The Saccharomyces cerevisiae Rad6 Postreplication Repair and Siz1/Srs2 Homologous Recombination-Inhibiting Pathways Process DNA Damage That Arises in asf1 Mutants

Abstract

The Asf1 and Rad6 pathways have been implicated in a number of common processes such as suppression of gross chromosomal rearrangements (GCRs), DNA repair, modification of chromatin, and proper checkpoint functions. We examined the relationship between Asf1 and different gene products implicated in postreplication repair (PRR) pathways in the suppression of GCRs, checkpoint function, sensitivity to hydroxyurea (HU) and methyl methanesulfonate (MMS), and ubiquitination of proliferating cell nuclear antigen (PCNA). We found that defects in Rad6 PRR pathway and Siz1/Srs2 homologous recombination suppression (HRS) pathway genes suppressed the increased GCR rates seen in asf1 mutants, which was independent of translesion bypass polymerases but showed an increased dependency on Dun1. Combining an asf1 deletion with different PRR mutations resulted in a synergistic increase in sensitivity to chronic HU and MMS treatment; however, these double mutants were not checkpoint defective, since they were capable of recovering from acute treatment with HU. Interestingly, we found that Asf1 and Rad6 cooperate in ubiquitination of PCNA, indicating that Rad6 and Asf1 function in parallel pathways that ubiquitinate PCNA. Our results show that ASF1 probably contributes to the maintenance of genome stability through multiple mechanisms, some of which involve the PRR and HRS pathways.

DNA replication must be highly coordinated with chromatin assembly and cell division for correct propagation of genetic information and cell survival. Errors arising during DNA replication are corrected through the functions of numerous pathways including checkpoints and a diversity of DNA repair mechanisms (32, 33, 35). However, in the absence of these critical cellular responses, replication errors can lead to the accumulation of mutations and gross chromosomal rearrangements (GCRs) as well as chromosome loss, a condition generally termed genomic instability (33). Genome instability is a hallmark of many cancers as well as other human diseases (24). There are many mechanisms by which GCRs can arise, and over the last few years numerous genes and pathways have been implicated in playing a role in the suppression of GCRs in Saccharomyces cerevisiae and in some cases in the etiology of cancer (27, 28, 33, 39-47, 51, 53, 56, 58, 60), including S. cerevisiae ASF1, which encodes the main subunit of the replication coupling assembly factor (37, 62).

Asf1 is involved in the deposition of histones H3 and H4 onto newly synthesized DNA during DNA replication and repair (62), and correspondingly, asf1 mutants are sensitive to chronic treatment with DNA-damaging agents (2, 30, 62). However, asf1 mutants do not appear to be repair defective and can recover from acute treatment with at least some DNA-damaging agents (2, 8, 30, 31, 54), properties similar to those described for rad9 mutants (68). In the absence of Asf1, both the DNA damage and replication checkpoints become activated during normal cell growth, and in the absence of checkpoint execution, there is a further increase in checkpoint activation in asf1 mutants (30, 46, 54). It has been suggested that asf1 mutants are defective for checkpoint shutoff and that this might account for the increased steady-state levels of checkpoint activation seen in asf1 mutants (8); however, another study has shown that asf1 mutants are not defective for checkpoint shutoff and that in fact Asf1 and the chromatin assembly factor I (CAF-I) complex act redundantly or cooperate in checkpoint shutoff (31). Furthermore, Asf1 might be involved in proper activation of the Rad53 checkpoint protein, as Asf1 physically interacts with Rad53 and this interaction is abrogated in response to exogenous DNA damage (15, 26); however, the physiological relevance of this interaction is unclear. Asf1 is also required for K56 acetylation of histone H3 by Rtt109, and both rtt109 mutants and histone H3 variants that cannot be acetylated (38) share many of the properties of asf1 mutants, suggesting that at least some of the requirement for Asf1 in response to DNA damage is mediated through Rtt109 (11, 14, 22, 61). Subsequent studies of checkpoint activation in asf1 mutants have led to the hypothesis that replication coupling assembly factor defects result in destabilization of replication forks which are then recognized by the replication checkpoint and stabilized, suggesting that the destabilized replication forks account for both the increased GCRs and increased checkpoint activation seen in asf1 mutants (30). This hypothesis is supported by other recent studies implicating Asf1 in the processing of stalled replication forks (16, 57). This role appears to be independent of CAF-I, which can cooperate with Asf1 in chromatin assembly (63). Asf1 has also been shown to function in disassembly of chromatin, suggesting other possibilities for the mechanism of action of Asf1 at the replication fork (1, 2, 34). Thus, while Asf1 is thought to be involved in progression of the replication fork, both the mechanism of action and the factors that cooperate with Asf1 in this process remain obscure.

Stalled replication forks, particularly those that stall at sites of DNA damage, can be processed by homologous recombination (HR) (6) or by a mechanism known as postreplication repair (PRR) (reviewed in reference 67). There are two PRR pathways, an error-prone pathway involving translesion synthesis (TLS) by lower-fidelity polymerases and an error-free pathway thought to involve template switching (TS) (67). In S. cerevisiae, the PRR pathways are under the control of the RAD6 epistasis group (64). The error-prone pathway depends on monoubiquitination of proliferating cell nuclear antigen (PCNA) on K164 by Rad6 (an E2 ubiquitin-conjugating enzyme) by Rad18 (E3 ubiquitin ligase) (23). This results in replacement of the replicative DNA polymerase with nonessential TLS DNA polymerases, such as REV3/REV7-encoded DNA polymerase ζ (polζ) and RAD30-encoded DNA polη, which can bypass different types of replication-blocking damage (67). The error-free pathway is controlled by Rad5 (E3) and a complex consisting of Ubc13 and Mms2 (E2 and E2 variant, respectively), which add a K63-linked polyubiquitin chain to monoubiquitinated PCNA, leading to TS to the undamaged nascent sister chromatid (4, 25, 65). Furthermore, in addition to modification with ubiquitin, K164 of PCNA can also be sumoylated by Siz1, resulting in subsequent recruitment of the Srs2 helicase and inhibition of deleterious Rad51-dependent recombination events (50, 52, 55), although it is currently unclear if these are competing PCNA modifications or if both can exist on different subunits in the same PCNA trimer. A separate branch of the Rad6 pathway involving the E3 ligase Bre1 monoubiquitinates the histone H2B (29, 69) as well as Swd2 (66), which stimulates Set1-dependent methylation of K4 and Dot1-dependent methylation of K79 of histone H3 (48, 49, 66). Subsequently, K79-methylated H3 recruits Rad9 and activates the Rad53 checkpoint (19, 70). Activation of Rad53 is also bolstered by Rad6-Rad18-dependent ubiquitination of Rad17, which is part of the 9-1-1 complex that functions upstream in the checkpoint pathway (17). Finally, Rad6 complexes with the E3 Ubr1, which mediates protein degradation by the N-end rule pathway (13).

