Tel1 associates with meiotic DSB hotspots genome-wide

To explore Tel1 interaction with chromosomes during meiosis, we N-terminally FLAG-tagged Tel1 at the native locus of the TEL1 gene (Methods). FLAG-Tel1 diploid cells progressed normally in meiosis I and II as monitored by the appearance of cells with two (MI) or four (MII) nuclei by DAPI staining (S1a Fig), both in a wild-type and in a SAE2 defective background (sae2Δ). This indicates that the FLAG tag does not impair Tel1 function in the meiotic DNA damage checkpoint activated by unresected DNA ends [38]. FLAG-Tel1 protein level increased slightly (~1.5 fold) during meiosis compared to pre-meiotic conditions (t = 0) reaching a maximum at 4 h, before decreasing again (S1b and S1c Fig), thus broadly following kinetics of meiotic DSB formation and repair (S1d and S1e Fig). However in a sae2Δ where unprocessed DSBs accumulate and a Tel1-dependent checkpoint is activated [38] (S1a Fig), Tel1 protein level increased ~2.5-fold by 4 h, and up to 3-fold by 8 h (S1b and S1c Fig). In a sae2Δ strain expressing catalytically dead Spo11-Y135F, where DSBs are not formed, the sae2Δ-dependent cell cycle arrest is alleviated (S1a Fig), and Tel1 protein level remained similar to wild-type (S1b and S1c Fig). Thus, in the sae2Δ background, unresected DSB ends activate the DNA damage checkpoint, leading to persistently elevated Tel1 protein level across meiosis.

In S. cerevisiae, average DSB hotspots activities vary over several orders of magnitude, the strongest hotspots being cleaved in ~15% of the cells [8]. Thus, the detection of protein interacting with meiotic DSBs is inherently limited by the frequency of hotspot cleavage. In order to determine whether Tel1 interacts with meiotic DSBs, we performed qPCR-based ChIP experiments of FLAG-Tel1 at two strong DSB hotspots (GAT1 and ARE1, ranked 1^th and 14^st [8]) relative to 2 control sites (CDC39 and MRE11). While no binding of FLAG-Tel1 at GAT1 and ARE1 was detected prior to meiotic DSB formation (time 0 h, Fig 1a), Flag-Tel1 was transiently recruited at DSB hotspots at 3 and 4 h (Fig 1a), coinciding with meiotic DSB formation [24]. No significant enrichment was detected at 5 h (Fig 1a). Thus, Tel1 transiently associates with meiotic DSB hotspots, in agreement with the short-lived nature of Spo11-DSBs in a wild-type background.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. FLAG-Tel1 associates with meiotic hotspots in a wild-type background.

a, Tel1 is transiently recruited to meiotic hotspots. Mean of %IP values of FLAG-Tel1 binding, quantified by ChIP-qPCR at the ARE1 and GAT1 hotspots relative to the CDC39 and MRE11 control sites. Points and error bars represent mean ± SD of three independent meiotic cultures (except for time 0h for which n = 2). P values are from paired ANOVA test between time 0h and other time-points, corrected via Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli for multiple comparisons test. ns (not significant); ^***, p < 0.001, ^****, p < 0.0001. b, ChIP-seq read coverage of FLAG-Tel1 (blue), Spo11-FLAG (orange) at 3 and 4 h into meiosis. A representative region of ChrVIII is shown. DSBs (red shaded area) and axis (grey shaded area) are represented by Spo11-oligos (red) and Rec8-HA ChIP-seq (grey), respectively. Horizontal axis: position on chromosome VIII. Vertical axis: Read coverage for each ChIP-seq subtracted from the input signal for Tel1 and Spo11. ChIP-seq for Rec8 (GSE69231) and Spo11 (GSE52863) as well as Spo11-oligos (GSE26449) were previously performed by Sun et al. and Pan J, et al., respectively, in a wild-type background [8,53]. c-d, Boxplots of Tel1 c) and Spo11 d) normalized counts per 1500 bp (see Methods) partitioned by DSB strength percentile (Spo11-oligos NormHPM). P values are from Kruskal-Wallis test between percentile 0.2 and all the other percentile, corrected via Dunn’s for multiple comparisons test. ns (not significant), ^**, p < 0.01; ^****, p < 0.0001. e, FLAG-Tel1 read densities (RPKM per 20-bp bins) of the 8^th and 9^th percentiles ±5 kb from DSBs midpoint at 3h (light blue line) and 4h (dark blue line) in SPM, identified by Spo11-oligos [8]. f, Spo11-FLAG read densities (RPKM per 20-bp bins) of the 8^th and 9^th percentiles ±5 kb from DSBs midpoint at 3h (light orange line) and 4h (brown line) in SPM, identified by Spo11-oligos [8].

https://doi.org/10.1371/journal.pgen.1011904.g001

To determine whether this localized enrichment reflects genome-wide recruitment, we performed ChIP-seq of FLAG-Tel1 at 3 h and 4 h during meiosis. At both time-points, Tel1 showed discrete ChIP-seq peaks that visually colocalize with Spo11 [39] binding sites and DSB peaks generated by sequencing of Spo11-oligos [8] (Fig 1b, pink shaded area, S2a Fig). A second category of Tel1 peaks also colocalized with the binding sites of Spo11-FLAG, but instead coincided with the binding sites of the axis protein Rec8 [40] (Fig 1b, grey shaded area), which define the chromosomes axis regions. Despite the visual colocalization of Tel1 and Spo11 signal with DSB hotspots at both 3 and 4h into meiosis, genome-wide analysis revealed no significant correlation with DSB hotspots strength as defined by Spo11-oligos (S2b and S2c Fig). The low signal-to-noise ratio precluded reliable genome-wide peak identification and high-resolution analyses. Nevertheless, a moderate but significant increase in Tel1 signal with hotspot strength emerged when considering the 20% strongest hotspots (8^th and 9^th percentiles, Figs 1c, S2d and S2e Fig r = 0.42 at 3h). Similarly, an increase in Spo11 signal [39] was only observed within this subset (Figs 1d and S2d, S2e r = 0,48 at 3h). Even within this subset, Tel1 and Spo11 signals correlated poorly (S2e-S2g Fig). Pile-up analyses further revealed that Tel1 signal at the 20% strongest hotspots displayed a bimodal distribution centered on hotspot midpoint at 3 and 4h, consistent with Tel1 binding to DSB ends that undergo resection in wild-type cells (Fig 1e). In contrast, Spo11 signal formed a single narrower peak centered on hotspots midpoints (Fig 1f), reflecting its catalytic role in DSB formation.

To conclude, our ChIP-qPCR experiments demonstrate that Tel1 interaction with DSBs hotspots is weak and transient during meiosis. While Tel1 and Spo11 ChIP-seq signals visually colocalize with DSB hotspots, there is no quantitative genome-wide association, likely highlighting distinct binding dynamics caused by specific functions at meiotic DSB hotspots.

Tel1 association with meiotic DSB hotspots is stabilized in the absence of resection

Given the weak and transient association of Tel1 with meiotic DSB hotspots in wild-type cells, and prior evidence suggesting Tel1/ATM preferentially binds unresected, protein-bound DSBs [35,36], we hypothesized that Tel1 binding would be enhanced in a sae2Δ mutant where impaired resection preserves Spo11-DSB intermediates. In this background, Spo11 remains covalently bound to DSB ends due to the abrogation of the Mre11-dependent endonucleolytic pathway that releases Spo11 from 5’ DNA ends [37]. The Tel1-dependent DNA damage checkpoint is activated [38], and only ~50% of the cells progress through the first meiotic division at 8 h (S1a Fig).

