Arabidopsis thaliana RD26 cDNA, isolated from dehydrated plants, encodes a NAC protein. Expression of the RD26 gene was induced not only by drought but also by abscisic acid (ABA) and high salinity. The RD26 protein is localized in the nucleus and its C terminal has transcriptional activity. Transgenic plants overexpressing RD26 were highly sensitive to ABA, while RD26-repressed plants were insensitive. The results of microarray analysis showed that ABA- and stress-inducible genes are upregulated in the RD26-overexpressed plants and repressed in the RD26-repressed plants. Furthermore, RD26 activated a promoter of its target gene in Arabidopsis protoplasts. These results indicate that RD26 functions as a transcriptional activator in ABA-inducible gene expression under abiotic stress in plants.
In this study, we show that a drought-inducible gene, RD26, encodes a NAC transcription factor. To investigate the in vivo functions of RD26, we generated transgenic Arabidopsis plants constitutively overexpressing or repressing RD26. Here, we show that RD26-overexpressing lines are hypersensitive to ABA and that many ABA- and abiotic stress-responsive genes were upregulated. On the contrary, RD26-repressed plants were less sensitive to ABA and the expression of ABA-responsive genes was repressed in the plants. Here, we suggest that RD26 participates in a novel ABA-dependent stress-signaling pathway.
RD26 encodes a NAC domain protein
The RD26 cDNA encodes a protein of 297 amino acids and is a member of the plant-specific NAC gene family. The N-terminal NAC domain contains predicted nuclear localization signals (NLS) between amino acid 76–88 and 113–179 (Figure 1a). A database search showed that there are 109 NAC genes in the Arabidopsis genome (Riechmann et al., 2000). The evolutionary relationship among 67 NAC proteins that showed the closest homology to RD26 was analyzed using the amino acid sequence of the NAC domain (Figure 1b). The phylogenetic analysis revealed that RD26 belongs to a group that includes ATAF1 and ATAF2. Northern blot and microarray analysis using the Agilent 22K Oligo DNA Microarray (K. Maruyama, JIRCAS, Tsukuba, Ibaraki, Japan unpublished data) showed that seven NAC genes that belong to the ATAF subfamily are responsive to ABA, dehydration and NaCl treatments. As none of the other NAC genes except for five genes, shown in Figure 1(b), were induced by these treatments (data not shown), only members of the ATAF subfamily may function in the ABA-dependent stress-signaling pathway. Although the C-terminal sequences of NAC proteins are divergent, two genes closest to RD26 show relatively high homology (∼30%) even in the C-terminal region (data not shown). The ANAC gene that has been reported to interact with a RING-H2 domain protein (Greve et al., 2003) is one of the closest homologues of RD26.
Expression of RD26
The RD26 cDNA was isolated by screening the cDNA library that had been prepared from Arabidopsis rosette plants dehydrated for 10 h, and the accumulation of RD26 mRNA was shown to be induced by dehydration (Yamaguchi-Shinozaki et al., 1992). We further analyzed the effects of various environmental stresses on the expression of the RD26 genes by RNA gel blot analysis. As shown in Figure 2(a), the RD26 gene was induced within 30 min after dehydration, NaCl and ABA treatments, and the mRNA level was continuously increased up to 24 h. RD26 was also responsive to methyl jasmonate, H2O2 and Rose Bengal, which are known to generate reactive oxygen species (ROS) (Green and Fluhr, 1995; Narusaka et al., 2003b). In the ABA-deficient aba2 mutant, the accumulation of RD26 mRNA during dehydration was markedly decreased suggesting that the dehydration-responsive expression of RD26 is regulated mainly by ABA (Figure 2b). On the contrary, in the NaCl-treated aba2 mutant, significant induction of RD26 was still observed (Figure 2b), implying the existence of an ABA-independent pathway of NaCl signaling for RD26 expression.
