You are on page 1of 8

Anal Bioanal Chem (2010) 398:219–226

DOI 10.1007/s00216-010-3788-3

REVIEW

The beautiful cell: high-content screening in drug discovery
Marc Bickle

Received: 28 February 2010 / Revised: 21 April 2010 / Accepted: 24 April 2010 / Published online: 25 June 2010
# Springer-Verlag 2010

Abstract The term “high-content screening” has become
synonymous with imaging screens using automated microscopes and automated image analysis. The term was coined
a little over 10 years ago. Since then the technology has
evolved considerably and has established itself firmly in the
drug discovery and development industry. Both the instruments and the software controlling the instruments and
analyzing the data have come to maturity, so the full
benefits of high-content screening can now be realized.
Those benefits are the capability of carrying out phenotypic
multiparametric cellular assays in an unbiased, fully
automated, and quantitative fashion. Automated microscopes and automated image analysis are being applied at
all stages of the drug discovery and development pipeline.
All major pharmaceutical companies have adopted the
technology and it is in the process of being embraced
broadly by the academic community. This review aims at
describing the current capabilities and limits of the
technology as well as highlighting necessary developments
that are required to exploit fully the potential of highcontent screening and analysis.
Keywords High-content screening . Imaging .
Drug discovery . Multiparametric . Cell-based assay

Introduction
In the early seventeenth century, Galileo Galilei improved
the design of the first telescope. Using his improved optics,
M. Bickle (*)
Technology Development Studio,
Max Planck Institute of Molecular Cell Biology and Genetics,
Pfotenhauerstrasse 108,
01307 Dresden, Germany
e-mail: bickle@mpi-cbg.de

he supported the heliocentric cosmology of Copernicus by
documenting the orbits of four of Jupiter’s moons and
observing all the phases of Venus. Thus, an optical
instrument allowed the precise description of a phenomenon changing fundamentally the view of our universe. In
the same century, Antonie van Leeuwenhoek and Robert
Hooke used another optical instrument to describe freeliving single cells and cells in multicellular organisms,
respectively. These discoveries spelled the beginning of
modern biology as we know it. As optical devices became
more and more powerful (Hubble telescope, electron
microscope), they enabled countless discoveries to be made
and have thus been instrumental in the advancement of
science in many disciplines. In a similar fashion, imaging
might revolutionize the drug screening industry in the
coming years and improve the efficacy of the drug
discovery pipeline, especially in respect to attrition rates
in clinical trials. The time point could not be better chosen
for an industry struggling with a loss of productivity
concomitant with rising costs of developing new drugs
[1–3]. The main reason for this potential to improve drug
discovery is the descriptive power of images and the
possibility to document spatiotemporal cellular features
quantitatively with computer-assisted image analysis.
Nonautomated and nonquantitative phenotypic screens
were performed much earlier than the advent of highcontent screening (HCS). Two very famous examples that
have come to shape our understanding of biological
processes were the Drosophila melanogaster embryo
screen performed by Nüsslein-Volhard and Wieschaus [4]
and the Caenorhabditis elegans screen performed by
Brenner [5] and colleagues. These phenotypic screens laid
the foundations for years of research and discoveries in
molecular developmental biology. The descriptions of the
mutants obtained have helped elucidate many biological
principles. Those screens exemplify the analytical power of

