The protein Swi6 in Saccharomyces cerevisiae is a cofactor in two complexes that regulate the transcription of the genes controlling the G1/S transition. It also ensures proper oxidative and cell wall stress responses. Previously, we found that Swi6 was crucial for the survival of genotoxic stress. Here, we show that a lack of Swi6 causes replication stress leading to double-strand break (DSB) formation, inefficient DNA repair and DNA content alterations, resulting in high cell mortality. Comparative genome hybridization experiments revealed that there was a random genome rearrangement in swi6Δ cells, whereas in diploid swi6Δ/swi6Δ cells, chromosome V is duplicated. SWI4 and PAB1, which are located on chromosome V and are known multicopy suppressors of swi6Δ phenotypes, partially reverse swi6Δ genome instability when overexpressed. Another gene on chromosome V, RAD51, also supports swi6Δ survival, but at a high cost; Rad51–dependent illegitimate recombination in swi6Δ cells appears to connect DSBs, leading to genome rearrangement and preventing cell death.
The multifunctional Swi6 protein in Saccharomyces cerevisiae is best known for its role in cell cycle regulation. It participates in the transcriptional control of over 100 genes necessary for the G1/S cell cycle transition (Koch et al., 1993; Iyer et al., 2001; Bean et al., 2005). Swi6 acts as a transcription cofactor in two different complexes regulating the expression of these genes. In the MBF complex (together with the Mbo1 protein), it controls the transcription of genes whose promoters contain the MluI cell cycle box. In the SBF complex, Swi6 cooperates with the Swi4 protein, which binds to promoters containing Swi4-binding cell cycle boxes (Moll et al., 1992). Swi6 shuttles between the nucleus and cytoplasm depending on its phosphorylation status (Sidorova et al., 1995; Sidorova and Breeden, 1997; Kim et al., 2010). Phosphorylated Swi6 is exported from the nucleus in an Msn5-dependent manner, while its import depends on Ess1, which facilitates its Cdc14-dependent dephosphorylation (Queralt and Igual, 2003; Atencio et al., 2014).
In addition to its regulation of genes required for the progression of the S phase, such as DNA replication or repair genes, Swi6 controls growth rate-dependent genes and contributes to the cellular response to various stresses (Fazio et al., 2008). swi6Δ cells do not display G1 delay upon oxidative stress provoked by lipid hydroperoxide, likely due to the downregulation of dozens of genes whose expression in wild-type (WT) cells is elevated under such treatment (Fong et al., 2008). Swi6 not only regulates the transcriptional response to oxidative stress but also detects it using a unique cysteine residue as a redox sensor (Chiu et al., 2011). Swi6 is also involved in the unfolded protein response and in the cell wall integrity pathway (Scrimale et al., 2009; Baetz et al., 2001; García et al., 2004; Kim et al., 2010).
Our previous study identified SWI6 among genes linked to oversensitivity to radiomimetic zeocin [i.e. genes important for surviving double–strand break (DSB) stress] (Krol et al., 2015). Here, we show that cells lacking SWI6 display various genome instability phenotypes: increased sensitivity to genotoxic agents, high levels of DNA damage upon exposure to zeocin and ineffective repair of this damage, and the accumulation of genome rearrangements leading to aberrant DNA content. In haploid swi6Δ cells, DNA rearrangements occur randomly, whereas in homozygous diploid swi6Δ/swi6Δ cells, chromosome V is duplicated. Moreover, regardless of ploidy, SWI6 deletion is accompanied by increased SWI4 and RAD51 transcription. We also found that the genome instability caused by SWI6 deletion could be suppressed by the overexpression of SWI4. We believe that the increased DNA rearrangement in cells lacking SWI6 is due to the accumulation of the recombinase Rad51 [a single-stranded (ss)DNA-binding protein responsible for homology searching during recombination (Sung, 1994)] and the diminished level of Srs2 (an Rad51 translocase), which together redirect DSB repair to an illegitimate recombination subpathway. This possibility is supported by the lower frequency of point mutations in the swi6Δ mutant than in the wild type (WT), likely due to Rev7 decay (Gong and Siede, 2009). Rev7 is a regulatory subunit of DNA polymerase ζ, which is responsible for most of the spontaneous mutations in haploid strains (Northam et al., 2010; Stone et al., 2012). Thus, the high genome instability of swi6Δ is likely due to the switching of DNA repair pathways during replication stress from Rev7-dependent translesion synthesis (TLS) to a damage avoidance pathway that relies on recombination. Since Srs2 is depleted and Rad51 is overexpressed in these mutants, the cells handle elevated levels of DSBs with the available resources, that is, the deleterious Rad51-dependent route of homologous recombination, causing genome instability.
Genomic data suggest a protective role of Swi6 against DSB stress
Zeocin is a radiomimetic bleomycin that is widely used in various anti-cancer therapies. As a chemoenzyme, zeocin intercalates into double-stranded nucleic acid structures and performs a redox reaction with a metal ion (Cu2+ or Fe3+), O2 and a one-electron reducer as cofactors, resulting in a variety of cleavage products that convert rapidly into DSBs (Sleigh, 1976; Chen et al., 2008). Zeocin, similar to other bleomycins, is not very specific and can introduce lesions into a broad range of macromolecules. It can perturb lipids, destroy proteins, and impair protein synthesis via the cleavage of selected RNA. The most frequent damage caused by zeocin is DSBs (Abraham et al., 2003; Povirk et al., 1979). We recently performed a genome-wide study seeking the cellular mechanisms responsible for survival of the genotoxic stress caused by zeocin treatment (Krol et al., 2015).
Among homozygous diploid yeast knockouts, we identified 133 strains with oversensitivity to zeocin, including swi6Δ/swi6Δ (Krol et al., 2015). This strain is particularly interesting because it belongs to a very limited group of knockout strains with high sensitivity to DSBs that are induced both chemically and by the in vivo overexpression of homing endonucleases (A.S., M.S., K.K., I.R., unpublished data). This group also includes strains lacking XRS2 or RAD52, whose products are crucial in DSB repair. The Xrs2 protein subunit of the MRX complex participates in DSB recognition, stabilization and resection, and Rad52 is a recombination mediator (Maher et al., 2011). One of our previous whole-genome screens also identified the swi6Δ/swi6Δ strain as a spontaneous mutator (Alabrudzinska et al., 2011), indicating an important role of Swi6 in maintaining genome stability not only under genotoxic stress but also during unperturbed cell growth.
SWI6 is essential for normal cell growth
The features of swi6Δ mutants include limited growth, long doubling time and high frequency of spontaneous suppressor mutations. The slope of the exponential phase growth curve of swi6Δ and swi6Δ/swi6Δ mutants was much less steep than that of the WT (Fig. 1A), corresponding to an almost tripled doubling time and entry to the stationary growth phase at lower cell densities (Fig. 1A). These parameters were consistent with the decreased viability of the swi6Δ mutant. The number of dead cells in the swi6Δ cell population was higher than in the control and increased with time (Fig. 1B). Moreover, swi6Δ cells have abnormally elongated morphology and larger volume than the cells of isogenic WT strains, and thus, optical density may be a misleading measure of cell growth (compare panels A and C in Fig. 1).
