Singlet oxygen and photo-oxidative stress management in plants and algae

Authors

  • HEIDI K. LEDFORD,

    Corresponding author
    1. Department of Plant and Microbial Biology, University of California, Berkeley, CA 94720–3102, USA
      Heidi K. Ledford. E-mail: hledford@nature.berkeley.edu
    Search for more papers by this author
  • KRISHNA K. NIYOGI

    1. Department of Plant and Microbial Biology, University of California, Berkeley, CA 94720–3102, USA
    Search for more papers by this author

Heidi K. Ledford. E-mail: hledford@nature.berkeley.edu

ABSTRACT

Photosynthetic organisms constantly face the threat of photo-oxidative stress from fluctuating light conditions and environmental stress. Plants and algae have developed an array of defences to protect the chloroplast from reactive oxygen species. Genetic and physiological studies have shown that antioxidant responses are important to high-light acclimation, both by directly scavenging or quenching reactive oxygen intermediates and by contributing reducing power for alternative electron transport pathways and excess energy dissipation. At present, the signalling events leading to up-regulation of antioxidant defences in high light remain a mystery. Recent advances toward understanding acclimation to oxidative stress in both photosynthetic and non-photosynthetic model organisms may illuminate how plants and algae respond to high-light stress. Although the role of hydrogen peroxide in high-light acclimation has been investigated, less is known about responses to singlet oxygen, a form of reactive oxygen that poses a significant threat specifically to photosynthetic organisms. This review will discuss some intriguing new findings in that area, focusing on recent findings regarding the nature of singlet-oxygen responses in the chloroplast.

Abbreviations
ROS

reactive oxygen species

1O2*

singlet oxygen

Chl

chlorophyll

LOOH

lipid hydroperoxide

H2O2

hydrogen peroxide

HL

high light

LL

low light

PS

photosystem

Pchlide

protochlorophyllide

npq1

non-photochemical quenching-1 mutant

lor1

loroxanthin-1 mutant

cbn1

chlorophyll b-less mutant

Yap1

S. cerevisiae yeast activator protein-like protein

Gpx3

S. cerevisiae glutathione peroxidase 3 protein

DCMU

3-(3,4-dichlorophenyl)-1,1-dimethylurea

Gpxh

C. reinhardtii glutathione peroxidase gene

FLU/flu

Arabidopsis thaliana fluorescent PROTEIN/mutant

EX/EX/ex

A. thaliana executer PROTEIN/GENE/mutant.

INTRODUCTION

The combination of highly reactive intermediates and an oxygen-rich environment is necessary for photosynthesis, but can be lethal when the balance between the rate of light collection and light energy use is perturbed. Unfortunately, nearly every imaginable environmental change can upset this balance, resulting in the production of reactive oxygen species (ROS) (reviewed in Asada 1996). Left unchecked, photo-oxidative stress can lead to loss of protein function, membrane integrity, and eventual cell death. A carefully orchestrated network of defences protects chloroplasts from oxidative stress and allows plants and algae to rapidly acclimate to changing light conditions (reviewed in Golan et al. 2004).

Although high levels of ROS are deadly, sublethal levels of ROS can serve as signals, prompting cells to prepare for sustained oxidative stress. These warning systems and their receptors have been well characterized in microbial model systems such as the budding yeast Saccharomyces cerevisiae, and can activate defences that are specific to the ROS signal (recently reviewed in Ikner & Shiozaki 2005). Despite the tight link between photo-oxidative stress and ROS, little is known about how photosynthetic organisms sense and respond to oxidative signals. This review will summarize some of what is known about responses to ROS during photo-oxidative stress, focusing on recent findings regarding singlet oxygen (1O2*) – a highly reactive form of oxygen that is of particular consequence to photosynthetic organisms.

