Estimation of accuracy in determining protein backbone conformations from NOE data and empirical φ, ψ probability distributions

Abstract

Listen

Translate article

Comprehensive computational experiments were carried out to examine systematically the potential for deriving protein backbone conformations using NOE data in conjunction with the probability distribution function of known backbone conformations of amino acid residues in proteins. It is shown that, in combination with the backbone conformational probability distribution, the measurement of NOE distances with moderate precision, consistent with the precision of modem NMR methods, is adequate to produce good local conformations of the protein backbone. The apparent need for much higher precision is not a fundamental property of the problem; it arose in previous treatments only from the mathematical statement of the problem and the approaches used. To demonstrate the improved precision resulting from combined use of NOE data and the φ, ψ probability distribution, a computer program, FISINOE, was developed, and control calculations with both simulated and experimental NOE data for bovine pancreatic trypsin inhibitor were performed. An ensemble of conformations obtained from the X-ray Protein Data Bank was used to obtain an approximate probability distribution for existing conformations of amino acid residues in proteins. A comprehensive statistical analysis of the results clearly shows that the set of sequential d connectivities (data on the presence or absence of sequential NOE cross peaks related to nearest-neighbor residues), combined with prior information about existing conformations, is sufficient to obtain a satisfactorily determined local conformation of the protein backbone in most cases; in some cases it would be helpful to use, as additional information, the cross-peak intensity between the NH and C αH intraresidual protons. Comparative studies indicate that the stability and accuracy in estimating the φ and ψ angles by the FISINOE program exceed those of other traditional approaches. The principal advantages of the FISINOE method are discussed, and the extension of this method for the determination of protein three-dimensional structures is outlined.