Sgs1 BLM independent role of Dna2 DNA2 nuclease at DNA double strand break is inhibited by Nej1 XLF

The two major pathways of DNA double strand break (DSB) repair, non-homologous end-joining (NHEJ) and homologous recombination (HR), are highly conserved from yeast to mammals. Regulated 5’ DNA end resection is important for repair pathway choice and repair outcomes. Nej1 was first identified as a canonical NHEJ factor involved in stimulating the ligation of broken DNA ends and, more recently, it was shown to be important for DNA end-bridging and inhibiting Dna2-Sgs1 mediated 5’ resection. Dna2 localizes to DSBs in the absence of Sgs1 through interactions with Mre11 and Sae2 and DNA damage sensitivity is greater in cells lacking Dna2 nuclease activity compared to sgs1∆ mutants. Dna2-Sae2 mediated 5’ resection is down-regulated by Nej1, which itself interacts with Sae2. The resection defect of sae2 ∆ and the synthetic lethality of sae2 ∆ sgs1 ∆ are reversed by deletion of NEJ1 and dependent on Dna2 nuclease activity. Our work demonstrates the importance of Nej1 in inhibiting short-range resection at a DSB by Dna2-Sae2, a critical regulatory mechanism that prevents the formation of genomic deletions at the break site.


INTRODUCTION
DNA double-strand breaks (DSBs) can be repaired by two central pathways, non-homologous end joining (NHEJ) and homologous recombination (HR).NHEJ mediates the direct ligation of DNA ends without the requirement for end processing, whereas HR requires 5' end resection.
Both the initiation of resection and end-bridging at a DSB are important for repair pathway choice and down-stream processing.Once resection initiates at a DSB, repair by canonical NHEJ is no longer an option.Therefore, a complex network of proteins has evolved to regulate this key step, which by extension governs repair pathway choice.yKu70/80 (Ku) and Mre11-Rad50-Xrs2 (MRX) are the first complexes that localize to a DSB and both are important for recruiting the NHEJ factor, Nej1 [1][2][3][4].Cells lacking NEJ1 are as defective in end-joining as ku70Δ and dnl4Δ [3,5,6].After localization, Nej1 stabilizes Ku, which protects the DNA ends from nucleolytic degradation, and promotes Lif1-Dnl4 mediated ligation [2,3,7,8].Nej1 also inhibits Sgs1-dependent 5' DNA resection through a mechanism in which Nej1 binds to both Mre11 and Sgs1, inhibiting Dna2 localization [4,9].While Nej1 functions antagonistically to MRX in the context of 5' resection, Nej1 aligns with MRX functionality in DNA end-bridging.The structural features of the MRX complex are critical for end-bridging, and bridging defects associated with NEJ1 deletion are additive with a rad50 mutant that is proficient in resection, but defective in end-bridging [4,[10][11][12][13].
Sae2, the yeast homologue of human CtIP, interacts with the MRX-complex and initiates resection by activating Mre11 3' to 5' endonuclease which results in Ku dissociation from the DNA ends [14,15].After resection initiates by Sae2-dependent Mre11 activation, long-range resection proceeds from two functionally redundant 5' to 3' nucleases, Dna2 in complex with Sgs1 helicase or Exo1 [15,16].Dna2 is mutated in a myriad of human cancers but understanding its entire function at DSBs has been challenging because DNA2 is an essential gene involved in Okazaki fragment processing, therefore it cannot be deleted [17][18][19][20][21].The lethality of dna2Δ is supressed by disruption of PIF1 helicase [22].Dna2 has both 3' to 5' and 5' to 3' endo/exonuclease activity and 3' to 5' helicase activity.It is an inefficient helicase, thus Sgs1 provides this activity during long-range resection [17,23,24].Dna2 is recovered at a HO-induced DSB and resection defects at the break in dna2Δ pif1-m2 isogenic cells were equivalent +/-SGS1 [16].This epistatic relationship for DNA2 and SGS1 provided support to the model that Dna2-Sgs1 function as complex in DSB repair.However, these data do not account for the greater sensitivity of dna2-1 sgs1Δ to IR and UV compared to single mutants, with nuclease deficient dna2-1 (P504→S) showing greater sensitivity than sgs1Δ [25].While Dna2 has an important role in 5' resection, there is currently a gap understanding its entire functionality and regulation at the DSB.
In the absence of Sae2 or Mre11 nuclease activity, resection initiates primarily through Dna2 activity, not through Exo1 [26][27][28].However, the ability of these long-range nucleases to serve as compensatory back-ups and initiate resection largely depends on Ku binding.The DSB sensitivity and resection deficiency of mre11-nuclease dead mutants are alleviated by disrupting Ku.When DNA ends are exposed from the loss of Ku, they are rapidly recognized as substrate for Exo1 nuclease.This two-step model for the initiation and then long-range resection depends on different nucleases, however current data does not exclude a function for Dna2 in both.Like Mre11, Dna2 has 3' to 5' endonuclease activity and the nuclease activities of Dna2 and Mre11 show functional redundancy for processing the ends of DSBs after radiation treatment [24].Moreover, when NEJ1 is deleted and Dna2 levels increase, 5' resection increases similarly in wild type and nuclease-dead mre11-3 mutant cells.How Nej1 functions to down-regulate Dna2-dependent 5' resection remains obscure, but the mechanism is likely more intricate than simply blocking Dna2-Sgs1 interactions with Mre11 [4].For example, Sae2 was recently shown to stimulate the nuclease and helicase activity of Dna2-Sgs1, although this needs to be demonstrated in vivo at the break site [29,30].The relationship between Nej1 and Sae2 warrants investigation for this reason and because like MRX and Nej1, Sae2 has a role in DNA end-bridging at DSBs [31], a function conserved in humans and with Ctp1 in fission yeast [32,33].
Here we determine the implications of regulating Nej1 and Dna2 levels at a DSB in DNA processing and the physiological importance of Nej1-dependent inhibition of Dna2-Sae2 interaction.In nuclease deficient dna2-1 mutants, NHEJ mediated repair increases after DSB formation as do the levels of Nej1 and Ku binding.We had previously shown the inverse, where Dna2 levels increase at DSBs in nej1Δ mutants.In the current work we find Nej1 interacts with Sae2 and down-regulated Sae2 binding to both MRX and Dna2.Deletion of both NEJ1 and SAE2 showed epistatic end-bridging defects, increased 5' resection, and the development of large deletions at the break site compared to sae2Δ single mutant cells.Lastly, we show that deletion of NEJ1 can reverse the synthetic lethality of sae2Δ sgs1Δ through a mechanism dependent on the nuclease activity of Dna2.

