Pfam (http://www.sanger.ac.uk/Software/Pfam/) and SMART (http://smart.embl-heidelberg.de/) protein databases enlist more than 90 Arabidopsis (Arabidopsis thaliana) arm repeat proteins. Based on their sequence homology, these proteins can be grouped into several different subfamilies such as impotin-α, kinesin, and U-box protein families (Coates, 2003). However, the functions of the Arabidopsis and other plant arm repeat proteins have not been characterized in detail except those of ARC1 and PHOR1. ARC1 interacts with an S-locus receptor kinase of Brassica (Gu et al., 1998) and has been demonstrated to be a positive regulator of the self-incompatibility response (Stone et al., 1999). A recent study shows that ARC1 promotes ubiquitination and proteasomal degradation of compatibility factors in pistil (Stone et al., 2003). The potato arm repeat protein PHOR1, on the other hand, is involved in GA signaling (Amador et al., 2001). Antisense suppression of its expression reduces GA sensitivity and plant height, whereas its overexpression increases GA sensitivity and internode length.

The BTB (broad-complex, tramtrak, and bric-a-brac) domain is another evolutionarily conserved protein-protein interaction domain (Bardwell and Treisman, 1994; Zollman et al., 1994). The approximately 120-amino acid motif, also known as poxvirus and zinc finger (POZ) domain, was first identified in a group of poxvirus proteins and in Drosophila zinc finger proteins. Subsequently, it has been found that the BTB/POZ domain is present in 5% to 10% of zinc finger transcription factors and in some actin-binding proteins or ion channels (Aravind and Koonin, 1998; Collins et al., 2001). Arabidopsis genome contains approximately 80 BTB domain proteins. However, only three of them have been reported to date: NPH3 and RPT2, signal transducers of phototrophic response (Motchoulski and Liscum, 1999; Sakai et al., 2000), and NPR1/NIM1, a regulator of gene expression during systemic acquired response (Cao et al., 1997; Ryals et al., 1997).

Plant hormone abscisic acid (ABA) controls various aspects of plant growth and development (Finkelstein et al., 2002). It inhibits germination and postgermination growth at high concentrations, although it is necessary for normal seedling growth. It regulates seed maturation process and prevents embryos from precocious germination. During vegetative growth, ABA plays an essential role in adaptation to various abiotic stresses such as drought, high salinity, and cold (Xiong et al., 2002). Extensive genetic and biochemical studies have been done to identify the regulatory components of various aspects of ABA response. As a consequence, a large number of ABA-signaling components have been reported that include transcription factors, kinases, phosphatases, RNA-binding proteins, G-proteins, and secondary messengers (Finkelstein et al., 2002; Xiong et al., 2002).

During vegetative growth, ABA controls the expression of numerous genes associated with adaptive responses to drought and other abiotic stresses (Ramanulu and Bartels, 2002; Shinozaki et al., 2003). The ABA regulation of stress-responsive genes is largely mediated by cis-regulatory elements sharing the CACGTGGC consensus. Previously, we and others identified a small subfamily of basic Leu zipper (bZIP) class transcription factors that interact with the elements (Choi et al., 2000; Uno et al., 2000). We subsequently showed that the factors, named as ABFs (i.e. ABF1–ABF4) or AREBs (i.e. AREB1–AREB3), are involved in ABA and various abiotic stress responses (Kang et al., 2002; Kim et al., 2004). In particular, ABF2/AREB1, which will be referred to as ABF2 hereafter, regulates seedling growth rate and plays an essential role in Glc-induced developmental arrest process. Its overexpression phenotypes, such as altered ABA sensitivity and multiple stress tolerance, also suggest that it is involved in ABA and stress responses. We are interested in delineating the ABA-signaling pathway(s) that lead to the ABF-dependent ABA/stress-responsive gene expression in vegetative tissues. Toward this end, we carried out two-hybrid screens to isolate proteins that interact with ABF family members to modulate their activities. Here, we report that an arm repeat and BTB/POZ domain protein interacts with ABF2. In vivo analysis of its function revealed that the ABF2-interacting protein is a novel ABA-signaling component that regulates seed germination, seedling growth, Glc response, and ABA and/or stress responses.

Figure 1. Summary of the two-hybrid screens to isolate ABF2-interacting proteins. A, Schematic diagram of ABF2 and the fragments used in the two-hybrid screens. The regions conserved among ABF family members are shown as boxes. S and T denote Ser and Thr residues, (more ...)

