High-throughput, single-cell-encoded colony imaging reveals infection-linked heterogeneity

Abstract

Phenotypic heterogeneity within bacterial populations shapes adaptation, antibiotic tolerance, and infection dynamics, yet colony-level quantification remains difficult at scale and on dark/selective media. We present HeteroTyper, an automated, high-throughput time-lapse platform that measures colony phenotypes across >100 plates per run on both transparent and dark media. HeteroTyper includes a ready-to-use computational pipeline that can be deployed on common plate-imaging setups and across agar types, delivering appearance times as a proxy for lag times, growth trajectories, colony size, intensity, and morphology from routine image series. Automated counts closely matched manual counts, supporting accuracy and scalability. Using Yersinia pseudotuberculosis as a model, HeteroTyper resolves phenotypic heterogeneity in vitro, with early and late phases showing broader appearance time distributions than mid-log phases. It also reveals organ-specific heterogeneity in vivo, with mesenteric lymph node isolates displaying the broadest appearance time distributions compared to spleen and liver. Appearance time was negatively associated with final colony size and intensity, linking growth initiation to downstream colony outcomes. Together, these distribution-level measurements indicate that bacterial adaptation during infection is compartmentalized by tissue microenvironments and prior history, and establish HeteroTyper as a practical, generalizable route to quantify infection-relevant phenotypic diversity on solid media at routine, high throughput.