苯胺及其衍生物联合毒性的研究毕业论文

苯胺及其衍生物联合毒性的研究毕业论文目录1绪论 ）的幂值分别计算（表1）。Logistic方程提供了一个良好的适合马拉硫磷（r2=0.987），对硫磷（r2=0.987），和胡椒基丁（r2=0.998）浓度-反应数据。这展出的两个有机磷农药相似的毒性特点。相对于有机磷物质，胡椒基丁的毒性想到较弱，而且幂值大约是有机磷物质的一半。储存器（盒子）的分配根据IAI的分析图，将不同浓度的有机磷酸脂配置到同一暗盒中，在暗盒中可近似地计算出某一附加浓度值来确定其毒性。用附加浓度来建立有机磷酸酯的联合毒性分析图，准确度是由两种毒性相当的有机磷酸酯多次联合实验后来确定的。用浓缩反映来评价二元混合物在数学统计上是难以分辨的。因此，马拉硫磷和对硫磷对最终混合物的毒性作用结果需用单一的有机磷酸酯来建分析图。有机磷酸脂抑制乙酰胆碱酯酶活性的普通作用模式可根据实验数据来确定。对比之下，胡椒基丁并没有抑制乙酰胆碱酯酶的活性。因此胡椒基丁可配置到它对应的暗盒中，在暗盒里，混有这种成分的混合物的毒性与使用了反应添加剂的混合物的毒性完全复合。化学相互作用我们假设胡椒基丁可以以一种形式和有机磷酸酯的成分完全混合，并可以减小它的毒性。图2直接表明了胡椒基丁撤销有机磷酸脂对乙酰胆碱酯酶的抑制能力。胡椒基丁经过长期对有机磷酸脂的毒性作用后可证明为拮抗作用，然后做出马拉硫磷的浓度反应曲线（图3A）和对硫磷（图3b）的浓缩反应曲线。胡椒基丁的抵制作用是用来表K-函数的浓度（如图4）在混合溶液中，这些K-函数常被用于最终IAI模型中，他可以减小被胡椒基丁浓度控制的马拉硫磷和对硫磷的有效浓度。混合物毒性的估算根据实验确定的30种成分的三元混合物与被指定的毒物进行了实验测定和比较，绘出浓度加法模型图（等式2），反应加法模型图（等式3），综合加法模型图（等式4），IAI模型图（等式6）。并不是浓度加法，反应加法和综合加法模型图能描述混合物的毒性（r2<0.10）。而是所有模型都远远超出其混合毒性。因此，IAI模型图对各种毒性混合物可提供准确的数据。83%的混合元素在表2中能精确估算出其毒性。讨论这项研究的结果表明，毒物动力学相互作用可以运用到综合加法模型图中来估算混合毒性。最近的研究表明，浓缩及反应加法模型都能用于创造简单加法模型，并用此模型来计算无相互化学作用的混合物的毒性（Altenburgeretal.,2005年；OlmsteadandLeBlanc，2005年；Teuschler等，2004年）。我们可以根据活性毒药和混合成分之间的联合作用建立其结构模型。根据定义，建立混合毒性模型时，因为简单的加和作用使化学相互作用表现的很不正常。量化这些相互作​​用，预期要增大混合物毒性，必须先确定选择适当的模型估算添加量，并准备解释其相互作用的结果。美国环保局评估混合毒性建议使用浓度模型（2000）。浓度加法模型与反应加法模型相比是一种更趋向保守估计混合物毒性浓度反的方法（Drescher和Boedecker，1995）。不过，不加区别的应用浓度加法模型会导致缺乏正确的机械基础，这就增加了混合物毒性的的不确定因素。综合加法描述了在近期研究中（Altenburger，2005年；OlmsteadandLeBlanc，2005年）提供了一种基础的方法来评估混合物的毒性。起初，类似于这种机制的化学反应都被置于群体或暗匣中。每个暗盒的毒性可以参考浓度加法模型，不同暗盒的总毒性可以参考反应加法模型（图6）。综合加法模式由Altenburger等人创建（2005年），奥姆斯特德和勒布朗（2005年）定理在概念上是等同，在计算方法上有点轻微的差别。综合加法模型在评估非相互作用的化学混合物毒性上市一个重大的进步。但是，这种模式并不是具备管理之间的相互作用，化学物质影响该混合物的毒性。在环境中两个或更多的化学物质之间发生协调作用的可能性很大，也许只有研究混合毒性才能解释这种现象。明确界定的互动关系的例子包括：预暴露在kepone的四氯化碳对肝的毒性（Klingensmith和Mehendale，1982年）和涉及激素受体拮抗剂和激素合成抑制剂的相互作用（木和勒布朗，2004年）。相互作用时可利用机制作用推出的化学成分。例如，用于本研究中的细胞色素P450酶抑制剂增效醚是假设对抗通过降低其代谢活化抵消马拉硫磷和对硫磷的毒性。