The more detailed spatial expression pattern of RD26 was determined by histochemical β-glucuronidase (GUS) staining of transgenic plants that harbored a RD26 promoter::GUS reporter construct. Figure 2(c) shows the expression of the RD26 promoter::GUS fusion gene in transgenic Arabidopsis plants raised under normal and water-deficit conditions. Weak expression of the RD26 promoter::GUS fusion gene was observed in several regions of rosette plants raised under normal growth conditions (Figure 2c, upper). By contrast, the dehydration of transgenic Arabidopsis plants strongly induced GUS expression in all organs and tissues (Figure 2c, lower), indicating that the RD26 promoter functions in all vegetative tissues of Arabidopsis plants during dehydration.
Figure 2(d) summarizes the sequences of various cis-acting elements observed in the promoter region of RD26. We found four ABREs, one dehydration responsive element (DRE), two recognition sites for MYB and one recognition site for MYC (reviewed by Busk and Pages, 1998; Finkelstein et al., 2002; Shinozaki et al., 2003). The RD26 promoter also includes a W-box, a recognition site of WRKY transcription factors. Members of the WRKY transcription-factor family appear to be involved in the regulation of various plant-specific physiological processes such as pathogen defense, senescence and trichome development (Eulgem et al., 2000). Additionally, the RD26 promoter has two AS-1 motifs known as oxidative stress-responsive elements (Lam et al., 1989). Therefore, in the ROS-responsive promoters, these sequences may function as cis-acting elements (Garreton et al., 2002).
Nuclear localization of GFP–RD26 fusion protein
The detection of putative NLS implies that RD26 may be localized to the nucleus. To examine the localization of the RD26 protein, we transiently expressed GFP–RD26 fusion protein in the epidermal cells of onion and observed the tissue by confocal microscopy. GFP–RD26 was localized in the nucleus (Figure 3a, i–iii), while GFP–ΔRD26 that lacks the NAC domain of RD26 was localized in the cytoplasm and nucleus (3a, iv–vi). These data showed that the NAC domain of RD26 is necessary for nuclear localization of RD26.
The C-terminal domain of RD26 has transactivation activity
Several NAC family genes have been suggested to function as transcriptional factors (Aida et al., 1999; Duval et al., 2002; Xie et al., 1999, 2000). We examined the transactivation activity of RD26 using a yeast system. GAL4 DNA binding domain–RD26 fusion proteins were expressed in yeast and assayed for their ability to activate transcription from the GAL4 upstream activation sequence and promote yeast growth in the absence of histidine (Figure 3b). The full-length and C-terminal part of RD26 showed obvious activation capacity (Figure 3b, rows 2 and 4), while the N-terminal part that included the NAC domain sequence did not (Figure 3b, rows 1 and 3). The region of amino acid 164–272 still promoted growth as vigorously as that observed with the intact C-terminus (Figure 3b, row 5). However, further deletion of this region completely abolished transactivation activity (Figure 3b, row 6), suggesting that the region of amino acid 217–272 is a potential transactivation domain.
Growth phenotypes of RD26-overexpressed transgenic lines
To analyze the biological functions of RD26, we generated transgenic Arabidopsis plants in which RD26 was overexpressed (35S::RD26). The coding region of RD26 was fused to the 35S promoter of the cauliflower mosaic virus and the omega sequence of the tobacco mosaic virus, which elevates the translation level of the transgene. We examined the expression levels of the transgene in the 35S::RD26 plants (RD26-overexpressed plants hereafter) by RNA gel blot analysis using an RD26-specific probe (Figure 4a).
The growth of the transgenic plants was compared with that of the wild-type plants at 3 weeks after sowing (Figure 4b). The RD26-overexpressed plants growing on germination medium (GM) agar plates had slightly chlorotic leaves with a larger leaf blade and shorter petiole than the wild-type plants (Figure 4b). When grown on soil, the RD26-overexpressed plants expressing a higher level of the transgene were much smaller than the wild-type plants, and had markedly reduced floral apical dominance, shorter bolts (Figure 4c), fewer flowers, and reduced seed yield (data not shown). The S3 line that expressed a lower level of transgene showed an intermediate phenotype (Figure 4b,c), suggesting that these morphological phenotypes were caused by overexpression of RD26.
