Site-specific gene integration in rice genome mediated
by the FLP–FRT recombination system
Soumen Nandy1 and Vibha Srivastava1,2,*
1Department of Crop, Soil & Environmental Sciences, University of Arkansas, Fayetteville, AR, USA
2Department of Horticulture, University of Arkansas, Fayetteville, AR, USA
Received 5 September 2010;
revised 14 October 2010;
accepted 18 October 2010.
*Correspondence (Tel: 479 575 4872; fax:
479 575 7465; email vibhas@uark.edu)
Keywords: FLP–FRT, Cre-lox, DNA
recombination, site-specific gene
integration, gene targeting, rice
transformation.
Summary
Plant transformation based on random integration of foreign DNA often generates
complex integration structures. Precision in the integration process is necessary to
ensure the formation of full-length, single-copy integration. Site-specific recombina-
tion systems are versatile tools for precise genomic manipulations such as DNA exci-
sion, inversion or integration. The yeast FLP–FRT recombination system has been
widely used for DNA excision in higher plants. Here, we report the use of FLP–FRT
system for efficient targeting of foreign gene into the engineered genomic site in
rice. The transgene vector containing a pair of directly oriented FRT sites was intro-
duced by particle bombardment into the cells containing the target locus. FLP activity
generated by the co-bombarded FLP gene efficiently separated the transgene con-
struct from the vector-backbone and integrated the backbone-free construct into
the target site. Strong FLP activity, derived from the enhanced FLP protein, FLPe, was
important for the successful site-specific integration (SSI). The majority of the trans-
genic events contained a precise integration and expressed the transgene. Interest-
ingly, each transgenic event lacked the co-bombarded FLPe gene, suggesting
reversion of the integration structure in the presence of the constitutive FLPe expres-
sion. Progeny of the precise transgenic lines inherited the stable SSI locus and
expressed the transgene. This work demonstrates the application of FLP–FRT system
for site-specific gene integration in plants using rice as a model.
Introduction
Transgene integration mediated by site-specific recombina-
tion (SSR) systems is an effective approach for generating
precise single-copy integrations in higher plants (reviewed
by Srivastava and Gidoni, 2010). Most plant species can
be transformed by Agrobacterium or particle bombard-
ment–mediated gene delivery; however, the rate of single-
copy integrations in these protocols is fairly low (Dai et al.,
2001; Travella et al., 2005; De Buck et al., 2009). More-
over, as the introduced DNA integrates randomly into the
genome, it is impossible to predict the structure of the
transgene locus. Therefore, it is highly desirable to develop
strategies for controlling the transgene integration pro-
cess. These strategies will not only improve the rate of
single-copy integrations; they will also allow rapid charac-
terization of the locus, and precise full-length integration
of large constructs containing multiple transgene units into
a single site, a feature most desirable for engineering poly-
genic traits. As precise single-copy loci tend to express at
higher levels compared to complex locus and maintain
expression levels (correlated with gene dosage) through
successive generations (Day et al., 2000; Schubert et al.,
2004; Chawla et al., 2006; De Buck et al., 2007; Nanto
et al., 2009; Akbudak et al., 2010), site-specific inte-
gration (SSI) is an attractive feature for transformation
technologies. While homologous-recombination-mediated
gene targeting is now possible in higher plants (Terada
et al., 2002; Shukla et al., 2009; Townsend et al., 2009),
well-characterized SSR systems continue to serve as excel-
lent tools as they carry out a simple and accurate reaction
in a variety of plant cells, without causing ‘off-target’
genetic aberrations (Coppoolse et al., 2003; Ream et al.,
2005). Several SSR systems are functional in plants, among
ª 2010 The Authors
Plant Biotechnology Journal ª 2010 Society for Experimental Biology and Blackwell Publishing Ltd 1
Plant Biotechnology Journal (2010), pp. 1–11 doi: 10.1111/j.1467-7652.2010.00577.x
which P1 phage Cre-lox and yeast FLP–FRT are the two
most widely used and extensively characterized systems
(Luo and Kausch, 2002; Ow, 2002; Srivastava and Gidoni,
2010). A number of SSR systems, including Cre-lox and
FLP–FRT, are able to efficiently excise DNA fragments from
transgene locus via a recombination between two sites
placed in the same orientation (Grønlund et al., 2007;
Gidoni et al., 2008; Thomson et al., 2009, 2010). How-
ever, only a few SSR systems have been successfully used
for SSI of foreign gene, which involves a recombination
between two sites in trans. Of these, Cre-lox has so far
displayed a superior SSI efficiency in plant cells (Srivastava
and Gidoni, 2010). Because Cre-mediated recombination
between two identical lox sites is freely reversible, use of a
combination of mutant sites is critical in stabilizing the
integration structures (Albert et al., 1995). Developing
alternative systems for SSI application is highly desirable to
incorporate additional features in the transformation plat-
form such as marker removal. Selection of SSI events
involves incorporation of a selectable marker gene, which
cannot be removed from the genome unless a removal-
strategy is incorporated in the transformation vector.
