Conversion of a glutamate dehydrogenase into methionine/norleucine dehydrogenase by site‐directed mutagenesis

Abstract

In earlier attempts to shift the substrate specificity of glutamate dehydrogenase (GDH) in favour of monocarboxylic amino-acid substrates, the active-site residues K89 and S380 were replaced by leucine and valine, respectively, which occupy corresponding positions in leucine dehydrogenase. In the GDH framework, however, the mutation S380V caused a steric clash. To avoid this, S380 has been replaced with alanine instead. The single mutant S380A and the combined double mutant K89L/S380A were satisfactorily overexpressed in soluble form and folded correctly as hexameric enzymes. Both were purified successfully by Remazol Red dye chromatography as routinely used for wild-type GDH. The S380A mutant shows much lower activity than wild-type GDH with glutamate. Activities towards monocarboxylic substrates were only marginally altered, and the pH profile of substrate specificity was not markedly altered. In the double mutant K89L/S380A, activity towards glutamate was undetectable. Activity towards L-methionine, L-norleucine and L-norvaline, however, was measurable at pH 7.0, 8.0 and 9.0, as for wild-type GDH. Ala163 is one of the residues that lines the binding pocket for the side chain of the amino-acid substrate. To explore its importance, the three mutants A163G, K89L/A163G and K89L/S380A/A163G were constructed. All three were abundantly overexpressed and showed chromatographic behaviour identical with that of wild-type GDH. With A163G, glutamate activity was lower at pH 7.0 and 8.0, but by contrast higher at pH 9.0 than with wild-type GDH. Activities towards five aliphatic amino acids were remarkably higher than those for the wild-type enzyme at pH 8.0 and 9.0. In addition, the mutant A163G used L-aspartate and L-leucine as substrates, neither of which gave any detectable activity with wild-type GDH. Compared with wild-type GDH, the A163 mutant showed lower catalytic efficiencies and higher K(m ) values for glutamate/2-oxoglutarate at pH 7.0, but a similar k(cat)/K(m) value and lower K(m) at pH 8.0, and a nearly 22-fold lower S(0.5) (substrate concentration giving half-saturation under conditions where Michaelis-Menten kinetics does not apply) at pH 9.0. Coupling the A163G mutation with the K89L mutation markedly enhanced activity (100-1000-fold) over that of the single mutant K89L towards monocarboxylic amino acids, especially L-norleucine and L-methionine. The triple mutant K89L/S380A/A163G retained a level of activity towards monocarboxylic amino acids similar to that of the double mutant K89L/A163G, but could no longer use glutamate as substrate. In terms of natural amino-acid substrates, the triple mutant represents effective conversion of a glutamate dehydrogenase into a methionine dehydrogenase. Kinetic parameters for the reductive amination reaction are also reported. At pH 7 the triple mutant and K89L/A163G show 5 to 10-fold increased catalytic efficiency, compared with K89L, towards the novel substrates. In the oxidative deamination reaction, it is not possible to estimate k(cat) and K(m) separately, but for reductive amination the additional mutations have no significant effect on k(cat) at pH 7, and the increase in catalytic efficiency is entirely attributable to the measured decrease in K(m). At pH 8 the enhancement of catalytic efficiency with the novel substrates was much more striking (e.g. for norleucine approximately 2000-fold compared with wild-type or the K89L mutant), but it was not established whether this is also exclusively due to more favourable Michaelis constants.