HO endonuclease-induced recombination in yeast meiosis resembles Spo11-induced events

Abstract

In meiosis, gene conversions are accompanied by higher levels of crossing over than in mitotic cells. To determine whether the special properties of meiotic recombination can be attributed to the way in which Spo11p creates double-strand breaks (DSBs) at special hot spots in Saccharomyces cerevisiae, we expressed the site-specific HO endonuclease in meiotic cells. We could therefore compare HO-induced recombination in a well-defined region both in mitosis and meiosis, as well as compare Spo11p- and HO-induced meiotic events. HO-induced gene conversions in meiosis were accompanied by crossovers at the same high level (52%) as Spo11p-induced events. Moreover, HO-induced crossovers were reduced 3-fold by a msh4Δ mutation that similarly affects Spo11p-promoted events. In a spo11Δ diploid, where the only DSB is made by HO, crossing over was significantly higher (27%) than in mitotic cells (≤7%). This single meiotic DSB failed to induce the formation of a synaptonemal complex. We also show that HO-induced gene conversion tract lengths are shorter in meiotic than in mitotic cells. We conclude that a hallmark of meiotic recombination, the production of crossovers, is independent of the nature of Spo11p-generated DSBs at special hotspots, but some functions of Spo11p are required in trans to achieve maximum crossing over.

Meiotic recombination in Saccharomyces cerevisiae differs from mitotic recombination in several respects. Beyond the fact that meiotic gene conversions occur 100 to 10,000 times more often than equivalent spontaneous mitotic events, the individual events themselves have different outcomes. Most notably, meiotic gene conversions have a higher association with crossing over than is seen in mitotic cells (reviewed in refs. 1 and 2). In addition to generating genetic diversity, crossing over is critical in ensuring accurate chromosome segregation during meiosis (3, 4).

The origins of most spontaneous gene conversion events in mitosis are unknown, but they are certainly different from the way in which double strand breaks (DSBs) are induced by the meiosis-specific Spo11 protein (5, 6). DSBs created by Spo11p are generated at many nearby sites in meiotic hotspots (7, 8). These hotspots represent a very special chromosomal context influenced by perhaps 10 other genes, the deletion of which abolishes or severely reduces DSB formation (reviewed in refs. 1, 3, 4, 9, and 10). Deletion of genes encoding Mre11p, Rad50p, and Xrs2p causes changes in the chromatin structure of hotspots (11). Similarly, the absence of protein components such as Hop1p that comprise the axial elements lying between sister chromatids also reduces DSB formation (12).

Other mutations reduce the proportion of meiotic gene conversions accompanied by meiotic crossovers without affecting DSB formation. The zip1Δ or zip2Δ mutations, which eliminate formation of the central element of the synaptonemal complex (SC), have this effect (13, 14), providing support for the general view that crossover regulation depends on this structure (3, 4). Deletion of the mismatch repair-related MSH4 and MSH5 genes also reduces the proportion of gene conversions accompanied by crossing over without reducing the total number of events (15, 16).

In the absence of Spo11-induced DSBs, meiotic crossing over is abolished. X-irradiation of such cells can produce an increase in exchange events, which is probably due to the induction of DSBs (17); but that study could not determine whether repair of such breaks is “meiotic-like,” that is, whether the frequency of crossing over associated with each repair event is higher than in mitotic cells.

The increase in crossover-associated gene conversion can be explained in several ways that are not mutually exclusive. For example, the specific DNA ends produced by Spo11p could predetermine the fate of these events. It might also depend on the creation of a special chromatin environment at hotspots by other proteins (e.g., ref. 11) that would subsequently influence the way recombination proceeds. Alternatively, it could depend on diffusible proteins that associate with DSB ends, for example, by a meiosis-specific DNA strand exchange protein such as Dmc1p (18), or that enhance stabilization of intermediate structures such as Holliday junctions that could be resolved by crossing over (reviewed in ref. 1).

To explore the differences between mitotic and meiotic recombination we have made use of the site-specific HO endonuclease to create identical DSBs in the two types of cells, at the same well-defined 2-kb region. HO-induced events in meiosis can also be compared directly with Spo11p-induced recombination in the same region. HO endonuclease creates 3′-ended 4-bp overhanging ends at a large, defined recognition site (19), whereas Spo11p appears to produce 5′-ended, 2-bp overhangs at many sites within hotspot regions (7, 8). Thus we can also evaluate how the DSB ends themselves influence recombination. We have previously shown that expression of HO from a meiosis-specific SPO13 promoter (SPO13∷HO) caused the appearance of DSBs at approximately the same time as those created by Spo11p and that the kinetics of appearance of recombinants by HO was similar to those of normal meiotic events (20). We report here that HO-induced recombination in meiotic cells has many of the characteristics of Spo11-induced events.

