Converting GLX2-1 into an active glyoxalase II

Abstract

Arabidopsis thaliana glyoxalase 2-1 (GLX2-1) exhibits extensive sequence similarity with GLX2 enzymes but is catalytically inactive with SLG, the GLX2 substrate. In an effort to identify residues essential for GLX2 activity, amino acid residues were altered at positions 219, 246, 248, 325, and 328 in GLX2-1 to be the same as those in catalytically active human GLX2. The resulting enzymes were overexpressed, purified, and characterized using metal analyses, fluorescence spectroscopy, and steady-state kinetics to evaluate how these residues affect metal binding, structure, and catalysis. The R246H/N248Y double mutant exhibited low level S-lactoylglutathione hydrolase activity, while the R246H/N248Y/Q325R/R328K mutant exhibited a 1.5-2-fold increase in k(cat) and a decrease in K(m) as compared to the values exhibited by the double mutant. In contrast, the R246H mutant of GLX2-1 did not exhibit glyoxalase 2 activity. Zn(II)-loaded R246H GLX2-1 enzyme bound 2 equiv of Zn(II), and (1)H NMR spectra of the Co(II)-substituted analogue of this enzyme strongly suggest that the introduced histidine binds to Co(II). EPR studies indicate the presence of significant amounts a dinuclear metal ion-containing center. Therefore, an active GLX2 enzyme requires both the presence of a properly positioned metal center and significant nonmetal, enzyme-substrate contacts, with tyrosine 255 being particularly important.

Figure 1

Crystal structure of Arabidopsis thaliana GLX2-5 and human GLX2 (1, 15). Numbering scheme corresponds to the sequences in Figure 2.

Figure 2

Alignment of predicted plant glyoxalase II’s from A. thaliana. The * marks the metallo-β-lactamase fold motif. The # marks the highly-conserved metal binding residues. The Δ marks the substrate binding residues.

Figure 3

Computational model of the active site of GLX2-1 overlapped with the active site of GLX2-5. Metal binding ligands are the same in both models except residue Arg246. Gray residues indicate the metal binding sites of GLX2-1, and black residues indicate the metal binding site of GLX2-5.

Figure 4

¹H NMR spectrum of the 2Co-R246H mutant of GLX2-1 in 10 mM MOPS, pH 7.2, containing 10% D₂O. The enzyme concentration in the samples was ~ 1 mM. The * represents peaks that were solvent-exchangeable. Peak e in the spectrum decreased by 1/2 when the sample was exchanged in 90% D₂O.

Figure 5

Fluorescence emission spectra of wild-type GLX2-1 and GLX2-1 mutants. The concentration of samples was 10 μM, and the buffer was 10 mM MOPS, pH 7.2. An excitation wavelength of 295 nm was used.

Figure 6

EPR spectra from GLX2-1 and the H238R mutant of GLX2-5. (A) As-isolated wild-type GLX2-1. (B) As-isolated H238R mutant of GLX2-5. (C) Simulation of the g_eff. = 1.775 - 1.945 signal, assuming interacting Fe(III) (S = 5/2, isotropic g = 2.0, D = 2 cm⁻¹, E/D = 1/3) and Fe(II) (S = 2, g = 1.92, 2.01 and 2.01, D = 15 cm⁻¹, E/D = 0.085), with J_{Fe(II)-Fe(III)} = 18 cm⁻¹ and an inter-iron distance r_{Fe(II)-Fe(III)} = 3.5 Å. The discrepancy in lineshape at the high-field side of the middle resonance is likely due to broadening in the experimental spectrum due to strains in D and/or J. (D) H238R mutant of GLX2-5 after addition of 1.5 eq. Fe(II) and 1.5 eq. Zn(II). (E) Computed spectrum, assuming S = 5/2 (Fe(III)), g = 2, D = 1 cm⁻¹, E/D = 0.2. (F) Difference spectrum generated by subtraction of B from D. Experimental spectra were recorded at 2 mW microwave power, 10 K, and 12 G (1.2 mT) magnetic field modulation at 100 kHz.

Figure 7

Fluorescence emission spectra of wild-type GLX2-5 and the H238R mutant of GLX2-5. The concentration of samples was 10 μM, and the buffer was 10 mM MOPS, pH 7.2. An excitation wavelength of 295 nm was used.

Figure 8

(Left) Homology model of GLX2-1 showing a potential remote metal binding site consisting of Asp96, Asp98, and His179. (Right) Remote metal binding sites of human GLX2 on subunits A and B (15). The large dark sphere is a solvent molecule.