Saccharomyces cerevisiae chromatin-assembly factors that act during DNA replication function in the maintenance of genome stability

Abstract

Some spontaneous gross chromosomal rearrangements (GCRs) seem to result from DNA-replication errors. The chromatin-assembly factor I (CAF-I) and replication-coupling assembly factor (RCAF) complexes function in chromatin assembly during DNA replication and repair and could play a role in maintaining genome stability. Inactivation of CAF-I or RCAF increased the rate of accumulating different types of GCRs including translocations and deletion of chromosome arms with associated de novo telomere addition. Inactivation of CAF-I seems to cause damage that activates the DNA-damage checkpoints, whereas inactivation of RCAF seems to cause damage that activates the DNA-damage and replication checkpoints. Both defects result in increased genome instability that is normally suppressed by these checkpoints, RAD52-dependent recombination, and PIF1-dependent inhibition of de novo telomere addition. Treatment of CAF-I- or RCAF-defective cells with methyl methanesulfonate increased the induction of GCRs compared with that seen for a wild-type strain. These results indicate that coupling of chromatin assembly to DNA replication and DNA repair is critical to maintaining genome stability.

Maintaining the stability of the genome is crucial for cell survival and normal cell growth. The presence of specific genome rearrangements and the ongoing accumulation of genome rearrangements are seen in many types of cancer cells (1–5). Similarly, the inheritance of genome rearrangements underlies other human genetic diseases (6, 7). Studies with model organisms have identified multiple mechanisms by which genome rearrangements arise and multiple pathways that act to maintain genome stability (8). A number of studies have suggested that spontaneous genome rearrangements result from errors during DNA replication that possibly lead to stalled or broken replication forks (9–15). A number of pathways seem to act on these errors to promote their correct repair and prevent their conversion into genome rearrangements (8, 16). Among these pathways are (i) at least three different checkpoints that act during S phase, (ii) recombination pathways similar to those that promote break-induced replication, (iii) a pathway that prevents de novo addition of telomeres to broken DNAs, and (iv) possibly mismatch repair that prevents recombination between divergent DNA sequences (8). An alternative source of DNA damage that can lead to genome instability is degradation of telomeres in the absence of telomerase (8, 16, 17); these degraded chromosomes seem to be acted on by many of the same pathways that have been suggested to act on DNA-replication errors. Interestingly, the human homologs of many of the Saccharomyces cerevisiae genes that function in suppression of genome instability and human genes encoding interacting proteins have been implicated in pathways that suppress the development of cancer (18–28).

DNA replication and chromatin assembly are coordinated, and in human cells S phase is also not completed in the absence of chromatin assembly (29–32). At least two chromatin-assembly complexes (CACs), chromatin-assembly factor I (CAF-I) and replication-coupling assembly factor (RCAF), function in the assembly of chromatin linked to DNA synthesis (33–39). CAF-I is a three-subunit complex, consisting of CAC1–CAC3, in S. cerevisiae, that interacts with histones H3 and H4 and targets them to DNA through an interaction between CAC1 and proliferating cell nuclear antigen (PCNA) (34, 40–42). CAF-I can promote the assembly of nucleosomes in vitro (33, 36, 40). Mutations that inactivate CAF-I cause defects in silencing at telomeres and mating-type loci as well as mild sensitivity to UV irradiation but do not effect viability or generally cause sensitivity to other DNA-damaging agents or hydroxyurea (36, 40). Recently it was shown that expression of a dominant-negative CAC1 protein in mammalian cells seems to induce DNA damage during S phase and activation of S-phase checkpoints (31). Interestingly, several genetic studies have suggested that CAC1–CAC3 may also have independent functions (43–47). RCAF consists of antisilencing function 1 (ASF1) and histones H3 and H4 and can also promote assembly of nucleosomes in vitro (36). RCAF interacts with CAF-I through an interaction between ASF1 and CAC2, and biochemical studies have indicated that RCAF and CAF-I cooperate in in vitro chromatin-assembly assays (36, 48, 49). Mutations in ASF1 cause weak defects in silencing as well as a broader range of defects than CAF-I defects including slow growth and sensitivity to a broader range of DNA-damaging agents and hydroxyurea (36, 50). The slow-growth phenotype of asf1 mutants is associated with a defect in transiting S phase, and asf1 mutants show a defect in response to hydroxyurea treatment similar to that seen in checkpoint-defective mutants (36, 51, 52). Consistent with this phenotype, ASF1 physically interacts with the checkpoint protein RAD53, and asf1 mutations cause partial defects in some checkpoint responses to hydroxyurea (51, 53). Inactivation of both CAF-I and RCAF results in stronger defects than inactivation of either complex alone (36, 50, 54), which suggests that CAF-I and RCAF have distinct roles in chromatin assembly as well as likely a joint role implied by the physical interaction between the two complexes.

