Successive and Targeted DNA Integrations in the Drosophila Genome by Bxb1 and φC31 Integrases

Abstract

At present φC31 is the only phage integrase system available for directionally regulated site-specific DNA integration in the Drosophila genome. Here we report that mycobacteriophage Bxb1 integrase also mediates targeted DNA integration in Drosophila with high specificity and efficiency. By alternately using Bxb1 and φC31, we were able to carry out multiple rounds of successive and targeted DNA integrations in our genomic engineering founder lines for the purpose of generating complex knock-in alleles.

THE serine family of phage integrases such as φC31 are highly useful due to their capability of mediating site-specific and unidirectional DNA integration in heterologous systems (Groth and Calos 2004). In the past few years, φC31-mediated site-specific DNA integration has gained wide applications in Drosophila for efficient and targeted transgenesis (Groth et al. 2004; Bateman et al. 2006; Bischof et al. 2007; Markstein et al. 2008; Ni et al. 2008). In particular, we and several other groups have developed approaches that combine φC31-mediated DNA integration with gene targeting for achieving directed and efficient modifications of endogenous genomic loci in Drosophila (Gao et al. 2008; Choi et al. 2009; Huang et al. 2009a,b; Weng et al. 2009). For example, in our genomic engineering approach, a “founder line” is first generated by homologous recombination-based gene targeting that effectively replaces the target gene with a φC31-attP (“attPC”) integration site. The target locus can then be modified into virtually any desirable knock-in alleles through φC31-mediated integration of corresponding DNA constructs into the founder line (Huang et al. 2009a,b). However, DNA integration effectively destroys the original attPC site by converting it into φC31-attR (“attRC”) and φC31-attL (“attLC”) sites (Groth and Calos 2004), preventing further DNA integrations into the target locus. Nonetheless, successive DNA integrations into a target locus can be highly desirable when making sophisticated knock-in alleles that are best done by integrating multiple constructs (Ow 2007). Although it is possible to carry out such successive DNA integrations by adding extra attPC or attBC (i.e., φC31-attB) sites on the integration construct, in practice we found that the φC31 often carried out random and promiscuous recombination among multiple attBC and/or attPC sites in Drosophila (J. Huang and Y. Hong, unpublished data), making the process highly inefficient and unreliable. Thus, an additional phage integrase is necessary for successive DNA integrations in a target locus.

Mycobacteriophage Bxb1 integrase (Ghosh et al. 2003; Nkrumah et al. 2006) is a serine integrase that has been shown capable of efficient site-specific integration in heterologous systems, including malaria, plants, and mammalian cells. In addition, characterized Bxb1-attP (“attPX”) and Bxb1-attB (“attBX”) integration sites not only are distinct from attPC and attBC (Ghosh et al. 2003; Nkrumah et al. 2006), but also are small sizes of ∼50 bp, which will leave small footprints before and after integrations. To test whether Bxb1 could mediate DNA integration in Drosophila, we made a transgenic vector pAttPX that carries a 52-bp attPX (“attPX-52”) (Figure 1A) and a removable w+ marker flanked by loxP sites. Through standard P-element transposition process, we obtained five independent host lines, all coincidently carrying the attPX-52 on the third chromosome. Two of them, the attPX-52#1^[w+] and attPX-52#3^[w+] lines, were converted to w^[−] by excising the w+ marker through Cre/loxP recombination (Figure 1B) (Materials and Methods). We then made a test integration construct, pGE-attBX-GFP, which carries the 46-bp attBX site and a UAS-GFP reporter (Figure 1B), and a construct pET11Bxb1polyA for in vitro synthesis of Bxb1 mRNA (Materials and Methods). pGE-attBX-GFP/Bxb1 mRNA mixtures were prepared and injected into the homozygous attPX-52 embryos using the same protocol of φC31-mediated DNA integration (Groth et al. 2004). We obtained 16 candidate lines from attPX-52#1^[w−] and 9 candidate lines from attPX-52#3^[w−] (Table 1). They were all mapped correctly to the third chromosome except one from each host line that was mapped to non-third chromosomes. These two non-third chromosome lines were likely due to pseudo-integrations.

Figure 1 .

Bxb1-mediated DNA integration in Drosophila. (A) Map of pAttPX (5.984 kb). pAttPX is a P-element-based transforming vector. (B) Bxb1-mediated DNA integration. w+ marker is first excised from the attPX-52^[w+] host line by Cre-mediated recombination between two flanking loxP sites. pGE-attBX-GFP plasmid is then integrated into the attPX-52^[w−] host line via Bxb1-mediated recombination between attPX and attBX. The integration converts attPX to attRX and attLX. 3′P and 5′P, 5′ and 3′ P-element sequences; w+, hsp70::white+ marker with glass multimer reporter (GMR) enhancer (Huang et al. 2008, 2009b); PX, attPX; BX, attBX; RX, attRX; LX, attLX; Amp^R, ampicillin-resistant gene.

