A Calmodulin-binding/CGCG Box DNA-binding Protein Family Involved in Multiple Signaling Pathways in Plants* □ S

We reported earlier that the tobacco early ethylene-responsive gene NtER1 encodes a calmodulin-binding protein (Yang, T., and Poovaiah, B. W. (2000) J. Biol. Chem. 275, 38467–38473). Here we demonstrate that there is one NtER1 homolog as well as five related genes in Arabidopsis . These six genes are rapidly and differentially induced by environmental signals such as temperature extremes, UVB, salt, and wounding; hormones such as ethylene and abscisic acid; and signal molecules such as methyl jasmonate, H 2 O 2 , and salicylic acid. Hence, they were designated as AtSR1–6 ( Arabidopsis thaliana signal-responsive genes). Ca 2 (cid:1) /calmodulin binds to all AtSRs, and their calmodulin-binding regions are located on a conserved basic amphiphilic (cid:2) -helical motif in the C terminus. AtSR1 targets the nucleus and specifically recognizes a novel 6-bp CGCG box (A/C/ G)CGCG(G/T/C). The multiple CGCG cis -elements are found in promoters of genes such as those involved in ethylene signaling, abscisic acid signaling, and light signal perception. The DNA-binding domain in AtSR1 is located on the N-terminal 146 bp where all AtSR1-re-lated proteins share high similarity but have no similarity to other known DNA-binding proteins. The calmodulin-binding nuclear proteins isolated from wounded leaves exhibit specific CGCG

Plants are constantly exposed to a variety of adverse environmental conditions such as temperature extremes, UV light, salt, and pathogen attacks. Thus, plants have to endure these stresses by modulating the expression of specific genes. Regulated gene expression is one of the most complex activities in cells. It involves many transcription factors that contribute to the basal transcription machinery or mediate gene regulation in response to developmental, environmental, or metabolic cues. Based on data from the Arabidopsis genome project, it was predicted that there would be more than 1709 transcription factor genes (about 6.7% of total 25,498 genes) that encode proteins with significant similarity to known classes of plant transcription factors classified by conserved DNA-binding domains. However, less than 10% of these factors have been genetically characterized (1).
Accumulating evidence indicates that Ca 2ϩ -mediated signaling is involved in the transduction of physical signals such as temperature, wind, touch, light, and gravity; oxidative signals such as those arising from pathogen attacks; and hormone signals such as ethylene, abscisic acid (ABA), 1 gibberellins, and auxin (2)(3)(4)(5)(6)(7). All these signals have been shown to trigger changes in amplitude or oscillation in cytosolic free Ca 2ϩ level. Recently, the signal-induced nuclear free calcium changes were also observed (8). Free Ca 2ϩ changes are sensed by a number of Ca 2ϩ -binding proteins that usually contain a common structural motif, the "EF-hand," a helix-loop-helix structure (9). One of the best characterized Ca 2ϩ -binding proteins is calmodulin (CaM), a highly conserved and multifunctional regulatory protein in eukaryotes. Its regulatory activities are triggered by its ability to modulate the activity of a certain set of CaM-binding proteins after binding to Ca 2ϩ , and thereby generating physiological responses to various stimuli (10 -15).
The CaM-regulated basic helix-loop-helix family of transcription factors was reported in mammals, where CaM inhibits the protein-DNA interaction by competing with the DNA-binding domain in certain proteins (16). In plants, TGA3, a member of a family of basic leucine zipper transcription factors, showed the Ca 2ϩ /CaM enhanced-binding activities to C/G box (17). However, the CaM-binding property of TGA3 was not defined. We cloned and characterized an early ethylene-responsive gene (NtER1) in tobacco that encodes for a CaM-binding protein (18). Bouche et al. (19) reported that a Brassica homolog Bn-CAMTA is a CaM-binding protein with nonspecific DNA-binding activity. They also showed that one of the Arabidopsis homologs (AtCAMTA1) encodes a CaM-binding protein with a transcription activation domain.
The Arabidopsis genome has one NtER1 homolog (AtSR1) and five related genes (AtSR2-6). Here we report that these six genes exhibit rapid and differential response to environmental stimuli such as UV, extreme temperatures, high salt concentration, and physical wounding; hormones such as ethylene and ABA; and signal elicitors such as methyl jasmonate (MJ), H 2 O 2 , and salicylic acid (SA). Furthermore, we demonstrate that calcium/CaM binds to a 23-mer peptide in all AtSRs that corresponds to the CaM-binding region of NtER1. We also show that AtSR1 targets the nucleus and has the specific binding activity to a novel DNA element (A/C/G)CGCG(G/T/C), referred to as "CGCG box."

