The redox imbalanced mutants of Arabidopsis differentiate signaling pathways for redox regulation of chloroplast antioxidant enzymes.

A network of enzymatic and nonenzymatic antioxidants protects chloroplasts from photooxidative damage. With all enzymatic components being nuclear encoded, the control of the antioxidant capacity depends on chloroplast-to-nucleus redox signaling. Using an Arabidopsis (Arabidopsis thaliana) reporter gene line expressing luciferase under control of the redox-sensitive 2-cysteine peroxiredoxin A (2CPA) promoter, six mutants with low 2CPA promoter activity were isolated, of which five mutants show limitations in redox-box regulation of the 2CPA promoter. In addition to 2CPA, the transcript levels for other chloroplast antioxidant enzymes were decreased, although a higher oxidation status of the ascorbate pool, a higher reduction state of the plastoquinone pool, and an increased oxidation status of the 2-Cys peroxiredoxin pool demonstrated photooxidative stress conditions. Greening of the mutants, chloroplast ultrastructure, steady-state photosynthesis, and the responses to the stress hormone abscisic acid were wild type like. In the rosette state, the mutants were more sensitive to low CO2 and to hydrogen peroxide. Comparison of gene expression patterns and stress sensitivity characterizes the mutants as redox imbalanced in the regulation of nuclear-encoded chloroplast antioxidant enzymes and differentiates redox signaling cascades.

Sequence analysis of plant plastomes and genomes revealed that all genes for chloroplast antioxidant enzymes and all enzymes involved in the biosynthesis and the regeneration of low M r antioxidants are nuclear encoded. Therefore, regulation of gene expression depends on chloroplast-to-nucleus signaling. Depending on the deviation from the cellular redox homeostasis different signaling pathways have been hypothesized Dietz, 1999a, 2005;Pfannschmidt et al., 2001;Apel and Hirt, 2004;Foyer and Noctor, 2005). Class 1 signaling originates from specific redox pairs in the photosynthetic electron transport chain, such as reduced and oxidized plastoquinone. Class 2 signaling depends on the redox state of stromal redox pairs, such as the thioredoxin, GSH, and Asc pools, and class 3 signaling is triggered by ROS (Dietz, 2003;Pfannschmidt, 2003). Experimentally, nuclear expression of chloroplast proteins has been shown to tentatively correlate with redox shifts in the thioredoxin, NAD(P)/NAD(P) H, plastoquinone, GSH, and Asc pools (summarized in Baier and Dietz, 2005;Foyer and Noctor, 2005) on a high background level, which is coordinated with chloroplast development and light signals (Pena-Ahumada et al., 2006).
The Arabidopsis (Arabidopsis thaliana) gene for 2-Cys Prx-A (2CPA; Baier and Dietz, 1997) serves as a model in the analysis of chloroplast-to-nucleus redox signaling (Baier et al., 2004b). It encodes a chloroplast peroxidase (Kö nig et al., 2002), which protects the photosynthetic membrane against oxidative damage (Baier and Dietz, 1999b). The 314 bp promoter core coordinates 2CPA transcription with leaf development (Baier et al., 2004b). Upstream of the core promoter, a 216 bp cis-element provides responsiveness to redox signals (Baier et al., 2004b) and regulates 2CPA expression depending on the redox state of the electron acceptors of PSI (Baier et al., 2004b). Putative signals are the redox states of ferredoxin (Fd), stromal thioredoxins, NAD(P) H, and the Asc and the GSH systems (Baier and Dietz, 2005), which are class 2 redox signals according to the classifications by Pfannschmidt (2003) and Dietz (2003).
The induction of genes encoding antioxidant enzymes upon oxidative stress conditions was observed in various studies with transgenic plants and mutants (Baier et al., 2000;Mittler et al., 2004;Vandenabeele et al., 2004). In signal transduction, chloroplast redox signals interplay with the antioxidant activity of the regulated genes. For example, antisense suppression of 2CPA expression specifically increases expression of sAPx, tAPx, and MDHAR, which counteract the loss of 2-Cys Prx function (Baier et al., 2000). However, tight linkage between enzymatic and nonenzymatic reactions (Foyer et al., 1994;Fridlyand and Scheibe, 1999) limits experimental signal differentiation. Thus, the first models on chloroplast-to-nucleus signal transduction were based on selective experimental data that need further substantiation by genetic means (Pfannschmidt, 2003;Baier and Dietz, 2005;Foyer and Noctor, 2005;Baier et al., 2006).
To address the redox signaling pathway(s) involved in the expressional control of the chloroplast antioxidant system, a mutant screen was performed in Arabidopsis and mutants were isolated based on decreased transcriptional activity of 2CPA. Low transcript levels of other genes with compensatory function indicate expressional coregulation. Here, five mutants are presented in which redox regulation of 2CPA through the 216 bp redox active promoter domain is affected, and one with decreased activation of the core promoter. It is shown that the mutational defects impact on the regulation of various genes for chloroplast antioxidant enzymes and cause higher sensitivity of the mutants to hydrogen peroxide (H 2 O 2 ) and low CO 2 .

