CDC7 and DBF4 encode the essential Cdc7-Dbf4 protein kinase required for DNA replication in eukaryotes from yeast to human. Cdc7-Dbf4 is also required for DNA damage-induced mutagenesis, one of several postreplicational DNA damage tolerance mechanisms mediated by the RAD6 epistasis group. Several genes have been determined to function in separate branches within this group, including RAD5, REV3/REV7 (Pol ζ), RAD30 (Pol η), and POL30 (PCNA). An extensive genetic analysis of the interactions between CDC7 and REV3, RAD30, RAD5, or POL30 in response to DNA damage was done to determine its role in the RAD6 pathway. CDC7, RAD5, POL30, and RAD30 were found to constitute four separate branches of the RAD6 epistasis group in response to UV and MMS exposure. CDC7 is also shown to function separately from REV3 in response to MMS. However, they belong in the same pathway in response to UV. We propose that the Cdc7-Dbf4 kinase associates with components of the translesion synthesis pathway and that this interaction is dependent upon the type of DNA damage. Finally, activation of the DNA damage checkpoint and the resulting cell cycle delay is intact in cdc7Δ mcm5-bob1 cells, suggesting a direct role for CDC7 in DNA repair/damage tolerance.
ONE of the most important aspects of a cell's life cycle is the accurate replication, segregation, and structural maintenance of its genome. Saccharomyces cerevisiae CDC7 encodes the catalytic subunit of a protein kinase that is involved in two of these processes, namely DNA replication and DNA repair (reviewed in Sclafani 2000). In DNA replication, Cdc7 protein is an essential regulator of this process and is thought to control initiation of replication by phosphorylating the Mcm2 protein, thereby activating the MCM helicase complex (reviewed in Bell and Dutta 2002). The requirement of Cdc7 in DNA repair was first suggested by the observation that the cdc7-1 mutant is defective in induced mutagenesis when treated with different DNA-damaging agents, including UV light, methyl methanesulfonate (MMS), and N-methyl-N′-nitro-N-nitrosoguanidine (MNNG; Njagi and Kilbey 1982; Kilbey 1986). Furthermore, it has been determined that overexpression of CDC7 causes an increase in induced mutation frequency (Sclafani et al. 1988), and both hyper- and hypomutagenic alleles of cdc7 have been identified (Hollingsworth et al. 1992). On the basis of its DNA damage-induced mutagenesis phenotype and UV survival epistasis analysis, CDC7 has been assigned to the RAD6 epistasis group of DNA repair genes in S. cerevisiae (Njagi and Kilbey 1982).
The RAD6 epistasis group controls a poorly understood DNA repair pathway composed of several genes that, when mutated, result in sensitivity to a variety of DNA-damaging agents and, in many cases, also cause defects in damage-induced mutagenesis. At the core of this epistasis group is the Rad6 protein, a ubiquitin-conjugating enzyme that is required for all DNA damage tolerance processes (reviewed in Friedberg et al. 1995). Rad6 interacts with Rad18 protein, and it is thought that this complex is recruited to sites of DNA damage by the single-stranded DNA-binding activity of Rad18. Once there, the Rad6-Rad18 complex mediates DNA damage tolerance mechanisms by modifying the replication fork via its ubiquitin-conjugating activity, by protein degradation (Bailly et al. 1994), and/or as a signaling mechanism (Hofmann and Pickart 1999). It has been determined that the downstream components of this pathway are separated into more than one distinct branch, resulting in different mechanisms of DNA damage tolerance. Several models have been proposed for the genetic interactions between members of the RAD6 epistasis group, namely RAD5, MMS2, POL30, RAD30, and REV3, and their roles in error-free or error-prone processes of DNA damage tolerance (McDonald et al. 1997; Ulrich and Jentsch 2000; Xiao et al. 2000).
The genes that are involved in error-free mechanisms of DNA damage tolerance, which are thought to occur via a DNA strand copy-choice process, include RAD5, MMS2, and POL30 (reviewed in Broomfield et al. 2001). Mutations in these genes lead to a deficiency in postreplication repair (PRR) activity following UV irradiation (Torres-Ramos et al. 1996, 2002), while showing little defect in DNA-damage-induced mutagenesis (Johnson et al. 1992; Torres-Ramos et al. 1996; Broomfield et al. 1998). RAD5 encodes a 134-kD protein with a putative helicase domain and a cysteine-rich sequence motif (RING finger; Johnson et al. 1992). Also, Rad5 has been shown to form a homodimer and to mediate an interaction between the UBC13-MMS2 and RAD6-RAD18 complexes (Ulrich and Jentsch 2000). MMS2 encodes a ubiquitin-conjugating enzyme variant protein that, in conjunction with the Ubc13 protein, forms a complex capable of assembling polyubiquitin chains linked through the K-63 residue of ubiquitin (Hofmann and Pickart 1999). Ubiquitin conjugation via K-63 is thought to have a specific signaling role in DNA damage tolerance, as a UbK63R mutation was shown to have DNA repair defects that fall within the RAD6 epistasis group, while having no obvious impairment in protein degradation (Spence et al. 1995). POL30 encodes proliferating cell nuclear antigen (PCNA), the processivity factor in eukaryotic DNA replication that is also involved in a variety of DNA repair processes, including nucleotide excision repair, base excision repair, and mismatch repair (Warbrick 2000). Mutational analysis of this gene identified the pol30-46 allele, which shows increased sensitivity to DNA damage, but is normal for growth (Ayyagari et al. 1995). Genetic analysis of this mutant indicated that its DNA repair defects are specific to the RAD6 epistasis group (Torres-Ramos et al. 1996) and that it functions in a branch separate from RAD5 (Xiao et al. 2000). Because pol30-46 strains show no defect in DNA-damage-induced mutagenesis, it was suggested that POL30 is involved in error-free DNA damage tolerance (Torres-Ramos et al. 1996). However, it does not rule out the possibility that it might also play a role in error-prone mechanisms, and recent work has characterized a different allele of POL30, pol30 (K164R), which is specifically defective in DNA-damage-induced mutagenesis and is epistatic to both rev3Δ and rad30Δ (Stelter and Ulrich 2003). This evidence suggests that PCNA is also involved in translesion synthesis (TLS) and that the pol30-46 mutation knocks out a function of POL30 specific to error-free processes of DNA damage tolerance.
