1-s2.0-S0166445X15300540-main.pdf (1.564Mb)

(1)

AquaticToxicology168(2015)48–59

ContentslistsavailableatScienceDirect

Aquatic Toxicology

jou rn al h om ep a ge :w w w . e l s e v i e r . c o m / l o c a t e / a q u a t o x

Environmental chemicals active as human antiandrogens do not activate a stickleback androgen receptor but enhance a feminising effect of oestrogen in roach

Anke Lange

^a,∗

, Marion Sebire

^a,b

, Pawel Rostkowski

^c,1

, Takeshi Mizutani

^d

, Shinichi Miyagawa

^d

, Taisen Iguchi

^d

, Elizabeth M. Hill

^c

, Charles R. Tyler

^a,∗

aUniversityofExeter,Biosciences,CollegeofLife&EnvironmentalSciences,ExeterEX44QD,UnitedKingdom

bCefasWeymouthLaboratory,BarrackRoad,TheNothe,Weymouth,DorsetDT48UB,UnitedKingdom

cUniversityofSussex,SchoolofLifeSciences,BrightonBN19QJ,UnitedKingdom

dOkazakiInstituteforIntegrativeBioscience,NationalInstituteforBasicBiology,NationalInstitutesofNaturalSciences,andDepartmentofBasicBiology, SchoolofLifeScience,SOKENDAI(GraduateUniversityforAdvancedStudies),5-1Higashiyama,Myodaiji,Okazaki,Aichi444-8787,Japan

a r t i c l e i n f o

Articlehistory:

Received29July2015 Receivedinrevisedform 16September2015 Accepted22September2015 Availableonline26September2015

Keywords:

Feminisation Antiandrogen Oestrogen Mixture Fish

Endocrinedisruption

a b s t r a c t

Sexualdisruptionis reportedinwildfishpopulationslivinginfreshwatersreceiving dischargesof wastewatertreatmentworks(WwTW)effluentsandisassociatedprimarilywiththefeminisationof malesbyexposuretooestrogenicchemicals.Antiandrogenscouldalsocontributetothefeminisationof malefish,buttherearefarlessdatasupportingthishypothesisandalmostnothingisknownfortheeffects ofoestrogensincombinationwithantiandrogensinfish.Weconductedaseriesofinvivoexposuresintwo fishspeciestoinvestigatethepotencyonreproductive-relevantendpointsoftheantiandrogenicantimi- crobialstriclosan(TCS),chlorophene(CP)anddichlorophene(DCP)andtheresin,abieticacid(AbA),all foundwidelyinWwTWeffluents.Wealsoundertookexposureswithamixtureofantiandrogensanda mixtureofantiandrogensincombinationwiththeoestrogen17␣-ethinyloestradiol(EE2).Instickleback (Gasterosteusaculeatus),DCPshowedatendencytoreducespiggininductioninfemalesandrogenisedby dihydrotestosterone(DHT),butthesefindingswerenotconclusive.Inroach(Rutilusrutilus),exposuresto DCP(178days),oramixtureofTCS,CPandAbA(185days),ortothemodelantiandrogenflutamide(FL, 178days)hadnoeffectongonadalsexratiooronthedevelopmentofthereproductiveducts.Exposure toEE2(1.5ng/L,185days)inducedfeminisationoftheductsin17%ofthemalesandinthemixtureof antiandrogens(TCS,CP,AbA)incombinationwithEE2,almostall(96%)ofthemaleshadafeminised reproductiveducts.Insticklebackandrogenreceptor(AR␣andAR␤)transactivationassays,themodel antiandrogens,FLandprocymidoneinhibited11-ketotestosterone(11-KT)inducedreceptoractivation, butnoneofthehumanantiandrogens,TCS,CP,DCPandAbAhadaneffect.Thesedataindicatethatantimi- crobialantiandrogensincombinationcancontributetothefeminisationprocessinexposedmales,but theydonotappeartoactthroughtheandrogenreceptorinfish.

Introduction

Endocrinedisruptingcompounds(EDCs)derivefrom(primar- ily)anthropogenic,industrial,agriculturalanddomestic sources

∗Correspondingauthors.Tel.:+441392724450;fax:+441392724000.

E-mailaddresses:[email protected](A.Lange),[email protected] (C.R.Tyler).

1 Presentaddress:NorwegianInstituteforAirResearch,DepartmentofEnviron- mentalChemistry,Norway.

andtheyhavethecapacitytointerferewithreproductivedevelop- mentandfunctioninawiderangeofspecies.Wildlifeassociated withfreshwaterecosystemsisespeciallyatriskofEDCexposureas aquaticenvironmentsactasarepositoryforawiderangeofchemi- calpollutants.Manyofthesechemicalsaredischargedviaeffluents fromwastewatertreatmentworks(WwTW),andgloballyexposure toWwTWeffluentshasbeenassociatedwithavarietyofdeleteri- ouseffectsonreproductioninfish(Joblingetal.,1998;Gravatoand Santos,2003;MillsandChichester,2005;Pottingeretal.,2013a;

Blazeretal.,2014).Todate,mostofthefocusonEDCshasbeen onoestrogensandtherearenowprovenlinksbetweenestrogen http://dx.doi.org/10.1016/j.aquatox.2015.09.014

(2)

exposure and a range of feminisation responses in ﬁsh. These responsesincludeelevatedconcentrationsofthefemaleegg-yolk precursor vitellogenin (VTG) in males and immature females, developmentofafemale-likeovariancavityinthetestisofmales, andintersexcharacterisedbythepresenceofbothmaleandfemale sexcellscontainedwithinthesamegonad.Thesefeminisingeffects havebeenlinkedtoreducedgametequalityandthereisconcern aboutpopulationleveleffects(Kiddetal.,2007;Harrisetal.,2011;

Langeetal.,2011).

Many of thefeminised effects seen in wildpopulations can beinducedbycontrolledexposuretooestrogensandtheirmix- tures.However,inthelastdecade,antiandrogenshaveemerged as anotherclass of EDCs that potentially contributeto adverse health effectsin humanand wildlife.Antiandrogensmaycause effectsthrough avarietyof differentmechanisms,includingvia actingasandrogenreceptor(AR)antagonists,thus,inhibitingAR- dependentgeneexpression,orbyalteringthebiosynthesisand/or excretionofnaturalhormones(Wilsonetal.,2008).Thereisevi- dencethatexposureofrodentstoantiandrogensduringcriticallife periodsthatincludesexualdifferentiation,foetallifeandmatura- tion,canhaveeffectsonmaledevelopment(Hotchkissetal.,2008;

Christiansenetal.,2009;Rideretal.,2010).Similarlyforﬁsh,there isevidencederivedfromlaboratory-basedinvivoexposuresthat someantiandrogenscansuppresstheeffectsofandrogensinmales, thus,contributingtodemasculinising/feminisingeffects.Reported effectsincludeinductionofintersexualityinmalemedaka(Oryzias latipes)andovarianatresiainfemalemedaka,reducedspermcount andreducedsecondarysexcharacteristicsinmalefatheadminnow (Pimephalespromelas)andmaleguppy(Poeciliareticulata),altered reproductivebehavioursinmalestickleback(Gasterosteusaculea- tus) and guppy, and reduced spiggin (an androgen-dependent proteinusedfornestconstruction)productioninmalestickleback (Makynenetal.,2000;BaatrupandJunge,2001;Bayleyetal.,2002;

KinnbergandToft,2003;Kiparissisetal.,2003;Jensenetal.,2004;

Kangetal.,2006;Martinovi ´cetal.,2008;Sebireetal.,2008,2009).