The progressive phenotypic changes in swi6Δ strains from the yeast knockout collection (Giaever et al., 2002) suggests that they acquire additional mutations. To verify this behavior, we constructed such a strain de novo by gene replacement, introducing a disruption cassette into the WT strain. However, when we attempted to construct these strains in the presence of a plasmid carrying an original copy of SWI6, we noted that homozygous diploid swi6Δ/swi6Δ cells did not lose that plasmid, even after 10 passages, while 35% of heterozygous diploid SWI6/swi6Δ cells lost the plasmid under these conditions (Fig. 1D). The inability to lose the functional SWI6 gene provided on the episomal plasmid from swi6Δ cells was compatible with the tetrad analysis of heterozygous diploid SWI6/swi6Δ (Fig. 1E). We found that typically no more than two spores from each tetrad were viable, which implies that SWI6 is an essential gene. Unexpectedly, one of the resulting haploids (7D) showed G-418 resistance due to the swi6::kanMX4 allele presence in the SWI6 genomic loci. This strain displayed slowest growth and atypical DNA content, as revealed by flow cytometry (Fig. 1E, right panel).
We also noticed that the sporulation efficiency depended on the SWI6 copy number. In SWI6/swi6Δ cells, sporulation decreased significantly (15-fold) compared to that of the WT diploid, whereas the swi6Δ/swi6Δ strain did not sporulate at all (Fig. 1F). The presence of SWI6 on the plasmid increased the sporulation efficiency of the SWI6/swi6Δ strain 10-fold and enabled a low level of sporulation in swi6Δ/swi6Δ (3% of WT level). On the other hand, overexpression of SWI6 in the WT strain resulted in a reduction of tetrad formation (Fig. 1F). Thus, we conclude that the copy number of SWI6 per cell influences the efficiency of sporulation.
Seemingly confusing results concerning SWI6 gene essentiality, namely, the viability of the swi6Δ deletion strain obtained by gene replacement methodology versus the very limited ability of dissected spores to sprout into swi6Δ haploids suggests that SWI6 is an ‘evolvable essential gene’. This new scientific terminology was introduced this year (Rancati et al., 2018); however, the phenomenon itself has been frequently observed previously. As Rancati et al. state, “We define evolvable essential genes as essential genes that can be acutely removed from the genome without causing stereotypical cell death; instead, a subset of cells with these genes deleted can undergo short-term adaptive evolution and spontaneously acquire compensatory mutations that suppress the lethal phenotype.” Moreover, the authors proposed five ‘context-dependent gene essentiality’ categories, among them a ‘karyotype-dependent essentiality’, which to our knowledge best fits SWI6. This category encompasses genes that, similar to SWI6, are essential for the euploid cell but nonessential for aneuploid cells in which the suppressor gene is overexpressed. Our data presented later in this paper suggest that RAD51 is a multicopy suppressor of the SWI6 mutation.
Further data suggest that SWI6 should be tightly regulated, and its copy number might not only influence sporulation efficiency between homozygous and heterozygous diploid mutants (Fig. 1F) but also influence the zeocin sensitivity of WT strains bearing an additional episomal copy of SWI6 (Fig. 2A). Moreover, the introduction of an episomal copy of SWI6 into a swi6Δ or swi6Δ/swi6Δ mutant does not fully restore the WT phenotype, which suggests that the mutant strain quickly acquires additional mutations that SWI6 cannot complement (Fig. 2A). Thus, a lack of SWI6 leads to immediate genome destabilization.
Strains lacking SWI6 display genome instability phenotypes
To demonstrate that SWI6 deletion strains are genetically unstable, we performed a series of experiments. First, we tested the sensitivity of swi6Δ and swi6Δ/swi6Δ strains to various genotoxic agents, namely, 100 mM hydroxyurea, 0.02% methyl methanesulfonate, 10 µg/ml camptothecin, and 2.5 and 5 µg/ml zeocin. As shown in Fig. 2A, both haploid and diploid mutants were oversensitive to all genotoxic compounds tested. This sensitivity was somewhat complemented by episomally encoded SWI6. Moreover, an additional copy of SWI6 in the WT strain increased its sensitivity to genotoxic agents (Fig. 2A), suggesting that to survive genotoxic stress, cells must tightly control the level of Swi6.
Next, we tested the level of spontaneous mutagenesis in SWI6-deleted strains in a forward mutation assay using two mutagenesis markers, CAN1 and URA3. We constructed de novo strains bearing appropriate mutagenesis markers (i.e. CAN1 and URA3) for swi6Δ haploids, and the heterozygous mutagenesis markers CAN1/can1Δ and URA3/ura3Δ for the swi6Δ/swi6Δ diploid. As expected based on genomic screen data (Alabrudzinska et al., 2011), the frequency of canavanine-resistant (CanR) clones was almost 10-fold higher, and the frequency of 5-fluoroorotic-acid-resistant (FOAR) clones was almost 30-fold higher in swi6Δ/swi6Δ than in SWI6/SWI6 strains (Fig. 2B). However, unexpectedly, the haploid swi6Δ strain displayed the opposite effect; it behaved as an anti-mutator (Fig. 2B). Our further experiments investigated this phenomenon. Please also note the two orders of magnitude higher number of mutations in diploid than in haploid WT strains, which reflects the differences in the mutation-generating mechanisms: point mutations in haploids versus the 100-fold more frequent DNA rearrangements leading to loss of heterozygosity (LOH) in diploids (Ohnishi et al., 2004; Hiraoka et al., 2000).
Strains lacking SWI6 accumulate DSBs
The puzzling results of the spontaneous mutagenesis assay prompted us to test whether the DNA repair abilities of the swi6Δ strains were unaffected. As both haploid and homodiploid swi6Δ strains responded to genotoxic treatment with increased mortality, we performed a DNA damage recovery test to evaluate the nuclear genome integrity after genotoxic treatment via analyzing chromosomes separated by pulse field gel electrophoresis (PFGE). In this assay, we separated chromosomes from untreated cells, from cells subjected to zeocin treatment and from zeocin-treated cells that were further grown for 3 h to allow DNA damage repair. Because zeocin exerts its strongest effect on the DNA of exponentially growing cells, such cells were used in all experiments and were further synchronized by nocodazole to facilitate chromosome detection by PFGE. As shown in Fig. 2C, unlike the control strains, neither swi6Δ nor swi6Δ/swi6Δ strains could rebuild zeocin-damaged chromosomes during recovery.
To gain insight into aberrant DNA structures developed in chromosomes during DSB repair, which might accumulate with hampered repair, we performed a chromosome comet assay (Lewinska et al., 2014; Adamczyk et al., 2016; Krol et al., 2017). We analyzed a band representing two chromosomes, VII and XV, cut from the gel after PFGE. All types of DNA structures were more frequent in the mutant than in the control strains. In the WT samples, short or long linear DNA strands predominated, whereas samples from swi6Δ mutants contained structures representing gapped DNA strands, circular DNA, Y-type structures and various branched DNA structures. We also found differences between swi6Δ- and swi6Δ/swi6Δ-derived samples (Fig. 2D). Both mutants accumulated gapped DNA strands and short DNA fragments, suggesting the formation of DSBs, but the repair intermediates differed. The swi6Δ-derived samples contained small DNA circles and Ψ-type structures reflecting fork regression byproducts or fork collapse (Keyamura et al., 2016; Nguyen et al., 2017). The swi6Δ/swi6Δ-derived samples contained large DNA circles and elaborated branched DNA structures, which are possible recombination intermediates resulting from multi-invasion-induced rearrangement (Piazza et al., 2017; Nguyen et al., 2017).
