Creating strains in which HO endonuclease is the sole initiator of meiotic recombination

We created strains in which HO endonuclease is the sole source of meiotic DSB activity by integrating a DNA cassette, P_SPO13-HO, which contains the HO ORF downstream of promoter sequences for the meiosis-specific SPO13 gene, as was done by a prior study to assess HO-mediated meiotic DSBs (Malkova et al. 1996b). We integrated P_SPO13-HO at the LYS2 locus in strains missing both endogenous HO activity and SPO11 (Figure 1A).

A spo13 mutation was used to facilitate analysis of interhomolog recombination in a spo11 strain background. The spo13 mutation allows the isolation of viable spore products from diploid cells that fail to initiate or complete recombination, because spo11 spo13 meiotic cells undergo a single equational division to produce two diploid (dyad) spores instead of four haploid spores (Klapholz and Esposito 1980). Genetic markers can be inspected in the diploid spores to determine if interhomolog recombination has occurred.

Robust HO endonuclease activity was detectable in P_SPO13-HO-containing meiotic cells, as meiotic spore products frequently carried a chromosome III that had undergone interhomolog recombination at the MAT locus, which carries the sequence that HO endonuclease targets (HO cs) (Haber 2012). In the absence of meiotic recombination at MAT, a spo11 spo13 (MATa/MATα) meiotic cell will produce two MATa/MATα diploid spores (Figure 1B). Indeed, almost all of the dyad spores that result from control spo11 spo13 diploids (with no HO endonuclease expression) fail to mate with either MATa or MATα cells, as predicted for the MATa/MATα genotype (Table 1). By contrast, approximately 40% of dyads from strains carrying P_SPO13-HO carry at least one diploid spore that is phenotypically either MATa or MATα (Table1), reflecting meiotic interhomolog recombination at MAT. These data are similar to the findings of (Malkova et al. 1996b), where 29% of otherwise recombination-less (rad50 mutant) meioses involving P_SPO13-HO expression and a single cleavable MAT locus displayed interhomolog conversion at MAT; this prior study also determined that every HO-associated meiotic conversion event arose from interhomolog recombination, vs. intra-chromosomal recombination using one of the two silent mating type loci.

We observed similar levels of meiotic recombination at MAT in each of four P_SPO13-HO strains containing chromosome IV-targeted HO cs loci (generated for experiments discussed below; Table 1). Diploid strains carrying P_SPO13-HO display very little mating type switching during vegetative growth (see Materials and Methods), indicating that HO endonuclease activity in our P_SPO13-HO strains is nearly completely meiosis-specific.

HO forms DSBs at HO cs loci on chromosome IV during spo11 meiosis

P_SPO13-HO strains were used to investigate whether the position of a meiotic DSB along the chromosome is a variable in assuring homologous centromere pairing and SC assembly. Toward this end, we integrated an HO cs at nine distinct locations on chromosome IV, including a site positioned 250 nucleotides to the left of CENIV (HO cs2; Figure 1A; Materials and Methods). Except for cs1 and cs2, the chromosomal position of each of our HO cs integration sites is associated with a high probability of DSB formation by Spo11 during wild-type meiosis (Blitzblau et al. 2007; Pan et al. 2011). Neither cs1 nor cs2 are positioned at sites that have been found to be strongly enriched for meiosis-specific axis proteins but the position of cs2 has been found to be strongly enriched for the meiosis-specific cohesin subunit, Rec8 (Panizza et al. 2011).

We directly evaluated HO-mediated DSB formation in six spo11 null strains, each homozygous for a different chromosome IV HO cs (Figure 2). These strains were additionally homozygous for a rad51 null mutation, in order to slow or abolish the completion of DNA repair and thereby allow us to evaluate the maximal level of HO-mediated DSB formation at these chromosomal sites during meiosis. Genomic DNA was isolated from HO cs-containing strains at 0, 12, 18 and 24 hr of sporulation; prior studies using this strain background indicate that most cells at the 24 hr time point have reached or progressed beyond late meiotic prophase (Voelkel-Meiman et al. 2012). Genomic DNA was digested using restriction enzymes that create a 2-7 kb DNA fragment encompassing the HO cs, and probed for HO cs-associated sequences using a 500 bp probe on a Southern blot.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

HO-mediated DSBs at HO cs sequences on chromosome IV in spo11 rad51 meiosis. (A) Southern blot analysis shows DNA cleavage by HO endonuclease at HO cs sequences on chromosome IV in spo11 spo13 rad51 diploid strains (LY491, LY456, LY492, LY481, LY459, LY457 and LY458; see Figure 1 for HO cs positions, Table S4 for strain genotypes). Samples were collected and processed at 0, 12, 18 and 24 hr after placement in sporulation medium. Genomic DNA was digested with restriction enzymes that target sites flanking each of the HO cs loci within a 10 kb region; DNA fragment sizes (kb) are displayed next to blots. Fragments were visualized using a probe that hybridizes to the natMX4 sequence adjacent to each HO cs. In the absence of HO and before entry into meiosis in the presence of HO (0 hr of sporulation), the probe detects a single large fragment that corresponds to an intact fragment of DNA containing the HO cs. In the presence of HO endonuclease, a faster migrating (smaller) fragment is also seen for each of the cut sites beginning at 12hr, except for cs6. The absence of HO-mediated DSBs at cs6 was verified using two different restriction enzymes (See Materials and Methods). (B) Bar graph shows the percent of DNA cut by HO endonuclease within P_SPO13-HO spo11 spo13 rad51 meiotic nuclei at the indicated time points. Values were calculated by dividing the intensity of the smaller fragment with the sum of the intensities of the smaller and larger fragment. Bars depict the range given by two experiments.

