Featured organism: Danio rerio, the zebrafish

The zebrafish has long been a favourite model for the study of vertebrate development. Here we provide an overview of the current state of knowledge and resources for the study of this fish, with comments on the future direction of zebrafish genomics from Professor Mark Fishman and Dr Stephen Wilson. Copyright © 2000 John Wiley & Sons, Ltd.


Background
Zebra®sh (Danio rerio) are small, y3 cm long. They are native to streams in India and are commonly kept as pets. The males are slender and torpedoshaped, with black longitudinal stripes and usually a gold colouration on the belly and ®ns. Females are fat when laden with eggs and have little, if any, gold on their undersides.
Zebra®sh are easy to raise, with a short generation time of 3 months, and the females can lay hundreds of eggs at weekly intervals. Fertilization is external, allowing easy access to embryos for observation and manipulation. Developing embryos are easily studied under a dissecting microscope since they are transparent. Zebra®sh embryos develop rapidly (in 2±4 days), with a beating heart and visible erythrocytes by 24 h. Another advantage of studying the zebra®sh is that, in contrast to other ®sh, which can be triploid or tetraploid (which makes genetic analysis dif®cult), it maintains the diploid state.
The zebra®sh has been shown to be a useful model for the development of several complex tissues such as the kidney (Drummond, 2000), the olfactory system (Ardouin et al., 2000) and the visual system (Saszik et al., 2000). In the study of haematopoiesis, zebra®sh mutants have been observed to demonstrate strikingly similar phenotypes to those in human diseases (Amatruda and Zon, 1999).

Tools for study
Despite the lack of a technology for gene knockout by homologous recombination, zebra®sh researchers have amassed huge numbers of mutant lines. The bulk of these were created in a large-scale mutagenesis project using the chemical mutagen, ethylnitrosourea (ENU), followed by screening for developmental defects (Haffter et al., 1996;Driever et al., 1996). ENU has the advantage of generating small mutations (compared with those caused by irradiation), which often affect only one gene. These studies revealed mutants defective in the development of many organ systems, such as the retina (Malicki et al., 1996), the inner ear (Whit®eld et al., 1996) and the cardiovascular system (Stainier et al., 1996). After exhaustive study of the mutants produced from these experiments, it became clear that saturation had not been reached and the initiation of a second large-scale mutagenesis screen for zebra®sh, the`Tu È bingen 2000' screen, was announced recently. The screen will be undertaken by Artemis Pharmaceuticals in collaboration with labs at the Max-Planck-Institutes of Developmental Biology and Immune Biology, the University of Heidelberg, the Howard Hughes Medical Institute in Boston and University College London (see this issue, pp. 232±234, for an interview with Dr Stefan Schulte-Merker of Artemis Pharmaceuticals on the mutagenesis screen).
A zebra®sh radiation hybrid (RH) panel (commonly known as the`Goodfellow' panel; Kwok et al., 1998) was produced. This has been used as the basis for a map designed to aid the positional cloning of the genes mutated in the strains from the mutagenesis screens (Geisler et al., 1999).
