Structural dependencies of interresidue scalar coupling (h3)J(NC') and donor (1)H chemical shifts in the hydrogen bonding regions of proteins

Abstract

A study is presented of the structural dependencies for scalar J-coupling and the amide donor (1)H chemical shifts in the hydrogen bonding regions of proteins. An analysis of the interactions between the donor hydrogen and acceptor oxygen orbitals in an N-H...O=C moiety suggests that there are three major structural factors for (15)N-(13)C coupling across hydrogen bonds: (1) the H...O' internuclear separation r(HO)('), (2) the H...O'=C' angle theta(2), and (3) indirect contributions involving the oxygen loan pair electrons should lead to a dependence on the H...O'=C'-N' dihedral angle rho. Density functional theory (DFT) and finite perturbation theory (FPT) were used to obtain the Fermi contact (FC) contributions to interresidue coupling in formamide dimers with systematic variation of these structural parameters. The computed (h3)J(NC)(') exhibit good correlations with cos(2) theta;(2) combined with an exponential dependence on r(HO)('). The correlation is further improved by including a dependence on the dihedral angle rho. For each of the 34 H-bonds having observable interresidue coupling in the immunoglobulin binding domain of streptococcal protein G, a formamide dimer was generated from the crystallographic structure with energy-optimized donor H-atom positions. The computed coupling constants are in reasonable agreement with the experimental, and there are excellent correlations with the simple equations involving theta;(2) and r(HO) if alpha-helix and beta-sheet regions are treated separately. This dichotomy is removed by introducing the dependence on the dihedral angle rho. Justification for the use of formamide dimers is provided by almost identical interresidue coupling constants for larger sequences extracted from the X-ray structure. The amide donor (1)H chemical shifts, which were based on DFT and GIAO (gauge including atomic orbital) methods, are in poorer agreement with the experimental data but exhibit excellent correlation with r(HO)('), theta(2), and rho.