然而，一些互动显然对某些机制成分无反应，综合加法模式有可能找出些意想不到的相互作用。实际上，实验结果预测模型的重大偏差就意味着互动。只有通过遵循模式推理或实验量化与团之间的相互作用来确定源的相互作用。毒物动力学的相互作用可以通过一个质“重证据”的方法或定量的方法评估混合物性质。这两种方法在概念上非常相似，两者都改变了化学物质的有效浓度。不过，不同的办法的运用其表现也不同，“重证据”的办法（穆塔兹和德金，1992年;赫兹伯格等人修改，1999年）是目前推荐的EPA的混合物毒性准则（2000年）。简言之，就互动而言，界定一种化学品在另一种上所产生效应根据预测所含的化学物质相互作用幅度（实验确定或默认值）为2293nm。危害药品（暴露量除以参考剂量或参考浓度）在混合物中各化学品乘以相互作用项。减小危害后总结得出混合物的风险指数（赫兹伯格和麦克唐内尔，2002年）。危害指数为量纲，只是提供了一个与相关的危险混合物一般估计。它有助于查明有潜在危险的混合物，但它不提供一个精确混合毒性的计算。另外，Mu和LeBlanc于2004年严格描述了该定量方法，它是基于对K-值，或K-函数的概念最早由芬尼于1942年提出。根据浓度-反应曲线，量化随着增加一种化学品的诱导浓度逐步转变。这项研究的主要目标是确定减小函数（即K-函数）是否可以用综合加法模式来增强化学互动对混合物成分的毒性影响。这项工作的第二个目的是加强我们对如何建立化学品类机理或暗匣及混合物的函数的理解。例如，有证据表明，某些化学品表现为稳定的互动模式（Durkinetal.1995年）。根据一个暗匣对另一个的化学作用我们可以看出这种稳定性增加了产生K-函数的可能。然而，同类型的互动并不意味着化学品可以表现出同样规模的互动。在目前的研究中，胡椒基丁对马拉硫磷和对硫磷都表现为抵抗作用，不过，两有机磷酸酯之间对立程度明显不同，生成的K-函数具体到每一个不同的有机磷酸脂。运用K-函数，在马拉硫磷、胡椒基丁的相互作用下，整个有机磷酸脂暗匣远远低于混合物毒性。另外，一些有机磷酸酯不需代谢活化，p450s都能去除其毒性。这些化合物可适当予以分配到有机磷酸酯暗盒与胡椒基丁混合来计算联合后有机磷酸酯的毒性，但他们需要用K-函数来表示他们的协调作用，而不是对立的。早在60年前布利斯（1939年）用IAI模式积分得到精确数据后就确定了这三个混合物概念。IAI模型为三元混合物毒性提供了合理的预测方法。该模型与加法基本模式相比是一个重大的改进。在观察和模拟结果期间出现变异可能是由于以下几个因素。不同的有机生物反应导致了固有生物变异，试验期间也可能观察到若干变异。假设K-函数运用到二元高阶化学混合物时是不完全正确的。进一步用复杂的混合物测试的IAI模型将有助于阐明基本原则和限制K-函数的相关应用。这种模式是比较易于应用，并通常可以从标准浓度-效应分析输入参数。不过，化学品之间的相互量化作用需要严格的实验要求。未来的研究可能揭示不管是否有限或实验是否定指标都能提供互动量化所需的资料。这种毒性互动是不太常见的（赫兹伯格和麦克多内尔，2002年），但仍然是混合毒性重要贡献者。IAI模型可以提高化学混合物危害和风险评估的准确率，减少在估算混合物毒性的不确定性。致谢这项研究是由环保局科学院AchieveResults提供R829358和NIEHS的培训中心提供ES07046的资金。作者衷心感谢AllenOlmstead博士和王桂荣女士为他们提供援助和咨询。附录B外文原文TOXICOLOGICALSCIENCES87(2),520-528(2005)doi:10.1093/toxsci/kfi247AdvanceAccesspublicationJuly7,2005AnIntegratedAdditionandInteractionModelforAssessingToxicityofChemicalMixturesCynthiaV.RiderandGeraldA.LeBlance1DepartmentofEnvironmentalandMolecularToxicology,NorthCarolinaStateReceivedMay2,2005;acceptedJune24,2005ABSTRACT

