Direct Coupling Between Meiotic DNA Replication and Recombination Initiation

See allHide authors and affiliations

Science  27 Oct 2000:
Vol. 290, Issue 5492, pp. 806-809
DOI: 10.1126/science.290.5492.806


During meiosis in Saccharomyces cerevisiae, DNA replication occurs 1.5 to 2 hours before recombination initiates by DNA double-strand break formation. We show that replication and recombination initiation are directly linked. Blocking meiotic replication prevented double-strand break formation in a replication-checkpoint–independent manner, and delaying replication of a chromosome segment specifically delayed break formation in that segment. Consequently, the time between replication and break formation was held constant in all regions. We suggest that double-strand break formation occurs as part of a process initiated by DNA replication, which thus determines when meiotic recombination initiates on a regional rather than a cell-wide basis.

During meiosis in most organisms, division of the diploid genome among haploid gametes is accompanied by frequent recombination between homologous parental chromosomes. Recombination occurs after meiotic DNA replication but before the first meiotic division. In the yeast Saccharomyces cerevisiae, blocking meiotic replication has been shown to prevent recombination (1–4), but the connection between these two fundamental processes remains unknown. A replication dependence of meiotic recombination would be expected if replication-inhibited cells could not form the double-strand breaks (DSBs) that initiate recombination. This could result either from a direct coupling between replication and DSBs or from checkpoint systems that sense incomplete replication and prevent DSB formation.

To test these suggestions, we examined DSBs in cells undergoing meiosis in the presence of 100 mM hydroxyurea (HU), a concentration that prevents DNA replication (4). In such conditions, meiotic progression is normally blocked before the first nuclear division (meiosis I) by the MEC1-dependent checkpoint system; progression is restored in mec1-1 mutants (4). Wild-type cells sporulated in HU did not form DSBs, and meiotic progression was blocked (5). By contrast, about 40% of mec1-1 mutant cells progressed through the meiosis I in the presence of HU, but DSBs still did not form (Fig. 1). Thus, the failure to form DSBs without replication is not due to a MEC1-dependent checkpoint block to meiotic progression, making it likely that replication is directly required for DSB formation. This conclusion is reinforced by the finding thatclb5 clb6 double mutants, which prevent meiotic replication without inducing the MEC1 block (4), also fail to form DSBs (6).

Figure 1

DNA replication is required for DSB formation. A checkpoint-deficient mec1-1/mec1-1 diploid was sporulated in the absence (−HU) or presence of 100 mM hydroxyurea (+HU), which blocks meiotic DNA replication (4, 26). (A) Meiotic progression, monitored by fluorescence microscopy of 4′,6′-diamidino-2-phenylindole (DAPI)–stained cells. Cells with more than one nucleus were scored as having completed the first meiotic division. (B) DSB formation. Blots contain digested DNA extracted at the indicated time during meiosis (26). The rightmost lane contains meiotic DNA from an untreated wild-type control (MJL1071).

To further examine the relation between replication and DSBs, we used two methods to delay replication in the left arm of chromosomeIII (chrIII-L). One approach used an ars305 ars306 ars307 triple mutation to inactivate all meiotic replication origins on chrIII-L (7) (Fig. 2A). The ars305 ars306 ars307chrIII-L is replicated passively by forks initiating atARS309 on the right arm (chrIII-R) or further to the right, at least 126 kb from the left-hand telomere. On the basis of a fork progression rate of 2 kb/min (8), replication of thears305 ars306 ars307 chrIII-L should take at least 40 min longer than when these origins are present and active. The other approach used a reciprocal translocation (his4::TEL1-L, Fig. 2A), replacing the first 70 kb of chromosome III with the first 4 kb of chromosomeI (TEL1-L). This places TEL1-L next to the HIS4-CEN3 region, which undergoes frequent DSBs (9) and contains the early-firing origin ARS306(10). Yeast telomeres are late replicating and impose this property on nearby sequences (11, 12), so this translocation should delay replication in the HIS4-CEN3region.

