No single way to understand singlet oxygen signalling in plants

Abstract

When plant cells are under environmental stress, several chemically distinct reactive oxygen species (ROS) are generated simultaneously in various intracellular compartments and these can cause oxidative damage or act as signals. The conditional flu mutant of Arabidopsis, which generates singlet oxygen in plastids during a dark-to-light transition, has allowed the biological activity of singlet oxygen to be determined, and the criteria to distinguish between cytotoxicity and signalling of this particular ROS to be defined. The genetic basis of singlet-oxygen-mediated signalling has been revealed by the mutation of two nuclear genes encoding the plastid proteins EXECUTER (EX)1 and EX2, which are sufficient to abrogate singlet-oxygen-dependent stress responses. Conversely, responses due to higher cytotoxic levels of singlet oxygen are not suppressed in the ex1/ex2 background. Whether singlet oxygen levels lower than those that trigger genetically controlled cell death activate acclimation is now under investigation.

Introduction

Reactive oxygen species (ROS) are produced continuously as unavoidable by-products of aerobic metabolism (Halliwell, 2006). Oxygen in its triplet ground state is chemically inert but it can be converted to a ROS either by electron transfer—leading to, by stepwise reduction, the formation of the superoxide radical (O₂^•−), hydrogen peroxide (H₂O₂) and hydroxyl radical (HO^•)—or energy transfer—which leads to the formation of singlet oxygen (¹O₂; Foote, 1968; Gollnick, 1968). During evolution, the biological activities of ROS seem to have undergone several modifications. The continuous release of highly reactive ROS has necessitated the development of ROS scavengers to minimize the cytotoxic effects of ROS in the cell (Apel & Hirt, 2004). At the same time, the detection of rapid changes in ROS concentrations that result from metabolic disturbances or external factors has been used by cells to activate stress-related responses and to readjust their homeostasis (Gechev et al, 2006). Finally, the discovery of genetically controlled ROS production by enzymes such as NADPH oxidases has assigned a surprisingly large diversity of biological activities to ROS. These activities affect a range of processes, including the defence reactions of plants and vertebrates against pathogens (Sagi & Fluhr, 2006), the regulation of cell expansion and development (Foreman et al, 2003; Gapper & Dolan, 2006), and plant–fungus mutualistic interactions (Takemoto et al, 2006; Tanaka et al, 2006).

In plants, chloroplasts and peroxisomes are the main sites of ROS production (Asada, 2006; Foyer & Noctor, 2003). The enhanced generation of ROS in these cellular compartments has been attributed to the disturbance of light-driven photosynthetic electron transport by various environmental factors that trigger stress responses (Niyogi, 1999). Under these environmental conditions, plants are exposed to light intensities that exceed their capacity to assimilate CO₂ and lead to the over-reduction of the electron-transport chain, and ultimately can result in the inhibition of photosynthesis. Plants use two strategies to protect their photosynthetic apparatus: first, the thermal dissipation of excess excitation energy in the photosystem II (PSII) antennae—known as non-photochemical quenching (Muller et al, 2001); and second, the transfer of electrons from PSII to alternative sinks, thereby sustaining partial oxidation of PSII acceptors and preventing photoinactivation of PSII—known as photochemical quenching. Photochemical quenching can be achieved in chloroplasts through, for example, the direct reduction of oxygen to O₂^•− by reduced electron-transport components associated with PSI (Asada, 1999; Rizhsky et al, 2003) and by reactions linked to the photorespiratory cycle that result in the enhanced production of H₂O₂ in peroxisomes (Kozaki & Takeba, 1996). ¹O₂ is continuously produced by PSII through energy transfer from excited chlorophyll to oxygen (Gollnick, 1968; Krieger-Liszkay, 2005). The quenching of ¹O₂ has been linked primarily to the turnover of the D1 protein of the PSII reaction centre (Telfer et al, 1994; Trebst, 2003), although tocopherols and carotenoids also participate in preserving PSII from photoinactivation (Havaux et al, 2005; Krieger-Liszkay & Trebst, 2006; Telfer, 2005).

