Three-dimensional localization of fluorescent proteins in living Escherichia coli

Abstract

3D localization of fluorescent proteins (FPs) in living bacteria has been challenging due to the low signal-to-background ratio of the FPs and the relatively uncertain positioning of the cells in the optical reference system. Using mother-machine microfluidic devices together with deep learning, we present an approach that enables accurate 3D localization of FPs in Escherichia coli over long periods. We describe a method to simulate ground truth training data for the deep learning network based on background models generated from experimental data. We test the method by studying how chromosomal loci are relocated in 3D over the E. coli cell cycle. Since the cells are radially symmetric, we expect the same width and height distribution of fluorophores if the 3D positions are correctly determined. We observe this pattern experimentally for all the labelled loci on the chromosome. Interestingly, some loci are located exclusively in the periphery of the nucleoid, while others are more confined to the core of the nucleoid. This method enables studying any chromosomal loci inside living E. coli cells in high-throughput.