The redox-sensitive transcription factor Rap2.4a controls nuclear expression of 2-Cys peroxiredoxin A and other chloroplast antioxidant enzymes

Background The regulation of the chloroplast antioxidant capacity depends on nuclear gene expression. For the 2-Cys peroxiredoxin-A gene (2CPA) a cis-regulatory element was recently characterized, which responds to photosynthetic redox signals. Results In a yeast-one-hybrid screen for cis-regulatory binding proteins, the transcription factor Rap2.4a was isolated. Rap2.4a controls the transcript abundance of the prominent chloroplast antioxidant enzyme through binding to the CGCG core of a CE3-like element. Rap2.4a activity is regulated by dithiol/disulfide transition of regulatory cysteinyl residues and subsequent changes in the quaternary structure. The mid-point redox potential of Rap2.4a activation is -269 mV (pH 7.0). Conclusion The redox sensitivity of Rap2.4a establishes an efficient switch mechanism for redox control of nuclear gene activity of chloroplast antioxidants, in which Rap2.4 is a redox-sensor and a transducer of redox information.


Background
In photosynthesis, excess excitation energy supports formation of reactive oxygen species (ROS) [1] which can damage metabolites, enzymes and structures [2]. Antioxidant enzymes detoxify ROS, dissipate excess energy and regenerate the electron acceptors NADP + and thioredoxin. As part of the acclimation to unfavourable growth conditions, expression of antioxidant enzymes increases under moderate stress conditions [3]. Under severe stress conditions gene expression decreases [4]. In addition, various antioxidant enzymes, such as ascorbate peroxidases (APx) [5] and peroxiredoxins [6], are inactivated. Accumulating ROS decrease the photosynthetic activity [1] and activate cytosolic defence mechanisms [1,7,8].
In CuZn-superoxide dismutase (Csd) knock-down lines of Arabidopsis, photooxidative stress alters strongest the expression pattern of chloroplast proteins [9]. Consistently, in 2-Cys peroxiredoxin (2-CP) antisense lines the imbalance in the chloroplast redox poise induces expres-sion of chloroplast APx and monodehydroascorbate reductase [10]. In planta analysis of 2CPA promoter regulation [11] demonstrated that nuclear transcription of chloroplast antioxidant enzymes responds to chloroplast signals. The redox state of the plastoquinone pool [12], the redox state of low molecular weight antioxidants [13], the acceptor availability at photosystem I [4,11] and ROS [7] have been postulated to signal the chloroplast redox poise. Signal transduction through ROS, oxylipins, protochlorophyllides, metabolic coupling by carbohydrates and the redox poise of NAD(P) + /NAD(P)H, MAPK cascades and ABA have been indicated [14][15][16]. Presently, signal transduction is under intensive investigation [16,17]. First results demonstrate the signalling function of Mg 2+protoporphyrins and the ABA-triggered transcription factor ABI4 in correlation of nuclear gene expression with chloroplast development upon greening [14][15][16]. However, the precise molecular mechanisms regulating nuclear expression of chloroplast antioxidant enzymes in green tissues in a redox-dependent manner are still elusive.
Mutants screened for low expression of the nuclear encoded chloroplast 2CPA (rimb-mutants) differentiated transcriptional regulation of chloroplast antioxidant enzymes from typical responses to ROS accumulation, such as the induction of lipoxygenase-2 (Lox2), ascorbate peroxidase-2 (Apx2), BAP1 and Fer1 [18] [Heiber et al., unpublished data]. Many genes for cytosolic antioxidant enzymes, such as Apx2, are gradually induced to very high levels. The expression intensity correlates with the availability of the regulating transcription factor [8]. In contrast, expression of most chloroplast antioxidant enzymes is induced up to a certain stress level, but decreased in response to severe oxidative stress conditions, such as application of high concentrations of H 2 O 2 [4,9,19] [Heiber et al., unpublished data], it is hypothesized that either a plus-minus regulator or interacting antagonistic signalling pathways control gene expression.
In respect of transcriptional regulation, the 2CPA is one of the best studied genes encoding a chloroplast antioxidant enzyme. Transcription is strongest in young developing tissues [20]. On top of the developmental regulation, the transcription intensity correlates with the acceptor availability at photosystem-I (PS-I) [11], which defines the reduction states of NADP + and thioredoxins [21]. In planta promoter analysis demonstrated that photosynthetic redox signal is sensed in a distinct promoter region. The target motif is located upstream of the 314 bp core promoter, that correlates 2CPA expression with chloroplast development [11]. Various nuclear encoded chloroplast proteins are co-regulated with 2CPA [18]. Piippo et al. [4] postulated that the reducing site of PS-I actually is a major signal initiation point in chloroplast-to-nucleus signaling.
A 216 bp redox-sensitive cis-regulatory region has been identified in the 2CPA promoter. It responds to the chloroplast redox signals [11]. Since sequence analysis gave no indication for interaction with a known redox-regulated transcription factor, a yeast-one-hybrid screen was performed to identify cis-regulatory proteins. Here isolation and characterization of the transcription factor Rap2.4a is presented. Rap2.4a is redox-sensitive, binds to a CE3-like element in the redox-sensitive promoter region and regulates transcription of 2CPA. Analysis of Rap2.4a-KO lines demonstrated that the transcription factor also impacts on expression of other nuclear encoded chloroplast antioxidant enzymes and protects plants against mild stresses, such as fluctuating environmental light conditions.

