https://orcid.org/0000-0002-3948-4554
Abstract
Ultraviolet-induced DNA lesions impede DNA replication and transcription and are therefore a potential source of genome instability. Here, we performed serial transfer experiments on nucleotide excision repair-deficient (rad14Δ) yeast cells in the presence of chronic low-dose ultraviolet irradiation, focusing on the mechanisms underlying adaptive responses to chronic low-dose ultraviolet irradiation. Our results show that the entire haploid rad14Δ population rapidly becomes diploid during chronic low-dose ultraviolet exposure, and the evolved diploid rad14Δ cells were more chronic low-dose ultraviolet-resistant than haploid cells. Strikingly, single-stranded DNA, but not pyrimidine dimer, accumulation is associated with diploid-dependent fitness in response to chronic low-dose ultraviolet stress, suggesting that efficient repair of single-stranded DNA tracts is beneficial for chronic low-dose ultraviolet tolerance. Consistent with this hypothesis, homologous recombination is essential for the rapid evolutionary adaptation of diploidy, and rad14Δ cells lacking Rad51 recombinase, a key player in homologous recombination, exhibited abnormal cell morphology characterized by multiple RPA–yellow fluorescent protein foci after chronic low-dose ultraviolet exposure. Furthermore, interhomolog recombination is increased in chronic low-dose ultraviolet-exposed rad14Δ diploids, which causes frequent loss of heterozygosity. Thus, our results highlight the importance of homologous recombination in the survival and genomic stability of cells with unrepaired lesions.
Introduction
Ultraviolet (UV) radiation in sunlight is the primary cause of exogenous DNA damage. UV radiation induces DNA photoproducts, such as cyclobutane pyrimidine dimers (CPDs) and pyrimidine-pyrimidone (6–4) photoproducts (Friedberg et al. 2006; Rastogi et al. 2010). These dimers create structural distortions in duplex DNA strands, thereby impeding DNA replication and transcription. Nucleotide excision repair (NER) is a highly conserved DNA repair mechanism that efficiently removes various helix-distorting lesions, such as UV-induced DNA photolesions, oxidative DNA damage, and various aromatic amines (German 1995; Friedberg et al. 2006; Scharer 2013; Spivak 2015). NER is highly efficient in removing these lesions. However, it is difficult to remove all lesions if lesions occur at sites about to be replicated; hence, DNA damage-related replication stress is a potential source of genome instability, which is a hallmark of cancer cells (Cleaver 2005; Halazonetis et al. 2008; Marteijn et al. 2014). To overcome the problem of replication forks encountering unrepaired DNA lesions, cells utilize several DNA damage response mechanisms, including DNA damage tolerance (DDT), homologous recombination (HR), and DNA damage checkpoint (DDC) pathways, which ensure complete DNA replication by effectively bypassing fork-blocking lesions, repairing single-stranded DNA (ssDNA) gaps behind replication forks, and regulating cell-cycle progression (Lopes et al. 2006; Cimprich and Cortez 2008; Ciccia and Elledge 2010; Covo et al. 2012; Jossen and Bermejo 2013; Saugar et al. 2014; Yin and Petes 2015; Branzei and Szakal 2016). These DNA damage response pathways either individually or in corroboration prevent fork stalling and/or DNA damage accumulation, allowing cells to maintain genomic integrity under various genotoxic stress conditions.
In natural environments, cells are continuously exposed to very low UV irradiation. Therefore, in addition to traditional approaches involving acute UV light exposure, understanding the cellular response to chronic low-dose UV (CLUV) exposure is an important complementary approach to help clarify the biological significance and crosstalk of each DNA damage response pathway. Previous studies on budding yeast have revealed that the DDT pathway is essential for promoting cell growth under CLUV exposure (Hishida et al. 2009). In CLUV-exposed DDT-deficient cells, bulk DNA replication is completed by other replication-bypass mechanisms, possibly involving re-priming downstream of the damage; however, ssDNA gaps accumulate, resulting in DDC-dependent cell-cycle arrest in the G2 phase (Hishida et al. 2009, 2010). In humans, low-dose UV irradiation induces ssDNA in DNA polymerase η-deficient XP-V cells, leading to S/G2 arrest (Quinet et al. 2014). These findings indicate that CLUV-induced lesions can potentially block replication fork progression, although NER can efficiently remove UV-induced DNA damage. Thus, resolving DNA replication malfunctions through DDT is crucial for adapting to CLUV irradiation. In contrast, NER-defective rad14Δ yeast cells, which lack the yeast ortholog of human xeroderma pigmentosum complementation group A (XPA), do not impair the viability in the presence of CLUV, albeit with a modest delay in growth rate; however, they accumulate irreparable UV-photoproducts and show enhanced mutagenesis (Hishida et al. 2009; Haruta et al. 2012). Thus, the NER pathway plays a critical role in the maintenance of genome stability rather than cell viability during CLUV exposure, which likely reflects the clinical phenotypes of NER-deficient XP patients, such as severe predisposition to UV-induced skin cancer (Berneburg and Lehmann 2001; Heinen et al. 2002; Cleaver 2005; Marteijn et al. 2014).