Thehighpropensityforsimultaneousexposuretomultipleenvironmentalchemicalsnecessitatesthedevelopmentanduseofmodelsthatprovideinsightintothetoxicityofchemicalmixtures.Inthisstudy,wedevelopedamathematicalmodelthatcombinesconceptsofconcentrationaddition,responseaddition,andtoxicokineticchemicalinteractiontoassesstoxicityofchemicalmixtures.Aternarymixtureofacetylcholinesteraseinhibitingorganophosphates(malathionandparathion)andtheP450inhibitorpiperonylbutoxidewasusedtomodeltoxicity.Concentration-responsecurvesweregeneratedforindividualchemicalsaswellasformixturesofthechemicalsusingacutetoxicitytestswithDaphniamagna.Thetoxicityofbinarycombinationsofmalathionandparathionadheredtotheprinciplesofconcentrationaddition.Thecontributionofpiperonylbutoxidetomixturetoxicitywasintegratedusingamodelforresponseaddition.Piperonylbutoxidealsomodifiedthetoxicityoftheorganophosphatesbyinhibitingtheirmetabolicactivation.Theantagonisticeffectsofpiperonylbutoxidetowardstheorganophosphateswerequantifiedascoefficientsofinteractions(K-functions)andincorporatedintothemixturemodel.Finally,toxicityoftheternarymixturewasmodeledat30differentmixtureformulationsusingthreeadditivemodelsthatassumednointeraction(concentrationaddition,responseaddition,andintegratedaddition)andusingtheintegratedadditionandinteraction(IAI)model.Toxicityofthe30mixtureswasthenexperimentallydeterminedandcomparedtomodelresults.OnlytheIAImodelaccuratelypredictedthetoxicityofthemixtures.TheIAImodelholdspromiseasameansforassessinghazardofcomplexchemicalmixtures.

KeyWords:synergy;cumulativetoxicity;predictivemodel;toxicodynamic;hazardassessment;riskassessment.

INTRODUCTION

Surveysofagriculturalandurbanstreamsandgroundwaterhavebroughtpublicattentiontowidespreadchemicalmixturecontamination(Battaglinetal.,2003;Kolpinetal.,2002).Theinfinitenumberofpotentialchemicalcombinations(intermsofbothconstituentsandconcentrationsofconstituents)limitstheutilityofstandardtoxicitytestingmethodsforestablishinghazardassociatedwithchemicalmixtures.Modelingapproachescouldaugmentthestandardtoxicitytestingparadigmwhenevaluatinghazardsassociatedwithexposuretochemicalmixtures.Chemicalconstituentsofamixturecanelicitsimilaraction,dissimilaraction,orinteraction(Bliss,1939;Casseeetal.,1998).Modelsofmixturetoxicityhavefocusedprimarilyonquantifyingthe"no-interaction"scenarios,whilecasesofinteractionoftenappearasqualitativeobservations(HertzbergandMacDonell,2002).Concentrationaddition(Loeweadditivity)andresponseaddition(Blissindependence)(Grecoetal.,1992)arecommonlyusedtomodelthetoxicityofnon-interactingchemicalswithinamixture.