To characterize the DNA lesions, we analyzed the number of repair foci formed by Rad52–YFP and Rfa1–YFP in parallel. Both fusion proteins are recruited to DNA lesions and thus can be used to monitor their frequency (Lisby et al., 2001). Rad52 binds DNA lesions processed by homologous recombination, and Rfa1 binds preferentially to ssDNA (Lisby et al., 2004). We found that the number of spontaneously formed Rad52 foci was higher in the swi6Δ and swi6Δ/swi6Δ mutants than in the WT but, in contrast to the WT, did not increase after zeocin treatment (Fig. 2E,F). Notably, the frequency of Rad52 foci in the swi6Δ and swi6Δ/swi6Δ mutants was similar to their frequency in the zeocin-treated WT. In all strains, the number of Rfa1 foci increased upon zeocin treatment, but in diploid strains, the difference between untreated and treated cells was more pronounced (Fig. 2G). While the number of repair foci differed between strains, the number of foci in individual cells with foci did not significantly change even after zeocin treatment (data not shown).
Lack of SWI6 affects nuclear DNA content
DSBs are very toxic DNA lesions because they can lead to aberrant and unfavorable DNA structures, such as recombination intermediates. Moreover, especially in cases of DSB stress, these lesions can lead to DNA rearrangement (Malkova and Haber, 2012; Skoneczna et al., 2015). The elevated frequency of Rad52 foci in swi6Δ strains, which reflects DSB accumulation, suggested that these strains might also accumulate DNA rearrangements, which should be reflected in DNA content changes. We verified this idea by analyzing the genomic DNA of the respective strains by flow cytometry, which can detect large DNA rearrangements in the genome as shifts in the flow cytometry profiles. As shown in Fig. 3A, compared to WT cells, both swi6Δ and swi6Δ/swi6Δ mutants displayed aberrant DNA content, such as aneuploidization and polyploidization, which were reversed by introducing SWI6 on a plasmid (Fig. 3A, right panel; see Fig. S1 for details). To verify the flow cytometry results, we analyzed 4′,6-diamidino-2-phenylindole (DAPI)-stained nuclear DNA by fluorescence microscopy. We found swi6Δ deficiency to be linked to various anomalies in nuclear morphology and to changes in the number of nuclei per cell (Fig. 3B,C).
Since swi6Δ mutants undergo aneuploidization, we asked whether the copy number of any specific DNA fragment(s) increased in all cells or whether the DNA amplification was random. We performed comparative genome hybridization (CGH), which can detect an increase or decrease in the copy number of any region of the genome common to the whole population of cells of the analyzed strain with respect to that of the control strain. As shown in Fig. 3D, no variations in the copy number of any genome region occurred in haploid swi6Δ cells, so aneuploidization in this strain must have resulted exclusively from random genome amplification. In contrast, in diploid swi6Δ/swi6Δ, we detected full duplication of chromosome V. Interestingly, the level of mitochondrial DNA in swi6Δ was almost twofold higher than in the haploid WT. Thus, the genomic changes shaping the haploid and diploid genome of swi6Δ mutants are distinct.
Chromosome V bears genes that, when overexpressed, can suppress swi6Δ phenotypes
The duplication of chromosome V in swi6Δ/swi6Δ cells suggested that this chromosome might contain genes that suppress swi6Δ phenotypes when duplicated. Indeed, two such genes, PAB1, encoding poly(A)-binding protein, which is involved in the control of poly(A) tail length and mRNA export, and SWI4, encoding the DNA-binding subunit of the SBF complex, were described previously as multicopy suppressors of various swi6 mutations (Amrani et al., 1997; Dunn et al., 2005; Brune, 2005; Sidorova and Breeden, 1993; Breeden and Mikesell, 1991). Episomally expressed PAB1 is a dosage suppressor of the bck2Δ swi6-ts mutant (Flick and Wittenberg, 2005), whereas overexpression of SWI4 suppresses the SWI6-dependency of HO transcription in vivo (Sidorova and Breeden, 1993). We tested whether these genes might also suppress the genome instability of swi6Δ strains. As shown in Fig. 4A, both PAB1 and SWI4 supported the growth of swi6Δ mutants, but only SWI4 supported the growth of swi6Δ/swi6Δ cells under genotoxic stress. Moreover, SWI4 alone, but not PAB1, suppressed the genomic rearrangements that led to DNA content anomalies in SWI6-deleted strains (Fig. 4B; see Fig. S2 for details). Therefore, Swi4 and Pab1 are partial suppressors of swi6Δ and can replace only some Swi6 functions.
Accumulation of Rad51 is likely responsible for DNA rearrangements affecting the genome of swi6Δ mutants
Since the additional copy of chromosome V influenced phenotypes associated with SWI6 deletion, we analyzed other genes located on this chromosome and noted several genes involved in various DNA repair pathways. RAD51 drew our particular attention because it encodes a recombinase that forms a helical filament with ssDNA, called an invasive strand, which is responsible for homology searches during recombinational repair of DSBs (Sung, 1994). We assumed that increased DNA rearrangement could be triggered by increased Rad51 levels, so we verified the level of Rad51 by western blot analysis. As shown in Fig. 5A, the Rad51 protein level was significantly increased in swi6Δ/swi6Δ cells. Interestingly, the haploid swi6Δ mutant, with no additional copy of chromosome V, exhibited similar Rad51 overproduction (Fig. 5A). Thus, the increased Rad51 level cannot be attributed only to the RAD51 copy number.
In the next step, we verified the expression level of RAD51 in mutant cells by quantitative real-time PCR (qPCR). We found that the level of RAD51 mRNA increased 6-fold in swi6Δ and 10-fold in swi6Δ/swi6Δ compared to the respective WT controls (Fig. 5B). Thus, duplication of the RAD51-bearing chromosome was not the only cause of genome instability suppression in the swi6Δ/swi6Δ mutant. Chromosome V duplication is probably beneficial, but increased expression of the RAD51 gene also exerted suppression.
To exclude the possibility that the phenotype of the swi6Δ/swi6Δ mutant did not result from chromosome V duplication, we constructed a BY4743 strain bearing chromosome V trisomy (BY4743 chr.V). We crossed strain BY4742 with the BY4741-derivative strain with documented chromosome V disomy [i.e. the laboratory-evolved strain evo-High-pH described in Yona et al. (2012)]. In Fig. S3A–D, we show that chromosome V trisomy itself does not significantly influence the genome stability phenotype. The sensitivity to zeocin of the diploid with chromosome V trisomy was the same as that of the control BY4743 (Fig. S3D). The frequency of spontaneous and zeocin-induced Rad52 foci was even lower (by 2-fold) than in the diploid BY4743 (Fig. S3C). The strain with the chromosome V trisomy displayed a DNA content shift (see flow cytometry histogram in Fig. 3B), which suggests aneuploidy, but it did not display polyploidy, which is typical for the swi6Δ/swi6Δ strain. In the DNA content histogram of the BY4743 chr.V strain, there are no additional peaks typical for a euploid with a DNA content higher than 4C. In contrast to the swi6Δ/swi6Δ mutant, in which multinuclear cells were very common (Fig. 2B,C), chromosome V trisomy alone did not lead to the formation of multinuclear cells, as revealed by microscopic observation of DAPI-stained cells (Fig. S3A). Thus, the lack of SWI6, but not the additional copy of chromosome V, was responsible for the genome instability phenotype of the swi6Δ/swi6Δ mutant.
Next, we asked whether SWI4 acts as a swi6Δ suppressor by increasing the expression of RAD51. Swi4 acts as a transcription factor in concert with Swi6 but also regulates the expression of some genes alone (Sidorova and Breeden, 1993; MacIsaac et al., 2006). If RAD51 expression is Swi4 dependent, the suppression of Swi4 in swi6Δ mutants may rely on this relationship. As shown in Fig. 5B, the level of SWI4 mRNA was 4–5 times higher in both haploid and diploid swi6Δ mutants than in the respective WT strains. However, in swi4Δ, the Rad51 protein level was only slightly lower than in the WT (Fig. 5A), and the strong overexpression of SWI4 [from the GAL1 promoter (Sidorova and Breeden 1993)] hardly influenced the Rad51 level in the control strain bearing empty plasmid (Rose et al., 1987) (Fig. 5C). Thus, Swi4-regulated RAD51 transcription was not substantiated.
