The available data show that recombinase-direct site integration can place a single-copy nonrearranged DNA fragment into the target site at a practical frequency. Moreover, a high percentage of those in-sertions express the transgene at a predictable and reproducible level.
The problem that lies ahead is whether a target site, once it is found favorable for transgene expression, can accept additional transgene
molecules. As mentioned earlier, site-specific recombination is a two-step process: the first two-step is to create a site, and the second two-step is to deliver the desired transgene. In the absence of homologus recombina-tion, single-copy target lines must be generated by conventional ran-dom integration. These target lines must be physically characterized and functionally tested for transgene integration and expression. If the target site can accommodate only a single delivery event, it is difficult to justify investing the time and labor into screening for a suitable tar-get. However, if additional DNA can be appended onto the existing target site, not only would it make the initial screening for target sites worth the effort, but it would also permit the construction of transgene clusters where collections of desirable traits reside. Clustering trans-genes, as opposed to scattering them at various places in the genome, would facilitate the introgression of large gene sets to field cultivars.
The idea of gene stacking rests on the concept that the integrating DNA brings along a different recombination site, such that after in-sertion of the new recombination site into the genome, the new recom-bination then becomes the new target for the next round of integration (Ow and Medberry, 1995). The following section describes two strate-gies for stacking through the introduction of fresh target sites.
Stacking via a Reversible Recombination System
Baszczynski and colleagues (Baszczynski, Bowen, Drummond, et al., 2001) described a strategy that uses heterospecific recombina-tion sites. These sites, with mutarecombina-tions in the spacer region, recombine with sites of identical spacer sequence, but not with sites of different spacer sequence. Figure 3.4 shows six heterospecific FRT sites, FRT, FRT5, FRT*, FRT2, FRT2*, FRTz, a collection of promoters, Px, Py, Pz, a collection of trait genes, G1, G2, G3, and two selectable mark-ers, M1 and M2. Here, G1, G2, and G3 are promoter-containing genes, but M1 and M2 represent promoterless coding regions.
The process begins with a genomic target consisting of Px-FRT-M1-FRT5-G1, using Px-M1 for selection (Figure 3.4a). The stacking of G2 to the G1 locus is accomplished with a first cassette: FRT-M2-FRT*-G2-Py-FRT2-FRT5 (Figure 3.4b). In an exchange reaction in-volving FRT × FRT and FRT5 × FRT5, the resulting genomic struc-ture would be Px-FRT-M2-FRT*-G2-Py-FRT2-FRT5-G1 (Figure 3.4c).
The FLP recombinase can be provided, for example, by transient ex-pression from a FLP-expressing plasmid. To stack G3 onto the G2-G1 locus, a second cassette would be used, bearing FRT2-M1-FRT2*-G3-Pz-FRTz-FRT5 (Figure 3.4d). Upon FRT2 × FRT2 and FRT5 × FRT5 recombination, the genomic structure would be Px-FRT-M2-FRT*-G2-Py-FRT2-M1-FRT2*-G3-Pz-FRTz-FRT5-G1 (Figure 3.4e).
At this point, the locus is not only stacked with G1, G2, and G3, but also with Px, Py, Pz, M1, and M2. The authors proposed using chimeraplasty to convert the sequences of FRT* and FRT2* to FRT and FRT2, respectively. Chimeraplasty is a technique in which RNA-DNA chimeric oligonucleotides are introduced into the cell to mutate homologous target DNA (for review, see Rice et al., 2001). Subse-quent FLP-mediated FRT × FRT and FRT2 × FRT2 recombinations would delete, respectively, M1 and M2, to generate the structure Px-FRT-G2-Py-FRT2-G3-Pz-FRTz-FRT5-G1 (Figure 3.4f). To deliver another trait gene to the locus, a third cassette would have to make use of FRTz and FRT5, and so on.
There are several concerns about this strategy. One is the limited availability of heterospecific sites. Even if lox and RS sites were in-corporated into this strategy to expand the repertoire of recombina-tion targets, it would not be long before new sites become unavail-able. Second, chimeraplasty occurs at a frequency of 10–4in tobacco and maize. It was successful for converting herbicide-sensitive genes to herbicide-resistant alleles (Beetham et al., 1999; Zhu et al., 1999).
However, unlike herbicide resistance, FRT spacer mutations are not selectable phenotypes. Therefore, this marker removal step should be viewed with some skepticism. Third, Figure 3.4 shows the clustering of quite a bit of extra DNA fragments in addition to the chosen trait genes. The FRT sequences are rather short and may be considered negligible, but the promoters are of considerable size.
