DNA constructs and transgenic lines:

The structure and components of the piggyBac splice site (PBss) gene trap element and the piggyBac transposase (PBase; including their Drosophila transformation vector counterparts, C4-PBss and PW8-PBase, respectively) are shown in Figure 1. The piggyBac transposon (5′ and 3′ sequences) as well as the heat-shock piggyBac transposase sequences derived from the cabbage looper moth (Tricoplusia ni) were provided by R. A. Harrel (USDA). In the EGFP sequences, which were derived from Hsp70-3X-flu-EGFP-flu and obtained from M. Zecca (Columbia University), the EGFP coding sequence is fused to three copies of an epitope from the influenza hemagglutinin protein (Chalfie et al. 1994; Zecca and Struhl 2002). The PBss construct contains a 3X-flu-EGFP ORF but will be referred to as EGFP throughout this manuscript. The splice acceptor site and engineered branch site were designed on the basis of known splice acceptor sequences in D. melanogaster (Mount et al. 1992) and created as an oligo (MWG Biotech, High Point, NC). Primers corresponding to other portions of the constructs were designed to incorporate appropriate restriction sites for efficient cloning into pBluescript SK+ to create pB-PBss. A KpnI/PvuII fragment from pB-PBss was inserted into Casper4 (where P-element lacZ sequences were removed) cut with KpnI/StuI, creating C4-PBss. An EcoRI/NdeI (T4 polymerase filled) fragment containing the heat-shock piggyBac transposase (hs-PBase) derived from pBdSac (Handler and Harrell 1999) was cloned into EcoRI/SmaI-cut pW8 vector, creating pW8-PBase. Molecular cloning was performed using standard procedures (Sambrook et al. 1989). Transgenic lines were created by co-injecting each construct separately with P-element Δ2/3 helper plasmid. At least four lines for each construct were obtained. None of the C4-PBss transformants expressed EGFP.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.—

Design and function of the piggyBac transposon gene trap system. (A) 5′ and 3′ piggyBac transposon sequences flank an engineered 3′ intronic splice acceptor site fused to an enhanced green fluorescent protein (EGFP) coding region and an α-tubulin 3′-untranslated region containing a poly(A) (pA) site for transcriptional termination. These sequences were integrated into the fly genome by cloning into a Casper4 vector with a mini-white selectable marker (C4-PBss). The driver line (PBase) was created by cloning a heat-shock promoter-piggyBac transposase into a PW8 vector. This was then integrated into the fly genome and selected for using white+ selection. (B) Mobilization of the PBss construct is accomplished by crossing flies carrying this construct with those carrying a heat-shock-driven piggyBac transposase (PBase) and applying heat during germline formation. Reinsertion of a properly oriented PBss sequence into an intron of gene X will result in F₂ heterozygote flies expressing protein X-EGFP or EGFP at the appropriate spatial and temporal parameters dictated by the trapped gene. Flies homozygous for a PBss gene insertion may result in null mutations or EGFP-tagged proteins, allowing the assessment of gene function and/or protein domain localization. EGFP sequences contain an in frame 3X-flu tag sequence at the 5′ end, which adds a 3X-flu tag to the N terminus of the translated EGFP sequence. SA, splice acceptor site; SD, splice donor site.

Mobilization and screening:

For the initial mobilization experiment, C4-PBss flies were crossed to hs-PBase flies and heat-shocked at 37° for 30 min for 3 successive days starting ∼48 hr after egg laying (AEL). Screening was accomplished by taking 10 male progeny from this cross carrying the C4-PBss construct (differences in eye color allowed us to follow the original insertion) and crossing them to ∼50 yw virgin females. Cages were constructed to allow collection of embryos from this cross on apple juice agar plates. These plates were changed every morning and examined periodically during embryonic and larval development under a UV dissecting scope for EGFP-positive progeny. EGFP-positive embryos and larvae were transferred to standard culture vials to complete development followed by crossing to balancer stocks. The segregation of inserts relative to balancer chromosomes was used to map inserts to chromosomes. Homozygous lethal inserts were maintained over balancers.

