Using a model system, we have shown that replicative senescence is accompanied by a 16-fold increase in base substitution and frameshift mutations near a chromosome end. The increase was dependent on error-prone polymerases required for the mutagenic response to DNA lesions that block the replication fork.
SACCHAROMYCES cerevisiae cells lacking telomerase, a ribonucleoprotein complex required for telomere replication, experience progressive telomere degradation that culminates in replicative senescence (Lendvay et al. 1996). Deletions that encompass the CAN1 locus located ∼32 kb from the telomere on the left arm of chromosome V accumulated during senescence (Hackett et al. 2001; Hackett and Greider 2003) and have been attributed to replication fork stalling (Motegi et al. 2006). Since replication fork stalling also generates mutations (Quah et al. 1980), we investigated the effect of telomere dysfunction on the generation of mutations at the CAN1 locus.
Mutation rate analysis:
We examined the behavior of mutants defective for EST2, which encodes the catalytic subunit of telomerase (Counter et al. 1997)(Table 1). We determined the mutation rate at the CAN1 locus using an assay that selects against deletions in serial cultures of wild-type and est2 mutant cells. We observed no significant change from the wild-type rate of CAN1 mutation in est2 cultures before senescence (P = 0.07), a 16-fold increase during senescence (P < 0.0001), and the restoration of wild-type levels upon recovery (P = 0.09) (Figure 1A). No senescence-dependent changes in the mutation rate at the CYH2 locus, located 310 kb from its telomere on the left arm of chromosome VII, were observed in est2 mutant cultures before (P = 0.68), during (P = 1.0), or after replicative senescence (P = 1.0) (Figure 2), suggesting that the mutagenic effect is restricted to telomere-proximal sequences. The can1 mutation spectrum observed for senescent est2 mutant cells was similar to that of wild type (Table 2), suggesting that the mechanism of senescence-dependent mutagenesis in est2 cells may be similar to the mechanism of spontaneous mutagenesis in wild-type cells. Since 50–70% of spontaneous mutagenesis has been attributed to the action of error-prone polymerases (Quah et al. 1980), we investigated whether they were involved in the mechanism underlying senescence-dependent mutagenesis. We examined the effects of mutations in the RAD30, REV7, and REV1 genes, which are required for error-prone polymerase function in yeast (Haracska et al. 2000; Johnson et al. 2000; Goodman 2002; Prakash et al. 2005). Mutation rates in est2 rad30 cultures before (P = 0.32), during (P = 0.34), and after senescence (P = 0.15) were not significantly different from those in est2 cultures (Figure 1, A and B), suggesting that Pol η does not contribute to senescence-dependent mutagenesis. In contrast, the rev1 and rev7 mutations completely suppressed senescence-dependent mutagenesis, as the CAN1 mutation rates were not significantly different from those of rev1 and rev7 mutants before (P = 0.46, rev1; P = 0.53, rev7), during (P = 0.84, rev1; P = 0.09, rev7), or after senescence (P = 1.0, rev1; P = 0.2, rev7) (Figure 1, A and B). Southern blot analysis of representative canavanine-resistant mutants collected from senescent cells revealed that all had unrearranged can1 loci, consistent with base substitution and frameshift mutations (D. Meyer and A. Bailis, unpublished data). These data suggest that Rev1 and DNA polymerase ζ are required for senescence-dependent mutagenesis, perhaps through mutagenic bypass of DNA replication lesions generated during replicative senescence. The minimal effects of the rad30, rev1, and rev7 mutations on the growth, senescence, and recovery of est2 rad30, est2 rev1, and est2 rev7 cells (Figure 3, A and B) suggest that the mutation of telomere-proximal sequences does not contribute to the initiation of or recovery from senescence.
