Screening bioactive constituents from herbal medicines or traditional Chinese medicines
according to traditional phytochemical approach is time-consuming and laborious. The objective of
the present study is to provide a strategy for rapid screening of active candidates from plant extract
based on the Quantitative Patteren-Activity Relationship. The chromatographic fingerprint (Patteren) data sets were obtained
by our previous proposed method and the antioxidant activity was measured by Ferric Reduction Ability Power assay.
The correlation model was built by partial least squares and bioactive components were screened out by target projection.
Variable selection and model evaluation were performed by 4-fold cross validation. The optimal model required 5
PLS components. The calibration and prediction multiple correlation coefficients were 0.990 and 0.975, respectively. And
the calibration and prediction errors were 23.2 and 36.4, respectively. The projection result showed that the antioxidant
activity was highly correlated to some variables/peaks, indicating that they are main antioxidants. 3'-Hydroxypuerarin, puerarin,
3'-methoxypuerarin and 6''-O-D-xylosylpuerarin were identified to be four main antioxidants by comparing their
TOF MS patterns and retention times. Moreover, the model was found to predict accurately the antioxidant capacity of the
Radix puerariae samples through their chromatographic fingerprint only. The results obtained in this work show clearly
how simple and effective our proposed method is in bioactive constituent screening as well as drug lead compound discovery
from herbal medicines.
Keywords: Quantitative patteren-activity relationship, bioactive constituent screening, antioxidant activity, target projection,
radix puerariae isoflavones.
Rights & PermissionsPrintExport