34
1 A Spring-Loaded Mechanism Governs the Clamp-Like 1 Dynamics of the Skp Chaperone 2 3 Daniel A. Holdbrook 1 , Björn M. Burmann 2 , Roland G. Huber 1 , Maxim V. 4 Petoukhov 4 , Dmitri I. Svergun 4 , Sebastian Hiller 2,* , Peter J. Bond 1,3,* 5 6 7 1 Bioinformatics Institute (A*STAR), 30 Biopolis Str, #07-01 Matrix, 138671 8 Singapore 9 2 Biozentrum, University of Basel, Klingelbergstrasse 70, 4056 Basel, Switzerland 10 3 Department of Biological Sciences, National University of Singapore, 14 Science 11 Drive 4, 117543 Singapore 12 4 European Molecular Biology Laboratory Hamburg, c/o DESY, Notkestrasse 85, 13 22607 Hamburg, Germany 14 15 16 *Correspondence 17 Dr. Peter J. Bond 18 ([email protected]) 19 Prof. Dr. Sebastian Hiller 20 ([email protected]) 21 22 23

A Spring-Loaded Mechanism Governs the Clamp-Like Dynamics ... · 2 24 Abstract 25 The trimeric periplasmic holdase chaperone Skp binds and stabilizes unfolded 26 outer membrane proteins

