The DNA Damage Checkpoint Response Requires Histone H2B Ubiquitination by Rad6-Bre1 and H3 Methylation by Dot1*

The cellular response to DNA lesions entails the recruitment of several checkpoint and repair factors to damaged DNA, and chromatin modifications may play a role in this process. Here we show that in Saccharomyces cerevisiae epigenetic modification of histones is required for checkpoint activity in response to a variety of genotoxic stresses. We demonstrate that ubiquitination of histone H2B on lysine 123 by the Rad6-Bre1 complex, is necessary for activation of Rad53 kinase and cell cycle arrest. We found a similar requirement for Dot1-dependent methylation of histone H3. Loss of H3-Lys79 methylation does not affect Mec1 activation, whereas it renders cells checkpoint-defective by preventing phosphorylation of Rad9. Such results suggest that histone modifications may have a role in checkpoint function by modulating the interactions of Rad9 with chromatin and active Mec1 kinase.

The cellular response to DNA lesions entails the recruitment of several checkpoint and repair factors to damaged DNA, and chromatin modifications may play a role in this process. Here we show that in Saccharomyces cerevisiae epigenetic modification of histones is required for checkpoint activity in response to a variety of genotoxic stresses. We demonstrate that ubiquitination of histone H2B on lysine 123 by the Rad6-Bre1 complex, is necessary for activation of Rad53 kinase and cell cycle arrest. We found a similar requirement for Dot1-dependent methylation of histone H3. Loss of H3-Lys 79 methylation does not affect Mec1 activation, whereas it renders cells checkpoint-defective by preventing phosphorylation of Rad9. Such results suggest that histone modifications may have a role in checkpoint function by modulating the interactions of Rad9 with chromatin and active Mec1 kinase.
Eukaryotic cells respond to DNA damage by activating a variety of DNA repair pathways and by triggering the DNA damage checkpoint, a surveillance mechanism required to control cell cycle progression in response to genotoxic stress (1).
A considerable amount of information is now available relative to the key protein factors, which act in eukaryotic repair pathways. In the last few years, a lot has been learnt on the signal transduction cascade leading to cell cycle arrest following DNA damage (2,3). However, the physical and functional interplay connecting activation of the checkpoint response with specific repair mechanisms has started to emerge only recently (3)(4)(5). Since damage recognition and repair can be strongly influenced by the chromosome organization (6,7), it is now essential to extend this knowledge considering the context of a chromatin environment. Critical checkpoint factors are phosphorylated in response to DNA damage, and the order of function in the pathway has been mainly inferred by monitoring their phosphorylation state (8). In the yeast Saccharomyces cerevisiae, the Mec1-Ddc2 and the Rad17-Mec3-Ddc1 complexes are independently loaded in proximity of a DNA lesion (9,10). Phosphorylation of Ddc2 is the first biochemical event detectable after checkpoint activation, and this modification does not require any checkpoint factor other than the Mec1 kinase itself (11)(12)(13). The Rad17-Mec3-Ddc1 complex, which shows a structural similarity to PCNA, 1 is also loaded near a DNA lesion and is phosphorylated on Ddc1 by the kinase activity of Mec1. Another critical target of Mec1 is the protein Rad9, whose phosphorylation may modulate its local concentration on DNA and is required for the recruitment and activation of Rad53 kinase (14 -17). This is a key step in the signal transduction cascade, and it is generally used as a marker to monitor checkpoint activation (18).