Concentrationadditionmodelsrelyupontheassumptionthatmixturecomponentscontributetotoxicitythroughacommonmechanismofaction.Calculatingmixturetoxicitybaseduponconcentrationadditionrequiresassessingtherelativecontributionofeachconstituenttothetotaltoxicantpool.Thetoxicityofthispoolisthenmodeledasasingletoxicant.Concentrationadditionisthebasisofthe"toxicequivalency"approachcommonlyusedtoassesstoxicityofchemicalsofthesameclasssuchasdioxins(Safe,1990).Ampleevidencesupportstheuseoftheconcentrationadditionmodelforassessingmixturestoxicityoflike-actingchemicals(Altenburgeretal.,2000;Deneeretal.,1988;Knemann,1981).Theresponseadditionmodel,alsoreferredtoastheindependentjointactionmodel,hasbeenusedtocomputetoxicityofmixtureswhenchemicalconstituentshavedifferentmechanismsofaction(Backhausetal.,2000;Walteretal.,2002).Intheresponseadditionmodel,combinedeffectsofthechemicalsarebasedupontheprobabilitythatindividualconstituentsofthemixturewillaffecttheexposedorganisms.

Theconcentrationadditionandresponseadditionmodelsarelimitedintheirapplicationtocomplexmixturesinthattheydonotaddresschemicalinteractions.Toxicokineticinteractionscanoccurbetweenchemicalsinwhichonechemicalalterstheeffectiveconcentrationofanother(AndersenandDennison,2004).Alternatively,toxicodynamicinteractionscanoccurbetweenchemicalsinwhichonechemicalinfluencestheresponseoftheorganismtoanotherchemical(AndersenandDennison,2004).Bothtoxicokineticandtoxicodynamicinteractionscansignificantlyimpactthetoxicityofchemicalmixtures.TheimportanceofaddressingchemicalinteractionswashighlightedbytheUSEPAintheirrecommendationsforevaluatingriskassociatedwithchemicalmixtures(USEPA,2000).

Recently,Altenburgeretal.(2005)andOlmsteadandLeBlanc(2005)demonstratedthatconcentrationadditionandresponseadditionmodelscouldbeintegratedintoacomprehensivemodelforuseinevaluatingtoxicityofnon-interactingchemicalmixtures.Theintentofthepresentstudywastoexpandthisapproachtoincorporateinteractionsamongchemicalconstituentswhentheyarepredictedtooccur.Importantissuesaddressedinthisworkinclude:(1)evaluatingwhethersingleinteractionmodifierscanbeappliedtoclassesofchemicalsand(2)establishingwhetherclearlydefinedbinaryinteractionspersistinhigherordercombinations.Thestrengthoftheintegratedadditionandinteraction(IAI)modelwasassessedbycomparingmodelresultstoexperimentallydeterminedtoxicityof30differentderivationsofaternarymixture.

MATERIALSANDMETHODS

Daphnidculture.

AlltoxicologicalexperimentswereperformedwiththedaphnidDaphniamagna.Daphnidswereacquiredfromlong-standingculturesinourlaboratorythatwereoriginallyobtainedfromtheUSEnvironmentalProtectionAgency,Mid-ContinentEcologyDivision–Duluth,MN.Daphnidsweremaintainedinreconstituteddeionizedwater(192mg/lCaSO4·H2O,192mg/lNaHCO3,120mg/lMgSO4,8.0mg/lKCl,1.0μg/lseleniumand1.0μg/lvitaminB12).Culturesweremaintainedin1-literbeakersatadensityof50daphnids/lmediumandculturemediumwaschangedthreetimesperweek.Adultdaphnidswerediscardedafterthreeweeksandreplacedwithneonates.Culturebeakersandallexperimentsweremaintainedinincubatorswitha16/8-hlight/darkcycleataconstanttemperatureof20°C.Culturedaphnidswerefed2.0ml(1.4x108cells)oftheunicellulargreenalgaeSelenastrumcapricornutumand1.0ml(4mgdryweight)ofTetrafinfishfoodsuspension(PetInternational,Chesterfill,NewSouthWales,Australia).TheSelenastrumwasculturedinthelaboratoryusingBold'sbasalmedium.

Acutetoxicityassays.

