Journal of the Association for Laboratory Automation http://jla.sagepub.com/ Image-Based Fluidic Sorting System for Automated Zebrafish Egg Sorting into Multiwell Plates Siegfried F. Graf, Sebastian Hötzel, Urban Liebel, Andreas Stemmer and Helmut F. Knapp Journal of Laboratory Automation 2011 16: 105 DOI: 10.1016/j.jala.2010.11.002 The online version of this article can be found at: http://jla.sagepub.com/content/16/2/105 Published by: http://www.sagepublications.com On behalf of: Society for Laboratory Automation and Screening Additional services and information for Journal of the Association for Laboratory Automation can be found at: Email Alerts: http://jla.sagepub.com/cgi/alerts Subscriptions: http://jla.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav >> Version of Record - Apr 1, 2011 What is This? Downloaded from jla.sagepub.com by guest on October 11, 2013 Original Report Image-Based Fluidic Sorting System for Automated Zebrafish Egg Sorting into Multiwell Plates otzel,2 Urban Liebel,2 Andreas Stemmer,3 Siegfried F. Graf,1 Sebastian H€ and Helmut F. Knapp1* 1  Microfluidics & Liquid Handling, Centre Suisse d’Electronique et de Microtechnique, Alpnach, Switzerland 2 Institute of Toxicology and Genetics, Karlsruhe Institute of Technology, Karlsruhe, Germany 3 Nanotechnology Group, Department of Mechanical and Process Engineering, ETH Zurich, Zurich, Switzerland Keywords: toxicity testing, cell sorting, automated, multiwell plate, Zebrafish, danio rerio T he global demand for the reduction of animal testing has led to the emergence of Zebrafish eggs/larvae as model organisms to replace current adult animal testing in, for example, toxicity testing. Because of the egg size (diameter 1.6 mm) and the relatively easy maintenance of Zebrafish farms the eggs also offer high-throughput screening (HTS). However, the current bottleneck for HTS is the cost-efficient placing of individual organisms into single wells of a multiwell plate (MWP). The system presented here is capable of storing, sorting, and placing individual organisms in a highly reproducible manner. In about 11 min a complete 96-MWP is filled, which corresponds to about 8 sec per egg. The survival rate of fertilized transgenic and wild-type eggs was comparable to the one of the control (control 6.7%, system 7.6%). Furthermore, it was also possible to place dechorionated eggs into individual wells. The results demonstrate that the cost efficient system works gentle and reliable enough to disburden scientists from the exhausting and monotonous job of placing single eggs into single wells, such that they can concentrate on the scientific aspects of *Correspondence: Helmut F. Knapp, Ph.D., Microfluidics & Liquid  Handling, Centre Suisse d’Electronique et de Microtechnique, CH-6055, Alpnach, Switzerland; Phone: þ41.416.727.523; Fax: þ41.416.727.500; E-mail: helmut.knapp@csem.ch 2211-0682/$36.00 Copyright Screening c 2011 by the Society for Laboratory Automation and doi:10.1016/j.jala.2010.11.002 their experiments and create results with a higher statistical relevance. ( JALA 2011;16:105–11) INTRODUCTION Manipulation of single cells or single organisms is a fast growing market not least because the REACH initiative was started in 2007. The goal of the REACH initiative is to test each compound (from which more than 1 ton is manufactured or imported in Europe) for its toxicity.1 At the same time, the number of vertebrate mammals used for toxicity testing should be reduced. The two previous facts call for high throughput and ethically justifiable toxicity screening methods. In recent years, fertilized Zebrafish eggs established themselves as a model system and are accepted as an alternative to mammalian testing.2e4 Another large advantage is that fertilized Zebrafish eggs up to 5 days after fertilization are not subject to the same regulatory requirements as adult mammals. Further advantages are the egg’s transparency and the fact that development occurs entirely outside the mother’s body. Finally, Zebrafish facilities are relatively easy to maintain and because of the size of an egg (diameter !1.6 mma, see Fig. 1), dispensing them in up to 384-well plates is possible.5 a http://www.neuro.uoregon.edu/k12/FAQs.html. JALA April 2011 105 Original Report Figure 1. A. Fertilized (except top right) Zebrafish eggs after 7 h, (B) different batch with fertilized Zebrafish eggs after 11 h. Two white eggs are dead. Scale bar indicates 10 mm for both images. In current toxicity tests, fertilized Zebrafish eggs are dispensed into multiwell plates (MWPs). Test compounds with different dilutions are pipetted to the individual fertilized Zebrafish egg in the single