|Home | About | Journals | Submit | Contact Us | Français|
The Rad6-Rad18 complex mono-ubiquitinates proliferating cell nuclear antigen (PCNA) at the lysine 164 residue after DNA damage and promotes DNA polymerase η (Polη)- and Polζ/Rev1-dependent DNA synthesis. Double-strand breaks (DSBs) of DNA can be repaired by homologous recombination (HR) or non-homologous end-joining (NHEJ), both of which require new DNA synthesis. HO endonuclease introduces DSBs into specific DNA sequences. We have shown that Polζ and Rev1 localize to HO-induced DSBs in a Mec1-dependent manner and promote Ku-dependent DSB repair. However, Polζ and Rev1 localize to DSBs independently of PCNA ubiquitination. Here we provide evidence indicating that Rad18-mediated PCNA ubiquitination stimulates DNA synthesis by Polζ and Rev1 in repair of HO-induced DSBs. Ubiquitination defective PCNA mutation or rad18Δ mutation confers the same DSB repair defect as rev1Δ mutation. Consistent with a role in DSB repair, Rad18 localizes to HO-induced DSBs in a Rad6-dependent manner. Unlike Polζ or Rev1, Polη is dispensable for repair of HO-induced DSBs. Ku and DNA ligase IV constitute a central NHEJ pathway. We also show that Polζ and Rev1 act in the same pathway as DNA ligase IV, suggesting that Polζ and Rev1 are involved in DNA synthesis during NHEJ. Our results suggest that Polζ-Rev1 accumulates at regions near DSBs independently of PCNA ubiquitination and then interacts with ubiquitinated PCNA to facilitate DNA synthesis.
DNA damage that stalls DNA replication poses a major threat to genomic stability and cellular viability. To tolerate such replication blocks, cells are equipped with specialized DNA polymerases that can bypass DNA damage by a process known as translesion synthesis (TLS) [1, 2]. The TLS DNA polymerases include DNA polymerase η (Polη) and Polζ. Polη, encoded by RAD30 in budding yeast and XPV in human, performs relatively error-free bypassing of pyrimidine dimers [3, 4]. Polζ consists of the catalytic subunit Rev3 and the accessory subunit Rev7 [5-7]. In budding yeast, Polζ interacts physically with Rev1 , a deoxycytidyl transferase that inserts a C opposite an abasic site . It has been proposed that Rev1 carries out the insertion step across from the lesion and that Polζ subsequently extends from the lesion . Recent evidence indicates that Polη and Polζ-Rev1 are also involved in other repair pathways than TLS [8, 10, 11].
In budding yeast, Rad6 and Rad18 are required for replication of damaged DNA templates . Rad6, a ubiquitin-conjugating enzyme, exists in vivo in a tight complex with Rad18 [12, 13]. Proliferating cell nuclear antigen (PCNA) is the eukaryotic sliding clamp required for processive DNA synthesis . The Rad6-Rad18 complex mediates mono-ubiquitination at the lysine 164 residue of PCNA  and promotes DNA synthesis by Polη and Polζ after DNA damage [16, 17]. However, some instances of DNA synthesis may not require modification of lysine 164 . PCNA is loaded onto DNA by replication factor C, which couples ATP hydrolysis with the opening and closing of the PCNA ring around the DNA . PCNA loaded on DNA can be efficiently ubiquitinated by the Rad6-Rad18 complex [19, 20]. Ubiquitination of PCNA might create a platform to recruit the TLS polymerases [21, 22]. Alternatively, ubiquitinated PCNA could function as a processivity factor for the TLS polymerases . We have shown that Polζ and Rev1 localize to DSBs and contribute to Ku-dependent repair of DSBs . Polζ and Rev1 are found to associate with DSBs in a Mec1-dependent manner but independently of PCNA ubiquitination . However, it remains to be determined whether PCNA ubiquitination activates Polζ/Rev1-dependent DNA synthesis in DSB repair.
In this study, we have investigated whether Rad18 or PCNA ubiquitination is required for Polζ/Rev1-dependent DSB repair. Although accumulation of Polζ and Rev1 at DSBs does not require PCNA ubiquitination, DSB repair activities of Polζ and Rev1 are fully dependent on Rad18 function. Consistent with the role in DSB repair, Rad18 localizes to sites near DSB lesions. These results suggest that Rad18 activates the Polζ/Rev1 pathway after DSB induction.
