Gamma-aminobutyric acid (GABA), an important bioactive compound, is synthesized through the decarboxylation of L-glutamate (L-Glu) by glutamate decarboxylase (GAD). The use of lactic acid bacteria (LAB) as catalysts opens interesting avenues for the biosynthesis of food-grade GABA. However, a key obstacle involved in the improvement of GABA production is how to resolve the discrepancy of optimal pH between the intracellular GAD activity and cell growth. In this work, a potential GAD candidate (LpGadB) from Lactobacillus plantarum was heterologously expressed in Escherichia coli. Recombinant LpGadB existed as a homodimer under the native conditions with a molecular mass of 109.6 kDa and exhibited maximal activity at 40 degrees C and pH 5.0. The K-m value and catalytic efficiency (k(cat)/K-m) of LpGadB for L-Glu was 21.33 mM and 1.19 mM(-1)s(-1), respectively, with the specific activity of 26.67 mu M/min/mg protein. Subsequently, four C-terminally truncated LpGadB mutants (GadB(Delta C10), GadB(Delta C11), GadB(Delta C12), GadB(Delta C13)) were constructed based on homology modeling. Among them, the mutant GadB(Delta C11) with highest catalytic activity at near-neutral pH values was selected. In further, the GadB(Delta C11) and Glu/GABA antiporter (GadC) of Lactococcus lactis were co-overexpressed in the host L. lactis NZ3900. Finally, after 48 h of batch fermentation, the engineered strain L. lactis NZ3900/pNZ8149-gadB(Delta C11)C yielded GABA concentration up to 33.52 g/L by applying a two-stage pH control strategy. Remarkably, this is the highest yield obtained to date for GABA from fermentation with L. lactis as a microbial cell factory.