Singlet oxygen production by photosystem II is caused by misses of the oxygen evolving complex

Summary Singlet oxygen (1O2) is a harmful species that functions also as a signaling molecule. In chloroplasts, 1O2 is produced via charge recombination reactions in photosystem II, but which recombination pathway(s) produce triplet Chl and 1O2 remains open. Furthermore, the role of 1O2 in photoinhibition is not clear. We compared temperature dependences of 1O2 production, photoinhibition, and recombination pathways. 1O2 production by pumpkin thylakoids increased from −2 to +35°C, ruling out recombination of the primary charge pair as a main contributor. S2QA − or S2QB − recombination pathways, in turn, had too steep temperature dependences. Instead, the temperature dependence of 1O2 production matched that of misses (failures of the oxygen (O2) evolving complex to advance an S‐state). Photoinhibition in vitro and in vivo (also in Synechocystis), and in the presence or absence of O2, had the same temperature dependence, but ultraviolet (UV)‐radiation‐caused photoinhibition showed a weaker temperature response. We suggest that the miss‐associated recombination of P680 +QA − is the main producer of 1O2. Our results indicate three parallel photoinhibition mechanisms. The manganese mechanism dominates in UV radiation but also functions in white light. Mechanisms that depend on light absorption by Chls, having 1O2 or long‐lived P680 + as damaging agents, dominate in red light.


Introduction
Molecular oxygen (O 2 ) in its ground state is a biradical, as the two unpaired electrons have parallel spins (thus, O 2 is a triplet). Exchange of energy and spin with a molecule in a triplet excited state can turn O 2 to a highly reactive singlet state ( 1 O 2 )for a review, see Schweitzer & Schmidt (2003). 1 O 2 causes cellular damage by oxidizing biomolecules containing double bonds (Schweitzer & Schmidt, 2003;Halliwell & Gutteridge, 2015;Di Mascio et al., 2019). 1 O 2 also generates cellular signals leading to acclimation to high light and/or to cell death via apoptotic pathways in photosynthetic organisms (Lee et al., 2007;Ramel et al., 2012b;Crawford et al., 2018).
Both photosystem I (PSI) and photosystem II (PSII) produce 1 O 2 (Macpherson et al., 1993;Cazzaniga et al., 2012), the majority being produced by the PSII core after a charge recombination reaction (Telfer et al., 1994;Ramel et al., 2012a); however, contrasting data have also been published (Santabarbara et al., 2007). The excited triplet state of the PSII reaction center Chl(s) ( 3 P 680 ) can react with the ground-state O 2 , which results in the formation of 1 O 2 and the singlet ground state of P 680 . 3 (Rutherford, 1989;Tyystjärvi & Vass, 2004), but have been suggested to be responsible for 1 O 2 production in weak light (Keren et al., 1997). The primary pair, in turn, recombines with nanosecond kinetics and may produce 3 P 680 when Q A is reduced (Vass & Styring, 1993). The P 680 + Q A − pair is available for recombination only if the O 2 -evolving complex (OEC) of PSII fails to reduce P 680 A transient failure of the OEC to reduce P 680 + is called a miss, and c. 8% of charge separations fail in this way (Forbush et al., 1971;Isgandarova et al., 2003;Pham et al., 2019). In this case, charge separation leads to reduction of Q A , followed by recombination of P 680 + Q A − . Misses represent reaction equilibria between S-state advancement and recombination reaction in normal, active PSII. The percentage of misses is similar in the thermophilic cyanobacterium Cyanosiphon merolae and in spinach, although the redox potential of the Q A /Q A − pair is much less negative in C. merolae than in spinach (Pham et al., 2019). Furthermore, the time constant of the recombination of P 680 + Q A − is 100-200 μs in Tris-washed thylakoids, which lack a functional donor side (Renger & Wolff, 1976); the kinetics in active PSII are not known. A 100-200 μs time constant of the P 680 + Q A against the majority of S-state transitions. These considerations imply that a miss occurs because the rate of S-state advancement in a fraction of OECs is so slow that P 680 + Q A − recombines, not because the recombination occasionally wins competition with normal advancement of the S-state. Electron paramagnetic resonance (EPR) measurements (Han et al., 2012) and modeling of flash O 2 data (Pham et al., 2019) indicate that most misses occur in the S 2 to S 3 transitions, but other S-states may also fail to advance.
To better understand 1 O 2 production and photoinhibition, we measured their temperature dependences and compared them with temperature dependences of recombination pathways. The results indicate that the miss-associated recombination reaction is crucial for the formation of 1 O 2 . The temperature dependence of photoinhibition, in turn, was similar but not identical under different wavelengths, suggesting that several mechanisms contribute to photoinhibition. In white light, the contribution of the manganese (Mn) mechanism (Hakala et al., 2005) was estimated to be 63-68%. Under monochromatic light, contributions from two other mechanisms (a 1 O 2 -dependent and a P 680 + -dependent mechanism) increase toward longer wavelengths.

