RESULTS AND DISCUSSION

Insertional mutagenesis and enhancer detection for nonmodel insects: To develop insertional mutagenesis systems for nonmodel arthropods, we employed the widely applicable transposable elements Hermes, Minos, Mos1, and piggyBac. Due to the absence of marked “balancer” chromosomes in most nonmodel arthropods, several reliable and distinguishable transformation markers are necessary when insertional mutagenesis and enhancer detection screens are combined with binary ectopic expression systems. To clearly identify, separate, and stably establish novel mutator insertion lines without the need of balancers, we have used the three independent and distinguishable fluorescent transformation markers 3xP3-ECFP, 3xP3-EYFP, and 3xP3-DsRed (Hornet al. 2002) to mark (i) the jumpstarter, (ii) the mutator, and (iii) the reporter strains of enhancer detection expression systems (Figures 1 and 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

—Crossing scheme for the identification of new mutator insertions and adult enhancer activities without the use of balancer chromosomes (see text for details). The use of distinguishable fluorescent transformation markers makes it possible to follow the different jumpstarter (3xP3-ECFP), mutator (3xP3-EYFP), and reporter (3xP3-DsRed) elements independently. To demonstrate the presence of the different elements, fly heads were photographed without filter and with ECFP-, EYFP-, or DsRed-specific filters (Hornet al. 2002). From G₂ males carrying the mutator on the X chromosome and a jumpstarter on an autosome, all male progeny showing EYFP fluorescence must carry new mutator insertions on autosomes. Some of them also detect adult enhancer activities, like the male showing segmental fluorescence in the abdomen. Note that it is not important for the scheme on what autosome the jumpstarter or reporter are inserted. For the integrated mutator-reporter {GAL4Δ+UAS}, G₂ males were crossed to white instead of reporter strain virgins.

After crossing jumpstarter and mutator strains (Figure 1, G₁), insects carrying both elements can be identified on the basis of the eye-specific ECFP and EYFP expression (Figure 1, G₂). In the next generation, the 3xP3-ECFP-marked jumpstarter can be crossed out to allow stable inheritance of a novel 3xP3-EYFP-marked mutator insertion. At the same time a 3xP3-DsRed-marked reporter can be crossed in to detect adult enhancer activities that mediate expression of a heterologous transactivator encoded by the mutator (Figure 1, G₃). When both mutator and reporter are based on the same transposon, the 3xP3-DsRed-marked reporter can be crossed out again to allow for molecular analysis of the novel insertion site. Since each construct can be followed independently, there is no need for balancer chromosomes. Moreover, the dominant fluorescent marker serves as a visible label for the novel insertions in both larval and adult stages and therefore facilitates stock keeping. Males and females carrying a novel insertion can be mated and their progeny analyzed for recessive phenotypes. Furthermore, the transposon insertion molecularly tags the mutated gene, which assists in its cloning.

Jumpstarter elements: Jumpstarter strains provide transposase activity to mobilize nonautonomous mutator elements. The jumpstarter constructs therefore contain an active transposase gene. However, to keep jumpstarter strains stable, the transposable element backbone that is used to introduce the active transposase gene into the genome should be derived from a different transposon family, so that cross-mobilization can be excluded. To generate a jumpstarter for mutators based on the hAT element Hermes, we chose the TTAA element piggyBac, and to create jumpstarters for piggyBac-based mutators, we used the hAT element Hermes or the Tc1/ mariner elements Mos1 and Minos. To drive expression of the transposase gene either the inducible Drosophila hsp70 promoter (Liset al. 1983) or the constitutive Drosophila α1-tubulin promoter (Theurkaufet al. 1986) was used (Table 1).