Analyzing the effect of RAD51 overexpression on swi6Δ mutant phenotypes showed that RAD51 supported the growth of mutant cells challenged with zeocin (Fig. 5D). RAD51 overexpression restored the genomic DNA flow cytometry profile of swi6Δ/swi6Δ to the WT profile and slightly improved the genomic DNA flow cytometry profile of swi6Δ cells (Fig. 5E; see Fig. S4 for details). Moreover, we determined that swi6Δ mutant cells could lose the pRS415-SWI6 plasmid in the presence of the pAT5 [ADH1pRAD51] plasmid (Tovmasyan 2014) (Fig. 5F) but not when pRS415-SWI6 was accompanied by an empty vector (Mumberg et al., 1995). All these data support the conclusion that RAD51 overexpression is crucial for the survival of swi6Δ strains.
Decreased levels of Srs2, the Rad51 translocase, contribute to genome rearrangements in swi6Δ strains
Increased Rad51 levels can be especially mutagenic when the Rad51 filament forms without proper control (Esta et al., 2013). The protein that limits its formation on DNA is the translocase Srs2 (Krejci et al., 2003; Burgess et al., 2009). Since recombination control in swi6Δ mutants seemed not to be tight, we tested the level of Srs2 in these cells. Indeed, Fig. 5G shows that the level of Srs2 protein was decreased in swi6Δ and swi6Δ/swi6Δ mutants relative to the average level of this protein in the respective WT strains. Moreover, overproducing Rad51 in the srs2Δ/srs2Δ strain mimics the swi6Δ/swi6Δ genome instability phenotypes to some extent (Fig. 6; see also Figs S5 and S6). These strains grew very poorly and displayed aberrant DNA content and increased numbers of multinuclear cells and cells containing nuclei with abnormal morphology.
It could be expected that SRS2 overexpression reduces the genome instability of swi6Δ strains. However, the high overproduction of SRS2 (from the GAL1 promoter) in swi6Δ cells influenced their DNA content profiles but did not improve them (Fig. 5E). Instead, such overproduction resulted in oversusceptibility to genotoxic stress and limited growth capacity of swi6Δ/swi6Δ mutants even under physiological conditions (Fig. 5D). The collected data suggest the indispensability of Rad51-dependent recombination for the survival of cells lacking SWI6.
Considering the important role of Swi6 in the cell cycle and in the stress response, it seems clear that a lack of SWI6 causes genome instability. However, for a long time, swi6Δ phenotypes linked to genome instability had not been described, and SWI6 was not defined as an essential gene. Why were these facts overlooked? It is possible this is because swi6Δ cells are highly mutable. To survive, they quickly adapt to changing environmental conditions. A lack of SWI6 is almost immediately compensated for by advantageous rearrangement of the genome (Fig. 3) in at least one cell of the population, producing a fitter dominant clone. Frequent DNA rearrangements and changing chromosome numbers are non-unique mechanisms that evolved to bypass the need to maintain a stable and functional genome to escape and survive stress. These mechanisms may be more common in yeast cells than in other eukaryotic cells (Hughes et al., 2000; Yona et al., 2012) but occur even in human cells. Aneuploidy is a typical hallmark of various human cancer cells (Thompson et al., 2010; Lee et al., 2016; Potapova et al., 2013). As in swi6Δ cells, the genome instability of cancer cells is due to endogenous stress caused by particular mutations (Coyle et al., 2017; Kobayashi et al., 2017; Yang et al., 2017; Margetis et al., 2017).
We asked what initiates the genome instability of SWI6-deficient cells. Our data showed that both haploid and diploid swi6Δ mutants display common phenotypes: (1) growth defects, for example, extended generation time, quick reaching of the stationary phase and increased cell mortality (Fig. 1); (2) a sporulation defect; (3) decreased survival under genotoxic stress (Fig. 2A); (4) inefficient DSB repair (Fig. 2C); and (5) spontaneous accumulation of DSBs (Fig. 2F). These data suggest that these cells experience endogenous genotoxic stress and have an error-prone DNA damage response that leads to DNA rearrangement (Fig. 3).
Considering the crucial role of Swi6 in the G1/S phase transition, we assumed that replication defects due to insufficient access to proteins involved in DNA synthesis would lead to replication stress. This assumption is supported by the increased number of Rfa1 foci in SWI6-deficient strains and the greater increase upon zeocin treatment (Fig. 2G). An elevated number of Rfa1 foci represents an increased number of ssDNA zones that need coating, which, in swi6Δ strains, immediately causes replication problems. The resulting replication stress is likely a source of DNA damage, presumably DSBs, reflected by the increased recruitment of Rad52 to the damage sites (Fig. 2F). The DNA damage level in swi6Δ appears to be high, as the Rad52 foci number does not increase after DSB induction by zeocin, suggesting that ‘all hands are already on deck’.
The supposition that the source of DNA stress in swi6Δ mutants is replication block was supported by the results of the chromosome comet assay (Fig. 2D). This methodology enables access to the DNA structures in chromosomal samples (Krol et al., 2017; Lewinska et al., 2014). In the swi6Δ-derived samples, we observed replication bubbles and Ψ-type structures (the latter formed during fork regression). The gapped DNA and short DNA fragments enriched in swi6Δ- and swi6Δ/swi6Δ-derived samples indicate DSBs to be a major DNA lesion type in these strains. The presence of various recombination intermediates in both swi6Δ- and swi6Δ/swi6Δ-derived samples suggested that recombination is involved in the repair of this damage. Moreover, haploid and diploid mutant strains accumulated distinct recombination intermediates, which was consistent with the results of the forward mutation assay. As shown in Fig. 2B, the lack of SWI6 exerts opposite effects on the spontaneous mutagenesis of haploid and diploid cells. In swi6Δ/swi6Δ, the level of spontaneous mutations was approximately one order of magnitude higher than in the diploid WT, which is characteristic of strains with defects in DNA repair, but in swi6Δ, the frequency of mutations was lower than in the haploid WT, and almost negligible. The vast difference in the spontaneous mutagenesis level between haploid and diploid control was also visible. To understand this difference, one must consider what exactly is measured in the mutagenesis test. In both haploid and diploid, one is looking for mutagenic events that lead to the loss-of-function phenotype of the mutagenesis marker. In the CAN1 or URA3 locus of haploids, the loss of function might be caused by point mutations, such as base substitutions (transversions and transitions), insertion or deletion events, and inversions or transpositions with summed frequencies of ∼2 per 106 cells. However, in diploids, the CAN1/can1Δ or URA3/ura3Δ marker loci allow detection of the loss-of-function phenotypes that results not only from point mutations (the frequency of these events is the same as in haploid cells) but also from at least the more-frequent LOH events (which are at least an order of magnitude more common). Thus, the much higher number of spontaneous DNA changes in diploid cells is due to LOH through gene conversion, allelic crossover and chromosome loss events (Ohnishi et al., 2004; Hiraoka et al., 2000).