wells. The MWP is then periodically imaged and analyzed.6 The large amount of data is locally or remotely stored and analyzed. Finally, fast search engines allow one to retrieve the collected data. In summary, a complete line-up of systems and techniques from different disciplines is required to perform a low cost high throughput toxicity test as shown below: 1. Inexpensive and efficient Zebrafish facilitydto get a large amount of eggs; 2. Inexpensive sorting systemdto place individual eggs into a MWP; 3. Inexpensive lab robotic systemdto handle the plates and deliver test compounds into wells; 4. Powerful automated imaging systemdto periodically observe the eggs quality; 5. Efficient data storage/handling; 6. Efficient image analysis systemdto analyze images; and 7. Efficient search algorithmsdto retrieve results. 106 JALA April 2011 Of the list above, many of the items listed are available. More specifically, inexpensive, efficient Zebrafish facilities can be established and maintained by following certain guidelines.7 Lab robots and powerful automated imaging system are commercially available although they are still expensive. Efficient data storage and data distribution is possible but also an expensive task. Systems for image and visualization analysis8 are available, their performance however corresponds to the available computer power. There are also several search engines available such as Harvester,9 which combine results from several other search engines and databases. The main remaining bottleneck is the inexpensive sorting system. Currently, most of the Zebrafish eggs are placed into well plates manually. Two skilled technicians can fill 30 96-well plates in about 3 h, which corresponds to 12 min per plate on the average. Unfortunately, this highly monotonous task is prone to errors and can typically not be performed by a single person for more than 3 h. Another example comes from Stern et al.10 who screened about 16,000 compounds in 16 weeks, while 5000 larvae were collected weekly. They were quality controlled and then 20 larvae were manually transferred with a chemical spatula in each well of a 48-well plate. Several attempts have been made to automate the singularization of Zebrafish eggs or larvae into MWPs. A possibility is to use a robotic pick and place approach for sorting eggs from a petri dish into the well plate. One such system is the CellBot (Fully Automated Single Cell Handling Platformb) from the Centre  Suisse d’Electronique et de Microtechnique, where a delta robot, capable of several cycles per second, is combined with a fully motorized inverted light microscope (iMic from Till Photonics, Germany). The robot and the microscope move above and beneath the sample platform, which is stationary. Because the petri dish and the well plate are never moved during the sorting process, this arrangement prevents sloshing of the sample liquid. The inverted light microscope scans the petri dish and checks for eggs, if one is found, this egg is removed by a pipette tool attached to the delta robot and placed into the well plate. Unfortunately, such a pick and place system is costly, rather slow, and uses a large portion of lab space. Further options are the COPAS and BioSorter (both from Union Biometrica, Holliston, MA), which are specially designed for sorting of large biological entities (up to 1500 mm). These systems are capable of singularizing Zebrafish eggs into 96-well plates in about 2 min.11 However, these systems were designed for high throughput flow cytometry and are therefore equipped with costly optics, not necessary for simply singularizing large entities into MWPs. Additionally, both systems are not capable of checking the quality of the entities sorted, such as shape, specific features, or damage. The system presented here, named ZebraFactor, is an inexpensive sorting device for large biological entities, such as Zebrafish eggs and larvae, Xenopus oocytes, pollen, and cell clusters. Sorting is done using a vision system taking b http://www.csem.ch/docs/Show.aspx?id¼7824. Original Report complete images of the entities, combined with fast vision algorithms capable of identifying a set of characteristics of an entity in real time as they are moved past the vision system. Identified single entities can then be extracted from the sorting system and transferred to a container, such as an individual well in a well plate, or to a subsequent device. Here, we will demonstrate the direct sorting and individualization of Zebrafish eggs into the wells of a 96-well plate. Moreover, the system also has been used for sorting and feeding Xenopus oocytes into an automated microinjection module, a