The cellular response to DNA damage is probably influenced by the structure of chromatin, much like all other DNA transactions (19,20), and it is tempting to speculate that local changes in chromatin structure near a DNA lesion may facilitate its identification contributing to the recruitment of repair and checkpoint proteins. Histones are the targets of a number of covalent modifications, such as acetylation, methylation, phosphorylation, and ubiquitination, which influence a variety of DNA metabolic processes. It is well established that, in mammalian cells, Ser 139 in the C-terminal tail of the H2AX histone variant is phosphorylated in response to DSBs, treatments with genotoxic agents, and stalled replication forks (21,22). Similarly, phosphorylation of Ser 129 in the C-terminal tail of H2A occurs both in budding and fission yeast following DNA damage (23). In Saccharomyces cerevisiae, the only histone species currently known to be ubiquitinated is histone H2B (24). Protein ubiquitination requires the activity of three different enzymes. A ubiquitin-activating enzyme (UBA or E1) covalently binds an ubiquitin molecule and transfers it to a ubiquitin-conjugating enzyme (UBC or E2). This physically associates with a ubiquitin ligase (E3), which allows conjugation of the ubiquitin moiety to specific substrates. It is well established that a single E2 can play relevant roles in a variety of pathways by interacting with different E3 enzymes. A number of studies have implicated ubiquitinated H2B in heterochromatic gene silencing and in the control of gene expression of several yeast genes (25). It has been shown that the activity of the GAL1 promoter is regulated through ubiquitination of H2B. Rad6 plays a major role in this process, which requires both the Gal4 activator and Bre1, one of the Rad6-associated E3 ligases (26). Rad6 is an evolutionarily conserved ubiquitinconjugating enzyme (E2) with multiple roles in postreplication repair (PRR), telomere silencing, protein degradation, and sporulation (26 -30). All cellular functions of Rad6 require its E2 activity, in fact mutations of Cys 88 in the active site confers a rad6⌬ phenotype (31). The RAD6 gene is the prototype of the PRR epistasis group, whose components act on stalled replication forks in the presence of DNA damage, allowing repair and resumption of DNA replication (27,28). PRR not only requires the action of Rad6, but also the product of the RAD18 gene. It has been shown that Rad6 physically interacts with Rad18, which acts as an E3 with Rad6 in PRR (27,28,32). Additional functions of Rad6 seem to require different specific E3 enzymes, such as Ubr1 in protein degradation (30) and Bre1 in the regulation of gene expression and silencing (26,33,34). The Rad6-Bre1 complex is responsible for ubiquitination of histone H2B-Lys 123 , and this modification is instrumental for the methylation of histone H3-Lys 4 and H3-Lys 79 by the Set1 and Dot1 DNA methyltransferases, respectively. The level of methylation at these residues strongly influences silencing and promoter activation (33)(34)(35)(36)(37).
In this paper, we demonstrate that Rad6 plays a new and key role in the activation of the DNA damage checkpoint in response to a variety of genotoxic stresses. Surprisingly, Rad18 is not acting with Rad6 in checkpoint signaling, but such a role is carried out by Bre1, through ubiquitination of histone H2B. Moreover, DNA damage signaling to Rad53 kinase requires methylation of H3-Lys 79 by Dot1, indicating a novel role of the histone code in the cellular response to DNA lesions. Loss of such histone modifications causes a checkpoint defect by impairing the function of Rad9; in these conditions, Mec1 kinase is normally activated.
Cell Cycle Blocks and UV Irradiation-Cells were grown in YEPD medium at 28°C to a concentration of 5 ϫ 10 6 cells/ml and arrested with nocodazole (20 g/ml) or ␣-factor (20 g/ml). 50 ml of exponentially growing or G 1 /G 2 -arrested cells were spun, resuspended in 500 l of fresh YEPD, and plated on a Petri dish (14-cm diameter). Plates were quickly irradiated with a Stratalinker at 100 J/m 2 . Immediately after treatment, cells were resuspended in 20% trichloroacetic acid for protein extract preparation.
Treatment with Genotoxic Agents-Exponentially growing cells were spun and resuspended in fresh YEPD medium containing 0.02% methyl methane sulfonate (MMS) or 200 mM hydroxyurea (HU) and incubated for 3 h before trichloroacetic acid protein extract preparation. Zeocin (ZEO) and 4-nitroquinoline N-oxide were used at 200 and 2 g/ml, respectively; cells were kept in the presence of the drugs for 30 and 15 min, respectively. Trichloroacetic acid extracts were then prepared as described above.
Synchronization Experiments-Cell synchronization in G 1 was obtained by treating exponentially growing cell cultures with 2 g/ml ␣-factor. Yeast cells were synchronized in G 2 by treating exponentially growing YEPD cell cultures with 2 g/ml nocodazole in 1% Me 2 SO. For G 2 -M checkpoint assay, the cells were stained with 4Ј,6-diamidino-2phenylindole, and nuclear division was monitored by microscopic analysis. UV and MMS treatments were performed as described previously (4). Flow cytometric DNA quantitation was performed with a FACScan cytofluorimeter after staining with Sytox Green (Molecular Probes, Inc., Eugene, OR).

