Conditional oxidative stress responses in the Arabidopsis photorespiratory mutant cat2 demonstrate that redox state is a key modulator of daylength-dependent gene expression, and define photoperiod as a crucial factor in the regulation of H2O2-induced cell death
Institut de Biotechnologie des Plantes, Université de Paris Sud XI, 91405 Orsay cedex, France
Photorespiration is a light-dependent source of H2O2 in the peroxisomes, where concentrations of this signalling molecule are regulated by catalase. Growth of Arabidopsis knock-out mutants for CATALASE2 (cat2) in ambient air caused severely decreased rosette biomass, intracellular redox perturbation and activation of oxidative signalling pathways. These effects were absent when cat2 was grown at high CO2 levels to inhibit photorespiration, but were re-established following a subsequent transfer to air. Growth of cat2 in air at different daylengths revealed that photoperiod is a critical determinant of the oxidative stress response. Decreased growth was observed in 8-h, 12-h and 16-h photoperiods, but lesion development was dependent on long days. Experiments at different light fluence rates showed that cell death in cat2 was linked to long days and not to total light exposure or the severity of oxidative stress. Perturbed intracellular redox state and oxidative signalling pathway induction were more prominent in short days than in long days, as evidenced by glutathione status and induction of defence genes and oxidative stress-responsive transcripts. Similar daylength-dependent effects were observed in the response of mature plants transferred from short days in high CO2 conditions to ambient air conditions. Prior growth of plants with short days in air alleviated the cat2 cell-death phenotype in long days. Together, the data reveal the influence of photoperiodic events on redox signalling, and define distinct photoperiod-dependent strategies in the acclimation versus cell-death decision in stress conditions.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Reactive oxygen species availability is regulated by a complex network of antioxidant enzymes (Mittler et al., 2004). Numerous genes encode enzymes involved in H2O2 processing through reduction (peroxidases) or dismutation (catalases), and the specific functions of these genes remain in many cases to be elucidated. Catalase was shown to play a key role in leaf redox homeostasis in barley by Kendall et al. (1983), who identified a catalase-deficient mutant through a ‘photorespiratory’ screen. This finding identified catalase as necessary for metabolism of H2O2 produced by the peroxisomal glycolate oxidase reaction. The screening strategy employed was based on similar principles to that used earlier to isolate photorespiratory mutants in Arabidopsis (Somerville, 1986). Despite the identification of a number of mutants with phenotypes conditional on photorespiratory activity, no catalase mutant was identified in Arabidopsis by this forward genetics approach. Subsequent to the identification of the barley catalase mutant, tobacco antisense lines were produced for the major leaf isoform of catalase, and these lines have proved to be useful systems in which to examine the consequences of increased availability of endogenous H2O2 (Dat et al., 2001). Recent work has applied a similar approach by using antisense and RNAi technology to decrease catalase in Arabidopsis leaves. These lines have been used to assess how high light-induced H2O2 production modulates the transcriptome (Vandenabeele et al., 2004; Vanderauwera et al., 2005).
Like all higher plants studied so far, the Arabidopsis genome contains three catalase genes. Until now no catalase knock-outs were described, and therefore it remained uncertain what the functional importance of the individual catalase genes is in Arabidopsis growth and development (Frugoli et al., 1996). In particular, specific roles of the different Arabidopsis catalase genes in photorespiration have yet to be defined. Therefore catalase knock-out mutants could allow elucidation of the specificity or redundancy of catalase gene function. Here, we report on the analysis of two independent T-DNA insertion lines in which CAT2 gene expression is disabled. We show that CAT2 plays an indispensable role in preventing redox perturbation under ambient air conditions, but that its expression is not necessary for optimal growth or redox homeostasis in the absence of photorespiration. Most importantly, we show that the functional consequences of CAT2 deficiency are conditioned by photoperiod. Cell death is only observed during air exposure over long days. This is linked to photoperiod rather than total light exposure or the degree of oxidative stress, and is associated with weakened induction of defence-related genes compared with air exposure over short days. The data suggest that photoperiodic events define distinct signalling strategies in response to oxidative stress, and that photoperiod-dependent execution pathways and/or withdrawal or failed induction of defence pathways are key mechanisms underlying cell-death pathways in long days.
Identification of cat2 as a photorespiratory mutant
To assess the functional importance of the CAT2 gene in controlling leaf redox state, we identified and obtained two allelic mutants from the SALK collection (Alonso et al., 2003) with T-DNA insertions in the CAT2 coding sequence (Figure 1a). Polymerase chain reaction (PCR) analysis of plants grown from T3 seeds identified T-DNA homozygous plants using specific primers. The two allelic lines were designated cat2-1 and cat2-2, and T4 seeds obtained from identified homozygotes were used for all further analyses. PCR confirmed the presence of the T-DNA and showed that no fragment could be amplified using CAT2-specific primers located either side of the insertion position (Figure 1b). RT-PCR analysis of CAT2 transcripts showed that no signal was detected in the cat2 mutants (Figure 1c). These analyses suggest that both allelic lines represent true knock-outs for CAT2 protein function. Consistent with this notion, enzyme assays showed that both cat2 lines had only 10–15% of wild-type activity, emphasizing the importance of the CAT2 gene in determining high leaf catalase activities (Figure 1d). Despite this marked decrease in extractable catalase activity, no appreciable increase in extractable global leaf H2O2 was measurable (data not shown).
