DFT calculations of comparative energetics and ENDOR/Mossbauer properties for two protonation states of the iron dimer cluster of ribonucleotide reductase intermediate X

Two models (I and II) for the active site structure of class-I ribonucleotide reductase (RNR) intermediate X in subunit R2 have been studied in this paper, using broken-symmetry density functional theory (DFT) incorporated with the conductor like screening (COSMO) solvation model and with the finite-difference Poisson-Boltzmann self-consistent reaction field (PB-SCRF) calculations. Only one of the bridging groups between the two iron centers is different between model-I and model-II. Model-I contains two mu-oxo bridges, while model-II has one bridging oxo and one bridging hydroxo. These are large active site models including up to the fourth coordination shell H-bonding residues. Mossbauer and ENDOR hyperfine property calculations show that model-I is more likely to represent the active site structure of RNR-X. However, energetically our pK(a) calculations at first highly favored the bridging oxo and hydroxo (in model-II) structure of the diiron center rather than having the di-oxo bridge (in model-I). Since the Arg236 and the nearby Lys42, which are very close to the diiron center, are on the protein surface of RNR-R2, it is highly feasible that one or two anion groups in solution would interact with the positively charged side chains of Arg236 and Lys42. The anion group(s) can be a reductant, phosphate, sulfate, nitrate, and other negatively charged groups existing in biological environments or in the buffer of the experiment. Since sulfate ions certainly exist in the buffer of the ENDOR experiment, we have examined the effect of the sulfate (SO42-, surrounded by explicit water molecules) H-bonding to the side chain of Arg236. We find that when sulfate interacts with Arg236, the carboxylate group of Asp237 tends to be protonated, and once Asp237 is protonated, the Fe(III) Fe(IV) center in X favors the di-oxo bridge (model-I). This would explain that the ENDOR observed RNR-X active site structure is likely to be represented by model-I rather than model-II.