Production of Reactive Oxygen Species by Photosystem II as a Response to Light and Temperature Stress

The effect of various abiotic stresses on photosynthetic apparatus is inevitably associated with formation of harmful reactive oxygen species (ROS). In this review, recent progress on ROS production by photosystem II (PSII) as a response to high light and high temperature is overviewed. Under high light, ROS production is unavoidably associated with energy transfer and electron transport in PSII. Singlet oxygen is produced by the energy transfer form triplet chlorophyll to molecular oxygen formed by the intersystem crossing from singlet chlorophyll in the PSII antennae complex or the recombination of the charge separated radical pair in the PSII reaction center. Apart to triplet chlorophyll, triplet carbonyl formed by lipid peroxidation transfers energy to molecular oxygen forming singlet oxygen. On the PSII electron acceptor side, electron leakage to molecular oxygen forms superoxide anion radical which dismutes to hydrogen peroxide which is reduced by the non-heme iron to hydroxyl radical. On the PSII electron donor side, incomplete water oxidation forms hydrogen peroxide which is reduced by manganese to hydroxyl radical. Under high temperature, dark production of singlet oxygen results from lipid peroxidation initiated by lipoxygenase, whereas incomplete water oxidation forms hydrogen peroxide which is reduced by manganese to hydroxyl radical. The understanding of molecular basis for ROS production by PSII provides new insight into how plants survive under adverse environmental conditions.


INTRODUCTION
Photosystem II (PSII) is water-plastoquinone oxidoreductase embedded in the thylakoid membrane that catalyzes light-driven H 2 O oxidation to O 2 and plastoquinone (PQ) reduction to plastoquinol (PQH 2 ; Dau et al., 2012;Vinyard et al., 2013;Nelson and Junge, 2015;Suga et al., 2015;Najafpour et al., 2016). In this reaction, primary charge separation between the chlorophyll monomer (Chl D1 ) and pheophytin (Pheo D1 ) of D1 protein forms 1 [Chl D1 •+ Pheo D1 •− ] radical pair which is fast stabilized by the oxidation of the weakly coupled chlorophyll dimer P D1 and P D2 (P680) forming 1 [P680 •+ Pheo D1 •− ] radical pair (Cardona et al., 2012). 1 [P680 •+ Pheo D1 •− ] radical pair is stabilized by the electron transport from Pheo D1 to the tightly bound plastoquinone Q A forming Q A •− and from the redox active tyrosine residue D1:161Y (Y Z ) to P680 •+ forming Y Z • . Electron transport form Q A •− to loosely bound plastoquinone Q B and the reduction of Y Z • by the proton-coupled electron transport from the Mn 4 O 5 Ca cluster forms reducing and oxidizing equivalent at Q B and Mn 4 O 5 Ca cluster, respectively. When two reducing equivalents are formed at Q B site, its protonation forms plastoquinol (PQH 2 ) which is liberated to PQ pool via channels (Lambreva et al., 2014). Formation of four oxidizing equivalents in the Mn 4 O 5 Ca cluster causes four-electron oxidation of two H 2 O to O 2 which is released via channels into the lumen (Vogt et al., 2015).
Light-driven processes comprising both energy transfer and electron transport are accompanied by formation of reactive oxygen species (ROS). In the energy transfer, singlet oxygen ( 1 O 2 ) is formed by the energy transfer from triplet chlorophyll to O 2 (Triantaphylides and Havaux, 2009;Pospíšil, 2012;Fischer et al., 2013). In electron transport, ROS are formed by the consecutive one-electron reduction of O 2 and by the concerted two-electron oxidation of H 2 O on the PSII electron acceptor and donor sides, respectively (Pospíšil, 2009). The one-electron reduction of O 2 forms superoxide anion radical (O 2 •− ) which dismutes spontaneously or enzymatically to hydrogen peroxide (H 2 O 2 ) and subsequently is reduced to hydroxyl radical (HO • ) via Fenton reaction. The two-electron oxidation of water forms H 2 O 2 which is oxidized and reduced to O 2 •− and HO • , respectively. Nonenzymatic and enzymatic scavenging systems have been engaged to eliminate ROS and thus control level of ROS formed under various types of abiotic (adverse environmental conditions such as high light, high and low temperatures, UV-radiation, and drought) and biotic (herbivores and pathogens such as viruses, bacteria, and fungi) stresses.
Under moderate stress, when scavenging system maintains ROS level low, ROS serves as signaling molecules which activate an acclimation response and programmed cell death (Apel and Hirt, 2004;Dietz et al., 2016). Several lines of evidence have been provided that ROS play a crucial role in intracellular signaling from the chloroplast to the nucleus under high light (Gollan et al., 2015;Laloi and Havaux, 2015) and high temperature (Sun and Guo, 2016). However, due high reactivity of ROS toward proteins and lipids, ROS diffusion is limited. It seems to be unlikely that ROS might transmit signal from the chloroplast to the nucleus. It is considered that products of protein oxidation and lipid peroxidation might serve as signaling molecules (Fischer et al., 2012). As ROS formed by energy transfer ( 1 O 2 ) and electron transport (H 2 O 2 ) are produced simultaneously, it seems to be likely that their action in signaling pathways interferes. It was demonstrated that H 2 O 2 antagonizes the 1 O 2 signaling pathways in the flu Arabidopsis mutant (Laloi et al., 2007).
Under severe stress, when scavenging system is unable to sufficiently eliminate undesirable ROS formation, PSII proteins and lipids might be oxidized by ROS. Several lines of evidence were provided in the last three decades on the oxidative damage of PSII proteins by ROS under high light (Aro et al., 1993) and high temperature (Yamamoto et al., 2008). It is widely accepted that 1 O 2 is major ROS responsible for oxidative modification of PSII proteins. Contrary, H 2 O 2 has low capability to oxidize PSII protein; however, when free or protein-bound metals are available, HO • formed by Fenton reaction oxidizes nearby proteins. It has to be pointed that experimental evidence on PSII protein oxidation was obtained in vitro and thus it remains to be clarified whether oxidative modification of PSII proteins by ROS occurs in vivo. Apart to involvement of ROS in PSII protein damage, the inhibition of de novo protein synthesis by ROS was proposed under high light (Nishiyama et al., 2006) and high temperature (Allakhverdiev et al., 2008). Whereas PSII protein oxidation is widely described, limited evidence has been provided on lipid peroxidation near PSII. It was shown that 1 O 2 formed in PSII initiates lipid peroxidation in the thylakoid membrane (Triantaphylides et al., 2008).
In this review, an update on the latest findings on molecular mechanism of ROS formation at high light and high temperature is presented. In spite of the fact that molecular mechanism of ROS formation is substantially different at high light and high temperature, high light regularly combined with high temperature might bring about more serious impact on ROS formation.