THE USUAL SUSPECTS: REACTIVE OXYGEN SPECIES IN THE CHLOROPLAST

Oxidative stress in the chloroplast comes in a variety of forms and originates from several locations. Protein-bound chlorophyll (Chl) molecules in thylakoid membranes are potent photosensitizers (Foote 1976). Environmental stress that upsets the balance between light harvesting and energy utilization lengthens the lifetime of 1Chl*, increasing the likelihood that 1Chl* will undergo intersystem crossing to form 3Chl* (van Mieghem et al. 1995; Noguchi 2002). 3Chl* is longer-lived than 1Chl*, and reacts more readily with ground-state 3O2. The physical interaction between 3Chl* and oxygen produces 1O2*, liberating oxygen from the spin restriction that normally limits its reactivity with singlet-state biological molecules (Green & Hill 1984). 1O2* is highly reactive and can modify lipids (Girotti & Kriska 2004), nucleic acids (Martinez et al. 2003), and proteins (Davies 2004).

1O2-specific fluorescent probes have detected 1O2* in leaves exposed to high light (HL) (Hideg et al. 1998; Fryer et al. 2002), and some 1O2* is probably generated in photosystem (PS) II even in the absence of obvious environmental stress (Keren, Gong & Ohad 1995). While precise estimates of 1O2* lifetimes in vivo vary, the general agreement is that 1O2* is too reactive to travel far from its site of origin before reacting with another biological molecule (Moan 1990). 1O2* generated in the chloroplast is therefore most likely to directly affect lipids and membrane proteins located near the site of 1O2* production. This places the reaction centre of PSII in particular jeopardy, and there is evidence that 1O2* contributes to photodamage at the reaction centre of PSII, leading to photoinhibition (Mishra & Ghanotakis 1994). Severe 1O2* stress may also inhibit the repair of the D1 protein in the PSII reaction centre (Nishiyama et al. 2004). The result is an accelerating cycle of damage in which 1O2* -mediated impairment of PSII increases the likelihood of more 1O2* production in the antenna (Hideg et al. 1998).

1O2* can add directly to the double bonds of polyunsaturated fatty acids to form lipid peroxides (LOOH) (Girotti & Kriska 2004). Other oxygen radicals, including LOOH, can also trigger lipid peroxidation via a number of mechanisms, the best-studied of which is hydrogen abstraction followed by addition of oxygen to the resulting carbon radical (Porter 1984). In either case, close packing of polyunsaturated fatty acid molecules in the chloroplast membrane favours the initiation of lipid peroxidation chain reactions that emanate from the initiation site and propagate until terminated by antioxidants or substrate depletion (Girotti & Kriska 2004). LOOH are more polar than their precursors and can eventually compromise membrane integrity. Susceptibility of a membrane to lipid peroxidation is enhanced by the presence of polyunsaturated lipids (Alexander-North et al. 1994) – a hallmark of chloroplast membranes.

1O2* stress is more likely to occur in PSII, where the lifetime of 1Chl* is generally longer than in PSI, but PSI faces problems of its own. High reduction potential on the acceptor side of PSI has been hypothesized to reduce oxygen to superoxide, particularly when the concentration of carbon dioxide is limiting, although measurements of oxygen consumption in this manner are often indirect (Radmer, Kok & Ollinger 1978; Asada 2000). The same occurs in vitro at the acceptor side of PSII (Pospíšil et al. 2004). Superoxide accumulation in HL has been observed in mesophyll tissue (Fryer et al. 2002), but superoxide is rapidly converted to hydrogen peroxide (H2O2) either spontaneously or with the aid of superoxide dismutases (Asada 2000). H2O2 itself is a relatively stable molecule that can travel freely across membranes. This means that H2O2 would be able to diffuse throughout the cell, spreading damage but possibly also signalling stress and recruiting defences (see Bechtold et al. 2005).

H2O2 can also lead to the production of hydroxyl radical, one of the most reactive forms of ROS. Hydroxyl radical can be formed from H2O2 via the Haber–Weiss reaction, catalysed by iron (Imlay & Linn 1988). In fact, the destructive properties of H2O2 and superoxide result primarily from their role in hydroxyl radical production rather than from direct damage (Imlay & Linn 1988). H2O2 and superoxide have both been detected in plant leaves following HL treatment (Fryer et al. 2002).