Table 1 . Bxb1-mediated DNA integration in attPX-52 host lines and in crb-PX genomic engineering founder lines.

Host line	Location of attPX-52	DNA injected	Embryos injected	Larvae surviveda	Adults surviveda	% integration efficiencyb
attPX-52#1	Third chromosome	pGE-attBX-PC	750	500 (67)	412 (55)	3.6 (15/412)c
attPX-52#3	Third chromosome	pGE-attBX-PC	600	396 (66)	273 (46)	2.9 (8/273)c
crb-PXGE#24[w−]	Third chromosome	pGE-attBX-PC	1850	522 (28)	254 (14)	1.2 (3/254)
crb-PXGE#24[w−]	Third chromosome	pGE-attBX-G80EYC	1350	722 (53)	466 (35)	0.4 (2/466)

^aNumbers in parentheses are percentages.

^bIntegration efficiency is calculated according to Groth et al. (2004).

^cIn each of these integration experiments, we discovered a single nonspecific integrant line (excluded from the table) based on the chromosomal mapping.

We used PCR analyses to confirm that the remaining third chromosome integration lines resulted from targeted integration of pGE-attBX-GFP. As shown in Figure 1B, the recombination between attPX and attBX sites will generate two new sites, attRX (i.e., Bxb1-attR) and attLX (i.e., Bxb1-attL) in the genome (Figure 1B). A diagnostic PCR spanning over the potential attLX showed a specific PCR product of the correct size in all third chromosome candidates. We sequenced PCR products from four candidates and confirmed the expected attLX site in each one of them (supporting information, Figure S1A). Overall, attPX-52#1^[w−] and attPX-52#3^[w−] flies showed comparable integration efficiencies of 3.6 and 2.9%, respectively (Table 1). The pseudo-integration rate of Bxb1 in the Drosophila genome can be estimated at ∼6–11% (1/16 in attPX-52#1^[w−] and 1/9 in attPX-52#3^[w−], respectively; Table 1).

We then carried out successive and targeted DNA integrations in one of our genomic engineering founder lines, crb^GX#24[w−], by alternately using φC31 and Bxb1 integrases. With the help of Bxb1, we first wanted to create an attPX integration site in the wild-type crb locus to insert extra transgenic fragments that need to be closely linked with crb in our future genetic assays. To do so, we generated the construct pGE-attB-crb^rescue-PX, which carries the ∼12-kb crb genomic DNA (gDNA) that was deleted in the founder lines (Huang et al. 2009b), and an attPX-52 site located at the 3′ end of gDNA (Figure 2A). φC31-mediated integration of pGE-attB-crb^rescue-PX into crb^GX#24[w−] generated a new crb-PX^GE#24[w+] founder line (Figure 2A). Similar to the pGE-attB-crb^rescue construct that contains only the crb genomic DNA (Huang et al. 2009b), integration of pGE-attB-crb^rescue-PX fully rescued crb^GX#24[w−] to being homozygous viable, healthy, and fertile, demonstrating that the attPX-52 at the 3′ end of crb locus did not interfere with the normal expressions of host locus. More importantly, these flies no longer carry a functional attPC site, but an attPX-52 site (Figure 2A).

Figure 2 .

Successive DNA integration in genomic engineering founder lines using φC31 and Bxb1. (A) Genomic engineering founder line crb^GX#24[w−] was first converted to crb-PX^GE#24[w+] via φC31-mediated integration of pGE-attBX-crb^rescue-PX. In crb-PX^GE#24[w+], crb deletion was fully rescued and the original attPC site was destroyed (i.e., converted into attLC and attRC), while a new attPX-52 site was inserted right after the 3′ end of rescued crb locus. The w+ and vector sequences (e.g., Amp^R, etc.) were then removed to generate crb-PX^GE#24[w−], which was ready for the Bxb1-mediated integration of pGE-attBX-PC. In crb-PC^GR#24[w−], integration of pGE-attBX-PC effectively removed the attPX-52 site but added a new attPC site. Once the w+ in crb-PC^GR#24[w+] was removed, the φC31-attP site could be used for the third round of DNA integration using φC31 integrase (not depicted here). (B) crb-PX^GE#24[w−] was also used to integrate the plasmid pGE-attBX-G80EYC via Bxb1-mediated DNA integration. pGE-attBX-G80EYC does not carry the extra loxP site so the w+ in crb-80EYC^GR[w+] is not removable. 5′ and 3′, the 5′ and 3′ flanking genomic DNA of crb.

