The RAD6/BRE1 Histone Modification Pathway in Saccharomyces Confers Radiation Resistance Through a RAD51-Dependent Process That Is Independent of RAD18
John C. Game, Marsha S. Williamson, Tatiana Spicakova, J. Martin Brown

Abstract

We examine ionizing radiation (IR) sensitivity and epistasis relationships of several Saccharomyces mutants affecting post-translational modifications of histones H2B and H3. Mutants bre1Δ, lge1Δ, and rtf1Δ, defective in histone H2B lysine 123 ubiquitination, show IR sensitivity equivalent to that of the dot1Δ mutant that we reported on earlier, consistent with published findings that Dot1p requires H2B K123 ubiquitination to fully methylate histone H3 K79. This implicates progressive K79 methylation rather than mono-methylation in IR resistance. The set2Δ mutant, defective in H3 K36 methylation, shows mild IR sensitivity whereas mutants that abolish H3 K4 methylation resemble wild type. The dot1Δ, bre1Δ, and lge1Δ mutants show epistasis for IR sensitivity. The paf1Δ mutant, also reportedly defective in H2B K123 ubiquitination, confers no sensitivity. The rad6Δ, rad51null, rad50Δ, and rad9Δ mutations are epistatic to bre1Δ and dot1Δ, but rad18Δ and rad5Δ show additivity with bre1Δ, dot1Δ, and each other. The bre1Δ rad18Δ double mutant resembles rad6Δ in sensitivity; thus the role of Rad6p in ubiquitinating H2B accounts for its extra sensitivity compared to rad18Δ. We conclude that IR resistance conferred by BRE1 and DOT1 is mediated through homologous recombinational repair, not postreplication repair, and confirm findings of a G1 checkpoint role for the RAD6/BRE1/DOT1 pathway.

RECENT research in eukaryotes has demonstrated a much greater role than was initially perceived for histone modifications in basic cellular processes, including transcription, gene silencing, control of carcinogenesis, and responses to DNA damage. As part of this, we reported that Saccharomyces strains deleted for any of several genes involved in histone modifications are substantially more sensitive than wild type to the lethal effects of ionizing radiation (IR) (Game et al. 2005). The mutants included strains deleted for the DOT1 gene, which encodes the methylase that acts on the lysine 79 residue (K79) of the histone H3 protein (Feng et al. 2002; van Leeuwen et al. 2002), as well as histone H3 mutants in which wild-type Dot1p cannot act because its target lysine is replaced with another amino acid. These findings complemented information from other laboratories that implicates histone H3 lysine 79 methylation in controlling the DNA damage checkpoint induced by ultraviolet radiation and other agents in yeast (Giannattasio et al. 2005; Wysocki et al. 2005) and in damage recognition by the checkpoint protein 53BP1 in mammalian cells (Huyen et al. 2004).

Substantial information is available indicating that the DOT1-mediated methylation of H3 K79 is dependent on the prior modification of histone H2B involving ubiquitination of lysine 123 in Saccharomyces (Briggs et al. 2002; Ng et al. 2002a) or lysine 120 in mammals (Kim et al. 2005). Recently, it was shown that H3 K79 trimethylation and some di-methylation is dependent on H2B K123 ubiquitination, whereas mono-methylation of K79 still occurs fully even in mutants that fail to modify H2B K123 (Shahbazian et al. 2005). The Rad6 ubiquitin conjugase and the Bre1 ubiquitin ligase together ubiquitinate H2B K123 (Robzyk et al. 2000; Hwang et al. 2003; Wood et al. 2003a). In addition, the LGE1 gene product has been found to complex with Bre1 protein and is required for its function (Hwang et al. 2003), and mutants involving some members of the RNA polymerase II-associated PAF1 complex, specifically deletions of the RTF1 and PAF1 genes, have also been reported to abolish H2B K123 ubiquitination (Ng et al. 2003a; Wood et al. 2003b). Most recently, the Bur1/Bur2 cyclin-dependent protein kinase has also been implicated in H2B K123 ubiquitination through its role in activating the Rad6 protein by phosphorylation (Wood et al. 2005).

Given this information, and to better understand the role of the RAD6 gene in different DNA repair pathways, we chose to study the X-ray sensitivity of additional Saccharomyces histone modification mutants, including those with reported defects in H2B K123 ubiquitination and H3 K79 methylation and those involved in methylation elsewhere on histone H3. In addition, we constructed double-, triple-, and multiple-mutant strains involving H2B K123 ubiquitination and H3 K79 methylation mutations combined with each other and with key mutations in previously known DNA repair pathways. We assessed IR sensitivity in these strains to determine epistasis relationships for this phenotype both within the proposed BRE1/DOT1-mediated histone modification pathway and between this pathway and others to identify the probable IR repair processes involved.

With the exception of paf1 deletion strains, we found increased sensitivity to X-rays in all the mutants that we tested that are reported to affect histone H3 K79 methylation. We also found that set2 mutants, which fail to methylate histone H3 lysine 36, show mild X-ray sensitivity, whereas mutants that abolish histone H3 lysine 4 methylation retain wild-type resistance to X-rays. We obtained evidence that genes required for histone H3 K79 methylation predominantly fall into a single RAD6-dependent IR epistasis group that falls outside the well-known family of recovery processes mediated by RAD6/RAD18-dependent postreplication/translesion synthesis mechanisms. Instead, these histone modification genes appear to function in a process that facilitates RAD51-dependent homologous recombinational repair (HRR), although they are not completely required for such repair since significant RAD51-dependent IR resistance remains in dot1Δ, bre1Δ and related mutants. We show, in agreement with evidence from others, that some aspects of the DNA damage response cell-cycle checkpoints are abrogated in mutants unable to methylate histone H3 K79, and discuss this as a possible cause of their IR sensitivity.

MATERIALS AND METHODS

Yeast strains:

As described earlier (Game et al. 2005), our starting strains were from the library of ∼4700 individual haploid deletion strains in the α mating type (background strain BY4742) created by an International Consortium and obtained from Research Genetics, Huntsville, Alabama (now Invitrogen Life Technologies). The genotype of strain BY4742 and the construction of the deletion strains have been described (Brachmann et al. 1998; Winzeler et al. 1999). Information is also available at the Saccharomyces Genome Deletion Project website at http://www-sequence.stanford.edu/group/yeast_deletion_project/deletions3.html. We also used our background-isogenic strains MW5067-1C and g1201-4C, described earlier (Game et al. 2005), as a wild type for survival curves and a wild-type MATa parent for initial crosses with MATα mutants from the deletion strain library, respectively. For crosses involving rad51, we primarily used a rad51∷URA3 disruption null allele (originally obtained from Vladimir Larionov) that we had backcrossed eight times into the library background to give expected unlinked nonisogenicity <1%. This enabled us to use the URA3 marker in place of KanMX4 to quickly distinguish rad51 from other mutants in crosses. In the text, we refer to the rad51∷URA3 allele as rad51null and the rad51∷KanMX4 replacement allele from the library as rad51Δ. Strains containing either of these rad51 alleles show equivalent survival curves, as shown in Figure 7.

Genetic methods and media:

Genetic methods including tetrad dissection were as described (Sherman et al. 1982). Cultures were routinely incubated at 30°. Rich media (YPD) and supplemented minimal media were prepared as described (Sherman et al. 1982). To induce meiosis, we incubated cultures for 4 or more days, usually at 30°, on solid Fogel's sporulation medium. This contains 9.65 g potassium acetate, 1 g glucose, 2.5 g yeast extract (Difco), and 2% agar per liter. To score geneticin (GEN) resistance, hygromycin B (HYG) resistance, or nourseothricin (NAT) resistance, we used YPD plates separately supplemented with geneticin (Sigma, St. Louis), hygromycin B (Research Products International), or nourseothricin (Werner BioAgents) added from filter-sterilized solution shortly before pouring plates to give a final concentration of 150 μg/ml (GEN), 300 μg/ml (HYG), or 100 μg/ml (NAT), respectively.