Due to the role of the PRR pathways at stalled replication forks and a recent study implicating the Rad6 pathway in the suppression of GCRs (39), we examined the relationship between these ubiquitination and sumoylation pathways and the Asf1 pathway in order to gain additional insights into the function of Asf1 during DNA replication and repair. Our findings suggest that Asf1 has multiple functions that prevent replication damage or act in the cellular responses to replication damage and that these functions are modified by and interact with the PRR pathways. The TLS PRR pathway does not appear to be involved, and both a Dun1-dependent replication checkpoint and HR are important for preventing the deleterious effects of PRR and Asf1 pathway defects. We hypothesize that this newly observed cooperation between Asf1 and the PRR pathways may be required for resolving stalled replication forks, leading to suppression of GCRs and successful DNA replication.

MATERIALS AND METHODS

Strains and media.

S. cerevisiae strains were grown in yeast extract-peptone-dextrose (YPD) medium or synthetic complete medium lacking the appropriate amino acid. G418-resistant colonies were selected on YPD plates containing 200 mg/liter of Geneticin. All of the strains used were derived from the S288c strain RDKY3615 (MATa ura3-52 leu2Δ1 trp1Δ63 his3Δ200 lys2ΔBgl hom3-10 ade2Δ1 ade8 hxt13::URA3) either by crossing with other RDKY3615 derivatives or by standard gene disruptions (7). The TAF tag added at the C terminus of PCNA has the structure His₆-FLAG₃-TEV cleavage site-(protein A)₂, and the His₆-FLAG₁ tag is a derivative of the TAF tag lacking the TEV site and the two protein A sequences (9, 10, 71). The detailed genotypes of the strains are listed in Table 1.

TABLE 1.

S. cerevisiae strains used in this study^a

Strain	Relevant genotype	Reference or source
RDKY3615	MATaura3-52 leu2Δ1 trp1Δ63 his3Δ200 lys2ΔBgl, hom3-10 ade2Δ1 ade8 hxt13::URA3	7
RDKY4755	RDKY3615 asf1::HIS3	46
RDKY5516	RDKY3615 rad6::HIS3	27
RDKY5519	RDKY3615 rad5::KAN-MX	27
RDKY5517	RDKY3615 rad18::KAN-MX	27
YKJM1585	RDKY3615 ubc13::TRP1	39
YKJM2135	RDKY3615 mms2::TRP1	39
RDKY5523	RDKY3615 rad30::HIS3	27
RDKY5521	RDKY3615 rev3::HIS3	27
RDKY6491	RDKY3615 rev3::HIS3 rad30::KAN-MX	This study
YKJM2233	RDKY3615 bre1::HIS3	39
YKJM2179	RDKY3615 siz1::TRP1	39
RDKY5028	RDKY3615 srs2::HIS3	This study
RDKY3636	RDKY3615 rad51::HIS3	41
RDKY4421	RDKY3615 rad52::HIS3	41
RDKY4423	RDKY3615 rad59::TRP1	41
RDKY5873	RDKY3615 pol30-119.LEU2	This study
RDKY5755	RDKY3615 asf1::TRP1 rad6::HIS3	This study
RDKY5753	RDKY3615 asf1::HIS3 rad5::KAN-MX	This study
RDKY5757	RDKY3615 asf1::HIS3 rad18::KAN-MX	This study
RDKY5884	RDKY3615 asf1::HIS3 ubc13::TRP1	This study
RDKY5886	RDKY3615 asf1::HIS mms2::TRP1	This study
RDKY5807	RDKY3615 asf1::TRP1 rev3::HIS3	This study
RDKY5888	RDKY3615 asf1::TRP1 rad30::HIS3	This study
RDKY6493	RDKY3615 asf1::TRP1 rev3::HIS3 rad30::KAN-MX	This study
RDKY5761	RDKY3615 asf1::TRP1 bre1::HIS3	This study
RDKY5759	RDKY3615 asf1::HIS3 siz1::TRP1	This study
RDKY5890	RDKY3615 asf1::TRP1 srs2::HIS3	This study
RDKY5874	RDKY3615 asf1::HIS3 pol30-119.LEU2	This study
RDKY4841	RDKY3615 asf1::HIS3 rad51::TRP1	This study
RDKY4855	RDKY3615 asf1::HIS3 rad52::TRP1	This study
RDKY4853	RDKY3615 asf1::HIS3 rad59::TRP1	This study
RDKY4803	RDKY3615 asf1::HIS3 dun1::TRP1	46
RDKY5863	RDKY3615 rad6::HIS1 dun1::KAN-MX	This study
RDKY5892	RDKY3615 rad5::KAN-MX dun1::TRP1	This study
RDKY5894	RDKY3615 rad18::KAN-MX dun1::TRP1	This study
RDKY6563	RDKY3615 pol30-119.LEU2 dun1::KAN-MX	This study
RDKY5859	RDKY3615 siz1::TRP1 dun1::KAN-MX	This study
RDKY5896	RDKY3615 bre1::HIS3 dun1::KAN-MX	This study
RDKY5869	RDKY3615 asf1::TRP1 rad6::HIS3 dun1::KAN-MX	This study
RDKY5900	RDKY3615 asf1::HIS3 rad5::KAN-MX dun1::TRP	This study
RDKY5902	RDKY3615 asf1::HIS3 rad18::KAN-MX dun1::TRP	This study
RDKY6615	RDKY3615 asf1::HIS3 pol30-119.LEU2 dun1::KAN-MX	This study
RDKY5861	RDKY3615 asf1::HIS3 siz1::TRP1 dun1::KAN-MX	This study
RDKY5865	RDKY3615 asf1::TRP1 bre1::HIS3 dun1::KAN-MX	This study
RDKY6862	RDKY3615 ubc13::TRP1 rad52::KAN-MX	This study
RDKY5906	RDKY3615 asf1::HIS3 rad5::KAN-MX rad52::TRP1	This study
RDKY5908	RDKY3615 asf1::HIS3 rad18::KAN-MX rad52::TRP1	This study
RDKY6863	RDKY3615 asf1::HIS3 ubc13::TRP1 rad52::KAN-MX	This study
RDKY3739	RDKY3615 dun1::HIS3	43
RDKY3749	RDKY3615 rad53::HIS3 sml1::KAN	43
RDKY5965	RDKY3615 DDC2-GFP::hph	This study
RDKY5967	RDKY3615 asf1::TRP1 DDC2-GFP::hph	This study
RDKY5969	RDKY3615 rad6::HIS3 DDC2-GFP::hph	This study
RDKY5971	RDKY3615 asf1::TRP1 rad6::HIS3 DDC2-GFP::hph	This study
RDKY6325	RDKY3615 rad5::KAN-MX DDC2-GFP::hph	This study
RDKY6324	RDKY3615 rad5::KAN-MX asf1::TRP1 DDC2-GFP::hph	This study
RDKY5946	RDKY3615 POL30-TAF::KAN-MX	This study
RDKY6313	RDKY3615 asf1::TRP1 POL30-TAF::KAN-MX	This study
RDKY6314	RDKY3615 rad6::HIS3 POL30-TAF::KAN-MX	This study
RDKY6315	RDKY3615 asf1::TRP1 rad6::HIS3 POL30-TAF::KAN-MX	This study
RDKY5976	RDKY3615 pol30-119-TAF::KAN-MX::LEU2	This study
RDKY6362	RDKY3615 asf1::TRP1 pol30-119-TAF.KAN-MX::LEU2	This study
RDKY6363	RDKY3615 rad6::HIS3 pol30-119-TAF.KAN-MX::LEU2	This study
RDKY6364	RDKY3615 asf11::TRP1 rad6::HIS3 pol30-119-TAF::KAN-MX::LEU2	This study
RDKY6649	RDKY3615 POL30-His6-FLAG::KAN-MX	This study
RDKY6663	RDKY3615 asf1::TRP1 POL30-His6-FLAG::KAN-MX	This study
RDKY6647	RDKY3615 rad6::HIS3 POL30-His6-FLAG::KAN-MX	This study
RDKY6777	RDKY3615 asf1::TRP1 rad6::HIS3 POL30-His6-FLAG::KAN-MX	This study

