Oxygen and ROS in Photosynthesis

Oxygen is a natural acceptor of electrons in the respiratory pathway of aerobic organisms and in many other biochemical reactions. Aerobic metabolism is always associated with the formation of reactive oxygen species (ROS). ROS may damage biomolecules but are also involved in regulatory functions of photosynthetic organisms. This review presents the main properties of ROS, the formation of ROS in the photosynthetic electron transport chain and in the stroma of chloroplasts, and ROS scavenging systems of thylakoid membrane and stroma. Effects of ROS on the photosynthetic apparatus and their roles in redox signaling are discussed.


Introduction
Molecular oxygen (O 2 ) is a natural acceptor of electrons in the respiratory pathway of aerobic organisms and in many other biochemical reactions. In its ground state, O 2 has two unpaired electrons with parallel spins in two separate π antibonding orbitals. Thus, ground-state O 2 is a triplet diradical. According to Pauli's principle, O 2 reacts slowly with many biomolecules because spin restriction causes a kinetic barrier, as almost all biomolecules are in the singlet state, having paired electrons with opposite spins. The kinetic barrier of O 2 is removed either by spin inversion of one unpaired electron or by a partial reduction in O 2 . Spin inversion requires the absorption of energy and converts the triplet state of O 2 to the singlet state of molecular oxygen ( 1 O 2 ). All other forms of active oxygen are produced via an electron transfer mechanism. 1 O 2 and partially reduced forms of oxygen have a higher reactivity towards many organic molecules than the ground state of oxygen and they are collectively called reactive oxygen species (ROS).
ROS can be classified as radical and non-radical species. In addition, ROS can be roughly divided into free ROS, small molecules composed of oxygen and hydrogen only, and incorporated ROS, in which oxygen is bound to other molecules to form reactive oxygen derivatives. Furthermore, a family of reactive species containing nitrogen moieties associated with oxygen are classified as reactive nitrogen species, and reactive oxygen derivatives like lipid peroxyl radicals (LOO • ) can be classified as both ROS and reactive lipid species. The free ROS are 1 Table 1 presents the most important reactive species containing active oxygen according to this classification.
The second reaction is more common because singlet excited states ( 1 S*) are usually short-lived and because only a few dye molecules have a large enough energy gap between the 1 S* and triplet states ( 3 S) to convert O 2 to 1 O 2 [15]. 3 Chl reacts rapidly with O 2 with a second-order rate constant close to 10 9 M −1 s −1 , and the relative quantum yield of 1 O 2 generation by chlorophyll a (Chl a) was around 80% when meso-tetraphenylporphyrin and tetra(p-sulfophenyl) porphyrin were used as standards [16]. The spin transition O 2 → 1 Σ + g O 2 is associated with the absorption band of gaseous O 2 at 762 nm. Absorption at 1268 nm, in turn, was found for the transition O 2 → 1 ∆gO 2 in liquids and in the atmosphere [17].
1 O 2 can be produced in the reaction of O 2 with an R= 3 O* (Reaction (10)), which can be formed by the decomposition of ROOOOR (Reaction (9)) [33]. Both excited singlet states of oxygen are metastable and can lose excitation energy via radiative and non-radiative pathways. The latter is physical quenching of 1 O 2 . The radiative deactivation is the transition of 1 ∆gO 2 to O 2 associated with light emission (hν; Reaction (14)). 1 The phosphorescence spectrum has a major maximum at 1268 nm [37]. The phosphorescence is extremely weak, as the deactivation of 1 O 2 mostly proceeds non-radiatively due to the collision of 1 O 2 with another molecule. The quantum yield of luminescence is from 10 −6 to 10 −3 [32]. Non-radiative deactivation mechanisms include electronical-to-vibrational energy transfer, charge-transfer-induced quenching and electronic energy transfer. In the deactivation of 1 Σ + gO 2 and 1 ∆gO 2 by an electronic-to-vibrational process, the excitation energy of 1 O 2 is converted into vibration of the O 2 molecule and a quencher molecule Qr (Reaction (15)). 1 (15) where E mn is the energy difference between the reactants and the products and v is the vibrational energy level of a molecule; m and n are vibrational modes. The deactivation of 1 O 2 by collisions of 1 O 2 with other molecules limits the lifetime of 1 O 2 in many solvents. The lifetime of 1 O 2 for many organic solvents is within 8-100 µs. The substitution of hydrogen with deuterium in the solvent molecule leads to a significant increase in the lifetime of 1 O 2 , usually by a factor of ten or more [17,[38][39][40]. The second order rate constant for the deactivation of 1 O 2 via an electronic-to-vibrational process varies widely, from 10 −2 to 10 6 M −1 s −1 [17].
Plants 2020, 9, 91 5 of 63 In addition to the electronic-vibrational non-radiative deactivation, 1 O 2 can be deactivated via charge-transfer-induced quenching (Reaction (16)) and an electronic energy transfer mechanism (Reaction (17)). 1 where A is an acceptor. Molecules with high triplet energies (more than 94 kJ mol −1 ) and low oxidation potential (midpoint redox potential (E m ) around 1.9 V vs. Normal Hydrogen Electrode (NHE)) can deactivate 1 O 2 with the charge-transfer mechanism. Second-order rate constants for deactivation of 1 O 2 via the charge-transfer mechanism are within 10 3 to 10 9 M −1 s −1 [17]. In the charge-transfer mechanism, the 1 (O 2 A) complex finally dissociates to A and O 2 without charge separation. Electronic energy-transfer quenching of 1 O 2 occurs via the interaction of molecules with a lower triplet state energy than the energy of 1 O 2 . The deactivation of 1 O 2 via the electronic energy-transfer mechanism is very efficient and its second-order rate constant is close to the diffusion-controlled limit. Carotenoids including β-carotene and lutein are the most efficient quenchers of 1 O 2 , and the second order rate constant for many carotenoids is about 10 10 M −1 s −1 [17,41].

Chemical Reactions of 1 O 2
The term "chemical deactivation" of 1 O 2 can be applied to reactions in which the products have less reactivity and toxicity towards cell metabolism than 1 O 2 . The redox potential relative to NHE for the pair 1 [42]. 1 O 2 is an electrophilic agent and reacts with electron-rich organic molecules via three well-known mechanisms. The ene reaction (Alder-ene reaction) is associated with the formation of a hydroperoxide (Reaction (18)).
Plants 2020, 9, x FOR PEER REVIEW 5 of 61 charge-transfer mechanism are within 10 3 to 10 9 M −1 s −1 [17]. In the charge-transfer mechanism, the 1 (O2 A) complex finally dissociates to A and O2 without charge separation. Electronic energy-transfer quenching of 1 O2 occurs via the interaction of molecules with a lower triplet state energy than the energy of 1 O2. The deactivation of 1 O2 via the electronic energy-transfer mechanism is very efficient and its second-order rate constant is close to the diffusion-controlled limit. Carotenoids including βcarotene and lutein are the most efficient quenchers of 1 O2, and the second order rate constant for many carotenoids is about 10 10 M −1 s −1 [17,41].

Chemical Reactions of 1 O2
The term "chemical deactivation" of 1 O2 can be applied to reactions in which the products have less reactivity and toxicity towards cell metabolism than 1 O2. The redox potential relative to NHE for the pair 1 O2/O2 •− is 0.65 V [42]. 1 O2 is an electrophilic agent and reacts with electron-rich organic molecules via three well-known mechanisms.
Plants 2020, 9, x FOR PEER REVIEW 5 of 61 charge-transfer mechanism are within 10 3 to 10 9 M −1 s −1 [17]. In the charge-transfer mechanism, the 1 (O2 A) complex finally dissociates to A and O2 without charge separation. Electronic energy-transfer quenching of 1 O2 occurs via the interaction of molecules with a lower triplet state energy than the energy of 1 O2. The deactivation of 1 O2 via the electronic energy-transfer mechanism is very efficient and its second-order rate constant is close to the diffusion-controlled limit. Carotenoids including βcarotene and lutein are the most efficient quenchers of 1 O2, and the second order rate constant for many carotenoids is about 10 10 M −1 s −1 [17,41].

Chemical Reactions of 1 O2
The term "chemical deactivation" of 1 O2 can be applied to reactions in which the products have less reactivity and toxicity towards cell metabolism than 1 O2. The redox potential relative to NHE for the pair 1 O2/O2 •− is 0.65 V [42]. 1 O2 is an electrophilic agent and reacts with electron-rich organic molecules via three well-known mechanisms.
Plants 2020, 9, x FOR PEER REVIEW 5 of 61 charge-transfer mechanism are within 10 3 to 10 9 M −1 s −1 [17]. In the charge-transfer mechanism, the 1 (O2 A) complex finally dissociates to A and O2 without charge separation. Electronic energy-transfer quenching of 1 O2 occurs via the interaction of molecules with a lower triplet state energy than the energy of 1 O2. The deactivation of 1 O2 via the electronic energy-transfer mechanism is very efficient and its second-order rate constant is close to the diffusion-controlled limit. Carotenoids including βcarotene and lutein are the most efficient quenchers of 1 O2, and the second order rate constant for many carotenoids is about 10 10 M −1 s −1 [17,41].

Chemical Reactions of 1 O2
The term "chemical deactivation" of 1 O2 can be applied to reactions in which the products have less reactivity and toxicity towards cell metabolism than 1 O2. The redox potential relative to NHE for the pair 1 O2/O2 •− is 0.65 V [42]. 1 O2 is an electrophilic agent and reacts with electron-rich organic molecules via three well-known mechanisms.

Lifetime and Diffusion Distance of 1 O2
Solvents and other deactivating compounds play significant roles in controlling the lifetime of 1 O2 and the lifetime, in turn, determines both the diffusion distance and the ability of 1 O2 to react with other substances. The lifetimes of 1 O2 in a pure lipid membrane and in the thylakoid membrane have been estimated to be 7 µs and 70 ns, respectively [51]. Thus, the respective diffusion distances, approximated using the diffusion coefficient of O2, are 220 and 5.5 nm. The very short lifetime in the thylakoid membrane may be caused by a high concentration of compounds deactivating 1 O2. In a nerve cell, the lifetime of 1 O2 is about 200 ns, which leads to a diffusion distance of about 270 nm [52]. The lifetimes and diffusion distances of 1 O2 in different tissues have been recently reviewed [23].

Lifetime and Diffusion Distance of 1 O2
Solvents and other deactivating compounds play significant roles in controlling the lifetime of 1 O2 and the lifetime, in turn, determines both the diffusion distance and the ability of 1 O2 to react with other substances. The lifetimes of 1 O2 in a pure lipid membrane and in the thylakoid membrane have been estimated to be 7 µs and 70 ns, respectively [51]. Thus, the respective diffusion distances, approximated using the diffusion coefficient of O2, are 220 and 5.5 nm. The very short lifetime in the thylakoid membrane may be caused by a high concentration of compounds deactivating 1 O2. In a nerve cell, the lifetime of 1 O2 is about 200 ns, which leads to a diffusion distance of about 270 nm [52]. The lifetimes and diffusion distances of 1 O2 in different tissues have been recently reviewed [23].
where DHA and PQ are dehydroascorbate and plastoquinone, respectively. The second-order rate constant for the reaction of 1 O 2 with AscH 2 depends on pH and varies from 10 5 M −1 s −1 to 10 8 M −1 s −1 [49]. Prenyllipids like PQH 2 -9 and α-tocopherol react with 1  Solvents and other deactivating compounds play significant roles in controlling the lifetime of 1 O 2 and the lifetime, in turn, determines both the diffusion distance and the ability of 1 O 2 to react with other substances. The lifetimes of 1 O 2 in a pure lipid membrane and in the thylakoid membrane have been estimated to be 7 µs and 70 ns, respectively [51]. Thus, the respective diffusion distances, approximated using the diffusion coefficient of O 2 , are 220 and 5.5 nm. The very short lifetime in the thylakoid membrane may be caused by a high concentration of compounds deactivating 1 (27) and (28)), which makes O 2 •− a much stronger deprotonation agent than would follow from its basicity.
which sum up to The equilibrium constant of Reaction (29) is about 10 9 [54]. Therefore, in the deprotonation process the pKa value of O 2 •− should be considered equivalent to a base with pKa 24 [53,54]. The E m of the  (30)).
O 2 •− can be formed in a potentially important equilibrium reaction with semiquinone anion radicals (Q •− ) with the formation of the respective quinone Q (Reaction (31)).
on pH because the protonation of O2 •− determines the rate. Dismutation can be considered as a twostep reaction: protonation of O2 •− , Reaction (41) and a radical-radical reaction between O2 •− and HO2 • or between two molecules of HO2 • -Reactions (42) and (43) The second-order rate constant of the dismutation of O2 •-has a maximum (10 8 M −1 s −1 ) at pH 4.8, equal to the pKa value of HO2 • . The rate constant decreases with increasing pH and becomes very low, around 0.3 M −1 s −1 , at alkaline pH. At physiological pH, the rate constant is about 10 5 M −1 s −1 [68]. The enzymatic dismutation of O2 •− is catalyzed by superoxide dismutase (SOD, EC 1.15.1.1). The SOD-catalyzed reaction proceeds as a sequence of oxidation and reduction of O2 •− by a metal ion (M) of the SOD enzyme, Reactions (44) and (45).
In addition to nucleophilic reactions with organic molecules, O2 •− can bind to both transition metals and to metal complexes. For example, in PSII, the interaction of O2 •− with a ferrous heme iron (47) O2 •− is a powerful nucleophile in aprotic medium and can be involved in nucleophilic reactions 1 with various organic compounds. O2 •− reacts with alkyl halides (RX), acyl halides and acyl 2 anhydrides to form ROO • intermediates through nucleophilic substitution reactions [55]. O2 •− can 3 add to positively charged carbon-carbon double bonds [71] and carbon-nitrogen double bonds [72].  19 (48) In addition to nucleophilic reactions with organic molecules, O 2 •− can bind to both transition metals and to metal complexes. For example, in PSII, the interaction of O 2 •− with a ferrous heme iron leads to the formation of a ferric-peroxo ((Fe 3+ )-OO − ) complex which can be protonated to a ferric-hydroperoxo ((Fe 3+ )-OOH) complex, Reaction (49) and (50) [72].
where L is a ligand. cell in DMF was found to be 76 min at 0 °C for 0.1 M O2 •− , and around 35 h for the O2 •− concentrations from 0.001 to 0.01 M [54]. In cells, the lifetime of O2 •− is efficiently controlled by SOD, and the lifetime will depend on SOD activity. In the periplasm of Escherichia coli, the lifetime of O2 •− was estimated to be longer than 0.6 s using the rate of O2 •− formation and the rate constant of its dismutation. The diffusion distance was calculated as 35 µm, assuming a general diffusion coefficient of small molecules of about 10 5 cm 2 s −1 [73].

