Search

Organisms across all domains of life have developed elegant strategies to produce specialized metabolites, also referred to as natural products, for defense, structure, and communication. Natural products provide novel inspiration for pharmaceutical development given their potent biological activities and structural diversity. Recent advancements in DNA sequencing technology and the development of predictive bioinformatics tools have unveiled a vast collection of biosynthetic machinery responsible for generating this structural diversity that may be harnessed by synthetic chemists. Many of the enzymes encoded by biosynthetic genes catalyze chemical reactions that are difficult to replicate with the same selectivity by traditional synthetic methods. An example of this discrepancy is illustrated by the current enzymatic and chemical methods known to generate a highly reactive ortho-quinone methide (o-QM) intermediate. The utility of o-QMs in total synthesis has been hampered by complications that surround the preparation of their precursors, the harsh generation methods, and poor chemo-, regio-, and stereoselective control. In contrast, multiple enzymes have been reported to catalyze o-QM formation under mild conditions with remarkable selectivity. Chapter 1 of this dissertation examines several different enzymatic strategies developed to generate o-QMs. Chapter 2 describes progress towards the biomimetic total synthesis of (–)-chlorizidine A that aims to synthetically replicate the final biosynthetic intramolecular cyclization via an o-QM intermediate. Chapter 3 reports the discovery of tetrachlorizine, a novel tetrachlorinated marine natural product, and the identification of its associated biosynthetic gene cluster. Biochemical characterization of a pivotal flavin-dependent oxidase revealed a dehydrogenation reaction that proceeds via an unprecedented o-QM mechanism. This oxidase was then repurposed to generate an isolable o-QM, an extremely rare chemical motif. Chapter 4 investigates the structure-function relationships of two homologous flavin-dependent oxidases with divergent reaction selectivities and highlights their favorable biocatalytic properties. This chapter also discloses the first reported bacterial enzymes capable of performing oxidative cyclization reactions to produce cannabinoids.

It has been shown that genomes of bacteria, especially members of the Streptomyces genus, harbor unprecedented numbers of biosynthetic gene clusters (BGCs) potentially encoding novel compounds. However, cloning and controlled expression of these large BGCs in a heterologous host require tedious optimization on a case-by-case basis. This dissertation presents a synthetic biology platform to rapidly reconstitute BGCs by refactoring and physically piecing DNA fragments together in a hierarchical manner. The two core technologies are (i) a high-throughput DNA assembly pipeline for high GC organisms and (ii) synthetic genetic elements to control gene expression in Streptomyces. As a proof of concept, a small library of synthetic gene clusters was constructed to encode ent-atisanoic acid, a late-stage intermediate of the neuroprotectant serofendic acid. We successfully controlled the relative expression level of individual genes, identified the tailoring enzyme required for the oxidation as well as demonstrated the utility of this DNA assembly pipeline. Next, we rationally optimize isoprenoid biosynthesis by perturbing relative expression of eight enzymes in the methylerythritol phosphate (MEP) pathway. One of the Design of Experiment (DoE) methods called Plackett-Burman design was used to guide the optimization effort for this eight-gene system. A five-level Plackett-Burman design was used to guide the design of 125 unique synthetic gene clusters encoding the MEP pathway, which was required to fully screen the effects of expression of each gene on the output measured by isoprenoid titer. In the eight-gene pathway, each gene has one of five expression levels. Total screening of the entire pathway variants has revealed a surprising degree of robustness in actinobacterial secondary metabolism. In sum, the DNA assembly pipeline will become a powerful tool to fuel future rational optimization efforts of multi-gene systems, including large BGCs.