RESULTS
Rad6 is implicated in ubiquitination of nucleosomes, and rad6⌬ cells are extremely sensitive to a variety of genotoxins This work (27,34). Moreover, Rad6 E2 activity is necessary for postreplication repair (31). Therefore, we considered Rad6 a good candidate to integrate checkpoint activation and DNA repair with chromatin modifications. We tested the functionality of the DNA damage checkpoint by monitoring phosphorylation of Rad53 and its ISA, in exponentially growing wild-type (WT) and rad6⌬ cells after UV treatment. As shown in Fig. 1A, rad6⌬ cells are defective both in Rad53 phosphorylation and in the ISA assay, following UV irradiation. We recently showed that a nucleotide excision repair-defective rad14⌬ strain could not activate the G 1 and G 2 checkpoints after UV treatment, whereas the intra-S checkpoint was still proficient (4). The defect observed in exponentially growing rad6⌬ cells was stronger than that found in a rad14⌬ strain, although the cell cycle distribution measured by FACS analysis was equivalent in the two mutant populations (data not shown). Surprisingly, no checkpoint alteration was detectable in other PRR-defective mutant strains such as rad18⌬ and rad5⌬.
We then tested the role of RAD6 in checkpoint activation in G 1 -and G 2 -arrested cells (Fig. 1B). Rad53 phosphorylation after UV irradiation is strongly compromised in rad6⌬ G 1blocked cells, whereas it is only partially defective in G 2 cells.
Quantification of Rad53 kinase activity with the ISA assay demonstrated that activation of the G 1 checkpoint in rad6⌬ cells is reduced by 90% compared with the WT level, whereas in G 2 Rad53 activity is diminished only to 50%. The different requirements for RAD6 in the G 1 and G 2 phases of the cell cycle indicates that a pathway, at least partially independent on RAD6, is active in G 2 cells. Rad53 phosphorylation was found to be fully proficient in rad18⌬ and rad5⌬ strains both in G 1and G 2 -arrested cells. These findings indicate that Rad6 function in checkpoint signaling is uncoupled from its function in PRR.
We extended our analysis to a variety of genotoxins, other than UV irradiation. As it is shown in Fig. 1C, rad6⌬ cells are defective in Rad53 phosphorylation also in response to MMS, HU, and ZEO treatments that cause DNA alkylation, replication fork arrest, and DSBs, respectively. Also in this assay, rad18⌬ cells show a behavior indistinguishable from that of a WT strain.
Rad6 is an E2 enzyme with multiple functions in several cellular processes, and Rad18, its E3 in PRR, does not seem to be involved in checkpoint function. We thus tested the response to DNA damage in S. cerevisiae strains carrying deletions in all putative E3-encoding genes present in the yeast genome (data not shown). The only E3 gene required for phosphorylating Rad53 in response to different genotoxins is BRE1 (Fig. 2A). The phenotype observed in bre1⌬ cells parallels the one found in rad6⌬ strains, in exponentially growing cells ( Fig. 2A), and in G 1 -or G 2 -arrested conditions, where Rad53 phosphorylation is abolished or strongly reduced, respectively (Fig. 2B). We next analyzed the effect of BRE1 deletion on the control of cell cycle progression following DNA damage. WT, bre1⌬ cells, and a checkpoint-deficient control strain (mec3⌬) were treated with UV light and released from a G 1 arrest, and their S phase entry and cell cycle progression were monitored by FACS analysis. Fig. 3A shows that bre1⌬ cells, similarly to a mec3⌬ strain, do not arrest at the G 1 -S transition in response to UV irradiation. In fact, they start to accumulate a 2C DNA content already 75 min after release from the G 1 block, whereas S phase entry is delayed in WT cells. bre1⌬ cells also failed to delay the kinetics of bud emergence after UV irradiation in G 1 , similarly to what is found in the checkpoint-defective mec3⌬ strain (38) (data not shown). BRE1 deletion also impairs the intra-S DNA damage checkpoint. In fact, when bre1⌬ cells are released from a G 1 arrest in the presence of sublethal concentrations of MMS, they fail to slow down S phase progression, similarly to mec3⌬ cells (Fig. 3A). The