RESULTS
Sgs1-independent recruitment of Dna2 nuclease is inhibited by Nej1 at DSB Nej1 promotes NHEJ and down-regulates 5' DNA resection by inhibiting Sgs1-dependent recruitment of Dna2 to a DSB [4,9].Deletion of SGS1 reverses hyper-resection and the large genomic deletions that form at a DSB break in nej1∆ mutants.To determine whether loss of Dna2 nuclease activity shows similar, we combined nuclease dead dna2-1 with nej1∆ [17,18].
Resection was measured by quantitative PCR-based approach and relies on a RsaI cut site 0.15kb from the DSB as previously described [9,31].RsaI will be unable to cleave if resection has proceeded past the cut site as single-stranded DNA is present and the region can be amplified by PCR (Figure 1A).5' resection was reduced more in dna2-1 compared to sgs1∆ mutants, and the rate in dna2-1 nej1∆ was far below wild type and sgs1∆ nej1∆ (Figure 1B).We performed chromatin immuno-precipitation (ChIP) at the HO-induced DSB and observed that Dna2 recruitment was similarly compromised in sgs1∆ and dna2-1 mutant cells (Figure 1C), however Sgs1 recruitment remained at wild type levels in dna2-1 (Figure S1A).Dna2 binding to Sgs1 is inhibited by Nej1 and consistent with previous reports, its recruitment to a DSB increases in nej1∆ and is reduced, but not gone, in sgs1∆ mutants (Figure 1C) [4,9].Strikingly, Dna2 can be recruited to a DSB independently of Sgs1 via a pathway that is also inhibited by Nej1 as its recovery in sgs1∆ nej1∆ was above wild type and similar to the level in nej1∆ mutants (Figure 1C).
The level and functionality of these factors at a DSB dramatically impacts repair pathway choice.We previously showed that overexpression of Dna2 resulted in decreased Nej1 recovered at the DSB [4].Under conditions of Dna2 deficiency in cells harboring dna2-1, Nej1 recovery increased as did the rate of NHEJ repair (Figures S1B, 1D).By contrast, Nej1 levels were unaltered in sgs1∆, which is notable given Nej1 physically interacts with Sgs1 but not Dna2 [4].In NEJ1+ cells, end-joining as determined by survival on galactose inversely correlates with 5' DNA resection, increasing in sgs1∆ and even more so in dna2-1 mutants.As Nej1 is essential for NHEJ, growth on galactose was markedly reduced in all mutant combinations containing nej1∆ (Figure 1D).
Determining the mating type of survivors that grow on galactose provides a physiological readout of DNA processing events and genomic alterations occurring at a HO-induced DSB located within MATα1 and adjacent to MATα2 (Figure 1E).Expression of these genes regulate the mating type by activating alpha-type genes and inhibiting a-type genes.Extensive resection leads to large deletions (>700 bp) and disruption of both α1 and α2 genes, with a-like survivors (red), whereas small deletions or insertions lead to sterile type survivors (yellow).Repair events resulting in α-type survivors show no sequence change at the HO recognition site and usually develop from disruption in the HO endonuclease (gray).In WT cells and sgs1∆ mutants, the majority of survivors are sterile, and contain small insertions or deletions in the MATα1 gene (Figure 1E) [9,34].By contrast, dna2-1 survivors were primarily MATα, indicating an increase in the rate of canonical NHEJ.The cut efficiency in dna2-1 mutants was similar to wild type 2 hours after galactose induction, however HO-cutting was completely abrogated in dna2-1 survivors (Table S4), which is consistent with canonical NHEJ repair and survivors having alterations in the endonuclease.
As with wild type, mre11-3 and exo1∆ single mutant survivors had small indels (Figure 2C) and end-joining increased more when these mutations were combined with dna2-1 compared to double mutant combinations with sgs1∆ (Figure 2D).While overall survival decreased when NEJ1 was deleted in combination with these double mutants (Figure 2D), the percentage of survivors with large deletions increased (Figures 2C, D).Survival was equally low in all mutant combinations with nej1∆ (Figure 2D), however survivors harboring dna2-1 showed an increase in non-mutagenic NHEJ, with fewer indels and no large deletion compared to triple mutant combinations with sgs1∆ (Figure 2C).
To determine whether 5' resection and HO survival differences in mre11-3 and exo1∆ were related to Dna2 localization +/-SGS1 status, ChIP was performed on Dna2 in the different mutant combinations.Deletion of NEJ1 increased the recovery level of Dna2 at the break in all mutant backgrounds (Figure 2E).Even though the rate of resection was lower in exo1∆ sgs1∆ mutants compared to sgs1∆, the level of Dna2 recovered was similar (Figure 2E).Consistently, deletion of Exo1 did not alter 5' resection or Dna2 levels in otherwise wildtype or nej1∆ mutants (Figures 2B, 2E).By contrast, the recovery of Dna2 in mre11-3 sgs1∆ was higher than in sgs1∆ (p-value < 0.05), and similar to wild type (Figures 2E, 1C).Overall, these data suggest that Sgs1-independent recruitment of Dna2, which is inhibited by Nej1, involves a factor linked to Mre11 nuclease activity.This led us to investigate a potential role for Sae2, which was recently shown to increase at the DSB in mre11-3 mutants [27].
