1. Chromosomes and Gene Expression
Download icon

Local chromosome context is a major determinant of crossover pathway biochemistry during budding yeast meiosis

  1. Darpan Medhi
  2. Alastair SH Goldman
  3. Michael Lichten  Is a corresponding author
  1. National Cancer Institute, United States
  2. The University of Sheffield, United Kingdom
Research Article
  • Cited 9
  • Views 1,276
  • Annotations
Cite this article as: eLife 2016;5:e19669 doi: 10.7554/eLife.19669

Abstract

The budding yeast genome contains regions where meiotic recombination initiates more frequently than in others. This pattern parallels enrichment for the meiotic chromosome axis proteins Hop1 and Red1. These proteins are important for Spo11-catalyzed double strand break formation; their contribution to crossover recombination remains undefined. Using the sequence-specific VMA1-derived endonuclease (VDE) to initiate recombination in meiosis, we show that chromosome structure influences the choice of proteins that resolve recombination intermediates to form crossovers. At a Hop1-enriched locus, most VDE-initiated crossovers, like most Spo11-initiated crossovers, required the meiosis-specific MutLγ resolvase. In contrast, at a locus with lower Hop1 occupancy, most VDE-initiated crossovers were MutLγ-independent. In pch2 mutants, the two loci displayed similar Hop1 occupancy levels, and VDE-induced crossovers were similarly MutLγ-dependent. We suggest that meiotic and mitotic recombination pathways coexist within meiotic cells, and that features of meiotic chromosome structure determine whether one or the other predominates in different regions.

https://doi.org/10.7554/eLife.19669.001

eLife digest

Inside the cells of many species, double-stranded DNA is packaged together with specialized proteins to form structures called chromosomes. Breaks that span across both strands of the DNA can cause cell death because if the break is incorrectly repaired, a segment of the DNA may be lost. Cells use a process known as homologous recombination to repair such breaks correctly. This uses an undamaged DNA molecule as a template that can be copied to replace missing segments of the DNA sequence. During the repair of double-strand breaks, connections called crossovers may form. This results in the damaged and undamaged DNA molecules swapping a portion of their sequences.

In meiosis, a type of cell division that produces sperm and eggs, cells deliberately break their chromosomes and then repair them using homologous recombination. The crossovers that form during this process are important for sharing chromosomes between the newly forming cells. It is crucial that the crossovers form at the right time and place along the chromosomes.

Chromosomes have different structures depending on whether a cell is undergoing meiosis or normal (mitotic) cell division. This structure may influence how and where crossovers form. Enzymes called resolvases catalyze the reactions that occur during the last step in homologous recombination to generate crossovers. One particular resolvase acts only during meiosis, whereas others are active in both mitotic and meiotic cells. However, it is not known whether local features of the chromosome structure – such as the proteins packaged in the chromosome alongside the DNA – influence when and where meiotic crossover occurs.

Medhi et al. have now studied how recombination occurs along different regions of the chromosomes in budding yeast cells, which undergo meiosis in a similar way to human cells. The results of the experiments reveal that the mechanism by which crossovers form depends on proteins called axis proteins, one type of which is specifically found in meiotic chromosomes. In regions that had high levels of meiotic axis proteins, crossovers mainly formed using the meiosis-specific resolvase enzyme. In regions that had low levels of meiotic axis proteins, crossovers formed using resolvases that are active in mitotic cells. Further experiments demonstrated that altering the levels of one of the meiotic axis proteins changed which resolvase was used.

Overall, the results presented by Medhi et al. show that differences in chromosome structure, in particular the relative concentration of meiotic axis proteins, influence how crossovers form in yeast. Future studies will investigate whether this is observed in other organisms such as humans, and whether local chromosome structure influences other steps of homologous recombination in meiosis.

https://doi.org/10.7554/eLife.19669.002

Introduction

The transition from the mitotic cell cycle to meiosis involves substantial changes in mechanisms of DNA double strand break (DSB) repair by homologous recombination (HR). Most mitotic HR repairs spontaneous lesions, and most repair products are non-crossovers (NCOs) that do not involve exchange of flanking parental sequences (Kadyk and Hartwell, 1992; Ira et al., 2003; Pâques et al., 1998). In contrast, meiotic recombination is initiated by programmed DSBs (Cao et al., 1990; Sun et al., 1989) that often are repaired as crossovers (COs) between homologous chromosomes (homologs), with exchange of flanking parental sequences. Inter-homolog COs, combined with sister chromatid cohesion, create physical linkages that ensure faithful homolog segregation during the first meiotic division, avoiding chromosome nondisjunction and consequent aneuploidy in gametes (reviewed by Hunter, 2015).

The DSBs that initiate meiotic recombination are formed by Spo11 in complex with a number of accessory proteins, and will be referred to here as Spo11-DSBs (reviewed by Lam and Keeney, 2015). Spo11-DSBs and resulting recombination events are non-uniformly distributed in the genomes of organisms ranging from budding yeast to humans (Baudat and Nicolas, 1997; Blitzblau et al., 2007; Buhler et al., 2007; Fowler et al., 2013; Gerton et al., 2000; Hellsten et al., 2013; Pratto et al., 2014; Singhal et al., 2015; Smagulova et al., 2011; Wijnker et al., 2013). In budding yeast, this non-uniform distribution of Spo11-DSBs is influenced by meiosis-specific proteins, Red1 and Hop1, which are components of the meiotic chromosome axis. The meiotic chromosome axis coordinates sister chromatids and forms the axial element of the synaptonemal complex, which holds homologs in tight juxtaposition (Hollingsworth et al., 1990; Page and Hawley, 2004; Smith and Roeder, 1997). Spo11-DSBs form frequently in large (ca 50–200 kb) 'hot' domains that are also enriched for Red1 and Hop1, and these 'hot' domains are interspersed with similarly-sized 'cold' regions where Spo11-DSBs are infrequent and Red1/Hop1 occupancy levels are low (Baudat and Nicolas, 1997; Blat et al., 2002; Blitzblau et al., 2007; Buhler et al., 2007; Panizza et al., 2011). Normal Spo11-DSB formation requires recruitment of Spo11 and accessory proteins to the meiotic axis (Panizza et al., 2011; Prieler et al., 2005), and Red1/Hop1 are also central to mechanisms that direct Spo11-DSB repair towards use of the homolog as a recombination partner (Carballo et al., 2008; Niu et al., 2005; Schwacha and Kleckner, 1997). Other eukaryotes contain Hop1 analogs that share a domain, called the HORMA domain (Rosenberg and Corbett, 2015), and correlations between these meiotic axis proteins and DSB formation are observed in fission yeast, nematodes and in mammals (Fowler et al., 2013; Goodyer et al., 2008; Wojtasz et al., 2009). Thus, most meiotic interhomolog recombination occurs in the context of a specialized chromosome structure and requires components of that structure.

Meiotic recombination pathways diverge after DSB formation and homolog-directed strand invasion. In budding yeast, about half of meiotic events form NCOs via synthesis-dependent strand annealing, a mechanism that does not involve stable recombination intermediates (Allers and Lichten, 2001a; McMahill et al., 2007) and is suggested to be the predominant HR pathway in mitotic cells (Bzymek et al., 2010; McGill et al., 1989). Most of the remaining events are repaired by a meiosis-specific CO pathway, in which an ensemble of meiotic proteins, called the ZMM proteins, stabilize early recombination intermediates and promote their maturation into double Holliday junction joint molecules (Allers and Lichten, 2001a; Börner et al., 2004; Lynn et al., 2007; Schwacha and Kleckner, 1994). These ZMM-stabilized joint molecules (JMs) are subsequently resolved as COs (Sourirajan and Lichten, 2008) through the action of the MutLγ complex, which contains the Mlh1, Mlh3, and Exo1 proteins (Argueso et al., 2004; Khazanehdari and Borts, 2000; Wang et al., 1999; Zakharyevich et al., 2010, 2012). MutLγ does not appear to make significant contributions to mitotic COs (Ira et al., 2003). A minority of events form ZMM-independent JMs that are resolved as both COs and NCOs by the structure-selective nucleases (SSNs) Mus81-Mms4, Yen1, and Slx1-Slx4, which are responsible for most JM resolution during mitosis (Argueso et al., 2004; Santos et al., 2003; De Muyt et al., 2012; Ho et al., 2010; Muñoz-Galván et al., 2012; Zakharyevich et al., 2012; reviewed by Wyatt and West, 2014). A similar picture, with MutLγ forming most meiotic COs and SSNs playing a minor role, is observed in several other eukaryotes (Berchowitz et al., 2007; Holloway et al., 2008; Plug et al., 1998).