Materials and Methods

Strains used in this study are derivatives of those used previously (21). Diploid strain WYL237 has a genotype MATa/MATα ura3/ura3 trp1/trp1 met13-4/met13-4 lys2C/lys2C ade1/ade1 CAN1/can1 CYH2/cyh2 and contains a well-defined URA3-LEU2-ADE1 interval on chromosome III (Fig. 1A). A 2.2-kb ADE1 gene was inserted at the AseI site of LEU2 located 1076 bp centromere proximal to the Asp718 site in LEU2 (19), whereas the opposite site of LEU2 is bounded by a large URA3-marked deletion removing part of the HIS4 gene and all of the intervening 17 kb to the XhoI site distal to LEU2 (E. J. Louis and J.E.H., unpublished observation). The leu2-R allele contains a 4-bp fill-in of the EcoRI site of LEU2 (22). The leu2-K allele is a 4-bp deletion of the KpnI site, located 399 bp from EcoRI (22). A 1.7-kb TRP1 marker was inserted next to CEN3. WYL191 is isogenic to WYL237, except that the LEU2 gene in this strain is flanked by the large heterologies marked by URA3 and ADE1 (Fig. 1B). Diploid strain DAM454 has the genotype MATa-inc/MATα-inc ade1/ade1 ura3/ura3 lys2∷SPO13∷HO/lys2∷SPO13∷HO. The LEU2 gene is flanked by large heterologies (Fig. 1D). The URA3 marker was replaced by ura3-52, by an interchromosomal gene conversion event, selected using 5-fluoroorotic acid (23). One copy of LEU2 contains an HO endonuclease cut site (leu2-cs) at the Asp718 site (24), and another copy of LEU2 is wild type.

Figure 1.

(A) A well-defined interval on chromosome III in which a 1.8-kb LEU2 gene is flanked closely by URA3 and ADE1 genes. The positions of leu2-K and leu2-R are shown by o and x, respectively. (B) Recombination in strain WYL 191 is largely confined to the 1.8-kb interval containing the 2-kb LEU2 gene bounded between flanking heterologies, a 20-kb deletion (URA3) and a 2.2-kb insertion (ADE1). (C) A 117-bp HO cleavage site (leu2-cs, arrow) was inserted into the LEU2 gene at Asp718. Sites of Spo11p cleavage are located approximately 400 bp centromere-distal to leu2-cs. The opposite homologue carries leu2-R. All strains expressing HO contain mutations at MAT that prevent HO cleavage. (D) DAM454 and DAM457 carry leu2-cs and a wild-type LEU2 allele. (E) DAM530 has the same chromosome III structure as that shown in D, except that the HML locus on the leu2-cs-containing chromosome was disrupted by the insertion of URA.

Diploid DAM518 is a leu2-R isogenic derivative of DAM454 created by gene replacement (21) (Fig. 1C). DAM 518 also contains a spo13Δ mutation introduced as previously described (20). SPO11 was deleted by the method of Wach et al. (25) to produce DAM502. msh4Δ derivatives were created using a plasmid provided by N. Hollingsworth (State University of New York, Stony Brook, NY). DAM 516 was created by a crossing an isogenic msh4Δ derivative of DAM454 and a backcrossed MATα-inc msh4Δ strain. Strain DAM509 is identical to DAM518 except that it carries a GAL∷HO gene integrated at ade3 (26) and is SPO13/spo13Δ. DAM528 is identical to DAM518 except that it lacks the lys2∷SPO13∷HO construct, is SPO13/spo13Δ and its MATa-inc parent is isogenic to a sibling of that used to construct DAM518. DAM535 is identical to DAM528 except it is homozygous for spo13Δ. DAM457 is identical to DAM454 except that it lacks the lys2∷SPO13∷HO construct and its MATα-inc parent was a sibling of that used to construct DAM454. DAM530 is identical to DAM454, except that it is homozygous for spo13Δ and spo11Δ and it carries an insertion of URA3 at HML (27) on the chromosome carrying leu2-cs (Fig. 1E).

Galactose inductions, yeast spreads and immunocytological analysis, and genetic and DNA methods have been described previously (20, 24, 28). The statistical significance of various comparisons was evaluated by using the G-test (29), with a program provided by Ed Louis (University of Leicester, Leicester, U.K.).

