Optimization of tetrastigma hemsleyanum extraction process based on GA-BPNN model and analysis of its antioxidant effect

Abstract

Tetrastigma hemsleyanum (Tetrastigma hemsleyanum Diels et Gilg) is a valuable traditional Chinese medicine with various applications. In this study, we aimed to optimize the extraction process for the total extraction yield of five flavonoid components, namely kaempferol, quercetin, rutin, kaempferol-3-O-rutinoside, and astragalin from the Tetrastigma hemsleyanum root (THR), and explore its potential molecular mechanisms in treating oxidative diseases as well as antioxidant activity. To achieve these objectives, we employed the genetic algorithm-back propagation neural network (GA-BPNN), the Box-Behnken design (BBD) with 4-factors and 3-levels to establish the optimal ethanol extraction process for the total extraction yield of the 5 components. Using public databases, the “component core targets-disease core target genes” networks were built, as well as molecular docking. Furthermore, DPPH was used to examine the antioxidant activity of the extracts obtained from THR under the optimal extraction process. The experimental value of the total extraction yield of the 5 components achieved a maximum of 788.12 mg/kg when the ethanol concentration was 73%, the solid-liquid ratio was 26 g/mL, and the ultrasonic duration was 30 min, and the ultrasonic temperature was 76 °C. When docked with protein molecules such as 6Y8I, quercetin, and other components received moderate to high scores. When the total concentration of the 5 components was 3.033 μg/mL, the DPPH radical scavenging rate was 89.81%. Compared with the BBD method, the GA-BPNN method is more efficient and reliable for optimizing the extraction process of active ingredients in THR because of its good data-fitting ability.