Interactions between cat2 and pp2a-b′γ mutations during growth in LD
First, plants were analyzed by following growth in LD at moderate light, conditions in which pathogenesis responses are activated by oxidative stress in cat2. All three mutant lines were smaller than the wild-type, with the double mutant showing a significant decrease relative to cat2 (Fig. 1a). Lesions were clearly apparent in both cat2 backgrounds, and slightly more extensive in cat2 pp2a-b′γ. The pp2a-b′γ single mutant also presented some lesions, although to a much lesser extent than cat2 (Fig. 1a). There was an overall correlation between lesion extent and expression of PATHOGENESIS-RELATED GENE1 (PR1). This SA marker gene was strongly induced in cat2, but not significantly in pp2a-b′γ (Fig. 1b). In the double mutant, PR1 expression was increased to about two-fold the cat2 values (Fig. 1b). Analysis of SA and related defense compounds showed a similar pattern, with some induction in the pp2a-b′γ mutant and stronger induction in cat2. The double mutant generally showed the most marked accumulation of these compounds, and this was particularly apparent for camalexin, a major phytoalexin in Arabidopsis (Glawischnig, 2007), which was about three-fold higher in cat2 pp2a-b′γ than in cat2 (Fig. 1c). Together, these observations show that loss of PP2A-B′γ function produces the activation of defense responses, as reported previously (Trotta et al., 2011a), and that the mutation reinforces these responses in cat2.
Figure 1. Growth and pathogenesis-related responses in Arabidopsis Col-0, cat2, pp2a-b′γ and cat2 pp2a-b′γ growing in long days (16 h light : 8 h dark). (a) Representative photographs of plants and quantification of rosette mass and lesion area. (b) PATHOGENESIS-RELATED GENE1 (PR1) expression measured by quantitative PCR (qPCR). (c) Salicylic acid (SA) and related defense compounds. Data are means ± SE of at least 10 plants (a) or three to four biological replicates (b, c). * and +, significant difference relative to Col-0 or to cat2 (P < 0.05). FW, fresh weight; nd, not detected.
Download figure to PowerPoint
In our hands, changes in H2O2 are negligible in cat2, either because of rapid metabolism by alternative pathways, technical difficulties in the specific quantification of this reactive molecule, or both (Queval et al., 2008; Mhamdi et al., 2010a,c). Increased accumulation of oxidized glutathione is a well-described response in catalase-deficient plants (Smith et al., 1984; Willekens et al., 1997), and is the clearest biochemical marker of oxidative stress in cat2 (Queval et al., 2007; Mhamdi et al., 2010a,c). This key determinant of intracellular thiol-disulfide status is clearly implicated in pathogenesis and other plant stress responses (Ball et al., 2004; Parisy et al., 2007), and is functionally important in determining the activation of phytohormone signaling and related phenotypes in cat2 grown in LD (Han et al., 2013a,b). Like glutathione, ascorbate is a marker of tissue redox status, and has been implicated in responses to biotic stress (Pastori et al., 2003). The reduced and oxidized forms of these two molecules were therefore quantified as markers of cellular redox state.
In all mutants, ascorbate was slightly more oxidized than in Col-0, but, in agreement with previous studies, its status was much less affected than glutathione by the cat2 mutation (Fig. S2). In cat2, glutathione was increased c. 2.5-fold, and this was associated with a drop in the redox state from over 90% to c. 60% reduced (Fig. S2). Slight oxidation of glutathione was also observed in pp2a-b′γ, but the combination of the two mutations did not cause further oxidation or accumulation relative to cat2 (Fig. S2).
One feature of the cat2 mutant is its conditional photorespiratory nature. This means that this line shows little sign of oxidative stress when grown at high CO2 or low light, conditions in which photorespiration is negligible or slow. At low light, the pp2a-b′γ mutant showed a similar decrease in growth to that observed at moderate light (Fig. S3). By contrast, the impact of the cat2 mutation on rosette size or glutathione status was much less marked than at the standard irradiance (compare data in Figs 1 and S3). In low light conditions, no lesions were detectable on the leaves of any of the mutants. Experiments in which plants were grown at high CO2 also revealed that no lesions were observed on single or double mutants (data not shown). Thus, whether or not PP2A-B′γ was functional, lesions triggered by the cat2 mutation were conditionally dependent on a threshold rate of photorespiratory H2O2 production, and the oxidative stress that resulted from it.