Because chromatin assembly is important for DNA replication and repair, correct chromatin assembly may also be important for maintaining genome stability. Because defects in CAF-I and RCAF cause defects in S-phase and/or DNA-repair defects, we tested whether mutations in the genes encoding these two chromatin-assembly factors cause genome instability. The results reported here indicate that CAF-I and RCAF defects cause the accumulation of DNA damage resulting in increased rates of accumulating genome rearrangements. Our results also support the view that CAF-I and RCAF play different roles in suppression of genome instability, consistent with the view that these complexes have both distinct and common roles in chromatin assembly.

Experimental Procedures

General Genetic Methods. Media for propagation of strains and determining gross chromosomal rearrangement (GCR) rates were as described (11, 55). All S. cerevisiae strains were propagated at 30°C except for rfc5-1 and dpb11-1 mutants, which were grown at 23°C. Gene disruptions all were made by standard PCR-based gene-disruption methods, and correct gene disruptions were verified by PCR as described (10). The hta1S129STOP and hta2S129STOP mutations were inserted by using pop-in pop-out plasmids and verified by sequence analysis (56). The sequences of primers used to generate disruption cassettes and confirm disruption of indicated genes are available on request. All strains were derived from the S288c parental strain RDKY3023 (MATa, ura3-52, leu2Δ1, trp1Δ63, his3Δ200, lys2ΔBgl, hom3-10, ade2Δ1, ade8) and in addition contained the hxt13::URA3 insertion used in the GCR assay. Relevant genotypes of these strains are RDKY3615 wild type; RDKY3733 sml1::KAN; RDKY4753 cac1::TRP1; RDKY5003 cac2::HIS3; RDKY5005 cac3::HIS3; RDKY5001 cac1::TRP1 cac2::HIS3; RDKY5009 cac1::TRP1 cac3::HIS3; RDKY5007 cac2::TRP1 cac3::HIS3; RDKY4755 asf1::HIS3; RDKY4779 cac1::TRP1 asf1::HIS3; RDKY5011 cac2::TRP1 asf1::HIS3; RDKY5013 cac3::HIS3 asf1::TRP1; RDKY3719 rad9::HIS3; RDKY3723 rad24::HIS3; RDKY3814 sgs1::HIS3; RDKY3727 rfc5-1; RDKY4538 dpb11-1; RDKY4757 rad9::HIS3 cac1::TRP1; RDKY4761 rad24::HIS3 cac1::TRP1; RDKY4765 cac1::TRP1 sgs1::HIS3; RDKY4769 rfc5-1 cac1::HIS3; RDKY4773 dpb11-1 cac1::HIS3; RDKY4759 rad9::HIS3 asf1::TRP1; RDKY4763 rad24::HIS3 asf1::TRP1; RDKY4767 asf1::HIS3 sgs1::TRP1; RDKY4771 rfc5-1 asf1::HIS3; RDKY4775 dpb11-1 asf1::HIS3; RDKY4343 pif1-m2; RDKY4224 tlc1::TRP1; RDKY3735 sm1l::KAN mec1::HIS3; RDKY3731 tel1::HIS3; RDKY3739 dun1::HIS3; RDKY3745 chk1::HIS3; RDKY4781 pif1-m2 cac1::TRP1; RDKY4785 cac1::TRP1 tlc1::HIS3; RDKY4789 sml1::KAN mec1::HIS3 cac1::TRP1; RDKY4793 tel1::HIS3 cac1::TRP1; RDKY4801 dun1::HIS3 cac1::TRP1; RDKY4807 chk1::HIS3 cac1::TRP1; RDKY4783 pif1-m2 asf1::HIS3; RDKY4787 asf1::HIS3 tlc1::TRP1; RDKY4791 sml1::KAN mec1::TRP1 asf1::HIS3; RDKY4795 tel1::HIS3 asf1::TRP1; RDKY4803 asf1::HIS3 dun1::TRP1; RDKY4809 asf1::HIS3 chk1::TRP1; RDKY4857 hta1 S129STOP hta2 S129STOP; RDKY3641 lig4::HIS3; RDKY3640 yKU80::HIS3; RDKY4421 rad52::HIS3; RDKY4823 hta1 S129STOP hta2 S129STOP cac1::TRP1; RDKY4827 lig4::HIS3 cac1::TRP1; RDKY4831 cac1::TRP1 yKU80::HIS3; RDKY4843 cac1::TRP1 rad52::HIS3; RDKY4825 hta1 S129STOP hta2 S129STOP asf1::HIS3; RDKY4829 lig4::HIS3 asf1::TRP1; RDKY4833 yKU80::HIS3 asf1::TRP1; RDKY4855 asf1::HIS3 rad52::TRP1; RDKY4853 asf1::HIS3 rad59::TRP1; RDKY4861 sml1::KAN cac1::TRP1; and RDKY5075 cac1::TRP1 cac2::HYG cac3::HIS3. We were unable to construct a cac1 cac3 asf1 triple mutant. In selected cases, we tested whether ARS CEN plasmids expressing either ASF1 or CAC1 under control of their native promoters complemented asf1- and cac1-induced GCR rates, respectively, to ensure that observed increased GCR rates were not due to accumulation of second-site mutations.

