ERnet: a tool for the semantic segmentation and quantitative analysis of endoplasmic reticulum topology for video-rate super-resolution imaging

Abstract

The topology of endoplasmic reticulum (ER) network is highly regulated by various cellular and environmental stimuli and affects major functions such as protein quality control and the cell's response to metabolic changes. The ability to quantify the dynamical changes of the ER structures in response to cellular perturbations is crucial for the development of novel therapeutic approaches against ER associated diseases, such as hereditary spastic paraplegias and Niemann Pick Disease type C. However, the rapid movement and small spatial dimension of ER networks make this task challenging. Here, we combine video-rate super-resolution imaging with a state-of-the-art semantic segmentation method capable of automatically classifying sheet and tubular ER domains inside individual cells. Data are skeletonised and represented by connectivity graphs to enable the precise and efficient quantification and comparison of the network connectivity from different complex ER phenotypes. The method, called ERnet, is powered by a Vision Transformer architecture, and integrates multi-head self-attention and channel attention into the model for adaptive weighting of frames in the time domain. We validated the performance of ERnet by measuring different ER morphology changes in response to genetic or metabolic manipulations. Finally, as a means to test the applicability and versatility of ERnet, we showed that ERnet can be applied to images from different cell types and also taken from different imaging setups. Our method can be deployed in an automatic, high-throughput, and unbiased fashion to identify subtle changes in cellular phenotypes that can be used as potential diagnostics for propensity to ER mediated disease, for disease progression, and for response to therapy.

Intracellular FUS protein accumulation leads to cytoskeletal, organelle and cellular homeostasis perturbations

“…Many of the above results point towards a dysregulation of the ALP which is supported by previous studies showing aberrant FUS aggregation disrupts the latter (Baskoylu et al, 2022; Soo et al, 2015). We observe cytoplasmic to nuclear translocation of TFEB in the presence of FUS accumulation, which we relate to the higher intralysosomal pH measured in FUS cells.…”

“…Despite using non-neuronal HEK293T and COS7 cells, we believe the underlying mechanisms observed should hold true for a more physiologically relevant model. In support of this, it was recently shown that there is impaired autophagy and neuronal dysfunction in a P525L-FUS knock in C. elegans model (Baskoylu et al, 2022). Moreover, our findings align with the fact that autophagy dysfunction occurs early on in disease pathology, which acts as an accelerator for neurodegeneration (Cortes and la Spada, 2019).…”

“…We show that clustering of lysosomes and mitochondria at the perinuclear area of cells is greater in P525L-, compared to WT-FUS and control cells (Figure S6). We have recently found that lysosomes are responsible for the maintenance of the tubular ER network in the periphery of cells (Lu et al, 2022(Lu et al, , 2020. Thus, clustering of lysosomes in the perinuclear area may significantly impair the tubular ER network in the cell periphery.…”

SwinCell: a transformer-based framework for dense 3D cellular segmentation

“…It enables accurate quantification of cell volume, shape, and intercellular spatial relationships, providing essential insights into biological processes. Particularly in areas like drug development, where the efficacy of treatments can be profoundly influenced by the 3D arrangement of cells and tissues, precise segmentation algorithms are indispensable 6,7 . Therefore, the development of advanced 3D cell segmentation algorithms is pivotal, given the intrinsic three-dimensional nature of biological tissue and cellular structures.…”

nERdy: network analysis of endoplasmic reticulum dynamics

“…In [27], the authors developed a deep residual-network model, ER-net, for the segmentation of ER. In follow-up work, ERnet-v2 [28], the authors used a Swin Transformer-based model to extract tubules, sheets, and sheet-based-tubules (SBTs) from the ER structure and provided quantitative measures to understand the topology of the ER network in a supervised manner. Garcia et al [29] developed a pipeline to quantitatively measure the areas of rough and smooth ER to assess the impact of different pharmacological perturbations on ER morphology.…”