Arabidopsis thaliana lines were grown in a 10 h light/14 h dark regime on soil as described in Juszczak et al. (2012). For the experiments depicted in Figures 7–9, the plants were grown in a 10 h light/14 h dark regime on 50% MS-medium supplemented with 0.5% sucrose as optimized for growth and expression of genes for chloroplast proteins in Heiber et al. (2014). For induction of flowering, 4 week old plants were transferred to a day/night regime of 16 h light/8 h dark.

rimb6 is an ethyl methanesulfonate mutant of T19-2 (Heiber et al., 2007), which expresses luciferase under control of the 2CPA promoter (Baier et al., 2004). The T-DNA insertion line SALK_111420 (Alonso et al., 2003) and the pathogen signaling mutants were obtained from NASC (http://www.arabidopsis.info). Homozygosity of the T-DNA insertion was confirmed by PCR (Primers Supplementary Table 1). Plant lines were crossed by transferring the pollen of the father plant on the stigma of an emasculated mother plant. The T₁ and T₂ offspring was tested for presence of the T-DNA by PCR. For differentiation of wildtype and cpr1-4 alleles, cDNA or genomic DNA was amplified with the primers CPR1-4 CAPS-F TTGATCTTGCCTTGGAAGAG and CPR1-4 CAPS-R ACAAGGCTACTCACAACGAG by PCR (30 cycles: 30 s 94°C, 30 s 56°C and 30 s 72°C) and digested with Fok-I, which cuts the 391 bp PCR product for the wildtype allele into 233 and 158 bp fragments and leaves the PCR product for the cpr1-4 allel intact. 2% (w/v) agarose gels were run for distinguishing between the 391 bp mutant and the 233 bp wildtype fragments.

For HL treatment, the plants were exposed to 800 μmol quanta m⁻² s⁻¹ for 4 h after 1 h at normal light intensity. H₂O₂ was applied by floating Arabidopsis seedlings, which were germinated on MS-medium supplemented with 0.5% sucrose (Heiber et al., 2007), in liquid MS-medium supplemented with 10 μM H₂O₂ and 0.5% (w/v) sucrose. H₂O₂ was added 1 h after onset of light. For the controls and the HL-treatments depicted in Figure 8, the plants were floated for the same time on MS-medium supplemented only with 0.5% sucrose.

The surfaces of mature leaves of 6-week-old rimb6 and T19-2 plants (grown under short-day conditions) were analyzed by scanning electron microscopy as described in Hiltscher et al. (2014).

A rimb6 mapping population was generated by crossing rimb6 (in the Col-0 background of the line T19-2) (Heiber et al., 2007) to Ler. The F₂ population was scored for low luciferase activity (Heiber et al., 2007) and the dwarf phenotype. For mapping with SSLP and CAPS markers (Jander et al., 2002), genomic DNA was extracted from 220 individual F₂ seedlings. Marker information was taken from the Monsanto Arabidopsis Polymorphism Collection (Jander et al., 2002) and Bell and Ecker (1994).

For high-throughput sequencing, DNA was extracted from a pool of 78 and a pool of 130 plants from the F₂ mapping population that exhibited reduced luciferase expression. For DNA extraction, 1 g of plant material for each pool was homogenized in 10 ml ice-cold nuclei extraction buffer [10 mM Tris-HCl (pH 9.5), 10 mM EDTA (pH 8.0), 100 mM KCl, 500 mM sucrose, 4 mM spermidine, 1 mM spermine and 0.1% (v/v) β-mercaptoethanol] and filtered through two layers of Miracloth (Calbiochem, MERCK, Germany). The samples were gently mixed in 2 ml lysis buffer (10% (v/v) Triton X-100 in nuclei extraction buffer) for 2 min on ice. Following a 10 min centrifugation at 2,000 g at 4°C, the sedimented nuclei were re-suspended in 500 μl CTAB buffer [100 mM Tris-HCl (pH 7.5), 0.7 M NaCl, 10 mM EDTA, 1% (v/v) β-mercaptoethanol and 1% (w/v) CTAB] and incubated for 30 min at 60°C. The samples were mixed by inversion for 5 min at room temperature following the addition of 350 μl chloroform-isoamyl alcohol (24:1). After 10 min centrifugation at 3,300 g, the DNA was precipitated in a 1:1 mixture of the upper phase and isopropanol by 3 min centrifugation at 15.700 g, washed in 75% (v/v) ethanol, dissolved in 50–100 μl DNase free water containing 10 μg/ml RNAseA and incubated for 20 min at 65°C prior to storage at −20°C.

