Zurich Open Repository andArchiveUniversity of ZurichMain LibraryStrickhofstrasse 39CH-8057 Zurichwww.zora.uzh.ch

Year: 2020

Regulation of the MLH1-MLH3 endonuclease in meiosis

Cannavo, Elda ; Sanchez, Aurore ; Anand, Roopesh ; Ranjha, Lepakshi ; Hugener, Jannik ; Adam,Céline ; Acharya, Ananya ; Weyland, Nicolas ; Aran-Guiu, Xavier ; Charbonnier, Jean-Baptiste ;

Hoffmann, Eva R ; Borde, Valérie ; Matos, Joao ; Cejka, Petr

Abstract: During prophase of the first meiotic division, cells deliberately break their DNA1. These DNAbreaks are repaired by homologous recombination, which facilitates proper chromosome segregation andenables the reciprocal exchange of DNA segments between homologous chromosomes2. A pathway thatdepends on the MLH1–MLH3 (MutL) nuclease has been implicated in the biased processing of meioticrecombination intermediates into crossovers by an unknown mechanism3,4,5,6,7. Here we have biochem-ically reconstituted key elements of this pro-crossover pathway. We show that human MSH4–MSH5(MutS), which supports crossing over8, binds branched recombination intermediates and associates withMutL, stabilizing the ensemble at joint molecule structures and adjacent double-stranded DNA. MutSdirectly stimulates DNA cleavage by the MutL endonuclease. MutL activity is further stimulated byEXO1, but only when MutS is present. Replication factor C (RFC) and the proliferating cell nuclearantigen (PCNA) are additional components of the nuclease ensemble, thereby triggering crossing-over.Saccharomyces cerevisiae strains in which MutL cannot interact with PCNA present defects in formingcrossovers. Finally, the MutL–MutS–EXO1–RFC–PCNA nuclease ensemble preferentially cleaves DNAwith Holliday junctions, but shows no canonical resolvase activity. Instead, it probably processes mei-otic recombination intermediates by nicking double-stranded DNA adjacent to the junction points9. AsDNA nicking by MutL depends on its co-factors, the asymmetric distribution of MutS and RFC–PCNAon meiotic recombination intermediates may drive biased DNA cleavage. This mode of MutL nucleaseactivation might explain crossover-specific processing of Holliday junctions or their precursors in meioticchromosomes4.

DOI: https://doi.org/10.1038/s41586-020-2592-2

Posted at the Zurich Open Repository and Archive, University of ZurichZORA URL: https://doi.org/10.5167/uzh-199557Journal ArticlePublished Version

Originally published at:Cannavo, Elda; Sanchez, Aurore; Anand, Roopesh; Ranjha, Lepakshi; Hugener, Jannik; Adam, Céline;Acharya, Ananya; Weyland, Nicolas; Aran-Guiu, Xavier; Charbonnier, Jean-Baptiste; Hoffmann, Eva R;Borde, Valérie; Matos, Joao; Cejka, Petr (2020). Regulation of the MLH1-MLH3 endonuclease in meiosis.Nature, 586(7830):618-622.DOI: https://doi.org/10.1038/s41586-020-2592-2

1

RegulationoftheMLH1-MLH3endonucleaseinmeiosis1

2

EldaCannavo1#,AuroreSanchez1#,RoopeshAnand1#,LepakshiRanjha1,Jan-3

nikHugener2,CélineAdam3,4,AnanyaAcharya1,2,NicolasWeyland5,Xavier4

Aran-Guiu6, Jean-Baptiste Charbonnier7,8, Eva R. Hoffmann6,9, Valérie5

Borde3,4,JoaoMatos2andPetrCejka1,26

7

Affiliations:8

1Institute forResearch inBiomedicine,Universitàdella Svizzera italiana (USI),9

FacultyofBiomedicalSciences,Switzerland10

2Department of Biology, Institute of Biochemistry, Eidgenössische Technische11

Hochschule(ETH),Zürich,Switzerland12

3InstitutCurie,PSLResearchUniversity,CNRSUMR3244,Paris,France13

4ParisSorbonneUniversité,Paris,France14

5InstituteofMolecularCancerResearch,UniversityofZürich,Zürich,Switzerland15

6GenomeDamageandStabilityCentre,SchoolofLifeSciences,UniversityofSus-16

sex,Brighton,UK17

7I2BC, iBiTec-S, CEA, CNRS UMR 9198, Université Paris-Sud, Gif-sur-Yvette,18

France;19

8UniversitéParisSud,Orsay,France20

9DNRFCenterforChromosomeStability,DepartmentofCellularandMolecular21

Medicine,FacultyofHealthandMedicalSciences,UniversityofCopenhagen,Co-22

penhagen,Denmark23

24

#Theseauthorscontributedequally.25

26

Materials&Correspondence:PetrCejka, Institute forResearch inBiomedicine,27

ViaVincenzoVela6,6500Bellinzona,Switzerland;28

E-mail:petr.cejka@irb.usi.ch29

30

2

SUMMARY31

32

Duringprophaseofthefirstmeioticdivision,cellsdeliberatelybreaktheirDNA1.33

TheseDNAbreaksarerepairedbyhomologousrecombination,whichfacilitates34

proper chromosome segregation andenables reciprocal exchangeofDNA seg-35

mentsbetweenhomologouschromosomes,promotingthusgeneticdiversityin36

theprogeny2.Geneticandcellulardataimplicatedapathwaydependentonthe37

MLH1-MLH3(MutLγ)nucleaseinthebiasedprocessingofmeioticrecombination38

intermediatesintocrossovers,butmechanismsthatleaditsactivationwereun-39

clear3-7.Here,webiochemicallyreconstitutedkeyelementsofthispro-crossover40

pathway.First,weshowthathumanMSH4-MSH5(MutSγ),whichwasknownto41

supportcrossingover8,bindsbranchedrecombinationintermediatesandphysi-42

callyassociateswithMutLγ.Thishelpsstabilizetheensembleatjointmolecule43

structuresandadjacentdsDNA.Second,weshowthatMutSγdirectlystimulates44

DNAcleavagebytheMutLγendonuclease,whichdemonstratesanunexpecteddi-45

rectfunctionforMutSγintriggeringcrossing-over.Third,wefindthatMutLγac-46

tivityisfurtherstimulatedbyEXO1,butonlywhenMutSγispresent.Fourth,we47

alsoidentifythereplicationfactorC(RFC)andtheproliferatingcellnuclearanti-48

gen(PCNA)asadditionalcomponentsofthenucleaseensemble,andshowthatS.49

cerevisiae strains expressing PCNA-interacting peptide (PIP) box-likemutated50

MutLγpresentstrikingdefectsinformingcrossovers.Finally,weshowthatthe51

MutLγ-MutSγ-EXO1-RFC-PCNA nuclease ensemble preferentially cleaves DNA52

withHolliday junctions,butshowsnocanonicalresolvaseactivity. Instead, the53

multilayerednucleaseensemblelikelyprocessesmeioticrecombinationinterme-54

diatesbynickingdsDNAadjacenttothejunctionpoints9.SinceDNAnickingby55

MutLγisdependentonitsco-factors,theasymmetricdistributionofMutSγand56

RFC-PCNAonmeioticrecombinationintermediatesmaydrivebiasedDNAcleav-57

age.ThisuniquemodeofMutLγnucleaseactivationmightexplaincrossover-spe-58

cificprocessingofHollidayjunctionswithinthemeioticchromosomalcontext4.59

60

61

3

MAINTEXT62

63

MutLγisanATP-stimulatedendonuclease64

TostudyhumanMutLγ(MLH1-MLH3),weexpressedandpurifiedtheheterodi-65

merfrominsectcells(ExtendedDataFig.1a,b).Similarlytothemismatchrepair66

(MMR)-specific MutLα (MLH1-PMS2)10 and yeast Mlh1-Mlh311,12, the human67

MLH1-MLH3complexnon-specificallynickeddouble-strandedsupercoiledDNA68

(scDNA)withmanganeseasametalco-factor(Fig.1a,ExtendedDataFig.1c-e)13.69

MutationsintheconservedmetalbindingDQHA(X)2E(X)4EmotifofMLH3abol-70

ishedtheendonuclease,indicatingthattheDNAcleavageactivitywasintrinsicto71

theMutLγheterodimer(Fig.1a).ATPpromotedthenucleaseactivity>2-fold(Fig.72

1a,ExtendedDataFig.1f).Almostnonucleaseactivitywasobservedwithmagne-73

sium(ExtendedDataFig.1g),whichisbelievedtobethespecificmetalco-factor10.74

ExperimentswithvariousATPanalogsrevealedthatATPhydrolysisbyMLH1-75

MLH3wasrequiredforthemaximalstimulationofDNAcleavage(ExtendedData76

Fig. 1h,i). TheN-termini of bothMLH1 andMLH3 proteins contain conserved77

WalkermotifsimplicatedinATPbindingandhydrolysis14.Todefinewhetherthe78

ATPaseofMLH1,MLH3orbothsubunitsoftheheterodimerpromotesitsnucleo-79

lyticactivity,wepreparedtherespectivehMutLγvariantswithmutationsinthe80

ATPasemotifs of either subunit individually or combined (ExtendedData Fig.81

1j,k)14.WeobservedthattheintegrityoftheATPasedomainofMLH1,andtoa82

muchlesserdegreeofMLH3,promotedthenucleaseactivityofMLH1-MLH3(Ex-83

tended Data Fig. 1l,m). Without ATP, the ATPase-deficient variants of MLH1-84

MLH3boundandcleavedDNAsimilarlyasthewildtypecomplex(ExtendedData85

Fig.1l-n).TheMutLγcomplexdidnotcleaveoligonucleotide-basedHollidayjunc-86

tions(HJ)DNA(ExtendedDataFig.1o).87

88

MutLγandMutSγbindtoDNAjunctions89

Yeast and human MutLγ complexes bind DNA with a preference towards90

branched structures suchasHolliday junctions11,15. Similarly, recombinanthu-91

manMutSγwasshowntobindHJs8.WefoundthatthehumanandyeastMutSγ92

complexesboundevenbetterHJprecursorssuchasD-loops(ExtendedDataFig.93

4

2a-f).ThisbindingpreferenceagreeswiththeproposedearlyfunctionofMutSγ94

tostabilizenascentstrandinvasionintermediatesthatmatureintosingle-endin-95

vasions, which helps ensure their crossover designation16. In contrast, single-96

strandedDNA(ssDNA)ordsDNAwasnotboundbyMutSγ(ExtendedDataFig.97

2b-f). Electrophoretic mobility shift assays demonstrated that the MutSγ and98

MutLγ complexesmoderately stabilized each other at the DNA junctions (Ex-99

tendedData Fig. 3a-e). Accordingly, the respectivehumanor yeastMutSγ and100

MutLγcomplexesdirectlyphysicallyinteract(Fig.1bandExtendedDataFig.3f-101

h)17. The very slowmigrationof theprotein-DNA complexeswas indicativeof102

multiple units of the heterocomplexes bound to the DNA substrate (Extended103

DataFig.3i,j),asshownpreviouslyforyeastMutLγ11.Wenotethatthepresence104

ofDNAjunctionswasessentialforstableDNAbinding(ExtendedDataFig.3a,e),105

whichsupportsamodelwherethebranchedDNAstructureservesasanucleation106

pointforaMutSγ-MutLγfilamentthatthenextendstotheadjacentdsDNAarms18.107

108

MutSγpromotestheMutLγnuclease109

PreviousinvivoexperimentsimplicatedMutSγinthestabilizationofnascentDNA110

jointmolecules early in themeiotic pro-crossover pathway8,17,19, but whether111

MutSγisdirectlyinvolvedlaterinnucleolyticprocessingwasnotclear.Usingour112

reconstitutedsystem,weobserved~3-foldstimulationoftheMLH1-MLH3endo-113

nucleasebyMSH4-MSH5(Fig.1c,ExtendedDataFig.4a-c),whichwasdependent114

ontheMLH3metalbindingmotif(ExtendedDataFig.4b).ATPpromotedDNA115

cleavagebytheMutSγ-MutLγensemble,andasinreactionswithMutLγalone,the116

maximalnucleaseactivitywasobservedwhenATPhydrolysiswaspossible(Ex-117

tendedDataFig.4d).AsMLH1andMLH3,alsoMSH4andMSH5proteinscontain118

conservedATPasedomains14,16.TheATPbinding/hydrolysismotifsinMLH1and119

MSH5werebothcrucial,whilethemotifinMSH4waslessimportantandinMLH3120

appeareddispensable(Fig.1d,e,ExtendedDataFig.4e,f).TheATPasemotifmu-121

tationsinMSH4orMSH5insteaddidnotaffectthecapacityofthetwosubunitsto122

formacomplexorbindDNA(ExtendedDataFig.4g,h).Thestimulatoryeffectwas123

likely facilitated by direct physical interactions between the cognate heterodi-124

