A high throughput system for transposon tagging and promoter trapping in tomato


*For correspondence (fax +972 8934 4181; e-mail avi.levy@weizmann.ac.il).


We describe new tools for functional analysis of the tomato genome based on insertional mutagenesis with the maize Ac/Ds transposable elements in the background of the miniature cultivar Micro-Tom. 2932 F3 families, in which Ds elements transposed and were stabilized, were screened for phenotypic mutations. Out of 10 families that had a clear mutant phenotype, only one mutant was Ds-tagged. In addition, we developed promoter trapping using the firefly luciferase reporter gene and enhancer trapping, using β-glucuronidase (GUS). We show that luciferase can be used as a non-invasive reporter to identify, isolate and regenerate somatic sectors, to study the time course of mutant expression, and to identify inducible genes. Out of 108 families screened for luciferase activity 55% showed expression in the flower, 11% in the fruit and 4% in seedlings, suggesting a high rate of Ds insertion into genes. Preferential insertion into genes was supported by the analysis of Ds flanking sequences: 28 out of 50 sequenced Ds insertion sites were similar to known genes or to ESTs. In summary, the 2932 lines described here contain 2–3 Ds inserts per line, representing a collection of approximately 7500 Ds insertions. This collection has potential for use in high-throughput functional analysis of genes and promoter isolation in tomato.


Efficient tools for forward and reverse genetics are invaluable to determine gene function. Such tools are available in maize and petunia, using insertional mutagenesis with native tranposons ( Koes et al. 1995 ; Mena et al. 1996 ; Walbot 1992). In Arabidopsis, forward and reverse genetics have been implemented with the T-DNA of Agrobacterium tumefaciens ( Feldman et al. 1989 ; Gaymard et al. 1998 ; Krysan et al. 1996 ) and with transposons from heterologous species such as the maize Ac/Ds ( Fedoroff & Smith 1993; Sundaresan et al. 1995 ) and En/Spm transposons ( Aarts et al. 1993 ; Wisman et al. 1998 ). Recently, databases of transposon insertion sites have been produced in Arabidopsis that enable rapid identification of knockouts ( Parinov et al. 1999 ; Speulman et al. 1999 ; Tissier et al. 1999 ). In tomato, insertional mutagenesis has been performed mostly with the Ac/Ds elements. These elements were shown to be active ( Yoder et al. 1988 ), and patterns of Ac/Ds transposition in this species were described ( Carroll et al. 1995 ; Osborne et al. 1991 ; Rommens et al. 1992 ; Yoder 1990). A number of tomato genes have been isolated with Ac/Ds, such as the Cf-9 ( Jones et al. 1994 ) and Cf-4 loci ( Takken et al. 1998 ), controlling resistance to various races of Cladosporium fulvum;Dwarf, a gene encoding a cytochrome P450 homologue ( Bishop et al. 1996 ), DCL, which controls chloroplast development ( Keddie et al. 1996 ) and the feebly gene which is involved in metabolism and development ( Van der Biezen et al. 1996 ). In all cases genes were tagged by targeted tagging, by taking advantage of the preferential transposition of Ac/Ds to sites close by ( Carroll et al. 1995 ; Healy et al. 1993 ) and of the linkage of the target to the previously mapped Ds elements ( Knapp et al. 1994 ; Thomas et al. 1994 ). Despite these successes in transposon tagging in tomato, there is still a need for more efficient forward and reverse genetics tools, especially in light of the present release of tomato ESTs into sequence databases (www.tigr.org/tdb/lgi/index/html). It has been proposed that the miniature cultivar Micro-Tom is well suited for large-scale mutagenesis in tomato owing to its small size, rapid life cycle, easy transformability, and efficient activity of the Ac/Ds elements ( Meissner et al. 1997 ).