Isolation of Mutants Affected in 2CPA Gene Expression
To isolate mutants affected in transcriptional regulation of a chloroplast antioxidant enzyme, a mutant screen was performed in transgenic Arabidopsis expressing luciferase under control of the 2CPA promoter (Baier et al., 2004b). Previous characterization of the reporter gene line (Baier et al., 2004b) demonstrated that transgene regulation copies 2CPA transcript abundance regulation. For mutagenesis, the line T19-2 was selected, which carries a single insertion of the reporter gene on the bottom arm of chromosome III close to bacterial artificial chromosome MQC12.
Seeds of the homozygous reporter gene line T19-2 were treated for 5 h with 50 mM ethyl methanesulfonate to introduce random point mutations. Twentythousand seedlings of the M2 progeny were grown aseptically in constant light. At 10 d, the seedlings were individually screened for decreased in vivo luciferase activity. At this stage of seedling development, in wild-type plants 2CPA promoter activity is strong in primary leaves and hardly detectable in cotyledons (Baier et al., 2004b). Plants with less than two-thirds of the reporter gene activity of the parental line were selected for further analysis (Fig. 1). To exclude artifacts due to lower luminescent areas or unspecific defects in chloroplast development, only plants were selected that were indistinguishable from the parental line at screening age in respect of size and color. The inheritance of the low reporter gene activity was tested in the progenies under the same conditions. In seven lines the decrease in luciferase expression was confirmed ( Fig. 1).
Scoring the luciferase activity in the F2 progeny of the backcrosses to the parental line T19-2 in 10-d-old seedlings demonstrated segregation of single recessive Mendelian loci (Supplemental Fig. S1). Strongest penetrance of the phenotype was observed in rimb1, rimb6, and rimb7 (Supplemental Fig. S1).
Because redox imbalanced (rimb) 2 and rimb3 were originally isolated from the same seed pool, they were tested for genetic independence by mapping the mutations to chromosome arms. One-thousand seedlings of the F2 population of the mutants with Arabidopsis var. Landsberg erecta (Ler) wild type were screened Figure 1. Gene expression in 10-d-old rimb mutants. The relative luciferase activity of 2CPATLuc and the transcript abundance of the endogenous 2CPA gene in 10-d-old seedlings expressing luciferase under control of the 2CPA promoter. The luciferase activity was determined in vivo 5 min after spraying the seedlings with luciferin (n 5 28-45; mean 6 SE). Significant differences to T19-2 (Student's t test; P , 0.001) are indicated by asterisks. For RT-PCR analysis the cDNA samples were standardized on actin-2 transcript amount. and rescreened for low luciferase activity and genotyped with simple sequence length polymorphic markers designed against sequence length polymorphisms between the Arabidopsis ecotypes Columbia-0 and Ler. Mapping of rimb3 to the middle section of the bottom arm of chromosome IV and rimb2 to the top of the top arm of chromosome III demonstrated genetic independence.

Trans-Activity of the Mutations
To test whether the mutations act in trans, reverse transcription (RT)-PCR analysis of 2CPA transcript amount was performed with cDNA samples standardized on actin-2 transcript amount (Fig. 1). Like the reporter gene activity, the transcript levels of the endogenous 2CPA gene were decreased in 10-d-old seedlings grown under screening conditions. Due to different locations of the reporter and the 2CPA gene on chromosome III and lack of transcript sequence similarity, it can be concluded that the mutations act on the regulatory elements of the 2CPA promoter, which the reporter gene and the endogenous gene have in common.

Photosynthetic Electron Transport in the rimb Mutants
Regulation of 2CPA transcription correlates with the photosynthetic electron pressure at the acceptor site of PSI (Baier et al., 2004b). To analyze whether the mutants are affected in photosynthetic electron transport, steady-state chlorophyll-a fluorescence patterns were determined. The effective quantum yield (F PSII 5 F V #/ F M #) reflects the efficiency by which PSII transforms excitation energy into chemical energy (Schreiber and Bilger, 1993). In 3-week-old rimb mutants, F PSII was determined throughout the day with single saturating light flashes (every 60 min, beginning 30 min after onset of light). All mutants, except the highly chlorotic rimb4, showed a wild-type-like F PSII around 0.7 (Fig. 2). In rimb4, it was reduced to 0.387 to 0.466 (Fig. 2), demonstrating a severe mutational impact on photosynthetic performance. As photosynthetic electron transport controls 2CPA promoter activity (Baier et al., 2004b), rimb4 was excluded from the further analysis.

Target Element in the 2CPA Promoter
In the 2CPA promoter, regulation of the 314 bp minimal promoter coordinates the expression intensity with chloroplast development, while redox regulation takes place on the 216 bp redox box located upstream of the minimal promoter (Baier et al., 2004b). To distinguish whether the mutants are affected in regulation of the minimal promoter or in regulation of the redox box, mesophyll protoplasts isolated from 3-week-old mutants and the parental line were transfected with equal amounts of b-glucuronidase (GUS)reporter gene plasmids encoding glucuronidase under control of either the 314 bp minimal promoter (D1) or the 530 bp promoter fragment (D3), which includes the 216 bp redox box (Baier et al., 2004b). The GUS activity was quantified after 6 h incubation at a light intensity of 50 mmol quanta m 22 s 21 (Fig. 3). For comparison, the protoplast preparation and DNA quality were standardized by setting the ratio of D3 and D1-GUS expression in T19-2 protoplasts to 100%. In rimb1, rimb2, rimb3, rimb5, and rimb7 less activation of the longer promoter fragment than the 314 bp promoter of the redox box resulted in lower D3/D1 ratios. In rimb6 the D3/D1 ratio was similar to T19-2, demonstrating that the mutational defects do not specifically affect regulation of the redox box (Fig. 3).

Abscisic Acid Responsiveness of the Mutants
In wild-type plants, the 216 bp redox-sensitive promoter fragment also mediates abscisic acid (ABA) responsiveness (Baier et al., 2004b). To test whether low 2CPA expression resulted from increased ABA responsiveness, 9-d-old seedlings grown in 50 mmol quanta m 22 s 21 were screened for luciferase activity 3 h after application of 40 mM ABA to 100 mL medium. Like in T19-2, the luciferase activity was decreased in all mutants by approximately 60% (Fig. 4). In addition, no differences between T19-2 and the mutants were observed in physiological tests for ABA responsiveness, such as germination on 0 to 1 mM ABA (Koornneef et al., 1984) and wilting (Finkelstein, 1994), with six independent seed batches (Fig. 4).

Light Regulation of 2CPA Promoter Activity
Under low light conditions of 50 mmol quanta m 22 s 21 1 h after onset of illumination, the reporter gene activity was not significantly different from that in the parental line T19-2 (Fig. 5A). Two hour illumination with 400 mmol quanta m 22 s 21 increased the relative reporter gene activity in rimb1, rimb2, and rimb7, demonstrating light intensity-dependent induction of gene expression (Fig. 5A). In seedlings adapted to 100 mmol quanta m 22 s 21 , which is in our growth chambers a light intensity yielding optimal biomass production, 5 h illumination with 400 mmol quanta m 22 s 21 decreased the reporter gene activity stronger in all mutants than in T19-2, indicating stronger responses to photosynthetic signals (Fig. 5B).