The NIH Institute-wide Zebra®sh Genome Initiative, overseen by the Trans-NIH Zebra®sh Coordinating Committee, was started in September 1998. A YAC library has been created, which consists of 17 000 clones, and provides y®ve-fold coverage of the zebra®sh genome (Zhong et al., 1998). There is also a P1 library, which consists of 4±5 genome equivalents, with clones having an average insert size of 115 kb (Amemiya and Zon, 1999). The initiative is also funding the creation of a 0.5±1.0 cM microsatellite map (Shimoda et al., 1999), mapping of the`Goodfellow' and`Ekker' (Hukriede et al., 1999) radiation hybrid panels, which are to be populated with y10 000 expressed sequence tags (ESTs), and a meiotic map (Kelly et al., 2000), which is to include markers in common with the RH maps. To supply those creating the maps with markers, there is also funding for the generation of y100 000 ESTs from oligo-®ngerprinted cDNA libraries representing several zebra®sh embryo developmental stages and different tissues. Another project in the Genome Initiative is to create deletion mutants of the entire genome. This uses an approach in which c-irradiated sperm are used to fertilize wild-type eggs (Fritz et al., 1996). DNA from gynogenetic haploid progeny of females produced from this fertilization is used as the template in multiplex PCR with primers for any chosen locus. Females carrying the required deletion are identi®ed due to failure to amplify that locus in a fraction of their progeny. The mutants produced will be placed in the Zebra®sh Stock Center at the University of Oregon, Eugene (see Web-based Resources), which is also supported by the NIH initiative.
Tools available for functional analysis in zebra-®sh include a selection of ectopic expression vectors (Hyatt and Ekker, 1999), with the option of having the gene of interest under the control of tissue speci®c or inducible promoters (Lekven et al., 2000). One particular example, which allows targeted gene expression in the zebra®sh, is called Gal4-UAS (Scheer and Campos-Ortega. 1999). This system involves two vectors, which are used to create separate transgenic zebra®sh lines. The ®rst carries the yeast Gal4 transcriptional activator, under the control of a given promoter; the second carries the gene of interest fused to the Gal4 DNAbinding motif. The zebra®sh are then crossed to achieve spatial transcription of the chosen gene. Being transparent, zebra®sh are ideal for expression pattern monitoring using lacZ or GFP fusion constructs. Alternative overexpression systems can use RNA or DNA injection (Krauss et al., 1993;Linney et al., 1999) and even large clones such as BACs have been used to rescue mutant phenotypes . This type of approach is crucial in identifying the gene in a chosen mutant that is responsible for the observed phenotype.
A recent addition to the battery of techniques for knocking out gene function in zebra®sh is RNAi. This strategy has been used in Caenorhabditis elegans, Drosophila and mouse and has now been shown to work in zebra®sh (Li et al., 2000;Wargelius et al., 2000). If this can be shown to work reliably, it could open an exciting new avenue for zebra®sh genomics, in which a null phenotype can be achieved for a chosen gene of interest. Another similar approach is to block gene expression using morpholino antisense oligonucleotides. These modi®ed oligonucleotides form more stable triple-helices with DNA than normal oligonucleotides (Lacroix et al., 2000). As yet, though, no indepth study of this method in zebra®sh has been published.