Another application of insertional mutagenesis is to combine a reporter gene within the non-autonomous mobile element (T-DNA or transposon) as a tool for discovering genes and/or transcriptional regulators, such as enhancers and promoters. Enhancers can be detected by cloning a reporter gene in between the borders of a mobile element and downstream of a weak constitutive promoter ( Fedoroff & Smith 1993; Wilson et al. 1989 ), the reporter being activated upon insertion near an enhancer. Promoters can be detected by cloning a promoterless reporter in between the borders of the mobile element ( Sundaresan et al. 1995 ), the reporter being activated upon insertion downstream of a promoter and in the correct orientation, thus generating transcriptional or translational fusion. Both enhancer and gene-trapping methods enable the presence of genes and their patterns of transcriptional regulation to be detected, independently of a mutant phenotype. The β-glucuronidase (GUS) reporter gene has been used in most works on enhancer and gene trapping in plants ( Maes et al. 1999 ; Sundaresan 1996). One problem with GUS staining ( Jefferson et al. 1987 ) is the destructive nature of the staining and destaining procedure. Non-invasive and non-destructive reporter genes such as the luciferase or GFP genes have not yet been widely used for gene trapping in plants.

We report on the production and analysis of a collection of 2932 families of miniature tomatoes containing stabilized insertions of Ds elements. This collection was screened for mutant phenotypes and for enhancer and gene trapping with the GUS and luciferase reporter genes. The high frequency of luciferase-trapped genes and the sequencing of transposon-flanking regions both indicate that the Ds elements preferentially insert into genes. We discuss the utilization of this system for high-throughput insertional mutagenesis and for non-invasive gene trapping in tomato.


Establishing the Ds insertion collection

The constructs used for preparing the Ds insertion collection in Micro-Tom are shown in Fig. 1. Among those, Ds378–GUS, an enhancer trap, and Bam35S–Ac, a stable transposase source, were previously transformed and shown to be active in Micro-Tom ( Meissner et al. 1997 ). The Ds251–LUC construct ( Fig. 1) was built for gene trapping with the luciferase reporter gene and was also transformed into Micro-Tom. These constructs contain the NPTII gene as a transformation marker and/or as a re-insertion marker. The indole acetamide hydrolase (iaaH) gene confers sensitivity to NAM and was used as a negative selection marker to select against Bam35s–Ac and thus obtain stable transposition events. The ALS gene confers resistance to 100 p.p.b. chlorosulfuron in plants carrying an unexcised Ds element and confers resistance to 3 p.p.m. chlorosulfuron in plants where the Ds element was excised. F1 seeds of transposase × Ds plants were produced by crossing the transposase plants (Bam35S–Ac) with 12 independent Ds378–GUS-transformed T1 plants and 15 independent Ds251–LUC-transformed T1 plants. The number of Ds inserts in each one of the T1 Ds parents used in the crosses was determined by Southern blotting and was found to vary from one to seven in the different plants (data not shown). F1 seeds were obtained from all the crosses involving Ds251–LUC. However, for a reason which is unclear, only one of the Ds378–GUS plants gave rise to fertile F1 seeds, while embryos were aborted in the other crosses. Somatic activity of the Ds element could be detected through increased resistance to chlorosulfuron in F1 plants compared to the Ds parent, and thus selection for somatic activity could be carried out in F1 seedlings by germinating and selecting the 100 p.p.b. chlorosulfuron-resistant F1 seedling. Another indication of transposition activity is that somatic GUS or luciferase sectors could be detected in F1 plants but not in the parents (data not shown). A total of 1768 F1 chlorosulfuron (100 p.p.b.)-resistant plants from the cross with Ds-GUS, and 971 F1 plants from the crosses with Ds-LUC, were grown and seeds were harvested from each F1 plant individually. F2 seedlings were selected for germinally stable transposition: excision (ALSr), re-insertion (Hygror or Kanar), and stabilization (Namr). Following this selection, 1451 out of 19 005 F2 seedlings were obtained from the cross with Ds-GUS (7.6%), and 1481 out of 20 619 F2 seedlings were obtained from the crosses with Ds-Luc (7.2%). The majority of these plants correspond to independent transposition events as determined by the fact that in most cases they originated from different F1 plants. For F1 plants which gave rise to more than one stably transposed seedling, Southern blot analysis indicated that in about half of the cases siblings corresponded to independent transposition events. Considering that half of the excised Ds elements do not re-insert ( Meissner et al. 1997 ) and that transposase was counter-selected (1/4 of the F2 progenies survived NAM selection), this means that excision rates were very high, in the 60% range (7.5 × 2 × 4).

