Regulation of Mus81-Eme1 structure-specific endonuclease by Eme1 SUMO-binding and Rad3ATR kinase is essential in the absence of Rqh1BLM helicase.

The Mus81-Eme1 structure-specific endonuclease is crucial for the processing of DNA recombination and late replication intermediates. In fission yeast, stimulation of Mus81-Eme1 in response to DNA damage at the G2/M transition relies on Cdc2CDK1 and DNA damage checkpoint-dependent phosphorylation of Eme1 and is critical for chromosome stability in absence of the Rqh1BLM helicase. Here we identify Rad3ATR checkpoint kinase consensus phosphorylation sites and two SUMO interacting motifs (SIM) within a short N-terminal domain of Eme1 that is required for cell survival in absence of Rqh1BLM. We show that direct phosphorylation of Eme1 by Rad3ATR is essential for catalytic stimulation of Mus81-Eme1. Chk1-mediated phosphorylation also contributes to the stimulation of Mus81-Eme1 when combined with phosphorylation of Eme1 by Rad3ATR. Both Rad3ATR- and Chk1-mediated phosphorylation of Eme1 as well as the SIMs are critical for cell fitness in absence of Rqh1BLM and abrogating bimodal phosphorylation of Eme1 along with mutating the SIMs is incompatible with rqh1Δ cell viability. Our findings unravel an elaborate regulatory network that relies on the poorly structured N-terminal domain of Eme1 and which is essential for the vital functions Mus81-Eme1 fulfills in absence of Rqh1BLM.

Introduction

Structure-specific DNA endonucleases are cornerstones in the proper execution of DNA replication, repair and recombination; yet they harbor the potential for causing genome instability. Controlling these enzymes is essential to ensure efficient processing of appropriate substrates while preventing counterproductive targeting of other similar DNA structures. The Mus81-Eme1 structure-specific endonuclease (SSE) emerged from early on as a key player in the processing of recombination intermediates that form during homology directed repair of two-ended double strand breaks or during the rescue of stalled or broken replication forks [1-7]. Over this last decade, control mechanisms of Mus81-Eme1 that are tightly linked to cell cycle progression have been identified.

In Saccharomyces cerevisiae (S. cerevisiae), Mus81-Mms4EME1 is stimulated at the G2/M transition by Cdc5PLK1-, Cdc28CDK1- and Dbf4-dependent phosphorylation of Mms4EME1 [8-12]. In human cells, catalytic upregulation of MUS81-EME1 is driven by complex formation with the SLX4 nuclease scaffold. This regulation is mediated by direct interaction between MUS81 and SLX4, which is strongly stimulated at the G2/M transition by phosphorylation of SLX4 by CDK1 [13]. Reminiscent of what has been observed in S. cerevisiae, maximal processing of joint molecules such as Holliday junctions by MUS81-EME1 also correlates with hyperphosphorylation of EME1, possibly by CDK1 or PLK1 [11]. Whether phosphorylation of EME1 contributes to the stimulation of MUS81-EME1 in human cells remains to be formally demonstrated. These mechanisms ensure that joint molecules such as Holliday junctions, D-loops or replication intermediates at under-replicated loci are efficiently resolved in mitosis before chromosome segregation [14]. By restricting the catalytic stimulation of Mus81-Eme1 to late stages of the cell cycle these control mechanisms also ensure that joint molecules and replication intermediates get a chance to be processed by more conservative non-endonucleolytic mechanisms that rely on their unfolding by RecQ helicases such as the BLM helicase in human cells and its Sgs1 and Rqh1 orthologs in Saccharomyces cerevisiae (S. cerevisiae) and Schizosaccharomyces pombe (S. pombe), respectively. These temporal controls further prevent the accumulation of hyper-activated Mus81-Eme1 in S-phase and the risk of the unscheduled processing of replication intermediates. The importance of such control mechanisms is underscored by the marked genomic instability that is caused by the premature stimulation of Mus81-Mms4 in budding yeast cells that produce an Mms4 mutant that mimics a constitutively phosphorylated Mms4 protein [9]. Interestingly, SUMOylation and ubiquitination of Mms4 were recently shown to specifically target phosphorylated Mms4 for degradation by the proteasome, further ensuring that hyperactivation of Mus81-Mms4 is restricted to mitosis [15]. In human cells, SLX4-MUS81 complex formation induced by premature activation of CDK1 results in the unscheduled processing of replication intermediates genome wide and chromosome pulverization [13].

In S. pombe, upregulation of Mus81-Eme1 also relies on the phosphorylation of Eme1 by Cdc2CDK1 [16]. However, in contrast to what has been described in S. cerevisiae, phosphorylation of Eme1 by Cdc2CDK1 primes Eme1 for further DNA damage checkpoint-mediated phosphorylation in response to DNA damage. This elaborate control mechanism ensures that Mus81-Eme1 is rapidly hyperactivated in response to DNA damage in late G2 and mitosis and is critical to prevent gross chromosomal rearrangements in absence of the BLM-related helicase Rqh1BLM [16].

To gain further insight into the molecular mechanisms involved in the control of Mus81-Eme1 in S. pombe, we undertook in silico analyses of a relatively short N-terminal domain of Eme1 that we found to be essential for cell viability in the absence of Rqh1BLM. We identified Rad3ATR consensus phosphorylation sites and two SUMO interacting Motifs (SIM1 and SIM2) within that domain. We demonstrate that Eme1 is a direct substrate for Rad3ATR both in vitro and in vivo and show that phosphorylation of Eme1 by Rad3ATR plays a prominent role in the catalytic stimulation of Mus81-Eme1 in response to DNA damage. Full-fledged stimulation of Mus81-Eme1 also relies on phosphorylation of Eme1 by Chk1 but, apparently, only when it occurs in conjunction with Rad3ATR-mediated phosphorylation of Eme1. We provide genetic evidence that both Rad3ATR—and Chk1-mediated phosphorylation of Eme1 are independently critical for cell fitness in the absence of Rqh1BLM. Remarkably, Chk1-mediated phosphorylation of Eme1 is lost when SIM2 is mutated, while mutating SIM1 has no impact. Both SIMs are important for cell fitness in absence of Rqh1BLM, and abrogating phosphoregulation of Mus81-Eme1 and mutating SIM1 and SIM2 recapitulates the synthetic lethality observed by deleting the N-terminus of Eme1 in the absence of Rqh1BLM.

