RecQ DNA helicases and Topo III topoisomerases have conserved genetic, physical, and functional interactions that are consistent with a model in which RecQ creates a recombination-dependent substrate that is resolved by Topo III. The phenotype associated with Topo III loss suggests that accumulation of a RecQ-created substrate is detrimental. In yeast, mutation of the TOP3 gene encoding Topo III causes pleiotropic defects that are suppressed by deletion of the RecQ homolog Sgs1. We searched for gene dosage suppressors of top3 and identified Pif1, a DNA helicase that acts with polarity opposite to that of Sgs1. Pif1 overexpression suppresses multiple top3 defects, but exacerbates sgs1 and sgs1 top3 defects. Furthermore, Pif1 helicase activity is essential in the absence of Top3 in an Sgs1-dependent manner. These data clearly demonstrate that Pif1 helicase activity is required to counteract Sgs1 helicase activity that has become uncoupled from Top3. Pif1 genetic interactions with the Sgs1–Top3 pathway are dependent upon homologous recombination. We also find that Pif1 is recruited to DNA repair foci and that the frequency of these foci is significantly increased in top3 mutants. Our results support a model in which Pif1 has a direct role in the prevention or repair of Sgs1-induced DNA damage that accumulates in top3 mutants.
DNA helicases and topoisomerases represent two classes of enzymes that are important for maintaining the fidelity of genetic information after DNA damage and during chromosome replication and segregation. A combined helicase–topoisomerase activity has been evolutionarily conserved across many species (Duguet 1997; Wu et al. 1999). These activities exist in a single polypeptide in reverse gyrase, an enzyme that is unique to archaebacteria (Declais et al. 2000; Duguet et al. 2001). In eubacteria and in eukaryotes ranging from yeast to humans, two separate protein families, RecQ DNA helicases and DNA topoisomerases, specifically Topoisomerase III-like members, exhibit genetic and physical interactions consistent with a conserved RecQ–Topo III protein complex (reviewed in Bachrati and Hickson 2003).
The RecQ DNA helicase family, named for the prototypical member from Escherichia coli, includes Sgs1 from Saccharomyces cerevisiae, Rqh1 from Schizosaccharomyces pombe, and Blm, Wrn, RecQL, RecQL4, and RecQL5 from humans (Gangloff et al. 1994b; Murray et al. 1997; Stewart et al. 1997; Aravind et al. 1999; Bachrati and Hickson 2003). Multiple lines of evidence suggest a role for RecQ during homologous recombination. In all species studied, mutation of a RecQ homolog results in hyperrecombination and genome instability. In humans, mutations in Blm, Wrn, and RecQL4 cause the genetic disorders Bloom, Werner, and Rothmund–Thomson, respectively, which are characterized by an increased predisposition to cancer (reviewed in Bachrati and Hickson 2003). In vitro, RecQ DNA helicases unwind duplex DNA in a 3′ to 5′ direction with a marked preference for forked DNA substrates and can disrupt joint DNA molecules via branch migration (Bennett et al. 1998, 1999; Harmon and Kowalczykowski 1998; Constantinou et al. 2000; Karow et al. 2000). Sgs1 and Blm physically interact with the DNA strand exchange protein Rad51 in their respective organisms, pointing to an intimate relationship between RecQ and homologous recombination machinery (Wu et al. 2001). Furthermore, defects in both S. pombe cells that lack Rqh1 and mammalian cells that lack Wrn are suppressed by heterologous expression of a bacterial Holliday junction resolvase, indicating that recombination intermediates accumulate in the absence of RecQ (Doe et al. 2000; Saintigny et al. 2002).
Topoisomerase III was first identified in S. cerevisiae because mutations in the TOP3 gene cause hyperrecombination (Wallis et al. 1989). Loss of Topo III function in E. coli similarly causes genomic instability (Schofield et al. 1992) and Topo III is essential in fission yeast (Goodwin et al. 1999; Maftahi et al. 1999). There are two mammalian isoforms. In mice, Topo IIIα deficiency is embryonic lethal while Topo IIIβ deficiency causes multiple organ defects and reduced life span (Li and Wang 1998; Kwan and Wang 2001). Topo III belongs to the type 1A subfamily of topoisomerases whose members catalyze ssDNA passage events and are characterized by the ability to relax negatively supercoiled DNA (Champoux 2001). However, in vitro and in vivo evidence indicate that Topo III does not efficiently remove negative supercoils, raising the possibility that it is not a typical topoisomerase (DiGate and Marians 1988; Wallis et al. 1989; Kim and Wang 1992).
The first clues that RecQ helicases and Topo III function together came from studies in S. cerevisiae. Mutant top3 yeast cells exhibit pleiotropic defects that include slow growth accompanied by the accumulation of cells in G2/M, increased recombination between direct repeats, chromosome segregation defects, meiotic failure, and hypersensitivity to agents that create DNA lesions or arrest DNA replication (Wallis et al. 1989 ; Gangloff et al. 1994b; Chakraverty et al. 2001). Mutations in the SGS1 gene were identified as suppressors of top3 mutant slow growth (Gangloff et al. 1994b). An sgs1 mutation by itself causes pleiotropic defects reminiscent of the top3 phenotype, albeit milder in severity (Gangloff et al. 1994b; Watt et al. 1995). Epistasis analysis indicates that Sgs1 functions upstream of Top3 and the two proteins physically interact (Gangloff et al. 1994b; Bennett et al. 2000; Fricke et al. 2001). These observations led to a model in which Sgs1 creates a DNA substrate that requires Top3 activity for its resolution.
In vitro, E. coli RecQ and Topo III demonstrate a concerted DNA strand passage activity that neither enzyme can perform alone. Together, RecQ and Topo III catenate/decatenate or remove positive supercoils from covalently closed dsDNA molecules (Harmon et al. 1999, 2003). Evidence supports a model in which Topo III recognizes the DNA structure at the unwound DNA fork bound to RecQ and is stimulated to perform sequential ssDNA passage events on opposing DNA strands, thus effectively achieving a dsDNA strand passage event. A similar concerted enzymatic activity is seen cross-species with bacterial RecQ and budding yeast Top3 as well as with mammalian Blm–Topo III, indicating that this functional interaction is conserved (Harmon et al. 1999; Wu and Hickson 2002, 2003). Furthermore, the conservation of a protein–protein interaction domain in eukaryotic RecQ homologs that specifically interacts with Topo III speaks to the biological importance of a concerted RecQ–Topo III activity.
The in vivo substrate for RecQ–Topo III is not known. Since the top3 mutant phenotype is largely suppressed by sgs1 in yeast, the severe defects caused by loss of Top3 are likely due to incomplete processing of intermediates that arise when Sgs1 acts alone. In bacteria and yeast, Topo III mutant defects are also suppressed when homologous recombination is eliminated (Gangloff et al. 1999; Zhu et al. 2001; Oakley et al. 2002; Shor et al. 2002; Laursen et al. 2003). Furthermore, top3 mutant sensitivities to agents that cause DNA lesions or replication fork stall are suppressed by sgs1 (Chakraverty et al. 2001) and it has been suggested that Sgs1 functions in the stabilization of stalled replication forks (Cobb et al. 2003). These observations contributed to the proposal that RecQ–Topo III is important for the resolution of recombination intermediates that form during DNA replication. Further support for this notion comes from recent studies of the Holliday junction dissolution seen in vitro with the Blm–TopoIII complex (Wu et al. 2005).
To further our understanding of RecQ–Topo III biology, we searched for gene dosage suppressors of top3 in S. cerevisiae and identified a single gene, PIF1, which encodes a DNA helicase that acts with polarity that is opposite to that of Sgs1. Although Pif1 overexpression suppresses multiple top3 defects, it exacerbates sgs1 and sgs1 top3 defects. We find that Pif1 helicase activity is essential in top3 mutants and that this requirement is dependent upon both an active Sgs1 DNA helicase and homologous recombination. Homologous recombination proteins form nuclear DNA repair foci spontaneously during S phase and in response to DNA damage (Lisby et al. 2001; Lisby et al. 2003) and here we show that Pif1 localizes to these same DNA repair foci. The frequency of Pif1 foci is significantly increased in top3 mutant cells, suggesting that Pif1 is recruited to sites of DNA damage when Sgs1 acts without Top3. Our results demonstrate that the Pif1 DNA helicase is required to prevent or repair recombination-dependent DNA damage that arises when Sgs1 helicase activity is uncoupled from Top3.
