The changing conditions of zebrafish mutants

In PNAS, Ni et al. (1) describe a gene trap construct that makes conditional inactivation of genes in zebrafish a reality. This represents a major milestone in zebrafish research. The relatively recent and rapid emergence of zebrafish as a vertebrate model organism has fundamentally been because of its utility in genetic approaches. The generation of mutations and subsequent screening for phenotypes in zebrafish requires significantly fewer resources than screens of equivalent numbers in mice or other mammalian models. Because of this advantage, it prompted two major phenotypic screens in the early 1990s that generated and identified hundreds of embryonic mutations (2, 3). These chemically induced alleles became a resource that propelled scientific advances in the zebrafish research community for nearly two decades. Zebrafish mutants have contributed biological insights ranging from basic mechanisms of gastrulation (4) to human pigmentation (5). However, the Achilles' heel of classical genetics, including zebrafish genetics, is a result of what Sean Carroll describes as the “genetic toolkit” (6), i.e., the same genetic pathways are often deployed in multiple different tissues at different times during development and adult homeostasis. The result is that, if you knock out a gene that has multiple functions, the massive pleiotropic changes that occur in the organism make it difficult to impossible to understand the specific role of that gene in, say, eye development or function. Mouse researchers devised an important method for circumventing this problem: conditional inactivation (7). By using homologous recombination, researchers can flank mouse exons with loxP recombination sites; then, by expressing the cre recombinase in a temporal or tissue specific manner, inactivate the gene by deleting the targeted exon(s). Unfortunately, neither of the two essential components of this strategy, ES cells or homologous recombination, are currently options for zebrafish researchers, so other strategies need to be pursued.

Nineteen years ago, Lin et al. demonstrated the first efficient insertion of a DNA element into the zebrafish genome via an integrase by using pseudotyped retroviruses (8). It was later shown that those insertions could be mutagenic, and the first large-scale screen for zebrafish mutations that used an inserted DNA element as the mutagen was performed (9). Since that time, there have been significant advancements in zebrafish transgenesis, particularly the use of the transposable element Tol2 (10), and many recent efforts have focused on making Tol2 an efficient mutagen. Most of these strategies rely on the basic idea of “gene-trapping,” whereby a strong splice acceptor is inserted between two exons, capturing the normal mRNA message and resulting in a premature truncation of the protein (usually in conjunction with a visible reporter gene such as GFP) (11–13). The article by Ni et al. (1) describes a major advancement in the utility of Tol2-based gene traps. Inside the Tol2 vector, they incorporated a “flippable” trap construct. If the transposon integrates in an intron in one orientation, a “strong” splice acceptor linked to mCherry captures nearly 100% of the mRNA splicing, and, because of multiple polyA sites, kills the wild-type message (Fig. 1A). Expression of the gene is detected by red fluorescence in the fish, which conveniently also indicates where the gene is expressed in the embryo. “Leakiness” (i.e., splicing around the trap) has been a chronic problem in gene-trap constructs, and it appears that, by using a highly efficient splice acceptor and multiple polyadenylation sites, Ni et al. (1) have found a combination that efficiently kills the vast majority of the wild-type transcript. However, if the integration is in the other orientation, the trap does not function, creating a “neutral” integration in the gene (Fig. 1B). The result of integrations in this orientation is the mRNA is spliced around the trap. These neutral alleles should function normally in vivo.

Fig. 1.

The sequence for creating a conditional allele. (A) The gene trap randomly lands in a gene capturing the normal mRNA message, truncating the protein and (if in the correct reading frame) generating a fusion to mCherry. (B) Converting the mutant allele back to “WT” functionality is achieved by expressing flp ubiquitously in the fish with the goal of getting a flipped allele transmitted through the germline. For simplicity, only one series of recombinases is shown, with flp being used first, but cre or flp can be used in the first step, followed by the other recombinase in the second step. (C) Fish with the neutral integration are bred with fish carrying the neutral integration and cre expressed tissue-specifically (in this case in the eye). Twenty-five percent of the resulting embryos will have a tissue-specific KO of the trapped gene. (D) Another option with this construct is to directly breed KO mutants together with a tissue specific recombinase to correct the mutation in only the desired tissue.

The primary advantage of the “flip trap” is that the construct also includes frt and loxP recombination sites flanking the bidirectional trap. Having these sites in the construct provides a critical but previously elusive feature to zebrafish mutant alleles: conditional inactivation (Fig. 1C) or conditional activation (Fig. 1D). Expression of cre or flp recombinase will cause an inversion of the trap, converting it from neutral to mutagenic (or mutagenic to neutral) in a controlled manner, giving these zebrafish mutants the same flexibility as a mouse conditional KO. Thus, if cre or flp is driven in a tissue-specific or temporal fashion, the problems of pleiotropic effects can be minimized, much as they are in mouse. In fact, because you can also convert a mutagenic allele to a neutral one, it allows for conditional rescue (Fig. 1D). For example, returning function of a gene only in the eye to demonstrate that expression of a gene anywhere else is not essential for the fish. This is an added flexibility not typically found in most mouse conditional alleles. In addition, because the construct contains frt and loxP sites, the construct can be flipped twice, meaning the desired alleles can be created regardless of the original orientation of the integration, so integrations identified, not by gene expression, but by genomic mapping of the insertion site, can still be used for either type of experiment.

By developing the FT1 cassette, Ni et al. (1) radically expand what is possible in zebrafish genetic research, making bona fide conditional alleles possible. However, there are still some remaining obstacles

Ni et al. describe a gene trap construct that makes conditional inactivation of genes in zebrafish a reality.

left to overcome to bring zebrafish on par with mouse reverse genetics. First, a broad array of tissue specific cre and flp drivers are necessary to properly exploit the power of conditional alleles. The mouse community has invested tremendous resources in developing a wide variety of cre drivers (14), and similar efforts are needed in the zebrafish community. There are already some efforts under way to create such resources (http://crezoo.crt-dresden.de/crezoo/ and http://zcre.org.uk/), but many more cre and flp transgenic lines are needed to expand the utility of conditional alleles.

The second major hurdle is efficient targeting of specific alleles. Tol2 mutagenesis still requires random integration, limiting conditional alleles to randomly hit genes containing introns. Some form of homologous recombination or site-specific targeting of constructs is still needed if complete coverage of the zebrafish genome with conditional alleles is the ultimate goal. There are promising results in other systems with the use of nuclease-mediated recombination, and this may also be a possible mechanism in zebrafish, but it remains to be demonstrated. Still, the future of conditional alleles in zebrafish is bright and getting brighter.