Biology subjects

Kaminski, C. F.

Publications and source records attributed to Kaminski, C. F..

At least 19 recordsLinked to original sources

bioRxiv2026

Antarctic fish cell cultures show adaptation of organelle morphology and dynamics to extreme cold

In the Antarctic Southern Ocean, cold-blooded animals have evolved to live at stable temperatures of 0{+/-}2 {degrees}C. This extremely low temperature affects their biology at every scale, from protein folding to development. However, how animal (sub)cellular organisation and dynamics are adapted to near-0 {degrees}C temperatures has not been studied. We therefore established methods to culture and fluorescently label cells from the Antarctic plunderfish Harpagifer antarcticus and a temperate comparator species, the shanny Lipophrys pholis. By imaging these cultures live at physiological temperatures, we found that subcellular organisation is broadly conserved in H. antarcticus, featuring known membranous organelles and biomolecular condensates that remain dynamic, with mitochondria in H. antarcticus and L. pholis moving at similar speeds. However, we also identified differences in organelle properties between H. antarcticus and L. pholis, including lysosomal enlargement and mitochondrial morphology changes. These differences may be functionally linked to protein misfolding and slow embryonic development in Antarctic species.

cell biology↗

bioRxiv2026

α-Synuclein Facilitates Spontaneous Dopamine Release in a Calcium- and Phosphorylation-Dependent Manner

-Synuclein (aSyn) is central to Parkinsons disease pathogenesis, yet its native physiological role at the presynapse remain poorly defined. Here, super-resolution imaging in dopaminergic neurons reveals that endogenous aSyn localises within nanometres of L-type voltage-gated calcium channels (LTCC), with closer proximity under both spontaneous neuronal activity and stimulated conditions compared to when extracellular calcium is chelated. Blocking Ca2+/calmodulin-dependent kinase II (CaMKII) reduces aSyn clustering at LTCC under spontaneous activity, suggesting that calcium entry and downstream calcium-dependent kinase activity contribute to aSyn localisation. Moreover, quantitative single-molecule analyses indicate that calcium increases the abundance of both total and serine129 phosphorylated (pS129) aSyn in synaptosomes under spontaneous conditions, and NMR analysis reveals that both calcium and S129 phosphorylation increase the binding affinity of aSyn to synaptic vesicles. Functional assays further demonstrate that LTCC blockade elevates intracellular DA levels exclusively in the presence of aSyn under spontaneous but not stimulated conditions. Finally, biochemical fractionation and multi-colour single-molecule imaging reveal that aSyn preferentially associates with small vesicles that are not obligately coupled to full-fusion associated recycling pools. These results suggest that aSyn acts as a calcium- and phosphorylation-regulated modulator of spontaneous DA release through pathways that are largely independent of full-fusion recycling mechanisms, and that pS129 aSyn is not solely a pathological marker but may also reflects physiological regulation. Together, these insights provide a framework for understanding how therapeutic strategies targeting aSyn may impact its normal synaptic functions.

neuroscience↗

bioRxiv2026

Discovery of TDP-43 aggregation inhibitors via a hybrid machine learning framework

TAR DNA-binding protein 43 (TDP-43) aggregation is a hallmark of several neurodegenerative diseases, including amyotrophic lateral sclerosis and frontotemporal dementia. Recent therapeutic efforts have highlighted the potential of small molecules capable of inhibiting TDP-43 aggregation; however, no effective treatments currently exist. Here, we developed a hybrid machine learning approach combining graph neural network (GNN) embeddings with traditional chemical descriptors and biological target annotations. Using XGBoost as the final classifier enabled model interpretability through SHAP analysis, allowing the identification of key chemical features and target annotations associated with TDP-43 anti-aggregation activity. Complementary Monte Carlo Tree Search analysis highlighted specific chemical substructures linked to predicted activity. By screening an external library of 3,853 small molecules, the model identified two compounds not previously evaluated against TDP-43 aggregation, namely berberrubine and PE859. Molecular docking analysis revealed that both compounds interact favourably with the TDP-43 RNA recognition motif (RRM) domain through distinct binding modes. Experimental validation showed that both compounds significantly reduced TDP-43 aggregation in HEK cells. Further testing in Caenorhabditis elegans expressing human TDP-43 demonstrated that PE859 significantly rescued locomotor defects, while berberrubine showed partial improvement. This work establishes a hybrid machine learning approach for accelerating small molecule drug discovery, yielding two promising therapeutic candidates for TDP-43 proteinopathies.