Sae2-dependent recruitment of Dna2 is inhibited by Nej1 Sae2 interacts with the MRX complex to activate Mre11 endonuclease and more recently it was also shown to stimulate the nuclease activity of Dna2 [15,29,30].While Sae2 recruitment is abrogated in mre11Δ, levels increase in mre11-3 (Figure 3A) [27].Thus, we determined whether the presence of Dna2 at the DSB in mre11-3 depended on Sae2.Indeed, the recruitment of Dna2 HA significantly decreased in mre11-3 sae2Δ and sae2Δ (Figure 3B).As Nej1 inhibits Dna2, we next determined Dna2 recovery when NEJ1 was deleted in combination with sae2Δ.
Underscoring the relationship between Dna2 and Sae2, increased recovery of Dna2 in mre11-3 nej1Δ double mutants was Sae2 dependent (Figures 2E and 3B).In the course of our investigation, we also observed that Sae2 levels increased in nej1Δ (Figure 3A), and the increase was additive in combination with mre11-3.Conversely, Sae2 recovery in dna2-1 remained indistinguishable from wild type as it did in sgs1Δ and exo1Δ mutants (Figures S1D, E).Thus, Nej1 appears regulate Sae2 levels as a way to inhibit resection at DSBs.When measured directly, we saw that hyper 5' resection in nej1∆ was reversed by deletion of SAE2 (Figure 3C).Similar to dna2-1, the rate of survival and non-mutagenic end-joining increased in sae2Δ mutants (Figure 3D).However, unlike dna2-1, large deletion developed when sae2Δ was combined with nej1∆ (Figures 1E, 3E).Previously, we showed that both 5' resection levels and DNA end-bridging defects contribute to the development of large deletions around a DSB [4].
End-bridging was measured using a strain with TetO at 3.2 Kb upstream and the LacO array at 5.2Kb downstream of the cut site.These cells express GFP-tagged TetR and mCherry-tagged LacO, enabling us to visualize the two sides of the break site by fluorescence microscopy (Figure 3F).In asynchronous cells, the distance between the GFP and mCherry foci was measured prior to (0 hr) and after (2 hr) DSB induction.Wild-type cells did not have a significantly different mean distance (0.24 and 0.28µm respectively), indicative of successful DNA end-bridging.
End-bridging was defective in sae2Δ, but not in dna2-1 mutant cells (Figures 3F, S2).Comparing mre11-3 and sae2Δ +/-NEJ1 in 5' resection and end-bridging provides support for a model where both processes safeguard against the development of large deletions.In mre11-3 mutants, resection proceeds but end-bridging is not disrupted whereas in sae2Δ mutants, bridging is disrupted but 5' resection is also low (Figures 3C-F), and neither single mutants show large deletions.By contrast, large deletions do form when both mutants are combined with the deletion of NEJ1.The end-bridging defect in sae2Δ mutants was epistatic with nej1Δ, yet the rate of 5' resection increased in sae2Δ nej1Δ compared to sae2Δ, and large deletions also formed.Consistent with this model, mre11-3 nej1Δ and mre11-3 sae2Δ had similar endbridging defects, however both 5' resection and Dna2 levels remained low and large deletions did not develop in mre11-3 sae2Δ survivors.These data underscore the role of Sae2, together with Nej1, in end-bridging and also suggest an additional role for Sae2 in resection that goes beyond activation of Mre11 nuclease, as resection decreases more in sae2Δ compared to mre11-3 mutants.
We then determined whether the interplay between these factors also involved physical interactions by performing yeast-two-hybrid (Y2H) [4,35].Sae2 was overexpressed as HAtagged prey and Nej1, the MRX subunits, and Dna2 domains -1-450 aa (N-term), 451-900 aa (nuclease) and 901-1522 aa (helicase) were expressed as LexA-tagged bait [4], under control of a galactose-inducible promoter (Figure S3).Consistent with previous reports [36], Sae2 physically interacts with each component of MRX complex (Figure 3G).We also find Sae2 binds Nej1 more strongly than its binds the MRX components, and it physically interacts with the Nterminal regulatory region and nuclease domains of Dna2.Interestingly, Sae2-Dna2 and Sae2-MRX interactions increased in nej1Δ mutants (Figure 3G), which was similar to previous observations with Dna2-Sgs1 and Dna2-Mre11, which also increased when NEJ1 was deleted [4].Nej1 inhibits interactions between nucleases and nuclease regulators, however one difference to note is that Nej1 and Sae2 interact physically (Figure 3G), whereas Nej1 and Dna2 did not [4].Taken together, these data support a model whereby Nej1 inhibits Dna2 recruitment to DSBs through multiple pathways.In addition to Nej1 inhibiting Sgs1-dependent Dna2 recruitment, Nej1 interacts with Mre11 and Sae2 inhibiting Dna2 binding with these factors (Figure 3H and [4]).