EXPERIMENTAL PROCEDURES
Plant Materials and Treatments-Arabidopsis thaliana ecotype Columbia were grown in a 1:1 mixture of soil mix and vermiculite under a 14-h photoperiod/10-h dark at 20 -22°C in a greenhouse or growth room. The 3-week-old seedlings were subjected to various treatments. Temperature stress plants were incubated at 4 or 42°C; physical wounding leaves were crushed using blunt forceps; UVB plants were exposed to two 15-watt UVB (280 -320 nm) lamps (F15T8.UVB 15 watts; UVP Inc.) at a dose of 2 kJ/m 2 after an irradiation period of 5 min; salt stress 200 mM NaCl was applied to soil; ethylene and MJ plants were placed in 4-liter sealed jars with 100 ppm C 2 H 4 or 0.2 M MJ, some plants were treated with 50 ppm MCP for 2 h prior to ethylene treatment; ABA, H 2 O 2 , and SA plants were sprayed with 100 M ABA, 10 mM H 2 O 2 , or 400 M SA in buffer (10 mM Tris-HCl, pH 7.2) with 1% Triton X-100, the control plants were sprayed with the buffer alone. All chemicals were purchased from Sigma. All the treatments were performed at room temperature except the temperature stress treatment. After each treatment, whole plants were collected and immediately frozen in liquid nitrogen and stored at Ϫ80°C until RNA extraction.
Cloning of AtSR1 cDNA and Gene Construction-The 5Ј end 488 bp of AtSR1 was cloned by PCR from an Arabidopsis cDNA ZAPII expression library using a gene-specific primer (AtSR1-P1, Table I) and T3 primer in the vector. The largest fragment was subcloned into pCR2.1 vector (Invitrogen) and sequenced from both sides. The full coding region of AtSR1 was cloned by PCR from the library with two genespecific primers (AtSR1-P2/P3) based on the 5Ј-cloning results (forward primer) and data base predicted cDNA sequence (reverse primer). The Pfu DNA polymerase (Invitrogen) was used for PCR amplification to maintain a high fidelity of amplification. The amplified fragment was subcloned into pET101 expression vector as described by the manufacturer (Invitrogen) and was sequenced from both sides.
The templates coding the N-terminal and C-terminal deletion mutants of AtSR1 were produced by PCR amplification from the cDNA with AtSR1-specific primers cloned into the pET101 expression vector (Invitrogen). The primers (Table I) were AtSR1-P2/P1 for C-terminal deletion ⌬C-(147-1032), and AtSR1-P4/P3 for ⌬N-(1-146). ⌬C was fused to the His 6 tag in pET101 by C-terminal fusion, and ⌬N was inserted to pET101 without fusion with the His 6 tag. The nucleotide sequences of the cloned fragments derived by PCR amplification were determined from both sides.
Preparation of Recombinant Proteins and CaM Binding Assay-AtSR1 and deletion mutants were expressed in Escherichia coli strain BL21(DE3) pLysS. The recombinant AtSR1 and ⌬N were extracted and purified with CaM-Sepharose column (Amersham Biosciences) essentially as described (18), and the recombinant ⌬C was purified with HisTrap column as described by manufacturer (Amersham Biosciences). The amount of protein was estimated by the method of Bradford using a protein assay kit (Bio-Rad). The proteins were separated by SDS-PAGE, electrotransfered onto polyvinylidene difluoride membrane (Millipore), and incubated with 35 S-labeled recombinant CaM with 0.1 mM CaCl 2 or 0.5 mM EGTA as described (18). The membrane was washed with 25 mM Tris-HCl, pH 7.5, and either 0.1 mM CaCl 2 or 0.5 mM EGTA, or 0.5 mM MgCl 2 , and then exposed to x-ray film overnight.
Gel Mobility Shift Assay-The synthetic peptides were prepared using an Applied Biosystems peptide synthesizer 431A in the Laboratory of Bioanalysis and Biotechnology, Washington State University. Samples containing 240 pmol (4 g) of bovine CaM (Sigma) and differing amounts of purified synthetic peptides in 100 mM Tris-HCl, pH 7.2, and either 0.1 mM CaCl 2 or 0.5 mM EGTA in a total volume of 30 l were incubated for 1 h at room temperature. The samples were analyzed by nondenaturing PAGE as described (18).
Transient Transformation Assays with GFP Fusion Constructs-The full-length AtSR1 or ⌬C-(147-1032) or ⌬N-(1-146) were amplified by PCR amplification with Pfu DNA polymerase using the gene-specific primers listed in Table I. The primers were AtSR1-P5/6 for AtSR1, AtSR1-P5/7for ⌬C, and AtSR1-P8/6 for ⌬N. In the 5Ј of each primer, an adaptor sequence of GTCTAGCGGATCC was added to create an artificial BamHI site. The 3Ј was created in-frame fusion with the GFP reading frame. The amplified fragments were subcloned into the pBluescript KS vector, and the DNA was sequenced from both sides for verification. These plasmids were digested with BamHI and ligated with the BamHI-digested psmGFP, which has a BamHI site between cauliflower mosaic virus 35S promoter and GFP (20). psmGFP was used as a control. GFP expression was monitored by a transient assay using leaves of 3-week-old Arabidopsis seedlings. Plasmid DNA was introduced by particle bombardment using the method described by Christou (21). Seven-m gold spheres were coated with plasmid DNA and accelerated by a 7-kV discharge toward the leaves placed on agar Petri plates. After bombardment, the leaves were kept in the dark for 24 h prior to confocal microscopic examination. The images were processed using a Bio-Rad MRC 1024 confocal laser scanning system with a Nikon microscope. The leaves were directly examined on a glass slide using argon laser (488 nm) for green fluorescence.