bioinformatics↗

bioRxiv2025

SIMple: A fibre-based platform for accessible structured illumination microscopy

Structured illumination microscopy can be used to achieve optical sectioning and super resolution in fluorescence images, reducing out-of-focus light and increasing the resolution beyond the diffraction limit, without the need for specialised detection optics. However, the complex illumination path is difficult to build and align. We present an illumination path based on fibre-optic components for both splitting and phase-shifting the illumination light. This enables a SIMple and compact "Plug&Play" modality which substantially reduces the time and alignment required when adding the optics to an existing widefield instrument. The system is capable of optical sectioning imaging at camera-limited frame-rates using multiple excitation wavelengths simultaneously, as demonstrated by imaging fixed and live biological samples at 561 and 491 nm. Super-resolution imaging of fixed samples on a very compact, self-contained microscope is also demonstrated: illumination is coupled in by fibre to a lightweight frame with dimensions of just 300 x 450 x 300 mm3, enabling easy transportation and use in laboratories with limited space. Characterisation of the system using bead analysis shows a resolution of 168 and 172 nm at 491 and 561 nm, respectively, an improvement by a factor of 1.91 and 1.92 compared to widefield, with a field of view of 100 x 100 {micro}m2.

bioengineering↗

bioRxiv2025

Brain cholesterol metabolites cause significant neurodegeneration in human iPSC-derived neurons

Disrupted cholesterol metabolism is increasingly recognised as a contributing factor in neurodegeneration; however, the specific effects of key brain-derived cholesterol metabolites, 24S-hydroxycholesterol (24S-HC) and 27-hydroxycholesterol (27-HC), remain poorly understood. Using human iPSC-derived i3 cortical neurons, we demonstrate that both 24S-HC and 27-HC significantly impair neuronal calcium signalling by elevating resting calcium levels, reducing spike amplitude, and disrupting network synchrony. These functional deficits are accompanied by widespread organelle dysfunction. Both oxysterols induce mitochondrial fragmentation, decrease spare respiratory capacity, and impair lysosomal degradation. Notably, 27-HC uniquely triggers lysosomal swelling and membrane permeabilisation. Additional signs of cellular stress, including axonal swellings and elevated endoplasmic reticulum calcium levels, were also observed. Furthermore, both 24S-HC and 27-HC were found to directly interact with alpha-synuclein (aSyn), promoting its accumulation in cellular models. In contrast, cholesterol itself had minimal impact, highlighting the distinct toxicity of its hydroxylated metabolites. Together, these findings reveal a mechanistic link between oxysterol accumulation and neuronal dysfunction, supporting the hypothesis that elevated levels of 24S-HC and 27-HC, commonly observed in Parkinsons and Alzheimers disease, may actively drive neurodegenerative processes. Targeting oxysterol metabolism may therefore represent a promising therapeutic avenue for intervention in neurodegenerative disorders.

neuroscience↗

bioRxiv2025

Deep learning for fluorescence lifetime predictions enables high-throughput in vivo imaging

Fluorescence lifetime imaging microscopy (FLIM) is a powerful optical tool widely used in biomedical research to study changes in a samples microenvironment. However, data collection and interpretation are often challenging, and traditional methods such as exponential fitting and phasor plot analysis require a high number of photons per pixel for reliably measuring the fluorescence lifetime of a fluorophore. To satisfy this requirement, prolonged data acquisition times are needed, which makes FLIM a low-throughput technique with limited capability for in vivo applications. Here, we introduce FLIMngo, a deep learning model capable of quantifying FLIM data obtained from photon-starved environments. FLIMngo outperforms other deep learning approaches and phasor plot analyses, yielding accurate fluorescence lifetime predictions from decay curves obtained with fewer than 50 photons per pixel by leveraging both time and spatial information present in raw FLIM data. Thus, FLIMngo reduces FLIM data acquisition times to a few seconds, thereby, lowering phototoxicity related to prolonged light exposure and turning FLIM into a higher throughput tool suitable for analysis of live specimens. Following the characterisation and benchmarking of FLIMngo on simulated data, we highlight its capabilities through applications in live, dynamic samples. Examples include the quantification of disease-related protein aggregates in non-anaesthetised Caenorhabditis (C.) elegans, which significantly improves the applicability of FLIM by opening avenues to continuously assess C. elegans throughout their lifespan. Finally, FLIMngo is open-sourced and can be easily implemented across systems without the need for model retraining.

