By David M. Evans
July 20, 2005 | The goal of drug discovery is to identify drug compounds that interact reliably and efficiently with an intended target, producing a well-defined therapeutic response and avoiding off-target binding that could cause toxicity and unwanted side effects. The more information researchers have about a drug target, the better able they are to develop potent and safe drug candidates.
The combination of two powerful new technologies that allow scientists to probe gene function at the level of the single cell and to visualize the cellular and subcellular effects of gene silencing provides an innovative platform for identifying and validating novel drug targets. By merging RNA interference (RNAi) technology and high-content cellular analysis, researchers are able to perform high-throughput phenotype profiling, linking gene expression to biochemical signaling pathways in the cell and, ultimately, to cell behavior.
RNAi is a molecular technique used to silence the expression of specific genes, which allows biologists to dissect disease-related biochemical pathways and to identify disease-linked drug targets (see “Running Interference,” December 2004 Bio•IT World, page 22). RNAi-mediated gene silencing involves introducing small RNA sequences into cultured cells, often via automated, robotic systems.
RNAi results can be measured using a range of cellular and subcellular parameters, such as alterations in cell cycle, cellular changes associated with apoptosis (programmed cell death), localization and transport of signaling factors, or changes in the size and morphology of intracellular compartments and organelles. Detecting and quantifying these molecular endpoints in real time in individual cells provides important contextual information about cell function and enables predictions about the effects of a drug compound on cellular pathways and cell survival.
| ||RGB IMAGE: Original fused image |
from the IN Cell 1000, showing
Rhodamine-labeled siRNA (red)
against green fluorescent protein
(green) localizing to the nuclear
membrane. The nucleus is
stained with DAPI (blue).
To fully appreciate the results produced by an RNAi assay and to be able to extract the information needed to establish a direct link between gene silencing and phenotypic outcome requires the detection of many different physicochemical events in individual cells and the ability to compare results between cells and cell types. Automated microscopy for cellular imaging combined with advanced computational algorithms is enabling researchers to capture, store, and analyze thousands of images from a single screening assay, quantify the cellular and subcellular events captured in the images, and make sense of the multiparametric data sets generated on image analysis.
Automated high-content screening (HCS) systems that integrate high-resolution, confocal imaging, and online, multi-wavelength fluorescent image analysis (exploiting the advantages of novel dye technologies and multiplexed assay formats) enable high-throughput screening of live-cell assays in 96- or 384-well formats. HCS provides the imaging and analytical power needed to detect even subtle changes in cell activity, morphology, and localization and to differentiate phenotypic endpoints.
Essential to the widespread acceptance and utility of high-content image analysis in drug discovery is its ease of use. Biologists with no particular programming knowledge need to be able to generate images, perform measurements using a wide variety of data parameters, and create customized analytical protocols capable of mining the data to answer questions relevant to very specialized areas of research.
Visualizing Gene Function
Once a researcher has successfully shut down a particular disease-related gene, he or she can zoom in on individual cells using a sensitive, high-speed camera to observe what happens as each cell reacts to the change in gene expression.
Cells may begin rounding up, a change in shape associated with the early stages of apoptosis. Or a fluorescently tagged cell-surface receptor may be activated, allowing the researcher to watch in real time as a receptor complex moves across the cell membrane and delivers its payload to the cytoplasm. Or an intracellular signaling molecule tagged with green fluorescent protein could light up as it interacts with another component of the signaling pathway, perhaps relaying instructions to the nucleus to activate or deactivate a particular gene.
The invaluable aspect of this technology is not the astounding visual images it produces, but rather the abundant and diverse data that can be distilled from those images — data that afford a better understanding of what is happening in the cell in response to an event or stimulus.
Each image can generate nearly 4 megabytes of data. Consider that the system produces multiple images for each well, with 384 wells per plate, and the massive volume of data available becomes evident. However, using the advanced analytical tools available for extracting and analyzing the data sets, researchers need not be overwhelmed by the amount of data. Imagine a biologist browsing the aisles of a supermarket with a recipe in hand, selecting only the specific ingredients needed for the recipe and perhaps sampling some others that could be used to create variations of the final product or could be saved for future applications.
A Range of Real-World Applications
GE Healthcare’s IN Cell Analyzer 3000 is an automated microscopy system capable of high-throughput subcellular imaging and online image analysis. It combines a line-scanning confocal imager with dual laser light scanners and three charge-coupled device cameras for simultaneous, multi-wavelength image acquisition. By controlling environmental factors such as temperature, humidity, and carbon dioxide, it enables robotic screening in live-cell assays.
High-speed imaging algorithms drive image acquisition and analysis, allowing the system to process a 96-well microplate in less than 3 minutes and a 384-well plate in under 10 minutes. The IN Cell Developer Toolbox software package enables measurements of a broad range of cellular events, including receptor activation, toxicity and apoptosis, signaling pathway activation, cell cycle status, organelle integrity and translocation, neuronal morphology, cell morphology, and overall cellular health.
In a typical HCS scenario performed in a large pharmaceutical company, the optimal infrastructure for storing and processing images so that the data will be accessible to multiple users across a global organization would involve an interface between the imaging system and a centralized server. The server would periodically poll the imager and, if it finds stored images, pull them off the system and store them on the server. In this way, the images would be accessible to far-flung research groups that can then query the raw image data, manipulate the data as their needs demand, and extract information relevant to their research focus.
Software tools specifically designed for high-content image analysis, such as Developer Toolbox, provide a selection of segmentation, preprocessing, and postprocessing tools. They give the user the flexibility to define the parameters, calculations, and output formats desired, to control the sequence of steps in an analytical routine, and to incorporate user-defined macros as needed to customize an analysis protocol.
Biologists can exploit the flexibility of the data analysis software to design their own algorithms from the building blocks provided, customizing the data-mining tools to optimize data output. For example, one researcher may want to visualize the cell nucleus and observe microtubule formation and mitotic events, while another might seek discrimination of the outer cell membrane or want to compare fluorescent emissions at different wavelengths to distinguish between signals generated simultaneously in a multiplexed assay.
The potential applications of HCS are virtually limitless. When run at high throughput, a cellular imaging system is ideal for use with neurite growth assays, for example. In these experiments, neurons are grown in culture in microplates and are exposed to a differentiation factor. By evaluating images from multiple sectors of each well, one can determine how many cells of a given cell type respond to the differentiation factor by sending out neurites, and one can quantify and compare the length of these axonal extensions.
Performing RNAi assays in cancer cells is another good example of the value of high-throughput HCS. By co-culturing two different cell types — a cancer cell and a normal cell — and knocking down the expression of a single gene in each well, the assay can reveal which loss of gene activity kills the cancer cell while leaving the normal cell unaffected, thereby identifying a potential target for drug discovery.
Lower-throughput HCS enables higher-resolution image acquisition and allows for visualization of subcellular events such as chromatin condensation, organelle redistribution, and mitochondrial activity.
The potential applications and ultimate value of HCS and cellular image analysis are limited only by the imagination and expertise of the drug discovery groups using them to probe gene function and cell behavior.
David M. Evans, Ph.D., is head of drug discovery at the Cancer Drug Development Lab, Translational Genomics Research Institute, Gaithersburg, Md.