The importance of the CAT2 gene in leaf redox homeostasis was first examined by growing mutants at air levels of CO2 (8-h/16-h day/night regime). Plants were grown at a standard irradiance of 200 μmol quanta m−2 sec−1, which is about 50% saturating for photosynthesis in the Columbia ecotype of Arabidopsis thaliana cultivated in controlled environment growth chambers (Veljovic-Jovanovic et al., 2001). Under these conditions, both cat2 lines showed severely decreased growth relative to the wild type (Figure 2a, upper panel). When plants were grown with high levels of CO2 (about 12 times that of air level), no difference in wild-type and cat2 phenotype and growth was apparent (Figure 2a, middle panel). Decreased growth in air was associated with induction of H2O2-responsive transcripts (Figure 2b, condition 1), whereas this induction was abolished when cat2 was grown at high levels of CO2 (Figure 2b, condition 2). When plants grown in high levels of CO2 were transferred to air for 4 days, accumulation of oxidative marker transcripts resumed (Figure 2b, condition 3). CAT2 function is therefore necessary for redox homeostasis under physiological conditions, and its absence allows cellular redox state to be perturbed in a controlled manner by modifying external CO2 concentration. In agreement with other data on catalase expression patterns (Frugoli et al., 1996; Zimmermann et al., 2006), CAT2 and CAT3 were the most highly expressed genes in Arabidopsis leaves, with CAT1 transcripts being much less abundant. Despite the absence of CAT2 function, no compensatory induction of CAT1 or CAT3 transcripts was observed in either mutant in any of the above conditions (Figure 2c).
Transfer to air of wild-type Arabidopsis grown in high levels of CO2 did not cause changes in any of the three catalase transcripts (Figure 3a). Despite this, total extractable catalase activities were induced twofold (Figure 3b). This increase was absent in cat2 mutant lines (Figure 3b), underscoring the role of CAT2 in photorespiration.
To further explore how redox perturbation linked to CAT2 deficiency affects gene expression, we analysed transcriptomic changes by non-targeted transcript profiling. We performed a cDNA-amplified fragment length polymorphism (AFLP) analysis that monitored the expression of approximately 2200 non-redundant transcripts in wild type and cat2 mutants grown in air, at high levels of CO2, or after transfer to air for 4 days following growth at high levels of CO2. The cDNA-AFLP technique, which essentially consists of the generation of unique restriction fragments from reverse-transcribed mRNA, delivers quantitative gene expression profiles in a rapid and reproducible way. After data processing and statistical analysis, 215 fragments were considered as being differentially expressed. A relevant selection of differentially expressed fragments were excised, and direct sequencing of the PCR products resulted in good-quality sequences for approximately 70% of the fragments. The identities of sequenced fragments are provided throughout the figures. When considering only the wild-type response, adaptive quality-based clustering of differentially expressed transcript fragments revealed three prominent clusters which grouped transcripts that are repressed (48 transcripts) or induced (69 transcripts) by high CO2 levels, and 26 transcripts that are induced after the transfer to ambient air conditions (Figure S1). Hierarchical clustering of expression values of differentially expressed fragments in both wild type and cat2 mutants identified two clusters that are specifically induced in cat2 plants when grown in, or transferred to, air (Figure 4). These genes hence show a similar expression pattern to the H2O2-responsive marker transcripts analysed by qRT-PCR (Figure 2).
Photoperiod is a key modulator of the cat2 phenotype
Interestingly, and in contrast with previously reported high light-induced cell death in catalase-deficient plants (Dat et al., 2003; Vandenabeele et al., 2004; Willekens et al., 1997), and despite the decreased growth and marked induction of H2O2-inducible transcripts (Figures 2 and 4), we did not observe any lesions in cat2 mutants exposed to the above-described conditions (air levels of CO2, 400 μL L−1; short-day regime, 8-h day/16-h night). However, when cat2 was grown at the same irradiance with longer days (12- and 16-h light, respectively), lesions were observed and these were most extensive under a 16-h photoperiod growth regime (Figure 5a). Cell death in cat2 was localized and not associated with overall changes in leaf chlorophyll or protein contents (data not shown). With longer days total catalase activity increased in both wild type and cat2, accompanied in wild-type plants by increased CAT2 transcripts (Figure 5b,c). Despite a small relative increase in longer days, total catalase activity in cat2 remained at 12–17% of wild-type activity for all daylengths (Figure 5c).
The extensive lesions in cat2 grown over long days did not prevent flowering and seed setting, although seed production was severely decreased relative to wild type (data not shown). The cat2 mutant showed a wild-type phenotype when grown at high levels of CO2 in both 12- and 16-h photoperiods, with rosette biomasses that were similar to wild type (data not shown).
Redox-photoperiod interactions in cat2
We explored processes underlying the photoperiod-dependent phenotype of cat2 grown in air. Firstly, redox co-factors and antioxidant redox buffers were measured in the three photoperiods at standard irradiance (Figure 6a). This analysis revealed that NAD and NADP pools remained at least as reduced in cat2 as in wild-type plants under all conditions, suggesting that no generalized oxidation occurs in cat2 in response to increased availability of photorespiratory H2O2. Ascorbate pools were decreased under 8-h and, to a lesser extent, under 16-h photoperiods, although no significant effect was observed with 12-h days. The most dramatic changes occurred in glutathione, a key thiol/disulfide buffer in most cells (Dröge, 2002; Foyer and Noctor, 2005). Under all photoperiods the leaf glutathione pool was more oxidized in cat2 than in the wild type, and this was accompanied by glutathione accumulation. For 12-h days only, increases in glutathione were partly caused by significant increases in the reduced form of glutathione (GSH). The most striking photoperiod-dependent trend in cat2 was a decrease in total glutathione with increasing daylength, which was accompanied by a decrease in the disulphide form of glutathione (GSSG) with photoperiods of 12 and 16 h compared with an 8-h photoperiod. Expression analysis of four H2O2 marker transcripts (OXI1, HSP17.6, GPX6, GSTF8) revealed that upregulation in cat2 depended on daylength in a similar way to glutathione contents: i.e. markedly induced in short days, with all four transcripts progressively less strongly upregulated in cat2 mutants as the growth daylength increased (Figure 6b).