mers,asyeastMsh4-Msh5didnotpromotethenucleaseactivityofhumanMutLγ125

5

(ExtendedDataFig.4i,j).MutSγ-MutLγexhibitednodetectablestructure-specific126

nucleaseorresolvaseactivity(ExtendedDataFig.4k,l).127

To assess how the other human MutS homologue complexes, compared to128

MutSγ,promotetheactivityofMutLγ,wesupplementedtheMLH1-MLH3nucle-129

ase reactions with the MMR factors MutSa (MSH2-MSH6) or MutSb (MSH2-130

MSH3).MutSb,butnothMutSa, couldstimulatetheMLH1-MLH3nucleasetoa131

similarlevelasMutSγ(ExtendedDataFig.5a-c)13.Thisagreeswithpreviousex-132

perimentsshowingthatyeastMutLγcouldpartiallysubstituteMutLαinthere-133

pairofinsertion/deletionmismatchesinMMR20.Thesedataalsounderpinthein-134

volvementofMutLγinthemetabolismoftrinucleotiderepeatslinkedtoseveral135

neurodegenerativediseases,aswellas rareMLH3mutations found inpatients136

withhereditarynonpolyposiscolorectalcancer(HNPCC)/Lynchsyndromechar-137

acterizedbymicrosatelliteinstability13,21,22.138

139

EXO1promotestheMutSγ-MutLγnuclease140

Genetic experimentswith budding yeast revealed a structural (nuclease-inde-141

pendent)functionofExo1intheMlh1-Mlh3pro-crossoverpathway5.Theeffect142

wasdependentonitsdirectinteractionwiththeMlh1subunitofMutLγ5,23,butit143

wasunclearwhether the interplaydirectlyaffects theMlh3endonuclease,and144

whetherthisfunctionisconservedinhighereukaryotes.Totestfortheeffectof145

EXO1 on the nuclease activity ofMLH1-MLH3,we used the nuclease-deficient146

EXO1(DA)varianttopreventdegradationoftheresultingnickedDNA(Extended147

Data Fig. 6a). We observed no stimulation of the MLH1-MLH3 nuclease by148

EXO1(DA)alone,butEXO1(DA)promotedDNAcleavage~2-3-foldwhenMSH4-149

MSH5waspresent(Fig.1fandExtendedDataFig.6b,c).Morethan40%DNAwas150

cleavedusingonly20nMconcentrationofthemulti-proteinensemble(Fig.1f).151

IncontrasttoMSH4-MSH5thatmoderatelystabilizedMLH1-MLH3onDNA,we152

detectednosuchcapacityofEXO1(DA)(ExtendedDataFig.6d).YeastExo1(DA)153

couldnotsubstitutehumanEXO1(DA)inthenucleaseassays(ExtendedDataFig.154

6e), inagreementwithadirectphysical interactionbetweenhumanEXO1(DA)155

andMLH1-MLH3(Fig.1g).Finally,EXO1(DA)didnotpromotethenucleaseactiv-156

ityofMLH1-MLH3withMutSb(ExtendedDataFig.6f),indicatingthatEXO1likely157

6

specificallystimulatestheendonucleaseactivityofMutSγ-MutLγinvolvedinmei-158

oticrecombination.159

160

PCNApromotestheMLH3nucleaseensemble161

WenextsetouttotestwhetherMLH1-MLH3withitsco-factorscancatalyzeDNA162

cleavageunderphysiologicalconditionsinmagnesium.Whilealmostnonuclease163

activityofMLH1-MLH3aloneinmagnesiumwasobserved,weaknickingwasseen164

upon adding MSH4-MSH5, and the reactions were further stimulated by165

EXO1(DA)(Fig.2a).AsRFC-PCNAareknowntodirecttheMutLaendonucleasein166

mismatchrepair10,wetestedfortheireffectonMutLγ.Notably,weobservedad-167

ditional~2-foldstimulationofDNAcleavagewhentheRFC-PCNAcomplexwas168

included(Fig.2a,ExtendedDataFig.7a-e).Nucleolyticcleavagewasdependent169

ontheintegrityoftheMLH3metal-bindingmotif(ExtendedDataFig.7b).Yeast170

RFCcouldpartiallysubstituteforhumanRFC(ExtendedDataFig.7c),inaccord171

withthecapacityofyeastRFCtoloadhumanPCNAonscDNA10,24.Incontrast,no172

stimulationwasdetectedwhenRFCwasomitted fromthereactionmixtureor173

whenusing relaxedDNA (ExtendedData Fig. 7b,f), indicating that PCNAmust174

likelybeactively loadedonto scDNAbyRFC10,24.Nostimulationwasobserved175

whenusingyeastinsteadofhumanPCNA(ExtendedDataFig.7b),orinreactions176

withmanganese(ExtendedDataFig.7g).Thus,whileMutLγonitsownisapoor177

nuclease that requires manganese, MutSγ, EXO1 and RFC-PCNA activate it to178

cleaveefficientlyinabuffercontainingphysiologicalmagnesium,andthereaction179

isnolongerstimulatedbyadditionalmanganese(Fig.2b).180

RFC-PCNAcouldalsopromotethenucleaseactivityofMutLγalone,although181

toalesserextent(ExtendedDataFig.7h),suggestingthatMutSγandEXO1(DA)182

arenotstrictlyrequiredtomediatethestimulatoryeffectofRFC-PCNA.Inaccord,183

wefoundthatMutLγdirectlyphysicallyinteractswithPCNA(Fig.2c).ATPwas184

necessary for thenucleaseactivityof theensembleand forPCNA loading, and185

couldnotbereplacedbyADPorAMP-PNP,showingthatATPhydrolysiswasre-186

quired(Fig.2d,ExtendedDataFig.7i).Incontrasttothereactionsinmanganese,187

the integrity of the ATPasemotifs of all four MutSγ andMutLγ subunits was188

7

requiredformaximalcleavageactivityinmagnesium(Fig.2e,ExtendedDataFig.189

7j).190

Notably, the nuclease ensemble preferentially cleaved plasmid-length DNA191

withpalindromicrepeatsformingaHJ-likestructure(cruciformDNA,Fig.2f),in192

agreementwiththebindingpreferenceoftheMutSγandMutLγheterodimersto193

theserecombinationintermediates(ExtendedDataFig.2b-f)8,11,15.However,the194

activityofthecomplexonthecruciformDNAprimarilyyieldednickedproducts195

(Fig.2f),unlike canonicalHJ resolvases thatgive rise to linearDNAuponcon-196

certedcleavageofbothDNAstrandsatthejunctions25.Wealsonotethatwedid197

notobserveanycleavageofmodelHJorD-loopoligonucleotide-basedsubstrates198

(ExtendedDataFig.7k).OurdatasuggestthatthehMutLγensembleprocesses199

recombination intermediates by resolution-independent nicking, in agreement200

withresultsobtainedfromsequencingofheteroduplexDNAarisingduringmeio-201

sisinyeastcells,whichindicatedcleavagebynickingsomedistanceawayfrom202

theDNAjunctionpoints9.203

Interactionswith PCNA are oftenmediated by a PIP-boxmotif.We supple-204

mentedtheMutLγensemblenucleaseassayswithaPIP-boxpeptidederivedfrom205

p2126,oracontrolpeptidewithkeyresiduesmutated.ThecompetingPIP-box206

peptideeliminatedthestimulatoryeffectofRFC-PCNA,whilethecontrolpeptide207

hadnoeffect(Fig.3a),demonstratingthatthePCNAfunctioninstimulatingthe208

nucleaseactivityoftheMutLγensembleisdependentonaninteractionviaaPIP-209

boxlikemotif.WenextanalyzedseveralmutantsofconservedPIP-box-likese-210

quencesinMLH1,MLH3andEXO1(DA)27-29(Fig.3b).Therespectivemutations211

didnotnotablyaffectthenucleasereactionspersewithoutPCNAorthecapacity212

tobindDNA,but themutantsbecamepartly refractory to stimulationbyRFC-213

PCNA, in particularwhen the PIP-box-likemutations ofmultiple factorswere214

combined(Fig.3c,ExtendedDataFig.8a-e).Furthermore,thecorrespondingmu-215

tationsintheyeasthomologuesofMlh1andMlh3(Fig.3b)resultedinmeiotic216

defects,as indicatedbyadecrease inthe frequencyofcrossoversat theCEN8-217

THR1intervalleadingtochromosomenon-disjunctionandreducedsporeviabil-218

ity(Fig.3d,e,ExtendedDataFig.9a).WenotedthatthestabilityofMlh1Pwasre-219

ducedinvivo,whichmaycontributetotheveryseverephenotype,butthelevels220

ofMlh3Pwerecomparabletowildtype(ExtendedDataFig.9b,c).Wealsofound221

8

yeastMlh1andMlh3asapartofacomplexwithRfc1inmeioticcellsatthetime222

ofjointmoleculeresolution(ExtendedDataFig.9d).Finally,usingchromatinim-223

munoprecipitationandsynchronousmeioticyeastcultures,weobservedanen-224

richmentofRfc1atbothnaturalandengineeredDSBhotspotsinlatemeioticpro-225

phaseatthetimewhenjointmoleculesareresolvedintocrossovers(Fig.3f).This226

coincideswiththeaccumulationofMlh3atthesamehotspots30.Theaccumula-227

tionofRfc1atsitesofrecombinationwasindependentofMlh3,suggestingthatit228

mayberetainedfromanearlierstepofDNAsynthesis(ExtendedDataFig.9e).229

WedemonstratedherethatMutSγ,EXO1andRFC-PCNAarerequiredtoacti-230

vatethenucleaseactivityofMutLγunderphysiologicalconditionswithmagne-231

sium,butwefailedtodetectanycanonicalHJresolvaseactivity.Rather,ourdata232

suggestthatthenucleaseensembleprocessesmeioticjointmoleculeintermedi-233

ates by biased resolution-independent nicking of dsDNA in the vicinity ofHJs.234

SinceHJsaresymmetricandtheirresolutioncanyieldbothcrossoversandnon-235

crossovers25, how is the crossover bias established?AsMutSγ likely stabilizes236

asymmetricHJprecursors, itmaybeasymmetricallydistributedat themature237

jointmoleculescontainingHJs(ExtendedDataFig.10).Similarly,PCNAislikely238

loadedasymmetricallyatjointmoleculestofacilitateDNAsynthesisbypolymer-239

ased,oratstranddiscontinuitiesbeforetheligationofdoubleHJstakesplace.We240

propose that the asymmetricpresenceof theMutLγnuclease co-factorsmight241

providethesignaltoguaranteebiased,crossover-specificprocessingofmeiotic242

jointmoleculeintermediates(ExtendedDataFig.10).InthismodeltheMMRre-243

actionthatlimitsmutagenesisduringDNAreplication11isrepurposedinmeiotic244

recombinationto insteadpromotediversityamongtherecombinedDNAmole-245

cules.246

247

248

REFERENCES249

250

1 Keeney, S., Giroux, C. N. & Kleckner, N. Meiosis-specific DNA double-251

strand breaks are catalyzed by Spo11, a member of a widely conserved 252

protein family. Cell 88, 375-384 (1997). 253

2 Hunter, N. Meiotic Recombination: The Essence of Heredity. Cold Spring 254

Harb Perspect Biol 7, doi:10.1101/cshperspect.a016618 (2015). 255

9

3 Nishant, K. T., Plys, A. J. & Alani, E. A mutation in the putative MLH3 256

endonuclease domain confers a defect in both mismatch repair and meiosis in 257

Saccharomyces cerevisiae. Genetics 179, 747-755, 258

doi:10.1534/genetics.108.086645 (2008). 259

4 Zakharyevich, K., Tang, S., Ma, Y. & Hunter, N. Delineation of joint 260

molecule resolution pathways in meiosis identifies a crossover-specific 261

resolvase. Cell 149, 334-347, doi:10.1016/j.cell.2012.03.023 (2012). 262

5 Zakharyevich, K. et al. Temporally and biochemically distinct activities of 263

Exo1 during meiosis: double-strand break resection and resolution of double 264

Holliday junctions. Mol Cell 40, 1001-1015, 265

doi:10.1016/j.molcel.2010.11.032 (2010). 266

6 De Muyt, A. et al. BLM helicase ortholog Sgs1 is a central regulator of 267

meiotic recombination intermediate metabolism. Mol Cell 46, 43-53, 268

doi:10.1016/j.molcel.2012.02.020 (2012). 269

7 Svetlanov, A., Baudat, F., Cohen, P. E. & de Massy, B. Distinct functions of 270

MLH3 at recombination hot spots in the mouse. Genetics 178, 1937-1945, 271

doi:10.1534/genetics.107.084798 (2008). 272

8 Snowden, T., Acharya, S., Butz, C., Berardini, M. & Fishel, R. hMSH4-273

hMSH5 recognizes Holliday Junctions and forms a meiosis-specific sliding 274

clamp that embraces homologous chromosomes. Mol Cell 15, 437-451, 275

doi:10.1016/j.molcel.2004.06.040 (2004). 276

9 Marsolier-Kergoat, M. C., Khan, M. M., Schott, J., Zhu, X. & Llorente, B. 277

Mechanistic View and Genetic Control of DNA Recombination during 278

Meiosis. Mol Cell 70, 9-20 e26, doi:10.1016/j.molcel.2018.02.032 (2018). 279

10 Kadyrov, F. A., Dzantiev, L., Constantin, N. & Modrich, P. Endonucleolytic 280

function of MutLalpha in human mismatch repair. Cell 126, 297-308, 281

doi:10.1016/j.cell.2006.05.039 (2006). 282

11 Ranjha, L., Anand, R. & Cejka, P. The Saccharomyces cerevisiae Mlh1-Mlh3 283

heterodimer is an endonuclease that preferentially binds to Holliday junctions. 284

