Dose-dependent effects of ¹O₂ in chloroplasts are determined by its timing and localization of production

In plants, ¹O₂ is primarily generated in chloroplasts and can act as a signal. Its effects are not only dose dependent, but also rely on localization and timing of its production.

Abstract

In plants, highly reactive singlet oxygen (¹O₂) is known to inhibit photosynthesis and to damage the cell as a cytotoxin. However, more recent studies have also proposed ¹O₂ as a signal. In plants under stress, not only ¹O₂ but also other reactive oxygen species (ROS) are generated simultaneously, thus making it difficult to link a particular response to the release of ¹O₂ and establish a signaling role for this ROS. This obstacle has been overcome by the identification of conditional mutants of Arabidopsis thaliana that selectively generate ¹O₂ and trigger various ¹O₂-mediated responses. In chloroplasts of these mutants, chlorophyll or its biosynthetic intermediates may act as a photosensitizer and generate ¹O₂. These ¹O₂-mediated responses are not only dependent on the dosage of ¹O₂ but also are determined by the timing and suborganellar localization of its production. This spatial- and temporal-dependent variability of ¹O₂-mediated responses emphasizes the importance of ¹O₂ as a highly versatile and short-lived signal that acts throughout the life cycle of a plant.

Introduction

Plants performing oxygenic photosynthesis must cope with photo-oxidative stresses throughout their life cycle. They utilize light as a primary energy source and synthesize and accumulate large amounts of chlorophyll (Chl) in chloroplasts to absorb light energy. If this light energy cannot be dissipated through photosynthetic electron transport, the excited Chl may turn into a potent photosensitizer that generates highly reactive singlet oxygen (¹O₂) (Apel and Hirt, 2004; Krieger-Liszkay et al., 2008; Foyer and Noctor, 2009; Triantaphylidès and Havaux, 2009). Plants have evolved various protection mechanisms to alleviate photodynamic damage caused by ¹O₂ (Foyer and Noctor, 2009; Li et al., 2009; Dogra et al., 2018). However, if this balance is disturbed when plants are subjected to environmental stresses, ¹O₂ is generated. For a long time, ¹O₂ had been recognized as detrimental to plants because of its oxidative damage effects, but more recent findings also suggest ¹O₂ as a signal (Li et al., 2009; Triantaphylidès and Havaux, 2009; Fischer et al., 2013; Laloi and Havaux, 2015; Noctor and Foyer, 2016; Foyer et al., 2017). Another emerging concept is that all types of oxidative modification/damage are involved in signaling (Foyer et al., 2017).

For various reasons it has been difficult to assess and verify such a signaling capacity of ¹O₂. (i) Besides ¹O₂, there are other chemically distinct reactive oxygen species (ROS) such as hydrogen peroxide (H₂O₂), superoxide (O₂·⁻), and hydroxyl radical (·OH), whose concentrations in plants under stress increase almost simultaneously with those of ¹O₂, thus making it difficult to define the biological effects of ¹O₂ and to link it to a particular cellular response (Apel and Hirt, 2004; Foyer and Noctor, 2009; Noctor et al., 2018). (ii) ¹O₂ may be formed at different intracellular and intraorganellar sites, and its biological impact may vary depending on where it is generated (Apel and Hirt, 2004; Noctor and Foyer, 2016). (iii) ¹O₂ interacts with proteins, lipids, carbohydrates, and nucleic acids (Sies and Menck, 1992; Davies, 2003; Triantaphylidès et al., 2008) and in this way may irreversibly inactivate and/or destroy the target (Elstner, 1982). It is not known yet how such a highly reactive molecule may initiate signaling rather than causing oxidative damage. (iv) Plants are exposed to diverse environmental conditions whose impact may rapidly change. (v) Even though the activity of ¹O₂ can be studied under steady-state conditions, it is difficult to separate primary from secondary effects of ¹O₂. To overcome these obstacles, one needs to induce the production of ¹O₂ non-invasively (Apel and Hirt, 2004).

In the meantime, experimental approaches have been developed that overcome most of these obstacles. Using these approaches, ¹O₂ has been recently demonstrated to act as a highly versatile signal that induces a wide range of stress responses throughout the life cycle of a plant (Fischer et al., 2013; Laloi and Havaux, 2015; Foyer et al., 2017; Noctor et al., 2018).

Formation of singlet oxygen

Ground state triplet molecular oxygen (³O₂) may be converted to the highly reactive ¹O₂ either metabolically or photochemically (Halliwell and Gutteridge, 1999; Apel and Hirt, 2004; Krieger-Liszkay et al., 2008; Triantaphylidès and Havaux, 2009; Fischer et al., 2013; Laloi and Havaux, 2015). Knowledge about the former reaction is still scarce (Kanofsky and Axelrod, 1986; Steinbeck et al., 1992; Khan and Kasha, 1994), whereas photochemical formation of ¹O₂ has been the subject of intense research (Krieger-Liszkay et al., 2008; Foyer and Noctor, 2009; Triantaphylidès and Havaux, 2009; Fischer et al., 2013; Laloi and Havaux, 2015). Energy transfer by an excited photosensitizer reverses the spin direction of one of the two outermost valence electrons of triplet oxygen that occupy separate orbitals with parallel spins, and allows the pairing of these electrons (Fig. 1). This spin reversal and pairing of electrons dramatically enhances the reactivity of oxygen (Fig. 1). In plants, ¹O₂ is primarily generated in chloroplasts due to the photosensitizing activity of tetrapyrroles. The syntheses of these porphyrins share a common pathway up to the formation of protoporphyrin IX (ProtoIX), when metals are inserted (Tanaka and Tanaka, 2007). Afterwards, the pathway diverges into two major branches, with the Fe branch being directed to the synthesis of hemes and phycobilins, and the Mg branch leading to the formation of Chls. These porphyrins have the potential to act as photosensitizers and transfer the excitation energy directly to ground state triplet oxygen, leading to the formation of singlet oxygen (Fig. 1) (Apel and Hirt, 2004; Krieger-Liszkay et al., 2008; Triantaphylidès and Havaux, 2009).