Effect of the pp2a-b′γ mutation in SD, conditions that are non-permissive for the activation of pathogenesis responses in cat2
Because the above data reveal a functional interaction between the pp2a-b′γ and cat2 mutations on the pathogenesis responses in LD, we focused the remainder of our study on establishing whether PP2A-B′γ could be involved in regulating these responses in cat2 grown in SD. When plants were grown in SD at the same irradiance as in LD (Fig. 1), oxidative stress in cat2 produced a decrease in growth, but no lesions were observed and PR1 was not induced (Fig. 2). Lesions were also absent on the leaves of pp2a-b′γ. By contrast, the cat2 pp2a-b′γ double mutant showed significant lesion formation and induction of PR1 (Fig. 2). Unlike cat2, the double mutant also showed significant DAB staining (Fig. 2).
Figure 2. The pp2a-b′γ mutation allows intracellular oxidative stress to trigger lesions and PATHOGENESIS-RELATED GENE1 (PR1) expression responses in Arabidopsis in otherwise non-permissive conditions (short days; 8 h light : 16 h dark). Data are means ± SE of at least 10 plants (lesions) or five biological replicates (PR1). Photographs at the bottom show staining of tissues with 3,3′-diaminobenzidine (DAB). Note that the top and bottom series of photographs show different plants; nd, not detected.
Download figure to PowerPoint
Redox analysis revealed that the appearance of lesions in the double mutant was not associated with marked increases in oxidation of either ascorbate or glutathione relative to cat2 (Fig. 3). Similar to effects observed in LD (Fig. S2), the double mutant showed a slight tendency to increased oxidation of ascorbate (Fig. 3). The pp2a-b′γ mutation did not affect glutathione status in the Col-0 background. In cat2 in SD, glutathione was even more oxidized than in LD (Fig. 3; compare with data in Fig. S2), but neither the oxidation state nor the accumulation was further increased by the secondary pp2a-b′γ mutation. Indeed, the lesions observed in SD in the double mutant (Fig. 2) were associated with a more reduced glutathione pool than in cat2 (Fig. 3), in which no lesions were apparent in this condition (Fig. 2).
Figure 3. Major leaf antioxidant pools in Arabidopsis Col-0, cat2, pp2a-b′γ and cat2 pp2a-b′γ growing in short days. White bars, reduced forms; black bars, oxidized forms. Data are means ± SE of three biological replicates. Where significant, * and + indicate difference relative to Col-0 or to cat2 (P < 0.05) for each form. Numbers above the bars indicate the percentage of ascorbate or glutathione in the reduced form. FW, fresh weight.
Download figure to PowerPoint
To identify the metabolic changes underlying the lesions induced in cat2 pp2a-b′γ in SD, non-targeted profiling of almost 100 metabolites was performed by GC-TOF-MS (Table S2). Using this technique, we have previously identified a specific metabolite signature induced by intracellular oxidative stress in an LD- and SA-dependent manner (Chaouch et al., 2010). This signature shows considerable overlap with that induced by bacterial challenge, and its main features are absent in both cat2 grown in SD and in a cat2 sid2 mutant in LD (Chaouch et al., 2010, 2012). Hierarchical clustering of all metabolites quantified in samples taken in SD revealed that the single cat2 and pp2a-b′γ mutations had only a minor impact (Fig. 4, left). By contrast, a marked effect on metabolite profiles was observed in the double mutant (Fig. 4, left). This involved the accumulation of a large subcluster of compounds that accumulated much less or not at all in the cat2 single mutant. Comparison with a previously described dataset revealed that most of these compounds were those that accumulated in the cat2 single mutant in LD, but not in cat2 sid2 (Fig. 4). Thus, the pp2a-b′γ mutation allows intracellular oxidative stress to induce metabolic responses in SD that are only observed in LD when PP2A-B′γ is functional, and these responses are dependent on the isochorismate pathway of SA synthesis. Indeed, SA itself was among the metabolites that were most clearly dependent on the cat2–pp2a-b′γ interaction (Fig. 4). Others included nicotinic acid, which has been implicated in pathogenesis responses (Noctor et al., 2011), and gluconic acid, which is strongly induced by bacterial challenge as well as in cat2 in LD (Chaouch et al., 2012; Tables 1, S2).
