|Home | About | Journals | Submit | Contact Us | Français|
Sgs1 is a member of the RecQ family of DNA helicases, which have been implicated in genomic stability, cancer and ageing. Srs2 is another DNA helicase that shares several phenotypic features with Sgs1 and double sgs1srs2 mutants have a severe synthetic growth phenotype. This suggests that there may be functional overlap between these two DNA helicases. Consistent with this idea, we found the srs2Δ mutant to have a similar genotoxin sensitivity profile and replicative lifespan to the sgs1Δ mutant. In order to directly test if Sgs1 and Srs2 are functionally interchangeable, the ability of high-copy SGS1 and SRS2 plasmids to complement the srs2Δ and sgs1Δ mutants was assessed. We report here that SGS1 is a multicopy suppressor of the methyl methanesulphonate (MMS) and hydroxyurea sensitivity of the srs2Δ mutant, whereas SRS2 overexpression had no complementing ability in the sgs1Δ mutant. Domains of Sgs1 directly required for processing MMS-induced DNA damage, most notably the helicase domain, are also required for complementation of the srs2Δ mutant. Although SGS1 overexpression was unable to rescue the shortened mean replicative lifespan of the srs2Δ mutant, maximum lifespan was significantly increased by multicopy SGS1. We conclude that Sgs1 is able to partially compensate for the loss of Srs2.
The ability of DNA helicases to unwind DNA enables replication, transcription, recombination and repair of DNA. The fundamental importance of DNA helicases is highlighted by a variety of human genetic disorders resulting from defective helicase function (1). Although the clinical features of these disorders differ, they all exhibit genomic instability and a predisposition to cancer. One class of DNA helicases, the RecQ family (named after its homology with the prototypical bacterial RecQ helicase), has attracted particular interest recently (2). Defects in three of the five known human RecQ family members manifest as Bloom’s syndrome (BLM) (3), Rothmund–Thomson syndrome (RECQ4) (4) and Werner’s syndrome (WRN) (5). All three disorders show genomic instability associated with a predisposition to various cancers, while Rothmund–Thomson syndrome and Werner’s syndrome display features of premature ageing. At the cellular level, cells from Werner’s syndrome sufferers display genomic translocations and deletions (6), and a decreased replicative lifespan in vitro (7, reviewed in 8). Cells from Bloom’s syndrome patients similarly exhibit chromosomal rearrangements and deletions, but are characterised by a uniquely high frequency of sister chromatid exchanges (reviewed in 9).
In order to shed light on how mutations in human RecQ helicases result in genomic instability, much recent attention has been directed at RecQ orthologues in lower organisms. The genomes of Escherichia coli, Saccharomyces cerevisiae and Schizosaccharomyces pombe each contain only a single predicted RecQ helicase, thus facilitating analysis of their cellular function(s). Mutations in E.coli recQ, S.cerevisiae SGS1 and S.pombe rqh1+ all result in increased recombination (10–15), suggesting a conserved role of RecQ helicases in regulating recombination. Interestingly, the hyper-recombination phenotype of budding yeast sgs1 mutants can be complemented by the human WRN and BLM genes (12). In addition to the hyper-recombination phenotype, sgs1 mutants are sensitive to a range of genotoxins and display a reduced replicative lifespan (defined as the number of buds produced by a mother cell) (12,16–24). The human BLM gene has been shown to complement the hydroxyurea sensitivity (12) and short lifespan (22) of the sgs1 mutant, further reinforcing the notion of a conserved function of RecQ helicases. Current thinking suggests that this function may be the prevention of inappropriate recombination at stalled replication forks arising during S phase (25).
The S.cerevisiae SRS2 gene was discovered as a suppressor of the UV sensitivity of rad6 and rad18 mutants (26), and also independently in a genetic screen which isolated and characterised yeast hyper-recombination mutants (27). No obvious human homologue of Srs2 has yet been found, but the sequence of the SRS2 gene does show homology to the bacterial UvrD and Rep helicases (28) and an S.pombe orthologue has been identified recently (29). One function of Srs2 is to channel the repair of DNA lesions into the RAD6 post-replication repair pathway (30). In the absence of Srs2, DNA lesions are instead channelled into the RAD52 homologous recombination repair pathway (31). Srs2 has also been implicated in other DNA repair processes such as non-homologous end joining (32) and single-strand annealing (33).