Although it is often difficult to distinguish cause and effect in the complex oxidative landscape of the chloroplast, to treat all ROS as the same problem is to miss the specificity underlying their localization and chemical reactivity. H2O2, for example, is less reactive than 1O2* or hydroxyl radicals, and may travel further from its point of origin. H2O2 could potentially move systemically to signal stress and activate defences, whereas 1O2* and hydroxyl radicals are not likely to progress far in the crowded interior of a cell before reacting with another molecule. A sensor for 1O2* would therefore need to be located near the site of 1O2* generation to provide optimal sensitivity and specificity. Superoxide is a charged molecule, and consequently has difficulty travelling across membranes. Compartmentalization of superoxide production and defence responses is therefore especially important. Lipid hydroperoxides are more likely to damage a membrane than superoxide, but may also impact the aqueous compartment by depleting thiols and affecting redox-buffering capacity (Morgan, Dean & Davies 2004).

Recent studies in yeast emphasize the specificity of oxidative-stress responses to individual forms of ROS. Analysis of deletion strains that were more susceptible to oxidants including H2O2, diamide (which generates superoxide), and cumene hydroperoxide (an organic peroxide that can trigger lipid peroxidation) showed that while there were some overall functions that are necessary for general oxidative stress responses, many responses were remarkably specific to the form of oxidative stress applied (Higgins et al. 2002; Thorpe et al. 2004). For example, the ability to repair damaged DNA and to degrade damaged proteins was necessary for wild-type levels of resistance to all forms of oxidative stress tested (Higgins et al. 2002). However, deletion strains deficient in mitochondrial electron transport were more sensitive to H2O2, whereas mutants deficient in lipid and carbohydrate metabolism were more sensitive to lipid peroxidation (Thorpe et al. 2004). Such differences mean that while there are general responses to oxidation, dissecting the tangled web of ROS defences also requires addressing the specificity of the ROS trigger.

OXIDATIVE-STRESS MANAGEMENT

Through over 2 billion years of fine-tuning (Des Marais 2000), oxygenic photosynthetic organisms have developed a panoply of protective mechanisms both to prevent the formation of ROS and to detoxify ROS once they have formed. A plant suddenly exposed to HL, for example, may reduce the amount of light absorbed by changing the orientation of leaves or chloroplasts (Kasahara et al. 2002) or by shrinking PSII antenna size (Escoubas et al. 1995). Concurrently, alternative electron transport pathways may redirect excess electrons either to water via the water–water cycle (Asada 2000; Rizhsky, Liang & Mittler 2003) or to oxygen via a chlororespiratory (Cournac et al. 2002; Aluru & Rodermel 2004) or photorespiratory pathway (Wingler et al. 2000; Haupt-Herting & Fock 2002). By capitalizing on these pathways, the plant may prevent a build-up of electrons in the electron transport chain by providing alternative electron acceptors when carbon dioxide is limiting. In addition, non-photochemical energy dissipation mechanisms are rapidly activated to thermally quench excess absorbed energy before it is used to initiate electron transfer (Holt, Fleming & Niyogi 2004).

HL-stressed plants also often accumulate a variety of small-molecule and enzymatic antioxidants that can detoxify ROS. Water-soluble antioxidants including ascorbate and glutathione scavenge H2O2 and LOOH, serving both as cofactors for peroxidases and as direct reductants. Lipid-soluble antioxidants such as carotenoids and tocopherols protect the membranes and are thought to be particularly important in the chloroplast, where oxidizable polyunsaturated lipids abound. Mutants lacking carotenoids not only cannot photosynthesize, but cannot survive exposure to even very low levels of light (Sager & Zalokar 1958; Anderson & Robertson 1960). Arabidopsis thaliana plants grown in HL accumulate more ascorbate and α-tocopherol and have higher ascorbate peroxidase activity than plants grown in low light (LL) (Müller-Moulé, Golan & Niyogi 2004). Membranes are also protected by enzymatic peroxidases such as glutathione peroxidases and phospholipid hydroperoxide glutathione peroxidases. The latter can directly reduce phospholipid hydroperoxides to relatively non-toxic alcohols. Chlamydomonas reinhardtii, Synechocystis sp. PCC 6803, and A. thaliana each increase transcript abundance of some glutathione peroxidases in response to a shift from LL to HL (Hihara et al. 2001; Rossel, Wilson & Pogson 2002; Ledford et al. 2004).