Figure 1.

Schematic representation of constructs transformed into Micro-Tom.

Constructs Bam35S–Ac and Ds378–GUS were used for enhancer trapping ( Fedoroff & Smith 1993). Sequences similar to Ac are shown in grey, with the terminal inverted repeats shown as grey arrows. Constructs are flanked by the right (RB) and left (LB) borders of their respective T-DNA. The β-glucuronidase gene (GUS) is fused to Ac weak promoter in Ds378–GUS. The luciferase gene in Ds251–LUC is fused to the Ac promoterless fragment from 1 to 251. Resistance to basta (Barr), kanamycin (Kanar) or hygromycin (Hygror) is conferred by the phosphinothricin, neomycin phosphotransferase or aminocyclitol phosphotransferase genes, respectively. Chlorosulfuron resistance (Chlorosulfuronr) is obtained upon excision of the Ds element from the Ds378–GUS, Ds251–LUC constructs and activation of a mutated acetolactate synthase gene from Arabidopsis ( Fedoroff & Smith 1993). Sensitivity to naphtalene acetamide (NAMs) is conferred by the indole acetic hydrolase gene.

Recessive mutations may be observed in F2 plants only if an early transposition event occurred in a given F1 plant and was transmitted to both male and female gametes. Although we have detected such case (one of the chlorophyll mutants described below), we have focused on screening for mutants in F3 families. F3 seeds from the 2932 F2 plants that were selected for stable transposition were collected for each plant separately, and F3 seeds sown for each family and screened for phenotypic mutations. Approximately 13 plants were grown for each of the 2932 families: In total the plants grown were: 18 863 F3 plants derived from 1451 different F2 plants of cross Ds378–GUS × Bam35S–Ac, and 19 253 F3 plants derived from 1481 different F2 plants for crosses of Ds251–LUC × Bam35S–Ac.

Mutant screening

The F3 families described above were screened for obvious phenotypes as seen only by visual observation. A total of 10 mutants were found, five with the Ds-GUS cross and five with the crosses with Ds-LUC ( Fig. 2). Six were chlorophyll mutants (e.g. Figure 2c,d); two mimicked the symptoms of TYLCV (tomato yellow leaf curl virus)-infected plants (e.g. Figure 2b); one mimicked hypersensitive response (not shown); and one showed narrow cotyledons and was almost leafless at the mature plant stage (e.g. Figure 2e–g). The flower of the leafless mutant had four fused ovary styles and stigma (data not shown). Eight out of 10 mutants showed 3 : 1 segregation, indicating that the mutation was caused by a single recessive allele. Southern blot analysis was done to test whether the mutants were caused by Ds insertion: DNA was extracted from 13 individuals of each of the mutant families, digested, blotted and probed with a Ds probe. To distinguish homozygote- from heterozygote-dominant plants, seeds from each one of the 13 plants of the mutant family were collected separately and sown. Co-segregation of the Ds element with the mutant phenotype was found only for the chlorophyll mutant shown in ( Fig. 2h,i) (data not shown).

Figure 2.

Mutant phenotypes obtained by insertional mutagenesis with Ac/Ds.

(a) Wild type; (b) mimicry of tomato yellow leaf curl virus symptoms; (c) chlorophyll mutant with white leaves throughout the plant; (d) upper new leaves are yellow and turn green when mature; (e) mutant with abnormal growth showing thin, elongated cotyledons in young seedlings compared to (f) wild-type seedlings. Later during development the plant grown from (e) is almost leafless with thin, simple, wiry leaves (g) and flowers with fused ovaries (not shown). (h–i) Chlorotic throughout development with stronger phenotype during fruit set; (h) chlorophyll mutant showing white leaves in the absence of transposase and (i) sibling showing variegated leaves when transposase is present.

