| |
Speaker 1: Dr Adrien Hallou
University of Cambridge
Spatial mechano-transcriptomics of the early mouse embryo
Abstract:
Cell morphology, mechanical forces, and gene expression act together to orchestrate cell fate decisions and tissue morphogenesis during embryonic development. Analysis into the nature of this cooperation is therefore required to fully understand the nature of the highly complex mechanisms which sculpt the developing embryo in space and time. Here, we present a new and unique computational approach combining image-based mechanical force inference and spatial transcriptomics and demonstrate its applicability to derive at the tissue level single cell mechanical and morphometric information alongside gene expression levels. Using a E8.5 mouse embryo seqFISH data set, we show that an integrated analysis of these modalities enables the prediction of tissue compartment boundaries on the basis of spatially resolved and correlated mechanical tension profiles and gene expression patterns, as well as the identification of gene modules predictive of the mechanical state of a cell. Our method can be applied to any spatial transcriptomics data set with sufficient cell membrane segmentation quality, enabling further analysis into the complex interactions between cell morphology, mechanical forces, and gene expression in a variety of systems and organisms.
|
| |
Speaker 2: Alison Farrar
University of Oxford
Bacterial ribosome phenotypes as a rapid antibiotic susceptibility test
Abstract:
Bacteria divide and evolve at staggering time scales, which leads to the development of antimicrobial resistance. Their essential role in living cells has made ribosomes a key target for antimicrobial discovery, with more than 24 classes of antibiotics specifically inhibiting ribosome function. When exposed to antibiotics, susceptible cells develop characteristic ribosomal spatial organizations that can be used as a phenotype. To visualize intracellular ribosome and nucleoid distributions, we used widefield fluorescence microscopy with ribosome-targeting fluorescent DNA probes and stained the nucleoid. We use images of antibiotic-treated E. coli to train a convolutional neural network to identify the ribosome phenotypes of ciprofloxacin-treated bacteria. As a proof of principle, this neural network was then applied to images of E. coli from ciprofloxacin-resistant clinical isolates, which should appear more like untreated E. coli (Zagajewski et al., MedRxiv 2023). This talk will also cover the motivations and preliminary results of the “Infection Inspection” citizen science project.
|
