Development Of A New Method For Evaluating Quality Grade Of Tea Based On Three-way Chromatographic Fingerprints

Tea is a special and preponderant agricultural product in China.According to the fermentation degree,tea can be classified as:green tea,white tea,wulong tea,yellow tea,dark tea and black tea.Among them,green tea is one of the most popular tea with the largest production,consumption and export volume.In the tea market,the price of tea is closely related to its quality grade.There are many phenomena of pass low-grade tea off as high-grade tea to gain profit in the tea market,which not only disrupts the transaction order of tea market,but also infringes on the interests of consumers.At present,sensory evaluation is the most important method to evaluate the quality grade of green tea.It is essential to develop an objective,steady and accurate method for assessing the grades of green tea.In this study,we acquired the three-way chromatographic fingerprints and metabolic fingerprint of green tea by high performance liquid chromatography-diode array detector(HPLC-DAD)and ultra-performance liquid chromatography/quadrupole time-of-flight mass spectrometry(UPLC-QTOF-MS).Based on the collected data,the discriminant model of green tea quality grade was established.The study provided basic data and method guidance for quality control of green tea.The main research conclusions are as follows:1.The three-way chromatographic fingerprints of Laoshan green tea in different picking seasons were collected by HPLC-DAD.After data preprocessing,a total of 55valid variables were obtained by using multivariate curve resolved-alternate least squares(MCR-ALS)analysis.Based on the relative concentration data of 55 variables,principal component analysis(PCA),partial least squares discriminant analysis(PLS-DA)and support vector machine(SVM)were used to distinguish between summer and autumn green tea from Laoshan.The scores plot of PCA showed the clustering trend by picking season with partial overlapping.The recognition rates of calibration model,cross validation model and external validation model in PLS-DA and SVM analysis were identical,which were 100%,91.3%and 100%,respectively.Based on PLS-DA model,three variable selection methods[(Regression coefficients,RC),(Variable importance in projection,VIP),(Selectivity ratio,SR)]were used to select the characteristic variables that had important contributions to the classification of summer tea and autumn tea.The new model(VIP-PLS-DA)established by those characteristic variables selected based on VIP method has the same predictive ability as the original PLS-DA model.Gallocatechin,proanthocyanidin B1,myricetin and vitexin glycoside were identified as characteristic markers.2.The three-way chromatographic fingerprints of different grades green tea were measured by HPLC-DAD with two extraction conditions(water extraction,70%methanol extraction)and resolved by MCR-ALS.Then,the discriminant models of different grades green tea were established based on the relative concentrations of these resolved components and chemical pattern recognition methods(PCA,PLS-DA,SVM).For PCA model,the variance interpretation rates of the first two principal components with methanol extraction were significantly higher than those with water extraction.For PLS-DA model,the total recognition rates of training set with two extraction methods was both 87.5%,and the total recognition rate of the cross-validation model with the water extraction(62.5%)was lower than that with the methanol extraction(71.4%),and the total recognition rate of the testing set with the water extraction was(75%)higher than that with methanol extraction(50%).For SVM model,the best classification accuracy rate of methanol extraction(83.9%)was higher than the accuracy rate of water extraction(78.6%).These data indicated that the model established with methanol extraction was better than that with water extraction,and it was more suitable to evaluate the quality grade of green tea.3.Non-targeted UPLC-QTOF-MS combination with multivariate statistical analysis methods were used to establish models for distinguishing between four different grades of green tea.PCA,PLS-DA,OPLS-DA can effectively distinguish between four different grades tea samples.PCA analysis results:R~2X=0.718,R~2Y=0.627(ESI+);R~2X=0.664,R~2Y=0.612(ESI-);PLS-DA analysis results:R~2X=0.567,R~2Y=0.718,Q~2=0.627(ESI+);R~2X=0.625,R~2Y=0.664,Q~2=0.612(ESI-);18different compounds were selected as differential metabolites by the VIP value in OPLS-DA models and t-test(VIP>1,P<0.05).Finally,VIP-OPLS-DA discriminant models in positive and negative ion modes were established by 18 identified variables to verify the validity of these differential variables.