• Open Access

A Cre::FLP fusion protein recombines FRT or loxP sites in transgenic maize plants

Authors


  • Novel materials described in this publication may be available for non-commercial research purposes on acceptance and signing of a material transfer agreement. In some cases, such materials may contain or be derived from materials obtained from a third party. In such cases, the distribution of material will be subject to the requisite permission from any third-party owners, licensors or controllers of all or parts of the material. Obtaining any permissions will be the sole responsibility of the requestor.

* Correspondence (fax +1 515 270 3444; e-mail alex.lyznik@pioneer.com)

Summary

The coding sequences of Cre (site-specific recombinase from bacteriophage P1) and FLP (yeast 2-µm plasmid site-specific recombinase) were fused in frame to produce a novel, dual-function, site-specific recombinase gene. Transgenic maize plants containing the Cre::FLP fusion expression vector were crossed to transgenic plants containing either the loxP or FRT excision substrate. Complete and precise excisions of chromosomal fragments flanked by the respective target sites were observed in the F1 and F2 progeny plants. The episomal DNA recombination products were frequently lost. Non-recombined FRT substrates found in the F1 plants were recovered in the F2 generation after the Cre::FLP gene segregated out. They produced the recombination products in the F3 generation when crossed back to the FLP-expressing plants. These observations may indicate that the efficiency of site-specific recombination is affected by the plant developmental stage, with site-specific recombination being more prevalent in developing embryos. The Cre::FLP fusion protein was also tested for excisions catalysed by Cre. Excisions were identified in the F1 plants and verified in the F2 plants by polymerase chain reaction and Southern blotting. Both components of the fusion protein (FLP and Cre) were functional and acted with similar efficiency. The crossing strategy proved to be suitable for the genetic engineering of maize using the FLP or Cre site-specific recombination system.

Introduction

Site-specific recombination has become an important molecular tool for the production of transgenic organisms (Branda and Dymecki, 2004; Sorrell and Kolb, 2005). A new generation of transgenic crop plants has been developed that incorporate recent advances in DNA recombination technology (Ebinuma et al., 2005; Tzfira and White, 2005; Lyznik et al., 2007). In particular, simple, two-component (recombinases and their cognate target sites), unrestrictive (recombining target sites in different orientations on a single or two different DNA molecules regardless of their topological structures) and flexible (catalysing DNA insertions, inversions, deletions) site-specific recombination systems, such as Cre-lox, FLP-FRT or R-RS, are gaining popularity in biotechnological applications. Cre-mediated (Cre, site-specific recombinase from bacteriophage P1) chromosomal excisions produce marker-free transgenic crop plants, such as maize (Zhang et al., 2003), rice (Hoa et al., 2002; Sreekala et al., 2005), potato (Cuellar et al., 2006) and even strawberries (Schaart et al., 2004), amongst others (Hare and Chua, 2002). The R-RS system has been used to produce marker-free transgenic plants as a component of the multi-autotransformation (MAT) procedure (Ebinuma et al., 2005). Chromosomal excision reactions catalysed by site-specific recombinases may also be used to eliminate multicopy transgenes, thus simplifying the subsequent breeding steps of transgenic events. This approach was originally demonstrated in wheat and maize plants containing elements of the Cre-lox system (Srivastava et al., 1999; Srivastava and Ow, 2001b). In addition, site-specific integration into pre-inserted mutant loxP sites has been accomplished in tobacco and rice (Albert et al., 1995; Srivastava and Ow, 2001a). This procedure may contribute to the consistent and reliable expression of foreign genes in plants (Day et al., 2000). Excisions catalysed by the yeast FLP-FRT system (FLP, yeast 2-µm plasmid site-specific recombinase) have been demonstrated in maize (Lyznik et al., 1996; Kerbach et al., 2005), rice (Radhakrishnan and Srivastava, 2004; Hu et al., 2008) and other plant species (Kilby et al., 1995; Sonti et al., 1995; Gidoni et al., 2001). However, its use for crop genetic improvement has been delayed, or even questioned, because of inconsistent results produced by FLP (Kerbach et al., 2005). The kinetic properties of FLP recombinase, when compared with those of Cre, may require more rigorous experimental conditions (Ringrose et al., 1998), but, once fulfilled, the performance of FLP is comparable with that of other recombination systems, as demonstrated in this article.

The availability of different site-specific recombination systems creates new opportunities for the genetic manipulation of plant genomes. Recombination-mediated cassette exchange (RMCE) protocols have been used to integrate foreign DNA into pre-selected chromosomal locations of mammalian cells (Feng et al., 1999; Baer and Bode, 2001) and plant cells (Nanto et al., 2005; Louwerse et al., 2007). If using just one recombination system, RMCE requires a pair of mutated, incompatible target sites on an incoming DNA vector and a chromosomal integration site (for example, FRT1 and FRT5 or loxP and lox2272) (Kolb, 2001). Cross-reactivity and reduced overall activity of the mutated sites may compromise the efficiency of RMCE (Kim et al., 2006). Two recombinases with their cognate target sites (e.g. two systems that do not interfere with each other) should alleviate some limitations of RMCE. Indeed, the combined action of FLP and Cre on their respective FRT and loxP sites produced chromosomal integration at a frequency of 61%–78% of selected embryonic stem cells (Lauth et al., 2002). The Cre::FLP fusion protein described in this paper may provide a suitable reagent to execute such experiments, provided that both FLP and Cre perform well.

