Abstract Glutamate decarboxylase (GAD) transforms L-glutamate into γ-aminobutyric acid (GABA) with the consumption of a proton. GAD derived from lactic acid bacteria exhibits optimum activity at pH 4.0–5.0 and significantly loses activity at near-neutral pH. To broaden the active range of the GAD GadB1 from Lactobacillus brevis Lb85 toward a near-neutral pH, irrational design using directed evolution and rational design using site-specific mutagenesis were performed. For directed evolution of GadB1, a sensitive high-throughput screening strategy based on a pH indicator was established. One improved mutant, GadB1T17I/D294G/Q346H, was selected from 800 variants after one round of EP-PCR. It exhibited 3.9- and 25.0-fold increase in activity and catalytic efficiency, respectively at pH 6.0. Through site-specific mutagenesis, several improved mutants were obtained, with GadB1E312S being the best one. The combined mutant GadB1T17I/D294G/E312S/Q346H showed even higher catalytic efficiency, 13.1- and 43.2-fold that of wild-type GadB1 at pH 4.6 and 6.0, respectively. The amount of GABA produced in gadB1T17I/D294G/Q346H-, gadB1E312S- and gadB1T17I/D294G/E312S/Q346H-expressing Corynebacterium glutamicum ATCC 13032 from endogenous L-glutamate increased by 9.6%, 20.3% and 63.9%, respectively. These results indicate that these mutations have beneficial effects on expanding the active pH range and on GABA biosynthesis, suggesting these GadB1 variants as potent candidates for GABA production.