Site-Specific Cassette Exchange Systems in the Aedes aegypti Mosquito and the Plutella xylostella Moth

Genetically engineered insects are being evaluated as potential tools to decrease the economic and public health burden of mosquitoes and agricultural pest insects. Here we describe a new tool for the reliable and targeted genome manipulation of pest insects for research and field release using recombinase mediated cassette exchange (RMCE) mechanisms. We successfully demonstrated the established ΦC31-RMCE method in the yellow fever mosquito, Aedes aegypti, which is the first report of RMCE in mosquitoes. A new variant of this RMCE system, called iRMCE, combines the ΦC31-att integration system and Cre or FLP-mediated excision to remove extraneous sequences introduced as part of the site-specific integration process. Complete iRMCE was achieved in two important insect pests, Aedes aegypti and the diamondback moth, Plutella xylostella, demonstrating the transferability of the system across a wide phylogenetic range of insect pests.


Introduction
The ability to introduce exogenous DNA segments into predetermined locations in insect genomes is crucial for insect synthetic biology. In particular, it is important for insects intended for potential field release that the synthesis method can exclude extraneous sequences such as antibiotic resistance genes and other plasmid-derived sequences not required in the final product [1].
Currently, non-autonomous transposable elements, such as piggyBac, are widely used for germ-line transformation of insects. The piggyBac transformation system inserts transgenes specifically into TTAA tetranucleotide target sites [2]. These are present in large numbers in insect genomes, leading to partially-random integration of piggyBac-based constructs. Insertion at different sites may have different fitness costs [3], furthermore the phenotypic expression of a transgene is affected by its insertion site, a phenomenon known as a genomic position effect [2]. Consequently, partially-random insertion allows the production of a panel of lines with independent insertion events having slightly different properties, from which lines with the preferred properties can then be selected. In contrast, re-use of the same insertion site would allow better comparability and consistency between insertions of different constructs, and a deep knowledge base could be developed for selected insertion sites [4]. This would be highly desirable for synthetic biology, with its emphasis on standardisation and characterisation of genetic elements.
The FC31-att system involves recombination between the specific attachment sites (attP and attB), producing hybrid sites attL and attR [5]. These hybrid sites are no longer recognised by the integrase making the integration stable and efficient for transgenesis. The specific target sequences are sufficiently complex that they do not occur naturally in insect genomes, though 'pseudo-attP sites' exist in Ae. aegypti that allow off-target insertions at very low frequency [6]. This site-specific integration systems require the initial integration of the target sequence into the host's genome as a 'docking site'. Transgenes of interest can then be inserted into the 'docking site' in the presence of the appropriate recombinase. Differences attributed to the transgenic modifications can be directly compared since position effects would remain constant between integrated transgenes.
FC31-att integration has been successfully demonstrated in a diverse range of species, including fruit flies [4,7], mosquitoes [6,8,9], moths [10], human cells [11], plants [12], mice [13] and zebrafish [14]. Despite its transferability, a limitation of the FC31-att system is that the whole attB plasmid inserts into the insect genome including antibiotic resistance genes residing in the plasmid backbone. For insects intended for potential field release in particular, it is important that the integration method excludes sequences such as antibiotic resistance genes and other plasmid-derived sequences not required in the final product [15][16][17][18].
To overcome this, recombinase-mediated cassette exchange (RCME) is of great value [19,20]. RMCE is an integration strategy that makes use of site specific recombinases, in which a donor sequence and a target sequence are both flanked by site-specific recombination sites. Double recipricol recombination between the two sites results in the integration of the DNA segment between them as long both sites are heterospecific (cannot recombine together). RMCE can be achieved by using inverted pairs of recombination sites flanking a cassette to be inserted. This cassette may be interchanged with another, flanked by the corresponding recomination sites that has been pre-inserted into the genome, in the presence of the recominase [19]. Since only the cassette of interest is inserted, no extraneous sequence need be included in the final insertion. This sequence conservative integration makes RMCE an attractive strategy for site-specific genome engineering of insects intended for field release.
The recombinases widely used for RMCE are the Cre [21] and FLP [22] recombinases, which promote recombination between the short nucleotide sequences loxP (34 bp) and FRT (48 bp) respectively. The recombination reaction is reversible in the presence of these recombinases, such that the rate of RMCE is low, whilst recombination is hampered by the thermodynamically disfavoured insertion of any donor cassette larger than the acceptor cassette [19]. As such, the use of Cre and FLP recominases is undesirable for the creation of efficient and stable RMCE systems. RMCE mediated by Cre-loxP [21] or Flp-FRT [22] systems have been demonstrated in mammalian cells [23][24][25], plants [26] and prokaryotes [27,28]. In insects, Cre-and Flp-RMCE was developed in Drosophila melanogaster [29,30]. Flp-RMCE was also shown in the silkworm, Bombyx mori [31]and Cre-RMCE in the tephritid fuit fly, Anastrepha suspense [32].
FC31 site-specific integrase can overcome the problem of poor integration efficiency encountered by reversble recombinases because the attL and attR sites formed after recomination are inert to further recombinase exposure. An efficient RMCE system using FC31-integrase (FC31-RMCE) was developed to allow for negative marker selection of transgenic D. melanogaster using two pairs of inverted recombination sites [33]. FC31-RMCE was also shown in the silkworm, Bombyx mori [34].
We tested the transferability of the FC31-RMCE system to mosquitoes by developing the system in the dengue-transmitting mosquito, Ae. aegypti. Furthermore, we developed an alternative two-step integration-excision RMCE system, integrase-recombinase mediated cassette exchange (iRMCE), which utilises FC31-att integration [5] of exogenous genetic material, followed by Cre-loxP or FLP-FRT excision of plasmid backbone sequences. Components of this novel system are interchangeable, and experiments were conducted to demonstrate the reproducibility of this new iRMCE system in two different insect species, the Ae. aegypti mosquito, and the agriculturally important pest moth, Plutella xylostella.