Results

Eme1 N-terminus is essential in absence of Rqh1BLM

To gain further insight into the mechanisms underlying the catalytic stimulation of Mus81-Eme1 in response to DNA damage, we searched for domains of Eme1 that are essential in absence of Rqh1BLM while dispensable for the intrinsic catalytic activity of Mus81-Eme1. Interestingly, we found that deleting a relatively short N-terminal domain (residues 1–117) of Eme1 is synthetic lethal with rqh1Δ (). Importantly, this domain does not contain the Cdc2CDK1 sites that we had previously reported to be involved in the stimulation of Mus81-Eme1 in response to DNA damage and to be critical for cell viability in absence of Rqh1BLM [16]. A detailed in silico analysis of the first 117 residues of Eme1 led to the identification of two SQ/TQ Rad3ATR consensus sites (S23Q and T50Q) and two putative SUMO-Interacting Motifs (SIMs), hereafter named SIM1 and SIM2, which matched the described (V/I)-X-(V/I)-X-(V/I/L) consensus sequence () [17]. These observations suggested that Eme1 might be a direct substrate for Rad3ATR and possess SUMO-binding properties. We tested these predictions in the following experiments.

Eme1 N-terminal part (1–117) is essential in absence of Rqh1BLM.

A- Tetrad analysis of an eme1 x rqh1Δ mating, germinated at 30°C. Boxes below dissections indicate the genotypes of each spore. B- Schematic of Eme1 protein. The yellow box depicts the first 117 amino-acid residues required for cell survival in absence of Rqh1BLM. The serine and threonine residues responding to Rad3ATR-consensus phosphorylation sites are depicted in red and the SUMO-Interacting Motifs are represented by blue squares. The serine residues targeted by Cdc2CDK1 are depicted in green.

Eme1 is a direct target of Rad3ATR kinase

Eme1 contains in total eight putative Rad3ATR consensus phosphorylation sites (S23Q, T50Q, S126Q, T145Q, T215Q, S313Q, T384Q, S458Q) (). To determine whether Eme1 is a direct substrate of Rad3ATR we set up in vitro kinase assays with recombinant Mus81-Eme1 and Rad3ATR. Recombinant Mus81(6His)-(MBP)Eme1 was produced in E. coli and affinity purified on Ni++ and amylose resins (). Recombinant (GFP)Rad3ATR was instead transiently overproduced in yeast cells exposed to bleomycin to induce DNA damage and activate Rad3ATR. We used a chk1Δ cds1Δ rad3Δ mutant strain to eliminate the possibility of endogenous checkpoint kinases co-purifying with (GFP)Rad3ATR. As shown in Eme1 was efficiently phosphorylated by (GFP)Rad3ATR. In contrast, an Eme18AQ mutant where all eight SQ/TQ sites are mutated to AQ is barely detected on the autoradiograph. To explore which SQ/TQ sites were important for phosphorylation we tested the impact of mutating Rad3ATR phosphorylation sites in three different clusters (). The strongest effect was seen with mutations in Cluster 1, whilst mutations in Clusters 2 and 3 had a milder impact (). Noticeably, in vitro Rad3ATR-mediated phosphorylation of Eme1 does not require the priming by Cdc2CDK1. Taken together, these data indicate that the S23Q and T50Q sites in the N-terminus of Eme1 are most critical for its in vitro phosphorylation by Rad3ATR.

Rad3ATR directly phosphorylates Eme1 in vitro.

A- Recombinant mus81 was co-expressed with either wild-type eme1, eme1, eme1 or eme1 and purified from bacterial cultures (purification scheme is depicted on the left). Samples were loaded on an 4–12% Nupage Bis-Tris polyacrylamide gel alongside a Rainbow (Amersham) protein size marker. B- Representative autoradiography of 32P labeled Eme1. Rad3ATR in vitro kinase assays were carried out on full-length recombinant Mu81-Eme1 complexes containing either wild-type Eme1 or Eme1 mutated for S23/T50 (cluster1*) or S126/T145/T215 (cluster2*) or S313/T384/S458 (cluster3*) or S23/T50/S126/T145/T215/S313/T384/S458 (eme18AQ). Samples were loaded on a 3–8% Nupage Tris-Acetate polyacrylamide gel alongside a PageRuler (ThermoFisher) protein size marker. C- Relative band intensity of phosphorylated Eme1 (n = 3). All clusters contribute to Eme1 Rad3ATR-dependent phosphorylation with S23/T50 (cluster1*) being the most important. D- Cultures from cdc25-22 TAP-eme1 and cdc25-22 TAP-eme1 were synchronized at the G2/M transition and released for one cell cycle. Total proteins were extracted at each indicated time point of the time course and analyzed by Western blot using an antibody raised against the Flag tag of Eme1. Ponceau stained membranes are depicted as loading control. E- Western blot on total lysates from untreated or 40 μM CPT–treated cells of the indicated background. Tubulin is used as a loading control. F- Scheme depicting the relative contributions of Cdc2CDK1, Rad3ATR and Chk1 kinases in wild-type, chk1Δ and eme1 backgrounds. Note: TAP- = 2xProtA-TEVsite-2xFlag-.

To investigate whether Eme1 is a substrate for Rad3ATR in vivo, we generated an eme1 mutant strain in which all SQ/TQ sites are mutated to AQ (). As expected, whereas the cell-cycle dependent phosphorylation profile of Eme18AQ is comparable to that of the WT protein (), we observed a strong reduction of the CPT-induced phosphorylation (). However, DNA damage-dependent phosphorylation was not totally abolished in the Eme18AQ background (). We suspected that this residual phosphorylation of Eme18AQ was catalyzed by Chk1 (). Accordingly, we observed a complete loss of CPT-induced phosphorylation of Eme18AQ in eme1 chk1Δ cells ().

Overall, our data strongly indicate that Eme1 is phosphorylated by both Chk1 and Rad3ATR following activation of the DNA damage checkpoint.

Phosphorylation of Eme1 by Rad3ATR is crucial in absence of Rqh1BLM

We previously showed that in vivo DNA damage-induced hyperphosphorylation of Eme1 is strictly subordinated to prior Cdc2CDK1-dependent phosphorylation [16]. Accordingly, mutating four CDK consensus target sites totally abrogates not only the cell-cycle dependent phosphorylation of the resulting Eme14SA protein but also its phosphorylation in response to DNA damage [16]. Importantly, the same study found that while an eme1 single mutant displays no abnormal phenotype, an eme1 rqh1Δ double mutant is extremely sick. These observations suggest that the Cdc2CDK1-dependent phosphorylations of Eme1 are required for its phosphorylation by both Rad3ATR and Chk1, and that these events are critical in the absence of Rqh1BLM.