Table 1. The most induced metabolites in the leaves of the Arabidopsis cat2 pp2a-b′γ double mutant in short days (SD). Fifty-three significantly different metabolites were identified in cat2 pp2a-b′γ relative to Col-0 (Table S2b). The 10 most induced of these are shown below
|Short days (this study)||Long days (Chaouch et al., 2012)|
| cat2 ||cat2 pp2a-b′γ||cat2 pp2a-b′γ|| cat2 |
|rel. Col-0||rel. Col-0||rel. cat2||rel. Col-0|
Figure 4. Gas chromatography-time of flight-mass spectrometry (GC-TOF-MS) analysis of Arabidopsis leaves reveals that the pp2a-b′γ mutation allows the induction of cat2 metabolite profiles in short days that are only otherwise induced in long days in an ISOCHORISMATE SYNTHASE1 (ICS1)-dependent manner. Left heatmap, display of identified metabolites in the four plant lines grown in short days. Right heatmap, corresponding data from Chaouch et al. (2010), replotted in a new form, for Col-0, cat2 and cat2 sid2 grown in long days. White rows indicate metabolites that were not detected in at least one sample in the previous analysis. Graphs on the right show selected metabolites that are induced in cat2 pp2a-b′γ in short days, with a comparison of their response in cat2 and cat2 sid2 in long days (means ± SE of three biological replicates).
Download figure to PowerPoint
The combined presence of the two mutations produced striking changes in many amino acids. In our conditions, although minor increases in homoserine and tryptophan were apparent, the pp2a-b′γ single mutant showed amino acid profiles that were similar to those of the wild-type (Fig. S4). When introduced into the cat2 background, the pp2a-b′γ mutation markedly stimulated amino acid accumulation. In some cases, the main effect was to reinforce increases that were already evident in cat2 (e.g. glutamate, glutamine, Cys, Met, Ser, O-acetylserine). In many cases, however, the two mutations acted together to produce increases that were minor or negligible in the single mutants. Striking examples of compounds that responded in this way were isoleucine, ornithine, Thr, lysine, β-alanine, GABA, phenylalanine, Tyr, tryptophan and arginine (Fig. S4), the last two featuring among the most strongly induced significant metabolites observed in cat2 pp2a-b′γ (Table 1).
To further investigate the role of SA, this compound was quantified by HPLC in plants grown in SD, together with the phytoalexin, camalexin. Despite the evident disulfide stress in cat2 (Fig. 3), the amounts of SA and camalexin were not significantly different from those in Col-0 (Fig. 5). Some induction of SA was observed in pp2a-b′γ, but the accumulation was much more marked in the double mutant, where total SA reached levels observed in cat2 in LD (Fig. 1). Camalexin accumulation above basal Col-0 levels was only significant in cat2 pp2a-b′γ, in which levels of this compound were similar to those in cat2 in LD (Fig 5; compare with Fig. 1). To establish whether the lesions associated with pathogenesis responses in cat2 pp2a-b′γ in SD were dependent on the SA pathway, triple mutants were produced that additionally carried either sid2, defective in SA synthesis (Wildermuth et al., 2001), or npr1, defective in SA signaling (Cao et al., 1994). In cat2 pp2a-b′γ npr1, lesions were significantly decreased, whereas, in cat2 pp2a-b′γ sid2, they were absent (Fig. 6). As reported previously for cat2 growing in LD, lesion formation in cat2 pp2a-b′γ in SD could be prevented simply by treating plants with myo-inositol (Fig. 6), a compound that is implicated in the SA signaling pathway (Meng et al., 2009; Chaouch & Noctor, 2010; Donahue et al., 2010). Together, these results show that the lesions produced in cat2 pp2a-b′γ in SD were dependent on the SA pathway in a similar manner to the lesions observed in the cat2 single mutant in LD.
Figure 5. Loss of PP2A-B′γ function allows the cat2 mutation to trigger the accumulation of salicylic acid (SA) and camalexin in Arabidopsis growing in short days. Data are means ± SE of four biological replicates. * and +, significant difference relative to Col-0 or to cat2 (P < 0.05). FW, fresh weight.
Download figure to PowerPoint
Figure 6. Complete or partial genetic reversion of the Arabidopsis cat2 pp2a-b′γ lesion phenotype in short days by sid2 and npr1 mutations, and chemical blocking of the lesions by myo-inositol (MI) treatment. For lesion quantification, data are means ± SE (n = at least 10 plants). Examples of lesion-forming areas are indicated on the photographs by white arrows. nd, not detected.
Download figure to PowerPoint
Proteome and phosphoproteome modifications induced by the pp2a-b′γ mutation during oxidative stress
Next, we assessed how the features related to defense signaling were reflected in the proteome, and particularly the phosphoproteome, of cat2 pp2a-b′γ double mutant plants. First, we analyzed the pattern of phosphoproteins in total soluble and membrane fractions isolated from leaves of Col-0, cat2, pp2a-b′γ and cat2 pp2a-b′γ grown in SD. SDS-PAGE and subsequent quantitative ProQ staining of phosphoproteins, followed by Sypro staining of total proteins, showed no major changes between the cat2 and pp2a-b′γ single mutants and Col-0 (Fig. S5). By contrast, a set of highly phosphorylated bands emerged in the soluble proteome of cat2 pp2a-b′γ double mutants (Fig. S5). None of the genotypes showed any major changes in the membrane fractions (Fig. S5).