Of the different jumpstarter constructs generated (see materials and methods), the heat-shock-inducible Hermes jumpstarter Bac{3xP3-ECFP, hsp70-Hermes} and piggyBac jumpstarter Mi{3xP3-DsRed, hsp70-piggyBac}, as well as the constitutive piggyBac jumpstarters Her{3xP3-ECFP, αtub-piggyBacK10} and Mos{3xP3-ECFP, αtub-piggyBacK10} were tested, respectively, for remobilization of genome-integrated, nonautonomous Hermes or piggyBac elements (see materials and methods). In this assay, all jumpstarter constructs proved functional (data not shown). Since stable and strongly expressed jumpstarter strains can be preselected before starting insertional mutagenesis screens, we chose to mark most of the jumpstarter constructs with the less-sensitive marker 3xP3-ECFP (Horn and Wimmer 2000).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

—Molecular structure of non-species-specific constructs for insertional mutagenesis and enhancer detection used in the Drosophila pilot screen. The piggyBac-jumpstarter element was integrated with a Hermes vector, whereas mutator and reporter elements used in this study were based on piggyBac vectors. For details on the different constructs see Table 1 and text.

At what phylogenetic distance the Drosophila α1-tubulin promoter will stimulate sufficient transposase expression to mobilize mutator elements is difficult to estimate. However, the Drosophila hsp70 promoter-based jumpstarters are also likely to work in non-Dipteran insects, since the hsp70 promoter has been shown to mediate heat-shock-inducible gene expression in the lepidopteran silkmoth, B. mori (Uhlířovaet al. 2002). Since 3xP3-ECFP is a poor marker for Tribolium castaneum transgenesis (M. Klingler, personal communication) and since this might also hold true for other insect species, Mi{3xP3-DsRed, hsp70-piggyBac} was marked with the more easily selectable marker 3xP3-DsRed (Hornet al. 2002). Moreover, this jumpstarter was specifically designed to contain the hsp70 promoter toward the end of the transposon, which should allow for enhancer trap effects at different insertion sites. Therefore, even if the heat-shock promoter should be nonfunctional, the basal promoter region should suffice to obtain some functional jumpstarter strains based on suitable germline enhancers close to the insertion. It is interesting to note that even without heat-shock treatment, most Dm[Mi {3xP3-DsRed, hsp70-piggyBac}] strains mobilized nonaucating that transposase activities are heat-shock independent and probably enhancer driven.

Mutator elements: To allow for effective insertional coverage in transposon mutagenesis screens, we marked the mutator elements with 3xP3-EYFP, which represents a highly sensitive transformation marker (Hornet al. 2000; Horn and Wimmer 2000). When transposon mutagenesis is used in vivo to identify genes expressed in certain larval or adult tissues, novel insertions and enhancer activities should be detected simultaneously and noninvasively. To avoid frequent switching between different filter sets while screening, we also employed EYFP as a reporter gene either within the mutator itself or within separate reporter elements (see below). This excludes, however, in vivo enhancer detection in tissues such as the eyes and central nervous system and other tissues in which the artificial 3xP3 promoter drives marker gene expression (Hornet al. 2000).

To allow enhancer detectors to be directly employed for misexpression studies, we included genes in the mutators that encode heterologous transactivators (Brand and Perrimon 1993; Belloet al. 1998). These primary reporter genes are controlled by the weak minimal promoter of the P-element transposase gene (Rørth 1998), which is transcriptionally silent unless activated by an enhancer element (O'Kane and Gehring 1987). The mutator elements are constructed with the basal promoter toward the end of the transposon (Figure 2). Genomic integration of the mutator near a tissue-specific enhancer will then allow the transactivator to be expressed in patterns similar to that of the gene normally under control of the detected enhancer.

In contrast to Brand and Perrimon (1993), we did not use the full-length yeast GAL4 gene when constructing GAL4-based mutators, but employed the coding region of the GAL4 deletion variant II-9 (Ma and Ptashne 1987). This deletion variant (GAL4Δ) contains both the amino-terminal DNA-binding domain and the carboxy-terminal activation domain of GAL4. GAL4Δ is almost as good an activator as GAL4 itself but supposedly more stable (G. Struhl, personal communication). GAL4Δ-based mutators were generated in both piggyBac and Hermes backgrounds (see materials and methods), but only Bac{3xP3-EYFP, p-GAL4Δ-K10}, hereafter abbreviated as {GAL4Δ}, was used in this study.

