Telomerase Deficiency Affects the Formation of Chromosomal Translocations by Homologous Recombination in Saccharomyces cerevisiae

Telomerase is a ribonucleoprotein complex required for the replication and protection of telomeric DNA in eukaryotes. Cells lacking telomerase undergo a progressive loss of telomeric DNA that results in loss of viability and a concomitant increase in genome instability. We have used budding yeast to investigate the relationship between telomerase deficiency and the generation of chromosomal translocations, a common characteristic of cancer cells. Telomerase deficiency increased the rate of formation of spontaneous translocations by homologous recombination involving telomere proximal sequences during crisis. However, telomerase deficiency also decreased the frequency of translocation formation following multiple HO-endonuclease catalyzed DNA double-strand breaks at telomere proximal or distal sequences before, during and after crisis. This decrease correlated with a sequestration of the central homologous recombination factor, Rad52, to telomeres determined by chromatin immuno-precipitation. This suggests that telomerase deficiency results in the sequestration of Rad52 to telomeres, limiting the capacity of the cell to repair double-strand breaks throughout the genome. Increased spontaneous translocation formation in telomerase-deficient yeast cells undergoing crisis is consistent with the increased incidence of cancer in elderly humans, as the majority of our cells lack telomerase. Decreased translocation formation by recombinational repair of double-strand breaks in telomerase-deficient yeast suggests that the reemergence of telomerase expression observed in many human tumors may further stimulate genome rearrangement. Thus, telomerase may exert a substantial effect on global genome stability, which may bear significantly on the appearance and progression of cancer in humans.


Introduction
The incidence of cancer increases with age, suggesting that physiological changes associated with aging contribute to carcinogenesis [1]. Cellular proliferation over the course of a lifetime coincides with a progressive loss of DNA from the ends of chromosomes, or telomeres, which culminates in a cessation of cell division known as replicative senescence [2,3]. In yeast, replicative senescence appears to be concurrent with telomere dysfunction and crisis, a precipitous loss of cell viability [4]. However, in vertebrates the subsequent loss of the tumor suppressors p53 or Rb in a rare subset of cells is required to escape replicative senescence and undergo the further telomere shortening that leads to telomere dysfunction and crisis [5][6][7][8][9]. Such critical shortening of the telomeres in the cells of late-generation mice (G6-G8) unable to replicate their telomeres can lead to a loss of telomere function that potentiates the development of epithelial cancers [5,10]. These tumors display aneuploidy and non-reciprocal translocations that are also characteristic of the epithelial tumors of adult humans. Importantly, critically shortened telomeres are also observed in precancerous lesions, further suggesting a link between telomere shortening and the development of cancer [11][12][13][14][15][16].
Replicative senescence is prevented through the action of the ribonucleoprotein complex telomerase, which acts to maintain telomeres at a stable length [4,17,18]. However, telomerase is not expressed in most normal human somatic cells making them vulnerable to the aging related replicative senescence and potential telomere dysfunction during crisis that has been implicated in carcinogenesis [1,19]. Work done in yeast, plants and mice has demonstrated that loss of telomere function leads to genome destabilizing events, the study of which gives mechanistic insight into the relationship between telomere dysfunction and carcinogenesis [5,[20][21][22][23][24][25][26][27][28]. For instance, a burst of mutagenesis concurrent with crisis and its attendant telomere dysfunction [20,26] could contribute to the critical number of mutations thought to be necessary to elicit carcinogenesis [29,30]. Dysfunctional telomeres are also subject to telomere-telomere fusion that initiates the breakage-fusion-bridge cycle [31] and propagates a wave of genome destabilizing events typical of tumor cells [32]. In yeast, homologous recombination (HR) between allelic sequences near telomeres is stimulated during crisis [23,28], which results in loss of heterozygosity that also contributes to tumor formation in humans [33]. Crisis-associated HR may also project instability away from telomeres if it stimulates interactions between dispersed, repetitive sequences such as Alu elements [34]. Such events have been observed to generate deletions and translocations associated with cancer [35,36].
Recent discoveries have suggested that telomerase may have effects on global genome stability, particularly with regard to responses to DNA double-strand breaks (DSB) [37][38][39][40][41]. In particular, telomerase-deficient mice display hypersensitivity to ionizing radiation and delays in DSB repair that may reflect a defect in HR [37]. While it remains unclear if the effect of telomerase on DSB repair is direct or indirect, these results suggest that telomerase may affect the generation of tumorigenic genome rearrangements.
In this manuscript we describe the effect of the loss of telomerase on the formation of chromosomal translocations by HR in budding yeast. We observed a crisis-associated increase in the rate of spontaneous translocation when one of the substrates was located near a telomere that was similar to increases in the rate of mutation and interhomolog recombination observed previously at this and other loci [20,23,26,28]. This is consistent with the gathering evidence for a substantial increase in genome instability concomitant with crisis. In addition to these crisisassociated effects we have also shown that loss of telomerase results in decreased frequencies of translocation following DSB formation adjacent to the translocation substrates, before, during and after crisis. This defect in DSB repair reflects a constitutive sequestration of the central HR protein, Rad52, at telomeres, and a decrease in its recruitment to DSBs, suggesting that telomerase exerts an indirect effect on global genome stability. These results suggest that the resumption of telomerase function in premalignant cells could stimulate genome rearrangement that may contribute to the progression toward tumor formation in humans.

