Antimicrobial resistance (AMR) is a global public health crisis with few options for control. As such, moving to predictive frameworks for identification of emerging bacterial strains capable of rapidly evolving AMR for early intervention is key. Although antimicrobial tolerance and persistence are thought to be precursor phenotypes for AMR, little evidence exists to support their importance in real-world scenarios. Here we leveraged national genomic surveillance data of the diarrhoeal pathogen Shigella sonnei (n=3745) to agnostically identify common genetic signatures among lineages convergently evolving toward AMR (n=15) using bacterial genome-wide association. This revealed an association of an AMR trajectory with a multi- and highly variable second copy of metG, borne on a phage-plasmid we called pWPMR2. Further bioinformatic analyses revealed that pWPMR2 was present across clinical isolates of other enteric pathogens globally, including previous major outbreaks. And, that the mechanism of bearing additional metG copies on mobile genetic elements was present across multiple bacterial phyla. Subsequent functional microbiology and experimental evolution studies revealed that the expression of additional metG, particularly the mutated version on pWPMR2, created a sub population of cells with persister phenotypes that predispose them to the evolution of resistance to third generation cephalosporins. This highlights that the provision of metG in trans predisposes bacteria to AMR with real world impacts, likely across a broad range of clinically relevant pathogens. As well as offering a warning sign for emerging AMR lineages, our approach is a timely exemplar of how genomic epidemiology frameworks can rapidly guide functional microbiology studies in the coming era of routine genomic surveillance.