Preparation of Nuclear Proteins-Nuclear protein extracts were prepared from 3-week-old Arabidopsis plants after the wounding treatment for 4 h or control plants that grew in normal conditions. Nuclear proteins were extracted from harvested samples (30 g) following protocol described by Green et al. (22). The nuclear proteins were further purified with CaM-Sepharose column (Amersham Biosciences) according to Yang and Poovaiah (23).
Gel Retardation Assays-The oligo selection procedure was performed as described by Wang et al. (24). Briefly, a pool of doublestranded random oligo molecules was labeled by primed synthesis of the random oligo OS-1 using the OS-2 primer ( Table I). The labeled probe was purified by 8% non-denaturing PAGE. After gel retardation assays, the retarded DNA was eluted and labeled by PCR amplification using the OS-3 primer ( Table I). The amplified probe was purified by electrophoresis and used as a probe for the next round of selection with gel retardation assays. All other probes were labeled by primed synthesis of the synthesized oligos (containing an OS-2 complementary adaptor sequence in 3Ј) using the OS-2 primer. The labeled probes were purified by electrophoresis.
DNA binding assays were performed in a 20-l reaction mixture containing 25 mM HEPES/KOH, pH 7.5, 40 mM KCl, 0.1 mM EDTA, 10% glycerol, 1 mM dithiothreitol, 5 g/ml antipain, and 5 g/ml leupeptin, 5 g of poly(dI-dC), 3 g of recombinant proteins or 20 g of nuclear proteins or 2 g of CaM-binding nuclear proteins and 1-2 ng of labeled double-stranded DNA fragments (24). DNA-protein complexes were allowed to form at room temperature for 20 min and were separated on a 10% PAGE gel in 0.5ϫ TBE at 4°C.
RNA Extraction and RT-PCR-RNA isolation was performed as described previously by Yang and Poovaiah (18). RT-PCR analysis was performed using gene-specific primers, which were designed from the least conserved regions in the central portion of each of the AtSR genes except AtSR1. Instead, the gene-specific primers for AtSR1 were designed from the 5Ј-region where the differences were observed between this study and the GenBank TM prediction. The forward/reverse primers are as follows: AtSR1-A/B, AtSR2-A/B, AtSR-3A/B, AtSR4-A/B, AtSR5-A/B, and AtSR6-A/B. The actin 8 gene (AtACT8) was used as a positive internal control. The PCR primers for detection of AtACT8 mRNAs were AtACT8-A/B. All primers are listed in Table I. Two g of total RNA were treated with 1 unit of RNase-free DNase (Invitrogen) for 10 min at 37°C followed by 5 min at 90°C to inactivate the DNase. The reverse transcription was carried out using 0.5 g of oligo(dT) 15 as a primer in a 20-l reaction mixture as described by the manufacturer (Invitrogen). The PCR was performed in a 25-l reaction mixture containing 1 l of reverse-transcribed cDNA as the template, two genespecific primers (0.5 M each), and 1.5 mM MgCl 2 . To maintain the amplification of the internal control and AtSRs within the exponential phase, the number of PCR cycles was adjusted to 25 cycles for AtACT8 and 34 cycles for all AtSR genes or otherwise indicated. The PCR products for each pair of primers were subcloned into pCR2.1 (Invitrogen) and sequenced first time to confirm the specificity of PCR amplification. The experiments were repeated three times. The amplified PCR products (9 l) were electrophoresed on a 1.5% (w/v) agarose gel, stained with ethidium bromide, and scanned using an image analyzer.
Real Time PCR-The Real Time quantitative PCR was performed in the PE Biosystems GeneAmp 5700 sequence detection system using the SYBR green detection as recommended by the manufacturer. Each reaction (25 l) contained 2.5 l of the 10ϫ SYBR green buffer; 200 nM dATP, dGTP, and dCTP; 400 nM dUTP; 2 mM MgCl 2 ; 0.625 units of Amplitaq Gold DNA polymerase; 250 nM forward and reverse primers (listed in Table I), and 1 l of the cDNA from reverse transcription. AtACT8 was used as an internal control. The reactions were performed in a MicroAmp 96-well plate capped with MicroAmp optical caps. The reaction mixtures were incubated at 95°C for 5 min, and followed by 40 cycles of 15 s at 95°C, 30 s at 60°C, and 30 s at 72°C. The data generated from the SYBR green detection were analyzed as described by Schmittgen et al. (25). Briefly, we used ⌬⌬C t method, in which the end point C t is defined as the PCR cycle number that crosses an arbitrarily placed signal threshold. The fold change of the gene expression was calculated by 2 Accession Numbers-GenBank TM accession numbers for the genes are as follows: AF506697 (AtSR1), AF253511 (NtER1), AF096260 (LeER66), AF303397 (EICBP), S48041 (parsley CG-1), AU174776 (rice EST), BF278589 (cotton EST), AV835190 (barley EST), BE341351 (Sorghum EST), BE341351 (potato EST), BAA74856 (human1, KIAA0833), BAA74932 (human2, KIAA0909), and AAM10969 (Brassica BnCAMTA).