Subsequent mobilization experiments were used to determine the approximate frequency of gene trapping events. This was accomplished by crossing hs-PBase flies to flies with one of the initial PBss insertions, scarface^PBss, and either heat-shocking larvae as described above or heat-shocking adult male hs-PBase; scarface^PBss flies. For each heat-shock protocol, 10 male flies were crossed to ∼50 yw virgin females. Progeny from these crosses were screened as described above. Numbers of EGFP-positive embryos carrying new EGFP expression patterns were counted over a 5- to 6-day period.

Insertion site identification and analysis:

Insertion sites were identified by extracting DNA from PBss-trapped flies and performing inverse PCR using previously described methods (http://www.fruitfly.org/about/methods/inverse.pcr.html#5′pz) with some modifications. Following extraction, DNA was further purified by phenol/chloroform extraction, precipitation, and resuspension using standard methods (Sambrook et al. 1989). Modification to the inverse PCR method involved digesting 10 μl of DNA with NdeI or XbaI, heat-inactivation of the enzyme, increasing the volume to 400 μl, and ligating overnight at 4°. Ligations were then precipitated and resuspended in 50 μl of 10 mm Tris/Cl, pH 8.0. PCR amplification was accomplished by using 2 μl of ligation reaction, PBss-specific primers (PBssi1 and PBssi2), and Expand Long Template (Roche Applied Science, Indianapolis). Reaction conditions were as follows: 94° for 2 min and 30 cycles of 94° for 15 sec, 55° for 30 sec, 72° for 3.5 min, followed by 72° for 10 min. Single bands were obtained for all of the lines and were sequenced by the Columbia University DNA Sequencing Facility using the same primers. In addition to the PBss insertions listed in Table 1, we identified the insertion site of our starter insertion, C4-PBss, to be at 88D.

To determine the number of PBss insertions in each line, ∼2 μg of DNA derived from each line, including the transposase line, PBase, and the starter transposon, C4-PBss, was digested with either EcoRI and NdeI or SacI and NdeI and probed with a ³²P-labeled SacI/EcoRI EGFP probe derived from the PBss plasmid. Autoradiography was performed using Kodak X-ray film and an X-omat developer (Kodak).

Transcript analysis:

RNA was extracted from three different PBss-trapped lines, scarface^PBss, Akap200^PBss, and Vha100-2^PBss, using Trizol Reagent (Life Technologies, Frederick, MD); reverse transcribed with MMLV reverse transcriptase (New England Biolabs, Beverly, MA); and amplified using Taq Polymerase (Roche Applied Science) with gene- and PBss-specific primers (sequences available upon request). Sequencing was performed by the Columbia University DNA Sequencing Facility using the same primers.

Immunolocalization experiments:

Embryos and discs used for antibody staining were prepared using standard methods. Primary antibodies used included: mouse anti-Wg, guinea pig Anti-Hth, mouse anti-Elav, and Texas red-labeled anti-HRP. Secondary antibodies used included: Texas red-conjugated goat anti-mouse and CY5-conjugated goat anti-guinea pig antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA).

Microscopy:

Embryos and discs stained with antibodies were visualized and photographed on a Bio-Rad confocal imager using Bio-Rad 2000 software (Bio-Rad Laboratories, Hercules, CA).

Design and construction of the PBss gene trap vector:

The PBss construct is a small mobile element created for the purpose of trapping a gene's transcriptional expression profile by reporting EGFP, which becomes fused to the gene's native transcript by mRNA splicing (Figure 1). The piggyBac transposon is derived from the cabbage looper moth, Tricoplusia ni. This element was chosen because of its loose DNA target sequence preference, TTAA (Cary et al. 1989; O'Brochta and Atkinson 1996; Handler and Harrell 1999; Lobo et al. 1999; Li et al. 2001). Similar to P elements, piggyBac requires transposon-derived sequences, designated as 5′ and 3′, at each end for transposition to occur. A piggyBac-derived transposase (PBase) catalyzes the transposition of elements containing these sequences (Figure 1A). Between the piggyBac 5′ and 3′ sequences, a consensus splice site was created upstream of an EGFP gene followed by an α-tubulin 3′-untranslated region (pA), for transcriptional termination (Figure 1A). An ATG initiation codon for the EGFP gene was placed in the first translated frame since the majority of known introns in Drosophila end in this frame (Long and Deutsch 1999; Long et al. 2003). The piggyBac transposase element is composed of a heat-shock (hs) promoter fused to the piggyBac transposase coding sequence (hs-PBase; Figure 1A; Handler and Harrell 1999).

