A SPECTRAL CORRELATION-FUNCTION FOR EFFICIENT SEQUENTIAL NMR ASSIGNMENTS OF UNIFORMLY N-15-LABELED PROTEINS

A new computer-based approach is described for efficient sequence-spec ific assignment of uniformly N-15-correlated [H-1,H-1]-NOESY spectra a re divided up into two-dimensional H-1-H-1 strips which extend over th e entire spectral width along one dimension and have a width of ca. 10 0 Hz, centered about the amide proton chemical shifts along the other dimension. A spectral correlation function enables sorting of these st rips according to proximity of the corresponding residues in the amino acid sequence. Thereby, starting from a given strip in the spectrum, the probability of its corresponding to the C-terminal neighboring res idue is calculated for all other strips from the similarity of their p eak patterns with a pattern predicted for the sequentially adjoining r esidue, as manifested in the scalar product of the vectors representin g the predicted and measured peak patterns. Tests with five different proteins containing both alpha-helices and beta-sheets, and ranging in size from 58 to 165 amino acid residues show that the discrimination achieved between the sequentially neighboring residue and all other re sidues compares well with that obtained with an unguided interactive s earch of pairs of sequentially neighboring strips, with important savi ngs in the time needed for complete analysis of 3D N-15-correlated [H- 1,H-1]-NOESY spectra. The integration of this routine into the program package XEASY ensures that remaining ambiguities can be resolved by v isual inspection of the strips, combined with reference to the amino a cid sequence and information on spin-system types obtained from additi onal NMR spectra.