a winged helix domain in human mus81 binds dna and modulates

A winged helix domain in human MUS81 binds DNAand modulates the endonuclease activity of MUS81complexesAndrew J Fadden1 Stephanie Schalbetter2 Maureen Bowles1 Richard Harris3

John Lally1 Antony M Carr2 and Neil Q McDonald13

1Structural Biology Laboratory Cancer Research UK London Research Institute 44 Lincolnrsquos Inn FieldsLondon WC2A 3LY UK 2Genome Damage and Stability Centre University of Sussex Brighton BN1 9RQ UKand 3Institute of Structural and Molecular Biology University College London and Birkbeck College MaletStreet London WC1E 7HX UK

Received August 2 2012 Revised July 5 2013 Accepted July 31 2013

ABSTRACT

The MUS81-EME1 endonuclease maintainsmetazoan genomic integrity by cleaving branchedDNA structures that arise during the resolution ofrecombination intermediates In humans MUS81also forms a poorly characterized complex withEME2 Here we identify and determine the structureof a winged helix (WH) domain from human MUS81which binds DNA WH domain mutations greatlyreduce binding of the isolated domain to DNA andimpact on incision activity of MUS81-EME1EME2complexes Deletion of the WH domain reducesthe endonuclease activity of both MUS81-EME1and MUS81-EME2 complexes and incisions madeby MUS81-EME2 are made closer to the junctionon substrates containing a downstream duplexsuch as fork structures and nicked Hollidayjunctions WH domain mutation or deletion inSchizosaccharomyces pombe phenocopies theDNA-damage sensitivity of strains deleted formus81 Our results indicate an important role forthe WH domain in both yeast and human MUS81complexes

INTRODUCTION

The XPF family of eukaryotic DNA junction endonucle-ases plays crucial roles in maintaining genomic stability byfunctioning in multiple DNA processing pathways (1) Inhumans at least four heterodimeric complexes containing

XPF family members have been identified These includeXPF-ERCC1 MUS81-EME1 and the largely uncha-racterized MUS81-EME2 complex as well as FANCM-FAAP24 although this complex has no demonstrablenuclease activity to date (12) The catalytically activesubunits of these complexes have a central nucleasedomain distantly related to prokaryotic type II restrictionenzymes and two C-terminal helix-hairpin-helix motifsthat co-operate as a functional (HhH)2 domain to bindDNA (34) A similar domain structure is found for thenon-catalytic partner of the endonuclease complexeswhich are necessary for DNA junction recognition andnucleolytic activity (4ndash6)The MUS81 endonuclease when associated with the

non-catalytic partner EME1 is able to efficiently cleavea variety of three- and four-way junctions containing aduplex downstream from a nick in vitro (7) These struc-tures include forks 30 flaps nicked Holliday junctions andD-loops (48) Such MUS81-EME1 substrates can formduring mitosis and fission yeast meiosis and duringprocessing of damaged replication forksStructural characterization of MUS81-EME1 has

revealed a distinctive orientation of its nuclease and(HhH)2 domains and showed an essential contributionfrom both (HhH)2 domains towards DNA junctionbinding (4) These studies lacked the amino-terminal(N-terminal) extensions of both subunits WithinMUS81 its N-terminal extension contains a single HhHmotif which is believed to be involved in proteinndashproteininteractions rather than DNA binding and has recentlyshown to be capable of binding Flap endonuclease 1(FEN1) (9) This study also showed that N-terminal

To whom correspondence should be addressed Tel +44 207 269 3259 Fax +44 207 269 3258 Email neilmcdonaldcancerorguk

The authors wish it to be known that in their opinion the first three authors should be regarded as Joint First Authors

Present addressRichard Harris Biochemistry Department Albert Einstein College of Medicine Yeshiva University 1300 Morris Park Avenue BronxNY 10461 USA

Published online 27 August 2013 Nucleic Acids Research 2013 Vol 41 No 21 9741ndash9752doi101093nargkt760

The Author(s) 2013 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby30) whichpermits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited

Downloaded from httpsacademicoupcomnararticle-abstract412197411276949by gueston 30 March 2018

fragments of MUS81 can bind DNA and stimulate theflap endonuclease activity of FEN1 The N-terminalregion also interacts with SLX4 a protein that isthought to act as a scaffold for the structure-specificendonucleases MUS81-EME1 XPF-ERCC1 and SLX1for recruitment to the repair of interstrand crosslinksand restart of damaged replication forks (10ndash12) TheN-terminus was also proposed to contain a Bloomsyndrome (BLM) protein-interacting domain (13)Here we identify for the first time a winged helix

domain (WH domain) within the N-terminal region ofhuman MUS81 that binds DNA increases the activityof MUS81-EME1EME2 complexes and influences theincision position of MUS81-EME2 but not MUS81-EME1 complexes on synthetic forks 30 flaps and nickedHolliday junctions Additionally in MUS81-EME2complexes it stimulates the cleavage of splayed armsMutations in the WH domain in Schizosaccharomycespombe render a similar sensitivity profile to DNAdamaging agents as a mus81 deleted strain implying thatthis domain has a critical role in DNA repair that has beenretained in human MUS81

MATERIALS AND METHODS

Protein purification

MUS81-EME1EME2 complexes (NCBI accessionnumbers MUS81NP_0794042 EME1 NP_001159603EME2 NM_001257370) (14) were produced inEscherichia coli BL21 (DE3) Rosetta pLysS (Stratagene)using the dicistronic expression plasmid derived frompGEX-KG MUS81-EME1EME2 wild-type and mutantcomplexes were purified as follows Cultures were grownin Luria-Bertani at 37C and induced with 25 mMIsopropyl-b-D-thiogalactoside (IPTG) at 18C overnightBacteria were harvested by centrifugation at 4000 g for15min at 4C and the pellets resuspended in 50mMsodium phosphate (pH 8) 300mM NaCl 2mMb-mercaptoethanol 10 glycerol 02 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate(CHAPS) (buffer A) supplemented with 10mMbenzamidine 1mM phenylmethylsulfonyl fluoride(PMSF) and protease inhibitors (Roche) Bacteria werelysed by sonication Cell debris was cleared by centrifuga-tion at 29 220 g for 60min at 4C and the clarified lysatewas mixed with glutathione-Sepharose 4B (GE HealthcareLifesciences) for 90min at 4C Unbound proteins werecollected and the affinity resin was washed extensivelywith buffer A Human MUS81 (hMUS81) complexeswere eluted with elution buffer [buffer A containing20mM reduced glutathione (pH 8)] The eluted proteinswere analysed by SDSndashPAGE hMUS81 complexes wereconcentrated by ultrafiltration and loaded onto a 1ml ofHiTrap Heparin-Sepharose column pre-equilibrated withbuffer A The hMUS81 complexes were eluted using aNaCl gradient (up to 1M NaCl in buffer A) over 10column volumes on an AKTA fast performance liquidchromatography system (GE Healthcare Lifesciences)Peak fractions from the Heparin-Sepharose columnwere pooled and concentrated by ultrafiltration The

concentrated sample was loaded onto a Superose 12 HR10300 column (GE Healthcare Lifesciences) pre-equilibrated with buffer A After analysis bySDSndashPAGE aliquots of complex-containing fractionswere flash-frozen in liquid nitrogen for assaying

hMUS81 WH domain purificationCultures were grown at 37C and induced by addition ofisopropyl-b-D-thiogalactoside (final concentration250 mM) and further incubation at 37C for 120minBacteria were harvested by centrifugation at 4000 gfor 15min at 4C Cells were resuspended in 50mMsodium phosphate (pH 7) 500mM NaCl 5mMdithiothreitol (DTT) (buffer A) supplemented with10mM benzamidine 1mM PMSF and 01mgmlDNase and lysed by sonication Cell debris was clearedby centrifugation at 29 220 g for 30min at 4C and theclarified lysate was mixed with glutathione-Sepharose 4B(GE Healthcare Lifesciences) for 90min at 4C Unboundproteins were collected and the affinity resin was washedextensively with buffer A hMUS81 WH domain proteinwas eluted from the affinity resin by addition of GST-tagged 3C protease (PreScission protease GEHealthcare Lifesciences) hMUS81 WH domain proteinwas concentrated by ultrafiltration and loaded onto apre-equilibrated Superdex 75 HR 10300 column (GEHealthcare Lifesciences) pre-equilibrated with 25mMsodium phosphate (7) 250mM NaCl 1mM DTT(buffer B) Fractions containing the hMUS81 WHdomain monomer were pooled and concentrated by ultra-filtration to 10mgml hMUS81 WH domain proteinsfor circular dichroism (CD) and DNA binding weregrown in Luria-Bertani media Buffer A was 50mMTris (pH 8) 500mM NaCl 5mM DTT and buffer Bwas 25mM Tris (8) 250mM NaCl 1mM DTT Thepoint mutations were made using the Quikchange sitedirected mutagenesis system (Stratagene) The MUS81WH deletion mutants were made using overlap PCR tosynthesise open reading frames devoid of the WH domainsequence (128ndash230) and the products were inserted intothe dicistronic expression plasmid via the EcoR1 andXho1 sites hMUS81 WH domain protein (residues 128ndash230) was expressed in E coli strain FB810 (BL21 (DE3)recA-) from a pET41a plasmid (Novagen) Uniformly 15Nlabelled WH domain protein was produced in minimalmedium M9 containing 1 gl (15NH4)2SO4 (SigmaAldrich) as the nitrogen source The 13C15N labelledWH domain was produced in a similar fashion with13C6-D-glucose (Sigma Aldrich) as the only carbonsource