J Biol Chem 289, 5674-5686, doi:10.1074/jbc.M113.533810 (2014). 285

12 Rogacheva, M. V. et al. Mlh1-Mlh3, a meiotic crossover and DNA mismatch 286

repair factor, is a Msh2-Msh3-stimulated endonuclease. J Biol Chem 289, 287

5664-5673, doi:10.1074/jbc.M113.534644 (2014). 288

13 Kadyrova, L. Y., Gujar, V., Burdett, V., Modrich, P. L. & Kadyrov, F. A. 289

Human MutLgamma, the MLH1-MLH3 heterodimer, is an endonuclease that 290

promotes DNA expansion. Proc Natl Acad Sci U S A 117, 3535-3542, 291

doi:10.1073/pnas.1914718117 (2020). 292

14 Sonntag Brown, M., Lim, E., Chen, C., Nishant, K. T. & Alani, E. Genetic 293

analysis of mlh3 mutations reveals interactions between crossover promoting 294

factors during meiosis in baker's yeast. G3 (Bethesda) 3, 9-22, 295

doi:10.1534/g3.112.004622 (2013). 296

15 Claeys Bouuaert, C. & Keeney, S. Distinct DNA-binding surfaces in the 297

ATPase and linker domains of MutLgamma determine its substrate 298

specificities and exert separable functions in meiotic recombination and 299

mismatch repair. PLoS Genet 13, e1006722, 300

doi:10.1371/journal.pgen.1006722 (2017). 301

16 Nishant, K. T., Chen, C., Shinohara, M., Shinohara, A. & Alani, E. Genetic 302

analysis of baker's yeast Msh4-Msh5 reveals a threshold crossover level for 303

meiotic viability. PLoS Genet 6, doi:10.1371/journal.pgen.1001083 (2010). 304

10

17 Santucci-Darmanin, S. et al. The DNA mismatch-repair MLH3 protein 305

interacts with MSH4 in meiotic cells, supporting a role for this MutL 306

homolog in mammalian meiotic recombination. Hum Mol Genet 11, 1697-307

1706 (2002). 308

18 Manhart, C. M. et al. The mismatch repair and meiotic recombination 309

endonuclease Mlh1-Mlh3 is activated by polymer formation and can cleave 310

DNA substrates in trans. PLoS Biol 15, e2001164, 311

doi:10.1371/journal.pbio.2001164 (2017). 312

19 Kneitz, B. et al. MutS homolog 4 localization to meiotic chromosomes is 313

required for chromosome pairing during meiosis in male and female mice. 314

Genes Dev 14, 1085-1097 (2000). 315

20 Flores-Rozas, H. & Kolodner, R. D. The Saccharomyces cerevisiae MLH3 316

gene functions in MSH3-dependent suppression of frameshift mutations. Proc 317

Natl Acad Sci U S A 95, 12404-12409 (1998). 318

21 Lipkin, S. M. et al. MLH3: a DNA mismatch repair gene associated with 319

mammalian microsatellite instability. Nat Genet 24, 27-35, 320

doi:10.1038/71643 (2000). 321

22 Wu, Y. et al. A role for MLH3 in hereditary nonpolyposis colorectal cancer. 322

Nat Genet 29, 137-138, doi:10.1038/ng1001-137 (2001). 323

23 Dherin, C. et al. Characterization of a highly conserved binding site of Mlh1 324

required for exonuclease I-dependent mismatch repair. Mol Cell Biol 29, 907-325

918, doi:10.1128/MCB.00945-08 (2009). 326

24 Pluciennik, A. et al. PCNA function in the activation and strand direction of 327

MutLalpha endonuclease in mismatch repair. Proc Natl Acad Sci U S A 107, 328

16066-16071, doi:10.1073/pnas.1010662107 (2010). 329

25 Rass, U. et al. Mechanism of Holliday junction resolution by the human 330

GEN1 protein. Genes Dev 24, 1559-1569, doi:10.1101/gad.585310 (2010). 331

26 Bruning, J. B. & Shamoo, Y. Structural and thermodynamic analysis of 332

human PCNA with peptides derived from DNA polymerase-delta p66 subunit 333

and flap endonuclease-1. Structure 12, 2209-2219, 334

doi:10.1016/j.str.2004.09.018 (2004). 335

27 Lee, S. D. & Alani, E. Analysis of interactions between mismatch repair 336

initiation factors and the replication processivity factor PCNA. J Mol Biol 337

355, 175-184, doi:10.1016/j.jmb.2005.10.059 (2006). 338

28 Liberti, S. E. et al. Bi-directional routing of DNA mismatch repair protein 339

human exonuclease 1 to replication foci and DNA double strand breaks. DNA 340

Repair (Amst) 10, 73-86, doi:10.1016/j.dnarep.2010.09.023 (2011). 341

29 Genschel, J. et al. Interaction of proliferating cell nuclear antigen with PMS2 342

is required for MutLalpha activation and function in mismatch repair. Proc 343

Natl Acad Sci U S A 114, 4930-4935, doi:10.1073/pnas.1702561114 (2017). 344

30 Sanchez, A. A., C; Rauh, F; Duroc, Y; Ranjha, L; Lombard, B; Mu, X; Loew, 345

D; Keeney, S; Cejka, P; Guérois R; Klein F; Charbonnier, JB; Borde, V. 346

Mechanism of in vivo activation of the MutLγ-Exo1 complex for meiotic 347

crossover formation. bioRxiv 2019.12.16.876623, 348

doi:https://doi.org/10.1101/2019.12.16.876623 (2019). 349

350

351

11

FIGURELEGENDS352

353

Figure1.MSH4-MSH5andEXO1(DA)interactwithandpromotetheMLH1-354

MLH3endonuclease.a,Top,alignmentofthemetalbindingmotifinMLH3.Con-355

servedresiduesarehighlightedinred.Substitutionmutationsusedinthisstudy356

areinitalics.Bottom,nucleaseassayswithMLH1-MLH3and2.7kbp-longscDNA,357

with5mMmanganeseacetate,withoutorwith0.5mMATP,at37°C.b,Protein358

interaction assays with immobilized MSH4-MSH5 (bait, 220 nM) and MLH1-359

MLH3(prey).The10%polyacrylamidegelwasstainedwithsilver.c,Nuclease360

assayswithMutSγandMutLγ(5mMmanganeseacetate,30°C,0.5mMATP).Av-361

eragesshown;n=3;errorbars,SEM.d,Top,alignmentofMSH5andMSH4ATPase362

domains.Conservedresiduesarehighlightedinred.Alaninesubstitutionsusedin363

thisstudyareinitalics.Bottom,nucleaseassayswithMutSγATPasevariants(5364

mMmanganeseacetate,30°C,0.5mMATP).Averagesshown;errorbars,SEM;365

n=3.e,Top,alignmentofMLH1andMLH3ATPasedomains.Conservedresidues366

arehighlightedinred.Alaninesubstitutionsusedinthisstudyareinitalics.Bot-367

tom,nucleaseassayswithMutLγATPasevariants(5mMmanganeseacetate,30368

°C,0.5mMATP).Averagesshown;errorbars,SEM;n=3.f,Quantitationofkinetic369

nuclease assays with MutSγ and MutLγ, without or with nuclease-deficient370

EXO1(DA)(20nM).Theassayswerecarriedoutat30°Cwith5mMmanganese371

acetateand2mMATP.Averagesshown;errorbars,SEM;n=3.g,Proteininterac-372

tionassayswithimmobilizedMutLγ(bait)andEXO1(prey).The10%polyacryla-373

midegelwasstainedwithsilver.374

375

Figure2.RFC-PCNApromoteDNAcleavagebytheMutLγ-MutSγ-EXO1(DA)376

ensemble.a,Nucleaseassayswith5.6kbp-longscDNAandindicatedhumanpro-377

teinswascarriedoutwith5mMmagnesiumacetateand2mMATPat37°C.A378

representative experiment is shown at the bottom, a quantitation (averages379

shown;n=7;errorbars,SEM)atthetop.b,Representativenucleaseassayscarried380

outwith5mMmagnesiumacetate,5mMmanganeseacetateor5/1mMmagne-381

sium/manganeseacetate,respectively,asindicated,withindicatedproteins,con-382

taining 2 mM ATP and incubated at 37 °C. c, Protein interaction assays with383

12

immobilizedMLH1-MLH3(bait)andPCNA(prey).The10%polyacrylamidegel384

wasstainedwithsilver.d,Nucleasereactionsasinpanela,lane5,buteitherwith-385

outATP,withATPorwithnon-hydrolysableAMP-PNP(2mM).Averagesshown;386

errorbars, SEM;n=4.e,Nuclease reactionswithMLH1-MLH3(50nM),MSH4-387

MSH5(50nM),EXO1(DA)(50nM)andyRFC-hPCNA(50-100nM,respectively),388

lane2.Lanes3-7containinsteadMLH1-MLH3orMSH4-MSH5variants(50nM)389

deficientinATPhydrolysis,asindicated.SeeschemesinFig.1d,eforthespecific390

mutations.Reactionswerecarriedoutwith5mMmagnesiumacetateand2mM391

ATPat37°C.Averagesshown;errorbars,SEM;n=4.f,Representativenuclease392

reactionswithMLH1-MLH3,MSH4-MSH5,EXO1(DA)andyRFC-hPCNA,asindi-393

cated,with3.5kbp-longdsDNAeithercontaining(left)ornot(right)DNArepeat394

formingHJ-likecruciformDNA.Averagesshown;errorbars,SEM;n=9.395

396

Figure3.ThestimulationoftheMLH3nucleaseensemblebyPCNArequires397

aPIPboxmotifandisconservedinevolution.a,NucleaseassayswithMLH1-398

MLH3,MSH4-MSH5,EXO1(DA),andyRFC-hPCNA,asindicated,with5mMmag-399

nesiumacetateand2mMATPat37°C.Thereactionsweresupplementedwitha400

p21PIP-boxwildtypeormutatedcontrolpeptide,whereindicated(670nM,~5-401

foldoverKdofwildtypepeptideforPCNA)26.Averagesshown;errorbars,SEM;402

n=4.b,AlignmentofPIP-box-likemotifsfromvarioushumanorS.cerevisiaepro-403

teins.Residuesmorelikelytobeconservedarehighlightedinred.Wildtypehu-404

manandyeastEXO1,MLH3/Mlh3andMLH1/Mlh1weremutatedtocreatere-405

spective(P)variantswithindicatedresiduesubstitutions(Ainitalics).c,Nuclease406

assays with MLH1-MLH3, MSH4-MSH5, EXO1(DA), and yRFC-hPCNA, as indi-407

cated,with5mMmagnesiumacetateand2mMATPat37°C.Whereindicated,408

wild type MLH1 was replaced with MLH1P (Q562A-I565A-F568A), wild type409

MLH3 with MLH3P (Q341A-V344A-F347A), and EXO1(DA), with EXO1(DA)P410

(D173A-Q788A-L791A).Averagesshown;errorbars,SEM,n=5.d,Recombination411

frequency,expressedasamapdistanceincentimorgans,wasassayedinthewild412

typeS.cerevisiaestrain,mlh1Dandmlh3D,andinstrainscomplementedwithan413

untaggedconstructexpressingwildtypeMlh1,Mlh1P(Q572A-L575A-F578A)or414

Mlh3P(Q293A-V296A-F300A).Allvariantswereexpressedfromtheendogenous415

gene locus. Averages shown; error bars, SD; n≥900 from 3 independent416

13

experiments.e,FrequencyofchromosomeVIIInon-disjunctioninstrainsasde-417

scribedinpaneld.Averagesshown;errorbars,SD;n≥900from3independent418

experiments.f,Rfc1-TAPlevelsattheindicatedmeioticDSBhotspotsrelativetoa419

negativecontrolsite(NFT1)wereassessedbyChIPandqPCRduringameiotic420

time-course (synchronized pCUP1-IME1 cells).Averagesshown;errorbars,SD;421

n=2.ThecartoonillustratesthepositionofsitesanalyzedbyqPCRrelativetothe422

meioticchromosomestructure.423

424

425

14

METHODS426

427

Preparationofexpressionvectors428

To prepare the MSH4-MSH5 expression vector, the MSH4-STREP and MSH5-429

8xHISconstructswerecodon-optimizedforexpressioninSpodopterafrugiperda430

Sf9cellsandsynthesized(GenScript).ThegeneswereamplifiedbyPCRusingM13431

forwardandreverseprimers(seeSupplementaryDataTable1forsequencesof432

alloligonucleotides)anddigestedwithSalIandHindIII(forMSH4)orSmaIand433

KpnI(forMSH5)restrictionendonucleases(NewEnglandBiolabs).Digestedfrag-434

ments were ligated into corresponding sites in pFBDM (Addgene) to obtain435

pFBDM-hMSH4co-STREPandpFBDM-hMSH5co-HIS,respectively.Bothplasmids436

werethendigestedwithBamHIandHindIII(NewEnglandBiolabs)andligatedto437

generate pFBDM-hMSH4co-STREP-hMSH5co-HIS. To prepare expression con-438

structscodingfortheATPasevariantsofhMSH4andhMSH5,therespectivecon-439

servedresiduesintheWalkerAmotifs(seeref16)weremutatedbyQuikChange440

IIsite-directedmutagenesiskit(AgilentTechnologies).ToprepareMSH4G685A,441

the pFB-hMSH4co-STREP-hMSH5co-HIS vector was mutated with primers442

HMSH4G685A_FO and HMSH4G685A_RE. This created pFB-hMSH4coG685A-443

STREP-hMSH5co-HIS. To prepare MSH5G597A, HMSH5G597A_FO and444

HMSH5G597A_RE primers were used to create pFB-hMSH4co-STREP-445

hMSH5coG597A-HIS.Wealsoprepareda construct combiningbothmutations,446

buttheresultingmutantcomplexwasnotstableandcouldnotbepurified.447

TopreparetheMLH1andMLH3expressionvectors,bothgeneswereamplified448

byPCRfrompFL-his-MLH3co-MLH1cocontainingbothMLH1andMLH3genes,449

whichwerecodon-optimizedforinsectcellexpression.ToamplifyMLH1,FLAG-450

hMLH1co_FO and hMLH1co_RE primers were used. The PCR product was di-451

gested withNheI and XbaI (New England Biolabs) and inserted in pFB-MBP-452

MLH3-his11 creatingpFB-FLAG-hMLH1co(thesequenceofMLH3wasremoved453

during this step). Similarly, MLH3 was amplified using MLH3co_FO and454

MLH3co_RE.ThePCRproductwasdigestedwithNheIandXmaI(NewEngland455

Biolabs)andinsertedintopFB-MBP-MLH3-HIS,generatingpFB-MBP-hMLH3co.456

Thesequenceofnon-optimizedMLH3wasremovedduringthisstep.Theconsen-457

sus metal-binding motif in MLH3 is DQHAADE (conserved residues458

15

underlined)3,10.Topreparethenuclease-deadvariant,thesequenceofwildtype459

MLH3inpFB-MBP-MLH3co-HISwasmutatedusingprimersHMLH33ND_FOand460

HMLH33ND_RE. This created a sequence with 3 point mutations including461

D1223N,Q1224KandE1229K(NKHAADK,mutatedresiduesinitalics),andthe462

resulting vector was pFB-MBP-MLH3co3ND-HIS. To disrupt the ATPase of463

hMLH114,thepFB-FLAG-hMLH1cowasmutatedusingprimersHMLH1E34A_FO464

andHMLH1E34A_RE.ThiscreatedpFB-FLAG-MLH1E34A.Tomutatethecorre-465

sponding conserved residue inMLH3, thepFB-HIS-MBP-MLH3cowasmutated466

usingprimersHMLH3E28A_FOandHMLH3E28A_RE.ThiscreatedpFB-HIS-MBP-467

MLH3E28A.TopreparetheMLH1Pvariant,thepFB-FLAG-hMLH1coplasmidwas468

mutatedusingHMLH1_PIP1_3AFOandHMLH1_PIP1_3AREprimers.Toprepare469

the MLH3P variant, the pFB-FLAG-hMLH3co plasmid was mutated using470

HMLH3_PIP_3AFOandHMLH3_PIP_3AREprimers.471

TopreparepFB-hEXO1-FLAG,thesequencecodingforwildtypehumanEXO1472

(orEXO1[DA],containingtheD173Amutationinactivatingitsnuclease)wasam-473

plifiedbyPCRusingprimersHEXO1_FOandHEXO1_RE,andrespectivevectors(a474

kindgiftfromStefanoFerrari,UniversityofZurich)31astemplates.ThePCRprod-475

uctsweredigestedbyBamHIandXmaI(NewEnglandBiolabs),andclonedinto476

correspondingsitesinpFB-MBP-Sae2-HIS32(thesequenceofMBP-Sae2wasre-477

movedduringtheprocess,FLAG-tagwasaddedtotheC-terminusandaHIS-tag478

fromtheoriginalconstructwasnottranslatedduetoaStopcodon).479

TopreparepFB-hMSH2-FLAG, thesequencecoding forMSH2wasamplified480

from pFB-hMSH233 using primers HMSH2FLAG_FO and HMSH2FLAG_RE. The481

PCRproductwasdigestedbyBamHIandXhoI(NewEnglandBiolabs),andcloned482

into corresponding sites inpFB-MBP-Sae2 (the sequenceofMBP-Sae2was re-483

movedduringtheprocess,FLAG-tagwasaddedtotheC-terminusofMSH2anda484

HIS-tagfromtheoriginalconstructwasnottranslatedduetoaStopcodon).485

To prepare pFB-HIS-yMLH1, pFB-GST-yMLH111 was digested using BamHI486

(NewEnglandBiolabs)toremovetheGSTtag.Thisprocedureleftbehindasingle487

BamHIsite.TwocomplementaryoligonucleotidesHis-ForandHis-Revwerean-488

nealedtoeachother,andclonedintotheBamHIsite.Thisintroducedasequence489

codingfor8xHIStagbeforetheyeastMLH1genecreatingpFB-HIS-yMLH1.490

16

TopreparepFB-MBP-yMLH3,a terminationcodonwas introducedafter the491

yMLH3gene inpFB-MBP-yMLH3-HIS11sothattheHIS-tagwouldnotbetrans-492

latedwiththeyMlh3protein.Thiswascarriedoutbysite-directedmutagenesis493

usingforwardprimer329andreverseprimer330.494

TopreparetheexpressionvectorforyeastMsh4,theyeastMSH4genewasam-495

plifiedfromthegenomicDNAoftheS.cerevisiaeSK1strainusingforwardprimer496

258andreverseprimer259b.Thereverseprimerintroducedthesequencefor497

the C-terminal STREP affinity tag. The amplified product was digested with498

BamHIandHindIII(NewEnglandBiolabs)andclonedintocorrespondingsitesof499

pFB-GST-MLH111tocreatepFB-yMSH4-STREP.TheyeastMSH5genewasampli-500

fiedfromthegenomicDNAoftheS.cerevisiaeW303strainusingforwardprimer501

265andreverseprimer266.TheMSH5genewas thencloned intoBamHIand502

XhoIrestrictionsitesofpFB-MBP-MLH3-HIS11tocreatepFB-yMSH5-HIS.503

504

PurificationofhumanMLH1-MLH3505

The bacmids and baculoviruses were prepared individually using pFB-FLAG-506

hMLH1co andpFB-HIS-MBP-hMLH3co vectors according tomanufacturer’s in-507

structions(Bac-to-Bacsystem,LifeTechnologies).Spodopterafrugiperda9(Sf9)508

cellswereseededat500,000cellsperml16hbefore infection.Thecellswere509

thenco-infectedwithbothbaculovirusesand incubated for52hat27 °Cwith510

constantagitation.Thecellswerethenharvested(500xg,10min)andwashed511

oncewithPBS(137mMNaCl,2.7mMKCl,10mMNa2HPO4,1.8mMKH2PO4).The512

pelletsweresnap-frozeninliquidnitrogenandstoredat-80°C.Allsubsequent513

stepswerecarriedoutoniceorat4°C.Thepelletswereresuspendedin3vol-514

umesof lysisbuffer[50mMTris-HClpH7.5,1mMdithiothreitol(DTT),1mM515

ethylenediaminetetraacetic acid (EDTA), 1 mM phenylmethylsulfonyl fluoride516

(PMSF),1:400(volume/volume)proteaseinhibitorcocktail(Sigma,P8340),30517

μg/ml leupeptin (Merck)] and incubated for 20min with continuous stirring.518

Next,1/2volumeof50%glycerolwasadded,followedby6.5%volumeof5M519

NaCl(finalconcentration305mM).Thesuspensionwasfurtherincubatedfor30520

minwithcontinuousstirring.Thecellsuspensionwascentrifugedfor30minat521

48,000xg toobtainsolubleextract.Thesupernatantwas transferred to tubes522

containingpre-equilibratedAmyloseresin(NewEnglandBiolabs,4mlper1lof523

