Arabidopsis Cys2/His2 Zinc-Finger Proteins AZF1 and AZF2 Negatively Regulate Abscisic Acid-Repressive and Auxin-Inducible Genes under Abiotic Stress Conditions

In plants, abiotic stresses induce various physiological changes and growth inhibition that result in adaptive responses to these stresses. However, little is known about how such stresses cause plant growth inhibition. Many genes have been reported to be repressed in plants under abiotic stress conditions. ZPT2 (for petunia [ Petunia hybrida ] zinc-ﬁnger protein 2)-related proteins with two Cys2/His2-type zinc-ﬁnger motifs and an ethylene-responsive element binding factor-associated amphiphilic repression motif are thought to function as transcriptional repressors. To characterize the roles of this type of transcriptional repressor under abiotic stress conditions, we analyzed the functions of two Arabidopsis ( Arabidopsis thaliana ) ZPT2 -related genes that were induced by osmotic stress and abscisic acid: AZF1 (for Arabidopsis zinc-ﬁnger protein 1) and AZF2 . The nuclear localization of these two proteins was observed in the roots under control conditions, and the accumulation of AZF2 was clearly detected in the nuclei of leaf cells under stress conditions. Transgenic plants overexpressing AZF1 and AZF2 were generated using stress-responsive promoters or the GVG chemical induction system. The overexpression of these genes caused severe damage to plant growth and viability. Transcriptome analyses of the transgenic plants demonstrated that AZF1 and AZF2 repressed various genes that were down-regulated by osmotic stress and abscisic acid treatment. Moreover, many auxin-responsive genes were found to be commonly down-regulated in the transgenic plants. Gel mobility shift assays revealed that both the AZF1 and AZF2 proteins bound to the promoter regions of these down-regulated genes. These results indicate that AZF1 and AZF2 function as transcriptional repressors involved in the inhibition of plant growth under abiotic stress conditions.

Plants are exposed to various environmental stress conditions, such as drought, high salt, and low tem-perature. In response to these stresses, plants have evolved a number of mechanisms to achieve an optimal adaptation to adverse conditions. Transcriptional modulation is thought to be one of the most important ways that plants respond and adapt to stress conditions. A number of genes have been reported to be induced or repressed in plants under stress conditions (Kreps et al., 2002;Yamaguchi-Shinozaki and Shinozaki, 2006). Various transcription factors have been shown to be involved in stress responses, such as those from the dehydration-responsive elementbinding protein (DREB), ethylene-responsive element binding factor (ERF), zinc-finger, WRKY, MYB, basic helix-loop-helix, and basic domain-leucine zipper (bZIP) families. These transcription factors function as transcriptional activators or repressors and control downstream gene expression in stress signal transduction pathways (Chen et al., 2002;Yamaguchi-Shinozaki and Shinozaki, 2006).
The plant hormone abscisic acid (ABA) plays important roles in the acquisition of dehydration and desiccation tolerance in vegetative tissues and during seed development (Finkelstein et al., 2002;Yamaguchi-Shinozaki and Shinozaki, 2006). ABA treatment induces not only stress tolerance but also growth inhibition in plants. Many of the drought-or salt stress-inducible genes are also regulated by ABA (Yamaguchi-Shinozaki and Shinozaki, 2006). A conserved cis-element known as the ABA-responsive element (ABRE) has been identified in the promoters of ABA-inducible genes (Busk and Pagès, 1998;Leung and Giraudat, 1998). Although various genes are induced by ABA, many ABA-down-regulated genes have also been reported (Hoth et al., 2002;Seki et al., 2002;Takahashi et al., 2004).
Recently, transcriptional repressors have been found to play important roles in modulating plant defenses against biotic and abiotic stresses (Kazan, 2006). In plants, the ERF-associated amphiphilic repression (EAR) motif (L/FDLNL/FXP) containing a DLN box was first identified in the C-terminal regions of the class II ERF transcription factors and the Cys2/His2-type zinc-finger proteins. This motif was shown to be essential for the repressive activity of these transcription factors (Ohta et al., 2001). Recently, it was also reported that the EAR motifs of the Novel Interactor of JAZ transcriptional repressor and the Auxin/Indole-3-Acetic Acid (AUX/ IAA) proteins are responsible for the interaction between those proteins and the Groucho/Tup1-type corepressor TOPLESS (Szemenyei et al., 2008;Pauwels et al., 2010). Many Cys2/His2-type zinc-finger proteins have functional EAR motifs and are thought to function as transcriptional repressors (Ohta et al., 2001).
The Cys2/His2-type zinc finger, which is also referred to as the classical or TFIIIA-type finger, is one of the best-characterized DNA-binding motifs among the eukaryotic transcription factors (Laity et al., 2001). This motif can be represented as CX 2-4 CX3FX5LX2HX 3-5 H and consists of approximately 30 amino acids and two pairs of conserved Cys and His residues that bind tetrahedrally to a zinc ion (Pabo et al., 2001). Many Cys2/His2-type zinc-finger proteins in plants have structural features that are unique to plant zinc-finger proteins (Takatsuji, 1999). Most fingers have a highly conserved QALGGH motif in the zinc-finger helices. In multiple-fingered proteins, each zinc-finger motif is separated by long spacers with a variety of lengths and sequences; in yeast and animals, Cys2/His2-type fingers are mostly clustered and separated by a short spacer (six to eight amino acids) known as an HC link (Klug and Schwabe, 1995). The plant zinc-finger proteins are thought to recognize target sequences and to regulate gene expression in a plant-specific manner (Takatsuji, 1999;Sakamoto et al., 2004).
The first Cys2/His2-type zinc finger protein found in plants as the DNA-binding protein was ZPT2-1 (previously named EPF1) in petunia (Petunia hybrida; Takatsuji et al., 1992). ZPT2-related genes encode twofingered proteins and include 14 members in petunia (Kubo et al., 1998;Takatsuji, 1999). Many other ZPT2related zinc-finger proteins have been reported in different plant species, including wheat (Triticum aestivum), petunia, Arabidopsis (Arabidopsis thaliana), and rice (Oryza sativa). One of the Arabidopsis ZPT2-related genes, STZ/ZAT10 (for salt tolerance zinc finger/zinc finger of Arabidopsis 10), was cloned by complementation of the salt-sensitive phenotype of a yeast calcineurin mutant, and its expression conferred salt tolerance in wild-type yeast (Lippuner et al., 1996). The expression of STZ has been shown to respond to drought, salt, cold, and ABA (Sakamoto et al., 2000;Gong et al., 2001;Lee et al., 2002). The constitutive expression of STZ was found to result in growth suppression and an enhanced adaptation of plants to drought and osmotic stresses (Sakamoto et al., 2004;Mittler et al., 2006). AZF1 (for Arabidopsis zinc-finger protein 1) and AZF2 belong to a subset of the Arabidopsis ZPT2-related genes and are closely related to STZ in structure (Englbrecht et al., 2004). Their expression is induced by drought, salt, and ABA (Sakamoto et al., 2000). AZF1 and AZF2 both contain a functional EAR motif and appear to function as transcriptional repressors, as does STZ (Ohta et al., 2001;Lee et al., 2002;Sakamoto et al., 2004). These proteins bind to A (G/C)T repeats within an EP2 sequence that is known to be a target sequence of several petunia ZPT2 proteins (Sakamoto et al., 2004). The preceding reports have suggested potential roles for AZF1 and AZF2 under abiotic stress conditions, thus necessitating further characterization of their functions as transcriptional repressors.
In this study, we analyzed the roles of AZF1 (At5g67450) and AZF2 (At3g19580) during abiotic stresses using transgenic plants expressing AZF1 and AZF2 under the control of stress-responsive or glucocorticoid-inducible promoters. We found that AZF1 and AZF2 repressed the expression of various genes, including osmotic stressand ABA-repressive genes and auxin-inducible genes.