We then tested the integration efficiency of attPX-52 in the homozygous lines of crb-PX^GE#24[w−] by carrying out the second-round integrations of a small plasmid pGE-attBX-PC bearing both an attBX and an attPC site (Figure 2A). crb-PX^GE#24[w−] showed Bxb1-mediated integration efficiency of 1.2% (Table 1). The reduced efficiency of Bxb1-mediated integration in genomic engineering founder lines is consistent with what we observed with φC31-mediated integration (Huang et al. 2009b). The reduced Bxb1 integration efficiency in the crb founder line could be due to locus-specific chromosomal effects, or it is also possible that Bxb1 favors attPX sites associated with P-element insertion over the sites that are arbitrarily inserted into the genome via homologous recombination. In crb-PC^GR[w+] founder lines (Figure 2A), pGE-attBX-PC integration destroys the attBX site but simultaneously adds a new attPC site for the third-round integration to be mediated by φC31.

Finally, we carried out the Bxb1-mediated integration of the construct pGE-attBX-G80EYC into crb-PX^GE#24[w−]. The 11.3-kb pGE-attBX-G80EYC carries an ∼5-kb insert containing ey-Cre (Newsome et al. 2000) and tub-Gal80 (O’Donnell et al. 1994; Lee and Luo 1999) (Figure 2B). Likely due to its larger size, the integration efficiency of pGE-attBX-G80EYC is approximately two- to threefold lower than that of pGE-attBX-PC at 0.4% in crb-PX^GE#24[w−] (Table 1). In crb-G80EYC^GR[w+] flies (Figure 2B), integration of pGE-attBX-G80EYC placed cy-Cre and tub-Gal80 modules at the endogenous locus of crb to conveniently manipulate the expression of other transgenes when assaying the function of certain crb mutants (J. Huang and Y. Hong, unpublished data) (Figure 2B).

In summary, we confirmed that Bxb1 phage integrase can mediate efficient site-specific DNA integration in the Drosophila genome and demonstrated its application in carrying out successive DNA integrations together with φC31 to generate complex genomic engineering alleles. Currently, we are also adding a FRT site to pGE-attB-crb^rescue-PX to generate a rescued crb^FRT-PX founder line specifically for making conditional alleles carrying mutations on the C terminus of Crb (see Figure S2). One advantage of this strategy is that the constructs for making conditional alleles need to contain only an ∼1.5-kb crb gDNA, which is easier to make and more efficient to integrate. In addition, conditional alleles will not be expressed until the FRT recombination (Figure S2), so potential dominant-lethal alleles can also be readily generated. In general, by alternately using φC31 and Bxb1, a virtually unlimited number of DNA fragments can be integrated into a single locus in a successive and controlled fashion, making it possible to generate at a target locus some extremely large and/or complex knock-in alleles. Because of their small sizes, attPX-52 and attB-46 can be easily incorporated into any existing constructs and vectors by simple ligation of oligonucleotides or PCR products carrying their sequences. Although the integration efficiency of Bxb1 is lower than φC31 (Groth et al. 2004), especially when using the vasa-φC31 system (Bischof et al. 2007), we expect that the efficiency of Bxb1-mediated integration in Drosophila can be readily optimized through similar measures done on φC31 (Bischof et al. 2007), such as optimizing the Bxb1 codon usage, adding a nuclear localization signal sequence to help its entry into the nuclei, and providing germline-specific expression of Bxb1 using transgenes similar to vasa-φC31 (Bischof et al. 2007). Alternatively, as shown in Figure 2A, by integrating an attPC site via Bxb1-mediated integration, the attPX founder lines can be easily converted to attPC founder lines for higher integration efficiency. It is noteworthy that we also tested the phage integrases R4 (Olivares et al. 2001) and TP901 (Stoll et al. 2002) for DNA integration in Drosophila but did not obtain positive results (J. Huang and Y. Hong, unpublished data).

In addition, genomic engineering and similar approaches all require the generation of founder lines through gene targeting, which is the most critical and time-consuming step (Huang et al. 2009a). Gene targeting in Drosophila requires transgenic lines that carry the homologous DNA fragment (“donor DNA”) to be excised out later to induce homologous recombination. Integrase-mediated DNA integration would allow targeted insertion of a donor DNA into precharacterized chromosomal locations, offering high efficiency of donor DNA excision that facilitates homologous recombination and would eliminate the time-consuming process of chromosomal mapping and sorting of donor transgenic lines. However, approaches such as genomic engineering require the donor DNA construct to bear at least one φC31-attP site for later integration of knock-in constructs, effectively excluding the use of φC31 in making transgenic donor lines. It is now possible to generate transgenic donor lines via Bxb1-mediated integration without interfering with the use of φC31 in final genomic engineering founder lines. We are in the process of generating an extended array of attPX host lines using pAttPX and will systematically test and select them on the basis of the efficiencies of DNA integration, donor DNA excision, and gene targeting. Such precharacterized attPX host lines will greatly facilitate the transgenic donor lines that are optimized for gene targeting.

