Enhancer and gene traps for molecular imaging and genetic analysis in zebrafish


  • Le A. Trinh,

    Corresponding authorCurrent affiliation:
    1. Molecular and Computational Biology, University of Southern California, Los Angeles, California, USA
    • Division of Biology, California Institute of Technology, Beckman Institute (139-74), Pasadena, California, USA
    Search for more papers by this author
  • Scott E. Fraser

    1. Division of Biology, California Institute of Technology, Beckman Institute (139-74), Pasadena, California, USA
    Current affiliation:
    1. Molecular and Computational Biology, University of Southern California, Los Angeles, California, USA
    Search for more papers by this author

Author to whom all correspondence should be addressed.

Email: ltrinh@caltech.edu


Enhancer and gene trapping methods are highly effective means for the identification and functional analysis of transcriptionally active genes. With recent advances in fluorescent proteins and transposon based integration technologies, a growing family of trapping approaches has been developed in zebrafish, offering powerful tools to both visualize and functionally dissect gene networks during development. Coupled with the intrinsic advantages of the zebrafish model system, creative genetic engineering of trap vectors has enabled high-resolution molecular imaging and genetic manipulations. This review highlights the different enhancer and gene trap approaches that have been developed in zebrafish and offers insights into the strengths, limitations and experimental strategies for their application to enrich our knowledge of gene function and the cellular processes they control.


Enhancer and gene trap approaches have played central roles in defining the expression patterns of genes and gene function (Stanford et al. 2001; Brickman et al. 2010). Enhancer traps (ET), in which a basal promoter driving a reporter gene (e.g. lacZ or green fluorescent protein [GFP]) is integrated randomly into the genome, offer a simple and direct means to assay the regulatory modules that control gene expression (Bellen 1999). First introduced in Drosophila (O'Kane & Gehring 1987), an ET reports gene regulatory action but does not provide insight into gene function. To meet this need, the gene trap (GT) has evolved to offer information on both spatiotemporal expression patterns and gene function (Gossler et al. 1989; Joyner et al. 1992; Skarnes et al. 1992; Skarnes et al. 2004). With the advent of genetically encoded fluorescent reporters, GT have evolved further, providing functional fusion proteins that are tagged with a vital marker and can thus be used in cell biology and in vivo molecular imaging studies.

The advantages of zebrafish as a model organism for studying vertebrate development has also lent themselves to the use of gene trapping approaches. Zebrafish are amenable to genetic techniques, such as mutagenesis and transgenesis. GT, which relies on the random insertion of a reporter throughout the genome to capture transcriptionally active loci, is a form of transgenesis. With the development of transposon base integration technologies, insertions of reporters have become highly efficient in zebrafish making high-throughput GT screens feasible (Kawakami 2004a; Kawakami et al. 2004b; Urasaki et al. 2006). Additionally, the embryos are transparent, which allows researchers to apply optical imaging techniques to follow developmental processes in real-time. ETs and GTs using fluorescent reporters provide the biological label to report gene expression, cell behavior, and tissue morphogenesis in the developing embryo. In combination, these features have allowed the evolution of ET and GT vectors into approaches for both molecular imaging and mutagenesis in zebrafish.

Despite the advantages of the zebrafish model system, the lack of embryonic stem (ES) cells and homologous recombination technology has prevented targeted knockout to create gene deletions. Alternative reverse genetic knockout technologies such as anti-sense morpholino and site-specific nucleases have been developed as an attempt to make up for this absence (Doyon et al. 2008; Meng et al. 2008; Bedell et al. 2011; Huang et al. 2011; Sander et al. 2012). As part of the alternative technology, the mutagenic capability of GTs serves as mutant resources for researchers. Additionally, GT vectors that can circumvent the lack of homologous recombination with recombinase-mediated cassette exchange (RMCE) can provide an alternative avenue for targeted genetic manipulation to be feasible in zebrafish (Trinh et al. 2011).

This review focuses on the recent progress in ET and GT approaches in zebrafish and how they can be applied for dynamic imaging and genetic analysis of embryogenesis. For perspective, an overview of basic vector design for ET and GT methods for efficient integration of vectors are discussed. The various ET and GT vectors that are currently available in zebrafish and their capabilities are highlighted, with an emphasis on the potential application of the different approaches for molecular imaging.

