Trans Events Associated With Crossovers Are Revealed in the Absence of Mismatch Repair Genes in Saccharomyces cerevisiae

Abstract

Genetic analysis of recombination in Saccharomyces cerevisiae has revealed products with structures not predicted by the double-strand break repair model of meiotic recombination. A particular type of recombinant containing trans heteroduplex DNA has been observed at two loci. Trans events were originally identified only in tetrads in which the non-Mendelian segregations were not associated with a crossover. Because of this, these events were proposed to have arisen from the unwinding of double Holliday junctions. Previous studies used palindromes, refractory to mismatch repair, as genetic markers whereas we have used a complementary approach of deleting mismatch repair proteins to identify heteroduplex DNA. We found that the markers occurred in trans and were associated with crossovers. In both mlh1Δ and msh2Δ strains, the frequency of trans events associated with a crossover exceeded that predicted from the random association of crossovers with noncrossover trans events. We propose two different models to account for trans events associated with crossovers and discuss the relevance to wild-type DSB repair.

MEIOTIC recombination is initiated by the formation of DNA double-strand breaks (DSBs; reviewed in Keeney 2001). The results of physical and genetic tests of the original model (Szostak et al. 1983) proposing such breaks and their repair have confirmed some aspects of the model and required modification of others (Figure 1). For example, DSBs have been found over the entire genome and correlate with hotspots of gene conversion and crossing over (Gerton et al. 2000). Detailed characterization of DSBs at the ARG4 recombination hotspot (Sun et al. 1989) have shown DSBs to be resected to generate 3′-single-stranded overhangs, which are, on average, 440 bp in length (Sun et al. 1991). Physical experiments have also demonstrated that one of the two 3′-single-stranded overhangs invades the homolog (Hunter and Kleckner 2001), the second strand is captured, and a double Holliday junction is formed (Collins and Newlon 1994; Schwacha and Kleckner 1994, 1995). The original DSB repair model of Szostak et al. (1983) proposed that crossovers and noncrossovers came from the random cleavage of the two Holliday junctions (Figure 1E). However, Gilbertson and Stahl (1996) provided genetic data indicating that noncrossovers do not arise by the random resolution of double Holliday junctions. The authors found that noncrossover tetrads with half conversions (also known as postmeiotic segregations; see Figure 1) of both markers failed to conform to the standard double-strand break repair (DSBR) model. In particular, rather than having the predicted two sectored spore colonies (resulting from heteroduplex formation on both broken and unbroken strands; Figure 1, H and I), frequently only one spore colony contained the half conversion for both genetic markers (Figure 1J), indicating that only the broken strand contains heteroduplex DNA. When two genetic markers were found to segregate on a single chromatid, the cosegregation of the two markers was divided into two types: those in which the genetic markers occurred in recombinant fashion (trans, Porter et al. 1993, Figure 1J, markers 2 and 3) and those in which the genetic markers remained in the parental configuration (cis, Figure 1J, markers 3 and 4). Cis events occur when both markers are on the same side of the DSB (Porter et al. 1993). The trans events, however, can be explained only by the formation of a DSB between the genetic markers. Because trans events were preferentially associated with the noncrossover configuration of the flanking markers, Gilbertson and Stahl (1996) proposed that trans events came from the topoisomerase-mediated resolution of the double Holliday junction. More recently, it has been suggested that trans events represent unligated Holliday junctions that are unwound with the strands reannealed to their original partner (H. M. Foss and F. Stahl, personal communication). The trans events were rarely associated with a crossover at ARG4. Indeed, the frequency of such events was explicable by the occurrence of a trans event associated with an incidental crossover. Subsequently, physical experiments have demonstrated that noncrossovers and crossovers derive from separate pathways, with noncrossovers suggested as arising from synthesis-dependent single-strand annealing (SDSA; Allers and Lichten 2001; Borner et al. 2004). Some models of SDSA can also generate trans heteroduplex DNA without crossovers (Paques and Haber 1999). A recent study, however, suggested that trans events could also be associated with crossovers at a frequency greater than that predicted by coincidence alone (Merker et al. 2003).

