Mutual, spatially limited control of meiotic DNA break formation by Mre11–Rad50–Nbs1 DNA repair complex and Tel1 (ATM) protein kinase

Abstract

Meiotic recombination is initiated by DNA double-strand breaks (DSBs); factors that control DSB frequencies are important to produce viable progeny. In many organisms, the ATM (Tel1) protein kinase prevents excessive meiotic DSBs, especially nearby DSBs on the same chromatid. Normally, two close DSBs are less frequent than expected from independence, a feature called DSB interference, which is lost in tel1Δ mutants. In the fission yeast Schizosaccharomyces pombe, high-level DSB formation depends on linear elements, Hop1, and meiotic cohesin complexes; we show here that these complexes impart competition between nearby DSB sites. When these complexes are impaired, Tel1 substantially represses DSB formation, and in its absence, two close DSBs on the same chromatid occur frequently and manifest high negative interference. After mitotic DNA damage, the conserved Mre11–Rad50–Nbs1 (MRN) complex is required for DNA resection, and the Tel1 kinase activity is needed to complete DSB repair. We found that during meiosis mre11Δ and rad50Δ mutants, like tel1Δ mutants, lack DSB interference and display highly negative DSB interference in meiotic complex mutants. Thus, MRN at a DSB site appears critical for Tel1 function in meiosis and reveals a complex interplay of positive and negative factors controlling meiotic DSB formation.

Graphical Abstract

Graphical Abstract.

Introduction

Meiosis produces haploid gametes from diploid progenitor cells by following DNA replication with two successive cell divisions. Preceding the first cell division is meiotic recombination, in which programmed DNA double-strand breaks (DSBs) are created as an initiation point for the engagement of homologous chromosomes necessary for an effective meiosis. The repair of the DSBs has two beneficial consequences: reciprocal genetic rearrangements, called crossovers, create genetically diverse progeny, and the tension created by crossovers aids the proper segregation of chromosomes [1]. The mechanisms that regulate DSB formation have been studied in many organisms but are not fully understood [2]. Here, we examine this regulation in a simple eukaryote, the fission yeast Schizosaccharomyces pombe.

The protein complexes involved in breaking meiotic DNA are highly conserved and well characterized [3], the most prominent being Spo11 (called Rec12 in S. pombe), which contains the active site for DSB formation [4–9]. After replication, meiotic cohesin complexes, containing the meiosis-specific subunits Rec8 and Rec11 in S. pombe, are loaded onto sister chromatids and hold them together [10–16]. Cells lacking cohesin have reduced but substantial DSBs, reduced crossovers, and aberrant segregation, leading to poor viability [11, 13, 17–20]. In S. pombe, an important early event is the phosphorylation of Rec11 by the casein kinases Hhp1–Hhp2, which is essential for the interaction of Rec11 with the linear element (LinE) component Rec10; DSBs formed by Rec12 require this interaction [21, 22]. Rec10 also binds to and colocalizes with the meiotic HORMAD protein Hop1 [23, 24], which is required for high-level recombination and DSB formation in most organisms studied [25]. Artificial targeting of Hop1 promotes DSB formation in normally “cold” DSB regions of S. cerevisiae [26].

The LinEs, composed of Rec25, Rec27, Mug20, and Rec10, localize to stretches of the meiotic chromosomes but do not run the entire length [23, 27, 28, 29]. They are functionally related to the synaptonemal complex (SC) proteins of many eukaryotes. The SC is a tripartite structure in which a set of proteins comprising the axial element (AE) localizes along the chromosome axis of each pair of sisters (homologs). The AEs are connected by the central elements (CE), creating a connection between the two homologs [30]. The components of the SC are well conserved, though the timing of SC formation varies among species. S. pombe does not have a complete SC; rather, the LinEs are functionally equivalent to AEs [23], and they share similarity in protein structure with the AE coiled-coil proteins of other species [31, 32]. LinE formation is cohesin-dependent in S. pombe, like chromosome axis formation in many species examined [10, 11, 12, 23, 33–36]. Cohesin and LinE deletion mutants in S. pombe have similar meiotic phenotypes: reduced DSBs and a loss of crossovers in a region-dependent manner, the exception being a rec10 null mutant, which is nearly as DSB- and recombination-deficient genome-wide as a rec12 null mutant [18, 28, 29, 37].

After a DSB is made, the highly conserved Mre11–Rad50–Nbs1 (MRN) complex initiates DSB repair via the nuclease activity of Mre11 [38, 39]. Coupled with the necessary accessory protein Ctp1, Mre11 clips one DNA strand and removes short ssDNA oligonucleotides covalently linked at their 5′ end to Spo11 (or Rec12 in S. pombe) [40–42]. This creates a 3′ ssDNA end that can be lengthened by further resection of the 5′ end by MRN and Exo1 exonuclease [43–46]. Aided by the DNA strand-exchange proteins Rad51 and Dmc1, this long 3′-ended ssDNA invades an intact dsDNA molecule (homolog or sister) for repair by homologous recombination [47].

MRN mutants in different species have varying DSB phenotypes. Both Mre11 and Rad50, but not Nbs1, are required for meiotic DSB formation in the budding yeast S. cerevisiae and the nematode Caenorhabditis elegans [48–51]. In these species, Mre11 is present at the time of DSB formation at special chromosomal sites (hotspots) at which DSBs (and recombinants) arise at high frequency [52]. However, in Coprinus cinereus, Arabidopsis thaliana, S. pombe, Tetrahymena thermophila, and mice, meiotic DSBs are formed in MRN mutants [53–57], though at a slightly reduced level in S. pombe. In all MRN mutants tested where DSBs are made, they are not repaired due to the lack of DNA resection, there are few or no crossovers, and meiotic viability is poor [58]. DSBs can be repaired through an alternative pathway of non-homologous end joining (NHEJ), in which, instead of strand invasion of duplex DNA, two broken DNA ends are ligated together [59]. NHEJ is thought to be restricted during meiosis by the presence of Spo11 (Rec12) bound to the DSB, by MRN binding to DSB ends and promoting homologous recombination [60], and by phosphorylation of SC components [61].

