A screen for mutants of budding yeast defective in meiotic gene conversion identified a novel allele of the POL3 gene. POL3 encodes the catalytic subunit of DNA polymerase δ, an essential DNA polymerase involved in genomic DNA replication. The new allele, pol3-ct, specifies a protein missing the last four amino acids. pol3-ct shows little or no defect in DNA replication, but displays a reduction in the length of meiotic gene conversion tracts and a decrease in crossing over. We propose a model in which DNA synthesis determines the length of strand exchange intermediates and influences their resolution toward crossing over.
HOMOLOGOUS recombination plays a critical role in maintaining genome integrity throughout cell division. In meiosis, homologous recombination is essential for proper homolog pairing and for the correct segregation of chromosomes at the first meiotic division. In vegetative cells, recombination plays an important role during DNA replication by providing a mechanism to bypass DNA lesions and other obstacles that block replication fork progression. Homologous recombination also provides a means to generate new combinations of genetic markers through gene conversion and crossing over, thereby generating genetic diversity among different individuals in the same population.
Numerous insights into the mechanism of homologous recombination have been obtained from meiotic studies using the convenient model organism, Saccharomyces cerevisiae (Paques and Haber 1999). Meiotic recombination in budding yeast is initiated by the formation of DNA double-strand breaks (DSBs) at recombination hotspots. The strands with 5′ ends at the site of the break are processed to expose single-stranded tails with 3′ termini (Figure 1). DSB formation and 5′-end resection are followed by invasion of an intact nonsister chromatid by only one of the two single-stranded tails (Hunter and Kleckner 2001). Single-end invasion results in hybrid DNA (hDNA) in which the two strands in a duplex are of different parental origin. If the two parental duplexes are genetically different within the region of strand exchange, the resulting hDNA contains mismatched base pairs and is referred to as heteroduplex DNA.
Single-end invasion intermediates can be channeled toward either of two repair pathways. In the first pathway, described by the DSB repair model (Szostak et al. 1983; Sun et al. 1989), DNA synthesis, capture of the second end, and ligation generate a double Holliday junction intermediate with asymmetric hDNA (i.e., hDNA on only one of the two duplexes) on each side of the DSB and on each chromatid (Figure 1, left). If a Holliday junction undergoes branch migration, then hDNA will be formed on both duplexes (symmetric hDNA). Eventually, double Holliday junction intermediates are resolved by cutting, at each junction, either both outside strands or both inside strands. Cutting of the two junctions in opposite directions generates crossovers, while cutting in the same direction generates noncrossovers.
The second pathway is described by the synthesis-dependent strand annealing (SDSA) model (Paques and Haber 1999) and supported by recent meiotic studies (Allers and Lichten 2001; Merker et al. 2003). According to the SDSA model (Figure 1, right), the invading strand is extended by DNA synthesis and then subsequently displaced. The newly synthesized DNA strand then anneals to the single-stranded tail on the other side of the break. DSB repair is completed by DNA synthesis and ligation. If extension of the invading strand creates a single-stranded tail that is longer than the exposed complement on the opposite side of the break, a flap structure with an exposed 3′ end will result after the two strands anneal and this flap will need to be removed to allow ligation. In this model, asymmetric hDNA occurs only on one side of the break and only on the chromatid that suffered the DSB. The SDSA repair pathway does not lead a priori to the formation of double Holliday junctions and therefore produces only noncrossover products. Although SDSA is a common mode of DSB repair in vegetative cells (Paques and Haber 1999), the extent to which it contributes to meiotic recombination remains unclear.
Although early intermediates in meiotic recombination have been characterized by extensive genetic, biochemical, and physical analyses, the later steps remain obscure. For example, it is not understood what determines the channeling of single-end invasion intermediates into either of the two repair pathways. In the DSB repair pathway, it is not known what controls the elongation of strand exchange intermediates, second-end capture, formation of Holliday junctions, branch migration, or junction resolution. This study provides insight into some of these issues.
We describe a novel allele of the POL3 gene of S. cerevisiae coding for the catalytic subunit of DNA polymerase δ. This mutation has little or no effect on DNA replication or DNA damage repair in vegetative cells. However, during meiotic recombination, the mutant produces shorter strand exchange intermediates and fewer crossover products. These results lead us to propose a role for Polδ in meiotic recombination beyond an involvement in simple gap repair.
