruit Tel1 and telomerase. In addition, MRX preferentially binds short telomeres (Negrini et al., 2007; McGee et al., 2010). The conclusion is thus that telomerase should preferentially elongate short leading-strand telomeres. Indeed, short telomeres have been shown to be preferentially elongated, in a Tel1-dependent fashion (see below; Teixeira et al., 2004; Chang et al., 2007).

Cell cycle dependency

As mentioned before, a fusion between Cdc13 and the telomerase subunit Est1 can elongate telomeres, even in the presence of cdc13-2 mutations or Est1 alleles unable to interact with Est2, the catalytic subunit of telomerase (Evans & Lundblad, 1999). In addition, a Cdc13–Est2 fusion allows the cells to replicate telomeres in the absence of the essential Est1 subunit. Thus, the role of the Cdc13–Est1 interaction is to recruit telomerase (Est2).

Using either cells with an inducible DSB (Diede & Gottschling, 1999) or with a critically short telomere (Marcand et al., 2000), it was possible to demonstrate that telomerase is not active in cells at the G1 phase of the cell cycle. ChIP experiments were used to determine the cell cycle dependency for the recruitment of the various telomere replication factors to the telomeres (Fig. 2). Cdc13 was found to be associated with telomeric DNA throughout the cell cycle, but its levels increase in late S phase, together with the appearance of the long G-overhangs (Taggart et al., 2002; Fig. 2a and b). Similarly, Est1 and Est3 are only observed at this cell cycle stage. Surprisingly, however, Est2, which encodes the catalytic subunit of telomerase, was found telomere-associated throughout the cell cycle, although its levels also increase in late S/G2 (Taggart et al., 2002). These results reflect the fact that there are two different pathways of telomerase recruitment (remarkably, both totally dependent on the integrity of the TLC1 RNA): Recruitment of Est2 in G1 requires an interaction between Yku80 and a specific stem-loop structure in TLC1 (Fisher et al., 2004). The recruitment in late S/G2 is Est1 and Cdc13 dependent (Chan et al., 2008; Fig. 2c).

It is still unclear what is the significance of the presence of Est2 at the telomeres in G1, as at this time, telomerase cannot add nucleotides to the unprocessed telomeres. Mutations that eliminate the interaction between Yku80 and TLC1 have only a very modest effect in the recruitment of Est2 (Peterson et al., 2001; Fisher et al., 2004). It is remarkable that Ku plays a dual role as a DNA binding protein that recognizes the dsDNA at the end of chromosomes, but it also functions as a specific RNA binding protein. Recent work has shown that these two activities are mutually exclusive, suggesting a new recruitment model in which Ku is responsible for importing telomerase into the nucleus and retaining it there via its interactions with TLC1 (Pfingsten et al., 2012). Once in the nucleus, Ku may be handed off from TLC1 to the telomeric DNA, where Cdc13 interacts with Est1 to secure telomerase to the telomere. Sir4, which binds Ku, may also play some role in this mechanism, as it also interacts with Cdc13, and both Est1 and Sir4 interact with the nuclear envelope protein Mps3 (Pfingsten et al., 2012), suggesting a role in perinuclear tethering. This model, however, does not relate to the cell cycle phase at which this recruitment may occur. The Cdc13 interaction with Est1 may need activation to promote telomerase recruitment and may also activate the new or extant telomerase at late S (Evans & Lundblad, 2002; Fig. 2c).

Microscopic observation of fluorescently labeled TLC1 in single living cells (Gallardo et al., 2011) showed that telomerase is mobile throughout the cell cycle, but its movement decreases, and its intensity increases, at late S phase. The surge in Est1 and Est3 at this stage of the cell cycle suggests a model in which telomerase is assembled in situ when all the subunits converge on the telomeric DNA. The recruitment of Est3 to telomerase was shown to be Est1 dependent (Tuzon et al., 2011). However, expression of Est1 in G1, which results in Est1 and Est3 recruitment to the extant Est2-TLC1, does not allow telomerase activation at that phase of the cell cycle, indicating that additional conditions must be met (Osterhage et al., 2006). These could be CDK1-related (e.g. phosphorylation of one of the proteins by the activated CDK1 may be a prerequisite), or the telomeric DNA may need to be in a particular molecular configuration (e.g. C-strand resected) for telomerase to act. Interestingly, the Rif1 and Rif2 proteins may also restrict G1 activation (Gallardo et al., 2011).

