|Home | About | Journals | Submit | Contact Us | Français|
In Saccharomyces cerevisiae, a DNA damage checkpoint in the S phase is responsible for delaying DNA replication in response to genotoxic stress. This pathway is partially regulated by the checkpoint proteins Rad9, Rad17 and Rad24. Here, we describe a novel hypermutable phenotype for rad9Δ, rad17Δ and rad24Δ cells in response to a chronic 0.01% dose of the DNA alkylating agent MMS. We report that this hypermutability results from DNA damage introduction during the S phase and is dependent on a functional translesion synthesis pathway. In addition, we performed a genetic screen for interactions with rad9Δ that confer sensitivity to 0.01% MMS. We report and quantify 25 genetic interactions with rad9Δ, many of which involve the post-replication repair machinery. From these data, we conclude that defects in S phase checkpoint regulation lead to increased reliance on mutagenic translesion synthesis, and we describe a novel role for members of the S-phase DNA damage checkpoint in suppressing mutagenic post-replicative repair in response to sublethal MMS treatment.
The DNA damage response (DDR) consists of a highly coordinated network of cellular processes tasked with maintaining genomic integrity despite continual damage from a wide variety of endogenous and exogenous agents. A critical step in this response is cell cycle arrest, in which a damage-induced signal triggers a checkpoint at G1, intra-S, or G2/M[1–3]. Notably, mutations in genes involved in DDR checkpoints are associated with predisposition to cancer in mammals (e.g. ATM, BRCA1, p53) .
Methylmethanesulfonate (MMS) is a monofunctional alkylating agent which generates methylated DNA lesions and triggers checkpoint activation; it is commonly referred to as “radiomimetic”. Multiple pathways coordinate to repair MMS lesions, which include direct reversal (dependent on the MGT1 alkyltransferase), base and nucleotide excision repair, post-replication repair, and homologous recombinational repair. In response to sublethal doses of DNA alkylating agents, budding yeast synchronize into a lengthened S-phase due to an intra-S phase checkpoint that is dependent on RAD53 and MEC1. While the budding yeast checkpoint adapter Rad9 is required for DNA damage-induced arrest in the G1 and G2/M phases, its role in intra-S is not absolute, since deletion of RAD9 is associated with partial loss of S phase slowing in response to MMS. Members of the RAD24 epistasis group (RAD24, RAD17, DDC1 and MEC3) exhibit a similar partial defect in S phase slowing[7–9]; however members of this group can enhance the MMS sensitivity of rad9Δ.
RAD9 has homology to the mammalian BRCA1 gene. Like BRCA1, Rad9 has BRCT and Tudor domains, which are important for protein-protein interactions mostly involved in DNA repair or cell cycle regulation. Rad9 serves as an adapter in the Mec1/Tel1-dependent checkpoint response to DNA damage. An early step in the cellular response to DNA damage is modification of histone tails near the site of damage (e.g. methylation, MEC1- or TEL1-dependent phosphorylation). Rad9p is subsequently recruited to the damaged site (through the association of its Tudor domains with phosphorylated histone H2A and methylated histone H3) and oligomerizes via its BRCT domains. Once recruited to the damaged site, Rad9p is also phosphorylated in a Mec1/Tel1-dependent manner, and its phosphorylated S/T-Q residues create a binding site for the FHA domain of the checkpoint effecter kinase Rad53[12–17]. Thus, oligomeric assembly of phosphorylated Rad9p is likely to serve as a platform for the enrichment of Rad53p and stimulation of its trans-autophosphorylation and phosphorylation by Mec1p and Tel1p. These phosphorylation events activate Rad53p and allow it to trigger downstream events in the DDR[18,19].