The genes thought to mediate the error-prone process of DNA damage tolerance include RAD30 and REV3. They both encode DNA translesion polymerases that are capable of replicating DNA past a damaged template (TLS; reviewed in Kunz et al. 2000; Prakash and Prakash 2002). RAD30, which encodes DNA polymerase η (Pol η), was identified in a search for homologs of the UmuC and DinB genes of Escherichia coli (McDonald et al. 1997; Roush et al. 1998). Deletion of RAD30 results in increased sensitivity to UV and MMS exposure, and it was determined that RAD30 constitutes a branch within the RAD6 epistasis group that is separate from RAD5 (McDonald et al. 1997; Roush et al. 1998). Pol η is remarkable for its accurate and efficient replication past a cis-syn thymine-thymine (T-T) dimer (Johnson et al. 1999b), and deficiencies in the human RAD30 homolog were found to be responsible for the variant complementation group of xeroderma pigmentosum syndrome (Johnson et al. 1999a; Masutani et al. 1999b). Biochemical analysis has shown that Pol η has low processivity and low fidelity on undamaged template, but is capable of nucleotide insertion across a variety of DNA lesions with both mutagenic and nonmutagenic consequences (Minko et al. 2000; Washington et al. 2000; Yuan et al. 2000; Johnson et al. 2001); genetic evidence also indicates that Pol η TLS activity is required for bypass of a variety of DNA lesions and that it contributes toward MNNG-induced mutagenesis (Haracska et al. 2000; Bresson and Fuchs 2002).
REV3, which encodes the catalytic subunit of DNA polymerase ζ (Pol ζ; Morrison et al. 1989), was identified in a screen for mutants that resulted in a low frequency of UV-induced mutagenesis (Lemontt 1971). The Rev3 protein, together with the Rev7 protein, forms the heterodimeric Pol ζ, which was shown to be a translesion polymerase capable of bypassing a cis-syn (T-T) dimer (Nelson et al. 1996). Recently, biochemical characterization of Pol ζ revealed it to be a high-fidelity DNA polymerase that is very inefficient at bypassing template lesions (Johnson et al. 2000), but highly proficient at extending 3′ ends opposite DNA lesions (Guo et al. 2001; Haracska et al. 2001b, 2003). Genetic analysis, however, indicates that Pol ζ is required for the bypass of a variety of lesions, including a T-T pyrimidine (6-4) pyrimidone dimer [(6-4) T-T dimer; Baynton et al. 1998; Nelson et al. 2000; Lawrence 2002). In light of this evidence, the current model of TLS proposes that one or more DNA polymerases are required for this process, resulting in both mutagenic or nonmutagenic bypass, and that this is mainly a consequence of the type of lesion on the DNA template (Kunz et al. 2000; Broomfield et al. 2001; Prakash and Prakash 2002).
The role of CDC7 in the RAD6 pathway and within its error-free and error-prone branches is currently unknown. The studies done so far suggest that it plays a role in TLS mechanisms. However, given that the limited analysis of CDC7 participation in DNA damage tolerance has focused on its defects in induced mutagenesis and has been carried out using only hypomorphic alleles, it does not rule out the possibility that it might also be required for replication restart in error-free processes. The isolation of the mcm5-bob1 allele (Jackson et al. 1993; Hardy et al. 1997), which allows for the deletion of CDC7, provides us with a tool to carry out an extensive analysis of the genetic relationships between CDC7 and other members of the RAD6 epistasis group.
MATERIALS AND METHODS
Yeast strains, media, and plasmids:
Yeast strains were grown in yeast extract/peptone/dextrose (YPD) with 2% glucose or in synthetic defined minimal media supplemented with appropriate amino acids and 2% glucose (Sclafani et al. 1988). All yeast strains used in this study are listed in Table 1. Strains 888, 889, and 890 were obtained from the Saccharomyces Genome Deletion Project and are in the S288c genetic background (Winzeler et al. 1999). All other strains are congenic with A364a (Hartwell 1967). Standard genetic methods were used for strain construction and tetrad analysis (Burke et al. 2000). Transformation of yeast strains was performed by the lithium acetate method (Ito et al. 1983).
rev3Δ::KanMX4, rad5Δ::KanMX4, and rad30Δ::KanMX4 disruption fragments were generated by PCR amplification of the gene locus using template genomic DNA from strains 888, 889, and 890. Genomic DNA was isolated as described (Lee 1992). The PCR primers used were REV3A, REV3D, RAD5A, RAD5D, RAD30A, and RAD30D from the Saccharomyces Genome Deletion Project.
rev3Δ, rad5Δ, and rad30Δ strains were generated by transforming strain yLPB11 with the respective gene disruption fragment, selecting for G418R. Heterozygote diploids were identified by Southern genomic hybridization. Diploids were sporulated and dissected to generate haploid strains of the genotype desired. Gene disruptions were confirmed again by Southern genomic hybridization. At least two independent isolates were generated for each genotype.