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

1

ASpring-LoadedMechanismGovernstheClamp-Like1

DynamicsoftheSkpChaperone2

3

DanielA.Holdbrook1,BjörnM.Burmann2,RolandG.Huber1,MaximV.4

Petoukhov4,DmitriI.Svergun4,SebastianHiller2,*,PeterJ.Bond1,3,*5

6

7

1BioinformaticsInstitute(A*STAR),30BiopolisStr,#07-01Matrix,1386718

Singapore9

2Biozentrum,UniversityofBasel,Klingelbergstrasse70,4056Basel,Switzerland10

3DepartmentofBiologicalSciences,NationalUniversityofSingapore,14Science11

Drive4,117543Singapore12

4EuropeanMolecularBiologyLaboratoryHamburg,c/oDESY,Notkestrasse85,13

22607Hamburg,Germany14

15

16

*Correspondence17

Dr.PeterJ.Bond18

([email protected])19

Prof.Dr.SebastianHiller20

([email protected])21

22

23

2

Abstract24

ThetrimericperiplasmicholdasechaperoneSkpbindsandstabilizesunfolded25

outermembraneproteins(OMPs)aspartofbacterialOMPbiogenesis.Skpbinds26

clientproteinsinitscentralcavity,therebyreducingitsbackbonedynamics,but27

itisunknownwhichmolecularmechanismsgovernSkpdynamicsandhowthe28

chaperoneadaptstodifferentlysizedclients.Here,weemployacombinationof29

microsecond-timescalemoleculardynamics(MD)simulation,small-angleX-ray30

scattering(SAXS)andNMRspectroscopytorevealthatSkpisremarkably31

flexible,andfeaturesamolecularspring-loadedmechanisminits“tentacle”arms32

thatenablesswitchingbetweentwodistinctconformationsonsub-millisecond33

timescales.Theconformationalswitchisexecutedaroundaconservedpivot34

elementwithinthecoiledcoilstructuresofthetentacles,allowingexpansionof35

thecavityandthusaccommodationofdifferentlysizedclients.Thespring-loaded36

mechanismshowshowachaperonecanefficientlymodulateitsstructureand37

functioninanATP-independentmanner.38

39

40

3

Theoutermembrane(OM)ofGram-negativebacteriafunctionsasageneral41

selectivitybarrier.Itprovidesthecellwithprotectionfrompotentiallyharmful42

agentsintheenvironment,whileallowingusefulmoleculestoenter.Inorderto43

fulfillthisfunction,transmembraneβ-barrelOMPsactasselectiveandnon-44

selectiveporins.Theirabsenceseverelyinhibitsnutrientcollectionandtherefore45

growth.OMPsaresynthesizedinthecytosolofthebacterialcell,andassuch46

theyfollowacomplexbiogenesispathway[1].UnfoldedOMPstraverseboththe47

hydrophobicinnermembraneandthehydrophilicperiplasmbeforetheyare48

foldedandintegratedintotheOM[2–4].Duetotheirarchitectureas49

transmembraneproteins,OMPsarepronetomisfoldingandaggregationinthe50

aqueous,butcrowdedperiplasmiccompartment.Undesirableintra-andinter-51

molecularinteractionsintheperiplasmaretypicallypreventedbyanetworkof52

periplasmicchaperonesincludingSurA,DegPandSkp[5,6].Theseaidinthe53

propertransport,foldingandinsertionofOMPsintotheOM.Skpbindsto54

unfoldedOMPsastheyemergefromtheSectransloconattheinnermembrane.55

ThebindingofSkppreventsprematurefoldingofOMPs,andpromotestheir56

releasefromtheSeccomplexoncetranslocationiscomplete.SkpholdstheOMP57

inafoldingcompetentstate[7],priortoarrivalandreleaseattheOM[8].58

59

Skpisatrimerwitha“jellyfish”-likearchitecture[9,10].Asmallβ-sheet“head”60

domainisthemajorsiteofassociationbetweenthemonomers,whilethreelong,61

hairpin-shapedα-helical“tentacles”or“arms”definetheouterboundariesofa62

centralcavity,characteristicofaclamp-likebindingsitetypicallyobservedin63

chaperones[11].Unlikethestructureofprefoldin[12],aeukaryoticchaperone64

withrelatedarchitecture,inthestructuresofSkptheα-helicalextensionsmay65

4

interactwithoneanotheratthetips.Eachtentacleiscomposedofashortα-helix66

(α1)thatleadsfromtheβ-sheetheadintotwoextendedantiparallelα-helices67

(α2andα3).TheOMPsubstratesofSkparediverseinshapeandsize.The68

smallestoftheseOMPs,suchasOmpA,havean8-strandedβ-barrelwith69

diameterof3.0nm,whereasthelargest,LptD,is24-stranded[13].Studiesof70

Skp–OmpXconformationanddynamicshaveshownhowSkpcanbind8-71

strandedOmpXandtOmpA[14],butitisyetunclearhowSkpisabletoadapt72

sufficientlytoaccommodatedifferentlargersubstrates.NMRstudieshave73

identifiedahingeregionaroundVal42/Phe50betweenα-helix1and2thatmay74

allowthetipsofthetentaclestomoveawayfromoneanother,thereby75

expandingthesizeofthecentralcavitytoallowlargersubstratestoenter[14].76

Suchsubstantialflexibilityintheα-helicalregionsofSkpisalsoconsistentwith77

absenceofelectrondensityforpartsofthetentaclesinbothoftheX-ray78

structures,andwithstructuraldisparitiesobservedtowardsthetipswhen79

comparingtheresolvedsubunits.80

81

Here,weutilizeextended,microsecond-timescale,atomic-resolutionMD82

simulationsofsubstrate-free(apo)trimericSkptoexploreitsconformational83

landscape,andcomparetheresultswithdatafromsmall-angleX-rayscattering84

(SAXS)andamide–amidedistancesobtainedfromNMRspectroscopy.Unbiased85

simulationsamplingisusedtodemonstratethattheSkptrimerisextremely86

flexible,revealingthefullrangeofpossiblemotionofthetentacles.Basedon87

theseobservations,weproposehypotheticalclosed-andopen-statemodelsand88

identifyaswitchingmechanismthatleadstolargevariationsinthevolumeofthe89

Skpcentralcavityviatheexchangeofahelicalkinkbetweenhelixα2andα3.90

5

ThisallowsustodefinethelimitsofSkp’sconformationalmobility,which91

explainsitscapacityforbindingsubstratesofvariablesize.Subsequentfittingof92

simulatedconformationalensemblestodatafromSAXSmeasurementssupports93

thenotionthatapoSkpexistsinadynamicequilibriumbetweenopenandclosed94

statesinsolution.Finally,comparisonofthesimulationswithNOESY95

experimentaldataconfirmsthatapoconformationsobservedspectroscopically96

arewellrepresentedinthesimulationtrajectories,andthattheyaredistinct97

fromthesubstrate-boundformofSkp.98

99

Results100

OpeningoftheSkpCavityviaSeparationoftheTentacles101

TheX-raystructureofSkp(PDBID:1SG2)representsa“closed”stateofthe102

protein,withatip-to-tipdistanceof<0.78nmmeasuredbetweentheCα103

carbonsofAla76[9].Complementedbyapartiallymodeledthirdsubunit,the104

1SG2structureservedastheinitialseedfor15independentMDsimulation105

replicasofatleast100nsinlength.Thesesimulationsservedtosearchfor106

possibleopenstates.Startingfromthe“closed”state,dissociationofthehelical107

tips,asidentifiedbyatip-to-tipdistance>1.4nm,wasobservedinthreeofthe108