Transcriptomic footprinting by cDNA-AFLP of 5-week-old wild-type and cat2 plants grown with 8-, 12- or 16-h daylengths was performed. Approximately 150 transcript fragments were already differentially expressed between photoperiods in wild-type plants (Figure S2). When transcript profiles of both wild-type and cat2 samples were analysed, 197 of approximately 2500 transcripts tested were differentially expressed, and hierarchical clustering revealed six clusters containing genes in which expression was affected by the cat2 mutation (Figure 7). Four of these were clusters in which photoperiod-dependent expression in wild-type plants was perturbed in cat2. Clusters 1 and 2 contain transcripts that are repressed under all conditions in cat2. Because expression of these genes is photoperiod-dependent in the wild type, repression in cat2 relative to wild type is stronger for long days (cluster 1) or short days (cluster 2). Clusters 3 and 4 contain genes that were induced in a relatively photoperiod-independent manner in cat2. Clusters 5 and 6 contain transcripts in which induction in cat2 depends on photoperiod. The induction of genes contained in cluster 5 is impaired in long-day conditions and is hence similar to H2O2-inducible marker transcripts measured by qRT-PCR in the same conditions (Figure 6b), as well as to glutathione accumulation (Figure 6a), whereas cluster 6 identifies a number of genes in which induction in cat2 requires long-day conditions. Taken together, these data suggest that photorespiratory H2O2-dependent gene expression is significantly influenced by the photoperiod in which plants are growing.
Cell death in cat2 with long days is linked to photoperiod, and not to total daily light fluence or the degree of oxidative stress
To further analyse the influence of photoperiod on lesion formation, we investigated the effects of daily fluence rate on the cat2 phenotype in 8- or 16-h light regimes. Irradiances were adjusted to produce the same total light fluence for short days (8 h at 400 μmol m−2 sec−1) as that received by plants in standard conditions with long days (16 h at 200 μmol m–2 sec−1), or the same total light fluence for long days (16 h at 100 μmol m–2 sec−1) as that received by plants in standard conditions with short days (8 h at 200 μmol m−2 sec−1). No cell death was observed for short days at high irradiance, even though this condition caused severe oxidative stress, as evidenced by decreased ascorbate and marked accumulation of glutathione primarily in the oxidized form (Figure 8, left). In cat2 with long days at 100 μmol m−2 sec−1, glutathione was less dramatically perturbed than for short days at 400 μmol m−2 sec−1. Despite this, visible lesions developed on the leaves of cat2 mutants with long days, and cell death was evident from trypan blue staining (Figure 8, centre). For long days, therefore, appreciable cell death was observed at a fourfold lower irradiance and twofold lower daily fluence rate than in conditions that caused no detectable cell death for short days. Because increases in irradiance from 100 to 400 μmol m−2 sec−1 are expected to markedly increase photosynthetic activity in Arabidopsis plants cultivated in growth chambers, in which photosynthesis does not saturate until about 500 μmol m−2 sec−1 (Veljovic-Jovanovic et al., 2001), these data suggest that cell death in cat2 is not simply the result of daylength-dependent differences in total photorespiratory H2O2 production.
We further analysed the long photoperiod requirement for lesion formation by supplementing short-day conditions at standard irradiance (200 μmol m−2 sec−1) with a lower irradiance for a further 8 h. A supplementary irradiance (30 μmol m−2 sec−1) was chosen that sustains low rates of Calvin cycle activity and, therefore, photorespiratory H2O2 production (Veljovic-Jovanovic et al., 2001) while still being high fluence for some photoperiod signalling pathways (Genoud et al., 1998). Growth of plants over long days at this low irradiance did not cause perturbations in ascorbate or glutathione (Figure 9b, centre), and no cell death was observed (Figure 9a, centre). Although this low irradiance does not in itself cause measurable oxidative stress in cat2 mutants, when it was used to supplement oxidative stress conditions with short days, visible lesions appeared on the leaves and patches of dead cells were evident from trypan blue staining (Figure 9a, right). Cell death induced by supplementary irradiance was not correlated with oxidative stress intensity: in agreement with the data of Figures 6 and 8, accumulation of oxidized glutathione was less dramatic than during growth over short days (Figure 9b, compare left and right panels).