Plasmid constructs
All plasmid sequences are provided in Table A in S1 File. Plasmids were constructed using standard molecular techniques from existing DNA constructs.

Aedes aegypti strains and rearing
The Ae. aegypti wild-type strain used was colonised in 1975 from Jinjang, Selangor, Malaysia. The strain was reared at 26°C (±1°C) and 70% (±10%) relative humidity with a 12:12 hour light:dark cycle. Larvae were fed on finely ground TetraMin Ornamental Fish Flakes (Tetra GmbH), and adults on a 10% sucrose solution with 14 U ml -1 penicillin and 14 μg ml -1 streptomycin. Adult females were fed on defibrinated horse blood screened for pathogens (TCS Biosciences).

Plutella xylostella strains and rearing
The P. xylostella wild-type strain was obtained from Syngenta laboratories. The strain was reared at 27°C (±1°C) and 60% (±2%) relative humidity with a 12:12 hour light:dark cycle. Adults were fed 7.5% sucrose solution with 100 μg ml -1 tetracycline and 0.07% methylhydroxybenzoate. Eggs were collected on cabbage extract-coated Parafilm strips, which were moved to a plastic beaker containing beet armyworm artificial diet (catalogue number F9219B, Bio-serv).

Generating transgenic strains
Injection solutions for transformation contained the transgenic DNA plasmid and the capped mRNA helper suspended in a buffer. The buffer had a final concentration of 5 mM KCl and 0.1 mM NaH 2 P0 4 , pH 6.8. DNA plasmids were purified using EndoFree Plasmid Maxi Kit (Qiagen). Helper piggyBac transposase transcribed from OX3081 [8], site-specific FC31 integrase [6], the FLP [38] and Cre [39] recombinases were provided as capped mRNA transcribed with the mMESSAGE mMachine T7 Kit (Ambion) and purified using the MEGAclear Kit (Ambion); sequences included the Drosophila melanogaster vasa gene's 3'UTR [40].
Pre-blastoderm Ae. aegypti embryos were microinjected as described [41] using a BA400 Motic light microscope (Microscope Services Ltd), an MN-151 micromanipulator (Narishige), and an Eppendorf FemtoJet microinjector air-pump (Eppendorf). To remove the oil, cover slips of the newly injected embryos were placed vertically in water for up to three hours, and then placed in fresh water for four days, giving the embryos sufficient time to fully develop before hatching.
Pre-blastoderm P. xylostella embryos were injected as described [42]. Glass slides of newly injected embryos were placed in a sealed box for~48 hours, then transferred to a beaker containing 30 ml of the beet armyworm artificial diet to hatch.
Up to 100 female G0s were pooled in cages with 1:1 wild-type males and their offspring were screened for the transgenic marker. For FC31-mediated integration and FC31-RMCE, 2-3 G0 males were pooled in small pots with wild-type females (1:10 males:females). After three days, insects from up to ten pots were pooled into each cage and offspring from these cage were screened for the transgenic marker. For Cre or FLP injections, males were pooled into 1-4 G0 males per pot with wild-type females (1:10 males:females) and offspring were collected from these pots for screening. Female G0 survivors from the same injections were pooled together into cages with up to 100 G0 females as before with wild-type males (1:1). Following a blood meal, these G0 females were individually placed in tubes for egg collection, and all offspring were screened. In all cases, G1 offspring were screened from two gonotrophic cycles.
For FC31-mediated integration and FC31-RMCE, and for Cre/FLP-mediated excision, a mixure of hemizygous and homozygous embryos were injected with the recombinase and donor construct. G0s lacking any transgenic marker were discarded.
The number of recombination events were defined as the number of G0 pools that produced transgenic offspring. Therefore potential multiple transgenic G1s from the same G0 pool were conservatively assumed to represent only a single transformation event from one G0 parent.
We calculated the minimum recombination efficiencies as the proportion of injected survivors (G0s) that produced transgenic offspring. This results in a lower calculated efficiency than if only fertile G0 individuals were considered. Multiple transgenic G1s from the same G0 parent or pool were assumed to represent a single recombination event.

