Pollen- and seed-mediated transgene flow is a concern in plant biotechnology. We report here a highly efficient ‘genetically modified (GM)-gene-deletor’ system to remove all functional transgenes from pollen, seed or both. With the three pollen- and/or seed-specific gene promoters tested, the phage CRE/loxP or yeast FLP/FRT system alone was inefficient in excising transgenes from tobacco pollen and/or seed, with no transgenic event having 100% efficiency. When loxP-FRT fusion sequences were used as recognition sites, simultaneous expression of both FLP and CRE reduced the average excision efficiency, but the expression of FLP or CRE alone increased the average excision efficiency, with many transgenic events being 100% efficient based on more than 25 000 T1 progeny examined per event. The ‘GM-gene-deletor’ reported here may be used to produce ‘non-transgenic’ pollen and/or seed from transgenic plants and to provide a bioconfinement tool for transgenic crops and perennials, with special applicability towards vegetatively propagated plants and trees.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
In the greater public discussion of plant biotechnology, transgene flow is a serious concern, with transgene containment being the ultimate goal (Dale et al., 2002; Konig, 2003). However, gene transfer through pollination routinely occurs in many commercially important plants, and between these crops and their wild relatives (Conner et al., 2003; Stewart et al., 2003). It has been reported that cross-pollination between commercial canola fields can occur over considerable distances (Rieger et al., 2002; Conner et al., 2003), and pollen-mediated gene flow from creeping bentgrass can take place as far away as 21 km (Watrud et al., 2004). Seed dispersal is another important vehicle in the spread of transgenes and, in some cases, seeds can be dispersed via spillage during handling and transportation (Stewart, 2004). The lack of effective biological containment technologies to prevent pollen- and seed-mediated gene flow has hindered the commercialization of many economically important, long-lived, genetically modified (GM) perennials, such as trees, that freely outcross in nature (Bradshaw and Strauss, 2001).
There are several candidate technologies for transgene containment with pollen or seed as targets. Inserting transgenes into the chloroplast genome provides an excellent method to reduce pollen-mediated gene flow (Daniell et al., 2005a,b), but the technology is not effective in addressing the seed-mediated gene flow problem, and some species have biparental chloroplast inheritance. Although ‘terminator gene’ technology (Oliver et al., 1998) may reduce the seed-mediated spread of GM genes, its potentially negative perceived impact on farmers in developing countries has created obstacles for its commercialization (Giovannetti, 2003). Male and female sterility technologies can be used to reduce the gene flow problem from vegetatively propagated trees (Khan, 2005; Ruiz and Daniell, 2005), but the anticipated adverse ecological effects of large-scale plantations of sterile plants, such as sterile trees, are of concern, because fauna that feed on pollen, seed or fruit may be negatively impacted, probably resulting in a loss of biodiversity in these ecosystems.
To develop a highly efficient method for the targeting of GM gene removal from pollen and/or seed of GM crops, we designed several novel gene cassettes using components from both FLP/FRT and CRE/loxP recombination systems. With the loxP-FRT hybrid sequences as the recognition sites to flank all transgenes, we demonstrated that the expression of either the FLP or Cre gene alone could lead to 100% efficiency in deleting all functional transgenes from pollen, or from both pollen and seed, of transgenic tobacco plants based on the examination of more than 25 000 progeny seedlings per transgenic event.
Fusion gene construction
We constructed a series of fusion gene cassettes using components from both FLP/FRT and CRE/loxP systems and inserted them into pBIN19 Ti-plasmid vector (Bevan, 1984). Our hope was that a combined use of both FLP/FRT and CRE/loxP systems might enhance the efficiency of deletion of transgenes from target organs. Because the DNA sequences within the two T-DNA borders of pBIN19 Ti-plasmid were non-essential for Agrobacterium infection of plant tissues (Xiang et al., 1999), we deleted most of these sequences before introducing the fusion gene cassettes into the T-DNA region of pBIN19.
