A versatile strategy for gene trapping and trap conversion in emerging model organisms

Although established genetic models offer unmatched resources for genetic analysis, there is strong motivation to develop genetic tools in new species. This motivation stems from the diversity that is evident in development, morphology and physiology, which means that many questions cannot be addressed in the few well-established models. The development of transgenesis in emerging animal models, such as Nematostella vectensis (Renfer et al., 2010), Parhyale hawaiensis (Pavlopoulos and Averof, 2005), Tribolium castaneum (Berghammer et al., 1999) and Ciona intestinalis (Sasakura et al., 2007), represents the first step for establishing sophisticated genetic techniques in these species. One such technique, gene trapping, captures gene expression at the site of transgene insertion. Transposon-mediated gene trapping allows the implementation of unbiased genetic screens to identify new genes and provides valuable markers for in vivo imaging and phenotypic characterisation (Bellen, 1999; Bellen et al., 1989).

In Drosophila, gene traps are also used to generate GAL4 drivers, powerful tools that exploit endogenous genes to direct gene expression with spatial and temporal specificity (Brand and Perrimon, 1993). In principle, gene trapping can be used to introduce a wide range of genetic tools, such as alternative expression drivers, recombinases, specialised markers and knockdown constructs, into a trapped locus. In practice, this is difficult to achieve because each application relies on a different transgene construct, and traps derive from unique insertions that cannot be reproduced with each construct. Replacing one type of construct with another, at a given locus, is possible but technically challenging and restricted to highly developed genetic models (Sepp and Auld, 1999).

We present a new approach termed integrase-mediated trap conversion (iTRAC) that allows primary gene traps to be adapted for diverse applications through transgene conversion. The approach uses a primary exon-trapping vector, based on the Minos transposon, that incorporates an attP docking site for the ϕC31 integrase. Once a trap has been generated and selected, a virtually unlimited range of secondary constructs carrying the cognate attB site can be integrated specifically into the docking site at the trapped locus (Fig. 1). As a proof of principle, we demonstrate iTRAC in Parhyale hawaiensis, a crustacean that has emerged as an attractive model for developmental studies (Browne et al., 2005; Extavour, 2005; Gerberding et al., 2002; Liubicich et al., 2009; Ozhan-Kizil et al., 2008; Pavlopoulos and Averof, 2005; Pavlopoulos et al., 2009; Price et al., 2010; Rehm et al., 2009; Vargas-Vila et al., 2010).

The 1.3 kb PhHsp70a fragment (accession FR749989) was isolated by inverse PCR from Parhyale genomic DNA using a previously described approach (Pavlopoulos and Averof, 2005; Pavlopoulos et al., 2009) and cloned upstream of the DsRed/SV40polyA reporter cassette to obtain plasmid pSL(PhHsp70a-DsRed).

A 230 bp SpeI fragment containing the ϕC31 attP site from pTA-attP (Groth et al., 2000) was cloned into the SpeI site of pSL(PhHsp70a-DsRed) to generate pSL(attP;PhHsp70a-DsRed). The attP-PhHsp70a-DsRed construct was cloned as an AscI fragment into the Minos vectors pMi{3xP3-DsRed} and pMi{3xP3-EGFP} (Pavlopoulos and Averof, 2005; Pavlopoulos et al., 2004), generating pMi{3xP3-DsRed;attP;PhHsp70a-DsRed} and pMi{3xP3-EGFP;attP;PhHsp70a-DsRed}.

The transcription initiation and splice sites of PhMS, PhHS and PhHsp70a were mapped by 5′ RACE from transgenic animals carrying stable insertions of PhMS-DsRed (Pavlopoulos and Averof, 2005), PhHS-DsRed (Pavlopoulos et al., 2009) and PhHsp70a-DsRed (Distal^DsRed trap), respectively, using SMART-RACE (Clontech) and a reverse primer targeting the coding sequence of DsRed (5′-CTTGGTCACCTTCAGCTTGGCGGT-3′).

The EGFP coding sequence was excised as a NcoI-NotI fragment from pMi{3xP3-EGFP} and cloned into NcoI and NotI cut pSL(PhHS-DsRed) (Pavlopoulos et al., 2009) to generate pSL(PhHS-EGFP).

Plasmid pBS(MiL;attB;PhHS-EGFP) was generated as follows: we removed the right inverted repeat of Minos from pMi{3xP3-DsRed} by AvrII and NheI digestion and religation; we excised DsRed with SgrBI and NotI and replaced it with a SgrBI-NotI fragment of pSL(PhHS-EGFP) carrying PhHS-EGFP; we then excised 3xP3 using SalI and replaced it with a SalI fragment containing attB from plasmid pTA-attB (Groth et al., 2000).