To better understand the factors that promote the unique biochemistry of CO formation during meiosis, in particular MutLγ-dependent JM resolution, we considered two different hypotheses. In the first, expression of meiosis-specific proteins and the presence of high levels of Spo11-DSBs results in nucleus-wide changes in recombination biochemistry, shifting its balance towards MutLγ-dependent resolution of JMs, wherever they might occur. In the second, local features of meiotic chromosome structure, in particular enrichment for meiosis-specific chromosome axis proteins, provides an in cis structural environment that favors MutLγ-dependent JM resolution. However, because Spo11-DSBs form preferentially in Red1/Hop1-enriched regions, and because these proteins are required for efficient Spo11-DSB formation and interhomolog repair, it is difficult to distinguish these two models by examining Spo11-initiated recombination alone.

To test these two hypotheses, we developed a system in which meiotic recombination is initiated by the sequence- and meiosis-specific VMA1 derived endonuclease, VDE (Gimble and Thorner, 1992; Nagai et al., 2003). VDE initiates meiotic recombination at similar levels wherever its recognition sequence (VRS) is inserted (Fukuda et al., 2008; Neale et al., 2002; Nogami et al., 2002). VDE- catalyzed DSBs (hereafter called VDE-DSBs) form independent of Spo11 and meiotic axis proteins. However, like Spo11-DSBs, VDE-DSBs form after pre-meiotic DNA replication and are repaired using end-processing and strand invasion activities that also repair Spo11-DSBs (Fukuda et al., 2003; Neale et al., 2002). We examined resolvase contributions to VDE-initiated CO formation, and obtained evidence that local enrichment for meiotic axis proteins promotes MutLγ-dependent CO formation; while recombination that occurs outside of this specialized environment forms COs by MutLγ-independent mechanisms. We also show that CO formation at a locus, and in particular MutLγ-dependent CO formation, requires Spo11-DSB formation elsewhere in the genome.

Results

Using VDE to study meiotic recombination at ‘hot' and 'cold’ loci

The recombination reporter used for this study contains a VDE recognition sequence (VRS) inserted into a copy of the ARG4 gene on one chromosome, and an uncleavable mutant recognition sequence (VRS103) on the homolog (Figure 1). Restriction site polymorphisms at flanking HindIII sites, combined with the heterozygous VRS site, allow differentiation of parental and recombinant DNA molecules. This recombination reporter was inserted at two loci: HIS4 and URA3, which are 'hot' and 'cold', respectively, for Spo11-initiated recombination and Red1/Hop1 occupancy (Borde et al., 1999; Buhler et al., 2007; Panizza et al., 2011; Wu and Lichten, 1995; also see Figure 4A and Figure 4—figure supplement 1, below). Consistent with previous reports, Spo11- DSBs and the resulting crossovers, are about five times more frequent in inserts at HIS4 than at URA3 (Figure 1—figure supplement 1A). When VDE is expressed, ~90% of VRS sites at both loci were cleaved by 7 hr after initiation of sporulation (Figure 2A), consistent with previous reports that VDE cuts very effectively (Johnson et al., 2007; Neale et al., 2002; Terentyev et al., 2010). Thus, in most cells, both sister chromatids are cut by VDE (Gimble and Thorner, 1992; Neale et al., 2002). In contrast, Spo11-DSBs infrequently occur at the same place on both sister chromatids (Zhang et al., 2011). While the consequences of this difference remain to be determined, we note that inserts at both HIS4 and URA3 are cleaved by VDE with equal frequency (Figure 2A). Thus, any effects due simultaneous sister chromatid-cutting should be equal at the two loci.

Figure 1 with 1 supplement see all
Inserts used to monitor VDE-initiated meiotic recombination.

The HIS4 and URA3 loci are denoted throughout this paper in red and blue, respectively, and are in Red1/Hop1 enriched and depleted regions, respectively (see Figure 4A and Figure 4—figure supplement 1, below). (A) Left—map of VDE-reporter inserts at HIS4, showing digests used to detect recombination intermediates and products. One parent (P1) contains ARG4 sequences with a VDE-recognition site (arg4-VRS), flanked by an nourseothricin-resistance module [natMX, (Goldstein and McCusker, 1999)] and the Kluyveromyces lactis TRP1 gene [KlTRP1, (Stark and Milner, 1989)]; the other parent (P2) contains ARG4 sequences with a mutant, uncuttable VRS site [arg4-VRS103, (Nogami et al., 2002) flanked by URA3 and pBR322 sequences. Digestion with HindIII (H) and VDE (V) allows detection of crossovers (CO1 and CO2) and noncrossovers (NCO); digestion with HindIII alone allows detection of crossovers and DSBs. P2, CO1 and CO2 fragments are drawn only once, as they are the same size in HindIII digests as in HindIII + VDE digests. Right—representative Southern blots. HindIII-alone digests are probed with a fragment (probe 2) that hybridizes to the insert loci and to the native ARG4 locus on chromosome VIII; this latter signal serves as a loading control (LC). Times after induction of meiosis that each sample was taken are indicated below each lane. (B) map of VDE-reporter inserts at URA3 and representative Southern blots; details as in (A). Strain, insert and probe details are given in Materials and methods and Supplementary file 1.

https://doi.org/10.7554/eLife.19669.003
Figure 2 with 1 supplement see all
VDE-initiated recombination occurs at similar levels at the two insert loci.

(A) Cumulative DSB levels are similar at the two insert loci. The fraction of uncut VRS-containing chromosomes (Parent 1) was determined by subtracting the amount of the NCO band in HindIII + VDE digests from the amount of the Parent 1 + NCO band in HindIII digests. (B) Non-cumulative VDE-DSB frequencies, measured as fraction of total lane signal, excluding loading controls, in HindIII digests. (C) Crossover (average of CO1 and CO2) and noncrossover frequencies, measured in HindIII-VDE digests. Solid lines—recombinants from cells expressing VDE; dashed lines—Spo11-initiated crossovers from vde- strains, measured in HindIII-VDE digests and thus corresponding to VDE-resistant products (see also Figure 1—figure supplement 1C). Values are the average of two independent experiments; error bars represent range. Representative Southern blots are shown in Figure 1 and Figure 1—figure supplement 1C.

https://doi.org/10.7554/eLife.19669.005

DSBs appeared and disappeared with similar timing at the two loci (Figure 2B), with measures of insert recovery (Figure 2—figure supplement 1A) and levels of interhomolog recombinants relative to cumulative VDE-DSB levels (Figure 2—figure supplement 1B) indicating that ~70% of VDE DSBs are repaired by interhomolog recombination. The remaining VRS-containing inserts appear to be lost, consistent with high levels of VDE activity preventing recovery of inter-sister recombinants. Thus, the two VDE recombination reporter inserts undergo comparably high levels of meiotic recombination initiation, regardless of the local intrinsic level of Spo11-initiated recombination.

When VDE-DSBs are repaired by interhomolog recombination, VRS sequences are converted to VRS103, and become resistant to digestion by VDE. We therefore used HindIII/VDE double digest to score recombinants that are resistant to VDE cleavage (Figure 1). Comparing the levels of such recombinants in VDE-expressing and vde∆ strains indicates that Spo11-initiated events comprise only a few percent of the recombinants scored in VDE-expressing strains (Figure 2C, Figure 1—figure supplement 1, data not shown). VDE-initiated recombinants formed at high frequencies at both HIS4 and URA3, and NCOs exceeded COs by approximately twofold at HIS4 and threefold at URA3 (Figure 2C). These values are within the range observed in genetic studies of Spo11-induced gene conversion in budding yeast (Fogel et al., 1979), but differ from the average of near-parity between NCOs and COs observed in molecular assays (Lao et al., 2013; Martini et al., 2006). This is consistent with earlier findings, that cutting both sister chromatids at a DSB site is associated with a reduced proportion of COs among repair products (Malkova et al., 2000).

MutLγ makes different contributions to VDE-initiated CO formation at the two insert loci

While VDE-initiated recombination occurred at similar levels in inserts located at HIS4 and at URA3, we observed a marked difference between the two loci, in terms of the resolvase-dependence of CO formation (Figure 3). At the HIS4 locus, COs were reduced in mlh3∆ mutants, which lack MutLγ, by ~60% relative to wild type. In mms4-md yen1∆ slx1∆ mutants, which lack the three structure selective nucleases active during both meiosis and the mitotic cell cycle (SSNs, triple mutants hereafter called ssn mutants), COs were reduced by ~30%, and by ~75% in mlh3 ssn mutants. Thus, like Spo11-initiated COs, VDE-initiated COs in inserts at HIS4 are primarily MutLγ-dependent, and less dependent on SSNs. In contrast, COs in inserts located at URA3 were reduced by only ~ 10% in mlh3, by ~40% in ssn mutants, and by ~60% in mlh3 ssn mutants, so that the final level of residual COs was the same as at HIS4. Thus, SSNs make a substantially greater contribution to VDE-initiated CO formation at URA3 than does MutLγ, and MutLγ’s contribution becomes substantial only in the absence of SSNs.