Non-invasive gene trapping with the luciferase reporter gene

One hundred and eight F3 families derived from the Ds-luciferase parents and corresponding to independent and stable transposition events were screened for luciferase expression in various plant organs by imaging with an ultra-low light-cooled charge coupled device (CCD) camera. The same plants were screened for luminescence at the seedling, flower and young fruit stages. Different expression patterns were obtained, for example in hypocotyls, anthers, sepals or fruits ( Fig. 3). The frequency of occurrence of light signals was 4, 11 and 55% in the seedlings, fruits and flowers, respectively ( Table 1). Organ-specific signals, those only in flowers, only in fruits or only in seedlings, were found at a frequency of 48, 2 or 1%, respectively ( Table 1).

Figure 3.

Gene-trapping expression patterns as detected with the luciferase reporter gene.

Each row represents the imaging of a given organ from F3 plants of the crosses Ds251–LUC × Bam35S–Ac where selection for stabilized Ds transposition was made, and after treatment with luciferin. Left column, imaging of the organ under light conditions; right, the same object as seen after imaging in total darkness. Imaging of seedlings grown vertically in agar plates (a) led to the identification of one individual with a luminescent hypocotyl (b). Imaging of a flower (c) showing strong luciferase activity in the anthers, a weak activity in the petals and no activity in sepals (d). Imaging of a young fruit (e) where luciferase activity is restricted to the sepals (f). Imaging of the two halves of a young fruit (g) where luciferase activity is stronger in the core of the fruit rather than in other parts (h).

Table 1. . Organ-specific patterns of luminescence in 108 F3Ds251–LUC families
OrganOrgan-specific patterns of luminescence       Total (% luminescent)
  • a

    Seedlings were grown in normal conditions or tested after cold or heat shock as described in the text. Two seedlings were shut off after cold treatment and one was shut off after the heat treatment.

  • + indicates that luminescence was observed; – indicates a lack of luminescence.

Seedlings a++++4
Total luminescent522172014360
% luminescent4821620140 

Unlike GUS, which is usually destructive, luciferase is a non-destructive and probably non-invasive reporter gene. This opens a number of possibilities, such as regenerating a somatic sector, following gene expression in real time, and analysing inducible genes. Examples of such applications are given below. Imaging of F1 plants derived from crosses between Ds-luciferase and transposase, enabled to identify luminescent sectors, which correspond to insertions into genes. One such cotyledon sector, which emitted a particularly strong light signal, was cut under sterile conditions and was regenerated into a mature plant. All the organs of the regenerated plant were screened and found to express luciferase ( Fig. 4). The promoter that was trapped in this plant seems to be constitutive, and was stronger than the 35S promoter when the expression of the 35S-luciferase-positive control plants and that of the regenerated plant were compared side by side (data not shown). In one F3 seedling luciferase activity was detected in the root. This type of activity was very specific, localized in the elongating zone of the root just above the root tip ( Fig. 5a,b). The luminescent root was manually bent into a horizontal position, and the petri dish with the seedlings was grown further, vertically, for 24 h and then imaged again. Gravitropism caused the root to bend downward and gene activity could again be localized to the bending and elongating region ( Fig. 5c,d), showing that gene expression can be followed over time.

Figure 4.

Regeneration of luminescent sectors.

The luciferase assay being non-destructive, it is possible to isolate luminescent sectors from within a chimeric tissue and to regenerate whole plants from these sectors. This was done with a luminescent sector from the cotyledons of an F1 seedling of the cross Ds251–LUC × Bam35S–Ac (a,b). Staining of several organs was done after regeneration and activity was discovered in roots (c,d); flowers (e,f); young fruits (g,h) and mature fruit (i,j).

Figure 5.

Trapping genes and following expression in real time.