To investigate further Tel1 localization along chromosomes during meiosis, we assessed Tel1 binding genome-wide by ChIP-seq in the sae2Δ background at 4 and 6 h (experiment 1), and at 6h on two additional independent meiotic time-courses (experiment 2), compared to 0 h. As a control we also mapped Spo11-FLAG binding sites at 6h, in the same experimental conditions. As for FLAG-Tel1 ChIP-seq experiments, Spo11-FLAG cultures were treated with formaldehyde, thereby detecting not only covalent interactions of Spo11 with unresected DSB sites but also non-covalent interactions [41]. Before the onset of meiosis, neither FLAG-Tel1 nor Spo11-FLAG were detected at specific chromosomal regions beside peaks corresponding to “hyper-ChIPable” regions that were common to both strains, and to the untagged control [42] (S3a Fig). 4 h after meiotic entry, Tel1 shows discrete ChIP-seq peaks that increased significantly at 6 h (Figs 2a and S3e), when the large majority of DSBs has been formed and remained unrepaired in the sae2Δ background [37].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Flag-Tel1 localizes to meiotic hotspots and axis sites in sae2Δ background.

a, Read coverage of FLAG-Tel1 (at 0h, 4h, 6h (exp1, n = 1) and 6h (exp2, n = 2)), Spo11-FLAG (n = 2), DSBs (CC-seq) [43] at 6 h and Rec8-HA ChIP-seq at 4 h. Horizontal axis: position on chromosome VIII. Vertical axis: Read coverage for each ChIP-seq subtracted from the untagged signal and the signal at 0 h for Tel1 and Spo11. Tel1 and Spo11 ChIP-seq and CC-seq were performed in a sae2Δ background. Rec8 ChIP-seq performed in a wild-type background (GSE69231) is from Sun et al. [53]. b, zoom-in on a region of ChrVIII (rectangle on panel a) illustrating the overlap of FLAG-Tel1 signal with Spo11-FLAG and Rec8 signals (grey shaded area), and FLAG-Tel1 signal with Spo11-FLAG and DSB sites (red shaded areas). c, Table of overlap between axis, DSBs, Tel1 and Spo11 peaks relative to each other’s. Percentages are indicated for the features in the first line relative to features in the first column. “Alone” in the bottom line refers to peaks overlapping with none of the other features. d, e, Venn diagrams showing the overlap between ChIP–seq peaks (Spo11, Tel1 and Axis) and CC-seq peaks (DSBs). Numbers are indicated using the color code of each individual feature. f, Tel1 recruitment to meiotic hotspots is dependent on Spo11 catalytic activity. FLAG-Tel1 recruitment measured by ChIP-qPCR at the ARE1, ATF2 and GAT1 hotspot relative to three control sites (CDC39, DRS2 and YCR013C) in sae2Δ ndt80Δand sae2Δ ndt80Δ spo11-Y135F mutant. Error bars indicate SD from 3 (time 0 h) and 5 (time 6 h) independent meiotic cultures. P values are from paired ANOVA test between time 0 h and 6 h, corrected via Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli for multiple comparisons test. ns (not significant), ^****, p < 0.0001.

https://doi.org/10.1371/journal.pgen.1011904.g002

To determine whether Tel1 and Spo11 colocalized with DSB sites, we compared ChIP-seq signals with DSBs mapped at 6 h in the sae2Δ background by CC-seq, a technique used to map sites of DNA–protein covalent complexes genome-wide [43]. About 3,000 DSB hotspots were identified by CC-seq [43], that spatially and quantitatively correlated with DSB hotspots identified by Spo11-oligo immunoprecipitation [8,43]. Consistent with the Tel1 signal at DSB hotspots in the wild-type background—but now with improved resolution—Tel1 peaks at 4 and 6h visually colocalize with both Spo11-FLAG binding sites and DSB peaks generated by CC-seq (Fig 2b, pink shaded areas). Smoothing the 6 h FLAG-Tel1 signal with windows of increasing size revealed alternative domains of higher and lower signal that mirror peaks previously observed on DSB maps [8,44], and recapitulated by CC-seq (S3b Fig). Similarly to ChIPs performed in wild-type cells (Fig 1a), a second category of Tel1 peaks also colocalizes with the binding sites of Spo11-FLAG and of Rec8 [40] (Fig 2b, grey shaded area), which define the chromosomes axis regions. Tel1 and Spo11 ChIP-seq signals at 6 h correlated well between duplicates (S3c and S3d Fig, r2 = 0.9 and 1 respectively), therefore data were pooled for subsequent finer-scale analyses. Tel1 and Spo11 peaks were determined with MACS2 and compared to each other, to DSB hotspots mapped at 6 h in the sae2Δ [43], and to axis coordinates (defined as the regions presents in at least two of the Red8, Hop1 and Red1 binding sites, S4a Fig). We detected 3,103 and 3,042 peaks of Tel1 and Spo11, respectively (S4b Fig). Aggregating signals across the genome confirmed a strong correspondence between Tel1, Spo11, DSB hotspots and axis regions (Figs 2c-2e and S4c), with 81% of Tel1 peaks overlapping Spo11 peaks (p < 10^-15), 59% overlapping with DSB hotspots (p < 10^-16), and 47% with axis sites (p < 10^-13)(Fig 2c). Tel1 peaks not colocalizing with any others features were rare (7%, Fig 2c), and of lower intensity (p < 0.0001, S5a Fig). In agreement with previous work performed in a wild-type background [39], DSBs were depleted in axis regions (p < 10^-11, S4d Fig) and hotspots overlapping axis sites were weaker than those overlapping Tel1 or Spo11 peaks (P < 0.0001, S5b Fig). Finally, DSB hotspots not associated with any features were of significantly low strength (P < 0.0001, S5b Fig), and visual examination showed that those hotspots often overlapped with a Tel1 signal too weak to be identified as such with our peak-calling parameters.

Tel1 association with meiotic DSB hotspots is dependent on Spo11 catalytic activity

Mre11 is a phosphodiesterase that functions as part of the Mre11/Rad50/Xrs2 (MRX) complex, which acts as a key sensor of DSBs and facilitates DNA repair through its coordinated nuclease activities [45]. While involved in the resection step of meiotic DSB ends [46], Mre11 interacts with DSB hotspots independently of DSB formation [47,48] and is required for DSB formation itself [49,50]. To ask if Tel1 association with DSB hotspots strictly depends on Spo11-DSBs, or if Tel1 may have functions upstream DSB formation (like Mre11 does), we tested its chromatin binding in the catalytically dead spo11-Y135F mutant. Because of the precocious cell-cycle progression caused by the absence of DSBs in the sae2Δ spo11-Y135F mutant, and Tel1 reduced protein abundance in this background (S1b and S1c Fig), ChIP experiments were performed in a ndt80Δ mutant background, where the cells arrest in late prophase I in a checkpoint-independent manner [51,52]. In this background, where Tel1 protein level remained elevated at 6 h (S6a and S6b Fig), Tel1 recruitment to ATF2, ARE1 and GAT1 hotspots was significantly reduced in the absence of Spo11 catalytic activity (Fig 2f). Thus, in contrast to Mre11, recruitment of Tel1 to meiotic DSB hotspots is dependent on the catalytic activity of Spo11.

