The Active Components Identification And Compatibility Research Of Curcuminoids In Curcuma Longa L

Traditionally, active compounds were discovered from natural product extractsby bioassay-guided fractionation, which was with high cost and low efficiency. Awell-trained Support Vector Regression (SVR) model based on Mean Impact Value(MIV) analysis was used to identify lead active compounds prior to the separationprocess. The use of the proposed approach could provide an efficient and economicalapproach for drug discovery from natural products. The main research works are asfollows:1. Components-activity relationship of curcuminoids: The tube width ε, thepenalty parameter C and the kernel function parameter g were simultaneously selectedfollowing leave-one-out cross validation by Particle Swarm Optimization (PSO).When ε was0.0100, C was1.0309and g was0.1000; the mean square error for crossvalidation (CVMSE) was the smallest. We used the RPAs of twenty-sixchromatographic peaks as input variables and inhibition ratios on HeLa cells as outputvariables. The results revealed that the ε-SVR-RBKF-PSO model of all the sets hadthe best model performance with a high correlation coefficient (Q=0.9812) and a lowmean square error (MSE=0.0013) between the experimental and predicted values.2. Active components identification of curcuminoids: The two new training setswere obtained when every independent variable increased and decreased10percentrespectively, which were used for simulation according to the fitting model. The meanof the difference values of the two simulation results was calculated by the number ofsamples, namely MIV. Finally, the sequence of the independent variables was sortedaccording to their absolute MIVs. Eight constituents possessing the high absoluteMIV were identified to have significant cytotoxicity. The IC50values of curcumin,demethoxycurcumin and bisdemethoxycurcumin were9.93±0.41,7.32±0.45and13.072.68μg/ml on MTT assays, respectively. However, the result was differentfrom that obtained from MIV methods above, namely, the cytotoxicity ofdemethoxycurcumin was below that of curcumin. We assumed that the antagonismbetween some certain component or components in curcuminoids anddemethoxycurcumin resulted in the difference. 3. The research of compatibility and interaction of curcuminoids: ResponseSurface Method (RSM) was applied to optimize the compatibility of three activecomponents operating on Design Expert software. The optimum mixture ratio wasconfirmed as follows:18.91:9.97:5.34and the value of IC50was5.00±1.26μg/ml,which indicated that the bioactivity increased based on the compatibility. We foundthat curcumin and demethoxycurcumin largely increased cytotoxicity.Bisdemethoxycurcumin slightly increased cytotoxicity, but it is absolutely necessaryin the study of interaction of curcuminoids.