Results

HO-Induced Recombination in Meiosis Resembles SPO11-Mediated Events.

We have examined recombination in a well-defined interval on the left arm of chromosome III, in which a 1.8-kb LEU2 gene is flanked closely by URA3 and ADE1 genes that can be used to determine how often crossing over accompanies gene conversion in the LEU2 locus (Fig. 1A). Spo11-mediated events were first analyzed by examining random spore segregants from a diploid carrying the leu2-K and leu2-R alleles, as previously described (21). Previous studies, using strains closely related to those used here, have shown that virtually all Leu2⁺ recombinants arise by gene conversion rather than by reciprocal exchange between the two (21). The strong bias in which crossover chromatid contains the Leu2⁺ gene convertant is consistent with previous analyses of heteroallelic recombination, supporting the conclusion that crossover-associated gene conversions usually involve heteroduplex DNA that covers only one of the two heteroallelic sites (30, 31). Among Leu2⁺ meiotic gene convertants in strain WYL237, 60% were accompanied by crossing over between the URA3 and ADE1 flanking markers (Table 1A).

Table 1.

The frequency of crossing over associated with recombination at the LEU2 locus

	Spores with no crossing over			Spores with crossing over
A. Analysis of random spores from WYL237 bearing no heterologies around LEU2 (Fig. 1A)
Random spores	His−Ura+Ade−	His−Ura−Ade+	Total	His−Ura+Ade+	His−Ura−Ade−	Total	Total	% crossing over
Leu−	84	99	183	28	23	31	214	14
Leu+	17	81	98	127	14	141	239	60
B. Analysis of random spores from WYL191 containing heterologies flanking LEU2 (Fig. 1B)
Random spores	His+Ura−Ade−	His−Ura+Ade+	Total	His+Ura−Ade+	His−Ura+Ade−	Total	Total	% crossing over
Leu−	379	331	710	33	23	56	766	7
Leu+	50	245	295	457	45	502	797	63
C. Analysis of Leu+ random spores from strains with and without SPO13∷HO (Fig. 1C)
Strain	His+Ura−Ade−	His−Ura+Ade+	Total	His+Ura−Ade+	His−Ura+Ade−	Total	Total	% crossing over
DAM528	144	8	152	28	115	143	295	49
SPO11
DAM509	145	4	149	7	119	126	275	46
SPO11
SPO13∷HO

To facilitate subsequent analysis of crossing over by Southern blot or PCR approaches, we then created strains in which the flanking URA3 and ADE1 regions were present as large heterologies, leaving the 1.8-kb leu2 region surrounded by a 20-kb deletion and a 2.2-kb insertion (WYL191; Fig. 1B). The presence of these heterologies had no significant effect on the association of gene conversion and crossing over, as 63% (502/797) of the Leu⁺ heteroallelic recombinants exhibited an exchange of flanking markers. Among Leu⁻ spores, most of which had not undergone a recombination event in this interval, only 7% (56/766) exhibited crossing over of flanking markers.

To be able to compare Spo11p- and HO-mediated recombination, we further modified this region by replacing the leu2-K allele with the 117-bp cleavage site for HO endonuclease, yielding the leu2-cs allele (Fig. 1C). First, we examined random spores from diploids heterozygous for leu2-cs and leu2-R, which are separated by 399 bp (Fig. 1C). The rate of Leu2⁺ recombinants in a SPO11 strain DAM528 heteroallelic for leu2-cs and leu2-R (0.24 ± 0.1%) was similar to the rate of Leu2⁺ recombinants obtained in isogenic strains with leu2-K and leu2-R (0.3 ± 0.1%). When the leu2-cs/leu2-R diploid expressed both SPO11 and SPO13∷HO (DAM509) the rate of Leu2⁺ recombinants increased 10-fold to 2.5 ± 0.4%. In both of these diploids approximately half of the Leu2⁺ random spores exhibited a crossover of the flanking heterologous markers (49% for DAM528 and 46% for DAM509) (Table 1C).

