Meiotic prophase length modulates Tel1-dependent DNA double-strand break interference

During meiosis, genetic recombination is initiated by the formation of many DNA double-strand breaks (DSBs) catalysed by the evolutionarily conserved topoisomerase-like enzyme, Spo11, in preferred genomic sites known as hotspots. DSB formation activates the Tel1/ATM DNA damage responsive (DDR) kinase, locally inhibiting Spo11 activity in adjacent hotspots via a process known as DSB interference. Intriguingly, in S. cerevisiae, over short genomic distances (<15 kb), Spo11 activity displays characteristics of concerted activity or clustering, wherein the frequency of DSB formation in adjacent hotspots is greater than expected by chance. We have proposed that clustering is caused by a limited number of sub-chromosomal domains becoming primed for DSB formation. Here, we demonstrate that DSB clustering is abolished when meiotic prophase timing is extended via deletion of the NDT80 transcription factor. We propose that extension of meiotic prophase enables most cells, and therefore most chromosomal domains within them, to reach an equilibrium state of similar Spo11-DSB potential, reducing the impact that priming has on estimates of coincident DSB formation. Consistent with this view, genome-wide maps of Spo11-DSB formation generated in the absence of Tel1 are skewed towards regions that load pro-DSB factors early—revealing regions of preferential priming—but this effect is abolished when NDT80 is deleted. Our work highlights how the stochastic nature of Spo11-DSB formation in individual cells within the limited temporal window of meiotic prophase can cause localised DSB clustering—a phenomenon that is exacerbated in tel1Δ cells due to the dual roles that Tel1 has in DSB interference and meiotic prophase checkpoint control.


INTRODUCTION
two-fold) in the sae2∆ tel1∆ ndt80∆ triple mutant, with total (DSB I + DSB II) levels reaching 144 ~46% of total DNA (Fig 1h). Notably, DSB I signals, as measured with the MXR2 probe, 145 showed partial smearing down the gel suggesting a general increase in hotspot width, perhaps 146 caused by the increase in the frequency of hyper-localised coincident cutting by Spo11 that 147 arises within hotspots 41,62 (Fig 1d). 148 149 DSB frequencies measured around the ARE1 locus were both increased by deletion of NDT80 150 in the sae2∆ tel1∆ background, reaching levels higher than those previously reported for when 151 Ndt80 is present 56 (Fig S2b-f). At the YCR061W locus, the effect at individual hotspots varied 152 (Fig S3b-f). Hotspot 'N' was increased by TEL1 deletion, but not further by NDT80 deletion 153 (Fig S3d), whereas hotspot 'Q' was increased more by NDT80 deletion than by TEL1 deletion 154 (Fig S3f). Notably, deletion of TEL1 leads to the formation of a previously undetectable hotspot 155 "O" (Fig S3c) flanking the YCR061WI probe (also detected in genome-wide CC-seq 39 maps 156 of Spo11 DSBs; Fig S3a), and this hotspot was increased a further two-fold upon NDT80 HIS4 probes on the left and right of the locus respectively (Fig 1c-h; see Extended methods, 185 "Calculation of DSB interference" for full description). Because DSBs and DCs accumulated 186 over time (Fig 1f-g,j), to simplify analysis and reduce sampling error, the 6 and 8 hour time 187 points were averaged, and then this average value was calculated across a number of 188 independent experimental repeats made in both the NDT80+ control (n=6) and ndt80∆ mutant 189 (n=5) backgrounds (Fig 1h,k). 190 191 Aggregation of additional observations made in this study with prior measurements 56 192 reinforced the prior conclusion: that is, in the presence of Tel1, a similar frequency of DCs 193 were observed to those that were expected (Fig 1l), suggesting no interference over this short 194 distance even though Tel1 is present (and thus the formation of DCs inhibited; Fig 1m). TEL1 195 deletion led to a ~1.5-fold increase in the frequency of single DSBs (Fig 1h), but a 196 disproportionate ~10-fold increase in the frequency of DCs (Fig 1k)-demonstrating not just 197 Tel1's inhibitory role, but also how observed DCs then exceed by ~3-fold those expected by 198 chance alone (Fig 1l), leading to a negative interference calculation (Fig 1m). 199 200 Remarkably, in the presence of Tel1-but now in the absence of Ndt80-although single DSB 201 frequency increased a small amount (Fig 1h), DC frequency was unchanged (Fig 1k), and at 202 a lower frequency than expected (Fig 1l), leading to positive interference (Fig 1m). Moreover,203 in the absence of Tel1 and Ndt80, single DSB frequencies increased further (Fig 1h), but 204 without any increase in DCs relative to tel1∆ (Fig 1k), leading observed and expected 205 frequencies of DCs to be similar (Fig 1l), and therefore, an absence of interference (Fig 1m).