Content and Redox State of Low Molecular Weight Antioxidants
The redox state and the pool size of low M r antioxidants and cellular thiols are indicators for the cellular redox state (Foyer et al., 1994). In all mutants with low 2CPA promoter activity the Asc pool was more oxidized than in the parental line (Fig. 6B). In rimb1, rimb3, rimb5, rimb6, and rimb7, the Asc pool size was increased (Fig. 6A). The GSH pool size was decreased in rimb1 (Fig. 6C), while in rimb2, rimb3, and rimb5 trends toward slightly less GSH were observed. However, in all mutants except rimb1, the GSH pool was kept reduced, demonstrating that the shift in the cellular redox poise was almost specific to Asc, while the redox state of the major cellular thiol pool was kept widely balanced.

Content and Redox State of 2-Cys Prx
In all rimb mutants, consistent with decreased transcript abundance, the 2-Cys Prx protein level was decreased. As a marker for oxidative stress inside chloroplasts, the redox state of 2-Cys Prx was determined by one-dimensional (1-D) and two-dimensional (2-D) gel electrophoresis. Depending on the redox state and the regeneration potential by thiol-disulfide isomerases such as thioredoxins, 2-Cys Prx dimers can be covalently linked by none, one, or two disulfide bonds (Kö nig et al., 2002). For analysis of intermolecular disulfide bonds, proteins were immediately extracted in SDS-containing buffer and analyzed on 20% polyacrylamide gels. Compared to the parental line T19-2, in the rimb mutants increased relative amounts of dimers were detected, which are linked by only one Figure 4. Screening of the rimb mutants for increased ABA sensitivity. A, In vivo luciferase activity after 4 h illumination at 50 mmol quanta m 22 s 21 in presence and absence of ABA. After 10 d growth on 100 mL solid Murashige and Skoog medium supplemented with 0.5% (w/v) Suc in microtiterplates, 20 mL of 40 mM ABA were applied to half of the seedlings 1 h after onset of light (mean 6 SD; n 5 36). Significant changes (Student's t test; P , 0.05) are indicated by asterisks. B, Sensitivity of seed germination to ABA. Root emergence was quantified after 7 d on media containing 0 to 1 mM ABA, which inhibits seed germination (six independent seed batches with at least 150 seeds). C, Percentage of seedlings with wilted cotyledons 4 h after opening the petri dishes at a relative humidity of 50%. The plant material was adapted to high humidity by growth on sterile media in water saturated petri dishes (n 5 370-488 cultivated in eight batches; mean 6 SD). Figure 3. Relative activation of the redox box. The relative GUSreporter gene activity controlled by the 530 bp 2CPA promoter fragment containing the redox-sensitive promoter domain (D3) was quantified in protoplasts isolated from the parental line T19-2 and the mutants and standardized on the GUS-reporter gene activity measured under control of the 314 bp minimal promoter (D1) relative to the expression intensity in T19-2 (mean 6 SD). Significant changes (Student's t test; P , 0.05) are indicated by an asterisk.
disulfide bond (monodisulfide dimer; Fig. 7). In rimb1, rimb2, rimb3, and rimb5, the amount of noncovalently linked 2-Cys Prx (dissociated into monomers) was decreased, demonstrating a higher propensity to form intermolecular disulfide bonds, such as by insufficient reductive regeneration of the active site (Kö nig et al., 2002. In contrast, in rimb6 and rimb7 the amount of the noncovalently linked 2-Cys Prx form was increased, indicating either higher 2-Cys Prx regeneration or accumulation of sulfinic acid derivates of 2-Cys Prx protein. For differentiation between reduced and oxidized 2-Cys Prx, the proteins were separated electrophoretically according to their pI after reductive cleavage of the disulfide bonds with dithiothreitol (DTT) and freezing of the thiol groups by iodoacetamide (Sheehan, 2006). For reduced 2-Cys Prx an pI (IEP) of 6.54 was determined. In the parental line, 45% of the protein had an IEP of 6.32 and a small portion of 2-Cys Prx one of about 6.21. Oxidation of fresh protein extracts with H 2 O 2 partially oxidized the protein, resulting in IEPs of 6.21, 6.12, and 6.05.
In all rimb mutants the amount of the oxidized 2-Cys Prx form with an IEP of 6.21 was increased and that of the reduced form immobilizing at pH 6.54 was decreased. In addition, in rimb2, rimb5, rimb6, and rimb7 oxidized 2-Cys Prx forms accumulated, which were only detected in H 2 O 2 -treated protein samples of the parental line T19-2, but not in untreated T19-2 samples. Figure 6. Content and redox states of Asc and GSH in 3-week-old rimb mutants grown on soil under long-day illumination (14 h light/10 h dark; n 5 6-10; mean 6 SD). Significant differences to T19-2 (Student's t test; P , 0.05) are indicated by asterisks. Figure 5. Light regulation of 2CPA promoter activity. A, Relative luciferase activity in the rimb mutants after 2 h illumination of 10-dold seedlings adapted to 50 mmol quanta m 22 s 21 with 400 mmol quanta m 22 s 21 standardized on the luciferase activity observed in the parental line T19-2 (white bars) compared to the relative luciferase activity prior to the increased light treatment (black bars; mean 6 SD). Significant changes (Student's t test; P , 0.05) are indicated by an asterisk. B, Relative luciferase activity in the rimb mutants after 5 h illumination of 10-d-old seedlings adapted to 100 mmol quanta m 22 s 21 with 400 mmol quanta m 22 s 21 standardized on the luciferase activity observed in the parental line T19-2 (mean 6 SD). Significant changes (Student's t test; P , 0.05) are indicated by asterisks.

Growth Habit of the Mutants
Although indistinguishable at screening age, from 2 weeks onwards the mutants grew slower than the parental line under long days (14 h light). In plants older than 3 to 4 weeks, under long-day illumination, early leaf senescence (rimb1) and chlorosis (rimb3 and rimb5) developed randomly and starch (especially in rimb5 and rimb6) and anthocyanins (especially in rimb6) accumulated (Table I). The strongest phenotype was observed in rimb4, which turned highly chlorotic soon after expansion of the primary leaves. Indicating severe pleiotropic defects, including decreased photosynthetic activity (Fig. 2), rimb4 was excluded from further analysis. By contrast, when grown under shortday illumination (10 h light), only subtle phenotypic differences were observed in the other rimb mutants (data not shown).