MATERIALS AND METHODS
Yeast extract–peptone–dextrose (YPD), synthetic complete (SC), and 5-flouroorotic acid (5-FOA) media were made as described (Sherman et al. 1986) with twice the recommended amount of leucine. Sporulation medium was prepared as described (Klapholz and Esposito 1982).
Strains, plasmids, and genetic manipulations:
The yeast strains used in this study are listed in Table 1. Mating, sporulation, and dissections were performed by standard procedures (Sherman et al. 1986). For segregation analysis of synthetic gene interactions, a minimum of 80 tetrads were dissected in each case. Unless otherwise indicated, cells were maintained at 30°. Temperature-sensitive strains were grown at 23° and 37° for permissive and restrictive conditions, respectively. For microscopic examination of fluorescently tagged proteins, cells were grown at 23° to allow for optimal formation of the fluorophore. Transformation of DNA into yeast was performed by standard methods (Gietz and Woods 2002). Genetic manipulations involving polymerase chain reaction (PCR)-generated DNA fragments were confirmed by DNA sequence analysis. Relevant plasmid details are summarized in Table 2. Oligonucleotide sequences are given in supplemental Table 1 at http://www.genetics.org/supplemental/.
Plasmid pWJ659 (gift from J. Weinstein) was made by subcloning a blunt-ended 2.6-kb XbaI–FspI TOP3 fragment from pWJ171 (Wallis et al. 1989) into the SmaI site of pRS415 (Sikorski and Hieter 1989). Respectively, plasmids pWJ1092 and pWJ1114 were made by subcloning TOP3 from pWJ659 into YCp50 (Rose et al. 1987) and pRS306 (Sikorski and Hieter 1989) via BamHI and HindIII sites. To make plasmid pWJ1189 (CEN–URA3–ADE2–TOP3), the BglII ADE2 fragment from pRS417 (Brachmann et al. 1998) was subcloned into the BamHI site of pWJ1092.
The top3–E447K,S583L conditional allele (top3-ts) was isolated from a genetic screen. Briefly, a randomly mutagenized pool of TOP3 open reading frames (ORFs) was generated by PCR (primers TOP3 + 124-R and TOP3 − 84-F) and cotransformed with SnaBI- and PvuII-gapped pWJ1092 into a top3∷TRP1 haploid, thus allowing recombination-mediated formation of a top3 mutant plasmid library in vivo. Transformants were assayed for growth and MMS sensitivity at 23° and 37° and plasmids containing top3-ts candidates rescued for further analysis. Bona fide top3-ts candidates were integrated at the endogenous locus via standard allele replacement methods (described in Rothstein 1991). The integration plasmid pWJ1203 was made by subcloning the E447K, S583L missense mutations from the original library isolate pWJ1130 into pWJ1114 via BssHII and BsiWI sites. The top3–E447K,S583L strain has the desired ts phenotype and both missense mutations are required for this phenotype. To mark the TOP3 locus with TRP1, a 931-bp fragment from intergenic region iYLR237 was amplified by PCR (primers F-XbaI-iYLR237 and R-ClaI-iYLR237), digested with XbaI and ClaI and subcloned into pRS304 (Sikorski and Hieter 1989) to make pWJ1174, which was then linearized and integrated at the endogenous iYLR237 locus. To show hypomorphism, top3–E447K,S583L was subcloned from pWJ1203 into pRS425 (Christianson et al. 1992) via BamHI and SalI sites to make the top3-ts high-copy vector pWJ1372.
The top3–Y356F catalytically inactive allele was first constructed in plasmid pWJ1074 (see below) and the Y356F mutation subcloned into pWJ659 via SphI and BlpI sites to make plasmid pWJ1085. Using BamHI and HindIII sites the top3–Y356F allele was subcloned from pWJ1085 into pRS306 (Sikorski and Hieter 1989) to generate vector pWJ1113 that was then integrated at the endogenous TOP3 locus yielding a TOP3–URA3–top3–Y356F repeat (strain U515). Recombinant pop-outs selected on 5-FOA only gave TOP3 isolates, suggesting that top3–Y356F is lethal. Plasmid pWJ659 transformed into U1515 provides wild-type TOP3 in trans and allowed successful recovery of top3–Y356F integrants. The top3–Y356F + pWJ659 strain was crossed to a wild-type strain. Loss of pWJ659 from the heterozygous diploid followed by sporulation and segregation analysis demonstrated that a top3–Y356F strain is inviable. top3–Y356F spore products germinate and form microcolonies (2–100 cells) that do not grow upon restreak. Plasmid shuffle to replace pWJ659 with pWJ1189 yielded the haploid top3–Y356F strain covered by plasmid-borne TOP3 that was used in the gene dosage suppressor screen (W3040-9A). Consistent with top3–Y356F lethality, strain W3040-9A cannot lose pWJ1189, as evidenced by lack of growth on 5-FOA.
The pif1∷URA3K.l. and rrm3∷URA3K.l. strains were made by described methods (Reid et al. 2002a,b) using the appropriate intergenic primers (Research Genetics) and URA3K.l. pop-outs selected on 5-FOA. The pif1∷KAN-MX deletion allele was amplified with intergenic primers (Research Genetics) from the appropriate deletion consortium library strain (Winzeler et al. 1999) and transferred to W1588-4C as described (Reid et al. 2002a,b).
The catalytically inactive pif1–K264R allele was amplified from plasmid pWJ1280 (see below) and integrated at the endogenous locus via allele replacement methods (Erdeniz et al. 1997) using primers C-PIF1-484F and D-PIF1-919R. The pif1-m2 allele was made by oligonucleotide-directed mutagenesis/PCR fusion using primers C-PIF1-us45-F, delATG2-Rev, delATG2-For, and Pif1-302R-D and integrated by similar methods. The sgs1-K706R, sgs1-ΔN82, and sgs1-ΔN82, K706R strains were gifts from J. Weinstein.
Plasmids containing galactose-inducible genes were made as follows. Plasmid pWJ1047 contains a galactose-inducible promoter followed by an HpaI restriction site that is flanked by A- and B-adaptamer sequence (gift from R. Reid). Respectively, A- and B-adaptamers correspond to the sequences appended to the N- and C-terminal ORF primers available from Research Genetics (Hudson et al. 1997). The appropriate ORF primers were used to amplify TOP3, PIF1, and RRM3, and the resulting PCR products cotransformed into yeast with HpaI-linearized pWJ1047 to make vectors pWJ1067, pWJ1279, and pWJ1281, respectively, by in vivo recombination. The top3–Y356F allele was made via oligonucleotide-directed mutagenesis/PCR fusion (primers TOP3-YF-fwd, TOP3-YF-rev, and Research Genetics ORF primers for TOP3) and subcloned into plasmid pWJ1047 to make plasmid pWJ1074 via adaptamer-mediated in vivo recombination. The pif1–K264R missense mutation was made by oligonucleotide-directed mutagenesis/PCR fusion with primers PIF1-333-F, PIF1-K264R-R, PIF1-K264R-F, and PIF1-1467-R, and the resulting DNA fragment digested with PmeI and SacII and subcloned into pWJ1279 to make pWJ1280. Replacement of URA3 with HIS3 using the marker-swap cassette from pUH (Cross 1997) converted plasmids pWJ1047, pWJ1067, and pWJ1074 to plasmids pWJ1345, pWJ1346, and pWJ1347, respectively.