Hydrogen Peroxide, H2O2
H2O2 is the result of two-electron reduction of O2 and considered a major biological ROS. In cells, H2O2 is mostly present in the neutral form because its pKa is 11.8. H2O2 is a strong, two-electron oxidant with a standard redox potential (E0′) of 1.32 V at pH 7.0. However, H2O2 reacts slowly or does not react with most biological molecules, including low-molecular-weight antioxidants, due to a high activation energy barrier [74]. Even if a reaction with H2O2 is thermodynamically favorable, it may be very slow.
Low-potential compounds reduce H2O2 with one electron, as the redox potential of H2O2/HO • is 0.3 V [12]. H2O2 can also act as an electrophile due to the polarizability of the O-O bond. H2O2 has a permanent dipole moment of 2.26 Debye. The O-O bond is relatively weak and susceptible to homolysis. H2O2 is decomposed by heating, radiolysis, photolysis, or by reaction with redox active transition metals [74].

Formation of H2O2.
Reduction of O2 to O2 •− followed by its dismutation (Reaction (5)) is the main pathway for the formation of H2O2 in cells.
H2O2 can be formed via oxidation of a quinol by O2. For example, hydroanthraquinone is widely used for the commercial synthesis of H2O2, Reaction (51) [75]. (51) H2O2 can be produced by a reaction of 1 O2 or O2 •-with an electron donor, like AscH2, Reactions (24) and (34), respectively [47,64].  (24) and (34), respectively [47,64]. The main reaction of 1 O 2 with PQH 2 in methanol was found to result in the formation of PQ and H 2 O 2 , Reaction (25) and the amount of H 2 O 2 produced was essentially the same as the amount of oxidized PQH 2 [48].
No direct formation of H 2 O 2 from H 2 and O 2 is expected in aerobic cells because the production of hydrogen requires anaerobic conditions [77]. H 2 is consumed by the bidirectional hydrogenase in green algae [78], but an enzyme-catalyzing Reaction (52) has not been found.

Reactions of H 2 O 2
Most biological molecules that do not bind transition metal ions do not react directly with H 2 O 2 . However, thiol and cysteine residues of proteins, as well as low-molecular-weight thiols, can directly react with H 2 O 2 [67]. The reaction of H 2 O 2 with thiols (RS) depends strongly on the pK a value of the thiol, because the reaction exclusively proceeds via the thiolate anion to form sulfenic acid (RSOH), Reaction (53). The rate constants of Reaction (53) range from 0.16 to 10 7 M −1 s −1 . Sulfenic acids have a lower pKa than the corresponding thiols [79]. Sulfenic acid can react with another thiol or H 2 O 2 to give the corresponding disulfide (RSSR) or sulfinic acid (RSO 2 − ), Reactions (54) and (55), respectively. However, the second-order rate constant of Reaction (55) is about 10 3 times lower than that of Reaction (54) [74,80].
The second-order rate constants of Reaction (53) for free GSH, cysteine and thioredoxin (TRX) are 0.89 M −1 s −1 , 2.9 M −1 s −1 and 1.05 M −1 s −1 , respectively [74,81]. However, H 2 O 2 can react efficiently with peroxiredoxins (PRX); the second order rate constant is 10 7 -10 8 M −1 s −1 [74] Most biological molecules that do not bind transition metal ions do not react directly with H2O2. However, thiol and cysteine residues of proteins, as well as low-molecular-weight thiols, can directly react with H2O2 [67]. The reaction of H2O2 with thiols (RS) depends strongly on the pKa value of the thiol, because the reaction exclusively proceeds via the thiolate anion to form sulfenic acid (RSOH), Reaction (53).
Carbonic anhydrase significantly accelerates Reaction (57) [83]. Transition metals (M) like iron and copper react with H 2 O 2 via the Fenton mechanism, in which the transition metal cleaves the O-O bond to form HO • and HO − , Reaction (58).
Rate constants of the Fenton reaction depend on the metal or metal complex and are in the 5-20 × 10 3 M −1 s −1 range [84][85][86][87]. In addition to Reaction (58), the interaction of H 2 O 2 with transition metals or metal complexes leads to the formation of a higher oxidation state of the metal as L-M(H 2 O 2 ) n+ , L-M (n+2)+ or L-MO (n+2)+ Reactions (59)-(61) respectively, where L is a ligand of the metal [74,88,89].
H 2 O 2 reacts rapidly with heme peroxidases, for example, myeloperoxidase and lactoperoxidase, with a second order rate constant in the range of 10 7 -10 8 M −1 s −1 [90]. The rate constant of the reaction of H 2 O 2 with ascorbate peroxidase (APX) was estimated to be 10 7 M −1 s −1 with K m for H 2 O 2 of 80 µM [91,92].
where A is a reductant, e.g., AscH 2 . Catalase (CAT)-dependent scavenging of H 2 O 2 occurs via a ping-pong mechanism, Reaction (65) and (66), where one H 2 O 2 molecule is used as an electron donor.  Another well-known means of HO • generation is through the photolysis of oxygen-containing species. In aqueous solution, the nitrate anion (NO 3 − ) can absorb UV radiation and produce HO • , Reactions ( (67) and (69). The formation of HO • is also observed upon photolysis of the nitrite ion (NO 2 − ), Reactions (68) and (69) [102,103]. Due to the requirement of short-wavelength UV radiation, this process does not occur in biological systems.
The photolysis of a H 2 O 2 molecule gives two HO • with a quantum yield of approximately 0.5 in aqueous solutions, Reaction (70). H 2 O 2 photolysis requires UV-C radiation because the molar absorption coefficient of H 2 O 2 is very low above 300 nm. H 2 O 2 photolysis is an effective way of generating HO • in aqueous solutions [102,104].
Another potential source of HO • is O 3 . The addition of an electron to an O 3 molecule leads to the decomposition of O 3 to HO • and O 2 via the formation of an ozonide anion radical [105]. O 3 can also be decomposed to O 2 and HO • via reduction by exited chlorophyll (Chl*), Reaction (71) [102].
However, O 3 has not been found inside plant cells.
HO • can be also produced in a radical-radical reaction of HO 2 • with RO 2 • , Reaction (75) [102].  (76); addition to double bonds with the formation of a hydroxylated radical (77); Another potential source of HO • is O3. The addition of an electron to an O3 molecule leads to the decomposition of O3 to HO • and O2 via the formation of an ozonide anion radical [105]. O3 can also be decomposed to O2 and HO • via reduction by exited chlorophyll (Chl*), Reaction (71) [102].
Formation of an aromatic-OH adduct due to a reaction of an aromatic compound with HO • is one of the methods for HO • detection with high-performance liquid chromatography-mass spectrometry. For example, HO • can react with phenylalanine to form isomers of tyrosine, Reaction (80) [107]. Isomers of tyrosine are rather stable and not normally present in proteins, and can serve as HO • traps in biological samples [108]. (77) electron transfer reactions leading to the formation of a neutral radical (78) or a cation radical (79) [106]; SCN − is the thiocyanate ion.
Another potential source of HO • is O3. The addition of an electron to an O3 molecule leads to the decomposition of O3 to HO • and O2 via the formation of an ozonide anion radical [105]. O3 can also be decomposed to O2 and HO • via reduction by exited chlorophyll (Chl*), Reaction (71) [102].
Formation of an aromatic-OH adduct due to a reaction of an aromatic compound with HO • is one of the methods for HO • detection with high-performance liquid chromatography-mass spectrometry. For example, HO • can react with phenylalanine to form isomers of tyrosine, Reaction (80) [107]. Isomers of tyrosine are rather stable and not normally present in proteins, and can serve as HO • traps in biological samples [108]. (79) Formation of an aromatic-OH adduct due to a reaction of an aromatic compound with HO • is one of the methods for HO • detection with high-performance liquid chromatography-mass spectrometry. For example, HO • can react with phenylalanine to form isomers of tyrosine, Reaction (80) [107]. Isomers of tyrosine are rather stable and not normally present in proteins, and can serve as HO • traps in biological samples [108]. Plants 2020, 9, x FOR PEER REVIEW 13 of 61 (80) HO • interacts with many metal (M) cations via an electron transfer Reaction (81), with a rate constant of ~10 8 M −1 s −1 [100].
HO • initiates lipid peroxidation, resulting in hydrogen abstraction from a pentyl group of an unsaturated fatty acid, and the formation of a radical that interacts with O2 to form an ROO • with a rate constant of ~10 8

Lifetime and Diffusion Distance of HO •
The lifetime of HO • in aqueous solution has been estimated to range from picoseconds to nanoseconds. The self-diffusion coefficient of HO • in water has been estimated to be 2.8 × 10 −5 cm 2 s −1 , and consequently the diffusion distance of HO • is a few molecular diameters from the site of origin [110,111]. (80) HO • interacts with many metal (M) cations via an electron transfer Reaction (81), with a rate constant of~10 8 M −1 s −1 [100].
HO • initiates lipid peroxidation, resulting in hydrogen abstraction from a pentyl group of an unsaturated fatty acid, and the formation of a radical that interacts with O 2 to form an ROO • with a rate constant of~10 8 M −1 s −1 [109], Reaction (82).  (81), with a rate constant of ~10 8 M −1 s −1 [100].
HO • initiates lipid peroxidation, resulting in hydrogen abstraction from a pentyl group of an unsaturated fatty acid, and the formation of a radical that interacts with O2 to form an ROO • with a rate constant of ~10 8

Lifetime and Diffusion Distance of HO •
The lifetime of HO • in aqueous solution has been estimated to range from picoseconds to nanoseconds. The self-diffusion coefficient of HO • in water has been estimated to be 2.8 × 10 −5 cm 2 s −1 , and consequently the diffusion distance of HO • is a few molecular diameters from the site of origin [110,111].

Lifetime and Diffusion Distance of HO •
The lifetime of HO • in aqueous solution has been estimated to range from picoseconds to nanoseconds. The self-diffusion coefficient of HO • in water has been estimated to be 2.8 × 10 −5 cm 2 s −1 , and consequently the diffusion distance of HO • is a few molecular diameters from the site of origin [110,111].

Production of ROS in Chloroplasts
Chloroplasts have a high metabolic activity accompanied with intensive formation of redox active compounds, which are able to react with oxygen to produce ROS. Most ROS production in the chloroplast occurs by the components of the light reactions. Photorespiration is responsible for 70% of total H 2 O 2 production in the leaves of C3 plants [112,113], but this reaction runs in peroxisomes outside of the chloroplast.