Chloroplast Ultrastructure
Redox imbalances can affect chloroplast development, which can influence gene expression of chloroplast proteins (Rizhsky et al., 2003). Therefore, the chloroplast ultrastructure was analyzed in the rimb mutants. In young mutant leaves, which show the most pronounced differences in 2CPA transcript amount compared to the parental line, the chloroplast ultrastructure was wild type like. Small starch granules were observed in chloroplasts of rimb1 and rimb5 (data not shown). At the mature stage, transitory starch accumulated in all rimb mutants. Most starch was observed in chloroplasts of rimb5, in which one big starch granule with low electron density was formed, and in rimb6, in which several granules with high electron density were observed. In parallel to starch accumulation, thylakoid stacking was decreased by one to three layers in rimb5 and rimb6, while rimb2 and rimb3 showed slightly stronger (2-3 layers) grana stacking than T19-2 (data not shown).

Performance of the Mutants in Low CO 2
To test the performance of the mutants under stress conditions, 4-week-old plants were exposed to 50 ppm CO 2 for 8 and 16 h, respectively. After the treatment, the plants were transferred to standard growth conditions (350 ppm CO 2 ). Reduced CO 2 availability restricts ribulose-1,5-bisphosphate carboxylation and favors ribulose-1,5-bisphosphate oxygenation. Here, the low CO 2 treatment was started with the onset of light to avoid carbohydrate biosynthesis prior to activation of ribulose-1,5-bisphosphate oxygenation.
Low CO 2 supports H 2 O 2 and superoxide production by photorespiration (Ogren, 1984) and photoreduction of O 2 (Fridlyand and Scheibe, 1999). Increased ROS generation demands for extra antioxidant protection. Consistently, the expression of many antioxidant enzymes, including 2CPA, is induced in wild-type plants and protects the leaves against photooxidative damage (Baier et al., 2004b;Wormuth et al., 2006).
Two days after the 8 h CO 2 deprivation treatment, strong chlorosis was observed in rimb2 and rimb3 (Fig.  8A, left). To a lesser extent the chlorophyll content   (Fig. 8A, right) in rimb1, rimb2, rimb3, and rimb6, all leaves, which were exposed to low CO 2 in the premature state, were highly chlorotic. Ninety-six hours after the treatment (data not shown) the leaves were completely bleached, while less leaves were affected in the parental line. Leaves formed during the postfumigation period and the leaves of rimb5, which had highest starch contents (Table I), developed no visible damage.

Effect of H 2 O 2 Application on the Mutants
Because low CO 2 availability can increase H 2 O 2 production (Ogren, 1984), the mutants were tested for their sensitivity to exogenously supplied H 2 O 2 . Leaf discs of 5-week-old plants were floated for 8 h on 20 mM H 2 O 2 . F PSII was analyzed as an indicator for chloroplast damage. In the mutants, H 2 O 2 application decreased F PSII stronger than in the parental line (Fig.  9A). In addition, chlorophyll degradation was enhanced (except in rimb7; Fig. 9B), demonstrating that the mutants are more sensitive to H 2 O 2 . Rimb5, which barely responded to low CO 2 (Fig. 7), showed highest sensitivity to H 2 O 2 .
To analyze whether the higher sensitivity to H 2 O 2 might be caused by decreased catalase levels, catalase protein amounts were detected immunochemically in samples standardized on the same protein amount. In  The catalase protein levels were analyzed immunochemically by western-blot analysis of 10 mg protein extracted from mature leaves of 3-week-old plants (C) and the protein amount was calculated relative to T19-2. D, Guaiacol peroxidase activity as indicator for extracellular and cytosolic peroxidase activity in 3-week-old rimb mutants harvested 1 h after onset of light (n 5 6). Significant differences to T19-2 (Student's t test; P , 0.05; mean 6 SD) are indicated by asterisks.
the rimb mutants the catalase levels were wild type like or slightly increased (Fig. 9C). In addition, guaiacol peroxidase activity was increased in the mutants, demonstrating higher extracellular and cytosolic peroxidase activity (Fig. 9D), while 2-Cys Prx protein levels were decreased (Fig. 7).

Gene Expression in Mature Leaves
Transcript Amount Regulation Patterns in Wild-Type Arabidopsis To identify candidate genes for transcript analysis in the mutants, transcript amount data for 223 putatively redox regulated genes were selected from 249 publicly available Affimetrix Chip experiments with Arabidopsis premature and mature leaves. Pearson correlation coefficients were calculated for all combinations of selected genes, which cover all genes for Halliwell-Asada-Cycle enzymes (chloroplast superoxide dismutases, Asc peroxidases, GR, and MDHAR and DHAR), Prxs, the small electron carrier proteins thioredoxins, glutaredoxins, and cyclophilins and nuclear-encoded components of the photosynthetic electron transport chain. In addition, a set of genes for cytosolic enzymes with expressional response to the carbohydrate availability (e.g. triose phosphate isomerase and Glc-6-P dehydrogenase) and extraplastidic antioxidant enzymes (e.g. cytosolic GR and the three catalases) were included as references.
The transcript abundances of components of the Halliwell-Asada-Cycle, e.g. CuZn-SOD Csd2 (At2g28190) and tAPx (At1g77490) and sAPx (At4g08390), were negatively correlated with 2CPA transcript levels. This indicates either independent regulation or strong compensatory responses similar to gene expression regulation observed in transgenic and mutant plants with decreased activity of one specific antioxidant enzyme (Baier et al., 2000;Mittler et al., 2004;Vandenabeele et al., 2004). A distance tree drawn by equivalently weighting the correlation coefficients for all pairs of genes clustered the transcript abundance regulation of these genes into several distantly related groups (Supplemental Fig. S2). Bioinformatic comparison of promoter sequences by sequence similarity analysis by Cis-Regulatory Element Detection Online (http://mips.gsf.de/proj/regulomips) and comparison of putative transcription factor binding sites after Mat-Inspector analysis (Cartharius et al., 2005) supported the hypothesis that each gene is regulated by an individual pattern of signals and, thus, is suited for mutant-based signal transduction analysis.
The 2CPA transcript amount was decreased in all rimb mutants demonstrating persistence of the mutational defects in the rosette stage. However, consistent with a low expression intensity of 2CPA in older tissues (Baier et al., 2000), the relative differences between T19-2 and the mutants were less than at screening age ( Fig. 1). In parallel to low 2CPA expression, the transcript levels of Cyp20-3, Trx-x, PrxIIE, chloroplast MDHAR, and NADPH-MDH were decreased in all rimb mutants (highest Spearman correlation). The mRNA levels of Lhcb2.2 were unchanged and that of Fd (At1g60950), Trx-m3 (At2g15579), cytGR (At3g54660), Lox2 (At3g45140), and peroxisomal Cat2 (At4g35090) were increased. The other genes showed mutant-specific expression regulation (Fig. 10). The strongest relative decrease in transcript abundance in all rimb mutants was observed for Cyp20-3. The Pearson correlation coefficient, which quantitatively weights the similarity in transcript abundance, was highest for gECS (K corr 2CPA 5 0.841).
Comparing the mutants for transcript abundance regulation, Pearson correlation coefficients higher than 0.97 were observed between rimb1, rimb2, and rimb3, between rimb3 and rimb7, and between rimb6 and rimb7. Considering only the genes for chloroplast antioxidant enzymes the transcript abundance regulation of rimb1 and rimb2 (K corr 5 0.627) separated from that of rimb3, rimb5, and rimb7 (K corr 5 0.713-0.742), while rimb6 showed the most specific gene expression pattern.