Mobilization, screening, and identification of transgenic flies:

To create transgenic flies carrying these constructs, PBss was cloned into the Casper4 vector (C4-PBss) and PBase was cloned into the PW8 vector (PW8-Pbase; Figure 1A) to allow white+ selection of transformed progeny. All of the w⁺ C4-PBss transformants failed to express EGFP. Once multiple stable lines were established for each construct, they were crossed to each other and heat-shocked three times at 37° for 30 min for 3 successive days after larval development had begun (∼24 hr AEL; Figure 1B and see materials and methods for details). EGFP-positive embryos and larvae were isolated with the use of a UV light source and a dissecting microscope. From this initial screen we isolated 25 EGFP-expressing lines from ∼24,000 embryos and larvae. EGFP-expressing lines were balanced and a subset of these lines was used for subsequent analysis (Table 1 and see below).

To determine if PBss insertions can be remobilized, two additional crosses were performed (see materials and methods). One experiment involved mobilizing PBss during larval development from a previously trapped gene, CG11066^PBss, which we have named scarface^PBss (see below). A second screen involved mobilizing PBss in adult male flies carrying the same scarface^PBss insertion. In both screens, only embryos were screened for new EGFP+ patterns, thus limiting the number of new insertions obtained. Nevertheless, both experiments generated novel EGFP+ insertions at similar frequencies, indicating that PBss elements can be remobilized during both larval and adult stages. The combined number of new EGFP+ patterns identified in these two experiments was 4 of ∼14,000 embryos. Since scarface^PBss represents a single insertion as determined by Southern analysis (see materials and methods), we are confident that new expression patterns represent remobilization events. Many of these stocks have been maintained in the presence of the hs-PBase element without a loss of EGFP expression, suggesting that PBss elements transpose rarely or not at all in the presence of hs-PBase, as long as no heat shock is provided (see Figure 1).

Insertion site identification:

Southern blot analysis was used to confirm that all EGFP+ lines contain only a single PBss element (data not shown). BLAST analysis (Altschul et al. 1990) of sequenced inverse PCR products was used to determine the sites of PBss insertion within a subset of the trapped lines. Table 1 lists information on the trapped genes. Analysis of the insertion sites has confirmed that PBss always inserts into a TTAA sequence, with no other obvious sequence preferences (Table 1). This sequence preference for piggyBac elements is consistent with previous observations and supports the idea that PBss will be able to insert widely throughout the genome, with minimal preferences for specific sequences (Cary et al. 1989; Elick et al. 1996; Handler and Harrell 1999).

Examination of the insertion sites listed in Table 1 illustrates that PBss transposed to all chromosome arms and, as expected, usually to a predicted intron. In addition, most of the insertions are lethal as homozygotes. One initially surprising finding from this analysis was the viability of an insertion into the first intron of the Act5c gene (Table 1). However, the Act5c gene has a second promoter that is downstream of the PBss insertion site (Bond and Davidson 1986; Chung and Keller 1990). The presence of an alternative promoter downstream of the PBss element can account for the viability of this insertion as a homozygote. These results suggest that transcripts initiating from the more distal promoter of the Act5c gene are not required for viability.

mRNA splicing utilizes the PBss splice acceptor:

To determine the ability of the PBss element to splice as designed we used RT-PCR to analyze the mRNA transcripts from three of the trapped lines (Figure 2; Table 1). Results from this analysis revealed that the engineered splice acceptor site upstream of EGFP in PBss was utilized in all three cases (Figure 2). In two of these three examples, the PBss element inserted into an intron and the splice donor site was derived from an upstream exon as anticipated (Figure 2, A and B). In one case PBss inserted into an exon of the gene Vha100-2 (Figure 2C). Although the engineered splice acceptor site was used in this case, the splice donor site was a cryptic site present within the 3′ piggyBac sequences (Figure 2C). The insertion into Vha100-2 illustrates that insertions into exons can, in some instances, also lead to a productive gene trap.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.—