Members of the RAD24 epistasis group comprise a damage-specific DNA clamp known as the 9-1-1 complex, which is involved in DNA damage checkpoint regulation. The 9-1-1 clamp is composed of three subunits, Rad17, Ddc1 and Mec3. It is loaded on to the damage site by the alternative heteropentameric replication factor C (RFC) complex, in which one subunit, Rfc1, is replaced by the checkpoint-specific subunit Rad24. Mec1-dependent phoshorylation and activation of Rad9 and Rad53 is severely reduced in rad17, mec3, ddc1 and rad24 mutants. Putative functions of the 9-1-1 complex involve activation of Mec1 kinase activity and recruitment of other factors that could propagate the checkpoint response pathway or facilitate the processivity of the replication fork[21,22].
Both RAD9 and the RAD24 group encode for proteins that are required for efficient S-phase checkpoint regulation in response to alkylation damage, and the role of this checkpoint is believed to be to allow a damaged cell time to repair DNA lesions prior to the arrival of the replication fork. If lesions are left unrepaired, cells utilize one of three independent post-replication repair (PRR) mechanisms to bypass the lesion. In the first PRR mechanism, a switch to an error-prone translesion synthesis (TLS) polymerase occurs, which is triggered by a Rad6-Rad18 mediated mono-ubiquitination of PCNA. One of the TLS polymerases is the Polζ complex, composed of Rev3, Rev7, Rev1, and likely additional proteins. Polζ is able to replicate over a damaged template much more efficiently than major replicases, inserting a noncognate nucleotide. A second mechanism employs polyubiquitination of PCNA by the Mms2-Ubc13-Rad5 complex, which promotes error-free lesion bypass through a mechanism involving regression of the replication fork. A third mechanism depends on Rad52, which promotes homologous recombination (HR) between sister chromatids. Genetic interactions between RAD9 and PRR genes (e.g. MMS2, REV3) have been reported.
In this study, we describe a novel hypermutable phenotype for mutants lacking RAD9 or members of the RAD24 epistasis group. We show that the phenotype occurs exclusively when cells are treated with a chronic low-dose treatment of MMS, and not when a higher dose is applied. Importantly, we demonstrate that different doses of MMS yield different effects on the cell cycle distribution, a phenomenon which is responsible for the dose-dependent hypermutability of S-phase checkpoint mutants. We show that the hypermutable phenotype of rad9Δ cells is dependent on rev3Δ, indicating that the mutability of such cells is due to hyperactivation of the error-prone post-replication repair pathway. Consistent with (and extending) previous work linking RAD9 to the PRR pathway, we show that RAD9 interacts with a large number of PRR genes that function in both error-prone (REV1, REV3, REV7) and error-free (RAD5, MMS2, UBC13) pathways, and present a model in which RAD9 plays a role in channeling lesions at the replication fork.
YEPD and dropout media have been previously described. MMS was purchased from Sigma (Cat# M4016). YEPD and synthetic plates containing MMS were freshly prepared the evening prior to use. Magic medium (SC-Leu-His-Arg; 200mg/L G418, 60mg/L L-Canavanine) used in the synthetic interaction screen was prepared according to Pan et al.
S. cerevisiae strains used in this study are listed in Table 1. Strain BY4741 was purchased from Open Biosystems. Yeast strains with the designation yMP have been previously described. All gene disruptions were achieved by homologous recombination at their chromosomal loci by standard PCR-based methods. Briefly, a deletion cassette with a 0.5 kb region flanking the target ORF was amplified by PCR from the corresponding xxxΔKANr strain of the deletion array (Open Biosystems) and transformed into the target strain for gene knockout. The primers used in the gene disruptions are designed using 20 bp sequences which are 0.5 kb upstream and downstream of the target gene.