In strains devoid of P_SPO13-HO, or at the zero-hour time point in strains carrying P_SPO13-HO, a single DNA fragment of predicted size was detected (Figure 2A). In strains carrying P_SPO13-HO and chromosome IV-targeted HO cs1, cs2, cs4, cs5 or cs7, a faster migrating fragment, corresponding to the product of an HO-mediated DSB, was also detected (Figure 2A). In these five strains, the percentage of DNA cut by HO endonuclease ranged from ∼20–60% (Figure 2B). These numbers suggest that HO endonuclease maximally cuts one or two out of the four available chromatids in a given meiotic nucleus. We note the possibility that this calculation underestimates the true number of HO-mediated DSBs, as some DSBs may be undetectable due to hyperresection of sequences flanking the DSB site (which has been noted to occur in rad51 mutants (Shinohara et al. 1992). However, our probe did not detect faster migrating HO cleavage products, which would be expected observable intermediates in the case of substantial resection activity.

Unexpectedly, DSB formation at HO cs6 was not detected in Southern blotting experiments with either of two different restriction enzymes (Figure 2A, B; Materials and Methods). HO endonuclease is active in the cs6 strain, based on observed meiotic recombination at the MAT locus that was comparable to strains carrying the other HO cs loci (Table 1). The position of cs6 on chromosome IV has been previously classified as a frequent target of Spo11 (Blitzblau et al. 2007; Pan et al. 2011); genetic background may affect the accessibility of cs6 DNA to the HO endonuclease in our strain.

HO promotes interhomolog recombination on chromosome IV in spo11 meiotic nuclei

Meiotic DSBs preferentially engage the homolog over the sister chromatid for repair (Schwacha and Kleckner 1994; Schwacha and Kleckner 1997; Hong et al. 2013). To ask whether HO-mediated meiotic DSBs on chromosome IV engage the homologous chromosome for repair when Spo11 is absent, we assessed interhomolog crossing over between genetic markers that flank an HO cs (Figure 3A). The LEU2 gene was introduced near to the centromere on one copy of chromosome IV, either 705 bases to the right of CEN4 in spo11 spo13 strains homozygous for HO cs5, cs6 or cs7 or 2.8 kilobases to left of CEN4 in analogous strains homozygous for HO cs2. In all strains, the THR1 gene was inserted at coordinate 1,416,000 on the right arm of the chromosome IV carrying LEU2+. The LEU2 and THR1 genetic markers thus allowed us to measure the frequency of HO-mediated crossing over.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

HO-mediated interhomolog recombination during meiosis. (A) Assay for interhomolog recombination at HO cs loci on chromosome IV in spo11 spo13 diploids. LEU2 is inserted at 450 kb (705 bp to the right of CEN4) and THR1 is inserted at 1,416 kb on chromosome IV for all strains, except for cs2 where LEU2 is at 447 kb (2.8 kb to the left of CEN4). A single HO cs is integrated between LEU2 and THR1. spo11 spo13 diploid cells undergo a single equational division during meiosis, resulting in spores carrying a chromatid from each parental homolog. In the absence of HO, each spore receives one LEU2 THR1 chromatid and one chromatid with neither marker (“No Recombination” column). In the presence of HO mediated DSBs, interhomolog reciprocal crossover recombination can result in a spore lacking the THR1 marker on chromosome IV (accompanied by a sister spore with two THR1 markers; upper right dyad). Because half of the crossover events will be invisible by this assay (lower right dyad), the percentage of apparent reciprocal crossovers were calculated by dividing twice the number of observed Thr- spores by the number of total 2-spore viable dyads (Table 2 and Table 4). (B) The upper bar graph plots the percentage of apparent crossing over calculated as described in (A) (n > 100 2-spore viable dyads assayed; Table 2) in various strains that carry no chromosome IV HO cs (LY208), or that carry distinct chromosome IV HO cs locations (left to right: LY208, LY555, LY207, LY324, LY322). The lower bar graph plots the percentage of apparent crossing over in mutant strains carrying HO cs5 (left to right: LY207, LY459, LY290, LY393, LY904, LY939, LY935, LY957 and LY910). Vegetative cultures of control expressing no HO nor HO cs (LY407) as well as LY208, LY555, LY324, LY322 and LY207 were independently evaluated for the uniform presence of the LEU2 and THR1 marker by assessment of growth of >400 single colonies on selective media.

In the absence of HO, spo11 spo13 dyads heterozygous for LEU2 near the centromere and THR1 on the arm of chromosome IV would be expected to undergo equational division to give diploid dyads containing spores that are each heterozygous for the LEU2 and THR1 insertions (Figure 3A, “No Recombination” outcome). An HO-mediated, interhomolog crossover recombination event at an HO cs to the right of the centromere and between the LEU2 and THR1 markers, however, can result in a situation where both chromosomes IV carry one chromatid that is devoid of the THR1 insertion (Figure 3A, “Recombination” outcome). Half of the diploid dyads arising from meioses involving such recombinant chromosomes IV will carry a spore that is phenotypically Leu+ Thr-. Figure 3A illustrates the phenotypes of each spore in such recombinant dyads. Note that a reciprocal crossover at cs2, which is positioned to the left of CENIV, will result in “recombinant” dyads that contain a Leu-, Thr+ spore.

We measured percent of meioses with an apparent reciprocal crossover event by dividing twice the number of “recombinant” dyads (dyads containing a Leu+, Thr- spore for non-cs2 strains and containing a Leu-, Thr+ spore for the cs2 strain) by the total number of dyads analyzed (Figure 3B, Table 2). Based on this calculation, HO-mediated, interhomolog crossover recombination occurred in ∼21% of meioses for HO cs5, 18% of meioses for HO cs7, and 19% of meioses for HO cs2 (closest to CENIV) (Table 2, Figure 3B). Consistent with the absence of detectable DSBs at HO cs6, interhomolog crossover recombination could be detected at this cut site in only 1.3% of meioses (Table 2, Figure 3B).