Transcript and predicted protein fusions for (A) scarface^PBss, (B) Hsc70-4^PBss, and (C) Vha100-2^PBss. RT-PCR transcript analyses were performed for all three trapped genes revealing two expected [scarface^PBss (A) and Hsc70-4^PBss (B)] and one unexpected [Vha100-2^PBss (C)] mRNA fusion product with the EGFP coding region of PBss. EGFP sequences contain an in frame 3X-flu tag sequence at the 5′ end, which adds a 3X-flu tag to the N terminus of the translated EGFP sequence. SA, splice acceptor site; SD, splice donor site.

Expressional analysis:

Some examples of the EGFP expression patterns obtained from these screens are shown in Figure 3. Close inspection of embryos from some PBss-trapped lines revealed that EGFP was localized to specific cellular compartments (Figure 3, A, D, and F), as opposed to a fairly uniform cellular distribution, which is expected for EGFP alone. This suggests that in these cases EGFP is fused to an endogenous protein with specific localization properties. Sequencing of the insertion sites has revealed that this is in fact the case for some lines (Table 1 and see below).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.—

Embryonic EGFP expression patterns of six representative PBss-tagged genes. (A) TMII^PBss, tropomyosin II protein. (A_i) A close-up of the anterior region of a TMII^PBss embryo. (B) Act5C^PBss, actin5c protein. (B_i) A close-up of the anterior region of an Act5C^PBss embryo. (C) l(2)01424^PBss, a putative translational initiation factor. (D) scarface^PBss (CG11066), a putative trypsin-like serine protease. (E) Vha100-2^PBss, a V-type ATPase. (F) Akap200^PBss, a kinase anchor protein.

Two lines in particular, Akap200^PBss and scarface^PBss, exhibit interesting subcellular localization properties at high magnification (Figures 4 and 5, respectively). The Akap200 gene encodes an A kinase accessory protein shown to interact with signaling pathway components. The protein is predicted to be membrane bound on the basis of the presence of a myristoylation sequence in its N terminus (Li et al. 1999; Rossi et al. 1999). Previous work has shown that the Akap200 protein is localized at the cell surface in developing germline cells (Jackson and Berg 2002). Analysis of the EGFP expression in developing germline cells of Akap200^PBss/+ animals revealed strong expression in the germarium of the developing oocytes as well as in the nervous system (Figures 3 and 4). Closer inspection of EGFP expression in the developing nervous system of Akap200^PBss/+ embryos suggests that it is localized to the membrane of developing axons of the ventral nerve cord as judged by costaining with a pan-axonal marker, anti-HRP (Figure 4). These results suggest a putative role for this Akap200 in participating in signaling in developing or mature axons, a previously unknown function for this protein.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.—

Immunofluorescence staining of Akap200^PBss embryos, germarium cells, and third instar larval wing and leg discs. Akap200^PBss EGFP-expressing embryos were stained with HRP (red) to label axons and Elav (blue), a pan-neural nuclear marker, to label neurons. (A) Ventral view of Akap200^PBss embryos. (B) Side view of Akap200^PBss embryos. (C) High magnification view of the developing ventral nerve cord. (C_i) Akap200^PBss EGFP alone. (C_ii) HRP alone. (C_iii) Elav alone. (D) Akap200^PBss germline cells exhibiting EGFP expression in germarium cells. (E) Third instar wing and leg discs expressing Akap200^PBss EGFP and stained with an extracellular marker, Wg (red), and a nuclear marker, Hth (blue).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.—

Immunofluorescence staining of scarface^PBss third instar larval wing discs with markers labeling nuclear (Homothorax, Hth) and extracellular (Wingless, Wg) compartments. (A) Wing, haltere, and leg discs. (B) Close-up of notum section of wing disc. (C) High magnification view of anterior section of wing disc. (C_i) scarface^PBss EGFP alone. (C_ii) Wg alone. (C_iii) Hth alone.