Photoperiod modulates defence signalling during short-term exposure to oxidative stress
Plants experience oxidative stress resulting from changes in environmental conditions that can occur over a relatively short time period (hours or days), and their capacity to initiate appropriate responses is likely to be a key factor in determining the outcome of the stress. To examine the significance of photoperiod during short-term exposure of mature plants to oxidative stress, we took advantage of the conditional photorespiratory cat2 phenotype. Plants were first grown to 5 weeks of age over short days (8-h day/16-h night) at high levels of CO2. In these conditions, cat2 rosette morphology and biomass production were identical to wild type (Figure 2). Plants were then transferred to air, at which point the photoperiod was either maintained at 8 h or increased to 16 h (Figure 10a). To specifically compare cat2 versus wild-type responses, plants were sampled after 3 days exposure to air 4 h into the photoperiod. At this point, lesions on cat2 leaves were minimal with a 16-h photoperiod and absent with an 8-h photoperiod. Over the following week, some cell death appeared on leaves of plants in the 16-h photoperiod, but did not appear on plants transferred to air under an 8-h light period (data not shown). Figure 10b shows that accumulation of glutathione was very marked in short days and was accompanied by dramatic accumulation of the glutathione synthesis precursors, γ-EC and cysteine. In contrast, induction of the glutathione biosynthesis pathway was much less marked in plants transferred to air in long days. cDNA-AFLP analysis of these samples and subsequent clustering of differentially expressed genes, as described above, revealed a set of transcripts with clear differential cat2/wild-type expression depending on the photoperiod during air exposure (Figure 10c). This cluster showed a pattern of stronger induction in cat2 in short days than in long days, and included all but one of the genes identified as more strongly expressed in cat2 continuously grown over short days (Figure 7, cluster 5). These data show that preferential H2O2 induction of glutathione accumulation and gene expression occurs during exposure to oxidative stress in short days, and is independent of gross changes in morphology or developmental stage.
Response of cat2 grown over short days to transfer to long days
Because of the photoperiod effects on cell-death phenotype and gene expression in cat2, we investigated how the cat2 phenotype in long days is modulated by prior exposure to photorespiratory H2O2 production in short days. For this, plants grown for 5 weeks in an 8-h photoperiod were transferred to a 16-h photoperiod (with air levels of CO2 throughout). Growth of plants was monitored for a further 2 weeks (7 weeks after germination), and redox state and gene expression were analysed 1 week after the transfer (6 weeks after germination). Firstly, this analysis showed that whereas the growth rate of the wild type was not affected by transfer to long days, the growth rate of cat2 showed a tendency to increase, and this occurred without appreciable formation of lesions (Figure 11a–c). Reduced ascorbate was decreased in cat2 in both photoperiods, but this was only statistically significant after transfer to long days (Figure 11d). Whereas no significant change in oxidized ascorbate occurred in cat2 in an 8-h photoperiod, the dehydroascorbate pool was significantly decreased after transfer to a 16-h photoperiod (Figure 11d). In an 8-h photoperiod the leaf glutathione pool was markedly increased in cat2, the result of a highly significant increase in GSSG (Figure 11e). After transfer to long days, GSSG remained much higher in cat2 than in the wild type, but the glutathione pool became more reduced than in the 8-h photoperiod, as reflected by a significant increase in the level of GSH compared with the wild-type value (Figure 11e). A cDNA-AFLP analysis revealed that, among transcripts that showed significant differential expression between wild-type and cat2 samples from the two conditions (8-h photoperiod versus 8-h photoperiod + 1 week with a 16-h photoperiod), most showed similar daylength-dependent, but genotype-independent, patterns in wild-type and cat2 leaves (Figure 11f). These included groups of transcripts that were more abundant for short and long days in both genotypes, respectively. Four clusters pointed to significant genotype-dependent effects on gene expression. Cluster 1 contains genes that are induced in cat2, and include several previously found to show short-day-specific induction (see Figure 10). Strikingly, several genes upregulated after transfer of wild-type plants to long days (cluster 2) were much less induced in cat2. Inversely, a number of other genes not strongly induced by the transfer in the wild type were strongly upregulated in cat2 (cluster 4). Cluster 3 groups transcripts in which photoperiod-dependent expression was inversed in cat2 compared with the wild type.
CAT2 encodes the photorespiratory catalase in Arabidopsis
Modelling analyses suggest that photorespiration is one of the highest capacity H2O2-producing pathways in C3 photosynthetic cells (Noctor et al., 2002), and isolation of mutants in barley demonstrated the key role of catalase in the detoxification of H2O2 generated by this pathway (Kendall et al., 1983). However, although a photorespiratory screen (high CO2/air) was used to identify several mutants for the photorespiratory carbon and nitrogen recycling pathway (Somerville, 1986), no Arabidopsis catalase mutant has been reported using this forward genetics approach. More recent studies have extended the number of photorespiratory genes identified in Arabidopsis (Boldt et al., 2005; Igarashi et al., 2003), and the present study adds the CAT2 gene to this list, demonstrating that catalase genes are not functionally redundant in Arabidopsis. The photorespiratory role of CAT2 is evidenced, firstly, by the finding that the cat2 phenotype can be annulled by high levels of CO2 (Figure 2) or by irradiances that drive only slow rates of photorespiration (Figure 9). Secondly, high levels of CO2 have been shown to cause decreases in total catalase activity in tobacco (Havir and McHale, 1989), and the data of Figure 3 show that CAT2 expression is required to observe increases in total catalase activity when plants grown at high CO2 levels are transferred to air. As increased activity was not accompanied by increased CAT2 transcripts, it is likely to have occurred through post-transcriptional regulation, as described in rye (Schmidt et al., 2006). A specific role for CAT2 in processing photorespiratory H2O2 is in agreement with its developmental and cellular expression patterns (Frugoli et al., 1996; Zimmermann et al., 2006). Although the CAT3 gene is highly expressed in leaves, expression is predominant in non-photosynthetic vascular tissues (Zimmermann et al., 2006), and CAT3 is therefore unlikely to play a major role in metabolizing photorespiratory H2O2. In agreement with this, homozygous cat3 insertion mutant lines did not show decreased growth when grown in conditions where these effects were evident in cat2 mutants (data not shown).