Differential association of Tel1 and Spo11 with DSB hotspots and axis regions

Our above analysis demonstrates that Tel1 peaks colocalize with Spo11 not only on DSB hotspots but also on the chromosomes axis regions (Fig 2b-2e). In order to improve our mechanistic understanding of meiotic DSB formation, we compared Tel1 and Spo11 ChIP-seq signal from the 6 h time-point (n = 2) at DSB hotspots and axis sites. Tel1 normalized signal at DSB hotspots was significantly higher than at axis sites (Fig 3a) and the strength of Tel1 signal correlated strongly with CC-seq signals (R² = 0.72, pearson = 0.85, Figs 3b and S7a). Thus, in the sae2Δ background, DSB hotspot strength is a major factor determining Tel1 association to DSB hotspots, suggesting a uniform binding affinity to any meiotic DSB formed, something we did not measure by ChIP-seq in the wild-type background. Heatmaps centered on DSB hotspots mid-points further confirmed the positive correlation between Tel1 peaks and DSB hotspots strength (S8a Fig). Interestingly, Tel1 signal strength was not well correlated with the strength of axis sites as defined by Hop1 ChIP-seq read counts [53] (R² = 0.06, pearson = 0.24, Figs 3c, S7b and S8b). On the contrary, we observed a broader and stronger Spo11 signal at axis sites compared to DSB hotspots (Fig 3d), in agreement with the self-assembly of the DSB machinery (including Spo11 dimers) on the chromosome axis before and independently of DSB formation [54–56]. Accordingly, Spo11 signal strength correlated better with axis sites (R² = 0.41, pearson = 0.64, Figs 3f, S7b, and S8b) than with DSB hotspots (R² = 0.27, pearson = 0.51, Figs 3e, S7a, and S8a), consistent with DSB independent, non-covalent interactions of Spo11 with DSB hotspots being predominantly detected when performing ChIP-seq of Spo11 with formaldehyde cross-linking [41]. Nevertheless, the strongest DSB hotspots (9^th percentile, S7a Fig lower panel, S8a Fig) still exhibited a significantly higher Spo11 signal (p < 10^-4) relative to weaker DSB hotspots. To note, Spo11 ChIP-seq signal-to-noise at DSB hotspots was largely increased in the sae2Δ background compared with experiments performed across time of DSB formation (3h to 5h) in a wild-type background [39] (S8a Fig). Thus, likewise Tel1, transient interactions of Spo11 with DSB hotspots are enhanced and preserved in the sae2Δ background where HR is halted at the DSB stage. On the contrary, Spo11 recruitment to axis sites was not impacted by the absence of resection (S7b and S8b Figs, compare sae2Δ signal-to-noise at 6h (pearson = 0.64) with SAE2 + 3h (pearson = 0.78), once again in agreement with Spo11 assembly on the chromosome axis before and independently of DSB formation and resection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Flag-Tel1 association with axis sites is dependent on DSB formation.

a, FLAG-Tel1 read densities (RPKM per 20-bp bins) ±5 kb from DSBs (blue line) and axis (grey line) midpoint, identified respectively by CC-seq(43), and by the intersect of Rec8, Red1 and Hop1 ChIP-seq (S4a Fig) [53]. b, Scatter plot between Tel1 signals (mean RPKM per 20-bp bins) quantified over 600pb regions centered on hotspots (93% of hotspots are < 600 bp), and DSB signals (CC-seq - HPM: hits per million mapped reads). c, Scatter plot between Tel1 signals (mean RPKM per 20-bp bins) quantified over 2kb regions centered on axis midpoint, and Hop1 signals (see methods - Read density maps). d, Spo11-FLAG read densities (RPKM per 20-bp bins) ±5 kb from DSBs (red line) and axis (yellow line) midpoint. e, Scatter plot between Spo11 signals (mean RPKM per 20-bp bins) quantified over 600pb regions centered on hotspots, and DSB signals (CC-seq - HPM). f, Scatter plot between Spo11 signals (mean RPKM per 20-bp bins) quantified over 2kb regions centered on axis midpoint, and Hop1 signals. The coefficient of determination (R²) and the Pearson correlation coefficient (Pearson) were measured and mentioned for each Scatter plot. g, The tethered loop-axis model for DSB formation (‘TLAC’). DSB formation occurs when DSB hotspots localised in chromatin loops are tethered to axis sites [10–12]. h, Tel1 recruitment to axis sites is dependent on Spo11 catalytic activity. FLAG-Tel1 recruitment measured by ChIP-qPCR at the GRR1 and YBP1 axis sites relative to three control sites (CDC39, DRS2 and YCR013C) in sae2Δ ndt80Δand sae2Δ ndt80Δ spo11-Y135F strains. Error bars indicate SD from 3 (time 0 h) and 5 (time 6 h) independent meiotic cultures. P values are from paired ANOVA test between time 0 h and 6 h, corrected via Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli for multiple comparisons test. ns (not significant), ^*, p < 0.05; ^****, p < 0.0001.

https://doi.org/10.1371/journal.pgen.1011904.g003

Tel1 association with axis sites is dependent on Spo11 catalytic activity

The stronger correlation of Tel1 signal with DSB hotspots than with axis sites is compatible with Tel1 being directly recruited to Spo11-DSBs, and Tel1 interaction with axis sites being indirect, caused by the spatial proximity of DSBs with axis sites mediated by loop tethering [10–12] or extrusion [16] (Fig 3g). To verify this hypothesis, we performed ChIP-qPCR of FLAG-Tel1 at two axis sites, GRR1 and YBP1, sae2Δ ndt80Δ and sae2Δ ndt80Δ spo11-Y135F mutant backgrounds. Tel1 recruitment to axis sites could be detected in the sae2Δ ndt80Δ background (Fig 3h), albeit with lower efficiency compared to DSB hotspots—in agreement with differences between signals strength at DSB hotspots and axis sites already observed by ChIP-seq (Fig 3a). Importantly, no significant recruitment of Tel1 at both axis sites was measured in the spo11-Y135F mutant (Fig 3h). Thus, Tel1 association with axis sites requires Spo11-DSBs formation. Altogether, these results are consistent with a model in which Tel1 binding to the chromosomal axis is mediated by the formation of Spo11-DSBs within the context of the TLAC [10–12] or loop extrusion [16], bringing DSB hotspots, Spo11, Tel1 and the chromosome axis in close proximity.

Xrs2 C-terminus is required for Tel1 recruitment to meiotic DSBs and activation of the DNA damage checkpoint

In mammalian cells and Xenopus extracts, Nbs1 C-terminus interacts with ATM and was proposed to be necessary for ATM recruitment to DSBs and activation [33,34,57]. In S. cerevisiae, Tel1 recruitment to mitotic DSBs and activation of its kinase activity are abolished in a xrs2–11 mutant, where Xrs2 is truncated for 162 amino-acids at its C-terminus [32] (Fig 4a). We deleted the corresponding amino-acids of XRS2 in the SK1 background to determine its impact on Tel1 recruitment to meiotic DSB hotspots and on the meiotic DNA damage checkpoint. Another Xrs2 C-terminal truncation mutant (xrs2–664) was previously shown to suppress the meiotic checkpoint activated by unresected DSB ends [58]. In the sae2Δ xrs2–11 mutant, similarly to the sae2Δ tel1Δ, approximately 80% of cells progressed through the first and second meiotic divisions by 8 h showing that the meiotic DNA damage checkpoint is abrogated (Fig 4b). Thus, in meiotic cells, activation of the DNA-damage checkpoint by unprocessed DSBs requires the Xrs2 C-terminal domain, presumably because of its function in mediating Tel1 recruitment to DSBs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Tel1 association with meiotic DSBs is dependent on Xrs2 C-terminal domain.

a, Schematic representation of Tel1, Xrs2 and mutants generated in this study: xrs2-11, xrs2-K846A/F847A and tel1-ΔC. The HEAT (Huntington’s elongation factor 3, a subunit of protein phosphatase 2A, TOR1) repeats, FAT (FRAP/ATM/TRRAP) domain, FATC (FAT C-terminal) domain of Tel1 and FHA (forkhead-associated), MID (Mre11 interaction) and TID (Tel1 interaction) domains of Xrs2, are indicated. b, Meiotic progression monitored by DAPI staining. The percentage of cells with 2 (MI) or 4(MII) nuclei at indicated times after entering meiosis is indicated. About 200 cells were scored for each time point. c,d, Tel1 recruitment to meiotic hotspots is partially dependent on Xrs2. FLAG-Tel1 recruitment measured by ChIP-qPCR at the ARE1, ATF2 and GAT1 hotspots relative to three control sites (CDC39, DRS2 and YCR013C) in sae2Δ ndt80Δand sae2Δ ndt80Δ xrs2-11 strains (c) and sae2Δ ndt80Δ xrs2-K/F strains (d). Error bars indicate SD from 3 (time 0 h) and 5 (time 6 h) independent experiments. The same %IP values are presented in c and for the sae2Δ ndt80Δcontrol. P values are from paired ANOVA test between time 0h and 6h, corrected via Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli for multiple comparisons test. ns (not significant), ^***, p < 0.001; ^****, p < 0.0001. P value ns but close to the significance are indicated.

https://doi.org/10.1371/journal.pgen.1011904.g004

To test whether Tel recruitment to meiotic DSB hotspots requires Xrs2-C terminal domain, we performed ChIP-qPCR of FLAG-Tel1 at ARE1, ATF2, GAT1 and HIS4::LEU2 hotspots at 6 h in the sae2Δ xrs2–11 ndt80Δ mutant background, to prevent precocious meiosis I exit. In this background, Tel1 protein abundance was maintained at a similar level to the sae2Δ ndt80Δ control (S6a and S6b Fig). Importantly, no significant enrichment of Tel1 to DSBs hotspots was measured in the sae2Δ ndt80Δ xrs2–11 mutant (Fig 4c). We thus conclude that Xrs2 C-terminus mediates Tel1 recruitment to meiotic DSBs. Collectively, these results demonstrate that Xrs2 C-terminal domain is required for Tel1 recruitment to meiotic DSBs, which is in turn essential for activation of the meiotic DNA damage checkpoint.