Figure 1.—

Modified double-strand break repair model by Szostak et al. (1983). The five markers used in this study to span the DSB are shown as solid circles and correspond to HPH, BIK1-939, his4-ATC, HIS4-1605, and NAT (Figure 2). The two distal circles indicate flanking markers used for monitoring crossing over. Only the two interacting chromatids are shown for clarity. (A) DSB formation initiates meiotic recombination. (B) Both ends of the DSB are processed to produce 3′-single-stranded overhangs. (C) Invasion of the intact homologous chromosome by the 3′-single-stranded overhangs occurs asynchronously such that initially only one of the 3′-single-stranded overhangs invades. This is the first stage during which heteroduplex DNA is formed. This intermediate may also lead to synthesis-dependent single-strand annealing. (D) DNA synthesis displaces one of the resident strands to form a D-loop. Capture of the D-loop by the second end primes DNA synthesis and ligation completes the joint molecule (double Holliday junction). (E) This intermediate is thought to be resolved only as crossovers as shown in F and G. The genetic configuration of the resulting tetrad is shown to the right. Arrows indicate which chromatid gives rise to what spore. The two parental DNA strands not shown correspond to the spores containing all solid (top) or all open (bottom) information. Unrepaired heteroduplex DNA is indicated by sectored spore colonies. The noncrossover resolutions in H and I are rare. (J) Tetrads displaying half conversions of markers placed on opposite sides of the DSB on a single chromatid are common. Co-half conversions for markers 2 and 3 on the opposites of the DSB are in the trans configuration while markers 3 and 4 on the same side of the DSB are in cis.

To elucidate the nature of trans events, we used mlh1Δ and msh2Δ strains in which the repair of mismatches in heteroduplex DNA is compromised. MSH2 and MLH1 are orthologs of the bacterial MutL and MutS that repair replication errors (reviewed in Schofield and Hsieh 2003). MSH2 and MLH1 have central roles in heteroduplex repair in both vegetative and meiotic cells (reviewed in Surtees et al. 2004). In addition, MLH1 is also required for normal levels of crossovers (Hunter and Borts 1997). Hence, deleting MSH2 or MLH1 allows the detection of heteroduplex DNA as many of the meiotic products contain spore colonies that are sectored for the genetic markers. We found that trans events associated with crossovers were frequent in both mutant strains. We propose two different models to account for trans events associated with crossing over and discuss the relevance to wild-type DSB repair.

MATERIALS AND METHODS

Strains:

All of the strains have been described previously (Hoffmann et al. 2003, 2005). ERY103 is the diploid wild-type strain generated from mating EY97 to EY128. ERY112 and ERY102 are isogenic derivatives of ERY103 where EY97 and EY128 have been deleted for MLH1 or MSH2, respectively. In Figures 1 and 3, all relevant genetic markers of EY97 are presented as solid circles (i.e., HPH, his4-ATC, BIK1, and NAT) and the alleles of EY128 are presented as open circles (i.e., hph⁻, BIK1-939, HIS4-1605, and nat⁻).

Genetic analysis:

Only tetrads in which BIK1-939 and his4-ATC co-converted were used in this study (Hoffmann et al. 2005, accompanying article in this issue). Tetrads in which the conversions of BIK1-939 and his4-ATC were caused by the independent repair of two or more DSBs were not included, as discussed previously (Hoffmann et al. 2005).

Statistical analysis:

Distributions of events were compared using the G-test of homogeneity (e.g., Sokal and Rohlf 1995). Pairwise comparisons of proportions were carried out using the χ² contingency test. Comparison of expected and observed proportions was analyzed using the χ² goodness-of-fit test (available from http://faculty.vassar.edu/lowry/VassarStats.html). For all statistical comparisons, P < 0.05 was considered significant, except when multiple data sets were analyzed using pairwise comparisons. In such cases, the Dunn-Sidak adjustment of a = 0.05 was used (Sokal and Rohlf 1995) to avoid a type I error, as applied previously (Hoffmann et al. 2003).

RESULTS

Rationale:

To identify trans events (two genetic markers segregating on a single chromatid in a recombinant fashion; Figure 1J), the genetic markers must flank the DSB and remain unrepaired (Porter et al. 1993; Gilbertson and Stahl 1996). To this end, we modified the interval around the HIS4 recombination hotspot to contain single nucleotide changes (Figure 2). The HIS4 DSB has previously been demonstrated to be the major DSB within an ∼10-kb region (Hoffmann et al. 2005). Thus, repair of other DSBs should not interfere with the analysis. To follow heteroduplex DNA, we deleted either MLH1 or MSH2, rendering the mispairs in heteroduplex DNA partially refractory to repair.

