Does Acinetobacter calcoaceticus glucose dehydrogenase produce self-damaging H2O2?

Abstract

The soluble glucose dehydrogenase (sGDH) from Acinetobacter calcoaceticus has been widely studied and is used, in biosensors, to detect the presence of glucose, taking advantage of its high turnover and insensitivity to molecular oxygen. This approach, however, presents two drawbacks: the enzyme has broad substrate specificity (leading to imprecise blood glucose measurements) and shows instability over time (inferior to other oxidizing glucose enzymes). We report the characterization of two sGDH mutants: the single mutant Y343F and the double mutant D143E/Y343F. The mutants present enzyme selectivity and specificity of 1.2 (Y343F) and 5.7 (D143E/Y343F) times higher for glucose compared with that of the wild-type. Crystallographic experiments, designed to characterize these mutants, surprisingly revealed that the prosthetic group PQQ (pyrroloquinoline quinone), essential for the enzymatic activity, is in a cleaved form for both wild-type and mutant structures. We provide evidence suggesting that the sGDH produces H2O2, the level of production depending on the mutation. In addition, spectroscopic experiments allowed us to follow the self-degradation of the prosthetic group and the disappearance of sGDH's glucose oxidation activity. These studies suggest that the enzyme is sensitive to its self-production of H2O2. We show that the premature aging of sGDH can be slowed down by adding catalase to consume the H2O2 produced, allowing the design of a more stable biosensor over time. Our research opens questions about the mechanism of H2O2 production and the physiological role of this activity by sGDH.

Keywords: hydrogen peroxide production; protein stability; soluble glucose dehydrogenase.

Figure 1. Non-linear regression on glucose and maltose range for wild-type sGDH and Y343F or D143E/Y343F mutants at 37°C

The oxidation activities of sGDH wild-type and mutants were determined by spectrometry to follow the reduction of 2,6-dichlorophenolindophenol (DCIP) (ε₆₀₀ = 21.6 mM⁻¹·cm⁻¹) at 600 nm, using phenazine methosulfate (PMS). Sugar concentrations varied between 0 and 400 mM for maltose and 0 and 800 mM for glucose. Curves in A, B, D, E were fitted according to (eqn 1); curve C according to (eqn 2) and curve F according to (eqn 3) (see Material and Methods section).

Figure 2. Active site of the holo sGDH wild-type with glucose

The figure was prepared by the authors based on the structure factors and refinement, published by Oubrie et al. in 1999 (PDB 1CQ1) [30], obtained from the PDB-REDO [31] server. The electron density (grey mesh) shown is calculated from a Polder Omit Map (PHENIX) contoured with a cutoff of 5σ. The positions of the protein chain (ribbon diagram), calcium atom (dark blue), PQQ (blue ball and stick model), and glucose binding (yellow ball and stick model) are shown.

Figure 3. Dimeric interface of the wild-type sGDH

(A) The electrostatic charge was calculated with the macromolecular electrostatics calculation plugin Plug-APBS available under PyMOL [33]. A well ordered and extensive water network (yellow spheres) is present between the two monomers, as seen here in a lateral view of the dimer interface. (B) Residues K272 and D396, from opposing subunits, form a salt bridge, identified with PISA interface [32], a web-server available via PDBe. The residues in blue belong to subunit A, those in yellow to subunit B. The region between residues N262 and D273, referred to as the 4CD loop (red), is stabilized by calcium ions (blue) present at the dimeric interface.

Figure 4. Active site of the holo sGDH wild-type

The active site of PQQ from the wild-type enzyme at pH 8. The density at 5 σ from a Polder Omit Map (PHENIX) is superimposed onto a model of cleaved PQQ. Residues close to the cofactor are labeled, and show little reorganization compared with the structures published by Oubrie et al. When compared with the structure in Figure 2, the cleavage of PQQ is clearly demonstrated and the pyrrole ring is seen to be almost perpendicular to the quinone. Similar cleavage was observed in all measured structures in this study. Electron density is observed directly above the cleaved quinone, which could possibly be attributed to a molecule of H₂O₂, although there is no direct evidence to substantiate this hypothesis. Consequently, this density has intentionally been left unmodeled in the refined structure.

Figure 5. Sensitivity of sGDH to H₂O₂ at pH 7.5 (A) and pH 5.0 (B)

The k_obs were calculated from the decreasing exponential fit at pH 7.5 and at pH 5.0; values are reported in Table 2.

Figure 6. Evidence of aging and catalase effect

Evidence for the aging of sGDH at 25°C, observed through three different parameters: spectroscopic properties at 340 nm (A), retention of activity (B), and the relative stability of modified electrodes measured using chronoamperometry at +0.3V vs. Ag/AgCl in a 50 mM TRIS buffer solution (pH 7.5) containing 50 mM glucose under air at 25°C (C). The time-dependent changes are shown in the presence (indicated by bold lines) or absence (represented by dashed lines) of 10,000 U catalase. The figures maintain the same color scheme throughout the three graphs.