The DNA quantity was determined with a Qubit fluorometer (Life Technologies, Germany). The quality was checked by electrophoresis on a 1.2% (w/v) agarose gel. An indexed paired-end DNA library was prepared for each sample according to Rowan et al. (2015) using 400 ng DNA for each pool and selecting for an insert size of 200–500 bp, and sequenced using a HiSeq2000 system (Illumina, San Diego, CA) with 2 × 100 bp reads.

The adapter sequences were clipped from raw reads, which were then filtered for quality, trimmed to a minimum length of 75 bp, before aligning to the Arabidopsis thaliana reference genome, allowing for a maximum of 10% mismatches and 7% gaps using the SHORE and GenomeMapper software programs (Ossowski et al., 2008; Schneeberger et al., 2009b). The alignments were corrected using paired-end information before polymorphism detection using SHORE. Finally, the allele frequencies of Col-0 and Ler-0 were determined using SHOREmap (Schneeberger et al., 2009b) and the boost function was applied to determine the final mapping interval. An annotated list of all SNPs was obtained using SHOREmap after filtering out the Ler polymorphisms to obtain a final list of candidate mutations.

For confirmation, the candidate gene was amplified with gene-specific primers (Supplementary Table 1) and OptiTaq polymerase (Roboklon, Germany) from the mutant and its parental line. Sanger sequencing (Sanger et al., 1977) of the PCR-products was performed by Eurofins MWG Operon (Ebersberg, Germany).

The shoots of 3–5 individual plants (per replicate) were pooled and immediately frozen in liquid nitrogen. Total RNA was extracted from 100 mg ground plant material using the GeneMatrix Universal RNA Purification Kit (Roboklon, Germany) with on-column treatment with RNase-free DNaseI (Fermentas, Germany). Oligo-d(T) primed cDNA was synthesized from 1 μg RNA with the High Capacity Reverse Transcription Kit (Applied Biosystems, Carlsbad, CA).

Real-time quantitative polymerase chain reactions (qRT-PCR) were performed as described in Hiltscher et al. (2014) or in a final volume of 20 μl containing 50 ng cDNA, 16 mM ammonium sulfate, 0.1 M Tris-HCl (pH 8.3), 0.01% (v/v) Tween-200, 2 mM MgCl₂, 0.1 mM dNTP, 0.2 μl 10X SYBR Green (Sigma-Aldrich, Germany), 0.2 U OptiTaq Polymerase (Roboklon) and 0.3 μM gene-specific primers (Supplementary Table 1). Technical triplicates were run on a CFX96 Real-time System (BioRad, Hercules, CA) for at least three independent biological samples using 40 cycles 95°C/15 s, 60°C/30 s and 72°C/30 s after 5 min incubation at 95°C. Each assay included a standard curve of four serial dilution points of Col-0 cDNA (300 ng–300 pg) and a non-template control. Exon–intron border spanning primers were designed using QUANTPRIME (Arvidsson et al., 2008). Primer specificity was assessed by inspection of the melting curves after 40 cycles. The Cq values were determined using the regression model within the CFX Manager software v3.0 (BioRad, Hercules, CA) and analyzed with respect to amplification efficiency. The transcript levels were normalized on the geometric mean transcript level of the constitutively expressed genes At3g18780 (actin2), At5g15710 (F-box), and At5g08290 (YLS8) (Czechowski et al., 2005).