For in vivo identification of an enhancer's activity, GAL4Δ needs to drive the expression of a visibly detectable secondary reporter such as EYFP, whose coding region must therefore be placed under the control of GAL4-binding sites (referred to as UAS, for upstream activation sequence; Brand and Perrimon 1993). This can either be provided by a separate reporter element (Figure 1 and see below) or somehow be integrated into the mutator element itself. To employ EYFP as both a marker for novel insertions and as a reporter for enhancer detection, we placed an additional UASp promoter (Rørth 1998) upstream of the 3xP3 promoter to generate Bac{UASp-3xP3-EYFP, p-GAL4Δ-K10}, hereafter referred to as {GAL4Δ+UAS}. Effective adult enhancer detection by this mutator could be observed when establishing transgenic strains, since many of them detected different enhancer activities (Figure 3). The high efficiency of visibly detecting enhancer activities with {GAL4Δ+UAS} might be due to the fact that in this construct the basal P-element promoter is actually placed at both ends of the mutator. This allows the genomic integration region to be scanned for enhancers with differently oriented basal promoters. In addition, the 3xP3 promoter (Hornet al. 2000) also occasionally serves as an enhancer detector. However, only the enhancer acting on the promoter driving GAL4Δ will serve as an expression tool for effector constructs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

—Examples of adult enhancer activities isolated with the mutator-reporter element {GAL4Δ+UAS}. (A and B) Strain MM12.III detects an enhancer activity for the distal and medial part of the proboscis, the thorax, and the legs, (C) strain MM12.II for the abdomen and a specific pattern in the legs, as well as (D) strain MM12.II_2 for the basal part of the proboscis and the maxillary palps. Note that in this particular insertion strain the eye-specific marker expression is suppressed.

At rare insertion sites of {GAL4Δ+UAS}, the 3xP3-mediated eye expression of EYFP can actually be suppressed while another enhancer drives EYFP expression (Figure 3D). Should no defined sex-chromosomal insertions (Figure 1 and see below) be available in nonmodel organisms, such partially suppressed insertions could actually serve as ideal launching pads for insertional mutagenesis. After remobilization from such an insertion site, the restoration of eye-specific EYFP expression indicates that the mutator element has moved its position and thus novel insertions can be isolated. Similarly, the loss or change of a particular enhancer activity at the launching pad insertion could also serve this purpose.

We employed the bacterial-viral fusion tTA (Gossen and Bujard 1992), which has successfully been used to generate tetracycline-controlled binary expression systems in Drosophila (Belloet al. 1998; Thomaset al. 2000; Heinrich and Scott 2000; Horn and Wimmer 2003) as another transcriptional activator. The major advantage of this system is that targeted gene expression can be additionally controlled by a food supplement. The primary spatiotemporal control is provided by the detected enhancer as in the GAL4/UAS system, but here a secondary temporal control due to tetracycline-dependent inactivation of tTA makes it possible to switch off the system. Tetracycline and tTA form a complex, which prevents tTA from binding to its response element, which therefore becomes inactive (Gossen and Bujard 1992). The tTA-based mutator used in this study, Bac{3xP3-EYFP, p-tTA-K10}, is hereafter abbreviated as {tTA}.

As in the case of {GAL4Δ+UAS}, an integrated mutator-enhancer-detector for the binary tTA expression system was generated by placing a TRE promoter upstream of the 3xP3 promoter. However, when strains were generated with this Bac{TRE-3xP3-EYFP, p-tTA-K10} mutator (abbreviated {tTA+TRE}), enhancer activities were not detected. When driven by defined tTA driver constructs, {tTA+TRE}-mediated EYFP expression was actually enhanced compared to simple TRE-EYFP reporter constructs (data not shown). This suggests a positive feedback loop between the TRE sites and the basal promoter driving tTA expression and indicates that both TRE and tTA are functional in this construct. The lack of enhancer detection suggests a lower sensitivity of the tTA/ TRE system compared to GAL4/UASp. {tTA+TRE} was therefore not used further in this study.

For the mutators {GAL4Δ}, {GAL4Δ+UAS}, and {tTA}, several X chromosomal insertions were obtained either directly after injection or after remobilization from autosomal insertion sites. For each mutator, three independent remobilizable and homozygous viable X chromosomal insertion strains were used in the Drosophila pilot screen (see below). During these pretests one X chromosomal {tTA} insertion was identified that could not be efficiently remobilized. The respective strain was not further analyzed but excluded from this study.