Natural products, also called secondary or specialised metabolites, are compounds that are produced by cells and are intended to have an effect outside of the producer. These compounds often have potent biological activities, and humankind has been using them as medicines in their crude forms for centuries. Even today, natural products are an important source and inspiration for drug development. An especially rich source of novel natural products are marine sponges. For many compounds isolated from sponges, though, it has been shown by genome sequencing and other methods that they are produced by symbiotic microorganisms. The genes necessary for the production of secondary metabolites are often clustered together in microorganisms, a phenomenon generally called a biosynthetic gene cluster (BGC). This clustering allows for the bioinformatic identification of BGCs, so-called genome mining, and subsequent investigation of natural product biosynthesis. The focus of this thesis lies on cyanobacteria. These fascinating microorganisms are not only exceptional producers of specialised metabolites, they have also been shown to occasionally produce compounds with high similarity to natural products isolated from sponges. The goal of this thesis was to further explore the biochemical potential of cyanobacteria in natural product research. This was done by either identifying new natural products and their biosynthetic gene clusters (chapter 2, 4 and 5), by characterising novel biochemistry and new enzymes involved in natural product biosynthesis (chapter 2, 3 and 5) or by testing the potential of cyanobacteria to heterologously express biosynthetic gene clusters (chapter 5). Chapter 2 focuses on the discovery and structure elucidation of new cytotoxic cyanobacterial meroterpenes termed merosterols. Their structure was determined with a combination of nuclear magnetic resonance spectroscopy (NMR), prediction of 13C chemical shifts, and a comparison of measured and predicted circular dichroism spectra. Additionally, the biosynthetic gene cluster responsible for merosterol production was identified. The substrate of the terpene cyclase was investigated by testing different surrogates with an intrinsic tryptophan fluorescence assay. The terpene cyclase was investigated in vitro and the structure of the formed product was determined by NMR analysis. The molecule consists of a polycyclic sterol-like terpene part annelated to an aromatic benzoic acid. A notable feature is the adopted opposite stereochemistry of the one of common steroids. Known terpene cyclases of the same class consist of two or three domains, and are often membrane bound proteins. The merosterol terpene cyclase on the other hand is soluble and smaller than the known enzymes. Structural 10 modelling proposed a minimalistic monodomain architecture, which was confirmed with an X-ray crystal structure of the enzyme, as further described in chapter 3. In chapter 4 the biosynthetic potential of cyanobacteria is further explored. Analysis of cyanobacterial genomes revealed a gene cluster putatively encoding a nonribosomal peptide synthetase/polyketide synthase hybrid accompanied by seven Fe(II)/α-ketoglutarate-dependent enzymes. Bioinformatic analysis suggested that six of them are halogenases, an unprecedented number of halogenases in a single gene cluster. Mass spectrometry-guided screening of cyanobacterial extracts led to the discovery and structure elucidation of the novel, extensively halogenated nonribosomal peptide/polyketide hybrids termed aranazoles. Chapter 5 is based on previous observations that cyanobacteria can produce similar secondary metabolites as symbionts of marine macroorganisms such as sponges. A putative gene cluster encoding genes for the biosynthesis of an indolocarbazole was identified in the genome of 'Candidatus Entotheonella serta'. It contains a glutathione sulphur transferase, a class of enzymes that has not yet been described in the context of indolocarbazoles. Investigation of the presence of homologous genes in cyanobacterial genomes led to the identification of similar gene clusters in two Fischerella species. Even though no activity was detected for the glutathione sulphur transferase so far, it was confirmed that the gene cluster is indeed responsible for the biosynthesis of indolocarbazoles. Detected products included the already known indolocarbazole arcyriaflavine A and a novel compound with two additional oxygens. This product can be explained by the presence of an additional cytochrome P450 enzyme encoded in the cyanobacterial gene clusters. Oxygenations of indolocarbazoles catalysed by P450s have only been described in the context of environmental DNA libraries. Otherwise only hydroxylations catalysed by FAD-dependent enzymes have been described so far for these compounds. All in all, this thesis provides further evidence that cyanobacteria are not only capable of producing unique novel secondary metabolites, but also have a repertoire of natural products closely resembling compounds, which are produced by symbionts of marine macroorganisms.

Traditionally, natural products have been isolated by bioactivity-guided fractionation. This strategy, however, led to the frequent re-isolation of known metabolites. An alternative strategy for the discovery of novel specialized metabolites is genome mining. Genome mining is an in-silico natural product discovery strategy which uses genetic information to identify natural product biosynthetic gene clusters (BGCs). The identified BGCs are annotated and the information about the BGC constitution is used to predict the structure of the corresponding secondary metabolite. Based on these predictions the corresponding natural product can be isolated in a targeted fashion. Trans-acyltransferase polyketide synthases (trans-AT PKSs) are giant modular multi-domain enzymes that are responsible for the biosynthesis of complex polyketides. These bacterial assembly-lines, ubiquitously distributed across the bacterial branch of life, were shown to produce bioactive polyketides of pharmaceutical and agricultural value. Phylogenetic analysis of ketosynthase (KS) domains from characterized PKSs, responsible for polyketide chain elongation, revealed a correlation between KS domains that phylogenetically cluster together and the substrates they recognize. Based on the correlation between phylogenetic origin and substrate specificity, referred to as the trans-AT PKS