The gene cassette pLF (LF is an abbreviation for the loxP-FRT fusion sequence) (see Figure 1a), which served as a control, contained the GUS::NPTII fusion gene (Fabijanski et al., 2001) and two sets of 86-bp loxP-FRT sequences to flank all transgenes. As a recombinase gene was not included in this cassette, excision was not expected to occur in pLF plants. Cassettes pF_polB-FLP and pF_polseed-FLP contained the 48-bp FRT recognition sites, and pLF_polB-FLP, pLF_polL-Cre, pLF_pol-Cre + FLP and pLF_polseed-FLP contained the 86-bp loxP-FRT (34 bp for loxP, 48 bp for FRT and 4 bp for the spacer between loxP and FRT) recognition sites. We used pollen-specific BGP1 (Xu et al., 1993) and LAT52 (Twell et al., 1990) and pollen- and seed-specific PAB5 (Belostotsky and Meagher, 1996) gene promoters to control the expression of the FLP gene. The BGP1 gene promoter was used to drive FLP expression (BGP1-FLP-nos) in pF_polB-FLP and pLF_polB-FLP. The pollen- and seed-specific PAB5 gene promoter was chosen to control FLP expression (PAB5-FLP-nos) in pF_polseed-FLP and pLF_polseed-FLP. Gene cassettes pF_polB-FLP vs. pLF_polB-FLP and pF_polseed-FLP vs. pLF_polseed-FLP were designed to determine the effect of the loxP-FRT fusion sequences on the FLP-mediated excision efficiency. The cassettes pF_polB-FLP and pF_polseed-FLP contained the native FLP/FRT system, whereas pLF_polB-FLP and pLF_polseed-FLP contained both loxP from the CRE/loxP system and FRT from the FLP/FRT system as recognition sequences. The cassettes pLF-polB-FLP, pLF_polL-Cre and pLF_pol-Cre + FLP were used to study the effect of the simultaneous expression of both FLP and Cre genes on the excision efficiency.
Figure 1b shows a schematic diagram of transgene deletion from pollen and seed using gene cassette pLF_polseed-FLP as an example. Transgenic plants hosting pLF_polseed-FLP would be expected to have transgenes present in all organs during the plant life cycle, except in seed and pollen. Because of pollen- and seed-specific FLP expression, all functional transgenes including the FLP gene, except an 86-bp non-protein encoding the loxP-FRT fusion sequence, should be excised from pollen and seed. The 8.7-kb excised DNA sequence, consisting of one loxP-FRT site, the PAB5-FLP-nos gene and the 35S-GUS::NPTII-nos gene, should be destroyed by nonspecific nucleases in the cell (Wilson, 1975; Srivastava and Ow, 2003). All other gene cassettes shown in Figure 1a, except cassette pLF, should work in the same fashion as pLF_polseed-FLP, although their excision efficiencies may be different.
Production and verification of transgenic plants
We produced more than 35 independent transgenic tobacco events for each gene cassette, except pLF, using an Agrobacterium-mediated plant transformation method (Li et al., 1992). Putative transgenic plants were initially screened for kanamycin resistance and β-glucuronidase (GUS) activity, and then confirmed with Southern blot hybridization (data not shown). Reverse transcriptase-polymerase chain reaction (RT-PCR) analysis of representative transgenic events showed that FLP or Cre genes, under the control of BGP1, LAT52 and PAB5 gene promoters, were transcribed in pollen and/or seed, but not in leaves or other non-reproductive organs (Figure 2).
Histochemical characterization of GUS gene expression in transgenic plants
Detailed histochemical analysis for GUS activity of T0 plants and T1 progeny of representative transgenic events from each group was performed. Because the GUS gene was under the control of the constitutive 35S cauliflower mosaic virus (CaMV) gene promoter (Odell et al., 1985), GUS expression served as an excellent marker for the presence of transgenes in the cell. We examined GUS activity in vegetative organs and young flowers of transgenic plants hosting gene cassettes pLF_polB-FLP, pLF_polL-Cre and pLF_polseed-FLP, and observed GUS activity in all tissues of leaves, stems, roots and young flowers (Figure 3a–d). These results revealed that neither the FLP nor Cre gene under the control of the LAT52, BGP1 or PAB5 gene promoters was active in vegetative organs or in young flowers. However, based on whole-tissue GUS staining patterns, we cannot exclude the possibility that excision may have occurred in a small number of cells at a low frequency in these organs.