compounds are screened that influence molecular events in cells within organisms.cenix-bioscience. HCS has the potential to reduce the attrition rate of drugs in clinical trials. thereby leading to the silencing of that particular gene. By careful documentation of the physiological effects of RNAi. assays are being developed using biologically more relevant assays such as mixed cell culture. These include morphological assays such as translocation assays. on the one hand. and statistical analysis of multiparametric data is beyond the scope of this review and the reader is referred to other recent publications [6–13]. 21]. phenotypic screens have the potential to identify in a targeted fashion chemical compounds that modulate molecular pathways in a cellular context. Only a few drugs against novel targets appear on the market every year and there is great hope that RNA interference (RNAi) combined with HCS will increase the number of potential targets. by being more predictive of in vivo outcomes than biochemical assays. One of the major drawbacks of RNAi technology is the problem of off-target effects [29. Bickle the development of drugs with alternative modes of action to those of the existing drugs on the market. disease models. Drug development and HCS Since its inception 10 years ago. and differentiated stem cells [9. neurite outgrowth or tube formation. infection assays. biomarkers. with a single technology. by chemical modifications of the RNAi oligos [31–33]. it is essential to discover and validate novel targets to allow M. It was later shown that the off-target effect is due to micro-RNA-like effects. HCS has penetrated all stages of the drug discovery pipeline. by screening in silico for better sequence stretches to target and. [30] reported in 2003 that siRNA induced silencing of many different genes besides the intended target. Thus. The concentration of each oligo is thereby diminished below the level for the off-target effect. but also assays where only a subpopulation of cells are targeted [14–16]. In vivo applications are still new and suffer from siRNA stability problems. RNAi is a natural biological process in which small interfering RNAs (siRNAs) are loaded into the RNAinduced silencing complex and degrade complementary messenger RNAs. The readouts and the statistical analysis of the data also need to be improved to improve the correlation between in vitro assays and clinical outcome. 30]. Many efforts have been undertaken since then to improve the quality of synthetic RNAi oligos. 23]. primary cells. The community is currently trying to establish standards. Lastly. RNAi and HCS are therefore very popular methods for target discovery and validation and Cenix Bioscience (http://www. This review aims at describing how HCS can and is applied in the drug discovery process and gives an overview of current practice in the field. Through the development of quantitative image analysis and the application of robust statistics. the phenotypic descriptors have become unbiased and comparable across several laboratories within global pharmaceutical companies.220 careful phenotypic descriptions of mutants. HCS and target discovery and validation For companies carrying out target-based primary screens. This area is under intensive investigation as a tremendous advantage to both target validation and RNAi therapeutics is expected to be achieved by improvements in in vivo technology. In a similar fashion. Jackson et al. and rapid elimination in urine [24]. By analyzing cellular physiological characteristics after RNAi-mediated gene silencing. whereby homology of the first eight nucleotides with the 3′ untranslated region of other transcripts led to either cleavage of the messenger RNAs or translational silencing. Discussion of technical aspects such as reagents. delivery to the target tissue problems. Furthermore. Another alternative is the use of RNAi pools synthesized by enzymatic digestion of in vitro generated transcripts (endoribonuclease-prepared siRNA) [34]. To fully realize this potential.com/) has become a highly successful contract research organization providing siRNA high-content screens as a service for target validation to the pharmaceutical industry. and cellular differentiation [17–19]. but the cumulative . For this purpose. a vast range of biological processes can be tested. to make the data resulting from HCS fully available to many laboratories. A common denominator is that HCS is a powerful analytical technology yielding biologically relevant. one can assess inhibition of potential targets in various pathways [22. RNAi can also be applied in vivo to assess potential targets in mouse disease models. novel targets can be discovered and validated and side effects modeled [25]. statistically robust data that are amenable to high throughput. allowing insights to be gained into molecular events within cells in organisms. The reasons are manyfold and specific to each stage of the process. Typical examples are cell cycle assays. 20. high-content analysis is particularly well suited owing to the high granularity of multiparametric analysis of cellular phenotypes. The underlying assumption in applying RNAi to target validation is that silencing of a gene mimics its inhibition by a small chemical compound. Therefore. in drug discovery. paying off the significant investment required to take up the technology. Various large-scale RNAi screening strategies have been devised in the past 10 years and several human genome-wide libraries are available from different vendors [26–28]. on the other hand. HCS allows novel types of assay to be screened.

