Peptide models XXXII. Computed chemical shift analysis of penetratin fragments

Abstract

Connections between the three-dimensional structure and the computational chemical shifts were investigated for the third helix of the Antennapedia homeodomain (called penetratin). Since the whole peptide cannot be studied at an acceptable ab initio level of theory, this molecule was disintegrated into fragments (mono-, tri- and octapeptides of the N- and C-terminus). Chemical shielding anisotropy tensors have been determined using the GIAO-RHF formalism employing four different basis sets: 3-21G(d), 6-31+G(d), 6-311++G(d,p) and 6-311+G(2d,p) for the one-residue-long fragments of penetratin. For three-residue-long fragments, as well as for the N- and C-terminal octapeptide part of penetratin, the 6-31+G(d) basis set was used. In examining the computed chemical shifts, we have concluded that they depend significantly on the backbone conformation of the amino acid residues, as expected. Correlation of the chemical shift values of small split-valence basis set (3-21G(d)) and those of the highest TZ2P-type basis set (6-311+G(2d,p)) is quite good for C α, C′ and H α and moderately significant for C β, N and NH nuclei. Thus, a small split-valence basis set may also be successfully used for the chemical shielding calculations of longer peptides. We have also noticed that the differences between the chemical shifts of all 16 amino acid residues remain almost the same at each basis set as function of the size of the theory set for Cα, Hα, N and C′ nuclei. The conformation or constitution of the residue influences this only to a small degree. The chemical shifts of the mono- and tripeptides were compared to the N- and C-terminal octapeptides, which represent longer and more ‘realistic’ systems. We have concluded that the mono- and tripeptide fragments can be used reliably to ‘predict’ the chemical shielding values of the octapeptides only for Cα, Hα and Cβ nuclei because their N, NH and C′ chemical shifts do not appear to model the longer system well. We have noticed that N and NH data of the tripeptides are significantly closer to the values of the N- and C-terminal octapeptide than the chemical shieldings of monopeptides. In contrast, the Cα, Hα, C′ and Cβ chemical shifts of the octapeptides do not have an unambiguous preference for values similar to the tripeptides or monopeptides. All experimental 1H chemical shift values of penetratin [J. Pept. Sci. 8 (2002) 151] were compared with their computed counterparts (N-, C- and (C−1)-terminal octapeptide). Since penetratin was composed of 13C and 15N at natural abundance levels, the latter types of shielding data are not available for comparison. By examining experimental and theoretical Hα chemical shift data, we have concluded that the correspondence of these values is moderate. The chemical shielding values of penetratin were also compared with values in known databases, and we achieved better results. However, the databases could be used only for the four common backbone conformers (α L, β L, γ L, ε L). In contrast, the chemical shift can be calculated for all backbone conformers. Furthermore, we believe that improvements to the computational method will allow the calculated chemical shielding values to approach experimental values.