^aAll strains listed are isogenic to RDKY3615, except for the indicated additional mutations.

Characterization of GCR rates.

The GCR assay used in this study measures the simultaneous loss of CAN1 and an inserted URA3 gene, both of which are located on the nonessential left end of chromosome V (7). All GCR rates were determined by fluctuation analysis using two or more independent clones to measure the rate of accumulation of Can^r, 5-fluoroorotic acid-resistant (5FOA^r) progeny during the growth of the strains being characterized. The average GCR rates from two or more experiments using at least five cultures for each clone are reported, as previously described (7, 42).

Sensitivity to chronic exposure to HU.

Cells were grown in YPD medium to log phase. Serial dilutions were made and spotted onto YPD plates containing 50 mM or 100 mM hydroxyurea (HU), which were then incubated for 2 to 3 days at 30°C.

Sensitivity to acute treatment with HU.

Cells were grown to log phase in YPD, and serial dilutions of the culture were plated onto YPD plates to determine the concentration of viable cells. HU was then added to the culture at a final concentration of 200 mM and left for 2 h, after which serial dilutions of the cultures were plated onto YPD plates. The plates were incubated for 2 to 3 days at 30°C. Survival was calculated by determining the percentage of viable cells present after HU treatment compared to the same culture prior to treatment. These experiments were performed in duplicate.

α-Factor arrest.

Cells were grown in YPD medium to log phase. α-Factor (Sigma) was added to the culture at a concentration of 7 μg/ml. The culture was then incubated for 2 h at 30°C with shaking, and arrest was monitored by phase-contrast light microscopy. The cells were then harvested by centrifugation, washed with water, and resuspended in fresh YPD.

Analysis of PCNA modifications.

For analysis of TAF-tagged strains, α-factor-arrested cells were washed with water and released into YPD or into YPD supplemented with methyl methanesulfonate (MMS). At specific time points, cells were pelleted and resuspended in ice-cold lysis buffer 1 (150 mM NaCl, 50 mM Tris buffer [pH 7.5], 10 mM NaF, 1 mM EDTA, 0.1% NP-40, 10 mM N-ethylmaleimide, 10 mM iodoacetate, protease inhibitors) and lysed using a bead beater. Cleared lysates were analyzed by Western blotting using peroxidase-antiperoxidase antibodies that recognize the protein A tags present at the C terminus of PCNA. Alternatively, for His-FLAG-tagged Pol30 strains, α-factor-arrested cells were washed with water and released into YPD or into YPD supplemented with MMS at the indicated concentrations for 2 h. Cells were then frozen overnight at −80°C before being lysed as described above. Pol30 was immunoprecipitated with anti-FLAG beads (Sigma) for 2 h, and samples were analyzed by Western blotting. To enhance detection of ubiquitin, the filters were denatured for 30 min at room temperature in 6 M guanidine-HCl-50 mM Tris buffer (pH 7.5)-5 mM dithiothreitol before being blocked with 5% bovine serum albumin. Antiubiquitin antibodies were from Santa Cruz Biotechnology and Covance, anti-SUMO antibodies were from Rockland Inc. and Abcam, and anti-FLAG antibodies were from Sigma. SuperSignal West Dura and SuperSignal West Femto Western blot reagents (Pierce) and maximum-sensitivity film (Kodak) were used for detection.

FACS analysis.

Cells were fixed in 70% ethanol for 1 hour at room temperature, harvested by centrifugation, and resuspended in 50 mM sodium citrate buffer, pH 7.0. After sonication and centrifugation, the cells were once again resuspended in sodium citrate buffer and treated with 250 μg/ml RNase A (U.S. Biochemicals) and 1 mg/ml proteinase K (Sigma) overnight at 37°C. The cells were then harvested by centrifugation, resuspended in 1 ml of sodium citrate buffer containing 1 μM Sytox green (Molecular Probes), and incubated at room temperature for 2 h. Samples were analyzed using a BD fluorescence-activated cell sorting (FACS) scan (FACS Vantage SE). The G₂/M content of the cells was calculated as previously described (30) using WinMDI2.8 software.

Quantification of Ddc2-green fluorescent protein (GFP) foci.

Ddc2 foci were quantified as described previously (30). Briefly, cells were grown in YPD medium to log phase and examined live by using a DeltaVision Restoration confocal microscope (Applied Precision). Images were collected in 0.2-μm z sections to allow viewing of the entire content of the cells. Two hundred to 300 cells were imaged and counted for each experiment. Softworx software (Applied Precision) was used for image analysis. Experiments were done in duplicate.

RESULTS

The PRR pathways and HR interact with the damage that occurs in asf1 mutants and results in GCRs.

Previous studies have shown that like Asf1, Rad18 and Rad5 play a role in preventing GCRs (39, 46). Therefore, we wanted to examine whether Asf1 and RAD6 epistasis group proteins function in the same pathway or in interacting pathways. We found here that rad5 and rad18 mutants had high GCR rates, in contrast to the low GCR rate of a rad6 single mutant (Table 2), in agreement with previously reported results (39). To examine whether Asf1 and the RAD6 epistasis group proteins function in the same or in interacting pathways, double mutants containing combinations of mutations affecting ASF1 and different PRR components were constructed and their GCR rates were determined by fluctuation analysis (7). We examined combinations of an asf1 deletion with rad6, rad18, rad5, ubc13, mms2, and bre1 deletions (Table 2). Interestingly, deletion of these genes reduced the GCR rate of an asf1 deletion strain to near-wild-type levels. This was also the case for deletions of SIZ1 and SRS2, which encode proteins involved in sumoylation of PCNA and subsequent inhibition of HR. Because asf1 mutants primarily have increased rates of accumulation of the de novo telomere addition class of GCRs (46), it is this class of GCRs that is predominantly suppressed.

TABLE 2.