17

Sf9culture)andincubatedfor1hwithcontinuousagitation.Theresinwascol-524

lectedbyspinningat2,000xgfor2minandwashedextensivelybatchwiseand525

onadisposablecolumn(10ml,ThermoFisher)withAmylosewashbuffer [50526

mMTris-HClpH7.5,1mMβ-mercaptoethanol(β-ME),1mMPMSF,10%glycerol,527

300mMNaCl].ProteinwaselutedwithAmyloseelutionbuffer[50mMTris-HCl528

pH 7.5, 0.5 β-ME, 1 mM PMSF, 10% glycerol, 300 mM NaCl, 10 mMmaltose529

(Sigma)]andthetotalproteinconcentrationwasestimatedbyBradfordassay.To530

cleaveoffthemaltosebindingtag(MBP),1/6(weight/weight)ofPreScissionpro-531

tease(PP)34,withrespecttototalproteinconcentrationintheeluate,wasadded532

andincubatedfor1h.Next,thecleavedamyloseeluatewasdilutedbyadding1/2533

volumeofFLAGdilutionbuffer(50mMTris-HClpH7.5,1mMPMSF,10%glyc-534

erol,300mMNaCl)tolowertheconcentrationofβ-ME.Thedilutedeluatewas535

then incubated batchwise for 1 hwith pre-equilibrated anti-FLAGM2 affinity536

resin(Sigma,A2220,0.8ml).TheresinwaswashedextensivelywithFLAGwash537

buffer(50mMTris-HClpH7.5,0.5mMβ-ME,1mMPMSF,10%glycerol,150mM538

NaCl).ProteinwaselutedwithFLAGwashbuffercontaining150ng/μl3xFLAG539

peptide(Sigma),aliquoted,frozeninliquidnitrogenandstoredat-80°C.Thefinal540

constructcontainedaFLAGtagattheN-terminusofMLH1.Theyieldfrom1lcul-541

turewas~0.5mgandtheconcentration~2μM.AllMLH1-MLH3mutantswere542

expressedandpurifiedusingthesameprocedure.543

544

PurificationofhumanMSH4-MSH5545

The human MSH4-MSH5 complex was expressed from a dual pFB-hMSH4co-546

STREP-hMSH5co-HISvectorinSf9cellsusingtheBac-to-Bacsystemasdescribed547

above.Allpurificationstepswerecarriedouton iceorat4°C.Thecellpellets548

wereresuspendedin3volumesofnickel-nitriloaceticacid(NiNTA)lysisbuffer549

[50mMTris-HClpH7.5,2mMβ-ME,1mMPMSF,1:400(volume/volume)prote-550

aseinhibitorcocktail(Sigma,P8340),30μg/mlleupeptin(Merck),20mMimid-551

azole]andincubatedfor20minwithcontinuousstirring.Next1/2volumeof50%552

glycerolwasadded, followedby6.5%volumeof5MNaCl(finalconcentration553

305mM),andthesuspensionwasfurtherincubatedfor30minwithcontinuous554

stirring.Toobtainsolubleextract,thesuspensionwascentrifugedat48,000xg555

for 30 min. The soluble extract was transferred to a tube containing pre-556

18

equilibratedNiNTAresin(Qiagen,4mlper1lSf9cells)andincubatedfor1hwith557

continuousmixing.TheNiNTAresinwascollectedbycentrifugationat2,000xg558

for2min.Theresinwaswashedextensivelybatchwiseandonadisposablecol-559

umnwithNiNTAwashbuffer(50mMTris-HClpH7.5,2mMβ-ME,300mMNaCl,560

1mMPMSF,10%glycerol,20mM imidazole).ProteinwaselutedwithNiNTA561

washbuffercontaining250mMimidazole.Theelutedsamplewasincubatedwith562

pre-equilibratedStrep-TactinSuperflowresin(Qiagen,0.7ml) for90minwith563

continuousmixing.Theresinwastransferredtoadisposablecolumnandwashed564

extensivelywithStrepwashbuffer(50mMTris-HClpH7.5,2mMβ-ME,300mM565

NaCl,1mMPMSF,10%glycerol).ProteinwaselutedwithStrepwashbuffercon-566

taining2.5mMd-Desthiobiotin(Sigma,D1411)andstoredat-80°Caftersnap567

freezing in liquidnitrogen.The finalconstructcontainedaSTREPtagat theC-568

terminusofMSH4andaHIS-tagattheC-terminusofMSH5.Thevariantsofthe569

MSH4-MSH5complexwerepurifiedusingthesameprocedure.Wenotethatthe570

doublemutantMSH4G685A-MSH5G597Aheterodimerwasnotstableandcould571

notbepurified.572

573

PurificationofhumanEXO1(DA)574

ThepFB-EXO1(D173A)-FLAGvectorwasusedtopreparerecombinantbaculovi-575

rusandtheproteinwasexpressedinSf9cellsasdescribedabove.Frozencellpel-576

letwasthawedandresuspendedin3pelletvolumesoflysisbuffer[50mMTris-577

HClpH7.5,0.5mMβ-ME,1mMEDTA,1:400(volume/volume)proteaseinhibitor578

cocktail(Sigma,P8340),0.5mMPMSF,20µg/mlleupeptin].Thecellsuspension579

wasincubatedwithgentlestirringfor10min.1/2volumeof50%glyceroland580

6.5%volumeof5MNaCl(finalconcentration305mM)wereadded.Thesuspen-581

sionwasincubatedfor30minwithstirring.Theextractwasthencentrifugedat582

48,000xgfor30min.Thesolubleextractwasaddedtopre-equilibratedM2anti583

FLAGaffinityresin(Sigma,A2220,2mlresinforpurificationfrom1lSf9cellcul-584

ture)andincubatedbatchwisefor45min.Thesuspensionwasthencentrifuged585

(2,000xg,5min),thesupernatant(FLAGflowthrough)removed,andtheresin586

wastransferredtoadisposablechromatographycolumn.Theresinwaswashed587

with50resinvolumesofTBSbuffer(20mMTris-HClpH7.5,150mMNaCl,0.5588

mMβ-ME,0.5mMPMSF,10%glycerol)supplementedwith0.1%NP40.Thiswas589

19

followedbywashingwith10resinvolumesofTBSbufferwithoutNP40.EXO1-590

FLAGwaselutedwithTBSbuffersupplementedwith150ng/μl3xFLAGpeptide591

(Sigma, F4799). Fractions containing detectable protein (as estimated by the592

Bradfordmethod)werepooled,appliedonadisposablecolumnwith1mlpre-593

equilibratedBiorex70resin(Bio-Rad),andflow-throughwascollected.Thesam-594

plewasthendilutedbyadding1volumeofdilutionbuffer(50mMTris-HClpH595

7.5,5mMβ-ME,0.5mMPMSF,10%glycerol).DilutedFLAG-EXO1wasappliedon596

1mlHiTrapSPHPcolumn(GEHealthcare)pre-equilibratedwithSbufferA(50597

mMTris-HClpH7.5,75mMNaCl,5mMb-ME,10%glycerol)at0.8ml/min.The598

columnwaswashedwith20mlSbufferA,andelutedwith8mllinearsaltgradi-599

entinSbufferA(75mMto1MNaCl).Peakfractionswerepooled,aliquoted,fro-600

zeninliquidnitrogenandstoredat-80°C.Theprocedureyieldedaround~0.15601

mgofproteinfrom1lofSf9culture,withanapproximateconcentrationof~1μM.602

603

PurificationofhumanMSH2-MSH6andMSH2-MSH3heterodimers604

TopreparetheMSH2-MSH6heterodimer,theSf9cellswereco-infectedwithre-605

combinant baculoviruses prepared from pFB-hMSH2-FLAG and pFB-hMSH6-606

HIS33vectors.Thepurificationwascarriedoutat4°Coronice.Thecellpellets607

wereresuspendedin3volumesoflysisbuffer[50mMTris-HClpH7.5,1:400[vol-608

ume/volume]proteaseinhibitorcocktail(Sigma,P8340),1mMPMSF,60μg/ml609

leupeptin,0.5mMβ-ME,20mMimidazole].Thesamplewasincubatedwhilestir-610

ringfor20min.1/2volumeof50%glycerolwasadded,followedby6.5%volume611

5MNaCl(finalconcentration305mM).Thecellsuspensionwasincubatedfor30612

minwithstirring.Toobtainsolubleextract,thesuspensionwascentrifuged(30613

min,48,000xg).Thesupernatantwasmixedwithpre-equilibrated2mlNiNTA614

resin(purificationfrom800mlSf9cells)andincubatedbatchwisefor1h.The615

resinwasthenwashedbatchwiseandoncolumnwithwashbuffer[30mMTris-616

HClpH7.5,1:1,000(volume/volume)proteaseinhibitorcocktail(Sigma,P8340),617

15μg/mlleupeptin,0.5mMβ-ME,0.5mMPMSF,20mMimidazole,300mMNaCl,618

10%glycerol].Boundproteinwaselutedwithelutionbuffer[30mMTris-HClpH619

7.5, 1:1,000 (volume/volume) protease inhibitor cocktail (Sigma, P8340), 15620

μg/mlleupeptin,0.5mMβ-ME,0.5mMPMSF,300mMimidazole,150mMNaCl,621

20

10%glycerol].Thepooledfractionsweredilutedwith7volumesofdilutionbuffer622

(30mMTris-HClpH7.5,15μg/mlleupeptin,0.5mMβ-ME,0.5mMPMSF,150623

mMNaCl,10%glycerol),andmixedwith0.7mlpre-equilibratedanti-FLAGM2624

affinity gel (Sigma). The suspensionwas incubated batchwise for 60min. The625

samplewascentrifuged(5min,1,000g)andresinwastransferredtoadisposable626

chromatographycolumn.The resinwas thenwashedextensivelywithdilution627

buffer.Theheterodimerwaselutedwithdilutionbuffersupplementedwith200628

μg/ml 3x FLAG peptide (Sigma). Eluates containing protein were pooled, ali-629

quoted,frozeninliquidnitrogenandstoredat-80°C.TheMSH2-MSH3heterodi-630

merwaspreparedusingthesameprocedure,usingpFB-hMSH3-HIS35.631

632

PurificationofyeastMsh4-Msh5633

BaculovirusesexpressingMsh4andMsh5werepreparedindividuallyusingthe634

Bac-to-BacsystemandpFB-yMSH4-STREPandpFB-yMSH5-HISvectors.Sf9cells635

wereco-infectedwithoptimizedratiosofbothvirusestoexpressbothproteins636

togetherasaheterodimer.Thecellswereharvested52hafterinfection,washed637

withPBS,andthepelletswerefrozeninliquidnitrogenandstoredat-80°Cuntil638

use.Thesubsequentstepswerecarriedoutoniceorat4°C.Thecellpelletwas639

resuspended in lysisbuffer [50mMTris-HClpH7.5,2mMβ-ME,1mMEDTA,640

1:400(volume/volume)proteaseinhibitorcocktail(Sigma,P8340),1mMPMSF,641

30μg/mlleupeptin,20mMimidazole]for20min.Then,50%glycerolwasadded642

toafinalconcentrationof16%,followedby5MNaCltoafinalconcentrationof643

305mM.Thesuspensionwasincubatedforfurther30minwithgentleagitation.644

Thetotalcellextractwascentrifugedat48,000xgfor30mintoobtainsoluble645

extract.TheextractwasthenboundtoNiNTAresin(Qiagen)for60minbatchwise646

followedbyextensivewashingwithNiNTAwashbuffer(50mMTris-HClpH7.5,647

2mMβ-ME,300mMNaCl,10%glycerol,1mMPMSF,10μg/mlleupeptin,20mM648

imidazole) both batchwise and on a column. The heterodimer was eluted by649

NiNTAelutionbuffer(NiNTAwashbuffercontaining250mMimidazole).Theelu-650

atewasfurtherincubatedwithpre-equilibratedStrep-TactinSuperflowresin(Qi-651

agen)for60minbatchwise.Theprotein-boundresinwasthenwashedintwose-652

quentialsteps;firstwithSTREPwashbufferI(50mMTrispH7.5,2mMβ-ME,10653

%glycerol,1mMPMSFand300mMNaCl)andthenwithSTREPwashbufferII654

21

(50mMTrispH7.5,2mMβ-ME,10%glycerol,1mMPMSFand50mMNaCl).655

TheheterodimerwaselutedwithSTREPwashbufferIIcontaining2.5mMd-Des-656

thiobiotin(Sigma).Theeluatewasthenappliedonapre-equilibratedHiTrapQ657

HPcolumn(GEHealthcare).ThecolumnwaswashedwithSTREPwashbufferII658

andproteinwaselutedwithalineargradientofNaCl(50to600mM)inSTREP659

wash buffer II. Collected fractionswere analyzed on SDS-PAGE, peak samples660

were pooled, aliquoted and stored at -80 °C. The final construct contained a661

STREPtagattheC-terminusofyMsh4andaHIS-tagattheC-terminusofyMsh5.662

Theprocedureyielded~0.15mgofproteinfrom4lofSf9culture,withanap-663

proximateconcentrationof~1μM.664

665

PurificationofyeastMlh1-Mlh3666

TheyMlh1-yMlh3heterodimerwasexpressedusingpFB-HIS-yMLH1andpFB-667

MBP-yMLH3and theBac-to-Bac systemandpurifiedusingaffinity chromatog-668

raphy11.Briefly,thecellswereresuspendedinlysisbuffercontaining50mMTris-669

HClpH7.5,1mMDTT,1mMEDTA,1:400(volume/volume),proteaseinhibitor670

cocktail(Sigma,P8340),1mMPMSF,30μg/mlleupeptinandincubatedfor20671

min.Subsequentlyglycerol[finalconcentration16%(volume/volume)]andNaCl672

(finalconcentration305mM)wereadded.Uponfurther incubationfor30min673

andcentrifugation(48,000xg,30min),theclearedextractwasthensubjectedto674

affinitychromatographywithAmyloseresin(NewEnglandBiolabs),theMBPtag675

wascleavedwithPreScissionproteaseandtheheterodimerwasfurtherpurified676

onNi-NTAagarose(Qiagen)11.Thefinaleluatewasdialyzedinto50mMTris-HCl677

pH7.5,5mMβ-ME,10%glycerol,0.5mMPMSFand300mMNaCl.Aliquotswere678

flashfrozenandstoredat-80°Cuntiluse.Thepurificationyielded~1mgprotein679

from2.4lcultureandtheconcentrationwas5.9μM.680

681

PurificationofyeastandhumanRFC,PCNAandtheKuheterodimer682

HumanPCNAwasexpressedinE.colicells(1l)frompET23C-his-hPCNAvector683

(akindgiftfromUlrichHuebscher,UniversityofZurich).Transformedcellswere684

grown to OD 0.5, and induced with 0.5 mM isopropyl β-D-1-thiogalactopyra-685

noside(IPTG)for3.5hat37°C.Cellswerelysedbysonicationinlysisbuffer(20686

22

mMTris-HClpH7.5,250mMNaCl,2mMβ-ME,5mMimidazole,1mMPMSF,687

1:250SigmaproteaseinhibitorcocktailP8340).Thelysatewasclearedbycen-688

trifugation(48,000xg,30min)andboundto2mlNiNTAresin(Qiagen)for1h689

batchwise.Resinwaswashedwithwashbuffer(20mMTris-HClpH7.5,250mM690

NaCl,2mMβ-ME,30mMimidazole,1mMPMSF),andPCNAwaselutedwithelu-691

tionbuffer(washbuffersupplementedwith400mMimidazole).Thesamplewas692

dilutedtoconductivitycorrespondingto100mMNaCl,andloadedonHiTrapQ693

column.Thecolumnwasdevelopedbyasaltgradient(100mMto1MNaCl)in20694

mMTris-HClpH7.5,2mMβ-MEand10%glycerol.ThefractionscontainingPCNA695

werepooled, aliquoted and stored at -80 °C. Yeast PCNAwasprepared asde-696

scribed36.697

ToexpresshumanRFC,theSf9cells(1.4l)wereinfectedwithrecombinant698

baculovirus prepared with vector pFBDM-MBP-RFC1-RFC2-3-4-His-5 (a kind699

giftfromJosefJiricny,ETHZurich,RFC1subunitMBP-taggedandRFC5subunit700

his-tagged).Thepurificationwascarriedoutat4°Coron ice.Thecellpellets701

wereresuspendedin3volumeslysisbuffer[50mMTris-HClpH7.5,2mMβ-ME,702

1:300(volume/volume)proteaseinhibitorcocktail(Sigma,P8340),1mMPMSF,703

30μg/mlleupeptin,15mMimidazole).Thecellswereletswellingonicefor20704

minandmixedoccasionally.Afterwards,1/2volumeof50%glycerolwasadded,705

followedby6.5%volume5MNaCl(finalconcentration305mM),andthesus-706

pensionwasincubatedwhilestirringfor30min.Thecellsuspensionwascentri-707

fugedat55,000xg for30mintoobtainsolubleextract.Thesupernatantwas708

mixedwithpre-equilibrated2mlNiNTAresin(Qiagen)andbatchincubatedwith709

gentleagitationfor1h.Theresinwaswashed3timesbatchwise,andwith15710

resinvolumesoncolumnwithwashbufferI(50mMTris-HClpH7.5,2mMb-ME,711

0.25MNaCl,10%Glycerol,1mMPMSF,20mMimidazole).TheRFCcomplexwas712

elutedwithNiNTAelutionbuffer(50mMTris-HClpH7.5,2mMb-ME,250mM713

NaCl,10%glycerol,1mMPMSF,300mMimidazole).Theeluatewasdirectlyap-714

plied in flowon1.5mlamyloseresin(NewEnglandBiolabs)pre-equilibrated715

withwashbuffer2(50mMTrispH7.5,2mMb-ME,0.1MNaCl,10%Glycerol,1716

mMPMSF),andwashedwithwashbuffer2.TheRFCcomplexwaselutedwith717

washbuffer2supplementedwith10mMmaltose.Theproteinconcentrationwas718

estimated by the Bradfordmethod, and the samplewas incubatedwith 20%719

23

(w/w) Prescission protease for 2 h at 4 °C. The samplewas then applied on720

HiTrapQcolumn(0.5ml/min),pre-equilibratedinQbufferA(50mMTrispH721

7.5,5mMb-ME,0.1MNaCl,10%Glycerol),thecolumnwasthenwashedwithQ722

bufferA,andelutedinthesamebufferwithasaltgradient(0.1Mto1M)in6723

columnvolumes.ThefractionscontainingRFCwerepooled,aliquoted,frozenin724

liquidnitrogenandstoredat-80°C.Thepurificationyielded~1mlof8.5μM725

RFC.726

YeastRFCwasexpressedinE.colicells(4l)transformedwithpEAO271(akind727

giftfromE.Alani,CornellUniversity).CellsweregrowntoOD0.5,andinduced728

with0.5mMIPTGfor3hat37°C.Cellswereresuspendedinlysisbuffer(60mM729

HEPES-NaOHpH7.5,250mMNaCl,2mMβ-ME,0.5mMEDTA,1:250Sigmapro-730

teaseinhibitorcocktailP8340,1mMPMSF,10%glycerol)anddisruptedbyson-731

ication.Theclearedextractwasloadedon5mlHiTrapSPcolumn,washedwith732

bufferSPA(30mMHEPES-NaOHpH7.5,300mMNaCl,2mMβ-ME,0.5mMEDTA,733

1mMPMSF,10%glycerol)andelutedwithasaltgradient(300mMto600mM734

NaCl). Eluted fractions were analyzed by polyacrylamide gel electrophoresis,735

pooledanddilutedtoconductivitycorrespondingto110mMNaCl.Thediluted736

samplewasappliedonHiTrapQcolumn,andelutedin110to600mMNaClgra-737

dientin30mMHEPES-NaOHpH7.5,2mMβ-ME,1mMPMSFand10%glycerol.738

Theeluatewasaliquotedandstoredat-80°C.ThepreparationoftheyeastKu739

heterodimerwasdescribedpreviously37.740

741

Nucleaseassays742

Thereactions(15µl)werecarriedoutin25mMTris-acetatepH7.5,1mMDTT,743

0.1mg/mlbovineserumalbumin(BSA,NewEnglandBiolabs),andasindicated744

manganeseormagnesiumacetate(5mM),ATP(concentrationsasindicated,GE745

Healthcare,27-1006-01)andplasmid-basedDNAsubstrate[100ngperreaction,746

either2.7kbp-longpUC19(Fig.1andrelatedExtendedDataFigures),5.6kbp-747

longpFB-RPA2(Figures2and3,andrelatedExtendedDataFigures),10.3kbp-748

longpFB-HIS-MBP-hMLH3co(ExtendedDataFig.4j),orpAG25(Addgene)orcru-749

ciformpIRbke8mut25(Fig.2f,ExtendedDataFig.4k)].Inexperimentswith32P-750

labeled oligonucleotide-based DNA the substrate concentration was 1 nM, in751

24

molecules.Whereindicated,ADP(AlfaAesar,J60672),AMP-PNP(TorontoChem-752

ical,A634303)orATP-γ-S(Cayman,14957)wereusedinsteadofATP.Wherein-753

dicated,thereactionsweresupplementedwithPIPboxpeptidederivedfromp21754

(GRKRRQTSMTDFYHSKRRLIFS) or control peptidewith key residuesmutated755

(underlined,GRKRRATSATDFYHSKRRLIFS)(Genecust).Thereactionbufferwas756

assembledonice,andtherecombinantproteinswerethenaddedonice(MLH1-757

MLH3proteinwasalwaysaddedlast).Thereactions,unlessindicatedotherwise,758

wereincubatedfor60minat30°Cor37°C.Thereactionsweresupplemented759

withproteinstorageordilutionbuffertocompensateforcomponentsintroduced760

with recombinantproteins ineachparticularexperiment, this resulted in final761