Results
Chromosomal translocations form spontaneously by HR in response to telomere dysfunction when one substrate is proximal to a telomere In aging mice and human cells lacking telomerase, the formation of translocations has been observed in response to telomere dysfunction during crisis, and is thought to be involved in carcinogenesis [1,10,19]. It has been shown previously that yeast mutants carrying a null allele of EST2, which encodes the reverse transcriptase subunit of telomerase [42], display crisis-associated increases in the rates of spontaneous mutation and gross chromosomal rearrangement at a locus that lies near a telomere, but not at a locus that lies proximal to the centromere [26]. This suggests that crisis may promote instability in a restricted region of the genome. Since telomerase deficiency has also been associated with a crisis-associated increase in spontaneous recombination between homologs for sequences that are located proximal to the telomere [23,28], and spontaneous translocation formation involving telomere-proximal sequences has been observed in telomerase-deficient yeast cells [20], we sought to determine the impact of crisis on the rate of spontaneous translocation formation by HR. Spontaneous translocation formation has been shown to occur by HR between dispersed, repetitive genomic sequences in humans [35,36], which could be a means of spreading the destabilizing effects of dysfunctional telomeres to regions that do not lie near a telomere.
Inter-chromosomal recombination was measured in diploids as the rate of formation of an intact HIS3 gene by HR between a 59 truncated his3 allele (his3-D59) located at the LEU2 locus on one copy of chromosome III, and a 39 truncated his3 allele (his39-D39) located at either the HIS3 locus on one copy of chromosome XV, or at the CAN1 locus on one copy of chromosome V [43]. The his3-D59 and his3-D39 alleles share 311 bp of homology, but neither shares sufficient homology with his3D200 [44], an allele present at the HIS3 locus on one or both copies of chromosome XV, to generate an intact HIS3 gene by HR.
Consistent with previous experiments demonstrating crisisassociated genome instability at telomere proximal loci [20,23,26,28], spontaneous inter-chromosomal recombination increased during crisis when one of the recombination substrates was telomere proximal (Table 1). In est2D/est2D mutant homozygotes, the rate of inter-chromosomal recombination when the his3-D39 substrate was at the CAN1 locus, 32 kb from the telomere on the left arm of chromosome V, was not significantly different from wild type before crisis, but increased to 17.3-fold over wild type during crisis, and decreased to near wild type levels after recovery. However, when the his3-D39 substrate was at the HIS3 locus that lies 360 kb from the telomere on the right arm of chromosome XV, the rate of spontaneous inter-chromosomal recombination was not significantly different form wild type before, during or after crisis. These results demonstrate that spontaneous inter-chromosomal recombination increases during crisis in telomerase deficient cells, but suggests that this increase requires that at least one of the substrates be located proximal to the telomere at the CAN1 locus, where recombinagenic lesions may accumulate during crisis. While crisis-associated lesion formation may be restricted to telomere proximal regions, such as CAN1, the location of his3-D39 near a telomere does not affect its inherent capacity to participate in HR as the rates of interchromosomal recombination in wild type cells were not significantly different when his3-D39 was at CAN1, or HIS3. Since the difference in the rates of inter-chromosomal recombination observed in senescent est2D/est2D homozygotes having the his39-D39 substrates at HIS3 or CAN1 did not differentially influence the kinetics of the initiation of, or recovery from crisis (Figure 1a), the stimulation of inter-chromosomal recombination appears not to be a determinant of cell viability.
His + recombinants recovered from ten independent wild type and senescent est2D/est2D cultures that carried the his3-D39 substrate at either the HIS3 or CAN1 locus, were subjected to genomic Southern blot and chromosome blot analyses to determine the precise nature of the recombination events as described previously [43]. All recombinants from wild type and est2D/est2D strains carrying the his3-D39 substrate at either HIS3 or CAN1 possessed an intact HIS3 gene (D. Meyer and A. Bailis, unpublished observations) and revealed very similar patterns of product formation; 8/10 to 10/10 of the recombinants possessed either the 0.78 Mb tXV:III or 0.64 Mb tV:III primary translocation chromosomes predicted to be formed during the creation of an intact HIS3 gene, while 4/10 to 5/10 also carried the 0.63 Mb tIII:XV or 0.26 Mb tIII:V chromosomes predicted for the products of reciprocal translocation ( Figure 2). The few His + recombinants whose karyotypes differ from these dominant patterns were observed in equal proportions among the wild type and est2D/est2D recombinants (D. Meyer and A. Bailis, unpublished observations) and reflect additional rearrangements subsequent to the initial translocation event [43]. These results demonstrate that spontaneous inter-chromosomal recombination between the his3-D59 and his3-D39 substrates generates translocations, and that these events are similar regardless of the location of the his3-D39 substrate, or whether they are occurring in wild type or senescent est2D/est2D homozygotes. Strains carrying spontaneous translocations have been shown previously to retain intact recombination substrates, suggesting that spontaneous translocation may occur by a conservative mechanism [43]. Chromosome blot analysis of the spontaneous wild type and est2/est2 His + recombinants suggested that the majority (7/10) possessed an intact his3-D59 substrate, an intact his3-D39 substrate, or both. This was indicated by the presence of sequences homologous to HIS3 on intact copies of the 0.32 Mb chromosome III and the 0.58 Mb chromosome V ( Figure 2). Genomic Southern blot analysis of BamHI digested DNA confirmed the presence of intact his3-D59 and, or his3-D39 substrates at the LEU2 and CAN1 loci, and revealed an intact his3-D39 substrate at the HIS3 locus in several recombinants (D. Meyer and A. Bailis, unpublished observations). This indicates that spontaneous translocation formation occurs by the same conservative mechanism in wild type and senescent est2D/est2D homozygotes, regardless of the locations of the substrates. These results also suggest that crisis in telomerase deficient diploids can result in telomere proximal instability that leads to increased spontaneous translocation by HR.