Chemicalsusedinmixtureanalyses(malathion,parathion,andpiperonylbutoxide)wereacquiredfromChemServices(WestChester,PA).Absoluteethanolwasusedasthecarrierforallofthechemicals.Alltoxicityassessmentswereinitiatedwithneonatal(24hold)daphnids.Eachtreatmentconsistedoftwo50mlbeakerscontaining40mlofexposuremediumand10neonates.Selanastrum(7x106cells)andfishfoodhomogenate(0.2mgdryweight)wereprovidedtoeachbeakerasfoodatthestartofeachexposure.Allbeakers,includingcontrols,contained0.01%carrier(ethanol).Beakerswerelabeledonthebottomandrandomlyrearranged,sothattheexposureconcentrationineachbeakerwasnotknowntotheinvestigatorwhenassessingresponseoforganisms.At48h,neonateswereevaluatedforresponse.Theresponseendpoint,immobilization,wasjudgedbytheinabilityoftheneonatetooccupythewatercolumnduring10sofobservation.

Acetylcholinesteraseanalyses.

AcetylcholinesteraseactivitywasmeasuredaccordingtoEllmanetal.(1961)asmodifiedforusewithmicrotiterplates(Fisheretal.,2000)withminoradditionalmodifications.Exposuregroupsconsistedofthree250mlbeakerscontaining200mlsolutionand40neonates(24hold).Algae(1.4x107cells)andfishfood(0.4mgdryweight)wereaddedtoeachbeakeronceperday.Solutionswererenewedat24h.Followingthe48-hexposureperiod,neonatesweretransferredto1.5mlmicrofugetubes.Mediawasremovedfromtubes;neonateswererinsed,andhomogenizedin35μlicecold0.02Mphosphatebuffer,pH8.0with1%Triton-X-100usingaTeflonpestle.Anadditional315μlphosphatebuffer,pH8.0withoutTriton-X-100wasthenaddedandsamplesweremixed.Sampleswerecentrifugedat14,000xgfor4minat4°Candsupernatantwastransferredtoacleanpre-cooledmicrofugetube.Approximately100μlofthesupernatantwasstoredat–20°Cforproteinanalysis.Thefollowingsolutionswereaddedtoeachwellina96-wellplate:100μlof8mM5,5'-dithio-bis(2-nitrobenzoate)(D-1830Sigma),50μlsupernatant(phosphatebufferwith0.1%Triton-X-100wasusedforsupernatantblanks),50μlof16mMacetylthiocholineiodide(A-5751Sigma).Absorbancewasmeasuredkineticallyfor15minat420nmusingaFusionUniversalMicroplateAnalyzer(PerkinElmer,Boston,MA).ProteinwasmeasuredaccordingtoBradford(1976)usingBio-RadProteinAssaydyeconcentrate(Hercules,CA)andastandardcurvegeneratedwithbovineserumalbumin.Themolarextinctioncoefficient(13,300M–1·cm–1)(Massonetal.,2004)wasusedtocalculatetheamountofyellowanion,5-thio-2-nitrobenzoate,formedover15minandthisratewasnormalizedtotheamountofproteinaddedtotheassay(nmol/min/mg).AnalysesofvarianceandTukey-KramerHSDwereusedtodetermineifsignificant(p0.05)differencesexistedbetweentreatments.

(1)

whereRistheresponse(%immobilization),Cisthechemicalconcentration,isthepowerorslopeofthecurve,andEC50istheexposureconcentrationelicitingimmobilizationin50%ofexposedanimals.Theseindividualconcentration-responsecurvesweresubsequentlyusedinmixturemodelingasdescribedbelow.

MixtureModeling

Concentrationaddition.

AccordingtoOlmsteadandLeBlanc's(2005)integratedadditionmodel,likeactingchemicalsareassignedtoacommoncassette(i.e.,grouping).Toxicityassociatedwiththecassetteisthencalculatedusingaconcentrationadditionapproach.Accordingly,malathionandparathionwereassignedtoacommoncassette,theorganophosphate(OP)cassette.ToestablishwhetherthetoxicityofthechemicalswithintheOPcassetteconformedtoaconcentrationadditionmodel,fiveratios(Table2)ofthechemicals(malathion:parathion)wereeachtestedatsixdifferentconcentrations.Parathionconcentrationswereexpressedintermsofmalathionequivalents.AllfiveratioswereequitoxicbaseduponcharacterizationofthetoxicityoftheindividualOPs.Thesixconcentrationsofeachbinarymixtureusedintheexperimentswereselectedtodefinetheconcentration-responsecurveforthemixture.Thejointtoxicityofthesebinarymixturesoflike-actingchemicalswascomputedusingthefollowingequation(OlmsteadandLeBlanc,2005):