To examine the FLP- or CRE-mediated excision of the transgenes in pollen, we conducted histochemical staining of GUS activity in pollen at different stages of flower development. As shown in Figure 3e,f, GUS activity was observed in both immature and mature pollen grains of plants hosting cassette pLF (containing the 35S-GUS::NPTII-nos gene but lacking an FLP or Cre gene). For transgenic events that harboured cassettes pLF_polB-FLP, pLF_polL-Cre and pLF_polseed-FLP, GUS activity was detected in immature pollen (Figure 3g), but disappeared in mature pollen (Figure 3h), suggesting that the transgenes were deleted from pollen. In transgenic plants hosting cassette pLF_polseed-FLP (containing the pollen- and seed-specific PAB5-FLP gene), no GUS activity was observed in the progeny plants derived from self-pollination (Figure 3l) or reciprocal cross-pollination with wild-type plants (Figure 3m,n), suggesting that the transgenes were excised from both pollen and seed. In contrast, GUS activity was detected in progeny seedlings of pLF plants when selfed (Figure 3i) and crossed with wild-type plants (Figure 3j,k), indicating that no excision occurred when neither FLP nor Cre was included.
Determination of efficiency of transgene excision from pollen and seed
To characterize the excision efficiency of each gene cassette, we vegetatively propagated all transgenic events and used these T0 plants to conduct self- or cross-pollination (between transgenic and wild-type plants) for each transgenic event obtained. We analysed the GUS gene expression in T1 progeny seedlings and used the number of GUS-positive progeny seedlings from each self- or cross-pollination to calculate the excision efficiency. Table 1 shows the total number of transgenic events analysed for each gene cassette, the average number of progeny seedlings per event examined for GUS activity, the ranges and means of the excision efficiencies observed, the number of transgenic events with 100% excision efficiency, and the average number of progeny seedlings analysed for those exhibiting 100% excision efficiency. None of the pF_polB-FLP or pF_polseed-FLP transgenic events (containing the FRT sequences as recognition sites) achieved 100% excision efficiency for transgenes from pollen and/or seeds. By contrast, a significant number of transgenic plants hosting cassettes pLF_polB-FLP or pLF_polseed-FLP (containing the loxP-FRT fusion sequences as recognition sites) were 100% efficient in deleting all functional transgenes from pollen and/or seed. For the transgenic events that exhibited 100% excision efficiency, vegetatively propagated T0 plants were used to conduct additional self- or cross-pollination with wild-type plants to produce more than 25 000 T1 progeny seeds for each event. As shown in Table 1, 13 of 42 pLF_polB-FLP plant events and eight of 45 pLF_polseed-FLP transgenic events were 100% efficient in deleting all transgenes from pollen and/or seed. Because cassettes pLF_polB-FLP and pLF_polseed-FLP contained loxP-FRT fusion sequences as recognition sites, whereas cassettes pF_polB-FLP and pF_polseed-FLP contained only FRT sequences as recognition sites, it is clear that the combined use of the loxP and FRT sequences (loxP-FRT fusion) as the recognition sites dramatically enhances the excision efficiency.
Table 1. Transgene excision efficiencies in pollen and seed as determined by the analysis of β-glucuronidase (GUS) activity in T1 seedlings
Two sets of loxP-FRT (LF) or FRT (F) in direct orientation were used to flank all functional transgenes delivered into host plants.