Telomerase deficiency causes a defect in the generation of translocations subsequent to DSB formation
The efficiency of translocation was shown previously to increase 10 7 -fold when HO endonuclease catalyzed DSBs are introduced adjacent to both his3-D59 and his3-D39, suggesting that translocation formation by HR between repetitive sequences may be very efficient in conditions that generate multiple genomic DSBs [43]. In that study, mutations in several DNA repair genes were found to confer increases in the rate of spontaneous translocation similar to that conferred by the est2D allele, but had distinct effects on DSB-stimulated translocation. Therefore, we also examined the effect of the est2D mutation on DSB-stimulated translocation ( Table 2). Surprisingly, we found that with the his3-D39 substrate at the CAN1 locus, the frequency of DSB-stimulated translocation in est2D/est2D homozygotes was reduced 12-to 20-fold, regardless of whether the cells were from pre-senescent, senescent or postsenescent cultures. Since ascertaining the frequency of DSBstimulated translocation is dependent upon the number of cells that survive DSB formation, the observation that plating efficiency following DSB formation in the est2D/est2D homozygote was reduced very little (Table 3) suggests that our assessment of the effect of est2D on translocation frequency was unaffected [43]. This demonstrates that the absence of telomerase leads to a defect in DSB-stimulated translocation regardless of the growth stage of the culture, and suggests that telomerase is required for efficient DSB repair by HR. Upon examining DSB-stimulated translocation in cells where the his3-D39 substrate was at the HIS3 locus, we found that the est2D allele had a similar effect, as the frequency was reduced 14-to 60-fold in the est2D/est2D homozygote regardless of whether the cells were pre-senescent, senescent or post-senescent. This demonstrates that the loss of telomerase has similar, inhibitory effects on DSB-stimulated translocation regardless of the location of the substrates. Similar results were obtained with diploids homozygous for a null allele of the TLC1 gene, which encodes the RNA subunit of telomerase [45], suggesting that the entire telomerase complex exerts an effect on DSB-stimulated translocation. Together, these results suggest that telomerase supports DSB-stimulated translocation formation by HR in yeast, Figure 1. Viability in serial liquid culture of telomerase deficient diploid cells carrying translocation substrates is influenced by rad52D but not by rad1D or rad59D. Each strain possessed the his3-D59 substrate at the LEU2 locus on chromosome III, and the his3-D39 substrate at either the HIS3 locus on chromosome XV or the CAN1 locus on chromosome V. The position of the substrates in each strain is denoted by a superscript. Serial liquid growth was performed as described previously [26]. Mean viabilities62 SE from at least eight independent cultures were plotted for each day of serial growth of strains of each indicated genotype.  (a) Chromosomal products expected from recombination between the his3-D59 substrate at the LEU2 locus on one copy of chromosome III and the his3-D39 substrate at the CAN1 locus on one copy of chromosome V. Both substrates are oriented toward the telomeres on the left arms of their respective chromosomes and contain a 117 bp HO endonuclease recognition sequence as depicted. The black bars depict the extent of the 1.8 kb HIS3 probe used to localize HIS3 sequences to individual chromosomes on Southern blots by hybridization. Recombination between the substrates yields an intact HIS3 gene on a tV:III translocation chromosome, and may also yield a tIII:V reciprocal product. (b) CHEF gels and blots of chromosomes prepared from a His 2 parent strain carrying the his3-D39 substrate on chromosome V and the his3-D59 substrate on chromosome III, and His + recombinants derived from it. Separated chromosomes were visualized by ethidium bromide staining, blotted to nylon and probed with the 1.8 kb BamHI HIS3 genomic clone. Positions of the recombination substrates and products are shown. The presence of the substrates and products in the parent and recombinant strains are denoted. The HIS3 probe hybridizes with the his3D200 allele on both copies of chromosome XV. (c) Chromosomal products expected from recombination between the his3-D59 substrate at the LEU2 locus on one copy of chromosome III and the his3-D39 substrate at the HIS3 locus on one copy of chromosome XV. Both substrates are oriented toward the telomere on their respective chromosomes and contain a 117 bp HO endonuclease recognition sequence as depicted. Recombination between the substrates yields an intact HIS3 gene on a tXV:III translocation chromosome, and may also yield a tIII:XV reciprocal product. (d) CHEF gels and blots of chromosomes prepared from a His 2 parent strain carrying the his3-D39 substrate on chromosome XV and the his3-D59 substrate on chromosome III, and His + recombinants derived from it. The gel and blot were prepared and displayed as above. doi:10.1371/journal.pone.0003318.g002 regardless of where the substrates are located in the genome, or the growth stage of the culture.
Ten independent His + recombinants recovered from both wild type and senescent est2D/est2D homozgotes that had undergone HO-catalyzed DSB formation adjacent to the recombination substrates were subjected to the same genomic Southern blot and chromosome blot analyses used to characterize the products of spontaneous recombination ( Figure 2). Like in the previous analysis, the genomic Southern blots revealed that all of the recombinants displayed bands consistent with the presence of the intact HIS3 sequence predicted for inter-chromosomal recombination (D. Meyer and A. Bailis, unpublished observations). Chromosome blot analysis of the recombinants revealed that 9/ 10 to 10/10 possessed the 0.78 Mb tXV:III or 0.64 Mb tV:III primary translocation chromosomes formed while creating an intact HIS3 gene, while 4/10 to 6/10 also carried the 0.63 Mb tIII:XV or 0.26 Mb tIII:V chromosomes predicted for products of reciprocal translocation ( Figure 2). As above, the few recombinants that did not display these dominant karyotypic patterns displayed evidence of additional genome rearrangements subsequent to translocation formation [43]. These results suggest that, like spontaneous recombination, the repair of HO-stimulated DSBs by HR between non-homologous chromosomes creates similar translocation chromosomes regardless of the location of the his3- The median translocation frequency was determined for each strain from at least 12 independent cultures. Fold differences from the median frequency obtained with the wild type strain with the same substrates, located at the same genomic positions, and at equivalent time-points are indicated in parentheses. The 95% confidence intervals are indicated in brackets. Pre-senescence corresponds to approximately 25 generations, senescence corresponds to approximately 60 generations, and postsenescence corresponds to approximately 100-125 generations. b Translocation substrates were as described in the Materials and Methods. The frequencies for the rad1D/rad1D (XV/III) , rad52D/rad52D (XV/III) and rad59D/rad59D (XV/III) strains have been previously reported [43].  Table 3. Plating efficiency of wild type and mutant diploids before and after the creation of multiple double-strand breaks by HO endonuclease a .
