In eukaryotic cells, S phase can be reversibly arrested by drugs that inhibit DNA synthesis or DNA damage. Here we show that recovery from such treatments is under genetic control and is defective in fission yeast rqh1 mutants. rqh1+, previously known as hus2+, encodes a putative DNA helicase related to the Escherichia coli RecQ helicase, with particular homology to the gene products of the human BLM and WRN genes and the Saccharomyces cerevisiae SGS1 gene. BLM and WRN are mutated in patients with Bloom‘s syndrome and Werner's syndrome respectively. Both syndromes are associated with genomic instability and cancer susceptibility. We show that, like BLM and SGS1, rqh1+ is required to prevent recombination and that in fission yeast suppression of inappropriate recombination is essential for reversible S phase arrest.
Multiple cellular mechanisms ensure the accurate transmission of genetic material from one generation to the next. These include enzymes which repair specific DNA lesions after DNA damage, as well as regulators that coordinate DNA repair and cell cycle progression. Mutations abolishing many of these processes result in genomic instability and cause cancer‐prone syndromes in humans, including xeroderma pigmentosum (excision repair), colorectal cancer (mismatch repair) and ataxia telangiectasia (checkpoint control) (see Bootsma et al., 1995; Chung and Rustgi, 1995; Enoch and Norbury, 1995; Lehmann and Carr, 1995; Chu and Mayne, 1996). Chromosomes are likely to be particularly vulnerable to DNA damage during replication, as DNA is decondensed and partially single stranded and therefore highly accessible to damaging agents. DNA replication is, therefore, subject to exquisite regulation which ensures its coordination with DNA repair and other cell cycle processes. Control of DNA replication initiation has been studied extensively and a picture of the molecular mechanisms involved is beginning to emerge (see Peeper et al., 1994; Chevalier and Blow, 1996).
Replication elongation is also likely to be highly regulated, although much less is known about these controls. The presence of DNA damage or limiting concentrations of nucleotides in the cell are known to block DNA elongation (Vassilev and Russev, 1984; Friedberg et al., 1995). Although arrest of replication elongation could be due to mechanical or chemical blocks to DNA polymerization, it is equally possible that the pausing and resumption of DNA replication are regulated by external controls, like many other cell cycle processes (Hartwell and Weinert, 1989; Murray, 1992). Indeed, elegant studies in the budding yeast Saccharomyces cerevisiae have demonstrated that the slowing of DNA replication in response to DNA damaging agents requires gene products previously shown to be involved in other cell cycle checkpoints (Paulovich and Hartwell, 1995; Longhese et al., 1996). Here we demonstrate that reversible S phase arrest also requires protective functions that are distinct from cell cycle checkpoint controls.
We have previously studied S phase arrest in the fission yeast Schizosaccharomyces pombe, by isolating mutants that are sensitive to hydroxyurea (HU), a drug that blocks DNA replication by depleting deoxynucleotides. Our screening strategy was based on the observation that checkpoint‐defective cells undergo an aberrant mitosis (‘cut’) when treated with HU (Enoch and Nurse, 1990). Under the same conditions normal cells cease DNA synthesis and arrest cell division, displaying an elongated cell morphology. By screening for mutants that ‘cut’ in HU, a number of checkpoint‐defective HU‐sensitive (hus) mutants were identified (Enoch et al., 1992). Several of the mutants were found to be allelic to previously known rad (radiation‐sensitive) genes, including rad3+, shown independently by Al‐Khodairy and Carr to be required for checkpoint control (Al‐Khodairy and Carr, 1992). rad3+ is related to the human ataxia telangiectasia gene, ATM (Bentley et al., 1996), mutation of which causes a variety of severe symptoms, including increased rates of cancer (see Lehmann and Carr, 1995).
A single allele of hus2, hus2‐22 (rqh1‐h2), was also identified in the screen for hus mutants (Enoch et al., 1992). For reasons that will become clear below, we have renamed this gene rqh1+. Like the other hus mutants, rqh1− cells undergo an aberrant ‘cut’‐like mitosis in HU and are also radiation sensitive. Here we show that, unlike checkpoint mutants, rqh1− cells arrest DNA replication and cell division normally in response to HU but then display significant defects in chromosome segregation in the subsequent mitosis. We propose that rqh1+ is required for recovery from S phase arrest. Our studies also indicate that rqh1+ is required for recovery from cell cycle arrest induced by DNA damage.
We have sequenced rqh1+ and found that it encodes a putative DNA helicase related to the products of the human BLM (Bloom‘s syndrome) and WRN (Werner's syndrome) genes (Ellis et al., 1995; Yu et al., 1996) and to the S.cerevisiae SGS1 gene (Gangloff et al., 1994; Watt et al., 1995). Bloom‘s and Werner's syndromes are rare congenital diseases associated with genomic instability and significant cancer predisposition (German, 1993; Yu et al., 1996). Sgs1 has been identified as a protein that interacts genetically and physically with type I and type II topoisomerases (Gangloff et al., 1994; Watt et al., 1995; Lu et al., 1996). The rqh1+, BLM, WRN and SGS1 genes encode proteins of similar sequence and length and contain a central domain closely related to the DNA helicase domain of the Escherichia coli RECQ gene product.
