Optimization of Protein Extraction from Rapeseed Oil Cake by Dephenolization Process for Scale-Up Application Using Artificial Neural Networks.

Abstract

Rapeseed proteins, due to their quality and wide availability, have great potential for application in human nutrition. However, their high content of antinutritional compounds poses significant economic and environmental challenges for food industry applications. To overcome these obstacles, various extraction and modification techniques, including enzymatic and ultrasound-assisted methods, were used to enhance protein functionality and improve both nutritional and sensory properties. In this study, the effects of dephenolization on the structural, physicochemical, and functional properties of rapeseed protein isolate obtained from defatted rapeseed cake were investigated through four different protocols. All obtained protein isolates (PIs) exhibited high protein purity (above 65%), with a notable difference in extraction yield. Furthermore, the extraction process was optimized using an artificial neural network (ANN) model, which demonstrated high predictive performance. The optimal extraction conditions for the dephenolization of rapeseed oil cake were 84% ethanol concentration, a solid-to-liquid ratio of 1/60 w/v, and 15 min of ultrasound treatment, resulting in an impressive protein purity of 90.68% with a yield of 29.17%. The obtained proteins were further characterized and compared in terms of protein profile (FTIR and SDS-PAGE), amino acid composition, solubility, and digestibility. The protein isolate (PI) obtained under optimized conditions displayed superior functional properties, underscoring the relevance and necessity of a data-driven, mathematical approach for scale-up and industrial implementation.