Abstract
A novel method of structure prediction for membrane-bound proteins is reviewed. The approach is based on a sequence-function analysis, secondary structure prediction and subsequent geometry optimization. The prediction of the structure of ligand-gated P2X4 receptor subunit, which is a membrane-embedded cation channelforming protein with extracellularly occuring ATP binding sites, is shown as an example. The potential Nglycosylation sites of the glycoprotein, location of the five disulfide bridges, and phospholipid-dependent protein kinase C (PKC) phosphorylation attachment sites were determined by sequence-function analysis. Subsequently an attempt was made to predict its conformation using homology-based comparative modeling and threading; however, the modeling could not be accomplished. Because of this, secondary structure prediction of the protein was carried out. The input coordinates of the spatial structure were obtained by a profile-based neural network prediction method. The resulting secondary structure was converted into a three-dimensional geometry. The secondary and tertiary structures were optimized by the quantum chemistry RHF/3-21G minimal basic set and all-atom molecular mechanics AMBER96 force field. The predicted shape is similar to the shape of the experimentally obtained monomeric structure of the classical ion channel, the K+ion channel from Streptomyces lividans (KcsA channel), and agrees with the P2X shape proposed by biological experimenters. The geometry optimized structure of the P2X4 receptor is freely available (Protein Data Bank format) from the authors on e-mail request (magp@medizin.unileipzig. de).
Keywords: Proteomics, structural bioinformatics, ATP-activated purinergic receptor, P2X receptor, cation channel, KcsA protein