Genomic DNA exctraction
Genomic DNA was isolated from single insects using the Machery-Nagel Nucleospin Kit (Machery-Nagel) according to the manufacturer's protocol and eluted in 50 μl endonuclease free water.

Adaptor-based amplification of flanking genomic nucleotide sequences
To obtain the genomic nucleotide sequences flanking the transposed piggyBac-based construct, an adaptor-mediated PCR amplification technique [42] was performed using primers specific for the adaptor and piggyBac sequence ( Table B in S1 File). PCR products were cloned using the GeneJET PCR Cloning Kit (Fermentas), and resulting minipreps sequenced by GATC Biotech.

PCR analysis of integration and excision events
PCRs were carried out using DreamTaq polymerase (Fermentas) according to the manufacturer's protocol, with 35 cycles of: 15 seconds at 94°C; 30 seconds at 55-60°C (primer-dependant); 72°C for 30 seconds per DNA kilobase, and; 7 minutes at 72°C. Primers are listed in Table C in S1 File.

Development of RMCE docking lines
Wild-type Ae. aegypti embryos were co-injected with a docking construct and capped piggyBac transposase mRNA (embryos injected: 1000, G0 survival: 25%. Three independent Ae. aegypti docking strains for FC31-RMCE (OX4372A, OX4372F, and OX4372I) were developed that showed 1:1 transgenic:non-transgenic Mendelian inheritance pattern of the docking constructs' fluorescent markers consistent with the presence of a single insertion.
Six independent Ae. aegypti OX4476 docking strains for iRMCE (embryos injected: 2000, G0 survival: 25%) were developed, each originating from separate G0 pools. Of the six Ae. aegypti OX4476 strains, one had a fitness cost associated with the transgene (only 18% of emerging adult offspring of hemizygous parents were transgenic, rather than the expected 50%), and one had multiple insertions. The remaining four showed typical 1:1 Mendelian inheritance of the construct. Ae. aegypti OX4476C, OX4476F, and OX4476H strains were selected for use in subsequent experiments.
Wild-type P. xylostella embryos were co-injected with docking construct OX4540 and the piggyBac transposase helper plasmid [42] (embryos injected: 1868, G0 survival: 47%); postinjection survival was equivalent to that previously reported [42]. Out of 22 independent P. xylostella OX4540 strains obtained, two were sterile and Mendelian inheritance data showed that five had multiple insertions and two had a lower proportion of transgenic progeny than non-transgenic progeny (39% and 42% transgenic progeny), which may indicate a fitness cost associated with the transgene. The remaining thirteen strains showed a typical 1:1 Mendelian inheritance of the construct and were therefore thought to have only one insertion. P. xylostella OX4540A was picked randomly from these thirteen strains.
Genomic sequences flanking the transgenic docking constructs were determined for all transgenic strains except Ae. aegypti OX4372F, Ae. aegypti OX4476H, and P. xylostella OX4540A (Table 1), which are thought to have inserted into highly repetitive sequences that are difficult to amplify [43]. Flanking sequences were different in each strain and, as expected [44], all carried the TTAA tetranucleotide sequence at the construct-genomic junctions. Obtained sequences were compared to the published Ae. aegypti sequence databases, and all sequences aligned to genomic regions with no known coding sequences.