Nuclear magnetic resonance spectroscopy

Nuclear magnetic resonance (NMR) spectra wereacquired at 298K (except where indicated) on a VarianUnity PLUS spectrometer (operating at a nominal 1H fre-quency of 500MHz) equipped with a triple resonanceprobe including a Z-axis pulse field gradient coil Dataacquisition and processing leading to structure calcula-tions are described in Supplementary InformationAtomic coordinates of the final 20 simulated annealing

9742 Nucleic Acids Research 2013 Vol 41 No 21


MUS81 conformers and the list of experimental restraints(accession code RCSB 103460) have been deposited at theRCSB Protein Data Bank Chemical shifts for resonanceassignments for the MUS81 have been deposited at theBioMagResBank (accession code 17324)

Nuclease assays

Oligonucleotides used to make the structures weredescribed previously (15) XO1 and X26 were labelledon the 50-terminus with [g32P] Complementary oligo-nucleotides were added to the labelled oligonucleotidesin two molar excess to form the structures and purifiedon a 10 non-denaturing polyacrylamide gel in TBEReactions were carried out at 25C for 90min inreaction buffer containing 50-32P-labelled substrate in60mM sodium phosphate (pH 74) 5mM MgCl2 1mMDTT and 100 mgml bovine serum albumin (BSA) in atotal volume of 20 ml Incision products were separatedon a sequencing gel (Sequagel National Diagnostics) for2 h The gel was removed and dried and products werevisualized by auroradiography or a STORM pho-sphorimager (Molecular Dynamics) Quantification ofdata was carried out using ImageQuant TL software(GE healthcare Life Sciences)

Fluorescence anisotropyFluorescence polarization anisotropy experiments werecarried out as described previously (14)

In vivo analysis in S pombe

A mus81 base strain was constructed in S pombe which iscompatible for recombination-mediated cassette exchange(16) The open reading frame of mus81 as well as 160 bp ofupstream sequence was replaced with the ura4+ geneflanked by loxP and loxM3 sites Wild type and mutantmus81 constructs were re-introduced into the endogenouslocus by recombination-mediated cassette exchange usingloxP- and loxM3-flanked mus81 on a Cre-recombinase ex-pression plasmid This allows expression of mutants andwild-type at the native locus by the mus81 promoter Thepoint mutations were introduced by site-directed muta-genesis A WH domain deletion mutant mus81-DWHwas constructed by fusion PCR by combining residues1ndash116 and 215ndash608 and therefore eliminating residues117ndash214 All mus81 variants were confirmed bysequencing after integration

For DNA damage-sensitivity assays cells were grownat 30C overnight in yeast extract (YE) media seriallydiluted and spotted onto agar plates starting with 105

cells YEA plates containing methyl methanesulfonate(Fluka 64294) camptothecin (98 Acros Organics7689-03-4) 4-nitroquinoline 1-oxide (Fluka 73265)Hydroxyurea (98 Sigma H8627) at the indicated con-centrations were used Plates were incubated for 4 daysat 30C

Whole cell protein extracts were prepared by TCAprecipitation For the chromatin fractionation assaychromatin was separated by centrifugation through asucrose cushion as in Liang and Stillman 1997 (17)except the buffer for cell wall digestion with Zymolyase

and lysing enzymes was 50mM sodium citrate 40mMEDTA 12M sorbitol

RESULTS

Identification of an uncharacterized amino-terminaldomain within human MUS81

The major portion of the human MUS81 (Accession codeNP_0794043) amino-terminus has few characterizeddomains (Figure 1A) Our sequence alignments and struc-ture predictions identified a conserved region of 100amino acids (residues 128ndash230 from human MUS81)present in most eukaryotic MUS81 orthologues exceptingnematode and fruit fly species (Figure 1B) Analysis ofthis region by the Phyre recognition server (18) gave abest fold prediction with an E-value of 034 as a WHtertiary fold A recombinant form of residues 128ndash230from human MUS81 was prepared and characterized(see Supplementary Methods) Size exclusion chromatog-raphy and CD revealed the protein migrated as an115 kDa protein and exhibited CD spectra consistentwith a predominantly helical secondary structure (datanot shown)

Structure of the hMUS81 WH domain

The 3D structure of residues 128ndash230 from hMUS81 wasthen determined by heteronuclear NMR (Figure 1C andSupplementary Table S1) The structure revealed a helicaldomain that adopts a WH tertiary fold (Figure 1D)commonly found in a large number of DNA-bindingproteins (19) The hMUS81 WH domain contains twofunctionally important elements of WH proteinsnamely the recognition helix a3 and the szlig-hairpin lsquowingrsquomotif (Figure 1D) The short wing motif of hMUS81 iswell ordered even in the absence of DNA Strikingly therecognition helix (helix a3) is at least two turns shorterthan many other WH domain recognition helices andits amino-terminal end is highly mobile despite thepresence of several prolines and a buried W182 sidechainSequences immediately preceding the recognition helix(helix a2a loop) segregate into two clade-specificclusters those belonging to higher vertebrate MUS81and those of lower vertebrates and eukaryotes(Figure 1B) Many MUS81 invariant residues are buriedincluding three that precede helix a1 (Y130 P132 andS146 Figure 1B) suggesting the angle between theamino-terminus and helix a1 is important The hMUS81WH domain is highly basic (pI 100) with 12 ArgLyssidechains that are distributed throughout the sequenceThese include two arginines within the recognition helix(R186 and R191 as discussed later) suggesting a possiblecharge complementarity appropriate for binding DNA(Figure 2A)Structural comparisons with the hMUS81 WH domain

indicate the closest similarity is to the manganese-bindingMnTR (Z-score of 91 and rmsd of 21 A over 66 C-alphaatoms) a metal-dependent bacterial repressor protein be-longing to a DtxR-like family of transcriptional regulators(2021) (Supplementary Figure S1 Figure 2A) These WHproteins bind DNA using a canonical DNA interaction

Nucleic Acids Research 2013 Vol 41 No 21 9743


seen previously (eg HNF-3g and E2F) involving majorgroove recognition by helix a3 and phosphodiesterbinding by one of the wing motifs (22) The similarity tohMUS81 WH domain strongly suggests it that too iscapable of a similar DNA-binding function Analysis ofthis structural similarity revealed a short motif T-X-X-G-X-Hy-AG (Hy=hydrophobic residue) within helix a4that together with preceding hydrophobic residues inthe recognition helix a3 provides a sensitive fingerprintfor identifying closely related WH proteins to hMUS81(Supplementary Figure S1) In hMUS81 T206 acts asan N-cap to helix a4 hydrogen bonding to mainchain

amides and also across to a carbonyl of the helix a3-szlig1loop similar to the metal-dependent transcriptional regu-lator WH domains The WH motif also includes L205 andG209 from hMUS81 that play important structural rolesat the amino-end of helix a4 and contact conserved buriedresidues from helix a1 A213 is also frequently a smallsidechain and packs against Y146 sidechain of helix a1

The hMUS81 WH domain binds DNA

Residues 125ndash244 of MUS81 were previously reported tobind to the RecQ helicase homologue BLM (13)

A

B

C D

Eme1

Mus811 551

1 583

1 379 Eme2

ERKXXXD

BLM binding244125

SLX4 binding861

WH

128 230

interaction

N

C

recognition helix (α3)