Oxidative signalling is activated specifically in photorespiratory conditions in cat2
Catalases are highly expressed high-capacity antioxidative enzymes in leaves, but their activity can be decreased by several types of stress conditions (Schmidt et al., 2006; Volk and Feierabend, 1989). Recently, Arabidopsis lines produced with a cat2 RNAi sequence have been used for transcriptomic profiling in response to high levels of light (Vandenabeele et al., 2004; Vanderauwera et al., 2005). The present study, performed at moderate or low irradiances, shows that absence of CAT2 expression leads to the activation of oxidative signalling specifically under photorespiratory conditions. This does not cause generalized cellular oxidation, as shown by the maintained or even increased reduction state of pyridine nucleotide pools, but it does cause dramatic adjustments in the glutathione pool. These observations, similar to those reported for barley catalase mutants and tobacco antisense lines (Noctor et al., 2002; Smith et al., 1984; Willekens et al., 1997), suggest that H2O2 processing in cat2 occurs through increased engagement of reductive H2O2 processing pathways, which depend on ascorbate and thiols (Davletova et al., 2005; Dietz, 2003; Rodriguez Milla et al., 2003). Enhanced flux through these pathways is associated with activation of oxidative signalling through induction of the OXI1 H2O2-responsive kinase, which is implicated in some pathogen responses and root hair development (Anthony et al., 2004, 2006;Rentel et al., 2004). Because of this, the cat2 mutants represent a model system in which to study oxidative signalling strength and the impact of redox perturbation. Although our data show that CAT2 is necessary for optimal growth and seed production under conditions of active photorespiration, absence of CAT2 function is not lethal under conditions of moderate growth irradiance. Extensive lesions were observed for long days, but this did not prevent flowering, and no lesions were observed for short days.
Glutathione is implicated in redox homeostasis, and signalling linked to meristem growth and plant interactions with symbionts and pathogens, as well as in stresses such as ozone, cold and excess light (Ball et al., 2004; Bick et al., 2001; Frendo et al., 2005; Gomez et al., 2004; May et al., 1996; Parisy et al., 2007; Sen Gupta et al., 1991; Vanacker et al., 2000; Vernoux et al., 2000). The accumulation of GSSG and total glutathione observed in cat2 mimics effects observed in some of these stress conditions. Signalled activation of the glutathione synthesis pathway in cat2 probably occurs to promote maintenance of GSH concentration and/or glutathione redox potential under conditions that lead to increases in GSSG (Meyer and Hell, 2005; Noctor, 2006). Dynamic changes in glutathione status could influence cell function through multiple mechanisms, including glutathionylation reactions, calcium signalling and activation of transcription factors (Dixon et al., 2005; Evans et al., 2005; Ito et al., 2003; Meyer and Hell, 2005; Michelet et al., 2005; Noctor, 2006). It is striking that perturbation of the glutathione pool was most dramatic in cat2 with short-day conditions, where no lesions were observed. For long days, accumulation of GSSG and the response of glutathione synthesis were damped, despite the exposure of plants to photorespiratory H2O2 for a greater length of time. Although stress-induced changes in both glutathione and ascorbate status have been shown to correlate with cell viability in plants (Kranner et al., 2002, 2006), the present data indicate that there is no simple relationship between glutathione oxidation or tissue GSSG content and cell death, and that plant cells are able to tolerate marked accumulation of GSSG. Thus, although intracellular thiol oxidation may promote the decreased rosette biomass observed in cat2 grown in air at all photoperiods, it is not sufficient to elicit cell death. From work on the catalase-deficient barley mutant, it was concluded that GSSG accumulation occurred both within and outside the chloroplast (Smith et al., 1985). It remains to be established whether this is also the case in the Arabidopsis cat2 mutant, or whether GSSG accumulates preferentially in specific compartments under some conditions.
Coordinated induction of genes encoding metabolite modification enzymes in cat2
Transcriptomic analysis revealed that the best-represented functional class of strongly induced genes in cat2 encoded enzymes involved in metabolite modification. Glutathione-S-transferases (GSTs) are encoded by a superfamily of more than 40 genes in Arabidopsis (Wagner et al., 2002). Certain GSTs, particularly those of the phi family, are active against organic peroxides, and are H2O2-inducible (Chen et al., 1996; Kovtun et al., 2000). A phi-family GST (GSTF8) was induced up to threefold in catalase mutants in air. GSTU24, a GST of the tau family, was much more rapidly and dramatically induced. Representative GSTs of this family were shown to catalyze conjugation (transferase) reactions but are much less active as peroxidases, and GSTU24 is among several GSTs induced by herbicide and xenobiotic treatment (Mezzari et al., 2005; Wagner et al., 2002). This suggests that induction of this gene could be linked to conjugation of secondary metabolites or stress-induced catabolites. Consistent with an important role for metabolite conjugation in the H2O2 response, three UDP-glycosyl transferases were among other genes most strongly induced in cat2. Certain UDP-glycosyl transferases (UGTs) are known to play a role in the H2O2 and pathogen response (Langlois-Meurinne et al., 2005), although in many cases the specific targets of different Arabidopsis enzymes are yet to be well characterized. One of the UGTs strongly induced in cat2 has been shown to be involved in callose deposition at the phragmoplast (Hong et al., 2001). Other induced genes in cat2 encoded a putative cytochrome P450 and, intriguingly, a lactamase. Lactamases are important enzymes in one mechanism of bacterial resistance against antibiotics such as penicillin (Thomson and Bonomo, 2005). Little is known about lactamase functions in plants, but lactam metabolites include anti-herbivore compounds the biological activity of which can be modulated by hydroxylation or glycosylation (Bailey and Larson, 1991; Baumeler et al., 2000). Another lactam metabolite is produced during chlorophyll degradation, when pheophorbide a is converted to red chlorophyll catabolite by monooxygenase activity (Hörtensteiner et al., 1998).