ΦC31-RMCE in Ae. aegypti
Ae. aegypti OX4372 docking strains were co-injected with donor plasmid OX4312 and capped FC31 integrase mRNA. G0s were pooled and minimum recombination efficiency was calculated ( Table 2). The number of recombination events were defined as the number of G0 pools that produced transgenic offspring carrying the Hr5ie1-AmCyan marker (blue bodies). Potential multiple RMCE G1s from the same G0 pool were conservatively assumed to represent only a single RMCE event from one G0 parent.
No FC31-mediated integration was indicated in G1 offspring from injected OX4372F insects.
The cuticular expression of the AmCyan and the absence of the DsRed2 marker in OX4372I G1 offspring (Fig 1) indicated germline recombination between both pairs of attachment sites and thus complete cassette exchange, creating new strain OX4372I[OX4312]. This was confirmed by amplifying and sequencing across both FC31 attachment sites and fluorescent marker junctions (Fig A in S1 File).
All G1 transformants from OX4312 injections into OX4372A resulted in the presence of cuticular expression of DsRed2 and patchy AmCyan expression in the optic nerves. This indicated incomplete cassette exchange due to recombination of only a single pair of attachment sites (Fig 1b), creating 'OX4372A[OX4312]-incomplete'. This incomplete recombination event was confirmed by positive amplification of the intact attP-DsRed2 3' junction, and also of the recombined attR-AmCyan 5' junction (see Fig A in S1 File). The OX4372A[OX4312]-incomplete strain showed reduced expression of the docking cassette's DsRed2 marker, as we have often seen in other FC31 integrations of complete plasmids leading to insertions carrying two 3xP3-based markers [8]. Individuals of this partially recombined line, OX4372A[OX4312]-incomplete, were crossed to each other and their offspring, a mixure of hemizygous and homozygous embryos, were injected with additional FC31 integrase mRNA. This resulted in loss of the DsRed2 marker and brighter expression of the AmCyan, which indicated a second recombination event between the remaining attB and attP sites, resulting in complete cassette exchange and creating OX4372A[OX4312].
Recombination efficiency was much higher during this second round of injection at 18.5%, perhaps due to the close proximity of the two attachment sites, such that the recombinase had a smaller search volume of potential sites. Complete cassette exchange was confirmed in line Ae. aegypti OX4372A[OX4312] by positive amplification followed by sequencing of the recombined attR-AmCyan 5' and 3' junctions, and absence of the attP-DsRed2 5' and 3' junctions (see Fig A in S1 File).
iRMCE:ΦC31-integration followed by Cre/FLP excision Phase I: FC31-mediated integration. Docking strains were co-injected with the donor construct and capped FC31 integrase mRNA for FC31-mediated integration.
Aedes aegypti. The G0 Ae. aegypti OX4476 docking strains injected with OX4714 were pooled and minimum recombination efficiency calculated (as previously described), Table 3.
Canonical FC31 integration of OX4714 into Ae. aegypti OX4476C and OX4476F lines (Fig 2) was observed by the presence of the eye-specific red and blue fluoresnent markers and was confirmed by positive PCR amplification, and subsequent sequencing of attL and attR junctions (Fig B in S1 File). FC31 integration of OX4714 into OX4476H was also confirmed (data not shown), however the integrated line produced very few offspring and was later lost.