Although transient expression delivers a strong burst of recombinase activity (Zuo et al., 2001; Endo et al., 2002; Mlynarova and Nap, 2003; Wang et al., 2005; Cuellar et al., 2006; Kondrak et al., 2006), crossing strategies rely on stable, relatively weaker, expression that may result in a diminished efficiency of recombination. This may be compensated by a large number of progeny plants with each F1 plant being a potential recombination event. Using genetic crosses for self-pollinating plants with bisexual flowers can be quite cumbersome (rice, wheat, tobacco, Arabidopsis) (Kilby et al., 1995; Hoa et al., 2002; Hu et al., 2008), but such an approach is particularly suitable for genetic engineering and subsequent breeding of transgenic maize plants (Zhang et al., 2003). There is no need to design elaborate vector constructions and selection schemes (e.g. positive/negative selection) for the production of transgenic/recombination-competent plants. In addition, the introduction of a recombinase through crossing may be accomplished as part of the routine breeding activities that follow transformation. We show that such a strategy is feasible in maize for both Cre- and FLP-mediated chromosomal recombination.

Results

Experimental design

Two chromosomes carrying a recombination substrate (either PHP 18656 or PHP 18524) and a Cre::FLP expression unit (PHP 17783) were brought together in the zygote as a result of egg cell fusion (fertilization) with one sperm of a pollen grain. Parental plants bearing the PHP 18656 vector (FT plants) delivered the FLP excision substrate, whereas plants bearing the PHP 18524 vector (LX plants) provided the Cre excision substrate (Figure 1). Activation of the Ds-Red2 marker, as the footprint of FLP-mediated excision, was designed to monitor FLP activity. Excisions catalysed by Cre were detected by polymerase chain reaction (PCR), and activation of the yellow fluorescent protein (YFP) marker indicated either inversions or random re-integration of the episomal product of Cre-mediated excisions. Inversions and excisions were mutually exclusive in activating the expression of YFP.

Figure 1.

Schematic diagram of the FLP (PHP 18656) and Cre (PHP 18524) recombination substrates and products. Expression of marker genes Zs-Yellow and Ds-Red2 from the site-specific recombination substrates and/or products produces yellow (YFP+) and red (Red2+) fluorescence phenotypes, respectively. A fully functional YFP gene is flanked by two FRT sites (white arrows represent the 599-bp inverted repeat from 2-µm yeast plasmid) that are embedded into the first maize ubiquitin intron in the PHP 18656 vector. This genetic element separates the Ds-Red2 coding sequence from its maize ubiquitin promoter (ubi promoter). A selectable marker gene (moPAT) is placed outside of the excision cassette and is driven by its own rice actin promoter. It should be noted that the Ds-Red2 gene is activated at the excision footprint in the FLP-mediated recombination reaction, whereas Zs-Yellow gene activity is not affected by the FLP reaction. The PHP 18524 substrate contains two T-DNAs integrated in a head-to-head configuration, thus allowing for either Cre-mediated excisions or inversions. The red lines indicate diagnostic products of DNA restrictions used for Southern blot analysis. Polymerase chain reaction (PCR) amplification products are marked by black arrows. The Zs-Yellow gene becomes active as a part of the Cre-produced episomal product, or can be activated by Cre-mediated inversion if this reaction precedes excision.

The expression of the Cre::FLP fusion protein was controlled by a modified maize ubiquitin promoter and the potato pinII terminator (see ‘Experimental procedures’ section for details). Two single-locus (segregating 1 : 2 : 1 in the T1 selfed generation) and single-copy (as determined by the quantitative PCR zygosity screen of the T1 segregating population) transgenic events were selected for the subsequent genetic crosses. The CF21 event was identified as a weak expressor of the Cre::FLP fusion, whereas the CF42 event provided FLP activity comparable with our other best FLP-expressing transgenic plants (Figure 2, plant FP20). The strong FLP activity produced a uniform population of the Red2+ kernels on crossing to the FT01 tester plants (Table 1, Figure 2). The CF21 event was used as a control treatment to correlate our observations with the level of FLP activity in maize plants.

Figure 2.

Ds-Red2 expression in the F1 kernels produced by crossing parental plants containing the FLP excision substrate (FT) and either the Cre::FLP fusion construct (CF) or the Ubi::FLP vector (FP). Yellow fluorescence was produced by the functional YFP gene on the FLP substrate. Red fluorescence originated from the excision footprints produced by FLP. The maternal parent plants are listed first. The Cre::FLP fusion event produced FLP activity comparable with that of the FLP gene driven by the maize ubiquitin promoter in the FP20 event (smaller photograph in the bottom left-hand corner).

Table 1.  Phenotypic variation in the F1 kernels of the Cre::FLP crosses. Kernels were scored on the basis of the presence of yellow fluorescence (YFP+) or red fluorescence (Red+) in either embryos or endosperm. Chimeric phenotypes for the expression of fluorescence proteins were scored as positive. Two transgenic events containing the FLP substrate (FT01 and FT09) were crossed reciprocally with the Cre::FLP plants identified as either maternal (♀) or paternal (♂) parents. For the LX18 crosses, only sectors of yellow fluorescence within endosperm or embryos were identified in the F1 generation
 FT01FT09LX18
YFP+Red+DarkYFP+Red+DarkYFP+Dark
Cre::FLP♀
CF42  011087  060369Sectors220
CF21416  270539 35 00167
Cre::FLP♂
CF42  010996  025664Sectors 81
CF21714 1380369 15140120

