Establishment of an enhancer trap system with Ds and GUS for functional genomics in rice

Abstract

To develop an efficient means of enhancer trapping, a two-element system employing Ds and an Ac transposase (AcTPase) gene was tested in rice. We generated 263 transgenic rice plants, each of which harboured the maize transposable element Ds together with a GUS coding sequence under the control of a minimal promoter ( Ds-GUS), and a gene that confers resistance to the herbicide chlorsulfuron. Among the 263 lines generated, 42 were shown to have a single copy of the Ds-GUS element. Four single-copy lines were crossed with each of six transgenic plants that carried the AcTPase gene. Excision of the Ds-GUS in leaves of F₁ plants was detected in eight combinations out of seventeen examined. The frequency of transposition of Ds-GUS in germ cells in the F₁ plants was examined using 10,524 F₂ plants, and 675 (6%) were judged to be transposants. Their frequencies differed among F₁ plants depending on the AcTPase x Ds-GUS cross considered, and also among panicles on the same F₁ plant. This suggests that Ds-GUS tends to transpose during panicle development. Southern analysis with a GUS probe showed different band patterns among transposants derived from different panicles. Therefore, the transposants derived from different panicles must have arisen independently. Transposants showing tissue-specific GUS activities were obtained, and enhancers thus trapped by the Ds-GUS element were identified. These results demonstrate that the system is suitable for the isolation of large numbers of independent Ds-GUS transposants, and for the identification of various tissue-specific enhancers. The Ds-GUS lines generated in this study offer a potentially powerful tool for studies on the functional genomics of rice.

Introduction

Rice is one of the most important suppliers of nutrients for humans, and is a model plant with several advantages for molecular genetic studies (Shimamoto 1995; Izawa and Shimamoto 1996). It has a relatively small genome (430 Mb), and there are a large number of established markers useful for mapping and subsequent positional cloning of genes of interest. In spite of these advantages, however, rice still lacks an efficient technique for cloning genes responsible for mutant phenotypes. Though the number is growing, relatively few genes have been disabled and cloned using the endogenous retrotransposon Tos17 (Hirochika 2001) and by positional cloning approaches (Song et al. 1995; Ashikari et al. 1999; Takahashi et al. 2001; Yamanouchi et al. 2002; Li et al. 2003; Sasaki et al. 2003). The Tos17 system is the most useful tagging system to date. However, the efficiency of tagging of newly emerging mutations has been estimated to be around only 10% (H. Hirochika, personal communication). In Arabidopsis, the T-DNA of Agrobacterium and heterologous transposable elements are often used as insertional mutagens, and target genes have been successfully cloned (Bhatt et al. 1996; Azpiroz-Leehan and Feldmann 1997). Therefore, it is obvious that efficient gene-tagging systems are necessary to facilitate the cloning of mutant genes, and for studies on the functional genomics of rice. In this species, about 18,000 insertional lines have so far been generated using a gene trap system based on T-DNA; about 7,000 of these have been examined, and 1.6–2.1% of tested organs showed reporter gene activity (Jeon et al. 2000).

The maize transposable element Ac / Ds is known to be active in many heterologous plant species, including rice (Enoki et al. 1999). In rice, it was first reported that autonomous Ac had transposed during shoot regeneration and leaf development (Izawa et al. 1991), and this discovery was followed by the detection of transposition of non-autonomous Ds in the presence of the Ac transposase gene (Shimamoto et al. 1993). Ds has been successfully used as an insertional mutagenic fragment, as well as a vehicle for enhancer trap systems, in Arabidopsis (Ito et al. 2002). An enhancer trap system is a modified version of a gene-tagging system. The enhancer trap vector contains the coding sequence of a reporter gene, such as a β-glucuronidase (GUS) gene, with a minimal promoter, in addition to a selectable marker gene. Insertion of such a reporter gene in or near an enhancer element of a gene may cause activation of the reporter gene by the enhancer and/or disruption of the inserted gene. The genes can be identified if disruption of the gene causes a phenotypic change, and can be cloned using the vector as a tag. Even if no phenotypic changes are observed, the gene can be identified by screening for tissue-specific expression of the reporter. Thus, this system can be utilized to identify organ-, tissue- or cell-type-specific enhancers or genes, as well as for conventional gene tagging.

In rice only limited attempts have been made to use such a system as a tool for gene tagging. There have been only a few reports which have shown transposition of autonomous Ac, or several types of modified Ds, using a small number of primary transformants (Izawa et al. 1997; Chin et al. 1999; Solis et al. 1999; Nakagawa et al. 2000). Chin et al. (1999) demonstrated efficient transposition of Ds and obtained several lines that displayed tissue-specific reporter gene activity using a gene trap vector. Because Ac/Ds tends to transpose preferentially, but not always, to sites close to its original position on the same chromosome (Smith et al. 1996; Machida et al. 1997; Nakagawa et al. 2000; Ito et al. 2002), one needs to use multiple Ds transgenic plants with independent insertions in various chromosomal positions. To apply Ac / Ds as a tool for functional analysis of the rice genome, it is also necessary to select a large number of transposants by an efficient method. Recently, a number of Ds transposants were obtained from many independent transformants, but neither the original positions of the Ds nor the patterns of transposition in plants were analysed (Upadhyaya et al. 2002).