The longest open reading frame of the Group 2 clones encoded a protein containing 705 amino acid residues. The open reading frame was missing the initiation codon. Database search and subsequent isolation and sequencing of the full-length cDNA revealed that the protein consists of 710 amino acid residues with a predicted molecular mass of 78 kD (Fig. 1C). The ABF2-interacting protein, ARIA, has nine copies of arm repeat in the N-terminal half, with arm 1, 8, and 9 being less well conserved. Additionally, it has a BTB/POZ domain in the C-terminal region. The gene encoding ARIA (At5g19330) is composed of 19 exons, and ARIA exhibits the highest sequence identity (59%) to another Arabidopsis arm repeat protein (At5g13060) of unknown function.

Figure 2. Expression patterns of ARIA. A, RNA gel-blot analysis. RNA was isolated from seedlings treated with 100 μm ABA, 250 mm NaCl, cold (24 h at 4°C), or dehydration (withholding from water for 2 weeks). Bottom section shows the ethidium bromide-stained (more ...)

ABF2 is a transcription factor and, as shown in Figure 2C (top section), is localized in the nucleus. We noticed that ARIA has a nuclear-localization signal near its N terminus (Fig. 1C), suggesting that it may be localized in the nucleus. To determine the intracellular localization of ARIA, transgenic plants harboring an ARIA-green fluorescent protein (GFP) fusion construct were generated, and the localization of the fusion protein was determined. Figure 2C (middle section) shows that GFP was localized in the nucleus, indicating that ARIA is nuclear localized. GFP was also detected in the periphery of cells. It appears that ARIA is localized in the cell membrane as well.

ARIA overexpression lines did not exhibit significant growth phenotypes under normal conditions except slightly (approximately 1 h) delayed germination (data not shown). However, ARIA overexpression affected ABA sensitivity during germination. ABA dose response analysis (Fig. 3A) showed that germination of 35S-ARIA transgenic seeds was more severely inhibited by ABA than wild-type seeds, especially at medium concentrations (1 and 2 μm) of ABA. Thus, ARIA overexpression enhanced ABA sensitivity during seed germination. In addition, germination of the transgenic seeds was more sensitive to mannitol, Glc, and NaCl (Fig. 3B), indicating that ARIA overexpression resulted in hypersensitive response to high osmolarity.

Figure 3. Phenotypes of 35S-ARIA plants. A, ABA dose response of germination. Seeds were cold treated for 5 d at 4°C and plated on Suc-free MS medium containing various concentrations of ABA. Germination (full emergence of radicle) was scored after 3 d. (more ...)

We also investigated the responses of 35S-ARIA seedlings to various abiotic stresses and found that they are less sensitive to high salt. For example, the survival rate of wild-type plants at 100 mm NaCl was 55%, whereas those of 35S-ARIA plants were 81% (AR40) and 72% (AR32), respectively (Fig. 3C). At 125 mm NaCl, 38% (AR40) or 36% (AR32) of the transgenic plants survived, whereas the wild-type survival rate was 11%. Thus, ARIA overexpression lines were more tolerant to high-salinity conditions.

Figure 4. Phenotypes of an aria knockout mutant. A, Schematic diagram of T-DNA insertion mutant. Top, The position of T-DNA insertion is presented. Bottom, Expression levels of ARIA in wild type (Col-0) and the aria mutant (ARK10) plants determined by RT-PCR. B, (more ...)

Germination assay (Fig. 4B) showed that the mutant seeds germinated more efficiently than wild-type seeds under normal growth conditions, although the degree of difference was not high. Postgermination growth of the aria mutant was also more efficient; i.e. aria seedlings were larger than wild-type plants, as shown in Figure 4C. They developed normally, however, and the fully-grown mutant seedlings were of similar size to the wild-type plants, indicating that the mutation affected the growth of young seedlings only. Together, the observations demonstrate that ARIA is a negative regulator of seed germination and young seedling growth.

The aria mutant also exhibited altered ABA response. ABA dose response analysis of germination (Fig. 4D) revealed that the mutant seed germination was less sensitive to ABA inhibition than wild-type seeds at high concentrations of ABA (i.e. 2 and 5 μm), indicating that their germination was partially insensitive to ABA. Similarly, primary root elongation of aria plants was less sensitive to ABA inhibition than wild-type plants at higher ABA concentrations (i.e. 2, 5, and 10 μm; Fig. 4E).