Because neither single mutant showed marked changes in protein profiles, two-dimensional gel analysis was used to identify the phosphoproteins observed in cat2 pp2a-b′γ. This was performed in two successive steps. First, oligomeric protein complexes in total soluble leaf extracts were separated by CN-PAGE, followed by SDS-PAGE in the second dimension (Fig. 7), and stained with ProQ followed by Sypro. To detect the phosphopeptides, the gels were silver stained and the spots of interest were identified by MS and analysis of MS/MS spectra, allowing Ser, Thr and Tyr phosphorylation as possible modifications (Table 2; see Table S3 for technical details).
Figure 7. Proteomic analysis of leaves of the Arabidopsis cat2 pp2a-b′γ double mutant growing in short days. (a, b) Representative Sypro-stained (a) and ProQ-stained (b) two-dimensional gels depicting proteins in the total foliar soluble fractions isolated from wild-type Col-0 and cat2 pp2a-b′γ plants. Oligomeric protein complexes were separated by clear-native polyacrylamide gel electrophoresis (CN-PAGE), followed by sodium dodecylsulfate (SDS)-PAGE in the second dimension. (c–e) Magnified frames from ProQ- and Sypro-stained gels showing areas with protein spots of higher or lower intensity in cat2 pp2a-b′γ relative to Col-0. The protein spots identified are indicated by sequential numbers in each frame, with each number indicating the same protein in cat2 pp2a-b′γ and Col-0. The proteins were identified by in-gel trypsin digestion and subsequent analysis by a nano-high-performance liquid chromatography-electrospray ionization-tandem mass spectrometry (nanoHPLC-ESI-MS/MS) system. Details concerning the identified proteins and phosphopeptides are presented in Table 2.
Download figure to PowerPoint
A number of spots with altered protein and/or phosphopeptide abundance between cat2 pp2a-b′γ and wild-type plants were observed (Fig. 7, Table 2). Proteins with different abundance included S-adenosyl-l-homocysteine hydrolase (SAHH) (spot 3, frame c), glutamine synthetase 1;1 (GLN1;1) and serine:glyoxylate aminotransferase (SGAT)/hydroxypyruvate reductase (spots 1 and 2, frame d), reflecting the marked changes observed in amino acid metabolism (Fig. S5), as well as copper/zinc superoxide dismutase (CSD2) and glutathione S-transferase PHI 2 (GSTF2) (spot 5 and 6, frame e), related to cellular redox homeostasis (Fig. 7, Table 2). These proteins have also been found previously to be affected in the pp2a-b′γ single mutant grown under different light and humidity conditions (130 μmol photons m−2 s−1, 50% relative humidity; Trotta et al., 2011a,b). The two-dimensional gel approach also identified a distinct spot containing AtMDAR2 (Arabidopsis thaliana monodehydroascorbate reductase2) (spot 3, frame d) in cat2 pp2a-b′γ, whereas, in the wild-type, this spot was barely visible (Fig. 7). In agreement with increased PR1 transcripts (Fig. 2) and SA contents (Fig. 5), the double mutant also showed a marked accumulation of the SA-dependent proteins, PR2 and PR5 (spots 4 and 7, frame d). PR2 co-migrated with ACC oxidase 2 (ACO2), an ethylene-forming enzyme, which, like PR2, was not visible in Col-0 extracts (Fig. 7, Table 2).
With the exception of AtMDAR2, all the above-mentioned proteins contained phosphorylated peptides, and were thus identified as phosphoproteins (Table 2). Phosphopeptides specific for cat2 pp2a-b′γ were observed in SAHH and GSTF2, as well as in GLN1;1, PR2, PR5 and ACO2, proteins that were detectable only in the mutant extracts (Fig. 7, Table 2). SGAT, by contrast, contained two phosphopeptides that were detected in the wild-type, but not in cat2 pp2a-b′γ (Table 2).
Among the proteins recognized by ProQ, protein disulfide isomerase 2 (PDI2), Calreticulin 1 (CRT1) and CRT2 (spots 1, 5 and 6, respectively, frame c) showed a marked up-regulation of the protein level in cat2 pp2a-b′γ compared with wild-type plants (Fig. 7). MS revealed no phosphorylated residues in CRT1, whereas PDI1 and CRT2 contained phosphorylated peptides that were present in cat2 pp2a-b′γ, but not in the wild-type (Table 2). An increased phosphorylation level of elongation factor 1-α (EF1α) was also observed, indicative of regulatory changes at the level of protein synthesis in the cytosol as well (Fig. 7, Table 2).
