Introduction
Most clinically important natural products, including many antibacterials, antifungals, immunosuppressants and antitumor agents, are produced by complex, slow-growing microorganisms such as actinobacteria, myxobacteria and filamentous fungi. Intensive and sustained efforts since the mid-1980s have revealed the genetic basis and enzymatic logic for biosynthesis of many of these natural products1. Genes that direct the biosynthesis of such metabolites in bacteria are usually organized into clusters of polycistronic operons on the chromosomes of the producing organism. Manipulation of such gene clusters has led to the engineered biosynthesis of many new natural products1. However, the genetic intractability of the producing organism and the polycistronic operons within such clusters can present significant obstacles to genetic manipulation. Furthermore, there are often substantial challenges associated with developing scalable fermentation processes for secondary metabolite production from natural producers. Thus, the rapid growth rate and genetic tractability of E. coli make it an attractive heterologous host for the expression of natural product biosynthetic gene clusters. In this issue of Nature Chemical Biology, Watanabe et al.2 report the cloning and heterologous expression in E. coli of a monocistronic reconstituted form of the ecm gene cluster from Streptomyces lasaliensis that directs the biosynthesis of the antitumor nonribosomal peptide echinomycin.
Over the past five years, several groups have reported the production in E. coli of complex polyketide and nonribosomal peptide metabolites, biosynthetic intermediates and shunt metabolites through the expression of reconstituted partial biosynthetic gene clusters that they cloned from bacilli, gamma proteobacteria and slow-growing actinobacteria3, 4, 5, 6, 7. These studies all used E. coli strains that had been engineered to ensure that polyketide synthases (PKS) and nonribosomal peptide synthetases (NRPS) were efficiently phosphopantetheinylated, a post-translational modification required for these multienzymes to be active. In some cases, investigators also introduced genes for carboxylation of acyl coenzyme A precursors of polyketides, and other genetic changes, into the expression host. Because of the large size or number of genes expressed, in most cases they used several plasmids containing mutually compatible origins of replication to assemble the reconstituted partial gene clusters. Most of these studies used synthetic polycistronic operons under the control of an inducible T7 promoter to express biosynthetic genes from one or more compatible plasmids in the engineered E. coli hosts.
Much recent interest has focused on bioactive metabolites produced by microbial symbionts of marine animals8. The frequent inability to culture such microbes poses enormous difficulties for the sustainable and economically viable production of these metabolites. A recent report has demonstrated heterologous production of ribosomally synthesized patellamide cyclic peptide natural products by cloning and expression in E. coli of a biosynthetic gene cluster from Prochloron didemni, the unculturable cyanobacterial symbiont of the marine ascidian Lissoclinum patella9. This may pave the way for heterologous production of other complex natural products from unculturable marine bacteria, including nonribosomal peptides and polyketides, using the engineered E. coli hosts. Similarly, the heterologous expression in Pseudomonas putida and E. coli of natural and synthetic operons from slow-growing and genetically intractable myxobacteria that direct biosynthesis of the hybrid polyketide–nonribosomal peptide myxochromides and epothilones has also recently been reported10, 11.
The modular strategy described by Watanabe et al. provides a new means by which to reconstruct these biosynthetic pathways in E. coli. To reconstruct the ecm cluster on three mutually compatible plasmids, they used 16 monocistronic gene 'cassettes', each containing a biosynthetic, self-resistance or post-translational modification gene having an upstream T7 promoter and ribosome binding site, a downstream T7 transcription terminator, and compatible flanking SpeI and XhoI restriction sites (Fig. 1). This approach results in the transcription of 16 monocistronic mRNAs, thereby offering several potential advantages over previous approaches to metabolite production in heterologous hosts, which mostly rely on natural or synthetic polycistronic operons for biosynthetic gene expression3, 4, 6, 7, 9, 10. The new approach facilitates manipulation of gene clusters to examine the function of individual biosynthetic genes and engineer the production of new metabolites, by overcoming the necessity of constructing in-frame deletions within genes in operons to ensure there are no polar effects on the expression of downstream genes. It should allow rapid combinatorial assembly of different subsets of biosynthetic genes from individual clusters as well as coexpression of genes from different clusters. The authors demonstrated the power of the approach for examining biosynthetic gene function by constructing a set of expression plasmids containing only 15 of the 16 genes required for echinomycin production; the ecm18 gene cassette encoding a methyltransferase was omitted from one of the plasmids (Fig. 1). Transformation of E. coli with this plasmid set resulted in the production of triostin A instead of echinomycin. Incubation of triostin A with purified recombinant Ecm18 and S-adenosyl methionine (SAM) confirmed that it catalyzes the unprecedented conversion of the disulfide group in this metabolite to the unusual methyl thioacetal group of echinomycin (Fig. 1). This transformation is proposed to proceed via initial methylation of one of the sulfur atoms of the disulfide linkage in triostin A to yield the corresponding sulfonium ion, followed by deprotonation of the methylene group attached to the sulfonium ion to give an ylide. Subsequent rearrangement of this ylide would yield the methyl thioacetal group of echinomycin. At present, whether an ylide intermediate is in fact involved in this transformation, and the mechanism of its possible rearrangement, are unclear. The rearrangement could proceed via a concerted mechanism as proposed by the authors, but elimination of a thiolate from the ylide intermediate to yield a sulfonium ion that could undergo nucleophilic attack at the unsaturated carbon by the thiolate is an equally plausible mechanism. Further studies will be required to elucidate the mechanistic details of this fascinating transformation.
Figure 1: Strategy for reconstruction of the ecm cluster on three plasmids using monocistronic gene cassettes.
Expression of the reconstructed cluster in E. coli results in production of echinomycin (red), whereas expression of the reconstructed cluster minus the ecm18 gene results in production of triostin A (blue). The authors confirmed the function of Ecm18 as a SAM-dependent methyltransferase that catalyses the conversion of triostin A to echinomycin in vitro using the purified recombinant enzyme.
Katie Ris
Full size image (82 KB)Though the work of Watanabe et al. represents a significant advance in the development of technology for heterologous expression of biosynthetic gene clusters in E. coli, there is still a long way to go before such technology can be applied routinely for the large-scale production of complex natural products and as a general tool for characterization and manipulation of biosynthetic pathways. Isolated yields of echinomycin from the heterologous production system are relatively low (300 
g l-1) compared with those in the natural producer. Also, production of soluble, active biosynthetic proteins by expression of genes from producing organisms such as actinobacteria and myxobacteria in E. coli can often be difficult. However, the application of systematic approaches to elucidate the origins of such problems may facilitate engineering of improved E. coli strains for the heterologous production of structurally complex bioactive metabolites from slow-growing or unculturable sources.