The ATM-binding segment of NBS1 has been narrowed down to a sequence motif termed FxF/Y located within a conserved 20 amino-acids C-terminal domain (Fig 4a) [33,59]. In S.cerevisiae, mutation of the corresponding residues (K846A/F847A) reduced Xrs2 interaction with Tel1, and partially affected telomere length [59], but its impact on Tel1 recruitment to telomeres or DSBs has not been investigated [59]. We introduced the corresponding mutations at the endogenous XRS2 locus with and without the N-terminal FLAG-tag and measured meiotic progression and Tel1 recruitment to meiotic DSB hotspots. Similarly to tel1Δ or xrs2–11 cells, the xrs2-K846A/F847A (xrs2-KF) mutant abolished the meiotic DNA-damage checkpoint induced by unresected DNA ends in the sae2Δ background (Fig 4b). No significant recruitment of FLAG-Tel1 to ATF2 and GAT1 was measured in the sae2Δ xrs2-KF ndt80Δ mutant compared to the sae2Δ ndt80Δ control (Fig 4d). To note however, we measured a weak but significant enrichment at the ARE1 and HIS4::LEU2 hotspots (Fig 4d), indicating that Tel1 may still associates−albeit weakly−with DSB ends in the sae2Δ xrs2-KF mutant background.

Given the low frequency of DSB formation at meiotic DSB hotspots in the cell population, direct comparisons between mutants is challenging. Therefore, we asked if Tel1 recruitment to an HO-induced break was similarly impacted by xrs2–11 and xrs2-KF mutants. To this end, we introduced FLAG-Tel1 and XRS2 mutants in JKM139 strain, where expression of the HO endonuclease is placed under the control of a galactose-inducible promoter leading to generation of a single DSB at the MAT locus, that cannot be repaired by HR because the homologous donor loci HML or HMR are deleted [60]. Tel1 binding was measured before, 90 min and 180 min after HO induction, in close proximity (0.15kb) to the DSB site and 10kb away (S9 Fig). In this system, as expected, Tel1 recruitment was more easily detected, with ~10 fold higher %IP values (0.35-1% at 3 h) than those obtained at strong hotspots in sae2Δ meiotic cells (0.04-0.08% at 6h). Surprisingly, while the xrs2–11 mutation totally abolished Tel1 recruitment as previously reported [32], the xrs2-K846/F847 mutation had only a modest, non-significant impact on Tel1 recruitment to HO-induced breaks (S9b Fig). Thus, unlike unresected Spo11-DSBs where Tel1 recruitment is significantly reduced in the xrs2-K846/F847 mutant, Tel1 recruitment to HO-induced breaks is only mildly affected.

Tel1 interaction with Xrs2 also requires the C-terminal half of Tel1 FATC domain [61], which is specifically conserved among PIKKs [62]. Deletion of the last nine amino-acids of the FATC domain at the extreme C-terminus of Tel1 was reported to cause defects in both DNA damage response and telomere homeostasis, and to abolish Tel1 recruitment to HO breaks [61]. We deleted this short stretch of amino-acids to make the tel1-ΔC mutant (aa 2778–2787, Fig 4a). Surprisingly, Tel1-ΔC protein level was already diminished under premeiotic conditions and failed to increase during meiosis, even in the presence of the ndt80Δ mutation (S6a and S6b Fig). Similarly to the sae2Δ tel1Δ, sae2Δ xrs2–11 and sae2Δ xrs2-KF mutants, the tel1-ΔC mutation abolished the delayed nuclear division observed in sae2∆ cells (Fig 4b) and Tel1 recruitment to meiotic DSB hotspots was markedly reduced (S6c Fig). Whether the FATC domain is directly required for Tel1 association with meiotic DSBs remains unclear, as the pronounced reduction in Tel1 binding may primarily result from the substantial decrease in Tel1 protein abundance in the tel1-ΔC mutant, thereby confounding results interpretation (see discussion).

Xrs2 C-terminal domain is required for Tel1 modulation of DSB interference

Meiotic DSB interference is controlled by Tel1 and prevents the formation of coincident DSBs on the same DNA molecule [24]. DSB interference can be measured by physical analysis: positive interference values indicate DSB interference (i.e., coincident DSBs occur less frequently than if they were independent of each other’s), values close to zero indicate no DSB interference (i.e., independence), and negative values manifest concerted DSB formation [24]. In the absence of Tel1, interference is lost and coincident cutting is observed between hotspots separated by up to 70–100kb [24]. Over medium range (for example between the widely-characterized HIS4::LEU2 hotspot and a second site that maps within the leu2::hisG locus), DSB interference is positive, and loss of of Tel1 leads to loss of interference with interference values close to zero [24]. Over short distances (for example between sites I and II within the HIS4::LEU2 locus, Fig 5a), DSB interference is also mechanistically modulated by Tel1, but measured interference value are close to zero in TEL1 cells because of an underlying clustering of DSBs(24,63). In the absence of Tel1, coincident DSBs are formed at those adjacent hotspots more frequently than if they were occurring independently of each other, leading to abundant ~2.4kb double-cuts detected on gels, and negative interference values [24,63].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. DSB interference is abolished in xrs2-11, but not in the xrs2-K846/F847 (KF) mutant.

a, Diagram of HIS4::LEU2 locus showing DSB sites I and II positions, fragment sizes and probes for assaying DSBs and double-cuts (DCs) frequencies. b, c, PstI-digested (b) or undigested (c) genomic DNA isolated from the indicated time-points after induction of meiosis was fractionated by electrophoresis, transferred to nylon membrane and hybridized with probes as indicated (Right probe: MRX2, Central probe: LEU2). P, PstI-digested parental DNA; U, uncut parental DNA. DSB signals are marked with open triangles. Double-cuts: DCs. Parts of a unique membrane were cropped for presentation purpose. d, Quantification of DSBs and double-cuts (DCs) signal in b,c (average 6 and 8 h, n = 3, mean + -SEM). e, Interference values (I = 1-(observedDC/expectedDC_exp), see methods).

https://doi.org/10.1371/journal.pgen.1011904.g005

To test if Tel1 recruitment to meiotic DSBs by Xrs2 is required to modulate interference, we prepared genomic DNA from sae2Δ, sae2Δ tel1Δ and sae2Δ xrs2–11 mutant meiotic time-courses, and measured frequencies of DSBs at sites I and II within the HIS4::LEU2 locus (Fig 5b and 5d), as well as frequencies of double-cuts between sites I and II on undigested genomic DNA (Fig 5c and 5d). Similarly to the absence of Tel1, the frequency of double-cuts was elevated in the sae2Δ xrs2–11 mutant compared to the sae2Δ control (Fig 5c and 5d), leading to negative values of DSB interference (Fig 5e). To determine if middle range interference was also dependent on Xrs2 C-terminus, we probed the interval between leu2::hisG and HIS4::LEU2 separated by 20kb on chromosome 3. Similar to the sae2Δtel1Δ mutant, double-cuts were increased between leu2::hisG and HIS4::LEU2 in the xrs2–11 mutant (S10a and S10c Fig), and calculation of interference led to an interference value close to zero (S10d Fig). Finally, in agreement with the severe reduction in Tel1 protein level in the sae2Δ tel1-ΔC mutant (S7a and S7b Fig), both DSBs and double-cuts were increased at HIS4::LEU2 in this mutant, leading to a negative value of DSB interference (S11a-S11d Fig). Altogether, both the xrs2–11 mutant and the tel1-ΔC mutant phenocopy a tel1 null, conferring defects in DSB interference.