In this study we introduced the enhancer trap vector Ds-GUS (Fedoroff and Smith 1993) into rice, and analysed the frequency and timing of transposition of the Ds-GUS element after crossing insertion lines with a line expressing the Ac transposase (AcTPase). To establish a set of independent insertion lines carrying the Ds-GUS at different positions distributed over most or all rice chromosomes, we chose 42 original transgenic lines, each of which harboured a single copy of the Ds-GUS. Mapping of the original Ds-GUS sites (T-DNA integration sites) showed that ten rice chromosomes out of twelve harboured one or more Ds-GUS insertions at various positions. Crossing of these Ds-GUS lines with the AcTPase lines induced excision and transposition of the Ds-GUS in leaves and in germ line cells at high frequency, resulting in the identification of many independent transposants in the F₂. Furthermore, we showed that simple and efficient selection methods based on herbicide and antibiotic resistance can be utilized to screen for stable transposants of the Ds-GUS. We further identified several transposants that showed tissue-specific GUS activity, and we present examples of the molecular identification of the trapped enhancers. These results demonstrate that we have succeeded in establishing an efficient experimental system capable of generating a large number of Ds-GUS transposant lines and various tissue-specific enhancers. These lines should serve as useful resources for functional genomics in rice.

Materials and methods

Enhancer trap vectors

Plasmids bearing Ds-GUS and 35S-AcTPase T-DNAs were obtained from Dr. N. V. Fedoroff (Fedoroff and Smith 1993). The 35S-AcTPase T-DNA plasmid was modified to make it suitable for rice transformation and selection. First, the tms2 gene driven by the CaMV 35S promoter (35S-tms2) was replaced with a bialaphos resistance gene driven by the same promoter. To do so, a 3′ portion of the AcTPase gene was first amplified by PCR using the primer pair Ac1 (5′-TGAAGCTTTGATATGCACAA-3′) and Ac3 (5′-CCTCTAGAGGTACCCTGCAGGTCGGTAACGGTCGGTAAA-3′). The amplified PCR fragment was digested with Xba I, and then ligated to the 35S-AcTPase T-DNA plasmid digested with the same enzyme. The resultant plasmid was digested with Kpn I, and then blunt-ended with T4 DNA polymerase, followed by digestion with Pst I. This DNA was ligated to the Eco RI (filled in with T4 DNA polymerase)-Pst I fragment of pARK8, which contains the bialaphos resistance gene driven by the CaMV 35S promoter (35S-bar). The resultant plasmid, named pBAc1, contains the 35S-bar in place of the 35S-tms2 in the original plasmid. Then the CaMV 35S promoter of pBAc1 was replaced by an ubiquitin promoter. For this purpose, pBAc1 was digested with Pst I, and then blunt-ended with T4 DNA polymerase, followed by digestion with Bam HI. This DNA was ligated to the Hind III (filled in with T4 DNA polymerase)-Bam HI fragment of pUBA, which contains the ubiquitin promoter. The resulting plasmid, named pBAc3, contains the bialaphos resistance gene driven by the ubiquitin promoter in place of the 35S-tms2 in the original plasmid (Fig. 1b).

Fig. 1a–c

Strategy used for enhancer trapping. a Structure of Ds-GUS T-DNA and positions of probes and primers. The arrows indicate positions and directions of primers. crALS, the coding sequence of the chlorsulfuron resistance gene; 19S-Hyg, the hygromycin resistance gene driven by the CaMV 19S promoter; 2×35S, two tandemly repeated copies of the CaMV 35S promoter; Km, the kanamycin resistance gene; RB, the right border of T-DNA; LB, the left border of T-DNA. Restriction enzyme sites: Bg, Bgl II; E, Eco RI; P, Pst I. b Structure of 35S-AcTPase T-DNA. 35S-AcTPase; the AcTPase gene driven by the CaMV 35S promoter; Ubi-bar, the bialaphos resistance gene driven by the ubiquitin promoter; Km, the kanamycin resistance gene; RB, the right border of T-DNA; LB, the left border of T-DNA. c Strategy used for enhancer trapping. Six of the 35S-AcTPase lines were crossed with each of the four Ds-GUS lines. F₁ plants were examined for somatic excision in leaves. F₂ seeds were subjected to screening for heritable transposition events by PCR, and for chlorsulfuron resistance to examine the frequency and timing of transposition in each F₁ plant. In the F₃ generation, transposants selected by PCR were subjected to screening for tissue-specific GUS expression to identify enhancer trapping events

Transformation of rice

Oryza sativa cv. Nipponbare was transformed by the method described by Hiei et al. (1994). When pBAc3 was used for transformation, bialaphos (5 mg/l) was used instead of hygromycin for the selection of transformants.

Southern analysis

DNAs were isolated from rice leaves with a cetyltrimethylammonium bromide (CTAB)-based method (Murray and Thompson 1980). DNAs were digested with restriction enzymes, and blotted onto a nylon membrane. Hybridization and detection were carried out using ECL direct labelling and detection systems (Amersham).