Increased genomic instability in BLM− cells is caused at least in part by elevated levels of genetic exchange, particularly between sister chromatids. Exchanges between sister chromatids and homologs can be readily detected cytologically in BLM− cells and such exchanges are further stimulated by exposure to DNA damaging agents (Krepinsky et al., 1979, 1982; Heartlein et al., 1987; Kurihara et al., 1987). Recombination is also elevated in sgs1− cells (Gangloff et al., 1994; Watt et al., 1996). Here we show that rqh1+ is also required to prevent recombination, particularly during S phase arrest. Thus, the function of this subfamily of DNA helicases in regulating genetic exchange and maintaining genomic stability has been highly conserved during evolution. Moreover, prevention of inappropriate recombination is required to ensure that cell cycle progression can resume normally if S phase has been interrupted.
rqh1+ encodes a RecQ‐like DNA helicase
The rqh1+ gene, previously called hus2+, was cloned from a genomic library by complementation of the UV sensitivity of the rqh1‐h2 (hus2‐22) mutant (see Materials and methods). Two overlapping complementing clones were isolated, both containing the same open reading frame. The sequence of the open reading frame predicts rqh1+ to encode a large, 1328 amino acid protein containing a 325 amino acid domain that shows sequence homology with the helicase domains of RecQ‐like DNA helicases. This sequence was subsequently reported by the Sanger Center S.pombe genome sequencing project, accession No. Q09811. Our sequence, which agrees with that from the Sanger Center, has DDBJ/EMBL/GenBank accession No. Y09426. Related proteins are the human Blm, Wrn and RecQL proteins, S.cerevisiae Sgs1 and E.coli RecQ (Figure 1A). Figure 1B shows a sequence alignment of the helicase domains of the members of the RecQ‐like family of helicases, including Rqh1. The Rqh1 helicase domain shows most similarity to the Blm helicase domain (55% amino acid identity) and least similarity to the Wrn helicase domain (40% amino acid identity). While all six members of the RecQ helicase family share sequence homology within the core helicase domain, Rqh1 appears to belong to a subfamily consisting of Rqh1, Blm, Wrn and Sgs1. These four proteins are approximately the same length, considerably longer than the E.coli prototype. They also show an extended region of sequence homology and have a similar positioning of the helicase domain (see Figure 1A). In addition, although the N‐ and C‐termini of Rqh1, Blm, Wrn and Sgs1 show little sequence similarity, they are all rich in charged and polar amino acids, especially serines, and have patches of acidic residues (Figure 1A) (Ellis et al., 1995; Rothstein and Gangloff, 1995; Yu et al., 1996).
Analysis of the rqh1 deletion strain
The rqh1 null phenotype was determined by disruption of the rqh1+ gene in the S.pombe genome. Most of the rqh1+ open reading frame, including the entire N‐terminus, all of the region encoding the helicase domain and half of the C‐terminus, was replaced with the ura4+ gene (see Figure 2A and Materials and methods). The strain carrying the rqh1 deletion (rqh1Δ) is viable, although in rich medium, cultures of rqh1Δ cells, like those of rqh1‐h2, have a 38% longer doubling time than wild‐type cells (data not shown; Enoch et al., 1992). rqh1Δ cells show no other obvious defects under normal growth conditions. Appropriate crosses established that rqh1Δ and rqh1‐h2 are allelic.
We have previously shown that the rqh1‐h2 mutant is sensitive to the DNA replication inhibitor hydroxyurea (HU) and to UV irradiation (Enoch et al., 1992). The HU sensitivities of rqh1Δ cells (TE767, see Table 1) and rqh1‐h2 cells (TE232, see Table 1) were compared. During a 10.5 h incubation in 10 mM HU, wild‐type cells remained largely viable, but the viability of rqh1Δ and rqh1‐h2 cells dropped >50‐fold (Figure 2B). The sensitivities of rqh1Δ, rqh1‐h2 and wild‐type cells to UV irradiation were also examined. Unlike wild‐type cells, both rqh1Δ and rqh1‐h2 cells were markedly sensitive to low doses of UV irradiation, with the viability of rqh1− cells dropping >500‐fold upon irradiation at 200 J/m2 (Figure 2C). Since rqh1Δ and rqh1‐h2 cells behaved identically in these assays, we believe that rqh1‐h2 is a null allele.
rqh1Δ cells undergo aberrant mitoses upon treatment with HU or UV irradiation
rqh1‐h2 mutants were initially identified in a screen for checkpoint mutants that were unable to arrest the cell cycle in response to HU. Like known checkpoint mutants, rqh1 mutants are hypersensitive to HU and HU‐treated rqh1‐h2 cells undergo septation in the absence of chromosome segregation (Enoch et al., 1992). This results in cells where the septum has either bisected the single nucleus or divided the cell such that one daughter is anucleate. We refer to this as the ‘cut’ phenotype, as it resembles the phenotype of previously described cut mutants, which have defects in chromosome segregation (Hirano et al., 1986). In contrast, wild‐type cells arrest the cell cycle when they are incubated in HU and become elongated, because they continue to grow without dividing.