Figure 2.—

The HIS4 region on chromosome III (Crick strand), shown in opposite from the conventional orientation. (A) The two haploid parents EY97 (top chromatid) and EY128 (bottom chromatid). The direction of transcription is indicated by the tapered end. The boxes labeled NAT and HPH represent insertions introduced in EY97 but not in EY128. EY97 is auxotrophic for histidine synthesis as the start codon of HIS4 has been changed from ATG to ATC (his4-ATC allele). HIS4 and BIK1 of parent EY128 contain silent single nucleotide changes at HIS4-1605 and BIK1-939. Arrows indicate the position and relative intensities of double-strand breaks in the region (Hoffmann et al. 2005).

Identification of trans events:

We analyzed a total of 1130 tetrads containing four viable spores for non-Mendelian segregation (NMS) of the his4-ATC allele in the mlh1Δ and msh2Δ strains. Of these, 116 in mlh1Δ and 96 in msh2Δ contained a NMS at his4-ATC (Table 1). All of these were analyzed for segregation at BIK1-939 and HIS4-1605. We eliminated all tetrads in which more than two chromatids were recombinant for any of the five genetic markers, as these were likely to have been caused by two independent DSB repair events (co-events). We also excluded tetrads in which BIK1-939 and his4-ATC showed NMS but segregated in opposite directions (e.g., 3:5 and 5:3). Such events cannot arise from repair of a single DSB (Hoffmann et al. 2005). Finally, we also excluded all events in which NAT, HYG, or HIS4-1605 showed co-events. Altogether, 39/116 and 38/96 events for mlh1Δ and msh2Δ, respectively, were discarded. These frequencies were similar to previously observed rates (Hillers and Stahl 1999; Merker et al. 2003). Of the tetrads that were considered to be attributable to a single DSB repair event, ∼67% and 70% were one sided in mlh1Δ and msh2Δ, respectively.

TABLE 1.

Classification of DSB repair events atHIS4

						Double Holliday junctionf		Transg
	Totala	Excluded eventsb	One sidedc	Two sidedd	co-HCe	CO	NCO	CO	NCO
Wild type	144	58	29	57	—	—	—	—	—
mlh1Δ	116	39	52	25	23	4	0	10	9
msh2Δ	96	38	41	17	10	2	1	2	5

^aThe number of HIS4 NMS that were analyzed.

^bThe number of events in which more than two chromatids were recombinant or where his4-ATC and BIK1-939 segregated in a fashion not predicted by a single DSB repair event. This category also included potential two-sided events where NAT, HPH, or HIS4-1605 showed a co-event.

^cNMS of his4-ATC but not BIK1-939.

^dTwo-sided events where his4-ATC and BIK1-939 co-converted in a fashion predicted by various models for a single DSB repair event.

^eCo-half conversions. Tetrads in which both markers showed half conversions and could thus be used to identify trans events.

^fCo-half conversion events in which the half conversions of BIK1-939 and his4-ATC were on two separate chromatids, consistent with resolution of double Holliday junctions as proposed in the model of Szostak et al. (1983). CO and NCO denote the number of tetrads with an associated crossover or noncrossover, respectively.

^gCo-half conversion events in which the half conversions of BIK1-939 and his4-ATC occurred on a single chromatids.

Of the remaining two-sided events, only those that contained a half conversion at both his4-ATC and BIK1-939 (Table 1 and Figure 3) could distinguish trans events from conventional co-conversions. The mlh1Δ strain contained 23 such events and the msh2Δ strain contained 10. Of these, seven tetrads (four from mlh1Δ and three from msh2Δ) contained the co-half conversions on two separate chromatids, consistent with the model by Szostak et al. (1983)(Figure 1, F–I). Only one of these had the flanking markers in the noncrossover configuration (Figure 3, A5). Hence, the lack of evidence for noncrossovers derived from double Holliday junction resolution (Figure 1, H and I) concurred with previous observations (Gilbertson and Stahl 1996). The remaining six tetrads in which the half conversions of BIK1-939 and his4-ATC occurred on two separate chromatids had their flanking markers in the crossover configuration, consistent with their having arisen via resolution of a double Holliday junction (Figure 3, A1–4). A total of 26 (19 from mlh1Δ and 7 from msh2Δ) of the 33 tetrads had co-half conversions on a single chromatid (Figure 3); all of these were in trans. Although it is possible that a DSB other than that at HIS4 contributes to non-Mendelian segregation of BIK1-939, this is unlikely. The nearest DSB was >2 kb from BIK1-939 and was much weaker than that of the HIS4 DSB (Figure 2; Hoffmann et al. 2005). In addition, the non-Mendelian segregation frequencies of HPH were much lower than those of his4-ATC (compare 1% to 20% NMS and 1% to 18% NMS in the mlh1Δ and msh2Δ strains, respectively). We have previously shown that events involving both HPH and BIK-939 are exceedingly rare (Hoffmann et al. 2005).