Table 2 and Figure 3 refer to LEU2-THR1 recombinants as “apparent” crossovers, because the Leu+, Thr- spores that we have presumed to be due to interhomolog crossover recombination could instead arise from an interhomolog recombination event that is nonreciprocal, in which sequences encompassing THR1 are converted without an accompanying crossover. In the case of an interhomolog conversion without a reciprocal crossover, the Leu+, Thr+ sister spore in the recombinant dyad is expected to carry one chromatid devoid of the THR1 insert. Whereas in the case of a reciprocal crossover, both chromatids in the Leu+, Thr+ spore within a recombinant dyad are expected to contain the THR1 insert. We used PCR to detect the presence or absence of the THR1 insert in Thr+ spores of 44 recombinant dyads for cs5-carrying strains, and to detect the presence or absence of the LEU2 insert in Leu+ spores of 54 recombinant dyads from cs2-carrying strains. We found that the 71% (31 out of 44) of Thr+ spores from cs5 recombinant dyads carried only THR1 insert-carrying chromosomes IV, indicating that 71% of recombinant dyads in this strain result from a reciprocal crossover event. ∼30% of Thr+ spores were heterozygous for THR1 insert-carrying chromosome IV, indicating that the recombinant spore in these dyads likely resulted from a nonreciprocal interhomolog recombination event instead of a reciprocal crossover. In an analogous analysis for cs2, we found that both chromosomes IV carried the LEU2 insert in only 52% (28 out of 54) of Leu+ spores from cs2 recombinant dyads. Thus, about half of the recombinant dyads from the cs2 strain result from a reciprocal crossover event, while the remainder apparently derive from a nonreciprocal recombination event.

The higher fraction of nonreciprocal interhomolog recombination events in cs2 vs. cs5 strains may be due to the proximity of the LEU2 marker to the DSB site: the LEU2 marker is approximately 2.5 kilobases from cs2, whereas the THR1 insert is ∼580 kilobases from cs5. On the other hand, it is remarkable that ∼30% of interhomolog recombination events in cs5 strains involve conversion of sequences ∼580 kilobases from the DSB site. Such remarkably long conversion tracts have been previously associated with HO-mediated DSB repair in spo11 meiotic nuclei (Malkova et al. 1996b; Malkova et al. 2000) and have been proposed to arise through a Break-Induced Replication (Kellis et al. 2004) process whereby a strand invasion intermediate is extended by replication through the end of the chromosome instead of undergoing second-end capture (Malkova et al. 1996a; Kraus et al. 2001; Malkova and Ira 2013).

In sum, as was reported in (Malkova et al. 2000) for the endogenous HO cs on chromosome III, our data suggest that HO-mediated meiotic DSBs on chromosome IV can engage a non-sister chromatid for their repair. Furthermore, our data indicate that a non-sister chromatid is utilized for HO-mediated meiotic DSB repair in at least 20% of meioses, and that a substantial fraction (∼30%) of HO-mediated interhomolog events involve extremely long conversion tracts.

Dmc1 is dispensable for HO-mediated interhomolog recombination in meiotic cells

We next asked whether HO-mediated interhomolog recombination events in spo11 meiosis rely on meiosis-specific strand exchange machinery. In rad51 single or rad51 dmc1 double mutants, the viability of diploid dyad spores is dramatically decreased, from 74% in the spo11 spo13 control meiocytes expressing HO, to 38% in the analogous strain missing Rad51, and to 34% in the combined absence of Rad51 and Dmc1 (Table 3). Genetic removal of HO endonuclease from the rad51 single or rad51 dmc1 double mutant strain restored dyad spore viability to 76% and 77%, respectively (Table 3), indicating that the spore lethality observed in our HO-expressing, rad51 strains is due to a failure to repair HO-mediated DSBs.

We also found that HO-mediated interhomolog crossovers at cs5 are nearly abolished in the absence of Rad51. Apparent crossovers between the LEU2 and THR1 markers flanking HO cs5 decreased from 20.8% observed among two-spore viable dyads from spo11 spo13 control meiocytes expressing HO, to 0% and 1.4% observed among two-spore viable dyads from spo11 spo13 rad51 and spo11 spo13 rad51 dmc1 strains expressing HO, respectively (Figure 3B, Table 2). HO-mediated meiotic interhomolog conversion events at the MAT locus also disappeared in spo11 spo13 rad51 meiotic cells expressing HO: 99.0% and 96.4% of two-spore viable dyads from spo11 spo13 rad51 and spo11 spo13 rad51 dmc1 strains, respectively, exhibited two non-mating spores (n = 105; Table S1A), similar to the proportion of non-mating spores from spo11 spo13 strains that lack meiotic HO expression altogether (Table 1). Furthermore, only 2 out of 181 (1.1%; Table S1B) one-spore viable dyads from P_SPO13-HO spo11 spo13 rad51 meiotic cells and 5 out of 363 (1.4%; Table S1B) one-spore viable dyads from P_SPO13-HO spo11 spo13 rad51 dmc1 meiotic cells exhibited the capacity to mate; these few MAT homozygotes may have arisen as a consequence of meiotic recombination or chromosome loss. Similarly, only 4 out of 181 (2.2%) and 1 out of 363 (0.3%) one spore viable dyads from P_SPO13-HO spo11 spo13 rad51 strains and P_SPO13-HO spo11 spo13 rad51 dmc1 strains, respectively, exhibited a non-parental configuration of the LEU2 and THR1 markers that flank cs5 on chromosome IV.

The low spore viability and absence of interhomolog recombination when Rad51 is missing from HO-expressing, spo11 spo13 meiocytes indicate that Rad51 is essential for robust repair of HO-mediated DSBs during meiosis, which aligns with the expectation that strand exchange is a requirement for meiotic recombination. Most strikingly, in contrast to Spo11-mediated meiotic DSBs, which are capable of utilizing a Dmc1-only pathway for repair by homologous recombination when Rad51 is absent (Bishop et al. 1992; Shinohara et al. 1997a), our data reveal that HO-mediated meiotic DSBs are incapable of undergoing homologous recombination using a Dmc1-only pathway.