rqh1Δ cells, like rqh1‐h2 mutants, show the ‘cut’ phenotype when incubated in HU, while wild‐type cells do not (compare Figure 3A and B). The ‘cut’ cells are first observed after 7 h incubation in HU (see Figure 3E) and are noticeably elongated, indicating that the aberrant mitosis occurs after a cell cycle delay (the cells in Figure 3A and B had been incubated in HU for 9 h). This is in contrast to checkpoint mutants, which do not elongate at all in HU because cell cycle progression is not delayed. Cells with the ‘cut’ phenotype also accumulate much more rapidly in cultures of checkpoint mutants than they do in rqh1Δ cultures, being readily observed after only 3 h incubation in HU (Enoch et al., 1992).
As shown in Figure 3B, by the time the rqh1Δ cells cut, wild‐type cells are no longer particularly elongated, suggesting that they have re‐entered the cell cycle after a period of cell cycle arrest. To investigate whether this was the case, wild‐type (TE271, see Table I) and rqh1‐h2 (TE232, see Table I) cells were incubated in 10 mM HU and the number of cells in the culture and the percentage of cells showing the ‘cut’ phenotype were analyzed at regular intervals. As Figure 3E shows, the cell number continues to increase in cultures of both wild‐type and rqh1− cells for the first 3 h after addition of HU. This is because exponentially growing S.pombe cells are predominantly in the G2 phase of the cell cycle, so after addition of HU these cells divide once before they encounter the S phase block. For the next 4 h neither wild‐type nor rqh1− cells divide, showing that the cell cycle is blocked. After ∼7 h in HU, however, the number of cells in both the wild‐type and rqh1− cultures starts to increase once more, showing that the cells have overcome the HU‐induced block and have re‐entered the cell cycle. This establishes that HU delays rqh1− and wild‐type cell cycles to the same extent. Thus, by definition, rqh1− cells do not have a defect in the S–M checkpoint (Hartwell and Weinert, 1989). Resumption of the cell cycle after 7 h in HU is consistent with previous studies showing that HU only blocks cell division temporarily in S.pombe (Sazer and Nurse, 1994). Microscopic analysis shows that rqh1− cells showing the ‘cut’ phenotype only start to accumulate as wild‐type and rqh1− cells re‐enter the cell cycle (Figure 3E). Thus rqh1− cells may ‘cut’ because they are unable to segregate chromosomes after S phase arrest.
We also investigated the morphology of rqh1Δ cells (TE767, see Table I) after UV irradiation (see Materials and methods). Figure 3C shows rqh1Δ cells 13 h after irradiation at 150 J/m2. Like rqh1Δ cells treated with HU, UV‐irradiated rqh1Δ cells ‘cut’. In contrast, UV‐irradiated wild‐type cells and unirradiated rqh1Δ cells do not show the ‘cut’ phenotype (Figure 3D and data not shown). Like HU‐treated cells, UV‐irradiated rqh1− cells that are ‘cutting’ are elongated compared with irradiated wild‐type cells (compare Figure 3C and D), indicating that aberrant mitosis has taken place after a cell cycle delay. Cell number measurements confirm that, as in HU‐treated cells, the aberrant mitosis in UV‐irradiated rqh1− cells does not occur until wild‐type cells have recovered from UV‐induced arrest and re‐entered the cell cycle (data not shown). This suggests that, like rqh1− cells incubated in HU, UV‐irradiated rqh1− cells may have a chromosome segregation defect but are not checkpoint defective.
‘Cut’ rqh1− cells have completed DNA replication
Previously described S–M checkpoint mutants ‘cut’ in HU without replicating their DNA. To investigate whether ‘cut’ rqh1− cells have completed DNA replication, we examined the DNA content and cell morphology of wild‐type and rqh1− cells during recovery from HU arrest. Cultures of wild‐type (TE271, see Table I) and rqh1‐h2 (TE232, see Table 1) cells were incubated in medium containing 10 mM HU for 4 h, then filtered into fresh medium without HU. Samples were taken from the culture at 20 min intervals after removal of HU and fixed for FACS analysis, to determine the DNA content of the cells, and examined microscopically for ‘cuts’ (see Materials and methods). As shown in Figure 4A, wild‐type and rqh1− cells resume and complete DNA replication with similar kinetics after HU arrest. In both wild‐type and rqh1− cells DNA replication starts between and 20 and 40 min after removal of HU and is complete after 80 min. Figure 4B shows that rqh1− cells showing the ‘cut’ phenotype were first observed 80 min after removal of HU and were still accumulating at 140 min, well after DNA replication was complete. Thus rqh1− cells are able to arrest the cell cycle upon inhibition of S phase and can complete bulk DNA replication in a normal manner once HU has been removed from the culture. However, even though DNA has been replicated, normal chromosome segregation cannot take place in many of the cells. These results suggest that correct chromosome segregation after S phase arrest requires special functions that are absent in rqh1 mutant cells.