PRR and HR defects have opposite effects on the GCR rates of asf1 mutants

Relevant genotype	Wild type		asf1Δ
	Strain	Mutation rate,a Canr 5FOAr	Strain	Mutation rate,a Canr 5FOAr
Wild type	RDKY3615	3.5 × 10−10 (1)	RDKY4755	2.1 × 10−8 (62)
rad5Δ	RDKY5519	1.6 × 10−9 (5)	RDKY5753	3.3 × 10−9 (9)
rad6Δ	RDKY5516	2.8 × 10−9 (8)	RDKY5755	1.8 × 10−9 (5)
rad18Δ	RDKY5517	1.5 × 10−9 (4)	RDKY5757	2.0 × 10−9 (6)
bre1Δ	YKJM2233	<9.9 × 10−10 (3)b	RDKY5761	1.3 × 10−9 (4)
ubc13Δ	YKJM1585	1.3 × 10−9 (4)b	RDKY5884	2.4 × 10−9 (7)
mms2Δ	YKJM2135	<2.6 × 10−10 (1)b	RDKY5886	1.8 × 10−9 (5)
pol30-119	RDKY5873	8.3 × 10−10 (2)	RDKY5874	1.9 × 10−9 (6)
rad30Δ	RDKY5523	3.5 × 10−10 (1)c	RDKY5888	1.0 × 10−8 (30)
rev3Δ	RDKY5521	1.8 × 10−10 (0.5)c	RDKY5807	2.3 × 10−8 (65)
rad30Δ rev3Δ	RDKY6491	7.9 × 10−10 (2)	RDKY6493	1.1 × 10−8 (32)
siz1Δ	YKJM2179	1.3 × 10−9 (4)b	RDKY5759	3.2 × 10−9 (10)
srs2Δ	RDKY5028	<3.1 × 10−10 (1)	RDKY5890	1.44 × 10−9 (4)
rad51Δ	RDKY3636	3.5 × 10−9 (10)	RDKY4841	7.2 × 10−8 (205)
rad52Δ	RDKY4421	4.4 × 10−8 (126)	RDKY4855	4.4 × 10−7 (1,257)
rad59Δ	RDKY4423	7.5 × 10−9 (21)	RDKY4853	1.4 × 10−7 (400)

^aNumbers in parentheses indicate the increases (n-fold) in mutation rate compared to that for the wild type.

^bRate reprinted from reference 39.

^cRate determined by Meng-Er Huang in this laboratory (27).

We also examined the effects of the pol30-119 (PCNA K164R) mutation on the GCR rates of asf1 mutants. This mutation alters lysine 164, the single site of Rad6-dependent ubiquitination and major site of sumoylation of PCNA (23), thus eliminating ubiquitination by the Rad6 pathways and sumoylation by the Siz1 pathway. We found that the GCR rates of the asf1 pol30-119 double mutants were also reduced to near-wild-type levels (Table 2), demonstrating that the suppression of GCRs by mutations in the Rad6 and Siz1 pathways is likely mediated through a failure to modify PCNA. If this suppression of GCRs was related to TLS, we would expect that deleting translesion polymerases such as polζ (rev3) or polη (rad30) would have a similar effect. However, we found that deletion of REV3, RAD30, or both in an asf1 mutant had little effect on the GCR rate of an asf1 mutant (Table 2), demonstrating that suppression of GCRs is independent of translesion DNA synthesis. We did not examine a rev1 mutation, as Rev1 is known to function in the same pathway as polζ (20, 36).

The above data are consistent with the possibility that the PRR and Siz1 pathways and Asf1 function in parallel but related pathways that either are important for normal replication to prevent replication damage from occurring or process normally occurring replication damage. In the absence of both Asf1 and PRR or the Siz1 pathway, the resultant replication damage could possibly be processed in a GCR-free way by HR. We therefore examined the GCR rates of an asf1 mutant containing additional mutations in the HR genes RAD51, RAD52, and RAD59. In each case, there was a synergistic increase in the GCR rate (Table 2), suggesting that the replication damage caused by an asf1 mutation is processed mainly in an error-free way by HR. Interestingly, we found that an asf1 rad18 rad52 triple mutant had a significantly lower GCR rate than the asf1 or rad52 single mutants or the asf1 rad52 double mutant (Table 3), indicating that in the absence of HR, the replication damage caused by the asf1 mutation leading to the formation of GCRs is dependent on Rad18. Surprisingly, the GCR rates of asf1 rad5 rad52 and asf1 ubc13 rad52 triple mutants were increased significantly compared to those of the single or double mutants, suggesting a Rad18-independent role for the Rad5-Ubc13 polyubiquitination pathway (Table 3). These results suggest that the damage caused by the absence of Asf1 appears to be processed either in a GCR-prone way by the Rad6-Rad18 pathway or in a GCR-free way by HR and Rad5-dependent pathways.

TABLE 3.

Rad5 suppresses GCRs in asf1 mutants in the absence of Rad52

Relevant genotype	Wild type		rad52Δ
	Strain	Mutation rate,a Canr 5FOAr	Strain	Mutation rate,a,b Canr 5FOAr
Wild type	RDKY3615	3.5 × 10−10 (1)	RDKY4421	4.4 × 10−8 (126)
asf1Δ	RDKY4755	2.1 × 10−8 (62)	RDKY4855	4.4 × 10−7 (1,257)
asf1Δ rad5Δ	RDKY5753	3.3 × 10−9 (9)	RDKY5906	1.9 × 10−6 (5,286)
asf1Δ rad18Δ	RDKY5757	2.0 × 10−9 (6)	RDKY5908	8.8 × 10−9 (29)
asf1Δ ubc13Δ	RDKY5884	2.4 × 10−9 (7)	RDKY6863	1.6 × 10−6 (4,629)

^aNumbers in parentheses indicate the increases (n-fold) in mutation rate compared to that for the wild type.

^bThe GCR rates of rad5Δ rad52Δ and rad18Δ rad52Δ double mutants were previously published in reference 39 and are 2.4 × 10⁻⁹ and 3.1 × 10⁻⁹, respectively. The GCR rate of the ubc13Δ rad52Δ double mutant was determined here to be 4.3 × 10⁻⁸.

Dun1 suppresses GCRs in Asf1- and PRR or HR suppression (HRS) pathway-defective mutants.

We have previously shown that a Rad53-Dun1-dependent checkpoint is required for proper S-phase progression in asf1 mutants (30). While deletion of DUN1 in an asf1 mutant does not cause an increase in the GCR rate (46), it was of interest to determine whether Dun1 plays a role in regulating the response to the DNA damage that occurs in double mutants with mutations in both asf1 and the PRR pathway.

As shown in Table 4, deletion of DUN1 in mutants lacking either of the Rad6 pathway proteins, the Pol30 K164 modification site, or Siz1 did not cause an increase in GCR rates compared to those in the dun1 single mutant. When a mutation in ASF1 was introduced into these double mutants, the GCR rates increased substantially, with the exception of the asf1 bre1 dun1 triple mutant. However, BRE1, unlike RAD6 and the PRR and HRS pathway genes, is required primarily for checkpoint activation (19, 70), and hence combining a bre1 mutation with other mutations affecting the RAD53 pathway might not cause increased checkpoint defects compared to that caused by a dun1 mutation. In the cases of the other triple mutants, increased GCR rates were observed, and the triple mutants showed decreased viability and slower growth than the single and double mutants (discussed below), indicating that Dun1 is required to suppress the GCRs and possibly other DNA damage occurring in double mutants containing defects in both the Asf1 and PRR pathways.