Plating efficiency (%) D39 substrate, or whether it is occurring in wild type or senescent est2D/est2D homozygotes. Unlike the strains bearing translocations arising by spontaneous inter-chromosomal recombination, strains carrying DSB-stimulated translocations were previously found not to retain the his3-D59 or his3-D39 substrates [43]. In this study, a similar, non-conservative pattern of substrate utilization was observed for all of the His + recombinants recovered after HO-stimulated DSB formation, as chromosome blot analysis failed to reveal the presence of HIS3 sequences on any intact copy of chromosome III or V (Figure 2). These results were verified by genomic Southern blot analysis, which also showed that his3-D39 was absent from chromosome XV in all of the recombinants (D. Meyer and A. Bailis, unpublished observations). Therefore, as for spontaneous translocation formation, DSBstimulated translocation appears to occur by a similar mechanism in wild type and senescent est2D/est2D homozygotes, regardless of where the his3-D39 substrate is located.
Epistatic interactions of est2D are similar to those of rad52D with respect to DSB-stimulated translocation formation The genetic control of DSB-stimulated translocation formation was investigated previously [43], and found to involve a distinctive group of genes including, RAD1, which encodes a subunit of an endonuclease involved in nucleotide excision repair and HR [46,47], RAD52, which encodes the central HR protein [48,49] that can anneal single-stranded DNA [50], and RAD59, a RAD52 paralog [51,52]. When an epistasis analysis was performed, rad1D was found to be epistatic to rad52D, while combining rad52D and rad59D had a greater effect than either single allele [43]. This suggests that Rad1 and Rad52 may function sequentially in translocation formation, while Rad52 and Rad59 may perform at least partially discrete functions in this process.
Since the effects on DSB-stimulated translocation formation conferred by the est2D allele suggest that it is involved in the repair of DSBs by HR, we examined its relationship with the rad1D, rad52D and rad59D alleles by performing an epistasis analysis ( Table 2). We used strains in which the his3-D39 substrate was located at the HIS3 locus so that the results could be compared more directly to those obtained previously for the rad1D/rad1D, rad52D/rad52D, and rad59D/rad59D homozygotes [43]. The frequency of translocation in the est2D/est2D rad1D/rad1D double homozygote was not significantly different from that of the rad1D/ rad1D homozygote before, during or after crisis ( Table 2), indicating that rad1D is epistatic to est2D. This suggests that, like Rad52, Est2 may function sequentially with Rad1. Further, since the frequency of translocation is the same in pre-senescent, senescent and post-senescent cells, Rad1 appears to exert the same effect throughout the period of serial growth. Similarly, the est2D/ est2D rad59D/rad59D double mutant displays essentially the same frequency of translocation in pre-senescent, senescent and postsenescent cells, but here the frequency is significantly lower than in either the est2D/est2D or rad59D/rad59 single homozygotes ( Table 2). This suggests that, like Rad52, Est2 performs a function in DSB-stimulated translocation that does not overlap with Rad59, and that Rad59 plays the same role throughout the period of serial growth. Importantly, the defects in DSB-stimulated translocation in both double mutants had less than a two-fold effect on plating efficiency subsequent to DSB formation, indicating that our determinations of translocation frequency were minimally affected [43]. We also found that the kinetics of crisis in the est2D/est2D rad1D/rad1D and est2D/est2D rad59D/rad59D double homozygotes (Figure 1c) were similar to those of the est2D/est2D homozygote (Figure 1b), suggesting that the defects in DSB-stimulated translocation in these strains do not substantially affect their growth properties.
Since est2D behaves much like rad52D with respect to its interactions with rad1D and rad59D we, therefore investigated the interaction between est2D and rad52D. One complication associated with this analysis is that combining est2D and rad52D has a delayed, synthetically lethal effect, such that cultures of est2D/est2D rad52D/rad52D double homozygotes do not survive much beyond the 30 th generation (Figure 1c), likely reflecting the necessity for Rad52 in the recovery from crisis [53]. However, at 25 generations cell viability in the est2D/est2D rad52D/rad52D cultures is sufficient to determine the frequency of DSB-stimulated translocation (Figure 1c). Interestingly, the translocation frequency in pre-senescent est2D/est2D rad52D/rad52D cells was 154-fold below that of pre-senescent est2D/est2D cells, and 42-fold below the level in rad52D/rad52D homozygotes ( Table 2). Plating efficiency of the est2D/est2D rad52D/rad52D double homozygotes following DSB formation was also significantly reduced (Table 3), presumably due to defective repair of DSBs at the MAT loci on both copies of chromosome III [43]. However, this decrease is unlikely to explain the difference in the frequencies of DSBstimulated translocation in the est2D/est2D rad52D/rad52D double and rad52D/rad52D single homozygotes as their plating efficiencies are not significantly different following DSB formation [43]. Together, these results suggest that Est2 and Rad52 play distinct roles in DSB-stimulated translocation.