rqh1+ is required for accurate chromosome segregation, especially after HU arrest
The above experiments establish that rqh1− cells are defective in chromosome segregation after HU treatment. Similar aberrant mitoses are observed in fission yeast top2 and cut mutants, which lack functions required for chromosome segregation (Uemura and Yanagida, 1984; Hirano et al., 1986). To analyze the chromosome segregation defect in rqh1 mutants further, we measured the rate of loss of a non‐essential minichromosome from rqh1− and wild‐type cells.
rqh1Δ and wild‐type strains of S.pombe containing the ade6‐M210 allele at the genomic ade6 locus and a single copy of the non‐essential, centromeric minichromosome Ch16 (Niwa et al., 1986) were constructed (TE786 and TE788 respectively, see Table I and Materials and methods). Since Ch16 contains the ade6‐M216 allele, which shows intragenic complementation with ade6‐M210 (Leupold and Gutz, 1964), cells containing the minichromosome are ade+, but those that have lost Ch16 are ade−. The chromosome loss rates of wild‐type and rqh1− cells were calculated by determining the proportion of cells that became ade− in a known number of generations (see Materials and methods). As shown in Figure 5A, under normal growth conditions the rate of chromosome loss for wild‐type cells was 1.2×10−4 per generation, in close agreement with the rate previously observed by Niwa et al. (1986). In contrast, rqh1− cells were 15 times more likely to lose Ch16, with a loss rate of 1.8×10−3 per generation. These results suggest that rqh1− cells lack functions required for accurate chromosome segregation even under normal growth conditions, which may account for the longer doubling time of cultured rqh1− cells (see above). Chromosome loss rates were also measured in cells following a 4 h incubation in 10 mM HU (for details see Materials and methods). Under these circumstances the wild‐type loss rate was elevated 4.7‐fold to 5.6×10−4 per generation, while the rate of chromosome loss from rqh1− cells was elevated to a rate of 1.5×10−2 per generation. Thus the chromosome loss rate from HU‐treated rqh1− cells was more than 8‐fold higher than untreated rqh1− cells and nearly 30‐fold higher than the rate of loss from HU‐treated wild‐type cells (Figure 5A). Indeed, minichromosome loss rates after HU treatment of rqh1− cells may be even higher than those figures stated here. After a 4 h incubation in HU ∼65% of the rqh1− cells were unable to recover from the S phase arrest and many went on to form the ‘cut’ phenotype. Since the ‘cut’ phenotype could be an extreme manifestation of a chromosome segregation defect, many of the cells that did not survive the incubation in HU may have shown chromosome loss. As we can only measure loss rates for those cells that survive, our numbers may represent an underestimation for minichromosome loss after HU treatment. UV‐irradiated rqh1− cells also display significantly elevated rates of chromosome loss (data not shown). Thus rqh1+ function appears to be required for proper chromosome maintenance under normal growth conditions and is even more important after S phase arrest or DNA damage.
rqh1+, like BLM, negatively regulates recombination
As described above, the rqh1+ gene product is structurally similar to the product of the BLM gene. BLM− cells show substantially elevated levels of recombination, with a particularly striking increase in sister chromatid exchange (SCE). SCE levels in BLM− cells are further elevated by treatment with agents that induce DNA damage (Krepinsky et al., 1979, 1982; Heartlein et al., 1987; Kurihara et al., 1987). Recombination rates are also elevated in S.cerevisiae cells lacking the SGS1 gene (Gangloff et al., 1994; Watt et al., 1996). To determine whether rqh1− cells have a similar defect, the rate of mitotic recombination between ade6 heteroalleles was measured in homozygous rqh1− and wild‐type diploids. Stable wild‐type and rqh1‐h2 diploid cells were constructed that were heterozygous at the ade6 locus, carrying different ade6 alleles with mutations at either end of the ade6 gene (TE745 and TE747 respectively; see Table I and Materials and methods). Homologous recombination occurring between the heteroalleles results in restoration of the ade+ phenotype. The levels of recombination in these wild‐type and rqh1− cells under normal growth conditions and after a 4 h incubation in HU were calculated by determining the proportion of cells that became ade+ in a known number of generations (see Materials and methods).
The levels of recombination occurring under normal growth conditions were almost identical in wild‐type and rqh1− cells, with rates of 7.50×10−7 and 7.65×10−7 per generation respectively (Figure 5B, left panel). This is comparable with rates determined in S.cerevisiae for similar intervals (Watt et al., 1996). After a 4 h incubation in 10 mM HU, recombination was stimulated in both cultures, but the effect was much more dramatic in rqh1− cells (Figure 5B, right panel; note that a different scale is used in each panel). While the wild‐type recombination rate increased 80‐fold to 6.14×10−5, the rqh1− rate increased >800‐fold to 6.36×10−4. Thus rqh1+, like BLM and SGS1, prevents recombination, particularly after S phase arrest. Like the minichromosome loss data, our figures for recombination rates after HU treatment of rqh1− cells may be lower than the true rates, since the cells that died upon HU treatment may have the highest levels of recombination.