15simulations,inamannerthatwasinsensitivetoinitialconditions[8].These109

separationsofthetwosubunitstowardsan“open”stateoccurredinallthree110

casesafter>80nsofsimulation.111

112

Followingtheobservationthatthetentacle-likearmswereabletoseparatefrom113

oneanotheronthenanosecondtimescale,thedynamicsofSkpwere114

investigatedfurtherintwosignificantlyextendedsimulations,of1μsinlength115

6

each,beginninginthe“open”conformation.Additionally,wemeasuredSAXS116

profilesofasampleofhighlypureapoSkpatasynchrotronbeamline.Duringthe117

two1μsMDsimulations,thetipswereobservedtospontaneouslyre-associate118

anddisassociateatthreedifferenttimepoints.There-associationofthetipswas119

apparentinthemeasuredradiusofgyration(Rgyr)ofSkp(Figure1A),which120

droppedbelowtheRgyroftheX-raystructure(3.0nm)at~250nsinone121

simulationandat~80nsand~850nsintheother.TheaverageRgyrof3.28nm122

observedinthe1μsMDsimulations(Figure1B)agreeswellwithexperimentally123

measuredRgyrof3.3nmfortheapo-Skptrimerinsolution,determined124

previouslybyneutronscattering[15].ItthusalsoagreeswellwiththeRgyrof3.6125

nmdeterminedbySAXS,consideringthatthescatteringofthehydrationshell126

likelyadds~0.6nmtothetriaxialdimensionsoftheproteininSAXS127

measurements[16,17].128

129

TheexperimentaldeterminationsofRgyrbySAXSandneutronscatteringare130

ensembleaveragesandthusmaskindividual,short-livedconformationalstates.131

Indeed,thestructuresdeterminedbyMDsimulationfeatureawiderangeof132

individualRgyrvalues,rangingbetween2.9and3.7nm(Fig.1A).Thereby,the133

largervaluesofRgyrcorrespondedtoconformationsofSkpwherethehelical134

armsareprojectedawayfromthetrimericaxisofsymmetry,insomecases135

dramaticallyexposingthelargecentralcavity(Figure1C–I).Previous136

simulationsthatconstrainedSkptoexploreaparticularRgyrhintedatsuch137

expansionofthecentralcavity[15].Ontheotherhand,thesmallervaluesofthe138

RgyrcorrespondtoconformationalstatesofSkpwherethetipsofallthreehelical139

armsarere-associatedwithoneanother(Figure1C–II).Inthedistributionoftip-140

7

to-tipdistances,theseclosedconformationspopulateadistinctstateat2nm,141

wellseparatedfromtheconformationalensemblecontinuumofthepartiallyand142

fullyopenstate(Figure1D).WhentheSkparmsareseparated,thetip-to-tip143

distanceistypicallybetween3to7nm(Figure1D),withthedistributionof144

distancespeakingat6nm.Themaximaldistancebetweenthetipsoftwo145

subunitsencounteredinfreesimulationwas7.2nm.Overall,thesedatathus146

revealanextraordinarydegreeofflexibilityoftheSkparmsintheapostate,in147

agreementwithsolutionNMRdynamicsmeasurements[14].Thesemotions148

averageoutonthesub-mstimescale,resultinginNMR-spectroscopically149

equivalentresonancesfortheentiretrimer.150

151

“Spring-Loaded”DynamicsCharacterizetheHelicalArms152

Ithasbeensuggestedthatapivotelementexistsaroundahighlyconserved153

phenylalanineresidue(Phe50)inhelixα2,allowingforaconformationalchange154

andincreasingthevolumeofthecentralcavity[8].Inagreementwiththis155

hypothesis,multipleobservationsofaspontaneousconformationalchangeand156

rotationaroundPhe50wereobservedduringsimulation.Thischangeinvolved157

theexchangeofahelicalkinkfrominitiallybenthelixα2(Figure2A-I)to158

initiallystraighthelixα3(Figure2A-II),andresultedinlateralprojectionofthe159

tipofthehelicalarmfromthethree-foldaxisofsymmetry.Anexampleofthis160

transitionisillustratedinSupplementaryMovie1.Thekinkinhelixα3wasmost161

oftenaccommodatedfromAla100toAsp105.Intotal,theconformational162

transitionatthehelicalkinkwasobservedtwelvetimesduringthesimulation163

trajectories(FiguresS1),butneveroccurredsimultaneouslyintwoorthree164

subunits.Inallobservedcases,theexchangeofthekinkfromhelixα2toα3165

8

exhibitedall-or-nothingmechanics,andhadamaximumlifetimeof~60ns166

(FigureS1).Thekinkexchangehadasimultaneouseffectonthephysical167

dimensionsoftheSkptrimer,withadirectlycoupledincreaseintheRgyrfrom168

0.15to0.30nm.Thisopenedthecavityandmaythuswellplayarolein169

substratebinding.170

171

DefiningtheMaximalCapacityoftheSkpCavityintheOpenState172

Toisolatethebiologicallyrelevant,dominantconcertedmotionsofSkp,we173

removedthehigh-frequencybackgroundnoisebyperformingprincipal174

componentanalysis(PCA)onthecombined(2x1μs)MDtrajectoriesofthe175

entire,trimericSkpassembly(Figure3).Theexchangeofthehelicalkink176

betweenhelixα2andα3didnotappearinthelowestfrequencymodes(Figure177

3A,FigureS2),indicatingthatthislarge,switchableinter-subunitmotionofan178

armfromSkpoccursindependentlyoftheotherlocalsubunitdynamics.This179

findingimpliesthatallostericcommunicationbetweenSkpsubunitsisabsent.In180

lightofthis,wethereforeinvestigatedtheinternalarmdomainmotionsby181

performingPCAonan“artificial”6µstrajectorycomposedoftheindividual182

subunittrajectoriesfromeachsimulation(i.e.3subunitsx2simulationsx1μs183

each).Principalcomponent1(PC1),thedominantmotionoftheSkpsubunit184

(Figure3Bi),accountedforoverhalfofthetotalstructuralvariance.This185

componentinvolvedoutwardprojectionofthetipsofthetentacles,awayfrom186

thethree-foldsymmetryaxisofSkp,withaconcurrentexchangeofthehelical187

kinkfromhelixα2toα3,asdescribedabove,whilstmaintainingastablehead188

domainconformation(FigureS3).Thespring-loadedmovementisthusakey189

elementofSkptentacledynamics.190

9

191

Inordertoestimatethepossiblerangeinsizeofthecentralcavity,two192

structuralstates,termed“extremeclosed”and“extremeopen”,wereconstructed193

byapplyingC3symmetrytothetwoextremestructuresofPC1(Figure4A).In194

bothcases,aleast-squares-fittotheβ-sheetheaddomainoftheX-raystructure195

wasperformedinordertopositioneachsubunitintheextrememodels.The196

transitionbetweenthesestatesisillustratedinSupplementaryMovie2.While197

thetipsoftheheliceswereobservedtoextendupto7.2nmfromoneanother198

duringfreesimulation,thesymmetric“extremeopen”modelfeaturesadistance199

betweentipsof12nm.ThedimensionsoftheresultingcavityenclosedbySkp200

wereestimatedbyexpandingavirtualsphereatintervalsalongtheaxisof201

symmetry.Inthe“extremeclosed”model,amaximumradiusof1.75nmforthe202

expandingspherewasachievedatadepthof~2.0nmbeneaththeheaddomain203

(Figure4B).Thispositioncorrespondspreciselytotheheightofthekinkinhelix204

