Abstract
We previously described a method for targeted homologous recombination at the yellow gene of Drosophila melanogaster. Because only a single gene was targeted, further work was required to show whether the method could be extended to become generally useful for gene modification in Drosophila. We have now used this method to produce a knockout of the autosomal pugilist gene by homologous recombination between the endogenous locus and a 2.5-kb DNA fragment. This was accomplished solely by tracking the altered genetic linkage of an arbitrary marker gene as the targeting DNA moved from chromosome X or 2 to chromosome 3. The results indicate that this method of homologous recombination is likely to be generally useful for Drosophila gene targeting.
TO carry out gene targeting in Drosophila (Rong and Golic 2000) a donor construct carrying DNA from the gene to be targeted is randomly inserted into the genome by P-element-mediated transformation. Then, a site-specific recombinase (FLP) and a site-specific endonuclease (I-SceI) are used to generate, in vivo, an extrachromosomal DNA molecule that carries a double-strand break (DSB) within the gene of interest. The presence of this DSB stimulates homologous recombination between the excised donor and the homologous chromosomal target locus (Figure 1). Our initial experiments were based on recombination events that rescued a mutant allele of the Drosophila yellow gene. Homologous recombination between the donor and target locus was frequent, occurring approximately once in every 500 female gametes. We recovered several classes of recombinant at the target locus that included allelic substitutions and integration of donor DNA. The promise of this technique is that, starting with only a DNA sequence, animals with specific mutations can be generated and their phenotypes studied to deduce the normal function of the mutated gene. It is therefore important to know whether this method can be generally applied for targeted gene modification in Drosophila.
One issue of concern is whether genes that are not located near telomeres (as is yellow) can be targeted. Engels (2000) raised the possibility that the targeted recombination that we observed relied on a type of DNA repair termed break-induced replication (BIR; Malkovaet al. 1996). He proposed that a single one-ended homologous exchange occurred between the chromosomal target locus and the excised and cut donor molecule, leaving the recombinant chromosome with a truncated terminus. To be recovered as a viable product this chromosome would be repaired by BIR, wherein the broken terminus invades the homolog, prompting unscheduled replication to the end of the chromosome (Figure 2). Since the yellow gene that we targeted lies only ∼110 kb from the X chromosome telomere, it is not unreasonable to imagine that a chromosome break at this location could be repaired by replication to the end of the chromosome. Additionally, yellow targeting was much more efficient in the female germline (with two X chromosomes) than the male germline (with one X), and Engels’ BIR model, wherein repair of a oneended recombination event relies on replication templated from a homolog, provides an explanation for this difference. As a practical matter, the most significant implication of Engels’ model is that if targeting must involve BIR, then it is likely that only genes situated near telomeres can be successfully targeted because of the requirement for continuous replication to the end of the chromosome.
A second question is whether methods can be developed to produce targeted mutation and whether this can be done without relying on knowledge of the phenotype produced by mutation in the target locus (Anonymous 2000). Homologous recombination reactions similar to that which we used at yellow have been modified to generate targeted mutations in Esherichia coli, in yeast, and in mice (Shortleet al. 1982; Shen and Huang 1986; Thomas and Capecchi 1987). It is likely that similar methodology can be used in Drosophila. The work presented here demonstrates the use of one such method in Drosophila for targeted gene modification.
MATERIALS AND METHODS
DNA constructs: A 2.5-kb NruI-PstI fragment of the pugilist (pug) genomic region was cloned into pBS(Not), which is a derivative of pBluescript II KS (+) from Stratagene. In pBS(Not), the ApaI site in the multiple cloning site was changed to a NotI site by site-directed mutagenesis. An I-SceI recognition site was synthesized as two oligos: Apa-I-plus, 5′-gctagggataacagggtaatggcc; and Apa-I-minus, 5′-attaccctgttatcc ctagcggcc. They were allowed to anneal and were cloned into the unique ApaI site in the 2.5-kb pug fragment. The same pug fragment, now flanked by NotI sites, was cloned into the NotI site in the P-element vector pP[>whs·N>], which is a modified form of pP[>whs>] (Golic and Lindquist 1989), to produce the donor construct pP[>whs·MTHΔ>] for pug targeting. This construct was transformed into flies by standard P-element transformation.
—The targeting scheme. The donor construct is diagrammed at the top. FLP-mediated excision and I-SceI-mediated cutting produce the extrachromosomal targeting molecule shown. A 2.5-kb internal fragment of the pug gene was used to construct the P[>whs·MTHΔ>] donor. It lacks 5′ regulatory sequences and the first, part of the fourth, and the fifth exons. This targeting molecule is expected to recombine with the endogenous pug locus, as shown, to produce a tandem duplication. Arrowheads at the donor site represent P-element ends. The direction of pug transcription is from left to right. The restriction sites and the location of the probe used in Southern blot analysis are shown with sizes indicated in kilobases. R, EcoRI; B, BamHI.
