Abstract:

An enzymatic membrane reactor designed based on the combination of membrane filtration and Protease N Amanocatalyzed hydrolysis was used to prepare egg-white hydrolysate (EWH) with antithrombin activity. A L18(37×21) mixed-level orthogonal array design was used to investigate the effects of substrate concentration, enzyme dosage, pH, temperature, filtration flow rate and length of hydrolysis on the antithrombin activity of EWH. This was followed by the stimulation and predication of EW hydrolysis using back propagation (BP) neural networks to obtain the optimal values of the above parameters. The results showed that the predicted value of IC50 of EWH obtained under the following optimized conditions: substrate concentration 1%, enzyme dosage 1%, filtration flow rate 10 mL/min, and length of hydrolysis 4 h was 10.43 mg/mL, 4.03% lower then the actual value. This demonstrates good reliability of BP neural networks in optimizing egg white protein hydrolysis.

Key words: enzymatic membrane bioreactor, egg white, back propagation (BP) neural network, antithrombin activity, optimization