Gene trap configurations

Three main classes of GT approaches using transposable elements or retroviral constructs to randomly integrate a reporter into the genome have been developed: enhancer (ET), gene (GT) and protein traps (PT) (Fig. 1). Each of these classes has essential elements in vector design that enable the annotation of regulatory DNA, the transcriptome, proteome or mutagenesis of the genome. ETs report transcriptional activity with limited mutagenic capabilities. This is achieved by randomly integrating a weak basal promoter upstream of a reporter gene encoding β-galactosidase or a fluorescent protein, into the genome (Fig. 1a). Expression of the reporter gene reflects the activity of nearby enhancers that can act at a distance on the weak promoter. Traditional GTs reveal both endogenous gene expression patterns and gene function by creating a defined mutation upon integration. Mutagenesis is enabled by the integration of a promoterless reporter encoded within a 3′ exon sequence downstream of a splice acceptor (SA) (Fig. 1b). The 3′exon contains termination and polyadenylation signals to create a truncation of the trapped gene. In the absence of a promoter, the expression of the reporter is dependent on upstream sequences and the SA. Mutagenesis is accomplished by creating a 5′ fusion of the trap gene with the reporter at the SA junction. In contrast, PTs report both transcriptional activity and full-length protein expression by creating a protein fusion of the reporter and the full-length trapped gene. Full-length protein tagging is achieved by using an internal exon reporter that incorporates both SA and splice donor (SD) sequences flanking the fluorescent reporter (Fig. 1c). The coding sequence for the reporter of a PT vector is devoid of a start and stop codon and therefore, will only be expressed when an in-frame fusion transcript of the trap gene and the reporter exon is made. The main distinction between a traditional GT and its variant, the PT, is that GTs are intended to create a truncated protein fusion of the reporter and the trapped gene, while PTs create full-length protein fusions that can be mutagenic at low rates they are not intended to truncate the trap protein and be mutagenic (Morin et al. 2001; Clyne et al. 2003; Buszczak et al. 2007). Variations of these three main configurations for vector designs have been used for screening efforts in zebrafish. The different components that distinguish the ETs, GTs, and PTs reported in zebrafish are discussed in the following sections.

Figure 1.

Schematic of enhancer trap (ET) and gene trap (GT) configurations. (a) An ET vector contains a basal promoter with a minimal activity upstream of the reporter gene (green rectangle) with a polyadenylation (pA) signal (green triangle). Insertion of the ET vector close to endogenous transcriptional regulatory elements of gene X (blue oval) will lead to the expression of the reporter (green) and gene X (blue). (b) The GT vector contains a splice acceptor (SA) (beige rectangle) upstream of a promoterless reporter (green rectangle) with a (pA) signal (green rectangle), also referred to as 3′ exon due to absence of a start codon in the reporter. Transcription of the reporter is dependent on the upstream regulatory element of gene X but translation is dependent on proper splicing of the splice acceptor to the 5′ exons of gene X (blue-green rectangle). (c) The protein trap (PT) vector contains an SA and splice donor (SD) site flanking an internal exon reporter. The internal exon is devoid of a start and stop codon. Integration in an intron with the correct orientation and reading frame leads to a fusion transcript of the 5′ and 3′ exons flanking the reporter exon (blue-green rectangle). Basic elements of enhancers and gene trap vectors are flanked by transposable elements (TE) for random insertion into the genome (exons, blue rectangle).

Transposable systems for integration

While ET and GT approaches have been extensively developed in Drosophila and mammalian systems (Bellen 1999; Skarnes 2005), ETs and GTs became feasible in zebrafish through the development of transposable systems that enable the efficient integration of DNA sequences into the zebrafish genome (Kawakami 2004a; Urasaki et al. 2006). The most widely used, the Tol2 transposable system developed by Kawakami and colleagues have been reported to have a germline transmission frequency >50% with multiple insertions per founder (Kawakami et al. 2004b; Balciunas et al. 2006). These high rates of germline insertion make it feasible to perform the high-throughput screens necessary to achieve genome-wide coverage from random approaches such as ET and GT. An added advantage is that the Tol2 transposon elements have been shown to mobilize in species ranging from Drosophila and Xenopus to chicken, mouse, and human (Kawakami 2007; Sato et al. 2007; Shibano et al. 2007; Urasaki et al. 2008), making it easy to deploy ET and GT vectors developed with the Tol2 system in zebrafish to other species.

In their efforts to dissect the functional elements of the Tol2 transposable system, Kawakami and colleagues created one of the first GT vectors in zebrafish (Kawakami et al. 2004b). The GT vector, T2KSAG, contained a SA from the rabbit β-globin gene, a promoterless GFP gene, and the SV40 polyA signal (Fig. 2a). In their pilot screen with the T2KSAG vector, they identified 36 independent expression patterns in the F1 embryos of 156 injected parental (F0) fish. These expression patterns were linked to endogenous transcripts by 5′ Rapid Amplification of cDNA Ends (RACE) analysis, which identified the transcripts fused to the GFP sequence, demonstrating that the rabbit β-globin SA can be functional in zebrafish. While the trap lines did not lead to any detectable phenotypes in the homozygous, the ability to generate stable insertions at a high rate (~51% F0 individuals had germline insertions and an estimated trapping rate of 8% in F1) using the Tol2 system set the groundwork for future GT vectors based on this transposable system.