wing

N

C

α1α2α4

szlig1

szlig2

TThuman 130 140 150 160 170 180 190 200 210 220 230

human Y P S W L G R G VILL LYR L K ELLQR A R S R ALRS L NLV R RY LT ELAQGS W A H AR V EHLNPNGHHF T E C QKSP VAPG A P P HR L THQPA S PE L KLAESEGLSLLNVGIGchimp Y P S W L G R G VILL LYR L K ELLQR A R S R ALRS L NLV R RY LT ELAQGS W A H AR V EHLNPNGHHF T E C QKSP VAPG A P P HR L THQPA S PE Kmacaque Y P S W L G R G VILL LYR L K ELLQR A R S R ALRS L NLV R RY LT ELAQGN W A H AR M EHLNPNGYYF T E C QKAP VAPG A P P HR L THQPA S PE L KLAESEGLSLLNVGIGcow Y P S W L G R G AVLL LYR L K ELLQR A R S R ALRS L NLV R RY LT ELAQGS W A H AR Q EHLNPSGQGF T E C PKVP VAPG A P P HR L THQPA S PQ L KLADSEGLGLLNVGSGdog Y P S W L G R G AVLL LYR L K ELLQR A R S R ALRS L NLV R RY LT ELAQGS W A H AR L ERLNPSGRSF T E C QKAP VALG T P P HR L THQPA S PE L KLAESEGLSSLDVGFRrabbit Y P S W L G R G AVLL LYQ L K ELLQR A R S R ALRS L NLV R RY LT ELAQGG W A H AR L EHL T E C QMAA VTPG S S P HK L AHQPA S AE L KLAESGPRSVDAGVGmouse Y P S W L G Q G EILL LYR L K ELLQK A R S K ALRS L NLI G RY LT ELAQSVG W A N AR Q EHLNSDGHSF T E C QKTP VVPG S P P HR L THRPA A PE L KLAEAEGLSTRHAGFRrat Y P S W L G Q G EILL LYR L K ELLQK A R S R ALRG L NLV R RY LT ELAQNAG W A N AR Q EHLNSDGHSF T E C QKTP VVPE S P P HR L THRPA A PE L KLAEAEGLSTLNTAFQopossum Y P S W L G R A ALLL LYK L K ELLKR E R S Q ALRS L NLI K RY LT ELCRGS R P H AR I EQLDPRSQGS T D C QEAP VALG SS P P HR L TQRPA S PG L ELAEAEGPTEPPARIQplatypus Y P S W L G R G VLLL MYR L R ELQQR E G R G ALRS L NLV R RQ RL GLVQGS R P R PY L EQRRPSGLSF S E C RESP VVPS ST P P RR L THRPDRQV A PE L PGGQGRTGELGGRGzebrafish Y P S W L G K G AVLL LYR M R ELQTE Q K S T SVST I ELV K RY LT SLAEKRE V Q R GY T HIQMPGSKGF F N A PLCE SFTVPDLG KY A S QK THNPA S DQ L KLDSEETGTLHEDVDxenopus Y P S W L G R G AVLL LYR M K ELQKE Q K S T SVST I ELV K RY LT ELAQQRE V Q R GY T DTKSPSSRGY T A A PLCD SFTLTDPN KY A T QK I THSPA S EK L RLEEAEDQGRADGFPfugu Y P S W L G K G AVLL LFR M K ELQAE Q K S T SVST I NLV K RY LT AVAEKRE V Q R GY A ESQNPGSKGF F M A QFCD SFTVPDLG KY A S QK I THNPA S ED L RLDSEHVKKSSTEVNtetraodon Y P S W L G K G AVLL LYR M K ELQTE Q K S T SVST I NLV K RY LT VLAEKRE V Q R GY A ESQISGSKGF F M A QYCD SFTVVNLG KY A S QR I THNPA S ED L RLDSVEHVKEASTEVAmedaka Y P S W L G R G AVLL LYR M K ELQAE Q K S T SVST I NLV K RY LT ALARKRE V Q R GH A QSQVPDGRSY F L A PLCD SFTVPDPG KY A S QK L THSPA S EE L RLAAERQELEstickleback Y P S W L G K G AVLL LYR M K ELQAE Q K S T SVST I NFV K RY LT ALAELQD V Q R GY T QMQVPGSKGY F M A LLCD SFSVPDLG KY A S RK I THNPA S ED L RLESVEHDPKDVPciona1 Y P S W L G K G ALLI LSR M K ELQRE Q K S T SMAT I ELV K KY IT ELAQQRT V V N AY A NKASENSFGY K A A TFCT SFTVPDPG QY A S NK K HNNPA E DS L KLDIENQNKGITNTciona2 Y P S W L G K G ALLI LYR M K ELQRE Q K C T SMGG I NLV K KY LT QLAEKRE V V N AY A DHTSVEGLGY K A A KFST SFSVPDSG QY A S AK N HSNPA H DA I KLDQKQESLNQDLTKRpombe Y P S W L G Y G SILC LYM A K QIVTM Q S M T AMKT I NLV Q KY LT EVCIRKV V S R AY A LNKHEF T P A PYCD SFGSATDRN RY A S TK Y TGHPS C DD E RLAKVDDSFQRKHTVScerevisiae Y P S W L G K G AILL LLE V K QIIEV G H E G SIAA K SLV E RY LT ELTKTRK I K R GY S LNAIPRG S E A KYSD CMTPNFSTK FY A S KH L EGRPK S EE V SLKTADGISFPKENEEPNEYSVTRNashbya Y P S W L G K G AILL ALE L K EIIDA A V E S AIKV I DLM E RY VT QMAEKRK V R R AY G LGCPSRG T E A KYCD SFVSNPLTR FY A T DH L QGRPR I EA E TLKHADSVIFPEDCPYcandida1 Y P S W L G H G AILI LYL M K RIIAG T R E S SAKT E NLI T IY LT SLAQKPK V K R GF A YDKDRNG T E A PYCD SFSNNAASK FY A S NH S TGRSPK F DE V QLKTAIGLSSPPEGSKemericella Y P S W L G L G ALIL LSS M K ELIEK Q S K T SMKI L ELV T KY LT EVAKPKP A A R PF A LDENSNQS T A A PYCD SFTVPSDPT FF A N TK Y RGHPLK S EE W AIKKTAQKSIQNTLaspergillus Y P S W L G L G ALLL LAT L K QLIEK Q S K T SMKT L ELV E RY LT EVAKAKP V S R PY G LDENASQG T A A PYCD SFTAPPDPS FY A N QK Y HGRPLR A EE W RIKKTLPGGDLN

α1 α2 α3 β1 β2 α4

FEN1 binding 551

Figure 1 Human MUS81 contains a WH domain (A) Domain structure of human MUS81 and its partners EME1 and EME2 The red boxindicates the catalytic motif within the nuclease domain (mauve) HhH motifs are shown in green Regions involved in known proteinndashproteininteractions are indicated (B) Structure-based sequence alignment of selected eukaryotic MUS81 WH domains secondary structures above thesequences are from the NMR structure Mutated residues in human helix a3 are indicated by blue triangles (C) Ensemble of NMR structures of WHdomain The domain comprises four helices (residues 137ndash150 160ndash169 183ndash191 207ndash220) and two szlig strands (residues 194ndash197 and 202ndash205) (D) Arepresentative WH domain structure indicating the location of the recognition helix and wing motif



However we were unable to demonstrate this interactionusing purified recombinant proteins (Fadden data notshown) The close similarity to transcriptional repressorsstrongly suggests that the hMUS81 WH domain woulduse a similar duplex DNA-binding mechanism NMRchemical shift titration with N15-labelled MUS81 WHdomain and duplex DNA shifted residues within regions131ndash142 (amino-terminus and helix a1) and 186ndash191 (rec-ognition helix a3 including R186 L188 and R191) thusimplicating these regions in DNA binding (SupplementaryFigure S2A and B)

The contribution of different regions of the WH domainto DNA binding was measured by fluorescencepolarization (Figure 2B) Carboxy-terminal truncationsto the WH domain resulted in a slight increase in DNAbinding (C 128ndash221) (Figure 2B) However truncationof both termini reduced DNA binding (NC 137ndash221) dra-matically Together these results suggest a role for boththe amino-terminus and the recognition helix of the WHdomain in binding DNA

To establish whether MUS81 WH domain is a majorgroove binding canonical WH domain individual

mutations within helix a3 (W182A R186A R191AR196A and R202A) were prepared These mutationsmarkedly reduced DNA binding (Figure 2B) CombiningR186A and R191A mutations from the recognition helixincreased the dissociation constant 10-fold stronglyimplicating helix a3 as a crucial element that engagesdouble-stranded DNA in a similar manner to canonicalWH domains

The WH domain modulates the incision activity ofMUS81 complexes

To study the effect of the WH domain of MUS81 onincision activity we prepared recombinant MUS81-EME1 and MUS81-EME2 complexes and compared theendonuclease activity of 25 fmol of each on splayed arm30-flap fork and Holliday junctions (Figure 3A) Thecatalytically impaired MUS81 D339N mutation (2324)inactivated both MUS81-EME1 and MUS81-EME2complexes (Figure 3A) confirming that the nucleaseactivity was specific To compare the effects of removalof the WH domain on the endonuclease activity of

B

A

R186

R191

N

hMus81 WH domain MntR (2HYG)(Pennella amp Giedroc 2005)

DtxR (1C0W)(Pohl et al 1999)

0

10

20

30

40

50

60

70

80

WT

(128

-230

)ΔC

(128

-221

)ΔN

ΔC (1

37-2

21)

W18

2A

R186

A

R191

A

R196

A

R202

AR1

86 R

191

Dis

soci

atio

n co

nsta

nt (

Kd)

Kd =66 μM +- 065 se

[Mus81 WH] microM

0 20 40 60

Flu

ores

cenc

e A

niso

topy

0020040060080101201401601802022024

Figure 2 The hMUS81 WH domain recognition helix binds DNA (A) Comparison of close structural relatives of hMUS81 WH domain withselected members of the DtxR repressor family The recognition helix (cyan cylinder) and wing motif (green szlig-strands) are shown with selected basicresidues represented as sticks DtxR is bound to dsDNA in the canonical WH domain DNA-binding mode (B) Dissociation constants (Kd) obtainedfor hMUS81 WH domain and selected mutants binding to duplex DNA (units mM se= standard error) Values were obtained from fluorescenceanisotropy curves using 50 nM DNA and purified hMUS81 WH domains