The conserved protein kinases ATR and ATM are necessary for responding to mitotic DNA damage and activating a checkpoint by phosphorylating multiple proteins to halt the cell cycle and thus allow DNA repair. MRN plays a critical role in recruiting and activating ATM and ATR to sites of DNA damage [62]. In meiosis, both ATR and ATM have roles in responding to and regulating DSB formation [63–65], and in both budding yeast and mice, ATM (called Tel1 in yeasts) is needed for full-length DNA resection [46, 66]. Additionally, ATM (Tel1) restricts the number of DSBs during meiosis [65, 67–73]; in budding yeast total DSBs increase 2-fold and in mice 10-fold. ATM (Tel1) prevents the same chromatid from being broken multiple times at or near a given site, as frequent, closely spaced DNA double-cuts are seen in yeast and mice ATM mutants [67, 73–76]. In wild-type meiosis, the formation of one DSB interferes with the formation of another DSB nearby, a feature known as DSB interference. This feature requires Tel1 in both budding and fission yeasts [67, 73]. Budding yeast with mutations in Xrs2 (the Nbs1 homolog) lose DSB interference [77], and mice with a meiotic knockout of Mre11 have frequent, closely spaced (<1 kb) double-cut DNA during meiosis [56], evidence that MRN plays a role in ATM (Tel1) activation during meiosis. ATM and ATR phosphorylate Hop1 to stimulate inter-homolog DSB repair, but this phosphorylation is dispensable for DSB formation [78, 79].

We analyzed DSB formation in mutants of cohesin, LinEs, and Hop1 in S. pombe, with and without Tel1 present. We observed that DSBs were strongly reduced, and surprisingly, some, but not all, of the DSBs were restored when Tel1 was removed. Tel1 kinase activity reduced DSB formation when the cohesin or LinE complexes were aberrant or missing. Strong DSB hotspots compete with weaker ones [73, 80–84], and here we show DSB competition is dependent on both cohesin and LinEs. Additionally, the amount of double-cut DNA was very strongly increased in cohesin, LinE, or Hop1 mutants without Tel1, resulting in very strong negative DSB interference; that is, instead of a first DSB interfering with and preventing a second DSB, the presence of a DSB stimulated the formation of a second. We also show that mre11Δ and rad50Δ, but not nbs1Δ or certain point mutants of the MRN complex, have more double-cut chromatids, resulting in spatially limited negative DSB interference like tel1Δ. When mre11Δ or rad50Δ was combined with a cohesin mutant, the phenotype was again like that of tel1Δ – increased double-cut chromatids and strong negative DSB interference. This suggests that, as seen in mice [56], the MRN complex, but not MRN nuclease activity, is needed to activate Tel1 to restrict meiotic DSBs. We infer that MRN activates the kinase activity of Tel1 to block further DSB formation after a limited number of DSBs are formed in a LinE-bound cluster of hotspots. These results support our previous DSB hotspot clustering model for the control of meiotic DSB formation in S. pombe [73] (see the “Discussion” section).

Materials and methods

Schizosaccharomyces pombe growth materials and methods

Media for cell growth were yeast extract liquid (YEL) and modified Edinburgh minimal medium number 2 (EMM2*) or their agar-containing derivatives [85]. Sporulation, including mating, was on sporulation agar (SPA). Defined media were supplemented with all required nutrients at 100 µg/ml; for adenine-requiring strains, YEL was supplemented with adenine and uracil at 100 µg/ml. Methods were as in [85].

DNA DSB analysis

Schizosaccharomyces pombe strains were cultured and induced for meiosis as previously described [86]. Briefly, haploid pat1-as (L95G)::kanMX6 strains were grown to saturation in 5 ml YEL at 30°C, then inoculated 1:100 into EMM2* and grown 2–3 days at 30°C. Cultures were diluted to OD₆₀₀ of 0.1 in 125 ml EMM2* and grown at 25°C, with shaking, until OD₆₀₀ reached 0.35–0.4 (~16–20 h, depending on the strain genotype). Cells were washed once and resuspended in 125 ml EMM2*–NH₄Cl and incubated at 25°C, with shaking, for 18 h. The ATP analog inhibitor 3-MB-PP1 (Toronto Research Chemicals) was added to 25 µM to start synchronous meiosis. Thirty milliliters of cells were harvested at 0, 7, and 8 h and washed once in cold 50 mM EDTA at 4°C. The supernatant was removed, and the cell pellets were frozen in liquid nitrogen and stored at −80°C until DNA extraction. Cells were embedded in agarose plugs, and DNA was extracted in the plugs to prevent DNA breakage [86]. The agarose-embedded DNA was digested with NotI restriction enzyme, and the DNA separated by pulsed-gel electrophoresis in 0.8% Certified Megabase Agarose (Bio-Rad) in 0.5× TBE at 14°C. The electrophoresis conditions were 26 h at 6 V/cm with either a 7.9-s initial switch time to 54.2-s final switch time (linear ramp) for Fig. 2, or 2.2-s initial switch to 93.7-s final switch (linear ramp) for all other figures. Southern blot hybridization was as described [86]; oligos used for [³²P]-labeled DNA probes are listed in Supplementary Table S2. The blots were scanned on a Typhoon Odyssey (GE Healthcare) and analyzed with ImageQuant XL (GE Healthcare) software. Quantifications of DSBs in Figs. 1, 2, and 4A and Supplementary Figs. S1, S3, S5, S6A, and S7 are the means of data at 0, 7, or 8 h. Quantifications of DSBs in Figs. 3, 4B, 5, and 6 and Supplementary Figs. S2, S4, S6B, and S8 are the means of data at 7 and 8 h combined.

Figure 2.

Meiotic DSB competition is impaired in cohesin and linear element mutants. A strong DSB hotspot (ura1::hph) was introduced 15 kb from the strong wild-type hotspot mbs1 on Chromosome I to analyze the effect of cohesin and LinE mutations on DSB competition. Meiotic DNA was analyzed, by Southern blot hybridization, from rec11Δ, rec10-144, and rec27Δ strains, with mbs1 only (left three lanes for each strain) or mbs1 with ura1::hph (right three lanes), either with (A) or without (B) Tel1 present. The frequency of DSBs at mbs1 is shown in the graphs (as % of total DNA). For each strain, the mean and SEM (or range) are shown from 2 to 7 experiments. The [³²P]-labelled DNA probe was located on the right end of the 0.5 Mb NotI J fragment, as indicated by the star in scheme on the right. Two-tailed P-values calculated by unpaired t-test were ns (not significant, >.05), *(<.05), **(<.005), and ***(<.0005).