Figure 2.

Schematic of variations on gene trap (GT) approaches. (a) The T2KSAG vector consist of a splice acceptor (SA) (beige rectangle), the green fluorescent protein (GFP) coding sequence (green rectangle), and an SV40 polyA signal (green triangle), flanked by the Tol2 transposable elements (TE, grey rectangle). Insertion of the T2KSAG vector into an intron leads to the expression of a fusion gene product of the 5′ exons (blue rectangle) of the trap gene and the GFP reporter. Alternatively, enhancers (blue oval) upstream of the T2KSAG vector can drive expression of the start codon containing GFP reporter, leading to expression of GFP without trapping the gene. (b) The FlExTrap vector contains a SA (beige rectangle) upstream of the mCherry reporter exon (red rectangle) with a polyA signal from the BHG gene (red triangle). Insertion into an intron of the same phase leads to the trapping of the gene (Trapping insertion). Heterotypic FRT (purple triangles) and lox (orange triangles) sites flank the reporter exon to enable inversion of the SA-containing cassette to rescue the trap locus. (c) The Gene-breaking (RP2) vector contains a 5′ trapping and a polyA-trapping cassette. The 5′ trapping cassette consists of a SA (beige rectangle) upstream of the red fluorescent protein (mRFP) reporter exon (red rectangle) with a polyA signal (red triangle) and transcriptional terminator (beige hexagon). The polyA-trapping cassette consists of the carp β-actin promoter (blue oval) upstream of the GFP coding sequence with a downstream splice donor (SD, beige rectangle). The 5′trapping cassette creates a truncated fusion of the N-terminus of the trap gene and mRFP, while the polyA-trapping cassette expresses GFP ubiquitous. All three vectors use the Tol2 transposable elements (TE, grey rectangle) for random insertion into the genome (exons, blue rectangle).

In addition to Tol2, the Sleeping Beauty (SB) transposable system has been used for integration of both ET and GT vectors in zebrafish (Balciunas et al. 2004; Sivasubbu et al. 2006). SB is an engineered Tc1-like element reconstructed from ancient fish fossil sequence (Ivics et al. 1997). While transgenesis efficiency with the SB elements are several fold higher than standard plasmid injections (Davidson et al. 2003), SB-base DNA integration has not been as efficient as Tol2-based vectors (Balciunas et al. 2006). However, the integration rates are sufficient for testing of vectors as Balciunas and colleagues used the SB system to dissect the minimal sequence of the ef1a promoter that can function in an ET (Balciunas et al. 2006). Additionally, the SB system was used to test variations on GT approach to increase the mutagenic capacity of GT vectors by “gene-breaking” (see “GT for functional analysis” section) (Sivasubbu et al. 2006).

Retroviral vectors, the most widely used approach in mouse, are an additional means for the random integration of ET and GT constructs. The only retrovirus system shown to efficiently integrate in zebrafish is based on the Moloney murine leukemia virus (MLV) pseudotyped with the vesicular stomatitis virus (VSV) envelope glycoprotein G (Gaiano et al. 1996a; Gaiano et al. 1996b; Amsterdam et al. 1997; Amsterdam et al. 1999; Amsterdam et al. 2004; Golling et al. 2002). It was originally developed as an insertional mutagenesis technology for efficient DNA integration in zebrafish (Gaiano et al. 1996a,b). Pseudotyped MLV has the advantage of being the most efficient insertional system in vertebrates, averaging over 40 proviral copies per cell (Chen et al. 2002; Golling et al. 2002). While retroviral vectors have high integration efficiency and have been used extensively for GTs in ES cells, they have become less popular in zebrafish due to the difficulty in creating high-titer viruses, cargo size limitations and the ease of using transposable systems. In their test for germline transmission of the pseudotyped MLV, Chen and colleagues reduced the GT reporter to a small FLAG epitope as larger reporters such as β-galactosidase and fluorescent proteins reduced viral titers (Chen et al. 2002). Additionally, they found that including additional FLAG epitopes to enable bidirectional gene trapping significantly reduced viral titers. Reduced viral titers presumably would reduce integration efficiency. However, Ellingsen and colleagues were able to create an ET in which the proximal gata2 promoter, a fragment of 1024 bp, was used to drive YFP expression with the pseudotyped MLV (Ellingsen et al. 2005). Using the gata2 ET vector, they were able to generate virus titers in the 10CFU/mL range and found YFP expression patterns due to enhancer trapping in one out of three founders screened (Ellingsen et al. 2005). This high rate of enhancer trapping indicates that retroviral based vectors can be as productive in ET and GT screens as transposable systems.

