No file available
To read the file of this research,
you can request a copy directly from the authors.
The Roboscope: Smart and Fast Microscopy for Generic Event-Driven Acquisition
Preprints and early-stage research may not have been peer reviewed yet.
Abstract
Automation of fluorescence microscopy is a challenge for capturing rare or transient events in biology and medicine. It relies on smart devices that integrate and interpret the observed data, and react to the targeted biological event. We report on the Roboscope, a novel autonomous microscope combining sequence interruption and deep learning integration, allowing generic event-driven acquisitions. This system distinguishes itself by its adaptability to various experiments, quick capture of dynamic events, and minimal data greediness - training with less than 100 images per class. The Roboscope's capability is demonstrated in non-synchronized cells by capturing the metaphase, a 20-minute event happening once per day or less. Conversely, double thymidine-block synchronisation, despite occurring during DNA replication, may perturb mitotic-spindle mechanics. The Roboscope's versatility and efficiency offer significant advancements to tackle the current challenges of cell biology, spreading out advanced microscopy methods to fundamental research as well as high content screening and precision medicine.
ResearchGate has not been able to resolve any citations for this publication.
Quantitative microscopy workflows have evolved dramatically over the past years, progressively becoming more complex with the emergence of deep learning. Long-standing challenges such as three-dimensional segmentation of complex microscopy data can finally be addressed, and new imaging modalities are breaking records in both resolution and acquisition speed, generating gigabytes if not terabytes of data per day. With this shift in bioimage workflows comes an increasing need for efficient orchestration and data management, necessitating multitool interoperability and the ability to span dedicated computing resources. However, existing solutions are still limited in their flexibility and scalability and are usually restricted to offline analysis. Here we introduce Arkitekt, an open-source middleman between users and bioimage apps that enables complex quantitative microscopy workflows in real time. It allows the orchestration of popular bioimage software locally or remotely in a reliable and efficient manner. It includes visualization and analysis modules, but also mechanisms to execute source code and pilot acquisition software, making ‘smart microscopy’ a reality.
How can inter-individual variability be quantified? Measuring many features per experiment raises the question of choosing them to recapitulate high-dimensional data. Tackling this challenge on spindle elongation phenotypes, we showed that only three typical elongation patterns describe spindle elongation in C. elegans one-cell embryo. These archetypes, automatically extracted from the experimental data using principal component analysis (PCA), accounted for more than 95% of inter-individual variability of more than 1600 experiments across more than 100 different conditions. The two first archetypes were related to spindle average length and anaphasic elongation rate. The third archetype, accounting for 6% of the variability, was novel and corresponded to a transient spindle shortening in late metaphase, reminiscent of kinetochore function-defect phenotypes. Importantly, these three archetypes were robust to the choice of the dataset and were found even considering only non-treated conditions. Thus, the inter-individual differences between genetically perturbed embryos have the same underlying nature as natural inter-individual differences between wild-type embryos, independently of the temperatures. We thus propose that beyond the apparent complexity of the spindle, only three independent mechanisms account for spindle elongation, weighted differently in the various conditions. Interestingly, the spindle-length archetypes covered both metaphase and anaphase, suggesting that spindle elongation in late metaphase is sufficient to predict the late anaphase length. We validated this idea using a machine-learning approach. Finally, given amounts of these three archetypes could represent a quantitative phenotype. To take advantage of this, we set out to predict interacting genes from a seed based on the PCA coefficients. We exemplified this firstly on the role of tpxl-1 whose homolog tpx2 is involved in spindle microtubule branching, secondly the mechanism regulating metaphase length, and thirdly the central spindle players which set the length at anaphase. We found novel interactors not in public databases but supported by recent experimental publications.
Cell segmentation is a critical step for quantitative single-cell analysis in microscopy images. Existing cell segmentation methods are often tailored to specific modalities or require manual interventions to specify hyper-parameters in different experimental settings. Here, we present a multimodality cell segmentation benchmark, comprising more than 1,500 labeled images derived from more than 50 diverse biological experiments. The top participants developed a Transformer-based deep-learning algorithm that not only exceeds existing methods but can also be applied to diverse microscopy images across imaging platforms and tissue types without manual parameter adjustments. This benchmark and the improved algorithm offer promising avenues for more accurate and versatile cell analysis in microscopy imaging.
Light-sheet microscopes enable rapid high-resolution imaging of biological specimens; however, biological processes span spatiotemporal scales. Moreover, long-term phenotypes are often instigated by rare or fleeting biological events that are difficult to capture with a single imaging modality. Here, to overcome this limitation, we present smartLLSM, a microscope that incorporates artificial intelligence-based instrument control to autonomously switch between epifluorescent inverted imaging and lattice light-sheet microscopy (LLSM). We apply this approach to two unique processes: cell division and immune synapse formation. In each context, smartLLSM provides population-level statistics across thousands of cells and autonomously captures multicolor three-dimensional datasets or four-dimensional time-lapse movies of rare events at rates that dramatically exceed human capabilities. From this, we quantify the effects of Taxol dose on spindle structure and kinetochore dynamics in dividing cells and of antigen strength on cytotoxic T lymphocyte engagement and lytic granule polarization at the immune synapse. Overall, smartLLSM efficiently detects rare events within heterogeneous cell populations and records these processes with high spatiotemporal four-dimensional imaging over statistically significant replicates.
