• Keine Ergebnisse gefunden

The ChemScreen project to design a pragmatic alternative approach to predict reproductive toxicity of chemicals

N/A
N/A
Protected

Academic year: 2022

Aktie "The ChemScreen project to design a pragmatic alternative approach to predict reproductive toxicity of chemicals"

Copied!
10
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

The ChemScreen project to design a pragmatic alternative approach to predict reproductive toxicity of chemicals

Bart van der Burg

a,∗

, Eva Bay Wedebye

b

, Daniel R. Dietrich

c

, Joanna Jaworska

d

, Inge Mangelsdorf

e

, Eduard Paune

f

, Michael Schwarz

g

, Aldert H. Piersma

h,i

, E. Dinant Kroese

j

aBioDetectionSystemsBV,Amsterdam,TheNetherlands

bDanishTechnicalUniversity,Lyngby,Denmark

cUniversityofKonstanz,Konstanz,Germany

dProcter&Gamble,Brussels,Belgium

eFraunhofer-ITEM,Hannover,Germany

fSimpple,Tarragona,Spain

gUniversityofTübingen,Tübingen,Germany

hNationalInstituteforPublicHealthandtheEnvironment,Bilthoven,TheNetherlands

iUtrechtUniversity,Utrecht,TheNetherlands

jTNO,Zeist,TheNetherlands

Keywords:

Invitro Insilico Integratedtesting Readaccross Reproductivetoxicity Endocrinedisruption

a b s t r a c t

Thereisagreatneedforrapidtestingstrategiesforreproductivetoxicitytesting,avoidinganimaluse.

TheEUFrameworkprogram7projectChemScreenaimedtofillthisgapinapragmaticmannerprefer- ablyusingvalidatedexistingtoolsandplacetheminaninnovativealternativetestingstrategy.Inour approachwecombinedknowledgeoncriticalprocessesaffectedbyreproductivetoxicantswithknowl- edgeonthemechanisticbasisofsucheffects.Weusedinsilicomethodsforprescreeningchemicals forrelevanttoxiceffectsaimingatreducedtestingneeds.Forthosechemicalsthatneedtestingwe havesetupaninvitroscreeningpanelthatincludesmechanistichighthroughputmethodsandlower throughputassaysthatmeasuremoreintegrativeendpoints.Insilicopharmacokineticmoduleswere developedforrapidexposurepredictionsviadiverseexposureroutes.Thesemodulestomatchinvitro andinvivoexposurelevelsgreatlyimprovedpredictivityoftheinvitrotests.Asafurtherstep,wehave generatedexampleshowtopredictreproductivetoxicityofchemicalsusingavailabledata.Wehave executedformalvalidationsofpanelconstituentsandalsousedmoreinnovativemannerstovalidate thetestpanelusingmechanisticapproaches.Weareactivelyengagedinpromotingregulatoryaccep- tanceofthetoolsdevelopedasanessentialsteptowardspracticalapplication,includingcasestudiesfor read-acrosspurposes.Withthisapproach,asignificantsavinginanimaluseandassociatedcostsseems veryfeasible.

1. Introduction

Thecomplexityofthesystemofriskassessmentofchemicals doesnotallowrapidevaluationofbasictoxicityprofilesofchemi- cals.Becauseofthis,thetoxicologicalpropertiesofmostindustrial chemicalsthatareincommonusearelargelyunknown[1].New legislationsuchas REACHaims toend this unacceptablesitua- tionbymodernizingandstreamliningchemicalriskassessment[2].

Correspondingauthor.Tel.:+31204350750;fax:+31204350757.

E-mailaddress:bart.van.der.burg@bds.nl(B.vanderBurg).

However,thisapproachisunlikelytobesuccessfulwithoutincor- poratingalternative,integratedtestingstrategiesinwhichcostly andtime consuminganimaltestsarereplacedtoalargeextent byrapidandcost-effectivealternativetestingmethods.Aninte- gratedtestingstrategy (ITS)isamethodtocombinetestresults fromdifferentsources, includingnon-testinformation (existing data,insilicoextrapolationsfromexistingdataormodelling)to giveacombinedtestresult[3]Inclusionofinsilicoandinvitrotest resultsandpre-existingdatainprincipleallowsmorerapideval- uationoftoxicologicalpropertiesofchemicals.Suchmethodsare particularlyneededforreproductivetoxicitytestingofchemicals.

Reproductivetoxicityassessmentisimportantforbothmanand

Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-0-291169 Erschienen in: Reproductive Toxicology ; 55 (2015). - S. 114-123

https://dx.doi.org/10.1016/j.reprotox.2015.01.008

(2)

theenvironmentandusesrelativelylargenumbersofexperimental animalsandtime-andresourceintensivetestingprocedures[4,5].

Unfortunately,thereareveryfewifanyalternativemethodsthat areabletocoverthiscomplexity.TheChemScreenprojectaimed tofillthisgapandselectsuitabletestsandplacetheminamore generalalternativetestingstrategy.

Becauseofthecomplexityoftheprocessofmammalianrepro- duction intact organisms are often regarded to be essential in assessingreproductivetoxicity ofcompounds.Eventhen,it has beenshownthatlargespeciesdifferencesexistandinterspecies extrapolation of developmental toxicity typically usually is not muchhigherthan60%betweensinglespecieswhenusingapical endpointsin animals[6].Nevertheless,resultsfromtheFrame- workprogram(FP)6programReProTectveryclearlyshowedthat an in vitro test battery covering only part of thereproductive cycleprocessescanprovideverypromising resultswithrespect toreproductivetoxicitytesting [7].It hasbeenarguedthatthe use of in vitro tests that assess mechanisms of toxicity may formthebasisfora newparadigmintoxicitytestingand could alsoimprovethepossibilitytoextrapolatebetweenspeciessince pathwaysof toxicity share many similarities betweendifferent species [8,9]. This mechanism-based approach of toxicity test- ingwasoneofthecornerstonesoftheChemScreenprogram.In 2005wealreadydevelopedapanelofmechanism-basedCALUX assays to assess hormonal activityof compounds [10], a panel which has shown tobe highly predictive for suchactivities in experimentalanimals[11,12].Someofthesemechanisticassays alsoformedapartofthepromising batteryoftestsusedinthe ReProTect[7].

WiththisinmindtheChemScreenprojecthasbeendesigned, aiming at further simplification of the screening battery by increaseduseofpreferablyhighthroughputmechanisticscreening assays[13].Weaimedtogeneratearapidscreeningsystem,that isrelativelysimple,cost-effective,andcanbeimplementedprefer- ablywithinthetight timescheduleoftheREACHprogram.The toolshouldbeflexibleandadaptableforapplicationsbeyondthis specificprogram,suchasforsafedesignpurposes,orprioritiza- tion.Tofillthegapofsuitablealternativemethodsforreproductive toxicitytestingweusedanovelhighthroughputapproachcom- bining insilico/invitromethods. Inthis approach wecombined knowledge of critical processes affected by reproductive toxi- cantswithknowledge onthemechanisticbasis of sucheffects.

Toreachourgoalsthefollowingobjectiveswereformulated,cor- respondingtothemajorChemScreenworkpackages (WPs;see Fig.1):

1.Establishinsilicoprescreeningmethodsprioritizinginvitrotox- icitytesting.

2.Establishdatabasesandinsilicomethodstoidentifypotential reproductivetoxicants.

3.Establishmentofsensitiveparametersandmediumthroughput invitroassays.

