Novel terpenes generated by heterologous expression of genes encoding bacterial terpene synthases in an engineered Streptomyces host

In 2008, we first described the results of a Pfam search of predicted terpene synthases in the current bacterial genome databases using a hidden Markov model (HMM) that was based on a group of aligned protein sequences of plant terpene synthases, resulting in the discovery of a large number of previously unrecognized bacterial terpene synthases. We have now extended these results using a newly trained third-generation HMM which was generated from the previously recognized 140 presumptive terpene synthases to thereby reveal 262 presumptive terpene synthases out of 8,759,463 predicted proteins from the combined public and in-house databases of bacterial genome sequences. Genes encoding numerous presumptive terpene synthase have been efficiently expressed in a heterologous host derived from an engineered S. avermitilis from which the endogenous natural product biosynthetic genes had been deleted, leading to the production and identification of numerous terpene hydrocarbons and alcohols, most of which had not previously been observed in bacterial cultures. While the majority of the terpenes so produced were identical to known fungal or plant terpenes, to date we have already identified and structurally characterized 13 entirely new cyclic terpenes, including 2 sesquiterpenes and 11 diterpenes, using the method of heterologous Streptomyces expression. These studies suggest that genes encoding terpene synthases are widely distributed in bacteria and that these terpene synthase genes represent a fertile source for discovery of new natural products.