The biological significance of DNA damage-induced gene expression in conferring resistance to DNA-damaging agents is unclear. We investigated the role of DUN1-mediated, DNA damage-inducible gene expression in conferring radiation resistance in Saccharomyces cerevisiae. The DUN1 gene was assigned to the RAD3 epistasis group by quantitating the radiation sensitivities of dun1, rad52, rad1, rad9, rad18 single and double mutants, and of the dun1 rad9 rad52 triple mutant. The dun1 and rad52 single mutants were similar in terms of UV sensitivities; however, the dun1 rad52 double mutant exhibited a synergistic decrease in UV resistance. Both spontaneous intrachromosomal and heteroallelic gene conversion events between two ade2 alleles were enhanced in dun1 mutants, compared to DUN1 strains, and elevated recombination was dependent on RAD52 but not RAD1 gene function. Spontaneous sister chromatid exchange (SCE), as monitored between truncated his3 fragments, was not enhanced in dun1 mutants, but UV-induced SCE and heteroallelic recombination were enhanced. Ionizing radiation and methyl methanesulfonate (MMS)-induced DNA damage did not exhibit greater recombinogenicity in the dun1 mutant compared to the DUN1 strain. We suggest that one function of DUN1-mediated DNA damage-induced gene expression is to channel the repair of UV damage into a nonrecombinogenic repair pathway.
THE expression of radiation-inducible genes is enhanced when cells are exposed to either UV or ionizing radiation. In prokaryotes, radiation induces the SOS response and triggers the expression of genes directly involved in DNA recombination, DNA repair, and cell cycle arrest (for review, see Friedberget al. 1995). Inducible recombination in prokaryotes functions in postreplication repair of UV-induced DNA lesions (Ruppet al. 1971). The identification in Saccharomyces cerevisiae of both radiation-inducible genes (Ruby and Szostak 1985) and genes that arrest the cell cycle at specific cell cycle checkpoints (Weinertet al. 1994) underscores the importance of understanding whether radiation resistance in eukaryotes is conferred by elevating the expression of specific repair genes.
In S. cerevisiae, ∼1% of the yeast genome is estimated to encode DNA damage-inducible genes (Ruby and Szostak 1985). These genes are involved in DNA metabolism, such as RNR3 (large subunit of ribonucleotide reductase; Elledge and Davis 1990), RNR2 (small subunit of ribonucleotide reductase; Elledge and Davis 1987), CDC9 (DNA ligase; Barkeret al. 1985), and CDC17 (DNA polymerase α; Johnstonet al. 1987), and in the recombinational repair of double-strand breaks, such as RAD54 (Coleet al. 1987) and the yeast recA homolog, RAD51 (Aboussekhraet al. 1992; Basileet al. 1992). Although it would seem logical that radiation-induced gene expression of DNA repair genes would enhance DNA repair, its biological significance in conferring radiation resistance is unclear. First, RAD1 (Friedberget al. 1995) and RAD52 (Coleet al. 1989), genes that participate directly in the repair of UV or ionizing radiation-induced damage, respectively, are not induced after the exposure of radiation in mitotic cells but are constitutively expressed. Second, deletion of the damage response element (DRE) sequences upstream of either RAD2 (Siedeet al. 1989) or RAD54 (Coleet al. 1989) does not confer enhanced radiation sensitivity in growing cultures after exposure to UV or ionizing radiation, respectively. Thus, the basal level of expression of DNA repair genes appears sufficient to maintain wild-type levels of DNA damage resistance in mitotic yeast.
The identification of protein kinase mutants defective in the transcriptional induction of RNR3 and hyper-sensitive to DNA-damaging agents, however, has established a biological function for the DNA damage-inducible response in DNA repair. These protein kinases include the serine-threonine kinases Rad53/Sad1 (Allenet al. 1994) and Dun1 (Zhou and Elledge 1993), the ataxia-telangiectasia mutated (ATM)-like kinase, Mec1 (Weinertet al. 1994), and the casein kinase I isoform Hrr25 (Hoet al. 1997). The Dun1 kinase is hypostatic to Rad53/Sad1 and Mec1 protein kinases in the pathway for DNA damage induction (Allenet al. 1994), but its epistatic relationship with Hrr25 is unknown (Hoet al. 1997). A subset of DNA-damage inducible genes can still be induced in dun1 mutants; for example, induction of RAD51, a gene involved in recombinational repair, is DUN1 independent but RAD9 dependent (Aboussekhraet al. 1996). Further studies have indicated that the pathways for DUN2 (POLϵ)-dependent transcriptional induction and RAD9-mediated cell cycle arrest are independent pathways that respond to DNA damage (Navaset al. 1996).