Defects in DNA repair mechanisms can lead to DNA damage accumulation in genomic DNA, which can then lead to cancer development (Halazonetis et al. 2008; Tian et al. 2015). However, how cells lacking DNA repair pathways achieve both the stress tolerance and genomic instability, to promote their survival and evolution is unclear. To elucidate the cellular tolerance mechanism that facilitates DNA replication through unrepaired UV lesions, we cultured NER-deficient rad14Δ cells by serially transferring them under prolonged CLUV exposure. We found that haploid rad14Δ cells become diploid within 50 generations after CLUV exposure, and the evolved diploid rad14Δ cells were more tolerant to CLUV than the original haploid cells. Importantly, in response to CLUV stress, the ability to process ssDNA, but not pyrimidine dimers, affects diploid-dependent fitness. We also found that although HR provides a growth advantage of diploids over haploids during CLUV exposure, cell proliferation in this situation leads to a frequent loss of heterozygosity (LOH). Thus, the HR-dependent recovery from CLUV-induced stress contributes to cell proliferation and mutagenesis, which could provide important insights into understanding how DNA repair-deficient cells adapt to their environment.
Materials and methods
Yeast strains
All yeast strains used in this study are derivatives of BY4741 and BY4742 and are listed in Supplementary Table 1. All deletion mutants were created by replacing the relevant open reading frames in these strains with selectable markers (Longtine et al. 1998). RFA1-YFP was created using a previously described polymerase chain reaction (PCR)-based method (Reid et al. 2002). Standard genetic procedures were used for strain construction and medium preparation (Amberg et al. 2006). Yeast cells were routinely grown in YPD medium (1% yeast extract, 2% peptone, 2% dextrose), containing 0.003% adenine sulfate (YPDA). Synthetic complete (SC) medium was used in strain selection and recombination and mutation frequency analysis. For the G1 synchronization of cells, the cultures were treated for 2 h with 10 μg/ml of α-factor (Sigma-Aldrich).
Growth medium and conditions
Strains were grown in yeast extract-peptone-dextrose medium containing 0.01% adenine sulfate (YPDA) to saturation (2 × 108 cells/ml) at 30°C. A fraction of the population was periodically transferred to the fresh YPDA medium (1 × 105 cells/ml. The cultures (4 ml) were then poured into standard 90 mm Petri dishes (Sterilin; Bibby Sterilin Ltd.) covered with a transparent plastic lid. During each passage, the cultures were incubated with horizontal shaking at 50 rpm under continuous UV irradiation (GL-10 germicidal lamp; Toshiba) for 24 h. The working dose of CLUV irradiation was ∼0.16 J/m2/min. UV irradiance was measured using a UVX radiometer UVX-25 (Ultraviolet Products).
Mutation frequency at CAN1 locus
Mutation frequencies were measured as previously described (Haruta et al. 2012). Briefly, cells were grown in the YPDA medium under CLUV exposure and collected from each passage. The appropriate dilutions of cells were incubated at 30°C for 3 days on a synthetic complete (SC) medium and the canavanine (60 µg/ml)-supplemented SC medium to measure the number of total viable cells and CanR mutants, respectively. Plates were incubated at 30°C for 3 days. The mutation frequency was calculated by dividing the number of CanR colonies/ml by the number of viable cells/ml, and the average was calculated for at least 3 independent sets of experiments.
Spot assay: sensitivity to CLUV irradiation
Cells were grown from single colonies to saturation (2 × 108 cells/ml) at 30°C. Tenfold serial dilutions of the cultures were spotted on the YPDA plates, covered with transparent plastic caps, and incubated for 3 days at 30°C in the presence of CLUV irradiation.
Cell viability
Cells were diluted with distilled water to an appropriate concentration. Total cell number was determined by microscopy. Cell counting was performed using hemocytometer at 400×magnification using a Zeiss Axioplan2 microscope. Colony-forming units (cfu) were measured by spreading 0.1 ml of serially diluted samples on YPDA plates. The plates were incubated at 30°C for 2 days, and colonies were counted manually. The viability was calculated at the indicated times as the number of viable cells (cfu) divided by the number of total cells.
Dot-blot analysis
Cells were grown in 5 ml aliquots and harvested at the indicated time points. DNA extracts were prepared as previously described (Giavara et al. 2005), resuspended in an alkaline buffer (final concentration; 0.4 M NaOH, 10 mM EDTA), and incubated at 100°C for 10 min. The samples were then spotted on a Hybond-N+ membrane (GE Healthcare) using a dot-blot apparatus (Bio-Rad) according to the manufacturer's instructions and dried at 80°C for 2 h. Then, membranes were blocked with 1.25% milk at 25°C for 1 h and probed overnight with a mouse anti-CPD monoclonal antibody (TDM2; Cosmo Bio). The dots were visualized using chemiluminescence after incubating with secondary horseradish peroxidase (HRP)-conjugated antibodies (GE Healthcare) and the Western Blot Ultra-Sensitive HRP substrate (Takara Bio). The relative intensity of each dot was quantified using ImageJ software (NIH) to compare with the initial amount of CPD per µg haploid rad14Δ DNA.
Preparation of yeast extracts and western blotting
Yeast cell extracts were prepared from yeast cultures (2 × 108 cells) using the trichloroacetic acid (TCA) method, as previously described (Hishida et al. 2006). TCA-precipitated protein extracts were separated using 6% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride membranes, and detected with anti-Rad53 (yC-19) antibody (Santa Cruz Biotechnology) using HRP-conjugated anti-goat IgG (SC2768) as a secondary antibody (Santa Cruz Biotechnology).
Flow cytometry
Yeast cells were fixed in 70% ethanol. After counting the cell number, 5 × 106 cells were washed with 50 mM sodium citrate/HCl (pH 7.5) and treated with 0.25 mg/ml RNase A (Novagen) and 20 mg/ml proteinase K (Roche) for 1 h at 50°C. Cells were then washed with 50 mM sodium citrate/HCl (pH 7.5), sonicated, stained with 1 mg/ml propidium iodide (PI; Sigma) for 1 h, and analyzed using a Becton Dickinson Cant II system using fluorescence-activated cell sorting (FACS) software (BD Biosciences).
Evaluation of chromosome loss and mitotic recombination/mutation frequencies
Diploid zygotes were isolated from freshly mated strains grown on the solid SC medium lacking histidine and uracil. After growth for 3–4 days on the solid YPDA medium at 30°C, diploid cells were grown from single colonies to the early-log phase, and 1 × 106 cells/ml cultures were further incubated in the presence or absence of CLUV irradiation for 24 h. Cells from each culture were washed, diluted, and cultured on YPDA plates and canavanine-containing SC plates to measure the total cell number and CanR cell number, respectively. After culturing for 3 days at 30°C, the canavanine-containing plates were replica-plated on the SC medium lacking uracil. The chromosome loss frequency was calculated by dividing the number of CanR and Ura− colonies/ml by the number of viable cells/ml. CanR Ura+ colonies were further analyzed through colony PCR at the CAN1/HIS3 heteroallele to distinguish between mitotic conversion (LOH) and spontaneous mutation events. CF1 (5′-GCTACGTATGATAATAGCCCGCC-3′) and CR1 (5′-GCTACGTATGATAATAGCCCGCC-3′) oligonucleotides were used to amplify the CAN1/HIS3 locus. The CAN1 allele gives a PCR product of 2.8 kbp, whereas the HIS3 allele gives a PCR product of 1.5 kbp. Therefore, a heterozygous CAN1/HIS3 diploid shows bands of both sizes, but a homozygous HIS3/HIS3 diploid shows a single 1.5 kbp band. The average of at least 3 independent experiments were calculated.
Other methods
Cell synchronization and fluorescence microscopy were performed, as previously described (Hishida et al. 2002).
Results
Effect of CLUV irradiation on genome stability in NER-deficient rad14Δ cells
We have previously shown that CLUV exposure does not impair the viability of haploid rad14Δ cells during 24 h; however, it increases the mutation frequency compared to that in the wild-type cells (Hishida et al. 2009; Haruta et al. 2012), implying that the ability to overcome replication problems ensures cell growth. To further examine the effects of prolonged CLUV exposure on rad14Δ cell viability and genome stability, we performed serial transfer experiments under CLUV exposure over a long period, where the cells were passaged for 5 days under CLUV exposure by transferring 0.1% culture volume into a fresh YPDA medium every 24 h (corresponding to ∼11 generations daily). Both wild-type and rad14Δ cells displayed similar colony-forming units throughout the experiment (Fig. 1a), indicating that prolonged CLUV exposure does not impair the viability of NER-deficient cells.
Haploid rad14Δ populations converge toward diploids under CLUV irradiation. a) Wild-type and rad14Δ cells grown to saturation were diluted daily to 1 × 105 cells/ml using the fresh YPDA medium and grown under CLUV exposure at 30°C. The plating efficiency was determined daily. b) CLUV-induced mutation frequency in CAN1. Cells were subcultured in the presence of CLUV. Data points are the average of at least 3 independent experiments. Error bars indicate the standard error (SE) for each data point. c) DNA content in the presence of CLUV. PI-stained cells were analyzed using FACS. d) DNA content in the rad14Δ haploid and 4 evolved rad14Δ diploid clones (AD1−AD4) analyzed using FACS. e) The relevant strains were spotted at 10-fold serial dilutions onto YPDA plates, and incubated in the presence or absence of CLUV at 30°C for 3 days.
Next, we examined canavanine resistance-inducing forward mutations in the CAN1 gene in wild-type and rad14Δ cells. In wild-type cells, CLUV did not increase the mutation frequency over time (Fig. 1b). Interestingly, although CAN1 mutation frequency increased in rad14Δ cells upon CLUV exposure during early time points (∼33 generations), it gradually decreased to near basal levels (∼55 generations; Fig. 1b). These results suggest that prolonged CLUV stress can alter the DNA damage response in NER-deficient cells, which may impact the mutation frequency.
Haploid rad14Δ populations converge toward diploidy during CLUV exposure
During the serial transfer experiments, we found that CLUV-exposed rad14Δ cells increased in size more than the unexposed rad14Δ haploids (data not shown). Since cell size is often correlated with the ploidy level, we measured the DNA content kinetics in CLUV-exposed rad14Δ cells using FACS. Wild-type populations displayed a haploid DNA content profile with 2 peaks, 1C and 2C, corresponding to the G1 and G2/M phases of the cell cycle, respectively (Fig. 1c), throughout the experiments, indicating that wild-type ploidy is stable under CLUV exposure. In contrast, 3 peaks were observed in rad14Δ populations 3 days after CLUV exposure (∼33 generations), and subpopulations with 2 peaks (2C and 4C) became dominant by 5 days (∼55 generations; Fig. 1c), indicating that the entire haploid population became diploid. This is consistent with the rapid decrease in mutation frequency later during the experiment, as diploid cells possess 2 copies of the reporter gene (CAN1/CAN1), and loss-of-function mutations in both CAN1 alleles are required for canavanine resistance.
