Automated detection of apoptotic versus nonapoptotic cell death using label-free computational microscopy

Deep learning-based image classification reveals heterogeneous execution of cell death fates during viral infection

Cell fate decisions, such as proliferation, differentiation, and death, are driven by complex molecular interactions and signaling cascades. While significant progress has been made in understanding the molecular determinants of these processes, historically, cell fate transitions were identified through light microscopy that focused on changes in cell morphology and function. Modern techniques have shifted toward probing molecular effectors to quantify these transitions, offering more precise quantification and mechanistic understanding. However, challenges remain in cases where the molecular signals are ambiguous, complicating the assignment of cell fate. During viral infection, programmed cell death (PCD) pathways, including apoptosis, necroptosis, and pyroptosis, exhibit complex signaling and molecular cross-talk. This can lead to simultaneous activation of multiple PCD pathways, which confounds assignment of cell fate based on molecular information alone. To address this challenge, we employed deep learning-based image classification of dying cells to analyze PCD in single herpes simplex virus-1 (HSV-1)-infected cells. Our approach reveals that despite heterogeneous activation of signaling, individual cells adopt predominantly prototypical death morphologies. Nevertheless, PCD is executed heterogeneously within a uniform population of virus-infected cells and varies over time. These findings demonstrate that image-based phenotyping can provide valuable insights into cell fate decisions, complementing molecular assays.

Label-free detection and simultaneous viability determination of CTCs by lens-free imaging cytometry

The detection of extremely rare circulating tumor cells (CTCs) in peripheral blood and simultaneously identifying their viabilities are significant for cancer diagnosis and prognosis as well as monitoring the efficacy of personalized treatment. A lens-free imaging system features high-resolution images taken over a large field of view (FOV), which has great potential for CTC detection and viability determination. But current still lens-free systems restrict the application for CTC detection in real samples due to the inherent limitations of lens-free technology: (1) the location of cells in the FOV will affect the imaging; (2) the extremely rare CTCs probably did not exist in one observation. In this paper, we realized the detection of CTCs in whole blood and the simultaneous determination of their viabilities by lens-free imaging cytometry. Our in-flow system plus a large FOV range of lens-free imaging highly increased the detection rate of rare CTCs with a high throughput of 150,000 cells per minute and improved the recognition efficiency for blood cells, living/dead CTCs by using a cell tracing-assisted deep learning algorithm. With this method, the average precision of blood cells, living/dead lung cancer cells A549, and living/dead colon cancer cells SW620 reached 98.80%, 97.88%, 97.93%, 97.72%, and 98.60%, respectively. Our system got a highly consistent result with the manual counting method using fluorescent staining (Pearson's r 99.93% for SW620) and can easily detect as few as 10 dead or living CTCs from 100,000 white blood cells (WBCs). Finally, real clinical samples were detected in our system. Both dead and living CTCs were found in all six advanced-stage cancer patients, and the number of living CTCs per million WBCs ranged from 13 to 39, more than that of the dead CTCs (5 to 25), while none of the CTCs were detected in six healthy control subjects. Moreover, we also found that CTCs died very quickly after leaving the human body, indicating that CTCs should be studied as soon as possible after sampling. Although this method is implemented for CTCs, it can also be used for the detection of other rare cells.

Novel Thienopyrimidine-Hydrazinyl Compounds Induce DRP1-Mediated Non-Apoptotic Cell Death in Triple-Negative Breast Cancer Cells

Apoptosis induction with taxanes or anthracyclines is the primary therapy for TNBC. Cancer cells can develop resistance to anticancer drugs, causing them to recur and metastasize. Therefore, non-apoptotic cell death inducers could be a potential treatment to circumvent apoptotic drug resistance. In this study, we discovered two novel compounds, TPH104c and TPH104m, which induced non-apoptotic cell death in TNBC cells. These lead compounds were 15- to 30-fold more selective in TNBC cell lines and significantly decreased the proliferation of TNBC cells compared to that of normal mammary epithelial cell lines. TPH104c and TPH104m induced a unique type of non-apoptotic cell death, characterized by the absence of cellular shrinkage and the absence of nuclear fragmentation and apoptotic blebs. Although TPH104c and TPH104m induced the loss of the mitochondrial membrane potential, TPH104c- and TPH104m-induced cell death did not increase the levels of cytochrome c and intracellular reactive oxygen species (ROS) and caspase activation, and cell death was not rescued by incubating cells with the pan-caspase inhibitor, carbobenzoxy-valyl-alanyl-aspartyl-[O-methyl]-fluoromethylketone (Z-VAD-FMK). Furthermore, TPH104c and TPH104m significantly downregulated the expression of the mitochondrial fission protein, DRP1, and their levels determined their cytotoxic efficacy. Overall, TPH104c and TPH104m induced non-apoptotic cell death, and further determination of their cell death mechanisms will aid in the development of new potent and efficacious anticancer drugs to treat TNBC.

