Enhancer trap is a powerful approach for identifying tissue- and stage-specific gene expression in plants and animals. For Arabidopsis research, β-glucuronidase (GUS) enhancer-trap lines have been created and successfully used to identify tissue-specific gene expression in many plant organs. However, limited applications of these lines for seed germination research have been reported. This is probably due to the impermeability of the testa to the GUS substrate. By focusing on the stages between testa and endosperm rupture, we were able to circumvent the testa barrier to the GUS substrate and observe diverse tissue-specific gene expression during germination sensu stricto. One hundred and twenty-one positive subpools of 10 lines out of 1130 were isolated. Approximately 4500 plants from these subpools were grown in a greenhouse and one to seven individual plants exhibiting GUS expression in seeds were isolated for each subpool. This library of the Arabidopsis seed enhancer-trap lines is an efficient tool for identifying seed germination-associated genes. The individual lines from this library will be provided to the international seed biology research community. International collaboration to identify the trapped genes using genome-walking PCR and to characterize the gene functions using knockout plants will significantly enhance our understanding of the molecular mechanisms of seed germination.
The Arabidopsis testa mutants have advanced our knowledge of seed biology, revealing that the chemical and physical properties of Arabidopsis testa significantly affect seed dormancy. The critical genes which determine the properties of testa are expressed in the endothelium of developing seeds (Debeaujon et al., 2001, 2003; Sagasser et al., 2002).
In mature Arabidopsis seeds, the testa is no longer a living tissue. The major sites of gene expression in imbibed Arabidopsis seeds are in the internal living tissues such as the embryo and the endosperm. Therefore, to comprehensively understand the mechanism of seed germination, it is necessary to characterize gene expression in these two tissues during imbibition. cDNA microarrays and proteomics have provided emerging data on gene expression in germinating Arabidopsis seeds (Gallardo et al., 2001, 2002; Ogawa et al., 2003). Although these are robust approaches to identify seed-expressed genes, the tissue specificity of the expression of these genes is unknown.
Enhancer- and gene-trap techniques using the GUS reporter gene have successfully been applied to identify tissue- and stage-specific gene expression in Arabidopsis leaves and inflorescences (Campisi et al., 1999; He et al., 2001) and carrot somatic embryos (Ko and Kamada, 2002). This technique has also been used to identify embryo- and endosperm-specific gene expression in developing Arabidopsis seeds (Stangeland et al., 2003). However, to our knowledge, only one successful application of gene-trap technique for seed germination studies in Arabidopsis has been reported (Dubreucq et al., 2000). The GUS substrate cannot penetrate the testa of Arabidopsis seeds to probe the sites of gene expression (i.e. the embryo and the endosperm) in germinating seeds of Arabidopsis because the testa is impermeable to the substrate. Therefore, it may be difficult to detect GUS activity and identify seed germination-associated genes in germinating seeds of the enhancer-trap lines. To investigate the mechanism of seed germination sensu stricto, it is essential to characterize gene expression in imbibed seeds before radicle protrusion, which is the first visible sign that germination is complete.
We characterized the morphological changes during Arabidopsis seed germination and identified the lag phase between testa and endosperm rupture. This lag phase has also been observed in lettuce (Cantliffe et al., 1984; Pavlista and Haber, 1970; Pavlista and Valdovinos, 1978) and tobacco seeds (Leubner-Metzger, 2002; Leubner-Metzger et al., 1996). We focused on the lag phase between testa and endosperm rupture and successfully detected GUS activity in the embryo and the endosperm of germinating Arabidopsis seeds. Here, we report the isolation of many positive enhancer-trap lines showing tissue-specific GUS expression in seeds and demonstrate the feasibility of using these lines for identification and characterization of seed-expressed genes.
Arabidopsis seed germination and the timing of GUS staining
The predominant tissue in mature Arabidopsis seed is the embryo whose large cotyledons are the major sites of seed reserves, which are mobilized during seedling establishment. Arabidopsis seed contains only a single cell layer of endosperm tissue that is poorly visible when performing an ordinary germination test (Figure 1a). However, the endosperm is distinguishable from the embryo using scanning electron microscopy (SEM) (Figure 1b). The endosperm and the embryo have distinct cell surface structures: the epidermal cells of the radicle are elongated while the endosperm cells appear more hexagonal (Figure 1c,d).
