We studied the response of nucleotide excision repair (NER)‐defective rad14Δ cells to UV irradiation in G1 followed by release into the cell cycle. Only a subset of checkpoint proteins appears to mediate cell cycle arrest and regulate the timely activation of replication origins in the presence of unrepaired UV‐induced lesions. In fact, Mec1 and Rad53, but not Rad9 and the Rad24 group of checkpoint proteins, are required to delay cell cycle progression in rad14Δ cells after UV damage in G1. Consistently, Mec1‐dependent Rad53 phosphorylation after UV irradiation takes place in rad14Δ cells also in the absence of Rad9, Rad17, Rad24, Mec3 and Ddc1, and correlates with entry into S phase. Two‐dimensional gel analysis indicates that late replication origins are not fired in rad14Δ cells UV‐irradiated in G1 and released into the cell cycle, which instead initiate DNA replication from early origins and accumulate replication and recombination intermediates. Progression through S phase of UV‐treated NER‐deficient mec1 and rad53 mutants correlates with late origin firing, suggesting that unregulated DNA replication in the presence of irreparable UV‐induced lesions might result from a failure to prevent initiation at late origins.
Maintaining the integrity of genetic information is fundamental for the life of cells and the survival of species. DNA damage, whether it results from spontaneous hydrolytic events, radiation or chemical mutagens, must be recognized and repaired efficiently. Unrepaired DNA or inaccurate repair of DNA damage can lead to genomic instability, mutations and ultimately cancer (for a review see Hartwell and Kastan, 1994). Since DNA is constantly suffering damage, eukaryotic cells respond to DNA insults by restoring the correct DNA structure through several repair systems and by delaying cell cycle progression through surveillance mechanisms known as checkpoints. The balanced and concerted action of DNA repair and checkpoint mechanisms prevents accumulation of mutations and ensures genetic integrity (for reviews see Carr and Hoekstra, 1995; Paulovich et al., 1997b; Longhese et al., 1998; Weinert, 1998).
The cell cycle can be transiently arrested at different stages, depending on the phase at which DNA alterations occur. In budding yeast, at least three DNA damage checkpoints have been identified, which inhibit the G1–S transition (Siede et al., 1993, 1994), slow down progression through S phase (Paulovich and Hartwell, 1995) and delay entry into mitosis (Weinert and Hartwell, 1988, 1993; Weinert et al., 1994), when DNA is damaged during G1, S or G2 phases, respectively. Due to the vast array of DNA insults, multiple sensors may be required to recognize specific DNA perturbations at different cell cycle stages. Proteins encoded by the RAD9 gene and by the genes of the RAD24 epistasis group, including RAD17, RAD24, MEC3 and DDC1, are required specifically for proper DNA damage response, and are proposed to act at an early step of DNA damage recognition at any stage of the cell cycle (Weinert and Hartwell, 1988; Siede et al., 1993; Weinert et al., 1994; Longhese et al., 1996a, 1997; Paulovich et al., 1997a; Paciotti et al., 1998). The finding that Ddc1, Mec3 and Rad17 interact physically with each other provides the evidence that these putative sensor proteins, inferred from genetic studies to operate in one pathway, do indeed interact biochemically (Paciotti et al., 1998; Kondo et al., 1999). Conversely, the DNA replication proteins Polϵ, Dpb11 and Rfc5 appear to sense both replication blocks and DNA damage specifically during DNA synthesis (Araki et al., 1995; Navas et al., 1995, 1996; Sugimoto et al., 1996, 1997). Once DNA perturbations are sensed, the signalling process involves a protein phosphorylation cascade propagated through the two protein kinases Mec1 and Rad53 (Allen et al., 1994; Weinert et al., 1994; Paulovich and Hartwell, 1995). Mec1 is a member of a large lipid kinase family, which includes Schizosaccharomyces pombe Rad3 (Bentley et al., 1996; Cimprich et al., 1996; for review see Carr, 1997), as well as mammalian ATM (ataxia‐telangiectasia mutated; Savitsky et al., 1995), ATR (AT and rad‐related; Bentley et al., 1996) and the catalytic subunit of DNA‐dependent protein kinase (DNA‐PK; Jeggo et al., 1995). After DNA damage, the Mec1 protein kinase, whose activity might be regulated by sensor proteins, is required to phosphorylate and activate several substrates, including the protein kinase Rad53 (Zheng et al., 1993; Sanchez et al., 1996; Sun et al., 1996). Since both Rad9 and Ddc1 are phosphorylated (directly or indirectly) by Mec1, these proteins are suggested to modulate the recruitment of Mec1 into a catalytically active complex and its specificity towards downstream effectors (Emili, 1998; Paciotti et al., 1998; Sun et al., 1998; Vialard et al., 1998). Rad53 and Mec1 are necessary for phosphorylation of target proteins in response to checkpoint activation. For example, the Rad53‐dependent inhibition of CLN1‐2 transcription through phosphorylation of the transcriptional regulator Swi6 has been proposed to account for delaying G1–S transition after DNA damage in G1 (Sidorova and Breeden, 1997). Moreover, Mec1, but not Rad53, is required to phosphorylate the anaphase inhibitor Pds1, which is a component of the DNA damage checkpoint acting specifically in G2 (Yamamoto et al., 1996; Cohen‐Fix and Koshland, 1997). Finally, components of the DNA replication apparatus could be effectors of the DNA damage checkpoint operating specifically during DNA synthesis. Indeed, mutations affecting the large subunit of RP‐A and DNA primase impair the cellular response to DNA alterations during S phase (Longhese et al., 1996b; Marini et al., 1997). The reduced rate of ongoing DNA synthesis in the presence of DNA damage may reflect controls at the level of origin initiation and/or a decrease in the rate of elongation of nascent strands. Recent studies indicate that both treatment with hydroxyurea, which blocks the progression of DNA replication, and chronically damaging DNA molecules during S phase with methyl‐methane sulfonate (MMS) selectively inhibit the firing of late replication origins in a RAD53‐dependent manner (Santocanale and Diffley, 1998; Shirahige et al., 1998).
