Saccharomyces cerevisiae as a Model to Study Replicative Senescence Triggered by Telomere Shortening

Abstract

In many somatic human tissues, telomeres shorten progressively because of the DNA-end replication problem. Consequently, cells cease to proliferate and are maintained in a metabolically viable state called replicative senescence. These cells are characterized by an activation of DNA damage checkpoints stemming from eroded telomeres, which are bypassed in many cancer cells. Hence, replicative senescence has been considered one of the most potent tumor suppressor pathways. However, the mechanism through which short telomeres trigger this cellular response is far from being understood. When telomerase is removed experimentally in Saccharomyces cerevisiae, telomere shortening also results in a gradual arrest of population growth, suggesting that replicative senescence also occurs in this unicellular eukaryote. In this review, we present the key steps that have contributed to the understanding of the mechanisms underlying the establishment of replicative senescence in budding yeast. As in mammals, signals stemming from short telomeres activate the DNA damage checkpoints, suggesting that the early cellular response to the shortest telomere(s) is conserved in evolution. Yet closer analysis reveals a complex picture in which the apparent single checkpoint response may result from a variety of telomeric alterations expressed in the absence of telomerase. Accordingly, the DNA replication of eroding telomeres appears as a critical challenge for senescing budding yeast cells and the easy manipulation of S. cerevisiae is providing insights into the way short telomeres are integrated into their chromatin and nuclear environments. Finally, the loss of telomerase in budding yeast triggers a more general metabolic alteration that remains largely unexplored. Thus, telomerase-deficient S. cerevisiae cells may have more common points than anticipated with somatic cells, in which telomerase depletion is naturally programed, thus potentially inspiring investigations in mammalian cells.

Keywords: DNA damage checkpoints; DNA replication; Saccharomyces cerevisiae; replicative senescence; telomerase; telomeres.

Figure 1

Telomere structure in S. cerevisiae (A) and H. sapiens (B). (A) S. cerevisiae telomeric dsDNA is coated by Rap1, that tethers Rif1 and Rif2, two important regulators of telomere length homeostasis and telomere protectors. Sir2-4 also bind Rap1, but are not represented. The ssDNA is bound by Cdc13, which interacts with Stn1 and Ten1 forming a complex called CST. It protects the telomeres, regulates the overhang length and telomere length homeostasis. More details can be found in Wellinger and Zakian (2012). (B) In H. sapiens and other mammals, telomeric DNA is coated by a complex dubbed “shelterin” composed of TRF1, TRF2, RAP1, TIN2, TPP1, and POT1. TRF1 and TRF2 share with budding yeast Rap1 the DNA binding domain. Mammalian RAP1 ortholog is not directly bound to telomeric DNA, but is associated through an interaction with TRF2. TIN2, TPP1, and POT1 have no obvious orthologs in S. cerevisiae. CTC1, STN1, and TEN1 compose the CST complex in H. sapiens. Although distantly related, CST also protects the telomeres, regulates telomerase, and is involved in DNA replication, including at non-telomeric sites. More details in Palm and de Lange (2008), Giraud-Panis et al. (2010), Jain and Cooper (2010).

Figure 2

The DNA-end replication problem in eukaryotes and the rate of telomere shortening. (A) In many eukaryotes, including budding yeast and humans, telomeres end with a 3′ single-stranded overhang. (B) Most telomeric DNA is replicated by replication forks that arise unidirectionally from subtelomeric elements (Miller et al., ; Makovets, ; Sfeir et al., 2009). Thus, most TG-rich strands, containing the 3′ end, are replicated by the lagging synthesis machinery (green), and most CA-rich strands, containing the 5′ end, are replicated by the leading synthesis machinery (red). (C) The presence of the overhang implies an asymmetry between the two template strands and is expected to result in an asymmetry in the length of the semi-conservative DNA replication products. On one hand, the lagging strand telomere is expected to conserve the longest strand containing the 3′ end (but this has not been addressed experimentally). The 5′ newly synthesized lagging strand starts with an RNA primer of a few ribonucleotides. The positioning of this last Okazaki fragment and subsequent removal of the RNA primer is expected to determine the length of the overhang of this strand. In mammals, the positioning of the first RNA primer is random and its removal is delayed (Chow et al., 2012). The enzyme(s) involved in the RNA primer maturation are currently unknown. On another hand, the leading strand synthesis (red) presumably stops prematurely due to a lack of template that corresponds to the length of the overhang. Thus, this telomere is shorter than the parental telomere and the lagging telomere. (D) 5′–3′ Resection of the CA-rich strands was shown to occur in both leading and lagging telomeres in mammals (Wu et al., 2012). In S. cerevisiae, this resection is probably limited to the leading telomere, as represented (Faure et al., 2010). (E) C-strand fill-in compensates for this resection. This synthesis likely requires the polymerase alpha/primase with the synthesis of an RNA primer that must be processed, similarly to the lagging telomere. Thus, the generation of the overhangs is different between leading and lagging telomeres and these processes are expected to modulate telomere-shortening rate and subsequently the onset of senescence.

