Gene vector and transposable element behavior in mosquitoes

Creating the technology to produce transgenic insects has been a goal of insect molecular geneticists for years because of the great need to have effective tools to find, isolate and analyze insect genes and to genetically modify insects for the purposes of insect control. Currently, there are at least four gene-vector systems derived from Class II transposable elements that can be employed to generate transgenic nondrosophilid insects[Hermes, Mos1 (mariner), Minos and piggyBac]. (Class II transposable elements move by the element precisely excising itself from its current location and reinserting itself in a new chromosomal location. This mechanism of movement is referred to as a cut-and-paste movement and is conservative in that it does not involve the replication of the element.) These systems are proving useful but their behavior can be unpredictable in the sense that, short of directly testing its mobility in the germ line, there are no good indicators of the potential effectiveness of a specific element in a species. While Hermes, Mos1,Minos and piggyBac have excellent host ranges, it is still impossible to predict their behavior in untested species. Elements may function efficiently in one species but inefficiently or not at all in close relatives. Hermes, for example, efficiently (>50%) transforms Drosophila melanogaster but transforms Aedes aegypti with frequencies of less than 10% (Jasinskiene et al., 1998; O'Brochta et al., 1996). piggyBac can efficiently transform Anopheles albimanus but appears to be much less inefficient in Anopheles gambiae (Grossman et al., 2001; Perera et al.,2002). mariner (Mos1) moves efficiently in Drosophila mauritiana yet is almost immobile in D. melanogaster (Bryan et al.,1987; Lidholm et al.,1993). Himar1 from Haematobia irritans moves efficiently in bacteria but has yet to show any evidence for mobility in insects (Lampe et al., 2000). In none of these cases was the mobility properties of the element predicted prior to use. Clearly, our understanding of these elements remains incomplete,and one aspect that remains largely unexplored is the behavior of these elements after they have become integrated into a foreign genome.

The post-integration behavior of insect gene vectors is particularly relevant to a number of future applications of this technology. Stability of integrated vectors is of concern to some researchers because genotypes created using these vectors may need to be maintained without change over time and would be of limited use in the laboratory or field if they were changing position within a genome or being lost from it. To those responsible for assessing the risks associated with releasing transgenic insects into the environment the issue of stability has additional significance. Instability,particularly if it is unpredictable, increases the uncertainty associated with describing accurately the phenotype of the insect and any attempt to model the behavior of a transgene over time in the genetically modified species.

The use of gene vectors for the purposes of creating insects with novel genotypes and phenotypes is only one of their many uses. Under some conditions, the transposable element-based gene vectors can be used to find and isolate genes by a variety of methods such as enhancer trapping and transposon mutagenesis, all of which rely on post-integration mobility or remobilization of the element. (Remobilization will be used to refer to the transposition of an insect gene vector after it has become integrated into a host's genome.) Futuristic plans for manipulating the genotypes and phenotypes of wild populations of insects such that they are no longer considered `pest'species requires the introduction and rapid transmission of transgenes through a population. Successful implementation of such a widespread genetic transformation might be accomplished by linking the transgene to an active transposable element and taking advantage of the ability of the element to spread or drive through populations under certain conditions. Hence, assessing the potential of existing insect gene vectors to serve as genetic drives requires that the post-integration behavior of the elements be thoroughly investigated. Here, we review and report on the post-integration behavior of the Hermes, Mos1, piggyBac and Minos elements in mosquitoes.

The Hermes element has been shown to transpose into the germ lines of several insect species by a cut-and-paste (conservative) mechanism of transposition (Michel et al.,2001; O'Brochta et al.,1996). In these insects, only those sequences on the vector-containing plasmid that are flanked by the terminal inverted repeats of the Hermes elements were integrated into the insect genome. Furthermore, target-site duplications 8 bp long were created at the site of insertion that conformed to the consensus sequence of 5′-GTnnnnAC-3′ observed for other hAT elements. Canonical cut-and-paste type transposition events, however, have never been observed in the germ lines of mosquitoes transformed using Hermes vectors. For both A. aegypti and Culex quinquefasciatus, germ-line integrations have involved not only all of those sequences delimited by the terminal inverted repeats of Hermes but also sequences flanking the element on the vector-containing plasmid(Allen et al., 2001;Jasinskiene et al., 1998, 2000; Kokoza et al., 2000; Moreira et al., 2000; Pinkerton et al., 2000).

