Effects of experimentally achievable improvements in the quality of NMR distance constraints on the accuracy of calculated protein structures.

New methods for collecting cross-relaxation data from proteins and nucleic acids make it possible to improve the accuracy and precision of interproton distance measurements used as input for NMR solution structure determinations. It thus is of interest to determine whether such experimentally achievable improvements in input distance constraints have significant effects on the precision and accuracy of the resulting structures. To answer this question, we have turned to a computational procedure involving the use of data simulated from a known structure, in order to allow unambiguous assessments of accuracy. The approach to improved distances evaluated here is that afforded by magnetization exchange network editing (MENE); MENE pulse sequences break the network of cross-relaxation interactions into regions that are manipulated so as to defeat certain spin-diffusion terms. A target structure was prepared from the X-ray structure of a small protein, turkey ovomucoid third domain (OMTKY3). A normal NOESY spectrum and two varieties of MENE spectra, BD-NOESY and CBD-NOESY, were simulated by means of complete relaxation matrix analysis. These results were used to create different input data sets with the same number of constraints (perfectly accurate distances derived from the target structure, more accurate distances derived from the MENE simulations, and less accurate distances derived from the NOESY simulation), and these, interpreted at different levels of precision, were used as input for solution structure calculations. The results showed that the use of more precise input data measurably improves the local precision and accuracy of calculated structures, but only if the more precise data include the actual target distance. Incorporation of the experimentally achievable, accurate distances with higher precision afforded by the MENE pulse sequences into the set of input distances was found to improve the accuracy of the resulting structures, particularly in terms of side-chain conformation.