The rad9Δ double deletion library was constructed using the dSLAM methodology. The pooled heterozygous diploid deletion library was a gift from Jef Boeke (Johns Hopkins). A rad9Δ deletion cassette with a 1.5 kb region flanking the RAD9 ORF was amplified by PCR (forward primer: 5′-AGCTCTTGAACAACATACTCTCAG-3′; reverse primer: 5′-GAGATTCATCAAACAGATTGATCGC-3′) and transformed into the library. Selection of the rad9Δ::URA3 diploids was performed on synthetic defined medium plates without uracil (SD-URA). Diploids were subsequently sporulated via replication onto SPO plates and incubation at room temperature for 5 days. Spores were replicated onto Magic Medium (MM) –URA plates to select MATa rad9Δ::URA3 double mutant haploid cells. Haploid double-deletion cells were replicated onto complete synthetic medium with or without 0.01% MMS. Clones exhibiting sensitivity to MMS were streaked for single colonies on complete SD-medium. Eight colonies per candidate were subsequently grown overnight in synthetic liquid medium to saturation, and 2μl of saturated culture were spotted on complete SD-medium +/− 0.01% MMS and scored after 2–3 days. For each candidate, UPTAG and DOWNTAG barcodes were sequenced to identify the corresponding gene deletion using primers and methods previously described[33,34].
MMS kill curves were performed as previously described[3,7]. Briefly, log-phase cells (5 × 107 cells) were harvested from YPD medium and resuspended in 10 ml YPD with a specified concentration of pre-diluted liquid MMS solution. One MMS solution was used for all cultures in a single experiment to ensure identical MMS concentration across all cultures, and control strains (wildtype, xxxΔ, and rad9Δ) were always run on the same day as the double mutants to control for day-to-day variation in MMS preparations. Cultures were incubated at 30°C, and aliquots were taken out at given intervals. The cells were resuspended in PBS + 5% sodium thiosulfate (to inactivate the MMS). Cells were sonicated, and cell concentrations were assessed using a Coulter Counter. Viability was determined by plating serial dilutions of cultures onto YPD and scoring the number of colony-forming units (CFU) after 3–4 days at 30°C. Viability was calculated as CFU/total cells. Cell cycle distributions were determined by flow cytometry of propidium iodide (PI)-stained cells using a method described previously. Distributions of PI-stained cells were assessed using a Beckman Dickson Calibur flow cytometer.
Log-phase cells grown at 30°C in YEPD were harvested, sonicated, and counted using a Coulter Counter. 1 × 108 cells were resuspended in 2 ml PBS, sonicated, and serially diluted. Dilutions were spread onto fresh YEPD plates and exposed to gamma irradiation using a Mark II cesium-137 irradiator (JL Shepherd & Associates) operated at a dose rate of 800 cGy/minute. Following IR, plates were immediately transferred to an incubator, and allowed to grow for 3 days at 30°C. Viability was calculated as CFU/total cells. Control cells were always irradiated on the same day as mutant strains, and three independent isolates were tested for each mutant strain over a three day period.
MMS-induced mutation and sister chromatid exchange (SCE) rates were measured as previously described. Briefly, mutation rates were measured by selection for Canavanine resistance (due to forward mutation of the CAN1 gene). Mutation rates were determined in both the BY4741 and A364a backgrounds. SCE rates were measured in the A364a background, previously engineered to carry a SCR::URA3 sister chromatid recombination substrate [36,37]. SCE and mutation rates were measured simultaneously (i.e. side by side on the same days with the same cell cultures) for these studies, and controls were always examined concurrently on the same day alongside mutant strains. MMS treatment of cells was performed exactly the same as described in section 2.4. Following inactivation of the MMS by resuspension of cells in PBS + 5% sodium thiolsulfate, cells were serially diluted and plated onto SD-Arg-Ser + 60 mg/L canavanine medium (for measurement of mutation rates), SD-His medium (for measurement of SCE rates), and YPD medium (for viability). Plates were incubated at 30°C for 3 days, and numbers of mutants/recombinants were assessed by the number of CFUs on the respective selection plates. Mutation rates were expressed as canavanine-resistant cells per 106 viable cells. SCE rates were expressed as His+ cells per 106 viable cells. For both mutation and SCE, the rates after MMS induction were determined by subtracting the observed numbers of mutants or recombinants in the starting culture (i.e. pre-MMS exposure) from the number observed post-MMS exposure.