Two forms of the Hermes element have been used in germline transformation experiments in mosquitoes, and both exhibit non-canonical integration patterns. The two types of Hermes elements that have been employed differ only in the terminal nucleotide of the right-hand inverted terminal repeat (ITR) sequence (Warren et al., 1994). These alternate forms of the right ITR, referred to as B5 and B6, involve a G-to-C transversion in the terminal nucleotide and originated during the isolation of two independent genomic DNA fragments from Musca domestica containing a Hermes element(Warren et al., 1994; Fig. 1). Elements containing the B5 end (with a terminal G) have been used in most of the studies of Hermes in insects (Allen et al., 2001; Michel et al.,2001; Pinkerton et al., 1996, 2000; Sarkar et al., 1997a,b),while the element containing the B6 ITR (with a terminal C) has been used in A. aegypti transformation experiments reported by others(Jasinskiene et al., 1998; Kokoza et al., 2000; Moreira et al., 2000).

We have examined the functionality of the Hermes B6 ITR in D. melanogaster to determine if it could also transpose by a cut-and-paste mechanism. A donor plasmid containing the B6 end and flanking M. domestica genomic sequence (Fig. 1) was constructed and is identical in every other way to the donor plasmid employed in previous studies. Hermes element plasmid-to-plasmid transposition assays were performed in developing D. melanogaster embryos using established procedures(O'Brochta et al., 1996). Transient plasmid-toplasmid transposition assays in embryos report the activity of a transposable element in somatic nuclei, since these vastly outnumber the germ-line nuclei in these embryos. Transpositions recovered from these assays are therefore most likely to originate from events occurring in somatic nuclei. We also repeated identical transposition assays using the B5 Hermes ITR, and, as previously shown, this Hermes element transposed by a cut-and-paste mechanism in Drosophila embryos(Table 1). By contrast, even though 10 times as many Hermes B6-containing donor plasmids were screened, no transposition events were recovered(Table 1). These experiments were repeated in developing A. aegypti embryos, and while Hermes B5 transposition events were recovered with a frequency of 0.4×10^-2 per donor plasmid screened, no transpositions were recovered when the B6 form of Hermes was used in this species. All interplasmid transposition events recovered using the Hermes B5 element were canonical cutand-paste events.

The donor plasmid containing the Hermes B6 element also contains different M. domestica genomic DNA at the right-hand end of the Hermes element compared with the Hermes B5 donor element. To determine whether the genomic DNA flanking the B6 element was negatively influencing the transposition behavior of the element, the element was converted to a B5 element by changing the terminal C to a G. This mutated B6 element (B6mut) was identical to the original B5 element, except for this mutation, but contained the flanking genomic DNA next to the B6 element(Fig. 1). When this mutated B6 element was used in transposition assays performed in Drosophila embryos, transpositions of Hermes into the target plasmid were recovered and confirmed both by diagnostic restriction digests and sequencing of the transposition junction fragments with the target plasmid(Table 1). The frequency of transposition was similar to that observed for the B5 form of Hermes, indicating that restoring the terminal nucleotide of Hermes B6 to G restores the activity of the element. These data also indicate that the M. domestica genomic sequence flanking both the B5 and B6 elements plays little or no role in determining the activity of the elements.

The functionality of the Hermes B6 element was tested further in D. melanogaster by using it as a germ-line transformation vector following the addition of the mini-white gene from this species. No transgenic individuals were recovered from 71 G₀ adults and 7532 G₁ progeny. Previously, we have shown that Hermes B5-mediated transformation of Drosophila occurs at frequencies of approximately 20-60% (Michel et al.,2001; O'Brochta et al.,1996). Therefore, the failure to obtain germ-line transformants using a B6-containing element indicates that this form of the element is inactive in the Drosophila germ-line.

The mobility properties of Hermes are distinctly altered when introduced into A. aegypti. In this species, the sequence requirements for element movement are altered, as indicated by the functionality of the B6 element, and germ-line integration reactions appear to rely on a mechanism other than simple cut and paste. Movements in the soma, as indicated by plasmid-to-plasmid assays, do appear to use a canonical cutand-paste mechanism.