207
To test whether similar effects were observed elsewhere, we also measured DSB and DC 208 formation between the three main hotspots (labelled 'E', 'F', and 'I') flanking the BUD23-ARE1 209 locus on chromosome III 56 (Fig S2a). Although other minor DSBs (and thus DCs) are also 210 visible, their low cutting frequency and the relatively high lane background precluded their 211 accurate measurement in this study. Deletion of NDT80 increased single DSB frequencies in 212 both the presence and the absence of Tel1 (Fig S2b-f) but without any major changes in DC 213 frequencies relative to the large effect caused by TEL1 deletion (Fig S2g-j). In agreement with 214 the measurements made at HIS4::LEU2 above, these effects altered DSB interference (Fig Importantly, the concomitant increases in both single DSB frequencies (Fig 2c) and DC 254 frequencies (Fig 2f) upon NDT80 deletion, generated no changes in the ratios of observed to 255 expected DC formation (Fig 2g), and as a result no change in measurements of DSB 256 interference between these loci (Fig 2h). 257 258 In agreement with these findings, measuring DSB and DC frequencies and DSB interference 259 between the ARE1 and YCR061W loci, separated by ~14 kb (Fig S4a), demonstrated that 260 although DC frequencies were modestly increased upon NDT80 deletion (both in the presence 261 and absence of Tel1; Fig S4b), this did not change the measurements of DSB interference 262 (Fig S4c).

264
Taken together, these observations underscore the view that whilst Tel1-dependent DSB 265 interference acts over both short and medium scales, the observation of negative interference 266 over short distances in tel1∆ mutants appears uniquely influenced by NDT80 status.

268
Deletion of NDT80 alters the genome-wide DSB distribution in the absence of Tel1 269 We recently developed covalent-complex sequencing (CC-seq), a high-resolution and 270 genome-wide sequencing method to detect and characterise the covalent Spo11-DSB 271 intermediates that accumulate in meiosis when SAE2 is deleted 39 (Fig 3a). Based on the 272 observations made above, we next sought to use CC-seq to explore the effects that Ndt80 273 and Tel1 may have on DSB formation at a genome-wide scale.