To investigate the functional relevance of phosphorylation of Eme1 by Rad3ATR, we introduced eme1 mutations in the rqh1Δ background. While we observed no obvious phenotype for the eme1 single mutant, the eme1 rqh1Δ double mutant displayed pronounced growth and colony formation defects compared to the rqh1Δ single mutant (Figs ). This genetic interaction was further exacerbated by exposure to genotoxic agents (). Noteworthy, introducing the Eme1 SQ/TQ cluster mutations in the rqh1Δ background resulted in smaller colonies compared to the parental cells (). In line with the prominent contribution of Cluster 1 to Rad3ATR-mediated phosphorylation of Eme1 in vitro (), we noticed a more pronounced effect of Cluster 1 mutations on colony size in absence of Rqh1BLM compared to the other two cluster mutations (). Overall, our data indicate that direct phosphorylation of Eme1 by Rad3ATR is important for cell viability in the absence of Rqh1BLM.

Negative genetic interaction between eme1 and rqh1Δ.

A- Tetrad analysis of an eme1 x rqh1Δ mating, germinated at 30°C. Boxes below dissections indicate the genotypes of each spore. B- Exponentially growing eme1, rqh1Δ and eme1 rqh1Δ cells were seeded on YES plates by micromanipulation and allowed to grow for 3 days at 32°C. C- Average percentage (± s.d.) of viable colonies (n = 3 independent experiments). Statistical significance is measured with one-way ANOVA followed by Tuckey post-test.

Rad3ATR direct phosphorylation of Eme1 contributes to the catalytic stimulation of Mus81-Eme1

We have previously shown that Rad3ATR contributes to the catalytic stimulation of the HJ-resolvase activity of Mus81-Eme1 [16]. We inferred at that time that the Rad3ATR-Chk1 axis was involved in this catalytic control. The finding that Rad3ATR can directly phosphorylate Eme1, and that this phosphorylation is crucial in absence of Rqh1BLM, prompted us to investigate whether it contributed to the catalytic stimulation of Mus81-Eme1.

As previously reported [16], hyperphosphorylation of Eme1 following activation of the DNA damage checkpoint by CPT correlates with increased HJ-resolvase activity of Mus81-Eme1 complex isolated from fission yeast cells expressing TAP-tagged Eme1 (Figs ). Accordingly, no catalytic stimulation is detected when Mus81-Eme1 is recovered from rad3Δ cells ([16] and Figs ). Remarkably, we find that the Mus81-Eme18AQ complex is not stimulated following CPT treatment despite a fully functional DNA damage checkpoint and Chk1-mediated phosphorylation of Eme1 in eme1 cells (Figs ). In contrast to the complete loss of catalytic stimulation of Mus81-Eme1 in absence of Rad3ATR, some degree of catalytic stimulation of Mus81-Eme1 remains detectable in chk1Δ cells following CPT treatment (Figs ). Taken together, these data demonstrate that full-fledged catalytic stimulation of Mus81-Eme1 relies on phosphorylation of Eme1 by both Rad3ATR and Chk1 kinases (). They also suggest that Chk1-mediated phosphorylation contributes to full activation of Mus81-Eme1 only when combined with phosphorylation of Eme1 by Rad3ATR.

Rad3ATR phosphorylation of Eme1 contributes to the catalytic stimulation of Mus81-Eme1 resolvase activity.

A- 32P-labeled (red dot) HJs were incubated for the indicated times with wild type Mus81–Eme1 or mutant Mus81-Eme18AQ complexes recovered from untreated or 40 μM CPT–treated WT, rad3Δ, eme1 or chk1Δ mutant cells as described in Materials and Methods. Reaction products were analyzed by neutral PAGE. Comparable amounts of protein samples were used in each reaction after normalization of their relative concentration (see Materials and Methods and ). B- Quantification of product formation, represented as a percentage of total radiolabeled DNA, in three independent experiments including the one in (A) (See ). C- Average (± s.d.) fold stimulation of HJ resolution by Mus81–Eme1 following CPT–treatment (n = 3 independent experiments, see ). The histogram shows the ratio of the HJ-resolvase activity of Mus81–Eme1 from CPT-treated cells over that of Mus81–Eme1 from untreated cells. D- Model for the Cdc2CDK1-, Rad3ATR- and Rad3ATR-Chk1-dependent phosphorylation of Eme1 and DNA damage-induced catalytic stimulation.

Eme1 contains bona-fide SIMs

Having confirmed our predictions that Eme1 is directly phosphorylated by Rad3ATR and shown that this is critical for catalytic stimulation of Mus81-Eme1 and cell fitness in absence of Rqh1BLM, we next undertook the analysis of the predicted SUMO-binding properties of Eme1 mediated by the putative SIM1 and SIM2 motifs in the N-terminal domain of Eme1 (Figs ).

The SUMO-binding capacities of SIM1 and SIM2 were assessed by a yeast two-hybrid assay against the unique S. pombe SUMO ortholog Pmt3. A fragment of Eme1 (Eme11-130) containing SIM1 and SIM2 displayed strong binding to Pmt3, confirming that the N-terminus of Eme1 possesses SUMO-binding properties (). Introducing point mutations in the conserved aliphatic residues of SIM1 strongly impaired interaction with Pmt3 while mutations in SIM2 had a milder effect (). Mutations in both SIMs led to complete loss of interaction with Pmt3 (). Our data confirm that the N-terminal domain of Eme1, which is essential for cell viability in absence of Rqh1BLM (), contains bona fide SIMs that jointly contribute to the SUMO-binding properties of Eme1, with a predominant contribution made by SIM1.

Eme1 contains bona-fide SIMs.

A- A fragment of Eme1 (eme11-130) containing SIM1 and SIM2 was used as bait in a Yeast two-hybrid assay for interaction with SUMOPMT3. The SIMs consensus sequences are indicated as well as the mutations introduced in each of them. B- Exponentially growing wild type, eme1Δ, eme1, eme1 and eme1 cells were seeded on YES plates by micromanipulation and allowed to grow for 3 days at 32°C. C- Average percentage (± s.d.) of viable colonies (n = 3 independent experiments). Statistical significance is measured with one-way ANOVA followed by Tuckey post-test. D- Five-fold dilutions of wild-type, rqh1Δ, eme1, eme1 and eme1 cells were plated on medium supplemented or not with the indicated concentrations of CPT followed by incubation at 30°C. E- Exponentially growing rqh1Δ, eme1, eme1 and eme1 as well as two independent clones of eme1 rqh1Δ, eme1 rqh1Δ and eme1 rqh1Δ were seeded on YES plates by micromanipulation and allowed to grow for 3 days at 32°C. F- Average percentage (± s.d.) of viable colonies (n = 3 independent experiments). Statistical significance is measured with one-way ANOVA followed by Tuckey post-test.

