A robust feeding control strategy adjusted and optimized by a neural network for enhancing of alpha 1-antitrypsin production in Pichia pastoris.

Highlights • Novel on-line μ-stat approach is presented for regulating methanol feeding rate in fermenter. • Methanol feeding was controlled in fed-batch based on the on-line ammonia consumption rate. • MLP3 neural network was used to reconstruct the controller. • The designed controller was used for A1AT production process control in P. pastoris. • Control and maintaining μ at the optimal level led to increase in target protein production. Abstract Exact monitoring of the key process variables, such as specific growth rate (μ), is essential for effective process control, and maintenance of the optimal conditions for the production of recombinant proteins. In this study, a novel and confident on-line μ-stat approach for setting up methanol feeding strategy is described that is based on the consumption of ammonium hydroxide as feedback control. For this purpose a new and robust control with a time-based stable control structure was used, which had the ability to reconstruct the controller. Moreover, the MLP3 neural network was applied to adjust and optimize the performance of the robust control system. To evaluate the cited strategy, the overproduction of human recombinant alpha 1-antitrypsin (A1AT) under the control of alcohol oxidase 1 promoter (p AOX1) was investigated. On-line control of the μ at the optimal amount (0.03 1/h) led to 120 g/L dry cell weight and 324 mg/L of A1AT concentration. In comparison with traditional protocols of methanol feeding, the obtained product concentration demonstrated a significant improvement. It is suggested that this new presented method has a remarkable potential to be used for the overexpression of different recombinant proteins in the host Pichia pastoris as well as for process development in other hosts. [ABSTRACT FROM AUTHOR]