To follow the fate of Hermes elements in transgenic mosquitoes and to examine their remobilization potential, an autonomous form of the B5 element was constructed(Guimond et al., 2003) and introduced into A. aegypti. As observed for previous germ-line integrations of Hermes in mosquitoes, integration of this element occurred by a mechanism other than cut-and-paste integration. This alternative mechanism resulted in the entire autonomous element being integrated together with sequences on the vector-containing plasmid that flank it(Fig. 2). This line was inbred for several generations, and transposable element display (TE display)analysis was used to examine the rates and patterns of transposition in this species, as was done with this same element in D. melanogaster(Guimond et al., 2003; Fig. 3). TE display permits the movement of Hermes to be monitored in both the germ line and the soma(Guimond et al., 2003). This method permits each of the unique junction fragments between the right and left ends of the element and flanking genomic DNA to be visualized as a gene amplification product. Transposition events are easily recognized by the amplification of junction fragments, each with a unique molecular mass. The TE display method is also sensitive enough to detect somatic movement of an element. Somatic remobilization during development results in a clone of cells that contain a new genotype with respect to the location of the element. Consequently, an insect in which there is somatic movement of the element as well as germ-line movement will yield a complex pattern of amplification products following TE display (Fig. 3). Elements inherited vertically through the germ line will be present in all cells of the insect (if there is no excision) and will yield an abundance of template from which amplification products will arise, while those elements transposing somatically will result in relatively less abundant template and products. In a TE display, somatic transposition events are seen as less-abundant PCR products relative to those arising from the vertically inherited element (Guimond et al.,2003). Somatic remobilization results in an insect that is a mosaic of genotypes. Somatic transpositions occurring early in development will be abundantly represented in the pool of template DNA, while those arising late will be under-represented. Somatic transpositions will therefore yield PCR products in the TE display method with varying abundance. As shown in Fig. 4A, Hermes B5is clearly mobile in somatic nuclei in this transgenic line of A. aegypti. Only a few weakly labeled fragments were seen in both wild-type untransformed mosquitoes and in transgenic mosquitoes containing the Hermes element into which the actin5C:EGFP (enhanced green fluorescent protein) gene but not the Hermes transposase gene had been inserted. Since the latter does not encode its own transposase, it is incapable of mediating its own movement through the A. aegyptigenome.

A number of well-isolated bands were excised, cloned, and their sequence determined, verifying the presence of Hermes terminal sequences and unique flanking DNA resulting from the transposition of the parental element(Fig. 4). In some cases, BLASTX analysis revealed similarities to sequences from gag proteins (L15 in Fig. 4) and integrase proteins(L16 in Fig. 4) encoded by retroviruses. The 8 bp of sequence immediately flanking the HermesITR displayed similarity to the 8 bpconsensus target-site duplication(5′-GTnnnnAC-3′) characterized for other hAT elements. The presence of these fragments indicates the transposition of the autonomous Hermes element in new sites in the genome of A. aegyptisomatic cells. Similar sequences indicative of cut-and-paste transposition of the Hermes element in these mosquitoes were also obtained from TE display analysis of the left-hand Hermes end(Fig. 4).

These data demonstrate that an autonomous Hermes element is capable of transposing in the soma of A. aegypti and does so in a way that is typical of Class II transposable elements. Following the initial integration of Hermes into the germ line of A. aegypti, this element becomes highly stable. By contrast, despite repeated efforts, we have never been able to detect the subsequent remobilization of this autonomous Hermes element in the germ line of A. aegypti. Thus, despite the transposase being placed under the control of the hsp70 promoter,remobilization of this autonomous Hermes element differs both between Drosophila and Aedes germ lines and between the soma and germ line of Aedes, indicating that regulation of Hermestransposition is controlled to some degree post-transcriptionally.