Dna2 plays a major role in resection at DSB that is inhibited by Nej1
We observed that recruitment of Dna2 at DSB is partially dependent on Sae2 and Sgs1, and both pathways are inhibited by Nej1.Furthermore, the rate of resection is severely hampered in cells harboring the dna2-1 mutation which is modestly alleviated by NEJ1 deletion, indicating both Dna2 levels and nuclease function are critically important for processing events at a DSBs.
Taken together we were prompted to determine whether nej1Δ, which removes Dna2 inhibition, would alleviate sae2Δ sgs1Δ synthetic lethality (SL) [27].We crossed sae2Δ nej1Δ with sgs1Δ nej1Δ and spores with the triple mutant combination grew remarkably well (Figure 4A).Similar to nej1Δ and the sae2Δ nej1Δ with sgs1Δ nej1Δ double mutants, the triple mutant showed a low level of survival and an accumulation of large deletions under chronic HO induction (Figures 4B, C).The rate of resection in the triple mutant was indistinguishable from wild type and significantly higher than levels in sae2Δ and sgs1Δ single mutants (Figure 4D).
Moreover, Dna2 HA recruitment to a DSB in sae2Δ sgs1Δ nej1Δ mutants was similar to WT (Figure 4E), and above levels in sae2Δ +/-NEJ1 or sgs1Δ (Figures 1C, 3B).To determine whether the reversion sae2Δ sgs1Δ lethality from NEJ1 deletion was Dna2 nuclease dependent, we generated heterozygous diploids for sae2Δ, sgs1Δ, nej1Δ and dna2-1 and upon tetrad dissection no viable spores with the quadruple mutant combination were recovered (Figure 4F).Taken together, our data support a model whereby Dna2 is recruited to a DSB through three pathways all under inhibition by Nej1.Dna2 localizes primarily through binding with Sae2 and Sgs1, thus deletion of both results in lethality.Removal of Nej1, which binds Mre11-C , allows Dna2 recruitment through Mre11-Dna2 interactions with cell viability in sae2Δ, sgs1Δ, nej1Δ mutants now dependent on the nuclease activity of Dna2 (Figure 4G).