checkpoint defect of bre1⌬ cells is further sup- FIG. 1. RAD6 is required for Rad53 activation in response to DNA damage, whereas RAD18 and RAD5 are dispensable. A, exponentially growing cultures of the strains K699 (WT), YMG121 (rad6⌬), YMG120 (rad18⌬), YMG119 (rad5⌬), and YMIC12H6 (rad14⌬) were irradiated with 100 J/m 2 . Protein extracts were prepared and separated by SDS-PAGE. Western blot was stained with ␣-Rad53 antibodies (left) or treated for the ISA assay (right). B, the same yeast cultures analyzed in A were blocked in G 1 or G 2 phases of the cell cycle and UV-irradiated with 100 J/m 2 . Western blot (top) and ISA assay (bottom) are shown. C, exponentially growing cultures of the strains K699 (WT), YMG121 (rad6⌬), and YMG120 (rad18⌬) were spun and resuspended in fresh YEPD medium containing MMS, HU, or ZEO, as described under "Experimental Procedures." The same cultures were also treated with UV (100 J/m 2 ) as described. Rad53 activation was monitored by Western blotting by using specific antibodies. ported by evaluating the kinetics of Rad53 phosphorylation. In fact, when WT cells are arrested with ␣-factor, UV-irradiated, and then released from the G 1 block, Rad53 is fully phosphorylated already at the earliest time point (Fig. 3B). On the contrary, Rad53 is only partially modified even 30 min after release in bre1⌬ and checkpoint-defective mec3⌬ cells. Whereas bre1⌬ cells are impaired in the G 1 -S and intra-S checkpoints, they are proficient in the G 2 -M response; indeed, they normally delay nuclear division after UV irradiation (Fig. 3C). The proficiency of the G 2 -M checkpoint in bre1⌬ cells may explain their low sensitivity to UV, MMS, and HU treatments (data not shown).
Bre1 is a RING finger protein with structural similarities with other ubiquitin ligases, and it was recently found as the E3 involved in the ubiquitination of H2B-K123 by Rad6 (33,34). This H2B modification has been linked to transcriptional regulation in S. cerevisiae (26). Since the Rad6-Bre1 complex targets the Lys 123 residue in histone H2B, we directly tested the response to DNA damage in a strain carrying the htb1-K123R mutation. As shown in Fig. 4A, htb1-K123R G 1 -arrested cells, in which H2B cannot be ubiquitinated by the Rad6-Bre1 complex, fail to phosphorylate Rad53 following UV irradiation. Similarly, htb1-K123R mutant cells are defective in activating Rad53 in response to MMS (Fig. 4B) and other genotoxic agents (data not shown). Accordingly, htb1-K123R mutant cells are defective in arresting the G 1 -S transition following UV damage (G 1 checkpoint) and in slowing down DNA synthesis in the presence of MMS (intra-S checkpoint). Fig. 4C shows that whereas WT cells delay entry into S phase after UV irradiation, compared with untreated conditions, htb1-K123R cells are impaired in postponing DNA replication, like a checkpoint-defective strain. Similarly, the htb1-K123R strain is unable to delay bud emergence after UV treatment in G 1 (data not shown).
Moreover, htb1-K123R cells also fail to slow down the kinetics of DNA synthesis in the presence of MMS. These data demonstrate that histone H2B-K123 ubiquitination by Rad6 and Bre1 is a prerequisite for checkpoint activation.
Studies on the transcriptional regulation of the GAL1 promoter showed that ubiquitination of H2B-K123 by the Rad6-Bre1 complex is instrumental for methylation of histone H3 on Lys 4 and Lys 79 residues by the Set1 and Dot1 DNA methyltransferases, respectively (33)(34)(35)(36). H3-Lys 36 is also methylated, but this modification is not dependent on H2B-Lys 123 ubiquitination and is performed by Set2 (39). We deleted the genes coding for the three DNA methyltransferases responsible for H3 methylation and tested the response to DNA damage in strains carrying these mutations. As shown in Fig. 5A, exponentially growing dot1⌬ cells are defective in phosphorylating Rad53 following UV, MMS, and ZEO treatments, whereas set1⌬ and set2⌬ strains behave as WT cells in checkpoint activation. Similarly to what was found in rad6⌬ and bre1⌬ cells, the dot1⌬ mutant strain is fully impaired in phosphorylating Rad53 in G 1 -arrested cells (Fig. 5B).