4.Establish a high throughputmechanistic pathway screen for reproductivetoxicants.

5.Establishmethodstopredictinvivoreprotoxicity.

Furthermoreampleattentionwaspaidtovalidationandappli- cation of thetest methods developed. The ChemScreen project startedatJanuary1,2010.Theprogramaimedtocollaboratively generateaninnovativetestingstrategycombininguniqueexper- tiseoftheparticipants.Toattainthislevelofinteraction,frequent meetings, and workshops were held, and a highprofile Scien- tificAdvisoryBoardwasinstalled tohelpguidethis process.In theinternationaladvisoryboardmajorstakeholders(JRC/ECVAM, OECD, US EPA, Industry, ECETOC) were represented. Beginning 2010theChemScreenprojectenteredanimportanttransatlantic

Fig.1.GraphicalrepresentationofChemScreen’smajorcomponents.Thedottedlines indicateareaswherealternativemethodsarerelativelywelldevelopedorunder constructioninvarious(FP)programsinwhichpartnersparticipate.Allworkpack- ages(WPs)areindicated,exceptWP6(ICTtools),WP7(dissemination)andWP8 (management).

collaborationwiththeUSEPANationalCenterforComputational Toxicology (NCCT), and the Texas Indiana Virtual Star Center (TIVSC).Agreementsonscientificcollaboration,data-andchemical sharingwereestablished.Herewewillpresentasummaryofobjec- tivesanddiscussobtainedresultswithintheworkpackages,and theoverallpicturethatemergesfromthis.Specificmaterialsand methodologiescanbefoundinthereferencedindividualpapers withinthisissue.

2. Resultsanddiscussion

2.1. Insilicoprescreeningmethodsprioritizinginvitrotoxicity testing

AcrucialaspectoftheREACHprogramisprioritization[2].With, asitnowturnsout,about50,000chemicalsneedinganupdateof toxicologicalinformationwithinadecade,thepotentialamountof safetytestingisenormous,whichrequiresthatchemicalsofhighest concernareidentifiedforprioritizedtesting.Anelaboratequantita- tivestructure-activityrelationship(QSAR)moduleaimedtoallow predictionofthe“leadinghealth effect”asrequestedinREACH, avoidingunnecessarytestingofcompoundsforwhichtheleading healtheffecthasbeenestablished.Forexample,compoundswith mutagenicpropertieswillnotbeevaluatedforreproductivetoxic- itysincefurtherinformationonitspotentialcarcinogenicityisof higherprioritywhenconsideringleadinghealtheffects.Therefore, inthispartoftheprojectweconcentratedontheestablishment andselectionofinsilicoprescreeningmethodstocategorizechem- icals, using QSARs. This wasdone for major classes of toxicity prioritizedinREACH(carcinogenesis,mutagenesisandreproduc- tive(CMR))toxicity[2]. Positivepredictionsforcarcinogenicity,

(3)

Fig.2. Predictions usingQSAR analysisusingpre-screeningtools for human health effectsapplied to over 70,000 REACHpre-registrationchemicals[Wedebye etal.,2015].

mutagenicityand reproductivetoxicity canbeappliedto avoid unnecessaryanimaltestingofcompounds.Asnotestingforrepro- ductiveeffectsshouldbeperformedinREACHonknowngenotoxic carcinogensor germ cellmutagens withappropriate risk man- agementmeasuresimplemented,asstatedintheREACHannexes VIII-X,predictionsforgenotoxiccarcinogensandgermcellmuta- genscanfurthermorebeappliedtoavoidtestingforreproductive toxicity.TomaketheQSARscreeningascomprehensiveaspossi- bleitwasbasedonasmanyaspossibleofthe143,835chemicals, whichwerepre-registeredbeforeDecember2008.Thepredictions werebasedonstructureinformationgeneratedbytheComputa- tionalToxicologyGroupwithintheJointResearchCenter(JRC)for 80,413pre-registeredsubstances(PRS),ofwhich70,983chemicals weresuitable forgenerationofQSAR predictionsin theapplied QSARsoftware[14].ResultsareshowninFig.2showthatmore than25%ofthechemicalshavestructuralalertsofgenotoxiccar- cinogensandgermcell mutagens,which willexcludethem for beingtestedforreproductivetoxicity.Aspartoftheinvitropre- screen,it wasalsoenvisagedtodevelopanexposuremoduleto identifychemicalsthatwillhavenooranegligiblesystemicavail- ability (and anticipating that gonads and embryos will not be exposed)afterexposureviadifferentroutes.Tothateffect,liter- aturesearcheshavebeenperformedtoinvestigateononehand whetherspecificchemicalpropertiescanbeidentifiedtopredict internalexposureafterexposureviatheoral,respiratoryandder- malroutes,and ontheotherhandwhich(publiclyavailable)in silicotoolsexistleadingtoreliablepredictionofnegligibleinter- nalexposure.Basedonthecurrentknowledge,nocut-offsbased onphysicochemical properties couldbe identifiedfor thethree mainexposureroutesbeloworabovewhichinternalexposureis negligible[Buistand Verweij,unpublishedresults].REACHdoes providephysicochemicalcriteriatowaivetoxicitytestingviathe respiratoryroute,butthesecriteriacannotwaivetestingviaother routesofexposure.Therefore,itwasconcludedthatinclusionof thesecriteriainamoduletopredictinternalexposureaimedat excludingreproductivetoxicitytestingbasedonphysicochemical criteriaisnotaviableoption.Instead,thepharmacokineticmodels willnowbeusedexclusivelyinextrapolationofinvitrotoinvivo data.

2.2. Establishdatabasesandinsilicomethodstoidentify potentialreproductivetoxicants

Anextstepinourstrategywastofurtherexpandthepossibili- tiesofidentifyingpotentialreproductivetoxicantswithoutfurther testing.Databases wereexpandedthat categorizetheeffects of knownreproductive toxicants. These databases can be usedto derive so-called read across information that allows extrapola- tionfromaknowntoxicanttootherrelateduntestedmolecules.

Thiscan be complemented with otherin silico,or in vitro test information onthesecompounds. We expandeddatabases that containinformationrelevanttoreproductivetoxicityintheRep- Dose(repeateddose)[15]andFeDTex(fertilityanddevelopmental toxicity)databases[16].First,reproductivetoxicitystudieswere identifiedfrompeerreviewed publications.Allavailablestudies onreproductivetoxicitywereselectedfordataentry.Inparticu- lar,theamountofchemicalsinFeDTexweremorethandoubledin thecourseofChemScreen,mountinguptoover300.Theexpanded databaseswereusedtoidentifyredundantdatacomingfromani- malstudiesandthuspossibilitiestoreducetesting.Evaluatingthe responsivenessofthedifferentgenerations,acomparisonofthe doselevelsattheNo-Observed-EffectLevel(NOEL)wasperformed.

ThesedatasuggestatendencyofF0andF1beingmostlyequally sensitiveinone-andmulti-generationalstudies,whereasF2seems tobelessindicativeforNOELderivation[16].Thisconfirmsearlier workon2-generationstudydatabasesbyPiersmaetal.[17].