Due to the high recombination rate at the MCM5 locus, proximal to the rDNA region on chromosome XII, it was important to determine the identity of the allele present, MCM5 or mcm5-bob1, in the strains isolated. The original mcm5-bob1 mutation (Hardy et al. 1997) ablates an Eco57I restriction site. This can be used as a diagnostic test on PCR fragments amplified using primers internal to the MCM5 open reading frame, MCM5-Fwd (5′-CACCACTTCCTCCATTTCCACC-3′) and MCM5Rev (5′-CCCCAGATTTAGTGAATAAGAGCCC-3′). When no MCM5 strains were isolated, mcm5-bob1 strains were transformed with pRS306-MCM5, linearized with MluI, selecting for Ura+. This generates a gene duplication, with one MCM5 copy, which complements the mcm5-bob1 mutation. Gene duplication was confirmed by Southern genomic hybridization.
All plasmids used in this study are listed in Table 2. Plasmid pRS306-MCM5 was constructed by cloning a 5.4-kb XhoI/NotI fragment from plasmid pCH802 into the XhoI/NotI sites of pRS306 (Sikorski and Hieter 1989). Plasmid pLPB25 was constructed by cloning a 1.1-kb BamHI/SmaI fragment from plasmid pBL230-46 into the BamHI/SmaI restriction sites in pRS306. Plasmid pLPB26, which introduces a unique NheI restriction site at codons eight and nine of the POL30 open reading frame, was derived from pLPB25 using PCR-overlap mutagenesis (Ho et al. 1989). PCR was carried out using the outside primers M13Fwd (5′-TGTAAAACGACGGCCAGT-3′) and M13Rev (5′-TCACACAGGAAACAGCTATGAC-3′), complementary to the pRS306 backbone, and internal mutation primers POL30Nhe-Fwd (5′-GAAGAAGCtagCCTTTTCAAG-3′) and POL30Nhe-Rev (5′-CTTGAAAAGGctaGCTTCTTC-3′) (lowercase letters indicate silent mutations that introduce a NheI restriction site).
To obtain pol30-46 strains, plasmid pLPB26 was linearized with NheI and transformed into ura3-− strains yLPB18, yLPB21, and yLPB24, selecting for Ura+. This results in a duplication of the POL30 locus, with one of the copies being pol30-46. After growth in YPD, recombinant Ura− clones were selected for on SD − Ura + 5-fluoroorotic acid (5-FOA) media. Integration of the pol30-46 allele was verified by PCR amplification and sequencing of the POL30 locus. PCR amplification and sequencing were carried out using the POL30A and POL30D primers from the Saccharomyces Genome Deletion Project. cdc7Δ::HIS3 mcm5-bob1 pol30-46 strains were also generated by mating strain yLPB26 with strain yLPB62. Diploids were sporulated and dissected, and double mutants were selected. The presence of the pol30-46 mutation was followed by PCR amplification and sequencing, as stated above.
Plasmid pLPB29 was generated by cloning a 3.6-kb MscI/XbaI insert from plasmid pPD61 into the 7.3-kb backbone fragment of plasmid pPD328. This generates a full-length 3× hemagglutinin(HA)-RAD53 gene construct in a pRS305 (Sikorski and Hieter 1989) plasmid backbone. Plasmids pPD61 and pPD328 were a generous gift from Paul Dohrmann of this laboratory. Integration of plasmid pLPB29 at the RAD53 locus was achieved by linearizing the plasmid with MscI and transforming into leu2− strains selecting for Leu+. This generates a RAD53 duplication, with one copy tagged with three HA epitopes. Gene duplication was confirmed by Southern genomic hybridization.
DNA damage survival analysis:
Cells were grown to midlogarithmic phase (between 1 and 5 × 107 cells/ml) in YPD. Cell numbers were determined with a Coulter (Hialeah, FL) Multisizer II using a 100-μm orifice. For UV survival analysis, appropriate dilutions were plated in triplicate on YPD plates and were either untreated (0 J/m2 control) or treated with specific UV doses with a 254-nm source at a fluency rate of 100 or 500 μW/cm2, as measured with a UVP radiometer. Plates were incubated at 30° for 2–3 days, after which colonies were counted. UV exposure and plate incubation were carried out in the dark to avoid light-induced repair. The data presented represent the mean of at least three independent experiments. For MMS survival analysis, two different assays were carried out. For the first assay, 10-fold serial dilutions were spotted onto YPD plates either with no MMS for a control or with specific amounts of MMS added to it. All cultures were diluted to an initial concentration of 2 × 107 cells/ml (10° data point). For the second assay, 5 × 107 cells were resuspended in 5 ml of 0.1 m NaPO4 buffer, pH 7.0, and treated with 0.5% MMS. Samples were removed before (0 min) and at different time points after addition of MMS. The samples were diluted sequentially into 10% sodium thiosulfate (to inactivate the MMS) and water and then plated on rich media to determine survival.
