A new model for mapping the peptide backbone: predicting proton chemical shifts in proteins

This paper describes a methodology that correlates experimental chemical shifts (at the alpha proton) of proteins with their geometrical data (both dihedral angles and distances) obtained from 13 representative proteins, which are taken from the Protein Data Bank (PDB) and the BioMagRes Data Bank (BMRB). To this end, the experimentally measured proton chemical shifts of simple amides have been correlated with DFT-based calculated structures, at the B3PW91/6-31G* level. This results in a series of mathematical relationships, which are extrapolated to the above-mentioned proteins giving rise to a modified equation for such skeleta. It is relevant to note that the equation is also supported by a clear comparison with NMR data of a protein beyond the chosen set, such as insulin, even with lower errors. The model also relates the dependence of chemical shifts on hydrophobic and anisotropic effects at the amino acid residues.