275
Taking the lead from prior work that mapped the transient Spo11-oligo intermediates liberated 276 from Spo11-DSB ends in wild-type cells 38,60 , we first simplified the data into a set of ~3400 277 Spo11-DSB hotspots characterised by their local enrichment of reads (Fig S5a). The locations 278 of these hotspots overlapped well (>85% congruence) with prior hotspot positions called from 279 Spo11-oligo data in wild-type cells 38,60 (Fig S5b-c). Residual differences are likely caused by 280 a combination of methodological (Spo11-oligo seq vs CC-seq) and real (SAE2+ vs sae2∆ 281 genotypes, and presence/absence of tags on Spo11 itself) effects, and were 282 disproportionately associated with weaker hotspots (Fig S5d-f). Notably, only a minority 283 (32/3473; <1%) of hotspots called from the CC-seq data were also present in a sae2∆ ndt80∆ 284 spo11-Y135F control sample in which the catalytic activity of Spo11 is disabled (Fig S5g), and 285 these were all weak (Fig S5h), underscoring the utility of CC-seq for measuring bona fide  Fig 3b), but slightly less so in the sae2∆ tel1∆ and sae2∆ tel1∆ ndt80∆ samples (Pearson R=0.92 ; Fig 3c), suggesting again that the impact that Ndt80 has is more 291 significant in the absence of Tel1. As expected from the highly correlated Pearson values, at 292 broad scale, hotspot-strength distributions were visually almost indistinguishable between the 293 four datasets when plotted along a representative chromosome (chromosome VII ; Fig 3d). 294 However, plotting a smoothed ratio of hotspot strength revealed spatial patterns influenced by 295 the presence of Ndt80 that were much stronger in the absence of Tel1 (Fig 3e; Fig S6). 296 297 To characterise these effects on each chromosome, ratios of hotspot strengths ±NDT80 were 298 represented as heatmaps binned at 50 kb scale (Fig 3f-g), and plotted centred on the 299 centromere consistent with prior representations 83 . Effects of Ndt80 in the presence of Tel1 300 were relatively modest and did not display a clear spatial pattern with respect to chromosome 301 features such as telomeres and the centromere (Fig 3f). By contrast, in the absence of Tel1, 302 the presence of Ndt80 led to a dramatic enrichment of Spo11-DSB signal in centromere-303 proximal regions-notably encompassing the entirety of the three shortest chromosomes (I, 304 III, and VI), and the entire region of chromosome XII left of the rDNA array (Fig 3g). These 305 observations suggest that NDT80 deletion in the tel1∆ background promotes genome-wide 306 redistribution of Spo11 activity, generating a more uniform pattern-and preventing bulk 307 enrichment of Spo11 activity in these largely centromere-proximal regions.

309
To understand how this pattern of enrichment might be explained by other features of Spo11-310 DSB formation, we compared our fold ratios ±NDT80 in the tel1∆ background to the time that Rec114-an essential pro-DSB factor-associates with meiotic chromosomes 83 (Fig 3h). 312 Remarkably, regions of Spo11-DSB formation that are enriched in the sae2∆ tel1∆ strain are 313 similar to regions that load Rec114 early (Fig 3g-h). Given that Rec114 is essential for Spo11-314 DSB formation 3,12,20,21,84,85 , we propose that in the shorter prophase experienced by sae2∆ 315 tel1∆ cells (data above), DSB formation is enhanced in the subset of chromosome domains in 316 which Rec114 first associates. We further propose that it is this effect that drives the negative 317 DSB interference (DSB clustering) that we have measured over short distances 56 .

319
Tel1 activity patterns DSB hotspot strength across the genome 320 We thus next sought to take advantage of the NDT80 deletion-induced meiotic prophase arrest 321 to characterise the specific genome-wide effects caused by loss of Tel1-dependent DSB 322 interference (Fig 4). Previous analysis of Spo11-oligo patterns in the presence and absence 323 of Tel1 revealed spatially localised correlated changes in DSB hotspot strengths that decayed 324 with distance (adjacent hotspots either went up or down in a correlated manner), with local 325 inhibition also patterned locally by the insertion of strong DSB hotspots 60 . Globally, however, 326 DSB hotspot strengths measured using Spo11-oligo data in the presence and absence of Tel1 are highly correlated (R=0.97 ; Fig 4a), suggesting relatively weak global effects. By contrast, 328 deletion of TEL1 affected CC-seq (sae2∆ background) hotspot strengths more severely 329 (R=0.91 in NDT80+ ; Fig 4b), likely driven at least in part by the tel1∆-dependent alterations in 330 prophase length described above. Nevertheless, even in the absence of Ndt80, CC-seq DSB 331 hotspot strengths ±TEL1 were less similar in the CC-seq sae2∆ data (R=0.94 in ndt80∆; Fig   332   4c) than in the published Spo11-oligo data 60 .

334
It is important to note that in all cases, these Pearson correlation values are high, and 335 consistent with this, like with ±NDT80 comparisons, broad-scale hotspot-strength distributions 336 were almost visually indistinguishable from one another between the paired ±TEL1 dataset 337 comparisons when plotting along a representative chromosome (e.g. chromosome IV ; Fig 4d). 338 However, plotting a smoothed ratio of hotspot strengths revealed a very different picture ( Fig   339   4e; Fig S7a). Whereas effects on Spo11-oligo hotspot strength ±TEL1 were relatively weak ). The most dramatic effects were often observed towards the ends of many 345 chromosomes-where the distribution of DSBs was enhanced in the presence of Tel1, as was 346 the relative proportion of DSBs forming on the entirety of chromosome 12 (Fig 4f-g).

