sign in sign up
<<
Home , Holliday junction

Molecular Cell Article

Cooperative Control of Holliday Junction Resolution and DNA Repair by the SLX1 and MUS81-EME1

Dennis Castor,1,4 Nidhi Nair,1,4 Anne-Ce´ cile De´ clais,2 Christophe Lachaud,1 Rachel Toth,1 Thomas J. Macartney,1 David M.J. Lilley,2 J.Simon C. Arthur,3 and John Rouse1,* 1MRC Protein Phosphorylation and Ubiquitylation Unit 2CR-UK Structure Research Group 3Division of Cell Signalling and Immunology College of Life Sciences, University of Dundee, Dundee DD1 5EH, Scotland, UK 4These authors contributed equally to this work *Correspondence: [email protected] http://dx.doi.org/10.1016/j.molcel.2013.08.036 This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-No Derivative Works License, which permits non-commercial use, distribution, and reproduction in any medium, provided the original author and source are credited.

SUMMARY causes pronounced hypersensitivity to agents that induce DNA ICLs (Fekairi et al., 2009; Mun˜ oz et al., 2009; Svendsen et al., Holliday junctions (HJs) are X-shaped DNA struc- 2009). The importance of SLX4 in ICL repair in humans is under- tures that arise during , scored by the observation that biallelic mutations in SLX4/FANCP which must be removed to enable cause Fanconi anemia (FA) (Kim et al., 2011; Stoepker et al., segregation. The SLX1 and MUS81-EME1 nucleases 2011), a cancer predisposition syndrome accompanied by devel- can both process HJs in vitro, and they bind in close opmental, skeletal, and hematological defects (Auerbach, 2009). proximity on the SLX4 scaffold, hinting at possible Despite the clear importance of the SLX4 complex in DNA repair, little is known about the underlying molecular mecha- cooperation. However, the cellular roles of mamma- nisms. For example, it is not yet known if SLX1 is involved in lian SLX1 are not yet known. Here, we use mouse ICL repair, partly because depletion of SLX1 from human cells de- genetics and structure function analysis to investi- stabilized SLX4, preventing the functional analysis of SLX1 (Mu- gate SLX1 function. Disrupting the murine Slx1 and n˜ oz et al., 2009). Furthermore, it is not yet known how SLX4 af- Slx4 genes revealed that they are essential for HJ res- fects the associated nucleases, and there has been some olution in mitotic cells. Moreover, SLX1 and MUS81- debate about whether the exquisite hypersensitivity of SLX4- EME1 act together to resolve HJs in a manner that defective cells reflects the loss of regulation of one or more of requires tethering to SLX4. We also show that these nucleases. In this light, a recent study concluded that the SLX1, like MUS81-EME1, is required for repair of role of SLX4 in ICL repair involves XPF-ERCC1 only, because a DNA interstrand crosslinks, but this role appears to fragment of SLX4 lacking amino acids 1–499, that did not interact be independent of HJ cleavage, at least in mouse with XPF-ERCC1, did not rescue the mitomycin-C (MMC) sensi- tivity of Slx4 hypomorphic MEFs (Crossan et al., 2011). However, cells. These findings shed light on HJ resolution in the first 499 amino acids of SLX4 also contain two ubiquitin-bind- mammals and on maintenance of genome stability. ing domains that are vital for ICL repair but that are not required for SLX4 to interact with XPF (Kim et al., 2011; Stoepker et al., 2011). INTRODUCTION Two studies used SLX4 deletion mutants lacking the C-terminal helix-turn-helix (HtH) domain to investigate the importance of SLX4 coordinates a multiprotein complex that is important for SLX1 binding for ICL repair. One study found that the HtH deletion DNA repair. In metazoans, this complex includes three struc- mutant fully rescues the sensitivity of Slx4 hypomorphic MEFs ture-selective nucleases: XPF-ERCC1, MUS81-EME1, and (Crossan et al., 2011), whereas the other showed that this mutant SLX1 (Andersen et al., 2009; Fekairi et al., 2009; Mun˜ oz et al., only partly rescued the MMC sensitivity of SLX4-defective FA 2009; Saito et al., 2009; Svendsen et al., 2009). Together these cells (Kim et al., 2013). Therefore, the functional relevance of nucleases confer the complex with the ability to cleave a wide the binding of nucleases to SLX4 in ICL repair remains unclear. range of branched DNA structures in vitro, which mimic DNA The SLX4 complex is capable of processing Holliday junctions intermediates that occur during the repair of damaged DNA and (HJs) in vitro (Fekairi et al., 2009; Mun˜ oz et al., 2009; Svendsen broken DNA replication forks. Both MUS81-EME1 and XPF- et al., 2009). HJs are four-way DNA junctions at which ERCC1 are required for the repair of DNA interstrand crosslinks two chromatids are topologically intertwined. These structures (ICLs) in mammalian cells, and the latter is also required for repair arise during homologous recombination (HR), a process required of UV-induced lesions (Ciccia et al., 2008). Depletion of SLX4 from for repairing unscheduled double-strand breaks (DSBs) or human cells using siRNA duplexes does not affect UV repair but damaged replication forks in mitotic cells. HJs are also key

Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors 221 Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

