Bifurcation drives the evolution of assembly-line biosynthesis

Abstract

Reprogramming biosynthetic assembly-lines is a topic of intense interest. This is unsurprising as the scaffolds of most antibiotics in current clinical use are produced by such pathways. The modular nature of assembly-lines provides a direct relationship between the sequence of enzymatic domains and the chemical structure of the product, but rational reprogramming efforts have been met with limited success. To gain greater insight into the design process, we wanted to examine how Nature creates assembly-lines and searched for biosynthetic pathways that might represent evolutionary transitions. By examining the biosynthesis of the anti-tubercular wollamides, we uncover how whole gene duplication and neofunctionalization can result in pathway bifurcation. We show that, in the case of the wollamide biosynthesis, neofunctionalization is initiated by intragenomic recombination. This pathway bifurcation leads to redundancy, providing the genetic robustness required to enable large structural changes during the evolution of antibiotic structures. Should the new product be non-functional, gene loss can restore the original genotype. However, if the new product confers an advantage, depreciation and eventual loss of the original gene creates a new linear pathway. This provides the blind watchmaker equivalent to the design, build, test cycle of synthetic biology.

Fig. 1. Intermediate chemotypes in the evolution of natural product biosynthetic gene clusters (BGCs).

a A cartoon demonstrating the three main models of gene cluster evolution depicting the transition of a fictitious BGC from chemotype A to chemotype B (coloured circles). Evolutionary processes are represented by dashed lines. b The structural diversity of the wollamides and desotamides. The d-ornithine and glycine residues, which define the wollamides and desotamides, respectively, are highlighted. The table on the right-hand side shows the variable positions of the various wollamide and desotamide congeners (NFK N-formyl kyunerine, Wol wollamide, Dsa desotamide).

Fig. 2. The genetic basis for bifurcated biosynthesis of desotamide and wollamide by Streptomyces. sp. MST-110588.

a The architecture of the desotamide producing dsa BGC from Streptomyces scopuliridis SCSIO ZJ46. b The architecture of the wollamide and desotamide producing wol BGC from Streptomyces sp. MST110588. c The proposed biosynthetic pathway of the wollamides and desotamides. The first three condensations are catalysed by WolI and WolH. The final two condensations are catalysed by WolG1 to produce desotamides or WolG2 to produce wollamides.

Fig. 3. Heterologous production of the wollamides by overexpression of wolG2.

a The extracted ion chromatograms (EIC) for masses corresponding to desotamide A ([M + H] = 697.40) and wollamide A ([M + H] = 754.46) from wild-type Streptomyces sp. MST-110588 producing both compounds. b The EICs of desotamide A and wollamide A from Streptomyces sp. MST-70754 producing desotamide A only. c The EICs of desotamide A and wollamide A from Streptomycessp. MST-70754/pBO1-wolG2 producing both compounds. Schematic representations of the assembly lines and overexpressed genes are also shown. (dsa desotamide, wol wollamide).

Fig. 4. Evidence for the intragenomic recombination of adenylation domains.

a DNA phylogeny of the external region of the adenylation domains, showing the close relationship of the major parent, wolG2 A2, and the recombinant, wolG1 A2. b DNA phylogeny of the internal region of the adenylation domains, showing the close relationship of the minor parent, rmoG A, and the recombinant, wolG1 A2. c Nucleotide identity between the three adenylation domains based on a 50 nucleotide sliding window. d Topology of an adenylation domain based on ref. , showing the predicted recombination sites Y116 and G343. Features are colour-coded according to panel a.

Fig. 5. Substrate specificities of wild-type and recombinant adenylation domains.

a Relative activities of adenylation domains as determined by hydroxylamine trapping assays. Each amino acid and enzyme combination was tested in triplicate (n = 3); the mean value was calculated and the standard deviation is shown. b Structural homology model of the synthetic recombinant adenylation domain. The model is colour-coded depending on the origin of the peptide sequence, pink for WolG2A2 and blue for RmoGA. The P-loop is highlighted in green.

Fig. 6. The evolution of biosynthetic assembly-lines.

a Proposed evolutionary history of the wollamide (wol) and desotamide (dsa) BGCs. b An updated model for the evolution of biosynthetic assembly-lines.