Fig. 1.

Formation of singlet oxygen (¹O₂). The biradical ³O₂ has two unpaired electrons with parallel spins. To oxidize a non-radical molecule, ³O₂ needs to react with a partner that provides a pair of electrons with parallel spins. Pairs of electrons usually have opposite spins and thus restrict the ability of ³O₂ to react with other molecules. Energy transfer from an excited photosensitizer (e.g. ³Chl) transforms ³O₂ into ¹O₂ by reversing the spin direction of one of the two unpaired electrons and allowing their pairing.

Chl, heme, and phycobilin are usually bound to proteins and, in this state, may use various quenching mechanisms to dissipate excess absorbed light energy. Thus, formation of ¹O₂ by these porphyrins is usually compromised under non-stressful conditions. Their biosynthetic intermediates, however, occur mostly in a free form and are potentially much more destructive when illuminated (Mochizuki et al., 2010). To avoid extensive photo-oxidative damage by these intermediates, plants prevent their accumulation by strictly controlling tetrapyrrole biosynthesis and catabolism. One important element of this regulation is the negative feedback inhibition of the first step of tetrapyrrole biosynthesis, formation of δ-aminolevulinic acid (ALA), by two effecter molecules, the FLU protein and heme, that both interact with different parts of Glu-tRNA reductase (Vothknecht et al., 1998; Meskauskiene et al., 2001; Goslings et al., 2004; Levicán et al., 2007), the first enzyme committed to tetrapyrrole synthesis (Vothknecht et al., 1998; Meskauskiene et al., 2001; Cornah et al., 2003). In the dark, the Mg branch leads only to the formation of protochlorophyllide (Pchlide), the immediate precursor of chlorophyllide (Chlide). The subsequent step from Pchlide to Chlide requires light (Griffiths, 1978; Apel et al., 1980). Once a critical level of Pchlide has been reached in the dark, ALA synthesis slows down and accumulation of Pchlide stops. Only after re-exposure to light, when Pchlide is photoreduced to Chlide, does Chl biosynthesis resume. Pchlide has been implicated in activating the FLU-dependent suppression of ALA formation, thereby allowing the Mg branch to regulate the initial step of tetrapyrrole biosynthesis, whereas the Fe branch controls this step via heme (Vothknecht et al., 1998; Goslings et al., 2004).

When plants are exposed to severe environmental stresses that interfere with photosynthetic electron transport, ¹O₂ production may also be initiated in chloroplasts by the photosensitizing activity of protein-bound Chl (Krieger-Liszkay et al., 2008; Li et al., 2009; Triantaphylidès and Havaux, 2009; Fischer et al., 2013; Laloi and Havaux, 2015 ). Upon light absorption, Chl changes from a ground state to the singlet excited state, ¹Chl. The fate of the excitation energy in this Chl may vary. Usually the absorbed light energy is transferred from Chl of the light-harvesting antenna complexes to the reaction center (RC) Chl where it drives photosynthetic electron transport. However, under high light, excessive excitation energy can also be dissipated as heat when the excited light-harvesting Chl interacts with carotenoids, or it can decay via formation of the excited triplet state of Chl (³Chl) (Krieger-Liszkay et al., 2008; Li et al., 2009). ³Chl has a longer life time than ¹Chl (Krieger-Liszkay, 2005), thereby allowing its excitation energy to be transferred to ground-state ³O₂ that produces ¹O₂ if no efficient quenchers are close enough to compete for this energy (Krieger-Liszkay et al., 2008). In the antenna, the ³Chl is in the close vicinity of various carotenoids that are able to quench ³Chl directly, and generation of ¹O₂ is usually suppressed. In contrast, in the RC of PSII the special Chl, P680, is not in close contact with carotenoids, and the β-carotene associated with the RC of PSII is unable to quench the excitation energy of ³P680; thus, generation of ¹O₂ is favored (Trebst, 1999; Umena et al., 2011) (Fig. 2A). Even though most of Chl is found in the antenna, ¹O₂ is mainly produced in the RC of PSII (Krieger-Liszkay et al., 2008). Whenever the electron acceptor of PSII remains reduced and is unable to accept electrons originating from excited P680 (e.g. under high light, low temperature, or drought conditions) (Fig. 2A), the excited triplet state of P680 may then act as a photosensitizer and produces ¹O₂. Under the stress conditions, ³Chl can also be generated in the PSII RC by charge recombination reactions (back-flow of electron transfer and charge separation reactions) and acts as a photosensitizer that produces ¹O₂ (Krieger-Liszkay et al., 2008). This ¹O₂ is believed to interact primarily with its nearest target, the D1 protein of PSII RC that binds the P680 Chl, thereby inactivating this protein and inhibiting PSII (Keren et al., 1997; Szilárd et al., 2005; Ohad et al., 2011; Vass, 2012; Kale et al., 2017) (Fig. 2A).

Fig. 2.

Schematic diagrams that illustrate how the PSII reaction center (RC) either in its active state in the grana core (A, B) or during its repair in grana margins (B, C) is thought to generate ¹O₂. (A) Light energy absorbed by PSII is trapped within the PSII RC whenever the electron acceptor of PSII remains reduced under various stress conditions and is unable to accept electrons which originated from the excited P680 Chl of the RC. This favors the transformation of the short-lived singlet state P680 into the more stable triplet state P680 that may then act as a photosensitizer and generate ¹O_2. This ¹O₂ is believed to interact primarily with its nearest target, the D1 protein of the PSII RC that binds the P680 Chl. At higher concentrations, such as under high light stress, ¹O₂ may also interact with β-carotene that is associated with PSII. Some of the oxidative breakdown products of β-carotene may act as a signal. (B) Schematic diagram showing the different regions of thylakoid membranes. Active PSII is localized in the grana core region, whereas the repair of damaged PSII takes place within the grana margin. (C) During the repair of PSII, the damaged D1 protein is degraded and replaced by newly synthesized D1 polypeptides. During the reassembly of active PSII, the insertion of D1 and Chl needs to be strictly co-ordinated to avoid photo-oxidative damage caused by the photosensitizing activity of unbound Chl. The Chl may be derived from the damaged PSII and/or is synthesized de novo within the grana margin region. A perturbation of this reconstitution is expected to allow a transient accumulation of either free Chl or its precursors (e.g. Pchlide or ProroIX) that may generate ¹O₂. PSII within the grana margin co-localizes with the FtsH protease, EX1, FLU, and enzymes of Chl synthesis. The onset of ¹O₂-mediated and EX1-dependent signaling strictly depends on an FtsH-dependent decline of the EX1 protein.