AsignificantoverlapwithRepDosewascreatedtoallowcom- parisonofinvivodatawiththesamechemicalandevaluatepossible predictivityofrepeateddosetoxicityforreproductivetoxicity.Tar- getorgansandeffectsobservedwithinrepeateddosetoxicitywere exploredforanyrelationshiptoeffectsoccurringinreproductive toxicitystudies.Whetherandhowminimaleffectconcentrationsof repeateddosetoxicitycanbeextrapolatedtoreproductivetoxicity appearedtodependonthechemicalclassofthecompoundsstud- ied.Forexample,noneoftherepeateddosestudiesonchlorinated phenols(includingphenolasparentalsubstance)providedanindi- cationforreproductivetoxiceffects.,Themajoreffectsobserved afterrepeatedexposurewerebodyweightlossrelatedtoreduced availabilityofenergyandincreasedcatabolismaswellaschanges

(4)

inliverweightandpathology.Thereproductivetoxicityobserved mightalsoberelatedtoinsufficientenergysupplyduetouncoupled oxidativephosphorylationresultingingrowthretardationoreven increasedprenatalmortality.Themechanismof toxicityis thus likelytobethesameinrepeateddoseaswellasreproductivetoxic- itybutwithoutmechanisticknowledgeitisdifficulttopredictsuch effectsbasedofrepeateddosetoxicityobservationsonly.Foralco- holicsolvents,phthalatesandglycoletherscomparableconclusions havebeendrawn[Batkeetal.,unpublishedresults].

UsingthestructuraldataofthePRS(seeSection1)areproduc- tivetoxicityscreenwascarriedoutaswell,howeveronlycoveringa smallpartofreproductivetoxicity(teratogeniceffectsbyalimited numberofmechanisms).Thelimitedamountofcompoundsavail- ablefortrainingandvalidationincombinationwithmanydifferent mechanismsandstructuralalertshamperdevelopmentofgener- allyapplicableQSARmodelsinthisfield[18].

Severaltoolsweredevelopedthatcanassisttopredictingenvi- ronmental reproductive toxicity of chemicals on a mechanistic basisaimingatmorestraightforwarddatainterpretation[Dietrich et al., unpublished results]. By focusing on molecular mecha- nismsconservedinvariousspecies(e.g.variousreceptor-mediated processes) possibilities wereexplored toextrapolate molecular screeningdatatoeffectsinaquaticorganisms(particularlyfish)and thusforecotoxicologicaleffects.Forthis,extrapolationmethods havebeendesigned toestablishifthereareconserved molecu- larmechanismsinaquaticorganismsresponsivetotheactionof differentchemicals. Asone of thefirst stepsanexpansion and restructuringof thelargeEDUKON ecotoxicology database[19]

hasbeencarriedoutaswellasaninventoryofpossiblechemo- biological interactions thatcan berelated todistinct molecular mechanisms.Allrelevanttechnicalprerequisitestoscreenandto usetheEDUKON DBasintended withintheprojectdescription werefulfilled,andtheintegrationofapproximately1400newdata setsof peer-reviewedstudies intotheDB wasaccomplishedin accordancewiththeintentiontointegratethenewest scientific findingsintotheDBandtosetittothecurrentstateofscience.

Theheterogeneityofthedatainvolvingmanyspeciesandeffects togetherwiththecomplexitiesofreproductivetoxicity,however, hamperedthegenerationoffirmconclusionsontheapplicability ofmechanisticinsightsinenvironmentalriskassessment[Dietrich etal.,unpublishedresults].

2.3. Establishsensitiveparametersandamediumthroughput invitroassaypanel

Theaimofthispartoftheprojectwasthedesignofarepro- ductivetoxicityscreen,consistingofanumberofinvitroassays thatarerepresentativeofthoseparametersinreproductivetox- icitythatarecrucialforreproductivehazardassessment.Oneof thestartingpointswasarathercomplexbatteryoftestsoriginally developedintheFP6ReProTectprojectusingprimarymaterialand cellsfromvarious sources(e.g.ratembryos,oocytes,spermato- cytes,etc)[7].Thefirstapproachin simplificationofthetesting battery was directed towards inclusion of integrative medium throughputassays thatcanberunin aroutinelaboratoryenvi- ronment,avoidingassayswithdifferentiated(primary)tissuesas presentintheReProTectbattery.Thiscanthenbecombinedwith morehighthroughputassays(seenextsection)therebyenhanc- ingthroughputandfurthersimplifyingtestprocedures[20].The nextstepwouldbetovalidatethis batteryand selectthemost promisingassaysneededtogiveproperpredictionsofreproductive toxicityofchemicals.Ourapproachinthatsenseisanintermedi- atebetweenthisReProTectapproachandthatofToxCast,which initiallyfocused ona much largervarietyof lessselective high throughputscreening(HTS)assays[21].Weexpectedthatasmaller batterywouldreducenoiseandallowmorestraightforwarddata

interpretation.Validationstepssubsequentlycanleadtoidentifi- cationofmissingendpoints,andthroughinclusionofadditional teststhisbatterycanbefurtherimprovedwithtime.

Atthestartoftheprojectweenvisionedthatonewaytosimplify testingschemesistheapproachofidentifyingcriticalendpoints forreproductivetoxicityusingliteratureand databasesearches, andfocusingondevelopmentoftestsfortheseendpoints.Sofar mostendpointsidentifiedwerequitegenericandnon-specificin nature,whichmaymeanthatscreeningpanelsshouldalsoinclude quitesimple,relativelynon-specifictests(cytotoxicity)inaddition to more mechanisticassays. We have therefore includedsepa- ratecytotoxicityendpointsinallourscreening assaysandhave extendedthenumberofmedium-andhighthroughputmechanis- ticscreeningassays.

Firstofall,progresswasobtainedwiththeoptimizationand amendmenttohigherthroughputofteststhataimedtobeincluded inthefinalbattery,namelyagenomics-basedimprovedembry- onicstemcelltest(EST),thezebrafishembryotoxicityassay(ZET) and steroidogenesis assays, in particular four CYP17/19 assays.

In theembryonic stem cell test,transcriptomics readouts were employedtoimprovethemechanisticreadoutofthesystemand improveitspredictivevalue.TheCYP17/19assayswerebeingopti- mizedandmicrosomesversuswholecellassayswerecompared fortheirpredictivevalue[22].Theembryonicstemcellassaywas appliedtothecategoryofphthalatesandpredictivitywithinthis class of compoundsona comparativebasis withexisting tran- scriptomicspatternsof otherphthalates wasshown[23,24].All above-mentioned tests were included in a first, relatively low throughputversionoftheChemScreenbattery[25].

2.4. Establishahighthroughputmechanisticpathwayscreenfor reproductivetoxicants

Amajorfocuswasontheestablishmentofmechanistictests andadaptingthemtohighthroughputscreeningmethodspredic- tiveofreproductivetoxicpotential,basedoninsightinmolecular mechanismsthatarerelevanttoreproductivetoxicity.Theseassays compriseapanelofhighlyspecifichumanCALUX®reportergene assays,whichincludesaselectionoutofabout50differentassays.

ThispanelisbasedonthehumanU2-OScelllinewhicharemod- ifiedtoexpressreportergenesforseparatepathways[10,11].The highselectivityisreachedthroughtheuseofreportergeneswith syntheticpromotersconsistingofminimalpromoterelementscou- pledtomultimerised,highlyselectiveresponseelementsdriving theexpressionofthereportergeneluciferase.Whenrelevantto thepathwayofinterest,specificreceptorsareintroducedtoacti- vatethecognatereportergene.Thepanelofselectiveassayswas designedtocreateverylowlevelsoffalse-positives.Thepanelcon- tainsasinglereportergeneassayperpathway.Thisisdifferentin themuchlargerToxCastassaypanelthattypicallyincludesmulti- pleassaysperendpoint[26].