By FACS analysis, we observed that dot1⌬ cells display a deficiency in the intra-S checkpoint, whereas they are only partially defective in delaying the G 1 -S transition (Fig. 6A). On the other hand, bre1⌬ cells are defective both in the G 1 -S and intra-S checkpoints, similarly to what was observed in mec3⌬ cells (Figs. 3A and 6A). Since Bre1 is required for both H3-Lys 79 methylation by Dot1 and H3-Lys 4 methylation by Set1, we supposed that H3-Lys 4 methylation might play a redundant role in checkpoint function. set1⌬ cells are not deficient in the G 1 checkpoint; in fact, these cells delay entry into S phase after UV treatment, similarly to a WT strain (data not shown) (40). However, when DOT1 and SET1 deletions are combined, the   FIG. 3. bre1⌬ cells are defective in the G 1 -S and intra-S DNA damage checkpoint. A, K699 (WT), YFL224 (bre1⌬), and YMIC4E6 (mec3⌬) cells were arrested in G 1 phase with ␣-factor. Cultures were treated with 50 J/m 2 or left untreated and then released from the G 1 arrest. For intra-S checkpoint analysis, 0.02% MMS was added to the medium after the release. Samples were taken every 15 min for FACS analysis. B, Rad53 phosphorylation state was monitored by Western blotting after UV irradiation from proteins sample taken at the indicated times from the experiment described in A. C, the same cultures used in A were arrested in G 2 with nocodazole and released after UV treatment (50 J/m 2 ); the percentage of binucleate cells was scored microscopically.
double mutant cells display a strong G 1 -S checkpoint defect, similarly to a bre1⌬ strain (Fig. 6A). The same conclusion can be drawn by analyzing the kinetics of Rad53 phosphorylation following UV irradiation (Fig. 6B). Therefore, H3-Lys 79 methylation by Dot1 is required for proper checkpoint activation, whereas the minor role of H3-Lys 4 methylation by Set1 can be unmasked only when the function of Dot1 is impaired. The importance of H3-Lys 79 methylation is directly demonstrated by the finding that exponentially growing h3-K79A and dot1⌬ mutant cells show a similar defect in Rad53 phosphorylation after UV and MMS treatments (Fig. 6C). Since h3-K79A mutant cells are strongly defective in silencing and cannot be arrested by ␣-factor treatment, the G 1 checkpoint cannot be directly tested in such a genetic background.
We were interested in how histone modifications control checkpoint activation. Information on the mechanism could be obtained by identifying which step of the phosphorylation cascade is blocked in bre1⌬ and dot1⌬ cells. Ddc2 modification is the first detectable biochemical event of the checkpoint signal transduction cascade. Fig. 7A shows that Ddc2 is normally phosphorylated in response to UV lesions, both in bre1⌬ and dot1⌬ strains. Similar results were found for Ddc1 phosphorylation (Fig. 7B). These observations indicate that Bre1 and Dot1 are not required for the activation of Mec1 kinase or the loading of the PCNA-like complex near the lesion. On the other hand, bre1⌬ and dot1⌬ G 1 -arrested cells are totally deficient in phosphorylating Rad9; in fact, after UV irradiation, only the unphosphorylated form can be detected in such mutants (Fig. 7C). Altogether, these findings indicate that histone modifications specifically influence the DNA damage-induced phosphorylation of Rad9, the adaptor protein necessary for Rad53 activation. DISCUSSION Cells are continuously under the assault of endogenous and exogenous genotoxins that challenge the integrity of DNA. To cope with such a formidable task, cells have evolved surveillance mechanisms and a variety of DNA repair systems. The DNA damage checkpoint responds to genomic insults and replication stress by integrating several DNA transactions (transcription, repair, replication, and recombination) with the regulatory circuits controlling cell cycle progression.
The definition of the signal, which activates the checkpoint cascade and the functional and physical interplay between checkpoint factors and repair mechanisms is a major issue in the field. These interconnections are best known in the case of the repair of DSBs where repair and checkpoint factors aggregate into cytological visible foci (41). The importance of chromatin structure for efficient DSBs repair is underscored by the   FIG. 4. Lysine 123 of H2B is required for Rad53 activation, G 1 -S and intra-S response to DNA damage. Exponentially growing cultures of the strains K699 (WT), YMG121 (rad6⌬), Y133 (htb1-K123R), and its WT control Y131 (HTB1) were divided in two aliquots. One of them was arrested in G 1 and UV-irradiated with 100 J/m 2 (A); the other was treated with MMS (0.02%) for 3 h (B). Rad53 phosphorylation was detected on Western blots using ␣-Rad53-specific antibodies. C, one aliquot of the same cultures used in A was arrested in G 1 , mock-or UV-irradiated (50 J/m 2 ), and released in fresh medium while another aliquot was released in the presence of MMS. Cell cycle progression was monitored by FACS analysis. rapid phosphorylation of H2AX at the foci (22,42). Such modification is required for prolonged maintenance of DSB foci, whereas it appears to be dispensable for the initial recognition of the lesion (42)(43)(44)(45)(46).