Figure 1.

Region-specific meiotic DSBs are increased by tel1Δ in cohesin and linear element mutants. Schizosaccharomyces pombe strains with the pat1-as (L95G) mutation were induced into synchronous meiosis at 25°C. DNA was harvested from pre-meiotic cells at 0 h and in meiotically induced cells at 7 and 8 h, when DSB formation is complete and MI begins; rad50S strains were used to assay total (accumulated) DSBs. Agarose-embedded DNA was digested with NotI and separated by pulsed-field gel electrophoresis (PFGE). Images are of Southern blot hybridizations with [³²P]-labelled DNA probes. DSBs were measured in cohesin mutant rec11Δ and LinE mutants rec10-144 and rec27Δ across an ~0.5 Mb section of Chromosome I (designated Regions 1 and 2), in both tel1⁺ and tel1Δ strains. DSB frequencies at 0, 7, and 8 h (as % of total DNA) are represented in the graphs beneath the blot images: upper graph, the sum of DSBs in Region 1; lower graph, the sum of DSBs in Region 2. For each strain, the mean and SEM are shown from three to six experiments. The [³²P]-labelled DNA probe was located on the left end of the 1 Mb NotI E fragment, as indicated by the star in the scheme on the right. Two-tailed P-values calculated by unpaired t-test were ns (not significant, >.05), *(<.05), **(<.005), and ***(<.0005).

Figure 4.

Loss of Hop1 leads to DSB repression by Tel1 and highly negative DSB interference without Tel1. (A) DSBs were measured in the HORMAD mutant hop1Δ rad50S, with and without Rec11 and Tel1. NotI-digested DNA was Southern blot-hybridized with a [³²P]-labelled DNA probe on the left end of the 1 Mb NotI E fragment on Chromosome I (see Fig. 1), and DSBs were measured across an ~0.5 Mb (designated Regions 1 and 2). DSB frequencies (as % of total DNA) are represented in the graphs beneath the Southern blot images: upper graph, Region 1; lower graph, Region 2 (see Fig. 1 to compare with hop1⁺). For each strain, the mean and SEM are shown from 5–6 experiments. (B) The frequency of double-cut DNA was determined in hop1Δ rad50S and hop1Δ rad50S rec11Δ in both tel1⁺ and tel1Δ strains. DNA digested with NotI was separated by PFGE, blotted, and hybridized with a [³²P]-labelled DNA probe located in the middle of the 0.5 Mb NotI J fragment, between the inserted DSB hotspot ura1::hph and mbs1, to detect DNA chromatids that were cut at both DSB hotspots. The mean DSB frequencies from the 7- and 8-h timepoints of ura1::hph DSBs (green), mbs1 DSBs (blue), and observed DNA double-cuts at both ura1::hph and mbs1 (red) are shown on the graph (as % of total DNA). (C) The double-cut frequencies are very low, and the observed (red) and expected (% ura1::hph x % mbs1; gray) double-cut frequencies (as % of total DNA) are replotted on a separate graph for clarity. Observed and expected values are the mean of the 7- and 8-h timepoints. For each strain, the mean and SEM are shown from three to six experiments. (D) DSB interference, calculated as 1 – (observed double-cut DNA/expected double-cut DNA), for each strain is plotted on the graph; the line in each box indicates the median value; the minimum and maximum values measured are at the bottom and top of each box. Interference was not determined (ND) for strains in which the expected double-cut DNA was <0.1%.

Figure 3.

Frequent double-cut DNA and strong negative DSB interference are observed in cohesin and LinE tel1Δ double mutants. (A) The frequency of double-cut DNA was determined in the cohesin mutant rec11Δ and LinE mutants rec10-144 and rec27Δ, in both tel1⁺ and tel1Δ strains. DNA digested with NotI was separated by PFGE, blotted, and hybridized with a [³²P]-labelled DNA probe (indicated by star on the right) located in the middle of the 0.5 Mb NotI J fragment, between the inserted DSB hotspot ura1::hph and mbs1, to detect DNA chromatids that were cut at both DSB hotspots. The mean DSB frequencies from the 7- and 8-h timepoints of ura1::hph DSBs (green), mbs1 DSBs (blue), and observed DNA double-cuts at both ura1::hph and mbs1 (red) are shown on the graph (as % of total DNA). (B) The double-cut frequencies are very low, and the observed (red) and expected (% ura1::hph x % mbs1; gray) double-cut frequencies (as % of total DNA) are replotted on a separate graph for clarity. Observed and expected values are the mean of the 7- and 8-h timepoints. For each strain, the mean and SEM are shown from three to six experiments. (C) DSB interference, calculated as 1 – (observed double-cut DNA/expected double-cut DNA), for each strain is plotted on the graph; the line in each box indicates the median value; the minimum and maximum values measured are at the bottom and top of each box. Interference was not determined (ND) for strains in which the expected double-cut DNA was <0.1%.

Figure 5.

DSB interference is dependent on Mre11 and Rad50, but not Nbs1. (A) The frequency of double-cut DNA was detected in various mutants of the MRN DNA repair complex. NotI-digested DNA was Southern blot hybridized with a [³²P]-labelled DNA probe located in the middle of the 0.5 Mb NotI J fragment (see Fig. 3), between the inserted DSB hotspot ura1::hph and mbs1, to detect DNA chromatids that were cut at both DSBs. The mean DSB frequencies from the 7- and 8-h timepoints of ura1::hph DSBs (green), mbs1 DSBs (blue), and observed DNA double-cuts at both ura1::hph and mbs1 (red) are shown on the graph (as % of total DNA). (B) The double-cut frequencies are very low, and the observed (red) and expected (% ura1::hph x % mbs1; gray) double-cut frequencies (as % of total DNA) are plotted on a separate graph for visibility. Observed and expected values are the mean of the 7- and 8-h timepoints. For each strain, the mean and SEM are shown from three or four experiments. (C) DSB interference, calculated as 1 – (observed double-cut DNA/expected double-cut DNA), for each strain is plotted on the graph; the line in each box indicates the median; the minimum and maximum values measured are at the bottom and top of each box.