A C-terminal fusion of YFP to PIF1 was made at the endogenous locus via described methods (Reid et al. 2002a) using primers C-Pif1-g-up, C-Pif1-gfuse-up, C-Pif1-g-down, and C-Pif1-gfuse-down. Similarly, a C-terminal fusion of YFP to RRM3 was made at the endogenous locus using primers C-Rrm3-g-up, C- Rrm3-gfuse-up, C- Rrm3-g-down, and C- Rrm3-gfuse-down. Plasmid pWJ1326 was made by replacing the LEU2 marker of plasmid pCox4–DsRed.T4 (Bevis and Glick 2002) with HIS3 using the marker-swap cassette from pLH (Cross 1997).
For construction of CFP–NUP49, first a plasmid expressing Nup49 from its endogenous promoter was constructed by amplifying genomic NUP49 (primers Nup49fwd and Nup49bwd), digesting the DNA fragment with ScaI and MslI followed by subcloning into the SmaI site of pRS415 (Sikorski and Hieter 1989) and subsequent subcloning into pRS413 (Sikorski and Hieter 1989) via SacI and XhoI sites to make plasmid pWJ1315. Next, CFP coding sequence was fused to the 5′-end of endogenous NUP49 as described (Reid et al. 2002a) using primers Nup49fwd, Nup49-Tup, Nup49-Tdown, and Nup49down. Integration/pop-out of the URA3K.l. marker was performed under cover of pWJ1315 since the CFP–URA3K.l.–CFP–NUP49 fusion is a null allele of an essential gene. The resulting CFP–NUP49 fusion was incorporated into plasmid pWJ1315 via gap repair to make plasmid pWJ1323. Plasmid pWJ1348 was made by subcloning CFP–NUP49 from pWJ1323 into pRS416 (Sikorski and Hieter 1989) via BamHI and SalI sites.
Construction of the NOP1 plasmids was as follows. Plasmid pWJ1249 contains an HpaI restriction site flanked by C- and D-adaptamer sequence (gift from R. Reid). C- and D-adaptamers correspond to the sequences appended to the intergenic primers available from Research Genetics. First, a plasmid expressing Nop1 from its endogenous promoter was made via adaptamer-mediated in vivo recombination by cotransforming NOP1 that had been amplified from genomic DNA using the appropriate intergenic primers (Research Genetics) and HpaI-linearized pWJ1249. NOP1 was subsequently subcloned from pWJ1249 into pRS413 via SacI and XhoI sites to make plasmid pWJ1277. Next, CFP coding sequence was integrated at the 3′-end of endogenous NOP1 as described (Reid et al. 2002a) using primers Nop1-g-up, Nop1-gfuse-up, Nop1-gfuse-down, and Nop1-g-down. Integration was performed in a diploid since the NOP1–CFP–URA3K.l.–CFP haploid is inviable. NOP1–CFP was then amplified from the integrated repeat in two overlapping parts (primers Nop1-g-up, XFP-3′int, XFP-5′int, and Nop1-g-down) and the PCR products cotransformed with ClaI and SexAI gapped pWJ1277 to generate the NOP–CFP plasmid pWJ1299 by a three-part in vivo recombination. Plasmid pWJ1327 was made by subcloning NOP1–CFP from pWJ1299 to pRS416 (Sikorski and Hieter 1989) via XhoI and NotI sites. Plasmid pWJ1321 was made by gap-repair/exchange of CFP (pWJ1299) with DsRed. pWJ1322 was made by subcloning DsRed–NOP1 from pWJ1321 into pRS416 (Sikorski and Hieter 1989) via SacI and XhoI sites.
Screen for high-copy suppressors of top3Δ slow growth:
A homozygous top3Δ diploid was used for two reasons: (1) the likelihood of recovering recessive loss of function mutations that suppress top3 (e.g., sgs1) as false positives is reduced and (2) the greater severity of the top3 phenotype in diploids vs. haploids is more amenable to suppressor analysis. The top3Δ diploid is so sickly that it must be maintained by providing a plasmid-borne copy of wild-type TOP3 and a plasmid shuffle strategy was used for the screen. A homozygous top3∷TRP1 null diploid harboring plasmid pWJ1189 (CEN–URA3–ADE2–TOP3) was transformed with a high-copy yeast genomic library contained in vector YEp51B (2μ-LEU2) (Akada et al. 1997). Normally, the top3 diploid cannot lose pWJ1189 as evidenced by sensitivity to 5-FOA, a compound that counterselects URA3. Library transformants were selected on SC-Leu in the presence of uracil to permit random loss of pWJ1189 in isolates containing suppressors of top3 and subsequently replica plated to 5-FOA-Leu. Transformants that gave rise to 5-FOAr colonies were assayed for ability to grow on SC-Ade to confirm loss of pWJ1189. 5-FOAr, Ade− transformants were chosen as candidate suppressors for further analysis. The HCS2 clone isolated in this screen contains TOP3 and serves as a control plasmid in several experiments. A single plasmid-linked, non-TOP3 high-copy suppressor, designated clone 1 (Figure 1A), was obtained. Clone 1 also suppresses top3Δ slow growth in a haploid. Plasmid pWJ1246 deletes the 1277-bp SacII fragment from clone 1 encompassing the C-terminal half of PIF1 and does not suppress top3 defects. Plasmid pWJ1286 was made by subcloning the pif1–K264R mutation from pWJ1280 into clone I via Bsu361 and PmeI sites. We found that pWJ1286 suppresses top3 defects as well as clone 1 in top3 PIF1 strains.
To test whether pWJ1286 can suppress in the complete absence of wild-type Pif1 (top3Δ pif1Δ strains), we employed a plasmid shuffle strategy analogous to the original screening process. First, a top3Δ pif1Δ strain harboring pWJ1189 (TOP3) was generated by transformation of a top3Δ pif1Δ heterozygote with pWJ1189 followed by sporulation and selection of the desired haploid segregant. Next, the top3Δ pif1Δ pWJ1189 haploid was transformed with YEp51B, clone 1, or pWJ1286. The resulting transformants were tested for their ability to lose pWJ1189 by growing first on SC-Leu and then replica plating to 5-FOA-Leu. Only clone 1 allowed loss of pWJ1189; growth on 5-FOA for pWJ1286-containing isolates was nonexistent and indistinguishable from YEp51B-containing isolates.
Screen for high-copy suppressors of top3-ts slow growth:
This screen was performed in a top3-ts (top3–E447K,S583L) homozygous diploid using the same 2μ genomic library described above. Since the top3-ts diploid grows as well as a wild-type strain at 23°, there is no need to provide a plasmid-borne copy of wild-type TOP3, thus simplifying the screening method. A top3-ts homozygote was transformed with the YEp51B genomic library and transformants selected on SC-Leu at 23°. Colonies were subsequently replica plated to a fresh SC-Leu plate and incubated at 37°. Transformants exhibiting robust growth at the restrictive temperature were chosen as candidates. One plasmid-linked, non-TOP3 high-copy suppressor was identified and is designated clone 2 (Figure 1A).
Screen for high-copy suppressors of top3–Y356F lethality:
A top3–Y356F haploid is viable only when wild-type TOP3 is provided in trans. A top3–Y356F haploid harboring pWJ1189 was transformed with the YEp51B high-copy library and suppressors sought via a plasmid shuffle strategy as described above for the top3Δ high-copy suppressor screen. Recessive chromosomal loss-of-function mutations were frequently isolated as false-positive candidates and were eliminated by standard complementation testing. All candidates eliminated by this criterion were determined to be sgs1 mutations. No plasmid-linked, non-TOP3 high-copy suppressors were identified in this screen.
Cells were processed for fluorescent microscopy and mounted on slides as described previously (Lisby et al. 2001). Live cell images were captured with a cooled Orca-ER CCD camera (Hamamatsu, Japan) mounted on a Zeiss Axioplan 2 microscope (Carl Zeiss, Thornwood, NY) using a Plan-Apochromat 100×, 1.4 NA objective lens and 100 W mercury arc lamp (Osram, Munich, Germany). For each field of cells, 11 or 13 fluorescent images at each of the relevant wavelengths were obtained at 0.3-μm intervals along the z-axis to allow inspection of all focal planes of each cell. Images were acquired and pseudo-colored using OpenLab software (Improvision, Lexington, MA) and prepared for publication in Adobe Photoshop (Adobe, San Jose, CA). The fluorophores used in this study and the filters used to visualize them are as described (Lisby et al. 2003).