STZ Family Genes Are Conserved across Divergent Plant Species and Are Differentially Expressed under Various Stress Conditions
STZ/ZAT10 shares a high level of homology with five proteins among the 18 members of the ZPT2related genes that encode proteins with two Cys2/ His2-type zinc fingers in Arabidopsis: AZF1, AZF2, AZF3, ZAT6, and ZAT8 (Supplemental Fig. S1, A and B; Englbrecht et al., 2004;Sakamoto et al., 2004;Ciftci-Yilmaz and Mittler, 2008). We collectively named these six genes the STZ family. To determine whether STZrelated genes are conserved in plants other than Arabidopsis, we searched for orthologous sequences in angiosperms, gymnosperms, pteridophytes, and bryophytes using the Phytozome (http://www.phytozome. net/index.php) and National Center for Biotechnology Information BLAST (http://blast.ncbi.nlm.nih.gov/ Blast.cgi) services. Many genes that were orthologous to the STZ family were found in these plants, and a neighbor-joining tree analysis was performed based on the sequences of three protein regions: the L box, two zinc fingers, and the DLN box (Supplemental Figs. S1A and S2). The tree showed that the STZ family-type genes were conserved across divergent plant species, especially in angiosperms, and that they were likely to be evolutionarily separated from the other homologs in Arabidopsis and their orthologs (Supplemental Fig. S2).
In this study, we initially focused on AZF1, AZF2, and ZAT8 because we had already reported the functions of STZ/ZAT10 under stress conditions (Sakamoto et al., 2004). We used northern-blot analysis to compare the expression patterns of the six genes of the STZ family in response to various stresses (Supplemental Fig. S3). In agreement with previously reported results (Sakamoto et al., 2004), the expression of AZF1 and AZF2 was induced by drought, salt, and ABA treatment, although the amount of AZF1 transcripts was relatively small. However, the expression level of ZAT8 was extremely low, and we could not detect a clear induction of ZAT8 under stress conditions. Therefore, we performed further research on AZF1 and AZF2.

Subcellular Localization Analysis of the AZF1 and AZF2 Proteins
The AZF1 and AZF2 proteins have been shown to localize to the nucleus (Sakamoto et al., 2004). We Figure 1. Induction and cellular localization of the AZF1-GFP and AZF2-GFP proteins in the roots and leaf epidermis. A, Confocal images of GFP fluorescence in the roots of transgenic Arabidopsis plants harboring the 35Spro:GFP, AZF1pro: AZF1-GFP, and AZF2pro:AZF2-GFP constructs. Two-week-old plants were treated with or without 100 mM ABA for 5 h. Bars = 50 mm. B, Confocal images of GFP fluorescence in the leaf epidermis of AZF2pro:AZF2-GFP plants. Two-week-old plants were treated with liquid MS medium with or without 150 mM NaCl or 100 mM ABA for 30 and 10 h, respectively. Arrows indicate the guard cells. Bars = 10 mm. C, Root cells of AZF2pro: AZF2-GFP plants stained with DAPI were photographed using differential interference contrast, GFP fluorescence, and DAPI fluorescence microscopy. The arrows indicate the nuclei. Bar = 50 mm. confirmed that all of the STZ-group proteins localized to the nucleus in Arabidopsis protoplasts (Supplemental Fig. S4). To identify the tissue specificity of the AZF1 and AZF2 proteins under nonstress and stress conditions in plants, we generated transgenic Arabidopsis plants that contained the AZF1-GFP and AZF2-GFP fusion genes driven by unique promoters (AZF1pro: AZF1-GFP and AZF2pro:AZF2-GFP) and analyzed the resulting fluorescence following exposure to nonstress and stress conditions. In both transgenic plants, the fluorescence was localized to the nuclei of the roots under nonstress conditions and was not detected in any aerial plant parts (Fig. 1A). The fluorescent signals of each protein were predominantly detected in the zones of maturation and were not observed in the zones of division and elongation in roots. In contrast, the AZF2-GFP signals were detected in the root tips when the plants were treated with exogenous ABA. The AZF2-GFP signals were also observed in leaves when the plants were treated with high salt and ABA (Fig. 1B). When the plants were exposed to high-salt stress, nuclear localization of the AZF2-GFP protein was observed in both epidermal and guard cells. However, in response to ABA treatment, the nuclear localization was observed mainly in guard cells. In root epidermal cells, colocalization of the AZF2-GFP signals with 4#,6diamidine-2#-phenylindole dihydrochloride (DAPI)stained nuclear DNA was observed (Fig. 1C).
We also examined the tissue specificity of AZF1 and AZF2 expression by evaluating the expression of a GUS reporter gene driven by the AZF1 and AZF2 promoters in transgenic plants in response to drought and ABA treatment. A histochemical analysis of GUS activity in rosette plants revealed that AZF1 and AZF2 were expressed mainly in roots, excluding the zones of division and elongation, under normal growth conditions (Supplemental Fig. S5A). AZF1 expression was restricted to roots under both stress and nonstress conditions. The expression of AZF2 was weakly detected in the cotyledons, even in the absence of stress, and was induced in the rosette leaves in response to drought stress. AZF2 expression was also detected in the flowers, specifically in the petals and stamens, irrespective of drought stress (Supplemental Fig. S5B). AZF2 promoter activity was induced in the roots and aerial plant parts by the application of exogenous ABA (Supplemental Fig. S5C). In the roots, a marked induction of AZF2 expression by ABA was detected in the zones of division. In the leaves, ABA enhanced the expression of AZF2 in the guard cells to a greater extent than did the other treatments.