(2)whereRistheresponsetothemixture,Ciistheconcentrationofchemicaliinthemixture,EC50iistheconcentrationofchemicalithatcausesa50%response,and'istheaveragepowerassociatedwiththechemicalsinthecassette.Theaveragepowerwasusedbecausechemicalswithinacassetteshouldhavesimilarslopes,aswasthecasewithmalationandparathion.Concentration-responseresultsfromeachbinarymixturewerethenusedtocalculateEC50valuesasdescribedforindividualchemicals.Analysesofvariancewereperformedtodetectsignificant(p0.05)differencesamongthefiveratiosusingSAS8.2software(SASInstitute,Cary,NC).

Responseaddition.

TheconceptofresponseadditionwasusedbyOlmsteadandLeBlanc(2005)tocomputethejointtoxicityassociatedwiththedifferentchemicalcassetteswithinamixture.Theresponseadditionmodelwasusedbecauseeachcassetteisassumedtoelicitaresponsethroughdifferentmechanisms.Theresponseadditionmodelcanbedepictedas:

(3)hereRrepresentstheresponsetothemixtureandRiistheresponsetochemicalsincassettei.

Equations2and3wereintegratedtoestablishtheresponseassociatedwithindividualcassetteswithinamixtureandtosumtheresponsesassociatedwiththecassettes(OlmsteadandLeBlanc,2005).Theresultingequationisacombinationofconcentrationandresponseadditionequations:

(4)

Chemicalinteractions.

TheabilityofonechemicalinthemixturetomodifytheeffectiveconcentrationofanotherwasdefinedbycoefficientsofinteractionsorK-functions(Finney,1942;MuandLeBlanc,2004).Specifically,K-functions,definedthedegreetowhichtheconcentrationofPBOinthemixturealteredtheeffectiveconcentration(i.e.,oxonmetabolite)ofeitherorganophosphateinthemixture.K-functionsweredescribedbyexperimentallyderivingtheeffectofconcentrationsofPBOontheEC50valuesderivedforeachorganophosphate.K-functionswerecalculatedforeachofthePBOconcentrationswiththefollowingequation:

(5)whereEC50OPistheconcentrationoforganophosphatethatimmobilized50%oftheexposedanimalsandEC50OP+PBOxistheEC50oftheorganophosphatewhenexposureoccurredinthepresenceofxconcentrationofPBO.TheseK-functionswerethenplottedagainsttheconcentrationofPBOfromwhichtheywerederived.ThelogisticequationthatdefinedthisrelationshipwasusedtocalculateK-functionswhenmodelingmixturetoxicity.K-functionswereintegratedintothismodeltodescribetoxicokineticinteractionsbetweenPBOandtheorganophosphates:

(6)whereka,irepresentsafunctiondescribingtheextenttowhichchemicala(PBO)presentinthemixtureatconcentrationCaalterstheeffectiveconcentrationofchemicali(malathionorparathion).

Theresponsetothirtycombinationsofthethreechemicalswascomputedusingtheconcentrationadditionmodel(Equation2),theresponseadditionmodel(Equation3),theintegratedadditionmodel(Equation4),andtheIAImodel(Equation6).Inaddition,theactualtoxicityofthe30mixtureswasmeasuredandresultswerecomparedtothefourmodelresults.The30mixtureformulationsweredesignedsothattheratioofthethreechemicalsvariedamongthemixtureformulations.Modelpredictionswerecomparedtoexperimentaldatausingcoefficientsofdetermination(r2;Zar,1996).Anr2valueof0.70orgreaterwasconsideredagoodfitoftheobserveddatatothemodel(QualityAmerica,2004).