100 mg plant material were immediately frozen in liquid nitrogen and lyophilized in 1 ml 80% methanol containing 10 μM ribitol (as an internal standard) and 0.5 g zirconia glass beads (Carl Roth, Germany) in a FastPrep™ Instrument (Qbiogene, Germany) using 45 cycles of 6.5 m s⁻¹ followed by a 15 min incubation at 70°C with continuous shaking at 1,400 rpm. Following 20 min centrifugation at 13,000 g, the supernatant was dried in a nitrogen stream in glass vials. To extract metabolites, 50 μl of 20 mg/l methoxylamine hydrochloride (in pyridine) were added to the samples prior to incubation at 90 min at 37°C. After addition of 50 μl N-methyl-N-[trimethylsilyl] trifluoroacetamide they were incubated for 30 min at 37°C.

One μl of each sample was analyzed in a TraceGC gas chromatorgraph coupled to a PolarisQ ion trap mass spectrometer and an AS2000 auto sampler (Thermo Finnigan, Germany). Injection was performed at 250°C (splitness mode) and separation was achieved on a 30 m × 0.25 mm Equity-5 column with 0.25 μm coating (Supleco, Bellefonte, CA, USA) at an interface temperature of 250°C and an ion source temperature of 200°C in a constant flow of helium carrier gas of 1 ml min⁻¹. Following 2 min constant heating at 80°C, the oven temperature was raised to 300°C with a speed of 3°C min⁻¹. Mass spectra were recorded in a range of 50–550 m z⁻¹. Metabolites were identified by comparison with the NIST98 (NIST, Gaithersburg, MD) database, pure standards and by using the Golm Metabolome Database (Kopka et al., 2005). Metabolite peak relative areas were quantified using the processing setup implemented in the Xcalibur software (Thermo Finnigan, Germany) and normalized to the peak area of the internal standard ribitol. Differences between T19-2 and cpr1-4 were evaluated based on the P-value of pairwise t-tests (P < 0.1).

The concentration of ascorbate and glutathione were determined in 8–12 biological replicates per genotype and treatment as described in Baier et al. (2000). The redox state was calculated by dividing the concentration determined for the oxidized form by the total concentration (oxidized plus reduced forms) for the same extract. The DHA/Asc and GSSG/2x GSH ratios were calculated by dividing the concentration of the oxidized form by the concentration determined for the reduced form.

For quantification of phytohormone levels approximately 150 mg frozen plant material was extracted in a FastPrep®-24 instrument (MP Biomedicals, USA) at 5 m s⁻¹ for 60 s in 2 ml screw-cab-tubes containing 1.25 g of 2.8–3.3-mm-diameter Zirconox beads, (Mühlmeier Mahltechnik, Germany) and 1 ml ethylacetate, including 20 ng D4-SA (OlChemIm Ltd., Czech Republic) as an internal standard. After centrifugation (10 min at 15,000 g at 4°C), the supernatant was transferred to a 2 ml reaction tube and the pellet was extracted a second time with 1 ml pure ethylacetate. Supernatants from both extractions were combined and dried in a vacuum concentrator (concentrator 5301, Eppendorf, Germany). The residue was eluted in 400 μl of 70% methanol with 0.1% formic acid (v/v) at room temperature. The extract was centrifuged again for 10 min at 15,000 g and 4°C, and 200 μl of supernatant were transferred to HPLC-vials. Analysis was performed using a 7 μl injection into a UPLC-ESI-MS/MS (Synapt G2-S HDMS; Waters, Milford, USA). Chromatography was performed on a C18 column (Acquinity UPLC BEH-C18, ø 2.1 × 50 mm, with a particle size 1.7 μm) at 30°C and a flow rate of 250 μl/min. Water and methanol [each containing 0.1% formic acid (v/v)] were used as solvents in a gradient (methanol: 0 min: 30%, 1 min: 30%, 4.5 min: 90%, 8 min: 90%, 9 min: 30%) with a 3 min equilibration time between the runs. Tandem mass spectrometry was performed in negative ionization mode with parent/daughter ion selections of 137/93 (SA), 141/97 (D4-SA), 209/59 (JA), 215/59 (D6-JA), 263/153 (ABA), 269/153 (D6-ABA), 322/130 (JA-Ile), 328/130 (D6-JA-Ile). Peak areas of daughter ions were integrated using MassLynx™ Software (version 4.1, Waters) and the amount of phytohormones was calculated according to the internal standard.