Mathematical analysis of UV survival curves:
When the sensitivity of the double-mutant strain is greater than that of either single mutant, meaning that they are not epistatic, it is possible to determine the expected interaction between the two mutants if their relationship is additive using the natural logarithm of the surviving fraction (−ln S) for each mutant (Brendel and Haynes 1973). This is given by the equation
If the observed −ln Sd.m. is greater than expected, as determined by the equation, one can conclude that the interaction between the two mutations is synergistic.
Fluorescence-activated cell sorter analysis:
Cells were grown at 30° in 20 ml YPD to a density of 1–2 × 107 cells/ml. A sample for each culture was removed and processed for fluorescence-activated cell sorter analysis (FACS) as previously described (asynchronous time point; Ostroff and Sclafani 1995). The rest of the cultures were treated with synthetic α-factor at 10 μm for 2 hr. Cell synchrony was monitored by phase-contrast microscopy at 400× magnification (90–95% unbudded cells indicated α-factor arrest). A sample for each culture was removed and processed for FACS (α-factor time point). The remainder of the cultures was split into two equal parts, washed to remove the α-factor, and resuspended in an equal volume of sterile water. One aliquot for each culture was exposed to 50 J/m2 of UV, at a fluency rate of 1000 μW/cm2, in a 100- × 15-mm petri plate with shaking to keep the cells in suspension; the second aliquot was treated equally, except for no UV exposure. The cultures were spun down, resuspended in 10 ml YPD, and incubated at 23°. UV exposure and subsequent incubation were carried out in the dark. Samples were removed from each culture every 20 min and processed for FACS.
Rad53 protein Western blot analysis:
Strains with a 3× HA-Rad53 construct at its chromosomal locus were grown in 20 ml YPD to a density of ∼2 × 107 cells/ml. Each culture was split into two aliquots, washed, and resuspended in 10 ml sterile water. For each culture, one aliquot was exposed to 100 J/m2 of UV as described for FACS protocol; the second aliquot was a no UV control. After UV exposure, each aliquot was spun down, resuspended in 10 ml YPD, incubated at 23° for 40 min in the dark, and then processed for yeast protein extracts. Cells were washed in 2 ml PK lysis buffer [50 mm Tris (pH 7.6), 50 mm NaCl, 0.1% Triton X-100, 0.1% Tween 20, 1 mm EDTA] and then resuspended in 500 μl PK lysis buffer with 1.7 mg/ml phenylmethanesulfonyl fluoride (Sigma, St. Louis) and 500 μl 0.5-mm glass beads (Biospec Products, Bartlesville, OK) in 1.5-ml screw-cap tubes (Sarstedt, Newton, NC). Cells were lysed by agitation in a Mini-Beadbeater-8 (Biospec Products) with two 2-min bursts interspersed by 1 min on ice. The bottom of each tube was punctured by a needle, and the cell lysate was collected by spinning into a new Eppendorf tube. The lysates were spun at 14,000 × g for 15 min at 4° to remove insoluble material. An aliquot of the soluble protein was used to determine protein concentration by a BCA protein assay (Pierce, Rockford, IL). The remainder was combined with 5× SDS sample buffer (1× final) and boiled immediately for 5 min. Protein extracts (150 μg) were resolved by 7.5% SDS-PAGE, transferred onto nitrocellulose membrane, and probed with 12CA5 anti-HA mouse monoclonal antibody (Boehringer Mannheim, Indianapolis) at 1:1000 dilution. Secondary horseradish peroxidase-conjugated goat anti-mouse antibody (Jackson ImmunoResearch, West Grove, PA) was used at 1:3000 dilution. Immunoblots were visualized with an ECL chemiluminescence kit (Perkin-Elmer Life Sciences, Norwalk, CT).
mcm5-bob1 has no effect on DNA damage survival:
To determine the role of CDC7 within the RAD6 pathway, we carried out a genetic analysis between CDC7 and various members of the RAD6 epistasis group. To avoid the problems that arise from using cdc7 hypomorphic point mutations in this kind of analysis, we exploited the fact that the presence of the mcm5-bob1 mutation permits the deletion of CDC7, which, otherwise, is an essential gene. Previous characterization of the mcm5-bob1 mutation, compared to wild-type (WT) cells, indicated that it causes a slight decrease in the time it takes for yeast cells to enter the S phase of the cell cycle, but has no obvious impairment on the growth of the cells (Hardy et al. 1997). Since we planned to take advantage of the mcm5-bob1 mutation to delete CDC7, we first examined what effect mcm5-bob1 has on DNA damage survival. We found that mcm5-bob1 cells are no more sensitive to UV irradiation than are WT cells (Figure 1), and the same is true for MMS exposure (data not shown). Comparison of a rad5Δ strain with a rad5Δ mcm5-bob1 strain showed that the two strains exhibit the same degree of sensitivity to UV (Figure 1) and MMS exposure (data not shown), indicating that mcm5-bob1 has no interaction with rad5Δ in response to DNA damage. This was also true for combinations of mcm5-bob1 with rad30Δ, rev3Δ, or pol30-46 (data not shown).
We conclude that the mcm5-bob1 mutation has no effect in the response to DNA damage exposure, and that we can use it as genetic tool to study the interaction between cdc7Δ and other mutations in the RAD6 epistasis group. Given this, for simplicity, we omit reference to the mcm5-bob1 allele whenever presenting data on cdc7Δ mcm5-bob1 strains.