We then analyzed recombination in this interval by tetrad analysis, comparing strain DAM457 expressing only SPO11 and DAM454 expressing both SPO11 and SPO13∷HO, both diploids carrying leu2-cs and LEU2 (Fig. 1D and Table 2). In the absence of SPO13∷HO, 3% of tetrads contained three LEU2 and one leu2-cs and another 2% had the reverse 1:3 gene conversion pattern, as expected if either leu2 segment could be cleaved by Spo11p. The majority of Spo11p-induced events in this region most likely to occur in the promoter region of LEU2 gene, which is a prominent hot spot for the formation of meiotic DSBs located 400 bp distal to the HO cleavage site (32). In the SPO11 SPO13∷HO strain, 12% of tetrads were 3 LEU2:1 leu2-cs, whereas only 1% were 1:3 events. This is the expected result if SPO13∷HO-induced DSBs in the leu2-cs allele are produced more frequently than SPO11-induced events in LEU2. We estimate that at least 70% of the 3 LEU2:1 leu2 gene conversions in the SPO13∷HO strain were induced by HO, in agreement with the random spore events discussed above. We will discuss later the 12% 4:0 events induced by HO, as they originate from DSBs of both sister chromatids and proved to have distinctly different behaviors.

Table 2.

Effect of SPO13∷HO on meiotic recombination involving leu2∷cs

Strain	Leu+∶Leu−	Total number of tetrads	Tetrads with no crossover between HIS4 and ADE1	Tetrads with crossover between HIS4 and ADE1*	% crossover	Tetrads with coconversion of ADE1
DAM 457	2∶2	286	264	21	7	1
SPO11	3∶1	10 (3%)	4	4	50	2
	4∶0	0	—	—		—
	0∶4	1	—	—		1
	1∶3	5 (2%)	0	4		1
	Total	302
DAM 454	2∶2	457	421	36	8	0
SPO11	3∶1	70 (12%)†	28	38‡	52§	4
SPO13∷HO	4∶0	74 (12%)†	28	18	23¶	28
	0∶4	0
	1∶3	4 (1%)	0	4
	Total	605

^*Parental ditype (PD), tetratype (TT), and nonparental ditype (NPD) asci were identified with respect to the flanking ADE1 and HIS4 markers (see Fig. 1D). Tetrads in which ADE1 was coconverted were not considered. In (2∶2), (3∶1), and (1∶3) cases, all crossover tetrads are TT. Among the 18 4∶0 tetrads from DAM454, 15 were TT and 3 were NPD. 

^† Statistically significant difference (P < 0.001) from the absence of SPO13∷HO. 

^‡ In 34 of 38 cases a Leu⁺ spore experiencing gene conversion contained a crossover product, whereas in 4 cases the crossover did not involve the LEU2-containing chromatid. 

^§ The frequency of crossing over associated with conversion to Leu⁺ was determined after omitting four tetrads in which the adjacent ADE1 marker was coconverted and four tetrads in which crossing over did not involve the gene-converted chromatid. 

^¶ For the 4∶0 tetrads, the frequency of crossing over was calculated on the basis of two gene conversion events per tetrad. Thus there were 15 TTs, with one crossover out of a possible two and three NPDs with two crossovers, of a total of 92 possible exchange events in 46 tetrads. 

In the case of 3 LEU2:1 leu2-cs tetrads in the SPO11 SPO13∷HO diploid DAM454, 58% were tetratype tetrads with respect to the flanking ADE1 and HIS4 markers; that is, two of the four chromatids had undergone a crossing over in the LEU2 interval. The four tetrads in which the ADE1 marker was coconverted were not considered. In 34 of 38 tetratype asci, crossing over involved the Leu2⁺-containing chromatid. After correcting for the four unassociated crossovers, we calculate that the proportion of gene conversions associated with crossovers was 52% (34/66), in good agreement with random spore measurements. These results show that HO-induced events in meiotic cells exhibit the same high level of crossing over as those created by SPO11.

About 6% (4/70) of the 3 LEU2:1 leu2 gene conversion events in the SPO11 SPO13∷HO strain DAM454 were coconverted for the adjacent ADE1 marker, inserted 1076 bp centromere proximal to the HO cleavage site. This is not significantly different from the 2/10 coconversions found among 3 LEU2:1 leu2 tetrads in the SPO11 strain DAM457, lacking SPO13∷HO, where the presumed site of DSBs is located 400 bp distal to leu2-cs. Among tetrads without a leu2 gene conversion ADE1 conversion was rare (Table 2), suggesting that most conversions of ADE1 were dependent on the DSBs that converted leu2.

SPO13∷HO-Induced Meiotic Events Are Different from GAL∷HOInduced Mitotic Recombination.

