High-Content Screening for Quantitative Cell Biology

Label-free detection of cellular drug responses by high-throughput bright-field imaging and machine learning

In the last decade, high-content screening based on multivariate single-cell imaging has been proven effective in drug discovery to evaluate drug-induced phenotypic variations. Unfortunately, this method inherently requires fluorescent labeling which has several drawbacks. Here we present a label-free method for evaluating cellular drug responses only by high-throughput bright-field imaging with the aid of machine learning algorithms. Specifically, we performed high-throughput bright-field imaging of numerous drug-treated and -untreated cells (N = ~240,000) by optofluidic time-stretch microscopy with high throughput up to 10,000 cells/s and applied machine learning to the cell images to identify their morphological variations which are too subtle for human eyes to detect. Consequently, we achieved a high accuracy of 92% in distinguishing drug-treated and -untreated cells without the need for labeling. Furthermore, we also demonstrated that dose-dependent, drug-induced morphological change from different experiments can be inferred from the classification accuracy of a single classification model. Our work lays the groundwork for label-free drug screening in pharmaceutical science and industry.

Data-analysis strategies for image-based cell profiling

Image-based cell profiling is a high-throughput strategy for the quantification of phenotypic differences among a variety of cell populations. It paves the way to studying biological systems on a large scale by using chemical and genetic perturbations. The general workflow for this technology involves image acquisition with high-throughput microscopy systems and subsequent image processing and analysis. Here, we introduce the steps required to create high-quality image-based (i.e., morphological) profiles from a collection of microscopy images. We recommend techniques that have proven useful in each stage of the data analysis process, on the basis of the experience of 20 laboratories worldwide that are refining their image-based cell-profiling methodologies in pursuit of biological discovery. The recommended techniques cover alternatives that may suit various biological goals, experimental designs, and laboratories' preferences.

Workflows for microscopy image analysis and cellular phenotyping

In large scale biological experiments, like high-throughput or high-content cellular screening, the amount and the complexity of images to be analyzed are steadily increasing. To handle and process these images, well defined image processing and analysis steps need to be performed by applying dedicated workflows. Multiple software tools have emerged with the aim to facilitate creation of such workflows by integrating existing methods, tools, and routines, and by adapting them to different applications and questions, as well as making them reusable and interchangeable. In this review, we describe workflow systems for the integration of microscopy image analysis techniques with focus on KNIME and Galaxy.

Live-single-cell phenotypic cancer biomarkers-future role in precision oncology?

The promise of precision and personalized medicine is rooted in accurate, highly sensitive, and specific disease biomarkers. This is particularly true for cancer-a disease characterized by marked tumor heterogeneity and diverse molecular signatures. Although thousands of biomarkers have been described, only a very small number have been successfully translated into clinical use. Undoubtedly, there is need for rapid, quantitative, and more cost effective biomarkers for tumor diagnosis and prognosis, to allow for better risk stratification and aid clinicians in making personalized treatment decisions. This is particularly true for cancers where specific biomarkers are either not available (e.g., renal cell carcinoma) or where current biomarkers tend to classify individuals into broad risk categories unable to accurately assess individual tumor aggressiveness and adverse pathology potential (e.g., prostate cancer), thereby leading to problems of over-diagnosis and over-treatment of indolent cancer and under-treatment of aggressive cancer. This perspective highlights an emerging class of cancer biomarkers-live-single-cell phenotypic biomarkers, as compared to genomic biomarkers, and their potential application for cancer diagnosis, risk-stratification, and prognosis.

Phenotypic assays for Mycobacterium tuberculosis infection

Tuberculosis (TB) is still a major global threat, killing more than one million persons each year. With the constant increase of Mycobacterium tuberculosis strains resistant to first- and second-line drugs, there is an urgent need for the development of new drugs to control the propagation of TB. Although screenings of small molecules on axenic M. tuberculosis cultures were successful for the identification of novel putative anti-TB drugs, new drugs in the development pipeline remains scarce. Host-directed therapy may represent an alternative for drug development against TB. Indeed, M. tuberculosis has multiple specific interactions within host phagocytes, which may be targeted by small molecules. In order to enable drug discovery strategies against microbes residing within host macrophages, we developed multiple fluorescence-based HT/CS phenotypic assays monitoring the intracellular replication of M. tuberculosis as well as its intracellular trafficking. What we propose here is a population-based, multi-parametric analysis pipeline that can be used to monitor the intracellular fate of M. tuberculosis and the dynamics of cellular events such as phagosomal maturation (acidification and permeabilization), zinc poisoning system or lipid body accumulation. Such analysis allows the quantification of biological events considering the host-pathogen interplay and may thus be derived to other intracellular pathogens. © 2017 International Society for Advancement of Cytometry.