MATERIALS AND METHODS
Isolation and mapping of the pol3-ct mutant:
The pol3-ct mutant was isolated using the homothallic (HO) strain S2702 (MATa/MATα and homozygous for ade2-1 HIS4-ura3-Stu-his4-260 leu2-3,112 lys2-1 thr1-4 trp1-289 ura3-1). After induction of sporulation and spore enrichment (Rockmill et al. 1991) spores were mutagenized with ultraviolet (UV) light to 50% survival and plated on YPD medium. Spore colonies were replica plated to sporulation medium and then to medium lacking uracil to assay the production of uracil prototrophs. The pol3-ct mutant was identified on the basis of a strong decrease (∼10-fold) in meiotic Ura+ prototroph formation.
Mapping of the pol3-ct mutation was carried out using multiply marked strains (X4119-15D, STX145-15D, STX82-3A, STX153-10C, STX75-3C, STX147-4C, X4120-19D, STX-6C, and STX155-9B) obtained from the Yeast Genetic Stock Center. Strains from our own collection carrying mutations in various meiotic genes (e.g., MSH5 and REC102) were also used.
Strains and plasmids:
Strains used to characterize the effect of pol3-ct in meiosis are derived from the haploid strains AS4 (MATα trp1-1 arg4-17 tyr7-1 ade6 ura3 MAL2) and AS13 (MATa leu2-Bst ura3 ade6) (Nag et al. 1989). AS4 is a HIS4 strain, while the PD strains are his4 strains derived from AS13 by transformation. PD75 is his4-ACG (a T to C transition in the initiating codon), PD5 is his4-519 (a 1-bp insert at +473), PD22 is his4-712 (a 1-bp insert at +1396), PD24 is his4-713 (a 1-bp insert at +2270), and DNY25 is his4-lopc (insertion of a palindromic sequence at the SalI site in HIS4). pol3-ct derivatives of these strains were constructed by the two-step transplacement procedure (Rothstein 1991). In the first step, strains were transformed with HindIII-digested LMδ1 (see below). Then Ura− derivatives of Ura+ transformants were selected on 5-fluoroorotic acid, creating strains LMP6 (AS4), LMP5 (PD75), LMP2 (PD5), LMP3 (PD22), LMP4 (PD24), and LMP8 (DNY25). Replacement of the POL3 allele by pol3-ct was verified by DNA sequencing.
Plasmid GR160 (obtained from Roland Chanet) containing the wild-type POL3 gene in YCp50 was used to complement the pol3-ct defect. The POL3 gene is toxic in many constructs in Escherichia coli (R. Chanet, personal communication), which probably accounts for our inability to clone POL3 on the basis of complementation of the mutant defect.
Plasmid LMδ1 was designed to allow replacement of the wild-type copy of POL3 at the endogenous locus by the pol3-ct allele. The 3′ end of the POL3 gene was amplified from a pol3-ct strain using primers p2862 (TAGTAGGAATTCTTGCTTCTGTCCGTCGTGA) and pR4173 (TAGAAGTCGACTAGCGCCCGAAGTCCTCACA). p2862 anneals starting at position +2503 within POL3, while pR4173 anneals downstream of POL3. The amplified fragment is 1311 bp in length. Underlined nucleotides within primers indicate restriction sites for EcoRI (p2862) and SalI (pR4173) that have been added at the 5′ end of the primers to clone the amplified fragment into plasmid pRS306 (New England Biolabs, Beverly, MA) treated with EcoRI and SalI. The HindIII site used for targeting is located 260 bp from the SalI site.
All phenotypic characterizations of pol3-ct were performed in the AS strain background except for mitotic prototroph formation, which was assayed in both AS strains and strain BR2495 (Rockmill and Roeder 1990) and its pol3-ct derivative LMR1/2. BR2495 has the following genotype: MATa/MATα his4-280/his4-260 leu2-27/leu2-3,112 arg4-8/ARG4 thr1-1/thr1-4 cyh10/CYH10 ade2-1/ade2-1 ura3-1/ura3-1 trp1-1/trp1-289.