The Mos1 mariner element (Mos1) from D. maurutiana has a very broad host range and has been used to create transgenic microbes, protozoans, insects and vertebrates(Lampe et al., 2000). Consistent with these empirical findings of widespread functionality is the distribution pattern of mariner elements in the genomes of plants and animals. More than 15% of the 400 insect species examined for the presence of mariner elements tested positive(Robertson, 1993). In some cases, the presence of the element within a species seems best explained by horizontal transfer from another organism by an unknown mechanism followed by element amplification by transposition. These data suggest that mariner elements are autonomous recombination systems requiring few or no host factors for their movement. In vitro studies with purified transposase, donor elements and target sequences seem to support this conclusion, since transposition of Mos1 occurs in vitro in only the presence of transposase, Mg²⁺ and target DNA(Lampe and Robertson,1996).

Plasmid-based mobility assays in developing insect embryos demonstrated the ability of Mos1 to function in A. aegypti as well as in other insects (Coates et al., 1995, 1997). Transposition followed the typical pattern of cut and paste and resulted in the creation of characteristic TA target-site duplications. Mos1 was used to create transgenic insects by Lidholm et al.(1993), who successfully transformed D. melanogaster using Mos1-based vectors carrying the D. melanogaster mini-white gene. While Mos1 functioned as a gene vector in this species, the rate of integration was considerably lower (approximately 5%) than that observed when P, Hermes and hobo elements were used as vectors in this species. Integrations of Mos1 into the germ line of D. melanogaster were always by the canonical cut-and-paste mechanism. Mos1 has also been used repeatedly to create transgenic A. aegypti, beginning with Coates et al.(1998), who observed a rate of integration of approximately 4%. During the transformation of A. aegypti, Coates et al.(1998) found that three of the four initial lines contained Mos1 elements that had integrated in a cut-and-paste manner; however, one of the lines contained the Mos1element together with flanking plasmid DNA sequences, similar to what is routinely seen following Hermes integration in the germ line of this species. Germ-line transformation of A. aegypti using Mos1can also be achieved by co-injecting a Mos1 vector and purified Mos1 transposase protein instead of a helper plasmid(Coates et al., 2000). Interestingly, in this experiment, all (7/7) of the integration events involved not only the Mos1 element but also sequences flanking the element on the original vector-containing plasmid. Therefore, under some conditions, the integration behavior of Mos1 resembles the Hermes element in that transpositional recombination appears to lose some fidelity and the ITRs fail to accurately delimit the sequences involved in the reaction.

The Himar1 element, reconstructed from the closely related Cpmar1 element from Chrysoperla plorabunda using a deduced consensus sequence of a mariner-like element in H. irritans,also shows some unexpected host-dependent behavior. Although this element is functional in vitro and in a wide range of prokaryotes, it appears to be completely immobile in insects (Lampe et al., 2000). Robust efforts to create transgenic D. melanogaster by embryo microinjection have repeatedly failed and are generally consistent with the observations of others using Mos1 as a primary transformation vector. Even mutants of the Himar1 element isolated from Escherichia coli that are hyperactive in microbes are immobile in insects, suggesting that activity per se appears not to be the factor limiting or preventing the activity of this element in insects(Lampe et al., 2000).

The remobilization potential of existing Mos1-based gene vectors is very low in both D. melanogaster and A. aegypti. Lidholm et al. (1993) reported somatic mosaicism in ≤1% of the progeny containing a Mos1 vector. This is in stark contrast to the extreme instability of the white peach(w^pch) allele of D. mauritiana, which contained the first-isolated mariner element. The marinerelement present in w^pch was active in the presence of Mos1 transposase and resulted in somatic mosaicism in 100% of the progeny. Consequently, the high degree of stability of the Mos1 vectors used by Lidholm et al.(1993) was quite unexpected. Other descriptions of the post-integration stability of Mos1 vectors followed (Lohe et al., 1995; Lozovsky et al., 2002). Lozovsky et al. (2002)reported testing six Mos1 vectors in which the transgenic marker gene varied in length and position within the vector. Following the introduction of these elements into D. melanogaster using Hermes vectors,attempts were made to remobilize the Mos1 elements by providing functional transposase in trans. Of the four vectors that permitted somatic movement to be assessed, only one showed any evidence of somatic movement based on mosaicism in the eye and in this case at a rate of 0.23%. Using more-sensitive PCR-based methods to detect Mos1 movement(excision) throughout the body, only two of the four vectors tested displayed evidence of somatic movement. The other two vectors tested appeared to be completely immobile following their introduction into D. melanogaster.