Formation of 1 O 2 in the Stroma
The chloroplast stroma is not considered as a significant source of 1 O 2 , although disintegration of the antenna complexes under stress conditions and disturbances in Chl synthesis and the accumulation of its precursors may lead to 1 O 2 production in the stroma [114]. The lack of FLU, a nuclear-encoded chloroplast protein that plays a key role during the negative feedback control of Chl biosynthesis, leads to the accumulation of protochlorophyllide in plastids and, consequently, to photosensitized generation of 1 O 2 [115]. It has been recently shown that lipoxygenase localized in the chloroplast is responsible for 1 O 2 formation [116]. Lipoxygenase initiates lipid oxidation to corresponding lipid peroxides, which decompose to lipid peroxyl radicals (LOO • ). LOO • can react with each other, forming a cyclic endoperoxide (dioxetane) intermediate. Dioxetane, in turn, can decompose via the Russel mechanism to form 1 O 2 , Reaction (9).
Theoretically, the Haber-Weiss mechanism (Reactions (3) and (4)) can cause the formation of 1 O 2 in the stroma, but the rate of this reaction is expected to be low due to very efficient scavenging of O 2 •− by chloroplasts [12] and the low rate constant of Reactions (3) and (4)  Ferredoxin (Fd) and free flavins (FL) and flavoenzymes are considered as the main stromal components involved in O 2 reduction and ROS formation. Fd is involved in electron transfer from the acceptor side of PSI to NADP + in a reaction catalyzed by Fd-NADP + reductase (FNR) (EC 1.18.1.2) [117]. Fd is a soluble 10 kDa protein [118] containing a 2Fe-2S center [119]. The leaf-type Fd from higher plants has an E m vs. NHE (at pH 8.0) from −390 to −425 mV [120]. The redox potential of Fd ox /Fd red is much more negative than the redox potential of  (83) and (84).  [125]. Recent studies show that O 2 reduction in a thylakoid suspension in the presence of Fd is a result of O 2 reduction by both a membrane-bound reductant and Fd. The distribution of electron flow from Fd and membrane-bound reductant to O 2 is sensitive to light intensity and NADP + but not to Fd concentration. Furthermore, Fd stimulates the reduction in O 2 by membrane-bound reductants [126]. Interestingly, NADP + very strongly inhibits O 2 reduction by Fd but stimulates O 2 reduction by a thylakoid-membrane-bound reductant [126]. These results suggest that Fd has a minor role in the direct reduction of O 2 in vivo.
Catalase, added to a suspension of illuminated thylakoid membranes, almost completely suppressed Fd-dependent O 2 consumption, suggesting that H 2 O 2 is the final product of O 2 reduction by Fd. This is clear from the well-known stoichiometry between O 2 consumption and O 2 evolution in isolated thylakoids when a reduction in O 2 occurs by electrons originally arising from water-splitting in PSII without any electron acceptors or ROS traps, Reactions (85)- (89). In this case, H 2 O 2 is produced via dismutation of O 2 •− [127,128].
The second-order rate constant of Reaction (92) is estimated to be only 2.5 × 10 2 M −1 s −1 [133,134]. The formation of FLH • (Reaction (92)) can also initiate complex autocatalytic FL oxidation. In solution, some amount of the FLH • is formed in a mixture of oxidized and reduced FL (FLH 2 ) via an equilibrium reaction, Reaction (94).
The next steps of the autocatalytic process are described by Reactions (96) (100) and (101).
The second-order rate constant for the reaction of O 2 with FLH • (Reaction (96)) is around 10 4 M −1 s −1 and that of the reaction of O 2 with FL •− (Reaction (97)) is much larger, around 10 8 M −1 s −1 [133]. The E m at pH 7 for FL/FLH • of FL mononucleotide is estimated to be −313 mV vs. NHE, which is more negative than the redox potential of O 2 /O 2 •− (−160 mV) in aqueous solution [134,136]. As predicted from the redox potentials, the reaction of FL •− with O 2 is thermodynamically favorable and the rate of oxidation of FLs by O 2 via autocatalytic mechanisms can strongly depend on both the stability and pKa value of FLH • . Free flavins can therefore be involved in the formation of ROS in the chloroplast stroma. Flavoenzymes can also be involved in the production of O 2 •− and H 2 O 2 in chloroplasts.
The reactivity of FLs in flavoenzymes is modulated by the protein environment of reduced FLs and the second-order rate constant of O 2 reduction by flavoenzymes varies from 2 M −1 s −1 to 2 × 10 6 M −1 s −1 [134]. The redox potential of flavoenzymes vs. NHE varies from −16 to −263 mV and −60 to −231 mV for FL/FLH • and for FLH • /FLH 2 , respectively [134]. The reactivity of flavoenzymes towards O 2 may differ by several orders of magnitude between flavoenzymes having similar redox potentials [134]. Such very high differences are due to the protein environment, which affects both O 2 movement and the binding of O 2 to the active site. In addition, the polarity of the protein environment in the active site can change the redox potential of becomes very negative (≈−600 mV vs. NHE) in a non-polar solvent [55]. The reduction in O 2 by a FL in a non-polar active site is thus unlikely [134]. The reactivity of flavoenzymes towards O 2 depends on the stabilization of the semiquinone in the active site because semiquinones show higher reactivity with O 2 than fully reduced FLs [133]. The reactivities of flavoenzymes with O 2 can be limited by a substrate that acts as a specific electron acceptor of the flavoenzyme. Some flavoenzymes, like glucose oxidase and xanthine oxidase, employ O 2 as a natural acceptor, forming both H 2 O 2 and O 2 •− with high efficiency [137,138]. The mechanism of O 2 reduction by flavoenzymes has recently been reviewed in detail [139].
Some stromal flavoenzymes, such as FNR, monodehydroascorbate reductase (MDAR), glutathione reductase (GR) and glycolate oxidase, can efficiency reduce O 2 to O 2 •− in the absence of the specific substrate [12,140]. The flavoenzymes are reduced and oxidized by their specific electron donors and acceptors with high rates. For example, MDAR is reduced by NAD(P)H with a second-order rate constant of 1.8 × 10 8 M −1 s −1 [141], and the reduced MDAR can be oxidized by monodehydroascorbate radical (AscH • and its anionic form (Asc •− ), abbreviated as MDA) with a second-order rate constant of 2.6 × 10 8 M −1 s −1 [142].  [147]. Thus, flavoenzymes may contribute to the high rates of O 2 photoreduction in chloroplasts.
In cyanobacteria [148,149], flavodiiron proteins reduce oxygen to water without ROS production. A substantial fraction of the total photosynthetic electron flow may be directed to this route [150]. Flavodiiron proteins have later been found from all oxygenic phototrophs except for angiosperms and some non-green algae (for review, see [151] (58)) if the scavenging of H 2 O 2 by the antioxidant enzymes is not fast enough for the efficient removal of H 2 O 2 . The Fenton reaction is possible in the chloroplast because up to 80% of cellular Fe in leaf cells is found in chloroplasts [153]. The involvement of free Fe in Fenton reaction is limited, since the Fe is stored in a redox inactive form as ferritin [154,155]. However, Fe can be activated and released from ferritin via interaction of ferritin with O 2 •− [156]. In addition to free transition metals, Fd can be involved in the production of HO • [157,158]. The second-order rate constant for the reaction of reduced Fd with H 2 O 2 was found to be 5 × 10 3 M −1 s −1 [121], which is two orders of magnitude higher than the second-order rate constant of HO • production in the Fenton reaction, 84 M −1 s −1 [84].

Formation of ROS in Thylakoid Membranes
The PETC employs three membrane protein complexes: PSI, PSII, their respective light-harvesting complexes (LHCI and LHCII), and the cytochrome b6/f complex (Cyt b6f, a plastoquinol-plastocyanin-oxidoreductase). Electron transfer between the complexes involves two mobile electron carriers, PQ and plastocyanin (PC). The liposoluble PQ mediates electron flow from PSII to Cyt b6f complex, and the water-soluble lumenal protein PC mediates electron flow from Cyt b6f to PSI. ROS are formed in several sites of the PETC, including PSII, PSI, the PQ pool and the light-harvesting complexes (LHC).

Formation of 1 O 2 in Thylakoids
The production of 1 O 2 in plants occurs mainly by the interaction of O 2 with excited states of Chls (where 1 Chl* and 3 Chl are the excited singlet and triplet states of Chl, respectively) via spin-conserved reactions, (Reactions (102) and (103)).
Reaction (102) is negligible because the lifetime of 1 Chl* is very short (~10 −8 s) [137]. The lifetime of 3 Chl is around 10 −3 s under anaerobic conditions [16,45]. In solution, the quenching of 3 Chl by O 2 mostly occurs via 1 O 2 generation with a second-order rate constant of 2 × 10 9 M −1 s −1 [16]. 3 Chl is formed both in the LHCs and in the reaction centers (RC) of PSII and PSI. In the LHCs, 3 Chl is formed by intersystem crossing (ISC) from 1 Chl* [114,159] and in the RCs by charge recombination. In PSII, the charge recombination between P680 + (the primary donor) and Q A − (bound quinone) produces 3 P680 [114,160]. 3 P680 is formed through a time-dependent "virtual triplet state" of the primary radical pair P680 + Pheophytin (Pheo) − [161]. A triplet state of P700, the primary donor of PSI, can also be formed via charge recombination [162].
Chls are mostly bound to LHCII and CP47 and CP43 proteins of PSII and the PSA A/B proteins of PSI. According to the high concentration of Chl in chloroplasts, around 60 mM [163], a significant formation of 3 Chl via ISC in LHCs should be observed. However, there is no experimental evidence for the production of 1 O 2 by the formation of 3 Chl in LHCs in vivo [114]. The formation of both 3 Chl and 1 O 2 has been observed in isolated LHCs. The formation of 1 O 2 was found in isolated LHCII with an electron paramagnetic resonance (EPR) measurement of 2,2,6,6-tetramethylpiperidine as a spin trap of 1 O 2 [164,165]. The appearance of long-lived 3 Chl in LHCs has been suggested to result from a small population of Chls that are substantially uncoupled from the matrix of LHC [166,167]. In a reconstructed Chl-protein complex, light-dependent 1 O 2 formation is lower by a factor of four compared to free Chl [168]. As the isolated protein does not contain pigments that would effectively quench 3 Chl or 1 O 2 , it was suggested that the low 1 O 2 formation is caused by the tight packing of Chl molecules inside the hydrophobic zone of the pigment-protein complex where the interaction of 3 Chl with O 2 is limited [168]. The same situation can probably be realized in LHCs where Chls are tightly packed. Furthermore, LHCs contain carotenoids that efficiently quench 3 Chl. In addition, the highly efficient transfer of excitation energy to the RC lowers the probability of ISC. Thus, 1 O 2 can only be formed in LHCs in sites where Chl is weakly bound to the protein matrix and 3 Chl cannot be efficiently quenched by carotenoids.
The main source of 1 O 2 appears to be O 2 reacting with 3 P680. As in the case of antenna complexes, the formation of 1 O 2 in the RC also depends on two factors: the lifetime of 3 P680 and the probability that O 2 reacts with 3 P680. Assuming that the accessibility of O 2 to 3 P680 is not changed significantly in different conditions; the yield of 1 O 2 generation in the RC of PSII is mainly limited by the rate of 3 P680 formation. The formation of 3 P680 proceeds via charge separation and charge recombination.
The formation of the excited singlet state 1 P680* is followed by the formation of the radical pair  [190], histidine [191][192][193] and Singlet Oxygen Sensor Green [190,194,195]. The 1270 nm luminescence has been used to measure 1 O 2 generation by isolated RC complexes [171,196]. The methods of 1 O 2 measurement were recently reviewed [23]. The absolute rate of 1 O 2 production can be estimated by comparing the signal strength (e.g., 1270 nm luminescence intensity, fluorescence yield, the yield of 2,2,6,6-tetramethylpiperidine-1yl) oxyl or EPR signal amplitude) with the signal obtained from a sensitizer chemical with a known 1 O 2 yield. The main limitation of such an estimation is that, in photosynthetic material, 1 O 2 may be effectively quenched before reacting with the sensor, and therefore all estimates of 1 O 2 yield of photosynthetic material represent a lower limit. At the PPFD (photosynthetic photon flux density) of 2000 µmol m −2 s −1 , histidine-dependent O 2 uptake measurements showed that isolated PSII RCs (6 Chl/RC, [197]) produce 1 O 2 at a rate of 4000 µmol 1 O 2 (mg Chl) −1 h −1 , with a quantum yield 0.16 [191]. The yield of 1 O 2 per 3 P680 is very high, as the quantum yield of 3 P680 formation in the same preparations was 0.3 [191,198]. 2,2,6,6-tetramethylpiperidine measurements at the same PPFD showed that isolated thylakoid membranes produced 3.73 × 10 −7 1 O 2 molecules per Chl molecule s −1 , and the quantum yield of 1 O 2 formation was 2.59 × 10 −4 [185]. Assuming 490 and 173 Chls per PSII and PSI, respectively [197], and a PSII:PSI ratio of 1 [199], the ratio of Chl to RC of PSII is 663 for a plant thylakoid membrane. Thus, isolated RCs and thylakoids produce, at PPFD 2000 µmol m −2 s −1 , 21,600 and 0.89 1 O 2 per RC per h, respectively. The large difference probably reflects differences in both the actual 1 O 2 production rate and in the experimental method. In cyanobacteria illuminated at PPFD 2300 µmol m −2 s −1 in deuterium oxide, a decrease in O 2 concentration in the presence of histidine showed 1 O 2 production of approximately 27 µmol (mg Chl) −1 h −1 [192], suggesting that 1 O 2 production in vivo may actually be of the same order of magnitude as the maximum rate of O 2 evolution. A similar conclusion was drawn from measurements with isolated RCs [191].
Inactivation of the oxygen-evolving complex (OEC) of PSII leads to an increase in the redox potential of the Q A /Q A − pair, so that 3 P680 is no longer formed, and therefore virtually no 1 O 2 can be produced through recombination reactions. However, even in this situation, some 1 O 2 production can be expected because inactivation of the Mn-cluster leads to the oxidation of organic molecules, presumably by P680 + or TyrZ •+ (the redox active tyrosine residue in the D1 protein), and the formation of organic hydroperoxides [200]. Recently, 1 O 2 formation has been detected in Mn-depleted PSII membranes and correlated with R • formation on the donor side of PSII. It was proposed that 1 O 2 is formed via the Russell mechanism, Reactions (9), (104) and (105) [201,202].
Thus, the formation of 1 O 2 associated with PETC can proceed both via the interaction of O 2 with 3 Chl and decomposition of ROOOOR (Reaction (9)). The formation of 1   PSI is not considered as a site of 1 O2 production, although theoretically the formation of 3 Chl can occur through charge recombination between P700 + and its electron acceptors. In isolated PSI membrane fragments, the recombination of [P700 + A0 − ] in the presence of dithionite was found to lead to the formation of the triplet state in PSI RCs with a quantum yield of approximately 30% [203]. In PSI particles, the flash-induced absorption changes at 820 nm are attributed to the formation of 3 P700 via conversion of the cation-anion biradical pair [P700 + A0 − ], with a yield approaching approximately 50% for 10 ns [204]. It was found that an increase in absorption at 820 nm is immediately followed by a multiphasic  PSI is not considered as a site of 1 O 2 production, although theoretically the formation of 3 Chl can occur through charge recombination between P700 + and its electron acceptors. In isolated PSI membrane fragments, the recombination of [P700 + A 0 − ] in the presence of dithionite was found to lead to the formation of the triplet state in PSI RCs with a quantum yield of approximately 30% [203].
In PSI particles, the flash-induced absorption changes at 820 nm are attributed to the formation of 3 P700 via conversion of the cation-anion biradical pair [P700 + A 0 − ], with a yield approaching approximately 50% for 10 ns [204]. It was found that an increase in absorption at 820 nm is immediately followed by a multiphasic decay, including a major fast phase within 5-10 µs and an intermediate phase (about 10-15% of the signal) within 2 ms [204]. Interestingly, O 2 does not affect the decay. It can be speculated that this indicates that O 2 is unable to efficiently quench 3 P700 in isolated PSI complexes and thereby produce 1 O 2 . It seems that, even if 3 P700 is formed in PSI, its quenching by O 2 is minimized. It has also been suggested that charge recombination mainly occurs between P700 + and phylloquinone A 1 − , which minimizes the formation of 3 P700 triplet [205]. 1 O 2 could be detected in PSII membrane fragments and PSII core complexes but not in PSI particles under the same conditions [181]. However, isolated PSI-LHCI supercomplexes of Arabidopsis produced 1 O 2 at a rate of approximately one tenth of that measured in PSII-LHCII supercomplexes, and the rate of 1 O 2 production by PSI-LHCI supercomplexes of the low-carotene szl1 mutant was approximately one fourth of that measured in PSII-LHCII supercomplexes [206]. There is also some evidence that the Fe-S centers of PSI produce 1 O 2 [207].