FLP-mediated excisions in the F1 and F2 progeny

Homozygous plants from two different PHP 18656 events (FT01 and FT09) were crossed to transgenic plants carrying the Cre::FLP fusion (Table 1). The CF42 plants were homozygous and the CF21 plants were hemizygous for the transgenic locus. Within each group of plants, reciprocal crosses were made in order to evaluate whether the FLP-mediated excisions were affected by the source of Cre::FLP activity, which could be delivered through pollen grains or produced in egg cells. All F1 kernels (embryos and endosperm) originating from the crosses with the FT transgenic parental plants should produce yellow fluorescence unless the YFP gene is excised and lost. In such a case, red fluorescence should be observed instead. For the CF42 crosses, red fluorescence (either uniform or dotted expression) was observed in all F1 kernels, indicating that the FLP-mediated excisions were ubiquitous in these kernels (Table 1). In contrast, uniform expression of YFP was found in a large fraction of the F1 kernels in the progeny of CF21 plants that delivered a weak expression of Cre::FLP. The dark phenotypes may represent poorly developed kernels, accidental modifications of the excision site or pollen contamination. There was no evident difference in the frequency of Ds-Red2 gene activation between reciprocal crosses, suggesting that the active FLP protein could be provided by either pollen or egg cells (Table 1).

F1 kernels were germinated to produce leaf tissue samples for PCR analysis. The quantitative end-point PCR analysis was used to demonstrate the FLP-mediated excision of YFP in F1 seedlings (Figure 3). The YFP panel in Figure 3 contains a selected set of samples, illustrating our scoring criteria for the YFP copy number estimations. As expected, one copy of FLP (a part of the Cre::FLP fusion) was identified in all but one of the DNA samples. A single copy of the moPAT marker (a part of the recombination substrate not flanked by FRTs) was also found in all but a few DNA preparations. However, a number of DNA samples from the CF42 crosses were scored as containing less than one copy of YFP (Figure 3). The fraction of F1 plants identified as Red2+ that had lost the YFP gene was calculated (see Table 2, top section). The stringent threshold cut-off value for calling the YFP null genotype (YFP–) may result in an underestimation of the frequency of complete FLP-mediated excisions in the F1 plants. No YFP-negative genotypes were identified from the CF21 crosses, thus confirming our earlier observations that this particular Cre::FLP event does not support efficient FLP-mediated site-specific recombination in the F1 generation (Table 2, bottom section).

Figure 3.

Quantitative end-point polymerase chain reaction (PCR) analysis of DNA from the F1 progeny of the FT01 × CF42 cross. The PCRs were set for the amplification of internal fragments of coding sequences from the Cre::FLP, moPAT and YFP genes (mFLP, moPAT and YFP, respectively). The mFLP and moPAT panels illustrate a uniform amplification of the marker genes not affected by the FLP-mediated recombination, whereas the YFP panel demonstrates different amplification rates of the YFP coding sequence subjected to the excisions catalysed by the Cre::FLP fusion recombinase.

Table 2.  Polymerase chain reaction (PCR) genotyping of the F1/Red2+ (top section) and F1/YFP+ (bottom section) kernels from the CF42 and CF21 crosses, respectively. Quantitative PCR for the YFP coding sequence produced ambivalent results in a few DNA samples whose zygosity could not be called (presumably highly chimeric plants for YFP)
CrossNo. of plantsFLP+YFP+YFP–moPAT+
CF42 × FT015757262456
FT01 × CF425252282352
CF42 × FT095656341956
FT09 × CF425050291850
CF21 × FT01301230 030
FT01 × CF21281528 028
CF21 × FT09251825 025
FT09 × CF21181418 018

F1 seedlings were grown to maturity, test crossed to non-transgenic plants (HiII) and the F2 kernels were analysed for segregation of the YFP and Red2 markers (Figure 4). The E1 ear was produced from a YFP+ kernel originating from the CF21 × FT09 cross. The YFP+ phenotype segregated at 1 : 1, providing no evidence of FLP activity (121 YFP+ : 133 Dark). The D4 ear was produced on a Red2+ plant from the CF42 × FT09 cross. About 50% of all kernels showed a uniform red fluorescence in this and other Red2+ plants (D4 ear: 165 Red2+ : 206 Dark). The approximate 1 : 1 segregation ratio and no evidence of yellow fluorescence in the F2 kernels may have indicated the complete excision of the YFP gene, which was subsequently lost. Other Red2+ plants from the same cross (CF42 × FT09) produced both YFP+ and Red2+ kernels in varied numbers (D5 ear: 97 Red2+ : 52 YFP+ : 157 Dark). They were considered to be chimeric plants for the FLP-mediated excision product. The YFP coding sequences were identified in a number of DNA preparations from the Red2+ seedlings by PCR (Figure 4b). From a random sample of 117 F1 progeny plants that were test crossed to non-transgenic plants, 70 ears produced the Red2+ kernels only, 46 ears contained both the Red2+ and YFP+ phenotypes and one ear contained YFP+ kernels only. The frequency of complete FLP-mediated excision (Red2+ phenotypes) was estimated to be about 40%–60% based on both the F1 PCR and F2 segregation analyses.

Figure 4.

