Anticancer component identification from the extract of Dysosma versipellis and Glycyrrhiza uralensis based on support vector regression and mean impact value†
Abstract
The good therapeutic effect of herbal medicines depends on their abundant components and it's extremely necessary to find out the main bioactive ingredients. In this paper, the extract of Dysosma versipellis and Glycyrrhiza uralensis was studied for the first time by chemometrics. A HPLC-UV method was developed and validated to establish fingerprint spectra of 46 batches of different samples and a total of 45 common components of all samples were quantitatively and qualitatively analyzed using HPLC-UV and UPLC-Q-TOF-MS/MS, respectively. The anticancer effect of the extract was obtained by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay on HeLa cells. After that, a support vector regression (SVR) model optimized by particle swarm optimization (PSO) was constructed to depict the relationship between the chemical constituents and anticancer effect of the extract. Then the mean impact value (MIV) method was introduced to evaluate the bioactivity of the concerned components based on the optimal SVR–PSO model. The results showed that the developed model has an excellent fitting accuracy and generalization ability, and a ranking of the components for their anticancer activity was obtained. The employed strategy provides an efficient and convenient access to active anticancer constituents from the extract of D. versipellis and G. uralensis. The identified components provide explicit guidance for screening anticancer compounds and the developed model can be used for predicting the activity of new samples.