Investigations on the structure, biosynthesis and biology of antibacterial cyclic lipopeptides

Abstract

The study was aimed to find new antibacterial cyclic lipopeptides in <i>Pseudomonas</i> bacteria using genome mining techniques. Application of this genome-driven approach led to the <i>in-silico</i> identification of several orphan biosynthetic gene clusters in plant- and insect-associated <i>Pseudomonas</i> strains. The strain <i>Pseudomonas</i> sp. SH-C52 was chosen for an in-depth analysis due to the presence of a unique gene cluster coding for an unprecedented lipo-dipeptide, a glycosyltransferase and a novel class of FAD-dependent monooxygenase. <br /> In a first step, using gene knockout experiments and comparative metabolite profiling, it could be demonstrated that the gene cluster is responsible for the production of the known compound SB-253514 (brabantamide A), and its minor congeners SB-253517 (brabantamide B) and SB-253518 (brabantamide C). <br /> The obtained brabantamides were subsequently subjected to a broad antimicrobial screening which revealed significant calcium-independent activities towards selected clinical relevant Gram-positives, e.g. <i>Bacillus subtilis</i> 168 and <i>Arthrobacter crystallopoietes</i>. The minor metabolites brabantamides B and C exhibited a broader antibiotic spectrum including activity towards <i>B. megaterium</i> and <i>Micrococcus luteus</i>. Thereby, especially brabantamide B, which contains an unsaturated fatty acid side chain, exhibited a twofold higher antibacterial potency in comparison with brabantamide A. Thus, a correlation between the side chain and the observed activity became apparent and provided first insights into structure-activity relationships. <br /> First preliminary mode of action studies with brabantamide A involving model membrane studies could show that this compound class acts via a defined target and that the antibiotic effects are not mediated by unspecific detergent-like membrane permeabilisation. Complementary luciferase reporter gene assays indicated that brabantamides interfere with peptidoglycan or trigger stress on the cell wall. Furthermore, in cytotoxicity assays brabantamide A was proven to be not toxic to eukaryotic cell lines at antibiotically active concentrations. <br /> Due to the uniqueness of the 5,5-bicyclic carbamate structure of brabantamides, whose assembly require a massive rearrangement of the carbon skeleton of a linear dilipopeptide precursor, the second part of the study was focussed on their intriguing biosynthesis. Thus, a biosynthesis hypothesis was developed involving a Baeyer-Villiger reaction mechanism, mediated by the novel monooxygenase BraC. In a series of extensive feeding experiments employing ¹³C-labelled precursors the hypothesis was tested and finally confirmed. <br /> The third part of this study dealt with the potent antimicrobial lipodepsipeptide empedopeptin A. This lipopeptide was isolated along with two new minor congeners, empedopeptin B and C. The peptide sequence of empedopeptin A but not its cyclisation scheme could be confirmed by NMR and MS experiments. In a collaborative effort, also the specific mode of action of empedopeptin A was examined. It could be demonstrated that empedopeptin A inhibits cell wall biosynthesis through Ca²⁺-dependent complex formation, predominantly with peptidoglycan precursor lipid II, and with lower affinity also with undecaprenyl pyrophosphate and teichoic acid.