Green deep eutectic solvent deproteinization of edible insect polysaccharides: Optimization using artificial neural...

The purification of polysaccharides extracted from protein-rich matrices remains challenging due to strong carbohydrate-protein interactions. In this study, a sustainable and efficient deproteinization strategy was developed using a menthol-based deep eutectic solvent (DES) for purifying Protaetia brevitarsis seulensis polysaccharide (PBSP). Systematic DES screening and single-factor optimization were followed by multivariate modeling using response surface methodology (RSM) and an artificial neural network coupled with a genetic algorithm (ANN-GA). The ANN-GA model exhibited superior predictive accuracy with lower error metrics and higher R² values than RSM. Under optimal conditions, the DES process achieved a polysaccharide retention rate (RR) of 95.30% and a deproteinization rate (DP) of 86.95%, corresponding to a comprehensive score (CS) of 91.12%. In contrast, the conventional Sevag method yielded a lower CS (68.53%) even after three extraction cycles and required approximately tenfold more solvent. Relative to PBSP, the DES-deproteinized polysaccharide showed approximately 61% higher total sugar and 75% lower protein contents, exhibiting superior purification efficiency compared with the Sevag-deproteinized polysaccharide. Structural characterization by molecular weight, FT-IR, NMR, and monosaccharide composition analyses demonstrated that DES treatment effectively removed proteins while preserving molecular integrity and the α/β-linked glucan backbone. Comparative evaluation across four edible insect polysaccharides from Oxya japonica, Tenebrio molitor, Zophobas morio, and Locusta migratoria, further confirmed the broad applicability of the developed DES process. Overall, this DES-based method provides a green, solvent-efficient, and scalable alternative for polysaccharide deproteinization, with strong potential for application to other insect-derived and protein-bound polysaccharides.