α2.Fromthispointonwards,theradiustapers,decreasingtowardsthetipsof205

thehelicalarms.Inthe“extremeopen”model,theupperpartofthecentral206

cavity,immediatelybelowtheheaddomain,hassimilardimensionstotheclosed207

model.However,theradiusofthespherecontinuestoincreaselinearly,208

reachingamaximumradiusof3.2nmata4.0nmdistancefromtheheaddomain.209

UsingNMR-baseddistancesmeasurements,itwaspreviouslyshownthatwhen210

boundtoSkp,8-strandedOmpX,oneofthesmallestpossiblesubstrates,adopts211

toafirstorderapproximationasphericalensembleofconformerswitharadius212

of2.1nm.Thisfluidglobulestatecanalreadybenearlyaccommodatedwithin213

the“extremeclosed”stateofapoSkp,whereaslargersubstratesmightbe214

accommodatedintheSkpcavitybygradualopeningofthearmstowardsthe215

10

“extremeopen”state.Extrapolatingundertheassumptionofequalmassdensity216

fromunfoldedOmpXtothelarge22-strandedsubstrateFhuA,thelatter217

polypeptidewouldadoptasphericalvolumewith3.5nmradius.Thislarge218

substratemighteventhereforebeaccommodatedatthelowerendofthecavity219

intheextremeopenstate;conceivably,anadditionalSkptrimermayalsobe220

recruitedforbinding[7].221

222

SAXSRevealsaDynamicEnsembleofOpenandClosedStates223

ForadescriptionoftheSkpapostate,theexperimentallydeterminedSAXS224

intensitywascomparedwiththeoreticalcalculations(Figure5).Becausethe225

individualconformersofthesimulationtrajectoriesfeaturesubstantial226

variabilityintheconformationsadopted,theircalculatedintensitiesvary(see227

FigureS4fortherangeoftheoreticalscatteringintensitiesandindicative228

structures),andnointensitycomputedfromasinglestructuredescribedthe229

experimentalSAXSdatawithinexperimentalerror.Likewise,thescattering230

computedfromtheX-raycrystalstructuredidnotagreewiththeSAXSdata,as231

evidencedbyadiscrepancyinthegoodnessoffit,withχ2=5.9.Abetter232

agreementwiththeexperimentaldatawasobtainedbyallowingformixturesof233

individualconformers,i.e.linearcombinationsofthecomputedpatterns.The234

ensembleoptimizationmethod(EOM)[18,19]usesageneticalgorithmto235

recombinemodelsfromapoolofstructures,untilanoptimalfittothe236

experimentalSAXScurveisobtained.Inordertoincludeasmanymaximally237

openstatesinthedataset,thepoolofSkpstructuresfromthesimulation238

trajectorieswasenhancedwithanequalnumberofhypotheticalsymmetrical239

structures,createdbyapplyingC3symmetrytoarandomselectionofindividual240

11

subunitconformations.ThecombinedpoolcontainedstructuresthathadRgyr241

valuesintherange2.75to4.25nm(Figure5A,greencurve).Thesestructures242

generatedafittotheexperimentalcurvewithaχ2=1.0(Figure5B).243

Representativestructuresobtainedfromthefitweresimilartothe“extreme244

open”and“extremeclosed”models(Figure5i,Figure5ii).Thus,theSAXS245

experimentsdirectlyindicatethatapoSkpinsolutionadoptsaconformational246

ensemble,composedofvariablyopenedandclosedstructuresandtheir247

intermediatestates.248

249

Whilethefittotheexperimentaldatawasimprovedconsiderablybyconsidering250

anensembleofstructures,thereremained,however,adeviationatthelow251

anglesoftheSAXSmeasurements,indicatingadearthoflargerstructuresinthe252

dataset.Inordertoincreasethenumberofstructuresinthepoolwithalarger253

Rgyr,anew,independentrandompool(Figure5A,redcurve)wasgenerated254

usingRANCH(RANdomCHain)withinEOM[18,19],allowingthepositionsofthe255

helical“tentacles”tovarywithrespecttothefixed“head”domain.EOMwas256

againthenemployedtoselectanoptimizedsetofmodelsconsistentwiththe257

data(Figure5A,pink/cyancurve).ThisapproachyieldedafittotheSAXSprofile258

withaχ2=1.5(Figure5B),wherebytherepresentativestructuresalsoincluded259

openandclosedconformations(Figure5iii,Figure5iv,Figure5v)further260

confirmingthestructuralvariabilityofSkpinsolution.Inthiscase,theselection261

frequencyofstructuresindicatedabimodaldistribution(Figure5A,pink/cyan262

curve).ThefirstpeakcorrespondedtoaRgyrof~3.2nm,similartothatinthe263

MD-generatedpool(Figure5A,greencurve).TheRgyrforthesecondpeakranged264

12

between3.8and4.1nm(Figure5A),andwasthuscomparabletotheRgyrof4.1265

nmobtainedforthe“extremeopen”modelofSkpgeneratedviaPCA.266

267

NOESYExperimentsConfirmDistinctDynamicsofApoSkp268

ThestructuralconfigurationofSkpinitsapostatewasinvestigatedfurtherby269

thecalculationofasubsetofinter-backboneamidehydrogendistancesforeach270

subunitconfigurationobservedinthesimulationtrajectories(Figure6A,Table271

S1).Thesedistanceswerecomparedtodistanceassignmentsderivedfrom272

NOESYspectroscopyforbothapo-SkpandSkpboundtoanOmpXsubstrate273

(Figure6B,C).Anarrow,unimodaldistributionwasobtainedforthemean274

deviationofbackbonedistancesofthetentacledomainacrosssimulatedSkp275

comparedtotheapoSkpNOEs,peakingat<0.8Å.Incontrast,thedeviationfrom276

theOmpX-boundSkpNOEdataexhibitedabroaderbimodaldistribution,which277

onlypartiallyoverlappedwiththeapodata,andextendedfurtherout,upto~1.3278

Å.MeasurementoftheequivalentdistancesfortheX-raycrystalstructure279

revealedthatthethreesubunitsofSkpdeviatetovaryingdegreesfromboththe280

apo-andsubstrate-boundNOEdata.However,inallcases,thecrystallographic281

deviationsareincreasedcomparedtothoseforthemostfrequentconformations282

observedinthesimulations.Thus,thesimulatedensemble,characterizedbya283

rangeofopenandclosedstates,isbestrepresentedbyligand-free,apoSkpin284

solution.ItsconformationislikelymodulatedbythepresenceofOmpX,possibly285

biasingtheensembletowardscollapsedstatesthatprotectthesubstrate.286

287

288

13

Discussion289

Viaacombinationofsimulationandexperiment,wehaveshownherethat290

ligand-free,apoSkpexistsasadynamicensembleofmultipleconformational291

states,withvariableaccesstothecentralcavity.Interconversionbetweenopen292

andclosedstatesisfast,withdramaticchangesalreadyonthemicrosecond293

timescale,asreflectedintheobservedfrequenciesoftip-to-tipseparation,294

helicalexchangebetweenhelixα2andα3,andconcomitantchangesinprotein295

Rgyrandcavityvolume.ThisfindingisinfullagreementwithpreviousNMR296

measurementsofSkpbackbonedynamics,showingthatcomplete297

conformationalaveragingisobtainedinatmost1ms[14].Thesimulationsthus298

revealanovelmechanismforchaperonecavityexpansion,involvingspring-299