Figure 3 with 2 supplements see all
Different resolvase-dependence of crossover formation at the two insert loci.

(A) Crossover frequencies (average of CO1 and CO2) measured as in Figure 2C from HIS4 insert-containing mutants lacking MutLγ (mlh3), structure-selective nucleases (mms4-md yen1 slx1) or both resolvase activities (mlh3 mms4-md yen1 slx1). (B) Crossover frequencies in URA3 insert-containing strains, measured as in panel A. Values are the average of two independent experiments; error bars represent range. (C) Final crossover levels (average of 8 and 9 hr values for two independent experiments), expressed as percent of wild type. Note that, in mlh3 mutants, crossovers in HIS4 inserts are reduced by nearly 60%, while crossovers in URA3 inserts are reduced by less than 10%. (D) Final noncrossover levels, calculated as in C, expressed as percent of wild type. Representative Southern blots are in Figure 3—figure supplement 2.

https://doi.org/10.7554/eLife.19669.007

At both insert loci, ssn and mlh3 ssn mutants accumulated DNA species with reduced electrophoretic mobility (Figure 3—figure supplement 2). These slower-migrating species contain branched DNA molecules, as would be expected for unresolved joint molecules (D. M., unpublished observations). Steady state VDE-DSB and final NCO levels were similar in all strains (Figure 3D, Figure 3—figure supplement 1), indicating that resolvases do not act during the initial steps of DSB repair, and consistent with most meiotic NCOs forming by mechanisms that do not involve Holliday junction resolution (Allers and Lichten, 2001a; De Muyt et al., 2012; Sourirajan and Lichten, 2008; Zakharyevich et al., 2012).

Altered Hop1 occupancy in pch2 mutants is associated with altered MutLγ– dependence of VDE-initiated COs

The marked MutLγ-dependence and –independence of VDE-initiated COs in inserts at HIS4 and at URA3, respectively, are paralleled by the levels of occupancy of the meiotic axis proteins Hop1 and Red1 (Panizza et al., 2011; Figure 4A, Figure 4—figure supplement 1A). To test further the suggestion that differential Hop1 occupancy is correlated with differences in CO formation at these loci, we examined the resolvase-dependence of VDE-initiated COs in pch2∆ mutants. Pch2 is a conserved AAA+ ATPase that maintains the nonuniform pattern of Hop1 occupancy along meiotic chromosomes (Börner et al., 2008; Joshi et al., 2009). The different Hop1 occupancies seen in wild type were preserved early in meiosis in pch2∆ mutants (Figure 4A, Figure 4—figure supplement 1A), consistent with previous findings that, in pch2 cells, Spo11-DSB patterns are not altered in most regions of the genome (Vader et al., 2011). By contrast, at later times (4–5 hr after initiation of meiosis), pch2∆ mutants displayed reduced Hop1 occupancy at HIS4, more closely approaching the lower occupancy levels seen throughout meiosis at URA3 (Figure 4A; Figure 4—figure supplement 1A).

Figure 4 with 2 supplements see all
pch2∆ mutants display altered Hop1 occupancy and crossover MutLγ-dependence.

(A) Hop1 occupancy at insert loci, determined by chromatin immunoprecipitation and quantitative PCR. Top—cartoon of insert loci, showing the location of primer pairs used. Bottom—relative Hop1 occupancy, expressed as the average ratio of immunoprecipitate/input extract for both primer pairs (see Materials and methods for details). Values are the average of two independent experiments; error bars represent range. (B) VDE-initiated CO frequencies measured as in Figure 2C at HIS4 (top) and URA3 (bottom) in pch2∆ (solid diamonds), pch2∆ mlh3∆ (open diamonds), and pch2∆ mms4-md yen1 slx1 (half-filled diamonds) mutants. Crossovers from wild type (solid line), mlh3∆ (dotted line) and mms4-md yen1 slx1mutants (dashed line) from Figure 3 are shown for comparison. Values are from two independent experiments; error bars represent range. Representative Southern blots are in Figure 4—figure supplement 2. (C) Extent of CO reduction in mlh3∆ mutants, relative to corresponding MLH3 strains. (D) Extent of CO reduction in mms4-md yen1 slx1 (ssn) mutants, relative to corresponding MMS4 YEN1 SLX1 strains. For both (C) and (D), PCH2 genotype is indicated at the top; values are calculated as in Figure 3C.

https://doi.org/10.7554/eLife.19669.010

The altered Hop1 occupancy in pch2∆ was accompanied by altered resolvase contributions to VDE-initiated COs (Figure 4B,C,D). MutLγ contributions decreased at HIS4 and increased at URA3, and the majority of COs were MutLγ-independent at both insert loci. In contrast, SSN contributions increased slightly at HIS4, and remained unchanged at URA3. Thus, in pch2∆ mutants, the similarity of Hop1 occupancy at later times in meiosis is paralleled by a shift towards more similar contributions of MutLγ to VDE-initiated COs at HIS4 and URA3. Finally, VDE-induced DSB dynamics and NCO levels were similar in PCH2 and pch2∆ strains, except that NCO levels at both loci were reduced in pch2∆ mms4-md yen1∆ slx1∆, suggesting a greater role for SSNs in NCO formation in the absence of Pch2 (Figure 4—figure supplement 1B,C).

Spo11-DSBs promote VDE-initiated, MutLγ-dependent COs

All of the experiments reported above used cells with wild-type levels of Spo11-DSBs. While VDE-DSBs form at similar levels and timing in SPO11 and spo11 mutant cells (Johnson et al., 2007; Neale et al., 2002; Terentyev et al., 2010), features of VDE-DSB repair, including the extent of end resection, are strongly influenced by the presence or absence of Spo11-DSBs (Neale et al., 2002). To determine if other aspects of VDE-initiated recombination are also affected, we examined VDE-initiated recombination in a catalysis-null spo11-Y135F mutant, hereafter called spo11. In spo11 mutants, VDE-DSB dynamics and NCO formation were similar in inserts at HIS4 and URA3, were comparable to those seen in wild type (Figure 5—figure supplement 1), and were independent of HJ resolvase activities (Figure 5—figure supplement 1). In contrast, the absence of Spo11-DSBs substantially reduced VDE-induced COs, resulting in virtually identical CO timing and levels at the two loci (Figure 5A). Unlike the ~60% reduction in COs seen at HIS4 in SPO11 mlh3∆ (Figure 3C), final CO levels were similar in spo11 mlh3Δ and spo11 MLH3 strains, at both HIS4 and URA3, and similar CO reductions were observed at both loci in spo11 ssn mutants (Figure 5B,C). Thus, processes that depend on Spo11-DSBs elsewhere in the genome are important to promote VDE-initiated COs, and appear to be essential for MutLγ-dependent CO formation.

Figure 5 with 2 supplements see all
VDE-initiated COs are reduced and are MutLγ-independent in the absence of Spo11 activity.

(A) VDE-initiated crossover frequencies, measured as in Figure 2C in spo11-Y135F strains (dark solid lines) in inserts at HIS4 (red) and at URA3 (blue). Data from the corresponding SPO11 strains (dotted lines, from Figure 2C) are presented for comparison. (B) COs in HIS4 inserts in spo11 strains that are otherwise wild-type (spo11) or lack either Mutlγ or structure-selective nucleases. (C) As in B, but with inserts at URA3. Values are from two independent experiments; error bars represent range. Representative Southern blots are in Figure 5—figure supplement 2.

https://doi.org/10.7554/eLife.19669.013

Discussion

Local chromosome context influences meiotic CO formation

We examined the contribution of different Holliday junction resolvases to VDE-initiated CO-formation in recombination reporter inserts at two loci, HIS4 and URA3, which are 'hot' and 'cold', respectively, for Spo11-inititiated recombination and for occupancy by the meiotic chromosome axis proteins, Hop1 and Red1. VDE-initiated COs at HIS4 are similar to those initiated by Spo11, in that most depend on MutLγ. In contrast, VDE-initiated COs at the 'cold' locus, URA3, more closely resemble mitotic COs, which are independent of MutLγ, but are substantially dependent on SSNs (Ho et al., 2010; Ira et al., 2003; Muñoz-Galván et al., 2012). Locus-dependent differences in MutLγ-dependence are reduced in pch2∆ mutants, as are differences in Hop1 occupancy at later times in meiosis I prophase. Based on these findings, we suggest that local chromosome context exerts an important influence on the biochemistry of CO formation during meiosis, and that factors responsible for creating DSB-hot and -cold domains also create corresponding domains where different DSB repair pathways are dominant. An attractive hypothesis (Figure 6) is that regions enriched for meiosis-specific axial element proteins create a chromosomal environment that promotes meiotic DSB formation, limits inter-sister recombination, preferentially loads ZMM proteins (Joshi et al., 2009; Serrentino et al., 2013), and is required for recruitment of MutLγ. In such regions, where most Spo11-dependent events occur, recombination intermediates will have a greater likelihood of being captured by axis-associated ZMM proteins, and consequently being resolved as COs by MutLγ. Regions with lower axial element protein enrichment are less likely to recruit ZMM proteins and MutLγ; DSB repair and CO formation in these regions are more likely to involve non-meiotic mechanisms. In short, the meiotic genome can be thought of as containing two types of environment: meiotic axis protein-enriched regions, where 'meiotic' recombination pathways predominate; and meiotic axis protein-depleted regions, in which recombination events more closely resemble those seen in mitotic cells.