Enhancer traps as cell-type reporters

Initially developed in Drosophila to detect genomic elements that regulate transcription, ETs have evolved into tools to mark specific cell-types and manipulate gene expression in a spatiotemporal restricted manner. In zebrafish, a number of ETs screens have been performed to create a large library of several hundred transgenic lines that serve as cell- or tissue-specific fluorescent reporters and drivers to control expression of effector genes (Ellingsen et al. 2005; Scott et al. 2007; Asakawa et al. 2008b; Nagayoshi et al. 2008; Distel et al. 2009; Kondrychyn et al. 2009; Poon et al. 2010; Kondrychyn et al. 2011; Scott & Baier 2009). All the ETs reported in zebrafish contain a fluorescent protein as the reporter of transcriptional activity. The fluorescent reporter allows cells to be followed through development as they undergo morphogenesis and differentiation.

While ET screens cannot be targeted to specific cell or tissue types, a number of screens have been performed in which only lines showing reporter expression in tissue of interest were propagated (Scott et al. 2007; Distel et al. 2009; Scott & Baier 2009; Poon et al. 2010). For ET screens in which the goal is to generate fluorescent reporters or drivers in specific cell-types, selectively maintaining lines with specific tissue expression minimize the workload. However, this approach limits the number of enhancers that can be mapped as enhancers controlling similar expression patterns are ignored and only those that drive cell- or tissue-specific expression would be isolated.

The basal promoter used in an ET vector can affect trapping efficiency and observed expression patterns. A number of basal promoters have been used in zebrafish for ET screens: ef1a, keratin8 (krt8), keratin4 (krt4), gata2, hsp70, c-fos, E1b, thymidine kinase (TK), and a hybrid TK, carp β-actin promoter (TKBA) (Balciunas et al. 2004; Ellingsen et al. 2005; Scott et al. 2007; Nagayoshi et al. 2008; Kondrychyn et al. 2009; Ogura et al. 2009). The frequency of generating unique expression patterns with constructs using these promoters range from 10% to 58%. The efficiency of the basal promoter for enhancer trapping may reflect their responsiveness to genomic enhancers as a direct comparison of three basal promoters, hsp70, c-fos, and E1b, show differences in the frequency of unique expression patterns and biases for specific tissue expression of ET lines isolated (Scott & Baier 2009). The E1b promoter exhibited a strong bias for traps with expression in cranial ganglia, while c-fos showed the lowest non-specific expression patterns. Basal non-specific background expression in particular tissues such as muscle and dermis has also been reported for some promoters (Scott et al. 2007; Asakawa & Kawakami 2008a; Asakawa et al. 2008b). Non-specific background expression is a major problem for ET lines that are used to drive expression of effector genes in binary systems as it limits the tissue specificity of effector gene expression.

Fluorescent reporters in cell and tissue-specific ETs have used as vital markers to visualize the cellular dynamics involved in both development and disease. Time-lapse analysis of a neurophil specific enhancer showed that neurophils migrate to sites of inflammation upon infection of Mycobacterium marinum. This cellular response induces granuloma formation (Meijer et al. 2008). Similar imaging with confocal microscopy of the hoxb4a:YFP ET showed that hoxb4 expression is mosaic in subpopulations of neurons in rhomobomere 7 and 8 (Ma et al. 2009). The mosaic expression is reflected at the endogenous mRNA level. However, the ET line captures only approximately 70% of hoxb4 mRNA expression at a given time. The discordance between endogenous mRNA and the YFP reporter expression suggest that in some cells hoxb4 enhancer elements can not drive expression of the ET construct. Alternatively, there is a temporal difference between the expression of the endogenous mRNA and protein reporter.

Enhancer traps to drive effector gene expression

The binary Gal4-UAS system can be used in combination with ET and GT to drive expression of effector genes. This system relies on tissue-specific expression of the yeast Gal4 transcriptional activator and a second effector transgene under the control of the Upstream Activation Sequence (UAS) that is bound by Gal4 (Brand & Perrimon 1993; Duffy 2002). When the driver and effector components are bought together in the same cell through genetic crosses, the effector is transcribed only in cells expressing the Gal4 protein. This has been a powerful approach for manipulating gene expression in Drosophila and has been the driving force for the creation of libraries of Drosophila ET lines to drive Gal4 expression in specific tissues (Duffy 2002).