Signaling versus damaging: the role of singlet oxygen

For a long time, photo-oxidative damage of plant cells had been considered to be the only biological effect of ¹O₂ production (Elstner, 1982; Davies, 2003; Krieger-Liszkay et al., 2008). Thus, it came as a surprise when ¹O₂ was also shown to act as a signal that triggered specific nuclear gene expression changes and greatly impacted the phenotype of the affected plant (Kim et al., 2008; Krieger-Liszkay et al., 2008; Triantaphylidès and Havaux, 2009; Fischer et al., 2013; Laloi and Havaux, 2015; Wang and Apel, 2016; Foyer et al., 2017).

A signaling role for ¹O₂ was first demonstrated in the conditional fluorescent (flu) mutant of Arabidopsis (op den Camp et al., 2003; Wagner et al., 2004). FLU acts as a negative regulator of tetrapyrrole biosynthesis, and flu seedlings lacking this regulator are unable to restrict the accumulation of the Chl precursor Pchlide in the dark. When these seedlings are transferred to the light, they rapidly bleach and die due to the photosensitizing activity of excess amounts of free Pchlide (Meskauskiene et al., 2001; op den Camp et al., 2003) (Fig. 3A–D). In the flu mutant, the amount of ¹O₂ generated is proportional to the amount of Pchlide and thus to the duration of dark treatment (Laloi and Havaux, 2015). When grown under continuous light, the flu mutant looks exactly like the wild-type plant as Pchlide is immediately photoreduced to Chlide. These properties make the flu mutant an ideal tool for generation of ¹O₂ and study of ¹O₂-induced stress responses. In the flu mutant, shortly (~30 min) after the release of ¹O₂, the chloroplast starts to lose its integrity (the chloroplast becomes leaky as the stromal protein appears in the cytosol), and afterwards rupture of the central vacuole occurs followed by final collapse of the cell (Kim et al., 2012). However, all these drastic phenotypic changes disappear in the flu/ex1 double mutant although it overaccumulates Pchlide in the dark and generates a similar amount of ¹O₂ to that generated in the parental flu line during re-illumination (Wagner et al., 2004; Kim et al., 2012) (Fig. 3). In addition to EX1, an EX1-like protein, dubbed EX2, was identified that, unlike EX1, did not impact the ¹O₂-mediated cell death and growth inhibition responses of the flu mutant but greatly affected ¹O₂-mediated nuclear gene expression changes (Lee et al., 2007).

Fig. 3.

Dose-dependent ¹O₂-mediated responses of the flu mutant. Various concentrations of ¹O₂ were reached by either changing the duration of the dark treatment (A–D) or using different light intensities during illumination of etiolated seedlings (E). Seven-day-old light-grown seedlings of the wild type (WT), flu, and flu/ex1 were grown for 7 d under continuous light (90 μmol photons m⁻² s⁻¹) at 22 °C in soil, and put in the dark for different periods of time (A–D) and re-exposed to light for 24 h (A–C). With increasing duration of the dark period, the amount of the photosensitizer Pchlide (indicated by arrows) increased (D). Up to an 8 h dark period ¹O₂-mediated growth inhibition (A), photoinhibition of PSII as revealed by transient Chl fluorescence changes (B), and cell death as shown by trypan blue staining (C) in flu were abrogated in flu/ex1 seedlings, suggesting that these responses of the flu mutant were triggered by EX1-dependent signaling. The EX1 dependency of these responses was gradually lost in flu plants that were kept in the dark for 16 h or 24 h. R.F.U., relative fluorescence unit. (E) Four-day-old etiolated seedlings grown on 1/2 Murashige and Skoog (MS) (0.8% agar, 0.5% sucrose) were exposed for 2 d to different light intensities as indicated at the bottom of the figure. ¹O₂-mediated cell death was revealed by trypan blue staining. At the highest light intensity, flu but also flu/ex1 seedlings initiated a cell death response, whereas at an intermediate light intensity this response was abrogated in flu/ex1 but not in flu. At a very low light intensity, none of these seedlings showed a cell death response, and flu and flu/ex1 seedlings looked similar to green wild-type control seedlings. Images of transient fluorescence were taken with a FluorCam 800MF system (Photon Systems Instruments) following manuals provided by the manufacturer. Trypan blue staining of dead cells and determination of Pchlide was performed as described by op den Camp et al. (2003).

Many genes are up-regulated in response to the release of ¹O₂ in the flu mutant (op den Camp et al., 2003; Gadjev et al., 2006; Dogra et al., 2017). The genes that are up-regulated prior to chloroplast leakage are directly affected by the release of ¹O₂. In contrast, the genes whose expression changed after chloroplast leakage are only indirectly affected as the loss of chloroplast integrity seems to enhance the photosensitizing activity of membrane-bound Chl and to amplify drastically the production of ¹O₂ that subsequently causes massive non-enzymatic lipid peroxidation and photo-oxidative damage (Kim and Apel, 2013). For most ¹O₂-responsive genes, the up-regulation is first seen shortly after the loss of chloroplast integrity (Kim and Apel, 2013). A large number of these genes have been annotated as being under hormonal control (Baruah et al., 2009a). As the concentrations of phytohormones such as jasmonic acid (JA), ethylene, and salicylic acid (SA) increase soon after the onset of ¹O₂ production (op den Camp et al., 2003; Danon et al., 2005), the enhanced expression of the majority of ¹O₂-responsive genes seems to be triggered by these phytohormones (Kim and Apel, 2013). Nevertheless, the ¹O₂-mediated transcriptional changes rely on the presence of EX1/EX2 protein. Without these proteins, all subsequent ¹O₂ signaling-mediated stress responses of the flu mutant are abolished (Wagner et al., 2004; Kim et al., 2012).