Plant responses to singlet oxygen

One of the difficulties in studying the biological activity of a particular ROS originates from the fact that several chemically distinct ROS are generated simultaneously in cells under stress, therefore making it almost impossible to link a particular stress response to a specific ROS. However, to study the specific activity of ¹O₂, this problem has been overcome by using the conditional flu mutant of Arabidopsis, which allows the induction of only ¹O₂ within plastids in a non-invasive and controlled manner.

FLU is a nuclear-encoded protein that is tightly associated with plastid membranes. It regulates the Mg²⁺ branch of tetrapyrrole biosynthesis by interacting directly with glutamyl tRNA reductase (Meskauskiene et al, 2001; Meskauskiene & Apel, 2002). In contrast to the wild type, etiolated flu seedlings are no longer able to restrict the accumulation of protochlorophyllide (Pchlide)—the immediate precursor of chlorophyllide (Chlide)—which is a potent photosensitizer that generates ¹O₂ in the light (Fig 1, D; Meskauskiene et al, 2001; op den Camp et al, 2003). As a result, when these seedlings are transferred from the dark to the light they rapidly bleach and die (Fig 1, D→L; Meskauskiene et al, 2001). However, the flu mutant remains viable when it is grown from germination under continuous light (Fig 1, LL). Under these conditions, Pchlide is immediately photoreduced to Chlide and therefore does not accumulate. Provided the flu plants are kept under continuous light, no obvious differences are observed between them and the wild type. This conditional phenotype of the flu mutant has been exploited to study the physiological role of ¹O₂ by growing flu plants initially under continuous light—until they reach the developmental stage of interest—then transferring them to the dark and re-exposing them to light. The release of ¹O₂ in the flu mutant has been determined directly by using the ¹O₂-specific probe DanePy (Kalai et al, 1998; op den Camp et al, 2003). Seedlings of flu bleach and die when they are kept under repeated 16 h:8 h light:dark cycles (Fig 1, L/D), whereas mature plants grown initially under continuous light develop necrotic lesions on their leaves and stop their growth immediately after being moved to daily dark:light cycles (op den Camp et al, 2003).

Figure 1.

Cytotoxicity versus signalling of ¹O₂ in Arabidopsis seedlings. In etiolated flu and executer1/flu (ex1/flu) seedlings, similar excess amounts of free protochlorophyllide (Pchlide) accumulate that, on excitation with blue light, emit a strong red fluorescence (D). In pre-illuminated flu seedlings exposed to 16 h: 8 h light:dark cycles (L/D) inactivation of the EX1 protein abrogates bleaching. However, in etiolated flu seedlings, which accumulate four- to fivefold the amounts of free Pchlide in seedlings grown in continuous light and transferred to the dark for 8 h, the ¹O₂-mediated collapse during re-illumination is not suppressed by the ex1 mutation (D→L). All three plant lines grew equally well under continuous light (LL). Adapted from Przybyla et al (2008) and reprinted with permission from Wiley–Blackwell Publishing Ltd. Wt, wild type.

These stress responses could result either from physicochemical damage caused by ¹O₂ or by the activation of a genetically determined stress-response programme. Initially, we addressed this question by using experiments in which plants were grown under continuous light and then moved to one 8 h dark period before re-exposure to light.

