Calculated and experimental ¹H, ¹³C and ¹⁹F chemical shifts were compared in BKM-824, a cyclic bradykinin antagonist mimic, c[Ava¹-Igl²-Ser³-DF5F⁴-Oic⁵-Arg⁶] (Ava=5-aminovaleric acid, Igl=&alpha-(2-indanyl)glycine, DF5F=pentafluorophenylalanine, Oic=(2S,3aS,7aS)-octahydroindole-2-carboxylic acid). The conformation of BKM-824 has been studied earlier by NMR spectroscopy (M. Miskolzie et al., J. Biomolec. Struct. Dyn. 17, 947-955 (2000)). All NMR structures have qualitatively the same backbone structure but there is considerable variation in the side chain conformations. We have carried out quantum mechanical optimization for three representative NMR structures at the B3LYP/6-31G* level, constraining the backbone dihedral angles at their NMR structure values, followed by NMR chemical shift calculations at the optimized structures with the 6-311G** basis set. There is an intramolecular hydrogen bond at Ser³ in the optimized structures.

The experimental ¹³C chemical shifts at five C^α positions as well as at the C^β, C^γ and C^δ position of Ava¹, which forms part of the backbone, are well reproduced by the calculations, confirming the NMR backbone structure. A comparison between the calculated and experimental H^β chemical shifts in Igl² shows that the dominant conformation at this residue is gauche. Changes of proton chemical shifts with the scan of the χ₁ angle in DF5F⁴ suggest that χ₁=180°. The calculated ¹H and ¹³C chemical shifts are in good agreement with experiment at the rigid residue Oic⁵. None of the models gives accurate results for Arg⁶, presumably because of its positive charge. Our study indicates that calculated NMR shifts can be used as additional constraints in conjunction with NMR data to determine protein conformations. However, to be computationally effective, a database of chemical shifts in small peptide fragments should be precalculated.