Expansion of the Stereochemical Space of Triterpenes by Mining Noncanonical Oxidosqualene Cyclases Across the Diversity of Green Plants

Triterpenoids and steroids are structurally complex polycyclic natural products with potent biological functions, for example, as hormones. In all eukaryotes, the carbon skeletons of these compounds are generated by oxidosqualene cyclases, which carry out a polycyclization cascade to generate four or five rings with up to nine stereogenic centers in a targeted manner. The tight stereochemical control of this cascade reaction severely limits the stereochemical space accessible by known oxidosqualene cyclases. Considering that naturally occurring hormone stereoisomers have markedly different biological activities, finding ways to produce stereoisomers of triterpenes would be highly desirable to open new avenues for developing triterpenoid and steroid drugs. Here, we present a plant kingdom-wide sequence mining approach based on sequence similarity networks to search for noncanonical oxidosqualene cyclases that might produce triterpene stereoisomers. From 1,891 oxidosqualene cyclase sequences representing the diversity of green plants, six candidates were selected for functional evaluation by heterologous production in <i>Nicotiana benthamiana</i>. Of these six candidates, three produced rare or previously inaccessible triterpene stereoisomers, namely, (3<i>S</i>,13<i>S</i>)-malabarica-17,21-diene-3β,14-diol, 19-<i>epi</i>-lupeol, and a previously unknown hopanoid stereoisomer that we call protostahopenol. Site-directed mutagenesis revealed key residues important for catalytic activity. The sequence similarity network mining strategy employed here will facilitate the targeted discovery of enzymes with unusual activity in higher organisms, which are not amenable to common genome mining approaches. More importantly, our work expands the accessible stereochemical space of triterpenes and represents the first step to the development of new triterpenoid-derived drugs.