Srs2 shares a number of interesting features with Sgs1. Both Sgs1 and Srs2 have been demonstrated to be 3′→5′ DNA helicases in vitro (34,35). The SGS1 and SRS2 gene products are generally expressed at low abundance but are up-regulated during the S phase of the cell cycle, and both have been implicated in the intra-S checkpoint response (18,36,37). Furthermore, sgs1 and srs2 mutants have a hyper-recombination phenotype (13,27). Most recently, sgs1Δ and srs2Δ strains have both been shown to have a shortened replicative lifespan and a greater propensity to undergo terminal cell-cycle arrest as mitotic G2/M intermediates (38). Deletion of both the SGS1 and SRS2 genes causes a severe synthetic growth defect, with low cell viability and a high incidence of G2/M terminal cell-cycle arrest, which can be suppressed by deletion of genes involved in homologous recombination (20,38,39). This suggests that unrestrained recombination causes the poor growth of the sgs1srs2 double mutant and is consistent with the idea that Srs2, like Sgs1, may act to prevent inappropriate recombination at stalled replication forks.
It was initially proposed that the Sgs1 and Srs2 helicases perform redundant functions in DNA replication and RNA pol I transcription (40). However, the DNA replication and RNA pol I defects reported are in fact likely to be secondary consequences of unrestrained recombination (20). Nevertheless, the original suggestion of functional overlap between these two helicases (40) remains possible. Consistent with this idea, we found the srs2Δ mutant to have a similar genotoxin sensitivity profile and replicative lifespan to the sgs1Δ mutant. Moreover, increased gene dosage of SGS1 partially restored genotoxin resistance and increased the maximum lifespan in the srs2Δ mutant. This indicates that Sgs1 can partially compensate for the loss of Srs2.
Yeast strains and plasmids used in this study are described briefly in Table Table1.1. The BY4743 strain was obtained from ATCC (Manassas, TN), and all other strains were obtained from Research Genetics (Groningen, The Netherlands). All deletion mutants were prepared by deletion module PCR using the Kan MX4 cassette (Saccharomyces Genome Deletion Project).
The pYES2 vector was obtained from Invitrogen (Groningen, The Netherlands). The pQE30 plasmid was obtained from Qiagen (Dorking, UK). The YCplac33 and YEplac195 vectors were gifts from Dr Alan Boyd (University of Edinburgh, Edinburgh, UK). The pNF3000 plasmid, which carries the RAD3 gene controlled by its endogenous promoter in the YEp24 vector, was generously provided by Drs Paula Fischaber and Errol Friedberg (University of Texas Southwestern Medical Center, Dallas, TX).
The YCplac33-SRS2 and YEplac195-SRS2 plasmids were constructed as follows. The 3525 bp SRS2 open reading frame (ORF) plus 610 bp upstream and 132 bp downstream was amplified from yeast genomic DNA (Promega, Southampton, UK) using the following primers: CATGAACCTGGATCCATATATAGAAATCGG (sense, BamHI site underlined) and CATTAAGGGTACCTAACACACCATCCACATTTC (antisense, KpnI site underlined). The PCR product was digested with BamHI and KpnI and ligated into BamHI/KpnI-digested YCplac33 and YEplac195 vectors. The pYES2-SRS2 plasmid was constructed by amplifying the 3.525 kb SRS2 ORF from YEplac195-SRS2 using the primers GGGAGGATCCATGTCGTCGAACAATGATCTTTG (sense, BamHI site underlined) and GAAACTCGAGCTAATCGATGACTATGATTTCAC (antisense, XhoI site underlined). The PCR product was digested with BamHI and XhoI and ligated into BamHI/XhoI-digested pYES2 vector.
The YEplac195-SGS1 and YEplac195-SGS1 K706→A constructs were made by subcloning the 4.8 kb SalI fragment containing the SGS1 gene from YCplac33-SGS1 and YCplac33-SGS1 K706→A (23) into SalI-digested YEplac195. The pYES2-SGS1 plasmid was constructed by amplifying the 4.344 kb SGS1 ORF from yeast genomic DNA using the following primers: GTACACACAAGGGATCCATGGTGACGAAGC (sense, BamHI site underlined) and TTTCCTCGAGTCACTTTCTTCCTCTGTAGT (antisense, XhoI site underlined). The PCR product was digested with BamHI and XhoI and ligated into BamHI/XhoI-digested pYES2 vector.
The pQE30-SGS1(1-400) plasmid was constructed as follows. The 1200 bp fragment of SGS1 was amplified from YCplac33-SGS1 using the following primers: GTACACACAAGGGATCCATGGTGACGAAGC (sense, BamHI site underlined) and TTCTTCTGGTACCACCTACGGATAATCGTC (antisense, KpnI site underlined). The PCR product was digested with BamHI and KpnI and ligated into BamHI/KpnI-digested pQE30 plasmid.