DISCUSSION
To address pathways involved in the expressional regulation of chloroplast antioxidant enzymes, a mutant screen was performed in Arabidopsis, expressing luciferase under control of the 2CPA promoter. Six mutants were isolated with trans-active mutational defects in activation of 2CPA transcription (Fig. 1). All mutations were recessive (Supplemental Fig. S1), showed gradual differences in penetrance (Supplemental Fig.  S1), and resulted in individually different phenotypes (Table I) and expression patterns (Fig. 10).

The rimb Mutants Represent a Novel Type of Mutant
Screening for decreased expression of a chloroplast antioxidant enzyme distinguishes the rimb mutants from mutants like rax1 (Ball et al., 2004) and axl8 (Rossel et al., 2006), which express cytosolic Asc peroxidase APx2 stronger than wild type. Rax1 carries a mutation in g-glutamyl-cysteine synthase (Ball et al., 2004), limiting the first step of GSH biosynthesis (Mullineaux and Rausch, 2005). Thus, APx2 expression is induced by the decreased availability of the low M r antioxidant GSH (Ball et al., 2004). In contrast, in the rimb mutants 2CPA promoter activity was decreased despite a higher oxidation state of the Asc pool (Fig. 6) and of oxidation of 2-Cys Prx protein (Fig. 7), which is a chloroplast-spe-cific marker for oxidative stress (Baier and Dietz, 1997;Kö nig et al., 2003), demonstrating that gene expression regulation is uncoupled from redox sensing.
In knockdown lines of chloroplast CuZn superoxide dismutase (SOD-kd; Rizhsky et al., 2003) the expression of various nuclear-encoded chloroplast proteins involved in photosynthesis was decreased in response to oxidative chloroplast damage. In contrast, in the rimb mutants expression of cab genes (e.g. Lhcb2.2; Fig. 10) and chloroplast ultrastructure (data not shown) were unaffected. The quantum yields of PSII (FPSII 5 F V / F M ) were wild type like, indicating high photosynthetic electron transport efficiency and a high oxidation state of intrinsic components of the photosynthetic electron transport chain (Schreiber and Bilger, 1993;Fig. 2). Consistent with not or much less inhibited photosynthesis in the rimb mutants, unlike the SOD-kd, only a subset of nuclear-encoded chloroplast proteins was affected in gene expression regulation (Fig. 10). Five mutants, namely rimb1, rimb2, rimb3, rimb5, and rimb7, were specifically limited in the activation of the recently identified 216 bp redox-sensitive promoter domain (Baier et al., 2004b;Fig. 3). The low activation of the redox box in five rimb mutants and the limitation in regulation of the minimal promoter in rimb6 resulted in decreased transcript abundance of several, but not all tested nuclear-encoded chloroplast enzymes (Fig. 10), suggesting that both types of mutants represent novel classes of Arabidopsis mutants suited for studying chloroplast-to-nucleus signaling.

rimb6
Transcript abundance regulation of rimb6 was most distinct from that of the other rimb mutants (Fig. 10). Since the activity of the 530 bp promoter element (D3) was decreased to the same extend as the 314 bp distal promoter element (D1; Fig. 3), rimb6 is not primarily affected in redox-box regulation. Comparison with transgene regulation of the parental line ( Figs. 1 and 3) suggests regulatory defects in activation of the 314 bp promoter core, which contains R, G, and I boxes known for coordination of nuclear gene expression with chloroplast development upon greening and in dependence of light availability (Donald and Cashmore, 1990). Chloroplast-nucleus coordination in greening Arabidopsis seedlings has been intensively studied in recent years using mutant approaches. However, unlike, for example, the genomes uncoupled mutants (Mochizuki et al., 2001), cab-underexpressed1 (Li et al., 1995, and immutans (Wu et al., 1999), rimb6 was not affected in greening (Fig. 9B) and chloroplast development (data not shown), and, unlike the fluorescence mutant (Meskauskiene et al., 2001), rimb6 did not show defects in chlorophyll biosynthesis upon greening of etiolated seedlings (data not shown). rimb6, like the other rimb mutants, is also distinct from Arabidopsis mutants with altered regulation of nuclear expression of chloroplast proteins, such as the high chlorophyll fluorescence mutants, which were screened for high chlorophyll fluorescence (Meurer et al., 1996), because photosynthetic electron transport (Fig. 2) was wild type like. It is therefore concluded that rimb6 represents a novel mutant type with disturbed coordination of nuclear expression of specific chloroplast proteins, such as MDHAR, Cyp20-3, and sAPx (Fig. 10).