Reporter elements: To easily detect enhancer activities by {GAL4Δ} or {tTA} insertions, the transactivator needs to drive the expression of a secondary reporter, e.g., EYFP for in vivo analysis or the bacterial lacZ gene for in situ analysis (O'Kane 1998). For this purpose, we generated separate reporter elements by placing these secondary reporter genes under the control of the UASp promoter (Rørth 1998) or the TRE promoter (Gossen and Bujard 1992). Of the different 3xP3-ECFP- and 3xP3-EYFP-marked reporters generated, Bac{3xP3-EYFP, UASp-EYFP-K10}, Her{3xP3-ECFP, UASp-EYFP-K10}, Bac {3xP3-ECFP, UASp-lacZ-K10}, and Her{3xP3-ECFP, UASp-lacZ-K10} were injected and the resulting strains were successfully tested (Table 1).

When testing strains for the reporter Her{3xP3-ECFP, UASp-EYFP-K10} by using defined GAL4 driver strains, we noted that not only EYFP was expressed as expected, but also ECFP (data not shown). This indicates that UAS-bound GAL4 can also activate transcription at the 3xP3 promoter, despite the respective downstream and distant location (see Figure 2 for composition of UAS reporters). In contrast, strains carrying the reporter Her{3xP3-ECFP, UASp-lacZ-K10} did not mediate GAL4-driven ECFP expression, suggesting that the larger distance between the UASp sites and the 3xP3 promoter (due to the longer lacZ gene; Figure 2) prevents activation of the 3xP3 promoter by UASp-bound GAL4. In these constructs, 3xP3 promoter activity seems to drive only the fluorescent marker placed directly downstream, which suggests that 3xP3 represents a proximal promoter element that cannot function as an enhancer element at greater distances.

Nonetheless, these results indicate that in the mutator element {GAL4Δ+UAS}, positive feedback loops can occur, which could allow strong and enduring enhancer detection after the loop has first been initiated. {GAL4Δ+ UAS} could thus serve as an “enhancing reporter” that would allow (i) a more sensitive detection of enhancers and (ii) the detection of enhancers active at earlier stages. The enhancer activity might actually have already stopped, but it is still detectable due to the positive feedback loop of the reporter. Hassan et al. (2000) have previously described the use of two independent constructs, an enhancer-GAL4 and a UAS-GAL4 construct, to create a positive feedback loop for fate mapping and signal amplification. {GAL4Δ+UAS} differs in that it combines enhancer detection and signal amplification within one single construct. The positive feedback can be an advantage, since it allows screening for genes active at early stages of organogenesis during late developmental stages of that same organ. However, for tissues that are sensitive to high levels of GAL4Δ expression, this enhancing reporter might be detrimental and respective enhancers will not be detectable due to toxicity.

Strains with the 3xP3-DsRed-marked reporters Bac {3xP3-DsRed, UASp-EYFP-K10}, Bac{3xP3-DsRed, UASp-lacZ-K10}, Bac{3xP3-DsRed, TRE-EYFP-SV40}, and Bac{3xP3-DsRed, TRE-lacZ-SV40} have been further used in this study (Figure 1 and 2). When we used EYFP as the reporter gene (Figure 1), adult enhancer activities could be detected noninvasively while screening for novel 3xP3-EYFP-marked insertions without the need for switching fluorescence filter sets. This allowed straightforward screening of enhancer activities for specific tissues without the need of establishing individual insertion strains beforehand. Nevertheless, the maturation time for internal cyclization and oxidation causes a delay of several hours before EYFP fluorescence can be detected (Daviset al. 1995). This usually presents no problem when screening for larval or adult enhancer activities, but when we were interested in embryonic enhancer activities, we applied the more sensitive and faster-responding reporter gene lacZ. This, however, required the establishment of individual insertion strains first, since the embryos do not survive the necessary fixation and staining procedures.

Pilot screen in D. melanogaster: To test the balancer-free insertional mutagenesis scheme for enhancer detection (Figure 1), we performed a small pilot screen in D. melanogaster. We decided against Mos1-based mutators, since it was shown that genomic insertions of transgenic Mos1 constructs can rarely be remobilized in Drosophila (Lozovskyet al. 2002). We decided against Hermes-based mutators, despite the fact that remobilization actually works well in Drosophila (data not shown), only because Hermes insertions can be remobilized by hobo elements (Sundararajanet al. 1999). Since not all Drosophila stocks are free of hobo (Blackmanet al. 1989), novel Hermes insertions might become unstable in future analyses. For the application of Hermes-based insertional mutagenesis in Drosophila, special care should be taken to use hobo-free strains only.