To summarize, the current model for telomere replication (Fig. 2) assumes that telomere elongation is coordinated with chromosomal replication (Shore & Bianchi, 2009). However, precise details of the timing are lacking. Late in S phase, the C-strand is processed, creating G-overhangs (Dionne & Wellinger, 1996, 1998; Frank et al., 2006; Fig. 2b). The MRX-Tel1 pathway plays a role in this event (Ritchie & Petes, 2000; Faure et al., 2010; Gao et al., 2010), which seems to be controlled/affected also by the Tel1 and CDK1 kinases and the Rap1/Rif proteins (Gardner et al., 1999; Craven & Petes, 2000; Frank et al., 2006). The newly formed G-overhang is covered by CST. This could be promoted by phosphorylation of Cdc13 or other telomeric proteins by the Tel1 (or Mec1) kinase (Tseng et al., 2006, but see Gao et al., 2010). The CST in turn recruits telomerase by interactions between Cdc13 and Est1 (Fig. 2c). Telomerase adds TG repeats to the G-rich strand (Fig. 2d). When does the moving fork interact with telomerase is not clear. Interaction of the CST with the DNA polymerase alpha complex (approaching the telomere with the moving fork) may bring to an end the TG-strand extension and promote lagging-strand synthesis of the CA-rich strand by polymerase delta (Diede & Gottschling, 1999; Shore & Bianchi, 2009). Alternatively, the CST may recruit polymerase alpha/primase independently of the moving fork, and the telomeric Okazaki fragments may ligate to those created by the moving lagging strand (Fig. 2d). A confounding factor is that short telomeres seem to affect the timing of firing of the distal origins of replication, with short telomeres replicating earlier in S phase (Bianchi & Shore, 2007ab).

The role of the RPA complex at telomeres is still controversial: the CST has been defined as a ‘telomere-specific RPA-like complex’. Although RPA has been shown to be present at late S at telomeres (Schramke et al., 2004), its presence could just reflect the presence of the chromosomal lagging-strand replication machinery. However, recent evidence supports a more direct role of RPA in lagging-strand synthesis at telomeres (Luciano et al., 2012). RPA binds to the two daughter telomeres during telomere replication but depends on the MRX complex for its binding to the leading-strand telomere. Moreover, RPA specifically co-precipitates with Ku and seems to associate with Cdc13 and Est1 (Luciano et al., 2012). Another issue that has not been sufficiently explored is the order and timing of chromatin remodeling once the telomere has been replicated: the newly replicated telomere must regain its special telomere chromatin composition and structure.

Telomere localization and transcription

Telomere tethering

Yeast chromosomes in the nucleus are not randomly located: all centromeres are clustered around the spindle pole body (yeast spindle organizer), whereas all telomeres are embedded in the nuclear envelope (Therizols et al., 2010). Telomeres play a central role in determining this nuclear configuration (reviewed in Taddei & Gasser, 2012). Several microscopic techniques have demonstrated that telomeres are found in clusters around the nuclear periphery (Palladino et al., 1993; Gotta & Gasser, 1996). Interestingly, the number of clusters is far lower than the number of telomeres: the 32 telomeres of an haploid yeast cell are usually seen in 3–6 clusters (Gotta et al., 1996). These foci move with a constant random motion that is more constrained than that of a nontelomeric locus (Schober et al., 2008; Therizols et al., 2010).

Many results suggested a correlation between perinuclear position and silencing in yeast. The telomeric clusters at the nuclear envelope are enriched for SIR proteins (Gotta et al., 1996). Mutations in Ku or SIR components partially affect telomere position (Laroche et al., 1998; Hediger et al., 2002). However, only a double mutant sir4Δ yku70Δ shows completely delocalized telomeres, demonstrating that redundant anchoring mechanisms are at play (Hediger et al., 2002). The association of telomeres to the nuclear periphery requires at least two nuclear envelope proteins, Esc1 and Mps3. Esc1 interacts with the C-terminus of Sir4, competing with the Yku80 protein, which binds the same region (Andrulis et al., 2002; Taddei et al., 2004). Mps3 resides in the nuclear periphery and has an N-terminal acidic extension that protrudes toward the nuclear interior and is also capable of binding both Ku and Sir4; deleting this extension prevents telomere tethering, although the cells remain viable (Bupp et al., 2007). Some results suggest that the Ku-dependent pathway tends to dominate during G1, while the Sir4/Esc1-dependent tethering pathway is preponderant during S phase (Hediger et al., 2002). As cells prepare to enter mitosis during G2, the telomeres lose their peripheral localization; perinuclear positioning is re-established in early G1 phase (Smith et al., 2003). Interestingly, it is possible that different telomeres differ in their dependence on these two pathways: during G1, for example, the tethering of the telomere VI-right depends primarily on the Ku pathway and that of the telomere VI-left primarily on the Sir4/Esc1 pathway (Bystricky et al., 2005).