Altogether, and at first sight, these findings support the idea that Tel1 recruitment to meiotic DSB hotspots, through its interaction with Xrs2, is required to modulate DSB interference and activate the meiotic DNA damage checkpoint. Surprisingly however, double-cuts were barely detectable in the sae2Δ xrsK846A/F847A mutant (Figs 5a-5d, S10a and S10c), and DSB interference was retained, with values similar to the sae2Δ control (Figs 5e and S10d). Therefore, while mutations in the FxF/Y motif significantly reduce Tel1 recruitment to DSB hotspots (including HIS4::LEU2, Fig 4d) and abolish the Tel1-dependent meiotic DNA damage checkpoint (Fig 4b), these residues are not required for DSB interference. This unexpected result where DSB interference requires a region in Xrs2 C-terminal domain, but not the FxF/Y motif—a motif that when mutated severely reduced but did not totally abolish Tel1 binding to meiotic DSBs—nuances the simple “reactive” model [18] where Tel1 recruitment to DSB ends via the Xrs2 C-terminus is required for DSB interference (see discussion). We propose that the low but significant binding of Tel1 at meiotic DSBs measured in the xrsK846A/F847A mutant, insufficient to sustain the DNA damage checkpoint, is sufficient to modulate interference.

Tel1 protein level varies in a resection- and meiotic progression-dependent manner

We first show that steady-state Tel1 protein slightly varies during meiosis, roughly following meiotic DSB formation and repair. We can distinguish 2 phases in the variation of Tel1 protein level: an increase concomitant with the window of DSB formation (from 2 to 4 h in a wild-type background), followed by a slight decrease when DSBs are repaired (from 4 to 6 h) and a sharper drop when cells exit meiotic prophase (between 6 and 8 h). Thus, Tel1 protein abundance aligns with that of its known meiotic substrates Rec114 and Hop1 [21,64], possibly reflecting a coordinated regulation during meiosis. In vegetative cells, Tel1 kinase activity is activated in response to DNA damage through post-translational modifications, recruitment to damage sites, and interactions with sensor complexes. To our knowledge, dynamic changes of Tel1 protein level in response to DNA damage has not been reported in vegetative cells. The mechanisms regulating Tel1 protein abundance in meiosis (transcriptional or post-transcriptional), and whether increased Tel1 level correlates with enhanced kinase activity, remain to be elucidated. Tel1 downregulation at later time-point, when cells exit meiotic prophase in the wild-type, may be caused by proteolysis through the 26S proteasome, a major degradation system during meiotic divisions [65].

In sae2Δ resection-deficient meiotic cells, Tel1 level is significantly increased compared to wild-type, and remains elevated at later time-points (8 h). Because this increase is abolished in the absence of DSB formation or in mutants that abolish the Tel1-dependent DNA damage checkpoint, we presume that both the accumulation of unresected Spo11-DNA ends and the activation of the DNA damage checkpoint contribute to steady Tel1 level in the sae2Δ background. Surprisingly, we observed a significant reduction in Tel1 protein level in the tel1-ΔC mutant compared to wild-type Tel1, whereas earlier studies reported only a slight decrease [61]. The reason for this discrepancy is unclear, but may be caused by different experimental conditions or the genetic background. Tel1-ΔC protein level does not increase during meiosis, and given the marked reduction in protein abundance, we were unable to conclude whether this domain is required for interaction with DSB hotspots. Given that the Tel1 FATC domain was previously shown to be required for Tel1 localisation to DSB ends [61], Tel1 binding to DSBs may actively promote its stability through a positive feedback mechanism, a regulatory loop that would be disrupted in the tel1-ΔC mutant.

Tel1 associates genome-wide with meiotic DSB hotspots

In order to dissect Tel1 function at meiotic DSBs and in particular in DSB interference, it was fundamental to determine in first instance whether Tel1 is recruited to meiotic DSB hotspots, and if so, whether Tel1 association to meiotic hotspots precedes, like Mre11 does [47], or requires DSB formation. In mitotic cells, Tel1 forms foci in response to γ-rays [66] and localizes to HO-induced DSBs through its interaction with the MRX complex [32]. Based on these mitotic properties, Tel1 localization to meiotic DSBs was likely and anticipated. However, while HO DSBs are formed in each cell, meiotic DSBs are formed asynchronously, in at most ~15% of the cells for the strongest hotspots at any given time, making it challenging to detect association with DSBs. Thus, although we can detect Tel1 recruitment by ChIP-qPCR in a wild-type background at strong DSB hotspots, this association is weak and transient, supporting the idea that Tel1 binds to DSB hotspots in response to DSB formation only. Consequently, correlations between Tel1 enrichment and DSB strength were obscured in our genome-wide analysis, with only the most active hotspots exhibiting detectable Tel1 recruitment.

As expected, Tel1 signal at DSB hotspots is significantly increased in the sae2Δ background at 4 h, and increased further at 6 h, consistent with the accumulation of unresected DSBs in this background, and with the preferential binding of Tel1 to “blocked” DNA ends [35,36]. Despite Tel1 recruitment in the sae2Δ background being unphysiological, the higher Tel1 signal-to-noise ratio in the sae2Δ background enables more detailed and quantitative analyses of Tel1 localization and dynamics at DSB hotspots, and demonstrates an excellent correlation between Tel1 signal strength and DSB hotspots strength, something that could not be measured in the wild-type.

In both wild-type and sae2Δ background Tel1 form a bimodal signal around DSB hotspot centres, while Spo11 signal forms a single peak on DSB hotspots. This difference can be explained by the distinct roles and recruitment mechanisms of these proteins. Spo11 single peak around DSBs hotspots reflects its direct involvement in DSB catalysis, while Tel1 recruitment to the regions flanking DSBs is consistent with its role in sensing DSBs formed by Spo11. However, and interestingly, Tel1 signal in the wild-type is broader than in the sae2Δ background (~4kb versus ~2kb, S3f Fig), suggesting that Tel1 also binds to resected single-strand DNA ends.

Although Spo11 is likewise enriched at DSB hotspots and its ChIP-seq signal is also markedly improved in the sae2Δ background, its signal remains lower than that of Tel1, and the correlation between Spo11 binding and hotspot strength is correspondingly weaker. This discrepancy is also consistent with Tel1 interaction with DSB hotspots being dependent on Spo11 catalytic activity, while Spo11 was proposed to interact non-covalently with DSBs hotspots independently of DSB formation [41]. To note, Spo11 covalent binding to DSBs in a sae2Δ background may prevents adapters ligation during DNA sequencing library preparation, and favor detection of non-covalent interactions. However, this possible technical limitation should impact similarly both Tel1 and Spo11 ChiP-seq libraries preparation.

Tel1 associates genome-wide with chromosomal axes

Meiotic DSB hotspots are located on chromatin loops, with cohesins positioned at their base, on the chromosome axes [10]. Several DSB proteins, including Spo11 itself, essential for DSB formation, are also located on chromosome axes [10,12,41,54], leading to a model where DSB hotspots sequences interact with axis sites prior to DSB formation (TLAC). This model has been proposed to be critical for DSB formation, also establishing a proximity with the homologous chromosome ensuring the interhomolog bias during DSB repair [12,13]. DSB-dependent interaction of Tel1 with axis sites provides direct support for the TLAC model or to alternative, non-exclusive models such as loop extrusion [16]. It further suggests that broken DNA ends are not immediately released from the axis when DSBs are formed. Furthermore, our ChIP-seq data reveal localization of Tel1 signal on axis regions at both 3 and 4h (Fig 1b), indicating that DNA end resection may take place within the axis-associated chromatin environment. Tel1 localization to the axis may be crucial for phosphorylation of various targets, such as Rec114 and Hop1 that are required respectively for downregulation of DSB formation [21] and for the homologous bias [64,67]. It will be particularly relevant to determine whether Tel1 interaction with the axis is required for those phosphorylation events.

