Computational prediction and validation of putative ketosteroid isomerase (KSI) structural genomics proteins

Abstract

Ketosteroid Isomerase (KSI) performs an important biological function in the metabolism of many bacteria by degrading steroids. Pseudomonas putida KSI (PpKSI) is strikingly efficient and selective. Three putative KSIs identified from Structural Genomics were analyzed by THEMATICS (THEoretical Microscopic Anomalous TItration CurveS) and POOL (Partial Order Optimum Likelihood) and then characterized in vitro to determine their KSI activity. A putative KSI from Mycobacterium tuberculosis (MtKSI) was predicted to have ketosteroid isomerase‐like activity but putative KSIs from Mesorhizobium loti and Pectobacterium atrosepticum were predicted by these methods not to have such activity. Biochemical experiments reveal that the putative KSI MtKSI does indeed possess KSI activity, although with reduced efficiency compared to PpKSI. The two proteins predicted to be misannotated, MlKSI and PaKSI, exhibited no detectable KSI activity. To investigate this lower efficiency in the correctly annotated KSI, we engineered the MtKSI active site to resemble more closely that of PpKSI, and found that most of these mutations alone or in tandem significantly lowered rather than increased activity.