HIGH LIGHT
When light energy which is driving force for photosynthetic reactions exceeds the photosynthetic capacity, a lightinduced decline in photochemical activity in PSII denoted as photoinhibition occurs. Limitations in the energy transfer and electron transport result in the generation of ROS. Limitation in energy transfer occurs, when the excess energy absorbed by chlorophyll in the PSII antennae complex is not fully utilized in the PSII reaction center by charge separation. Under these conditions, singlet chlorophyll might be converted to deleterious triplet chlorophyll. To prevent formation of triplet chlorophyll, quenching of singlet chlorophyll to heat is maintained directly by xanthophylls or indirectly by the rearrangement of Lhcb protein by PsbS (Ruban et al., 2012). However, when quenching of singlet chlorophyll is not sufficient, singlet chlorophyll is converted to triplet chlorophyll which transfers energy to O 2 forming 1 O 2 . Limitation in electron transport on the PSII electron acceptor side is accompanied by full reduction of PQ pool. As the Q B site becomes unoccupied by PQ due to the full reduction of PQ pool, forward electron from Q A to Q B is blocked. Under these conditions, back electron transport from Q A •− to Pheo and consequent recombination of Pheo

Singlet Oxygen
Singlet oxygen is formed by the triplet-triplet energy transfer from triplet chlorophyll or triple carbonyl to O 2 . Triplettriplet energy transfer from triplet chlorophyll to O 2 occurs in both the PSII antennae complex and the PSII reaction center. In the PSII antennae complex, triplet chlorophyll is formed by the photosensitization reaction, whereas in PSII reaction center triplet chlorophyll is formed by the charge recombination of triplet radical pair 3 [P680 •+ Pheo •− ]. Triplet-triplet energy transfer from triplet carbonyl to O 2 proceeds during lipid peroxidation initiated by ROS formed by light. Whereas 1 O 2 formation by the energy transfer from triplet chlorophyll is well documented and represents the main source of 1 O 2 at high light, 1 O 2 formation by the energy transfer from triplet carbonyls is rarely evidenced and has marginal contribution to the overall 1 O 2 formation.