We conducted careful inspection of morphological changes in imbibed wild-type Arabidopsis seeds under a dissection microscope every 3 h and we also fixed seeds at different stages of germination for SEM to observe fine structural changes. The first apparent morphological change in germinating seed was testa rupture which generally occurred just above the embryonic axis, indicating that the expansion of the embryonic axis was a major driving force to open the testa. Testa rupture started 18 h after incubation at 22°C. There was at least a 12-h delay before endosperm rupture (or radicle emergence; Figure 2). During this period, the endosperm enclosing the embryo gradually elongated and emerged from the opened testa, and the radicle tip subsequently penetrated the endosperm layer in the micropylar region (endosperm cap; Figure 1e). Thus, the elongation of the endosperm was observed between the occurrence of testa and endosperm rupture, suggesting that endosperm weakening occurs and allows the radicle to elongate inside the endosperm before the completion of germination. We observed a similar pattern of seed germination in the Arabidopsis enhancer-trap lines (Thomas Jack lines, 1130 pools of 10 lines, CS31086, Arabidopsis Biological Resource Center at Ohio State University) and focused on this lag phase for GUS staining. By staining the seeds showing testa rupture (22-h imbibition at 22°C), we were able to detect tissue-specific GUS expression in seeds. The application of GUS substrate after testa rupture allowed penetration of GUS substrate into the endosperm and the embryo early enough to detect gene expression during germination sensu stricto.
Tissue-specific GUS expression in germinating seeds
The Arabidopsis enhancer-trap population was screened for seed-expressed genes. A total of 121 positive pools showing GUS expression in germinating and germinated seeds were identified. GUS activity expressed in seeds was generally strong and easily detected using a dissection microscope. The patterns of GUS staining in seeds varied. In some lines, the whole seed (except for testa) showed GUS activity (Figure 3a) including the embryo and the endosperm (Figure 3b,c, respectively), whereas other positive lines displayed tissue-specific GUS expression in the axis, hypocotyl and hypocotyl plus cotyledons (Figure 3d–f, respectively). One line displayed GUS expression exclusively at the tips of the cotyledons (Figure 3g). Post-germinative GUS expression was also detected (Figure 3h).
Overall, GUS expression localized in the micropylar region (endosperm plus radicle tip) was the predominant pattern in germinating Arabidopsis seeds (Figure 4). Of the 121 positive subpools, 91 subpools exhibited GUS activity exclusively in the micropylar region of the seed. Micropylar-localized GUS staining was found mostly in seeds with a large testa rupture, however, a small percentage of seeds with a slight testa rupture also showed micropylar staining. GUS staining was detected in both the endosperm cap and the radicle tip in some lines. It is possible that the GUS signal in these lines was transferred between tissues in close contact to each other. Other lines having micropylar-localized GUS activity exhibited the expression in either an embryo-specific or an endosperm-specific manner. In embryo-specific expression lines, the GUS activity was detected only in the radicle tip, which was enclosed by the endosperm layer (Figure 4a–c). The GUS staining in these seeds was less intense compared with other lines because the site of activity was still enclosed by the unstained endosperm layer (Figure 4b). However, when the embryo was dissected from the seed, an intense color was observed in the radicle tip, which confirmed the specificity of embryo expression (Figure 4c). In contrast, in other lines with endosperm cap-specific GUS expression, the GUS signal was clearly visible from the outside of testa. The specificity of GUS expression in these lines was confirmed by dissecting seeds into micropylar and lateral halves and also into embryo and endosperm parts (Figure 4d–g).
It is possible that the micropylar-localized expression is a reflection of limited infiltration of the GUS substrate during staining. The micropylar region of the seed is the first to take up the substrate because testa rupture is initiated in this part of the seed. To examine this possibility, we scratched and removed the testa at the lateral part of the seed and then stained with GUS substrate. The GUS activity was still restricted to the micropylar region in scratched seeds, while the control tetrazolium staining was detected in the entire seed except for the testa (Figure 4h). These results indicated that the micropylar-localized GUS staining was not a consequence of limited substrate uptake but that it represents tissue-specific gene expression. Diverse patterns of tissue-specific GUS expression in germinating and germinated Arabidopsis seeds are summarized in Figure 5.
To isolate individual positive plants showing GUS expression in seeds, 30 plants from each positive subpool were grown in the greenhouse and seeds were harvested from each plant. The second screening for GUS expression using these seeds allowed us to isolate one to seven GUS-positive individual plants for each subpool. The seed-GUS-expressing lines have been kept as a library.