Singlet-oxygen-responsive gene network

Under these experimental conditions, several global gene-expression studies have been carried out with flu plants using Affymetrix GeneChip arrays (Laloi et al, 2006, 2007; Lee et al, 2007; op den Camp et al, 2003). Within 30 min of the release of ¹O₂ in plastids, approximately 300 nuclear genes are rapidly upregulated at least threefold in flu plants relative to the wild type. On the basis of comparative transcriptome analyses, we identified groups of genes that have been shown previously to also be under the control of phytohormones, in particular, abscisic acid, ethylene, methyl jasmonate and salicylic acid (Danon et al, 2005; Ochsenbein et al, 2006). Previous attempts to modulate the triggering of stress responses in flu plants by using phytohormone-related mutants revealed that the ethylene, methyl jasmonate and salicylic acid pathways do have a role in mediating the cell-death response of flu (Danon et al, 2005; Ochsenbein et al, 2006). Furthermore, we identified genes that seem to be more strictly under the control of ¹O₂ (Gadjev et al, 2006; Laloi et al, 2006; op den Camp et al, 2003). In addition, approximately 50 genes encoding putative transcription factors showed a threefold increase in expression within 30 min of the release of ¹O₂. These include ethylene responsive factors, WRKY transcription factors, zinc-finger proteins and several DNA-binding proteins. Many genes involved in putative signal-transduction pathways and calcium regulation, such as protein kinases, calcium and calmodulin-binding proteins, were also identified (Danon et al, 2005; Laloi et al, 2006; Lee et al, 2007; op den Camp et al, 2003). We are now using genetic screens to identify modulators and constituents of the ¹O₂ signalling network. We have searched for ¹O₂-responsive and ¹O₂-specific promoters that, in combination with the luciferase (LUC) reporter gene, can be used in the flu background to monitor ¹O₂-dependent activation of nuclear genes. A LUC reporter gene line has been mutated with ethyl methane sulphonate (EMS), and second-site mutants with a constitutively upregulated LUC expression have been isolated. The preliminary characterization of these mutants supports our view that ¹O₂ signalling does not operate as an isolated linear pathway, but rather merges in a complex signalling network that integrates various cues and relays these to the nucleus (A. Baruah, K. Simkova, C.L. & K.A., unpublished data).

The finding that ¹O₂ can selectively activate nuclear genes, which are not—or only poorly—responsive to O₂^•− or H₂O₂, has also been reported for mammals (Klotz et al, 2003), the unicellular algae Chlamydomonas (Leisinger et al, 2001) and the phototrophic bacterium Rhodobacter sphaeroides (Anthony et al, 2005). In Chlamydomonas, a glutathione peroxidase homologous gene (GPXH) was shown to be strongly induced by ¹O₂-generating photosensitizers, but only weakly induced by O₂^•− or H₂O₂ (Fischer et al, 2005; Leisinger et al, 2001). A second gene, encoding heat-shock protein (HSP)70A, was shown to be induced by both ¹O₂ and H₂O₂. The study of its promoter has revealed that activation by ¹O₂ might require promoter elements that are different from those used by H₂O₂ (Shao et al, 2007). In Rhodobacter sphaeroides, the group IV sigma factor sigma(E) is required to activate a specific transcriptional response to ¹O₂, thereby protecting cells against this ROS (Anthony et al, 2005; Campbell et al, 2007; Glaeser & Klug, 2005). Protein synthesis patterns were also distinct from those after H₂O₂ treatment, but similar to those after high light exposure (Glaeser et al, 2007). Although the response to ¹O₂ is regulated primarily by sigma(E), mutant strain analysis revealed the existence of sigma(E)-independent pathways (Glaeser et al, 2007). In addition, the phrA gene, which encodes a photolyase, is regulated by both ¹O₂ and H₂O₂ in a sigma(E)-dependent manner, but not by treatment with paraquat, which generates O₂^•− that subsequently dismutates to H₂O₂ (Hendrischk et al, 2007).