pLF_pol-Cre + FLP
Table 1 also shows that transgenic events hosting pLF_pol-Cre + FLP (with the loxP-FRT fusion sequences as recognition sites and both FLP and CRE recombinase genes expressed simultaneously) had an average excision efficiency of 11% from a total of 39 independent transgenic events, whereas transgenic plants hosting pLF_polB-FLP (containing the loxP-FRT fusion sequence with BGP1-FLP-nos) and pLF_polL-Cre (containing the loxP-FRT fusion sequence with LAT52-Cre-nos) had average excision efficiencies of 79% based on a total of 42 independent transgenic events and 33% from 38 independent events, respectively. Furthermore, none of the pLF_pol-Cre + FLP transgenic events showed 100% efficiency in deleting transgenes from pollen. In contrast, 31% of pLF_polB-FLP events and 13% of pLF_polL-Cre events were 100% efficient based on more than 25 000 T1 progeny seedlings examined per event. These results demonstrate that simultaneous expression of FLP and CRE leads to a marked decrease in excision efficiency. In addition, the data in Table 1 reveal that, of the three gene promoters used, BGP1 appeared to be more effective than LAT52 or PAB5, because pLF_polB-FLP plants had an average excision efficiency of 79% based on a total of 42 independent events, whereas pLF_polL-Cre and pLF_polseed-FLP plants had efficiencies of 33% from a total of 38 independent events and 55% based on 45 independent events, respectively.
To confirm the high excision efficiencies observed, we vegetatively propagated representative pLF_polB-FLP, pLF_polL-Cre and pLF_polseed-FLP plants. Although we observed 100% excision efficiencies from pollen and/or seed in a number of transgenic plants hosting two copies of pLF_polB-FLP, pLF_polL-Cre or pLF_polseed-FLP, the plants used for propagation were those that contained a single copy of the transgenes. Using sexually and/or vegetatively propagated plants, we conducted additional self-pollination and cross-pollination with wild-type plants. As shown in Table 2, for plants hosting pLF (no recombinase gene included), we observed a positive GUS stain in about 75% of the progeny when self-pollinated, consistent with the expected 3 : 1 simple Mendelian ratio for genetic segregation. When pollen grains from pLF plants were used to cross with wild-type plants or wild-type plant pollen grains were used to pollinate pLF plants, approximately 50% of the progeny were GUS positive. These results show that, as expected, no excision of transgenes occurred in plants hosting pLF.
Table 2. Analysis of transgene excision efficiencies of representative plants*
Transgenic events hosting a single copy of the transgenes were analysed.
A detailed explanation of the gene cassettes used is given in Figure 1.
Transgenic plants were pollen donors and wild-type plants were seed parents for these crosses.
Transgenic plants were seed parents and wild-type plants were pollen donors for these crosses.
‘Expected’ ratio for GUS− : GUS+ was derived according to Mendel's law of genetic segregation assuming that there was no excision of transgenes.
‘Observed’ number of GUS− : GUS+ progeny seedlings was obtained from each self-pollination or cross-pollination experimentally. If the number of GUS+ progeny seedlings was zero, the excision efficiency in the pollen and/or seed of the relevant parental plant(s) was 100%.
1 : 3
776 : 2127
1 : 1
987 : 1028
1 : 1
1009 : 985
No FLP gene/no excision in either pollen or seed
1 : 3
12693 : 12287
1 : 1
25883 : 0
1 : 1
11944 : 11718
Pollen-specific FLP expression/100% excision in pollen but no excision in seed
1 : 3
11892 : 12001
1 : 1
27954 : 0
1 : 1
19765 : 19041
Pollen-specific FLP expression/100% excision in pollen but no excision in seed
1 : 3
12376 : 12964
1 : 1
26556 : 0
1 : 1
12356 : 12583
Pollen-specific CRE expression/100% excision in pollen but no excision in seed
1 : 3
11468 : 11292
1 : 1
24398 : 0
1 : 1
16790 : 17657
Pollen-specific CRE expression/100% excision in pollen but no excision in seed
1 : 3
24103 : 0
1 : 1
25223 : 0
1 : 1
31790 : 0
Pollen- & seed-specific FLP expression/100% excision in both pollen & seed
1 : 3
32990 : 0
1 : 1
26343 : 0
1 : 1
32120 : 0
Pollen- & seed-specific FLP expression/100% excision in both pollen & seed
GUS, β-glucuronidase; WT, wild-type.
