Process modeling and optimization of high yielding L-methioninase from a newly isolated Trichoderma harzianum using response surface methodology and artificial neural network coupled genetic algorithm

Abstract

Abstract A high yielding L-methioninase producing fungus was isolated from soil samples and was identified as Trichoderma harzianum. The enzyme was purified using chromatographic techniques and the purified enzyme had an apparent molecular mass of 48 kDa on SDS PAGE. The purified L-methioninase showed a specific activity of 74.4 U/mg, which is the highest among other L-methioninase, reported and can be a potential candidate as an anticancer agent. The media components such as lactose, L -methionine, KH2PO4, K2HPO4 and Zinc chloride was found to have a significant effect on enzyme production by classical optimization method. Response surface methodology and artificial neural network linked genetic algorithm was employed to develop an optimized medium for L-methioninase production. Maximum enzyme production obtained using RSM was 30.2 U/ml at medium composition of 12.5 g/l of lactose, 10 g/l of L -methionine, 2 g/l of KH2PO4, 4 g/l of K2HPO4 and 0.0125 g/l of zinc chloride. ANN model was found to be superior to RSM with a higher coefficient of determination (R2) of 0.995, lower RSME (0.306) and MSE (0.093). A higher enzyme production of 33.32 U/ml was achieved using ANN-GA optimized medium; 13.9 g/l of lactose, 11.37 g/l of L -methionine, 1.58 g/l of KH2PO4, 3.98 g/l of K2HPO4 and 0.01 g/l of zinc chloride, which is in agreement with the predicted value of 33.76 U/ml.