Breeding Of γ-Aminobutyric Acid-Producing Lactobacillus And Optimization Of Fermentation Conditions

γ-amino butyric acid(GABA) is a major inhibitory neurotransmitter in the central nervous system. In this paper, GABA-producing strain breeding and fermentationg process were studied.A GABA-yielding strain hjxj-01 was isolated from the milk samples and identified as Lactobacillus brevis. It was shown that Lactobacillus brevis hjxj-Ol had the ablity of producing GABA. Then, UV and ⁶⁰Co y-rays were used to treat the original strain. After several times of mutagenesis, a mutant strain hjxj-08119 was bred by GABA resistance selection. Moreover, the GABA-producing capacity could be maintained stably after 12 generations. The fermentation results indicated that the average yield of GABA was 17 g/L, which was increased 142.9% compared with that of the origin strain hjxj-Ol.Artificial neural network (ANN) and particle swarm optimization (PSO) were applied to optimize GABA production by Lactobacillus brevis in shake flask cultures. Firstly, glucose, sodium glutamate and MnSO₄H₂O, which influenced GABA production positively were screened from 15 related factors by using Plackett-Burman design. The reasonable ranges of these three factors were determined by single factors experiment. Then according to hybrid design, 11 samples were selected for training network, and ANN was modeled. The results indicated that hybrid design was more rational than random design, and the model built by hybrid design shown good performance. The average error between the experimental value and predictive values was 2.37%. Finally, based on the model, the optimized condition was predicted by particle swarm optimization. Verification experiment showed that the optimized condition leaded the average yield of GABA to be increased from 17.06g/L to 33.4g/L, and in the optimized condition, the relative discrepancy was only 1.49% between the experimental value and the predictive value of GABA yield.The effects of culture condition on GABA yield of Lactobacillus hjxj-08119 inbatch fermentation was investigated. Evaluation of flux distribution for glutamate metabolism was used first to determine which metabolic pathways, significant for optimization in theory, should be controlled. More specifically, control for the local pathways from glutamate to GABA and from GABA to succinate were very important with regard to the GABA molar yield and GABA production rate. Subsequently, twenty three factors, including pH, temperature, pyridoxal phosphate(PLP), 2-ketoglutarate, etc, which might influence the activity of glutamate decarboxylase, 4-aminobutyrate aminotransferase and succinate semialdehyde dehydrogenase were investigated employing Plackett-Burman designs (PBD), hybrid designs (HD), artificial neural networks (ANN) and particle swarm optimization (PSO). According to the PBD analysis, FeCl₃, PLP and pH were proved to be significant parameters for the GABA molar yield. When the condition was controlled as pH=5.08, c(FeCl3)=5.1 mmol/L and c(PLP)=0.71 mmol/L after 20h cluture in batch fermentation, the GABA molar yield and GABA production rate between 20h and 40h were to be increased by 13.1% and 45.9% (from 71% to 80.27% and from 8.25 mmol/(L·h) to 12.04 mmol/(L-h) respectively compared with those obtained in the original condition. Based on the optimal condition in batch fermentation, the fed- batch fermentation was investigated too. The addition of glucose and L-MSG could result in accumulation of GABA to reach as high as 107.5g/L after fermentation culture of 91.5 h.A non-structured model was proposed to simulate the growth of Lactobacillus hjxj-08119, the consumption of glucose, the consumption of L-MSG and the accumulation of GABA in batch fermentation. The parameters of fermentation kinetics (μ_max,α , β , m₁ , m₂, Y_x/s1 , Y_p/s2 and Y_s1/s2) was estimated usingPSO.