NMR assignments of 15N sparsely labeled large and glycosylated proteins
Nkari, Wendy Karwitha
MetadataShow full item record
Assignment of protein NMR resonances has historically been done using triple resonance techniques that require samples uniformly labeled with 15N, 13C, and sometimes 2H. However, particular proteins, like glycosylated mammalian proteins, cannot usually be prepared using bacterial hosts needed for facile expression with such labeling. Proteins >40 kDa and lacking 2H enrichment suffer from resonance overcrowding and signal loss during coherence transfer steps. Sparse labeling, where a single isotopically labeled amino acid (usually 15N) is used, presents an alternative that allows economical expression in non-bacterial hosts and provides improved resolution. To exploit the well-resolved 15N-1H amide cross-peaks, a new resonance assignment strategy involving amide hydrogen-deuterium exchange has been developed. First amide exchange rates for a continuously exchanging native protein sample are determined from sequential HSQC spectra. A time series of aliquots is simultaneously sampled from an identical exchanging sample, rapidly quenched and denatured, followed by examination using HSQC spectra obtained under slow back-exchange conditions. Exchange rates extracted from loss of intensities are then used to pair cross-peaks in the native and denatured spectra. One approach to identifying cross-peaks in the denatured spectrum uses a digested fully protonated protein sample. The peptides are HPLC separated and MS sequenced, and HSQC spectra of the labeled peptides are obtained. Cross-peaks of the separated peptides and the fully denatured protein overlay, allowing sequential assignment of the denatured spectrum and transfer of those assignments by exchange correlation to the native spectrum. In an effort to eliminate digestion and separation of peptides, i-1 NOEs for the labeled residue in the intact denatured protein have been used to assign the preceding residue using fingerprint side-chain patterns. This additional information limits possible positions for the labeled site in the protein sequence. As a supplementary measure, chemical shift prediction methods are used to assign resonances. The combined strategy has been successfully applied to galectin-3 (16 kDa), to produce assignments that agree with those determined by traditional approaches. The same approach has been applied to a larger (18 kDa) disulfide cross-linked protein, the dendritic cell-specific intracellular adhesion molecule-3-grabbing nonintegrin, and to an uncharacterized glycosylated protein, fucosyltransferase III (37 kDa).