NaClconcentrations~30mM.Thereactionswereterminatedwith5µlSTOPso-762

lution(150mMEDTA,2%SDS,30%glycerol,0.01%bromophenolblue),1µlpro-763

teinaseK(Roche,03115828001,18mg/ml)andfurtherincubatedfor60minat764

50°C.Thereactionproductswerethenseparatedbyelectrophoresisin1%aga-765

rose(Sigma,A9539)containingGelRed(Biotium)inTAEbuffer.UsingBio-Rad766

SubCellGTsystem(gellength26cm),theseparationwascarriedoutfor90min767

at120V.Thegelswerethenimaged(InGenius3,GeneSys).Theresultswerequan-768

titatedusingImageJandexpressedas%ofnickedDNAversusthetotalDNAin769

eachparticular lane;anynickedDNApresent incontrol (noprotein)reactions770

wasremovedasabackground.Forgelsourcedata,seeSupplementaryFigure1.771

772

Electrophoreticmobilityshiftassays773

TheDNAbindingreactionswerecarriedout in15μlvolume inbindingbuffer774

containing25mMHEPESpH7.8,5mMmagnesiumchloride,5%(volume/vol-775

ume)glycerol,1mMDTT,50μg/mlBSA,6.6ng/μldsDNA(inassayswithyeast776

proteins)or3.3ng/μldsDNA(inassayswithhumanproteins)ascompetitor(50777

bp-long),0.5nMDNAsubstrate(32P-labelled,inmolecules)andrespectivecon-778

centrationsofrecombinantproteins(yeastorhumanMSH4-MSH5complexand779

theirvariants,MLH1-MLH3andvariants).Theoligonucleotide-basedDNAsub-780

stratesweressDNA(labelledoligonucleotidePC1253),dsDNA(labelledPC1253781

andPC1253C),Y-structure(labelledPC1254andPC1253),HJ(labelledPC1253782

andPC1254,PC1255andPC1256)andD-Loop(labelledBB,andBT, INVaand783

INVb).MgCl2wasreplacedby3mMEDTAwhereindicated.Thereactionswere784

25

assembledandincubatedonicefor15min,followedbytheadditionof5μlEMSA785

loadingdye(50%glycerol,0.01%bromophenolblue).Theproductsweresepa-786

ratedon6%nativepolyacrylamidegel(19:1acrylamide-bisacrylamide,BioRad)787

onice.Thegelsweredriedon17CHRpaper(Whatman),exposedtostoragephos-788

phor screens (GEHealthcare), and scannedbyPhosphorimager (TyphoonFLA789

9500,GEHealthcare).ThequantitationwascarriedoutbyImageQuantsoftware790

(GE Healthcare) and graphs were plotted using Prism software (Prism 8,791

Graphpad).Forgelsourcedata,seeSupplementaryFigure1.792

Forthe"super-shift"assayscomprisingyMlh1-yMlh3andyMsh4-yMsh5,there-793

actionswerecarriedoutasmentionedabove(withmagnesiumorEDTA,asindi-794

cated),exceptthattheproductswereseparatedon0.6%agarosegelinTAEbuffer795

at4°C(1h,100V).ThegelsweredriedonDE81paper(Whatman)andscanned796

asabove.Inthesuper-shiftassayswithMLH1-MLH3,MSH4-MSH5andEXO1,the797

reactionbufferadditionallycontained75mMNaCland10μMATP.TheDNAbind-798

ingassayswithyKu70-80werecarriedoutsimilarly,withoutsaltandATP,and799

wereincubatedfor30minat30°C.800

801

Proteininteractionassays802

Totestforprotein-proteininteractions,recombinant“bait”proteinwasimmobi-803

lizedonbeadscoupledtoaspecificantibodyandincubatedwiththe“prey”pro-804

tein.Afterremovalofunboundproteinbybeadswashing,proteinswereeither805

detectedbysilverstainingorbywesternblot.Forgelsourcedata,seeSupplemen-806

taryFigure1.807

TotestfortheinteractionbetweenMLH1-MLH3andMSH4-MSH5,0.7µganti-808

MLH1 antibody (Abcam, ab92312)was captured on 15µl Protein Gmagnetic809

beads (Dynabeads, Invitrogen) by incubating in 50µl PBS-T (PBSwith 0.02%810

Tween-20) for60minwithgentlemixingat regular intervals.Thebeadswere811

washed3timesonmagneticrackswith150µlPBS-Ttoremoveunboundanti-812

bodies.Thebeadswerethenmixedwith165nMrecombinantMLH1-MLH3and813

220nMMSH4-MSH5in50µlbindingbufferI(25mMHEPESpH7.8,3mMEDTA,814

1mMDTT,50µg/mlBSA,80mMNaCl)and incubatedon ice for45minwith815

gentleagitationatregularintervals.Beadswerethenwashed3timeswith150µl816

26

washbufferI(25mMHEPESpH7.8,3mMEDTA,1mMDTT,0.02%Tween-20,817

80mMNaCl)andproteinswereelutedbyboilingthebeadsinSDSbuffer(50mM818

Tris-HCl pH 6.8, 1.6% sodium dodecyl sulphate, 100 mM DTT, 10% glycerol,819

0.01%bromophenolblue)for3minat95°C.Theeluatewasseparatedona10%820

SDS-PAGEgelandproteinsweredetectedbysilverstaining.Toperformtheex-821

perimentreciprocally,5µganti-HISantibody(Genscript,A00186)wascaptured822

onProteinGbeads(Dynabeads,Invitrogen)asdescribedabove.Therecombinant823

protein complexes, as above,were then addedand incubated in50µl binding824

bufferII(25mMHEPESpH7.8,3mMEDTA,1mMDTT,50µg/mlBSA,80mM825

NaCl) for 45 min with gentle agitation at regular intervals. Beads were then826

washed3timeswithwashbufferII(25mMHEPESpH7.8,3mMEDTA,1mM827

DTT,80mMNaCl,0.1%TritonX-100).Thesubsequentstepswerecarriedoutas828

describedabove.Totestforspecies-specificinteractionsasshowninExtended829

DataFig.3g,thesameprocedurewasfollowedexcept100nMofeitherhuman830

MSH4-MSH5oryeastMsh4-Msh5wasincubatedwith400nMMLH1-MLH3.To831

testfortheinteractionbetweenyeastMlh1-Mlh3andMsh4-Msh5,10µlProtein832

G beads were used to capture 1 µg anti-STREP antibody (Biorad, MCA2489).833

yMsh4-yMsh5(120nM)wasincubatedwiththebeadsin60µlbindingbufferIII834

(25mMTris-HClpH7.5,3mMEDTA,1mMDTT,20mg/mlBSA,60mMNaCl)for835

60minwithcontinuousmixing.Next,thebeadswerewashed3timeswith150µl836

washbufferIII(25mMTris-HClpH7.5,3mMEDTA,1mMDTT,120mMNaCl,837

0.05%TritonX-100).300nMyMlh1-yMlh3wasthenaddedtotheresuspended838

beadsin60µlbindingbufferIII,andincubatedforadditional60minwithcontin-839

uousmixing.Beadswerewashed3timeswith150µlwashbufferIIIandboiled840

afterwardsfor3minat95°CinSDSbuffertoelutetheproteins.Theproteincom-841

plexesweredetectedbywesternblotwithanti-HISantibody(Genscript,A00186).842

TotestfortheinteractionbetweenMLH1-MLH3andEXO1,0.33µganti-MLH1843

antibody (Abcamab223844)was capturedon10µl proteinGmagneticbeads844

(Dynabeads,Invitrogen)byincubatingin50µlPBS-T(PBSwith0.1%Tween-20)845

for2hat4°Cwithgentlemixingatregularintervals.Thebeadswerewashed4846

timesonmagnetic rackswith150µlPBS-T to removeunboundantibody.The847

beadswerethenmixedwith1µgrecombinantMLH1-MLH3and0.5µgEXO1in848

27

200µlbindingbufferI(25mMTris-HClpH7.5,3mMEDTA,1mMDTT,20µg/ml849

BSA,300mMNaCl)andincubatedonicefor2hwithgentleagitationatregular850

intervals.Beadswerethenwashed4timeswith300µlwashbufferI(50mMTris-851

HClpH7.5,3mMEDTA,1mMDTT,300mMNaCl,0.05%TritonX-100)andpro-852

teinswereelutedbyboilingthebeadsinSDSbuffer(50mMTris-HClpH6.8,1.6%853

sodiumdodecylsulphate,100mMDTT,10%glycerol,0.01%bromophenolblue)854

for3minat95°C.Theeluatewasseparatedona10%SDS-PAGEgelandproteins855

weredetectedbysilverstaining.856

TotestfortheinteractionbetweenhumanMLH1-MLH3andhumanPCNAor857

EXO1,1µganti-MLH1antibody(Abcamab223844)wascapturedon15µlprotein858

Gmagneticbeads(Dynabeads,Invitrogen)byincubatingin50µlPBS-T(PBSwith859

0.1%Tween-20)for1hatroomtemperaturewithgentlemixingatregularinter-860

vals.Thebeadswerewashed3timesonmagneticrackswith150µlPBS-Ttore-861

moveunboundantibody.Thebeadswerethenmixedwith1.5µgeachrecombi-862

nantMLH1-MLH3andPCNAorEXO1,in60µlbindingbufferI(25mMTris-HCl863

pH7.5,3mMEDTA,1mMDTT,20µg/mlBSA,60mMNaCl)andincubatedonice864

for1hwithgentleagitationatregularintervals.Beadswerethenwashed4times865

with150µlwashbufferI(50mMTris-HClpH7.5,3mMEDTA,1mMDTT,120866

mMNaCl,0.05%TritonX-100)andproteinswereelutedbyboilingthebeadsin867

SDSbuffer(50mMTris-HClpH6.8,1.6%sodiumdodecylsulphate,100mMDTT,868

10%glycerol,0.01%bromophenolblue)for3minat95°C.Avidin(Sigma,A9275,869

110ng/µl)wasaddedtotheeluateasastabilizer.Theeluatewasseparatedona870

10%SDS-PAGEgelandproteinsweredetectedbysilverstaining.871

872

Yeastmanipulations873

AllyeaststrainsarederivativesoftheSK1backgroundandarelistedinSupple-874

mentaryDataTable2.Yeaststrainswereobtainedbydirecttransformationor875

crossing to obtain the desired genotype. The following alleles have been de-876

scribed previously: mlh1Δ, mlh3Δ as well as spore-autonomous fluorescent877

markerforthelivecellrecombinationassays38,39.878

YIplac211 plasmid derivatives carryingMLH1 (pYIplac211-MLH1, pML535) or879

MLH3(pYIplac211-MLH3,pML536),aswellastherespectivepromoter(~500bp880

28

upstreamofATG)andterminator(~200bpdownstreamofSTOP)regionswere881

used tocomplementmlh1∆ormlh3∆mutantstrains, respectively.pYIplac211-882

MLH1andpYIplac211-MLH3werelinearizedandintegratedinthepromoterre-883

gion of the respective genomic loci. pYIplac211-MLH1Q572A-L575A-F578A(pML538),884

encodingMlh1P,andpYIplac211-MLH3Q293A-V296A-F300A(pML540),encodingMlh3P,885

weregeneratedbyrestrictiondigest-mediatedinsertionofasyntheticfragment886

carrying the respectivemutations into pYIplac211-MLH1or pYIplac211-MLH3.887

PCR-basedC-terminaltaggingofMLH1andMLH3wasperformedusingstandard888

procedures40.889

Rfc1wasC-terminally taggedwithTAPtag.TheMlh1-HAandthe internally890

FLAG-Myc-taggedMlh3constructsweredescribedpreviously30,41.Transformants891

wereconfirmedusingPCRdiscriminatingbetweencorrectandincorrectintegra-892

tionsandsequencing.Allexperimentswereperformedat30°C.Twodifferentap-893

proacheswereusedformeiosisinduction.Inthefirstone,cellsweregrowninSPS894

presporulationmediumandtransferred insporulationmediumasdescribed42.895

Forhighlysynchronouscopper-induciblemeiosis,theprocedurewasdescribed43.896

Briefly, cellswere grown in YPD to exponential phase. Exponentially growing897

yeastwereinoculatedatOD600=0.05intoreducedglucoseYPD(1%yeastextract,898

2%peptone,1%glucose)andgrowntoanOD600=11-12for16-18h.Cellswere899

washed, resuspended in sporulation medium (1.0% [w/v] potassium acetate,900

0.02%[w/v]raffinose,0.001%polypropyleneglycol)atOD600=2.5.After2h,901

copper(II)sulfate(50µM)wasaddedtoinduceIME1expressionfromtheCUP1902

promoter.903

904

Analysisofrecombinationusingspore-autonomousfluorescence905

Thespore-autonomousfluorescenceanalysisofrecombinationwasperformedas906

described39,withminormodifications.Diploidyeastcellcolonieswerestreaked907

onYP2%glycerolplates,grownfor48h,andsinglecolonieswereexpandedtwicein908

YPDplatesat30°Cfor24h.Cellswerethentransferredtosporulationmedium909

plates(SPM,2%KAc)andincubatedat30°Cfor48h.Sporeswereresuspended910

inSPM,brieflysonicatedandtransferredontoPoly-L-Lysinecoatedmicroscopy911

slides.ImageswerecapturedinfourchannelsusingaWide-fieldDeltaVisionmul-912

tiplexedmicroscopewitha60x1.4NADICOilPlanApoNobjectiveandapeco.edge913

29

5.5cameraunderthecontrolofSoftworx(AppliedPrecision).Imageswerepro-914

cessed in Fiji and the pattern of spore fluorescence in tetrads was manually915

scored.Onlytetradswitheachfluorescentmarkeroccurringintwosporeswere916

includedinthefinalassay.Recombinationfrequency,expressedasmapdistance917

incentimorganswascalculatedusingtheStahllabonlinetools(https://elizabeth-918

housworth.com/StahlLabOnlineTools/)44. Three replicates using independent919

cloneswereanalyzed.≥900tetradswerescoredforeachgenotype.920

921

Analysisofsporeviability922

Sporeviabilitywasdeterminedbymicrodissectionof≥156sporesfromatleast923

twoindependentexperimentsafterinductionofmeiosisonSPMplatesat30°C924

for24h.925

926

Proteinstabilityanalysesbywesternblotting927

Proteinextractsfromyeastwereperformedusingthetrichloroaceticacid(TCA)928

method45.Briefly,exponentiallygrowingcultureswereharvestedanddisrupted929

usingglassbeadsin10%TCA.Precipitateswerecollectedbycentrifugation,re-930

suspendedin2xNuPAGEsamplebuffer(Invitrogen),andneutralizedwith1M931

Tris.Sampleswerethenboiledat95°Cfor5min,clearedbycentrifugation,and932

separatedinNuPAGE3-8%Tris-Acetategels(Invitrogen).Aftergelelectrophore-933

sis,proteinsweretransferredontoPVDFmembranes(GEHealthcare).Antibodies934

targeting the following tags or proteins were used: mouse anti-FLAG HRP-935

conjugated (1:15,000, A8592-1MG, Sigma), rabbit anti-FLAG (1:2000, F7425-936

.2MGSigma),rabbitanti-Crm1(1:5000,agiftfromK.Weis,ETHZurich),swine937

anti-rabbitHRP-conjugated(1:5000,P0399,Dako).Forgelsourcedata,seeSup-938

plementaryFigure1.939

940

Co-immunoprecipitationandWesternblotanalysis941

1.2x109cellswereharvested,washedoncewithPBS,andlysedin3mllysisbuffer942

[20mMHEPES-KOHpH7.5,150mMNaCl,0.5%TritonX-100,10%glycerol,1943

mMMgCl2,2mMEDTA;1mMPMSF;1xCompleteMiniEDTA-Free(Roche);1X944

PhosSTOP(Roche);125U/mlbenzonase]withglassbeadsthreetimesfor30sin945

aFastprepinstrument(MPBiomedicals,SantaAna,CA).Thelysatewasincubated946

30

1h at 4 °C. 100μl of PanMouse IgGmagnetic beads (ThermoScientific)were947

washedwith100μllysisbuffer,preincubatedin100μg/mlBSAinlysisbufferfor948

2hat4°Candthenwashedtwicewith100μllysisbuffer.Thelysatewascleared949

bycentrifugationat13,000xg for5minand incubatedovernightat4°Cwith950

washedPanMouse IgGmagneticbeads.Themagneticbeadswerewashed four951

timeswith1mlwashbuffer[20mMHEPES-KOHpH7.5,150mMNaCl,0.5%Tri-952

tonX-100,5%Glycerol,1mMMgCl2,2mMEDTA,1mMPMSF,1xCompleteMini953

EDTA-Free(Roche)].Thebeadswereresuspendedin30μlTEV-Cbuffer(20mM954

Tris-HCl pH 8, 0.5mM EDTA, 150mMNaCl, 0.1%NP-40, 5% glycerol, 1mM955

MgCl2,1mMDTT)with3μlTEVprotease(1mg/ml)andincubatedfor2hat23956

°Cunderagitation.Theeluatewastransferredtoanewtube.Beadseluatewas957

heatedat95°Cfor10minandloadedonpolyacrylamidegel[4-12%Bis-Trisgel958

(Invitrogen)]andruninMOPSSDSRunningBuffer(LifeTechnologies).Proteins959

werethentransferredtoPVDFmembraneusingTrans-Blot®Turbo™Transfer960

System(Biorad)at1Aconstant,upto25Vfor45min.Proteinsweredetected961

usingc-Mycmousemonoclonalantibody(9E10,SantaCruz,1:500),HA.11mouse962

monoclonalantibody(16B12,Biolegend,1:750)orTAPrabbitmonoclonalanti-963

body(Invitrogen,CAB1001,1:4,000).TheTAPantibodystilldetectstheCBP(Cal-964

modulinBindingProtein)moietyafterTEVcleavageoftheTAPtag.Signalwas965

detectedusingtheSuperSignalWestPicoorFemtoChemiluminescentSubstrate966

(ThermoFisher).ImageswereacquiredwithaChemidocsystem(Biorad).Forgel967

sourcedata,seeSupplementaryFigure1.968

969

Chromatinimmunoprecipitationandreal-timequantitativePCR970

Foreachmeiotictimepoint,2x108cellswereprocessedasdescribed46,withthe971

followingmodifications:lysiswasperformedinlysisbufferplus1mMPMSF,50972

μg/mlaprotininand1xCompleteMiniEDTA-Free(Roche),using0.5mmzirco-973

nium/silicabeads(BiospecProducts,Bartlesville,OK).Thelysatewasdirectlyap-974

pliedon50μlPanMouseIgGmagneticbeads.Beforeuse,magneticbeadswere975

blockedwith5μg/μlBSAfor4hat4°C.976

QuantitativePCRwasperformedfromtheimmunoprecipitatedDNAorthewhole977

cellextractusingaQuantStudio5Real-TimePCRSystemandSYBRGreenPCR978

mastermix(AppliedBiosystems,ThermoScientific)asdescribed46.Resultswere979

31

expressedas%ofDNAinthetotalinputpresentintheimmunoprecipitatedsam-980

pleandnormalizedbythenegativecontrolsiteinthemiddleofNFT1,a3.5kb981

longgene.Forthemeiotictime-courseinFigure3f,thedatawerefurthernormal-982

izedbythevalueatthe2htime-point(timeofmeiosisinductionbycopperaddi-983

tion).PrimersforGAT1,BUD23,HIS4LEU2,AxisandNFT1havebeendescribed30.984

985

ADDITIONALREFERENCES986

987

31 El-Shemerly, M., Hess, D., Pyakurel, A. K., Moselhy, S. & Ferrari, S. ATR-988

dependent pathways control hEXO1 stability in response to stalled forks. 989

Nucleic Acids Res 36, 511-519, doi:10.1093/nar/gkm1052 (2008). 990

32 Cannavo, E. & Cejka, P. Sae2 promotes dsDNA endonuclease activity within 991

Mre11-Rad50-Xrs2 to resect DNA breaks. Nature 514, 122-125, 992

doi:10.1038/nature13771 (2014). 993

33 Iaccarino, I., Marra, G., Palombo, F. & Jiricny, J. hMSH2 and hMSH6 play 994

distinct roles in mismatch binding and contribute differently to the ATPase 995

activity of hMutSalpha. EMBO J 17, 2677-2686, 996

doi:10.1093/emboj/17.9.2677 (1998). 997

34 Anand, R., Pinto, C. & Cejka, P. Methods to Study DNA End Resection I: 998

Recombinant Protein Purification. Methods Enzymol 600, 25-66, 999

doi:10.1016/bs.mie.2017.11.008 (2018). 1000

35 Palombo, F. et al. hMutSbeta, a heterodimer of hMSH2 and hMSH3, binds to 1001

insertion/deletion loops in DNA. Curr Biol 6, 1181-1184, doi:10.1016/s0960-1002

9822(02)70685-4 (1996). 1003

36 Biswas, E. E., Chen, P. H. & Biswas, S. B. Overexpression and rapid 1004

purification of biologically active yeast proliferating cell nuclear antigen. 1005