A

GGACATCTT

GTACCTCGA

CATGGAGCT

Wild-type MUS81-EME2 ΔWH MUS81-EME2

CCAGTGCCTTGCTAGGTCACGGAACGAT

GGACATCTT

GTACCTCGA

CATGGAGCT

MUS81-EME1

WT

ΔWH

D33

9N

WT

D33

9N

EME1 EME2

WT

D33

9N

WT

D33

9N

WT

D33

9N

WT

D33

9N

ΔWH

ΔWH

ΔWH

ΔWH

ΔWH

WT

ΔWH

D33

9N

WT

D33

9N

ΔWH

EME1 EME2 EME1 EME2 EME1 EME2

SA 3rsquo flap fork HJ

- - - -

225

76

2601401007050403525

15

10

GST-MUS81

EME2

GST-MUS81EME1

WT

WT

EME1 EME2

WT

WT

EME1 EME2

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

SA 3rsquo flap

--

B

GGACATCTT

GTACCTCGA

CATGGAGCT

TACCAGTGCCTTGCTAATGGTCACGGAACGAT


-

∆WHDWTbuffer

CGTTCCGT

S 3rsquo 5rsquo F HJ

nHJ

mH

J


nHJ

mH

J


nHJ

mH

JC D

Figure 3 The WH domain modulates the endonuclease activity of MUS81 complexes (A) Effect of WH domain on incisions by MUS81 complexes(A) SDSndashPAGE gels showing purified wild-type (WT) MUS81-EME1 and MUS81-EME2 proteins The anomalous migration pattern of EME1 andhas been observed before (2) (B) Sequencing gels showing products of incision reaction by 25 fmol purified recombinant MUS81-EME12 wild-typeor WH domain complexes and the MUS81 D339N dead mutant on 75 fmol substrates as indicated labelled with 32P (asterisk) () is proteinbuffer only SA is splayed arm HJ Holliday junction Arrows show the positions of the incisions made Schematic below shows the cleavage patternof MUS81 at the junction on model 30 flap complexes Large and small arrows indicate the relative intensities of the incisions made by the MUS81complexes on the 30 flap (C) Sequencing gel showing products of incision reaction by purified recombinant MUS81-EME2 wild-type or WD domaina3 helix R186191A mutants on 30 flaps and forks (D) Sequencing gels showing incision sites of MUS81-EME2 and MUS81WH-EME2 on varioussubstrates S is splayed arm 30 is 30 flap 50 is 50 flap F is fork HJ is Holliday junction nHJ is nicked Holliday junction mHJ is mobile Hollidayjunction



MUS81-EME1EME2 complexes the data for splayedarm 30 flap and fork substrates from Figure 3A werequantified by densitometry and the proportion of eachband calculated as a percentage of the total present inthe lane (Supplementary Table S2) In general MUS81-EME2 complexes converted more of its substrates toproducts than the same concentration of MUS81-EME1suggesting that MUS81-EME2 is intrinsically more active(Figure 3A and Supplementary Table S2) This is in agree-ment with another study (14) Furthermore bothcomplexes cleaved substrates with a duplex downstreamfrom a junction in preference to splayed arms Under theconditions of these experiments which contained 25 fmolof protein complexes and 75 fmol of substrate MUS81-EME1 complex was unable to cleave the splayed arm(Figure 3A and Supplementary Table S2) whereas thesame amount of MUS81-EME2 converted nearly half ofit into incision products (Figure 3A and SupplementaryTable S2) Deletion of residues 128ndash230 corresponding tothe WH domain from MUS81-EME2 reduced totalcleavage of the splayed arm substrate and the incisionsoccurred at the same position as wild-type (Figure 3A andSupplementary Table S2)

On both 30 flaps and forks MUS81-EME1 made threeincisions of similar intensity upstream from the junctionand deletion of the WH domain resulted in decreased totalincision activity (Supplementary Table S2) This contrastswith previous studies where it was shown that removal of246 amino acids from the N-terminus of MUS81 did notimpact on the incision activity of recombinant MUS81-EME1 complexes (425) although the EME1 used inthese studies was derived from a shorter transcriptvariant (26) More than 95 of the 30 flap and fork sub-strate is cleaved by wild-type MUS81-EME2 under theseconditions and deletion of the WH domain reduced thisto 89 and 82 for the 30-flap and fork respectivelyMUS81-EME2 makes incisions further upstream fromthe junction than MUS81-EME1 (Figure 3A) on thesesubstrates and the nicked Holliday junction (Figure3C) Deletion of the WH domain results in movement ofthe dominant incision site closer to the junction and indi-cates that stabilization of DNA binding of the complex bythe WH domain of MUS81 has a role in positioning theincision sites in MUS81-EME2 Both MUS81-EME1 andMUS81-EME2 cleaved intact Holliday junctions less wellas observed previously for MUS81-EME1 (Figure 3A) (4)and there was no cleavage of single-stranded DNA (datanot shown) The MUS81 R186AR191A double mutationwithin the helix a3 of the WH domain which reduced theDNA-binding affinity (Figure 2B) also showed a similarincision pattern to the WH deletion mutant (Figure 3C)These data imply that DNA binding by the WH domain inMUS81 increases the activity of both MUS81 complexesin vitro and allows incisions to be made further from thejunction by MUS81-EME2 complexes on 30 flaps forksand nicked Holliday junctions

The substrate preferences of MUS81-EME2 have notbeen previously characterized so we tested the incisionactivity of MUS81-EME2 and the MUS81WH domaindeletion mutant on a range of substrates that included amobile Holliday junction (X26) containing 26 centrally

placed complementary base pairs allowing the junctionto migrate This structure is thought to be more represen-tative of an in vivo migratable Holliday junction and wasshown previously to be weakly cleaved by purifiedMUS81-EME1 (2) Here we show that WT MUS81-EME2 did not cleave the mobile HJ and instead preferredto cleave substrates with a duplex downstream from anick or discontinuity Surprisingly fewer incisions weremade on a 50 flap indicating that its upstream duplex maybe inhibiting binding of MUS81-EME2 to the down-stream single strand

Role of the WH domain in vivo

The structural similarities to repressors the WH domainDNA-binding properties and effects on the endonucleolyticincision combined provide compelling evidence that inhuman MUS81 the WH domain of MUS81 binds duplexDNA Human EME1 and EME2 are both distant homo-logues of S pombe Eme1 although human EME2 is muchshorter (2) To address the role of theWHdomain in vivo wemutated S pombe mus81 while keeping the gene expressedby its own promoter at the native locus WH domaindeletion and the double point mutation D395AD396A (acatalytically dead form of mus81) (27) reduced S pombeviability slightly and conferred similar levels of sensitivityto chronic exposure toDNA-damaging agents as deletion ofthe entiremus81 gene (Figure 4) N-terminal substitutions inregions implicated in DNA binding by the human WHdomain were tested Mutation of the tetra-basic motifRKRK (residues 113ndash116) to alanines did not sensitize thecells to DNA damage However mutating YR at 122ndash123 inthe N-terminal region directly adjacent to helix a1 increasedsensitivity to all DNA-damaging agents to levels similar tothat of mus81-DWHD The double mutants R165AR168Aand H189AK192A located just outside the region of helixa3 had no effect on the DNA damage sensitivityReplacement of K176 and K181 [corresponding to R186and R191 in human (Figure 1)] with alanine also results inno change in sensitivity However substitution of K176 andK181 (KE) with glutamate which enhances the effect oflosing the basic sidechains resulted in an intermediate sen-sitivity to HU and 4NQO but not to CPT or MMS Thissuggests that these residues are involved in processing asubset of structures produced during the repair of lesionsinduced by these agents The KE mutant is proficient inmeiosis with a spore viability similar to wild-type incontrast to mus81 and mus81-DWHD which displayedspore viabilities of lt1 of wild-type (data not shown)Combining the K176AK181A double mutation withR165AR168A (KER) results in a moderate increase in sen-sitivity to HU and 4-NQO and a similar phenotype tomus81-DWH These results show that the WH domain ofMus81 is crucial for resistance to DNA-damaging agentsin S pombe and implies that this in vivo function has beenretained in mammals

WH mutations do not affect protein localization

The N-terminus of Mus81 was shown to be required fornuclear localisation in Saccharomyces cerevisiae (28)We therefore investigated the localization of mus81-WH



