RESIDUES RESPONSIBLE FOR DISTINCT BIOLOGICAL FUNCTIONS ARE CHARACTERIZED BY DIFFERENT STATISTICAL FEATURES OF SEQUENTIAL AND SPATIAL NEIGHBORHOODS - A THERMOLYSIN EXAMPLE

An investigation of the functional topography of thermolysin was carri ed out using frequency analysis of its primary and tertiary structures . The statistical validity of this approach was estimated for the enzy me active site, the substrate-binding pocket, the inter-domain interfa ce and calcium-binding sites' predictions. We showed that frequency an alysis of primary structure could be employed to predict the localizat ion of contiguous parts of the inter-domain interface. The same approa ch appears to be unsuitable to a search for conformation-dependent enz yme active sites and substrate-binding pockets. In contrast, frequency analysis of the spatial neighborhood is not effective for predicting the inter-domain interface as distinct from the active site, substrate -binding pocket and calcium-binding sites. These differences should be taken into account when investigating and understanding protein struc ture-function relationships.