Histochemical staining for O2- and H₂O₂ and semiquantitative analysis was performed with nitroblue tetrazolium (NBT) and 3,3-diaminobenzidine (DAB) as described in Juszczak et al. (2016).

All quantitative data were subjected to statistical analysis using the two-tailed, pairwise t-test, ANOVA (Bonferroni/Tukey testing), pairwise t-test, X²-test or the pairwise Welch's test.

The growth habit of the rimb6 mutant was indistinguishable from its parental line T19-2 during germination and at the seedling age. In the rosette stage, the leaves showed reduced expansion, the leaf margins were curled (Figure 1A, Supplementary Figure 1) and growth was slowed relative to T19-2 (Figures 1C,D). In mature leaves, the epidermal pavement cells were 30 ± 9% smaller and had fewer lobes (Figure 1E). However, rimb6 did not show apoptotic mesophyll clefts as observed in rimb1 (Hiltscher et al., 2014), which was isolated in the same screening approach (Heiber et al., 2007).

The onset of bolting was similar in rimb6 and T19-2, when 4-week old plants were shifted to long-day conditions (14 h light/10 h dark), demonstrating that meristem reprogramming from vegetative to generative growth is unaffected in the mutant. The primary inflorescences of rimb6 were shorter, had fewer branches and flowers than wildtype (wt) Col-0 or T19-2 (Figures 1A,B; Supplementary Figure 2). Many secondary shoots were released shortly after bolting, giving rimb6 a bushy habitus (Supplementary Figure 1).

The RIMB6 locus was mapped by a combination of SSLP mapping (Jander et al., 2002) and high-throughput sequencing (Schneeberger et al., 2009a). Plants showing the rimb6 dwarf phenotype were selected from the F₂ population derived from a cross of rimb6 and wildtype plants of the Arabidopsis thaliana accession Landsberg erecta (Ler). The rimb6 mutation was localized in the phenotyped F₂ population with simple sequence length polymorphism (SSLP) markers on chromosome IV between the markers NGA8 (Bell and Ecker, 1994) and CER46127 (Jander et al., 2002).

Illumina sequencing of two pools of F₂ plants with low luciferase activity—one with 78 and one with 130 individuals—showed enrichment of Col-0 alleles (>90%) on chromosome IV between 7 and 9 Mb. After removing known Col/Ler single nucleotide polymorphisms (SNPs), 12 putative mutations remained (Supplementary Table 2). Three of the mutations were G/A substitutions and one was a T/A substitution. The other eight mutations were putative deletions. G/A substitutions result from chemically induced C/T substitutions and are most typical (>99%) for mutagenesis by ethyl methansulfonate (Greene et al., 2003). The G/A mutations at the positions 7,078,331 and 7,442,672 were non-synonymous and were investigated further as potential candidates. The mutation at position 7,442,672 was confirmed by sequence comparison of PCR products amplified from genomic DNA of rimb6 and T19-2. The observed mutation was only found in rimb6-derived PCR products and absent from those amplified from the non-mutagenized line, T19-2. The other polymorphisms were found in both rimb6 and T19-2 and could therefore not be causal for the phenotype. The candidate mutation changes a tryptophan codon, TGG, in the mRNA encoding the CONSTITUTIVE EXPRESSOR OF PATHOGENESIS RESPONSE GENES 1 (CPR1) (At4g12560) into a stop codon, TGA, and terminates translation after 286 amino acids (Figure 2A). Wildtype CPR1 antagonizes effector triggered immunity (ETI) in the cytosol by mediating the proteasomal degradation of R-proteins, like SNC1 and RPS2 (Gou et al., 2009; Cheng et al., 2011). As one of its basic functions, besides antagonizing ETI induction, CPR1 controls microtubule arrangement (Han et al., 2015). The truncated CPR1 protein produced by the rimb6 mutation lacks the C-terminal FBA domain (Figure 2B), which binds target proteins in an E3-ubiquitin ligase complex and guides R-proteins (Gou et al., 2009; Cheng et al., 2011) and other proteins (Wang et al., 2014) toward ubiquitinylation and degradation.