The recombination frequency between the δ-repeats at the SUP4 locus was determined as described (Rothstein et al. 1987; Wallis et al. 1989). W1588 (W303) strains have ochre-suppressible mutations at ADE2 and CAN1. The URA3 selectable marker is inserted between δ-repeats 4 and 5 at the SUP4 locus, thus linking it to the SUP4-o dominant allele. To measure the recombination frequency between δ-repeats, simultaneous loss of URA3 and SUP4-o was assessed by monitoring adenine prototrophy and both canavanine and 5-FOA resistance. For determination of the recombination frequency between rDNA repeats at the rDNA locus, an RDN∷URA3 assay (Zou and Rothstein 1997) was crossed into the appropriate genetic backgrounds, single segregants were picked from an SC-Ura plate, grown overnight in liquid SC-Ura, and then plated to 5-FOA medium to determine the frequency of URA3 loss. The total number of colony-forming units (cfu) analyzed was determined by direct plating to SC and the deletion frequency calculated accordingly. A recombination assay was placed at the CUP1 locus via integration of XbaI-linearized plasmid pL1580 (gift of R. Keil) and crossed into the appropriate mutant backgrounds. Similar to the method used to measure recombination at rDNA, the recombination frequency between CUP1 repeats was determined by measuring 5-FOA resistance as described (Keil and McWilliams 1993). All recombination assays were performed in the relevant null mutant backgrounds with the exception of the SUP4-o assay in which the pif1-m2 allele was used because the petite phenotype of pif1Δ interferes with the red/white color assay that is used to assess adenine prototrophy.
For determination of the doubling time of yeast strains, cultures were maintained in mid-log phase in SC-Leu and cell density was monitored over time with a spectrophotometer. Analysis of cell-cycle distribution was performed on log-phase cultures by microscopic examination. Cells were scored as G1 (unbudded), S (small budded), or G2/M (large budded). Classification as S vs. G2/M was based on both position of the nucleus as determined by a nuclear marker (Rad52 or Nup49) or DAPI staining and bud size. Cells with a bud less than or equal to one-third the size of the mother and a nucleus still in the mother cell were classified as S phase. Cells with a bud greater than one-third the size of the mother with the nucleus at or in the neck were classified as G2/M. Nuclear position was used as the overriding criterion. To determine the effect of p2μ-PIF1 on cell-cycle distribution, over 750 cells were viewed for each culture. To determine plating efficiency, quadruplicate cultures of cells were grown to mid-log phase in SC-Leu, quantitated in a hemacytometer, 200 cells plated per SC-Leu plate, and the plates incubated for 3 to 4 days. Plating efficiency was calculated as the number of cfu divided by the number of cells plated. In this same experiment, the frequency of spontaneous petite formation (mitochondrial deficiency) was determined by replica plating from SC-Leu to media containing glycerol (a nonfermentable sugar) as the sole carbon source. Since petites cannot grow on nonfermentable carbon sources, the frequency of petite formation could be determined as the number of cfu that do not grow on glycerol divided by the total cfu. To examine the effect of co-overexpressing TOP3 and PIF1 alleles in a wild-type strain background, a galactose-inducible series of vectors was used. The TOP3 series (pWJ1345, pWJ1346, and pWJ1347) was transformed into a wild-type MATa strain, and the PIF1 series (pWJ1047, pWJ1279, and pWJ1280) was transformed into a wild-type MATα strain. The two sets of resulting transformants were then mated to each other and diploids were selected on SC-Ura,His and streaked onto SGal-Ura,His medium. Similarly, pGal-RRM3 (pWJ1281) was transformed into the wild-type MATα strain and diploids were made to test the effect of co-overexpressing RRM3 and the TOP3 plasmid series.
Gamma irradiation was administered using a Gammacell-220 60Co irradiator (Atomic Energy, Ottawa, Canada). To determine gamma-ray sensitivities of strains, cells were grown to mid-log phase and then plated on YPD. YPD plates were exposed to varying doses (0–80 krad) of gamma. For analysis of Rad52 and Pif1 focus formation in response to gamma rays, cultures were grown to mid-log phase and exposed to 4 krad of gamma rays and aliquots of cells were taken and subjected to fluorescence microscopy for a 3-hr time course. Two trials yielded similar results and the data were compiled with over 2100 cells viewed postirradiation.
A search for high-copy suppressors in three different top3 mutant backgrounds:
A top3Δ strain exhibits multiple defects including slow growth, sensitivity to DNA damaging agents, a cell-cycle defect characterized by an accumulation of cells in G2/M, reduced cell viability, and hyperrecombination at repetitive sequences. Three top3 allelic backgrounds were used to search for high-copy suppressors. The first is a top3 null allele (top3Δ). The next is a hypomorphic top3-ts conditional allele that we isolated, which behaves like wild type at the permissive temperature and top3Δ at the restrictive temperature. The third is a catalytically inactive top3–Y356F allele, which contains phenylalanine substituted for the active site tyrosine that covalently binds to the free DNA end during the Top3 catalytic cycle (Bennett and Wang 2001). We found that this top3–Y356F strain is inviable. Gene dosage suppressors of top3 mutant slow growth—or in the case of top3–Y356F, lethality—were isolated as described in materials and methods. In the top3Δ background, 60,000 transformants yielded 27 plasmid-linked suppressors, 26 of which contained TOP3. One plasmid, designated clone 1, contains a genomic fragment from chromosome 13 (Figure 1A). Figure 1B shows suppression of top3Δ slow growth by clone 1. In the top3-ts background, 600,000 transformants yielded 218 plasmid-linked suppressor clones, 217 of which contained TOP3. One isolate, designated clone 2, contains a genomic fragment from chromosome 13 that is distinct from, but overlaps, clone 1 (Figure 1A). In the top3–Y356F background, 500,000 transformants yielded 203 plasmid-linked suppressors, all of which contained genomic clones of TOP3.
PIF1 encodes a 5′-to-3′ DNA helicase that functions in both nuclear and mitochondrial DNA metabolism (reviewed in Bessler et al. 2001) and is the only complete open reading frame common to clones 1 and 2. Deletion of the C-terminal half of PIF1 removes conserved helicase domains that are essential to catalytic activity and eliminates the ability of clone 1 to suppress top3 slow growth, confirming that PIF1 is the high-copy suppressor (Figure 1A and data not shown). The PIF1 gene has two distinct start codons (Schulz and Zakian 1994). Protein translated from the first ATG includes a mitochondrial targeting signal sequence, and protein translated from the second ATG lacks this sequence and is destined for the nucleus. Absence of mitochondrial Pif1 leads to mitochondrial DNA loss and a mitochondrial-deficient or petite yeast strain. Nuclear roles ascribed to Pif1 include maintenance of telomeric and ribosomal DNA.
PIF1 is a gene dosage suppressor of multiple top3 mutant defects:
Consistent with the finding that high-copy PIF1 suppresses top3Δ slow growth as evidenced on plate assays (Figure 1B), high-copy PIF1 shortens top3Δ doubling time in liquid culture (Figure 1C). The generation time of the top3Δ strain is 334 min and this is reduced to 194 min by p2μ-TOP3. High-copy PIF1 shortened the doubling time of the top3Δ strain to 244 min, demonstrating partial suppression of top3Δ slow growth. In contrast, the generation time of the wild-type strain (156 min) is slightly lengthened by p2μ-PIF1 (181 min) in our strain background (W303; W1588). This is consistent with previous reports that Pif1 overexpression causes slow growth in wild-type cells (Lahaye et al. 1991). In addition, we found that high-copy Pif1 (p2μ-PIF1) reduced the plating efficiency of wild-type cells (W1588-4C) from 100 ± 0.5% to 51 ± 2%. On the other hand, Pif1 overexpression did not reduce plating efficiency (39 ± 4%) in top3Δ cells (W1958 segregants), which normally have a plating efficiency of 33 ± 4%.