Automated analysis of high-content microscopy data with deep learning

Existing computational pipelines for quantitative analysis of high‐content microscopy data rely on traditional machine learning approaches that fail to accurately classify more than a single dataset without substantial tuning and training, requiring extensive analysis. Here, we demonstrate that the application of deep learning to biological image data can overcome the pitfalls associated with conventional machine learning classifiers. Using a deep convolutional neural network (DeepLoc) to analyze yeast cell images, we show improved performance over traditional approaches in the automated classification of protein subcellular localization. We also demonstrate the ability of DeepLoc to classify highly divergent image sets, including images of pheromone‐arrested cells with abnormal cellular morphology, as well as images generated in different genetic backgrounds and in different laboratories. We offer an open‐source implementation that enables updating DeepLoc on new microscopy datasets. This study highlights deep learning as an important tool for the expedited analysis of high‐content microscopy data.

High content analysis in amyotrophic lateral sclerosis

Amyotrophic lateral sclerosis (ALS) is a devastating disease characterized by the progressive loss of motor neurons. Neurons, astrocytes, oligodendrocytes and microglial cells all undergo pathological modifications in the onset and progression of ALS. A number of genes involved in the etiopathology of the disease have been identified, but a complete understanding of the molecular mechanisms of ALS has yet to be determined. Currently, people affected by ALS have a life expectancy of only two to five years from diagnosis. The search for a treatment has been slow and mostly unsuccessful, leaving patients in desperate need of better therapies. Until recently, most pre-clinical studies utilized the available ALS animal models. In the past years, the development of new protocols for isolation of patient cells and differentiation into relevant cell types has provided new tools to model ALS, potentially more relevant to the disease itself as they directly come from patients. The use of stem cells is showing promise to facilitate ALS research by expanding our understanding of the disease and help to identify potential new therapeutic targets and therapies to help patients. Advancements in high content analysis (HCA) have the power to contribute to move ALS research forward by combining automated image acquisition along with digital image analysis. With modern HCA machines it is possible, in a period of just a few hours, to observe changes in morphology and survival of cells, under the stimulation of hundreds, if not thousands of drugs and compounds. In this article, we will summarize the major molecular and cellular hallmarks of ALS, describe the advancements provided by the in vitro models developed in the last few years, and review the studies that have applied HCA to the ALS field to date.

Genetically Encoded Calcium Indicators as Probes to Assess the Role of Calcium Channels in Disease and for High-Throughput Drug Discovery

The calcium ion (Ca²⁺) is an important signaling molecule implicated in many cellular processes, and the remodeling of Ca²⁺ homeostasis is a feature of a variety of pathologies. Typical methods to assess Ca²⁺ signaling in cells often employ small molecule fluorescent dyes, which are sometimes poorly suited to certain applications such as assessment of cellular processes, which occur over long periods (hours or days) or in vivo experiments. Genetically encoded calcium indicators are a set of tools available for the measurement of Ca²⁺ changes in the cytosol and subcellular compartments, which circumvent some of the inherent limitations of small molecule Ca²⁺ probes. Recent advances in genetically encoded calcium sensors have greatly increased their ability to provide reliable monitoring of Ca²⁺ changes in mammalian cells. New genetically encoded calcium indicators have diverse options in terms of targeting, Ca²⁺ affinity and fluorescence spectra, and this will further enhance their potential use in high-throughput drug discovery and other assays. This review will outline the methods available for Ca²⁺ measurement in cells, with a focus on genetically encoded calcium sensors. How these sensors will improve our understanding of the deregulation of Ca²⁺ handling in disease and their application to high-throughput identification of drug leads will also be discussed.