The T-DNA insertion line SALK_111420 carries an insertion in the first exon of CPR1 (Figure 2A). The homozygous line, which lacks detectable CPR1 expression (Figure 2D), was phenotypically similar to rimb6 (Supplementary Figure 2). Allelism of rimb6 was confirmed by crossing rimb6 (cpr1-4) with the T-DNA insertion line. rimb6 x SALK_111420 and SALK_111420 x rimb6 F₁ plants showed the mutant phenotype (Supplementary Figure 1), while reciprocal crosses of rimb6 and its non-mutagenized parental line T19-2 did not. The rimb6 mutant was renamed cpr1-4 and the T-DNA insertion line SALK_111420 as cpr1-5.

qRT-PCR analysis with primers binding to the 3′-part of the At4g12560 transcript (= 5′-part of the oligo-dT-primed cDNA) (Supplementary Table 1) demonstrated that cpr1 transcripts were present at lower levels in cpr1-4 than the CPR1 transcripts in wt plants (Figure 2D), indicating a feed-back of early translation termination on transcript stability.

In wildtype plants, CPR1 antagonizes induction of defense marker genes, e.g., PHYTOALEXIN DEFICIENT 4 (PAD4), SALICYLIC ACID DEFICIENT 2 (SID2) and PATHOGENESIS RELATED GENE 2 (PR2) (Gou et al., 2009), in the absence of pathogen stimuli by supporting degradation of R-proteins, such as SNC1 and RPS2 (Cheng et al., 2011; Gou et al., 2012). These genes were highly expressed in cpr1-4 and cpr1-5 (Figure 2E), but barely detectable in T19-2 and the background line of cpr1-5 (SALK-wt), demonstrating that both alleles constitutively activate defense gene expression like previously described for other cpr1-alleles (Gou et al., 2009).

To prove the causality between 2CPA miss-regulation and the cpr1-4 mutation, the cpr1-4 mutant was transformed with a construct expressing wildtype CPR1 under control of the CPR1 promoter (pCPR1::CPR1 in cpr1-4). The segregating T₂ population was screened with a CAPS marker (cleaved amplified polymorphic sequence) that amplifies a product where the wt CPR1 allele has a FokI cleavage site and cpr1-4 does not (Figure 3A). Four T₂ lines were selected, which express the wildtype allele, and four lines without the pCPR1::CPR1 construct. Only the four T₂-lines expressing the transgene showed wildtype growth and development, demonstrating complementation of the cpr1-4 mutant phenotype.

The activity of the transgene was analyzed by qRT-PCR (Figure 3B). The CPR1 transcript level was 3.6-fold higher in the CPR1-transgenic lines than in the cpr1-4 mutant, indicating that approximately 2/3 of the CPR1 transcripts in the pCPR1::CPR1 transformants encode wildtype CPR1. In the complemented lines, the 2CPA transcript levels were by average 1.6-fold higher than in cpr1-4 and the PAD4 transcript levels and the PR1 transcript levels were significantly lower than in the cpr1-4 mutant, and in the range of wildtype plants (blue lines in Figure 3B), demonstrating that CPR1 regulates 2CPA and the two ETI genes PAD4 and PR1 inversely.

The relevance of CPR1 for chloroplast function tempted us to compare the carbohydrate and amino acid profiles in 28-day-old cpr1-4 and T19-2 1 and 5 h after exposure to light by gas chromatography coupled to mass spectrometry (GS-MS) (Figure 4). The time-points address activation of light metabolism, which often diminishes the carbohydrate pools transiently, before carbohydrates and secondary photosynthates accumulate (Gibon et al., 2004; Zeeman et al., 2007; Figure 4, left).

The concentrations of glucose, fructose and pyruvate were lower in cpr1-4 than in T19-2 1 h after onset of light (Figure 4, right). On the contrary, the concentrations of glucose-6-P, fructose-6-P and malate were all increased in cpr1-4, reflecting that the hexose energization status (hexose-P/hexose-ratio) and the reduction state of the malate/oxaloacetate system were both high. The levels of N-containing putrescine, aspartate, asparagine, threonine and leucine were also higher in cpr1-4.