Our results show that rqh1+ is required to prevent excessive recombination after S phase arrest. High levels of recombination could cause the chromosome segregation defect in HU‐treated rqh1− cells, as it may be impossible to segregate sister chromatids that are linked via recombination intermediates.
rqh1− cells are defective in recovery from S phase arrest
The rqh1‐h2 mutant was first identified in a screen for HU‐sensitive checkpoint mutants. Here we describe characterization of the rqh1 deletion strain and the rqh1‐h2 mutant, which appears to be a null allele (Figure 2). Like checkpoint mutants, rqh1− cells are hypersensitive to HU and rqh1− cells incubated in HU display the ‘cut’ phenotype (Figures 2 and 3). However, unlike checkpoint mutants, rqh1− cells arrest cell division in response to HU. Moreover, when released from the S phase arrest, rqh1− cells complete bulk DNA replication in an apparently normal fashion (Figure 4). Nevertheless, rqh1− cells are unable to segregate their chromosomes in the mitosis after S phase arrest, leading to the ‘cut’ phenotype. These experiments show that rqh1 mutants are not checkpoint defective, but rather are unable to recover from S phase arrest. Thus reversible S phase arrest entails more than stopping and resuming DNA synthesis. Additional processes, which are dependent on rqh1+, are required to ensure normal segregation of chromosomes in the mitosis that follows arrest of DNA replication. rqh1+ may be required either while DNA replication is arrested or after it resumes, to ensure that the arrest can be reversed.
Under normal growth conditions rqh1− mutants have a modestly elevated generation time and show enhanced rates of minichromosome loss. Although we were not able to observe elevated recombination levels in untreated rqh1− cells, we have only analyzed recombination between one pair of heteroalleles and only over a small interval and thus our assay may not have been sensitive enough to detect increased recombination. It is possible that elevated recombination levels might be revealed if larger intervals were examined. Thus we believe that rqh1+ could have a function during the normal cell cycle which becomes essential after S phase arrest or DNA damage. Alternatively, rqh1+ might play a different role during the normal cell cycle, unrelated to its function after S phase arrest or DNA damage.
Genetic control of S phase arrest and recovery
Analysis of the fission yeast checkpoint genes has previously shown that normal S phase arrest requires at least two genetically controlled processes; arrest of mitosis by negative regulation of the cyclin dependent kinase Cdc2 and induction of functions that are necessary to allow recovery of cells after S phase arrest (Enoch et al., 1992; Al‐Khodairy et al., 1994). Both recovery and checkpoint control are abolished by mutations in checkpoint rad genes (hus1+, rad1+, rad3+, rad9+, rad17+ and rad26+). However, mutations in cell cycle control genes (cdc2+, cdc25+, weel+ and mik1+) only abolish the checkpoint function (Enoch and Nurse, 1990; Enoch et al., 1991, 1992; Sheldrick and Carr, 1993). Here we show that recovery, but not checkpoint control, is abolished by mutation of the rqh1+ gene. A possible model linking checkpoint control and recovery is presented in Figure 6. The checkpoint rad gene products are proposed to function early during S phase arrest, possibly generating a signal that engages the cell cycle checkpoint and activates rqh1+‐dependent recovery processes. Thus checkpoint control and recovery are both dependent on checkpoint rad gene function, although they are independent processes. Interactions between the checkpoint rad gene products and downstream gene products could be direct or indirect.
rqh1− cells are also sensitive to UV irradiation. Again this is not due to a checkpoint defect, as rqh1− cells arrest upon UV irradiation and only ‘cut’ when they re‐enter the cell cycle and try to segregate their chromosomes. Possibly this UV sensitivity is also due to a defect in S phase recovery, as UV‐induced lesions could be inhibiting ongoing DNA replication. Previous studies have shown that mutants lacking S phase recovery functions are particularly sensitive to UV during S phase (Al‐Khodairy et al., 1994).
It is remarkable that arrest and recovery in response to HU or UV irradiation in fission yeast requires two genes, rad3+ and rqh1+, related to human genes mutated in two different cancer‐prone syndromes, ataxia telangiectasia and Bloom's syndrome. This suggests that reversible S phase arrest may be a critical aspect of the response to DNA damage and prevention of cancer in higher eukaryotes. The fission yeast mutants will provide valuable tools for further investigation of the molecular mechanisms involved in this process.
