An Age-Induced Switch to a Hyper-Recombinational State

See allHide authors and affiliations

Science  26 Sep 2003:
Vol. 301, Issue 5641, pp. 1908-1911
DOI: 10.1126/science.1087706


There is a strong correlation between age and cancer, but the mechanism by which this phenomenon occurs is unclear. We chose Saccharomyces cerevisiae to examine one of the hallmarks of cancer—genomic instability— as a function of cellular age. As diploid yeast mother cells aged, an ∼100-fold increase in loss of heterozygosity (LOH) occurred. Extending life-span altered neither the onset nor the frequency of age-induced LOH; the switch to hyper-LOH appears to be on its own clock. In young cells, LOH occurs by reciprocal recombination, whereas LOH in old cells was nonreciprocal, occurring predominantly in the old mother's progeny. Thus, nuclear genomes may be inherently unstable with age.

Age may be the greatest carcinogen: Cancer incidence increases exponentially near the end of human life (1). Chromosomal abnormalities are a hallmark of most tumors, and it is widely held that genomic instability is a prerequisite for tumorigenesis (2). In older individuals, there is evidence for increased genomic instability, even in noncancerous cells (3). Although numerous hypotheses exist to explain the association between aging and genomic instability (1), these have been difficult to test. To develop a mechanistic understanding of age-related genomic instability, we asked whether such a phenomenon occurs in a model biological system, the budding yeast Saccharomyces cerevisiae.

Heterozygosity was created in a diploid strain by the insertion of a marker gene in one copy of a locus. Loss of heterozygosity (LOH) at the locus was detected when a genetic alteration occurred in which the marker was “lost.” Although LOH in yeast can arise by multiple mechanisms, spontaneous LOH in wild-type cells occurs primarily through mitotic recombination (4). Recombination is presumed to be initiated by DNA damage along the chromosome and is typically accompanied by LOH at all centromere-distal loci (5, 6); accordingly, distal markers were more likely to undergo spontaneous LOH (table S1). Therefore, in order to maximize the chance of observing LOH events, we inserted markers distally on the two longest chromosome arms: at the SAM2 locus on the right arm of chromosome IV and at the MET15 locus on the right arm of XII, about 1 and 2 Mb, respectively, from their centromeres (7). Marker genes affecting colony color when lost were inserted at these loci (8, 9).

The number of daughter cells produced before death by a yeast (mother) cell defines her life-span (10). In order to determine whether genomic instability, manifested as LOH, was affected by a mother cell's increasing age, we isolated by micromanipulation every daughter cell produced from a mother and allowed each daughter to form a colony (11). When the life-spans of the mother cells were complete, daughter colonies were assayed for LOH by changes in colony color. LOH was readily observed in the progeny of aging mothers by the appearance of uniformly colored colonies, or colored sectors within colonies. LOH events resulting in sectored daughter colonies were scored as half-, quarter- or eighth-sectors, which are consistent with the daughter cell or its progeny experiencing an LOH event one, two, or three generations after separation of the daughter from the original mother cell (Fig. 1, A and B).

Fig. 1.

Pedigree analysis reveals age-induced LOH. (A) Daughter colonies were replica-plated to media that permit the detection of loss at MET15 (light gray) and loss at SAM2 (dark gray) (8). Top: Colonies produced by the 12th to 41st (left to right, top to bottom) daughters of a single representative mother cell. Bottom: A schematic representation of the pedigree obtained from these colonies; circles represent colonies derived from individual daughter cells. Colonies with eighth-, quarter-, or half-sectors of colored cells are represented; two LOH events occurred in the 23rd daughter colony. Dashed circles represent daughter cells that did not form a colony. (B) Pedigrees of 40 individual mother cells of strain UCC809, distributed on the y axis with the same representation as in (A); x-axis position indicates the time of separation of the daughter cell from the mother. Double lines precede colonies from daughter cells that could not be separated from the dead mother. (C) The fraction of daughter colonies with LOH events (SAM2 or MET15) for a given mother's age. (D) The number of cell divisions before the first LOH event occurs or between subsequent LOH events at the MET15 or SAM2 locus is presented as a frequency distribution for 40 pedigrees.