To generate pGAL-ACT1, the ACT1 coding sequence was amplified by polymerase chain reaction (PCR) and cloned into YCpG22 . YCpG22, containing a GAL1-10 promoter region, is a derivative of YCplac22 . YCpA-GAL-HO was described previously . YCpT-GAL-HO is obtained from J.A. Nickoloff. Epitope tagging of RAD18 was performed by a PCR-based strategy . Cells expressing myc-tagged Rad18 protein are as resistant to UV irradiation as wild-type cells, indicating that myc-tagged Rad18 protein is fully functional (data not shown). Cells carrying dnl4Δ, pol4Δ, rad1Δ, rad6Δ, rad30Δ or siz1Δ mutation were obtained by transformation with PCR fragments marked with HphMX4 , KanMX4 , TRP1  or URA3 . Disruption of each gene was confirmed by PCR. Constructions of other mutations are described elsewhere [8, 29]. The strains used here are listed in Table 1.
Sensitivity to HO-induced DSBs was examined as previously described [8, 30]. Cells were transformed with the YCpA-GAL-HO plasmid, and grown overnight in sucrose-medium selectable for the plasmid. Cultures were serially diluted in water and aliquots of diluted cultures were spread out on plates containing glucose (2%) or sucrose-galactose (2%) medium selectable for the plasmid. HO expression from YCpA-GAL-HO is inducible in sucrose-galactose medium, but is repressed by the addition of glucose. Independent transformants were examined at least three times, and viability was calculated in a condition where 100-200 colonies are formed on a plate. Dilution experiments showed that survivors form colonies proportionally if less than 2×106 cells were spread out on a plate. Sensitivity to HO-induced DSBs results in a reduced plating efficiency on inducible galactose plates relative to repressible glucose plates. In this assay, cells suffer continuous HO expression. However, cells can survive after HO expression if they carry mutations in the HO recognition sequence or lose the HO cleavage site at the ADH4 locus. Rad52-dependent recombination could delete the HO recognition site, because cells contain no donor sequence containing the HO cleavage site. We note that all the genes distal to the ADH4 locus on chromosome VII are not essential for cell proliferation. During NHEJ, annealing of ssDNA regions can delete non-homologus sequences , resulting in loss of the HO cleavage site. In addition, annealing of ssDNA tracts could generate gaps, which are filled in by DNA synthesis. Subsequently, error-prone DNA synthesis might introduce mutations at the HO cleavage site. Therefore, cell survival after HO expression depends on DSB repair processes. However, this assay system fails to monitor simple joining of DSB ends, which regenerates the HO cleavage site.
Chromatin immunoprecipitation was performed as previously described  using anti-myc (9E10) antibodies. The PCR reaction was performed under non-saturating conditions, in which the rate of PCR amplification was proportional to the substrate concentration and cycling. The sequences of primers for the HO set at the ADH4 locus were 5’-TCTATTAATGAGCCGAGACCGGTA-3’ and 5’-CGCATGTGAATGACACACGAAAGT-3’, those for the SMC2 locus were 5’-AAGAGAAACTTTAGTCAAAACATGGG-3’ and 5’-CCATCACATTATACTAACTACGG-3’.