Genetic approaches have verified the importance of some antioxidants for survival of HL stress. The npq1 lor1 double mutant in C. reinhardtii is unable to synthesize lutein as a result of the lor1 mutation, and the npq1 mutation prevents HL-induced violaxanthin de-epoxidation to zeaxanthin (Niyogi, Björkman & Grossman 1997). The npq1 lor1 double mutant undergoes photobleaching and lipid peroxidation in HL, eventually leading to cell death (Baroli et al. 2004). The npq1 lor1 double mutant is sensitive to superoxide stress generated in the chloroplast via treatment with metronidazole and to 1O2* stress generated by exposure to the photosensitizing dye, rose bengal (Baroli et al. 2004). Absence of the thermal dissipation component of non-photochemical quenching in the double mutant due to the lack of zeaxanthin does not explain HL sensitivity, since the npq5 mutant also lacks thermal dissipation but is not particularly HL sensitive (Elrad, Niyogi & Grossman 2002). Furthermore, creating a triple mutant with npq2, which prevents epoxidation of zeaxanthin to violaxanthin and constitutively increases the concentration of zeaxanthin, restores wild-type levels of resistance to rose bengal and HL (Baroli et al. 2003). A concomitant reduction in antenna size is not sufficient to fully explain the restoration of growth in HL, as HL sensitivity is only partially rescued in triple mutants containing npq1, lor1, and cbn1, a Chl b-less mutant with reduced antenna size (Baroli et al. 2003). The most likely explanation for HL sensitivity in npq1 lor1 is that reduced antioxidant capacity in the npq1 lor1 mutant leads to enhanced sensitivity to photo-oxidative stress.

Just as it is difficult to tease apart the role of each individual ROS, overlapping functions of antioxidants often complicate studies of their function in photo-oxidative stress prevention. The npq1 lor1 example also highlights the power of double mutants for unmasking the role of specific antioxidants or other protective mechanisms. Neither npq1 nor lor1 alone have an obvious HL phenotype, but the double mutant undergoes photo-oxidative bleaching (Niyogi et al. 1997). Given the number of different antioxidant systems present in photosynthetic organisms, it is perhaps not that surprising that mutations affecting activity of a single antioxidant often do not have an obvious impact on HL acclimation. In A. thaliana, the npq1 mutant also lacks the ability to convert violaxanthin to zeaxanthin, but is able to grow normally in HL (Niyogi, Grossman & Björkman 1998). When the npq1 mutation is combined with vtc2, a mutant that contains less than 30% of the wild-type ascorbate content, the resulting double mutant experiences more photo-oxidative stress in HL than either single mutant alone (Müller-Moulé, Havaux & Niyogi 2003). Similarly, absence of tocopherol in A. thaliana mutants did not obviously enhance sensitivity to photo-oxidative stress, and these mutants exhibited only mild deficiencies in photosynthetic capacity and lipid peroxidation (Porfirova et al. 2002; Bergmüller, Porfirova & Dörmann 2003). This was a surprise given the efficacy with which tocopherols can scavenge and quench 1O2* and LOOH in membranes, and the subsequent finding that mutants with reduced tocopherol levels also had seed germination and seedling development defects suggested that tocopherol was more important in preventing lipid peroxidation in seeds and seedlings than in coping with photo-oxidative stress (Sattler et al. 2004). Double mutants with reduced tocopherol and glutathione content, however, grew more slowly and had lower photosynthetic efficiency than either single mutant, indicating that the presence of at least one of the two antioxidants is required for normal growth and photosynthesis (Kanwischer et al. 2005). It is likely that tocopherol will emerge as an important player in the photo-oxidative stress response whose absence, like that of lutein and zeaxanthin, is felt primarily when other compensatory mechanisms are also lacking.

Phenotypes that are not apparent under laboratory conditions may also emerge when mutants are tested in the field. This was the case for npq4, an A. thaliana mutant that is deficient in thermal dissipation of excess energy. In the laboratory, npq4 exhibited an increase in photoinhibition in HL, but no obvious growth defect (Li et al. 2000, 2002). Under the naturally fluctuating light conditions in the field, however, fitness of npq4 was reduced by 30–50% (Külheim, Ågren & Jansson 2002). It is possible that the value of having multiple protective mechanisms only becomes clear in a fluctuating natural environment, and this may also explain the lack of a phenotype in some antioxidant-deficient mutants.