In haploid and diploid cells, spontaneous mutation spectra differ (predominantly point mutations versus DNA rearrangements, respectively), as do processes that lead to these mutations. The cellular process responsible for generating point mutations is error-prone DNA synthesis relying on TLS DNA polymerases (Skoneczna et al., 2015; Gao et al., 2017). DNA polymerase ζ is the only TLS DNA polymerase in yeast that synthesizes longer DNA stretches (Northam et al., 2010). This feature, together with the low fidelity of this enzyme, makes it responsible for the majority of point mutations (Stone et al., 2012). However, in swi6Δ cells, this repair pathway cannot be used effectively because DNA polymerase ζ does not function in these cells. The Rev7 protein, the regulatory subunit of DNA polymerase ζ, is barely detectable in swi6Δ (Gong and Siede, 2009). The shortage of the Rev7 protein and the resulting lack of DNA polymerase ζ activity are likely responsible for the extremely low mutation rate in swi6Δ (Fig. 2B). To repair DNA lesions and complete DNA synthesis, another pathway must be activated, such as the DNA damage avoidance pathway, which relies on the activity of proteins involved in homologous recombination. Notably, several high-throughput screens analyzing the epistatic relation between genes showed negative genetic correlation or even synthetic lethality between RAD51 and REV7, and between RAD51 and REV3 (Collins et al., 2007; Costanzo et al., 2010; Ball et al., 2009; Pan et al., 2006), which supports this conjecture.
TLS deficiency in swi6Δ cells leads to replication block followed by DSB formation. Since the DNA damage response in swi6Δ cells is also affected by the altered expression of Swi6-dependent genes, the DSBs become recombinogenic. Traces of DNA rearrangements were visible in the flow cytometry profiles and the CGH results (Fig. 3A,D). DNA rearrangements could be detected in the forward mutation assay in diploid cells but not in haploid cells, where the loss of a partial or whole chromosome is lethal, which explains the contradictory mutagenesis results (Fig. 2B). The translocation of chromosomal fragments might result in canavanine or 5-FOA resistance in haploid mutants but will also induce the mitotic spindle checkpoint due to chromosome imbalance. Indeed, the flow cytometry profile of an asynchronous culture of swi6Δ cells displayed not only a shift to aneuploidy but also an accumulation of G2 cells (Fig. 3A). Miles and coworkers, who obtained similar data, explained this phenotype by the swi6Δ mutant losing the ability to transition to quiescence (Miles et al., 2016). Instead, we favor the hypothesis that the flow cytometry profile changes reflect the genome instability of the strains and subsequent division problems.
The mutagenesis assay and flow cytometry results suggested illegitimate recombination as a source of genome instability. This hypothesis was supported by the DSB accumulation seen in swi6Δ mutants. DSBs can lead to genome rearrangements (Malkova and Haber, 2012; Chan et al., 2007). The flow cytometry results and CGH experiments uncovered different patterns of genomic changes in haploids and diploids, that is, random changes in the haploid genome versus chromosome V duplication in the diploid genome. Chromosome V was duplicated in each of three independently prepared swi6Δ/swi6Δ strains, revealing the high level of rearrangements in such strains. At least a 10-fold increase in mutation frequency was detected in swi6Δ/swi6Δ strains, even though the mutagenesis markers are located on the duplicated chromosome V, which means that mutagenesis is even higher, as loss of function is frequent in strains containing more than one copy of a functional mutagenesis marker.
The TLS deficiency in swi6Δ mutants means that there is a requirement for an alternative DNA repair pathway to avoid DNA damage-induced replication block (see the proposed model in Fig. 7). This alternative damage avoidance pathway relies on homologous recombination and uses the sister chromatid as a template for DNA synthesis (Halas et al., 2011; Skoneczna et al., 2015). We showed that the balance of Rad51 and Srs2 proteins engaged in this repair pathway was disturbed in swi6Δ mutants (Fig. 5A,G), causing illegitimate recombination leading to DNA content changes (Figs 3 and 6) and growth defects in the strain (Fig. 1). In haploids, the homologous sequences necessary for recombination repair are available only when the synthesis of certain DNA sequences is already finished, so replication blocks often lead to fork collapse. The accumulation of replication bubbles and Ψ-type structures in chromosomal samples from swi6Δ cells indicates that there are replication fork problems (Fig. 2D). As previously shown (Pan et al., 2006), rad51Δ and swi6Δ are synthetically lethal. We demonstrated that only Rad51 overproduction allows swi6Δ mutants to survive because it permits the use of deleterious homologous recombination routes. A plasmid loss experiment proved that swi6Δ cells could lose the pRS413-SWI6 plasmid only when a plasmid overexpressing Rad51 was present (Fig. 5F). Therefore, RAD51 gene overexpression not only suppresses swi6Δ cell growth but actually keeps this mutant alive.
The random chromosomal rearrangement in haploid swi6Δ cells (Fig. 3A,D) frequently led to cell death (Fig. 1). Diploid swi6Δ/swi6Δ cell genomes are more plastic and these cells tolerate DNA rearrangements better because they contain two copies of the genome, facilitating adjustment to stressful conditions. These cells accumulate aberrant DNA structures that are different from those in swi6Δ, mostly recombination intermediates such as large DNA circles or elaborated recombination intermediates, which resemble the structures developed on DNA during multi-invasion-induced rearrangement (Piazza et al., 2017; Nguyen et al., 2017) (Fig. 2D). The tolerance of DNA rearrangement by the swi6Δ/swi6Δ strain allows it to counteract SWI6 deletion by means of duplication of chromosome V, which contains RAD51, the gene necessary for survival (Figs 3D and 5F). As chromosome V contains two other genes that suppress swi6Δ/swi6Δ cell growth (SWI4 and PAB1), the choice of duplicated chromosome is especially justified from the perspective of adaptive evolution. The duplication of chromosome V in the swi6Δ/swi6Δ mutant is an example of a duplication that enables cells to survive under strong selective pressure, usually a transient solution that facilitates further adaptation, as proposed by Yona and coworkers (Yona et al., 2012). We showed that of two previously known swi6Δ multicopy suppressors, only SWI4 could restore the DNA content profile of swi6Δ and swi6Δ/swi6Δ strains. We excluded Swi4-dependent RAD51 expression as the molecular mechanism of this suppression, but its molecular basis remains unknown and needs further exploration. Swi6 and Swi4 influence the expression of a few hundred genes, including those that encode proteins necessary for DNA synthesis and the DNA damage response. Shortages of these transcripts slow replication and may lead to replication fork block or even fork collapse, providing a wealth of possible explanations for this biological phenomenon. Which of these explanations will prove true remains unknown.
MATERIALS AND METHODS
Strain and plasmid construction
Yeast strains were the derivatives of the BY474X strain and are listed in Table S1. Strains YBI11 and YBI16 were constructed by gene replacement using the swi6Δ::kanMX4 disruption cassette in strains YAS288 and YAS281, respectively. Strain YMS32 resulted from the cross of parental strains YBI11 and YBI16. Strain YKK49 is a BY4742 derivative with a swi6Δ::kanMX4 disruption. YKK50 is a swi6Δ::kanMX4/swi6Δ::kanMX4 diploid resulting from a cross of the YKK49 and BY4741 swi6Δ strains from a knockout yeast strain collection (Open Biosystems). YBI19 (SWI6/swi6Δ::kanMX4) was prepared by the elimination of one genomic copy of SWI6 from the BY4743 diploid and its replacement with the swi6Δ::kanMX4 allele, with pRS415-SWI6 present in the cell.
