posted on 2024-07-13, 05:46authored byMitchell Polley
The engineering of catalytic function or substrate specificity in enzymes has been an interesting field of Protein Science over the past decade. There have been many successful examples, with the publication of several proven stochastic stratagems for doing such. Despite this, it is apparent that there is an almost complete lack of generic tools for the rational redesign of functional protein sites. In this context, a survey of known high-resolution protein crystal structures was undertaken, focussing on those protein structures for which the identities of functional residues could be ascertained. The survey resulted in a database of over one hundred thousand amino acid residues and a second database of amino acids of known, specific function. The distributions of side-chain and main-chain dihedral angles were calculated and analysed statistically. Particular emphasis was placed on cross comparison of functional residues against non-functional, or structural, residues. From this investigation emerged a new set of definitions for rotameric nomenclature, based-upon the empirical data obtained. Using the new empirical definitions, it was ascertained that the distributions of dihedral angles among functional residues often differ from those distributions observed in structural residues. Subsequent analysis of the data also revealed that catalytic residues that perform similar roles in their respective proteins often show conserved dihedral geometries, to such a remarkable degree that family-like clustering could be observed. Compilation of the empirical rotamer definitions revealed that protein geometry, long viewed as being of infinite complexity, could be approximated by just under 1,500 conformational states. Backbone dihedral angle distributions were utilised for the creation of a similarity index to compare the similarities of observed backbone distributions between amino acids of differing identities. This index was later derived into a homology matrix, which can be employed for sequence alignment and homology-based structural alignment with a bias towards similarities of permitted backbone geometries. It was also found that the backbone similarity indices also displayed a correlation to the genetic code, with amino acids of a single point mutation difference in their coding triplets displaying a propensity for high backbone similarity. Overall the study generated several generic tools that will be invaluable for rational engineering of protein function and specificity.
History
Thesis type
Thesis (PhD)
Thesis note
Submitted for the Degree of Doctor of Philosophy, Swinburne University of Technology, 2002.