Molecular analyses of targeting events: Southern blot analyses were performed as described (Rong and Golic 1998). To verify the restoration of the ApaI site in the targeted pug allele, primers proxH3 (Rong and Golic 1998) and pugseq1.rev (5′-ttcagacgctccaacactga-3′) were used. PCR reactions were done using genomic DNA from flies homozygous for the targeted pug alleles. PCR products were digested with ApaI and run on a gel along with uncut controls.
Fly stocks and crosses: Description of stocks not provided here can be found at http://flybase.bio.indiana.edu. The deficiency that deletes the pug+ region is in the stock Df(3R)cu/TM6B, Tb, provided by the Bloomington Stock Center. The stock v; pugDrv18 (Rong and Golic 1998) was used to provide the pug null allele. For targeting we used insertions of the P[>whs·MTHΔ>] donor on either chromosome X or 2. We used either a chromosome 2 or 3 that carries both the 70FLP and 70I-SceI transgenes. Crosses generated flies with a single copy donor element and 70FLP and 70I-SceI. All flies carried either the w1 or the w1118 null alleles on their X chromosomes. For most screening crosses, the heat-shocked flies were mated to y1 w1; Sb/TM6, Ubx flies, and the whs gene was mapped with respect to chromosome 3 in all progeny with pigmented eyes. In a small number of cases the heat-shocked flies were mated to y1 w1 70FLP3F; Sb/TM6, Ubx and the offspring were screened for a white+ nonmosaic phenotype, which could indicate that the whs gene was no longer flanked by FLP recombination targets (FRTs), as expected for a targeting event. Positives were retested for FLP-mediated white mosaicism and mapped with respect to chromosome 3. The reported numbers are corrected for the fact that in some crosses only half the tested flies had the appropriate genotype.
RESULTS
Insertional, or ends-in, targeting of an internal gene fragment could conceivably generate a targeted gene knockout. The expected product would be a tandem partial duplication of the target gene, with both copies defective because they each lack a portion of the gene (as shown in Figure 1). We tested this targeting scheme by using it to generate a null mutation in the pug gene, which encodes a homolog of the trifunctional form of the enzyme methylenetetrahydrofolate dehydrogenase. Null mutations in this gene produce a recessive eye color defect (Rong and Golic 1998). The gene is located at 86C on the right arm of chromosome 3, ∼20 Mbp from the nearest telomere. A 2.5-kb internal fragment of pug was engineered to carry a recognition site for the I-SceI endonuclease. This pug fragment was then placed into a modified form of the P-element vector P[>whs>] (Golic and Lindquist 1989) to produce the donor construct for pug targeting (Figure 1).
The whs gene of this vector is a hypomorphic allele of the white eye color gene and serves as a w+ marker for transformation and for the recovery of targeting events. The pug gene fragment and whs are flanked by direct repeats of the FRT. When FLP mediates recombination between these FRTs, both the whs gene and the partial pug gene are excised from the chromosome on a circular DNA molecule. Subsequent I-SceI cutting generates the recombinogenic pug targeting molecule. Homologous recombination between this molecule and the resident pug locus is expected to carry the whs gene into pug (Figure 1). If the donor is located on a chromosome other than chromosome 3, a targeting event can be recognized by the altered genetic linkage of whs that will result. The modified pug allele produced by targeted insertion of this donor molecule is expected to be mutant.
—Break-induced replication model for yellow targeting (adapted from Engels 2000). This model proposes that in a y1 female, the extrachromosomal y+ donor pairs with a chromosomal y1 allele, and DNA to the left of the break recombines with the chromosome (A). This gives rise to a truncated X chromosome missing all DNA distal to yellow. The reciprocal product is an acentric fragment, which is lost as a result of exonucleolytic degradation or failure to segregate when the cell divides. It is then proposed that the broken end of the centric X chromosome invades the homolog at y, initiating DNA replication toward the telomere (B and C). This produces two intact X’s, one of which carries a targeted y allele. The resulting cell should be homogyzous for all the alleles distal to y, exemplified by both X telomeres having the same shade (D). The yellow gene is represented as a rectangle, the rest of the chromosome as a single line. Telomeres are represented by triangles; centromeres are indicated as circles. The asterisk indicates the site of the point mutation in y1. Left-facing arrows indicate leading-strand DNA synthesis. Right-facing arrows indicate lagging-strand synthesis.