Mapping of insertions

Sequences flanking Ds-GUS T-DNA insertions were amplified by TAIL-PCR (Liu et al. 1995) or adaptor-ligated PCR, cloned, and sequenced using an ABI377 autosequencer (ABI). Two sets of primers specific for Ds-GUS T-DNA were used: the first set comprised ALS2 (5′-GTTGATCTCTTCATCATTCAATGG-3′) and its nested primers ALS3 (5′-AGCTTAACTAGTAAACTAAAGTAGTC-3′) and ALS4 (5′-AATATTACATGAATAATCCAAAACGAC-3′), while CGN3 (5′-GCTGCATTAATGAATCGGCCAAC-3′) and its nested primers CGN2 (5′-GGAGAGGCGGTTTGCGTATTG-3′) and CGN1 (5′-GAGACCTCAATTGCGAGCTTTC-3′) made up the other set. For adaptor-ligated PCR, genomic DNAs were digested with Bcl I, Nsi I, Pst I, Sal I or Xho I, and ligated to an adaptor which was generated by annealing the complementary oligonucleotides 5′-GTACATATTGTCGTTAGAACGCGTAATACGACTCACTATAGGGAN-3′ and 5′-NTCTCCCTATAGTGAGTCGTATTACGCGTTCTAACGACAATATGTAC-3′ (N indicates sequences that are cohesive to a corresponding cleaved restriction site when annealed). After ligation, PCR was carried out with the Ds-GUS specific primer ALS2 and the adaptor-specific primer CPC1 (5′-GTACATATTGTCGTTAGAACGCGTAATACGACTCA-3′), and then with the nested primers ALS4 and CPC2 (5′-CGTTAGAACGCGTAATACGACTCACTATAGGGAGA-3′). PCR products were cloned into pBluescript (Stratagene) and sequenced with the ABI377 autosequencer. A database search was carried out at the DDBJ (http://spiral.genes.nig.ac.jp/homology/welcome-e.shtml) and at the RGP (http://rgp.dna.affrc.go.jp/).

Detection of somatic excision and heritable transposition

DNAs were isolated from a leaves of F₁ plants by a simple CTAB-based method. A young leaf (~2 cm long) was homogenized with a motor-driven pestle in 100 μl of CTAB buffer (100 mM TRIS-HCl pH 8.0, 20 mM EDTA, 3% CTAB, 1.4 M NaCl, 1% polyvinylpyrrolidone K 30). The homogenate was incubated at 60°C for 30 min after the addition of another 400 μl of the CTAB buffer. After incubation, 500 μl of chloroform was added to the homogenate and gently mixed for 30 min, followed by centrifugation at 15,000 rpm for 10 min at room temperature. The supernatant was transferred to a new tube, and DNAs were precipitated with an equal volume of isopropanol. DNAs were dissolved in 20 μl of 10T-0.1E (10 mM TRIS-HCl pH 7.6, 0.1 mM EDTA) supplemented with a small amount of RNase A. PCR was carried out for 30 cycles of 94°C for 30 s, 60°C for 30 s and 72°C for 1 min, followed by a final 5-min incubation at 72°C. The reaction mixture contained 1×ExTaq buffer, 2.5 mM MgCl₂, 0.25 mM dNTPs, each primer at 0.2 μM, and 0.5 U of ExTaq (Takara Shuzo) in a volume of 20 μl. The following primers were used: 35S-F2 (5′-GTGGATTGATGTGATATCTC-3′), ALS1 (5′-CTAAAGCTTCGACGAGGATA-3′), GUS-F (5′-ACGCTCACACCGATACCATC-3′), GUS-R (5′-CAACGCTGATCAATTCCACA-3′), NOS-R2 (5′-ATCGCAAGACCGGCAACAGG-3′), H-1 (5′-AACCAGTTGTTGATCTGCTT-3′) and H24-1 (5′-GCTCTTCTGAGGAGGACCAA-3′ (Ito et al. 1999).

Selection for chlorsulfuron resistance

F₃ seeds were surface sterilized with 1% sodium hypochlorite and rinsed several times with sterilized water. The seeds were then placed on an MS plate (MS basal salts and vitamins, 3% sucrose, 0.8% agar, pH 5.8) containing various concentrations of chlorsulfuron and 50 mg/l hygromycin, and incubated at 28°C under continuous light for 2 weeks. To screen for transposants, F₂ seeds were cultured on an MS plate containing 1 mg/l chlorsulfuron and 50 mg/l hygromycin for 2 weeks. Resistant seedlings were then transferred to soil.

Enhancer trap screening

F₃ seeds were surface sterilized with 1% sodium hypochlorite and rinsed several times with sterilized water. The sterilized seeds were imbibed in water overnight at 28°C, or placed on MS plates containing 50 mg/l hygromycin, and incubated at 28°C for 2 weeks. Hygromycin-resistant plants were transferred to soil and grown on a cycle of 14 h at 28°C in the light and 10 h at 25°C in the dark.

Samples were incubated in X-gluc solution containing 1 mg/ml 5-bromo-4-chloro-3-indolyl-β-D-glucuronide and 0.5% Triton X-100 in 50 mM phosphate buffer (pH 7.0) at 37°C overnight, fixed (in 5% formaldehyde, 5% acetic acid, 20% ethanol), and cleared in ethanol.

Cloning of enhancers

TAIL-PCR and adaptor-ligated PCR were carried out as described above, except that the following Ds-GUS specific primers were used: Ds1 (5′-GGTGAAACGGTCGGGAAACTAG-3′), Ds2 (5′-CTACCGTTTCCGTTTCCGTTTAC-3′) and Ds3 (5′-GTTTTGTATATCCCGTTTCCGTTC-3′). The PCR products were directly sequenced on an ABI3100 autosequencer. A database search was carried out at the DDBJ (http://spiral.genes.nig.ac.jp/homology/welcome-e.shtml) and the RGP (http://cdna01.dna.affrc.go.jp/cDNA/ and http://rgp.dna.affrc.go.jp/). Insertion of the Ds-GUS was confirmed by PCR using the primer combination P1772-F1 (5′-GTAGGTAGTATCTGTATGCA-3′) and Ds3 for P1772, and the primer pair P2503-F1 (5′-TTAATCGCACTCGTGCAAAG-3′) and Ds3 for P2503 (see below).