The optical microscope has revolutionized biology since at least the 17 th Century. Since then, it has progressed from a largely observational tool to a powerful bioanalytical platform. However, realizing its full potential to study live specimens is hindered by a daunting array of technical challenges. Here, we delve into the current state of live imaging to explore the barriers that must be overcome and the possibilities that lie ahead. We venture to envision a future where we can visualize and study everything, everywhere, all at once – from the intricate inner workings of a single cell to the dynamic interplay across entire organisms, and a world where scientists could access the necessary microscopy technologies anywhere.
Single-molecule localization microscopy enables three-dimensional fluorescence imaging at tens-of-nanometer resolution, but requires many camera frames to reconstruct a super-resolved image. This limits the typical throughput to tens of cells per day. While frame rates can now be increased by over an order of magnitude, the large data volumes become limiting in existing workflows. Here we present an integrated acquisition and analysis platform leveraging microscopy-specific data compression, distributed storage and distributed analysis to enable an acquisition and analysis throughput of 10,000 cells per day. The platform facilitates graphically reconfigurable analyses to be automatically initiated from the microscope during acquisition and remotely executed, and can even feed back and queue new acquisition tasks on the microscope. We demonstrate the utility of this framework by imaging hundreds of cells per well in multi-well sample formats. Our platform, implemented within the PYthon-Microscopy Environment (PYME), is easily configurable to control custom microscopes, and includes a plugin framework for user-defined extensions.
Light microscopy is a powerful single-cell technique that allows for quantitative spatial information at subcellular resolution. However, unlike flow cytometry and single-cell sequencing techniques, microscopy has issues achieving high-quality population-wide sample characterization while maintaining high resolution. Here, we present a general framework, data-driven microscopy (DDM) that uses real-time population-wide object characterization to enable data-driven high-fidelity imaging of relevant phenotypes based on the population context. DDM combines data-independent and data-dependent steps to synergistically enhance data acquired using different imaging modalities. As a proof of concept, we develop and apply DDM with plugins for improved high-content screening and live adaptive microscopy for cell migration and infection studies that capture events of interest, rare or common, with high precision and resolution. We propose that DDM can reduce human bias, increase reproducibility, and place single-cell characteristics in the context of the sample population when interpreting microscopy data, leading to an increase in overall data fidelity.
Monitoring the proteins and lipids that mediate all cellular processes requires imaging methods with increased spatial and temporal resolution. STED (stimulated emission depletion) nanoscopy enables fast imaging of nanoscale structures in living cells but is limited by photobleaching. Here, we present event-triggered STED, an automated multiscale method capable of rapidly initiating two-dimensional (2D) and 3D STED imaging after detecting cellular events such as protein recruitment, vesicle trafficking and second messengers activity using biosensors. STED is applied in the vicinity of detected events to maximize the temporal resolution. We imaged synaptic vesicle dynamics at up to 24 Hz, 40 ms after local calcium activity; endocytosis and exocytosis events at up to 11 Hz, 40 ms after local protein recruitment or pH changes; and the interaction between endosomal vesicles at up to 3 Hz, 70 ms after approaching one another. Event-triggered STED extends the capabilities of live nanoscale imaging, enabling novel biological observations in real time.
A common goal of fluorescence microscopy is to collect data on specific biological events. Yet, the event-specific content that can be collected from a sample is limited, especially for rare or stochastic processes. This is due in part to photobleaching and phototoxicity, which constrain imaging speed and duration. We developed an event-driven acquisition framework, in which neural-network-based recognition of specific biological events triggers real-time control in an instant structured illumination microscope. Our setup adapts acquisitions on-the-fly by switching between a slow imaging rate while detecting the onset of events, and a fast imaging rate during their progression. Thus, we capture mitochondrial and bacterial divisions at imaging rates that match their dynamic timescales, while extending overall imaging durations. Because event-driven acquisition allows the microscope to respond specifically to complex biological events, it acquires data enriched in relevant content. Event-driven acquisition uses neural-network-based recognition of specific biological events to trigger switching between slow and fast super-resolution imaging, enriching the capture of interesting events with high spatiotemporal resolution.
Microscopy image analysis has recently made enormous progress both in terms of accuracy and speed thanks to machine learning methods and improved computational resources. This greatly facilitates the online adaptation of microscopy experimental plans using real-time information of the observed systems and their environments. Applications in which reactiveness is needed are multifarious. Here we report MicroMator, an open and flexible software for defining and driving reactive microscopy experiments. It provides a Python software environment and an extensible set of modules that greatly facilitate the definition of events with triggers and effects interacting with the experiment. We provide a pedagogic example performing dynamic adaptation of fluorescence illumination on bacteria, and demonstrate MicroMator’s potential via two challenging case studies in yeast to single-cell control and single-cell recombination, both requiring real-time tracking and light targeting at the single-cell level. In microscopy, applications in which reactiveness is needed are multifarious. Here the authors report MicroMator, a Python software package for reactive experiments, which they use for applications requiring real-time tracking and light-targeting at the single-cell level.
Synchronous cell populations are commonly used for the analysis of various aspects of cellular metabolism at specific stages of the cell cycle. Cell synchronization at a chosen cell cycle stage is most frequently achieved by inhibition of specific metabolic pathway(s). In this respect, various protocols have been developed to synchronize cells in particular cell cycle stages. In this review, we provide an overview of the protocols for cell synchronization of mammalian cells based on the inhibition of synthesis of DNA building blocks—deoxynucleotides and/or inhibition of DNA synthesis. The mechanism of action, examples of their use, and advantages and disadvantages are described with the aim of providing a guide for the selection of suitable protocol for different studied situations.
