Extending digital biology: Bacterial survival and morphological heterogeneity under antibiotic stress

Bacterial morphology is modified by antibiotic stress depending on the antibiotic mechanism of action, while changes in morphology can contribute to cell survival. Here we present a workflow that generates morphological signatures for the progeny of individual bacterial cells across 168 antibiotic conditions. The workflow uses stationary microfluidic droplets to encapsulate and grow bacteria and confocal microscopy to image the droplet content. A custom image analysis pipeline labels morphologies in a subset of images to train a neural network. The network yields multidimensional morphological signatures from 82,000 droplets, showing the co-existence of different morphologies even for the progeny of individual cells. The signatures differ with antibiotic type and concentration, providing a way to distinguish antibiotics by their mode of action. Combined with digital detection of bacterial survival within droplets, this workflow provides a quantitative platform to investigate antibiotic resistance and identify the antimicrobial activity of unknown compounds.