Statistical optimization of γ-aminobutyric acid production by response surface methodology and artificial neural network models using Lactobacillus fermentum isolated from palm wine

The aim of this study was to screen and optimize gamma amino butyric acid (GABA) production by Lactobacillus fermentum (Lb. fermentum) in soybean milk and De Man-Rogosa Sharp (MRS) media. The GABA yield was confirmed by thin layer chromatography and quantitatively estimated by ninhydrine method. Response surface methodology (RSM), a statistical tool was applied for optimization of GABA yield. The accuracy of RSM predicted results were demonstrated by a non-statistical (hypothetical) model using Artificial Neural Network (ANN). On TLC plate GABA produced by Lb. fermentum showed same Rf value (0.54) compared with standard GABA. Lb. fermentum produced 6.94 g/L and 3.07 g/L of GABA in MRS media and soymilk respectively. After optimization there was 1.7 fold increase in the GABA yield in optimized conditions when compared to unoptimized soybean milk. In single variable optimization Monosodium glutamate (MSG), glucose and incubation period (IP) were found to favor for GABA production. Through RSM model the maximum predicted GABA yield (5.34 g/L) was observed with a concentration of 1.5% of MSG and 1.19% of glucose with 48 h of incubation. This is the first report on production of GABA in soybean milk by Lb. fermentum isolated from palm wine. Production of GABA by Lb. fermentum in soymilk as a basal substrate, is economical in comparison with synthetic media (MRS). This convincing results from this study could be a touchstone for exploring Lb. fermentum strain to obtain GABA enriched functional food for human consumption.