The plasmids used in this study are listed in Table S2. To construct a plasmid overexpressing the SRS2 gene, the PCR-amplified SRS2 coding sequence obtained by using specific primers (Table S3) was introduced at flanking BamHI and HindIII restriction enzyme sites. The BamHI–HindIII fragment was cloned into the yeast centromeric vector pRS415 under the ADH1p promoter. The plasmid pRS423-PAB1 was constructed by subcloning the EcoRI-NotI DNA fragment carrying the PAB1 gene from pPS1601 (Valentini et al., 2002) into the pRS423 vector. Plasmids pRS413-SWI6 and pRS416-SWI6 were constructed by subcloning the SWI6 from pRS415-SWI6 (Chiu et al., 2011) into pRS413 and pRS416 (Sikorski and Hieter, 1989), respectively.
Other resources, such as primers, antibodies, kits and software used in this study, are listed in Table S3.
Culture media and growth conditions
YPD medium contained 1% yeast extract (Difco, Mt Pritchard, NSW, Australia), 2% bactopeptone (Difco) and 2% glucose (POCh, Gliwice, Poland). YPD-GPS medium was YPD medium supplemented with G-418 (100 μg/ml; Calbiochem, Darmstadt, Germany), penicillin (50 μg/ml; Polfa Tarchomin, Poland) and streptomycin (50 μg/ml; Sigma-Aldrich, St Louis, MO). SC medium contained 0.67% yeast nitrogen base (Difco) and 2% glucose and was supplemented with all amino acids, uracil and adenine (Formedium, Hunstanton, UK). The minimal yeast nitrogen base medium contained 0.67% yeast nitrogen base (Difco) and 2% glucose. The solid medium also contained 2.5% agar (Difco). Liquid cultures were grown with agitation at ∼200 r.p.m. (New Brunswick Scientific, Edison, NJ); the swi6Δ mutants were grown at 23°C, and other strains at 28°C. The media were sterilized with an EnbioJet microwave autoclave (Enbio Technology, Kosakowa, Poland).
Determining the growth rate
Yeast strain cultures were grown at 23°C with shaking to the exponential phase and then diluted to a density of ∼104 cells/ml in the same medium and cultivated under the same conditions. Every 2 h for at least 36 h, the number of cells/ml in each culture was counted by using a hematocytometer, and the optical density at 600 nm (OD600) values of the cultures were measured. Then, the OD600 values and cell counts were plotted as a function of time in Microsoft Excel.
Phloxine B staining of dead cells
The freshly prepared yeast strains were inoculated into YPD liquid medium and cultivated overnight at 23°C with shaking (150 r.p.m.) until they reached a density of 1×107–2×107 cells/ml. Then, 2 μl of phloxine B (from a 10 mg/ml stock) was added to 1 ml of yeast culture, and the cells were grown for an additional 40 min under the same conditions. After incubation, the cells were spread on glass slides, covered with cover glasses and observed under 200× magnification with a fluorescence microscope (Axio Imager.M2, Zeiss, Oberkochen, Germany). The phloxine B-labeled dead cells were observed by using a 46HE filter set, and the total numbers of cells were counted in bright-field images. Then, the frequency of dead cells was calculated. The results were the average of two independent experiments in which at least 500 cells were analyzed. The procedure was repeated after one (5 days) and two passages (10 days).
Plasmid loss assay
Strain BY4741 was first transformed with the appropriate plasmids (i.e. with pRS315 or pAT5 and with pRS413 or pRS413-SWI6). Then, in the strains with different plasmid combinations, the SWI6 gene was replaced with the swi6::kanMX4 disruption cassette via gene replacement. The resultant strains were passaged three times on YPD (without selection). Next, 100 clones of swi6Δ strains carrying certain plasmid sets, namely, pRS415 and pRS413, pAT5 and pRS413, pRS415 and pRS413-SWI6, or pAT5 and pRS413-SWI6, were replicated and grown on YPD supplemented with G418 and on SC −Leu medium to select clones possessing pRS415-derived plasmids and on SC −His medium to select clones possessing pRS413-derived plasmids. The number of clones growing on each medium was counted, and the frequency of plasmid loss was determined.
Sporulation efficiency assay
Diploid strains pre-grown on YPD medium were placed onto sporulation medium (0.1% yeast extract, 1% potassium acetate, 0.05% glucose, 2% agar) for 14 days at 28°C. Cells were then suspended in water. Cells and asci were counted in a cell counting chamber (at least 300 cells per probe), and the frequency of asci among the total cells was expressed as percent of asci relative to all cells counted and relative to WT. The mean and s.d. were calculated from the data for at least three cultures for each strain.
Following sporulation (see above), cells were suspended in 0.5 mg/ml Zymolyase 100 T solution in 50 mM Tris-HCl pH 7.5, and incubated for 10 min at 37°C. Tetrads were dissected using a Singer MSM200 micromanipulator (Singer Instruments, Roadwater Watchet, Somerset, UK) on YPD plates, and spores were grown for 3 days at 23°C.
Sensitivity drop tests
To estimate the level of sensitivity to genotoxic stress, we performed a drop assay. Each deletion strain was grown in YPD liquid medium with shaking at 23°C to a density of ∼1×107–2×107 cells/ml. The cells were then spun down (850 g for 1 min), washed with sterile water, and adjusted to a density of 3.3×107 cells/ml via resuspension in the same solution. Five 6-fold serial dilutions were performed for each strain, and 3.3 μl of each dilution was spotted onto Omnitray (Nunc, Roskilde, Denmark) plates containing YPD medium supplemented with appropriate concentrations of a specified genotoxic agent or on YPD medium for the dilution control. The plates were then incubated at 23°C for 2–3 days. The growth on each plate was documented with an Epson Perfection 2480 Photo scanner (Seiko Epson Corporation, Suwa, Japan).
Spontaneous mutagenesis assay
This test is based on the loss of function of a certain marker gene. To allow the detection of mutations in the forward mutation assay, strains with introduced mutagenesis markers were freshly prepared. The URA3 and CAN1 genes served as mutagenesis markers in haploids, and heterozygous CAN1/can1::LEU2 and URA3/ura3Δ loci were created in diploids.
The CAN1 gene encodes arginine permease, which is responsible for the uptake of arginine from growth media. The mutagenesis test relies on the similarity between arginine and its toxic analog canavanine. Cells with a functional CAN1 gene can uptake canavanine and incorporate it into proteins, causing dysfunction and subsequent cell death. Cells with loss-of-function mutations of the CAN1 gene can grow in the presence of canavanine. Therefore, this test enables the positive selection of mutants. Mutations in the URA3 gene, the second mutagenesis marker, are selected on 5′-fluoroorotic acid (5-FOA), which serves as an alternative substrate for Ura3. This enzyme usually converts orotidine-5′-phosphate into UMP, but the alternative substrate, 5-FOA, is converted into 5-fluorouracil, which is toxic. This assay also enables the positive selection of mutants, as only the cells with loss of function of the URA3 gene can survive in the presence of 5-FOA.
The frequency of spontaneous mutagenesis was estimated by a semiquantitative drop assay. Independent cultures of the analyzed strains were grown for 48 h in 1 ml of YPD medium with shaking at 23°C in MASTERBLOCK® 96-well deep well microplates (Grainer Bio-One, Oberösterreich, Austria). The cells were then spun down by centrifugation (1810 g for 1 min), resuspended in 500 µl of water and sonicated three times for 3 s each in a Branson 2800 ultrasonic bath (Branson Ultrasonics Corp., Danbury, CT). Then, 150-µl aliquots of cells were placed in a 96-well microtiter plate (Medlab-Products Sp. z. o.o., Warsaw, Poland), and the cell suspension was serially 6-fold diluted; 33 µl of the cell suspension from the first row of the microtiter plate for haploids or 3.3 µl for diploids was spotted onto Omnitray plates (Nunc) containing an appropriate selection medium [SC arginine +canavanine (60 µg/ml) or SC +5-FOA (1 mg/ml)] for mutant selection. Then, 3.3 µl of the cell suspension from rows 4 to 7 was spotted on the control plates containing SC medium and incubated at 23°C for 2–3 days. The number of colonies grown on each type of plate as countable spots were totaled, and the mutation frequencies were calculated considering the dilution factor. At least 48 independent cultures of each strain were assayed in two independent biological experiments, and the median was calculated to obtain the final mutagenesis frequency estimation. The P-values were calculated by using the Mann–Whitney U-test.