For plants hosting one copy of pLF_polB-FLP (containing the pollen-specific BGP1-FLP-nos gene), 50% of progeny seedlings were GUS positive if the plants were pollinated with wild-type pollen (Table 2). If the pollen of pF_polB-FLP plants was used to pollinate wild-type plants, the resulting T1 progeny seedlings were 0% GUS positive, demonstrating that transgenes were deleted from pollen. Similar results were also obtained for the plants that hosted one copy of pLF_polL-Cre (the pollen-specific LAT52-Cre-nos gene). For pLF_polseed-FLP plants (containing the pollen- and seed-specific PAB5 gene promoter used to drive FLP expression), none of the T1 progeny seedlings from either self-pollinated or reciprocal wild-type crosses were GUS positive. The results demonstrate that pLF_polB-FLP, pLF_polL-Cre and pLF_polseed-FLP are 100% efficient for the deletion of all transgenes from pollen and/or seed based on more than 25 000 T1 seedlings examined per event.
Molecular characterization of transgene excision from pollen and seed
The deletion of all transgenes from pollen and/or seed was verified using Southern hybridization and highly sensitive PCR techniques. Southern blot analysis confirmed the presence of the GUS and FLP genes in vegetative organs of T0 pLF_polB-FLP, pLF_polL-Cre or pLF_polseed-FLP plants, but these genes were absent in progeny from self-pollinated pLF_polseed-FLP plants or cross-pollinated pLF_polB-FLP or pLF_polL-Cre plants with wild-type plants being pollen recipients (Figure 4a,b). In addition, using the genomic DNA isolated from T0 pLF_polB-FLP and pLF_polseed-FLP plants as templates and a set of oligos annealing the DNA sequences outside the two loxP-FRT sites for PCRs, we amplified 7.5- and 8.7-kb DNA products, respectively (Figure 4c).
DNA sequencing analysis confirmed that the 7.5- and 8.7-kb products contained pLF_polB-FLP and pLF_polseed-FLP (Figure 1a). Using the same oligos and the genomic DNA from T1 seedlings from crosses of pLF_polB-FLP (male) × wild-type (female) plants and from self-pollinated pLF_polseed-FLP plants, we amplified a 0.2-kb fragment instead of 7.5- or 8.7-kb fragments (Figure 4c). This 0.2-kb fragment was not found in wild-type plants or in vegetative organs of T0 pLF_polB-FLP or T0 pLF_polseed-FLP plants. Furthermore, DNA sequencing analysis also revealed that the 0.2-kb signal consisted of a single loxP-FRT site and the deletion junction located within the two loxP-FRT direct repeats (Figure 4d). Disappearance of the 7.5- or 8.7-kb PCR products and appearance of the 0.2-kb post-excision PCR product in progeny further confirmed that all transgenes were deleted from the pollen and/or seed.
As shown here, with the combined use of loxP and FRT sequences (86 bp in length) as the flanking sites for FLP or CRE recombinase, we have developed an exceptionally efficient system for deleting all functional transgenes from pollen and/or seed of transgenic plants (Figure 5). Based on the analysis of more than 25 000 progeny for each representative transgenic event, efficiencies for the automatic deletion of all functional transgenes from pollen and/or seed are as high as 100% under glasshouse conditions. In addition, we have demonstrated that the high excision efficiency trait remains stable in vegetatively propagated progeny plants. Our results suggest that the ‘GM-gene-deletor’ system reported here should be readily applicable for the production of ‘non-transgenic’ pollen and seed from vegetatively propagated GM plants that are important to forest, ornamental and paper industries.