To explore the ploidy changes during CLUV exposure, we mixed equal numbers of 2 independent LEU2- and HIS3-labeled MATarad14Δ haploids and subcultured the mixtures for 5 days in the presence of CLUV irradiation. Most resulting diploid cells were either LEU+ or HIS+, and the ratio of the 2 populations remained the same throughout the experiment, whereas LEU+ HIS+ diploids were hardly detected during CLUV exposure (Supplementary Table 2). In addition, the evolved diploids were still MATa, because the majority of their cells responded to the α-factor. These results suggest that diploidization of haploid cells due to mating-type switching or cell–cell fusion is extremely rare events. In this context, a previous study showed that spontaneous diploidization due to whole-genome duplication (endoreduplication) is a relatively common event in haploid yeast populations (∼10−4 per generation), and the relative fitness of these new diploids determines their fate within the population (Harari et al. 2018). Taken together, we speculate that the diploidization of haploid cells is mainly caused by endoreduplication and that a small subpopulation of diploid cells spontaneously generated in the original haploid population might have outcompeted the rest of the population under CLUV exposure because they had a significant growth advantage over haploid cells.
Diploids are better adapted to CLUV stress than haploids
To test whether diploid rad14Δ cells are more resistant to CLUV than haploid cells, we isolated 4 evolved diploid clones (MATa/MATa) (Fig. 1d), designated as AD1 to AD4, from cell cultures independently passaged for 5 days under CLUV exposure, and tested their CLUV sensitivity by the spot assay. We note that the CLUV sensitivity of cells in solid spot assays differs from that in liquid serial-transfer assays in the following ways: (1) the liquid assay shows less CLUV sensitivity than the solid assay because the UV light is shielded to some extent by the medium and (2) in the solid assay, the ploidy of cell population is not replaced during CLUV exposure. The spot assay revealed that haploid rad14Δ cells displayed increased sensitivity to CLUV irradiation, but all the evolved diploid rad14Δ clones were more resistant to CLUV irradiation than the original haploid rad14Δ strain (Fig. 1e). Similar to the evolved diploids, canonical rad14Δ diploids (MATa/MATα), generated by mating between rad14Δ haploids of opposite mating types, also showed similar CLUV resistance (Fig. 1e). These results suggest that the diploid itself has a significant fitness advantage for CLUV tolerance. Thus, adaptive changes in ploidy in response to CLUV stress might be determined by the beneficial effects of increased ploidy.
ssDNA, but not pyrimidine dimer, accumulation contributes to a CLUV tolerance for diploid cells under CLUV stress
To investigate whether haploid and diploid rad14Δ cells differ in UV-induced pyrimidine dimer yield, cells were exposed to CLUV, and the CPD content of genomic DNA was measured through a dot-blot assay using the anti-CPD antibody. The number of CPDs per µg DNA was higher in both CLUV-exposed haploid and diploid rad14Δ cells than that in the respective nonirradiated control cells, and no significant differences between haploid and diploid cells were noted (Fig. 2, a and b), indicating that diploid-specific CLUV tolerance is not due to differences in pyrimidine dimer yield. Notably, the amount of CPDs that accumulated in genomic DNA remained constant after the first day under CLUV exposure (Fig. 2b), in contrast to the previous data observed after acute UV irradiation, where the number of CPDs in rad14Δ cells linearly increased with increasing acute UV irradiation (Hishida et al. 2009). This probably reflects the different UV irradiation setting conditions. CLUV irradiation of rad14Δ cells is expected to cause old DNA strands to continually accumulate CPDs. However, this accumulation is not detected in the serial-transfer experiment because only a very small fraction of old DNA is present at the beginning of each round. Indeed, of the DNA purified from CLUV-irradiated cells, half of the DNA strands are due to de novo DNA synthesis during the last cell cycle. Half of the opposite strand also originated from DNA synthesis 1 generation earlier. Thus, even though the cells were cultured under CLUV for 24 h, the majority of DNA is not exposed to CLUV irradiation throughout the experiment, but rather derived from newly synthesized, less UV-damaged strands within a few generations. Thus, it is most likely that in rad14Δ cells, a dynamic equilibrium is maintained between de novo DNA synthesis and increasing amounts of CPDs without reaching saturation levels during CLUV exposure. These results imply that cell proliferation itself increases the tolerance to CLUV stress, which may explain why NER-deficient cells remain viable during CLUV exposure.
Accumulation of pyrimidine dimers does not contribute to the selective advantage of diploid cells against CLUV stress. a) Dot-blot analysis of DNA extracted from CLUV-exposed cells using an anti-CPD antibody. The haploid and diploid rad14Δ cells were subcultured in the presence of CLUV, as in Fig. 1a and analyzed daily. The genomic DNA was extracted, spotted onto membranes, and then subjected to immunoblotting with the TDM2 antibody. b) CPD accumulation was measured as the relative intensities of dot blots. The CPD levels were normalized against those in total DNA and expressed as relative intensity compared with the initial damage in haploid rad14Δ cells (time 0). SEM (at least n = 3) is indicated. c) CLUV-exposed rad14Δ haploid and diploid cells with RPA-YFP foci. Cells were passaged for 4 days under CLUV exposure by transferring 0.1% of the culture volume into fresh YPDA medium every 24 h, fixed, and analyzed using fluorescence microscopy. Scale bar, 10 µm. d) Cells subcultured under CLUV exposure were collected daily and examined using fluorescence microscopy. The graph shows the percentage of cells with RPA foci. Error bars represent the SE of 4 independent experiments.