—Cytological mapping of targeted insertion. Chromosomes from flies homozygous for a targeted pug allele were probed with labeled white gene DNA (random-primed pP[>whs>] DNA). Two signals were detected: one at the 3C locus, which is the endogenous white locus, and the other at 86C, which is the targeted pug locus. Chromosome landmarks are indicated.
To carry out targeting we produced flies that had the transgenic pug donor construct, on either the X chromosome or chromosome 2, and the heat-inducible 70FLP and 70I-SceI transgenes. These flies were heat shocked as described (Rong and Golic 2000) and testcrossed to flies carrying the w1 mutation. The vast majority of progeny were white eyed, owing to FLP-mediated excision and loss of the FRT-flanked whs gene (Golic and Lindquist 1989). In most cases, when offspring with pigmented eyes were observed the whs gene was mapped to detect its mobilization to chromosome 3. A small number of crosses were screened with an alternative technique that is explained in the discussion.
A total of 455 vials (generating 100-150 progeny per vial) were set up to screen for mobilization of whs in the female germline: we recovered three independent cases of whs mobilization to chromosome 3. Chromosome in situ hybridization was used to locate the whs gene on chromosome 3. In two cases, whs was located at 86C, the normal location of pug (Figure 3). In the third case whs was located in cytological region 97 and was not characterized further.
We also tested for targeting in the male germline. We recovered no pug targeting events from 308 vials, although we did recover one example of nontargeted mobilization (to chromosome 2).
Southern blot analysis was performed on one of the targeted alleles to confirm its molecular structure. DNA from homozygous flies was digested with restriction enzymes EcoRI and BamHI and hybridized with a probe made from the 2.5-kb pug fragment (Figure 1). As expected, the wild-type 9-kb band was converted into two bands of 7 and 10 kb (Figure 4). Hybridization with a white probe detected only the 10-kb band and the endogenous white bands (not shown).
—Southern blot analysis of a pug targeting event. Genomic DNA was digested with EcoRI and BamHI and blotted to a membrane. The membrane was hybridized with a 2.5-kb pug probe (Figure 1). Lane 1, molecular markers with indicated sizes; lane 2, pug+ control showing the endogenous 9-kb band; lane 3, DNA from flies homozygous for the targeted pug allele showing, as predicted, the 7-kb and the 10-kb fragments.
To assess the phenotype of the two pug alleles produced by targeting we generated the following three genotypes: (1) flies homozygous for the targeted alleles; (2) flies that carried the targeted alleles over a pug deficiency; and (3) flies that carried the targeted alleles heterozygous with a previously isolated pug null allele (Rong and Golic 1998). For both targeting events, all three genotypes displayed the pug-null phenotype of slightly reduced pigmentation in the center of the eye (not shown). By this assay, the alleles generated by targeted integration of the donor are functional nulls.
Further analysis suggested that the pug targeting events arose by precise homologous recombination. The I-SceI cut site in the donor was inserted at a unique ApaI site in the 2.5-kb pug fragment, destroying the ApaI site in the process. The DSB repair model for recombination (Szostaket al. 1983) predicts that the DSB generated by I-SceI will be enlarged to a gap by exonuclease activity. During integration, this gap will be repaired by DNA synthesis that uses the wild-type target locus as a template. It was therefore expected that the ApaI site would be restored in the targeting process, leading to an ApaI site in each of the two pug copies. This was confirmed by Southern blot analysis of ApaI-digested genomic DNA, or by ApaI digestion of PCR-amplified DNA fragments (not shown). In summary, cytological, phenotypic, and molecular data support the conclusion that these represent cases of targeted modification of the pug locus.
DISCUSSION
The general application of targeted mutagenesis: The most important aspect of our results is the demonstration that a gene far removed from the nearest telomere can be successfully targeted. Since the pug gene is very much closer to centric heterochromatin than it is to a telomere it seems likely that most euchromatic genes will be susceptible to targeted modification by this technique.
These experiments also show that nontargeted insertions, although they do occur, are not so frequent as to be a significant nuisance. Here, the ratio of targeted to nontargeted recombinants was 2:1 in females (or 1:1 overall). In the previous yellow targeting experiments, targeted recombinants also outnumbered nontargeted events (Rong and Golic 2000). Nontargeted events that arise on chromosomes other than the target chromosome can be easily detected by mapping the whs marker gene. Southern blot analysis can be used to detect and discard nontargeted events that are linked to the target chromosome and to confirm the expected structure of targeting events.