After observing that rad9 mutants, which fail to arrest the cell cycle at the G2 checkpoint (Weinert and Hartwell 1988), exhibit higher frequencies of chromosomal rearrangements (Fasulloet al. 1998), we investigated both the recombination and radiosensitive phenotypes of dun1 mutants. We demonstrate that the dun1 rad52 and dun1 rad9 mutants exhibit a synergistic increase in UV sensitivity and that dun1 mutants exhibit enhanced frequencies of spontaneous and UV-induced recombination. We suggest that the DUN1 pathway for DNA damage-induced gene expression participates in the RAD3 pathway, and that DNA damage tolerance in dun1 mutants is partially conferred by RAD52-dependent recombinational mechanisms.
MATERIALS AND METHODS
Methods: Yeast extract, peptone, dextrose (YPD), synthetic dextrose (SD), synthetic complete (SC)-ADE, and SC-HIS are described by Sherman et al. (1982). YP(A)D contains YPD supplemented with 80 mg/liter of adenine. Ura– isolates were selected on fluoroorotic acid (FOA) medium (Boekeet al. 1984). Yeast transformations were done according to Chen et al. (1992). Petite colonies were scored on YP plates supplemented with 2% lactic acid (pH 5.5).
Yeast strains and plasmids: Yeast strains are described in Table 1. All yeast strains are derived from W303 (Thomas and Rothstein 1989). The rad52, rad9, rad18, and rad1 mutations contain insertions of TRP1, URA3, LEU2, and LEU2, respectively. The dun1Δ mutant (YA145), kindly provided by S. Elledge, contains dun1-Δ100::HIS3. In the plasmid pZZ66, the dun1-Δ100::HIS3 allele is present on a BamHI restriction fragment (Zhou and Elledge 1993). The dun1-Δ100::his3::URA3 mutation was constructed by inserting the 1.1-kb URA3 fragment in the internal HindIII sites of HIS3. After digestion with BamHI, the dun1-Δ100::his3::URA3 fragment was then introduced into YA145 by one-step gene disruption (Rothstein 1983) to generate YD123. The UV sensitivity of the Ura+ His– transformants after 120 J/m2 exposure was then confirmed. Double mutants were made by crossing the appropriate single mutants, dissecting tetrads of the resulting diploids, and identifying meiotic segregants that had the desired auxotrophic phenotype. The rad9 dun1 rad52 triple mutant (YD106) was made by crossing the rad9 single mutant (YA132) with the dun1 rad52 double mutant (YD103), and dissecting tetrads from the sporulated diploid.
Construction of strains to monitor mitotic intrachromosomal and heteroallelic recombination: YA156 (YKH12a) and YA157 (YKH12α), kindly provided by L. Symington, contain a nontandem duplication of ade2 that was generated upon integration of the pKH9 (URA3) plasmid (Figure 1). One copy of ade2 contains the nonrevertible allele ade2-a and the other copy of ade2 contains the nonrevertible allele ade2-n (Huang and Symington 1994). Haploid strains containing the integrated pKH9, including the dun1 mutant (YD100), the rad52 (YD102), and rad1 (YD108) mutants, and the double dun1 rad52 double mutant (YD110) are meiotic segregants that were obtained after genetic crosses and tetrad dissections. An Ade+ Ura– recombinant of YD100, YD101, was used in genetic crosses for making strains that monitor heteroallelic recombination.
Heteroallelic recombination was monitored in diploid strains containing ade2-a and ade2-n on different homologs. To identify Ura– Ade– haploids that contained either ade2-a or ade2-n, we obtained FOAR isolates of YA156 and YA157. Primers 5′ CGCTATCCTCGGTTCTGCAT 3′ and 5′ TAACGC CGTATCGTGATTAA 3′ were used to amplify the ade2 allele. Digestion of the PCR-amplified product with either AatII or NdeI restriction endonucleases indicated whether the ade2 allele contained one or both of the restriction sites. YD114 and YD115 (Table 1) containing ade2-a and ade2-n alleles, respectively, were mated to generate the diploid YD116; a strain with the identical genotype was made by Bai and Symington (1996). The ade2-a and ade2-n haploids containing dun1, rad52, and rad1 mutations were generated by genetic crosses of YD114 and YD115 with YD101, YD108, or YD102 and sporulation.