Forward mutation, mitotic recombination, and irradiation assays:
Forward mutation to canavanine resistance and mitotic intragenic recombination were determined by fluctuation tests using the method of the median. Rates reported are the average of three (30°) or two (18°) independent experiments, each performed with nine independent 5-ml cultures set up from 2-day-old colonies and incubated at 30° or 18°.
Cells in stationary phase (UV) or exponentially growing (γ-rays) were washed in 0.9% NaCl and plated at appropriate dilutions on YPD and synthetic medium. UV irradiation was performed using a 264-nm source delivering 1 J/m2/sec. γ irradiation was performed using a 137Cs source. Cells were treated with 0, 100, 200, and 400 Gy. Survival was determined as the number of cells forming colonies on YPD medium after a given dose of irradiation divided by the number of colonies derived from cells not irradiated. Each experiment was carried out at least three times with qualitatively similar results.
For meiotic analysis of AS strains, cells were incubated in 5 ml of liquid YPD medium overnight, washed once with water, and then resuspended in 10 ml liquid sporulation medium and incubated at 18° for 5 days. Tetrads were treated with zymolyase in 1 m sorbitol for 10 min and dissected on YPAD plates. After 3 days, they were replica plated to appropriate omission media to follow the segregation of nutritive markers. Data were used from four-spore-viable tetrads only, which composed ∼60% of tetrads for both wild-type and pol3-ct strains. To compare nonparental ditype (NPD) ratios between wild-type and pol3-ct strains, a contingency chi-square test was performed for each interval, using the observed and expected NPD values for wild type and mutant.
Isolation and genetic mapping of the pol3-ct mutant:
A novel screen for mutants defective in meiotic gene conversion was devised. The starting strain is a HO diploid homozygous for the ura3-1 allele at the URA3 locus on chromosome V and for a ura3-Stu mutation inserted at the HIS4 locus on chromosome III. This strain was sporulated, and the spores were mutagenized with UV light to 50% survival. Due to mating-type switching followed by mating between cells of opposite mating type, the cells in spore colonies are diploid and homozygous at all loci except MAT. Since any newly induced mutations are homozygous, recessive mutations can be readily detected. The diploids still carry ura3-1 and ura3-Stu alleles and thus can be assayed for meiotic gene conversion by screening for the production of Ura+ prototrophs. A conversion-defective mutant that showed a 10-fold reduction in Ura+ prototrophs was chosen for further study.
Attempts to clone the gene from a yeast genomic library by complementation of the conversion defect proved unsuccessful. We therefore mapped the gene by tetrad analysis. The mutant was crossed to ura3-1 strains carrying 48 different markers dispersed throughout the yeast genome (Figure 2). Haploid segregants were scored for a defect in meiotic gene conversion by mating to tester strains carrying ura3-Stu and the conversion-defective allele; the resulting diploids were sporulated and screened for the production of uracil prototrophs. This mapping localized the mutation to the left arm of chromosome IV, 22 cM centromere proximal to MSH5 and 30 cM distal to MBP1.
In the region defined by mapping, the POL3 gene, which encodes the catalytic subunit of DNA polymerase δ (Polδ), was the only obvious candidate for the mutated gene. Consistent with this possibility, a plasmid carrying a wild-type copy of POL3 complements the mutant defect. When the POL3 gene was amplified from the mutant strain by PCR and sequenced, a point mutation was found in the POL3 open reading frame at position +3268, 11 bp before the last nucleotide (Figure 3A). This allele changes a leucine codon (TTA) to a stop codon (TAA). As a consequence, the protein is missing the last four amino acids (LSKW). The mutant was therefore named pol3-ct, for POL3 C-terminal truncation. The mutation does not lie in any of the conserved domains of the protein; in particular, pol3-ct does not affect the putative zinc finger domains also located near the carboxy terminus of the protein.
pol3-ct does not show growth or DNA repair defects:
Since POL3 is an essential gene responsible for the polymerase and proofreading activities of Polδ (Hindges and Hubscher 1997), it was important to determine the effects of pol3-ct in vegetative cells. Growth of pol3-ct mutants is normal at 18° and 30°; pol3-ct grows slightly slower than wild type at 37° (doubling time of 150 min for POL3 vs. 165 min for pol3-ct). These results indicate that DNA synthesis is not impaired in pol3-ct mutants as it is in thermosensitive mutants of POL3 that show cell cycle arrest when switched to nonpermissive temperatures (Conrad and Newlon 1983).