For this same leu2 interval, we examined HO-induced recombination in mitotic cells, using a galactose-inducible HO gene that was also present in DAM509. The results were different from those induced in meiosis by SPO13∷HO. We selected DSB-initiated repair events arising in the G₂ stage of the mitotic cell cycle (equivalent to the time when recombination happens in meiosis), by plating cells on galactose-containing medium. We considered colonies that were sectored Leu⁺/Leu⁻, where only one of the HO-cut chromatids had converted to LEU2 whereas the second leu2-cs allele was uncut or coconverted to leu2-R. The majority of these events apparently occurred at the G₂ stage, although some of them could have occurred by repairing the HO-induced DSB in subsequent cell cycles. These colonies were analyzed by Southern blots to identify crossovers. Gene conversions coming from the G₂ phase of the mitotic cell cycle are rarely (≤7%) associated with crossing over (3/46 cases).

The gene conversion tract lengths of HO-induced events in mitotic cells were longer than those in meiotic cells, in agreement with previous studies of conversion tract lengths (1). Forty-eight percent (26/54) of all HO-induced mitotic gene conversions examined in G₁ cells resulted in coconversion of the 2.2-kb ADE1 marker, compared with only 6% (4/70) of HO-induced gene conversion in meiosis (Table 2).

We also examined spontaneous mitotic recombination between leu2-K and leu2-R in strain WYL191 with the arrangement of markers shown in Fig. 1B, to compare it to HO-induced recombination. Only 4 of 107 independent Leu⁺ colonies selected in mitosis were homozygous for the distal his4 marker, indicative of a crossover event in G₂ cells. Correcting for the fact that only half of the crossover events should be recovered as His4⁻ because of the random segregation of chromatids in mitosis (31), we estimated that 7% of gene conversions were accompanied by crossing over. Thus HO-induced and spontaneous mitotic recombination events are similar to each other, and both are statistically significantly different from what occurs for both Spo11p- and HO-induced gene conversions in meiosis in the same interval.

Analysis of SPO13∷HO-Induced Recombination in the Absence of the MSH4.

We then asked whether the high level of crossing over of HO-induced events in meiosis depended on the MSH4 gene function, as do SPO11-induced events. Msh4 and Msh5 proteins are involved in regulating the proportion of gene conversions that are associated with crossovers, and mutations in these genes reduce crossing over about 2- to 3-fold without affecting the total number of gene conversions (15, 16). To carry out these experiments, we used diploids homozygous for spo13Δ, which rescues spore viability in recombinationless mutants like spo11Δ (33, 34) and for mutations such as msh4Δ that significantly reduce crossing over (15, 16). spo13Δ causes meiotic cells to undergo only a single meiotic division and to produce two diploid spores.

Before analyzing the effects of msh4Δ on HO-induced meiotic events, we confirmed that spo13Δ did not affect HO-promoted recombination. We dissected the dyads produced by strain DAM518 (spo13Δ SPO11 SPO13∷HO) and determined the frequency of dyads in which one spore had undergone a gene conversion resulting in the formation of a Leu⁺ spore, whereas the other spore was still heteroallelic for leu2-cs and leu2-R (confirmed by the ability of such diploids to give rise to Leu⁺ papillae). These dyads have had a single gene conversion and are therefore the equivalent of a 3:1 tetrad. In DAM518 the frequency of such dyads was 6% (13/223), whereas strain DAM535, carrying SPO11 without SPO13∷HO, gave only 1% (2/209) dyads with one Leu2⁺ spore (P < 0.01) (Table 3A). Thus, as with the SPO13 tetrads analyzed above, approximately 80% of the events in DAM518 are likely to have been induced by SPO13∷HO.

Table 3.

Analysis of Leu⁺-containing dyads from spo13Δ diploids

A. Analysis of unselected dyads from diploids with and without SPO13∷HO
Strain	Relevant genotype	Leu+ spore-containing dyads		Dyads containing only Leu− spores	Total
		Dyads with a single gene conversion†	Others‡
DAM518	SPO11 SPO13∷HO	13 (6%)*	24	186	223
	spo11Δ SPO13∷HO
DAM535	SPO11	2 (1%)	3	204	209
	spo11Δ

B. Analysis of Leu+/Leu− dyads obtained from asci plated on growth medium
Strain	Relevant genotype	Analysis of ADE1–HIS4 crossing over among Leu+-containing dyads§
		No crossover	Crossover	% crossover	Coconversion of HIS4	Coconversion of ADE1
DAM518	SPO11	12	13	52¶	1	0
	spo11Δ
DAM502	spo11Δ	30	11	27	5	3
	spo11Δ
DAM516	SPO11 msh4Δ	37	8	18	1	3
	spo11Δ msh4Δ

