Metabologenomics: Discovery of new natural products and their biosynthetic gene clusters by genome-informed metabolomics

Over the past decade, the genomics revolution has provided a glimpse into the vast, untapped metabolic potential of microbial genomes. Simultaneously, the field of metabolomics, propelled by advances in LC-MS and informatics, now allows for high-throughput, semi-quantitative characterization of metabolites. Focusing on actinomycetes we present a new approach, metabologenomics, marrying metabolomics and genomics to connect exported metabolites to their biosynthetic gene clusters. With the high mass accuracy afforded by FT-Orbitrap instrumentation (< 3 ppm), 2,521 metabolite components were identified from the extracts of 178 actinobacterial growths. Of these, 110 were confidently identified as known natural products by searching an aggregated database of 9,817 actinomyctete natural products. Detected metabolites and families of related gene clusters (GCFs) were used for a binary correlation scoring system, to evaluate the likelihood that a cluster could be responsible for the production of a particular secondary metabolite. Output scores ranged from 0 to over 300. Known GCF-metabolite pairings were used to calibrate score interpretation, with 27 known gene cluster families correctly paired with published biosynthetic clusters: these included oxyetracycline (score = 270), actinomycin D (score = 204) and rimocidin (score = 210). From the highest scoring pairs, we characterized a new metabolite and named it rimosamide (score = 264).