SUMOylation of Yku80 and of Sir4 may play a regulatory role in telomere length maintenance and tethering, although the details are still unclear (Ferreira et al., 2011; Hang et al., 2011). Loss of SUMOylation abolishes tethering without affecting TPE (Ferreira et al., 2011), demonstrating that tethering and TPE are separable.

It is very likely that the Ku pathway for tethering involves additional components. Identifying these proteins is complicated by the fact that Ku plays additional roles in transcriptional silencing (Gravel et al., 1998), telomerase recruitment (Stellwagen et al., 2003), and specification of replication timing (Cosgrove et al., 2002). In addition, Mps3 is also able to interact directly with telomerase via its connections with Est1 (Antoniacci et al., 2007) and is required for the tethering of Yku80 and TLC1 (Schober et al., 2009).

Interestingly, a deletion of YKU80 shows increased recombination between interstitial and telomeric regions, suggesting that Ku80 tethering sequesters this region and prevents recombination (Marvin et al.,2009ab). Tethering of telomeres to the nuclear envelope also reduces the efficiency of the homology search when a DSB is created close to a telomere (Therizols et al., 2006; Agmon et al., 2013).

Establishing a causal relationship between subnuclear organization and transcriptional repression has been difficult, because all the mutants that alter the position of silent domains also affect silencing. In many cases, peripheral localization of DNA within the yeast nucleus has been shown to reinforce transcriptional silencing. For example, artificial localization to the periphery enhances transcriptional repression at a compromised silencer (Andrulis et al., 1998). However, positioning and silencing can be separated: for example, repression can be maintained at an intact silencer that is released from the nuclear periphery (Gartenberg et al., 2004). In a series of clever experiments, Taddei et al. (2004) showed that perinuclear chromatin anchoring can occur prior to or independently of transcriptional repression. Moreover, as explained above, SUMOylation seems to be required for tethering but not for TPE (Ferreira et al., 2011). Silencing and anchoring, however, are carried out by the same set of proteins (Ku and SIR) and reinforce each other: Ku-mediated anchoring of Rap1-bound telomeres allows Sir4 recruitment, which in turn increases the recruitment of Sir2 and Sir3, spreading the silencing along the chromatin. As silent chromatin spreads, the interactions of Sir4 to the nuclear enveloped are reinforced (Taddei et al., 2004).

Telomere transcription: TERRA

In addition to the telomeric DNA repeats and the telomere-associated proteins, it is possible to detect noncoding telomeric repeat-containing RNA (TERRA) at the telomeres (Fig. 3). TERRA is conserved from yeast to humans; it encompasses both telomeric and subtelomeric regions and is transcribed by the RNA polymerase II (Luke et al., 2008; Bell et al., 2010; Schoeftner & Blasco, 2010; Arora et al., 2011; Iglesias et al., 2011). In yeast, telomeric RNA ranges from a few hundred nucleotides to as much as 1200 nt and is rapidly degraded by the Rat1 exonuclease, which is in charge of degrading all mRNAs (Luke et al., 2008). The TERRA transcripts overlap the Y′ elements, which are able to encode a helicase that is expressed in cells exposed to stress or upon loss of telomerase activity (Yamada et al., 1998). Interestingly, a similar protein is expressed from subtelomeric regions in S. pombe (Hansen et al., 2006). In telomerase-negative cells, Y′ transcripts have been shown to be amplified by a retrotransposon-mediated mechanism (Maxwell et al., 2004). However, the possible interactions between Y′ transcripts, Y′-encoded helicases and TERRA remain unexplored.

Figure 3.

Proposed regulation of telomere length by TERRA. TERRA is expressed from a telomere-located promoter. At telomeres of normal length, Rat1 constantly degrades the newly made RNA. At short telomeres, in contrast, TERRA is highly expressed. The mechanism controlling this fact remains enigmatic. It is still unclear, for example, whether transcription requires removal of telomeric proteins (“??”). Recent work has shown that TERRA forms foci together with telomerase and are jointly recruited, by a still unknown mechanism, to the same short telomere that expressed TERRA.