every protein participating in the cellular process under study is potentially a target. Only a few examples of primary high-content screens have been published and most of these screens are of modest size. 51]. and tertiary assays can be used with fewer and more promising compounds. A drug having adverse effects on a subset of the population (e. For a long time HCS was not applied to primary screens. Subtle effects of drugs on a subset of the cellular population can be assessed. stability. because the throughput was deemed too low and the technology too expensive. and precipitation. compounds are improved in regard to their efficacy. lead series can be tested simultaneously for on-target (efficiency) and off-target effects (specificity). In spite of all these advances and the reduction in off-target effects. First. toxicity. This reduces costs and cycle time. The system biology cellular overview provided by HCS allows the development of compounds to be guided in a more directed manner. HCS often yields more sensitive readouts. reducing attrition rates at the hit-verification stage [47]. Such effects remain completely hidden in a homogeneous assay and will only become apparent at much later stages of the drug development pipeline. as the throughput of traditional techniques. The possibility of entirely omitting target validation is a great advantage. and suffers from a lack of a clear definition as to what criteria need to be fulfilled for a target to be considered validated. 36–42]. hit selection is very efficient. Early ADMET data were traditionally only available later. By multiplexing HCS assays. because phenotypes are scored with several parameters. one can survey a large proportion of the network of pathways potentially affected by the compounds studied. it is only in recent years that HCS has been applied for primary screens. It is now becoming apparent that. specificity. The complexity and redundancy of signaling pathways pose significant challenges for the development of effective and specific drugs. when quantitative cellular phenotypes are exploited. Therefore. Having access to this kind of information is very important for decisions during structure–activity relationship studies of lead candidates to ensure the compound remains on the target and off-target effects can be avoided. as multiple readouts are performed in a single experiment. During lead optimization. imaging is very efficient and could save both cost and time. Third. in a single run. it is still necessary to use several independent siRNAs to validate the observed phenotype [35]. The multiparametric subcellular resolution provided by HCS allows detailed mechanistic studies of the mode of action of lead candidates. one can discover valuable new targets for drug development. Target validation is costly..The beautiful cell: high-content screening in drug discovery effect of all the oligos on the target gene is sufficient for silencing. Cellular pathways have multiple readouts and cross talk to other pathways. compounds are assayed directly for the biological outcome under study and no a priori information about targets is necessary. When carrying out RNAi experiments and validating the results carefully. Owing to these properties. but were not published. distribution. with a single instrument.g. it is possible to devise screens that assay concomitantly several modes of action or toxicity. Other large primary screens have been carried out in the pharmaceutical industry (personal communications). Second. reducing cost and timelines. forming networks of signaling pathways. This is likely due to the fact that the time to develop drugs is lengthy and the technology is still young. This is the case for HCS. allowing better informed decisions in lead prioritization. Furthermore. The hit rates are therefore generally higher in cellbased assays than in biochemical assays. metabolism. as homogeneous assays. A . Assays have been developed that score toxicity in a much more refined fashion than just looking at cell death [50. compounds identified in cellular assays are known to penetrate cells and therefore fulfill the minimal requirements of absorption. Thus. owing to the multiplexing and multiparametric characteristics of HCS. time-consuming. undergoing cell division) can be identified using a cell-by-cell analysis. with some notable exceptions [14. is much lower [49]. secondary assays can be incorporated into the primary screen. because of patenting and commercial reasons. A very important condition for a technology to be useful for lead optimization is that it yields statistically robust data. Fourth. no drug that was discovered in a primary high-content screen has yet received market approval. HCS in primary screening To the best of my knowledge. The capacity to score toxic compounds in high throughput immediately after a primary homogeneous screen is a great advantage. as a cell-by-cell analysis is possible [48]. such as liposome suspensions. the rate of false positives tends to be lower. when screening. Cell-by-cell analysis yielding subpopulation analysis can be very important for lead optimization [52]. cell entry. resulting in secondary screening activity at the primary screen stage [43–46]. and physicochemical properties in iterative rounds of synthesis and biological testing. 221 HCS in lead optimization HCS is very well suited for lead optimization and was initially developed for this application. and IC 50 curves can be generated for many parameters. excretion. and toxicity (ADMET). The ability to combine several readouts in a single assay reduces the cost and time to reach decisions during these cycles. allowing classification of compounds in a lead series according to several criteria.