Asystematicapproachhasbeentakendevelopingthescreening panelusingverysensitiveandselectivereportergenes,covering a wide range of receptors and signaling pathways that poten- tially are involved in reproductive toxicity [20]. The approach takenaimedtoprovideseveraladvantages ontopofalow rate offalsepositivity.First,byprovidingamechanisticbasisregula- toryacceptanceshouldbefacilitated.Second,thismechanisticbasis willprovideinputfordecisionsonfurthertestrequirementsand riskassessment.Third,cost,speed,robustness,andquantification areadvantageousinthesetypesofassays.Finally,thisapproach facilitatespredictionsforbothhuman-andecotoxicologicalprop- erties ofchemicals,when differentprediction modelsare used.

Oneapproach takentodevelop additional reportergene assays forChemScreenistousecentralintracellularpathwaysthatare involved inregulatingtranscriptionin amammaliancell.These

(5)

pathwaysincludebesidesnuclearhormonereceptorsalsoreporter geneassaysforquitegenericresponsestotoxicants,includingcyto- toxicity,apoptosisandgenotoxicity.Theseresponsescanalsobe relevantinreproductivetoxicity,however.Acontrolcellline,called cytotoxCALUXhasbeengeneratedwhichconstitutivelyexpresses theluciferasegene[27].Tovalidatethisapproachofusinglimited setsofselectiveassaysitwasshownthatasingleselectiveand extensivelyvalidated CALUX assay(ERalphaCALUX)gave com- parableresultsasabattery ofToxCasttestsmeasuringestrogen receptorpathway-mediatedendpoints[VanderBurgetal.,unpub- lishedresults].

InadditiontothesehighlyselectiveCALUXHTSreportergene assays, other assays have been established in more complex, murineembryonicstem(ES)cellsbyintroducingreportersystems forsignalingpathwayscontrollingkeydifferentiationpathwaysin embryonicdevelopment(so-calledReProGloassays)[28].Despite severalattemptstointroducereportergenesotherthanthatmea- suringwnt-pathwayactivation,asusedintheoriginalReProGlo assay,noneofthemsucceeded[Uibeletal.,unpublishedresults].

AllCALUXandReProGloassayshavenowbeenautomatedin 384and 96wellsformat,respectively,anda seriesofreference compoundshasbeenusedtodeselectlessinformativeassaysand identifymissingendpoints.Datastorageandanalysishasbeenset upandmorethan250compoundshavebeenscreenedintheCALUX panelandasubsetintheReProGloassay.Becauseoftheneedofa practicaltooltostoreandprocessthelargeamountofhighthrough- putdatathispartofthetoolwasgenerated.Itprovidesthefeatures requiredtostore,manageandquicklyanalyzeassaydatagenerated inhighHTSbioassaysallowingaccesstoprimarydata.Thistoolwas successfullyconstructedandinstalledandallCALUXHTSscreening datahavebeenuploaded.Finally,theCALUXHTSpanelendoge- nouslyexpresseslittlemetabolicactivitybutcanberunwithand withoutS9metabolicfractions[27].

Ourapproachfromthebeginningonwardsfocusedoninclusion ofalimitednumberoftestswithrelevantmodesofaction.Thisis differentfromtheToxCasttestpanelthatincludesawiderrange ofmechanismswithmultipletestspermechanisticendpoint[21].

Resultssuggestthatitmaynotbeneededtoincludea rangeof assaysforeachendpointtomakegoodpredictionsofinvivoactiv- ityofchemicals,aswasshownfortheestrogenreceptorreporter geneassays [20].Thisis consistentwiththeearliernotionthat transactivationofERsinoneassay,ERsubtypeorspeciescanbe extrapolatedtootherspeciesorsubtypesofERsforthepurposeof chemicalscreeninginthecurrentpracticeofriskassessmentrely- ingonrelativelyuninformativeanimaldata[29].Alsoinapanel theassays performedverywellpredictingreproductivetoxicity inmoregeneralterms[25]whichagainsuggeststhatarelatively simplescreeningpanelcansufficeforthispurpose.

2.5. Establishmethodstopredictinvivoreprotoxicity

Asanextstepinourtestingstrategywehavebeenestablishing integrativemethodstopredictinvivoreproductivetoxicitycon- centrationsusinginvitrobenchmark(threshold)concentrationsas astartingpoint,alsoreferredtoasreversedosimetry[30].Predic- tionofthecorrectinvivodoselevelatwhichadverseeffectscanbe expectedisoneofthekeyissues.Highthroughputpharmacokinetic modelshavebeensetupdescribingthebioavailabilityviarelevant routes(oral,dermal,inhalation)andthepharmacokineticsofthe mostrelevantreferencechemicalclasses.ArapidPBPKmodeling strategyhasbeendevelopedtotranslateinvitroconcentrationsto invivomaternaldoselevels([31];Fig.3).Insteadofbuildingcus- tommodelsforeachchemical(class)asplanned,itwaschosento developonegenericPBPKmodel frameworkdescribingallrele- vantcompartments,physiologicalprocessesandexposureroutes.

Thegenericmodelwithits(chemical-independent)anatomicaland

physiologicalparametervaluesinprincipleappliestoallchemicals.

Thepharmacokineticsofspecificcompoundshasbeenaddressed byprovidingchemical-specificparameters(relatingtopermeabil- ity,partitioningandclearance)thatcanbeobtainedbyeitherin silicopredictionsorroutineinvitromeasurements,asinputtothe genericmodel.Tothisend,analgorithm hasbeenimplemented topredictthebindingandpartitioningbehaviorofcompoundsin water,proteinandlipidphasesofblood(plasmaandredbloodcells) andtissues(interstitialfluidandcells).Thisalgorithmtakesasinput readilyobtainablechemicalproperties.Itcanalsobeusedtopredict bindingandpartitioningbehaviorinculturemediumifitspHand proteinandlipidfractionsareknown,andthustocorrectfordiffer- encesinunboundconcentrationsbetweeninvitroandinvivo.This genericframeworkismorecompatiblewiththerapidscreening characteroftheChemScreentoolasawhole,andallowspharma- cokineticpredictionofamuchlargernumberofcompoundsina relativelyshorttime[31].

In addition, to assess dermal exposure routes a specialized methodologyforinvitro—dermalinvivoextrapolationwasdevel- oped [32]. The method enables the calculation of an external exposuredoseleadingtosystemictoxicitybycombininginvitro experimentaldatarelevanttothetoxicantaffectedpathwayswith insilicopredictionsof skinpermeationand whole-body ADME.

Themethodisinlinewiththecurrentdevelopmentofmechanis- ticmethodsusedtopredictinvivotoxicityfromsetsofreference data,whichmaycombinedatafrominvivoexperiments,invitro cell-basedassaysand/orcomputationalresults.Duetoincomplete understandingofthemechanismsoftoxicityanddatalimitations, determinationoftherepresentativeconcentrationinvitrowillmost likelyneedtorelyonastatisticalapproachofchoosingthelowest valueofEC50distribution.Theuseoftheinternalsystemicconcen- trationexternaltoestimateasafedosehasbeencomparedtoan approachbasedoninvivodose(LOEL).ResultssuggestthatLOELs maybetooconservativeforuseinpredictionoftoxicityandrisk assessment[32].

