Construction of a Basis Set of Signature Pockets of an Enzyme Functional Class by Structural Alignment of Multiple Binding Surfaces: Metalloendopeptidase and NAD Binding Proteins

Abstract

To understand the structural basis of protein function and to infer the biological role of a protein, we developed an algorithm for the construction of a basis set of signature pockets that are characteristic of a protein function. The signature pockets are constructed by identifying structurally conserved surface elements across binding surfaces of the same enzyme functional class. Signature pockets are then selected to form a minimalistic basis set representing the full ensemble of surfaces that an enzyme functional class can sample. By accurately locating elements on the binding surfaces that are invariant to conformational fluctuation, the signature pockets provide information on key players in enzyme function. A collection of signature pockets form a minimalistic basis set, which can be used for protein function prediction through database search. Our approach avoids the problems when an entire active site is used as a template due to conformational changes because of the dynamical nature of protein binding events. Our approach also avoids the problems when only a few key residues are used as a structural template, which often results in numerous false positives when predicting enzyme function. In addition, our method does not assume a priori a single structural template for representation of a functional class of proteins. Instead, a minimal set of distinct signature pockets are constructed to form a basis set that is able to characterize the full ensemble of binding surfaces that are capable of the specific enzyme function. We describe in detail how this approach is applied to accurately infer functional roles of the metalloendopeptidase family, which descend from a common ancestor, and of the NAD binding proteins, which have diverse evolutionary origins.