Deep learning is a subdiscipline of artificial intelligence that uses a machine learning technique called artificial neural networks to extract patterns and make predictions from large data sets. The increasing adoption of deep learning across healthcare domains together with the availability of highly characterised cancer datasets has accelerated research into the utility of deep learning in the analysis of the complex biology of cancer. While early results are promising, this is a rapidly evolving field with new knowledge emerging in both cancer biology and deep learning. In this review, we provide an overview of emerging deep learning techniques and how they are being applied to oncology. We focus on the deep learning applications for omics data types, including genomic, methylation and transcriptomic data, as well as histopathology-based genomic inference, and provide perspectives on how the different data types can be integrated to develop decision support tools. We provide specific examples of how deep learning may be applied in cancer diagnosis, prognosis and treatment management. We also assess the current limitations and challenges for the application of deep learning in precision oncology, including the lack of phenotypically rich data and the need for more explainable deep learning models. Finally, we conclude with a discussion of how current obstacles can be overcome to enable future clinical utilisation of deep learning.
In many phenomena of biological systems, not a majority, but a minority of cells act on the entire multicellular system causing drastic changes in the system properties. To understand the mechanisms underlying such phenomena, it is essential to observe the spatiotemporal dynamics of a huge population of cells at sub-cellular resolution, which is difficult with conventional tools such as microscopy and flow cytometry. Here, we describe an imaging system named AMATERAS that enables optical imaging with an over-one-centimeter field-of-view and a-few-micrometer spatial resolution. This trans-scale-scope has a simple configuration, composed of a low-power lens for machine vision and a hundred-megapixel image sensor. We demonstrated its high cell-throughput, capable of simultaneously observing more than one million cells. We applied it to dynamic imaging of calcium ions in HeLa cells and cyclic-adenosine-monophosphate in Dictyostelium discoideum , and successfully detected less than 0.01% of rare cells and observed multicellular events induced by these cells.
The development of digital pathology and progression of state‐of‐the‐art algorithms for computer vision have led to increasing interest in the use of artificial intelligence (AI), especially deep learning (DL)‐based AI, in tumor pathology. The DL‐based algorithms have been developed to conduct all kinds of work involved in tumor pathology, including tumor diagnosis, subtyping, grading, staging, and prognostic prediction, as well as the identification of pathological features, biomarkers and genetic changes. The applications of AI in pathology not only contribute to improve diagnostic accuracy and objectivity but also reduce the workload of pathologists and subsequently enable them to spend additional time on high‐level decision‐making tasks. In addition, AI is useful for pathologists to meet the requirements of precision oncology. However, there are still some challenges relating to the implementation of AI, including the issues of algorithm validation and interpretability, computing systems, the unbelieving attitude of pathologists, clinicians and patients, as well as regulators and reimbursements. Herein, we present an overview on how AI‐based approaches could be integrated into the workflow of pathologists and discuss the challenges and perspectives of the implementation of AI in tumor pathology.
Fluorescence Lifetime Imaging Microscopy (FLIM) is a robust tool to measure Förster Resonance Energy Transfer (FRET) between two fluorescent proteins, mainly when using genetically-encoded FRET biosensors. It is then possible to monitor biological processes such as kinase activity with a good spatiotemporal resolution and accuracy. Therefore, it is of interest to improve this methodology for future high content screening purposes. We here implement a time-gated FLIM microscope that can image and quantify fluorescence lifetime with a higher speed than conventional techniques such as Time-Correlated Single Photon Counting (TCSPC). We then improve our system to perform automatic screen analysis in a 96-well plate format. Moreover, we use a FRET biosensor of AURKA activity, a mitotic kinase involved in several epithelial cancers. Our results show that our system is suitable to measure FRET within our biosensor paving the way to the screening of novel compounds, potentially allowing to find new inhibitors of AURKA activity.
Le fuseau mitotique assure la ségrégation des chromatides sœurs et le maintien de la poïdie des cellules filles. Le fuseau est composé de microtubules dynamiques (qui polymérisent et dépolymérisent continuellement), de nombreux moteurs moléculaires, d'agents de réticulations et de régulateurs. Bien que la structure du fuseau au niveau moléculaire soit connue, son fonctionnement reste délicat à comprendre, et nécessite la prise en compte de la dynamique de ses composants et leurs interactions. Les approches utilisées pour répondre à ces problématiques sont jusqu'à maintenant plutôt des approches in silico et in vitro. Il manque aujourd'hui une caractérisation de la mécanique du fuseau dans son contexte physiologique. Nous proposons une méthode non invasive basée sur de l'analyse d'image, combiné à une modélisation heuristique pour mesurer les paramètres mécaniques durant toute la division. Nous suivons les pôles du fuseau marqués par protéine fluorescente avec un taux acquisition rapide et une bonne résolution spatiale ce qui nous permet d'accéder aux fluctuations de longueur du fuseau in vivo. Avec la transformée de Fourier aux temps courts, nous calculons leurs densités spectrales de puissances — leurs signatures mécaniques. Ces spectres sont alors ajustés avec un modèle Kelvin — Voigt avec inertie (un ressort, un amortisseur et un terme inertiel en parallèle). Nous avons validé la méthode par des expériences numériques où nous retrouvons les évolutions des paramètres sur des données simulées et la calibration a été réalisée par l'utilisation de la rupture du fuseau induite par micro chirurgie laser ou par la génétique. Nous avons caractérisé le fuseau de l'embryon unicellulaire du nématode C. elegans. La méthaphase apparaît dominée par l'amortisseur, ce qui est cohérent avec la lente élongation du fuseau que nous observons. Mais contraste l'idée répandue de l'existence d'un mécanisme de maintien de la longueur du fuseau durant la métaphase. Au passage en anaphase, les trois paramètres mécaniques chutent, avant de réaugmenter environ 50 secondes après la transition pour réatindre un régime dominé de nouveau par l'amortisseur, ce qui suggère que les microtubules interpolaires jouent un rôle mineur durant l'élongation du fuseau en début d'anaphase. Dans la perspective de comprendre le lien entre la mécanique du fuseau et les interactions des acteurs moléculaires, nous avons partiellement supprimé un gène par sous-structure du fuseau. Nous avons alors retrouvé des comportements connus avec une perspective augmentée offerte par notre méthode. Cette méthode, ne va pas seulement permettre la compréhension fondamentale de la mécanique du fuseau, en remplaçant la modélisation du fuseau basé uniquement sur la longueur, mais aussi d'aller vers la prise en compte de la robustesse de fonctionnement du fuseau mitotique face aux défauts tel que la polyou l'aneuploïdie.