Transformations:

To facilitate scoring multiple deletion mutations in crosses, for several relevant genes we replaced the KanMX4 cassette that was used to create the original deletion library with cassettes containing LEU2 (obtained from James A. Brown, Stanford University) or a hygromycin B (HYG) or nourseothricin resistance (NAT) gene (obtained from Beth Rockmill, Yale University), using described cassettes (Goldstein et al. 1999) and a standard transformation procedure (Ito et al. 1983). To restore BRE1 to bre1Δ mutant strains, we transformed with a CEN URA3 plasmid containing BRE1, obtained from James A. Brown, using a lithium acetate procedure (Gietz et al. 1995).

Determining X-ray sensitivity:

As described earlier (Game et al. 2005), for X-ray exposures we used a Machlett OEG 60 X-ray tube with a beryllium window and a Spellman power supply operated at 30 kV and 15 mA to deliver a dose rate of 1.3 Gy (130 rad)/sec of “soft” X-rays. To determine initially whether a mutant strain was likely to exhibit IR sensitivity, we essentially followed the spot-testing procedure described previously (Game et al. 2005). To quantify the degree of sensitivity, we obtained X-ray survival curves using log-phase cells from overnight liquid YPD cultures, freshly sonicated to reduce any clumpiness, as described in the same article. Colonies were counted after incubation for 4–6 days at 30°. We obtained survival curves for at least two separate strains for most of the single-, double-, or multiple-mutant genotypes that we present here, and in many cases additional survival assays (not shown) over part or all of the dose range served to confirm our findings. For the most part, we find good agreement in X-ray sensitivity between different spore clones with the same genotype in the same genetic background. We prefer to present individual survival curves instead of averaging measurements at each dose from separate curves, both because dose points within a curve are related by serial dilutions and because their statistical robustness will vary from curve to curve on the basis of colony counts as well as on the accuracy of the unirradiated control. This means that error bars calculated for a mean value based on separate curves can be misleading (see Game et al. 2005). In addition, despite our isogenic genetic background, we prefer to obtain confirmatory survival curves using separate spore clones rather than repeating the same strain, as a better control for modifier mutations that might arise. Clearly, taking average values for separate strains would obscure any variability that we hope to expose.

Ultraviolet radiation treatments:

Log-phase cells were prepared for UV survival as for X-ray curves. They were irradiated on YPD plates using a shielded apparatus containing five General Electric G8T5 tubes giving most of their radiation at 254 nm. Plates were incubated in the dark and colonies counted as for X-ray curves.

Cell cycle checkpoint studies:

Standard methods (Day et al. 2004) were adapted as follows: To study the IR-induced G1 checkpoint, cells were arrested at G1 using α-factor (Zymo Research). One microliter of 10 mm α-factor was added to 2 ml of log-phase cells shaken in liquid YPD at OD ∼0.2 at 30°. After 1.5 hr, a second microliter was added and synchrony was assessed microscopically after another 1.5 hr. The cultures were split: one-half was irradiated using a 137Cs source (Mark 1 model 3 from J. L. Shepherd, San Francisco; dose rate 28.4 Gy/min) and one-half was mock treated. After irradiation, cells were released from the block and 0.25-ml aliquots were fixed in 70% ethanol at 15-min intervals. Fixed cells were spun down and washed with 1 ml 0.05 m sodium citrate. Cell pellets were mixed with 0.5 ml 0.05 m sodium citrate containing 0.25 mg/ml RNase A and incubated at 50° for 1 hr. After addition of propidium iodide (16 μg/ml final concentration), samples were incubated at room temperature for 30 min, briefly resonicated, and analyzed by flow cytometry (Nash et al. 1988) with a FACSCalibur machine.

To study the IR-induced G2 checkpoint, nocodazole (Sigma) was added (15 μg/ml final concentration) to midlog-phase cultures (OD ∼0.2) shaking in liquid YPD and cells were incubated for 2.5 hr at 30° to achieve >90% large buds. Cultures were split and then mock treated or irradiated with a 137Cs source (see above), nocodazole was removed by resuspending in sterile water, and cells were then resuspended in fresh YPD and shaken at 30°. Aliquots of 250 μl were fixed in 70% ethanol at 30-min intervals, spun down, resuspended in PBS (120 mm sodium chloride; 2.7 mm potassium chloride; pH 7.3 with 10 mm phosphate buffer), and incubated with DAPI (1 μg/ml final concentration) (Williamson and Fennell 1979) at room temperature for 20 min. Percentages of uninucleate and binucleate cells were assessed by fluorescence microscopy.

RESULTS

IR survival of mutants separately blocked in histone H3 K4, K36, and K79 methylation:

In addition to histone H3 K79, two other histone H3 lysine residues, K4 and K36, are known to be methylated in both Saccharomyces and higher eukaryotes (Roguev et al. 2001; Strahl et al. 2002; Lee and Skalnik 2005). We studied mutants blocked in each of these methylations to determine if they too played a role in IR resistance, as is the case for H3 K79 methylation. H3 K4 methylation resembles that of H3 K79 in being dependent on prior ubiquitination of histone H2B K123 for di- and trimethylation of the lysine residue (Dover et al. 2002; Sun and Allis 2002; Shahbazian et al. 2005). Methylation of H3 K4 is carried out by the SET1 protein complex, also known as COMPASS, which is thought to include at least eight component proteins (Miller et al. 2001; Roguev et al. 2001; Krogan et al. 2002; Schneider et al. 2005). Information is available concerning the IR sensitivity ranking for homozygous diploid deletion mutants involving six of the COMPASS-encoding genes relative to the rest of the mutants in a pooled deletion library after a single dose (200 Gy) of IR (Brown et al. 2006). These mutants are deleted for BRE2, SDC1, SHG1, SPP1, SWD1, and SWD3, respectively. This assay involves microarray hybridization to assess the relative prevalence of molecular markers for each mutant relative to the whole pool (Birrell et al. 2001; Game et al. 2003; Brown et al. 2006). While the assay is less rigorous than survival curves, none of the six COMPASS-component mutants tested in this way came within the top 20% of mutants ranked in the pool for IR sensitivity (Brown et al. 2006), collectively arguing strongly against a significant role for the COMPASS complex in ensuring diploid survival after IR. Additional observations based on qualitative assays of replica plates with patches of haploid cultures of the same mutants also showed no evidence of sensitivity. To confirm lack of sensitivity, we assayed survival vs. dose for one of these mutants, the MATα haploid deleted for the SWD1 gene, and found sensitivity equivalent to that of wild type (Figure 1).

Figure 1.—

Survival vs. X-ray dose for haploid mutant strains separately affected in methylation of three sites on histone H3. Two set2Δ strains, blocked in H3 K36 methylation, are compared to swd1Δ and dot1Δ strains, blocked in H3 K4 and H3 K79 methylation, respectively. A wild-type and a rad51Δ haploid strain are included for comparison. The strains share the same genetic background, and the dot1Δ mutant shows X-ray sensitivity equivalent to that of previously published dot1Δ strains in this background (Game et al. 2005).