Telomerase deficiency leads to increased association of Rad52 with the telomere
The results of the genetic experiments described above suggest that Est2 may operate like Rad52 in DSB-stimulated translocation, but that these are separate functions. The ability of Rad52 to anneal complementary single-stranded DNA molecules [50] is thought to play a critical role in DSB-stimulated translocation formation by promoting single-strand annealing (SSA) [43]. Rad52 has been shown to physically associate with DSBs, which is thought to precede their repair by HR [54,55]. While telomerase has also been shown to bind to chromosomal DSBs, this has only been documented when TG-repeat sequences typical of telomeres were adjacent to the break sites [56]. Further, it is unclear how telomerase, complete with its TLC1 encoded RNA template would promote DSB-stimulated translocation formation by HR. In order to investigate the physical basis of how telomerase affects translocation, we used chromatin-immunoprecipitation (ChIP) to track the association of MYC-tagged Est2 protein with a marked telomere, adh4::URA3::TELVII [57], and with the his3-D39 substrate at the HIS3 locus before and after DSB formation in wild type cells (Figure 3a). These experiments were performed in haploid strains that possessed recombination substrates at the HIS3 and LEU2 loci, and displayed frequencies of translocation formation that were very similar to those observed in diploids (D. Meyer and A. Bailis, unpublished observations). Similar to previously published results [58], we observed a 9-to 14-fold enrichment of telomere sequences over background at all time points, indicating stable association of Est2 with the telomere (Figure 3d). In contrast, we saw no enrichment of his3D39 sequences above background at any time after DSB formation, indicating that Est2 was not present. Tagging EST2 with MYC epitope sequences did not affect frequencies of DSB-stimulated translocation or the growth properties of the cells (D. Meyer and A. Bailis, unpublished results). These results strongly suggest that Est2, and by extension telomerase exerts its effect on DSBstimulated translocation without binding to the recombination substrates.
Since our genetic data indicate that telomerase may function similarly to Rad52 in DSB-stimulated translocation, but the ChIP results indicate that its effect is likely to be indirect we addressed the possibility that telomerase influences the behavior of Rad52. We investigated this by using ChIP to compare the interaction of FLAG-tagged Rad52 with the marked telomere and the his3-D39 substrate in wild type and pre-senescent est2D mutant strains before and after DSB formation (Figure 3b and 3c). Translocation formation in these strains occurred at frequencies that were very similar to those observed in wild type and est2/est2 homozygous diploids (D. Meyer and A. Bailis, unpublished observations). Unlike Est2-MYC in wild type cells, Rad52-FLAG appeared to localize weakly to telomeres, as indicated by the modest, two-fold enrichment of telomere sequences above background at all time points (Figure 3e). In contrast, Rad52-FLAG localized strongly to his3-D39 subsequent to DSB formation, as indicated by a robust, five-to 10-fold enrichment of his3-D39 sequences above background, which is similar to previously published results [54,55]. Importantly, Rad52-FLAG displayed near maximal localization to his3-D39 one hour after DSB formation, suggesting that there was rapid association of substantial quantities of protein to the recombination substrates. This is consistent with the high frequencies of DSB-stimulated translocation observed in wild type cells (Table 2; D. Meyer and A. Bailis, unpublished observations).
The ChIP experiments with pre-senescent est2D mutant strains revealed changes in the localization of Rad52-FLAG to the marked telomere and the his3-D39 substrate (Figure 3c) that are consistent with the reduced frequency of DSB-stimulated translocation in est2D/est2D homozygotes described above (Table 2). In contrast to the weak localization of Rad52-FLAG with the telomere in wild type cells (Figure 3b), Rad52-FLAG displayed a robust localization with the telomere in the est2D mutant, as enrichment of telomere sequences was six-to nine-fold above background levels at all time points (Figure 3f). Notably, these cells did not display the loss of viability indicative of the onset of crisis (Figure 1b; D. Meyer and A. Bailis, unpublished results), suggesting that Rad52 may localize to telomeres in the est2D mutant prior to telomere recovery by HR [53]. The increased localization of Rad52-FLAG to telomeres in the est2D mutant correlated with reduced localization to his3-D39 at all time points . ChIP analysis of the association of Est2 and Rad52 with a telomere and a recombination substrate in wild type and presenescent est2 mutant strains. (a) Association of Est2 with a telomere and a recombination substrate in wild type cells. Wild type cells containing the EST2-MYC18-TRP1 allele were collected before (0 hours) and at various times after (0.5, 1, 2, and 4 hours) the induction of HO endonuclease and subjected to ChIP using anti-Myc antibody [74]. Retained DNA was subjected to pairwise multiplex PCR using primers specific for SAM1 (S) and the his3-D39 recombination substrate at the HIS3 locus (H), and RAD59 (R) and the marked telomere on the left end of chromosome VII (T). DNA collected following ChIP of extract from a strain that contained an untagged EST2 gene was subjected to PCR using the R and T primer sets to yield a no tag control (NT). PCR of whole cell extract using the S, H, R and T primer sets was used as a PCR control (C). PCR products were run on 1.5% agarose gels and stained with ethidium bromide. (b) Association of Rad52 with a telomere and a recombination substrate in wild type cells. Wild type cells containing the RAD52-FLAG-KAN-MX allele were collected before and at various times after the induction of HO endonuclease and subjected to ChIP using anti-FLAG antibody. PCR products were generated and displayed as above. (c) Association of Rad52 with a telomere and a recombination substrate in est2 mutant cells. est2 mutant cells containing the RAD52-FLAG-KAN-MX allele were collected before and at various times after the induction of HO endonuclease and subjected to ChIP using anti-FLAG antibody. PCR products were generated and displayed as above. after DSB formation, as enrichment of his3-D39 sequences was only two-to six-fold above background (Figure 3f). Critically, since peak signal level was not reached until two hours after DSB formation, these data suggest that twice as much time was required to localize half as much Rad52-FLAG to the recombination substrate in est2D mutant cells as in wild type. Cutting by HO endonuclease at his3-D39 and his3-D59 was greater than 90% complete by 30 minutes in both wild type and est2D mutant cells (D. Meyer and A. Bailis, unpublished observations), suggesting that the differences in localization of Rad52 to his3-D39 were not due to differences in the levels of substrate cutting by HO endonuclease. Since Rad52 is critical for DSB-stimulated translocation formation (Table 2) [43], these data are consistent with telomerase deficiency reducing translocation frequency by sequestering Rad52 to telomeres.