Interactions of ¹O₂ signalling with other ROS

These results strongly suggest that, in various photosynthetic organisms, ¹O₂ and other ROS affect nuclear gene expression through distinct signalling pathways. These signalling pathways do not always operate independently and might interact with each other. In Arabidopsis, evidence supporting such interactions has been provided by non-invasively modulating the concentration of H₂O₂ in transgenic mutant plants that overexpress the plastid-specific thylakoid-bound ascorbate peroxidase (Murgia et al, 2004). The overexpression of this H₂O₂ scavenger enhanced the intensity of ¹O₂-mediated stress responses in the flu mutant, suggesting that H₂O₂ either directly or indirectly antagonizes ¹O₂-mediated signalling (Laloi et al, 2007). Similarly, pre-treatment of Chlamydomonas reinhardtii with low concentrations of the ¹O₂-generating photosensitizer rose bengal induced a rapid and transient acclimation to subsequently more intense ¹O₂-producing stress but, conversely, increased its sensitivity to paraquat (Ledford et al, 2007). Therefore, ¹O₂-mediated signalling might interact with H₂O₂-derived signals, suggesting that cells not only discriminate between different ROS but also that their responses to a given ROS can be modified by the concomitant perception of a second ROS. This discovery might explain previous observations suggesting that H₂O₂ can protect plants from high light stress. It seems that sensing the relative quantities of different ROS help to fine-tune the response to a specific stress.

Signalling versus cytotoxicity

An extensive second-site mutant screen of flu plants was undertaken to identify suppressor mutants that, during the 8 h dark period, accumulate similar excess amounts of free Pchlide as the original flu line but, on re-illumination, suppress ¹O₂-mediated growth inhibition and cell death of mature flu plants and/or the bleaching of flu seedlings. So far, we have focused our studies on one particular group of suppressor mutants, known as executer (ex). Allelism tests and mapping of the mutated genes revealed that they represent only a single nuclear locus, EX1, which encodes a plastid protein unrelated to known proteins (Wagner et al, 2004). Under non-permissive dark:light conditions the ex1/flu double mutant generates similar amounts of ¹O₂ as the parental flu line, but behaves similarly to wild-type seedlings as they remain viable and do not bleach (Fig 1, L/D), whereas mature plants continue to grow. It seems that because of the ex1 mutation, the flu plants have lost the ability to perceive the presence of ¹O₂ in their chloroplasts and to activate ¹O₂-mediated response programmes. These results show that bleaching of flu seedlings, growth inhibition and cell death of mature flu plants cannot be ascribed to photo-oxidative damage by ¹O₂, but instead reflect the ¹O₂-mediated activation of genetically controlled response programmes that require the activity of the EX1 gene. Enhanced levels of ¹O₂ within plastids of the flu mutant not only trigger marked phenotypic changes but also modulate nuclear gene expression. Indeed, as previously mentioned, within the first 30 min of the release of ¹O₂ rapid changes in nuclear gene expression occur. Inactivation of EX1 attenuated the upregulation of ¹O₂-responsive nuclear genes but did not fully eliminate these changes. A second, related nuclear-encoded protein, known as EX2, has been identified that is also implicated in the changes in signalling of ¹O₂-dependent nuclear gene expression (Lee et al, 2007). On inactivation of EX2 in the flu mutant, additional ¹O₂-responsive genes emerge and genes already upregulated in flu plants are either stimulated further or downregulated. The primary function of EX2 seems to be that of a modulator attenuating and controlling the activity of EX1. When both EXECUTER proteins are inactive in the ex1/ex2/flu triple mutant, most of the ¹O₂-responsive gene transcripts are close to wild-type level (Lee et al, 2007).

Both EX1 and EX2 are plastid-localized proteins that associate with thylakoid membranes (Lee et al, 2007). Therefore, the sensing of ¹O₂ seems to take place within the plastid, in close vicinity to its production sites within plastid membranes of the Arabidopsis flu mutant (Przybyla et al, 2008; op den Camp et al, 2003). In wild-type plants, ¹O₂ can be produced in the reaction centre of PSII and in the antenna system (Krieger-Liszkay, 2005). Whether these different production sites will affect differentially the sensing and signalling of ¹O₂ is not yet clear. Indeed, a diffusion distance significantly larger than what was previously thought has recently been reported for ¹O₂ in rat nerve cells (Skovsen et al, 2005). This indicates that sensing of ¹O₂ might be distant from the production site. In Chlamydomonas cells under high light-stress (3,500 μmol m⁻² s⁻¹), a small fraction of ¹O₂ produced from the PSII reaction centre has been reported to leave the chloroplast and activate the nuclear gene GPXH (Fischer et al, 2007). Whether EXECUTER proteins of Chlamydomonas are involved in mediating the ¹O₂ effect has not yet been addressed.