supervised methods need human input to create classes in a teaching data set and the algorithms select the parameters that best describe the classes. different adsorption and elimination properties and kinetics. All the methods cluster parameters with various statistical methods allowing the determination of the mode of action [43. 57]. Unfortunately. The differences arise mainly from the quality of the objectives and their numerical aperture and the quality of the light path. To determine the pathway affected by a compound and thus its mode of action. advanced statistical tools are used. A review of statistical methods in HCS would go beyond the scope of this review. by moving a Nipkow disk into the light path. 47. Statistics in HCS is a growing field. but the methods tend to be not very powerful. The BDPathway is therefore a very flexible system. as more images can be acquired by an automated microscope than by a human being. Algorithms functioning in this manner are. The software design will determine the flexibility and the user-friendliness of the system. As the names suggest. Service quality is another factor that is often considered when purchasing a new microscope. First. as it can switch between wide-field and confocal mode. Given the central role of the liver in adverse drug reaction. all microscopy experiments should be carried out with automated microscopes. support vector machines. For in vitro toxicology to be predictive of clinical toxicity. automated microscopes have been constantly improved and all systems currently on the market are very reliable and robust. for instance. more complex. to improve the scientific value of microscopy experiments. the most common in vitro toxicology system developed is hepatoxicity using either primary hepatocytes or hepatic cell lines. The clustering analysis is generally performed on a data set obtained with known inhibitors of various pathways. the risk of introducing bias through the experimenter choosing which pictures to take is eliminated. owing to different metabolism.e. Cells should be grown using sandwich cultures to obtain full polarization and the gas mixture should be controlled to have low oxygen partial pressure to reflect the natural environment of hepatocytes. the experimental data are statistically more robust. and refined methods being developed. 46. Bickle primary cells or differentiated stem cells should ensure that the cells studied are relevant to hepatic toxicity. nearest neighbors.. hierarchical clustering. they tend to miss hits or incorrectly assign clusters. Second. Furthermore. culture conditions need to reflect the natural environment of the hepatocytes. Second. These algorithms are considered less biased than supervised ones and allow the data to reveal the full spectrum of cellular responses. There is little difference in optical performance between the various widefield systems. with new. . the microscopes being equipped with incubation chambers and onboard injection. and lack of genetic variation. neural networks. The analysis aims at finding parameters that predict toxicity before actual cell death occurs.222 fingerprint that represents the desired cellular phenotype can be derived from the multiplexed multiparametric readouts. Another potential approach. automated microscopes provide statistically robust data that are unbiased by human intervention. Arguably. This type of analysis will certainly grow in the future and allow very precise measurements of the action of compounds on various tissues. and decision trees. and k nearest neighbors. The multiplexing capacity of microscopic techniques allows several pathways involved in toxicological processes to be assayed [50. These methods are well suited when finding defined phenotypes is the goal. There are currently over seven vendors of HCS microscopes offering both wide-field systems and confocal systems. Often personal preferences in software design determines the preference for a platform. Typical unsupervised algorithms are self-organizing maps. Unsupervised methods require no or only very minimal input from human operators. Since the early days of HCS. there are concerns about the translatability of animal results to humans. Using M. i. is quantitative multiparametric analysis of tissue sections [58]. 53–55]. the cells under study should reflect the physiological characteristics of organs. Many systems offer livecell imaging options. HCS in toxicology HCS has also been deployed in toxicology in the hope of minimizing the number of animal experiments and replacing them with in vitro cellular studies. The main difference between the platforms is the software controlling the microscopes and the image analysis software. the analysis of cellular systems needs to fulfill certain criteria. By sampling images randomly within the wells of microtiter plates. microscopic cellular analysis became feasible at the very large scale of screening. Supervised and unsupervised methods exist for classifying multiparametric screening results. A multiparametric image analysis is then performed on the data set and the data are analyzed by various methods. time is more efficiently utilized if the scientist does not spend hours in front of microscopes acquiring images. but they have been reviewed elsewhere [56]. The BDPathway is unique of its kind. First. Third. Current instrumentation and software With the development of fully automated microscopes and fully automated image analysis.

cells are under oxidative stress with the partial oxygen pressure of the normal atmosphere. although their price being higher. and lastly the laser point scanning TCS SP5 from Leica. . Much development has been invested in the software both controlling the instruments and also analyzing the images. will also have the flexibility of optional spinning disk confocality like the BDPathway. This is the first HCS instrument to have EMCCD cameras. although developments are currently under way. the market will be smaller. like in the case of the IN Cell Analyzer 2000. The whole process is programmed in advance with scheduling software. The instruments have become more user-friendly and faster. All new instruments will profit from advances in CCD camera technology. The OPERETTA. with larger chips. The instrument has three electronmultiplying CCD (EMCCD) cameras for parallel acquisition of up to three channels. The culture conditions also need to be improved in regarding the gas mixture cells receive. With increasing experience. users have started wanting to be more in control of the image analysis and are less reliant on turnkey solutions provided by HCS microscopes. More relevant biology HCS has matured considerably in the first 10 years since its inception. A new interesting confocal instrument is the CellVoyager from Yokogawa. It is to be expected that new confocal instruments will also appear on the market. sensitive acquisition with onboard injection for live-cell drug discovery screens. In current systems. Future instrumentation In 2009. With the IN Cell Analyzer 2000. allowing better characterization of compounds in living cellular systems. For live-cell imaging. Currently. importantly. Such instruments will likely have a large impact on drug discovery. it is important to work with objectives with the highest numerical aperture and the highest transmittance. Cells grown in two dimensions are not polarized and do not express the same genes as in their physiological environment [59]. imposing tighter constraints on the optical quality of the microscope. With the OPERETTA. the laser point scanning ImageXpressULTRA from Molecular Devices. Screening cancer cell lines in 2D cell cultures on a hard substrate does not reflect the cellular physiological processes in an organism sufficiently well.The beautiful cell: high-content screening in drug discovery live-cell imaging does not depend solely on the ability to incubate cells while imaging. only four dedicated confocal systems are on the market. faster cameras and. In consequence. and incubator and one for data handling. The laser line scanning IN Cell Analyzer 3000 from GE Healthcare. This instrument was designed with a special eye for livecell imaging in drug discovery. the plate is then moved to the imaging platform for acquisition. thereby providing wide-field as well as confocal imaging. Image-based autofocus is simply not an option when performing live-cell experiments. but lacks the capacity to be integrated with peripheral robotics. The CellVoyager is therefore a very sophisticated instrument providing fast. The software of the instruments has also evolved tremendously during the past 10 years. For HCS to fulfill its promise of reducing the attrition rate of the drug development process. Furthermore. The light path can be diverted around the spinning disk. one for controlling the microscope. for instance. a laser-based autofocus system. image analysis is made much easier by the possibility to evaluate several algorithms by mousing over their tab and looking in real time how the masks shift. two new wide-field systems were introduced on the market and it is to be expected that more new instruments with the latest upgrades in microscopy will appear in the coming years. and target activation. screenable assays covering many aspects of cell biology have been developed. a sensitive camera. All future instruments can be expected to be equipped with a powerful multicore workstation to provide the necessary computing power and speed to acquire and analyze images in a rapid fashion. it is possible to quickly scan a well at low magnification. many image analysis solutions provide tools for flexible segmentation and parameter extraction next to turnkey solutions. which feeds plates directly into an acclimatized injection platform capable of injecting in different plate formats. and. with more sensitive. The CellVoyager has two work stations. which is currently only designed for wide-field imaging. The TCS SP5 has been adapted from a conventional microscope to a HCS microscope. nuclear translocation. 223 This feature allows the best performing algorithm to be chosen and its parameters to be adjusted very easily. more relevant biological systems need to be screened. Instruments have become more powerful and more versatile. Cells suffer considerably from light. the Yokogawa spinning disk OPERA from PerkinElmer. and image analysis solutions for the quantification of the data are functioning well. before acquiring high-resolution images. by introducing 4D and 5D microscopy. There are image analysis solutions for many of the common drug discovery assays such as neurite outgrowth. lenses and filters have also improved and the loss of light should be reduced in the new instruments. The microscope is equipped with an integrated incubator. injection platform. The instrument has a Yokogawa spinning disk with four solidstate lasers and a 100-W halogen light for bright-field illumination.