2.6. Validation

Acceptance byregulatoryagencies is a major bottleneckfor applicationofanyalternativetesting strategy.Therefore,during theprojectampleattentionhasbeenpaidtovalidationandregu- latoryacceptanceofthetestmethodsdevelopedinChemScreen.

Thishasbeendoneinthecontextoftheringtrails,butalsoformal validationsusingtraditionalmethodshavebeenundertaken.Close contactshavebeenestablishedwithrelevantECVAMandOECD workinggroups,scientistsandregulators.Inselectingteststobe includedinthebattery,methodswerepreferredthatwereeither validatedorintheprocessofbeingvalidated,includingtheESTtest [33],thezebrafishembryotest[34],andreportergeneassaysfor estrogens[35,36]andandrogens[37].TheERalpha-andARCALUX assayalreadywereprevalidatedinthecontextoftheReProTectFP6 program,andsubsequentlyhavebeenthisissuetoEURL-ECVAM andOECDtoallowformalvalidation,andregulatoryacceptance [35,36,38,39].Sinceourbatteryoftestsofreportergeneassaysis verycomparabletotheselattertests,thevalidatedtestsmaybe usedas“validationanchors”inascreeningbattery,i.e.givingcon- fidenceintheotherbatteryconstituents.Itwouldalsobepossible todesignmoresimpleandrapidvalidationproceduresforthese additionalbatteryconstituentssincethetestproceduresofallthese testsessentiallyarethesame.

An important element in the project from the beginning onwardswastheexecutionofdedicatedringtrails(alsoreferred toasfeasibilitystudies)totestandimprovethetestbatteryasa whole.Withinthebudgetaryconstraints,wechosetogoforrela- tivelysmalldedicatedtrailsratherthananunbiasedlargescreening withmanyassaysandmanychemicals.Thefirstringtrailusingthe

(6)

Fig.3.SchematicrepresentationofkeyelementsofthehighthroughputPBPKmodel.FordetailsseetextandBosgraetal.,2015.

completepreselectedtestbatteryshowedthattestingalimitedlist ofcompoundsforwhichsufficientinvivoinformationisavailable, canbeavaluablestepintheidentificationofkeyelementsinabat- teryaswellasmissingones[25].Correctpredictionsweremade for11outof12chemicalswhilechemicalthatwasmissedcould beexplained bythelackofrelevantmechanismofactionbeing presentinthetestingbattery.Interestingly,thepredictionsofthe highthroughputCALUXpanelalone,whencombinedwithPBPK modelingwasexactlythesame([25];Table1)

QSAR and related chemoinformatics tools alone clearly are insufficient for reliable predictions of reproductive toxicity of chemicals.Similarly,biologicalmethodsinvitroscreeningmeth- odshavetheirlimitations.Whencombined,predictionscangreatly improve[40].Therefore,weassessedfurtheroptionstocombine structuralpropertiesincombinationwithinvitrodatatoidentify substanceswithreproductivetoxicproperties[41].Byusingthe OECDToolbox[42]andexpertjudgmentweidentified89chemi- calcategories.Inaddition275chemicalswereidentifiedwithany effectonthereproductivesysteminmalesandfemaleanimals.We combinedtheseresultswithalargedatasetonestrogenandandro- genreceptor activityusing CALUXassays todetermineifthese hormonaleffectsarerelatedtothereproductiveaberrations.The invivoandinvitroeffectswerefoundnottobeconfinedtodistinct chemicalcategories,showingthelimitationsofactivitypredictions onchemicalstructureonly.Whilethechemicalclasseswerenot predictiveforaninvivoeffect,aclearcorrelationwasseenbetween theinvivoreproductiveaberrationsandactivityinthereporter geneassayonestrogenicactivity[41].

Afurtherapplicationthatwasexploredusingafeasibilitystudy was the possibility of read across using the ChemScreen bat- tery:whenanuntestedquerychemicalisstructurallysimilarto areproductivetoxicant,thebatterycouldbeusedtoconfirmthese reproductiveproperties,andtoavoidanyfurtherinvivostudies.

Ideally,thisbatteryshouldthenalsobecapabletocorrectlydis- tinctreproductivetoxicantsfromnon-reproductivetoxicants,even whenbothhaveaclearstructuralsimilarity.Thisstudyshowedthat thebatterywasabletomakethisdistinctionusingthreedifferent chemicalclassesasanexample.Thisprovidesimportantopportu- nitiesforapplicationsunderREACHwhereread-acrossprocedures arebeingconsideredasanimportantelementofintegratedtesting toavoidorreduceanimalexperimentation[43].Heretheweight ofevidenceofavailablerepeateddosetoxicitydataforbothsource and query chemical, theirstructural similarity,as wellas their batteryresultscouldbeusedtojustifywaivinganinvivostudy [42].Alternativelyoradditionally,linkagetoregulatoryapplication mayusecombinationofinvitrotestresultswithexistingrepeated dosestudydatapotentialreproductivetoxicantscouldbeidentified aheadofinvivostudies,allowingprioritizationsofanimaltesting [41].

2.7. Mechanisticvalidationandlinkagetoadverseoutcome pathways

Because of the strong interest in molecular screening approaches asdevelopedin ChemScreenand ToxCast,currently there are significant activities ongoing to improve linkage of

(7)

Table1

ResultsfromthefirstChemScreenringtrail(adaptedfromPiersmaetal.[25]).

comp oun d Toxi city in vivo

EST ZET

ReProGlo

Cyp17 Cyp19 CALUX CALUX PBPK

Entire battery

Cyclosporin A

Monoethylhexyl phthalate Sodium valproate D-mannitol Flusilazole Glufosinate ammonium Methoxy acetic acid

Retinoic acid Dioctyltin chloride Endosulfan Diethylstilbestrol Methylmercury chloride

Gray:positiveresult;white:negativeresult;EST:embryonicstemcelltest;ZET:zebrafishembryonictest;CALUX:CALUXHTSpanelofreportergeneassays;battery:

combinedpredictionsofalltests.

mechanisticin vitrotoolsto resultsin animalexperiments, e.g.

throughthedesignofso-calledAdverseOutcomePathways[44].

ChemScreen’sscientistshaveembarkedonthesenewandimpor- tantactivitiesthoughttobeessentialmissinglinksinapplicationof molecularscreeningtools.Wehavebeenactivelyengagedinpro- motingregulatoryacceptance ofthetoolsdeveloped,alsousing alternativevalidationmethodsthatprovidelinkagetothemecha- nismofactionofchemicalsleadingtoadversity.

Tothisend,inanextfeasibilitystudysetsofchemicalswere selectedthatcaninduceeithersexorgan(SO)deformitiesorneural tubedeformities(NTD).Ontheonehandthesewereusedtostudy thedifferentadverseoutcomepathwaysinvolvedinneuraltube defects[45],whileontheotherhandthesesetsofchemicalswere usedtofurthervalidateandspecifytheapplicationdomainsofthe differentelementsinthehighthroughputCALUXtestbattery[20].

AnalysisoftheresultsoftheCALUXhighthroughputpanelshows thatitisverysuitabletoidentifynotonlyendocrineactivecom- pounds,butalsocanpredictchemicallyinducedSOdeformations withover80%accuracy[20].Thisisconsistentwiththestudyof Lewinetal.[41],showingaclearcorrelationbetweentheinvivo reproductiveaberrationsandactivityinthereportergeneassayon estrogenicactivity.