Failure to recover from S phase arrest is associated with increased recombination
S phase arrest enhances loss rates of a non‐essential minichromosome and elevates recombination in rqh1 mutants. These results suggest that the role of rqh1+ in recovery from S phase arrest could be to prevent excess recombination. High levels of recombination may explain the ‘cut’ phenotype of rqh1− cells after treatment with HU; if cells enter mitosis with sister chromatids entangled by unresolved recombination intermediates, subsequent chromosome segregation would be difficult or impossible. This model is consistent with studies showing that certain types of entangled chromosomes are not sensed by cell cycle checkpoints. For example, S.pombe and S.cerevisiae mutants lacking topoisomerase II also ‘cut’, as mitosis is initiated even though sister chromatids are catenated (Uemura and Yanagida, 1984; Holm et al., 1985; Uemura et al., 1987).
rqh1+ could prevent recombination by a number of mechanisms. It could be a negative regulator of recombinases, with a particularly important function during S phase arrest, when a considerable amount of recombinogenic single‐stranded DNA is present in the cell. Alternatively, the absence of rqh1+ could cause DNA damage, leading to the generation of recombinogenic lesions. For example, it has been proposed that the RecQ helicases function together with topoisomerases to resolve replication intermediates at the end of S phase (Rothstein and Gangloff, 1995). It will be important to investigate interactions between rqh1+ and S.pombe topoisomerases, particularly since the ‘cut’ phenotype observed after HU treatment resembles the phenotype caused by reduced topoisomerase II function (Uemura and Yanagida, 1984, 1986; Uemura et al., 1987).
The role of DNA helicase enzymatic activity in the regulation of recombination by the RecQ‐related helicases remains to be determined. Although Sgs1 has been shown to be a helicase biochemically, an S.cerevisiae sgs1 mutant lacking helicase activity is wild‐type with regard to its interactions with top1 and top3 mutants (Lu et al., 1996). However, the effect of these mutations on recombination is not known. Possibly helicase activity may be required for some but not all of the in vivo functions of these proteins.
Control of recombination by a conserved family of RecQ helicase‐related proteins
Based on structural features, rqh1+ encodes the fourth member of a subfamily of putative DNA helicases related to the product of the E.coli RECQ gene. In addition to the helicase domain, the four members of this subfamily are approximately the same size and have other structural similarities. The other members of the subfamily are the human Blm and Wrn proteins and the S.cerevisiae Sgs1 protein (Figure 1). BLM−, sgs1− and rqh1− cells show enhanced levels of genetic exchange, suggesting that the function of the RecQ‐related proteins has been conserved during evolution. Loss of BLM function is associated with increased rates of carcinogenesis, suggesting that proper regulation of genetic exchange plays a vital role in the maintenance of genetic integrity. The Wrn protein also plays a role in maintenance of the genome and tumor prevention, although it may not regulate genetic exchange, as SCE is not elevated in WRN− cells (Bartram et al., 1976; Darlington et al., 1981; Gebhart et al., 1988).
Although both SGS1 and rqh1+ prevent recombination, loss of the gene has different biological consequences in the two yeasts, as rqh1− fission yeast are UV sensitive (Figure 2) and sgs1− cells are not (Watt et al., 1996). As we show here, the fission yeast rqh1+ gene also plays a vital role in recovery from S phase arrest, while SGS1 is not known to function in cell cycle control. Thus each yeast provides a unique opportunity to investigate the consequences of unregulated genetic exchange and, therefore, to understand the function of this new class of tumor suppressors.
Materials and methods
Genetic and molecular methods
Plasmids and strains were constructed using standard techniques (Sambrook et al., 1989; Moreno et al., 1991). All strains and plasmids used in this report are listed in Table I. Unless otherwise stated, ‘wild‐type’ refers to strain TE271, which is 972 h−; ‘rqh1‐h2’ refers to strain TE232 and ‘rqh1Δ’ refers to strain TE767 (see Table I). Schizosaccharomyces pombe medium was prepared as previously described (Moreno et al., 1991).
Molecular analysis of rqh1+
The rqh1‐h2 (hus2‐22) strain was transformed with the S.pombe genomic libraries pURSP1 and pURSP2 (Barbet et al., 1992) and clones that complemented the radiation sensitivity of this strain were isolated, as described in Murray et al. (1992). Two independent overlapping clones were isolated, one of ∼4850 bp (rqh1A) and one of ∼6000 bp (rqh1B). 5151 bp of the rqh1B insert was sequenced using nested deletions and shown to contain an open reading frame of 3987 bp with no introns. Subsequently this sequence was reported by the Sanger Center S.pombe genome sequencing project, accession No. Q09811. Our sequence, which agrees with that from the genome project, has also been submitted to the DDBJ/EMBL/GenBank database, accession No. Y09426.
Sequence alignments of the predicted rqh1+ gene product with other members of the RecQ‐like DNA helicase family were carried out using the DNASTAR Megalign program. Similarities between the helicase domains of the members of this family were calculated using the GCG Bestfit program.