The plasmid pNN287, containing tandem his3 fragments (Fasullo and Davis 1987), was introduced into W303 by selecting for Ura+ transformants (YD122). YD122 was then crossed with the dun1-Δ100::his3::URA3 mutant, and a meiotic segregant containing both dun1 and the integrated pNN287 was identified (YD123). Verification that His+ recombinants of YD122 resulted from SCE was demonstrated by the linkage of HIS3 and URA3 contained on pNN287. FOAR isolates of these His+ recombinants should be both Ura– and His–. Of 105 His+ recombinants, ∼95% of the FOAR were His–, confirming that SCE occurred to generate the original recombinants.
Determining rates of spontaneous recombination and mutagenesis: Rates of spontaneous, mitotic recombination were determined by the method of the median as described by Lea and Coulson (1949) and executed by Esposito et al. (1982). Eleven independent colonies were used for each calculation. Three independent experiments were performed for each strain, and the significance of the differences was determined by the nonparametric statistical Whitney-Mann U-test (Zar 1996). Recombination rates were determined using colonies arising on YP(A)D medium, on which Ade+ cells have no selective advantage, and both ADE2 and ade2 colonies appeared white.
Quantitating radiation sensitivity and radiation stimulation of recombination: To quantitate UV or γ-ray sensitivities, strains were grown in YPD to saturation, plated directly on YPD medium after appropriate serial dilution, and irradiated. A 254-nm germicidal lamp (2 J/m2/sec) was used for UV irradiation. A nordion 1.8-kCi 137Cs irradiator was used as a γ-ray source at 7.8 krads/hr. Colony-forming units (CFU) were counted after 3 days of growth and again after 1 wk. At least three independent experiments were performed for each indicated dose. Statistical differences were determined by the two-tailed t-test (Zar 1996).
To quantitate recombination after radiation or chemical exposure, yeast cultures were either grown to an A600 of 0.5–1 for log phase cells or to an A600 of 4 for stationary phase cells. To arrest cells at the G2 phase of the cell cycle, cells were grown to an A600 of 0.5–1 in YPD, nocodazole [methyl-5-(2-thienylcarbonyl)-H-benzimidazole-2-yl-carbamate] (25) was added to a final concentration of 15 μg/ml, cells were incubated at 30° for 3 hr, and cell cycle arrest was confirmed as previously described (Fasulloet al. 1998). Irradiation protocols are described in Fasullo and Dave (1994). After irradiation, cells were plated for viability and for the generation of Ade+ or His+ recombinants. Colonies were counted after 3 days of incubation at 30° and again 1 wk later. For all experiments, measurements of the DNA damage-associated recombination were calculated by subtracting the spontaneous frequency from the frequency obtained after exposure to the agent, as in previous studies (Fasullo et al. 1994, 1998). Statistical differences were determined by the two-tailed t-test (Zar 1996).
UV sensitivity of dun1 and assignment of DUN1 to the RAD3 epistasis group: The UV sensitivity of the dun1 haploid mutant (YA145) was compared to the UV sensitivities of the rad52 (YA133), rad9 (YA132), rad1 (YA131), and rad18 (YA154) haploid mutants (Figure 2). The dun1 mutant exhibited a level of UV sensitivity greater than wild type (P < 0.01) but not significantly different (P > 0.2) from the rad52 mutant defective in recombinational repair. All other single mutants exhibited greater levels of UV sensitivity than the dun1 mutant. To exclude the possibility that the dun1 mutant contained an extragenic suppressor that increased UV resistance, we backcrossed the dun1 mutant with the parental Rad+ W303 strain and determined the UV sensitivity at 30 mJ/m2 of 10 independent haploid segregants containing the dun1::HIS3 disruption. All 10 dun1 haploid segregants exhibited similar UV sensitivity (44% average survival), and all 10 DUN1 haploid segregants exhibited similar UV sensitivity (80% average survival), indicating that the original dun1 mutant did not contain an unlinked extragenic suppressor. Thus, the dun1 mutant is UV sensitive but is significantly more resistant than the mutants defective in the RAD3 or RAD6 pathways.