In order to extend our knowledge on the interplay between DNA repair, chromatin regulation, and checkpoint activation, we analyzed a possible role for RAD6 in the cellular response to DNA damage. Rad6 is an E2 enzyme that, together with Rad18, a ubiquitin ligase (E3), plays a fundamental role in PRR by modulating ubiquitination of PCNA and possibly other substrates (47,48). When we analyzed the checkpoint response to UV irradiation of a rad6⌬ strain, we observed a clear defect in the activation of the Rad53 kinase. This was particularly evident in cells damaged during the G 1 phase of the cell cycle, but it was also manifest in exponentially growing cells. In G 2arrested cells, activation of Rad53 is only partially affected by the loss of Rad6 function, indicating the possible existence of a partially overlapping Rad6-indepedent pathway acting on the G 2 -M checkpoint. Deletion of RAD6 greatly reduces Rad53 phosphorylation in response to a variety of genotoxic agents (UV, 4-nitroquinoline N-oxide, MMS, and ZEO), which cause a multiplicity of DNA lesions. Moreover, the finding that rad6⌬ FIG. 6. Dot1 is required for the intra-S response to MMS and cooperates with Set1 in the G 1 -S checkpoint. Logarithmically growing WT (K699), YFL224 (bre1⌬), YFL234 (dot1⌬), and YMG198 (set1⌬ dot1⌬) were synchronized with ␣-factor and resuspended in YEPD after UV irradiation (50 J/m 2 ) or in medium with MMS (0.02%). Cell cycle progression was monitored by FACS analysis (A). Protein extracts prepared from cell samples were collected at the indicated times were subjected to Western blot analysis with anti-Rad53 antibodies (B). C, UCC7201 (WT), UCC7202 (dot1⌬), and UCC7223 (h3-K79A) strains were treated with MMS or UV-irradiated as described in Fig. 5A. Protein extracts were analyzed by Western blotting. cells are impaired in checkpoint activation in response to HU indicates that RAD6 is playing a role also in the proper response to replication forks arrest. RAD6 is the founding member of the PRR epistasis group, but, surprisingly, we found that its role in checkpoint activation is unlinked from the one in PRR. In fact, deletion of other PRR genes, such as RAD18 and RAD5, did not cause any checkpoint defect.
Rad6 exerts its ubiquitin conjugation function, interacting with different E3 proteins, such as Rad5, Ubr1, and Bre1, in a multiplicity of cellular pathways (27,30,33). We analyzed UV-induced Rad53 phosphorylation in yeast strains deleted in genes coding for all putative E3 proteins; this allowed us to identify the ubiquitin ligase working with Rad6 in checkpoint activation. We found that deletion of BRE1 impairs checkpoint activation similarly to rad6⌬. Bre1 has been recently reported to be the E3 involved in the monoubiquitination of histone H2B on the Lys 123 residue during GAL1 promoter activation (26). We showed that bre1⌬ cells are defective in delaying entry into S phase in response to UV irradiation and fail to slow down the kinetics of DNA synthesis in the presence of a sublethal concentration of MMS. The G 1 -S and intra-S checkpoint defects of bre1⌬ cells are similar to what found in mec3⌬ cells, whereas the G 2 -M checkpoint is proficient in bre1⌬ strains; this observation might explain the low sensitivity to genotoxic treatments displayed by bre1⌬ cells.
The importance of H2B-Lys 123 ubiquitination in checkpoint activation is demonstrated by the observation that the checkpoint defects of a htb1-K123R strain mimic those observed in rad6⌬ and bre1⌬ cells. Recent work demonstrated that, at specific promoters, H2B-K123 ubiquitination is followed by H3 methylation at Lys 4 and Lys 79 by the Set1 and Dot1 methyltransferases, respectively (33)(34)(35)(36)(37). When we deleted the genes coding for the methyltransferases that modify histone H3, we found a marked checkpoint impairment in dot1⌬ strains, whereas SET1 and SET2 deletions per se did not cause any evident checkpoint defect. Loss of H3-Lys 79 methylation, caused by a point mutation, mimics the effect of the DOT1 deletion, arguing that the important target of Dot1 in checkpoint activation is H3-Lys 79 . However, H3-Lys 4 methylation by Set1 partially cooperates with H3-Lys 79 by Dot1 in controlling the G 1 -S checkpoint, since the checkpoint defect in set1⌬ dot1⌬ double mutant cells is more evident than that observed in the dot1⌬ strain. DOT1 is required for silencing in yeast, and either loss of its function or overproduction of its gene product leads to disruption of silencing (37). The checkpoint defect described so far is instead specific for a dot1⌬ strain; in fact, overexpressing Dot1 does not affect Rad53 phosphorylation (data not shown), suggesting that the deficiency in checkpoint function is not due to disruption of silencing.