Young leaves were cut into pieces 1 cm in length and floated on water for the indicated periods at 28°C in light. RNA isolation and RT-PCR were carried out as described previously (Ito et al. 2001) using the primer pairs P1772a-F2 (5′-GGAAGCCTATCGGAGTCTTC-3′) and P1772a-R2 (5′-AAAGAGCCTTGCACAACATG-3′) for AK073621, P2503-F3 (5′-GCAGTACACCAAGGACTCTG-3′) and P2503-R3 (5′-CACACCCGTCAGAAATCCTC-3′) for OsLEA3, and P2503-F4 (5′-CAAAATGTCCTCAGTAGTTG-3′) and P2503-R4 (5′-GGCAAGCTGAGACAGCATAT-3′) for AK062488 (see below).

Results

Generation of Ds-GUS lines and 35S-AcTPase lines of rice

We utilized as enhancer-trap vectors two plasmids developed by Fedoroff and Smith (1993) with some modifications (Fig. 1a, b). One is a vector containing Ds between a cauliflower mosaic virus (CaMV) 35S promoter and the coding region of a chlorsulfuron resistance gene, inserted in a T-DNA region (Fig. 1a). The Ds contains a GUS coding sequence under the control of a minimal version of the CaMV 35S promoter, and a hygromycin resistance gene. This Ds is called Ds-GUS. The other vector carries an Ac transposase (AcTPase) gene driven by the CaMV 35S promoter (35S-AcTPase) (Fig. 1b). This vector was modified from the original AcTPase vector (Fedoroff and Smith 1993) as follows. The 35S-tms2 cassette in the original vector was replaced by a gene that confers resistance to the herbicide bialaphos, which is controlled by a maize ubiquitin promoter, so that the 35S-AcTPase gene and the bialaphos resistance gene are now located in the T-DNA region. Thus, the hygromycin resistance and the bialaphos resistance can be used to check for the presence of the Ds-GUS and the AcTPase gene, respectively. Furthermore, the chlorsulfuron resistance can serve to monitor excision of the Ds-GUS from its initial position in the vector.

The Ds-GUS T-DNA was introduced into rice by Agrobacterium -mediated transformation (Hiei et al. 1994). We obtained 263 transgenic rice plants, and the copy number of the integrated T-DNA was estimated by Southern analysis (Fig. 2a) using the probes shown in Fig. 1a. Since probes R and L detect a right end and a left end, respectively, of the T-DNA including the chlorsulfuron resistance gene, Ds-GUS lines that showed Eco RI fragments of the same sizes as those in the vector must contain the complete sequence of the Ds-GUS and the flanking chlorsulfuron resistance gene. The number of Bgl II or Pst I bands indicates the copy number of the Ds-GUS T-DNA. Forty-two lines were judged to contain a single copy of the Ds-GUS and the chlorsulfuron resistance gene necessary for its transposition and selection. Fifty lines were judged to contain two copies, at least one of which retained the complete region. The other 171 lines contained more than three copies or no complete region.

Fig. 2a, b

Original Ds-GUS lines of rice. a Estimation of the copy number of Ds-GUS T-DNA. Genomic DNAs isolated from six original Ds-GUS lines were digested with each of three restriction enzymes, blotted onto nylon membrane and hybridized with the probe R. Bg, Bgl II; E, Eco RI; P, Pst I. b Mapping of Ds-GUS T-DNA insertion sites. Sequences flanking twenty Ds-GUS T-DNA insertion sites were amplified, sequenced and mapped on rice chromosomes in silico. The numbered vertical lines indicate the twelve rice chromosomes. The filled triangles indicate chromosomal positions at which the Ds-GUS T-DNA was located in each original line. The Ds-GUS lines used in this study are boxed

We generated six lines of rice that contained 35S-AcTPase T-DNA. Southern analysis suggested that the numbers of integrated T-DNA copies ranged from one to six depending on the line considered (data not shown).

Both Ds-GUS lines and 35S-AcTPase lines were self-pollinated, and seeds were collected for further studies.

Mapping of the original Ds-GUS insertions on rice chromosomes

To determine map positions of the Ds-GUS inserts in the original lines, the sequences flanking the Ds-GUS T-DNA were amplified by TAIL-PCR or adaptor-ligated PCR, and sequenced. A database search was carried out using the amplified flanking sequences as queries. If the sequence matched a rice genome sequence that had already been mapped, this simultaneously localized the original Ds-GUS insertion. With this method we were able to map the Ds-GUS sequences in twenty lines. These sequences were distributed in various locations on ten of the twelve rice chromosomes (Fig. 2b).

