Reactive Oxygen Species in Plant Cell Death

CELL DEATH IN PLANTS

Paradoxically, death is an integral part of life. Cell death is essential for growth and development of eukaryotes, by maintaining tissue and organ homeostasis in concert with cell proliferation, growth, and differentiation. Until recently, the wide variety of cell death types reported in the literature was mostly caged in two semantic categories: apoptosis and necrosis. Discrimination between these two forms was based on the presence or absence of specific biochemical and molecular hallmarks, such as DNA laddering, cytochrome c release, caspase involvement, ATP depletion, cytoplasmic swelling, and loss of membrane integrity (Pennell and Lamb, 1997). However, over the last decade this arbitrary division had clearly become too simplistic and a more accurate description of plant cell death needed to be established (van Doorn and Woltering, 2005). To avoid a Babel confusion of languages, we propose necrosis to depict accidental cell death caused by extrinsic factors, such as phytotoxic accumulation of specific molecules after a traumatic stress event. Thus, plant cell death through necrosis is passive, indiscriminate, and often follows irreversible injury. It is characterized by a progressive loss of membrane integrity that results in swelling of the cytoplasm and release of cellular constituents. Accordingly, the terms active or programmed cell death (PCD) define any form of cell death involving a single or a series of molecular and cellular orderly processes mediated by intracellular death programs, regardless of the (external) trigger or the hallmarks it exhibits (Jacobson et al., 1997; Dangl et al., 2000). Although both cell death events are quite well defined in animals, in plants there seems to be much more overlap between the phenotypic and molecular hallmarks of necrosis and PCD, thus making it harder to discriminate between the two events. In plants, PCD can occur during many developmental processes and abiotic stress conditions. For instance, PCD has been described during hypoxia stress (Drew et al., 2000), extremes in temperature (Vacca et al., 2004), and ozone (Langebartels et al., 2002). Developmental PCD has been reported during tracheary element formation (Kuriyama and Fukuda, 2002), seed development, germination (Souter and Lindsey, 2000; Young and Gallie, 2000), and senescence processes (Gunawardena et al., 2004). However, cell death events remain best described during incompatible plant-pathogen interactions that form the basis for the hypersensitive response (HR; Pennell and Lamb, 1997). This multitude of plant PCD events clearly illustrates a functional analogy between cell death across kingdoms: sculpting and deleting structures, eliminating cells to control cell quality and quantity after trauma, and producing differentiated cells without organelles (Jacobson et al., 1997). But, is there more than just a functional similarity between PCD in plants and animals? In animals, reactive oxygen species (ROS), such as hydrogen peroxide (H₂O₂), superoxide ion, and nitric oxide (NO) are well-recognized triggers of cell death (Jabs, 1999). In contrast, involvement of these molecules during plant PCD was, for a long time, rather hypothetical. In addition to the lack of accurate and quantifiable detection methods, the inherent toxic nature of ROS masked their underlying function in various signaling networks. Only recently, the prominent role of ROS has been revealed in the induction, signaling, and execution of plant cell death. The initial phenotypic and pharmacological evidence for ROS signaling during plant PCD has been confirmed genetically with the identification of Arabidopsis (Arabidopsis thaliana) mutants incapable either of arresting ROS-driven PCD or of undergoing a PCD despite high ROS levels. Recently, the first genes involved in ROS perception and signal transduction have been identified and now we are faced with the challenge of uncovering the other players in the gene regulatory network of ROS-dependent cell death. In this update, we will describe the current knowledge on ROS homeostasis and cell death events in plants.