intermediates during meiotic recombination (Schwartz and gene by eliminating the transcription start site and the remainder of Heyer, 2011; West, 2009). Ultimately HJs must be removed to the first exon (Figure S1A). Gene disruption was confirmed by enable , and two distinct modes of Southern blotting and PCR (Figures S1B and S1C). Both Slx1+/ HJ removal have been identified in mammalian cells (Schwartz and Slx1/ mice were born at Mendelian frequencies (Table S1) and Heyer, 2011). The first pathway involves the dissolution of without overt morphological, developmental, or hematological de- double HJs (dHJs) by the BTR complex (BLM-TOPIII-RMI1- fects. Adult mice were fertile (data not shown). SLX1 protein was RMI2). The coupled helicase and topoisomerase activities of undetectable in extracts of Slx1/ mouse embryonic fibroblast BLM and TOPIII, respectively, disassemble HJs, resulting exclu- extracts (MEFs) (Figure 1A) or testis extracts (Figure S1D) by west- sively in noncrossover products (Chaganti et al., 1974; Wu and ern blotting. The expression levels of SLX4, ERCC1, and MUS81 Hickson, 2003, 2006). This pathway dominates in mitotic cells, proteins, however, were normal in Slx1/ MEFs (Figure 1A). possibly because minimizing crossovers lowers the incidence We also disrupted the murine Btbd12 (Slx4) gene (Figure S2A); of loss of heterozygosity (LOH) that would increase disease gene disruption was confirmed by Southern blotting, PCR (Fig- risk and impair organism fitness (LaRocque et al., 2011). ures S2B and S2C), and western blotting (Figure S2D). The Alternatively, HJs can be resolved by nucleases (Schwartz and expression levels of SLX4-associated proteins ERCC1 and Heyer, 2011;West,2009).Dependingonthe symmetry ofthe cleav- MUS81 were normal in MEFs from the Slx4/ mice but SLX1 age, crossover or noncrossover products may occur. Cells from protein was undetectable by western blotting (Figure 1A). This Bloom syndrome (BS) patients lacking BLM show a large increase suggests that SLX4 regulates SLX1 protein stability, and there- in the frequency of sister chromatid exchanges (SCEs), which are fore SLX4 null mice lack both SLX1 and SLX4 proteins. Initially, thought to result from the crossovers generated by nucleolytic res- no viable Slx4/ offspring were obtained from crossing Slx4 het- olution of dHJs that escape dissolution (Chaganti et al., 1974; erozygotes. However, after backcrossing the heterozygotes five Wechsler et al., 2011). To date, three nuclear HJ resolving activities times, we obtained viable Slx4/ mice, albeit at sub-Mendelian have been identified in mammalian cells: MUS81-EME1, SLX1, and frequencies (Table S2). Slx4/ mice were on average around GEN1 (Bailly et al., 2010; Fekairi et al., 2009; Ip et al., 2008; Mun˜ oz 10%–15% smaller than heterozygotes or wild-type mice at the et al., 2009; Svendsen et al., 2009; Taylor and McGowan, 2008; age of 6 weeks (data not shown). No overt developmental or Wechsler et al., 2011; West, 2009). GEN1 cleaves HJs symmetri- morphological defects were observed. Although mating Slx4/ cally to produce nicked linear duplex products (Ip et al., 2008; males with Slx4/ females resulted in viable progeny, testes in Rass et al., 2010), whereas SLX1 introduces a mixture of symmetric males were on average 47% smaller than in wild-type mice at and asymmetric cuts across the junction (Fekairi et al., 2009; Mu- 10 weeks of age (data not shown). Smaller testis size is in line n˜ oz et al., 2009; Svendsen et al., 2009). In contrast to GEN1 and with two previous reports describing a hypomorphic Slx4 mouse SLX1, MUS81-EME1 does not cleave intact HJs efficiently,butpre- strain made by the European Conditional Mouse Mutagenesis fers nicked junctions and recombination intermediates such as Program (EUCOMM) (Crossan et al., 2011; Holloway et al., extended D-loop structures (Boddy et al., 2001; Ciccia et al., 2011). It is worth noting that the EUCOMM mice were reported 2003; Doe et al., 2002; Gaillard et al., 2003; Whitby et al., 2003). to exhibit phenotypes reminiscent of those seen in FA patients The elevated SCE frequency in BS cells requires GEN1 and such as aplastic anemia (Crossan et al., 2011), even though MUS81-EME1 (Wechsler et al., 2011). However, although GEN1 mouse knockouts of other FA genes did not recapitulate most cleaves intact HJs efficiently in vitro, MUS81-EME1 does not. of these phenotypes (Bakker et al., 2013). However, we saw no Instead MUS81-EME1 shows a strong preference for nicked evidence of an FA-like syndrome in the Slx4/ or Slx1/ mice HJs, suggesting that it might act on junctions that are subjected we generated in this study (data not shown). to prior nicking by a different . One possible candidate is SLX1 because it can process intact HJs efficiently in vitro. Fur- SLX1 Nuclease Activity Is Required for Repair thermore, SLX1 and MUS81-EME1 bind to the HtH motif and the of DNA ICLs SAP domain of SLX4, respectively (Fekairi et al., 2009; Kim et al., We next investigated if SLX1 is involved in ICL repair assessed by 2013; Svendsen et al., 2009). The close proximity of the SLX4 SAP hypersensitivity to genotoxins that induce ICLs. As shown in Fig- and HtH domains places SLX1 close enough to MUS81-EME1 on ure 1B, Slx1/ MEFs are hypersensitive to ICL-inducing agents SLX4 to suggest the possibility of cooperation between the two such as nitrogen mustard (HN-2) and MMC, and embryonic nucleases. In this sense, one function of SLX4 in its capacity as a stem cells (ESCs) from Slx1/ mice were also hypersensitive scaffold would be to facilitate serial processing of HJs, a possibility to MMC (Figure S3A). The sensitivity of Slx1/ MEFs to agents raised previously (Svendsen et al., 2009). In this study, we investi- that induce ICLs was much less pronounced than Slx4/ MEFs, gated whether SLX1 is involved in HJ resolution and ICL repair, and probably because SLX4 binds to several nucleases involved in we tested potential cooperation with MUS81-EME1 and the func- ICL repair in addition to SLX1. Slx1/ MEFs and ESCs were not tional significance of binding of these nucleases to SLX4. more sensitive to camptothecin (CPT), ionizing radiation (IR), hydroxyurea (HU), or UV light than wild-type cells (Figures S3A RESULTS and S3B). Defects in the repair of ICLs often result in chromo- some abnormalities (Auerbach, 2009). In this light, we observed Disruption of the Murine Giyd2/Slx1 and Btbd12/Slx4 a slight increase in the number of chromosome abnormalities Genes such as chromatid breaks and radial structures in Slx1/ MEFs There is currently no information on SLX1 function in mammals. To exposed to MMC compared with wild-type cells, and a much study the invivo roles of SLX1, wedisruptedthe murineGiyd2 (Slx1) larger increase in Slx4/ MEFs (Figure S3C).

222 Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

A Slx1-/- Figure 1. SLX1 Is Involved in the Repair of WT #1 #2 Slx4-/- DNA ICLs 191- (A) Western blot analysis of MEF extracts from SLX4 wild-type (WT), Slx1/, and Slx4/ mice. Two 39- separate Slx1/ MEF clones were tested. (See ERCC1 also Figures S1 and S2.) 64- (B) Clonogenic survival analysis of MEFs exposed MUS81 to genotoxins. For each genotype, cell viability of 39- untreated cells is defined as 100%. Data are SLX1 28- represented as mean ± SEM, n = 3. (See also 39- Figure S3.) GAPDH (C) HEK293 cells stably expressing tetracycline- inducible (‘‘+’’) GFP-tagged mouse SLX4 were B 100 100 transfected with mouse SLX1 (WT), SLX1 active site mutant E79A, or left untransfected (‘‘’’). Cell WT extracts were subjected to western blotting (left panel). Anti-FLAG immunoprecipitates were incu- Slx1-/- #1 10 10 bated with a synthetic radiolabeled four-way -/- Slx1 #2 junction containing a 12 bp homologous core. % survival Slx4-/- Reaction products were subjected to denaturing 1 PAGE (right panel). 1 / 0 10203040 0.00 0.25 0.50 0.75 1.00 (D) Western blot analysis of Slx1 MEFs com- MMC (ng/ml) HN-2 (µM) plemented with untagged versions of wild-type SLX1, SLX1 E79A, or empty vector (top panels). ip: FLAG Extracts were subjected to immunoprecipitation C GFP-SLX4: - + ++ with anti-SLX4 antibodies, and precipitates were GFP-SLX4: - + ++ FLAG-SLX1: -- WT E79A probed with the antibodies indicated (bottom panels). *, IgG light chain. FLAG-SLX1: -- E79AWT (E) Clonogenic survival analysis of complemented 191- GFP Slx1/ MEFs from (D) exposed to increasing

A doses of MMC. For each genotype, cell viability of 39- T C G SLX1 T untreated cells was defined as 100%. Data are G 28- HJ A T represented as mean ± SEM, n = 3. 39- A GAPDH * A G C A

homologous core C C G T MEFs were infected with viruses express- ing untagged SLX1 wild-type or E79A, or -/- DEWT Slx1 100 with empty virus; mutation of E79 in SLX1 empty empty SLX1 E79A SLX1 did not affect interaction with SLX4 191- (Figure 1D). As shown in Figure 1E, wild- SLX4 10 type SLX1, but not the E79A mutant, 39- rescued the MMC hypersensitivity of SLX1 % survival / 28- Slx1 MEFs. Together, these data pro- extracts 39- 1 vide evidence that SLX1 is involved in GAPDH 010203040 MMC (ng/ml) repair of DNA ICLs in mammals. 191- SLX4 WT + empty vector SLX1 Is Required for the -/- 39- Slx1 + empty vector Nucleolytic Processing of HJs SLX1 Slx1-/- + SLX1 SLX1 is capable of processing HJs in vitro

ip: SLX4 28- * Slx1-/- + SLX1 E79A (Fekairi et al., 2009; Mun˜ oz et al., 2009; Svendsen et al., 2009). We next investi- gated whether SLX1 is required for HJ We next tested if the nuclease activity of SLX1 is required for resolution in vivo using the elevated SCE frequency observed cellular resistance to ICL-inducing agents. To this end we in BLM-depleted cells as readout (Figure 2A). To this end, we mutated a highly conserved residue, Glu79, in the SLX1 URI- used shRNA-expressing to deplete BLM from type nuclease domain to alanine. This mutation abolished the Slx1/ MEFs (Figure S4A) (Sfeir et al., 2009). As shown in Fig- activity of FLAG-tagged mouse SLX1 immunoprecipitated from ure 2B, BLM protein was undetectable in MEF extracts when HEK293 cells in the cleavage of a radioactively [50-32P]-labeled cells were infected with expressing the BLM shRNA, HJ with a core that could undergo a number of steps of branch but not when empty virus was used. Depletion of BLM from migration, thereby presenting all possible dinucleotides at the wild-type MEFs caused an increase in SCE frequency from 8.3 point of exchange (Figure 1C) (Mun˜ oz et al., 2009). Next, Slx1/ SCE per metaphase to 18 SCE per metaphase (Figure 2C), but

Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors 223 Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

AB

CD

E

F

Figure 2. Defective HJ Resolution in Slx1–/– Cells (A) Representative image of a metaphase spread for SCE analysis in MEFs after depletion of BLM with shRNA-expressing retroviruses. SCE events are indicated by arrowheads. (B) Western blot analysis of MEFs, of the genotypes indicated, infected with retroviruses expressing a BLM-specific shRNA (+) or with virus prepared with empty vector as control (). (legend continued on next page)

224 Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

in two separate clones of Slx1/ MEFs, the SCE frequency after SCE frequency induced by MMC in cells lacking both MUS81 BLM depletion was diminished to around 12 SCE per meta- and SLX1 was indistinguishable from control cells (Slx1+/ phase. A similar defect in SCE was observed in Slx4/ MEFs Mus81+/) and the Slx1/ or Mus81/ single knockouts. We (Figure 2C), in agreement with a previous analysis of BS cells also compared the MMC sensitivity of Slx1/ Mus81/ MEFs depleted of SLX4 (Wechsler et al., 2011). The SCE defect in with the respective single knockouts in clonogenic survival Slx1/ MEFs depleted of BLM was rescued by expression of assays. As shown in Figures 3B, the MMC hypersensitivity of wild-type SLX1 but not by the nuclease-inactive SLX1 E79A Mus81/ cells was slightly more pronounced than Slx1/ cells, mutant (Figure 2D). Therefore, SLX1 is required for nucleolytic and Slx1/ Mus81/ cells were not more sensitive than the resolution of HJs in vivo. most sensitive of the single knockouts (Mus81/). Similar We also analyzed SCEs formed during the repair of ICLs. results were obtained using alternative clones of MEFs (Fig- Exposure of wild-type MEFs to MMC caused a dose-dependent ure S4C). Taken together, these observations suggest that increase in SCE frequency, but Slx1/ MEFs were indistinguish- SLX1 and MUS81-EME1 are epistatic in terms of ICL repair, at able from wild-type cells in this regard (Figures 2E and 2F). In least judged by hypersensitivity to MMC, but they are not contrast, the MMC-induced increase in SCE frequency was required for SCE induced by ICLs in MEFs. reduced by more than 30% in Slx4/ MEFs (Figures 2E and We went on to explore the possibility that SLX1 and MUS81- 2F). Thus, although SLX4 contributes to the generation of SCE EME1 are also epistatic in the resolution of HJs. To this end, we during ICL repair, SLX1 does not appear to be involved. This depleted BLM from MEFs of various genotypes using shRNA- could be due, in principle, to redundancy between SLX1 and expressing retroviruses (Figure 3C). Depletion of BLM from nucleases such as MUS81-EME1, as discussed below. control MEFs caused an increase in SCE frequency from around 7 SCE per metaphase to around 20 SCE per meta- SLX1 Is Epistatic with MUS81-EME1 in ICL Repair and phase. In Slx1/ or Mus81/ single knockout MEFs, the HJ Resolution SCE frequency after BLM depletion was diminished to around The phenotypes associated with disruption of Slx1 in mice 10 SCE per metaphase, and no further decrease was observed described above are somewhat reminiscent of those reported in Slx1/ Mus81/ double knockout cells leaving the residual for Mus81/ mice (Dendouga et al., 2005). In both cases, the levels of SCE unaffected (Figure 3D). Therefore, SLX1 and mice are viable and fertile with no gross abnormalities, but MUS81-EME1 are epistatic with regard to the resolution of MEFs are hypersensitive to ICL-inducing agents. Cells from HJs that escape dissolution, suggesting that they function in both mice also show a higher than normal level of chromosome the same pathway. abnormalities after exposure to MMC. These observations, together with the physical proximity of SLX1 and MUS81-EME1 SLX4 Mutations that Prevent Interaction with SLX1 bound to SLX4, suggest that these two nucleases might act in and/or MUS81-EME1 the same pathway, perhaps cooperatively. In order to test this The data above suggest that SLX1 and MUS81-EME1 act possibility, we generated Slx1/ Mus81/ double knockout together in HJ resolution, and we next set out to test if this re- mice to enable epistasis analysis. These mice were born at Men- quires the tethering of the two nucleases to SLX4 where they delian frequencies (Table S3), showing no developmental, bind close together. The SLX1-interacting region of SLX4 has morphological, or hematological abnormalities. Moreover, the been localized to a small C-terminal fragment containing the fertility of Slx1/, Mus81/,orSlx1/ Mus81/ littermates, HtH motif (Figure 4A) (Fekairi et al., 2009; Svendsen et al., or testis size in males, is not significantly different from wild- 2009). A 200 amino acid fragment containing the HtH domain type mice (data not shown). To study HJ resolution, we isolated was subjected to saturated mutagenesis, followed by reverse MEFs from Slx1/ Mus81/ embryos, and as control we used yeast two-hybrid screening in order to find SLX4 mutations MEFs from single knockout embryos (Slx1/ Mus81+/ and that abolish interaction with SLX1 (data not shown). Two such Slx1+/ Mus81/). Double heterozygous (Slx1+/ Mus81+/) em- mutations were identified: a single point mutation in a highly bryos were used as negative control (Figure S4B). conserved Cys residue in the SLX4 HtH domain (C1536R), and As mentioned earlier, SLX1 had no effect on SCEs generated a small deletion mutant lacking six amino acids (Cys1536- in MEFs exposed to MMC (Figures 2E and 2F), and similar data Thr1541; Figure 4A). Full-length SLX4 bearing these mutations, were reported for Mus81/ cells (McPherson et al., 2004). How- or wild-type SLX4, was stably expressed in Slx4/ MEFs. As ever, it was possible this is because SLX1 acts redundantly with shown in Figure 4C, SLX1 is undetectable in extracts of Slx4/ MUS81-EME1. However, as shown in Figure 3A, the increase in MEFs, suggesting the stability of SLX1 requires that it binds to

(C) Scatterplot of SCE frequencies in MEFs depleted of BLM as described in (B). Fifty metaphases were analyzed for each condition, and significance was calculated using one-way ANOVA (***p < 0.0001) followed by Bonferroni’s Multiple Comparison Test. Each point represents the total number of SCEs in a single mitotic spread. The horizontal line in each data set represents mean SCE frequency. (D) Same as (C) except that Slx1/ MEFs infected with viruses expressing SLX1, SLX1 E79A, or empty virus were analyzed. Significance was calculated as above; ***p < 0.0001; n.s., nonsignificant. (E) Scatterplot of SCE frequencies in wild-type, Slx1/, and Slx4/ MEFs after exposure to increasing concentrations of MMC. Fifty metaphases were counted for each condition. Each point represents the total number of SCEs in a single mitotic spread. The horizontal line in each data set represents mean SCE frequency. (F) Same as (E) except that the data were plotted as fold increase over untreated cells. Data are represented as mean ± SEM. Experimental significance was calculated using an unpaired t test; **p < 0.01; ***p < 0.001.

Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors 225 Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

A SLX4. Stable expression of wild-type SLX4 restored normal *** levels of SLX1 in Slx4/ MEFs. However, the SLX4 mutants 30 C1536R or D1536-1541, expressed at levels similar to wild- type SLX4, did not restore SLX1 expression (Figure 4C). Further- 20 more, no SLX1 could be detected in SLX4 immunoprecipitates from cells expressing these mutant forms of SLX4, but MUS81 10 and ERCC1 were present at normal levels (Figure 4C). These

SCE per metaphase data indicate that the interaction of SLX4 with SLX1 is required 0 Slx1: +/- -/- +/- -/- Slx4-/- +/- -/- +/- -/- Slx4-/- for SLX1 stability, and therefore cells expressing SLX4 mutants Mus81: +/- +/- -/- -/- +/- +/- -/- -/- that cannot interact with SLX1 are essentially null for SLX1 - MMC + MMC (10 ng/ml) expression. Nonetheless, these mutants can be used to investi- gate if SLX1 contributes to the roles of SLX4 in DNA repair. B We also set out to generate SLX4 mutants that cannot interact 100 with MUS81-EME1. A previous study deleted the entire SAP Slx1+/- Mus81+/- domain of SLX4 to abolish interaction with MUS81-EME1, and Slx1-/- Mus81+/- the resulting SLX4 deletion fragment could only partly rescue the MMC sensitivity of FANCP cells (Kim et al., 2013). However, 10 +/- -/- Slx1 Mus81 even though in budding yeast Slx4 does not interact with Mus81- Eme1 % survival Slx1-/- Mus81-/- Mms4 (Schwartz et al., 2012), we found that deleting the Slx4-/- entire SAP domain of yeast Slx4 perturbs its function (data not 1 shown). These data indicate that the SAP domain of yeast Slx4 0 10203040 is functionally important independent of MUS81 interaction, MMC (ng/ml) and it is likely that the same is true of human SLX4. On this basis we sought to engineer more subtle changes in mouse SLX4 that C Slx1+/- Slx1-/- Slx1+/- Slx1-/- would abolish the interaction with MUS81-EME1. A fragment Mus81+/- Mus81+/- Mus81-/- Mus81-/- Slx4-/- containing the SAP domain of SLX4 was subjected to saturated shBLM: - + - + - + - + - + 191- mutagenesis, reverse yeast two-hybrid and alanine scanning BLM (data not shown). These experiments revealed several mutations 39- in SLX4 that abolished interaction with MUS81-EME1: Y1340A, GAPDH L1348A, and a combination of E1351A and L1352A (Figure 4B). Full-length SLX4 bearing these mutations, or wild-type SLX4, were stably expressed in Slx4/ MEFs. As shown in Figure 4D, *** D the SLX4 Y1340A, L1348A, and E1351A L1352A mutants were 30 unable to interact with MUS81-EME1 in coimmunoprecipitation experiments, but their ability to interact with SLX1 and ERCC1 20 was unaffected.

10 Binding of SLX1 and MUS81-EME1 to SLX4 Is Essential for HJ Resolution SCE per metaphase 0 We next examined whether the SLX4 mutants that are unable to Slx1: +/- -/- +/- -/- Slx4-/- +/- -/- +/- -/- Slx4-/- Mus81: +/- +/- -/- -/- +/- +/- -/- -/- interact with SLX1 or MUS81-EME1 are able to rescue the / + empty virus + shBLM defects observed in Slx4 cells, starting with analysis of ICL repair. As shown in Figure 5A, the SLX4 C1536R and D1536- Figure 3. SLX1 and MUS81 Act Epistatically in HJ Resolution 1541 mutants, that are incapable of interacting with SLX1, only / (A) Scatterplot of SCE frequencies in MEFs of the genotypes indicated after partly rescued the MMC hypersensitivity of Slx4 MEFs. The exposure to 10 ng/ml MMC. Fifty metaphases were counted for each condition residual sensitivity of these cells was similar to that of Slx1 / and significance was calculated using a one-way ANOVA (***p < 0.0001) fol- cells (Figure 1B), consistent with a failure of C1536R and lowed by Bonferroni’s Multiple Comparison Test. Each point represents the D1536-1541 SLX4 mutants to restore SLX1 expression in total number of SCEs in a single mitotic spread. The horizontal line in each data Slx4/ cells (Figure 4C). On the other hand, the SLX4 L1348A set represents the mean SCE frequency. (B) Clonogenic survival analysis of MEFs of the genotypes indicated exposed and E1351A L1352A mutants, both of which fail to interact with to MMC. For each genotype, cell viability of untreated cells was defined as MUS81-EME1, were able to fully rescue the MMC hypersensitiv- 100%. Data are represented as mean ± SEM, n = 3. (See also Figure S4.) ity of Slx4 / MEFs (Figure 5B). Referring to the SLX4 C1536R (C) Western blot analysis of MEFs, of the genotypes indicated, infected with mutant hereafter as SLX4SLX1 and the SLX4 E1351A L1352A retroviruses expressing a BLM-specific shRNA (+). Viruses prepared with as SLX4MUS81, we combined these mutations to generate empty vector served as control (). SLX4SLX1 MUS81 (Figure S5), and found that this compound (D) Cells from (C) were analyzed for SCE frequencies. Fifty metaphases were / counted for each condition and significance was calculated as in (A). Each mutant restored MMC hypersensitivity of Slx4 cells to the SLX1 point represents the total number of SCEs in a single mitotic spread. The same degree as the SLX4 mutant (Figure 5C). These data horizontal line in each data set represents mean SCE frequency. suggest that MMC hypersensitivity of cells lacking SLX4 is at

226 Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

A

B

C D

Figure 4. SLX4 Mutations that Prevent Interaction with SLX1 and MUS81 (A) Alignment of the C terminus of SLX4 from different species. The HtH domain is highlighted in red. The black line above the alignment refers to the six amino acid deletion in SLX4 that abolished interaction with SLX1, and the asterisk indicates Cys1536. Jalpred3-based secondary structure prediction is indicated below the alignment with barrels representing a helices and the arrow a b sheet. M.m., Mus musculus; H.s., Homo sapiens; X.t., Xenopus tropicalis; G.g., Gallus gallus; D.r., Danio rerio; D.m., Drosophila melanogaster. (B) Alignment of the SAP domain of SLX4 from different species (highlighted in green). Tyr1340, Leu1348, Glu1351, and Leu1352 are indicated with asterisks. S.c., ; S.p., Saccharomyces pombe. (C) Slx4/ MEFs were infected with retroviruses expressing SLX4 wild-type (SLX4), SLX4 C1536R, or SLX4 D1536-1541. Extracts were subjected to western blot analysis (upper panels) or immunoprecipitation with anti-SLX4 antibodies (lower panels). (D) Same as (C) except that Slx4/ MEFs were infected with retroviruses expressing SLX4 wild-type (SLX4), SLX4 bearing alanine mutations at Y1340, L1348, or E1351+L1352.

Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors 227 Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

A 100 B 100 Figure 5. HJ Resolution Requires Binding of SLX1 and MUS81-EME1 to SLX4 (A) Clonogenic survival analysis of Slx4/ MEFs stably expressing SLX4 wild-type (SLX4), SLX4 10 10 C1536R, or SLX4 D1536-1541, exposed to MMC. For each genotype, cell viability of untreated cells was defined as 100%. Wild-type MEFs and % survival % survival 1 1 Slx4/ MEFs infected with empty virus were used as controls. Data are represented as mean ± 0 10203040 0 10203040 SEM, n = 3. MMC (ng/ml) MMC (ng/ml) (B) Same as (A) except that SLX4 bearing alanine mutations at L1348 or E1351+L1352 were WT + empty vector WT + empty vector examined. (C) Clonogenic analysis as in (A) and (B) except -/- -/- Slx4 + empty vector Slx4 + empty vector that SLX4 mutations C1536R (SLX4SLX1) and Slx4-/- + SLX4 Slx4-/- + SLX4 E1351A+L1352A (SLX4MUS81) were combined to prevent interaction with both SLX1 and MUS81 Slx4-/- + SLX4 C1536R Slx4-/- + SLX4 L1348A (SLX4 SLX1 MUS81). (See also Figure S5.) Slx4-/- + SLX4 Δ1536-1541 Slx4-/- + SLX4 E1351A L1352A (D) Scatterplot of SCE frequencies in MEFs in- fected with retroviruses expressing a BLM-spe- cific shRNA or control virus (). The following MEFs were examined: Slx4/ MEFs infected with 100 C SLX4 wild-type or mutants of SLX4, i.e., SLX4SLX1, WT + empty vector SLX4MUS81, or SLX4SLX1 MUS81. Wild-type MEFs Slx4-/- + empty vector and Slx4/ MEFs infected with viruses prepared 10 Slx4-/- + SLX4 from empty vector were used as control. Fifty Slx4-/- + SLX4SLX1 metaphases were counted for each condition and significance was calculated using one-way -/- MUS81

% survival Slx4 + SLX4 1 ANOVA (***p < 0.0001) followed by Bonferroni’s Slx4-/- + SLX4SLX1 MUS81 Multiple Comparison Test. Each point represents the total number of SCE in a single mitotic spread. 0 10203040 The horizontal line in each data set represents MMC (ng/ml) mean SCE frequency.

D *** 30 n.s. *** repair in MEFs (Figure 5C). These data suggested strongly that the role of SLX1 and MUS81-EME1 in ICL repair is inde- 20 pendent of HJ resolution at least in cells from mice, and we set out to investigate this possibility.

10 An Hje-SLX1 Fusion Protein with a Preference for HJs SCE per metaphase The data above show that SLX1 and 0 MUS81-EME1 are both involved in HJ shBLM: - + +++ ++ resolution, judged by measuring the empty empty empty SLX4 SLX4SLX1 SLX4MUS81 SLX4SLX1 MUS81 increased SCE frequency caused by -/- WT Slx4 BLM depletion. However, SLX1 and MUS81-EME1 are both capable of least partly due to lack of SLX1 and that the interaction of SLX4 cleaving a range of branched DNA structures besides HJs with MUS81-EME1 is not required for ICL repair in MEFs. in vitro. This consideration, together with the observation that We went on to test if the SLX4SLX1 and SLX4MUS81 mutants are BLM can act on a similar range of branched substrates (Moha- defective in HJ resolution. As shown in Figure 5D, the elevation in ghegh et al., 2001; van Brabant et al., 2000), raised the possibility SCE frequency triggered by BLM depletion is diminished in that the SCEs in BLM-depleted cells arise from nucleolytic Slx4/ MEFs compared with wild-type MEFs. This defect is processing of structures other than four-way junctions. In an rescued by expression of wild-type SLX4, but not by the attempt to investigate this possibility, we set out to see if the SLX4SLX1, SLX4MUS81 mutants or the SLX4SLX1 MUS81 compound cleavage of HJs by SLX1 and MUS81-EME1 is physiologically mutant. Therefore, it is likely that the binding of both SLX1 and relevant. With this in mind, we mutated the SLX1 nuclease MUS81-EME1 to SLX4 is essential for HJ resolution. In contrast, domain (R38A E79A) to inactivate it and fused this SLX1 deriva- the binding of SLX1, but not MUS81-EME1, is important for ICL tive, which should maintain the ability to bind SLX4, to an

228 Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

archaeal resolving enzyme Hje. This enzyme was reported to back to wild-type levels (Figure 6E). Taken together, these have a marked preference for HJs compared with other resol- data strongly suggest that the formation of SCE in BLM- vases (Kvaratskhelia and White, 2000). We next asked if the depleted cells, which was shown to require binding of SLX1 resulting fusion protein, referred to as Hje-SLX1 (Figure 6A), and MUS81-EME1 to SLX4, results from the resolution of HJs. can rescue defects associated with loss of SLX1. We used an Finally, we tested the ability of the Hje-SLX1 fusion protein to analogous fusion construct containing a catalytically inactive reverse the ICL repair defects in cells lacking SLX1. Intriguingly, D39A Hje mutant as a negative control. We confirmed that the Hje-SLX1 fusion protein was incapable of rescuing the MMC FLAG-tagged versions of the Hje-SLX1 fusion protein, and the hypersensitivity of Slx1/ MEFs in contrast to wild-type SLX1 catalytically inactive derivative, bind to SLX4 in coimmunopreci- (Figure S6D). These data strengthen the notion that the role of pitation experiments in HEK293 cells (Figure S6B). SLX1 in ICL repair in MEFs involves cleavage of DNA structures Next, anti-FLAG immunoprecipitates were incubated with other than HJs. radiolabeled DNA substrates. We used a four-way junction with a 12 bp homologous core (HJ), a replication fork (RF) analog DISCUSSION (a three-way junction where one strand is discontinuous at the junction), and a 50 flap. Reaction products were separated by In this study, we demonstrate that SLX1 is involved in cellular re- under denaturing conditions. As shown in sponses to DNA damage in mammals. Consistent with its re- Figure 6B, both SLX1 and the Hje-SLX1 fusion exhibited strong ported ability to process HJs in vitro (Fekairi et al., 2009; cleavage activity on the four-way junction substrate. Consistent Mun˜ oz et al., 2009; Svendsen et al., 2009), we found that with previous reports, SLX1 cleaved the four-way junction at SLX1, like MUS81-EME1, plays a major role in the nucleolytic multiple positions (Fekairi et al., 2009; Mun˜ oz et al., 2009; Svend- resolution of HJs in vivo. SLX1 and MUS81-EME1 are both sen et al., 2009); the Hje-SLX1 fusion protein also cleaved at required for resolving the majority of HJs that escape dissolution several points, but these were different from SLX1 (Figure 6B; by BLM, and they act epistatically in this regard. We also found Figure S6A). Nonetheless, nondenaturing gel electrophoresis that the tethering of SLX1 and MUS81-EME1 to SLX4, where showed that Hje-SLX1 introduced symmetric cleavages within they bind in close proximity, is essential for the ability of these the homologous core of the HJ, resulting in the production of nucleases to promote HJ resolution. The simplest interpretation linear duplex DNA as required for the productive resolution of of these findings is that SLX1 and MUS81-EME1 act coopera- the junction (Figure 6C). The Hje-SLX1 derivative, which is inac- tively in the processing of HJs, as previously postulated by tivated by the D39A mutation, showed no detectable activity to- others (Svendsen et al., 2009). In this model SLX1 acts on intact ward the HJ substrate (Figure 6B). Although all of the FLAG HJs to create nicked junctions, which are then cleaved by immunoprecipitates exhibited some nonspecific activity toward MUS81-EME1 (Boddy et al., 2001; Ciccia et al., 2003; Doe the RF substrate (Figure 6B, asterisk), a weak SLX1-specific et al., 2002; Svendsen et al., 2009; Whitby et al., 2003). Close cleavage product was observed in FLAG-SLX1 immunoprecipi- juxtaposition of SLX1 and MUS81-EME1 on SLX4 might position tates that was not present in SLX1 E79A precipitates (Figure 6B, the respective active sites to favor coordinated first-strand arrowhead). This product was not observed in Hje-SLX1 precip- cleavage by SLX1 and second-strand cleavage by MUS81- itates, indicating that this fusion protein cannot cleave replication EME1. This idea might help to solve the longstanding dilemma forks. The 50 flap substrate was cleaved by FLAG-SLX1 one of how MUS81-EME1 participates in HJ resolution in vivo nucleotide before the junction, and by contrast, the Hje-SLX1 when the recombinant enzyme only weakly cleaves intact HJs fusion exhibited only very weak activity three nucleotides from in vitro compared with nicked HJs (Boddy et al., 2001; Ciccia junction (Figure 6B; Figure S6A). These experiments demon- et al., 2003; Doe et al., 2002; Whitby et al., 2003). strated that Hje-SLX1 binds to SLX4 and acts as a potent HJ The model whereby SLX1 and MUS81-EME1 act together in resolving enzyme with restricted substrate specificity compared HJ resolution is based on measuring SCE frequency upon BLM with SLX1. depletion. But do the SCEs observed in BLM-depleted cells result from the cleavage of HJs? It is possible that the SCEs HJs Are Physiological Substrates of SLX1 resulting from depletion of BLM reflect instead the cleavage and MUS81-EME1 of branched DNA structures other than HJs. This is a particular We next examined whether the Hje-SLX1 fusion could restore concern because BLM, like MUS81-EME1 and SLX1, is active the cellular functions perturbed in cells lacking SLX1. Untagged toward a range of branched DNA structures in vitro (Moha- versions of Hje-SLX1 and the inactive D39A derivative were sta- ghegh et al., 2001; van Brabant et al., 2000). Furthermore, the bly expressed in Slx1/ MEFs (Figure S6C, left panels). Cells cleavage of the precursors to HJs such as extended D-loop were then depleted of BLM. Strikingly, the Hje-SLX1 fusion pro- structures can produce crossovers that would manifest as tein, but not the D39A version, restored SCE frequency in BLM- SCEs (Osman et al., 2003). Thus, it is possible that the SCEs depleted Slx1/ MEFs similar to wild-type SLX1 (Figure 6D). in BLM-depleted cells occur as a result of the cleavage Expression of Hje-SLX1 in cells that were not depleted of of recombination structures other than HJs. However, the BLM had no apparent effect on SCE levels (Figure 6D). The demonstration that the Hje-SLX1 fusion protein with increased Hje-SLX1 fusion and the inactive D39A derivative were also specificity for HJs can reverse the SCE defect in Slx1/ and stably expressed in Slx1/ Mus81/ MEFs (Figure S6C, right Slx1/ Mus81/ cells strongly suggests that SCE forma- panels). This experiment revealed that Hje-SLX1 but not the tion in BLM-depleted cells reflects HJ cleavage by SLX1 and D39A version restored SCE frequency after BLM depletion MUS81-EME1.

Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors 229 Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

A SLX1 R38A/E79A * * Hje-SLX1 fusion: FLAG Hje nuclease

B Holliday junction RF 5’flap

* * *

C ladder SLX1 SLX1 E79A Hje-SLX1 Hje(D39A)-SLX1 ladder SLX1 SLX1 E79A Hje-SLX1 Hje(D39A)-SLX1 ladder SLX1 SLX1 E79A Hje-SLX1 Hje(D39A)-SLX1 Hje-SLX1 Hje(D39A)-SLX1 marker SLX1 SLX1 E79A

A T C HJ- G T G A T A A DS- G C A

homologous core C C * G T

*** D n.s. E *** n.s. n.s. 30 30 ***

20 20

10 10 SCE per metaphase SCE per metaphase 0 0 shBLM: - + - ++ shBLM: - + +++ empty empty Hje- Hje- Hje(D39A)- empty empty empty Hje- Hje(D39A)- SLX1 SLX1 SLX1 SLX1 SLX1 WT Slx1-/- Slx1+/- Mus81+/- Slx1-/- Mus81-/-

Figure 6. HJs Are Physiological Substrates of SLX1 and MUS81 (A) Schematic diagram of a Hje-SLX1 fusion protein. The SLX1 nuclease domain was mutated (R38A E79A) to inactivate it and fused to an archaeal resolving enzyme Hje known to cleave Holliday junctions preferentially. (B) HEK293 cells stably expressing GFP-tagged SLX4 were transfected with SLX1 wild-type or SLX1 E79A, Hje-SLX1, or Hje (D39A)-SLX1. Anti-FLAG immunoprecipitates were incubated with synthetic DNA structures: mobile Holliday junction (b strand-labeled Jbm5), replication fork (RF)-like structure (b strand- labeled), or 50 flap (a3 strand-labeled). Reaction products were subjected to denaturing PAGE. Sequencing ladders were obtained by subjecting the corre- sponding substrates to Maxam and Gilbert‘s purine-specific reaction. Lower exposure is shown for the RF sequencing ladder. (See also Figure S6.) (C) Same as (B) except the products from the HJ reaction were subjected to nondenaturing PAGE. HJ and duplex DNA (DS) of the same size as the predicted products of symmetrical cleavage of the HJ were run in parallel as markers. (D) Scatterplot of SCE frequencies in MEFs infected with retroviruses expressing a BLM-specific shRNA (+) or with control virus (). The following MEFs were examined: Slx1/ MEFs infected with Hje-SLX1, Hje (D39A)-SLX1, or empty virus. Wild-type MEFs infected with empty virus were used as control. Fifty metaphases were counted for each condition and significance was calculated using one-way ANOVA (***p < 0.0001; n.s., nonsignificant) followed by Bonferroni’s Multiple Comparison Test. Each point represents the total number of SCEs in a single mitotic spread. The horizontal line in each data set represents mean SCE frequency. (E) Same as (D) except that Slx1/ Mus81/ MEFs expressing Hje-SLX1, Hje (D39A)-SLX1, or empty virus were examined.