RD26-overexpressed plants were hypersensitive to ABA
To evaluate the effect of RD26 overexpression on the ABA sensitivity, we germinated the RD26-overexpressed plants on or transferred them to GM media containing various concentrations of ABA. During the germination process, no obvious difference was observed between the RD26-overexpressed plants and vector control plants (data not shown). These seedlings were transferred to GM agar media containing various concentrations of ABA. Upon transfer to a medium supplemented with ABA, the seedlings of the RD26-overexpressed plants became bleached and their root growth was inhibited to a greater extent than that of the vector control plants (Figure 5a). On the medium containing 1.0 μm ABA, root growth of vector control plants was 82% of that in the medium without ABA, whereas that of the RD26-overexpressed plants was 54% of that in the medium without ABA (Figure 5b). On the medium containing 3.0 μm ABA, the root growth of the RD26-overexpressed plants was reduced to 45%. The root growth of the vector control plants in the medium containing ABA was 72% of that in the medium without ABA. These results demonstrate the hypersensitivity of the RD26-overexpressed plants to ABA.
Although we also tested the sensitivity of the plants to osmotic stresses such as high-salinity and drought, no significant tolerance was observed in our experimental conditions (data not shown).
RD26 protein transactivates the promoter of target gene in vivo
Using a yeast system we found that the RD26 protein may act as a transcription activator. To test this hypothesis, we preliminarily checked upregulated genes in the RD26-overexpressed plants using the ∼7000 Arabidopsis full-length cDNA microarray (Seki et al., 2002), and found that the most strongly upregulated gene in the RD26-overexpressed plants encoded a glyoxalase I family protein (GLY) (data not shown). We also carried out RNA gel blot analysis to confirm this result. The GLY gene was upregulated in both the ABA-untreated and treated RD26-overexpressed plants (Figure 6a, lanes 1–6). Dehydration, NaCl and ABA treatments also induced expression of this gene (Figure 6b), and the induction was slower than that of RD26 (Figures 2a and 6b).
To determine whether the RD26 protein is capable of transactivating a promoter of the GLY gene, we performed transactivation experiments using protoplasts prepared from Arabidopsis T87 cultured cells. Protoplasts were cotransfected with a GUS reporter construct that had a deletion in the promoter region of the GLY gene and an effector plasmid containing RD26 cDNA fused to the 35S promoter and Ω sequence (Figure 6c,d). Expression of the GUS reporter gene containing the −410 or −215 bp region of the GLY promoter was moderately activated by adding ABA (Figure 6e). Co-expression of the RD26 protein strongly transactivated the −410 or −215 bp GLY promoter, and the expression level of the GUS gene was markedly elevated by ABA treatment (Figure 6e). However, the −86 bp promoter region was not activated by ABA nor was it activated by co-expression of the RD26 protein (Figure 6e). These results suggest that the RD26 protein functions as a transcription activator involved in the ABA-responsive expression of the GLY gene.
Construction of RD26-repressed plants – loss-of-function phenotype of RD26
Although we obtained two RD26 T-DNA insertion mutant lines, no obvious change in the phenotype was observed in our experimental conditions (data not shown). As reviewed previously (Zhang, 2003), the negative results may be attributed to the potential functional redundancy among NAC genes that belong to the ATAF subfamily. Actually, as shown in Figure 1(b), there were at least six NAC genes responsive to abiotic stress and ABA treatment other than RD26, suggesting that they work redundantly. Therefore, to minimize the effects of phenotypic masking caused by potential functional redundancy, we used a recently established dominant repression technique exploiting a repression domain, SRDX, derived from the EAR-motif of superman, a TFIIIA-type zinc finger repressor (Hiratsu et al., 2003). By using several well-studied transcription factors such as EIN3, PAP1 and AtMYB23, Hiratsu et al. (2003) demonstrated that translational fusion between these genes and the repression domain suppressed the expression of their target genes and caused loss-of-function phenotypes. Furthermore, ectopic expression of the CUC1 chimeric repressor, which had the repression domain, resulted in cup-shaped or fused cotyledons, the phenotype of cuc1/cuc2 double mutant (Aida et al., 1997); this demonstrated that the repression domain can convert a transcriptional activator into a strong suppressor which produces loss-of-function phenotypes even in the presence of redundant transcription factors (Hiratsu et al., 2003).