Protein Expr Purif 6, 763-770, doi:10.1006/prep.1995.0007 (1995). 1006

37 Reginato, G., Cannavo, E. & Cejka, P. Physiological protein blocks direct the 1007

Mre11-Rad50-Xrs2 and Sae2 nuclease complex to initiate DNA end 1008

resection. Genes Dev, doi:10.1101/gad.308254.117 (2018). 1009

38 Arter, M. et al. Regulated Crossing-Over Requires Inactivation of 1010

Yen1/GEN1 Resolvase during Meiotic Prophase I. Dev Cell 45, 785-800 1011

e786, doi:10.1016/j.devcel.2018.05.020 (2018). 1012

39 Thacker, D., Lam, I., Knop, M. & Keeney, S. Exploiting spore-autonomous 1013

fluorescent protein expression to quantify meiotic chromosome behaviors in 1014

Saccharomyces cerevisiae. Genetics 189, 423-439, 1015

doi:10.1534/genetics.111.131326 (2011). 1016

40 Wild, P. et al. Network Rewiring of Homologous Recombination Enzymes 1017

during Mitotic Proliferation and Meiosis. Mol Cell 75, 859-874 e854, 1018

doi:10.1016/j.molcel.2019.06.022 (2019). 1019

41 Duroc, Y. et al. Concerted action of the MutLbeta heterodimer and Mer3 1020

helicase regulates the global extent of meiotic gene conversion. Elife 6, 1021

doi:10.7554/eLife.21900 (2017). 1022

42 Murakami, H., Borde, V., Nicolas, A. & Keeney, S. Gel electrophoresis 1023

assays for analyzing DNA double-strand breaks in Saccharomyces cerevisiae 1024

32

at various spatial resolutions. Methods Mol Biol 557, 117-142, 1025

doi:10.1007/978-1-59745-527-5_9 (2009). 1026

43 Chia, M. & van Werven, F. J. Temporal Expression of a Master Regulator 1027

Drives Synchronous Sporulation in Budding Yeast. G3 (Bethesda) 6, 3553-1028

3560, doi:10.1534/g3.116.034983 (2016). 1029

44 Stahl, F. W. & Lande, R. Estimating interference and linkage map distance 1030

from two-factor tetrad data. Genetics 139, 1449-1454 (1995). 1031

45 Matos, J. et al. Dbf4-dependent CDC7 kinase links DNA replication to the 1032

segregation of homologous chromosomes in meiosis I. Cell 135, 662-678, 1033

doi:10.1016/j.cell.2008.10.026 (2008). 1034

46 Borde, V. et al. Histone H3 lysine 4 trimethylation marks meiotic 1035

recombination initiation sites. EMBO J 28, 99-111, 1036

doi:10.1038/emboj.2008.257 (2009).1037

1038

MAINTEXTSTATEMENTS1039

1040

Acknowledgements1041

ThisworkwassupportedbygrantsfromtheSwissNationalScienceFoundation1042

(31003A_17544)andERC(681-630)toP.C.,InstitutCurieandCNRStoV.B.,by1043

AgenceNationaledelaRecherche(ANR-15-CE11-0011)toV.B.andJ.-B.C.,bythe1044

NovoNordiskFoundation(NNF15OC0016662)andERC(724718)toE.R.H,and1045

theSwissNationalScienceFoundation(155823and176108)toJ.M.Wethank1046

JosefJiricny(ETHZurich)andmembersoftheCejkalaboratoryforhelpfulcom-1047

mentsonthemanuscriptandNeilHunterforcommunicatingresultspriorto1048

publication.1049

1050

Conflictofinterest1051

Theauthorsdeclarenoconflictofinterest.1052

1053

Authorcontributions1054

E.C.,A.S.,R.A.andP.C.planned,performedandanalyzedthemajorityoftheexper-1055

imentsandwrotethepaper.L.R.andA.A.performedmostoftheexperimentswith1056

yeastrecombinantproteinsandelectrophoreticmobilityshiftassays.N.W.per-1057

formed experiments to define simultaneousDNAbinding byMLH1-MLH3 and1058

MSH4-MSH5.J.H.performedexperimentswithyeastmlh1andmlh3variantsmu-1059

tated inPIP-box-likesequences,andthedatawereanalyzedtogetherwith J.M.1060

ChipexperimentsandRfc1-Mlh1andRfc1-Mlh3pulldownassayswerecarried1061

33

outbyC.A.,thedatawereanalyzedtogetherwithV.B.J-B.C.helpedpreparethe1062

MLH1-MLH3expressionconstruct anddesignedexperimentswith thePIP-box1063

peptide.X.A-G.andE.R.H.preparedtheMSH4-MSH5expressionconstruct.Allau-1064

thorscontributedtopreparethefinalversionofthemanuscript.1065

1066

DataAvailabilityStatement1067

Allrelevantdatageneratedoranalyzedduringthisstudyareincludedinthispub-1068

lishedarticleanditssupplementaryinformationfile.1069

1070

EXTENDEDDATAFIGURELEGENDS1071

1072

Extended Data Figure 1. ATP hydrolysis promotes MLH1-MLH3 to nick1073

scDNA.a,AschemeofMLH1andMLH3constructs.Themaltose-bindingprotein1074

(MBP)onMLH3wascleavedduringproteinpurification.b,RecombinantMLH1-1075

MLH3usedinthisstudy.The4-15%gradientpolyacrylamidegelwasstainedwith1076

CoomassieBlue.c,NucleaseassaywithMLH1-MLH3and2.7kbp-longsupercoiled1077

DNA(scDNA)asasubstrate.Thereactionwith5mMmanganeseacetatewasin-1078

cubatedwithoutATPat37 °C.d,Quantitationofassayssuchas inc.Averages1079

shown;errorbars,SEM;n=3.e,NucleaseassaywithMLH1-MLH3(300nM)and1080

2.7kbp-longscDNA.LinearDNAwasusedasamarker.Theassaywascarriedout1081

at37°Candcontained5mMmanganeseacetateandATP(0.5mM).TheMLH1-1082

MLH3nucleaseintroducesnicksindsDNAbutdoesnotlinearizedsDNA.f,Quan-1083

titationofnucleaseassayswithMLH1-MLH3withoutorwithATP(0.5mM),inthe1084

presenceofmanganese(5mM).Averagesshown;errorbars,SEM;n=4.g,Nucle-1085

aseassaywithMLH1-MLH3and5mMmagnesiumacetate.Thereactionbuffer1086

containedATP(0.5mM).Theassaywascarriedoutat37°C.Theheterodimer1087

exhibitsbarelydetectablenucleaseactivityinmagnesium.h,Nucleaseassaywith1088

MLH1-MLH3andvariousnucleotidecofactors(ADP,ATPandnon-hydrolysable1089

ATPanalogsATP-γ-SandAMP-PNP,all0.5mM).Theassaywascarriedoutat371090

°Cwith5mMmanganeseacetate.Thepanelshowsarepresentativeexperiment.1091

i,Quantitationofnucleaseassayssuchasinpanelh,supplementedwithvarious1092

nucleotide co-factorsand their analogs (0.5mM).Averages shown;errorbars,1093

34

SEM; n=4. j, Purified MLH1-MLH3 variants used in this study. MLH1(EA),1094

MLH1(E34A); MLH3(EA), MLH3(E28A); MLH3(3ND), MLH3(D1223N-Q1224K-1095

E1229K).The4-15%gradientpolyacrylamidegelwas stainedwithCoomassie1096

Blue. k,AlignmentofMLH1andMLH3ATPasemotifs.Conservedresiduesare1097

highlightedinred.AlaninesubstitutionsinMLH3andMLH1mutantsusedinthis1098

studyareinitalics.l,NucleaseassaywithwildtypeMLH1-MLH3andindicated1099

variantsdeficientinATPhydrolysis,withoutorwithATP(0.5mM).Theassaywas1100

carriedoutat37°C,with5mMmanganeseacetate.Thepanelshowsarepresenta-1101

tiveexperiment.m,Quantitationofnucleaseassaysasshowninpanell,without1102

orwithATP(0.5mM),witheitherwildtypeorMLH1-MLH3variantsmutatedin1103

conservedATPasedomain residues.Averages shown; errorbars, SEM;n≥4.n,1104

ElectrophoreticmobilityshiftassaywithindicatedMLH1-MLH3variants,oligo-1105

nucleotide-basedHJasthesubstrate, intheabsenceofATPandnomagnesium1106

(with3mMEDTA).Asterisk(*)indicatesthepositionoftheradioactivelabel.A1107

representative experiment is shown at the bottom, a quantitation (averages1108

shown,n=3;errorbars,SEM)atthetop.o,NucleaseassayswithwildtypeMLH1-1109

MLH3onoligonucleotide-basedDNAsubstrates(Hollidayjunction,HJandnicked1110

Hollidayjunction,nickedHJ).Theasteriskindicatesthepositionoftheradioactive1111

label.Theassaywascarriedoutat37°C,with5mMmanganeseormagnesium1112

acetate,asindicated,withATP(1mM).Theproductswereanalyzedby10%na-1113

tivepolyacrylamidegelelectrophoresis.1114

1115

ExtendedDataFigure2.HumanandyeastMutSγcomplexespreferentially1116

bindbranchedDNAintermediates.a,RecombinanthumanMSH4-MSH5used1117

in this study.b,Electrophoreticmobility shiftassayswithhumanMSH4-MSH51118

andindicatedDNAsubstrates.Asterisk(*)indicatesthepositionoftheradioac-1119

tivelabel.Theassayswerecarriedoutinabuffercontaining2mMmagnesium1120

acetatewithout ATP. c, Quantitation of DNA binding assays such as shown in1121

panelb.Averagesshown;errorbars,SEM;n=3.d,Electrophoreticmobilityshift1122

assayswith yeastMsh4-Msh5and indicatedDNA substrates.Asterisk (*) indi-1123

catesthepositionoftheradioactivelabel.Theassayswerecarriedoutinabuffer1124

containing2mMmagnesiumacetatewithoutATP.e,Quantitationofexperiments1125

such as shown in panel d. Averages shown; error bars, range; n=2. f,1126

35

QuantificationofelectrophoreticmobilityshiftassayswithyeastMsh4-Msh5and1127

indicated DNA substrates, without magnesium (with 3 mM EDTA). Averages1128

shown;errorbars,range;n=2.1129

1130

ExtendedDataFigure3.MutSγandMutLγphysically interactandmoder-1131

atelystabilizeeachotheratDNAjunctions.a,Toinvestigatetheinterplayof1132

MutLγandMutSγatDNAjunctions,weperformedelectrophoreticmobilityshift1133

assayswitheitherorbothcomplexesundermorestringentconditions(75mM1134

NaCl,2mMmagnesiumacetate), separatedon0.6%agarosegels.Under these1135

conditions,MSH4-MSH5lostthecapacitytostablybindHJs/D-Loops,butcould1136

helpstabilizetheMutSγ-MutLγcomplex.ThebindingofMutLγalonewasnotsta-1137

ble,asevidencedbyaweakprotein-DNAbandandthepresenceofsmearinthe1138

lanesindicativeofcomplexesthatdissociatedduringelectrophoresis.Theaddi-1139

tionofMutSγresultedinamoderatestabilizationoftheprotein-DNAcomplex,1140

and aminor super-shift in electrophoreticmobility of the stable protein-DNA1141

band(indicatedbytheredandbluearrows).Shownarerepresentativeexperi-1142

ments.b,Electrophoreticmobilityshiftassaysasinpanela,butwithoutmagne-1143

sium(with3mMEDTA).c,Quantitationofassayssuchasshowninpanelb.The1144

Yaxis indicatesrelativeprotein-DNAcomplexstability,obtainedupondividing1145

theprotein-DNAbandintensity(seeblueorredarrowsinpanelb)bytheinten-1146

sityoftheradioactivesignalinthelaneabovethefreesubstrateband,butbelow1147

theprotein-DNAband.Averagesshown;errorbars,SEM;n=5.d,Assaysasina,1148

with humanMutLγ and either human or yeastMutSγ. The supershiftwas ob-1149

servedonlywhenthecognatehumancomplexeswerecombined.e,Electropho-1150

reticmobilityshiftassaysasina,butwithyeastMutLγandMutSγcomplexes.f,1151

ProteininteractionassayswithimmobilizedMLH1-MLH3(bait)andMSH4-MSH51152

(prey).The10%polyacrylamidegelwasstainedwithsilver.g,Proteininteraction1153

assayswithimmobilizedhumanMSH4-MSH5oryeastMsh4-Msh5thatwereused1154

asbaits,andhumanMLH1-MLH3(prey).Theelutedproteinswereanalyzedby1155

silver staining. Although interaction between yeast Msh4-Msh5 and human1156

MLH1-MLH3wasstilldetected,itwasweakerthantheinteractionbetweenthe1157

cognateMSH4-MSH5andMLH1-MLH3complexes.h, Protein interactionassay1158

with immobilized yeast Msh4-Msh5 (bait) and yeast Mlh1-Mlh3 (prey). The1159

36

eluted proteinswere analyzed bywestern blotting. i, Electrophoreticmobility1160

shift assayswithMLH1-MLH3andMSH4-MSH5, as indicated, andoligonucleo-1161

tide-based HJ DNA substrate. 32P-labeled λDNA/HindIII digest was used as a1162

marker.TheDNA-boundMLH1-MLH3andMSH4-MSH5speciesmigratehighup1163

ontheagarosegelwheretheresolutioncapacityislimited.j,Electrophoreticmo-1164

bilityshiftassaywithyeastKu70-Ku80heterodimerandHJDNAsubstrate.Ku1165

bound thedsDNAends of the fourHJ arms, resulting in up to4 heterodimers1166

bound to the DNA substrate (lanes 5-7). Comparisonwith λ DNA/HindIII and1167

panelirevealedthattheKu-DNAcomplexmigratesmuchfasterthanDNA-bound1168

MLH1-MLH3andMSH4-MSH5.ThissuggeststhatmultipleunitsofMLH1-MLH31169

andMSH4-MSH5bindDNA.1170

1171

Extended Data Figure 4. MSH4-MSH5 promotes DNA cleavage by MLH1-1172

MLH3,butthecomplexdoesnotexhibitresolvaseactivity.a,Quantitationof1173

kineticnucleaseassayswithMLH1-MLH3(50nM)withoutorwithMSH4-MSH51174

(50nM)using5.6kbp-longscDNA.Theassayswerecarriedoutat30°Cinthe1175

presenceof5mMmanganeseacetateand2mMATP.Averagesshown;errorbars,1176

SEM;n=3.b,NucleaseassayswithMSH4-MSH5andeitherwildtypeMLH1-MLH31177

or nuclease-dead MLH1-MLH3 (D1223N-Q1224K-E1229K, 3ND). The assays1178

werecarriedoutat30°Cinthepresenceof5mMmanganeseacetateand0.5mM1179

ATP. c, Quantitation of nuclease assays with variousMLH1-MLH3 andMSH4-1180

MSH5concentrations,asindicated.Theassayswerecarriedoutat30°Cinthe1181

presenceof5mMmanganeseacetateand0.5mMATP.Averagesshown;error1182

bars,SEM,n=3.Theefficiencyofnucleasecleavagewasgenerallydependenton1183

theconcentrationsused.Whenusing50nMMLH1-MLH3,themaximalcleavage1184

efficiencywasachieved togetherwith50nMMSH4-MSH5,no further increase1185

whenusing100nMMSH4-MSH5wasobserved.Thissuggeststhatbothheterodi-1186

mersmayformastoichiometriccomplex.Viceversa,whenusing50nMMSH4-1187

MSH5,afurtherincreaseofDNAcleavagewasobservedwhenMLH1-MLH3con-1188

centrationsexceeded50nM,whichisinagreementwiththecapacityofMLH1-1189

MLH3tocleaveDNAonitsown.d,QuantitationofnucleaseassayswithMLH1-1190

MLH3andMSH4-MSH5,as indicated, in thepresenceofvariousnucleotideco-1191

factors or their analogs (2mM). The assays were carried out at 30 °C in the1192

37

presenceof5mMmanganeseacetate.Averagesshown;errorbars,SEM;n≥4.e,1193

RepresentativenucleaseassayswithMSH4-MSH5andvariantsofMLH1-MLH31194

deficientinATPhydrolysis,asindicated.Theassayswerecarriedoutat30°Cin1195

thepresenceof5mMmanganeseacetateand0.5mMATP.f,Representativenu-1196

cleaseassayswithMLH1-MLH3andvariantsofMSH4-MSH5deficientinATPhy-1197

drolysis,asindicated.Theassayswerecarriedoutat30°Cinthepresenceof51198

mMmanganeseacetateand0.5mMATP.g,RecombinantMSH4-MSH5anditsvar-1199

iants used in this study.MSH4(G685A),MSH4(GA);MSH5(G597A),MSH5(GA).1200

The 4-15% gradient polyacrylamide gel was stained with Coomassie Blue. h,1201

Quantitation of electrophoreticmobility shift assayswithMSH4-MSH5 and its1202

ATPasemotifmutant variants. Oligonucleotide-basedHJwas used as the sub-1203

strate.Asterisk(*)indicatesthepositionoftheradioactivelabel.ATPwasnotin-1204

cludedinthebindingbuffer.ThemutationsdidnotaffectthecapacityofMSH4-1205

MSH5tobindDNA.Averagesshown;errorbars,SEM;n=3.i,Nucleasereactions1206

werecarriedoutwithyeastorhumanMutSgandMutLgcomplexes,asindicated1207

(50nM),with2.7kbp-longscDNAsubstrate.WhilehumanMutSgpromotedDNA1208

cleavagebyhumanMutLg(comparelanes2and3),yeastMutSgdidnotnotably1209

promoteDNAcleavagebyhumanMutLg(comparelanes2and5),andrecipro-1210

cally,humanMutSgdidnotpromoteDNAcleavagebyyeastMutLg(comparelanes1211

7 and 8). j, Quantitation of nuclease assayswith human and yeastMutSg and1212

MutLgcomplexesasinpaneli,butwith10.3kbp-longscDNAsubstrate.k,Cleav-1213

ageofpIRbke8mutcruciformDNA(invertedrepeatsfoldingbacktoformaHol-1214

lidayjunctionstructure)byMutSgandMutLgcomplexes.Thequantitationbelow1215

thelanesrepresentsanaveragefromtwoindependentexperiments.Simultane-1216

ouscleavageofbothstrandsatthejunctionpointwouldleadtolinearDNA.No1217

linearDNAwasobservedwithMutSgandMutLg, indicatinga lackofcanonical1218

resolvaseactivity.l,Representativenucleaseassayswithindicatedproteinsand1219

oligonucleotide-basedHJDNA.Asterisk(*)indicatesthepositionoftheradioac-1220

tivelabel.NoDNAcleavagewasobserved,indicatingalackofstructure-specific1221

DNAcleavageactivityontheoligonucleotide-basedsubstrate.Theproductswere1222

analyzedby15%denaturingpolyacrylamidegelelectrophoresis.1223

1224

38

ExtendedDataFigure5.MutSbbutnotMutSastimulatesMutLgtoasimilar1225

extentasMutSg.a,RecombinantMutSb(MSH2-MSH3)usedinthisstudy.b,Re-1226

combinant MutSa (MSH2-MSH6) used in this study. c, Nuclease assays with1227

MLH1-MLH3,MSH4-MSH5,andMSH2-MSH3orMSH2-MSH6,as indicated.The1228

assayswerecarriedoutat30°Cinthepresenceof5mMmanganeseacetateand1229

0.5mMATP.Arepresentativeexperimentisshownatthebottom,aquantitation1230

(averagesshown;n=3;errorbars,SEM)atthetop.1231