We have limited knowledge about the signals that elicit DNA damage checkpoints. Indeed, the DNA damage checkpoint activation probably requires the recognition of particular DNA structures or alterations in the genome at any stage of the cell cycle. For example, genotoxic agents cause many types of primary damage that can be converted to secondary lesions during DNA replication. In fact, a replication fork encountering a bulk in the template may stop, and resuming replication downstream of the damage might generate a daughter strand gap. Moreover, DNA replication across a single‐strand nick is likely to cause replication fork collapse and to produce double‐strand breaks (Kuzminov, 1995). Therefore, primary DNA lesions by themselves, secondary structures generated during DNA replication and/or the actual process of repair could be sensed as checkpoint triggering signals. In order to study the influence of DNA damage processing on checkpoint activation, we analysed the response of NER‐defective cells to DNA damage in G1 followed by release into the cell cycle. Saccharomyces cerevisiae cells lacking the RAD14 gene product, which encodes the structural and functional homologue of human XPA, are defective in NER. Irradiation of rad14Δ cells arrested in G1 by α‐factor treatment with low doses of UV results in the introduction of irreparable covalent changes in the genome and slows down cell cycle progression after release from α‐factor arrest (Siede et al., 1994). The cell cycle delay caused by UV damage in G1, which is RAD9 dependent in a NER‐proficient background, was shown to be RAD9 independent in NER‐defective cells (Siede et al., 1994). This suggests that NER‐processed UV photoproducts, but not unexcised dimers, might be able to activate the RAD9‐dependent checkpoint response. In this view, the inhibition of cell cycle progression observed after release from α‐factor block of NER mutants UV‐irradiated in G1 could be due to physical interference of unrepaired DNA damage with enzymes involved in DNA metabolism. Alternatively, checkpoint genes other than RAD9 might be able to recognize lesions and slow down cell cycle progression in the absence of DNA repair.
Here we provide evidence that a DNA damage checkpoint mechanism acts during DNA replication of irreparably UV‐damaged templates and we try to address some of the several unanswered questions about the molecular basis of this control. We show that RAD53 and MEC1 mediate DNA damage‐induced cell cycle arrest of rad14Δ mutants, whereas neither RAD9 nor the RAD24 group of genes are required for this arrest. The finding that Mec1‐dependent Rad53 phosphorylation in response to UV irradiation takes place in rad14Δ cells, in either the presence or the absence of Rad9, Rad17, Rad24, Mec3 and Ddc1, further supports the notion that a Mec1‐ and Rad53‐dependent checkpoint is active in the absence of a functional excision repair pathway. In addition, we show by two‐dimensional gel analysis that rad14Δ cells, after UV irradiation in G1 and release into the cell cycle, do not immediately arrest cell cycle progression but initiate DNA replication from early origins and, in the attempt to replicate damaged DNA templates, accumulate replication and recombination intermediates. Conversely, initiation of DNA replication from late origins is inhibited in a MEC1‐ and RAD53‐dependent manner, suggesting that the firing of late origins in mec1 and rad53 mutants might allow DNA replication in the presence of irreparable UV‐induced lesions.
RAD53 and MEC1, but not RAD9 and the RAD24 group of genes, are required for DNA damage‐induced cell cycle arrest in rad14Δ mutants
Saccharomyces cerevisiae cells missing the RAD14 gene product, which is essential for NER, after UV treatment in G1 and release into the cell cycle show slow cell cycle progression compared with wild type (Siede et al., 1994). It has been shown previously that in contrast to excision repair‐proficient cells, the UV‐induced delay of cell cycle progression in rad14Δ mutants takes place also in the absence of Rad9 (Siede et al., 1994). This cell cycle delay might be due to physical interference of unrepaired photoproducts with replication or, alternatively, might be controlled by checkpoint genes other than RAD9. To determine whether the slowing of cell cycle progression in rad14Δ mutants was an actively regulated process, we introduced mutant or null alleles of several checkpoint genes in the rad14Δ background and analysed cell cycle progression after UV irradiation in G1 followed by release into the cell cycle. ρ° strains were derived from each double mutants to avoid mitochondrial DNA contribution to cell DNA content measured by flow cytometry. These strains were synchronized in G1 by α‐factor treatment, UV‐irradiated with 5 J/m2 and released from the pheromone block. This low UV dose was sufficient to induce an apparent G1 or early S phase arrest in the rad14Δ mutant as determined by FACS analysis (Figure 1), whereas it was not able to delay bud emergence in this strain (data not shown). Under the same conditions, cell cycle progression of the wild‐type strain was not significantly affected compared with the unirradiated control. As shown in Figure 1, both the rad14Δ rad9Δ and rad14Δ ddc1Δ double mutants showed a delay of cell cycle progression after UV irradiation in G1 and release into the cell cycle which was indistinguishable from that observed for the rad14Δ single mutant. Similar results were obtained when the same analysis was performed for rad14Δ rad17Δ, rad14Δ rad24Δ and rad14Δ mec3Δ double mutants (data not shown). Therefore, not only RAD9, but also the genes of the RAD24 epistasis group were dispensable for UV‐induced cell cycle arrest in the rad14Δ mutant. Since RAD9 and the RAD24 group of genes are believed to act in different branches of the DNA damage checkpoint response (Lydall and Weinert, 1995; de la Torre‐Ruiz et al., 1998; Paciotti et al., 1998), we further analysed whether the rad14Δ rad9Δ ddc1Δ triple mutant was still able to delay cell cycle progression in response to UV irradiation in G1 (Figure 1). The behaviour of the rad14Δ rad9Δ ddc1Δ triple mutant was the same as that of the rad14Δ single mutant, thus excluding the possibility that the RAD9 gene could compensate for the lack of the RAD24 group of genes, and vice versa, in the absence of NER. In contrast, as shown in Figure 1, the rad14Δ mec1‐1 and rad14Δ rad53 double mutants entered S phase between 30 and 60 min after UV irradiation and replicated most if not all of their DNA during the subsequent 60–90 min. Therefore, UV‐induced cell cycle arrest in the absence of NER is controlled by a mechanism depending on the MEC1 and RAD53 genes, whereas neither RAD9 nor the RAD24 group of genes appear to have a role in slowing down cell cycle progression in the presence of irreparable lesions. These data indicate that Mec1, which is thought to act upstream of Rad53, might become activated even in the absence of Rad9 and the Rad24 group of proteins. In this view, in NER‐deficient strains, either the high concentration of unrepaired UV photoproducts or the secondary lesions arising during replication of damaged templates might spontaneously generate checkpoint signals that in turn directly or indirectly activate Mec1.