ROS SIGNALLING

While the importance and abundance of antioxidants in HL emphasizes the necessity of fending off ROS, discoveries made over the past few decades have demonstrated that ROS are not only destructive, but can also be important signals. At sublethal levels, ROS have been shown to activate defence responses in model organisms such as S. cerevisiae and Escherichia coli. Activating these defences enhances survival during subsequent oxidative stress from the same ROS. Given the link between ROS and HL stress, it is worth exploring whether similar responses are important for HL acclimation. Illuminating a single leaf on an A. thaliana plant results in H2O2 accumulation and increased ASCORBATE PEROXIDASE 2 (APX2) gene expression in unexposed leaves, raising the possibility that H2O2 acts as a signal during HL acclimation (Karpinski et al. 1999). The redox state of the plastoquinone pool has been proposed as a regulator of H2O2-mediated HL acclimation (Karpinska, Wingsle & Karpinski 2000), but how these signals are converted into changes in gene expression is not yet clear.

Classic examples of H2O2 and superoxide stress in S. cerevisiae and E. coli illustrate the value of ROS as signalling intermediates. The S. cerevisiae H2O2-responsive transcription factor Yap1 (Yeast activator protein 1) controls the transcription of genes involved in H2O2 stress responses via a redox-regulated domain comprised of four oxidizable cysteines (Wood, Andrade & Storz 2003; Wood, Storz & Tjandra 2004). Interestingly, these cysteines are not directly oxidized by peroxide, but instead mediate interactions with Gpx3, a glutathione peroxidase that directly interacts with the redox-regulated domain to produce an intramolecular disulfide bond (Delaunay et al. 2002). Involvement of Gpx3 in activation of Yap1 revealed a novel function for antioxidants as important signalling intermediates. Oxidation of the disulfide bonds triggers nuclear accumulation of the Yap1 transcription factor by masking a nuclear export signal (Wood et al. 2004). When the disulfide bonds are reduced, conformational changes re-expose the nuclear export signal, driving distribution of Yap1 to the cytoplasm where it is inactive.

ROS signalling may also be important for photosynthetic organisms, although at this stage no ROS ‘sensors’ have been identified. The development of microarrays for several photosynthetic model organisms opens up new avenues of exploring ROS responses. By tracking changes in transcript abundance following shifts from LL to HL, HL-response regulons have been identified in cyanobacteria, green algae, and land plants (Hihara et al. 2001; Huang et al. 2002; Rossel et al. 2002; Im et al. 2003; Kimura et al. 2003). Transcriptional profiles of plant tissue exposed to H2O2 and superoxide are now also available (Desikan et al. 2001; op den Camp et al. 2003, Vandenabeele et al. 2003, 2004). Promoter analysis of genes that were repressed or induced by HL provides a tantalizing beginning to understanding the regulation of HL responses and awaits functional confirmation (Rossel et al. 2002). Analysing the promoters of genes whose expression levels are co-regulated under these stress conditions may provide a starting point to identify the factors that regulate these responses.

RADICAL APPROACHES FOR A NON-RADICAL PROBLEM

Little is currently known about how responses to oxidative stress impact HL responses. Traditional methods of studying individual sources of oxidative stress rely on treatment with exogenous chemicals and may fail to reflect physiologically relevant factors such as localization and duration of stress under natural conditions. Furthermore, 1O2*, a ROS that is of particular relevance to the chloroplast, has received comparatively little attention. 1O2* is ephemeral, difficult to detect, and most often generated through the use of photosensitizing dyes such as rose bengal and methylene blue, the localization of which does not precisely match that of Chl. As a result, less is known about 1O2* stress than H2O2 and superoxide stress in any model system. Given the relevance of 1O2* to the chloroplast and the specificity of ROS responses, it is important to overcome the challenges of studying 1O2*.

Evidence for an overlap between HL and 1O2* responses has been accumulating. HL-grown cells of the green alga Dunaliella bardawil, which accumulates β-carotene in response to HL, are more resistant to rose bengal than LL-grown cells (Jimenez & Pick 1993). The C. reinhardtii npq1 lor1 double mutant lacking two carotenoids found exclusively in the chloroplast is sensitive to both HL and rose bengal (Baroli et al. 2003). Transcript abundance of a glutathione peroxidase gene in C. reinhardtii, Gpxh, increases in response to both rose bengal and HL (Leisinger et al. 1999; Ledford et al. 2004).