TABLE 4.

Suppression of GCRs in asf1 and PRR double mutants requires DUN1

Relevant genotype	Wild type		asf1Δ
	Strain	Mutation rate,a Canr 5FOAr	Strain	Mutation rate,a Canr 5FOAr
Wild type	RDKY3615	3.5 × 10−10 (1)	RDKY4755	2.1 × 10−8 (62)
dun1Δ	RDKY3739	7.3 × 10−8 (200)	RDKY4803	6.9 × 10−8 (197)
rad6Δ	RDKY5516	2.8 × 10−9 (8)	RDKY5755	1.8 × 10−9 (5)
rad18Δ	RDKY5517	1.5 × 10−9 (4)	RDKY5757	2.0 × 10−9 (6)
rad5Δ	RDKY5519	1.5 × 10−9 (5)	RDKY5753	3.3 × 10−9 (9)
bre1Δ	YKJM2233	<9.9 × 10−10 (3)b	RDKY5761	1.3 × 10−9 (4)
siz1Δ	YKJM2179	1.3 × 10−9 (4)b	RDKY5759	3.2 × 10−9 (10)
rad6Δ dun1Δ	RDKY5863	2.2 × 10−8 (64)	RDKY5869	3.14 × 10−7 (897)
rad18Δ dun1Δ	RDKY5894	2.9 × 10−8 (84)	RDKY5902	3.06 × 10−7 (874)
rad5Δ dun1Δ	RDKY5892	2.3 × 10−8 (66)	RDKY5900	8.6 × 10−7 (2463)
pol30-119 dun1Δ	RDKY6563	3.3 × 10−8 (94)	RDKY6615	1.33 × 10−7 (380)
bre1Δ dun1Δ	RDKY5896	1.6 × 10−9 (5)	RDKY5865	1.2 × 10−8 (25)
siz1Δ dun1Δ	RDKY5859	5.2 × 10−8 (234)	RDKY5861	4.6 × 10−7 (1320)

^aNumbers in parentheses indicate the increases (n-fold) in mutation rate compared to that for the wild type.

^bRate reprinted from reference 39.

The highest GCR rate we observed was in the asf1 rad5 dun1 triple mutant, suggesting that Rad5-mediated error-free repair plays the most critical role in processing damage caused by the absence of Asf1. Interestingly, the asf1 siz1 dun1 triple mutant also showed a greatly increased GCR rate compared to the single or double mutants, confirming that the Rad6 PRR and Siz1 pathways play similar roles in suppressing GCRs in asf1 mutants. These data imply that both ubiquitination and sumoylation of PCNA are required to process the damage in asf1 mutants, and in the absence of these pathways a Dun1-dependent checkpoint becomes activated in order to suppress GCRs.

Increased HU and MMS sensitivity in Asf1 and Rad6 PRR pathway double mutants.

We have previously proposed that Asf1 plays a critical role at stalled replication forks (30), and we and others have shown that in the absence of Asf1 there is spontaneous accumulation of checkpoint complexes on DNA during S phase (30, 54). To examine whether Asf1 and the PRR pathways have overlapping roles in processing stalled replication forks, we monitored the growth of rad6, rad5, rad18, bre1, and siz1 mutants (containing the mutation alone or in combination with an asf1 mutation) on media containing 50 mM or 100 mM HU (Fig. 1A). While asf1 mutants grew slower on HU-containing media, they were not killed and they recovered fully after a few days of incubation at 30°C, as we have previously reported (30). In contrast, a rad6 mutant was extremely sensitive to chronic exposure to HU (Fig. 1A) and did not grow during extended incubation (not shown). Similar extreme sensitivity to HU was also observed for the asf1 rad6 double mutant. The rad18 and rad5 mutants were similar to the asf1 single mutant, with slow growth but full recovery. In contrast, the siz1 and bre1 single mutants did not show any growth defect on HU plates. The asf1 rad18, asf1 rad5, and asf1 bre1 double mutants demonstrated a marked decrease in their ability to grow on HU-containing media compared to the single mutants, suggesting that the Rad6 PRR pathways as well as the Bre1-mediated histone modification and Rad53 checkpoint responses may play separate but related roles at stalled replication forks. Interestingly, the asf1 siz1 double mutant did not have an exacerbated phenotype compared to the asf1 single mutant, demonstrating that Asf1 functions independently of PCNA sumoylation in response to HU. We also observed virtually the same genetic interactions between an asf1 mutation and rad5, rad6, rad18, and bre1 mutations when MMS sensitivity was examined (Fig. 1B) (note that a siz1 mutation was not examined).

FIG. 1.

HU and MMS sensitivities of mutants lacking Asf1 and PRR pathway components. (A) Sensitivity of mutants to chronic exposure to HU. Serial dilutions of cells were plated onto YPD medium containing HU as indicated and incubated at 30°C for 2 to 3 days. (B) Sensitivity of mutants to chronic exposure to MMS. Conditions were the same as for panel A except that the plates contained the indicated concentrations of MMS instead of HU. (C) Mutants lacking Asf1 and PRR components are not sensitive to killing by acute exposure to HU. Cells were exposed to 200 mM HU in culture for 2 h, washed, and plated on YPD. The data are expressed as the percentage of viable cells (CFU) that survive acute HU treatment relative to the viable cells present in the culture at the time of addition of HU. The small increase in the number of CFU after treatment reflects the fact that there is some cell division that occurs during the early phase of growth after the addition of HU. Error bars indicate standard deviations.

To examine whether the growth defect on HU plates was due to a checkpoint defect resulting from the inactivation of the Rad6 and Asf1 pathways, we treated the single and double mutants with 200 mM HU for 2 hours and determined the percent survival when the treated cells were replated onto normal yeast medium (Fig. 1C). As we have shown previously, unlike the checkpoint-defective rad53 mutants (12), asf1 mutants had wild-type levels of survival after acute HU treatment (30). The rad6, rad18, rad5, and bre1 single mutants were also not killed by acute HU treatment. There was also no HU-dependent killing observed for the asf1 bre1, asf1 rad18, and asf1 rad5 double mutants, indicating that these mutants are not checkpoint defective. The asf1 rad6 double mutant had somewhat reduced survival compared to the asf1 and rad6 single mutants; however, this survival was much higher than that of the rad53 mutant, indicating that if there is a checkpoint defect in this double mutant, it is only a partial defect.

Dun1 is required for normal rates of S-phase progression of PRR pathway mutants.