Eme1 SIMs are required in absence of Rqh1BLM

To assess the functional relevance of SIM1 and SIM2, we generated mutant strains harboring the point mutations, described in in SIM1 (eme1), SIM2 (eme1) or both SIMs (eme1) of Eme1. None of the three eme1, eme1 and eme1 mutant strains presented any obvious growth defect or reduced fitness compared to a WT strain in absence of exogenous DNA damage () and following CPT treatment (). Since eme1 rqh1Δ double mutants are non-viable (), we next assessed the importance of SIM1 and/or SIM2 for cell viability in absence of Rqh1BLM. As shown in , while mutating SIM1 in the rqh1Δ background does not reduce colony formation capacities of the resulting eme1 rqh1Δ double mutant compared to the rqh1Δ single mutant, it leads to a marked increase in the proportion of elongated and sick cells (). In contrast, mutating SIM2 strongly impairs the ability of eme1 rqh1Δ to form viable colonies () in addition to causing a strong increase in the number of sick cells (). Simultaneously mutating both SIMs did not further impair colony formation capacities compared to eme1 rqh1Δ cells (). However, it had an additive effect regarding the proportion of elongated and sick cells ().

We further looked at cell fitness following chronic exposure to CPT. Loss of Rqh1BLM in eme1 mutant slightly exacerbated CPT sensitivity compared to rqh1Δ. This effect was slightly more pronounced for eme1 rqh1Δ mutants while the eme1 rqh1Δ mutants displayed the steepest increase in CPT sensitivity compared to rqh1Δ (). Overall, these data demonstrate that both SIMs contribute to the essential role of the Eme11-117 N-terminal domain in absence of Rqh1BLM, with a more prominent contribution made by SIM2.

Contribution of SIM1 and SIM2 to Eme1 phosphorylation and Mus81-Eme1 stimulation

The importance of the Eme1 SIMs in absence of Rqh1BLM prompted us to assess possible functional ties between the SUMO-binding properties of Eme1 and its phosphorylation. Interestingly, whereas mutating the SIMs had no obvious effect on the phosphorylation profile of Eme1 throughout the cell cycle (), it substantially reduced phosphorylation levels in response to CPT (). It also reduced the catalytic stimulation of the Mus81-Eme1 complex, compared to the WT complex, following CPT treatment (Figs ). This suggested that mutating the SIMs did not abrogate phosphorylation of Eme1 by Rad3ATR, which we showed results in complete loss of DNA damage induced stimulation of Mus81-Eme1 (Figs ). In line with this, Eme1 is efficiently phosphorylated by Rad3ATR in vitro (). To assess whether the residual CPT-induced mobility shift of Eme1 is due to phosphorylation by Rad3ATR in vivo, we generated an eme1 mutant. Mutating all Rad3ATR consensus phosphorylation sites and both SIMs resulted in the complete loss of CPT-induced mobility shift of Eme1 (), confirming that Eme1 remains phosphorylated by Rad3ATR in response to DNA damage. These data suggest that the SIMs promote Chk1-mediated phosphorylation of Eme1 and as such contribute to the catalytic stimulation of Mus81-Eme1, which is critical for cell fitness in absence of Rqh1BLM [16]. To gain further insight into the respective contribution of SIM1 and SIM2 to Chk1-mediated phosphorylation of Eme1 we compared the CPT-induced mobility shifts of Eme1 and Eme1 mutants (). While mutating SIM1 had barely any impact, mutating SIM2 substantially reduced phosphorylation levels in response to CPT (). Based on these findings and the severe growth defect of an eme1 rqh1Δ double mutant (Figs ), we propose that phosphorylation of Eme1 by Chk1, promoted by SIM2, represents an additional phosphorylation-based layer of control of Mus81-Eme1 that is critical for cell viability in absence of Rqh1BLM ().

Mutations within the SIMs differently affect Eme1 phosphorylation.

A- Western blot detection of TAP-Eme1 from cdc25-22 TAP-eme1 and cdc25-22 TAP- eme1 synchronized at the G2/M transition and released for one cell cycle. Total proteins were extracted at each indicated time point of the time course and analyzed by Western blot using an antibody raised against the Flag tag of TAP-Eme1. Cdc2 is used as a loading control. B- Western blot detection of TAP-Eme1 from untreated or 40 μM CPT–treated WT, rqh1Δ and eme1 cells. Cdc2 is used as a loading control. C- 32P-labeled (red dot) HJs were incubated for the indicated times with Mus81-Eme1 complex recovered from untreated or 40 μM CPT–treated eme1 cells as described in Materials and Methods. Reaction products were analyzed by neutral PAGE. Comparable amounts of protein samples were used in each reaction after normalization of their relative concentration (see Materials and Methods and ). A first graph below the gel represents the quantification of product formation, as a percentage of total radiolabeled DNA, in three independent experiments including the one shown above the graph (See ). A second graph shows the average (± s.d.) fold stimulation of HJ resolution by Mus81-Eme1 following CPT–treatment (n = 3 independent experiments, see ). The histogram shows the ratio of the HJ-resolvase activity of Mus81–Eme1 from CPT-treated cells over that of Mus81–Eme1 from untreated cells. D- Western blot detection of TAP-Eme1 from untreated or 40 μM CPT–treated WT, eme1, eme1 and eme1 cells. Ponceau stained membranes are depicted as loading control. E- Western blot detection of TAP-Eme1 from untreated or 40 μM CPT–treated WT, eme1 and eme1 cells. Cdc2 is used as a loading control. F- Model summarizing our findings on the relative contributions of Rad3ATR, Chk1 and the SIMs to phosphorylation of Eme1 and catalytic stimulation of Mus81-Eme1 in response to DNA damage. Note: TAP- = 2xProtA-TEVsite-2xFlag-.