After 5 h of light exposure, the glucose and fructose availabilities were restored to control levels in cpr1-4 (Figure 4, right), but the pyruvate shortage remained and the glucose-6-P, fructose-6-P and serine levels were still increased. At this time point, the aspartate concentration had fallen to that of the control plants and the levels of the aspartate-derived amino acids threonine and asparagine were still increased. In addition to malate, citrate, isocitrate and α-ketoglutarate levels were enriched and stress metabolites such as ornithine, arginine, proline and glutamine and the pentose phosphate cycle intermediate erythrose-4-P accumulated. Taken together, these results show that the reduction status of metabolites, the hexose energization and the amination status were increased in cpr1-4.

The SA concentration was increased in cpr1-4 (Figure 5A) as reported for cpr1-2/cpr30 (Gou et al., 2009). Accumulation of SA at the youngest tested stage (Figure 5A), demonstrated early activation of SA-biosynthesis in cpr1-4. The concentrations of the wounding hormone jasmonate and its isoleucine-conjugate did not differ significantly between T19-2 and cpr1-4 (Figure 5A). The levels of abscisic acid (ABA), which is a repressor of 2CPA expression and of many other nuclear genes for chloroplast proteins and plastid genes (Baier et al., 2004; Staneloni et al., 2008; Yamburenko et al., 2013), were also similar at 14 and 28 d, and decreased in cpr1-4 at 42 d (as compared to T19-2), excluding any relevance of ABA with respect to low 2CPA activation in cpr1 in early stages.

To analyse how CPR1 affects defense and plastid antioxidant signaling, we compared the expression of 2CPA, MDAR and tAPX (PAS genes) with PAD4, SID2, and PR2 (defense genes) in leaves of 14-, 28-, and 42-day-old cpr1-4 and T19-2 by qRT-PCR (Figure 5B; data for PAS-gene regulation in cpr1-5: Supplementary Figure 3). In cpr1-4, the transcript levels of the three PAS genes were slightly reduced in 14- and 28-day-old plants, but were similar to T19-2 at 42 days. The expression of the three defense genes was higher in cpr1-4 mutants than in T19-2 at all developmental time points (Figure 5B). For PAD4 and SID2, the difference was strongest in 28-day-old plants.

To study the consequences of the cpr1-4 mutation and insufficient antioxidant protection on ROS-metabolism, we analyzed the levels of the two major ROS, O2- and H₂O₂ (Figure 6). Generation of O2- at the thylakoids (Mehler, 1951) increases upon excess excitation pressure (Foyer et al., 1994). cpr1-4 showed higher O2- and H₂O₂ levels in the cotyledons of 14-day-old plants than T19-2 (Figure 6A). At an age of 28 d, some staining patterns indicated higher ROS-levels in cpr1-4 (Figure 6A). Quantification of a series of plants showed high variability and indistinguishable mean values between T19-2 and cpr1-4 (Figure 6B). Later, at 42 d, the O2- and H₂O₂ levels again increased to significantly higher ones than in T19-2 (Figure 6B).

ZAT10, which responds to various types of ROS of plastidic and extra-plastidic origin, such as to ozone, H₂O₂ and O₂⁻ (Rossel et al., 2007), was strongly induced in 28-day-old cpr1-4 plants (Figure 6C) and in cpr1-5 plants of all age (Supplementary Figure 3). The ROS-inducible gene BAP1 responds to transfer of excess energy from pigments to oxygen at the thylakoid membrane (op den Camp et al., 2003). It was expressed more strongly in leaves of 14- and 28-day-old cpr1-4 plants than those of T19-2 (Figure 6C) and highly accumulated in 14, 28, and 42 day old cpr1-5 as compared to the respective wildtype (Supplementary Figure 3).

