摘要
Nuclear Magnetic Resonance(NMR) is powerful to interrogate protein structures and dynamics residue by residue at the atomic resolution but impeded in large proteins due to fast relaxation of NMR signals and difficult in backbone assignment. Though signals of 13Cα correlated triple resonance experiments are visible for medium or even very large proteins, the backbone assignment remains a great challenge because of the severe chemical shift degeneracy of 13Cα. Here, we present for the first time a back-fitting algorithm integrated with previously presented covariance NMR method to discriminate these 13Cα degenerated peaks by utilizing peak shape similarity along the common 13Cα dimension in pairs of triple resonance experiments as a new filter. Even for two fully degenerated 13Cα nuclei in spectra, the peak shapes can still be different based on their slightly chemical shift differences and local structural properties. Our algorithm identifies over 90% artificial peaks caused by chemical shift degeneracy and allows sequential backbone assignment in the covariance spectrum without precisely peak picking. The versatility and advantage of this approach is demonstrated in the backbone assignment of a 42 k Da maltose binding protein(MBP) complexed with β-cyclodextrin with a very high accuracy of 97.7% using only HN(CO)CA and HNCA experiments. Our results pave the way to study structures and dynamics of large proteins residue by residue.
Nuclear Magnetic Resonance(NMR) is powerful to interrogate protein structures and dynamics residue by residue at the atomic resolution but impeded in large proteins due to fast relaxation of NMR signals and difficult in backbone assignment. Though signals of 13Cα correlated triple resonance experiments are visible for medium or even very large proteins, the backbone assignment remains a great challenge because of the severe chemical shift degeneracy of 13Cα. Here, we present for the first time a back-fitting algorithm integrated with previously presented covariance NMR method to discriminate these 13Cα degenerated peaks by utilizing peak shape similarity along the common 13Cα dimension in pairs of triple resonance experiments as a new filter. Even for two fully degenerated 13Cα nuclei in spectra, the peak shapes can still be different based on their slightly chemical shift differences and local structural properties. Our algorithm identifies over 90% artificial peaks caused by chemical shift degeneracy and allows sequential backbone assignment in the covariance spectrum without precisely peak picking. The versatility and advantage of this approach is demonstrated in the backbone assignment of a 42 k Da maltose binding protein(MBP) complexed with β-cyclodextrin with a very high accuracy of 97.7% using only HN(CO)CA and HNCA experiments. Our results pave the way to study structures and dynamics of large proteins residue by residue.
引文