Using Molecular Networking for Microbial Secondary Metabolite Bioprospecting

Abstract

The oceans represent an understudied resource for the isolation of bacteria with the potential to produce novel secondary metabolites. In particular, actinomyces are well known to produce chemically diverse metabolites with a wide range of biological activities. This study characterised spore-forming bacteria from both Scottish and Antarctic sediments to assess the influence of isolation location on secondary metabolite production. Due to the selective isolation method used, all 85 isolates belonged to the phyla Firmicutes and Actinobacteria, with the majority of isolates belonging to the genera Bacillus and Streptomyces. Based on morphology, thirty-eight isolates were chosen for chemical investigation. Molecular networking based on chemical profiles (HR-MS/MS) of fermentation extracts was used to compare complex metabolite extracts. The results revealed 40% and 42% of parent ions were produced by Antarctic and Scottish isolated bacteria, respectively, and only 8% of networked metabolites were shared between these locations, implying a high degree of biogeographic influence upon secondary metabolite production. The resulting molecular network contained over 3500 parent ions with a mass range of m/z 149-2558 illustrating the wealth of metabolites produced. Furthermore, seven fermentation extracts showed bioactivity against epithelial colon adenocarcinoma cells, demonstrating the potential for the discovery of novel bioactive compounds from these understudied locations.

Keywords: Antarctica; bacteria; bioprospecting; molecular networking; secondary metabolites.

Figure 1

Neighbour-joining tree based on almost complete 16S rRNA gene sequences, showing the phylogenetic relationships between 81 Scottish and Antarctic bacterial isolates (KP011, KP122, KP130 and KP133 were excluded from the tree) with NCBI accession numbers in brackets. Bootstrap support >50% based on a neighbour-joining analysis of 1000 resampled datasets is shown. Scale bar, 0.1 substitutions per nucleotide position. * strains subjected to metabolite comparison through molecular networking. Branches are colour-coded according to their isolation location (Scotland, Green; Antarctica, Blue) for each bacterial isolate.

Figure 2

Molecular network of 3558 parent ions produced by 38 bacterial strains. Nodes colours are based on bacterial isolation location, where nodes representing parent ions are produced by strains isolated from Antarctic sediment, Scottish sediment or both. Grey nodes represent media components and the node size reflects the number of strains that produced each parent ion. Nodes highlighted in coloured boxes represent parent ions that are identified as the previously discovered metabolites (A) desferrioxamine E (B) surfactin C14 and (C) surfactin C15.

Figure 3

Surfactin C14 (A) MS/MS parent ion fragmentation for this study (B) MS/MS parent ion fragmentation for the molecular networking standard (C) major fragment ion; surfactin C15; (D) MS/MS parent ion fragmentation for this study; (E) MS/MS parent ion fragmentation for the molecular networking standard; (F) major fragment ion and desferrioxamine E (G) MS/MS parent ion fragmentation for this study; (H) MS/MS parent ion fragmentation for the molecular networking standard; (I) major fragment ions.

Figure 4

Relative proportions of location-specific parent ions identified by molecular network analysis (Figure 2).

Figure 5

Molecular network with 315 parent ions produced by ten Bacullus licheniformis strains. Nodes representing ions produced by strains isolated from Scotland (KP30, KP30, KP56, KP62, KP93) and Antarctica (KP120, KP121, KP126, KP126, KP128) are green and blue, respectively. Nodes representing parent ions produced by strains isolated from both Scotland and Antarctica are orange and media components are grey. The node size reflects the number of strains that produce each parent ion from the smallest node (strain specific) to the largest node (10 strains).

Figure 6

Fragmentation of parent ions that were tentatively assigned as antimycins showing fragmentation of (a) parent ion m/z m/z 521.24854 (Antimycin A3 or A7, C₂₆H₃₆N₂O₉) and (b) parent ion m/z 493.21734 (Antimycin A5, C₂₄H₃₂N₂O₉). MS² spectra (i) and MS³ (ii) are annotated to illustrate the predicted fragmentation patterns of parent antimycins.

Figure 7

Micrographs (taken with EVOS XL Core Imaging system, 100× magnification in phase contrast) of Caco-2 cells seeded in 96 microplate wells after 48 h of exposure to metabolite extract of strain KP087 (A) at 10 µg/mL (viability test showed 37% viability for this well) and medium blank (B) at 10 µg/mL (viability test showed 100% viability for this well). Formation of apoptotic bodies can be seen in (A) as small black dots, as well as areas free of cells in the lower part of the picture as cells could no longer attach to plate. Cells exposed to medium blank (B) show full confluence where single cells cannot be distinguished.