Cre-lox-mediated marker removal is a popular strategy
owing to its high efficiency (Gilbertson, 2003; Gidoni
et al., 2008). Based on which, a marker-free site-specific
gene integration strategy has been proposed that involves
the use of an alternative SSR system for site-specific gene
integration followed by marker removal by highly efficient
system such as Cre-lox (Srivastava and Ow, 2004; Darbani
et al., 2007; Nanto and Ebinuma, 2008; Fladung and Bec-
ker, 2010). Because a number of studies have shown that
FLP–FRT recombination occurs precisely in a variety of
plant species (Luo and Kausch, 2002; Gidoni et al., 2008),
it is an obvious choice for developing a new application.
However, initial attempts on using FLP–FRT for site-specific
gene integration in plants were not successful (Kerbach
et al., 2005; Srivastava and Gidoni, 2010).
Two different strategies for site-specific gene integration
have been developed, each producing the same end prod-
uct, the vector-backbone-free precise integration of the
defined DNA segment. The recombinase-mediated cas-
sette exchange (RMCE) strategy is based on the dual
recombination between a pair of recombination sites
resulting in the exchange of DNA cassettes between the
target locus and the donor DNA. To prevent the excision
or inversion of the DNA cassettes, two incompatible
recombination sites are used in the RMCE strategies,
e.g. the use of a wild-type lox and a mutant lox site in
Cre-lox-based RMCE strategy (Louwerse et al., 2007). The
co-integration strategy, on the other hand, consists of a
single recombination between two recombination sites in
trans resulting in the integration of the circular DNA mole-
cule into the genomic target. The circular molecule is gen-
erated de novo in the cell via recombination between two
directly oriented sites in the introduced DNA, separating
the gene cassette from the vector-backbone (Srivastava
and Ow, 2002). Recently, Li et al. (2009) showed FLP–FRT-
mediated gene integration in soybean, and Fladung et al.
(2010) presented a preliminary report on FLP–FRT-medi-
ated gene integration in hybrid aspen. Both groups used
the RMCE strategy based on the recombination between a
pair of FRT sites, resulting in the exchange of DNA cas-
settes between the target locus and the donor DNA.
Here, we demonstrate the use of FLP–FRT system for
the co-integration of a circular DNA into the genomic tar-
get site via a single FRT · FRT recombination in rice gen-
ome. In the research leading up to the present work, we
found that the co-integration strategy based on the
expression of recombinase gene from the target locus was
highly successful with Cre-lox system but not with the
FLP–FRT system. The expression of the FLP gene was
severely down regulated when an FRT site was incorpo-
rated between the promoter and the coding region of the
FLP gene, a component of the original co-integration strat-
egy developed using the Cre-lox system (Srivastava and
Ow, 2002). By supplying strong FLP activity through
co-bombardment of FLP gene, SSIs were obtained at
workable efficiency in two different target lines used in
the study. This work describes a robust platform for plant
transformation employing the biolistic gene gun for
DNA delivery and FLP–FRT recombination system for gene
integration.