It is possible to chemically induce 1O2* production in the chloroplast. Herbicides such as 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) and bromoxynil (Krieger-Liszkay & Rutherford 1998; Fufezan, Rutherford & Krieger-Liszkay 2002) can promote 1O2* production in the chloroplast, but they do so by blocking electron transport. Alternatively, inhibitors of Chl biosynthesis can be used to trigger accumulation of photosensitizing porphyrin-containing Chl biosynthetic intermediates. Leisinger et al. (2001) studied 1O2* responses in C. reinhardtii by genetically or chemically inhibiting conversion of protoporphyrin IX to Mg-protoporphyrin IX, resulting in accumulation of the photosensitizer protoporphyrin IX, and found that Gpxh gene expression was enhanced by protoporphyrin IX accumulation in the light (Leisinger et al. 1999, 2001). This gene also confers resistance to lipid peroxidation when overexpressed in tobacco (Yoshimura et al. 2004). Analysis of the Gpxh promoter revealed a 1O2* -responsive element with similarity to cyclic-AMP responsive elements (Leisinger et al. 2001). This was one of the first clues that 1O2* stress in the chloroplast can lead to changes in nuclear gene expression, and future work promises to uncover more about the signalling pathway involved.

While the use of dyes and inhibitors has contributed to our understanding of 1O2* responses, recent creative approaches using A. thaliana mutants have dramatically changed our conception of 1O2* stress in plants by showing that the lipid peroxidation and cell death that accompanies 1O2* exposure may be genetically programmed rather than the result of direct oxidation. In A. thaliana, as in other angiosperms, protochlorophyllide (Pchlide) oxidoreductase is light dependent, meaning that dark-grown, etiolated seedlings fail to produce Chl and that light-grown plants will not synthesize Chl at night (Huq et al. 2004; Sperling et al. 1997). Pchlide is a powerful photosensitizer and if it were allowed to accumulate overnight, each daybreak would bring a fresh jolt of 1O2*. Arabidopsis thaliana uses a negative feedback regulation of aminolevulinic acid production to keep the concentration of Pchlide and other Chl biosynthetic intermediates, like Pchlide, in check (Sperling et al. 1997; Meskauskiene et al. 2001). In A. thaliana, a protein named FLUORESCENT (FLU, named for the mutant phenotype) mediates this negative feedback by direct interaction with glutamyl-tRNA reductase to prevent the formation of aminolevulinic acid (Meskauskiene & Apel 2002; Goslings et al. 2004). The A. thaliana flu mutant lacks this feedback inhibition and accumulates Pchlide in the dark, giving flu seedlings a characteristic Pchlide fluorescence following exposure to blue light (Meskauskiene et al. 2001). Shifting flu plants from the dark to LL results in photosensitized production of 1O2* by Pchlide accumulated in the chloroplast (op den Camp et al. 2003). Repeatedly shifting flu plants from dark to LL causes abrupt growth arrest and accumulation of foliar necrotic lesions. Cell death resulted from 1O2* production, as evidenced by partial rescue of protoplasts made from flu plants grown in the presence of vitamin B6 (Danon et al. 2005), which also protects Cercospora fungi from 1O2* produced by its own photosensitizing toxin, cercosporin (Ehrenshaft et al. 1999).

The flu mutant afforded a rare opportunity to study 1O2* stress using an endogenous, chloroplast-localized source of 1O2*. Microarray analysis of gene expression at time points following a transition to LL revealed a subset of 70 genes that were up-regulated in the flu mutant but not induced by treating flu with paraquat (op den Camp et al. 2003), a superoxide-producing compound. Paraquat treatment was an important control, because most if not all photosensitizers can also produce small amounts of superoxide in addition to 1O2* (Foote 1968). Although it is still possible that 1O2* produces other ROS intermediates in the flu mutant, this collection of 1O2*-inducible genes demonstrates that 1O2* produced in the chloroplast can impact nuclear transcription in ways that are distinct from other ROS.