DISCUSSION
Here we show Nej1 is a general inhibitor of 5' resection by regulating the recruitment of central nucleases and their binding partners to a DSB.We show that Dna2 is recruited to a DSB by Sae2, independently of Sgs1, and that Nej1 physically interacts with Sae2.Nej1 not only inhibits Dna2 interactions with Sgs1 and MRX [4], but the data presented here show Nej1 also inhibits MRX-Sae2 interaction and Sae2-dependent Dna2 recruitment.Moreover, deletion of NEJ1 and SAE2 show epistatic end-bridging defects, raising the possibility that their interaction might also have functional relevance for this important mechanism in DSB repair.

The relationship between Dna2 and NHEJ factors
The 5' resection defects in dna2-1 mutants support a Sgs1-independent role for Dna2 in DSB repair and indicate that only part of Dna2 functionality in resection is manifested by deletion of SGS1.Resection was more defective in dna2-1 compared to sgs1Δ even though Dna2 levels were similarly reduced in dna2-1 and sgs1Δ (Figures 1B, C).Moreover, Dna2 is still recruited to a DSB, albeit at lower levels, in the absence of Sgs1.Underscoring the inhibitory relationship between Nej1 and Dna2, when NEJ1 was deleted in sgs1Δ mutants, Dna2 levels increased to above wild type indicating that Nej1 inhibits Dna2 recruitment through Sgs1-independent pathway(s) (Figure 1C and [4]).Double mutant combination of either dna2-1 or sgs1Δ with EXO1 deletion also support an Sgs1independent function for Dna2 at the DSB.5' resection was markedly reduced in both exo1Δ dna2-1 and exo1Δ sgs1Δ as the two long-range resection pathways were disrupted.However, deletion of NEJ1 reversed the resection defect more in exo1Δ sgs1Δ nej1Δ triple mutants (Figures 2B).There appears to be little or no genetic interaction between NEJ1 and EXO1 in DSB repair as hyper-resection in nej1Δ mutants was not reversed by the deletion of EXO1 and large deletions persisted (Figures 2B, D) [9].Another difference to highlight is that exo1Δ dna2-1 survivors contained very few mutations at the break site whereas exo1Δ sgs1Δ survivors contained a high percentage of small indels indicative of some short-range resection, independently of NEJ1 status (Figure 2C).Of note, exo1Δ dna2-1 double mutants were viable, which is in contrast to the lethal interaction previously reported for exo1Δ dna2Δ pif1-m2 [16].
How the pif1-m2 mutation might impact the genetic interaction between DNA2 and EXO1 is not clear, however this result suggests that the genetic lethality does not stem from loss of Dna2 nuclease activity.
Our data also suggests that Dna2 and Nej1 have an antagonistic relationship rather than a unidirectional one where Nej1 inhibits Dna2, as the levels of Nej1 and Ku70 increase at DSBs in dna2-1 (Figures S1B-C).These results suggest that increased levels of Ku and Nej1 and delayed resection allow more opportunity for NHEJ.Indeed, survival of dna2-1 mutants increases under HO-DSB induction when NHEJ is proficient, as in NEJ1+ (Figures 1C).Canonical NHEJ factors down-regulate 5' resection at DSBs, however with the exception of Ku, mechanistic understanding about how these factors contribute to repair pathway choice remains fairly obscure.We show Nej1, which is known to contributes to Ku stability [3], has different genetic requirements for inhibiting resection compared to those previously identified for Ku.For example, even though nej1Δ, like ku70Δ, suppresses the sensitivity mre11-3 and sae2Δ, (Figure 2A, 3C and [37,38]), the rescue of mre11-3 by ku70Δ depends on Exo1 nuclease, yet rescue by nej1Δ is Exo1-independent [9,37,38].
Nej1 and Sae2 in end-bridging Nej1 physically interacts with Sae2 and down regulates Sae2 interactions with MRX.Consistent with recent reports, Sae2 has a role in end-bridging (Figure 3F; [31][32][33]).In contrast to the additive end-bridging defects for rad50Δ nej1Δ [4], sae2Δ end-bridging defects are epistatic with nej1Δ.Given the physical interaction between Nej1 and Sae2, it is possible that the two factors could function as a complex in end bridging.It might be informative to determine whether Sae2 and Nej1 can form hetero-oligomers, distinguishing a sub-population of Sae2 involved in DNA end-bridging apart from Sae2 homo-oligomers involved in Mre11 activation and checkpoint signalling [39,40].Importantly, none of the other factors involved in 5' resection, when mutated, showed a defect in end-bridging including exo1Δ, sgs1Δ, mre11-3, or dna2-1 (Figure S2).These data are in line with previous work showing the structural integrity of the MRX complex, which remains intact in mre11-3, is important for its end-bridging function [12,13,41].

