Research Article

P[acman]: A BAC Transgenic Platform for Targeted Insertion of Large DNA Fragments in D. melanogaster

See allHide authors and affiliations

Science  15 Dec 2006:
Vol. 314, Issue 5806, pp. 1747-1751
DOI: 10.1126/science.1134426


We describe a transgenesis platform for Drosophila melanogaster that integrates three recently developed technologies: a conditionally amplifiable bacterial artificial chromosome (BAC), recombineering, and bacteriophage ΦC31–mediated transgenesis. The BAC is maintained at low copy number, facilitating plasmid maintenance and recombineering, but is induced to high copy number for plasmid isolation. Recombineering allows gap repair and mutagenesis in bacteria. Gap repair efficiently retrieves DNA fragments up to 133 kilobases long from P1 or BAC clones. ΦC31-mediated transgenesis integrates these large DNA fragments at specific sites in the genome, allowing the rescue of lethal mutations in the corresponding genes. This transgenesis platform should greatly facilitate structure/function analyses of most Drosophila genes.

Drosophila is an important model organism for studying biology and disease, and new tools are continually being developed to facilitate this research (1, 2). A major advance was the development of P-element–mediated transformation after the injection of plasmids into Drosophila embryos (3). Hence, P-element vectors have been engineered for numerous applications (4). However, P-element–mediated transformation has a number of limitations: inability to clone large DNA fragments in available P-element vectors, difficulties in manipulating large DNA fragments, inability to transfer large DNA fragments into the fly genome, and failure to target DNA to specific sites in the genome.

Cloning large DNA fragments in high–copy-number plasmids, such as typical P-element vectors, is inefficient because large fragments are unstable at high copy number in bacteria. Hence, low–copy-number vectors, including P1 (5) and bacterial artificial chromosome (BAC) (6) vectors, were developed to stably maintain large cloned DNA fragments. Unfortunately, low–copy-number vectors hamper sequencing, embryo injection, and other manipulations requiring large amounts of plasmid DNA. An elegant solution is a conditionally amplifiable plasmid that has two origins of replication (ori's): oriS for low-copy propagation, typical for P1 and BAC vectors; and oriV, which can be experimentally induced to high copy number (7). Hence, the introduction of conditionally amplifiable BAC features into fly transformation vectors is a first key step to manipulating large DNA fragments in Drosophila.

Cloning of large DNA fragments is limited by conventional methods that rely on restriction enzymes and DNA ligases, hampering analyses of large genes and gene complexes. Recently, efficient in vivo cloning technologies using enhanced and regulated recombination systems, commonly known as recombineering, have been developed (8). Recombineering greatly facilitates the retrieval of DNA fragments through gap repair and their subsequent site-directed mutagenesis. Because recombineering is based on homologous recombination, restriction enzymes and DNA ligases are not required. Recombineering is widely used by mouse geneticists to generate transgenic and knockout constructs. Recombineering-mediated mutagenesis is much more efficient with low-copy plasmids (8). Hence, using recombineering in a conditionally amplifiable BAC should greatly facilitate the gap repair of large DNA fragments and subsequent mutagenesis at low copy number.

P-element–mediated transformation is limited by DNA size, precluding the study of large genes (>40 kb) and gene complexes. In addition, more than 75% of P elements insert in regulatory elements of genes (9), often disrupting genes in subtle ways (10). Moreover, P elements are subject to position effects: the effect of a local chromosomal environment on the levels or patterns of transgene expression. This necessitates the generation and characterization of several transgenes for each DNA construct studied. Hence, site-specific integration would greatly facilitate structure-function analysis of transgenes, permitting direct comparison of differently mutagenized DNA fragments integrated at the same site in the genome. Recently, site-specific integration using the integrase of bacteriophage ΦC31 has been demonstrated in Drosophila (11). ΦC31-integrase mediates recombination between an engineered “docking” site, containing a phage attachment (attP) site, in the fly genome, and a bacterial attachment (attB) site in the injected plasmid. Three pseudo-attP docking sites were identified within the Drosophila genome, potentially bypassing desired integration events in engineered attP sites. Fortunately, they did not seem receptive to attB plasmids, because all integration events were at the desired attP sites (11). Thus, recovery of large DNA fragments by gap repair into a low-copy plasmid containing an attB site, followed by ΦC31-mediated transformation, might allow the integration of any DNA fragment into any engineered attP docking site dispersed throughout the fly genome.