Triplet Chlorophyll
Light energy absorbed by chlorophylls is transferred from the PSII antennae complex toward the PSII reaction center (van Amerongen and Croce, 2013). However, when energy transfer is limited, chlorophylls might serve as photosensitizers which form 1 O 2 by the energy transfer from their triplet state to O 2 ( Figure 1A). To prevent this, chlorophylls are coupled with carotenoids which have capability to quench triplet chlorophylls. Carotenoids consist of carotenes (β-carotene) and their oxygenated derivatives xanthophylls (lutein, zeaxanthin; Domonkos et al., 2013). In the PSII antennae complex, lutein and zeaxanthin play a crucial role in triplet chlorophyll quenching (Dall'Osto et al., 2006, 2012. Whereas lutein is permanently coordinated to Lhcb proteins, zeaxanthin is accumulated under high light by the reversible de-epoxidation of violaxanthin and is either free in the thylakoid membrane or bound to Lhcb protein (Havaux and Niyogi, 1999;Pinnola et al., 2013). Four xanthophyll binding sites were documented in the monomeric (Lhcb4-6) and the trimeric (LHCII) antenna proteins of PSII (Liu et al., 2004). Xanthophylls bound in both L1 (lutein) and L2 (lutein in LHCII and lutein or zeaxanthin in monomeric Lhcb4-6 proteins) sites can efficiently quench the neighboring triplet chlorophylls. Lutein in L1 (Lut620) and L2 (Lut621) are coupled with chlorophylls Chl610-Chl614 and Chl602-Chl604, respectively. The quenching of triplet chlorophylls 602 and 603 by lutein in L2 is highly efficient, whereas lutein in L1 site had no effect on quenching of triplet chlorophyll 612 (Ballottari et al., 2013). To maintain effective quenching of triplet chlorophyll by carotenoids, carotenoids has to be properly distanced and oriented from chlorophylls. Triplet-triplet energy transfer from chlorophylls to carotenoids is mediated by Dexter mechanism (Dexter, 1953), which needs overlap between the electron clouds of the donor and acceptor. When distance or orientation of carotenoid and chlorophyll is changed, the capability of carotenoids to quench excitation energy of triplet chlorophylls is diminished (Cupellini et al., 2016). Under such conditions, when O 2 is in the proximity of triplet chlorophyll, the transfer of excitation energy from triplet chlorophyll to O 2 forms 1 O 2 . Comparison of the monomeric and the trimeric antenna proteins of PSII showed that the monomeric antenna proteins (Lhcb6 > Lhcb5 > Lhcb4) produced more 1 O 2 as compared to trimeric antenna proteins (LHCII; Ballottari et al., 2013).
When electron transport on the PSII electron acceptor side is limited due to the slow electron transport to the Q A and Q B , several types of charge recombination of [P680 •+ Q A •− ] and 1 [P680 •+ Pheo D1 •− ] radical pairs occur. Whereas [P680 •+ Q A •− ] radical pair recombines solely to the ground state P680, primary radical pair 1 [P680 •+ Pheo D1 •− ] formed by the reverse Frontiers in Plant Science | www.frontiersin.org electron transport from Q A •− to Pheo D1 either recombines to the ground state P680 or converts to the triplet radical pair 3 [P680 •+ Pheo D1 •− ] by change in the spin orientation. Recombination of triplet radical pair 3 [P680 •+ Pheo D1 •− ] forms triplet chlorophyll 3 P680 * delocalized on the weakly coupled chlorophyll dimer P D1 and P D2 (Fischer et al., 2013;Telfer, 2014). Evidence has been provided that triplet state is localized on the Chl D1 at low temperature (Noguchi et al., 2001). The formation of 3 Chl D1 was proposed to occur either directly by the charge recombination of the triplet radical pair 3 [P680 •+ Pheo D1 •− ] or by the triplet energy transfer from 3 P680 * to Chl D1 . As two β-carotenes (Car D1 and Car D2 ) are distanced from chlorophyll dimer P D1 and P D2 , β-carotenes are not able to quench triplet chlorophyll 3 P680 * ( Figure 1B).

Triplet Carbonyl
Lipid peroxidation initiated by radical ROS (O 2 •− , HO • ) forms the primary and the secondary lipid peroxidation products. The primary lipid peroxidation product are lipid hydroperoxides (lipid hydroperoxy fatty acids, LOOH) which decompose to the secondary lipid peroxidation products lipid hydroxides (hydroxy fatty acids, LOH), reactive carbonyl species (RCS), and electronically excited species. Hydrogen abstraction from polyunsaturated fatty acid by HO • forms lipid alkyl radical (L • ) which interacts with O 2 forming lipid peroxyl radical (LOO • ). Lipid peroxyl radical abstracts hydrogen from the adjacent polyunsaturated fatty acid forming LOOH. Lipid hydroperoxide is stable; however, under oxidizing or reducing condition it is oxidized or reduced to LOO • or alkoxyl radical (LO • ). Cyclization or recombination of LOO • forms high energy intermediates, dioxetane, or tetroxide. High energy intermediates are highly unstable and decomposite to triplet excited carbonyls ( 3 L * ) which might transfer triplet energy to O 2 forming 1 O 2 . Alternatively, tetroxide might directly decompose to 1 O 2 via the Russell mechanism. Evidence has been provided that 1 O 2 is formed through lipid peroxidation under light stress in spinach PSII membranes deprived by the Mn 4 O 5 Ca cluster (Yadav and Pospíšil, 2012a). The authors demonstrated that the oxidation of lipids by highly oxidizing P680 •+ and TyrZ • caused 1 O 2 formation via the Russell mechanism. It has to be noted that amount of 1 O 2 formed by the triplet-triplet energy transfer from triplet chlorophyll is considerably higher than from triplet carbonyl.