loadeddynamicsofthetentacles,wherebyahelicalkinkisexchangedbetween300

helicesα2andα3.Thismechanismservestoallowthetipsofeachsubunitto301

projectawayfromthecentralaxisofsymmetry,whilstensuringminimal302

disruptionoftheproteinstructuralfold.Itwillbeofgreatinteresttoassess303

whetherarchitecturallysimilar(butevolutionarilydistinct)eukaryotic304

chaperonessuchasTim9/10orprefoldinhaverelatedfunctionalmechanisms305

[20,21].306

307

TheabilityofSkptoadapttoavarietyofdifferentlysizedsubstratesiskeytoits308

roleinperiplasmicOMPtrafficking.Thisfunctionalrequirementismetby309

variabilityintheshapecreatedbythehelicaltentacles.Thereby,thephysical310

limitoftheexpansionoftheSkpcavityisfargreaterinsizethandisplayedin311

availableX-raycrystallographicstructures,andthussufficienttoaccommodate312

Skpsubstratesabove8strands.Skpsubstantiallyaltersitsconformationupon313

14

substratebindingbyreducingbackbonedynamicsandflexibilityinthepivot314

region[14].Thesubstrate-boundstateofSkpwaspreviouslyfoundtobe315

structurallydistinctfromtheconformationsexploredbyapoSkp,indicatinga316

substrate-inducedconformationalchange.Thelarge-scalemovementsofthe317

subunitsrelativetooneanotheroccurindependentlyofboththeexchangeofthe318

helicalkinkandofoneanother.Suchloose,dynamicassociationoftheSkpcavity319

withsubstrateensuresthattheOMPisheldinafoldingcompetentstate,without320

over-stabilizingnon-nativecontactsoftheOMPbackbone[7].TheabilityofSkp321

toexpanditscavityinthiswaylikelyhasfunctionalimpactbothinthecapture322

andreleaseofdiverseOMPsubstrates,andmightalsoenableaccesstothe323

centralcavityfortheOMP-insertingBAMcomplex[22–24].324

325

326

15

Acknowledgements327

ThisworkwasfundedbytheSwissNationalScienceFoundationandA*STAR.328

ExperimentalSAXSintensityofSkpandthestructuralmodelsgeneratedbyEOM329

aredepositedinSASBDBdatabase(entriesarebeingassigned).330

331

AuthorContributions332

DAHandPJBperformedthecomputationalexperiments;BMB,MVP,andDISand333

SHperformedtheSAXSandNMRexperiments;allauthorsanalyzedthedata;PJB,334

DAH,andSHwrotethepaperwiththecontributionsfromBMB,RGH,MXP,and335

DIS;PJBandSHdesignedthestudy.336

337

Competingfinancialinterests.Theauthorsdeclarenocompetingfinancial338

interests.339

340

Materials&Correspondence.Correspondenceandrequestsformaterialsmay341

bedirectedtoProf.Dr.SebastianHiller([email protected])andDr.342

PeterJ.Bond([email protected]).343

344

345

16

Methods346

SimulationSystemConfiguration347

AcrystalstructureofSkp(PDBID:1SG2)[9]wasusedasaninitialconfiguration.348

TheresiduenumberingconventionusedhereinisbasedonUniProtentry349

sp|P0AEU7|21-161,wherethefirstresidue,Ala21,isthestartofthemature350

proteinchain.Thehelicaltipresiduesofthethirdsubunit,fromMet60toAla95,351

areunresolvedintheX-raystructure.Themissingresidues,therefore,were352

modeledwithMODELLER[25],usingthetworesolvedsubunitsastemplates.353

Theinitialtip-to-tipdistanceofthemodeledsubunitwithregardstotheother354

twosubunitswas2.53and1.74nm,respectively.15independentsimulations,355

eachof100nsinlength,wereinitiallyperformedofSkp,eitherinisolation,orin356

thepresenceoflipidAorlipidA+KDOboundattheputativeLPSbindingsiteon357

eachsubunitoftheSkphomotrimer,asdescribedinapreviousstudy[8].In358

accordancewiththeconclusionsofthispreviousstudy,thesimulationsoflipid-359

boundSkpshowedbroadlysimilardynamicstothelipidfreesimulations.360

Snapshotsofthe“openstate”ofSkpdefinedfromthesepreliminaryshorter361

simulationswereusedtoinitialize2x1µssimulations.Thiswasdefinedasthe362

largestdistancebetweentheCαofAla76residues(locatedatthetipofeach363

subunit).TheopenSkpstructuresweresolvatedinarhombicdodecahedronbox,364

containing~110,000watermoleculesanda0.15MNaClsolution.Position365

restraintsof1,000kJmol-1nm-2wereappliedfor5nstotheCαatomsofthe366

protein,priortoperforming1μsproductionsimulations.367

368

SimulationParametersandAnalysis369

17

SimulationswereperformedwiththeGROMACSsimulationpackage[26,27],370

usingtheCHARMM22forcefieldparameterset,incorporatingtheCMAP371

potentialcorrections[28,29],asdescribedpreviously[8].Allsimulationswere372

performedintheNPTensembleatatemperatureof298Kandpressureof1atm.373

Thetemperaturewascontrolledwiththevelocity-rescalethermostat[30],and374

pressurebytheParrinello-Rahmanbarostatusingisotropiccoupling[31,32].A2375

fstimestepwasusedtointegratetheequationsofmotion,andtheLINCS376

algorithmwasusedtoconstrainallbondlengths[33].A1.2nmcut-offwasused377

forLennard-Jonesinteractions,withthepotentialsmoothlyswitchedoff378

between1.0and1.2nm.Electrostaticinteractionswerecalculatedusingthe379

Particle-Mesh-Ewaldalgorithmwithareal-spacecut-offof1.2nm[34].VMDwas380

usedforvisualizationandcreatingimages[35].TheVMDplugin,Bendix,was381

usedtocalculatethedegreeofbendinginthehelicesandforthegenerationof382

imagesofkinkedhelices[36].OtheranalyseswereperformedusingGROMACS383

toolsandin-housescriptsthatutilizedthecapabilitiesofMDAnalysis[37]and384

MODELLER[25].385

386

SAXSMeasurementsandPrimaryProcessing387

SAXSmeasurementswereperformedattheP12beamlineofEMBL(DESY388

Hamburg)[38]coveringtherangeofmomentumtransfer0.01<q<0.44Å-1(q=389

4πsin(θ)/λ,where2θisthescatteringangleandλ=1.2ÅistheX-ray390

wavelength).SamplesofSkpwerepreparedinthebuffercontaining25mM391

Hepes(pH7.5),150mMNaCl,1mMDTT.Datawasacquiredinarangeof392

proteinconcentrationsfrom5.8to0.6mg/mlandanalyzedusingtheATSAS393

softwarepackage[39].Theprimarydataprocessingwasperformedusing394

18

PRIMUS[40].TheforwardscatteringI(0)andtheradiiofgyrationRgwere395

evaluatedusingtheGuinierapproximation[41],assumingthatatverysmall396

angles(s<1.3/Rg),theintensityisrepresentedasI(s)=I(0)exp(−(sRg)2/3).397

ThemaximumdimensionsDmaxwerecomputedusingtheindirecttransform398

packageGNOM[42],whichalsoprovidesthedistancedistributionfunctionp(r).399

400

ComparisonofMD-DerivedConformerstoSAXS401

CRYSOL[43]wasusedtocalculatetheformfactorsforthestructuresderived402

fromMDtrajectories.Defaultvalueswereusedforthesolventdensity(0.334403

e/Å3).Alltheoreticalscatteringcurvesderivedfromsimulationstructureswere404

fittedtotheexperimentaldata.405

406