Recovery after zeocin treatment assay
The assay was performed as in Krol et al. (2015), except that the cells were grown at 23°C. Briefly, yeast strains were cultured in 50 ml of YPD liquid medium at 23°C with shaking to a density of ∼1×107–2×107 cells/ml. Each culture was then divided into 2:1 proportions. The smaller subculture served as a control and was supplemented with 15 μg/ml nocodazole (Sigma-Aldrich) and incubated for 3 h at 23°C with shaking. The larger subculture was supplemented with 100 μg/ml zeocin (InvivoGen, San Diego, CA) and incubated for 1 h at 23°C with shaking. The cells were then spun down (1610 g for 1 min),), washed twice and resuspended in fresh medium. The suspension was divided again into two subcultures of equal volume. One served as a zeocin-treated probe, and 15 μg/ml nocodazole was added to the other. Afterwards, the cells were incubated with shaking for an additional 3 h at 23°C to allow recovery from genotoxic stress. Samples of 107 cells from each of the three cultures were examined by PFGE to determine their chromosomal integrity.
Separation of yeast chromosomes by pulse field gel electrophoresis
Yeast chromosome integrity was analyzed as described previously (Krol et al., 2017). Yeast cells were embedded in 20-μl plugs of low-melting-point InCert agarose (Lonza, Basel, Switzerland) and digested with Zymolyase 100T (BioShop, Burlington, ON, Canada), and then with proteinase K (Sigma-Aldrich) and RNase A at 30°C with gentle rotation (4 r.p.m.) on an SB3 rotator (Bibby Sterlin, Stone, UK). Plugs were placed in the wells of a gel prepared from 0.7% D5 agarose (Conda, Torrejon de Ardoz, Madrid, Spain) in 1× TAE and sealed with the same agarose. Electrophoresis was performed for 24 h in 1× TAE buffer at 6 V/cm and 12°C with the ramping set to 0.8, the angle set to 120°, and the switch time set to 65–90 s in a CHEF Mapper® XA Pulsed-Field Electrophoresis System (Bio-Rad, Hercules, CA). After electrophoresis, the DNA was stained with 0.5 μg/ml ethidium bromide (Sigma-Aldrich) for 30 min with gentle rocking, washed twice with water for 15 min, and documented by using 302-nm UV light for DNA visualization and a charge-coupled device camera (FluorChem Q Multi Image III, Alpha Innotech, San Leandro, CA).
Chromosome comet assay
Aberrant DNA structures in the chromosomal bands excised from the PFGE gel were visualized via a chromosome comet assay (see Fig. S7 for graphical description of principles of the method and some examples of assay results), as described in Krol et al. (2017). In short, the chromosomal bands were placed distally on poly-L-lysine-coated microscopic slides (CometSlide, Trevigen, Gaithersburg, MD) and shielded with 40 μl of 0.6% New Sieve low-melting-point agarose (Conda). After agarose solidification, the slide was placed in 30 mM NaOH (POCh) and 1 mM EDTA, pH>12, for 10 min to denature the chromosomal DNA. Electrophoresis was performed for 15–20 min at 0.1 A for 15 min under denaturing conditions in 30 mM NaOH and 1 mM EDTA, pH>12. Then, agarose neutralization and DNA precipitation were performed by soaking 3 times for 30 min each in N/P solution (50% ethanol (Polmos, Warsaw, Poland), 1 mg/ml spermidine (Sigma-Aldrich) and 20 mM Tris-HCl, pH 7.4). Chromosomal DNA was stained with the fluorescent dye YOYO-1 (Thermo Fisher Scientific) by spotting the staining solution [0.25 mM YOYO-1, 2.5% DMSO (Sigma-Aldrich), 0.5% sucrose (Schwarz/Mann, Orangeburg, NY)] on the slide. DNA structures were visualized with an Axio Imager.M2 fluorescence microscope equipped with a 38HE filter set and an AxioCam MRc5 Digital Camera (Zeiss, Oberkochen, Germany). Images were collected at 1000× magnification, archived and processed using Axio Vision 4.8 (Zeiss).
Determination of Rad52 and Rfa1 foci frequency by fluorescence microscopy
The analyzed strains were transformed with the pWJ1344 plasmid carrying a RAD52–YFP fusion (Torres-Rosell et al., 2007). The transformants were grown to the exponential phase (1×107 cells/ml) in SC liquid medium at 23°C. An aliquot of each culture was collected to assess the frequency of spontaneously formed Rad52–YFP foci. Zeocin was added to the remaining culture at 100 μg/ml, and the cells were incubated for an additional hour under the same conditions. The zeocin-induced Rad52 foci were examined under 1000× magnification with an Axio Imager.M2 fluorescence microscope (Zeiss, Oberkochen, Germany) with a 46HE filter set and under Nomarski optics and captured with an AxioCam MRc5 Digital Camera operated by Axio Vision 4.8 software (Zeiss) with the following exposure times: DIC (100 ms) and Rad52–YFP (1500 ms). Images were processed and enhanced identically with Axio Vision 4.8. The numbers of cells and of Rad52 foci in the cells were counted, and the average percentage of cells with Rad52 foci and the average number of Rad52 foci per cell were calculated after screening at least 200 cells in each of three biological repeats, for a total count of at least 600 cells. The results are presented as the mean±s.d., and P-values were calculated with a Student's t-test.
The Rfa1–YFP foci were analyzed as described above, except that the RFA1-YFP::LEU2 fusion was introduced into the native RFA1 locus by gene replacement. In this experiment, the exposure times were 100 ms for DIC and 1500 ms for Rad52–YFP.
DNA content analysis by flow cytometry
The DNA content of yeast cells was measured by flow cytometry as previously described (Krol et al., 2015), with slight modifications. The cells from 1 ml of a yeast exponential culture (0.5×107–1×107 cells/ml) were spun down (19,300 g for 1 min) and subjected to permeabilization and fixation via suspension in 1 ml of chilled (−20°C) 80% ethanol (Polmos, Warsaw, Poland). The suspensions were held at room temperature for at least 2 h. The fixed cells were then washed twice in FACS buffer [0.2 M Tris-HCl (Sigma-Aldrich) pH 7.4 and 20 mM EDTA (Merck, Darmstadt, Germany)] and incubated at 37°C for 2 h in FACS buffer with 1 mg/ml RNase A (Sigma-Aldrich) to remove the RNA. Then, the cells were washed with phosphate-buffered saline (PBS) and stained overnight at 4°C in the dark with 100 μl of propidium iodide solution (50 μg/ml in PBS; Calbiochem). After the addition of 900 μl of PBS, the cells were sonicated three times for 3 s each in a Branson 2800 ultrasonic bath, to avoid cell clumping, immediately before flow cytometry analysis of the DNA content. This analysis was performed with a FACSCalibur analyzer (Becton-Dickinson, Franklin Lakes, NJ). A total of 10,000 cells in each sample were counted. Several (4–10) biological repeats were performed for each test, and a representative FACS profile is shown.