RESULTS
Structure of AtSRs-The Arabidopsis data base was searched using NtER1 amino acid sequences. One NtER1 homolog (AtSR1) and five related proteins (AtSR2-6) were found in the Arabidopsis genome, with gene identification numbers At2g22300, At5g09410, At3g16940, At5g64220, At1g67310, At4g16150, respectively. Based on the alignment of the amino acid sequences of AtSR1-6, some variations were observed in the N-terminal portions of AtSR1 and AtSR6, whereas AtSR2-5 showed higher similarity to each other (see Fig. 1 in the Supplemental Material). To verify the cDNA sequence of AtSR1, a gene-specific primer (AtSR1-P1) and T3 primer in the vector were used to clone the 5Ј region of AtSR1. The longest DNA fragment (ϳ490 bp) was subcloned, and DNA sequencing revealed that the first 54 amino acids in AtSR1 were different from the prediction in GenBank TM but showed a linear similarity with other AtSRs N-terminal regions (see Fig. 1 in the Supplemental Material). Comparison of AtSR1 cDNA sequences with the genomic sequence indicated the differences resulted from the different RNA splicing sites in introns 1 and 2 (data not shown). Further cloning of the full cDNA showed that sequences in other region were the same as predicted. Reddy et al. (26) reported the isolation of a clone (EICBP, a partial cDNA sequence was reported in GenBank TM ) which had the same cDNA sequence as the GenBank TM prediction, suggesting this gene could have two types of transcripts. AtSR1 has 3302 nucleotides, and its largest open reading frame encodes a protein composed of 1032 amino acids with a predicted molecular mass of approximately 116 kDa. The predicted AtSR1 is an acidic and hydrophilic protein (pI 5.3) with no obvious membrane-spanning domains, and an overall secondary structure of ␣-helices. AtSR1 contains several noteworthy structural features as follows: 1) N-terminal 62-79 has a typical bipartite nuclear targeting signal (27); 2) C-terminal 900 -922 has an almost identical amino acid sequence as the characterized CaM-binding region in NtER1 (18); 3) the central portion 661-726 has two ankyrin-like repeats which is a motif known to be responsible for mediating protein-protein interactions (28); 4) the C-terminal 853-896 has two IQ motifs, which is a CaM-binding motif in many proteins (29); and 5) the C-terminal 1003-1019 has an acidic domain with 11 acidic amino acids.
The amino acid sequences of AtSR6 were corrected based on the reported EST (GenBank TM accession number T04795) (see Fig. 1 in the Supplemental Material). AtSR2-6 had predicted lengths ranging from 852 (AtSR3) to 1035 amino acids (AtSR5) with pI values of 5.2-8. Overall similarity among six homologs ranges between 43 and 78%. They shared very high similarity in the N-terminal portion and in the C-terminal portions but not in the central portions (see Fig. 1 in the Supplemental Material). In the C terminus, all six AtSRs and NtER1 showed over 75% similarity and 65% identity, especially in the CaMbinding region with greater than 90% similarity and 79% identity. In the N-terminal portions, AtSR1-6 had over 66% similarity and 50% identity. Similar to AtSR1, AtSR2-5 have a predicted nuclear targeting signal sequence in the N terminus, one or two ankyrin repeat(s) in the center portion (except AtSR6), and more than two IQ motifs around the CaM-binding region.
All six AtSRs have just one copy each in the Arabidopsis genome. Five Arabidopsis chromosomes have one AtSR gene each, except chromosome 5 which has two, separated by about 20,000 kb. Phylogenetic analysis revealed that AtSR1-6 could be grouped into four subfamilies (SF); SF1 (AtSR1), SF2 (AtSR2, 4), SF3 (AtSR 3, 6), and SF4 (AtSR5) (data not shown). The overall conserved structure of all the AtSRs suggests that they may diverge from a single ancestral origin. AtSR2 and -4 share the highest similarity and are both located on chromosome 5, which suggests that they evolved by gene duplication most recently.
AtSRs Are CaM-binding Proteins- Fig. 1A shows the alignment of the CaM-binding region of AtSR1 with other AtSRs, tobacco NtER1 and tomato LeER66. The predicted CaM-binding domains (23 amino acids) corresponding to the NtER1 CaM-binding region are highly conserved. For example, AtSR1 has only two conserved amino acid sequence substitutions as compared with the counterpart of NtER1 in this portion (amino acids 900 -922). Helical wheel projection in GCG 10 (version 10 of the GCG program) revealed that AtSR1-6 had the basic amphiphilic ␣-helix structure (data not shown), a typical secondary structure for most characterized CaM-binding proteins (18,30,31). To determine that the AtSRs are CaM-binding proteins, the full-length AtSR1 and two truncated constructs were expressed in E. coli. The ⌬C-(147-1032) was fused to a His tag in the C terminus and was purified by His-Trap column chromatography. The full-length and ⌬N-(1-147) recombinant proteins were purified by a CaM-Sepharose column. The recombinant proteins were then subjected to a CaM binding assay. The results revealed that CaM binds to both AtSR1 and ⌬N, but not ⌬C, in the presence of 0.1 mM CaCl 2 (Fig. 1B). No CaM binding was observed for all proteins when 0.1 mM CaCl 2 was replaced by either 0.5 mM calcium chelator EGTA (Fig. 1B) or other divalent ions such as 0.5 mM MgCl 2 (data not shown). Therefore, CaM binding to AtSR1 was Ca 2ϩ -dependent, and the CaM-binding region was within the C-terminal residues 147-1032. Furthermore, four peptides (representing four SFs) corresponding to the putative CaM-binding domains of AtSR1,2,3,5 were synthesized. Gel mobility shift assays revealed that CaM bound to all of them in a Ca 2ϩ -dependent manner. Two examples (AtSR1 and AtSR3) are shown in Fig.  1C. These results indicated that AtSR1-6 were all Ca 2ϩ -dependent CaM-binding proteins. Because many IQ motifs were CaM-binding domains, the peptides corresponding to the two IQ motifs of AtSR1 were used for the mobility shift assay. CaM did not bind to these two peptides, either in the presence of CaCl 2 or EGTA (data not shown).