SUMO-binding capacities and phosphorylation cooperate for cell viability in absence of Rqh1BLM

To further assess whether we have unraveled layers of regulation of Mus81-Eme1 that all contribute in their own way to cell survival in absence of Rqh1BLM, we undertook genetic analyses by combining the mutations that impair the SIMs (i.e. eme1) with those that abrogate the cell cycle and DNA damage-dependent phosphorylations of Eme1 (i.e. eme1). As expected, we observed no constitutive Cdc2CDK1-dependent and DNA damage-induced mobility shift of the Eme1 mutant protein (). The resulting eme1 mutant strain displays a slightly reduced ability to form viable colonies in absence of exogenous stress (). Remarkably, while eme1 rqh1Δ () and eme1 rqh1Δ [16] double mutant strains are sick but viable, we were unable to generate viable eme1 rqh1Δ cells (). We next crossed eme1 and rqh1Δ cells to determine to which extent direct phosphorylation of Eme1 by Rad3ATR and the SIM related functions (including phosphorylation by Chk1) contribute to the essential functions of Mus81-Eme1 in absence of Rqh1BLM. As shown in , we were also unable to recover eme1 rqh1Δ cells. This is reminiscent of the synthetic lethal interaction between eme1 and rqh1Δ and suggests that the Eme11-117 N-terminal domain is involved in three regulatory processes that each make key contributions to the essential functions fulfilled by Mus81-Eme1 in absence of Rqh1BLM ().

The combined loss of Eme1 phosphorylation and SUMO-binding properties is lethal in absence of Rqh1BLM.

A- Western blot detection of TAP-Eme1 on total lysates from untreated or 40 μM CPT–treated from WT or eme1 cells. An Eme14SA+ protein shows no constitutive Cdc2CDK1-dependent (-CPT) and DNA damage-induced mobility shift (+CPT). Tubulin is used as a loading control. Note: TAP- = 2xProtA-TEVsite-2xFlag- B- Exponentially growing eme1, eme1 and eme1 cells were seeded on YES plates by micromanipulation and allowed to grow for 3 days at 32°C. C- Average percentage (± s.d.) of viable colonies (n = 3 independent experiments). Statistical significance is measured with one-way ANOVA followed by Tuckey post-test. D- Tetrad analysis of an eme1 x rqh1Δ mating, germinated at 30°C. Boxes below dissections indicate the genotypes of each spore. E- Tetrad analysis of an eme1 x rqh1Δ mating, germinated at 30°C. Boxes below dissections indicate the genotypes of each spore.

Model for the synthetic lethality resulting from combined loss of Eme1 phosphorylation and SUMO-binding properties in absence of Rqh1BLM.

See text for details.

Discussion

In this study we have identified three layers of control of Mus81-Eme1 that are critical in absence of Rqh1BLM (Figs ). A first layer relies on the phosphorylation of Eme1 by Rad3ATR in response to DNA damage and is critical for the catalytic stimulation of Mus81-Eme1 (Figs ). Another layer that drives catalytic stimulation of Mus81-Eme1 relies on the phosphorylation of Eme1 in response to DNA damage by Chk1 (Figs ). It is noteworthy that both these layers require that Eme1 is first phosphorylated by Cdc2CDK1 [16]. The third layer relies on newly described SIM1- and SIM2-dependent SUMO-binding properties of Eme1. Importantly, our results reveal that there is some crosstalk between these layers with, on one hand, Chk1-mediated stimulation of Mus81-Eme1 requiring that Eme1 is also phosphorylated by Rad3ATR and, on the other, SIM2 contributing to phosphorylation of Eme1 by Chk1 (Figs ).

In our initial study we had suggested that catalytic stimulation of Mus81-Eme1 in response to DNA damage relied on the phosphorylation of Eme1 by a classical DNA damage checkpoint signaling mechanism and that it was ultimately driven by phosphorylation of Eme1 by Chk1 [16]. With the demonstration that DNA damage-induced stimulation of Mus81-Eme1 strictly relies on direct phosphorylation of Eme1 by Rad3ATR (Figs ), we are providing new insight into a control mechanism of Mus81-Eme1 that turns out to be substantially more elaborate than initially anticipated.

Indeed, mutating all eight Rad3ATR SQ/TQ phosphorylation sites in Eme1 prevents phosphorylation by Rad3ATR in vitro and in vivo () and fully abrogates DNA damage-induced stimulation of the resulting Mus81-Eme18AQ complex (Figs ), even though Eme18AQ still undergoes Chk1-mediated phosphorylation (). These data point towards an essential role of direct phosphorylation by Rad3ATR and question whether phosphorylation by Chk1 makes any contribution at all to the catalytic stimulation of Mus81-Eme1. However, we also found that in absence of Chk1, Mus81-Eme1 is very poorly stimulated by CPT treatment even though Eme1 is still directly phosphorylated by Rad3ATR (Figs ). Based on these findings, we propose a model whereby full-fledged stimulation of Mus81-Eme1 in fact relies on phosphorylation of Eme1 by both Rad3ATR and Chk1 (Figs ). The complete lack of stimulation of the Mus81-Eme18AQ complex, despite Chk1-mediated phosphorylation of Eme18AQ suggests that stimulation by Chk1 only occurs when Eme1 is also phosphorylated by Rad3ATR. However, we cannot formally exclude at this stage that the 8AQ mutations could have a knock-on effect that impairs phosphorylation by Chk1 (without any obvious impact on the mobility shift of Eme1).

It is noteworthy that all Rad3ATR SQ/TQ phosphorylation sites are located before the coiled-coil domain of Eme1 and sit therefore within an intrinsically disordered part of Eme1. This is particularly relevant for the first two phosphorylation sites found in cluster 1 which are the only two within the Eme11-117 domain. They are the ones that also contribute the most to the phosphorylation of Eme1 by Rad3ATR in vitro and cell fitness in absence of Rqh1BLM (). It will be important to further investigate what structural impact phosphorylation of the N-terminus of Eme1 may have or whether it drives conformational changes by promoting the association with a coactivator and how this stimulates Mus81-Eme1. Hints that the poorly structured N-terminal domain of Eme1 may negatively impact the catalytic activity of Mus81-Eme1 may be seen in the increased in vitro activity that results from clipping off large N-terminal domains of human MUS81 and EME1 that are dispensable for the endonuclease function of the complex [18]. This would also be reminiscent of the auto-inhibition of human MUS81-EME1 by the N-terminal Helix-hairpin-Helix domain within MUS81 that is relieved upon association of the N-terminus of MUS81 with SLX4 [19]. Cryo-EM studies of XPF-ERCC1 also revealed how conformational changes imposed by DNA binding relieved auto-inhibition by the N-terminal helicase domain of XPF [20]. Members of the XPF-family of SSEs, which all carry their catalytic-relevant functions in the C-terminal part of their subunits, appear to have evolved in a way that provides their N-terminal domains with regulatory functions mediated through controlled conformational changes. Based on our findings, such conformational changes driven by phosphorylation seem like a plausible explanation in the case of Eme1.