Effects of AZF1 and AZF2 Gene Overexpression in Transgenic Plants
To investigate the roles of AZF1 and AZF2 in plants under stress conditions, we previously attempted to generate transgenic Arabidopsis plants overexpressing AZF1 or AZF2 under the control of an enhanced 35S promoter (Sakamoto et al., 2004). However, the transformation efficiency of these plants was extremely low, and we obtained neither AZF1-nor AZF2-overexpressing plants (Sakamoto et al., 2004). Therefore, we decided to generate transgenic plants overexpressing AZF1 or AZF2 driven by a stress-inducible RD29A promoter (Kasuga et al., 1999). To evaluate the expression levels of AZF1 and AZF2 in the transgenic plants under high-salt conditions, 16 independent transgenic lines for each transgenic plant were subjected to RNA gel-blot analyses. Three lines from each transgenic plant in the T2 generation that demonstrated elevated salt-induced expression levels of AZF1 or AZF2 were selected for further analyses ( Fig. 2A). Moreover, we generated transgenic plants that overexpressed AZF2 under the control of its own promoter. From 24 independent lines, we selected T2 lines that exhibited higher expression levels of AZF2 under salt-stress conditions compared with the vector control. Finally, we selected three stable lines from the T3 generation of the AZF2 transformants for further studies (Fig. 2B). Under nonstress conditions, these RD29Apro:AZF1, RD29Apro:AZF2, and AZF2pro:AZF2 plants displayed similar growth and morphological phenotypes (Supplemental Fig. S6). The salt sensitivities of these three transgenic plants were compared with those of the vector control plants. All of the transgenic plants displayed salt-hypersensitive phenotypes (Fig. 2, C-E). Whereas more than 60% of the vector control plants survived after salt-stress treatments, the survival rates of the RD29Apro:AZF1 and RD29Apro:AZF2 plants were less than 35% (Fig. 2, F and G). Up to 25% of the AZF2pro:AZF2 plants survived, an approximately 40% decrease compared with the vector control plants (Fig. 2H). These phenotypes appeared to depend on the levels of AZF1 and AZF2 expression. We also generated transgenic AZF1pro:AZF1 (T3 generation) plants in a manner similar to that described for the AZF2pro:AZF2 plants. However, the levels of saltinduced AZF1 expression in these transgenic plants were not significantly higher than those detected in the vector control plants (data not shown), which is consistent with the low inducibility of AZF1 observed under various stress conditions, as shown in Supplemental Fig. S3. We were unable to detect clear differences in the extent of sensitivity to salt stress between the AZF1pro:AZF1 and vector control plants (data not shown).
Next, we generated transgenic plants carrying AZF1 and AZF2 under the control of a glucocorticoid-inducible promoter (Aoyama and Chua, 1997). Treatment with 10 mM dexamethasone (DEX) was used to induce expression in these plants. Three independent lines that demonstrated high DEX-induced AZF1 (pTA7002: AZF1) and AZF2 (pTA7002:AZF2) expression were selected for further analyses (Fig. 2I). The growth of the pTA7002:AZF1 and pTA7002:AZF2 plants on germination medium (GM) agar plates supplemented with or without 1 mM DEX after sowing was compared with that of the vector control plants. Both of the transgenic lines on DEX-containing plates displayed dwarfed growth with small curled leaves, but no phenotypes that were clearly different from the wild type were observed in the absence of DEX (Fig. 2J). For the salt-sensitivity tests, plants grown on GM agar plates were transferred to 0.53 Murashige and Skoog (MS) agar plates containing 1 mM DEX or 0.01% ethanol Each lane contained 20 mg of total RNA from 2-week-old plants that had been treated with or without 200 mM NaCl. rRNAs are shown as equal loading controls. C to E, Evaluation of salinity stress in plate conditions. Plants grown on GM plates were transferred onto a MS medium plate supplemented with or without 175 mM NaCl. The plates were observed for approximately 9 to 10 d after the transfer. F to H, Percentages of the plants that survived and SD (error bars) calculated from three independent experiments (n . 30).
The asterisks indicate significantly lower survival rates compared with the vector control as determined by Student's t-test (* P , 0.05, ** P , 0.01). I, Expression of AZF1 and AZF2 in the pTA7002:AZF1, pTA7002:AZF2, and vector control plants. Each lane contained 20 mg of total RNA from 3-week-old plants that had been treated with or without 10 mM DEX. rRNAs are shown as equal loading controls. J, Growth phenotypes of pTA7002: AZF1, pTA7002:AZF2, and vector control plants after 2 weeks of growth on GM agar plates supplemented with or without 1 mM DEX.
with or without NaCl. In the absence of DEX, no clear differences between the vector control and the transgenic plants were observed, irrespective of the application of NaCl treatment. In contrast, these transgenic plants exhibited severe salt-sensitive phenotypes following treatment with a combination of DEX and 175 mM NaCl, which indicated that the overexpression of AZF1 and AZF2 enhanced salt sensitivity in the transgenic plants (Supplemental Fig. S7).

