W::Neo: A Novel Dual-Selection Marker for High Efficiency Gene Targeting in Drosophila

We have recently developed a so-called genomic engineering approach that allows for directed, efficient and versatile modifications of Drosophila genome by combining the homologous recombination (HR)-based gene targeting with site-specific DNA integration. In genomic engineering and several similar approaches, a “founder” knock-out line must be generated first through HR-based gene targeting, which can still be a potentially time and resource intensive process. To significantly improve the efficiency and success rate of HR-based gene targeting in Drosophila, we have generated a new dual-selection marker termed W::Neo, which is a direct fusion between proteins of eye color marker White (W) and neomycin resistance (Neo). In HR-based gene targeting experiments, mutants carrying W::Neo as the selection marker can be enriched as much as fifty times by taking advantage of the antibiotic selection in Drosophila larvae. We have successfully carried out three independent gene targeting experiments using the W::Neo to generate genomic engineering founder knock-out lines in Drosophila.


Introduction
We have recently developed a new approach termed genomic engineering that combines the gene targeting with phage integrase QC31-mediatd DNA integration for the purpose of directed, efficient and versatile modifications of endogenous genomic loci in Drosophila [1,2]. Genomic engineering is a two-step process. First, a ''founder'' knock-out is generated by homologous recombination (HR)-based gene targeting that deletes the target locus and effectively replaces it with a QC31-attP integration site. Second, the target locus can then be modified into virtually any desirable knock-in alleles through QC31-mediated integration of corresponding DNA constructs into the founder line [1,2]. We have also developed an additional integrase system for making sophisticated knock-in mutants by successive and targeted DNA integrations in genomic engineering [3]. Although generating novel knock-in alleles through site-specific DNA integration is extremely efficient compared to the HR-based knock-in/knock-out process, in genomic engineering and several similar approaches [1,2,4,5,6] a founder knock-out line must be first generated through HR-based targeting. In Drosophila, the frequency of HR for a given target locus can vary from ,10 22 to ,10 26 , i.e. an approximately 10,000-fold difference [2,7]. For target loci that are of ,10 24 HR frequency, targeting experiments can be highly time and labor intensive. Therefore, more efficient and reliable gene targeting remains crucial for approaches like genomic engineering.
General HR-based gene targeting in Drosophila [8,9] requires several rounds of genetic crosses including targeting cross, screening cross, and mapping cross ( Figure 1A) [7].
Transgenic flies of targeting construct were first generated to carry the donor DNA as a chromosomal insertion (''P[donor]'') flanked by FRT sites. To initiate the homologous recombination, the donor DNA in P[donor]/hs-FLP hs-I-SceI of targeting cross progeny is excised and linearized by heatshock-induced expression of Flipase (FLP) and the restriction enzyme I-SceI. In screening cross, heatshocked P[donor]/hs-FLP hs-I-SceI targeting females are crossed with w [2] males so potential targeting mutant progeny may be recovered based on the dominant w+ marker (i.e. red eye). Mapping cross will be used to further genetically map and confirm the targeting mutants. In order to improve the scalability and throughput in these major genetic crosses, we have in the past developed several measures such as optimizing targeting vectors and hs-FLP hs-I-SceI stocks and introducing a UAS-Rpr negative selection marker ( Figure 1A) [7]. These improvements have already yielded a high success rate for a number of targeting experiments [2,7]. Nonetheless, for targeting experiments of ,10 24 HR frequency, .10 5 progeny from screening cross have to be screened visually based on eye color marker w+. This time and labor intensive process directly limits the scale of targeting experiments.
To solve this problem, increasing the HR frequency by the means such as target-specific zinc finger nuclease (ZFN) likely presents one of the most promising approaches [10]. Nonetheless at present target-specific ZFNs or similar nucleases require extensive testing and refining efforts that to large degree may offset the benefits of increased HR frequency [11]. As an alternative, more efficient screening methods can also significantly increase the success rate of gene targeting. To this end, we developed an approach to directly enrich the targeting mutants by introducing a dominant selection marker Neo [12] in addition to the w+.