Construction of the rqh1 deletion mutant
The disruption construct was created by replacing the 3.6 kb NheI–AgeI fragment of rqh1B (see Figure 2A) with a linker containing a NotI site:
This construct was unable to complement the HU sensitivity of the rqh1‐h2 allele. An insert bearing the rqh1 deletion was removed from the pUR19 vector by digestion with SacI and SphI and ligated into SacI and SphI digested pUC19. The 1.7 kb ura4+ gene was then inserted into the NotI digested rqh1 deletion construct as described previously (Barbet et al., 1992). This plasmid (pTE436, see Table I) was digested with SacI and SphI and the linear disruption construct, consisting of the ura4+ gene flanked by rqh1+ sequences, was isolated. This DNA was used to disrupt one copy of the rqh1+ gene by the one‐step disruption method in an h+/h+ ura4‐D18/ura4‐D18 stable diploid (TE480, see Table I; a generous gift from G.Cottarel). The homologous integration event was confirmed by Southern blotting. This strain was crossed to an h−/h− ura4‐D18/ura4‐D18 stable diploid (TE558, see Table I) to generate a sporulating diploid heterozygous for the rqh1 deletion. These diploids were sporulated, tetrads were dissected and haploids that were HU sensitive and ura+ were identified. The HU‐sensitive and ura+ phenotypes were found to co‐segregate and to segregate 2:2 in all of the tetrads analyzed. One such haploid was picked for further analysis and Southern blot analysis was used to confirm deletion of the rqh1+ gene (TE767, see Table I). Further crosses established that rqh1Δ was allelic to rqh1‐h2.
Analysis of HU and UV response
UV and HU sensitivity were determined as previously described (Al‐Khodairy and Carr, 1992; Enoch et al., 1992). Cells were fixed for microscopy and analyzed for ‘cut’ formation as previously described (Enoch et al., 1992). Unless otherwise noted, cells were grown on rich medium at 29°C. The number of colonies growing on two plates was counted for each viability measurement and each number was expressed as a proportion of the number of viable cells for that cell type at the start of the experiment. In every case the starting measurement corresponded to at least 480 cells. To investigate cell number increase in the presence of HU, cultures were grown to early log phase and HU was added to a final concentration of 10 mM. Samples were removed at the indicated times for analysis of cell number using a hemocytometer and scoring of ‘cuts’. To examine ‘cut’ formation after UV irradiation, 1×107 cells from an early log phase culture were plated on each of two 23 cm2 plates. The plates were irradiated at 150 J/m2 and cells were then washed off the plates using 20 ml pre‐warmed medium and incubated at 29°C. Samples were removed and examined microscopically for ‘cuts’. To analyze recovery from S phase arrest (Figure 4), cultures were grown to early log phase and then arrested by the addition of HU to a final concentration of 10 mM. Cells were incubated for 4 h, filtered onto 0.45 μM HA filters (Millipore, Bedford, MA) and resuspended in fresh medium. Samples were removed at the indicated time points for FACS analysis (Shazer and Sherwood, 1990) and scoring of ‘cuts’.
Determination of chromosome loss rates
The rate of loss of Ch16 (Niwa et al., 1986) from wild‐type and rqh1− cells under normal growth conditions was calculated using two different methods. In the first method, chromosome loss was measured in the progeny of a single ade+ cell after a known number of generations. In the second method, the increase in the number of cells that had lost Ch16 in a population of cells was determined after approximately one generation.
For the first method, TE786 and TE788 strains (see Table I) were grown in the absence of adenine to select for the presence of Ch16 and then streaked onto YE plates. After 3 days growth at 29°C, whole colonies of cells were dispersed in 1 ml YE medium. Cell number was measured using a hemocytometer and used to calculate the total number of cells in the colony, which can be used to determine the number of generations since the original ade+ progenitor. The cells were then diluted into YE medium, sonicated briefly and plated onto YE plates at a concentration of ∼1000 cells/plate. We did not see any difference between the growth rates of ade+ and ade− cells on YE plates. After 2.5–3 days growth at 29°C the cells were replica‐plated onto EMM plates supplemented with 7 μg/ml adenine, on which ade− cells are a pink color. Two days later the proportion of ade− colonies was calculated. Only those colonies which were completely pink (and had therefore arisen from a single ade− progenitor cell) were scored. For each wild‐type colony dispersed in YE medium, ∼37 000 colonies were screened for their ade phenotype; for each rqh1Δ colony, ∼6000 colonies were screened. Rates of loss of Ch16 were determined for two wild‐type colonies and 10 rqh1Δ colonies.
For the second method, cultures of cells were grown under selective conditions to ensure maintenance of Ch16 to an OD595 of 0.05–0.15. The cells were then filtered and resuspended in an equal volume of YE medium and incubated for 6 h. Samples were removed at the beginning and the end of the incubation period and cell number and the proportion of ade− segregants was determined as described above. In each of two experiments ∼37 000 wild‐type colonies were counted from the first time point and 40 000 from the last time point, and 37 000 rqh1Δ colonies were counted from the first time point and 15 000 from the last time point.
The rate of loss of Ch16 in each type of experiment was calculated using the following formula, adapted from Murakami et al. (1995).
where R0 and Rn are the proportion of ade+ cells 0 and n generations after removal of selection respectively. In experiments using a single colony, R0 was taken as 1, since the single progenitor cell for each colony was ade+.