Exo1 has been suggested to be necessary for destabilizing the CAN1 locus during senescence by promoting exonucleolytic degradation from the telomere (Hackett and Greider 2003). We observed no significant change in the rates of CAN1 mutation in est2 exo1 cultures before (P = 0.89) or after senescence (P = 1.0) from that observed in exo1 cultures, but did see a significant 10-fold increase during senescence (P < 0.0001) (Figure 1, A and B). These data suggest that Exo1-dependent nucleolytic degradation is not required to observe senescence-dependent increases in CAN1 mutation rate.
In addition to mutagenesis, we also examined the rate of gross chromosomal rearrangement (GCR) in est2 mutants grown serially over time. GCR is defined as an event that leads to the simultaneous loss of CAN1 and a URA3 marker inserted at the HXT13 locus that lies between CAN1 and the telomere on the left end of chromosome V (Chen and Kolodner 1999). The GCR rate was only 2-fold over that in wild type before senescence, 383-fold over that in wild type during senescence, and decreased to wild-type levels upon recovery from senescence (Table 3), all consistent with previously published reports (Hackett et al. 2001; Myung et al. 2001; Pennaneach and Kolodner 2004).
Consistent with CAN1 mutagenesis REV1 and REV7 were found to be important in determining the GCR rate during replicative senescence. GCR rates in est2 rev1 and est2 rev7 mutants were only 2- to 4-fold higher during senescence than in rev1 and rev7 mutants (Table 3) that did not undergo senescence (Figure 3A). These results are consistent with Pol ζ and Rev1 being required for senescence-dependent GCR. Strikingly, Rad30 was found to be required to suppress GCR as the rate in est2 rad30 cells was 37-fold over that in wild type before senescence, 2673-fold over that in wild type during senescence, and at wild-type levels upon recovery (Table 3). Interestingly, this increase is nearly completely dependent on Rev7 as the GCR rate in senescent est2 rad30 rev7 cells was <2-fold greater than that in the est2 rev7 mutant (Table 3). Therefore, like senescence-dependent mutagenesis at the CAN1 locus, senescence-dependent GCR requires Rev1 and Pol ζ.
Our data suggest that Rev1- and Pol ζ-dependent mutations and GCR at telomere proximal loci are an important consequence of telomere dysfunction. The involvement of Pol η, Pol ζ, and Rev1 suggests that the postreplication repair machinery (Minesinger and Jinks-Robertson 2005) may be responding to the failure of bidirectional DNA replication in the region (Plosky and Woodgate 2004; Lehmann 2005). Strathern and colleagues have reported the involvement of Pol ζ in generating mutations associated with the recombinational repair of an enzyme-catalyzed double-strand break (Rattray et al. 2002), suggesting that a similar process may be involved in generating mutations at sequences lying near the uncapped telomeres of senescent cells (Dubois et al. 2002). However, such a mechanism is unlikely to involve extensive exonucleolytic degradation from the telomere as loss of Exo1, a factor involved in the degradation of uncapped telomeres (Hackett and Greider 2003), does not reduce senescence-dependent mutation (Figure 1B) or GCR (Table 3). Further, this mechanism would likely require homologous recombination between sister chromatids, as the CAN1 gene is a unique sequence in the genome. Perhaps similar forces underlie some of the increased genome instability and cancer in somatic cells of elderly people (Lengauer et al. 1998; Johnson et al. 1999).
We thank D. Gottschling, J. McDonald, R. Woodgate, and V. Lundblad for supplying strains and plasmids. We also thank J. Termini, T. O'Connor, R.-J. Lin, and the members of the Bailis lab for productive discussions. Our thanks also go to several anonymous reviewers for their valuable comments. This work was supported by funds from the Beckman Research Institute of the City of Hope, the National Institutes of Health (GM057484 to A.B.), and the Department of Defense Breast Cancer Research Program (W81XWH0410407 to D.M.).
Communicating editor: E. Alani
- Received November 11, 2006.
- Accepted November 17, 2006.
- Copyright © 2007 by the Genetics Society of America