Characterization of GCR Rates and Breakpoints. All GCR rates were determined independently by fluctuation analysis two or more times by using either 5 or 11 cultures, and the average value is reported (11, 55). Statistical significance was evaluated by the Mann–Whitney test by using programs available at http://faculty.vassar.edu/lowry/vshome.html. The effect of methyl methanesulfonate (MMS) treatment on cell survival and GCR frequency was determined as described (57). The sequences of independent rearrangement breakpoints were determined and classified as described (11, 55).

Results

Chromatin-Assembly Factors Suppress Genome Instability. Mutations in each of the three genes encoding components of CAF-I were tested for their effect on the GCR rate (Table 1). A cac1 mutation increased the GCR rate by 340-fold, and the resulting GCRs were a mixture of de novo telomere additions and translocations with microhomology or nonhomology breakpoints (Table 2). In contrast, mutations in the CAC2 and CAC3 genes encoding the other two subunits of CAF-I increased the GCR rate but not to the same level as that caused by a cac1 mutation (P = 0.0003 and 0.0005, respectively). Interestingly, compared with the increased GCR rate caused by a cac1 mutation, the cac1 cac2 double-mutant strain had a decreased GCR rate (P = 0.0036), and the cac1 cac3 double-mutant strain had an increased GCR rate (P = 0.0001). The cac2 cac3 double-mutant strain had a GCR rate that was not different from that caused by a cac2 mutation (P = 0.42). The cac1 cac2 cac3 triple mutant had a GCR rate that was similar to the GCR rate of the cac2 and cac3 single mutants (P = 0.43 and 0.25, respectively) and the asf1 (see below; P = 0.1) mutant; this GCR rate was reduced compared with the GCR rate of either the cac1 or cac1 cac3 mutant strains. One interpretation of these results is that the increased GCR rate of the cac1 single mutant and the cac1 cac3 double mutant may in part be due to the aberrant activity of subcomplexes of CAF-I containing CAC2 and possibly ASF1 (see below). It should be noted that other studies have also suggested that the different CAF-I subunits may also have distinct functions in addition to a shared function implied by their presence in the CAF-I complex (43,–47), and defects in these different functions could also contribute to the differences between the effects of the cac1, cac2, and cac3 mutations seen here.

Table 1. Effect of CAF-I and RCAF defects on the rate of accumulating GCRs.

	Wild type		asf1Δ
Relevant genotype	Strain no.	Mutation rate (Canr 5-FOAr)	Strain no.	Mutation rate (Canr 5-FOAr)
Wild type	3615	3.5 × 10-10 (1)	4755	2.5 × 10-8 (71)
cac1Δ	4753	1.2 × 10-7 (343)	4779	3.9 × 10-8 (111)
cac2Δ	5003	3.0 × 10-8 (87)	5011	1.5 × 10-8 (43)
cac3Δ	5005	1.1 × 10-8 (32)	5013	1.2 × 10-8 (35)
cac1Δ cac2Δ	5001	5.1 × 10-8 (145)		ND
cac1Δ cac3Δ	5009	6.9 × 10-7 (1,997)		ND
cac2Δ cac3Δ	5007	3.4 × 10-8 (97)		ND

The numbers in parentheses are the fold increases in rate relative to that of the wild-type strain. The GCR rate with RDKY5075 (cac1Δ cac2Δ cac3Δ) was 1 × 10^-8 (29). ND, not determined.

Table 2. Structure of observed rearrangement breakpoints generated from strains defective in CAF-I and RCAF.

Relevant genotypes	Strain no.	Telomere addition	Translocation/deletion, non, micro
Wild type*	3615	5	1, 0
cac1Δ sml1Δ	4861	6	2, 2
cac1Δ	4753	7	2, 1
asf1Δ	4755	9	0, 2
cac1Δ asf1Δ	4779	4	2, 4
tel1Δ*	3731	0	0, 6
mec1Δ sml1Δ*	3735	9	0, 0
tel1Δ cac1Δ	4793	2	3†, 4
mec1Δ sml1Δ cac1Δ	4789	9	0, 0
lig4Δ‡	3641	6	0, 0
cac1Δ lig4Δ	4827	20	0, 0

“non, micro” indicates the number of nonhomology and microhomology breakpoints observed, respectively.