To generate RD26-repressed plants, we fused the RD26 coding region to the 12 amino acid-repression domain of SRDX (LDLDLELRLGFA) in frame and expressed the chimeric RD26 repressor under the control of the 35S promoter (Figure 7a). None of the 42 T1 35S::RD26SRDX seedlings that we analyzed had the cup-shaped or fused cotyledons observed in the CUC1-repressed plants (35S::CUC1SRDX; Hiratsu et al., 2003), suggesting that the overexpression of RD26SRDX confers an RD26-dependent phenotype. The transgenic plants expressing the RD26SRDX chimeric suppressor had longer petiole and smaller leaf blades than vector control plants (Figure 7b). After the transfer to soil, the 35S::RD26SRDX plants (RD26-repressed plants hereafter) grew normally (data not shown). The morphological changes in the leaves of the RD26-repressed plants were the inverse of those in the RD26-overexpressed plants that had larger leaf blades and shorter petioles (Figure 4). Furthermore, the expression of the GLY gene, the target of RD26, was repressed in the RD26-repressed plants (Figure 6a, lanes 7–10). These observations indicate that RD26 was efficiently repressed by the EAR motif and produced the loss-of-function phenotype.
RD26-repressed plants were insensitive to ABA
As RD26 has been shown to function as a transcriptional activator in ABA signaling, we analyzed the sensitivity of the RD26-repressed plants to ABA. As observed in the experiments of the RD26-overexpressed plants, there was no obvious difference in sensitivity between the RD26-repressed plants and vector control plants during the germination process (data not shown). Then seedlings of the 35S::RD26SRDX (RD26-repressed), 35S::RD26 (RD26-overexpressed), and vector control plants were transferred to GM agar plates containing 10 μm of ABA. One week after the transfer, the RD26-repressed plants grew normally on the ABA-containing plates, but the RD26-overexpressed plants and vector control plants became bleached (data not shown). After 3 weeks, the RD26-repressed plants were still green and continued to grow on the ABA plates, while the vector control plants showed suppressed growth or wilted (Figure 8). In contrast to the RD26-repressed plants, the RD26-overexpressed plants became bleached earlier than the vector control plants (data not shown), and were completely bleached 3 weeks after transfer (Figure 8). These results indicate clearly that RD26 acts positively in ABA signaling.