Isolation of Rap2.4a
To identify cis-regulatory proteins involved in transcriptional regulation of 2CPA, a yeast-one-hybrid (Y1H) screen was performed with the 216 bp redox-active 2CPA promoter domain [11] and its flanking regions. The bait element was cloned into the vector pONE-1 upstream of the Gal1,10 minimal promoter and the HIS3-cDNA. The vector was transformed into the yeast strain Y187. Preys were provided by co-transformation of Y187 with a cDNA library derived from an Arabidopsis thaliana cell suspension culture [22]. Strongest interaction with the 2CPA promoter DNA was observed with pAct2-clone1, which encodes a fusion protein of the Gal4-activation domain (AD) and the AP2-type transcription factor Rap2.4a (At1g36060; type Ib-ERF) [23] spaced by 11 amino acids (AD-Rap2.4a).
The interaction of the AD-Rap2.4a-fusion protein with the bait was confirmed on media supplemented with 40 mM 3-amino-1,2,4-triazol (3AT), which is an inhibitor of Hisbiosynthesis. To exclude epigenetic regulation, the yeast strain Y187 was retransformed with E. coli-amplified prey and bait constructs. Survival on 40 mM 3AT confirmed the strong interaction between the transcription factor and the target element.

Localization of the Rap2.4a-binding site in the 2CPA promoter
With five overlapping DNA-fragments (F1, F2, F3, F4 and F5) covering the Y1H-bait, the binding site of Rap2.4a was mapped to the 13 bp overlap of F4 and F5 by EMSA under non-reducing conditions (Fig. 1A). The Rap2.4a target sequence was confirmed with a synthetic double-stranded 13 bp oligonucleotide (Fig. 1B). Heterologously expressed Rap2.6 (At1g43160) (Fig. 1A), which shares 80 % sequence identity with Rap2.4a in the DNA-binding site (Fig. 1C), and control lysates of E. coli, which were transformed with an empty expression vector (data not shown), did not shift any 2CPA promoter fragment. Alter-native to monitoring the gel shifts by immunodetection of His-tagged proteins, the interaction between the bait element and Rap2.4a was analysed by detection of biotinylated PCR-products using horseradish peroxidisecoupled streptavidin (data not shown). Here, immunodetection of the proteins was chosen as routine method, since both methods showed the interaction of DNA and proteins, but immunodetection of His-tags turned out to be easier and more efficient to apply.
Pattern analysis by MatInspector [24] predicted a coupling element 3 (CE3)-like motif (CACGCGATTC) in the 13 bp target sequence. The motif deviates in the two bases following the CGCG-core from the typical CE3-element [25] (ACGCGTGTC). Replacing the CGCG-core by TTGT abolished binding of Rap2.4a to double-stranded 20 bp oligonucleotides (data not shown), like single nucleotide substitutions of C 3 and G 4 did (Fig. 1D). It is concluded that C 3 and G 4 of the CGCG-core are essential for Rap2.4a binding.
In vitro characterisation of DNA binding of recombinant Rap2.4a to the redox box of the 2CPA promoter The 2CPA promoter region used in the Y1H-screen was amplified into 5 fragments by PCR (F1 -F5). Electrophoretic mobility shift assay (EMSA) was performed with 2.5 μg heterologously expressed His-tagged Rap2.4a or Rap2.6. The proteins were detected with anti-His antibody on Western-blots. (B) EMSA with a synthetic double-stranded oligonucleotide corresponding to 13 bp overlap of the fragments F4 and F5 (C) Similarity between AP-domain of Rap2.4a and other AP2-transcription factors according to PHYLIP. The maximal sequence variation is 22 %. (D) EMSA with the wild-type CE3-like element, its mutagenised variants MutA -MutE and an ABRE with His-tagged Rap2.4a followed by immunodetection with anti-His-antibody.