^*Data are from Myung et al. (55).

^†One case was a deletion that had 143 bp of non-chromosome V sequences inserted at the breakpoint.

^‡Data are from Myung et al. (16).

To analyze the role of the RCAF complex in suppression of genome instability, a mutation in the ASF1 gene was tested for its effect on the GCR rate. An asf1 mutation increased the GCR rate by 70-fold (Table 1), and the resulting GCRs were a mixture of mainly de novo telomere-addition GCRs and a low proportion of translocations with microhomology breakpoints (Table 2). The GCR rate of the asf1 cac1 double-mutant strain was reduced relative to that observed in the cac1 single-mutant strain (P = 0.012) but was not significantly different than the GCR rate of the cac1 cac2 double mutant (P = 0.32). The GCR rate of the asf1 cac2 double mutant was somewhat less than but not significantly different than that of either the asf1 or cac2 single mutants (P = 0.13 and 0.5, respectively); these results are consistent with the observation that ASF1 and CAC2 interact and may function together (48, 49). These results further support the idea that the increased GCR rate of the cac1 single mutant may in part be due to the aberrant activity of ASF1 as well as CAC2 (see above). The GCR rate of the asf1 cac3 double mutant was not different than that of the cac3 mutant (P = 0.41) and somewhat less than that of the asf1 mutant (P = 0.028).

cac1 and asf1 Mutations Seem to Activate Different Checkpoints. The increased GCR rate caused by cac1 and asf1 mutations could occur if defects in chromatin assembly due to the cac1 and asf1 defects cause either replication defects or damage to the newly replicated DNA. If this is true, cac1 and asf1 mutations might show synergistic interactions with mutations that cause defects in the replication and DNA-damage checkpoints that act during S phase to suppress spontaneous GCRs (55, 57). To investigate this possibility, the effect of combining cac1 and asf1 mutations with mutations that inactivate different checkpoint sensor functions was tested (Table 3). Combining a cac1 mutation with mutations that inactivate different DNA-damage and intra-S DNA-damage checkpoint functions (rad9, rad24, and sgs1) resulted in a synergistic increase in the GCR rate. However, combining a cac1 mutation with either rfc5-1 or dpb11-1 mutations that cause defects in the replication checkpoint did not increase the GCR rate above that caused by the cac1 mutation (P = 0.13 and 0.1, respectively). In contrast, when an asf1 mutation was combined with rad9, rad24, sgs1, rfc5-1, or dpb11-1 mutations, a synergistic increase in the GCR rate was observed. We do not know why the rfc5-1 and dpb11-1 mutations showed somewhat different effects when combined with the asf1 mutation, although we note that neither of these mutations is a complete loss-of-function mutation. These results suggest that the errors that lead to increased GCR rates in a cac1 mutant are recognized by the DNA-damage checkpoints, whereas the errors caused by an asf1 mutation are recognized by both the DNA-damage and replication checkpoints.

Table 3. Interaction between CAF-I or RCAF defects and checkpoint defects.

	Wild type		cac1Δ		asf1Δ
Relevant genotype	Strain no.	Mutation rate (Canr 5-FOAr)	Strain no.	Mutation rate (Canr 5-FOAr)	Strain no.	Mutation rate (Canr 5-FOAr)
Wild type	3615	3.5 × 10-10 (1)*	4753	1.2 × 10-7 (343)	4755	2.5 × 10-8 (71)
rad9Δ	3719	2.0 × 10-9 (6)*	4757	4.0 × 10-7 (1,142)	4759	1.0 × 10-7 (285)
rad24Δ	3723	4.0 × 10-9 (11)*	4761	4.8 × 10-7 (1,371)	4763	2.0 × 10-7 (571)
sgs1Δ	3814	7.7 × 10-9 (22)†	4765	6.3 × 10-7 (1,800)	4767	9.7 × 10-8 (278)
rfc5-1	3727	6.6 × 10-8 (189)*	4769	1.3 × 10-7 (371)	4771	8.6 × 10-7 (2,457)
dpb11-1	4538	9.0 × 10-8 (257)*	4773	1.2 × 10-7 (342)	4775	2.3 × 10-7 (657)
mec1Δ sml1Δ‡	3735	6.8 × 10-8 (194)*	4789	5.2 × 10-7 (1,486)	4791	1.7 × 10-7 (486)
dun1Δ	3739	7.3 × 10-8 (208)*	4801	5.6 × 10-7 (1,600)	4803	6.9 × 10-8 (197)
chk1Δ	3745	1.3 × 10-8 (37)*	4807	8.0 × 10-7 (2,285)	4809	6.9 × 10-8 (197)

The numbers in parentheses are the fold increases in rate relative to that of the wild-type strain.