The effect of high-copy PIF1 on the cell cycle was examined by microscopy as described in materials and methods (Table 3). The cell-cycle profile of a log-phase wild-type culture is unaltered by p2μ-PIF1. Mutant top3Δ cells accumulate in G2/M and p2μ-PIF1 partially rescues this defect. These data suggest that Pif1 overexpression in top3 cells either decreases the length of G2/M and/or increases the chance of survival through G2/M.
Mutant top3 strains are highly sensitive to the DNA-damaging agents hydroxyurea (HU) and methylmethane sulfonate (MMS). High-copy PIF1 suppresses these sensitivities in the top3Δ strain at low drug concentrations, while at higher concentrations no suppression is evident (Figure 2), consistent with partial suppression of top3Δ defects.
Next, we tested the effect of PIF1 overexpression on the cellular localization of the homologous recombination protein Rad52. Rad52 typically exhibits diffuse nuclear localization but redistributes to discrete foci within the nucleus at sites of DNA double-strand breaks during S phase. These foci are thought to be sites of active DNA repair(Lisby et al. 2001, 2003). The majority of the G2/M cells that accumulate in top3Δ log-phase cultures each contain a single Rad52 focus, suggesting the presence of DNA damage (data not shown). High-copy Pif1 (p2μ-PIF1) does not alter the percentage of unbudded cells that have Rad52 foci in wild-type or top3Δ cells (Table 4). Pif1 overexpression also does not affect the percentage of Rad52 foci in wild-type budded cells. However, p2μ-PIF1 does reduce the incidence of Rad52 foci in top3Δ budded cells from 64 to 30%, suggesting that excess Pif1 suppresses the top3 mutant either by preventing DNA damage or promoting the repair of DNA damage.
Although Pif1 was not recovered in our screen for high-copy suppressors of lethality in the integrated top3–Y356F strain, Pif1 overexpression does strongly suppress the lethality caused by top3–Y356F overexpression (pGal-top3–YF) in a wild-type TOP3 strain (Figure 3). Galactose-induced expression of top3–YF, but not TOP3, is toxic, consistent with previous reports (Bennett and Wang 2001; Oakley et al. 2002). In contrast, in the presence of galactose, cotransformants of pGal-top3-YF and pGal-PIF1 grow as well as those containing the empty vector controls.
PIF1 gene dosage suppression of top3 defects is not general to all Pif1 DNA helicase family members. Rrm3, also a 5′ to 3′ DNA helicase, is the closest homolog to Pif1 in S. cerevisiae (reviewed in Bessler et al. 2001). Overexpression of RRM3 (pGal-RRM3) does not suppress and, to the contrary, exacerbates top3Δ slow growth (data not shown). Similarly, pGal-RRM3 does not suppress and instead exacerbates the toxicity of top3–Y356F overexpression in a wild-type strain background (data not shown).
The requirement of Pif1 catalytic activity for dosage suppression of top3 defects:
To test whether Pif1 helicase activity is necessary for gene dosage suppression of top3 defects, we substituted arginine for the lysine at amino acid position 264 (pif1-KR) in the high-copy PIF1 vectors. This conserved residue is essential for catalytic activity in all helicases tested (Gorbalenya and Koonin 1993) and pif1–K264R behaves like a null allele in vivo (Zhou et al. 2000).
Characterization of pif1-KR overexpression in wild-type cells:
Similar to that seen with p2μ-PIF1, p2μ-pif1-KR slightly lengthens the doubling time of the wild-type strain (W1588-4C) from 156 to 185 min (compare to 181 min for p2μ-PIF1), demonstrating a negative effect even in the absence of helicase function. Also similar to p2μ-PIF1, p2μ-pif1-KR does not affect the cell-cycle profile or Rad52 focus frequency in the wild-type strain (Table 3 and Table 4). There are, however, some differences between pif1-KR and PIF1 overexpression in wild-type cells. While p2μ-PIF1 decreases the wild-type plating efficiency from 100 ± 0.5% to 52 ± 2%, p2μ-pif1-KR decreases it only to 72 ± 8%, indicating that PIF1 toxicity in wild-type cells is partly, but not entirely, due to excess helicase activity. On the other hand, for mitochondria, p2μ-pif1-KR has a toxic effect that p2μ-PIF1 does not exhibit. The frequency of spontaneous petite formation in the wild type (W1588-4C) is 1.1 ± 0.1% and this is increased to 30 ± 3% by p2μ-pif1-KR, while p2μ-PIF1 has little or no effect (0.5 ± 0.1%).
Characterization of pif1-KR overexpression in top3 cells:
The p2μ-pif1-KR plasmid suppresses defects as well as p2μ-PIF1 in a top3Δ PIF1 strain background. It also reduces the doubling time of the top3Δ PIF1 cells from 335 to 246 min (compare to 244 min with p2μ-PIF1) and partially suppresses the G2/M cell-cycle defect (Table 3) and the frequency of Rad52 foci in budding cells (Table 4). These results might suggest that the mere presence of excess Pif1 protein, and not Pif1 catalytic activity, is important for suppressing top3Δ defects. However, we asked whether p2μ-pif1-KR could suppress top3Δ in the absence of the endogenous chromosomal PIF1 by assaying pCEN-TOP3 plasmid loss in a top3Δ pif1Δ strain transformed with either empty vector, p2μ-PIF1 or p2μ-pif1-KR (see materials and methods). Only p2μ-PIF1 was able to suppress the TOP3 requirement (data not shown). Thus, p2μ-pif1-KR does not suppress top3 defects in a top3Δ pif1Δ background. These results demonstrate that p2μ-pif1-KR cannot suppress top3Δ unless at least one functional copy of PIF1 is present.
Pif1 catalytic activity is required for pGal-PIF1 suppression of the toxicity associated with top3–Y356F overexpression in a wild-type strain (Figure 3). This result further supports the importance of Pif1 catalytic activity for the suppression of top3 defects. Moreover, cotransformants containing pGal-top3-YF and pGal-pif1-KR actually fare worse than those containing pGal-top3-YF and the empty vector control. The synergy caused by the combination of these two catalytic inactive proteins suggest that Pif1 and Top3 compete for the same substrate (see discussion).
In addition, our characterization of pif1-KR overexpression in the top3 background demonstrates that gene dosage suppression must be unrelated to Pif1 mitochondrial function. Similar to what was seen in the wild type, in top3Δ cells (W1958 segregants), p2μ-pif1-KR has a toxic effect on mitochondria with petite formation frequencies of 1.6 ± 0.9%, 1.2 ± 0.5%, and 38.5 ± 2.5% for empty vector, p2μ-PIF1 and p2μ-pif1-KR, respectively. Thus p2μ-PIF1 and p2μ-pif1-KR suppress top3 equally well, despite this negative effect of excess pif1-KR on mitochondria. Given these data, as well as our observation that complete mitochondrial deficiency (rho0) does not suppress top3 (data not shown), gene dosage suppression of top3 defects must, by default, be related to Pif1 nuclear role(s).
The double pif1 top3 mutant is synthetic lethal in an Sgs1-dependent manner:
The genetic interactions among pif1Δ, top3Δ, and sgs1Δ alleles were examined by standard segregation analysis (Figure 4A). No viable pif1Δ top3Δ haploids were isolated, demonstrating synthetic lethality. pif1Δ top3Δ spores germinate with wild-type frequency but grow to microcolonies of 2 to 100 swollen large-budded cells suggestive of failure during G2/M. pif1–K264R top3Δ is similarly synthetic lethal, demonstrating that Pif1 catalytic activity is required in the absence of Top3 (data not shown). In addition, the pif1-m2 allele that selectively eliminates nuclear Pif1 is synthetic lethal with top3Δ, demonstrating that a nuclear function of Pif1 is important in the absence of Top3 (data not shown). In contrast, sgs1Δ pif1Δ top3Δ and sgs1Δ pif1Δ segregants are viable and grow as well as the pif1Δ mutant alone (Figure 4A). pif1Δ, sgs1Δ, and sgs1Δ pif1Δ strains were assayed for HU and MMS sensitivity (Figure 4B and Figure 6D, rows 1–4). The pif1Δ strain is mildly sensitive to these drugs compared to the sgs1Δ strain. Taking into account the smaller colony size caused by the pif1Δ petite phenotype, the sgs1Δ pif1Δ double mutant resembles the sgs1Δ mutant. These results indicate that there is no sgs1Δ pif1Δ synthetic interaction. Taken together, these genetic analyses reveal that Pif1 DNA helicase activity is essential for dealing with nuclear DNA damage that arises in the top3 mutant strain when Sgs1 is present.