AtSR1 Targets the Nucleus-A search of GenBank TM revealed that several partial clones from both dicots and monocots had over 66% similarity and 50% identity with AtSRs N-terminal portion (AtSR1, amino acids 13-134). These plants included parsley, potato, cotton, rice, barley and sorghum. This portion also showed over 56.6% similarity and 42.4% identity with two predicted proteins, KIAA0833 and KIAA0909, based on cDNA isolated from adult human brains. Alignment of these sequences indicated that they had the similar secondary structure with several predicated ␣-helices and two ␤-sheets, as well as several positive charged amino acids (more than 10 net positive charges) in this portion. In particular, they all have conserved bipartite nuclear localization signals (AtSR1, amino acids 62-79).
We further selected AtSR1 for detailed studies on its subcellular localization by making the full-length of AtSR1, ⌬N-(1-146) and ⌬C-(147-1032) with GFP fusion constructs. Transient transformation into Arabidopsis leaves was performed by DNA bombardment, and the image was analyzed 24 h after transformation using a confocal microscope. The green fluorescence was throughout the cytoplasm for the GFP control construct. However, both AtSR1:GFP fusion and ⌬C:GFP fusion predominantly were localized to the nucleus. In contrast, the ⌬N:GFP fusion was visualized as patches in the cytoplasm (Fig. 2). Thus AtSR1 targets nuclei, and the nuclear localization signals are within N-terminal 146 amino acids. Furthermore, the fact that all other AtSRs and related proteins have conserved bipartite nuclear localization signals suggests that they all are nuclear proteins.
AtSR1 Recognizes Specific DNA Elements-Parsley CG-1 (147 amino acids) is a partial clone with high similarity with N-terminal portion of AtSRs. da Costa e Silva (32) reported that parsley CG-1 bound to a DNA fragment CGCGTTTA-ATCTCCAACAAACCCCTTCTAG in which CGCG was crucial for DNA binding. The gel retardation assay showed that neither full-length AtSR1 nor deletion mutants bound to this DNA fragment (data not shown). In order to test whether the nuclear protein AtSR1 had specific interacting DNA elements, an oligo selection procedure was used with a pool of 30 completely random sequences of oligonucleotides. Because the putative DNA-binding domain was located in the N terminus, the recombinant AtSR1⌬C-(147-1032) was used for gel retardation assays to avoid the potential negative effects of other domains on DNA binding. The poly(dI-dC) was used as a nonspecific competitor. After three rounds of selection, gel-retarded oligo DNA molecules were amplified, and then a library enriched in DNA inserts containing specific sequences recognized by ⌬C was established.
The DNA sequencing revealed that half of the positive clones had a common DNA element of 6-bp ACGCGG (or CCGCGT). However, AtSR1 also bound to other fragments with ACGCGT (30%), CCGCGG (10%), ACGCGC (or GCGCGT) (5%), CCGCGC (or GCGCGG) (5%). They share a consensus sequence CGCG in the middle. Mutations of any nucleotide of CGCG abolished the DNA binding (Fig. 3A). Nonetheless, CGCG alone is not sufficient for DNA binding. The minimum DNA-binding elements are 6-bp CGCG box, (A/C/G)CGCG(C/ G/T). Some variations were observed in 1st and 6th nucleotides of CGCG boxes. No T was observed in position 1, and no A was observed in position 6. Substitutions with T in position 1 or with A in position 6 abolished the DNA binding (Fig. 3A). Note that a faint band with a faster mobility was visualized besides a strong protein-DNA complex band with a slower mobility (Fig. 3A). One possible explanation is that AtSR1⌬C binds to the target DNA fragment with both monomer and dimer (or oligomer) but mainly with dimer (or oligomer).
To determine whether full-length AtSR1 had the same binding specificity as ⌬C, the recombinant AtSR1 and ⌬N-(1-146) were subjected to gel retardation assays. The results show that AtSR1 had a DNA binding activity to probe 1 similar to ⌬C. However, ⌬N did not interact with probe 1 (Fig. 3B), indicating that AtSR1 had only one specific DNA-binding region that was within the N-terminal 146 amino acids. To demonstrate that AtSR1 was in fact the protein present in the mobility-shifted band, the different amounts of ⌬C protein were added in reactions with the same amount of probe 1 for gel retardation assay. The DNA-protein complex intensity was correlated with the increasing protein amount. No DNA-protein complex was observed without addition of the protein, confirming the presence of ⌬C protein in the complex (Fig. 3C). The ⌬C binding activity to the CGCG box was further demonstrated in competitive gel retardation assays (Fig. 3D). Formation of the DNAprotein complexes between labeled probe 1 and ⌬C was subjected to specific competition by wild type CGCG fragments (unlabeled probe 1). However, the mutated CGCG fragments (unlabeled probe 6) were not capable of competing for binding to ⌬C (Fig. 3D). To examine the effect of CaM binding on DNA binding activity, CaCl 2 , EGTA, CaCl 2 /CaM, and EGTA/CaM was added in the reactions with full-length AtSR1 and probe 1, respectively. No obvious effect was observed on DNA binding activity by any of them, which suggested that CaM binding had no effect on DNA binding (data not shown).