Different resolvase functions in different genome domains.

(A) Early crossover decision model for meiotic recombination (Bishop and Zickler, 2004; Hollingsworth and Brill, 2004) illustrating early noncrossover formation, a major pathway where recombination intermediates form in the context of ZMM proteins and are resolved by MutLγ to form crossovers, and a minor pathway where ZMM-independent intermediates are resolved by SSNs as both crossovers and noncrossovers. (B) Division of the meiotic genome into meiotic axis-protein-enriched 'hot' domains (red) that are enriched for Red1 and Hop1, and 'cold' domains where Red1 and Hop1 are depleted. VDE DSBs (yellow stars) can be directed to form efficiently in either domain, but only VDE DSBs that form in 'hot' domains can be recruited to the meiotic axis. (C) DSBs in 'hot' domains can form joint molecules (red star) in the context of ZMM proteins and the synaptonemal complex, and thus can be resolved by MutLγ-dependent activities. DSBs in 'cold' domains form joint molecules (blue star) outside of this structural context, and are resolved by MutLγ-independent activities.

https://doi.org/10.7554/eLife.19669.016

The observation that some COs at HIS4 are SSN-dependent, even though most are MutLγ-dependent (Figure 3), indicates that this division is not absolute. In addition, it is important to keep in mind that ChIP-based values for meiotic axis protein-enrichment and molecular measures of CO resolvase-dependence are both population-based averages, and do not detect cell-to-cell heterogeneity. It is possible that meiotic axis protein enrichment at HIS4 varies across a population, and most SSN-dependent COs form in cells where HIS4 is not meiotic axis protein-enriched. Alternatively, it is possible that meiotic axis protein enrichment at HIS4 is uniform across a population, but that MutLγ is recruited to JMs with less than unit efficiency, and that when MutLγ is not recruited, SSNs resolve JMs. Finally, it is important to recognize that, while meiotic axis protein occupancy is an attractive candidate as a determinant of resolvase contributions to VDE-induced CO formation, other explanations are possible. It is possible that the associations seen at HIS4 and URA3, rather than being directly causative, reflect another underlying aspect of meiotic chromosome structure or function, and that other differences between these two loci cause the observed differences in resolvase usage.

While the current study is the first to directly query the effect of chromosome context on JM resolution, others have obtained results that are consistent with an effect of local chromosome context on meiotic DSB repair. Malkova and coworkers used the HO endonuclease to initiate recombination in meiotic cells at LEU2, also a ‘hot’ locus (Panizza et al., 2011; Wu and Lichten, 1995). The resulting COs were dependent on Msh4, a ZMM protein, to the same degree as are Spo11-induced COs, suggesting that these nuclease-induced COs at the axis enriched LEU2 locus were the products of ZMM/MutLγ-dependent JM resolution (Malkova et al., 2000). Serrentino et al. (2013) showed that enrichment for the budding yeast ZMM protein, Zip3, at DSB sites is correlated with interhomolog CO levels. Specialized chromosome elements also impact meiotic recombination in budding yeast: COs are differentially reduced relative to NCOs near telomeres (Chen et al., 2008); and interhomolog recombination is inhibited near centromeres (Chen et al., 2008; Lambie and Roeder, 1988, 1986; Vincenten et al., 2015). Locus-specific differences in CO/NCO ratios also have been observed in mouse meiosis (de Boer et al., 2015), locus-specific differences in partner choice have been reported in S. pombe (Hyppa and Smith, 2010), and crossover suppression by centromeres is observed in many species (Talbert and Henikoff, 2010).

Consistent with the suggestion that different meiotic recombination uses different mechanisms in different regions, the meiotic genome also appears to contain regions that differ in terms of the response to DNA damage. Treatment of meiotic yeast cells with phleomycin, a DSB-forming agent, triggers Rad53 phosphorylation, as it does in mitotic cells, while Spo11-DSBs do not (Cartagena-Lirola et al., 2008). This suggests that Spo11-DSBs form in an environment that is refractory to Rad53 recruitment and modification, but that there also are environments where exogenously-induced damage can trigger the mitotic DNA damage response. In light of this suggestion, it is interesting that the meiotic defects of spo11 mutants in a variety of organisms are often only partially rescued by DSBs caused by exogenous agents (Bowring et al., 2006; Celerin et al., 2000; Dernburg et al., 1998; Loidl and Mochizuki, 2009; Pauklin et al., 2009; Storlazzi et al., 2003; Thorne and Byers, 1993). While other factors may be responsible for the limited rescue observed, we suggest that it reflects the random location of exogenously-induced DSBs, with only a subset forming in regions where repair is likely to form interhomolog COs that promote proper homolog segregation.

The interplay of resolvase activities is chromosome context-dependent

Although we observe marked differences in the contributions of different resolvases to VDE-induced CO formation at HIS4 and at URA3, there is no absolute demarcation between MutLγ and SSN activities at the two loci. At HIS4, where MutLγ predominates, ssn mutants still display a modest reduction in VDE-initiated COs when MutLγ is active, but an even greater relative reduction in the absence of MutLγ. These findings are consistent with previous studies suggesting that, in the absence of MutLγ, SSNs serve as a back-up that resolves JMs to produce both COs and NCOs (Argueso et al., 2004; De Muyt et al., 2012; Zakharyevich et al., 2012). Our current data indicate that the converse may also be true, since at URA3, MutLγ appears to make a greater contribution to CO formation in the absence of SSNs than in their presence. However, in our studies, JMs are more efficiently resolved in mlh3∆ mutants than in ssn mutants, which display persistent unresolved JMs. Therefore, if MutLγ acts as a back-up resolvase, it can do so in only a limited capacity, possibly reflecting a need for a specific chromosome structural context in which MutLγ can be efficiently loaded and activated. The absence of such a meiosis-specific chromosome context may explain why MutLγ does not appear to contribute to CO formation during the mitotic cell cycle (Ira et al., 2003), although lower MLH3 expression in mitotic cells (Primig et al., 2000) may also reduce its contribution.

Both VDE-induced and Spo11-induced COs form at significant frequencies in mlh3∆ ssn mutants, which lack all four of the HJ resolvase activities thought to function during meiosis (Figure 3; Argueso et al., 2004; Zakharyevich et al., 2012). These residual crossovers may reflect the activity of a yet-unidentified JM resolvase; they may also reflect the production of half-crossovers by break-induced replication (Ho et al., 2010; Kogoma, 1996; Llorente et al., 2008) or by other mechanisms that do not involve dHJ-JM formation and resolution (Ivanov and Haber, 1995; Mazón et al., 2012; Muñoz-Galván et al., 2012). Alternatively, long-tract NCO gene conversion events that include flanking heterologous sequences might be responsible for the products, scored in our molecular assays as COs, that are independent of both MutLγ and SSNs.

Genome-wide Spo11-DSBs promote VDE-initiated COs and are required for chromosome context-dependent differentiation of VDE DSB repair

In catalysis-null spo11-Y135F mutants, most VDE-DSBs are repaired by interhomolog recombination (Figure 5, Figure 5—figure supplement 2), indicating that a single DSB can efficiently search the meiotic nucleus for homology. However, VDE-promoted COs are substantially reduced in spo11 mutants (Figure 5), as has been observed with HO endonuclease-induced meiotic recombination (Malkova et al., 2000). Moreover, in spo11 mutants, virtually all VDE-initiated COs are MutLγ-independent (Figure 5, Figure 5—figure supplement 2), and thus more closely resemble COs that form in mitotic cells. Because patterns of Hop1 occupancy are not markedly altered in spo11 mutants (Franz Klein, personal communication), these findings indicate that, in addition to the local effects of meiotic chromosome structure suggested above, meiotic CO formation is affected by processes that require Spo11-DSBs elsewhere in the genome.