The feasibility of identifying and characterizing the trapped genes
To evaluate the feasibility of the enhancer-trap approach for identifying seed-expressed genes, gene expression in one of the positive lines TMH1 that showed root-specific GUS expression was characterized. All TMH1 progeny exhibited GUS expression when germinated. GUS expression detected in the next generation of seeds indicated that the T-DNA lines were stable transformants. To identify the T-DNA insertion site in TMH1, genome-walking PCR was conducted using T-DNA right border-specific primers and adapter primers (see Experimental procedures). An approximately 0.7 kb DNA fragment was amplified from genome-walking PCR, sequenced, and found to contain the T-DNA right border (data not shown). The DNA sequence flanking the T-DNA right border was analyzed using the SIGnAL T-DNA Express Arabidopsis Gene Mapping Tool (http://signal.salk.edu/cgi-bin/tdnaexpress) and predicted to be in the Arabidopsis chromosome V. The putative insertion site was located in the intergenic region between At5g02750 (zinc finger protein) and At5g02760 (protein phosphatase 2C) (Figure 6a). To verify the insertion site, PCR was conducted using a T-DNA-specific primer (TSP) and a gene-specific primer (GSP) designed to part of the intergenic sequence of the putative insertion site. A DNA fragment of the predicted size (0.6 kb) was amplified from the genomic DNA of TMH1 but not from wild-type Arabidopsis genomic DNA (Figure 6b). The DNA sequence of the PCR product verified the junction between the genomic DNA and the T-DNA, confirming the predicted insertion site (data not shown).
The insertion site was detected in the intergenic region of the two genes At5g02750 and At5g02760, indicating that GUS expression detected in the TMH1 is most likely driven by the enhancer(s) of one of these two candidate genes. RT-PCR was conducted to examine the potential mRNA expression of these two genes in wild-type Arabidopsis seeds. No RT-PCR product was obtained with At5g02750-specific primers, whereas a product of the predicted size (1.1 kb) was amplified with At5g02760 primers (Figure 6c, RT). The DNA sequence of the RT-PCR product of At5g02760 matched the coding region of the genomic DNA sequence of the same gene. Both sets of primers amplified a DNA fragment of the predicted size (0.7 and 1.4 kb for At5g02750 and At5g02760, respectively) from wild-type Arabidopsis genomic DNA (Figure 6c, gDNA). These results indicate that the At5g02760 gene is expressed in Arabidopsis seeds. Thus, isolation of an enhancer-trap line carrying GUS expression in seeds, and characterization of the T-DNA insertion site and gene expression analysis allowed us to identify the seed-expressed protein phosphatase 2C gene. This demonstrates the feasibility of using the enhancer-trap lines isolated from our large-scale screening to identify tissue-specific genes in Arabidopsis seeds. The T-DNA insertion sites identified for some other lines are shown in Figure 7.
Recent publications demonstrate that Arabidopsis can also be used as a powerful tool for seed germination research (Clerkx et al., 2004; Sattler et al., 2004; Yamauchi et al., 2004). We have screened Arabidopsis enhancer-trap lines with detailed attention to seed morphology during germination and GUS staining. Morphological characterization of germinating seeds led to identification of a lag phase between testa and endosperm rupture, suggesting that the Arabidopsis endosperm cap also provides mechanical resistance to the radicle and is weakened before radicle protrusion. It is plausible that specific sets of genes are activated in the endosperm cap and the radicle tip to complete the prerequisite processes of radicle emergence (i.e. endosperm weakening and generation of growth potential of the radicle). The predominant GUS expression pattern detected in the enhancer-trap Arabidopsis seeds was micropylar-specific. One identified T-DNA insertion site was in the vicinity of the β-xylosidase gene (Figure 7), although its expression needs to be verified. Hydrolases such as xylosidase, which could be involved in cell wall modification and could cause endosperm weakening and embryo expansion, are possibly expressed during germination. In tomato seeds, xyloglucan endotransglycosylase/hydrolase (XTH) genes are expressed exclusively in the endosperm cap and the radicle of germinating seeds (Chen et al., 2002). Further characterization of the remaining lines may identify other cell wall-modifying proteins.
Expression of hydrolases and other cell wall proteins occurs in a micropylar-specific manner in multiple species including tomato (endo-β-1,3-glucanase and chitinase, Wu et al., 2001; endo-β-mannanase, Nonogaki et al., 2000; XTH, Chen et al., 2002; expansin, Chen and Bradford, 2000), tobacco (endo-β-1,3-glucanase, Leubner-Metzger et al., 1996), coffee (endo-β-mannanase, da Silva, 2002) and Arabidopsis (extensin, Dubreucq et al., 2000). It is plausible that the activation of the micropylar region of seed is a widespread phenomenon for the induction of seed germination. However, the activation mechanism is unknown. The regulatory mechanism controlling expression of the genes encoding the hydrolases and other cell wall proteins is not yet fully understood.