Initially, the YEplac195-SGS1 plasmid was modified to remove a NotI restriction site carried over from the original cloning vector. Transposon-based mutagenesis was then performed on this modified plasmid using the E::Z™ In-frame linker insertion kit (Cambio, Cambridge, UK). This utilises the Tn5 in vitro transposition system (41). Mutagenesis was performed in vitro by incubating 0.03 pmol of EZ::TN<NotI/KAN-3> transposon and plasmid DNA with 1 µl of Tn5 transposase enzyme at 37°C for 2 h. The reaction was then stopped and the reaction mixture transformed into XL1 Blue E.coli (Stratagene, Amsterdam, The Netherlands). All resultant Kan+ Amp+ colonies were pooled and grown up prior to plasmid DNA isolation. The KanR gene was spliced out by digesting the midiprep with NotI, leaving a 57-bp sequence at the site of transposon insertion. This sequence codes for a 19-amino acid read-through insertion, which maintains the reading frame and so avoids creation of truncation mutants. The NotI-cut plasmid was gel purified, re-ligated in vitro and re-transformed into XL1 Blue E.coli. The resulting Amp+ colonies were used to construct both the random mutational library and the methyl methanesulphonate (MMS) loss-of-function library.
For the random mutational library, 190 Amp+ colonies were selected at random and grown up individually. Of these, PCR analysis identified 80 colonies that contained plasmids harbouring the 57-bp insertion within the SGS1 gene. These colonies were individually grown up prior to isolation of their plasmid DNA. The position of transposon insertion was determined by digesting plasmid minipreps with SalI and NotI.
For the MMS loss-of-function library, all original Amp+ colonies harbouring transposon-mutagenised YEplac195-SGS1 were pooled and grown up. Their DNA was then isolated and transformed into the sgs1Δ mutant strain. Ura+ transformants (500) were then picked and patched individually onto 10 × 10 grids on SD-URA plates. Each plate also contained positive (sgs1Δ-YEplac195-SGS1) and negative (sgs1Δ-YEplac195) controls for comparison. Plates were incubated overnight at 30°C, and then replica plated onto plates containing 0.023% MMS and left for 2 days. Loss-of-function mutations (i.e. plasmids no longer able to rescue MMS sensitivity in the sgs1Δ strain) were identified on replica plates, and the corresponding colonies picked from the master plate. These were grown up, pooled and their DNA recovered and transformed into XL1 Blue E.coli. Forty-two Amp+ colonies were selected for further analysis of transposon insertion by restriction digest analysis. The remaining colonies were pooled, and their DNA isolated by midiprep.
This was performed as described previously for sgs1Δ, using identical drug concentrations (23). For quantitation of strains grown on solid media supplemented with MMS, yeast strains were grown up in minimal media for 2–3 days at 30°C, diluted and then spread on solid media plates containing increasing doses of MMS. After 3 days, the percent viability of each strain was expressed as number of colonies obtained relative to that on the control (no MMS) plate. For quantitation of MMS sensitivity of strains in liquid media, yeast strains were grown to an OD600 of 0.5–0.6 before addition of 0.01–0.12% MMS for 60 min. Cells were washed in 10% sodium thiosulfate, diluted and spread onto solid media plates. After 3 days, the percent viability of each strain was expressed as number of colonies obtained relative to that obtained for the control samples.
Yeast extracts were prepared by bead-beating in 8 M-urea-containing 2D-gel extraction buffer (Bio-Rad, Hemel Hempstead, UK) and run on SDS–polyacrylamide gels prior to transfer onto nitrocellulose. Recombinant his-tagged Sgs1(1-400) protein was purified from E.coli transformed with the pQE30-SGS1(1-400) plasmid using Ni2+-NTA-agarose (Qiagen). The purified protein was then used for antibody production in rabbits by following a published method (42). The resulting Sgs1p 857 antiserum was then purified using protein G–Sepharose to yield the IgG fraction used for western blotting. The Srs2p (yA-19) and Sir2p (yN-19) goat polyclonal antibodies were obtained from Autogen Bioclear (Mile Elm, UK). All primary antibodies were routinely used at dilutions of 1:500. Immunoreactive bands were detected with peroxidase-conjugated secondary antisera and visualised using enhanced chemiluminescence.
This was performed as described previously (23). Briefly, strains were grown at 30°C until they reached an OD600 of 0.6–1.0. One microlitre of culture was streaked onto plates and left at 30°C for 1–2 h. After this time, cell doublets were moved to uninhabited regions of the plate. When these budded again, (newly formed) virgin yeast cells were removed by micromanipulation to a new location. All future buds produced by these daughter cells were micromanipulated away, and catalogued. The plates were incubated at 30°C during working hours, and moved to 4°C overnight. Lifespan was defined as number of daughter cells removed from the mother cell. All lifespans were observed a minimum of twice.
This was performed as described previously (23).