Discussion
This manuscript describes evidence that telomerase affects the formation of chromosomal translocations by HR. We found that the rate of spontaneous translocation increased 17.3 fold during crisis in telomerase deficient cells with the his3-D39 substrate located at the CAN1 locus near the telomere on the left arm of chromosome V, but was not significantly increased before, or after (Table 1) crisis. Further, this crisis-associated increase was not observed when the his3-D39 substrate was located at the HIS3 locus that is distant from the telomere on the right arm of chromosome XV, suggesting that the recombinagenic effect of crisis may be restricted to telomere proximal regions of the genome. These results are consistent with the observation of spontaneous translocation formation involving sequences at other telomere proximal loci in telomerase-deficient cells [20]. Effects on heteroallelic recombination have been suggested to be the result of crisis-associated exonucleolytic degradation from uncapped telomeres, which can confer a recombinagenic state on telomere proximal sequences [23]. However, the retention of telomere proximal sequences in the form of a reciprocal translocation chromosome ( Figure 2) in a significant fraction of the recombinants suggests that crisis-associated translocations need not arise exclusively from exonucleolytic degradation from the telomere. We speculate that DNA replication lesions, such as those thought to stimulate crisis-associated mutagenesis at telomere proximal loci [26] may also contribute to spontaneous translocation formation in senescent cells. Similar events in senescent human cells that have entered crisis could generate translocations by HR between abundant and widely dispersed repetitive sequences, such as Alu elements, which are found near telomeres and can participate in translocation formation by HR [59].
Translocation formation has been shown previously to increase many-fold subsequent to the formation of DSBs adjacent to both recombination substrates, consistent with its occurrence by SSA, a particularly efficient mechanism of HR [43,60]. In contrast to its effect on spontaneous translocation formation, telomerase deficiency decreased the frequency of DSB-stimulated translocation by as much as 250-fold ( Table 2), suggesting that telomerase contributes to DSB repair by HR. Additionally, in telomerase deficient cells the frequency of DSB-stimulated translocation was reduced before, during and after crisis, and regardless of whether either of the recombination substrates was located near a telomere. This suggests that telomerase exerts a genome-wide effect on DSB repair that is independent of its role in forestalling crisis. Analysis of the epistasis interactions between est2D and several HR gene mutations ( Table 2) suggested that telomerase plays a role in DSBstimulated translocation formation by SSA that is similar to that of Rad52 but is distinct from it. The observation that telomerase deficiency also reduces translocation formation following the creation of a DSB adjacent to only one recombination substrate (D. Meyer and A. Bailis, unpublished observations) suggests that telomerase deficiency may have a more general effect on DSB repair by HR. These results are reminiscent of the ionizing radiation sensitivity and DNA repair defects displayed by telomerase deficient mice [37].
Analysis of the association of Rad52 with a telomere and a recombination substrate by ChIP in wild type and est2D/est2D mutant cells showed that telomerase deficiency results in a crisisindependent sequestration of Rad52 to the telomere that affects its ability to localize to the his3-D39 recombination substrate at the HIS3 locus after DSB formation (Figure 3). Both the maximum level, and rate of accumulation of Rad52 at the recombination substrate were reduced. Given the central role of Rad52 in DSBstimulated translocation formation [43], this is consistent with the reduced frequencies of translocation observed in telomerasedeficient cells (Table 2). However, given the proximity of the CAN1 locus to the telomere on the left end of chromosome V it is notable that translocation involving the his3-D39 substrate is as diminished when it is located there as it is when it is located at the telomere distal HIS3 locus. This suggests that CAN1, which is 32 kb from the end of the chromosome and greater than 20 kb from the subtelomeric domain may be too distant from the telomere to be influenced by the Rad52 sequestered there. Alternatively, the reduced frequency of translocation may reflect the necessity for sufficient levels of Rad52 at both substrates to observe wild type frequencies of DSB-stimulated translocation.
Observing the sequestration of Rad52 to telomeres before any diminution of proliferative capacity in est2D/est2D diploid (Figure 1a) or est2D haploid [26] mutants is detected suggests that telomeres may be recognized as DNA lesions before they undergo the HR that facilitates recovery from crisis [53]. The functional significance of this early association of Rad52 with telomeres is suggested by the greatly accelerated crisis exhibited by est2D/est2D rad52D/rad52D double homozygotes (Figure 1c), which may reflect a role for Rad52 in accommodating to defective telomere replication. These results are consistent with the observation of H2AX foci at telomeres in pre-senescent human mammary epithelial cells [61], which suggests that telomerase deficient yeast and human cells are involved in an essentially constitutive response to DNA damage at their telomeres. Relieved of this burden, malignant cells that have reactivated telomerase may have an enhanced capacity to repair DSBs, which may, perhaps help to explain their characteristic abundance of genome rearrangements.
While we have demonstrated that telomerase deficiency can affect the localization of Rad52 to DSBs, our genetic results suggest that Rad52 is unlikely to be the only important HR protein whose behavior is affected. The synergistic reduction in translocation frequency observed when the est2D and rad52D alleles are combined in double mutants ( Table 2), suggests that telomerase deficiency may affect the availability of factors other than Rad52 for DSB-stimulated translocation formation. Null alleles of both RAD59, and the central mismatch repair gene, MSH2, reduce DSB-stimulated translocation formation to approximately the same extent as rad52D, suggesting that they also encode required factors [43]. Importantly, like rad52D [53], msh2D [62] and rad59D [63] have been shown to affect HR-dependent telomere recovery, consistent with their action at telomeres. Perhaps the loss of telomere integrity that sequesters Rad52 at telomeres in telomerase deficient cells also sequesters Msh2 and/or Rad59, which would be consistent with the very low levels of DSBstimulated recombination observed in rad52D/rad52D rad59D/ rad59D double homozygotes [43]. We are currently investigating the possibility that telomerase deficiency affects the localization of Rad59 and Msh2 to recombination substrates, and whether such a change might be the result of altering their association with telomeres.