Singlet oxygen measurements
Pumpkin thylakoids (100 μg Chl ml −1 ) were incubated for 5 min in a photoinhibition buffer (40 mM HEPES-potassium hydroxide (pH 7.4), 1 M betaine monohydrate, 330 mM sorbitol, 5 mM MgCl 2 , and 5 mM NaCl) in darkness and then illuminated for 2 min with strong light, and the light-induced changes in O 2 concentration in the absence and presence of 20 mM L-histidine (Sigma-Aldrich) were recorded (Telfer et al., 1994;Rehman et al., 2013). An optode (Firesting O 2 FSO2-0x with OXSP5 sensor spots; PyroScience GmbH, Aachen, Germany), a homemade cuvette, and a 10 W cold-white LED (PPFD 2000 μmol m −2 s −1 ; for the spectrum, see Supporting Information Fig. S1) were used for measurements at 5, 15, 25, and 35°C. The optode was calibrated with air-saturated water. A measurement at −2°C, where water cannot be used for calibration, and a comparison measurement at 15°C, were done using an O 2 electrode (Hansatech, King's Lynn, UK) and a slide projector equipped with a halogen lamp (PPFD 3000 μmol m −2s −1 ; for the spectrum, see Fig. S1). The O 2 electrode was calibrated with air-saturated photoinhibition buffer that remained liquid at −2°C, before and after addition of solid sodium dithionite to remove O 2 . With both devices, the rate of 1 O 2 production was calculated as the difference in the rate of O 2 consumption in the presence and absence of 20 mM histidine. The temperature dependence of the reaction between histidine and 1 O 2 was tested by illuminating (PPFD 2000 μmol m −2 s −1 ) 1 μM rose bengal in the presence of 20 mM histidine. No significant consumption of O 2 was observed in the absence of histidine.

Recombination reactions
Thermoluminescence was measured with a homemade luminometer from pumpkin thylakoids (600 μg Chl ml −1 ) as previously described (Tyystjärvi et al., 2009), in the presence and absence of 20 μM 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU), with a heating rate of 0.56°C s −1 . A 1 J xenon pulse was fired at −10°C. The rate constants of the S 2/3 Q A/B − recombination reactions were calculated as functions of temperature with COPASI (Hoops et al., 2006) assuming that each rate constant depends on temperature according to the Arrhenius equation. Three competing recombination routes (direct, indirect, and excitonic) were assumed for the analysis of the Q band (Rappaport & Lavergne, 2009), whereas the B band was analyzed as a single reaction (Randall & Wilkins, 1945;Tyystjärvi & Vass, 2004).