While Tel1 enrichment by ChIP at DSB hotspots is significantly higher than at axis sites, Spo11 signal correlates better with axis sites. This observation is in agreement with Spo11 binding to axis sites independently of DSB formation that has been proposed [14,54], but not yet demonstrated. This observation also reinforces the “platform” model where multiple Spo11 dimers assemble with other pro-DSB factors into DNA-driven condensates and catalyze DSB formation upon loop tethering to the axes sites [55,56].

Finally, ChIP-seq analysis in the sae2Δ background at the 6-hour timepoint captures proteins associated with DSBs hotspots and axis regions at a stage when breaks remain unrepaired and unresected. DSB mapping showed a strong overlap in both the localization and strength of hotspots between wild-type and sae2Δ cells [43,63]. However, the persistence of unresected DSBs in the sae2Δ mutant leads to activation of the meiotic recombination checkpoint, blocking meiotic progression that may also in turn impact DSB formation, potentially affecting the interpretation of our results. Thus, while performing ChIP-seq in a sae2Δ is valuable for detecting otherwise transient interactions with DSBs in the wild-type, it primarily provides a static snapshot rather than reflecting the full dynamics of DSB processing and repair.

Tel1 association with meiotic DSB hotspots and Tel1-dependent DSB interference require Xrs2 C-terminal domain

MRX binding to sites of damage precedes Tel1 localization, and Tel1–Xrs2 interaction is required for Tel1 recruitment to DSBs in vegetative cells [32,33,57,66]. Tel1 binding to meiotic DSB hotspots is also dependent on Xrs2 C-terminal region, indicating that Tel1 association to Spo11-DSBs is essential for Tel1 multiple functions. In agreement with this hypothesis, both the DNA damage meiotic checkpoint and DSB interference are lost in the xrs2–11 mutant, leading to the proposal that Tel1 recruitment to meiotic DSBs is required for both functions. Interestingly, the xrs2-K846/F847 mutant behaves like a null mutation with regard to the meiotic DNA damage checkpoint, Tel1 recruitment to meiotic DSBs is significantly reduced but residual binding is observed at two of the four hotspots tested by ChIP-qPCR, and DSB interference is maintained. In vegetative cells, the xrs2-K846/F847 mutant was reported to behave almost like a null mutant regarding the DNA damage response, but to cause only a modest defect in telomeres length homeostasis [59]. Tel1 recruitment to telomeres in a xrs2-K846/F847 mutant was not investigated [59], but Xrs2 interaction with Tel1 was not completely abolished (tested with an Xrs2 peptide bearing the xrs2-K846/F847 mutation), suggesting that residual binding and recruitment of Tel1 may occur and be sufficient for some telomeric functions. In our experiments, Tel1 interaction with DSB hotspots may occur transiently in the sae2Δ xrs2-K846/F847 mutant, more efficiently at some hotspots than others, enough to modulate DSB interference, but insufficiently to activate and maintain meiotic DNA damage signaling. Interestingly the xrs2-K846/F847 mutant has only a modest, non-significant impact on Tel1 recruitment to HO-induced DSBs in vegetative condition. The differing impact of the xrs2-K846/F847 mutation on Tel1 recruitment to HO-induced DSBs versus unresected Spo11-DSBs may stem from several factors: the inherently weaker Tel1 ChIP signal at meiotic DSBs, the contrasting dynamics between the two systems (with a single HO-induced DSB compared to up to 250 Spo11-induced DSBs per meiotic cell), and the distinct processing of these breaks—HO-induced DSBs are resected, while Spo11-DSBs remain unresected in a sae2Δ background. This result indicates that resection modulates Tel1 recruitment, and that Xrs2 FxF/Y motif is more critical for Tel1 association with unresected ends, a hypothesis that would requires further investigation. Altogether, we propose that Tel1 binds transiently to meiotic DSBs in the xrs2-K846/F847 mutant, enough for catalytic activities towards specific targets and interference, but insufficiently for activation of the meiotic DNA damage checkpoint. In this scenario, Tel1-Xrs2 interaction through the FxF/Y motif would be necessary to stabilize activated Tel1 at sites of DNA damage, critical for Tel1-mediated DNA damage checkpoint activation, but not for DSB interference. In line with previous hypotheses [33], another FxF/Y motif within Xrs2 C-terminus may be required for Tel1 full recruitment and activities. To conclude, we propose that Xrs2 interaction with Tel1 at meiotic DSBs modulates Tel1 activation, with different levels of activation being required for various Tel1-dependent processes. While checkpoint activation requires robust DSBs binding (and likely strong kinase activity), DSB interference appears to be less dependent on Tel1 retention to DSBs and may also rely on kinase-independent functions.

Strains and media

All S.cerevisiae strains used in this study are derivatives of the SK1 diploid used for its efficient meiotic entry and progression. Strain genotypes are listed in S1 Table.

To generate the FLAG-Tel1 SK1 strain, a URA3 cassette was integrated upstream the TEL1 promoter. The 5’ region of 3xFLAG-TEL1 from the FLAG-Tel1 strain generated by Seidel et al.[68] was amplified by PCR and transformed in the URA3-TEL1 SK1 strain. After selection on 5-FOA plates, the FLAG-tag insertion was verified by PCR screening and sequencing.

Strains with deletion and mutations were constructed by one-step allele replacement. The xrs2–11::kanMX6, xrs2-KF::KANMX6 and tel1ΔC::KANMX6 strains were made by transformation of a PCR product generated using pFA6a-kanMX6 and pFA6a-NatMX6 as template. Oligonucleotides are listed in S2 Table.

Meiotic Time courses

Pre-culture of 8 – 10 colonies were performed in YPD + AU (Yeast Extract 1%, Peptone 2%, Glucose 2%, Adenine and Uracil 0.0137%) for 30h at 30°C, 250rpm agitation. 0.3 OD/ml of cells were incubated in 200ml pre-sporulation medium (BYTA: Yeast Extract 1%, Tryptone 2%, Potassium Acetate 1%, Potassium Hydrogen Phtalate 0.05M, Antifoam 0.001%) for 16h at 30°C with 250rpm agitation. Cells were collected and washed with water, and resuspended at 2.5 OD/ml in in pre-warmed sporulation medium (SPM: 0.3% Potassium Acetate, 0.02% Raffinose, amino acids mix and Antifoam 0.001%) before incubation at 30°C with 250rpm agitation. Sample were collected at different time point and treated accordingly to the following experiments.

Meiotic progression

50μl of cells were collected at different time-points and fixed in ethanol 70%. Meiotic progression was monitored by DAPI staining and counting nuclei content by fluorescence microscopy. ~ 200cells were counted for each condition.

Southern blot

Genomic DNA was extracted from samples of synchronously sporulating cultures as previously described [69]. Briefly, spheroplasts were prepared in 1M sorbitol, 0.1M EDTA, 0.1M NaHPO4 pH7.5, 1% β-mercaptoethanol and 200µg/ml zymolyase 100T for 30min at 37°C, and lysed by incubation with SDS (0.5% final) and proteinase K (200µg/ml final) for 1hour at 60°C. DNA was isolated from protein by adding equal amount of phenol:chloroform:isoamyl alcohol. DNA was precipitated by adding one-tenth volume of 3M sodium acetate pH5.2 and an equal volume of ethanol 100%. Precipitates were washed in 70% ethanol and resuspend overnight in TE 1X at 4°C. RNA was digested by RNAse A (100µg/ml final) for 60min at 37°C and DNA was reprecipitated with sodium acetate and ethanol before suspension in 1xTE buffer.

DSBs and double cuts were detected and measured as previously described [24]. Briefly, to measure meiotic DSBs, DNA was digested with the appropriate restriction enzyme (see S2 Table), whereas DCs were measured on undigested DNA. Digested and undigested DNA were fractioned in 0.7% agarose – TBE 1X gel for ~18h, transferred to nylon membrane under denaturing condition and hybridized with the appropriate probe for detection and quantification. DSBs and DCs level were quantified as previously described [24] using the Fiji Software.