EX1-dependent signaling operates only in response to rather minor increase in ¹O₂ production, when light-grown flu plants are transferred to the dark for up to 8 h, and are re-exposed to light (Wagner et al., 2004; Wang et al., 2016) (Fig. 3A–D). If the duration of the dark treatment is prolonged to 16 h or longer, flu seedlings accumulate more Pchlide and, upon re-illumination, generate higher amount of ¹O₂. Under these latter conditions, ¹O₂-mediated responses become EX independent (Fig. 3), thus re-emphasizing the fact that EX-dependent signaling in flu mutants occurs only at low, non-toxic concentrations of ¹O₂. Because of its high reactivity and short half-life, primary reactions of ¹O₂ that initiate signaling should be localized near the site of ¹O₂ formation. Thus, in the case of EX1-dependent signaling, ¹O₂ should be generated close to where EX1 is localized.

EX1 is present in thylakoid membranes and is restricted to the non-appressed margins of grana stacks (Wang et al., 2016) (Fig. 2B, C). Grana margins play an important role during repair of damaged PSII (Khatoon et al., 2009; Puthiyaveetil et al., 2014; Yoshioka-Nishimura et al., 2014). Active PSII is present in the core of tightly compressed grana membranes (Aro et al., 1993). Following inactivation, PSII needs to be translocated to the grana margins where it is disassembled and the damaged D1 and—to a lesser extent—also the D2 protein of the PSII RC are cleaved by the FtsH protease (Fig. 2C) (Lindahl et al., 2000; Bailey et al., 2002; Dogra et al., 2018). During the reassembly of active PSII, these proteins are replaced by D1 and D2 polypeptides synthesized on membrane-bound ribosomes attached to non-appressed stroma thylakoids (Mullet et al., 1990). In grana margins, EX1 forms part of a larger physical unit that contains PSII, the FtsH protease, the NADPH-Pchlide oxidoreductases (PORs) B and C that catalyze the light-dependent reduction of Pchlide to Chlide in light-grown plants, and two elongation factors (Wang et al., 2016). This localization of EX1 suggests that ¹O₂ is produced close to where the disassembly and reassembly of PSII take place. Most probably this ¹O₂ is generated by the photosensitizing activity of Chl or some of its intermediates that are used for the reconstitution of PSII (Fig. 2C). Turnover of Chl in light-adapted plants is confined to the core of PSII (Feierabend and Dehne, 1996), suggesting that newly synthesized Chl in green plants is primarily used for the assembly of active PSII. Enzymes of Chl biosynthesis and the FLU protein are highly enriched in grana margins and co-localize with the site of PSII repair (Wang et al., 2016). The recycling and de novo synthesis of Chl must be tightly controlled and synchronized with the synthesis of the PSII RC proteins D1 and D2 to minimize the disruptive effect of unbound, photoreactive Chl and its intermediates ( Mullet et al., 1990; Müller and Eichacker, 1999). A slight disturbance of this co-ordinated assembly process is likely to enhance the level of unbound Chl or its intermediates and to generate ¹O₂ (Fig. 2C). As shown by the example of the flu mutant, this ¹O₂ may trigger EX-dependent signaling (Wagner et al., 2004; Kim et al., 2012; Wang et al., 2016). In the flu mutant, the spatial distribution of the photosensitizer does not exactly match that of the EX1 protein. Whereas EX1 is confined to grana margins, Pchlide is evenly distributed in margin and core regions of grana stacks and is also found in stroma lamellae (Wang et al., 2016). This difference in the distribution of photosensitizer and EX1 re-emphasizes the notion that the amounts of ¹O₂ generated in the flu mutant after an 8 h dark–light shift are too low to cause apparent photo-oxidative damage to a plant. As the biological activity of ¹O₂ depends on the presence of EX1, this protein seems to act as a sensor of ¹O₂ that amplifies its potential signaling effect.

With the onset of ¹O₂-mediated signaling in the flu mutant, there is a rapid decline of EX1 that depends on the FtsH protease (Wang et al., 2016; Dogra et al., 2017). Generation of ¹O₂ without a decline of EX1 is not sufficient to trigger ¹O₂ signaling (Fig. 2C). As FtsH also cleaves the two PSII RC proteins D1 and D2 (Lindahl et al., 2000; Bailey et al., 2002; Kato et al., 2009), EX1-dependent signaling seems not only spatially but also functionally linked to the repair of PSII (Fig. 2C). The simultaneous onset of ¹O₂ production and decline of EX1 indicates that EX1 by interacting with ¹O₂ becomes susceptible to proteolytic attack by FtsH. It seems conceivable that EX1 acts as a negative regulator that needs to be removed by proteolytic cleavage to activate the ¹O₂-dependent signaling pathway. However, experimental evidence does not support such a notion. In flu/ex1 plants, ¹O₂-mediated signaling is only active after EX1 is expressed in the complemented mutant line (Wang et al., 2016; Dogra et al., 2017). At present, it is not known how FtsH modifies EX1 and whether proteolytic breakdown products of EX1 interact with downstream signaling components within the plastid or outside in the extraplastidic cytoplasm or in the nucleus.