Synthetic enhancement genetics can be used to examine how mutations in two genes interact to modulate a phenotype and to uncover useful information about the functions of the interacting genes and their relationship. We performed a genetic screen to identify second site mutations that enhance the DNA damage sensitivity of the rad9Δ mutant to chronic sublethal (0.01%) MMS treatment. We utilized a screening protocol derived from the dSLAM procedure. Briefly, a rad9Δ::URA3 query construct was introduced to a haploid-convertible heterozygous diploid yeast knockout library pool by integrative transformation. Following sporulation, the haploid double mutants carrying both the rad9Δ allele and a second gene disruption were selected and subsequently screened for sensitivity on synthetic complete media plates containing 0.01% MMS. A total of 27,000 colonies were screened from the double deletion library. From this, 337 individual double deletion mutants were found to be sensitive to MMS, of which 202 unique double mutants were identified by sequencing of the flanking barcode regions.
Our initial screen was not exclusive for the enhancement phenotype we sought, since all single mutants conferring significant sensitivity to MMS also would be recovered (in addition to the desired rad9Δ–interacting genes). Thus we performed a quantitative counter screen comparing the sensitivity of each double mutant candidate (xxxΔ rad9Δ) to the original single mutants (xxxΔ) from the deletion library by assessing viability following a 5-hour exposure to 0.01% MMS in liquid rich medium. This counter screen identified a subset of 25 yeast gene disruptions that significantly enhance the sensitivity of the rad9Δ mutant, such that the xxxΔ rad9Δ double mutant was more sensitive to MMS than either the rad9Δ or the xxxΔ single mutants.
To reconfirm the enhanced sensitivity, we reconstructed individual gene deletions of the 25 genes in a wild type or rad9Δ background by mating each single deletion strain to a rad9Δ strain. Three independent segregants of each double or single mutant were subjected to a second round of MMS liquid kill curve testing. All 25 of the double mutants (xxxΔ rad9Δ) exhibited a 5-fold or greater enhanced sensitivity to MMS than either single mutant (xxxΔ or rad9Δ) (p<0.01) (Table 2 and Supplemental Figure S1). These genes comprised a number of different functional categories (Figure 1), and 15 out of the 25 interactions were previously unobserved (indicated by an asterisk in Figure 1). The severity of the interaction with rad9Δ varied significantly among these categories (Table 2). Notably, genes involved in PRR exhibited the highest degree of interaction with rad9Δ (Table 2 and Supplemental Figure S1). Genes involved in homologous recombination repair (HR) and resolution of HR intermediates as well as direct reversal of alkylation (MGT1) and other aspects of DNA repair (IXR1) also enhanced the sensitivity of rad9Δ.
Interestingly, we also identified a number of genes not previously known to function in the DDR, including ISW1 (chromatin remodeling), POT1 (fatty acid metabolism), BBC1 (localized to actin patches), MSN1 (transcription), and two uncharacterized genes, YIL158W and UIP5. In order to determine whether these interactions displayed general DNA damage sensitivity or MMS-specific sensitivity, we tested whether these genes enhanced rad9Δ sensitivity to ionizing radiation as well. Four of the candidates (BBC1, ISW1, YIL158W and POT1) displayed cross-sensitivity to ionizing radiation, suggesting that these genes are important for surviving DNA damage in rad9Δ cells (Supplemental Figure S2).
Cells have multiple repair options available for handling any single lesion; however the cellular mechanism for choosing which pathway to utilize is poorly understood. In light of results from our screen and recent data linking checkpoint genes to choice of PRR mechanism, we sought to explore the role of the S. cerevisiae checkpoint gene RAD9 in such a function. We hypothesized that if RAD9 contributed to regulation of mutagenic versus error-free PRR in the S phase, then the rad9Δ mutant might exhibit a hypermutable phenotype when treated with the DNA alkylating agent MMS. To test this prediction, we exposed cells to 0.01% MMS in liquid culture for 5 hours and assayed induction of mutations by measuring forward mutation to canavanine resistance. As shown in Figure 2A, the rad9Δ mutant shows significant elevation of MMS-induced mutation rate compared to wild type (p≤0.01).
