Chemometrics-enhanced high performance liquid chromatography-ultraviolet detection of bioactive metabolites from phytochemically unknown plants.

Abstract This work describes the use of Colubrina greggii as a model to investigate the use of chemometric analysis combined with data from a leishmanicidal bioassay, using Principal Component Analysis (PCA) and Orthogonal Projections to Latent Structures (O-PLS), to detect biologically active natural products in crude extracts from plants having little or no phytochemical information. A first analysis of the HPLC-UV profiles of the extract and its semi-purified fractions using both Principal Component Analysis (PCA) and Orthogonal Partial Least Squares (O-PLS) indicated that the components at t R 48.2, 48.7, 51.8 min correlated with the variation in bioactivity. However, a further O-PLS analysis of the HPLC-UV profiles of fractions obtained through a final semi-preparative HPLC purification showed two components at t R 48.7 and 49.5 min which correlated with the variation of the bioactivity in a high performance predictive model, with high determination coefficient, high correlation coefficient values ( R 2 and Q 2  = 0.99) and a low root mean square error (RMSE = 0.018). This study demonstrates that the association of chemometric analysis with bioassay results can be an excellent strategy for the detection and isolation of bioactive metabolites from phytochemically unknown plant crude extracts.