Evidence for Mos1 remobilization in A. aegypti was first obtained by injecting embryos from transgenic lines containing cinnabar (cn)-marked Mos1 vectors with plasmids expressing Mos1 transposase (N.J., C.J.C. and A.A.J., unpublished data). Adults developing from injected embryos were backcrossed to kh^w, and progeny with pigmented eyes different from the parental phenotype were considered possible products of transposition. Of the 24 413 progeny with pigmented eyes examined, 2938 had phenotypes distinguishable from the parentals. A sample of these exceptional progeny was used to establish 24 families whose progeny was examined by Southern blot analysis. One line contained the Mos1 vector in a new location, indicating that a transposition event had occurred and demonstrating the potential of Mos1 to be remobilized in this species(Fig. 5).

An alternative strategy for testing Mos1 remobilization was used by Wilson et al. (2003). Two transgenic lines were created using Mos1 vectors containing the cn gene from D. melanogaster in an A. aegyptistrain homozygous for kh^w. To examine the remobilization potential of these elements in A. aegypti, heterozygotes were created with a third transgenic line containing a piggyBac vector into which Mos1 transposase gene under the regulatory control of the hsp82 promoter of Drosophila pseudoobscurra had been placed. Progeny from these heterozygotes was examined, and red-eyed progeny with an eye color phenotype different from the heterozygote parental insects was selected as putative transposition events and analyzed further using TE display (Guimond et al.,2003). Of the 14 000 red-eyed progeny examined by Wilson et al.(2003), only one contained a remobilized Mos1 vector (Fig. 6). These investigators also tested for somatic mobility of the Mos1 elements by performing TE display on Mos1-cn/hsp-Mos1heterozygotes. Approximately 50% of the 30 individual heterozygotes analyzed had evidence of at least one somatic remobilization event. In those individuals that had evidence for somatic movement, fewer than five events were detected. Wilson et al.(2003) did not estimate a frequency of somatic transposition but it appeared to be at least 10-fold lower than that observed with an autonomous Hermes element in D. melanogaster (Guimond et al.,2003).

The stability of Mos1 in A. aegypti was similar to that described in D. melanogaster by others(Lidholm et al., 1993; Lohe et al., 1995; Lozovsky et al., 2002). The unusual post-integration stability of Mos1 in D. melanogaster appears not to be a specific host effect but instead appears to be a general characteristic of the Mos1 element and observed in all species of insects into which it is inserted.

The basis of this post-integration stability remains unknown, although the current model proposes that there are critical sequences and spacing of sequences within Mos1 that are important in determining the efficiency of movement (Lohe and Hartl,2002). Lozovsky et al.(2002) found only one element that could be remobilized at detectable levels, and this particular vector consisted of almost two complete elements oriented as direct repeats and each flanking the marker gene. These investigators suggested that there are critical spacing requirements between certain sequence components of the element. Inserting marker genes and transgenes disrupts this spacing and impairs the element's ability to remobilize.

Lohe and Hartl (2002) also suggest that Mos1 movement in insects is dependent upon the presence of at least three blocks of sequences located throughout the element. These critical sequences are required in cis and are located at least 200 bp from the ends of the element. The `critical sequence' hypothesis of Lohe and Hartl (2002) is somewhat contentious for two reasons. First, while others have described the role of sequences other than the ITRs in the transposition process of other elements,these sequences have always been immediately sub-terminal and not dispersed throughout the element as proposed by these investigators. In addition, the cis-sequence requirements of Mos1 have been determined in vitro, and only approximately 60 bp and 30 bp of the left and right ends, respectively, of the element are essential for in vitrotransposition activity of a 1.1 kb vector(Tosi and Beverly, 2000). The minimized element created by Tosi and Beverly(2000) was only twofold less active than an unmodified element under the same conditions. Clearly, the data of Lohe and Hartl (2002) and Tosi and Beverly (2000)indicate an influence of the host on the activity and behavior of the element that is consistent with the observed immobility of functional and hyperactive Himar1 elements in insects.