Meiotic DSB repair occurs concurrently with homolog pairing and synapsis (Börner et al., 2004; Padmore et al., 1991), and efficient homolog synapsis requires wild-type DSB levels (Henderson and Keeney, 2004), indicating that multiple interhomolog interactions along a chromosome are needed for stable homolog pairing. To account for the reduced levels and MutLγ-independence of VDE-initiated COs in spo11 mutants, we suggest that a single VDE-DSB is not sufficient to promote stable homolog pairing, and that additional DSBs along a chromosome are needed to promote stable homolog pairing, which in turn is needed to form ZMM protein-containing structures that stabilize JMs and recruit MutLγ. However, the 140–190 Spo11-DSBs that form in each meiotic cell (Buhler et al., 2007) are also expected to induce a nucleus-wide DNA damage-response, and to compete with other DSBs for repair activities whose availability is limited, and both have the potential to alter recombination biochemistry at VDE-DSBs (Johnson et al., 2007; Neale et al., 2002). Thus, while we believe it likely that defects in homolog pairing and synapsis are responsible for the observed impact of spo11 mutation on VDE-initiated CO formation, it remains possible that it is due to changes in DNA damage signaling, repair protein availability, or in other processes that are affected by global Spo11-DSB levels.

Concluding remarks

We have provided evidence that structural features of the chromosome axis, in particular the enrichment for meiosis-specific axis proteins, create a local environment that directs recombination to 'meiotic' biochemical pathways. In the remainder of the genome, biochemical processes more typical of mitotic recombination function. In other words, the transition to meiosis from the mitotic cell cycle does not involve a global inhibition of 'mitotic' recombination pathways. These 'mitotic' mechanisms remain active in the meiotic nucleus, and can act both in recombination events that occur outside of the local 'meiotic' structural context, and in recombination in spo11 mutants. It is well established that local chromosome context influences the first step in meiotic recombination, Spo11-catalyzed DSB formation (Panizza et al., 2011; Prieler et al., 2005). Our work shows that it also influences the last, namely the resolution of recombination intermediates to form COs. It will be of considerable interest to determine if other critical steps in meiotic recombination, such as choice between sister chromatid and homolog as a DSB repair partner, or the choice between NCO and CO outcomes, are also influenced by local aspects of interstitial chromosome structure.

In the current work, we focused on correlations between local enrichment for the meiosis-specific axis protein Hop1 and Holliday junction resolution activity during CO formation. Other HORMA domain proteins, including HIM-3 and HTP-1/2/3 in C. elegans, ASY3 in A. thaliana and HORMAD1/2 in M. musculus, also have been reported to regulate recombination and homolog pairing (Ferdous et al., 2012; Fukuda et al., 2010; Kim et al., 2014; Wojtasz et al., 2009), suggesting that HORMA domain proteins may provide a common basis for the chromosome context-dependent regulation of meiotic recombination pathways in eukaryotes.

Materials and methods

Yeast strains

All yeast strains are of SK1 background (Kane and Roth, 1974), and were constructed by standard genetic crosses or by direct transformation. Genotypes and allele details are given in Supplementary file 1. Recombination reporter inserts with arg4-VRS103 contain a 73nt VRS103 oligonucleotide containing the mutant VDE recognition sequence from the VMA1-103 allele (Fukuda et al., 2007; Nogami et al., 2002) inserted at the EcoRV site in ARG4 coding sequences within a pBR322-based plasmid with URA and ARG4 sequences, inserted at the URA3 and HIS4 loci, as described (Wu and Lichten, 1995). Recombination reporter inserts with the cleavable arg4-VRS (Neale et al., 2002) were derived from similar inserts but with flanking repeat sequences removed, to prevent repair by single strand annealing (Pâques and Haber, 1999). This was done by replacing sequences upstream and downstream of ARG4 with natMX (Goldstein and McCusker, 1999) and K. lactis TRP1 sequences (Stark and Milner, 1989) respectively (see Supplementary file 1 legend for details). The resulting arg4-VRS and arg4-VRS103 inserts share 3.077 kb of homology.

VDE normally exists as an intein in the constitutively-expressed VMA1 gene (Gimble and Thorner, 1993), resulting in low levels of DSB formation in presporulation cultures (data not shown), probably due to small amounts VDE incidentally imported to the nucleus during mitotic growth (Nagai et al., 2003). To further restrict VDE DSB formation, strains were constructed in which VDE expression was copper-inducible. These strains contain the VMA1-103 allele (Nogami et al., 2002), which provides wild type VMA1 function, but lacks the VDE intein and is resistant to cleavage by VDE. To make strains in which VDE expression was copper-inducible, VDE coding sequences on an EcoRI fragment from pY2181 (Nogami et al., 2002); a generous gift from Dr. Satoru Nogami and Dr. Yoshikazu Ohya) were inserted downstream of the CUP1 promoter in plasmid pHG40, which contains the kanMX selectable marker and a ~1 kb CUP1 promoter fragment (Jin et al., 2009), to make pMJ920, which was then integrated at the CUP1 locus.

Sporulation

Yeast strains were grown in buffered liquid presporulation medium and shifted to sporulation medium as described (Goyon and Lichten, 1993), except that sporulation medium contained 10 uM CuSO4 to induce VDE expression. All experiments were performed at 30°C.

DNA extraction and analysis

Genomic DNA was prepared as described (Allers and Lichten, 2000). Recombination products were detected on Southern blots containing genomic DNA digested with HindIII and VDE (PI-SceI, New England Biolabs), using specific buffer for PI-SceI. Samples were heated to 65°C for 15 min to disrupt VDE-DNA complexes before loading; gels contained 0.5% agarose in 45 mM Tris Borate + 1 mM EDTA (1X TBE) and were run at 2 V/cm for 24–25 hr. DSBs were similarly detected on Southern blots, but were digested with HindIII alone as previously described (Goldfarb and Lichten, 2010), and electrophoresis buffer was supplemented with 4 mM MgCl2. Gels were transferred to membranes and hybridized with radioactive probe as described (Allers and Lichten, 2001a, 2001b), and were imaged and quantified using a Fuji FLA-5100 phosphorimager and ImageGauge 4.22 software. HindIII-VDE gel blots were probed with ARG4 sequences from −430 to +63 nt relative to ARG4 coding sequences (Probe 1, Figure 1). To correct for the low level of uncut VDE sites present in all VDE digests (see Figure 1), NCO frequencies measured from these digests were adjusted by subtracting the frequency of apparent NCOs in 0 hr samples. HindIII gel blots were probed with sequences from the DED81 gene (+978 to +1650 nt relative to DED81 coding sequence), which is immediately upstream of ARG4 (Probe 2, Figure 1). Digests of sae2∆ strains (Figure 1—figure supplement 1) were probed with nt 3149–4351 of pBR322.

Chromatin immunoprecipitation and quantitative PCR

Cells were formaldehyde-fixed by adding 840 μl of a 36.5–38% formaldehyde solution (Sigma) to 30 ml of meiotic cultures, incubating for 15 min at room temperature, and quenched by the addition of glycine to 125 mM. Cells were harvested by centrifugation, resuspended in 500 μl lysis buffer (Strahl-Bolsinger et al., 1997) except with 1 mg/ml Bacitracin and complete protease inhibitor cocktail (one tablet/10 ml, Roche 04693116001) as protease inhibitors, and cells were lysed at 4°C via 10 cycles of vortexing on a FastPrep24 (MP Medical) at 4 M/sec for 40 s, with 5 min pauses between runs. Lysates were then sonicated to yield an average DNA size of 300 bp and clarified by centrifugation at 21,130 RCF for 20 min. 1/50th of the sample (10 µl) was removed as input, and 2 μl of anti-Hop1 (a generous gift from Nancy Hollingsworth) was added to the remainder (~490 μl) and incubated with gentle agitation overnight at 4°C. Antibody complexes were purified by addition of 20 μl of 50% slurry of Gammabind G Sepharose beads (GE Healthcare 17088501), with further incubation for 3 hr at 4°C, followed by pelleting at 845 RCF for 30 s. Beads were then processed for DNA extraction (Blitzblau et al., 2012; Viji Subramanian and Andreas Hochwagen, personal communication). Beads were washed with 1 ml lysis Buffer and once each with 1 ml high salt lysis buffer (same as lysis buffer except with 500 mM NaCl), 1 ml ChIP wash buffer (10 mM Tris, 0.25M LiCl, 0.5% NP-40, 0.5% sodium deoxycholate, 1 mM EDTA) and 1 mL 10 mM Tris, 1 mM EDTA; all washes were done for 5 min at room temperature. DNA was then eluted from beads by adding 100 ml 10 mM Tris, 1 mM EDTA, 1% SDS and incubating at 65°C for 15 min. Beads were then pelleted by a short spin at 16,363 RCF and the eluate transferred to a fresh tube. Beads were washed again in 150 ml 10 mM Tris, 1 mM EDTA, 0.67% SDS, mixed and pelleted again. Both eluates were pooled and crosslinks reversed for both immunoprecipitated (IP) and input samples by incubating overnight at 65°C. 250 ml 10 mM Tris 1 mM EDTA, 4 ml 5 mg/ml linear acrylamide (20 mg) and 5 ml 20 mg/ml Proteinase K (100 mg) was added, and samples were incubated at 37°C for 30 min for immunoprecipitates and 2 hr for input samples. 44 µl 5M LiCl was then added to immunoprecipitates, and DNA was precipitated by adding 1 ml ice cold ethanol, incubating at −20°C for 20 min, and centrifugation at 21,130 RCF for 20 min. For input samples, 44 ml 5M LiCl was added, followed by extraction with an equal volume of phenol:chloroform:isoamyl alcohol (25:24:1) and centrifugation at 16,363 RCF for 10 min. The aqueous layer was transferred to a fresh tube and DNA was precipitated from input samples as with immunoprecipitate samples.

qPCR analysis of purified DNA from input and immunoprecipitated samples used primer pairs that amplify two regions: chromosome III coordinates 65350–65547 and 68072–68271, Saccharomyces Genome Database, flanking the HIS4 gene, and chromosome V coordinates 115119–115317 and 117728–117922, flanking the URA3 gene (see Figure 1—figure supplement 1). Chromosome coordinates are from the Saccharomyce cerevisiae reference genome (Engel et al., 2014). Primers and genomic DNA from input and immunoprecipitated samples were mixed with iQ SYBR green supermix (Biorad) and analyzed using a Biorad iCycler.