Tischer and Gergely review the cell biology behind microtubule poisons and their clinical use in cancer patients.
The dependence on history of both present and future dynamics of life is a common intuition in biology and in humanities. Historicity will be understood in terms of changes of the space of possibilities (or of “phase space”) as well as by the role of diversity in life’s structural stability and of rare events in history formation. We hint to a rigorous analysis of “path dependence” in terms of invariants and invariance preserving transformations, as it may be found also in physics, while departing from the physico-mathematical analyses. The idea is that the (relative or historicized) invariant traces of the past under organismal or ecosystemic transformations contribute to the understanding (or the “theoretical determination”) of present and future states of affairs. This yields a peculiar form of unpredictability (or randomness) in biology, at the core of novelty formation: the changes of observables and pertinent parameters may depend also on past events. In particular, in relation to the properties of synchronic measurement in physics, the relevance of diachronic measurement in biology is highlighted. This analysis may a fortiori apply to cognitive and historical human dynamics, while allowing to investigate some general properties of historicity in biology.
A microtubule-based bipolar spindle is required for error-free chromosome segregation during cell division. In this review I discuss the molecular mechanisms required for the assembly of this dynamic micrometer-scale structure in animal cells.
In dividing vertebrate cells multiple microtubules must connect to mitotic kinetochores in a highly stereotypical manner, with each sister kinetochore forming microtubule attachments to only one spindle pole. The exact sequence of events by which this goal is achieved varies considerably from cell to cell because of the variable locations of kinetochores and spindle poles, and randomness of initial microtubule attachments. These chance encounters with the kinetochores nonetheless ultimately lead to the desired outcome with high fidelity and in a limited time frame, providing one of the most startling examples of biological self-organization. This chapter discusses mechanisms that contribute to accurate chromosome segregation by helping dividing cells to avoid and resolve improper microtubule attachments.
We present a variety of new architectural features and training procedures that we apply to the generative adversarial networks (GANs) framework. We focus on two applications of GANs: semi-supervised learning, and the generation of images that humans find visually realistic. Unlike most work on generative models, our primary goal is not to train a model that assigns high likelihood to test data, nor do we require the model to be able to learn well without using any labels. Using our new techniques, we achieve state-of-the-art results in semi-supervised classification on MNIST, CIFAR-10 and SVHN. The generated images are of high quality as confirmed by a visual Turing test: our model generates MNIST samples that humans cannot distinguish from real data, and CIFAR-10 samples that yield a human error rate of 21.3%. We also present ImageNet samples with unprecedented resolution and show that our methods enable the model to learn recognizable features of ImageNet classes.
Precise positioning of the mitotic spindle is important for specifying the plane of cell division, which in turn determines how the cytoplasmic contents of the mother cell are partitioned into the daughter cells, and how the daughters are positioned within the tissue. During metaphase in the early C. elegans embryo, the spindle is aligned and centered on the anterior-posterior axis by a microtubule-dependent machinery that exerts restoring forces when the spindle is displaced from the center. To investigate the accuracy and stability of centering, we tracked the position and orientation of the mitotic spindle during the first cell division with high temporal and spatial resolution. We found that the precision is remarkably high: the cell-to-cell variation in the transverse position of the center of the spindle during metaphase, as measured by the standard deviation, was only 1.5% of the length of the short axis of the cell. Spindle position is also very stable: the standard deviation of the fluctuations in transverse spindle position during metaphase was only 0.5% of the short axis of the cell. Assuming that stability is limited by number fluctuations, we infer that the centering machinery has a large number of independent force generating units, on the order of one thousand, consistent with the several thousand of astral microtubules in these cells. Astral microtubules grow out from the two spindle poles, contact and push against the cell cortex, and then shrink back shortly thereafter. The high stability of centering can be accounted for quantitatively if the astral microtubules buckle as they exert compressive, pushing forces onto the cortex. We thus propose that the large number of microtubules in the asters provides a highly precise mechanism for positioning the spindle while its assembly is completed during metaphase prior to the onset of anaphase.