Rad53 is phosphorylated in rad14Δ mutants in response to UV damage in G1 followed by release into the cell cycle
Rad53 is phosphorylated in response to DNA damage and its phosphorylation correlates with the activation of DNA damage checkpoints (Sanchez et al., 1996; Sun et al., 1996). Since the UV‐induced cell cycle arrest in rad14Δ mutants is dependent on RAD53 and MEC1, we would expect UV damage in G1 to cause Rad53 phosphorylation in both wild‐type and rad14Δ cells. To explore this possibility, exponentially growing wild‐type and rad14Δ cells were synchronized in G1 by α‐factor treatment, UV‐irradiated and released from the pheromone arrest. As shown in Figure 2A and B (top), low doses of UV irradiation caused rad14Δ cells to arrest with an approximately 1C DNA content, but did not consistently delay bud emergence compared with wild‐type cells under the same conditions. Phosphorylated Rad53 was detectable immediately after release from α‐factor in both wild‐type and rad14Δ UV‐irradiated cell cultures (Figure 2B, bottom), further supporting the notion that a Rad53‐dependent checkpoint pathway is operating in excision repair‐deficient cells. The persistence of Rad53 phosphorylated forms reflected differences in DNA damage‐induced cell cycle delay in wild‐type and rad14Δ cells. In fact, 135 min after UV irradiation and release from the α‐factor block, the amount of Rad53 phosphorylated forms in wild‐type cells progressively decreased concomitantly with nuclear division (Figure 2B). Conversely, in rad14Δ cells that arrested cell cycle progression after UV treatment in G1, the phosphorylated forms of Rad53 persisted until the end of the experiment (210 min; Figure 2B).
Irreparable DNA lesions promote Rad53 phosphorylation in different checkpoint mutants
Rad53 phosphorylation has been shown to be dependent on Mec1, Rad9, Rad17, Rad24, Mec3 and Ddc1 (Zheng et al., 1993; Navas et al., 1996; Sanchez et al., 1996; Sun et al., 1996; Paciotti et al., 1998; Figure 3). If checkpoint signals could activate Mec1 in NER‐deficient mutants even in the absence of RAD9 and the RAD24 group of genes, we would expect deletion of RAD14 to be able to promote UV‐induced phosphorylation of Rad53 in cells lacking any of the above checkpoint proteins but Mec1. We therefore analysed Rad53 phosphorylation in protein extracts from different rad14Δ checkpoint mutants after synchronization in G1 by α‐factor, UV irradiation and release from the pheromone arrest. As shown in Figure 3A, deletion of RAD14 gene promotes phosphorylation of Rad53 in strains lacking the above listed proteins. In fact, Rad53 became phosphorylated after UV damage in G1 in rad9Δ rad14Δ, ddc1Δ rad14Δ, mec3Δ rad14Δ, rad24Δ rad14Δ, rad17Δ rad14Δ and ddc1Δ rad9Δ rad14Δ cells. As expected, Rad53 phosphorylation was instead dramatically reduced in rad9Δ and, to an even greater extent, in ddc1Δ, rad17Δ, mec3Δ and rad24Δ single mutants, and it was completely eliminated in the rad9Δ ddc1Δ double mutant after UV irradiation in G1 (Figure 3A). Since Mec1 is necessary together with Rad53 to arrest cell cycle progression in rad14Δ cells, the UV‐induced Rad53 phosphorylation in rad14Δ cells was expected to depend on Mec1. Indeed, we failed to detect any DNA damage‐induced phosphorylation of Rad53 in rad14Δ mec1‐1 double mutants as well as in rad14Δ rad9Δ ddc1Δ mec1‐1 quadruple mutants after UV irradiation in G1 (Figure 3B). Since Rad53 phosphorylation correlates with checkpoint activation, the above data are consistent with the activation of a RAD53‐ and MEC1‐dependent DNA damage checkpoint response in NER‐deficient background despite the lack of RAD9 and the RAD24 group of genes.