Oxygen Reduction in PETC
The first evidence that O 2 can accept electrons from PETC was observed by Mehler who found that O 2 was consumed and H 2 O 2 evolved under the illumination of broken chloroplasts [208]. Light-dependent O 2 consumption as an indicator of Mehler's reaction has been reported also in vivo in algae and cyanobacteria [209,210], and in isolated intact chloroplasts with the capacity for CO 2 fixation [211,212] [224,226,228] Removal of HCO 3 − bound to acceptor side of PSII.
(0.2-2) × 10 −1 [222] Reoxidation of Q A − by PQ that binds to an empty Q B site. in a hydrophobic environment is so low that Q A − would be a poor reductant, although its lifetime is long enough for chemical reactions (Table 2). However, the participation of Q A − in O 2 reduction has been suggested [160,[234][235][236]. O 2 •− production was found to increase in the presence of an inhibitor of electron transfer at the Q B site of PSII, DCMU (3-(3,4-di-chlorophenyl)-1,1-dimethyl urea), which was explained by the fact that DCMU increases the lifetime of Q A − [236].
The production of O 2 •− by both PSII with a functional Mn-cluster and by Ca 2+ and Cl − -depleted PSII was detected using 5-diethoxyphosphoryl-5-methyl-1-pyrroline-N-oxide as a spin trap [235]. The generation of O 2 •− at Q A may occur due to the flexibility of the redox potential of Q A /Q A − , that has been reported to range from −80 to −200 mV (Table 2) and to depend on the structural and functional state of PSII. A shift in the redox potential of the Q A /Q A − to a negative direction may cause the enhanced production of O 2 •− in a PsbS (a chloroplast-localized protein required for NPQ) knock-out mutant [237]. Double-reduced Q A has a very negative redox potential (Table 2) and can reduce O 2 . However, Q A 2− can only be formed by chemical treatment or by strong illumination in the absence of O 2 [238].
Thus, the involvement of Q A 2− in O 2 reduction seems unlikely.
Reduced Q B is less likely to reduce O 2 than Q A − because the redox potential of Q B /Q B − is around −45 mV (Table 2). Although Q B − has a much longer lifetime than Q A − when electron donation from Q A − does not occur (Table 2), the quinone in the Q B site is involved in proton-coupled electron transfer, and the redox potential of Q B − becomes positive when the quinone is protonated ( Table 2). The possible generation of O 2 •− by quinones at Q A and Q B pockets is illustrated in Figure 2A.  In untreated chloroplasts, cyt b559 is found in high potential (HP), intermediate potential (IP) and low potential (LP) forms [239]. A very low potential (VLP) form was observed in isolated RCs of PSII [240] and seems to be an isolation artefact. The Em values of the three forms of cyt b559 are (see review [239]): The ratio between the forms is variable and depends on isolation procedure. For example, modification of the donor side of PSII by removal of the Mn-cluster leads to the conversion of the HP form to the LP form [241]. In intact chloroplasts, the ratio of HP to LP forms was found to be 58 to 31, with respective redox potentials of 383 and 77 mV [242]. In untreated PSII membranes, the ratio HP:IP:LP was estimated as 44:31:25, with redox potentials of 375, 228 and 57 mV, respectively [243,244]. In isolated thylakoid membranes, 85% of cyt b559 was in the HP form [245]. The values measured from intact chloroplasts may best reflect the situation in vivo.
Cyt b559 has been suggested to be involved in cyclic electron transfer around PSII, where PQ bound in the specific binding pockets QD and QC acts as both an electron donor and an electron acceptor [239]. This idea is supported by the finding that PQH2 can reduce cyt b559 in both intact chloroplasts [246] and in PSII RC preparates [247]. The photoreduction of cyt b559 was found in isolated thylakoids and was inhibited by DCMU [248]. In Triton X-100-solubilized PSII particles,  In untreated chloroplasts, cyt b559 is found in high potential (HP), intermediate potential (IP) and low potential (LP) forms [239]. A very low potential (VLP) form was observed in isolated RCs of PSII [240] and seems to be an isolation artefact. The E m values of the three forms of cyt b559 are (see review [239]): The ratio between the forms is variable and depends on isolation procedure. For example, modification of the donor side of PSII by removal of the Mn-cluster leads to the conversion of the HP form to the LP form [241]. In intact chloroplasts, the ratio of HP to LP forms was found to be 58 to 31, with respective redox potentials of 383 and 77 mV [242]. In untreated PSII membranes, the ratio HP:IP:LP was estimated as 44:31:25, with redox potentials of 375, 228 and 57 mV, respectively [243,244]. In isolated thylakoid membranes, 85% of cyt b559 was in the HP form [245]. The values measured from intact chloroplasts may best reflect the situation in vivo.
Cyt b559 has been suggested to be involved in cyclic electron transfer around PSII, where PQ bound in the specific binding pockets Q D and Q C acts as both an electron donor and an electron acceptor [239]. This idea is supported by the finding that PQH 2 can reduce cyt b559 in both intact chloroplasts [246] and in PSII RC preparates [247]. The photoreduction of cyt b559 was found in isolated thylakoids and was inhibited by DCMU [248]. In Triton X-100-solubilized PSII particles, which mostly have the LP form of cyt b559, short-chain PQs stimulated both photoreduction and dark oxidation of cyt b559 [248].
The involvement of cyt b559 in electron transfer reactions of PSII indicates that cyt b559 is a redox active component that can potentially reduce O 2 . It has been shown that fast, dark reoxidation of the PQ pool in thylakoid membranes is not caused by direct oxidation of PQH 2 by O 2 , and it was suggested that the LP form of cyt b559 can transfer an electron to O 2 and thereby act as a PQH 2 :O 2 oxidoreductase [248]. In isolated PSII membranes, O 2 has been shown to compete with prenylquinones for oxidation of the LP form of cyt b559, suggesting that LP cyt b559 can form O 2 •− [249]. Exogenously added short-chain quinones significantly enhance O 2 •− production by PSII [245]. This finding was interpreted to indicate that these quinones reduce LP cyt b559, which then undergoes spontaneous autoxidation, resulting in O 2 •− formation.
However, the reduction in O 2 by LP cyt b559 is thermodynamically unfavorable, taking into account that the redox potential of the LP form is usually within 20-110 mV in untreated membranes, although sometimes an LP form with a negative potential is observed [239]. A possible alternative solution is that cyt b559 mediates the formation of semiquinones at Q D and Q C sites. Experimental evidence of the reduction in O 2 by a loosely bound plastosemiquinone anion radical (PQ •− ) at the Q C site was provided by Yadav et al. [250]. The authors showed that PQ •− can be formed by a one electron reduction in PQ at the Q B site and one electron oxidation of PQH 2 by cyt b559 at the Q C site. Because a PQ molecule has been crystallographically detected in the Q C site [251], PQ might be tightly bound within the Q C pocket and act as an electron carrier from cyt b559 to P680. The environment of the Q D pocket is probably flexible and lipophilic and can facilitate a PQ/PQH 2 exchange. In this case, the PQ pool can serve as an electron donor for cyt b559. It might be proposed that the formation of O 2 •− in a cyt b559-dependent pathway couples cyt b559 and quinones and depends on the redox state of the PQ pool ( Figure 2A). The rate constant of cyt b559-mediated reduction of O 2 was estimated to be about 10 −6 M −1 s −1 inside the thylakoid membrane, assuming that the reaction proceeds as a second-order chemical reaction [252]. The formation of O 2 •− in PSII causes the formation of H 2 O 2 via spontaneous dismutation (Reaction (5)) [234] or via a cyt b559-dependent catalytic reaction [253]. In isolated RCs of PSII, cyt b559 was found to exhibit SOD activity [217]. As proposed by Pospíšil [160], the catalytic formation of H 2 O 2 by cyt b559 proceeds as a two-step reduction-oxidation reaction involving two molecules of O 2 •− . The first step is reduction of cyt b559 (Fe 3+ ) to cyt b559 (Fe 2+ ), Reaction (106 during the two-electron oxidation of water by the Mn-cluster at the transition from the S 2 state to the S 0 state [234,254]. However, the two-electron oxidation of water during the transition of the Mn-cluster from S 3 to S 1 does not result in the formation of H 2 O 2 [255]. In PSII, the Fenton mechanism involving a metal (M) cation, Reaction (58), can also function, leading to the formation of HO • .
In PSII, HO • can be formed both in the dark and in the light. HO • formation was shown when PSII membrane particles were heated in the dark [256]. The authors suggested that this process is associated with heat-induced changes of the PSII donor side and proceeds via the Fenton mechanism. The formation of HO • was suppressed by CAT and metal chelators, indicating that the appearance of HO • is related to the decomposition of H 2 O 2 . However, a high concentration of CAT, around 5000 U/mL, was required to suppress the appearance of HO • . Exogenous calcium and chloride prevented the appearance of HO • . Furthermore, no HO • -related EPR signal was observed after removal of the Mn-cluster by Tris-treatment of PSII membranes [257]. These data confirm that the Mn-cluster is likely involved in HO • formation in PSII under heat stress in the dark.
The light-dependent formation of HO • occurs in untreated PSII membranes [235,[258][259][260]. Experimental results suggest that HO • can be produced in the light by two pathways: firstly, by the well-known Fenton mechanism and secondly, by the reduction of a peroxide bound to the non-heme iron on the acceptor side of PSII [72]. The formation of HO • at the non-heme iron is initiated by the binding of O 2 •− and formation of an O 2 •− -iron complex that can be protonated to a ferric-hydroperoxo complex, Reactions (49) and (50). The ferric-hydroperoxo complex can be decomposed via reduction by Q A − with the formation of HO • and a ferric-oxo ((Fe 3+ )-O − ) complex, Reaction (108).
where L is a ligand. Reaction (108) can be considered as a Fenton reaction proceeding with a bound hydroperoxide. The possible sites of formation in PSII are shown in Figure 2B. The formation of bound hydroperoxides has been found to occur on the donor side of the PSII ( Figure 2C). The mechanism is associated with the formation of a long-lived species, having a high positive redox potential in PSII. In PSII membranes holding an intact Mn-cluster, the O 2 consumption rate is very low, around 1 µmol O 2 (mg of Chl) −1 h −1 [128], but the rate becomes 6-fold higher in alkaline-treated and Mn-depleted PSII membranes [261]. O 2 consumption was found to be associated, at least partially, with the generation of a component with positive charge(s) on the donor side of PSII, as the electron donors diphenylcarbazide and ferrocyanide suppressed the rate of O 2 consumption caused by disruption of the donor side of PSII. A further study revealed that the removal of Mn from the OEC of PSII leads to O 2 photoconsumption with a maximum at the first flash, with a yield comparable to the yield of O 2 evolution on the third flash measured in the PSII samples before Mn removal [262]. Inactivation of the OEC can lead to the formation of both P680 •+ and TyrZ • . In the absence of electron donation from the OEC, both will have a long lifetime, and will therefore be able to interact with surrounding molecules such as Chls, carotenoids and amino acids. Based on these results, it has been proposed that the formation of peroxides on the donor side of PSII proceeds via a radical chain mechanism, starting with P680 •+ (or TyrZ • ), Reactions (104), (105) and (109).
The evidence of ROOH production on the donor side of PSII was obtained using the specific fluorescence probe SPY-HP [200]. In this work, highly lipophilic peroxides (LOOH) and relatively hydrophilic ones (ROOH), were distinguished by the rate of reaction with Spy-HP. The formation rates of both LOOH and ROOH were estimated to be 0.022 µmol LOOH (µmol RC) −1 s −1 and 1.11 µmol ROOH (µmol RC) −1 s −1 , respectively [200]. The formation of carbon centred radicals, in turn, was found in PSII membranes with EPR spin-trapping technique when PSII membranes were treated by high light and heating. It has recently been found that exposure of Mn-depleted PSII membranes to high light results in the formation of protein radicals located mainly in the D1, D2, CP43 and CP47 proteins [202]. The formation of protein radicals is suppressed by diphenylcarbazide, indicating that protein radicals were formed by the oxidation of proteins by P680 •+ or TyrZ • . The formation of protein radicals was correlated with the formation of hydroperoxides measured with the SPY-HP probe. The formation of R • can initiate chain propagation reactions, and thereby lead to accumulation of ROOH (Reactions (105) and (109)).