The cell cycle is an evolutionarily conserved process necessary for mammalian cell growth and development. Because cell-cycle aberrations are a hallmark of cancer, this process has been the target of anti-cancer therapeutics for decades. However, despite numerous clinical trials, cell-cycle-targeting agents have generally failed in the clinic. This review briefly examines past cell-cycle-targeted therapeutics and outlines how experience with these agents has provided valuable insight to refine and improve anti-mitotic strategies. An overview of emerging anti-mitotic approaches with promising pre-clinical results is provided, and the concept of exploiting the genomic instability of tumor cells through therapeutic inhibition of mitotic checkpoints is discussed. We believe this strategy has a high likelihood of success given its potential to enhance therapeutic index by targeting tumor-specific vulnerabilities. This reasoning stimulated our development of novel inhibitors targeting the critical regulators of genomic stability and the mitotic checkpoint: AURKA, PLK4, and Mps1/TTK.
Observer bias and other "experimenter effects" occur when researchers' expectations influence study outcome. These biases are strongest when researchers expect a particular result, are measuring subjective variables, and have an incentive to produce data that confirm predictions. To minimize bias, it is good practice to work "blind," meaning that experimenters are unaware of the identity or treatment group of their subjects while conducting research. Here, using text mining and a literature review, we find evidence that blind protocols are uncommon in the life sciences and that nonblind studies tend to report higher effect sizes and more significant p-values. We discuss methods to minimize bias and urge researchers, editors, and peer reviewers to keep blind protocols in mind.
We propose a new framework for estimating generative models via an
adversarial process, in which we simultaneously train two models: a generative
model G that captures the data distribution, and a discriminative model D that
estimates the probability that a sample came from the training data rather than
G. The training procedure for G is to maximize the probability of D making a
mistake. This framework corresponds to a minimax two-player game. In the space
of arbitrary functions G and D, a unique solution exists, with G recovering the
training data distribution and D equal to 1/2 everywhere. In the case where G
and D are defined by multilayer perceptrons, the entire system can be trained
with backpropagation. There is no need for any Markov chains or unrolled
approximate inference networks during either training or generation of samples.
Experiments demonstrate the potential of the framework through qualitative and
quantitative evaluation of the generated samples.
Quantitative microscopy relies on imaging of large cell numbers but is often hampered by time-consuming manual selection of specific cells. The 'Micropilot' software automatically detects cells of interest and launches complex imaging experiments including three-dimensional multicolor time-lapse or fluorescence recovery after photobleaching in live cells. In three independent experimental setups this allowed us to statistically analyze biological processes in detail and is thus a powerful tool for systems biology.
Probe photobleaching and a specimen's sensitivity to phototoxicity severely limit the number of possible excitation cycles in time-lapse fluorescent microscopy experiments. Consequently, when a study of cellular processes requires measurements over hours or days, temporal resolution is limited, and spontaneous or rapid events may be missed, thus limiting conclusions about transduction events. We have developed ALISSA, a design framework and reference implementation for an automated live-cell imaging system for signal transduction analysis. It allows an adaptation of image modalities and laser resources tailored to the biological process, and thereby extends temporal resolution from minutes to seconds. The system employs online image analysis to detect cellular events that are then used to exercise microscope control. It consists of a reusable image analysis software for cell segmentation, tracking, and time series extraction, and a measurement-specific process control software that can be easily adapted to various biological settings. We have applied the ALISSA framework to the analysis of apoptosis as a demonstration case for slow onset and rapid execution signaling. The demonstration provides a clear proof-of-concept for ALISSA, and offers guidelines for its application in a broad spectrum of signal transduction studies.
Expression of cyclins at the translational level is generally studied by immunoblotting lysates of cells synchronized in the cycle. Most methods used to synchronize transformed cells induce growth imbalance. The aim of the present study was to analyze levels of cyclins B1, A, E, and D3 in the respective phases of the cycle in synchronized human leukemic MOLT-4 cells, correlate them with total cellular protein content (reflecting growth imbalance), and compare the synchronized cells with cells from unperturbed, asynchronous cultures. Expression of cyclins detected immunocytochemically in individual permeabilized cells was analyzed by multiparameter flow cytometry, which made it possible to relate position of the cell in the cell cycle with cyclin expression. Cells synchronized at the G1-S boundary by thymidine, mimosine, or aphidicolin had about 40% increased total protein and 4-5 fold higher levels of cyclins E and B1 compared to their G1 counterparts from unperturbed cultures. Expression of cyclin A in synchronized cells was 2-fold higher, while expression of cyclin D3 was essentially unaltered. The synchronized cells traversing S phase after release from the block had elevated but decreasing levels of cyclins E, B1, and A. Although the cyclin expression of cells reentering G1 was similar to that of their counterparts from asynchronous cultures, the total protein content was still elevated by about 30%. The data indicate that due to different degrees of imbalance in total protein and individual cyclin content, levels of cyclins detected by immunoblotting of cell lysates from synchronized cultures may not be representative of their expression in unperturbed cells. The elevated level of cyclin B1 in the cells arrested at the G1-S boundary may reflect the increased half-life of this protein, stabilized as the result of the overexpression of cyclin E.