In our Drosophila pilot screen, we chose the 3xP3-EYFP-marked piggyBac-based mutators {GAL4Δ}, {GAL4Δ+UAS}, and {tTA}, for which we used three homozygous X chromosomal launching integrations each. To remobilize these mutators from the X chromosomes, we selected three different strains carrying the homozygous constitutive jumpstarter pHer{3xP3-ECFP, αtub-piggyBacK10} on the second or third chromosome. This allowed us to dispense with heat-shock protocols. In addition, to detect potential enhancer activities by novel {GAL4Δ} and {tTA} insertions, we employed homozygous strains carrying, on the second or third chromosome, the 3xP3-DsRed-marked reporters Bac{3xP3-DsRed, UASp-EYFP-K10}, Bac{3xP3-DsRed, UASp-lacZ-K10}, or Bac{3xP3-DsRed, TRE-EYFP-SV40}, respectively (Figure 2).

The different fly strains were crossed as depicted in Figure 1: For each mutator, nine different mutator-jumpstarter-strain combinations were set up (Figure 1, G₁ cross; see also legend to Table 2). To maximize identification of independent transposition events for each combination, 27 single males carrying both jumpstarter and mutator were crossed to virgins of the respective EYFP-reporter strains Dm[Bac{3xP3-DsRed, UASp-EYFP-K10}]orDm[Bac{3xP3-DsRed, TRE-EYFP-SV40}] (Figure 1, G₂ cross) or a white strain in the case of {GAL4Δ+ UAS}. All male G₃ progeny that show the mutator marker 3xP3-EYFP must carry novel autosomal insertions, since the originally X chromosomal launching insertion is not paternally inherited by males. This scheme, therefore, allows for straightforward identification of novel insertion events, which in some cases also led to the detection of adult enhancer activities that could be identified concurrently (Figure 1, G₃).

For each mutator, the results of nine combinations of jumpstarter and mutator strains are presented together, since no significantly different performance rates were observed between the individual combinations (Table 2): Simple excision events observed in the female progeny (loss of 3xP3-EYFP marker) were between 48 and 100%. The observed transposition frequency was high with an average >10%, thus indicating that piggyBac remobilization was highly effective. In ∼80% of the single G₂ crosses, novel autosomal insertions could be identified. This represents an average jumping rate for piggyBac mutators that lies within the best rates observed for P-element constructs (Berg and Spradling 1991). Due to concomitant jumpstarter presence in some of the mutator-carrying G₃ males, not all of the 80% could be used to establish stable new insertion lines. However, novel insertions could be stably established from ∼50% of the single G₂ crosses. This suggests that piggyBac can serve as an excellent insertional mutagenesis agent in D. melanogaster and might actually outperform P elements.

Adult and larval enhancer detection: In ∼50% of the crosses with the GAL4-based mutators {GAL4Δ} and {GAL4Δ+UAS}, adult enhancer activities could be detected (Table 2). These enhancers activated gene expression in diverse body parts. Head-specific, thorax-specific, abdominal-specific, or leg-specific enhancers were observed (Figure 3). This indicates that EYFP reporters can be used efficiently to noninvasively detect adult enhancer activities; thus, genes expressed specifically in adult organs can easily be screened for without having to establish numerous insertion lines beforehand. This will be of importance for nonmodel organisms, since keeping a substantial number of independent insertion lines might not be manageable for arthropod species that are more complicated than Drosophila to rear.

{GAL4Δ+UAS} shows slightly reduced jumping rates and transposition frequencies (Table 2). This might be due to some slight toxicity when positive feedback loops cause high GAL4Δ expression. Since this integrated {GAL4Δ+UAS} mutator-reporter allows enhancer detection without crossing to separate reporter strains, we also investigated the established {GAL4Δ+UAS} insertion lines for larval enhancer activities. Figure 4 shows that various tracheal, fat body, or muscle enhancers were detected, indicating that {GAL4Δ+UAS} also serves as an excellent enhancer detector for larval tissues in Drosophila.