To understand how Rad6-Bre1, Dot1, and the histone code could influence activation of the checkpoint, we analyzed at what step of the cascade histone modifications are required for the transmission of the signal. Since modification of Rad53 is a late event in the pathway, we monitored the phosphorylation state of other critical factors. Two protein complexes, Mec1-Ddc2 and Rad17-Mec3-Ddc1 are independently loaded near DNA lesions (9,41). Active Mec1 kinase phosphorylates Ddc2, Ddc1, and Rad9. This last event is required for oligomerization of Rad9 and activation of Rad53. We have previously shown that processing of UV-induced lesions by nucleotide excision repair factors is required for Ddc2 and Ddc1 phosphorylation, and it is a prerequisite for loading of the Mec1-Ddc2 and the PCNA-like complexes to UV-damaged chromosomes (4). Here we found that, in bre1⌬ and dot1⌬ cells, Ddc2 and Ddc1 are phosphorylated after genotoxic stress, indicating that Mec1 can be activated even in the absence of chromatin modifications induced by the Rad6-Bre1 complex or Dot1. On the contrary, both mutant strains were unable to hyperphosphorylate Rad9 after treatments with various genotoxic agents. Altogether, these findings indicate that the signal transduction cascade is blocked between phosphorylation of the PCNA-like complex and that of Rad9. These epigenetic modifications of histones are therefore not needed for the activation of Mec1 kinase or for the loading of the PCNA-like complex to the site of lesion, whereas they are essential for allowing Mec1 to phosphorylate Rad9.
Rad9 is a checkpoint mediator that physically associates with regions corresponding to HO-induced DSBs (49) and is required to transduce the signal to Rad53. Rad9 mutants lacking multiple phosphorylation sites fail to oligomerize on DNA and do not associate tightly to DSBs, indicating a close connection between Rad9 phosphorylation and its assembly at the sites of DNA damage (49 -51). Having found only unphosphorylated Rad9 in genotoxin-treated bre1⌬ or dot1⌬ G 1 -arrested cells, we infer that recruitment of Rad9 near DNA lesions is impaired in such mutant backgrounds. We could not directly test this hypothesis using HO-induced DSBs, because this type of damage specifically activates the G 2 -M checkpoint (52), which is only partially affected in the absence of Bre1 or Dot1. While this manuscript was in preparation, in vitro data were published showing that a recombinant glutathione S-transferase-Rad9 fusion interacts with calf thymus histone H3 (53). The same paper shows that mammalian 53BP1 fails to congress to sites of damage in the absence of H3 methylation, strongly supporting our physiological data. Since in exponentially growing yeast cells 90% H3-Lys 79 is constitutively methylated (37), nonmutually exclusive hypothesis can be put forward to explain our observations. (i) Rad9 may constitutively interact with methylated H3, leading to a high Rad9 concentration close to DNA. Upon lesion detection and Mec1 activation, Rad9 can then be promptly phosphorylated, leading to oligomerization and accumulation at the site of damage. (ii) DNA insults trigger Mec1 activity and a specific ubiquitination of histone H2B inducing a localized change in the level of H3-Lys 79 methylation. This modification allows Rad9 hyperphosphorylation and recruitment to the site of lesion. (iii) Genotoxic agents may generate additional modifications of chromatin structure (54), which expose methylated H3-Lys 79 , leading to Rad9 recruitment and activation.
Our findings that H2B-Lys 123 ubiquitination and H3-Lys 79 methylation influence the DNA damage response suggest that specific checkpoint proteins and/or additional factors may be able to sense variations in chromatin structure or topology. However, the precise understanding of how the histone code influences dynamic changes in chromatin structure or proteinprotein and protein-DNA interactions near DNA damage sites remains a major challenge for future research.