Mixture designs for exploring class diversity and metabolite fingerprinting: an efficient column chromatographic strategy.

The effects of four solvents, hexane, dichloromethane, ethyl acetate, methanol, and their mixtures on the separation of metabolites in crude extracts of Erythrina speciosa Andrews leaves were investigated using two strategies for open column chromatography. The classical extraction procedure was compared with mobile phases prepared according to a mixture design in order to explore the effects of solvent interactions on metabolite separations. Principal component analysis was used to compare the UV spectra obtained from RP-HPLC-DAD and to estimate the number of independent factors contained in the chromatographic data of the extracts. The results showed that, in addition to solvent polarity, solvent mixtures play an important role in metabolite separation. When pure solvents are used, larger groups of similar spectra are observed in the factor analysis score graphs indicating the same or a limited number of metabolite classes. In contrast solvent mixtures produced score graphs with a larger number of clusters indicating greater metabolic diversity. Besides resulting in more peaks than the pure solvents the chromatographic data of the design mixtures resulted in larger numbers of significant principal components confirming the greater chemical diversity of their extracts. Thus, if the objective of an analysis is to obtain metabolites of the same class, one should use pure solvents. On the other hand, binary and ternary solvent mixtures are recommended for more efficient investigations of class diversity and richer metabolite fingerprints.