Early origin firing is not inhibited in rad14Δ mutants released into the cell cycle after UV damage in G1
Since the delay of cell cycle progression in the presence of irreparable covalent changes in the DNA molecules is mediated through the activation of a MEC1‐ and RAD53‐dependent checkpoint response, the accumulation of unrepaired pyrimidine dimers and/or secondary DNA lesions generated during the attempt to replicate a damaged template might activate the checkpoint pathway. To explore these possibilities further, we decided to determine whether rad14Δ mutant cells, incapable of dimer excision, immediately arrested the cell cycle or were able to initiate replication of a damaged template after UV irradiation in G1 followed by release into the cell cycle. Since FACS analysis was not sensitive enough to distinguish whether cell cycle progression of UV‐treated rad14Δ was blocked in G1 or in early S phase, we analysed the firing of the early‐replicating ARS305 origin by 2‐D gel electrophoresis (Friedman and Brewer, 1995). To this end, rad14Δ cells were synchronized in G1 phase by α‐factor treatment and one half of the culture was released from the pheromone block, whereas the other half was irradiated prior to release. After release into the cell cycle, samples from unirradiated or UV‐irradiated cells were collected at different times and pooled together (see legend to Figure 4 and Materials and methods). Digested chromosomal DNA was separated by 2‐D gel electrophoresis and hybridized to an ARS305 probe (see Materials and methods). As shown in Figure 4, both UV‐irradiated and unirradiated rad14Δ cells showed hybridization signals for the bubble arc, indicating that DNA replication started from the ARS305 origin independently of DNA damage. Therefore, the presence of irreparable UV photoproducts in G1 does not prevent the initiation of DNA replication from an early origin of replication. The signals for the Y‐arc, normally generated when the analysed origin is also passively replicated by forks initiated from origins outside, was reproducibly increased in the irradiated rad14Δ cells compared with the unirradiated culture (Figure 4). This accumulation of simple Y‐DNA intermediates might arise from the blockage of replication fork progression at the unexcised UV photoproducts. Moreover, a strong spike signal was detected specifically in the rad14Δ irradiated culture, indicating that fully replicated but connected X‐DNA molecules were being accumulated during the attempt to replicate UV‐damaged DNA template (Figure 4). These UV‐induced X‐shaped molecules might represent recombination intermediates. In fact, it is thought that newly synthesized DNA following UV irradiation in NER‐defective cells contains gaps opposite dimers, which might be cured by sister chromatid recombination (Lehmann, 1972; Wolff et al., 1974, 1977; Clark and Hanawalt, 1984; Berger and Edenberg, 1986; Kadyk and Hartwell, 1992, 1993; Paulovich et al., 1998). To investigate this possibility further, we analysed whether the RAD52 gene, which encodes a central recombination protein (for review see Petes et al., 1991), was required to generate the X‐shaped molecules in rad14Δ cells. To this end, rad14Δ rad52Δ double mutants were synchronized in G1 with α‐factor, UV‐irradiated before release from the pheromone block, and 2‐D gel analysis was performed as described above. As shown in Figure 4, hybridization signals for both bubble and Y‐arcs were detectable with the ARS305 probe in the absence of Rad52, whereas the spike representing X‐shaped DNA molecules was absent. Therefore, the DNA damage‐induced X‐shaped molecules accumulated in rad14Δ cells are likely recombination intermediates whose formation is dependent on Rad52.
Mec1 and Rad53, but neither Rad9 nor Ddc1, mediate inhibition of late origin firing in rad14Δ mutants after UV irradiation in G1
FACS analysis data showed that the absence of Rad53 and Mec1 allows rad14Δ cells to replicate DNA in the presence of UV‐induced lesions, suggesting that these checkpoint proteins might regulate the initiation and/or the elongation step of DNA replication of UV‐damaged templates. As shown in Figure 5, 2‐D gel analysis did not show significant differences for the early origin ARS305 between the rad14Δ single, the rad14Δ rad53, rad14Δ mec1‐1 double and the rad14Δ ddc1Δ rad9Δ triple mutants after UV irradiation in G1 followed by release into the cell cycle, thus indicating that none of the checkpoint mutations analysed affects initiation from early origins. Since Rad53 regulates the timing of replication initiation during normal cell growth and blocks initiation of replication from late origins in MMS‐treated cells (Shirahige et al., 1998), we investigated whether this was the case also in the presence of irreparable UV‐induced dimers in the genome by analysing the activation of the efficient late‐firing origin ARS501. We found that the presence of UV‐induced damage selectively blocked initiation of DNA replication from this late origin in rad14Δ cells. In fact, as shown in Figure 5, 2‐D gel analysis of rad14Δ DNA after UV irradiation in G1 did not show any bubble arc with the ARS501 probe, whereas this was detectable in the unirradiated rad14Δ cells. A similar behaviour was observed also for irradiated ddc1Δ rad9Δ rad14Δ triple mutants, indicating that the DDC1 and RAD9 genes are not required to inhibit late origin firing in the presence of UV‐induced lesions (Figure 5). In contrast, 2‐D gel analysis of irradiated rad14Δ rad53 and rad14Δ mec1‐1 double mutants clearly showed the bubble arc ARS501 hybridization signal, similar to that found for this late origin in the unirradiated cultures and for the early ARS305 origin in both unirradiated and irradiated cells (Figure 5). Therefore, both Mec1 and Rad53 are necessary to block initiation of DNA replication from late origins in the presence of UV‐induced lesions. Moreover, an increase of X‐ and Y‐shaped DNA molecules was detected with the ARS501 probe in irradiated rad14Δ rad53 and rad14Δ mec1‐1 double mutants compared with the irradiated rad14Δ cells (Figure 5). The accumulation of such DNA molecules, similar to that detected for the early ARS305 origin, might be induced by replication forks starting from the late ARS501 origin in the absence of MEC1 and RAD53. In this view, activation of DNA replication from the late ARS501 origin or from origins elsewhere in the genome, which may be fired in rad53 and mec1 mutants, might increase the substrates for recombination events and the number of replication forks that become blocked at the UV photoproducts.