To test for a role for H3 K36 methylation in IR resistance, we studied the set2Δ mutant, since Set2p is responsible for methylating this residue (Strahl et al. 2002). We observed that the set2Δ MATα haploid strain showed mild X-ray sensitivity in survival curves, which was reproducible in set2Δ spore clones derived from a backcross of this strain to wild type (Figure 1). A clear segregation for sensitivity in this cross was difficult to observe on replica plates, although segregation for a borderline X-ray-sensitive phenotype was apparent after 1540 Gy of X-rays. A homozygous set2Δ/set2Δ diploid strain constructed from our spore segregants also showed a small increase in sensitivity compared to a wild-type diploid (not shown). In addition, the set2 diploid from the deletion library pool showed a ranking of 105 for relative growth after IR treatment (Brown et al. 2006), consistent with mild sensitivity. We conclude that methylation of histone H3 K36 plays at least a minor role in resistance to radiation.

Mutants defective in histone H2 K123 ubiquitination are X-ray sensitive:

We reported earlier (Game et al. 2005) that either yeast strains deleted for the DOT1 gene, whose product methylates the histone H3 K79 residue (Ng et al. 2002b), or yeast strains in which the H3 K79 residue is altered to another amino acid, showed sharply increased X-ray sensitivity compared to wild type. At the same time, work from other laboratories showed that methylation of histone H3 K79 as well as H3 K4 is dependent on prior ubiquitination of histone H2B at residue K123 (Briggs et al. 2002; Ng et al. 2002a). The H2B K123 ubiquitination reaction has been shown to result from the combined action of the RAD6-encoded ubiquitin conjugase and the BRE1-encoded ubiquitin ligase (Robzyk et al. 2000; Hwang et al. 2003). IR sensitivity in rad6 mutants was first reported in 1968 (Cox and Parry 1968) and is well known (Game and Mortimer 1974; Lawrence 1994) but has been thought previously to result from the interaction of Rad6p with Rad18p and their joint role in the ubiquitination of proliferating cell nuclear antigen (PCNA) (Hoege et al. 2002). The dot1Δ mutant's IR sensitivity implied that the role of RAD6 in H2B K123 ubiquitination might also contribute to the sensitivity conferred by rad6Δ, and we anticipated that bre1Δ itself should confer X-ray sensitivity comparable to that conferred by dot1Δ but less than that conferred by rad6Δ. In addition, deletion mutations involving the LGE1, RTF1, and PAF1 genes have also each been reported to abolish H2B K123 ubiquitination (Hwang et al. 2003; Ng et al. 2003a; Wood et al. 2003b) and hence might also be expected to confer IR sensitivity. While Lge1p directly interacts with Bre1p, both Rtf1p and Paf1p are members of the separate Paf1/RNA polymerase II complex and may have additional effects on other histone modifications (Mueller and Jaehning 2002; Krogan et al. 2003). We therefore tested X-ray sensitivity in bre1Δ, lge1Δ, rtf1Δ, and paf1Δ deletion strains.

Initial plate tests indicated that MATα haploid strains carrying any of bre1Δ, lge1Δ, or rtf1Δ showed clearly increased X-ray sensitivity compared to wild type, whereas the paf1Δ mutant showed at most marginal sensitivity. The bre1Δ, lge1Δ, and rtf1Δ diploids from the deletion library pool show rankings of 36, 51, and 32, respectively, for relative growth after IR treatment (Brown et al. 2006), consistent with sensitivity and in the same range as the dot1Δ mutant (rank 42). The paf1Δ mutant, which has a slow-growth phenotype (Shi et al. 1996), is not ranked in the pool assay. As done with other mutants from the library (Game et al. 2005), we backcrossed each of the mutant MATα haploid strains to a wild-type strain (g1201-4C) carrying the same genetic background as the deletion library to confirm haploidy and to test whether the X-ray-sensitive phenotype cosegregates with the geneticin-resistance phenotype marking the known deletion mutation. Results are shown in Table 1, where it can be seen that a cosegregation for both phenotypes occurs in crosses of bre1Δ, lge1Δ, and rtf1Δ, confirming that the deletion itself is responsible for conferring the X-ray sensitivity. For paf1Δ, no segregation for X-ray sensitivity was apparent, but a convincing cosegregation was observed for the geneticin-resistance marker and a slow-growth phenotype conferred by the original mutant (see Shi et al. 1996).

View this table:
TABLE 1

Meiotic spore viability and cosegregation data for four deletion-mutant heterozygous diploids

Next, to quantify sensitivity, we performed X-ray survival curves for at least two haploid strains carrying each of the deletion mutations, using the original strains from the deletion library and one or more spore clones derived from the crosses to wild type. Results are shown in Figures 2 and 3, where an additional dot1Δ mutant survival curve is shown for comparison. It can be seen that the bre1Δ, lge1Δ, and rtf1Δ mutations confer sensitivity comparable to that seen in dot1Δ strains, consistent with a repair defect that in each case arises from abolition of the Dot1p-mediated histone H3 K79 methylation. In the case of BRE1, we further confirmed that the deletion itself conferred the IR sensitivity by transforming a bre1Δ strain with a plasmid containing the BRE1 gene and finding that this restored wild-type resistance (Figure 2). Surprisingly, however, the paf1 deletion mutant shows a wild-type response to IR, in contrast to the other three mutants (Figure 3). We considered that the lack of sensitivity of paf1Δ might arise from a secondary mutation in the mutant strain acting as a suppressor or modifier, but rejected this as unlikely when we found a uniform lack of sensitivity in paf1Δ spore clones segregating from a cross with wild type, as judged from irradiated replica plates. A survival curve of one of these paf1Δ spore clones shown with the original mutant in Figure 3 resembles wild type, and a curve from another spore clone (not shown) was equivalent. We note that Paf1p has multiple functions in addition to facilitating H2 K123 ubiquitination (Chang et al. 1999; Krogan et al. 2003; Mueller et al. 2004; Sheldon et al. 2005), and it seems possible that the slow-growth phenotype of paf1Δ counteracts the expected IR sensitivity, as discussed later.

Figure 2.—

Survival vs. X-ray dose for two bre1Δ and two lge1Δ haploid deletion strains. Wild-type, rad51Δ, and dot1Δ haploids are included for comparison. In addition, a curve for the bre1Δ strain g1329-26A transformed with a plasmid containing the BRE1 gene and its native promoter (pJB200 from James Brown at Stanford University) is shown.

Figure 3.—

Survival vs. X-ray dose for two rtf1Δ and two paf1Δ haploid deletion strains. Wild-type, rad51Δ, and dot1Δ haploids are included for comparison.

Mutants defective in histone H2 K123 ubiquitination or H3 K79 methylation interact epistatically for IR sensitivity:

If IR sensitivity in mutants defective in H2B K123 ubiquitination arises from their downstream effects on H3 K79 methylation, then combining bre1Δ and lge1Δ with each other or with dot1Δ should add no additional sensitivity. We constructed strains with each of the double-mutant genotypes involving these three genes, as well as triple-mutant strains. Figures 4–6 show that all three genes interact epistatically. The data are compelling for bre1Δ lge1Δ (Figure 4), bre1Δ dot1Δ (Figure 5), and the bre1Δ dot1Δ lge1Δ triple-mutant strain (Figure 6). For dot1Δ lge1Δ strains, we observed some scatter among strains of equivalent genotype, with two strains showing possibly increased sensitivity and a third falling closer to the single mutants (Figure 6).

Figure 4.—

Survival vs. X-ray dose for two bre1Δ lge1Δ double-mutant haploid deletion strains, shown with representative bre1Δ and lge1Δ single mutants (see Figure 2). Wild-type and rad51Δ haploids are included for comparison.