(a) Tomato seedlings were grown vertically and screened for luminescence. (b) After imaging in complete darkness, one seedling was identified as luciferase positive in the root, in the elongation zone, just above the root cap. Arrows indicate the position of the signal. In order to test whether expression was limited to the elongation zone, the positive root was lifted, laid horizontally and imaged again after 24 h. Cell elongation could be observed in the kink of the root formed as a result of gravitropism (c). The luminescent signal was again restricted to the newly formed elongation zone (d).

In order to identify if an inducible promoter was trapped among the 108 F3 families tested, we imaged the seedlings under normal conditions, after 30 h at 4°C and 12 h recovery, and after 20 h at 42°C with 12 h recovery. Four positive (light-emitting) seedlings could be seen when grown under normal conditions (16 h day, 25°C), two were shut off after the cold treatment and one more was shut off after the heat treatment. No cold or heat upregulated promoter was detected in this collection.

Enhancer trapping

Fruits from 70 different F2 plants, derived from cross Ds378–GUS × Bam35S–Ac and selected for stable transposition events, were screened for GUS activity, that is, for insertions near enhancers. Four fruits showed positive GUS staining with various expression patterns: in the vascular tissue ( Fig. 6a); throughout the fruit ( Fig. 6b) or in the outer layers of the pericarp ( Fig. 6c,d). Progeny of these plants were further tested to determine whether GUS activity is restricted to fruits or can also be detected in other tissues such as roots, leaves and stems. In two lines ( Fig. 6a,b), GUS activity was restricted to fruits and in the other two ( Fig. 6c,d) it was found to be also in the roots, stems and leaves (data not shown). Note that negative controls were not coloured under the staining conditions used (data not shown). Interestingly, a gradient of blue colour was observed in mutants ( Fig. 6c,d); this gradient was observed again when mutants were tested in the next generation.

Figure 6.

X-Gluc staining of fruits from F2 plants derived from the cross Ds378–GUS × Bam35S–Ac. A variety of staining patterns are found; in one line staining was localized only to the vascular cells of fruits from the breaker stage (a); in other lines staining was observed throughout the pericarp (b) or only in the outer layer of the pericarp (c,d).

Sequencing Ds insertion sites

The high frequency of luciferase-trapped genes ( Table 1) suggests either a high copy number of Ds insertions and/or preferential insertion into genes. Copy number was determined by Southern blot analysis for 30 out of the 108 F3 Ds-LUC families that were used in the trapping experiments. On average, the copy number of new Ds insertions was 2–3 (ranging from 1–4) in F3 plants derived from F2 plants where Ds had been stabilized (data not shown). Taken together, the 108 F3 families constitute a collection of approximately 250 Ds insertions. Preferential insertion into genes was tested by sequencing Ds flanking regions in this collection. Sequence analysis of the 50 DNA fragments derived from inverse PCR is summarized in Table 2. Out of 50 Ds flanking sequences, 16 are similar to known genes, 12 are similar to ESTs with no known function, six are in the T-DNA region and 16 are sequences with no significant homology to sequences from the GenBank database. Only three sequences (#2, #5, #7) were almost identical (within range of sequencing error) to known genes from tomato, indicating that sequencing insertion sites enabled the discovery of new tomato genes.

Table 2. . Analysis of Ds flanking sequences
 No.Nucleotide similarityEaProtein similarityEb
  1. a Expect value: statistical significance threshold for reporting matches against database sequences obtained after B lastn program search.

  2. b Expect value obtained after B lastx program search.