At first glance, this result contradicts a previous report by Barbour et al. that the rad9Δ mutant is not hyper-mutable in the presence of MMS. However, we subsequently noted that the MMS exposures were very different between these two studies; in our study, cells were exposed for 5 hours to 0.01% MMS, whereas in the Barbour et al. study, cells were exposed to a higher concentration of MMS (0.05%) for half an hour. We hypothesized that this critical difference in exposure might explain the discordant results in the two studies. To test this, we measured MMS-induced mutation rates in the same strains under the two conditions (5 hours at 0.01% MMS vs 0.5 hours at 0.05% MMS; see Figure 2A). Consistent with the report of Barbour et al., we saw no hypermutable phenotype of rad9Δ at the 0.05% MMS dose, demonstrating the dose-dependence of the rad9Δ hypermutable phenotype (Figure 2A).
To test whether the rad9Δ MMS dose-dependent hypermutable phenotype was specific to the BY4741 strain background, we repeated the mutagenesis studies in the A364a strain background. As shown in Figure 2B, the MMS dose-dependent hypermutable phenotype is recapitulated in the A364a background; furthermore, the hypermutable phenotype could also be detected at a ten-fold lower MMS exposure (0.001%) (Figure 2C). We conclude that the hypermutable phenotype of rad9Δ in MMS is dose-dependent and is not unique to the BY4741 strain background.
To determine whether the hypermutable phenotype was dependent on the canonical REV3-dependent error-prone PRR pathway, we tested whether the recovery of can1 mutants in the rad9Δ background was REV3-dependent. We constructed a rad9Δ rev3Δ double mutant and repeated the mutagenesis experiment. As shown in Figure 2C, the hypermutable phenotype of rad9Δ cells is dependent on REV3, demonstrating that the hypermutable phenotype is due to increased activity of the error-prone REV3-dependent branch of the PRR pathway.
In addition to mutagenic damage tolerance mechanisms, PRR can also employ homologous recombination (HR), which can be tested by measuring sister chromatid exchange (SCE) rates. Thus, we asked whether rad9Δ has an effect on SCE induction in the presence of MMS. We observed that wild type and rad9Δ cells exhibited no significant difference in SCE induction in either the 0.01% (5 hours) or the 0.05% (0.5 hours) MMS conditions (Figure 2B). Thus, we conclude that while rad9Δ mutation affects MMS-inducible mutation rates, there is no effect on the rate of MMS-inducible SCE in these cells.
One possible explanation for the dose-dependence of the rad9Δ hypermutable phenotype is that in 0.05% MMS, lesion density is high enough to produce multiple lesions in a short track of DNA, which is more likely to degrade or be processed to a DSB than to induce mutagenic trans-lesion synthesis. However, our data do not support this model. For example, if more DSB were being produced in the 0.05% versus 0.01% MMS conditions, we would expect to see higher rates of sister chromatid recombination in the former. In contrast, we see a higher level of sister chromatid exchange induction in the 0.01% MMS conditions (Figure 2B). Also not consistent with there being more DSB in the 0.05% conditions, the 0.01% MMS condition introduces more lethal damage (DSBs are lethal in haploid yeast cells), evidenced by lower survival of rad9Δ, indicating that a concentration of 0.01% MMS at 5 hrs is a higher effective dose than 0.05% MMS at ½ hour. Thus it is unlikely that the possibility of fewer DSBs at the lower concentration explains the hypermutable phenotype.
