This study reports an unusual ploidy-specific response to replication stress presented by a defective minichromosome maintenance (MCM) helicase allele in yeast. The corresponding mouse allele, Mcm4Chaos3, predisposes mice to mammary gland tumors. While mcm4Chaos3 causes replication stress in both haploid and diploid yeast, only diploid mutants exhibit G2/M delay, severe genetic instability (GIN), and reduced viability. These different outcomes are associated with distinct repair pathways adopted in haploid and diploid mutants. Haploid mutants use the Rad6-dependent pathways that resume stalled forks, whereas the diploid mutants use the Rad52- and MRX-dependent pathways that repair double strand breaks. The repair pathway choice is irreversible and not regulated by the availability of repair enzymes. This ploidy effect is independent of mating type heterozygosity and not further enhanced by increasing ploidy. In summary, a defective MCM helicase causes GIN only in particular cell types. In response to replication stress, early events associated with ploidy dictate the repair pathway choice. This study uncovers a fundamental difference between haplophase and diplophase in the maintenance of genome integrity.
AMONG the genetic and epigenetic changes to genomes, changes in ploidy are the most drastic, and as such, polyploidy is not tolerated by most animal species (Li et al. 2009a). A recent study of tetraploid yeast suggests that the deleterious effects of ploidy change are due to the uncoordinated scaling of the spindle pole body, spindle, and kinetochore, thus resulting in genetic instability (GIN) (Storchova et al. 2006). However, ploidy changes occur in every sexual cycle of all eukaryotes and are associated with the inclusion or exclusion of an entire set of chromosome homologs that significantly alters the DNA repair capacity. Little is known about whether DNA damage response is regulated differently in haplophase and diplophase during sexual cycles.
DNA replication stress, induced by oncogene activation, genotoxic stress, or defects in the DNA replication machinery, is believed to cause GIN that accelerates tumorigenesis (Halazonetis et al. 2008). However, DNA replication stress does not always lead to increased mutation rates or aneuploidy. In metazoans, multiple factors may affect GIN as a consequence of DNA replication stress but only in some cell types because different cells proliferate at different rates under different cellular contexts (Sarkisian et al. 2007). Therefore, it is difficult to compare directly the GIN of different cell types induced by the same replication stress, and dissect the underlying mechanisms for the differences in GIN. Saccharomyces cerevisiae is an excellent model for studying the mechanisms and pathways leading to GIN, and an often-used model for cell type-specific regulation. Yeast naturally exists in three cell types: haploids with two mating types, MATa, MATα and MATa/α diploid, which is the default state in the wild. These cell types have different properties, most of which can be attributed to the different genotypes at the mating type locus (Friis and Roman 1968; Durand et al. 1993; Galitski et al. 1999; Barbour and Xiao 2006; Valencia-Burton et al. 2006; Meyer and Bailis 2008).
Repair pathways may be distinctly regulated in different cell types. Double strand breaks (DSBs) are repaired by two main pathways, nonhomologous end-joining (NHEJ) and homologous recombination (HR), which have distinguishable mutagenic potential (Takata et al. 1998; P'ques and Haber 1999). Yeast mainly uses the HR pathway. In diploid yeast, NHEJ is severely disabled through the repression of NEJ1, a key component of NHEJ, by the transcriptional repressor, Mata1–Matα2 (Frank-Vaillant and Marcand 2001) encoded by the MATa and MATα genes. While human somatic cells use NHEJ as the main pathway to repair DSBs (Mao et al. 2008), mouse embryonic stem (ES) cells display enhanced HR capacity (Shrivastav et al. 2008). Furthermore, the choice between NHEJ and HR for DSB repair is also cell cycle regulated through CtIP/Ctp1/Sae2 (Limbo et al. 2007; Yun and Hiom 2009). However, little is known about the cell type-specific regulation of damage repair other than DSBs such as those induced by replication defects (Barbour and Xiao 2006; Shrivastav et al. 2008; Jain et al. 2009).