C. Analysis of unselected dyads containing a single gene conversion event
Strain	Genotype	Total number of dyads analyzed	Gene conversion of one spore in the dyad		Among gene conversions of one spore, coconversion of HIS4‖	Among gene conversions of one spore, coconversion of ADE1‡‡
			Single gene conversion, with Leu+ spore	Single gene conversion, coconversion of leu2-R
DAM518	SPO11	223	13/28 (42%)	15/28 (58%)	3/28 (11%)	1/9 (11%)
	spo11Δ
DAM502	spo11Δ
	spo11Δ	344	4/30 (13%)**	26/30 (87%)**	12/30 (40%)**	3/9 (33%)

^† Dyads in which one spore had undergone a gene conversion resulting in Leu⁺ phenotype while the other spore was still heterozygote for leu2-cs and leu2-R (confirmed by the ability of such diploids to give rise to Leu⁺ papillae).  

^* indicates a statistically significant difference from the absence of SPO13∷HO (P < 0.01). 

^‡ Dyads in which both spores undergone gene conversion events (analogous to 4∶0 tetrads) or dyads in which a Leu⁻ spore did not give rise to Leu⁺ papillae because of reductional chromosomal division, chromosome missegregation, or crossing over in the interval between the LEU2 gene and the centromere. 

^§ Sectored colonies arising from the germination of two spores in an ascus were identified after germination on YEPD medium by replica plating them to medium lacking leucine. Leu⁺/Leu⁻ sectored colonies were then analyzed to show that the Leu⁻ half still retained the leu2-cs allele; hence there was a single gene conversion event in this meiosis. 

^¶ Statistically significant difference from the spo11Δ strain (P < 0.05) and the msh4Δ strain (P < 0.01). 

^‖ One of three His⁻ segregants obtained from DAM518 was Leu⁺. The other two His⁻ segregants obtained from DAM518 and all 12 His⁻ segregants obtained from DAM502 were Leu⁻. 

^**Indicates a statistically significant difference from the Spo11⁺ strain (P < 0.01). 

^‡‡ Only a subset of events were analyzed to determine if ADE1 was heterozygous or homozygous. 

A larger number of Leu⁺ recombinants arising from a single HO cleavage were then obtained by plating complete dyads without microdissection and analyzing Leu⁺/Leu⁻ sectored colonies where the Leu⁻ sector was still heterozygous for leu2-cs and leu-R (Table 3B). Southern blot and genetic analysis showed that 52% of the predominantly SPO13∷HO-induced events in strain DAM518 were associated with crossing over (Table 3B), similar to the 52% we observed in tetrad analysis performed in Spo13⁺ diploids (Table 2). Again, we observed that HO-induced gene conversions are associated with crossing over in meiosis almost 8 times more frequently than in mitosis, again suggesting that the high level of crossing over is not a special feature of SPO11-induced events.

Having shown that spo13Δ diploids behaved similarly to wild-type diploids in the way HO-induced events were repaired, we analyzed Leu⁺-containing dyads obtained from meiosis of spo13Δ msh4Δ diploids carrying leu2-cs and leu2-R. As shown in Table 3B, a msh4Δ mutation reduces SPO13∷HO-induced crossovers by about 3-fold (from 52% to 18%), similar to the reductions that have been observed for SPO11-induced events in previous studies (15, 16). The overall frequency of Leu⁺-containing dyads was not significantly different between msh4Δ mutants (4.4 ± 1.7%) and wild-type controls (6.6 ± 2.7%).

SPO13∷HO Induces Gene Conversions with a High Proportion of Crossovers, Even in the Absence of SPO11.

After showing that Spo11p is not required at the site of the DSB (i.e., in cis) to ensure high levels of crossing over, we asked whether some function of Spo11p is nevertheless needed to ensure the full level of crossing over associated with gene conversion (i.e., that Spo11p plays a role in trans). We used the spo13Δ spo11Δ SPO13∷HO diploid DAM502, heterozygous for leu2-cs and leu2-R. Here, the only DSBs in meiosis come from HO cleavage. Crossing over in those dyads with one Leu2⁺ spore was reduced from 52% to 27% (Table 3B). Thus in the absence of Spo11p, crossing over still occurs at a level 4 times higher than in mitotic cells, but 2-fold lower than in a SPO11 strain. This 2-fold reduction could be due to a modulating effect of other nearby recombination events that would be created by Spo11p and/or to the fact that spo11Δ mutants fail to form SC. Our result is analogous to previous studies showing that zip1Δ or zip2Δ mutations, which prevent formation of SC in SPO11 diploids, reduce the proportion of gene conversions accompanied by crossing over by about 2-fold (13, 14).