AZF1 and AZF2 Down-Regulate Many ABA-and Osmotic Stress-Repressive Genes
To unravel the transcriptional networks of AZF1 and AZF2, we first compared the expression profiles of the pTA7002:AZF1 and pTA7002:AZF2 plants with those of the vector control plants using the Arabidopsis 3 Oligo Microarray (Agilent Technologies), which contains more than 44,000 genes. For the microarray analyses, we used RNA extracted from plants treated with or without 10 mM DEX for 24 h. In the pTA7002:AZF1 plants, a total of 468 and 1,076 genes were downregulated and up-regulated with a ratio of greater than 2, respectively. In the pTA7002:AZF2 plants, 1,672 and 1,774 genes were down-regulated and up-regulated with a ratio of greater than 2, respectively. Because these zinc-finger proteins have been reported to function as transcriptional repressors (Ohta et al., 2001;Sakamoto et al., 2004), we focused on the genes for which expression was repressed in the pTA7002:AZF1 and pTA7002: AZF2 plants. Using Genevestigator (https://www. genevestigator.com/gv/index.jsp), we determined the expression profiles of the down-regulated genes in the pTA7002:AZF1 and pTA7002:AZF2 plants. The top 100 down-regulated genes in the respective overexpressors included many genes that demonstrated reduced expression levels in response to osmotic stresses and/or ABA (Fig. 3A). Thus, many stress-repressive genes were down-regulated in both transgenic plants overexpressing AZF1 and those overexpressing AZF2. In contrast, we were unable to detect any distinct relationship between the up-regulated genes in the AZF1 and AZF2 overexpressors and osmotic stress and/or ABA-responsive genes (Supplemental Fig. S8A).
We also performed microarray analyses to identify genes that were down-regulated in response to high-salt stress in the AZF2pro:AZF2 plants. Both the AZF2pro:AZF2 and vector control plants were treated with a high concentration of salt (200 mM NaCl) or water for 5 h prior to the microarray experiments. Overall, 89 and 70 genes were down-regulated with a ratio of greater than 2 in the high-salt-and watertreated AZF2pro:AZF2 plants, respectively, compared with the vector control plants (Supplemental Table S1). The group of 89 genes included many osmotic stressand ABA-repressive genes (Fig. 3A). Overall, 31 and 27 genes were down-regulated more than 2-fold in the plants that were treated with ABA and high salt, respectively . Moreover, among the 89 genes, two sets of 24 genes were down-regulated in the pTA7002:AZF1 and pTA7002:AZF2 plants, respectively, and 18 of these 89 genes were down-regulated in both types of transgenic plants (Supplemental Table  S1). In contrast, the repressed genes in the watertreated transgenic plants did not exhibit clear, common expression patterns following exposure to the various stimuli (Supplemental Fig. S8B).
The functional characteristics of these down-regulated genes were analyzed using The Arabidopsis Information Resource Gene Ontology annotation search tool (http://www.arabidopsis.org/tools/bulk/go/index.jsp) and the Pfam batch sequence search tool (http://pfam. janelia.org/search#tabview=tab1). We obtained the ontological profiles of the 89 genes that were downregulated in the AZF2pro:AZF2 plants and the top 100 genes that were down-regulated in the pTA7002:AZF2 and pTA7002:AZF1 plants (Fig. 3, B-D). We found that many genes known to be involved in transcription and various metabolic pathways, such as carbohy-drate, lipid, and secondary metabolism, were downregulated in these transgenic plants.

AZF1/AZF2-Regulated SAUR Genes in Plants
It is notable that many of the auxin-related genes were down-regulated in the AZF2pro:AZF2 plants that were treated with high salt, and most of these genes were small auxin-up RNA (SAUR) genes (Supplemental Table S1). Among the 18 genes that were commonly down-regulated in the pTA7002:AZF1 and pTA7002: AZF2 plants, 15 were SAURs, and almost all were repressed in response to stress. To validate the microarray results, we selected five genes (SAUR16, -20, -21, -26, and -63) whose expression was down-regulated due to the overexpression of AZF1 and AZF2. Using quantitative real-time (qRT)-PCR, we found that the expression levels of these SAUR genes were significantly reduced in the AZF2pro:AZF2 plants under high-salt conditions; the lowest expression for most of these genes was found in the transgenic lines displaying the highest expression of the AZF2 transcript (Fig. 4).
The expression analyses using the transgenic plants overexpressing AZF1 and AZF2 suggest that AZF1 and AZF2 control the expression levels of the SAUR genes. The SAUR genes form a large family in Arabidopsis that includes over 70 members (Hagen and Guilfoyle, 2002). A total of 27 SAURs were downregulated by either AZF1 or AZF2, a number that constitutes approximately 40% of the SAUR family members. The SAUR gene family can be phylogenetically classified into three clades, and we determined the positions of these 27 genes in a phylogenic tree (Fig. 5). Interestingly, 25 of the 27 SAUR genes were included in clades I or II. Clade I contains 22 genes, 11 of which were down-regulated in AZF2pro:AZF2, pTA7002:AZF1, or pTA7002:AZF2 plants. A total of 12 of the 22 clade I genes, including SAUR62, -63, -64, -65, -66, -67, and -68, are auxin-up-regulated genes (Redman et al., 2004;Nemhauser et al., 2006). The second clade consists of 16 genes, including 14 that were downregulated in AZF1 or AZF2 overexpressors: SAUR13, -15, -23, -27, and -28 are up-regulated by auxin. Although the third clade consists of 28 members and is the largest of the three clades, it contains only two genes (SAUR46 and -58) that were down-regulated by AZF1 or AZF2 and only four genes (SAUR34, -35, -45, and -46) that were up-regulated by auxin.
The expression profiles of all the SAUR genes in the roots and vegetative rosettes were obtained from the eFP browser (http://bbc.botany.utoronto.ca/efp/cgibin/efpWeb.cgi). The genes in clades I and II had a tendency to display higher expression levels in the leaves and lower expression levels in the roots under nonstress conditions (Supplemental Fig. S9A). In contrast, the genes in clade III demonstrated expression patterns opposite to those of the genes in the other two clades. Our Web-based transcriptome analyses of the SAURs indicated that many of the genes in clades I and II were repressed by ABA and down-regulated by osmotic stress, but the genes in clade III did not demonstrate this tendency (Supplemental Fig. S9B). This classification of the SAUR genes revealed that AZF1 and AZF2 could specifically down-regulate the expression levels of the members of clades I and II. Both clades are likely to contain genes that respond to auxin, that show lower expression levels in the roots and higher expression levels in the leaves under nonstress conditions, and that are down-regulated by osmotic stress and ABA.