Sae2-dependent recruitment of Dna2 is inhibited by Nej1
As MRX is required for the recruitment of other processing factors to the DSB, mre11-3 was used in this study because in contrast to mre11Δ the complex is proficient in localizing and endbridging in this nuclease-dead allele [12,13,41].Consistent with previous reports, we find Dna2 recruitment to DNA ends did not depend on Mre11 nuclease activity, however levels were reduced in sae2Δ (Figure 3A; [42]).This differs slightly from previous work, which showed Dna2 is reduced at 2 hours, but not at 3 hrs, in sae2Δ mutant cells [42].This discrepancy could result from slight variations in the experimental set up because in the same study Ku70 was not altered 3 hrs after HO induction in sae2Δ mutants, yet we find that Ku70 levels increase slightly (Figure S1C).Our results are however in line with work showing KU70 deletion suppresses the damage sensitivity of sae2Δ and Ku70 levels increase at a DSB in ctp1Δ mutants in S. pombe [37,38,43].
Recent in-vitro studies have shown that human CtIP has a stimulatory effect on Dna2 nuclease [29,30].We observed a physical interaction between Sae2 and Dna2 which was inhibited by Nej1 (Figure 3G).Dna2 enrichment levels at a DSB corresponded well with resection levels in the various mutant backgrounds (Figures 3B, C), and support a model for Sae2 recruitment of Dna2 to DSBs.Indeed, there was a marked decrease in Dna2 in sae2Δ mutant cells compared to mre11-3 (Figures 2E, 3B).These data are consistent with previous work showing Dna2 recruitment to DSBs occurs through interactions with CtIP in humans [44].Nej1 inhibited Dna2 recruitment in the various mutant backgrounds.Compared to sae2Δ mutants, the level of Dna2 increased in sae2Δ nej1Δ, however recovery levels remained lower than in mre11-3 nej1Δ (Figures 2E, 3B).Our data do not contradict the role of Sae2 as a nuclease [28], but further study of the interplay between Nej1, Dna2 and Sae2-mutants (D285P/K288P and E161P/K163P) is imperative.
Dna2 recovery and 5' resection in sae2Δ sgs1Δ nej1Δ was restored to wild type levels, which was quite surprising given sae2Δ sgs1Δ is synthetic lethal (Figure 4) [27].It is puzzling that the recovery of Dna2 and the rate of resection is greater in sae2Δ sgs1Δ nej1Δ compared to sae2Δ nej1Δ.These data might indicate that the presence of Sgs1 is inhibitory to the initiation of resection when SAE2 has been deleted.We previously showed that both Sgs1 and Dna2 interact directly with Mre11 [4].Thus, one possibility is that the relative amount of Dna2 binding directly to Mre11 is dynamic and altered in sae2Δ nej1Δ mutants and impacted differently by the presence of Sgs1.For example, Dna2-Mre11 might function to initiate resection in sae2Δ, whereas Dna2 in complex with Sgs1 at the DSB functions in long-range resection and is less efficient in initiating it.In the absence of Mre11 nuclease activity however, Sgs1-Dna2 would become very important for resection, and as previously shown, this interaction is also inhibited by Nej1 [4,9].Consistent with such a model, both Dna2 and 5' resection levels increase in mre11-3 sae2Δ nej1Δ, relative to mre11-3 sae2Δ.Suppression of sae2Δ sgs1Δ lethality, has also been seen upon deletion of other factors that inhibit resection, KU70 and RAD9.Like triple mutant combinations with ku70Δ or rad9Δ, resection in sae2Δ sgs1Δ nej1Δ was similar to wild type at later time points.The increased resection and rescue of sae2Δ sgs1Δ upon NEJ1 deletion might result from decreased endjoining, as this would also be reduced in ku70Δ.However, two results suggest that the loss of end-joining itself does not rescue the lethality of sae2Δ sgs1Δ.First, deletion of DNL4 ligase does not rescue sae2Δ sgs1Δ and in sae2Δ sgs1Δ rad9Δ triple mutants, NHEJ occurs.
We show Dna2 can be recruited to the DSB site by Sae2, and like its recruit through Mre11 and Sgs1, Dna2-Sae2 is inhibited by Nej1.We also report well-regulated and balanced interplay between Nej1, Sae2 and Dna2 in end-bridging and end-resection.This work underlines a central function for Dna2 nuclease in DSB repair and accentuates the role Nej1 as a major regulator of NHEJ/repair pathway choice.S2 Primers Primers used in this study Table S3 Yeast strains

KEY RESOURCE
Yeast strains used in this study Table S1 All the yeast strains used in this study are listed in Table S1 and were obtained by crosses.The strains were grown on various media in experiments described below.For HO-induction of a DSB, YPLG media is used (1% yeast extract, 2% bactopeptone, 2% lactic acid, 3% glycerol and 0.05% glucose).For the continuous DSB assay, YPA plates are used (1% yeast extract, 2% bacto peptone, 0.0025% adenine) supplemented with either 2% glucose or 2% galactose.For the mating type assays, YPAD plates are used (1% yeast extract, 2% bacto peptone, 0.0025% adenine, 2% dextrose).For yeast 2-hybrid assays, standard amino acid drop-out media lacking histidine, tryptophan and uracil is used and 2% raffinose is added as the carbon source for the cells.