We have shown that Polζ (the Rev3-Rev7 complex) and Rev1 are required for Ku-dependent repair of DSBs , using an experimental system in which HO endonuclease induces a single DSB at the ADH4 locus (Fig. 1A). Ku and DNA ligase IV constitute a central NHEJ pathway [32, 33]. In budding yeast, DNA ligase IV is composed of the catalytic subunit Dnl4 (Lig4) and the co-factor Lif1 . To further support that Polζ and Rev1 act in DSB repair, we examined whether REV1 and REV3 act in the same pathway as DNL4 (Fig. 1B). In budding yeast, DSBs are repaired primarily by the Rad52-dependent HR pathway; therefore, the efficiency of NHEJ can be assessed in a rad52Δ background [30, 34, 35]. In the assay system (Fig. 1A), wild-type and rad52Δ mutant cells retain 20% and 0.2% viability after HO expression, respectively . Cells carrying the rad52Δ mutation or those containing the rev1Δ, rev3Δ, dnl4Δ or hdf1Δ mutation were transformed with the GAL-HO plasmid. Transformed cells were grown in sucrose to maintain the GAL promoter inactive. Aliquots of the culture were then plated on medium containing galactose to induce HO expression or medium containing glucose to repress HO expression. Consistent with the idea that Ku and DNA ligase IV constitute a central NHEJ pathway, introduction of the hdf1Δ single, dnl4Δ single or hdf1Δ dnl4Δ double mutation into rad52Δ mutants caused a further 20-fold reduction in survival after DSB induction. The rev1Δ or rev3Δ mutation decreased viability of rad52Δ mutants by 5-fold , but neither mutation decreased viability of rad52Δ dnl4Δ double mutants. In budding yeast, Rad1 contributes to DSB repair independently of Ku or Rad52 . If rad52Δ hdf1Δ cells were survived by DSB repair, introduction of the rad1Δ mutation should further reduce the cell viability. Consistently, rad52Δ hdf1Δ rad1Δ triple mutants were more sensitive (3-fold) than rad52Δ hdf1Δ double mutants (Fig. 1B). Taken together, these observations support the possibility that Polζ and Rev1 contribute to DNA synthesis in NHEJ of HO-induced DSBs.
The viability assay here depends on the hypothesis that cells survived from continuous HO expression after aberrant repair of the HO cleavage site. However, some rad52Δ hdf1Δ rad1Δ triple mutants carrying the GAL-HO plasmid proliferated on galactose medium. We addressed how the rad52Δ hdf1Δrad1Δ triple mutants survived on galactose medium. There are two explanations for cell survival without DSB repair. One possibility could be that cells lost the intact GAL-HO plasmid; for example, they might have received mutations in the GAL-HO construct or reversion mutations in the ade1 mutation on the chromosome. The GAL-HO plasmid used for the viability assay carries ADE1 as a selection marker. We thus re-tested whether rad52Δ hdf1Δ rad1Δ survivors retain viability on galactose medium by introducing a TRP1-marked GAL-HO plasmid (Table 2). Another possibility could be that cells obtained mutations in genes involved in galactose metabolism, thereby causing a defect in HO expression. We addressed this possibility by introducing a TRP1-marked GAL-ACT1 plasmid (Table 2). ACT1 encodes actin, which is one of major components of the cytoskeleton. Overexpression of ACT1 from the GAL promoter blocks cell proliferation . Most of the rad52Δ hdf1Δ rad1Δ survivors carrying the GAL-ACT1 plasmid failed to proliferate on galactose medium. One out of 39 tested survivors proliferated well on galactose, suggesting that this survivor acquired mutation(s) affecting galactose metabolism. Correspondingly, this survivor retained viability on galactose even after transformation with the TRP1-marked GAL-HO plasmid. Besides, only one survivor retained viability on galactose medium after introduction of the TRP1-marked GAL-HO plasmid. Thus, rad52Δ hdf1Δ rad1Δ survivors principally lost the intact GAL-HO plasmid for cell proliferation, suggesting that rad52Δ hdf1Δ rad1Δ mutants are defective in repair of HO-induced DSBs.
We also examined survivors from each of rad52Δ single, rad52Δ rev1Δ double or rad52Δ hdf1Δ double mutants (Table 2). All the rad52Δ survivors carrying the GAL-ACT1 plasmid did not grow well on galactose. However, most survivors carrying the TRP1-marked GAL-HO plasmid retained viability on galactose medium. These results suggest that rad52Δ survivors largely received mutations in the HO recognition site. Similar results were obtained with the rad52Δ rev1Δ survivors. Notably, half of rad52Δ hdf1Δ survivors appeared to acquire mutations at the HO restriction site. These results are consistent with the current view that the Ku-dependent NHEJ pathway plays an important role in DSB repair in the absence of the HR pathway and Rad1 contributes to DSB repair as a third pathway in budding yeast.