Here we describe new vectors that overcome the limitations associated with P-element–mediated transgenesis. We developed P/ΦC31 artificial chromosome for manipulation (P[acman]), a conditionally amplifiable BAC vector that contains recognition sites for both P-transposase– (3) and ΦC31-mediated integration (11). P[acman] permits recombineering-mediated cloning of any genomic DNA fragment from Drosophila P1 or BAC clones (1215) and enables the transfer of large DNA fragments into the fly genome. The ability to easily manipulate these DNA fragments through recombineering and to introduce them into specific sites in the fly genome will greatly facilitate and accelerate in vivo genetic manipulations of Drosophila.


Construction of a BAC transgenesis vector for Drosophila. The motivation to construct P [acman] grew from our inability to clone 29- and 39-kb DNA fragments in existing P-element vectors for the transformation rescue of mutations in the gene senseless2 (16). We were unable to identify restriction sites for cloning the entire gene. Moreover, recombineering-mediated gap repair in existing high-copy P-element vectors was unsuccessful. This was consistent with published data demonstrating that gap repair in high-copy or medium-copy plasmids has an upper size limit of 25 and 80 kb, respectively (17). Because high-copy P elements have a substantial size, gap repair is limited to fragments of about 20 kb. We hypothesized that a low-copy plasmid might alleviate size limitations and improve the stability of large DNA fragments. Hence, we added P-element components to a chloramphenicol-resistant conditionally amplifiable BAC (7), resulting in P[acman]-CmR (Fig. 1A). The P-element components include the 5′P and 3′P termini required for P-transposase–mediated integration, a multiple cloning site (MCS), and the white+ marker. The conditionally amplifiable BAC contains two origins of replication: oriS and oriV (7). oriS keeps P[acman] at low copy number for the stability of large cloned inserts and efficient recombineering, whereas oriV permits copy-number induction for high-yield DNA preparation for sequencing and embryo injections. We created transgenic flies containing P[acman]-CmR using P-transposase–mediated integration. To facilitate cloning by gap repair from a variety of sources, we also replaced the ampicillin-resistance marker into P[acman]-CmR, resulting in P[acman]-ApR. Both plasmids can be used to clone any DNA fragment from a variety of donor vectors.

Fig. 1.

P[acman]: BAC transgenesis for Drosophila. (A) P[acman]contains P-element transposase sites (3′P and 5′P), white+, and an MCS. This P element is inserted in the conditionally amplifiable BAC, containing a low-copy origin of replication (oriS) and a copy-inducible origin of replication (oriV). (B) P[acman] is linearized between both homology arms (LA and RA) and transformed into recombineering bacteria containing P1 or BAC clones. Integration into the germ line of white flies is mediated by P-element–mediated transformation.

Recombineering-mediated gap repair into P[acman]. P[acman] was used to retrieve fragments by gap repair. For each gap repair, we designed four primer sets (fig. S1). Two homology arms, located at either end of the DNA fragment, were cloned into P[acman] (fig. 1B). Linearization between both homology arms and the subsequent transformation of the linearized construct into recombineering-competent bacteria containing the necessary genomic clone allow retrieval of the DNA fragment by gap repair. Gap repair was performed with the use of two similar strategies, relying on Red recombination functions (fig. S2) (17, 18). Both strategies were used to retrieve DNA fragments ranging from 9.4 to 39 kb, from different donor plasmids (Table 1). Colony polymerase chain reaction (PCR) screening identified correct recombination events at both junctions after gap repair. After plasmid copy-number induction, DNA fingerprinting and sequencing demonstrated that the desired fragment was obtained. Hence, the methodology is reliable and large DNA fragments can easily be retrieved.