Identification of target genes of RD26 using the Agilent 22K Oligo DNA Microarray
To identify the target genes of the RD26 protein, we used the Agilent Arabidopsis 2 Oligo Microarray (Agilent Technologies, Inc., Palo Alto, CA, USA) which covers over 21 000 Arabidopsis genes. RNAs prepared from 2-week-old seedlings of the RD26-overexpressed, RD26-repressed and vector control plants with or without ABA treatment (100 μm ABA for 5 h) were used for microarray analysis. In the microarray analysis, RD26 cDNA was highly overexpressed in 35S::RD26 plants (ratio to non-treated control plants: 11.8) and 35S::RD26SRDX plants (ratio to non-treated control plants: 13.7; ratio to ABA-treated control plants: 4.3) (Table 1). Genes whose expression levels were significantly higher in the RD26-overexpressed plants are summarized in Table 1. Twenty genes were upregulated in the RD26-overexpressed plants and 15 of them were downregulated in ABA-treated RD26-repressed plants, suggesting that RD26 regulates the expression of these genes. In addition, most of the upregulated genes were responsive to ABA and abiotic stresses (Table 1). The overexpression of RD20 in the RD26-overexpressed plants is notable. RD20 encodes a Ca2+-binding protein that has been shown to respond to ABA and dehydration, and is supposed to function in ABA-mediated stress-signal transduction (Takahashi et al., 2000). The GLY gene, demonstrated as the direct target of RD26 in Figure 6, is also listed. Glyoxalase is one of the glutathione-dependent detoxification enzymes (Dixon et al., 1998). The glutathione transferase and aldo/keto reductase family gene upregulated in the RD26-overexpressed plants has also been shown to play a role in detoxification of ROS (Oberschall et al., 2000). Interestingly, many defense- and senescence-related genes were found among the genes upregulated in the RD26-overexpressed plants, for example, SAG13, a marker gene of senescence (Lohman et al., 1994). The ELI3-2 gene encoding a cinnamil-alcohol dehydrogenase is known not only as a defense-related gene (Kiedrowski et al., 1993), but also as a senescence-associated gene, SAG25 (Quirino et al., 1999). Most of the remaining genes are related to the metabolism of cellular components. These genes may be involved in morphological changes in transgenic plants.
Table 1. Genes upregulated or downregulated in the RD26-overexpressed or RD26-repressed plants identified by microarray analysis
aExperiment 1: upregulated in non-treated 35S::RD26 plants (non-35S::RD26/non-vector).
bExperiment 2: downregulated in ABA-treated 35S::RD26SRDX plants (ABA-vector/ABA-35S:: RD26SRDX).
cExperiment 3: downregulated in non-treated 35S::RD26SRDX plants (non-vector/non-35S::RD26SRDX).
dStress response analyzed by microarray analysis (Seki et al., 2001; K. Maruyama, unpublished data). D, drought; N, high salinity; A, ABA; C, cold.
eValues represent the mean of two replicates for two experiments (i.e. individual RNA preparations from two different lines).
We characterized RD26, a new member of the NAC-domain gene family. It is one of the RD (responsive to dehydration) genes we isolated by differential screening (Taji et al., 1999; Yamaguchi-Shinozaki et al., 1992). We found that RD26 protein functions as a transcriptional activator not only in yeast (Figure 3) but also in Arabidopsis protoplasts (Figure 6). Deletion analysis showed that the region of amino acid 214–272 is necessary for transcriptional activation, although this region has no characteristic structure. Furthermore, we confirmed that the RD26 protein contains a functional NLS in the NAC domain, suggesting that the RD26 functions in the nucleus (Figure 2a). These data are consistent with the previous observations that the other NAC genes function as transcriptional activators (Aida et al., 1999; Duval et al., 2002; Xie et al., 1999, 2000).
The expression of RD26 was induced strongly by dehydration, high salinity and ABA treatments. The dehydration-responsive expression of RD26 was found to be regulated mainly by ABA, but ABA is not needed in the response of RD26 to NaCl. The 1.5-kb promoter region of RD26 contains four ABREs, one MYC recognition site, two MYB recognition sites, and one DRE. ABRE and MYC and MYB recognition sites are involved in ABA-responsive gene expression (reviewed by Busk and Pages, 1998; Finkelstein et al., 2002; Shinozaki et al., 2003). Therefore, these cis-acting elements regulate the ABA-dependent expression of RD26. RD26 was also responsive to jasmonic acid and ROS-related reagents. We found two W-boxes and two AS-1 motifs in the RD26 promoter. These cis-elements may act as defense- and oxidative stress-responsive cis-acting elements, respectively, suggesting that they may regulate the biotic stress responses of RD26.