Theseeffectsarepredominantlyderivedforexposurestothemodel antiandrogenﬂutamide(FL)andtootherantiandrogensatconcen- trationsthatfarexceedthosemeasuredinaquaticsystems,albeit thereisevidenceforsomeeffectsofantiandrogensinﬁshforenvi- ronmentallyrelevantexposures(e.g.Sebireetal.,2009;Sebireetal., 2011;Greenetal.,2015).

Globally,antiandrogenicactivitieshavenowbeendetectedin effluents,surfacewatersandsedimentsusinginvitrobasedrecep- torARassays, suchasARtransactivation assaysor yeast-based transcriptionalactivationassays(Tollefsenetal.,2007;Urbatzka etal.,2007;Shietal.,2009;Hill etal.,2010;Rostkowskietal., 2011;Zhaoetal.,2011;Belletetal.,2012;Alvarez-Mu ñozetal., 2015).InanextensivesurveyofWwTWeffluentsintheUK,sig- nificantantiandrogenicactivitywasidentified(between21.3and 1231␮g flutamide equivalents L⁻¹) in all samples investigated (EnvironmentAgency,2007).Furthermore,amodellingstudyhas correlatedfeminisedfishinUKriverswithpredictedantiandrogen contentboth aloneand incombination withoestrogens(Jobling etal.,2009).Compoundsknowntobeantiandrogenicincludesome pesticides (e.g. procymidone, vinclozolin, linuron),pharmaceut- icals(e.g.FL,cyproteroneacetate),andsomeindustrialchemicals suchasphthalatesorpolybrominateddiphenylethers.Ourrecent studies,however,indicatethatthesecompoundsmaynotbesig- nificantcontributorstobioavailableantiandrogensinfishlivingin UKrivers.Usingabioassay-directedanalyticalapproach,wehave identifiedtheantimicrobialstriclosan(TCS),chlorophene(CP)and dichlorophene(DCP),ingredients in avariety of householdand personalcareproducts,togetherwithresinacids,naturallyoccur- ringcomponentsofwoodandbark,asamongtheantiandrogensin WwTWeffluentsthatbioconcentrateinfishbileatconcentrations tensofthousandsgreaterthanintheeffluentitself(Rostkowski

etal.,2011).DuetotheiroccurrenceinWwTWeffluentsandtheir abilitytobioconcentrate,thesecompoundsareconsideredtobe bioavailabletofish.Theantimicrobialsarepresentineffluentsat ngtolow␮g/Lconcentrationsandforresinacidsfromlowngup tomg/Lconcentrations.Allthesecompounds havebeenshown topossesssimilartohigherantiandrogenicpotenciesinvitroon thehumanARwhencomparedwiththestandardantiandrogenic compoundFL(Rostkowskietal.,2011).

The aim of this study was to investigate the potency on reproductive-relevant endpoints in fish of some of the antiandrogenic antimicrobials (DCP, CP, TCS) and resin acids present in WwTWeffluents, includingasmixtures, and in combination with theoestrogen 17␣-ethinyloestradiol (EE2). Thiswas done principallythroughaseriesofinvivoexperimentsinwhichfish wereexposedtoantiandrogensatenvironmentalconcentrations not exceeding the maximum antiandrogenic activity identified for WwTWeffluents intheUK. Inthefirst study,theabilityof DCPtoinhibitspiggininduction(awell-establishedandsensitive biomarkerfor(anti)androgens) wasassessed infemalestickle- backs androgenised by exposureto dihydrotestosterone (DHT).

TwofurtherexperimentsinvestigatedtheeffectsofDCP,themodel antiandrogenFL,a mixtureofTCS,CPand abieticacid(AbA)or ofamixtureofantiandrogens(TCS,CP,AbA)incombinationwith theenvironmentaloestrogen,EE2onreproductivedevelopment inroach(Rutilusrutilus),aspeciesthathasreceivedsomeofthe mostextensiveworkforunderstandingthefeminisingeffectsof environmentaloestrogens.Finally,sticklebackARtransactivation assayswereappliedtosupportamechanisticunderstandingfor theeffectsoftheantimicrobialantiandrogensseenintheinvivo studies.

MaterialandMethods

2.1. Fishhusbandryandchemicalorigin

Mixed sex populations of three-spined stickleback were obtainedfromPrioryFisheries(Cullompton,UK)inNovember2008 andmaintainedinthelaboratoryunderconstantwatertempera- ture(10–12^◦C)andphotoperiod(12L:12D)forfourmonthsprior tothestartoftheexperiment.Theﬁshwerefeddailywithfrozen gamma-irradiatedbloodworm(TropicalMarineCentre,Chorley- wood,UK).

Pre-spawning, sexually maturemale andfemale roach were obtainedfromtheEnvironmentAgency’sNationalCoarseFishFarm (Calverton,Nottinghamshire,UK)inMay2009andbroughttothe laboratory.Fishwereseparatedbysexandmaintainedat15–16^◦C andaphotoperiodmatchingthedaylengthatthetimeofsampling (16L:8D).Spawningwasinducedbyasingleintraperitonealinjec- tionofcarppituitaryextract(CPE,CalvertonFishFarm)dissolved inphysiologicalsaline,usinganestablishedmethodforinducing spawningofadultﬁshandensuressynchronousgametecollection (Joblingetal.,2002).24haftertheinjectionwithCPE,ﬁshweredry strippedoftheirgametesandeggsfertilisedinvitro.

Chlorophene(CP,95%purity),dichlorophene(DCP,97.5%),17␣- ethinyloestradiol(EE2,»98%),dihydrotestosterone(DHT,≥97.5%), ﬂutamide(FL,≥99%),triclosan(TCS,≥97%),testosterone(T,≥98%), 11-ketotestosterone (11-KT, ≥98%),bicalutamide (≥98%),bis(2- hydroxyphenyl) methane (98%), oestrone (99%), progesterone (≥99%)and4-(4-chlorophenoxy)phenol(97%)wereobtainedfrom Sigma–Aldrich(Gillingham,UK).Abieticacid(85%)wasobtained fromAcros Organics(Geel, Belgium)and procymidone (PROCY,

>98%) from Fluka. Stock solutions of chemicals were prepared in HPLC grade acetone or ethanol (both Fisher Scientiﬁc UK, Loughborough,UK).

(3)

50A.Langeetal./AquaticToxicology168(2015)48–59

Table1

Averagemeasuredchemicalconcentrationsinexposuretanks.

Treatment(nominalexposureconcentration;␮g/L) Meanmeasuredexposureconcentrations(␮g/L)±SEM(n)

DHT FL DCP AbA CP TCS EE2

ExposureofandrogenisedfemalesticklebacktoDCP

DWC ≤LOD ≤LOD ≤LOD n.a. n.a. n.a. n.a.

SC ≤LOD ≤LOD ≤LOD n.a. n.a. n.a. n.a.

DHT(5) 4.20±0.12(8) ≤LOD ≤LOD n.a. n.a. n.a. n.a.

DHT(5)+FL(150) 3.70±0.21(8) 132.44±2.72(8) ≤LOD n.a. n.a. n.a. n.a.

DHT(5)+DCP(0.1) 3.58±0.26(8) ≤LOD 0.041±0.010(8) n.a. n.a. n.a. n.a.

DHT(5)+DCP(1) 3.73±0.22(10) ≤LOD 0.570±0.060(8) n.a. n.a. n.a. n.a.

DHT(5)+DCP(10) 4.05±0.14(8) ≤LOD 7.020±0.736(9) n.a. n.a. n.a. n.a.

ExposureofroachtoDCPorFL

DWC n.a. ≤LOD ≤LOD n.a. n.a. n.a. n.a.

SC n.a. ≤LOD ≤LOD n.a. n.a. n.a. n.a.

DCP(1) n.a. ≤LOD 0.36±0.05(16) n.a. n.a. n.a. n.a.

