Fibronectin type III (FN-III) domains are autonomously folded modules found in a variety of multidomain proteins. The 10th FN-III domain from fibronectin (fnFN10) and the 3rd FN-III domain from tenascin-C (tnFN3) have 27% sequence identity and the same overall fold; however, the CC' loop has a different pattern of backbone hydrogen bonds and the FG loop is longer in fnFN10 compared to tnFN3. To examine the influence of length, sequence, and context in determining dynamical properties of loops, CC' and FG loops were swapped between fnFN10 and tnFN3 to generate four mutant proteins and backbone conformational dynamics on ps-ns and mus-ms timescales were characterized by solution (15)N-NMR spin relaxation spectroscopy. The grafted loops do not strongly perturb the properties of the protein scaffold; however, specific effects of the mutations are observed for amino acids that are proximal in space to the sites of mutation. The amino acid sequence primarily dictates conformational dynamics when the wild-type and grafted loop have the same length, but both sequence and context contribute to conformational dynamics when the loop lengths differ. The results suggest that changes in conformational dynamics of mutant proteins must be considered in both theoretical studies and protein design efforts.

Using a data set of 16 proteins, a neural network has been trained to predict backbone 15 N generalized order parameters from the three-dimensional structures of proteins. The final network parameterization contains six input features. The average prediction accuracy, as measured by the Pearson correlation coefficient between experimental and predicted values of the square of the generalized order parameter is > 0.70. Predicted order parameters for non-terminal amino acid residues depends most strongly on local packing density and the probability that the residue is located in regular secondary structure.