Source data

Numerical values underlying all graphs are contained in Supplementary file 2.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
    The Mus81/Mms4 endonuclease acts independently of double-Holliday junction resolution to promote a distinct subset of crossovers during meiosis in budding yeast
    1. T de los Santos
    2. N Hunter
    3. C Lee
    4. B Larkin
    5. J Loidl
    6. NM Hollingsworth
    (2003)
    Genetics 164:81–94.
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
    Purification and characterization of VDE, a site-specific endonuclease from the yeast Saccharomyces cerevisiae
    1. FS Gimble
    2. J Thorner
    (1993)
    The Journal of Biological Chemistry 268:21844–21853.
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
    Sister chromatids are preferred over homologs as substrates for recombinational repair in Saccharomyces cerevisiae
    1. LC Kadyk
    2. LH Hartwell
    (1992)
    Genetics 132:387–402.
  55. 55
    Carbohydrate metabolism during ascospore development in yeast
    1. SM Kane
    2. R Roth
    (1974)
    Journal of Bacteriology 118:8–14.
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
    Repression of meiotic crossing over by a centromere (CEN3) in Saccharomyces cerevisiae
    1. EJ Lambie
    2. GS Roeder
    (1986)
    Genetics 114:769–789.
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
    Changes in protein composition of meiotic nodules during mammalian meiosis
    1. AW Plug
    2. AH Peters
    3. KS Keegan
    4. MF Hoekstra
    5. P de Boer
    6. T Ashley
    (1998)
    Journal of Cell Science 111 (Pt 4 ):413–423.
  81. 81
  82. 82
  83. 83
  84. 84
    Multiple pathways of recombination induced by double-strand breaks in Saccharomyces cerevisiae
    1. F Pâques
    2. JE Haber
    (1999)
    Microbiology and Molecular Biology Reviews 63:349–404.
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
  90. 90
  91. 91
  92. 92
  93. 93
  94. 94
  95. 95
  96. 96
  97. 97
  98. 98
  99. 99
  100. 100
    Stage-specific effects of X-irradiation on yeast meiosis
    1. LW Thorne
    2. B Byers
    (1993)
    Genetics 134:29–42.
  101. 101
  102. 102
  103. 103
  104. 104
  105. 105
  106. 106
    Factors that affect the location and frequency of meiosis-induced double-strand breaks in Saccharomyces cerevisiae
    1. TC Wu
    2. M Lichten
    (1995)
    Genetics 140:55–66.
  107. 107
  108. 108
  109. 109
  110. 110

Decision letter

  1. Bernard de Massy
    Reviewing Editor; Institute of Human Genetics, CNRS UPR 1142, France

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Local chromosome context is a major determinant of crossover pathway biochemistry during budding yeast meiosis" for consideration by eLife. Your article has been favorably evaluated by Kevin Struhl (Senior Editor) and three reviewers, one of whom, Bernard de Massy (Reviewer #1), is a member of our Board of Reviewing Editors. The following individual involved in review of your submission has agreed to reveal their identity: Joao Matos (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

In this study Medhi et al. have analyzed the contribution of different Holliday junction resolvases to VDE-initiated CO-formation in recombination reporters inserted at HIS4 and URA3 loci. One known and documented difference between these two loci is the occupancy for the chromosome axis protein Hop1, high at HIS4 and low at URA3. The authors find that VDE-initiated COs at HIS4 are strongly dependent on MutLγ, similar to those initiated by Spo11. VDE-induced COs at URA3 are more dependent on SSNs. The authors also show that pch2∆ mutants, which have smaller differences in Hop1 occupancy between HIS4 and URA3, display smaller locus-dependent differences in formation of COs by MutLγ. The authors conclude that local chromosome context is important for the biochemistry of CO formation during meiosis.

The data is of high quality and the striking differences between insertions at the HIS4 and URA3 loci are clearly documented. These results provide significant, interesting and novel insights into the regulation of meiotic recombination by properties of chromosome structure.

Essential revisions:

In order to fully validate the interpretations, some additional experiments are needed using the tools that the authors have used in the present experiments. In addition, the authors should not over interpret the data which is based on the comparison of two loci (HIS4 and URA3) with a correlation with HOP1 occupancy. There are a large number of possible differences between these loci aside from the ones the authors focus on, and it remains possible that the direction of the correlation seen with these two loci remains purely coincidental. Pch2 mutant fits the expectation but absence of Pch2 has likely several other consequences. The Abstract and conclusions should be modified accordingly. Use of SEM should me revised.

1) In order to evaluate the generality of their findings (see general comment above), the authors could compare published ChIP-seq data of Hop1 with available genome wide recombination maps from resolvase mutants.

2) One important experiment missing in this paper is to demonstrate the requirement for ssn in pch2 mutant and thus to analyze intermediates, COs and NCOs in pch2 mms4 yen1 slx1 mutant.

3) Since there are only two replicate datasets for several analyses, error bars should show range rather than SEM for the time-course plots. Bar graphs should be replaced with univariate scatter plots, but if the authors wish to retain the bar graphs, then error bars should show range, not SEM. This paper in PLoS Biology provides an excellent discussion of pitfalls for bar graphs and suggests other strategies for data display: http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1002128

4) The authors should explicitly indicate that in most meiosis VDE cuts both sister chromatids and that the consequences of this on pathway choices are unknown.

5) In the subsection “Local chromosome structure influences meiotic CO formation”, second paragraph: There is little or no mention of prior studies of context-dependence for crossover-noncrossover likelihood (Mancera et al. 2008; Serrentino et al. 2013; deBoer et al. 2015) or interhomolog vs. intersister partner choice (Hyppa & Smith 2010; Fowler et al. 2014). The Serrentino paper is mentioned briefly in passing in the preceding paragraph, but I doubt a reader would realize from this mention that paper had documented differences in crossover vs. noncrossover outcome between different loci. There are also studies documenting the different recombination behavior for DSBs within pericentromeric regions (Chen et al. 2008; Vincenten et al. 2015). These prior studies should be discussed in comparison to the context dependence documented here. (This is also relevant to the statement in the subsection “Concluding remarks”, at the end of the first paragraph.)

Reviewer #1:

Taking advantage of DSBs induced by the endonuclease VDE in S. cerevisiae together with genetic and molecular analysis of recombination intermediates during meiosis, Medhi et al. show distinct requirements for resolution of recombination intermediates in correlation with chromosomal context, i.e. the enrichment for the axis protein Hop1: in a Hop1-enriched region, MutLg (Mlh1, Mlh3, Exo1) is the main resolvase involved, in a Hop1-poor region, resolution is mostly dependent on structure specific nucleases (Mus81-Mms4, Yen1, Slx4-Slx1). This pattern is altered in pch2 mutants where Hop1 appears to be more homogenously distributed.

The experimental strategy, experiments and interpretations are very clear and convincing. These results provide significant, interesting and novel insights into the regulation of meiotic recombination by properties of chromosome structure.

Major comment:

The only experimental validation missing in this paper is to demonstrate the requirement for ssn in pch2 mutant and thus to analyze intermediates, COs and NCOs in pch2 mms4 yen1 slx1 mutant.

General comments:

1) Several observations and interpretations should be more precisely discussed:

How these observations relate to previous genetic studies on the effect of MutLg and ssn on CO in different intervals?

2) How do the authors explain that the ratio NCO/CO (for VDE induced events) is similar is both contexts (His4 vs. Ura3)? According to the model proposed, higher density of meiosis axial elements should lead to increase capture of ZMM and decreased NCO/CO.