Figure 5.—

Survival vs. X-ray dose for two bre1Δ dot1Δ double-mutant haploid deletion strains, shown with representative bre1Δ and dot1Δ single mutants (see Figure 2). Wild-type and rad51Δ haploids are included for comparison.

Figure 6.—

Survival vs. X-ray dose for three lge1Δ dot1Δ double-mutant haploid deletion strains and two bre1Δ lge1Δ dot1Δ triple-mutant strains, shown with a representative bre1Δ and lge1Δ single mutant (see Figure 2). Wild-type and rad51Δ haploids are included for comparison.

Both rad6Δ and rad51Δ are epistatic to dot1Δ:

Most previously characterized mutants that show substantial X-ray sensitivity in Saccharomyces either are defective in HRR, mediated by RAD51, RAD52, and related genes, or are defective in one or more aspects of postreplication repair/translesion synthesis that are dependent on the RAD6 and RAD18 genes.

Mutants in the latter group, including rad6Δ and rad18Δ, confer additional sensitivity in double-mutant combinations with rad51Δ (McKee and Lawrence 1980; Game 2000; see Figure 7), supporting the view that these repair processes are essentially separate. However, rad6Δ mutants show substantially greater X-ray sensitivity than rad18Δ mutants (see Figure 8), although rad6Δ and rad18Δ mutants are equally defective in ubiquitination of PCNA, which is a prerequisite for the subsequent steps of postreplication repair/translesion synthesis (Hoege et al. 2002; Stelter and Ulrich 2003; Haracska et al. 2004). This suggests an additional role for RAD6 in mediating IR resistance outside the PCNA ubiquitination pathway. Further support for a separate role for RAD6 in DNA transactions may come from the fact that rad6 mutants are completely defective in meiotic division and fail to commit to meiotic recombination (Game et al. 1980), whereas rad18 mutants show little if any meiotic phenotype (Game and Mortimer 1974; Dowling et al. 1985).

Figure 7.—

Survival vs. X-ray dose for dot1Δ rad6Δ, dot1Δ rad51∷URA3 and two rad6Δ rad51∷URA3 double-mutant haploid deletion strains together with dot1Δ, rad6Δ, and rad51∷URA3 single mutants and a wild-type strain. A rad51Δ BY4742 strain carrying the standard deletion library replacement cassette, which exhibits IR sensitivity equivalent to that conferred by the rad51∷URA3 disruption allele, is also shown.

Figure 8.—

Survival vs. X-ray dose for a bre1Δ rad18Δ double-mutant strain compared to two bre1Δ and two rad18Δ single mutants. A wild-type and a rad6Δ strain are included for comparison. This is the median rad6Δ curve of seven obtained; see text.

Given the X-ray sensitivity of bre1Δ, we anticipated that this additional role for RAD6 could be mediated by its involvement in H3 K79 methylation through its function with BRE1 in H2B K123 ubiquitination, and this in turn could involve the RAD51-dependent HRR pathway. We therefore constructed double mutants involving dot1Δ with rad6Δ and with rad51Δ. Figure 7 shows that dot1Δ confers no additional X-ray sensitivity when combined with either rad6Δ or rad51Δ. However, we and others have observed that rad6Δ single mutants tend to vary in radiation sensitivity and to quickly pick up modifier mutations, especially in the SRS2 gene (Schiestl et al. 1990). To address possible variation here, we performed seven survival assays involving six rad6Δ strains. In comparing double mutants with rad6Δ, we show either the curve with the median IR sensitivity (Figures 8, 9, 12, and 17) or a rad6Δ strain from the same cross as the double mutant to which we compare it (Figure 7). Also, we show both the most sensitive and the least sensitive of the six strains in Figure 10. The dot1Δ rad6Δ double mutant (g1238-2B, Figure 7) has a sensitivity equivalent to the most related rad6Δ single mutant (g1238-7B, Figure 7) and very similar to that of the median rad6Δ strain (MW5094-8A, shown on the same scale in Figure 17). Hence, we conclude that DOT1 mediates a pathway of radiation resistance that requires the RAD6 gene but also facilitates HRR, thus demonstrating a role for RAD6 in enabling effective HRR.

Figure 9.—

Survival vs. X-ray dose for two dot1Δ rad18Δ and one lge1Δ rad18Δ double-mutant strain compared to dot1Δ, lge1Δ, and rad18Δ single mutants. A wild-type and a rad6Δ strain are included for comparison.

Figure 10.—

Survival vs. X-ray dose for two dot1Δ lge1Δ rad18Δ and one dot1Δ bre1Δ rad18Δ triple-mutant strain and a dot1Δ bre1Δ lge1Δ rad18Δ quadruple mutant. These strains are compared with wild type, the four component single mutants, and two rad6Δ strains representing the most and least sensitive full curves of seven rad6Δ curves obtained; see text.

The bre1Δ, lge1Δ, and dot1Δ mutations add extra IR sensitivity when combined with the rad18Δ mutation:

Given that the HRR mutation rad51Δ is epistatic to dot1Δ, we expected that the latter mutation would confer increased sensitivity in double-mutant combinations with rad18Δ, since RAD18 is known to act in postreplication repair (PRR) and itself interacts additively with mutants in HRR (McKee and Lawrence 1980; Game 2000). Figures 8 and 9 show that there is a strong, rather similar increase in sensitivity in each of the double mutants that we constructed involving rad18Δ with bre1Δ, lge1Δ, or dot1Δ compared to the component single mutants. This both confirms that histone H3 K79 methylation is not involved in PRR and supports the functional separation of PRR from the HRR pathway. Figure 10 shows that the dot1Δ bre1Δ rad18Δ and dot1Δ lge1Δ rad18Δ triple mutants as well as a quadruple mutant involving bre1Δ, lge1Δ, and dot1 with rad18Δ fall within the range of these doubles, further confirming epistasis of bre1Δ, lge1Δ, and dot1Δ. It is noteworthy that these strains, and specifically the bre1Δ rad18Δ double mutant (Figure 8), which is defective in ubiquitination of two separate repair-involved targets of the RAD6 ubiquitin ligase, resemble the rad6Δ single mutant in sensitivity. The median curve in Figure 8 as well as the rad6Δ curves in Figure 10 confirm that the additional sensitivity of the rad6Δ single mutant compared to the rad18Δ mutant can be accounted for by the role of RAD6 in the BRE1-mediated histone ubiquitination step. However, as noted below, we also tested the N-end rule protein ubiquitination activity of RAD6 for a possible effect on IR resistance by studying ubr1Δ mutant strains.

A role in IR repair for the RAD6-dependent UBR1 ubiquitin ligase:

The UBR1 gene encodes the ubiquitin ligase that interacts with Rad6p in its major role in poly-ubiquitinating proteins targeted for degradation according to the N-end rule (Dohmen et al. 1991). This pathway is not specific to DNA repair, but ubr1Δ mutants have been found to affect chromosome stability, probably through an indirect effect on sister-chromatid cohesion by affecting the degradation pathway for cohesin (Rao et al. 2001). The ubr1Δ diploid from the deletion library pool showed a ranking of 38 for relative growth after IR treatment (Brown et al. 2006), consistent with IR sensitivity, and we found a mildly increased sensitivity in ubr1Δ haploid survival curves, as shown in Figure 11. A mild sensitivity on plate tests appeared to cosegregate with the ubr1Δ allele in crosses (not shown). To determine if the sensitivity is manifested through an effect on the RAD18 or BRE1 pathways or perhaps neither of these, we made double and triple mutants involving ubr1Δ, rad18Δ, and bre1Δ. We found little or no increased sensitivity in bre1Δ ubr1Δ double mutants (Figure 11), but a significant increase in rad18Δ ubr1Δ doubles (Figure 12). This enhancement of rad18Δ sensitivity is consistent with a role for UBR1 in HRR, as might be expected from the reported effects of ubr1Δ on chromosome stability and cohesin degradation (Rao et al. 2001). Given the mild sensitivity of ubr1Δ, it is less compelling that BRE1 is really epistatic to UBR1. However, the rad18Δ bre1Δ ubr1Δ triple mutants shown in Figure 12 resemble the rad18Δ bre1Δ double mutant as well as the rad6Δ single mutant. We expect this triple-mutant genotype to mimic rad6Δ since it should lack all three known ubiquitination activities that RAD6 mediates, but a potential contribution to IR sensitivity from ubr1Δ in the triple mutant might be difficult to discern in the context of the high sensitivity of the rad18Δ bre1Δ double mutant, which already resembles rad6Δ (see above).

