Towards Classification and Functional Description of Enzymes: A case study of feruloyl esterases

The prediction of enzyme functionality from sequence or structure data remains a challenging task that can be best addressed by studying the structure-function relationships determined from previously available information. This thesis work was focused on developing a reliable classification and functional description for the feruloyl esterase (FAE) enzyme family, whose members’ possess both structural and catalytic promiscuity. To establish functional subgrouping of feruloyl esterases a combination of computational and experimental resources was used. The major challenge for FAEs, which often share little sequence similarity to each other and show varied substrate specificity catalyzing the conserved reaction involving an ester bond, is to represent the function in a computationally accessible format. For the analysis of FAEs with overlapping and unique specificity to individual substrates there is a need to capture the chemical function in terms of overall substrate specificity. To meet this requirements, the classification of FAEs was performed by incorporating the information of sequence properties, common-feature based pharmacophore models and the knowledge of active-residue constellations of the FAE binding pockets. Using machine learning techniques an automated descriptor-based classification system for FAEs was proposed that resulted into 12 FAE families. Based on catalytic residue constellations these families were sub-grouped into 32 functionally distinct sub-families. The biological relevance of the descriptor based classification system was validated with experimental data obtained from biochemical and biophysical characterization of FAEs. Challenges in the selection of the appropriate docking algorithm and scoring function combination for the prediction of substrate specificity of FAEs were addressed using molecular docking approaches. The evaluation of 88 docking algorithm-scoring function combinations from leading commercial docking programs for substrate specificity predictions revealed large differences in their performances that could be attributed to the differences in properties of the target proteins. Using the combination of in silico approaches and enzymology, structure-function relationships of FAEs were probed, especially in case of an exceptional Multiple Nucleophilic Elbowed Esterase (MNEE) from Sorangium cellulosum with four functionally distinct and catalytically promiscuous active-sites. Finally, this thesis demonstrates the application of structure-function relationship studies to obtain insights on the promiscuity of enzymes in their evolutionary path and to explain their structure-activity changes in immobilization based biosynthetic reactions