Improved Glutamate Decarboxylase By Computer-Aided Enzyme Design

Glutamate decarboxylase (GAD) is an essential enzyme widely distributed in nature from microorganisms to plants and animals. It is the key enzyme for the biosynthesis of GABA, which is useful as a functional bioactive component in food and pharmaceutical, for its anti-hypertensive and analgesic properties as well as calming effects.Currently the mass production of GABA in bioreactors is still limited due to the bottleneck caused by the low substrate solubility at the acidic pH optimum of GAD and the inherently thermal unstable nature of this enzyme. Based on the prospects in utilizing Lactobacillus brevis GAD for the biosynthesis of GABA, we employed bioinformatics methods such as homology modeling, molecular docking and computational enzyme design to create a more efficient biocatalyst for GABA preparation, thus breaking through the above mentioned bottleneck.Firstly, a pH-sensitive colorimetric assay was established to quantitatively measure GAD activity in bacterial cell extracts using a microplate format. The assay is based on the color change of bromocresol green due to an increase in pH as protons are consumed during the enzyme-catalyzed reaction. Bromocresol green was chosen as the indicator because it has a similar pKa to the acetate buffer used. The corresponding absorbance change at620nm was recorded with a microplate reader as the reaction proceeded. The enzymatic reaction rate could be calculated using the formula:v=2.4×106×dA/dt (μmol·min-1). This is a simple, economical assay that can be carried out in robotic high-throughput devices in directed evolution experiments for the rapid determination of GAD activity.Secondly, we constructed the homology model of Lactobacillus brevis GAD with E. coli GadB as the template using both MODELLER and Swiss Model programs. Then PROCHECK, ERRAT and ProSA were used to evaluate the quality of resulting models. The analysis of enzyme active site revealed the following functional residues:Phe65and Thr215constituted the hydrophobic substrate entrance; Asp248and Lys279directly contributed the catalysis; Serl27, Asp248, His278, Lys279and an a-helix near the PLP phosphate group were responsible for keeping the favorable position and orientation of the cofactor PLP in the active center.Thirdly, we modeled the geminal diamine intermediate consisting of the substrate, PLP and Lys279according to the reaction mechanism, and then docked it into GAD homo logy model using ROSETTALIGAND. A total of5000docking trajectories were generated and the best scored enzyme-substrate complex showed that the hydrogen bonds from Gln166and Thr64were of vital importance in substrate binding. Besides, the positon of a-carboxylate group of the substrate was almost pepenticular to the pyridine ring of PLP, which was rotated by approximately6°with respect to its original position, with the PLP phosphate moiety acting as an anchor. These geometries are in accord with the characteristics often observed in PLP dependent decarboxylases. This work not only revealed the binding mode of substrate in the GAD active site and the distribution of functional residues, but also provided a structure model for further study on the enzyme structure-function relationship as well as enzyme design.Fourthly, we speculated that the substrate entrance was probably blocked by a C-terminal tail of GAD based on the homology model, which could be the reason for no detectable enzymatic activity at neutral pH. Site-directed mutagenesis was performed to delete14C-terminal residues to generate a mutant, designated as GAD△C, which exhibited extended activity at pH6.0compared to the wild type enzyme. Comparison of the UV-visible, fluorescence and circular dichroism spectra of the mutant with those of the wild type revealed that the microenvironment of the active site had been changed and the "blocking" effect might be eliminated. These results provided evidence for the important role of C-terminal region in the pH-dependent regulation of enzymatic activity, and the resulting mutant would be useful in a bioreactor for continuous production of GABA.Lastly, RosettaDesign algorithm was employed with the aim of improving the thermostability of GAD. Sequence space was searched with an iterative Monte Carlo procedure, replacing a single amino acid rotamer at a time, and reevaluating the energy. With global enregy minimum found, twenty point mutations were suggested by the program:A63N, T64Q, C66S, I87W, I98W, I105W, K138G, M185Y, M212W, Y216W, L226G, V229G, P240G, S249W, F256Y, V283G, W292G, C379V, K402H and K413G. Site-directed mutagenesis was used to generate each of these mutant enzymes. After an initial screening for the enzymatic activity and a further thermal denaturation experiment, the variant C379V was selected out, which increased the T1/2by5℃, and the catalytic efficiency was enhanced by19%compared with those of the wild-type GAD. Increased hydrophobic interactions brought about by this mutation was speculated as the main reason for the improved properties. This work created a more efficient biocatalyst for GABA preparation and built the basis for the computational thermostabilization of other enzymes for industrial use.