Superoxide Anion Radical
Superoxide anion radical is formed by the one-electron reduction of O 2 on the PSII electron acceptor side (Figure 2). Pheophytin (Pheo D1 •− ), tightly bound plastosemiquinone (Q A •− ), loosely bound plastosemiquinones (Q B •− or Q C •− ), free PQ (PQ •− ), and ferrous iron of LP form of cyt b 559 were proposed to serve as electron donors to O 2 (Ananyev et al., 1994;Cleland and Grace, 1999;Pospíšil et al., 2004Pospíšil et al., , 2006Yadav et al., 2014). As Pheo D1 •− has highly negative redox potential, the reduction of O 2 by Pheo D1 •− is thermodynamically feasible; however, its short lifetime makes the diffusion limited reduction of O 2 less reasonable. Contrary, plastosemiquinones (Q A •− , Q B •− ) does not fulfill thermodynamic criteria due to their more positive redox potential, whereas they accomplish the kinetic criteria due their long lifetime. However, due to the different concentration of O 2 and O 2 •− , the standard redox potential of O 2 /O 2 •− redox couple is shifted according Nernst equation to more positive and thus the reduction of O 2 by plastosemiquinones becomes feasible (Pospíšil, 2009). The observation that exposure of isolated D1/D2/cyt b 559 complexes which lacks Q A to high light causes a significant rate of cytochrome (III) reduction revealed that Pheo D1 •− has capability to reduces O 2 . The detection of O 2 •− in isolated thylakoids by a voltammetric method showed O 2 •− production by the tightly bound plastosemiquinone Q A •− (Cleland and Grace, 1999). Experimental evidence has been recently provided on the reduction of O 2 by the loosely bound plastosemiquinones (Yadav et al., 2014). The authors demonstrated that plastosemiquinone is formed by the oneelectron reduction of plastoquinone at the Q B site and the oneelectron oxidation of plastoquinol by cyt b 559 at the Q C site. Apart to cofactors involved in the linear transport, the ferrous heme iron of LP form of cyt b 559 was shown to reduce O 2 forming O 2 •− (Pospíšil et al., 2006). It has been demonstrated that PsbS knock-out rice mutants produced more O 2 •− compared to WT under high light (Zulfugarov et al., 2014). The authors proposed that the lack of PsbS may cause shift in the midpoint redox potential of Q A /Q A •− redox couple to more negative value and thus enhance O 2 •− production by Q A •− . The D1 protein phosphorylation which is associated with the migration of damaged PSII complexes from the grana to the stroma lamellae during D1 protein repair cycle was shown to decrease O 2 •− production (Chen et al., 2012). The author proposed that the D1 protein phosphorylation causes conformation change of D1 protein and thus modifies the binding of loosely bound plastosemiquinone to Q B site. Consequently, the alternation of Q B site brings about the decrease in O 2 •− formed by the loosely bound plastosemiquinone Q B •− . In agreement with this proposal, it has been recently demonstrated that O 2 •− production is enhanced in STN8 kinase knock-out rice mutants under high light (Poudyal et al., 2016). It has been proposed that enhancement in O 2 •− production is due to the absence of conformational changes caused by STN8 kinaseinduced phosphorylation. Using PsbY knock-out Arabidopsis plants, it has been shown that redox potential property of cyt b 559 is controlled by PsbY protein (von Sydow et al., 2016). It has to be explored whether PsbY protein controls O 2 •− production.