Previous studies have shown that ssDNA gaps accumulate during the S phase after acute UV irradiation, suggesting that the UV-induced lesions would cause replication blockage, leading to the formation of postreplication ssDNA gaps (Lopes et al. 2006; Quinet et al. 2014). To examine whether the ssDNA gap levels affect cellular tolerance to CLUV stress, we observed the subcellular localization of Rfa1-YFP, a large subunit of the RPA complex labeled with yellow fluorescent protein (YFP). Rfa1–YFP forms foci (RPA foci) in response to replication stress or DNA damage and serves as a reliable indicator of DNA damage bypass tracts (Lisby et al. 2004; Wong et al. 2020). Although only few wild-type haploid cells had RPA foci during CLUV exposure, an increased number of rad14Δ haploids with RPA foci were observed upon CLUV treatment (Fig. 2, c and d). Interestingly, the number of rad14Δ haploids with RPA foci gradually decreased during CLUV exposure (82.5% on day 1 to 42.9% on day 4; Fig. 2d), which coincided with the emergence of evolved diploid cells. Consistently, the number of rad14Δ diploids with RPA foci showed approximately 2-fold decrease compared to that of rad14Δ haploids, when cells were incubated under CLUV irradiation for 1 day and did not change over time (Fig. 2d). Thus, it is likely that the decrease in the CLUV-induced number of cells with RPA foci may be correlated with an increase in ploidy. Taken together, these results imply that the CLUV resistance of NER-deficient cells involves their ability to process ssDNA generated during replication through unrepaired lesions.
HR is required for population conversion from haploidy to diploidy in CLUV-exposed rad14Δ cells
RPA-coated ssDNA accumulation triggered by replication stress can potentially undergo DDC activation and HR repair (Fanning et al. 2006; Gangavarapu et al. 2007; Yin and Petes 2015). To investigate the potential roles of DDC and HR in CLUV tolerance of rad14Δ cells, we constructed rad14Δ mec1Δ and rad14Δ rad51Δ strains in haploid and diploid backgrounds. The mec1Δ strain also carries an sml1Δ mutation, which suppresses the lethality of mec1Δ. CLUV weakly affected the growth of DDC-deficient (mec1Δ) and HR-deficient (rad51Δ) haploids, but rad14Δ mec1Δ and rad14Δ rad51Δ haploids exhibited higher sensitivity to CLUV than their respective single mutants (Fig. 3a). Given that the synergistic effect of combining them on CLUV sensitivity is due to the lack of 2 different pathways by which cells either eliminate or tolerate lesions, both the HR and DDC pathways may play important roles in CLUV tolerance in NER-deficient cells. Remarkably, while mec1Δ and rad14Δ mec1Δ diploids were more resistant to CLUV than their haploid counterparts, the CLUV sensitivity of rad51Δ and rad14Δ rad51Δ diploids was similar or slightly higher than that of the respective haploids (Fig. 3a). These results suggest that HR is not only crucial for CLUV resistance but also confers a growth advantage of diploids over haploids during CLUV exposure. To confirm this, we measured the DNA content of rad14Δ mec1Δ and rad14Δ rad51Δ haploids throughout the serial-transfer experiment under CLUV condition. We applied ∼0.2 J/m2/min UV irradiation dose, corresponding to approximately one-eighth of the canonical CLUV irradiation dose, to ensure the growth of both strains in the liquid assay. We found that rad14Δ mec1Δ haploid populations became diploid in response to CLUV irradiation, whereas rad14Δ rad51Δ populations remained haploid (Fig. 3b). These results suggest that the rapid conversion to diploid is due to differences in their CLUV sensitivity between haploid and diploid cells as well as the ploidy change observed in rad14Δ cells and that HR contributes to conferring a growth advantage of rad14Δ diploids under CLUV conditions.
HR is required for CLUV tolerance in rad14Δ cells. a) The indicated strains were spotted onto YPDA plates and incubated under CLUV exposure. UV irradiation doses are 1.6 J/m2/min (CLUV) and 0.2 J/m2/min (CLUV ×1/8). b) rad14Δ mec1Δ sml1Δ and rad14Δ rad51Δhaploids were subcultured for 5 days under CLUV exposure, and the cell-cycle profiles of these cultures were determined using FACS analysis.
CLUV causes late S/G2 cell-cycle arrest and cell death in rad14Δ rad51Δ cells
The above results demonstrate that HR plays a critical role in CLUV resistance in both haploid and diploid rad14Δ cells. Therefore, we examined the effect of the rad51Δ mutation on the growth and viability of CLUV-exposed haploid and diploid rad14Δ cells. The liquid growth assay revealed that rad14Δ and rad51Δ haploid/diploid cells had a modest growth delay in the presence of CLUV (Fig. 4a). In contrast, CLUV significantly impaired the growth of rad14Δ rad51Δ haploids/diploids (Fig. 4a). Microscopic analyses revealed that, 8 h after CLUV irradiation, 56% rad14Δ rad51Δ diploids were large-budded cells with a single nucleus at the bud neck, which is the characteristic G2/M phase morphology, and only 7% cells were in the S phase (Fig. 4b and Supplementary Fig. 1). The remaining 37% rad14Δ rad51Δ cells exhibited an abnormal morphology characterized by protruded and/or multiple buds (Fig. 4b and Supplementary Fig. 1). In contrast, upon CLUV exposure, the percentage of large-budded cells increased in both rad14Δ and rad51Δ diploids compared to that in the wild-type cells, but abnormal cells were rare (Fig. 4b). In addition, the viability of rad14Δ rad51Δ diploids significantly decreased 24 h after initiating CLUV treatment (Fig. 4c). Similar behaviors were observed in each of the haploid strains (Supplementary Fig. 2, a and b). Taken together, these results suggest that HR is essential for CLUV tolerance in NER-deficient haploids/diploids.