Another pivotal finding of this study is the identification of two SIMs (SIM1 and SIM2) in the Eme11-117 N-terminal domain. Noteworthy, as observed for the eme1 and eme1 phosphorylation mutants, mutating the SIMs of Eme1 severely impacts cell viability in absence of Rqh1BLM () but does not lead to any increased sensitivity to CPT (). This contrasts with the acute hypersensitivity of mus81Δ and eme1Δ null cells and indicates that the Eme1 SIM-dependent functions are intimately linked to the control of Mus81-Eme1 in relation to Rqh1BLM functions.

We show that mutating the SIMs results in a marked decrease in DNA damage induced phosphorylation of Eme1 by Chk1 and tampers catalytic stimulation of Mus81-Eme1 (Figs ). The drop in catalytic stimulation is less extensive than what might be expected from the pronounced reduction in DNA damage induced phosphorylation of Eme1. It is noteworthy that Eme1 remains phosphorylated by Rad3ATR and there remains the possibility that it undergoes residual Chk1-mediated phosphorylation that does not induce a detectable mobility shift but contributes to catalytic stimulation. Importantly, our genetic analyses indicate that the SIMs fulfil pivotal functions beyond catalytic control of Mus81-Eme1. Indeed, whereas Eme1 mutants unable to undergo catalytic stimulation (eme1 or eme1) or to bind SUMO (eme1) are viable in absence of Rqh1BLM ([16], Figs ), Eme1 mutants that combine both defects (eme1 and eme1) are not (). Such non-catalytic functions of the SIMs may drive the efficient recruitment and stabilization of Mus81-Eme1 at sites where it is needed to process secondary DNA structures that accumulate in absence of Rqh1BLM. In line with this, it was recently proposed that the human SLX4 nuclease scaffold that targets the XPF-ERCC1, MUS81-EME1 and SLX1 SSEs to specific genomic loci contains several SIMs that are involved in its own recruitment to telomeres, PML bodies and DNA damage [21-23]. An alternative and radically opposite explanation could be that the SIMs of Eme1 are involved in a process that negatively controls Mus81-Eme1 by sequestering the nuclease in subnuclear compartments, away from DNA secondary structures such as replication intermediates that could otherwise get opportunistically processed. In absence of Rqh1BLM such structures would accumulate and their premature endonucleolytic processing would be deleterious to the cell. Such compartmentalisation has been observed in human cells with the nucleolar accumulation of MUS81-EME1 in S-phase and its relocalisation out of the nucleolus to sites of DNA damage in replicating cells following UV-irradiation [7].

Remarkably, while we find that SIM1 and SIM2 cooperate to bind SUMO (), we provide evidence that they also fulfil independent functions. First, mutating the strong SUMO binding SIM1 causes the accumulation of sick cells in absence of Rqh1BLM but it barely impacts colony formation, whereas mutating the much weaker SUMO-binder SIM2 significantly reduces colony formation in addition to causing the accumulation of sick cells. Mutating both SIMs synergistically increased the proportion of sick cells (Figs ). Furthermore, we also found that SIM2 but not SIM1 promotes Chk1-dependent phosphorylation of Eme1 in response to DNA damage (). These results not only suggest that each SIM fulfils different functions, they also question whether some of those fulfilled by SIM2 might extend beyond SUMO-binding when putting into perspective the SIM2-specific phenotypes and its relatively poor affinity for SUMO compared to SIM1. In line with this, we found that Eme1 still undergoes DNA damage-triggered phosphorylation in mutant cells that lack the Pli1 and Nse2 SUMO E3 ligases or that do not produce SUMO (). At first glance, this might indicate that the processes that lead to the phosphorylation of Eme1 in response to DNA damage do not involve SUMO, including those that rely on SIM2. However, the possibility remains that in absence of SUMO, DNA damage-induced phosphorylation of Eme1 is entirely driven by Rad3ATR, which does not rely on Eme1 SIMs (). Further work is therefore needed to formally establish whether SIM2 promotes phosphorylation of Eme1 by Chk1 through interaction with SUMO or not. While we cannot exclude that the mutations introduced in SIM2 induce structural changes that impact more than just SUMO-binding, should SUMO not be involved there remains the exciting possibility that SIM2 could drive the association of Eme1 with a partner that contains a SUMO-like domain (SLD) [24]. The only SLD-containing protein described so far in S. pombe is the Rad60 genome stability factor that contains two SLDs, each of which interact with different players of the SUMO pathway [25]. Interestingly, the presumed S. cerevisiae Rad60 ortholog Esc2 has been reported to interact with Mus81 via its SLDs and to stimulate the Mus81-Mms4 complex [26]. In addition, Esc2 was recently found to promote the degradation of phosphorylated Mms4 [15]. We can exclude similar scenarios involving Rad60 as none of the SIMs were found to modulate the levels of Eme1 or phosphorylated Eme1. However, based on such functional promiscuity between Mus81-Mms4 and the SLD-containing Esc2 protein, it is tempting to see Rad60 as the ideal candidate for a SIM2-mediated partner of Eme1 that would promote Chk1-dependent phosphorylation of Eme1 in response to DNA damage. This would not be the first example of regulatory processes that involve similar players but different outcomes for Mus81-Eme1 and Mus81-Mms4.

Overall, our findings show that the poorly structured N-terminal domain of Eme1 harbors essential regulatory functions of Mus81-Eme1, the control of which appears to be remarkably more elaborate than initially described. With the demonstration that it relies on three regulatory layers that together contribute to the vital functions it fulfils in cells lacking Rqh1BLM, we are setting the basis for new lines of investigation that should contribute to a better understanding of the contributions made by Mus81-Eme1 in the maintenance of genome stability.

Materials and methods

Fission yeast strains, media, techniques and plasmids

Fission yeast strain genotypes are listed in . Media and methods for studying S. pombe were as described elsewhere [27].

The eme1 mutants (eme1, eme1, eme1, eme1, eme1) were generated as follows. The Eme1 genomic locus from strain PH81 (h+, leu1-32 ura4-D18 TAP-eme1 mus81:13Myc-KanMX6), which produces N-terminally TAP-tagged Eme1, was subcloned into a TopoTA vector (Invitrogen). Point mutations were introduced on that TopoTA-EME1 vector by using a Multiprime site-directed mutagenesis kit (Stratagene). Mutations were confirmed by DNA sequencing. The mutated Eme1 genomic locus from the TopoTA-Eme1 vector was used as template for PCR using primers forward 5’–acccatctctcacctaacc– 3’ and reverse 5’- cagtattagcttacagcc– 3’. The PCR fragment was then used to transform strain PH41, in which a URA4 cassette replaces the start codon of EME1 gene. 5′-FOA–resistant clones were selected and confirmed as TAP-eme1 mutant producing strains by genomic DNA sequencing.