The absence of Spo11p also affects the lengths of gene conversion tracts of HO-induced meiotic gene conversions. In the spo13Δ spo11Δ SPO13∷HO diploid DAM502, 9% (30/344) of the dyads had a single gene conversion of leu2-cs (Table 3C), compared with 13% (28/223) in the Spo11⁺ diploid, but there was a significant reduction in the percentage of Leu2⁺ segregants among single gene conversion events (13% in spo11Δ (4/30) instead of 42% (13/28) in SPO11). Most of the reduction in Leu2⁺ recombinants was due to enhanced coconversion of leu2-cs to leu2-R in the spo11Δ diploid compared with the SPO11 strain, so that the gene convertants were Leu2⁻. In addition to the increased coconversion of the leu2-R marker 400 bp away from the HO cleavage site, the spo11Δ SPO13∷HO strain also showed a significant (P < 0.01) increase in the frequency of coconversions of the adjacent 20-kb heterology extending to HIS4 (40% versus 11% in SPO11 SPO13∷HO). This same trend was found when we analyzed a subset of events for coconversions of the adjacent ADE1 insertion by Southern blots (Table 3C).

The long conversion tracts extending 20 kb distal to the leu2 region could in fact be much longer, extending all the way to the end of the chromosome. To examine how long those conversion tracts arising in the spo11Δ spo13Δ SPO13∷HO strain could become, we used a diploid in which the leu2-cs-containing chromosome carried URA3, inserted at the HML locus, 12 kb from the chromosome end (Fig. 1E). Of 110 dyads analyzed, 13 were identified as having one His⁻ and one His⁺ spore. Of these, two were the result of chromosome loss, as they had also lost MATa on the opposite side of the centromere. By Southern blot analysis two others were shown to be reciprocal crossovers, but nine were the result of a nonreciprocal event that is consistent with a break-induced replication mechanism (35, 36) that apparently copied the His⁻ chromosome all the way to the end. This demonstrates that break-induced replication can occur in meiotic cells. These data reveal that HO-induced meiotic recombination in the absence of SPO11 is significantly different from what is seen in a SPO11 spo13Δ SPO13∷HO strain: the proportion of crossovers accompanying gene conversion is reduced, the length of gene conversion tracts is greatly increased, and many breaks are apparently repaired by nonreciprocal break-induced replication rather than by gene conversion.

SPO13∷HO spo11Δ Mutants Do Not Form Visible SC.

spo11Δ cells fail to form the tripartite synaptonemal complex (SC) that has been implicated in the control of crossing over (37). Hence it was of great interest to know whether a single DSB induced by SPO13∷HO in a spo11Δ strain would be sufficient to promote SC formation. We examined two independent spo13Δ spo11Δ SPO13∷HO diploids and their isogenic spo11Δ and SPO11 controls lacking SPO13∷HO by immunostaining of chromosome spreads (28), using the synapsis-specific anti-Zip1 antibody (38). Whereas SPO11 diploids showed normal levels of partial and complete SCs at 5, 6, and 7.3 h after induction (30–65%), SC-like structures were found exceedingly rarely in the spo11Δ strains (0.2–1%), regardless of the presence of the SPO13∷HO construct. More than 1000 nuclei were inspected per time point. Many of the spread spo11Δ nuclei contained a Zip1 positive polycomplex (PC) and occasionally rigid linear structures, which, because they did not correspond to DAPI-labeled chromatin, were interpreted as fragments of the PC. Thus, the induction of a single DSB by SPO13∷HO was not sufficient to induce formation of extensive SCs. This result raises the possibility that lack of synaptonemal complex by itself is responsible for the very unusual long gene conversion tracts seen in spo11Δ SPO13∷HO cells, as well as for crossover levels that are 2-fold lower than maximum.

Cleavage of Both Sister Chromatids Causes a Disruption of Meiotic Recombination.

In normal meiosis, there are few, if any, tetrads in which both sister chromatids are cleaved simultaneously at the same hot spot, to produce a 4:0 tetrad by two independent gene conversion events. This is also evident from the fact that Spo11-mediated DSBs created in a haploid strain undergoing meiosis can be repaired by homologous recombination, using their sister chromatids as templates (39); such a repair would be impossible if both chromatids were cut, in the absence of a homologous chromosome. Therefore the consequences of cleaving both sister chromatids of one homologue at the same site have not previously been assessed.