Hydrogen Peroxide
Hydrogen peroxide is formed by the one-electron reduction of O 2 •− and the two-electron oxidation of H 2 O on the PSII electron acceptor and donor sides, respectively (Figure 2). Hydrogen peroxide formation by the one-electron reduction of O 2 •− occurs as dismutation or is maintained by plastosemiquinone. In the dismutation, two O 2 •− are simultaneously reduced and oxidized forming H 2 O 2 and O 2 , respectively. In the spontaneous dismutation, the interaction of two O 2 •− is restricted due to repulsion of the negative charge on the molecule, whereas the interaction of the protonated form of superoxide, hydroperoxyl radical (HO 2 • ), either with O 2 •− or HO 2 • is feasible. Spontaneous dismutation has been recently monitored by real-time detection of H 2 O 2 in PSII membrane under high light using highly sensitive and selective osmium-horseradish modified electrode (Prasad et al., 2015). In the enzymatic dismutation, reduction and oxidation of O 2 •− is associated with the redox change of the redox active metal center which serves as a superoxide oxidase (SOO) and superoxide reductase (SOR), respectively. It was demonstrated that the interaction of O 2 •− with the non-heme iron results in the oxidation of the ferrous iron and the formation of ferric-peroxo species which is protonated to ferric-hydroperoxo species (bound peroxide; Pospíšil et al., 2004) (Figure 3A). Evidence has been provided that the ferric and ferrous heme irons of cyt b 559 exhibit the SOO and the SOR activities, respectively (Tiwari and Pospíšil, 2009;Pospíšil, 2011). Apart to dismutation, free PQ •− in PQ pool was proposed to participate in H 2 O 2 formation. Hydrogen peroxide was shown to be formed by reduction of O 2 •− by free PQ •− (Borisova-Mubarakshina et al., 2015). The authors showed that H 2 O 2 formed in PQ pool regulates the size of PSII antenna complex at high light. Furthermore, evidence has been provided that H 2 O 2 might be formed by reduction 1 O 2 of by PQH 2 (Khorobrykh et al., 2015). It was demonstrated that 1 O 2 generated by photosensitizer Rose Bengal interacts with PQH 2 forming H 2 O 2 . The authors proposed that H 2 O 2 formed by reduction of 1 O 2 by PQH 2 in the thylakoid membrane might cause dimerization of the protein kinase STN7 and thus activates the enzyme.
Hydrogen peroxide formation by the two-electron oxidation of H 2 O is maintained by the Mn 4 O 5 Ca cluster when the complete four-electron oxidation of H 2 O to O 2 is limited. Whereas all four manganese are redox active in four-electron oxidation of H 2 O to O 2 , the incomplete oxidation of H 2 O to H 2 O 2 involves two redox active manganese. The two-electron oxidation of H 2 O has been proposed to involve the transition from either S 2 to S 0 state or S 1 to S −1 state. Evidence has been provided that release of chloride from its binding site near to the Mn 4 O 5 Ca cluster enhanced H 2 O 2 formation (Bradley et al., 1991;Fine and Frasch, 1992;Arato et al., 2004). A nucleophilic attack of hydroxo group on oxo group was proposed as an attractive model for formation of hydroperoxo species. It is proposed that nucleophilic attack of hydroxo group coordinated to Mn(4) and Cl(2) and oxo group coordinated to Ca forms hydroperoxo intermediate (Figure 3B). The hydroxo group is formed by deprotonation of the H 2 O substrate coordinated to to Mn(4) and Cl(2), whereas the oxo group is formed by double deprotonation of H 2 O substrate coordinated to Ca. A nucleophilic attack of manganese-coordinated hydroxo group on the calcium-coordinated electrophilic oxo group forms a peroxide intermediate that substitutes Cl(2) in coordination to Mn(4). Chloride controls accessibility of H 2 O substrate to Mn(4) and the nucleophilicity of hydroxo group and thus interaction of hydroxo and oxo groups. Water substrate, which serves as a precursor for the hydroxo group, enters into the catalytic site, when the Cl(2) binding site becomes opened to the solvent H 2 O due to its release.

Hydroxyl Radical
Hydroxyl radical is formed by the one-electron reduction of H 2 O 2 formed on the both PSII electron acceptor and donor sides (Figure 2). Hydroxyl radical formation by the one-electron reduction of free H 2 O 2 and bound peroxide on the PSII electron acceptor side was shown to be maintained by free iron and the non-heme iron, respectively (Pospíšil et al., 2004). The authors demonstrated that the reduction of bound peroxide (ferric ironhydroperoxo intermediate) formed by the interaction of O 2 •− with the ferrous non-heme iron forms HO • via ferric iron-oxo intermediate ( Figure 3A).
Hydroxyl radical formation by the one-electron reduction of H 2 O 2 on the PSII electron donor side is likely to be maintained by manganese. From thermodynamic point of view, the reduction of H 2 O 2 by manganese is not feasible. It was proposed that the reduction of H 2 O 2 by manganese becomes thermodynamically more favorable by (1) the coordination of manganese to the protein due to the decrease in the redox potential of manganese and (2) the pH decrease in the lumen due to the increase in the standard redox potential of H 2 O 2 /HO • redox couple (Pospíšil, 2012). It was demonstrated that PSII membranes depleted by chloride shows higher HO • formation compared to control PSII membranes . Based on the observation that HO • formation was not completely suppressed by exogenous SOD, the authors proposed that HO • is formed by reduction of H 2 O 2 produced by the incomplete water oxidation on the PSII electron donor side.

