Harnessing Enzyme Promiscuity: Beyond Lock and Key

It is widely thought that enzymes are only capable of performing highly specific reactions on specific substrates. This, however, is a dramatic simplification. In fact, enzymes are often capable of catalyzing the reaction of a range of substrates that are only similar to their native substrate. In the industrial biology community there is a growing interest in using this promiscuity to genetically engineer organisms to produce nonnative or even nonbiological products. To facilitate this work, a program known as BNICE was developed to use enzyme promiscuity to computationally predict possible enzymatic pathways that can generate a specified compound. BNICE previously required that researchers manually develop and input the reaction rules defining potential biological transformations. This is time-consuming, error-prone, and forces a researcher to study a limited set of reactions. In this work, we describe a method to automatically generate these reaction rules by mining databases of enzymatic reactions and solving a set cover optimization problem. The results of this work will assist with ongoing research into novel routes to produce commodity chemicals, biomimetic catalyst design, and the low-cost production of pharmaceuticals, and will serve as a platform for future enzymatic research.