biophysics↗

bioRxiv2024

A high-resolution microscopy system for biological studies of cold-adapted species under physiological conditions

The fundamental processes governing life are sensitively dependent on temperature. Whilst much is known about the constraints on how proteins operate at 37{degrees}C, little knowledge exists about how biological function is maintained sub-zero temperature conditions, where proteins are less stable and oxidative damage is high. However, almost 90% of habitable environments on Earth are permanently below 5{degrees}C (i.e. the deep sea and polar regions). This means that we do not understand how a large and diverse proportion of the global biome functions. To address this question at the cellular level, tools are required for imaging biological systems at high resolution under physiological conditions. This poses severe technical challenges that cannot be addressed with traditional optical microscopy techniques. High-resolution imaging objectives require short working distances and the use of immersion media, which lead to rapid heat transfer from the microscope to the sample. This affects the viability of live specimens and the interpretability of the results when the sample function optimally at low temperatures. Condensation and temperature-induced shrinking of components pose further challenges, reducing image resolution and contrast. Here, we address these issues and provide a method for high-fidelity imaging of live biological samples at temperatures of around, or below, 0{degrees}C. Our method is compatible with different microscopy modalities, including super-resolution imaging. It relies on hardware additions to traditional microscopy systems that can be straightforwardly implemented, namely, a cooling collar, 10% ethanol as an immersion medium, and nitrogen flow to mitigate condensation. We demonstrate the method in live cell cultures derived from Antarctic fish species and highlight the need to maintain physiological conditions for these fragile biological samples. Future applications are diverse and include evolutionary biology and the study of cold-adapted organisms, as well as cellular biophysics and several applications in biotechnology.

biophysics↗

bioRxiv2024

FLIMPA: A versatile software for Fluorescence Lifetime Imaging Microscopy Phasor Analysis

Fluorescence lifetime imaging microscopy (FLIM) is an advanced microscopy technique capable of providing a deeper understanding of the molecular environment of a fluorophore. While FLIM data were traditionally analysed through the exponential fitting of the fluorophores emission decays, the use of phasor plots is increasingly becoming the preferred standard. This is due to their ability to visualise the distribution of fluorescent lifetimes within a sample, offering insights into molecular interactions in the sample without the need for model assumptions regarding the exponential decay behaviour of the fluorophores. However, so far most researchers have had to rely on commercial phasor plot software packages, which are closed-source and rely on proprietary data formats. In this paper, we introduce FLIMPA, an opensource, stand-alone software for phasor plot analysis that provides many of the features found in commercial software, and more. FLIMPA is fully developed in Python and offers advanced tools for data analysis and visualisation. It enhances FLIM data comparison by integrating phasor points from multiple trials and experimental conditions into a single plot, while also providing the possibility to explore detailed, localised insights within individual samples. We apply FLIMPA to introduce a cell-based assay for the quantification of microtubule depolymerisation, measured through fluorescence lifetime changes of SiR-tubulin, in response to various concentrations of Nocodazole, a microtubule depolymerising drug relevant to anti-cancer treatment.

biophysics↗

bioRxiv2024

Enzyme-Responsive DNA Condensates

Membrane-less compartments and organelles are widely acknowledged for their role in regulating cellular processes, and there is an urgent need to harness their full potential as both structural and functional elements of synthetic cells. Despite rapid progress, synthetically recapitulating the nonequilibrium, spatially distributed responses of natural membrane-less organelles remain elusive. Here we demonstrate that the activity of nucleic-acid cleaving enzymes can be localised within DNA-based membrane-less compartments by sequestering the respective DNA or RNA substrates. Reaction-diffusion processes lead to complex nonequilibrium patterns, dependent on enzyme concentration. By arresting similar dynamic patterns, we spatially organise different substrates in concentric sub-compartments, which can be then selectively addressed by different enzymes, demonstrating spatial distribution of enzymatic activity. Besides advancing our ability to engineer advanced biomimetic functions in synthetic membrane-less organelles, our results may facilitate the deployment of DNA-based condensates as microbioreactors or platforms for the detection and quantitation of enzymes and nucleic acids.

