Understanding and Exploiting Enzyme Promiscuity for Metabolic Engineering

While the traditional view of biochemistry says one enzyme catalyzes one reaction, in reality enzymes are far from perfect catalysts. Many enzymes are known to be promiscuous, i.e., to react on chemically similar compounds.  To date, only a modest understanding of the extent and nature of enzyme promiscuity has been incorporated into our view of metabolism and metabolic engineering. Enzyme promiscuity represents both an opportunity and a challenge in metabolic engineering.  Enzyme promiscuity is an exciting opportunity as it provides a potential path for synthesizing chemicals that would not be possible with canonical enzymatic reactions.  We can utilize a much larger set of biochemical reactions than has been documented. However, enzyme promiscuity is also a daunting challenge.  An organism’s native metabolic network may contain undocumented side reactions known as ‘underground metabolism.’  As we engineer organisms to produce chemicals, underground metabolism can frustrate efforts in unpredicted ways by introducing reactions that reduce yields or cause toxicity, especially in engineered systems, where substrate concentrations may be much higher than evolutionary optimal concentrations promoting promiscuous reactions.

To address these opportunities and challenges, the Tyo lab has actively developed cheminformatics algorithms to understand and predict enzyme substrate promiscuity.  Our cheminformatics approach is unique and complementary to bioinformatics approaches, as we only analyze the compounds structure, not the enzyme.  We rely on digital representations of compounds to make our predictions.  Using a chemical fingerprinting representation, we have developed machine-learning algorithms that use existing enzyme/substrate information to predict reactions with new compounds.  Our tools consistently perform with high accuracy.  In parallel, we are using the Biochemical Network Integrated Computer Explorer (BNICE) to map underground metabolism.  We have predicted the products of undocumented enzymatic side reactions on the genome scale.  By knowing probable chemical structures of promiscuous products, we enable targeted metabolomics to experimentally verify underground reactions.  Once underground reactions are identified, we can intervene to increase productivity and avoid toxicity.

These computational tools will significantly empower the future of metabolic engineering.  By allowing the biosynthesis of a broad set of new compounds, we will generate new products and materials that were previously unthinkable.  Likewise, by uncovering undocumented enzymatic reactions that affect metabolism will give us evolutionary insight into the arrangement of metabolic networks and critical information to guide troubleshooting efforts in metabolic engineering efforts.