Evolution of chemical diversity by coordinated gene swaps in type II polyketide gene clusters

Abstract

Natural product biosynthetic pathways generate molecules of enormous structural complexity and exquisitely tuned biological activities. Studies of natural products have led to the discovery of many pharmaceutical agents, particularly antibiotics. Attempts to harness the catalytic prowess of biosynthetic enzyme systems, for both compound discovery and engineering, have been limited by a poor understanding of the evolution of the underlying gene clusters. We developed an approach to study the evolution of biosynthetic genes on a cluster-wide scale, integrating pairwise gene coevolution information with large-scale phylogenetic analysis. We used this method to infer the evolution of type II polyketide gene clusters, tracing the path of evolution from the single ancestor to those gene clusters surviving today. We identified 10 key gene types in these clusters, most of which were swapped in from existing cellular processes and subsequently specialized. The ancestral type II polyketide gene cluster likely comprised a core set of five genes, a roster that expanded and contracted throughout evolution. A key C24 ancestor diversified into major classes of longer and shorter chain length systems, from which a C20 ancestor gave rise to the majority of characterized type II polyketide antibiotics. Our findings reveal that (i) type II polyketide structure is predictable from its gene roster, (ii) only certain gene combinations are compatible, and (iii) gene swaps were likely a key to evolution of chemical diversity. The lessons learned about how natural selection drives polyketide chemical innovation can be applied to the rational design and guided discovery of chemicals with desired structures and properties.

Keywords: evolution; gene cluster; natural products; polyketide.

Fig. 1.

Phylogeny of CLF protein sequences from 544 putative type II PKS gene clusters. Red numbers represent key ancestors. Leaf colors represent phylum of origin. Gene clusters clade by polyketide chain length (noted on the right).

Fig. 2.

Phylogeny of 78 CLF protein sequences from the reference set and selected orphan genes. Accessory genes identified in the same gene cluster (within 30 kb) as the CLF are shown at each leaf. Leaf colors represent phylum of origin. Node support for the CLF phylogeny is shown as Bayesian posterior probabilities.

Fig. 3.

Coevolution of type II PKS KS with partner genes. (A) Schematic illustrating two nucleotide records and the clustered (within 30 kb) KS + partner on each record. (B) KS1-KS2 pairwise amino acid identities are plotted vs. pairwise identities of a clustered partner. (C–I) Correlation of evolutionary histories of the KS with partner genes.

Fig. 4.

Evolution of type II PKS gene clusters by coordinated gene swaps. The tree traces the key ancestors from the most anciently diverged type II PKS gene clusters (Top) to the more recently diverged C20 systems, such as oxytetracycline and landomycin (Bottom). Highlighted in red are key ancestors, whose gene cluster architecture is inferred on the left. Representative polyketide structures from each clade are shown, and activity sites for KR (gold), TcmN cyclase (green), and KSIII (light blue) are shown on the chemical structures.