In our study, the deletion of all functional transgenes from pollen and seed was confirmed using three different techniques: histochemical assay for GUS activity, Southern blot hybridization and PCR. Histochemical assay of GUS activity offers a simple, easy and highly sensitive method for both the initial screen of transgene activity in pollen and seed, and for the determination of the excision efficiency in large numbers of progeny seedlings from self- or cross-pollinated events. PCR amplification provides a highly sensitive method to verify the deletion of all functional transgenes from pollen and seed, and Southern blot hybridization offers additional confirmation. It should be noted that, although all functional transgenes were deleted, a short, non-native, non-protein encoding an 86-bp loxP-FRT sequence remained behind in pollen and seed. However, the potential for adverse environmental and health effects from this non-expressed DNA sequence should be minimal or relatively easy to determine. Moreover, removal of all foreign DNA sequences, including the loxP-FRT recognition sequence, may be possible with the use of a recombinase that either recognizes related native plant sequences or cuts outside of their recognition sequences (Keenan and Stemmer, 2002).
Because the efficiencies of systems with FRT, loxP or the loxP-FRT fusion as flanking sites were compared side by side, our results clearly demonstrate that the loxP-FRT fusion sequences as recognition sites dramatically enhance the FLP- or CRE-mediated excision efficiency. We speculate that the fused loxP-FRT sequences may enhance the alignment of the recognition sequences, DNA bending or cleavage, or the formation of a Holliday junction or DNA–recombinase complex (van Duyne, 2001; Chen and Rice, 2003), thus leading to improved efficiency. In addition, the presence of loxP on one side and FRT on the other side of the flanking DNA sequence may enhance the excision efficiency, as the use of loxP on one side and FRT on the other results in higher rates of FLP- or CRE-mediated recombination (Lauth et al., 2002). However, when FLP and CRE were expressed simultaneously with loxP-FRT fusion sequences as the recognition sites (pLF_pol-Cre + FLP), the excision efficiency was decreased relative to pLF_polL-Cre and pLF_polB-FLP events (Table 1). Because we originally thought that the combined use of the CRE/loxP and FLP/FRT systems would enhance the excision efficiency, this result was surprising and unanticipated. It is possible that both FLP and CRE may bind to the same loxP-FRT molecule and such binding may affect the formation or stability of the DNA–recombinase complex (van Duyne, 2001; Chen and Rice, 2003), thus reducing the excision efficiency. However, additional experiments, preferably using a bacterial system, are needed to test this possibility.
Transgene containment technologies with little ‘leakage’ are highly desirable to avoid potentially ecologically important rare hybridization and introgression events (Stewart et al., 2003). Although the system reported by Mlynarova et al. (2006), using a CRE/loxP system to delete all functional transgenes from pollen to produce ‘non-transgenic’ pollen, was high, as suggested by Haygood et al. (2004), a higher excision efficiency may be needed under large-scale field conditions. Our data demonstrate that the ‘GM-gene-deletor’ system, with a unique use of loxP and FRT sequences, is 100% efficient based on more than 25 000 progeny examined, a significantly higher efficiency than that observed by Mlynarova et al. (2006). Furthermore, the ‘GM-gene-deletor’, when the PAB5 gene promoter is used to control FLP or CRE expression, can delete all functional transgenes from both pollen and seed, whereas the Mlynarova system deletes transgenes from pollen only. Thus, the ‘GM-gene-deletor’ system reported here should offer a better technique than the Mlynarova system to address the GM gene flow problem.
A modified version of the ‘GM-gene-deletor’ technology could also be useful in preventing gene flow from both pollen and seed of sexually propagated crops, such as sunflower, canola and sorghum, which outcross to wild relatives (Conner et al., 2003; Stewart et al., 2003). If a non-phytotoxic, chemically inducible RNAi-FLP gene is introduced into the pLF_polseed-FLP cassette, FLP expression in pollen and seed could be conditionally repressed (via a spray application of an induction agent), which should prevent the deletion of transgenes from pollen and seed, and allow the production of certified or registered seed stocks. Without application of the induction agent, sexually produced plants would produce transgene-free pollen and/or seed in the subsequent generation. This system may be especially applicable for genetically encoded and potentially controversial traits, such as industrial or pharmaceutical proteins.