Figure 3.—

Classification of co-half conversion events. Only tetrads in which one or two chromatids were recombinant were included (Hoffmann et al. 2005). (A) Tetrads containing the half conversions at BIK1-939 and his4-ATC on two separate chromatids consistent with the DSB repair model of Szostak et al. (1983). (B) Tetrads that contained both half conversions on the same chromatid in trans. The number of each class of tetrads observed in the two mutant strains is given at the right. Genetic markers are, as in Figure 2, from left to right: HPH (H), BIK1-939 (9), his4-ATC (A), HIS4-1605 (1), and NAT (N).

It is also possible that the trans events arose from two independent DSB repair events at HIS4. Therefore, we calculated the expected number of trans events, given the individual frequencies of the following (as calculated in Hoffmann et al. 2005): (1) that two independent events occurred (0.04 and 0.03, for the mlh1Δ and msh2Δ strains, respectively); (2) that both DSB repair events were one sided (0.33 and 0.42, respectively); (3) that only two chromatids were recombinant (0.29 and 0.06, respectively); (4) that both genetic markers showed half conversions (0.70 and 0.57, for the mlh1Δ and msh2Δ strains, respectively); and (5) that both half conversions were in trans rather than in cis (0.5). Thus, the total numbers of expected trans tetrads were 0.76 (585 × 0.0013) and 0.12 (545 × 0.0002) for the mlh1Δ and msh2Δ strains, respectively. The observed numbers of trans events were 25- and 58-fold higher. We conclude that it is unlikely that the trans events originated from two independent DSB repair events.

Trans events associated with crossovers represent a significant proportion of events in mismatch-repair-defective strains:

Both mismatch-repair-defective strains contained a significant number of trans co-conversions associated with crossovers. In the mlh1Δ strain, 10 of 19 trans tetrads (Table 1) were associated with a crossover. In the msh2Δ strain, 2 of the 7 tetrads were associated with crossovers. The proportions correspond to those observed by Merker et al. (2003), who found 10 trans events of which 4 were associated with a crossover. Using the above calculations and the frequency with which his4-ATC NMS events are associated with crossovers in the general population (0.40 and 0.55, respectively, for the mlh1Δ and msh2Δ strains; Hoffmann et al. 2005), one would expect 0.3 and 0.07 tetrads, respectively. Thus, mlh1Δ and msh2Δ strains were increased 33- and 29-fold, respectively, for trans events associated with a crossover.

Repair of a single DSB can lead to gene conversions accompanied by a crossover. However, crossovers can also be “incidental” to the gene conversion event when the crossover and the gene conversion are the result of two independent DSB repair events. To demonstrate that the crossovers associated with the trans events were not incidental, the frequency of crossovers associated with trans events should be greater than the frequency of incidental crossovers in the total population. To determine if this was the case, the number of incidental crossovers associated with a half conversion his4-ATC was evaluated. Incidental crossovers were deemed to be those that did not occur immediately adjacent to the half conversion (Figure 3; in tetrad B8 the crossover did not map between his4-ATC and HIS4-1605 as expected; Merker et al. 2003). This number was divided by the total number of half conversions of his4-ATC. Such calculations yield the frequencies of incidental crossovers as 8.7% and 9.6% for the mlh1Δ and msh2Δ strains, respectively (Table 2). Thus, the frequency of trans events associated with a crossover was at least 6- and 3-fold higher in the mlh1Δ and msh2Δ strains, respectively, than the rate of incidental crossovers in the general population (P < 0.05, goodness-of-fit test for combined data). These values are similar to those observed previously using palindromes as genetic markers in an otherwise wild-type background (6.6-fold increase; Merker et al. 2003).

TABLE 2.

Incidental crossovers

	CO-Ia	Crossoverb	Totalc	Incidentald (%)
Wild type	5	45	77	6.5
mlh1Δ	6	25	69	8.7
msh2Δ	5	24	52	9.6

^aIncidental crossovers (see text and Figure 3).

^bAll tetrads in which crossovers occurred (see text).

^cAll tetrads in which his4-ATC showed a half conversion.

^dIncidental crossover frequencies (given as percentages) were calculated as CO-I divided by the total.

