基因毒性杂质之结构警示(欧洲)

EUR23844EN-2009DevelopmentofstructuralalertsfortheinvivomicronucleusassayinrodentsRomualdoBenignia,CeciliaBossaa,OlgaTcheremenskaiaaandAndrewWorthbaIstitutoSuperiorediSanita’,EnvironmentandHealthDepartment,Rome,ItalybInstituteforHealth&ConsumerProtection,EuropeanCommission-JointResearchCentre,Ispra,ItalyThemissionoftheIHCPistoprovidescientificsupporttothedevelopmentandimplementationofEUpoliciesrelatedtohealthandconsumerprotection.TheIHCPcarriesoutresearchtoimprovetheunderstandingofpotentialhealthrisksposedbychemical,physicalandbiologicalagentsfromvarioussourcestowhichconsumersareexposed.EuropeanCommissionJointResearchCentreInstituteforHealthandConsumerProtectionContactinformationAddress:TP582E-mail:andrew.worth@ec.europa.euTel.:+390332789566Fax:+390332786717http://http://ecb.jrc.ec.europa.eu/qsar/http://ec.europa.eu/dgs/jrc/LegalNoticeNeithertheEuropeanCommissionnoranypersonactingonbehalfoftheCommissionisresponsiblefortheusewhichmightbemadeofthispublication.AgreatdealofadditionalinformationontheEuropeanUnionisavailableontheInternet.ItcanbeaccessedthroughtheEuropaserverhttp://europa.eu/JRC52274EUR23844ENISSN1018-5593Luxembourg:OfficeforOfficialPublicationsoftheEuropeanCommunities©EuropeanCommunities,2009ReproductionisauthorisedprovidedthesourceisacknowledgedPrintedinItalyABSTRACTInvivomutagenicityandcarcinogenicitystudiesareposingahighdemandfortest-relatedresources.Amongthesestudies,themicronucleustestinrodentsisthemostwidelyused,asfollowuptopositiveinvitromutagenicityresults.Arecentsurveyofthe(Q)SARmodelsformutagenicityandcarcinogenicityhasindicatedthatno(Q)SARmodelsforinvivomicronucleusareavailableinthepublicdomain.Therefore,thedevelopmentandextensiveuseofestimationtechniquessuchas(Q)SARs,read-acrossandgroupingofchemicals,promisestohaveahugeanimalsavingpotentialforthisendpoint.Inthisreport,wedescribetheidentificationofstructuralalertsfortheinvivomicronucleusassay,andprovidethelistofunderlyingchemicalstructures.Thesestructuralalertsprovideacoarse-grainfilterforthepreliminaryscreeningofpotentialinvivomutagens.LISTOFABBREVIATIONSEPAEnvironmentalProtectionAgencyEUEuropeanUnionFDAFoodandDrugAdministrationHOMOHighestOccupiedMolecularOrbitalISSIstitutoSuperiorediSanita’JRCJointResearchCentreLUMOLowestUnccupiedMolecularOrbitalOECDOrganisationforEconomicCooperationandDevelopment(Q)SAR(Quantitative)Structure-ActivityRelationshipREACHRegistrationEvaluationandAuthorisationofCHemicalsROCReceiverOperatingCurveSAStructuralAlertSA_BBBenigni-Bossastructuralalertsformutagnicity/carcinogenicityinToxtreeSA_MicStructuralalertsrefersfortheinvivomicronucleusassayinToxtreeSA_ProtStructuralalertsforproteinbindingintheOECDQSARToolboxCONTENTS1.Introduction....................................................................................................62.Structuralalerts..............................................................................................83.Developmentofstructuralalertsfortheinvivomicronucleusassay..............104.Finalconsiderations......................................................................................205.References....................................................................................................21Appendix1..............................................................................................................2361