Figure 6.

mre11Δ and rad50Δ, like tel1Δ, have strong negative DSB interference in double mutants with rec11Δ. (A) The frequency of double-cut DNA was detected in various mutants of the MRN DNA repair complex in a cohesin-deficient rec11Δ strain and a rec11Δ tel1-kd strain for comparison. NotI-digested DNA was Southern blot hybridized with a [³²P]-labelled DNA probe located in the middle of the 0.5 Mb NotI J fragment (see Fig. 3), between the inserted DSB hotspot ura1::hph and mbs1, to detect DNA chromatids that were cut at both DSBs. The mean DSB frequencies from the 7- and 8-h timepoints of ura1::hph DSBs (green), mbs1 DSBs (blue), and observed DNA double-cuts at both ura1::hph and mbs1 (red). (B) The double-cut frequencies are very low, and the observed (red) and expected (% ura1::hph x % mbs1; gray) double-cut frequencies (as % of total DNA) are plotted on a separate graph for visibility. Observed and expected values are the mean of the 7-and 8-h timepoints. For each strain, the mean and SEM are shown from three or four experiments. (C) DSB interference, calculated as 1 – (observed double-cut DNA/expected double-cut DNA), for each strain is plotted on the graph. The line in each box indicates the median; the minimum and maximum values measured are at the bottom and top of each box. Interference was not determined (ND) for strains in which the expected double-cut DNA was <0.1%.

DSB interference calculation

DSB interference (I) was calculated in the same manner as genetic interference, using the equation I = 1 – (observed double-cut frequency/expected double-cut frequency), where expected double-cut frequency is (frequency of ura1::hph DSBs) × (frequency of mbs1 DSBs). When double-cut DSBs (dc) were very frequent, the frequency of dc was added to both hotspots: I = 1- [obs. dc/(ura1::hph DSB + dc) x (mbs1 DSB + dc)].

Intragenic gene conversion assay

Schizosaccharomyces pombe strains with either ura1::hph or ura1-171 alleles were mated, and the spores analyzed as described [85]. Briefly, 100 μl of each saturated YEL culture was harvested, mixed, washed three times in dH₂O, and spotted on SPA with adenine and uracil (100 µg/ml), and incubated at 25°C for 3–4 days. Spores were harvested into 0.5 ml dH₂O, glusulase (β-glucuronidase/arylsulfatase; Millipore Sigma) was added 1:100, and the suspension incubated at 30°C for 6 h. 0.5 ml of 60% EtOH was added, and the mixture incubated at room temperature for 15 min. Spores were washed three times in dH₂O and resuspended in 1 ml of dH₂O and stored at 4°C.

tel1-kd strain construction

The 1.8-kb HindIII fragment containing the ura4⁺ gene was amplified by polymerase chain reaction (PCR) from pYF20 with oligos OL5254 and OL5255 containing 80 nt DNA sequence from the Tel1 ORF (Supplementary Table S2), from nucleotides 7787–7866 and 7969–8048. Transformation of strain GP7080 with the PCR fragment generated a tel1::ura4⁺ substitution at nucleotides 7867 to 7968 of the Tel1 ORF, verified by PCR, by selecting for Ura⁺ colonies, creating strain GP9400. Plasmid pFA6-tel1-kd-natMX6 (created and sent to us by Dr. Israel Salguero in the lab of Dr. Stephen Jackson) was used as the template with oligos OL5256 and OL5257 to generate a PCR fragment with mutations G2629D, D2630A, N2635K, and D2649E at the active site kinase residues of Tel1, but without the linked natMX6 cassette. This PCR product (see Supplementary Table S2) was used to transform strain GP9400 (tel1::ura4⁺) to tel1-kd by selecting for FOA-resistant colonies. The tel1-kd integration was verified by sequencing, creating strain GP9402.

hop1-5A construction

We cloned the 1851 bp ORF of hop1, plus 433 bp upstream and 476 bp downstream, into pBR322, and, using site-directed mutagenesis with Q5 (New England Biolabs), we made the following nucleotide changes, with corresponding amino acid changes: T399G (S103A), T909G (S247A), T927G (S253A), A978G (T270A), and A1017G G1018C (S283A). The sequence was verified, and plasmid DNA was digested with Bpu10I and HaeII and used to transform GP565 hop1::ura4⁺ to hop1-5A and FOA-resistance. The hop1-5A integration was verified by sequencing, creating strain GR570.

Results

Tel1 partially represses DSB formation in LinE and cohesin mutants; DSB competition depends on LinEs and cohesin

We analyzed meiosis in strains with the pat1-as (L95G) mutation, which allowed nitrogen-starved cells to be synchronously induced into meiosis at 25°C with the addition of an ATP-analog inhibitor (3MB-PP1) [87]. We could then measure the maximal DSB frequencies in strains with the MRN mutation rad50S, in which DSBs are not resected and thus accumulate [48, 88]. In the cohesin subunit mutant rec11Δ and the LinE mutants rec10-144 (G727E) and rec27Δ, a three- to four-fold reduction of DSBs was seen at multiple hotspots (Fig. 1, designated Regions 1 and 2) across 0.5 Mb on Chromosome I. In double mutant strains with tel1Δ, Region 1 DSBs (upper graph) were significantly increased ~2-fold compared to the tel1⁺ single mutants but were still only 50% of wild-type. Region 2 DSBs (lower graph), however, were not significantly increased in the tel1Δ background, indicating this repression of DSBs by Tel1 is region (or site) specific. Similar results (Supplementary Fig. S1) were seen with another cohesin subunit mutant (rec8Δ) and LinE mutants (rec25Δ and mug20Δ). No DSBs were seen in a rec12-164 (Y98F) tel1Δ mutant, as expected because Rec12 activity is essential for meiotic DSB formation [6, 37]. These results show that Tel1 partially prevents DSB formation in strains lacking intact cohesin and LinEs.

To determine whether these effects on DSB formation extend to recombination, we assayed intragenic gene conversion in the ura1 gene. Note that S. pombe intragenic recombination occurs nearly exclusively by gene conversion [89, 90]. Crossing ura1-61 and ura1-171 generated 210 Ura⁺ spores/10⁶ total spores, in agreement with previous results [91]. Substitution of the first 1.8 kb of the ura1 ORF with a hygromycin-resistance cassette [92], which removes the position of the ura1-61 mutation, created a DSB and recombination hotspot (ura1::hph, see Fig. 2), as ura1::hph x ura1-171 crosses generated 2400 Ura⁺ spores/10⁶ total spores (Table 1). rec11Δ and rec10-144 mutations decreased ura1 gene conversion (~2-fold versus wt), while double mutations with tel1Δ (and the triple mutant rec11Δ rec10-144 tel1Δ) increased Ura⁺ frequency by 5-fold (~2.5-fold versus wt), reflecting the increased DSBs in these mutants (Fig. 1).

