The disulphide beta-cross: from cystine geometry and clustering to classification of small disulphide-rich protein folds.

Small disulphide-rich protein folds (SDFs) tend to have less, regular secondary structure than larger protein folds and are thus problematic in protein structure taxonomy and prediction. We report regularities for disulphide-bridged beta-sheet and for cystine clustering that are particularly relevant to such proteins. The repertoire of cystine conformations results in preferences in disulphide distribution between/within beta-sheets. For example, disulphides seldom bridge between beta-sheets with antiparallel orientation for the flanking polypeptide segments, as the separations between packed sheets are such that the only rotamers that straddle them easily are those that generally require parallel orientation. A left-handed chirality preference is described for the intervening connection for disulphides bridging between berta-strands in different sheets in such a parallel orientation. Geometrical analysis of clusters of two cystine residues has shown that closely clustered cystine residues tend to have approximately orthogonal relative orientation. A positive orientation of this type is most often accommodated by a recurrent motif of disulphide-bridged beta-sheet that we call the disulphide beta-cross. The consensus features of this motif are described. It occurs in non-homologous proteins with a variety of folds, subsuming partial similarities previously noted by several other workers. Further examples are discussed, such as a two-cystine/two-beta-hairpin assembly common to hirudin and the three-fingered toxin folds. We suggest that the consensus features enable it to act as a good folding nucleus. We classify similar three-cystine arrangements that may be described as a ladder or stack, that tend to contain a disulphide beta-cross and that recur in folds that can otherwise be quite different. The preferences for disulphide-beta-sheet distribution and for cystine clusters contribute to an array of partial similarities for SDFs, many of which incorporate the disulphide beta-cross. It is suggested that SDF taxonomy cannot properly be considered without using both the relationship between clustered cystine residues and that between cystine residues and the regular secondary structures that they connect (here, we study beta in particular). The implications for SDF classification are demonstrated.