Genes with CGCG cis-Elements-A search of the data base revealed that cis-acting elements ACGCGG/CCGCGT were present in the promoter regions of about 130 genes (more than two copies) in Arabidopsis genome. Some of these genes are listed in Table II. The promoter regions are assumed to be within ϳ1 kb upstream of the starting transcription site (for the known genes) or the first ATG (for the predicted genes). These genes are related to ethylene signaling (EIN3) and ABA signaling (a putative ABA responsive protein), light perception (phytochrome A, phyA), stress responsive such as the DNA repairing protein, heat shock protein, touch protein (TCH 4), and CaM-regulated ion channel. CaM genes (CaM2 and CaM3) and AtSR6 also contains CGCG cis-elements in their promoter regions.
In order to confirm that AtSR1 bound to those promoters in vitro, we selected the promoter regions of EIN3 (from Ϫ330 to Ϫ295), CaM2 (from Ϫ221 to Ϫ199), and phyA (from Ϫ162 to Ϫ104) for gel retardation assays (Fig. 3E). The results revealed that AtSR1 had the specific binding activity to these DNA fragments. However, no binding activities were detected in all CGCG mutants (Fig. 3E).
Expression Patterns of AtSRs-The expression level of AtSR1-6 is relatively low in plants. By using RT-PCR, the signal was detected around 42 cycles under normal growth conditions, whereas the exponential stage for internal control AtACT8 was detected around 25 cycles (Fig. 4). All AtSRs had higher expression in stem and flowers, as well as siliques (except AtSR6) at 6-week stages. This was consistent with the expression pattern of tobacco NtER1 and tomato LeER66. NtER1 was highly expressed in senescent leaves and flowers (18), and LeER66 had more expression during fruit ripening (33). Several AtSRs (AtSR1,3,4) also had a relatively higher expression in 4-day-and 7-day-old seedlings, which suggested that they had some role in early stages of development. How-ever, the expression patterns of AtSR1-6 differed between other tissues and developmental stages. Generally the roots exhibited a higher expression of AtSRs than leaves both at the 2-week stage and the 6-week stage, suggesting an important role for AtSRs in root growth and development. AtSR1,2,5,6 were highly expressed in roots at both the 2-week stage and the FIG. 3. AtSR1 recognizes specific DNA elements by a gel retardation assay. A, AtSR1⌬C specifically binds to a novel CGCG box. All probes are listed below. The CGCG box is underlined, and the mutated nucleotides are in lowercase letters. B, AtSR1 has the specific binding activity similar to ⌬C. ϩ indicates that the recombinant protein was added in DNA binding assay; Ϫ indicates that no protein was added. C, AtSR1 binds to DNA in a dose-dependent manner. The intensity of the DNA-protein complex correlated with increasing ⌬C protein amounts. D, competitive gel retardation assays to detect the binding activities specific to a CGCG box (probe 1). Binding reactions contained a molar excess (10and 100-fold) of competitor DNA: 1, CGCG fragment (unlabeled probe 1); 6, CGCG mutant (unlabeled probe 6); Ϫ, no competitor. E, recognition of CGCG cis-elements of phytochrome A (phyA), ethylene-insensitive gene 3 (EIN3), and calmodulin 2 (CaM2). The probes corresponding to three promoters are listed at right. The CGCG boxes are underlined. The mutated nucleotides are in lowercase letters. The dashes indicate sequences identical to the wild type. 6-week stage, whereas the AtSR4 expression was higher at the 6-week stage than the 2-week stage. AtSR5 exhibited the highest expression level in leaves, whereas AtSR6 showed very little expression at both the 2-week stage and the 6-week stage. AtSR4 was barely detected in 2-week-old leaves, but it was highly expressed in leaves at the 6-week stage.
Expressions of AtSRs Are Differentially Regulated by Multiple Signals-Tobacco NtER1 and tomato LeER66 were ethylene-responsive genes. To study the effects of ethylene on the expression of AtSRs, 3-week-old seedlings were either treated with ethylene or with 1-methylcyclopropene (MCP) for 2 h prior to ethylene treatment which is known to block ethylene action. Only AtSR1,2,5 responded to ethylene treatment. Treatment with MCP for 2 h prior to ethylene treatment blocked their induction (data not shown), indicating that AtSR1,2,5 were ethylene-responsive genes. The time course of the induction showed that all three AtSRs were induced within 15 min after treatment and reached their peaks after 30 min of treatment. Among these genes, AtSR1 was highly induced as compared with AtSR2,5. Application of MJ showed a similar effect on AtSR induction as ethylene (Fig. 5A).
