黄酮类醛糖还原酶抑制剂的定量构效关系研究.pdfVIP

CODEN JYYHE6／ISSN 1001—4160 ComputersandAppliedChemistry 2009，Vo1．26，No．12，1579—1582 QSARstudiesonflavonoidsasaldosereductaseinhibitors XuJie，HouJing，HeLin，LiuM ingandLuoHaibin (SchoolofPharmaceuticalSciences，SunYat—SenUniversity，Guangzhou，510060，Guangdong，China) Abstract：Aldosereductase(AR1isallenzymeincarbohydratemetabolismthatconvertsglucosetosorbitolIthaslongbeenbelievedtoberesponsiblefor diabeticcomplicationsinvolvinganumberoforgansFlavonoidsandtheirderivativesindicateinhibitoryactivitiesagainstAR Inthepresentwork．a3D OSAR r3Dquantitativestructure．activityrelationship)studyon76structurallyrelatedflavonoidsasARinhibitorswasperformed．AreasonableCoMSIA(comparative molecularsimilarityanalysis)mode1withthecombinedsteric，electrostatic，hydrophobic，andH—bonddonorfieldsachievedthehighestcross—validated regressioncoemcient( =0666andconventionalregressioncoefficientrz：0．918)，inwhichtheelectrostaticandH—bondinteractionsareidentifiedasthe underlyingfactorfortheinhibitoryactivitiesbetweenAR andthoseflavonoidsmodulatednon．triviallyby stericandhydrophobiclnteractionsTheaccuracy andpredictivepowerofthederivedCoMSIAmodelwereverifledbyleave—nine—out，Y-randomization，andtest，setvalidations．Resultsofthestudycanaidthe designofhovelAR inhibitorswithhighamnitv． Keywords：MLR(MultipleLinearRegression)，NLMR(NonlinearMultivariableRegression)，ANN(artificialneuralnewtork)，alkylbenzenes，boilingpoint， molarvolume inhibitorysutdiesagainstvariousenzymes．Theaccuracy and ln1tr0QUCtlon predictivepowerofthederived3D—OSARmodelwereverified Diabeticpatientssufferfrom diabetescontro1andcomplica— by leave—nine—outy-random ization．andtest—setvalidations．In tionssuchasneuropathy，nephropathy，retinopathy，a