The role of TERRA in telomere biology remains enigmatic. It has been proposed that TERRA levels may regulate telomere length; for example, increased levels of TERRA (e.g. in rat1-1 cells grown at semi-permissive conditions) lead to shorter telomere length (Luke et al., 2008). Similarly, expression of a strong inducible promoter at the subtelomeric region leads to telomere shortening (Sandell et al., 1994). On the other hand, there are reports of telomere defects caused by reduced TERRA levels (Azzalin et al., 2007; Deng et al., 2009). A complicating issue is that TERRA expression, and thus probably telomere regulation, appears to be differently affected by the subtelomeric structure of each individual chromosomal arm. TERRA expression in both X and XY′ types of repeats is repressed by a Rap1-based pathway, but only the first type is also repressed by SIR proteins (Iglesias et al., 2011).

In humans, TERRA is tightly regulated by the nonsense-mediated decay machinery, which degrades mRNA molecules with mutations (Chawla & Azzalin, 2008), and the DNA methyltransferases DNMT3b and DNMT1 methylate TERRA promoters within CpG islands and thus downregulate its expression (Nergadze et al., 2009). TERRA transcription is promoted by the MLL histone methyltransferase (Caslini et al., 2009).

The potential mechanism of action of TERRA is not yet clear: TERRA-like RNA oligonucleotides inhibit telomerase activity in vitro (Schoeftner & Blasco, 2008; Redon et al., 2010), suggesting a direct regulation of telomerase, perhaps through inhibition of the polymerization reaction of telomerase. However, RNase H inhibits TERRA overexpression effect, suggesting that TERRA anneals with the telomeric DNA (Luke et al., 2008). Recent work has shown that in cells lacking telomerase activity, TERRA plays also a role in delaying senescence by promoting alternative lengthening of telomeres (ALT; elongation of telomeres by HR), whereas senescence is accelerated in cells unable to recombine (Balk et al., 2013).

A recent publication by the Chartrand group followed endogenous TERRA expression in single yeast cells using RNA fluorescence in situ hybridization (FISH). They found that TERRA expression is induced by telomere shortening, leading to the accumulation of TERRA molecules into a single perinuclear focus. Simultaneous timelapse imaging of telomerase RNA and TERRA revealed telomerase nucleation on TERRA foci in early S phase. Their results suggest that the TERRA foci may act as scaffolds for the recruitment of telomerase molecules and trigger the formation of telomerase clusters (which they call T-Recs). The TERRA–telomerase cluster is subsequently recruited to the short telomere from which TERRA molecules originate, suggesting that TERRA plays a role in the recruitment of telomerase to short telomeres (Cusanelli et al., 2013).

Regulation of telomere length

Telomere length is remarkably variable between organisms, from a few hundred nucleotides in S. cerevisiae to tens of kb in mice. Despite this variability, cells expressing telomerase keep telomeres at a very constant length. Due to the ease with which it can be genetically manipulated and its rapid growth, yeast has greatly contributed to our understanding of telomere length homeostasis.

Uniform telomere length homeostasis is achieved by a balance between shortening and elongating mechanisms within the cell. Telomere shortening takes place naturally by the ‘end-replication problem’ (the inability to fill-in gaps left at the telomere ends by removal of the RNA primer) and by end resection or partial uncapping episodes. In cells with extremely long telomeres, however, an additional mechanism exists that can shorten telomeres to the wt length within one or just a few generations. This mechanism was termed TRD, or telomeric rapid deletion (Li & Lustig, 1996; Bucholc et al., 2001), and is the result of HR between telomeric repeats, which generates chromosomes with short telomeres and a telomeric DNA circle (Bucholc et al., 2001). Two elongating mechanisms exist for telomeres: HR (also known as ALT) and telomerase-mediated telomere elongation. Telomerase-driven telomere elongation must be regulated to attain the uniform wild-type telomere size.

Early work suggested a model in which telomerase activity is regulated in such a way that it slows down with telomere size (Marcand et al., 1999). Sophisticated experiments carried out by the Lingner laboratory, in which telomere elongation events could be followed in individual cells during a single cell cycle, showed that not all telomeres are elongated in each cell cycle. Rather, telomeres with short TG tracts tend to be preferentially chosen for elongation (Teixeira et al., 2004). However, the extent of elongation is independent of the original TG tract length. Moreover, if a strain with two telomerases that differ in their RNA templates is used, it is possible to show that multiple rounds of association and dissociation can take place on a single telomere on a single cell cycle (Chang et al., 2007). Tel1 (the yeast ATM orthologue, or Mec1/ATM in its absence) plays a pivotal role in increasing telomerase processivity at very short telomeres (Chang et al., 2007), which are preferentially chosen with the help of the kinase(s) (Arneric & Lingner, 2007). In the absence of the Rif proteins, the frequency (but not the extent) of elongation events is increased (Teixeira et al., 2004). Based on these experiments, Lingner et al. proposed that the telomere may exist in either an extendible or a nonextendible state. Three models have been put forward for molecular mechanisms of the extendible/nonextendible states (Teixeira et al., 2004; Shore & Bianchi, 2009):