Five of the rimb Mutants Are Affected in Redox Regulation of Gene Expression
The five other rimb mutants were limited in activation of the promoter elements that provide responsiveness to redox signals correlating with the electron pressure at the acceptor site of PSI (Baier et al., 2004b;Fig. 3). Consistent with this conclusion, a 5-fold increase of light on top of limited light availability (Fig.  5A) activated 2CPA promoter activity by stimulating the promoter core, while increased light availability on top of the regular growth light intensity (Fig. 5B) resulted in a lower level of relative reporter gene activation in rimb1, rimb2, rimb3, rimb5, and rimb7 than in the parental line.
Previously it was shown that redox regulation of the 2CPA redox box depends on ABA (Baier et al., 2004b). ABA-dependent suppression of 2CPA promoter activity (T19-2 in Fig. 4A) suggested that in the rimb mutants low 2CPA expression may result from increased responsiveness to ABA. However, no difference in ABA responsiveness of 2CPA promoter activity (Fig. 4A) and ABA-regulated phenotypes such as wilting (Fig. 4C) and germination (Fig. 4B) were observed in any rimb mutant.
The rimb mutants showed higher sensitivity to an increase in the light intensity (Fig. 5B), to low CO 2 (Fig.  8, A and B), and to H 2 O 2 (Fig. 9A) and developed stress phenotypes with leaf age (Table I). Consistently, increased oxidation of the Asc pool (Fig. 6B) and accumulation of oxidized 2-Cys Prx (Fig. 7) mirrored redox imbalances. In rimb1 and rimb3 the portion of noncovalently linked 2-Cys Prx was decreased, demonstrating that the redox imbalances mainly impacted on reductive regeneration of the thiol residues by Figure 10. Transcript analysis in the rimb mutants by gene-specific RT-PCR. The transcript abundances of selected genes were analyzed in the rimb mutants and the parental line T19-2 by RT-PCR in cDNA samples standardized on actin-2 transcript amount. The numbers give the means of the percentage of signal intensity relative to the T19-2 in three independent experiments. For Fd and Trx-m3 very strong induction was detected, tentatively quantified and labeled with and ., respectively.
thiol-disulfide isomerase, such as thioredoxin (König et al., 2002). In contrast, in rimb6 and rimb7, determination of the dimerization status and the pIs indicated increased amounts of monomers, in which the cysteinyl residues are hyperoxidized and cannot form disulfide bonds .
In rimb5, rimb6, and rimb7 increased amounts of 2-Cys Prx forms with IEPs between 6.05 and 6.22 are consistent with accumulation of Asc, which is an alternative marker for oxidative stress (Foyer et al., 1994). Stronger oxidation of the Asc pool and increased oxidation of 2-Cys Prx, but decreased transcript levels of various chloroplast antioxidant enzymes in rimb1, rimb2, rimb3, rimb5, and rimb7 (Fig. 10), illustrates defects in redox signaling.
Pearson correlation analysis of the transcript levels of nuclear-encoded chloroplast antioxidant enzymes grouped rimb1 with rimb2 and rimb3 with rimb5 and rimb7, indicating two subtypes of mutants. Stronger decreased transcript levels of 2CPA, sAPx, tAPx, gECS, and Csd2 suggest that the penetrance of the mutational defects on transcript abundance regulation is stronger in rimb1 and rimb2. It is concluded that rimb1 and rimb2 are mutated in more central elements of the signal transduction pathway or affected in stronger regulators than rimb3, rimb5, and rimb7.

Defects in Redox-Box Regulation Do Not Affect Typical ROS Pathways
Depending on the extent of deviation from cellular homeostasis, various kinds of signaling cascades have been postulated to mediate chloroplast-to-nucleus redox signaling. These range from sensing of the redox state of the plastoquinone pool to responsiveness to ROS signals (summarized in Pfannschmidt et al., 2001;Baier and Dietz, 2005;Foyer and Noctor, 2005). To date, most attention has been given to ROS-induced signaling cascades, which are involved in responses to pathogens and photooxidative stress (Mahalingam et al., 2003;Mateo et al., 2004;Suzuki and Mittler, 2005). Results from genome array hybridizations allowed selection of marker genes for specific types of ROS signals (op den Camp et al., 2003;Suzuki and Mittler, 2005;Gadjev et al., 2006). The increased oxidation states of the Asc pool (Fig. 6) and of 2-Cys Prx protein (Fig. 7) display deviation from the cellular redox homeostasis (Fig. 6). Up-regulation of transcript levels of Cat2 and Lox2 ( Fig. 10; Desikan et al., 2001;Kiddle et al., 2003), of the singlet oxygen marker BAP1 (At3g61190) and of the superoxide/H 2 O 2 marker gene Fer1 (At5g01600; op den Camp et al., 2003; data not shown) demonstrated activation of stress-induced genes in all rimb mutants. In addition, the transcript levels of Fd (At1g60950) and thioredoxin m3 (At2g15570) were increased, which encode chloroplast redox proteins involved in distribution of electrons from the photosynthetic electron transport chain, while 2CPA, PrxIIE, Cyp20-3, MDHAR, and Trx-x transcript levels were decreased (Fig. 10).
Lox2 expression is induced, for example, by insufficient antioxidant protection by Asc (Kiddle et al., 2003) and in light regimes promoting PSI activity (Fey et al., 2005), stimulating superoxide formation in the Mehler reaction (Mehler, 1951), and resulting in oxidation of the GSH pool (Fey et al., 2005). In signal transduction, jasmonate transmits the redox signals and activates Lox2 via the NPR1-dependent signaling pathway (Spoel et al., 2003). Consistent with the conclusion on independent redox signal transduction mechanisms drawn from the transcript abundance pattern depicted in Figure 10, 2CPA transcription is insensitive to jasmonate (and salicylate) (Baier et al., 2004b).
Compared to ROS signaling in pathogenesis and under excess photosynthetic excitation energy (summarized in Mullineaux et al., 2006;Torres et al., 2006), little is known about the mechanisms controlling nuclear expression of chloroplast antioxidant enzymes upon less severe redox imbalances (Baier and Dietz, 2005). Recently, Davletova et al. (2005) reported that in response to insufficient cytosolic H 2 O 2 detoxification, the chloroplastic H 2 O 2 scavenging system collapses by oxidative inhibition of the enzyme function. However, the steady-state level of APx activity was elevated during early light stress (Davletova et al., 2005). Since the activity of sAPx and tAPx is highly sensitive to H 2 O 2 (Asada, 2000;Miyake et al., 2006), it implicates expressional induction of chloroplast APx. In the rimb mutants, the redox shift in the Asc pool (Fig. 6) and increased oxidation of 2-Cys Prx protein (Fig. 7) demonstrated intracellular redox imbalances. Like in 2-Cys Prx antisense lines, which resemble the rimb mutants in respect to (1) low 2CPA expression (Figs. 1, 8C, and 10), (2) the shift in the redox state of the Asc pool ( Fig.  6), (3) an unchanged redox state of the GSH pool (Fig.  6), and (4) wild-type-like chlorophyll and protein levels in young tissues (Table I), an induction of sAPx, MDHAR, and/or tAPx transcript levels can be expected (Baier et al., 2000). However, only in rimb7 higher transcript levels were observed for the two APx genes (Fig. 10). In parallel, this mutant showed highest 2CPA transcript levels and strongest 2-Cys Prx reduction in the rosette state (Figs. 7 and 10), demonstrating that in rimb7 the (developmental) penetrance of the mutational defects is weak.
In the other rimb mutants induction of sAPx, tAPx, and MDHAR expression was either missing or only the transcript level of one of the genes was increased (Fig. 10). As shown in the Asc biosynthetic mutant vtc1 (Kiddle et al., 2003), tissue levels of Asc correlate with sAPx transcript levels and Asc decreases tAPx transcript abundance. Comparing vtc1 with wild-type plants, and adg1 and pgr1 mutants, Wormuth et al. (2006) described recently an 86% correlation of 2CPA transcript abundance with the oxidation state of the Asc pool. The Asc level was at least slightly increased in the rimb mutants and the Asc pool more oxidized (Fig. 6). Consequently, in the rimb mutants the correlation between Asc availability and the expression of sAPx, tAPx, and 2CPA (Kiddle et al., 2003;Wormuth et al., 2006) was broken. In contrast, the correlation between the Asc pool size and redox state was maintained for Lox2 expression (Kiddle et al., 2003). It is concluded that a specific signaling pathway is affected in the rimb mutants, which is distinct from induction of Lox2, Cat2, Fd, and Trx-m3.