Variations on the Gal4-UAS system have been combined with ETs in zebrafish (Scott et al. 2007; Asakawa & Kawakami 2008a; Asakawa et al. 2008b; Distel et al. 2009; Ogura et al. 2009). To map neural circuits in zebrafish, Gal4-VP16 ET lines have been generated in combination with a UAS:Kaede reporter (Scott et al. 2007; Scott & Baier 2009). The photoconvertible Kaede reporter provides an added level of control for visualization of labeled cells as ultraviolet light can irreversibly convert a subset of UAS-driven Kaede cell from green to red fluorescence. To block synaptic transmission in zebrafish, a UAS-driven tetanus toxin transgene and Gal4 ET lines have been used to study neural function (Asakawa et al. 2008b). The ability to drive expression of effectors with tissue specific Gal4 ET lines adds to the arsenal of available tools for the analysis of gene function in zebrafish.

A drawback of using Gal4-VP16 in ET lines is that a high level of expression can lead to lethality (Koster & Fraser 2001; Scott et al. 2007). To avoid toxicity of Gal4-VP16, Gal4-activator derivatives have been generated for ET lines, Gal4FF, KalTA4 and Gal4-NFκB (Asakawa et al. 2008b; Distel et al. 2009; Ogura et al. 2009). These derivatives rely on modifying the transcriptional activation module fused to Gal4 to attenuated activation and reduce non-specific protein–protein interactions. Attenuated transcriptional activation of Gal4 derivatives can be compensated by using multiple UAS repeats to increase the expression of UAS-driven effector genes (Koster & Fraser 2001). For low expressing enhancers, UAS repeats can proportionally increase the expression of UAS-driven reporters to provide a bright fluorescent signal for visualization and imaging. However, repetitive UAS transgenes are subject to abrogation (Goll et al. 2009; Akitake et al. 2011), therefore, it is important to balance the negative effects of high expression levels and imaging needs.

In addition to UAS effectors, the development of photosensitive fluorescent proteins allows ET lines to be deployed for functional analysis in a spatiotemporal matter. The photosensitized fluorescent protein, KillerRed (KR) generates reactive oxygen species (ROS) in cells when exposed to intense green or white light illumination, resulting in damage to cells and death (Teh et al. 2010). A membrane tagged KR (mem-KR) has been used to generate ET lines that label cell morphology due to the accumulation of the mem-KR signal on the cell membrane (Teh et al. 2010). High intensity light can induce cell damage in mem-KR ET lines in a dose-dependent manner, allowing for functional analysis of KR-expressing cells. Targeting the time and location of light exposure can allow the tuning of cell death to ask questions about cell autonomy, neighbor interaction and role of ROS in specific organ formation.

Gene trap for functional analysis

A major distinction between ET and GT approaches is the mutagenic capability of GT vectors. A number of variations on the basic designs of GT vectors that allow for mutagenesis have been used in zebrafish. The first is the traditional SA-containing vectors designed to report integration events downstream of the promoter (Skarnes et al. 2004; Skarnes 2005). As previously mentioned, this approach was used to dissect the functional elements of the Tol2 transposable system (Kawakami et al. 2004b). The presence of a start codon in the GFP reporter for the T2KSAG vector allowed trapping of genes when the vector inserted upstream of the initiation codon. In these insertional events, the GT would not create a mutagenic truncation of the trap locus (Fig. 2a). These results suggest that a SA-containing vector would be more mutagenic with reporters that do not contain a start codon.

A variation on the traditional SA-containing vector, referred to as Flip and Excision (FlEx) in cell culture (Schnutgen et al. 2003), combines elements to enable bidirectional trapping and conditional mutagenesis of GT locus by Flp and Cre mediated recombination has been adapted to zebrafish (Ni et al. 2012). The FlExTrap approach uses heterotypic Flp- and Cre-recombination sites flanking the SA-containing element to enable inversion of the SA-containing cassette (Fig. 2b). This allows the GT vector to function regardless of the starting orientation of the insertion. Non-trapping insertional events can be converted to the trapping orientation by Cre- or Flp-mediated recombination. Additionally, insertional events that are mutagenic can be rescued by inversion of the GT cassette (Fig. 2b). This approach provides flexibility for genetic manipulation. Theoretically using ligation mediated (LM)-PCR-base screening, high numbers of insertional events can be identified, similar to retroviral mutagenesis. However, by focusing on insertional events regardless of reporter expression, the potential associated with using reporter expression to identify spatiotemporal expression patterns and subcellular localization of the trap gene would not be fully leveraged. In practice, Ni and colleagues used the mCherry reporter in the SA-containing cassette to screen for expression of a mCherry fusion protein instead of LM-PCR base screening (Ni et al. 2012).