Formation of Reduced Forms of Oxygen
The acceptor side of PSI is believed to be the predominant site of O 2 reduction in thylakoid membranes, as O 2 reduction depends on the PSI activity (see reviews [12,[213][214][215]). It has been shown that both the photoreduction of cytochrome c and photooxidation of epinephrine, which have been used as traps for O 2 •− , were inhibited by SOD. This indicates that the reduction in O 2 proceeds via univalent reduction, and O 2 •− was identified as the primary product in illuminated thylakoids [124,263]. The predominant role of PSI in O 2 reduction was shown in experiments with specific inhibitors that block PETC at different sites, and using a PSI-deficient mutant. The photoproduction of O 2 •− in thylakoids is inhibited by DCMU and can be restored by the addition of AscH 2 and N,N,N ,N -tetramethyl-p-phenylenediamine, to provide electron donation to PC and P700, respectively [124,263,264]. This indicates that the contribution of PSII to the photoproduction of O 2 in thylakoids is small . A slight influence of O 2 on the steady-state level of Chl fluorescence in a PSI-deficient mutant of Oenothera sp. was attributed to insignificant leakage of electrons from PETC to O 2 , due to the suppression of Mehler's reaction [265]. On the other hand, a significant rate of O 2 reduction by thylakoids was observed in the presence of dibromothymoquinone (DBMIB) and dinitrophenylether of 2-iodo-4-nitrothymol (DNP-INT), inhibitors of PQH 2 oxidation by Cyt b6f, [128,266]. It was found that the contribution of other sites of PETC, besides PSI, to O 2 reduction increased with an increase in light intensity, and at high intensities achieved 60% of total O 2 reduction in PETC. These data suggest that PSI is not the only site of O 2 reduction in thylakoid membranes, but other sites of PETC can contribute to O 2 reduction [128]. Thus, experiments with isolated PSI membranes can provide more correct measurements of activity of PSI in the photoproduction of O 2 •− .
The electron transport chain within PSI (Figure 3) consists of two quasisymmetrical branches (A and B) containing six Chl, two phylloquinones (A 1 ), and three 4Fe-4S clusters (F X , F A , and F B ). Two Chl a molecules have been assigned to the spectroscopically characterized primary acceptor A 0 . Another pair of Chl a molecules is located between P700 and A 0 and assigned as accessory Chls that may participate in excitation and/or electron transfer (for more details, see review [267]). The acceptor side of PSI is believed to be the predominant site of O2 reduction in thylakoid membranes, as O2 reduction depends on the PSI activity (see reviews [12,[213][214][215]). It has been shown that both the photoreduction of cytochrome c and photooxidation of epinephrine, which have been used as traps for O2 •− , were inhibited by SOD. This indicates that the reduction in O2 proceeds via univalent reduction, and O2 •− was identified as the primary product in illuminated thylakoids [124,263]. The predominant role of PSI in O2 reduction was shown in experiments with specific inhibitors that block PETC at different sites, and using a PSI-deficient mutant. The photoproduction of O2 •− in thylakoids is inhibited by DCMU and can be restored by the addition of AscH2 and N,N,N',N'tetramethyl-p-phenylenediamine, to provide electron donation to PC and P700, respectively [124,263,264]. This indicates that the contribution of PSII to the photoproduction of O2 in thylakoids is small. A slight influence of O2 on the steady-state level of Chl fluorescence in a PSI-deficient mutant of Oenothera sp. was attributed to insignificant leakage of electrons from PETC to O2, due to the suppression of Mehler's reaction [265]. On the other hand, a significant rate of O2 reduction by thylakoids was observed in the presence of dibromothymoquinone (DBMIB) and dinitrophenylether of 2-iodo-4-nitrothymol (DNP-INT), inhibitors of PQH2 oxidation by Cyt b6f, [128,266]. It was found that the contribution of other sites of PETC, besides PSI, to O2 reduction increased with an increase in light intensity, and at high intensities achieved 60% of total O2 reduction in PETC. These data suggest that PSI is not the only site of O2 reduction in thylakoid membranes, but other sites of PETC can contribute to O2 reduction [128]. Thus, experiments with isolated PSI membranes can provide more correct measurements of activity of PSI in the photoproduction of O2 •− .
The electron transport chain within PSI ( Figure 3) consists of two quasisymmetrical branches (A and B) containing six Chl, two phylloquinones (A1), and three 4Fe-4S clusters (FX, FA, and FB). Two Chl a molecules have been assigned to the spectroscopically characterized primary acceptor A0. Another pair of Chl a molecules is located between P700 and A0 and assigned as accessory Chls that may participate in excitation and/or electron transfer (for more details, see review [267]).  [145,[268][269][270][271]; (C) possible means of ROS formation in PSI [152,266,[272][273][274][275]. PC is plastocyanin; P700 is a dimer of Chl a molecules, the primary electron donor; A0A and A0B are Chl a molecules located in branches A and B, respectively, both act as primary electron acceptors; A1A and A1B are phylloquinone molecules located in branch A and B, respectively, both acting as electron acceptors; FX, a 4Fe-4S cluster, a secondary electron acceptor; FA and FB, 4Fe-4S clusters, terminal electron acceptors.
The mechanism of O2 reduction in PSI is still under debate. It was suggested that O2 •− production within the thylakoid membranes most likely occurs via autooxidation of the membrane-bound primary electron acceptors in PSI, possibly 4Fe-4S clusters (FX, FA, and FB) [152]. The Em of FA/FA − and FB/FB − and  [145,[268][269][270][271]; (C) possible means of ROS formation in PSI [152,266,[272][273][274][275]. PC is plastocyanin; P700 is a dimer of Chl a molecules, the primary electron donor; A 0A and A 0B are Chl a molecules located in branches A and B, respectively, both act as primary electron acceptors; A 1A and A 1B are phylloquinone molecules located in branch A and B, respectively, both acting as electron acceptors; F X , a 4Fe-4S cluster, a secondary electron acceptor; F A and F B , 4Fe-4S clusters, terminal electron acceptors.