In order to determine whether the suppressed GCR phenotype of double mutants harboring an asf1 deletion and mutations in various genes in the PRR pathways was connected to the rate of S-phase progression, we monitored S-phase progression by FACS of α-factor-arrested G₁ cells that were released into standard yeast medium. We previously found that asf1 mutants progress through S phase at the same rate as wild-type cells but show slowed S-phase progression when defects in replication checkpoint and intra-S checkpoint components, particularly the Rad53-Dun1 pathway, are present (30). In the present study, we examined the effect of mutations in ASF1 and/or DUN1 on S-phase progression of rad6, rad5, rad18, bre1, and siz1 mutants. S-phase progression was evaluated based on the percentage of cells in the G₂/M phase of the cell cycle at 40 min after α-factor release (Fig. 2).

FIG. 2.

S-phase progression of asf1, PRR, and Siz1 pathway mutants. The rates of S-phase progression of the indicated mutants after release from α-factor arrest were monitored using FACS analysis. The proportion of cells in G₂/M was determined at the 40-min time point, at which at least 80% of wild-type cells have reached G₂/M. The values presented are the averages of those from at least three experiments, and the error bars indicate the standard deviations.

The rad5 and rad18 single mutants had both mostly completed DNA replication by 40 min after α-factor release, with kinetics similar to that of the asf1 mutant (90%, 93%, and 89%, respectively). The asf1 rad5 and asf1 rad18 double mutants also showed no significant reduction in S-phase progression rates (90% and 84%, respectively). In contrast, the rad5 dun1 and rad18 dun1 double mutants exhibited markedly slower progression through S phase (45% for both), indicating that Dun1 is required for normal S-phase progression in the absence of Rad5 or Rad18. Deletion of ASF1 in the rad5 dun1 and rad18 dun1 double mutants did not decrease S-phase progression further (65% and 51%), in contrast to the elevated GCR rates observed in these triple mutants. In none of the cases examined was there a delay in exiting G₁ phase. These results are consistent with the idea that asf1, rad5, and rad18 mutations cause similar defects in DNA replication and that a Dun1 checkpoint is required for normal S-phase progression in the presence of these defects.

In contrast to the rad5 and rad18 mutants, the rad6 and bre1 mutants had small but significant defects in progression through S phase (56% and 65%, respectively), which were exacerbated by an asf1 deletion (35% and 52%). Similarly, rad6 dun1 and bre1 dun1 double mutants also had S-phase progression defects (32% and 44%); however, the asf1 bre1 dun1 triple mutant did not have an additional S-phase progression defect (52%) compared to that of the asf1 bre1 double mutant. We were unable to measure the S-phase progression of the asf1 rad6 dun1 triple mutant due to extremely slow growth and poor viability of this mutant. The facts that asf1 rad6 and asf1 bre1 double mutant cells have slower S-phase progression rates that the respective single mutants and that a dun1 mutation does not further exacerbate the S-phase progression defect of an asf1 bre1 double mutant is consistent with the ideas that Bre1 plays a role in Rad53 activation and asf1 mutants require a Rad53 checkpoint for proper S-phase progression. The bre1 checkpoint defect likely accounts for the S-phase progression defect of the asf1 rad6 double mutant.

The asf1 siz1 double mutant had only a small reduction in the rate of S-phase progression compared to the siz1 single mutant (68% versus 75%); however, in the absence of DUN1, the siz1 single and asf1 siz1 double mutants showed a further decrease in the rate of S-phase progression (siz1 dun1, 50%; asf1 siz1 dun1, 55%). These results indicate that the function of Siz1 is important for normal rates of S-phase progression of asf1 mutants but possibly not as important as proper checkpoint function or the activity of Rad6.

Rad6 PRR pathway mutants have elevated levels of Ddc2 foci.

Using a functional GFP-tagged version of the Ddc2 protein that is recruited to sites of DNA damage during checkpoint activation, we and others have previously shown that asf1 mutants have elevated levels of Ddc2 foci, reflecting DNA damage and checkpoint activation, predominantly in S-phase cells, and have further suggested that in asf1 mutants checkpoint activation is required to stabilize stalled or damaged replication forks (30, 54). Because PRR seems to play a role in repairing damage that occurs in the absence of Asf1, we monitored the formation of Ddc2 foci in Rad6 PRR pathway mutants (Fig. 3). Mutations in RAD5, RAD6, RAD18, and BRE1 all caused elevated levels of Ddc2 foci. Mutations in RAD5 and RAD18 caused levels of foci that were similar to that in an asf1 mutant, and the asf1 rad5 and asf1 rad18 double mutants were the same as the respective single mutants, suggesting that asf1, rad5, and rad18 mutations cause defects in the same process. The asf1 rad6 and asf1 bre1 double mutants had higher levels of Ddc2 foci than the respective single mutants. This increase in focus formation likely reflects the combination of rad6 and bre1 checkpoint defects with the increase in DNA damage present in an asf1 mutant, as previously observed for rad9 and dun1 checkpoint-defective mutations (30). We also analyzed Rad53 phosphorylation and found that Rad6 pathway PRR mutants had increased levels of Rad53 phosphorylation, similar to that of asf1 mutants (data not shown). These findings parallel the results of the FACS experiments indicating that asf1 mutations and Rad6 PRR pathway mutations cause the same requirements for checkpoint functions in order to maintain normal S-phase progression.

FIG. 3.

Spontaneous checkpoint activation in mutants lacking Asf1 and PRR pathway components. Checkpoint activation was monitored in live cells by counting the number of small budded S-phase cells with Ddc2.GFP foci using confocal microscopy. The experimental results are quantified for single and double mutants and are expressed as the percentage of cells containing Ddc2.GFP foci. The values presented are the averages of those from at least three experiments, and the error bars indicate the standard deviations.

Asf1 cooperates with Rad6 in ubiquitination of PCNA.