Given the link between 1O2* and lipid peroxidation, it seemed likely that cell death in the flu mutant involved 1O2*-mediated membrane damage. Localization of the photosensitizer to chloroplast membranes coupled with the short half-life of 1O2* suggested that the molecules most immediately modified by photosensitization of Pchlide would likely be localized in or close to the plastid membrane. Lipid peroxidation profiles of the flu mutant revealed a surprise – while the flu mutant was accumulating LOOH, the type of peroxide formed was indicative of enzyme-mediated lipid peroxidation rather than direct modification of unsaturated fatty acids by 1O2* or other ROS (op den Camp et al. 2003). The two sources of lipid peroxidation can be distinguished by the composition of lipid hydroperoxides that accumulate: ROS-mediated lipid peroxidation is characterized by the accumulation of racemic mixtures, whereas enzymatic peroxidation produces structurally and entantiomerically uniform reaction products such as (13S)-hydroperoxides (Feussner & Wasternack 2002). One specific isomer of oxygenated linolenic acid, 13-HOTE, accumulated rapidly in re-illuminated flu plants. The uniformity of the LOOH population raises the possibility of a lipid-based, 1O2* signalling system in the flu mutant (op den Camp et al. 2003).

The flu system is elegant, but the reality is that wild-type A. thaliana plants do not accumulate Pchlide and are unlikely to experience 1O2* stress so uniformly throughout the leaves of the plant. Furthermore, although 1O2* generation in flu occurs in the chloroplast, the location of Pchlide may not precisely match that of Chl. Pchlide localizes to the prolamellar bodies of the chloroplast in dark-grown flu seedlings (op op den Camp et al. 2003), but subchloroplastic localization of Pchlide in light-grown plants has not been established. Despite these possible limitations, several lines of evidence suggest that lessons learned from the flu mutant may be translatable to HL-stimulated production of 1O2*.

Recent attempts to establish oxylipin profiles of plants experiencing HL stress have revealed a transient increase in 13-HOTE accumulation 30 h after shifting plants from LL to HL. Racemic mixtures of LOOH appeared only when wild-type or the HL-sensitive mutant chaos was exposed to HL coupled with carbon dioxide limitation to encourage photorespiration (Montillet et al. 2004). It is possible that the accumulation of 13-HOTE in HL-exposed plants and in flu occurred via the same lipoxygenase-mediated mechanism. 13-HOTE also accumulated as an early response to cadmium stress and during treatment with harpin, a protein that elicits a hypersensitive response in A. thaliana (Montillet et al. 2004). Both cadmium and the hypersensitive response may trigger 1O2* production as well as other ROS (Sandermann 2000; de León et al. 2002; Faller, Kienzler & Krieger-Liszkay 2005), but it remains possible that lipoxygenase activity is a general response to oxidative stress in the chloroplast, and not necessarily specific to 1O2*.

Mutagenesis of the flu mutant and subsequent isolation of second-site mutations that allow flu to survive dark to light transitions has also revealed parallels between flu responses to Pchlide-generated 1O2* and wild-type responses to 1O2* produced by HL. One such suppressor, executer (ex), prevents both cell death and growth arrest in the flu mutant (Wagner et al. 2004). Because 1O2* production in flu ex double mutants equals that in flu single mutants, and because ex is a recessive (loss-of-function) mutation, it is more likely that the chloroplast-targeted EX protein has a regulatory rather than antioxidant function. In a wild-type FLU background, the ex mutation partially protects cut leaves from ion leakage (measured as an estimate of cell viability) in response to DCMU treatment plus illumination (Wagner et al. 2004). Protection of cell viability was not absolute, however, and higher doses of DCMU killed ex at a rate similar to that of wild type. Furthermore, photosynthetic efficiency declined in the ex mutant just as quickly as it did in wild type. Future studies of the ex mutant and its responses to HL will clarify what role, if any, EX plays in HL responses.