In our initial experiments, we compared flu and ex1/flu mutants treated with single or repetitive dark periods of 8 h at the rosette leaf or seedling stages. Under these conditions, as described above, introduction of the ex1 mutation suppresses growth inhibition and seedling lethality of the flu mutation. However, etiolated ex1/flu seedlings, which accumulate four- to fivefold the amounts of free Pchlide in flu and ex1/flu seedlings grown in continuous light and transferred to the dark for 8 h, bleach and die in a manner similar to the flu single mutant seedlings when transferred to light (Przybyla et al, 2008; Fig 1, D→L). This indicates that the amount of ¹O₂ produced in flu plants after an 8 h dark:light shift does not reach a cytotoxic level and induces genetically controlled stress programmes, whereas etiolated flu seedlings transferred to light generate cytotoxic amounts of ¹O₂ (Fig 2). In agreement with this idea, measurements of oxidation products of polyunsaturated fatty acids show that flu plants re-illuminated after being kept in the dark for 8 h accumulate enzymatically produced oxylipins, such as 13-hydroxy octadecatrienoic acid (13-HOT), 13-hydroxy octadecadienoic acid (13-HOD), oxophytodienoic acid and jasmonic acid, but not the products of non-enzymatic lipid peroxidation, such as 10-HOD, 12-HOD, 10-HOT and 15-HOT (Przybyla et al, 2008). By contrast, etiolated flu seedlings moved into the light accumulate large amounts of the non-enzymatically formed oxylipins 10-HOD, 12-HOD, 10-HOT and 15-HOT, which indicates that, under these more harsh conditions, toxic effects of ¹O₂ prevail (Fig 2; Przybyla et al, 2008).

Figure 2.

Proposed model of plant responses to various levels of ¹O₂. Constantly changing environmental factors can lead to the generation of variable levels of ¹O₂ that can be mimicked in the flu mutant by modulating the length of exposure to darkness. Increasing dark incubation of flu plants leads to a proportional increase of free protochlorophyllide in the dark and of ¹O₂ generation on illumination. When the production of ¹O₂ is low, an acclimatory response might be activated that allows a plant to survive at subsequent higher levels of ¹O₂. Greater amounts of ¹O₂, for example in flu plants treated with 8 h of darkness and re-illuminated, trigger genetically controlled cell death through EXECUTER-dependent pathways. Most enzymatically oxidized lipid derivatives are produced under these conditions. When the level of ¹O₂ increases further, plants die owing to chemical damage caused by ¹O₂ and a massive production of non-enzymatically oxidized lipid is observed. EX, EXECUTER protein; PCD, programmed cell death.

Perspectives

An interesting remaining question is whether ¹O₂ levels lower than those produced in flu mutants after 8 h of darkness induce qualitatively different changes in gene expression that might lead to acclimation (Fig 2). This question could be addressed by first pre-exposing flu mutants to short dark periods that give rise to low concentrations of ¹O₂ during re-illumination and then applying a much harsher stress treatment. Recent studies of Chlamydomonas indicate that sub-lethal stress leads to a transient or moderate elevation of ¹O₂, and results in the activation of a subset of ¹O₂-responsive genes and protection against a subsequent more severe stress (Ledford et al, 2007). The identification of differentially affected genes under conditions that induce genetically controlled cell death or acclimation will be important for modifying genetic constraints that determine the adaptability of plants to environmental changes.

Chanhong Kim

Rasa Meskauskiene

Klaus Apel

Christophe Laloi