Mcm4 is a subunit of the hexameric MCM replication helicase (Bochman and Schwacha 2008). Mcm4Chaos3 is a cancer susceptible allele of Mcm4 that predisposes mice to mammary gland tumors (Shima et al. 2007b). Previously we showed that the effects of Mcm4Chaos3 in mice was recapitulated in diploid yeast strains bearing the corresponding mutation, allowing us to study the mechanism and consequence of replication stress-induced GIN in yeast (Li et al. 2009b). Our initial study showed that haploid yeast bearing the mcm4Chaos3 allele was grossly normal (Shima et al. 2007a). The unexpected result was that GIN and the checkpoint-dependent cell cycle delay was a diploid-specific outcome. Clearly, important diploid-specific phenotypes may have been overlooked in the past because haploid mutants are commonly used in yeast genetic studies.
In this study, we show that the mcm4Chaos3 haploid not only exhibits no cell cycle delay but also no obvious GIN, although both haploid and diploid mutants show evidence of compromised replication. We demonstrate that the different outcomes of replication stress are associated with distinct repair pathways activated in the haploid and diploid mutants. The haploid mutants use the Rad6-dependent pathways that resume stalled forks whereas the diploid mutants use the Rad52- and MRX-dependent pathways that repair double strand breaks. The repair pathway choice is regulated neither by the availability of different repair enzymes nor by MAT locus heterozygosity, but rather by ploidy itself. Distinct from the effect of geometric constraints shown in polyploids (Storchova et al. 2006), the diploid-specific defect shown here is not enhanced by increased ploidy. This study reveals a fundamental difference between haplophase and diplophase on the maintenance of genome integrity. It provides a model to study the cell type-specific GIN outcomes and the repair pathway choices in response to a cancer susceptible helicase defect.
MATERIALS AND METHODS
Yeast strains and plasmids:
Isogenic haploid W303 yeast strains mcm4Chaos3 were constructed as described (Shima et al. 2007a). All strains used in this study are in the W303 background and listed in supporting information, Table S1. The strain background carries the rad5-535 point mutation. Replacing this allele with RAD5 had little effect on ploidy difference (data not shown).
The MATa/Δ and MATΔ/α diploids were constructed by disruptions of the MAT locus using the pFP19 plasmid, a gift from Hannah Klein. Ectopic expression of NEJ1 was performed using PMV01 plasmid with the empty vector PMV04 as control; both plasmids were gifts from James Haber. The MATa/α haploid (nonmater) was constructed by transforming either MATa haploid with BE96 or MATα haploid with BE97. The BE96 and BE97 plasmids were gifts from Hannah Klein.
Flow cytometric analysis:
Approximately 1 × 107 cells were collected from log-phase cultures and processed as described (Clarke et al. 2001). DNA was stained with Sytox Green (Molecular Probes, Eugene, OR) and profiles were analyzed using a Becton Dickinson (San Jose, CA) LSR II with a 530/30BP channel filter and BD FACS DiVa software Becton Dickinson.
Growth curve and doubling time:
Saturated cell cultures were diluted 25 times in YPD medium and then grown at 30° for 24 hr. The absorbance at 600 nm was measured every 10 min using the microplate reader Tecan M200. Growth rates and doubling times were calculated by the maximum slope plotted in log scale. For each experiment where doubling times of different strains were compared, all strains were processed simultaneously in three independent trials that routinely showed variations in doubling times of <0.1 hr.
Intrachromosomal recombination assay:
Each strain carried the recombination reporter leu2-ri∷URA3∷leu2-bsteii, which had a heteroallelic duplication of LEU2, with URA3 inserted between the two LEU2 genes. Gene conversion was determined by fluctuation tests measuring Leu+ Ura+ frequency. The deletion frequency was determined by fluctuation tests measuring fluoro-orotic acid resistance rates. Each test was performed with 10 colonies and repeated twice for each strain (Xu et al. 2004).
Determination of spontaneous mutation frequency:
The forward mutation rate at the CAN1 1ocus was determined by standard methods (Sia et al. 1997), using at least 12 independent cultures for each rate estimate. Frequencies were calculated from the occurrence of canavanine-resistant mutants using the method of the median (Lea and Coulson 1949).