Cell synchronization

For synchronization of cells by cdc25-22 block and release, cells containing the temperature-sensitive cdc25-22 allele were grown to exponential phase at permissive temperature (25°C) and shifted at restrictive temperature (36°C) for 3.5 h to arrest the cell cycle in G2. Upon release to permissive temperature (25°C), the cells synchronously enter the cell cycle. Cells were collected and processed every 20 min. Progression into S phase was monitored microscopically by counting cells that contained septa using calcofluor (Sigma) staining, the appearance of which correlates with S phase.

For DNA damage studies, Bleomycin (Merck) was added to cells arrested at the G2/M transition and further incubated at 36°C for 1,5h.

Colony formation assay

Fresh cultures were re-seeded on YES plates by micromanipulation and allowed to grow for 3 days at 32°C. At least 3 independent experiments were performed and averaged. Statistical significance (1 way ANOVA and Tuckey test) is displayed on each graph.

Yeast two-hybrid

Yeast strains were derived from EGY48 (MATα, ura3, his3, trp1, and LexA-LEU2) containing the pSH18-34 (LexA-LacZ, URA3, and amp) plasmid. The fission yeast complementary DNA (cDNA) coding for the first 130 amino acid was cloned into the pJG4-5 (B42-AD, TRP1, and amp) and the pEG202 (LexADNA-BD, HIS3, and amp) and cotransformed into EGY48 + pSH18-34 strain and plated onto -URA-TRP-HIS medium. To monitor protein interaction, clones were spotted onto 3% Gal-URA-TRP-HIS-LEU and 3% Gal-URA-TRP-HIS-Xgal (80 μg/ml) plates. Plates were incubated at 30°C for 2 to 4 days.

Protein extraction, immunoprecipitation and immunoblotting

Cellular lysates were prepared from exponentially growing cell cultures treated with 40μM camptothecin (Sigma) or 5μg/ml bleomycin (Merck). Denatured cell lysates were prepared by TCA precipitation. Cells were suspended in 20% TCA and lysed mechanically using glass beads (Sigma). Following centrifugation, the TCA precipitate was suspended in SDS-PAGE loading buffer (Invitrogen) containing Tris-base. Protein extracts were directly resolved on Tris-acetate 3–8% polyacrylamide NuPAGE gels (Invitrogen). Proteins were transferred to a nitrocellulose Hybond-C membrane (Invitrogen). The membrane was blocked in PBS-T milk 5% and probed by using anti-Flag (Sigma F1804) antibody (1:5,000 dilution), anti Cdc2 (Santa-Cruz sc-53) antibody (1,1000 dilution), anti-tubulin alpha T5168 (Sigma).

Recombinant MBP-Eme1-Mus81-6-His production and purification

The cDNA of eme1 and mus81 were subcloned into the pMBP-parallel1 and pCDFDuet-1 plasmids respectively using in-fusion (Takara) cloning system. A 6-His tag was inserted in frame with mus81 ORF for C-terminal of the protein. Plasmids were co-transformed in Rosetta (DE3) pLysS cells. The expression of MBP-Eme1 and Mus81-6His was carried by growing the cells into auto-induced media (Formedium) at 37°C. Cells were harvest at 4°C, resuspended in PBS 1X and kept at -20°C. Lysis buffer 2X (100mM Tris-HCl, pH = 8.0, 300mM NaCl, 20% glycerol, 0,2% NP-40, 2mM PMSF, 2mM ß-Mercaptoethanol, protease inhibitor cocktail complete EDTA-free (Roche), 10 mg/mL lysozyme, 20mM imidazole) was added to lysed cells before incubation for 20 min at 4°C. The lysate was cleared by centrifugation before incubation on Ni2+ agarose beads (Qiagen) for 2h at 4°C. The beads were washed with 5 volumes of lysis buffer 1X and eluted with lysis buffer 1X supplemented with 250mM imidazole. Eluted complexes were incubated with amylose beads (NEB) for 2h at 4°C. The beads were washed 3 times in lysis buffer 1X and twice in kinase buffer (25mM HEPES-KOH, pH = 7.5, 50mMKCl, 10mM MgCl2, 10mM MnCl2, 2% glycerol, 0,1% NP-40, 50mM NaF, 1mM Na3VO4, 50mM ß-glycerophosphate, 1mM DTT). Complexes were eluted in kinase buffer supplemented with 10mM maltose. Proteins were aliquoted, snap-frozen in liquid nitrogen and stored at -80°C for long term storage.

GFP-Rad3ATR production and in vitro kinase assay

GFP-Rad3ATR was transiently expressed from the full-length nmt41 promoter from cds1Δ chk1Δ rad3Δ cells treated 2h with bleomycin. Cells pellets were disrupted using a Ball Mill (Retsch) in presence of liquid Nitrogen. Resulting powder was resuspended in 2 volumes/weight of lysis buffer (50mM Tris-HCl, pH = 8.0, 500mM NaCl, 10% glycerol, 1% NP-40, 50mM NaF, 1mM Na3VO4, 50mM ß-glycerophosphate (Sigma), 2mM PMSF, 1mM DTT, protease inhibitor cocktail complete EDTA free (Roche)). Lysates were cleared by centrifugation before incubation on GFP-TRAP agarose beads (Chromotek) at 4°C for 1h. Beads were washed three times with lysis buffer and twice with kinase buffer (25 mM HEPES-KOH, pH = 7.5, 50mM KCl, 10mM MgCl2, 10mM MnCl2, 2% glycerol, 0,1% NP-40, 50mM NaF, 1mM Na3VO4, 50mM β-glycerophosphate, 1mM DTT). GFP-Rad3ATR was kept attached on beads for the following kinase assays.

GFP-TRAP-bound GFP-Rad3ATR was resuspended in kinase buffer supplemented 100 μM of cold ATP before addition of 10 μCi γ32P-ATP and substrates. After 30 min at 30°C, reactions were stopped by the addition of 15 μL of 4X SDS sample buffer. Samples were denatured and resolved by SDS-PAGE electrophoresis. Following Coomassie staining, gel was dried and expose with phosphorimager.