Our results address the concerns of whether this method can be used for targeted mutagenesis. We generated a directed mutation without relying on prior knowledge of the mutant phenotype. Instead, we recovered targeting events by following the segregation of an easily visible genetic marker. In addition, our results demonstrate the production of a functional null allele, even though the gene is partially duplicated. The right-hand copy is almost certainly a null owing to the absence of upstream regulatory elements and the first exon of the pug gene. The phenotypic studies suggest that the left-hand copy is also a functional null, although it is possible that the phenotypic assay may not be sensitive to a small amount of pug activity. Even so, it is not unreasonable to imagine that, by strategically placing point mutations in the donor, the two copies would assuredly be nulls. Therefore, we believe that this targeting method provides a general tool for directed mutagenesis of the Drosophila genome.
The mechanism of targeted homologous recombination: The recovery of targeted alleles at pug does not rule out the possibility that some of the targeting events we previously recovered at yellow did arise by homologous recombination and BIR. The BIR model for yellow targeting provided an explanation for the observed female:male difference in targeting efficiency. It predicts that there should be no difference between the sexes in the efficiency of targeting an autosomal gene. In the experiments reported here, the pug targeting events recovered from females do outnumber those recovered from males, but the number of events is too few to reasonably conclude that this difference is significant, and thus cannot be used to address the BIR model.
Targeting of yellow was substantially more efficient (1 in 4 vials for the female germline; Rong and Golic 2000) than targeting of pug (1 in 228 vials). This could be taken as support of the hypothesis that the majority of yellow targeting events relied on the mechanism of BIR. According to this view, pug targeting is less efficient because of the much greater gene-telomere distance involved and the correspondingly lower probability of a replication fork sucessfully proceeding to the end of the chromosome. On the other hand, it seems at least equally likely that the explanation for this disparity owes to the different amounts of donor:target homology in the two experiments—8 kb in the yellow experiments vs. 2.5 kb in the pug targeting experiments reported here. In other targeting systems, the extent of donor:target homology does influence the frequency of homologous recombination (Shen and Huang 1986; Deng and Capecchi 1992; Dray and Gloor 1997). Further experiments (Engels 2000) are required to determine whether BIR is part of the mechanism of Drosophila gene targeting.
Targeting methodology: Altered genetic linkage was used to detect pug targeting events in most of the work reported here. The essence of this screen is the detection of marker gene movement from one chromosome to another. This works well but is time consuming since it requires mapping of the donor P element and construction of special strains so that the donor chromosome or its homolog is dominantly marked.
To speed the process of gene targeting we have begun to use an alternative method to detect targeted homologous recombination events (Figure 5). The whs marker has an easily scored cell-autonomous phenotype. The infrequent progeny of targeting crosses that exhibit the w+ phenotype arise as a result of the donor element escaping FLP-mediated excision or by mobilization of the donor to the target site (or a nontargeted site). The original donor is distinguished from cases of mobilization by the fact that, in the former case, the whs marker gene is flanked by FRTs, but in the latter it is not (see Figure 1). FLP will cause eye-color mosaicism when whs is flanked by FRTs but not when there is only a single FRT adjacent to whs. This difference can be employed to aid in the detection of gene targeting events. Flies that carry the donor insertion, 70FLP, and 70I-SceI are heat shocked and crossed to flies that carry 70FLP. The progeny of this targeting cross are heat shocked during the first few days of development to induce FLP synthesis. The mosaic or nonmosaic phenotype of the infrequent w+ flies tells whether they carry the original donor element or a targeted insertion. Molecular characterization by Southern blotting can then be used to confirm that the recovered events resulted from homologous recombination at the target locus. One of the two pug targeting events was recovered using this method.
—A rapid targeting scheme. In this example a donor element on chromosome 2 is used for targeting pug on chromosome 3. If the donor parent in G0 is heterozygous for that insertion (as indicated here) then only half of the progeny will carry the donor element. This may often be the case if the donor element has not yet been mapped. Many of the G1 flies that receive the donor element will be white-eyed owing to the high rate of somatic excision and loss, but these are still suitable for use in the targeting cross. S2CyO is a chromosome 2 balancer. HS, heat shock.
This alternative procedure significantly simplifies and speeds the process of gene targeting by allowing targeting to be carried out without knowing the location of the donor element. The process can be made even easier by using a 70FLP line that has strong constitutive expression of FLP. We have generated such a line (unpublished results). This eliminates the need for heat-shocking the progeny of the targeting cross for the mosaic/nonmosaic test.
The crosses to obtain targeting can be carried out contemporaneously with crosses to map the donor element. The purpose of mapping is to provide the information needed to remove, by crossing, the remnant donor P element after targeting has been achieved.
Acknowledgments
We thank Julie Feder for the plasmid pP[>whs·N>]. This work was supported by National Science Foundation grant MCB-9728070.
Footnotes
-
Communicating editor: R. S. Hawley
- Received October 18, 2000.
- Accepted December 4, 2000.
- Copyright © 2001 by the Genetics Society of America