Current Status of Genome Knowledge
The zebra®sh genome is y1700 Mb in size, as 25 chromosomes. The basis for all of the maps of the genome consists of 25 linkage groups on which the 25 centromeres have been placed (Johnson et al., 1996). There are radiation hybrid (RH) and meiotic maps of the genome onto which CA repeat markers, mutations and ESTs have been placed. There are currently 3000 microsatellites and over 1000 genes and ESTs on the meiotic map (Kelly et al., 2000), and 3000 ESTs have been placed on the`Goodfellow' RH panel, 1000 of which have also been mapped on the`Ekker' RH panel (Zon Lab web page).
Comparison of the zebra®sh gene map to a selection of mammalian genomes  yielded signi®cant evidence of synteny. A comparison of the zebra®sh HOX clusters to those of other vertebrates (Amores et al., 1998) provided evidence for a genome duplication event, which is thought to have occurred after the divergence of ray-®nned and lobe-®nned ®shes but before the teleost radiation. Further comparison of the seven zebra®sh HOX complexes with the four in Fugu con®rmed gene loss as a dominant feature of the evolution of tetrapod and teleost HOX clusters (Aparicio, 2000).

Future aims
Professor Mark Fishman is the director of the Cardiovascular Research Center at the Massachusetts General Hospital. His group is applying a genetic approach to the study of the earliest steps of development of cardiovascular form and function in zebra®sh.
He feels that the key to the zebra®sh is the genetic screen. He did the ®rst large-scale screens in collaboration with Wolfgang Driever (and in parallel with Professor Christiane Nu È sslein-Volhard's group) and they were remarkably informative. They were astounded by how modular the genetic disruptions were, i.e. that single gene mutations deleted or perturbed single units of form (e.g. removing a chamber or a valve of the heart). This demonstrated the unitary logic of organ assembly. The next step will be to clone these genes, which will provide the necessary handles on molecular pathways for these units, such as thè ventricular pathway'. A second consideration in his work is that organs are functioning devices, and embryonic survival depends upon certain of them, such as the cardiovascular system, kicking in from day one. The zebra®sh is quite viable and vigorous without a functioning cardiovascular system because it survives by diffusion from the water during the ®rst few days. This is in distinction to the mouse, in which secondary degeneration begins immediately upon cessation of cardiac function. Thus, for the ®rst time, it is possible to analyse the genes that drive function in the embryo. So, he can now see a way to answer questions like:`What makes the ®rst heart beat?';`How does contraction drive cellular development?'; and`Does blood¯ow cause or modify vessel formation?'. He also sees similar questions being answered with regard to the onset of function for renal, gut and other organ systems. The phenotypes of many of these mutations provide eerily accurate phenotypic models of complex human diseases, for which we have few candidate genes. In this regard, he feels that the zebra®sh will be the salvation of physiology and the way to search the human genome project for real gold.
Professor Fishman's view is that the main value of mapping and sequencing of the genome is that it will expedite mutation cloning. However, he notes that zebra®sh genome comparisons with other species will provide good starting points for analysis of regulatory elements, and insight into how their divergence might cause large-scale structural changes during evolution.
He anticipates that new screens will be applied to the understanding of organ formation and physiology, evolutionary biology and other areas. Behaviour, learning, and memory are already the focus of some early phase studies and suppressor and enhancer screens will help re®ne pathway descriptions. In addition, the relatively ready penetrance of many chemicals into the viable embryo makes them targets for chemical screens, relevant to drug discovery and toxicology.
Dr Stephen Wilson is a Wellcome Senior Research Fellow and Reader at the Department of Anatomy and Developmental Biology at University College London. His group is studying the mechanisms underlying the patterning of cells and tissues in the embryonic zebra®sh, with a focus on the embryonic development of the forebrain.
He feels that genome sequencing is a major priority for the community for several reasons, but perhaps primarily to facilitate the rapid cloning of genes affected by mutations isolated in a wide variety of genetic screens, using a candidate gene approach.
He sees full-length cDNA sequencing as an essential parallel project to genome sequencing to facilitate the identi®cation of coding sequences within the genome. This will be essential to aid in the cloning of mutations and for functional analysis of the genome, as a wide variety of mis-expression strategies now exist in zebra®sh, provided that one has access to full-length coding sequence. Generation of new libraries, from wild-type and transgenic lines, is likely to be essential for this project. As further cDNA resources become available, such as Unigene sets, then the generation of extremely wellcharacterized arrayed libraries will be possible.
He also feels that zebra®sh are perhaps the most suitable vertebrate model system (owing to their transparency and availability in huge numbers) in which to perform large-scale expression pro®ling analysis of genes by in situ hybridization. Using whole embryos at a variety of developmental stages, it is possible to rapidly, accurately and economically determine sites and times of expression of thousands of genes during embryonic and larval development. This type of analysis has been initiated in several groups, but Stephen's feeling is that it needs to be expanded and funded on a larger scale.
From these large-scale expression pro®ling projects, large numbers of genes with highly speci®c expression patterns will be isolated. In his opinion it will be important to establish selected transgenic GFP lines that label speci®c cell populations in the ®sh by using regulatory sequences that control the expression of these genes. Such lines will be essential for detailed phenotypic analysis (by cross-ing them to mutant lines), for detailed screening procedures and for the isolation of cell type-speci®c RNA populations that may be underrepresented in more general libraries.
Stephen points out that if antisense morpholinos turn out to reliably inhibit gene function, then as soon as full-length cDNA sequences are obtained, large-scale`knock-out' analysis becomes possible. This would certainly not replace true genetic approaches, but may be an initial ®rst indicator of gene function. He also notes that existing screens have isolated only a fraction of the mutant lines that can be obtained and so it is clear that further directed screens are essential. These should include screens by morphology, by gene and protein expression (using wild-type and transgenic-GFP lines), by behaviour, physiology and as many other criteria as investigators can establish.
In addition, Stephen asserts that more knowledge of anatomy, physiology, cell biology and behaviour of developing and adult ®sh, organized and presented on the web via ZFIN (see Web-based Resources), will be required to fully make sense of gene function analysis studies.