A third approach used in zebrafish referred to as “gene-breaking” (GB) combines a traditional SA-containing element with a polyA-trapping element to optimize mutagenesis and ease of identifying the trap gene (Sivasubbu et al. 2006; Clark et al. 2011). In polyA trapping, a promoter upstream of the reporter dissociates the expression of the reporter from the trap gene (Brickman et al. 2010). An SD sequence downstream of the reporter allows capture of exons 3′ of the insertion site (Fig. 2c). PolyA trapping vectors have been used in ES cells to drive expression of selectable markers, allowing for identification of loci that are transcriptionally inactive in ES cells (Niwa et al. 1993; Salminen et al. 1998; Zambrowicz et al. 1998). In zebrafish, polyA trapping with a fluorescent reporter (also referred to as a “gene-finding cassette”) allows for screening of integration events by fluorescent microscopy without relying on the expression of the trap locus (Sivasubbu et al. 2006; Clark et al. 2011). In addition, capture of the polyA enables cloning of the trap locus by 3′RACE, an approach that is more robust than 5′RACE.

Two GB constructs have been reported in zebrafish (Sivasubbu et al. 2006; Clark et al. 2011). The initial GB construct, pT2/PAT6, used the carp β-actin promoter to drive expression of GFP with a downstream SD (Sivasubbu et al. 2006). The GFP reports insertional events; however, the SA-containing cassette for 5′ trapping did not encode a visible reporter, preventing the visualization of the spatiotemporal expression patterns of the trap locus. A second-generation GB construct, referred to as RP2, added a number of elements to allow both visualization of N-terminal protein expression and conditional rescue of the trap locus (Fig. 2c) (Clark et al. 2011). To report the spatiotemporal expression of the trap locus, a red fluorescent protein (mRFP) reporter was added downstream of the SA. This created a truncated fusion protein of the N-terminus of the trap gene and mRFP (Fig. 2c). The polyA trapping element with GFP under the control of the carp β-actin promoter serves as a reporter of integration events (Fig. 2c). The presence of loxP sites flanking the GB construct permits selective rescue of the trap locus by Cre recombination. Using the RP2 approach, Clark and colleagues generated a library of over 300 GT lines. Eight of the 35 lines that were tested had viability or morphological defects as homozygous. Knockdown of wildtype transcript in homozygous GT larvae ranged from 77% to 99% irrespective of whether the line exhibited visible phenotype. This knockdown efficiency was attributed to the presence of a strong polyA-trapping element and suggested that the RP2 approach can complement existing genetic knockdown resources in zebrafish.

In addition to efficient mutagenesis, the RP2 vector was designed to create an in-frame translational fusion of the trap gene with the mRFP reporter, as the coding sequencing for mRFP in the SA-containing cassette was devoid of a start codon (Clark et al. 2011). This provided a direct mRFP tag of the N-terminus of the trap gene (Fig. 2c). In some cases, the mRFP truncated fusion protein exhibited distinct subcellular localization. Whether the subcellular localization patterns recapitulate the endogenous protein will be dependent on the insertion of each trap locus. The mRFP truncated fusion protein may not recapitulate endogenous protein localization in cases where the protein localization signal is contained in the C-terminal domain, are mis-folded or targeted for degradation.

Protein trap for molecular imaging and functional analysis

Although GT approaches were initially designed for mutagenesis, an alternative strategy of protein trapping has the potential of converting GTs into tools for molecular imaging. To tag full-length proteins via GT, PT vectors use an internal exon encoding a fluorescent reporter that is devoid of a start and stop codon (Fig. 1c). This generates an in-frame full-length fluorescent fusion protein of the trap locus when the integration event occurs between exons that are in the same reading frame as the reporter exon. The strategy permits full-length protein localization and dynamics studies to be performed on trap lines that are not feasible with traditional 5′ tagging that leads to a truncated fluorescent fusion protein. The retention of all regulatory sequences of the endogenous gene, including both 5′ and 3′ UTRs ensures that all transcriptional and translational regulatory elements that could affect the expression levels, localization and function of the protein are fused to the fluorescent reporter.

In zebrafish, the FlipTrap approach combines PT with traditional SA-containing GT by conditional recombination of the PT cassette with the Cre-lox system (Fig. 3) (Trinh et al. 2011). In the forward orientation, the FlipTrap vector consists of an internal exon encoding the yellow fluorescent protein, citrine, flanked by a SA and SD sequences from the zebrafish ras associated domain family 8 (rassf8) gene (Fig. 3a). Insertion of the FlipTrap vector into an intron of the same phase as the reporter by Tol2 transposition leads to the splicing of the citrine sequence into frame with the endogenous mRNA and hence, the formation of a full-length fluorescent fusion protein (Fig. 3a). In the reverse orientation, the vector contains a 3′ exon encoding the red fluorescent protein, mCherry with a termination and polyadenylation signal. Two pairs of heterotypic lox sites flanking the internal citrine exon and the 3′mCherry exon allow for flipping of the 3′mCherry exon and excision of the citrine and SD sequences in the presence of Cre recombinase (Fig. 3b). Heterotypic Flp Recognition Target (FRT) sites flanking the citrine and mCherry cassette permit the excision and replacement of the FlipTrap insertion by another DNA cassette in the presence of Flp recombinase (Fig. 3c).