8. Conclusions HCS has been widely accepted by the pharmaceutical industry as a valuable tool to study biological responses to drug candidates at various stages of the drug development pipeline. as academic institutions are less reliant on vendors. and culture conditions [60. by improving drug candidates emerging from the discovery pipeline. Carnero A (2009) Clin Transl Oncol 11:651–658 Evensen L. References 1. 3. Gubler H. At the moment. Adams CP. on the other hand. through either multithreading or faster clock frequencies of chips. resulting in multimodal distributions [70]. 11. Beibel M. the full description of cellular populations will be exploited. A more refined analysis. bringing to bear their biological know-how in their assays. which are rich in leukocytes and platelets. 12. Researchers can obtain the buffy coats.224 Incubation chambers need to be devised with a controlled atmosphere and gas mixture. Another possibility to increase computing power is cloud computing. analyzing every cell is too expensive computationally and averages of populations are analyzed. It is nevertheless important that the cellular pathway targeted is pathologically relevant and that the cells used are physiologically relevant. Parker CN (2010) J Biomol Screen 15:95–101 Thomas N (2010) J Biomol Screen 15:1–9 . thus. In stem cell work. the use of primary cells or stem cells is promising. 6. It is expected they will cut costs. more and more cluster farms will be built. on the other hand. where the interplay of all the genes in the genome in a given process is analyzed in a quantitative fashion. which allow for a certain amount of expansion before terminal differentiation [62]. Computing power One restricting factor in HCS is the computing power available to handle large data sets and analyze large quantities of images. owing to its ability to analyze subpopulations. it is easy to distinguish feeder cells from stem cells morphologically or by introduction of cell-type-specific fluorescent markers. HCS has also started entering the academic world in recent years. 5. to genetic background variations and. by using more sophisticated algorithms. Wieschaus E (1980) Nature 287:795–801 Brenner S (1974) Genetics 77:71–94 Conrad C. HCS is well suited for mixed cell cultures for organotypic systems. Sharma P. As the number of transistors in computers continues to double every 2 years (Moore’s law). 10. more objects can be more precisely segmented. to differences in isolation. and. HCS is an ideal tool for systems biology. The aims in academic institutions are twofold: on the one hand. due. resulting in a better description of cellular behavior. This is rarely true and discrete subpopulations are more the norm. to develop completely novel solutions. Automated image analysis is a compromise between speed and precision. 69]. 61]. The most common use of human primary cells is blood cells as they are the most easily accessible [63–67]. Both have challenges that need to be addressed. academic institutions are also performing drug discovery campaigns. 2. One valuable source of primary cells is progenitor cells. from blood donor centers to study various aspects of the immune system or for toxicology studies. who have their commercial constraints. the processing power will increase. it is likely that with diminishing costs. 9. Gabriel D. There are many commercial suppliers of primary cells or biobanks and it is also possible to set up collaborations with hospitals to obtain primary human cells. higher precision will be affordable. which will probably become more common in the future [68. Gehin P. Bickle underlying assumption when averaging a cell population is that the parameters are normally distributed. Gerlich DW (2010) J Cell Biol 188:453–461 Niederlein A. Furthermore. 7. With increasing computer power. 4. on the one hand. would be more accurate. Lorens JB (2010) Cytom A 77A:41–51 Daub A. Finkbeiner S (2009) Curr Opin Neurobiol 19:537–543 Kummel A. A great concern with primary cells is their variability. The M. White D. The academic world is sure to contribute some exciting new developments in the near future. In conclusion. Brantner VV (2010) Health Econ 19:130–141 Pearson H (2006) Nature 444:532–533 Hughes B (2008) Nat Rev Drug Discov 8:93–96 Nusslein-Volhard C. Another corollary of increased computing power is the possibility to carry out cell-by-cell analysis of entire populations. even using quantiles of the population. In this respect. Link W. Future improvements in the cellular systems screened and computing power will increase further the value of HCS for drug discovery. Micklem DR. With increased computing power. it will be interesting to analyze in future years the impact that imaging screens have had on the drug discovery pipeline. the full benefits of HCS are rarely exploited. Marc B (2009) Comb Chem High Throughput Screen 12:899–907 Zanella F. Meyenhofer F. With increased computing speed. purification. The accurate mimicking of pathological situations is a field that needs to be developed to get away from the simple cell lines and simple cell culture conditions currently used. with more and more institutes purchasing an automated microscope and the corresponding peripheral robotics.