Some T-DNA insertion sites that we identified for the enhancer-trap lines exhibiting micropylar-specific GUS expression were found in the vicinity of transcription factor genes, with one insertion site located between two transcription factors (homeobox-leucine zipper protein and basic helix-loop-helix proteins; Figure 7). Further characterization of these lines could potentially provide useful information concerning upstream events during seed germination. Some of these transcription factors might be involved in determining tissue- and stage-specific expression of cell wall proteins in germinating seeds. GUS signals in these lines were also detected in rosette and cauline leaves, inflorescences and siliques, indicating that some of the transcription factors which are potentially important for seed germination could also play roles in other stages of plant development (P.-P.L., N.K., T.M.H., J.R.H. and H.N., unpublished data). Only three of the 12 lines that we are currently characterizing exhibited GUS expression only in seeds, indicating that genes essential for seed germination may not necessarily be seed-specific.
Using the enhancer-trap screening protocol, we have created a library of 121 individual lines carrying GUS expression in seeds. Campisi et al. (1999) conducted segregation analyses for kanamycin resistance and suggested that no more than 25% of the enhancer-trap lines are likely to contain more than one insertion. We conducted Southern blot analyses for 40 lines and found that 24 plants contained a single insertion with the rest carrying mainly two or three insertions. At least 60% of our lines are most likely single insertion lines. For the lines with more than a single insertion, backcrossing to the wild type will be necessary to identify which insertion is responsible for seed-specific expression (He et al., 2001).
As our experiments have demonstrated, it is feasible to utilize these lines for identifying seed-expressed genes. One line showed a T-DNA insertion site in the coding region of the gene, which potentially prevents gene expression (Figure 7, gamma-adaptin). This potential knockout line could be directly utilized for gene function analysis. The function of other identified genes can also be characterized by obtaining the knockout plants from available resources. However, characterization of all the lines will require a tremendous amount of work. Genome-walking PCR and gene function analyses for all the isolated lines will require much time. Detailed gene expression analyses under different conditions such as dormant or non-dormant states and different hormonal treatments will also need to be carried out for each line to examine the physiological roles of the gene products. Therefore, we will not analyze all the isolated lines of the library but will focus on the genes that we have already identified. The remaining lines will be donated to the Arabidopsis Biological Resource Center and provided to the international research community. The detailed information on this library is provided on the website NSF-funded Integrative Seed Biology at Oregon State University (http://www.science.oregonstate.edu/isb/).
Plant materials and growth conditions
For germination, Arabidopsis thaliana (wild type Columbia-0) seeds were placed in 9 cm plastic petri dishes on two layers of filter paper (no. 2; Whatman Inc., Clifton, NJ, USA) moistened with 4 ml water and incubated at 4°C for 3 days in the dark and at 22°C for 3 days under the light. The seeds were examined for testa rupture and radicle protrusion through the endosperm under a dissection microscope every 3 h.
Scanning electron microscopy
Germinating and germinated Arabidopsis seeds at different stages were fixed in 10 mm potassium phosphate buffer, pH 7.0 containing 4% (w/v) paraformaldehyde overnight and then dehydrated in an increasing ethanol series (30–100% v/v). Specimens in 100% ethanol were critical point dried with carbon dioxide in a Balzer CPD-020 dryer (Balzers Union, Ltd, Balzers, Liechtenstein) according to Anderson (1951). The dried specimens were mounted on an aluminum planchette and coated with approximately 10 nm of 60/40% Au/Pd using an Edwards S150B sputter coater (Edwards High Vacuum, Ltd, West Sussex, UK) operating at 1 × 10−2 Torr, 5 mbar argon pressure, 1.5 kV, 20 mA plasma current, for 60 sec. Examination was performed using the AmRAY 3300FE SEM (AmRay, Bedford, MA, USA) in the Electron Microscope Facility, Department of Botany and Plant Pathology, Oregon State University.
