An infrastructure for high-throughput microscopy: instrumentation, informatics, and integration.

High-throughput, image-based cell assays are rapidly emerging as valuable tools for the pharmaceutical industry and academic laboratories for use in both drug discovery and basic cell biology research. Access to commercially available assay reagents and automated microscope systems has made it relatively straightforward for a laboratory to begin running assays and collecting image-based cell assay data, but doing so on a large scale can be more challenging. Challenges include process bottlenecks with sample preparation, image acquisition, and data analysis as well as day-to-day assay consistency, managing unprecedented quantities of image data, and fully extracting useful information from the primary assay data. This chapter considers many of the decisions needed to build a robust infrastructure that addresses these challenges. Infrastructure components described include integrated laboratory automation systems for sample preparation and imaging, as well as an informatics infrastructure for multilevel image and data analysis. Throughout the chapter we describe a variety of strategies that emphasize building processes that are scaleable, highly efficient, and rigorously quality controlled.

Identifying off-target effects and hidden phenotypes of drugs in human cells.

We present a strategy for identifying off-target effects and hidden phenotypes of drugs by directly probing biochemical pathways that underlie therapeutic or toxic mechanisms in intact, living cells. High-content protein-fragment complementation assays (PCAs) were constructed with synthetic fragments of a mutant fluorescent protein ('Venus', EYFP or both), allowing us to measure spatial and temporal changes in protein complexes in response to drugs that activate or inhibit particular pathways. One hundred and seven different drugs from six therapeutic areas were screened against 49 different PCA reporters for ten cellular processes. This strategy reproduced known structure-function relationships and also predicted 'hidden,' potent antiproliferative activities for four drugs with novel mechanisms of action, including disruption of mitochondrial membrane potential. A simple algorithm identified a 25-assay panel that was highly predictive of antiproliferative activity, and the predictive power of this approach was confirmed with cross-validation tests. This study suggests a strategy for therapeutic discovery that identifies novel, unpredicted mechanisms of drug action and thereby enhances the productivity of drug-discovery research.