The genetic interaction of Pif1 with the Sgs1–Top3 pathway is not a general property of the Pif1 DNA helicase family. The rrm3Δ top3Δ combination is synthetic lethal; however, this lethality is not suppressed by sgs1Δ. (data not shown). Furthermore, in agreement with previous reports, the sgs1Δ rrm3Δ combination is inviable (Tong et al. 2001; Ooi et al. 2003; Schmidt and Kolodner 2004; Torres et al. 2004).
Sgs1 catalytic activity is required for Pif1–Top3 genetic interactions:
We further examined the effects of sgs1Δ as well as other several sgs1 mutations, on Pif1–Top3 genetic interactions. The sgs1-ΔN82 mutation deletes the N terminus of Sgs1 that physically interacts with Top3, mimicking a top3 mutation even in the presence of wild-type TOP3 (Duno et al. 2000; Fricke et al. 2001). The sgs1-K706R (sgs1-KR) mutation eliminates Sgs1 helicase activity by substituting arginine for the conserved lysine at amino acid position 706 (Lu et al. 1996).
sgs1Δ suppresses many top3 defects, including slow growth and sensitivity to HU and MMS. Both sgs1Δ and sgs1Δ top3Δ strains are mildly sensitive to HU and MMS and p2μ-PIF1 does not suppress but, to the contrary, exacerbates these drug sensitivities (Figure 2). These results indicate that Pif1 interaction with the Sgs1–Top3 pathway is specific to DNA damage that arises when Sgs1 acts in the absence of Top3.
In the sgs1-ΔN82 strain, p2μ-PIF1 partially suppresses slow growth (Figure 5A) and MMS sensitivity (Figure 5B, rows 7 and 8). In contrast, when Sgs1 catalytic activity is eliminated in the context of this mutation (sgs1-ΔN82, KR) p2μ-PIF1 compromises growth (Figure 5A) and increases MMS drug sensitivity (Figure 5B, rows 9 and 10). These results corroborate that Pif1 overexpression is beneficial specifically when Sgs1 helicase activity is uncoupled from Top3.
sgs1-KR suppresses pif1Δ top3Δ lethality, demonstrating that the pif1Δ top3Δ synthetic interaction is due to Sgs1 helicase activity (Figure 6A). We next looked for genetic interactions between the different sgs1 mutant alleles and pif1Δ in a wild-type TOP3 background. The sgs1-ΔN82 pif1Δ combination is synthetic lethal and this is suppressed by removal of Sgs1 catalytic activity (sgs1-ΔN82, KR) (Figure 6B and 6C). Taking into account the smaller colony size caused by the pif1Δ petite phenotype, both the pif1Δ sgs1-KR and pif1Δ sgs1-ΔN82, KR double mutants exhibit growth and MMS sensitivity similar to the respective sgs1 single mutants (Figure 6D, rows 5–8). Thus the synthetic interaction between pif1Δ and the Sgs1–Top3 pathway also occurs specifically when Sgs1 helicase activity is uncoupled from Top3.
Taken together, these genetic analyses strongly suggest that Pif1 is important when the Sgs1 DNA helicase creates a substrate that is not accessible to resolution by the Top3 topoisomerase (see discussion).
The Pif1–Top3 genetic interaction is recombination dependent:
Eliminating homologous recombination suppresses the top3 growth defect (Oakley et al. 2002; Shor et al. 2002). We found that p2μ-PIF1 does not suppress rad52 top3 or rad52 slow growth, consistent with the notion that PIF1 high-copy suppression of top3 defects does not occur in the absence of homologous recombination (data not shown). In addition, rad51Δ, rad52Δ, rad54Δ, or rad55Δ deletions suppress the pif1Δ top3Δ synthetic defect, demonstrating that pif1Δ top3Δ synthetic lethality is recombination dependent (Figure 7 and data not shown). Neither rad9 nor mad2 mutation suppresses pif1Δ top3Δ, indicating that the synthetic defect is independent of the DNA damage and spindle checkpoints (data not shown). We found that the top3Δ rrm3Δ and sgs1Δ rrm3Δ synthetic interactions are similarly recombination dependent and checkpoint independent (data not shown).
sgs1, top3, sgs1 top3, pif1, and rrm3 mutants had different effects on recombination between repeated sequences. Three distinct repetitive loci, CUP1, rDNA, and SUP4, were examined. The sgs1 and top3 mutations cause hyperrecombination at all three loci (Table 5 and Wallis et al. 1989; Shor et al. 2002). Mutation of rrm3 causes hyperrecombination at rDNA and CUP1 (Table 5 and Keil and McWilliams 1993), but has no effect at the SUP4 locus (Table 5). Mutation of pif1 causes hyperrecombination at CUP1 and SUP4, but has little effect at the rDNA locus (Table 5). Thus, while sgs1 and top3 appear to have a general effect on recombination between repeated sequences, pif1 and rrm3 have locus-specific effects that differ from each other. These results corroborate that Pif1 and the Sgs1–Top3 pathway do not share a global common function and are consistent with Pif1 and Rrm3 also having distinct functions.
Pif1 localizes to DNA repair foci in top3 mutant cells:
Pif1 localization was studied in live cells using a Pif1–YFP fusion protein. PIF1–YFP exhibits no synthetic interaction with top3Δ and no associated mitochondrial deficiency, demonstrating that it produces functional Pif1 protein (data not shown). We found that Pif1 is present in the mitochondria as seen by colocalization with the mitochondrial protein Cox4 (Figure 8A). Pif1 also exhibits nuclear localization and is largely concentrated in the nucleolus as evidenced by colocalization studies with the nuclear membrane marker Nup49 and the nucleolar protein Nop1 (Figure 8B). Occasionally, Pif1 spontaneously redistributes to a single nuclear focus that is outside of the nucleolus. These spontaneous Pif1 foci are rare in G1 (0.6%) cells, but increase in frequency as cells enter the S (7%) and G2/M (12%) phases of the cell cycle (Table 6). These results suggest that Pif1 is recruited to nuclear foci during DNA replication and/or chromosome segregation.
Colocalization studies with Rad52–CFP revealed that almost all nuclear Pif1 foci (>96%) colocalize with a Rad52 focus at all phases of the cell cycle (Table 6). Conversely, only a subset of the Rad52 foci colocalize with a Pif1 focus (35% during S phase and 77% during G2/M phase). The extent of Pif1 and Rad52 focus colocalization was unaffected by strain background (top3 mutant vs. wild type) or cell treatment (ionizing radiation). Overall, more spontaneous Pif1 foci are detected in G2/M cells and the extent of Pif1 focus colocalization with Rad52 is highest in G2/M cells. This may suggest that recruitment of Pif1 to the DNA repair focus is a relatively late event that occurs after Rad52 recruitment. However, Pif1 foci still form in a rad52Δ strain and Rad52 foci still form in a pif1Δ strain, indicating that Pif1 and Rad52 can participate in independent DNA repair pathways (data not shown).