CLUV-induced cell-cycle arrest and abnormal morphology in rad14Δ rad51Δ cells. a) Growth curves of wild-type (circle), rad14Δ (square), rad51Δ (triangle), and rad14Δ rad51Δ (diamond) cells in the presence of CLUV irradiation. Haploid (left panel) and diploid (right panel) cells were grown under CLUV exposure and the plating efficiency was determined using samples collected every 4 h. The number of cells is represented as relative colony-forming unit (= 1 at time 0). SEM (n = 3) is indicated. b) Cells were incubated in the presence or absence of CLUV for 8 h, and at least 100 cells per each strain were counted under a microscope. The percentages of diploid cells with no bud (G1), small bud (S), large-bud (G2/M), and protruded or multiple buds (others) were scored. c) Cell viability of diploid strains was calculated as the number of viable cells (colony-forming units) divided by the total number of cells. Error bars indicate the standard error (SE) of 4 independent experiments. *P < 0.01, **P < 0.001; n.s.: not significant (P > 0.05).
CLUV causes irreparable ssDNA accumulation in rad14Δ rad51Δ cells
If HR is essential for repairing ssDNA gaps generated in CLUV-exposed rad14Δ cells, HR defects may lead to the accumulation of unrepaired ssDNA, which might manifest as higher levels of RPA foci and affect their distributions. Therefore, we investigated the impact of ssDNA accumulation on subcellular RPA foci localization in rad14Δ rad51Δ diploids. Diploid cells in the early log phase were CLUV-irradiated for 8 h and then subjected to fluorescence microscopy. We found that spontaneous RPA foci increased in cells lacking Rad51 but not Rad14 (Fig. 5, a and b), consistent with previous studies showing that HR defects significantly increase the frequency of spontaneous genomic instability (Chen and Kolodner 1999; Huang et al. 2002). Under CLUV exposure, RPA foci increased in rad14Δ and rad51Δ cells but not in wild-type cells (Fig. 5, a and b); 30% rad14Δ cells and 56% rad51Δ cells had RPA foci and most contained 1 or 2 RPA foci (Fig. 5b). In contrast, CLUV-exposed rad14Δ rad51Δ diploids yielded strikingly different fluorescence images from their single mutants. Almost rad14Δ rad51Δ cells (93%) contained RPA foci, half of which included multiple dispersed foci (Fig. 5, a and b). These characteristics were also observed in haploid rad14Δ rad51Δ cells (Supplementary Fig. 2c).
CLUV-induced RPA foci. The indicated diploid strains were grown for 8 h at 30°C in the presence or absence of CLUV irradiation. The samples were analyzed for the presence of RPA foci using fluorescence microscopy. Representative images are shown in (a). Multiple foci-containing cells are indicated by arrowheads. Scale bar, 10 µm. b) The percentages of cells with RPA foci were scored; white and black portions of bars depict the percentage of cells with single or 2 and multiple (≥3) RPA foci, respectively. The results represent the average of at least 3 independent measurements. Error bars indicate the standard error (SE) of 3 independent experiments.
To further investigate whether cells with multiple RPA foci can proceed to the next cell cycle, we used haploid cells to synchronize CLUV-exposed cells in G1. Haploid cells were cultured for 8 h under CLUV exposure and then treated with α-factor for 2 h in the absence of CLUV to synchronize cells in the G1 phase. After treatment with α-factor, most wild-type, rad14Δ, and rad51Δ cells were arrested in the G1 phase, as revealed by a single 1C peak in FACS analysis and their shmoo-like morphology (Fig. 6, a–c), indicating that most NER- and HR-deficient cells can be transmitted through mitosis into the next cell cycle regardless of whether they have RPA foci or not. In contrast, the number of G1-arrested rad14Δ rad51Δ cells was less than 37% (Fig. 6c), and the percentage of cells with multiple RPA foci (56%) was higher after α-factor treatment than that before α-factor treatment (38%) (Figs. 5b and 6d). We note that the broad 2C peak was evident in the CLUV-exposed rad14Δ rad51Δ cells. This may be due to the multibudding yeast cells, which may be accompanied by erroneous mitotic exit and abnormal DNA synthesis. In addition, the level of the phosphorylated Rad53 isoform, a DDC activation marker, was slightly higher in rad14Δ and rad51Δ cells than in wild-type cells during CLUV treatment and was greatly elevated in rad14Δ rad51Δ cells (Fig. 6e). Taken together, these results demonstrate that multiple RPA foci in CLUV-exposed rad14Δ rad51Δ cells reflect increased levels of irreparable ssDNA gaps, possibly due to impaired DNA replication recovery upon CLUV exposure.