^*Data are from Myung et al. (55).

^†Data are from Myung et al. (57).

^‡GCR rate of the sml1 strain was 3.1 × 10^-10 (1).

cac1 and asf1 Mutations Interact Differently with Mutations in Genes Encoding Components of the Checkpoint Signal Transduction Cascade. Signals generated by activation of different cell-cycle checkpoints are transduced to effector proteins by a signal transduction cascade (52). To determine which checkpoint transducer functions play a role in suppressing cac1-induced GCRs, double mutants containing mutations in the CAC1 gene and genes encoding transducer proteins were analyzed (Table 3). When a cac1 mutation was combined with mec1, dun1,or chk1 mutations, a synergistic increase in the GCR rate was observed. The rearrangement breakpoints formed in the cac1 mec1 strain were all de novo telomere-addition events (Table 2) consistent with previous observations that mec1 mutations result in a large increase in the rate of de novo telomere additions (55). This result is consistent with the idea that a checkpoint involving MEC1, DUN1, and CHK1 acts in suppressing cac1-induced GCRs. This role of MEC1, DUN1, and CHK1 parallels their role in the DNA-damage checkpoints and the observed interaction between cac1 mutations and intra-S DNA-damage checkpoint sensor-defective mutations.

When an asf1 mutation was combined with either a mec1 or chk1 mutation, a modest synergistic increase in the GCR rate was observed (Table 3). In contrast, the GCR rate of the dun1 asf1 strain was not increased compared with that of the dun1 strain (P = 0.07). This suggests that the MEC1 CHK1 signal transduction cascade branch is most important in suppression of asf1-induced GCRs. The lack of an interaction between asf1 and dun1 mutations could suggest that this signal transduction cascade branch is less important for suppressing asf1-induced GCRs. However, it is known that asf1 mutations cause a partial checkpoint defect in response to hydroxyurea including a defect in activating the RAD53 kinase and, by inference, potentially a defect in activating DUN1 because it lies downstream of RAD53 in regard to some checkpoint responses (36, 51, 58). Thus, an alternative explanation for the lack of an interaction between asf1 and dun1 mutations is that asf1 mutations significantly inactivate the RAD53 DUN1 checkpoint branch. We previously suggested that three different checkpoints function to suppress genome instability and that each functions through distinct but partially overlapping signal transduction cascade components (55, 57). Thus, an alternative explanation for the limited interaction between an asf1 mutation and mec1, chk1, and dun1 mutations is that an asf1 defect may result in damage that activates multiple checkpoints and hence multiple signal transduction cascades. Consequently, inactivating only one signal transduction cascade may cause little defect in suppressing the genome instability induced by an asf1 mutation.

cac1 and asf1 Mutations Interact Differently with Mutations in Genes Encoding Telomere Maintenance Functions. Combining a tel1 mutation with a cac1 mutation resulted in an ≈2- to 3-fold reduction of the GCR rate compared with a cac1 single mutant (Table 4). Analysis of GCR breakpoints formed in the cac1 tel1 strain revealed a decrease in the proportion of de novo telomere-addition events, although the number of events analyzed was small (Table 2). Calculating the rate of accumulating de novo telomere additions and translocations in the cac1 tel1 strain suggested that inactivation of TEL1 exclusively reduced the rate of de novo telomere additions. This result indicates that a TEL1-dependent pathway is important for the formation of de novo telomere-addition GCRs, consistent with the role of TEL1 in telomere maintenance (16, 55, 59). As predicted by this observation, combining a pif1-m2 mutation (16, 60) with a cac1 mutation resulted in a synergistic increase in the GCR rate, whereas combining a tlc1 mutation (61) with a cac1 mutation reduced the GCR rate, which indicates that GCR formationinthe cac1 strain by de novo telomere addition is suppressed by PIF1 and requires telomerase.

Table 4. Interaction between CAF-I or RCAF defects and defects in telomere maintenance functions or DSB repair.