CDC7 and REV3 belong to the same pathway in response to UV treatment:
To determine which branch of the RAD6 epistasis group CDC7 belongs to, we carried out a UV survival epistasis analysis on strains combining cdc7Δ with a rad5Δ, rad30Δ, rev3Δ, or pol30-46 mutation. Briefly, an equal number of cells for each of the strains were plated on YPD, after which they were exposed to different UV doses and incubated in the dark to determine cell survival. By comparing the phenotype of double mutations with that of the single mutant, it was determined that the cdc7Δ rev3Δ strain is no more sensitive than a cdc7Δ or rev3Δ strain alone, suggesting that cdc7Δ and rev3Δ are epistatic in response to UV damage (Figure 2A). In contrast, mathematical analysis (Table 3 and see materials and methods) of the single- and double-mutant survival data for rad5Δ in conjunction with cdc7Δ revealed that the cdc7Δ rad5Δ (Figure 2C) strain shows a synergistic response in UV sensitivity relative to the single-deletion strains in that the −ln S for the double-mutant (−ln Sd.m.) strain is greater than expected for an additive interaction (Brendel and Haynes 1973).
The analysis of survival data comparing the interaction between cdc7Δ and rad30Δ or pol30-46 is not as straightforward (Table 3). While the observed −ln Sd.m. is greater than expected, the difference is not as large as when comparing cdc7Δ and rad5Δ. Furthermore, while for the cdc7Δ rad5Δ and cdc7Δ pol30-46 double-mutant strains the difference between the observed and expected −ln Sd.m. increases with higher UV doses, this is not so for the cdc7Δ rad30Δ strain. Nevertheless, the data indicate that the double-mutant strains are more sensitive than either single-mutant strain and the interaction between cdc7Δ and pol30-46 or rad30Δ is at least additive, if not synergistic. These results, together with previously published data, suggest that, in response to UV exposure, CDC7, RAD5, POL30 (as indicated by the pol30-46 allele), and RAD30 constitute separate branches of the RAD6 epistasis group.
CDC7 represents a distinct RAD6 branch in response to MMS treatment:
It has been determined that CDC7 is required for DNA-damage-induced mutagenesis resulting from UV, MMS, MNNG, and EMS treatment (Njagi and Kilbey 1982), whereas REV3 is dispensable in MNNG (Xiao et al. 1999) and, possibly, EMS-induced (Prakash 1976) mutagenesis. This suggests that the requirement of CDC7 and REV3 for mutagenesis in response to different types of DNA-damaging agents is not always the same. Thus, we decided to investigate the genetic relationships between CDC7 and the other members of the RAD6 epistasis group in response to treatment with MMS.
To examine the interaction between cdc7Δ and rad5Δ, we used a qualitative serial dilution assay on rich media plates that contained specific amounts of MMS, compared to media with no MMS. The sensitivity of cdc7Δ in this assay is relatively mild and is notable only starting at MMS concentrations between 0.005 and 0.01% (data not shown). rad5Δ strains, on the other hand, are notably sensitive to MMS concentrations between 0.0005 and 0.001% (Figure 3 and data not shown). The double-mutant strain, however, shows a 5- to 10-fold increase in sensitivity with respect to rad5Δ, with notable killing at 0.0002% MMS (Figure 3). This shows that, as in response to UV, cdc7Δ and rad5Δ show a synergistic interaction upon MMS treatment.
The results for the plate assay examining the relationship between cdc7Δ and pol30-46 were not as obvious. While the double mutant was more sensitive than either single mutant (data not shown), the difference was not as striking as above, which made interpretation of the results difficult. In light of this, we decided to carry out a quantitative assay, where we treated cells in suspension with 0.5% MMS for increasing periods of time, at which point an aliquot was removed, diluted, and plated on rich media to determine cell survival (Figure 4A). The same was done to examine the interaction between cdc7Δ and rev3Δ (Figure 4B) or rad30Δ (Figure 4C).
We find that the MMS sensitivity of the pol30-46 strains generated here is not as strong as in previously published reports (Xiao et al. 2000; Broomfield and Xiao 2002). We note, however, that there are several differences between the strains used, including genetic background and the method used to introduce the pol30-46 mutation (see materials and methods). Furthermore, differences in the MMS reagents used could be responsible for the discrepancy observed, as the UV sensitivity of the pol30-46 strains is similar in both studies. When we examine the interaction between cdc7Δ and pol30-46, we find that the double-mutant strain is significantly more sensitive than either single mutant (Figure 4A). Analysis of the survival data (Table 4) indicates that this interaction is slightly stronger than additive at lower MMS doses (10 and 20 min), but no more so at a higher dose (30 min). Therefore, as observed above in response to UV damage, we conclude that the interaction between cdc7Δ and pol30-46 in response to MMS treatment is at least additive.
Analysis of the MMS sensitivity of the cdc7Δ rev3Δ strain revealed a very different picture from what was observed in response to UV damage. In this case, the double mutant is significantly more sensitive than either single mutant, which exhibit very similar killing profiles in response to MMS exposure. The survival data for the double mutant (Table 4) also fluctuate between being slightly more and slightly less than what is expected of an additive effect. However, we conclude that cdc7Δ and rev3Δ exhibit an additive interaction.