FL(150) n.a. 136.79±5.77(16) ≤LOD n.a. n.a. n.a. n.a.

FL(450) n.a. 355.14±21.88(16) ≤LOD n.a. n.a. n.a. n.a.

Exposureofroachtoamixtureofantiandrogens(AAmix)and/orEE2

DWC n.a. n.a. n.a. ≤LOD ≤LOD ≤LOD ≤LOD

AAmix(50each) n.a. n.a. n.a. 6.40±0.79(20) 23.86±1.25(20) 7.11±1.05(20) ≤LOD

3ng/LEE2(3) n.a. n.a. n.a. ≤LOD ≤LOD ≤LOD (1.60±0.60)x10⁻³(2)

AAmix(50each)+EE2(3) n.a. n.a. n.a. 3.87±0.67(19) 23.18±2.06(19) 9.10±1.87(19) (1.29±0.05)x10⁻³(11)

n.a.:notapplicable;limitofdetection(LOD):0.12ngDCP/L;0.21ngFL/L;0.21ngDHT/L;0.50ngEE2/L;0.15ngCP/L,0.12ngTCS/L,0.22ngAbA/L.DWC,dilutionwatercontrol;SC,solventcontrol;AAmix,Aba+CP+TCS(50␮g/L each).

(4)

2.2. SticklebackexposuretoDCP

In thesticklebackexperiment, females were exposed for 21 days,basedontheOECDGuidancedocument148fortheandro- genised female stickleback screen (AFSS) (OECD, 2011). The exposureconsistedofseventreatmentseachinduplicatewitheach tankcontaining10ﬁshinavolumeof10L.Fishweremaintained inaﬂow-throughsystemwithincomingwateratambienttemper- ature(15±0.5^◦C)anda12L:12LDphotoperiodthroughout.There wasadailyexchangeoftwotankvolumesofwaterandthechem- icaldosingsolutionsweredeliveredtothetanksusingperistaltic pumps.

FishweresimultaneouslyexposedtoDHT(nominal5␮g/L)and oneofthreeconcentrationsofDCP(nominal0.1,1.0or10␮g/L) ortheantiandrogen-positivecontrolFL(nominal150␮g/L).The experimentincludedadilutionwatercontrol(DWC),asolventcontrol(SC)andanandrogen-positivecontrol(DHTalone).Chemical stocksolutionswerepreparedinethanolandfurtherdilutedwith dilutionwaterbeforedosingintotheexposuretanks.Theﬁnalsol- ventexposureconcentrationwas0.0003%. Thedosingsolutions wererenewedevery2daysandtheﬂowratesmonitoredatthe sametime.Watersamplesweretakenforchemicalanalysisfrom eachtankonaweeklybasis.Fishwerefeddailywithfrozenblood- worm.

2.3. Roachexposures

Twoexposureswereconducted withroach.Threedayspost fertilisation(dpf),fertilisedeggsweredeployedintoglassaquaria andexposedunderflow-throughconditions,induplicatetanks,as describedbelow.Embryoshatched7–10dpfandthefishwerepro- visionedwithdietaryrequirementsaccordingtotheirage(Paull et al., 2008).Briefly, roach were fed withCyprico Crumble EX dryfood(CoppensInternationalbv,Helmond,TheNetherlands) and at all lifestages, thediet was supplemented initially with freshlyhatchedbrineshrimp(Artemiasp.)nauplii,giventosati- ationand,asthefishgrewbeyond80dayposthatch(dph)with frozengamma-irradiatedbrineshrimpandbloodworm(Chirono- mussp.;TropicalMarineCentre,Chorleywood,UK)andTetraMin dry coldwater flake food (Tetra Werke, Melle, Germany). Fish weremaintainedatambientwater temperatures(18±2^◦CMay toOctober,13±2^◦COctobertoDecember,8±2^◦Cforexperiment 1and21±2^◦CMay–Octoberand17±1^◦COctober–Novemberfor experiment2).Thephotoperiodregimewasadjustedtostimulate ambientseasonalchanges(16L:8DbetweenMayandAugustafter whichitwasreducedinastepwisemannerto12L:12D).Thechem- icaldosingsolutionsweredeliveredtothetanksusingperistaltic pumpsandthedosingsolutionswererenewedevery3–4daysand theflowratesmonitoredatthesetimes.Watersamplesforchemi- calanalysisweretakenregularlyfromeachtankforanalysis(once ortwicepermonth).

Inexperiment1,approximately400eggswereplacedintoeach tankandexposedcontinuouslytooneofthreeconcentrationsof DCP(nominal1,10or30␮g/L)oroneoftwoconcentrationsofFL (nominal150or450␮g/L)until172daysposthatch(dph).Dilution waterandsolvent-dosedtankswereincludedascontrols.Initially, fertilisedeggs weredeployedinto 20L of water and thewater volumewassubsequentlyincreasedto45Lat30dph,70Lat80 dphand145Lat145dph.At32and136dph,stockingdensities inalltankswereadjustedtoensurethefishbiomasswasequiv- alentoverthestudyperiod.From136dphonwards,therewere 100fishpertankensuringsufficientnumbersforbiologicalsamp- ling.At172dph,fishweresampledfromeachtankforbiological analyses.Throughouttheexposure,thereweretwotankvolume exchangesofwaterdaily.StocksolutionsofDCPandFLwerepre- paredinethanol,dilutedindilutionwaterbeforesuppliedtothe

exposuretanks.Thesolventdosingtoallthetankswaslessthan 0.00075%(v/v).

Inexperiment2,initially,approximately200eggswereplaced intoeachtankexposedcontinuouslytoEE2(atnominal3ng/L),a mixtureofthethreeantiandrogensAbA,CPorTCS(eachatnominal 50␮g/L),oramixtureofthethreeantiandrogens(eachatnominal 50␮g/L)in combination withEE2(atnominal3ng/L). Fish weremaintainedin50Lofwaterthroughoutandtherewasone tankvolumeexchangeof waterdaily. At55dph, stockingden- sitiesinalltankswereadjustedto100fishpertankandat185 dph,fishweresampledfromeachtankforbiologicalanalyses.In thisexperiment,asolvent-freeapproachwasadoptedforwhich chemicalstocksolutionswerepreparedinacetone.Solvent-free dosingstockswerepreparedevery3–4daysbyaddingtheappro- priatevolume(<0.5mL)of acetonestockintoglassvessels.The acetonewasallowedtoevaporateovernightandsubsequently,the requiredvolumeofwaterwasaddedbeforethesolutionwasstirred for2hfollowedby30minsonicationbeforeconnectingintothe flow-throughsystem.Controlsweremaintainedindilutionwater.

Themeasuredchemicalconcentrationsforeachofthesestudies aregiveninTable1.At29dph,onecontroltankwaslostdueto mortalitiesandreplacedusingﬁshfromthereplicatetank.

