Unravelling the functional diversity of type III polyketide synthases in fungi

Abstract

Type III polyketide synthases (T3PKSs) are enzymes that produce diverse compounds of ecological and clinical importance. While well-studied in plants and bacteria, only a handful of T3PKSs from fungi have been characterised to date. Here, we developed a comprehensive workflow for kingdom-wide characterisation of T3PKSs. Using publicly available genomes, we mined more than 1000 putative enzymes and analysed their active site architecture and genomic neighbourhood. From there, we selected 37 representative PKS candidates for cell-free expression and prototyping with a diverse set of Coenzyme A activated substrates, revealing unique patterns in substrate and cyclisation specificity, as well as the preferred number of malonyl-Coenzyme A extensions. Using the 341 enzyme-substrate pairs generated in this study, we trained a machine learning model to predict T3PKS substrate specificity and experimentally validated it with an extended panel of non-natural substrates. We anticipate that the model will be useful for in silico screening of T3PKSs, while the insight into the product scope of these enzymes offers interesting starting points for further exploration.