Both cytotoxic and signaling effects of ¹O₂ can result in plant cell death, and which of the two effects prevails depends on the amount of ¹O₂. When the amount of ¹O₂ increases only slightly or moderately, its signaling effects prevails. In contrast, cytotoxic effects become dominant if the amount of ¹O₂ increases drastically. The loss of chloroplast integrity and rupture of the central vacuole can be mediated by signaling effects of ¹O₂ (Kim et al., 2012; Woodson et al., 2015). It is likely that these processes can also be induced by the cytotoxic effects of ¹O₂, but evidence is needed. To distinguish the signaling and cytotoxic effects of ¹O₂, two hallmarks are generally used: the impact of EX1 mutation and the prevalence of either enzymatic or non-enzymatic lipid peroxidation (Przybyla et al., 2008). If the ¹O₂-induced stress responses can be suppressed by EX1 mutation, and/or the enzymatic but not non-enzymatic lipid peroxidation occurs predominantly, these responses are generally attributed to signaling effects of ¹O₂, and otherwise to its cytotoxic effects. 13-HOT and 13-HOD are representatives of enzymatic lipid peroxidation, while 10-HOT, 10-HOD, 12-HOD, and 15-HOD are typical products of non-enzymatic lipid peroxidation (Przybyla et al., 2008). The etiolated flu and flu/ex1 seedlings accumulate ~3 times more Pchlide compared with seedlings that are grown under continuous light and treated with 8 h darkness. Thus, upon illumination, the etiolated flu and flu/ex1 seedlings generate a much higher amount of ¹O₂ that exceeds the threshold concentration of ¹O₂ which induces EX1-dependent signaling. In the etiolated seedlings of both flu and flu/ex1, non-enzymatic lipid peroxidation prevails and the EX1 mutation is unable to suppress the cell death response. In contrast, in 8 h dark-treated flu seedlings, lipid peroxidation occurs almost exclusively enzymatically, and both the lipid peroxidation and the cell death response can be suppressed by EX1 mutation (Przybyla et al., 2008).

Different roles of ¹O₂: quality control versus induction of cell death

Once a signaling role for ¹O₂ had been established in the flu mutant, interest in studying the biological activity of ¹O₂ markedly increased. Other experimental systems have now also been established that non-invasively induce generation of ¹O₂. One of them is the ferrochelatase2 (fc2) mutant of Arabidopsis. In the fc2 mutant, the photosensitizing activity of the tetrapyrrole intermediate, ProtoIX, generates ¹O₂ (Woodson et al., 2015). Similarly, down-regulation of a tobacco ferrochelatase by RNAi increases the level of ProtoIX that generates ¹O₂ and leads to formation of necrotic leaves in the transformed tobacco (Papenbrock et al., 2001). Moreover, when plants are treated with peroxidizing herbicides that inhibit Protox, the enzyme that catalyzes the oxidization of Protogen to ProtoIX, they start to overaccumulate ProtoIX and show severe photo-oxidative damage (Sandmann and Boger, 1988; Becerril and Duke, 1989; Watanabe et al., 1998).

Studies of the fc2 mutant have implicated the release of ¹O₂ with activation of a ubiquitin-dependent quality control pathway that has been proposed to remove damaged chloroplasts selectively from plant cells (Woodson et al., 2015). The fc2 mutant of Arabidopsis lacks one of two ferrochelatases that catalyze formation of protoheme by inserting Fe²⁺ into ProtoIX. While fc2 mutant plants grown under continuous light turn green like the wild type, they become pale and form abnormally small leaves under an 8 h light/16 h dark regime (Woodson et al., 2015). This phenotypic change had been attributed to the generation of ¹O₂ by the photosensitizing activity of ProtoIX (Scharfenberg et al., 2015; Woodson et al., 2015). Upon transfer to dark, the fc2 mutant starts to overaccumulate ProtoIX and after 50 min of darkness reaches an ~10-fold higher maximum than the wild type, but that declines afterwards over the next 30 min. The concentration of ProtoIX at the end of the 16 h dark period has not been reported and thus it is not known whether generation of ¹O₂ in fc2 mutants grown under an 8 h light/16 h dark cycle is indeed caused by increased levels of ProtoIX. In addition, the transformed tobacco leaves with reduced expression of a tobacco ferrochelatase accumulate ProtoIX under light but not in the dark (Papenbrock et al., 2001). Thus, it is necessary to determine the ProtoIX concentration of the fc2 mutant at the end of the dark period and at selected time points during the light period. In etiolated fc2 seedlings, ProtoIX reaches the same low level as in wild-type control plants (Scharfenberg et al., 2015), whereas the concentration of Pchlide is higher than in the wild type but not as high as in similarly treated flu plants (Scharfenberg et al., 2015; Woodson et al., 2015). Thus, as pointed out by Scharfenberg et al. (2015), the fc2 mutant displays a weak flu phenotype when grown in the dark, suggesting that generation of ¹O₂ in this mutant is due to the photosensitizing activity of Pchlide. This interpretation is supported by the fact that a ¹O₂-mediated loss of chloroplast integrity and a subsequent collapse of the affected cell appears not only in the flu mutant but also in the fc2 mutant (Kim et al., 2012; Woodson et al., 2015). Among second-site mutations of fc2 that restore the ability to green when grown under a dark/light regime, the PUB4 gene encoding a E3 ubiquitin ligase has been identified (Woodson et al., 2015). When grown under non-permissive dark/light conditions, the fc2/pub4 double mutant generates ¹O₂ and its chloroplasts appear stressed with a distorted membrane system, but unlike chloroplasts of the parental fc2 line they remain intact and are not degraded. Thus, ubiquitination of chloroplast proteins seems to be an important step during ¹O₂-mediated dismantling and subsequent degradation of chloroplasts. Whether in fc2 mutants this ubiquitin-dependent breakdown of chloroplasts signifies activation of a quality control pathway that is mediated by ¹O₂ is difficult to judge. As the release of ¹O₂ in flu and fc2 mutants induces chloroplast leakage followed by the disruption of the central vacuole and the collapse of the cell (Kim et al., 2012; Woodson et al., 2015), it sounds impossible to identify a quality control pathway that selectively removes individual damaged chloroplasts in a collapsed cell. However, in fc2 plants grown under continuous light that are not expected to overaccumulate ProtoIX and Pchlide, evidence for the operation of a ubiquitin-dependent quality control pathway has been obtained (Woodson et al., 2015). These plants show, relative to wild-type plants, a reduced growth and slightly impaired photosynthetic electron transport (Scharfenberg et al., 2015). These differences between the mutant and wild type may explain why in fc2 the number of damaged chloroplasts is higher than in the wild type. In the fc2/pub4 double mutant, the number of damaged chloroplasts returns to the wild-type level, suggesting that ubiquitination is indeed involved in recognizing damaged chloroplasts and subsequently allowing their degradation to proceed. However, so far it remains unclear whether activation of this pathway and/or damage of chloroplasts in these plants are caused by ¹O₂.