Multiple RPA foci reflect increased levels of irreparable ssDNA gaps. a) The indicated haploid strains were grown under CLUV irradiation for 8 h and then treated with α-factor in the absence of CLUV irradiation to arrest the cell cycle at the G1 phase. The samples taken before or after α-factor treatment were subjected to FACS analysis. b) The same experiment was conducted as in (a). The samples taken before or after treatment with α-factor were analyzed for the presence of RPA foci using fluorescence microscopy. Multiple foci-containing cells are indicated by arrowheads. Scale bar, 10 µm. Samples taken after treatment with α-factor were analyzed for c) cell-cycle distribution or d) the presence of RPA foci using fluorescence microscopy. At least 100 cells per each strain were counted. Error bars indicate the standard error (SE) of 3 independent experiments. e) The haploid strains were cultured under CLUV exposure for the indicated time periods. The protein extracts were prepared and separated by 6% SDS-PAGE, followed by western blotting with anti-Rad53 antibody.
Interhomolog recombination is elevated in CLUV-exposed rad14Δ diploids
These data demonstrate that HR-dependent repair plays a key role in CLUV tolerance and genome stability in CLUV-exposed rad14Δ cells. One explanation for the higher CLUV tolerance in diploids is that interhomolog HR events prevent ssDNA gap accumulation. To determine whether interhomolog HR increases in diploid rad14Δ cells during CLUV exposure, we assessed the frequency of mitotic recombination events between chromosome homologs. A tester strain containing CAN1/HIS3 heteroalleles at the CAN1 locus on the right arm of chromosome V and HOM3/URA3 heteroalleles at the HOM3 locus on its left arm was used (Fig. 7a). Diploid rad14Δ cells were grown for 24 h under CLUV exposure and then plated on a medium containing canavanine. Canavanine-resistant colonies were then replica-plated on a complete medium lacking uracil to distinguish mitotic recombination from chromosome loss (Fig. 7a (1)). The CanR Ura+ colonies were further analyzed using colony PCR at the can1 allele to distinguish between mitotic conversion and spontaneous mutation. In this experiment, the presence of the CAN1 fragment indicates can1 mutation, whereas the loss of the CAN1 fragment indicates LOH through interhomolog gene conversion or allelic crossover (Fig. 7a (2-a, 2-b)). The CLUV-induced CanR frequency in rad14Δ diploids showed 10-fold increase compared to that in the absence of CLUV (Fig. 7b). Most CanR cells (96%) developed due to LOH, and the rest (4%) developed due to mutation events (Fig. 7c). Thus, CLUV-exposed rad14Δ diploids significantly increased the rates of interhomolog HR events.
Interhomolog recombination is elevated in CLUV-exposed rad14Δ diploids. a) Schematic representation of the chromosome V homolog used to monitor interhomolog recombination. CanR colonies were classified into Ura+ or Ura−. (1) Loss of chromosome V containing URA3 and CAN1 produced a Ura- colony. (2-a) Gene conversion between HIS3/CAN1 heteroalleles or (2-b) CAN1 marker mutation produced an Ura+ colony. These events can be further distinguished by PCR analysis of the HIS3/CAN1 locus. b) The indicated strains were grown for 1 day at 30°C in the presence or absence of CLUV. The CanR mutation frequency was calculated. Error bars indicate the standard error (SE) of 4 independent experiments. c) The percentage of CanR clones with chromosome loss, mutation, or LOH. The percentage of CanR Ura- clones was calculated by subtracting the frequency of CanR Ura+ clones from that of CanR cells. Chromosomal DNA from the obtained CanR Ura+ mutants was analyzed to distinguish between CAN1 mutation and LOH using PCR. At least 100 CanR colonies of each strain were analyzed. L and S depict large and small CanR colonies, respectively.
We also examined the effect of RAD51 deletion on the CanR frequency of the tester strains. CanR frequency increased in diploid rad51Δ cells, even in the absence of CLUV irradiation (Fig. 7b), possibly due to the defects in repairing endogenous DNA damage as described above. CLUV irradiation significantly increased CanR frequency in rad14Δ rad51Δ cells, but not in rad51Δ cells (Fig. 7b). Interestingly, we found that CanR colonies of rad14Δ rad51Δ cells were heterogeneous in size: 20% were large and 80% were small (Fig. 7c). Most large colonies were classified as can1 mutants, whereas approximately 40% small colonies were classified as developed due to chromosome loss, and the remaining were classified as developed due to LOH (Fig. 7c). Since all LOH cells derived from the rad14Δ population formed large colonies, rad14Δ rad51Δ LOH cells might have some kind of chromosome instability, such as large-scale chromosome deletions and/or chromosome rearrangements, which is associated with reduced proliferation capacity. It should be noted that the CanR frequency in CLUV-exposed rad14Δ rad51Δ cells may have been underestimated as rad14Δ rad51Δ cells display severe growth defects and lose viability upon CLUV treatment. Taken together, these results suggest that CLUV-induced replication stress in NER-deficient cells cause large-scale chromosome rearrangement and chromosome loss due to irreparable ssDNA accumulation when the HR pathway is impeded.