The post-integration behavior of Mos1 in A. aegypti also revealed an aspect of this element that could be relevant to future applications. Of the somatic and germ-line Mos1 remobilization events recovered and analyzed by Wilson et al.(2003), approximately 25%(4/17) were instances of where the element transposed into a copy of itself. In all of the experiments in their study, the element was in a heterozygous condition, and therefore only during mitosis in diploid cells is a second copy of the element present to serve as a target for Mos1 transposition. It therefore appears that Mos1 can transpose sometime between the beginning of S phase and the end of metaphase. Transposition during S phase has been reported for the Ac/Ds element of maize and it is thought to play an important role in the increase in element copy number over time. Because cut-and-paste transposition is a conservative mechanism leading to no net increase in copy number, Class II elements must exploit in some way the host replicative machinery in order to increase in copy number within a genome. Replication may rely on gap repair mechanisms following element excision (Gloor et al., 1991; Lohe et al., 2000; Nassif et al., 1994) or by timing transposition to occur during chromosome replication(Fedoroff, 1989; Greenblatt and Brink, 1963; Wilson et al., 2003). The remobilization events reported by Wilson et al.(2003) may have taken place in polyploid cells in which there would be many copies of the Mos1element to serve as targets. Hence, the somatic transposition events detected,isolated and analyzed by TE display would each have come from a single polyploid cell. It is not clear whether the method of TE display has the level of sensitivity that would permit the detection of transposition events represented by perhaps only hundreds of copies of the template. In any case,however, the pattern of transposition of Mos1 indicates a distinctly nonrandom distribution. Preferential integration of the element into copies of itself on sister chromatids or homologous chromosomes suggests that the element may have a propensity to undergo local hopping whereby an element is more likely to jump to linked sites, often only kilobases or less from the initial integration site. Furthermore, the targeting of the element itself reflects a bias in the target site selection process that is similar to that seen during the process of element homing in which sequences contained on the element direct the transposition process to homologous sequences. Additional experiments would be needed in order to specifically address these hypotheses.

piggyBac vectors have been used to transform a wide range of insects, including representatives from the orders Diptera, Lepidoptera,Coleoptera and Hymenoptera. The Diptera have included the mosquitoes A. aegypti (Kokoza et al.,2001; Lobo et al.,2002), Anopheles stephensi(Nolan et al., 2002), Anopheles albimanus (Perera et al., 2002) and Anopheles gambiae(Grossman et al., 2001), with transformation efficiencies ranging from approximately 1% in A. gambiae to 40% in A. albimanus. Initial reports described all piggyBac integrations in mosquitoes as arising from perfect cut-and-paste transposition reactions into TTAA target sites that were duplicated during integration. More recently in A. aegypti,exceptional transposition events have been recovered and described. Adelman et al. (in press) created six transgenic lines of A. aegypti containing piggyBac vectors with EGFP under the regulatory control of the 3XP3 promoter. Two of the four lines analyzed further arose from the insertion of a single vector. The remaining two lines arose from the insertion of multiple copies of the piggyBac vector in tandem arrays. In addition, these complex integrations of the piggyBac vector were accompanied by the incorporation of helper-plasmid sequences containing piggyBactransposase. These transgenics appeared to arise from either the non-canonical transposition of piggyBac or by recombination mechanisms unrelated to transposition.

After screening the progeny of approximately 900 G₀A. aegypti adults homozygous for the kh^wallele, we recovered four independent transgenic lines expressing the kynurenine hydroxylase gene (cn) from D. melanogaster(D.A.O'B., N.S. and P.W.A., unpublished data). Two of these lines resulted from cut-and-paste transposition events in which only sequences delimited by the ITRs transposed. The remaining two lines arose from non-canonical reactions and resulted in the transposition of sequences flanking the element on the original donor plasmid. These events resembled Hermesintegration events in this species, in which flanking DNA present on the donor plasmid is invariably transferred during the transpositional recombination reaction in the germ line during transformation. In both cases, only a single element integrated and, while the extent of the flanking DNA has not been fully determined, it includes at least 200 bp of vector sequences at both ends.