Rad53 phosphorylation of rad14Δ cells UV‐irradiated in G1 requires entry into S phase
The observation that rad14Δ cells UV‐irradiated in G1 initiate DNA replication from early origins when released into the cell cycle suggested that unprocessed DNA damage in NER‐defective cells might require DNA replication to be sensed and to activate the Mec1‐ and Rad53‐dependent checkpoint response. If this were the case, we would expect to abolish Rad53 modification by preventing rad14Δ cells from entering S phase after UV treatment in G1. We therefore analysed Rad53 phosphorylation after UV irradiation in G1 and release from α‐factor at restrictive temperature in strains carrying a temperature‐sensitive allele of the CDC4 gene, which is required for initiation of S phase by promoting proteolysis of the specific inhibitor of cyclin B‐dependent kinases, p40SIC1 (Schwob et al., 1994). As shown in Figure 6, DNA damage‐induced phosphorylation of Rad53 in rad14Δ cells after UV treatment in G1 depended on entry into S phase. In fact, it was abolished in cdc4 rad14Δ double mutant cells released from α‐factor at 37°C, and it was restored by further deletion of the SIC1 gene (Figure 6), which allows cdc4 cells to enter S phase at restrictive temperature (Schwob et al., 1994). Conversely, Rad53 was readily phosphorylated in NER‐proficient wild‐type and cdc4 mutants UV‐irradiated in G1 and released from α‐factor at 37°C.
Genotoxic insults induce many different forms of DNA damage that may undergo structural transformations and metabolic processing during the cell cycle, thus generating secondary lesions that further challenge checkpoint and DNA repair pathways. It is still unknown whether DNA damage by itself is sufficient to activate a checkpoint response or whether DNA damage must be processed in order to induce a checkpoint‐mediated cell cycle arrest. DNA intermediates and/or DNA–protein complexes, which can form during different stages of DNA repair, recombination or replication, are good candidates for structures monitored by checkpoints. For example, single‐stranded DNA regions could be sensed as checkpoint signals (Garvik et al., 1995) and checkpoint components may be directly involved in processing damaged DNA to its single‐stranded form (Lydall and Weinert, 1995).
Some evidence suggests that DNA damage per se might not be sufficient for checkpoint activation and that the primary damage must be processed to yield DNA intermediates. For example, in Escherichia coli, the presence of UV‐induced DNA lesions is not sufficient to cause SOS induction. Rather, the SOS‐inducing signal is generated when cells attempt to replicate the damaged DNA, leading to the accumulation of single‐stranded regions (Sassanfar and Roberts, 1990). Moreover, if NER‐deficient human XPA− cells are UV irradiated, an increase in p53 protein level is not detectable under conditions in which cells cannot replicate DNA, whereas it is induced if DNA replication is allowed, suggesting that the primary damage must be modified by DNA replication in order to activate the checkpoint response (Nelson and Kastan, 1994).
Genes from the different DNA repair pathways do not appear to be required to slow down DNA synthesis in the presence of DNA lesions (Paulovich et al., 1997a). Conversely, mutations in NER genes cause further slowing of cell cycle progression after UV irradiation in G1 compared with excision repair‐proficient cells (Siede et al., 1994). A fraction of excision repair‐defective mutants can survive UV irradiation due to highly efficient mechanisms that enable them to tolerate large amounts of unrepaired damage during DNA replication. This accommodation of lesions might account for the slowing of S phase and involve mutagenic and/or recombinogenic pathways that may be regulated by checkpoint proteins.
A Mec1‐ and Rad53‐dependent checkpoint arrests cell cycle progression of UV‐damaged rad14Δ cells in early S phase
We found that early origins are fired in rad14Δ cells after UV irradiation in G1 followed by release into the cell cycle, suggesting that the presence of unexcised UV photoproducts does not prevent cells from entering S phase. In fact, we observe replication bubbles in 2‐D gels of rad14Δ DNA which was prepared from cells released into the cell cycle after UV irradiation in G1, and probed for the early ARS305 replication origin. Altogether, our data indicate that UV‐induced DNA damage in G1 in NER‐defective cells causes an early S phase arrest. We have investigated whether this arrest was due to genetically controlled mechanisms, and our analysis shows that Rad53 and Mec1 are necessary to slow down DNA synthesis in rad14Δ mutants after UV irradiation in G1. The activation of a Mec1‐ and Rad53‐dependent checkpoint in the presence of irreparable lesions is further confirmed by the finding that Rad53 is phosphorylated in rad14Δ cells released into the cell cycle after UV irradiation in G1 and that Mec1 is absolutely required for this phosphorylation. Therefore, in the absence of NER, a DNA damage checkpoint is acting specifically during DNA synthesis and is responsible for the S phase arrest. In this view, cells lacking the possibility of processing UV photoproducts might not be able to activate the checkpoint pathway until they enter S phase, suggesting that unprocessed DNA damage might require DNA replication to be sensed and to activate the MEC1‐ and RAD53‐dependent checkpoint response. This hypothesis is supported by the finding that Rad53 phosphorylation in rad14Δ cells UV‐irradiated in G1 does not take place when entry into S phase is prevented by Cdc4 inactivation. Neither RAD9 nor the RAD24 group of genes, which are needed to slow down S phase progression in excision repair‐proficient cells, seem to have any role in regulating the rate of ongoing S phase in cells that are unable to remove UV‐induced DNA damage. The persistence of damage‐induced Rad53 phosphorylated forms in rad14Δ mutants also carrying the rad9Δ, rad24Δ, rad17Δ, mec3Δ or ddc1Δ mutation is consistent with the hypothesis that the lack of Rad9 and the Rad24 group of proteins does not impair the activation of a Mec1‐ and Rad53‐dependent checkpoint in the absence of NER. Therefore, Rad9 and the Rad24 group of proteins appear to be dispensable for checkpoint response when chromosomes containing irreparable UV‐induced dimers experience DNA replication. Since the checkpoint‐mediated S phase arrest in mutants incapable of dimer excision requires MEC1 and RAD53, we propose that DNA replication of a damaged template might spontaneously generate some checkpoint signals able to activate Mec1, either directly or through other DNA damage sensors, therefore allowing Rad53 phosphorylation. Since replication proteins like DNA polymerase ϵ, Rfc5, Dpb11 and Rpa1 have been implicated in DNA damage checkpoint response specifically acting during S phase (Araki et al., 1995; Navas et al., 1995, 1996; Longhese et al., 1996b; Sugimoto et al., 1997), it is possible that once DNA replication is initiated, the DNA damage signalling system works through these replication factors in sensing and transducing signals to Mec1.