Figure 11.—

Survival vs. X-ray dose for two ubr1Δ haploid deletion strains and two bre1Δ ubr1Δ double mutants, compared with two ubr1Δ and two bre1Δ single mutants. A wild-type and a rad51Δ haploid strain are also shown.

Figure 12.—

Survival vs. X-ray dose for two haploid ubr1Δ rad18Δ double mutants and two bre1Δ ubr1Δ rad18Δ triple-mutant strains, compared with two rad18Δ and two ubr1Δ single mutants. A wild-type strain, a bre1Δ rad18Δ double mutant, and a rad6Δ strain (see also Figure 8) are also shown.

The rad5Δ mutation is additive for IR sensitivity with bre1Δ and with rad18Δ:

The Rad5 protein acts downstream from Rad18p in the ubiquitination steps of PCNA and thereby plays a major role in postreplication repair (Hoege et al. 2002; Torres-Ramos et al. 2002). However, while rad5Δ and rad18Δ interact epistatically with respect to UV sensitivity (Johnson et al. 1992; this study; data not shown), we observed an additive response for IR sensitivity (Figure 14), in agreement with other reports (Friedl et al. 2001; Chen et al. 2005). In addition, Chen et al. (2005) presented data showing that RAD5 has another function that contributes to IR resistance independently of its PCNA-modifying role and is probably related to an MRE11/RAD50/XRS2-mediated repair activity. To study RAD5 in relation to the BRE1/DOT1 pathway, we constructed rad5Δ bre1Δ and rad5Δ dot1Δ double mutants. Figures 13 and 14 show that the rad5Δ mutation adds sensitivity to bre1Δ and dot1Δ as well as to rad18Δ. When taken with data for rad18Δ combined with bre1Δ and dot1Δ (Figures 8 and 9), this implies that RAD5, RAD18, and BRE1/DOT1 mediate three at least partly independent IR resistance mechanisms. Surprisingly, the dot1Δ rad5Δ rad18Δ triple shows only slight sensitivity beyond each of the component double mutants (Figure 14). It is difficult to assess the significance of this, but it seems less sensitive than would be expected from double-mutant data. Roles for Rad5p in more than one IR repair pathway might account for this, as discussed later.

Figure 13.—

Survival vs. X-ray dose for two haploid rad5Δ mutant strains, a bre1Δ rad5Δ double mutant and a dot1Δ rad5Δ double mutant. A bre1Δ and a dot1Δ mutant are shown for comparison, together with wild type and a rad51Δ strain.

Figure 14.—

Survival vs. X-ray dose for a rad5Δ rad18Δ double mutant and a dot1Δ rad5Δ rad18Δ triple mutant, compared to the component single mutants and a dot1Δ rad18Δ and dot1Δ rad5Δ double mutant (see also Figures 9 and 13). A wild type and a rad51Δ curve are also shown.

IR epistasis and colony-size effects of rad50Δ with bre1Δ and dot1Δ:

We constructed double and triple mutants involving bre1Δ, dot1Δ, and rad50Δ to test whether the Mre11/Rad50/Xrs2 complex (MRX) is involved in repair affected by histone H3 K79 methylation. Figure 15 shows that combining bre1Δ, dot1Δ, or both mutants with rad50Δ adds no further sensitivity to rad50Δ alone, as might be expected from the role of MRX in HRR (Bressan et al. 1999; Game 2000) and our double-mutant data with rad51Δ. As discussed later, there is no support for a separate IR damage repair role involving nonhomologous end-joining (NHEJ) from these data, since the strains in Figure 15 have survival curves equivalent to those of the rad51Δ mutant included for comparison. However, we did observe a strong effect of the rad50Δ bre1Δ double mutant and the rad50Δ bre1Δ dot1Δ triple-mutant genotypes on the colony size of meiotic spore clones, which was sharply reduced compared to that of other spore clones in the same cross. This implies a slow-growth phenotype presumably caused by interaction of rad50Δ with bre1Δ and confirms similar findings from large-scale random spore analysis (Tong et al. 2004). Since this phenotype was absent in our rad50Δ dot1Δ double-mutant spore clones, it is presumably not mediated by abrogation of H3 K79 methylation. However, synthetic lethality has been reported (Tong et al. 2004) between rad50Δ and two mutants for genes in the COMPASS complex, swd3Δ and bre2Δ, responsible for methylating the histone H3 K4 residue (Krogan et al. 2002). Since di- and trimethylation of this residue depends on histone H2 K123 ubiquitination, it is plausible that the slow-growth phenotype of rad50Δ bre1Δ double mutants also arises from the impact of bre1Δ on H3 K4 methylation.

Figure 15.—

Survival vs. X-ray dose for two rad50Δ mutants and a dot1Δ rad50Δ and a bre1Δ rad50Δ double mutant. A rad51Δ mutant, a wild type, and the dot1Δ and bre1Δ single mutants are included for comparison.

A role for the BRE1/DOT1 pathway in IR-damage-induced checkpoint control:

While this work was in progress, several reports suggested that histone H2B K123 ubiquitination and histone H3 K79 methylation are important for checkpoint arrest after DNA damage. It was recently shown that the 53BP1 checkpoint protein in mammalian cells recognizes and binds to methylated histone H3 K79 residues and that the methylation is important for attracting 53BP1 to double-strand-break (DSB) sites (Huyen et al. 2004). Saccharomyces Rad9 protein, which has a central role in establishing checkpoint delays after irradiation (Weinert and Hartwell 1988, 1989; Siede et al. 1993), shares homologous domains with 53BP1, including a recently described major domain very similar to the Tudor domain in 53BP1 that interacts with methylated mammalian H3 K79 (Alpha-Bazin et al. 2005). Hence, the mammalian 53BP1 findings are strongly suggestive of a role for H3 K79 methylation in RAD9-mediated checkpoints in yeast. In addition, others have shown directly that rad6Δ, bre1Δ, and dot1Δ mutants reduced or abolished the checkpoint delay seen in wild type after UV- and chemical DNA-damaging treatments in G1 and intra-S phase cells without affecting the G2 checkpoint (Giannattasio et al. 2005). These authors also showed that phosphorylation of Rad9 protein was reduced or abolished in these mutants after similar DNA-damaging treatments, leading in turn to defective activation of Rad53 checkpoint protein. Most recently, Wysocki et al. (2005) identified dot1Δ in a screen for mutants that abrogate the G1 damage checkpoint. Surprisingly, however, using more qualitative tests, these authors found little evidence of IR sensitivity in the dot1Δ mutant. Genetic differences in the strain backgrounds used probably account for these different results (see discussion). To investigate whether the substantial IR sensitivity of bre1Δ and dot1Δ mutants in our strains is conferred through an effect on checkpoint controls, and to extend the epistasis analysis, we studied double mutants involving rad9Δ as well as tested directly for abrogation of checkpoints in bre1Δ and dot1Δ mutants using IR.

