Some observations on peptide chain models in relation to crystallographic data for gramicidin B and insulin

Abstract

As chemical data for protein molecules accumulate, it becomes clear that any theory of the configuration of protein chains has to be integrated with evidence of considerable chemical complexity, both in side-chain distribution, interaction and cross-linking. We have recently considered chain models in relation to two crystal structures, gramicidin B and insulin, for both of which chemical and crystallographic evidence is more than usually detailed. Gramicidin B is interesting as a cyclic peptide of size intermediate between amino-acids and proteins (Dubos 1939; Gregory & Craig 1948). A combination of chemical and crystallographic data (see table 8) suggests that the molecule has a weight 3800 and contains about 32 residues, two of which are ethanolamine and some of which are D-amino acid residues. Further, the intensity distribution of X-ray reflexions from crystals of this peptide strongly suggests that the molecule consists of two chains, anti-parallel with one another and linked end to end, these molecules being packed in a roughly hexagonal array. This is illustrated by the calculation of two possible electron density projections shown in figure 39. The chain structures seen here have dimensions corresponding to chain folding of the same order of magnitude as that proposed in α -keratin, 14 to 15 residues in about 22 Å, or 1 residue in 1⋅4 to 1⋅6 Å. It is possible to construct models of molecules with these dimensions, based either on the α -helix model proposed by Pauling, Corey & Branson (1951), or folded chains of the Astbury type (1941), provided that considerable deviations are allowed for residues linking the two chains and for the ethanolamine residues. Such deviations do not apparently prevent the structure from having a marked similarity, from an X-ray scattering point of view, to α -folded synthetic peptides (figure 40).