In vivo function of C 3 G 4 in Rap2.4a-regulation of 2CPA transcription
The regulatory function of Rap2.4a on the 2CPA promoter was tested by transient Rap2.4a over-expression (35S:Rap2.4a) in Arabidopsis mesophyll protoplasts which were transfected with 2CPA wt :YFP. Standardized on co-transfected 35S:CFP, 16 h Rap2.4a over-expression resulted in ca. 5-fold higher YFP activity than in an empty vector control ( Fig. 2A) demonstrating that Rap2.4a is an activating transcription factor.
To test the in vivo function of C 3 G 4 of the CE3-like element on Rap2.4a activation of 2CPA transcription, Arabidopsis mesophyll protoplasts were transfected either with reporter constructs expressing YFP under control of the wild-type promoter (2CPA wt :YFP) or a mutagenized pro-moter (2CPA mut :YFP), in which TT substituted for C 3 G 4 (Fig. 2B). After normalization on the expression of cotransfected 35S:CFP reference constructs the YFP/CFP expression ratio of the TT-variant (2CPA mut :YFP) was decreased by 34 % compared to 2CPA wt :YFP after 16 h incubation (Fig. 2B) confirming the regulatory function of the two nucleotides in stabilization of the interaction between the transcription factor and the promoter, however it did not fully omit 2CPA promoter activation.

Redox regulation of Rap2.4a-dependent 2CPA transcription
Luciferase reporter elements have a lower stability and higher sensitivity and time-resolution than fluorescence proteins [29]. Hence, redox regulation of the 2CPA promoter was studied in a transgenic 2CPA:luciferase line [11]. The luciferase activity was 1.   (Fig. 4 left). After application of mild H 2 O 2 concentrations (Fig. 4 right) only high molecular weight bands were detected. Treatment with 5 mM H 2 O 2 resulted in a complete loss of Rap2.4a signals on the Western-Blots indicating that either high molecular mass complexes were formed, which did not migrate into the gel any more or that large protein aggregates were formed that there were beyond the transfer limits of Western-blotting. To test whether aggregates were formed and whether they were reversible, the samples were treated with 5 mM of the reductant β-mercaptoethanol minutes after the incubation with 5 mM H 2 O 2 . With β-mercaptoethanol, high amounts of monomeric Rap2.4a were detected demonstrating that H 2 O 2 by its own formed reversible high molecular weight complexes (Fig. 4 right).
Cysteine residues are typical targets for redox regulation of proteins. Rap2.4a contains 3 cysteine residues at the positions 113, 286 and 302, which are located outside of the DNA-binding motif (aa 141 -209) in the N-and C-terminal domains. They are not conserved in the Ib-ERF-subfamily of AP2-family of transcription factors and specific to Rap2.4a (data not shown). By equilibrium-based redox titration the midpoint redox potential of -269 mV was determined for the transition from Rap2.4a monomers to dimers (Fig. 4B). The relative amount of oligomers started to increase at redox potentials higher than -262 mV (Fig.  4B) consistent with the observation that oligomerization occurred at oxidizing conditions (Fig. 4A). Due to the difficulties to blot the aggregates quantitatively (see Fig. 4A) and the high number of different aggregates formed (Fig.   4A), it is impossible to determine the precise redox poise for the induction of aggregation. From the observations presented in Fig. 4A and 4B, it is suggested that it is a gradual multi-step process promoted by highly oxidizing conditions.