3) Why JMs are detected in ssn mutant and not or weakly in Mlh3 mutant? In addition, if intermediates are distinct (nicked single/double HJ?) how could resolvases partially substitute, how could Mlh be used as a backup? The discussion in the first paragraph of the subsection “The interplay of resolvase activities is chromosome context-dependent”, does not clarify these questions.

In the absence of Spo11: CO are Mlh-independent, how is Hop1 or Hop1 phosphorylation distributed?

Reviewer #2:

Much information has been accumulated concerning the genetic control of meiotic recombination pathways that lead to formation of crossover or noncrossover recombination products. However, much of this work tends to focus on one or two individual genomic loci, so it has remained unclear to what extent observed genetic pathways operate uniformly (or not) across the genome. This intriguing paper documents a substantial difference between two positions in the yeast genome for use of MutLgamma and other Holliday junction resolving activities. The experimental design is clever, taking advantage of site-specific DSBs made by the VDE nuclease. The data are of excellent quality and the striking differences between insertions at HIS4 and URA3 are clearly documented. This is an interesting and important study.

There were a few concerns about the framing of some of the interpretations and about how some of the data are displayed. While these are important points, they can be easily remedied with text fixes.

1) A significant concern has to do with interpreting differences between the HIS4 and URA3 loci in terms of chromosome structure and/or abundance of Hop1 or other chromosome axis proteins. The main issue is that the authors are trying to argue based on a two-point correlation, but there are a large number of possible differences between these loci aside from the ones the authors focus on, and it remains possible that the direction of the correlation seen with these two loci remains purely coincidental. The pch2 experiments are consistent with the authors' preferred interpretation, but fall short of providing convincing support: while it is true that Hop1 occupancy changes, it is also the case that other things probably change as well in the absence of Pch2. So, it is a stretch to ascribe the effects of this mutation as being solely or even primarily via effects on Hop1 localization. What this paper clearly demonstrates is context dependence for recombination protein utilization. The title of the paper is thus spot-on. But the Discussion needs to be more cautious in interpreting the possible causes of the context dependence, because there is just not enough information here to make strong conclusions.

2) In the subsection “Local chromosome structure influences meiotic CO formation”, second paragraph: There is little or no mention of prior studies of context-dependence for crossover-noncrossover likelihood (Mancera et al. 2008; Serrentino et al. 2013; deBoer et al. 2015) or interhomolog vs. intersister partner choice (Hyppa & Smith 2010; Fowler et al. 2014). The Serrentino paper is mentioned briefly in passing in the preceding paragraph, but I doubt a reader would realize from this mention that paper had documented differences in crossover vs. noncrossover outcome between different loci. There are also studies documenting the different recombination behavior for DSBs within pericentromeric regions (Chen et al. 2008; Vincenten et al. 2015). These prior studies should be discussed in comparison to the context dependence documented here. (This is also relevant to the statement in the subsection “Concluding remarks”, at the end of the first paragraph.)

3) Throughout: SEM is used for error bars, but in most cases this is for experiments with only two replicates. SEM is rarely a good choice for displaying information about experimental variability, but is particularly fraught (as is SD) with just two measurements. It would be better to use error bars to indicate range when there are only two measurements, and SD (or range) for instances where there are larger numbers of replicates. A related point: in Figure 2—figure supplement 1A, it's a little unclear if the error bars are SEM here, but if so, the data seem to indicate >100% recovery of arg4-VRS103 for at least one of the experiments at URA3. Is this right? In any case, it would be better to use univariate scatterplots to show the individual values from each replicate rather than using bar graphs with SEM error bars (or more complicated error bars as in Figure 3. If the authors wish to stick to bar graphs, then the error bars should show the range. This paper in PLoS Biology provides an excellent discussion of pitfalls for bar graphs and suggests other strategies for data display: http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1002128

Reviewer #3:

In this study Medhi et al. have analyzed the contribution of different Holliday junction resolvases to VDE-initiated CO-formation in recombination reporters inserted at HIS4 and URA3. HIS4 is "hot" whereas URA3 is "cold" for occupancy by the chromosome axis proteins Hop1 and Red1.

The authors find that VDE-initiated COs at HIS4 are strongly dependent on MutLγ, similar to those initiated by Spo11. VDE-induced COs at URA3 are more dependent on SSNs. The authors also show that pch2∆ mutants, which have smaller differences in Hop1 occupancy between HIS4 and URA3, display smaller locus-dependent differences in formation of COs by MutLγ. The authors conclude that local chromosome context is important for the biochemistry of CO formation during meiosis.

The data is of high quality and the work is overall interesting. If the authors address the issues listed below, I would strongly recommend it for publication in eLife.

1) The authors infer general chromosome behavior from 2 loci. It would be important to expand the analysis to at least another pair of "hot" and "cold" loci. Alternatively, the authors could compare published ChIP-seq data of Hop1 with available genome wide recombination maps from resolvase mutants.

2) Since VDE cuts both sister chromatids, could the authors reduce VDE expression to obtain a more comparable type of cleavage to Spo11. Otherwise, it may be inappropriate to extrapolate the obtained results to those of naturally occurring DSBs.

3) In Figure 3, the authors show that CO formation at HIS4 is reduced by 60% in mlh3 mutants, 30% in ssn mutants and 75% in a combination of the two. How are COs made in such mutant and why isn't the reduction at least the sum of both mlh3 and ssn mutants (expected ~90% reduction)? The exact opposite is observed at URA3: The mlh3 sse mutant is more defective than the sum of the two pathways.

4) In the subsection “Altered Hop1 occupancy in pch2 mutants is associated with altered MutLγ− dependence of VDE-initiated Cos”, last paragraph: Why would COs that are Mlh3-dependent increase at URA3 relative to HIS4 (37% vs. 20%) in pch2∆ mutants. If the Hop1 occupancy is similar, as argued by the authors, why is this value so different?

5) The finding that MLH3-dependent COs are absent in spo11 mutants is not surprising. However, it is surprising that the formed COs are also largely SSN-independent (~70%). How do the authors explain this? The authors should analyze the mlh3∆ ssn∆ mutant. It is possible that SSNs can backup completely for MLH3 in this case (and MLH3 provides a significant backup for SSNs). Which would suggest that actually Mlh1-Mlh3 does at least part of the job in the WT, but one does not see it because SSNs can compensate for it.

6) How do the authors explain that the dependency on pathway usage at "hot" and "cold" regions is so limited. 60% vs. 30% contribution is not what one would expect if the genome is really partitioned and this is a major biochemical determinant of pathway usage. The authors should discuss this in more detail.

7) CO and NCO frequency is the same at URA3 and HIS4 (Figure 2). If at one locus HJ cleavage is mediated by MutLγ and at the other by SSNs, wouldn't one expect a different NCO/CO ratio? Nucleases have been shown by the Lichten lab (Dayani 2011) to generate a mixture of both, while MutLγ would generate exclusively COs (Zakharyevich 2012).

https://doi.org/10.7554/eLife.19669.019

Author response

[…]

Essential revisions:

In order to fully validate the interpretations, some additional experiments are needed using the tools that the authors have used in the present experiments. In addition, the authors should not over interpret the data which is based on the comparison of two loci (HIS4 and URA3) with a correlation with HOP1 occupancy. There are a large number of possible differences between these loci aside from the ones the authors focus on, and it remains possible that the direction of the correlation seen with these two loci remains purely coincidental. Pch2 mutant fits the expectation but absence of Pch2 has likely several other consequences. The Abstract and conclusions should be modified accordingly. Use of SEM should me revised.

Please see below for additional work and manuscript changes that address these concerns. We have rewritten the Abstract to modify its overall emphasis, and have also added text to the Discussion that explicitly addresses the above concern about coincidence:

“The observation that some COs at HIS4 are SSN-dependent, even though most are MutLγ-dependent (Figure 3), indicates that this division is not absolute. […] It remains possible that the association seen at HIS4 and URA3, rather than being directly causative, reflects another underlying aspect of meiotic chromosome structure or function, and that other differences betweenthese two loci cause the observed differences in resolvase usage.”

It should be noted that an emerging consensus is that Pch2’s primary activity involves remodeling HORA-domain proteins (see Rosenberg and Corbett, JCB 2015 for discussion). This makes it likely that the varied meiotic phenotypes of pch2 mutants are all a consequence of altered Hop1 distributions, but of course altering Hop1 occupancy will affect many different meiotic processes; thus, the text above, in particular the last two sentences.

1) In order to evaluate the generality of their findings (see general comment above), the authors could compare published ChIP-seq data of Hop1 with available genome wide recombination maps from resolvase mutants.