Redox modulation of photoperiod-dependent expression and development
The conditional photorespiratory phenotype of cat2 enabled us to examine the interaction between endogenous ROS availability and daylength in conditions in which the cat2 phenotype (decreased growth and/or cell death) is expressed or not at the morphological level. These experiments demonstrate that ROS availability modulates the expression of genes that are modulated by daylength in the wild type. This effect was particularly evident in the experiments of Figures 7 and 11, suggesting that redox effects on photoperiod signalling are closely related to ROS modulation of the developmental programme. Of particular interest are the changes in senescence-associated genes. ROS and associated systems are implicated in the regulation of senescence (Navabpour et al., 2003; Zimmermann et al., 2006). A lipase/esterase-thioesterase induced in cat2 is similar to an ethylene-inducible lipase described in Citrus sinensis (Zhong et al., 2001), and its expression pattern closely matched that of a putative ACC oxidase (Figure 11). Similarly, NITRILASE4 has strong specificity for β-cyano-l-alanine and is induced in senescing leaves, a function that may be important in detoxification of cyanide produced in ethylene biosynthesis (Piotrowski et al., 2001). Genes encoding a protein similar to SAG102, and a branched chain amino acid transaminase possibly involved in senescence (Taylor et al., 2004), were strongly induced in the wild type after transfer to long days, but were much less induced in cat2 (Figure 11). The observations in cat2 suggest that excess availability of intracellular ROS produces partial inversion of senescence programmes. Although some ethylene-associated genes were induced in cat2 in short days, elements of the senescence programme induced in the wild-type by transfer to long days were downregulated in cat2.
Photoperiod is a key determinant of the response to oxidative stress
Many genes that are most strongly induced in cat2 are also induced by salt, high light, ozone and pathogens (http://www.genevestigator.ethz.ch). As noted above, many of these genes may be crucial in modulating the strength of ROS-induced metabolite signals. Metabolites known to have important interaction with ROS in controlling redox homeostasis and cell death include lipid peroxides, salicylic acid and related molecules, chlorophyll catabolites and other secondary metabolites (Lamb and Dixon, 1997; Yao and Greenberg, 2006). In addition to regulation of their production, the biological activity of these compounds is controlled by metabolism and/or conjugation through enzymes such as thioredoxin-dependent peroxidases, GSTs with peroxidase or conjugase activity, UGTs, reductases and oxidases/hydroxylases. It is striking that genes induced in cat2 encoding such enzymes were found to be strongly induced in short days and were much less upregulated in long days, when cell death occurs. This impaired upregulation with long days was observed when the photoperiod influence was examined in plants grown from germination in air, and also in short-term experiments in which plants were exposed to oxidative stress after prior growth in high levels of CO2 to eliminate possible effects of differences in developmental stage or phenotype. That these genes showed a similar expression pattern to oxidative activation of the glutathione synthesis pathway suggests that long-day plants such as Arabidopsis possess distinct, coordinated photoperiod-determined strategies in response to stress. With short days, increased H2O2 availability elicits potent redox signalling that is linked to homeostasis and acclimation, whereas in long-day conditions, homeostatic and defence responses are muted. Thus, photoperiod modulates the role of H2O2 in stress signalling, and so dictates different strategies in response to stresses that involve an oxidative component. It might be noted that redox state was least affected by the cat2 mutation in intermediate daylengths (Figure 7a). This suggests that there could be an optimum photoperiod for redox homeostasis in Arabidopsis, although maintenance of a comparatively reduced state did not prevent some cell death in a 12-h photoperiod. Furthermore, cell death showed a quantitative dependence on photoperiod (Figure 5) but not on total light fluence (Figure 8). As daylength is a critical determinant of the flowering transition in Arabidopsis (Putterill et al., 2004), different strategies in response to oxidative stress could be hardwired into the genetic control of vegetative growth versus reproductive programmes. It is possible that oxidative stress responses during the vegetative stage are directed towards survival, even at the expense of slower growth, whereas in conditions that promote flowering cell death programmes become activated.
Short day-dependent acclimation of cat2 to long days
The Arabidopsis LESION SIMULATING DISEASE 1 (lsd1) mutant lacks a functional zinc-finger protein that is required to prevent the spread of lesions in long days (Dietrich et al., 1994). As in cat2, lesions do not form in short days. Also similar to cat2, the phenotype of lsd1 is aggravated by active photorespiration (Mateo et al., 2004). However, key differences between the two mutants exist. Firstly, although lsd1 has lower catalase activities than wild-type Arabidopsis, the decrease is much less marked than in cat2, and is linked to changes in CAT1 expression (Mateo et al., 2004). Secondly, lsd1 (a kind gift from M. Torres and J. L. Dangl, University of North Carolina) showed little or no phenotype when grown in our growth chambers over short days, conditions in which growth was markedly decreased in cat2. Thirdly, this mutant shows spreading cell death on transfer from short to long days (Dietrich et al., 1994). In contrast, on transfer to long days following previous growth over short days in air, cat2 grew at similar rates to the wild type, and lesions only slowly appeared on the very oldest leaves (Figure 11). This response was associated with cat2-specific induction of a nucleoside diphosphate kinase involved in pyrimidine nucleotide synthesis, lack of induction of senescence-associated genes, and with maintenance of the expression of several genes and high glutathione levels induced by pre-exposure to H2O2 in short days (Figures 11). These effects could be related to mechanisms that drive acclimatory resistance in plants exposed to sublethal doses of the superoxide generator, methyl viologen (Vranováet al., 2002), and the absence of lesions on the youngest and newly-produced leaves suggest that systemic signals are involved.