RESULTS

IndividualChemicalToxicityAnalyses

TheIAImodelrequirestoxicitydescriptionfortheindividualchemicalswithinamixture.Concentration-responsecurvesweregeneratedformalathion,parathion,andpiperonylbutoxide(Fig.1)fromwhichEC50valuesandcorresponding95%confidenceintervals,andpowerofthecurves()werederived(Table1).Thelogisticequationprovidedagoodfittothemalathion(r2=0.987),parathion(r2=0.987),andpiperonylbutoxide(r2=0.998)concentration-responsedata.Thetwoorganophosphatesexhibitedsimilartoxicitycharacteristics.Piperonylbutoxidewasconsiderablylesstoxicascomparedtotheorganophosphatesandhadapowerapproximatelyone-halfthatoftheorganophosphates.

CassetteAssignment

AccordingtotheIAImodel,theorganophosphateswouldbeassignedtothesamecassetteandtoxicityassociatedwiththecassettewouldbeassessedusingaconcentrationadditionapproach.Thevalidityofusingconcentrationadditiontomodelthetoxicityassociatedwiththeorganophosphatecassettewasdeterminedusingseveralcombinationsofthetwoorganophosphatesdeemedtobeequitoxicbaseduponconcentrationadditivity.Indeed,theconcentration-responseassessmentsofthesebinarymixtureswerestatisticallyindistinguishable(Table2).Therefore,thecontributionsofmalathionandparathiontothetoxicityofthefinalmixturesweremodeledasasingleorganophosphatecassette.

Thecommonmodeofactionoftheorganophosphates—theinhibitionofacetylcholinesteraseactivity—wasconfirmedexperimentally(Fig.2).Incontrast,piperonylbutoxidedidnotinhibitacetylcholinesteraseactivity.Piperonylbutoxidewas,therefore,assignedtoitsowncassettewherethetoxicityofthismixturecomponentwasintegratedintothetoxicityofthemixtureusingtheresponseadditionmodel.

ChemicalInteraction

Wehypothesizedthatpiperonylbutoxidewouldinteractwiththeconstituentsoftheorganophosphatecassetteinamannerthatwouldmodifythetoxicityassociatedwiththiscassette.Theabilityofpiperonylbutoxidetoabrogatetheacetylcholinesterase-inhibitingpotentialofeachorganophosphatewasdemonstrateddirectly(Fig.2).Theantagonisticeffectofpiperonylbutoxideonthetoxicityoftheorganophosphateswasfurtherdemonstratedbytheprogressiveshiftingoftheconcentration-responsecurvesformalathion(Fig.3A)andparathion(Fig.3B).Thismodifyingeffectofpiperonylbutoxidewasquantifiedasconcentration-dependentK-functions(Fig.4).TheseK-functionswereusedinthefinalIAImodeltomodifytheeffectiveconcentrationsofmalathionandparathionasdictatedbytheconcentrationofpiperonylbutoxideinthemixture.

MixturesToxicityAssessment

Thetoxicityof30combinationsoftheternarymixture(Table3)wasexperimentallydeterminedandcomparedtopredictedtoxicityusingtheconcentrationadditionmodel(Equation2),theresponseadditionmodel(Equation3),theintegratedadditionmodel(Equation4),andtheIAImodel(Equation6).Neithertheconcentrationaddition,responseadditionnorintegratedadditionmodelsaccuratelydescribedthetoxicityofthemixtures(r2<0.10).Rather,allmodelsgrosslyoverestimatedmixturetoxicity(Figs.5A–5C).However,theIAImodelprovidedagood(r2=0.716)assessmentofthetoxicityofthevariousmixtureformulations(Table3,Fig.5D).Toxicitywasaccuratelyestimatedwithinafactorof2for83%ofthemixtureformulations.

DISCUSSION

Theresultsofthisstudydemonstratethattoxicokineticinteractionscanbeincorporatedintoanintegratedadditionmodeltoassessmixturetoxicity.Recentstudieshaveshownthatconcentrationandresponseadditionmodelscanbeusedincombinationtocreateacomprehensiveadditivemodeltocalculatethetoxicityofnon-interactingchemicalmixtures(Altenburgeretal.,2005;OlmsteadandLeBlanc,2005;Teuschleretal.,2004).Here,webuilduponthatmodelingframeworkbyincorporatingtoxicokineticinteractionsbetweenmixtureconstituents.

