Backbone torsional potential functions for rotations about N--Calpha and Calpha--C bonds in dipeptide model systems in relation to nuclear magnetic resonance and infra red spectral data.

Abstract

A comparative study of the conventional three‐fold torsional potential functions (referred to as “Old V”), the potential function having zero φ barrier and a twofold ψ‐barrier, proposed from this laboratory (“Int V”), and those developed recently in this laboratory from the study of dipeptide conformations in protein crystal structures (“New V”) has been carried out by comparing theoretically calculated and experimentally observed NH‐CαH coupling constants from n.m.r. studies and the relative occupancy of hydrogen‐bonded conformations from i.r. studies of dipeptide model systems. It is observed that no fully satisfactory agreement with both n.m.r. and i.r. data is possible in any one of the three functions. For the hydrogen bond‐disrupting solvent DMSO, (NH‐CαH) coupling constant values were predicted equally well in all three. However, only for New V are values of the constants A (= 8.62), B (= ‐2.33), C (= 1.51) as obtained by the least‐squares method in the expression J(θ) = A cos2θ+ B cos θ+ C sin2θ such that J‐values in excess of 10Hz, observed experimentally in some cases, are permissible. However, the same potential function yields unsatisfactory agreement in the case of hydrogen bond‐promoting solvent CCl4. This was interpreted to be because the potential function concerned was obtained from data in solid state where solvent interaction is absent.In view of the lack of satisfactory agreement with the solution data for all three potential functions, it was concluded that solvent effects other than the promotion of intramolecular hydrogen bonding, which were ignored in the present study, must be taken into account for better prediction of experimental data from theoretical potential functions.