Pulse Field Gel Electrophoresis (PFGE)

DNA was prepared in plug as previously described [70]. Plugs were sealed in 1% agarose – 0.5X TBE gel, and chromosomes were fractioned using a CHEF-DRIII PFGE system (Bio-Rad) using the following program: 14°C; 6 V/cm; switch angle 120°; switch time of 30 – 30s for 3 h and time 3 – 6s for 20 h for detecting double-cuts (FRM2 probe), and 14°C; 6 V/cm; switch time of 20 – 50s for 28h for quantifying DSBs at HIS4::LEU2 (CHA1 probe). DNA was transferred to nylon membrane under denaturing condition and hybridized with the FRM2 or CHA1 probe (see S2 Table). DSBs and DCs level were quantified by ImageJ.

Signal quantification and DSB interference

DSB interference values between DSB hotspots I and II at HIS4::LEU2 were calculated as previously described [24]. Briefly, double-cuts (DCobs, “observed” DC) were quantified using a central probe (LEU2) measuring double-cut frequency between the two hotspots. This signal was multiplied by three to take in account that this probe hybridises to three parental genomic locations (HIS4::LEU2, leu2::hisG, nuc1::LEU2). Expected double-cuts (DCexp) were calculated by multiplying the frequencies of DSBs measured at DSB I and II using a MRX2 probe. However, because of double-cut formation, the DSB II signal measured using the MRX2 probe is underestimated. To compensate for this loss, the DCobs signal was added to DSB II. DSB interference were calculated as followed: 1 – DCexp/ DCobs. DSB interference values between HIS4::LEU2 and Leu2::hisG were calculated as previously described [24].

Western blot

Proteins were isolated by trichloroacetic acid (TCA) precipitation method. 10ml of sporulating cells were treated with 250µl of TCA10% and mechanically lysed by Precellys agitation (4 cycles of 30sec at 6000rpm). TCA pellets were extracted from lysate, and dissolved in STE buffer (2% SDS, 0.5M Tris HCl pH8.1, 10mMEDTA, Bromophenol Blue) with heat at 90°C for 5min. After 3 min of centrifugation at 3000rpm at RT, whole cell supernatant was transferred in new tubes and stored at -20°C.

Protein samples were loaded on a 6% SDS-polyacrylamide gel and blotted to 0.45µm nitrocellulose membrane. Membranes were saturated with 5% BSA – PBST. FLAG-Tel1 was detected with an anti-FLAG mouse antibody (M2 monoclonal, Sigma, 1/3000) and revealed with anti-mouse (Dako P0447, 1/5000) HRP conjugated secondary antibodies. Signal was detected using the HCL kit (ECL Primer Western Blotting Detection Reagent, Amersham) and a Chemidoc system (Biorad). Signal were quantified with the ImageLab Software. Flag-Tel1 signal was normalized to ponceau staining before standardisation to time 0 h. For ndt80 mutant experiments, Flag-Tel1 protein extracts were treated as above except that Flag-Tel1 was detected using a goat anti-mouse IgG Dylight 800 antibody (Invitrogen, SA5–10036). Flag-Tel1 signal was normalized to Rap1 protein level using a Rabbit polyclonal Rap1-antibody (Homemade, V.GELI).

Chromatin immunoprecipitation (ChIP)

For each meiotic time point, 2 x 10⁸ cells were fixed in paraformaldehyde (1% final). FLAG-Tel1 cells were fixed for 30min, while FLAG-Spo11 strain were fixed for 15min, quenched by glycine for 5min (125mM final) and washed in TBS 1X three times. Cells were lysed using mechanical lysis (Precellys, 6 cycles at 6.5M/sec 30sec) together with a Lysis buffer (50mM HEPES-KOH, 140mM NaCl, 1mM EDTA pH8, 1% Triton X100, 0.1% Sodium Deoxycholate) complemented with protease inhibitor cocktail tablet (Roche, Ref 05 892 791 001), PMSF (1mM final), pepstatin (0.1mM final) and leupeptin (1mM final). After cell lysis, chromatin was sheared by sonication into 250 – 500pb fragment (30sec ON/30sec OFF, 7 cycles, 4°C, Diagenode Bioruptor Pico). 50µl was kept aside as the input. For Flag-Tel1 ChIP, samples were incubated with 2µl of anti-FLAG antibody (M2 monoclonal, Sigma, Ref F1804) at 4°C for 2 h 30min on a rotating wheel then 20µl of pre-equilibrated Protein-G Dynabeads (Invitrogen, 1003D) were adding and incubated for an additional 2h30min at 4°C on a rotating wheel. IP samples were washed twice with lysis buffer, twice with 500mM NaCl containing lysis buffer, twice with LiCl buffer (10mM Tris-HCl pH8, 0.25LiCl, 1mM EDTA pH8, 0.5% Triton X-100, 0.5% Sodium Deoxycholate) and once with 1XTE buffer. Chromatin was eluted from magnetic beads in elution buffer (TE 1X, 0.5% SDS) at 65°C for 10min. DNA-protein interaction was disrupted by incubation at 65°C overnight for IP and input samples. RNA was digested by RNAse A treatment (20µg total, 30min at 37°C) and proteins by proteinase K treatment (100µg total, 1hour at 42°C). DNA was purified on column (MSB Spin PCRapace, Invitek Molecular, ref: 1020220300).

To detect enrichment of proteins in proximity of HO endonuclease cut sites, cells were growth overnight in YPD at 30°C under agitation. Cells were washed once with YP-raff and incubated for an additional 5 hours in YP-raff, diluted to OD₆₀₀ = 0.1 in YP-raff and incubated for 3 generations at 30°C. At OD₆₀₀ = 0.8 galactose was added to the final concentration of 2% to induce expression of the HO endonuclease. At time 0, 90’ and 180’, 5.10⁷ cells were cross-linked with 1% formaldehyde for 15 min at RT, quenched for 5 min with 0.125 M Glycine, washed 3X with cold 1X TBS and pellets were snap-frozen in liquid nitrogen. ChIPs were performed as described above.

Quantitative PCR (qPCR) on ChIP samples

qPCR was performed on ChIP samples at control sites (within CDC39, MRE11, DRS2, YCR013C), DSB hotspots sites (ARE1, ATF2, GAT1) and axis sites (YBP1, GRR1) (see S2 Table) using the TB Green Premix Ex Taq II (Tli RNAse H Plus, Takara). Each sample and input were quantified relatively to a standard curve. We calculated the percent IP value (%IP) by dividing the IP signal by the input signal. We take in account the dilution of the input for qPCR analysis (20 fold), and the difference of volume between the amount of input versus the one used for IP (10 fold) to calculate the real %IP.

ChIP-seq

Tel1 and Spo11 ChIP-seq were performed following the ChIP-qPCR protocol with some modifications. For both FLAG-Tel1 and FLAG-Spo11 strain, cells were fixed in paraformaldehyde for 30min. Protein G magnetic beads were washed (0.1M NaHPO4 pH7.5, 0.01%Tween20, pH8.2) and saturated with BSA (0.1M NaP, 0.01% Tween20, 5mg/ml BSA) for 3 h at 4°C. Anti-FLAG antibody (M2 Monoclonal, Sigma) was pre-incubated with saturated-magnetic beads for 4 h at 4°C. Cell lysis was performed as for the ChIP-qPCR protocol. Chromatin shearing was performed in Diagenode tube 15ml with beads (900 – 1000mg) by 7 cycles of sonication (30sec ON/30sec OFF, 4°C). Chromatin was immunoprecipitated with antibody-protein G mix by overnight incubation at 4°C. Samples were washed as for the ChIP protocol, and DNA was eluted with IPure v2 Diagenode kit. DNA quantity was measured by QuBit and fragmentation quality was controlled by Agilent chip (Bio-analyser). DNA was sequenced with a NovaSeq sequencer (Illumina) using a 100nt paired-end protocol (TruSeq) at the Institut Curie sequencing facility (Paris, France).