Results
Molecular strategy
The target construct, pNS5, contains a target FRTL site
between the promoter and the coding sequence of FLP
gene (NCBI accession no. NC_001398), while the donor
construct, pAM18, contains a cassette flanked by FRT and
FRTR (Figure 1a, b). The FRTL and FRTR, each contain a sin-
gle-base mutation at -10 position in the left or the right
FLP-binding sequence as described by Senecoff et al.
(1988). FLP-mediated recombination between FRT and
FRTR is expected to occur rapidly resulting in the genera-
tion of a vector-backbone-free donor-circle containing
FRTR (Figure 1c). FLP-mediated integration of the donor
ª 2010 The Authors
Plant Biotechnology Journal ª 2010 Society for Experimental Biology and Blackwell Publishing Ltd, Plant Biotechnology Journal, 1–11
Soumen Nandy and Vibha Srivastava2
circle would then generate a precise co-integration struc-
ture consisting of transcriptional fusion of the selectable
marker gene (NPT) with the Ubi promoter in the genomic
target site (Figure 1d). The direct integration of the
pAM18 DNA into the target locus is highly unlikely as the
inter-molecular FRTL · FRTR reaction is kinetically disfa-
voured over the intra-molecular FRTR · FRT reaction
(Huang et al., 1991). However, if in rare cases the integra-
tion of pAM18 DNA occurred, the resulting SSI structure
will not be stable as it will contain pAM18-backbone
flanked by two FRT sites, which would rapidly recombine
out as a result of the highly efficient FRT · FRT recombina-
tion. The resulting SSI structure would be identical to that
generated by the integration of the donor circle (Fig-
ure 1d). Within this structure, there are distinct 5¢ and 3¢
junctions spanning FRTL+R and FRT sites. Formation of
FRTL+R, a relatively inactive site, is expected to stabilize
the integration locus. The target construct, pNS5, also
contains a hygromycin resistance gene, hpt, as a transfor-
mation selection marker, and a heat shock cre gene for
the excision of the lox-flanked marker genes. To allow
removal of the marker genes flanking the gene-of-interest
(e.g. GUS), pNS5 construct is flanked by two loxP sites.
Integration of the gene-of-interest that is also flanked by
inversely oriented loxP sites into the pNS5 target will gen-
erate the SSI structure in which the DNA fragment on
either side of the gene-of-interest will be flanked by
directly oriented loxP sites. Thus, all target locus-specific
genes and the NPT marker gene could be deleted by heat-
induction of Cre-lox recombination. Previous studies have
demonstrated the efficacy of soybean heat shock 17.5E
promoter, utilized in this construct, for effectively regulat-
ing cre expression for DNA excision (Zhang et al., 2003;
Wang et al., 2005). Thus, the long-term goal of pNS5-
target lines is to develop the method of marker-free site-
specific gene integration based on the strategy described
UBI FLP35S::hpt HSP::Cre L+R RBLB NPT Ubi::GUS
SSI
locus
6.8 1.72.8
p1 p2 p3 p4
0.7 0.5
UBI FLP35S::hpt HSP::CrepNS5 L RBLB
3>0.5 (a) 3.5 >1.3
NPT Ubi::GUS
R
(b) (c)
(d)
(e)
Ubi FLPe
pAM18
Donor
circle
pUbiFLPe
R R R
R R R R
E E7.8 E0.8 E0.9
E E3.5 E0.9
Probe 1
Probe 4
Probe 2Probe 3
R
>1.3
Figure 1 Design of DNA constructs and molecular strategy of site-specific gene integration. (a) pNS5 target construct contains three transcription
units flanked by loxP sites (open triangles): hygromycin phosphotransferase (hpt) gene driven by CaMV 35S promoter, cre gene driven by soybean
heat shock 17.5 E promoter (HSP), and FLP gene driven by maize ubiquitin (Ubi) promoter. Target FRTL site (shaded triangle) is incorporated into
the leader sequence of FLP gene. RB and LB represent T-DNA borders. The two loxP sites are designed to enable future marker excision using the