FlexibilityAssessmentbyEnsembleOptimizationMethod407

EnsembleOptimizationMethod(EOM)[18,19]hasbeenappliedtocharacterize408

conformationalvariabilityofSkpinsolution.Thepositionsandorientationsof409

thethreehelicalarms(Asn20–Ala115)wererandomizedwithrespecttothe410

restofthestructuretogenerateapoolof10,000models.Thegeneticalgorithm411

wasappliedtoselectfromthepoolanoptimizedensemblethatbestfitthe412

experimentalSAXSdata.413

414

ProteinExpression,PurificationandIsotopelabeling415

SkpcontaininganN-terminalhexa-histidinetagandlackingitssignalsequence,416

andOmpXobtainedfrominclusionbodies,wereexpressedandpurifiedas417

describedpreviously[14,44].[U–2H,15N,13C]-labeledSkpwasobtainedby418

growingtheexpressioncellsinM9-minimalmedia[45]supplementedwith419

19

(15NH4)Cl,D–[2H,13C]–glucoseandD2O.[U–2H]-labeledOmpXwasobtainedby420

theadditionofD–[2H]–glucoseandD2OtoM9minimalmedium.Isotopeswere421

purchasedfromSigma-AldrichorCambridgeIsotopeLabs.422

423

NMRSpectroscopy424

NMRexperimentsofhumanSkpandSkp/OmpXwereperformedinNMRbuffer425

containing25mMMES,150mMNaClpH6.5.Themeasurementswererecorded426

at304KonaBrukerAscendII700MHzspectrometerequippedwitha427

cryogenicallycooledtriple-resonanceprobe.The3D[1H,1H]-NOESY-15N-TROSY428

experiments[46–48]wererecordedwithamixingtimeof100msresultingina429

totalexperimenttimeof141.5hforSkpinitsapoformandfor112.5hforSkp-430

OmpX.Theinterscandelaywassetto0.95s.Inthedirectdimension,1024431

complexpointswererecordedinanacquisitiontimeof91ms,multipliedwitha432

75°-shiftedsinebell,zero-filledto2048pointsandFouriertransformed.Inthe433

nitrogenindirectdimension,90complexpointsweremeasuredwithamaximal434

evolutiontimeof40ms,multipliedwitha75°-shiftedsinebell,zero-filledto256435

pointsandFouriertransformed.Intheprotonindirectdimension,150complex436

pointsweremeasuredwithamaximalevolutiontimeof18msforSkpinitsapo437

form,multipliedwitha75°-shiftedsinebell,zero-filledto512pointsandFourier438

transformed.Forallspectraapolynomialbaselinecorrectionwasappliedinall439

dimensions.NMRdatawereprocessedusingPROSA[49]andanalyzedwith440

CARA[50]andXEASY[51].441

442

ComparisonofSimulationandNMRobserveddistances443

Asubsetofamidehydrogen-hydrogeninteractionswaschosenforcomparisonof444

20

simulationandNOESYdata.Thesubsetwaschosenbycomparingthedistances445

oftheextremestructuresofPC1andPC2totheX-raystructure.Twoamide446

hydrogenswereselectedtobeinthissubsetiftheywere<6Åinanystructure,447

andiftheirdistancedifferedby>1ÅcomparedtotheX-raystructure.Onlythose448

residuepairsforwhichaclearcrosspeak/diagonalpeakquotientforeither449

apoSkporSkp-OmpXcouldbemeasuredwereusedinthefinalanalysis.The450

distance,𝑑,betweentwoamidehydrogens,𝑖and𝑗,wasdeterminedusingthe451

followingEquation: 𝑑!" = 𝑐 ∙ 𝑃!"!!!,where𝑃isthecrosspeak/diagonalpeak452

quotientand,𝑐isaglobalconstantobtainedbyminimizingtheaveragedistance453

betweentheNMR(bothapoSkpandSkp/OmpX)andsimulationdata,withthe454

additionalconstraintthatnoNOEdistancecouldbe>6Å.Inanygivensimulation455

frame,thedeviationfromexperimentforanatompairwasmeasuredasthe456

absolutedifferencebetweenthesimulationdistanceandthedistanceestimated457

fromtheNOEsignal.ForamidehydrogenpairswithoutanNOEsignal,the458

deviationwas0ifthesimulateddistancewas>6Å,andotherwisewascalculated459

astheabsolutedifferencefrom6Å.460

461

21

References462

1. RigelNW,SilhavyTJ.Makingabeta-barrel:Assemblyofoutermembrane463

proteinsinGram-negativebacteria.CurrOpinMicrobiol.ElsevierLtd;464

2012;15:189–193.doi:10.1016/j.mib.2011.12.007465

2. RuizN,KahneD,SilhavyTJ.Advancesinunderstandingbacterialouter-466

membranebiogenesis.NatRevMicrobiol.2006;4:57–66.467

doi:10.1038/nrmicro1322468

3. HaganCL,SilhavyTJ,KahneD.β-BarrelMembraneProteinAssemblyby469

theBamComplex.AnnuRevBiochem.2011;80:189–210.470

doi:10.1146/annurev-biochem-061408-144611471

4. NeupertW,HerrmannJM.Translocationofproteinsintomitochondria.472

AnnuRevBiochem.2007;76:723–749.473

doi:10.1146/annurev.biochem.76.052705.163409474

5. SchleiffE,BeckerT.Commongroundforproteintranslocation:access475

controlformitochondriaandchloroplasts.NatRevMolCellBiol.Nature476

PublishingGroup;2011;12:48–59.doi:10.1038/nrm3027477

6. KnowlesTJ,TuckerAS,OverduinM,HendersonIR.Membraneprotein478

architects :theroleoftheBAMcomplexinoutermembraneprotein479

assembly.2009;7.doi:10.1038/nrmicro2069480

7. ThomaJ,BurmannBM,HillerS,MüllerDJ.Impactofholdasechaperones481

SkpandSurAonthefoldingofβ-barrelouter-membraneproteins.Nat482

StructMolBiol.NaturePublishingGroup;2015;advanceon:795–802.483

doi:10.1038/nsmb.3087484

22

8. BurmannBM,HoldbrookDA,CallonM,BondPJ,HillerS.Revisitingthe485

InteractionbetweentheChaperoneSkpandLipopolysaccharide.BiophysJ.486

2015;108:1516–1526.doi:10.1016/j.bpj.2015.01.029487

9. KorndörferIP,DommelMK,SkerraA.Structureoftheperiplasmic488

chaperoneSkpsuggestsfunctionalsimilaritywithcytosolicchaperones489

despitedifferingarchitecture.NatStructMolBiol.2004;11:1015–1020.490

doi:10.1038/nsmb828491

10. WaltonTa,SousaMC.CrystalstructureofSkp,aprefoldin-likechaperone492

thatprotectssolubleandmembraneproteinsfromaggregation.MolCell.493

2004;15:367–74.doi:10.1016/j.molcel.2004.07.023494

11. StirlingPC,BakhoumSF,FeiglAB,LerouxMR.Convergentevolutionof495

clamp-likebindingsitesindiversechaperones.NatStructMolBiol.496

2006;13:865–870.doi:10.1038/nsmb1153497

12. SiegertR,LerouxMR,ScheuflerC,HartlFU,MoarefiI.Structureofthe498

MolecularChaperonePrefoldin:UniqueInteractionofMultipleCoiledCoil499

TentacleswithUnfoldedProteins.Cell.2000;103:621–632.500

doi:10.1016/S0092-8674(00)00165-3501

13. GrussF,ZähringerF,JakobRP,BurmannBM,HillerS,MaierT.The502

structuralbasisofautotransportertranslocationbyTamA.NatStructMol503

Biol.NaturePublishingGroup;2013;20:1318–1320.504

doi:10.1038/nsmb.2689505

14. BurmannBM,WangC,HillerS.Conformationanddynamicsofthe506

periplasmicmembrane-protein-chaperonecomplexesOmpX-Skpand507

tOmpA-Skp.NatStructMolBiol.NaturePublishingGroup;2013;20:1265–508

23

72.doi:10.1038/nsmb.2677509

15. ZaccaiNR,SandlinCW,HoopesJT,CurtisJE,FlemingPJ,FlemingKG,etal.510

DeuteriumLabelingTogetherwithContrastVariationSmall-Angle511