Measurement of gross chromosome rearrangements:
The test strains carry a marker ∼10 kb from the telomere of ChrXV-L to select for gross chromosome rearrangements (GCR) events that result from break-induced replication in repairing DSBs. The GCR rate was measured on the basis of the previously reported protocol (Kanellis et al. 2007). Ten colonies from each strain were tested, and two rounds of independent experiments were conducted.
Cell viabilities were measured by first counting log phase cells in a hemacytometer before plating in triplicate on YEPD and counting visible colonies after 3 days of growth at permissive temperatures.
Mitotic recombination assay:
A standard assay for measuring mitotic recombination and chromosome loss was used (Hartwell and Smith 1985). The test strain was heterozygous for mutations in CAN1 and HOM3, two markers located on opposite arms of chromosome V. The haploid strain with the can1 mutation was resistant to canavanine (Canr) and the hom3 strain was auxotrophic for threonine (Thr−). Heterozygous diploid strains were Cans and Thr+. Mitotic recombination was scored by the Canr Thr+ phenotype. Over 90% of the Canr strains scored were Thr+.
Pulsed-field gel electrophoresis:
Standard pulsed-field gel electrophoresis (PFGE) procedures were used according to the manufacturer's instructions (Bio-Rad). Agarose plugs containing 3 × 108 cells/ml were loaded onto a 1% agarose gel in 0.5 × Tris–borate EDTA (TBE) buffer and electrophoresed at 6 V at an angle of 120° for 22 hr at 14° with an initial switch time of 50 sec and a final switch time of 90 sec.
Unusual ploidy effect: Haploid mcm4Chaos3 mutants are grossly normal without G2/M delay or obvious GIN:
We showed previously that mcm4Chaos3/Chaos3 homozygotes and mcm4Chaos3/Δ hemizygotes display a G2/M delay prior to anaphase. The G2/M delay depended on the DNA damage checkpoint gene RAD9 (Li et al. 2009b). However, in the mcm4Chaos3 haploid no cell cycle delay was observed (Figure 1A). The haploid mutant was indistinguishable from wild type with respect to doubling time (2.66 ± 0.02 vs. 2.62 ± 0.03 hr).
We previously showed that the diploid mutant displayed severe GIN with a 100-fold increase in mitotic recombination. We investigated the recombination rate in the mcm4Chaos3 haploid strain using a recombination reporter that measured intrachromosomal gene conversion and deletions between direct repeats (Xu et al. 2004). We found that the haploid mcm4Chaos3 mutant had wild-type levels of deletion events (6.18 ± 1.96 × 10−5 vs. 8.57 ± 1.95 × 10−5) and gene conversion (1.93 ± 0.48 × 10−5 vs. 1.11 ± 0.19 × 10−5). Thus, the haploid mcm4Chaos3 mutant did not display hyperrecombination.
We next examined the mutation frequency of mcm4Chaos3 haploid using the CAN1 forward mutation assay (Kokoska et al. 2000). Haploid mcm4Chaos3 only showed a subtle mutator phenotype, with a mutation frequency (1.1 ± 0.2 × 10−6) about 2.5-fold above wild type (3.9 ± 0.1 × 10−7). The slight increase of the mutation frequency in the mcm4Chaos3 haploid prompted us to examine the potential increase of GCR frequency in the ChrXV-L GCR strain (Kanellis et al. 2007). In this assay the loss of two selectable markers, the CAN1 and URA3 genes, ∼10 kb from the telomere of ChrXV-L was measured. We observed no dramatic increase of GCR in the mutant strain (10.0 ± 0.9 × 10−8) in comparison to wild type (6.0 ± 1.0 × 10−8).