The photoperiod–H2O2 signalling interaction in the control of acclimation versus cell death
An influence of photoperiod on the redox responses of wild-type plants to excess light and low temperature has previously been noted, pointing to an interaction between daylength and redox-mediated acclimation signals (Becker et al., 2006). The present data provide no evidence that acclimation or death in cat2 is simply related to the severity of oxidative stress or extent of redox perturbation. They reinforce current concepts that the different effects of ROS on plant cells are not mediated through damage or indiscriminate oxidation, but occur through interactions with signalling pathways (Foyer and Noctor, 2005). Specifically, our observations point to an important role for the photoperiod context in which plants are exposed to increased ROS availability. At least two non-exclusive mechanisms may explain the photoperiod-dependence of acclimation versus cell death (Figure 12). Absence of cell death in short days, even in the presence of marked redox perturbation, could be caused by the stronger upregulation of defence mechanisms and death suppressors than in long days (Figure 12, right). Alternatively, or in addition, cell death in long days could be mediated through an execution pathway that is absent or less active in short days (Figure 12, left). Other work has reported spontaneous lesion formation in mutants for the psi2 gene, which encodes a negative regulator of phytochrome signalling, and demonstrated interactions between salicylic acid signalling and phytochrome-dependent pathways (Genoud et al., 1998, 2002). Leaf bleaching in response to singlet oxygen is much diminished in the absence of EXECUTOR1 expression (Wagner et al., 2004). The present work raises the intriguing possibility that elements of photoperiod signalling pathways could play analogous roles to EXECUTOR1 in cell death promotion in response to excess H2O2.
Two A. thaliana allelic mutant lines carrying T-DNA insertions in the CAT2 gene (At4g35090) were identified using insertion mutant information obtained from the SIGnAL website at http://signal.salk.edu, and seeds were obtained from the Nottingham Arabidopsis Stock Centre (http://nasc.nott.ac.uk). In accession number SALK_057998.55.00.x the T-DNA is in the third exon from the 5′ end and homozygotes were designated cat2-1. Accession number SALK_076998.55.75.x carries the T-DNA in the fourth exon from the 5′ end and homozygotes were designated cat2-2. After identification of homozygotes as described below, all further analyses were performed on plants grown from T4 seeds.
Identification of homozygous cat2 knock-out mutants
Leaf DNA was amplified by PCR (30 sec 94°C, 30 sec 60°C, 1 min 72°C, 30 cycles) using primers specific for left T-DNA borders and the CAT2 gene (Table S1). For the left border, primers were NewLB1 and sensCAT2. Fragments obtained were sequenced to confirm the insertion site. Zygosity was analysed by PCR amplification of leaf DNA using NewLB1 with sensCAT2 (alleles carrying T-DNA), and sensCAT2 with revCAT2 or revbisCAT2 (wild-type allele).
Plant growth and sampling
Seeds were incubated for 2 days at 4°C and then sown in soil in 7-cm pots in a controlled environment growth chamber at the specified photoperiod at (unless stated otherwise) an irradiance of 200 μmol quanta m−2 sec−1 at leaf level, 20°C/18°C, 65% humidity, and given nutrient solution twice per week. The CO2 concentration was maintained at 400 μL L−1 (air) or 4500 μL L−1 (high CO2 levels). Samples were rapidly frozen in liquid nitrogen, and stored at −80°C until analysis. Each sample was either a mixture of material from two different leaves (enzyme and metabolite assays) or a pool of six leaves from different plants (transcript analysis and metabolite analyses of Figures 8 and 9). All samples were taken 4 h into the photoperiod, except for condition 3 of Figure 4. In this case, to allow analysis of short-term changes in transcript profiles, samples were also taken 7 h into the photoperiod, at 5 h after transfer of plants to air, which was performed 2 h into the photoperiod.
RT-PCR and qRT-PCR analysis
RNA was extracted using the kit NucleoSpin® RNA plant (Macherey-Nagel; http://www.mn-net.com) and reverse transcribed with the SuperScript™ First-Strand Synthesis System (Invitrogen, http://www.invitrogen.com). Semi-quantitative PCR analysis of CAT2 transcripts was performed using primers 3 and 5 (Table S1). ACTIN2 transcripts were measured as a control using primers 6 and 7 (Table S1). cDNAs were amplified using the conditions described above for analysis of genomic DNA except that the number of cycles was 25.
For quantitative real-time RT-PCR, a fluorescence threshold value of 0.2 (t) was chosen to allow all transcripts to be measured within the linear range of amplification. For each couple of oligonucleotide primers (Table S1), control experiments performed on genomic DNA dilutions from 6.25 pg to 102.4 ng showed that cycle threshold (Ct) was linearly proportional to DNA quantity within this range (corresponding to between 22 and 36 cycles). Only primer couples for which amplification occurred with an efficiency (E) of between 0.91 and 1.09 were used. Following amplification, products were denatured by heating from 60 to 95°C to check amplification specificity (a unique peak shows a unique amplification product). Results were processed with GeneAmp® 5700 sds Software to obtain Ct. Values were normalized by an internal reference (Ctr) according to the equation ΔCt = Ct – Ctr, and quantified as 2–ΔCt. A second normalization by a control (ΔCtc) (wild-type) ΔΔCt = ΔCt –ΔCtc produces a relative quantification: 2–ΔΔCt (Czechowski et al., 2004). To analyse transcripts, leaves were ground in liquid nitrogen and total RNA was extracted using the NucleoSpin® plant RNA kit (Macherey-Nagel). RNA (1 μg) was reverse-transcribed using the SuperScript™ First-Strand Synthesis System for RT-PCR (Invitrogen) and real-time PCR analysis was performed in duplicate on 1 ng cDNA using primers at 250 nm, 12.5 μl qPCR™ Mastermix Plus for SYBR® Green I (Eurogentec, http://www.eurogentec.com), final volume 25 μl, in a GeneAmp® 5700 thermocycler (Perkin-Elmer, http://www.perkinelmer.com). PCR conditions were 2 min at 50°C, 10 min at 95°C, and 40 cycles of 15 sec at 95°C and 1 min at 60°C.
