Computational and experimental functional annotation of putative ketosteroid isomerases

Abstract

Cholesterol degradation in bacteria can supply carbon and energy for the organism. One step in this pathway converts 5-androstene-3,17-dione to 4-androstene-3,17-dione, catalyzed by the enzyme Ketosteroid Isomerase (KSI). In Pseudomonas putida (Pp), PpKSI represents a classical example of electrostatic preorganization with a catalytic efficiency near the rate of diffusion; the homologous enzyme in Mycobacterium tuberculosis (Mt KSI), however, performs this reaction 1000-fold less efficiently. Two putative enzymes in Mt, Rv0760c and Rv2042c, were initially hypothesized to be Ketosteroid Isomerases. Structural and functional computation tools such as template-free modeling, ligand docking, molecular dynamics, and machine learning tools to investigate active-site residues, like Partial Order Optimum Likelihood (POOL) and Structurally Aligned Local Sites of Activity (SALSA), were used to gain insight into the function of Mt KSI. Rv0760c was predicted to have some KSI activity, with potential for Kemp Eliminase activity. Experimental kinetics assays served as validation to computational predictions and reengineering efforts found increases in catalytic rate for Rv0760c. Rv2042c showed limited similarity in computational comparisons to other KSI proteins.--Author's abstract