In summary, mcm4Chaos3 causes a cell cycle delay and severe GIN in diploid but not in haploid yeast. It is possible that the haploid mutant has a lower tolerance for GIN, and haploid cells experiencing mutations readily die. To test this possibility, we compared the viabilities of the haploid and diploid mutants. We found that the haploid mutant (86 ± 5%) had a better viability than the diploid mutant (59 ± 2.5%) (Li et al. 2009b). We also tested the possibility that chromosome aberrations or other mutations may suppress the growth defect in mcm4Chaos3 haploids. We examined the karyotype by genomic DNA comparative hybridization microarray (CGH), and did not find any aberrations (data not shown). We backcrossed the mcm4Chaos3 haploid to wild-type background and did not find the segregation of any growth defects (data not shown). Furthermore, defects of homozygous diploid mutants constructed by the mating of haploids were rescued by introducing a wild-type MCM4 (Li et al. 2009b). These disparate observations between the haploid and homozygous diploid mutant suggest an unusual dependence on ploidy for the manifestation of GIN stimulated by the mcm4Chaos3 allele.
mcm4Chaos3 haploid requires intact checkpoint functions for normal growth:
To investigate whether damage is also induced by mcm4Chaos3 in haploids, we constructed double mutants of mcm4Chaos3 with various checkpoint mutations. Since the defects of mcm4Chaos3 diploids are more severe at 37° (Li et al. 2009b), we used the higher temperature to impose greater stress on the haploid double mutants. If mcm4Chaos3 caused insufficient damage to activate checkpoint response, then the checkpoint response pathways would be dispensable. Conversely, if mcm4Chaos3 caused significant DNA damage in haploids, then cells with checkpoint mutations would fail to detect the damage generated by mcm4Chaos3, resulting in unrepaired damage and severe growth defects. Supporting the second possibility, mcm4Chaos3 showed synthetic growth defects with rad9Δ and mec1Δ, respectively (Figure 1, B and C), and the cell cycle distributions were also altered in the double mutants (Fig. S1A). These results indicate that mcm4Chaos3 also induces DNA damage in haploids.
Both haploid and diploid mutants suffer from replication stress:
MCM helicase has been shown to migrate with the elongation fork during DNA replication (Aparicio et al. 1997). To investigate whether mcm4Chaos3 mutant has compromised replication forks, we constructed double mutants of mcm4Chaos3 with mrc1Δ and tof1Δ, respectively. Mrc1 and Tof1 are replication fork stabilization proteins that are loaded onto DNA shortly after replication initiation and travel with the elongating fork (Katou et al. 2003). The synergistic growth defects of mcm4Chaos3 with mrc1Δ and tof1Δ (Figure 1, D–G) in both haploid and diploid suggest that mcm4Chaos3 causes replication fork defects that require fork stabilization in both cell types. However, this synthetic effect seems to be more severe in the haplophase (Figure 1, D and E) than in the diplophase (Figure 1, F and G).
Our results suggest that both haploid and diploid mutants suffer from replication stress and require the cooperation of fork stabilization proteins for survival. If both haploid and diploid mutants experience similar stress, then the ploidy difference may be caused by distinct downstream repair mechanisms that are more robust and efficient in haploid cells such that fork damages are repaired without causing GIN.
The diploid mutant requires the HR-dependent DSB repair pathway, while the haploid mutant does not:
Considering the 100-fold increase in mitotic recombination observed in the mcm4Chaos3/Chaos3 diploid strain (Li et al. 2009b), recombination-mediated replication is a likely mechanism for repairing the replication defects in the diploid mutant. To visualize the state of the chromosomes in the absence of damage repair, we placed mcm4Chaos3 into a recombination-deficient background (mcm4Chaos3/Chaso3 rad51Δ/Δ). The double mutant showed synthetic lethality at 37° (Figure 2A), arresting with about 4C DNA (Figure 2B). Fragmentation of chromosomes is observed on contour-clamped homogeneous electric fields (CHEF) gel (Figure S1B). This result indicates that homologous recombination (HR) is indispensable in the diploid for repairing DNA damage induced by mcm4Chaos3.