The consequences of unregulated progression into S phase in the presence of DNA damage are largely unknown. For example, DNA strand breaks, single‐stranded DNA regions and/or DNA recombination intermediates might result from replicating UV‐damaged DNA and might be recognized as checkpoint signals. Indeed, after UV irradiation of rad14Δ cells, we observe an enormous accumulation of Y‐ and X‐shaped molecules at the ARS305 origin‐containing fragment. Y‐DNA intermediates could form if replication forks initiated at the early ARS305 origin or at origins elsewhere in the chromosome are blocked at the damaged sites. Inhibition of DNA replication by DNA damage had previously been observed in E.coli as well as in mammalian cells (Larner et al., 1994). When E.coli cells are exposed to DNA‐damaging agents, DNA synthesis is transiently inhibited (Livneh et al., 1993) and this inhibition is mostly due to effects of the lesions on the progress of the replication fork (Rupp and Howard‐Flanders, 1968). Moreover, the capacity of lesions such as cyclobutane pyrimidine dimers to inhibit the progression of DNA polymerases is well established in vitro (for review see Friedberg et al., 1995). On the other hand, the UV‐induced accumulation of X‐shaped DNA molecules at the ARS305 origin‐containing fragment, which we detected as a spike signal on a 2‐D genomic DNA blot, could represent Holliday junctions (Zou and Rothstein, 1997). For example, recombination intermediates could be generated when stalling of the replication fork is caused by DNA polymerase encountering a lesion and DNA synthesis is resumed downstream of the damage. This results in the formation of a daughter strand gap that could be repaired by strand exchange with the undamaged sister molecule, thus generating a Holliday junction (Friedberg et al., 1995). The induction of recombination in cells incapable of excising UV‐induced pyrimidine dimers might increase the ability to circumvent DNA synthesis blockage, and therefore complete DNA replication in the presence of irreparable lesions. Several data in different organisms indicate that recombination mechanisms might enable cells to replicate their DNA after UV irradiation. Escherichia coli mutants, incapable of excising UV‐induced pyrimidine dimers, revealed a recombination mechanism involving sister strand exchange (Rupp and Howard‐Flanders, 1968; Rupp et al., 1971; Ganesan, 1974; Smith and Sharma, 1987). Moreover, in NER‐defective mammalian cells, newly synthesized DNA following UV irradiation is thought to contain gaps opposite dimers, which are then resolved by sister chromatid recombination (Lehmann, 1972; Wolff et al., 1974, 1977; Clark and Hanawalt, 1984; Berger and Edenberg, 1986). Finally, when NER yeast mutants encounter DNA damage, they experience the induction of replication‐dependent sister chromatid exchange, which requires Rad52 (Kadyk and Hartwell, 1993; Paulovich et al., 1998). The finding that the formation of these molecules is dependent upon the function of the recombination protein Rad52 in UV‐irradiated rad14Δ cells provides evidence that the observed X‐DNA forms are recombination intermediates, thus suggesting that UV‐induced lesions are channelled into the recombination repair pathway during DNA replication of a damaged template.
Role of Mec1 and Rad53 in regulating DNA synthesis in the presence of irreparable lesions
The replication of damaged DNA represents a major mechanism of genetic instability since the rate of DNA synthesis might determine the frequency by which DNA lesions are converted to permanent changes (for a review see Naegeli, 1994). The checkpoint control decreases the rate of ongoing DNA replication, avoiding the consequences of replicating a damaged template (Paulovich and Hartwell, 1995). Since cells entering S phase in the presence of DNA damage show an increased dependence on checkpoint function for survival, the slowing down of DNA replication probably allows cells to survive DNA damage better either by removing the damage more efficiently or by tolerating it better than would be possible during an unrestrained S phase. The regulated decrease in the rate of DNA replication on a damaged template could be due to a control of initiation and/or elongation of DNA replication, and/or to induction of alternative modes of replication slower than normal DNA synthesis and possibly dependent on recombinogenic mechanisms. One question is how the lack of MEC1 and RAD53 genes allows rad14Δ cells to succeed in replicating their chromosomes in the presence of irreparable damage. Although DNA lesions are thought to induce inhibition of replicon initiation (Larner et al., 1994) and chain elongation (Sarasin and Hanawalt, 1980), several data suggest that cells have the ability to replicate damaged chromosomes. Different mechanisms that allow DNA synthesis on templates containing blocking lesions have been hypothesized (for reviews see Echols and Goodman, 1991; Naegeli, 1994). One possibility is by continuing DNA replication across the template lesion (so‐called trans‐lesion synthesis). This process involves error‐prone DNA replication often resulting in mutagenesis (Rajagopalan et al., 1992). Since in mec1 rad14Δ and rad53 rad14Δ double mutants, UV‐induced mutation rates are reduced compared with those observed in rad14Δ single mutants under the same conditions (our unpublished observation), trans‐lesion DNA synthesis is not likely to be responsible for damage‐resistant DNA replication in the absence of MEC1 and RAD53.
Cells might also reinitiate DNA replication downstream of the block, thus producing a daughter‐strand gap opposite the site of damage, which subsequently might become filled in by recombinational repair. If the induction of replication‐dependent recombination facilitates the bypass of a bulge in the template and could account for the slow DNA synthesis in the presence of DNA damage, Rad53 and Mec1 might regulate the formation of recombination intermediates. However, we observe accumulation of X‐shaped DNA molecules in both the rad14Δ single mutant and the rad14Δ mec1 and rad14Δ rad53 double mutants, suggesting that Rad53 and Mec1 are not required for their formation. Nevertheless, we cannot exclude the possibility that Mec1 and Rad53 might regulate the stability, processing and/or resolution of the DNA recombination intermediates that arise during replication of a damaged template. If this were the case, the regulation of replication‐dependent recombination by the checkpoint pathway might allow cells to accommodate unrepaired DNA damage and may partially account for the checkpoint‐dependent slowing of DNA replication in the presence of DNA damage, as suggested previously by Paulovich et al. (1998).