In contrast to the Rad51-deficient context, no decrease in spore viability is associated with HO DSB activity at cs5 in the absence of Dmc1 (75% spore viability in spo11 spo13 dmc1 vs. 74% in spo11 spo13 DMC1 strains respectively; Table 3) and apparent crossovers at HO cs5 are only slightly decreased in spo11 spo13 meiotic cells missing Dmc1 (15.8% in dmc1 vs. 20.8% in DMC1 strains; Figure 3; Table 2). Furthermore, HO-mediated interhomolog recombination in spo11 spo13 meiotic cells appears unaffected by the individual removal of two factors that normally promote the use of the Dmc1 recombinase (Schwacha and Kleckner 1997; Callender et al. 2016): HO-mediated apparent interhomolog crossover levels are unchanged in spo11 spo13 meiotic cells missing the meiosis-specific kinase, Mek1 (20.8% in both spo11 spo13 control and spo11 spo13 mek1 meiotic cells; Figure 3, Table 2), and in the absence of the meiosis-specific chromosomal protein, Red1 (22.4% in spo11 spo13 red1 strains; Figure 3, Table 2).

Thus, in contrast to wild-type meiosis, where removal of factors that promote Dmc1 utilization causes a severe reduction in interhomolog crossovers (Rockmill and Roeder 1991; Bishop et al. 1992; Rockmill et al. 1995; Shinohara et al. 1997a; Bishop et al. 1999; Tsubouchi and Roeder 2003), HO-initiated interhomolog DSB repair in our experimental (spo11) strain is not dramatically affected by the loss of Mek1, Red1, or the Dmc1 recombinase, apparently because these HO-mediated DSBs can easily access a “Rad51-only” recombinase pathway.

Can HO-mediated interhomolog recombination in meiotic cells utilize Dmc1 when Rad51 recombinase activity is absent?

The data presented above establish a requirement for Rad51 and the lack of a requirement for Dmc1 in the interhomolog repair of HO-mediated DSBs in meiosis. We used a Rad51 separation-of-function genetic context to determine if the Dmc1 recombinase functions redundantly with Rad51 recombinase activity, or whether Dmc1 is completely unavailable to the repair of HO-mediated meiotic DSBs in spo11 mutants.

We analyzed HO-mediated meiotic DSB outcomes in the context of the rad51-II3A separation-of-function allele (a kind gift of D. Bishop; Cloud et al. 2012). rad51-II3A encodes a protein that lacks strand exchange activity and thus supports a Dmc1-Rad51 joint recombinase pathway but not a Rad51-only pathway. We observed robust HO-mediated interhomolog recombination in spo11 spo13 meiotic cells homozygous for rad51-II3A (15.7%; Figure 3, Table 2). Furthermore, we observed a strong diminishment in apparent interhomolog crossovers when Dmc1 is removed from the rad51-II3A strain, from 15.7% in rad51-II3A to 2% in rad51-II3A dmc1 (Figure 3, Table 2), indicating that the interhomolog recombination measured in rad51-II3A mutants is primarily due to Dmc1 recombinase activity.

Altogether these data indicate that HO-initiated, interhomolog recombination intermediates may primarily engage the Rad51 recombinase, but are capable of utilizing Dmc1 as a recombinase so long as Rad51 is also present.

spo11 spo13 rad51-II3A meiotic cells that express HO exhibit diminished spore viability relative to control strains (52% vs. 74% in the control; Table 3). This observation indicates that Dmc1 is only partially capable of rescuing genome-wide HO-mediated meiotic DSB repair when Rad51 recombinase activity is diminished. Consistent with this interpretation, we observe evidence of Dmc1-mediated recombinase activity even in meiocytes that presumably also sustained lethal unrepaired HO DSBs due to the absence of Rad51 activity: ∼29% (n = 189) of one-spore-viable dyads from rad51-II3A spo11 spo13 strains expressing P_SPO13-HO contain a spore that is homozygous at the MAT locus (presumably due to interhomolog recombination at MAT), in contrast to rad51 and rad51 dmc1 spo11 spo13 strains expressing P_SPO13-HO where only ∼1% of one-spore viable dyads carry a spore that is homozygous at MAT (Table S1B).

HO-mediated DSB repair in meiotic cells is influenced by Mre11 and Xrs2

The Mre11 nuclease, Rad50 ATPase, and FHA-containing Xrs2 protein can form a complex (“MRX”) that functions in DSB repair in mitotic cells, and has been implicated in both DSB formation and repair during meiosis (Borde 2007; Gobbini et al. 2016). We investigated the role of this complex in the repair of HO-mediated, meiotic DSBs in spo11 mutants by examining interhomolog recombination in spo11 spo13 mre11 and spo11 spo13 xrs2 meiotic cells that carry P_SPO13-HO and the HO cs5. We found that, even in the absence of HO expression, the mre11 and xrs2 mutations alone confer a substantial decrease in spore viability to spo11 spo13 strains (50% and 65%, respectively, vs. 89% for the spo11 spo13 control; Table 3). This data indicates that Mre11 and Xrs2 activities are critical for maintaining genomic integrity during a spo11 spo13 meiotic cell cycle even in the absence of meiotic DSBs, and is consistent with a prior finding that diminished Mre11 activity reduces the spore viability of a spo13 strain (Ajimura et al. 1993). In strains carrying P_SPO13-HO, we observed that the mre11 mutation confers a further reduction in spore viability (from 50% in the absence of P_SPO13-HO to 21% in the presence of P_SPO13-HO; Table 3), indicating that Mre11 is also important for the repair of HO-mediated meiotic DSBs. Zero out of thirteen two-spore viable dyads and only 3 out of 60 (5%) one-spore-viable dyads produced by the spo11 spo13 mre11 mutant exhibited a Leu+ Thr- marker configuration, but this dataset is too small to conclude that HO-mediated interhomolog and/or crossover repair is dependent on Mre11 in spo11 spo13 meiotic cells. The strikingly low spore viability might lead one to conclude that HO-mediated meiotic DSBs uniformly fail to repair in spo11 spo13 mre11 meiocytes expressing P_SPO13-HO. However, this may not be the case: 46% of two spore viable dyads and 45% of one spore viable dyads from spo11 spo13 mre11 mutants were found to exhibit at least one mating capable spore (Table S1). While diploid spores homozygous for MAT can arise from a spo11 spo13 meiosis as a consequence of chromosome loss, the near absence of such MAT homozygotes among two-spore-viable and one-spore-viable dyads from spo11 spo13 rad51 strains suggests that chromosome loss rarely occurs in these HO-expressing meiocytes and thus at least some HO-mediated DSBs in spo11 spo13 mre11 meiotic cells are likely repaired via interhomolog recombination. Thus, we conclude that Mre11 is not uniformly required for the interhomolog repair of HO-mediated DSBs during meiosis, but that a subset of HO-mediated meiotic DSBs cause inviable meiotic products when Mre11 is absent.