Phenotypic variants segregating in the F2 generation of test crosses. (a) F1 plants from the initial cross of CF × FT were grown to maturity and test crossed to non-transgenic plants (HiII). Ears were photographed under 470 ± 25-nm light. The E1 ear was produced on an F1/YFP+ plant from the cross of CF21, whereas ears D4 and D5 were obtained from the F1/Red2+ plants of the CF42 cross. (b) Polymerase chain reaction (PCR) amplification of the YFP coding sequence in DNA preparations from F1/YFP+ seedlings (YFP+ embryos, CF21 × FT09) and F1/Red2+ seedlings (Red2+ embryos, CF42 × FT09). Primers were designed to amplify a 659-bp PCR product. Lanes marked as E1, D4 and D5 correspond to the F1 ears shown in (a). Weak PCR amplification bands may represent chimeric plants for the YFP gene.

The F2 Red2+ seedlings produced the expected PCR amplification signal from the maize ubiquitin promoter and the Ds-Red2 coding sequence junction site (Figure 5a). They did not contain the YFP expression unit, and the original sequence around the left FRT site could not be amplified by primers placed on either side of the FRT site. The PCR products were sequenced to show that the FRT sites recombined as expected from the FLP-mediated reaction (data not shown). The FLP coding sequence segregated in the population of Red2+ seedlings. PCR results were verified by Southern blot analysis (Figure 5b). The 2.4-kb XhoI-PacI restriction fragment containing the excision footprint hybridized to the FRT probe in the Red2+ seedlings. The R1, R2 and R3 plants exemplified the FLP-mediated excisions, with the excision footprint segregating out from the Cre::FLP fusion gene. It should be noted that the 3.2-kb fragment from the Cre::FLP fusion was not found in the R1 plant originally scored by PCR as containing the FLP coding sequence. This may have indicated a possible PCR error that was identified by Southern blotting.

Figure 5.

Molecular analysis of the FLP recombination substrates and products in the F2 progeny of test crosses. (a) Polymerase chain reaction (PCR) amplification of the junction sites between the maize ubiquitin promoter and the Red2 coding sequence (Ubi::Red2) from the F2 Red2+ seedlings (R1–R6). The original sequence around the left side FRT site (left FRT) was amplified only in the FRT substrate-containing parental plants (FT) and in an F2 YFP+ seedling from the CF21 cross (Y1). The maize ubiquitin promoter/YFP coding sequence junction was amplified from the same DNA preparations (Ubi::YFP). The FLP coding sequence was found in the DNA preparations from a Cre::FLP parental plant (CF), one of the non-fluorescent F2 seedlings (NF), and was found segregating in the F2 Red2+ seedlings (R1, R4, R5). Lane HiII represents PCRs using DNA preparations from a non-transgenic plant. (b) Southern blot analysis of genomic DNA isolated from the F2 seedlings. The genomic DNA was digested with a pair of PacI and XhoI restriction enzymes and hybridized to the FRT probe. The 2.4-kb fragment diagnostic of the excision footprint was found in F2 Red2+ seedlings (R1–R5). The 3.2-kb fragment, also hybridizing to the probe, originated from the Cre::FLP vector and was identified in the parental CF plant (CF) and two F2 Red2+ plants (R4 and R5). Two fragments (3.4 and 4.0 kb) representing the FLP substrate in the FT42 parental plant (FT) were not detected in the F2 Red2+ plants. (c) All analysed F2 YFP+ seedlings originating from the CF42 cross contained the YFP coding sequence (YFP cd panel); none of them produced the PCR amplification product for the FLP coding sequence (FLP cd panel). All analysed F2 YFP+ seedlings originating from the CF21 cross contained the YFP coding sequence (YFP cd panel) and segregating FLP coding sequence (FLP cd panel).

The F2 YFP+ kernels were identified either in the control CF21 × FT09 crosses (ear E1, Figure 4) or in the ears produced by chimeric F1 plants of the CF42 × FT09 crosses (ear D5, Figure 4). These YFP+ kernels contained the original, not recombined, FRT excision substrate. Interestingly, no evidence of the Cre::FLP coding sequence was found in the F2 YFP+ kernels produced from F1 chimeric plants of the CF42 crosses, but the Cre::FLP gene segregated within the pool of the F2 YFP+ kernels from the CF21 cross (Figure 5c). In addition, when the F2 YFP+ plants (original CF42 × FT09 cross) were again crossed to the FLP-expressing FP20 parental plants (Figure 2), the Red2+ phenotype was produced in about 50% of F3 kernels (Table 3). Thus, the original FRT substrates were not mutated or otherwise rendered non-functional in the F1 generation. These observations indicate that the original FRT substrate can survive the presence of FLP during the development of chimeric F1 plants, but its transmission to the F2 progeny is limited by the strong expression of FLP during embryo development or seedling germination.

Table 3.  FLP-mediated excisions in the F3 progeny of crosses between the F2 YFP+ plants heterozygous for the FLP substrate and the FLP-expressing tester plants FP20. The letter E and number denote the individual F2 YFP+ kernels produced from the original F1 FT01 × CF42 crosses, as shown in Figure 4a (ear D5). These kernels did not contain the Cre::FLP fusion gene (Figure 5c). A 1 : 1 Red+/Dark segregation ratio indicated 100% FLP excision efficiency. The excision substrates did not recombine during the development of F1/F2 generation plants, but preserved their recombination capacity; no mutations accumulated at the FRT sites. They were recombined by FLP in the F3 progeny
CrossDarkRed+YFP+Chimeric
FT01 E.4.6 × FP2012621217 33
FT01 E.5.7 × FP20 695 660 23
FT01 E.1.8 × FP20 643 616 73
FT01 E.2.9 × FP20 556 553 13
FT01 E.2.11 × FP2012201157372
FT01 E.1.13 × FP2011221114163
FT01 E.8.14 × FP20 700 672 00