Whenusingmolecularscreeningtoolsitshouldbenotedthat couplingtheseresults1:1toapicalendpointsinexperimentalani- malscannotbereadilydone.Inthefirstplace,theremaybemultiple mechanismsleadingtothesameendpoint,whilethesamemech- anismmaybeinvolvedindifferentendpointtoxicitieswhentime andtissuedistributionisdifferent.Inaddition,asingletestonly givespartofthepuzzleandonlybatteriessuchastheChemScreen batterycanbeeffectiveinreplacinganimalstudies.Thevalidation ofsuchabatteryisachallenge,inparticularintheabsenceofgold standard,causedbyrelativelypoorpredictivityofanimaldatafor

humans[6].Aninterestingapproach inthislightwastocluster apicalreproductivetoxiceffectsofchemicalswhencomparingthis withdataofmolecularscreeningassays,resultinginastrongcorre- lationofrelevantCALUXassayswiththoseclusteredapicaleffects [20,41].Thisverylikelywillchallengethedesignofverysimple linearAOPsand suggestthat networkanalysisand/orclustered AOPelementsmaybeneededtolinkmechanisticteststoapical endpointsinanefficientmanner.

InthiscontextitisinterestingtonotethattheReProTecttest- ing battery was able to provide predictions ongender-specific effectsofthechemicals,andsometimeseventhetargetorgan[7].

Thiscould bedone becauseof inclusion of tests basedon pri- marymaterialofmaleandfemalereproductiveorgans.Noattempts weremadetopredictgender-specificeffectswiththeChemScreen battery,but it islikely tobemore difficultbasedontheapical testsincludedandthemechanistictestsbecauseoftheirlimited genderspecificity.EventheresultsofCALUXassays thatassess sexhormonereceptoractivationwherefoundtoberelevantfor predictingreproductiveeffectsinbothmaleandfemalerodents [41],whichisconsistentwithexpressionandfunctionofARand ERin both sexes. Therefore, although mechanistic tests canbe highlypredictiveforinvivotoxicity,exactpredictionoftheapi- calendpointsthatwillbeaffectedinanimalswillingeneralnotbe straightforward.Thesameproblems,however,occurinextrap- olatingbetweenanimalspecies,life stagesand organtoxicities.

Mechanistictestsdohavetheopportunitytomoredirectlylink tohumanriskassessment,however.Thiscanleadtobetterhuman riskassessment.Forinstance,estrogensignalingiscomplexand aberrantactivationisassociatedwithseveralhumandiseasesprop- ertiesnottestedinstandardanimaltests[46,47].Inthefuturemore mechanism-based risk assessment using tests providing more directmechanisticinformationonhumantissue-specificreceptor

(8)

activity and associated disorders may become an interesting opportunity.

2.8. ApplicationoftheChemScreentoolbox

AnimportantaimoftheEuropeanframeworkprogramsisdis- seminationofresultsandtheirapplication,preferablyleadingto newbusinessopportunitiesofsmallandmediumenterprises.To reachthisweworkedoncouplingdifferenttoolsanddatabasesin ordertodesign integratedtestingstrategiesthatcanbeusedin chemicalriskassessment.Thefollowingmodulesweredefined:

2.8.1. Module1.Insilicoprescreeninganddatacollectionmodule Thismoduleincludesinsilicoprescreening(QSAR)methodsto categorizechemicalsintomajorclassesoftoxicityprioritizedin REACH.Predictionsforcarcinogenicity,mutagenicityareapplied, whichincaseofpositivitywilldriveriskassessmentinthisdirec- tion avoiding reproductive toxicity testing according toREACH guidelines.Italsousesestablisheddatabasesandliteraturetofind anytestingresultsofthechemicalitselfandusechemoinformatics tofindstructuralanalogueswithreproductivetoxicitydatatobe usedassourcechemicalsinareadacrossapproach.

2.8.2. Module2.Invitroscreeningmodule

Thetest battery establishedwithinChemScreen canprovide confirmation ofalready predictedreproductive toxicity fromin module1,orgiveafirstalertofpotentialreproductivetoxicityinthe absenceofpredictionsfromtheseinsilicomethods.Thisscreening modelalready givesreliablepredictionsforstrongreproductive toxicants.Clearly,thispromisingscreeningmodulecanbefurther improvedwhenmoremechanismsofreproductivetoxicityareelu- cidated,allowingagreaterrelianceonthehighthroughputassays attheexpenseofthelowerthroughputones.

2.8.3. Module3.Kineticmodule

TheestablishedPBPKtoolslinkinvitroandinvivoplasmacon- centrations,and extrapolatethese toassociatedoral, dermalor potentiallyinhalationdoses.Thesearetobeappliedforvalidating andinterpretingmodule2results,i.e.toclassifychemicals,priori- tizechemicalstobetested,andforhelpingtodesignlimitedinvivo studiesforselectedchemicalsincaseconsiderednecessary.

2.8.4. Module4.Dataintegrationandriskassessment

AllelementsoftheChemScreenintegratedtestingstrategyhave beensetupandarein anadvancedstateofbeingcoupledand integrated. The efficacy of this integration in terms of making decisionsontheneedofsubsequentinvivoanimaltests,heavily dependsonthepurposeoftheinvestigation,e.g.whetheritisfor safedesignofnewchemicalsorforregistrationofexistingchem- icals.Forinstance,inthecontextoftheREACHtestingprogram readacrosscanbeaviableoption,e.g.toreducethereproductive toxicitytestingrequirementforchemicals(i.e.OECD421,or422) producedatmorethan10tons.Thecostsofthisinvitrotestingis around10timeslowerthantheinvivotesting,whichinitscheap- estversioncostsaround55,000Euro[48].Expertjudgmentstillis neededindifferentsituationstointerprettheoutcomes,e.g.tover- ifytheapplicabilitydomainsofthemodelsused,bothwithregard tostructuralaswellasbiologicalcharacteristics.Currently,more caseexamplesarebeingruntofurthervalidatetheseintegrated approaches.

Ourtestbatteryandintegratedtestingconceptwillbedissem- inatedfurther,andtrainingandsupportfacilitiesareofferedusing theworld-wideexpertiseavailabletodisseminatebioassays.The coordinatoratBioDetectionSystemsisthecontactpersonforthese activities(contactdetails;seewww.chemscreen.eu).

3. Conclusionsandfuturedirections

ThecombinedeffortsoftheChemScreenconsortiumhaveledto amodularintegratedassessmentstrategywhichincludesmethods forinsilicoprescreening,invitroscreening,methodstoextrapolate invitroresultstoinvivopredictions,andintegrativeonestocom- binethesedatawithpreexistingknowledgeonthetestchemical orrelatedchemicals.Thisworkconfirmsandextendstheearlier findingoftheFP6ReProTect project[7]that arelativelysimple batteryoftestscanbeusedsuccessfullytopredictreproductive toxicityofchemicalsseeTable1.Significantadvanceshavebeen madeparticularlyinthefurthersimplificationofthetestbattery, whilestillbeingabletoidentifyreproductivetoxicchemicalseffi- ciently,eitherinisolation[25],orinagroupingcontext[41,43].