Alignment and mapping

Sequencing quality control was determined using FastQC tool (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and aggregated across samples using MultiQC. Reads with Phred quality score less than 30 were filtered out. For Tel1 and Spo11 ChIP-seq in the sae2Δ mutant, paired-end reads were mapped to Cer3H4L2 saccharomyces cerevisiae genome (PMID: 31649282) using Bowtie2 (v2.3.4.3) [71] with default parameters. To allow complete comparison between WT and sae2Δ background (concern the chrIII only), paired-end reads from Tel1 ChIP-seq in WT cells were also mapped to Cer3H4L2 saccharomyces cerevisiae genome (PMID: 31649282) using Bowtie2 (v2.3.4.3) [71] with default parameters.

Only mapped reads were filtered in using samtools view (--exclude-flags 4)⁶⁷. Duplicate tags were removed using Picard Tools MarkDuplicates and mapped tags were processed for further analysis. For all subsequent analyses, mitochondrial genome, plasmid genome, and ribosomal DNA (rDNA) loci were blacklisted and removed. For visualization purpose, mapped tags were converted into BedGraph and normalized by their library size (FPKM) using bamCoverage [72].

Wig files from Hop1, Rec8 and Red1 ChIP-seq data in the WT background were obtained from published study (PMID: 26258962; GSE69231) transformed in bedGraph file with wig2bed (BEDOPS v2.4.41) and converted into the Cer3H4L2 by adding 3078 bp from the position 65684 and 1173 bp bp from the position 95042 on the chrIII.

Spo11 ChIP-seq reads from GSE52863 (PMID: 24635992 - WT background; 3h, 4h and 5h after meiotic induction) and the matched untagged sequencing reads were aligned to the yeast Cer3H4L2 genome (PMID: 31649282) using bowtie2 and duplicate sequencing reads were removed by samtools markdup. BedGraph files for each time point were created using bamCoverage (FPKM) [72].

Mre11 normalized log2 ratio from the ChIP-on-Chip published study (PMID: 21822291 - WT background; 3 h after meiotic induction; GSE30072) was transformed in bedGraph file and converted into the Cer3H4L2 by adding 3078 bp from the position 65684 and 1173 bp bp from the position 95042 on the chrIII.

For subsequent analysis, visualization, and read density maps (Figs 1, 2a–2e, 3a–3f, S2, S3e–S3f, S5, S7 and S8), signal (normalized counts) at each sample was first subtracted from the corresponding untagged (for the sae2Δ) or input signal (for the wild-type), and replicates were merged together when existing. For Tel1 and Spo11 ChIP-seq in the sae2Δ mutant, the signal at t 0h was further subtracted from the corresponding sample signal at t 6h before merging the replicates together.

We used Integrative Genome Viewer (IGV) v2.9.4 [73,74] or Integrative Genomics Viewer (IGB – v 10.0.1) [75] for visualization.

Correlation between biological replicates

The degree of correlation between biological replicates (S3c and S3d Fig) was visualized with a scatter plot and subsequent computation of Pearson’s correlation coefficient (FP) and Spearman’s rank correlation coefficient (ρ) using the average signals in individual 100-bp non-overlapping genomic windows.

Chromosome profiles

For S3b Fig, Cer3H4L2 mapped tags were counted within 1000 bp bins (beedtools makewindows -w 1000) [76] throughout the genome using featureCounts (v1.6.4) [77]. Signal values were then normalized by library size, smoothened by applying Hann function with different window sizes (2.5, 5, 10,25 and 50kb), and displayed using homemade R scripts.

Peak calling

To ensure greater precision in peak selection, Tel1 ChIP-seq binding sites in the sae2Δ background at 6h were determined with Macs2 callpeak [78] using 3 different p-value cut-off (--bw 300 --nomodel --keep-dup 1 -p [0.05, 1e-2 and 1e-4]) and untagged reads as the control. NarrowPeak from duplicate samples were merged for the three p-value conditions and only the smaller region from overlapping peaks was considering. For correlation analysis, regions with distance <350 bp were merged and peaks of the FLAG-Tel1 sae2Δ t-0h peaks (“hyper-ChIPable” regions) were substracted from the FLAG-Tel1 sae2Δ t-6h peaks list. Spo11 ChIP-seq binding sites, in the sae2Δ background, were determined with Macs2 bdgbroadcall. As above, peaks of the FLAG-Spo11 sae2Δ t-0h peaks were substracted from the FLAG-Spo11 sae2Δ t-6h peaks list. Red1, Hop1 and Rec8 ChIP-seq binding sites in the WT background were determined with Macs2 bdgbroadcall (-c 1.301 -l 100 -g 100). Axis sites were obtained by cross-referencing Red1, Hop1 and Rec8 binding sites at 1 bp minimum, considering sites that overlap at least two out of the 3 proteins binding sites (S4a Fig).

High-ChIPable regions

The three-negative control from FLAG-Tel1 sae2Δ, FLAG-Spo11 sae2Δ and sae2Δ strains (see S3a Fig) were analyzed with the Macs2 software [78] to call ChIP-seq broad peaks, subsequently merged to create the blacklist of high-ChIPable regions. 550 independent regions were found all over the genome.

For all peak calls described above, positions that overlap a hyper-ChIPable regions have been removed from the overlapping peak coordinate. Peaks > 200 bp were retained.

Intersection between dataset

Intersection between each dataset were counted as the overlaps between axis, DSBs, Tel1 and Spo11 peaks at 1 bp minimum using intersectBed [77]. To test the significance of the intersection between datasets (real), we generated 10 sets of random peaks (shuffled). Each random set contained the same number of regions as in the cognate set of DSBs or Tel1 peaks and had the same length distribution, but each region segment was randomly picked with the condition that the segments did not overlap with any actual DSBs or Tel1 peaks (shuffleBed [-chrom -noOverlapping]) [77].

Venn diagrams were performed with Venn Diagram Generator (https://academo.org/demos/venn-diagram-generator/). Boxplot were generated by GraphPad Prism v8.0.2 (GraphPad Software, Boston, Massachusetts USA, www.graphpad.com)

Read density maps

Read coverage plots and heatmap were obtained by extracting 5 kb on each side of the center of DSBs or axis sites and counting the reads per 20-bp bins using annotatePeaks.pl (HOMER) [79]. The - hist option was used to generate histograms of position dependent read counts relative to the center of peaks (Figs 1e, 1f, 3a and 3d) and – ghist to generate an output profile for each peaks used to generate heatmaps (S8 Fig). The regions were sorted according to the DSBs strength (NormHPM) for S8a Fig and Hop1 counts (RPKM) for the S8b Fig, and heatmaps were created using TreeView [80]. For scatter plot of Fig 3 and boxplot S5 Fig, the mean count per 20-bp were measured on a region of 600 bp centered on hotspot region (Figs 3b, 3e and S5a) and 2000 bp centered on axis midpoint (Figs 3c, 3f and S5b). For S5 Fig, DSB or Axis peaks were partitioned by strength percentile according to DSB hotspots strength (NormHPM) and Hop1 normalized count (RPKM) respectively. For scatter plot of S2b, S2c, S2f and S2g Figs and boxplot Figs 1c, 1d and S2d in WT strain, the read count per 1500-bp were measured on a region centered on hotspot region or the top20% Hotspot. To avoid any Tel1 or Spo11 signal due to the axis region, only regions of 1500 bp that overlap the axis by less than 10% are taken into account.

Boxplot were generated by GraphPad Prism v8.0.2 (GraphPad Software, Boston, Massachusetts USA, www.graphpad.com).

Statistics

For qPCR, statistical analysis was performed on paired time point (0 h versus other time point) or unpaired (wild-type versus mutants) using ANOVA corrected via Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli for multiple comparisons test (alpha = 0.05).

For boxplot (Figs 1c, 1d and S2d), statistical analysis was performed on unpaired data using Kruskal-Wallis test between percentile 0.2 and all the other percentile, corrected via Dunn’s for multiple comparisons test.

The significance of the difference between random (shuffled) and actual region (real) was performed using Mann Whitney test.

Error bars represent the mean ± SD of the indicated number of independent experiments. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 and ****p ≤ 0.0001