Figure 4 Mutations in the WH domain of S pombe mus81 phenocopies a mus81 deletion Top panel shows the S pombe WH domain sequence andstructural motifs predicted from the alignment in Figure 1 The mutated residues are indicated Bottom panel shows spot tests on agar platescontaining DNA damaging agents CPT campothecin MMS methylmethanesulfonate HU hydroxyurea 4NQO 4-Nitroquinoline 1-oxide



mutant complexes in S pombe cells by chromatin fraction-ation A C-terminal TAP tag was fused to the endogenouscopy of the mus81 mutants (Figure 5A) TAP-tagged wild-type Mus81-WH Mus81-KE and Mus81-KER proteinswere expressed at their expected sizes and at similar levels(Figure 5A) The tagged strains did not have a significantphenotype For the chromatin fractionation exponen-tially growing cells were harvested and the cell walldigested After lysis the whole cell extracts (W) wereseparated using sucrose cushion into soluble (S) and chro-matin fraction (C) (Figure 5B) The fractions of wild-typeand mutant C-terminally TAP-tagged strains wereanalysed by western blot (Figure 5B) The TAP-taggedMus81 Mus81-WH and Mus81-KER proteins eachaccumulated in the chromatin fraction However thelevel of Mus81-WH was reduced in the chromatinfraction compared with Mus81 and Mus81-KER but itis unlikely that this would cause the observed phenotypeas the Mus81-KER mutant with a similar phenotypeshows levels comparable with wild-type Mus81 in thechromatin fraction We conclude that the phenotype ofthe mus81-WH mutants was not due to mislocalizationof the proteins in the cell

DISCUSSION

A role for the WH domain in MUS81-EME2 complexes

We report the identification 3D structure and both in vitroand in vivo characterization of a previously unnoticed WHdomain within MUS81 In human recombinant MUS81this domain appears to enhance the incision activity ofboth MUS81-EME1 and MUS81-EME2 complexes andmodulate the positioning of MUS81-EME2 incisions on 30

flaps forks and nicked Holliday junctionsOur results for MUS81-EME1 contrasts with studies

where the endonuclease activity of the complex was notaffected by the removal of the N-termini of both MUS81and EME1 (42526) There are currently two acceptedtranscript variants of EME1 which have been reportedto display different levels of activity when recombinantforms of the MUS81-EME1 were expressed andimmunoprecipitated from Hela cells (26) although thesehad additional single amino acid additions and substitu-tions We have used the longer EME1583 transcript variantof EME1 isolated by Ciccia et al (2) which has a 13amino acid insert after residue 371 residing within the36R linker region (residues 368ndash403) identified by Changet al (4) and shown to be important for DNA binding andendonuclease activity Their experiments suggest thatlengthening the 36R linker by replacing it with the equiva-lent region in zebrafish reduced the endonuclease andDNA-binding activity of recombinant N-terminallytruncated MUS81-EME1570 complexes and it waspostulated that MUS81-EME1583 might be less activebecause of the length of the insert In our experimentsthe 13 amino acids insert might inhibit binding of the36R region of EME1 to DNA in such a way thatbinding of the WH domain to duplex DNA becomes dom-inant and would explain why removal of the WH domaindecreases the activity in the MUS81-EME1583 complexDetailed kinetic comparisons of recombinant versions ofboth EME1 transcript variants will be required to ascer-tain the contribution of the WH domain to the endonucle-ase activity of each MUS81-EME1 complexThe WH domain influences the positioning of the

incision site made by MUS81-EME2 on substrates suchas 30 flaps forks and nicked Holliday junctions thatcontain an upstream and downstream duplex anddeletion of the WH domain consistently moves themajor incision site towards the junction by up to fivebases (Figure 3 Supplementary Table S2) This isprobably a result of enhanced bending or opening of theduplex at the junction and given that the incision patternis exactly the same on all three of these substrates weconclude that the WH domain must bind duplex DNAor interact with subdomains in the MUS81-EME2complex or both Alternatively it is possible that it maycontribute to substrate recognition by stable binding ofthe downstream duplex as well as the MUS81 (HhH)2domain as modelled by Chang et al (4)The splayed arm substrate lacks a downstream duplex

and we find that in our assays the WH domain increasescleavage activity on splayed arms by MUS81-EME2 Theactivity of the MUS81-EME1 complexes is weaker for allthe substrates and cleavage of the splayed arm was not

anti-tubulin

control

W S C

WT WHD KER

W S C W S C W S C

PAP

anti-H3

A

80kDa

58kDa

WT

501

mus

81W

T

mus

81ΔW

HD

mus

81K

176E

K18

1E

mus

81K

176E

K18

1ER

165A

R16

8A

CTAP

B

Ponceau

Figure 5 Mutations in the WH domain do not affect cellular localiza-tion of Mus81-Eme1 (A) Top panel Immunoblot of whole-cell extractsfrom cells expressing C-terminally TAP-tagged Mus81 wild-type andmutant proteins Bottom panel Ponceau staining of the immunoblot toshow protein loading TAP-tagged proteins were detected with peroxidaseanti-peroxidase (PAP) The first lane contains extract from an untaggedcontrol strain (501) (B) Chromatin fractionation of TAP-taggedMus81+ Mus81-WH and Mus81-KER W whole extract S solublefractionsupernatant C chromatin fraction Samples were separated bySDSndashPAGE Control lanes contain an untagged wild-type strain TheMus81-TAP constructs were detected with peroxidase anti-peroxidase(PAP) The arrow indicates the TAP-tagged proteins (asterisk) is anon-specific band and (hash) is a presumed degradation product



detectable at the concentrations used N-terminallydeleted recombinant human MUS81-EME1570 complexeshave been shown to be capable of cleaving splayed arms athigh concentrations (100 nM) (4) therefore it is likely thatthe MUS81 (HhH)2 domain or other components of theC-terminal complex can interact weakly with the down-stream single strand of the splayed arm in MUS81-EME1It is conceivable that the WH domain may stabilize aweaker interaction of the C-terminal domains with thedownstream single strand in the splayed armA hypothetical model for the interaction of MUS81-

EME2 with a fork is shown in Figure 6 We note thatcleavage of the 50 flap by MUS81-EME2 is poor andthe WH domain reduces this activity We speculate thatthe presence of a second duplex with inappropriatepolarity results in a non-productive interaction thatblocks substrate incision The N-terminal truncatedhuman MUS81-EME1 used in (4) was unable to cleave50 flaps even at the same concentration (100 nM) thatproduced incisions on a splayed arm indicating that themechanism of inhibition by the second duplex does notinvolve the WH domain

Implications for the activity of MUS81 complexesin the cell

The N-terminus of hMUS81 containing the WH domain iswell conserved in S pombe and we found that the WHdomain is essential for both meiosis and DNA damagetolerance in this organism The role of the Mus81 WHdomain in meiosis in S pombe suggests that its DNAbinding activity is necessary for the processing of inter-mediates such as D loops or Holliday junctions in theabsence of a Yen1GEN1 resolvase homologue (29) Wefound here that point mutations in two residues (K176Eand K181E) of the WH domain predicted to be involvedin DNA binding have no effect on meiosis but show a

different drug sensitivity profile to WH domain deletionmutants implying that there may be a separation offunction which is manifested as recognition andcleavage of different structures during replication forkrepair Further examination of this finding is necessarybut beyond the scope of this study Enhanced bendingor opening of the DNA at the junction of the structuresby MUS81-EME2 may yield products with larger singlestrand gaps and affect their downstream processingpossibly indicating a link between the different drugsensitivities of the KE mutants in S pombe and the roleof the WH domain in MUS81-EME2 activity

We note that duplex DNA binding by N-terminal frag-ments of MUS81 has been suggested to stimulate FEN1cleavage of double flap substrates (930) Such an inter-action may link the WH domain of MUS81-EME1EME2complexes to Okazaki fragment processing during stalledreplication fork repair

Overall our data suggest that EME1 gene duplicationin metazoan genomes may have resulted in differences inprocessing of the substrates cleaved by MUS81 Furtherstructural and biochemical analyses are evidently requiredto reveal the basis for DNA junction recognition andcleavage by full-length MUS81-EME complexes contain-ing the WH domain

ACCESSION NUMBERS

Coordinates have been deposited at the RCSB proteindata bank as RCSB ID [rscb103460] and PDB ID code[2mc3]

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Onlineincluding [31ndash40]

MUS81

EME2

EME2-NLD

MUS81-ND

incision

MUS81

EME2

MUS81-ND

WHD

5rsquo

incision3rsquo

5rsquo

3rsquo

bending

MUS81 opening

i ii

Figure 6 Model for the interaction of recombinant MUS81-EME2 with synthetic substrates (i) Substrates such as 30-flaps replication forks andnicked Holliday junctions contain a downstream duplex which is bound by the (HhH)2 domain of MUS81 causing conformational changes whichtrigger cleavage For splayed arms this interaction is weak and not detectable without the MUS81-WH domain (ii) The WH domain of MUS81plausably binds the upstream duplex near or at the junction and together with interaction with other domains bends the downstream duplex andorunwinds the upstream duplex thus moving the incision position further from the junction MUS81-ND is MUS81 nuclease domain EME2-NLD isEME2 nuclease-like domain MUS81WHD is MUS81 WH domain On splayed arms the interaction of the MUS81-WH domain described abovemay stabilize the weak binding of the MUS81-EME2 C-terminal domains sufficiently to facilitate cleavage but is not tight enough to allow openingbending required to move the incision position