348
We propose that these chromosome-specific effects are the genome-wide signature of Tel1-349 dependent DSB interference-manifesting as spatially patterned changes in the population-350 average frequency of Spo11 DSB formation within hotspots (LLR and MJN; manuscript in 351 preparation). Importantly, although such effects are influenced to some degree by prophase 352 timing, they are in fact largely unaltered by changes in the length of meiotic prophase.

355
We previously established in S. cerevisiae that Spo11-DSBs are subject to distance-356 dependent interference via activation of the DNA-damage-responsive kinase, Tel1-part of a 357 negative-regulatory pathway that appears to be conserved in mice, flies and plants 63-68 . 358 Critically, due to its involvement in the DNA damage response, Tel1 has at least two 359 overlapping roles: DSB interference and regulation of meiotic prophase kinetics, but our 360 understanding of how these two roles intersected was unclear and largely unexplored.

362
To investigate the relationship between these two roles of Tel1, we have measured the 363 frequency of single and coincident Spo11-DSB formation arising at adjacent hotspots in the presence and absence of both Tel1 and Ndt80, the latter of which is a critical transcription 365 factor required for exit from meiotic prophase 81 . Importantly, deletion of NDT80 causes cells 366 to arrest in late meiotic prophase irrespective of the strength of checkpoint activation. In order 367 to estimate total Spo11-DSB formation potential, we have utilised strains in which Mre11-368 dependent nucleolytic processing of Spo11-capped DSB ends is abolished via deletion of the 369 activator, SAE2 29,31 , permitting total Spo11-DSB levels to accumulate.

371
When considering total Spo11-DSB levels, both TEL1 and NDT80 deletion independently 372 increased Spo11 activity, with the greatest DSB frequency arising when both genes were 373 deleted (Fig 1f-h; Fig S2d-f; Fig S3d- of Spo11-DSB formation must be rate-limiting. If it is the activation step that limits total DSB 400 potential, then the frequency of active domains must be limiting, meaning therefore that active domains vary in their chromosomal location across the cell population. These effects will 402 create a heterogeneous mixture of active and inactive subpopulations when considering any 403 given chromosomal region (Fig 5a). Critically, such heterogeneity will give rise to lower-than-404 expected measurements of DSB interference (Fig 5a, bottom).

406
In wild-type cells, the formation of such subdomains is likely to help disperse a limited amount 407 of DSB potential across the genome. However, in tel1∆, the absence of localised negative 408 regulation will permit efficient coincident cutting by Spo11 at all DSB hotspots located within 409 any local region of activation-detected as negative interference 56 (Fig 5a, bottom).  Furthermore, because of the dual role of Tel1 in both DSB interference and checkpoint 419 activation, loss of Tel1 leads to an accelerated exit from meiotic prophase (Fig 1b), 420 presumably due to a relatively earlier activation of Ndt80 and subsequent down-regulation of 421 Spo11-DSB formation 59 . Such effects of Tel1 loss are likely to be more significant in the sae2∆ 422 background, where DSB-dependent checkpoint activation is dependent on Tel1 70 , which is not 423 the case under conditions where Spo11 has been removed from DSB ends and ssDNA 424 resection has initiated 60,70 . Thus, the differential prophase timing that arises ±TEL1 in the 425 sae2∆ background potentially exacerbates the subpopulation effect. Our observations suggest 426 that by extending the length of prophase, NDT80 deletion can be used to limit effects caused 427 by differential prophase kinetics, homogenizing the DSB potential across the entire genome 428 and cell population (Fig 5b). We contend that this is particularly important when deleting TEL1, 429 or other factors, that influence the meiotic prophase checkpoint.