Double-mutant analysis with rad9Δ:

Since RAD9 has a well-established role in DNA-damage-induced checkpoint controls (Weinert and Hartwell 1988, 1989; Siede et al. 1993), we anticipated that if bre1Δ confers sensitivity through abrogating one or more RAD9-dependent checkpoints, epistasis between rad9Δ and bre1Δ would be observed, whereas an additive response would suggest a different mechanism. Mutations in the RAD9 gene are known to be IR sensitive (Cox and Parry 1968; Game and Mortimer 1974), but surprisingly, we could find little published information about the epistasis relationships of rad9 in combination with mutants in other RAD genes. An exception is the rad9Δ rad6Δ combination, which has been reported to show additive sensitivity compared to the single mutants (Schiestl et al. 1989). It has also been shown that activation of some Saccharomyces HRR proteins after radiation is dependent on an intact RAD9 gene (Bashkirov et al. 2000) and that the ATM-mediated checkpoint is required for normal function of the RAD54 HRR gene in chicken DT40 cells (Morrison et al. 2000). We tested the double-mutant rad9Δ bre1Δ and the triple-mutant rad9Δ bre1Δ dot1Δ in the deletion library background, where each of the single mutants is sensitive, to determine whether RAD9 affects the same pathway as BRE1 and DOT1, as might be expected on the basis of results with mammalian cells (Huyen et al. 2004).

We found that the rad9Δ single mutant is significantly more sensitive than the bre1Δ or dot1Δ strains, but that double or triple mutants involving rad9Δ with either or both of these two are no more X-ray sensitive than rad9Δ alone (Figure 16). This agrees with findings by Wysocki et al. (2005), who observed qualitatively that dot1Δ did not potentiate the sensitivity of rad9Δ in a different genetic background where there was little or no IR sensitivity of the dot1Δ single mutant. The epistasis implies either that the X-ray resistance mediated by BRE1 and DOT1 is dependent on a RAD9-mediated checkpoint or that the RAD9 checkpoint itself is partially dependent on an intact BRE1/DOT1-mediated H3 K79 methylation pathway. In the latter case, which is consistent with data from mammalian cells (Huyen et al. 2004), RAD9 function is presumably only partly dependent on H3 K79 methylation, since substantial RAD9-dependent resistance remains in bre1Δ and dot1Δ single mutants (Figure 16).

Figure 16.—

Survival vs. X-ray dose for two dot1Δ rad9Δ double-mutant strains and a dot1Δ bre1Δ rad9Δ triple mutant, compared to two rad9Δ single mutants. Wild type and the bre1Δ and dot1Δ single mutants are included for comparison.

On the basis of our observations that rad51null and rad9Δ are each epistatic to bre1Δ and dot1Δ, we expected that RAD9 itself would fall into the RAD51 epistasis group with respect to X-ray sensitivity. This would also be consistent with reported additivity between rad9Δ and rad6Δ (Schiestl et al. 1989) since known mutants in HRR such as rad51null show increased sensitivity in combination with rad6Δ (Figure 17; see also McKee and Lawrence 1980; Game 2000). Since we find no additivity between bre1Δ and rad9Δ, increased sensitivity contributed by rad6Δ in the rad6Δ rad9Δ double mutant seems likely to arise from the RAD18-dependent aspect of RAD6 repair. To test this, we studied double mutants involving rad9Δ with rad18Δ, rad51null, and rad5Δ as well as retesting the rad6Δ rad9Δ combination.

Figure 17.—

Survival vs. X-ray dose for two rad9Δ rad51∷URA3 double mutants and one rad9Δ rad6Δ double-mutant and two rad9Δ rad6Δ rad51∷URA3 triple-mutant strains, compared with a rad6Δ rad51∷URA3 double mutant (see also Figure 5), a wild type, and the three component single mutants.

We found a sharp increase in sensitivity in rad9Δ rad18Δ (Figure 18), a slightly lesser increase in rad9Δ rad5Δ (Figure 18), and, at most, only a slight increase in the rad9Δ rad51null strains (Figure 17), compared to the single mutants in each case. Hence RAD18-mediated repair seems to be largely independent of the RAD9-mediated checkpoint, whereas RAD51-mediated HRR is heavily dependent on RAD9 function. As expected from these observations as well as previous work (Schiestl et al. 1989), we found that the rad6Δ rad9Δ double mutant also shows strongly increased IR sensitivity compared to the single mutants. In fact, this double mutant is equivalent to rad6Δ rad51null double mutants (Figure 17), again suggesting that RAD9 is required for most or all RAD51-mediated IR recovery. The equivalent sensitivity of the rad9Δ rad18Δ double mutant to rad9Δ rad6Δ and rad6Δ rad51null, despite the lower sensitivity of rad18Δ compared to rad6Δ, also implies that RAD9 is required for the part of RAD6-mediated resistance that is independent of RAD18 and that this, in turn, is dependent on RAD51. The rad9Δ rad6Δ rad51Δ triple mutants in Figure 17 are possibly slightly more sensitive than the rad6Δ rad51Δ double mutants; hence a minor additional role for RAD9 outside either PRR or HRR cannot be excluded. Interestingly, in double-mutant combinations both rad9Δ and rad18Δ also show increased sensitivity with rad5Δ, but the rad5Δ rad9Δ rad18Δ triple mutants in Figure 18 are no more sensitive than the rad9Δ rad18Δ double mutants, perhaps implying that RAD5 acts in more than one pathway, as discussed later.

Figure 18.—

Survival vs. X-ray dose for two rad9Δ rad5Δ and one rad9Δ rad18Δ double mutant and two rad9Δ rad5Δ rad18Δ triple mutants. Each single mutant and wild type are also shown for comparison.

The dot1Δ and bre1Δ mutants are defective in the G1 but not the G2 IR-induced cell cycle checkpoint:

Wild-type yeast cells irradiated in the G2 phase of the cell cycle become arrested before proceeding through cell division. This arrest is dependent on the RAD9 gene and is important for subsequent cell survival: rad9 mutants substantially fail to arrest in G2 after irradiation, and this is thought to be part of the reason for their increased killing by IR compared to wild type (Weinert and Hartwell 1988). Using the rad9Δ mutant and wild type as controls, we tested haploid dot1Δ and bre1Δ mutants for an effect on the IR-induced G2 checkpoint by monitoring their ability to progress into mitosis following a nocodazole-induced accumulation in G2. Figure 19 shows that, as expected, without irradiation, all four strains promptly enter nuclear division when released from nocodazole. After 500 Gy of 137Cs gamma irradiation, however, wild-type, dot1Δ, and bre1Δ cells remain arrested, with little sign of division up to 90 min following irradiation. In contrast, rad9 shows significant escape from the G2 checkpoint, although at this dose it too shows delayed division compared to the unirradiated control. Results are similar after 1000 Gy of radiation (not shown), and together these data indicate that dot1Δ and bre1Δ mutations do not significantly abrogate the IR-induced G2 cell cycle checkpoint.

Figure 19.—

Effect on the IR-induced G2/M checkpoint of the bre1Δ, dot1Δ, and rad9Δ mutations compared to wild type. Cells were released from nocodazole synchronization into fresh medium at time zero after 500 Gy of 137Cs gamma irradiation, and in parallel without irradiation. The percentage of cells that have undergone nuclear division is shown for each culture at 30-min intervals thereafter. Strains used were g1295-21C bre1Δ, g1295-6C dot1Δ, g1304-19C rad9Δ, and BY4741 wild-type MATa.