Among the intriguing results to come from the extragenic suppressor screen was the finding that cell death and growth arrest in 1O2*-stressed flu mutants were genetically separable. This means that cell death and growth arrest, like lipid peroxidation, are programmed responses rather than direct consequences of 1O2* damage. Cell death in flu is accompanied by DNA fragmentation, a hallmark of programmed cell death (Danon et al. 2005). Programmed cell death in response to HL has not been yet been documented, although the dinoflagellate Peridinium gatunense undergoes apoptosis in response to ROS generated during CO2 limitation (Vardi et al. 1999). Among the genes induced by shifting flu from dark to light were genes related to salicylic acid, jasmonic acid, and ethylene biosynthesis or signalling (op den Camp et al. 2003; Danon et al. 2005). As in the programmed cell death accompanying a hypersensitive response to pathogen infection, salicylic acid and ethylene are required for cell death in the flu mutant following LL exposure (Danon et al. 2005). Cell death in flu is also partially dependent upon jasmonic acid, in contrast to superoxide-mediated cell death which is suppressed by jasmonic acid (see Kangasjärvi et al. 2005).

Identifying the ‘receptors’ for specific ROS is an important step towards understanding how oxidative stress is sensed in the chloroplast. No such receptor has been identified for 1O2* in any system. There are three possible mechanisms by which a receptor could sense 1O2*. First, 1O2* could directly modify a sensor protein that transduces the signal leading to changes in gene expression. Cysteine, histidine, methionine, tryptophan, and tyrosine residues are most sensitive to 1O2*, and modification of these residues could activate a protein sensor (Davies et al. 1999). Some of these amino acid modifications are specific to 1O2*, which could account for the specificity seen in 1O2* responses. The short lifetime of 1O2* means that a protein sensor must be localized very close to the point of origin. Although this could also confer a degree of specificity, a protein sensor would be limited in its flexibility to sense 1O2* from different sources.

As an alternative to direct modification of a protein, damage products resulting from 1O2* oxidation could serve as signalling intermediates. DNA damage is known to trigger a number of responses including apoptosis, but seems an unlikely candidate for signalling in the flu mutant since 1O2* generation occurs in membranes. LOOH are an intriguing possibility, especially in plants, both because of the localization of 1O2* stress within polyunsaturated fatty acid-rich environments such as the thylakoid membrane, and due to the known signalling capabilities of lipid breakdown products (Girotti & Kriska 2004). LOOH are longer lived than 1O2*, and can migrate through aqueous compartments to other membrane systems in vitro, possibly as a result of their enhanced polarity (Vila, Korytowski & Girotti 2001). Such transfer could propagate a LOOH signal between cellular compartments, and could also amplify toxicity of lipid peroxidation. Lack of direct 1O2* LOOH products in flu would seem to argue against this mechanism, but it may also be possible that levels of ROS-mediated lipid peroxidation below the detection threshold could act as a signal.

A third possibility is that 1O2* damage to lipids and proteins signals changes in gene expression by affecting the redox state of the cell. The redox state of the glutathione pool has been implicated in HL-accclimation, and transgenic Nicotiana tabacum plants with elevated chloroplast glutathione levels are more susceptible to photo-oxidative stress, possibly due to a disruption in the redox balance (Creissen et al. 1999; see Foyer & Noctor 2005). Glutathione redox state may also be important for regulating expression of the C. reinhardtii Rubisco large subunit in response to light (Irihimovitch & Shapira 2000). Protein peroxides and LOOH can both impact the redox state of the cell by depleting the overall thiol content (Morgan et al. 2004), but it is difficult to envision how this mechanism would account for the specificity seen in 1O2* responses.

At present, there is not enough data to distinguish definitively among these models. The flu mutant is an excellent place to begin to address these questions, with underlying consideration given to how results from flu could translate to HL stress responses. All of these roads will eventually lead back to the most important questions: how do plants and algae respond to their natural environment? This question will no doubt be addressed as genomic resources expand to encompass ecologically relevant photosynthetic organisms such as poplar (Brunner, Busov & Strauss 2004; Tuskan, DiFazio & Teichmann 2004), Prochlorococcus marinus (Dufresne et al. 2003, Rocap et al. 2003), and diatoms such as Thalassiosira pseudonana (Armbrust et al. 2004). Supplementing research in the laboratory with experiments performed on organisms in their natural environment or on economically significant species in managed field environments will open exciting new avenues of exploration in this arena.

ACKNOWLEDGMENTS

The authors thank Benjamin Gutman for critical reading of this manuscript. This work was supported by the National Institutes of Health (GM71908 and GM58799) and by the University of California Toxic substance Research and Teaching Program (grant no. 03T-1).

Ancillary