	Wild type		cac1Δ		asf1Δ
Relevant genotype	Strain no.	Mutation rate (Canr 5-FOAr)	Strain no.	Mutation rate (Canr 5-FOAr)	Strain no.	Mutation rate (Canr 5-FOAr)
Wild type	3615	3.5 × 10-10 (1)†	4753	1.2 × 10-7 (343)	4755	2.5 × 10-8 (71)
tel1Δ	3731	2.0 × 10-10 (0.6)†	4793	6.4 × 10-8 (183)	4795	2.0 × 10-9 (5.6)
pif1-m2	4343	8.3 × 10-8 (237)‡	4781	3.5 × 10-7 (1,000)	4783	2.7 × 10-7 (771)
tlc1Δ	4224	3.1 × 10-10 (0.9)‡	4785	2.0 × 10-9 (6)	4787	1.3 × 10-6 (3,700)
hta1 S129*hta2 S129*	4857	4.4 × 10-10 (1.3)	4823	9.0 × 10-9 (26)	4825	2.6 × 10-8 (74)
lig4Δ	3641	1.6 × 10-9 (5)‡	4827	2.1 × 10-7 (600)	4829	2.4 × 10-8 (69)
yKu80Δ	3640	7.8 × 10-10 (2)‡	4831	9.3 × 10-10 (2.7)	4833	1.0 × 10-9 (2.9)
rad52Δ	4421	4.4 × 10-8 (126)‡	4843	3.0 × 10-7 (857)	4855	4.4 × 10-7 (1,257)

The numbers in parentheses are the fold increases in rate relative to that of the wild-type strain.

^†Data are from Myung et al. (55).

^‡Data are from Myung et al. (16).

Combining an asf1 mutation with a tel1 mutation resulted in a large decrease in the GCR rate, indicating that a major proportion of the GCRs that occur in an asf1 mutant are TEL1-dependent. Because most of the GCRs in an asf1 mutant are telomere additions, this could reflect the role of TEL1 in telomere maintenance. Consistent with this hypothesis, combining an asf1 mutation with a pif1-m2 mutation resulted in a large increase in the GCR rate, indicating that most of the GCRs in an asf1 mutant are suppressed by PIF1 (16). Surprisingly, combining an asf1 mutation with a tlc1 mutation resulted in a large increase in the GCR rate rather than the predicted decrease in GCR rate if asf1-induced telomere additions require telomerase. It seems unlikely that the de novo telomere additions that occur in an asf1 mutant do not require the activity of telomerase. A possible explanation for this effect is that, in the absence of ASF1, chromatin structure may be altered near telomeres, resulting in increased disruption of telomere integrity in a tlc1 mutant leading to increased genome instability (16, 17). Consistent with this idea, it has been shown that an asf1 mutation enhances the growth defects caused by a cdc13 mutation that causes a defect in telomere maintenance (50). Alternatively, because this effect is similar to the increase in the GCR rate observed in a tel1 tlc1 double mutant compared with the respective single mutants (16), it is possible that asf1 and tel1 mutations may cause similar checkpoint defects in response to telomere damage.

Double-Strand Break (DSB)-Repair Pathways Are Required for Suppression and Formation of GCRs in cac1 and asf1 Strains. Previous studies have suggested that the break-induced replication pathway of recombination, presumably by promoting recombination with sister chromatids, plays a role in suppressing GCRs and that nonhomologous end joining (NHEJ) can act on damage that leads to GCRs and sometimes results in the formation of translocations (11, 16). The role of these pathways in the formation and suppression of GCRs observed in cac1 and asf1 strains was investigated, combining mutations that cause defects in different DSB-repair pathways with cac1 and asf1 mutations (Table 4). Break-induced replication is highly dependent on RAD52 but shows less dependence on other RAD50 epistasis group genes because of redundancy between different gene products or only a partial requirement at normal growth temperatures (62, 63). Combining a rad52 mutation with either a cac1 or asf1 mutation resulted in a synergistic increase in the GCR rate. Synergistic effects were also seen with rad51, rad54, rad55, rad57, and rad59 mutations (data not shown), which suggests that the GCRs induced by cac1 and asf1 mutations are suppressed partially by the homologous recombination. Inactivation of NHEJ by a lig4 mutation (64) in combination with cac1 mutation caused a significant increase in the GCR rate compared with that caused by a cac1 mutation (P = 0.028). The rearrangement breakpoints that occurred in a cac1 lig4 double mutant were all de novo telomere additions (Tables 2 and 4), indicating that cac1-induced translocation events require NHEJ. In contrast, combining a lig4 mutation with an asf1 mutation did not change the GCR rate compared with that seen in an asf1 single-mutant strain (P = 0.46). We did not analyze the GCRs formed in an asf1 lig4 double mutant because of the low rate of formation of translocations induced by an asf1 mutation (Table 2). These results suggest that ligase 4 plays a role in suppressing some of the cac1-induced GCRs but not in suppressing asf1-induced GCRs. Inactivation of the yKU80 gene in both cac1 and asf1 strains reduced the GCR rates observed to almost wild-type levels, which is consistent with previous observations that yKU80 is required for both efficient NHEJ and efficient de novo telomere addition (16), although it is also possible that asf1- and cac1-induced GCRs are lethal in the absence of Ku. Although our favored hypothesis is that recombination and NHEJ act to suppress GCRs induced by asf1 and cac1 mutations, it is also possible that in the absence of recombination or NHEJ, there are increased levels of broken DNAs, the repair of which is less efficient in the absence of ASF1 or CAC1.

