A MolecularsimulationasanaidtoexperimentalistsWilfredFvanGunsterenJozicaDolencandAlanEMarkComputerbasedmolecularsimulationtechniquesareincreasinglyusedtointerpretexperimentaldataonbiomolecularsystem ID: 499594
Download Pdf The PPT/PDF document "vailable online at www.sciencedirect.com" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
A vailable online at www.sciencedirect.com MolecularsimulationasanaidtoexperimentalistsWilfredFvanGunsteren,JozicaDolencandAlanEMarkComputer-basedmolecularsimulationtechniquesareincreasinglyusedtointerpretexperimentaldataonbiomolecularsystemsatanatomiclevel.Directcomparisonbetweenexperimentandsimulationis,however,seldomstraightforward.Theavailableexperimentaldataarelimitedin www.sciencedirect.comCurrentOpinioninStructuralBiology judgedandillustratethesewithreferencetorecentliterature.InterpretationofexperimentaldatausingWhenrelatingtheofapropertyoveragivenconformationaldistribution),whetherfromasimu-lation()ormeasuredexperimentally(),totheconformationaldistributionitself,threegeneralcasescanbedistinguished:doesnotreßecttheshapeof)asinsensitivetoconformation;see,forexample,reference[doesnotreßecttheshapeof)asdeterminedbyrarelysampledconformationswithsmall(irrelevant)Boltzmannweights;see,forexample,reference[reßectsthedominantconformationsofOnlyinthecaseQ3cancarryinformationrelevanttotheinterpretationofatamolecularlevel.Never-theless,agreementbetweenmaybeobtainedif:isinsensitiveto),thatis,matchesirrespectiveofthe)beingsimulated[(A2)Therearecompensatingerrorsinthesimulationmodel,procedureorexperimentalset-up[];seeD1-5below.(A3)Theexperimentaldataofinterest,,havebeenusedtobiasthesimulation.issensitivetothedistributionAgain,onlyinthecaseA4canthedegreeofagreementbetweensimexpbeusedtovalidatethesimulationand/ortointerprettheexperimentalresults.Failuretoobserveacorrelationbetweenthesimulationandexperimentcanalsobeduetomanyreasons:(D1)ThesimulationisinsufÞcientlyaccurate,thatis,(a)relevantdegreesoffreedomwereomitted;(b)theforceÞeldwasinsufÞcientlyaccurate;(c)approxi-mationsmadewhensolvingtheequationsofmotionweretoocrude;(d)inappropriatethermo-dynamicorspatialboundaryconditionswereused.(D2)Themeasuredisinaccurate.areaverageddifferentlywithrespecttotimeorspatialextent.(D4)Relatedbutdifferentquantitiesarecompared,forexample,atom-positionalßuctuationsversuscrys-tallographicBfactors.(D5)Differentsystemsarecompared(e.g.crystalversussolution),orsystemsstudiedunderdifferentthermodynamicconditions(e.g.temperature,pressure,pH,ionicstrength,etc.).Clearlytheinterpretationofexperimentalobservablesintermsofmolecularconformationsisnotstraightforward.Inparticular,thedegreeofagreementbetweencaneasilybeoverinterpretedandfrequentlyleadsresearcherstomakeunrealisticclaimsregardingthevalidityofasimulation(whenanapparentcorrelationbetweentheoryandexperimentisobserved)or,equallyproblematic,leadstotheunjustiÞeddisregardofsimu-lationresults(whennocorrelationisevident).Note,inalloftheabovedenotesanobservablequantity,thatis,apropertythatcanbemeasureddirectly.Suchprimaryexperimentaldatashouldnotbeconfusedwithsecondaryexperimentaldata,,thatis,dataderivedbyapplyingagivenprocedure,,basedonvariousassumptionsandapproximations:Forexample,whereaspeaklocationandintensityfromX-raydiffractionorNMRspectroscopicexperimentsrepresentprimarydata,molecularstructures,NMRorderparameters,andsoonaresecondary(derived)quantities.Comparisonsofsimulationstosecondary,non-observed,experimentalquantitiesreßect,atleastpartly,theapproximationsandassumptionsassociatedwiththecon-versionandmayinrealitycarrylittleexperimentalinfor-mation[].Clearly,whencoupling(restraining)asimulationtoasetofvaluesinordertoensurethattheconformationaldistributionsatisÞes,onlyprimaryexperimentaldatashouldbeused.Forfurtherdiscussiononthevalidationofmolecularsimulationsseereference[ExperimentallyobservablequantitiesBelow,weillustratetheuseofbiomolecularsimulationstointerpretarangeofexperimentallyobservedquantitiesintermsoftheconsiderationsoutlinedaboveusingexamplesfromtherecentliterature.Onequantitythatisrelativelyinsensitivetotheparticularconformationaldistributionisthesmall-angleX-rayscatteringintensity,fromwhichtheradiusofgyrationcanbederived.Asshowninreference[]verydifferentconformationaldistributionscangiverisetoagivenexperimentalobser-vation.Thisstudyalsoillustratesthedifferencebetweencomparingtoprimarydata(X-rayintensities)asopposedtosecondarydata(radiusofgyration).Questionablecom-parisonstosecondaryquantitiesaremadeinmanystu-dies.Forexamplein[]ßuorescencemeasurementsontheTrp-cageproteinareusedtoinferchangesinenthalpy,whichareusedtoinferchangesinthenumberofhydrogenbondsasafunctionoftemperature,whichareultimatelyrelatedtoresultsfromsimulations[].InasimilarmannerthevalidityofsimulationsperformedwithJ-couplingconstantswasassessedbycomparingtopro-posedhydrogenbondenergiesinsteadofprimaryquantities[].InsuchcasesthebasiccomparisonisTheoryandsimulation CurrentOpinioninStructuralBiology:149Ð153www.sciencedirect.com betweenalternativemodelingprocedures,notbetweensimulationandexperiment.Bycontrast,wherethecom-parisonistoprimaryexperimentalquantitiessuchasinfraredspectra[17,18]orNMRspectra[],adirectvalidationofthesimulationmodelsbyexperimentispossible.Theuseoftheso-calledphi-valuestorestrainorvali-datefoldingsimulationsdeservesaspecialcomment.Phi-valueanalysisisausefulframeworkwithinwhichtoanalyzeexperimentaldataonfoldingtransitionsinproteins[].Phi-valuesaresecondarydataderivedfromtheeffectofmutationsonstabilityandfoldingkinetics.Anequivalentquantitycannotbeextractedfromsimulations;insteadphi-valuesareoftendeÞnedintermsoftheso-calledÔnativecontactsÕ.Clearly,asthedeÞnitionsofthesetwoquantitiesareverydifferentandevenspandifferentranges,anyapparentcorrelationcanneitherbeusedtovalidatethesimulationnortointer-prettheexperiment(seeD4andD5above)[22Ð24].Acomparisonofsimulatedversusmeasuredfoldingratesforproteins[25,26]basedonmanyshortsimulationsisalsoproblematicowingtothedifferenttimescalesaccessibletosimulationsandexperiment(D3),thedeÞnitionofwhenthesystemisfoldedandassumptionsinregardtotheprobabilityoffoldingatlongtimescales.IncontrasttoX-rayorNMRstudiesthatgenerallyyieldthousandsofindividualparameters(diffractionintensi-tiesoratomÐatomresonances),insometypesofexper-imentonlyasinglequantityismeasured,forexample,atomicforcemicroscopy,ionicconductancemeasure-mentsorwideangleX-rayspectra.SuchdataaregenerallynotsuitedformodelvalidationowingtoinabilitytorelatethemeasurementtospeciÞcdegreesoffreedomwithinthemodelandthusthehighprobabilityofcompensationoferrors(A2above)[27Ð30].Thisproblemisfurtheraggravatedwhenthequantityinquestionrepresentssecondaryexperimentaldatasuchasinthecalculationoftheareaperlipid,NMRorderparametersalongthelipidtailsinmembranesystemsorevendensityproÞlesacrossmembranes[AconstantdifÞcultyinthestudyofbiomolecularsystemsincludingproteinshasbeenthelimitedavailabilityofexperimentaldatawithhighinformationcontent.Fortu-nately,therangeofexperimentalquantitiesthatarebeingcomparedtosimulationscontinuestoexpand.Forexample,theresponseofgroupsofatomswithinproteinstolocalelectricÞeldshasbeenmeasuredexperimentallyandanalyzedusingsimulations[].Comparisonsbe-tweenmeasuredandcalculatedRamanspectrahaveprovidednewinsightintothedynamicsofwatersur-roundingsugars[].Measurementsofßuorescenceani-sotropy[]andßuorescencelifetimes[]havealsobeendirectlycomparedtotheresultsofsimulationsprovidingnovelinsightsintolocaldynamicsaswellashelpingtovalidatethesimulationmodels.MDsimu-lationshaveshownmobilewatermoleculesinsidecavitiesinproteinsthatareresponsiblefordetectableNOEs[orelectrondensity[],whichcouldnotbetracedtoÞxedwatermolecules.Finally,wemustdrawattentiontoafewstudies[inwhichtheexperimentaldatathathavebeenusedtobiasthesimulationwhengeneratingaconfor-mationaldistribution)issubsequentlyusedtocom-(seeA3above).Suchcomparisonsdonotvalidatethesimulationnordemon-stratethattheconformationaldistributionreßectsexper-iment.Infact,restraining)inthesimulationtomatchcaneasilyintroduceartifactsin)eventhoughthecorrectaverageismaintained,enhan-cingthelikelihoodofinappropriateconclusionsbeingdrawn[Withoutquestionmolecularsimulationsplayacrucialroleinfacilitatingtheinterpretationofexperimentaldataattheatomiclevel.Nevertheless,itmustberememberedthatcomparedtothenumberofdegreesoffreedomwithinbiomolecularsystemsofinterest,experimentaldataareingeneralverylimitedinnature.Inaddition,almostallmeasuredpropertiesareinherentlyaveragedquantities.Asaconsequence,comparisonsbetweensimulationsandexperimentwhileseeminglystraightfor-wardcanbeeasilyoverinterpretedormisinterpretedbothintermsoftheapparentqualityofasimulationaswellasintermsoftheinabilityofthemodeltoreproducetheexperimentalobservableofinterest.Inparticular,cautionmustbeexercisedwhencomparingsimulationstosecondaryorderivedexperimentaldata.Theuseofsecondarydataasrestraintsinsimulationsoftenleadstoartifactsandshouldbeavoidedwherepossible.Inallcaseswhereexperimentaldataarecomparedwithorincorporatedintoasimulationcaremustbetakenthatthedatarelatetothesamemolecularsystemunderthesamethermodynamicconditionsandcorrespondtothesamephysicalquantitiesasinthesimulation.Finally,experimentaldatathathavebeenusedtobiasorrestrainasimulationcannotsubsequentlybeusedforvalidation.FailuretoobservethesesimpleconsiderationswillmakeitmorelikelythatexcessiveclaimsregardingthevalidityofsimulationresultsaremadeorforsimulationresultstoberejectedwithoutjustiÞcation.Moreseriouslyitwillmeanthatinappropriateorevenincorrectphysicalinterpretationsofexperimentaldatawillbepropagatedintheliterature.AcknowledgementsFinancialsupportbytheNationalCentreofCompetenceinResearch(NCCR)(StructuralBiology)oftheSwissNationalScienceFoundation(SNSF),theSlovenianResearchAgency(ARRS)andtheAustralianResearchCouncil(ARC)isgratefullyacknowledged.MolecularsimulationandexperimentvanGunsteren,DolencandMark151 www.sciencedirect.comCurrentOpinioninStructuralBiology 