Examination of these pedigrees revealed a marked change in LOH with the mother's age (Fig. 1C). Daughter colonies early in the life-span had no LOH events, whereas LOH was observed frequently in the colonies produced by daughters of old mothers. The first LOH events observed in the pedigrees of individual mothers did not occur until the mothers had gone through 23 cell divisions (median value); this late onset was observed at both loci analyzed (Fig. 1D, open bars). However, once an LOH event was observed in a lineage, subsequent LOH events were much more frequent, occurring in every third to fourth daughter colony (Fig. 1, B and D, solid bars). The rate of LOH per cell division in old cells was ∼40 to 200 times that of young cells (Table 1). The frequency of LOH remained constant as the mother cells continued to age: After the first event, there was no significant correlation between the age of the mother and the frequency of subsequent LOH events (MET15 P = 0.69 and SAM2 P = 0.39 for a nonparametric Spearman correlation coefficient).

Table 1.

Age increases the rate of LOH. The rate of LOH at the MET15 and SAM2 loci was calculated for young and old cells as described (11). The 95% confidence interval (CI) is based on the Poisson distribution.

Genotype LOH rate per 10,000 cell divisions (95% CI)
Young Old
Wild type 7 (5-10) 1 (0.5-2.0) 300 (100-500) 200 (50-400)
fob1Δ/fob1Δ 7 (4-10) 1 (0.4-3.0) 150 (90-230) 80 (30-200)
sir2Δ/sir2Δ 160 (120-200) 1 (0.4-3.0) 200View inline (50-300) View inline
  • View inline* The sir2Δ/sir2Δ rate of MET15 LOH in old cells was calculated by half-sector frequency.

  • View inline No sir2Δ/sir2Δ mother cell produced more than a single daughter colony with a SAM2 LOH event.

  • LOH was observed in a majority of the pedigrees: 25 of 40 mothers produced at least one LOH event. Those that did not were generally short-lived. Together, these results demonstrate that there is an age-induced onset of genomic instability in S. cerevisiae. Once an LOH event occurs in a pedigree, additional LOH is observed at a higher frequency for the duration of the mother's life-span. This suggests that as mother cells age, there is a switch from a state with a low spontaneous rate of LOH to a state of increased genomic instability.

    Extrachromosomal ribosomal DNA circles (ERCs) accumulate in aging mother cells and have been proposed to cause replicative senescence (12). To determine whether ERCs were responsible for the observed age-induced genomic instability, we genetically altered the accumulation of ERCs in mother cells and determined the effect on LOH. Cells with a mutant FOB1 gene have reduced levels of recombination at the ribosomal DNA (rDNA) locus and thus accumulate fewer ERCs, whereas mutants in SIR2 have higher rates of rDNA recombination and rapidly accumulate ERCs (1315). Consistent with published results (13), deletion of both copies of the FOB1 gene nearly doubled the life-span (fig. S1A). Nevertheless, the frequency and onset of age-induced LOH were unaffected in these cells. The median number of cell divisions in FOB1-deleted (fob1Δ/fob1Δ) mother cells before a first LOH event was detected was the same as in wild-type mothers: 25 divisions at MET15 and 24 divisions at SAM2. Also unchanged was the increased frequency with which subsequent LOH events occurred (Table 1 and fig. S1C). Overall, the absence of FOB1 affected only the total number of LOH events observed; the longer life-span translated into 38 of 40 mothers displaying age-induced LOH, and nearly three times as many LOH events were detected after the initial event in a lineage. Thus, reducing the level of ERCs prolonged life-span, but did not delay the onset nor decrease the frequency of age-induced genomic instability.