The rates of loss of the minichromosome from wild‐type cells using the first method were calculated as 1.2×10−4 and 5.0×10−5 for each of two separate experiments, those using the second method were 1.2× 10−4 and 2.5×10−4 for each of two separate experiments. We believe that differences in the rates between individual experiments are due to experimental error and are not significant. This is because the rates of loss of the minichromosome from rqh1Δ cells in the absence of HU calculated using the first method described above vary between 6.9× 10−4 and 3.36×10−3, even though each experiment was carried out using identical strains and techniques. Thus the two different methods of calculating the rate of loss of the minichromosome from cells under normal growth conditions give essentially the same results.
The overall loss rate was taken as the median value from 12 individual experiments for rqh1Δ cells and four individual experiments for wild‐type cells.
To determine chromosome loss rates after incubation in HU, cells were grown under selective conditions to early mid log phase as described above, filtered and resuspended in rich medium (time 0). HU was then added to the cultures to a final concentration of 10 mM and they were incubated at 29°C. After 4 h the cells were filtered again, resuspended in fresh medium without HU and incubated for a further 5 h (time n). Samples were removed at time 0 and time n and cell number and chromosome loss rates were measured as described above. Cell number was also determined in samples before and after each filtration step, so that final measurements of cell number increase could be corrected for losses that occurred during filtration. In each of two experiments ∼37 000 wild‐type colonies were counted from time 0 and 40 000 from time n; 37 000 rqh1Δ colonies were counted from time 0 and 13 000 from time n. The rate of chromosome loss from the HU‐treated cells was calculated using the formula described above. The number of generations, n, was calculated from the cell number increase between time 0 and time n, taking account of any change in dilution that occurred at the filtration step. The overall loss rate for HU‐treated wild‐type and rqh1Δ cells was taken to be the average of the loss rates determined in two separate experiments.
Measurement of recombination rates
The rate of mitotic recombination was calculated for the interval between two different ade6 mutations; ade6‐M26, which is a G→T substitution at G135, and ade6‐L469, which is a C→T substitution at C1467. Reciprocal crossing over between these heteroalleles within this interval will lead to one copy of the ade6 gene with both mutations and one wild‐type copy of ade6, and the cell with therefore become ade+. In order to calculate the rate of recombination under normal growth conditions, strain TE730 was crossed with strain TE744 and strain TE728 was crossed with strain TE725 (see Table I). The presence of the mat2‐102 allele in TE728 and TE730 prevents sporulation of the resulting diploids. Conjugants (TE747 and TE745 respectively, see Table 1) were streaked onto EMM plates supplemented with adenine, but not histidine or leucine; under these conditions only diploids resulting from conjugation can grow. After 3 days growth at 29°C, whole colonies of cells were dispersed in 330 μl YE medium. The number of cells in the colony (and therefore the number of generations that the single diploid progenitor cell had gone through) was estimated by determining cell number using a hemocytometer. To measure ade+ recombinants, samples were plated on EMM plates supplemented with 50 mg/l guanine, which prevents residual growth of ade− colonies (Grossenbacher‐Grunder and Thuriaux, 1981). Aliquots were also diluted and plated onto EMM plates supplemented with adenine to determine the total number of viable cells. All the plates were allowed to grow at 29°C for 3–5 days and the number of colonies on each plate was counted. The rate of recombination per generation for these cells was calculated using the formula given in the previous section, except under these circumstances R0 and Rn are the proportion of ade− cells 0 and n generations after the start of the experiment. Under the conditions described above R0 is 1, since the single progenitor cell for each colony was ade−. Rates were calculated for eight diploid wild‐type colonies and 10 diploid rqh1− colonies and the median value for each cell type was taken as the overall recombination rate.
The rate of recombination in cells treated with HU was determined as follows. Colonies of cells from freshly constructed diploids were grown in EMM supplemented with adenine to an OD595 of 0.1–0.25 (time 0). HU was then added to the culture to a final concentration of 10 mM and the cells were incubated at 29°C for a further 4 h (time n). Aliquots were removed from the cultures at time 0 and time n and percent recombinants was determined as described above. The rate of recombination per generation for these cells was calculated using the formula given in the previous section, where R0 and Rn are the proportion of ade− cells at time 0 and n respectively. The recombination rate in HU was calculated for four diploid wild‐type and five diploid rqh1− cultures and the median value was taken as the recombination rate in HU for that cell type.
We would like to thank Johanne Murray for many helpful comments and discussions and Jürg Kohli and James Haber for advice on experimental design. We would also like to thank Juanita Campos‐Torres for performing the FACS analysis, Hamid Ghazizadeh, Angela Vilché and Gladys Reimundo for excellent technical assistance and Fred Winston, Kristi Chrispell Forbes, Tim Humphrey, Cory Kostrub and Elizabeth Moynihan for valuable advice and comments on this manuscript. E.S. was supported by an EMBO long‐term fellowship and an HFSPO fellowship. Work in T.E.'s laboratory is supported by a grant from the NIH (GM50015).
- Copyright © 1997 European Molecular Biology Organization