.IntroductionMutagenicitytestingisanimportantpartoftheregulatoryhazardassessmentofchemicals.Itisundertakenfortwomainreasons:a)todetectchemicalsthatmightcausegeneticdamageingermcells,andthusincreasetheburdenofheritable(genetic)diseaseinthehumanpopulation;andb)todetectchemicalsthatmightbecarcinogenic(basedontheassumptionthatmutagenesis,forexampleinsomaticcells,isakeyeventintheprocessofcarcinogenesis).Sincenomethodisablealonetodetectallpossiblegenotoxicevents,awidearrayoftestsystemshasbeendevelopedandacceptedinternationallyinregulatoryschemes.Mostoften,thesemethodsareusedwithina2-tieredintegratedtestingapproach:Tier1includesinvivoassays,andTier2includesinvivoassays.Asamatteroffact,mutagenicitytestingwasthefirsttoxicityendpointforwhichinvivoassayswereacceptedforregulatorytesting,some25yearsago.Thelatterusuallycomprisebacterialmutagenicityandcytogeneticstests,althoughgenemutationtestinginculturedmammaliancellsissometimesalsoundertaken.Tier2ofthetestingstrategyinvolvestheuseofshort-terminvivostudies(usuallyabone-marrowcytogeneticsassay)toassesswhetheranypotentialforgenotoxicitydetectedattheTier1invivostageisactuallyexpressedinthewholeanimal.Thus,negativeresultsinvivoareusuallyconsideredsufficienttoindicatelackofmutagenicity,whereasapositiveresultisnotconsideredsufficienttoindicatethatthechemicalrepresentsamutagenichazard(i.e.itcouldbeafalsepositive).TheaboveapproachtogenotoxicitytestinghasbeenadoptedthroughouttheEU1,andhasbeenrecommendedinternationallyaspartofthestrategyforpredictingandquantifyingmutagenicandcarcinogenichazard(Ashbyetal.,1996;Combesetal.,2007;KirklandandSpeit,2008;Lilienblumetal.,2008).1http://guidance.echa.europa.eu/docs/guidance_document/information_requirements_r7a_en.pdf?vers=20_08_087AccordingtoanassessmentcarriedoutbytheformerEuropeanChemicalsBureau(ECB),theinvivomutagenicitystudies,shortlyfollowedbycarcinogenicity,areposinghighdemandfortest-relatedrecourses(Pedersenetal.,2003;VanderJagtetal.,2004).Amongthose,themicronucleustestinrodentsisthemostwidelyused,asfollowuptopositiveinvivomutagenicityresults.Arecentsurveyofthe(Q)SARmodelsformutagenicityandcarcinogenicity(performedjointlybyISSandtheJRC)hasindicatedthatno(Q)SARmodelsforinvivomicronucleusareavailableinthepublicdomain(Benignietal.,2007):therefore,thedevelopmentandextensiveuseofestimationtechniquessuchas(Q)SARs,read-acrossandgroupingofchemicals,mighthaveahugesavingpotentialforthisendpoint.Inthisreport,wedescribe:a)thecollectionofdataonchemicalstestedwiththeinvivomicronucleusassay;b)preliminaryanalysesofthedata;c)theidentificationofStructuralAlerts(SA)propertothistoxicologicalendpoint.First,somebackgroundinformationontheconceptofSAisprovided.82.StructuralalertsTheSAsforatoxicologicalendpointaremolecularfunctionalgroupsorsubstructuresknowntobelinkedtothattypeoftoxicity.TheSAsareacoarse-grainedapproachtotheuseofStructure-ActivityRelationships(SAR)tounderstandthetoxicitymechanismsandtopredictthetoxicactivityofchemicals.Becauseoftheirnature,theSAshavetheroleofpointingtochemicalspotentiallytoxic,whereasnoconclusionsorindicationsaboutnontoxicchemicalsarepossible(exceptbyexclusion)(BenigniandBossa,2006;BenigniandBossa,2008).AsetofchemicalscharacterizedbythesameSAconstituteafamily(class)ofcompoundsthatsharethesamemechanismofaction.ThereactivityofaSAcanbemodulatedorabolishedbytheremainingpartofthemoleculeinwhichtheSAisembedded.Atacoarse-grainlevel,suchmodulatingeffectscanberepresentedbyothermolecularsubstructures(e.g.,bulkygroupsorthotoanaromaticaminegroup)thatareknowntohaveaninfluenceonthereactivityoftheSA.Usually,theknowledgeonthemodulatingsubstructuresisquitelimitedformostoftheSAs,thusitprovideslimitedhelpindecidingwhichchemicalsinaclasswillactuallybetoxicandviceversa.ApowerfulgeneralizationoftheStructure-ActivityRelationshipsisprovidedbytheQuantitativeStructure-ActivityRelationship(QSAR)analysis,whichproducesamathematicalmodelthatlinksthebiologicalactivitytoalimitednumberofphysicalchemicalorothermolecularproperties(descriptors)withgeneralrelevance.Sincemostofthedescriptorshavecontinuousvalues,theQSARsprovidefine-tunedmodelsofthebiologicalactivity,andcangiveaccountofsubtledifferences.GeneralintroductionsonQSARaregivenelsewhere(HanschandLeo,1995,Hanschetal.,2002).ThustheSAsarenotadiscriminantmodelonthesamegroundoftheQSARmodels:thelatterproduceestimatesforbothpositiveandnegativechemicals,aswellasforthegradationoftoxicpotency.ThemainroleoftheSAsisthatofpreliminary,orlarge-scalescreenings.Theyareexcellenttoolsforcoarse-graincharacterizationofchemicals,including:descriptionofsetsofchemicals,preliminaryhazardcharacterization,categoryformationandprioritysetting(enrichment).Sincefine-tunedQSARsdonotexistformanytypesofchemicals,themodelsbasedonSAsholdaspecialplaceinpredictivetoxicology.The9knowledgeontheactionmechanismsasexemplifiedbytheSAsisroutinelyusedinSARassessmentintheregulatorycontext(see,forexample,themechanistically-basedreasoningaspresentedinWooetal.(2002).Inaddition,theSAsareatthebasisofpopularcommercial(e.g.,DEREK,byLhasaLtd.2)andnon-commercialsoftwaresystems(e.g.,Oncologic,byUSEnvironmentalProtectionAgency[EPA]3).Recently,asfollow-upofthecollaborationbetweenISSandJRC,arulebaseformutagensandcarcinogenshasbeendesignedandimplementedinthesoftwareToxtree1.51.Itusesastructure-basedapproachconsistingofanewcompilationofSAsforcarcinogenicityandmutagenicity.ItalsooffersthreemechanisticallybasedQSARsforcongenericclasses(aromaticaminesandaldehydes)(Benignietal.,2008a).Toxtree1.51isfreelyavailablefromtheJRCwebsite.42http://www.lhasalimited.org/3http://www.epa.gov/o ppt 关于艾滋病ppt课件精益管理ppt下载地图下载ppt可编辑假如ppt教学课件下载triz基础知识ppt /newchems/tools/oncologic.htm4http://ecb.jrc.ec.europa.eu/qsar/qsar-tools/index.php?c=TOXTREE103.Developmentofstructuralalertsfortheinvivomicronucleusassay3.1DataThecompilationofSAsfortheinvivomicronucleusassayinrodentsprovidedhere,isbasedonboththeexistingknowledgeonthemechanismsoftoxicactionandastructuralanalysisofthechemicalstestedintheassay.Theinvivomicronucleusdatainthepublicdomainisquitelimited.AsearchoftheChemicalCarcinogenesisResearchInformationSystem(CCRIS)attheToxnetwebsitewiththequery:“invivomicronucleus”pointsonlyto240chemicals.5Forthiswork,theremarkablylargercommercialdatabasebyLeadscopeInc.,called“FDASARGenetoxDatabase”wasused.6Thisdatabasecontainsmorethan700chemicalstestedininvivomicronucleuswithrodents,andincludesdatafromboththepublicdomainandtheUSFoodandDrugAdministration(FDA)files.Alargemajorityofdatawerebasedontheanalysisofmicronucleiinbonemarrowcells;fordetailsonthetechnique,seeforexample,KrishnaandHayashi(2000).3.1PreliminaryanalysesSincethemainroleoftheinvivomicronucleusassayinregulatoryschemesisthatofconfirming(ordisproving)thepositiveinvitroresults,itisofinteresttocheckhowtheinvivomicronucleusresultsrelatetotherodentcarcinogenicitydataandtotheprimaryinvitropredictiontest,i.e.,theSalmonellatyphimurium(Ames)test.TablesIandIIdisplaytherelationshipsbetweentheinvivomicronucleusadthetworeferencetests.TheresultsforrodentcarcinogenicityandtheAmestestwereretrievedfromthefreelyavailableISSCANv3adatabase,7whichischaracterizedby:5http://toxnet.nlm.nih.gov/cgi-bin/sis/search6http://www.leadscope.com/product_info.php?products_id=777http://www.iss.it/ampp/dati/cont.php?id=233&lang=1&tipo=711a)thehighqualityofbothchemicalandbiologicalinformation;b)theQSAR-readyformat(Benignietal.,2008b).Obviously,thetotalnumbersofchemicalsinthetwotablesarerelativeonlytothosechemicalstestedinbothsystems.TableI.ContingencytablecomparingtheresultsoftherodentcarcinogenicitytestwiththemicronucleustestCarcinogenicitytestMicronucleustestNegativePositiveTotalNegative301040Positive8657143Total11667183TableII:ContingencytablecomparingtheresultsoftheSalmonellatyphimuriumassaywiththemicronucleustestSalmonellaassayMicronucleustestNegativePositiveTotalNegative7436110Positive413475Total11570185TableIshowsthatistheinvivomicronucleusassayispoorlysensitivetotherodentcarcinogens:about60%oftherodentcarcinogensarenotdetectedbythemicronucleus.ThepoorsensitivityofthemicronucleusassaytopotentialgenotoxinsisalsoapparentfromTableII.ItshouldbeemphasizedthatthepresentresultsobtainedwiththelargeLeadscopemicronucleusdatabaseareinagreementwithpreviousanalysesbasedonsmallerdatasetsinthepublicdomain(Benigni,1995).12Inasecondroundofanalyses,theextenttowhichthemicronucleusdataarerelatedtowellestablishedindicatorsofDNAandproteinbindingwaschecked.Thisinviewoftheplethoraofthereportedmechanismsofmicronucleusinduction.Asamatteroffact,micronucleiaremarkersofbothaneugenic(changeinthechromosomesnumber,usuallybyloss)andclastogenic(chromosomebreakage)effects.Itisgenerallyassumedthatsucheffectsaregeneratedthrougharangeofdifferentpathways.Evidence(mainlygatheredfrominvitrostudies)indicatesthatmicronucleicanbeinducede.g.,bytypicalDNA-attackingagents(e.g.,alkylatingagentslikemethylmethanesulfonate),bymitoticspindlepoisons(e.g.,colcemide,vincristine),orbyinhibitorsofcytokinesis(e.g.,cytochalasinB).Thelattereffectsareprobablyduetointerferencewithproteins.Otherchemicalsarethoughttobeclastogenicthroughaspecificdisturbanceofcytokinesisduetolipophilicity(Dornetal.,2007).TherelativeinfluenceofDNAandproteinbindingonmicronucleusgenerationwascheckedbyrecordingthedistributionofstructuralalertsforthetwoeffectsintheLeadscopeinvivomicronucleusdatabase.AsprobesforDNAbinding,weusedthestructuralalertsforcarcinogenicity/mutagenicityimplementedinToxtree1.51.Asamatteroffact,thelargemajorityofthesealertsrefertogenotoxiccarcinogenicity,whichisassumedtobecausedthroughdirectinteractionwithDNA(BenigniandBossa,2008).Asprobesforproteinbinding,weusedthealertsimplementedintheOrganisationforEconomicCooperationandDevelopment(OECD)QSARToolbox.8Thesealertsweremainlydevelopedfromthemechanisticknowledgeonskinsensitization,andmodelthecovalentbindingtoproteins.TheresultsoftheaboveanalysisisdisplayedinFigure1asaROCgraph.Itappearsthatthestructuralalertsforcarcinogenicity/mutagenicitycorrelatetosomeextentwiththeinductionofmicronuclei,whereasthoseforproteincovalentbindingshownocorrelation(inthegraph,theyareonthediagonallinewhichrepresentsrandomresults).8http://www.oecd.org/document/23/0,3343,en_2649_34379_33957015_1_1_1_1,00.html13Figure1.ReceiverOperatingCurveshowingtheconcordanceoftwosetsofstructuralalertswiththeresultsoftheinvivomicronucleusassay(SA_BBreferstotheBenigni-BossaalertsinToxtree;SA_ProtreferstothealertsforproteinbindingintheOECDQSARToolbox)3.3StructuralAlertsforinvivomicronucleusassaySincetheaboveanalysespointedtogenotoxiceffectsasanimportantdeterminantofmicronucleiinduction,wedevelopedthelistofStructuralalertsforinvivomicronucleususingthecarcinogenicity/mutagenicityalertsinToxtreeasacore,andthensearchingforadditionalsubstructuresspecifictothemicronucleus-positivechemicals.FromtheToxtreealertsforcarcinogenicity/mutagenicity,weexcludedfouralertsspecificfornon-genotoxicmechanismsofcarcinogenicity.UsinglineardiscriminantanalysisasananalyticaltoolandROCplotsasagraphicaltool,aseriesofadditionalsubstructureswereadded/removedto/fromtheToxtreealertsinordertoincreasesensitivityandspecificity.Intheseexploratoryanalyses,we14screenedtheverylargecollectionofsubstructuralpatternsandfunctionalgroups(morethan27,000)containedinthesoftwareLeadscopeEnteprise2.4.15-6.Wealsore-checkedtheToolboxproteinbindingalertsforindividualsubstructuresrelatedwithmicronucleusinduction.TheresultistheoptimizedlistofalertsinAppendix1.TogetherwiththeToxtreealerts,itcontainsfiveadditionalsubstructuresidentifiedinthecourseofthisresearch.Forthesakeofclarity,thecodesofthealertsinToxtreearemaintained,whereasthefiveadditionalalertshavenewcodes.Figure2displaystheagreementbetweenthealertsforinvivomicronucleus,andtheexperimentalresultsforthisendpoint.Outof547negatives,thespecificityoftheSAsis0.57.Thesensitivityis0.65outof182positives.Theoverallaccuracyis0.59.Foracomparison,theROCgraphshowsthenewlydevelopedalertsformicronucleustogetherwiththoseforDNAandproteinbinding.ItappearsthattheperformanceofthefinallistofalertsisconsiderablyhigherthanthatoftheDNAbindingandProteinbindingalerts.TableIIIgivesthetruepositiveratefortheindividualalerts.15Figure2ReceiverOperatingCurveshowingtheconcordanceofstructuralalertsfortheinvivomicronucleusassaywiththeexperiemtnalresultsforthisassay(SA_MicreferstotheinvivomicronucleusalertsinToxtree)16TableIII:CharacterisationofStructuralAlerts.STRUCTURALALERTNo.SubstancesfiredNo.ofpositivesubstancesTruePositives(%)SA_1:acylhalides00SA_2:alkyl(C<5)orbenzylesterofsulphonicorphosphonicacid4375SA_3:N-methylolderivatives100SA_4:monohaloalkene33100SA_5:SorNmustard44100SA_6:propiolactonesorpropiosultones00SA_7:epoxidesandaziridines201260SA_8:aliphatichalogens35926SA_9:alkylnitrite11100SA_10:a,bunsaturatedcarbonyls581628SA_11:simplealdehyde9222SA_12:quinones9444SA_13:hydrazine600SA_14:aliphaticazoandazoxy00SA_15:isocyanateandisothiocyanategroups200SA_16:alkylcarbamateandthiocarbamate9222SA_18:PolycyclicAromaticHydrocarbons11100SA_19:heterocyclicPolycyclicAromaticHydrocarbons700SA_21:alkylandarylN-nitrosogroups6583SA_22:azideandtriazenegroups22100SA_23:aliphaticN-nitrogroup215017SA_24:a,bunsaturatedaliphaticalkoxygroup11100SA_25:aromaticnitrosogroup00SA_26:aromaticringN-oxide00SA_27:nitro-aromatic17212SA_28:primaryaromaticamine,hydroxylamineanditsderivedesters501938SA_28bis:aromaticmono-anddialkylamine5240SA_28tris:aromaticN-acylamine200SA_29:aromaticdiazo8450SA_30:coumarinsandFurocoumarins300SA_32:1,3-dialkoxy-