2.4. Analyticalchemistry

Measuredexposureconcentrationsweredeterminedinwater samplesfromeverytankusinggaschromatography–massspec- trometry(GC/MS).Duringeachexperiment,watersampleswere collectedfromtheexperimentaltanksintosolvent-cleanedglass flaskstodeterminetheactualexposureconcentrations.Immedi- atelyaftercollection,methanolandaceticacid(finalconcentrations 4 and 1%,respectively) wereadded toeach sample which was also spiked with an equal amount of internal standard (IS)as expectedforthetestcompound intheextractedvolume.Inter- nalstandardsincludedbicalutamide(forthedeterminationofFL), oestrone(forthedeterminationofEE2andAbA),progesterone(for thedeterminationofDHT),4-(4-chlorophenoxy)-phenol(forthe determinationofTCSandCP)andbis(2-hydroxyphenyl)methane (forthedeterminationofDCP).Watersampleswereextractedusing preconditionedOasisHLBcartridges(WatersCorporation,Milford, MA,USA),whichwerethenrinsedwithwater,driedundervac- uumandstoredat−80^◦Cuntilanalysis.Priortoanalysis,cartridges weredefrosted and extractseluted withmethanol.The solvent wasremovedundervacuumandthesampleextractssilylatedby theadditionof30␮Lbis(trimethylsilyl)trifluoroacetamide(BSTFA) and30␮Lpyridine(65^◦Cfor30min).Thesamplewasdrieddown undernitrogen,re-dissolvedin20␮LBSTFAand2␮Linjectedinto theGC–MS.SampleswereanalysedonaTraceGC(Thermoquest, Texas,USA)fittedwitha30mZebronZB-5MSfusedsilicacapillary column(30m×0.25mm×0.25␮mfilmthickness)connectedtoa Polaris-Qiontrapmassspectrometer(Thermo,Texas,USA).TheMS wasoperatedin50–650m/zfullscanpositiveionisationmodewith electronionisationat70eV.Them/zionswithmaximumintensity wereusedasquantifierionsandacalibrationcurvewasusedtocal- culatetheabsoluteamountofeachanalyteinthesampleextractin comparisonwiththeISresponse.

2.5. Fishsamplingandbiologicalanalyses

Allfishweresacrificedhumanelybyterminalanaesthesiawith benzocainefollowed bycervical dislocationasapprovedby the UKHomeOffice(Animals(ScientificProcedures)Act1986).Total length(stickleback) orstandard length(roach) and wet weight wererecordedtothenearest1mmand 0.01g,respectively.For sticklebacks(n=14–20pertreatment),kidneysweredissectedout, weighedandstoredat−20^◦Cuntilthemeasurementofspiggin

(5)

52 A.Langeetal./AquaticToxicology168(2015)48–59

usingan ELISA,as described by Katsiadaki et al. (2002).Roach (n=15fromeachtankinexperiment1andn=25fromeachtank inexperiment2)werepreservedintotoin4%paraformaldehyde forhistologicalanalysisoftheirgonads,embeddedinparafﬁnwax, sectionedto5–10␮mandstainedwithhematoxylinandeosin.The sectionswereanalysedforgermcellsandgonadalductformation, accordingtoPaulletal.(2008).

2.6. Invitrostudies

Tosupportinterpretationofthebiologicaleffectsdataforthe invivoexposures,weinvestigatedtheactivitiesoftheantimicro- bialantiandrogens,DCP,TCS,CP,andAbA,andFLinsticklebackAR transactivationassays.

HepG2 cells were seeded in 24-well plates at 5×10⁴ cells perwellinphenol-redfreeDulbecco’smodiﬁedEagle’smedium (Sigma) supplemented with 10% charcoal/dextran treated fetal bovineserum.After24h,thecellsweretransfectedwith200ng ofAR␣orAR␤,400ngofthereporterconstructMMTV-Lucand 100ngofpRL-TKusingFugeneHDtransfectionreagent(Promega).

Fivehoursaftertransfection,cells weredosedwithARagonists alone,orARagonistsandARantagonistsincombinationandincu- batedfor44h,whenthecells werecollected,andtheluciferase activitymeasuredusingtheDual-LuciferaseReporterAssaySystem (Promega).Promoteractivitywascalculatedasfirefly-/seapansy- luciferaseactivities.Allchemicalstestedweredissolvedanddiluted inDMSO.Thefinalsolventconcentrationwas0.1%DMSOforsingle chemicaltreatmentand0.2%forsimultaneousagonistandantag- onistexposure.Fulldetailsofthemethodologyareprovidedinthe Supplementarymaterial.

2.7. Dataanalyses

Unlessstatedotherwise,thedataarepresentedasmeans±SEM andaprobabilitylevelofp<0.05wasconsideredtobestatistically significant.Generallinearmodels(GLM)wereusedtoanalysefor effectsoftreatmentongrowthoffishwheretankwasincludedas arandomfactorinthemodel(nestedwithintreatment)inorderto adjustforpseudo-replicationoccurringduetomultiplemeasure- mentsbeingtakenwithinonetank.Anysignificantdifferenceswere determinedbyTukey’sHSDposthoctests.Althoughtankeffectsare accountedforinthefullstatisticalmodel,apairedt-testwasalso appliediftankeffectswerefoundtoestablishwhichtreatments wereresponsibleforthetankeffects.Thesameanalysiswasapplied totestforeffectsofDCPonspigginlevelsinthesticklebackexperi- ment.Estimatedmarginalmeans,whichareafunctionofthemodel parametersandareadjustedforthefactorsinthemodel,werecal- culatedandplottedforeachgroup.Apairedt-testwasusedtotest fordifferencesmeasuredchemicalconcentrationsbetweenrepli- catetanks.Toanalyseforrelationshipsbetweentreatmentandsex intheroachexperiments,chi-squaretestswereapplied.Statistical analyseswereperformedusingIBMSPSSStatisticsforWindows, Version21.0.0

Fortheinvitroexperiments,alltransfectionswereperformedin triplicateandtheassaysrepeatedatthreetimes.Dataarepresented asmean±SEMfromthreeseparateexperiments.Dose–response datausingafour-parametriccurveﬁtting,EC₅₀(foragonists)and IC50(forantagonists)wereanalysedusingGraphPadPrism(Graph PadSoftwareInc.,SanDiego,CA,USA).Relativepotenciesofthe agonistswereestablishedbycomparingtheirEC₅₀withtheEC₅₀ of11-KTandpotenciesoftheantagonistswerecalculatedrelative toFL.

Results

3.1. Waterchemistry

The mean measured exposure concentrations for the different experiments are shown in Table 1 and the average concentrations for each replicate tank are provided in Supple- mentary material (Tables S1–3). Recoveries were variable for theindividual compounds.For thesticklebackexposure, actual measured tank DHT concentrations across the different tanks werebetween3.58±0.26and4.20±0.12␮g/L(nominal5␮g/L), andforFL,132.44±2.72␮g/L(nominal150␮g/L).ForDCP,mea- suredconcentrationswere0.041±0.010␮g/L(nominal0.1␮g/L), 0.570±0.060␮g/L(nominal1.0␮g/L)and7.02±0.736␮g/L(nom- inal10␮g/L).MeasuredDCPconcentrationsdifferedsigniﬁcantly betweenthereplicatetanksforthemedium(t-test:p=0.028)and high(t-test:p=0.016)DCPtreatmentgroupsandforDHTconcen- trationsbetweenthetworeplicatetanksinthehighDCPtreatment groups(t-test:p=0.02).Allothermeasuredchemicalconcentra- tionsdidnotdiffersigniﬁcantlybetweenthetworeplicatetanks (t-test:0.102<p<0.696;seeSupplementarydata).

For the ﬁrst roach exposure, the measured DCP concentra- tionswere0.36±0.05␮gDCP/L(nominal1.0␮g/L),3.52±0.88␮g DCP/L(nominal10␮g/L),10.39±2.60␮gDCP/L(nominal30␮g/L), and for FL, 136.79± 5.77␮g FL/L (nominal 150␮g/L) and 355.14±21.88␮gFL/L(nominal450␮g/L).Therewerenostatis- ticallysigniﬁcantdifferencesinchemicalconcentrationsbetween thetworeplicatetanks(t-test:0.135<p<0.978;seeSupplementary data).

