Computational analysis of peptide and protein structure using data from NMR spectroscopy

Paul Barrett Laub, University of Pennsylvania


Structural characterization of several peptides and a protein was done over a wide range of detail by analysis of data from NMR spectroscopy. Particular emphasis was placed on efficient methods capable of obtaining structural detail appropriate to the biological questions asked. A model of the protein folding intermediate of horse cytochrome c was created as a pair of peptides corresponding to the C- and heme-bound N-terminal helices of the protein. Measurements of $\sp1$H T$\sb1$ relaxation rates were made to locate residues in the C-peptide closest to the heme in the peptide complex. A strong correlation was found between relaxation rate enhancement of protons and their distance from the heme in cytochrome c suggesting that the disposition of helices in the peptide complex is similar to that in the intact protein. Restrained molecular dynamics (r-MD) with distance restraints derived by qualitative interpretation of NOESY cross peak intensities was used to define the conformations of two peptides: CHIGPGRAFC, in its linear and disulfide-bridged form, and Ac-YTLDADF, when bound to the R1 subunit of ribonucleotide reductase. In the first peptide, subpopulations of $\beta$-turns at GPGR and at GRAF were found for the cyclic form in agreement with the temperature-dependence of amide proton chemical shift. No nonrandom structure was found for the linear form. Using transferred NOSEY spectroscopy, conformations of the backbone and sidechains of Ac-YTLDADF were defined well. A turn at residues TLDA was found; however, a bridging hydrogen bond is distinctly absent. Variation in $\sp1$H chemical shift in ubiquitin with the substitution of tryptophan for phenylalanine 45 indicates that only immediately adjacent residues are perturbed. Exploiting this, r-MD was used to refine the F45W structure locally, starting from the crystal structure of wild type ubiquitin and employing stochastic boundaries so that regions far from the site of the mutation could be held fixed. To determine the influence of distance bounds derivation on the refined structure, tandem refinements were done with restraints obtained qualitatively and by use of complete relaxation matrix analysis. Though both methods produced structures with very similar conformations, structures produced by the latter method had higher accuracy and precision.

Subject Area


Recommended Citation

Laub, Paul Barrett, "Computational analysis of peptide and protein structure using data from NMR spectroscopy" (1995). Dissertations available from ProQuest. AAI9532229.