Gas Chromatographic Fingerprint Analysis of Secondary Metabolites of Salvia Reuterana Combined with Antioxidant Activity Modelling Using Multivariate Chemometric Methods

Chromatographic fingerprinting is widely used for classification and authentication of several samples such as plant extracts. Gas Chromatography (GC) is a common instrument for fingerprinting of natural samples. On the other hand, plants are noticeably used as important sources of antioxidant compounds. Due to the various biological activities such as antioxidant activity, Salvia Reuterana has been used in Iranian traditional medicine among the 58 Salvia types [1]. In addition, due to the complexity of natural sample matrices and lack of selectivity in analytical instruments, multivariate chemometric methods have been largely used to extract maximum useful information from chromatographic fingerprints [2]. In the present work, a chemometrics-based strategy is proposed for GC fingerprints analysis of Salvia and modeling their antioxidant activity. On this matter, ultrasonic-assisted extraction-dispersive liquid-liquid microextraction (UAE-DLLME) was used for extraction of chemical components of twenty eight salvia samples from eight populations. The optimum extraction conditions were obtained using factorial based response surface methodology (RSM). The optimum extraction parameters were 60 mg of powdered aerial parts of dried Salvia sample, 2.5 mL of methanol as first extraction solvent, 45 min first sonication time, 40°C extraction temperature, 30 μL of tetrachloroethylene as preconcentration solvent, 2 min second sonication time and NaCl 7% (w/v) [3]. The GC profiles were arranged in a data matrix and this data matrix was mean-centered, baseline corrected and aligned using correlation optimized warping (COW) method before cluster analysis. The data matrix was then analyzed using principal component analysis (PCA), hierarchical cluster analysis (HCA) and k-nearest neighbors (kNN) clustering methods to explore similarities and dissimilarities among different Salvia samples according to their secondary metabolites. Evaluation of the PCA score plot (explaining 85.8% of variance accounted for three PCs) showed that Salvia samples belong to six clusters. In order to specify borderlines between classes, the degree of class separation (DCS) was calculated.These results were then confirmed by HCA and kNN and six clear-cut clusters were determined using HCA and kNN dendrograms. For antioxidant activity measurement, 30 μL of the extracted Salvia sample and 20 μL methanol (99.9%), were added to 5 mL of 1,1-diphenyl-2-picryl-hydrazyl (DPPH) (50 μM). After 30 minutes at room temperature, absorbance was measured at 517 nm [4]. The values of inhibitory concentration (IC50) were then arranged in a data vector. The profiles of preprocessed chromatograms were auto-scaled and modeled by partial least squares-regression (PLS-R) to correlate Salvia clusters to their antioxidant activity. The optimum number of LVs (respectively explaining 74.32% and 97.5% of variance accounted for six LVs for X- and y-blocks) in PLS model was determined using error rate (ER) values obtained by leave-one-out cross-validation. The classification figures of merit including calibration regression coefficient and RMSECV were respectively 0.9779 and 0.01674. The PLS score plot represented 6 different classes that confirmed the clusterig results obtained by PCA score plot, HCA and kNN. Finally, variables were decreased using variable importance in projection (VIP) method and the most important components which affect the antioxidant activity were detected. It is concluded that the poroposed strategy in this work can be used for classification of chromatographic fingerprints of natural samples and modelling of their antioxidant activity according to their class separation.