ROS HOMEOSTASIS, A DEDICATED BALANCE BETWEEN DEATH AND SIGNALING

Life under aerobic conditions is intimately linked with ROS production. Indeed, ROS are generated as by-products of the most essential energy-generating processes, photosynthesis and respiration. Together with an extensive battery of oxidases, chloroplasts, peroxisomes, and mitochondria are the main organellar ROS producers (Vranová et al., 2002; Apel and Hirt, 2004). Despite an efficient antioxidant machinery in these organelles, subtle changes in ROS homeostasis are inevitable (Foyer and Noctor, 2005). When the increase in ROS is relatively small, the housekeeping antioxidant capacity is sufficient to reset the original balance between ROS production and scavenging, thus reestablishing redox homeostasis. Under growth-limiting environmental conditions, this delicate redox balance is easily disturbed, potentially leading to a significant ROS accumulation. In fact, a 3- to 10-fold increase in ROS levels has been calculated under stress conditions (Polle, 2001). Thus, it is not surprising that a transient oxidative burst and a subsequent temporary shift in the intracellular redox state are common features of both biotic and abiotic stress responses (Dat et al., 2000; Mittler et al., 2004). Under unfavorable environmental conditions, such as temperature extremes, drought, or salt stress, the rate of carbon fixation is limited, causing an increase in photoinhibition potentially steering the photosystem toward overproduction of superoxide radicals and H₂O₂ (Foyer and Noctor, 2005). Similarly during ozone exposure, ROS are generated following the entry of ozone through the stomata and its conversion in the leaf apoplast, eventually leading to the formation of HR-like lesions (Pellinen et al., 1999; Rao and Davis, 2001). Furthermore, most incompatible plant-pathogen interactions are typified by a biphasic oxidative burst during which both infected and adjacent cells are sacrificed to limit pathogen spreading (Doke, 1997; Draper, 1997). The accumulation of ROS toward the apoplast is considered to originate mainly from an increased activity of apoplastic peroxidases, amine oxidases, and an NADPH-oxidase complex coupled to a decrease in cellular ROS-scavenging capacity (Bolwell, 1999; Mittler et al., 1999; Torres and Dangl, 2005).

At high concentrations (oxidative stress), the various ROS can certainly behave as extremely reactive molecules. Their ability to react indiscriminately with almost all cellular components provokes, among others, destructive protein modifications as well as mutagenic DNA strand breaks, purine oxidations, and protein-DNA cross links (Beckman and Ames, 1997; Berlett and Stadtman, 1997). Lipid peroxidation, in both cellular and organellar membranes, takes place when above-threshold ROS levels are reached, thereby not only directly affecting normal cellular functioning, but also aggravating the oxidative stress through production of lipid-derived radicals (Montillet et al., 2005). Consequently, the cellular damage resulting from high ROS levels shows some of the typical hallmarks of necrosis, because phytotoxic levels of oxidants indiscriminately attack cellular constituents, leading to membrane leakage and cell lysis. To avoid such collateral damage, a tight regulation of ROS homeostasis is necessary, which is provided by a complex gene network that operates in all subcellular compartments and through complex feed-forward/feed-back loops between oxidants and scavengers. As an illustration of this complexity, at least 158 genes in the Arabidopsis genome specifically master the spatio-temporal network of ROS production and scavenging (Mittler et al., 2004). Simultaneously, the tight ROS homeostasis regulation also creates a baseline on which spikes of ROS can signal in different cellular processes. In addition, depending on their type and their subcellular production sites, ROS may determine the specificity of ensuing cellular responses. Although the underlying signaling function of ROS has long been masked by the inherent toxic nature of oxidants, their triggering ability during various developmental processes and environmental stress responses is now widely recognized (Foyer and Noctor, 2005).

ROS SIGNALS INVOLVED IN PLANT CELL DEATH

The first indications for an involvement of ROS in PCD were based mainly on spatio-temporal correlations between increased ROS levels and cell death. The initial experimental evidence that ROS also acted as a signal in plant PCD was obtained in cell suspensions by demonstrating that H₂O₂-induced cell death could be blocked by cycloheximide and protease inhibitors (Levine et al., 1994). Since then, there has been ever growing support for a key role for ROS as triggers of PCD. Whereas initial results were generally obtained by exogenous application of oxidants, more recently, transgenic plants with perturbed levels of cellular antioxidants and mutants unable to stop the initiation or propagation of ROS-driven cell death have elegantly demonstrated the important role of ROS in the initiation phase of different plant PCD events (Mittler and Rizhsky, 2000; Lorrain et al., 2003).