Replication defects caused by a deficient Polδ can result in the accumulation of mutations (Datta et al. 2000) and/or DNA lesions that are potential recombination substrates (Aguilera and Klein 1988). In addition, Polδ has been proposed to play a role in DNA damage repair (Giot et al. 1997). We therefore examined the effect of pol3-ct on mutation rate, mitotic recombination, and sensitivity to DNA-damaging agents. These experiments were carried out at 30°, the optimum temperature for yeast cell growth. In addition, since the meiotic experiments described below were carried out at 18°, a subset of vegetative phenotypes was also examined at this temperature. We found no change in mutation rate at the CAN1 locus at 30° and a modest (but statistically significant) twofold increase at 18° (Figure 3B). There was no change in mitotic intragenic recombination at either temperature (Figure 3, C and D). The pol3-ct mutant does not display enhanced sensitivity to UV irradiation (Figure 3, E and F) or to γ irradiation (Figure 3G).
pol3-ct changes the conversion gradient at HIS4:
The pol3-ct allele was originally found to decrease meiotic intragenic recombination between heteroalleles of the URA3 gene, suggesting that pol3-ct affects gene conversion. To determine which step(s) of meiotic gene conversion might be altered, conversion was examined at the HIS4 recombination hotspot (Nag et al. 1989). These experiments were carried out at 18°, the temperature at which the frequency of non-Mendelian segregation is maximal in the strain background used for this analysis.
Non-Mendelian segregation is observed at high frequency at HIS4 (Table 1) due to the presence of a meiotic DSB site just upstream of the gene (Fan et al. 1995). hDNAs spanning regions of heterologies within HIS4 contain mismatches. Correction of these mismatches can lead to gene conversion (3:1 or 1:3 segregations), whereas unrepaired mismatches lead to postmeiotic segregations (sectored colonies). An important feature of the HIS4 meiotic hotspot is that it shows a conversion gradient with its high end at the 5′ end of the gene next to the initiation site (Detloff et al. 1992).
Remarkably, the conversion gradient within HIS4 is strongly accentuated in the pol3-ct background (Table 1; Figure 4A). No effect on non-Mendelian segregation frequencies was seen for the most 5′ marker, indicating that initiation of recombination at HIS4 is not altered in a pol3-ct background. In contrast, there is a decrease in gene conversion for all other markers in HIS4. The effect of pol3-ct increases progressively within HIS4 to a maximum of sixfold at the 3′ end of the gene. We therefore conclude that formation of hDNAs starting at the initiation site upstream of HIS4 is not impaired in the pol3-ct mutant. However, these hDNAs extend shorter distances into HIS4 in pol3-ct than in wild type.
Previous studies have shown that polarity gradients are largely abolished when mismatch repair is inhibited (Detloff et al. 1992; Alani et al. 1994). Two observations indicate that the effect of pol3-ct on the HIS4 conversion gradient is not due to an alteration in the efficiency of mismatch repair. First, for none of the his4 alleles examined does the pol3-ct mutation increase the frequency of 5:3 and 3:5 segregations (indicative of a failure to repair) relative to 6:2 and 2:6 segregations (indicative of repair). Second, a HIS4 mutation (his4-lopc) that usually escapes mismatch repair (even in wild type) shows a decreased frequency of non-Mendelian segregation in the pol3-ct mutant (Table 1).
The pol3-ct effect on gene conversion is not specific to the HIS4 locus:
The decrease in meiotic gene conversion in pol3-ct is not specific to HIS4. A twofold decrease in conversion was also found at the ARG4 and TYR7 loci (Table 1). At the MAT locus, parental alleles in diploids differ by a heterology of 700 bp. An mlh1 mutant, which is deficient in mismatch repair, shows high frequencies of postmeiotic segregation at MAT, indicating formation of hDNA with large loops at this locus (Wang et al. 1999). Conversion of this heterology is decreased sixfold in pol3-ct (Table 1), strengthening the conclusion that the decrease in gene conversion does not depend on the nature of the segregating mutations. No decrease in conversion was observed for the leu2-Bst marker. On the basis of the results obtained at HIS4, it is likely that the leu2-Bst marker is close to a recombination initiation site.