  1. The resection model: Resection of the C-strand may be more extensive if the TG tracts are short. This would generate a longer G-overhang to which more CST molecules would bind, thus increasing the chances of telomerase recruitment. Support for this model is supplied by experiments in which a single DSB is flanked by either short (80 bp) or a long (250 bp) TG tract: ChIP of Cdc13 showed higher levels in the short construct (Negrini et al., 2007).
  2. The activation model: Telomerase activity may become activated by short TG tracts. Support for this model comes from alleles of CDC13 (a member of the CST complex) that show stable, capped, and short telomeres, indicating that, as in tel1 mutants, telomerase is able to elongate extremely short telomeres (i.e. no Est phenotype is detected), but not those with sizes closer to that of wt cells (Meier et al., 2001). This type of mutants suggests that interactions with Est1 (Evans & Lundblad, 2002; DeZwaan & Freeman, 2009) may be present all the time but activated at the right circumstances.
  3. The recruitment model: According to this third possibility, the TG tract length regulates the association of telomerase with the telomere end. It is not clear what mechanism would preferentially recruit more telomerase to those telomeres having short TG tracts. Direct measurement by ChIP of protein levels at short vs long telomeres (Bianchi & Shore, 2007ab; Sabourin et al., 2007) supports the last model, as Est1, Est2, and Tel1, but not Cdc13, were present at higher levels at short telomeres. The presence of Tel1 at short telomeres also suggests that phosphorylation of some of the proteins involved may help in the recruitment. Cdc13 is phosphorylated by both Tel1 (and Mec1 in its absence; Tseng et al., 2006) and by CDK1 (Li et al., 2009). In both cases, it has been reported that phosphorylation promotes Est1 binding and thus telomerase recruitment. However, genetic analysis of site-specific mutants that abolish phosphorylation argues against this simple model (Gao et al., 2010). It has also been proposed that phosphorylation by Mec1 is intended to prevent telomere addition at spurious DSBs (Zhang & Durocher, 2010).

Tel1 activity thus appears to be critical for elongating short telomeres. In tel1Δ mutants, the telomeres are extremely short, but stable (Lustig & Petes, 1986), yet they do not bind Est1 or Est2 (Goudsouzian et al., 2006). It has been shown that Tel1 recruitment to telomeres takes place by interactions with the C-terminus of Xrs2 (Sabourin et al., 2007). But why would this interaction be favored at telomeres with short TG tracts? Marcand et al. (2008) suggested that the Rif proteins, and particularly Rif2, may inhibit the MRX–Tel1 pathway. In short telomeres, this inhibition would be reduced, allowing Tel1 recruitment. Supporting information to this idea comes from the fact that by ChIP, Rif2 levels, but not Rif1 levels, are lower in strains with short telomeres and that Tel1 no longer binds to a short telomere better than to one of normal size in rif2Δ cells (McGee et al., 2010). Additional support for this model comes from work by Hirano et al. (2009), showing that Rif2, and to a lower extent Rif1, blocks Tel1 (but not MRX) association at longer TG tracts by competing for Xrs2 binding. Telomeres are extremely long in rif1Δ and rif2Δ mutants (Hardy et al., 1992ab; Wotton & Shore, 1997). The double mutant, however, exhibits even longer telomeres (similar in size to those seen in mutants of the RAP1 gene lacking the C-terminal domain that interacts with both Rif proteins; Kyrion et al., 1992; Wotton & Shore, 1997).This is consistent with two separate mechanisms that control the activity of telomerase. It should be noted, however, that the molecular mechanism of action of Rif1 is unknown and that Rap1-dependent, Rif-independent mechanisms of telomere length control have been suggested in the past too (Negrini et al., 2007; Marcand et al., 2008).

Another negative regulator of telomerase is Pif1, a 5′–3′ helicase that seems to work by a mechanism that is independent of Rif1 and Rif2 (Schulz & Zakian, 1994). The helicase motif of Pif1 is required to prevent telomerase activity in vivo and in vitro, suggesting that Pif1 acts catalytically to prevent telomere elongation (Zhou et al., 2000; Boule & Zakian, 2006). In the absence of Pif1, Est1 levels increase at the telomeres, whereas its overexpression reduces the association of both Est1 and Est2 with the telomeres (Boule & Zakian, 2006). It has been thus proposed that the role of Pif1 is to actively displace telomerase from its substrate. However, the role of Pif1 is more complex: The protein is cell cycle regulated, peaking at late S/G2 and is preferentially recruited to short telomeres by an unknown mechanism. In its absence, Est2 binds equally well to short and normal-sized telomeres (Vega et al., 2007), suggesting that it participates in telomere length homeostasis. In addition, Pif1 prevents the addition of telomeres at chromosomal DSBs (Schulz & Zakian, 1994; Myung et al., 2001) thus avoiding gross chromosomal rearrangements and seems to be preferentially located at regions of the genome with G-quadruplex-forming potential, where it may facilitate DNA replication (Paeschke et al., 2013). Thus, Pif1 may have roles in general DNA replication (e.g. helping replicate secondary structures), as well as direct involvement in telomere length regulation.