Similar to Mre11, Xrs2 is not required for the interhomolog repair of HO-mediated meiotic DSBs, as is evident by the robust HO-mediated interhomolog recombination previously observed at MAT in SPO11 spo13 xrs2 strains (Malkova et al. 1996b), Indeed, similar to spo11 spo13 control strains that express P_SPO13-HO (Table 1), we found that sporulated P_SPO13-HO spo11 spo13 xrs2 strains give rise to a large fraction (31.5%; Table S1A) of two spore viable dyads in which one or both spores is mating-capable, indicative of robust meiotic interhomolog recombination at the MAT locus. In contrast to spo11 spo13 mre11 strains, however, meiotic expression of HO in spo11 spo13 xrs2 strains is associated with only a slight reduction in spore viability (from 65% in the absence of HO, to 55% in the presence of HO; Table 3), indicating that Xrs2 is less critical than Mre11 for the repair of that putative subset of HO-mediated DSBs that, left unrepaired, cause the low spore viability observed for spo11 spo13 mre11 strains.

Interestingly, the level of HO-mediated, apparent meiotic crossovers at HO cs5 is also significantly reduced when Xrs2 is absent, from 20.8% in the spo11 spo13 control to 5.5% in spo11 spo13 xrs2 meiotic cells (P = 0.0001; Figure 3, Table 2). Thus, while Xrs2 (and Mre11) are not essential per se for the interhomolog repair of HO-mediated meiotic DSBs, the strong diminishment in apparent meiotic crossover events in spo11 spo13 xrs2 strains may reflect a capacity of Xrs2 to drive HO-initiated interhomolog repair intermediates toward a crossover outcome.

Does HO-mediated, meiotic DSB repair in the absence of Spo11 rely on canonical meiotic recombination proteins that act downstream of strand exchange?

We asked whether the Spo11-independent repair of HO-mediated DSBs utilize meiosis-specific recombination proteins that act downstream of strand exchange. Removal of the SC associated crossover-promoting factors Zip1, Zip2, Zip4, Mer3, Zip3, Msh4-Msh5 or Mlh3 during otherwise wild-type meiosis leads to a ∼50–70% reduction in crossing over in otherwise wild-type meiotic cells (Sym et al. 1993; Ross-Macdonald and Roeder 1994; Hollingsworth et al. 1995; Hunter and Borts 1997; Agarwal and Roeder 2000; Novak et al. 2001; Börner et al. 2004; Tsubouchi and Roeder 2006; Voelkel-Meiman et al. 2016). Of these factors, Zip1, Zip2 and Zip4 are known to be absolutely essential for SC assembly, while at least some assembled SC has been observed in mutants missing Zip3 or Msh4-5 complexes (Agarwal and Roeder 2000; Novak et al. 2001). We found that removal of Zip1 does not reduce the level of HO-mediated interhomolog recombination observed at HO cs5 in spo11 spo13 meiotic cells (21% in zip1 vs. 20.8% in the control), and removal of Zip2 conferred only a mild (∼16%; P = 0.13) reduction in apparent interhomolog crossovers observed at HO cs5 (Figure 4, Table 4). However, removal of Zip3, Msh4, Mlh3, or Mer3 resulted in a significant reduction of HO-mediated, apparent interhomolog crossovers in spo11 spo13 meioses, by approximately 30–43% (P < 0.0001 for zip3, msh4, mlh3; P = 0.0095 for mer3; Figure 4, Table 4), suggesting that these meiosis-specific factors do influence the repair of a subset of HO-mediated recombination events. Taken together, these results suggest that HO meiotic recombination intermediates may be accessible to a subset of SC-associated recombination factors but not those that are absolutely critical for SC assembly.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

HO-mediated recombination in the absence of Spo11 does not rely heavily on canonical meiotic recombination factors. Bar graph shows the frequency of apparent crossovers in spo11 spo13 strains carrying P_SPO13-HO, an HO cs on chromosome IV, and mutant alleles of various recombination factors. Calculations were performed as described in Figure 3 (n > 250; precise values and strain names are reported in Table 4). Left part of graph shows data for strains carrying HO cs5, and the right 3 strains carry HO cs7. Significant deviations, relative to the wild-type value, were determined using Fisher’s Exact test (*P-value ≤ 0.05, **P-value ≤ 0.01, ***P-value ≤ 0.001).