development by us not presented here (XenoFactor: Fully Automated Microinjection Systemc). In the following, we present the sorting system in combination with our WellPlateFeeder (see Fig. 2). SYSTEM Hardware The ZebraFactor as shown in Figure 2 consists of two main units named CellSorter and WellPlateFeeder. The CellSorter is used to sort single entities such as Zebrafish eggs or larvae from the suspension, whereas the WellPlateFeeder places the single entities into a well of the MWP. The actions of the CellSorter and the WellPlateFeeder are synchronized to ensure correct feeding of entities into the well plate. After introducing the egg suspension into the CellSorter, the suspension is continuously moved by our novel concept of using a sliding and static ring as shown in Figure 3. The two rings define a circular fluidic channel for the egg suspension. The sliding ring is rotated by an electro motor and drags along the suspension buffer because of viscous drag forces. Additionally, a small friction force acts on the chorion of the egg. The sum of drag and friction forces results in rolling of the egg along the fluidic channel and past the cameras positioned to visualize a portion of the channel. This concept works most efficiently with objects larger than 200 mm (Xenopus laevis oocyte are up to 1300 mm, Zebrafish eggs are up to 1600 mm), and allows moving cells continuously without destroying them. Additionally this concept offers: 1. To store eggs until their usedcontinuous motion avoids adhesion; 2. To observe eggs over and over againdin connection with the vision system; 3. To deliver eggs on demand to a subsequent system; and 4. To load the CellSorter with additional eggs on the fly.d Figure 2. System overview of ZebraFactor consisting of the WellPlateFeeder (left) and the CellSorter (right) connected by tubing. Black bar indicates 50 mm. The vision system installed along the fluidic path consists of one or two inexpensive complementary metal oxide semiconductor cameras and a light emitting diode array for illumination. Additionally, excitation and emission filters can be introduced for using fluorescence signals for the identification of fluorescently labeled entities. In case of transparent samples such as Zebrafish egg or larvae, a one-camera setup is sufficient (as in Fig. 2), whereas for opaque samples such as Xenopus laevis oocytes or in case where the user is interested in the sample surface, a two-camera setup would be used (as in Fig. 4). The two cameras would be placed opposite to each other, which enables observing the whole surface of an entity. This configuration allows the system: 1. To inspect the whole surface of any kind of sample (opaque and transparent); 2. To inspect a cell compartment of a transparent sample; and 3. To inspect fluorescence signals. The ZebraFactor presented here is simply equipped with one camera and a light emitting diode-array illumination without using any excitation and emission filters because the processed Zebrafish eggs were not labeled. To remove a single egg from the CellSorter, buffer is redirected inside the channel and, by a gentle push, moves the egg to the WellPlateFeeder, or any other subsequent system such as the microinjection module mentioned above. In any case, light barriers are used to control the process of moving the egg from the CellSorter to the subsequent system. c http://www.csem.ch/docs/Show.aspx?id¼6247. d On the fly means, the system does not need to be stopped to add additional cells. Figure 3. Schematic of fluid and object motion as generated by the sliding ring concept. The egg is driven by drag and friction forces and rolls along the base of the sliding ring. JALA April 2011 107 Original Report The software to control the hardware is designed to store the analyzed image together with the batch and well plate number and well plate location. Further, egg collection time, stock number, genotype, and sorting time can be added to the XML database. These data allows one to keep track of the egg from breeding to the single larva analysis. Graphical User Interface Figure 4. CellSorter from Figure 2 with a two-camera configuration to sort opaque cells. The incorporated vision algorithm is based on a neuronal network, which allows the user to set and teach sorting and singularization criteria. With this kind of approach, the user can simply singularize eggs or larvae by size and shape or even sort by complex criteria, which currently can only be done manually. In summary, the vision algorithm enables: 1. To simply singularize single eggs or larvae from a suspension (e.g., by size and shape); 2. To sort eggs or larvae by complex criteria from