Histone H2A is phosphorylated in response to DNA damage and becomes localized to the site of DSBs in DNA, suggesting an involvement in DSB repair (56, 65). Mutations that eliminate the phosphorylation sites in histone H2A cause increased sensitivity to agents that cause DSBs (56). The effects of cac1 and asf1 mutations were tested in combination with hta1 and hta2 mutations that eliminate the histone H2A phosphorylation sites (Table 4). The hta1 and hta2 mutations had no effect on the GCR rate by themselves. When these mutations were combined with a cac1 mutation, the GCR rate was reduced significantly (P = 0.0001), yet they had no effect on the GCR rate when combined with an asf1 mutation (P = 0.1). These results suggest that phosphorylation of histone H2A in response to DNA damage contributes to the formation of GCRs when CAC1 is not functional.

cac1 and asf1 Mutations Increase the Frequency of DNA-Damage-Induced GCRs. Treatment of cac1 or asf1 strains with 0.05% MMS, an MMS concentration that only activates the intra-S checkpoint (66, 67), did not result in a further increase in the GCR frequency, whereas treatment of a wild-type control strain with 0.05% MMS significantly increased the GCR frequency (Fig. 1A) (68). However, when cac1 or asf1 strains were treated with higher doses of MMS (0.1% or 0.2%), which are concentrations that activate both the intra-S checkpoint and G₁ and G₂ DNA-damage checkpoints (66, 67), a large increase in GCR frequency was observed (Fig. 1B) (68). The cac1 mutant was no more sensitive to killing by all doses of MMS tested than the wild-type strain, whereas the asf1 mutant was 10- to 20-fold more sensitive to killing by MMS (data not shown), consistent with other studies (36, 50). These results suggest that chromatin assembly involving CAC1 and ASF1 plays an important role in the repair of high-dose but not low-dose MMS-induced DNA damage. Alternatively, it is possible that the intra-S checkpoint is already activated in cac1 and asf1 strains, resulting in suppression of low-dose MMS-induced GCRs. As a result, induction of GCRs is only observed in response to treatment with higher concentrations of MMS.

Fig. 1.

Induction of GCRs by DNA damage in CAF-I- and RCAF-defective strains. Log-phase cells of the wild-type (RDKY3615) and cac1 (RDKY4753) or asf1 (RDKY4755) mutant strains were treated with the indicated concentration of MMS in water for 2 h, washed, diluted into 10 times the starting volume of yeast extract/peptone/dextrose, and grown to saturation. The cells then were plated on appropriate medium to determine the frequency of Can^r 5-FOA^r cells present. Three to five independent cultures of each strain were used in each experiment, and each experiment was performed at least twice. The average fold increase in the frequency of GCRs relative to no-MMS treatment is reported. Results obtained at either 0% or 0.5% MMS (A) are plotted with a fold-induction scale (y axis) different from the complete data set for 0%, 0.05%, 0.1%, and 0.2% MMS (B).

Discussion

Previous studies have led to the hypothesis that errors during S phase can result in spontaneous genome rearrangements (9–15). To investigate this hypothesis further, we tested whether the chromatin-assembly factors CAF-I and RCAF, which are thought to function in the assembly of chromatin during DNA replication (33–39), are important for suppression of GCRs. Consistent with this idea, mutations in genes encoding components of these two chromatin-assembly factors resulted in increased spontaneous genome instability. In each case, the GCRs that resulted were translocations/deletions or terminal deletions of chromosome arms associated with de novo telomere addition driven by telomere maintenance functions. The cac1-induced (CAF-I) translocations/deletions seemed to be formed by NHEJ of broken chromosomes. Genetic analysis indicated that cac1-induced (CAF-I) GCRs were suppressed by DNA-damage checkpoints but not the replication checkpoint, which is consistent with recent observations that expression of a dominant-negative human CAC1 (p150) induced DNA-damage foci and altered chromatin structure during S phase (31). In contrast, asf1-induced (RCAF) GCRs were suppressed by both DNA-damage checkpoints and the replication checkpoint, consistent with previous observations that asf1 mutations seem to cause damage during S phase and that ASF1 alleviates histonemediated inhibition of DNA replication in vitro (36, 51, 69). In both cases, the GCRs seem to be suppressed by RAD52-dependent recombination and PIF1-dependent suppression of de novo telomere additions similar to that observed for spontaneous GCRs (16). Defects in CAF-I and RCAF also resulted in increased MMS-induced GCRs, consistent with the observation that asf1 mutations cause sensitivity to agents that induce DSBs and that CAF-I promotes chromatin assembly during DNA repair (35, 36, 42, 50).