To assess radiation-induced G1 arrest, we followed the cell cycle progression of dot1Δ and bre1Δ mutants released from α-factor-induced synchrony with and without 500 Gy of 137Cs gamma irradiation and, as reported by others (Giannattasio et al. 2005; Wysocki et al. 2005), observed a significant effect of both mutants in abrogating the arrest response seen in wild type (see Figure 20). In fact, both mutants are essentially equivalent in this phenotype to the rad9 strain. We used rad9 as a positive control because previous work (Siede et al. 1993) has shown that RAD9 is involved in the G1 checkpoint as well as in the G2 checkpoint. The three strains differ from wild type, where IR-induced arrest can clearly be seen. Despite differences in IR sensitivity, our findings of a G1 but not a G2 checkpoint defect in dot1Δ or bre1Δ mutants in the deletion library background are thus consistent with recent findings by others in the W303 background (Giannattasio et al. 2005; Wysocki et al. 2005).

Figure 20.—

Effect on the IR-induced G1/S checkpoint of the bre1Δ, dot1Δ, and rad9Δ mutations compared to wild type. Cells were released from α-factor synchronization into fresh medium at time zero after 500 Gy of 137Cs gamma irradiation, and in parallel without irradiation, and fixed and analyzed for DNA content with flow cytometry at 15-min intervals thereafter. Strains used for each genotype were as in Figure 19.

DISCUSSION

We have found that Saccharomyces mutants that are unable to ubiquitinate the histone H2B lysine 123 residue are substantially sensitive to ionizing radiation. It has been shown elsewhere that ubiquitination of H2B K123 is required for completion of the subsequent methylation of histone H3 at its lysine 4 and lysine 79 (but not lysine 36) residues (Briggs et al. 2002; Dover et al. 2002; Ng et al. 2002a; Sun and Allis 2002; Shahbazian et al. 2005). We reported earlier (Game et al. 2005) that abolishing methylation at H3 K79 also confers IR sensitivity, and it seems clear from data reported here and elsewhere that the IR sensitivity of mutants that abolish H2B K123 ubiquitination arises from this downstream effect on histone H3 K79 methylation. This is supported by the finding that the bre1Δ dot1Δ double mutant has the same sensitivity as the bre1Δ and dot1Δ single mutants (Figure 5). Moreover, the equivalent sensitivity of the bre1Δ and dot1Δ single mutants implies that all the sensitivity of mutants affected in H2B K123 ubiquitination can be accounted for in our strains by their secondary effects on H3 K79 methylation. This is consistent with the absence of the sensitivity of COMPASS mutants, which impact the other H3 methylation (K4) that is dependent on H2B K123 ubiquitination, and with separate IR sensitivity of set2Δ mutants, which impact H3 K36 methylation independently of H2B K123 ubiquitination. Since deleting BRE1 impacts di- and trimethylation rather than mono-methylation of H3 K79 (Shahbazian et al. 2005), the IR sensitivity of the bre1Δ mutant indicates that wild-type IR resistance depends on di- or trimethylation rather than on mono-methylation of H3 K79 by Dot1p.

Surprisingly, the paf1Δ mutant fails to show increased IR sensitivity although it too has been reported to block H2B K123 ubiquitination (Wood et al. 2003b) as well as H3 K79 methylation (Krogan et al. 2003). If checkpoint defects are responsible for the sensitivity of the other mutants that abolish H2B K123 ubiquitination, then possibly the more severe slow-growth phenotype of paf1Δ strains compared to related mutants (Mueller and Jaehning 2002; this study) relieves their IR sensitivity by providing adequate time for repair even in the absence of normal checkpoints. In the case of rad9 mutants, delaying the cell cycle artificially alleviates their IR sensitivity (Weinert and Hartwell 1988), and perhaps an analogous effect occurs spontaneously in paf1Δ strains. Alternatively, one of the other paf1Δ phenotypes may impact the need for H2B K123 ubiquitination or H3 K79 methylation in IR resistance in an unknown way. Although both Paf1p and Rtf1p are members of the PAF1/RNA polymerase II complex, their mutants differ in several aspects of their phenotypes, and surprisingly, it has been reported that knocking out RTF1 in a paf1Δ background substantially reverses the slow-growth phenotype (Mueller and Jaehning 2002). Work with double mutants involving paf1Δ may clarify the reason for its IR resistance.

A strong increase in IR sensitivity is seen when bre1Δ, dot1Δ, or lge1Δ are combined with rad18Δ, but no such increase is seen in double mutants that we tested with rad50Δ or rad51Δ. This represents convincing evidence that these histone modification genes are not required for any of the several aspects of postreplication repair/translesion synthesis that are dependent on RAD18-mediated ubiquitination of PCNA. Rather, they are required for effective homologous recombination repair as mediated by the RAD51 pathway, although clearly they are needed only for a subset of such repair, since their mutant strains are consistently less sensitive than rad51Δ or rad50Δ strains (Figures 2 and 15) and similar mutants (Game 2000).

There is also strong evidence from this study and others that bre1Δ and dot1Δ mutations abrogate the radiation-induced checkpoint control in G1 cells but leave the G2 checkpoint largely unaffected. As a cause of IR sensitivity, this seems paradoxical, given that HRR does not occur in haploid G1 cells. Also, our data suggest that in log-phase cultures the histone modification mutants, like others involved in HRR, are less affected in the initial slope of the curves representing mainly cells in G1 than they are in what is the tailed part of the curve in wild type, which represents mainly G2 cells (see Figures 2, 3, and 14; see also Game 1983 for review). However, an effect on the intra-S checkpoint, as has been reported elsewhere (Giannattasio et al. 2005; Wysocki et al. 2005), may contribute to the lethality that we observed, since HRR can occur not only in G2 but also in S-phase. Alternatively, an additional mutant effect may cause lethality through impacting HRR directly without being strictly dependent on checkpoint controls. The epistasis between rad9Δ and dot1Δ indicates that all the IR resistance conferred by Dot1p depends on Rad9p, but on the basis of our data and that of others (Bashkirov et al. 2000; Morrison et al. 2000), RAD9 itself may perhaps be considered a player in HRR as well as a checkpoint gene, since rad9Δ strains in our hands are as sensitive as mutants such as rad51Δ that effectively abolish HRR. In mammalian cells, the product of the 53BP1 gene (an ortholog of Saccharomyces RAD9) associates through its Tudor domain with methylated H3 K79, and it has been proposed that when DSBs occur, the preplaced methyl groups on this residue in the core of H3 become exposed near the DSBs, serving as a signal to bring Rad9p to the site (Huyen et al. 2004). Clearly, Rad9p function is not exclusively dependent on this signal, since much Rad9p-dependent resistance remains in dot1Δ strains. Present results suggest that alternative mechanisms may exist at different stages of the cell cycle, but it is also possible that multiple DSB-signaling pathways occur in parallel or that only certain subsets of DSBs are dependent on H3 K79 trimethylation to bring about interaction with Rad9p. The Dot1p methylase is important in differentiating euchromatin from heterochromatin (Ng et al. 2003b), and it seems possible that there are positional impacts on DSB repair that depend on this effect. Currently, rapid progress is being made in understanding the early steps in chromatin changes that occur at DSB sites and lead to the formation of foci containing phosphorylated histone H2A (in Saccharomyces) or H2A.X (in mammals) (Tsukuda et al. 2005; van Attikum and Gasser 2005; Nussenzweig and Paull 2006). However, the spatial, temporal, and functional relationships among these foci and H3 K79 trimethylation in determining IR resistance still remain unclear.

