Novel high-throughput cultivation techniques create a demand to pre-select strains for in-depth follow-up studies. We report a workflow to identify promising producers of novel natural products by systematically characterizing their metabolomes. For this purpose, 60 strains from four phyla (Proteobacteria, Bacteroidetes, Actinobacteria and Firmicutes) comprising 16 novel species and six novel genera were cultivated from marine and terrestrial sources. Their cellular metabolomes were recorded by LC-MS/MS; data analysis comprised databases MS/MS matching, in silico compound assignment, and GNPS-based molecular networking. Overall, 1052 different molecules were identified from 6418 features, among them were unusual metabolites such as 4-methoxychalcone. Only a minor portion of the 755 features were found in all phyla, while the majority occurred in a single phylogroup or even in a single strain. Metabolomic methods enabled the recognition of highly talented strains such as AEG42_45, which had 107 unique features, among which a family of 28 potentially novel and related compounds according to MS/MS similarities. In summary, we propose that high-throughput cultivation and isolation of bacteria in combination with the presented systematic and unbiased metabolome analysis workflow is a promising approach to capture and assess the enormous metabolic potential of previously uncultured bacteria.
Keywords: LC-MS/MS; difficult-to-grow bacteria; natural products; novel bacterial strains; untargeted metabolomics.