Cre-mediated excisions in the F1 and F2 progeny

Cre-mediated excisions were monitored by PCR using a pair of primers specific to the flanking sequences around the loxP sites in the PHP 18524 recombination substrate (Figure 1). The expected 902-bp amplification product of the Cre-mediated site-specific recombination was found in the F1 seedlings from the CF42 × LX18 reciprocal crosses (Figure 6, F1, panels a–f). No footprints of Cre-mediated excisions were identified in the crosses employing the CF21 parental plant (Figure 6, F1, panels g and h). Evidence of Cre-mediated excisions of YFP was found in about 70%–80% of analysed F1 plants (excision footprints were found in 36 of 47 plants containing the Cre::FLP gene and the moPAT marker for the loxP substrate). Southern blot analysis verified Cre activity in the F2 seedlings (Figure 6, F2). The 2.9-kb DNA fragments of the Cre-mediated excision products were identified as either the only fragments hybridizing to the moPAT probe (complete excision of two YFP alleles: seedlings p9, p15, p18) or in combination with the parental 6.1-kb fragment (seedlings p16, p28, p14, p23). As the intensities of the two bands were not equal, the 6.1-kb fragment may indicate chimeric seedlings for the excision product.

Figure 6.

loxP excision footprints in the F1 and F2 progeny of the LX18 crosses. A 902-bp polymerase chain reaction (PCR) amplification band of the loxP excision footprint was found in the F1 seedlings of LX × CF crosses (panels a–f). For each panel, the maternal plant is listed first. The control CF21 crosses did not produce amplification signals in the F1 progeny (panels g and h). Bottom: Southern blot analysis of the loxP excision footprints in the F2 generation. Genomic DNA was digested with a pair of Eco47III and SacI restriction enzymes and hybridized to the moPAT coding sequence probe. A band corresponding to two fragments of the same size (6.13 bp) was detected in the DNA preparations from the parental LX18 plant (LX), but not from the CF42 parental plant (CF). Lanes p16 to p23 contained DNA samples from the F2 seedlings of a test cross. Three DNA preparations (p9, p15 and p18) showed just one 2.9-kb band hybridizing to the probe, indicating the complete excision of two YFP alleles. The other samples (p16, p28, p14 and p23) produced two bands similar in size to the original excised YFP alleles.

The episomal excision product should produce the YFP+ phenotype on re-integration into chromatin structures. YFP+ kernels were not found in the population of F1 kernels. This indicates that the re-integration of the episomal Cre-mediated recombination products was not a frequent event.

Discussion

Site-specific recombinases have been successfully used for chromosomal deletions, inversions, integrations and translocations (Sorrell and Kolb, 2005). Although Cre recombinase has been evaluated extensively in plant species, including rice and maize (Srivastava and Ow, 2001a; Zhang et al., 2003; Sreekala et al., 2005), FLP-mediated recombination has produced inconsistent results in eukaryotic cells. The arguments against FLP stem from observations indicating that the synaptic complexes formed by Cre are more stable than the FLP complexes with FRT sites (Ringrose et al., 1998). Cre has a higher affinity than FLP for its target, and a higher cooperativity in the sequential binding of monomers to target sites. However, FLP can drive the recombination reaction to completion under optimal conditions, whereas Cre-mediated excisions usually do not exceed 75% of the available substrate (Buchholz et al., 1996). In plants, although maize cells were originally used to demonstrate the functionality of the FLP-FRT system (Lyznik et al., 1996), transgenic maize plants expressing FLP did not perform well in the excision of a FRT-flanked marker gene in genetic crosses (Kerbach et al., 2005). These observations were in contrast with the efficient excisions catalysed by Cre in the progeny of maize crosses (Zhang et al., 2003; Kerbach et al., 2005). Genetic crosses in tobacco and the application of heat-inducible FLP in Arabidopsis resulted in predominantly chimeric plants for the FLP excision products (Kilby et al., 1995; Gidoni et al., 2001; Latvala-Kilby and Kilby, 2006). However, it seems that the FLP-FRT system performed similarly to the Cre-lox system in rice (Radhakrishnan and Srivastava, 2004; Hu et al., 2008). In this paper, using the Cre::FLP fusion, we performed a direct comparison of FLP and Cre activity in maize under similar experimental conditions, and found similar expression levels from the same transformation events. Nevertheless, the linkage of two recombinases may selectively affect their performance, and different chromosomal locations of the recombination substrates for FLP and Cre may also influence the recombination rates. If a precise comparison of the performance of FLP and Cre in maize is required, additional work will be needed.

Nevertheless, this work demonstrates that both FLP and Cre can perform well in maize. Apparent inefficiencies of the FLP recombinase in plant transformation experiments may be related to particular experimental conditions. The production of transgenic, well-performing plants is a complex undertaking. The Cre::FLP fusion gene comprises two coding sequences: one for the Cre site-specific recombinase and the other for the FLP site-specific recombinase. They were codon modified for optimal performance in maize cells, and the potato ST-LS1 gene intron was incorporated into the Cre coding sequence to enhance its expression. The strong maize ubiquitin promoter was used to drive the expression of the fusion gene. Forty-four primary T0 transgenic events were produced (19 T0 plants were selfed; of these, eight events were single-copy insertions with three events showing moderate to strong activity of FLP and Cre in the leaf tissues). One transgenic event (CF42) with a strong expression of FLP and Cre and another event providing a weak expression of the Cre::FLP fusion gene (CF21) were selected. The latter failed to induce detectable site-specific recombination in the maize genetic crosses, and was used mostly as a control treatment. Such variability of FLP activity originating from different transformation events indicates that the overall effectiveness of FLP in plant cells is very sensitive to the expression levels or FLP protein accumulation. Therefore, it is difficult to compare our results with other biological systems that have been used to test the yeast FLP recombinase. It may be concluded that the FLP enzyme can produce chromosomal site-specific recombination events in maize at a frequency acceptable for practical applications.