Table 1.

Retrieval of genes in P[acman]. Genomic fragments containing genes of interest retrieved by gap repair into P[acman]-CmR or -ApR are shown. Donor plasmids, with clone coordinates, are P1, BACs, or ampicillin-modified BACs (BAC-ApR). Mutations in the corresponding genes were lethal or showed a phenotype. Rescue was obtained for most genes using P-element–mediated transformation. NA, not applicable.

sens2 sens2-22 22.3 kb P1 DS05421 CmR NA NA
sens2 sens2-29 28.9 kb P1 DS05421 CmR NA NA
sens2 sens2-39 38.9 kb P1 DS05421 CmR NA NA
CG10805 CG10805-S 9.7 kb P1 DS05421 CmR Lethal Yes
CG10805 CG10805-L 14.7 kb P1 DS05421 CmR Lethal Yes
dap160 dap160 10.9 kb P1 DS02919 CmR Lethal Yes
sens sens-S 12.1 kb BAC-ApR BACR17E13 CmR Lethal Yes
sens sens-L 18.1 kb BAC-ApR BACR17E13 CmR Lethal Yes
Dalpha7 Da7 29.4 kb BAC-ApR BACR02B03 CmR Phenotype Yes
Drp Drp 9.4 kb BAC BACR30P05 ApR Lethal YesView inline
Sec8 Sec8-L 12 kb BAC BACR02L23 ApR Lethal No
Eps15 Eps15-L 11.9 kb BAC BACR27P17 ApR Lethal Yes
  • View inline* From (25).

  • To test the functionality of the gap-repaired constructs, we used P-mediated transposition. Although germline transformation of the constructs was efficient with small inserts, the efficiency dropped for fragments over 20 kb. Transgenic flies containing the gap-repaired fragments in P[acman] were crossed to flies carrying mutations in the corresponding genes. Eight out of 9 fragments tested fully rescued the lethality or visible phenotype associated with the mutations, in the homozygous or transheterozygous condition (Table 1). One fragment (Sec8-L) did not rescue, although a smaller fragment encompassing the same gene did.

    ΦC31-mediated integration of P[acman]. Because P-transposase–mediated transformation has size limitations, we explored the possibility of using ΦC31-mediated integration to integrate larger constructs at specific sites within the genome. To combine the power of recombineering with ΦC31, we equipped P[acman] with an attB site, resulting in attB-P[acman]. To create genomic attP docking sites, we introduced an attP site and yellow+ into a minimal piggyBac transposon (19) (piggyBac-yellow+-attP) and integrated and remobilized this piggyBac in the Drosophila genome using piggyBac transposase. We isolated 34 homozygous viable insertions and determined their exact genomic location (Fig. 2A and table S1). Because piggyBac has a more random distribution than P-transposase (9), attB-P[acman] should integrate into both gene-poor and gene-rich regions.

    Fig. 2.

    P[acman] transgenesis in Drosophila using the ΦC31 system. (A) The piggyBac-y+-attP docking element was transformed or remobilized in the Drosophila genome to obtain multiple attP docking lines (VK lines). Locations are indicated on a schematic representation of the polytene chromosomes. White triangles represent insertions between, and black triangles within, annotated genes. VK line numbers (table S1) are indicated within the triangles. (B) attB-P[acman] can integrate at an attP docking site in the fly genome. (C) Correct integration events in docking sites are PCR-positive for the attR and attL assays, whereas original docking sites are PCR-positive for the attP assay. yw served as a negative control. Two docking sites on chromosomes X (VK6 and VK38), 2 (VK1 and VK2), and 3 (VK5 and VK13) were used. LM, length marker.