a mixed suspension (e.g., fertilized or unfertilized, damaged or healthy.); and 3. To easily teach the vision algorithm a new set of characteristics for sorting. The WellPlateFeeder is designed as a gantry robot, that is, the MWP is stationary and the feeding tool is moved from one well position to the next. The system is designed such that enough space is available to pick the well plate by a robot and place it onto, for example, an automated microscope for later analysis. The feeding tool consists of two pinch valves driven by electro motors. This allows one to keep the valves open or closed without any energy consumption and heat dissipation. To decide when to open or close the valves, a light barrier is placed between the valves. If the signal of the light barrier is interrupted, the first valve passed by the egg closes and the second valve opens. A dispensing pump is then used to dispense the amount of buffer needed to transfer the egg into the 96-well plate. The current system is optimized for 96-well plates. However, the same technique can be used to dispense single eggs into a 384-well plate or several eggs into 1 well of a 96-well plate. The complete system is designed such that with little effort it can be automatically primed, which takes less than 3 min. To clean the system at the end, the buffer is removed automatically within less than 2 min. Finally, the user has to clean the sorting part to avoid calcinations and salt crystals, which takes about 8 min. It is also possible to flush the system with pure or diluted ethanol for sterilization. Priming and cleaning only needs to be performed at certain intervals, for example, before and after filling many well plates in a day’s work. 108 JALA April 2011 The graphical user interface (GUI) is kept as simple as possible. In the standard view, the GUI offers a ‘‘Fill/Empty system’’ button, ‘‘Start/Stop,’’ and ‘‘Save image.’’ Information about the batch number and the well plate number is shown in the top left corner as illustrated in Figure 5. Additionally, a virtual MWP shows which wells are already filled with samples. In the idle mode the upper of the two images on the GUI shows the live image, whereas the lower one remains black. In the running mode, the upper image of the two images shows the analyzed egg. If a suitable egg is found and the system is ready for removal, a second image is taken just before removal and is shown in the lower image of the GUI (see Fig. 6). Additional tabs such as ‘‘TroubleShoot’’ and ‘‘Extra’’ are inserted into the GUI to assist if the system malfunctions or if manual steps are required (see Fig. 6). VALIDATION The system described above was validated with fertilized Zebrafish eggs. In the following, the methods and the results are presented. Figure 5. Graphical User Interface of the ZebraFactor: on the top left the batch number and corresponding well plate number are indicated. The overall number of buttons is kept low for simplicity: ‘‘Start’’ to start the process; ‘‘Light’’ to observe; what is going on in the sorter; ‘‘Save Image’’ to capture an image; ‘‘New MWP’’ to get a new MWP; ‘‘Next Pos’’ to move the WellPlateFeeder to the next MWP position; ‘‘Fill System’’ to fill the system; ‘‘Load Settings’’ if the settings file was changed. Two additional tabs ‘‘TroubleShoot’’ and ‘‘Extra’’ are available for more complex tasks. Original Report Figure 6. A. Screen shot taken after three eggs were placed. The first image shows the egg when it was detected and analyzed, the lower image shows the egg just before the removal. Below the ‘‘Save Image’’ button, information about the egg diameter in pixels, the detection score, and the processing time in milliseconds are given. On the bottom left, the past time is shown. B. The tab ‘‘TroubleShoot’’ contains solutions to problems: ‘‘PrimeP1’’ is used to remove bubbles from the pump P1; ‘‘BubbleRem’’ is used to remove bubbles; and ‘‘Unclogging’’ is used to unclog the system. C. The tab ‘‘Extra’’ contains buttons to control the individual devices such as pumps and valves. Method Adult wild-type and transgenic Zebrafish were raised at the Karlsruhe Institute of Technology following standard fish care and maintenance protocols.12 Zebrafish eggs from different batches were collected directly after fertilization. One-half of the eggs were kept in a petri dish for the control. The other half were used in the ZebraFactor to automatically fill 96-well plates. We followed the procedure described by Lammer et al.,6 except that eggs were dispensed at approximately 4 h instead of 3 h postfertilization (hpf). The time for each run was measured and the errors noted. Finally, the survival rate was