Post-integration remobilization of piggyBac has been reported in D. melanogaster (Horn et al.,2003) and Tribolium castaneum(Lorenzen et al., in press). Horn et al. (2003) described the remobilization of piggyBac by establishing transgenic lines of D. melanogaster containing a single insertion of a non-autonomous piggyBac vector containing enhancer-reporter genes (reporter lines)and other lines containing single insertions of either a Hermes,mariner or Minos element containing the piggyBactransposase gene under the regulatory control of a strong promoter(jumpstarter lines). Remobilization was induced by crossing reporter and jumpstarter lines and testing the progeny for reporter elements with new linkage relationships relative to the original element. These investigators found that, on average, 80% of the heterozygotes tested produced progeny that included a transposition event. All transposition events were canonical cut-and-paste recombination reactions. For T. castaneum,remobilization was achieved by injecting developing embryos from a transgenic line containing a single piggyBac element containing a unique reporter gene expression pattern (3xP3 EGFP) resulting from the local influence of a tissue-specific enhancer with a piggyBactransposase-expressing helper plasmid. G₁ individuals were screened for a modified reporter gene expression pattern. Of the 32 adults developing from embryos injected with helper plasmid, nine produced progeny that had an altered pattern of reporter gene expression relative to the parental insects,yielding an estimated rate of remobilization under these conditions of 28%.

Currently, there are no well-documented cases of piggyBacremobilization in mosquitoes. Adelman et al.(in press) observed what appeared to be severe instability of the arrays of piggyBac vectors in two of their transgenic lines. These arrays appeared capable of wholesale excision and loss in one generation. The role of piggyBac transposase in this process is not known. These investigators report no evidence for piggyBac transposition within the genome of these transgenics despite the presence of a piggyBac transposase source. We have constructed multiple reporter and jumpstarter lines of A. aegypti homozygous for kh^w (D.A.O'B., N.S. and P.W.A., unpublished data). Reporter lines were created using a non-autonomous piggyBacvector expressing the wild-type D. melanogaster cinnabar(cn) gene and also containing DsRed under the minimal,un-enhanced hsp70 promoter of D. melanogaster. The DsRed gene in this vector serves as a reporter of local enhancer activity. The piggyBac vector can serve as a reporter of remobilization in two ways. First, cn gene expression in transgenic A. aegypti is position dependent and will allow transposition events to be recovered by selecting progeny with an altered eye color phenotype. Second, the appearance of a novel pattern of DsRed expression in progeny relative to the parental line will indicate the relocation of the reporter element within the genome of the progeny. Jumpstarter lines were created using non-autonomous mariner vectors containing EGFPunder the regulatory control of the 3XP3 promoter and the piggyBac transposase gene under the regulatory control of the D. melanogaster hsp70 promoter. The hsp::pBactransposase gene in the jumpstarter element was the same gene contained on the helper plasmid used in the creation of the initial piggyBac-containing reporter lines. Heterozygous insects containing combinations of the piggyBacremobilization reporter elements and jumpstarter lines were created and backcrossed to kh^w. Progeny expressing cn and with a phenotype different from the heterozygous parent were selected as potential remobilization events. The genotype of these insects was determined by TE display using piggyBac-specific primers. To date,none of the A. aegypti progeny selected on the basis of an altered eye color phenotype (putative remobilizations) have yielded a pattern of PCR products from TE display that is consistent with germ-line or somatic movement of the element (Fig. 7). Only a limited number (3000) of progeny have been screened and 100 of these have been genotyped by TE display because they had an eye color phenotype that suggested the presence of a transposition event. The 3000 progeny screened arose from a cross between kh^w and heterozygotes for reporter and jumpstarter elements that had been heat-shocked once during larval development at 41°C for one hour. Under the limited conditions and scope of this experiment, piggyBac does not appear to be efficiently remobilized in A. aegypti.

Minos, although tested less extensively than other insect gene vectors, appears to have a broad host range. Evidence for transposition activity in Orthoptera, Lepidoptera and Diptera, including the mosquito A. stephensi (Catteruccia et al., 2000a,b; Klinakis et al., 2000; Zhang et al., 2002), has been reported. Recently, evidence for its ability to transpose in mammalian cells was reported (Drabek et al.,2003). Germ-line transposition rates in insects are similar to those reported for other elements. In the medfly (Ceratitis capitata), transgenics were recovered from 1-3% of the fertile G₀ adults (Loukeris et al.,1995) and this frequency can be improved by directly providing transposase mRNA during the embryo injection process(Kapetanaki et al., 2002). In all cases where it has been determined, Minos integrates using a cut-and-paste mechanism and there is no evidence for non-canonical integration events as reported for mariner, Hermes and piggyBac.