METHOD DETAILS Chromatin Immunoprecipitation
ChIP assay was performed as described previously [4].Cells were cultured overnight in YPLG at 25°C.Cells were then diluted to equal levels (5 x 10 6 cells/ml) and were cultured to one doubling (3-4 hrs) at 30°C.2% GAL was added to the YPLG and cells were harvested and crosslinked at various time points using 3.7% formaldehyde solution.Following crosslinking, the cells were washed with ice cold PBS and the pellet stored at -80°C.The pellet was resuspended in lysis buffer (50mM Hepes pH 7.5, 1mM EDTA, 80mM NaCl, 1% Triton, 1mM PMSF and protease inhibitor cocktail) and cells were lysed using Zirconia beads and a bead beater.Chromatin fractionation was performed to enhance the chromatin bound nuclear fraction by spinning the cell lysate at 13,200rpm for 15 minutes and discarding the supernatant.The pellet was re-suspended in lysis buffer and sonicated to yield DNA fragments (~500bps in length).The sonicated lysate was then incubated in beads + anti-HA/Myc Antibody or unconjugated beads (control) for 2 hrs at 4°C.The beads were washed using wash buffer (100mM Tris pH 8, 250mM LiCl, 150mM (HA Ab) or 500mM (Myc Ab) NaCl, 0.5% NP-40, 1mM EDTA, 1mM PMSF and protease inhibitor cocktail) and protein-DNA complex was eluted by reverse crosslinking using 1%SDS in TE buffer, followed by proteinase K treatment and DNA isolation via phenolchloroform-isoamyl alcohol extraction.Quantitative PCR was performed using the Applied Biosystem QuantStudio 6 Flex machine.PerfeCTa qPCR SuperMix, ROX was used to visualize enrichment at HO2 (0.5kb from DSB) and HO1 (1.6kb from DSB) and SMC2 was used as an internal control.

Tethering Microscopy
Cells derived from the parent strain JC-4066 were diluted and grown overnight in YPLG at 25°C to reach a concentration of 1x10 7 cells/ml.Cells were treated with 2% GAL for 2 hours and cell pellets were collected and washed 2 times with PBS.After the final wash, cells were placed on cover slips and imaged using a fully motorized Nikon Ti Eclipse inverted epi-fluorescence microscope.Z-stack images were acquired with 200 nm increments along the z plane, using a 60X oil immersion 1.4 N.A. objective.Images were captured with a Hamamatsu Orca flash 4.0 v2 sCMOS 16-bit camera and the system was controlled by Nikon NIS-Element Imaging Software (Version 5.00 Cells from each strain were grown overnight in 15ml YPLG to reach an exponentially growing culture of 1x10 7 cells/mL.Next, 2.5mL of the cells were pelleted as timepoint 0 sample, and 2% GAL was added to the remaining cells, to induce a DSB.Following that, respective timepoint samples were collected.Genomic DNA was purified using standard genomic preparation method by isopropanol precipitation and ethanol washing, and DNA was re-suspended in 100mL ddH 2 O. Genomic DNA was treated with 0.005μg/μL RNase A for 45min at 37°C.2μL of DNA was added to tubes containing CutSmart buffer with or without RsaI restriction enzyme and incubated at 37°C for 2hrs.Quantitative PCR was performed using the Applied Biosystem QuantStudio 6 Flex machine.PowerUp SYBR Green Master Mix was used to quantify resection at MAT1 (0.15kb from DSB) locus.Pre1 was used as a negative control.RsaI cut DNA was normalized to uncut DNA as previously described to quantify the %ssDNA / total DNA [31].
Continuous DSB assay and identification of mutations in survivors Cells were grown overnight in YPLG media at 25°C to saturation.Cells were collected by centrifugation at 2500rpm for 3 minutes and pellets were washed 1x in ddH 2 O and resuspended in ddH 2 O. Cells were counted and spread on YPA plates supplemented with either 2% GLU or 2% GAL.On the Glucose plates 1x10 3 total cells were added and on the galactose plates 1x10 5 total cells were added.The cells were incubated for 3-4 days at room temperature and colonies counted on each plate.Survival was determined by normalizing the number of surviving colonies in the GAL plates to number of colonies in the GLU plates.100 survivors from each strain were scored for the mating type assay as previously described [9].
Yeast 2-hybrid Various plasmids (Table S2) were constructed containing the gene encoding the region of the proteins -Sae2, Dna2, Mre11, Nej1, Rad50 and Xrs2, using the primers listed in Table S3.The plasmids J-965 and J-1493 and the inserts were treated with BamHI and EcoRI and ligated using T4 DNA ligase.The plasmids were sequence verified.Reporter (J-359), bait (J-965) and prey (J-1493) plasmids, containing the gene encoding the desired protein under a galactose inducible promoter, were transformed into JC-1280.Cells were grown overnight in -URA -HIS -TRP media with 2% raffinose.Next day, cells were transferred into -URA -HIS -TRP media with either 2% GLU or 2% GAL and grown for 6 hrs at 30°C.Cell pellets were resuspended and then permeabilized using 0.1% SDS followed by ONPG addition.β-galactosidase activity was estimated by measuring the OD at 420nm, relative β-galactosidase units were determined by normalizing to total cell density at OD600.Additionally, for drop assay cells were grown and spotted in five-fold serial dilutions on plates containing 2% galactose lacking histidine and tryptophan (for plasmid selection) and leucine (for measuring expression from lexAop6-LEU2).
Plates were photographed after 3-4 days of incubation at 30°C.