To determine whether the DUN1 gene participates in either the RAD3, RAD52, or RAD9 pathways for UV resistance, the UV sensitivity of haploid mutants containing combinations of the dun1 null mutation and the rad1, rad9, rad52, and rad18 mutations was quantitated (Figure 2). If DUN1 was important in one particular resistance pathway (epistasis group), then double mutants should exhibit no greater sensitivity than the most sensitive single mutant (Haynes and Kunz 1981). If DUN1 participated in a different UV repair pathway, then the double mutant should exhibit a synergistic (greater than additive) increase in UV sensitivity. The dun1 rad52 (YD103) dun1 rad9 (YD104), and dun1 rad18 (YD107) double mutants exhibited synergistic increases in UV sensitivities compared to the single mutants. However, there was no significant difference (P > 0.4) between the UV sensitivity of the dun1 rad1 double and the rad1 single mutant, even at relatively low levels of radiation exposure (10 and 20 J/m2). In comparison to the dun1 rad52 and dun1 rad9 double mutants, the dun1 rad52 rad9 triple mutant (YD106) exhibited a further synergistic increase in UV sensitivity, although the UV sensitivity was not as great as the rad1 mutant. This indicates that DUN1, RAD52, and RAD9 participate in three independent pathways that confer resistance to UV-induced DNA damage and that DUN1 is a member of the RAD3 epistasis group.
We asked whether the synergistic decrease of UV resistance in the dun1 rad52 mutants, compared to the single mutants, depended on haploidy or growth of cells in stationary phase. As observed for the rad52 dun1 haploid cells, the same synergistic decrease in UV survival, compared to rad52 (YD118) and dun1 (YD117) diploid mutants, was observed in rad52 dun1 diploid (YD119) cells (data not shown). This occurred when cells were grown to either log phase or stationary phase. This suggests that RAD52-dependent recombination pathways participate in the tolerance of UV lesions in both haploid or diploid dun1 mutants.
Sensitivity of the dun1 mutant to ionizing radiation: We asked whether a synergistic increase in ionizing radiation sensitivity, in comparison to the single mutants, also occurred in the dun1 rad52, dun1 rad9, and dun1 rad1 double haploid (Figure 3). In comparison to wild type, all single mutants exhibit γ-ray sensitivity, and rad1 and dun1 mutants exhibit the same low level of γ-ray sensitivity (P > 0.2). The γ-ray sensitivities of the dun1 rad9 and the dun1 rad1 double mutants are not different from the sensitivity of the single rad1 or rad9 mutants (P > 0.3). The γ-ray sensitivity of the dun1 rad52 double mutant was enhanced compared to the rad52 single mutant (P < 0.05) but the increase in sensitivity was not synergistic. These results suggest that unrepaired DNA lesions induced by ionizing radiation in the dun1 mutants are not channeled into either the RAD52 or the RAD6 pathways.
dun1 exhibits higher rates of spontaneous, mitotic heteroallelic recombination: The UV sensitivities of dun1 mutants suggest that either the RAD52 recombination pathway or the RAD6 error-prone pathway participate in UV resistance in the dun1 mutant. We therefore determined whether dun1 strains exhibit a RAD52-dependent, hyper-recombination phenotype. Recombination assays used included heteroallelic recombination, intrachromosomal recombination, and SCE (Figure 1), since UV exposure stimulates these mitotic recombination events (Davieset al. 1975; Kadyk and Hartwell 1993; Galli and Schiestl 1995; Fasulloet al. 1998).