Many of the cancers seen in aging adults originate in epithelial cells that lack telomerase activity [1]. Studies of the role of telomerase in the maintenance of genome integrity in mammalian cells have provided insight into the molecular aspects of the relationship between aging and cancer [19,27,64,65]. The work presented here suggests another mechanism by which telomerase deficient cells that enter crisis may suffer increased rates of spontaneous genome rearrangement that can, perhaps, lead to cancer. However, this work suggests that telomerase also promotes genome rearrangements by DSB repair. This could have a significant impact on the response of tumor cells to radiation and chemotherapy, as many tumors reactivate the expression of telomerase [66], and these treatments generate DSBs in mammalian cells [67][68][69]. Careful consideration of the multiple ways in which telomerase influences genome stability may, therefore, facilitate better prevention, diagnosis and treatment of cancer.
The RAD52-FLAG-KAN-MX allele was created by PCR using p3FLAG-KanMX as a template provided as a kind gift from Michael Grunstein. The can1::his3-D39-HOcs translocation substrate cassette was created by insertion of a HincII/PvuII fragment from pLAY500 containing the his3-D39-HOcs fragment from pLAY500 into the MscI site of CAN1 in pLAY475 to create can1::his3-D39-HOcs in pLAY588. The can1::his3-D39-HOcs fragment was released from pLAY588 by digestion with PstI, and cloned into YIp365R, which possesses a URA3 selectable marker, that had been digested with PstI to create pLAY589. pLAY589 was linearized by digestion with NaeI to target its transplacement into the CAN1 locus following transformation and selection for uracil prototrophy. Uracil prototrophs were plated to 5-FOA media to select for plasmid loss and the resulting 5-FOA r colonies assessed for canavanine sensitivity by replica plating to synthetic medium lacking arginine and containing canavanine at a concentration of 60 mg/ml. Canavanine resistant colonies were subjected to genomic Southern blot analysis to confirm integration of the can1::his3-D39-HOcs construct into the CAN1 locus (D. Meyer and A. Bailis, unpublished observations). Construction of the other translocation assay components; his3-D39-HOcs, leu2::HOcs-his3-D59 (300) , and trp1::GAL1-HO-KAN-MX were discussed previously [43].
All strains possess the his3-D59 substrate at the LEU2 locus on one copy of chromosome III [43]. The his3-D39 substrate is located either at the CAN1 locus on one copy of chromosome V (V/III) , or at the HIS3 locus on one copy of chromosome XV (XV/III) [43]. The his3-D59 and his3-D39 substrates share 311 bp of HIS3 coding sequence. These strains also possess either one (XV/III) , or two (V/III) copies of the his3-D200 allele at the HIS3 locus that is unable to form an intact HIS3 gene by recombination with either substrate.
The plasmids pSD196 and pSD120, containing, respectively the wild type EST2 and TLC1 genes inserted into the URA3 containing single-copy vector pRS416 [79], were generous gifts from Daniel Gottschling.

Serial liquid growth assay
Serial liquid growth was performed as described previously [28] and was initiated by inoculating five ml of YPD (1% yeast extract, 2% bacto peptone, 2% dextrose) with single colonies from which pSD196 or pSD120 had been lost following a period of nonselective growth. Upon reaching saturation following 24 hours of growth at 30uC hemocytometer counts were performed to determine the total number of cell bodies, after which 5610 5 cells were inoculated into 5 ml of fresh YPD liquid, and approximately 500 cells were plated to YPD agar and incubated at 30uC for 3 days. The resulting colonies were counted and divided by 500 to determine plating efficiency. Viability was calculated by multiplying the total number of cell bodies by the plating efficiency. This process was repeated following each day of serial growth. The viabilities reported for each genotype after each day of serial growth were the mean values62 standard errors calculated from at least eight independent cultures. Differences between mean viabilities that exceeded two standard errors were considered statistically significant at the 95% confidence interval.

Spontaneous translocation rates
Ten ml YPD cultures were inoculated with single colonies of cells from which pSD196 or pSD120 had been lost, and grown for 24 hours at 30uC. Approximately 1610 9 cells were plated to medium lacking histidine to select for translocations, and dilutions plated to YPD medium to determine viability. Rates were determined from at least 12 independent cultures by the method of the median [80]. The 95% confidence intervals were determined using a table [81]. Differences between rates for which the 95% confidence intervals did not overlap were considered statistically significant. This process was repeated at five successive growth intervals approximately 25 generations apart using single colonies that arose on the YPD viability plates. Selected His+ recombinants were subjected to genomic Southern blot analysis to verify the presence of wild-type HIS3 sequence, and to chromosome blot analysis to verify the presence of translocations as described below.

HO-stimulated translocation frequencies
One ml cultures of synthetic complete (SC) medium containing 3% glycerol and 3% lactate were inoculated with single colonies from which pSD196 or pSD120 had been lost, and incubated for 24 hours at 30uC. An aliquot of cells was then removed and dilutions plated to YPD to determine the number of viable cells before expression of HO endonuclease. Galactose was then added to the remaining cells to a final concentration of 2% to induce expression of the HO endonuclease. After four hours of expression, the cells were plated to medium lacking histidine to select for recombinants, and dilutions plated to YPD to determine viability. Translocation frequency was determined by dividing the number of histidine prototrophic colonies by the total number of viable cells plated. Median translocation frequencies from at least 12 independent cultures per growth period and for each genotype were reported and the 95% confidence intervals calculated using a table [81]. Differences between frequencies for which the 95% confidence intervals did not overlap were considered statistically significant. This process was repeated at five successive growth intervals, approximately 25 generations apart using single colonies that arose following plating of the cells that had not been exposed to galactose on to YPD. Selected His + recombinants were subjected to genomic Southern blot and chromosome blot analyses as described below.

