The ongoing antibiotic crisis, marked by the rise of multidrug-resistant pathogens and a sharp decline in the discovery of new antimicrobial compounds, poses a major challenge to modern medicine. Historically, microbes from the natural environment have been the cornerstone of antibiotic discovery, producing diverse secondary metabolites that have transformed clinical therapeutics. These bioactive molecules are encoded within biosynthetic gene clusters (BGCs)—co-localized genes that direct the production of secondary metabolites as ecological defense and adaptation strategies—an insight that led to the advent of genome mining and natural products discovery efforts. Yet, despite major advances, our understanding of how environmental context shapes biosynthetic potential remains limited.
Here, we address this gap through a global, systematic exploration of microbial antibiotic potential across more than 4090 metagenomes spanning natural, extreme, and built environments—including the International Space Station, Gowanus Canal and Ely Copper Mine (two EPA-designated Superfund sites), hypersaline lakes, and permafrosts with more than 37,288 BGCs belonging to more than 13,000 unique gene cluster families. By integrating large-scale metagenomic analyses with genome-resolved BGC mining, we map ecological and evolutionary signatures that drive secondary metabolite diversity. Our findings reveal that microbial communities from extreme or anthropogenically impacted niches harbor distinct BGC architectures and novel biosynthetic domain combinations, reflecting unique chemical innovation strategies shaped by environmental stress.
To move beyond prediction, we also introduce a multi-omics framework integrating metabolomics with metagenomics for direct compound-level validation. By linking BGCs to observed metabolites including untargeted and previously uncharacterized metabolites, we establish a robust methodology that connects genomic potential to realized compounds. While laboratory-based extraction and isolation remain essential for full compound characterization, our approach enables confirmation of bioactive molecule presence with measurable confidence, providing a crucial validation layer between in silico prediction and wet-lab discovery. This framework also uncovered “silent” or cryptic gene clusters, which can help with discovery of new classes of antibiotics.
Together, this study establishes the first global atlas of biosynthetic capacity across environmental gradients. To our knowledge, it represents the first large-scale integration of untargeted metabolomics for all-class compound validation, uncovering the hidden biosynthetic and chemical diversity of environmental microbiomes in modern urban environments.