High-throughput time-resolved morphology screening in bacteria reveals phenotypic responses to antibiotics

Abstract

Image-based high-throughput screening strategies for quantifying morphological phenotypes have proven widely successful. Here we describe a combined experimental and multivariate image analysis approach for systematic large-scale phenotyping of morphological dynamics in bacteria. Using off-the-shelf components and software, we established a workflow for high-throughput time-resolved microscopy. We then screened the single‐gene deletion collection of Escherichia coli for antibiotic-induced morphological changes. Using single-cell quantitative descriptors and supervised classification methods, we measured how different cell morphologies developed over time for all strains in response to the β-lactam antibiotic cefsulodin. 191 strains exhibit significant variations under antibiotic treatment. Phenotypic clustering provided insights into processes that alter the antibiotic response. Mutants with stable bulges show delayed lysis, contributing to antibiotic tolerance. Lipopolysaccharides play a crucial role in bulge stability. This study demonstrates how multiparametric phenotyping by high-throughput time-resolved imaging and computer-aided cell classification can be used for comprehensively studying dynamic morphological transitions in bacteria.

A genetic screen to identify factors affected by undecaprenyl phosphate recycling uncovers novel connections to morphogenesis inEscherichia coli

“…Historically, growth has served this function [e.g., (Typas et al, 2008, Nichols et al, 2011]. However, since most mutants grow normally, more recent efforts have attempted to systematically catalog morphological phenotypes (Campos et al, 2018, French et al, 2017, Zahir et al, 2019. Here we corroborate, extend, and add to the collective morphological record 19 genes (and implicate several others) whose mutants produce shape defects in a matter of hours using relatively inexpensive reagents (i.e., LB).…”

Image-Based Dynamic Phenotyping Reveals Genetic Determinants of Filamentation-Mediated β-Lactam Tolerance

“…Recently, we developed a method for high-throughput timeresolved imaging of bacteria in 96-well plates (Zahir et al, 2019). This method is ideal for recording dynamic single-cell responses of a large number of strains in the form of high-resolution phase contrast images.…”

“…A systematic genomewide phenotypic analysis of β-lactam tolerance was hitherto difficult due to lack of a suitable high-throughput assay for measuring killing kinetics. We applied a recently developed methodology for high-throughput microscopy of bacteria grown in 96-well plates (Zahir et al, 2019) and measured lysis kinetics of all mutants in the Keio collection (Baba et al, 2006). The image-based dynamic phenotyping of the response to β-lactams revealed several proteins that modulate the process of lysis.…”

Cheetah: a computational toolkit for cybergenetic control

“…The Otsu method was also not able to distinguish between alive and dead cells, resulting in measurements that combined both categories. To enable both segmentation and classification a more complex pipeline would have needed to be implemented 39 . When compared to the more accurate results generated by Cheetah, the Otsu method led to a slightly lower estimation of average mCherry fluorescence ( Figure 2F).…”