Effect of Rap2.4a deletion on gene expression using T-DNA insertion lines
Rap2.4a is expressed in roots and shoots (Fig. 5A left). Like the 2CPA transcript amount, Rap2.4a mRNA levels decreased in leaves upon application of 50 mM ascorbate (Fig. 5A middle) and sugars (Fig. 5A right).
The in planta function of Rap2.4a in 2CPA expression was assessed in T-DNA insertion lines of Arabidopsis thaliana. The Rap2.4a gene is interrupted upstream of the AP2-type DNA-binding site (Salk_091212; Rap2.4a-KO). Consequently, Rap2.4a-KO-lines lacked Rap2.4a mRNA (Fig.  5B). From the F2 progeny of the backcross and from the T2 progeny of the Salk-line several independent homozygous lines were selected for analysis.
The transcript levels of selected genes known to be induced by ROS, e.g. the transcription factor ZAT10 (At1g27730) and the cytosolic ascorbate peroxidase Apx2 (At3g09640) were slightly increased in Rap2.4a-KO, like the transcript levels of plastocyanin (PetE: At1g20340) and the ethylene-inducible DREBP-analogous Rap2.4b (At1g78080; Lin et al., 2007). The transcript amount of three other Ib-ERF transcription factors including the chloroplast targeted Ib-ERF Rap2.4c (At2g22200) [28], and that of chloroplast GR were only by 2 -19 % decreased compared to wild-type plants.

Effect of Rap2.4a disruption on environmental stability
After adaptation to controlled environmental conditions, the Rap2.4a-KO T-DNA-insertion lines showed only subtile phenotypes: The chlorophyll levels were slightly decreased and the leaf blades were 8 % larger (Fig. 6A and  6B). Three independent sets of 20 plants were pre-culti- Relative luciferase activity (mean ± SD) in mesophyll protoplasts from transgenic Arabidopsis expressing luciferase under control of the 2CPA promoter, 2 h after transfection with either 35S:Rap2.4a or 35S:CFP, and 1.5 h after application of 1 mM H 2 O 2 , DTT and ascorbate, respectively. Significance of difference (P > 0.01; Student's T-Test; n = 72) is indicated by asterisks. Quaternary structure regulation of Rap2.4a under oxidizing and reducing conditions     C vated for 6 weeks under controlled conditions (10 h continuously 100 μmol m -2 s -1 ). Afterwards, they were transferred to the greenhouse and exposed to naturally fluctuating light conditions. After two changes between two cloudy (maximum light intensity: 80 μmol quanta m -2 s -1 ; 14 h light) and two sunny days (maximum light intensity: 500 μmol quanta m -2 s -1 ; 14 h light), Rap2.4a-KO lines gradually developed stress phenotypes, such as severe chlorosis, stunted, thicker and less branched inflorescences and increased leaf blade areas (Fig. 6C). In average the chlorophyll contents were decreased by 27 ± 15 % (n = 60) compared to wild-type plants and the leaf blade area increased to 183 ± 63 % (n = 60). In total in 47 % of the Rap2.4a-KO plants the chlorophyll content was at least decreased by 50 %. In 61 % of the plants the leaf blade area of the largest 5 leaves was at least twice the size of the largest leaves of wild-type-plants grown in parallel (data not shown).