There may be additional applications for ‘GM-gene-deletor’ technology. First, non-transgenic organs or plants from transgenic plants could be produced, mitigating consumer concerns over transgene presence in food (Figure 5). For example, with a fruit-specific gene promoter to drive FLP gene expression, all transgenes in fruit could be deleted. Second, the ‘GM-gene-deletor’ could provide farmers with the ability to replant viable, but non-transgenic, seed harvested from transgenic plants. This benefit stands in contrast with the controversial ‘terminator seed’ technology (Shand, 2002; Giovannetti, 2003) that produces non-viable seed. The ‘GM-gene-deletor’ would have both technology restriction and biosafety utilities without seed sterility. Third, ‘GM-gene-deletor’ technology, when it can be used in grain crops, would eliminate the need for labelling and physical separation of transgenic from non-transgenic grains after harvest. Fourth, the combined use of loxP and FRT as the recognition sequences for FLP or CRE recombinases, the key element of the ‘GM-gene-deletor’ system, may also be able to enhance the efficiencies of CRE or FLP in some non-plant systems.
Construction of gene cassettes
To construct the fusion gene cassettes used in this study, we synthesized the loxP-FRT fusion sequences with a spacer and a multiple cloning site in the middle: 5′-GGGAATTCATAACTTCGTATAGCATACATTATACGAAGTTATGACTGAAGTTCCTATACTTTCTAGAGAATAGGAACTTCGGAATAGGAACTTCGGTACCTATGTCGACGAACTCGAGTAGAGCTCAAGGATCCTTATAACTTCGTATAGCATACATTATACGAAGTTATGACTGAAGTTCCTATACTTTCTAGAGAATAGGAACTTCGAATAGGAACTTCGCTAGCGG-3′. We synthesized the FRT recognition sequences with a spacer and a multiple cloning site in the middle: 5′-GGGAATTCGAAGTTCCTATACTTTCTAGAGAATA GGAACTTCGGAATAGGACTTCGGTACCTATGTCGACGAACTCGAGTAGAGCTCAAGGATCCTTGAAGTTCCTATACTTTCTAGAGAATAGGAACTTCGGAATAGGAACTTCGCTAGCGG-3′. The italic letters represent the loxP site (34 bp) and the bold letters represent the FRT sequence (48 bp). The four base pairs, GACT, between the loxP and FRT sites denote the spacer sequence. The DNA sequence flanked by two loxP-FRT sites or two FRT sites, GGTACCTATGTCGACGAACTCGAGTAGAGCTCAAGGATCCTT, is a multiple cloning site containing KpnI, SalI, XhoI, SacI and BamHI restriction enzyme sites. The 8-bp DNA sequence at the 5′-end contains an EcoRI site (GAATTC), and the 8-bp DNA sequence at the 3′-end contains an NheI site (GCTAGC).
We deleted most of these sequences within two T-DNA borders of the pBIN19 Ti-plasmid before introducing the fusion gene cassettes. The synthetic loxP-FRT fusion and FRT sequences were inserted into the pBIN19 vector separately within EcoRI and NheI, respectively. A GUS and NPTII fusion gene (Fabijanski et al., 2001), under the control of the globally active 35S CaMV gene promoter (Odell et al., 1985) (35S-GUS::NPTII-nos), was then cloned into two FRT or loxP-FRT sites. The NPTII gene (the kanamycin resistance gene) served as a marker gene for the selection of transgenic plants during transformation, and the GUS reporter gene provided a convenient tool to monitor the presence or absence of the transgenes in specific organs of transgenic plants. The 35S-GUS::NPTII-nos fusion gene was excised from pBluescript KS II with SalI plus XhoI, and cloned into pBIN19-LF, which was digested with SalI to produce the pBIN19-LF-GN construct (pLF).