cDNA-AFLP analysis was carried out in eight different conditions of CO2 concentration and photoperiod. For each condition, two wild-type samples were compared with a sample of each knock-out mutant line. Each sample consisted of pooled material from at least six different plants. Templates for cDNA-AFLP were prepared as described by Vuylsteke et al. (2007). Twenty-four randomly chosen primer combinations (BstT1-Mse12, BstT1-Mse14, BstT1-Mse21, BstT1-Mse24, BstT2-Mse42, BstT3-Mse14, BstT3-Mse21, BstT3-Mse33, BstT3-Mse44, BstT4-Mse11, BstT4-Mse12, BstT4-Mse13, BstT4-Mse14, BstT4-Mse21, BstT4-Mse22, BstT4-Mse23, BstT4-Mse31, BstT4-Mse32, BstT4-Mse33, BstT4-Mse34, BstT4-Mse41, BstT4-Mse42, BstT4-Mse43, BstT4-Mse44) were used and amplification products were separated on polyacrylamide gels. Gels were dried, exposed to Kodak BioMax MS films, and scanned in a PhosphorImager 445 SI (Amersham Biosciences, http://www.amersham.com). Gel images were processed with AFLP QuantarPRO (Keygene Products, http://www.keygene.com), a software application that allows accurate quantification of band intensities in DNA fingerprints. The intensity of around 2200 individual bands was determined and normalized for variations such as running or loading differences. The obtained raw expression data were further processed via the ARRAY-AN application, as described in Vandenabeele et al. (2003). To select for genes that displayed significant differences in means across the different groups of samples, analysis of variance (anova; P < 0.05) was used. A coefficient of variation (CV), calculated as the SD on all values of the different samples divided by the average expression of the different samples, was utilized as a selection criterion for differential expression. Differentially expressed transcript tags with a CV value higher than 0.5 were used for hierarchical clustering. All clustering figures contain the mean values calculated via anova. False-discovery rates were calculated from F distribution-based P-values using the q-value software implemented in the R statistical package (Storey and Tibshirani, 2003).
A selection of differentially expressed fragments was excised from the gels and reamplified by PCR with their respective selective cDNA-AFLP primers. In addition, the identity of approximately 40 fragments was obtained by comparing the generated gel patterns with those of gels obtained in previous experiments (Vandenabeele et al., 2003; Vanderauwera et al., 2005).
Trypan blue staining
Dead cells were stained in detached rosette leaves by a method modified from Wäspi et al. (2001). Leaves were incubated overnight at 37°C in lactophenol/trypan blue solution, prepared by dissolving 10 mg trypan blue and 10 g phenol in 10 ml distilled water, 10 ml glycerol and 10 ml lactic acid. Leaves were then incubated in chloral hydrate solution to remove pigments. Destained leaves were photographed under a magnifying glass or using differential interference contrast microscopy.
Approximately 200 mg leaf material was ground in liquid nitrogen, 100 mg insoluble PVP was added and the powder was extracted into 1.5 ml 0.1 m NaH2PO4 (pH 7.5), 1 mm EDTA. Samples were taken for chlorophyll analysis and the remainder centrifuged for 10 min. An aliquot of the supernatant was loaded onto NAP-5 columns and the desalted eluant used for enzyme assays. Catalase was assayed as previously described (Veljovic-Jovanovic et al., 2001). Protein was assayed using the BIORAD Bradford method. Chlorophyll was measured in 80% acetone at 663 and 645 nm.
Oxidized and reduced forms of glutathione, ascorbate, NAD and NADP were measured by plate-reader assay, and thiols were measured by HPLC separation of bimane derivatives followed by fluorescence detection, as described in Queval and Noctor (2007).
We thank the Salk Institute Genomic Analysis Laboratory for providing the sequence-indexed Arabidopsis T-DNA insertion mutants. Funding for the SIGnAL indexed insertion mutant collection was provided by the National Science Foundation. We thank Rokhaya Leye, Mélanie Béraud, Dorothée Thominet and Fanta Ouédraogo for technical assistance and the Nottingham Arabidopsis Stock Centre, UK, for supply of seed stocks. We are grateful to Claire Gachon and Mathilde Langlois-Meurinne for advice on qRT-PCR analysis, Roland Boyer for photography, Séverine Domenichini for microscopy, and Jean-Louis Prioul for help with statistical analysis. This work was partly supported by the Research Fund of Ghent University (Geconcerteerde Onderzoeksacties no. 12051403), the ‘Fonds voor Wetenschappelijk Onderzoek-Vlaanderen’ (no. G.0350.04), and the French Agence Nationale de la Recherche-GENOPLANTE programme ‘Redoxome’ (no. GNP0508 G).