Discussion
This study showed that haploid rad14Δ populations rapidly converge toward diploidy under CLUV exposure. Previous experimental studies have demonstrated that changes in ploidy often result in fitness benefits under certain stress conditions (Yona et al. 2012; Harrison et al. 2014; Storchova 2014; Selmecki et al. 2015; Voordeckers et al. 2015). Accordingly, the conversion of haploid rad14Δ cells to diploid cells promotes survival in response to CLUV irradiation. Notably, we showed that the amount of CPD per microgram DNA is the same in haploid and diploid cells, which means that the diploid has more CPD per cell. Nevertheless, the accumulation of RPA-coated ssDNA, possibly due to the replication of damaged template DNA, is reduced in diploids. Thus, our results demonstrated that the numbers of ssDNA, but not UV-induced photolesions, affect the diploid-dependent fitness for CLUV irradiation. These data imply that the CLUV resistance of NER-deficient cells involves their ability to repair the postreplicative ssDNA gaps rather than direct repair of the CPD itself.
RPA binds/protects ssDNA and RPA-coated ssDNA is an important intermediate for recruiting various DNA damage response factors, including DNA repair, tolerance, and checkpoint signaling. Our data showed that the RPA foci count increased in CLUV-exposed rad14Δ cells. Interestingly, most CLUV-exposed rad14Δ cells proceeded to the next cell cycle despite increased RPA foci levels, and had poor Rad53 phosphorylation, suggesting that CLUV-induced ssDNA in rad14Δ cells is insufficient to inhibit the cell-cycle progression. In this context, a recent report has demonstrated that most RPA-coated ssDNA accumulation during replication of damaged DNA template does not occur at stalled replication forks, but in daughter-strand gap regions behind the replication forks, defined as the postreplication territory (Wong et al. 2020). DNA fiber assay in humans has revealed that UV-blocked replication forks effectively restart through re-priming past the lesion, leaving only a small gap opposite the lesion (Quinet et al. 2016). Furthermore, the previous studies have shown that HR is stimulated by UV irradiation (Rupp et al. 1971; Kadyk and Hartwell 1993; Fasullo and Sun 2008; Covo et al. 2012; St Charles et al. 2012; Yin and Petes 2015). Therefore, we speculate that the increased number of fork-blocking lesions in CLUV-exposed NER-deficient cells would increase replication stress, thus forming postreplication ssDNA gaps instead of blocking replication forks. Under such conditions, these ssDNA gaps would not interfere with the overall genome replication progression and could be efficiently rescued by HR and DDT, both of which might contribute to the suppression of DDC activation. Consistent with this, our data revealed that rad14Δ cells lacking Rad51 recombinase exhibited G2 arrest and DDC activation upon CLUV exposure. Notably, several rad14Δ rad51Δ cells contained multiple RPA foci concomitant with abnormal cell morphology, which led to viability loss. These results indicate that multiple RPA foci reflect a higher number of un-resolved tracts of ssDNA. Thus, HR-dependent repair of the postreplicative ssDNA gaps plays a critical role in CLUV resistance of NER-deficient cells.
We showed here that CLUV-induced conversion of rad14Δ haploids toward diploids requires HR, and the diploid rad14Δ cells are more CLUV resistant than haploid cells. These results imply that HR may repair the ssDNA gaps more efficiently in diploids than in haploids. Although how HR is enhanced in diploids is unclear, 1 explanation for these results is that the extra chromosomal copies could serve as HR partners to improve the efficiency of damage tolerance. In diploid cells, sister chromatids are generally the preferred templates for HR-mediated recovery from replication stress, but homologous chromosomes can also be used to restore broken replication forks and/or repair ssDNA gaps (Carotenuto and Liberi 2010; Moynahan and Jasin 2010; Keyamura et al. 2016). Consistent with this, the exposure of rad14Δ diploids to CLUV markedly increased the rates of interhomolog HR events. Thus, it is likely that an increased interhomolog HR would contribute to the recovery from replication stress by providing additional routes for repairing ssDNA gaps left behind the replication fork in CLUV-exposed NER-deficient cells. While the above model represents the simplest explanation for the increased HR activity in diploids, it may be possible to improve CLUV resistance more efficiently using sister templates by somehow regulating HR activity for lesion bypass in a diploid-specific manner.
This study established that HR provides a growth advantage during DNA replication through unrepaired lesions, in which efficient ssDNA repair through interhomolog HR might contribute to the suppression of DDC activation and higher CLUV resistance to diploid cells than to haploid cells. However, using homologous chromosomes as repair templates have important consequences for genetic stability; LOH is a frequent outcome under CLUV exposure when the NER mechanism is compromised. Taken together, we conclude that although HR-dependent stress tolerance is initially beneficial to cell proliferation by reducing potentially lethal ssDNA accumulation and suppressing the DDC activation, it is also linked with the resultant genomic instability when lesions overwhelm cells. Although this study used CLUV-exposed NER-deficient cells as an HR function model to overcome replication problems caused by UV-induced lesions, it will improve our understanding of the mechanisms of the complementary relationship between DNA repair, DDT, and HR during chronic replication stress. Therefore, a similar argument that HR functions as a double-edged sword for replication stress tolerance can be made for the development of any cancer associated with defects in DNA repair pathways, which may have important implications for tumorigenesis and cancer treatment.
Data availability
Strains and plasmids are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article and figures.
Supplemental material is available at GENETICS online.
Acknowledgments
We thank members of the Hishida laboratory for discussions and technical assistance.
Funding
This work was supported by JSPS KAKENHI Grant Numbers JP17K07290 and JP23114007 (to TH).
Conflicts of interest
The authors declare that there is no conflict of interest.