Life in the absence of telomerase

Yeast strains that lack telomerase activity exhibit an ‘ever shortening telomere’ (Est) phenotype: with each generation, more and more cells stop dividing and senesce, until no further growth is observed (Lundblad & Szostak, 1989). Senescence can occur in cells in which the average telomere length is close to that of the wild type, indicating that a massive shortening of telomeres is not a prerequisite for senescence (Lundblad & Blackburn, 1993). Indeed, it is enough to have a single short telomere that cannot be elongated by telomerase to elicit the growth arrest, even in telomerase-proficient cells (Abdallah et al., 2009; Khadaroo et al., 2009).

From a population of senescing cells, rare survivors arise by a recombination-based mechanism, also known as ALT. Survivors are usually of two types: Type I, which amplify internal Y′ repeats, and Type II, which have enlarged TG repeats. Whereas Type I survivors are more common, they grow slowly and in liquid cultures are usually overgrown by Type II cells (Teng & Zakian, 1999; Grandin & Charbonneau, 2007). The central recombination protein Rad52 is essential for both classes, and usually no survivors are observed in its absence (Lundblad & Blackburn, 1993). Another essential protein is Pol32, a subunit of DNA polymerase Delta also involved in break-induced replication, a process in which a broken chromosomal end invades a different chromosome and copies its content until its end (Lydeard et al., 2007). This dependency suggests that both types of telomere maintenance pathways occur by recombination-dependent DNA replication. Interestingly, a Rad52- and Pol32-independent mechanism can be found at low frequency in survivors with extremely long telomeres (Grandin & Charbonneau, 2009; Lebel et al., 2009). The mechanism at action in these cells is still enigmatic.

Type I survivors grow slowly and maintain short telomeres with normal G-overhangs. The cells show massive amplification of their Y′ sequences in a tandem repeat array. Extra-chromosomal Y′ arrays can also be detected; these have been proposed to be intermediates in the process of recombination leading to Y′ amplification (Larrivee & Wellinger, 2006). In addition to Rad52 and Pol32, the appearance of Type I survivors requires the Rad51, Rad54, Rad55, and Rad57 recombination proteins (Le et al., 1999; Chen et al., 2001).

Type II survivors show an increase in the telomeric TG repeats; the length distribution is extremely variable, with some telomeres being very short and others extending for more than 10 kb. In these cells, TG circles are detected, presumably created by a rolling-circle mechanism followed by recombination of circles into the genome (Larrivee & Wellinger, 2006). Long telomeres in Type II survivors progressively shorten with cell growth and require constant recombination events to maintain cell growth. The genetic requirements of these cells differ from those of Type I survivors: they require the MRX complex, the Rad52 paralog Rad59, and the yeast ortholog of the WRN and BLM helicases, Sgs1 (Diede & Gottschling, 1999; Chen et al., 2001; Huang et al., 2001; Johnson et al., 2001). The extrachromosomal telomeric circles observed in Type II survivors could be created by the same mechanism that generates TRD (Li & Lustig, 1996; Natarajan & McEachern, 2002).

The choice of recombination mechanism to generate survivors is affected by the DDR: the Mec1 and 9-1-1 pathways control Type II survivor formation, by a mechanism that is independent of the Rad53 and Chk1 kinases (Grandin & Charbonneau, 2007). Interestingly, cdc13-1 cells lacking the Ku complex survive by a Type II mechanism independent of the MRX and Rad59, but dependent on Rad51 (Grandin & Charbonneau, 2003).

Systems biology of telomere biology

In the last decade, a revolution has taken place in biology, which turned the traditional reductionist approach of molecular biology upside down, to attempt a whole-encompassing, gestalt view of the cell. A flurry of genome-wide approaches were launched, in which a systematic approach was taken to try to map all the genes (genomics), RNA molecules (transcriptomics), proteins (proteomics), and metabolites (metabolomics) in a given organism. This ‘omics’ approach was driven by the sophisticated genetics of yeast, which allowed the construction of mutant collections, fusion protein collections, and many other tools. The systems biology revolution is still ongoing, and an effort is being made to map, for example, all the genetic and physical interactions in the yeast cells.