When using the integrated {GAL4Δ+UAS} mutator-reporter, an additional reporter construct is not needed. Only two elements, a jumpstarter and {GAL4Δ+UAS}, are required to screen for enhancers. In this case, {GAL 4Δ+UAS} can be combined with 3xP3-DsRed-marked jumpstarters to avoid using the weak 3xP3-ECFP marker, which seems not to be suitable for all insect species. {GAL4Δ+UAS} and the jumpstarter Mi{3xP3-DsRed, hsp70-piggyBac} might actually be an ideal starting system for a first insertional enhancer detection screen in a nonmodel arthropod.

In contrast, {tTA} mutators only rarely detected enhancers (2%; Table 2). This further indicates that the tTA/TRE system has a much lower sensitivity for enhancer detection compared to GAL4/UASp. tTA/TRE-based constructs are therefore not reliable enough to screen for tissue-specific enhancers. This is unlikely to be due to higher toxicity of the tTA transactivator, since {tTA} jumping rates and transposition frequencies were similar to {GAL4Δ}. The low sensitivity of the tTA/TRE system is probably due to a lack of effective expression amplification in this binary system.

Insertional mutagenesis to identify novel Drosophila gene functions: To determine if piggyBac insertional mutagenesis screens could advance the Drosophila gene disruption project, we further analyzed the autosomal {GAL4Δ} and {tTA} insertions. By segregation analysis, we identified the chromosomal location for 236 novel insertions and balanced them (see materials and methods). When testing for recessive lethality, 4 insertions showed semilethality (homozygous progeny <10%) and 14 insertions were homozygous lethal (Table 3). This corresponds to a frequency of 7.6% for lethal or semilethal insertions, which is slightly lower than that observed in mutagenesis screens with P elements (Bellen 1999; Peteret al. 2002). Of these 18 insertions, the genomic localization was determined by inverse PCR (Huanget al. 2000), sequencing, and BLAST searches against the Drosophila genome sequence (Release 2; http://www.fruitfly.org/annot/release2.html). Molecular data confirmed the genetically determined chromosomal localization. The insertion in line 128 actually lies in an as-yet-unaligned scaffold, which, based on our genetic data, should be part of the third chromosome (Table 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

—Examples of larval enhancer activities isolated with the mutator-reporter element {GAL4Δ+ UAS}. (A) Line 258 shows fluorescence in the tracheal system, (B) line 254 in the fat body, (C) line 249 in the posterior spiracles and in some unidentified lateral cells, (D) line 251 in muscles, (E) line 252 in the posterior spiracles and in a subset of the peripheral nervous system, which differs from marker-mediated fluorescence (Hornet al. 2000), and (F) line 257 in a subset of thoracic muscles.

Insertions could be identified in well-characterized genes, like schnurri (shn; line 166; Staehling-Hamptonet al. 1995) or tramtrack (ttk; line 150; Brown and Wu, 1993), but also in noncharacterized CG-number genes or in genes for which no mutations have previously been identified, e.g., line 52 insertion in gene HLH106 (Theopoldet al. 1996; Rosenfeld and Osborne 1998). Within the genes, the insertions were found in the 5′ UTR, in exons, in introns, and downstream of them, but also in intergenic regions. The frequent insertion within introns actually makes piggyBac an ideal candidate for the design of protein trap systems (Morinet al. 2001). For many of the targeted genes, no P-element alleles have previously been identified and sometimes the closest P-element insertion referenced in GadFly lies ∼30 kb away. Table 3 shows further details on the different insertions.

To ascertain if the recessive lethality was correlated with the piggyBac-mutator insertion, we performed excision experiments by crossing the lethal lines with a jumpstarter strain. We then isolated chromosomes carrying an excision event and tested them for lethality over the original insertion chromosome. In 9 of the 14 lethal insertion lines, the recessive lethality could be reversed, indicating that the mutator insertion was indeed the cause of the lethality phenotype. For the other 5 lines, the lethal mutation was not associated with the piggyBac insertion. This may be due to mutations in the background, since the employed chromosomes were not recently isogenized. Thus, 3.8% (9 out of 236) of the established novel insertion lines caused reversible lethal mutations, which is within the range observed in P-element mutagenesis screens (Peteret al. 2002).