HIGH TEMPERATURE
When PSII is exposed to high temperature, decline in the PSII activity denoted as heat inactivation occurs (Mathur et al., 2014). Heat inactivation occurs on the both PSII electron acceptor and donor sides. On the PSII electron donor side, heat inactivation is associated with the inhibition of water oxidation accompanied with release of PsbO, PsbP, and PsbQ proteins, calcium, chloride, and manganese from their binding sites (Coleman et al., 1988;Enami et al., 1994;Pospíšil et al., 2003;Barra et al., 2005). On the PSII electron acceptor side, heat inactivation is linked to the inhibition of electron transport from Q A to Q B (Pospíšil and Tyystjarvi, 1999). The authors demonstrated that increase in the midpoint redox potential of Q A /Q A •− redox couple is responsible for the inhibition of Q A to Q B electron transport. Contrary to high light, ROS formation at high temperature is not driven by energy absorbed by chlorophylls; however, it is associated with heat-induced structural and functional changes in the thylakoid membrane. On the PSII electron acceptor side, 1 O 2 is formed decomposition of high energy intermediates formed by lipid peroxidation. On the PSII electron donor side, incomplete H 2 O oxidation forms H 2 O 2 which is reduced by manganese to HO • via Fenton reaction.

Singlet Oxygen
Singlet oxygen is formed by the triplet-triplet energy transfer from 3 L * to O 2 produced by the decomposition of high energy intermediates, dioxetane, or tetroxide, formed during lipid peroxidation Pospíšil and Prasad, 2014). The observation that elimination of HO • formation by mannitol did not suppress 1 O 2 formation revealed that lipid peroxidation is unlikely initiated by HO • (Pospíšil et al., 2007). More recently, it has been demonstrated that inhibition of lipoxygenase by catechol and caffeic acid in Chlamydomonas cells prevented 1 O 2 formation (Prasad et al., 2016). Singlet oxygen was proposed to be generated at the lipid phase near the Q B site (Yamashita et al., 2008). It was pointed that PQH 2 formed by reduction of PQ by stromal reducing compound might cause ROS production which can damage D1 protein (Marutani et al., 2012).

Hydrogen Peroxide
Hydrogen peroxide is formed by the two-electron oxidation of H 2 O on the PSII electron donor side (Figure 4). It was proposed that the release of extrinsic proteins (PsbO, PsbP, and PsbQ) leads to the inadequate accessibility of water to the Mn 4 O 5 Ca cluster and consequently to the formation of H 2 O 2 (Thompson et al., 1989). Indeed, it was demonstrated using the amplex red fluorescent assay that exposure of PSII membranes to high temperature (40 • C) results in H 2 O 2 formation (Yadav and Pospíšil, 2012b However, when chloride is released from its binding site, the delivery of H 2 O to the Mn 4 O 5 Ca cluster is unrestricted and incomplete oxidation of O 2 to H 2 O 2 occurs. Crystal structure of PSII from cyanobacteria Thermosynechococcus vulcanus reveals that two chlorides are located at distances of 6.67 and 7.40 Å from the Mn 4 O 5 Ca cluster (Umena et al., 2011). To avoid oxidation of nearby amino acid, diffusion of H 2 O 2 into the lumen has to be restricted to the channels. As H 2 O 2 is larger polar molecule similar to H 2 O, it seems to be likely that H 2 O 2 diffuse into the lumen via water channels. However, when H 2 O 2 leaks from the water channels, it might interact with manganese and formed HO • .

Hydroxyl Radical
Hydroxyl radical is formed by the one-electron reduction of H 2 O 2 formed on the PSII electron donor side (Figure 4). It was demonstrated by the EPR spin trapping spectroscopy that the exposure of PSII membranes to high temperature results in HO • formation (Pospíšil et al., 2007). The authors showed that HO • production is completely suppressed by exogenous catalase and metal chelator desferal revealing that HO • is formed via the metal-catalyzed Fenton reaction. Furthermore, the observation that the addition of exogenous calcium and chloride prevented HO • formation reveals that HO • is produced by the Mn 4 O 5 Ca cluster. This proposal was confirmed by the observation that no HO • formation was observed in PSII membranes deprived by the Mn 4 O 5 Ca cluster (Yamashita et al., 2008). As the replacement of chloride by acetate at its binding site near to the Mn 4 O 5 Ca cluster and the blockage of water channel prevented HO • formation in a similar manner as H 2 O 2 formation, it was assumed that chloride plays a crucial role in HO • formation (Yadav and Pospíšil, 2012b). The authors proposed that H 2 O 2 formed by the incomplete H 2 O oxidation is reduced to HO • via the Fenton reaction mediated by free manganese released from the Mn 4 O 5 Ca cluster. The release of manganese from its binding site at high temperature was reported using atomic absorption (Nash et al., 1985) and EPR (Coleman et al., 1988;Pospíšil et al., 2003) spectroscopy. Detailed study using X-ray absorption spectroscopy showed that decomposition of the Mn 4 O 5 Ca cluster occurs in two steps (Pospíšil et al., 2003). In the first step, two manganese are released Frontiers in Plant Science | www.frontiersin.org from their binding sites into the lumen remaining two manganese connected by a di-µ-oxo bridge, whereas in the second phase the remaining two manganese are liberated form PSII.