Figure 2

DSB formation is delayed in late-replicating regions. (A) Map of chromosome III. Filled and open squares, functional and inactive origins, respectively, in ars305 ars306 ars307 mutants; zig-zag line, breakpoint of thehis4::TEL1-L translocation (27). (B) Meiotic replication intermediates in an ars305 ars306 ars307 strain. 2D gel blots with DNA extracted at the indicated times were probed with left arm (GLK1) and right arm (YCR54c) probes (27). “Y” arc, replication intermediates; “X” spike, recombination intermediates (28); vertical arrows, time of maximum intermediate levels. (C) DSBs on whole chromosomes. Pulsed-field gel blots with DNA extracted at the indicated times were probed with left-end probes (27). Vertical arrows, time of maximum DSB levels in each arm. (D) Replication intermediate and DSB quantitation. Dotted lines, replication intermediates at GLK1 (▴), atARS306 (♦), or at YCR54c (▵). Solid lines, DSBs on the left arm (▪) and on the right arm (□), measured on pulsed-field gels. The time that a strain initiates meiosis after transfer to sporulation medium varies between cultures. In the experiments presented here, measurements of the time of nuclear divisions indicate that the his4::TEL1-L strain initiated meiosis 1 half-hour later than did the other two strains (5).

We monitored meiotic replication on chromosome III by two-dimensional (2D) gel electrophoresis of replication intermediates (13). Replication occurred simultaneously at three locations on the normal chromosome III (Fig. 2D). By contrast, inars305 ars306 ars307 strains, replication on chrIII-L was delayed by 60 min relative to chrIII-R (Fig. 2, B and D), as expected if chrIII-L was replicated passively by forks initiating on chrIII-R. On his4::TEL1-L, replication in the HIS4-CEN3 region was delayed by 30 min relative to chrIII-R (Fig. 2D), and most ARS306 origin activity was eliminated (5), as expected for a telomere position effect on replication.

Delaying replication did not markedly affect DSB frequencies on chrIII-L, as measured from blots of pulsed-field gels (Fig. 2C). Maximum DSB levels on the ars305 ars306 ars307chrIII-L were identical to those on the normal chromosome (20% ± 6% compared with 20% ± 5%), as were break levels at individual sites (5). DSBs between HIS4 andCEN3 on his4::TEL1-L were modestly reduced (6 ± 2% compared with 12 ± 2% on a normal chromosome), the reduction being stronger near TEL1 than near CEN3. However, delaying replication locally had a distinct, reproducible effect on DSB timing (Fig. 2, C and D). DSBs formed simultaneously (1.5 to 2 hours after replication) in both arms of the normal chromosome III. By contrast, overall DSB formation in the late-replicating ars305 ars306 ars307chrIII-L was delayed by 30 min compared with chrIII-R (Fig. 2, C and D). Thus, the time interval between replication and DSB formation was maintained in both arms. Furthermore, the DSB delay in the originless chrIII-L varied with distance from active right-arm origins (Fig. 3), being greater at a centromere-distal site (YCL49c, 1-hour delay) than at a centromere-proximal site (YCL10c, 30-min delay). In thehis4::TEL1–L chromosome, DSBs formed 30 min later in the late-replicating his4-CEN3 interval than in the earlier replicating chrIII-R, again maintaining a 2-hour gap between replication and DSBs (Fig. 2D). By contrast, the time of DSB repair was independent of the time of break formation. Early- and late-forming DSBs disappeared at the same time (Figs. 2C and 3B), and X-shaped recombination intermediates formed at the same time in early- and late-replicating sequences in ars305 ars306 ars307mutants (Fig. 2B) and in all other diploids examined (5).

Figure 3

The DSB delay increases with distance from active origins. DSBs were measured at individual sites with conventional agarose gels (26). (A) DSBs at individual sites. Blots containing restriction-digested DNA from wild-type and ars305 ars306 ars307 cells at 0, 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, 6, and 7 hours in meiosis. Solid circles, DSBs near the left end of chrIII-L; gray circles, DSBs on chrIII-L near the centromere; open circles, DSBs on the right arm. DSB bands used for measurements are indicated. Vertical arrows, time of maximum DSB frequencies. (B) DSB quantification. Symbols are as in (A).

These data indicate that, irrespective of the time of replication, a constant period of time is maintained between replication and DSB formation in a region. DSB repair, by contrast, occurs at a specific time during meiosis, suggesting that repair is controlled on a cell-wide basis. Further evidence that DSB timing does not affect repair was provided by measuring recombination between leu2mutant alleles (leu2-K and leu2-R) located between HIS4 and CEN3 (see Fig. 2A). Leu+ meiotic recombinants were recovered fromhis4::TEL1–L, ars305 ars306 ars307, and control strains in 0.42, 0.45, and 0.40% of spores, respectively.

