Calculation of Structures from NMR Restraints

Abstract

When the NMR method for protein structure determination was introduced in the early 1980s the new approach met with enthusiasm amongst NMR spectroscopists, as well as scepticism and disbelief by structural biologists until the simultaneous but independent determinations of the three-dimensional structure of the protein tendamistat by X-ray crystallography [1] and NMR spectroscopy [2,3] yielded virtually identical results [4]. Since that time, NMR has become a firmly established method for determining the threedimensional structures of proteins. More than 7800 structures in the Protein Data Bank [5] of March 2009 have been determined by NMR ( Figure 5.1). This remarkable achievement would not have been possible without the development of sophisticated computational methods to compute three-dimensional protein structures from NMR-derived conformational restraints, and by increasingly automated approaches for analysing multidimensional NMR spectra.NMR structure calculations can be performed in several ways that differ essentially by the extent to which the analysis of the spectra is automated. In a basic structure calculation, all spectra are analysed by the spectroscopist who also interprets the data and provides the structure calculation program with geometric restraints in the form of allowed interatomic distance ranges, ranges of allowed torsion angle values, and Protein NMR Spectroscopy: Practical Techniques and Applications, First Edition. Edited

Recent Advances in NMR Protein Structure Prediction with ROSETTA

“…Restraints (also called "constraints" in Rosetta jargon) are used for model scoring and imply the use of an energy function. They are used to guide a structure search algorithm, such as simulated annealing, molecular dynamics, or Monte Carlo optimization, producing an ensemble of structures that best satisfy the NMR restraints [3]. However, for large proteins in solution, (>25 kDa) NMR datasets can be sparse because of low signal-to-noise ratios, low spectral resolution, or difficulties in obtaining an unambiguous assignment of the NMR signals [4,5].…”

Exploring the limits of solution-state NMR in application to challenging proteins in drug discovery

“…Despite a significant proportion of obtained structures of biomolecules deposited in PDB has been solved by NMR, upon examination of BMRB statistics (http://www.bmrb.wisc.edu/ref_info/), the following inference related to size distribution of the deposited structures can be made: 90% of the NMR derived structures are in the range of 2 kDa to 25 kDa with a maximum at 8 to 10 kDa with only a few structures with molecular weights exceeding 25 kDa (Guntert 2011). This paucity of NMR structures above 25 kDa reflects the increasing effort and cost involved in the structure determination.…”