Plating efficiencies
Plating efficiencies of cells collected from SC glycerol/lactate cultures before and after the addition of galactose were determined as described previously [43]. One ml SC glycerol/lactate cultures were inoculated with single colonies of cells that had lost pSD196 or pSD120 and grown for 24 hours at 30uC, at which point cell number was determined by hemocytometer count. Approximately 200-500 cells were plated onto YPD and incubated at 30uC for two days at which point the number of colonies that had arisen were counted. Plating efficiency was determined before the addition of galactose by dividing the number of colonies by the number of cell bodies that had been plated, and multiplying the quotient by 100. Galactose was added to the cultures and allowed to induce expression of HO endonuclease for four hours at 30uC, at which point cell number was assessed by hemocytometer. Approximately 200-500 galactose treated cells were plated to YPD, incubated at 30uC for two days, and the number of colonies counted. Plating efficiency after the addition of galactose was determined by dividing the number of colonies by the number of cells plated and multiplying the quotient by 100. The median plating efficiencies from at least seven independent trials were reported and the 95% confidence intervals determined from a table [81]. Differences between plating efficiencies for which the 95% confidence intervals did not overlap were considered statistically significant.

Contour-clamped homogenous electric field (CHEF) analysis
Chromosomes were prepared in agaraose plugs from selected His + recombinants by an established method [82]. Chromosomes were separated on 1% agarose gels using a Bio-Rad CHEF-DRII apparatus with the following parameters: 1 st Block, 14uC, 70 second switch time, 15 hours at 6 V/cm; 2 nd Block, 14uC, 120 second switch time, 11 hours at 6 V/cm. Chromosomes were visualized by staining with 1 mg/mL ethidium bromide for 30 minites and photographed. Ethidium stained gels were irradiated with 60 mJoules of UV in a UV Stratalinker (Stratagene) to nick the DNA, and destained for 30 minutes in distilled water. Chromosomes were transferred to a positively charged nylon membrane (Hybond N + , GE Healthcare) under denaturing conditions (0.4N NaOH, 1.5M NaCl) by capillary action, and then probed with a 1.8 kb BamHI HIS3 genomic clone. The HIS3 probe extends from 469 bp upstream of the HIS3 coding sequence to 634 bp downstream. The DNA fragment was labeled with 32 P by random priming using a Megaprime DNA labeling kit (GE Healthcare). Blots were exposed to film, and the film was developed using a Konica Minolta SRX-101A processor.

Chromatin immunoprecipitation (ChIP)
Five ml SC glycerol/lactate pre-cultures cultures were inoculated with wild type and est2 mutant spore colonies carrying the his3D39 recombination substrate at the LEU2 locus on chromosome III, the his3D59 substrate at the HIS3 locus on chromosome XV, the adh4::URA3-TELVII construct at the telomere on the left arm of chromosome VII and either EST2-MYC18-TRP1, RAD52-FLAG-KAN-MX or untagged alleles then grown for 24 hours at 30uC. The pre-cultures were used to inoculate fresh 300 ml SC medium glycerol/lactate such that the OD 600 reached 0.3-0.5 by the following day. Upon reaching the appropriate cell density, 45 ml of culture were removed (0 hours of induction), pelleted and immediately subjected to a ChIP procedure described previously [74] using the following critical reagents: a Millipore ChIP Assay Kit, Pierce Halt Protease Inhibitor Cocktail, Millipore Anti-Myc monoclonal antibody, and Sigma Anti-FLAG M2 monoclonal antibody. Sufficient 20% galactose was added to the remaining cell culture to reach a final concentration of 2% in order to induce expression of the HO endonuclease. Following 0.5, 1, 2 and 4 hours of induction at 30uC, 45 ml of cells were removed, pelleted and immediately subjected to the ChIP protocol.
Briefly, the pelleted cells were exposed to formaldehyde, washed, lysed in a buffer containing SDS and protease inhibitors, and the DNA sheared to ,500 bp by sonication. The whole cell extract was subjected to immuno-precipitation by incubating with 5 mg of anti-body at 4uC overnight and shaking. The next day, protein A agarose beads were added to the samples and incubated at 4uC for three hours. The beads were then washed successively in low salt, high salt, and LiCl solutions, and finally a TE buffer. DNA/protein/anti-body complexes were eluted from the beads with elution buffer (1% SDS, 0.1M NaHCO 3 ) and the DNA/ protein cross-links reversed with the addition of NaCl to a final concentration of 100 mM and incubation at 65uC for four hours. Proteins were then digested with proteinase K, EDTA and Tris-HCL pH 6.5 at 45uC for one hour, extracted with phenol:chloroform, and the DNA precipitated with ethanol. Selected DNA sequences were amplified by PCR using the following primer sets: a 150 bp HIS3 sequence was amplified with the primers HIS3-59 . PCR products were run on 1.5% agarose gels, stained with ethidium bromide and imaged using a Typhoon 8600. The image was then analyzed and signals quantitated using the ImageQuant 5.0 software package. RAD59 and TEL PCR signals obtained by ChIP with strains containing the EST2-MYC18-TRP1 or RAD52-FLAG-KAN-MX alleles were normalized to RAD59 and TEL PCR signals obtained by ChIP with strains that have untagged EST2 and RAD59 alleles. The normalized TEL PCR signals at each time point were divided by the normalized RAD59 signals at the same time points to obtain the T/R ratios reported in Figure 3. RAD59 was chosen as the control for telomeric DNA because RAD59 lies distal to the telomere and is not on any of the chromosomes that contain an HO cut site in our strains, therefore, it should not be bound by Est2 or Rad52. The HIS3 and SAM1 PCR signals obtained by ChIP at each time point following induction of HO endonuclease expression by galactose were normalized to the HIS3 and SAM1 PCR signals obtained from cells collected before the addition of galactose. The normalized HIS3 signals at each time point following HO endonuclease expression were divided by the normalized SAM1 signals from the same time points to obtain the H/S ratios reported in Figure 3. SAM1 was chosen as the control locus for HIS3 because it lies distal to the telomere and is not on any of the chromosomes that contain an HO cut site in our strains, therefore, it should not be bound by Est2 or Rad52. The mean T/ R and H/S ratios62 standard errors from five independent time courses were reported.