Hybrid-hybrid relaxation matrix structural refinement of duplex DNA from simulated 3D NOESY-NOESY data

A hybrid-hybrid relaxation rate matrix refinement of 3D NOESY-NOESY data was tested by simulated refinement calculations on a dodecamer duplex DNA structure. 'Experimental' 3D data sets were generated from the known target structure for the simulated refinement. Theoretical 3D NOE-NOE spectra were then calculated from several model-built structures. By scaling, the simulated experimental 3D data were merged with the theoretical 3D data to create a hybrid 3D NOESY-NOESY data set. This was then deconvoluted into a partial 2D hybrid volume matrix. The deconvoluted 2D hybrid NOESY data were then merged with the simulated 2D NOESY data (and, if available, additional experimental 2D data) to create a hybrid-hybrid 2D NOE volume matrix. Experimental 2D NOE data are not required for such a calculation. The hybrid-hybrid 2D data set was then used in an established 2D relaxation matrix procedure (MORASS) to produce distance constraints for restrained molecular dynamics refinement. The theoretical 3D data set and the model were refined in each iteration until agreement was obtained between the experimental and theoretical 3D NOESY-NOESY data sets. Two simulated 3D NOESY-NOESY data sets of the dodecamer duplex with 20% and 50% integration errors were used for refinement, containing 1667 and 1472 3D cross peaks, respectively. A total of 478 and 481 2D cross peaks were generated by deconvolution of the two 3D data sets, respectively. A simulated 2D NOE data set with 359 cross peaks (20% integration error) was used for comparison. Refinement calculations from three different models using all three data sets showed that the deconvoluted 3D NOESY-NOESY data can provide accuracy and precision comparable to or even better than those from 2D NOESY data. Importantly, the 3D data reproduced the sequence-specific variation in the helical parameters much better than the 2D data did. Computationally, this method does not require any significant increase in central processing unit time, storage or computer memory over the MORASS 2D protocol.