DAPI staining of yeast genomic DNA
Strains were grown to the exponential growth phase (1×107–2×107 cells/ml) in YPD liquid medium with shaking at 23°C. Afterwards, the cells were centrifuged and then permeabilized and fixed by resuspension in 1 ml of 80% ethanol and 15-min incubation at room temperature. The cells were spun down and resuspended in 0.5 ml of PBS, and 0.5 µl of DAPI from a 1 mg/ml stock was added before 15-min incubation in darkness at room temperature. The cells were washed twice with 1 ml of PBS and then resuspended in 50 µl of PBS. An aliquot of the cell suspension was dropped on a microscope slide and covered with a cover glass. The cells were examined with a fluorescence microscope (Axio Imager.M2, Zeiss, Oberkochen, Germany) with a 49 DAPI filter set or under Nomarski optics at 1000× or 630× magnification. Images were captured with an AxioCam MRc5 digital camera with an exposure time of 100 ms for DIC and DAPI. Images were processed and enhanced identically using Axio Vision 4.8 software.
Array-based CGH assay
Genomic DNA (0.5 μg) from the strains selected for DNA content comparison was labeled with either Cy3- or Cy5-dUTP with the SureTag DNA Labeling Kit. Equal amounts of labeled DNA of both strains were combined; hybridized to the Yeast (V2) Gene Expression Microarray, 8×15K (17 h at 67°C), and washed by using components of the Oligo aCGH Hybridization Kit. Reagent kits and instrumentation were supplied by Agilent Technologies Inc. (Santa Clara, CA), and the experiment was performed according to the manufacturer's protocols. After hybridization and washing, the slides were scanned with an Axon GenePix 4000B. Feature extraction was performed with GenePix Pro 6.1 and normalization with Acuity 4.0 (Molecular Devices, Sunnyvale, CA). CGH profiles with superimposed piecewise regression plots (to emphasize differences in DNA content) were generated by CGH-Explorer v3.2 (Lingjaerde et al., 2005).
For western blotting, 2×107 cells collected by centrifugation were lysed immediately by the addition of 50 μl of SDS sample buffer [50 mM Tris-HCl pH 7.5, 5% (w/v) SDS, 5% (v/v) glycerol, 50 mM DTT, 5 mM EDTA, 0.04% (w/v) Bromophenol Blue, 1 mM PMSF, 1 mM benzamidine, 10 mM N-ethyl maleimide, 2 μg/ml leupeptin, 2 μg/ml pepstatin A, 1 μg/ml chymostatin, 8.8 μg/ml aprotinin and 3 μg/ml antipain]. Samples were vigorously vortexed and frozen in liquid nitrogen. Immediately before use, the samples were boiled for 6 min. After centrifugation (19,300 g for 2 min), equal volumes of the SDS-lysed cell extracts were separated by SDS-PAGE (7% or 9% polyacrylamide gel, depending on the mass of the analyzed protein), and the proteins transferred onto PVDF membrane (Amersham). Blots were blocked for 2 h in 5% (w/v) nonfat dried milk and/or 3% (w/v) BSA in TBST [25 mM Tris-HCl pH 7.5, 137 mM NaCl, 27 mM KCl, 0.1% (v/v) Tween-20] before probing with primary antibodies. Rad51 protein was detected by incubating the membranes with rabbit polyclonal antibody anti-Rad51 (1:2000, Thermo Fisher, PA5-34905) followed by incubation with goat anti-rabbit IgG conjugated to horseradish peroxidase (HRP) (1:2000, DAKO, P0448). Actin was detected by using mouse anti-actin monoclonal antibody (1:5000, C4, CHEMICON, MAB-1501) followed by goat anti-mouse IgG conjugated to HRP (1:2000, DAKO, P0447) antibody. Srs2 was detected by using mouse anti-Srs2 monoclonal antibody (1:250, Abmart, X3-P12954) followed by incubation with goat anti-mouse IgG conjugated to HRP (1:2000, DAKO, P0448). Immunoreactive proteins on the blots were visualized by using chemiluminescent substrates for HRP (SuperSignal West Pico, PIERCE) and documented with a charge-coupled device camera (FluorChem Q Multi Image III, Alpha Innotech, San Leandro, CA) or exposed to FOTON XS-1N film. The results from three to five experiments were averaged to determine the relative protein levels. The resulting bands were quantified by using Image Quant 5.2 (Molecular Dynamics, Inc., Sunnyvale, CA). The protein level was normalized to that of Act1 or to unspecific bands.
Liquid cultures of yeast strains in the exponential phase (1×107–2×107 cells/ml) were harvested and stored at −80°C until use. RNA was extracted by the hot acid phenol method (http://younglab.wi.mit.edu/expression/totalRNAprep.html). The removal of genomic DNA contamination and cDNA synthesis (with 1 µg of total RNA as a template) were performed by using a Maxima First-Strand cDNA Synthesis RT-qPCR kit with DNase (Thermo Scientific) according to the manufacturer's protocol. A control PCR was performed without prior reverse transcription to ensure that there was no genomic DNA contamination. Reactions were performed with RT HS-PCR mix SYBR® A (A&A Biotechnology, Poland) and a LightCycler® 480 II Roche according to the manufacturer's protocols. The primers used for qRT-PCR (for RAD51 and SWI4 and for ACT1 to normalize the data) are listed in Table S3. They were analyzed for specificity and efficiency, and the PCR efficiency was at least 1.8. The crossing thresholds (CTs) were calculated by the second derivative method with the LightCycler Relative Quantification Software. The relative quantification was corrected for PCR efficiency. qPCR was performed with two biological and two technical replicates. Amplification efficiency was calculated using the LightCycler® 480 Software version 1.5 by the standard curve method with 5×, 25×, 50×, 100×, 200× and 400× dilutions. The fold difference in expression was calculated by the Livak method. Statistical significance was determined with a Student's t-test.
Quantification and statistical analysis
The t-test and Mann–Whitney U-test were performed with Origin Pro software.
Detailed descriptions of the size of the analyzed samples and the data processing are provided in the Materials and Methods section and in the figure descriptions. The significance thresholds used throughout the paper are *P<0.05, **P<0.01 and ***P<0.001.
We thank Linda Breeden (Fred Hutchinson Cancer Research Center, USA), Ian W. Dawes (School of Biotechnology and Biomolecular Sciences, University of New South Wales, Australia), Michal Dmowski (Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Poland), Aneta Kaniak-Golik (Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Poland), and Pamela Silver (Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, USA) for providing the plasmids and Orna Dahan (Department of Molecular Genetics, Weizmann Institute of Science, Israel) for providing the evoHigh-pH strain.
Conceptualization: A.S.; Methodology: K.K., J.A.-M., M. Skoneczny, M. Sienko, A.H., A.S.; Validation: A.S.; Formal analysis: K.K., J.A.-M., M. Skoneczny, J.J., A.S.; Investigation: K.K., J.A.-M., M. Skoneczny, M. Sienko, J.J., I.R., A.S.; Resources: A.S.; Data curation: M. Skoneczny, A.S.; Writing - original draft: K.K., A.S.; Writing - review & editing: M. Skoneczny, A.K., A.S.; Visualization: K.K., J.A.-M., M. Skoneczny, A.S.; Supervision: A.S.; Project administration: A.S.; Funding acquisition: A.S.
This work was supported by Narodowe Centrum Nauki [grant no. 2016/21/B/NZ3/03641 to A.S.].
The authors declare no competing or financial interests.