Analyses of Mutant Plants under Salt-Stress Conditions
Our microarray results indicated that the expression levels of many SAUR genes were down-regulated in the pTA7002:AZF2 and pTA7002:AZF1 plants. To further examine the effects of AZF1 and AZF2 on the expression levels of the SAUR genes, we obtained T-DNA insertion mutants of azf1 and azf2 in a Columbia background and generated double mutants. A semiquantitative reverse transcription-PCR analysis confirmed that AZF1 and AZF2 expression was com- Figure 5. Phylogenetic analysis of the SAUR family in Arabidopsis. The sequence alignment and phylogenetic tree were prepared using the ClustalX program (version 2.0) and MEGA4 software as described in Supplemental Experimental Procedures S1. The genes that were down-regulated in AZF2pro:AZF2, pTA7002:AZF2, and pTA7002:AZF1 plants are indicated by red, green, and yellow circles, respectively. pletely eliminated by the T-DNA insertion in the homozygous mutants (Fig. 6, A and B). Compared with the wild-type plants, all of the mutants displayed similar growth phenotypes on the GM agar plates and soil pots under normal conditions (Fig. 6, C and D). Double mutant plants tended to show delayed chlorosis or enhanced root growth compared with wild-type plants under high-salt conditions. However, those phenotypic differences were unstable and were difficult to evaluate statistically. We then analyzed the expression of SAURs under control and high-saltstress conditions in the mutants. The lack of AZF1 and AZF2 transcripts in the azf1 azf2 mutant resulted in increased levels of SAUR gene expression under high-salt conditions, and these expression patterns were not detected in the absence of salt stress (Fig. 6, E and F). However, the expression levels of the SAURs in the azf1 and azf2 single mutants were similar to those detected in wild-type plants. These results suggest that AZF1 and AZF2 coordinately repress the expression of these SAUR genes in response to osmotic stresses.
Binding of the AZF1 and AZF2 Proteins to the Promoter Regions of SAUR Genes As described above, many genes were down-regulated in the transgenic plants overexpressing AZF1 or AZF2, and the expression of many SAUR genes was commonly repressed in the transgenic plants. Using gel mobility shift assays, we examined whether AZF1 and AZF2 proteins could recognize the promoter regions of SAUR genes. Because it was very difficult to prepare full-length recombinant AZF1 and AZF2 proteins in bacterial expression systems, we decided to use the truncated forms of recombinant AZF1 and AZF2. Truncated AZF1 and AZF2 proteins including two canonical zinc-finger motifs were expressed as in-frame fusion proteins with the maltose-binding protein in Escherichia coli (Fig. 7A). We divided the 1,030-bp promoter regions of SAUR20 and SAUR63 into four 280-bp fragments that contained 30-bp overlap sequences at their 3# ends. Then, we examined whether the recombinant AZF1 and AZF2 proteins in azf2, and AZF1 and AZF2 in azf1 azf2 mutants and the expression of each gene in wild-type (WT) plants exposed to drought conditions for 2 h. Actin cDNA was amplified as a control. C, The plants were grown on GM plates for 3 weeks. D, Threeweek-old plants were transferred from GM plates to soil and grown for an additional 3 weeks. E and F, qRT-PCR analyses of the transcript levels of the selected AZF1/AZF2 downstream genes in mutant and wild-type plants. The highest expression level of each gene was designated as 100. The data represent means 6 SD of triplicate experiments.
could interact with the SAUR63 and SAUR20 promoter regions. Gel shift assays showed that the recombinant AZF1 and AZF2 proteins bound to the bp 2280 to 21 region of the SAUR63 promoter and to the bp 2780 to 2501 region of the SAUR20 promoter (Fig. 7B). Competition experiments showed that interactions between the proteins and the DNA fragments were competed gradually, depending on the concentration of unlabeled oligonucleotides, and disappeared following the addition of a 100-fold excess of competitors.