Description of the PBss insertion into CG11066/scarface:

On the basis of the phenotype of homozygous escapers of a PBss insertion into CG11066, we have named this gene scarface (see below). Here, we describe a developmental expression profile and preliminary analysis of the mutant. On the basis of BLAST comparisons (Altschul et al. 1990), scarface is most similar to trypsin-like serine proteases that are secreted or are associated with the outer membrane of cells. The scarface^PBss insertion was initially identified due to a strong larval EGFP expression pattern in the anterior and posterior termini of third instar larvae. Subsequent analysis showed EGFP expression in many tissues, including wing imaginal discs (Figure 5). Costaining with extracellular and nuclear markers, Wg and Hth, respectively, in the wing discs suggests that scarface may be membrane associated or secreted (Figure 5). scarface^PBss/+ adult flies exhibit EGFP expression in the proboscis and in the socket cells of all micro- and macrochaete analyzed (Figure 6, C–F, and data not shown). Finally, most scarface^PBss homozygotes are pupal lethal. However, some homozygote escapers occasionally eclose. These flies exhibit what appears to be “scarring” around the mouthparts that is indicative of necrotic tissue (Figure 6, G and H).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.—

EGFP expression and mutant phenotype of larval and adult scarface^PBss flies. (A) Anterior side view of a heterozygote scarface^PBss/+ larva. (B) Posterior side view of a heterozygote scarface^PBss/+ larva. (C) Ventral view of a heterozygote scarface^PBss/+ fly expressing EGFP in the adult proboscis. (D) Side view of a heterozygote scarface^PBss fly. (E) Close-up of an adult wing hinge showing expression in patches of specific cells in the hinge. (G) Close-up of an adult scarface^PBss leg indicating expression in the socket cells of the macrochaete. (G and H) Brightfield and EGFP expression images of an adult homozygote scarface^PBss/scarface^PBss fly exhibiting black, visibly necrotic tissue near the proboscis (arrow). (G_i and G_iii) Brightfield. (G_ii and G_iv) EGFP expression of the fly in G_i and G_iii, respectively. (H) Brightfield ventral view close-up of scarface^PBss head with arrow indicating necrotic tissue close to proboscis (arrow).

PBss design:

The PBss element was designed to trap a gene's expression pattern using the gene's normal transcriptional and splicing mechanisms. Because PBss has no promoter to initiate EGFP transcription, productive (EGFP+) insertions must occur downstream of an endogenous gene's promoter. A second requirement is that PBss must insert in such a manner to allow the in frame translation of EGFP. All of the insertions characterized here fulfill both of these requirements. As expected, most of these insertions were in introns. In these examples, expression of EGFP depended on mRNA splicing between an upstream exon and the engineered splice site 5′ to EGFP. For those insertions that occurred downstream of a coding exon, EGFP was expressed as a fusion protein. In contrast, for those insertions that occurred downstream of a noncoding exon, translation of EGFP was initiated from its own ATG. In one unusual case, we found an insertion into an exon. In this case, splicing still occurred and EGFP was expressed as a fusion protein. This example illustrates that some PBss insertions into exons can also be productive.

We also engineered a transcriptional stop sequence downstream of the EGFP sequences to terminate transcription of trapped genes. The reasoning behind this aspect of the design was the hope that most productive insertions would also cause a mutation, allowing a subsequent genetic analysis of the gene to be carried out. On the basis of the frequency of lethal and mutant insertions, it is likely that transcription termination is occurring due to the PBss insertion. This aspect of the PBss design contrasts with other gene trap vectors that have splice sites on both sides of the reporter gene, allowing the incorporation of the reporter sequences into an otherwise intact transcript (Zambrowicz et al. 1998; Lukacsovich et al. 2001; Morin et al. 2001). In the latter case, a mutation of the endogenous gene is much less likely. Thus, in these systems subsequent genetic analysis would require a second step to delete the element and neighboring chromosomal DNA to generate a mutation.