431
A key feature of our observations is that negative interference (and its abolition upon NDT80 432 deletion) was only detected over short distances-behaviour that is consistent with zones of We have also explored the changes in genome-wide patterns of Spo11-DSB formation that 440 arise in the presence and absence of Tel1, and how these differences are affected by NDT80 441 deletion (Fig 3). Importantly-and consistent with our hypothesis that accelerated exit from 442 prophase in sae2∆ tel1∆ accentuates the impact of subpopulation domains in which Spo11 is 443 active-deletion of NDT80 led to a much stronger change in the genome-wide pattern of DSB 444 formation in sae2∆ tel1∆ cells than in sae2∆ cells (Fig 3f-g). Such a difference is expected 445 due to the more limited temporal window of meiotic prophase that otherwise arises in the 446 absence of Tel1.   A second finding that emerges from our genome-wide studies, is that despite the influence 464 that temporal changes in meiotic prophase timing has on Spo11-DSB distribution, deletion of 465 TEL1 itself elicits a much stronger effect that is detectable both in the presence and absence 466 of Ndt80 (Fig 4f-g). We hypothesise that these strong Tel1-dependent changes are the  A feature-but also a limitation-of our analytical methods is the reliance on SAE2 deletion to 471 permit Spo11-DSB and Spo11 double-cut signals to accumulate without repair. On the one 472 hand, sae2∆ enables us to study mechanisms of DSB interference in the absence of other 473 regulatory pathways that are dependent upon and triggered after Spo11 removal (i.e.

474
homologue engagement 77,90 ), and which may otherwise obscure Tel1's influence. However, we cannot exclude that the accumulation of unrepaired Spo11 DSBs itself influences how the 476 system behaves, and as such, all observations must be interpreted with this in mind.

478
Looking more broadly, the regulatory feedback mechanisms discussed here are likely to 479 ensure that cells stay in a DSB-permissive state only for as long as needed-limiting the level 480 of DSB formation, and therefore recombination, required to facilitate accurate chromosome

511
Yeast strains 512 All the Saccharomyces cerevisiae yeast strains used in this study are in the SK1 background 513 as described in Table S1, and derived using standard techniques. Strains contained the