Crossing over is decreased in pol3-ct strains:
pol3-ct strains show wild-type levels of sporulation (∼30%) and spore viability (∼90%). These results, together with the studies of gene conversion described above, indicate that DSB formation and repair occur as efficiently in pol3-ct strains as they do in wild type. To gain insight into the resolution of recombination intermediates in the mutant, crossover formation was analyzed by measuring map distances in four genetic intervals on three different chromosomes. In all intervals, map distances are decreased, on average to 66% of the wild-type level (Table 2; Figure 4B). These results suggest a role for DNA synthesis in promoting crossover formation.
Crossovers show interference in pol3-ct strains:
Meiotic crossovers are nonrandomly distributed such that two crossovers rarely occur close together, a phenomenon known as crossover interference. The effect of the pol3-ct mutation on interference was investigated by measuring NPD ratios, which can be roughly defined as the frequency of double crossovers observed in a marked interval divided by the number expected in the absence of interference (Papazian 1952; Snow 1979). An NPD ratio of <1.0 indicates positive interference. Crossover interference is not decreased in the pol3-ct mutant (Table 2; Figure 4C). In fact, in both intervals tested, the NPD ratio in the mutant is lower than that in wild type (indicating stronger interference); this difference is statistically significant only in the LEU2-MAT interval (P < 0.05).
pol3-ct is a separation-of-function allele:
Previous studies of the role of Polδ in recombination have been complicated by the use of temperature-sensitive alleles, which have modest defects in DNA replication even at permissive temperatures. Nevertheless, it has been possible to show that Polδ is required in vegetative cells for gene conversion events induced by irradiation and for repair of a DSB at the MAT locus (Fabre et al. 1991; Holmes and Haber 1999). This study is the first to demonstrate a role for Polδ in meiotic recombination. Our data demonstrate that the pol3-ct mutant has little or no defect in DNA replication or DNA damage repair in vegetative cells. In addition, we have found that pol3-ct does not affect spore formation or spore viability, indicating that premeiotic DNA replication is efficient. Thus, pol3-ct acts as a separation-of-function allele that specifically impairs a function of Polδ involved in meiotic recombination. This characteristic of the pol3-ct allele makes it a unique tool to investigate the role of Polδ in meiotic recombination, unimpeded by the complications associated with temperature-sensitive mutations.
Possible roles for Polδ in determining gene conversion tract length:
Our results suggest that pol3-ct strains initiate the wild-type number of meiotic recombination events; however, the average length of hDNA formed is shorter than that in wild type. It is not clear why the pol3-ct mutation was originally identified on the basis of decreased prototroph formation between URA3 heteroalleles. Perhaps both URA3 alleles are far from a DSB site; in this case, shortening tract length might decrease the fraction of events that reach either mutation.
What does the decrease in gene conversion tract length in pol3-ct strains tell us about the role of wild-type Polδ in meiotic recombination? Recombination events channeled through the DSB repair pathway necessitate the synthesis of new DNA sufficient to replace the sequences removed by processing of DSBs to expose single-stranded tails (Figure 1, left). Therefore, in this pathway, the length of asymmetric hDNA should be at least equivalent to the lengths of the single-stranded tails present at DSB sites. In contrast, in the SDSA pathway, displacement of the invading strand might occur before DNA synthesis extends the full length of the single-stranded tail on the other side of the DSB site (Figure 5A). In this case, repair synthesis would have to be completed after annealing of the displaced strand to the complementary single strand. The length of hDNA would then be confined to the length of DNA synthesis that occurred on the invaded chromatid. In the pol3-ct mutant, DNA repair synthesis might frequently stop earlier than in wild type, perhaps due to weakened processivity of Polδ during recombination.
How might the pol3-ct mutation affect tract length for gene conversion events that proceed through the DSB repair pathway? We propose a model in which Polδ does not simply serve to fill in single-stranded gaps, but rather plays a determining role in hDNA extension. According to the DSB repair model, DNA synthesis primed from the 3′ end of the invading strand enlarges the D-loop; the displaced DNA subsequently captures the second single-stranded tail, which in turn primes repair synthesis. We propose that Polδ subsequently increases the length of asymmetric hDNA by promoting DNA synthesis that is coupled to extended removal of the resected strand either by strand displacement or by further 5′ to 3′ resection (Figure 5B). In this scenario, DNA synthesis would be de facto coupled with junction migration; consequently, asymmetric hDNA could be extended prior to the formation of ligated double Holliday junctions. This hypothesis would explain nicely why asymmetric hDNAs at HIS4 (and at other recombination hotspots such as ARG4 and CYS3) frequently span the entire gene while the initial single-stranded tails appear to be no longer than 800 nucleotides (Sun et al. 1991; Bishop et al. 1992; Detloff and Petes 1992; Vedel and Nicolas 1999).