Plasmid pBS(MiL;attB;PhHsp70a-EGFP) was generated by excising PhHS from pBS(MiL;attB;PhHS-EGFP) using SmaI and NcoI and replacing it by PhHsp70a obtained by SmaI and partial NcoI digestion of pMi{3xP3-DsRed;attP;PhHsp70a-DsRed).

Plasmid pMi{3xP3-DsRed;attP;PhHsp70a-DsRed} was injected with Minos transposase mRNA and stable transgenic lines were isolated as described previously (Pavlopoulos and Averof, 2005). Traps were imaged on a Leica MZ16F epifluorescence stereoscope.

Trapped genes were identified by 5′ RACE on each line using the SMART-RACE Kit (Clontech) with reverse primers DsRed-SMART (5′-CTTGGTCACCTTCAGCTTGGCGGT-3′) or hsp70aDsRedR (5′-GAGGCCATGGTTGTGGATT-3′). Additional cDNA sequences from the Distal locus were obtained by 3′ RACE on wild-type animals using forward primers targeting the sequences already determined (accession FR821313). In situ hybridisation using the Distal cDNA probe was carried out on wild-type embryos as described previously (Rehm et al., 2009). The Distal^DsRed line carries additional transgene insertions that do not give visible traps.

Splicing to PhHsp70a-DsRed and normal splicing at the Distal locus were measured by quantitative RT-PCR using a common forward primer targeting the Distal 5′ UTR (5′-TGACAGTCGCTGCGAAATAG-3′) and two reverse primers targeting DsRed (5′-GGGTGCTTCACGTACACCTT-3′) and a Distal 3′ exon (5′-GTCTGCTCGTCTTCCTTTGC-3′). RNA was isolated from populations of 20-30 heterozygous and homozygous Distal^DsRed embryos, reverse transcribed using oligo(dT) primers and amplified in triplicate on the MJ Research Opticon real-time PCR machine; PCR efficiency with each set of primers was 1.98 and 1.72, respectively. Trapped and normally spliced products were detected at a ratio of 0.09 (s.e.=0.18) in heterozygous embryos and 0.15 (s.e.=0.32) in homozygous embryos.

In vitro synthesised ϕC31 integrase mRNA was prepared as described previously (Groth et al., 2004). For the interplasmid assay, 1- or 2-cell stage Parhyale embryos were injected with plasmids carrying the attP and attB sites and ϕC31 integrase mRNA at 500 ng/μl, 500 ng/μl and 100 ng/μl, respectively. Surviving embryos were collected 24 hours after injection and nucleic acids were isolated by mechanical disruption in Holmes-Bonner buffer (100 mM Tris-HCl pH 7.5, 10 mM EDTA, 300 mM NaCl, 2% SDS, 7 M urea), triple phenol/chloroform extraction and ethanol precipitation. Recombination events were detected by PCR using two forward primers (5′-AGGAAGGGAAGAAAGCGAAA-3′ and 5′-CCAATTTCTATCTTAGCCCAACC-3′) and a common reverse primer (5′-GGGTGCTTCACGTACACCTT-3′), as illustrated in Fig. 4A.

Genomic integration via ϕC31 integrase was tested by injecting plasmid pBS(MiL;attB;PhHS-EGFP) and integrase mRNA (300 ng/μl and 100 ng/μl, respectively) into 1- or 2-cell stage embryos of a line carrying multiple attP insertions. Out of 512 injected embryos, 207 survived to late embryogenesis and 104 expressed EGFP after heat shock. Seventeen individuals with uniform expression were raised to adulthood; all produced progeny carrying the PhHS-EGFP marker. Wild-type embryos lacking attP insertions were injected as controls; out of 619 injected embryos, 255 survived as late embryos, but none expressed EGFP after heat shock. iTRAC was tested by injecting plasmid pBS(MiL;attB;PhHsp70a-EGFP) and ϕC31 integrase mRNA (300 ng/μl and 100 ng/μl, respectively) into 1- or 2-cell stage embryos of the Distal^DsRed line. Out of 389 injected embryos, 90 survived to late embryogenesis and 47 expressed EGFP in at least some limbs and in the characteristic pattern of Distal; 12 of these embryos showed bilateral replacement of DsRed fluorescence by EGFP. Four individuals were raised to adulthood, of which three gave rise to Distal^EGFP progeny.