Three principal lesions are able to trigger spontaneous HR: DSBs, stalled replication forks, and collapsed forks (Saleh-Gohari et al. 2005). DSB recognition and kinase activation of ATM/Tel1p are mediated through the Mre11-Rad50-Xrs2 (MRX) protein complex (Costanzo et al. 2004). An increase in DSBs would result in a greater requirement for MRX function. Consistent with this idea, we found that mcm4Chaos3 and rad50Δ were synthetically lethal at 37° in the diplophase (Figure 2C). A similar effect was also observed in mcm4Chaos3 with mre11Δ (Figure S2A). Therefore, double strand break repair (DSBR) is required for the mcm4Chaos3 diploid. The requirement for DSBR, the fragmentation of chromosomes (Figure S1B), the RAD9-dependent cell cycle delay, and the increased aneuploidy (Li et al. 2009b) strongly suggest that DSBs are formed in the diploid mutant.
Another DSB repair pathway independent of HR is NHEJ, which is sequestered in diploids (Valencia et al. 2001). It is likely that DSBs are also generated in the mcm4Chaos3 haploid, but NHEJ is the more efficient pathway for preventing DSBs from translating into GIN. To test this hypothesis, we constructed double and triple mutants of mcm4Chaos3 with DSB repair mutations in haploid yeast. Dnl4 (DNA ligase IV) is a key component of NHEJ pathway (Martin et al. 1999). The double mutant of mcm4Chaos3 with dnl4Δ or rad50Δ grew as well as wild type in the haplophase (Figure 2, D and E). Disruption of both the HR and the NHEJ pathways did not show synergistic defects with mcm4Chaos3 in the triple mutant (Figure 2E), suggesting that the haploid mutant does not require DSBR under the stress of the mcm4Chaos3 mutation. The dispensability of DSBR and the absence of a cell cycle delay (Figure 1A) suggest that the haploid mutant does not experience DSBs as the diploid mutant does, but experiences damage other than DSBs.
The haploid mutant requires the RAD6-dependent stalled fork resumption pathway, while the diploid mutant does not:
We suspected that stalled forks are the primary spontaneous damage in the mcm4Chaos3 mutants. Other than HR, cells may also resume replication at stalled forks via the RAD6-dependent pathway and the novel MGS1-dependent pathway (Barbour and Xiao 2003). To dissect these repair pathways in mcm4Chaos3 haploids, we constructed double and triple mutants of mcm4Chaos3 with mutations defective in repairing stalled forks. There was no observable synthetic growth defect in mcm4Chaos3 mgs1Δ rad51Δ (Fig. S2B), though mcm4Chaos3 showed synthetic growth defect with rad6Δ (Figure 2F) (Fan et al. 1996; Ulrich and Jentsch 2000). In contrast, the diploid mutant did not require the RAD6-dependent pathway (Figure 2G). Therefore, unlike the diploid mutant that relies solely on the DSBR pathway for survival, the haploid mutant is able to resume replication of the stalled forks presumably before the stalled forks degenerate into DSBs.
Damage in the diploid mutant cannot be repaired by the NHEJ pathway:
The diploid mutant showed a fork defect that required Mrc1 and Tof1 for fork stability (Figure 1, F and G) while the haploid showed spontaneous damage that manifested as stalled forks. It is likely that the substrates that activate DSBR in the diploid mutant may not be the conventional DSBs but may have been derived from collapsed forks. A fork collapse produces a one-ended DSB that has no second end with which to rejoin, and therefore not a substrate for NHEJ pathway (Shrivastav et al. 2008). To test this hypothesis, we activated the NHEJ pathways in the diploid by ectopically expressing the NEJ1 gene, which hypothetically would rescue the lethality of the mcm4Chaos3 rad52Δ double mutant if two-ended DSBs were created (Valencia et al. 2001). However, the activation of NHEJ did not repair the damage of the mcm4Chaos3 diploid (Figure 3A). Also, we did not observe on CHEF gel DNA smears that are indicative of spontaneous DSBs in the mcm4Chaos3 diploid mutant (Figure S2C), suggesting that the damage in mcm4Chaos3 diploid mutant is not conventional DSBs but collapsed stalled forks. The repair of collapsed forks in diploid would have a less stringent requirement for fork stabilizing proteins compared to stalled forks in haploid (Figure 1, D–G). Thus, stalled forks and collapsed forks (or one-ended DSBs) may be two outcomes of the replication stress caused by mcm4Chaos3 and they activate different downstream repair pathways (Figure 3E).