1232

ExtendedDataFigure6.StimulationofthenucleaseactivityofMutSg-MutLg1233

byEXO1(D173A).a,RecombinantEXO1(D173A),usedinthisstudy.The4-15%1234

gradientpolyacrylamidegelwasstainedwithCoomassieBlue.b,Nucleaseassays1235

with MLH1-MLH3 and MSH4-MSH5, as indicated, without (left) or with1236

EXO1(D173A)(right).Theassayswerecarriedoutat30°Cinthepresenceof51237

mMmanganeseacetateand0.5mMATP.Arepresentativeexperimentisshown1238

atthebottom,aquantitation(averagesshown;n=3;errorbars,SEM)atthetop.1239

c,NucleaseassayswithMLH1-MLH3and/orEXO1(D173A),asindicated.Theas-1240

sayswerecarriedoutat30°Cinthepresenceof5mMmanganeseacetateand0.51241

mMATP.Arepresentativeexperimentisshownatthebottom,aquantitation(av-1242

eragesshown;n=4;errorbars,SEM)atthetop.EXO1(DA)doesnotpromotethe1243

nucleaseofMLH1-MLH3alone.ThelimitedDNAcleavageinlane3likelyresults1244

fromresidualnucleaseactivityofEXO1(D173A)thatbecomesapparentathigh1245

proteinconcentrations(100nM)inthepresenceofmanganese.d,Quantitationof1246

electrophoretic mobility shift assays with MLH1-MLH3, MSH4-MSH5 and1247

EXO1(D173A),asindicated.Theprotein-DNAspecieswereresolvedin1%aga-1248

rosegels.Averagesshown;errorbars,SEM;n=5.EXO1(D173A)didnotnotably1249

affect DNA binding of MLH1-MLH3 andMSH4-MSH5. e, Nuclease assays with1250

MLH1-MLH3, MSH4-MSH5 with either human EXO1(D173A) or yeast1251

Exo1(D173A),asindicated.Theassayswerecarriedoutat30°Cinthepresence1252

of 5mMmanganese acetate and 0.5mMATP. A representative experiment is1253

shownatthebottom,aquantitation(averagesshown;n=5;errorbars,SEM)at1254

thetop.f,NucleaseassayswithMLH1-MLH3,MSH2-MSH3andEXO1(D173A),as1255

indicated. The assays were carried out at 30 °C in the presence of 5 mM1256

39

manganeseacetateand0.5mMATP.Arepresentativeexperimentisshownatthe1257

bottom,aquantitation(averagesshown;n=3;errorbars,SEM)atthetop.1258

1259

Extended Data Figure 7. RFC-PCNA promote the nuclease activity of the1260

MutSg-MutLg-EXO1(DA)ensemble.a,RecombinanthumanandyeastRFCand1261

PCNAusedinthisstudy.The4-15%gradientpolyacrylamidegelwasstainedwith1262

CoomassieBlue.b,NucleaseassayswithscDNAandindicatedproteins(all50nM,1263

excepthumanPCNA,100nM)werecarriedoutwith5mMmagnesiumacetate1264

and2mMATPat37°C.Arepresentativeexperimentisshownatthebottom,a1265

quantitation(averagesshown;n≥4;errorbars,SEM)atthetop.c,Experimentsas1266

inpanelb,comparingtheefficacyofhumanandyeastRFCasapartoftheMLH31267

nucleaseensemble.Averagesshown;n=4;errorbars,SEM.d,Nucleasereactions1268

containingMLH1-MLH3(50nM),MSH4-MSH5(50nM),EXO1(D173A)(50nM)1269

and yRFC-hPCNA (50-100 nM, respectively) (column 1), withoutMSH4-MSH51270

(column2)orwithouthEXO1(D173A) (column3).Reactionswere carriedout1271

with5mMmagnesiumacetateand2mMATPat37°C.Averagesshown;error1272

bars, SEM; n≥4. e, Kinetic nuclease assays withMLH1-MLH3 (50 nM), MSH4-1273

MSH5 (50 nM), EXO1(D173A) (50 nM) and yRFC-hPCNA (50-100 nM, respec-1274

tively),asindicated.Reactionswerecarriedoutwith5mMmagnesiumacetate1275

and2mMATPat37°C.Averagesshown;errorbars,SEM;n≥5.f,Nucleaseassays1276

withMLH1-MLH3 (50 nM),MSH4-MSH5 (50 nM), EXO1(D173A) (50 nM) and1277

hRFC-hPCNA(50-100nM,respectively),as indicated,withsupercoiled(left)or1278

relaxedDNA(right).Reactionswerecarriedoutwith5mMmagnesiumacetate1279

and2mMATPat37°C.Shownisarepresentativeexperiment.RFC-PCNAdonot1280

stimulate the cleavage of relaxed DNA. g, Nuclease assays with MLH1-MLH3,1281

MSH4-hMSH5,EXO1(D173A)withoutorwithyRFC-hPCNA,asindicated.Theas-1282

sayswerecarriedoutat37°Cinthepresenceof5mMmanganeseacetateand21283

mMATP.Arepresentativeexperimentisshownatthebottom,aquantitation(av-1284

eragesshown;n=3;errorbars,SEM)atthetop.Withoutmagnesium,nostimula-1285

tion of DNA cleavage by RFC-PCNAwas observed.h, Nuclease reactions with1286

MLH1-MLH3(50nM),MSH4-MSH5(50nM),EXO1(D173A)(50nM)andyRFC-1287

hPCNA(50-100nM,respectively),asindicated.Reactionswerecarriedoutwith1288

5mMmagnesiumacetateand2mMATPat37°C.Averagesshown;errorbars,1289

40

SEM;n≥5. i,NucleaseassayswithMLH1-MLH3(50nM),MSH4-MSH5(50nM),1290

EXO1(D173A) (50nM) and yRFC-hPCNA (50-100nM, respectively) and5mM1291

magnesiumacetate,eitherwithnonucleotideco-factor(lane2),withATP(2mM,1292

lane3)orADP(2mM,lane4).ATPisstrictlyrequiredforDNAcleavagebythe1293

nucleaseensemble.j,RepresentativenucleaseassayswithMLH1-MLH3(50nM),1294

MSH4-MSH5(50nM),EXO1(D173A)(50nM)andyRFC-hPCNA(50-100nM,re-1295

spectively),lane2.Lanes3-7containinsteadMLH1-MLH3orMSH4-MSH5vari-1296

antsdeficientinATPhydrolysis,asindicated.SeeFig.1d,eforthespecificmuta-1297

tions.Reactionswerecarriedoutwith5mMmagnesiumacetateand2mMATP1298

at37°C.Averagesshown;errorbars,SEM;n=4.k,Nucleaseassayswithindicated1299

oligonucleotide-basedsubstratescarriedoutat37 °C in thepresenceof5mM1300

manganeseacetateand2mMATP.Allproteins30nM,asindicated.Asterisk(*)1301

indicatesthepositionoftheradioactivelabel.Thereactionproductswereana-1302

lyzedona15%denaturingpolyacrylamidegel.NoDNAcleavagewasobserved.1303

1304

ExtendedDataFigure8.PIPbox-likemotifsinEXO1,MLH3andMLH1facil-1305

itatethestimulatoryeffectofRFC-PCNAonthehMLH3nucleaseensemble.1306

a,TheMLH1P-MLH3Pvariant(seeFig.3b)isnotimpairedinHJ-binding.Electro-1307

phoreticmobilityshiftassaywascarriedoutwith5ng/reactiondsDNAcompeti-1308

torand3mMEDTA(nomagnesium).Asterisk(*) indicates thepositionof the1309

radioactive label. b, TheMLH1PandMLH3P variant combinations are not im-1310

pairedinnucleaseactivitywithoutorwithMSH4-MSH5andEXO1(D173A)inthe1311

absenceofRFC-PCNA.Thenucleaseassayswereperformedwith5mMmanga-1312

neseacetateand2mMATPat37°C.Averagesshown;errorbars,SEM,n=3.c,1313

Nuclease assayswithMSH4-MSH5 (50nM), EXO1(D173A) (50nM) andyRFC-1314

hPCNA(50-100nM),andarespectiveMLH1-MLH3variant,asindicated(seeFig.1315

3b).Mutations in thePIP-box likemotifreducethestimulationof thenuclease1316

ensemblebyRFC-PCNA.Theassayswerecarriedoutwith5mMmagnesiumace-1317

tate and 2 mM ATP at 37 °C. Averages shown; error bars, SEM, n=5. d, The1318

EXO1P(D173A)variantwithmutatedPIP-boxmotif(seeFig.3b)isnotaffectedin1319

its ability to promote the nuclease of MLH1-MLH3 andMSH4-MSH5 (without1320

RFC-PCNA).Theassayswerecarriedoutwith5mMmanganeseacetateand2mM1321

ATPat37°C.Averagesshown;errorbars,SEM,n=4.e,TheEXO1P(D173A)variant1322

41

with mutated PIP-box motif (see Fig. 3b), in complex with MLH1-MLH3 and1323

MSH4-MSH5impairsthestimulatoryfunctionofyRFC-hPCNA(50-100nM).The1324

assayswerecarriedoutwith5mMmagnesiumacetateand2mMATPat37°C.1325

Averagesshown;errorbars,SEM,n=5.1326

1327

ExtendedDataFigure9.RFC-PCNApromotemeioticrecombinationinyeast1328

cells.a, Sporeviabilityupon tetradmicrodissection, analyzed in thewild type1329

strain,mlh1Dandmlh3D,andinstrainscomplementedwithaconstructexpress-1330

inguntaggedMlh1P(Q572A-L575A-F578A)orMlh3P(Q293A-V296A-F300A)at1331

theendogenouschromosomallocus.Atleast156sporesfrom2biologicalrepli-1332

cateswereanalyzedforeachgenotype.b,WesternblotanalysisofMlh1Pexpres-1333

sioninyeast.TCAextractswerepreparedfromexponentiallyproliferatingSK11334

strainsexpressingMLH1,MLH1-FLAGorMLH1P-FLAGfromtheendogenousgene1335

locus. ThePIP-box-likemutationaffects thestabilityof theFLAG-taggedMlh11336

protein.Blotswereprobedwithanti-FLAGantibody(Sigma,F7425).Crm1isa1337

proteinnormalizationcontrol.Asteriskdenotesacross-reactingband.c,Western1338

blotanalysisofMlh3Pexpressioninyeast.Asinb,butwithMLH3,MLH3-FLAGor1339

MLH3P-FLAG constructs. Blots were probed with anti-FLAG antibodies: Sigma1340

F7425(leftpanel);A8592(rightpanel).Crm1isaproteinnormalizationcontrol.1341

Mlh1-FLAGandMlh3P-FLAGshowedcomparableexpressionlevels.Asterisksde-1342

notecross-reactingbands.d,ApulldownofTAP-taggedyeastRfc1-5andassoci-1343

atedproteinsfrommeioticcellextractsfrompCUP1-IME1cells5h30minafter1344

theinductionofmeiosis.ThepresenceofMlh1-HAandMlh3-MycintheTEVelu-1345

atewasanalyzedbyWesternblotting.e,Rfc1-TAPlevelsat thethree indicated1346

meioticDSBhotspotsrelativetoanegativecontrolsite(NFT1)wereassessedby1347

ChIPandqPCRinndt80∆-arrestedcellsafter7hinmeiosis.Mlh3isnotrequired1348

fortherecruitmentofRFCtothemeioticDSBhotspots.MLH3:VBD2136;mlh3∆:1349

VBD2137.Averagesshow;errorbars,SD,n=2.1350

1351

ExtendedDataFigure10.Apossiblemodelforbiasedresolutionofrecom-1352

bination intermediates by the MLH3 nuclease ensemble. Meiotic dsDNA1353

breaks (a) are resected (1) and the resultingDNAoverhang invadesmatching1354

42

DNAonahomologouschromosome(2).TheunstableD-Loopintermediates(b)1355

are stabilized by MSH4-MSH5 (3), DNA synthesis by RFC-PCNA-Pold (4) and1356

branchmigration(5),leadingtomorestablestructurestermedsingle-endinva-1357

sions(c).Thisisfollowedbyasecondendcapture(6),andmoreDNAsynthesis1358

(7)leadingtoprecursorsofdoubleHollidayjunctions(d)andlatermatureddou-1359

bleHollidayjunctions(e).Asaresultoftheprevioussteps,MSH4-MSH5andRFC-1360

PCNAmaybepresentasymmetricallyatthe(d)or(e)intermediatesatthejunc-1361

tionspointsortheirvicinity.Theasymmetricpresenceoftheco-factorsthendi-1362

rectsandstimulatesthebiasedDNAcleavage(9)of(d)or(e)structuresbyMLH1-1363

MLH3-EXO1.Uponfinalprocessing(10)andligation(11),theultimateresultisa1364

DNAcrossovercharacterizedbyreciprocalexchangeoftheDNAarmsofthere-1365

combiningchromosomes.1366

1367

1368

Figure 1

a

1 2 3 4 5 6 Lane

No

pro

tein

No ATP

MLH1-MLH3 (300 nM)

No

pro

tein

MLH1-MLH3(3ND) (300 nM)

+ ATP

++ +

+––

–

scDNA

nicked

DNA

–

Mn2+

+ + + + – 80 220 400 400 – MLH1-MLH3 (nM)

1 2 3 4 5 6 7 8 9 Lane

kDa Ma

rke

r

116

97

66

45

MLH3

MSH4MSH5

MLH1

anti-HIS

MSH4-MSH5– – –+

b Pull down by anti-HIS

200

MSH4-MSH5

MLH1-MLH3

+++

–

scDNA

nicked DNA

–+

(both 50 nM)

No

pro

tein

c

1 2 3 4 Lane

0

10

20

30

40

50

scD

NA

cle

avag

e (

%)

Mn2+

MSH4-MSH5 (50 nM)

MLH1-MLH3

MLH1(EA)-MLH3

MLH1-MLH3(EA)

MLH1(EA)-MLH3(EA)

––––

++

++

– –

–

–– ––

–––

––

All (50 nM)No

pro

tein

e

Mn2+

0

10

20

30

40

scD

NA

cle

avag

e (

%)MLH1-MLH3 (50 nM)

MSH4-MSH5

MSH4(GA)-MSH5

MSH4-MSH5(GA)

–––

–––

++

+––

–

– ––

All (50 nM)No

pro

tein

d

Mn2+

0

10

20

30

40

scD

NA

cle

avag

e (

%)

+–

Input

250150

100

75

50

kDa

Ma

rke

r

37

anti-MLH1

EXO1

MLH3

MLH1

1 2 3 4 Lane

MLH1-MLH3

+ +–

anti-MLH1

EXO1

++ +

Pulldown

+––

MLH1-MLH3 (20 nM)

MSH4-MSH5 (20 nM)

Mn2+

0 20 40 600

10

20

30

40

50

Time (min)

scD

NA

cle

avag

e (

%)

EXO1(DA)

No EXO1(DA)

f

g

Figure 2

0

5

10

15

20

scD

NA

cle

avag

e (

%)

scDNA

nicked DNAlinear DNA

Plasmid

with HJ

Plasmid

without HJ

No

pro

tein

No

pro

tein

RFC-PCNA–– –

––+ +

+++–+

1 2 3 4 5 6 7 Lane

Mg2+

d

MLH1-MLH3 +

250

150

100

75

50

kDa

37

anti-MLH1

PCNA

MLH3

MLH1

++ –

anti-MLH1

PCNA

+

+

–

–

–

–

–

–

1 2 3 Lane

0

20

40

60

scD

NA

cle

avag

e (

%)

MLH1-MLH3

No

pro

tein

MSH4-MSH5

EXO1 (DA)

RFC-PCNA++++

++

+

++

+

–––

–––

––––

a

scDNA

nicked DNA

All proteins

(50 nM) except

PCNA

(100 nM)

Mg2+

All

proteins (50 nM)

except

PCNA (100 nM)

b

No

pro

tein

No

pro

tein

scDNA

nicked DNA

Mg

2+

Mn

2+

Mn

2+

Mg

2+

MLH1-MLH3

(300 nM)

Mn

2+

Mg

2+

Mn

2+

Mg

2+

MLH1-MLH3

MSH4-MSH5

EXO1(DA)

RFC-PCNA

1 2 3 4 5 6 7 8 Lane

Mg2+

scDNA

nicked DNA

No

pro

tein

ATP

AMP-PNP

++

––

–– –

–

1 2 3 4 Lane

0

20

40

60

scD

NA

cle

avag

e (

%)

MLH1-MLH3

MSH4-MSH5EXO1(DA)

RFC-PCNA

MLH1-MLH3

MSH4-MSH5EXO1(DA)

0

10

20

30

40

50

scD

NA

cle

avag

e (

%)

Pulldown

–

+–

No

pro

tein

MLH1-MLH3

MSH4-MSH5EXO1(DA)

RFC-PCNA

ML

H1

(EA

)

No

su

bstitu

tio

n

ML

H3

(EA

)

ML

H1

(EA

)-M

LH

3(E

A)

MS

H4

(GA

)

MS

H5

(GA

)

Su

bstitu

tio

ns

All

proteins (50 nM)

except

PCNA (100 nM)

scDNA

nicked DNA

c

ef

Mg2+

scD

NA

cle

avag

e (

%)

scD

NA

cle

avag

e (

%)

All

proteins (50 nM)

except

PCNA (100 nM)

1 2 3 4 5 Lane

scDNA

nicked DNA

0

10

20

30

40

scD

NA

cle

avag

e (

%)

MLH1-MLH3

++++–+++–– RFC-PCNA

No

pro

tein

PIP

PIP

mu

t

No

ne

Mg2+

++++–MSH4-MSH5

1 2 3 4 5 Lane

Competing peptide

a b

c

e

wild

typ

e

mlh1∆ mlh3∆

Complementation

Mlh

1

Mlh

1P

Mlh

3

Mlh

3P

d

0

5

10

15

20

Gen

eti

c d

ista

nce (

cM

)

No

ne

No

ne

wild

typ

e

mlh1∆ mlh3∆

Complementation

Mlh

1

Mlh

1P

Mlh

3

Mlh

3P

No

ne

No

ne

0

2

4

6

8

Ch

rom

oso

me

no

n-d

isju

ncti

on

(%

)

MSH4-MSH5

scDNA

nicked DNA

RFC-PCNA

MLH1P-MLH3P

EXO1

(DA)

EXO1

(DA)P

No

pro

tein

––

+ + + + + +– + + +– –

Mg2+

MLH1-MLH3

EXO1

(DA)

1 2 3 4 5 6 7 Lane

0

5

10

15

20

25

scD

NA

cle

avag

e (

%)

f

0

2

4

6

8

2 4 6 8

neg. control

HIS4LEU2

BUD23

GAT1

axis

Time in meiosis (h)

Re

lati

ve

en

ric

hm

en

t o

f R

fc1

++++–EXO1 (DA)

Figure 3

All

proteins (50 nM)

except

PCNA

(100 nM)

All

proteins (30 nM)

except

RFC(50 nM)

PCNA (100 nM)

Extended Data Figure 1

1 2 3 4 5 6 7 8 9 10 Lane

No

pro

tein

nicked DNA

scDNA

No

pro

tein

No ATP + ATP All proteins (300 nM)

MLH1-MLH3

MLH1(E34A)-MLH3

MLH1-MLH3(E28A)

MLH1(E34A)-MLH3(E28A)

++

++

–– – –

– –––––

–– –++

++

– –– – –

– –––––

a

50 100 200 400 600 800 (nM)

Mg2+

scDNA

nicked DNA

MLH1-MLH3

No

pro

tein

f

200

11697

66

45

31

22

14

kDa ML

H1

-ML

H3

ML

H1

(EA

)-M

LH

3

MLH3

(or variant)MLH1

(or variant)

ML

H1

-ML

H3

(EA

)

ML

H1

(EA

)-M

LH

3(E

A)

ML

H1

-ML

H3

(3N

D)

1 2 3 4 5 6 Lane

Ma

rke

r

j

1 2 3 Lane

scD

NA

lin

ea

rD

NA

linear DNA

scDNA

scD

NA

+ M

LH

1-M

LH

3

nicked DNA

b

m

Mn2+

Nicked

HJ

0 20 40 75 150

MLH1-MLH3

HJ

(nM)

Substrate

Free label Free label

Substrate

o

Mg2+Mg2+

g

1 2 3 4 5 6 7 Lane

Mn2+

1 2 3 4 5 6 Lane

AT

P

AD

P

AM

P-P

NP

No

co

facto

r

MLH1-MLH3

(200 nM)

scDNA

nicked DNA

Mn2+No

pro

tein

AT

P-γ

-S

200

11697

66

45

31

MLH3

MLH1

Ma

rke

r

1 2 Lane

hM

utL

γ

0 100 200 300 4000

10

20

30

40

hMLH1-hMLH3 (nM)

scD

NA

cle

avag

e (

%)

No ATP

No

pro

tein

50 100 200 400 (nM)

d

No ATP

scDNA

nicked DNA

MLH1-MLH3 Mn2+

Mn2+

1 2 3 4 5 Lane

c

MLH1-MLH3 (nM)

e

MLH1-MLH3

(300 nM)Mn2+

–ATP +ATP0

20

40

60

80

scD

NA

cle

avag

e (

%)

iMLH1-MLH3

(200 nM)

Mn2+

No

co

fac

tor

+A

TP

+A

DP

+A

MP

-PN

P

+A

TP

-γ-S

0

10

20

30

40

scD

NA

cle

avag

e (

%)

h

–

No ATP + ATP

MLH1-MLH3

MLH1(EA)-MLH3

MLH1-MLH3(EA)

MLH1(EA)-

MLH3(EA)

++

++

– –– – –

– –––––

–++

++

– –– – –

– –––––

All

(300 nM)

Mn2+

0

20

40

60

scD

NA

cle

avag

e (

%)

0 20 40 75 150

MLH1-MLH3

(nM)

1 2 3 4 5 Lane

Mn2+

Substrate

Free label

HJ

0 20 40 75 150

MLH1-MLH3

(nM)

k

l

1 2 3 4 5 Lane1 2 3 4 5 Lane

0

20

40

60

80

100

DN

A b

ind

ing

(%

)

ML

H1

-ML

H3

ML

H1

(E3

4A

)-M

LH

3

No

pro

tein

ML

H1

(E3

4A

)-M

LH

3(E

28

A)

ML

H1

-ML

H3

(3N

D)