The spontaneous hyper-recombination phenotype of the dun1 mutant was demonstrated by determining the rates of spontaneous Ade+ prototrophs that result from mitotic, homolog recombination between two ade2 alleles, ade2-n and ade2-a. These Ade+ prototrophs can be distinguished from ade2 strains by colony pigment; on YPD medium, Ade+ colonies are white while ade2 strains are red. Since YPD contains insufficient adenine to ensure equal growth rates of Ade+ and Ade– colonies, Ade+ white colony sectors outgrow the perimeter of the red colony, while off-white sectors that do not outgrow the red colonies are Ade– and petite. Dun1 (YD117) colonies accumulated white Ade+ sectors more readily than wild type, but dun1 rad52 (YD119) colonies did not (Figure 4). When colonies are grown on YP(A)D, in which there is no growth advantage for Ade+ sectors, the rate of generating Ade+ prototrophs was ∼5-fold above (P < 0.05) wild type (Table 2). Northern blots demonstrated that the level of ade2 RNA from either DUN1 or dun1 strains was equivalent (data not shown), indicating that the increase in recombination is not due to elevated ade2 transcription in the dun1 mutant. The higher rate of recombination in the dun1 mutant is RAD52 dependent; the rate of heteroallelic recombination in the dun1 rad52 double mutant is indistinguishable (P > 0.05) from that of the rad52 mutant and is reduced ∼15-fold compared to wild type (Table 2).
Since DUN1 participates in the RAD3 pathway, we asked whether the spontaneous hyper-recombination phenotype was dependent upon RAD1 (Table 2). Rates of mitotic heteroallelic recombination in the rad1 strain (YD120) were similar to that of wild type, as expected from previous studies (Monteloneet al. 1988). The hyper-recombination phenotype of the dun1 mutant was also observed in the rad1 dun1 double mutant (YD121), which is significantly different from wild type (P < 0.05) but not significantly different than dun1 (P > 0.05). Dun1 rad1 colonies also accumulate more white Ade+ sectors than wild type on YPD medium (Figure 4). Thus, the hyper-recombination phenotype of dun1 mutants is not dependent upon RAD1.
Aguilera and Klein (1988) identified two classes of mitotic hyper-recombination (hpr) mutants that exhibit elevated intrachromosomal recombination: those that result in increased gene conversion events, and those that result in reciprocal exchange events. We therefore determined whether Ade+ recombinants, occurring by heteroallelic recombination, resulted from gene conversion or reciprocal exchange events. Ten independent, spontaneous Ade+ recombinants isolated from wild type (YD116) and from dun1 (YD117) diploids were sporulated, and tetrads dissected. For all 20 diploids, Ade+::Ade– exhibited 2:2 segregation (data not shown). Ade– meiotic segregants from each Ade+ diploid were backcrossed to ade2-a and ade2-n haploids, YD114 and YD115. If the Ade+ diploid recombinant resulted from reciprocal exchange between the two alleles, then the ade2 meiotic segregants would contain both ade2-a and ade2-n alleles (ade2-a,n), and would not yield spontaneous Ade+ recombinants when backcrossed to YD114 and YD115. From each of the 20 ade2 meiotic segregants from the 10 dun1 recombinants and 10 wild-type recombinants, we obtained Ade– diploids from a single back-cross that yielded Ade+ recombinants. Conversion of ade2-a or ade2-n to ADE2 occurred in approximately the same ratio, 8:12. Thus, we conclude that in all 20 Ade+ recombinants from both wild-type and dun1 mutants, ADE2 resulted from gene conversion and not from reciprocal exchange of the two ade2 alleles. We did not address whether such gene conversion events were associated with flanking marker exchange. Our data are consistent with observations that mitotic, heteroallelic recombination predominately results from gene conversion events (reviewed in Peteset al. 1991).
dun1 mutants exhibit more intrachromosomal gene conversion events but fewer pop-out events: Since the recombination assay that measures heteroallelic recombination generally monitors gene conversion events, we also determined whether dun1 mutants exhibit a bias toward gene conversion events. We used an intrachromosomal recombination assay where pop-outs (loss of an integrated plasmid by either crossovers or gene conversion) could be readily detected (Figure 1). The assay consisted of direct repeats of ade2 that flank URA3 (Huang and Symington 1994). The rate of generating Ade+ recombinants was approximately the same in wild type (YA156) as in the dun1 mutant (YD100), but reduced 10-fold in both the rad52 and the rad52 dun1 mutants (Table 3). In wild type, 57% of Ade+ recombinants are Ura– and result from pop-outs; in dun1, rad1, and dun1 rad1 mutants, ∼5% were Ura–. Consistent with this observation, the rate at which the integrated pKH9 plasmid is lost in dun1 (YD100), as measured by the rate of FOAR colonies, was 6.1 × 10–6, ∼10-fold less than the rate of 5.9 × 10–5 observed in the wild-type strain YA156. This indicates that spontaneous, mitotic recombination in the dun1 mutant results in a higher proportion of gene conversion events and fewer popouts. The rad52 and the rad52 dun1 mutants exhibit higher percentages of pop outs, consistent with previous findings that gene conversion events are RAD52-dependent (Jackson and Fink 1981). Thus, spontaneous intrachromosomal recombination between direct repeats in the dun1 mutant is biased toward gene conversion events that are not associated with crossover events.