Deletion of the SGS1 gene in both haploid and diploid strains has been reported to result in a replicative lifespan of ~40% that of isogenic wild-type strains (22–24). To determine whether deletion of SRS2 has a similar effect, lifespan analysis was performed on srs2Δ and BY4743 (wild-type) isogenic diploid strains transformed with the empty YCplac33 vector (Fig. (Fig.1A).1A). Both strains were transformed with YCplac33 to allow direct comparison with previous findings (23). The mean lifespan of the BY4743-YCplac33 strain was 16.0 ± 1.0 (n = 76), compared with 7.2 ± 0.9 (n = 38) for the srs2Δ-YCplac33 strain. The maximum lifespans recorded were 44 and 23, respectively. The short lifespan of the srs2Δ mutant observed is comparable with that of the isogenic diploid sgs1Δ-YCplac33 strain, which has a mean lifespan of 7.4 ± 0.8 (n = 32) and maximum of 16 (23). A similar magnitude of shortened replicative lifespan has recently been reported for the haploid sgs1Δ and srs2Δ strains (38). Therefore, deletion of SGS1 or SRS2 reduces the replicative lifespan to a similar extent in both haploid and diploid backgrounds.
We have previously characterised the sensitivity of the sgs1Δ mutant to MMS (DNA methylating agent), 4-NQO (causes bulky-base adducts and oxidative damage), hydroxyurea (DNA synthesis inhibitor), camptothecin (Type I topoisomerase inhibitor) and mitoxantrone (Type II topoisomerase inhibitor) (23). To assess the sensitivity of the srs2Δ mutant to these drugs, the diploid srs2Δ mutant was transformed with YCplac33 (empty vector) and YCplac33-SRS2 and compared with the BY4743 isogenic wild-type strain transformed with YCplac33. Various dilutions of srs2Δ-YCplac33, BY4743-YCplac33 and srs2Δ-YCplac33-SRS2 strains were grown on plates containing the above drugs (Fig. (Fig.1B).1B). Similar to the sgs1Δ mutant, the srs2Δ-YCplac33 strain showed sensitivity to all DNA damaging agents tested. Transformed srs2Δ cells containing the YCplac33-SRS2 plasmid showed comparable drug sensitivity to the BY4743 wild-type strain under these conditions. The observation that plasmid-borne SRS2 can restore genotoxin resistance to isogenic wild-type levels thus confirms that deletion of the SRS2 gene is the sole defect in the srs2Δ mutant.
To assess the specificity of the drug sensitive phenotype of the sgs1Δ and srs2Δ strains, the MMS sensitivity of other yeast mutants lacking DNA helicases was assessed. The pif1Δ, rrm3Δ, yKU70Δ and yKU80Δ mutants were tested. Pif1 and Rrm3 are DNA helicases involved in regulating telomeric and mitochondrial DNA (reviewed in 43), and are also involved in the replication fork progression through the rDNA (44). The yKU70 and yKU80 gene products are involved in the repair of DNA double strand breaks (reviewed in 45) and physical and genetic interactions between RecQ helicases and Ku proteins have been demonstrated (46,47). It was found that the pif1Δ mutant showed a slight sensitivity to MMS, as compared with the isogenic wild-type strain. The rrm3Δ, yKU70Δ and yKU80Δ strains were not sensitive to MMS (results not shown). Therefore, MMS sensitivity appears to be a relatively specific phenotype for loss-of-function of either the Sgs1 or Srs2 helicases.
Based on the fact that either single mutant is viable, but the double mutant has a severe growth defect, it has been proposed that the Sgs1 and Srs2 helicases may be functionally redundant (40). We therefore reasoned that overexpression of either of these DNA helicases might compensate for loss of the other. To initially confirm overexpression, Sgs1 and Srs2 levels in vivo were assessed by western blotting. Protein was extracted from sgs1Δ-pYES2, sgs1Δ-pYES2-SGS1, srs2Δ-pYES2 and srs2Δ-pYES2-SRS2 cultures induced with galactose. Extracted protein samples were separated by SDS–PAGE, western blotted and probed for either Sgs1 or Srs2. An ~160 kDa band corresponding to Sgs1 was present in the sgs1Δ-pYES2-SGS1, but not sgs1Δ-pYES2, protein extract (Fig. (Fig.2A).2A). Similarly, an ~150 kDa band corresponding to Srs2 was present in the srs2Δ-pYES2-SRS2, but not srs2Δ-pYES2, protein extract (Fig. (Fig.2A).2A). For a control, all samples were probed for Sir2p, which was equally present in all protein extracts.