Mao M. Eckhardt F. Muller E. Volkov Y. Barabasz A. Schelter J. Schelter J. Weiler J. Socci ND. Wipf P. Putz G. Andree C. Wong STC (2009) J Biomed Inform 42:32–40 55. Bakal C. Jackson AL. Cuddihy M. Loo L-H. Abraham VC. Jackson AL. Liron Y. Altschuler SJ (2008) Proc Natl Acad Sci USA 105:19306–19311 53. Srivastava P. Chanda SK. Silber PM. Castoreno AB. Habermann B. Gerlich DW. Janjua K. Peng LF. Dorkin R. Krajewska M. Jan E. Kabbe R. Lee G. Csucs G (2009) J Comput Sci Syst Biol 2:219–239 57. Placantonakis DG. Ellenberg J. Lazo JS (2003) Chem Biol 10:733 42. Geiger B. Waring JF. Hyman AA. Besson D (2007) J Biomol Screen 12:1042–1049 48. Mikhailov SN. Buchholz F (2007) Nat Meth 4:337–344 225 35. Satagopam V. Ovcharenko D. Kim EH. Kam Z (2007) J Struct Biol 158:233 39. Labow M. Nielsen SJ. Crofton KM. Angeline A. Bus C. Huber W. Zhou C. Garippa RJ (2008) HCS for HTS. Durbin R. Bulkescher J. Kumar C. Kotov NA (2008) ACS Nano 2:928–938 58. Pedersen HC. Moiana A. Johnson JM. Robinette BL. Buisson B. Nielsen AF. Brisson M. Schneider R. Li AP (2002) Chem Biol Interact 142:57–71 61. Warrior U. Heninger A-K. Marshall W. Ovcharenko D. Brain P. Almholt DLC. Pau G. Que QQ. Rogers P. Martin RC. Khvorova A. Lloyd S. Mousses S (2009) Meth Mol Biol 563:275–287 26. Gabriel D (2009) J Biomol Screen 14:59–65 49. Pelkmans L. Zerial M (2005) Nature 436:78–86 29. Brosch R. Vaidya S. Ying X. Linke SP. Bartz SR. Grabner H. Alexandrov Y. Ibig-Rehm Y. Karpilow J. Garavaglia A. Smith LH. Mawji I. Riccardi G. Bernardi P. Fava E. Schelter J. Wild H. Ewann F. Mason AJ. Gun'ko YK. Meyer TF. Burchard J. Akentieva NP. Neumann B. Kreamer A. Abramov M. Erfle H. Engels JW. Echeverri CJ. Huang X. Koch C. Sauerwein R. Guo J. Marsollier L. Otto JC. Powers EA. Sakai Y. Dahl SW (2004) Assay Drug Dev Technol 2:7 38. Wu LF. Kelnar K. Walter T. Türbachova I. Feng Y (2008) Nat Chem Biol 4:59–68 46. McWhinnie E. Freudenrich TM. Dennis J. Natt F. Loechel F. Wengel J. Vogt A. Bernards R (2006) Nat Meth 3:777 36. Tao CY. Silver SJ. Yin Z. Reed JC. Root DE. Nichols K. Yin HH. Theis M. Rees E. Schmidt S. Schmeck C. Yan Y. Krog-Jensen C. Mickanin C. Kuzmin A.The beautiful cell: high-content screening in drug discovery 13. Grabner H. Laursen MB. Bender A. Ferrer M. Mull R. Wu LF. Pepperkok R. Kiefer J. Alves K. Covey DF. Kozak K. Davies AM. Osafune K. Howard-Cofield E. Hare T. Santini F (2008) J Biomol Screen 13:449–455 41. Bobkov GV. Chen S. Frenzel C. Machuy N. Kang S. Krajewski S (2009) J Histochem Cytochem 57:649–663 59. Cooley KA. Linsley PS. Studer L (2008) Cell Stem Cell 2:602–612 20. Maux D. Zaltsman AB. Lamprecht MR. Cetin C. Strulovici B. Ciro A. Zeps N. Hériché J-K. Baron B (2003) J Biomol Screen 8:489 37. Lee SY. Linsley PS (2006) RNA 12:1197–1205 32. Schmitz MHA. Papait R. Burchard J. Smicius R. Eils R. Boutros M. Echeverri CJ. ContrerasDominguez M. Kempf M. Moffat J. Evers AS. Marshak DR. Chau BN. Borowiak M. Tallarico JA. Brown D (2007) Cancer Res 67:10782–10788 24. Lam K. Wu LF. Ross DA. Berlin K. Garyantes T. Fenistein D. Reiser V. Li B. Caldarelli A. Maehr R. Slack MD. Fox JL. Unterreiner V. Kjems J (2009) Nucleic Acids Res 37:2867–2881 33. Agrawal A. Cavet G. Haskins JR. Lim L. Butcher S. Genovesio A. Krausz E. Jones TR. Engel S. Slack MD. Hillisch A. Altschuler SJ (2004) Science 306:1194 44. Olson AH. Wiley. Lee S. Hoboken 43. Heide M. Tarpley MG. Iacopetta B. Mundy WR (2010) Neurotoxicol Teratol 32:25–35 17. Zerial M. Shin WK. Delaage