Cytomics for discovering drugs

Abstract

There are two major fields of applications for the cytomics approach: Predictive medicine for preventative medicine and drug discovery. Drug discovery demands that thousands of compounds are rapidly tested for their biological activity in order to detect those rare events that may bear the potential of becoming useful pharmaceutics somewhere in the future (1, 2). This requires the combination of appropriate cell systems for testing, robust high-throughput technological platforms allowing for high-content screening (mostly imaging systems) as well as fast and stable automated image and data analysis systems enabling unbiased data scrutiny. Ideally, such cytomic oriented detection work flow should help to reduce the frequency of false positive hits (requiring substantial effort and costs for further testing) and false negative hits (leading to loss of potentially precious compounds). Evensen et al. (3) established a unique co-culture assay of endothelial cells and vascular smooth muscle cells for screening of anti-angiogenic drugs for cancer therapy. Their system enables investigation of drug efficacy on immature, VEGF- dependent, and growing endothelium as well as on mature, nonproliferating endothelium. The authors tested in a pilot study several anti-angiogenic drugs using a microtiter-plate imaging system. They found that tube total length is the most informative parameter derived from the images and they discovered by their quantitative assay some unexpected compound activity. Automated high-throughput imaging combined with quantitative analysis of the actin cytoskeleton opens the opportunity for rapid standardized determination of the effects of treatment or infection on cell structure. An image analysis tool for automated measurement of cytoskeleton modifications in a high-throughput imaging system is demonstrated by Weichsel et al. (4). They introduce the parameter image coherency and show that it is suitable to detect accurately global alterations in the cytoskeleton organization. De Vos et al. (5) focus their interest on the fully automated multivariate phenotypic classification of individual cell nuclei and subnuclear spots by automated classification and supervised machine learning. These authors recently developed controlled light exposure microscopy, a novel technology that strongly reduces photodamage by limiting excitation in parts of the image where full exposure is not needed (6). In their present publication, they applied fluorescence mosaic microscopy for image acquisition and ImageJ for image analysis of DAPI and histone gamma-H2AX (7) labeled cells. The authors developed a comprehensive work flow for high- content screening of the nuclear architecture of cell lines treated with a genotoxic agent without or in combination with ionizing or UV irradiation. This approach is suitable for high-throughput screening in drug discovery. Image quality is crucial in image-based screening and quantitation. Low quality images may render unreliable results and need to be further processed or discarded. Now Zeder et al. (8) developed an automated, artificial neural network based, quality assessment of autonomously acquired microscopic images for fluorescently stained bacteria. Their software proves to have a slightly higher success rate than human observers in classifying image qualities correctly. More importantly, it relieves the experimenter from monotonous work. In general, high-throughput high-content imaging systems acquire 2D images. However, intracellular structures are clearly located in a 3D environment that is a prerequisite for exerting their spatio-temporal function. Allalou et al. (9) developed a robust signal detection algorithm for 3D fluorescence microscopy consisting of a detector and a verifier to detect point-like signals in 3D images. The authors state that their approach is superior over hitherto used image analysis methods for 3D recognition. Finally, time-lapse microscopy of cell proliferation and gene expression is of importance in detecting long-term effects of drugs on cellular systems. However, the analysis of such serial images is tedious. Now Wang et al. (10) developed robust automated methods for image segmentation and dynamic lineage analysis for the single-cell fluorescence microscope. These tools are portable and applicable for diverse detection modalities and for various species. Standardization is essential in unifying data, for performing interlaboratory comparisons and for using different instruments to perform analysis on the same platform. These demands lead in the past to the development of the Flow Cytometry Data File Standard (FCS) with the latest upgrade being FCS 3.0. Now, The international society for the advancement of cytometry data standard task force (ISAC DSTF) presents the most recent FCS upgrade: FCS 3.1 (11) with several improvements and simplifications. In conclusion, this first issue of the year shows the broad range of developments in cytometry providing to improve drug discovery and to get a better hold on the heterogeneity of responses that unknown drugs may evoke in respective cellular test systems (12). I thank Dr. Jozsef Bocsi, Heart Center Leipzig, for his help with this manuscript.