The best studied examples of ROS-derived PCD are those following the typical biphasic oxidative burst during both the HR and ozone stress. The ensuing cell death is characterized by discrete cellular lesions that are generally preceded by the appearance of several hallmarks of PCD, such as chromatin condensation, DNA laddering, and cytochrome c release (Lam, 2004). Moreover, these features can be halted by administrating either high concentrations of antioxidants or inhibitors of both translation and transcription and of known signal transducing components, such as kinases or phytohormones. Accordingly, transgenic plants with low or high levels of several antioxidants (superoxide dismutase, catalase, and ascorbate peroxidase) also exhibit an altered response to both pathogen- and ozone-driven PCD, again demonstrating the crucial importance of a tightly orchestrated redox balance (Van Camp et al., 1994; Örvar et al., 1997; Mittler et al., 1999). Further evidence for a ROS-dependent PCD pathway in plants has been provided by transgenic plants deficient in catalase (Cat1AS), the major H₂O₂-scavenging enzyme. In Cat1AS leaves of tobacco (Nicotiana tabacum), accumulation of photorespiratory H₂O₂ induced an active cell death of leaf palisade parenchyma cells. In addition to the presence of several PCD hallmarks, this high light-driven cell death could be blocked by infiltration of various antagonists of HR-like PCD (Dat et al., 2003). Interestingly, we have also been able to uncouple a necrosis-like response, provoked by continuous high light-dependent H₂O₂ accumulation, from PCD triggered by a transient, nonlethal, in planta ROS accumulation. The short H₂O₂ pulse sufficient to induce PCD in leaves had a lipoxygenase-dependent oxylipin signature similar to that induced by a pathogenic elicitor (cryptogein). In contrast, the continuous H₂O₂ accumulation generated by long-term high light exposure or H₂O₂ feeding led to necrosis and ROS-mediated lipid peroxidation (Montillet et al., 2005). The existence of a dose-dependent ROS threshold below which PCD will be triggered and above which necrotic cell death will prevail might be the reason for the overwhelming effect that phytotoxic ROS levels have on PCD hallmarks. Alternatively, similarly to what is reported in animals, high doses of oxidants might inhibit components of the PCD pathway (Kazzaz et al., 1996). Complementary to the data provided by transgenic plants, a number of Arabidopsis mutants have demonstrated the specificity in the signaling capacity of different ROS in initiating PCD. In two mutants of Arabidopsis, lesion-stimulating disease1 (lsd1) and radical-induced cell death1 (rcd1), elevated ROS levels are necessary and sufficient to induce spreading of cell death (Jabs et al., 1996; Overmyer et al., 2000; Mateo et al., 2004). The conditional fluorescent (flu) mutant of Arabidopsis that generates singlet oxygen upon a dark-to-light shift initiates a cell death response immediately after the release of singlet oxygen (op den Camp et al., 2003). Singlet oxygen is interpreted differently at the molecular level than H₂O₂ and superoxide, thus providing further evidence for a selective signaling effect of various ROS in plants (Danon et al., 2005). The ozone-sensitive mutant vitamin c-1 (vtc-1), deficient in l-ascorbic acid, shows small patches of spontaneous lesions on its leaves (Pavet et al., 2005). These lesions were attributed to changes in redox homeostasis rather than to H₂O₂, as evidenced by lower ascorbate levels in vtc mutants than in the wild-type plants (Pavet et al., 2005). Finally, the localized expression pattern of the copper amine oxidase-encoding gene ATAO1, which produces H₂O₂ associated with PCD in lateral root cap cells and developing tracheary elements, also hints to H₂O₂ as a trigger for certain developmental cell death processes (Møller and McPherson, 1998). In summary, various ROS species and cellular redox changes can trigger different signaling cascades leading to PCD. However, the understanding of how ROS signals are perceived and transduced is still in its infancy.