Somatic excision of the Ds-GUS in F₁ leaves

To examine how frequently the Ds-GUS element can transpose in rice, we chose four lines containing single copies of the Ds-GUS (E029, E060, E150 and E161) and crossed them with each of six 35S-AcTPase lines (A006, A008, A015, A016 A017 and A018) (Fig. 1c). F₁ seeds were obtained from 17 of these crosses, and the resulting plants were used for the detection of somatic excision by PCR. For this purpose, DNAs were isolated from a single leaf of each F₁ plant. Primers were designed to flank the original Ds-GUS insert (Fig. 1a); one (35S-F2) anneals in the CaMV 35S promoter and the other (ALS1) anneals in the coding sequence of the chlorsulfuron resistance gene. Because the CaMV 35S promoter is repeated in tandem, there are two annealing sites for the primer 35S-F2. If the Ds-GUS has been excised, 0.6-kb and 0.9-kb bands are amplified from an inner site and an outer site, respectively. If the Ds-GUS has remained in place the size of the amplified band is about 6 kb, or no band is amplified due to the fact that long-range PCR is inefficient. Out of the 17 combinations examined, eight showed the 0.6-kb and 0.9-kb bands, whereas none of the other nine combinations, nor any of the parental lines, showed such bands (Fig. 3a). The Ds-GUS line E029 showed no excision with any of the 35S-AcTPase lines, and the 35S-AcTPase line A015 failed to induce excision of the Ds-GUS in any of the combinations examined. Nevertheless, these results demonstrated that many combinations of the Ds-GUS and the 35S-AcTPase permit excision of the former in leaves.

Fig. 3a–d

Somatic excision and heritable transposition of the Ds-GUS. a Detection of excision of the Ds-GUS in leaves of F₁ plants by PCR. The upper panel shows somatic excision of the Ds-GUS in F₁ plants. In some F₁ plants 0.6-kb and 0.9-kb bands appear. The parental lines used for each cross are indicated above the lanes. The two lower panels show the results of PCR analysis of the parental lines. The leftmost lanes contain a size marker (lambda DNA digested with Sty I). b Screening for germline transposants by PCR. The asterisks indicate transposants that show a Ds-GUS band and a positive control band (cont) but no untransposed Ds-GUS band (UT). Lanes that show all three bands indicate that the plants have Ds-GUS at the original position, and these were classified as non-transposants. Presence of the control band only indicates a plant without the Ds-GUS. c Confirmation of transposants selected for chlorsulfuron resistance by subsequent PCR. Plants with a Ds-GUS band and a positive control band (cont) but no untransposed Ds-GUS band (UT) were confirmed to be transposants. Plants with three bands are either non-transposant or carry both a transposed Ds-GUS and an untransposed Ds-GUS. d Independent transposants derived from different panicles of the same F₁ plant. Genomic DNAs of transposants (P2352–P5968) obtained from different panicles of the same F₁ plant (A008×E060) were digested with Sac I, blotted onto nylon membrane and hybridized with a GUS probe

Heritable transposition of the Ds-GUS

To examine the frequency and timing of germ-line transposition of the Ds-GUS in the F₁ plants, F₂ seeds were collected separately from each panicle of the F₁ plants. Using leaf DNAs from each F₂ individual, transposants were selected by PCR with three combinations of primers (Fig. 1a). The first combination of primers (GUS-F and NOS-R2) amplifies an inner portion of the Ds-GUS, and tests for the presence of the Ds-GUS. The second combination (ALS1 and GUS-R) amplifies a junction region between the Ds-GUS and the coding sequence of the chlorsulfuron resistance gene, and detects the presence of untransposed Ds-GUS, but not transposed Ds-GUS. The third combination (H24-1 and H-1; Ito et al. 1999), which amplifies a part of OSH1, an endogenous rice gene (Matsuoka et al. 1993), was used as a positive control for the PCR. If Ds-GUS is detected but untransposed Ds-GUS is not, the plant harbours only transposed Ds-GUS and it is scored as a transposant. A representative sample of the results of the PCR screen for transposants is shown in Fig. 3b and summarized in Table 1. Based on this screen, we identified 678 transposants (6%) among 12,123 F₂ plants derived from all 17 cross combinations. The 678 transposants were obtained from 143 panicles. The frequency of transposition differed greatly in the different F₁ plants, and differences in transposition frequency were even found among panicles from the same F₁ plant. Examples are shown in Table 2. Thus, in an F₁ plant derived from the cross between A008 and E060, 64% (16/25) of the seeds of panicle #19 were transposants, while no transposant was obtained from nine other panicles (Table 2). In an F₁ plant derived from the A017×E150 cross, the transposition frequency ranged from 0% in 13 panicles to 56% (5/9) in panicle #24 (Table 2). These results indicate that the Ds-GUS can transpose in the germ-line cells of many F₁ plants at a high frequency.

Table 1 Frequency of heritable transposition

Table 2 Number of transposants identified in each panicle

The differences in transposition frequency among panicles suggests that the Ds-GUS tends to transpose at certain stages during panicle development (see Discussion). This further suggests that transposants obtained from different panicles are independent. To check this, Southern analysis was carried out using DNAs isolated from the transposants derived from different panicles of the same F₁ plant. Most of the F₂ plants obtained from different F₁ panicles showed a unique band with a specific size when hybridized with a GUS probe (Fig. 3d), while most of the F₂ plants from the same panicle showed same sized bands (data not shown). These results indicated that transposants obtained from different panicles arise independently.

Chlorsulfuron resistance and screening of the transposants

The Ds-GUS in the vector prevents expression of the chlorsulfuron resistance gene because it is inserted between the CaMV 35S promoter and the coding sequence of the resistance marker (Fig. 1a), and it is expected that the excision of the Ds-GUS from its original position should activate the chlorsulfuron resistance gene and confer the resistance phenotype upon transposants. To confirm this, F₃ seeds derived from self-pollinated transposants were germinated on an MS plate containing various concentrations of chlorsulfuron. As a positive control, we used transgenic rice plants which harbour the chlorsulfuron resistance gene driven by the CaMV 35S promoter (35S-ALS). As a negative control non-transformed wild-type plants were used.

