February 1, 2015 (Vol. 35, No. 3)
Ken Doyle, Ph.D.
Better Molecular Labels and Other Advances Give a Richer Impression of Cellular Life
Many molecular and cell biologists share the ultimate dream—being able to follow individual biological molecules within a cell in real time and observe how they affect the cell’s morphology and other characteristics. This dream is now becoming a reality thanks to a combination of techniques: molecular labeling, high-resolution microscopy, robotics, and sophisticated image analysis software.
Used in combination to establish relationships between molecular events and cellular phenotypes, these techniques constitute what researchers call high-content analysis (HCA). An alternative name, high-content screening (HCS), has gained currency among scientists in the drug discovery industry. In general, HCS refers to more high-throughput-oriented HCA.
Various HCA refinements are being developed, as CHI’s annual High-Content Analysis meeting demonstrated. This event, which took place January 26–27 in San Diego, provided an opportunity to discuss new assays and probes, more advanced image analysis and data management, high-content screening of three-dimensional models, and ultra-high-resolution imaging.
Also on the agenda were the implications for various HCA-related issues. These ranged from the specifics of predictive toxicology and preclinical safety to more general topics, such as methods for increasing screening throughput and innovations in data analysis.
According to Mikael Persson, Ph.D., primary exploratory toxicologist at Lundbeck, predictive toxicology has two goals: bringing safe treatments to patients and reducing the attrition rates of new drugs. To achieve these goals, predictive toxicology relies on cell models that permit hundreds of drugs with known clinical toxicity profiles to be screened. With these models, the functional characteristics that determine success or failure can be identified. “The main idea,” said Dr. Persson, “is to learn from the past to better predict the future.”
Dr. Persson added that predictive toxicology is not likely to replace testing in animals. However, careful selection of endpoints in vitro can lead to better predictive success for some human toxicities than routine animal testing. For example, most HCS assays for prediction of liver toxicity can identify 40–60% of drugs that exhibit hepatotoxicity in humans with very few false positives. “More importantly,” continued Dr. Persson, “they also identify some of the rare toxicities that routine rodent toxicity testing does not.”
Better prediction of drug-induced liver damage in humans provides a better understanding of the safety margins required during drug development and clinical trials. “We can use this information to screen thousands of new compounds in our drug discovery projects and to guide chemists in actively choosing or generating the safest compounds,” explained Dr. Persson, who indicated that the assay developed in his group measures cytotoxicity, mitochondrial toxicity, and lysosomal activation.
Dr. Persson stated that HCS is being widely adopted in drug discovery for several reasons. Compared to traditional biochemical toxicity assays, HCS allows customized, multiparametric measurements at the single-cell level. This level of sensitivity allows researchers to tailor their measurements to endpoints and toxicity parameters most relevant to their study. In addition, HCS methods allow the measurement of several endpoints simultaneously.
Another advantage of HCS is speed. “A typical assay in a 96-well plate with 10 fields of view from each well and using four different dyes can usually be imaged and analyzed within 20 minutes,” concluded Dr. Persson.
Imaging Hardware and Software
pioneered the development of high-resolution imaging systems. Nikon’s primary focus is on stem cells, a specialty that is consistent with “expertise in live-cell imaging technologies built upon the Ti-E inverted microscope and Perfect Focus system,” said Ned Jastromb, senior application product manager.
Jastromb presented a case study that showcased the use of Nikon’s HCA platform to study phenotypic stem-cell differentiation. According to Jastromb, Nikon’s customers utilize the HCA platform for a broad range of experimental designs, from single drug discovery assays to multiwell plate analysis of fixed and live cells. He noted that any HCA system is only as a good as its image analysis software. Nikon’s NIS-Elements system controls hardware and includes many traditional analysis and measurement tools.
More recently, Nikon developed its HCA software bundle to provide a new user experience for customers migrating to high-content imaging. “Analysis routines in certain freeware packages are mature,” Jastromb elaborated, “so we wanted to make certain that our images could be used in those platforms as well.”
Open-Source Image Analysis
Image analysis software is so central to HCA that some researchers have made it their life’s work. Anne Carpenter, Ph.D., director of the Imaging Platform at the Broad Institute, presented the results of collaborative efforts in using HCA to study the mechanisms and potential treatments of disease.
Her method analyzes patterns in cell imaging that show distinct morphological alterations as a result of treatment with a chemical or genetic agent. “Our goal is to classify drug mechanisms of efficacy and toxicity, distinguish cancer-relevant proteins, and identify biomarkers of disease,” said Dr. Carpenter.