General information on zebra®sh
The Zebra®sh Information Network (ZFIN) http://z®sh.uoregon.edu/ This page contains a vast array of information on zebra®sh anatomy, genetic strains, maps, genes, ESTs and probes. There are news pages, a protocol book, a link to the zebra®sh newsgroup, laboratory and researcher lists and information about the stock centre.

Mapping and sequencing projects
The Zebra®sh webserver at the Cardiovascular Research Centre of Massachusetts General Hospital http://zebra®sh.mgh.harvard.edu/ This site provides access to zebra®sh genetic maps, made using microsatellites, as a searchable on-line database and ®gures of the maps of each of the 25 linkage groups. There is also an atlas of zebra®sh anatomy and information on a YAC library created by the group.
The WashU Zebra®sh Genome Resources Project http://z®sh.wustl.edu/ The group are developing ESTs from y50 000 zebra®sh cDNAs, derived from libraries produced from a selection of sources, such as adult brain, kidney and liver. The cDNA clones are available from Incyte (http://www.genomesystems.com) and from the Resource Centre/Primary Database (RZPD) section of the German Human Genome Project (http://www.rzpd.de). The project also involves mapping of the ESTs onto a radiation hybrid (RH) map. This page provides maps and a service in which the group will use your sequence or your primers to place a marker of your choice onto the map. They also provide the protocols they are using so that users can map a marker onto the panel themselves.
The Dawid Laboratory ± Zebra®sh Radiation Hybrid Mapping http://dir.nichd.nih.gov/lmg/lmgdevb.htm This group provides access to their radiation hybrid panel so that users can place a marker of interest onto the map. There are images of the maps of all 25 linkage groups and mapping data from a cDNA expression screen.
The Trans-NIH Zebra®sh Initiative http://www.nih.gov/science/models/zebra®sh/ This page contains information on funding opportunities, reports of previous meetings relating to the initiative, future meetings and courses and a list of zebra®sh links.

Gene indices
The TIGR Gene Index for Zebra®sh http://www.tigr.org/tdb/zgi/ This is a database of zebra®sh expressed sequences classi®ed into three categories: ESTs (partial, singlepass sequences from either end of cDNA clones); CT's (tentative consensuses, built from overlapping EST sequences); and ETs (mature transcripts, with untranslated regions where possible). The database can be searched with a gene name or identi®er, by tissue or library identi®er, or by using a BLAST search with a sequence of interest.
The NCBI Unigene Zebra®sh Sequences Collection http://www.ncbi.nlm.nih.gov/UniGene/Dr.Home.html A non-redundant set of zebra®sh genes generated from known genes and automated clustering of the EST data. These can be searched by keyword or cDNA library type.

Zebra®sh stock centres
The Zebra®sh International Resource Center http://z®sh.uoregon.edu/zf_info/stckctr/stckctr.html The center has a collection of wild-type, mutant and transgene-carrying ®sh and frozen sperm. The center provides information and advice on keeping zebra®sh and researchers can visit the center to learn techniques or screen the collection for mutants of interest.

Comparative and functional genomics projects
The Stanford Zebra®sh Genome Project http://zebra®sh.stanford.edu/genome/Frontpage.html This page has genetic maps of zebra®sh and the results of a comparison made with the human genome. There are also pages detailing the group's efforts towards the generation of new markers for use in the maps.
The Zebra®sh Deletion Project http://www.ciwemb.edu/labs/halpern/zdp.html Although these pages are still under construction, there is a tool available to look for deletions available in each linkage group and contact details for members of the group, to arrange to obtain mutants from their stocks.
The UCL Zebra®sh Research Group http://www.ucl.ac.uk/zebra®sh-group/ The group is involved in screening mutants from both of the Tu È bingen zebra®sh mutagenesis projects. They are also undertaking a smaller-scale mutagenesis screen of their own. There is also a small stock centre, with just over 100 lines, which are made available to the community.