synthetic biology↗

bioRxiv2024

Machine learning-based spike sorting reveals how subneuronal concentrations of monomeric Tau cause a loss in excitatory postsynaptic currents in hippocampal neurons

Extracellular recordings of neuronal activity constitute a powerful tool for investigating the intricate dynamics of neural networks and the activity of individual neurons. Microelectrode arrays (MEAs) allow for recordings with a high electrode count, ranging from 10s to 1000s, generating extensive datasets of neuronal information. Furthermore, MEAs capture extracellular field potentials from cultured cells, resulting in highly complex neuronal signals that necessitate precise spike sorting for meaningful data extraction. Nevertheless, conventional spike sorting methods face limitations in recognising diverse spike shapes, thereby constraining the full utilisation of the rich dataset acquired from MEA recordings. To overcome these limitations, we have developed a machine learning algorithm, named PseudoSort, which employs advanced self-supervised learning techniques, a distinctive density-based pseudo-labelling strategy, and an iterative fine-tuning process to enhance spike sorting accuracy. Through extensive benchmarking on large-scale simulated datasets, we demonstrate the superior performance of PseudoSort compared to recently developed machine learning-based (ML) spike sorting algorithms. We showcase the practical application of PseudoSort by utilising MEA recordings from hippocampal neurons exposed to subneuronal concentrations of monomeric Tau, a protein associated with Alzheimers disease (AD). Our results, validated against patch clamp experiments, unveil that monomeric Tau at subneuronal concentrations induces stimulation-dependent disruptions in both local and global activity of hippocampal neurons. Remarkably, patch clamp electrophysiology highlights the effect of combined Tau and neuronal stimulation treatment on excitatory postsynaptic currents, whereas PseudoSort excels in identifying neuronal clusters that exhibit diminished firing capacity following Tau treatment alone, i.e., in the absence of stimulation. This comprehensive approach validates the prowess of PseudoSort and unravels the intricate effects of Tau on neuronal activity, particularly in the context of AD.

neuroscience↗

bioRxiv2024

Graphene microelectrode arrays, 4D structured illumination microscopy, and a machine learning-based spike sorting algorithm permit the analysis of ultrastructural neuronal changes during neuronal signalling in a model of Niemann-Pick disease type C

Simultaneously recording network activity and ultrastructural changes of the synapse is essential for advancing our understanding of the basis of neuronal functions. However, the rapid millisecond-scale fluctuations in neuronal activity and the subtle sub-diffraction resolution changes of synaptic morphology pose significant challenges to this endeavour. Here, we use graphene microelectrode arrays (G-MEAs) to address these challenges, as they are compatible with high spatial resolution imaging across various scales as well as high temporal resolution electrophysiological recordings. Furthermore, alongside G-MEAs, we deploy an easy-to-implement machine learning-based algorithm to efficiently process the large datasets collected from MEA recordings. We demonstrate that the combined use of G-MEAs, machine learning (ML)-based spike analysis, and four-dimensional (4D) structured illumination microscopy (SIM) enables the monitoring of the impact of disease progression on hippocampal neurons which have been treated with an intracellular cholesterol transport inhibitor mimicking Niemann-Pick disease type C (NPC) and show that synaptic boutons, compared to untreated controls, significantly increase in size, which leads to a loss in neuronal signalling capacity.

