Optimization of ultrasound-assisted extraction of bitter compounds and polyphenols from willow gentian underground parts

Abstract

• Underground parts of Gentiana asclepiadea as a source of bioactive compounds. • Optimization and modeling of the extraction process by RSM and ANN approach. • Extraction of gentiopicroside, isoorientin, isovitexin, isogentisin and polyphenols. • The experimentally obtained values were in close agreement with the predicted. • The ANN model showed superior predictive capacity compared to the RSM model. Willow gentian ( Gentiana asclepiadea L., Gentianaceae) is a medicinal plant that is primarily used in the treatment of gastrointestinal and liver diseases in folk medicine. As a bitter raw material, it is also used in the food industry for liquors flavoring. In order to optimize the ultrasound-assisted extraction (UAE) of gentiopicroside, isovitexin, isoorientin, and isogentisin, as well as total polyphenols from the underground parts of willow gentian, the influence of process parameters was investigated. Extraction time (from 5 to 65 min), ethanol concentration (from 10 to 90% v/v), solid to liquid ratio (from 1:10 to 1:50 w/v), and extraction temperature (from 20 to 80 °C) were estimated at five levels as independent variables. The response surface methodology (RSM) and the artificial neural network (ANN) approach were employed to develop a predictive extraction model and to establish optimal extraction conditions to simultaneously maximize the extraction yield of all target compounds. Established optimal values of process parameters were: extraction time 50 min, ethanol concentration 53% (v/v), solid to liquid ratio 1:40 (w/v), and extraction temperature 65 °C. The obtained extraction yields under predicted optimal conditions for gentiopicroside (49.23 mg/g DW), isoorientin (1.00 mg/g DW), isovitexin (4.04 mg/g DW), isogentisin (1.14 mg/g DW), and TPC (10.03 mg of GAE/g DW) confirmed the adequacy of both RSM and ANN models. Compared with the RSM model, the ANN model fitted better with the experimental values of each compound (higher coefficient of determination and lower mean absolute deviation) indicating its higher predictive capacity. The developed high-accuracy extraction model provides the basis for further separation and purification, as well as for studying the bioactivity of selected compounds from underground parts of G. asclepiadea .