Table 1.

tel1Δ increases meiotic recombination in cohesin and linear element mutants

Parental strains (ura1::hph x ura1-171)	Ura+/106 viable spores
	tel1+	tel1Δ
rec+	2400 ± 210 (7)	1700 ± 70 (5)
rec11Δ	960 ± 40 (6)	5000 ± 310 (6)
rec10-144	1200 ± 50 (4)	5000 ± 150 (4)
rec10-144 rec11Δ	1200 ± 290 (3)	4800 ± 830 (4)
mbs1Δ	3100 ± 150 (3)	4100 ± 550 (3)
rec10-144 mbs1Δ	1200 ± 110 (3)	2700 ± 300 (3)

Crosses were homozygous for the indicated mutations in the left column and for tel1⁺ or tel1Δ as indicated in the middle and right columns; they were heterozygous for the indicated ura1 mutations. Intragenic recombination without the DSB hotspot ura1::hph was measured by crossing ura1-61 and ura1-171, which produced 210 ± 11 Ura⁺/10⁶ viable spores (n = 2), similar to results previously reported [91]. Data are the mean ± SEM from (n) crosses.

We next examined the phenomenon of DSB competition, in which addition of a strong DSB hotspot decreases the frequency of DSBs at a nearby hotspot (<200 kb away), as measured in a meiotic cell population [73, 80–84]. A LinE protein artificially tethered to a site creates a new DSB hotspot, but this hotspot did not impart competition on the surrounding DSBs [92]. This suggests that normal LinE loading, for example by cohesin, is important in establishing DSB competition. The inserted hotspot ura1::hph competes with the mbs1 hotspot 15 kb away in wt (Fig. 2A; [92]), but the two hotspots showed no DSB competition in rec11Δ, rec10-144, or rec27Δ mutants (Fig. 2A); the mbs1 DSB frequency was unchanged or slightly higher than without the inserted hotspot. The weak DSBs seen in these mutants might not be strong enough to establish competition, so we looked at DSBs in double mutants with tel1Δ; in previous assays, Tel1 did not affect DSB competition [72, 92], and our results here agree (Fig. 2B). However, none of the three rec mutants showed DSB competition, as the frequency of DSBs at mbs1 was not reduced but rather was slightly increased in the presence of ura1::hph (Fig. 2B). These results show that DSB competition is dependent on cohesin and LinEs, and both must be in place for proper DSB regulation.

We deleted mbs1 to determine if a competitive effect could be measured by the ura1 genetic assay. When mbs1 was deleted, the frequency of Ura⁺ recombinants in a ura1-171 x ura1::hph cross was increased modestly in wild-type (∼30%) and strongly in tel1Δ (2.5-fold, Table 1), demonstrating competition between mbs1 and the ura1::hph hotspot. Interestingly, the Ura⁺ frequency was reduced in rec10-144 tel1Δ mbs1Δ compared to rec10-144 tel1Δ mbs1⁺ (Table 1), in agreement with the DSB analysis: the presence of ura1::hph stimulates DSBs at mbs1, and mbs1 stimulates ura1⁺ recombination when cohesin and/or LinEs are absent.

Loss of either cohesin or LinEs in tel1Δ mutants results in strong negative DSB interference

Next, we examined DSB interference by assaying the frequency of DNA cut at both the mbs1 and ura1::hph hotspots, 15 kb apart. Like the wild-type strains, the rec11Δ, rec10-144, and rec27Δ mutants showed little double-cut DNA (Fig. 3A and B); however, the low frequency of DSBs and the very low expected frequency of double-cuts (<0.1%; Fig. 3B) did not allow reliable determination of interference (marked ND, Fig. 3C). However, when analyzed in a tel1Δ background, the frequency of double-cut DNA was remarkably high, even higher than with tel1Δ alone (∼3% versus 1.5%; Fig. 3A and B). DSB interference was also much more negative (−4, −8, and −2 in the three tel1 double mutants versus −1 in the tel1 single mutant, Fig. 3C), indicating that a DSB at one hotspot strongly stimulates a DSB at the second hotspot in the absence of both cohesin or LinEs and Tel1. Similar results (Supplementary Fig. S2A, B, and C) were seen with mutants of another cohesin subunit (rec8Δ) or LinEs (rec25Δ and mug20Δ), and in double cohesin-LinE mutants (rec11Δ rec27Δ and rec11Δ rec10-144), indicating cohesin and LinEs act together to control close DSBs.

Loss of Hop1 leads to DSB repression by Tel1 (but not region-specific) and highly negative DSB interference without Tel1

DSBs and meiotic recombination are significantly decreased by the deletion of meiotic HORMAD protein hop1Δ in S. pombe [42, 93], so we questioned whether this was also due to Tel1 repression. As in cohesin and LinE mutants (Fig. 1), the low DSBs in Region 1 in hop1Δ increased ~2-fold when tel1 was also deleted (Fig. 4A). Interestingly, the Region 2 DSBs also increased ~2-fold (Fig. 4A), which was not observed in cohesin or LinE mutants lacking Tel1 (Fig. 1). We next assayed double-cut DNA between the ura1::hph and mbs1 DSB hotspots. There was no observable double-cut DNA in hop1Δ, though again the very low expected frequency of double-cuts (<0.1%; Fig. 4C) did not allow reliable determination of interference; significant double-cuts were observed in hop1Δ tel1Δ (Fig. 4B and C), which resulted in highly negative DSB interference (Fig. 4D). Similar results were seen when the cohesin subunit Rec11 was deleted in hop1Δ and hop1Δ tel1Δ strains (Fig. 4B, C, and D), which suggests that cohesin and Hop1 both act to control close DSBs. Since Hop1 is a direct target of Tel1 [78], we made a phosphorylation-minus mutant of Hop1; five potential S/TQ sites [94], the consensus for Tel1 (ATM) phosphorylation sites [95], were changed from serine to alanine. These mutations had no observed effect on DSB formation (Supplementary Fig. S3), and DSB interference was still strongly positive (Supplementary Fig. S4C), suggesting that Hop1 is not a direct target, or not the only target, of Tel1 to restrict DSBs.