DNA damage-induced mitotic recombination in the dun1 mutant: If UV resistance in dun1 mutants is partially conferred by RAD52-dependent recombinational repair, then UV-induced mitotic recombination should be enhanced in dun1 mutants. Since the synergistic effect of dun1 and rad52 is observed in haploids, where heteroallelic recombination does not occur, as well as diploids, we investigated whether dun1 mutants exhibit an enhanced level of UV-induced unequal SCE (Table 4). We found no differences in the rate of spontaneous SCE in the dun1 mutant (1.1 ± 0.2) × 10–6, compared to the rate in wild type (1.4 ± 0.3) × 10–6. After UV irradiation, we observed significant (P < 0.05) two- to threefold increases in recombination, compared to wild type. However, if cells were arrested in G2 with the micro-tubule inhibitor nocodazole prior to irradiation, there was no difference between the stimulation of SCE events in the dun1 mutant compared to wild type. We speculate that dun1-enhanced UV-stimulated SCE depends on replication of the UV dimer.
Since spontaneous heteroallelic recombination is greater in dun1 diploids than wild type, we exposed both wild-type and dun1 diploid mutants to UV, γ-rays, and MMS to quantitate whether the stimulation of heteroallelic recombination was greater in dun1 cells for any particular DNA damaging agent (Table 5). Although the dun1-enhanced recombination is modest (approximately twofold), it is significantly different (P < 0.01) from wild-type levels of UV stimulation. However, there was no increase in the levels of γ-ray and MMS-stimulated heteroallelic recombination in the dun1 mutant (data not shown). Thus, the recombinogenicity of only a subset of DNA-damaging agents is enhanced in dun1.
Higher percentages of petites are generated in dun1 mutants: We observed a 10-fold higher percentage of petites generated in the dun1 (YA145) strain (7.1 ± 0.9% or 57/804 total) compared to the wild-type (YA103) strain (0.7 ± 0.4% or 13/1697 total). However, rates of spontaneous reversion to Trp+ and Ade+ prototrophy due to spontaneous mutagenesis are not higher in the dun1 mutant (data not shown). Since mitochondrial DNA repair is less efficient than genomic DNA repair (Yakes and van Houten 1997), higher spontaneous numbers of petites is consistent with the DNA repair defect.
Three independent pathways participate in the repair of UV-induced DNA damage in S. cerevisiae. These include pathways for excision repair, mutagenic repair, and recombinational repair. Which pathway is chosen to repair UV-induced DNA damage is not completely understood. DUN1 encodes a serine-threonine protein kinase that is activated by signal transduction pathways for sensing DNA damage. Dun1 mutants, which exhibit both UV and MMS sensitivity, are defective in DNA damage-inducible expression of ribonucleotide reductase (Zhou and Elledge 1993) and MAG1 (Liu 1997) and are partially defective in DNA damage-induced G2 checkpoint (Patiet al. 1997). In this study, we demonstrate that DUN1 participates in the RAD3 pathway for excision repair and that dun1 mutants exhibit higher rates of spontaneous gene conversion and enhanced UV-induced mitotic recombination. We suggest that the DUN1 pathway for DNA damage-induced gene expression channels the repair of UV damage into a nonrecombinogenic pathway.
The participation of DUN1 in the RAD3 epistasis group elucidates previous observations that DUN2 (POLϵ)-mediated transcriptional induction and RAD9-mediated cell cycle arrest are independent DNA damage-inducible pathways (Navaset al. 1996). The independence of these pathways was supported by the enhanced UV sensitivity of the dun2 rad9 mutant, compared to the single mutants. Based on the participation of DUN1 in the excision repair pathway, we speculate that the DUN2 pathway may activate other genes that participate in UV excision repair.