M. Lacapere J-J. Gao B. Carpenter AE. Kaludercic N. Peters J-M. Mitchison TJ. Wünsche A. Walter T. Rice J (2006) Assay Drug Dev Technol 4:153 45. Cattaneo E (2010) BMC Neurosci 11:7 14. Buchholz F. Leake D. Simonen M. O’Brien P. Kauschke SG. Rosenberg A. Nam JY. Towne DL. Conrad C. Li F. Kittler R. Brodin P (2009) PLoS Pathog 5:e1000645 19. Röhl I. Wolff M. Rodrigues CD. Hougham C (2006) Arch Toxicol 80:580–604 51. Held M. Perrimon N. Wrana J. Hewitt NJ. Lobell M. Hall J (2005) Nat Biotechnol 23:995–1001 34. Jarvis R. Root DE. Klingmueller U. Ilan M. Møller S. Heilker R (2008) J Biomol Screen 13:837–846 21. Sabatini DM (2009) Proc Natl Acad Sci USA 106:1826–1831 54. Bohl S. Sieckmann F. Olek S (2006) Epigenetics 1:55–60 62. Ma Y. Seo MJ. Burchard J. Downward J. Joly-Guillou M-L. Djaballah H. Christophe T. Chattopadhyaya J. Wong S (2008) BMC Bioinform 9:264 56. Bradley PL. Carret C. Detmers PA. Carralot J-P. Feng Y. Cole ST. Duval F. Martin C. Hojland T. Melton D (2009) Nat Chem Biol 5:258–265 22. Hunicke-Smith S. Chapuis C. Burns DJ (2008) J Biomol Screen 13:527–537 52. Greenwalt DE (2006) In: Taylor DL. Burchard J. Lim L. Hendrix M. Luty AJF. Nehrbass U. Oh T. Michaud M. Davidow L. Fackett A. Lingham R. Hahn WC. Huesken D. Santos AF. Terry R. Kim J. Goldstein S. Hacohen N. Herdewijn P. Woodle MC. Gubler H. Rong J. Mathey-Prevot B. Rogers P. Zhou X. Ellenberg J. Lum L. Leng N. Desbordes SC. Pan C. Paran Y. Praestegaard M. Haas J. Odadzic D. Xu G. Hellwag A. Hess S. Erfle H. Perrimon N. Lange J. Camnasio S. Jackson M. Roquemore EP. Jenkins JL. Bordet T. Baron U. Krejsa C. Datti A. Perlman ZE. Ellenberg J (2010) Nature 464:721–727 18. Byrne SJ. Habermann B. Durr O. Skovierová H. Martinez ED. Wagner J. Galea P. Linsley PS (2006) RNA 12:1179–1187 30. Hyer ML. Bolognini D. Ostuni MA. Schreiber SL. Chirn G-W. Hansen TB. Macrae C. Lu PY (2006) Drug Discov Today 11:67–73 25. Wenska M. Chang S-F. Honcharenko D. Verheijen JH (2008) ASSAY Drug Devel Technol 6:693–710 16. No Z. Babu BR. Neumann B. Xue J. Sachse C. Pepperkok R (2007) Nat Protoc 2:392 28. Fava E. Perrimon N. Pham H. Franke K. Eggert US. Reynolds A. Erck MGMv. Slabicki M. Gonçalves LA. Jackson AL. Korn B. Hayden MJ. Diaz D. Baum B. Hannus M. Rubin LL. Li F. Henderson CE. Mitchison TJ. Young DW. Mota MM (2008) PLoS Pathog 4:e1000201 23. Sabatini DM. Beachy PA. Jackson AL. Xie FY. Jeon HK. Prudencio M. Radio NM. Steidl E-M. Hudak E. Cleary M. Liebel U. Kashman Y. Hoffmuller U. Irwin W. Wang J. Meier H. Bjørn SP. Kobayashi SV. Bramsen JB. Slaughter M. Langkjaer N. Delaage P. Cohen D. Hoffman AF. Asselbergs F. Fraser AG. Kozak K. Brodte A. Pruss RM (2007) J Pharmacol Exp Ther 322:709–720 40. Barman J. Hinzen B (2006) Drug Discov Today 11:175–180 50. Pagliaro L. Real E. Rigamonti D. Linsley PS. Nichols A. Sun Y. Van Aerschot A. Held M. Brown D (2005) RNA 11:985–993 27. Warner J. Schimmer A (2008) Apoptosis 13:413–422 15. Scott A. Hannus M. Jackson AL. Sonnichsen B. Heyse S. Park EJ. Reinhardt M. Wunberg T. Drouot C. Giuliano KA (eds) High content screening: a powerful approach . Mann M (2009) Mol Cell Proteomics 8:443–450 60. Foley B. Hoyt J. Grenier JK. Kiger A. Taipale J. Johnson C. Altschuler SJ (2007) Nat Meth 4:445–453 47. Golland P. Poser I. Massaad C. Jahn-Hoffmann K. Lang P. Schumacher M. Li Z. Schelter J. Kelnar K. Gardina P. Guo J. Jessop RA. Surendranath V. Liebel U. Linsley PS (2003) Nat Biotechnol 21:635–637 31. Meloon B. Zhou X. Labow M.