In the SPO13∷HO diploids about 12% of the tetrads had four LEU2 spores (Table 2), apparently the consequence of two gene conversion events resulting from the cleavage of both sister chromatids. Although some events might have been the consequence of HO-induced recombination before meiosis, we could show that most of them most likely arose after premeiotic DNA replication, as 38% of them had one crossover and one noncrossover event and another 12% showed different coconversions of the flanking ADE1 or HIS4 regions. The recombination events in these cases appeared to be different from what was obtained in tetrads with a single SPO13∷HO-induced event. First, although 52% of tetrads with 3 LEU2:1 leu2-cs were accompanied by crossing over, only 23% were crossover-associated in the 4 LEU2:0 leu2-cs cases. This value was calculated on the basis that there are two conversion events per tetrad, including 15 with one crossover and 3 tetrads with 2 crossovers, of a total of 92 events in 46 tetrads. The calculated crossover rate of 23% is significantly different (P < 0.001) from the rate of 52% among the 3:1 tetrads. It is possible that this reduced level of crossing over accompanying gene conversions could be a reflection of the longer apparent gene conversion tracts among the 4:0 tetrads than the 3:1 tetrads, although no other data support this possibility.

The proportion of observed four-strand double crossovers in the 4:0 tetrads (6.5%) corresponds to the expected frequency, based on the calculated crossover rate of 0.23 among these 4:0 events and assuming that only four-strand double crossovers can result when the two sister chromatids are cleaved at the same site. This conclusion suggests that there is no interference in double crossovers in the interval containing leu2. We note that apparent interference in crossing over was seen previously by Kolodkin et al. (40), who studied GAL∷HO-induced 4:0 events at the MAT locus in meiotic cells; but in their case, only 4:0 tetrads were recovered, and it was not clear whether many of these events were initiated before premeiotic DNA replication.

Discussion

We have compared the outcomes of recombination events in the same chromosomal location, initiated by DSBs created by the same HO endonuclease, in mitotic and meiotic cells. This study examines the recombination induced by the same DSB under these two very different situations. We find that the fraction of gene conversions accompanied by crossing over is 8 times greater in meiosis than in mitotic cells and is comparable to Spo11p-generated events in the same well-defined genetic interval. We conclude therefore that two of the special properties of meiotic recombination, the frequent crossovers accompanying gene conversion and the shorter conversion tract lengths compared with mitotic events, are not dependent on initiation by Spo11p. Thus, neither the different overhanging ends of Spo11p- and HO-generated DSBs nor the specific sequences in which DSBs are made account for the difference between mitotic and meiotic gene conversions. Moreover, although the ability of Spo11p to create DSBs depends on more than 10 proteins, some of which have been shown to bind to DNA and in some cases alter the chromatin structure of a hotspot (11, 41), this special chromatin context does not appear to be required to achieve the characteristic meiotic outcome.

By creating an HO-induced DSB in the absence of any other meiotic DSBs, we also demonstrated that a single HO-induced gene conversion had about half the level of crossing over (27%) as was seen when Spo11p was also active (52%). Apparently a single HO-induced DSB is insufficient to create a visible extent of SC. One explanation for the reduced level of crossing over in the spo11Δ SPO13∷HO strain is that it lacks the synaptonemal complex that is somehow responsible for increasing the proportion of gene conversions accompanied by crossing over. The reduced level of crossing over we obtained in a spo11Δ SPO13∷HO strain is analogous to what has been seen previously in SPO11 strains lacking the SC proteins Zip1p and Zip2p (13, 14).

Our results suggest that the special features of meiotic recombination can be attributed to processes occurring after the DSB is generated. This assumption focuses our attention on meiosis-specific recombination proteins such as Dmc1p and/or meiosis-specific chromosome structures, including the synaptonemal complex, meiosis-specific cohesins (42), or other possible chromatin modifications. Further support for this conclusion is that the absence of Msh4p reduces crossing over of HO-induced gene conversions to the same extent that others have reported for Spo11p-mediated events. Thus we are now in a position to use several revealing assays, developed in mitotic cells using HO endonuclease (24, 43), to compare in detail the molecular mechanisms of meiotic and mitotic recombination. Further experiments should also establish whether HO-induced recombination is sufficient to drive reductional division, another key characteristic of meiosis.

Abbreviation

DSB

double strand break