Growth condition dependency is the major cause of non-responsiveness upon genetic perturbation

Investigating the role and interplay between individual proteins in biological processes is often performed by assessing the functional consequences of gene inactivation or removal. Depending on the sensitivity of the assay used for determining phenotype, between 66% (growth) and 53% (gene expression) of Saccharomyces cerevisiae gene deletion strains show no defect when analyzed under a single condition. Although it is well known that this non-responsive behavior is caused by different types of redundancy mechanisms or by growth condition/cell type dependency, it is not known what the relative contribution of these different causes is. Understanding the underlying causes of and their relative contribution to non-responsive behavior upon genetic perturbation is extremely important for designing efficient strategies aimed at elucidating gene function and unraveling complex cellular systems. Here, we provide a systematic classification of the underlying causes of and their relative contribution to non-responsive behavior upon gene deletion. The overall contribution of redundancy to non-responsive behavior is estimated at 29%, of which approximately 17% is due to homology-based redundancy and 12% is due to pathway-based redundancy. The major determinant of non-responsiveness is condition dependency (71%). For approximately 14% of protein complexes, just-in-time assembly can be put forward as a potential mechanistic explanation for how proteins can be regulated in a condition dependent manner. Taken together, the results underscore the large contribution of growth condition requirement to non-responsive behavior, which needs to be taken into account for strategies aimed at determining gene function. The classification provided here, can also be further harnessed in systematic analyses of complex cellular systems.

A Critical and Comparative Review of Fluorescent Tools for Live-Cell Imaging

Fluorescent tools have revolutionized our ability to probe biological dynamics, particularly at the cellular level. Fluorescent sensors have been developed on several platforms, utilizing either small-molecule dyes or fluorescent proteins, to monitor proteins, RNA, DNA, small molecules, and even cellular properties, such as pH and membrane potential. We briefly summarize the impressive history of tool development for these various applications and then discuss the most recent noteworthy developments in more detail. Particular emphasis is placed on tools suitable for single-cell analysis and especially live-cell imaging applications. Finally, we discuss prominent areas of need in future fluorescent tool development-specifically, advancing our capability to analyze and integrate the plethora of high-content data generated by fluorescence imaging.

Machine learning and computer vision approaches for phenotypic profiling

Grys et al. review computer vision and machine-learning methods that have been applied to phenotypic profiling of image-based data. Descriptions are provided for segmentation, feature extraction, selection, and dimensionality reduction, as well as clustering, outlier detection, and classification of data.

With recent advances in high-throughput, automated microscopy, there has been an increased demand for effective computational strategies to analyze large-scale, image-based data. To this end, computer vision approaches have been applied to cell segmentation and feature extraction, whereas machine-learning approaches have been developed to aid in phenotypic classification and clustering of data acquired from biological images. Here, we provide an overview of the commonly used computer vision and machine-learning methods for generating and categorizing phenotypic profiles, highlighting the general biological utility of each approach.

Effect-Size Measures as Descriptors of Assay Quality in High-Content Screening: A Brief Review of Some Available Methodologies

The field of high-content screening (HCS) typically uses measures of screen quality conceived for fairly straightforward high-throughput screening (HTS) scenarios. However, in contrast to HTS, image-based HCS systems rely on multidimensional readouts reporting biological responses associated with complex cellular phenotypes. Not only is the dimensionality in which the screens operate higher, but also the scale of the individual features describing the quantified phenotypic changes is often smaller than what is seen in one-dimensional HTS platforms. Therefore, the use of HTS-type quality measures to characterize HCS screens may severely underestimate the detection power of the assays used, and it may mislead the screening scientists regarding the necessary screen optimization. Also, traditional HTS-based measures are typically reported without any estimation of precision, which makes them unsuitable for computation of confidence intervals or for meta-analysis. This review summarizes the well-established statistical techniques for reporting effect sizes and argues that this broadly accepted methodology could be seamlessly integrated into HCS quality reporting, supplanting or even replacing measures such as Z' or S_w (assay window).

Single dish gradient screening of small molecule localization

Understanding trafficking in cells and tissues is one of the most critical steps in exploring the mechanisms and modes of action (MOAs) of a small molecule. Typically, deciphering the role of concentration presents one of the most difficult challenges associated with this task. Herein, we present a practical solution to this problem by developing concentration gradients within single dishes of cells. We demonstrate the method by evaluating fluorescently-labelled probes developed from two classes of natural products that have been identified as potential anti-cancer leads by STORM super-resolution microscopy.