Results and Discussion
We took advantage of the well established fact that Drosophila larvae are highly sensitive to G418 which is a drug related to neomycin and karamycin, but can be made resistant by the expression of neomycin resistance gene (Neo) [12]. By making candidate mutants neomycin-resistant (Neo+), G418 can be used to directly eliminate the vast number of screening cross progeny carrying no targeting events ( Figure 1A). This approach provides several advantages. First, G418 can be easily added to the fly food, thus is fully compatible with the current genetic cross schemes of gene targeting. Second, G418-sensitivity in Drosophila larvae is dosage dependent [12]. By administrating G418 at pre-determined concentrations it is possible to eliminate a large percentage (e.g. 90%-99%) but not all of the larvae, minimizing the risk of killing Neo+ targeting mutants while at the same time leaving enough number of survival larvae to ensure healthy growing conditions.
Although Neo+ would be an effective marker for enriching targeting mutants, w+ is still the most convenient marker for genetic mapping. To incorporate both w+ and Neo+ into the targeting mutants, we made a white::Neo (W::Neo) gene encoding a chimeric protein in which Neo is directly fused to the C-terminus of W+. This design also minimized the size of the w+ Neo+ dual selection marker for potentially more efficient molecular cloning, donor DNA excision and HR. To test the effectiveness of W::Neo for being both w+ and Neo+, we first generated a pKIKO-WN vector by replacing the w+ in an older targeting vector pKIKO [7] with W::Neo. Through standard P-element based transgenesis, we obtained several w+ transgenic lines of pKIKO-WN showing that W::Neo functioned as a normal w+ marker for producing red eye flies (see Figure 2A). We also picked one of the pKIKO-WN lines and confirmed that its w+ progeny showed clear resistance to neomycin compared to their w [2] siblings carrying no pKIKO-WN (Table S1). To test the effectiveness of W:Neo marker in targeting experiments, we carried out three ends-out (replacement) gene targeting experiments using W::Neo as a dual selection marker (Table 1). We first modified the pRK2 based gene targeting constructs of dArf6 [7] by replacing the w+ with W::Neo. We obtained 16 w+ transgenic donor lines by injecting 1150 embryos. As shown in Table S2, all but three lines showed clear resistance to G418 at 0.20 mg/ml concentration. The three lines that showed reduced resistance to G418 also showed failures in FRT or Cremediated excisions and dramatically reduced effectiveness of UAS-Rpr negative selection marker, suggesting these components were likely damaged during transgenic insertion. We picked line#22 (pArf6GX22) to carry out the targeting experiment. Based on the experiments in Figure 2B, we determined that 0.20 mg/ml should be the optimal G418 concentration that eliminates .90% of w [2] sibling larvae with apparently no effect on the survival of pArf6GX22/+ larvae.
Using our improved targeting stocks [7], we were able to collect 20,000 targeting females of pArf6GX22/hs-FLP hs-I-SceI from targeting crosses (See Figure 1A). To set up the screening cross, 12,000 targeting females were crossed with w; Gal4 4-77[w2] [7]. The Gal4 drives the expression of UAS-Rpr negative selection marker that eliminates .96% of false positives [7]. Half of the cross population were grown on normal food without G418. As previously reported we screened ,700,000 progeny and recovered five targeting mutants [7] (Table 2). Another half of the cross population containing 6,000 targeting females were grown on food containing 0.20 mg/ml G418. Based on the tests conducted in Table S3, we set up the crosses in G418 bottles with 30 females per bottle. As expected, G418 drastically reduced the number of progeny produced. In total we collected and screened only ,67,000 flies but recovered 23 targeting mutants ( Table 2). Extrapolating from such data, using W::Neo marker with G418 selection we enriched the targeting mutant frequency from 5/ (7610 5 ) to 23/(6.7610 4 ), an enrichment of approximately 48 times. The mere 67,000 flies we screened were effectively equivalent to .3610 6 screening cross progeny without G418 selection.
