Model-building of neurohypophyseal hormones

An attempt is made to construct models of the polypeptide backbone of the neurohypophyseal hormones. Using the sequences of six hormones and the empirical statistical approach (Kabat & Wu, 1972; Wu & Kabat, 1973), an initial selection is made of a set of φ, ψ angles satisfying all six sequences from a 20 × 20 × 20 table of φ, ψ values of tripeptides in 11 known proteins. Because of insufficient data at several positions, φ, ψ angles in the α-helical and β-sheet domains were considered as possible alternatives and the final selection was made to bring the disulfide bond as close as possible (model I). These initial values are then subjected to an energy minimization to refine the model; reasonable agreement with nuclear magnetic resonance data for possible φ values is obtained. Model II is constructed to be consistent with the available nuclear magnetic resonance data and the suggestion of a sequential Cu²⁺-binding site. Both models contain similar structural features and may be converted to one another with rotation of only a few angles. Interactive computer graphics was applied to construct model II, and its use in problems in this type is described. The relation of the experimental data to the various models that have been proposed for the neurohypophyseal hormones is discussed and an experiment is suggested that may permit a choice between them.