correlation rule, we developed the trans-AT PKS-derived polyketide predictor (TransATor) web application for the precise predictions of trans-AT PKS-derived polyketide structures. The TransATor web-application can be used to conduct in-silico dereplication studies. The TransATor-based in-silico annotation of a BGC to a known metabolite prevents the re-isolation of known metabolites, as exemplified by the isolation of tartrolon D from Gynuella sunshinyii. Moreover, TransATor predictions were employed to prioritize isolation efforts and, most importantly, to guide the purification and structure elucidation workflow. This has led to the isolation of a novel polyketide scaffold from Brevibacillus sp. Leaf182 and a new phormidolide-like polyketide, leptolyngbialide, from Leptolyngbia sp. PCC 7375. The prediction of KS substrate specificities furthermore resulted in the revision of the biosynthetic model for oocydin from Serratia plymuthica, the proposal of a model for sesbanimide biosynthesis in Azorhizobium doebereinerae and the reinvestigation and revision of the absolute stereochemistry of phormidolide isolated from a Leptolyngbia sp. ISBN3-Nov-94-8. A previous study on misakinolide-like polyketides suggested that the corresponding PKS family has evolved through the exchange, deletion or insertion of conserved sequential arrangements of PKS modules (PKS motifs). Driven by the hypothesis that trans-AT PKSs evolved in a mosaic-like manner, through combinatorial recombination of conserved motifs from different pathways, we analyzed all 1724 trans-AT PKSs deposited in GenBank for the presence of motifs shared between different PKSs. In total, 327 conserved sequential arrangements of PKS modules were identified. Based on these conserved motifs, also reflected in the polyketide structure, we were able to isolate new polyketides with conserved substructures. These included the novel glutarimide-containing polyketide secimide from the plant pathogen Pseudomonas syringae pv. syringae and the new phormidolide, gynuellalide, from G. sunshinyii. Moreover, the identification of a conserved PKS gene fragment encoding part of the PKS responsible for the biosynthesis of the cytotoxic spliceostatin in the plant pathogen Xanthomonas campestris pv. cannabis resulted in the identification of the pathogenicity factor of the bacterium. In addition, PKSs that share conserved motifs with other orphan PKSs but do not show any conserved motifs with known polyketides are good candidates for the discovery of structural novelty in the corresponding polyketides. The analysis of conserved PKS motifs shared between different PKSs furthermore paves the way towards evolutionary inspired engineering of non-natural hybrid PKSs to produce novel polyketides. 8 A complementary strategy to genome mining is the activation of silent BGCs. Here, we studied the potential of the actinobacterial signaling metabolite hormaomycin to activate silent BGCs in the producing organism and the model strain Streptomyces coelicolor. The effect of hormaomycin on the metabolome of the producing organism, Streptomyces griseoflavus W-384 was analyzed, and compared to the metabolome of hormaomycin production deficient- and hormaomycin overproducing mutants. Two novel metabolites, xanthoductin and xanthotryptin were only produced in the hormaomycin overproducing mutant. Moreover, the production of pharmaceutically important natural products was either strongly up- or downregulated in the presence of hormaomycin. In order to study its effect on other Streptomyces, we subjected S. coelicolor, either challenged with the hormaomcyin overproducer or purified hormaomycin to matrix-assisted laser-desorption ionization time-of-flight (MALDI-TOF) imaging mass spectrometry and nanospray desorption ionization mass spectrometry (nanoDESI) coupled to molecular network analysis. These analyses revealed that hormaomycin induces the production of siderophores, antibiotics, morphogens and pigments belonging to different classes of natural products. In addition, hormaomycin was able to trigger the production of unknown metabolites. These studies show that hormaomycin can be used as a tool to activate silent BGCs and to increase the production of pharmaceutically relevant specialized metabolites. Intrigued by the chemotype triggered by hormaomycin, we were interested in identifying its molecular target. Using a mutasynthesis strategy, hormaomycin analogs with either alkyne or azide handles were generated. These analogs were coupled to different tags and the obtained probes used for pull-down assays to identify the target of hormaomcyin. These assays revealed the the F-ATP synthase as the putative target molecule. Quantification of intracellular ATP levels in the presence or absence of hormaomycin and the phenotypic comparison of the effect of hormaomycin with the decoupling agent carbonylcyanide m-chlorophenylhydrazone supported our initial hypothesis that hormaomycin exerts its effect by interacting with the F-ATP synthase. Moreover, we conducted a high-throughput bipartite interaction screen of all strains of the At-LSPHERE strain collection (50,000 interactions) of bacterial leaf isolates to identify talented antibiotic producers. Linking these data with genome mining studies (AntiSMASH and BigScape analysis), resulted in the identification of antibiotic producers harboring BGCs that encode for putative biosynthetic and structural novelty. The most promising inhibiting strain, Brevibacillus sp. Leaf182 that contains a dozen BGCs was selected for bioactivity-guided fractionation to identify the antibiotic(s) responsible for the observed inhibition of one third of all strains of the strain collection. Isolation of the metabolites responsible for the majority of the observed antibiotic activity yielded tyrocidin analogs and marthiapeptide A. To our surprise, marthiapeptide A is putatively produced via the non-ribosomal route of peptide biosynthesis, which is remarkable given that two structurally close homologs were previously reported to be ribosomally-synthesized and post-translationally modified peptides. The biosynthesis of the same class of polythiazole/polyoxazole-containing 7-8 membered cyclopeptides from two biorthogonal pathways displays an extreme case of convergent evolution. Moreover, the isolation of a novel trans-AT PKS scaffold from the same organism highlights the huge untapped biosynthetic potential of bacterial isolates from the plant phyllosphere.