We then carried out two new targeting experiments against the Dscam locus using the pGX-attP-WN targeting vector ( Figure 1B). Dscam encodes a neuronal adhesion molecule of extraordinary diversity through alternative splicing ( Figure 3A) [13]. Based on the genomic engineering approach, we targeted the deletions of exon#4 and #17 to generate two different founder knock-out lines designated as Dscam-N and Dscam-C, respectively. All ten Dscam-N and three Dscam-C transgenic donor lines were resistant to at least 0.25 mg/ml G418 (data not shown). For the donor lines used for carrying out the targeting experiments, pDscam-NGX113 showed G418-resistance similar to pArf6GX22 whereas pDscam-CGX1 appeared to be sensitive to ,0.50 mg/ml G418 ( Figure 2B).  For the Dscam-N targeting, we set up a screening cross using 16,000 targeting females in 800 vials with normal food (i.e. 20 females per vial). After three days we transferred flies to bottles of ,0.20 mg/ml G418 food, at the density of 160 females per bottle. The flies were transferred to new G418 bottles every three to four days. The normal food culture yielded ,1.6610 5 progeny and we recovered 23 targeting candidates (Table 2). In contrast, G418 bottles yielded 3.3610 4 progeny but we recovered 162 targeting mutants. The Dscam-C targeting was carried out similarly. We recovered 3 targeting mutants from 1.9610 5 progeny yielded from normal food, and 3 from ,2.2610 4 progeny yielded from G418 food ( Table 2). Based on screening the progeny from normal food, the HR frequency for Dscam-N targeting can be estimated is ,1.4610 24 , while for Dscam-C is ,1.6610 25 . Overall, the enrichment of targeting mutants by G418 selection can be roughly estimated as 34 and 9 times in our Dscam-N and Dscam-C targeting experiments, respectively. These numbers are likely underestimated due to the fact we grew G418 bottles under extremely overcrowded conditions of 160 females per bottle due to constrained incubator space at the time of experiments. As expected, both Dscam-N and Dscam-C mutants are lethal and their lethality can be rescued by integrating back the deleted fragment of gDNA into their corresponding knock-out founder lines (data not shown).
It should be noted that the hsp70 promoter which drives the W::Neo expression in targeting mutants is not transcriptionally insulated and although its expression in eyes are boosted with an eye-specific GMR enhancer [7] (Figure 1B) its expression levels in other larval tissues still could suffer from chromosomal locationeffects. Therefore, one potential caveat with G418 selection is that it may be difficult to know the actual strength of G418 resistance of a given targeting mutant. To investigate this issue, we systematically and quantitatively measured the G418 resistance of dArf6, Dscam-N and Dscam-C transgenic donor lines and targeting mutants. As shown in Figure 2B, all the W::Neo lines showed resistance well above 0.20 mg/ml G418, better than the common FRT lines carrying hs-Neo [14]. Assuming the results from the three target loci are representative, the risk of killing the real mutants should be very low at the 0.20 mg/ml G418 concentration that we used. In addition, for a given target locus there appears to be a good correlation between the transgenic donor line and the targeting mutants in terms of their G418-resistances ( Figure 2B), consistent with the fact that in both lines the W::Neo is flanked by the same 59 and 39 gDNA which likely influence the expression of W::Neo the most. Therefore, by carefully testing the G418-resistenace in transgenic donor lines, it is possible to estimate the strength of G418-resistance of the future targeting mutants in order to optimize the G418 concentration for each individual targeting experiment. In general, 0.20 mg/ml of G418 seems to be working well in our experiments.
One practice we would recommend is to set up the screening crosses on normal food first, and transferring them to G418 food after one or two days. This method apparently gives healthier crosses on G418 food. In addition, the progeny from normal food come out earlier and can be used for a small to medium scale (,10 4 -10 5 flies) screening first. If no or not enough number of candidates are recovered, the target locus might be of low HR frequency and one may continue to screen the G418-selected progeny. In addition, despite that the hsp70 promoter used in pGX-attP-WN is constitutively expressed, its expression level can still be greatly increased (up to one hundred-fold) by heat-shock [12]. Although we did not carry out heat-shock treatments in screening crosses, it can be easily adapted into the protocol. Finally, for targeting loci that may severely represses hsp70 promoter, we are considering making modified targeting constructs that may feature stronger or insulated promoters.
In summary, we report here the successful applications of a novel w+/Neo+ dual selection marker that may effectively enrich the targeting mutants up to fifty times with the help of G418selection. Our new pGX-attP-WN targeting vector could significantly facilitate the large scale targeting experiments, making target loci of ,10 26 HR frequency much more experimentally accessible. Besides gene targeting, the W::Neo marker should be useful in routine Drosophila genetic crosses when both w+ and Neo+ are desirable for selecting a particular genotype.
Since all female candidates were discarded in targeting experiments, the adjusted HR frequency should be twice higher than listed here. c.
Screening crosses were set up on the normal food first, then transferred to G418 food after two days.