Wefoundthatevenwitharelativelysmallnumberoftests,either apicaltestscombinedwithmechanistictestsoranrelativelysmall numberofmechanistictestspredictivitiesrangingfrom74to94%

canbereached[20,25],whichiscomparabletothatobtainedwith muchlargerToxCastscreeningpanels[21],theReProTectbattery [7],orthezebrafishELStests[49],ahumanembryonicstemcell test[50].Thisisremarkable,sincetheconcordancebetweendif- ferenttestspecies,likerabbitandrathasbeenestimatedtobenot morethan60%[6].Itseemsthatthesevalidationresultsmaybe biasedtosomeextent,e.g.becauserelativelystronglyactingrepro- ductivetoxicchemicalsareoftenusedinthevalidation studies.

Therefore,tocreatemoreconfidencewehavealsodevelopedand appliedothermeansofvalidatingtestsusingmechanisticknowl- edgeonthetestsandusingselectedchemicalstotestwhetherthe predictedadverseoutcomesmatchwiththemechanismsincluded inthebattery[20].Moredetailedclassificationschemesofchemi- cals,linkedtobothchemicalandbiologicalpropertiesasdeveloped byWu etal.[40]maybeveryhelpfulinfurthervalidationand improvementofthetestbattery,throughidentificationofchemical classesormechanismsmissedbythebattery.

AlthoughtheChemScreenbatteryshowedaverygoodperfor- mance identifyingstrongreproductivetoxicants, it isuncertain howitwillperformwithchemicalsthathavealessexplicitactivity profile.Thisisrelevant,sincethemajorityofindustrialchemicals arelikelytohaverelativelynon-specificmodesofaction[51].There isnoreasontobelievethatthesechemicalswillbemissedbythe battery,sincerelevantpathwaysseemavailable.Alsopartofthe reproductivetoxicchemicalsusedinourfirstfeasibilitystudydo notactthroughspecificmechanisms,andneverthelessarepicked upthroughrelativelynon-specificcytotoxicityendpointsinourtest battery[25].

Clearlynotallpathwaysofreproductivetoxicityareknownat present,andthepanelofmechanisticassaysmaythereforebemiss- ingcompoundsactingvialesscommonmechanismsofaction.On theotherhand,itseemsalsolikelythatchemicalsoftenaddress multiplepathwaysoftoxicity,therebyreducingtheneedtohave allpossiblemechanismofactionincludedinafinalbattery.This issupportedbytheverygoodperformanceofourbattery,even when compared withmuch largerbatteries [21] or more inte- grativelowerthroughputbatteriesandtests[7,49,50].Atpresent thecurrent combination of apicaland HTStestsseems a prag- maticandeffectivechoicethatcanbefurtherimprovedwhenmore mechanismsofreproductivetoxicitybecomeknown.Forregula- toryacceptance,itmayalsobehelpfultoincludetestdataofthese moreapicaltestssinceinitiallytheymaygiveadditionalconfidence inthetestresultsandassistwithclassificationandlabeling,par- ticularlyintheabsenceofanimaldata.Itshouldbenotedthatthe majorityofREACHchemicalsareexpectedtoberelativelyinertand nottobereproductivetoxic.Becauseofthis,forregulatoryapplica- tionawidecoverageofpossiblemechanismsofactionoftoxicants isneeded,butwithoutgivingmanyfalse-positivesthatmaylead toadditional testing requirement.Our battery of veryselective

(9)

mechanistictestsincombinationwithapicaltestsandavailable animaltestdataseemsinthisrespectaneffectivetestingstrategy.

Withincreasedexperienceandconfidenceininvitrotestingthe weightoninvitrotestingandextrapolationtohumanratherthan animaldatawillbecomeincreasinglyimportant.

Conflictofinterest

Theauthorsdeclarethattherearenoconflictsofinterest.

Acknowledgments

Theworkdescribed in this manuscript couldnothave been exertedwithouttheeffortsof alladditionalChemScreen partic- ipantsnotspecificallymentioned. Wewouldparticularlyliketo sincerelythankouradvisoryboardmembersDrsDavidDix,Robert Chapin,MariaBondesson,MarkCronin,andNeilCarmichael for theirsupportandvaluablecontributiontothesuccessoftheChem- Screenproject.WearealsoverygratefulforthecontributionofDrs BurkhardFlick,ThomasSobanskiandSandervanderLindentothe finalsymposiumandapaneldiscussionthatprovidedelementsof thisreview.Thisworkwascarriedoutwithfinancialsupportfrom theCommissionoftheEuropeanCommunities,thecollaborative projectChemScreen(GA244236).

References

[1]EuropeanCommission.Whitepaper“strategyforafuturechemicalspolicy”.

COM2001;88:2001.

[2]EuropeanCommission(EC).Regulation(EC)No1907/2006oftheEuropeanPar- liamentandoftheCouncilof18December2006concerningtheRegistration, Evaluation,AuthorizationandRestrictionofChemicals(REACH),establishinga EuropeanChemicalsAgency,amendingDirectiveNo1488/94aswellasCouncil Directive76/769/EECandCommissionDirectives91/155/EEC,93/105/ECand 2000/21/EC.OffJEurUnionL20072007;136:3–280.

[3]HartungT,LuechtefeldT,MaertensA,KleensangA.Foodforthought...inte- gratedtestingstrategiesforsafetyassessments.ALTEX2013;30(1):3–18.

[4]VanderJagtK,MunnS,TorslovJ,deBruijnJ.Alternativeapproachescanreduce theuseoftestanimalsunderREACHEuropeanCommissionReportEUR21405;

2004.

[5]CostanzaR,HartungT.Re-evaluationofanimalnumbersandcostsforinvivo teststoaccomplishREACHlegislationrequirementsforchemicals—areportby thetransatlanticthinktankfortoxicology.ALTEX2009;26:187–208.

[6]Hurtt ME, Cappon GD, Browning A. Proposal for a tiered approach to developmentaltoxicitytestingforveterinarypharmaceuticalproducts for food-producinganimals.FoodChemToxicol2003;41(5):611–9.

[7]SchenkB,WeimerM,BremerS,vanderBurgB,CortvrindtR,FreybergerA, etal.TheReProTectFeasibilityStudy,anovelcomprehensiveinvitroapproach todetectreproductivetoxicants.ReprodToxicol2010;30:200–18.

[8]NationalResearchCouncil(USA).Scientificfrontiersindevelopmentaltoxico- logyandriskassessment.Washington,DC:NASPress;2000.

[9]NationalResearchCouncil(USA).Toxicitytestinginthetwenty-firstcentury:a visionandastrategyCommitteeonToxicityandAssessmentofEnvironmental Agents.Washington,DC:NationalResearchCouncil;2007.

[10]SonneveldE,JansenHJ,RitecoJAC,BrouwerA,VanderBurgB.Develop- mentofandrogen-andestrogen-responsivebioassays,membersofapanel ofhumancellline-basedhighlyselectivesteroidresponsivebioassays.Toxicol Sci2005;83:136–48.

[11]SonneveldE,RitecoJAC,JansenHJ,PieterseB,BrouwerA,SchoonenWG,etal.

Comparisonofinvitroandinvivoscreeningmodelsforandrogenicandestro- genicactivities.ToxicolSci2006;89:173–87.

[12]SonneveldE,PieterseB,SchoonenW,VanderBurgB.Validationofinvitro screeningmodels forprogestagenicactivities:inter-assaycomparisonand correlation withinvivoactivityinrabbits.ToxicolInVitro2011;25:545–

54.

[13]VanderBurgB,KroeseED,PiersmaAH.Towardsapragmatic alternative testingstrategyforthedetectionofreproductivetoxicants.ReprodToxicol 2011;31(May(4)):558–61.