1.vanGunsterenWF,BakowiesD,BaronR,ChandrasekharI,ChristenM,DauraX,GeeP,GeerkeDP,GlattliA,HunenbergerPHetal.Biomolecularmodeling:goals,problems,perspectivesAngewChemIntEd2.ZagrovicB,PandeVS:Howdoesaveragingaffectproteinstructurecomparisonontheensemblelevel?BiophysJ3.TrzesniakD,GlattliA,JaunB,vanGunsterenWF:NMRdataforbeta-peptidesusingmoleculardynamicsJAmChemSoc4.TrzesniakD,vanGunsterenWF:Catalyticmechanismofcyclophilinasobservedinmoleculardynamicssimulations:pathwaypredictionandreconciliationofX-raycrystallographicandNMRsolutiondataProteinSci5.LauAY,RouxB:Thefreeenergylandscapesgoverningconformationalchangesinaglutamatereceptorligand-bindingdomain:1203-1214.6.GlattliA,vanGunsterenWF:AreNMR-derivedmodelstructuresforbeta-peptidesrepresentativefortheensembleofstructuresadoptedinsolution?AngewChemIntEd7.ZagrovicB,JayachandranG,MillettIS,DoniachS,PandeVS:Howlargeisanalpha-helix?Studiesoftheradiiofgyrationofhelicalpeptidesbysmall-angleX-rayscatteringandmolecularJMolBiol8.ZagrovicB,vanGunsterenWF:ComparingatomisticsimulationdatawiththeNMRexperiment:howmuchcanNOEsactuallytellus?Proteins:StructFunctBioinf9.FuchsPFJ,BonvinA,BochicchioB,PepeA,AlixAJP,TamburroAM:KineticsandthermodynamicsoftypeVIIIbeta-turnformation:aCD,NMR,andmicrosecondexplicitmoleculardynamicsstudyoftheGDNPtetrapeptideBiophysJ:2745-2759.10.vanGunsterenWF,BonvinAMJJ,DauraX,SmithLJ:Aspectsofmodelingbiomolecularstructureonthebasisofspectroscopicordiffractiondata.InBiologicalMagneticResonance:StructureComputationandDynamicsinProteinNMR.EditedbyRamaKrishnaN,BerlinerLJ.Springer;1999:3-35.11.DauraX,AntesI,vanGunsterenWF,ThielW,MarkAE:TheeffectofmotionalaveragingonthecalculationofNMR-derivedstructuralpropertiesProteins:StructFunctGenet12.GlattliA,DauraX,SeebachD,vanGunsterenWF:Canonederivetheconformationalpreferenceofabeta-peptidefromitsCDJAmChemSoc:12972-12978.13.vanGunsterenWF:Ontestingtheoreticalmodelsbycomparisonofcalculatedwithexperimentaldata.InStudiesinPhysicalandTheoreticalChemistry:ModellingofMolecularStructuresandProperties.EditedbyRivailJ-L.Elsevier;1990:463-478.14.vanGunsterenWF,MarkAE:ValidationofmoleculardynamicsJChemPhys15.IavaroneAT,PatrikssonA,vanderSpoelD,ParksJH:FluorescenceprobeofTrp-cageproteinconformationinsolutionandingasphaseJAmChemSoc16.GsponerJ,HopearuohoH,CavalliA,DobsonCM,VendruscoloM:Geometry,energetics,anddynamicsofhydrogenbondsinproteins:structuralinformationderivedfromNMRscalarcouplingsJAmChemSoc15127-15135.17.MuYG,KosovDS,StockG:Conformationaldynamicsoftrialanineinwater.2.ComparisonofAMBER,CHARMM,GROMOS,andOPLSforceÞeldstoNMRandinfraredJPhysChemB:5064-5073.18.GnanakaranS,HochstrasserRM,GarciaAE:NatureofstructuralinhomogeneitiesonfoldingahelixandtheirinßuenceonspectralmeasurementsProcNatlAcadSciUSA:9229-9234.19.PeterC,RuepingM,WornerHJ,JaunB,SeebachD,vanGunsterenWF:Moleculardynamicssimulationsofsmallpeptides:canonederiveconformationalpreferencesfromROESYspectra?ChemEurJ20.ShowalterSA,BruschweilerR:ValidationofmoleculardynamicssimulationsofbiomoleculesusingNMRspinrelaxationasbenchmarks:applicationtotheAMBER99SBforceÞeldJChemTheorandComp21.ItzhakiLS,OtzenDE,FershtAR:Thestructureofthetransition-stateforfoldingofchymotrypsininhibitor-2analyzedbyproteinengineeringmethodsÑevidenceforanucleationÐcondensationmechanismforprotein-foldingJMolBiol22.SettanniG,RaoF,CaßischA:Phi-valueanalysisbymoleculardynamicssimulationsofreversiblefoldingProcNatlAcadSci23.SalvatellaX,DobsonCM,FershtAR,VendruscoloM:DeterminationofthefoldingtransitionstatesofbarnasebyusingPhi(I)-value-restrainedsimulationsvalidatedbydoublemutantPhi(IJ)-valuesProcNatlAcadSciUSA:12389-12394.24.GsponerJ,HopearuohoH,WhittakerSBM,SpenceGR,MooreGR,PaciE,RadfordSE,VendruscoloM:DeterminationofanensembleofstructuresrepresentingtheintermediatestateofthebacterialimmunityproteinIm7ProcNatlAcadSciUSA:99-104.25.SnowCD,NguyenN,PandeVS,GruebeleM:comparisonofsimulatedandexperimentalprotein-folding26.SnowCD,SorinEJ,RheeYM,PandeVS:Howwellcansimulationpredictproteinfoldingkineticsandthermodynamics?AnnuRevBiophysBiomolStruct2005,:43-69.27.HummerG,SzaboA:Freeenergysurfacesfromsingle-moleculeforcespectroscopyAccChemRes28.SotomayorM,SchultenK:Single-moleculeexperimentsinvitroinsilico:1144-1148.29.SotomayorM,VasquezV,PerozoE,SchultenK:IonconductionthroughMscSasdeterminedbyelectrophysiologyandBiophysJ30.SegaM,GarberoglioG,BroccaP,CantuL:structureofphospholipidbilayers:comparisonbetweenmoleculardynamicssimulationsandwide-angleX-rayJPhysChemB31.KlaudaJB,KucerkaN,BrooksBR,PastorRW,NagleJF:Simulation-basedmethodsforinterpretingX-raydatafromlipidbilayersBiophysJ:2796-2807.32.CourniaZ,UllmannGM,SmithJC:Differentialeffectsofcholesterol,ergosterolandlanosterolonadipalmitoylphosphatidylcholinemembrane:amoleculardynamicssimulationstudyJPhysChemB2007,111:1786-1801.33.SuydamIT,SnowCD,PandeVS,BoxerSG:ElectricÞeldsattheactivesiteofanenzyme:directcomparisonofexperimentwith:200-204.34.LerbretA,BordatP,AffouardF,GuinetY,HedouxA,PaccouL,PrevostD,DescampsM:Inßuenceofhomologousdisaccharidesonthehydrogen-bondnetworkofwater:complementaryRamanscatteringexperimentsandmoleculardynamicssimulationsCarbohydrRes35.SchroderGF,AlexievU,GrubmullerH:Simulationofßuorescenceanisotropyexperiments:probingproteinBiophysJ36.GroenhofG,Bouxin-CademartoryM,HessB,DeVisserSP,BerendsenHJC,OlivucciM,MarkAE,RobbMA:Photoactivationofthephotoactiveyellowprotein:whyphotonabsorptionTheoryandsimulation CurrentOpinioninStructuralBiology:149Ð153www.sciencedirect.com triggersalsomerizationofthechromophoreintheJAmChemSoc:4228-4233.37.NilssonL,HalleB:Molecularoriginoftime-dependentßuorescenceshiftsinproteinsProcNatlAcadSciUSA38.SomaniS,ChngC-P,VermaCS:Hydrationofahydrophobiccavityanditsfunctionalrole:asimulationstudyofhumanProteins:StructFunctBioinf:868-885.39.SchifferCA,vanGunsterenWF:Accessibilityandorderofwatersitesinandaroundproteins:acrystallographictime-averagingstudyProteins:StructFunctGenet40.PaciE,Lindorff-LarsenK,DobsonCM,KarplusM,VendruscoloM:TransitionstatecontactorderscorrelatewithproteinfoldingJMolBiol:495-500.41.Lindorff-LarsenK,BestRB,DePristoMA,DobsonCM,VendruscoloM:Simultaneousdeterminationofproteinstructureanddynamics:128-132.42.VendruscoloM:DeterminationofconformationallyheterogeneousstatesofproteinsCurrOpinStructBiol43.ScottKA,RandlesLG,MoranSJ,DaggettV,ClarkeJ:ThefoldingpathwayofspectrinR17fromexperimentandsimulation:usingexperimentallyvalidatedMDsimulationstocharacterizestateshintedatbyexperimentJMolBiol2006,359:159-173.44.TordaAE,ScheekRM,vanGunsterenWF:nuclearoverhausereffectdistancerestraintsappliedtoJMolBiolMolecularsimulationandexperimentvanGunsteren,DolencandMark153 www.sciencedirect.comCurrentOpinioninStructuralBiology