Since it has been documented that DNA repair and DNA damage tolerance mechanisms differ throughout the cell cycle[36,40], a second possible explanation for the dose-dependence of the rad9Δ hypermutable phenotype is that cell cycle distributions differ significantly between the 0.01% and 0.05% MMS conditions used in these studies. (We previously demonstrated that 0.015 – 0.03% MMS exposure induces a regulated slowing of S phase progression, termed the intra-S phase checkpoint[3,7,40]; the 0.05% dose used in the Barbour et al. study was not previously tested for S phase effects.) To test this hypothesis, we treated wild type and rad9Δ cells with either 0.01% or 0.05% MMS for a period of 5 hours, withdrawing cells at multiple time points throughout the treatment for assessment of cell cycle distribution by flow cytometry. As seen in Figure 3A, wild type cells treated with 0.01% of MMS accumulate in the S phase over the course of 5 hours, as previously described[3,7]. Also as previously described, rad9Δ cells treated with the same dose show reduced accumulation in the S phase, proceeding through to the G2 phase faster than wild type. In dramatic contrast, there is no observable accumulation of rad9Δ cells in the S phase during a 30 minute pulse of an asynchronous culture with 0.05% MMS (Figure 3A), the conditions used in the Barbour et al. study. Based on these data, the majority of MMS-induced DNA damage in our experiments (0.01% MMS) is introduced during the S phase of the cell cycle, whereas the majority of damage was induced outside of the S phase in the 0.05% MMS condition used in the Barbour et al. study. Moreover, if we treat cells with 0.05% MMS past the 30 minute pulse, we see a synchronization represented by a strong G1 peak (Figure 3A). We confirmed that these cells were accumulating in the G1 phase through the observation that the majority of cells at the higher dose remain unbudded (Figure 3B). However, a proportion of budded cells remain, suggesting that though replication is suppressed, it may not entirely be due to accumulation in the G1 phase. It is possible that a small proportion of cells progress into the S phase upon treatment with 0.05% MMS, but the replication forks may only progress for very small distances in response to high doses (producing a “G1-like” S-phase peak). Nonetheless, these results are consistent with the hypothesis that the defect in the intra-S phase checkpoint in rad9Δ cells leads to a higher mutation rate in the 0.01% MMS treatment (where cells are replicating), but not in the 0.05% MMS treatment (where replication is suppressed). Importantly, it has been demonstrated that cells are most susceptible to mutagenesis in the S phase of the cell cycle.
If the dose-dependent rad9Δ hypermutable phenotype were due to a defect in the intra-S phase DNA damage checkpoint, then we predicted that other mutations (e.g. rad17Δ and rad24Δ) affecting this checkpoint might also exhibit dose-dependent hypermutability in MMS. To test this prediction, we measured mutation and SCE induction in rad17Δ and rad24Δ mutants in 0.01% MMS. As shown in Figure 3C, the rad17Δ and rad24Δ mutants phenocopy rad9Δ, displaying hypermutability in response to a 5 hour exposure to 0.01% MMS, but no effect on MMS-induced SCE. Like rad9Δ, both rad17Δ and rad24Δ mutants display reduced accumulation in the S-phase after chronic MMS treatment, a phenotype that is not evident following a pulse of 0.05% MMS (Supplemental Figure S3). The observation that additional intra-S phase checkpoint-defective mutants exhibit similarly enhanced MMS-inducible mutation rates is consistent with a model wherein the hypermutable phenotype is a result of inappropriate S-phase progression in the presence of MMS-induced damage.
As described in this study, MMS dose has a profound impact on cell cycle distributions. In the previous study by Barbour et al., a rapid 30 minute pulse with 0.05% MMS is not associated with accumulation of cells in S phase (Figure 3), and base excision repair is likely to remove alkylation damage, in most cases prior to entry into the S phase (after the MMS is withdrawn). As a result, there is little consequence to the genetic integrity of the cell. However, if residual damage remains once cells enter the S phase, or if damage is introduced during the S phase (as is the case in the 0.01% MMS condition used in our experiments), replication forks encounter the damage and stall, and cells are forced to employ damage tolerance mechanisms, some of which are mutagenic. This hypothesis is consistent with studies of Ostroff et al. and Kadyk et al. that demonstrated that cells are most susceptible to UV-induced mutagenesis and sister chromatid exchange during the S phase of the cell cycle[37,40].
There are three potential outcomes for a stalled replication fork (Figure 4). First, DNA repair proteins (e.g. base excision repair) may remove the offending lesion, allowing the fork to resume replication. Second, the lesion can be tolerated (i.e. circumnavigated, rather than being removed) either by template switching (dependent on MMS2, UBC13, RAD5) or by mutagenic translesion synthesis (dependent on REV1, REV3, REV7). Third, the stalled fork may collapse, and occasional fork collapses are repaired by homologous recombination (HR) . MEC1 is required for stabilizing stalled forks; hence in the mec1 mutant stalled forks collapse irreversibly at high rates, resulting in rapid death.