    Deletion of both copies of the SIR2 gene reduced the average life-span of the LOH-detection strain by more than half (fig. S1A), as expected (14). In contrast to wild-type or fob1Δ/fob1Δ cells, MET15 LOH occurred in sir2Δ/sir2Δ cells at a very high frequency, regardless of age (Table 1 and fig. S1C). There was little change in this frequency once a first LOH event occurred in a mother's pedigree (Table 1 and fig. S1C). At the SAM2 locus, however, only a minority of mother cells produced even a single LOH event (8 of 35); no subsequent LOH event was observed in any lineage (15). This discrepancy is readily explained by a locus-specific recombination effect of SIR2. In sir2Δ/sir2Δ cells, mitotic recombination is greatly increased within the rDNA (16). LOH at MET15, which is distal to the rDNA array, increased as well (Table 1); this recombination increase was restricted to chromosome XII and did not affect LOH at SAM2 (Table 1). As sir2Δ/sir2Δ mother cells did not undergo a sufficient number of cell divisions (median life-span was 11 divisions) to achieve the age-induced increase in LOH, virtually no LOH was observed at SAM2. Overall, these data suggest that the onset and subsequent increased frequency of age-induced LOH operate on a different “clock” than does yeast replicative life-span. The aging-associated genetic instability represented by ERCs (17) is thus distinct from the age-induced LOH we observed, and ERCs are not responsible for the age-induced LOH.

    We characterized the mechanism of age-induced LOH by examining whether the age-induced LOH events were the result of chromosome loss. In addition to heterozygosity at the SAM2 locus on the right arm of chromosome IV, the left arm was marked by heterozygosity at the ho locus (ho/hoΔ::TRP1). Both ho alleles were still present in all SAM2 LOH clones. Similarly, no chromosome loss was detected when LOH occurred at MET15 [(15) and see below]. Thus, age-induced LOH is likely initiated by chromosomal damage and is not the result of chromosome nondisjunction during mitosis.

    We next determined whether LOH occurred by reciprocal or nonreciprocal events. Reciprocal recombination (crossing-over) in heterozygous cells results in a mother cell homozygous for one allele and a daughter cell homozygous for the other allele (Fig. 2A). In nonreciprocal events, one of the pair (mother or daughter) loses heterozygosity, while the other remains heterozygous (Fig. 2A). Reciprocal and nonreciprocal LOH events are readily distinguished by examination of half-sectored colonies (18, 19). We found that in young cells, spontaneous LOH at the MET15 locus occurred primarily through crossing-over (Fig. 2A). In contrast, the age-induced LOH events occurred predominantly by a nonreciprocal pathway (Fig. 2A). These results identify a mechanistic difference between the pathway of mitotic recombination normally responsible for LOH in young cells and the pathway that results in increased levels of LOH as cells age.

    Fig. 2.

    Age induces a switch in the mechanism of LOH. (A) Schematic representation of reciprocal and nonreciprocal events leading to LOH. Open or solid squares represent two different alleles at the same locus. Phenotypes of half-sectored colonies of strain UCC768, obtained either by plating cells in culture (young, n = 103) or from daughter colonies of old mothers generated by pedigree analysis (old, n = 74), were used to determine the proportion of reciprocal and nonreciprocal LOH events. (B) Strain UCC768 was subjected to pedigree analysis, and LOH events were analyzed further to identify the mechanism of MET15 LOH. The genotype on the right arm of chromosome XII is diagrammed, with the distance in kilobases between marked loci indicated below the diagram. Genotypes of LOH clones (n = 270) were determined by phenotype of the affected locus or by polymerase chain reaction (12). Open and solid squares represent the corresponding alleles at each locus. (C) The daughter bias of age-induced LOH is shown by the number of LOH events at the indicated locus that created homozygous mothers or homozygous daughters, identified by pedigree analysis. Data were pooled from pedigrees of five strains.