    To integrate attB-P[acman], we co-injected circular plasmid DNA and mRNA encoding ΦC31-integrase (11) into embryos carrying piggyBac-yellow+-attP (Fig. 2B). Integration results in flies with a “yellow+” body color and “white+” eye color phenotype. Because the attB site is upstream of the MCS and white+ is downstream of the MCS, only transgenic flies that are “yellow+ white+” should have undergone an integration event containing most of the injected DNA, including the cloned insert (Fig. 2B). Hence, successful integration events can be genetically traced. Moreover, because both transposons (P-element and piggyBac) are maintained after integration, the inserted DNA can be remobilized using the respective transposases.

    We tested whether different attP docking sites are equally receptive to attB-P[acman]-ApR. We focused on seven docking sites: two each on chromosomes X, 2, and 3, and one on chromosome 4. As shown in table S2, all but one of these docking sites were receptive, with similar integration efficiencies of 20 to 30%. We occasionally obtained efficiencies as high as 60%, similar to those in a previous report (11). However, in general we observed higher survival rates than Groth et al. (11). In our hands, such efficiencies were higher than the P-transposase–mediated integration of similarsized constructs. Because ΦC31-mediated–integration is site-specific, only a single insertion is needed, and the injection procedure can therefore be scaled down to 50 embryos or less. Integration of attB-P[acman]-ApR in the same docking site led to the same white+ expression level (fig. S3). This is convenient, because rare events that were not integrated in the proper site exhibited a different eye color and were therefore distinguished from true integration events. Integration of attB-P[acman]-ApR in a single docking site on chromosome 2 consistently caused a patchy red eye phenotype (fig. S3D), indicating position-effect variegation of white+ expression. This was not observed for the yellow+ marker present at the same docking site. Moreover, the insertion of attB-P[acman]-ApR into different docking sites resulted in different eye color phenotypes, indicating position effects on the expression level of the white+ marker (fig. S3). The main difference between ΦC31- and P-element–mediated integration is that position effects with the ΦC31 system are predictable, allowing for the selection of different but defined expression levels.

    To identify correct integration events, we developed PCR assays specific for the attP, attL (left attachment), and attR (right attachment) sites (Fig. 2C). Correct integration events were identified by the loss of the attP PCR product (specific for the original docking site) and the appearance of attL and attR PCR products (specific for the integration event). PCR analysis indicated that all but one of the insertions was correctly integrated. Moreover, correct integration events in homozygous viable docking sites maintained homozygous viability, demonstrating that the insertion does not detrimentally affect the local chromosomal environment. In numerous transformation experiments, in addition to proper integration events, we recovered only one event that converted a homozygous viable docking site into a recessive lethal locus. PCR analysis demonstrated that this event did not result in correct integration. Therefore, ΦC31-mediated integration is very efficient.

    Cloning of large DNA fragments into attB-P[acman]. To clone large DNA fragments into attB-P[acman]-ApR, we performed recombineering-mediated gap repair, as described above. We retrieved fragments up to 102 kb in length from single P1 or BAC clones (Table 2). Multiple genes and gene complexes, including bancal, Dscam, teashirt, and the Bearded and Enhancer of Spl complexes, were cloned into attB-P[acman]-ApR (Table 2). Unfortunately, some large genes were not contained within a single BAC (15). We therefore decided to reconstitute large genes through serial gap repair from two overlapping BAC clones into attB-P[acman]-ApR:Instep1, the smaller part is retrieved from one BAC, which is followed by step 2, retrieving the remainder of the gene (Fig. 3 and fig. S4). This two-step procedure was successful in retrieving a 133-kb fragment encompassing the Tenascin-major gene (Table 2).

    Fig. 3.