Even though fc2 resembles flu, the biological effects of ¹O₂ in these two mutants show a remarkable difference. In flu, ¹O₂-mediated phenotypic changes depend on EX1 and are completely abrogated in the flu/ex1 double mutant. Whereas in fc2, ¹O₂-induced changes are not affected by the absence of EX1 (Woodson et al., 2015). At the moment, it is difficult to explain this difference. Since fc2 mutants accumulate less Pchlide in the dark than flu mutants (Scharfenberg et al., 2015), they are not expected to release a higher amount of ¹O₂ that surpasses the signaling capacity of EX1, as shown in Fig. 3. It is not known whether FC2 is localized in the grana margin region close to EX1. As the intraorganellar location of ¹O₂ formation is expected to influence the biological effects of ¹O₂, the release of ¹O₂ at a different site away from EX1 may explain why activation of ¹O₂-mediated responses in the fc2 mutant does not depend on EX1.

¹O₂-induced cell death under severe high light stress

The third experimental system used to induce generation of ¹O₂ non-invasively and to study its signaling activity is the Chl b-deficient chlorina1 (ch1) mutant of Arabidopsis that is devoid of PSII antenna complexes (Dall’Osto et al., 2010; Ramel et al., 2013; Shumbe et al., 2016). Without these antenna complexes, the mutant lacks light-scavenging capacity and is highly sensitive to high light. In contrast to wild-type controls, ch1 plants exposed to a combination of high light (1000 µmol photons m⁻² s⁻¹) and low temperature (10 °C) suffer from severe photoinhibition of PSII and photo-oxidative damage due to an enhanced production of ¹O₂ in the RC of PSII (Ramel et al., 2013). The excitation energy trapped within the PSII RC favors the transformation of the short-lived singlet state of P680 into the more stable triplet state that allows the excited Chl to act as a photosensitizer that generates ¹O₂ (Fig. 2A). ¹O₂ oxidizes various chloroplast membrane constituents such as proteins (Davies, 2003), polyunsaturated fatty acids (Przybyla et al., 2008; Triantaphylidès et al., 2008), and carotenoids (Ramel et al., 2012a), causing photo-oxidative damage that ultimately may lead to the collapse of cells. This cell death response of the ch1 mutant to high light stress is not exclusively due to the toxicity of ¹O₂ but also involves genetically controlled stress-responsive components that are activated by the release of ¹O₂, and depends on the OXIDATIVE SIGNAL INDUCIBLE 1 (OXI1) Ser/Thr kinase (Shumbe et al., 2016; Foyer et al., 2017; Dogra et al., 2018). Expression of the OXI1 gene in the high-light-stressed ch1 mutant is strongly induced. Inactivation of OXI1 reduces the extent of photo-oxidative damage and cell death in the high-light-stressed ch1/oxi1 double mutant even though these plants generate similar amounts of ¹O₂ to those generated by the parental ch1 mutant (Shumbe et al., 2016). OXI1 had been shown earlier to be involved in H₂O₂-mediated signaling that controlled root hair growth and plant–pathogen interactions (Rentel et al., 2004). As OXI1-dependent responses triggered by H₂O₂ differ from those induced by ¹O₂, both ROS seem to activate different signaling pathways that converge on OXI1 but otherwise operate via distinct mechanisms and lead to different physiological responses of the affected plants.

At first glance, the ¹O₂-mediated cell death responses of ch1 and flu seem to be very similar. In both mutants, the release of ¹O₂ not only leads to cell death but also affects similar sets of ¹O₂-responsive nuclear genes, and these responses occur independently of H₂O₂-dependent signaling. However, initiation of the high-light-induced cell death in ch1 results primarily from photo-oxidative damage, whereas in flu the cell death response is under genetic control and is triggered by a rapid but minor increase of ¹O₂ that is too low to damage the cell directly (Kim et al., 2012; Shumbe et al., 2016; Wang et al., 2016). In ch1, the ¹O₂ level rises more gradually, with a production lasting as long as the light stress is maintained and the PSII RCs are still intact. During this period, increases of various hormones such as JA and ethylene form an integral part of the cell death-inducing mechanism (Shumbe et al., 2016). Also in flu, ¹O₂-mediated hormone changes have been reported (op den Camp et al., 2003; Danon et al., 2005; Przybyla et al., 2008). However, in flu, the enhanced hormone production and expression changes of most of the ¹O₂-responsive genes occur later than the ¹O₂-mediated loss of chloroplast integrity (Kim et al., 2012). Hence, these ¹O₂-responsive genes seem to be only indirectly affected by ¹O₂ and are probably activated during the loss of cellular integrity by hormones such as JA and ethylene (Kim and Apel, 2013). Generation of ¹O₂ in ch1 has been reported to occur within grana stacks of PSII (Laloi and Havaux, 2015), whereas in flu¹O₂ formation that induces cell death responses takes place within grana margins close to the repair site of damaged PSII (Wang et al., 2016). In flu, the ¹O₂-mediated rapid loss of chloroplast integrity and the subsequent cell death strictly depend on the two EX proteins that are localized within the grana margins, whereas in ch1 the high-light-induced cell death response is not affected by the absence of these proteins (Shumbe et al., 2016). Thus, even though the cell death responses of flu and ch1 are both triggered by ¹O₂, they are controlled by different mechanisms.

