Comparative DFT study on the α-glycosidic bond in reactive species of galactosyl diphosphates

Abstract

AbstractCorrect prediction of the structure and energetics along the reaction pathway of the formation or dissociation of the glycosidic bond in sugar phosphates is crucial for the understanding of catalytic mechanism and for the determination of transition state structures of sugar-phosphate processing enzymes. The performance of seven density functional theory (DFT) methods (BLYP, B3LYP, MPW1PW91, MPW1K, MPWB1K, M05 and M05-2X) and two wave function methods (HF and MP2) was tested using four structural models with the activated sugar-phosphate α-glycosidic linkage. The models were chosen based on the crystal structure of the retaining glycosyltransferase LgtC complex with methyl α-d-galactopyranose diphosphate and its 2-fluoro derivative. Results of the MP2 method were used as a benchmark for the other methods. Two structural trends were observed in the calculations: predicted length of the activated C1-O1 glycosidic bond of 1.49–1.63 Å was significantly larger than values of a standard C1-O1 glycosidic bond in crystal structures of carbohydrates (1.39–1.48 Å), and the calculated value depended on the DFT method used. The MPW1K, M05 and M05-2X functionals provided results in closest agreement with those from the MP2 method, the difference being less than 0.02 Å in the calculated glycosidic bond lengths. On the contrary, the BLYP and B3LYP functionals failed to predict sugar diphosphate in the (-sc) conformation as a stable structure. Instead, the only stationary points localized along the C1-O1 dissociation coordinate were oxocarbenium ions at the distance of approximately 2.8 Å. The M05-2X, MPW1K and MPWB1K functionals gave the most reasonable prediction of the thermochemical kinetic parameters, where the formation of the oxocarbenium ion has a slightly endothermic character (0.4–1.7 kJ mol−1) with an activation barrier of 7.9–9.2 kJ mol−1.