benzene6583SA_33:1-phenoxy-benzene5480SA_34:hacceptor-path3-hacceptor1635534SA_35:cxolane21943SA_36:carbodiimides221003.4FurtheranalysesonthealertsformicronucleusAstrikingevidenceinTableIIIistherelativelylowpercentageoftruepositivesidentifiedbymanySAs.Inotherwords,oftenthetoxicpotentialofthealertsisnottranslatedintoactualtoxicityintheexperimentalsystem.Foracomparison,theTruePositiveRateofthevariousalertsformutagenicity/carcinogenicityinToxtreeisremarkablyhigher,rangingfrom70to100%(BenigniandBossa,2008).TheaboveresultcontributestobetterunderstandtheevidenceinTablesIandII,whereitappearsthatthemicronucleusassayhasmanymorenegativesthanthecarcinogenicitybioassayandtheSalmonellamutagenicitytest.TableIIIindicatesthatthelowsensitivityofthemicronucleusassayislargelyduetothefactthatoften,18chemicalfunctionalitiesandsubstructureswhicharesupposedtobereactivedonotexerttheirpotentialreactivityinthisexperimentalsystem.Theissueofthelowsensitivityofthemicronucleusassayhasbeenrecognizedbyscientistsinvolvedinresearchaimedatimprovingtheavailableshort-termmutagenicityassays;asamatteroffact,validationoffurther,moresensitiveinvivoassays(e.g.,invivoCometassay)ispresentlyinprogress(KirklandandSpeit,2008).Inthecontextofthisresearch,weinvestigatedifageneraleffectofbioavailabilityonthelimitedsensitivityofmicronucleuswasapparent.Tothisaim,weconsideredtwochemicaldescriptorswellknownastobelinkedtobioavailability:logP(hydrophobicity)andMolarRefractivity(MR)(HanschandLeo,1995).ThetwodescriptorswerecalculatedwiththeC-QSARsoftware(Daylight,Inc.)9forallthechemicalsinthemicronucleusdatabase.Forthetwoparameters,TableIVreportstherangesofvaluesforpositiveandnegativemicronucleusresults.TableIV:RangesofC-logPandC-MRinchemicalsassayedwiththemicronucleustestC-logPC-MR(x10-1)MicronucleusNegatives-18.64–20.430.10–33.73MicronucleusPositives-9.58–15.230.15–32.91ItappearsthatthemicronucleuspositivescoveramorelimitedrangeoflogPvaluesthanthemicronucleusnegatives;however,theconsiderationofexclusionvaluesforlogPincombinationwiththeSAsdidnotimprovetheoverallperformance(resultsnotshown).WhereasnogeneraleffectoflogP(orMR)wasfound,analysesontheindividualchemicalclassesshowedthatlogPcut-offscanbeidentifiedfortheclassesofNitroaromatics(NegativesatlogP>0.0),AromaticDiazo(NegativesatlogP<3.7),9http://www.daylight.com/about/index.html19andOxolanes(NegativesatlogP>1.5).Theconsiderationofthesecut-offsincreasesthespecificityoftheSAsfrom0.57to0.60.Theaboveresultsuggestsapossiblestrategytounderstandandmodelingthemanynegativeresultsobservedwiththemicronucleus.Sincethebonemarrow(maintargetofthetest)isanorganeasilyaccessiblebythebloodstream,itcanbehypothesizedthatthelackofeffectshownbyseveralchemicalswithSAs(hencepotentiallyreactive)isduetothemanypossibletargetsforreactionencounteredintheinvivosituation;thisdiminishestheprobabilityforthechemicalsofreaching,andinteractingwiththemoleculartarget(s)ofthemicronucleustest.Forexample,highlyreactivechemicalswillprobablyreactwithanytargetencounteredintheirway(e.g.,proteins,water)beforereachingthebonemarrow.ThusitcanbeenvisagedthatQSARsforindividualchemicalclassesshouldbedeveloped,andthattheyshouldconsiderparameterslinkedtochemicalreactivity(suchasHOMOandLUMOenergies).ItcanbehypothesizedthatthemodelsderivedfromtheseQSARswillcontributetomodulatetheindividualSAs.204.FinalconsiderationsStructuralalert