The Rad6-Rad18 complex ubiquitinates PCNA at Lys-164, and thereby activates Polη and Polζ functions in budding yeast. Indeed, neither rad30Δ nor rev3Δ mutations increase sensitivity to MMS or UV light of cells carrying the ubiquitination-defective PCNA (pol30-K164R) mutation [16, 17]. We therefore examined the effect of pol30-K164R mutation on the DSB repair as above (Fig. 2A). Although the pol30-K164R mutation alone did not significantly affect cellular death after DSB induction (data not shown), it decreased viability of rad52Δ cells similarly to the rev1Δ mutation. The pol30-K164R mutation did not further enhance the viability loss in rad52Δ hdf1Δ double mutants. Introduction of the rev1Δ mutation did not affect viability in rad52Δ pol30-K164R double mutants. Thus, the pol30-K164R mutation behaves like the rev1Δ mutation in DSB repair. We next examined the effect of rad18Δ mutation on the DSB repair (Fig. 2B). The rad18Δ mutation conferred the same phenotype as the pol30-K164R mutation; both mutations increased cell death in rad52Δ mutants but did not affect viability in rad52Δ hdf1Δ double mutants after DSB induction. Moreover, introduction of the pol30-K164R mutation did not increase viability loss in rad52Δ rad18Δ double mutants. Again, rev1Δ mutation did not affect the rad52Δ rad18Δ double mutants. Cells carrying the rad52Δ rad6Δ double or rad52Δ rad6Δ rad18Δ triple mutation showed the same viability loss as rad52Δ rad18Δ double mutants, consistent with the view that Rad6 and Rad18 act as a complex in the same pathway (Fig. 2C). PCNA is also SUMOylated by the Siz1-Ubc9 complex at Lys-164 . Both PCNA ubiquitination and SUMOylation have been shown to contribute to Polζ/Rev1-dependent DNA synthesis . We monitored effects of siz1Δ mutation on cellular viability in rad52Δ and rad52Δ rad18Δ backgrounds (Fig. 2D). No apparent defect was associated with the siz1Δ mutation, suggesting that Siz1-dependent SUMOylation is not essential for repair of HO-induced DSBs. Together, these results support a model in which Rad6-Rad18 complex promotes the Polζ/Rev1-dependent DSB repair by ubiquitinating PCNA.
The Rad6-Rad18 complex binds to DNA [12, 13], and ubiquitinates PCNA loaded on DNA [19, 20]. To confirm the function of the Rad6-Rad18 complex in DSB repair, we investigated whether the Rad6-Rad18 complex localizes to the HO-induced DSB by chromatin immunoprecipitation (ChIP) assay. Cells expressing Rad18-myc were transformed with the GAL-HO plasmid . Transformants were grown initially in sucrose to repress HO expression and then incubated with galactose to induce HO expression. Cells were collected after a 3 hr incubation with galactose, and extracts prepared after formaldehyde cross-linking were subjected to immunoprecipitation with anti-myc antibodies. Co-precipitated DNA was extracted and amplified by polymerase chain reaction (PCR) using the HO1 primer set to amplify a region near the HO restriction site in the ADH4 locus on chromosome VII. As a control, PCR was performed to amplify the SMC2 locus on chromosome VI, which contains no cleavage site (Fig. 1A). PCR amplification with the HO1 primer set was detected in cells carrying the GAL-HO plasmid after incubation with galactose (Fig. 3A). In contrast, there was no PCR amplification from the SMC2 locus after incubation with galactose (Fig. 3A). PCR amplification was not observed in untagged cells or cells lacking an HO cleavage site at the ADH4 locus (data not shown). These results indicate that Rad18 associates with sites near the HO-induced DSB. To address the role of Rad6-Rad18 complex formation, we next examined the effect of rad6Δ mutation on Rad18 association with the HO-induced DSB. Rad18 association with the DSB was significantly decreased in rad6Δ mutants (Fig. 3A). This association defect was not due to decreased expression of Rad18; the rad6Δ mutation did not affect the expression level of Rad18 (data not shown). These results suggest that Rad18 associates with DSBs in the form of the Rad6-Rad18 complex, and support the role of the Rad6-Rad18 complex in DSB repair.