Rad6-dependent PRR is thought to be mediated in part by ubiquitination of PCNA (23). The Rad6-Rad18 pathway monoubiquitinates PCNA on K164, resulting in TLS, while addition of a polyubiquitin chain by Mms2-Ubc13-Rad5 leads to TS. Since ASF1 genetically interacts with components of the RAD6 pathway, we wanted to examine the effect of an asf1 deletion on PCNA ubiquitination in untreated cells as well as in MMS-treated cells. We made use of strains in which PCNA was tagged with the TAF tag (9), which is a C-terminal His₆-3×FLAG-2×protein A tag. In combination with the peroxidase-antiperoxidase antibody, the protein A components of this tag allow for detection of minute amounts of protein that might otherwise go undetected. We arrested cells in G₁ using α-factor and released them into YPD for 40 min. As shown in Fig. 4 and in Fig. S1 in the supplemental material, modification of PCNA in wild-type cells was low in G₁ but increased significantly in S-phase cells, probably due to stalled replication forks at sites of endogenous DNA damage. Modification of PCNA in an asf1 mutant was identical to that in the wild type, whereas it was strongly reduced in rad6 mutants (Fig. 4; see Fig. S1 in the supplemental material), consistent with previous reports that the Rad6 pathway is essential for PCNA modification. However, some PCNA modification persisted in rad6 mutants (Fig. 4; see Fig. S1 in the supplemental material). Furthermore, this low level of PCNA modification appeared to be dependent on Asf1, since PCNA modification was completely absent in a rad6 asf1 double mutant (Fig. 4; see Fig. S1 in the supplemental material). We next monitored PCNA modification in the presence of DNA damage by releasing α-factor-arrested cells into medium containing MMS. As shown in Fig. 5A, treatment with 0.1% MMS strongly induced modification of PCNA in wild-type cells (treatment with HU did not increase modification of PCNA [data not shown]). The MMS-induced modification of PCNA in asf1 mutants was identical to that in wild-type cells, whereas modification was reduced but not eliminated in rad6 mutants, and again the residual modification seen in the rad6 mutant was almost completely dependent on Asf1. Similar results were observed with 0.01%, 0.03%, and 0.3% MMS (data not shown). Even after extreme overexposure of the Western blots, no modification of PCNA was observed in cells harboring the pol30-119 allele, encoding PCNA-K164R, indicating that the residual Rad6-independent modification also modifies this residue (Fig. 5B). While it has generally been interpreted that Rad6-dependent mobility shifts of PCNA like those described above reflect ubiquitination of PCNA, it is possible that sumoylation was being observed in our experiments (23). Therefore, we tagged PCNA with a His-Flag tag and reexamined MMS-dependent S-phase modification of PCNA by immunoaffinity purification of PCNA followed by probing for PCNA modifications on Western blots with ubiquitin- and SUMO-specific antibodies (Fig. 6). The results showed that SUMO levels were generally unaffected in asf1 and rad6 single mutants and modestly reduced in the asf1 rad6 double mutant, whereas ubiquitin levels were reduced to differing degrees in asf1 and rad6 single mutants and the Rad6-independent ubiquitination was significantly reduced in the asf1 rad6 double mutant. Together, these results suggest that an additional Rad6-independent pathway plays a role in ubiquitination of PCNA and that this pathway is strongly dependent on Asf1.

FIG. 4.

Modification of PCNA in untreated cultures depends mostly on Rad6 and partially on Asf1. α-Factor-synchronized cultures of the indicated mutants were released into YPD for 40 min, and modification of PCNA was analyzed by Western blotting to detect the protein A tag. Upper panel, short exposure; lower panel, overexposed Western blot.

FIG. 5.

Modification of PCNA in MMS-treated cells depends on both Rad6 and Asf1 and the K164 residue of PCNA. (A) α-Factor-synchronized cultures of the indicated mutants were released into YPD supplemented with 0.1% MMS, and samples were taken at the indicated time points. Cells were lysed, and modification of PCNA was analyzed by Western blotting to detect the protein A tag. (B) α-Factor-synchronized cultures of the indicated strains containing the TAF-tagged pol30-119 allele were released into YPD supplemented with 0.1% MMS, and samples were taken at the indicated time points. Cells were lysed, and modification of PCNA was analyzed by Western blotting to detect the protein A tag.

FIG. 6.

Ubiquitination of PCNA in MMS-treated cells depends on both Rad6 and Asf1. Wild-type cells, rad6 and asf1 single mutants, and asf1 rad6 double mutants were grown to log phase and arrested in G₁ phase using α-factor. Subsequently, cultures were released into YPD containing the indicated amounts of MMS for 2 h. Equal numbers of cells were then lysed, and Pol30 was immunoprecipitated using FLAG antibodies to pull down His-FLAG-tagged Pol30 and analyzed by Western blotting using anti-SUMO, antiubiquitin, and anti-FLAG antibodies.

DISCUSSION

In the present study, we examined genetic interactions between the Asf1 histone chaperone pathway and the Rad6 PRR and Siz1/Srs2 HRS pathways. The results presented here are consistent with two possibilities; one is that there is increased DNA damage during DNA replication in the absence of Asf1 and that the PRR and Siz1/Srs2 HRS pathways play an important role in the cellular response to this damage, and another is that the Asf1 and PRR and Siz1/Srs2 HRS pathways cooperate in the cellular response to DNA damage that arises spontaneously during DNA replication. Several observations are consistent with this view: combinations of defects in the Asf1 and PRR pathways result in increased sensitivity to chronic exposure to HU and MMS; defects in the Asf1 pathway (30) and the Rad6 PRR pathways result in spontaneous checkpoint activation during S phase, and combinations of defects in the two pathways generally cause further increases in checkpoint activation; and finally, cells that are defective in Asf1 or individual branches of the PRR or Siz1 pathways show normal rates of S-phase progression, but this normal S-phase progression is now dependent on a Dun1-dependent checkpoint. It was previously suggested that Asf1 defects result in DNA damage that can underlie genome instability (30), and our observations indicate that the Rad6 PRR and Siz1/Srs2 HRS pathways coordinate the cellular response to the DNA damage that arises due to such Asf1 defects. Blocking the different Rad6-dependent pathways or the Siz1/Srs2 HRS pathway suppressed the increased GCR rates of asf1 mutants, and this effect was generally dependent on HR and a Dun1-dependent checkpoint, suggesting that under suppressive conditions asf1-induced DNA damage is channeled into HR in a checkpoint-dependent manner. Consistent with all of these types of genetic interactions, ubiquitination of PCNA, which is thought to modulate cellular responses to replication-blocking DNA damage (67), was more dependent on Rad6 in asf1 mutants than in wild-type cells, suggesting the existence of a Rad6-independent PCNA ubiquitination pathway. A model that summarizes our results and relates suppression of asf1-induced GCRs to the Rad6-PRR and Siz1/Srs2 HRS pathways is presented in Fig. 7.

FIG. 7.

Model for the involvement of Asf1 in PCNA modification. (A) When the replication fork encounters a lesion in the DNA, the Rad6-Rad18 pathway is responsible for bulk monoubiquitination of PCNA, obscuring the involvement of Asf1. Monoubiquitin can serve as a seed for further ubiquitination by the Rad5-Mms2-Ubc13 pathway. Sumoylation of PCNA also occurs but is not dependent on Rad6. (B) In the absence of Rad6, some ubiquitination of PCNA still takes place, and this is dependent on Asf1 and an unidentified E2/E3 pair. The sensitivity of our assays is not high enough to determine whether polyubiquitination can still take place in the absence of Rad6 (dashed arrow), but sumoylation of PCNA can still occur and is independent of either Rad6 or Asf1. (C) In the absence of Siz1, sumoylation of PCNA does not occur, and the antirecombination function of Srs2 is not activated. (D) In the absence of Rad5-Ubc13, ubiquitination of PCNA mediated by Rad6-Rad18 still takes place, but polyubiquitination does not appear to occur (dashed arrow). Under the circumstances described for panels B, C, and D, HR is stimulated, resulting in the suppression of asf1-induced GCRs, which also involves a Dun1-dependent checkpoint.