Bydefinition,chemicalinteractionsrepresentadeviationfromsimpleadditivitywhenmodelingmixturetoxicity.Toquantifytheseinteractions,theexpectedadditivetoxicityofthemixturemustfirstbedetermined.Choosingtheappropriatemodeltoassessadditivityisessentialforaccurateinterpretationofinteractionresults.USEPAguidelinesforassessingmixturetoxicitysuggestadefaultmodelofconcentrationaddition(2000).Thisrecommendationisbasedonatendencytowardsmoreconservativeestimatesofmixturetoxicitywithconcentrationadditionthanwithresponseadditionmodeling(DrescherandBoedecker,1995).However,indiscriminateapplicationofconcentrationadditionlacksasoundmechanisticbasisandthereforeincreasestheuncertaintyassociatedwithpredictingmixturetoxicity.Theintegratedadditionmodeldescribedinrecentworks(Altenburgeretal.,2005;OlmsteadandLeBlanc,2005)providesamechanism-basedalternativetoassessingmixturetoxicity.Initially,chemicalswithsimilarmechanismsofactionareplacedintogroups,orcassettes.Thetoxicitywithineachcassetteismodeledwithconcentrationadditionandoveralltoxicityofthedifferentcassettesisthenmodeledwithresponseaddition(Fig.6).TheintegratedadditionmodelspresentedbyAltenburgeretal.(2005)andOlmsteadandLeBlanc(2005)areconceptuallyequivalentanddifferonlyslightlyintheirmethodsofcalculation.Theintegratedadditionmodelrepresentsasignificantadvanceinassessingtoxicityofnon-interactingchemicalmixtures.Thismodel,however,isnotequippedtomanageinteractionsamongchemicalsthatimpacttoxicityofthemixture.

Thepossibilityofsignificantsynergisticinteractionsoccurringbetweentwoormorechemicalsintheenvironmentisperhapsthemostcompellingreasontostudymixturetoxicity.Well-definedexamplesofsynergyincludeenhancedhepatotoxicityofcarbontetrachloridewithpre-exposuretokepone(KlingensmithandMehendale,1982)andinteractionsinvolvinghormonereceptorantagonistsandhormonesynthesisinhibitors(MuandLeBlanc,2004).Interactionsoftencanbepredictedbasedonmechanismsofactionofconstituentchemicals.Forexample,theP450inhibitorpiperonylbutoxideusedinthepresentstudywashypothesizedtoantagonizethetoxicityofmalathionandparathionbydecreasingtheirmetabolicactivation.However,someinteractionswillnotbeapparentfromconstituentmechanismsofaction.Theintegratedadditionmodelhasthepotentialtoidentifytheseunexpectedinteractions.Ineffect,significantdeviationofexperimentalresultsfrommodelpredictionsimpliesinteraction.Oncethesourceoftheinteractionisidentified,eitherthroughinferenceorexperimentation,quantificationandincorporationoftheinteractionintothemodelfollow.

Toxicokineticinteractionscanbeincorporatedintomixtureassessmentsviaaqualitative"weightofevidence"approachoraquantitativeapproach.Thetwoapproachesareconceptuallyquitesimilarinthatbothmodifytheeffectiveconcentrationsofchemicalsinaneffectorconcentration-dependentmanner.However,theapproachesdiffersignificantlyintheirapplication.The"weightofevidence"approach(MumtazandDurkin,1992;modifiedbyHertzbergetal.,1999)iscurrentlyrecommendedintheEPAmixturetoxicityguidelines(2000).Briefly,interactiontermsthatdefinetheeffectofonechemicaluponanotheraregeneratedbaseduponthepredictedmagnitudeofinteraction(experimentallydeterminedordefaultvalue)asafunctionoftheconcentrationsoftheinteractingchemicals.Hazardquotients(exposureleveldividedbyreferencedoseorreferenceconcentration)ofindividualchemicalsinthemixturearemultipliedbytheinteractionterm.Themodifiedhazardquotientsarethensummedtoarriveatthehazardindexofthemixture(HertzbergandMacDonell,2002).Thehazardindexisdimensionlessandsimplyprovidesageneralestimateofthehazardassociatedwiththemixture.Itisusefulforidentifyingpotentiallyhazardousmixtures,butitdoesnotprovideanaccuratecalculationofmixturetoxicity.Alternatively,astrictlyquantitativeapproachwasdescribe