北京工业大学-亚硝基脲类抗癌药物的设计方法研究

DFTStudiesontheQuantitativeStructure-activityRelationshipofN-(2-chloroethyl)-N'-cyclohexyl-N-nitrosoureasasAnticancerAgentsCAOJun,JINShu-Bin,HANBing,LIUWei,ZHAOLi-Jiao,ZHONGRu-Gang（CollegeofLifeSciencesandBioengineering,BeijingUniversityofTechnology,Beijing100022,P.R.China）Abstract:N-(2-chloroethyl)-N'-cyclohexyl-N-nitrosoureas(CCNUs)arewidelyappliedasantitumoragents.However,researcheshaveindicatedthattheirdecompositionproductsareinclinedtocausesecondarycancer.Henceitisnecessarytoprobeintobothcarcinogenicandanticancermechanismofsuchcompoundssothateffectiveanticancerdrugswithlowertoxicitycanbeobtained.Inthepresentwork,theDFT-B3LYPmethodwiththebasissetof6-31+G(d,p)wasemployedtocalculatethemoleculargeometriesandelectronicstructuresof17CCNUs.ThechargesofC6and(miLogP)2wereselectedasquantumchemicaldescriptors.Theobservedanticanceractivityalongwiththetwodescriptorswasusedtoestablishthequantitativestructure-activityrelationship(QSARs)throughmultiplelinearregressionanalysis.ThisQSARmodelcanbeappliedtopredicttheanticanceractivityofCCNUs.Keywords:CCNUs,anticancer,DFT,QSAR0.INTRODUCTIONChloroethylnitrosoureas(CENUs)aresignificantinthetreatmentofhumanmalignancies,includingHodgkin’sdisease,braintumorsandlymphomas[1-3].CCNUsbelongtoanimportantfamilyofCENUs,whichundergodecompositionreactionsinaqueoussolutionandgeneratereactiveintermediatesknowntomodifyDNAandproteins[4,5].TheanticanceractivityofCCNUsisassociatedtheformationofDNAinterstrandcrosslinks(ICLs),whichisinitiatedbythedamagetotheO6-positionofguanine[6-8].ConsideringthealkylationandICLsofDNAresultedfromthedecompositionofCCNUs,muchattentionhasbeenpaidtothetwodecompositionpathwaysandtheproductsinordertoelucidatetheconnectionoftheirchemicalstructureswithanticanceractive.Tongandcoworkersputforwardanotherdecompositionpathway,asillustratedinpathway,whichproceededviaasupposedintermediate4,5-dihydro-1,2,3-oxadiazole[9].Insecondpathway,ethyleneglycolisfinallyyieldedbythedegradationofoxadiazole[10].Lownetal.[3]proposedthispathway,whichinvolvedthecyclizationofCCNUstoa2-(alkylimino)-3-nitroso-2-oxazolidine.Thisintermediatyieldethylaldehyde,whichalsohasthecapabilityofalkylatingDNA.Quantitativestructure-activityrelationship(QSARs)appliedelucidatestheunderlyingmodesofactionofchemicalsandtheircorrelationtothecharactersofthemolecularstructure[11].Inthiswork,thedensityfunctionaltheory(DFT)methodwasappliedtoderivequantumchemicaldescriptorsfortheQSARstudyof17CCNUs.ItwasexpectedthatthisworkcouldshedlightontheunderstandingoftheanticancermechanismofCCNUs,aswellastheanticancerdrugdesignofCCNUswithhighereffectandlowertoxicity.1.MODELSANDCOMPUTATIONSThebiologicaldatausedfortheQSARanalysisinthisstudyarethelog1/Cofthesecompounds.Cisthemolarconcentration(mol/kg)ofCCNUsproducingareductionof1000-foldinL-1210leukemiainmice(3logkill).ThestructuresofalltheCCNUswerebuiltwithGaussViewprogram.FullgeometricoptimizationofallstructureswereperformedbytheDFTmethodwiththehybridfunctionalB3LYP(Becke’sthree-parameterfunctionalemployingtheLee,YangandParrcorrelationfunctional)at6-31+G(d,p)theoreticallevel.Inpreviousresearches,DFT-B3LYPmethodwasdemonstratedtobeappropriatetoinvestigatedinstructure-propertyrelationshipofcarcinogensordrugs.Theanalyticalcomputationsofthevibrationalfrequencieswereperformedforallstationarypointstoverifythattheywereenergyminimumwithallpositivefrequency.BecausealltheinvivoprocessinvolvingthebiomoleculardamagebyCCNUstakeplaceinaqueoussolution,thesolventeffectofwateronallcompoundswastakenintoaccount.Allcomputationswerecarriedoutbyemployingtheself-consistentreactionfield(SCRF)methodusingtheconductor-likepolarizablecontinuummodel(CPCM).AllcalculationsfortheacquisitionofquantumchemicalparameterswereperformedwithGAUSSIAN03programpackageonanIntelSR870BH2server.Multiplelinearregressions(MLR)analysiswasusedtoderivethestructuraldescriptors,bywhichthelinearrelationshipbetweenthebioactivityindicesandthecomputationalparameterscouldbedeterminedreasonably.Thecorrelationbetweenthevariablesinthemodelwasestimatedbythecorrelationcoefficientr,whichdescribedthecorrelationbetweenthestructuralparameters.Whenrwasbelow0.8,itwasconsideredthattherewasn’tobviouscorrelationbetweenthemoleculardescriptorssothattheregressionmodelwasstableandcredible.AfterMLRanalysis,thedescriptors,whichhadhighercorrelationwiththedependentvariable,wereretainedandtheotherswereremovedfromthedescriptordataset.ThecomputationaldatasetwasanalysedbyusingthestatisticsprogrampackageSPSS(StatisticsPackageforSocialScience,13.0version).2.RESULTSANDDISCUSSIONThechemicalstructuresofthe17compoundsandtheirbiologicalactivityarelistedinTableI.TheobserveddataofLog1/Cwereobtainedfromtheliteraturesource18.Atotalof23typesofmoleculardescriptorswerecalculatedfortheQSARanalysis.TheelectronicdescriptorsoflocalchargesonO1,C2,N3,N4,O5,C6,C7,Cl8,N9andC10werecalculatedaccordingtoMullikenpopulationanalysis(MPA).Thehighestoccupiedmolecularorbitalenergies(EHOMO),thelowestunoccupiedmolecularorbitalenergies(ELUMO)andtheenergeticgapbetweenLUMOandHOMO(ΔE)wereobtainedbycompu