Automated Mapping of Natural Product Families Identifies Improved Antibacterials

Microbial natural products are the most valuable source of therapeutic antibacterials, featuring unique chemical structures that have arisen through eons of natural selection and environmental competition. The two most valuable classes of these natural products – the polyketides and nonribosomal peptides – are produced by massive multi-modular assembly-line enzymes whose domain combinatorialization has given rise to a plethora of diverse antimicrobial scaffolds, with entire families of evolved compounds arising from lone promiscuous assembly-lines. Though evolution has honed the potent bioactivities and structures of these metabolites towards niche-specific threats, this is often at odds with the needs for treating human diseases, wherein the abundance of a given natural product often does not correlate with efficacy towards targeted pathogens. Continued exploration of microbial secondary metabolites for improved antimicrobial agents will require systems-level interrogation of natural product families and classes to assess the breadth of nature’s evolved small molecule arsenal and selectively identify desired variants. Here, we outline an automated informatic discovery platform capable of simultaneously identifying and elucidating entire families of microbial natural products, including vast numbers of novel congeners. Through the use of database-dependent search algorithms, LC-MS/MS data from crude bacterial culture extracts was analyzed to reveal 80 new analogues of powerful antibacterial agents, guiding the isolation of novel variants with improved activity. Our results demonstrate that informatic discovery methodologies can provide a systems-level analysis of natural product taxonomy, enabling the charting and navigation of antimicrobial chemical space towards desired variants with superior pharmaceutical potential.