We observed the site-specific recombination of FRT and loxP sites catalysed by a dual-function Cre::FLP fusion recombinase. Although the specific mechanisms of site-specific recombination catalysed by FLP and Cre may differ, both recombinases form synaptic complexes by a sequential assembly of four monomers around two recombination target sites (Grindley et al., 2006). Protein–protein interactions are essential in bringing two recombination sites into an appropriate alignment and to bend the sites to produce active synaptic complexes (Luetke and Sadowski, 1995; Guo et al., 1999). The presence of additional polypeptides fused to the FLP or Cre monomers in the Cre::FLP fusion protein was not detrimental to the DNA binding properties of the monomers, the cooperative assembly of the tetrameric structures or the catalytic properties of synaptic complexes. Indeed, the synthesis of the Cre::FLP fusion gene was encouraged by earlier work which tested the fusion of steroid binding domains to FLP and Cre in order to regulate their activity (Logie and Stewart, 1995; Kellendonk et al., 1996). The FLP activity of our Cre::FLP fusion was comparable with the FLP recombinase activity in maize cells, as tested in transient and stable transformation experiments.

The crossing strategy for the induction of site-specific recombinations worked well for maize. The estimated efficiency of FLP-mediated complete excisions was about 40%–60% in the F1 population. A similar efficiency was observed for Cre-mediated excisions. Our experimental results are comparable with the Cre excisions in maize reported by Zhang et al. (2003), except that we found non-recombined substrates in about 40% of F1 plants. In chimeric maize plants, the original FLP substrates were found only in the F2 kernel populations that did not contain the Cre::FLP gene. Despite the long period of plant development and the presence of the Cre::FLP protein in all cells of the F1 plants, the non-recombined FRT substrates reached the F2 zygotes. This observation supports the conclusion that excisions occur mainly during the early stages of embryo development. Similar results have been reported for FLP-mediated site-specific recombination in rice (Hu et al., 2008). Whether such observations are related to the intrinsic properties of the maize ubiquitin promoter driving the expression of Cre::FLP, or indicate particular properties of the site recombination reactions, needs to be assessed further. The latter conclusion seems to be less likely, as the bombardment of leaf tissues with FLP routinely produces recombination events, and heat shock-inducible FLP expression is known to generate clonal sectors in somatic plant cells (Latvala-Kilby and Kilby, 2006). Evidently, fluctuations in maize ubiquitin promoter (otherwise considered as a constitutive promoter) activity within different tissues can have a significant impact on FLP recombinase performance in maize.

The overall results presented in this article demonstrate that both FLP- and Cre-mediated site-specific recombination, in conjunction with the crossing strategy, provide a practical method for the removal of undesirable DNA chromosomal fragments from the maize genome. The method does not require a dedicated selection for the recombination events. The Cre::FLP fusion protein is an effective reagent for the catalysis of site-specific recombination in maize.

Experimental procedures

Vector constructions

The maize codon-optimized moCre recombinase gene containing the ST-LS1 intron (Vancanneyt et al., 1990) was fused in frame with the maize codon-optimized FLPm recombinase gene. The stop codon of moCre was replaced by a 45-bp linker sequence encoding the GGGSGGGSGGGSDPT peptide fused directly to the start codon of FLPm. The sequence of the junction site was: GACGGCGGCGGTAGCGGCGGTGGCTCTGGCGGTGGCTCGGATCCAACAATG (the last ‘D’ Cre codon and the first ‘M’ FLP codon are shown in italic). The Cre::FLP fusion gene was driven by the maize ubiquitin promoter (Christensen and Quail, 1996) modified to contain the yeast 2-µm plasmid inverted repeat integrated into the 5′ untransformed region (5′ UTR) intron at the EcoRI restriction site. The maize codon-optimized moCAH gene (cyanamide hydratase) controlled by the maize ubiquitin promoter was used as a selectable marker gene. The 34-bp loxP site was also integrated between the moCAH gene and the left border.

