Genome Mining Reveals trans‐AT Polyketide Synthase Directed Antibiotic Biosynthesis in the Bacterial Phylum Bacteroidetes

Complex polyketides such as erythromycin, epothilone, and FK506 greatly contribute to human health. Bacteria synthesize these natural products through large, multifunctional enzymes called modular polyketide synthases (PKSs), which use many catalytic domains in sequence to generate polyketide chains in an assembly-line-like process. Because there is usually a strict correlation between the enzymatic architecture and the structure of the metabolite, it is often possible to predict the biosynthetic product from a PKS gene sequence. This feature has been successfully exploited for natural product discovery through bioinformatic analysis of bacterial DNA sequences. Considering the impressive recent developments in the field of sequencing technologies and the concomitant explosion in genomic data, it is anticipated that genome mining will play a key role in future drug discovery programs. A crucial prerequisite for the success of this strategy is for predictions to be made at a high level of confidence for a wide range of pathways. A recent unexpected discovery by our research group and others is the presence of a novel and evolutionarily distinct group of modular PKSs. These enzymes use free-standing acyltransferases (AT) to select polyketide building blocks and to attach them to the multifunctional PKS in trans, as opposed to cis-acting textbook PKSs that carry integrated AT domains. Such trans-AT PKSs have long been overlooked, as they are rare in actinomycetes that served as initial model organisms to study polyketide biosynthesis. However, a wide range of bioactive polyketides from diverse bacteria, including the clinically used antibiotics of the mupirocin and streptogramin series, are now known to be products of trans-AT PKSs. Trans-AT pathways represent highly interesting targets for natural product discovery for two reasons: they often occur in bacteria that have not been well studied in pharmacological screening programs, and the PKSs exhibit a remarkable array of unusual enzymatic features, such as new catalytic domains, unprecedented module architectures, and non-canonical biosynthetic transformations, which often translate into unusual chemistry. Another consequence of these peculiarities is that textbook collinearity rules cannot be applied to deduce structures from DNA data. However, on the basis of functional and phylogenetic studies, we recently developed a novel predictive approach that allows the assignment of polyketides to transAT PKS sequences with high confidence. This method is based on the observation that the substrate specificity of ketosynthase (KS) domains, which catalyze polyketide chain extensions in a Claisen-condensation-like reaction, correlates with its evolution. Thus, if the position of a particular KS in a phylogenetic tree is known, it can often be predicted whether its substrate carries a b-hydroxy group, an a,b-double bond, an a,breduced moiety, or an aor b-carbon branch. In this way, structural information for an entire polyketide chain can be obtained by combining the information of all KSs present in a PKS. The utility of these trans-AT collinearity rules was demonstrated by the isolation of novel thailandamide polyketides by genome mining of Burkholderia thailandensis. Since then, it was shown that the predictive power can be continuously increased by refining the phylogenetic analysis with KSs of newly sequenced gene clusters. In this work, we demonstrate the utility of trans-AT genome mining by assigning a PKS of Chitinophaga pinensis DSM 2588, a member of the chemically poorly studied bacterial phylum Bacteroidetes, to the biosynthesis of elansolids, antibacterial and cytotoxic agents that arise from an unusual series of post-PKS modifications. The genome of C. pinensis DSM 2588, a heterotrophic gliding bacterium consisting of long unicellular filaments, was found to contain two large gene clusters that encode a hybrid non-ribosomal peptide synthetase-PKS and a trans-AT PKS system (the latter termed els PKS). Intrigued by the fact that only few compounds have been previously reported from the genus Chitinophaga, we were interested in the product of the els pathway. The cluster (Table 1) spans 75.1 kb and contains 18 genes, of which six (elsIJNOPQ) encode multimodular PKS proteins, none of which harbors AT domains. Instead, elsA and elsB exhibit similarities to free-standing AT genes of transAT PKSs. ElsA also contains an additional C-terminal oxidoreductase domain that was recently shown to catalyze enoyl reductions in trans. The els polyketide was therefore expected to contain at least one reduced moiety. Several putative proteins (ElsDEKLM) were identified that are usually involved in the generation of b-carbon branches at the growing polyketide chain. 14] Further gene products resemble 4-hydroxybenzoate synthases (ElsH), cytochromes P450 (ElsF), methyltransferases (ElsR), and proteins with unknown function (ElsC and ElsG). The six large PKS proteins contain a total of 15 modules. The N terminus of ElsI harbors a module that consists of [a] Dr. R. Teta , E. J. N. Helfrich, S. K nne, A. Schneider, Prof. Dr. J. Piel Kekul Institute of Organic Chemistry and Biochemistry University of Bonn, Gerhard-Domagk-Strasse 1, 53121 Bonn (Germany) Fax: (+ 49) 228-739712 E-mail : joern.piel@uni-bonn.de [b] Dr. M. Gurgui, Dr. G. Van Echten-Deckert Kekul Institute LIMES Membrane Biology and Lipid Biochemistry University of Bonn, Gerhard-Domagk-Strasse 1, 53121 Bonn (Germany) [c] Dr. R. Teta , Prof. Dr. A. Mangoni Dipartimento di Chimica delle Sostanze Naturali Universit di Napoli “Federico II” Via D. Montesano 49, 80131 Napoli (Italy) Supporting information for this article is available on the WWW under http ://dx.doi.org/10.1002/cbic.201000542.