DSB interference depends on Mre11 and Rad50, but not on MRN-dependent resection

Recent studies have demonstrated that the MRN complex limits DSBs in mice [56], so we examined additional MRN mutants in S. pombe rad50⁺ strains. (For the previous DNA analyses, the rad50S mutation was used to accumulate DSBs and thus enable their maximal quantification.) As expected, all MRN mutants examined, including deletions, accumulated DSBs (Supplementary Fig. S5), though with modest reductions in DSB frequencies. rad50S mutants have strong positive DSB interference (Fig. 3) [73], as do nbs1Δ, mre11-D65N (nuclease-minus), and ctp1Δ mutants (Fig. 5A, B, and C). In contrast, both mre11Δ and rad50Δ had negative interference like tel1Δ (Fig. 5). Since clipping of the Rec12-oligo and resection are impaired in all these strains [41, 42], active resection and repair are not necessary for positive DSB interference. nbs1Δ retained positive interference, and we thus infer that Mre11 and Rad50 are still present at a DSB during meiosis and can recruit Tel1 in the absence of Nbs1 in S. pombe. Negative DSB interference was also observed in tel1-kd (Fig. 5A, B, and C), in which the kinase active site was inactivated, evidence that the kinase activity of Tel1 is necessary for DSB interference. The mre11Δ tel1Δ mutant was like either single mutant (Fig. 5A, B, and C), which suggests MRN and Tel1 act coordinately after a DSB is made.

DSB frequencies are similar in multiple MRN and NHEJ mutants

The similarities in phenotype of tel1Δ, mre11Δ, and rad50Δ suggest that MRN and Tel1 act together in response to DSBs. However, the mre11Δ and rad50Δ strains consistently had marginally fewer DSBs than a rad50S strain, and mre11Δ tel1Δ had fewer than rad50S tel1Δ (Fig. 1 and Supplementary Fig. S5). To address the possibility that some repair, and thus loss of observed DSBs, occurred through NHEJ in the absence of MRN, we studied pku70Δ derivatives, which lack NHEJ [96, 97]. pku70Δ had no significant effect on DSB formation or DSB interference in rad50S, rad50S tel1Δ, mre11Δ, or mre11Δ tel1Δ strains (Fig. 1 and Supplementary Fig. S6A and D). Therefore, NHEJ appears to be very infrequent in the absence of MRN during meiosis in S. pombe.

In vegetative cells, Tel1 is hyperactive in rad50S strains of S. cerevisiae and mice [98, 99]. There was no significant difference between rad50S strains with or without Tel1 (Figs. 1, 2, and 3A), but DSBs in the cohesin or LinE mutant rad50S backgrounds might be repressed by hyperactive Tel1. Therefore, it was necessary to assay DSBs in a different repair-deficient mutant, mre11-D65N [41, 42, 100]. When comparing the rad50S versus mre11-D65N derivatives, DSBs in Region 1 were increased significantly by mre11-D65N, ~2-fold, in rec27Δ (4.8 versus 9.2, P = .006, Fig. 1 and Supplementary Fig. S3), but not significantly increased in rec11Δ (7.6 versus 8.7, P = .57; Fig. 1 and Supplementary Fig. S7) or hop1Δ (4.9 versus 5.6, P = .55; Fig. 4A and Supplementary Fig. S3). Additionally, rad50S nbs1Δ rec11Δ triple mutant showed increased Region 1 DSBs compared to nbs1Δ rec11Δ (8.9 versus 4.9, P = .1; Supplementary Fig. S7), a characteristic of less Tel1 function and not hyperactivity caused by rad50S.

Loss of cohesin in mre11Δ and rad50Δ mutants results in strong negative DSB interference

We found mre11Δ rec11Δ and rad50Δ rec11Δ mutants had a three- to four-fold increase of DNA double-cut at both mbs1 and ura1::hph (Fig. 6A and B) compared to mre11Δ rec11⁺ and rad50Δ rec11⁺ (Fig. 5A and B), similar to observations in tel1-kd rad50S rec11Δ (Fig. 6A and B) compared to tel1-kd rad50S rec11⁺ (Fig. 5A and B), and earlier results with tel1Δ (Fig. 3A and B). All three double mutants had highly negative DSB interference (Fig. 6A, B, and C). Conversely, none of the rad50S rec11Δ, nbs1Δ rec11Δ, or mre11-D65N rec11Δ double mutants produced double-cut DNA at high frequency (Fig. 6A and B). The low frequencies of DSBs in rad50S rec11Δ and nbs1Δ rec11Δ mutants and the expected infrequent double-cut DNA (< 0.1%) do not allow DSB interference to be calculated (ND, Fig. 6C), but the mre11-D65N rec11Δ mutant had weaker positive DSB interference compared to mre11-D65N rec11⁺ (0.4 versus 0.9, Figs. 4C and 5C); the observed double-cut DNA was not significantly different than that expected in mre11-D65N rec11Δ, so it is unclear if there is weak positive or no interference. The mre11-D65N rec27Δ mutant did not show an interference decrease compared to mre11-D65N rec27⁺ (0.9 versus 0.9; Fig. 4C and Supplementary Fig. S4C). The rad50S nbs1Δ rec11Δ triple mutant also had increased double-cut DNA compared to rad50S rec11Δ or nbs1Δ rec11Δ  mutants and had highly negative DSB interference (Fig. 6A, B, and C). The rad50S and nbs1Δ mutations combined may create a less active and/or less stable MRN complex, a phenotype that is exacerbated by the removal of intact cohesin. If the rad50S mutation hyperactivated Tel1, there would be stronger interference and less double-cut DNA, not the observed increase, when rad50S is added to nbs1Δ rec11Δ.