230 Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

In this study we demonstrate that SLX1 is involved in ICL GEN1. It will be interesting to make the relevant combinations repair in mammalian cells, and similarly to HJ resolution, of gene knockouts in mice to investigate this. Given the impor- SLX1 appears to act epistatically with MUS81-EME1 in ICL tance of meiotic HJ resolution in heredity, and in preventing repair. However, several observations suggest that the epistatic defects in chromosome segregation, it will be vitally important interaction between SLX1 and MUS81-EME1 in ICL repair does to decipher the precise mechanisms underlying the resolution not involve the cleavage of HJs, at least in mice. First, the SCEs of meiotic HJs. formed in MMC-treated MEFs do not require SLX1 or MUS81- EME1, unlike the SCEs observed in BLM-depleted cells. Sec- EXPERIMENTAL PROCEDURES ond, whereas HJ resolution in BLM-depleted MEFs requires binding of both SLX1 and MUS81-EME1 to SLX4, cellular resis- Depletion of BLM from MEFs Using shRNA tance to MMC requires binding of SLX1 but not MUS81-EME1. The protocol was adapted from a previous report (Sfeir et al., 2009). Briefly, viruses expressing a BLM-specific shRNA were produced by cotransfecting Third, the Hje-SLX1 fusion protein, which is much more specific / 293T cells with pSUPER.retro.puro (Open Biosystems) expressing the BLM for HJs than SLX1, can rescue the SCE defect in Slx1 and target sequence 50-gga cct gct gga aga ttt a-30 and with pCMV-VSV-G and / / Slx1 Mus81 MEFs depleted of BLM, but it cannot rescue pCMV-Gag-Pol. MEFs were infected four times in 6 hr intervals with virus-con- the MMC hypersensitivity of these cells. These observations taining supernatant supplemented with 8 mg/ml polybrene. Twenty-four hours suggest that the role of SLX1 and MUS81-EME1 in the repair after the last infection, MEFs were selected with 3 mg/ml puromycin. Knock- of ICLs involves the processing of structures other than HJs. down efficiency was confirmed by western blot. How might SLX1 and MUS81-EME1 cooperate in ICL repair? MUS81-EME1 is known to act at the ‘‘unhooking’’ step of ICL SCE Assay Cells were treated for 24 hr with 10 mM BrdU before replacing the medium with repair, a process that involves incisions on either side of the fresh BrdU and the indicated dose of MMC for an additional 24 hr. When SCE ICL that produce DSBs (Hanada et al., 2006, 2007). However, analysis was performed after BLM knockdown, cells were selected for the up- DSB induction in cells exposed to ICL-inducing agents does take of virus in the presence of 3 mg/ml puromycin 24 hr after the last infection. not require SLX1 (data not shown). Therefore, the epistatic MEFs were treated with two cycles of BrdU as described above (Figure S4A). interaction of SLX1 and MUS81-EME1 in ICL repair is indepen- At 6 hr before the end of the second BrdU cycle, nocodazole was added at a m dent of DSB induction. One possible explanation is that SLX1- final concentration of 0.15 g/ml. Cells were harvested and prepared for meta- phase spreads. Spreads were stained with Hoechst 33258, followed by UV SLX4 processes recombination intermediates during the repair irradiation and staining with Giemsa according to standard protocols. Fifty of the DSBs generated by the pool of MUS81-EME1 that is not spreads were analyzed for each condition on a DeltaVision wide-field decon- bound to SLX4. volution microscope (Olympus). It is clear that the dominant mode of HJ removal in mitotic Full details of all other experimental procedures are given in Supplemental cells—at least those arising spontaneously—involves BLM, Information. and HJ resolution by nucleases acts to cleave junctions that escape dissolution. The BLM pathway has been supposed to SUPPLEMENTAL INFORMATION minimize the risk of loss of heterozygosity, since this pathway generates noncrossover outcomes exclusively (LaRocque Supplemental Information includes six figures, four tables, and Supplemental Experimental Procedures and can be found with this article online at http://dx. et al., 2011). However, in meiotic cells crossover formation is doi.org/10.1016/j.molcel.2013.08.036. necessary and should therefore involve nucleolytic HJ resolu- tion. Since SLX1 and MUS81-EME1 resolve the majority of ACKNOWLEDGMENTS HJs that escape dissolution in mitotic mammalian cells, one might predict that they might dominate in . However, We are grateful to James Hastie, Hilary MacLauchlan, and the Antibody Pro- although Mus81/ mice show meiotic defects, they are fertile, duction Team at Division of Signal Transduction Therapy, University of Dundee suggesting that other nucleases can mediate HJ processing in and to DNA Sequencing Service and Synthesis Service at CLS, University of Dundee. We thank CIRRU staff, especially Lorraine Malone the absence of MUS81-EME1 (Dendouga et al., 2005; Holloway and Carol Clacher, for help with animal husbandry and ESC derivation, and we et al., 2008). This is unlikely to be SLX1, since the fertility and are thankful to Gail Fraser and Elaine Forsyth for help with genotyping. We / / / / gonad size of Slx1 , Mus81 ,orSlx1 Mus81 mice thank Victoria McGuire for help with generating knockout mice and Catherine generated in this study were not significantly different from Lamm for histological analyses. We are grateful to Clare McGowan for the kind wild-type mice (data not shown). Since Slx4/ mice exhibit gift of Mus81 / mice and to Simon Boulton for sending the mice. We are subfertility and hypogonadism, but Slx1/ Mus81/ do not, grateful to members of the Rouse lab for critical reading of this manuscript. the third SLX4-associated nuclease, XPF-ERCC1, might be We thank Anton Gartner for communicating data prior to publication, useful discussions, and critical reading of the manuscript. We are also grateful to involved in meiotic HJ resolution. In this light, recent studies Steve West for communicating data prior to publication. This study was also in nematodes found that XPF/BLM and SLX1/MUS81-EME1 supported by the Medical Research Council (MRC), the pharmaceutical com- act on parallel pathways in worm meiosis (Agostinho et al., panies supporting the Division of Signal Transduction Therapy Unit 2013; O’Neil et al., 2013; Saito et al., 2013). Moreover, SLX4 (AstraZeneca, Boehringer-Ingelheim, GlaxoSmithKline, Merck KgaA, Janssen and XPF-ERCC1 are important for crossover generation in Pharmaceutica, and Pfizer) associated with the MRC PPU (D.C., N.N., R.T., Drosophila (Yildiz et al., 2002). It will be interesting to test if T.J.M., J.S.C.A., J.R.), and Cancer Research UK (A.-C.D. and D.M.J.L.). XPF-ERCC1 is involved in HJ resolution in mammals. However, / Received: April 26, 2013 since Slx4 mice can generate offspring, it must be that Revised: July 19, 2013 nucleases that do not associate with SLX4 can resolve HJs in Accepted: August 14, 2013 the absence of SLX4 complex components. This might be Published: September 26, 2013

Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors 231 Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