Although expression of SGS1 or SRS2 from the galactose-inducible pYES2 vector resulted in detectable levels of Sgs1 or Srs2 protein, it was not possible to detect Sgs1 by western blotting when SGS1 was expressed under its own promoter from the YEplac195 vector. We therefore attempted to genetically verify the overexpression of Sgs1 from this plasmid. Sgs1 helicase activity has been postulated to produce a substrate that is deleterious unless resolved by Top3 (11,48). Sgs1 overexpression should therefore be more harmful in top3Δ mutants than in TOP3 strains. To test this, top3Δ mutant and isogenic wild-type control strains were transformed with both low-copy (YCplac33) and high-copy (YEplac195) SGS1 plasmids (Fig. (Fig.2B).2B). Both strains gave viable Ura+ transformants when transformed with the YCplac33, YCplac33-SGS1, YEplac195 and YEplac195-SGS1 K706→A plasmids. However, transformation of both strains with YEplac195-SGS1 gave viable Ura+ transformants in the wild-type strain, but very few viable colonies with the top3Δ mutant. Therefore, expression of Sgs1 from the multicopy YEplac195 vector, but not the low-copy YCplac33 plasmid, was detrimental to the top3Δ mutant. This indicates that increased gene dosage of Sgs1 does indeed result in increased Sgs1 protein levels. Furthermore, the deleterious effects of higher Sgs1 levels in the top3Δ mutant required the helicase activity of Sgs1, consistent with the notion that the helicase activity of Sgs1 creates a substrate that Top3 resolves (11,48).
To assess if Sgs1 and Srs2 are functionally interchangeable, the ability of high-copy plasmids carrying SGS1 and SRS2 to rescue MMS sensitivity in the sgs1Δ and srs2Δ strains was assessed (Table (Table2).2). To control for non-specific effects of helicase overexpression, the high-copy pNF3000 plasmid carrying RAD3 (a 5′→3′ DNA helicase involved in excision repair) (49) was also compared. As a further specificity control, the effects of these plasmids on the (slight) MMS sensitivity of the pif1Δ mutant were also investigated. High-copy plasmids encoding SRS2 (both YEplac195-SRS2 and pYES2-SRS2) were able to complement the MMS sensitivity of the srs2Δ mutant, but had no obvious beneficial effects in either the sgs1Δ or pif1Δ mutants. Rather, Srs2 overexpression was found to have a dominant negative effect on growth of sgs1Δ, pif1Δ and wild-type strains. High-copy vectors expressing SGS1 (YEplac195-SGS1 and pYES2-SGS1) were able to complement the MMS sensitivity of both the sgs1Δ and srs2Δ mutant, but not the pif1Δ mutant. Sgs1 helicase activity was essential for complementation as YEplac195-SGS1 K706→A (expressing a helicase-defective allele which has some complementing activity in the sgs1Δ mutant) (23,50) was not able to suppress the MMS sensitivity of the sgs1Δ or srs2Δ mutant. The pNF3000 plasmid was unable to rescue the MMS sensitivity of all three mutants tested, although this plasmid has been demonstrated previously to complement the UV sensitivity of a rad3-1 strain (49).
The ability of SGS1 to rescue the sensitivity of the srs2Δ mutant to MMS and hydroxyurea is shown in Figure Figure3A.3A. This demonstrates that the ability of SGS1 to act as a multicopy suppressor of srs2Δ is not uniquely specific for sensitivity to DNA damaging agents, but is also seen for inhibitors of DNA replication. To quantify the SGS1 complementation of srs2Δ MMS sensitivity, srs2Δ-YEplac195-SGS1, srs2Δ-YEplac195 and BY4743-YEplac195 diploid strains were grown up, diluted and then spread onto plates containing increasing doses of MMS. After 3 days, the percent viability of each strain was expressed as number of colonies obtained relative to that on the control (no MMS) plate. It can be seen that at the 0.003% dose, where complete killing of srs2Δ-YEplac195 cells occurred, the YEplac195-SGS1 construct restored MMS resistance to ~60% of that seen in the isogenic wild-type (Fig. (Fig.3B).3B). A similar effect was observed when cells were exposed to MMS for 1 h before removal and inactivation of the drug (Fig. (Fig.3C).3C). Under these conditions, overexpression of SGS1 restored viability to 30–40% of wild-type levels at doses where complete killing of control srs2Δ-YEplac195 cells occurred.
Mutational analysis suggests that Sgs1 is a multifunctional protein, as mutant alleles of sgs1 have selective complementing ability in different phenotypic assays (17,21,23,51). We therefore set out to determine whether the domains of Sgs1 required for rescue of MMS sensitivity in the sgs1Δ mutant differ from those required for complementation of the srs2Δ mutant. In order to do this, we used in vitro transposon-scanning mutagenesis on the YEplac195-SGS1 plasmid to create two distinct libraries of sgs1 mutants containing in-frame 19-amino acid insertions (see Materials and Methods). The unselected mutational library was generated by individually growing up randomly chosen bacterial colonies and PCR screening for the presence of plasmids harbouring the 57-bp insertion within the SGS1 gene. Figure Figure4A4A shows the frequency and approximate location of in-frame transposon insertions in SGS1 in this library, which comprises 75 alleles. It can be seen that there is coverage of insertion events throughout the entire SGS1 gene with no obvious hotspots, suggesting that transposon insertion is basically random. To construct a library of sgs1 MMS loss-of-function plasmids (i.e. plasmids no longer able to rescue MMS sensitivity in the sgs1Δ strain), sgs1Δ cells were transformed with the entire original pool of transposon-mutagenised YEplac195-SGS1. Of the resulting Ura+ transformants, 500 were picked and grown up individually on 10 × 10 grids before replica plating onto MMS plates. One hundred colonies failed to grow on MMS plates; these were recovered from the master plate, and their pooled plasmid DNA isolated to create the MMS loss-of-function library. Restriction analysis of 42 alleles chosen at random from this library revealed a more restricted distribution of insertion events than in the unselected library (compare Fig. Fig.4A4A and B). Most notably, there was a complete absence of insertions in the C-terminal 250 amino acids of Sgs1, a domain known to be dispensable for MMS resistance (17,51) (Fig. (Fig.44B).