Discussion

Identifying the factors that influence meiotic DSB formation is essential for determining the mechanism of DSB control. The proper timing and placement of meiotic DSBs and the resulting recombination are highly dependent on protein complexes that organize and align chromosomes [25]. In all species studied, the chromosome axis, upon which the synaptonemal complex can be built, is essential for proper meiotic recombination, and meiotic cohesins are necessary for the establishment of the axis [10, 11, 23, 33–36]. Loss of chromosome axis proteins results in less frequent DSBs and recombination [17–20]. The fission yeast S. pombe linear elements with Hop1 form chromosome axis structures that are dependent on meiotic cohesin [10, 23, 35, 37]. Our results show that cells lacking cohesin, LinEs, or Hop1 have reduced DSBs by the action of Tel1 kinase (Figs. 1 and 4A), and in Tel1’s absence there is an increase in chromatids that are cut twice at nearby hotspots 15 kb apart. This suggests that proper loading and interaction of hotspot-bound LinEs and Hop1 are necessary for the formation of DSBs sensitive to competition and interference; in their absence, Tel1 is activated to repress some DSBs, particularly at nearby hotspots. Some DSB hotspots are completely dependent on cohesin and LinEs, but other DSBs are not, which indicates that some additional, unidentified factors are involved.

Previous studies demonstrated that crossovers are reduced in S. pombe cohesin, LinE, and Hop1 mutants, but not uniformly, as some genetic intervals have stronger reductions than others [17, 18, 28, 93, 94]. DSBs were not repressed by Tel1 at all hotspots, as some DSBs did not increase in frequency in cohesin or LinE tel1Δ double mutants; however, this was different in hop1Δ, as both Region 1 and Region 2 DSBs were increased by tel1Δ. The causes of the regional variability and potential differences among DSB hotspots are intriguing and remain to be determined. The previous conclusion, based on observations with the pat1-114 Ts mutant at 34°C, that LinEs are required for DSBs and are determinants of hotspot position is not strictly correct—DSBs in some regions can be formed at 25°C, although at reduced levels in the absence of LinEs, and Tel1 prevents their formation at high levels (Fig. 1). This agrees with microscopy of GFP-tagged LinE proteins that show aberrant morphology at high temperature compared to 25°C [101]. Of note, MRN mutants mre11Δ and rad50Δ also show stronger reductions of DSBs at 34°C [53] than at 25°C (Supplementary Fig. S5). Previous experiments showed that wt meiosis did not differ in many aspects of DSB formation, repair, and spore viability at 25°C and 34°C [41, 53, 102, 103, 104], but the mutants used here are more sensitive to temperature changes.

We pursued other factors that act with Tel1 to limit DSBs. An obvious candidate was the MRN complex. During mitotic DNA damage, MRN recruits and activates Tel1, which in turn phosphorylates itself and MRN [62, 105]. This interaction creates a feedback loop that enables the signal cascade initiated by Tel1 phosphorylation of multiple proteins to delay the cell cycle until DNA damage is repaired. Because of the interdependent nature of MRN and Tel1, use of MRN mutants to accumulate DSBs is potentially problematic. Tel1 is hyperactive in mice and S. cerevisiae rad50S strains [98, 99], which could result in the low DSBs seen in rad50S cohesin and LinE mutants (Fig. 1 and Supplementary Fig. S1). mre11-D65N rec11Δ and mre11-D65N hop1Δ mutants did show small increases in DSBs compared to rad50S rec11Δ and rad50S hop1Δ that were statistically insignificant; however, there was a significant increase in DSBs in rec27Δ mre11-D65N compared to rec27Δ rad50S. rad50S tel1Δ did not significantly increase DSB frequencies in Region 1 or 2 (Fig. 1), or at mbs1 or ura1::hph (Fig. 3A) compared to rad50S tel1⁺. Thus, while there is some evidence of Tel1 hyperactivation in rad50S, it is relatively minor, and rad50S and mre11-D65N both have equally strong positive DSB interference (Figs. 3C and 5C), suggesting similar Tel1 activity in response to meiotic DSBs. It is possible that other functions of Tel1 are affected differentially depending on the MRN mutant.

Recent studies of meiosis in budding yeast and mice have demonstrated Tel1 and MRN’s role in limiting multiple proximal DSBs (<1 kb apart) on the same chromatid, which can have deleterious effects [56, 67, 74–76]. Our results in S. pombe show that mre11Δ and rad50Δ result in phenotypes very similar to those of tel1Δ: wild-type positive DSB interference turns negative (Fig. 5), and DSBs are increased in cohesin mutant strains that lack MR (Mre11 or Rad50) (Fig. 6). However, these phenotypes are not displayed in MRN point mutants rad50S and mre11-D65N (nuclease-negative), ctp1Δ, or nbs1Δ,in which resection and clipping are impaired [106]. This suggests that MRN is first recruited to a DSB end and then recruits and activates Tel1 to stop further DBSs on the same DNA chromatid, independent of DNA resection; Tel1 is also recruited to mitotic DNA damage by nuclease-deficient MRN [107].

We were surprised that the nbs1Δ mutant still had positive DSB interference, as Nbs1 interacts directly with Tel1 [108], and mutants of Xrs2, the Nbs1 homolog, in budding yeast are defective in short-range (<1 kb) DSB interference [77]. Our results are consistent with studies in S. pombe that show MR alone is enough to recruit Tel1 to DSBs, as H2A phosphorylation by Tel1 was observed in nbs1Δ, but not in mre11Δ, mutants during mitotic DNA damage when Tel1 was overexpressed, and Mre11 is recruited to DSBs in the absence of Nbs1 [109]. Furthermore, Rec12-oligo clipping is partially active, and spore viability is higher, especially at 25°C, in nbs1Δ compared to mre11Δ or rad50Δ [41]. The rad50S nbs1Δ rec11Δ mutant had a significant increase in double-cut DNA that is indicative of no Tel1 action, but the nbs1Δ rec11Δ and rad50S rec11Δ mutants did not (Fig. 6A, B, and C). These results suggest an Mre11–Rad50 interaction with cohesin (or LinEs) and that this interaction is affected by the combination of nbs1Δ  and rad50S mutations. LinE formation is dependent on the presence of MRN (but not resection), as nuclear structures seen by electron microscopy form normally in a rad50S mutant but not in a rad50Δ mutant [106], suggesting an MRN–LinE interaction. In cohesin and LinE mutants, MRN is likely present at the initial DSB site, and lack of a proper chromosome axis may trigger MRN to coordinate with Tel1 to repress DSB formation. How this interaction occurs is of interest but not yet known.