Glc inhibits the shoot development (i.e. cotyledon greening, cotyledon expansion, and true leaf formation) at high concentrations, and the inhibition process is dependent on ABA (Jang et al., 1997; Leon and Sheen, 2003). To see whether ARIA is involved in the process, we determined the Glc sensitivity of aria plants. Figure 4F shows that cotyledon greening of wild-type plants was gradually inhibited as Glc concentration in the medium increased. The aria mutant plants were also responsive to Glc in a similar manner, but the degree of inhibition was lower than that of the wild-type plants. The differential response was not observed with mannitol, i.e. it was not osmotic response (data not shown). The result demonstrates that ARIA is a necessary component for the Glc inhibition of shoot development.

Figure 5. Expression of ABA-responsive genes in 35S-ARIA and aria plants. RNA levels were determined by RT-PCR using total RNA isolated from 2-week-old seedlings. The number of amplification cycles is different for different genes and for overexpression and knockout (more ...)

We described an arm repeat protein designated as ARIA, which specifically interacts with ABF2. In animals, arm proteins are involved in a variety of cellular functions such as cell contact, signal transduction, tumor suppression, and nuclear import (Hatzfeld, 1999; Andrade et al., 2001). In plants, their functions are largely unknown except for a few genes. ARC1 and PHOR1 are regulators of self incompatibility and GA signaling, respectively, as mentioned earlier (Stone et al., 1999; Amador et al., 2001). A tobacco arm repeat protein, NtPUB4, which interacts with a receptor-like kinase, has been suggested to be a developmental regulator (Kim et al., 2003). The majority (approximately 40%) of Arabidopsis arm repeat proteins, including ARC1 and PHOR1, contain the U-box motif found in a subclass of E3 unbiquitin ligases (Coates, 2003; Mudgil et al., 2004). Many of them therefore may participate in the ubiquitin-dependent protein degradation process. ARIA, however, does not contain the U-box motif. Instead, it possesses another conserved sequence motif, BTB/POZ domain. Sequence comparison indicates that only two Arabidopsis proteins, including ARIA, have both arm repeat and BTB/POZ domain. The BTB/POZ domain is found in many transcription factors and in some actin-binding proteins in animals (Aravind and Koonin, 1998; Collins et al., 2001). Most recent studies show that some BTB domain proteins are substrate-specific adapters for the CUL-3-based E3 ubiquitin ligases (van den Heuvel, 2004). Although arm repeat and BTB domain proteins play diverse roles, the basic functions of the two motifs are to mediate protein-protein interactions. Thus, ARIA has the potential to form complexes with other proteins or to function as a scaffold.

The physiological relevance of the ABF2-ARIA interaction was supported by their similar expression patterns. Expression of both ABF2 and ARIA is induced by ABA and high salt (Fig. 2; Choi et al., 2000). Both are highly expressed in vegetative tissues (especially in lateral roots, leaf vascular tissues, and guard cells) and in reproductive organs (i.e. anthers, filaments, stigma, and abscission zone). Furthermore, both proteins are localized in the nucleus (Fig. 2), although ARIA is also found in the cell membrane or cell wall region.

Our data on the in vivo function of ARIA further support the physiological significance of the ABF2-ARIA interaction. ARIA overexpression enhanced ABA and osmolarity sensitivities at the germination stage. During subsequent seedling growth, it enhanced salt tolerance. Disruption of its expression, on the other hand, promoted germination and/or seedling growth and impaired Glc response. Several of the 35S-ARIA and aria mutant phenotypes are similar to those of 35S-ABF2 and abf2 plants. For instance, delayed germination of overexpression lines, faster germination and/or growth of mutant seedlings, salt tolerance of overexpression lines, and Glc insensitivity of knockout mutants were also observed with ABF2 (Kim et al., 2004). Furthermore, we observe altered expression of several ABF2-responsive genes in 35S-ARIA and aria plants (Fig. 5), suggesting that ARIA affects ABF2-regulated gene expression.

Our results indicate that ARIA is a positive component of ABA signaling. ABA sensitivity was enhanced by its overexpression and impaired by its knockout mutation. Germination was delayed by its overexpression and promoted by its mutation. Also, other ABA-associated processes such as osmolarity sensitivity and sugar response were positively and negatively affected by ARIA overexpression and its mutation, respectively. Two observations are worthy to be mentioned regarding the role of ARIA in ABA response. First, most of the ARIA overexpression and knockout phenotypes are relatively weak or partial (Figs. 3 and 4), although they are consistently observed. This implies that the function of ARIA might be redundant. As mentioned before, there is an arm repeat/BTB domain protein in the Arabidopsis genome, which is highly homologous to ARIA not only in the amino acid sequence but also in its gene structure (data not shown), and thus, functional redundancy between the two proteins can be speculated. Another observation is that ARIA affects only a subset of ABA-dependent processes. ABA sensitivity during germination and young seedling growth was affected by ARIA. However, other ABA-dependent processes, such as stomatal closure and abiotic stress responses other than salt tolerance, were not significantly affected by it (data not shown).