Inthesecondroachexposure(mixtureexperiment),theace- tonestocksusedfordosingandthestocksofinternalstandards were analysed and concentrations conﬁrmed as 98% for TCS, 103% for EE2, 108% for AbA, 114% for CP, and for the internal standards as 123% (4-(4-chlorophenoxy) phenol) and 90%

(oestrone)ofnominalconcentrations.Intheexposuretanks,recov- eriesweremuchlower.Theactualmeasuredtankconcentrations werebetween3.87±0.67 and6.40±0.792␮g/L for AbA(nomi- nal50␮g/L),between23.18± 2.06and23.86±1.25␮g/L forCP (nominal50␮g/L),between7.11±1.05and9.10±.87␮g/LforTCS (nominal50␮g/L)andbetween1.29±0.05 and1.60±0.60ng/L for EE2 (nominal 3ng/L). Measured CP concentrations differed signiﬁcantlybetweenthebothreplicatetanksforboth,themix- tureofantiandrogenstreatment(t-test:p=0.031)andthemixture ofantiandrogensco-administeredwithEE2(t-test:p<0.001).All othermeasuredchemicalconcentrationsdidnotdiffersigniﬁcantly betweenthetworeplicatetanks(t-test:0.078<p<0.896;seeSup- plementary;TablesS1–S3).

3.2. EffectsofDCPonspigginproductioninmature,androgenised femalesticklebacks

Intotal,125fishwereanalysed(14–20pertreatment).Themean length,massandconditionfactorofthefishwere4.81±0.04cm, 1.23±0.03g and 1.09±0.01, respectively. There were no sta- tistically significant differences between treatment groups for length(F_6,7.041=0.700,p=0.660),weight(F_6,7.276=0.688,p=0.667) and conditionfactor (F6,7.008=0.573,p=0.743) (seeSupplemen- tary material; Fig.S1). With theexception for fish exposed to 10␮gDCP/L(co-administeredwithDHT),measuredspigginlevels did not differ significantly in fish sampled between the replicate tanks (0.027<p<0.948; see Supplementary material; Fig.

S2). Overall, treatmenthad signiﬁcanteffects onspiggin levels (ANOVAF6,7.047=36.721,p<0.0001).Post-hoccomparisonsusing the Tukey’s HSD test showed no differences in spiggin levels betweenwatercontrolandsolventcontrol.Incontrast,DHTexpo- sure signiﬁcantly induced spiggin levels compared with both

(6)

Fig.1.Spigginunits/gbodyweightinandrogenisedfemalethree-spinedsticklebackexposedtoDCP.FishweresimultaneouslyexposedtoDHT(nominal5␮g/L)andone ofthreeconcentrationsofDCP(nominal0.1,1.0or10␮g/L)ortheantiandrogen-positivecontrolFL(nominal150␮g/L).Theexperimentincludedadilutionwatercontrol (DWC),asolventcontrol(SC)andanandrogen-positivecontrol(DHTalone).Dataarepresentedasestimatedmarginalmeans±SEM(n=14–20ﬁshforeachtreatment).

Statisticallysigniﬁcantdifferencesbetweenexperimentalgroupsaredenotedbydifferentletters(p<0.05;GLMfollowedbyTukey’sHSDtest).DWC,dilutionwatercontrol;

SC,solventcontrol;DCPL,0.1␮gDCP/L;DCPM,1.0␮gDCP/L;DCPH,10␮gDCP/L.

controlsas expected. Co-administration of FLwith DHT signif- icantlyreduced spiggin levels compared withexposuretoDHT aloneasexpected.Spigginlevelsinthelow(0.1␮g/L)andmedium (1.0␮g/L)DCPtreatmentgroups(co-administeredwithDHT)were signiﬁcantlylowercomparedwiththeDHTgroup,buttherewere nodifferencesforthehigh(10␮g/L)DCPtreatmentgroup(Fig.1).

Therewerenotankeffects onspigginlevelsor ongrowth,but therewasatankeffectonconditionfactor(ANOVAF7,111=8.353, p<0.0001)(seeSupplementarymaterial;TableS4).Althoughtank effectsareaccountedforinthefullstatisticalmodel,apairedt-test wasalsoappliedtotestfordifferencesbetweenthereplicatetanks iftankeffectswereseen.Theobservedtankeffectsforcondition factorweredrivenbyonlytwotreatmentswhereastherewereno tankeffectsfortheremainingﬁvetreatments(0.234<p<0.611).

Thetankeffectswereaccountedforinthemodelanddonotmask theeffectsoftreatmentonconditionfactor.

3.3. InvivopotencyofFLandDCPongonadalsexual differentiationofroach

The mean length, mass and condition factor of male and female fish were 4.15±0.04cm/0.76±0.02g/1.05± 0.01 and 4.17±0.04cm/0.77±0.03g/1.05±0.01, respectively. Therewere nostatisticallysignificantdifferencesbetweentreatmentgroups forlength,weightandconditionfactor(seeSupplementarymate- rial;Fig.S3).Therewerenotankrelatedeffects ongrowth,but therewasatankeffectforconditionfactorinfemaleroach(ANOVA F_4,35=5.631,p=0.001)(seeSupplementarymaterial;TableS4).At 172dph,therewerenoeffectsofthehighestexposureconcentra- tionsforDCP(30␮g/L)orFL(450␮g/L)onthesexratioofexposed roachorondevelopmentofthereproductiveduct,asdetermined bygonadhistopathology.IntheDWC,57%and43%ofthefishcould

beassignedasmalesandfemales,respectively,andintheSC,73%

and27%asmalesandfemales,respectively(n=30ﬁshanalysed fromeachtreatment).IntheFLexposure(nominal450␮g/L),70%

oftheﬁshweremalesand30%females(n=30).Intheexposureto DCP(nominal30␮g/L),45%weremalesand55%females(n=29).

Norelationshipwasfoundbetweentreatmentandthesexofﬁsh (² (3,n=119)=3.278(p=0.351)).Nodifferenceswereobserved forductformationinanyofthetreatments(²(3,n=114)=5.487 (p=0.139));ductdevelopmentwasunclearfor5histologicalsam- plesduetotechnicalproblemswithtissueprocessing.

3.4. EffectsofantiandrogensaloneandincombinationwithEE2 ondevelopmentofthereproductiveductinroach

The mean length, mass and condition factor of male and female fish were 3.93±0.06cm/0.92±0.04g/1.42± 0.02 and 3.85±0.07cm/0.87±0.04g/1.42±0.02, respectively. There were nostatisticallysignificantdifferencesbetweentreatmentgroups forlength,weightandconditionfactor(seeSupplementarymate- rial; Fig. S4). There were no tank effects on morphometric endpointsmeasuredinmaleroach,butthereweretankeffectsfor growth(length(ANOVAF_4,78 =3.316,p<0.05)),weight(ANOVA F4,78=4.722,p<0.01)andconditionfactor(F4,78=6.396,p<0.001) infemaleroach(seeSupplementarymaterial;TableS4).Feminising effectsoftreatmentonreproductiveductswereanalysedformale fish(Fig.2).Therewasa significantrelationshipbetweentreat- mentandfeminisedductsinmaleroach(²(6,n=109)=88.345(p<

0.001)),i.e.feminisationofreproductiveductsinmaleroachwas associatedwithtreatment.Asigniﬁcantdifferencebetweenrepli- catetankswasobservedinthedevelopmentofreproductiveducts incontrolmalesonly(²(1,n=38)=4.479(p<0.05))withnosignif- icantdifferencesbetweenreplicatetanksfortheothertreatments

(7)

Fig.2.Effectsofearlylifeexposuretoamixtureofantiandrogens,EE2,oramixtureofantiandrogenstogetherwithEE2,onthedevelopmentofreproductiveductsin roach.(A–I)Representativehistologicalsectionsfortreatmentgroups:(A)ovaryofcontrolfemale,(B)testisofcontrolmale,(C)ovaryoffemaleexposedtothemixtureof antiandrogens,(D)testisofmaleexposedtothemixtureofantiandrogens,withanormalduct,(E)ovaryoffemaleexposedtoEE2,(F),testisofmaleexposedtoEE2,with anormalduct,(G)testisexposedtoEE2,withafeminisedduct,(H)ovaryoffemaleexposedtothemixtureofantiandrogensandEE2,and(I)testisinmaleexposedtothe mixtureofantiandrogensandEE2,withafeminisedduct.Scalebars:100␮m.(J)Proportionsofmale-andfemale-likereproductiveductsinmaleroachexposedtoamixture ofantiandrogens,EE2,oramixtureofantiandrogenstogetherwithEE2.Relationshipsbetweentreatmentandfeminisationofreproductiveductsinmaleroachwereanalysed bychi-squaretestandsigniﬁcantrelationshipsareindicated(*p<0.05and***p<0.001).Fullstatisticalresultsforchi-squaredtestsandrelationshipsbetweentreatmentand feminisationofreproductiveductsinmaleroachareprovidedinSupplementarymaterial(TableS6).Numbersineachbarrepresentthenumberofﬁshanalysed.