In summary, haploid and diploid yeast use distinct repair pathways to restore the replication defects created by mcm4Chaos3, resulting in dichotomous outcomes of GIN. What determines the choice of different repair pathways? Is it determined by the availability of repair pathway on an ad hoc basis or by an upstream-regulated process? To investigate whether haploid and diploid mutants could be coerced into using alternative repair pathways to repair their fork defects, we altered the availability of repair pathways in the haplophase and diplophase in the next set of experiments.
The choice of repair pathway is not reversible or determined by the availability of repair proteins:
The fork resumption pathway used in the mcm4Chaos3 haploid is also available in diploid (Barbour and Xiao 2006). The fact that mcm4Chaos3 is synthetically lethal with rad52Δ (Figure 3A) indicates that the damage generated in the diplophase could not be channeled to other pathways for repair when the HR pathway was blocked. Thus, the choice of using the HR pathway was not regulated by the availability of repair pathways.
Two DNA helicases, Sgs1 and Srs2, regulate the repair pathways that resume stalled forks (Barbour and Xiao 2003). Sgs1 and Srs2 process recombination intermediates formed during fork stalling and channel the aberrant structures into the RAD6-dependent pathway for repair. Indeed, as previously reported, sgs1Δ and srs2Δ appear to cause rampant recombination in haploids (Figure 3, B and C) (Gangloff et al. 2000). To investigate whether HR pathway is able to repair the DNA damage in the mcm4Chaos3 haploid, we introduced sgs1Δ and srs2Δ, respectively, into the haploid mutant strain to channel damage repair from the fork resumption pathway to HR. The mcm4Chaos3 srs2Δ haploid mutant showed synthetic growth defects at the restrictive temperature (Figure 3B). The effect of the mcm4Chaos3 and sgs1Δ double mutant was even more dramatic, showing synthetic lethality at the restrictive temperature (Figure 3C, see similar effect in mcm4Chaos3 top3Δ, Figure S2D) and arresting at late S or G2/M phase (Figure 3D) with fragmented chromosomes (Figure S1B). This synthetic lethality is suppressed by the deletion of RAD51 (Figure 3C), suggesting that repairing the DNA damage substrate created in sgs1Δ mcm4Chaos3 haploid by HR may be the cause of lethality. Thus, although the HR pathway is available, this pathway is unable to repair the DNA damage in haploid.
In summary, haploid and diploid mutants use distinct and noninterchangeable repair pathways to repair their fork defects (Figure 3E). (The dissection of additional repair pathways is shown in Figure S3.) This distinct and exclusive choice of repair pathways is most likely regulated by some upstream mechanism that is associated with haploid and diploid identities. The two obvious determinants for cell-type identity are mating type heterozygosity and ploidy itself.
The diploid-specific effect is not due to MAT locus heterozygosity:
Diploid yeast strains are more resistant than haploid strains to γ-rays, UV, and methyl methanesulphonate (MMS). This resistance is partly due to heterozygosity at the MAT locus (Heude and Fabre 1993; Barbour and Xiao 2006). The effect of MAT heterozygosity on increased resistance to DNA damage agents is dependent on the function of HR proteins (Saeki et al. 1980). To investigate whether the unusual diploid specificity of the mcm4Chaos3 phenotype is due to heterozygosity at the MAT locus, we constructed the double mutant mcm4Chaos3 rad51Δ in a haploid with MAT heterozygosity. The mcm4Chaos3 rad51Δ diploid was lethal at the restrictive temperature (Figure 2A) while the haploid double mutant was grossly normal (Figure 4A). Notably, the MATa/α mcm4Chaos3 rad51Δ haploid did not show any synthetic effect (Figure 4A). Therefore, the diploid-specific defect is not due to MAT heterozygosity.