The starch metabolism enzyme ADP glucose pyrophosphorylase large subunit 1 (AGPase) (spot 4, frame c) and an unknown protein At2g37660 (spot 5, frame d) were also detected by ProQ staining, but these proteins contained less phosphopeptides in cat2 pp2a-b′γ than in wild-type plants (Fig. 7, Table 2). In the intensively Sypro-stained spot 5 in frame (d), the protein with highest abundance was a chloroplastic 31-kDa RNA binding protein, but for which no phosphopeptides could be identified (Table 2). Spot 6 in frame (d) contained two phosphoproteins, HSP20-like protein and thiocyanate methyltransferase 1 (HOL1). Our previous study identified the myrosinase TGG1, a glucosinolate-degrading enzyme, as a distinctly up-regulated spot in pp2a-b′γ (Trotta et al., 2011a), and this spot was evident in cat2 pp2a-b′γ (Fig. 7).
Two spots, containing a plastid isoform of triose phosphate isomerase (PDTPI) (spot 1, frame e) and carbonic anhydrase1 (CA1) (spot 3, frame e), appeared to be clearly down-regulated and contained less phosphopeptides in cat2 pp2a-b′γ relative to wild-type plants (Table 2). Although CA1 has been implicated in SA-dependent responses, the triose phosphate isomerase catalyzes the conversion of glyceraldehyde 3-phosphate to dihydroxyacetone phosphate in the Benson–Calvin cycle.
Finally, we also found that CAT2 co-migrated as a high-molecular-weight oligomeric protein complex in the soluble leaf extract of wild-type plants, and also contained a phosphorylated Tyr residue at the C-terminal part of the protein (Fig. 7, spot 2, frame c). In cat2 pp2a-b′γ, the disappearance of CAT2 coincided with a decrease in the co-migrating SGAT (Fig. 7).
To determine whether the increased abundance detected at the protein level was accompanied by similar changes in transcript abundance, qPCR analysis of selected genes was performed. In agreement with the proteomics data (Table 2), GLN1;1 and GSTF2 transcripts were much more abundant in the double mutant than in other genotypes (Fig. 8a). This pattern was not observed for the transcripts of three other proteins found to accumulate in cat2 pp2a-b′γ (Fig. 8a), suggesting that their increased abundance could be related to regulation at the post-transcriptional level.
Figure 8. Quantification of transcripts encoding differentially abundant proteins identified on the two-dimensional gels (a) and genes involved in flowering control or photoperiod signaling (b). Data are for Arabidopsis grown in short days. For explanation of abbreviations, see Supporting Information Table S1. Data are means ± SE of three biological replicates. For ease of expression, FT and CO values have been multiplied by 10 000, and FLC and PHY values by 100. Where significant, * and + indicate difference relative to Col-0 or to cat2 (P < 0.05).
Download figure to PowerPoint
Flowering time and related gene expression in the different genotypes
Photoreceptors and regulation of flowering are among the possible pathways by which the PP2A-B′γ-dependent processes might mediate day length regulation of oxidative stress responses. This possibility was explored in two ways. First, an analysis was undertaken of the time taken to produce flowering stems in LD conditions, which promote the flowering program in Arabidopsis. No marked difference was observed between Col-0 and cat2, but the flowering stem appeared later in both the pp2a-b′γ single mutant and cat2 pp2a-b′γ than in Col-0 and cat2 (Fig. S6). Delayed flowering is in agreement with effects reported for the pp2a-b′γ mutant in a recent study of the role of PP2A subunits in flowering regulation (Heidari et al., 2013). Second, the expression levels of three genes involved in the control of flowering (CONSTANS (CO), FLOWERING LOCUS T (FLT), FLOWERING LOCUS C (FLC)) were determined in SD, conditions in which the pp2a-b′γ mutant plays an important role in controlling the outcome of oxidative stress. This analysis revealed very low expression in all genotypes, and little or only slight differences between them (Fig. 8b). Slightly higher levels of FLC expression were observed in the pp2a-b′γ genotypes, although this was only significant for cat2 pp2a-b′γ (Fig. 8b). These increases were much less apparent than increases in FLC expression reported in pp2a-b′γ in LD (Heidari et al., 2013). By contrast, clear differences were observed in phytochrome expression. In particular, PHYTOCHROME A (PHYA) transcripts were significantly less abundant in both pp2a-b′γ genotypes than in Col-0, with no difference between Col-0 and cat2 (Fig. 8b).