Discussion
2CPA transcription is redox-regulated on top of a strong developmental regulation, which correlates with chloroplast development and greening [4,11,20,30]. Redox-regulation is a fine-tuning mechanism which coordinates nuclear expression of the chloroplast protein with the actual environmental parameters [4,11,20,30]. In a onehybrid screen for cis-regulatory proteins binding to the redox-sensitive promoter element of the 2CPA gene, the transcription factor Rap2.4a (At1g36060) was isolated. Rap2.4a binds to a CE3-like element in a redox-dependent manner ( Fig. 1; Tab. 2) and activates 2CPA expression under control and slightly oxidizing conditions (Tab. 1).
Although the CE3-like element in the 2CPA promoter differs from ABRE only in one base (ACGC vs. ACGT), Rap2.4a specifically bound to the CE3-like element.
Phenotype of Rap2.4a-KO and wild-type plants

wild-type
Rap2.4a-KO Rap2.4a-KO wild-type A B C According to the nomenclature by Nakano et al. [23], who recently grouped 122 AP2-type transcription factors into 12 clades, the isolated transcription factor Rap2.4a is one of eight class-Ib-ERF proteins with conserved AP2domains, but highly variant N-and C-termini [23]. Two class-Ib-ERFs have so far been partially characterized: Rap2.4b (At1g78080) complements DREBP and blocks ethylene signalling if it is overexpressed in Arabidopsis [27]. Rap2.4c (At2g22200) is post-translationally targeted to chloroplasts where it may take over a specific, so far unknown function [28].
Rap2.4a binding to DNA is redox regulated. It is omitted under strongly reducing and strongly oxidizing conditions (Tab. 2). From in vitro analysis it has to be expected that the protein monomerizes or oligomerize, respectively, under these conditions. In between, under control and slightly oxidizing conditions, the transcription factor is in its dimeric state (Fig. 4). Since Rap2.4a activated the gene expression under these conditions (Tab. 1, Fig. 2), it is concluded that dimeric Rap2.4a stimulates promoter acitivity.
Redox regulation of proteins is often maintained though thiol-disulfide regulation. Within the Rap2.4 family of transcription factors Rap2.4a is characterized by a distinct signature of cysteinyl residues. While one cysteinyl residue, which is conserved in other group members, is missing, the cysteine residue at position 113, 286 and 302 are specific for Rap2.4a. Redox-dependent oligomerization may indicate intermolecular disulfide formation and/or structural changes that foster aggregation in hetero-or homo-complexes and finally inactivation (Fig. 4, Tab. 2).
In vivo and in vitro gene expression analysis ( Fig. 2 and 5; Tab. 1 and 2) demonstrated that Rap2.4a confers redox responsiveness to the 2CPA promoter by redox-dependent binding and activation ( Fig. 4 and Tab. 1 and 2). In addition, Rap2.4a availability impacts on the expression of various other nuclear encoded chloroplast proteins involved in adaptation of plants to environmental variation. Increased transcript levels of ROS-regulated ZAT10 [8] and stress-induced Rap2.4b [27] demonstrate that Rap2.4a function antagonizes activation of secondary signalling cascades which are activated at higher stress thresholds. Because no homologous Rap2.4a binding sites could have been identified in the promoters of coregulated genes (data not shown), it is assumed that the other antioxidant enzymes are indirectly co-regulated by a so far unknown mechanism. From comparison of Rap2.4a-KO lines with 2CPA antisense lines, cross talk by the availability of 2CPA mRNA or protein can be excluded. It is more likely that Rap2.4a triggers secondary transcription factors, which in turn activate the other nuclear genes for chloroplast antioxidant enzymes.
The midpoint redox potential of the activating transition from Rap2.4a monomer to dimer is -269 mV at pH 7.0 (Fig. 4B). It is more negative than the midpoint potential of glutathione (E hc = -230 mV), but less than that of most thioredoxins (-290 to -300 mV) [31,32]. Moderate oxidation of the glutathione pool, such as caused by photosynthetic activity and propagated by thioredoxins and the redox poise of the NAD(P) + /NADPH and the glutathione systems [15], may be sufficient to activate Rap2.4adependent gene expression. Stronger redox imbalances would inactivate Rap2.4a by aggregate formation (Fig. 4; Tab. 2) and consequently support accumulation of ROS (Fig. 5, 6, 7). While over-expression of Rap2.4a promotes 2CPA expression under control and mildly oxidizing conditions, under reducing conditions it does not impact on 2CPA expression (Tab. 1). It is concluded that Rap2.4a only confers its activating potential in its dimerized state (Fig. 7).
Rap2.4a impacts on nuclear expression of chloroplast proteins ranging from antioxidant enzymes to light-harvesting proteins (Fig. 5). The redox state of the plastoquinone pool, metabolic redox signals and ROS have been postulated to be signals indicating progressing deviation from metabolic equilibrium by increased photosynthetic electron pressure [12]. In Rap2.4-KO the transcript level of PetE, which responds to the redox state of the plastoquinone pool [12], was increased indicating a higher reduction state of the intersystem electron transport chain. Slightly increased transcript levels of the ROS-marker genes ZAT10, which encodes a strongly inducible transcription factor activating expression of cytosolic Apx [8], and its target gene Apx2, which encodes a ROS-inducible cytosolic antioxidant enzyme [33], in the Rap2.4-KO-line (Fig. 5) indicate that Rap2.4a antagonizes ROS-signalling in wild-type plants and differentiates Rap2.4a-dependent gene expression regulation from regulation of cytosolic antioxidant defence mechanisms. This observation is consistent with the analysis of the rimb-mutants, which were screened for decreased activation of the 2CPA promoter [18], suggesting that the regulation of genes encoding chloroplast antioxidant enzymes is independent from regulation of genes for cytosolic antioxidants. For transmission of ROS-signals, specific transcription factors, such as ZAT10 [12], have been described. Now, characterization of Rap2.4a shows the first mechanism which explains redox regulation of nuclear expression of chloroplast antioxidant enzymes.