DISCUSSION
Our study indicates that two Cys2/His2-type zincfinger proteins with EAR motifs, AZF1 and AZF2, can function as transcriptional repressors under osmotic stresses. Using transgenic plants overexpressing AZF1 or AZF2, we empirically demonstrated that the overexpression of AZF1 or AZF2 repressed the expression of many osmotic stress-and ABA-repressive genes in plants and severely affected seedling growth. Below, we discuss possible downstream targets shared by the AZF1 and AZF2 proteins.
Under abiotic stress conditions, such as drought, high salt, and low temperature, many genes have been reported to be repressed in plants (Hoth et al., 2002;Seki et al., 2002;Takahashi et al., 2004). ABA is a key regulator that mediates abiotic stress signaling pathways and induces not only stress tolerance but also growth inhibition (Finkelstein et al., 2002). Previously, we revealed that the transcription levels of AZF2 were markedly decreased in both areb1 areb2 abf3 and srk2d srk2e srk2i triple mutants using microarray analyses Nakashima et al., 2009;Yoshida et al., 2010). In this work, northern-blot analysis indicated that the induction of AZF2 expression by ABA was strong and persisted over time (Supplemental Fig.  S3). The AZF2 protein accumulated in the leaf guard cells following ABA treatment (Fig. 1B). Considering these results together with the presence of the ABREs in the AZF2 promoter, the expression of AZF2 appears to be regulated mainly by ABA, and this gene is one of the downstream targets of the AREB/ABF bZIP-type transcription factors.
Transgenic plants overexpressing AZF1 and AZF2 under the control of a glucocorticoid-inducible promoter showed severe growth retardation with morphological defects (Fig. 2J). This dwarf phenotype is consistent with the results obtained from the overexpression of Cys2/His2-type zinc finger genes, such as STZ and ZAT7 (Sakamoto et al., 2004;Mittler et al., 2006;Ciftci-Yilmaz et al., 2007). The reduced expression of genes involved in carbohydrate and lipid metabolism might lead to the growth inhibition observed in AZF1 and AZF2 overexpressors (Fig. 3, B-D). Although the number of photosynthesis-related genes with reduced expression was small, these genes' reduced expression may have also affected the growth of the plants. Previous microarray gene expression anal- Figure 7. Interactions between the recombinant AZF1 and AZF2 proteins and the promoter regions of the SAUR genes. A, Schematic representation of full-length and truncated AZF1 and AZF2. The black boxes indicate zincfinger (ZF) motifs. Amino acid numbers are from the full-length proteins. B, DNA-binding activities were analyzed by gel shift assays. 32 P-labeled 280-bp probes (0.2 ng) containing promoter fragments of SAUR63 (2280 to 21 bp) and SAUR20 (2780 to 2501 bp) were incubated with maltose-binding protein (MBP; 11 ng), MBP-AZF1b (200 ng), or MBP-AZF2b (13 ng). Unlabeled probes were used as competitors. The amounts of the competitors were 2 ng (103), 6 ng (303), and 20 ng (1003). The shifted bands are indicated by arrows.
ysis has shown that the expression of many genes involved in carbohydrate metabolism and photosynthesis was down-regulated under drought, high-salt, and cold stress conditions (Seki et al., 2002). Growth suppression following a reduction in these types of proteins is a critical phenomenon in plant stress responses.
Surprisingly, the plants that overexpressed AZF1 and AZF2 under the control of stress-responsive promoters showed a hypersensitive phenotype to salt stress (Fig. 2). Moreover, the high overexpression of AZF1 and AZF2 induced by DEX treatment was lethal to the plants even under nonstress conditions (Supplemental Fig. S7). Because the enhanced sensitivity to high salt was also promoted by the overproduction of AZF2 transcripts under the control of the native promoter of AZF2 (Fig. 2E), these results were probably not caused by the ectopic expression of AZF1 and AZF2; instead, they probably reflect functions of these proteins. The difficulty in generating transgenic plants that constitutively express AZF1, AZF2, and AZF3 under the control of the cauliflower mosaic virus 35S promoter (Sakamoto et al., 2004) suggests that the overexpression of these genes may have a harmful effect on plant growth or development. However, the overexpression of STZ has been shown previously to enhance the adaptation of a plant to drought stress (Sakamoto et al., 2004). Another group has reported that both the knockout and the overexpression of STZ improved the plant's adaptation to osmotic and salinity stresses (Mittler et al., 2006). The varying results between the overexpression of these two AZFs and STZ may reflect partial differences in their sets of downstream genes. In fact, the constitutive expression of STZ enhanced the expression of the oxidative stressresponsive genes ascorbate peroxidase2 (APX2), Fe-superoxide dismutase1, and APX1 (Mittler et al., 2006). However, the transcript levels of these reactive oxygen species defense genes were not altered in the AZF1 and AZF2 overexpressors used in this study, and few SAUR genes were down-regulated in the DEXinduced STZ overexpressors (data not shown). The expression of many stress-inducible genes, such as DREB1A, RD29A, KIN2, cold-regulated15A (COR15A), and COR15B, was also suppressed in the transgenic plants that had enhanced expression of AZF1 and AZF2 under the control of a glucocorticoid-inducible promoter. Therefore, we suspect that these types of suppression had a causal influence on the enhanced salt sensitivity of the transgenic plants. The short-term salt treatment and mild enhancement of AZF2 transcript levels might lead to almost no down-regulation of stress-inducible genes in plants that overexpress AZF2 under the control of its own promoter. We observed the stress responses of RD29Apro:AZF1, RD29Apro:AZF2, and AZF2pro:AZF2 plants under drought stress. We also analyzed the effects of ABA treatment on AZF2pro:AZF2 seedlings. However, these transgenic plants showed no clear differences from the control plants (data not shown). These observations suggest the strong influence of the overproduction of AZF1 and AZF2 transcripts under high-salt conditions. The analysis of microarray data from the Genevestigator database revealed that the expression levels of many osmotic stress-and ABA-repressive genes were down-regulated in both the AZF1 and AZF2 overexpressors (Fig. 3A). Although the transcript abundance of a large number of genes was both induced and repressed in the transgenic plants that overexpressed AZF1 and AZF2 under the control of a glucocorticoid-inducible promoter, the enhanced expression of AZF2 using its native promoter resulted in a much lower number of up-regulated than down-regulated genes. Moreover, the stress-responsive expression patterns of the up-regulated genes were ambiguous compared with those observed for the down-regulated genes (Supplemental Fig. S8A). The moderate induction of AZF2 using its own promoter might influence the clear appearance of gene suppression in plants. In agreement with the gain-of-function study, the suppression of gene expression caused by osmotic stress was partially attenuated in the azf1 azf2 double mutant plants (Fig. 6, E and F). These results indicate that AZF1 and AZF2 function as repressors to regulate the expression of specific genes under stress conditions. Although many osmotic stress-and ABA-repressive genes were down-regulated by AZF1 and AZF2, some were auxin response components (Fig. 3A). Many auxin-responsive genes were down-regulated in the transgenic plants that overexpressed AZF2 using its native promoter (Supplemental Table S1). These genes include IAA5, IAA29, auxin-regulated gene involved in organ size (ARGOS), and many SAUR genes. Aux/IAA proteins are proposed to be transcriptional repressors that play crucial roles in auxin signaling by interacting with auxin response factors (Guilfoyle and Hagen, 2007). The IAA5 gene is thought to be involved in the growth acceleration in response to auxin and brassinosteroids (Goda et al., 2002). The IAA29 gene has been shown to be a component of auxin-mediated elongation growth in shade avoidance responses (Tao et al., 2008). ARGOS is thought to transduce auxin signals to regulate cell proliferation and organ growth during organogenesis (Hu et al., 2003). SAUR genes are the early auxin-responsive genes that constitute a large multigene family in plants (Paponov et al., 2008). It is noteworthy that almost all of the SAUR genes commonly repressed by AZF1 and AZF2 belong to the groups indicating relatively high auxin responsiveness ( Fig. 5; Supplemental Fig. S9). Several of these SAUR genes have been shown to be expressed mainly in the elongation tissues of maize (Zea mays) and soybean (Glycine max), which suggests that they have important roles in auxin-mediated cell elongation in plants (Gee et al., 1991;Knauss et al., 2003). The high expression levels of the SAUR genes from clades I and II in shoots and their up-regulation in response to auxin also support the view that these genes are important for the stimulation of shoot elongation (Paponov et al., 2008; Supplemental Fig. S9). It has been reported that the transcript levels of AZF1 and AZF2 were minimally increased by the synthetic auxin 2,4-dichlorophenoxyacetic acid (Sakamoto et al., 2004). Using the eFP browser, we found that the expression of AZF1 or AZF2 was relatively constant, regardless of auxin (IAA) treatment. These findings indicate that AZF1 and AZF2 repress the expression of several genes that may be involved in auxin signaling and auxin-mediated plant growth under osmotic stress conditions.