REFERENCES Holloway, J.K., Mohan, S., Balmus, G., Sun, X., Modzelewski, A., Borst, P.L., Freire, R., Weiss, R.S., and Cohen, P.E. (2011). Mammalian BTBD12 (SLX4) Agostinho, A., Meier, B., Sonneville, R., Jagut, M., Woglar, A., Blow, J., protects against genomic instability during mammalian spermatogenesis. Jantsch, V., and Gartner, A. (2013). Combinatorial regulation of meiotic PLoS Genet. 7, e1002094. holliday junction resolution in C. elegans by HIM-6 (BLM) helicase, Ip, S.C.Y., Rass, U., Blanco, M.G., Flynn, H.R., Skehel, J.M., and West, S.C. SLX-4, and the SLX-1, MUS-81 and XPF-1 nucleases. PLoS Genet. 9, (2008). Identification of Holliday junction resolvases from humans and yeast. e1003591. Nature 456, 357–361. Andersen, S.L., Bergstralh, D.T., Kohl, K.P., LaRocque, J.R., Moore, C.B., and Kim, Y., Lach, F.P., Desetty, R., Hanenberg, H., Auerbach, A.D., and Sekelsky, J. (2009). Drosophila MUS312 and the vertebrate ortholog BTBD12 Smogorzewska, A. (2011). Mutations of the SLX4 gene in Fanconi anemia. interact with DNA structure-specific endonucleases in DNA repair and recom- Nat. Genet. 43, 142–146. bination. Mol. Cell 35, 128–135. Kim, Y., Spitz, G.S., Veturi, U., Lach, F.P., Auerbach, A.D., and Auerbach, A.D. (2009). Fanconi anemia and its diagnosis. Mutat. Res. 668, Smogorzewska, A. (2013). Regulation of multiple DNA repair pathways by 4–10. the Fanconi anemia protein SLX4. Blood 121, 54–63. Bailly, A.P., Freeman, A., Hall, J., De´ clais, A.C., Alpi, A., Lilley, D.M., Ahmed, S., Kvaratskhelia, M., and White, M.F. (2000). An archaeal Holliday junction and Gartner, A. (2010). The Caenorhabditis elegans homolog of Gen1/Yen1 resolving enzyme from Sulfolobus solfataricus exhibits unique properties. resolvases links DNA damage signaling to DNA double-strand break repair. J. Mol. Biol. 295, 193–202. PLoS Genet. 6, e1001025. LaRocque, J.R., Stark, J.M., Oh, J., Bojilova, E., Yusa, K., Horie, K., Takeda, J., Bakker, S.T., de Winter, J.P., and te Riele, H. (2013). Learning from a paradox: and Jasin, M. (2011). Interhomolog recombination and loss of heterozygosity in recent insights into Fanconi anaemia through studying mouse models. Dis. wild-type and Bloom syndrome helicase (BLM)-deficient mammalian cells. Model. Mech. 6, 40–47. Proc. Natl. Acad. Sci. USA 108, 11971–11976. Boddy, M.N., Gaillard, P.H., McDonald, W.H., Shanahan, P., Yates, J.R., 3rd, McPherson, J.P., Lemmers, B., Chahwan, R., Pamidi, A., Migon, E., Matysiak- and Russell, P. (2001). Mus81-Eme1 are essential components of a Holliday Zablocki, E., Moynahan, M.E., Essers, J., Hanada, K., Poonepalli, A., et al. junction resolvase. Cell 107, 537–548. (2004). Involvement of mammalian Mus81 in genome integrity and tumor Chaganti, R.S., Schonberg, S., and German, J. (1974). A manyfold increase suppression. Science 304, 1822–1826. in sister chromatid exchanges in Bloom’s syndrome lymphocytes. Proc. Mohaghegh, P., Karow, J.K., Brosh, R.M., Jr., Bohr, V.A., and Hickson, I.D. Natl. Acad. Sci. USA 71, 4508–4512. (2001). The Bloom’s and Werner’s syndrome proteins are DNA structure-spe- Ciccia, A., Constantinou, A., and West, S.C. (2003). Identification and charac- cific helicases. Nucleic Acids Res. 29, 2843–2849. terization of the human -eme1 endonuclease. J. Biol. Chem. 278, Mun˜ oz, I.M., Hain, K., De´ clais, A.C., Gardiner, M., Toh, G.W., Sanchez-Pulido, 25172–25178. L., Heuckmann, J.M., Toth, R., Macartney, T., Eppink, B., et al. (2009). Ciccia, A., McDonald, N., and West, S.C. (2008). Structural and functional Coordination of structure-specific nucleases by human SLX4/BTBD12 is relationships of the XPF/MUS81 family of proteins. Annu. Rev. Biochem. 77, required for DNA repair. Mol. Cell 35, 116–127. 259–287. O’Neil, N.J., Martin, J.S., Youds, J.L., Ward, J.D., Petalcorin, M.I., Rose, A.M., Crossan, G.P., van der Weyden, L., Rosado, I.V., Langevin, F., Gaillard, P.H., and Boulton, S.J. (2013). Joint molecule resolution requires the redundant McIntyre, R.E., Gallagher, F., Kettunen, M.I., Lewis, D.Y., Brindle, K., et al.; activities of MUS-81 and XPF-1 during Caenorhabditis elegans meiosis. Sanger Mouse Genetics Project. (2011). Disruption of mouse Slx4, a regulator PLoS Genet. 9, e1003582. of structure-specific nucleases, phenocopies Fanconi anemia. Nat. Genet. 43, 147–152. Osman, F., Dixon, J., Doe, C.L., and Whitby, M.C. (2003). Generating cross- overs by resolution of nicked Holliday junctions: a role for Mus81-Eme1 in Dendouga, N., Gao, H., Moechars, D., Janicot, M., Vialard, J., and McGowan, meiosis. Mol. Cell 12, 761–774. C.H. (2005). Disruption of murine Mus81 increases genomic instability and DNA damage sensitivity but does not promote tumorigenesis. Mol. Cell. Rass, U., Compton, S.A., Matos, J., Singleton, M.R., Ip, S.C., Blanco, M.G., Biol. 25, 7569–7579. Griffith, J.D., and West, S.C. (2010). Mechanism of Holliday junction resolution by the human GEN1 protein. Genes Dev. 24, 1559–1569. Doe, C.L., Ahn, J.S., Dixon, J., and Whitby, M.C. (2002). Mus81-Eme1 and Rqh1 involvement in processing stalled and collapsed replication forks. Saito, T.T., Youds, J.L., Boulton, S.J., and Colaia´ covo, M.P. (2009). J. Biol. Chem. 277, 32753–32759. Caenorhabditis elegans HIM-18/SLX-4 interacts with SLX-1 and XPF-1 and maintains genomic integrity in the germline by processing recombination inter- Fekairi, S., Scaglione, S., Chahwan, C., Taylor, E.R., Tissier, A., Coulon, S., mediates. PLoS Genet. 5, e1000735. Dong, M.Q., Ruse, C., Yates, J.R., 3rd, Russell, P., et al. (2009). Human SLX4 is a Holliday junction resolvase subunit that binds multiple DNA repair/ Saito, T.T., Lui, D.Y., Kim, H.M., Meyer, K., and Colaia´ covo, M.P. (2013). recombination endonucleases. Cell 138, 78–89. Interplay between structure-specific endonucleases for crossover control during Caenorhabditis elegans meiosis. PLoS Genet. 9, e1003586. Gaillard, P.H.L., Noguchi, E., Shanahan, P., and Russell, P. (2003). The endog- enous Mus81-Eme1 complex resolves Holliday junctions by a and coun- Schwartz, E.K., and Heyer, W.D. (2011). Processing of joint molecule interme- ternick mechanism. Mol. Cell 12, 747–759. diates by structure-selective endonucleases during homologous recombina- tion in . Chromosoma 120, 109–127. Hanada, K., Budzowska, M., Modesti, M., Maas, A., Wyman, C., Essers, J., and Kanaar, R. (2006). The structure-specific endonuclease Mus81-Eme1 pro- Schwartz, E.K., Wright, W.D., Ehmsen, K.T., Evans, J.E., Stahlberg, H., and motes conversion of interstrand DNA crosslinks into double-strands breaks. Heyer, W.D. (2012). Mus81-Mms4 functions as a single heterodimer to cleave EMBO J. 25, 4921–4932. nicked intermediates in recombinational DNA repair. Mol. Cell. Biol. 32, 3065– Hanada, K., Budzowska, M., Davies, S.L., van Drunen, E., Onizawa, H., 3080. Beverloo, H.B., Maas, A., Essers, J., Hickson, I.D., and Kanaar, R. (2007). Sfeir, A., Kosiyatrakul, S.T., Hockemeyer, D., MacRae, S.L., Karlseder, J., The structure-specific endonuclease Mus81 contributes to replication restart Schildkraut, C.L., and de Lange, T. (2009). Mammalian telomeres resemble by generating double-strand DNA breaks. Nat. Struct. Mol. Biol. 14, 1096– fragile sites and require TRF1 for efficient replication. Cell 138, 90–103. 1104. Stoepker, C., Hain, K., Schuster, B., Hilhorst-Hofstee, Y., Rooimans, M.A., Holloway, J.K., Booth, J., Edelmann, W., McGowan, C.H., and Cohen, P.E. Steltenpool, J., Oostra, A.B., Eirich, K., Korthof, E.T., Nieuwint, A.W., et al. (2008). MUS81 generates a subset of MLH1-MLH3-independent crossovers (2011). SLX4, a coordinator of structure-specific endonucleases, is mutated in mammalian meiosis. PLoS Genet. 4, e1000186. in a new Fanconi anemia subtype. Nat. Genet. 43, 138–141.

232 Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors Molecular Cell SLX1 and MUS81 in DNA Repair and HJ Resolution

Svendsen, J.M., Smogorzewska, A., Sowa, M.E., O’Connell, B.C., Gygi, S.P., West, S.C. (2009). The search for a human Holliday junction resolvase. Elledge, S.J., and Harper, J.W. (2009). Mammalian BTBD12/SLX4 assembles Biochem. Soc. Trans. 37, 519–526. a Holliday junction resolvase and is required for DNA repair. Cell 138, 63–77. Whitby, M.C., Osman, F., and Dixon, J. (2003). Cleavage of model replication forks by fission yeast Mus81-Eme1 and budding yeast Mus81-Mms4. J. Biol. Taylor, E.R., and McGowan, C.H. (2008). Cleavage mechanism of human Chem. 278, 6928–6935. Mus81-Eme1 acting on Holliday-junction structures. Proc. Natl. Acad. Sci. USA 105, 3757–3762. Wu, L., and Hickson, I.D. (2003). The Bloom’s syndrome helicase suppresses crossing over during homologous recombination. Nature 426, 870–874. van Brabant, A.J., Ye, T., Sanz, M., German, J.L., III, Ellis, N.A., and Holloman, Wu, L., and Hickson, I.D. (2006). DNA helicases required for homologous W.K. (2000). Binding and melting of D-loops by the Bloom syndrome helicase. recombination and repair of damaged replication forks. Annu. Rev. Genet. Biochemistry 39, 14617–14625. 40, 279–306. Wechsler, T., Newman, S., and West, S.C. (2011). Aberrant chromosome Yildiz, O., Majumder, S., Kramer, B., and Sekelsky, J.J. (2002). Drosophila morphology in human cells defective for Holliday junction resolution. Nature MUS312 interacts with the nucleotide excision repair endonuclease MEI-9 to 471, 642–646. generate meiotic crossovers. Mol. Cell 10, 1503–1509.

Molecular Cell 52, 221–233, October 24, 2013 ª2013 The Authors 233