531
Samples were centrifuged at 3000 x g for 4 minutes, supernatant was discarded and pellet 532 resuspended in 2 mL 50 mM EDTA, centrifuged again for 1 minute at 3000 x g, supernatant 533 discarded and pellet stored at -20 °C until use. For Spo11 CC-seq, 50 mL of culture was taken 534 at t = 6 hours. Samples were centrifuged at 3000 x g for 5 minutes, supernatant discarded and 535 pellet frozen at -20 °C until use. For FACS, 200 uL of culture was taken at t = 0, 2, 4, 6 and 8 536 hours after inducing meiosis, samples were centrifuged at 16,000 x g for 1 minute, supernatant 537 discarded, fixed in 1mL of 70% EtOH and stored at 4 °C until use. For DAPI staining, 195 uL 538 of culture was taken at t = 3, 4, 5, 5.5, 6, 7, 8, 9 and 10 hours after inducing meiosis. Cells 539 were fixed in 450 uL of 100% EtOH and stored at -20 °C until use. Samples were centrifuged at room temperature, 16,000 x g for 1 minute. Supernatant was 543 aspirated, pellet resuspended in 500 uL 10 mM Tris HCl pH 8.0 / 15 mM NaCl / 10 mM EDTA 544 pH 8.0 / 1 mg/mL RNase A and incubated at 37°C for 2 hours at 800 rpm on a Eppendorf 545 Thermomixer. Samples were then centrifuged at 16,000 x g for 1 minute, supernatant 546 aspirated, pellets resuspended in 100 uL of 1 mg/mL Proteinase K + 50 mM Tris HCl pH 8.0 and incubated at 50°C for 30 minutes at 800 rpm on a Eppendorf Thermomixer. Samples were 548 centrifuged and supernatant aspirated. Pellets were washed in 1 mL 1M Tris-HCl pH 8.0 and 549 then resuspended in 1 mL 50 mM Tris-HCl pH 8.0 + 1 uM Sytox green. Samples were stored 550 overnight at 4 °C and then sonicated at 20% amplitude for 12-14 seconds before being sorted 551 by flow cytometry (Accuri™ Flow Cytometers). shaking separated by a 5-minute rest and followed by a 5 minutes centrifugation at 14,000 568 rpm. DNA and RNA were extracted from 450 uL of the aqueous phase and precipitated with 569 45 uL of 3 M NaAc pH 5.2 and 500 uL of 100% EtOH, centrifuged at 14,000 rpm for 1 minute, 570 aspirated and washed with 1 mL 70% EtOH, pulsed down, air dried for 10 minutes and 571 resuspended in 450 uL of 1x TE (10 mM Tris / 1 mM EDTA pH 7.5) overnight at 4 °C. RNA 572 was digested with 50 uL of 1 mg/mL RNase A (10 mg/ml stock) for 1 hour at 37 °C. DNA was 573 precipitated by addition of 50 uL of NaAc pH 5.2 and 1 mL of 100% EtOH, mixed by inversion 574 and centrifuged for 1 min at 14,000 rpm. DNA was washed with 1 mL 70% EtOH, pulsed down, 575 air dried for 10 minutes, dissolved in 200 uL of 1x TE (10 mM Tris / 1 mM EDTA pH 7.5 prep 576 room solution) overnight at 4 °C. To measure the frequency of DSBs (single-cuts) gDNA was 577 digested with a restriction enzyme (as described in Table S2, digestion column). For 20 uL of 578 gDNA, 6 uL of H2O, 3 uL of enzyme buffer and 1 uL of enzyme was added and incubated 579 overnight at 37 °C. To quantify double DSB events (double-cuts), gDNA was left undigested 580 (Table S2). 581 582 583 DSB analysis by Southern blot 585 0.7% or 0.8% agarose gels were prepared for digested and undigested gDNA samples, 586 respectively. The gel was mixed with 125 uL EtBr (0.1 mg/mL) and allowed to set for 1 hour 587 at room temperature. 20 uL of digested sample + 1x loading dye or 10 uL of gDNA + 10 uL 588 water and 1x loading dye was loaded on wells. For the ladder, 10 uL of Lambda BstE II-digest 589 was used as ladder. DNA was separated at 45-50 V for 15-19 hours. Gels were imaged using 590 the Syngene InGenius bioimaging system. DNA was nicked by exposure to 1800 J/m2 UV in 591 a Stratalinker. Afterwards, the gel was soaked in denaturing solution (0.5 M NaOH, 1.5 M 592 NaCl), on a shaker for ~30 minutes. Tris-HCl pH 8 + 1% BME + RNase 10 ug/mL + water) and incubated for 2 hours at 37 °C.

603
Samples were inverted every 30 minutes. Solution 2 was aspirated, and plugs were covered 604 with 1 mL of solution 3 (0.25 M EDTA + 20 mM Tris-HCl pH 8 + 1% sodium sarcosine + 1 605 mg/mL proteinase K + water) and incubated overnight at 55 °C. Solution 3 was aspirated and 606 samples washed three times with 1 mL of 50 mM EDTA on a rotary wheel. EDTA was 607 aspirated and plugs were covered by 1 mL of storage buffer (50 mM EDTA, 50% glycerol) and 608 stored at -20°C until use. PFGE gel: 1.3% agarose gel was prepared using 150 mL of 0.5X 609 TBE (diluted from 5X TBE: 450 mM tris base + 450 mM boric acid + 10 mM EDTA pH8 + 610 water) and cooled down to 55 °C before use. The plugs were cut in half and washed in 2.5 mL 611 of 0.5X TBE on the rotary wheel for 15 minutes. The plugs were loaded in order onto the gel 612 comb and fixed with 1% agarose. A slice of mid-range PFG marker (#3425, NEB) was fixed 613 onto the first and last gel combs. Once set (10 min at room temperature), the 1.3% agarose 614 was poured covering the gel combs and allowed to solidify for 30 minutes. The gel comb was 615 removed, and wells filled with 1% agarose and allowed to set (10 min at room temperature), 616 then immersed into pre-cooled 0.5X TBE buffer for 15 minutes. For DNA fragments of ~150 617 kb, the gel was run 30-30s for 3 hours + 3-6s for 37 hours at 6 V/cm and 120 °C angle. After 618 electrophoresis, the gel was soaked in 150 mL distilled water + 7.5 uL of EtBr (0.1 mg/mL), 619 shaken for 20 minutes and then imaged using the Syngene InGenius bioimaging system. DNA was nicked by exposure to 1800 J/m2 UV in a Stratalinker, then soaked in a denaturing 621 solution (0.5 M NaOH, 1.5 M NaCl) whilst shaking for ~30 minutes.  Table S3.                 The number of biological repeats is indicated in figure legends. For Fig 1-2