This study and recent related work (Dover et al. 2002; Ng et al. 2002a; Giannattasio et al. 2005; Wysocki et al. 2005) have clarified the role of RAD6 in radiation resistance. The high IR and UV sensitivity of rad6 mutants has usually been attributed to the role of Rad6p in at least three forms of postreplication repair or translesion synthesis mediated by ubiquitination of PCNA through an interaction with Rad18p (Xiao et al. 2000; Broomfield et al. 2001; Hoege et al. 2002). However, this fails to account for the substantially greater IR sensitivity of rad6Δ strains compared to rad18Δ strains. The IR sensitivity of bre1Δ suggests that this extra sensitivity arises from the role of Rad6p in H2B K123 ubiquitination. This is convincingly confirmed by double-mutant analysis, since rad6Δ is epistatic to both bre1Δ and rad18Δ, while bre1Δ is additive with rad18Δ and the rad18Δ bre1Δ double mutant mimics the sensitivity of rad6Δ alone. The RAD6/BRE1/DOT1 pathway also provides a clear link for Rad6p to HRR, as indicated by epistasis of rad51Δ to dot1Δ. It is not yet clear if this role impacts HRR itself or is mediated entirely through checkpoint controls that may be prerequisites for HRR. Recently, Zhang and Lawrence (2005) have reported that the error-free mode of RAD18-dependent postreplicational repair frequently involves recombination between sister strands, at least in plasmid DNA. However, this RAD18-dependent process may perhaps be better regarded as an aspect of postreplication repair that depends on recombination, rather than part of HRR per se, in contrast to the separate RAD6/BRE1/DOT1 pathway, which is independent of RAD18.

The availability of strains separately mutant for UBR1, BRE1, and RAD18 enables us to determine which of the many phenotypes of rad6Δ mutants arise from each of the pathways that rad6Δ impacts. It will be instructive to determine if the ubr1Δ bre1Δ rad18Δ triple mutant truly mimics all the phenotypes of rad6Δ; if not, this will imply more roles for the Rad6 ubiquitin conjugase. Homozygous rad6 mutant diploids are able to undergo premeiotic DNA synthesis but are completely defective in sporulation, meiotic division, and in commitment to meiotic recombination (Cox and Parry 1968; Game et al. 1980), but little information has been available about the specific nature of the meiotic defect. There is evidence that the H2B ubiquitination function of Rad6p is important for the role of RAD6 in meiosis, since both rad6Δ and bre1Δ, as well as a histone H2B K123 substitution mutant, reduced the frequency of meiotic DSBs, at least in the SK1 genetic background (Yamashita et al. 2004). The set1Δ deletion mutant has been shown to confer meiotic defects broadly similar to those of bre1Δ (Sollier et al. 2004), implying that the meiotic phenotype of H2B ubiquitination mutants may be manifested through their effect on SET1-mediated histone H3 K4 methylation. Studying the dot1Δ mutant in meiosis should reveal whether H3 K79 methylation is similarly involved.

Uncertainty remains about the role of the RAD5 gene in IR resistance despite the recent finding that this role is independent of the Rad5p function in poly-ubiquitinating PCNA and results instead from a separate ATPase activity in the protein (Chen et al. 2005). These authors found a lack of additivity for IR sensitivity between rad5Δ and each of the MRX deletion mutants, but they found additivity between rad5Δ and rad51Δ and rad5Δ and rad52Δ (Chen et al. 2005). From this, one might expect that MRX deletion strains by themselves would be more IR sensitive than rad51Δ strains, but in fact the curves are equivalent (Game 2000; Chen et al. 2005; see Figure 15), implying complexity in pathway interactions. We have confirmed that rad5Δ adds sensitivity to rad51Δ (not shown) and find that it also adds sensitivity to rad9Δ and rad18Δ as well as to bre1Δ and dot1Δ (Figures 13 and 18). However, the rad5Δ rad18Δ dot1Δ triple mutant (Figure 14) is less sensitive than expected from these double-mutant combinations, and rad5Δ fails to add significant sensitivity to rad6Δ (not shown), again presenting complexity in interpretation. Finally, as shown in Figures 14 and 18, each of the three double mutants involving rad5Δ, rad9Δ, and rad18Δ are more sensitive than the component single mutants, yet the rad5Δ rad9Δ rad18Δ triple mutant is no more sensitive than the rad9Δ rad18Δ double mutant. Thus, the additivity between rad5Δ and rad18Δ is abolished in a rad9Δ background. Some of these paradoxical findings can be explained if RAD5 contributes to IR resistance through more than one pathway. If it partially impacts two pathways, then its deletion should cause at least some increased sensitivity when combined with single mutants in either pathway, but no further increase in a triple-mutant combination where both the pathways are already blocked by other mutations.

Our data differ from those of Wysocki et al. (2005) with respect to IR sensitivity of the dot1Δ single mutant. Using qualitative plate tests, these authors found little difference between wild type and dot1Δ at doses up to 900 Gy. The most likely explanation for the different findings is the different strain backgrounds used, which were W303 (Thomas and Rothstein 1989) in the work by Wysocki et al. (2005) vs. the S288C/deletion library background used here. There is good agreement with respect to the mutants' effects on the damage checkpoints themselves; hence it is likely that the two backgrounds differ in the relative influence of checkpoint defects on survival.

Our data do not address the role of the BRE1 and DOT1 in nonhomologous end-joining, since wild-type Saccharomyces repairs little if any X-ray-induced damage by NHEJ. Although Saccharomyces lacks the catalytic subunit of DNA protein kinase that is involved in mammalian NHEJ, it is still able to process some types of DSBs by NHEJ, such as those induced by restriction endonucleases (Boulton and Jackson 1996). However, mutants such as yku70Δ that are defective in NHEJ but not in HRR or PRR confer little or no IR sensitivity when they are present as single mutants, and only mild (Boulton and Jackson 1996; Siede et al. 1996) or no additional IR sensitivity (Milne et al. 1996; J. Game, unpublished observations) when present in double-mutant combinations with HRR mutants. Thus, an NHEJ defect, if present in bre1Δ or dot1Δ strains, would be unlikely to impact IR sensitivity.

We reported earlier that the dot1Δ deletion does not by itself confer any substantial UV sensitivity on our strains, and during this study we found that bre1Δ, rtf1Δ, and paf1Δ also confer no (or minimal) sensitivity. However, preliminary data (not shown) indicate a probable synergistic increase in UV sensitivity in a rad18Δ dot1Δ double mutant compared to either single mutant. HRR mutants as well as excision repair mutants are well known to interact synergistically with mutants in the RAD18 pathway with respect to UV sensitivity (Cox and Game 1974; Game 1983), so this observation is consistent with our other data indicating a role for DOT1 in HRR. Further understanding of the role of the BRE1 and DOT1 genes in UV repair requires studying double-mutant combinations with excision repair genes, since excision repair is the major mechanism of UV resistance in Saccharomyces.

Acknowledgments

We thank James A. Brown at Stanford University and Beth Rockmill at Yale University for generously sharing information, strains, plasmids, and cassettes. This work was supported by National Institutes of Health grant GM5997901 (to J.C.G.) and by National Cancer Institute Public Health Service training grant CA09302 and Department of Defense grant W81WH-04-1-0451 (to J.M.B.).

Footnotes

  • Communicating editor: S. T. Lovett

  • Received March 2, 2006.
  • Accepted June 3, 2006.

References

View Abstract