We note that the observed reduction in crossing over in zip3, msh4, mlh3 and mer3 mutants could reflect a selective diminishment of nonreciprocal recombination events that involve extremely long conversion tract lengths, instead of a diminishment in reciprocal crossovers. Under this scenario, the effect of Zip3, Msh4, Mlh3 and Mer3 meiotic crossover-associated proteins on HO-initiated meiotic recombination intermediates would be to promote and/or stabilize longer conversion tracts. To address this possibility, we used PCR analysis to assess the fraction of apparent crossovers resulting from a reciprocal vs. a nonreciprocal event in the zip3 mlh3 double mutant strain, which displays a ∼37% reduction in apparent interhomolog crossovers in our cs5 experimental strain. We found that 20% (4/20) of apparent interhomolog crossovers in zip3 mlh3 dyads (compared to 29% of apparent crossovers in wild-type) remain the result of long-tract interhomolog conversion events, which possibly arise through a BIR mechanism. This result indicates that the ∼37% reduction in apparent interhomolog crossovers caused by the combined absence of Zip3 and Mlh3 does not reflect a selective diminishment of nonreciprocal recombination events.

Factors that process joint molecule recombination intermediates independent of an SC-associated pathway include Mms4, the Slx1/Slx4 complex, and Yen1 (De Los Santos et al. 2003; Fricke and Brill 2003; Ip et al. 2008; Jessop and Lichten 2008). To assess if HO DSBs on chromosome IV rely on any of these so-called “class II” recombination factors, we examined HO-mediated crossing over at HO cs5 in spo11 strains deficient in meiotic Mms4 and Yen1 (P_CLB2-MMS4 and P_CLB2-MMS4 yen1 double mutants). We found that HO-mediated interhomolog recombination is not significantly diminished in the absence of these proteins (Figure 4; Table 4). Finally, we created a strain in which both SC-associated and SC-independent classes of recombination factors are missing. Together, Mlh3, Yen1, Mms4 and Slx1/Slx4 proteins have been implicated in ensuring the resolution of most Spo11- mediated crossovers during wild-type meiosis (Zakharyevich et al. 2012). Simultaneous removal of these four factors did not, however, significantly reduce HO-mediated, apparent crossovers at cs5, nor cs7 in our spo11 spo13 meiotic cells (Figure 4, Table 4).

The Sgs1 helicase has both pro- and anti-crossover activity and its absence allows multi-chromatid exchange events during meiosis (Rockmill et al. 2003; Jessop et al. 2006; De Muyt et al. 2012). Expression of sgs1-C795, a meiotic null allele (Mullen et al. 2000; Rockmill et al. 2003), results in a mild increase in HO-mediated apparent crossover recombination at HO cs5, from 20.8 to 25.9% P = 0.008; Figure 4, Table 4). Thus, Sgs1 may limit HO-mediated interhomolog recombination in spo11 meiotic cells.

These data, however, need to be considered in light of the unexpected result that mlh3 msh4 double mutants and our slx1 mlh3 yen1 P_CLB2-MMS4 quadruple mutant exhibit no significant decrease in HO-mediated crossing over, despite the reduction in crossing over observed in msh4 and mlh3 single mutants. While this observation could reflect bona fide genetic interactions between crossover promoting factors, it may instead indicate that non-specific strain background effects modulate the recombination phenotype.

Altogether, our data suggest that in meiotic cells devoid of Spo11, HO-mediated interhomolog recombination intermediates might engage with a subset, but not the full cohort of SC associated (class I) crossover factors. Furthermore, HO-mediated meiotic DSBs do not absolutely rely on either class I or class II crossover proteins for interhomolog repair.

An HO-mediated DSB located proximal or distal to the centromere is not sufficient to promote stable pairwise associations between homologous centromeres during meiosis

We targeted HO DSBs to budding yeast’s longest chromosome in order ask whether the position of an interhomolog recombination event with respect to the centromere determines homolog pairing outcomes in meiosis. One homolog pairing process is the recombination-dependent transition from homology-independent centromere “coupling” to homologous centromere pairing (Tsubouchi and Roeder 2005; Stewart and Dawson 2008). We used lacO-associated GFP-LacI and tetO-associated TetR-mCherry to assess whether HO-mediated recombination, proximal or distal to CENIV, promotes homologous centromere pairing in spo11 mutant meiosis.

lacO DNA sequences were integrated 705 nucleotides to the right of CENIV, and tetO DNA sequences were introduced at coordinate 1,242 kb on chromosome IV (See Materials and Methods). We visualized lacO and tetO DNA sequences on surface spread meiotic chromosomes by lacO- or tetO-binding proteins GFP-LacI or TetR-mCherry, expressed in trans within the same strains. These cytological tools were built into diploids carrying HO cs2, cs4, or cs5, and into a strain carrying seven active chromosome IV HO cs loci (although this strain is also heterozygous for cs6, which (presumably) is not actively cut; Figure 5A). Diploid strains were sporulated for ∼15 hr before preparing surface-spread nuclei for immunofluorescence; an antibody that targets the meiosis-specific chromosomal protein Hop1 was utilized to verify that nuclei had progressed into meiosis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

A centromere-proximal or distal HO DSB is not sufficient to pair homologous chromosomes. (A) Cartoons show the locations of lacO DNA sequences 705 bp to the right of centromere IV (green) and tetO DNA sequences at 1,242 kb on chromosome IV in strains used for cytology experiments. GFP-LacI and TetR-mCherry, expressed in trans, bind lacO and tetO respectively. Strains homozygous for a single HO cs (top) or carrying seven active HO cs loci (bottom) were utilized in cytological experiments. (B) Images show surface spread meiotic nuclei from wild-type strains (LY42; left column), and from spo11-Y135F strains expressing meiosis-specific HO endonuclease and carrying HO cs5 (LY887; center and right columns) at 15 hr. of sporulation. The presence of Hop1 (not shown) was used to select meiotic nuclei for pairing analysis. The distance between foci corresponding to LacI-GFP bound to lacO sequences near CEN IV (green), and TetR-mCherry bound to tetO sequences on the arm of chromosome IV were considered paired if foci center to foci center < 0.5 μm apart. Bar, 1μm. (C) Bar graphs display the average frequency of CEN IV pairing or chromosome IV arm pairing at 15 hr of sporulation in control (LY176) and chromosome IV HO cs – carrying strains (left to right: LY176, LY173, LY174, LY175, LY331and LY887). 100 meiotic nuclei per genotype were analyzed in triplicate (n = 300 total per genotype). Bars depict standard error of the mean. (D) Homologous and non-homologous centromere pairing between centromeres indicated on the x axis was assessed in spo11 mutant strains (left to right: LY303, LY176, LY356, LY357 and LY358) at 15 hr of sporulation (n > 100).