To investigate further the role of MAT heterozygosity in this diploid-specific defect, we constructed MATa/Δ and MATΔ/α diploids. The mcm4Chaos3 diploid that was hemizygous at the MAT locus showed chromosome fragmentation (Figure S1B) and was inviable at the restrictive temperature (Figure 4B), indicating that the damage could no longer be repaired. The observation that the growth defect was worse in diploids with hemizygous MAT than with heterozygous MAT suggests that the diploid-specific growth defects are not due to MAT heterozygosity. Rather, MAT heterozygosity is required for viability in mcm4Chaos3 diploid presumably because of its role in upregulating the HR pathway, as previously reported (Valencia-Burton et al. 2006).
To investigate the effect of MAT heterozygosity in repair pathway choice in the mcm4Chaos3 diploid, we measured the loss of heterozygosity (LOH) frequency of CAN1 with respect to HOM3 on the left arm of chromosome V (Hartwell and Smith 1985) at the permissive temperature. There was little difference in the LOH frequency between MATa/α mcm4Chaos3/Chaos3 (2.60 ± 1.60 × 10−3) and MATΔ/α mcm4Chaos3/Chaos3 (1.02 ± 0.49 × 10−3) strains, which was ∼100-fold elevated over that of the wild type (2.12 ± 0.11 × 10−5) (Li et al. 2009b). Nearly all LOH events in MATΔ/α mcm4Chaos3/Chaos3 were also due to mitotic recombination as they were in MATa/α mcm4Chaos3/Chaos3 (Li et al. 2009b). Even when the ability to perform HR was compromised in the MATΔ/α background (Figure 4B) (Valencia-Burton et al. 2006), the damage was still committed to HR repair independent of MAT heterozygosity, consistent with the previous observation (Figure 3E) that the repair pathway choice is not regulated by the availability of repair enzymes.
In sum, the diploid-specific defects and the repair pathway choice are not determined by MAT heterozygosity or the availability of repair enzymes, but by the ploidy.
The diploid-specific defects do not increase proportionally with ploidy:
The diploid-specific defect could be the effect of an increase in spontaneous DNA damage associated with the replication of an extra chromosome set or the geometric constraints in scaling the mitotic spindle with increased ploidy (Storchova et al. 2006). To investigate these possibilities, we constructed the triploid and tetraploid mutant and wild-type strains by mating MATa/Δ diploid with MATα haploid or MATΔ/α diploids. We found that none of the diploid-specific defects such as G2/M delay (Figure 5A), growth defect (Figure 5B), or increased doubling time (Figure 5C) are significantly enhanced with increased ploidy. We also constructed triploids and tetraploid mutants containing only one copy of MATa or MATα (with disrupted MAT heterozygosity). Similar to the diploid mutant, heterozygosity of the MAT locus was required for viability (Figure 5D), suggesting that the triploid and tetraploid cells may also use HR to repair the mcm4Chaos3-induced DNA damage. This result suggests that the ploidy effect in mcm4Chaos3 mutant either is not due to geometric scaling (Storchova et al. 2006) or has reached the threshold in the diploid.
Choice of repair pathway is under active control rather than passive shunting:
Cells develop multiple pathways to repair particular types of DNA damage. These pathways are distinct with regard to repair efficiency and mutagenic potential and must be tightly controlled to preserve viability and genomic stability. In this study, although the fork resumption pathways and the DSB repair pathways are available in both haploid and diploid, the choice is dictated by ploidy. We have shown that the haploid and diploid mutants are unable to use each other's designated repair pathways to repair their replication defects (Figure 3) and that these repair pathways do not randomly compete for substrates on an ad hoc basis.
There is substantial evidence that the choice of repair pathway may be passively shunted in particular cell types on the basis of availability (Gudmundsdottir et al. 2007). Why did our study reach a different conclusion from previous studies? We believe that the disparity lies in the difference in the initial processing of the spontaneous damage generated in mcm4Chaos3 haploid and diploid. These processed structures define the substrates for the repair pathway and dictate the repair pathway choice. In the diploid mutant, the spontaneous damage seems to be processed to DSBs, while the same damage is rescued directly without degenerating into DSBs in the haploid (Figure 3E). Our results reveal a complex upstream regulation of repair pathways under replication stress, which may be at least as complex as the end processing of DSBs (Mimitou and Symington 2008).