Screening of Arabidopsis seeds for GUS expression
The Arabidopsis enhancer-trap lines (Thomas Jack lines, 1130 pools of 10 lines, CS31086, ABRC; Campisi et al., 1999) were used for screening. Approximately 50 seeds from each subpool were placed on a small round filter paper (approximately 7 mm in diameter) moistened with water and placed on a metal stage for SEM specimens. The metal stage was cleaned thoroughly each time before placing another subpool of seeds to avoid contamination. One-hundred subpools of 50 seeds placed on small filter papers were incubated on two layers of larger (15 cm in diameter) filter papers placed in a plastic petri dish and incubated as described above. After 3-day pre-chilling at 4°C and 22 h incubation at 22°C, the small filter papers supporting the seeds were briefly blotted on dry filter papers to remove excessive water and then soaked in GUS substrate solution. GUS staining of seeds was performed as previously described (Weigel and Glazebrook, 2002) using 100 mm sodium phosphate buffer (pH 7.0) containing 0.1% (v/v) Triton X-100 and 2 mm X-Gluc (RPI Co., Mount Prospect, IL, USA). Staining was examined with a dissection microscope after overnight incubation at room temperature.
Genomic DNA was extracted from Arabidopsis leaves using phenol extraction according to the QUICK-PREP method described at http://www.biotech.wisc.edu/NewServicesandResearch/Arabidopsis/ and used for genome-walking experiments. Genome-walking PCR was performed using Genome Walker Kit (Clontech Laboratories, Inc, Palo Alto, CA, USA) according to the manufacturer's manual. Briefly, the genomic DNA of the enhancer-trap TMH1 line was digested with DraI. Adapter DNA provided in the kit was ligated to the DraI-digested genomic DNA fragments. The first PCR was conducted using Ex Taq DNA polymerase (Takara, Madison, WI, USA), an adapter primer (5′-GTAATACGACTCACTATAGGGC-3′) provided with the kit and a T-DNA right border-specific primer (RB1: 5′-TCTAGAGTCGACCTGCAGGCATGCAAGCTT-3′). Second round PCR was conducted using the nested adapter primer (5′-ACTATAGGGCACGCGTGGT-3′) and the nested T-DNA right border-specific primer (RB2: 5′-TCCCAACAGTTGCGCACCTGAATGGCGAAT-3′). The conditions used for the first PCR were: one cycle at 94°C (4 min), 80°C (2 min); seven cycles at 94°C (25 sec), 72°C (3 min); 32 cycles at 94°C (25 sec), 67°C (3 min) followed by one cycle at 67°C (7 min). The conditions used for the second PCR were: one cycle at 94°C (4 min), 80°C (2 min); five cycles at 94°C (25 sec), 72°C (3 min); 20 cycles at 94°C (25 sec), 67°C (3 min) followed by one cycle at 67°C (7 min).
PCR to verify the putative T-DNA insertion site
To verify the predicted T-DNA insertion site, a gene-specific primer (GSP: 5′-ACGTTCCAAGGCCACATGTG-3′) was designed for part of the intergenic sequence that was located upstream of the putative T-DNA insertion site and used in PCR with RB2 primer and genomic DNA from TMH1. The following conditions were used for PCR: the initial denaturation at 94°C (4 min), touchdown cycles [94°C (15 sec), 67→ 61°C (15 sec), and 72°C (30 sec)] (one cycle for each temperature) and 25 cycles at 94°C (15 sec), 60°C (15 sec) and 72°C (30 sec) followed by extension at 72°C (7 min).
RNA extraction and RT-PCR
Total RNA was extracted from 36-h-imbibed Arabidopsis seeds using standard phenol-SDS extraction (Sambrook et al., 1989). Two micrograms of DNase-treated total RNA was used for reverse transcription (RT) with a RETROscript Kit (Ambion, Austin, TX, USA). The RT product was subjected to PCR using the primers for At5g02750 (5′-ATGGAAGACGAAAACACCAC-3′ and 5′-TTACGCCACGTGTAACTCG -3′) and At5g02760 (5′-ATGGTTAAACCCTGTTGGAG-3′ and 5′-TCATGATGTTGAATGCATCG-3′). The following conditions were used for PCR: the initial denaturation at 94°C (4 min), touchdown cycles [94°C (15 sec), 62→ 56°C (15 sec), and 72°C (30 sec)] (one cycle for each temperature) and 25 cycles at 94°C (15 sec), 55°C (15 sec) and 72°C (30 sec) followed by extension at 72°C (7 min).
We are grateful to Dr Thomas Jack, Dartmouth College (Hanover, NH) and the Arabidopsis Biological Resource Center, Ohio State University for providing the Arabidopsis enhancer-trap population for our research. We thank Dr Kevin Ahern and Mr Soeldner, Oregon State University, for continuous efforts to maintain our undergraduate research program and for training undergraduate students in SEM, respectively. We also thank Dr John Harada, University of California, Davis, for his helpful discussions. This work was supported by the Howard Hughes Medical Institute Summer Undergraduate Research grant to T.M.H. and J.R.H. and the National Science Foundation grant IBN-0237562 to H.N.