ACKNOWLEDGEMENTS

The authors thank Alberto Ciccia and Steve West forproviding discistronic expression constructs forMUS81-EME1 and MUS81-EME2 complexes Theyalso thank Ian Hickson for providing recombinant BLMprotein Ms Sarah Curry who made the S pombe mus81-DD and mus81-R165168A mutants and helped to set upthe site-directed mutagenesis in S pombe and Ms RebeccaHaigh who made the mutant mus81-Y122AR123A

FUNDING

Funding for open access charge Cancer Research UKcore funding to the London Research Institute (toNQM) MRC grant [G1100074 to AMC]

Conflict of interest statement None declared

REFERENCES

1 CicciaA McDonaldN and WestSC (2008) Structural andfunctional relationships of the XPFMUS81 family of proteinsAnnu Rev Biochem 77 259ndash287

2 CicciaA ConstantinouA and WestSC (2003) Identificationand characterization of the human mus81-eme1 endonucleaseJ Biol Chem 278 25172ndash25178

3 NewmanM Murray-RustJ LallyJ RudolfJ FaddenAKnowlesPP WhiteMF and McDonaldNQ (2005) Structureof an XPF endonuclease with and without DNA suggests amodel for substrate recognition EMBO J 24 895ndash905

4 ChangJH KimJJ ChoiJM LeeJH and ChoY (2008)Crystal structure of the Mus81-Eme1 complex Genes Dev 221093ndash1106

5 NishinoT KomoriK IshinoY and MorikawaK (2005)Structural and functional analyses of an archaeal XPFRad1Mus81 nuclease asymmetric DNA binding and cleavagemechanisms Structure 13 1183ndash1192

6 TripsianesK FolkersG AbE DasD OdijkH JaspersNGHoeijmakersJH KapteinR and BoelensR (2005) The structureof the human ERCC1XPF interaction domains reveals acomplementary role for the two proteins in nucleotide excisionrepair Structure 13 1849ndash1858

7 EhmsenKT and HeyerWD (2009) A junction branch pointadjacent to a DNA backbone nick directs substrate cleavage bySaccharomyces cerevisiae Mus81-Mms4 Nucleic Acids Res 372026ndash2036

8 OsmanF and WhitbyMC (2007) Exploring the roles of Mus81-Eme1Mms4 at perturbed replication forks DNA Repair (Amst)6 1004ndash1017

9 ShinYK AmangyeldT NguyenTA MunashinghaPR andSeoYS (2012) Human MUS81 complexes stimulate flapendonuclease 1 FEBS J 279 2412ndash2430

10 FekairiS ScaglioneS ChahwanC TaylorER TissierACoulonS DongMQ RuseC YatesJR III RussellP et al(2009) Human SLX4 is a Holliday junction resolvase subunit thatbinds multiple DNA repairrecombination endonucleases Cell138 78ndash89

11 SvendsenJM SmogorzewskaA SowaME OrsquoConnellBCGygiSP ElledgeSJ and HarperJW (2009) MammalianBTBD12SLX4 assembles a Holliday junction resolvase and isrequired for DNA repair Cell 138 63ndash77

12 MunozIM HainK DeclaisAC GardinerM TohGWSanchez-PulidoL HeuckmannJM TothR MacartneyTEppinkB et al (2009) Coordination of structure-specificnucleases by human SLX4BTBD12 is required for DNA repairMol Cell 35 116ndash127

13 ZhangR SenguptaS YangQ LinkeSP YanaiharaNBradsherJ BlaisV McGowanCH and HarrisCC (2005)

BLM helicase facilitates Mus81 endonuclease activity in humancells Cancer Res 65 2526ndash2531

14 CicciaA LingC CoulthardR YanZ XueY MeeteiARLaghmani elH JoenjeH McDonaldN de WinterJP et al(2007) Identification of FAAP24 a Fanconi anemia core complexprotein that interacts with FANCM Mol Cell 25 331ndash343

15 ConstantinouA ChenXB McGowanCH and WestSC(2002) Holliday junction resolution in human cells two junctionendonucleases with distinct substrate specificities EMBO J 215577ndash5585

16 WatsonAT GarciaV BoneN CarrAM and ArmstrongJ(2008) Gene tagging and gene replacement using recombinase-mediated cassette exchange in Schizosaccharomyces pombe Gene407 63ndash74

17 LiangC and StillmanB (1997) Persistent initiation of DNAreplication and chromatin-bound MCM proteins during the cellcycle in cdc6 mutants Genes Dev 11 3375ndash3386

18 KelleyLA and SternbergMJ (2009) Protein structure predictionon the web a case study using the Phyre server Nat Protoc 4363ndash371

19 GajiwalaKS and BurleySK (2000) Winged helix proteinsCurr Opin Struct Biol 10 110ndash116

20 PennellaMA and GiedrocDP (2005) Structural determinants ofmetal selectivity in prokaryotic metalndashresponsive transcriptionalregulators Biometals 18 413ndash428

21 PohlE HolmesRK and HolWG (1999) Crystal structure ofa cobalt-activated diphtheria toxin repressor-DNA complexreveals a metal-binding SH3-like domain J Mol Biol 292653ndash667

22 WhiteA DingX vanderSpekJC MurphyJR and RingeD(1998) Structure of the metal-ion-activated diphtheria toxinrepressortox operator complex Nature 394 502ndash506

23 ChenXB MelchionnaR DenisCM GaillardPH BlasinaAVan de WeyerI BoddyMN RussellP VialardJ andMcGowanCH (2001) Human Mus81-associated endonucleasecleaves Holliday junctions in vitro Mol Cell 8 1117ndash1127

24 EnzlinJH and ScharerOD (2002) The active site of the DNArepair endonuclease XPF-ERCC1 forms a highly conservednuclease motif EMBO J 21 2045ndash2053

25 TaylorER and McGowanCH (2008) Cleavage mechanism ofhuman Mus81-Eme1 acting on Holliday-junction structures ProcNatl Acad Sci USA 105 3757ndash3762

26 BlaisV GaoH ElwellCA BoddyMN GaillardPHRussellP and McGowanCH (2004) RNA interference inhibitionof Mus81 reduces mitotic recombination in human cells MolBiol Cell 15 552ndash562

27 BoddyMN GaillardPH McDonaldWH ShanahanPYatesJR III and RussellP (2001) Mus81-Eme1 are essentialcomponents of a Holliday junction resolvase Cell 107 537ndash548

28 FuY and XiaoW (2003) Functional domains required for theSaccharomyces cerevisiae Mus81-Mms4 endonuclease complexformation and nuclear localization DNA Repair (Amst) 21435ndash1447

29 IpSC RassU BlancoMG FlynnHR SkehelJM andWestSC (2008) Identification of Holliday junction resolvasesfrom humans and yeast Nature 456 357ndash361

30 KangMJ LeeCH KangYH ChoIT NguyenTA andSeoYS (2010) Genetic and functional interactions betweenMus81-Mms4 and Rad27 Nucleic Acids Res 38 7611ndash7625

31 BoucherW (1996) AZARA v20 Department of BiochemistryUniversity of Cambridge UK

32 BrungerAT AdamsPD CloreGM DeLanoWL GrosPGrosse-KunstleveRW JiangJS KuszewskiJ NilgesMPannuNS et al (1998) Crystallography amp NMR system A newsoftware suite for macromolecular structure determination ActaCrystallogr D Biol Crystallogr 54 905ndash921

33 CornilescuG DelaglioF and BaxA (1999) Protein backboneangle restraints from searching a database for chemical shift andsequence homology J Biomol NMR 13 289ndash302

34 DelaglioF GrzesiekS VuisterGW ZhuG PfeiferJ andBaxA (1995) NMRPipe a multidimensional spectral processingsystem based on UNIX pipes J Biomol NMR 6 277ndash293

35 DossetP HusJC MarionD and BlackledgeM (2001) A novelinteractive tool for rigid-body modeling of multi-domain



macromolecules using residual dipolar couplings J BiomolNMR 20 223ndash231

36 KraulisPJ (1989) ANSIG A program for the assignment ofprotein 1H2D NMR spectra by interactive graphics J MagnReson 24 627ndash633

37 LingePJ and NilgesM (1999) Influence of non-bondedparameters on the quality of NMR structures a new force fieldfor NMR structure calculation J Biomol NMR 13 51ndash59

38 OttigerM DelaglioF and BaxA (1998) Measurement of J anddipolar couplings from simplified two-dimensional NMR spectraJ Magn Reson 131 373ndash378

39 RuckertM and OttingG (2000) Alignment of BiologicalMacromolecules in Novel Nonionic Liquid Crystalline Media forNMR Experiments J Am Chem Soc 122 7793ndash7797