The visual allure of microscopy makes it an intuitively powerful research tool. Intuition, however, can easily obscure or distort the reality of the information contained in an image. Common cognitive biases, combined with institutional pressures that reward positive research results, can quickly skew a microscopy project towards upholding, rather than rigorously challenging, a hypothesis. The impact of these biases on a variety of research topics is well known. What might be less appreciated are the many forms in which bias can permeate a microscopy experiment. Even well-intentioned researchers are susceptible to bias, which must therefore be actively recognized to be mitigated. Importantly, although image quantification has increasingly become an expectation, ostensibly to confront subtle biases, it is not a guarantee against bias and cannot alone shield an experiment from cognitive distortions. Here, we provide illustrative examples of the insidiously pervasive nature of bias in microscopy experiments – from initial experimental design to image acquisition, analysis and data interpretation. We then provide suggestions that can serve as guard rails against bias.
Standfirst:
Artificial intelligence and machine learning techniques are breaking into biomedical research and health care, which importantly includes cancer research and oncology, where the potential applications are vast. These include detection and diagnosis of cancer, subtype classification, optimization of cancer treatment and identification of new therapeutic targets in drug discovery. While big data used to train machine learning models may already exist, leveraging this opportunity to realize the full promise of artificial intelligence in both the cancer research space and the clinical space will first require significant obstacles to be surmounted. In this Viewpoint article, we asked four experts for their opinions on how we can begin to implement artificial intelligence while ensuring standards are maintained so as transform cancer diagnosis and the prognosis and treatment of patients with cancer and to drive biological discovery.
Aneuploidy is a ubiquitous feature of human tumors, but the acquisition of aneuploidy typically antagonizes cellular fitness. To investigate how aneuploidy could contribute to tumor growth, we triggered periods of chromosomal instability (CIN) in human cells and then exposed them to different culture environments. We discovered that transient CIN reproducibly accelerates the acquisition of resistance to anti-cancer therapies. Single-cell sequencing revealed that these resistant populations develop recurrent aneuploidies, and independently deriving one chromosome-loss event that was frequently observed in paclitaxel-resistant cells was sufficient to decrease paclitaxel sensitivity. Finally, we demonstrated that intrinsic levels of CIN correlate with poor responses to numerous therapies in human tumors. Our results show that, although CIN generally decreases cancer cell fitness, it also provides phenotypic plasticity to cancer cells that can allow them to adapt to diverse stressful environments. Moreover, our findings suggest that aneuploidy may function as an under-explored cause of therapy failure.
Artificial intelligence (AI) is rapidly reshaping cancer research and personalized clinical care. Availability of high-dimensionality datasets coupled with advances in high-performance computing, as well as innovative deep learning architectures, has led to an explosion of AI use in various aspects of oncology research. These applications range from detection and classification of cancer, to molecular characterization of tumors and their microenvironment, to drug discovery and repurposing, to predicting treatment outcomes for patients. As these advances start penetrating the clinic, we foresee a shifting paradigm in cancer care becoming strongly driven by AI.
Significance: AI has the potential to dramatically affect nearly all aspects of oncology—from enhancing diagnosis to personalizing treatment and discovering novel anticancer drugs. Here, we review the recent enormous progress in the application of AI to oncology, highlight limitations and pitfalls, and chart a path for adoption of AI in the cancer clinic.
Since its renaissance, deep learning (DL) has been widely used in various medical imaging tasks and has achieved remarkable success in many medical imaging applications, thereby propelling us into the so-called artificial intelligence (AI) era. It is known that the success of AI is mostly attributed to the availability of big data with annotations for a single task and the advances in high-performance computing. However, medical imaging presents unique challenges that confront DL approaches. In this survey article, we first present traits of medical imaging, highlight both clinical needs and technical challenges in medical imaging, and describe how emerging trends in DL are addressing these issues. We cover the topics of network architecture, sparse and noisy labels, federating learning, interpretability, uncertainty quantification, and so on. Then, we present several case studies that are commonly found in clinical practice, including digital pathology and chest, brain, cardiovascular, and abdominal imaging. Rather than presenting an exhaustive literature survey, we instead describe some prominent research highlights related to these case study applications. We conclude with a discussion and presentation of promising future directions.
Mitotic spindle microtubules (MTs) undergo continuous poleward flux, whose driving force and function in humans remain unclear. Here, we combined loss‐of‐function screenings with analysis of MT‐dynamics in human cells to investigate the molecular mechanisms underlying MT‐flux. We report that kinesin‐7/CENP‐E at kinetochores (KTs) is the predominant driver of MT‐flux in early prometaphase, while kinesin‐4/KIF4A on chromosome arms facilitates MT‐flux during late prometaphase and metaphase. Both these activities work in coordination with kinesin‐5/EG5 and kinesin‐12/KIF15, and our data suggest that the MT‐flux driving force is transmitted from non‐KT‐MTs to KT‐MTs by the MT couplers HSET and NuMA. Additionally, we found that the MT‐flux rate correlates with spindle length, and this correlation depends on the establishment of stable end‐on KT‐MT attachments. Strikingly, we find that MT‐flux is required to regulate spindle length by counteracting kinesin 13/MCAK‐dependent MT‐depolymerization. Thus, our study unveils the long‐sought mechanism of MT‐flux in human cells as relying on the coordinated action of four kinesins to compensate for MT‐depolymerization and regulate spindle length.