Our Minos gene-trapping vector carries the 3xP3-DsRed or 3xP3-EGFP transformation marker (Berghammer et al., 1999; Pavlopoulos and Averof, 2005), the ϕC31 attP site, and a ‘trapping element’ (described below) upstream of the DsRed reporter and the early mRNA polyadenylation sequence of SV40. The two types of trapping strategies commonly employed, i.e. enhancer trapping and exon trapping, make use of a core promoter or a splice acceptor to capture the activity of cis-regulatory elements or splice donors of trapped genes, respectively. To create a trapping system that is widely applicable, we sought to identify promoters or splice acceptors capable of gene trapping in a range of species. First, we tested two core promoter elements, the Drosophila hsp70 basal promoter and an artificial ‘super core promoter’ (which combines several core promoter motifs) (Juven-Gershon et al., 2006), for enhancer trapping activity in Drosophila, Tribolium and Parhyale, but neither was found to work across the species tested (Schinko et al., 2010) (data not shown). Next, we searched for core promoters and splice acceptors among sequences that lie upstream of Parhyale hsp70 genes. Among the sequences tested, two elements were capable of efficient gene trapping in Parhyale: the heat-inducible element PhHS (Pavlopoulos et al., 2009) and a fragment named PhHsp70a. Both fragments could drive expression patterns specific to individual transgene insertions without any need for a heat shock (Fig. 2B-L). Using 5′ RACE on cDNA prepared from transgenic lines, we determined that PhHS contains a core promoter upstream of the transcription start site and a large intron within the 5′ UTR, whereas PhHsp70a is a truncated 5′ UTR sequence with a splice acceptor site that becomes spliced to the exons of trapped genes (see Fig. S1 in the supplementary material). We decided to focus on exon trapping mediated by PhHsp70a.

In our first experiment using the Minos{3xP3-DsRed;PhHsp70a-DsRed} vector, we recovered at least six independent traps from ~250 injected Parhyale embryos. Using the same vector, we also obtained four independent exon traps in Drosophila from ~350 injected embryos, suggesting that this construct can mediate exon trapping efficiently in diverse arthropods.

A variety of exon traps have been generated to date using the PhHsp70a-DsRed trapping construct in Parhyale (Fig. 2). These include traps with expression in the central nervous system, mesoderm, appendages, mouthparts, gills and other patterns. Most lines have been propagated through many generations over 4-5 years, demonstrating that the transgenes are stable, with continuing activity and no detrimental effects on reproduction and survival.

In emerging model organisms, gene traps are likely to be first used as markers for visualising specific tissues or cell types, providing a means to follow dynamic cell behaviours, to study physiological processes in vivo and to assess phenotypes following experimental manipulations. Some of our traps mark well-recognised organs, such as the nervous system, gills or paragnaths (Fig. 2D,F,H,K), whereas others mark complex populations of cells and previously undescribed cell types. For instance, one trap marks a previously uncharacterised cell type on the dorsal epidermis of late embryos, juveniles and adults (Fig. 2C) that is associated with specific sensory or structural elements in the epidermis of Parhyale (Fig. 3A).

Another trap allows us to image cardiac function. Parhyale has a typical arthropod heart, consisting of a muscular tube with three pairs of lateral inflow valves and an anterior outflow valve (Fig. 3B). Using this trap, we were able to observe heart function and to visualise the opening and closing of valves in unprecedented detail (Fig. 3C and see Movie 1 in the supplementary material).

A third trap, which we named Distal^DsRed, marks the distal part of all Parhyale limbs in embryos, larvae and adults (Fig. 2E,L). We have used this line to monitor limb regeneration following amputation in Parhyale (Fig. 3D,D′).

In emerging model organisms, transposon-based exon trapping is one of the most straightforward ways to isolate genes through unbiased genome-wide screens. The trapped gene of interest can be cloned easily by primer extension on cDNA from the trapped line, even when the genome is unsequenced. To demonstrate this, we cloned a cDNA from the gene trapped in the Distal^DsRed line of Parhyale. The cDNA contains a long open reading frame with no similarity to known proteins. A corresponding probe revealed the same expression pattern as Distal^DsRed in embryos (see Fig. S2 in the supplementary material). Similarly, we have cloned cDNAs of several other genes trapped by PhHsp70a-DsRed in Parhyale and Drosophila. Sequencing of these cDNAs led to our discovery of trans-splicing in Parhyale (Douris et al., 2010).