The substrates for site-specific recombination, PHP 18656 and PHP 18524, were produced by in vitro co-integration of two vectors, each containing a single FRT or loxP site, respectively. The acceptor vectors were built on the pSB11 plasmid backbone (Komari and Kubo, 1999). For the FLP in vitro reaction, the moPAT expression unit (driven by the rice actin promoter) was integrated as the HindIII-XmaI fragment into the corresponding sites of pSB11. This vector was supplemented with the Ds-Red2 expression cassette containing the yeast 2-µm inverted repeat (FRT site) integrated into the first maize ubiquitin intron. The donor vector contained a YFP expression unit controlled by the maize ubiquitin promoter supplemented with the FRT site outside of the YFP cassette. For the Cre in vitro reaction, the HindIII-StuI fragment containing the loxP site was integrated into the HindIII and NotI sites of the pSB11 vector containing the moPAT expression unit driven by the maize ubiquitin promoter. The donor vector contained the YFP expression unit controlled by the maize ubiquitin promoter integrated into the HindIII and EcoRI sites of pUC18. The loxP site was integrated between the promoter and the YFP coding sequence at the EcoRI site of the first ubiquitin intron. The in vitro site-specific recombination reactions contained 25 mm Tris-HCl, pH 7.4, 1 mm ethylenediaminetetraacetic acid (EDTA), 1 mm dithiothreitol (DTT), 5% glycerol, 0.1 mm NaCl, 0.5 µg of an acceptor vector, 1 µg of a donor vector and either 1.4 µg of FLP protein or Cre protein in a total volume of 10 µL. Incubations were for 60 min at 30 °C. Two microlitres of the incubation mixture were used for the transformation of DH5αEscherichia coli competent cells (library efficiency DH5α competent cells; Invitrogen, Carlsbad, CA, USA) according to the manufacturer's specifications. Bacterial colonies, grown at 37 °C overnight in spectinomycin-containing (100 mg/L) Luria–Bertani (LB) medium, were transferred into 2 mL of ampicillin-containing (100 mg/L) liquid LB medium for the identification of double-antibiotic-resistant, co-integrative plasmids. Subsequently, electroporation was used to integrate the pSB11-based vectors into the super-binary vector pSB1 residing in Agrobacterium tumefaciens strain LBA 4404. The fluorescent protein genes were obtained from Clontech Laboratories, Palo Alto, CA, USA (Matz et al., 1999).

Production of transgenic plants

Zea mays (‘HiII’) immature embryos were transformed by a modified Agrobacterium-mediated transformation procedure described by Djukanovic et al. (2006). Immature embryos (10–12 days old) were dissected from sterilized kernels and placed into a liquid medium [4.0 g/L N6 Basal Salts (Sigma C-1416, Sigma-Aldrich, St. Louis, MO, USA), 1.0 mL/L Eriksson's Vitamin Mix (Sigma E-1511), 1.0 mg/L thiamine HCl, 1.5 mg/L 2,4-dichlorophenoxyacetic acid (2,4-D), 0.690 g/L l-proline, 68.5 g/L sucrose, 36.0 g/L glucose, pH 5.2]. After embryo collection, the medium was replaced with 1 mL of Agrobacterium suspension at an optical density (OD) of 0.35–0.45 at 550 nm. After incubation for 5 min at room temperature, the embryo suspension was poured on to a plate containing 4.0 g/L N6 Basal Salts (Sigma C-1416), 1.0 mL/L Eriksson's Vitamin Mix (Sigma E-1511), 1.0 mg/L thiamine HCl, 1.5 mg/L 2,4-D, 0.690 g/L l-proline, 30.0 g/L sucrose, 0.85 mg/L silver nitrate, 0.1 nm acetosyringone, 3.0 g/L Gelrite, pH 5.8. Embryos were incubated in the dark for 3 days at 20 °C, followed by 4 days of incubation in the dark at 28 °C, and then transferred on to new plates containing 4.0 g/L N6 Basal Salts (Sigma C-1416), 1.0 mL/L Eriksson's Vitamin Mix (Sigma E-1511), 1.0 mg/L thiamine HCl, 1.5 mg/L 2,4-D, 0.69 g/L l-proline, 30.0 g/L sucrose, 0.5 g/L 2-(N-morpholino)ethanesulphonic acid (MES) buffer, 0.85 mg/L silver nitrate, 3.0 mg/L Bialaphos for moPAT selection or 100 mg/L carbenicillin, 6.0 g/L agar, pH 5.8. Embryos were subcultured every 3 weeks until transgenic events were identified. Somatic embryogenesis was induced by transferring a small amount of tissue on to regeneration medium containing 4.3 g/L Murashige and Skoog (MS) salts (Gibco 11117; Gibco, Grand Island, NY, USA), 5.0 mL/L MS Vitamins Stock Solution, 100 mg/L myo-inositol, 0.1 µm abscisic acid (ABA), 1 mg/L indoleacetic acid (IAA), 0.5 mg/L zeatin, 60.0 g/L sucrose, 1.5 mg/L Bialaphos, 100 mg/L carbenicillin, 3.0 g/L Gelrite, pH 5.6. The plates were incubated in the dark for 2 weeks at 28 °C. All material with visible shoots and roots was transferred to medium containing 4.3 g/L MS salts (Gibco 11117), 5.0 mL/L MS Vitamins Stock Solution, 100 mg/L myo-inositol, 40.0 g/L sucrose, 1.5 g/L Gelrite, pH 5.6, and incubated under artificial light at 28 °C. One week later, plantlets were moved into glass tubes containing the same medium and grown until they were sampled and/or transplanted into soil.

PCR

DNA was extracted by placing two punches of leaf tissue, two stainless steel beads and 300 µL of Puregene® cell lysis solution (Qiagen, cat.# 158912, Qiagen, Valencia, CA, USA) into each tube of a Mega titre rack. The samples were homogenized on a Genogrinder at 1650 r.p.m. for 30–60 s, followed by incubation at 65 °C for 50 min. After the addition of 100 µL of Puregene® protein precipitation solution (Qiagen, cat.# 158912), the plates were centrifuged at 3500 g for 15 min at 15 °C. The supernatants (200 µL) were mixed with 200 µL of cold isopropanol and centrifuged at 3500 g for 15 min at 15 °C. The pellets were rinsed by adding 250 µL of 70% cold ethanol, dried overnight at room temperature and resuspended in 100 µL of TE buffer (10 mm Tris, 1 mm EDTA, pH 8.0). Where available, the DNA preparations for Southern blot analysis were used for PCR (Figure 5a). PCRs contained 200 ng of DNA template, 500 nm of each primer and 10 µL of 2 × RedExtractandAmpPCR mix (Sigma, St. Louis, MO, USA) in a total volume of 20 µL. The initial incubation was at 94 °C for 4 min, followed by 35 cycles at 94 °C for 1 min, 66 °C for 1 min and 72 °C for 1 min. Ten microlitres of each PCR were analysed on a 1% agarose gel stained with ethidium bromide.