Regulation of Several Chloroplast Antioxidant Enzymes Is Coupled
Even stronger than the 2CPA transcript levels, the relative transcript levels of Cyp20-3 were decreased, which encodes a cyclophilin promoting 2-Cys Prx activity (Laxa et al., 2007). The relative changes in transcript abundance were subjected to Pearson correlation analysis. Only a correlation coefficient of 10.491 was observed between 2CPA and Cyp20-3 in 3-week-old rimb mutants (Fig. 10), while it was 0.960 in wild-type plants ( Supplemental Fig. S2). The difference suggests that the two genes are transcriptionally coregulated, but independently posttranscriptionally regulated.
In addition to Cyp20-3, the transcript levels of thioredoxin-x, MDHAR, PrxIIE, and tAPx were decreased in the five mutants that are affected in redox-box regulation (Figs. 3 and 10). Lack of sequence similarities excludes posttranscriptional regulation, such as control by microRNAs or transcript stabilizing proteins Sunkar et al., 2006) and, therefore, demonstrates transcriptional coregulation.
The transcript levels of most other tested nuclearencoded chloroplast antioxidant enzymes (2CPB, PrxQ, sAPx, and Csd2) and redox-active proteins (MDH, CDSP32, and NTRC) were differentially regulated in the various mutants. For 2CPB, PrxQ, sAPx, Csd2, MDH, CDSP32, and NTRC increased transcript levels in some mutants showed that the transcript levels are not exclusively dependent on the mutated signaling pathway, consistent with hypothesis on regulatory networks by transcript correlation analysis in wildtype Arabidopsis (Supplemental Figs. S2). For example, the 2CPB promoter contains strong ethylene response elements (data not shown), by which gene expression can secondarily and independently be induced (Fig. 10). Thus, comparative transcriptome analysis of the rimb mutants provides insight into the cross talk between parallel induced, compensatory signaling cascades in addition to demonstrating common links between the expressional regulation of nuclear encoded chloroplast proteins.

CONCLUSION
This report describes the isolation and biochemical characterization of novel mutants with decreased expression of 2CPA and other genes for chloroplast antioxidant enzymes. Low expression of various genes with compensatory regulation in wild-type plants ( Figs. 1 and 10; Supplemental Fig. S2) demonstrates coregulation by common signaling pathways. For rimb1, rimb2, rimb3, rimb5, and rimb7 the mutational defects limited redox activation of the 2CPA promoter. Comparison of transcript abundance regulation in rimb1 and rimb2 with rimb3, rimb5, and rimb7 showed tight coregulation of MDHAR, gECS, Cyp20-3, Trx-x, and tAPx with 2CPA than sAPx and Csd2. The novel type of mutant presented here gives genetic evidence for the independence of lately defined redox signaling pathways (e.g. Pfannschmidt et al., 2001;Baier and Dietz, 2005;Foyer and Noctor, 2005;Gadjev et al., 2006) and differentiate signaling cascades. The cosuppression of genes for various chloroplast antioxidant enzymes under oxidizing conditions and the separation of redox-box regulation from induction of marker genes for ROS signaling, such as Cat2, Lox2, BAP1, and Fer1 (Kiddle et al., 2003;op den Camp et al., 2003), represent two steps toward understanding intercompartment redox signaling.