The mechanism of O 2 reduction in PSI is still under debate. It was suggested that O 2
•− production within the thylakoid membranes most likely occurs via autooxidation of the membrane-bound primary electron acceptors in PSI, possibly 4Fe-4S clusters (F X , F A , and F B ) [152]. The E m of F A /F A − and F B /F B − and F X /F X − vs. NHE were estimated to be −479, −539, and −650 mV, respectively ( Figure 3) [145]. The reduction in O 2 by F X is thermodynamically favorable but kinetically less likely than a reduction in O 2 by F A − or F B − , as the lifetime of F X − is less than 50 ns (Figure 3). When F A and F B clusters are reduced, the lifetime of F X is limited by charge recombination [P700 + F X − ] and estimated to be~250 µs [276].
Electron transfer from F B − to Fd occurs within 1 µs, and therefore the oxidation of F A − and F B − by O 2 in an aqueous region is not kinetically favorable in the presence of oxidized Fd (Figure 3). to 600 for isolated thylakoid membranes [124,264]. The rate of O 2 •− production is at least one order of magnitude higher in PSI subchloroplast fragments in the presence of the surfactant Triton X-100 than in its absence [278]. The K m value for O 2 in photoreduction by PSI was estimated to be 2-3 µM in both thylakoid membranes and PSI subchloroplast fragments and the second order rate constant for O 2 reduction by the electron acceptors of PSI was calculated to be 1.5 × 10 7 M −1 s −1 [278]. In another work, the K m value for O 2 was estimated to equal to~8 and~3 µM for thylakoids, in the absence and in the presence of Triton X-100, respectively [279]. Experiments with O 2 •− -dependent protein iodination showed that O 2 •− can also be produced in the aprotic interior of the thylakoid membrane close to the RC of PSI [272]. Thus, not only F A and F B , but also F X and A 1 , might be involved in O 2 •− production within the thylakoid membrane. It has been suggested that O 2 •− mediates cyclic electron transfer by donating electrons to Cyt b6f or to P700 + , and this cycle would explain why the observed rate of O 2 •− production is low in intact PSI [12]. It has recently been suggested that the reduction in O 2 •− by F A , F B and F X occurs in a lipophilic region [273]. According to the E m , Fx and A 1 would be favorable reductants of O 2 , even in an aprotic environment, as the E m values of the pairs A 1 /A 1 − located on the A-and B-branches of PSI electron transfer chain are −0.7 and −0.81 mV, respectively ( Figure 3). Indeed, phylloquinone A 1 stimulated the flash-induced photoconsumption of O 2 when added to thylakoid membranes from which A 1 had been partially removed [266]. It was suggested that A 1 could be the main reductant in O 2 •− production in PSI. However, results regarding the importance of the phylloquinone in O 2 •− production vary. Firstly, the stimulation of O 2 photoconsumption by addition of A 1 was observed only on the first flash [266]. The appearance of O 2 •− on the outside and inside of thylakoid membranes was tested with hydrophilic and lipophilic cyclic hydroxylamines that react with O 2 •− , forming nitroxide radicals with a specific EPR spectrum [274]. In this work, a significant effect of SOD on the formation of both hydrophilic and lipophilic nitroxide radicals suggested that 90% of O 2 •− is formed at the membrane surface or outside of the membrane. On the other hand, evidence of the participation of A 1 in O 2 •− formation was obtained with PSI complexes isolated from menB mutant, a phylloquinone-less knockout strain of the gene encoding 1,4-hydroxynaphthoyl-CoA-synthase of the cyanobacterium Synechocystis sp. PCC 6803. The mutant contains PQ at the phylloquinone-binding site A 1 [275]. In the mutant, the redox potential of PQ bound to the A 1 site was −553-−693 mV, close to the redox potential of F X /F X − and about 100 mV more positive than that of A 1 /A 1 − [281]. O 2 photoconsumption in isolated PSI complexes of the mutant was found to be slower than in the wild type [275]. The low rate of O 2 photoconsumption in the mutant was explained by the difference in the redox potentials of PQ and A 1 , and the results suggest that A 1 is the main site of O 2 reduction in PSI. N,N,N ,N -Tetramethyl-p-phenylenediamine and AscH 2 were used as electron donors. MDA is formed mainly by a reaction between APX and AscH 2 (Reactions (62)- (64)). MDA can also be formed by a reaction between AscH 2 and O 2 •− , and MDA is an effective electron acceptor of PSI, effectively competing with methyl viologen [282]. The reduction in MDA by PSI occurs via reduced Fd [283]. In summary, a number of AscH 2 -related reactions can influence the photoreduction of O 2 by PSI (Reactions (34), (62)- (64) and (110)- (113)), and the large number of reactions and reactants makes it difficult to estimate the importance of AscH 2 /MDA/DHA in O 2 reduction in PSI.
MDA + MDA + 2H + → AscH 2 + DHA (113) From data presented by Kozuleva et al. [275], the rate of O 2 •− production by PSI can be estimated to be 2.  [121]. However, the rate of O 2 reduction by PSI was saturated to above 20 µM of O 2 , with the second-order rate constant 1.5 × 10 7 M −1 s −1 at a high light intensity [278]. The reaction of O 2 with semiquinones having low redox potential proceeds with rate constants in the range of 10 8 -10 9 M −1 s −1 [61]. These data may suggest that cooperation between 4Fe-4S clusters and phylloquinones A 1 can provide flexibility for the O 2 •− formation inside and outside of the thylakoid membrane. In high light, O 2 •− formation by A 1 becomes more important, which leads to the accumulation of O 2 •− within the membrane.
It was shown that the formation of HO • in broken chloroplasts was suppressed by DCMU and it was suggested that HO • is predominantly produced in PSI via the reduction in H 2 O 2 by protein-bound iron in PSI, as the metal chelator Desferal did not suppress HO • production [157].  [286], and therefore the presence of H 2 O 2 in the vicinity of the A 1 site, with a much more negative redox potential (Figure 3), would lead to the formation of HO • in a Fenton-type reaction of H 2 O 2 with phylloquinone A 1 •− (Reaction (115)) [252]. PQ is a prenyllipid consisting of 2,3-dimethyl-1,4-benzoquinone and a side chain of nine isoprenyl units attached to Position 5. The total amount of PQ in leaves is in the range 25-40 molecules per P700 [248,[287][288][289]. PQ has been found in thylakoid membranes, the envelope of the chloroplast and plastoglobules. [290][291][292][293][294]. The ratio of PQ in the envelope and PQ in the thylakoid membrane was found to be 2:5 [293]. The PQ involved in electron transfer in the thylakoid membrane is called the photoactive PQ and its amount is in the range 6-15 PQ per P700, assuming that the ratio of P700 and Chl is 1/600 [248,288,[295][296][297][298]. PQ can be present in three forms: PQ, PQ •− , and PQH 2 . Both reduced forms can exist in protonated and deprotonated forms: PQH • or PQ •− , and PQH 2 , PQH − or PQ 2− . The pK 1 and pK 2 values of PQH 2 in aqueous solutions are 10.8 and 12.9; the pKa value of PQH • is 5.9 [299]. The above data were measured for plastoquinone-1, which has only one prenyl group attached to Position 5 of 2,3-dimethyl-1,4-benzoquinone.
Significant PSI-independent O 2 reduction was observed in a thylakoid suspension in the presence of the DNP-INT, that prevents the oxidation of PQH 2 by Cyt b6f [128,236,266]. In the work of Kruk et al. [266], significant O 2 reduction was also demonstrated in the presence of DBMIB, another inhibitor of oxidation of PQH 2 by the Cyt b6f. DBMIB was found to strongly inhibit O 2 •− production, whereas the formation of H 2 O 2 was only partially inhibited. Furthermore, the rate of H 2 O 2 production increased with the concentration of DBMIB [300]. On other hand, the removal, by a repeated freeze-thaw procedure, of PC, suppressed O 2 reduction by thylakoid membranes. In addition, the PC-inhibitor HgCl 2 significantly suppressed O 2 reduction [266]. These data may suggest that the suppression of PSI-independent O 2 reduction requires a strong inhibition procedure that may cause unspecific damage to the photosynthetic apparatus. In the work of Cleland and Grace [236], the production of O 2 •− in the presence of DNP-INT was attributed to O 2 reduction by Q A − . However, Khorobrykh and Ivanov [128] showed that PSI-independent O 2 consumption in thylakoids was suppressed by DCMU, and O 2 consumption by isolated PSII membranes was low.  (Figure 4) [128]. Thylakoid membranes have also been shown to accumulate H2O2 in the presence of cytochrome c that reacts with O2 •− and prevents the formation of H2O2 via superoxide dismutation (Reaction (31)) [284]. These data suggest that a considerable amount of H2O2 is generated inside the thylakoid membrane in the reaction of O2 •− with PQH2 ( [284], Figure 4), as earlier suggested by Khorobrykh and Ivanov [128]. These results contradict with the results of Asada et al. [124], where cytochrome c completely inhibited H2O2 formation by thylakoids. However, in a later work of Takahashi and Asada [272], the formation of H2O2 in the presence of cytochrome c was shown. It is possible that different light intensities caused the contradiction, as H2O2 formation appears to increase with light intensity [284].
The mechanism and efficiency of O2 reduction in the PQ pool are under debate. Autooxidation of PQH2 is one possible mechanism (Figure 4) but is it biologically significant? According to the redox potential, the reduction in O2 by both PQ •− and PQ 2− is thermodynamically favorable in aqueous solution since the redox potentials of PQ/PQ •− and PQ •− /PQ 2− are −165 and −274 mV, respectively [299]. The deprotonation of PQH2 or PQH • is essential for O2 reduction. Since PQ 2− is mostly protonated under physiological pH, PQ •− was considered the main form of reduced PQ that could be involved in O2 reduction in thylakoids. The reactions of semiquinones with O2 with formation of O2 •− are equilibrium reactions where the quinone can be reduced by O2 •− (Reaction (31)).
The equilibrium constant for the reaction of O2 with PQ •− , as determined by the equation (RT/F)lnK = E(O2/O2 •− ) − E(Q/Q •− ), where R is the gas constant, T is temperature and K is the equilibrium constant, and F is the Faraday constant, is 1.56 if the redox potentials of PQ/PQ •− and O2/O2 •− are −165 and −160 mV, respectively [61]. The forward and reverse second-order rate constants for the formation of O2 •− by PQ •− (Reaction (31)) are kforward ~ 10 8 M −1 s −1 and kreverse ~ 7 × 10 7 M −1 s −1 [61]. The equilibrium constant for Reaction (116) was estimated to be 10 −9.2 [301], which shows that the formation of PQH • via Reaction (116) is negligible. Thus, the apparent rate of O2 •− production in the PQ pool is determined by the rate of PQ •− appearance, rate of O2 •− production via reaction of O2 with PQ •− (Reaction (31)) and the rate of O2 •− removal from the equilibrium Reaction (31). In solvents with pure of PQH2 and PQ, the  (31)) [284]. These data suggest that a considerable amount of H 2 O 2 is generated inside the thylakoid membrane in the reaction of O 2 •− with PQH 2 ([284], Figure 4), as earlier suggested by Khorobrykh and Ivanov [128]. These results contradict with the results of Asada et al. [124], where cytochrome c completely inhibited H 2 O 2 formation by thylakoids. However, in a later work of Takahashi and Asada [272], the formation of H 2 O 2 in the presence of cytochrome c was shown. It is possible that different light intensities caused the contradiction, as H 2 O 2 formation appears to increase with light intensity [284]. The mechanism and efficiency of O 2 reduction in the PQ pool are under debate. Autooxidation of PQH 2 is one possible mechanism (Figure 4) but is it biologically significant? According to the redox potential, the reduction in O 2 by both PQ •− and PQ 2− is thermodynamically favorable in aqueous solution since the redox potentials of PQ/PQ •− and PQ •− /PQ 2− are −165 and −274 mV, respectively [299]. The deprotonation of PQH 2 or PQH • is essential for O 2 reduction. Since PQ 2− is mostly protonated under physiological pH, PQ •− was considered the main form of reduced PQ that could be involved in O 2 [61]. The equilibrium constant for Reaction (116) was estimated to be 10 −9.2 [301], which shows that the formation of PQH • via Reaction (116) is negligible. Thus, the apparent rate of O 2 •− production in the PQ pool is determined by the rate of PQ •− appearance, rate of O 2 •− production via reaction of O 2 with PQ •− (Reaction (31)) and the rate of O 2 •− removal from the equilibrium Reaction (31). In solvents with pure of PQH 2 and PQ, the apparent rate of O 2 •− production obviously results from the following reactions: In thylakoid membranes, PQ is reduced by PSII to PQH 2 in the light. This lowers the concentration of PQ, thereby preventing reaction (31)  of PQH 2 in different solvents, estimated from the initial rates, were found to range from 10 −2 to 10 −3 M −1 s −1 for both aqueous and aprotic solvents [248,252]. However, fast PQH 2 oxidation in organic solvent was observed after the addition of KOH [252]. This reaction likely results from the formation of PQ 2− with a very negative redox potential, −1.1 V for PQ/PQ 2− [301]. The rate constant of PQH 2 oxidation associated with O 2 reduction by the PQ pool in thylakoids was estimated to be~10 3 M −1 s −1 if the reaction occurs inside the thylakoid membrane [128], and a later work calculated this rate constant to be 1.21 × 10 −3 M s −1 while the rate of PQH 2 autoxidation was~10 −8 M s −1 [252]. These rates were calculated assuming that PQH 2 oxidation by O 2 is a second-order chemical reaction and the oxidation of PQH 2 occurs in the volume of thylakoid membrane. The steady-state concentration of PQ •− inside the thylakoid membrane produced via reaction (116) can be estimated to be about 10 −8 M. The following values were used in the calculations: amount of photoactive PQ, 14 × 10 −3 mol PQ (mol Chl) −1 [248]; volume of thylakoid membrane, 4.6 × 10 −6 L (mg Chl) −1 [304]; molar mass of Chl, 894 g mol −1 [304]; the equilibrium constant for reaction (116) was taken as 10 −10 ; and the ratio of PQ and PQH 2 was taken as 1/9. If O 2 •− is very rapidly removed and therefore Reaction (31)  after photoreduction [248], suggesting that light-dependent reaction(s) dominate in the O 2 -dependent oxidation of the PQ pool.
The PQ pool has also been found to scavenge 1 O 2 in thylakoid membranes [305][306][307], with the second-order rate constant of the reaction of 1 O 2 with PQH 2 , 0.97 × 10 8 M −1 s −1 in acetonitrile [50]. The reaction of 1 O 2 with PQH 2 in methanol was found to lead to the formation of H 2 O 2 (Reaction (119)) [48]. It was suggested that the reaction of 1 O 2 with PQH 2 is initiated by the formation of 1 O 2 in PSII and can also proceed inside the thylakoid membrane [48]. The formation of H 2 O 2 via oxidation of PQH 2 by 1 O 2 may occur in two ways. In the first one, 1 O 2 reacts with PQH 2 to form an unstable hydroperoxide adduct of the quinone ring (PQH 2 -OO), which directly decomposes to form H 2 O 2 and PQ (Reaction (119)). In the second way, the hydroperoxide adduct decomposes to form HO 2 • and PQH • (Reaction (120)). This indirect mechanism would be similar to that proposed for the oxidation of AscH 2 by 1 O 2 [47]. In the indirect mechanism, H 2 O 2 is produced by the oxidation of PQH 2 to PQH • by HO 2 • (Reaction (121)).
Thus, the PSI-independent O 2 reduction in the PQ pool may depend on 1 O 2 production in PSII, and this reaction can cause the formation of H 2 O 2 inside the thylakoid membrane. O 2 reduction, associated with the PQ pool, also occurs without any inhibitors [264]. The ratio of the rate of O 2 reduction in PSI and the rate of O 2 reduction in the PQ pool, in the absence of any inhibitors, reaches 1:1 at a high light intensity [264]. However, the rate of O 2 reduction by the PQ pool in the presence of DNP-INT is saturated at a low light intensity [128,264,308]. These data confirm that O 2 reduction in a PQ pool in thylakoids without any inhibitors can occur parallel to O 2 reduction in PSI. Interestingly, efficient O 2 reduction by the PQ pool is observed at pH 5.0 in the absence of inhibitors, but not in the presence of DNP-INT [128,264]. The simplest explanation for differences in O 2 reduction by the PQ pool in the absence and presence of DNP-INT, is to assume that the formation of O 2 •− occurs in PSI. O 2 •− can react with PQH 2 to form H 2 O 2 via Reaction (114) [264].
The autoxidation of PQH 2 does not imply the participation of any enzymes. However, the oxidation of the PQ pool with PTOX is widely discussed [309]. PTOX is a non-heme diiron quinol oxidase that oxidizes PQH 2 and reduces O 2 to H 2 O. PTOX is localized in the non-appressed regions of the thylakoid membrane [310]. It has been suggested that PTOX provides an alternative electron flow from the PQ pool to O 2 to prevent photoinhibition of PSII [311]. However, in higher plants, PTOX-mediated electron flow to O 2 is negligible [312] or its contribution is less than one percent of the total electron flow through PETC [313,314]. However, PTOX may depend on conditions, as high PTOX content and high PTOX activity were induced in the alpine species Ranunculus glacialis L. during growth in strong light [315]. The rate of PTOX-mediated electron flow is approximately 0.3 e − s −1 (P680) −1 [313]. This makes the rate of PQH 2 oxidation equal to 0.15 PQH 2 (P680) −1 s −1 , or 0.35 µmol PQH 2 (mg Chl) −1 h −1 , assuming that the ratio of PSII to Chl is 1:420. Thus, the rate of PQH 2 oxidation can be estimated to be 8.68 × 10 −5 M s −1 inside the thylakoid membrane [252]. The second-order rate constant of PTOX-mediated oxidation of PQH 2 inside the thylakoid membrane is 10.6 M −1 s −1 . In the light, the oxidation of PQH 2 by PTOX, associated with the reduction of O 2 to H 2 O, would not lead to consumption of O 2 because of its matching stoichiometry with O 2 production by PSII, Reactions (122) and (123).
PTOX-mediated electron flow to O 2 is assumed to produce no ROS. However, isolated PTOX can oxidize decylPQH 2 with the formation of O 2 •− or H 2 O 2 at pH 8.0 or in substrate-limiting concentrations [316]. The efficiency of ROS production by PTOX was estimated to be around 17% of the total O 2 -reduction activity of PTOX [316]. The rate of PTOX-mediated PQH 2 oxidation associated with the formation of H 2 O 2 was estimated to be 1.47 × 10 −5 M s −1 , with the second-order rate constant of 1.8 M −1 s −1 inside the thylakoid membrane [252]. Thus, the estimated rate of PTOX-mediated O 2 reduction is 100 times less than the rate of O 2 reduction by the PQ pool in illuminated thylakoids. Furthermore, if O 2 •− is formed by PTOX, PQ •− might also be formed.
PQ •− is also considered a source of O 2 •− production by Cyt b6f via the reaction of O 2 with PQ •− .
It has been suggested that PQ •− , generated via one-electron oxidation of PQH 2 at the Q O site by the 2Fe-2S cluster of the high-potential, Rieske iron-sulfur protein of the Cyt b6f (Reaction (124)), can be oxidized by the conversion of O 2 to O 2 •− [317].
Isolated Cyt b6f complexes have been shown to produce H 2 O 2 when decylPQH 2 and PC were used as an electron donor and electron acceptor, respectively [317]. It was suggested that •− formation is observed in the presence of DBMIB, which has been shown to bind to an iron-sulfur binding site and at a position distal to the iron-sulfur binding site in Cyt b6f. This indicates that the mechanism of H 2 O 2 production is related to the oxidation of PQH 2 at the Q O site of Cyt b6f. The production of O 2 •− in Cyt b6f was also shown with EPR spectroscopy [317].  Figure 5. The formation of HO • has never been detected in the PQ pool, although it is supposed to happen. In the chloroplast stroma, H 2 O 2 is efficiently scavenged, which would limit HO • formation. However, H 2 O 2 formed inside membranes by the PQ pool is not efficiently scavenged, and may therefore react with PQ •− to form HO • via the Fenton mechanism (Reaction (125)). 125) of the thylakoid membrane is 4.6 × 10 −6 L (mg Chl) −1 [304]. This rate is close to the rate of O2 •− production by the PQ pool in the presence of DNP-INT. As DNP-INT blocks the oxidation of PQH2 at the QO site, the oxidation of PQH2 in the QO site cannot be responsible for O2 reduction in the PQ pool in the presence of DNP-INT. The formation of PQ •− at the Qi site of Cyt b6f appears to cause O2 reduction in the PQ pool in the presence of DNP-INT (S. Khorobrykh and E. Tyystjärvi, unpublished data). The possible means of the reduction in O2 in Cyt b6f are shown in Figure 5. The formation of HO • has never been detected in the PQ pool, although it is supposed to happen. In the chloroplast stroma, H2O2 is efficiently scavenged, which would limit HO • formation. However, H2O2 formed inside membranes by the PQ pool is not efficiently scavenged, and may therefore react with PQ •− to form HO • via the Fenton mechanism (Reaction (125)). PQ •− + H2O2 → PQ + HO • + OH − (125)