Conclusion
Piippo et al. [4] postulated recently that the acceptor availability at photosystem I, which regulates 2CPA transcriptional activity [11], has a stronger impact on nuclear gene expression than the redox state of the plastoquinone pool. Here, it is shown that the transcription factor Rap2.4a is a redox regulator with redox sensory function. Rap2.4a is involved in the signalling pathway triggering the 2CPA redox box [11], which responds to the acceptor availability at photosystem I [11]. The sensitivity of Rap2.4a-KO lines (Fig. 6) demonstrates that Rap2.4a regulates tolerance against natural environmental fluctuation via the control of the nuclear expression of the chloroplast antioxidant and photosynthetic systems (Fig. 5). Rap2.4a acts as a plus/minus regulator ( Fig. 4 and 7, Tab. 1 and 2). Rap2.4a is activated by moderate redox imbalances, but inactivated upon severe stress (Tab. 1 and 2) consistent with decreased transcript levels for nuclear genes encoding chloroplast antioxidant after ROS-application [4,9,34]. It is concluded that Rap2.4a controls plant riskassessment. It can switch gene expression between induction of acclimation reactions and molecular stress avoidance, such as ROS-controlled inactivation of photosynthetic electron transport.
Redox regulation of gene expression by Rap2.4a Figure 7 Redox regulation of gene expression by Rap2.4a. Nuclear expression of chloroplast antioxidant enzymes is redox-regulated by the modulation of the expression and changes in the quaternary structure of Rap2.4a. If the redox poise shifts to more oxidized values, Rap2.4a expression is increased. Rap2.4a protein dimerizes in response to the redox-shift (Fig. 4) and activates 2CPA expression. Under severe oxidative stress Rap2.4a expression is decreased and the transcription factor loses its DNA affinity and oligomerizes.