Finally, the PBss vector described here uses a piggyBac-based transposition system. The reason for choosing this system was to allow for a broad and unbiased set of insertion sites within the Drosophila genome. Consistent with previous analysis of piggyBac transposition preferences, all of the insertions identified here inserted into a TTAA sequence (Table 1; Handler and Harrell 1999). Since the D. melanogaster genome is very AT rich, the chances that a TTAA sequence will be found within most introns is high. Although it is possible that piggyBac elements may also exhibit insertional hotspots, our preliminary findings suggest that this system will provide a larger and less-biased set of insertion sites than that provided by P-element-based vectors (Spradling et al. 1995).

The design of the PBss vector contrasts with another recently described gene trap vector, pGT1 (Lukacsovich et al. 2001). This construct is similar to ours in its ability to splice to upstream exons. However, it reports expression through a fusion protein with the yeast transcriptional activator, Gal4, whereas ours reports expression via EGFP. The choice of EGFP allows PBss insertions to be identified in live animals without any UAS-reporter construct. In addition, we have shown that some insertions generate fusion proteins that are localized to distinct cellular compartments. This feature of the PBss system can potentially provide information about the endogenous protein's subcellular localization. This observation also has implications for Gal4-based vectors. Because Gal4 must be nuclear to be functional, some Gal4 fusion proteins may not be active if they are localized to nonnuclear compartments.

PBss limitations and improvements:

On the basis of these results we suggest that, in its current state, the PBss element is a valuable functional genomics tool. However, our analysis also suggests some limitations and potential improvements that can be implemented in the future. One potential drawback of the current element is the apparently low frequency of transposition events measured in the remobilization experiments. In the three miniscreens described here we obtained a productive transposition frequency that ranged from ∼0.3 to ∼1 event per 1000 animals examined. This frequency in part reflects the fact that productive (i.e., EGFP-expressing) insertions are inherently low-frequency events. For a productive insertion to occur, the PBss element must insert in the correct orientation downstream of either a noncoding exon (and thus utilize the in frame ATG 5′ to EGFP) or a coding exon whose frame matches the EGFP frame (in this case, +1). Other nonproductive insertions would have been undetected in our screens. We emphasize, however, that despite the low frequency, identifying EGFP-positive embryos or larvae is very easy. This is especially true for the system described here because the starting elements (C4-PBss) do not express EGFP. Thus, following mobilization, any EGFP-expressing animal stands out and is easy to identify. Given this absence of background, the process of identifying productive insertions can also be automated using whole-embryo sorters, thus allowing the quick isolation of thousands of EGFP-expressing lines (Furlong et al. 2001).

Our observations also suggest some improvements that can be introduced to the PBss system. We were surprised to observe a cryptic splice donor site in the 3′ piggyBac sequences present in PBss. In the example analyzed here, the utilization of this splice site may have been required for EGFP to be translated in frame. However, there may be some cases when the utilization of this splice site could interfere with EGFP expression. To prevent this splice from occurring, it may be possible to mutate the essential “GT” dinucleotide at this cryptic splice donor site.

It is likely that the timing and/or length of heat shocks to induce expression of the piggyBac transposase can also be improved over the initial conditions tried here. It is possible that the PBss insertion we used as the starting point for two of the miniscreens (scarface^PBss) may, for some reason, be less able to excise than other insertions. A survey of additional PBss insertions may reveal a more prolific starting point for initiating new transposition events. The PBss element used here has the EGFP sequences in the +1 translational frame (relative to the engineered splice acceptor site). Thus, another potential improvement would be to simultaneously use three PBss elements that have the EGFP sequences in each of the three translational frames. Finally, it may be possible to include an IRES-Gal4 cassette downstream of EGFP. This would allow both the tagging of a gene and the subsequent ability to misexpress other gene products in the same pattern under Gal4 control. In preliminary experiments, however, we have been unable to find an IRES that functions in an unbiased manner in Drosophila embryos and imaginal discs (data not shown). Thus, additional surveys of IRES sequences must be carried out before this feature can be added to the PBss system.

We emphasize, however, that the PBss system as it stands is able to both tag and mutate endogenous Drosophila genes as designed. Thus, PBss is a functional genomics tool that is ready to be used by the Drosophila community.