The Asf1 and PRR and Siz1 pathways appear to play shared and distinct roles in the cellular responses to both spontaneous and induced damage that affects DNA replication. Mutations in ASF1 showed synergistic interactions with mutations in RAD5, RAD18, and BRE1, resulting in extreme sensitivity to chronic treatment with HU and MMS; rad6 mutations caused such extreme sensitivity that it was not possible to study the asf1 rad6 double mutant. A siz1 defect did not cause sensitivity to HU and did not affect the sensitivity caused by an asf1 mutation, indicating that the PCNA sumoylation pathway does not interact with the Asf1 pathway in this process. These results suggest that the Asf1 and Rad6 PRR pathways play distinct roles in response to the damage that results from chronic HU and MMS treatment. Like asf1 mutations, defects in Rad6 pathway genes and SIZ1 did not affect S-phase progression, but the rate of S-phase progression was now dependent on Dun1. An asf1 mutation did not affect the S-phase progression of rad5, rad18, or siz1 mutants, whereas combining an asf1 mutation with either a rad6 or bre1 mutation resulted in a lower rate of S-phase progression; the latter result likely reflects the checkpoint defects caused by rad6 and bre1 mutations (19, 70), as deletion of DUN1 in these mutants did not further increase the S-phase progression defect. Indeed, we have previously shown that checkpoint defects reduce the rate of S-phase progression of asf1 mutants (30). Similar to mutants with an asf1 mutation, mutants with mutations in the Rad6 pathway genes also had elevated levels of S-phase Ddc2 foci. asf1 rad5 and asf1 rad18 double mutants showed the same level of Ddc2 foci as the single mutants, whereas the asf1 rad6 and asf1 bre1 double mutants had further-elevated levels, probably reflective of the checkpoint defects caused by rad6 and bre1 mutations resulting in even higher levels of endogenous DNA damage than in either single mutant. Recent studies have shown that asf1 mutants accumulate Rad52 foci whereas PRR pathway mutants do not (3), supporting the view that the further-elevated levels of Ddc2 foci seen in asf1 rad6 and asf1 bre1 double mutants reflect an interaction between the increased damage that occurs in the absence of Asf1 and defects in checkpoints that act in processing this damage. Overall, our observations on S-phase progression and on S-phase Ddc2 foci suggest that Asf1 and the PRR pathways play important roles in minimizing replication stress in normally growing cells and in response to replication stress.

We have previously suggested that in the absence of Asf1, DNA damage occurs during S phase and results in formation of GCRs. The suppression of GCR rates in the asf1 Rad6 PRR and asf1 Siz1/Srs2 HRS pathway double mutants implies that the DNA damage that occurs in asf1 mutants could be recognized by the PRR pathways, which in some way allows the damage to be processed into translocations or healed by telomerase (59). Alternatively, defects in the PRR pathways might channel the asf1-induced DNA damage into pathways that suppress GCRs (Fig. 7). The results presented here indicate these suppressive pathways likely include HR and a Dun1-dependent checkpoint that possibly stabilizes replication forks. Defects in the PRR pathways would stimulate recombination by inhibiting the Siz1 pathway, and our results further suggest that such defects might alter processing of stalled replication forks by modifying the activity of the Rad5 pathway; the Rad6-Rad18 TLS pathway seems unlikely to be involved in formation of GCRs that arise in asf1 mutants, because elimination of the TLS polymerases does not suppress asf1 mutants. The observation that a rad52 mutation causes a synergistic increase in GCR rates in asf1 rad5 and asf1 ubc13 double mutants but not in an asf1 rad18 double mutant suggests that the Rad5-dependent PCNA ubiquitination pathway might have a role that is independent of the Rad6-Rad18 PRR pathway components that modify PCNA (5); this is consistent with our observation of residual PCNA ubiquitination in rad6 mutants. It is unlikely that this reflects Rad5 helicase activity, because the similar effects of rad5 and ubc13 mutations in asf1 rad52 double mutants implicates the ubiquitination pathway; consistent with this, RAD5 and rad5 helicase-defective plasmids equally complemented the asf1 rad5 rad52 triple mutant (data not shown). Interestingly, deletion of DUN1 causes a synergistic increase in the GCR rate of asf1 PRR pathway double mutants but not in that of the respective single mutants. This indicates that both the Asf1-dependent pathway and the PRR pathways must be disabled to produce a defect that triggers a requirement for a Dun1-dependent GCR-suppressing mechanism. Finally, the Bre1-dependent branch of the Rad6 pathway, which controls histone modifications (19, 29, 69, 70), is required for formation of GCRs that arise in mutants lacking Asf1. Although we currently do not understand how deletion of BRE1 results in suppression of GCRs, this finding is consistent with a previous report suggesting that Bre1 might be generally required for GCR formation by allowing GCR-prone DNA repair machinery access to DNA lesions through the modulation of chromatin structure (39).

It is generally accepted that ubiquitination of PCNA in PRR is dependent on the activity of the Rad6 pathways (23, 64, 67). However, we observed that while this was largely true during S phase and after treatment with MMS, there was a significant level of Rad6 pathway-independent ubiquitination present; this residual ubiquitination of PCNA in rad6 mutants has previously been unappreciated and might be detected only when exceptionally sensitive detection methods are used. In contrast, PCNA ubiquitination was much more dependent on Rad6 in asf1 mutants. This indicates that there is an Asf1-dependent PCNA modification pathway; however, we have not yet identified the proteins that promote the observed modifications. Whether this represents a specific Asf1-dependent pathway or whether it is the modification of chromatin or checkpoint functions by the Asf1 pathway that then allows a second ubiquitination pathway to act on PCNA is not clear. However, the fact that PCNA modification in asf1 mutants is more dependent on Rad6 provides a mechanism that could in part explain the observed consequences of PRR pathway defects in asf1 mutants.

It was initially thought that Asf1 functions only in the assembly of chromatin during DNA replication and repair. In recent years it has been appreciated that Asf1 also plays roles in disassembly of chromatin, in modification of chromatin by Rtt109, and in checkpoint shutoff in cooperation with the CAF-I complex (8, 11, 14, 22, 31, 61). It is possible that these different roles of Asf1 could explain the diversity of defects caused by asf1 mutations and the different types of interactions seen between asf1 mutations and PRR defects. For example, it is possible that the role of Asf1 as a chromatin “disassembly” factor (1) may be responsible for the critical role that Asf1 appears to play in replication fork progression. For instance, in the absence of Asf1 the chromatin in front of the replication fork may not be disassembled fast enough, leading to instability of replication forks that would underlie the slower S-phase progression seen in the absence of Dun1. Also, it is likely that in order to resolve stalled replication forks (caused by HU, for example), chromatin must be further disassembled to allow the checkpoint and repair machinery to assemble on the DNA, consistent with the “access, repair, restore” model of chromatin and DNA repair (21). Because the PRR proteins must be recruited to replicating DNA to bypass stalled replication forks, it is possible that disassembly of chromatin by Asf1 may facilitate the action of the PRR proteins underlying the observed collaboration between the Asf1 and PRR pathways. Interestingly, Asf1 is required for histone acetylation by Rtt109, and rtt109 mutations cause many of the same DNA damage sensitivity, increased GCR, and checkpoint activation defects as seen in asf1 mutants (14). This raises the possibility that the interaction between the Asf1 and the Rad6-Bre1 pathway may be mediated through the modification of chromatin (18, 69, 70). Clearly, a broader perspective on how Asf1 alters chromatin structure during normal cell growth and after DNA damage is warranted.