Under moderate light stress (400 µmol photons m⁻² s⁻¹/20 °C), ¹O₂ induces stress acclimation in ch1 that attenuates the ¹O₂-mediated cell death response during a subsequent severe high light stress (Shumbe et al., 2016). ¹O₂-mediated signaling under moderate light stress seems different from OXI1-dependent signaling in high-light-treated plants. Under severe light stress, the ¹O₂-mediated cell death is preceded by an enhanced expression of the OXI1 gene, whereas under moderate light stress expression of this gene is suppressed. At the same time, an enhanced expression of acclimation-specific genes under moderate light contrasts with the down-regulation of other ¹O₂-responsive genes that are activated under high light stress (Shumbe et al., 2016). Some of these latter genes are known to control the biosynthesis of JA. In high-light-treated ch1 plants, up-regulation of these genes correlates with high levels of JA, whereas in pre-acclimated ch1 plants this JA accumulation during high light treatment is suppressed (Shumbe et al., 2016). Exogenously applied JA restores the high light stress-induced cell death response in pre-acclimated ch1 plants. On the other hand, a genetic block of JA biosynthesis significantly reduces the extent of photo-oxidative damage and cell death during high light stress in a JA-deficient delayed-dehiscence2 (dde2)/ch1 double mutant (Ramel et al., 2013). Collectively, these results suggest that JA plays a dual role by promoting a cell death response and suppressing the effect of stress acclimation in high-light-treated ch1 mutant plants.

¹O₂-mediated dose-dependent and spatially resolved responses

So far, different ¹O₂-mediated reactions of plants have been attributed to different dose-dependent effects of ¹O₂. In high-light-treated ch1 plants, ¹O₂-mediated signaling at higher ¹O₂ concentrations contributes to the cell death response and induces an up-regulation of OXI1 gene expression, whereas under moderate light stress a lower level of ¹O₂ correlates with stress acclimation, suppression of OXI1 gene expression, and reduced photodamage (Shumbe et al., 2016). In light-grown flu seedlings transferred to the dark for up to 8 h and re-exposed to light, a lower amount of ¹O₂ activates EX1-dependent signaling and triggers a cell death response without obvious photo-oxidative damage (Kim et al., 2012; Wang et al., 2016). An extension of the duration of the dark period in the flu mutant enhances the accumulation of the photosensitizer Pchlide that upon re-illumination generates a higher amount of ¹O₂ and causes photo-oxidative damage (Fig. 3A–D). Under these latter conditions, the toxicity of ¹O₂ prevails and superimposes the EX1-dependent signaling induced by ¹O₂ (Fig. 3). It might be interesting to test the behavior of EX1-overproducing plants under these latter conditions. During illumination, the amounts of ¹O₂ produced in etiolated flu and flu/ex1 are also affected by light intensities. Under a higher light intensity (100 μmol photons m⁻² s⁻¹), a higher amount of ¹O₂ is produced. As a result, the cytotoxic effect of ¹O₂ prevails and both flu and flu/ex1 seedlings initiate a cell death response. However, under a low light intensity (10 μmol photons m⁻² s⁻¹), only a moderate amount of ¹O₂ is generated; the cell death response is abrogated in flu/ex1 but not in flu. At a very low light intensity (1 μmol photons m⁻² s⁻¹), the amount of ¹O₂ produced is too low to lead to a cell death response even in flu (Fig. 3E).

Since in flu a cell death response is triggered by a minor transient increase of ¹O₂ whereas in ch1 a cell death response is seen under a higher ¹O₂ production, ¹O₂-mediated cell death responses do not seem to be only dose dependent. In flu and ch1, ¹O₂ is generated at different sites within the chloroplast, suggesting that the spatial distribution of ¹O₂ production also influences the reaction of the plant. Because of its high reactivity and short half-life (~4 µs in water), the primary reaction of ¹O₂ is restricted to a small suborganellar area adjacent to the site of ¹O₂ generation (Redmond and Kochevar, 2006; Laloi and Havaux, 2015). Spatially resolved ¹O₂-initiated responses have been proposed to depend on a spatial resolution of the photosensitizer distribution (Redmond and Kochevar, 2006). However, in the case of the EX1-mediated cell death, this response of the flu mutant strictly depends on the presence of EX1 within grana margins and not on the distribution of the photosensitizer Pchlide (Wang et al., 2016). EX1 seems to operate as a sensor of minor increases of ¹O₂ concentrations that amplifies the potential biological activity of ¹O₂ and reveals its signaling capacity. Thus, it is the location of EX1 but not the spatial distribution of the photosensitizer that defines the site where the primary photo-oxidation by ¹O₂ and initiation of ¹O₂-mediated and EX1-dependent signaling occur (Wang et al., 2016).

With the onset of ¹O₂ formation in the flu mutant, there is a rapid decline of the EX1 protein in grana margins that depends on the metalloprotease FtsH (Wang et al., 2016). Hence, the proteolytically modified EX1 or its breakdown products may be expected to be directly involved in ¹O₂-mediated signaling and to interact with downstream signaling components. In ch1, ¹O₂ formation in grana stacks has been implicated in damaging active PSII and inhibiting photosynthesis (Shumbe et al., 2016). Under these severe high light stress conditions, ¹O₂ generates non-enzymatically a wide range of oxidation products, some of which may disseminate within the cell and act as a second messenger that triggers stress responses. For instance, oxidation of lipids has been shown to produce multiple reactive derivatives with strong electrophilic properties (Farmer and Mueller, 2013) that may activate redox-sensitive transcription factors. Oxidation of β-carotene by ¹O₂ in the PSII RC generates various oxidative breakdown products, some of which, such as β-cyclocitral (Ramel et al., 2012a, b) and dihydroactinidolide (Shumbe et al., 2014), are biologically active (Fig. 2A). Intriguingly, these carotenoid oxidation products do not trigger cell death but induce stress defense responses and acclimation.