Stacking via an Irreversible Recombination System
Although some recombination systems catalyze freely reversible reactions, many do not. Instead, the substrate sites, often known as attB and attP, are nonidentical. This necessitates that the product sites generated from an attB × attP reaction, attL and attR, are dissimilar in sequence to attB and attP. The recombination enzyme that promotes
FIGURE 3.4. Gene stacking strategy with a reversible recombination system.FRT, FRT5, FRT*, FRT2, FRT2*, FRTz are heterospecific recombination sites.Px, Py, Pz are promoters, G1, G2, G3 are trait genes, and M1, M2 are selectable markers. Baszczynski and colleagues (Baszczynski, Bowen, Drummond, et al., 2001) describedM1, M2 as lacking the ATG start codon, which is provided by fusion toPx, Py, or Pz. For simplicity, neither transcriptional nor translational fusions are specified here. Dotted lines indicate recombination between pairs of sites.
the attB × attP reaction, often referred to as the integrase, does not re-combine attL × attR. The lack of a readily reversible reaction gives a distinct advantage for employing such a system in DNA integration since integrated molecules are stable. Most important, however, an ir-reversible system permits a novel gene stacking strategy that is not achievable using only freely reversible systems. In fact, this is the un-derlying reason for this laboratory’s interest in the C31 recombina-tion system.
Figure 3.5 shows BB', PP', BP', and PB' as attB, attP, attL, and attB, respectively, G1, G2, G3, G4, and G5, as trait genes, and M1 and M2 as markers. However, unlike in Figure 3.4, M1 and M2 include functional promoters. The process begins with a single-copy trait gene linked to a marker: lox-M1-lox-G1-BB'-(inverted lox). The sin-gle-copy locus may be obtained by molecular screening. Alterna-tively, a complex multicopy integration pattern may be resolved by Cre-lox site-specific recombination into a single-copy state (Sriv-astava et al., 1999; Sriv(Sriv-astava and Ow, 2001). If a resolution-based strategy were used, the marker M1 would have been deleted, leaving a configuration consisting of lox-G1-BB'- (inverted lox). To append G2 to the G1 locus, the integrating plasmid with the structure PP'-G2-PP'-lox-M2 recombines with the genomic BB' target. The recombina-tion enzyme for the reacrecombina-tion, or the integrase, can be provided, for ex-ample, by transient expression from a cotransformed plasmid. Since either PP' can recombine with the single BB', two different integration structures would arise that are distinguishable by molecular analysis.
Figure 3.5b shows only the structure useful for further stacking, con-sisting of lox-M1-lox-G1-BP'-G2-PP'-lox-M2-plasmid backbone-PB'-(inverted lox). The Cre recombinase is introduced into the system to remove the unneeded DNA. The resulting structure becomes lox-G1-BP'-G2-PP'-(inverted lox). To stack G3, the construct BB'-G3-BB'-lox-M2 is introduced (Figure 3.5c). Analogous to the previous steps, the genome has only a single PP' site to recombine with either of the BB' sites on the plasmid. Recombination with the G3 upstream site produces the structure shown in Figure 3.5d. After removing the un-needed DNA, the locus containing G1, G2, and G3 is ready for the stacking of G4 (Figure 3.5e). In another variation, sets of inverted attB and attP sites, rather than sets of directly oriented sites, can also
FIGURE 3.5. Gene stacking strategy with irreversible recombination system.
Recombination sitesattB, attP, attL, and attB are shown as BB', PP', BP' and PB', respectively.G1, G2, G3, G4 and G5 are trait genes, and M1 and M2 are markers. However, unlike in Figure 3.4,M1 and M2 include functional promoters.
Dotted lines indicate recombination between pairs of sites.
be used. The sequence of events is analogous to those described for Figure 3.5.
Several features are worth noting. First, the vector for delivery of G4 is the same as the vector for delivery of G2. Likewise, the vector for delivery of G5 (Figure 3.5g) is the same as the vector for delivery of G3. In principle, the stacking process can be repeated indefinitely, alternating between the uses of two simple vectors. Second, the stacking from G2 onward requires only a single marker gene, and this is critical as it bypasses the need to continually develop new selec-table markers. Third, the trait genes, such as G1, G2, and so on, should not be narrowly interpreted as a single promoter-coding re-gion-terminator fragment. Each DNA fragment could not only be composed of multiple transgenes, but could also include border DNA that insulate its (their) expression from surrounding regulatory ele-ments. This may be useful when clustering transgenes that bring with them dominant cis-regulatory elements.
C31 SITE-SPECIFIC RECOMBINATION SYSTEM