One-hybrid screen of Arabidopsis thaliana cDNA library
The redox sensitive 2CPA promoter fragment [11] was amplified by PCR using the primers AAAAACTCGAGCAT-AATAATGAA and AAAAACTCGAGGCTTCTTCACTTA and cloned into the Xho I site of pONE1 [35] to generate the bait construct pONE1-2CPA. The yeast strain Y187 carrying pONE1-2CPA was transformed with 50 Ng of an Arabidopsis thaliana suspension culture cDNA library generated in pAct2 [22], selected on SD medium [36] lacking histidine, tryptophan and leucine and supplemented with 1 mM 3-aminotriazole (3AT). The clones were re-screened on SD medium containing 5 -40 mM 3AT. Plasmids were isolated according to standard protocols [11] and transformed into E. coli TOP10 for further analysis.

RNA isolation and RT-PCR
Total RNA was isolated as described in [20] and cDNA synthesis and RT-PCR performed in the linear phase of amplification according to [11] and [18] with gene specific primers, whose specificity was controlled by sequencing of the amplification products.

Polyacrylamide gel electrophoresis (PAGE)
SDS-PAGE was performed as described in [20]. For nonreducing gels no DTT was added to the loading buffer. For native gels, the running and loading buffers were prepared without SDS and the samples were not heated prior to sample loading.

Electrophoretic mobility shift assays (EMSA)
2.5 μg recombinant proteins and 20 ng DNA were separated on 6-7% native polyacrylamide gels (dilution of Rotiphorese-30 (Roth, Germany) in 0.5X TBE [36] at 100 V until the bromphenolblue dye was migrated approximately 2/3 down the length of the gel. After 20 min incubation in 0.5 × TBE at room temperature it was electrophoretically blotted on nitrocellulose membranes (for protein detection) or Hybond-N + (Amersham Biosciences, UK) (for nucleic acid detection) using a semi-dry blotter. The His-tagged proteins and biotinylated PCRproducts were detected with anti-His-antibody (Amer-sham Biosciences, UK), Light Shift Chemiluminescence EMSA Kit (Pierce, USA) and SuperSignal West Pico Chemiluminescent Substrate (Pierce, USA), respectively, according to the suppliers' protocols.
The promoter region -294 to -721 upstream of 2CPA translation start site was PCR amplified in five fragments using the primers AAACCATGGCATGCATAAGAGTC and TCCGGGAAATCCAGG for fragment 1, AAACCAT-GGCATGCATAAGAGTC and GATGACGGAGATGATG for fragment 2, TCTCCGTCATCGAAC and GCAGAGTTTCT-GGGT for fragment 3, AAACCATGGAATACCCAGAAACT and GCGTGACCGGAGACATG for fragment 4 and CTC-CGGTCACGCGATTCAAC and CTCTCTTCACTTGGTT-TAC for fragment 5. To generate double-stranded 13-20 bp DNA fragments, matching synthetic single-stranded oligonucleotides were annealed at room-temperature.

Analysis of the DNA-binding affinity of Rap2.4a
For quantification of the binding affinity, 2 Ng recombinant protein was incubated for 10 min with 100 pmol fragment F5 in 0.5 × TBE or 0.5 × TBE supplemented with 5 mM DTT or H 2 O 2 and analysed by agarose gel electrophoresis in 0.5 × TBE for unbound F5. Free F5 was quantified in 5 parallels as described for quantification of RT-PCR products. PCR amplification (10 cycles) was performed with F5 fragments eluted from the 120 -200 bp region of the gels using the QIAquick Gel Extraction Kit (Quiagen).

Fluorometric transactivation assays and localization studies
The Rap2.4a cDNA was cloned into Age I and Eco RI sites of 35SCFP-NosT vector [38] replacing CFP. In addition, 35SCFP-NosT vector was re-ligated after Age I/Eco RI digestion to serve as empty vector control. Isolation and saturating transfection of Arabidopsis thaliana mesophyll protoplasts with equal amounts of 35S:Rap2.4a, 35S:CFP-