Transformer-based spatial-temporal detection of apoptotic cell death in live-cell imaging

Intravital microscopy has revolutionized live-cell imaging by allowing the study of spatial-temporal cell dynamics in living animals. However, the complexity of the data generated by this technology has limited the development of effective computational tools to identify and quantify cell processes. Amongst them, apoptosis is a crucial form of regulated cell death involved in tissue homeostasis and host defense. Live-cell imaging enabled the study of apoptosis at the cellular level, enhancing our understanding of its spatial-temporal regulation. However, at present, no computational method can deliver robust detection of apoptosis in microscopy timelapses. To overcome this limitation, we developed ADeS, a deep learning-based apoptosis detection system that employs the principle of activity recognition. We trained ADeS on extensive datasets containing more than 10,000 apoptotic instances collected both in vitro and in vivo, achieving a classification accuracy above 98% and outperforming state-of-the-art solutions. ADeS is the first method capable of detecting the location and duration of multiple apoptotic events in full microscopy timelapses, surpassing human performance in the same task. We demonstrated the effectiveness and robustness of ADeS across various imaging modalities, cell types, and staining techniques. Finally, we employed ADeS to quantify cell survival in vitro and tissue damage in mice, demonstrating its potential application in toxicity assays, treatment evaluation, and inflammatory dynamics. Our findings suggest that ADeS is a valuable tool for the accurate detection and quantification of apoptosis in live-cell imaging and, in particular, intravital microscopy data, providing insights into the complex spatial-temporal regulation of this process.

Toward a predictive understanding of epithelial cell death

Epithelial cell death is highly prevalent during development and tissue homeostasis. While we have a rather good understanding of the molecular regulators of programmed cell death, especially for apoptosis, we still fail to predict when, where, how many and which specific cells will die in a tissue. This likely relies on the much more complex picture of apoptosis regulation in a tissular and epithelial context, which entails cell autonomous but also non-cell autonomous factors, diverse feedback and multiple layers of regulation of the commitment to apoptosis. In this review, we illustrate this complexity of epithelial apoptosis regulation by describing these different layers of control, all demonstrating that local cell death probability is a complex emerging feature. We first focus on non-cell autonomous factors that can locally modulate the rate of cell death, including cell competition, mechanical input and geometry as well as systemic effects. We then describe the multiple feedback mechanisms generated by cell death itself. We also outline the multiple layers of regulation of epithelial cell death, including the coordination of extrusion and regulation occurring downstream of effector caspases. Eventually, we propose a roadmap to reach a more predictive understanding of cell death regulation in an epithelial context.

The small molecule raptinal can simultaneously induce apoptosis and inhibit PANX1 activity

Discovery of new small molecules that can activate distinct programmed cell death pathway is of significant interest as a research tool and for the development of novel therapeutics for pathological conditions such as cancer and infectious diseases. The small molecule raptinal was discovered as a pro-apoptotic compound that can rapidly trigger apoptosis by promoting the release of cytochrome c from the mitochondria and subsequently activating the intrinsic apoptotic pathway. As raptinal is very effective at inducing apoptosis in a variety of different cell types in vitro and in vivo, it has been used in many studies investigating cell death as well as the clearance of dying cells. While examining raptinal as an apoptosis inducer, we unexpectedly identified that in addition to its pro-apoptotic activities, raptinal can also inhibit the activity of caspase-activated Pannexin 1 (PANX1), a ubiquitously expressed transmembrane channel that regulates many cell death-associated processes. By implementing numerous biochemical, cell biological and electrophysiological approaches, we discovered that raptinal can simultaneously induce apoptosis and inhibit PANX1 activity. Surprisingly, raptinal was found to inhibit cleavage-activated PANX1 via a mechanism distinct to other well-described PANX1 inhibitors such as carbenoxolone and trovafloxacin. Furthermore, raptinal also interfered with PANX1-regulated apoptotic processes including the release of the 'find-me' signal ATP, the formation of apoptotic cell-derived extracellular vesicles, as well as NLRP3 inflammasome activation. Taken together, these data identify raptinal as the first compound that can simultaneously induce apoptosis and inhibit PANX1 channels. This has broad implications for the use of raptinal in cell death studies as well as in the development new PANX1 inhibitors.