Genes1Medicago trunculata Pi transporter, MTAF0003552e-10A. thaliana Pi transporter, U623311e-64
2Solanum lycopersicum feebly gene, U356446e-29S. lycopersicum feebly prot, S706486e-06
3Tomato ovary EST 2658192e-16A. thaliana hypothetical. protein, AC0023401e-04
4Arabidopsis thaliana aap1 gene, X956226e-07  
5Lycopersicon esculentum 9.04 kb rDNA, X52215 0  
6  Spirodela polyrrhiza PDR5-like ABC, Z70524 1e-20
7L. esculentum auxin-induced proteinase2e-25  
inhibitor, L25128   
8  D. radiodurans hypothetical. protein 
9  L. esculentum polyprotein AF1190406e-11
10  A. thaliana putative Na+/H+ exchanging 
 protein, AL035394e-10    
11Potato patatin pseudogene SB6B5e-05  
12Solanum tuberosum mitoch trnC, X93575. 4e-09  
13  Anthoceros punctatus photosystem I P700 apoprotein A1, AB013664 9e-5
14Tomato EST 2438644.6e-12A. thaliana putative protein kinase, AC0042609e-10
 15L. esculentum ripening-related mRNA,0.004  
  Oryza sativa putative copia polyprotein,
EST17Tomato EST2464910.0049  
18Tomato EST 2658192.6e-17  
19Tomato EST 2612126.3e-05  
20Tomato EST 2482009.9e-05  
21A. thaliana EST P246759e-05  
22Tomato EST 2612121.2e-11  
23Tomato EST TC34150.00057  
24Tomato EST 2464914.9e-03  
25Tomato EST 2662451.2e-05  
26Tomato EST TC55320.004  
27Tomato EST 2616161.9e-05  
28Tomato EST2499461.2e-05  
In T-DNA 6 sequences   
Unknown 16 sequences   


Forward and reverse genetics with the Micro-Tom Ac/Ds system

We have described a new system for high-throughput insertional mutagenesis in tomato. Previous efforts to develop insertional mutagenesis in tomato have mostly focused on targeted transposon tagging ( Bishop et al. 1996 ; Jones et al. 1994 ). The system described here is well suited for non-targeted tagging as large populations can be screened; however the rate of tagged mutants is quite low (one in 10 in this work). The reason for the low tagging rate might be the strong activity of Ds elements, with excision rates in the 60% range.

The 2932 F3 stable insertion lines (38 000 plants) described here contain approximately 7500 Ds insertions (two to three copies per line on average) and might therefore be useful for reverse genetics. Assuming that about half the transposition events land in genes, as discussed below, we roughly estimate that the 2932 F3 lines contain around 3000 mutated genes. While only 10 families with a clear mutant phenotype were found, careful analysis, including measurement of quantitative traits, suggests that many subtle alterations in plant growth can be detected in this population (unpublished data) and therefore that it might be useful for forward genetics.

Sequencing of insertion sites also suggests that Ac/Ds insertional mutagenesis can be useful for reverse genetics in tomato because of a preferential insertion into genes: considering that the tomato genome is 10 times bigger than that of Arabidopsis, and assuming that the number of genes is in the same range as in Arabidopsis, the average gene density in tomato is much lower than that in Arabidopsis. Nevertheless the majority of the insertion sites (28 out of 50) corresponded to coding regions (known genes or ESTs), indicating a bias for insertion into coding regions. Out of the 28 (16 + 12) sequences that had a significant similarity to known genes or ESTs, only three are virtually identical to known tomato genes (the feebly gene, the rDNA, and the proteinase inhibitor). Other sequences were similar, but not identical, to other genes or ESTs from tomato, Arabidopsis, potato, or other plant species. This indicates that sequencing Ds insertion sites is an effective tool for the discovery of new genes. Among interesting hits are sequences similar to a Pi transporter, an Na+/H+ antiporter, and an ABC transporter. Interestingly, feebly had been also isolated by Ac/Ds tagging ( Van der Biezen et al. 1996 ), suggesting that it may be a hot-spot for Ac/Ds insertion. In our case the Ds insertion in feebly is in a different location. The proportion of insertions that were in the T-DNA and therefore are not useful for reverse genetics was six out of 50. This frequency (12%) is similar to that reported for Arabidopsis ( Parinov et al. 1999 ). Assuming that the tomato genome contains 20 000–30 000 genes, and that about half of the insertions are into genes, we roughly estimate that 200 000–300 000 Ds insertions, derived from several unlinked T-DNAs, would be sufficient to achieve a high (approximately 90%) chance of insertion into any specific target gene.