Fluorescence measurements in the light
Fluorescence parameters were measured during illumination from Synechocystis cells (optical density at 730 nm 0.8-1.1), pumpkin thylakoids (100 μg Chl ml −1 ), and detached pumpkin leaves with a PAM-2000 (Walz, Effeltrich, Germany) at 5-35°C. A saturating pulse was fired to calculate (F M − F 0 )/F M (=F V /F M ) after dark acclimation (30 s for thylakoids and 30 min for leaves). Thereafter, the sample was illuminated with white light (PPFD 750 μmol m −2 s −1 from a slide projector for Synechocystis, and PPFD 1500 μmol m −2 s −1 from a 500 W high-pressure xenon lamp with a water filter for pumpkin; for the spectra, see Fig. S1). To calculate (F M 0 − F)/F M 0 , saturating pulses were fired after 1 min for thylakoids and after 15, 30, and 45 min for Synechocystis cells and pumpkin leaves. To estimate the reduction state of Q A , 1 − qL (Kramer et al., 2004) was calculated, using an estimation (Oxborough & Baker, 1997) for F 0 0 (see Mattila et al., 2020).

Photoinhibition treatments
Pumpkin thylakoids (100 μg Chl ml −1 ), detached pumpkin leaves, or intact Synechocystis cells (optical density at 730 nm 0.8-1.1) were illuminated at various PPFD values, wavelengths, and temperatures, as indicated. Before a treatment, leaves were incubated overnight at PPFD 10-20 μmol m −2 s −1 with the petioles in a solution with 0.4 mg ml −1 lincomycin (Sigma-Aldrich). Lincomycin (0.4 mg ml −1 ) was added to Synechocystis cells right before the treatment. Thylakoids were illuminated in photoinhibition buffer, unless otherwise mentioned, and Synechocystis cells in BG-11. The samples were mixed during the treatments. Red (> 650 nm) or blue (400-450 nm) light was obtained with a 500 W high-pressure xenon lamp equipped with a long-pass or a short-pass edge filter (LL-650 and LS-450, respectively; Corion, Holliston, MA, USA). Monochromatic light was obtained with band-pass filters (full width at half maximum 10 nm; Corion; Newport, Irvine, CA, USA). UV radiation was obtained with VL-8.LC (365 and 254 nm) and VL-8.M (312 nm) lamps (Vilber Lourmat, Collégien, France; for the spectra, see Havurinne et al., 2021). White light was obtained from a 1000 W (Sciencetech, London, ON, Canada) or 500 W (Oriel Instruments; Newport) high-pressure xenon lamp (when measuring the temperature dependence of photoinhibition in thylakoids at 5°C intervals, and for the temperature dependence of photoinhibition in leaves), from a slide projector equipped with a low-voltage halogen lamp (Synechocystis), or from a 10 W cold-white LED (all other experiments). For laser-pulse-induced photoinhibition, pumpkin thylakoids (27 μl, 76 μg Chl ml −1 ) were illuminated in a 3 × 3 × 10 mm 3 cuvette with 532 nm, 4 ns, 12.5 mJ pulses from a Nd : YAG laser (Continuum, San Jose, CA, USA). The interval between the laser pulses was 0.1 s (240 flashes in total), 10 s (100 flashes) or 30 s (40 flashes).
α-Tocopherol, when used, was vigorously mixed in dimethyl sulfoxide and immediately added to thylakoid suspension, which was then vigorously mixed for 20 s. In control experiments, only dimethyl sulfoxide was added. Anaerobic conditions, applied when indicated, were achieved by flushing the sample continuously with nitrogen (N 2 ) gas. In control (aerobic) experiments, the sample was flushed with air. Freshly prepared 20 mM sodium bicarbonate was used to test recovery of PSII activity in isolated thylakoids after anaerobic photoinhibition.
Experiments with isolated thylakoids were always repeated under otherwise identical conditions in the dark, to determine the rate of dark inactivation of PSII.