Our genome-wide screen revealed extensive interactions between RAD9 and post-replication repair genes required for tolerating unrepaired DNA damage during the S phase. We can infer from the heightened importance of post-replication repair in rad9Δ cells that replication forks are encountering lesions more frequently in the rad9Δ mutant than in the wild type. This could be due to: i) a general decrease in the efficiency of repair or reversal of alkylation damage in the rad9Δ mutant, and/or ii) abnormal coordination between DNA replication and alkylation damage repair or reversal in the rad9Δ mutant, resulting in an increase in the number of lesions’ being encountered by replication forks. There are data suggesting that either or both of these mechanisms could occur, as discussed below.
RAD9 has been implicated in nucleotide excision repair of UV-damaged DNA[43,44]. Recent studies have suggested that RAD9 is required for repair of the transcribed strands and the non-transcribed strands of active genes (but not for repair of transcriptionally inactive DNA sequences), possibly through the up-regulation of genes involved in the repair process. There are no studies reported to look for a role, either direct or indirect, of RAD9 in promoting base excision repair or direct reversal of alkylation damage, and this would be an interesting area of follow-up investigation.
We previously showed that in the continuous presence of MMS, the rate of S phase progression is dramatically slowed by an intra-S phase checkpoint in wild type cells, suggesting the possibility that there may be coordination between DNA replication and repair. It is interesting to note that while mec1 and rad53 mutants show severe defects in S phase regulation in the presence of MMS, rad9Δ mutation confers a far more subtle defect . The basis of these two distinct phenotypes is not understood, and it is equally plausible that MEC1 (or RAD53) and RAD9 are involved in distinct mechanisms controlling S phase progression or that they are involved in the same mechanism, with the MEC1 and RAD53 mutations showing higher penetrance. Of note, the mec1 rad9Δ and rad53Δ rad9Δ double mutants are more sensitive to MMS then any of the single mutants , indicating that the survival-promoting functions of MEC1 and RAD53 do not lie completely within the same pathway as that for RAD9.
Elegant work by Tercero and Diffley investigated the underlying mechanism of S phase slowing in the presence of MMS, and the effects of mutations in RAD53 and MEC1. They showed that exposure to MMS reduces the rate of DNA replication fork progression to about 300 base pairs per minute, 5 to 10 times lower than fork rates in the absence of MMS[46,47]. However, they found that the slow fork rate progression does not require RAD53 or MEC1, indicating that the accelerated S phase is primarily a consequence of inappropriate initiation events observed in these mutants. Furthermore, the cytotoxicity of MMS in checkpoint mutants occurs specifically when cells are allowed to enter S phase with damage, at which time replication forks in checkpoint mutants collapse irreversibly at high rates. Hence, preventing damage-induced replication fork catastrophe seems to be a primary mechanism by which the MEC1-dependent checkpoint preserves viability in the face of DNA alkylation. (Of note, these studies were all performed in the presence of 0.033% MMS.)
The mechanism underlying the RAD9-dependent slowing of S phase progression in response to MMS has not been investigated, nor have fork elongation rates been measured in the rad9Δ mutant in the presence of low dose MMS (i.e. 0.01% MMS during active S phase). Hence, while the Tercero and Diffley  data eliminate the possibility of MEC1-dependent control of replication fork elongation (at least in 0.033% MMS), their data do not eliminate the possibility that fork elongation rate is controlled in a MEC1-independent mechanism. Hence, although speculative, it remains formally possible that slowing of S phase in response to MMS is due both to control of origin firing (MEC1-dependent) and control of elongation, and we hypothesize that control of elongation may be RAD9-dependent and required for efficient removal of alkylation damage, via a mechanism potentially analogous to transcription-repair coupling.