40 WishartDS BigamCG HolmA HodgesRS and SykesBD(1995) 1H 13C and 15N random coil NMR chemical shifts ofthe common amino acids I Investigations of nearest-neighboreffects J Biomol NMR 5 67ndash81



fragments of MUS81 can bind DNA and stimulate theflap endonuclease activity of FEN1 The N-terminalregion also interacts with SLX4 a protein that isthought to act as a scaffold for the structure-specificendonucleases MUS81-EME1 XPF-ERCC1 and SLX1for recruitment to the repair of interstrand crosslinksand restart of damaged replication forks (10ndash12) TheN-terminus was also proposed to contain a Bloomsyndrome (BLM) protein-interacting domain (13)Here we identify for the first time a winged helix

domain (WH domain) within the N-terminal region ofhuman MUS81 that binds DNA increases the activityof MUS81-EME1EME2 complexes and influences theincision position of MUS81-EME2 but not MUS81-EME1 complexes on synthetic forks 30 flaps and nickedHolliday junctions Additionally in MUS81-EME2complexes it stimulates the cleavage of splayed armsMutations in the WH domain in Schizosaccharomycespombe render a similar sensitivity profile to DNAdamaging agents as a mus81 deleted strain implying thatthis domain has a critical role in DNA repair that has beenretained in human MUS81

MATERIALS AND METHODS

Protein purification

MUS81-EME1EME2 complexes (NCBI accessionnumbers MUS81NP_0794042 EME1 NP_001159603EME2 NM_001257370) (14) were produced inEscherichia coli BL21 (DE3) Rosetta pLysS (Stratagene)using the dicistronic expression plasmid derived frompGEX-KG MUS81-EME1EME2 wild-type and mutantcomplexes were purified as follows Cultures were grownin Luria-Bertani at 37C and induced with 25 mMIsopropyl-b-D-thiogalactoside (IPTG) at 18C overnightBacteria were harvested by centrifugation at 4000 g for15min at 4C and the pellets resuspended in 50mMsodium phosphate (pH 8) 300mM NaCl 2mMb-mercaptoethanol 10 glycerol 02 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate(CHAPS) (buffer A) supplemented with 10mMbenzamidine 1mM phenylmethylsulfonyl fluoride(PMSF) and protease inhibitors (Roche) Bacteria werelysed by sonication Cell debris was cleared by centrifuga-tion at 29 220 g for 60min at 4C and the clarified lysatewas mixed with glutathione-Sepharose 4B (GE HealthcareLifesciences) for 90min at 4C Unbound proteins werecollected and the affinity resin was washed extensivelywith buffer A Human MUS81 (hMUS81) complexeswere eluted with elution buffer [buffer A containing20mM reduced glutathione (pH 8)] The eluted proteinswere analysed by SDSndashPAGE hMUS81 complexes wereconcentrated by ultrafiltration and loaded onto a 1ml ofHiTrap Heparin-Sepharose column pre-equilibrated withbuffer A The hMUS81 complexes were eluted using aNaCl gradient (up to 1M NaCl in buffer A) over 10column volumes on an AKTA fast performance liquidchromatography system (GE Healthcare Lifesciences)Peak fractions from the Heparin-Sepharose columnwere pooled and concentrated by ultrafiltration The

concentrated sample was loaded onto a Superose 12 HR10300 column (GE Healthcare Lifesciences) pre-equilibrated with buffer A After analysis bySDSndashPAGE aliquots of complex-containing fractionswere flash-frozen in liquid nitrogen for assaying

hMUS81 WH domain purificationCultures were grown at 37C and induced by addition ofisopropyl-b-D-thiogalactoside (final concentration250 mM) and further incubation at 37C for 120minBacteria were harvested by centrifugation at 4000 gfor 15min at 4C Cells were resuspended in 50mMsodium phosphate (pH 7) 500mM NaCl 5mMdithiothreitol (DTT) (buffer A) supplemented with10mM benzamidine 1mM PMSF and 01mgmlDNase and lysed by sonication Cell debris was clearedby centrifugation at 29 220 g for 30min at 4C and theclarified lysate was mixed with glutathione-Sepharose 4B(GE Healthcare Lifesciences) for 90min at 4C Unboundproteins were collected and the affinity resin was washedextensively with buffer A hMUS81 WH domain proteinwas eluted from the affinity resin by addition of GST-tagged 3C protease (PreScission protease GEHealthcare Lifesciences) hMUS81 WH domain proteinwas concentrated by ultrafiltration and loaded onto apre-equilibrated Superdex 75 HR 10300 column (GEHealthcare Lifesciences) pre-equilibrated with 25mMsodium phosphate (7) 250mM NaCl 1mM DTT(buffer B) Fractions containing the hMUS81 WHdomain monomer were pooled and concentrated by ultra-filtration to 10mgml hMUS81 WH domain proteinsfor circular dichroism (CD) and DNA binding weregrown in Luria-Bertani media Buffer A was 50mMTris (pH 8) 500mM NaCl 5mM DTT and buffer Bwas 25mM Tris (8) 250mM NaCl 1mM DTT Thepoint mutations were made using the Quikchange sitedirected mutagenesis system (Stratagene) The MUS81WH deletion mutants were made using overlap PCR tosynthesise open reading frames devoid of the WH domainsequence (128ndash230) and the products were inserted intothe dicistronic expression plasmid via the EcoR1 andXho1 sites hMUS81 WH domain protein (residues 128ndash230) was expressed in E coli strain FB810 (BL21 (DE3)recA-) from a pET41a plasmid (Novagen) Uniformly 15Nlabelled WH domain protein was produced in minimalmedium M9 containing 1 gl (15NH4)2SO4 (SigmaAldrich) as the nitrogen source The 13C15N labelledWH domain was produced in a similar fashion with13C6-D-glucose (Sigma Aldrich) as the only carbonsource

Nuclear magnetic resonance spectroscopy

Nuclear magnetic resonance (NMR) spectra wereacquired at 298K (except where indicated) on a VarianUnity PLUS spectrometer (operating at a nominal 1H fre-quency of 500MHz) equipped with a triple resonanceprobe including a Z-axis pulse field gradient coil Dataacquisition and processing leading to structure calcula-tions are described in Supplementary InformationAtomic coordinates of the final 20 simulated annealing




Nuclease assays









RESULTS












A

B

C D

Eme1

Mus811 551

1 583

1 379 Eme2

ERKXXXD

BLM binding244125

SLX4 binding861

WH

128 230

interaction

N

C


wing

N

C

α1α2α4

szlig1

szlig2

TThuman 130 140 150 160 170 180 190 200 210 220 230


α1 α2 α3 β1 β2 α4

FEN1 binding 551










B

A

R186

R191

N



0

10

20

30

40

50

60

70

80

WT

(128

-230

)ΔC

(128

-221

)ΔN

ΔC (1

37-2

21)

W18

2A

R186

A

R191

A

R196

A

R202

AR1

86 R

191

Dis

soci

atio

n co

nsta

nt (

Kd)

Kd =66 μM +- 065 se

[Mus81 WH] microM

0 20 40 60

Flu

ores

cenc

e A

niso

topy

0020040060080101201401601802022024




A

GGACATCTT

GTACCTCGA

CATGGAGCT



GGACATCTT

GTACCTCGA

CATGGAGCT

MUS81-EME1

WT

ΔWH

D33

9N

WT

D33

9N

EME1 EME2

WT

D33

9N

WT

D33

9N

WT

D33

9N

WT

D33

9N

ΔWH

ΔWH

ΔWH

ΔWH

ΔWH

WT

ΔWH

D33

9N

WT

D33

9N

ΔWH



- - - -

225

76

2601401007050403525

15

10

GST-MUS81

EME2

GST-MUS81EME1

WT

WT

EME1 EME2

WT

WT

EME1 EME2

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

SA 3rsquo flap

--

B

GGACATCTT

GTACCTCGA

CATGGAGCT



-

∆WHDWTbuffer

CGTTCCGT


nHJ

mH

J


nHJ

mH

J


nHJ

mH

JC D


















DISCUSSION







anti-tubulin

control

W S C

WT WHD KER

W S C W S C W S C

PAP

anti-H3

A

80kDa

58kDa

WT

501

mus

81W

T

mus

81ΔW

HD

mus

81K

176E

K18

1E

mus

81K

176E

K18

1ER

165A

R16

8A

CTAP

B

Ponceau











ACCESSION NUMBERS


SUPPLEMENTARY DATA


MUS81

EME2

EME2-NLD

MUS81-ND

incision

MUS81

EME2

MUS81-ND

WHD

5rsquo

incision3rsquo

5rsquo

3rsquo

bending

MUS81 opening

i ii




ACKNOWLEDGEMENTS


FUNDING



REFERENCES
















































Nuclease assays









RESULTS












A

B

C D

Eme1

Mus811 551

1 583

1 379 Eme2

ERKXXXD

BLM binding244125

SLX4 binding861

WH

128 230

interaction

N

C


wing

N

C

α1α2α4

szlig1

szlig2

TThuman 130 140 150 160 170 180 190 200 210 220 230


α1 α2 α3 β1 β2 α4

FEN1 binding 551










B

A

R186

R191

N



0

10

20

30

40

50

60

70

80

WT

(128

-230

)ΔC

(128

-221

)ΔN

ΔC (1

37-2

21)

W18

2A

R186

A

R191

A

R196

A

R202

AR1

86 R

191

Dis

soci

atio

n co

nsta

nt (

Kd)