We found that among meiotic nuclei from control spo11 strains lacking HO endonuclease, 25% exhibited homologous pairing between the centromere regions of chromosome IV (Figure 5B, C). This frequency of association is higher than expected if pairwise centromere associations between 32 individual chromosomes are random and completely independent of homology in spo11 mutants, thus we also examined pairing between the centromere regions of chromosomes III and V, and between centromeres III, V, and IV, in spo11 mutant meiotic cells. Consistent with previous studies that found a preferential association between centromeres belonging to chromosomes of similar size (Lefrançois et al. 2016), centromeres III and V were paired in 22% of surface-spread nuclei (n >100), while centromeres III and IV (which have a greater size differential) displayed a 15% pairing frequency (n > 100; Figure 5D).

Having established the baseline of centromere IV pairing in spo11 meiotic cells, we next asked whether strains with HO-mediated DSBs on chromosome IV display an increase in homologous centromere pairing. We found that spo11 mutants carrying a single HO cs in conjunction with HO endonuclease exhibited a similar frequency of paired centromeres IV as spo11 cells devoid of HO endonuclease, even when the HO cs is adjacent to the CENIV locus. Homologous centromeres IV were paired in 22%, 29% and 24% of meiotic cells at 15 hr of sporulation from strains carrying cs2, cs4, or cs5, respectively (n= 300; Figure 5B, C). Furthermore, homologous centromeres IV were paired in 21% of meiotic cells from strains carrying seven active chromosome IV HO cs loci (cs2, cs4, cs5, cs7, cs8, cs9 and cs10; n= 300; Figure 5B, C).

Similarly, pairing at a chromosome IV arm location, detected in about 8% of spo11 control cells devoid of HO endonuclease, was detected at approximately the same frequency in spo11 cells carrying single or multiple HO cs loci and meiotic HO endonuclease: Chromosome IV arm pairing was observed in 9%, 15%, and 11% of meiotic cells from HO-expressing strains carrying cs2, cs4 and cs5, respectively, and in 10% of the meiotic cells from a strain carrying seven active chromosome IV HO cs loci (Figure 5B, C).

We conclude that a single HO-mediated recombination event, while capable of promoting interhomolog recombination in ∼20% of meioses, is not sufficient to promote stable pairwise associations between homologous centromeres, even when positioned 250 nucleotides from the centromere.

An HO-mediated meiotic DSB fails to promote SC assembly in the context of a null or catalytically inactive spo11 allele, even when meiotic axis proteins are overexpressed

Initial steps in meiotic recombination are required for the assembly of SC in budding yeast, which, at early time points, occurs predominantly from centromere regions (Tsubouchi et al. 2008). An earlier study reported that HO-mediated recombination between homologous chromosomes III was unaccompanied by SC assembly in spo11 meiotic cells (Malkova et al. 2000), but the small size of chromosome III leaves open the possibility that a partial SC assembly event was missed in this prior experiment. We thus asked whether SC structure(s) assemble on the largest yeast chromosome (IV) in response to an HO-mediated DSB, and furthermore whether the distance from the centromere of the HO-mediated recombination event matters.

Surface-spread meiotic nuclei from HO-expressing and control spo11 strains at 15 hr of sporulation were labeled with antibodies against the SC transverse filament protein Zip1 to visualize SC, and with antibodies against the meiosis-specific chromosomal protein Hop1 to verify that nuclei had progressed into meiosis (Figure 6A). In spo11 meiotic nuclei, Zip1 localizes diffusely on surface-spread chromosomes, with several brighter foci likely corresponding to centromeres (Tsubouchi and Roeder 2005). A large fraction of spo11 control nuclei also display an aggregate of Zip1 called a polycomplex. Zip1’s distribution pattern on meiotic chromatin from spo11 strains expressing HO and carrying seven active HO cs loci (including cs2, positioned immediately adjacent to CENIV) appeared indistinguishable from spo11 control meiotic nuclei (Figure 6A, B). The absence of linear Zip1 structures on meiotic chromatin indicates that HO-mediated recombination is incapable of interfacing with and/or successfully activating the molecular pathway(s) that facilitate SC assembly in spo11 meiotic nuclei.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Synaptonemal complex does not assemble in response to an HO-mediated meiotic DSB. (A) Representative surface spread meiotic nuclei from SPO11 (top row; YAM424), or spo11 null strains carrying seven active HO cs loci (cs2, cs4, cs5, cs6, cs7, cs8, cs9, cs10; LY371) at 15 hr of sporulation (upper panel, second and third row). Lower panel displays a representative nucleus from the spo11-Y135F strain carrying HO cs5 and overexpressing Rec8 (LY890; lower panel, top row) and two representative nuclei from the spo11 null strain carrying seven active HO cs and overexpressing Rec8 (LY892, lower panel, bottom two rows) at 24 hr of sporulation in ndt80 strains. Zip1 (green) binds diffusely to and also assembles some bright foci on DAPI-stained meiotic chromatin (blue) from these spo11 strains, regardless of HO-induced meiotic DSBs; polycomplex aggregates of Zip1 (white arrowheads) are often observed. The presence of the meiosis-specific Hop1 protein or Rec8-MYC (red) is displayed in the third column. Bar, 1μm. (B) The proportions of nuclei (n = 50) with different Zip1 and Rec8-MYC distribution phenotypes at 24 hr of sporulation are plotted for a control strain missing P_SPO13-HO (LY893), the spo11 null strain carrying seven active HO cs loci (LY371), a Rec8-overexpression control spo11-Y135F strain with no chromosome IV HO cs (LY891), the spo11-Y135F strain carrying HO cs5 and overexpressing Rec8 (LY890) and the spo11 null strain carrying seven active HO cs loci and overexpressing Rec8 (LY892). Immunoblot in (C) shows Rec8-MYC levels in meiotic cells at 24 hr of sporulation from a control SPO11 ndt80 strain in which REC8 is untagged and which happens to also be homozygous for P_GAL-HOP1 (LY769), a control spo11-Y135F ndt80 strain homozygous for REC8-MYC (LY893), followed by REC8-overexpressing LY890, LY891 and LY892 strains. The same cultures were used to prepare meiotic surface spread nuclei analyzed at the 24 hr time point. Molecular weight indicators are given (kDa) to the left. (D) Graph plots Rec8-MYC protein levels from a strain carrying two endogenous copies of REC8-MYC and strains carrying 2µ-REC8-MYC. Tubulin levels were used to normalize Rec8-MYC levels across samples. The average of 3 replicates is plotted; bars give standard error of the mean.