Ionizing radiation causes DNA double-strand breaks and induces Rad52 focus formation in budding cells (Lisby et al. 2001). We asked whether Pif1 nuclear foci are similarly induced. Cells were exposed to 4 krad, a gamma ray dose that gives two DNA double-strand breaks per cell on average (Lisby et al. 2001) and monitored Rad52–YFP and Pif1–CFP focus formation for 3 hr postirradiation (Figure 8D). The percentage of unbudded (G1) cells with a Rad52 (4.2%) or Pif1 (3.1%) focus was low and remained relatively constant over the course of the experiment. Consistent with previously published results (Lisby et al. 2001), the frequency of budded (S/G2/M) cells with a Rad52 focus increased from 19% pre-irradiation to 85% at 150- and 180-min postirradiation. The frequency of budded cells with a Pif1 nuclear focus paralleled what was seen for Rad52, increasing from 9.5% pre-irradiation to 66% at 150- and 180-min postirradiation. These results suggest that Pif1 foci form in response to DNA double-strand breaks caused by gamma irradiation. However, a pif1Δ strain is not sensitive to 80 krad, the highest gamma ray dose tested, demonstrating that Pif1 is not required to survive this type of DNA damage.
Strikingly, in the top3Δ strain, the incidence of spontaneous Pif1 foci increases to 18% in S-phase cells (from 7% in the wild type) and 66% in G2/M cells (from 12% in the wild type) (Table 6 and Figure 8E). Except for the increased frequency of Pif1 foci in these top3 mutant cells, Pif1 localization appears wild type (Figure 8). Pif1 localization and the frequency of Pif1 nuclear foci in sgs1Δ and sgs1Δ top3Δ strains are similar to what is observed in the wild-type strain (data not shown). Taken together, these data suggest that the increased frequency of Pif1 recruitment to DNA repair foci in top3 cells must be due to DNA damage that occurs when Sgs1 acts in the absence of Top3.
In contrast to our results with Pif1 localization, a functional Rrm3–YFP fusion protein did not form visible foci in wild-type, sgs1Δ, or top3Δ cells. This suggests that the ability to form a nuclear focus is not general to this family of DNA helicases but is specific to Pif1.
While it is clear that Sgs1 and Top3 work together in a pathway that promotes genome stability, their cellular DNA substrate(s) remains an area of intense investigation. Previous genetic screens designed to provide clues to Sgs1–Top3 function have included searches for mutations that are synthetic lethal in combination with sgs1 (Mullen et al. 2001; Tong et al. 2001; Ooi et al. 2003; Schmidt and Kolodner 2004; Torres et al. 2004) and searches for mutational suppressors of top3 mutant slow growth (Gangloff et al. 1994b; Shor et al. 2002). Genes that synthetically interact with sgs1 include srs2, rrm3, and members of the mus81-mms4, slx5-slx8, and slx1-slx4 heterodimeric complexes. Mutations in these genes are also synthetic lethal with top3, indicating that these gene products likely act in pathways that have functional overlaps with the Sgs1–Top3 pathway. An exhaustive search for mutational suppressors of top3 slow growth identified, in addition to sgs1, members of the RAD52 epistasis group of proteins required for homologous recombination, contributing to the idea that Sgs1–Top3 is involved in the resolution of recombination intermediates (Shor et al. 2002).
A current model to explain RecQ–Topo III biology suggests that RecQ creates a DNA substrate that is resolved by Topo III. Furthermore, it appears that RecQ helicase activity is toxic in the absence of Topo III, since Topo III mutant defects are suppressed by mutation of RecQ homologs in budding and fission yeasts (Gangloff et al. 1994a; Murray et al. 1997; Stewart et al. 1997). In S. cerevisiae, a catalytically inactive top3 allele causes a more severe phenotype than the null allele and is also suppressed by sgs1, suggesting that the catalytically inactive Top3 protein binds to and stabilizes a toxic Sgs1-created substrate. In vivo and in vitro data support the notion that RecQ–Topo III is involved in the resolution of recombination intermediates at the end of DNA replication (reviewed in Heyer et al. 2003). We presume that the toxic DNA structure that persists in top3 mutants is a normally transient recombination intermediate of the Sgs1–Top3 pathway that arises during DNA replication.
The Pif1 DNA helicase genetically interacts with the Sgs1–Top3 pathway:
Here, we show that high-copy Pif1 can suppress top3 mutant defects (Figures 1–3⇑ and Tables 3–4) and that the pif1 top3 mutant combination is synthetic lethal (Figure 4). Furthermore, our data demonstrate that these interactions between Pif1 and the Sgs1–Top3 pathway are specific to the condition that occurs when Sgs1 is uncoupled from Top3. High-copy Pif1 does not suppress sgs1Δ or sgs1Δ top3Δ defects and even makes these strains more sensitive to DNA damaging agents (Figure 2C–2D). Furthermore, sgs1Δ suppresses pif1Δ top3Δ synthetic lethality and pif1Δ sgs1Δ mutants have no synthetic defect (Figure 4). These observations indicate that Pif1 and Sgs1 do not share a common function, which is not surprising given the opposite polarities of their helicase activities. Our results also demonstrate that Pif1 overexpression does not alleviate the need for the Sgs1–Top3 pathway. Rather, our data are consistent with Pif1 being required to deal with a toxic DNA structure that is created by Sgs1 and that accumulates when Top3 is absent.
Pif1 helicase activity is required to counteract Sgs1 helicase activity in the absence of Top3:
Pif1 helicase activity is important for the interaction of Pif1 with the Sgs1–Top3 pathway. In the complete absence of wild-type Pif1 (top3Δ pif1Δ strains), high-copy pif1-KR cannot suppress top3 defects, demonstrating that some Pif1 catalytic activity, provided by at least one copy of PIF1, is essential for the suppression of top3 defects. Pif1 catalytic activity is also required for gene dosage suppression of the toxicity caused by top3-YF overexpression in a wild-type strain (Figure 3). Furthermore, in top3 mutants, Pif1 catalytic activity becomes essential as evidenced by top3 pif1–K264R synthetic lethality. Together, these results establish that Pif1 helicase activity is absolutely required to promote survival in the absence of Top3.
We also show that the interaction of Pif1 with the Sgs1–Top3 pathway manifests specifically when Sgs1 helicase activity is uncoupled from Top3. For example, in the absence of Sgs1 catalytic activity (sgs1–K706R top3 and sgs1–K706R mutants), Pif1 overexpression is no longer beneficial and is instead toxic (Figure 5B, rows 5 and 6 and data not shown). Additionally, while an sgs1 mutation that does not physically interact with Top3 and mimics a top3 phenotype (sgs1-ΔN82) is suppressed by Pif1 overexpression, eliminating Sgs1 catalytic activity in this context (sgs1-ΔN82, K706R) renders excess Pif1 toxic (Figure 5A and 5B, rows 7–10). Thus, Pif1 gene dosage suppression of top3 defects is dependent upon an active Sgs1 helicase. Similarly, pif1 top3 synthetic lethality is evident only when the Sgs1 helicase is active since it is suppressed by introducing the sgs1–K706R mutation (Figure 6A). Consistent with this observation, the synthetic lethality of sgs1-ΔN82 pif1 is suppressed when the helicase activity of sgs1-ΔN82 is eliminated (sgs1-ΔN82, K706R) (Figure 6B and 6C). These results demonstrate that Sgs1 helicase activity that is uncoupled from Top3 is lethal in cells that lack Pif1. Taken together, our results establish that Pif1 helicase activity is essential to counteract Sgs1 helicase activity and promote survival in the absence of Top3.
Known roles of Pif1 cannot account for its genetic interaction with the Sgs1–Top3 pathway:
Roles ascribed to Pif1 to date include maintenance of rDNA and regulation of telomerase-dependent DNA replication (Schulz and Zakian 1994; Ivessa et al. 2000; Zhou et al. 2000; Mangahas et al. 2001; Myung et al. 2001). We found that deletion of the rDNA tandem array does not suppress pif1 top3 synthetic lethality or top3 slow growth (data not shown). We also found that mutations that eliminate telomerase-dependent telomere replication (e.g., est2 or ku70) (Myung et al. 2001) do not suppress pif1 top3 synthetic lethality or top3 slow growth (data not shown). These results demonstrate that the genetic interaction between Pif1 and the Sgs1–Top3 pathway is unrelated to the roles of Pif1 in rDNA maintenance and at telomeres. Thus Pif1 must have additional nuclear functions that account for the pif1 top3 genetic interaction.