ROS SIGNAL RELAY DURING PLANT CELL DEATH

In animal cells, mitochondria have long been recognized as central players in ROS-dependent apoptotic cell death. One key event is the release of cytochrome c necessary for caspase activation that precedes mitochondrial membrane depolarization and nuclear condensation as well as other hallmarks of apoptosis (Fleury et al., 2002). In plants, mitochondria may also serve as first relay stations where the initial alteration in ROS homeostasis is amplified, triggering cytochrome c release through mitochondrial transition pore opening and morphological changes (Tiwari et al., 2002; Dat et al., 2003; Casolo et al., 2005; Yoshinaga et al., 2005). The role of mitochondria is further corroborated by the up-regulation of both manganese superoxide dismutase and the alternative oxidase mitochondrial antioxidant genes early during ROS-driven PCD (Robson and Vanlerberghe, 2002; Dat et al., 2003). In plants, besides mitochondria, chloroplasts are also important ROS suppliers, and they may generate intermediate signals involved in PCD. For example, during cryptogein-induced PCD, an H₂O₂-dependent activation of lipoxygenases targets chloroplastic polyunsaturated fatty acids, releasing oxylipins that are sufficient for inducing PCD (Rustérucci et al., 1999; Maccarrone et al., 2000; Montillet et al., 2005). A plastid involvement in PCD is further substantiated by the ectopic expression in the chloroplasts of mammalian antiapoptotic B-cell leukemia/lymphoma (BCL2) members, which protect transgenic tobacco plants from herbicide-induced PCD (Chen and Dickman, 2004). Finally, recent data on the role of phytochrome signaling during the establishment of the HR clearly implicate the necessity for a chloroplastic factor in the pathway leading to the HR (Genoud et al., 2002; Karpinski et al., 2003). Thus, in addition to the recognized participation of mitochondria in PCD, an active involvement of chloroplast-derived signals is also primordial during plant PCD.

A strong interplay between ROS and other signaling molecules (phytohormones) exists during plant PCD (Overmyer et al., 2005). The fact that ROS-dependent PCD is associated with increased levels of both salicylic acid (SA) and ethylene and that the overproduction of an SA-degrading enzyme (catechol) in ethylene mutants tempers PCD, strongly positions both ethylene and SA within a positive feed-back cycle that promotes ROS-dependent cell death (de Jong et al., 2002; Moeder et al., 2002; Danon et al., 2005). On the other hand, jasmonic acid can either induce or act antagonistically in this oxidative-dependent cell death cycle depending on whether the initial ROS signal is singlet oxygen or superoxide. This different interaction with oxylipins adds another level of specificity and complexity to ROS signaling during PCD (Overmyer et al., 2000; Danon et al., 2005). An additional player in ROS-dependent PCD is NO that operates in a dose-dependent manner synergistically with H₂O₂ in HR cell death (Delledonne et al., 2001; Zago et al., 2006). Moreover, NO may directly interact with SA, because infiltration of tobacco leaves with NO results in a significant accumulation of SA (Durner et al., 1998). Thus, redox-dependent plant PCD involves a range of signaling molecules, but this interactive regulatory network is just starting to be elucidated (Fig. 1).

Figure 1.

Schematic representation of ROS-dependent cell death pathways. Environmental and developmental changes stimulate the production of ROS through various organelles and enzymes. This initial increase in ROS levels may be further enhanced by various relay and production centers. Sustained ROS production provokes the accumulation of phytotoxic levels of ROS leading to necrotic cell death. Transient increases may initiate signal transduction cascades involving cross talk with other phytohormones, SA, jasmonate, ethylene, and NO. This cross talk can amplify or diminish the ensuing response or channel the cascade through more specific transducer sensors and targets toward ROS-dependent and cell death-related gene expression. Besides MAPK-driven phosphorylation cascades, other regulatory posttranslational modifications, such as protein oxidation and nitrosylation might be involved in ROS-dependent cell death pathways. ETH, Ethylene; exe1, executer1; JA, jasmonic acid.