In vitro nuclease assay

TAP-Eme1 (2xProtA-TEVsite-2xFlag-Eme1) was affinity purified and used in nuclease assays on X12 mobile HJ as previously described [28]. Briefly, cell pellets were resuspended in 1 volume to weight lysis buffer (50mM Tris-HCl, pH = 8.0, 150 mM NaCl, 10% glycerol, 0,1% NP-40, 50mM NaF, 50mM β-glycerophosphate (Sigma), 2mM PMSF, protease inhibitor cocktail complete EDTA free (Roche)). Cells were subjected to mechanical lysis using a Ball Mill (MM400 Retsch). For this, usually, 4 to 5 ml of the cell suspension were poured into grinding chambers precooled in liquid nitrogen. The frozen cell pellet was disrupted by 2 agitation runs at 30hz. The resulting powder was resuspended in another 1 volume to weight of lysis buffer and centrifuged. Clear supernatant was loaded onto IgG sepharose beads (Cytiva) for 2h at 4°C (20 μl packed beads used for 4 ml of lysate). After extensive washes, proteins were eluted in presence of 60 μl of AcTEV protease for 1h at RT. In order to determine the relative amount of Mus81-Eme1 between different samples, 3 μl of each TEV eluate was treated with phosphatase before SDS-PAGE and Western blot analysis in order to collapse the Eme1 signal into a single band of dephosphorylated Eme1. The relative intensity of the dephosphorylated Eme1 band was quantified for each sample using the ImageLab software. Dilution folds were calculated and used to bring the concentration of each sample down to that of the least concentrated sample. 3 μl of each normalized sample was used in nuclease assays with 32P-labelled DNA substrates as previously described [28].

TCA (Sigma T6399)

Glass beads (Sigma G8772)

Calcofluor (Sigma 18909)

S-(+)-Camptothecin (Sigma C9911)

Bleomycin (Calbiochem 9041-93-4)

Hydroxyurea (Sigma H8627)

Protease inhibitor Cocktail EDTA-free (Roche 11873580001)

Anti Flag (Sigma F1804)

Anti Cdc2 (Santa-Cruz sc-53)

Anti tubulin (Sigma T5168)

GFP-TRAP magnetic-agarose (Chromotek gtma-20)

Ni-NTA agarose beads (Qiagen 70666–3)

Amylose resin (NEB E8021S)

Phosphatase (NEB P0753S)

IgG Sepharose 6 Fast Flow (Cytiva 17-0969-01)

AcTEV (Invitrogen 12575015)

The Rad3ATR-consensus sites are widespread throughout Eme1 sequence and were subdivided in three clusters (Upper panel).

All Rad3ATR-consensus sites are mutated in Alanine to generate eme18AQ mutant (Lower panel).

(PDF)

Click here for additional data file.

A- Exponentially growing cultures of eme1, rqh1Δ and eme1 cells were heated-fixed and observed by fluorescent-microscopy using DAPI staining. Cells were classified based on their morphologies. Small G2 cells (Type 1), elongated bi-nucleated and/or septated cells (Type 2) and sick cells (Type 3). B- Five-fold dilutions of cells with the indicated genotype were plated on medium supplemented or not with the indicated concentrations of CPT, HU and MMS followed by incubation at 30°C. C- Tetrad analysis of an eme1 x rqh1Δ mating, germinated at 30°C. To assess the impact of introducing the cluster mutations (C*) in an rqh1Δ background, the ratio of the colony diameter of the double mutants C* rqh1Δ mutants over that of the rqh1Δ mutant were calculated for each tetrad and plotted on the graph below the tetrads.

(PDF)

Click here for additional data file.

A- Western Blot on the indicated TEV-eluates after phosphatase treatment and normalization of their relative concentration, as described in Materials and Methods.

B- Neutral PAGE showing 32P-labeled (red dot) HJs incubated for the indicated times with Mus81–Eme1 complexes recovered from untreated or 40 μM CPT–treated TAP-eme1 (“wild type”) cells as described in Materials and Methods. Comparable amounts of TEV eluates were used in each reaction after normalization of their relative concentration (see Materials and Methods and ). Graphs below the autoradiographs represents the quantification of product formation, as a percentage of total radiolabeled DNA, for each experiment. Note: Schematic depicting the migration profile of uncleaved and cleaved radiolabelled HJs. Yellow dotted boxes represent areas of the gel used for quantification purposes and calculation of the percentage of cleavage product. The “Bckg” box corresponds to background signal that was subtracted from the “Uncut” and “Cut” signals. C- Same as A- but with Mus81–Eme1 from untreated or 40 μM CPT–treated TAP-eme1 rad3Δ cells. D- Same as A- but with Mus81–Eme1 from untreated or 40 μM CPT–treated TAP-eme1 eme1 cells. E- Same as A- but with Mus81–Eme1 from untreated or 40 μM CPT–treated TAP-eme1 chk1Δ cells. Note: TAP- = 2xProtA-TEVsite-2xFlag-.

(PDF)

Click here for additional data file.

A- Cells from exponentially growing cultures of the indicated genotypes were heated-fixed and observed by fluorescent-microscopy using DAPI staining (same as S2), counted and sorted depending on their morphologies.

B- Five-fold dilutions of cells with the indicated genotype were plated on medium supplemented or not with the indicated concentrations of CPT followed by incubation at 30°C.

(PDF)

Click here for additional data file.

Same as for S3B Fig but for Mus81-Eme1 complex from untreated or 40 μM CPT–treated eme1 cells.

(PDF)

Click here for additional data file.

Recombinant Mus81 was co-produced in E. coli with either wild-type Eme1 or Eme1 and purified as described in Fig 2A.

Rad3ATR in vitro kinase assays were carried out on the corresponding recombinant Mu81-Eme1 and Mus81-Eme1 complexes. A representative autoradiography of 32P labeled Eme1 is shown. Relative band intensity of phosphorylated Eme1 (n = 3) shows that Eme1 is efficiently phosphorylated in vitro by Rad3ATR.

(PDF)

Click here for additional data file.

Western blot detection of TAP-Eme1 from untreated or 40 μM CPT treated WT, pli1Δ, nse2, pli1Δ nse2 and pmt3Δ cells.

Western blot using an antibody raised against the Flag tag of TAP-Eme1. Tubulin is used as a loading control. Note: TAP- = 2xProtA-TEVsite-2xFlag-.

(PDF)

Click here for additional data file.

Complete list of S. pombe strains used in this study.

(XLSX)

Click here for additional data file.