Finally, uncontrolled firing of replication origins might account for the damage‐resistant DNA replication in the absence of MEC1 and RAD53. Our finding that unrepaired UV damage in G1 selectively blocks initiation of DNA replication from the late ARS501 origin in a MEC1‐ and RAD53‐dependent manner suggests that the unregulated initiation of DNA synthesis occurring in mec1 and rad53 mutants at late origins may be one of the mechanisms allowing DNA replication in the presence of irreparable UV photoproducts. Consistently, after UV irradiation in G1 and release into the cell cycle, the ddc1Δ rad9Δ rad14Δ triple mutant neither fires late origins nor progresses through S phase, further strengthening the correlation between unregulated activation of replication origins and progression through S phase. Therefore, when unrepaired UV‐induced dimers block progression of replication forks started at early origins, DNA replication may be further supported by initiation at late origins on the same chromosome. Since deletion of the B‐type cyclin gene CLB5 has been shown to reduce the efficiency of replication initiation at late chromosomal origins (Donaldson et al., 1998) and to cause a lengthened S phase (Epstein and Cross, 1992; Schwob and Nasmyth, 1993), we speculate that Clb5/Cdk1 may be one of the targets of the DNA damage checkpoint acting during DNA synthesis.
We cannot exclude the possibility that the DNA damage checkpoint pathway may control not only the initiation but also the elongation step of DNA synthesis, and the control of these two processes might have different requirements for checkpoint proteins. In contrast to what was observed in excision repair‐deficient cells, both ddc1Δ and rad9Δ NER‐proficient cells are partially defective in slowing down DNA synthesis in the presence of DNA damage, suggesting that Ddc1 and Rad9 might regulate only a subset of functions of the DNA damage checkpoint acting during replication. In this view, we propose that the checkpoint proteins Mec1, Rad53, Ddc1 and Rad9 are all involved in controlling the rate of fork progression, stopping the advancing replication forks until lesions are repaired, but only Mec1 and Rad53 might regulate replication initiation in the presence of DNA damage. In ddc1Δ and rad9Δ checkpoint mutants, unrestrained fork movement in excision repair‐proficient cells may be sufficient to promote DNA replication on damaged DNA, while in NER‐deficient strains the presence of large amounts of irreparable bulge in the template might obstruct ongoing replication forks. In this case, only rad53 and mec1 mutants, which are allowed to fire origins that in wild‐type cells are inhibited by DNA damage, may succeed in replicating DNA in the absence of NER.
The inability of rad9Δ and ddc1Δ mutants to promote DNA replication on irreparable UV‐damaged DNA might also be explained by the fact that Rad9 and Ddc1 do not have any role in sensing unrepaired UV dimers, which instead might be recognized, directly or indirectly, by Mec1 and Rad53. For example, Rad9 and the Rad24 group of proteins might detect DNA or protein–DNA structures generated during the repair process. Since in NER‐proficient strains these intermediates are likely to be processed quickly during repair, Rad9 and the Rad24 group of proteins could be required to sustain and/or amplify these transient signals, thus facilitating their recognition by Mec1.
Altogether, our data strongly support the notion that Mec1 and Rad53 play a primary role in the S phase damage‐sensing pathway. Given its structural similarity to DNA‐PK, Mec1 might directly recognize specific DNA or protein–DNA structures generated during DNA synthesis on damaged templates and/or the replication apparatus might be engaged in promoting Mec1 activation. It will be interesting to determine whether replication proteins such as DNA polymerase ϵ, Dpb11, Rfc5 and Rpa1 are required for the Mec1‐dependent inhibition of late origin firing, thus implying complex relationships between checkpoint pathways and replication machinery.
Materials and methods
Strains and media
Strains used in this study are listed in Table I. All yeast strains were derivatives of or were backcrossed at least three times to W303 (MATa or MATα, ade2‐1, trp1‐1, leu2‐3,112, his3‐11,15, ura3). To construct a RAD14 chromosomal deletion, the rad14Δ::kanMX4 cassette was constructed by PCR using plasmid pFA6a‐kanMX4 (Wach et al., 1994) as a template and oligonucleotides PRP11 (3′‐ttg cca cca tgc aga act tga atg gtg gtt ata tca acc cta agg aca agc gta cgc tgc agg tcg ac‐5′) and PRP12 (3′‐ttc ttc agt tct tag ccc gca gtc tgt aca tct tct tct ttg aat ttg ata tcg atg aat tcg agc acg‐5′) as primers. One‐step replacement of 638 bp of the RAD14 coding region with the KanMX4 cassette was carried out by transforming strain K699 with the rad14Δ::KanMX4 PCR product to give rise to strain YLL355. To construct strain YHN133, carrying the RAD52 chromosomal deletion, we transformed strain K699 with the BamHI linearized plasmid pSM21 (a gift from M.Fasullo, Loyola University, Chicago, IL), which carries a rad52Δ::TRP1 cassette cloned into the BamHI site of plasmid pBR322. The accuracy of the RAD14 and RAD52 gene replacements was verified by Southern blot analysis. Strains DMP2538/3A and DMP2538/11D are meiotic segregants of a cross between strains YLL355 and DMP2133/2D. Strain DMP2539/3D is a meiotic segregant of a cross of strains YLL355 and DMP2132/5A. Strain DMP2538/3A was crossed with strains YLL157, YLL244, DMP2141/1A, DMP2149/1D, YLL335, DMP2541/8A and YHN133 to generate meiotic segregants DHN133/5C, DMP2057/10A, DHN205/2C, DHN206/1A, DHN207/2D, DHN198/6D and DHN196/11C, respectively. Strain DHN219/8C was derived from a cross between strains DHN196/11C and SP1056, kindly provided by S.Piatti. Strain DHN133/5C was crossed with strain YLL262/2C to generate the meiotic segregants DHN179/1C, DHN179/4A and DHN200/11D. Strain DHN201/8A is a meiotic segregant of a cross between strains DHN198/6D and DHN179/4A. Strains DHN176/10B and DHN176/11C were obtained from a cross of strains DMP2371/9C and DMP2538/3A. Strains K699ρ°, YHN156, YHN158, YHN160, YHN159, YHN161 and YHN169 are ρ° derivatives of strains K699, DMP2539/3D, DMP2538/11D, DHN133/5C, DMP2057/10A, DHN179/1C and DHN198/6D, respectively. To generate ρ° strains, cells were plated on YEPD plates and 10 μl ethidium bromide (10 mg/ml) was spotted at the centre of the plate. After 3 days incubation at 25°C, small colonies from the middle of the zone between border and centre of the Petri dish were picked up and tested for growth on YEP‐ethanol plates (1% yeast extract, 2% bactopeptone, 50 mg/l adenine, 2% ethanol). Clones not growing on ethanol‐containing medium were tested for loss of mitochondria by DAPI staining and fluorescence microscopy.