DNA Constructs
The W::Neo marker was made by fusing the Neo coding sequence to the C-terminus of W+ in pKIKO vector [7] through overlapping PCR. Cloned W::Neo fragments were sequenced to ensure error-free PCR. pGX-attP-WN was made by replacing the coding sequence of w+ in pGX-attP with W::Neo. Targeting construct of dArf6 was described previously [7]. Molecular cloning of targeting constructs of Dscam-N and Dscam-C was carried out according to the protocols described in Huang et al [7]. Primers used for making targeting constructs are listed in Table S4. We used ''cis-analyst'' tool at http://www.fruitfly.org/seq_tools/ other.html to compare genomic sequences between Drosophila melanoganster and Drosophila pseudoobscura to identify apparently nonconserved non-coding regions for positioning the QC31-attP and loxP sites in the target locus.

Transgenesis and ends-out targeting
All transgenic flies were created using w 1118 stocks via the standard P-elements-based transgenic protocol. Most fly cultures and crosses were carried out at room temperature (,22uC) or 25uC. Ends-out gene targeting and PCR-verifications of targeting candidates were carried out as described in Huang et al [7]. Primers used for PCR verifications as shown in Figure 3B,C are listed in Table S4.

G418 treatment and tests
G418 (from Fisher Scientific) was directly added to microwavemelted fly food at ,50uC as described [12]. All G418 concentrations reported here were effective concentrations based on the manufacture specifications. To quantitatively measure the G418 resistance, males from the following stocks were crossed with virgin females of: w 1118 (wild type control); y w; pArf6 GX22 /TM3 (transgenic donor line used for dArf6 targeting); y w; pDscam-N GX113 /TM3 (transgenic donor line used for Dscam-N targeting); y w; pDscam-C GX1 /TM3 (transgenic donor line used for Dscam-C targeting); y w; dArf6 GX16[w+] /CyO (dArf6 founder knock-out line); y w; Dscam-N GX07[w+] /CyO (Dscam-N founder knock-out line); y w; Dscam-C GX101[w+] /CyO (Dscam-C founder knock-out line); y w; FRT-42D ubi-GFP NLS /CyO; For each cross, embryos were collected under 18uC for 24 hours and were split evenly into two vials containinig normal food (i.e. 0 mg/ml G418) and food of specified G418 concentration, respectively. On average approximately 200 embryos were placed in each vial. Adult w+ and w [2] progeny were counted from each vial within 18 days under 25uC. For either w [2] or w+ progeny, their survival rate in G418 selection is calculated as the percentage of (# from G418 food)/(# from normal food). Each test was carried out in at least triplicates. To calculate the w [2] survival rates in Figure 2B, we averaged the survival rates of all TM3/+ and CyO/+ cross progeny at a given G418 concentration.

Supporting Information
Table S1 W::Neo marker confers G418 resistance in transgenic flies. * In each vial, six males of y w/Y; pKIKO-WN#2/CyO were crossed with six y w; Pin/CyO virgin females and were grown for five days. Progeny were scored based on w+ marker. In the absence of G418 selection one third of the progeny were expected to be w [2] (i.e. w [2] /w+ = 50%). ** These experiments were repeated three times. In experiment #2 only two y w; pKIKO-WN#2/CyO females were used for each cross, hence the small number of progeny. Avg: Average. (DOC) Table S2 G418-resistance test of dArf6 transgenic donor lines. a: Tests were carried out similarly as in Table S1. b, c: hs-FLP and hs-Cre stocks used here constitutively express FLPase or Cre recombinase. In hs-FLP and hs-Cre crosses, constitutively expressed FLPase or Cre will excise the donor DNA, resulting in visible eye color variegation in the cross progeny. The degree of eye color variegation due to the loss of w+ was compared by estimating the percentage of white area in each eye. 0% white area suggests likely at least one of the two FRT or loxP sites were damaged in the transgenic donor DNA insertion. d: Gal4 2-21 drives UAS-Rpr expression of the transgenic donor DNA that results in 100% lethality in the cross progeny. Significantly reduced lethality suggests either damaged transgenic donor DNA insertion or severe repression of UAS-Rpr expression due to chromosomal location. e: These three lines likely have damaged transgenic donor DNA insertion based on their test results with hs-FLP, hs-Cre and Gal4 2-21 . They also show much higher percentage of w [2] progeny on 0.25 mg/ml G418, presumably due to the fact that reduced G418-resistance in the w+ progeny allowed better survival of w [2] siblings. f: The transgenic donor line used for targeting. n.d.: Not done Chr.: chromosomal location of the transgenic insertion. (DOC) Table S3 Optimal culture density for G418 selection in dArf6 targeting. Crosses were first set up in vials, and were then transferred to G418 bottles after one or two days. It appears that between 20 to 40 females per bottle, the yield of w+ candidates per 100 targeting females remains relatively stable. We decided that crosses of 30 targeting females per bottle appears to be a good compromise between achieving the maximum recovery of w+ candidates and minimizing the number of G418 bottles. (DOC) Table S4 Primers used for the generation and verification of Dscam-N and Dscam-C founder knock-lines. *: Primers for PCR verifications of dArf6 founder lines were in Huang et al [7]. w+: PCR for verifying founder knock-out lines that contain the w+ marker. w [2] : PCR for verifying founder lines that had their w+ marker removed by loxP recombination. n/a: not applicable. (DOC)