The cpr1-4 (rimb6) mutant was isolated for low activation of 2CPA promoter activity at the seedling stage and shown to be affected in expression of other PAS genes in 3 week old soil grown plants (Heiber et al., 2007). To test the target spectrum and the impact of elevated ROS levels, we investigated the regulation of a series of PAS genes, genes for light-harvesting complex proteins (LHCA and LHCB), for photosynthetic electron transport components (PET genes), ribulose-1,5-bisphosphate carboxylase small subunit (RBCS), sugar metabolism proteins (APL3 and STP1), stress marker genes (BAP1 and FER1) and extra-plastidic antioxidant enzymes (APX2 and CAT2) in T19-2 and cpr1-4 in response to externally applied H₂O₂ and to high light (HL) in 9-day-old seedlings on MS-medium optimized for seedling growth and minimal impact on greening and PAS gene expression (Heiber et al., 2014) (Figure 7). The cpr1-4 mutant showed significantly lower expression of genes for chloroplast peroxidases, monodehydroascorbate reductase (MDAR) and LHCB2.2 under control conditions. The transcript levels of the ROS-regulated genes, BAP1 (Bachmann et al., 2002; op den Camp et al., 2003; Spoel et al., 2003) and Fer1 (Bachmann et al., 2002; op den Camp et al., 2003; Spoel et al., 2003), were slightly (BAP1) or significantly (Fer1) increased in cpr1-4 mutants under control conditions (Figure 7).

CAT2 (catalase) and APX2 (ascorbate peroxidase), which encode ROS-inducible extra-plastidic antioxidant enzymes (Mullineaux et al., 2000; Du et al., 2008), responded to the HL and H₂O₂ treatment in T19-2, but showed no difference between T19-2 and cpr1-4 under control conditions.

In cpr1-4, the transcript levels of PrxQ, tAPX, and LHCB2.2 were lower in control plants and in H₂O₂- and HL-treated plants (Figure 7). For PrxQ and tAPx, the mRNA levels were slightly decreased in H₂O₂-treated plants and significantly lower in response to HL. The transcript levels of the two peroxiredoxin genes 2CPA and 2CPB, of sAPx and of CuZn superoxide dismutase 2 (CSD2) were lower in H₂O₂-treated cpr1-4 than in cpr1-4 under control conditions. Moderate HL resulted in higher transcript levels. Antagonistic regulation by light and H₂O₂ is consistent with the regulatory model postulated for 2CPA based on identification of distinct promoter motifs (Baier et al., 2004) and analysis of redox-box regulation by the transcription factor RAP2.4a (Shaikhali et al., 2008).

In 28-day-old sterile grown plants, the transcript levels of most PAS, PET and LHC genes and of RBCS were at least slightly more decreased after 4 h HL treatment than in the control plants harvested 1 h after onset of light (Figure 8). However, for tAPX, ECS, GR, PETC and PETM the HL effect was not distinguishable from the effect of additional 4 h at normal light intensity. sAPX was expressed at higher levels in cpr1-4 than in T19-2 1 h after onset of light and indistinguishable from T19-2 after 5 h in standard growth light or HL. Although sAPX and tAPX are coregulated by RAP2.4 transcription factors (Rudnik et al., 2017), sAPX is often inversely regulated to tAPX in response to stress or metabolite availability (Heiber et al., 2014; Juszczak et al., 2016). CSD2 transcript levels were not decreased 1h after onset of light and were less decreased after 5 h illumination in cpr1-4 than in T19-2.

Ascorbate biosynthesis depends on carbohydrate availability and stress activation (Bartoli et al., 2003; Pena-Ahumada et al., 2006; Heiber et al., 2014). The ascorbate levels were higher in 9-day-old cpr1-4 plants than in T19-2 (Figure 9). Four hours of incubation with H₂O₂ or illumination with 800 μmol photons m⁻² s⁻¹ (HL) had similar effects on ascorbate consumption and ascorbate oxidation in both genotypes (Figure 9 and Supplementary Figure 4). During further rosette development, ascorbate accumulated. In 28-day-old cpr1-4, the reduction state of the ascorbate pool was attenuated (Figure 9). However, the ascorbate pool size was decreased in cpr1-4 after 5 h illumination with growth light intensity (NL) and HL (Figure 9) demonstrating increased consumption.