Figure 3

Cellular responses to a DSB in S. cerevisiae. A DSB is rapidly recognized by MRX that recruits Tel1. In G1 phase of the cell cycle, the DSB is repaired mainly by NHEJ (not represented). In S/G2 phases of the cell cycle, the DSB is processed for repair by HR. The 5′–3′ resection is achieved by the concerted action of nucleases and helicases including Sae2, Sgs1, Dna2, and Exo1. The resulting ssDNA is then coated by RPA. In yeast, Rad52 then catalyzes the displacement of RPA by Rad51. The resulting Rad51 filament can then initiate homology search and strand invasion and prime DNA synthesis. Resolution of DNA structure is then required to conclude the gene conversion. The DNA checkpoint pathway starts by Tel1 recruitment, which modifies nearby nucleosomes (not represented) to form a first platform to recruit additional checkpoint components. The RPA-coated ssDNA forms a second platform that recruits Ddc2-Mec1, Dpb11, and the 9-1-1 checkpoint clamp (Ddc1-Rad17-Mec3), assisted by the checkpoint clamp loader (Rad24-Rfc2-5, not represented). Rad9 is then recruited and activated. Collectively, this edifice contributes to the efficient activation of Rad53 and Chk1 that will phosphorylate downstream targets.

Figure 4

Molecular response to a short telomere in the presence of telomerase in S. cerevisiae. The very first steps of DSB processing shown in Figure 3 also operate at short telomeres of wild-type cells, since a similar 5′–3′ resection takes place late S, by the time telomeres are replicated (Figure 2D). Coated with the telomere-specific Cdc13, the resultant overhang structure is required for the recruitment of telomerase to ensure telomere length homeostasis.

Figure 5

Molecular response to a short telomere in the absence of telomerase in S. cerevisiae. In the absence of telomerase, three distinct, but non-exclusive scenarios can be envisaged. (A) In the G1 phase of the cell cycle, short telomeres incur a risk of being recognized by the DSB sensing factors. Although in budding yeast, this may not be sufficient to trigger a G1 checkpoint (Zierhut and Diffley, ; Symington and Gautier, 2011), this probably contributes to cell cycle arrest in G1 of senescing cells in mammals. The main risk in budding yeast cells is the NHEJ pathway and the formation of telomere–telomere fusions. This will pose a problem in the next S-phase, since palindromic telomeric fusions may generate a replication stress and subsequent activation of a replication checkpoint. Then, if fusions resist to replication, in the next mitosis, the segregation of dicentric chromosomes may contribute to a delay of the M phase. Their resolution may generate immediate unviability or, if it occurs within telomeric repeats, it recreates short telomeres with uncertain end structures. (B) Telomeres form natural replication pauses and in case of telomere breaks during replication, the preferential mode of repair is telomerase-dependent re-elongation. In the absence of telomerase, it’s not known whether replication is made more difficult at the shortest telomeres by a specific processing of DNA ends. Nevertheless, the activation of the replication checkpoint in senescent cells may reflect a myriad of different structures occurring at telomeres when replicated in the absence of telomerase. (C) Finally, short telomeres may be processed as DSBs, whereby an extended 5′–3′ resection may expose subtelomeric ssDNA resulting in Mec1 activation as in Figure 4. Note that all of the three scenarios can contribute to the G2/M delay or arrest observed in senescent cells.