Our observations suggest that CAF-I and RCAF promote different chromatin-assembly functions. This conclusion is based on the observation that mutations in CAC1 and ASF1 each resulted in different GCR rates and showed different interactions with checkpoint defects, NHEJ defects, and defects in histone H2A phosphorylation. These conclusions are consistent with the results of other genetic studies that indicate that CAF-I and RCAF may have distinct functions and cooperate with each other (36–39, 50). We observed that cac2 and asf1 single mutants and the cac2 asf1 double mutant had similar GCR rates, which suggests a role for CAC2 in RCAF function consistent with the observed physical interaction between CAC2 and ASF1 (48, 49). Interestingly, asf1 and cac2 mutations reduced the GCR rate of a cac1 mutant strain, and a cac2 mutation reduced the GCR rate of the cac1 cac3 double mutant. This observation raises the possibility that some of the GCRs that occur in a cac1 mutant or a cac1 cac3 double mutant may result from aberrant reactions promoted by RCAF in the absence of a functional CAF-I complex and by aberrant reactions promoted by CAF-I subcomplexes. This idea is consistent with the hypothesis that CAF-I and RCAF are codependent, which is based on biochemical studies (36, 49, 69). An alternate possibility is that asf1 and cac2 mutations activate a checkpoint that partially suppresses the GCRs induced in cac1 and cac1 cac3 mutants. The hta1 S129STOP hta2 S129STOP mutations also reduced the GCR rate of a cac1 mutant strain, suggesting that phosphorylation of improperly assembled chromatin may also contribute to genome instability. Interestingly, induction of a dominant-negative CAC1 in mammalian cells induced phosphorylated histone H2Ax foci (31). It should also be noted that other studies have suggested that CAC1–CAC3 all seem to have distinct functions in addition to a common function (43–47), and it is known that a portion of CAC3 does not copurify with the CAF-I complex (49, 70). Thus, it is also possible that these differences in function could contribute to the differences in the effects of the cac1, cac2, and cac3 mutations seen here.

A model describing our observations is presented in Fig. 2. Our results suggest that failure to assemble chromatin properly during DNA replication results in spontaneous DNA damage that can be processed to yield GCRs. The differences in the interaction between cac1 (Fig. 2A) and asf1 (Fig. 2B) with mutations that cause defects in the different checkpoints that act in suppressing genome instability may reflect the induction of different types of damage by asf1 and cac1. The checkpoint and replication (S-phase) defects caused by ASF1 but not CAC1 mutations may also contribute to the differences in genetic instability caused by ASF1 and CAC1 mutations. Our genetic data also suggest that cac1-induced GCRs (and cac1 cac3-induced GCRs, where tested) may involve ASF1 and CAC2 action and histone H2A phosphorylation in the absence of normal CAF-I function. There are several possible explanations for this result: (i) aberrant reactions promoted by CAF-I subcomplexes or by ASF1, CAC2, and histone H2A phosphorylation in the absence of normal CAF-I function may result in increased damage or decreased repair; or (ii) possibly the absence of ASF1, CAC2, or histone H2A phosphorylation could activate a pathway (such as a checkpoint pathway) that then promotes suppression of cac1-induced GCRs. In all cases, the formation and suppression of the resulting GCRs seem to involve the same pathways as have been implicated in the formation and suppression of spontaneous GCRs (8, 16). These results suggest that some spontaneous GCRs may result from spontaneous errors in chromatin assembly and also raise the possibility that genetic defects in chromatin-assembly genes could promote genome instability in mammalian cells. A remaining question of interest is that of what roles other chromatin assembly and modification functions play in maintaining genome stability.

Fig. 2.

Model for the induction of GCRs by CAC1 and ASF1 defects. (A) In the absence of CAC1, aberrant chromatin assembly occurs on newly replicated DNA and results in DNA damage, possibly including DSBs. This damage may be mediated in part by the action of ASF1 (and CAC2) and histone H2A phosphorylation in the absence of normal CAF-I function. This DNA damage results in GCRs, which are suppressed in part by the DNA-damage checkpoints. (B) Induction of GCRs by ASF1 defects. In the absence of ASF1, aberrant chromatin assembly occurs during DNA replication and results in DNA damage, possibly including DSBs. Replication defects and checkpoint defects likely also occur. The resulting DNA damage and replication and checkpoint defects result in GCRs, which are suppressed in part by both the DNA-damage and replication checkpoints.