Transcript abundance analysis showed stronger expression of GR and slightly stronger expression of ECS (γ-glutamyl-cysteine synthase), which are involved in glutathione reduction and biosynthesis, respectively, in 9 day old cpr1-4 mutants in response to H₂O₂ application and lower expression in 28 day old cpr1-4 after 5 h illumination with standard light and high intensities demonstrating that the regulatory effect is light intensity independent. Quantification of the glutathione content and determination of the redox state of the glutathione pool (Figure 9) and calculation of the GSSG/GSH ratio (Supplementary Figure 4) demonstrated that the glutathione pool was, in contrast to the ascorbate pool, not significantly affected in cpr1-4.

Various PAS genes and pathogen response genes were regulated inversely in cpr1-4 as compared to T19-2 (Figure 6). To test whether the expression of the immune regulators affect PAS gene expression in absence of pathogens, as CPR1 does, we analyzed the transcript levels of 2CPA, MDAR, and sAPX in comparison to CPR1 and the defense genes PAD4 and PR2 during development in wt plants and the immune signaling mutants npr1 (Cao et al., 1997), pad4 (Jirage et al., 1999), and rps2 (Cheng et al., 2011) at three developmental stages.

CPR1 transcript levels decreased with age in Col-0 (Figure 10). PAD4 levels tended to increase. The wildtype CPR1 pattern was maintained in the pad4 and rps2 mutants. In the npr1 mutant, CPR1 transcript levels were increased as compared to Col-0 at an age of 28 d and 42 d. The PAD4 mRNA levels were elevated in npr1 at 28 d and in rps2 at 28 d and 42 d.

As a read-out for induction of immunity responses, PR2 transcript levels were analyzed. They increased in all genotypes at an age of 42 d, but reached higher levels in npr1 and rps2 and lower levels in pad4 compared to Col-0.

The PAS genes 2CPA, sAPX and MDAR showed highest expression at 28 days in Col-0. 2CPA expression, which was the reporter used for selecting the rimb-mutants (Heiber et al., 2007), was not significantly affected in any of the three immune mutants (Figure 10). sAPX showed slightly higher levels in npr1 and pad4 mutants in 14-day-old seedlings, but the general developmental pattern was maintained. MDAR expression showed the wt developmental pattern in pad4, but not in npr1 and rps2 (Figure 10). None of the genes for PAS enzymes was disregulated in these defense signaling mutants to the same extent as in cpr1-4 (Figures 5, 10). We conclude that the CPR1-mediated regulation of defense responses and PAS genes occur independently and this partially depends on the developmental stage of the plants.

To test the hypothesis, we compared 2CPA, PR1, and PR2 expression in wildtype plants, cpr1-4 and pad4 single and cpr1-4 x pad4 double mutants after 10 days of growth on 0.5 MS medium, which were the conditions for the rimb-mutant screen (Heiber et al., 2007). pad4 was crossed into the cpr1-4 background to avoid activation of the SA biosynthesis, while maintaining basic SA biosynthesis and SA sensitivity (Zhou et al., 1998). The pad4 mutation widely restored the leaf habitus and rosette growth defects of cpr1-4 (Figure 11). At an age of 28 days, the rosette diameter of cpr1-4 mutants was about 50% of wildtype Col-0 and had 29.5 ± 3.7% fewer leaves, while the double mutants were only 14.3 ± 2.4% smaller in diameter and had formed on average 2.1 ± 0.5 fewer leaves than Col-0.

Despite the high data noise due to low expression intensity of PR1 and PR2 in wildtype plants, the transcript levels of PR1 and PR2 were significantly (pairwise t-test; p < 0.01) increased in the cpr1-4 mutants. The strong induction effect was lost in cpr1-4 x pad4 double mutants, demonstrating that transcript accumulation of both genes depends on activation of the PAD4-mediated SA loop (Figure 11). On the contrary, lack of PAD4 did not (significantly) affect the cpr1-4 mutant effect on 2CPA expression (Figure 11), showing that low expression of 2CPA in the cpr1-4 mutant is independent from SA accumulation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.