NeutronScatteringSuggestsHowSkpCapturesandReleasesUnfolded512

OuterMembraneProteins[Internet].1sted.Isotopelabelingof513

biomolecules:labelingtechniques.ElsevierInc.;2015.514

doi:10.1016/bs.mie.2015.06.041515

16. PerkinsSJ.Proteinvolumesandhydrationeffects.EurJBiochem.516

1986;157:169–180.doi:10.1111/j.1432-1033.1986.tb09653.x517

17. PerkinsSJ.X-Rayandneutronscatteringanalysesofhydrationshells:A518

molecularinterpretationbasedonsequencepredictionsandmodellingfits.519

BiophysChem.2001;93:129–139.doi:10.1016/S0301-4622(01)00216-2520

18. BernadoP,MylonasE,PetoukhovMV,BlackledgeM,SvergunDI.521

Structuralcharacterizationofflexibleproteinsusingsmall-angleX-ray522

scattering.JAmChemSoc.2007;129:5656–5664.doi:10.1021/ja069124n523

19. TriaG,MertensHDT,KachalaM,SvergunDI.Advancedensemble524

modellingofflexiblemacromoleculesusingX-raysolutionscattering.IUCrJ.525

InternationalUnionofCrystallography;2015;2:207–217.526

doi:10.1107/S205225251500202X527

20. VainbergIE,LewisSA,RommelaereH,AmpeC,VandekerckhoveJ,Klein528

HL,etal.Prefoldin,achaperonethatdeliversunfoldedproteinsto529

cytosolicchaperonin.Cell.1998;93:863–873.doi:10.1016/S0092-530

8674(00)81446-4531

21. WebbCT,GormanMA,LazarouM,RyanMT,GulbisJM.Crystalstructureof532

24

themitochondrialchaperoneTIM9/10revealsasix-bladedalpha-533

propeller.MolCell.2006;21:123–133.doi:10.1016/j.molcel.2005.11.010534

22. GuY,LiH,DongH,ZengY,ZhangZ,PatersonNG,etal.Structuralbasisof535

outermembraneproteininsertionbytheBAMcomplex.Nature.Nature536

PublishingGroup;2016;advanceon:113–122.doi:10.1038/nature17199537

23. HanL,ZhengJ,WangY,YangX,LiuY,SunC,etal.StructureoftheBAM538

complexanditsimplicationsforbiogenesisofouter-membraneproteins.539

NatStructMolBiol.NaturePublishingGroup;2016;23:192–196.540

doi:10.1038/nsmb.3181541

24. BakelarJ,BuchananSK,NoinajN.Thestructureofthebetabarrel542

assemblymachinerycomplex.Science(80-).2016;351:180–186.543

doi:10.1007/s13398-014-0173-7.2544

25. SaliA,BlundellTL.ComparativeProteinModellingbySatisfactionof545

SpatialRestraints.JournalofMolecularBiology.1993.pp.779–815.546

26. VanDerSpoelD,LindahlE,HessB,GroenhofG,MarkAE,BerendsenHJC.547

GROMACS:Fast,flexible,andfree.JComputChem.2005;26:1701–1718.548

doi:10.1002/jcc.20291549

27. BerendsenHJC,vanderSpoelD,vanDrunenR.GROMACS:Amessage-550

passingparallelmoleculardynamicsimplementation.ComputPhys551

Commun.1995;91:43–56.doi:10.1016/0010-4655(95)00042-E552

28. MackerellaD,Jr,BashfordD,BellottM,DunbrackRL,EvanseckJD,etal.553

All-AtomEmpiricalPotentialforMolecularModelingandDynamics554

StudiesofProteins.JPhysChemB.1998;102:3586–3616.555

doi:10.1021/jp973084f556

25

29. LindahlE,BjelkmarP,LarssonP,CuendetMa.,HessB.Implementationof557

thecharmmforcefieldinGROMACS:Analysisofproteinstabilityeffects558

fromcorrectionmaps,virtualinteractionsites,andwatermodels.JChem559

TheoryComput.2010;6:459–466.doi:10.1021/ct900549r560

30. BussiG,DonadioD,ParrinelloM.Canonicalsamplingthroughvelocity561

rescaling.JChemPhys.2007;126:1–7.doi:10.1063/1.2408420562

31. ParrinelloM.Polymorphictransitionsinsinglecrystals:Anewmolecular563

dynamicsmethod.JApplPhys.1981;52:7182.doi:10.1063/1.328693564

32. NoséS,KleinML.Constantpressuremoleculardynamicsformolecular565

systems.MolPhys.1983;50:1055–1076.566

doi:10.1080/00268978300102851567

33. HessB,BekkerH,BerendsenHJC,FraaijeJGEM.LINCS:Alinearconstraint568

solverformolecularsimulations.JComputChem.1997;18:1463–1472.569

doi:10.1002/(SICI)1096-987X(199709)18:12<1463::AID-JCC4>3.0.CO;2-H570

34. EssmannU,PereraL,BerkowitzML,DardenT,LeeH,PedersenLG.A571

smoothparticlemeshEwaldmethod.JChemPhys.1995;103:8577–8593.572

doi:10.1063/1.470117573

35. HumphreyW,DalkeA,SchultenK.VMD:Visualmoleculardynamics.JMol574

Graph.1996;14:33–38.doi:10.1016/0263-7855(96)00018-5575

36. DahlACE,ChaventM,SansomMSP.Bendix:Intuitivehelixgeometry576

analysisandabstraction.Bioinformatics.2012;28:2193–2194.577

doi:10.1093/bioinformatics/bts357578

37. Michaud-AgrawalN,DenningEJ,WoolfTB,BecksteinO.MDAnalysis:A579

26

toolkitfortheanalysisofmoleculardynamicssimulations.JComputChem.580

2011;32:2319–2327.doi:10.1002/jcc.21787581

38. BlanchetCE,SpilotrosA,SchwemmerF,GraewertMA,KikhneyA,Jeffries582

CM,etal.Versatilesampleenvironmentsandautomationforbiological583

solutionX-rayscatteringexperimentsattheP12beamline(PETRAIII,584

DESY).JApplCrystallogr.InternationalUnionofCrystallography;2015;48:585

431–443.doi:10.1107/S160057671500254X586

39. PetoukhovMV,FrankeD,ShkumatovAV,TriaG,KikhneyAG,GajdaM,et587

al.NewdevelopmentsintheATSASprogrampackageforsmall-angle588

scatteringdataanalysis.JApplCrystallogr.InternationalUnionof589

Crystallography;2012;45:342–350.590

40. KonarevPV,VolkovVV,SokolovaAV,KochMHJ,SvergunDI,KonarevPV,591

etal.PRIMUS :aWindowsPC-basedsystemforsmall-anglescatteringdata592

analysisPRIMUS :aWindowsPC-basedsystemforsmall-anglescattering593

dataanalysis.JApplCrystallogr.2003;36:1277–1282.594

41. GuinierA.Small-AngleX-rayDiffraction:ApplicationtotheStudyof595

UltramicroscopicPhenomena.AnnPhys(Paris).1939;12:161–237.596

42. SvergunDI.Determinationoftheregularizationparameterinindirect-597

transformmethodsusingperceptualcriteria.JApplCrystallogr.598

InternationalUnionofCrystallography;1992;25:495–503.599

43. SvergunD,BarberatoC,KochMHJ.CRYSOL-aProgramtoEvaluateX-ray600

SolutionScatteringofBiologicalMacromoleculesfromAtomicCoordinates.601

JApplCrystallogr.1995;28:768–773.doi:10.1107/S1600536804004209602

44. TaferH,HillerS,HiltyC,FernándezC,WüthrichK.NonrandomStructure603

27

intheUrea-UnfoldedEscherichiacoliOuterMembraneProteinX(OmpX)†.604

Biochemistry.2004;43:860–869.doi:10.1021/bi0356606605

45. SambrookJ,FritschEF,ManiatisT.Molecularcloning.2nded.Coldspring606

harborlaboratorypressNewYork;1989.607

46. XiaYL,SzeKH,ZhuG.Transverserelaxationoptimized3Dand4DN-15/N-608

15separatedNOESYexperimentsofN-15labeledproteins.JBiomolNMR.609

2000;18:261–268.Available:<GotoISI>://000165485200009610

47. ZuiderwegER,FesikSW.Heteronuclearthree-dimensionalNMR611

spectroscopyoftheinflammatoryproteinC5a.Biochemistry.1989;28:612

2387–2391.doi:10.1021/bi00432a008613