Previous analyses of MMS sensitivity of a rad30Δ strain using a similar assay to the one used here have had conflicting outcomes; in one study, a rad30Δ strain is more sensitive than WT to MMS treatment (Roush et al. 1998), whereas in a second study the rad30Δ strain behaves no differently from a WT strain (Broomfield and Xiao 2002). We find that the rad30Δ strains generated in this report are as sensitive as WT to MMS killing. However, when combined with cdc7Δ, the double-mutant strain shows a significant increase in sensitivity compared to the cdc7Δ single mutant. This suggests that RAD30 plays a minor role in the response to MMS treatment and that this role is separate from CDC7. In conclusion, our analysis of the interactions between CDC7 and representative genes of distinct branches within the RAD6 pathway indicates a distinct role for CDC7 in response to MMS treatment.
The DNA damage checkpoint is intact in the absence of CDC7:
One explanation for the phenotypes of cdc7Δ strains in response to DNA-damaging agents would be the possible role of the Cdc7/Dbf4 protein complex in checkpoint function. This aspect of CDC7 function in genome maintenance is not well understood (Jares et al. 2000; Sclafani 2000). Initial studies with cdc7ts mutants demonstrated that the DNA damage checkpoint was intact (Siede et al. 1994; Ostroff and Sclafani 1995). However, because of the low sensitivity to UV light and possible leakiness of the hypomorphic alleles examined, we decided to reexamine the status of the DNA damage checkpoint in a cdc7Δ strain.
Previous analysis of a rad9Δ strain, which lacks a functional DNA damage checkpoint, showed that cells progress into S phase of the cell cycle independently of the presence of DNA damage. WT cells exposed to UV irradiation, on the other hand, showed a transient delay before progressing into S phase, when compared to unirradiated controls (Siede et al. 1993). We used a similar assay to determine if the G1/S cell cycle delay caused by exposure to DNA-damaging agents is still present in a cdc7Δ strain. To do so, logarithmically growing cultures were first synchronized in G1 using α-factor. The cultures were split into two aliquots and, immediately after release from the G1 arrest, one aliquot was exposed to UV light. Then, at various time points, samples were collected for analysis of DNA content, allowing us to determine their progress through the cell cycle. When exposed to UV light, WT cells exhibited a delayed entry into S phase of the cell cycle. Cells that have not been exposed to UV light reach the G2 phase of the cell cycle when UV-treated cells enter S phase (Figure 5A, compare WT + UV vs. −UV at 60 min). The same effect of UV exposure is observed for cdc7Δ cells (Figure 5B). It is not until 60 min after α-factor release that we begin to see a shift in the DNA peak of UV-treated cdc7Δ cells, at which point the nontreated control is clearly progressing through S phase. After some time, irradiated cdc7Δ cultures overcome the cell cycle block and resume normal growth, eventually reaching stationary phase.
Second, we wanted to determine if the delay in cell cycle entry correlated with activation of the DNA damage checkpoint in response to UV exposure. To that end, we examined the phosphorylation status of the Rad53 protein, a key component of G1/S, intra-S, and G2/M checkpoints in S. cerevisiae (reviewed in Nyberg et al. 2002). We found that in WT or cdc7Δ cells that were exposed to UV light (Figure 5C) there was an upward shift of the Rad53 band migration, indicative of hyperphosphorylation of the protein and activation of the DNA damage checkpoint. Taken together, we conclude that the DNA damage checkpoint is intact and that the sensitivity of cdc7Δ cells to DNA-damaging agents is a result of the lack of Cdc7 function in DNA repair/damage tolerance mechanisms.
The role of CDC7 in DNA damage tolerance is poorly understood. Previous data on induced mutagenesis and epistasis analysis indicated that CDC7 belongs to the RAD6 epistasis group, most likely within the TLS pathway. Accordingly, CDC7-mediated induced mutagenesis is restricted to the S phase of the cell cycle in agreement with its kinase activity profile (Ostroff and Sclafani 1995; Oshiro et al. 1999; Weinreich and Stillman 1999). It is also known that different alleles of cdc7 are either hyper- (cdc7-3, -4, -23) or hypomutagenic (cdc7-1, -7), even though they all exhibit reduced activity in DNA replication (Hollingsworth et al. 1992). Cdc7-Dbf4 kinase activity is required for mutagenesis, as a “kinase-dead” allele is defective in the process (Hollingsworth et al. 1992). This suggests a difference in affinity for downstream substrates of Cdc7 kinase in induced mutagenesis, although the identity of these is not known (Sclafani 2000). To gain a better understanding of the role of CDC7 in DNA damage tolerance, we have carried out an extensive analysis of the genetic interactions between CDC7 and members of the RAD6 epistasis group.
The mcm5-bob1 mutation does not affect CDC7-mediated DNA damage tolerance:
To avoid the problems of using a cdc7 hypomorphic allele in epistasis analysis, we took advantage of the fact that, in the presence of the mcm5-bob1 mutation, we are able to delete CDC7. Our analysis of the mcm5-bob1 mutation in response to DNA damage shows that it does not affect the sensitivity of yeast strains to UV or MMS, either by itself or in combination with other mutations in the RAD6 epistasis pathway (Figure 1). Furthermore, previous work in our lab has also determined that mcm5-bob1 has no effect on induced mutagenesis, either by itself or in combination with cdc7, suggesting that the bypass of CDC7 is specific to DNA replication (Pahl 1994). Thus we have taken advantage of the mcm5-bob1 mutation as a genetic tool to study the role of CDC7 in DNA damage tolerance.