This is a great idea. Unfortunately, currently there are not sufficient tetrad data to make a locus-by-locus comparison. The only available SSN mutant data are for a mms4 meiotic depletion strain (Oke et al., PLOS Genetics 2014), with a total of 596 crossovers in 7 tetrads analyzed, or a scored crossover density of about 1/17kb. Nishant K.T. and collaborators have (unpublished) crossover data for 19 mlh3 tetrads from an S288c-YJM789 hybrid strain that they have made available to us, but even at this higher crossover number (1224, 1CO/8 kb), there are not enough to confidently score differences between wild-type and mlh3 on a locus-by-locus basis. We are currently exploring strategies to divide the genome into bins with different Hop1 enrichment levels and examine relative Mlh3-dependence of crossovers in each bin, but this is a complex problem that is going to require considerable work before we even know if the current mlh3 dataset is of sufficient size.

2) One important experiment missing in this paper is to demonstrate the requirement for ssn in pch2 mutant and thus to analyze intermediates, COs and NCOs in pch2 mms4 yen1 slx1 mutant.

We did these experiments and they are presented in Figure 4 and Figure 4—figure supplements 1 and 2; corresponding text has also been changed (subsection “Altered Hop1 occupancy in pch2 mutants is associated with altered MutLγ−dependence of VDE-initiated Cos”, last paragraph).

3) Since there are only two replicate datasets for several analyses, error bars should show range rather than SEM for the time-course plots. Bar graphs should be replaced with univariate scatter plots, but if the authors wish to retain the bar graphs, then error bars should show range, not SEM. This paper in PLoS Biology provides an excellent discussion of pitfalls for bar graphs and suggests other strategies for data display: http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1002128

Error bars were changed to range in all but Figure 3C, D and Figure 4C, D, where error bars were removed (see below). Figure legends were appropriately changed.

We did not convert bar graphs to scatter plots, as they do not report primary data. Primary data are presented as line graphs and also in Supplementary file 2. The bar graphs are used to summarize features of the data, with the goal of visually communicating conclusions.

Weissgerber et al., cited in the reviewers’ comments, object to bar (and line) graphs because they do not “allow readers to critically evaluate continuous data”. In our paper, data are clearly presented in other figure panels and in a data supplement, so the interested reader has plenty of opportunity for evaluation. Weissgerber et al. dislike bar graphs because they conceal differences in distributions (including outliers), in sample size, and in relationships between dependent variables. In our data, there are no differences in sample size (2, in each case), in distribution relative to the mean (can’t be, with 2 values), and variables are independent. Therefore, the issues that motivate Weissgerber et al. are not relevant to our paper.

In the case of Figures 3 and 4, panels C and D, it is not possible to convert error bars to “range”. This is because the values are the mean of 8 and 9 hr samples in two independent experiments with the indicated mutant (all 4 values averaged), divided by a similar mean for the indicated wild-type strain. Since the values are the ratio of two means, range is not applicable. In addition, values are a mix of dependent (8 and 9 hr samples from the same time course) and independent values (everything else), so formally it is not legitimate to calculate standard deviations. Instead, we removed error bars and representations of significance from these panels, and figure legends have been appropriately adjusted. We believe that these bar graphs still have value (see below), in that they enhance comprehensibility. We would prefer to retain them, but will remove them if requested. If error bars are deemed necessary, then we can calculate standard deviations for these ratios (we agree that S.E.M. was not correct), keeping in mind that such a calculation is not strictly legitimate.

(The following contains Michael Lichten’s views, not necessarily those of the other authors, and is included here by way of discussion with the editor, editorial staff, and reviewers. Please feel free to delete it from the public review record or include it, as you see fit.)

Despite current trends to the contrary, there is a definite value to summary plots and statistics, if they enhance clarity and comprehensibility but are not the only form in which data are presented, and if they are not used to hide data features that are relevant to the analysis. Univariate scatter plots (a.k.a. “confetti plots”), currently so popular, actually can make data less comprehensible and more obscure, especially when sample sizes are so large that individual points cannot be distinguished. It is hoped that the suggestions of Weissgerber et al., which are certainly well taken, will not be blindly imposed on every paper that is submitted to eLife, but rather will be used in situations where they are appropriate.

4) The authors should explicitly indicate that in most meiosis VDE cuts both sister chromatids and that the consequences of this on pathway choices are unknown.

The following text was added:

“Thus, in most cells, both sister chromatids are cut by VDE (Gimble and Thorner, 1992; Neale et al., 2002). In contrast, Spo11-DSBs infrequently occur at the same place on both sister chromatids (Zhang et al., 2011). While the consequences of this difference remain to be determined, we note that inserts at both HIS4 and URA3 are cleaved by VDE with equal frequency (Figure 2A). Thus, any effects due simultaneous sister chromatid-cutting should be equal at the two loci.”

5) In the subsection “Local chromosome structure influences meiotic CO formation”, second paragraph: There is little or no mention of prior studies of context-dependence for crossover-noncrossover likelihood (Mancera et al. 2008; Serrentino et al. 2013; deBoer et al. 2015) or interhomolog vs. intersister partner choice (Hyppa & Smith 2010; Fowler et al. 2014). The Serrentino paper is mentioned briefly in passing in the preceding paragraph, but I doubt a reader would realize from this mention that paper had documented differences in crossover vs. noncrossover outcome between different loci. There are also studies documenting the different recombination behavior for DSBs within pericentromeric regions (Chen et al. 2008; Vincenten et al. 2015). These prior studies should be discussed in comparison to the context dependence documented here. (This is also relevant to the statement in the subsection “Concluding remarks”, at the end of the first paragraph.)

We agree that previous findings were given short shrift, and included more of this information in the paper. However:

1) Mancera et al. say that CO/NCO ratios differ at different loci, but this is likely a consequence of small sample size and uneven distribution of polymorphic markers.

2) Serentino et al. showed that three DSB sites with lower Zip3 occupancy/DSB ratios had fewer COs (measured by genetic distance, cM) than a DSB site with higher Zip3 occupancy. Please note that, because NCOs were not scored, this could be because of changes in CO/NCO or in IH/IS ratios. Their calculations used ssDNA data from Buhler et al. as a proxy for DSB levels; when the calculation is made using Pan et al.’s more recent Spo11-oligo data, the differences between loci become much less marked, and Zip3 occupancy correlates fairly well with Spo11 oligo levels (ML, unpublished).

Rather than include an extensive discussion of these issues, we wrote the following:

“Serrentino et al. (2013) showed that enrichment for the budding yeast ZMM protein, Zip3, at DSB sites is correlated with interhomolog CO levels. […] Locus-specific differences in CO/NCO ratios also have been observed in mouse meiosis (de Boer et al., 2015), locus-specific differences in partner choice have been reported in S. pombe (Hyppa and G. R. Smith, 2010), and crossover suppression by centromeres is observed in many species (Talbert and Henikoff, 2010).”

https://doi.org/10.7554/eLife.19669.020

Article and author information

Author details

  1. Darpan Medhi

    1. Laboratory of Biochemistry and Molecular Biology, Center for Cancer Research, National Cancer Institute, Bethesda, United States
    2. Sheffield Institute for Nucleic Acids, The University of Sheffield, Sheffield, United Kingdom
    3. Department of Molecular Biology and Biotechnology, The University of Sheffield, Sheffield, United Kingdom
    Contribution
    DM, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  2. Alastair SH Goldman

    1. Sheffield Institute for Nucleic Acids, The University of Sheffield, Sheffield, United Kingdom
    2. Department of Molecular Biology and Biotechnology, The University of Sheffield, Sheffield, United Kingdom
    Contribution
    ASHG, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  3. Michael Lichten

    Laboratory of Biochemistry and Molecular Biology, Center for Cancer Research, National Cancer Institute, Bethesda, United States
    Contribution
    ML, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    mlichten@helix.nih.gov
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9707-2956

Funding

National Cancer Institute (Intramural Research Program of the NIH through the Center for Cancer Research at the National Cancer Institute)

  • Michael Lichten

University of Sheffield (Graduate tuition grant)

  • Darpan Medhi

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Robert Shroff, Anuradha Sourirajan, Satoru Nogami, Yoshikazu Ohya, and Nancy Hollingsworth for generous gifts of strains and reagents, Viji Subramanian and Andreas Hochwagen for technical advice, and Dhruba Chattoraj, Julie Cooper, and Alex Kelly for comments on the manuscript. This work was supported by The University of Sheffield and the Intramural Research Program of the NIH through the Center for Cancer Research, National Cancer Institute.

Reviewing Editor

  1. Bernard de Massy, Institute of Human Genetics, CNRS UPR 1142, France

Publication history

  1. Received: July 14, 2016
  2. Accepted: November 17, 2016
  3. Accepted Manuscript published: November 18, 2016 (version 1)
  4. Version of Record published: January 9, 2017 (version 2)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 1,276
    Page views
  • 457
    Downloads
  • 9
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Chromosomes and Gene Expression
    Robert F Levendosky, Gregory D Bowman
    Research Advance
    1. Chromosomes and Gene Expression
    Cori K Cahoon, Diana E Libuda
    Insight