    Reconstitution of large fragments from two overlapping BACs by serial gap repair. Three homology arms are designed: LA1, RA1, and RA2, located at the left end, the region of overlap of the two BACs, and the right end of the desired DNA fragment, respectively. During step 1, the construct is linearized between LA1 and RA1, and the left segment of the gene is obtained from BAC 1, resulting in LA2. In step 2, the construct is linearized between LA2 and RA2, and the remaining segment of the gene is obtained from BAC 2 to reconstitute the entire gene.

    Table 2.

    Retrieval of genes in attB-P[acman]. Genome fragments containing genes of interest retrieved by gap repair into attB-P[acman]-ApR are shown. Donor fragments are from P1 or BACs, or from PCR. Clone coordinates of the donor clones are indicated. Mutations in all corresponding genes are lethal. Rescue is indicated for P-element– or ϕC31-mediated transformation. ND, not determined; NA, not applicable.

    Sec6 Sec6-S 2.8 kb PCR BACR27L09 Yes No Yes ND
    miR-4 miR-4 4.5 kb BAC BACR02J10 ND NA ND NA
    Sec8 Sec8-S 4.9 kb BAC BACR02L23 Yes Yes ND NA
    Eps15 Eps15-S 10.8 kb BAC BACR3B7 ND NA Yes ND
    Sec6 Sec6-L 11.5 kb BAC BACR27L09 Yes ND Yes ND
    sens sens-L 18.1 kb BAC BACR17E13 ND NA No NA
    miR-9a miR-9a 20.1 kb BAC BACR01D04 ND NA Yes ND
    Tsh Tsh-1 28.4 kb BAC BACR03L08 ND NA Yes Yes
    grp grp 29.8 kb P1 DS00592 ND NA Yes ND
    Brd-C Brd-C 37.2 kb BAC BACR01H12 ND NA ND NA
    bancal bancal 39.5 kb BAC BACR33D17 ND NA ND NA
    Dscam Dscam-1 73.3 kb BAC BACR26B18 ND NA Yes Yes
    E(Spl)-C E(Spl)-C 77.7 kb BAC BACR13F13 ND NA Yes Yes
    Tsh Tsh-2 86.4 kb BAC BACR03L08 ND NA Yes Yes
    Dscam Dscam-2 102.3 kb BAC BACR26B18 ND NA No NA
    ten-m ten-m 20 kb BAC BACR02D04 NA NA NA NA
    ten-m ten-m 133 kb BAC BACR22C11 ND NA Yes Yes

    ΦC31-mediated integration of large gap-repaired fragments. To obtain transgenic flies with attB-P[acman]-ApR constructs, we can use either P-transposase or ΦC31-integrase. We first tested P-transposition for three small fragments. Transgenic animals were obtained for all three (Table 2). Because P-transposase–mediated integration is inefficient for fragments larger than 20 kb, we switched to ΦC31-mediated integration. Several injection rounds suggested that the molar DNA concentration was critical. We empirically established that 75 ng of DNA per microliter of injection buffer (ng/μl) is the lower limit for attB-P[acman]-ApR (∼13 kb). Therefore, large (75- to 135-kb) supercoiled plasmid DNA (300 to 750 ng/μl) was co-injected with ΦC31 mRNA (250 to 500 ng/μl) (11). As shown in Table 2, we obtained transgenic animals for most constructs, including 73-, 76-, 86-, and 133-kb fragments. In these cases, the transformation efficiency was about 10% for medium (15 to 50 kb) and 2 to 4% for large plasmids (>50 kb). PCR analysis confirmed that all constructs integrated correctly. For two transgenes, 18 and 102 kb in length, we did not obtain transformants, illustrating that optimization for certain constructs might be required. Different gap-repaired fragments inserted in the same docking site differentially affected white+ expression, suggesting that DNA context is important. Finally, multiple fragments fully rescued two independent lethal mutations, in the homozygous or transheterozygous condition, in each of the four corresponding genes or gene complexes tested (Table 2). These data show that gap repair permits the cloning of BAC-sized fragments, large transgenes can be integrated site-specifically in the Drosophila genome using ΦC31, and mutations in essential genes and entire gene complexes can be rescued by large transgenic fragments.