    Mechanisms of nonreciprocal LOH can be distinguished by examination of multiple heterozygous loci along a chromosome. In general, gene conversion, small deletion, and point mutation do not affect markers that are more than 10 kb from the locus where LOH is detected (20). Conversely, break-induced replication (BIR) or chromosome truncation results in the loss of all markers distal from a breakpoint. BIR achieves homozygosity by copying from the homologous chromosome, typically all the way to the telomere (21), whereas a chromosome truncation is manifested by hemizygosity of distal markers. In order to identify the mechanism of age-induced nonreciprocal LOH, a diploid strain was created with heterozygous markers at six loci, including MET15, distributed along ∼2200 kb of the right arm of chromosome XII. Pedigree analysis of these mother cells revealed that age-induced LOH at MET15 occurred with the same late onset, increased frequency, and nonreciprocal character that we found earlier (15). Furthermore, in ∼99% of these LOH events, all the markers distal to MET15 also underwent LOH, including the most distal marker more than 300 kb from MET15 (Fig. 2B) (15). MET15 LOH events initiated at various locations along the chromosome between MET15 and the centromere (Fig. 2B). Quantitative Southern analysis demonstrated that in ∼95% of the events, the loci experiencing LOH remained at two copies per cell (15). Thus, BIR was the predominant pathway of LOH, with ∼5% of the events resulting from a chromosomal truncation. This suggests that age-induced LOH is the result of an age-dependent increase in DNA double-strand breaks and/or a decrease in the normal repair of such damage.

    Complete pedigree analysis also allowed us to determine whether LOH events occurred in mother or daughter cells. The continuous appearance of colored daughter colonies in a mother's life-span demonstrated that the mother cell had undergone LOH (e.g., Fig. 1B, mother 39). A completely colored daughter colony followed by daughter colonies with wild-type cells indicated that LOH occurred in the daughter cell only, while the mother remained heterozygous (e.g., Fig. 1B, mother 28). We presumed that BIR events would occur with equal likelihood in mother and daughter cells. In contrast to this expectation, we found a strong daughter bias (nearly 20-fold) for age-induced LOH at both the MET15 and SAM2 loci (Fig. 2C). This asymmetry presents an interesting biological situation for the mother cell, which may be viewed as a type of stem cell (22, 23). When age-induced LOH occurs, the genomic integrity of the mother appears to be maintained relative to the daughter, preserving the potential to produce subsequent daughters with a wild-type genome.

    What defect causes age-induced LOH? Our mechanistic characterization allows us to eliminate several possibilities. Nondisjunction, leading to chromosome loss, is not responsible. Telomere length is not affected by replicative age in S. cerevisiae (24), precluding telomere loss as the cause of chromosomal instability. An increase in interchromosomal fusions, resulting in dicentric chromosomes, could produce substrates for BIR, but such events do not explain the daughter bias of age-induced LOH.

    Rather, we postulate that aging mother cells accumulate damaged proteins over time (25), which effectively eliminates the normal function of a gene product required for genome integrity. This defect appears to thwart normal DNA damage detection, because, unlike young cells repairing an induced double-strand break (26), mother cell divisions producing a daughter with LOH lacked noticeable cell cycle delays or arrests (Fig. 1B).

    The daughter bias of age-induced LOH may be the result of broken chromosomes that persist immediately after cell division. If the centromere-containing and acentric fragments are partitioned into separate nuclei, with the acentric fragment tending to remain in the mother cell (27), repair from the homolog by BIR (accompanied by LOH) will become the primary DNA repair option in the daughter cell.

    Our results provide predictions about the mechanisms that underlie age-related genomic instability in eukaryotic cells, as well as a model system in which to test them. Ultimately, deeper understanding of this phenomenon in yeast may help to solve the link between oncogenesis and age in humans.

    Supporting Online Material

    Materials and Methods

    Fig. S1

    Tables S1 and S2

    References and Notes

    References and Notes

    View Abstract

    Stay Connected to Science

    Navigate This Article