In exon-trapped genes, splicing of the endogenous transcript onto the gene-trap cassette generates a chimaeric mRNA that expresses, at most, the N-terminal portion of the endogenous protein. If splicing to PhHsp70a-DsRed were 100% efficient, traps would be mutagenic and we would expect to detect a loss-of-function phenotype in animals homozygous for the trap. However, the majority of exon traps obtained using PhHsp70a-DsRed do not cause a detectable phenotype in homozygous animals. We used quantitative RT-PCR to examine splicing to PhHsp70a-DsRed relative to normal splicing at the Distal^DsRed locus. We found that only some transcripts are spliced to the PhHsp70a-DsRed trapping cassette, whereas a large proportion are still spliced onto the endogenous downstream exon. Thus, PhHsp70a-DsRed allows for sensitive trap detection with little disruption of endogenous gene function. This is helpful for maintaining stocks in the absence of balancer chromosomes. iTRAC, described below, provides the means to convert such traps into mutagenic insertions.

The attP site in the trapping vector is a platform for integrating new constructs into trapped loci so as to generate new markers and tools for genetic analysis. ϕC31 integrase has never been used before in Parhyale. First, we devised a rapid PCR-based assay for ϕC31 integrase activity in vivo, which demonstrated efficient integrase-dependent recombination of attP and attB sites across two plasmids injected into early Parhyale embryos (Fig. 4A).

Next, we tested the ability of attB-bearing plasmids to integrate at attP sites inserted in the Parhyale genome in the presence of ϕC31 integrase mRNA. A plasmid carrying attB and a heat-inducible EGFP marker (PhHS-EGFP) integrated with high efficiency into a transgenic line carrying multiple copies of the attP sequence (104 of 207 injected embryos expressed EGFP after heat shock). No integration events were recovered by injection into wild-type embryos (0 out of 255 embryos). Integration of the PhHS-EGFP transgene also occurred in the germline, as judged from its transmission to the next generation.

Finally, we were able to demonstrate the conversion of a DsRed exon trap into one that expresses EGFP in the same pattern. A plasmid carrying attB-PhHsp70a-EGFP was injected with ϕC31 integrase mRNA into early embryos of the Distal^DsRed transgenic line. A high proportion of these embryos (47 out of 90 survivors) showed EGFP fluorescence replacing DsRed fluorescence in at least a subset of limbs (Fig. 4B). In ~20% of injected survivors, iTRAC also occurred in the germline and a stable DsRed to EGFP conversion was evident in subsequent generations (Fig. 4C). In Distal^EGFP lines, DsRed expression could not be detected. These results indicate that iTRAC can be implemented with high efficiency in Parhyale.

Model organisms are defined by the experimental approaches they offer to address biological questions of broad interest. Candidate gene approaches have, until now, been the main avenue for applying knowledge gained in established models to other species of interest, but these approaches are biased and incomplete. The establishment of transgenesis in new species that span the phylogenetic tree of animals, from cnidarians to protostomes and chordates, sets the stage for developing forward and reverse genetics tools and approaches in a wide range of organisms. iTRAC provides a shortcut for implementing these approaches.

The versatility of iTRAC opens opportunities for a wide range of downstream applications: (1) the generation of markers for different types of microscopy, ranging from fluorescence-based live imaging to electron microscopy; (2) the implementation of binary systems for gene expression, such as the GAL4, LexA, tTA and Q systems; (3) the genetic marking and manipulation of clonal populations of cells using FLP or Cre recombinase; (4) the conversion of gene traps into gene knockouts, for instance by insertion of strong transcriptional terminators; (5) genetic cell ablation using cell-autonomously acting toxins; and (6) chromosome engineering. Conceivably, the same collections of traps could be used in the future to implement tools that are not yet realised. Thus, iTRAC may serve as a genetic Swiss army knife, allowing the exploitation of gene traps in a virtually endless number of ways.

The approach and the vectors presented here are likely to be applicable in a broad range of animal models, as all the constituents are known to work in widely divergent species: the Minos transposon is an excellent vector for gene trapping not only in arthropods, but also in mammals and in Ciona (Pavlopoulos et al., 2007; Sasakura et al., 2007); the ϕC31 integrase system has found application in Drosophila, zebrafish, Xenopus, mouse and human (Allen and Weeks, 2005; Groth et al., 2004; Groth et al., 2000; Lister, 2010), and we have shown here that it works very efficiently in Parhyale; the PhHsp70a element can mediate exon trapping in Parhyale and Drosophila and, given the wide conservation of splice acceptor sites, is likely to be useful more broadly.

We thank Yiannis Livadaras and Giorgos Tsoumbekos for help with Drosophila traps, Michele Calos for the ϕC31 integrase plasmids, Mingming Wu and Carsten Wolff for discussions on individual gene traps, and Maura Strigini for critical comments. This work was funded by the EPAN programme of the General Secretariat for Research and Technology (Greece) and by the Marie Curie RTN programme ‘ZOONET’ (European Union).