[14]Wedebye EB, Dybdahl M, Nikolov NG, Jónsdóttir SO, Niemelä JR. QSAR pre-screenof70,983substancesforgenotoxiccarcinogenicity,germcellmuta- genicity,developmentaltoxicityandendocrineactivityintheChemScreen project.ReprodToxicol2015;55:64–72.

[15]BitschA,JacobiS,MelberC,WahnschaffeU,SimetskaN,MangelsdorfI.REP- DOSE:adatabaseonrepeateddosetoxicitystudiesofcommercialchemicals—a multifunctionaltool.RegulToxicolPharmacol2006;46(3):202–10.

[16]SchulzF,BatkeM,MangelsdorfI,Pohlenz-MichelC,SimetskaN,LewinG.

Sensitivityofdifferentgenerationsanddevelopmentalstagesinstudieson reproductivetoxicity.ToxicolLett2014;226(Apr(2)):245–55.

[17]PiersmaAH,RorijeE,BeekhuijzenME,CooperR,DixDJ,Heinrich-HirschB,etal.

Combinedretrospectiveanalysisof498ratmulti-generationreproductivetox- icitystudies:ontheimpactofparametersrelatedtoF1matingandF2offspring.

ReprodToxicol2011;31(May(4)):392–401.

[18]NovicM,VrackoM.QSARmodelsforreproductivetoxicityandendocrinedis- ruptionactivity.Molecules2010;15(3):1987–99.

[19]http://edukon.biologie.uni-konstanz.de/edukon/index.php.

[20]Van der Burg B, Pieterse B, Buist H, Lewin G, Van der Linden SC, Man HY, et al. A high throughput screening system for predicting chemically-inducedendocrinedisruptionandsexorgandeformities.Reprod Toxicol2015;55:95–103.

[21]SipesNS,MartinMT,ReifDM,KleinstreuerNC,JudsonRS,SinghAV,etal.Predic- tivemodelsofprenataldevelopmentaltoxicityfromToxCasthighthroughput screeningdata.ToxicolSci2011;124(Nov(1)):109–27.

[22]RoelofsMJ,PiersmaAH,vandenBergM,vanDuursenMB.Therelevanceof chemicalinteractionswithCYP17enzymeactivity:assessmentusinganovel invitroassay.ToxicolApplPharmacol2013;268(May(3)):309–17.

[23]Schulpen SH,RobinsonJF,PenningsJL,vanDartelDA, PiersmaAH. Dose responseanalysisofmonophthalatesinthemurineembryonicstemcelltest assessedbycardiomyocytedifferentiationandgeneexpression.ReprodToxicol 2013;35(Jan):81–8.

[24]SchulpenSH,PenningsJL,TonkEC,PiersmaAH.Astatisticalapproachtowards thederivationofpredictivegenesetsforpotencyrankingofchemicalsinthe mouseembryonicstemcelltest.ToxicolLett2014;225(3):342–9.

[25]PiersmaAH,SchulpenSHW,UibelF,VanVugt-LussenburgB,BosgraS,Hermsen SAB,etal.Evaluationofanalternativeinvitrotestbatteryfordetectingrepro- ductivetoxicants.ReprodToxicol2013;38:53–64.

[26]DixDJ,HouckKA,MartinMT,RichardAM,SetzerRW,KavlockRJ.TheToxCast programforprioritizingtoxicitytestingofenvironmentalchemicals.Toxicol Sci2007;95(1):5–12.

[27]VanderLindenSC,vonBerghAR,vanVugt-LussenburgBM,JonkerLR,TeunisM, KrulCA,etal.Developmentofapanelofhigh-throughputreporter-geneassays todetectgenotoxicityandoxidativestress.MutatResGenetToxicolEnviron Mutagen2014;760:23–32.

[28]UibelF,MühleisenA,KöhleC,WeimerM,StummannTC,BremerS,etal.

ReProGlo:anewstemcell-basedreporterassayaimedtopredictembryotoxic potentialofdrugsandchemicals.ReprodToxicol2010;30(1):103–12.

[29]DangZ, RuS,WangW,Rorije E,HakkertB, VermeireT.Comparisonof chemical-inducedtranscriptionalactivationoffishandhumanestrogenrecep- tors:regulatoryimplications.ToxicolLett2011;201(Mar(2)):152–75.

[30]Wetmore BA. Quantitative in vitro-to-in vivo extrapolation in a high-throughput environment.Toxicology2014, http://dx.doi.org/10.1016/

j.tox.2014.05.012,pii:S0300-483X(14)00114-0.[Epubaheadofprint].

[31]BosgraS,WesterhoutJ.Interpretinginvitrodevelopmentaltoxicitytestbattery results:theconsiderationoftoxicokinetics.ReprodToxicol2015;55:73–80.

[32]DancikY,TroutmanJA,JaworskaJ.Estimationofsafeinvivodoseofder- mallyappliedchemicals—illustrationwithfourreproductivetoxicants.Reprod Toxicol2015;55:50–63.

[33]SeilerAE,SpielmannH.Thevalidatedembryonicstemcelltesttopredict embryotoxicityinvitro.NatProtoc2011;6(7):961–78.

[34]OECD.TestNo.236:FishEmbryoAcuteToxicity(FET)Test.OECD;2013.

[35]OECD. Test guideline No.455. Stably transfected human estrogen receptor-transcriptional activation assay for detection of estrogenic agonist-activityofchemicals.Paris:OrganisationforEconomicCooperation andDevelopment;2009.

[36]OECD.SPSFbytheNetherlands:U2-OScellsTranscriptionalERalphaCALUX®- assayforthedetectionofestrogenicandantiestrogenicchemicalsforinclusion inTG455/TG457.OECD;2013.

[37]OECD.SPSFbytheEuropeanCommission:performance-basedtestguideline onandrogenreceptortransactivationassays(ARCALUX).OECD;2013.

[38]VanderBurgB,WinterR,ManH-y,VangenechtenC,WeimerM,Berckmans P,etal.OptimizationandprevalidationoftheinvitroARCALUXmethodto testandrogenicandantiandrogenicactivityofcompounds.ReprodToxicol 2010;30:18–24.

[39]VanderBurgB,WinterR,WeimerM,BerckmansP,SuzukiG,GijsbersL, et al.Optimization andprevalidationoftheinvitroER␣CALUXmethod totestestrogenicandantiestrogenicactivityofcompounds.ReprodToxicol 2010;30:73–80.

Referenzen

ÄHNLICHE DOKUMENTE

This Southeast Asia-Europe research project will use a One Health approach to identify the major parameters responsible for the presence of animal-associated antimicrobial

Documentation effort: The V-Model software development process is partitioned into activities for requirements engineering, software architecture and design, and software

Based on (1) the OMEGA modeling environment (section 4.2), the following subsections relate the requirements outlined in the previous section to existing tool support, also for

• Global (static) data: The loader allocates a global data area (64 KB) and loads its address into register EDI. The first 8 bytes of the global data area hold the address- es of

The goal of this project is to write a compiler that translates a program of the toy language Mini to an intermediate representation (a CFG with IR instructions), performs

Discussions to introduce a range of further measures, such as traffic calming measures around the kindergarten, traffic lights at pedestrian crossings of the main

The GCC governments have implemented aggressive economic diversification and industrialization plans for more than a decade now. Accordingly, local industry growth, job

To keep the citric acid cycle going requires anaplerotic reactions such as the glyoxylate shunt to restore the cycle intermediates that are withdrawn for the biosynthesis of