The resulting pBIN19 derivative vectors were used to host the BGP1-FLP-nos, LAT52-Cre and PAB5-FLP genes. To amplify a 660-bp fragment of the BGP1 5′-upstream region, we used genomic DNA of Brassica campestris as the template, 5′-GCGGTACCTATCATTCCTTTAATTTCAAGG-3′ as the 5′-end primer and 5′-GTCTGCAGTTGGAGAGGAGATGGGGTTG-3′ as the 3′-end primer for a 40-cycle PCR: denaturation of the DNA template at 94 °C for 1 min, followed by primer annealing at 58 °C for 1 min and extension at 72 °C for 1 min. The KpnI- and PstI-digested 660-bp BGP1 5′-upstream fragment was fused to the 5′-end of the FLP-nos sequence in pBluescript KS II (kindly provided by Dr G. M. Wahl, Salk Institute, San Diego, CA, USA). The BGP1-FLP-nos fusion gene was excised from pBluescript KS II with KpnI and SacI digestion and ligated into the KpnI and SacI sites of the pBIN19-LF-GN vector, resulting in the pLF_polB-FLP construct. The LAT52-Cre-nos fusion gene was constructed using the same cloning strategy as for the BGP1-FLP-nos gene. The coding sequence of the Cre gene was kindly provided by Dr David Ow, USDA Plant Gene Expression Center, Albany, CA, USA. The LAT52 gene promoter and PAB5 gene promoter were cloned from tomato and Arabidopsis plants using PCR-based technology with the following two pairs of specific DNA primers: 5′-GGCGGATCCTATACCCCTTGGATAAG-3′ (KpnI site in bold), 5′-CGGCTGCAGAGCACAATAGCCTTTGCC-3′ (PstI site in bold) and 5′-GGCGGATCCGGCAAGACTCTTCGTCTTTG-3′ (KpnI site in bold), 5′-CGGCTGCAGGGGCAATTCCAGATGCAAC-3′ (PstI site in bold), respectively. The same strategy was used to construct all other gene cassettes shown in Figure 1a with the help of the restriction enzyme sites indicated. Most of these restriction enzyme sites were introduced during the cloning processes.
Plant transformation, growth conditions and histochemical staining
Agrobacterium tumefaciens LBA 4404 manipulation, Nicotiana tabacum leaf disc transformation and histochemical staining for GUS activity were performed as described previously (Li et al., 1992). Plants used for this study were grown in the glasshouse. All plants used for the experiments were T0 generation plants that had been vegetatively propagated.
Southern blot, PCR and DNA sequencing analysis
Southern blot hybridization was performed using 10 µg of PstI-, BamHI- or SalI-digested tobacco genomic DNA with α-32P-labelled FLP or GUS DNA fragments as hybridization probes (RadPrime Labelling Kit, GibcoBRL, Grand Island, NY). The hybridization signals were visualized using a Packard Cyclone Storage Phosphor System (Packard Instruments, Meriden, CT). Genomic DNA isolated from immature pollen (stage 5 flowers) (Koltnow et al., 1990) of T0 plants was used to establish the pre-excision signal, and from T1 seedlings from self- or cross-pollination with wild-type plants to establish the post-excision signal. Two oligos, 5′-GAACGTGGCGAGAAAGGAAGG-3′ and 5′-ACTGACAGAACCGCAACGTTG-3′, specific to the T-DNA sequences outside the two loxP-FRT fusions were used as primers for PCRs at 94 °C for 1 min, 60 °C for 1 min and 72 °C for 8 min for 40 cycles. DNA fragments were amplified by Ex Taq polymerase (TaKaRa, Madison, WI, USA) and sequenced using an Applied Biosystems 377 DNA sequencer (Perkin-Elmer, Wellesley, MA).
This work was supported by Connecticut Innovation, Inc. (USA) (to Yi Li, Richard McAvoy and Yan Wu), Consortium of Plant Biotechnology Research, Inc. (CPBR)/US Department of Energy (USA) (to Yi Li), the University of Connecticut Research Foundation and the University of Connecticut Storrs Agricultural Experiment Station (USA) (to Yi Li), NSF of China (no. 20050490) (to Yan Pei), and the University of Tennessee Institute of Agriculture (USA) (to C. Neal Stewart Jr.). The authors thank Dr G. Berkowitz, Mr W. Smith and Mr C. O'Donnell for editing, discussion and suggestions.