FC31-mediated insertions do not absolutely require an attP site; even if one is present in the genome a proportion of insertions are found at other sites, so-called pseudo-attP sites [45]. Recombination of the donor construct's attB site with endogenous sites that have little sequence identity with attP has been observed in Ae. aegypti, D. melanogaster, and mammalian cells [4,6,45,46]. Such non-attP insertions of OX4714 were observed in all injected Ae. aegypti OX4476 lines (H, C and F), indicated by the absence of the attL and attR junctions that are characteristic of FC31 integration (data not shown). These pseudo-attP insertions were discarded and are not included in the integration efficiency calculations.
Plutella xylostella. A total of 1082 embryos from the docking strain OX4540A were injected with OX4580 and FC31-integrase mRNA; of these, 332 individuals survived to pupation. Crosses to wild-type were set up at the pupal stage. Their progeny were screened for ZsGreen  expression (carried by OX4580). Out of 91 crosses, one generated individuals expressing ZsGreen. These were used to establish line OX4580[4540A] (Fig 3). The minimum calculated integration efficiency was 0.30%. Similar integration efficiencies (0.16% and 0.39%) for FC31 integration in P. xylostella were obtained previously in our lab (data not shown). Canonical FC31 integration of OX4580 into P. xylostella OX4540A was confirmed by positive amplification and subsequent sequencing of attL and attR recombination junctions (see Fig C in S1 File).
Phase II: FLP-and Cre-mediated excision. As a result of the FC31 integration, the donor plasmid backbone and the marker cassette from the 'docking' construct in the Ae. aegypti OX4476[OX4714] strains F and C, and the P. xylostella OX4540A[OX4580] strain, are flanked by loxP and FRT sites in the same orientation. In the presence of Cre or FLP, excision of the flanked sequences, and thus loss of the docking-construct's marker, was expected.
Aedes aegypti. To test excision efficiencies in Ae. aegypti, OX4476C[OX4714] and OX4476F [OX4714] lines were injected with FLP or Cre. Transgenic survivors were crossed to wild-type in small pools (2-5 injected G0s per pool). Individuals bearing excision events were identified through loss of the AmCyan marker (Fig 2). As anticipated, the number of false-positive excision events identified in the Ae. aegypti strains by fluorescence was high due to difficulty in correctly identifying AmCyan-positive (blue-eyed) individuals. PCR amplification across the excision junction followed by sequencing confirmed these excision events (see Fig B in S1 File). Excision did not appear to affect DsRed2 expression. For line OX4476F[OX4714], minimum excision efficiencies were estimated at 2.4% using Cre and 1.3% using FLP, whereas no excision was detected in line OX4476C[OX4714] ( Table 4).
P. xylostella. Individuals bearing excision events were identified through loss of the DsRed2 marker (Fig 3). PCR amplification across the excision junction (Fig C in S1 File) followed by sequencing confirmed these excision events. The minimum Cre excision frequency was higher in P. xylostella OX4540A[OX4580] than in any of the Ae. aegypti lines, with 90% (75/83) of the G0 survivor pools producing at least one larva showing excision using the same Cre recombinase mRNA source as used in Ae. aegypti (Table 4). The excision frequencies were calculated per G0 family as the proportion of G1 offspring in which excision had occurred. In the 75 G0 families, excision occurred at frequencies between 0.88-100%.
Similarly, the same batch of FLP recombinase mRNA was used to inject P. xylostella and Ae. aegypti, albeit at different times and by different people. As with Cre, a higher minimum FLPexcision frequency was observed in P. xylostella. All 63 of the Go survivor pools produced at least one larva showing excision (Table 4). In these 63 families, excision occurred at frequencies of between 0.47%-100%.
In conclusion, full iRMCE was achieved in the two insect species; successful FC31 integration was followed by Cre excision and also by FLP excision in Ae. aegypti and in P. xylostella.