The work presented here indicates a broader role for Pif1 in maintenance of nuclear genome integrity. Previous studies showed that lack of Pif1 does not affect chromosome loss, reciprocal recombination, or gene conversion (Schulz and Zakian 1994). However, at CUP1 and SUP4, two repetitive loci outside of the rDNA array, we find that pif1 mutants are hyperrecombinant, suggesting that Pif1 has a general role in the maintenance of repeated sequences (Table 5). We also examined pif1 mutants for sensitivity to several DNA damaging agents including HU that leads to stalled replication forks, gamma rays that create dsDNA breaks, and MMS that creates alkylated DNA adducts that are likely processed to dsDNA breaks. Although pif1 null mutants are mildly sensitive to HU (Figure 4B), otherwise wild-type cells that lack mitochondria (rho0) exhibit similar sensitivity, suggesting that pif1 HU sensitivity is due to its mitochondrial defect (unpublished). pif1 cells are MMS sensitive (Figure 4B) while rho0 cells are not (unpublished), demonstrating that nuclear Pif1 provides protection against MMS-induced DNA damage. In contrast, pif1 mutants are not gamma ray sensitive (results and data not shown). These observations indicate that there is something specific to the type of damage induced by MMS that calls for Pif1 and not the presence of dsDNA breaks per se. Interestingly, top3 mutants are also highly MMS sensitive but not gamma ray sensitive (unpublished), suggesting an overlapping role for Pif1 and Top3 in dealing with certain types of DNA damage.
The role of Pif1 in the Sgs1–Top3 pathway is downstream of recombination:
Several lines of evidence place the essential role of Pif1 in top3 mutants downstream of homologous recombination. While high-copy Pif1 suppresses a top3 mutant, it does not suppress the slow growth of rad52 top3 double mutants, suggesting that Pif1 gene dosage suppression of top3 defects does not occur in the absence of recombination. Even more convincing, pif1 top3 synthetic lethality is suppressed by deleting members of the Rad52 epistasis group of homologous recombination genes, demonstrating that the lethal event in pif1 top3 strains occurs downstream of recombination (Figure 7). These results indicate that the genetic interaction of Pif1 with the Sgs1–Top3 pathway is recombination dependent.
We find that the majority of G2/M cells that accumulate in a top3 culture contain a single Rad52 DNA repair focus, corroborating the idea that Sgs1 creates a toxic recombination intermediate that persists at the end of DNA replication in top3 mutants. Pif1 overexpression suppresses both the cell-cycle defect and the increased frequency of DNA repair foci in top3 mutants, suggesting that the mechanism of suppression is related to prevention, resolution, or repair of stalled recombination intermediates in top3 mutants (Table 3 and Table 4). Furthermore, when we examined Pif1 cellular localization, we found that Pif1 itself localizes to the frequent DNA repair foci found in top3 mutant budding cells (Figure 8 and Table 6). While 35% of Rad52 foci in top3 mutant S-phase cells contain Pif1, this number increases to 77% in G2/M cells, suggesting that Pif1 localization to the DNA repair center is a late event. These data are consistent with a scenario in which Pif1 is recruited to DNA repair centers in top3 mutant cells to deal with toxic DNA structures created by Sgs1.
What is the mechanism by which Pif1 helicase activity suppresses toxicity caused by Sgs1 in the absence of Top3?
An emerging theme has been that RecQ–Topo III is important for processing recombination intermediates that form during DNA replication (reviewed in Heyer et al. 2003). Double Holliday junctions are a proposed intermediate in the repair of collapsed replication forks as well as in the postreplication repair of gaps left as the replication fork bypasses DNA lesions. Classically, Holliday junctions are thought to be resolved by a resolvase activity that yields a recombinogenic crossover event half of the time. In vitro studies demonstrate a concerted Blm–Topo IIIα activity that resolves double Holliday junctions without crossing over and in vivo evidence in yeast suggests that Sgs1–Top3 possesses a similar activity (Ira et al. 2003; Wu and Hickson 2003; Wu et al. 2005). In the current model for Sgs1–Top3 resolvase activity, Sgs1 molecules act on opposing DNA strands to branch migrate two Holliday junctions toward each other, creating a hemicatenated strand interlink between two chromosomes that is a substrate for decatenation by Top3 (central panel in Figure 9 and reviewed in Heyer et al. 2003). In the absence of Sgs1 (sgs1 and sgs1 top3 strains), pathways that act in parallel provide alternate means of resolution (white box in Figure 9). In the absence of Top3, Sgs1 molecules act on the double Holliday junction to create a DNA structure that persists and causes characteristic top3 mutant defects (shaded box in Figure 9). Our results clearly demonstrate that Pif1 helicase activity is required to counteract Sgs1 helicase activity in top3 mutants.
How does Pif1 counteract Sgs1 activity? One possibility is that Pif1 strips Sgs1 protein from DNA, thus preventing the cell from committing to a defunct repair pathway in top3 mutants. Such a role in protein removal from DNA has been described for the Srs2 DNA helicase in disassembly of Rad51 filaments (Krejci et al. 2003; Veaute et al. 2003). Another possibility is that Pif1 recognizes and rewinds a toxic DNA substrate created by Sgs1, leaving it accessible to repair by other pathways. Consistent with this idea, Sgs1 and Pif1 unwind DNA with opposing polarities and Pif1 activity is stimulated by forked DNA structures (Lahaye et al. 1991; Lahaye et al. 1993; Bennett et al. 1998, 1999). Here, we show that Sgs1 helicase activity is required for the genetic interaction of Pif1 with Sgs1–Top3, and this lends credence to the idea that the Sgs1-created DNA substrate per se is important for the interaction. In further support of the idea that Pif1 acts on an Sgs1-created substrate, we show that excess pif1-KR suppresses top3 defects in cells with at least a single copy of Pif1 (top3 PIF1) but not in top3 pif1 cells, suggesting that as long as some functional Pif1 is present, the catalytically inactive protein stabilizes/protects the substrate for Pif1 until a functional Pif1 arrives to act at the site. Consistent with this idea, Pif1 does exhibit distributive activity in vitro with the protein readily binding/releasing its substrate (Lahaye et al. 1993).
These two proposed roles for the requirement of Pif1 in top3 mutants, removal of Sgs1 protein from its DNA substrate by Pif1 vs. the active rewinding of an Sgs1-created substrate by Pif1, are not necessarily mutually exclusive. It is also possible that Sgs1 normally remains bound at the ssDNA–dsDNA junction after DNA unwinding (branch migration) until Top3 arrives and that mere removal of Sgs1 from the DNA at this point permits passive rewinding (reverse branch migration). Consistent with this idea, in vitro studies demonstrate that Top3 specifically recognizes the complex of Sgs1 bound to its DNA substrate (Harmon et al. 1999, 2003) and our data suggest that Top3 and Pif1 compete for the same substrate (Figure 3). Since we find no evidence for direct protein–protein interaction between Sgs1 and Pif1 by two-hybrid analysis (unpublished), we favor the idea that Pif1 specifically recognizes and actively rewinds the DNA structure created by Sgs1. In vitro studies of possible mechanistic interactions between Pif1 and Sgs1 are needed to further elucidate this interesting genetic interaction that is apparent only in top3 mutant backgrounds.
We thank Ralph Keil, Robert Reid, Qi Feng, Justin Weinstein, and Hui Zou for plasmids and yeast strains. We thank Brent Wells for technical assistance with the screen for gene dosage suppressors of top3-ts and analysis of Rrm3 localization. We are grateful to Marco Foiani, Peter Thorpe, Michael Lisby, Erika Shor, and Robert Reid for critical reading of the manuscript. This work was supported by National Cancer Institute National Research Service Award grant CA90047 to M.W. and National Institutes of Health grant GM50237 to R.R.
Communicating editor: M. Lichten
- Received September 27, 2004.
- Accepted June 30, 2006.
- Copyright © 2006 by the Genetics Society of America