Kd =66 μM +- 065 se

[Mus81 WH] microM

0 20 40 60

Flu

ores

cenc

e A

niso

topy

0020040060080101201401601802022024




A

GGACATCTT

GTACCTCGA

CATGGAGCT



GGACATCTT

GTACCTCGA

CATGGAGCT

MUS81-EME1

WT

ΔWH

D33

9N

WT

D33

9N

EME1 EME2

WT

D33

9N

WT

D33

9N

WT

D33

9N

WT

D33

9N

ΔWH

ΔWH

ΔWH

ΔWH

ΔWH

WT

ΔWH

D33

9N

WT

D33

9N

ΔWH



- - - -

225

76

2601401007050403525

15

10

GST-MUS81

EME2

GST-MUS81EME1

WT

WT

EME1 EME2

WT

WT

EME1 EME2

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

SA 3rsquo flap

--

B

GGACATCTT

GTACCTCGA

CATGGAGCT



-

∆WHDWTbuffer

CGTTCCGT


nHJ

mH

J


nHJ

mH

J


nHJ

mH

JC D


















DISCUSSION







anti-tubulin

control

W S C

WT WHD KER

W S C W S C W S C

PAP

anti-H3

A

80kDa

58kDa

WT

501

mus

81W

T

mus

81ΔW

HD

mus

81K

176E

K18

1E

mus

81K

176E

K18

1ER

165A

R16

8A

CTAP

B

Ponceau











ACCESSION NUMBERS


SUPPLEMENTARY DATA


MUS81

EME2

EME2-NLD

MUS81-ND

incision

MUS81

EME2

MUS81-ND

WHD

5rsquo

incision3rsquo

5rsquo

3rsquo

bending

MUS81 opening

i ii




ACKNOWLEDGEMENTS


FUNDING



REFERENCES



















































A

B

C D

Eme1

Mus811 551

1 583

1 379 Eme2

ERKXXXD

BLM binding244125

SLX4 binding861

WH

128 230

interaction

N

C


wing

N

C

α1α2α4

szlig1

szlig2

TThuman 130 140 150 160 170 180 190 200 210 220 230


α1 α2 α3 β1 β2 α4

FEN1 binding 551










B

A

R186

R191

N



0

10

20

30

40

50

60

70

80

WT

(128

-230

)ΔC

(128

-221

)ΔN

ΔC (1

37-2

21)

W18

2A

R186

A

R191

A

R196

A

R202

AR1

86 R

191

Dis

soci

atio

n co

nsta

nt (

Kd)

Kd =66 μM +- 065 se

[Mus81 WH] microM

0 20 40 60

Flu

ores

cenc

e A

niso

topy

0020040060080101201401601802022024




A

GGACATCTT

GTACCTCGA

CATGGAGCT



GGACATCTT

GTACCTCGA

CATGGAGCT

MUS81-EME1

WT

ΔWH

D33

9N

WT

D33

9N

EME1 EME2

WT

D33

9N

WT

D33

9N

WT

D33

9N

WT

D33

9N

ΔWH

ΔWH

ΔWH

ΔWH

ΔWH

WT

ΔWH

D33

9N

WT

D33

9N

ΔWH



- - - -

225

76

2601401007050403525

15

10

GST-MUS81

EME2

GST-MUS81EME1

WT

WT

EME1 EME2

WT

WT

EME1 EME2

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

SA 3rsquo flap

--

B

GGACATCTT

GTACCTCGA

CATGGAGCT



-

∆WHDWTbuffer

CGTTCCGT


nHJ

mH

J


nHJ

mH

J


nHJ

mH

JC D


















DISCUSSION







anti-tubulin

control

W S C

WT WHD KER

W S C W S C W S C

PAP

anti-H3

A

80kDa

58kDa

WT

501

mus

81W

T

mus

81ΔW

HD

mus

81K

176E

K18

1E

mus

81K

176E

K18

1ER

165A

R16

8A

CTAP

B

Ponceau











ACCESSION NUMBERS


SUPPLEMENTARY DATA


MUS81

EME2

EME2-NLD

MUS81-ND

incision

MUS81

EME2

MUS81-ND

WHD

5rsquo

incision3rsquo

5rsquo

3rsquo

bending

MUS81 opening

i ii




ACKNOWLEDGEMENTS


FUNDING



REFERENCES





















































B

A

R186

R191

N



0

10

20

30

40

50

60

70

80

WT

(128

-230

)ΔC

(128

-221

)ΔN

ΔC (1

37-2

21)

W18

2A

R186

A

R191

A

R196

A

R202

AR1

86 R

191

Dis

soci

atio

n co

nsta

nt (

Kd)

Kd =66 μM +- 065 se

[Mus81 WH] microM

0 20 40 60

Flu

ores

cenc

e A

niso

topy

0020040060080101201401601802022024




A

GGACATCTT

GTACCTCGA

CATGGAGCT



GGACATCTT

GTACCTCGA

CATGGAGCT

MUS81-EME1

WT

ΔWH

D33

9N

WT

D33

9N

EME1 EME2

WT

D33

9N

WT

D33

9N

WT

D33

9N

WT

D33

9N

ΔWH

ΔWH

ΔWH

ΔWH

ΔWH

WT

ΔWH

D33

9N

WT

D33

9N

ΔWH



- - - -

225

76

2601401007050403525

15

10

GST-MUS81

EME2

GST-MUS81EME1

WT

WT

EME1 EME2

WT

WT

EME1 EME2

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

SA 3rsquo flap

--

B

GGACATCTT

GTACCTCGA

CATGGAGCT



-

∆WHDWTbuffer

CGTTCCGT


nHJ

mH

J


nHJ

mH

J


nHJ

mH

JC D


















DISCUSSION







anti-tubulin

control

W S C

WT WHD KER

W S C W S C W S C

PAP

anti-H3

A

80kDa

58kDa

WT

501

mus

81W

T

mus

81ΔW

HD

mus

81K

176E

K18

1E

mus

81K

176E

K18

1ER

165A

R16

8A

CTAP

B

Ponceau











ACCESSION NUMBERS


SUPPLEMENTARY DATA


MUS81

EME2

EME2-NLD

MUS81-ND

incision

MUS81

EME2

MUS81-ND

WHD

5rsquo

incision3rsquo

5rsquo

3rsquo

bending

MUS81 opening

i ii




ACKNOWLEDGEMENTS


FUNDING



REFERENCES















































A

GGACATCTT

GTACCTCGA

CATGGAGCT



GGACATCTT

GTACCTCGA

CATGGAGCT

MUS81-EME1

WT

ΔWH

D33

9N

WT

D33

9N

EME1 EME2

WT

D33

9N

WT

D33

9N

WT

D33

9N

WT

D33

9N

ΔWH

ΔWH

ΔWH

ΔWH

ΔWH

WT

ΔWH

D33

9N

WT

D33

9N

ΔWH



- - - -

225

76

2601401007050403525

15

10

GST-MUS81

EME2

GST-MUS81EME1

WT

WT

EME1 EME2

WT

WT

EME1 EME2

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

R18

6AR

191A

SA 3rsquo flap

--

B

GGACATCTT

GTACCTCGA

CATGGAGCT



-

∆WHDWTbuffer

CGTTCCGT


nHJ

mH

J


nHJ

mH

J


nHJ

mH

JC D


















DISCUSSION







anti-tubulin

control

W S C

WT WHD KER

W S C W S C W S C

PAP

anti-H3

A

80kDa

58kDa

WT

501

mus

81W

T

mus

81ΔW

HD

mus

81K

176E

K18

1E

mus

81K

176E

K18

1ER

165A

R16

8A

CTAP

B

Ponceau











ACCESSION NUMBERS


SUPPLEMENTARY DATA


MUS81

EME2

EME2-NLD

MUS81-ND

incision

MUS81

EME2

MUS81-ND

WHD

5rsquo

incision3rsquo

5rsquo

3rsquo

bending

MUS81 opening

i ii




ACKNOWLEDGEMENTS


FUNDING



REFERENCES





























































DISCUSSION







anti-tubulin

control

W S C

WT WHD KER

W S C W S C W S C

PAP

anti-H3

A

80kDa

58kDa

WT

501

mus

81W

T

mus

81ΔW

HD

mus

81K

176E

K18

1E

mus

81K

176E

K18

1ER

165A

R16

8A

CTAP

B

Ponceau











ACCESSION NUMBERS


SUPPLEMENTARY DATA


MUS81

EME2

EME2-NLD

MUS81-ND

incision

MUS81

EME2

MUS81-ND

WHD

5rsquo

incision3rsquo

5rsquo

3rsquo

bending

MUS81 opening

i ii




ACKNOWLEDGEMENTS


FUNDING



REFERENCES






















































ACCESSION NUMBERS


SUPPLEMENTARY DATA


MUS81

EME2

EME2-NLD

MUS81-ND

incision

MUS81

EME2

MUS81-ND

WHD

5rsquo

incision3rsquo

5rsquo

3rsquo

bending

MUS81 opening

i ii




ACKNOWLEDGEMENTS


FUNDING



REFERENCES















































ACKNOWLEDGEMENTS


FUNDING



REFERENCES















































a winged helix domain in human mus81 binds dna and modulates

Documents