To investigate the in vivo functions of RD26, we generated transgenic Arabidopsis plants overexpressing intact RD26 or RD26SRDX chimeric repressor. Considering the potential functional redundancy of NAC genes, an overexpression approach would be better than a loss-of-function approach such as knockout, antisense, and RNA interference. Especially, the dominant repression technique using the EAR-motif repression domain is a unique and powerful tool for the functional analysis of redundant transcription factors (Hiratsu et al., 2003). The overexpression and repression of RD26 resulted in an opposite phenotype (Figure 4 and 7). The seedlings of the RD26-overexpressed plants had larger leaf blades and shorter petioles while the RD26-repressed plants had smaller leaf blades and longer petioles. Typically, salt-stressed plants become smaller and have shorter petioles like the RD26-overexpressed plants (Burssens et al., 2000), suggesting that stress responsive signaling is constitutively activated in the RD26-overexpressed plants.
Direct analysis of the responses of transgenic plants to exogenous ABA revealed that the overexpression of RD26 increases the sensitivity of both seedlings and roots to ABA whereas the growth of RD26-repressed plants was not inhibited by a high concentration of ABA (Figures 5 and 8). These findings suggest that RD26 functions as a transcriptional activator in ABA signal transduction in vegetative tissue.
To investigate the target genes of RD26, we performed microarray analysis using the RD26-overexpressed plants. As might be expected, many ABA- and drought-responsive genes were upregulated in the RD26-overexpressed plants, and most of them were repressed in the ABA-treated RD26-repressed plants. For example, the RD20 gene suggested to function in ABA-mediated stress-signal transduction (Takahashi et al., 2000) was overexpressed in the RD26-overexpressed plants. Genes involved in the antioxidant defense system, such as the GLY gene, a glutathione transferase gene, and an aldo/keto reductase family gene were also upregulated in the RD26-overexpressed plants. Overexpression of defense- and senescence-responsive genes in the RD26-overexpressed plants is also notable. In both senescence and pathogen infection, the generation of several ROS plays a role in signal transduction or physiological processes (Overmyer et al., 2003). Recently, increasing evidence indicates that water stress-induced ABA accumulation triggers the generation of ROS, which, in turn, leads to the upregulation of the antioxidant defense system (Jiang and Zhang, 2003). Although, it is not yet clear whether such a stress-signal transduction system exists, the expression of RD26 was also induced by methyl jasmonate, H2O2 and Rose Bengal. Thus, RD26 is considered to play a key role in the crosstalk among defense, senescence, and ABA-mediated stress-signaling pathways.
We further demonstrated that RD26 transactivates its target gene in Arabidopsis protoplasts. The promoter of the GLY gene, upregulated in the RD26-overexpressed plants, was used for the transactivation experiment (Figure 6). The −410 and −215 bp GLY promoter fragments were activated moderately by ABA treatment and strongly by co-expression of the RD26 protein. However, the −86 bp GLY promoter region did not respond to the ABA treatment or the RD26 co-expression. The −410 bp GLY promoter contains two ABREs which interact with bZIP proteins such as rice TRAB1, Arabidopsis ABI5 and AREB/ABF proteins. However, deletion analysis of the GLY promoter showed that elimination of one of the ABREs did not affect the activation of the GLY promoter by ABA treatment or co-expression of RD26. A single copy of ABRE is not sufficient for ABA-responsive transcription, and at least one coupling element in the promoter regions is necessary for expression of ABA-inducible genes (Hobo et al., 1999; Narusaka et al., 2003a). These data suggest that the GLY gene may be expressed independently of the ABREs. Recently, RD26 and the closest homologues, At3g15500 and At1g52890, were found to recognize the ERD1 promoter (Tran et al., 2004) and the ‘CACG’ to be the core-binding motifs of these three proteins using a yeast system (Tran et al., 2004). Although the GLY promoter contains two CACG motifs, deletion analysis that a −86 bp GLY promoter fragment that included the two CACG motifs could not be activated by the RD26 protein. This result suggests that the −86 bp GLY promoter contains an additional cis-acting element recognized by RD26 together with the CACG core sequence, or that RD26 recognizes a sequence in the GLY promoter different from that identified in the ERD1 promoter.
