Proteomics, the three-dimensional structure determination of the non-membrane soluble proteins of a genome, is an emerging field in which nuclear magnetic resonance (NMR) spectroscopy can play a key role. Multidimensional (3D or 4D) NMR is used for solution structure determination of isotopically-labeled soluble proteins up to 50 kDa. The first step employs """"""""backbone"""""""" experiments to correlate inter- and intra-residue chemical shifts. These experiments are a bottleneck for assignment, the first step in structure determination. Adequate digital resolution must be obtained in each of the indirect dimensions, each increment typically requires several scans, and several different spectra must be acquired. Digital resolution in each dimension separately, is required for Fourier Transform (FT) analysis, as short signals give rise to broad unresolved lines in accordance with the time-frequency uncertainty principle. Most peaks must be resolved to arrive at a structure. This work proposes to develop"""""""" high-throughput 3D and 4D backbone experiments by using a different method of multidimensional NMR analysis developed at UC Irvine and called the Filter Diagonalization Method (FDM). Usable 2D 500 MHz NMR spectra of proteins have been obtained in two minutes, and a 3D spectrum obtained in less than 30 minutes, using FDM analysis. The resolution offered by these spectra is much better than that obtainable by any other known method, producing high-quality data sets from which assignments can be reliably made. These results suggest that high-throughput assignment is possible for proteins up to 200 residues, and that knowledge of the primary structure can be used to optimize the NMR experiments. In cases where a high-resolution crystal structure exists, the chemical shift assignment can be threaded over the assumed structure, allowing ligand binding and solution dynamics to be studied rapidly. In other cases, knowledge of the chemical shifts may be enough, in conjunction with computational and database approaches, to suggest a classification of the protein. Incorporation of the optimized FDM algorithm into the NMRPipe software suite, supported by NIH, will allow its use in many laboratories.