Modulation of Point Spread Function for Super-Resolution Imaging

High image resolution is desired in wave-related areas such as ultrasound, acoustics, optics, and electromagnetics. However, the spatial resolution of an imaging system is limited by the spatial frequency of the point spread function (PSF) of the system due to diffraction. In this article, the PSF is modulated in amplitude, phase, or both to increase the spatial frequency to reconstruct super-resolution images of objects or wave sources/fields, where the modulator can be a focused shear wave produced remotely by, for example, a radiation force from a focused Bessel beam or X-wave, or can be a small particle manipulated remotely by a radiation-force (such as acoustic and optical tweezers) or electrical and magnetic forces. A theory of the PSF-modulation method was developed, and computer simulations and experiments were conducted. The result of an ultrasound experiment shows that a pulse-echo (two-way) image reconstructed has a super-resolution (0.65 mm) as compared to the diffraction limit (2.65 mm) using a 0.5-mm-diameter modulator at 1.483-mm wavelength, and the signal-to-noise ratio (SNR) of the image was about 31 dB. If the minimal SNR of a "visible" image is 3, the resolution can be further increased to about 0.19 mm by decreasing the size of the modulator. Another ultrasound experiment shows that a wave source was imaged (one-way) at about 30-dB SNR using the same modulator size and wavelength above. The image clearly separated two 0.5-mm spaced lines, which gives a 7.26-fold higher resolution than that of the diffraction limit (3.63 mm). Although, in theory, the method has no limit on the highest achievable image resolution, in practice, the resolution is limited by noises. Also, a PSF-weighted super-resolution imaging method based on the PSF-modulation method was developed. This method is easier to implement but may have some limitations. Finally, the methods above can be applied to imaging systems of an arbitrary PSF and can produce 4-D super-resolution images. With a proper choice of a modulator (e.g., a quantum dot) and imaging system, nanoscale (a few nanometers) imaging is possible.

Automated detection of apoptotic bodies and cells in label-free time-lapse high-throughput video microscopy using deep convolutional neural networks

MOTIVATION

Reliable label-free methods are needed for detecting and profiling apoptotic events in time-lapse cell-cell interaction assays. Prior studies relied on fluorescent markers of apoptosis, e.g. Annexin-V, that provide an inconsistent and late indication of apoptotic onset for human melanoma cells. Our motivation is to improve the detection of apoptosis by directly detecting apoptotic bodies in a label-free manner.

RESULTS

Our trained ResNet50 network identified nanowells containing apoptotic bodies with 92% accuracy and predicted the onset of apoptosis with an error of one frame (5 min/frame). Our apoptotic body segmentation yielded an IoU accuracy of 75%, allowing associative identification of apoptotic cells. Our method detected apoptosis events, 70% of which were not detected by Annexin-V staining.

AVAILABILITY AND IMPLEMENTATION

Open-source code and sample data provided at https://github.com/kwu14victor/ApoBDproject.

DeXtrusion: automatic recognition of epithelial cell extrusion through machine learning in vivo

Accurately counting and localising cellular events from movies is an important bottleneck of high-content tissue/embryo live imaging. Here, we propose a new methodology based on deep learning that allows automatic detection of cellular events and their precise xyt localisation on live fluorescent imaging movies without segmentation. We focused on the detection of cell extrusion, the expulsion of dying cells from the epithelial layer, and devised DeXtrusion: a pipeline based on recurrent neural networks for automatic detection of cell extrusion/cell death events in large movies of epithelia marked with cell contour. The pipeline, initially trained on movies of the Drosophila pupal notum marked with fluorescent E-cadherin, is easily trainable, provides fast and accurate extrusion predictions in a large range of imaging conditions, and can also detect other cellular events, such as cell division or cell differentiation. It also performs well on other epithelial tissues with reasonable re-training. Our methodology could easily be applied for other cellular events detected by live fluorescent microscopy and could help to democratise the use of deep learning for automatic event detections in developing tissues.

Fluorescence-Based Portable Assays for Detection of Biological and Chemical Analytes

Fluorescence-based detection techniques are part of an ever-expanding field and are widely used in biomedical and environmental research as a biosensing tool. These techniques have high sensitivity, selectivity, and a short response time, making them a valuable tool for developing bio-chemical assays. The endpoint of these assays is defined by changes in fluorescence signal, in terms of its intensity, lifetime, and/or shift in spectrum, which is monitored using readout devices such as microscopes, fluorometers, and cytometers. However, these devices are often bulky, expensive, and require supervision to operate, which makes them inaccessible in resource-limited settings. To address these issues, significant effort has been directed towards integrating fluorescence-based assays into miniature platforms based on papers, hydrogels, and microfluidic devices, and to couple these assays with portable readout devices like smartphones and wearable optical sensors, thereby enabling point-of-care detection of bio-chemical analytes. This review highlights some of the recently developed portable fluorescence-based assays by discussing the design of fluorescent sensor molecules, their sensing strategy, and the fabrication of point-of-care devices.