The ATR homolog Mec1 plays a central role in DNA damage checkpoint responses in budding yeast . We have shown that Polζ and Rev1 associate with regions near DSBs in a Mec1-dependent manner . We therefore asked whether Mec1 also controls the Rad18 association with DSBs (Fig. 3B). However, Rad18 association with DSBs was not impaired in mec1Δ mutants. The ATM homolog Tel1 is also involved in the cellular response to DSBs, although Tel1 plays a minor role . Checkpoint response to DSBs is essentially abolished in mec1-81 tel1Δ double mutants . We then examined the effect of mec1-81 tel1Δ mutations on Rad18 association with DSBs (Fig. 3B). Rad18 associated efficiently with DSBs even in mec1-81 tel1Δ mutants. Thus, activation of the checkpoint responses is not required for Rad18 localization to the DSB.
Ubiquitination of PCNA activates the Polη (Rad30) pathway as well as the Polζ-Rev1 pathway in budding yeast [16, 17]. Similar to rev1Δ or rev3Δ mutation , rad30Δ mutation by itself did not affect cellular viability in the presence of Rad52 after DSB induction (data not shown). We then examined the effect of rad30Δ mutation on the DSB repair in rad52Δ mutants (Fig. 4A). Unlike rev1Δ or rev3Δ mutation, introduction of the rad30Δ mutation did not enhance viability loss in rad52Δ mutants. Moreover, the rad30Δ mutation did not increase cell death in rev1Δ rad52Δ double mutants. These results are consistent with the above finding that the viability loss in rev1Δ rad52Δ or rev3Δ rad52Δ double mutants is not distinguishable from that of pol30-K164R rad52Δ or rad18Δ rad52Δ double mutants (see Fig. 1 and Fig. 2). We note that rad52Δ rev1Δ rad30Δ triple mutants are more sensitive to UV light than either of rad52Δ rev1Δ or rad52Δ rad30Δ double mutants (data not shown). Thus, Rad30 does not play an apparent role in repair of the HO-induced DSB.
Pol4, a family X DNA polymerase, plays a critical role in DNA synthesis during NHEJ [39, 40]. To address the relationship between the Polζ-Rev1 and Pol4 pathways, we investigated the effect of rev1Δ and pol4Δ mutations on the DSB repair in rad52Δ mutants (Fig. 4B). Consistent with the previous finding , introduction of the pol4Δ mutation enhanced viability loss in rad52Δ mutants. Interestingly, the rev1Δ pol4Δ double mutation caused more significant viability loss in rad52Δ mutants than either of the rev1Δ or pol4Δ single mutation. These results suggest that Polζ-Rev1 and Pol4 act redundantly during DNA synthesis in repair of HO-induced DSBs.
Rad18-mediated PCNA ubiquitination has been proposed to promote Polζ- and Rev1-dependent DNA synthesis. PCNA ubiquitination has been shown to mediate Rev1-PCNA interaction . However, localization of Polζ and Rev1 to DSB lesions does not require Rad18 function or PCNA ubiquitination . It has thus remained possible that Polζ and Rev1 contribute to DSB repair independently of PCNA ubiquitination. In this study, we have used a system in which cells receive pure DSBs after HO expression and found that the Rad18-mediated PCNA ubiquitination activates the Polζ/Rev1 pathway in DSB repair as well. These observations support the model in which Polζ-Rev1 first accumulates at regions near DSBs and then interacts with ubiquitinated PCNA to facilitate DNA synthesis. Exposure to DSB inducing agents, for example ionization radiation (IR), also induces single-strand breaks, base modifications and DNA adducts, which can be repaired by the TLS pathway. For example, the introduction of a rev1Δ mutation significantly reduces viability of rad52Δ hdf1Δ double mutant cells after IR (data not shown). Therefore, it is difficult to characterize functions of Polζ-Rev1 and Rad18 in DSB repair using IR. HO endonuclease, which generates pure DSBs, enabled us to investigate a role of Polζ-Rev1 and Rad18 in DSB repair.