Western Blot
Cells were lysed by re-suspending them in lysis buffer (with PMSF and protease inhibitor cocktail tablets) followed by bead beating.The protein concentration of the whole cell extract was determined using the NanoDrop.Equal amounts of whole cell extract were added to wells of 10% polyacrylamide SDS gel.After the run, the protein were transferred to Nitrocellulose membrane at 100V for 80mins.The membrane was Ponceau stained (which served as a loading control), followed by blocking in 10% milk-PBST for 1hour at room temperature.Respective primary antibody solution (1:1000 dilution) was added and left for overnight incubation at 4°C.The membranes were then washed with PBST and left for 1 hour with secondary antibody.Followed by washing the membranes, adding the ECL substrates and imaging them.(A) Schematic representation of regions around the HO cut site on chromosome III.i) Schematic representation of mating type analysis of survivors from persistent DSB induction assays.The mating phenotype is a read out for the type of repair: α survivors (mutated HO endonucleasesite, grey), sterile survivors (small insertions and deletions, blue) and "a-like" survivors (>700 bps deletion, red).ii) The ChIP probe used in this study is 0.6kb from the DSB.The RsaI sites used in the qPCR resection assays, 0.15kb from the DSB, are also indicated.Both sides of HO cut site are tagged with GFP (5kb from cute site) and mCherry (9kb from cut site) repeats, which were used in end-tethering microscopy.
(D) Percentage cell survival upon chronic HO induction in cells mentioned in (B).
(E) Mating type analysis of survivors from persistent DSB induction assays mentioned in (D and described in A).
(C) Mating type analysis of survivors from persistent DSB induction assays mentioned in (D and described in 1A).
(D) Percentage cell survival upon chronic HO induction in cells mentioned in (A and B).
The fold enrichment is normalized to recovery at the SMC2 locus.
(D) Percentage cell survival upon chronic HO induction in cells mentioned in (C).
(E) Mating type analysis of survivors from persistent DSB induction assays mentioned in (D and described in 1A).
(C) Mating type analysis of survivors from persistent DSB induction assays mentioned in (B and described in 1A).
(D) qPCR-based resection assay of DNA 0.15kb away from the HO DSB, as measured by % ssDNA, at 0, 40, 80 and 150 mins.post DSB induction in cycling cells mentioned in (B).
(E) Enrichment of Dna2 HA at 0.6kb from DSB 0 hour (no DSB induction) and 3 hours after DSB induction in wild type (JC-4117), sae2Δ sgs1Δ nej1Δ (JC-5480) and no tag control (JC-727) was determined.The fold enrichment is normalized to recovery at the SMC2 locus.
(G) Model of DSB where Nej1 inhibits the physical association of Dna2-Mre11 and prevents Mre11-dependent Dna2 recruitment.
For all the experiments -error bars represent the standard error of three replicates.Significance was determined using 1-tailed, unpaired Student's t test.All strains compared are marked (P<0.05*;P<0.01**; P<0.001***) unless specified with a line, strains are compared to WT.
Table S1 Yeast Strains: The yeast strains used in this study are outlined.

(
E) Model of DSB where Nej1 inhibits the physical association of Dna2-Sae2 and prevents Sae2dependent Dna2 recruitment.For all the experiments -error bars represent the standard error of three replicates.Significance was determined using 1-tailed, unpaired Student's t test.All strains compared are marked (P<0.05*;P<0.01**; P<0.001***) and are compared to WT, unless specified with a line.

Figure 4 .
Figure 4. Dna2 plays a major role in resection at DSB that is inhibited by Nej1.

Figure 3 A
Figure 3

TABLE REAGENT
).All images were deconvolved with Huygens Essential version 18.10 (Scientific Volume Imaging, The Netherlands, http://svi.nl), using the Classic Maximum Likelihood Estimation (CMLE) algorithm, with SNR:40 and 50 iterations.To measure the distance between the GFP and mCherry foci, the ImageJ plug-in Distance Analysis (DiAna) was used[45].Distance measurements represent the shortest distance between the brightest pixel in the mCherry channel and the GFP channel.Each cell was measured individually and > 50 cells were analyzed per condition per biological replicate.
AUTHORS CONTRIBUTIONSA.M. and J.A.C. designed the research.A.M., and N.A. performed experiments and analyzed the data.A.M., N.A. and J.A.C wrote the manuscript.41

Table S4 :
Table of Primers and Probes used in these studies.Table of DSB cut efficiency.