Transposants derived from E060 and E150 showed strong resistance to chlorsulfuron (Fig. 4). They grew on plates containing 1 mg/l chlorsulfuron. However, transposants derived from E161 were as sensitive as the untransposed parental lines or non-transformed plants (Fig. 4). These results revealed that the transposants can show resistance to concentrations of chlorsulfuron that are more than 100 times that tolerated by the untransposed parental lines, but the level of chlorsulfuron resistance observed in the transposants greatly depends on the parental line from which they are derived.

Fig. 4

Chlorsulfuron resistance of transposants. P5669, P1863 and P3239 are transposants derived from crosses between A017 and E060, E150 and E161, respectively. The numbers at the bottom indicate the concentration of chlorsulfuron (mg/l) used

In order to screen many transposants for chlorsulfuron resistance, F₂ seeds were germinated and grown on an MS plate containing 1 mg/l chlorsulfuron and 50 mg/l hygromycin, and resistant seedlings were selected. Out of 11,026 F₂ seeds examined, 2202 (20%) showed resistance to both chlorsulfuron and hygromycin (Table 1). These resistant plants were obtained from 358 panicles. The frequency of transposants observed in the chlorsulfuron screen was higher than that found by PCR screening, because we preferentially used F₂ seeds derived from the F₁ plants which produced transposants at high frequency, as judged by PCR, such as A008×E060 and A008×E150. In this screen, the proportion of the transposants differed from plant to plant and from panicle to panicle, as in the PCR-based screen. In agreement with the above results, no transposants were obtained from E161 (Table 1). PCR analysis using primers ALS1 and 35S-F2, which amplify an empty donor site, showed that the resistant plants contained the chlorsulfuron resistance gene (data not shown), and PCR analysis with the three combinations of the primers used for PCR screening confirmed that the resistant plants were transposants in most cases (Fig. 3c). These results showed that the transposants could be effectively screened by testing for chlorsulfuron and hygromycin resistance.

GUS activities in the transposants

To test whether transposed Ds-GUS elements can trap enhancers and induce GUS activity, the transposants selected by the PCR were stained with X-gluc. We examined F₃ seeds imbibed overnight, 2-week seedlings grown on MS plates containing hygromycin, and various organs at the flowering stage. Parental lines in which the Ds-GUS had not transposed were also stained to evaluate background GUS activity in every organ examined.

The parental Ds-GUS lines showed weak GUS activity in various organs which were examined after overnight incubation. The parental Ds-GUS lines also showed high GUS activity in some organs, such as nodes at all stages, and lodicules, styles, anthers and pollen only after flowering. It would thus appear to be difficult to identify enhancers trapped in these organs at these stages. This pattern of background staining is due to the structure of the vector, and not to the position of T-DNA insertion, because all parental lines showed a similar staining pattern (data not shown).

Among 400 lines examined, we identified 44 lines that showed specific GUS activities in various organs. Examples are shown in Fig. 5. Transposant lines P2097, P4583, P2449, P1652 and P6129 showed GUS activities in developing floral organs. In P2097 and P4583, GUS staining was observed in entire anthers, while in P2449, the staining was restricted to the septa of the anthers (Fig. 5). In P1652, GUS staining was observed in the pollen, and in P6129, it was observed in the lodicules (Fig. 5). P5097 showed GUS staining in a young panicle with a length of about 2 mm (Fig. 5). P5816 showed GUS staining in the dorsal region of a developing embryo (Fig. 5). P1772, P2503, P4692 and P5664 showed GUS activity in leaves. In P1772 and P2503, the GUS staining was observed in the cut edge of a leaf (Fig. 5). Thus, wounding stress may have activated certain enhancers in these two lines. In P4652 and P5664, the GUS staining was observed in a ligular region including a ligule and auricles, but not in the leaf blade or leaf sheath (Fig. 5). P4555 showed GUS activity in a stem (Fig. 5). P4581 and P4583 showed GUS activity in roots with distinct regional specificities. In P4581, the staining was observed in an entire root (Fig. 5). In contrast, P4583 showed GUS staining specifically in the inner layers of a root (Fig. 5).

Fig. 5

Tissue-specific GUS activities detected in the transposants in the F₃ generation. P2097, P4583, P2449, P1652 and P6129 show GUS staining in different organs of young flowers. P5097 shows staining in the young inflorescence, P5816 in a restricted region in the developing embryo. In P1772 and P2503, staining is observed at the cut edges of the leaf blade. In P4692 and P5664, GUS staining is found in the ligular region of the leaf. In P4555 the stem, and in P4581 and P4583 the roots, are stained in specific patterns

Although we detected various tissue-specific enhancers, the enhancer trapping efficiency of 11% (44/400) in the transposants may have been overestimated. Because the transposants used for staining were not selected for the absence of the 35S-AcTPase, some of the GUS activities observed may have been caused by a second transposition in the F₃ plants that retained the 35S-AcTPase. In such a case, the staining pattern may be shared with neither their siblings nor progenies. Confirmation of the staining with plants that are free from 35S-AcTPase is therefore necessary.