Barbour et al. proposed that because rad9Δ is synergistic to both mms2 and rev3 with respect to killing by MMS, RAD9 likely functions as a separate branch of post-replication repair, independent of the REV3- and MMS2-associated branches and downstream of RAD18. Our observation that rad9Δ cells are hypermutable in 0.01% MMS is consistent with this model, in that loss of a parallel post-replication repair pathway could shuttle a higher percentage of DNA lesions into the mutagenic, REV3-dependent translesion synthesis pathway. There are alternative models.
One alternative model is that the RAD9 adapter acts at the replication fork at sites of damage to regulate how DNA damage is channeled through the various repair and tolerance pathways during the S phase so as to minimize genetic instability (Figure 4). Ill-described biochemical acrobatics must occur at the fork to elicit polymerase switching or template switching when DNA damage tolerance mechanisms are employed, and the mechanism of these switches is unclear. Under this alternative model, in the presence of MMS, RAD9 activity would strongly and actively promote use of non-mutagenic base excision repair (and/or alkylation reversal), while it would actively suppress mutagenic translesion synthesis. Hence, loss of RAD9 function would reduce the efficacy of base excision repair (Figure 4), thereby increasing the reliance of cells on template switching and translesion synthesis for survival; this would explain the synergy observed between the rad9Δ and both the mms2Δ and rev3Δ mutants, without evoking the need of a novel post-replication repair pathway. Additionally, loss of RAD9 function would result in derepression of translesion synthesis (Figure 4), leading to the hypermutable phenotype observed in our experiments. Hence, under this alternative model, RAD9 does not participate in a third branch of DNA damage tolerance, but rather it stabilizes the genome by maximizing the cell’s ability to employ non-mutagenic mechanisms (base excision repair and template switching) of repairing or tolerating lesions, while suppressing mutagenic translesion synthesis.
A second alternative model is that RAD9 acts to promote continuous DNA synthesis, potentially by limiting re-priming of stalled forks at MMS lesions. In this model, the observed hypermutability in MMS-treated rad9Δ cells may be due to an increased reliance on PRR to repair large ssDNA gaps resulting from discontinuous synthesis, which would explain the synergy between rad9Δ and both the MMS2 and REV3 branches of PRR.
Interactions between rad9Δ and HR repair genes are consistent with both models. In the first model, loss of RAD9 function would result in a higher frequency of fork stalling at unrepaired lesions, leading to an elevated probability of fork collapse and repair by HR genes[41, 50]. In the second model, a number of these HR genes (SGS1, MUS81, MMS4 and members of the RAD52 group) have been shown to promote gap repair and are epistatic to genes in the error-free branch of PRR[51, 52, 56]. Their synergy with rad9Δ may indicate an inability to repair large ssDNA gaps in rad9Δ cells. Recent work has shown that the choice among homologous recombination, translesion synthesis, and Rad52-related gap repair is dependent on a complex interplay between ubiquitination and sumoylation of PCNA. The role, if any, that RAD9 might play in mediating these signaling events is an intriguing avenue for further study.
BRCA1, one putative mammalian homolog of RAD9, also plays an important role in S-phase checkpoint regulation and genome stability . It is tempting to speculate that BRCA1 may also have synergistic interactions with homologs of PRR genes characterized in this study. If such synergistic interactions are evident in human cells, such genes may be modifiers of cancer penetrance for BRCA1 cases. Moreover, inhibition of PRR pathways may serve as an effective treatment mechanism for BRCA1 −/− tumors, comparable to the growing use of poly(ADP-ribose) polymerase (PARP) inhibitors, now in clinical trials for treating BRCA-deficient tumors .
We thank Jef Boeke for the diploid deletion library pool and Sean Wang for help with statistical analyses. We thank our anonymous reviewers for their insightful comments and suggestions. B.D.P. was supported by the FHCRC Dual Mentor Program and a U.S. Department of Defense Breast Cancer Research Program predoctoral fellowship. This work was supported by NIH grant R01 CA 129604.
Conflict of interest
The authors declare that there are no conflicts of interest.
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.