A specific interaction between Polδ and a component of the recombination apparatus might be required for synthesis-coupled junction migration. In the pol3-ct mutant, this interaction might not occur efficiently such that Polδ is removed upon reaching the first 5′ end, thus preventing further hDNA extension.
Possible roles for Polδ in meiotic crossing over:
In addition to altering the length of gene conversion tracts, pol3-ct also decreases the frequency of crossing over. How can this observation be reconciled with the models just presented (Figure 5)?
If single-end invasion intermediates sustain shorter tracts of DNA synthesis in pol3-ct (Figure 5B), this would facilitate strand displacement and favor the repair of DSBs through the SDSA pathway. Since SDSA is not associated with crossing over, a bias toward SDSA (vs. the DSB repair pathway) could account for the decrease in crossing in pol3-ct strains.
An alternative (not mutually exclusive) possibility is that pol3-ct decreases the probability of crossing over for events that proceed through the DSB repair pathway. Polδ might be involved in isomerization and/or the decision as to which strands are to be cut during Holliday junction resolution. For example, the resolvase that introduces nicks into strands of like polarity at Holliday junctions could be directed by Polδ to operate on strands with newly synthesized DNA. Such a bias would favor cutting the two junctions in opposite directions and thereby encourage crossover formation. Premature removal of Polδ would mean loss of the bias and a decrease in crossing over.
pol3-ct does not impair crossover interference:
A number of mutants (e.g., zip1, msh4) show a two- to threefold decrease in crossing over that is not associated with decreased initiation of recombination (Sym and Roeder 1994; Novak et al. 2001). These mutations also eliminate crossover interference. The pol3-ct mutation reduces crossing over without decreasing interference, raising the possibility that pol3-ct affects a different subset of crossovers from those eliminated by zip1 and msh4. The fact that interference appears slightly stronger in the pol3-ct mutant raises the possibility that pol3-ct specifically (or preferentially) reduces a subset of crossovers that do not normally exhibit interference. The existence of two crossover pathways, one subject to interference and the other free of interference, is consistent with recent observations (de los Santos et al. 2003).
Meiotic crossovers are nonrandomly distributed among chromosomes, such that every chromosome pair almost always undergoes at least one crossover (an obligate crossover) to ensure its correct segregation at meiosis I. Crossover interference and obligate crossing over are often assumed to be mechanistically related. Consistent with this hypothesis, mutations that decrease crossover interference also reduce spore viability due to the missegregation of nonrecombinant chromosomes (Sym and Roeder 1994; Chua and Roeder 1997; Novak et al. 2001). The wild-type level of spore viability observed in pol3-ct strains suggests that crossovers are properly distributed among chromosomes.
The pol3-ct mutation links replication and recombination:
Further studies using the pol3-ct mutant will lead to a better understanding of the mechanism by which Polδ influences gene conversion tract length and crossing over. This study casts new light on our growing awareness of the links between DNA replication and homologous recombination. It is now generally acknowledged that homologous recombination serves as a backup system supporting DNA replication when the template DNA is damaged or the replication machinery malfunctions (Kuzminov 2001). It is fascinating to unveil, on the other hand, that the outcome of homologous recombination depends largely on a major DNA replication protein.
We thank Roland Chanet and Tom Petes for kindly providing plasmids and yeast strains. L.M. especially thanks Eun-Jin Erica Hong for valuable comments on the manuscript and Michael Odell, Roberta Shew, and Nadia Shapiro for excellent technical assistance. This work was supported by the Howard Hughes Medical Institute, le Commissariat à l'Energie Atomique, and Electricité de France. L.M. was supported by postdoctoral fellowships from the Association pour la Recherche sur le Cancer and from the European Molecular Biology Organization.
- Received January 6, 2004.
- Accepted April 15, 2004.
- Genetics Society of America