Junction Mapper is a novel computer vision tool to decipher cell–cell contact phenotypes

Abstract

Stable cell–cell contacts underpin tissue architecture and organization. Quantification of junctions of mammalian epithelia requires laborious manual measurements that are a major roadblock for mechanistic studies. We designed Junction Mapper as an open access, semi-automated software that defines the status of adhesiveness via the simultaneous measurement of pre-defined parameters at cell–cell contacts. It identifies contacting interfaces and corners with minimal user input and quantifies length, area and intensity of junction markers. Its ability to measure fragmented junctions is unique. Importantly, junctions that considerably deviate from the contiguous staining and straight contact phenotype seen in epithelia are also successfully quantified (i.e. cardiomyocytes or endothelia). Distinct phenotypes of junction disruption can be clearly differentiated among various oncogenes, depletion of actin regulators or stimulation with other agents. Junction Mapper is thus a powerful, unbiased and highly applicable software for profiling cell–cell adhesion phenotypes and facilitate studies on junction dynamics in health and disease.

High content Image Analysis to study phenotypic heterogeneity in endothelial cell monolayers

“…Assessing IEJ at the single cell level is key to highlight subtle nuances in inter-endothelial cells connectivity within the same monolayer. Nonetheless, measuring and classifying junctional signal by image analysis is technically challenging and no currently available software is able to evaluate multiple images in a standardised way 29,30 . To overcome this obstacle, we have implemented within ECPT a junction analysis method (Appendix 1).…”

“…Although VE-cadherin junction are dynamic, previous studies showed the possibility to characterize the junction patterning with image analysis 25, 26 . However, most work carried on HUVEC did not study the differences in different organ-specific EC and did not study single-cell level heterogeneity.…”

“…To overcome this obstacle, we integrated within ECPT an automated way to analyse VE-cadherin pattern and junction morphology using machine learning capabilities of CPA. Previous studies on EC junction 18,25,26 did not analyse the proportion of different junction pattern, our method subdivides the junctional inter-cellular space in several discrete areas scored using an expert-trained machine learning algorithm (CPA with Fast gentle Boosting, Appendix 1). In line with the literature in the field, we classified each junction binning in three categories: jagged, linear or stabilised (Figure 4A).…”

Rac1-PAK1 regulation of Rab11 cycling promotes junction destabilization

“…In contrast to other GTPase disruption of keratinocyte junctions ( Braga et al, 2000 ; Brezovjakova et al, 2019 ; Frasa et al, 2010 ), Rac1 activation induces a striking phenotype: Cells flatten out with elongation of the contacting interface between neighbors (this work), and thickening of cadherin staining in the middle of junctions ( Lozano et al, 2008 ). The progressive disappearance of cadherin staining from cell corners suggests that E-cadherin delivery to junctions is misdirected to the center of contacts.…”

“…Reduced Pearson coefficient values may reflect the contribution of the cytosolic pool of catenins associated with distinct partners (i.e., not on vesicles). Unfortunately, the striking Rac1-specific phenotype at junctions (with augmented fluorescence signal in the middle of contacts; Lozano et al, 2008 ) makes it challenging to compare the relative cadherin-catenin association at junctions with the vesicular pool ( Brezovjakova et al, 2019 ). Similar analyses with E-cadherin immunoprecipitation from transduced (TAT-Rac1 Q61L ) or transfected cells (myc-Rac1 Q61L ) lysates did not show the release of catenins from the complex ( Fig.…”