CDC7 is specifically associated with error-prone mechanisms of DNA damage tolerance:
RAD5 and POL30 (as indicated by the pol30-46 allele) represent two error-free pathways for DNA damage tolerance that are thought to rely on recombination/copy-choice mechanisms and are inherently nonmutagenic. Analysis of strains that combine cdc7Δ with rad5Δ, or cdc7Δ and pol30-46, revealed that they are more sensitive to UV irradiation and MMS exposure, compared to the single-mutant strains. The cdc7Δ rad5Δ strain showed a synergistic increase in sensitivity in response to the UV irradiation and MMS exposure, whereas the interaction between cdc7Δ and pol30-46 gives only an additive decrease in cell survival.
The results of this genetic analysis, together with data in the literature (Ulrich and Jentsch 2000; Xiao et al. 2000), indicate that RAD5, POL30, and CDC7 all function in separate pathways for DNA damage tolerance in S. cerevisiae (Figure 6). Furthermore, the synergism between cdc7Δ and rad5Δ (Table 3 and Figure 3) indicates that the two pathways compete for a common substrate resulting from DNA damage. The additive interaction between cdc7Δ and pol30-46 (Tables 3 and 4) suggests that the affected pathways are independent from one another downstream of the point where they are blocked. However, it does not preclude the possibility that the initial substrate resulting from DNA damage is common (Cox and Game 1974). Finally, the data presented here infer that CDC7 function is restricted to the TLS branch of DNA damage tolerance, an inherently error-prone mechanism that can result in the introduction of mutations, consistent with previous analyses (Njagi and Kilbey 1982; Hollingsworth et al. 1992).
The genetic interaction between CDC7 and TLS polymerases is dependent on the nature of the DNA damage:
The biochemical characterization of Pol η, encoded by the RAD30 gene, and its role in xeroderma pigmentosum syndrome, suggest that it is a DNA polymerase specifically suited for the error-free bypass of cis-syn (T-T) dimers (Johnson et al. 1999a; Masutani et al. 1999a,b; Washington et al. 2000). However, it has been shown to contribute to the translesion of many DNA damage structures, including an O6-methylguanine, and N-2-acetylaminofluorene modified guanine, although with lower efficiency (Haracska et al. 2000; Bresson and Fuchs 2002). These characteristics could account for the phenotypes of a rad30Δ strain, which shows a significant sensitivity to UV irradiation, but not other DNA damaging agents (this study; Roush et al. 1998; Haracska et al. 2000; Xiao et al. 2000; Broomfield and Xiao 2002).
Our analysis of the genetic interaction between rad30Δ and cdc7Δ indicates that the double mutant shows an additive or even slightly stronger increase in UV sensitivity (Table 3). This suggests that RAD30 and CDC7 function in separate pathways dealing with UV damage substrates and could reflect a specificity of the RAD30 pathway for the bypass of cis-syn (T-T) dimers. The CDC7-mediated pathway, on the other hand, would deal primarily with other UV-induced damage structures. However, genetic studies have shown that RAD30 is also involved in mutagenic bypass of a (6-4) T-T dimer (Bresson and Fuchs 2002). Given that CDC7 is also required for UV-induced mutagenesis, it is possible that they function in separate pathways independently of the UV-induced substrate.
As mentioned above, the rad30Δ strain shows no increased sensitivity to MMS treatment, compared to a WT strain. However, the cdc7Δ rad30Δ strain is significantly more sensitive than a cdc7Δ strain to MMS. This suggests a very minor role of RAD30 in response to MMS. The phenotype detected here is similar to the observation made in the study of the role of RAD30 in MNNG-induced mutagenesis, which became apparent only when the rad30Δ was combined with a pol32Δ (Haracska et al. 2000). On the basis of our results, we propose that CDC7 and RAD30 function separately in response to MMS.
The analysis of a cdc7Δ rev3Δ strain indicates that cdc7Δ is epistatic to rev3Δ in response to UV irradiation (Figure 1), but shows an additive interaction in response to MMS treatment (Figure 3). This suggests that other cellular components contribute to the CDC7 pathway within the RAD6 epistasis group. One possibility is that Pol δ also contributes to CDC7-mediated DNA damage tolerance. POL32, a subunit of Pol δ, has been shown to be required for UV-, MMS-, and MNNG-induced mutagenesis, and pol3-13, a temperature-sensitive allele of the main subunit of Pol δ, is also defective in UV-induced mutagenesis. In addition, genetic analysis of pol32Δ and pol3-13 determined that these two genes are in the same pathway as REV3 (Giot et al. 1997; Haracska et al. 2000, 2001b; Huang et al. 2000) in response to UV damage. pol32Δ and rev3Δ have also been shown to be epistatic in response to MMS treatment. However, while POL32 is required for MNNG-induced mutagenesis, REV3 is not (Haracska et al. 2000; Huang et al. 2000).