The altered expression of several ABF2-regulated genes (Fig. 5) suggests that ARIA affects the ABF2-dependent gene expression. We do not know the biochemical mechanism of ARIA function at present. However, it can be speculated that it may function as a coactivator or repressor of ABF2. In animals, the arm protein, β-catenin, has been demonstrated to be a transcriptional coactivator; i.e. it translocates into the nucleus in response to a hormone signal and forms complexes with transcription factors to activate target gene expression (Polakis, 2000). The BTB/POZ domain, on the other hand, is known to mediate transcriptional repression by recruiting transcriptional corepressors, which, in turn, recruit histone deacetylase to suppress transcription (Collins et al., 2001). The BTB/POZ domain is also involved in protein degradation (van den Heuvel, 2004). Thus, ARIA might be involved in the stability control of ABF2 or other proteins that possibly might associate with it. Since ARIA possesses two protein-protein interaction domains, another possibility is that it may function as an adaptor for ABF2 to form a protein complex.

Plasmid rescue and insert DNA analysis were carried out as described (Choi et al., 2000). Sequencing of the plasmid DNA rescued from the positive clones revealed that three of them (clones 12, 20, and 24) encoded an arm repeat protein (At5g19330) and two of them (clones 17 and 27) encoded a transcription factor. The longest arm protein clone was missing the first five amino acid residues. Full-length gene was isolated by PCR using the primer set 5′-GGATCGTCTTTTACTTTGTGAACG-3′ and 5′-CATTCAAGACCGATTGTGATCAG-3′ and 1 μg of library DNA. The PCR product, which contains the entire coding region and 5′ (208 bases) and 3′ (24 bases) additional sequences, was cloned into the Zero Blunt TOPO PCR Cloning kit (Invitrogen, Carlsbad, CA) and sequenced fully. The correctness of its nucleotide sequence was confirmed by comparing it with the genomic sequence on the Arabidopsis database.

For binding assay, GST-ARIA fusion proteins (0.5 μg) were incubated with the glutathione-sepharose 4B resins for 1 h at 4°C in a binding buffer (50 mm Tris, pH 8.0, 100 mm NaCl, 10% glycerol, 0.5% Triton X-100, and 1 mm phenylmethylsulfonyl fluoride). In vitro-translated, ³⁵S-labeled ABF2 was then added, and incubation was continued for 2 h with constant rotation. The resins were washed five times with the binding buffer and resuspended in SDS-PAGE sample buffer. The proteins were separated on 15% SDS-polyacrylamide gel and visualized by autoradiography.

To investigate ABF2 localization, the coding region of ABF2 was inserted in front of the GUS-coding region of pBI221 in frame. Onion epidermal cells were then transiently transformed with the ABF2-GUS construct by particle bombardment using PDS 1000 (Bio-Rad, Hercules CA). GUS activity was determined by X-gluc staining after 24 h at 23°C. Nuclei were visualized by 4′,6-diamidino-2-phenylindole staining and observed under a fluorescence microscope (BX51, Olympus, Tokyo).

To establish aria mutant lines, four putative ARIA knockout mutant lines were obtained from the Arabidopsis Stock Center. The stock seeds were sown and grown on soil, and seeds were harvested from individual plants. To choose T-DNA insertion lines with single integration, segregation ratio of kanamycin resistancy (Kan^R) was tested, and homozygous sublines were established from those segregating at 3:1 ratio of Kan^R:Kan^S. Genomic DNA was isolated from the sublines and the integration of T-DNA at the annotated site was confirmed by the sequencing of PCR fragments. We were able to identify one insertion line (SALK_143439) with a single T-DNA insertion at the annotated site among the four putative lines. T-DNA is inserted at −379 from the translation start site. Expression analysis by RT-PCR showed that ARIA expression is abolished in the insertion line. For phenotype analysis, two sublines (ARK5 and ARK10) were used. Same results were obtained from them, and those from AR10 are presented.

For germination test, seeds collected at the same time were plated, unless stated otherwise, on MS medium supplemented with 1% Suc and other supplements (i.e. ABA, mannitol, Glc, and NaCl), and radicle emergence was examined at various time points. For ABA dose response analysis of germination, Suc was omitted from the media.

Phenotype analyses other than germination assay were performed on MS medium supplemented with 1% Suc and also with ABA, Glc, or mannitol as specified in the figure legend. For root elongation assay, plants were grown at vertical position.

Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession number NM121938.