In conclusion, the RD26 gene encoding a NAC-related transcription factor is responsive not only to dehydration, but also to NaCl, ABA and jasmonic acid treatments. The transgenic plants overexpressing RD26 cDNA were hypersensitive to ABA, and inversely, the transgenic plants with RD26 repressed were insensitive to ABA. The expressions of many ABA- and stress-induced genes including RD20 and GLY genes were upregulated in plants overexpressing RD26 and repressed in plants with RD26 repressed. In Arabidopsis protoplasts, RD26 activated a promoter of the GLY gene that was upregulated in plants overexpressing RD26. Thus, these results provide strong evidence for the involvement of RD26 in stress-responsive ABA signaling and the presence of a novel ABA-dependent signaling pathway.
Plants (Arabidopsis thaliana ecotype Columbia) were grown on 9-cm plastic pots filled with a 1:1 mixture of perlite/vermiculite (Kakiuchi No. 2 Compost; Kakiuchi Material Co., Ltd, Tokyo, Japan) and watered with 0.1% Hyponex (Hyponex, Osaka, Japan). Transgenic plants were grown on GM agar plates (Valvekens et al., 1988) containing 30 mg l−1 kanamycin (GMK). Two-week-old plate-grown plants were transferred to soil pots. The plants were then grown under long-day conditions (16-h-light/8-h-dark cycle) at 22°C.
Constructs and transformation of Arabidopsis
The 35S::RD26 plasmid was constructed by amplifying the entire coding region of RD26 by PCR with BamHI linker primers and cloned into a binary vector pBE2113Not (Liu et al., 1998) in the sense orientation. The RD26 promoter::GUS plasmid was constructed by amplifying a 2-kb DNA fragment upstream of the RD26 coding region by PCR with an upstream SalI linker primer and a downstream BamHI linker primer and cloned into the XbaI/BamHI site of pBI101.1. Arabidopsis plants were transfected with Agrobacterium tumefaciens strain GV3101 by the vacuum infiltration method (Bechtold et al., 1993).
RNA gel blot analysis
Three-week-old plants were harvested from GM agar plates, and then dehydrated on Whatman 3MM paper at room temperature with approximately 60% humidity under dim light. Plants subjected to treatment with hormones and salt stress were grown hydroponically in a solution containing 250 mm NaCl, 100 μm ABA, 50 μm IAA, 50 μm GA3, 50 μm 1-aminocyclopropane-1-carboxylic acid (ACC), 50 μm methyl jasmonate, 20 mm H2O2 and 5% glucose or water, under dim light. Cold treatment was applied under dim light by exposure of plants grown at 22°C to a temperature of 4°C. Plants subjected to the stress treatments for various periods were frozen in liquid nitrogen. Total RNA was isolated according to the method described by Nagy et al. (1988) and subjected to RNA gel blot analysis as described previously (Taji et al., 2002).
Transient expression of the sGFP protein in onion epidermal cells
The RD26 protein was fused to the C terminal of GFP by constructing a plasmid pGFP3BX which includes an sGFP coding region without a stop codon, and new cloning sites downstream of the sGFP coding region. The PCR fragment of sGFP (Chiu et al., 1996) amplified with an upstream BamHI linker primer 5′-CCCGGATCCATGGTGAGCAAGGGCGAG-3′ and a downstream noSTOP-BglII/XhoI linker primer 5′-CTGCAGCCCGGGCGGCCGCTCGAGATCTGTACAGCTC-3′ was digested with a BamHI and NotI, and replaced with a BamHI–NotI fragment of p35S-sGFP (Nakashima et al., 1997). The entire region or C terminal region (encoding amino acids 163–298, 487–894) of the RD26 coding sequence was amplified by PCR with BamHI linker primers and cloned into the BglII site of pGFP3BX in sense orientation to produce GFP–RD26 fusion proteins. These DNA constructs were introduced into onion epidermal cells using a pneumatic particle gun (PDS-1000/He; Bio-Rad Laboratory, Tokyo, Japan) as described previously (Ito and Shinozaki, 2002). After incubation at 22°C for 8–12 h, the tissues were stained with propidium iodide (10 μg ml−1) and the GFP fluorescence was observed in whole mounts under a confocal laser-scanning microscope (Zeiss LSM510; Carl Zeiss, Jena, Germany) as described previously (Ito and Shinozaki, 2002).