(8)

(seeSupplementarymaterial;Fig.S5&TableS5).Thedifferencefor ductsbetweenreplicatecontroltanksresultedbecauseitwasitwas notpossibletoclearlyidentifytheductsininallroachinthistreat- mentseventhoughwecouldclearlydeﬁnemalesbasedontheir gonadmorphologyandgermcells.(seeSupplementarymaterial;

Fig.S5&TableS5).

Thereproductiveductsof38of41malessampledfromthecon- troltankshaddevelopednormally(itwasnotpossibletoassignthe statusofthereproductiveductsforthefewremainingmaleﬁsh).All males(n=25)sampledafterexposuretothemixtureofTCS,CPand AbAhadnormal,male-likereproductiveducts.Therewerestatisti- callysigniﬁcantdifferencesintheproportionsoffeminisedductsin maleroachexposedtoEE2aloneorincombinationantiandrogens comparedtoroachthetreatmentsnotincludingEE2.

Nosignificantdifferencewasobservedinthedevelopmentof reproductiveductsbetweencontrolmalesandmalesexposedtoa mixtureofTCS,CPandAbA(²(1,n=63)=0.385(p=0.535)).After exposureto1.6ngEE2/L(averagemeasuredconcentration),23of thesampledfishweremales.Ofthese,70%ofmalegonadshad developednormalducts,whereas17%ofmalegonadshaddevel- opedfeminisedducts(itwasnotpossibletoassignthestatusof thereproductiveductsforthefewremainingmalefish).Thisfind- ingwasstatisticallysignificantdifferentfromcontrolmales(²(2, n=61)=7.866(p=0.02))andmalesexposedtoamixtureofTCS,CP andAbA(²(2,n=48)=6.528(p=0.038)).

ForthecombinationofthreeantiandrogenswithEE2,23ofthe sampledfishweremalesofwhich96%haddevelopedfeminised reproductiveducts(itwasnotpossibletoassignthestatusofthe reproductiveductsforthefewremainingmalefish).Thisfinding wassignificantlydifferentfrommalesintheotherthreetreatment groups(control:(² (2,n=61)=57.807(p=2.801×10⁻¹³));mix- tureofTCS,CPandAbA:(²(2,n=48)=45.997(p=1.028×10⁻¹⁰)) orEE2alone:(²(2,n=46)=29.462(p=4.004×10⁻⁷))).Theresults ofallstatisticalanalysesforthedevelopmentoffeminisedductsin maleroacharesummarisedinSupplementarymaterial(TablesS5

&S6).

3.5. Invitroinhibitionofandrogenreceptor(AR)activationby environmentalantiandrogens

Bothmodelantiandrogens,FLandPROCYinhibitedtheactiva- tionofbothsticklebackARsactivatedby11-KTorDHT.TheIC50

valueswere1.85×10⁻⁷MFL(11-KT,AR␣),2.25×10⁻⁷MFL(11- KT,AR␤), 5.24×10⁻⁷ MFL(DHT,AR␣), 1.78×10⁻⁷ MFL(DHT, AR␤), 2.72×10⁻⁶ MPROCY(11-KT,AR␣), 7.33×10⁻⁶ MPROCY (11-KT,AR␤), 1.52×10⁻⁶ MPROCY(DHT,AR␣)and5.69×10⁻⁶ MPROCY(DHT,AR␤)(Fig.3A&B).Incontrast,theenvironmen- talantiandrogensTCS,CP,DCPandAbAdidnotinhibitthe11-KT inducedactivationofsticklebackAR␣(Fig.3C)orAR␤(Fig.3D).

4. Discussion

Whilsttheeffectsofenvironmentaloestrogensonsexualdevel- opmentinfishhavebeenwellstudied,littleattentionhasthusfar beenpaidtoantiandrogensthataretakenupandbioconcentratein fish,includingforenvironmentallyrelevantmixturesorforcombi- nationswithrelevantenvironmentaloestrogens.Here,weexposed fishtocompoundsidentifiedinWwTWeffluentsandshowntobe antiandrogeniconthehumanARinvitro,toestablishtheireffectsin vivo(insticklebackandroach).Furthermore,weappliedstickleback ARtransactivationassaysinanattempttosupportamechanistic understandingfortheeffectsoftheantimicrobialantiandrogens seeninvivo.

Fortheinvivoexposures,concentrationsofthechemicalstocks (in solvent) measured prior to the start of the second roach

experiment,wereconsistentwithnominals.Inthetankexposures wheresolventwasusedasacarriertherecoveriesofthediffer- entchemicalswere72–84%forDHT,88%forFLand 47–70%for DCPin thesticklebackexperiment andbetween79and91%for FLand35–36%forDCPin theroachexperiment, buttheywere consistentwithin anyone treatment.In thesolventfree exposure(roachmixtureexperiment)recoveriesofthechemicalswere muchlowerat8–13%,14–18%and46–48%forAbA,TCSandCP, respectively.Possiblereasonsforthelowermeasuredconcentra- tionsoftheantiandrogenicchemicalsintheexposuretanksinclude thoserelatingtochemicalstability,adsorptionandlossesduring theextractionprocess.TCS,CPandDCPhaveallbeenshownpre- viouslytobephotodegradableintheaquaticenvironment,forTCS evenunderlowlightintensities(Werneretal.,1983;Mansﬁeld andRichard,1996;Latchetal.,2005;Loresetal.,2005;Aranami andReadman,2007;Fangetal.,2010).Someofthechemicalloss couldbeaccountedforbyadsorptiontothetankwallsasallthree antiandrogenshave highpredictedlogKow values(4.41CP,5.17 TCS,6.51AbA,ACD/Perceptasoftware—Classicedition,ACDLabs, Toronto,Canada).ForTCSandCP,theyarealsoknowntoadsorbto suspendedsolids(HazardousSubstancesDataBank,2006,2012) andthiscouldhaveoccurredduetothepresenceofuneatenfood andfaecesintheexposuretanks.Lossesofthechemicalsduring theextractionprocess are less likely asthe sameprotocol and equipmentwasusedforallthreeinvivoexperimentsandtherecov- eriesforDHT(76%)andFL(79–91%)weremuchhigherandsimilar topreviouslypublishedranges(Panteretal.,2004;Sebireetal., 2008;Pottingeretal.,2013b).Despitethelowrecoveriesforsome chemicalsthemeasuredconcentrationswererelativelyconsistent overthelongexposuretimesandtheyprovidedappropriatedos- ingregimensforthepurposeofthesestudies.Forsomechemicals (sticklebackexperiment:DCPinthe1and10␮g/Lexposuregroups andDHTinthe10␮gDCP/Lexposuregroup;roachexperiment2:

CPinthemixtureofCP,TCSandAbAgroups),thereweredifferences inmeasuredconcentrationsbetweenreplicatetanks.Althoughthis wasnotdirectlytakenintoaccountinthestatisticalmodel,compar- isonsofbiologicaleffectsinreplicatetankssuggestthat(withthe exceptionforsticklebackexposedto10␮lDCP/L)biologicaleffects seenwereasaconsequenceofthechemicaltreatments.