48. MarionD,DriscollPC,KayLE,WingfieldPT,Baxa,GronenbornaM,etal.614

Overcomingtheoverlapproblemintheassignmentof1HNMRspectraof615

largerproteinsbyuseofthree-dimensionalheteronuclear1H-15N616

Hartmann-Hahn-multiplequantumcoherenceandnuclearOverhauser-617

multiplequantumcoherencespectroscopy:applicationto.Biochemistry.618

1989;28:6150–6156.doi:10.1021/bi00441a004619

49. GüntertP,DötschV,WiderG,WüthrichK.Processingofmulti-dimensional620

NMRdatawiththenewsoftwarePROSA.JBiomolNMR.2:619–629.621

doi:10.1007/BF02192850622

50. KellerR.TheComputerAidedResonanceAssignmentTutorial.Ed.Rochus623

LJKeller.EditorialCantinaVerlag,Goldau,Swiss;2004.624

51. BartelsC,XiaT,BilleterM,GüntertP,WüthrichK.TheprogramXEASYfor625

computer-supportedNMRspectralanalysisofbiologicalmacromolecules.626

JBiomolNMR.Springer;1995;6:1–10.627

28

Figures628

629

Figure1.Openingandclosingofapo-Skpinsolution.A)Theradiusof630gyrationtimelinesforSkpintwo1µstrajectoriesareshowninredandblue.The631linesindicatedtotherightofthegraphrepresent:a*,the“extremeopen”model632(shownlaterinFigure4);b*,thelargestradiusofgyrationachievedduringfree633simulation;c*,theSAXSestimateoftheradiusofgyrationdescribedherein(after634removingtheinfluenceofthehydrationshell),alsoequaltotheestimate635providedin[15];andd*,theradiusofgyrationcalculatedforthecrystal636structure(PDB1SG2).B)DistributionoftheradiiofgyrationforSkpfromboth6371µstrajectories.Thepinklineisthecombineddistribution.Theothercolored638linescorrespondtothosedescribedinA.C)ThestructurescorrespondingtoI)639themaximumandII)theminimumradiusofgyrationachievedinthe640simulations.D)Thetip-to-tipdistancebetweenallpairsofsubunitsintheSkp641trimer.Thedistributioniscombinedforbothtrajectories.Thetip-to-tipdistance642wascalculatedasthedistancebetweentheCαatomsofAla76inallsubunitpairs.643Theline,d*,correspondstothevaluefromthecrystalstructure.644645

646

29

647648Figure2.Dynamicexchangeofakinkfromhelixα2toα3inthetentacle649domain.A)CylinderrepresentationofthesubunitinA,displayingthekinkin650helixα2(stateI),andhelixα3(stateII).TheothertwosubunitsofSkpare651displayedingreyforclarity.B)Estimatedkinkangleforhelixα2andα3forone652subunitoverthecourseofa1µssimulation.Thedegreeofbendingisshownin653thetopleftcorner.Anexchangeofkinkbetweenhelicesmaybeclearlyobserved654betweenthe400-500nstimeperiod.655656

30

657658Figure3.Large-scaledominantmotionsofboththetrimericcomplexand659individualsubunitsofSkp.A)TheextremeconformationsoftheentireSkp660trimeralongprincipalcomponentsi)PC1andii)PC2.PCAwasperformedforthe661combined(2x1μs)“openSkp”trajectories.Whencombined,PC1toPC4662explainedalmost90%ofthetotalvarianceinthedata.Betweenthetwo663individual1μstrajectories,therewassubstantialsimilarityinthesemodes664(covarianceoverlap,0.68),indicatingconvergenceinthesubspaceexplored.The665exchangeofthehelicalkinkbetweenhelixα2andα3(Figure2)appearedin666noneofthefirstfourlowfrequencymodes,andwasinsteadobservedinthe667higherfrequencymodes,PC5toPC9.B)Theextremeconformationsofthe668isolatedsubunit,alongprincipalcomponentsi)PC1andii)PC2.PCAwas669performedforthecombinedtrajectoriesofeachindividualproteinsubunit(6x1670μs),afterleast-squaresfittingtotherigidheaddomain.PC1representsamotion671

31

involvingexchangeofthehelicalkinkandoutwardprojectionofthetipsofthe672tentacle,andaccountedfor>50%ofthetotalvarianceinthedata.PC2accounted673foronly10%ofthedata,andinvolvesarotationofthehelicaltentaclesrelative674totheβ-sheetheadinanorthogonaldirectiontoPC1.InbothA)andB),the675arrowsindicatethedirectionandmagnitudeofmotion,andthepercentageof676thetotalvarianceexplainedbytheprincipalcomponentisshownbesideeach677figure.Notethedifferenceindistancescaleineachfigure.678679

680

32

681682Figure4.Hypotheticalmodelofthe“extremeopen”and“extremeclosed”683statesofSkp.A)The“extremeopen”and“extremeclosed”modelscreatedby684assumingtheextremestatesofprincipalcomponent1(PC1)ofthesubunit685dynamicsareappliedtoeachsubunitoftheSkptrimersimultaneously.B)The686largestsphereradius,alongthecavityaxis,calculatedforthe“extremeopen”and687“extremeclosed”statesshowninA).Theradiusofgyrationforthe“extreme688open”modelisindicatedinFigure1A(a*).689690

691

33

692693Figure5.EnsemblefittingtotheexperimentalSAXSintensities.A)Relative694frequencyofstructureswithaparticularRgyrintheMD(green)andRANCH(red)695pools.ThefrequencyofmodelsselectedfrombyEOMisshownfortheRANCH696pool(pink/cyan).B)TheoptimizedensembleofRANCH(redline)andMD697(green)generatedstructuresfittedtotheexperimentalSAXSprofile.698RepresentativestructuresselectedbyEOMareshownforboththeMD(iandii)699andRANCH(iii,ivandv)generatedmodels.700701

702

34

703704Figure6.NMRobservedbackboneamide-amidehydrogendistancesfor705apo-andOmpX-boundSkp,comparedtosimulatedapoSkp.A)The706backbonestructureofSkp,withresiduesinblueindicatingthosebetweenwhich707amidehydrogendistancesweremeasuredforcomparisonwithNOE708assignments.Ahydrogen-hydrogeninteractionwasselectedforanalysisifthe709pairofatomswere<6ÅfromoneanotherintheX-raystructureortheextreme710PC1andPC2structures.Asubsetofthesewasselectedfromthoseinteractions711thatdifferedby>1ÅbetweenthePCstructuresandtheX-raystructure.Ofthese,712onlythoseresiduepairsforwhichaclearcrosspeak/diagonalpeakquotientfor713eitherapoSkporSkp/OmpXcouldbemeasuredwereusedinthefinalanalysis.714B)Distributionacrossallsimulationtrajectoriesofmeandeviationsfromamide715hydrogendistancesfromNOESYspectraofapo-andOmpX-boundSkp.Deviation716fromexperimentforanatompairwasmeasuredastheabsolutedifference717betweenthesimulationdistanceandthedistanceestimatedfromtheNOEsignal.718ForamidehydrogenpairswithoutanNOEsignal,thedeviationwas0ifthe719simulateddistancewas>6Å,andotherwisewascalculatedastheabsolute720differencefrom6Å.DeviationsforeachsubunitoftheX-raystructureare721indicatedwithdashedlines.C)Assigned2Dstripsfrom3D15N-edited-[15N,1H]-722NOESYspectratakenattheindicatedpositionsof250µMSkpinitsapoform723(cyan)and250µMSkpwithboundOmpX(holoform,purple)inNMRbufferat72437°C.SpectrawererecordedwithaNOESYmixingtimeof100ms.Brokenlines725indicateNOEconnections.BrokencirclesindicatemissingNOEcross-peaksin726eithertheholoortheapo-state.**-denotescrosspeaksfromadjacentplanes.727728