Microtubules play essential roles in cellular organization, cargo transport, and chromosome segregation during cell division. During mitosis microtubules form a macromolecular structure known as the mitotic spindle that is responsible for the accurate segregation of chromosomes between the two daughter cells. This is accomplished thanks to finely tuned control of microtubule dynamics. Even small changes in microtubule dynamics during spindle formation and/or operation may lead to chromosome mis-segregation, chromosome instability and aneuploidy. These three events are directly correlated with human diseases like cancer and developmental defects. Precise measurements of microtubule dynamics in the spindle will allow us to discover new molecules involved in regulating microtubule dynamics and enable a deeper understanding of the mechanisms that underlie mitosis and cancer emergence and development. Moreover, many chemotherapeutic agents for cancer treatment are targeted to microtubules, so continued investigation of their dynamics with utmost precision will facilitate the development of new drugs. Measuring microtubule dynamics in the spindle has been a difficult task until recently. With the development of new and gentler microscopic techniques, and new computer programs, we can perform better and more accurate measurements of microtubule dynamics during mitosis.
The spindle segregates chromosomes at cell division, and its task is a mechanical one. While we have a nearly complete list of spindle components, how their molecular-scale mechanics give rise to cellular-scale spindle architecture, mechanics, and function is not yet clear. Recent in vitro and in vivo measurements bring new levels of molecular and physical control and shed light on this question. Highlighting recent findings and open questions, we introduce the molecular force generators of the spindle, and discuss how they organize microtubules into diverse architectural modules and give rise to the emergent mechanics of the mammalian spindle. Throughout, we emphasize the breadth of space and time scales at play, and the feedback between spindle architecture, dynamics, and mechanics that drives robust function.
The assembly of the mitotic spindle and the subsequent segregation of sister chromatids are based on the self-organized action of microtubule filaments, motor proteins, and other microtubule-associated proteins, which constitute the fundamental force-generating elements in the system. Many of the components in the spindle have been identified, but until recently it remained unclear how their collective behaviors resulted in such a robust bipolar structure. Here, we review the current understanding of the physics of the metaphase spindle that is only now starting to emerge.
Successive cell divisions during embryonic cleavage create increasingly smaller cells, so intracellular structures must adapt accordingly. Mitotic spindle size correlates with cell size, but the mechanisms for this scaling remain unclear. Using live cell imaging, we analyzed spindle scaling during embryo cleavage in the nematode Caenorhabditis elegans and sea urchin Paracentrotus lividus. We reveal a common scaling mechanism, where the growth rate of spindle microtubules scales with cell volume, which explains spindle shortening. Spindle assembly timing is, however, constant throughout successive divisions. Analyses in silico suggest that controlling the microtubule growth rate is sufficient to scale spindle length and maintain a constant assembly timing. We tested our in silico predictions to demonstrate that modulating cell volume or microtubule growth rate in vivo induces a proportional spindle size change. Our results suggest that scalability of the microtubule growth rate when cell size varies adapts spindle length to cell volume.
Cell synchronization is often achieved by transient inhibition of DNA replication. When cultured in the presence of such inhibitors as hydroxyurea, aphidicolin or excess of thymidine the cells that become arrested at the entrance to S-phase upon release from the block initiate progression through S then G2 and M. However, exposure to these inhibitors at concentrations commonly used to synchronize cells leads to activation of ATR and ATM protein kinases as well as phosphorylation of Ser 139 of histone H2AX. This observation of DNA damage signaling implies that synchronization of cells by these inhibitors is inducing replication stress. Thus, a caution should be exercised while interpreting data obtained with use of cells synchronized this way since they do not represent unperturbed cell populations in a natural metabolic state. This chapter critically outlines virtues and vices of most cell synchronization methods. It also presents the protocol describing an assessment of phosphorylation of Ser 139 on H2AX and activation of ATM in cells treated with aphidicolin, as a demonstrative of one of several DNA replication inhibitors that are being used for cell synchronization. Phosphorylation of Ser139 H2AX and Ser 1981ATM in individual cells is detected immunocytochemically with phospho-specific Abs and intensity of immunofluorescence is measured by flow cytometry. Concurrent measurement of cellular DNA content followed by multiparameter analysis allows one to correlate the extent of phosphorylation of these proteins in response to aphidicolin with the cell cycle phase.
We extend Generative Adversarial Networks (GANs) to the semi-supervised context by forcing the discriminator network to output class labels. We train a generative model G and a discriminator D on a dataset with inputs belonging to one of N classes. At training time, D is made to predict which of N+1 classes the input belongs to, where an extra class is added to correspond to the outputs of G. We show that this method can be used to create a more data-efficient classifier and that it allows for generating higher quality samples than a regular GAN.
High-content screening (HCS), which combines automated fluorescence microscopy with quantitative image analysis, allows the acquisition of unbiased multiparametric data at the single cell level. This approach has been used to address diverse biological questions and identify a plethora of quantitative phenotypes of varying complexity in numerous different model systems. Here, we describe some recent applications of HCS, ranging from the identification of genes required for specific biological processes to the characterization of genetic interactions. We review the steps involved in the design of useful biological assays and automated image analysis, and describe major challenges associated with each. Additionally, we highlight emerging technologies and future challenges, and discuss how the field of HCS might be enhanced in the future.
The microtubule-based metaphase spindle is subjected to forces that act in diverse orientations and over a wide range of timescales. Currently, we cannot explain how this dynamic structure generates and responds to forces while maintaining overall stability, as we have a poor understanding of its micromechanical properties. Here, we combine the use of force-calibrated needles, high-resolution microscopy, and biochemical perturbations to analyze the vertebrate metaphase spindle's timescale- and orientation-dependent viscoelastic properties. We find that spindle viscosity depends on microtubule crosslinking and density. Spindle elasticity can be linked to kinetochore and nonkinetochore microtubule rigidity, and also to spindle pole organization by kinesin-5 and dynein. These data suggest a quantitative model for the micromechanics of this cytoskeletal architecture and provide insight into how structural and functional stability is maintained in the face of forces, such as those that control spindle size and position, and can result from deformations associated with chromosome movement.