To investigate whether AtSRs responded to other signals, plants were subjected to abiotic stresses such as heat, cold, high NaCl concentration, UVB, and wounding, hormone ABA, as well as the signal elicitors H 2 O 2 and SA. In these experiments, plants were exposed to stress conditions for short periods of time (4 h) to avoid possible secondary effects. These stimuli differentially induced the expression of AtSR genes. ABA, H 2 O 2 , and SA showed similar patterns by inducing the expression of AtSR3,4,5,6 with no effect on AtSR1,2. Heat shock, cold, and UVB showed a separate pattern of expression by inducing AtSR1,2,3,5,6. However, none of these treatments showed any effect on the expression of AtSR4. Salt stress also had a similar effect but did not induce the expression of AtSR6 (Fig. 5A). No amplification were observed for AtSRs in all controls (data not shown) Furthermore, the physical wounding induced the expression of all six AtSRs. The quantitative expression profiles of AtSR1-6 following wounding were studied using real time RT-PCR. The results showed that the induction level varied among genes significantly. AtSR1,2 showed the fastest and highest induction. Their expression levels were increased by over 80-fold within 2 h. However, the induction of AtSR3,4 were just around 12-fold (Fig. 5B).
CaM-binding Nuclear Proteins Bind to CGCG cis-Elements-To confirm the presence of CGCG box-binding proteins in vivo, nuclear proteins were isolated from wounded leaves. The specific DNA binding activity to the DNA fragment Ϫ330 to Ϫ295 of EIN3 promoter was detected in the total nuclear proteins from wounded plants, but the nuclear proteins from untreated plants showed almost no DNA binding activity (data not shown). No protein-DNA complex was observed when the CGCG mutant was used as a probe. These results indicated that plant leaves contained nuclear CGCG-binding proteins which were induced by wounding. These are consistent with real time RT-PCR analysis which showed that wounding induced the expression of all AtSRs. To investigate whether the CGCG box-binding proteins were CaM-binding proteins, the total nuclear extract from wounded plants was further purified by CaM-Sepharose column. Out of 110 g of nuclear proteins, about 4 g of CaM-binding nuclear proteins were purified. The fraction of CaM-binding nuclear proteins was subjected to gel retardation assay. The specific DNA binding activities were dramatically increased for CaM-binding nuclear proteins (data not shown), indicating that CGCG box-binding nuclear proteins in vivo are CaM-binding proteins. DISCUSSION Earlier we reported that an early ethylene-responsive gene NtER1 in tobacco encodes a CaM-binding protein and suggested that this type of gene exists in other plants, such as Arabidopsis and tomato (18). Later, an Arabidopsis homolog (EICBP) was identified as an ethylene-responsive gene, and calcium/CaM was shown to bind to the recombinant protein (26). Also, a Brassica homolog BnCAMTA and one of the Ara-bidopsis homologs (AtCAMTA1) were classified as CaM-binding proteins with nonspecific DNA binding activity (19). The results described herein show that the six NtER1 homologs in Arabidopsis (AtSR1-6) are responsive to a variety of stimuli (Fig. 5). Each AtSR has a conserved structural feature with a DNA-binding region (CGCG domain) in the N terminus and a CaM-binding domain in the C terminus. The detailed study of AtSR1 revealed that the CGCG domain (amino acids 1-146 in AtSR1) targets the nucleus and specifically interacts with a novel CGCG box, a 6-bp double-strand DNA element (A/C/ G)CGCG(G/T/C) (Figs. 2 and 3). The optimal target sequence for AtSR1 is an ACGCGG (or CCGCGT) element. Further comparisons revealed that the CGCG domain protein is present in other plants, both dicots and monocots. They all have conserved CGCG domains similar to AtSR1, suggesting that they all bind to the CGCG box. However, they may have different target sequence preferences on the nucleotides of positions 1 and 6 in the 6-bp DNA element. For example, the five members of AtERF proteins (Arabidopsis ethylene-responsive binding factors) have distinct DNA-binding preferences on the GCC box (34). AtERF1, -2, and -5 appear to be most sensitive to the single-nucleotide substitutions within the GCC box sequence. By contrast, AtERF3 and -4 appear to be more flexible than others with respect to their target sequence preference.
The DNA-binding domain of AtSR1 does not exhibit similarity with any previously known DNA-binding motifs. However, this CGCG domain is highly conserved among all six members of AtSRs and related proteins in other plants, even sharing some similarity with two human proteins. The structural prediction suggests that this region possesses several ␣-helices and several basic amino acids (more than 10 net positive charges), which may be good candidates to form a DNA-interaction face. Therefore, the CGCG domain with a specific recognition of the novel CGCG box may represent a novel type of DNAbinding domain present not only in plants but also in humans. However, defining this DNA-binding domain will require further structural analysis of these proteins and DNA-protein interactions.