MOLECULAR TARGETS OF ROS SIGNALING DURING PLANT CELL DEATH

One of the first genes identified in ROS-dependent cell death was isolated through a genetic screen. The Arabidopsis lsd1 mutant is unable to cope with increased superoxide levels and develops spontaneous superoxide- and SA-dependent runaway lesions (Jabs et al., 1996). LSD1 encodes a zinc-finger protein that, together with two other zinc-finger proteins (LOL1 and LOL2), could act as a molecular rheostat, sensing changes in ROS homeostasis, thereby repressing a default death pathway or controlling an antideath mechanism through the regulation of cell death genes (Dietrich et al., 1997; Epple et al., 2003). The ozone-sensitive lesion mimic mutant, rcd1, also displays PCD lesions. RCD1 has been proposed to modulate the activity of target proteins through ADP ribosylation (Ahlfors et al., 2004; Overmyer et al., 2005). In a landmark paper, ROS-induced PCD has recently been shown not to be mediated by photooxidative damage but through a genetically determined PCD pathway (Wagner et al., 2004). In fact, inactivation of the Arabidopsis Executer1 gene completely abolished singlet oxygen-induced cell death (Wagner et al., 2004). This hitherto unknown nucleus-encoded chloroplastic protein might perceive nonscavenged singlet oxygen species within the chloroplast. Whether Executer1 is specifically involved in singlet oxygen perception or also in sensing other ROS species remains to be tested.

Other important stress signal transducers include mitogen-activated protein kinases (MAPKs) that act upstream of the oxidative burst during ozone treatment and the HR (Ren et al., 2002; Samuel and Ellis, 2002). However, MAPKs might also work downstream in ROS-dependent cell death events. The primary ROS-activated tobacco MAPK is the SA-induced protein kinase, which is required during harpin-dependent PCD (Samuel et al., 2005). A MAPKKK of alfalfa (Medicago sativa) activates cell death induced by H₂O₂ through a specific MAPK-scaffolding action (Nakagami et al., 2004). Overlapping features of animal and plant PCD have also inspired different functional approaches by introducing animal genes into plants to gather mechanistic insight into oxidative stress-dependent cell death events. No obvious homologs of the BCL-like animal cell death suppressors have been identified in plant genomes to date; nevertheless, protection against mitochondrial and chloroplast-derived ROS-dependent cell death is conferred by their overproduction in tobacco. Although functional plant homologs have been successful in protecting against ROS-mediated PCD in the transgenic lines, their existence remains unclear (Mitsuhara et al., 1999; Chen and Dickman, 2004). However, recently an evolutionarily conserved Arabidopsis BCL2-associated athanogene protein has been shown to be induced by H₂O₂ and capable of provoking PCD in both yeast (Saccharomyces cerevisiae) and plants (Kang et al., 2006).

CONCLUSION AND PERSPECTIVES

Research on redox-dependent PCD has demonstrated that when ROS production exceeds the cellular scavenging capacity, ROS levels may rise above two different threshold levels both leading to cell death. Unleashed ROS accumulation will build up to phytotoxic levels hereby indiscriminately attacking and damaging proteins, lipids, and DNA. This type of ROS-dependent cell death leads to a necrotic phenotype and is commonly encountered under conditions favoring leakage from electron transport chains together with reduced antioxidant potential. Under many abiotic stress conditions, cell death execution is steered through this ROS overaccumulation. When the accumulation of ROS is insufficient to kill the cell directly, changes in cellular redox homeostasis appear to switch on a signaling cascade, leading to PCD, as typically encountered during the HR.

The recent identification of plant genes involved in ROS-dependent PCD is merely the start of a new era during which unearthing the remainder of the gene regulatory network that controls this mechanism will be crucial. Initial optimism that functional equivalents of animal cell death genes could be identified in plants on the basis of sequence homologies has gradually declined. Until today, no homologs of recognized apoptotic regulators, such as caspases or BCL2-associated X protein, have been detected in plants. Thus, current stakes are on the outcome of ongoing mutant and functional screens. Additional suppressor screens for ROS-induced cell death and cloning of the remaining alleles already reported to revert or induce PCD in Arabidopsis mutants will increase the number of genes in the plant PCD pathway schemes. Finalized genome-wide transcriptome studies have identified hundreds of genes that are rapidly induced by ROS signals (Desikan et al., 2001; op den Camp et al., 2003; Vandenabeele et al., 2003; Vanderauwera et al., 2005; Gadjev et al., 2006). Within these inventories, a cohort of novel genes involved in early ROS perception and transduction will undoubtedly be discovered. Functional screens, in which these genes are either up- or down-regulated and scored for their potential to modulate ROS-dependent cell death, will assist in their identification. This approach, together with the characterization of target proteins that undergo regulatory posttranslational modifications, such as protein oxidation and nitrosylation during cell death, will lead to a better understanding of cell death in plants.