Telomere biology is benefitting from this approach, which greatly enlarged our knowledge about the regulation of telomere-related processes. The yeast genome has close to 6000 recognized genes. A collection of 4700 mutants was constructed by systematically deleting each individual nonessential gene in yeast (nonessential yeast mutant collection; Winzeler et al., 1999). This collection was later complemented by two additional libraries of mutants of all the essential genes (yeast has c. 1300 essential genes), in which either hypomorphic (Breslow et al., 2008) or temperature-sensitive alleles (Ben-Aroya et al., 2008) of the genes were created. These mutant collections allow researchers to carry out systematic mutant screens even if the phenotype of interest is not selectable. For example, three publications reported the systematic screening of the mutant collections, looking for those mutants that affect telomere length (telomere length maintenance or tlm mutants). In this brute-force approach, DNA was extracted from each individual yeast strain, and telomere length was measured by Southern blot (Askree et al., 2004; Gatbonton et al., 2006; Ungar et al., 2009). Together, these papers identified c. 400 genes affecting telomere length. This list starkly contrasts with the 30 or so genes known to do so at the time the screens were carried out (Askree et al., 2004) and stresses the central role played by telomere biology in the yeast life cycle, as c. 7% of the genome affects telomere biology. Moreover, it also demonstrate the complexity of the challenge: mutation in any of the TLM genes changes the final telomere size; as this size is determined by mechanisms that elongate or shorten telomeres (mechanisms that are positively and negatively regulated), this means that each of the 400 genes participates in determining the equilibrium between the two types of activity. Remarkably, however, in each genetic background, wt cells exhibit always telomeres of the same size; thus, in the tug-of-war between elongating and shortening mechanisms, the equilibrium is attained at the same telomere length. The genes uncovered in these screens, as expected, include those affecting DNA and chromatin metabolism, but almost all functions in the cell are also represented, including RNA and protein synthesis, traffic and modification, metabolic pathways, mitochondrial functions, etc. The challenge ahead, of course, is to determine how all these genes impinge on the telomere length determination.

The fact that a near-complete list of TLM genes is available opens the door for further exploration of telomere biology. Using computational approaches and the vast amount of information about protein–protein and genetic interactions in yeast, for example, network models of the telomere biology have been established, allowing their study (Rog et al., 2005; Shachar et al., 2008; Yosef et al., 2009).

Secondary screens were also carried out on the tlm mutant collection. In one of these, TLC1 RNA levels were measured in all tlm mutants, and 24 were found to affect telomere length via their effect on TLC1 levels (Mozdy et al., 2008). These results suggest that the level of telomerase RNA may be limiting in telomere length maintenance. A second screen explored the effect of starvation on telomere length (Ungar et al., 2011). Starved cells, or those exposed to the TorC1 inhibitor rapamycin, respond by dramatically shortening their telomeres. By screening the tlm mutants for those that do not respond to the starvation signals, it was found that the Ku heterodimer plays a central role in the starvation response. When cells are starved, Ku protein levels are reduced, affecting telomere length. This finding is particularly interesting in light of studies suggesting that calorie restriction may lengthen life span, whereas telomere attrition leads to cellular senescence (Ungar et al., 2011). In another study that followed the response of yeast telomeres to environmental stimuli, it was found that exposure to ethanol elongates telomeres, whereas caffeine and high temperature reduce telomere length (Romano et al., 2013). Again, a systematic screen of tlm mutants revealed that the Rap1/Rif1 pathway is necessary for the transduction of three different environmental signals. Interestingly, rif2Δ strains and mutants of the MRX/Tel1 pathway exhibited a stronger-than-expected response to ethanol, which causes elongation, indicating that these pathways have a role in preventing length-independent elongation of telomeres. Finally, Jin-Qiu Zhou et al. explored the tlm collection looking for mutants that affect the survival pathways in the absence of telomerase activity. The TLC1 gene was knocked out in 280 tlm mutants, and the patterns of senescence and survival were monitored. New functional roles were found for 10 genes that affect the emerging ratio of Type I vs. Type II survivors and 22 genes that are required for Type II survivor generation. For example, the Pif1 helicase and the INO80 chromatin remodeling complex reduced the frequency of Type I survivors, whereas the kinase, endopeptidase, and other proteins of small size (KEOPS) complex is required for Type II recombination (Hu et al., 2013).

