Multi-mode S-type ultrasound-assisted protein extraction from walnut dregs and in situ real-time process monitoring.

This study aimed to investigate the impact of multi-mode S-type ultrasound treatment on the protein extraction level of walnut dregs. The structural properties of the walnut protein (WP) were characterized, and the correlation between protein structure and extraction level was analyzed. The in situ real-time monitoring model for the ultrasound-assisted WP extraction process was established by a miniature fiber near-infrared (NIR) spectrometer. Results showed that the protein yield, purity, and comprehensive extraction index (CEI) of extracted WP were 71.07 %, 72.69 %, and 71.72, respectively, under optimal conditions (dual-frequency 20/28 kHz, ultrasonic treatment duration 30 min, and ultrasound power density 120 W/L). The secondary structure of extracted WP displayed that the proportion of α-helix and β-sheet reduced, while the contents of β-turn and random coil increased after ultrasonic treatment. Besides, sonication decreased the disulfide bond content and increased free sulfhydryl (-SH) and surface hydrophobicity compared to the control. The microstructures of WP confirmed that appropriate sonication could unfold the protein aggregates and reduce the particle size. The extraction level of WP is positively correlated with the -SH content (p < 0.01). The quantitative prediction model of Si-PLS for -SH content in the ultrasound-assisted WP extraction process was established and performed a good correction and prediction performance (Rc = 0.9736; RMSECV = 0.446 μmol/L; Rp = 0.9342; RMSEP = 0.807 μmol/L). This study exploited a high-efficiency way for the WP extraction industry, and provided theoretical support for the development of the intelligent system in industrial protein extraction process.