Probing the genome by gene trapping

Enhancer and promoter trapping have not been described previously in tomato. This feature may be particularly useful for the isolation of transcriptional regulatory regions, particularly in fruit. Root or leaf promoters can probably be borrowed from Arabidopsis trapping systems and retain their specificity; however, fleshy berry fruit types have a different structure, metabolism and development compared to siliques and probably have some unique sets of regulatory elements. Non-invasive gene trapping has not been used extensively in plants. The firefly luciferase gene used in this work proved to be a sensitive and non-invasive reporter, with many advantages compared to GUS: there is no endogenous background in the plant, and it is convenient to utilize – 20 min after the application of luciferin on the plant tissue it is possible to screen for luminescence. Furthermore, non-invasive trapping opens up many prospects such as isolation of inducible promoters; follow-up of gene expression in mutants in real time; and regeneration of somatic trapping events. This latter feature can be useful when dealing with species which cannot be propagated through seeds, or when searching for rare patterns of gene expression (each leaf represents a ‘field’ of sectors). The major limitations of gene trapping with luciferase are the requirements of a dedicated imaging system, a dark room and the high cost of the substrate.

A surprising finding was the overall high rate of luciferase-positive plants: 65 out of the 108 families screened (60%) ( Table 1). To explain this, we assume that the promoterless system used here was efficient and also that the Ds-luciferase preferentially inserts into genes. Promoterless gene trapping was used previously with T-DNA constructs, with the reporter cloned near the T-DNA border ( Maes et al. 1999 ). With Ac/Ds the presence of ATGs triplets at both termini of the element has caused concern that the reporter gene would not be translated in-frame, and thus that promoterless strategies may not work. For this reason a gene-trapping system was developed with an intron and a splice acceptor site upstream of a GUS reporter ( Sundaresan et al. 1995 ). The high frequency of luminescent traps in our experiment indicates that the ATGs near the transposon 5′ terminal region (nucleotides 1–251) did not prevent the efficient recovery of traps. This may be caused by a favourable nucleotide context that enables preferential initiation of translation from the ATG of luciferase ORF, or by other unknown mechanisms. The high rate of luminescent plants is consistent with the sequencing data showing that Ds preferentially inserts into genes ( Table 2). We do not understand the reason for the high frequency of luminescence signals in flowers ( Table 1); it may reflect the possibility that more genes are expressed in flowers than in other organs, or maybe gene expression is stronger in the flower than in other organs where weak signals might have been undetected, or flowers might contain more ATP than leaves fruits and roots (ATP is required for light emission and catalysis of luciferin) ( Aflalo 1991).

In summary, the utilization of a miniature tomato, together with the Ac/Ds trap elements that insert preferentially into genes, make the Micro-Tom–Ac/Ds system ideally suited to produce a large collection of insertions for reverse genetics in tomato and for promoter isolation. Sequencing of insertion sites is an effective way to identify new tomato genes. Subsequent analysis of homozygous mutants for these genes should greatly facilitate their functional analysis.

Experimental procedures

Constructs transformation

Constructs Ds378–GUS and Bam35S–Ac ( Fedoroff & Smith 1993) were kindly provided by Nina Fedoroff and Ds251–LUC was built as described below. These constructs were transformed in tomato as previously described ( Meissner et al. 1997 ).