The ttk insertion (line 150) is allelic to the lethal alleles ttk¹ and ttk^1e11. The associated lethality can be reverted by mutator excision; thus, the intron-localized line 150 represents a true allele of ttk. This indicates that piggyBac insertional mutagenesis screens can be applied to mutate, to isolate, and to identify specific gene functions. In contrast, line 166, with the insertion in an intron of shn, is not allelic to the lethal alleles shn¹ and shn⁰⁴⁷³⁸, and the lethality cannot be reverted by excision of the mutator. Therefore the lethality must derive from another recessive mutation on the chromosome.

To determine if {GAL4Δ} insertions could also serve as embryonic enhancer detectors, we crossed the lethal insertion lines (Table 3) to a strain carrying the reporter Bac{3xP3-DsRed, UASp-lacZ-K10} and performed X-gal stainings (O'Kane 1998) on their embryonic progeny. Figure 5 shows the detected enhancer activities for tissues like salivary glands, specific muscles, segmental patterns, etc. Despite the fact that line 150 represents a ttk allele, no ttk-like expression pattern was observed. In this case {GAL4Δ} does not detect the enhancers of the gene in which it is inserted. In contrast, in line 166 {GAL4Δ} does detect shn-specific enhancers and drives lacZ expression in shn-like patterns. Thus, despite not creating a shn allele, {GAL4Δ} is picking up the enhancers of the gene. {GAL4Δ} can therefore be used to isolate genes on the basis of embryonic enhancer activities. However, when comparing the endogenous expression pattern of shn (Staehling-Hamptonet al. 1995) with the GAL4-mediated lacZ expression, we recognized a significant time delay. While shn shows dorsal-specific expression during blastoderm stages, the matching enhancer-mediated lacZ expression cannot be detected until germ-band retraction (Figure 5A). shn is gut-specifically expressed during germ-band retraction, but the corresponding lacZ expression cannot be detected until dorsal closure and head involution are complete (Figure 5B). This delay might actually be the reason why most observed enhancer activities are detectable only at late embryonic stages (Figure 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

—Embryonic enhancer activities from lethal insertions (Table 2) of the {GAL4Δ} mutator element detected by X-gal staining. (A and B) Line 166 is an insertion into an intron of the shn gene. The insertion does not cause a mutation of the gene, but detects some of its enhancers. The enhancers are detected with significant delay: the dorsally restricted blastoderm expression can be detected in the amnioserosa at the germ-band retraction stage (A), and the endoderm-specific expression of shn at that stage can be detected in the gut after dorsal closure (B). (C) Line 43 shows an enhancer activity for the posterior spiracles, (D) line 51 for the salivary glands, (E) line 52 for a segment polarity pattern, (F) line 80 for the head and posterior spiracles, (G) line 127 for the pharynx, and (H) line 139 for specific muscles.

Molecularly precise cut-and-paste mechanism of piggyBac: When the genomic localization of the insertions was determined by inverse PCR, the obtained genomic 5′ and 3′ sequences always matched to the same insertion site, indicating that only single insertions have been observed so far despite the high transposition rate of piggyBac. This is consistent with the nonreplicative, conservative cut-and-paste transposition mode described for piggyBac (Loboet al. 1999). To molecularly analyze piggyBac excision sites, PCR reactions were performed on successfully excised chromosomes from lines 29, 51, and 139. Our data confirm that piggyBac excision is molecularly precise and does not leave a footprint behind, which is unique among eukaryotic class II transposons (Elicket al. 1996). The original TTAA target site, which is duplicated upon insertion of the piggyBac element, is left as a single TTAA site after the element has been excised.

Imprecise excision events have not as yet been detected. This is disadvantageous compared to P elements, since it is unlikely that small deletions can be generated with piggyBac. Nevertheless, this might be overcome by including P ends into piggyBac-based mutators. Furthermore, the precise excision might actually be an advantage in the case that piggyBac would show hot-spot behavior like P elements (Spradlinget al. 1999). This behavior causes problems with P elements, since often the induced mutation does not correspond to the final integration site but to a previous insertion site, where it causes a mutation by imprecise excision. Thus even if the P element is not finally localized at a hot spot, the caused mutation often maps to it. This makes the isolation of novel mutant insertions with P elements rather difficult. Molecularly precise excision events as observed with piggyBac will avoid these problems.