Although AZF2 transcription was clearly induced by osmotic stress and ABA treatment, the induction of AZF1 was slight (Supplemental Fig. S3). Using transgenic plants that contained the AZF1pro:AZF1-GFP and AZF2pro:AZF2-GFP fusion genes, we detected osmotic stress-induced AZF2 protein in leaves under high-salt conditions, but the induction of AZF1 was not clear (Fig. 1B). Hence, the repression activity of AZF1 under osmotic stresses may be much weaker than that of AZF2. In contrast to our expectations, the suppression of gene expression was much more reduced in the double mutant than in any single mutant (Fig. 6, E and F). Therefore, we speculate that the AZF1 and AZF2 proteins share several downstream genes and regulate gene expression in a coordinated manner under osmotic stresses. However, the azf1 and azf2 single and double mutant seedlings did not show obvious phenotypes in response to ABA treatment or osmotic stresses, such as drought, high salt, and cold, which suggests a high degree of functional redundancy among the various gene family members.
According to the data from the eFP database, the expression levels of the SAURs from clades I and II were low in roots and relatively high in shoots (Supplemental Fig. S9A), which was opposite to the expression patterns observed for AZF1 and AZF2 under normal growth conditions. Moreover, the expression of most of the SAUR genes, which belong to clades I and II, was down-regulated under drought, high salt, and ABA treatment conditions (Supplemental Table  S1; Supplemental Fig. S9B). Because expression analyses using the overexpressors and the mutant plants indicated that the expression of certain SAUR genes was at least partially regulated by both AZF1 and AZF2, it was intriguing to analyze whether the SAUR genes were the direct downstream targets of the AZF1 and AZF2 proteins. Gel shift assays showed that the AZF1 and AZF2 proteins interacted with promoter fragments of the SAUR63 and SAUR20 genes, respectively (Fig. 7). AZF1 and AZF2 contain the EAR repression domain and are thought to function as transcriptional repressors (Ohta et al., 2001;Sakamoto et al., 2004;Kazan, 2006). Therefore, our data suggest that the AZF1 and AZF2 proteins bind to the SAUR promoter regions to repress the expression of these genes.
It has been reported that the AZFs and STZ recognize the tandemly repeated A(G/C)T core sequences and that the sequences around the A(G/C)T repeats might influence binding (Sakamoto et al., 2004). The importance of the spaces and DNA sequences between the two core sites for the DNA-binding affinities of the petunia ZPT2-type proteins has also been reported (Takatsuji, 1999). We searched for the A(G/C)T core sequences within the promoter fragments of SAUR63 and SAUR20 and found a few sets of the A(G/C)T repeats in the fragments. Although it is possible that AZF1 and AZF2 recognize the promoter regions of SAUR63 and SAUR20 through those A(G/C)T repeats, further study is required to determine the binding sequences of the STZ family proteins within the native promoter regions in Arabidopsis.
The optimal binding affinities of the ZPT2-type proteins are greatly affected by the spacing between the units of the DNA-binding sites, whereas the recognition of the spacing and sequences can be partially relaxed, depending on the character of each protein (Takatsuji, 1999;Sakamoto et al., 2004). Therefore, it could be assumed that the STZ-related proteins have a dual nature in DNA recognition: (1) each member has several optimal, high-affinity binding sequences; and (2) there is combinatorial redundancy between the binding sites and STZ-related members, which leads to the complex overlapping of target genes. Thus, we propose that the overexpression of AZF1 and AZF2 preferentially repressed the expression of genes that had relatively ideal binding sequences for them and that this type of gene repression largely affected the gain-of-function phenotype. In contrast, the functional during osmotic stress conditions. Osmotic stress induces the expression of AZF1 and AZF2 via ABA-dependent or -independent pathways. The accumulation of AZF1 and AZF2 during stress results in the suppression of many ABA-repressive genes, including several auxin-responsive genes, such as SAURs. Consequently, AZF1 and AZF2 are assumed to partially regulate ABA and auxin signaling and to be involved in plant responses to osmotic stress. redundancy and compensatory functions among the homologous proteins might be expected to prevent the presentation of phenotypic differences in the mutants. We propose that when AZF1 and AZF2 are overexpressed in plants, they preferentially down-regulate many osmotic stress-repressive genes that are shared by both zinc fingers. Few of these shared genes are under the control of STZ. Despite the ability of AZF2 and STZ to bind to the same cis-acting element within an EP2 sequence (Sakamoto et al., 2004), differences in the downstream genes for which expression was preferentially regulated in the overexpressors can account for the different roles of AZF2 and STZ under specific stress conditions.
The phytohormone auxin has been recognized as a key player in the regulation of plant growth and development and in plant responses to environmental changes (Benjamins and Scheres, 2008;Tromas and Perrot-Rechenmann, 2010). Recent studies have suggested that there is a coordinated, generic "stressinduced morphogenic response" that is stimulated by abiotic stresses and that controls plant morphology (Potters et al., 2007). ABA and auxin are thought to play important roles in stimulating the stress-induced morphogenic response (De Smet et al., 2006;Zolla et al., 2010). It is proposed that AZF1 and AZF2 partially regulate cross talk between the ABA and auxin signal transduction pathways under stress conditions.
In this work, phylogenetic analyses revealed the existence of genes orthologous to AZF1 and AZF2 in angiosperms and many other plant species (Supplemental Fig. S2). Because those orthologs in bryophytes, pteridophytes, and gymnosperms belong to the clade in which almost all the proteins have three or more conserved fingers, ZPT2-related proteins might have evolved from zinc-finger proteins that have three or four fingers, such as At4g35280 and At5g56200. Therefore, it is suggested that the STZ-related genes, which consist of 18 members in Arabidopsis, have developed specific functions as transcriptional regulators.
In summary, we demonstrated that AZF1 and AZF2 can function as transcriptional repressors to inhibit the expression of osmotic stress-and ABA-repressive genes under abiotic stress conditions (Fig. 8). The highly induced expression of AZF2 in plants using its own promoter showed clear evidence that AZF2 represses gene expression under osmotic stress. AZF1 and AZF2 interacted with the promoter regions of several SAUR genes. These findings indicate the possibility that these proteins directly bind to the promoter regions of their target genes to regulate transcription. The gain-offunction phenotype suggests that the overexpression of AZF1 and AZF2 severely inhibits plant growth and development, leading to enhanced salt sensitivity in seedlings. In addition, AZF1 and AZF2 down-regulated the expression of many auxin-responsive genes, suggesting that the repression of these types of auxinrelated genes plays a role in plant responses to abiotic stresses. To cope with adverse environmental conditions, such as high salinity, drought, and low tem-perature, plants adjust by conserving energy and resources and reducing their growth. Therefore, the repression of gene expression, salt sensitivity, and inhibition of growth caused by the overexpression of AZF1 and AZF2 may be related to an adaptive response of plants to osmotic stresses. Because the seedlings of the AZF1 and AZF2 mutants showed no clear phenotype under osmotic stress, it is likely that functional redundancy and compensatory functions exist among the STZrelated gene family members. In the future, elucidating the combination of overlapping downstream targets between the AZF and ZAT genes will shed light on their complex interplay under various stress conditions.