Materials and Methods

Fly stocks and genetics

The following stocks were obtained from the Bloomington Stock Center: BL#766: y¹ w^67c2 P[Crey}1b; noc^Sco/CyO and BL#851: y¹ w^67c23 P[Crey}1b; D*/TM3, Sb. The genomic engineering founder line w; crb^GX#24w[−]/TM3 e Sb was previously described (Huang et al. 2009b).

DNA constructs

pAttPX was constructed by inserting the attPX-52 (5′ GTGGTTTGTCTGGTCAACCACCGCGGTCTCAGTGGTGTACGGTACAAACCCA) to a modified pKIKO (Huang et al. 2008) in which both sides of the FRT sites and I-SceI sites were removed. pGE-attBX-GFP was constructed by replacing the attBC-53 with FRT, I-SceI, attBX-46 (5′ GGCCGGCTTGTCGACGACGGCGGTCTCCGTCGTCAGGATCATCCGG), and UAS-GFP in pGE-attB (Huang et al. 2009b). pGE-attBX-PC was constructed by replacing the attBC-53 with attBX-46 and attPC-50 in pGE-attB^GMR. pGE-attBX-G80EYC was constructed by inserting attBX-46, tub1-Gal80, the FRT site, and ey-Cre into a modified pGE-attB in which attBC-53 was removed. The details of pGE-attBX-G80EYC construction will be described elsewhere. pET11Bxb1polyA was made by replacing the φC31 cDNA in pET11phiC31polyA (Groth et al. 2004) with a Bxb1 cDNA sequence (Ghosh et al. 2003; Nkrumah et al. 2006). pGE-attB-crb^rescue-attPX was made by inserting into pGE-attB^GMR (Huang et al. 2009b) an attPX-52 and a full-length crb gDNA that was deleted in the founder lines.

Generation of attPX-52 host lines

pAttPX plasmid and w¹¹¹⁸ stocks were used to generate attPX host lines in Table 1 via the standard P-element-based transgenic protocol. The w+ marker was removed from attPX-52 lines by crossing with y¹ w^67c23 P[Crey}1b; D*/TM3, Sb as previously described (Huang et al. 2009b).

Generation and verification of Bxb1-mediated integrations in attPX-52 host lines

Bxb1-mediated DNA integration was carried out using a previously published φC31 protocol (Groth et al. 2004), except that the DNA mixtures were prepared with Bxb1 mRNA synthesized from a pET11Bxb1polyA plasmid. The Bxb1 mRNA was synthesized on the basis of the published protocol (Groth et al. 2004), and DNA/mRNA mixtures were made at final concentrations of 0.2 μg/μl DNA and 1 μg/μl mRNA (Groth et al. 2004). Integration lines were identified by w+ and verified by PCR using primers JH112 (5′ CGCAAAGCTAGAGTTGTTGG) and JH117 (5′ GATCTCGAGCTGCAGATTC), which produce a 310-bp product that contains the potential attLX sequence resulted from attPX-52 and attBX-46 recombination. PCR products were sequenced to confirm the presence of attLX.

Generation of crb-PX^GE#24[w−]

Embryos from w; crb^GX#24w[−]/TM3 e Sb were injected with pGE-attB-crb^rescue-PX, using an established φC31-mediated DNA integration protocol (Groth et al. 2004). Initial w+ integration lines of crb-PX^GE#24[w+] were converted to w^[−] using hs-Cre lines as previous described (Huang et al. 2009b).

Generation and verification of Bxb1-mediated integrations in crb-PX^GE#24[w−]

Embryos from w; crb-PX^GE#24w[−]/TM3 e Sb were injected with pGE-attBX-PC and pGE-attBX-G80EYC, using a Bxb1-mediated DNA integration protocol as mentioned above. Integrations of pGE-attBX-PC and pGE-attBX-G80EYC were verified first by the screening and genetic mapping of a w+ marker and then by the PCR amplification of the integration loci (Figure 2 and Figure S1, B and C). Primers used for PCR amplifications were the following: for crb-PCR1, HJ658 (AACGAAAAGCAATGACAACTAGAA) and YH509 (GTTTGCTCAGCTTGCTTCGC); and for crb-PCR2, HJ658 (AACGAAAAGCAATGACAACTAGAA) and YH510 (CATTATTACCATCGTGTTTACTGTTTATTG).