Construction of plasmid Ds251–LUC

Ds251–LUC was built for promoter trapping with luciferase. For this purpose the 10 kb Asp718 fragment from plasmid 378 Ds–GUS ( Fedoroff & Smith 1993) was cloned into the same site of pUC19, giving rise to plasmid pAA1–18. The internal MscI–PacI fragment within the Ds element of 378 Ds–GUS was digested from pAA1–18, removing the Ac promoter the GUS and HPT genes. This fragment was replaced by the promoterless firefly luciferase gene linked to the NPTII gene giving rise to plasmid p22–8. The luciferase gene was obtained from the SalI–BglII fragment of construct pJD301, kindly provided by Virginia Walbot and described by Luehrsen et al. (1992) . The NPTII gene was derived from the 2.3 kb SacII–ClaI fragment from plasmid pGA492 ( An 1986). In the new Ds in p22–8, the 5′ of the promoterless luciferase gene was ligated downstream of nucleotide 251 from the 5′ end of Ac. The Asp718 fragment of p22–8, containing the ALS excision cassette ( Fedoroff & Smith 1993) and the new Ds–LUC promoter trap, was cloned into a polylinker that was inserted in the SLJ525 binary vector instead of the NPTII-containing HindIII–ClaI fragment (details of the intermediate steps of the cloning are available upon request). The resulting Ds251–LUC plasmid is shown in Fig. 1.

GUS staining

GUS activity in F2 plants was determined according to the histochemical procedure described by Jefferson et al. (1987) with some modifications. Fruits of the F2 plants were stained with X-Gluc (5-bromo-4-chloro-3-indolyl-β- d-glucuronic acid; Duchefa) at a concentration of 0.5 mg ml−1, supplemented with in 100 m m NaH2PO4 pH 8.0 (phosphate buffer), 15 m m EDTA, 5 m m ferrocyanide, 5 m m ferricyanide, and 20% methanol. Seedlings were stained with 100 m m NaH2PO4 pH 7.0 (phosphate buffer), 0.5 m m ferrocyanide, 0.5 m m ferricyanide, 0.1% Triton, and 10 m m EDTA. The clearing procedure was done as described ( Beeckman & Engler 1994).

Luciferase screening

The luciferase screening was done by imaging plant tissues in the dark chamber using a camera for ultra-low light imaging (Princeton Instruments Inc., USA). Image acquisition and processing was performed with IP lab software (Signal Analysis Corp., USA).

The imaged tissues were supplemented with 1 m m beetle luciferin (Promega) and 0.01% Triton, and were kept for 20 min in the dark after luciferin application to overcome the natural fluorescence emitted by the chlorophyll. The bioluminescence was imaged during the first 3 h of incubation with the substrate.

Isolation of transposon-flanking regions by long-range inverse PCR

Transposon-flanking regions were isolated by long-range inverse PCR adapted from Mathur et al. (1998) . Genomic DNA was extracted from pools of 5–13 Ds-insertion lines. Twenty DNA pools were prepared and 4 μg of each pool was digested overnight in 100 μl final volume with 20–50 U of restriction endonucleases SpeI, XbaI or Asp718. Following phenol chloroform extraction and ethanol precipitation, DNA was resuspended in 100 μl, and 30 μl were run on a 0.8% agarose gel to check their proper digestion. The other 70 μl were self-circularized in 500 μl final volume with 6 μl of T4 DNA ligase (Biolabs) overnight at 16°C. After ethanol precipitation, half the resuspended DNA was subjected to PCR amplification using two sets of nested primers specific of the Ds 5′ and 3′ termini, as previously described ( Gorbunova & Levy 1997). PCR reactions were performed in 50 μl using the Expand Long Template PCR system (Boehringer), as recommended by the supplier, with buffer 3. The second nested amplification was performed using 2 μl of a 500-fold diluted first PCR mixture. The isolated PCR fragments were either used directly as templates for sequencing with the internal Ds 5′ and Ds 3′ primers ( Gorbunova & Levy 1997), or cloned with the PGEM-T system (Promega). Sequences were submitted to databases at the National Center for Biotechnology Information ( www.ncbi.nlm.nih.gov) and the Tigr Institute for Genomic research ( www.tigr.org) for homology searching.


We are grateful to Yaron Sitrit for help and discussions, Efrat Rubin, X. Zhang, H. Rozen, A. Shipov and N. Avivi for excellent technical assistance, and G. Kazenelson and D. Aviv for help with the tissue culture. This work was supported by the National Plant Genome Center program of the Israeli Ministry of Science (grant 5922-2-96 to A.A.L.) and a Training and Mobility of Researchers Marie Curie postdoctoral fellowship from the European Union to V.C.