Differences in genetic interactions between rad9 and dun1 regarding sensitivity to UV and ionizing radiation: Since several models of DNA double-strand break repair involve DNA synthesis (Szostaket al. 1983; Malkovaet al. 1996), we expected that dun1 mutants would also be sensitive to ionizing radiation. However, we observed no synergism between rad9 and dun1 with respect to γ-ray sensitivity. Nonetheless, the dun1 γ-ray sensitivity is still consistent with DUN1 belonging to the RAD3 epistasis group, since dun1 mutants exhibit the same level of γ-ray sensitivity as rad1 mutants. Essentially identical γ-ray survival curves are obtained for single mutants dun1 and rad1 and for rad1 dun1 double mutants (Figure 3), for rad9 and dun1 rad9 (Figure 3), and for rad9 and rad9 rad1 (Fasulloet al. 1998). Further experiments are in progress to determine whether DUN1, similar to RAD1, participates in the single-strand annealing (SSA) mechanism for double-strand break repair (Fishman-Lobell and Haber 1992).
Recombination phenotypes of the dun1 mutants: Since tolerance to DNA damage in dun1 mutants is partially conferred by RAD52-dependent mechanisms, it is interesting that only a subset of spontaneous RAD52-dependent recombination events is enhanced in the dun1 mutant, compared to wild-type strains. While spontaneous heteroallelic and intrachromosomal gene conversion events are enhanced, intrachromosomal popouts are decreased, and spontaneous unequal SCE is unchanged in dun1 mutants. Although we do not understand why the proportion of intrachromosomal recombination events that result from exchange (pop-outs) is decreased in the dun1 mutant, a similar phenomena (Table 3) was observed for rad1 mutants, suggesting that DUN1 also participates in the SSA mechanism for spontaneous recombination. We speculate that the unchanged rate of spontaneous SCE in dun1 mutants, compared to the rate in wild type, results from a lower level of crossovers but a higher level of sister-chromatid gene conversions. However, due to the low frequency of spontaneous SCE, the percentages of spontaneous SCE that result from reciprocal exchanges and gene conversion events are unknown (Fasullo and Davis 1987; Kadyk and Hartwell 1992). Thus, detection of enhanced spontaneous recombination in dun1 mutants may depend on the recombination assays used to monitor gene conversion.
Which spontaneous DNA lesions are more recombinogenic in the dun1 mutant is unclear. Spontaneous DNA lesions include those that result from oxidation and alkylation damage (depurination). Since neither ionizing radiation nor MMS were more recombinogenic in dun1 mutants, dun1-enhanced spontaneous recombination may result from spontaneous DNA lesions similar to those induced by UV.
The synergistic increase in UV sensitivity in the rad52 dun1 double mutant, compared to the single mutants, for both haploids and diploids, suggests that one recombination mechanism by which UV damage is tolerated in dun1 mutants is SCE. Although in comparison to wild type, the frequency of UV-stimulated recombinants that result from unequal SCE increased a modest two- to threefold in the dun1 strain, a more direct physical measurement of SCE may demonstrate a greater enhancement. Less enhanced UV stimulation of heteroallelic recombination in the dun1 mutant is consistent with observations that sister chromatids are preferred substrates for DNA repair (Kadyk and Hartwell 1992). Thus, we suggest that higher levels of SCE and heteroallelic recombination constitute a subset of RAD52-dependent mechanisms that lead to DNA damage tolerance in the dun1 mutant.