Inthesticklebackinvivoexposuresbasallevelsofspigginwere

<70spigginunits/gbodyweightinthecontrolgroups,spigginwas inducedbyexposuretoDHTandtheinductionwasinhibitedinfish co-exposedtoDHTandFL,asexpected(OECD,2011).Co-exposure offemalesticklebacktoDHTandDCPcausedasignificantdecrease in spiggin levels for the low (0.1␮g/L) and medium (1.0␮g/L) concentrationofDCP,supportingthefindingsfromstudieswith recombinantyeasthARassays(Rostkowskietal.,2011)thatDCP canactasanantiandrogen.Incontrast,however,exposuretothe highconcentrationofDCP(10␮g/L),didnotinhibitspigginpro- duction.AntiandrogensthatactthroughtheARhavebeenshown toinduceamonotonicconcentration-dependentinhibitionofspig- ginintheAFSS(OECD,2010;Knagetal.,2013;Pottingeretal., 2013b),anditishardtoconcludethereforeonwhetherDCPcan convincinglybeascribedasantiandrogeninthesticklebackspig- ginbioassay.Takingintoaccountthechemistryresultsforthehigh DCPtreatmentgroup,theconcentrationofDHTonaverageappears tobehighercomparedwiththatdosedinthelowandmedium DCPcombinationtreatmentswhichcouldexplainwhythesetwo DCPtreatmentsresultedinaslightlylowerinductionofspiggin.

NomortalitieswereobservedinthehighdoseDCPtreatment,soit isunlikelythatthelackofspiggininhibitionwascausedbycyto- toxicity,supportedbytheconditionfactoroffishwhichdidnot differbetweenfishexposedtoDHTaloneandfishexposedtoDHT togetherwiththehighdoseofDCP.

Theperiodofsexdifferentiationiswhentheinterplaybetween oestrogens and androgens is critical in determining sexual

(9)

Fig.3.Dose–responsecurvesforexposuresto(A)ﬂutamide,(B)procymidoneand(C&D)theenvironmentalantiandrogensTCS,DCP,CPandAbA,insticklebackARreporter assays.Resultsareexpressedasmeans±SEM,n=3.Thecellswereco-exposedto11-KTataconcentrationofEC65foreachreceptorandtheY-axisindicatesfold-change comparedtotheactivityof11-KTalone.

developmentandthisperiodisthusmostlikelytobesusceptibleto effectsofEDCsthatactthoughtheoestrogenandandrogenaxes.In fish,whilsttheroleofoestrogensinovariandifferentiationiswell established,controversyexistsabouttheroleofandrogensinthe regulationofgonadalsexdifferentiation(Ijirietal.,2008).Inthe roachexperiments,theendpointsofgonadalsexandfeminisation ofreproductiveductswerechosenforstudybasedontheirknown responsestooestrogenexposure.Exposuretothemodelantian- drogenFLhadnoeffectonsexratiosoffishcomparedtocontrols, whichcompareswellwithastudyonsteelheadtroutexposedtoFL (Soweretal.,1983),butcontrastswithastudyonguppiesexposed toFL(Bayleyetal.,2002).FLhasbeenshownalsotoinhibittestic- ulargrowthand/orspermatogenesisinotherfishspecies(Baatrup and Junge,2001; Jensenet al., 2004; León et al.,2007).In our study,wewereunabletoidentifyfinescaledifferencesingerm cellsand/orsomaticcelldevelopmentinmalesduetoatechnical problemwiththegonadfixingconductedinanautomatedtissue processingsystem.Collectively,however,fromthepublisheddata, responsestoFLinfishappeartodifferbetweenfishspecieswhich mayrelatetodifferencesintheimportanceofandrogensinfish.

Speciessuchasguppyandsticklebackforinstance,showstrong androgen-relatedreproductivebehavioursand assuchmightbe moresusceptibletotheeffectsofantiandrogensonbehaviourthan forsomeotherfishspecies.Inmammals,thehydroxymetabolite ofFLisactiveontheARanddifferencesinmetabolismbetween fishmayaccountforsomeofthedifferencesintheirresponsesto FL.Similarly,theantiandrogens(DCPorthemixtureofCP,TCSand AbA)hadnoeffectonthesexratioorfeminisationofthereproduc- tiveductsinroach.Thisfindingalignswitharecentstudyfinding thatpharmaceuticalantiandrogensatenvironmentalconcentra- tionshadnoeffectsonsecondarysexualcharacteristicsinfathead

minnowanddonotinduceintersexinJapanesemedaka(Green etal.,2015).

Whenroachwereexposed toEE2aloneatanexposurelevel reportedforsomeUKWwTWeffluentsorsurfacewaters(Williams etal.,2003,2012),17%ofmalegonadshaddevelopedfeminised ductswhichalignswiththefactthatthedevelopmentofovarian cavitiesisamechanismdrivenbyoestrogens(Rodgers-Grayetal., 2001).Perhapsthemostintriguingfindingfromourstudiesisthat exposuretoamixtureofthreepotentialantiandrogensandEE2 togetherresultedinaconsiderablyenhancedfeminisingeffectof theoestrogenonductdevelopmentinmales(96%oftheexposed maleroach)despitethefactthatthiscombinationofantiandro- genshadnoeffectonthemaleductsintheabsenceofEE2.This findingisincontrastwitharecentreportwhichshowednoeffects ofco-exposureoftwopharmaceuticalantiandrogenstogetherwith oestrogens,eachattheirpredictedenvironmentalconcentrations, onfeminisation of fish(Green et al., 2015).However,we used a different setof antiandrogenic chemicals to that usedin the workreportedbyGreenetal.(2015),andtheiractionsaspoten- tialantiandrogensmaydiffer(seelater).Inourstudy,weexposed roachtoamixtureofthreeantiandrogenicchemicalsatconcen- trationswhich reflectedthetotallevelsofantiandrogenactivity commonlymeasuredinWwTWeffluents(EnvironmentAgency, 2007).Thisresultedinameasuredconcentrationoftotalantiandro- genicactivityinthecombinationexperimentof360␮gflutamide equivalents/L(calculatedonpotencyofindividualchemicalsina yeastreceptor transcription assay,YAS).As discussedby Green etal.(2015),thelikelihoodisthatthesuiteofantiandrogeniccom- poundsintheenvironmentcollectivelyresultinpotentiationof oestrogen-inducedfeminisationoffishratherthanafewselected contaminants. Interestingly, another study recently found that

(10)

co-treatmentof juvenile Murrayrainbowﬁsh(Melanotaenia ﬂu- viatilis) with FL and E2 did not lead to additive reproductive impairmentinthisspecies(Bhatiaetal.,2015).

Although the bioavailable ‘antiandrogens’ tested have been showntoinhibitthehumanARinvitro(Rostkowskietal.,2011), theydidnotappeartobeantiandrogenicinﬁshinvivo,toinhibit spiggininductionoraffectsex,evenasamixture.Thiswassup- portedbythetransactivationassaysusingsticklebackARswhere noneofthetestedcompoundsinhibitedtheactivationofAR.It couldbearguedthathepatocellularcarcinomacellsofthetransac- tivationsystemmightpotentiallymetabolisethetestcompounds resultinginnon-antiandrogenicmetabolites, butthis isunlikely asHepG2cells possessverylow tonodrug-metabolising activity(Fukudaetal.,2002;Rodríguez-Antonaetal.,2002;Wilkening etal.,2003).Itispossiblethatdifferencesinantiandrogenicactiv- itybetweenhumanandﬁshrelatetoarelativelylow sequence identitybetweenthehumanandteleostAR,whichisaround40%

(Olsson etal.,2005;Hossain etal.,2008).In theligandbinding domain,thesequencesimilaritiesareonly62–73%betweenhuman andvariousﬁshspecies(Touhataetal.,1999;Olssonetal.,2005;

Hossainetal.,2008;Oginoetal.,2009).Speciesdifferencesinbind- ingspeciﬁcitiesof(anti)androgenstotheARhavepreviouslybeen showninstudiescomparingdifferentspeciesincludingrainbow troutandgoldﬁshARs(WellsandVanDerKraak,2000)andrain- bowtrout,fatheadminnowandhumanAR(Wilsonetal.,2007).