Damage to PSII
PSII is the main producer of 1 O 2 in the chloroplast and a minor producer of other ROS (see Section 3.2), and therefore it is of great interest whether PSII is damaged by 1 O 2 . In isolated PSII core complexes, electron transfer activity is lost and pigments are bleached only in the presence of O 2, suggesting an effect of ROS [318]. Furthermore, PSII is sensitive to damage caused by externally applied 1 O 2 , as shown by a decrease in the quantum yield of PSII in lincomycin-treated tobacco leaves illuminated with the 1 O 2 sensitizer Rose Bengal [189].
The above results indicate that PSII can be damaged by 1 O 2 but do not prove that 1 O 2 produced by PSII is the agent of damage in the photoinhibition of PSII (for reviews, see [319,320]). The photoinhibition of thylakoid membranes does not depend on O 2 and has a similar action spectrum under aerobic and anaerobic conditions [321], and photoinhibition in lincomycin-treated spinach leaf disks is only slightly slower in CO 2 doped N 2 than in air [322]. The effects of both deuterium oxide and ROS scavengers (reviewed by [319]) are variable and may depend on the type of complex. Similarly, effects of intrinsic 1 O 2 quenchers and scavengers vary, as overproduction of the xanthophyll zeaxanthin protects against photoinhibition in vivo in the green alga Chlamydomonas reinhardtii [323] and the carotenoid-rich mutant ∆SigCDE of the cyanobacterium Synechocystis sp. PCC 6803 show protection against the damaging reaction of photoinhibition [193], whereas the same reaction is not more rapid in α-tocopherol-deficient mutants of Arabidopsis [324] and Synechocystis [325]. Further indirect evidence on the participation of 1 O 2 in photoinhibition of PSII also varies, as the modification of the recombination reactions of PSII toward non-1 O 2 -producing direction provides protection against photoinhibition [326], whereas the protection offered by NPQ is very limited [327], suggesting that the photoinhibition of PSII may not depend only on the excitation of Chl [320,328]. The photoinhibition-tolerant green alga Chlorella ohadii exhibits a recombination reaction model that is expected to lead to low 1 O 2 production [329].

Damage to PSI
PSI has long been known to become inhibited, at least in certain plants, at chilling temperatures [336], in a reaction that depends on electron transfer from PSII to PSI [337]. The damage targets the iron-sulfur centers of PSI, and the remaining inactive PSI still functions as an excitation energy quencher [338].  [339]. However, neither the identity of the inhibitory ROS nor the exact site and the mechanism of production are known. Damage to PSI can be specifically induced by the application of fluctuating light, either in the form of short (10-300 ms) strong flashes [340], or in the form of few-seconds-long, saturating but not very strong flashes fired on top of short-term exposure of the plant to weak, PSII-specific light [341].

Oxidation of Membrane Lipids by ROS
Unsaturated fatty acids of membrane lipids can become peroxidated in a reaction with 1 O 2 (Reaction (20)) or HO • (reaction (77)). Peroxidation by 1 O 2 dominates the non-enzymatic formation of lipid peroxides in leaves, whereas radical-induced peroxidation is more common in non-photosynthetic tissues [4].
Fatty acid peroxides, in turn, decompose either spontaneously or enzymatically to oxylipin carbonyls [342]. Tri-unsaturated fatty acids especially fragment to malondialdehyde that is highly reactive in its protonated dialdehyde form (O=CH-CH2-CH=O) [343]. Both malondealdehyde and acrolein, another highly reactive fragmentation product, are produced under non-stressed conditions but their concentrations increase during stress [344,345]. Lipid-peroxide-derived aldehydes and ketones like malondialdehyde function both as agents of damage and signaling molecules in Arabidopsis [344,346]. Due to their reactivity towards ROS, tri-unsaturated fatty acids may function as ROS sinks [347], and signaling by products of lipid oxidation may be essential for plant cells' ability to survive oxidative stress [346].

Damage to Stromal Proteins
The production of ROS in the chloroplast is expected to damage proteins of the compartment of origin. In thylakoid membranes, light-induced damage primarily targets the photosystems. In the stroma, several proteins are known to be targets of ROS damage. The inhibitory effects are often ascribed to the oxidation of cysteine residues.
ROS have a strong inhibitory effect on translation in cyanobacteria [348]. The mechanism of the inhibition by H 2 O 2 is the oxidation of cysteine residues and the subsequent formation of an intramolecular disulfide bond in translation elongation factor G [348], and the formation of a sulfenic acid and an intermolecular disulfide bond in elongation factor Tu [349,350]. The inhibition of translational elongation exerts its effect on the activity of PSII by inhibiting or slowing down the turnover of the D1 protein [351]. Similar ROS effects are expected in chloroplasts.
The Calvin-Benson cycle is inhibited by H 2 O 2 [352] with the ribulose-1,5-bisphosphate carboxylase oxygenase (rubisco) as the most important target of oxidation [353]. Analysis of the proteome of H 2 O 2 -treated chloroplasts revealed modified cysteine residues in both subunits of rubisco, Fd-dependent glutamate synthase, ferredoxin-NADP + oxidoreductase 1 (FNR1) and glyceraldehyde 3-phosphate dehydrogenase subunit B, and a similar analysis after methyl viologen treatment revealed oxidative changes in 24 chloroplast proteins and modified cysteines in rubisco large subunit, FNR1, myrosinase and NAD(P)-binding Rossman-fold-containing protein [353]. The authors suggested that, due to its large amount, rubisco functions as a redox buffer in the chloroplast.

Damage to Chloroplast DNA
ROS are known to react with DNA [2], and chloroplast DNA is not an exception. A comparison of the integrity of DNA of the chloroplasts of mesophyll and bundle sheath cells of maize, a C4 plant, offers insight into ROS damage within the chloroplasts [354]. In C4 plants, mesophyll cells carry out the photosynthetic electron transfer reactions that produce NADPH and ATP, but also ROS, whereas the bundle sheath chloroplasts are almost devoid of PSII that produces O 2 . A drastically larger amount of DNA damage, analyzed with a long-sequence-specific variant of polymerase chain reaction, was found in the chloroplast DNA of light-grown maize plants in mesophyll cells than in bundle sheath cells [354]. Interestingly, mitochondrial DNA showed a similar difference between mesophyll and bundle sheath mitochondria. Doping soil with Cr(VI) that causes ROS production in leaves also caused damage, visualized by staining with 4 ,6-diamidino-2-phenylindole, in the chloroplast DNA [355].

Detoxification of O 2 •− and H 2 O 2
Plants have evolved a multitude of enzymatic and non-enzymatic ROS-scavenging and quenching mechanisms. ROS-mediated signaling and ROS detoxification are coupled, as signaling is generally initiated by the oxidation of target molecules, that therefore also act as antioxidants (see reviews [10,356,357]). Here, we discuss the main scavenging mechanisms and antioxidant molecules controlling ROS in the chloroplasts, with emphasis on ROS detoxification in the thylakoid membrane, or stromal-scavenging mechanisms in its immediate vicinity. O 2 •− , produced in chloroplasts, is scavenged efficiently by copper/zinc SODs residing on the stromal face of the thylakoid membrane [12,358]. The dismutation reaction, catalyzed by CuZnSODs, is described in Reactions (5), (44) and (45). While SOD is the main catalyst, the dismutation reaction can also be catalyzed by redox reactive metals such as manganese [359,360], or it can occur non-catalytically [68]. O 2 •− can also oxidize two highly important chloroplast antioxidants, AscH 2 [62,64] (Reaction (34)) and GSH [66,67] (Reaction (37)). H 2 O 2 is reduced by AscH 2 in a reaction catalyzed by APXs (Reactions (62)-(64)) [93]. The net reaction of H 2 O 2 scavenging by AscH 2 can be summarized as (Reaction (126)).
The reaction produces water and MDA. Different APX isoenzymes are found in different chloroplast compartments. Stromal APXs and thylakoid APXs have specific roles in, e.g., plant development, but exhibit functional redundancy in ROS detoxification in mature leaves [361][362][363]. The rest of the ascorbate-glutathione cycle regenerates AscH 2 [12,364]. The first step is the reduction of MDA to AscH 2 by Fd red (Reaction (127)), MDA + Fd red + 2H + → AscH 2 + Fd (127) or by NADPH in a reaction catalyzed by MDA reductase (Reaction (128)).
The complete description of the catalytic cycle of reduction of MDA to AscH 2 is described in [12]. The MDA molecules that are not immediately reduced dismutate non-catalytically, forming AscH 2 and DHA (Reaction (113)).
NADPH, formed by the PETC, is then used by glutathione reductase to reduce GSSG back to GSH (Reaction (130)) thereby completing the ascorbate-glutathione cycle. The functions of APXs in plants have been reviewed by [365]. Because the electrons utilized in the reduction of O 2 to O 2 •− by PSI originate from water molecules broken down by PSII, and the end product of the production and scavenging of H 2 O 2 is water, the whole scavenging system is often referred to as the water-water cycle [12]. PRXs, particularly two-cysteine peroxiredoxins (2-Cys PRX), have been shown to function in conjunction with thylakoid APXs in downplaying H 2 O 2 accumulation during conditions causing oxidative stress in plants [366][367][368]. 2-Cys PRXs facilitate a peroxidative reduction of H 2 O 2 , utilizing electrons from NADPH in a reaction catalyzed by thioredoxin reductase C or, less efficiently, from reduced TRXs [366][367][368]. The catalytic cycle of peroxide detoxification by 2-Cys PRXs and their subsequent regeneration is described in detail in [369]. NADPH is produced both in the light, by PETC, and in the dark, by the oxidative pentose phosphate pathway [370], whereas TRXs are recycled to their reduced form by Fd produced in the light by PSI; the reduction in TRXs is catalyzed by thioredoxin reductases [371]. 2-Cys PRXs are not the only enzymes that can facilitate TRX-dependent H 2 O 2 detoxification, as glutathione peroxidases also utilize TRXs as substrates instead of reduced glutathiones in plant chloroplasts, and can likely initiate a similar cycle to 2-Cys PRXs [372][373][374][375]. Many other components have been suggested to take part in the recycling of the PRX/glutathione peroxidases-initiated H 2 O 2 detoxification cycle, such as glutaredoxin, cyclophilins and AscH 2 [369].