Cells were grown at 25°C in YEPD medium (1% yeast extract, 2% bactopeptone, 50 mg/l adenine, 2% glucose). Transformants carrying the KanMX4 cassette were selected on YEPD plates containing 400 μg/ml G418 (US Biological).
Cell synchronization in G1 was obtained by treatment of exponentially growing YEPD cell cultures with 2 μg/ml of α‐factor, followed by release in YEPD. α‐factor‐arrested cells were collected by centrifugation and 2.5×108 cells were spread on 14‐cm diameter YEPD plates, followed by UV irradiation with 5 J/m2 using a Stratalinker 2400 (Stratagene). Aliquots were taken and flow cytometric DNA quantitation was determined on a Becton‐Dickinson FACScan. Survival was measured as plating efficiency of appropriate dilutions of the cultures on YEPD plates. Cell density was determined by counting aliquots fixed in 3.7% formaldehyde.
Western blot analysis
For Western blot analysis, protein extracts were prepared by TCA precipitation as described previously (Longhese et al., 1997). Protein extracts were resolved by electrophoresis on 10% SDS–polyacrylamide gels and proteins were transferred to Protran membranes (Schleicher and Schuell). Rad53 was detected using anti‐Rad53 polyclonal antibodies kindly provided by C.Santocanale (ICRF, Clare Hall Laboratories, UK). Secondary antibodies were purchased from Amersham and proteins were visualized by an enhanced chemiluminescence system according to the manufacturer's instructions.
Two‐dimensional agarose gel electrophoresis analysis
Cell cultures (2 l; 1×107 cells/ml) were grown in YEPD medium, synchronized with 2 μg/ml (0.66 μg/ml for the bar1 strain DHN219/8C) of α‐factor and split in two. One half was not irradiated, released in 500 ml YEPD medium and samples for DNA preparation (10 ml) were removed every minute from 10 to 50 min after release from the pheromone block, supplied with 100 μl of 10% NaN3, pooled and chilled with 100 ml of frozen 0.2 M EDTA pH 8, 0.1% NaN3. The other half was centrifuged and cells were spread on 10 14‐cm diameter YEPD plates (1×109 cells/plate), UV‐irradiated with 16 J/m2, resuspended and pooled in 500 ml YEPD medium. The UV dose used in these experiments was adjusted to 16 J/m2 in order to obtain the same survival as in the experiments where only 2.5×108 cells were spread on 14‐cm diameter plates. Samples for DNA preparation were removed every 2 min from 10 to 100 min after UV irradiation and pooled as described above. DNA from pooled S phase cells was prepared as previously described (Hubermann et al., 1987; Brewer et al., 1992). DNA samples (5 μg) were analysed by 2‐D gel electrophoresis as described by Friedman and Brewer (1995). First dimension gels were 0.4% agarose and second dimension gels were 1.1% agarose. Fragments probed were a 4.2‐kb EcoRV ARS305 fragment of chromosome III and a 3.2‐kb XbaI ARS501 fragment of chromosome V (Donaldson et al., 1998). The ARS305 probe was PCR‐amplified from genomic DNA using primers ARS305F (3′‐att cgc ctg aca gga cg‐5′) and ARS305R (3′‐ata acg gag gcc gaa cc‐5′). The ARS501 probe was obtained by XbaI digestion of plasmid pUC‐ARS501 and gel‐purification. Primers ARS305F and ARS305R and plasmid pUC‐ARS501 were kindly provided by B.Brewer (University of Washington, Seattle, WA). Blots were exposed for 10–14 days as required.
We are grateful to B.Brewer, S.Hunt and M.Huang for teaching H.N. the 2‐D gel technique, for help with the preliminary 2‐D gel experiments and for the gift of ARS primers and plasmids. We wish to thank M.Fasullo for providing the pSM21 plasmid, C.Santocanale for antibody against Rad53, S.Piatti and V.Paciotti for critical reading of the manuscript, K.Shirahige and all the members of our laboratory for useful discussions and criticisms. This work was supported by grants from Associazione Italiana Ricerca sul Cancro and Cofinanziamento 1997 MURST‐Università di Milano to G.L. and CNR Target Project on Biotechnology Grant CT.97.01180.PF49(F).
- Copyright © 1999 European Molecular Biology Organization