Calculations of DSB interference 921
To study DSB interference between two hotspots, the observed frequency of double DSB 922 events that arise at the same molecule (observed DCs) was compared with the expected 923 frequency on the assumption of independence (expected DCs) as described in Fig S1f-j. 924 Such expected DC frequency was estimated from multiplication of the frequencies of single 925 DSB events between which DSB interference is studied. To study the strength of interference, 926 the coefficient of coincidence (CoC) was estimated by dividing the observed frequency of DCs 927 by the frequency of expected DCs and subtracting this value from 1 (Fig S1k). Positive values 928 close to 1 indicated strong interference, values close to zero indicated independence (no 929 interference) and negative values indicate concerted DSB activity (Fig S1l). DSB interference As an example, at HIS4::LEU2, the frequency of DCs between DSB I and DSB II was 934 measured with a central probe LEU2 (Fig 1c,i). The averaged observed DCs from time 6 and 935 8 hours, was then compared with the expected frequency of coincident cuts (also averaged 936 time 6 and 8 hours) obtained by multiplying the averaged frequency of DSB I (measured with 937 MXR2 probe) and DSB II (measured with HIS4 probe) (Fig 1c,d-e, l). Strength of interference  (Fig 1m). Using this method, one 941 interference measurement was produced for every repeat and then averaged. Standard 942 deviation and SEM was estimated and a two-tailed T-test performed to measure significant 943 differences between the strains as indicated in the figure legends.  instance, weak hotspots are more difficult to characterise than strong ones. Moreover, 952 quantification of DC molecules is challenging in the TEL1+ background because the level of 953 coincident DSBs is low and generally at or below the detection limit, therefore most of the 954 signal measured is background signal which sometimes can be higher than the calculated 955 expected random frequency (if any of the hotspots is weak) and thus leading to an 956 underestimate of interference strength (e.g. hotspot N and Q; Fig S3g-h). Finally, because the strength of DSB interference is calculated using the division of the observed frequency of 958 double-cut molecules by the frequency expected from independence, the result may be 959 inaccurate when the observed and expected values are close to zero because it produces a 960 disproportionate relative difference that may be artefactual. For example, for this reason the 961 strength of interference was excluded between NO at the YCR061W hotspot in the sae2Δ and 962 sae2Δ ndt80Δ (Fig S3l).

964
Another limitation of these techniques is that they only permit an estimate of the number of 965 broken chromatids and not how many times a chromatid has been broken, therefore, 966 quantification of the total frequency of DSBs may be underestimated if the frequency of double 967 events is high (as is the case of tel1Δ mutants). Furthermore, the direction and distance from 968 the probe to the hotspot also influences the accuracy of hotspot detection. For example, due 969 to hotspots having a width of 100-300 bp, when the inter-hotspot distance is very short (e.g. On the other hand, when the distance between the probe and the measured hotspot is large, 975 the presence of hotspots close to the probe will cause an underestimate of the real frequency 976 of DSBs that are further away. For instance, quantification of the main ARE1 hotspot "F" 977 slightly differs when measured from the right side of the DSB using PWP2 probe or from the 978 left with the TAF2 probe (Fig S2). In this example, measurement of F with TAF2 reported a 979 lower amount of F than PWP2 probe probably due to the presence of the strong hotspot E 980 prior to F, thus closer to the TAF2 probe.