Enzymatic Glycosylation of Phenolic Compounds Using BsGT-3 Based on Molecular Docking Simulation

Abstract

Genome projects reveal myriads of genes and their corresponding proteins. Each gene is annotated based on the amino acid sequence homology with the functionally characterized proteins [Siew and Fischer, 2003]. This classification method provides an initial starting point for the functional characterization of a particular gene or protein. However, because in vitro characterization of an individual gene is labor-consuming, a more simple method has been developed. One of them is the molecular modeling, which aids in the narrowing of the substrate range. UGTs are found ubiquitously from microorganisms to plants and mammals [Mackenzie et al., 1997; Vogt and Jones, 2000; Liang and Qiao, 2007], with more than 1,800 available in the carbohydrate-active enzyme database (http://www.cazy.org/). UGTs transfer a sugar group from the nucleotide sugar to small compounds including antibiotics, phenolics, and alkaloids. Attachment of the transferred sugar to these molecules enhances the water solubility and sometimes alters the biological activities [Vogt and Jones, 2000]. For example, in humans the absorption rate of glucose differs depending on the location of sugar in quercetin [Erlund, 2004]. In addition, the biological activity of small molecules is altered by the attachment position and the kind of sugar [Day et al., 2003]. Chemical attachment of the sugar to a certain form of compound is hindered by complex reactions, resulting in a low yield of the final compound. Although biological glycosylation using enzymes or cells became more attractive [Lim, et al., 2004; Ko, et al., 2006a], a series of cloning, expression, and substrate determination process is time-consuming. On the other hand, prediction of the substrate through the amino acid sequence could accelerate the whole process of UGT characterization. We used UGT (BsGT-3) from Bacillus subtilis and carried out molecular modeling, followed by molecular docking to predict the substrates of UGTs. The molecular structure of BsGT-3 was built using the homology-based molecular modeling method. The template protein used was a glycosyltransferase, deposited in the Protein Data Bank (PDB code: 1pn3.pdb). The structure of BsGT-3 was constructed using Modeller 8.2 (Accelrys, San Diego, CA). To refine the 3D structure of BsGT-3, energy minimization and molecular dynamics were performed. The protein was embedded in a 5-A shell of 1271 water molecules. In energy minimization, the force field was cvff provided by Accelys, and steepest descents were performed until maximum derivative of 0.001 kcal/molA was achieved. Subsequently, molecular dynamics was performed using a computational Grid system (http://www.mgrid.or.kr) at 300 K, 1 atm for 1 ns with 1 fs in each step. The output conformers were collected at every 400 fs. The energy profiles of 2,500 conformers were then analyzed by Analysis/InsightII, choosing 20 conformers at every 50 ps. Their superimposed structures showed the root mean square derivation of 0.8 A. Of the twenty conformers, the conformer showing the lowest total energy was evaluated by PROCHECK. The substrate for the docking study was built up using Sybyl 7.2 (Tripos, St. Louis, MO) on Pentium IV 3.2 Ghz Linux PC. Gasteiger-Huckel charge was given to the substrate, which was subjected to energy minimization by the conjugate gradient algorithm using Tripos force field. The minimization process was forced to stop either when the iteration number reached 1,000 or when the convergence criteria were met (maximum root mean square gradient 0.05 kcal/mol). The overall substrate-binding site of BsGT-3 is larger than the other UGTs from plants, and contains a conserved histidine residue (His16), which serves as a base to take a proton from a substrate as found in other UGTs [Shao et al., 2005; Offen et al., 2006]. His16 made array with *Corresponding author Phone: +82-2-450-3764; Fax: +82-2-3437-6106 E-mail: jhahn@konkuk.ac.kr