Extraction And Separation Of Flavonoids And Quantitative Structure-Activity Relationship Research

As a classical type important natural organic compounds produced during plants metabolism, flavonoid is of great significance to cure, treat and prevent tumour, senescence and cardiovascular diseases and has fairly good market prospect at present. Flavonoid has still been the potential source in the search of lead compounds and biologically active components and has also been the emphasis of the research, development and ultilization at home and aboard in the recent 30 years. In this thesis, flavonoids extracted from Honeysuckle and Chrysanthemum were firstly extracted by conventional extraction and such newly emerging methods as microwave-assisted extraction(MAE) and ultrasonic-assisted extraction. Through comparision of different methods and experimental design as well as analysis of experimental data, a optimal condition for extraction was obtained. Then the flavonoid extracts were further separated and analyzed. Apart from the above-mentioned work, the chemical shift(CS) of carbon-13 nuclear magnetic resonance(13C NMR) of flavonoid molecules was estimated and predicted through modeling of quantitative structure-spectrum relationship (QSSR) by atomic electro-negativity interaction vector(AEIV). The main contents and conclusions are provided as follows: 1) Honeysuckle flavonoids and Chrysanthemum flavonoids were extracted by different extraction ways, among which microwave-assisted extraction(MAE) and ultrasonic-assisted extraction were especially focused. Several experimental designs, such as Homogeous / uniform design, the orthogonal experimental design and central composite design, together with Statistical analysis and intuitionistic analysis were performed to get the upmost extraction efficiency and optimal condition. It is indicated after comparision that MAE and ultrasonic-assisted extraction were better than conventional extraction, for solvent was economized and extraction time was greatly shorten, thus resource and energy were saved remarkably. 2) Macro-reticular resin was employed to refine flavonoids extracted from Honeysuckle and Chrysanthemum. The purificatory extractions were distilled under low pressure to gain a concentrated solution, which was further dried into yellow powder under vacuum condition. Ultraviolet Absorption Spectrum was adopted for determination of such a purificatory extractions. The obtained spectral bands accord with characteristic spectral bands of flavonoids. The extractions were also analyzed by capillary electrophoresis. By comparing between crude extraction and the one after purification , it's found that impurity obviously decreased after purification. 3) Atomic electro-negativity interaction vector(AEIV) and atomic state index(ASI) are used for expression of chemical environment and atomic state of 255 carbon atoms in seventeen dihydroflavone. Using multiple linear regression (MLR) technique, the quantitative relationship model is established for AEIV and ASI and 13C-NMR chemical shifts of these atoms .The correlation coefficients (R) of model is 0.9816. To validate the prediction capabilities of the model for the external prediction set, the leave-one-out (LOO) cross-validation (CV) technique was applied, and the correlation coefficient RCV is 0.9806. The results show that AEIV and ASI can be used to express chemical environment of molecular and atomic state, the created models have good estimation stability and favorable predictive ability. 4) Based on three-dimensional structure of molecular, a novel 3D Holographic Vector of Atomic Interaction Field (3D-HoVAIF) was obtained through expression of molecule structure, which considering three kinds of non-bond interaction related with biologic activity, i.e. electrostatic, steric and hydrophobic interaction. 3D-HoVAIF was applied to study Quantitative Structure-Activity Relationship of inhibition of flavonoids on xanthine-oxidase. A model with 7 variables has been created by using multiple linear regression. The correlation coefficients (R) of model is 0.980. The correlation coefficient RCV is 0.939 through the leave-one-out (LOO) cross-validation (CV) technique. The results show that 3D-HoVAIF can be used to well express the structure of these compounds and the model has satisfactory estimation stability and favorable predictive ability.