    P[acman] provides numerous improvements when compared to current strategies for Drosophila transgenesis. First, DNA constructs larger than 100 kb can be retrieved from genomic P1 and BAC clones using recombineering-mediated gap repair. Fragments are retrieved into a plasmid fitted with an inducible oriV replication origin that allows easy preparation of large quantities of DNA for sequencing and Drosophila transgenesis. The retrieved fragments do not need to be resequenced because they are directly retrieved from the genomic clone without PCR amplification. Second, unlike P-transposase, ΦC31-integrase enables the integration of large fragments into the Drosophila genome. Because ΦC31-integrase catalyzes recombination between two ectopic attachment sites (attB and attP), transgenes are integrated at specific docking sites in the fly genome. This largely eliminates the problem of position effects, a highly desirable feature when comparing different mutagenized constructs derived from the same transgene for structure/function analysis. Finally, site-directed mutagenesis via recombineering is much more efficient in low-copy plasmids such as P[acman].

    We were able to clone fragments as large as 102 kb from single BACs. Only one report documents the gap repair of a similarly sized fragment from one BAC (20). In both cases, gap repair was successful because of the use of a low-copy vector. Indeed, gap repair into high-copy or medium-copy plasmids has an upper size limit of 25 and 80 kb, respectively (17). We also reconstituted one of the largest Drosophila genes, using serial gap repair, resulting in reconstitution of a 133-kb fragment. One recent report demonstrated a variation of seamless recombination of two large DNA fragments (20). The methodology described here may facilitate the cloning of even larger genomic fragments into P[acman], which may not be contained within single BAC clones in the genome tiling path set (15).

    The numerous docking sites (Fig. 2A and table S1) created in this work will have to be characterized in more detail to determine the expression levels of different genes that are inserted in the same site. It will also be important to determine whether adjacent enhancers or regulatory elements influence gene expression in each of the docking sites in order to identify sites that are as “neutral” as possible, a requirement important for the study of regulatory elements. The neighboring genome environment may also become important when overexpression or RNA interference transgenes are inserted. We feel that piggyBac-y+-attP insertions in intergenic regions may be the best candidates for these applications, although this will have to be determined experimentally.

    The circular ΦC31 bacteriophage genome is integrated into the linear genome of its host, Streptomyces lividans, by ΦC31-integrase (21). Thegenomeof ΦC31 is about 41.5 kb, which is larger than standard high-copy plasmids. This fact suggested that ΦC31-integrase would be useful to integrate circular DNA molecules up to 41.5 kb and potentially larger into the Drosophila genome. Indeed, we integrated fragments up to ∼146 kb at defined sites in the genome, which was previously impossible. Large gap-repaired fragments should complement many molecularly defined deficiencies (22), and subsequent mutagenesis should permit the analysis of genes within the deletion, obviating the need to recover mutations using conventional genetic screens or imprecise excision of mapped P elements. This approach also opens the road for clonal analysis of any gene that maps close to centromeric heterochromatin, because transgenic constructs for these genes can be recombined onto an FRT-containing chromosome. Finally, recombineering will allow the integration of any peptide or protein tag into the genomic rescue constructs to study protein localization and function in vivo.

    The methodology proposed here opens up a wide variety of experimental manipulations that were previously difficult or impossible to perform in Drosophila. Moreover, it is expected that it should be possible to adapt this methodology to other model organisms, because piggyBac transposes in many species, including mammals (23), and ΦC31 is operational in many species, including mammalian cells (24).

    Supporting Online Material

    Materials and Methods

    Figs. S1 to S4

    Tables S1 to S4


    References and Notes

    View Abstract

    Stay Connected to Science

    Navigate This Article