Interestingly,thelatterstudyreporteddifferencesintherelative orderofbindingofcompoundstotheARsandthereforeconcluded thatspecies-specificEDCsexist(Wilsonetal.,2007).Moreworkis clearlyrequiredtobetterestablishwhetherchemicalsmorewidely identifiedasantiandrogensonthemammalianARarealsoactive onfishARs.

Giventhatweshowthattheantiandrogenstesteddidnotappear tointeractwithﬁshARsintheinvitrotransactivationassays,they aremostlikely tohave contributed tothefeminisationof duct developmentinducedbyEE2invivothroughanothermechanism.

Onepossibilitywouldbethrougheffectsonsteroidogenesis.TCS hasbeenshowntoimpairthepituitary–gonadalpathwayinmale rats,viareduction of LH and cholesterol production, depressed StAR expression, and down-regulationof several key steroido- genicenzymes(CYP11a,CYP17,3␤-HSD,and17␤-HSD)(Kumar et al., 2009) resulting in an inhibition of androgen production.

Interestingly,studiesonratshaveshownthatTCSpotentiatesthe estrogeniceffectofEE2onuterinegrowthintheuterotrophicassay, whereasTCSalonehadnoeffects(Stokeretal.,2010;Louisetal., 2013).Twopossiblemechanismswereproposedforthepotenti- ationoftheoestrogenic effectof TCS,eitherthroughenhancing theinteractionofoestrogenswiththeoestrogenreceptor(ER)or increasingendogenous oestrogenlevelsby inhibitingoestrogen metabolism(Stokeret al.,2010).Bothmechanismsarepossible explanationsforthefeminisingeffectsofthetestcompoundson reproductiveductdevelopmentobservedinourstudyonroach.

AbA has previously been described as an inhibitor of 5␣- reductase(Rohetal.,2010),theenzymeconvertingtestosterone intoDHT,andinhibitionof5␣-reductaseinfatheadminnows(using dutasteride) results in effects onreproductive functions which areconsistentwithanantiandrogenicmode-of-action(Margiotta- Casalucietal.,2013).Tothebestofourknowledge,nodataare availablefortheeffectsofDCPandCPonsteroidogenesis.

Giventhatsomeofthechemicalsinvestigatedherearelikely tohaveanantiandrogenicmode-of-actionthroughactingonthe steroidogenicpathway,theratioofoestrogenstoandrogensmight play a role in the observed induction feminised ducts in male roach.Exposuretothemixtureofantiandrogenscouldalterthe balanceofthecirculatinglevelsofandrogensandoestrogens(as androgensaretheprecursorsofoestrogens),reducingtheoverall levelsofcirculatingsteroids,butnottheirratio.Exposuretoalow

concentration of exogenous oestrogen (EE2), would shift this ratiotowardshigheroestrogensresultinginfeminisedducts.Co- exposuretoantiandrogensandoestrogen,mightreducecirculating levels of androgenswhilst increasinglevels of oestrogens such shiftingtheandrogen:oestrogenratioinfavourofoestrogenactiv- itytoinducefeminisationofreproductiveductsinthemales.This howeverisahypothesisonlyatthistimeandfurtherworkisneeded toconﬁrmtheeffectofantiandrogensonpotentiationoffeminisa- tionofﬁshandtheirmodesofaction.Equallyitispossiblethatthe effectsseenaremanifestedthroughalterationstofeedbackmech- anismsonthehypothalamus–pituitary–gonad(HPG)axisaffecting thedynamicsfortestosteroneproduction andaromatisation(to oestrogen).

Inconclusion,ourresultsshowthatcompoundsthatarecon- tainedineffluentsandareantiandrogenictothehumanARmight notnecessarilybeantagoniststofishARsdespitetheirpotencyto bioaccumulateinfish.Nevertheless,amixtureofthese‘antiandrogens’togetherwithEE2inducedanenhancedfeminisingeffectof ductdevelopmentinmaleroachshowingacombinationeffectin thisspecies.Thesedataaddfurthertothehypothesisthatoestro- gens and antiandrogens act in combination to feminise males andimpairreproductivehealthinwildfishpopulations.Ourdata alsohighlighttheneedforhomologousassaysforextrapolating betweeneffectsinvitroandinvivo.Furthermore,itisimportantto considereffectsacrossmultipleinteractionsitesinordertotruly understandmixtureeffectsofchemicalsonanimalhealth.

Acknowledgments

WethankmembersoftheEnvironmentandEvolutionresearch groupattheUniversityofExeterfortheirsupport,IoannaKatsi- adaki(Cefas)foruseofthespigginELISA,AlanHenshawandhis teamatCalvertonFishFarm(UKEnvironmentAgency)fortheir supportintheroachwork;CrisBurgess,ErinReardon,Matthew Winter,EduardaSantosandFrancesOrton(UniversityofExeter) forhelpfuldiscussionsandadviceonstatisticalanalyses.Thiswork wasfunded bytheUKNaturalEnvironmental ResearchCouncil (NE/E016634/1andNE/E017363/1and)toCRTandEMH.Thiswork wasfurthersupportedbyDefra(UK),UK-JapanResearchCollabo- rationGrantsfromtheMinistryoftheEnvironment,Japan(MS,TI), andGrants-in-AidforScientiﬁcResearchfromtheJapanSocietyfor thePromotionofScience(MS,TI).

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.aquatox.2015.09.

014.

References

Alvarez-Mu ˜noz,D.,Indiveri,P.,Rostkowski,P.,Horwood,J.,Greer,E.,Minier,C., Pope,N.,Langston,W.J.,Hill,E.M.,2015.Widespreadcontaminationofcoastal sedimentsintheTransmancheChannelwithanti-androgeniccompounds.

Mar.Pollut.Bull.95,590–597.

Aranami,K.,Readman,J.W.,2007.Photolyticdegradationoftriclosaninfreshwater andseawater.Chemosphere66,1052–1056.

Baatrup,E.,Junge,M.,2001.Antiandrogenicpesticidesdisruptsexual characteristicsintheadultmaleguppy(Poeciliareticulata).Environ.Health Perspect.109,1063–1070.

Bayley,M.,Junge,M.,Baatrup,E.,2002.Exposureofjuvenileguppiestothree antiandrogenscausesdemasculinizationandareducedspermcountinadult males.Aquat.Toxicol.56,227–239.

Bellet,V.,Hernandez-Raquet,G.,Dagnino,S.,Seree,L.,Pardon,P.,Bancon-Montiny, C.,Fenet,H.,Creusot,N.,Ait-Aissa,S.,Cavailles,V.,Budzinski,H.,Antignac,J.P., Balaguer,P.,2012.Occurrenceofandrogensinsewagetreatmentplants inﬂuentsisassociatedwithantagonistactivitiesonothersteroidreceptors.

WaterRes.46,1912–1922.

Bhatia,H.,Kumar,A.,Du,J.,Chapman,J.C.,McLaughlin,M.J.,2015.Co-treatment withthenon-steroidalanti-androgendrug,ﬂutamideandthenatural