Despite our rapidly growing knowledge about the human genome, we do not know all of the genes required for some of the most basic functions of life. To start to fill this gap we developed a high-throughput phenotypic screening platform combining potent gene silencing by RNA interference, time-lapse microscopy and computational image processing. We carried out a genome-wide phenotypic profiling of each of the approximately 21,000 human protein-coding genes by two-day live imaging of fluorescently labelled chromosomes. Phenotypes were scored quantitatively by computational image processing, which allowed us to identify hundreds of human genes involved in diverse biological functions including cell division, migration and survival. As part of the Mitocheck consortium, this study provides an in-depth analysis of cell division phenotypes and makes the entire high-content data set available as a resource to the community.
Abnormal chromosome content - also known as aneuploidy - is the most common characteristic of human solid tumours. It has therefore been proposed that aneuploidy contributes to, or even drives, tumour development. The mitotic checkpoint guards against chromosome mis-segregation by delaying cell-cycle progression through mitosis until all chromosomes have successfully made spindle-microtubule attachments. Defects in the mitotic checkpoint generate aneuploidy and might facilitate tumorigenesis, but more severe disabling of checkpoint signalling is a possible anticancer strategy.
We previously reported the phenotype of depletion of polo-like kinase 1 (Plk1) using RNA interference (RNAi) and showed that p53 is stabilized in Plk1-depleted cancer cells. In this study, we further analyzed the Plk1 depletion-induced phenotype in both cancer cells and primary cells. The vector-based RNAi approach was used to evaluate the role of the p53 pathway in Plk1 depletion-induced apoptosis in cancer cells with different p53 backgrounds. Although DNA damage and cell death can occur independently of p53, p53-deficient cancer cells were much more sensitive to Plk1 depletion than cancer cells with functional p53. Next, the lentivirus-based RNAi approach was used to generate a series of Plk1 hypomorphs. In HeLa cells, two weak hypomorphs showed only slight G2/M arrest, a medium hypomorph arrested with 4N DNA content, followed later by apoptosis, and a strong Plk1 hypomorph underwent serious mitotic catastrophe. In well-synchronized HeLa cells, a medium level of Plk1 depletion caused a 2-h delay of mitotic progression, and a high degree of Plk1 depletion significantly delayed mitotic entry and completely blocked cells at mitosis. In striking contrast, normal hTERT-RPE1 and MCF10A cells were much less sensitive to Plk1 depletion than HeLa cells; no apparent cell proliferation defect or cell cycle arrest was observed after Plk1 depletion in these cells. Therefore, these data further support suggestions that Plk1 may be a feasible cancer therapy target.
Merotelic kinetochore orientation is a misattachment in which a single kinetochore binds microtubules from both spindle poles rather than just one and can produce anaphase lagging chromosomes, a major source of aneuploidy. Merotelic kinetochore orientation occurs frequently in early mitosis, does not block chromosome alignment at the metaphase plate, and is not detected by the spindle checkpoint. However, microtubules to the incorrect pole are usually significantly reduced or eliminated before anaphase. We discovered that the frequency of lagging chromosomes in anaphase is very sensitive to partial inhibition of Aurora kinase activity by ZM447439 at a dose, 3 microM, that has little effect on histone phosphorylation, metaphase chromosome alignment, and cytokinesis in PtK1 cells. Partial Aurora kinase inhibition increased the frequency of merotelic kinetochores in late metaphase, and the fraction of microtubules to the incorrect pole. Measurements of fluorescence dissipation after photoactivation showed that kinetochore-microtubule turnover in prometaphase is substantially suppressed by partial Aurora kinase inhibition. Our results support a preanaphase correction mechanism for merotelic attachments in which correct plus-end attachments are pulled away from high concentrations of Aurora B at the inner centromere, and incorrect merotelic attachments are destabilized by being pulled toward the inner centromere.
Asymmetric division of the C. elegans zygote is due to the posterior-directed movement of the mitotic spindle during metaphase and anaphase. During this movement along the anterior-posterior axis, the spindle oscillates transversely. These motions are thought to be driven by a force-generating complex-possibly containing the motor protein cytoplasmic dynein-that is located at the cell cortex and pulls on microtubules growing out from the spindle poles. A theoretical analysis indicates that the oscillations might arise from mechanical coordination of the force-generating motors, and this coordination is mediated by the load dependence of the motors' detachment from the microtubules. The model predicts that the motor activity must exceed a threshold for oscillations to occur.
We have tested the existence of a threshold by using RNA interference to gradually reduce the levels of dynein light intermediate chain as well as GPR-1 and GPR-2 that are involved in the G protein-mediated regulation of the force generators. We found an abrupt cessation of oscillations as expected if the motor activity dropped below a threshold. Furthermore, we can account for the complex choreography of the mitotic spindle-the precise temporal coordination of the buildup and die-down of the transverse oscillations with the posterior displacement-by a gradual increase in the processivity of a single type of motor machinery during metaphase and anaphase.
The agreement between our results and modeling suggests that the force generators themselves have the intrinsic capability of generating oscillations when opposing forces exceed a threshold.