Transactivation analysis in yeast
RD26 was examined for the presence of an activation domain using a yeast assay system. The complete region or partial region of the RD26 coding sequence was cloned into the DNA-binding domain vector pGBKT7 (Clontech, Palo Alto, CA, USA), so that a GAL4 DNA-binding domain-RD26 fusion protein would be produced when the yeast cells were transformed. We used the yeast strain AH109 harboring the LacZ and HIS3 reporter genes. The transformed yeast culture was dropped onto SD plates with or without histidine. The plates were incubated for 3 days and applied to a β-gal assay.
Transient expression assay using Arabidopsis protoplasts
Arabidopsis protoplasts were transfected by a method described previously (Ueda et al., 2001), with minor modifications. Arabidopsis ecotype Columbia T87 suspension-cultured cells (Axelos et al., 1992) were incubated in an enzyme solution [0.4 m mannitol, 1% (w/v) cellulase ‘Onozuka’ R-10 (Yakult, Tokyo, Japan), 0.25% (w/v) macerozyme R-10 (Yakult), 8 mm CaCl2, and 5 mm MES, pH 5.6] for 2 h at 25–28°C under gentle agitation and then passed through three layers of nylon mesh (108 μm pore). Protoplasts were collected by centrifugation at 600 g for 5 min, washed twice with solution A (0.4 m mannitol, 70 mm CaCl2 and 5 mm MES, pH 5.6) and resuspended in MaMg solution (0.4 m mannitol, 15 mm MgCl2 and 5 mm MES, pH 5.6). After the addition of plasmid DNA to 150 μl of protoplast solution, 65 μl of DNA uptake solution containing 40% (w/v) polyethylene glycol 3350, 0.4 m mannitol and 100 mm Ca(NO3)2 was added. The mixture was placed at room temperature for 20–60 min and then diluted with 5 ml of 0.4 m mannitol, 125 mm CaCl2, 5 mm KCl, 5 mm glucose and 1.5 mm MES, pH 5.6. Protoplasts were collected by centrifugation at 600 g for 5 min, resuspended in Murashige and Skoog (MS) cell culture medium (Murashige and Skoog, 1962) supplemented with 0.4 m mannitol, and incubated at 22°C for 16–20 h in the dark. GUS and LUC activities were assayed as described previously (Urao et al., 1996), with a multilabel counter (Wallac AVROsx 1420; Perkin Elmer, Boston, MA, USA).
Two-week-old seedlings of the 35S::RD26 (RD26-overexpressed), 35S::RD26SRDX (RD26-repressed), and vector control plants grown on the GMK plate were harvested directly or after ABA treatment for 5 h as described in Figure 2, and were subjected to microarray experiments using the Agilent Arabidopsis 2 Oligo Microarray (Agilent Technologies, Inc.). Total RNA isolated using Trizol reagent (Invitrogen, Carlsbad, CA, USA) was used for the preparation of Cy5-labeled and Cy3-labeled cDNA probes. All microarray experiments including data analysis were carried out as described previously (Hughes et al., 2001). The reproducibility of microarray analysis was assessed by dye swap in each experiment.
We thank Ms Setsuko Kawamura, Ms Hiroko Kobayashi and Ms Saho Mizukado for their excellent technical assistance. We are grateful to Ms Kyoko Yoshiwara for excellent maintenance of Arabidopsis T87 cell culture and support of microarray analysis. We also thank the Rice Genome Resource Center at NIAS for the use of the Arabidopsis microarray analysis system. This work was supported in part by BRAIN, CREST and Genome Project of RIKEN.