We have investigated double-cut DNA between DSB hotspots in S. pombe only at a relatively long distance (15–180 kb, Fig. 3 and Supplementary Fig. S8). It is not known if S. pombe shows the same frequent double-cutting within the same narrow DSB hotspot (35–200 bp) as seen in mice [56, 74]. The large amount of Spo11-oligos generated in an Mre11 conditional mutant in mice is not observed in any tested MRN mutant of S. pombe [41, 42, 106]. There is also a clear difference in the amount of excess DSBs prevented by Tel1—there is not a 10-fold increase in a tel1Δ mutant (Fig. 1) as seen in mice [65]. However, the loss of Mre11 in mice results in increased multiple cut chromatids, but loss of Mre11 nuclease activity does not [56]; this is similar to the phenotypes of the S. pombe mre11 mutants analyzed here.

The robust increase of double-cut chromatids in strains lacking both the pre-DSB action of cohesin, LinEs, and Hop1 and the post-DSB action of MRN and Tel1 is very interesting, as the double-cuts are much more frequent than expected from random breakage (Figs. 3, 6 and Supplementary Fig. S2); there are multiple possibilities for the cause of this high negative DSB interference. Since DSBs are measured in a population, they could result from “hot cells,” a subset of cells that are more active and have excess DSB formation compared to the other cells. A hot cell is more likely to have multiple DSBs—a DSB at mbs1 would frequently result in a second DSB at ura1::hph due to higher DSB activity in that “hot” cell. However, meiotic crossovers measured on two different S. pombe chromosomes occurred independently, providing evidence against hot cells in either tel1⁺ or tel1Δ [73]. The negative DSB interference also decreases as the distance between hotspot pairs increases [73] (Supplementary Fig. S8), which is not predicted by hot cells. In budding yeast, it was proposed that some chromosomal domains have higher intrinsic DSB activity, which leads to DSB clusters, and observed negative interference that is eliminated when meiotic prophase is extended [110]. The study here is limited to only a few hotspot pairs (Supplementary Fig. S8), and this region may have higher DSB potential than other parts of the genome, and the large increases of double-cut DNA may not occur between all hotspots. We prefer the interpretation that cohesin, LinEs, and Hop1, in addition to being involved in the initiation of DSBs, have a role in inhibiting multiple DSBs, structurally or through additional factors, such as inhibiting the DSB-forming complex. MRN could also provide a surveillance function, not only for DSBs but also for chromosome axis components, to prevent additional DSBs on a disorganized DNA substrate.

We propose that during S. pombe meiosis, the LinEs are loaded at a subset of potential DSB hotspots through Rec10’s interaction with phosphorylated cohesin subunit Rec11 ([21, 22] (Fig. 7). LinE-bound sites (DSB hotspots) are brought together through an undetermined mechanism, perhaps by the cohesin complex forming clusters at the same time as loading LinEs. Recent results in vitro with purified LinE complex also suggest that this could happen through phase-separated condensation of the LinE complex [111]. The in vitro purified LinE protein condensates form more readily when a LinE component (Rec27) can bind DNA. Self-assembly and condensation have been proposed and demonstrated for complexes of the SC and DSB complex in other species and proposed as a mechanism of DSB and crossover control [112, 113]. The proper interactions of the LinE proteins (and likely Hop1) with DNA, and subsequently with other close DNA-bound LinE complexes, are an important mechanism of DSB control in S. pombe. We do note, however, that because MRN is necessary for DSB DNA end tethering [114, 115] promoted by Tel1 [116], destabilized DNA ends in MRN- or Tel1-deficient strains conceivably could influence multiple DSBs on the same chromatids. When the cohesin or LinE complex is also missing and the chromosome axis structure is chaotic, perhaps the DSB is no longer recognized and multiple cuts are made.

Figure 7.

A model of DSB interference. Cohesin and the linear element complex cooperatively load onto meiotic DNA at sites of potential DSB hotspots. These protein complexes cluster together and bring the potential hotspots into physical proximity, allowing DSBs to be coordinately regulated over a limited distance. Cohesin, LinEs, and Hop1 recruit or activate the DSB-forming complex, and a DNA break is made (top left). In response to this break, the MRN-Ctp1 DNA repair complex is recruited to the DSB. We propose that MRN in turn recruits the Tel1 kinase, and Tel phosphorylates the DSB-forming complex to prevent further DSBs at other DSB hotspots in that cluster (top right). In the absence of MRN or Tel1, multiple DSBs are made preferentially in the same cluster, resulting in the observed negative DSB interference. In the absence of cohesin/LinEs, MRN and Tel are active and restrict some DSBs (bottom left). Without cohesin/LinEs/Hop1 (chromosome axis complexes) and MRN/Tel1, both the regulatory cluster and the inhibitory phosphorylation by Tel1 are absent, resulting in increased double-cut DNA and strong negative DSB interference (bottom right).

The loss of DSB competition in cells that lack either cohesin or LinEs (Fig. 2) and the high frequency of double-cut chromatids (Fig. 3) support a shared function in regulating DSBs spatially. The proposed clustering of LinEs initially occurs over 100–200 kb (∼20–40 cM), corresponding to the extent of DSB competition and interference [73] (Supplementary Fig. S8). This may result from LinEs and Hop1 interacting over the limited extent of chromatin loops within the Rec8 cohesin-mediated axis (via Rec11) [117]. In this scenario, a DSB is made, causing localization of MRN and Tel1, which then prevents more DSBs in the clustered interval (Fig. 7). In strains lacking cohesins (or LinEs or Hop1) and Tel1, close DSB hotspots are unrestricted and cooperatively loaded with DSB complexes and then broken at nearly the same time. The targets of Tel1 after DSB formation in S. pombe remain to be determined. In a phospho-proteomics study of meiotic S. cerevisiae cells, 19% of the DSB-dependent phosphorylation sites detected had the Tel1 consensus motif [118], and evidence suggests that the Rec114–Mei4–Mer2 complex, related to S. pombe Rec15–Rec7–Rec24, is one potential target [68, 119]. The requirement for Tel1 to limit meiotic DSBs appears to be conserved. Perhaps additional features of this mechanism are conserved as well, and the results reported here may apply to other species.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work was supported by the National Institutes of Health of the United States of America [NIH R35 GM118120 to G.R.S. and NIH P30 CA015704 to the Cellular Imaging Shared Resource, RRID:SCR_022609, of the Fred Hutchinson/ University of Washington/ Seattle Children’s Cancer Consortium].

Data availability

All data are contained within the manuscript and/or supplementary files.