strategy described earlier by Srivastava and Ow (2004). (b) Donor vector pAM18 contains a promoterless neomycin phosphotransferase II gene
(NPT) and b-glucuronidase gene (GUS) driven by Ubi promoter between FRTR and FRT sites (shaded triangle) in the pBluescript SK backbone (grey
line). (c) The structure of donor circle generated from pAM18 upon FLP-mediated FRTR · FRT recombination. The circle contains the gene construct
and FRTR without the vector-backbone. (d) Structure of site-specific integration (SSI) locus. Integration of donor circle via FRTL · FRTR recombina-
tion traps Ubi promoter to activate NPT gene and forms a unique FRTL+R and FRT sites (shaded triangles). EcoRI (E) and EcoRV (R) sites for each
construct and sizes of intervening fragments are shown. Shaded bars represent the DNA probe fragments used for Southern hybridization. PCR
primers (p1–p4) and the expected PCR product sizes are also indicated. (e) pUbiFLPe vector contains FLPe gene driven by Ubi promoter in a pBlue-
script SK backbone (grey line). Each transcription unit contains a transcription termination signal of nopaline synthase gene (nos 3¢).
ª 2010 The Authors
Plant Biotechnology Journal ª 2010 Society for Experimental Biology and Blackwell Publishing Ltd, Plant Biotechnology Journal, 1–11
FLP–FRT-mediated gene integration in plants 3
earlier (Srivastava and Ow, 2004). In the future experi-
ments, a modified donor construct, containing the GUS
gene flanked by inversely oriented loxP sites, will be used
to carry out marker excision from the SSI locus. In the pres-
ent work, however, the efficiency of FLP–FRT-mediated
gene integration was tested without any interference of
Cre-lox recombination. Therefore, pAM18 lacks loxP sites.
Development and characterization of target lines
Thirty-nine transgenic events with target construct, pNS5,
were obtained using Agrobacterium-mediated transforma-
tion. EcoRV-digested genomic DNA of each line was sub-
jected to Southern hybridization with hpt, cre and FLP
probe fragments (see Figure 1a). Subsequently, PCR was
carried out using p1 and p4 on selected lines, and the
resulting 0.5-kb PCR product was sequenced. Seven of 39
NS5-lines were found to contain a single-copy of pNS5
T-DNA and the expected FRTL site in the PCR product
(Figure S1; data not shown). Two of these lines, A and B,
were used in the subsequent experiments, and FLP activity
was supplied by co-transformation of FLPe gene as these
target lines failed to express sufficient levels of FLP activity
(described below). T1 seeds of these lines were used to
generate callus on hygromycin-containing media. Use of
hygromycin was intended to eliminate the wild-type segre-
gants among T1 progeny.
FLP expression analysis
To explore the basis of sub-optimal FLP expression in the
NS5-lines, two FLP constructs (pUbiFLP and pUbiFrtFLP)
differing only in the presence of the FRT site, were com-
pared (Figure 2a, b). Each contains a strong Ubi promoter.
Co-bombardment of pUbiFLP and pRP9 on rice callus gen-
erated strong GUS activity (Figure 2a), while single bom-
bardment of pRP9 did not display any GUS activity. Thus,
FLP expression originating from pUbiFLP catalysed a strong
FRT · FRT reaction. However, co-bombardment of pUb-
iFrtFLP and pRP9, on wild-type callus, did not generate
any GUS activity (Figure 2b), suggesting that insertion of
FRT between Ubi promoter and FLP coding region ham-
pers gene expression. Next, two GUS constructs, differing
only in the presence of the FRT site, were compared
(Figure 2c, d). The FRT-containing vector, pUbiFrtGUS,
displayed detectable GUS activity, although it appeared
weaker than that of the FRT-lacking vector, pUbiGUS.