To determine whether the same functional domains of SGS1 are also required for complementation of srs2Δ, the srs2Δ mutant was transformed with the MMS loss-of-function SGS1 library (more than 500 Ura+ transformants obtained). For a control, the srs2Δ mutant was also transformed with the unselected library of 75 transposon-mutagenised SGS1 plasmids characterised in Figure Figure4A4A (more than 200 Ura+ transformants obtained). Colonies were resuspended, diluted and spread on control (no MMS) and 0.0035% MMS plates (a dose at which the srs2Δ mutant fails to grow). For the unselected SGS1 plasmid library, 156 colonies were obtained in total, compared with 378 colonies obtained on the control (no MMS) plates. This verifies that insertion of the 57-bp linker sequence in the SGS1 gene is not detrimental per se. In contrast, for the MMS loss-of-function SGS1 plasmid library, none of the MMS loss-of-function plasmids (in the sgs1Δ background) could rescue MMS sensitivity in the srs2Δ mutant. Therefore, we conclude that the same domains of SGS1 are required to restore MMS resistance to both sgs1Δ and srs2Δ strains, suggesting that a common molecular mechanism of action underlies complementation in both mutants.
It has been reported that sensitivity to MMS is enhanced in diploid srs2Δ strains relative to haploids (28). This ploidy-specific phenotype has been interpreted as additional lethal recombination events occurring between homologous chromosomes as well as sister chromatids. We therefore set out to determine whether the ability of SGS1 to complement MMS sensitivity in the srs2Δ mutant is due to suppression of recombination between homologous chromosomes. Haploid sgs1Δ and srs2Δ strains were used for these studies and compared with the isogenic wild-type BY4741 strain. To directly compare any ploidy-specific phenotypes, both haploid and diploid BY4741, sgs1Δ and srs2Δ mutant strains were grown on plates ± MMS (Fig. (Fig.5A).5A). As reported previously (28), the diploid srs2Δ strain was more MMS sensitive than the haploid srs2Δ strain. The observed MMS sensitivity of strains (in order of most MMS sensitive) was as follows: srs2Δ diploid > sgs1Δ haploid = sgs1Δ diploid > srs2Δ haploid > BY4741 haploid = BY4741 diploid.
To test for any ploidy-specific effects of Sgs1 and Srs2 overexpression, haploid strains transformed with YEplac195, YEplac195-SGS1, YEplac195-SGS1 K706→A or YEplac195-SRS2 were grown on plates ± MMS. Both the haploid sgs1Δ-YEplac195 and srs2Δ-YEplac195 strains were sensitive to 0.01% MMS, whereas BY4741-YEplac195 was unaffected at these doses (Fig. (Fig.5B).5B). The ability of YEplac195-SGS1 (and not YEplac195-SGS1 K706→A) to suppress MMS sensitivity was reproducible in both the sgs1Δ and srs2Δ haploid strains. Similarly, YEplac195-SRS2 could rescue the MMS-sensitive phenotype of the haploid srs2Δ mutant, but again no beneficial effects in the sgs1Δ haploid were evident. This demonstrates that the ability of SGS1 to function as a multicopy suppressor of the srs2Δ mutant is not a ploidy-specific effect.
In addition to the MMS-sensitive phenotype, the ability of the YEplac195-SGS1 plasmid to rescue the shortened replicative lifespan of the srs2Δ mutant was investigated. Lifespan analysis was performed on the srs2Δ-YEplac195 and srs2Δ-YEplac195-SGS1 strains (Fig. (Fig.6).6). Average lifespans of the strains [expressed as mean number of buds removed ± standard error of the mean (SEM)] were 11.1 ± 1.2 (n = 37) for srs2Δ-YEplac195 and 13.3 ± 1.4 (n = 38) for srs2Δ-YEplac195-SGS1. The maximum lifespans recorded were 24 and 36, respectively. Although a Student’s t-test revealed there to be no significant difference between the mean lifespans, a χ2 test on the 10% longest-lived subpopulations (52) revealed that Sgs1 overexpression caused a statistically significant increase in srs2Δ lifespan (P < 0.02).