Figure 3.

Schematic of FlipTrap approach. (a) The FlipTrap vector consists of a citrine coding sequence (green rectangle) flanked by a splice acceptor (SA) and SD (beige rectangle). In the reverse orientation are mCherry (red rectangle) and polyA (red triangle) sequences. Heterotypic lox sites are shown for loxP (light grey triangles) and loxPV (dark grey triangles). Flp Recognition Target (FRT) sites (blue triangles) are positioned internal to the transposable elements (TE, grey rectangles). Insertion of the FlipTrap by transposition into the intron of an actively expressed gene leads to splicing of the citrine exon, allowing expression of a full-length fluorescent fusion protein when the insertion is in-frame with the trapped gene. (b) Schematic of conditional mutagenesis by Cre mediated recombination. Cre recombination of the lox sites (grey triangle) leads to “flipping” of the mCherry and polyA sequence into the forward orientation and excision of the citrine and SD sequences. Expression of the Cre induced mutant allele leads to the production of a truncated protein fused to mCherry. (c) Schematic of cassette exchange system in FlipTrap. The exogenous DNA cassette (X, grey rectangle) with SA (beige) sequence is flanked by FRT sites (dark blue triangles). In the presence of Flp recombinase and exogenous DNA, the FlipTrap cassette is replaced and converted to a driver line.

Screening for developmentally expressed genes with the FlipTrap vector in our laboratory have led to the creation of a library of fluorescently tagged full-length protein (Trinh et al. 2011). The FlipTrap line exhibits a wide array of distinct expression patterns that can be used to map cell lineages by molecular identity and study cellular events during development (Zigman et al. 2011; Palanca et al. 2012; Choe et al. 2013). This is best illustrated in confocal time-lapse microscopy of trap lines of proteins that exhibit dynamic subcellular localization patterns throughout the cell cycle (Fig. 4). Analysis of trap lines in genes known to be regulated through the cell cycle such as lamina-associated polypeptide 2βb (LAP2β), a component of the nuclear envelope (NE), and kinetochore associated 2-like (kntc2-l), a component of mitotic spindles, show the expected relocalization patterns during cell division. In the case of the GT(kntcl2)ct501a line, the citrine fusion protein is only visible upon condensation of the chromatin as cells divide, providing a tractable marker of mitotic cells in the developing embryo. Such trap lines provide a means to analyze the in vivo dynamics of cellular events in both wildtype and mutant phenotypes.

Figure 4.

In vivo dynamics of FlipTraps through the mitotic cycle in gastrulating zebrafish. Single focal planes of epiblast cells in Gt(lap2β-citrine)ct3b (top panels) and Gt(kntc2l-citrine)ct501a (bottom panels) embryos expressing membrane-mCherry (red) and H2B-cerulean (blue) to visualize the membrane and nuclei, respectively. White arrow highlight cells undergoing mitosis. The FlipTrap Citrine fusion proteins are seen in green. LAP2β is a component of the inner membrane of the nuclear envelope, while Kntcl2 is a component of the kinetochore, a site where spindle fibers attach to chromosomes. During mitosis, the nuclear envelope disassembles and disappears from around the condensing chromatin. LAP2β reappears around the chromatin during telophase. Kntc2l expression condenses around the chromosomes only in cells undergoing mitosis. The retention of native localization signals and function of the Citrine fusion proteins offers a means to analyze the in vivo dynamics of each of the trapped proteins in the living embryo. Scale bar, 10 μm.

While PT approaches are design to capture the entire transcript and provide localization information of the full-length protein, insertion of a fluorescent protein can lead to disruption of protein function. However, protein domain analyses of the FlipTrap lines indicate that the majority of insertion events (89.9%) occur between two exons that do not encode a single protein domain (Trinh et al. 2011), making it less likely that the fluorescent fusion leads to aberrant function or interactions. Characterization of FlipTrap lines in which the insertion events occur between two exons that encode a single protein domain suggest there is a strong correlation between integration events that disrupt protein domains and the presence of mutant phenotypes (Trinh et al. 2011). The strong experimental bias for insertional events that do not disrupt protein domains may be due to the screening protocol for visible fluorescent fusion protein expression. In an expression screen, integration events that would affect protein folding or stability would not be isolated.