The sequences of the primers used in the experiments were as follows: FLP excision footprint: CGTGTTTGTGTTAGATCCGTGCTG, GCCCTTGGTCACCTTCAG, 1465-bp product; Cre excision footprint: CACTCAGCAAGCTGGTACGATTGTAATA, CGCTGGTTGGTGTCCGTTAG, 902-bp product; YFP coding sequence: GCCTGAAGGAGGAGATGACCATG, CATCTTGCTGGGCACGCTCTTG, 563-bp product; FLP coding sequence: TCGGTCACGAGGCACTGGATGAT, CCCCAGTTCGACATCCTCTGCAA, 659-bp product; FRT site: CGTGTTTGTGTTAGATCCGTGCTG, CAGTTCGATGTAACCCACTCGTG, 1431-bp product; moPAT: GCTCGCCCTGGATTTTGGTTTTAG, CGAGACCTCCACCGTGAACTTC, 686-bp product.

End-point quantitative PCR was performed using 2 µL of DNA template (about 100 ng) isolated by a modified alkaline lysis method from one punch of leaf tissue (Truett et al., 2000). For each gene, a forward primer, a reverse primer and a 6-carboxyfluorescein (FAM)-based fluorogenic probe were used to quantify the targeted DNA sequence. The reactions were normalized to an amplification signal from an endogenous gene detected by a VIC-based fluorogenic probe (Applied Biosystems, Foster City, CA, USA). The amplification reactions for the marker genes were run simultaneously with the normalizing gene in a single tube reaction. On completion of quantitative PCR, all raw data were used to calculate the delta cycle threshold (dCT) values, and quick positive and negative quantitative PCR calls (if needed) were made according to the dCT estimations. A PCR running five cycles later than the normalizing gene was called negative.

Southern blotting

Leaf tissue (about 1–2 g fresh weight) was dried, ground and resuspended in 10 mL of Puregene® Cell Lysis Solution (Gentra Systems, Inc., Minneapolis, MN, USA). The DNA extracts were incubated for 30 min at 64 °C with shaking at 400 r.p.m., followed by centrifugation at 3220 g for 10 min, and the supernatants were mixed with 5 mL of phenol–chloroform (1 : 1) solution, followed by centrifugation at 3220 g for 10 min. DNA was precipitated by adding the same volume of isopropanol, spun down for 10 min at 3220 g and resuspended in 5 mL of TE buffer, pH 8.0, 0.4 mL of ethidium bromide (10 mg/mL) and 5 g of caesium chloride. The mixture was centrifuged overnight (12–17 h) at 390 000 g. The DNA extraction and ethidium bromide removal were essentially performed as described in Sambrook et al. (1989). The final DNA preparations were dissolved in TE buffer. Ten micrograms of DNA from each sample were digested overnight with selected restriction enzymes, and the DNA fragments were separated in a 1% agarose gel run at 35 mV overnight. The TurboBlotter and Blotting Stack (Schleicher & Schuell, Keene, NH, USA) were used to transfer DNA on to a nylon membrane, as described in the manufacturer's manual. The DNA fragments were attached to the membrane by ultraviolet (UV) irradiation at 1.2 kJ/m2 in a UV Stratalinker (Stratagene, Cedar Creek, TX, USA), and the blots were pre-hybridized for 2–3 h in 20 mL of ExpressHyb hybridization solution (Clontech Laboratories) at 65 °C. The random prime labelling system (Amersham Pharmacia Biotech, Piscataway, NJ, USA) was used with Redivue [32P]dCTP to produce radioactively labelled DNA fragments according to the supplied protocol. The FRT and moPAT probes were prepared by isolating the corresponding vector fragments. Hybridizations were incubated overnight at 65 °C. The blots were washed twice with 1% SSPE/0.1% sodium dodecylsulphate (SDS) solution for 15 min at 65 °C, followed by two additional washes with 0.1% SSPE/0.1% SDS solution under the same conditions (0.15 m NaCl, 0.01 m sodium phospate, 1mm EDTA). The blots were exposed to Kodak BioMax MR film overnight at –70 °C (Eastman Kodak Company, New York, NY, USA).

Monitoring fluorescence in plant tissues

Ears or kernels were illuminated by 36 light-emitting diodes (LEDs) emitting 470 ± 25-nm light (Edmund Optics Inc., New York, NY, USA). Pictures were taken by a digital camera with a red filter, allowing for a colour differentiation between YFP and DsRed2 fluorescence (orange for red fluorescence and yellow/green for yellow fluorescence). For a closer inspection, tissue samples were analysed under a Nikon SMZ1500 stereomicroscope (Nikon Instruments, Melville, NY, USA) with sets of filters designed by the manufacturer to monitor YFP (490–510-nm excitation, 520–550-nm emission barrier) and DsRed (530–560-nm excitation, 590–650-nm emission barrier) fluorescence.

Acknowledgements

The authors thank Jessica Stagg, Shifu Zhen and Susan Nilles for technical assistance in producing the transgenic plants.

Ancillary