Discussion
We successfully demonstrated two cassette exchange systems (FC31-RMCE and iRMCE) for the reliable and targeted transformation of insect species. We chose the Aedes aegypti mosquito and the diamondback moth (Plutella xylostella) due to their importance to agriculture and public health. This paper presents the first reported examples of RMCE in mosquitoes and in the pest moth P. xylostella. Furthermore, this is the first time that all three recombinases have been shown to work in P. xylostella, and that FLP has been shown to work in Ae. aegypti. Previous attempts at inducing FLP-mediated excision in Ae. aegypti were unsuccessful and included the use of a helper plasmid containing FLP under the control of the D. melanogaster hsp70 gene's promoter [47]. It is unclear whether the T7 3'UTR or the use of FLP mRNA in this study aided the successful FLP-mediated excision.
We further showed that iRMCE docking lines Ae. aegypti OX4476C and OX4476F, and P. xylostella OX4540A, as well as FC31-RMCE docking lines Ae. aegypti OX4372A and OX4372I, all carry functional and accessible attP-docking sites that can be used to integrate a panel of genes into the insect's genome. Further refinements to RMCE systems could also include incorporating methods to stabilise the piggyBac-based docking construct [48][49][50][51][52] to circumvent remobilisation risks for insect species in which remobilisation is known to occur, such as Anopheles stephensi [53].
The use of a single round of embryo injections makes the FC31-RMCE system less laborious for engineering transgenic insects. However, integration of transgenes in the first round, and subsequent excision of extraneous sequences in a separate step, as in iRMCE, is an attractive tool. Carefully designed donor cassettes could be used to compare transgenes in the presence or absence of additional elements that can later be excised in a separate step from the specific recombination sites. For example, a lethal transgene can be stably inserted into an insect genome if it contains its genetic repressor in the construct. Subsequently, this genetic repressor can be excised if flanked by two FRT or loxP sites. iRMCE worked well in two Orders of insects, indicating the robustness of the system. However, several elements may benefit from optimisation. For example, improved integration efficiency of FC31 integrase with the addition of a nuclear localised signal to the C-terminus (NLS-C) was shown in mammalian cells [54] and D. melanogaster [55], though not in P. xylostella (data not shown). Alternatively, the use of FC31 'self-docking' strains drastically improved the efficiency of the FC31-att system in Anopheles gambiae [55,56], and the use of hsp70 gene promoter and heat-shocking embryos can induce high Cre-mediated excision frequencies in D. melanogaster [57] and Ae. aegypti [47], but failed to induce FLP-mediated excision in Ae. aegypti [47].
In Ae. aegypti, FC31-att integration resulted in reduced expression of the docking cassette's fluorescent markers, consistent with previous observations in Aedes albopictus [8]. This effect may be due to transcriptional interference [58,59] or changes in chromatin structure resulting in polymerase interference [60,61], and indicates that even a modest change at the same site can affect transgene expression levels.
The variation in FC31 integration and FLP/Cre-mediated excision frequencies between docking lines carrying the same engineered molecular constructs suggests that the structure of flanking genomic sequences also affects the efficiency of these systems. Previous work in Ae. aegypti ('Aae') strains were injected with Cre and FLP in parallel for direct comparison. The number of Ae. aegypti pools with progeny apparently lacking the AmCyan marker (due to weak fluorescence) is shown ('marker loss'); this was not a problem with the P. xylostella ('Pxy') strain so values in this column reflect true marker loss. Excision events (E) were confirmed by PCR and sequencing, and the minimum calculated excision frequencies (Efficiency) are shown as a percentage. doi:10.1371/journal.pone.0121097.t004 Site-Specific Cassette Exchange Systems in Insect Species D. melanogaster [55], Ae. aegypti [6] and Ae. albopictus [8] also suggests that the efficiency of FC31-att integration is dependent on the chromosomal location of the attachment sites, though the precise molecular basis is not yet clear. The ability to directly compare and characterise components (such as lethal, suppressor or refractory effector elements) facilitates the fine-tuning of expressed phenotypes. This minimises the time-consuming nature of transforming and characterising transgenic insects using piggyBac-based transposons as otherwise a panel of insertion lines needs to be generated and analysed for each construct to understand the 'average' behaviour. However, without an effective cassette exchange system, organisms engineered using site-specific serine-integrases such as FC31-att will carry plasmid backbone sequences, including antibiotic resistance genes, introduced as part of the integration mechanism. This is undesirable, especially for strains potentially intended for field release.
Use of meganucleases such as TALENs and CRISPR/Cas9 [62] may allow targeted integration of a molecular construct at any genomic locus without inserting the extraneous plasmid sequences. Since integration efficiency may be low, insertion of an attP site may be attractive, and facilitate re-use of the insertion site. Targeted integration using a meganuclease was recently demonstrated in An. gambiae [63], albeit using a transgenic rather than endogenous target site. Though such sites, or pre-existing attP insertions, may not have the adjacent loxP or FRT sites, unwanted parts of the attB plasmid may still be removed post-integration if flanked by paired loxP or FRT sites.
A key tenet of synthetic biology is the use of standardised, characterised, modular components. One overlooked element of this is the insertion site [52]. The ability to re-use insertion sites will encourage the development of a set of insertions sites of known properties, with each further use of this set contributing further to the dataset. Desirable properties for an insertion site vary somewhat between applications, but baseline characterisation may include assessment of fitness, effects on life history parameters such as size, development time, longevity, male mating competiveness and female fecundity and fertility, as well as molecular characterisation. Pre-selection of insertion sites with appropriate characteristics in this regard would significantly facilitate the development of functional strains, including for potential field use.