Ethylene and ABA are two major plant hormones that affect almost all stages of plant development, such as seed germination, cell fate, fruit ripening, senescence, and abscission. They are also key regulators that mediate the response of a plant to biotic and abiotic stresses (35)(36)(37). Recently, it has been documented that MJ, H 2 O 2 , and SA play a role as defense signal molecules. The responses of plants to the environmental stresses such as heat shock, cold, UV, wounding, and high salt or pathogen and insect attacks are often mediated by these hormones and/or signal elicitors (5, 38 -44). Ca 2ϩ and Ca 2ϩbinding proteins such as CaM are important in plant defense response. The biotic and abiotic stresses trigger free calcium changes and CaM gene expression (2,3,(45)(46)(47)(48). The crosstalks between Ca 2ϩ /CaM and other signaling pathways such as oxidative signaling, auxin, ABA, ethylene action, and environmental stresses have been reported in recent years (4,5,18,23,31,44,49). However, our understanding of the complex signaling cascade is in its infancy.
In this study, we demonstrate that a CaM-regulated AtSR gene family responds differentially to multiple physical and chemical stimuli (Fig. 5). Each gene was induced by at least five different signals, and each signal triggered more than one gene expression (Fig. 5A). All of the genes were induced by physical wounding (Fig. 5B). It is well established that jasmonic acid plays a central role in plant responses to wounding. Hormones such as ABA and ethylene have also been proposed to play a role in wound signaling (43). Our results support that both jasmonic acid-dependent and -independent wound signaling pathways were rapidly activated to regulate AtSR expression following wounding. Furthermore, these genes respond to signals very rapidly (within minutes), which indicates that they are all early signal responsive genes. Early signal responsive genes are believed to play a prominent role in regulating lateresponsive genes (33,50). These results indicate that the multiple signal transduction pathways regulate the expression of the Ca 2ϩ /CaM-binding AtSR gene family, suggesting that AtSR is one of the early hubs for cross-talk among signaling pathways in plants.
Identification of the AtSR1-specific DNA-binding elements led us to investigate further the potential downstream regulatory genes. Several of these genes are involved in ethylene and ABA signaling, DNA repairs (UV damage), signal perception, and stress response (Table II), which is consistent with the AtSR gene expression profiles in response to signals (Fig. 5). For example, EIN3 is an ethylene-responsive transcription factor, that functions downstream of CTR1 and EIN2 (51). Three CGCG cis-elements are present in its promoter region. Phytochrome A is one of the best characterized photoreceptors, and the regulation of phyA expression has been intensively studied by characterizing several cis-acting elements, such as GT, PE3, and RE1 elements (52). Four CGCG cis-elements appear in Arabidopsis phyA. A CGCG cis-element was found inside the positive element (PE3) of phyA from monocots oat, rice, and maize, and another CGCG element also appeared in a negative element (RE1) in oat (52). It has been suggested that calcium/ CaM may be one of signaling intermediates in phototransduction (53). Our data provide additional clues for a role of calcium/CaM in phototransduction. Coincidentally, several lightresponsive cis-elements, such as I boxes (GATAA), CCAAT elements, and GT-1 sites (54), were found in the promoters of AtSR1,4,5,6 (data not shown). Gel retardation assays show that the recombinant AtSR1 specifically recognizes the wild type EIN3 and phyA promoter fragments but not the CGCG mutants (Fig. 3E). Furthermore, CaM-binding nuclear proteins purified from wounded plants had the specific DNA-binding activity to CGCG boxes. In contrast, DNA binding activities were barely detected for the nuclear proteins from control plants that were grown under normal conditions. Our data suggest that AtSR1 may regulate the expression of these genes in vivo. Isolation and characterization of the knockout mutants of AtSRs will provide direct evidence of downstream regulatory genes.
All AtSRs are shown to be calcium-dependent CaM-binding proteins (Fig. 1). However, the specific role of CaM-binding to AtSRs is not yet clear. It has been reported that Ca 2ϩ /CaM inhibits the protein-DNA binding by competing with the basic helix-loop-helix DNA-binding domain in certain transcription factors (16). The DNA-binding protein TAG3 also shows Ca 2ϩ / CaM enhanced-binding activities to C/G box (17). However, its CaM-binding domain has not been identified. In this study, we found that the CaM-binding region is far away from the DNAbinding region in all AtSRs. The possible role of Ca 2ϩ /CaM may be manifested in the control of interactions with other proteins or altering transcription activation of others. Recently, a transcription activation domain was mapped to a region just up- stream of ankyrin repeats in AtCAMTA1 by fusing with the DNA-binding domain of LexA transcription factor and testing in a yeast system (19). Identification of the AtSR1 recognition sequence as a CGCG box should help to define the importance of Ca 2ϩ /CaM-binding in transcription activation in plants.
In conclusion, multiple signals rapidly and differentially induce AtSR expression; AtSR-encoded proteins are Ca 2ϩ -dependent CaM-binding proteins; AtSR1 shows specific DNA binding activity to the CGCG box; and the CaM-binding nuclear proteins are able to specifically interact with the CGCG cis-elements. Based on these results and other studies, we propose that CaM-regulated AtSRs may serve as one of the early hubs in multiple signal transduction cascades by differentially responding to the multiple upstream signals. Furthermore, AtSRs recognize CGCG cis-elements and may regulate the downstream gene expression, which ultimately leads to the physiological responses to varieties of stresses (Fig. 6). Further characterization of the functional significance of AtSRs should provide a better understanding of the mechanisms of plant defense to biotic and abiotic stresses.