Mutant sgs1 and srs2 strains have been shown to exhibit hyper-recombination, defective intra-S checkpoint responses and shortened replicative lifespans (13,18,27,37,38). The sgs1Δ mutant is also known to be sensitive to a variety of genotoxic agents (12,16,17,19,23). We report here that the srs2Δ mutant is similarly sensitive to MMS, hydroxyurea, camptothecin, mitoxantrone and 4-NQO, reinforcing the phenotypic overlap with the sgs1Δ mutant. In addition to its genotoxin sensitivity, the sgs1Δ mutant has a shortened replicative lifespan (22–24). We found that the srs2Δ mutant has a mean replicative lifespan ~45% of that of the wild-type strain. This is comparable with the average lifespan recorded for the isogenic sgs1Δ mutant (43%) (23), revealing a similar magnitude of shortened lifespan for the diploid sgs1Δ and srs2Δ mutants.
These findings are in agreement with a recent report documenting the phenotypes of the haploid sgs1Δ and srs2Δ mutants (38). In that work, it was revealed that the previous characterisation of premature ageing in the sgs1 mutant as a phenocopy of normal ageing (24) was not entirely accurate. Instead, it was concluded that the short lifespan exhibited by both sgs1Δ and srs2Δ mutants was due to a combination of ageing-independent mitotic checkpoint arrest and accelerated normal ageing (38). To determine whether the increased G2/M checkpoint arrest phenotype was also apparent in our diploid strains, we followed the approach of McVey et al. (38) and re-analysed our lifespan data for incidence of terminally budded phenotypes. This analysis was performed blind as our lifespan studies were completed before publication of their report. The morphology of sgs1Δ and srs2Δ terminal phenotypes revealed a higher percentage of budded mitotic intermediates than that observed for the isogenic wild-type strain (data not shown), thus confirming the findings of McVey et al. (38). As the severe phenotype of sgs1srs2 double mutants can be suppressed by inactivation of Rad51, Rad52 or Rad57, it is likely that it results from unrestrained recombination (38), in keeping with the idea that both Sgs1 and Srs2 act to prevent such events during S phase.
It has been suggested that the Sgs1 and Srs2 helicases may be functionally redundant, based on the fact that either single mutant is viable, but the double mutant has a severe growth defect (40). Although complete redundancy can be ruled out due to the abundance of phenotypes exhibited by each single mutant (see above), the similarities of these phenotypes are consistent with a degree of functional overlap. Our finding that SGS1 can function as a multicopy suppressor of the MMS and hydroxyurea sensitivity of the srs2Δ mutant provides direct support for this notion. In contrast, high-copy SRS2 was unable to complement genotoxin sensitivity of the sgs1Δ strain. One explanation of this finding is that Sgs1 performs some cellular function(s) that cannot be substituted by Srs2, despite the ability of Sgs1 to functionally replace Srs2. However, it is worth noting that, as reported previously (53), we found that higher levels of Srs2 caused detrimental effects in the wild-type strain.
Sgs1 is a large protein that, in addition to the RecQ helicase domain, has distinct domains required for interactions with Top2, Top3 and Rad51 (11,54–57). Furthermore, mutant alleles of sgs1 have selective complementing ability in different phenotypic assays (17,23,50,51). Therefore, the domains of Sgs1 required for rescue of MMS sensitivity in the sgs1Δ background could potentially differ from those required for complementation of the srs2Δ mutant. To address this issue, we constructed libraries of SGS1 mutant plasmids containing small in-frame insertions. Unsurprisingly, a large proportion of loss-of-function mutants for MMS resistance in the sgs1Δ background contained transposon insertions in or around the helicase domain of SGS1. This is consistent with reports that the helicase activity of Sgs1 is required for dealing with MMS-induced DNA damage (17,21,23,51,58). Some insertions were also found N-terminal to the helicase domain, mapping close to the interaction sites for Top2 and Top3. This supports the observation that the N-terminus of Sgs1 is essential for MMS resistance (17). Interestingly, no MMS loss-of-function alleles mapped to the C-terminus of Sgs1, where the Rad51 interaction occurs (57). Again, this is consistent with reports that much of the C-terminus of Sgs1 (the final 254 amino acids) is dispensable for MMS resistance (17,51). It was found that none of these SGS1 loss-of-function plasmids tested could complement the srs2Δ mutant. Similarly, a point mutation causing inactivation of Sgs1 helicase activity (50) was unable to restore MMS resistance to the srs2Δ strain. This demonstrates that complementation of the srs2Δ mutant by high-copy SGS1 requires domains of Sgs1 that are directly involved in processing MMS-induced DNA lesions.