Plant Material and Growth Conditions
Arabidopsis (Arabidopsis thaliana) wild-type plants and mutants were grown on soil under controlled conditions (long-day illumination: 14 h light [70-100 mmol quanta m 22 s 21 ] and 23°C/10 h dark and 18°C; 45% to 55% relative humidity) unless otherwise specified. Leaf discs were cut with an 8 mm cork screw from the leaf blades of mature leaves of 5-week-old plants and incubated upside up on the indicated medium in the light (100 mmol quanta m 22 s 21 ). For the CO 2 fumigation experiment, the plants were transferred to a 28 L perspex chamber and fumigated with synthetic air controlled by a Millipore Tylan RO 7030 system. For in vitro growth, seeds were surface sterilized with 80% ethanol followed by 20% (v/v) household bleach, washed at least five times with sterile water, stratified at 4°C for 2 d in the dark, and germinated on Murashige and Skoog medium (Duchefa Biochemie BV) containing 0.5% or 1% (w/v) Suc, respectively. Plants were grown in constant light at 20°C and 10 to 400 mmol quanta m 22 s 21 or in cycling light as described for soil-grown plants. Effectors were applied to the medium.

Isolation of the rimb Mutants and Genetic Analysis
Approximately 200 mg seeds of a homozygous 2CPA:Luc reporter gene line (Baier et al., 2004b) were treated with a 50 mM ethyl methanesulfonate (Sigma) for 5 h. The mutagenized M1 seeds were sown in 100 pools on soil. The harvested M2 seeds were surface sterilized and grown in microtiter plates on Murashige and Skoog medium containing 1% Suc in continuous light. After 10 d growth a total of 20,000 M2 seedlings, representing 2,000 M1 plants were screened using a Victor Luminometer (Wallac, Perkin-Elmer). Five minutes after spraying with a 1 mM luciferin (Duchefa), 0.01% (v/v) Triton X-100 solution, luciferase activity was measured for 1 s. Seedlings showing lower luminescence than the parental line were selected for further analysis. The progenies were rescreened under the same conditions for confirmation of the inheritance of the low luciferase phenotype. To show the independence of the mutants, the mutations were mapped to chromosome arms from the F2 population of crosses to Arabidopsis Ler according to Jander et al. (2002). Selection and confirmation of low luciferase lines was performed as described for isolation from the mutagenized seedling population.

Gene Expression Analysis
RNA isolation and RT-PCR was performed as described in Baier et al. (2000). The cDNA samples were standardized on actin-2 transcript amount. The transcripts of the genes of interest were amplified using gene-specific primers. The primer sequences and Arabidopsis Genome Initiative codes are given in Table I. The signal intensities were quantified from the fluorogrammes of ethidium bromide stained DNA separated on agarose gels using the GELSCAN software package (BIOSCITECH) and normalized to the transcript abundance observed in the parental line. Representative examples from three independent experiments are shown.

Protein Extraction, 2-D and 1-D Gel Electrophoresis, and Western-Blot Analysis
For 1-D western-blot analysis mature leaves were extracted in 1.44% (w/v) glycine, 0.303% (w/v) Tris, 2% (w/v) SDS either in the presence or absence of 2% (v/v) 2-mercaptoethanol, separated in 20% resolving gels after addition of 1 volume 23 loading buffer and analyzed as described in Baier and Dietz (1997). Band intensities were quantified using the GELSCAN software package (BIOSCITECH).

Analysis of Expressional Regulation Patterns
For analysis of coregulation of 223 putatively redox-regulated genes, from the signal intensities published in 249 publicly available Affimetrix cDNA Chip experiments with premature and mature Arabidopsis leaf tissues (http://affymetrix.arabidopsis.info/AffyWatch.html), the Pearson correlation coefficients were calculated for all possible pairs of genes. For the distance plot (Supplemental Fig. S1) the data ranging from 21 to 11 were transformed by f(x) 5 (1 1 x)/2 and analyzed with PAUP matrix using PHYLIP online (http:// evolution.genetics.washington.edu/phylip/phylipweb.html) and plotted into a distance tree.

Metabolite Analysis
Asc, GSH, chlorophyll, pheophytin, and protein contents were determined as described in Baier et al. (2000) and guaiacol peroxidase activity as described in Baier and Dietz (1999b). Starch levels were measured as Glc after digestion with a-amylase and amylo-glucosidase as described in Baier et al. (2004a), with the exception that the starch content was standardized on fresh weight. H 2 O 2 amounts were quantified enzymatically at 436 nm in a reaction mixture containing 100 mM K 2 HPO 4 /KH 2 PO 4 pH 6.5 and 0.05% (v/v) guaiacol (Sigma) with 4 units horseradish peroxidase (Sigma).

Protoplast Isolation and Transfection
Protoplasts were isolated from 3-week-old T19-2 and the mutants as described in Seidel et al. (2004) and transfected in parallel with the reporter gene constructs D1 and D3 (Baier et al., 2004b) according to Seidel et al. (2005). After 6 h incubation at 20 mmol quanta m 22 s 21 , the glucuronidase activity was quantified according to Abel and Theologis (1998). To normalize protoplast and plasmid quality D3-driven glucuronidase activity was standardized on D1-driven activity and on the activity observed in the parental line T19-2.

Hormone Sensitivity Tests
For physiological ABA sensitivity tests, germination of 2-to 8-week-old seeds from six independent harvests was analyzed on Murashige and Skoog plates containing 1% (w/v) Suc and 0.4% (w/v) phytagel (Sigma) and supplemented with 0 to 1 mM ABA (Sigma). After 7 d the number of seeds with emerged roots was counted. Wilting was quantified by monitoring cotyledon wrinkling of seedlings grown in high humidity after exposure to 25% to 30% relative humidity.
Electron microscopy of Arabidopsis chloroplasts was performed with leaves of 5-week-old plants grown under long-day illumination as described by Kö nig et al. (2002) using the contrasting method B by Engels et al. (1997).

Chlorophyll-a Fluorescence Measurements
The quantum yield of PSII (FPSII 5 F V /F M ) was measured every hour beginning 30 min after onset of light under growth conditions using the Mini-PAM Fluorometer (Walz). Light flashes of 2,000 mmol quanta m 22 s 21 were given to analyze the maximum fluorescence. Calculation of photosynthetic parameters was performed as described by Schreiber and Bilger (1993).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Segregation analysis of the low luciferase phenotype in the F2 generation of the backcross of the rimb mutants to the parental line T19-2.
Supplemental Figure S2. Transcript amount correlation analysis.