counted approximately 18 h after fertilization. Additionally, for some fertilized eggs the chorion was removed with tweezers before the sorting. Petri dishes (Semadeni, Part No 5647) and MWPs (Semadeni, Part No 6233) were clean but not sterile. Buffer used for all the processes was embryo buffer E3þ.13 To handle the eggs, a disposable plastic pipette was used (Semadeni, Art No 2292). MWP filling took place at room temperature (about 23  C); however, the incubation was done at 28.8  C. Results To verify the system, five runs with transgenic (see Table 1) and two runs with wild-type (see Table 2) Zebrafish eggs were made. In Figure 7 dispensed single Zebrafish eggs in individual wells of a MWP are shown. In an additional experiment, after modifying the search parameters, sorting of dechorionated embryos was possible. Table 1. Results for transgenic Zebrafish eggs automatically placed by the ZebraFactor into 96-well plates, compared with control with similar amount of eggs kept in a petri dish Line transgenic Control Plate Plate Plate Plate Plate 1 2 3 4 5 Total Average Amount of embryos (entity) 449 98 96 96 97 97 484 96.8 Dead after 18 hpf (entity) (%) 30 6.7 10 9 5 8 5 37 7.4 10.2 9.4 5.2 8.2 5.2 7.6 7.6 hpf at placing (hpf) ca. 4 ca. ca. ca. ca. ca. 4 4 4 4 4 ca. 4 Errors (entity) Time for placing (hh:mm:ss) 5 0 0 1 5 00:12:35 00:12:23 00:17:24 00:11:06 00:12:41 11 2.2 01:06:09 00:13:14 JALA April 2011 109 Original Report Table 2. Results for wild-type Zebrafish eggs automatically placed by the ZebraFactor into 96-well plates, compared with control with similar amount of eggs kept in a petri dish Line wild type Control Amount of embryos (entity) 201 Dead after 18 hpf (entity) (%) 24 11.9 hpf at placing (hpf) ca. 4 Errors (entity) Time for placing (hh:mm:ss) Plate 1 Plate 2 111 99 11 14 9.9 14.1 ca. 4 ca. 4 10 10 00:15:55 00:12:54 Total Average 210 105 25 13 11.9 12.0 ca. 4 20 10 00:28:49 00:14:24 Twenty-four embryos with total three errors (none or two embryos in one well) were dispensed into wells. The placing into the well plate was gentle enough to avoid crushing the yolk bag as shown in Figure 8. All embryos survived (survival count after 100 hpf). DISCUSSION AND CONCLUSION The need for toxicity testing of compounds is increasing because the REACH initiative in Europe was launched in 2007. Numerous publications show that fertilized Zebrafish eggs and larvae are good model organisms for toxicity testing and can substantially reduce the amount of vertebrate mammals used for such tests. However, for efficient large scale toxicity testing, automated solutions are required. Firstly, lab automation systems to handle liquids and well plates are necessary. Secondly, software and hardware solutions to efficiently store, distribute, and analyze the large amount of data collected from the screens are needed. One remaining bottleneck in the whole screening process remains by getting the eggs or larvae from the breeding tank into a processable format such as the standardized 96- or 384-well plate. Figure 7. Single Zebrafish eggs dispensed into individual wells of a multiwell plate with the ZebraFactor. 110 JALA April 2011 Our sorting device for singularizing eggs from a suspension into a 96-well plate, named ZebraFactor, consisting of the CellSorter and WellPlateFeeder, fills a 96-well plate at an average of 13 min, which corresponds to about 8 sec per egg. The surviving rate is approximately the same as in the control group (wild type: ZebraFactor: 7.6% vs Control: 6.7%; transgenic: ZebraFactor: 12.0% vs Control: 11.9%). Furthermore, only by changing some parameters in the software, dechorionated embryos could be dispensed with a 100% survival rate. Until today, lab personal was only able to manually fill 96well plates 3 hs a day before exhaustion lowered their throughput and quality. Now, it is possible to use the ZebraFactor to perform this monotonous work with the same constant quality 24 hrs a day, 7 days a week. Moreover, the personnel can prepare and run the experiments while the ZebraFactor is filling further well plates. Thus, the automated sorting system presented here is expected to enable toxicity screening with several factors higher throughput than today. Several factors limit the minimal time of a cycle: (1) transfer distance and transfer speed between the CellSorter and the WellPlateFeederdthe shorter the distance or the faster the transfer speed, the shorter the cycle time; (2) pressure wave resulting from pinch valves in the interfacedthe faster the valve switches the higher the pressure wave traveling Figure 8. Dechorionated embryo after placement into a multiwell plate using the ZebraFactor. Original