Identification of enhancers trapped by the Ds-GUS

To identify enhancers trapped by the Ds-GUS element, we chose as examples two lines (P1772 and P2503), both of which displayed GUS activity in surfaces exposed by wounding. P1772 was derived from the parental line E060, in which the Ds-GUS was located on chromosome 1. P2503 was derived from E150, in which the Ds-GUS was located on chromosome 9. Sequences flanking the transposed Ds-GUS were amplified by TAIL-PCR or adaptor-ligated PCR, sequenced, and a database search was carried out using the flanking sequences as queries. The results showed that in P1772 the Ds-GUS was located in the BAC clone OSJNBa0018L16, which covers the region between 64.4 and 65.3 cM on chromosome 12 (http://rgp.dna.affrc.go.jp/). We searched for cDNAs located near the insertion site and identified a full-length cDNA, AK073621 (Fig. 6a). The Ds-GUS was located 1.4 kb downstream of the 3′ end of this cDNA. No cDNA has yet been identified or annotated within 10 kb on the opposite side of the insertion site. The expression pattern of this gene was examined by RT-PCR (Fig. 6b). RNAs were isolated from leaves at the indicated time after wounding. Expression was observed at 4 h and 24 h after cutting, while no signal was detected at 0 h (Fig. 6b). Because this expression pattern was consistent with the GUS activity observed in P1772 (Fig. 5), it was concluded that the Ds-GUS had trapped the enhancer of this gene in P1772.

Fig. 6a–e

Identification of enhancers trapped in P1772 and P2503. a Genome structure around the Ds-GUS insertion site in P1772. The open boxes indicate exons, the arrows mark the orientation of transcription. b RT-PCR analysis of AK073621. The upper (Southern blot) and lower (EtBr staining) panels show the expression of AK073621 and actin (internal control), respectively, at the indicated times (in h) after wounding; + and − indicate whether reverse transcriptase was added to the reaction mixture or not. c Genome structure around the Ds-GUS insertion site in P2503. The open boxes and the arrows indicate exons and the orientation of transcription, respectively. d RT-PCR analysis of AK062488. The upper and lower panels show the expression of AK062488 and actin, respectively, as indicated in b. e RT-PCR analysis of OsLEA3. The upper (Southern blot) and lower (EtBr staining) panels show the expression of OsLEA3 and actin, respectively, as indicated in b

In P2503, the Ds-GUS was located in the BAC clone OJ1362_G11, which has been mapped at 109 cM on chromosome 5 (http://rgp.dna.affrc.go.jp/). Two cDNAs were identified near the insertion site (Fig. 6c). These were the full-length cDNA AK062488 and OsLEA3 cDNA (Moons et al. 1997). The Ds-GUS was located 3.2 kb upstream of the 5′ end of AK062488 and 1.1 kb upstream of the 5′ end of the OsLEA3 cDNA (Fig. 6c). We examined the expression patterns of these genes and found that OsLEA3 was expressed transiently after wounding (Fig. 6e), while AK062488 was expressed in the absence of wounding stress, and actually showed decreased expression after wounding (Fig. 6d). Because the expression pattern of OsLEA3 was consistent with the GUS activity observed in P2503 while that of AK062488 differed from the profile of GUS activity (Fig. 5), it was concluded that the Ds-GUS element had trapped the enhancer of OsLEA3 in P2503.

Discussion

In this study we examined the utility of an enhancer trap vector based on Ac / Ds, GUS and a chlorsulfuron resistance gene, and showed that this system serves as an efficient system for enhancer trapping in rice. (1) We have shown that the Ds-GUS element can transpose in developing panicles at an acceptable frequency in the presence of a 35S-AcTPase gene (Fig. 3b, Table 1). Comparison of frequencies of transposition among panicles, and subsequent Southern analysis, revealed that the transposants obtained from different panicles are the products of independent events (Fig. 3d, Table 2). This makes it easy to obtain a large number of independent transposants by simple repetitive ratoon culturing of the F₁ plants as described by Chin et al. (1999). (2) We confirmed that excision of the Ds-GUS permits the expression of the chlorsulfuron resistance gene (Fig. 4, Table 1). This also enabled us to obtain a large number of transposants simply by screening the F₂ seeds for chlorsulfuron and hygromycin resistance. (3) Tissue-specific enhancer activities were detected in a proportion of the transposant lines by GUS staining (Fig. 5). This finding indicates that transposed Ds-GUS can reveal the activity of various enhancers in a dominant manner. (4) We identified some of the enhancers trapped by the Ds-GUS element. This clearly demonstrates the usefulness of our system for the molecular identification of enhancers. (5) We obtained stable Ds-GUS lines which carry single copies of the element in ten of the 12 rice chromosomes. Each line can be used for the efficient isolation of region-specific transposants. Efficient isolation of transposants harbouring Ds-GUS dispersed at various positions throughout the rice genome should therefore be feasible. Thus, the results obtained here show that a reliable system which allows one to generate a number of independent transposants and to further select various enhancer-trapped lines has been successfully established in rice. Successful identification of the trapped enhancers at the molecular level also demonstrated that the enhancer trapping system described here represents a marked improvement over those previously described for rice.