Plant Materials and Growth Conditions
All of the Arabidopsis (Arabidopsis thaliana) lines described herein were derived from the Columbia wild-type line. The T-DNA insertion lines of azf1 (SALK_133011) and azf2 (SALK_132562) were obtained from the Arabidopsis Biological Resource Center. The homozygous azf1 and azf2 mutants were isolated, and the azf1 azf2 double mutants in the same backgrounds as the single mutants were constructed by genetic crosses. The primers that were used to screen homozygous mutants were designed according to http:// signal.salk.edu/isect.2.html and are listed in Supplemental Table S2. Unless otherwise stated, the plants were grown on GM agar plates, as described previously (Yamaguchi-Shinozaki and Shinozaki, 1994).

Constructs and Generation of Transgenic Plants
The primers used in this analysis are listed in Supplemental Table S2. To generate the AZF1pro:AZF1 and AZF2pro:AZF2 constructs, 1.6-and 2.2-kb fragments containing the AZF1 and AZF2 upstream sequences, respectively, were amplified from wild-type genomic DNA by PCR using KpnI-SmaI linker primers. The resulting fragments were digested with KpnI and SmaI and cloned into the KpnI and SmaI sites of pGK-El2-35S-CsGFP (Qin et al., 2008) to produce pGK-AZF1pro:CsGFP and pGK-AZF2pro:CsGFP, respectively. The entire 0.7-kb coding region of AZF1 and 0.8-kb coding region of AZF2 were amplified by PCR using BamHI-EcoRV linker primers. The PCR products were digested with BamHI and EcoRV and cloned into the BamHI-EcoRV sites of pGK-AZF1pro:CsGFP and pGK-AZF2pro:CsGFP to produce pGK-AZF1pro: AZF1 and pGK-AZF2pro:AZF2. To prepare the AZF1pro:AZF1-sGFP and AZF2pro:AZF2-sGFP constructs, the coding regions of AZF1 and AZF2 without stop codons, respectively, were amplified from AZF1pro:AZF1 and AZF2pro:AZF2, respectively, using the primers containing the SmaI-BamHI linkers. The generated fragments were inserted into the SmaI-BamHI sites of pGK-AZF1pro:CsGFP and pGK-AZF2pro:CsGFP to produce pGK-AZF1pro: AZF1-sGFP and pGK-AZF2pro:AZF2-sGFP, respectively. To construct the PBI-RD29Apro:AZF1 and PBI-RD29Apro:AZF2 plasmids, the AZF1 and AZF2 coding sequences were amplified with the XbaI-SmaI linker primers and inserted into the XbaI and SmaI sites of the pBIRD29AAP-Not vector (Kasuga et al., 1999). To generate the glucocorticoid-inducible gene expression constructs (pTA7002:AZF1 and pTA7002:AZF2), the binary transformation plasmid pTA7002, which contains the two-component glucocorticoid system (Aoyama and Chua, 1997), was digested with SpeI. The XbaI fragments of the AZF1 and AZF2 coding regions were then cloned into the SpeI sites of the pTA7002 vectors. All of the PCR products that were cloned into the vectors were sequenced to confirm the absence of PCR errors. These constructs were used to transform Arabidopsis plants using the vacuum infiltration method with Agrobacterium tumefaciens C58 cells (Bechtold and Pelletier, 1998). The T2 or T3 seeds were used for the subsequent experiments.

GFP Fluorescence
Two-week-old transgenic plants harboring the AZF1pro:AZF1-sGFP or AZF2pro:AZF2-sGFP construct were removed from GM agar plates and treated with liquid MS medium with or without 100 mM ABA to observe GFP fluorescence in the roots. The plants were also treated with MS medium containing 150 mM NaCl or 100 mM ABA to observe GFP fluorescence in the leaves. To visualize the nuclei, the plants were incubated with 10 mg mL 21 DAPI for 15 min. The subcellular localization of GFP and DAPI was then visualized using a confocal laser-scanning microscope (LSM510; Zeiss).

RNA Preparation and Real-Time PCR
Total RNA was isolated from 2-week-old plants using the TRIzol reagent (Invitrogen). cDNA synthesis and real-time PCR analyses were performed as described previously (Qin et al., 2008). For each overexpressing transgenic line, at least two independent lines were analyzed, which represented biological replicates. For the T-DNA insertion lines, representative results from several independent samples were obtained. The primers used for the qRT-PCR are listed in Supplemental Table S2. To obtain quantitative data, three replicates were analyzed to characterize the expression data for each gene.

High-Salt Sensitivity of Transgenic Plants
For salt stress, 9-to 10-d-old transgenic plants grown on GM agar plates were transferred onto 0.8% agar plates (0.53 MS medium) with or without the addition of 175 mM NaCl. The plates were maintained at 22°C under a 16-hlight/8-h-dark cycle (60 6 10 mmol photons m 22 s 21 ) until visual symptoms were observed. For the stress assays, three independent lines for each transgenic plant were analyzed. The statistical significance of the values was determined using Student's t test.

Microarray Analysis
Genome-wide expression analyses were performed using the Arabidopsis 3 Oligo Microarray (Agilent Technologies), as described previously (Qin et al., 2008). Two-week-old AZF2pro:AZF2 and vector control plants were treated with water as the control or 200 mM NaCl for 5 h. Three-week-old pTA7002: AZF1, pTA7002:AZF2, and vector control plants were treated with 10 mM DEX solution for 24 h. The gene expression levels were compared between the transgenic and control plants, using two different lines for each transgenic plant. The reproducibility of the microarray analysis was assayed by including biological and technical (dye-swap) replicates in each experiment. The microarray data were analyzed statistically as described previously (Qin et al., 2008). All of the microarray data are available at the Web site http://www.ebi. ac.uk/arrayexpress/ under accession number E-MEXP-2922.

Preparation of Probes and Recombinant Proteins and Gel Mobility Shift Assays
The primers used in this analysis are listed in Supplemental Table S2. The probes for the gel mobility shift assays were prepared from the promoter regions of the SAUR20 and SAUR63 genes. DNA fragments of 280 bp were amplified by PCR from wild-type genomic DNA using XhoI or HindIII linker primers and were subcloned into the pSK vector. The DNA fragments digested from the plasmids were labeled with or without the Klenow fragment of DNA polymerase and [a-32 P]dCTP. Fragments encoding the truncated proteins containing the two canonical zinc-finger motifs of AZF1 and AZF2 were PCR amplified using BamHI-XbaI linker primers. The PCR fragments excised with BamHI and XbaI were cloned into pMAL-c2X (New England Biolabs). The resulting plasmids were transformed into Escherichia coli BL21 (DE3). The production and purification of the fusion proteins and the gel shift assays were performed according to the methods described by Sakamoto et al. (2004), with minor modifications. DNA-binding reactions were performed in 25 mM HEPES-KOH (pH 7.6), 40 mM KCl, 0.01 mM ZnCl 2 , 40 mg mL 21 poly(dIdC), 1% bovine serum albumin, and 1 mM dithiothreitol. After incubation for 20 min at the ambient temperature, the mixtures were subjected to electrophoresis on a 6% polyacrylamide gel in 0.53 Tris-borate/ EDTA at 150 V for 90 min and visualized by autoradiography.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Phylogenetic tree and alignment of the twofingered Cys2/His2-type zinc-finger proteins in Arabidopsis.
Supplemental Figure S2. Conservation of AZF/ZAT family proteins among plant species.
Supplemental Figure S3. RNA gel-blot analysis of AZF and ZAT expression in response to different treatments.
Supplemental Figure S4. Nuclear localization of the AZF and ZAT proteins.
Supplemental Figure S5. Patterns of AZF1 and AZF2 promoter-driven GUS expression in transgenic Arabidopsis plants.
Supplemental Figure S7. High-salt sensitivity of plants overexpressing AZF1 or AZF2 under the control of a DEX-inducible promoter.
Supplemental Figure S9. Expression patterns of SAUR genes in Arabidopsis seedlings.
Supplemental Table S1. Transcripts down-regulated in AZF2pro:AZF2 plants under high-salt conditions.
Supplemental Table S2. Primers used in this study.
Supplemental Experimental Procedures S1. Additional methods.