Spatially resolved responses may be triggered not only by ¹O₂ generated in grana margins and grana stacks. There are also other sites within the chloroplast that upon stress accumulate tetrapyrroles and release ¹O₂. The chloroplast envelope contains Pchlide (Pineau et al., 1986). In cotyledons, the import of PORA into plastids depends on Pchlide that interacts with the PORA precursor polypeptide during its uptake (Kim and Apel, 2004). In the outer plastid envelope protein16-1 (oep16-1) mutant, this import of PORA is impaired and free Pchlide starts to accumulate (Samol et al., 2011). Upon light exposure, it generates ¹O₂ and induces a cell death response that differs from that of the flu mutant (Samol et al., 2011). As mentioned above, the fc2 mutant displays a weak flu phenotype when grown under non-permissive dark/light conditions and induces a cell death response that unlike flu is not affected by the absence of EX1 (Woodson et al., 2015). As the intraorganellar location of ¹O₂ formation is expected to influence the biological effects of ¹O₂, the release of ¹O₂ at a site other than the grana margin could explain why the ¹O₂-mediated cell death responses in the oep16-1 and fc2 mutants differ from the cell death response of the flu mutant.

In addition to the location and the extent of ¹O₂ formation, the developmental stage at which ¹O₂ is produced also influences the specificity of ¹O₂-mediated responses. A development-dependent specification of ¹O₂-mediated responses was first documented in the flu mutant. At the seedling stage, the release of ¹O₂ induced a rapid bleaching of flu seedlings (op den Camp et al., 2003), in more mature flu plants the release of ¹O₂ triggered lesion formation in leaves (Wang et al., 2016), whereas in plants ready to bolt generation of ¹O₂ led to an immediate cessation of growth (op den Camp et al., 2003; Przybyla et al., 2008). At all three developmental stages, ¹O₂-mediated responses were dependent on EX1 and were abolished in the flu/ex1 double mutant (Wagner et al., 2004; Wang et al., 2016). In Arabidopsis wild-type plants, ¹O₂-mediated and EX-dependent signaling was shown also to be active during late embryogenesis prior to the onset of seed dormancy and to affect plastid development after seed germination (Kim et al., 2009). Etioplasts and chloroplasts of seedlings are derived from undifferentiated progenitors in embryos named proplastids that descend from maternal plastids (Possingham, 1980). During the transition from the morphogenic phase with rapid cell division to the maturation phase of embryogenesis, proplastids may either differentiate into functional photoheterotrophic chloroplasts or remain undifferentiated (Vicente-Carbajosa and Carbonero, 2005). With the onset of seed desiccation, thylakoid membranes of these photoheterotrophic chloroplasts disintegrate and release their Chl. At the same time, ¹O₂-mediated and EX-dependent signaling is initiated that pre-determines the fate of plastid differentiation by recruiting absisic acid that later on acts as a positive regulator of plastid formation in etiolated and light-grown seedlings. In wild-type seedlings lacking EX1 and EX2, chloroplast development in cotyledons is severely impaired. This is reflected in a reduced Chl and protein content and a much smaller size of chloroplasts in cotyledons of ex1/ex2 seedlings that resemble undifferentiated proplastids (Kim et al., 2009).

Outlook

As the biological effects of ¹O₂ are influenced not only by the location and extent of ¹O₂ production within chloroplasts but also change during plant development, ¹O₂ is expected to act as a highly versatile and short-lived signal throughout the life cycle of a plant and to give rise to a surprising variety of different signaling pathways. So far little is known about what distinguishes these different signaling pathways from each other. In the flu mutant, second-site genetic screens aimed at identifying constituents involved in ¹O₂-mediated signaling have only led to the discovery of the plastid-localized EX1 protein but failed to identify other signaling components (Wagner et al., 2004; Baruah et al., 2009a; Meskauskiene et al., 2009). In the fc2 mutant, similar second-site mutant screens have led to the identification of the E3 ubiquitin ligase that is required for the breakdown of chloroplasts but is unlikely to take part in cell death-inducing signaling (Woodson et al., 2015). In ch1, OXI1 acts as an enhancer of high-light-induced cell death rather than being an obligatory component of a ¹O₂-dependent cell death-inducing signaling pathway (Shumbe et al., 2016). These results suggest that the proposed ¹O₂-dependent signals are not transferred to the nucleus via a single linear signaling pathway that can easily be blocked genetically but rather through a more complex signaling network that is difficult to analyze by introducing only single gene mutations.

Unlike the flu, fc2, and ch1 mutant plants, in wild-type plants ¹O₂-mediated signaling does not operate alone but interacts with other signaling pathways that converge with ¹O₂-dependent signaling and delay or modify some of the ¹O₂-mediated responses seen in the three mutant lines (Baruah et al., 2009b; Kim and Apel, 2013). For instance, perturbations of cellular homeostasis prior to ¹O₂ production confer an enhanced stress resistance by activating acclimation that suppresses ¹O₂-mediated cell death responses without blocking ¹O₂-mediated expression changes of ¹O₂-responsive genes (Baruah et al., 2009b; Coll et al., 2009; Meskauskiene et al., 2009; Šimková et al., 2012). ¹O₂ itself may also modify consequences of higher ¹O₂ concentrations through an autoregulatory feedback control that induces acclimation (Ledford et al., 2007; Kim and Apel, 2013; Shumbe et al., 2016). Finally, hydrogen peroxide may antagonize ¹O₂-mediated signaling. Overexpression of a thylakoid-bound ascorbate peroxidase (tAPX, an H₂O₂ scavenger) in the flu mutant increases the expression of most of the ¹O₂-induced genes and enhances the ¹O₂-mediated cell death and growth inhibition phenotypes compared with the flu parental line (Laloi et al., 2007). Other recent examples of ¹O₂-mediated signaling in wild-type plants have indicated that it is not only restricted to light stress, but may also occur during wounding, pathogen attack, senescence, and drought stress (Mur et al., 2010; Vellosillo et al., 2010; Alboresi et al., 2011; Nomura et al., 2012; González-Pérez et al., 2011; Gutiérrez et al., 2014; Mor et al., 2014; Uberegui et al., 2015). ¹O₂ that triggers these responses may not only be generated photochemically in chloroplasts, but in some cases has been suggested to be formed metabolically in the absence of light and to emanate also from other subcellular compartments (Mor et al., 2014; Noctor and Foyer, 2016; Foyer et al., 2017). Collectively, these data emphasize the complexity of signaling events that must be dissected before the biological significance of ¹O₂-mediated signaling in wild-type plants can be fully understood.