neuroscience↗

bioRxiv2023

Molecular mechanisms of liposome interactions with bacterial envelopes

Although fusogenic liposomes offer a promising approach for the delivery of antibiotic payloads across the cell envelope of Gram-negative bacteria, there is still limited understanding of the individual nanocarrier interactions with the bacterial target. Using super-resolution microscopy, we characterize the interaction dynamics of positively charged fusogenic liposomes with Gram-negative (Escherichia coli) and Gram-positive (Bacillus subtilis) bacteria. The liposomes merge with the outer membrane (OM) of Gram-negative bacteria, while attachment or lipid internalization is observed in Gram-positive cells. Employing total internal reflection fluorescence microscopy, we demonstrated liposome fusion with model supported lipid bilayers. For whole E. coli cells, however, we observed heterogeneous membrane integrations, primarily involving liposome attachment and hemifusion events. With increasing lipopolysaccharide length the likelihood of full-fusion events was reduced. The integration of artificial lipids into the OM of Gram-negative cells led to membrane destabilization, resulting in decreased bacterial vitality, membrane detachment, and improved co-delivery of Vancomycin--an effective antibiotic against Gram-positive cells. These findings provide significant insights into the interactions of individual nanocarriers with bacterial envelopes at the single-cell level, uncovering effects that would be missed in bulk measurements. This highlights the importance of conducting single-particle and single-cell investigations to assess the performance of next-generation drug delivery platforms.

biophysics↗

bioRxiv2022

Multiparametric sensing of outer membrane vesicle-derived supported lipid bilayers demonstrates the specificity of bacteriophage interactions

The use of bacteriophage, viruses that specifically infect bacteria, as antibiotics has become an area of great interest in recent years as the effectiveness of conventional antibiotics recedes. The detection of phage interactions with specific bacteria in a rapid and quantitative way is key for identifying phage of interest for novel antimicrobials. Outer membrane vesicles (OMVs) derived from gram-negative bacteria can be used to make supported lipid bilayers (SLBs) and therefore in vitro membrane models that contain naturally occurring components of the bacterial outer membrane. In this study, we used Escherichia coli OMV derived SLBs and use both fluorescent imaging and surface sensitive techniques to show their interactions with T4 phage. We also integrate these bilayers with microelectrode arrays (MEAs) functionalised with the conducting polymer PEDOT:PSS and show that the pore forming interactions of the phage with the SLBs can be monitored using electrical impedance spectroscopy. To highlight our ability to detect specific phage interactions, we also generate SLBs using OMVs derived from Citrobacter rodentium, which is resistant to T4 phage infection, and identify their lack of interaction with phage. The work presented here shows how interactions occurring between phage and these complex SLB systems can be monitored using a range of experimental techniques. We believe this approach can be used to identify phage against bacterial strains of interest, as well as more generally to monitor any pore forming structure (such as defensins) interacting with bacterial outer membranes, and thus aid in the development of next generation antimicrobials.

bioengineering↗

bioRxiv2022

Mutation of the ALS/FTD-associated RNA-binding protein FUS alters axonal cytoskeletal organisation

Aberrant condensation and localisation of the RNA-binding protein fused in sarcoma (FUS) occur in variants of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). ALS is also associated with cytoskeletal defects, genetically and through observations of compromised axonal transport. Here, we asked whether compromised axonal cytoskeletal organisation is an early feature of FUS-associated ALS/FTD. We used an ALS-associated mutant FUS(P525L) and the FTD-mimic hypomethylated FUS, FUS(16R), to investigate the common and distinct cytoskeletal changes found in these two reported Xenopus models. Combining a novel atomic force microscopy (AFM)-based approach for in vitro cytoskeletal characterisation and in vivo axonal branching analysis, we found that mutant FUS reduced actin density in the dynamically remodelling growth cone, and reduced axonal branch complexity. We furthermore found evidence of an axon looping defect for FUS(P525L). Therefore, we show that compromised actin remodelling is potentially an important early event in FUS-associated pathogenesis. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/510780v2_ufig1.gif" ALT="Figure 1"> View larger version (56K): org.highwire.dtl.DTLVardef@1e9a92aorg.highwire.dtl.DTLVardef@161076eorg.highwire.dtl.DTLVardef@12edf0forg.highwire.dtl.DTLVardef@365785_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2022

α-synuclein fibril and synaptic vesicle interactions lead to vesicle destruction and increased uptake into neurons