Possible mechanisms for dun1-stimulated recombination: The RAD1 independence of the spontaneous hyper-recombination phenotype of the dun1 mutants suggests possible mechanisms for how dun1-enhanced recombination occurs. Similar to dun1 recombination phenotypes, spontaneous RAD1-independent intrachromosomal recombination (Rattray and Symington 1995) and UV-induced RAD1-independent unequal SCE (Paulovichet al. 1998) mostly occurs by RAD52-dependent gene conversion. Kadyk and Hartwell (1993) suggest that such SCE events may be initiated when UV lesions are bypassed by the replication machinery; the resulting recombinogenic single-strand gaps may then be processed into either daughter-strand gap, double-strand break, or replication slippage pathways. The cell cycle dependence of the dun1-enhanced UV stimulation also suggests that in dun1 mutants, some UV lesions are bypassed by the replication enzymatic machinery and the resulting recombinogenic single-strand gaps may trigger RAD52-dependent recombination. The RAD1-independence of the dun1 hyper-recombination phenotype differs from the RAD1-dependence of the hyper-recombination phenotypes of other DNA repair mutants. These include rad9 mutants (Fasulloet al. 1998) and rem mutants (Malone and Hoekstra 1984) defective in the Rad3 helicase (Monteloneet al. 1988). Rad1, as part of the Rad1-Rad10 endonuclease (Bardwellet al. 1994), may participate in mitotic recombination by excising nonhomologous DNA from recombinogenic double-strand breaks to form stable recombination intermediates (Fishman-Lobell and Haber 1992), or by processing DNA lesions into recombinogenic single-stranded DNA during excision repair (Monteloneet al. 1988). The first role of RAD1 may be important in hyper-recombination mutants where recombination between repeated sequences on nonhomologous chromosomes (ectopic recombination) is enhanced (Bailiset al. 1992; Fasulloet al. 1998). Since the dun1-enhanced recombination occurs between homologs or chromatids that are essentially identical, and the initiating lesion may be generated by DNA replication, RAD1 may not be needed to either initiate or process the recombination intermediates.
DNA damage-inducible gene expression and dun1 phenotypes: The DUN1 pathway for DNA damage-inducible gene expression controls the expression of several genes involved in DNA repair, rendering it difficult to ascribe all the dun1 phenotypes to a defect in the DNA damage inducibility of the RNR genes. However, the UV sensitivity of dun1 mutants can be suppressed by the overexpression of RNR1 (Zhaoet al. 1998), and cells exposed to hydroxyurea, an inhibitor of ribonucleotide reductase, also exhibit more mitotic recombination (Galli and Schiestl 1996). Both dun1 and cdc8 (deoxy-thymidylate kinase; Jonget al. 1984) mutants exhibit higher levels of mitotic recombination and spontaneous petites (Sclafani and Fangman 1984). Thus, by analogy with cdc8 phenotypes, nulceotide imbalance may result in some of the dun1 phenotypes.
Since the substrates for the Dun1 protein kinase have not been identified, we cannot exclude the possibility that some dun1 phenotypes directly result from failure to phosphorylate DNA repair proteins. For example, the Srs2/RadH/Hpr5 helicase, which exhibits amino acid similarity to UvrD, contains several potential phosphorylation sites for serine/threonine protein kinases (genetics computer group, protein motif comparison). Hpr5 mutants exhibit variable, elevated rates of mitotic gene conversion (Palladino and Klein 1992). Thus, a full explanation for the dun1 recombination pheno-types awaits the identification of the substrates of Dun1 kinase.
Mutants defective in other yeast protein kinases, including Cdc5 (Aguilera and Klein 1988) and Pkc1 (Huang and Symington 1994), exhibit spontaneous hyper-recombination phenotypes. Currently the interactions between these protein kinases and Dun1 is unknown. PKC1 is involved in stress responses that include heat and osmotic shock, and pkc1 mutants exhibit an extended S phase (Levin and Bartlett-Heubusch 1992). Heat shock may induce recombination by causing oxidative damage (Brennanet al. 1994; Davidsonet al. 1996). Thus, it would be interesting if pkc1 rad52 double mutants exhibited a synergistic sensitivity to environmental stress.
Summary: We have shown that DNA damage tolerance in dun1 mutants is partially conferred by RAD52-dependent recombination. This is the first demonstration that a defect in a protein kinase directly involved in the transcriptional induction of DNA damage-inducible genes also increases spontaneous, mitotic recombination in yeast. It will be important to determine whether other protein kinases involved in the DNA damage-inducible responses have similar phenotypes.
We thank R. Rothstein for the W303 derived strains that contain the rad1, rad52, rad9, and rad18 gene disruptions, S. Elledge and Z. Zhou for dun1 mutants, and L. Symington for YK12. We thank N. Faegerman, T. Hryciw, W. Xiao, and Z. Dong or carefully reading this manuscript. The work was supported by U.S. Public Health Service grant CA70105, and a grant from the Leukemia Research Foundation.
Communicating editor: M. Lichten
- Received November 18, 1998.
- Accepted April 15, 1999.
- Copyright © 1999 by the Genetics Society of America