When SGS1 was over expressed in the srs2Δ mutant, no significant increase in the mean replicative lifespan was evident. This contrasts with the ability of SGS1 overexpression to suppress MMS sensitivity of the srs2Δ mutant. One possible reason for this is that other important roles of Srs2 may still be lacking and contribute to the shortened mean replicative lifespan of the srs2Δ mutant. Alternatively, the beneficial effects of SGS1 overexpression may be outweighed by detrimental effects associated with high levels of Sgs1. Several observations support the latter theory. First, SGS1 overexpression can increase the maximum lifespan of the srs2Δ mutant by 50%. Secondly, SGS1 overexpression has been reported to cause detrimental effects (18,24). Thirdly, srs2Δ cells over expressing SGS1 actually show an increase in terminally budded morphologies as compared with control srs2Δ mutants containing empty vector (data not shown). It is remarkable that high-copy SGS1 can extend maximum lifespan in the face of such deleterious effects.
Proteins involved in recombinational repair (e.g. Rad52) and post-replication repair (e.g. Rad6 and Rad18) play important roles in processing DNA lesions induced by MMS, in an analogous manner to that proposed for the processing of UV-induced lesions (59). Srs2 is thought to channel lesions into post-replication repair pathways, whereas Sgs1 is thought to be more intimately involved in recombinational repair (2,57,60) (Fig. (Fig.7).7). The ability of Sgs1 overexpression to process MMS-induced DNA damage in the srs2Δ mutant implies that endogenous Sgs1 levels are normally limiting in the srs2Δ mutant. This effect of Sgs1 overexpression may be to increase the efficiency of Sgs1’s normal physiological function(s), thereby reducing the requirement for Srs2-dependent repair pathways. Alternatively, it may be that Sgs1 is capable of directly replacing Srs2, by processing substrates that would normally be dealt with by Srs2.
A variety of data support the notion that the sgs1Δ and srs2Δ phenotypes are a consequence of unrestrained recombination (20). Consistent with this, mutations in the RAD51 or RAD52 genes can suppress the MMS sensitivity of the srs2Δ mutant (61), whereas overexpression of RAD51 or RAD52 enhances the MMS sensitivity of the srs2Δ mutant (62). Therefore, in the absence of Srs2, the Rad52-dependent recombination repair pathway is over-active (31), leading to aberrant processing of recombination intermediates and genomic instability (Fig. (Fig.7).7). However, it is worth noting that, in addition to its well-established anti-recombinase function, Srs2 has recently been claimed to promote recombination in some circumstances (63). The absence of Sgs1 would also be predicted to cause aberrant processing of recombination intermediates as not only are the later steps of the homologous recombination repair pathway impaired (57,64), but also aberrant recombination intermediates cannot be ‘salvaged’ by Sgs1 (14). This model is consistent with observations that the synthetic growth defect in sgs1srs2 double mutants (40) can be rescued by mutation of genes involved in homologous recombination (20,38,39).
Although we favour the idea that Sgs1 overexpression rescues MMS sensitivity in the srs2Δ mutant by suppressing unrestrained recombination, it remains possible that the ability of Sgs1 overexpression to compensate for lack of Srs2 is unrelated to DNA recombination. This possibility is supported by the fact that inactivation of the homologous recombination pathway does not suppress the severe synthetic growth defects observed in the S.pombe srs2Δrqh1Δ double mutant (29). Nevertheless, the fact that genetic interactions between SGS1/rqh1+ and SRS2 are evident in two such evolutionarily divergent species further supports the notion of functional overlap between these DNA helicases. It will be of interest to assess if overexpression of rqh1+ can rescue any phenotypes of the S.pombe srs2Δ mutant.
These findings may have potentially important ramifications for studies of the human RecQ helicases. First, the physiological effects of overexpression of Sgs1 may warrant further characterisation, as higher levels of both the BLM and WRN helicases have been reported to exist in transformed cells and tumour cells, relative to those found in normal cells (65,66). Furthermore, the finding that RecQ helicases can show partial functional redundancy with other DNA repair pathways may be significant for the understanding of the functions of RecQ family members in maintenance of genomic stability. In addition to the functional overlap between Sgs1 and Srs2 demonstrated here, an extra copy of Ku70, a DNA repair helicase, can partially rescue the phenotypes of Drosophila lacking the RecQ helicase Dmblm (47). This functional overlap of DNA repair pathways implies that defects in human RecQ helicases would have more severe effects under conditions where the redundant pathways are limiting. This may be important in understanding the role of the human RecQ helicases in genomic stability, cancer and ageing.
We gratefully acknowledge Drs Paula Fischaber and Errol Friedberg for the generous gift of the pNF3000 plasmid. This work was supported by the Wellcome Trust in the form of Prize Studentships to H.W.M. and T.J.C.