Monomeric alpha-synuclein (aSyn) is a well characterised as a lipid binding protein. aSyn is known to form amyloid fibrils which are also localised with lipids and organelles in so called Lewy bodies, insoluble structures found in Parkinsons disease patients brains. It is still unclear under which conditions the aSyn-lipid interaction can start to become pathological. Previous work to address pathological interactions has focused on using synthetic lipid membranes, which lack the complexity of physiological lipid membranes which not only have a more complex lipid composition, but also contain lipid interacting proteins. Here, we investigate how either monomeric or fibrillar aSyn interact with physiological synaptic vesicles (SV) isolated from rodent brain. Using small angle neutron scattering and high-resolution imaging we observe that aSyn fibrils disintegrate SV, whereas aSyn monomers cause clustering of SV. Furthermore, SV enhance the aggregation rate of aSyn, however increasing the SV:aSyn ratio causes a reduction in aggregation propensity. SV lipids appear as an integrated part of aSyn fibrils and while the fibril morphology differs to aSyn fibrils alone, the core fibril structure remains the same. We finally demonstrate that lipid-associated aSyn fibrils are more easily taken up into cortical i3Neurons derived from induced pluripotent stem cells. Our study sheds light on differences between interactions of aSyn with synthetic lipid vesicles and physiological SV. We show how aSyn fibrils may enhance pathology by disintegrating SV, which in turn may have fatal consequences for neurons. Furthermore, disease burden may additionally be impacted by an increased uptake of lipid-associated aSyn by neurons, leading to more SV damage and enhancing aSyn aggregation.

neuroscience↗

bioRxiv2022

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

The molecular mechanisms that connect the formation of aberrant cytoplasmic FUS condensates to biological malfunction are incompletely understood. Here, we develop an approach to determine the intracellular FUS viscosity in live mammalian cells and find that ALS-related mutant P525L-FUS forms the most viscous condensates and has impaired cytoskeletal mechanoproperties and increased euchromatin formation. We further show that some of the main cellular organelles, e.g., actin/tubulin, lysosomes, mitochondria, the endoplasmic reticulum, are significantly functionally/structurally impaired in the presence of FUS. These may be related to defects in the tubulin network, as the latter facilitates transport, formation, fusion and fission of organelles. We observe significant increases in lysosomal biogenesis, size and pH; moreover, intracellular FUS accumulation significantly promotes cytoplasmic-to-nuclear translocation of TFEB, i.e., the master gene for inducing autophagy. However, despite these, increased autophagy needed for protein aggregate clearance is not observed to occur. Our study reveals that the formation of highly viscous FUS condensates significantly impacts cytoskeletal/organelle function and cellular homeostasis, which are closely associated with cell ageing. This raises the intriguing question as to whether mutant FUS activates similar cell processes as those during cellular senescence.

biophysics↗

bioRxiv2022

A multiplex platform to identify mechanisms and modulators of proteotoxicity in neurodegeneration

Neurodegenerative disorders are a family of diseases that remain poorly treated despite their growing global health burden. A shared feature of many neurodegenerative disorders is the accumulation of toxic misfolded proteins. To gain insight into the mechanisms and modulators of protein misfolding, we developed a multiplex reverse genetics platform. Using this novel platform 29 cell-based models expressing proteins that undergo misfolding in neurodegeneration were probed against more than a thousand genetic modifiers. The resulting data provide insight into the nature of modifiers that act on multiple misfolded proteins as compared to those that show activity on only one. To illustrate the utility of this platform, we extensively characterized a potent hit from our screens, the human chaperone DNAJB6. We show that DNAJB6 is a general modifier of the toxicity and solubility of multiple amyotrophic lateral sclerosis and frontotemporal dementia (ALS/FTD)-linked RNA-binding proteins (RBPs), including FUS, TDP-43, and hnRNPA1. Biophysical examination of DNAJB6 demonstrated that it co-phase separates with, and alters the behavior of FUS containing condensates by locking them into a loose gel-like state which prevents their fibrilization. Domain mapping and a deep mutational scan of DNAJB6 support the critical importance for DNAJB6 phase separation in its effects on multiple RNA-binding proteins. Crucially, these studies also suggest that this property can be further tuned to generate novel variants with enhanced activity that might illuminate potential avenues for clinical translation.

neuroscience↗

bioRxiv2022

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

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 cells 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.

bioengineering↗