Cell type-specific GIN:
Cell type-specific GIN is often a shunned topic because of its complexity. The mcm4Chaos3 yeast provides a tangible model to dissect the underlying causes. The haploid and diploid yeast are congenic (except for the mating type locus), can be grown in the same conditions, and suffer similarly from endogenous replication defects. These inherent properties allow us to investigate the difference in DNA damage response specific to cell-type identity between haploids and diploids. In diploid yeast, we previously showed that mcm4Chaos3 resulted in LOH and that a hypermutable subpopulation gained new traits such as aneuploidy and improved growth (Li et al. 2009b). Here, we showed that the haploid mcm4Chaos3 mutant had no growth defect or obvious GIN, although it also suffered from replication stress. This difference in outcome is associated with the regulation of repair pathways in different cell types. In the diploid mutant, the error prone HR repair pathway seemed to be the cause of hyperrecombination and aneuploidy, while in the haploid, the RAD6-dependent pathway repaired fork damage with sufficient fidelity without causing GIN.
HR is generally less mutagenic than NHEJ (Takata et al. 1998; P'ques and Haber 1999). However, in repairing mcm4Chaos3-induced DNA damage, HR in diploid is much more error prone than the fork resumption pathway in haploid. This difference in GIN may be due not to the repair pathway used but to the manner in which the initial damage is processed. It is known that DSBs pose the greatest challenge for cells in maintaining genome integrity. While replication defects in the diploid mutant are processed to DSBs, the haploid mutant managed to avoid the formation of DSBs. When DSBR is operating at a high level (100-fold above normal), even the HR pathway will generate severe GIN.
Haploid and diploid yeast have fundamental differences:
The dichotomous response to replication stress indicates that there are fundamental differences between haploid and diploid yeast. In the mcm4Chaos3 diploid, MAT heterozygosity confers the availability of the HR repair pathway (Figure 4B), but does not obligate the commitment of this repair pathway. Interestingly, under replication stress ploidy determines the repair pathway choice and the consequent GIN. This ploidy effect may have evolved as another layer of regulation to ensure genome integrity during changes in ploidy associated with the sexual cycle. The diploid-specific GIN and cell cycle delay seem to be unique to the mcm4Chaos3 allele, but have not been observed in other genotoxic stresses (data not shown) (Heude and Fabre 1993; Barbour and Xiao 2006). This apparently unique response could either be due to the underinvestigation of replication stress in diploids or the challenges of a specific helicase defect. In particular, defective replicative helicases are not known to expose single strand DNA and may have to activate the intra-S phase checkpoint differently (Branzei and Foiani 2005).
If the ploidy effect observed in this study is not due to geometric scaling, it may act through the dosage changes of certain genes. In yeast, ploidy-regulated gene expression has been observed for the G1 cyclins, Flo11 (Galitski et al. 1999), components of the cell wall (De Godoy et al. 2008), and cell surface proteins (Wu et al. 2010). However, these genomic scale comparisons between haploids and diploids were conducted in normal conditions (Galitski et al. 1999; De Godoy et al. 2008; Wu et al. 2010) and therefore not designed to identify the genes responsible for ploidy-specific stress response. The mechanisms that regulate homologous recombination for DSBR are conserved from yeast to human (Hartwell and Smith 1985; P'ques and Haber 1999; Limbo et al. 2007; Yun and Hiom 2009). We believe that the mechanisms that regulate repair pathway choices in response to replication stress may also be evolutionarily conserved.
We thank Eric Alani and John Schimenti for discussions. This work is supported by National Institutes of Health GM-072557 awarded to B.K.T.
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.110.125450/DC1.
↵1 Present address: Department of Biochemistry and Molecular Pharmacology, University of Massachusetts Medical School, Worcester, MA 01605.
Communicating editor: F. Winston
- Received November 27, 2010.
- Accepted January 11, 2011.
- Copyright © 2011 by the Genetics Society of America