Thus, insertion of FRT sequence downstream of promoter
confers a general suppressive effect on the gene activity.
However, in case of pUbiFrtFLP, a greater suppressive
effect, presumably as a result of the tight binding of FLP
protein to FRT site, was observed. This hypothesis is sup-
ported by the analysis of the NS5-lines, which contain a
barely detectable level of FLP transcript (as per RT-PCR,
data not shown). The formation of the FLP–FRT complex
on the NS-5 locus could severely suppress transcription
resulting in the sub-optimal FLP activity in the target lines.
Generation and characterization of SSI lines
To supply strong FLP activity for site-specific gene integra-
tion, FLP gene was co-bombarded with the donor con-
struct, pAM18. An enhanced version of FLP recombinase,
FLPe, that displays 3–5 times higher efficiency compared
Ubi FLP
pUbiFLP
Ubi FLP
pUbiFrtFLP
Ubi GUS
pUbiGUS
Ubi GUS
pUbiFrtGUS
Ubi GUS
pRP9
Block
30 – 50 +
Ubi GUS
pRP9
Block
0+
>50
10 – 15
(a)
(b)
(c)
(d)
Constructs Bombarded GUS activity
(No. of blue spots)
Figure 2 Effect of FRT site on gene expres-
sion. The effect of the presence of FRT site
in the leader sequence of FLP and GUS
genes was studied by bombardment of plas-
mid DNA into wild-type callus cells and
staining them for GUS activity 48 h later.
(a, b) Co-bombardment of FLP constructs
differing only in the presence of FRT site
(grey triangle) with the FLP recombination
target construct, pRP9. (c, d) Single bom-
bardment of GUS constructs differing only in
the presence of FRT site. GUS activity for
each construct is given as number of blue
spots per bombarded plate.
ª 2010 The Authors
Plant Biotechnology Journal ª 2010 Society for Experimental Biology and Blackwell Publishing Ltd, Plant Biotechnology Journal, 1–11
Soumen Nandy and Vibha Srivastava4
to the wild-type FLP (wtFLP) (Buchholz et al., 1998), was
used in the present experiment. FLPe was developed by
cycling mutagenesis using DNA shuffling approach and
therefore contains 95% sequence homology with the
wtFLP. Callus derived from T1 seeds of target lines A and
B was bombarded with an equimolar mixture of pAM18
and pUbiFLPe. Cultures of target lines were raised on
hygromycin-containing media to avoid wild-type segre-
gants within T1 seeds. The bombarded callus was selected
on geneticin (100 mg ⁄ L) to isolate the putative SSI events.
Ten geneticin-resistant events were obtained in two sepa-
rate experiments using target line A (A1–A10), and three
events using target line B (B1–B3) (Tables 1 and 2). Nine
of these lines expressed GUS activity (Table 2). Genomic
DNA from all 13 callus lines was analysed by PCR for the
presence of SSI junctions and with Southern hybridization
for the accuracy of the SSI structure. All but one line (B3)
amplified the 5¢ junction fragment (0.6 kb) in a PCR using
primers p1 and p2 (Figure 3a, Table 2), although A9
amplified a larger than expected fragment. Sequencing of
this fragment from the selected lines (A1, A2, B11 and
B12) revealed the presence of accurate recombination
footprint consisting of FRTL+R sequence (data not shown).
PCR on 3¢ junction, using primers p3 and p4, amplified
the expected 0.5-kb fragment from only nine of the 13
lines (Figure 3a; Table 2). Sequencing of this fragment
from the above-selected lines also revealed the presence
of the accurate recombination footprint consisting of FRT
sequence. Southern hybridizat