Role of ROS in Retrograde Signaling
Both 1 O 2 and H 2 O 2 formed in the thylakoid membrane were proposed to be involved in retrograde signaling (Dietz et al., 2016). Role of 1 O 2 in acclimation and programmed cell death was demonstrated in green algae (Erickson et al., 2015) and higher plants (Triantaphylides and Havaux, 2009;Laloi and Havaux, 2015). In higher plants, fluorescent (flu) and chlorina 1 (ch1) Arabidopsis mutants were advantageously used due to their high capability to form 1 O 2 . It was proposed that the 1 O 2 level determines whether acclimation response or programmed cell death is triggered (Laloi and Havaux, 2015). At low 1 O 2 level, acclimation response is mediated by β-cyclocitral formed by oxidation of β-carotene (Ramel et al., 2013a;Havaux, 2014). It was demonstrated that exposure of WT Arabidopsis plants to β-cyclocitral caused expression of 1 O 2 related gene (Ramel et al., 2012). In agreement with this finding, it was shown that concentration of β-cyclocitral is enhanced in ch1 Arabidopsis plants under acclimation (Ramel et al., 2013b). Further, evidence was provided on the role of jasmonic acid in acclimation response. It was demonstrated that jasmonatedeficient Arabidopsis mutant (delayed-dehiscence 2) was more resistant to light and jasmonate biosynthesis was pronouncedly lowered under acclimation (Ramel et al., 2013b). Based on these observations, the authors proposed that downregulation of jasmonate biosynthesis plays a crucial role in the triggering of acclimation response (Ramel et al., 2013c). At high 1 O 2 level, programmed cell death is dependent on the plastid proteins EXECUTER1 (EX1) and EXECUTER2 (EX2; Lee et al., 2007) and OXIDATIVE SIGNAL INDUCIBLE1 (OXI1) encoding an AGC kinase (Shumbe et al., 2016). Several lines of evidence on the involvement of EX1 and EX2 in programmed cell death were provided using flu Arabidopsis mutant (Lee et al., 2007). In this mutant, 1 O 2 is formed by triplettriplet energy transfer from the triplet chlorophyll precursor protochlorophyllide to O 2 (op den Camp et al., 2003). Even if EX1 and EX2 are located in chloroplast, it was proposed that jasmonic acid formed by 1 O 2 -initiated lipid peroxidation mediates genetically controlled programmed cell death response via these two plastid proteins (Przybyla et al., 2008). The initiation of 1 O 2 signaling has been recently demonstrated close to EX1 in the grana margins nearby the site of chlorophyll synthesis and 1 O 2 formation (Wang et al., 2016). As 1 O 2 signaling depends on the FstH protease, the authors proposed that 1 O 2 signaling is linked to D1 repair cycle. Apart to EX1 and EX2, it has been shown recently that OXI1 kinase is involved in 1 O 2 signaling in ch1 Arabidopsis mutant (Shumbe et al., 2016). In this mutant, 1 O 2 is formed by triplet-triplet energy transfer from the triplet chlorophyll formed in PSII to O 2 (Krieger-Liszkay, 2005). As OXI1 kinase is localized at the cytosol at the cell periphery or in the nucleus, it seems to be likely that oxylipins mediate signal transduction from chloroplast to cytosol (Shumbe et al., 2016).
Hydrogen peroxide formed under high light was demonstrated to play a crucial role in signaling associated with acclimation and programmed cell death (Foyer and Noctor, 2009;Karpinski et al., 2013;Gollan et al., 2015). It is well established that H 2 O 2 regulates expression of genes by the activation of protein kinase signaling pathways. It was proposed that precursor of jasmonic acid, 12-oxo phytodienoic acid (OPDA), mediates signal transduction from chloroplast to cytosol (Tikkanen et al., 2014). It has been recently demonstrated that H 2 O 2 formed in PQ pool triggers signal transduction from the chloroplast to the nucleus via protein kinase signaling pathways leading to the regulation of the PSII antenna size during the acclimation response (Borisova-Mubarakshina et al., 2015).
Our knowledge on the involvement of ROS in retrograde signaling at high temperature is highly limited. While the physiological relevance of light-induced 1 O 2 to acclimation and programmed cell death is described to some extent, no evidence was provided on the role of 1 O 2 formed under high temperature to plant stress response. However, it seems to be likely that 1 O 2 might oxidize lipid, protein or pigment forming specific oxidation products and thus initiates signal transduction from the chloroplast to the nucleus in the signaling cascade pathway. Contrary to 1 O 2 , H 2 O 2 was shown to be an important component in heat stress-activated gene expression. Hydrogen peroxide was demonstrated to be involved in the synthesis of heat shock proteins (Volkov et al., 2006). More experimental data are required to pronouncedly progress our understanding of multiple signaling pathways involved the in response to heat stress.

