Accuracy of bound peptide structures determined by exchange transferred nuclear Overhauser data: A simulation study

The exchange-transferred NOE method to determine the three-dimensional stru cture of peptides bound to proteins, or other macromolecular systems, is be coming increasingly important in drug design efforts and for large or multi component assemblies, such as membrane receptors, where structural analysis of the full system is intractable. The exchange-transferred nuclear Overha user effect spectroscopy (etNOESY) method allows the determination of the b ound-state conformation of the peptide from the intra-molecular NOE interac tions between ligand protons. Because only ligand-ligand NOEs are generally observable, the etNOESY method is restricted to fewer NOEs per residue tha n direct protein structure determination. In addition, the averaging of rel axation rates between free and bound states affects the measured cross-peak intensities, and possibly the accuracy of distance estimates. Accordingly, the study reported here was conducted to examine the conditions required t o define a reliable structure. The program CORONA was used to simulate etNO E data using a rate-matrix including magnetic relaxation and exchange rates for two peptide-protein complexes derived from the reference complex of cA MP-dependent protein kinase ligated with a 24-residue inhibitor peptide. Th e results indicate that reasonably accurate peptide structures can be deter mined with relatively few NOE interactions when the interactions occur betw een non-neighboring residues. The reliability of the structural result is s uggested from the pattern of NOE interactions. A structure with an accuracy of approximately 1.3 Angstrom rms difference for the main-chain atoms can be obtained when etNOE interactions between non-neighboring residues occur over the length of the peptide. The global precision is higher (approximate ly 0.9 Angstrom rms difference) but is not correlated to global accuracy. A local definition of precision along the backbone appears to be a good indi cator of the local accuracy.