There are no published data regarding the ability of Minos to be remobilized following its integration into the genome of any organism. Unpublished claims of efficient remobilization of Minos in D. melanogaster exist; however, the data supporting these claims have not been published(http://www.minosbiosystems.com/). Similarly, preliminary attempts have been made to remobilize integrated Minos elements in A. stephensi. Heterozygotes were created by crossing a Minos-containing vector and a line expressing Minos transposase. No progeny that had the Minos element in a new position as a result of transposition was recovered from these insects,indicating that under the conditions tested by the investigators remobilization did not occur (F. Catteruccia and A. Crisanti, Imperial College, personal communication).

Limited data currently exist about the remobilization potential of existing insect gene vectors after their integration into a foreign genome. Most evidence has been obtained from experiments with D. melanogaster and has shown that Hermes and piggyBac remobilization occurs readily in this species. The rates of remobilization of Hermes and piggyBac differ, although other aspects of the behavior of these systems, such as integration site specificity with respect to chromosomal regions and genes, have not been systematically compared. Mos1 stands in contrast to these elements in that it is almost immobile following its initial integration into the genome of D. melanogaster. Mos1 appears to contain more stringent size requirements as well as constraints on the presence and spacing of cis-critical sequences.

The behavior of current gene vectors in D. melanogaster does not necessarily reflect the behavior of these elements in mosquito genomes. Whereas Hermes, Mos1 and piggyBac are functional in mosquitoes, certain differences in behavior can be seen, both among these elements and between D. melanogaster and mosquitoes. All of the elements appear to have similar activities as primary transformation vectors in, for example, A. aegypti and yield transformants from 2-5% of the surviving injected embryos. All three elements have displayed non-canonical transposition behavior during the primary integration reaction. In the case of Hermes, all primary integration events involve a mechanism other than cut-and-paste transposition. Mos1 and piggyBac display cut-and-paste transposition activity, and evidence also exists for a non-canonical transposition reaction similar to that seen with Hermes. Although the nature of this alternative mechanism(s) is not known, it consistently results in the transposition and integration of non-vector sequences that flank the ITRs of the element in the donor plasmid. None of the elements, under the conditions tested, appears to remobilize efficiently in the germ line of this species. Only very rare germ-line transposition events have been detected with Mos1. Hermes and piggyBac remobilizations in the germ line have not been detected. However, Hermes and Mos1 do appear to remobilize in the soma of A. aegypti. Although the rates of somatic transposition are unknown, it is clear that for both species the remobilization events involve cut-and-paste transposition reactions. Currently, we have no evidence of somatic activity of piggyBac in the soma of A. aegypti. The behavior of Minos, although less well characterized, seems to resemble that of the other elements, with evidence for remobilization in D. melanogaster but not in mosquitoes.

The post-integration behavior of Hermes, Mos1, Minos and piggyBac is revealing interesting aspects of each of the elements and of the species in which they are being tested. First, these studies illustrate that there can be significant interactions between the transposable element and the host. Even though a transposable element system can be shown to function with minimal requirements in vitro, consisting of only transposase, Mg²⁺ and target DNA, the requirements for proper functioning in vivo may be more extensive. These studies also illustrate how different elements can be affected more or less by the same factors. Finally, it appears notable that the `unusual' behaviors of the existing insect gene vectors are pronounced in the germ line of A. aegypti and apparently not in the germ line of other insects. While more work needs to be done to explore the post-integration behavior of these elements, it is interesting to speculate that A. aegypti may possess a mechanism in its germ cells that helps it resist invasion by horizontally transferred DNA.

On a practical level, the apparent stability of the vectors tested in the germ line of A. aegypti indicates that the stability of the transgenics being created with these insects will be very high even in the presence of homologous functional transposase. These observations may be significant in assessing the risks associated with the creation and eventual release of transgenic mosquitoes into the environment.

The authors' research reported here was supported in part by the National Institutes of Health and The Wellcome Trust.