Role of ROS in Oxidative Damage
At high light, proteins and lipids might be oxidized by ROS formed in PSII. PSII proteins were evidenced to be oxidatively modified in the following order D1 > D2 > Cyt b559 > CP43 > CP47 > Mn 4 O 5 Ca cluster (Komenda et al., 2006). Amino acid oxidation at the lumen exposed AB-loop of D1 protein forms 24 kDa C-terminal and 9 kDa N-terminal fragments, whereas amino acid oxidation in the stromally exposed D-de loop of the D1 protein form 23-kDa N-terminal and 9-kDa C-terminal fragments (Edelman and Mattoo, 2008). Identification of naturally oxidized amino acid in D1 protein using mass spectrometry was shown nearby to the site of ROS production (Sharma et al., 1997;Frankel et al., 2012Frankel et al., , 2013. Whereas D1 protein oxidation was pronouncedly studied in vitro, limited evidence was provided on D1 protein oxidation in vivo (Shipton and Barber, 1994;Lupinkova and Komenda, 2004). Regardless of a broad range of evidence on PSII protein oxidation obtained in vitro, the plausibility of these processes in vivo has to be clarify. An efficient repair cycle for D1 protein, which includes proteolytic degradation of damaged D1 protein and its replacement with a newly synthetized D1 copy is essential for maintaining the viability of PSII Mulo et al., 2012;Jarvi et al., 2015). Apart to involvement of ROS in PSII protein damage under high light, ROS were shown to suppress the synthesis de novo of proteins with the elongation step of translation as primary target (Nishiyama et al., 2006). However, considering the limited ROS diffusion, it seems to be more likely that ROS produced in the stroma might oxidize the translational elongation factors involved in D1 repair cycle. Unbound chlorophylls released to the stroma from their binding sites during PSII protein damage or chlorophyll precursors during chlorophyll synthesis are likely candidates for 1 O 2 formation due to the lack of effective quenching of triplet excitation energy by carotenoids. To avoid 1 O 2 formation, unbound chlorophylls might be temporarily coordinated to early light-induced proteins (ELIPs). In agreement with this proposal, it was demonstrated that small CAB-like proteins prevent 1 O 2 formation during PSII damage, most probably by the binding of unbound chlorophylls released from the damaged PSII complexes (Sinha et al., 2012). Lipids associated with membrane proteins were shown to be oxidized by ROS. The initiation of lipid peroxidation by 1 O 2 comprises the insertion of 1 O 2 to double bond of polyunsaturated fatty acid, whereas HO • initiates lipid peroxidation by hydrogen abstraction from polyunsaturated fatty acid. It has been demonstrated that primary (LOOH) and secondary (LOH, RCS, and electronically excited species) lipid peroxidation products are formed at high light. Formation of hydroxy fatty acid was demonstrated in Arabidopsis plants (Triantaphylides et al., 2008). The authors showed that oxidation of polyunsaturated fatty acid by 1 O 2 leads to formation of LOOH which further forms LOH isomers (10-HOTE and 15-HOTE).
At high temperature, limited evidence was provided on the oxidation of proteins and lipids by ROS. It was demonstrated that exposure of thylakoid membranes to high temperature caused cleavage of D1 protein forming 9 kDa C-terminal and 23 kDa N-terminal fragments (Yoshioka et al., 2006). The authors demonstrated that FtsH protease is involved in the cleavage of the D1 protein at high temperature. Furthermore, it was reported that 1 O 2 formed at Q B site by the recombination of LOO • formed by the lipid peroxidation caused the D1 protein degradation by the interaction with D-de loop of the D1 protein in a similar manner as under high light (Yamashita et al., 2008). As experimental evidence for oxidative damage of PSII protein by endogenous ROS was obtained predominantly in vitro, it is unclear whether the PSII protein oxidation at high temperature occurs in vivo. Apart to involvement of ROS in PSII protein oxidation, the inhibition of de novo protein synthesis by ROS was proposed at high temperature (Allakhverdiev et al., 2008). Lipid peroxidation is associated with formation of RCS. It was demonstrated that malondialdehyde is formed in Arabidopsis plants exposed to heat stress (Yamauchi et al., 2008).

CONCLUSION AND PERSPECTIVES
Under environmental conditions, abiotic stresses adversely affect plant growth and survival. The impact of high light on the photosynthetic apparatus is considered to be of particular significance as light reactions of photosynthesis are inhibited prior to other cell functions are impaired. However, under environmental conditions, plants are exposed to combination of multiple stresses. High light stress is often associated with high temperature causing global warming which is one of the most important characteristics of accelerated climatic changes. Extensive research over the last 10 years focused on the structural and functional changes of the photosynthetic complexes in response to high light, high temperature or their combination. The exploration of molecular mechanism of ROS production by PSII helps to understand the adaptive processes by which plants cope with high light and high temperature stresses.