One of the trans events in the mlh1Δ strain contained an incidental crossover (Figure 3, B8) giving rise to a frequency of incidental crossovers among the trans events of 3.8% (1/26). This frequency was similar to that observed for the entire population (see above). This observation lent further support to the notion that the remaining crossovers were associated with the trans events.

DISCUSSION

Trans events associated with crossovers contribute a significant proportion of two-sided events:

The number of trans co-half conversion tetrads associated with a crossover was much larger than that observed by Gilbertson and Stahl (1996) at the ARG4 DSB hotspot. However, Merker et al. (2003) also observed such events in wild-type strains at similar frequencies, suggesting that trans events at HIS4 occur in both wild-type and mismatch-repair-defective strains. In the study reported here, only 17% (mlh1Δ) and 20% (msh2Δ) of the two-sided events associated with a crossover were those predicted by Szostak et al. (1983). The remainder were trans events associated with either a crossover (43% and 20% for the mlh1Δ and msh2Δ strains, respectively) or a noncrossover (39% and 50%, respectively). Finally, only the msh2Δ strain contained a two-sided event in which the two half conversions segregated on two different chromatids, but with the flanking markers in the noncrossover configuration (10%; 1/10; Table 1). Hence, the class of events predicted by the canonical DSBR model for the resolution of double Holliday junctions in the noncrossover mode is rare. Of all of the single DSBR events at HIS4 associated with a crossover, the proportions of trans crossovers were 13% (10/77) and 3% (2/58), respectively, in the mlh1Δ and msh2Δ strains. These proportions are underestimates due to the fact that mlh1Δ and msh2Δ strains have fewer detectable two-sided events compared to the wild-type strain (Hoffmann et al. 2005).

Origin of trans crossovers:

No mechanism has been suggested for the origin of trans events associated with crossovers. There are at least two plausible explanations for this type of event (Figure 4). Within the context of the model proposed by Szostak et al. (1983), migration of the Holliday junction established during the single-end invasion could lead to trans events associated with crossing over. However, migration of the Holliday junction must be constrained to being toward the DSB (Figure 4C) as migration in the opposite direction leads to symmetric heteroduplex DNA, for which we found no evidence (data not shown). The Holliday junction established after capture of the second end must not be mobile for similar reasons. Alternatively, establishing a replication fork with lagging-strand synthesis on the D-loop can explain the trans events associated with a crossover. In this model, the replication fork movement is coupled to D-loop migration (Figure 4, F and G). There is currently no experimental evidence to distinguish these scenarios.

Figure 4.—

Models of trans events associated with crossovers. A model involving restricted Holliday junction migration leading to trans heteroduplex DNA is illustrated in A–E. Single-end invasion (A) followed by DSB repair replication and second end capture (B) leads to the formation of the joint molecule (C). Migration of the Holliday junction established during the single-end invasion (right-hand side) leads to an intermediate in which heteroduplex is formed in trans (D). Resolution results in a tetrad in which a trans event was associated with a crossover (E). A model in which a replication fork is established is shown in F and G. Lagging-strand replication using the D-loop as template is coordinated with leading-strand synthesis (F). Further replication accompanied by D-loop migration (G) and capture of the second end establish a double Holliday junction similar to that in D.

Are trans events representative of wild-type DSB repair events?

Either the mutant strains allow us to observe the wild-type situation, by revealing heteroduplex DNA but without otherwise interfering with the mechanism of DSB repair, or Msh2p/Mlh1p have specific roles during meiotic DSB repair that preclude trans events in wild-type cells. Consistent with the proposal that absence of Msh2p/Mlh1p simply reveals the normal situation, trans events have been observed previously in wild-type strains marked with palindromes at frequencies similar to those reported here. However, since palindromes are refractory to repair by Msh2p/Mlh1p (Alani 1996), both experiments reflect what happens in the absence of mismatch repair.

Both Mlh1p and Msh2p may normally prevent trans events from occurring by preventing replication forks from being established and/or Holliday junctions from migrating. The latter is consistent with a role for Mlh1p in conjunction with Msh4/5p in stabilizing Holliday junctions (Hoffmann et al. 2005) as well as with the finding that Msh2p can bind Holliday junctions in vitro (Alani et al. 1997). We cannot exclude the possibility that mlh1Δ has a more severe phenotype than msh2Δ due to the low numbers of events in the msh2Δ strain. Since the structure of heteroduplex DNA within the double Holliday junction has not been characterized, it remains to be determined whether the pathways generating trans events are the result of Msh2p/Mlh1p-defective DSB repair or are a “normal” pathway for DSB repair in wild-type cells.