Spo11 requires 9 additional accessory proteins to initiate recombination during meiosis (Lam and Keeney 2014). A catalytically inactive mutant version of Spo11, Spo11-Y135F, is capable of localizing to chromatin during meiosis (Prieler et al. 2005). With the hope of targeting Spo11 partners to an HO-mediated DSB, we initially created a gene fusion between Spo11-Y135F and the N terminus of HO endonuclease in our HO cs-containing spo11 cells. However, the chimeric protein did not exhibit HO endonuclease activity, as indicated by the absence of meiotic interhomolog recombination at the MAT locus in strains carrying this fusion. We next analyzed strains sustaining HO-mediated, meiotic DSBs with spo11-Y135F expressed in trans. We found no evidence of homologous centromere pairing nor SC assembly in meiotic nuclei carrying meiotic HO-mediated DSBs at cs5, nor in strains carrying seven active cs loci and homozygous for spo11-Y135F (Figures 5, 6A), indicating that the Spo11-Y135F protein cannot confer to HO-mediated DSBs the capacity to promote meiotic chromosome pairing processes.

SC assembly requires the meiosis-specific cohesin, Rec8 (Klein et al. 1999). It was previously reported that artificial DSBs, applied to spo11 mutant meiotic cells that overexpress Rec8, lead to Zip1 linear assemblies which were interpreted to be tripartite SC (Brar et al. 2009). This result led us to wonder whether the overexpression of a meiosis-specific cohesin or another chromosome axis protein would enable HO-mediated DSBs to promote SC assembly. To address this question, we introduced a 2-micrometer plasmid carrying REC8-MYC driven by the HOP1 promoter in order to overexpress REC8-MYC in HO-expressing spo11 meiotic cells carrying no HO cs, a single HO cs, or seven active HO cs loci. A western blot demonstrated that Rec8-MYC is approximately five times more abundant in our REC8-MYC overexpression strains relative to the level produced by two chromosomal copies of REC8-MYC (Figure 6C). However, Zip1’s distribution on surface-spread chromosomes from these meiotic cells appeared dotty-diffuse regardless of whether the strain expressed higher than wild-type levels of Rec8-MYC (Figure 6A, B). Thus, the level of Rec8-MYC overexpression achieved in this experiment fails to bestow HO-mediated DSBs with the capacity to promote SC assembly.

We also asked whether the overexpression of the meiosis-specific chromosome axis proteins Hop1 and Red1 results in HO-mediated SC assembly in spo11 meiotic nuclei. A western blot confirmed that Hop1 is overexpressed at least ninefold relative to control cells in spo11 cells carrying a HOP1-RED1 2-micrometer plasmid (a kind gift of N. Hollingsworth; (Hollingsworth and Ponte 1997) (Figure S1). Immunofluorescence on surface-spread nuclei indicated that, with or without HO-mediated DSBs, spo11 cells carrying the HOP1-RED1 2-micrometer plasmid exhibited a dotty-diffuse Zip1 distribution on chromatin and/or a Zip1 polycomplex, with no linear SC-like structures (Figure S1).

Artificial DSBs supplied en masse by exposure to phleomycin influence SC protein distribution but fail to promote robust SC assembly in spo11 meiotic nuclei

We wondered whether the failure of our meiotic HO-mediated DSBs to support centromere pairing or SC assembly is due to the fact that these processes need a minimal threshold level of DSBs in order to progress. In order to address this question, we examined pairing and SC protein distribution in meiotic nuclei from spo11 mutants that had sustained multiple DSBs due to phleomycin exposure. Whereas surface-spread meiotic nuclei from spo11 mutants rarely exhibit more than one or two Rad51 foci, a significant increase in Rad51 foci (ranging between 5-55; P ≤ 0.001; Figure S2A) was observed upon exposing spo11 meiotic nuclei to an increasing series of phleomycin doses. The dozens of Rad51 foci observed reflect a snapshot of one stage in the repair of phleomycin-dependent DSBs, and are likely an underestimate of DSBs formed in this experiment. We observed a statistically insignificant increase in homologous pairing at centromeres and arm regions of chromosome IV in strains that experienced the highest dose of phleomycin (P = 0.378; Figure S2B). The vast majority of meiotic nuclei exposed to high doses of phleomycin display a “dotty-diffuse” distribution of SC structural proteins Zip1 and Ecm11 (similar to that seen in the negative control), although a small fraction (∼10%) of phleomycin-exposed nuclei display multiple short linear assemblies of coincident Zip1 and Ecm11 (Figure S2C, D). However, nuclei from cells that had been exposed to high doses of phleomycin only rarely display the Zip1 polycomplex structure that is frequently observed in control nuclei. Taken together, these results indicate that exposure of spo11 meiotic cells to phleomycin causes a change in the distribution of Zip1 protein such that Zip1 is less likely to form a polycomplex structure, and may weakly facilitate homologous associations between chromosomes and minimal SC-like assemblies, but does not trigger robust homologous synapsis.