ML

H1

-ML

H3

(E2

8A

)Wells

All p

rote

ins (

25

0 n

M)

1 2 3 4 5 6HJ

No Mg2+No ATP

Protein-

bound DNA

n

Extended Data Figure 2

MSH5MSH4

a

Ma

rke

r

200

116

97

66

45

hM

utS

γ

31

kDa

No

pro

tein

25 50 75 100

MSH4-MSH5 (nM)

Free DNA

Protein-bound

DNA

Free DNA

Protein-bound

DNA

Free DNA

Protein-bound

DNA

Wells

Wells

Wells

dsDNA

HJ

D-Loop

Mg2+

b

ssDNA

Free DNA

Protein-bound

DNA

Wells

0 25 50 75 1000

20

40

60

80

100

hMSH4-hMSH5 (nM)

DN

A b

ind

ing

(%

)

dsDNA

HJ

D-Loop

ssDNA

Mg2+

0 50 100 150 2000

20

40

60

80

yMsh4-yMsh5 (nM)

DN

A b

ind

ing

(%

)

HJ

D-Loop

ssDNA

dsDNA

Y-Structure

e

0 50 100 150 2000

20

40

60

80

100HJ

D-Loop

ssDNAdsDNA

Y-Structure

yMsh4-yMsh5 (nM)

DN

A b

ind

ing

(%

)

fNo Mg2+

Mg2+

Mg2+

25 50 100 200

yMsh4-yMsh5

(nM)

No

pro

tein

ssDNA

Y-StructureFree DNA

Protein-bound

DNA

Wells

Free DNA

Wells

Free DNA

Wells

Free DNA

Wells

Free DNA

Wells

D-Loop

HJ

dsDNA

Protein-bound

DNA

Protein-bound

DNA

Protein-bound

DNA

1 2 3 4 5 Lane

Protein-bound

DNA

c

d

1 2 Lane

MSH4-MSH5 (nM)

1 2 3 4 5 Lane

a

75 0 75 100 0 100 hSγ (nM)

No Mg2+

0 75 75 0 100 100 hLγ (nM)

No

pro

tein

1 2 3 4 5 6 7 Lane

Free DNA

DNA-Lγ

DNA-Sγ-Lγ

Unstable

DNA-proteinspecies

HJ

b

No Mg2+

23.19.4 6.6 4.4

2.3 + 2.0

No

pro

tein

1 2 3 4 Lane

kbp

λ D

NA

/Hin

dII

I

Both (100 nM)

No

pro

tein

1 2 3 4 5 6 7 8 Lane

23.19.4 6.6 4.4

2.3 + 2.0λ D

NA

/Hin

dII

I

2 4 8 15 30 60

yKu70-yKu80 (nM)

kbp

Mg2+

0.6

ij

1 2 3 Lane

MLH1-MLH3 (400 nM)

MSH4-MSH5 (100 nM)

yMsh4-yMsh5 (100 nM)– +

+ + ++

–– –

MLH3

MSH4, MSH5

MLH1

anti-HIS

yMsh4, yMsh5

f

Extended Data Figure 3

yMsh4-yMsh5

yMlh1-yMlh3

1 2 3 Lane

yMsh5-his

yMlh1-hisWB:

anti-HIS

+ +–+ + –

ySγ (nM)

yLγ (nM)No

pro

tein

75 100 0 0 75 100

0 0 75 100 75 100

Free DNA

Free DNA

DNA-Lγ or

DNA-Sγ

DNA-Sγ-Lγ

HJ

dsDNA

e

HJ

75 0 75 100 0 100 hSγ (nM)

0 75 75 0 100 100 hLγ (nM)

No

pro

tein Mg2+

Free DNA

DNA-Lγ

DNA-Sγ-Lγ

Unstable

DNA-proteinspecies

Free DNA

DNA-LγDNA-Sγ-Lγ

Unstable

DNA-protein

speciesD-Loop

1 2 3 4 5 6 7 Lane

Free DNA

dsDNA

d

hSγNo

pro

tein

1 2 3 4 5 6 Lane

hLγ

ySγ

+

+ +++

++

–

– – –– –

– –

Free DNA

DNA-Lγ or

DNA-Sγ

DNA-Sγ-Lγ

HJ

All proteins

(100 nM)

Mg2+

1 2 3 4 5 6 7 Lane

MLH1-MLH3

MSH4-MSH5

200

Pull down by

anti-MLH1

116

97

66

45

kDa Ma

rke

r

MLH3

MSH4MSH5

MLH1

anti-hMLH1

+ + +– – –– + + – +–

g

h

Mg2+

0

1

2

3

DN

A b

ind

ing

sta

bilit

y (

art

ific

ial u

nit

s)

hLγ

hSγ–

Protein

concentration

(nM)75 75 100 100

+ + + +–+ +

c

Human MLH1-MLH3 = hLγ

Human MSH4-MSH5 = hSγ

Abbreviations

MSH4-MSH5

1 2 3 4 5 6 7 Lane

Input

MLH1-MLH3

–+ +

+

Free

DNA

Free

DNA

Protein-bound

DNAProtein-

bound

DNA

MSH4(G685A)-MSH5

MSH4-MSH5(G597A)

scDNA

nicked DNA

No

pro

tein

MLH1-MLH3

(50 nM)

MSH4-MSH5–––

++

+– ––

– ––

All (50 nM)

1 2 3 4 5 Lane

No

pro

tein

MLH1-MLH3

MLH1(E34A)-MLH3

MLH1-MLH3(E28A)

MLH1(E34A)-MLH3(E28A)

–++

++

++

++

––

–– – –

–– –

– – –

––––

–

–

–––

––

scDNA

nicked DNA

No

MSH4-MSH5

50 nM

MSH4-MSH5 All (50 nM)

1 2 3 4 5 6 7 8 9 Lane

a

fe

i

b

MS

H4

-MS

H5

MS

H4

(GA

)-M

SH

5

MS

H4

-MS

H5

(GA

)

Ma

rke

r

200

11697

66

45

31

22

14

kDa

MSH4 (or variant)

1 2 3 4 Lane

MSH5 (or variant)

g

0

20

40

60

80

100

DN

A b

ind

ing

(%

)

h

MSH4-MSH5

HJ

MSH4(G685A)-MSH5

MSH4-MSH5(G597A)

++

+

– –––

––

0

20

40

60

80

scD

NA

cle

avag

e (

%)

MLH1-MLH3 (nM)

MSH4-MSH5 (nM)

00

05

0

50

50

50

50

50

50

50

50

25

50

10

0

20

0 00 0

25

50

10

0

MLH1-hMLH3

titrationMSH4-MSH5

titration

Mn2+

Mn2+

Mn2+

1 2 3 4 5 6 7 8 Lane

No

pro

tein

MLH1-MLH3

MSH4-MSH5

yMlh1-yMlh3

yMsh4-yMsh5

+ + +– – – –– – ––+ + +– – – – – + +– – – + + – –

scDNA

nicked DNA

Mn2+

Extended Data Figure 4

0 10 20 30 40 50 60

0

10

20

30

40

50

60

Time (min)

scD

NA

cle

avag

e (

%)

hMLH1-hMLH3

hMLH1-hMLH3 hMSH4-hMSH5

Mn2+

c

MLH1-MLH3 (50 nM)

MSH4-MSH5 (50 nM)

Substrate

Anticipated

resolvase

product

Free label

50

100

nt

10

20

30

40

––

+++

–

1 2 3 4 Lane

l

50 nt

50 nt50 nt

50 ntMSH4-MSH5 (100 nM)

1 2 3 4 Lane

MLH1-MLH3 (100 nM)

No

pro

tein

Cruciform

DNA

+ ++ +

––

scDNA

nicked DNA

0 23 40 0 Cleavage (%)

Ma

rke

r

linear DNA

k

MLH1-MLH3, MSH4-MSH5

Mn2+Mn2+

1 2 3 Lane

No

pro

tein

MSH4-MSH5++

scDNA

nicked DNA

All proteins

(50 nM)

Mn2+

Wild

typ

e M

LH

1-M

LH

3

ML

H1

-ML

H3

(3

ND

)

j

Mn2+

0

20

40

60

80

scD

NA

cle

avag

e (

%)

yMsh4-yMsh5

yMlh1-yMlh3

MSH4-MSH5

MLH1-MLH3

–

No

pro

tein

––– + + +

+ +

– –

–––

––+

+–– +

–

–

–

MLH1-MLH3

Both

(50 nM)0

20

40

60

scD

NA

cle

avag

e (

%)

MLH1-MLH3

MSH4-MSH5

– ++– – +

– ++– – +

– ++– – +

– ++– – +

– ++– – +

No c

ofa

ctor

ATP

AD

P

AM

P-P

MP

ATP

-γ-S

d

Mn2+

c

200

1169766

45

31

22

kDa Ma

rke

r

hM

utSb

MSH3MSH2

a

Ma

rke

r

hM

utSa

200

1169766

45

31

22

kDa

MSH6

MSH2

b

scDNA

nicked DNA

MSH2-MSH6 (hMutSa)

MSH2-MSH3 (hMutSb)

MSH4-MSH5 (hMutSγ)–––

–– –

––

––

–

–

++

+

– ++

+–

––

––

MLH1-MLH3

(50 nM)

No

pro

tein

All (50 nM)

1 2 3 4 5 6 7 8 Lane

Mn2+

0

20

40

60

80

scD

NA

cle

avag

e (

%)

Extended Data Figure 5

1 2 Lane

1 2 Lane

No

pro

tein

0

20

40

60scD

NA

cle

avag

e (

%)

+0

20

40

60

80

DN

A b

ind

ing

(%

)

No Mg2+

With Mg2+, with ATP

EXO1(D173A)

MLH1-MLH3

MSH4-MSH5

All (75 nM)

+ ++ +

+ +

+ ++ ++ +

––––––

d

Extended Data Figure 6

0

10

20

30

40

50

scD

NA

cle

avag

e (

%)

EXO1(D173A)

MLH1-MLH3

Both (100 nM)

+

No

pro

tein

+++

––

––

1 2 3 4 Lane

c

scDNA

nicked DNA

Mn2+

scDNA

nicked DNA

No

EXO1(DA)

EXO1(D173A)

(50 nM)

MLH1-MLH3

MSH4-MSH4––

+ + + ++ +––

––

+––

Both

(30 nM)

1 2 3 4 5 6 7 8 Lane

200

116

97

66

45

31

22

14

kDa Ma

rke

r

EX

O1

(D1

73

A)

a b

1 2 Lane

Mn2+

+ EXO1(D173A)

yExo1(D173A)+–

––––

–

scDNA

nicked DNA

MLH1-MLH3 MSH4-MSH5

Both (30 nM)

No

pro

tein

1 2 3 4 Lane

Both

(50 nM)

Mn2+

0

10

20

30

40

50

scD

NA

cle

avag

e (

%)

e

scDNA

nicked DNA

MLH1-MLH3

(20 nM)

MSH2-MSH3 (20 nM)

EXO1(D173A) (50 nM)+ +

+–

––

0

10

20

30

40

50

scD

NA

cle

avag

e (

%)

No

pro

tein

––

1 2 3 4 Lane

f

Mn2+

0

10

20

30

40

50

scD

NA

cle

avag

e (

%)

EXO1(D173A) (50 nM)

MSH4-MSH5

(50 nM)

No

pro

tein

MLH1-MLH3 (20 nM)

RFC (50 nM)

PCNA (100 nM)

–––

––

+––

+––

++–

+–+

+++

+++

–+

+++

+ –

1 2 3 4 5 6 7 8 9 10 Lane

scDNA

nicked DNA

Extended Data Figure 7

Mn2+

Mg2+

No

pro

tein

MLH1-MLH3 (50 nM)

MSH4-MSH5 (50 nM)

EXO1(D173A) (50 nM)

RFC (50 nM)

PCNA (100 nM)

ATP

ADP––

––+

+

1 2 3 4 Lane

scDNA

nicked DNA

Mg2+

a

kDa

yRfc1

250

Ma

rke

r

Ye

ast

RF

C

Hu

ma

n R

FC

Hu

ma

n P

CN

A

150

100

70

50

40

30

20

PCNA

RFC1

RFC2-5

1 2 3 4 LaneMLH1-MLH3

MSH4-MSH5

EXO1(D173A)

Yeast RFC

Yeast PCNA

Human PCNA

No

pro

tein

–

–––––

+

+

+

++

+

++

+

++

ML

H1

-ML

H3

+ MSH4-MSH5

+ EXO1(DA)

+ RFC-PCNA

MLH1-MLH3(3ND)+++ + +

––––––

–

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Lane

–––

–––

–––

–––

–––

–––

–––

–

+

+ +–

––

– ––

– –––– +

+++

++ +

+

+ + ++ + +

– – – – – –

–––––

+– + +– –

––

–+–+ +

+–

scDNA

nicked DNA

nicked DNA

scDNA

Mg2+

All proteins (50 nM)

except

PCNA (100 nM)

0

10

20

30

40

50

60

scD

NA

cle

avag

e (

%)

0

10

20

30

40

50

60

scD

NA

cle

avag

e (

%)

MLH1-MLH3

MSH4-MSH5

EXO1(D173A)

Human PCNA

Yeast RFC

Human RFC

No

pro

tein

+ + ++ + ++ + ++ + +

++

––––

––

––

––

All proteins

(50 nM)

except

PCNA

(100 nM)

Mg2+

scDNA

nicked DNA

1 2 3 4 Lane

b c

MLH1-MLH3, MSH4-MSH5

EXO1(D173A), RFC-PCNA

No

MS

H4

-MS

H5

No

EX

O1

(DA

)

0

10

20

30

40

50

scD

NA

cle

avag

e (

%)

Mg2+

d

e

f

g

MLH1-MLH3

MSH4-MSH5

EXO1(D173A)

RFC-PCNA+++– – –

+ + + + + ++ +

+ +– –– –

– – – –

Mg2+

0

10

20

30

scD

NA

cle

avag

e (

%)

h i

No

pro

tein

No

su

bstitu

tio

n

ML

H1

(EA

)-M

LH

3

ML

H1

-ML

H3

(EA

)

ML

H1

(EA

)-M

LH

3(E

A)

MS

H4

(GA

)-M

SH

5

MS

H4

-MS

H5

(GA

)

Mu

tan

t va

ria

nts

scDNA

nicked DNA

WT MLH1-MLH3

WT MSH4-MSH5

EXO1(D173A)

RFC-PCNA

––––

++++

– – –– –+ +

+++

+ +++

++

++

++

k

No

pro

tein

No

pro

tein MLH1-MLH3

–– ++ ++ +

+

1 2 3 4 5 6 Lane

EXO1(D173A)

Mg2+

+ ++ +

+ ++ +

MSH4-MSH5

RFC-PCNA

Mg2+

EXO1(D173A)

RFC-PCNA

MSH4-MSH5

MLH1-MLH3

j

MLH1-MLH3

MSH4-MSH5

EXO1(D173A)

RFC-PCNA

––––

+ ++ ++ +

+–

––––

+ ++ ++ +

+–

Supercoiled

DNA

Relaxed

DNA

sc/rel DNA

nicked DNA

Mg2+

0 20 40 600

10

20

30

40

50

60

Time (min)

scD

NA

cle

avag

e (

%)

MLH1-MLH3

MLH1-MLH3, MSH4-MSH5

MLH1-MLH3, MSH4-MSH5, EXO1(DA)

MLH1-MLH3, MSH4-MSH5, EXO1(DA), RFC-PCNA

HJ

D-Loop

0

20

40

60

scD

NA

cle

avag

e (

%)

b

Mn2+

No

pro

tein

EXO1(DA)

EXO1(DA)P

––

++–

– ––

(30 nM)

MLH1-MLH3

MSH4-MSH5

Both (30 nM)

1 2 3 4 Lane

0

5

10

15

20

25

scD

NA

cle

avag

e (

%)

scDNA

nicked DNA

d

scD

NA

cle

avag

e (

%)

MLH1P-

MLH3

MLH1-

MLH3P

MSH4-MSH5

RFC-PCNA

– + + + + + +– – + + +– –

scDNA

nicked DNA

Mg2+

MLH1-

MLH3

1 2 3 4 5 6 7 Lane

c

scD

NA

cle

avag

e (

%)

Extended Data Figure 8a

MLH1-MLH3 (30 nM)

MLH1P-MLH3P (30 nM)No

pro

tein

++–

–

HJ

Free DNA

Protein-bound

DNA

MSH4-MSH50

10

20

30

40

scD

NA

cle

avag

e (

%)

scDNA

nicked DNA

No

pro

tein

No

pro

tein

ML

H1

P-M

LH

3

ML

H1

-ML

H3

P

ML

H1

-ML

H3

ML

H1

P-M

LH

3

ML

H1

-ML

H3

P

ML

H1

-ML

H3

ML

H1

P-M

LH

3P

(200 nM) (30 nM)

+ + + +– – – –

MLH1-MLH3 variant

1 2 3 4 5 6 7 8 9 Lane

Mn2+

1.8-fold 1.5-fold 1.4-fold

0

10

20

30

40

scD

NA

cle

avag

e (

%)

scDNA

nicked DNA

No

pro

tein

MLH1-MLH3

MSH4-MSH5Both (50 nM)

RFC-PCNA– – – ++ –

EXO1

(DA)

EXO1

(DA)P

1 2 3 4 5 6 Lane

e

(30 nM)

Both

(30 nM)

scD

NA

cle

avag

e (

%)

scD

NA

cle

avag

e (

%)

scD

NA

cle

avag

e (

%)

scD

NA

cle

avag

e (

%)

+ + + +– – – –EXO1(DA)

– + + + + + + EXO1(DA)

No

pro

tein

Mg2+

All

proteins (50 nM)

except

PCNA (100 nM)

––

Re

lati

ve

en

ric

hm

en

t o

f R

fc1

0

1

2

3

4

5

MLH3 mlh3∆

neg. control

HIS4LEU2

BUD23

GAT1

mlh1∆ mlh3∆

Complementation

wild

typ

e

No

ne

Mlh

1

Mlh

1P

No

ne

Mlh

3

Mlh

3P

Extended Data Figure 9

a

e

0

20

40

60

80

100

Sp

ore

via

bilit

y (

%)

Rfc1-TAP

Input

Rfc1-TAP

Mlh1-HA

Rfc1

Mlh1-HA

–++ +

–+++

TEV

eluate

Mlh3-Myc

Input

Rfc1-TAP

Mlh3-Myc

–++ +

–+++

TEV

eluate

Rfc1

Rfc1-TAP

ML

H1

ML

H1

-FL

AG

mlh

1P-F

LA

G

ML

H3

ML

H3

-FL

AG

mlh

3P-F

LA

G

Mlh1-FLAG

Crm1

ML

H3

ML

H3

-FL

AG

mlh

3P-F

LA

G

Mlh3-FLAG

Crm1

***

Mlh3-FLAG*

Crm1

b c

d

Extended Data Figure 10