The KEOPS complex is a good example of a group of proteins with a central role in telomere biology identified in several genome-wide screens. The components of the KEOPS complex include Kae1, a putative endopeptidase with an unknown role which is absolutely conserved in Archaea, Bacteria, and Eukarya; Bud32, a serine/threonine kinase; Pcc1, the yeast homolog of the two human cancer testis antigens that are specifically expressed in different tumors but also in normal testes and ovaries (Kisseleva-Romanova et al., 2006); Gon7 (also referred to as Pcc2), a small protein with no known functional domains, found only in fungi; and Cgi121, whose human homolog binds the human Bud32, also known as the p53-related protein kinase (Miyoshi et al., 2003). For example, Downey et al. screened the nonessential mutant collection for mutants that partially suppress the temperature sensitivity of cdc13-1, caused by telomere uncapping. Among other suppressors, they identified KEOPS members and demonstrated that this complex is required for telomere capping and maintenance (Downey et al., 2006).

An additional genome-wide screen looked for mutants affecting the telomere three-dimensional configuration: A construct carrying a TATA-less galactose-inducible upstream activating sequence downstream of the URA3 gene is able to transcribe the URA3 gene only if folded back on itself. This is indeed the case of the construct integrated close to a telomere. A genome-wide screen for mutants affecting telomere fold-back identified 112 genes. Among various biologic processes uncovered, lysine deacetylation was found to be essential for the fold-back, through Rif2-dependent recruitment of the Rpd3L complex to telomeres. Absence of Rpd3 function generates increased susceptibility to nucleolytic degradation and the initiation of premature senescence, suggesting a protective role for Rpd3 deacetylation activity (Poschke et al., 2012).

As explained, the Lydall laboratory conducted a systematic screen for mutants that affect growth of a cdc13-1 allele. This screen identified 369 gene deletions that could be divided into eight different phenotypic classes. The results included many of the checkpoint-affecting genes expected, but also genes in a variety of unexpected categories, such as RNA metabolism, phosphate and iron homeostasis, etc. In addition, the screen identified a number of genes of previously unknown function renamed restriction of telomere capping (RTC) or maintenance of telomere capping (MTC; Downey et al., 2006; Addinall et al., 2008).

This screen was extended, by systematically looking for suppressors or enhancers of the yku70Δ mutation and comparing them to the results obtained for cdc13-1. By developing a sophisticated analysis, named quantitative fitness analysis, a detailed map of the genetic interactions of both capping proteins could be observed. In this analysis, mutations in some genes, such as those the nonsense-mediated decay pathway, were shown to suppress cdc13-1 but to enhance the phenotype of yku70Δ. The response to telomere uncapping was shown to be genetically complex, with many genes involved in a variety of processes affecting the outcome (Addinall et al., 2011).

Another genome-wide screen examined, in a systematic fashion, the kinetics of senescence, by crossing the est1Δ mutation to the nonessential mutant collection. As expected, the vast majority of gene deletions showed no strong effects on entry into/exit from senescence. However, c. 200 gene deletions (among them the well-characterized rad52Δ mutant) accelerated entry into senescence, and such cells often could not recover growth. A smaller number of strains (among them rif1Δ) accelerated both entry into senescence and subsequent recovery (Chang et al., 2011).

A screen for enhancers of the MMS sensitivity of tel1Δ uncovered a small number (13) of genes. These included Yku70, members of the 9-1-1 pathway, the CCR4-NOT deadenylase complex, nuclear pore components, and several histone deacetylases. Most of these mutants caused the MMS sensitivity due to their effects on telomeres (Piening et al., 2013).

The systems biology revolution is at its infancy; we can summarize at this stage that the genome-wide studies have vastly extended our view of telomere biology. The number of cellular processes affecting various aspects of telomere integrity, replication, length regulation, and structure is remarkable. Most of the genes uncovered in these screens are evolutionarily conserved and likely to act similarly in other organisms, including humans. A better understanding of the mechanisms regulating telomere biology will have significant medical implications, especially in the fields of aging and cancer. The yeast S. cerevisiae, as an easily manipulated organism that grows fast and has superb genetics and molecular biology, has contributed tremendously (and continues to do so) to our understanding of the basic mechanisms of the cells, including telomere biology.

Acknowledgements

I would like to thank all members of the Kupiec laboratory for encouragement and support and Tom Petes for comments on the manuscript. Research in the laboratory is supported by grants from the Israel Cancer Research Fund, the US-Israel Bi-national Fund, the Israeli Ministry of Science and Technology, and the Israel Science Foundation. We apologize to all our colleagues whose work was not quoted due to length constraints.