Over all, piggyBac insertions behave properly in Drosophila. The aforementioned X chromosomal insertion of {tTA} that was not efficiently remobilizable and line 233, in which the {tTA} mutator could not be excised (Table 3), are the only two cases we observed so far. We therefore conclude that piggyBac can serve as a reliable tool in insertional mutagenesis approaches for the Drosophila gene disruption project.

Concluding remarks: Despite the limitations of our small pilot screen, it can be expected that piggyBac-based mutator elements will allow the isolation of novel gene functions through the identification of insertions in previously untargeted gene loci of Drosophila. Even within our limited screen, which detected nine novel mutator-caused lethal insertions, genes and chromosomal regions were targeted to which no P element has gone before. Moreover, transposon mutagenesis with piggyBac mutators could even be carried out in the presence of P elements, as in FRT-based mosaic (Xu and Rubin 1993) or eye misexpression screens (Karimet al. 1996). This would facilitate the isolation of previously unidentified gene functions in organogenesis and other late developmental processes.

Furthermore, the presented piggyBac-based insertional mutagenesis and enhancer detection system has been generated using broad-range transposable elements and widely applicable fluorescent transformation markers (Hornet al. 2002). This allows for utilization of the system in nonmodel insects and will therefore enable the identification of gene functions in many different biological processes. Initial experiments to test for remobilization of genome-inserted piggyBac insertions have already been successfully performed in the red flour beetle T. castaneum (A. Berghammer and M. Klingler, personal communication). The visible markers will facilitate stock keeping of mutant and enhancer detector lines, which is especially important for “balancer-less” nonmodel arthropods. Identification and isolation of insertional mutations is much simpler than that of chemically induced mutations, since the mutated genes or detected enhancers are molecularly tagged. This will facilitate cloning of the affected genes and thus help to rapidly correlate sequence data with biological functions, which is of key importance for successful functional genomics approaches in insects.

In addition, insertional mutagenesis and enhancer detection systems will help to identify cis-regulatory sequences from important agricultural pest species such as the Mediterranean fruit fly Ceratitis capitata. The isolation of sex-specific enhancers would make it possible to develop female lethality systems, which could be employed to generate male-only strains (Heinrich and Scott 2000; Thomaset al. 2000). In addition, early embryonic enhancers could be isolated for the development of embryonic lethality systems to generate competitive males for a transgenic sterile insect technique (Horn and Wimmer 2003).

Moreover, a major advantage of this elaborate insertional mutagenesis system is that it not only identifies interesting enhancers, but also at the same time provides tools to drive gene expression. Once enhancers of interest have been identified, they can be used to express any cloned gene as an effector in the respective embryonic, larval, or adult tissues and the effect of the expression on the tissues can be examined. In medically important disease vectors like the yellow fever mosquito Aedes aegypti or in Anopheles malaria mosquitoes, this will make it possible (i) to identify gut or salivary gland specific enhancers and (ii) to employ them for the expression and examination of peptides regarding their ability to block the transmission of these diseases (Itoet al. 2002).

piggyBac seems a good candidate for insertional mutagenesis and enhancer detection screens in a broad range of insect species. It should be noted, however, piggyBac shows some phylogenetic distribution (Handler 2002) and might therefore not be usable in all insect species. Even though stable piggyBac integration can be obtained in species that carry closely related transposable elements (Handler and McCombs 2000), the introduction of an active piggyBac transposase in such a species might trigger an uncontrollable insertional mutagenesis. Moreover, insertions will be hard to characterize molecularly. Consequently, other broad-range transposable elements, like Hermes, Minos, and Mos1, should also be further developed for insertional mutagenesis and enhancer detection systems. These will present invaluable tools to correlate biological functions with the explosion of DNA sequence information soon to emerge from the field of insect genomics. Functional genomics will rely heavily on possibilities of identifying gene expression patterns and of characterizing mutations of the proposed genes to understand their molecular and physiological roles.