If simple re-joining of DNA ends does not occur, DSBs are processed by exonuclease activities, generating ssDNA tract at the DSB ends. The Rad6-Rad18 complex has been shown to bind to single-stranded DNA in vitro [12, 13]. This biochemical property of the Rad6-Rad18 complex is consistent with our observation that Rad18 localizes to DSBs. During the revision of this paper, Davies et al.  also reported that Rad18 localizes to DSBs. Requirement of Rad18 for DSB repair has been shown in chicken DT 40 cells , suggesting that Rad18 associates with DSB lesions in higher eukaryotes as well. The Rad6-Rad18 complex has been proposed to ubiquitinate PCNA loaded on DNA [19, 20], because only a small fraction of PCNA is modified after DNA damage. Even though Rad18 localizes to DSBs, PCNA modification has not been detected in cells suffering HO-induced or EcoRI-induced DSBs by immunoblotting analysis (data not shown). It is not known what fraction of PCNA is loaded on DNA at a DSB lesion or how many DSBs are generated after EcoRI expression. If only a few PCNA molecules are loaded on DNA, it might be difficult to detect PCNA ubiquitination.
Annealing of ssDNA tracts at DNA ends may generate gaps with 3’-termini. Replication factor C can load PCNA on gapped DNA substrates . The Rad6-Rad18 and Polζ-Rev1 complexes accumulate at sites near DSB ends . Once loaded, PCNA could be ubiquitinated by the Rad6-Rad18 complex. In turn, ubiquitinated PCNA might stimulate the activity of nearby Polζ or Rev1 . Previous studies have demonstrated that Pol4 plays an important role in gap-filling, by extending the DNA strand from mismatched 3’-termini [39, 40]. Similarly, Polζ efficiently extends from mismatched termini . Polζ is a processive enzyme , whereas Pol4 is not . Thus, Polζ-Rev1 and Pol4 could play distinct roles in gap-filling during NHEJ. Recent studies have established the model in which the cyclin-dependent kinase, Cdc28, stimulates resection of DNA ends to generate ssDNA tracts at DSBs [47, 48]. This model explains why NHEJ operates preferentially in G1. In the absence of DNA degradation, however, NHEJ could rejoin the DNA ends without gap-filling. Thus, DNA polymerase activities might be more dispensable for NHEJ in G1 phase than in other cell-cycle stages. Consistently, Rev1 and Rad18 do not play an apparent role in repair of HO-induced DSBs in G1 (data not shown), although Ku plays a critical role . Since Mec1 localizes to sites of DNA damage by interacting with RPA-coated ssDNA [50, 51], Mec1 associate with DSBs in S and G2/M phase more efficiently than in G1 phase . We have shown that Mec1-dependent phosphorylation promotes Polζ-Rev1 association with DSBs . Mec1-mediated Polζ-Rev1 accumulation might assure efficient gap-filling during NHEJ in S or G2/M phase.
PCNA ubiquitination activates the Polη pathway as well as the Polζ/Rev1 pathway [16, 17]. However, Polη (Rad30) is dispensable for our DSB repair assay. These findings suggest that PCNA ubiquitination does not always activate the Polη pathway. Association of Polη with the HO-induced DSB has not been detected by ChIP assay (data not shown), whereas Polζ and Rev1 association is observed . Association of human Polη with damage sites is dependent on PCNA ubiquitination [21, 22]. As discussed above, ubiquitination of PCNA has not been detected after DSB induction. It is therefore possible that each DSB lesion contains only a few ubiquitinated PCNA molecules. In this case, Polη might fail to accumulate at DSBs in budding yeast. By contrast, Polζ-Rev1 accumulates at DSBs independently of PCNA ubiquitination. Therefore, Polζ-Rev1 could interact with ubiquitinated PCNA at DSBs.
Although there are only two TLS polymerases (Polζ and Polη) in budding yeast, human cells express more varied TLS polymerases including Polκ and Polι . These TLS polymerases may play specific roles at sites of DNA damage, because they possess different biochemical properties. However, it still remains unclear how these TLS polymerases act coordinately in DNA repair. Our results suggest that ubiquitination-independent accumulation helps Polζ-Rev1 to act differently from Polη in budding yeast.
We thank Carol Newlon and Sean Gregory for critical reading and discussion, and J.A. Nickoloff for sending materials. This work was supported by NIH grant GM073876.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.