Dr. Carpenter’s research resulted in the development of CellProfiler, which she described as “free, open-source software designed to enable biologists without training in computer vision or programming to quantitatively measure phenotypes from thousands of images automatically.” A key feature of CellProfiler is its modularity—researchers can mix and match image analysis components to suit individual needs.
In addition to maintaining the CellProfiler project, Dr. Carpenter’s group focuses on data-mining technology. High-resolution microscopy generates vast quantities of rich data, but devising methods to take advantage of these data can be a challenge. In HCA, as in so many things, Dr. Carpenter noted, “a key challenge is also a major opportunity.”
Her group has been adapting machine-learning algorithms to suit the demanding needs of HCA. Some applications include identifying morphological biomarkers of disease, classifying the mechanisms underlying drug efficacy and toxicity, creating performance-diverse chemical libraries, and characterizing drug targets. “Ultimately, we hope to make perturbations in cell morphology as computable as other large-scale functional genomics data,” stated Dr. Carpenter.
Daniel LaBarbera, Ph.D., assistant professor at the University of Colorado, offered a new twist on a standard model in cancer research—the multicellular tumor spheroid (MCTS) model. This model, first described in 1970, has, according to Dr. LaBarbera, “proven to be superior in recapitulating the in vitro growth and phenotypes of in vivo human tumors,” permitting the study of cells in three-dimensional (3D) culture, and allowing greater complexity in modeling the behavior of tumor cells than can be accomplished in conventional monolayer tissue culture models.
Although MCTS has overcome many challenges to establish its role in high-throughput screening, some hurdles remain. “Most HCA imaging system software is optimized for 2D cell-based analysis,” said Dr. LaBarbera. However, he is optimistic that more instruments and software will be developed to analyze 3D model systems. Computational technology, he added, will also evolve to process and store the volume of data such systems generate.
Dr. LaBarbera’s group devised the 3D MCTS models to address these issues, beginning with well-characterized biomarkers engineered as surrogate reporters of complex disease phenotypes. This approach allows for HCA in live MCTS to distinguish local (in single cells), region-specific (individual Z-planes), and global (whole MCTS) effects of small molecules.
“Our goal is to identify potent small molecule agents and their respective molecular targets that could induce the reversion of epithelial-mesenchymal transition (EMT),” he explained. EMT is a well-characterized process implicated in many prominent human diseases including cancer progression and metastasis.
Dr. LaBarbera suggested that the true potential of the MCTS model has not been achieved. “In the future,” he concluded, “the MCTS model will transform patient care, allowing for individualized therapy.”
Alan Waggoner, Ph.D., director of the Molecular Biosensor and Imaging Center at Carnegie Mellon University, with colleagues Jonathan Jarvik, Ph.D., and Marcel Bruchez, Ph.D., developed a novel approach to studying G protein-coupled receptor (GPCR) signaling in live cells. Dr. Waggoner’s laboratory uses a type of biosensor that consists of two-part “fluoromodules.”
The first part of the biosensor is a fluorogen-activating protein (FAP). The second is a small-molecule fluorogen whose fluorescent intensity increases dramatically when bound by the FAP. These components allow the biosensor to detect receptor downregulation by monitoring changes in signal location and intensity.
“Genetically introduced sensors should provide a bright signal but not be so concentrated that they disturb what is to be measured,” said Dr. Waggoner, citing some of the challenges he faced in developing this technology. He also mentioned the difficulty of introducing biosensors into cells. Additionally, he noted that sensors have to work with a wide variety of cellular processes, including “protein-protein interactions, kinase/phosphatase activities, conformational changes, and gene switching.”
Given the key role that GPCRs play in drug discovery efforts, Dr. Waggoner’s biosensor research provides opportunities to discover drug-mediated receptor responses that other approaches can miss. His laboratory’s current efforts are directed toward a new assay that concomitantly measures receptor loss from the surface and receptor accumulation inside the cell.
Novel HCA Instrument
Thermo Scientific is showcasing a new instrument in the high-content analysis (HCA) area—the CellInsight™ CX7 HCA platform. Innovating off of last year’s introduction of the CellInsight™ CX5 high-content system, the new CX7 delivers two autofocus modes (software and laser) and built in multi-mode confocal, wide-field, and transmitted light capability. The platform also expands the range of compatible fluorochromes by extending excitation in the CFP/YFP ranges and in the Near IR range, according to Thermo officials.
“Leveraging the HCS Studio™ software, the CX7 can deliver flexibility to the scientist to address the increasing number of new cell models used in high-content analysis,” says Scott Keefer, product manager of the company’s high-content product portfolio. “The CX7 not only extends the capabilities of quantitative imaging to cell biologists, but complements the broad tool set that Thermo Scientific offers across a spectrum of cell-based assays.”