Detoxification of 1 O 2
Carotenoids and tocopherols are the main antioxidants against 1 O 2 in chloroplasts [43]. Carotenoids function in the NPQ of singlet excited Chl (reviewed in [376,377]), quench 3 Chl and quench and scavenge 1 O 2 . Each LHCII subunit contains two luteins, a neoxanthin and a violaxanthin/zeaxanthin [378]. All eight Chl a molecules of an LHCII subunit are positioned within close proximity to either of the two luteins or neoxanthin, which facilitates efficient 3 Chl-quenching especially by lutein, and lowers the probability of 3 Chl interaction with O 2 , quenching 95% of 3 Chl in LHCII [43,[369][370][371][372][373][374][375][376][377][378][379][380][381]. Violaxanthin and zeaxanthin are not likely to be involved in 3 Chl quenching in LHCII, as they are bound far from the Chl molecules [382,383]. However, zeaxanthin can quench 3 Chls in the monomeric Lhcb antenna subunits of PSII (Lhcb4-6) and in the dimeric Lhca subunits of PSI antennae [384]. In LHCII, zeaxanthin is specifically involved in NPQ. In high light, zeaxanthin is produced by violaxanthin de-epoxidase from violaxanthin through the intermediate antheraxanthin, and the newly formed zeaxanthin replaces violaxanthin in LHCII. A switch back to moderate light or darkness induces the epoxidation of zeaxanthin back to violaxanthin and the subsequent replacement of zeaxanthin with violaxanthin in LHCII [385].
The PSII core, consisting of the proximal antennae CP43 and CP47, the Mn-cluster and the RC (D1/D2/Cyt b559) [386,387], binds 11 β-carotenes, two of which are located in the RC [386]. The distance between these two β-carotenes and the RC Chl P680 is too long to allow the participation of the β-carotenes in quenching of 3 P680 [170,191,386,388]. However, there are indications that β-carotenes in other parts of the isolated PSII core are likely to quench 3 Chls [386].
The detoxification of 1 O 2 itself [192] by carotenoids occurs mainly through physical quenching via electronic energy transfer mechanism (Reaction (17), where A is a carotenoid). The resulting triplet state of the carotenoid ( 3 Car) dissipates its excitation energy via a nonradiative transition to its ground state [23,41,43,389]. Carotenoids can also take part in the chemical scavenging of 1 O 2 [43,390]. Oxidation products of β-carotene found in plants in high light suggest that 1 O 2 can oxidize the β-carotenes of PSII reaction centre [301,390]. β-cyclocitral (β-CC), a volatile product of oxidation of β-carotene by 1 O 2 , has been shown to be involved in cell signaling [391] (see Section 6.1). In LHCII, 1 O 2 produced by the interaction between O 2 and the residual 3 Chl that is not quenched by carotenoids, is rapidly inactivated, due to the abundance of carotenoids in LHCII and free carotenoids such as zeaxanthin in the surrounding lipid matrix [388,392].
Other antioxidants in the thylakoid membrane are not bound to LHCs or, in stroma, offer an even greater capacity for the physical or chemical quenching of 1 O 2 . Tocopherols, or specifically α-tocopherol, are considered as important antioxidants against 1 O 2 [393]. The rate constants of the physical quenching of 1 O 2 by tocopherols in organic solvents are significantly higher than those of chemical scavenging [389], suggesting that, similarly to carotenoids, the main quenching mechanism by α-tocopherol is physical quenching (Reaction (17)) [43]. However, the oxidation of α-tocopherol by 1 O 2 produces 8-hydroperoxy-tocopherone that can be re-reduced to α-tocopherol by AscH 2 [394,395], which lends the recyclability of the stromal ascorbate-glutathione cycle to 1 O 2 detoxification of the lipid phase. AscH 2 also has the capacity to scavenge 1 O 2 (reaction 24) that reaches the stroma [47]. Chloroplasts contain flavonoids in the envelope membrane, and they have the potential to quench 1 O 2 both physically and chemically [396,397]. Even though the relatively remote location from the most prominent 1 O 2 production sites does put their role as 1 O 2 antioxidants in question, flavonoids have been shown to be involved in lowering the amount of 1 O 2 in high light in vivo [397]. Other potential 1 O 2 antioxidants include polyunsaturated fatty acids [4], PQH 2 [48,[305][306][307]398] and isoprene [399,400].

ROS Produced by Plant Chloroplasts Function as Signaling Molecules
ROS are known to participate in retrograde signaling, acclimation to biotic or abiotic stresses, programmed cell death (PCD) and many other processes (for recent reviews, see [9][10][11][401][402][403][404][405][406][407]). Here, we aim to briefly summarize what is known (and what is not) about how chloroplast-derived ROS are sensed and how the signaling cascades are initiated. Signaling by ROS produced by enzymes like NADPH-oxidase (reviewed in [408]) will not be discussed here.

Signaling by 1 O 2
The lifetime of 1 O 2 in plant cells has not been measured, but is generally assumed to be too short (for review, see [23]; Section 2.1.4) to enable diffusion out of chloroplasts and, consequently, 1 O 2 itself is unlikely to function as a messenger molecule. Instead, the accumulation of β-CC (a reaction product of β-carotene and 1 O 2 ) has been shown to induce gene expression, leading to stress (e.g., high light) acclimation [391,409,410]. In theory, β-CC could directly travel to the nucleus and activate 1 O 2 responsive genes (for discussion, see [411]), however, direct evidence is lacking. Methylene Blue Sensitivity 1 protein might participate in transferring the signal from cytosol to the nucleus [412,413]. In addition, β-CC can be converted to water-soluble β-cyclocitric acid, which also could function as a signaling molecule [410]. Other oxidation products of 1 O 2 might have signaling functions, too [346,414]. Another 1 O 2 -induced pathway involves the Executer1 (EX1) (and possibly Executer2) proteins [415,416]. The oxidation of a tryptophan residue of EX1, presumably by 1 O 2 [417], leads to the degradation of EX1 by FtsH, a protease that is also important to the repair cycle of PSII [418]. Afterwards, a signaling cascade leading to PCD is activated [419]. EX1 is not simply a repressor of the PCD pathway [407], however, it is not understood how the degradation of EX1 leads to the induction of PCD. In addition to cell death, EX1 is important in systemic acquired acclimation [420].
The β-CC and EX1 pathways are thought to operate independently [391]. A possible explanation of the need for two pathways is that small amounts of 1 O 2 lead to acclimation responses while larger amounts initiate PCD (and still higher amounts cause damage and unregulated cell death [403]). Accordingly, under severe stress, the β-CC pathway, through Oxidative Signal-Inducible 1 kinase, may also lead to PCD, but even this route is EX1-independent [421,422]. Most β-CC is produced from the β-carotene located in the RC of PSII [390,423], and therefore in the grana core [424], whereas EX1 is located in grana margins [425]. As 1 O 2 is not expected to diffuse far, the site of production rather than the amount may determine which signaling pathway is activated.
Why do plants need to react differently to 1 O 2 produced in different sites? PSII repair occurs mainly in grana margins (for a review, see [426]), and it has been speculated that EX1 would activate PCD if PSII repair is impaired, possibly under adverse environmental conditions when loose Chls might produce 1 O 2 [425]. Chl turnover was shown to associate with the repair of PSII [427], implying 1 O 2 generation, even though Chl synthesis and degradation are tightly regulated and loose Chls are thought to be bound to specific proteins that prevent 1 O 2 production [428]. During a low light to high light transition, the FtsH-protease may get transiently inactivated (possibly indirectly by H 2 O 2 ; [429]), thus preventing activation of the EX1-induced PCD. This is in agreement with the view that the EX1 pathway does not respond to high light stress, but it is the β-CC pathway that initiates high light acclimation. Alternatively, the EX1 pathway might be important in plant defense against pathogens [402,430]. 1 O 2 produced by Chl catabolites has been proposed to be involved in the hypersensitive response [431], and similarly to the flu-mutant [115], 1 O 2 produced by Chl catabolites has been suggested to initiate the EX1 pathway also in the wild type [432]. Interestingly, NADPH-protochlorophyllide oxidoreductases were shown to associate with EX1 and FtsH [425], though the interactions may be weak or transient, as they are not always observed [433]. PSII is a target of many pathogens [434] and a non-functional PSII repair cycle might also be involved in plant immunity [435]. However, the physiological role of the EX1 pathway is still unclear.

Signaling by H 2 O 2
In contrast to 1 O 2 , the long lifetime of H 2 O 2 enables its function as a messenger molecule. Exposito-Rodriguez et al. [436] observed that photosynthesis-derived H 2 O 2 rapidly accumulated in the nuclei, and the addition of cytosolic H 2 O 2 scavengers did not prevent this. The authors proposed that H 2 O 2 originated from chloroplasts closely associated with the nucleus. The diffusion of H 2 O 2 through membranes is not extremely rapid [95], but the transport may be facilitated by (specialized?) aquaporins [95,437,438]. The formation of stromules has been observed under stress [439], and they have been suggested to allow for direct contact between chloroplasts and the nucleus [440]. Another hurdle that chloroplast-originated H 2 O 2 needs to overcome is that the powerful antioxidant systems of stroma (see Section 5.1) are believed to efficiently scavenge H 2 O 2 . Accordingly, it has been proposed that H 2 O 2 produced inside the thylakoid membranes (see Section 3.2) might have a great importance in signaling [441]. On the contrary, a meta-analysis of 79 transcriptomic studies concluded that ROS responses are determined by timing rather than the site of origin [442]. Therefore, H 2 O 2 may participate in multiple pathways, some of which are sensitive to the site of H 2 O 2 production [443].
H 2 O 2 is involved in many signaling pathways. For example, photosynthesis-derived ROS, probably H 2 O 2 , may induce enzymatic O 2 •− production by cytosolic NADPH-oxidases 408]. In addition, a reduced PQ pool was proposed to cause stomatal closure via H 2 O 2 accumulation [444].
Borisova-Mubarakshina et al. [445] showed evidence that H 2 O 2 regulates PSII antenna size in barley during long-term acclimation to high light. It is not clear what senses H 2 O 2 in plant cells. SAL1 (an inositol polyphosphate 1-phosphatase) degrades phosphoadenosine phosphate (PAP) in chloroplasts. The oxidation of cysteine residues of SAL1, e.g., under high light, probably by H 2 O 2 , leads to the inactivation of SAL1 and accumulation of PAP [446]. PAP can be transported into the nucleus and activate genes protecting plants from oxidative stress [447,448]. In addition, it has been proposed that a glutathione peroxidase [449], heat shock transcription factors [450], APX [362] and protein phosphatases (reviewed in [451]) might function as H 2 O 2 sensors.
In general, genes responding to 1 O 2 were found to differ from those known to be regulated by H 2 O 2 [391]. The available data suggest that H 2 O 2 actually antagonizes EX1-mediated 1 O 2 signaling [452,453]. On the other hand, H 2 O 2 and the β-CC-mediated 1 O 2 signaling pathways may converge at Oxidative Signal-Inducible 1 kinase [407,454]. β-CC also down-regulates SAL1 and up-regulates genes generating PAP [391,423], supporting the view that both H 2 O 2 -and β-CC-signaling pathways induce stress acclimation.  [458].

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.