REVIEWS Why zebrafish? A combination of two traits make zebrafish a favored vertebrate model system. The zebraflsh system allows simultaneous application of both experimental embryology and extensive genetic analysis, while the classical model organisms lend themselves to either sophisticated embryological manipulations (frog and chicken) or advanced genetics (mouse). Genetic experiments with zebrafish will help to identify genes involved in the formation of certain structures in the embryo. Here, we focus on how genetics can be done in zebrafish. The basic features of zebrafish embtyogenesis have been reviewed previously in T/GL2. Zebrafish (Brachy- danio rerlo) are small (3--4 cm long) tropical freshwater fish that originate from India. They reach sexual maturity at about three months. Females lay hundreds of eggs approximately every other week, a feature that greatly facili.tates genetic analysis. Eggs are fertilized externally and embryos are completely transparent, allowing one to follow the development of every indi- vidual cell3 (Fig. 1). Because the embryo is clear, such events as cell movements during gastrulation, the for- marion of the domains of the brain and the beating of the heart can be observed. Deviations from normal development can be analysed easily in the living embryo. Since zebmfish ate small, inexpensive to main- rain and can be bred in large numbers, several extensive genetic screens are currently under way in an effort to saturate the zebrafish genome for mutations that affect embryogenesis. Genetics in zebralbh, generating and identifying mutations 'Yeast-style'genetics: tricks that save time The diploid genome of mice or zebrafish presents a time-consuming problem to geneticists. After individual heterozygous founder animals have been generated, Ft families must be bred for several months and then siblings crossed to drive a recessive mutation to homozygosity and reveal its phenotype, In fish, alterna- tive approaches are feasible that circumvent this prob- lem. Methods adapted for zebraflsh by Streisinger and colleague# make it possible to induce haploid devel- opment or gynogenetic diploid development (Fig. 2), UV-inactivated sperm can be used to fertilize eggs in vitro and induce haploid development. Haploid em- bryos develop the basic structure of the body but have a variety of defects, including short tails, a deformed notochord and edema, and die around the fourth day after fertilization. Haploid embryos permit screening tot mutations that affect the formation of the body plan, but cannot be u:~ed for detailed screens, since even wild- type haploid embryos have abnormalities in many structures and in later developmental programs. Gynogenetic diploid embryos can be produced by subjecting eggs fertilized with UV-inactivated sperm to high hydrostatic pressure just after fertilization: this so-called early pressure (EP) treatment suppresses the second meiotic anaphase. Alternatively,a heat shock (HS) administered 15 min after fertilization suppresses the first mitotic division to induce gynogenesis. Half of the HS-treated embryos and half of the non-recombinant fraction of the EP-treated embryos will express the Zebrafish: genetic tools for studying vertebrate development WOLFGANG DRIEVER~DEREK STEMPLE~ AI.KXANDI~ SCHIER AND tn.l,~'NASOLNICA-IO~F.Z~.t Zebraflsh baee emered the aretm of vertebrate biology as a maiJstt~eammodel systent atul the use ofgeneffc tools in this tropical flsb should eahance our uuderstattding of vertebrate developme~ Thezebraflsb system aUows genetic expertme~s that are aotposstble in other vertebrates, aml the nmtatioNs isolated thusfar attest to its useJhhtess, complementing knowledgeobtained flom other model organisms. mutant phenotype. Unfortunately, some 10-20% of EP-treated embryos and more than 50% of HS-treated embryos may develop abnormally because of physical damage to the eggsS. Hence, the HS technique is not efficient in screening for mutations. On the other hand, the EP technique allows mutations close to the centro- meres to be identified easily (half of the progeny is mutanO. Identification of mutations in genes further from the centromere can be obscured by meiotic recombination and crossover interference (for example, after EP treatment, only 3% of the embryos from a heterozygous golden mutant female are homozygous6). Haploid embryos and EP-treated gynogenetic diploid embryos have been used by a group based in Eugene (OR, USA) to isolate a sigviflcant number of mutations. In addition, the EP technique provides a simple method of mapping genes relative to the eentromereCh the further a gene from the centromere, the smaller the fraction of mutant embryos produced. 'Drosophila-style'genetics: saturatton mutasenes~ Studies in Drosophila introduced file idea of satu- ration genetic screens to metazoan developmental biology. Genetic screens large enough to identify almost all mutable genes involved in embryonic development have yielded profound new insights into the molecular mechanisms that underlie development "~, Such screens involve two generations of breeding. Go males are .mutagenized and Ft founder animals estab- lished. From individual Ft founders, F 2 families are raised, half of which are heterozygous for the newly induced mutations, Sibling crosses among the F2 (or F 2 X Ft backcrosses) are performed to identify mutant phenotypes in the F 3 embryos, Most mutations that affect pattern formation and organogenesis are lethal at late embryonic or early larval stages, and thus are referred to as 'embryonic lethal' or si,;lply 'lethal' mutations. In vertebrates, such screens were considered im- possible until recently. In mice, the breeding of thousands of lines to drive mutations to homozygosity in a two-generation breeding schedule is possible, although expensive. Screening for embryonic mutant phenotypes, however, is extremely difficult, since TIG MAY 1994 VOL 10 No, 5