64. doi:10. Margot G. Garcia Loaiza B.226 63. Gerhard L. Gomez Gonzalez S (2010) J Digit Imaging (in press). Methods in molecular biology. Pelaez Echavarria A. Sriraman P. Fieles W. Snijder B. Imrich A. Pognan F. Ramo P. Venkatachalam V. Reynolds P (2010) J Biomed Inform 43:342–353 69. Sulahian TH. Totowa. vol 356. Abbot S. Montoya Munera E. M. Gerd S (2006) Cytom A 69A:200–202 Morelli J. Zhao L. Kobzik L (2009) PLoS ONE 4:e6209 Tobias W. Sacher R. Koester D. Bickle to systems cell biology and drug discovery. Ciaccio P (2006) Cell Biol Toxicol 22:15–27 Basiji DA. Ruiz Ibañez C. Rosenthal A. Pelaez Restrepo G. 66. Chiriac C. Barone L. Valencia Diaz E. Buehrle M. Morrissey P (2007) Clin Lab Med 27:653 67. Garcia Ruiz M. Beletskii MCA. Ortyn WE. Garcia Chaves A. Yu L (2005) Biotechniques 39:894–897 68. 65. Gutierrez Mazo J. Pelkmans L (2009) Nature 461:520–523 . Li MH.1007/s10278-009-9265-x 70. Liang L. Humana. Ramirez Giraldo J. Stanford J. Mork P. pp 121–128 DeLoid GM. Damm E-M. Liberali P.