Report through the system gets, which can led to malfunctioning of the CellSorter; and (3) speed of the WellPlateFeeder to move from one well to the nextdthe faster the WellPlateFeeder the shorter the cycle time. As mentioned above, an average automated filling of a 96-well plate with the ZebraFactor takes 13 min, which results to a cycle time per egg or well of 8 sec. We strongly believe that this time can be reduced to 4 sec, by making adaptions to the design of the fluidics and the control software. This would enable filling of a 96-well plate in about 6 min. We assume that the cycle time to fill into a 384-well plate will be even shorter because of smaller displacement distances. In near future, our system presented will not only be capable to sort into 96-well plates, but also into 384-well plates or to dispense 1e4 larvae into a single well of a 96-well plate. Because of the small size of the CellSorter, it could also be directly combined with an inverted microscope, such that the embryos are directly placed into a well plate mounted on the microscope. Finally, the CellSorter can also be combined with an automated microinjection system and then with a WellPlateFeeder. This would allow one to perform knockdown analysis and much more. A video of the performance of the ZebraFactor can be viewed at http://www.youtube.com/watch?v¼1ZtMQ-dI52E. ACKNOWLEDGMENTS Kros, A.; Meijer, A. H.; Metz, J. R.; van der Sar, A. M.; Schaaf, M. J.; Schulte-Merker, S.; Spaink, H. P.; Tak, P. P.; Verbeek, F. J.; Vervoordeldonk, M. J.; Vonk, F. J.; Witte, F.; Yuan, H.; Richardson, M. K. Zebrafish development and regeneration: new tools for biomedical research. Int. J. Dev. Biol. 2009, 53(5e6), 835e850. 3. DeMicco, A.; Cooper, K. R.; Richardson, J. R.; White, L. A. Developmental neurotoxicity of pyrethroid insecticides in Zebrafish embryos. Toxicol. Sci. 2010, 113(1), 177e186. 4. Langheinrich, U. Zebrafish: a new model on the pharmaceutical catwalk. BioEssays 2003, 25(9), 904e912. 5. Eimon, P. M.; Rubinstein, A. L. The use of in vivo Zebrafish assays in drug toxicity screening. Expert. Opin. Drug Metab. Toxicol. 2009, 5(4), 393e401. 6. Lammer, E.; Carr, G. J.; Wendler, K.; Rawlings, J. M.; Belanger, S. E.; Braunbeck, T. Is the fish embryo toxicity test (FET) with the Zebrafish (Danio rerio) a potential alternative for the fish acute toxicity test?. Comp. Biochem. Physiol. C, Toxicol. Pharmacol. 2009, 149(2), 196e209. 7. Bilotta, J.; Saszik, S.; DeLorenzo, A. S.; Hardesty, H. R. Establishing and maintaining a low-cost Zebrafish breeding and behavioral research facility. Behav. Res. Methods Instrum. Comput. 1999, 31(1), 178e184. 8. Walter, T.; Shattuck, D.; Baldock, R.; Bastin, M.; Carpenter, A.; Duce, S.; Ellenberg, J.; Fraser, A.; Hamilton, N.; Pieper, S.; Ragan, M.; Schneider, J.; Tomancak, P.; Heriche, J. K. Visualization of image data from cells to organisms. Nat. Methods 2010, 7(3), S26eS41. 9. Liebel, U.; Kindler, B.; Pepperkok, R.; William, E.; Balch, C. J. D. Bioinformatic ‘‘Harvester’’: a search engine for genome-wide human, The authors thank the Neuhauss group at the University Zurich, especially Kara Dannenhauer, for their constant support with fresh Zebrafish eggs. mouse, and rat protein resources. Methods Enzymol. 2005, 404, 19e26 Competing Interests Statement: The authors certify that they have no relevant 10. Stern, H. M.; Murphey, R. D.; Shepard, J. L.; Amatruda, J. F.; Straub, C. Academic Press. financial interests in this manuscript. T.; Pfaff, K. L.; Weber, G.; Tallarico, J. A.; King, R. W.; Zon, L. I. Small REFERENCES molecules that delay S phase suppress a Zebrafish bmyb mutant. Nat. Chem. Biol. 2005, 1(7), 366e370. 1. Lilienblum, W.; Dekant, W.; Foth, H.; Gebel, T.; Hengstler, J.; Kahl, R.; Kramer, P. J.; Schweinfurth, H.; Wollin, K. M. Alternative 11. Pulak, R. Techniques for Analysis, Sorting, and Dispensing of C. methods to safety studies in experimental animals: role in the risk assess- Humana Press. 2006, pp 275e286. 12. Westerfield, M. The zebrafish book. A guide for the laboratory use of zebrafish (Danio rerio). 4th ed. University of Oregon Press: Eugene; 2000. ment of chemicals under the new European Chemicals Legislation (REACH). Arch. Toxicol. 2008, 82(4), 211e236. elegans on the COPASÔ Flow-Sorting System, Strange, K., Ed. 2. Brittijn, S. A.; Duivesteijn, S. J.; Belmamoune, M.; Bertens, L. F.; 13. Trompouki, E.; Zon, L. I. Small Molecule Screen in Zebrafish and HSC Bitter, W.; de Bruijn, J. D.; Champagne, D. L.; Cuppen, E.; Flik, G.; Van- Expansion, in Cellular Programming and Reprogramming. 2010; 301e316. denbroucke-Grauls, C. M.; Janssen, R. A.; de Jong, I. M.; de Kloet, E. R.; JALA April 2011 111