Temporal pattern of transposition of the Ds-GUS element

The differences in transposition frequency observed among panicles from the same plant suggests that the Ds-GUS tends to transpose at certain stages during panicle development. If the Ds-GUS transposes before differentiation of the primordial cells of the panicles, 75% of the F₂ plants should be transposants with the same insertion pattern in many panicles (25% will lose the Ds-GUS by segregation). For example, in an extreme case, if the Ds-GUS transposes in a fertilized ovule of an F₁ plant, all the panicles will show a transposition frequency of 75% and the transposant seeds set will be identical. In contrast, if the Ds-GUS transposes after differentiation of the primordial cells of the panicles, the frequency of transposition will depend on the stage at which the Ds-GUS transposes during panicle development. In this case, the proportions of the transposants in panicles may differ from panicle to panicle, as was found in our system. If the Ds-GUS transposes at an earlier step in panicle development, the proportion of transposants will be higher than when the Ds-GUS transposes later. This is indeed what we found (Table 2). The variation in the timing of transposition also indicates that the transposants obtained from different panicles contained Ds-GUS elements that had reinserted at different chromosomal positions. This is a great advantage, allowing one to obtain large numbers of independent transposants in rice, because rice produces many panicles in ratoon cultures (Chin et al. 1999).

Transposition frequency and chlorsulfuron resistance

We observed a correlation between the frequency of transposition and the level of chlorsulfuron resistance. Thus, the transposants derived from E060 and E150, which showed a higher frequency of heritable transposition (11% and 15%, respectively, by PCR screening), showed greater resistance to the herbicide, while the transposants derived from E161, which showed a lower frequency of transposition (1%), were as sensitive as the non-transformed control plants (Fig. 4). One possible explanation for this correlation is that both characteristics are determined by the chromosomal structure of the site at which the original Ds-GUS is located. For the transposition of the Ds-GUS, the AcTPase needs to access the site where the original Ds-GUS is located. Similarly, for the expression of the chlorsulfuron resistance gene, the transcription machinery needs to access the empty donor site where the Ds-GUS is located before transposition. The accessibility of the site to these proteins may determine the frequency of transposition, as well as the level of expression of the resistance gene. Such epigenetic regulation of both transcription and transposition has also been reported in Arabidopsis (Miura et al. 2001). The chromosomal structure may thus regulate both transposition and gene expression. The plants generated in this study may also be useful for studying the influence of chromatin structure on gene expression and transposition.

Strategy for distributing the Ds-GUS element over whole rice chromosomes

To permit the isolation of many Ds-GUS insertional mutants and the trapping of various tissue-specific enhancers, it is important that the Ds-GUS can transpose to sites on all twelve chromosomes of rice. It is well known that Ac / Ds tends to transpose to linked sites in both maize and heterologous plant species, such as Arabidopsis and rice (Smith et al. 1996; Machida et al. 1997; Nakagawa et al. 2000; Ito et al. 2002). It would be difficult to obtain numerous lines habouring single transposed Ds-GUS elements in various independent chromosomal positions if a limited number of original Ds-GUS lines is used. This difficulty can be overcome by using multiple original lines, in which the Ds-GUS is distributed evenly at different positions on each chromosome. These lines can produce a number of transposants with reinserted Ds-GUS in various positions on each of the twelve chromosomes. Given that the size of the rice genome is 1522 cM (Harushima et al. 1998), and a single Ds-GUS line can be used for saturated transposition of a 20-cM interval (10 cM each direction), a total of 76 original lines with Ds-GUS dispersed in various chromosomal positions would be necessary. We generated 263 transgenic rice plants harbouring Ds-GUS, and identified 42 single-copy lines and 50 two-copy lines. Single-copy lines can easily be obtained from the two-copy lines in the next generation after segregation of the two copies by self-pollination. If the Ds-GUS elements in the available 92 lines are distributed evenly in the rice genome, saturated transposition can be carried out with these lines. We have therefore begun to map the elements in the 42 single-copy lines taking advantage of the availability of the genome sequence of rice. We amplified and sequenced the flanking sequences of the Ds-GUS T-DNA, and carried out a homology search in the database. The position of the Ds-GUS can thus be mapped on chromosomes if it has inserted within a genome sequence that has already been mapped. With this method, we have mapped the insertion sites of the Ds-GUS in 20 of the original lines to ten different chromosomes out of the twelve (Fig. 2b). These lines could be used for targeted mutagenesis or enhancer trapping in a specific chromosomal region.

Enhancers trapped by Ds-GUS insertion

In this study, we identified enhancer trapped lines which show the tissue-specific GUS activities at a relatively high frequency (Fig. 5). Since these lines may have inherited the 35S-AcTPase, and the GUS activities may have resulted from a second transposition of the Ds-GUS in the F₃ plants, it is necessary to confirm the enhancer activities in their siblings or their progenies which are free from the 35S-AcTPase. Alternatively, it is necessary to clone the enhancers and compare their expression with the GUS staining patterns. Successful identification of the enhancers in two lines out of two examined suggested that the GUS activities shown in Fig. 5 may indeed be representative of each line in many cases, and are not due to a second transposition of the Ds-GUS in the given organs. Once the enhancer activity is confirmed, the temporal and spatial activities of the enhancers, as well as the environmental conditions under which the enhancers are activated, can be studied in detail. In addition, the enhancers or their genes can be cloned easily by using the Ds-GUS as a tag, as we demonstrated. Such enhancer-trapped lines and their genes can be used as tissue-specific marker lines and molecular markers, respectively, in a wide range of studies on rice. In addition to these characteristic benefits of enhancer trapping, insertional mutants can be obtained in the same way as with conventional transposon tagging. We have screened some of the Ds-GUS transposed lines, and identified a mutant line which produced a giant embryo. This clearly shows that the system can be utilized for mutant isolation and analysis. This enhancer trap system should therefore serve as a very useful tool for functional genomics in rice.