THE DIMANGANESE(III,IV) OXIDATION-STATE OF CATALASE FROM THERMUS-THERMOPHILUS - ELECTRON NUCLEAR DOUBLE-RESONANCE ANALYSIS OF WATER AND PROTEIN LIGANDS IN THE ACTIVE-SITE

The H-1 hyperfine tensors of the dimanganese(III,IV) oxidation state of the non-heme-type catalase enzyme from the thermophilic bacterium Thermus thermophilus have been measured by electron nuclear double resonance (ENDOR) spectroscopy at pH 6.5-9. These were compared to model dimanganese(III,IV) complexes possessing six-coordinate N4O2, N3O3, and O6 atom donor sets to each Mn and mu-oxo and mu-carboxylato bridging ligands. The lack of N-14 hyperfine couplings in the enzyme suggests either O6 or O5N ligand donors to each Mn. Moreover, the two sigma coordination sites on Mn(III) directed at the d(z2) orbital cannot be occupied by N ligands. The H-1 ENDOR spectrum revealed two types of anisotropic tensors, attributable to two D2O-exchangeable protons on the basis of the magnitude of the electron paramagnetic resonance (EPR) line narrowing in D2O. All six of the H-1 hyperfine couplings are proposed to arise from a single displaceable water molecule in the active site, on the basis of their reversible disappearance, upon incubation in D2O or by precipitation from ammonium sulfate, and by simulation of the H-1 ENDOR spectrum. The Mn ions are coordinated predominantly by nonmagnetic O atoms lacking covalently bound protons in both alpha and beta positions. This implicates predominantly carboxylato-type ligands (Asp and Glu) and possibly a di-mu-oxo bridge between Mn ions. The latter is supported also by the presence of strong antiferromagnetic coupling. Comparison to other dimetalloproteins also possessing the four-helix bundle structural motif shows that the polyoxo(carboxylato) coordination in catalase differs significantly from the polyhistidine coordination adopted by the diiron(II,II) site in the O2-binding protein myohemerythrin, but resembles the polycarboxylato ligation adopted by the diiron(III,III) site of ribonucleotide reductase. The catalase H-1 ENDOR spectrum is essentially identical to that for the exchangeable protons in the active site of the diiron(II,III) state of uteroferrin, an acid phosphatase [Doi et al. (1988) J. Biol. Chem. 263, 5757-5763], and also for a polycarboxylato complex possessing the Mn2(mu-O)2 core with H-bonded water ligands. The H-1 ENDOR line shape in catalase could be simulated using a theoretical model suitable for multispin clusters. It treats the two Mn spins as point dipoles which are exchange-coupled. It includes both dipolar and isotropic ligand hyperfine couplings. Using this model, the position of the proton with the largest interaction could be located with respect to the Mn-Mn vector because of the extreme sensitivity of line shape to position. Four possible positions are predicted, differing solely by the undetermined sign of the experimental hyperfine tensor elements. Analysis of the H-1 ENDOR spectrum via angle selection using the EPR line shape anisotropy enabled determination of the location of the proton relative to the d(z2) orbital of Mn(III), which is orthogonal to the presumed Mn(mu-O)2 plane. Simulation of the angle-selected H-1 ENDOR spectrum revealed that only one proton position could account for the spectrum. This gives distances of 4.6 and 2.9 angstrom to Mn(III) and Mn(IV), respectively. These coordinates are compatible with a water or hydroxide molecule H-bonded to a terminal ligand on Mn(IV). A second proton exists at a greater distance. The structural basis for the lack of catalase activity in the Mn2(III,IV) oxidation state is suggested to originate from the kinetic inertness to substitution or reduction of the bridging oxo atoms in the Mn2(mu-O)2 core.