Relying on in-frame fusion in PT approaches can reduce the number of integration that would capture gene products. However, in eukaryotes, exons are predominantly symmetric, carrying the same phase on both ends (Long et al. 1995; Tomita et al. 1996; Fedorov et al. 1998; Sakharkar et al. 2002). Additionally, the distribution of intron phases is non-uniform, with a strong bias for phase-0 introns (~48%) (Long et al. 1995). This predicts that the probability of generating an in-frame fusion with a PT or GT vector designed with a symmetric frame zero reporter should be the same, 1 in 8 insertional events (1/2 in frame * 1/2 for intronic insertion * 1/2 in right orientation). The FlipTrap vector is predicted to have a similar trapping rate as the citrine exon was designed in the zero reading frame and symmetric (Trinh et al. 2011). While, a trapping rate that takes into account all insertional events (silent and visible traps) has not been assessed for this vector, the rate at which positive lines were isolated from F0 fish is in the same range as those reporter for GT screens, from 6.7% to 18.3% (Trinh et al. 2011). Additionally, insertion site analysis shows that 98.7% of the isolated FlipTrap lines had an insertion in phase 0 introns and generated in-frame fusions of the reporter exon with the trap gene (Trinh et al. 2011). These results suggest that PTs can be as efficient as GT approaches.

While full-length fusion protein libraries have proven to be an enormous resource for protein localization studies in Drosophila where the PT approach was initially developed (Morin et al. 2001; Clyne et al. 2003; Buszczak et al. 2007), a standard PT is fundamentally limited for gene function studies, as mutant alleles cannot be generated. To overcome this disadvantage, the FlipTrap vector has the ability to create conditional mutant alleles by Cre-lox recombination. Cre-mediated recombination of heterotypic lox sites results in the excision of the citrine and SD sequence and “flipping” of the mCherry sequence into the forward orientation (Fig. 3b). The “flipped” allele leads to the expression of a truncated protein of the trap gene that can be detected by the conversion from Citrine to mCherry fusion protein expression. Knockdown of wildtype transcripts in homozygous Cre induced mutant allele range from 70% to 97%, indicating that Cre-induced mutant alleles can lead to efficient knockdown of the trapped genes and appears to be similar to knockdown efficiency reporter for the RP2 system (Clark et al. 2011; Trinh et al. 2011). Whether a reduction in wildtype transcript will lead to mutant phenotype depends on two factors. First, the insertion site along the length of the trap gene will determine the potential for generating a mutant allele. Insertions within 3′ regions are more likely to generate minor truncations of the protein, resulting in no visible phenotypes or perhaps weak hypomorphs. Second, redundant loci are less likely to have an observable or measurable phenotype. These limitations are the same faced by traditional SA-containing GT vectors. However, the ability to generate mutant alleles from the FlipTrap lines by Cre-lox recombination enables conditional mutagenesis in zebrafish, providing an added level of spatial and temporal control similar to the FlExTrap approach.

An additional feature of the FlipTrap approach is the extensibility of each of the trap lines due to the ability for RMCE (Trinh et al. 2011). In the presence of Flp recombinase and exogenous DNA that contains the same FRT sites, the FlipTrap cassette can be replaced (Fig. 3c). Test experiments indicate that this process is highly efficient as 94% of the injected embryos exhibit cassette exchange with a reporter in the somatic tissue (Trinh et al. 2011). Furthermore, experiments in our laboratory indicate that cassette exchange in the germline can range from 20% to 30% (unpubl. data, Trinh and Fraser, 2012). These rates suggest that new variants of the FlipTrap lines can be generated at a reasonable efficiency and opens the possibility to convert each FlipTrap locus into drivers for any sequence of choice. For example, cell-type specific FlipTrap lines can be use to control the expression of Gal4, similar to ET lines. More importantly, these results suggest that future generation of ET, GT and PT vectors should incorporate the ability for RMCE to allow targeted genetic manipulation in the absence of homologous recombination.


Gene trapping in zebrafish has evolved from the simple annotation of the transcriptome to complex tools for imaging and functional analysis that takes advantage of the intrinsic features of the system. Unlike gene trapping in ES cells in which mutagenesis is the key, the ease of applying imaging approaches in zebrafish has allowed ET, GT, and PT lines to be used for labeling and manipulating specific cell-types. The potential to use GTs beyond mutagenesis of the trap loci will ensure the extensibility of this approach. In particular, ET and GT combined with binary systems such as Gal4-UAS provide the possibility of extending the use of each line with the creation of new UAS effector lines. Additionally, vector design elements that add components to enable cassette exchange will further extend trap lines. Taken together, the different ET, GT and PT lines provide an enormous resource for the zebrafish community. The initial descriptions of these lines have been superficial due to focus on vector design to enable efficient identification and mutagenesis of trapped genes. Future work will require detailed analysis of the lines that have isolated from the various collections and application of these enormous resources in specific biological questions.


We thank Rusty Lansford and Masahiro Kitano for helpful comments. Work related to gene trapping in our laboratory has been supported by NHGRI Center of Excellence in Genomic Science grant P50HG004071.