DSB frequencies are often measured with the use of mutants (rad50S, mre11S, sae2/com1) in which the protein that creates DSBs, Spo11p, remains covalently attached to break 5′ ends (14, 15). Failure to remove Spo11p results in the accumulation of unrepaired breaks, which are more easily detected and quantified. We observed, in sae2Δ orrad50S mutants, a marked effect of replication timing on DSB frequencies (Fig. 4). In ars305 ars306 ars307 sae2Δ strains, total DSBs on the late-replicating chrIII-L were reduced fourfold relative to asae2Δ strain with all origins intact (Fig. 4A). Breaks in the late-replicating HIS4-CEN3 region ofhis4::TEL1-L sae2Δ strains were reduced almost 10-fold (Fig. 4A); repression at individual sites increased with proximity to the new telomere (Fig. 4B). DSBs near the right-hand telomere of chromosome III were also differentially reduced in sae2Δ mutants (5). We speculate that thissae2Δ/rad50S-dependent reduction in late DSBs could be due to checkpoint systems that sense unprocessed DSBs and prevent later break formation. Alternatively, it could result from a failure to recycle a DSB-forming complex, thought to remain at DSB sites in the absence of processing (16). In either case, break measurements obtained with rad50S or sae2 mutants should be interpreted cautiously, because these mutants specifically reduce DSBs in late-replicating regions.

Figure 4

Breaks are reduced in late-replicating regions in the absence of DSB processing. (A) DSBs on whole chromosomes in sae2Δ strains (26). Pulsed-field gel blots with DNA prepared at 0 or 6 hours in meiosis, hybridized with a chromosome III right-end probe. Total DSB frequencies in each chromosome arm are given. (B)sae2Δ-specific DSB reduction at individual sites. Frequencies at individual sites were measured with blots of conventional agarose gels (27). Black bars, DSBs in theHIS4-CEN3 region; open bar, DSBs on the right arm. Loci are arranged in terms of increasing distance from the left-hand telomere. *, DSBs at his4 in his4::TEL1-L sae2Δstrains were below the limit of detection (0.05% of chromosomes); this ratio was calculated assuming 0.05% as an upper value.

In summary, our data confirm that meiotic replication is required for DSB formation. DSB formation is most likely directly coupled to replication because a fixed time period is maintained between these two events. DSB processing-defective mutants allow a further distinction between break formation in early- and late-replicating regions. Collectively, these findings identify meiotic DNA replication as the primary event governing DSB formation. Evidence for a replication-DSB interplay is also provided by recent findings that meiotic S phase is shortened in spo11 deletion mutants (17). We suggest that DNA replication of a region initiates a series of events that culminate, 1.5 to 2 hours later, in DSB formation. This is in contrast to what is seen in bacteria or in vegetatively growing yeast, where recombination-initiating lesions are formed by fork pausing and breakage during replication itself (18).

DNA replication has been suggested as playing an important role in several chromosome structural changes that occur during mitotic growth of S. cerevisiae, including the establishment of sister chromatid cohesion (19), the restoration of transcriptional silencing to mating-type gene cassettes (20), and the recruitment of chromatin-remodeling factors (21). Our studies identify meiotic recombination as another chromosomal transformation coupled to DNA replication and thus provide evidence for a mechanistic link between these two fundamental processes. In meiosis, the 1.5- to 2-hour period observed between replication and DSB formation might reflect the time needed to assemble protein complexes that participate in DSB formation (17, 22) or to establish interhomolog contacts needed for efficient DSB formation (17, 23, 24). Our findings indicate that such processes are coupled to replication on a regional basis, rather than by a cell-wide progression signal. The ability to create and distinguish between early- and late-replicating regions should allow the determination if other landmark events of meiosis, such as the assembly of early homolog pairing structures or the formation of synaptonemal complex, are also directly coupled to meiotic DNA replication and thus are affected by its timing.

  • * To whom correspondence should be addressed. E-mail: lichten{at}


Stay Connected to Science

Navigate This Article