The most striking observation from this study is the dependence of the genetic interactions between cdc7Δ and rad30Δ or rev3Δ on the type of DNA-damaging agent used. Most likely, this is a reflection of the variety of DNA damage structures that can arise from treatment with UV or MMS. As has been shown from in vitro and in vivo studies, Pol η (RAD30) and Pol ζ (REV3/REV7) show marked differences in dealing with specific DNA damage structures. In vitro studies, however, do not necessarily reflect what is happening inside the cell. For example, the interaction between Rad30 and PCNA has been shown to be essential for the function of the polymerase in vivo, but this requirement is not seen in in vitro bypass assays of a cis-syn (T-T) dimer (Haracska et al. 2001a). Also, both RAD30 and REV3 are required for bypass of a (6-4) T-T dimer, and it has been proposed that they function together in this process (Bresson and Fuchs 2002). However, the genetic analysis of rad30Δ and rev3Δ strains and the interaction between these two deletions do not support such a model (reviewed in Lawrence 2002).
The relationship between the many cellular components involved in TLS, as is understood now, is not very clear. The in vivo assays used to analyze the requirement of RAD30 and REV3 in the bypass of specific DNA damage structures can offer a strong insight into the mechanism of TLS. It seems clear that use of these assays to test the requirements of other TLS components, such as REV1, POL32, POL30, and now CDC7 and/or DBF4 will only add to our understanding of this important cellular process for dealing with the presence of DNA damage.
From the data presented here, we propose a model whereby the interaction between TLS components is dependent on the type of lesion encountered by the replication machinery. In the case of UV irradiation, or specific DNA damage structures resulting thereof, Cdc7 plays a role in the regulation of the Rev3/Rev7 pathway (Figure 6A). In other cases, such as alkylation damage, Cdc7 seems to be regulating a previously unidentified pathway (Figure 6B).
Activation of the DNA damage checkpoint in S. cerevisiae does not require CDC7:
Finally, we address the idea that Cdc7 is involved in the DNA damage checkpoint and the possible implications on the analysis of these results. Work in Xenopus laevis and Schizosaccharomyces pombe has shown that Cdc7 is important for checkpoint activation as a transducer and/or a target of checkpoint signaling (Jares et al. 2000; Snaith et al. 2000; Costanzo et al. 2003). Recently, it was shown that Cdc7/Dbf4 kinase is required for an etoposide-induced DNA damage checkpoint in the Xenopus system (Costanzo et al. 2003). Lack of checkpoint function is one explanation for the DNA damage sensitivity and mutagenesis phenotypes observed in cdc7 mutants. To eliminate this possibility, we examined the status of the DNA damage checkpoint in cdc7Δ strains and found that both the G1- to S-phase transition delay and the activation of Rad53 in response to UV exposure are intact (Figure 5). This is in agreement with experiments that show that the intraS-phase checkpoint is intact in cdc7Δ mcm5-bob1 cells (Weinreich and Stillman 1999) and with recent work showing that S. cerevisiae Cdc7 is not required for checkpoint activation, maintenance, or downregulation in response to hydroxyurea (HU) or MMS treatment using cdc7ts strains at the restrictive temperature (Tercero et al. 2003).
However, we cannot completely rule out the possibility that Cdc7 is a downstream target of the checkpoint. Dbf4 and Hsk1 (Sp Cdc7) are phosphorylated in a Rad53/Cds1 HU-treatment-dependent manner in S. cerevisiae and in S. pombe, respectively (Weinreich and Stillman 1999; Snaith et al. 2000). Furthermore, other checkpoint proteins have been shown to have a role in DNA damage-induced mutagenesis (Paulovich et al. 1998; Kai and Wang 2003). We recognize the possibility that the Cdc7 function in DNA damage tolerance is induced by checkpoint activation, but argue that this function is directly involved in the TLS mechanism.
What is the role of Cdc7 function in TLS?
Previous data from our laboratory, and the work presented here, reveal a role for Cdc7 protein and its kinase activity in TLS. In DNA replication, Cdc7 may phosphorylate Mcm2 protein, a subunit of the hexameric MCM complex. The mcm5-bob1 mutation bypasses the requirement of Cdc7 in DNA replication, but not in induced mutagenesis using cdc7ts, cdc7Δ, and cdc7Δ dbf4Δ strains. In addition, mcm5-bob1 on its own has no effect on cell survival or induced mutagenesis (Pahl 1994), and there is no evidence implicating the involvement of other MCM subunits in DNA damage tolerance. This suggests that the substrate for Cdc7-Dbf4 in TLS may be different from its substrate in DNA replication.
It is thought that TLS occurs via a DNA polymerase switch, whereby the replicative polymerase is substituted by another one capable of bypass, allowing the replication fork to progress through the damage (Kunz et al. 2000; Broomfield et al. 2001; Prakash and Prakash 2002). It has been proposed that, in some cases, this switch occurs more than once, as more than one polymerase might be required for efficient bypass. Although it is not known how this exchange occurs, or how it is regulated, it is reasonable to expect that it involves proteins already present at the replication fork and/or others that are brought to it when replication stalls. The target of Cdc7 phosphorylation is likely to be one of these—possibly the bypass polymerases themselves or an accessory protein, such as Pol32, Rev1, Rev7, or PCNA.
We thank Peter Burgers for the generous gift of the plasmid pBL230-46. We thank the University of Colorado Cancer Center Core facility for performing the FACS analysis. DNA samples were sequenced by the University of Colorado Cancer Center DNA Sequencing and Analysis Core facility, which is supported by the National Cancer Institute core support grant CA46934. This work was supported by grant GM35078 from the Public Health Service awarded to R.A.S. and in part by training grant T32-GM08730 from the National Institutes of Health.
Communicating editor: T. Stearns
- Received August 27, 2003.
- Accepted March 18, 2004.
- Genetics Society of America