Quantification of photoinhibition
Before and after treatments, light-saturated O 2 evolution was measured from aliquots of treated thylakoids or from thylakoids isolated from treated leaves, at 22°C, or from aliquots of illuminated Synechocystis suspension at 32°C, using an O 2 electrode (Hansatech, King's Lynn, UK) as previously described (Hakala et al., 2005) with artificial electron acceptors (0.5 mM 2,6dimethylbenzoquinone (DMBQ) with thylakoids; 0.5 mM 2,6dichlorobenzoquinone and 0.5 mM hexacyanoferrate(III) with Synechocystis). In some experiments, as indicated, PSII activity was estimated by measuring the fluorescence parameter F V /F M with a Fluorpen (Photon Systems Instruments, Brno, Czech Republic) after at least 5 min (thylakoids) or 30 min (leaves) dark incubation.
The rate constant of photoinhibition k PI was calculated by fitting the loss of O 2 evolution or decrease in F V /F M , as indicated, to the first-order reaction equation in SIGMAPLOT (Systat Software Inc., Palo Alto, CA, USA). In the case of thylakoids, the final k PI values were obtained by subtracting the first-order rate constant of dark inactivation from the raw k PI value.

Activation energy
The activation energy E a was calculated by fitting the dependence of the rate constant k on absolute temperature T to the Arrhenius equation by using linear regression of log e (k) to 1/T according to

Detection of carbon-centered radicals
We mixed 5.9 mg of α-(4-pyridyl 1-oxide)-N-tert-butylnitrone (POBN; Enzo Life Sciences Inc., New York, NY, USA) in 0.6 ml of thylakoid suspension (100 μg Chl ml −1 ) to get a final concentration of 50 mM POBN. POBN-R-adduct (the reaction product of POBN and a carbon (C)-centered radical) was detected before and immediately after illumination or dark incubation with an EPR spectrometer (Miniscope MS 5000; Magnettech GmbH, Berlin, Germany). The measurement parameters were as follows: 60 s sweep time (three technical repetitions) at 330-340 mT, 0.2 mT modulation, 100 kHz frequency, and 10 mW power. C-centered radicals were quantified by the height of the first positive peak at 334.5-334.8 mT of the EPR signal.

Results
Singlet oxygen production by thylakoid membranes shows a positive temperature dependence Isolated pumpkin thylakoids were illuminated in high light at −2, 5, 15, 25, and 35°C, and 1 O 2 production during the illumination was measured with the histidine method (Rehman et al., 2013). 1 O 2 production increased five-fold from −2 to +35°C, and the data showed a reasonable fit to the Arrhenius equation (Fig. 1). The reaction between 1 O 2 and histidine, probed by using rose bengal as a 1 O 2 sensitizer, showed much weaker temperature dependence (Fig. S2), indicating that the temperature dependence in Fig. 1 reflects the temperature dependence of 1 O 2 production by thylakoids and not that of the reaction between 1 O 2 and histidine.
Temperature dependence of singlet oxygen production matches that of the misses Photosystem II is responsible for most 1 O 2 produced by thylakoids (Cazzaniga et al., 2012). However, to understand which PSII charge recombination reaction is responsible for the 1 O 2 production, the temperature dependences of recombination reactions were compared with the observed temperature dependence of 1 O 2 formation. The sub-nanosecond recombination of P 680 + Pheo − is known to occur readily even at 77 K (Zabelin et al., 2016), indicating negligible E a , and the triplet state 3 P 680 has only c. 0.025 eV lower energy than the P 680 + Pheo − radical pair (Dau & Zaharieva, 2009), implying that the temperature dependence of triplet formation via recombination right after primary charge separation is negligible at physiological temperatures. Therefore, rapid recombination of the primary pair does not account for the observed 1 O 2 formation by PSII.
Next, recombination of the charge pairs consisting of a reduced quinone acceptor (Q A − or Q B − ) and a hole in the OEC in the state S 2 were studied with thermolumiscence. The rate constants of the S 2 Q A − → S 1 Q A and S 2 Q B − → S 1 Q B recombination reactions can be calculated from the thermoluminescence Q (with DCMU) and B (no DCMU) bands, respectively. Previously, it has been shown that three competing pathways operate for the S 2 Q A − → S 1 Q A recombination (Rappaport & Lavergne, 2009). The 'excitonic' pathway leads, via P 680 + Pheo − , to the short-lived singlet excited state of P 680 and produces the luminescence. The 'indirect' pathway also has the primary pair as an intermediate; it does not produce luminescence but can instead lead to the formation of 3 P 680 when the P 680 + Pheo − pair recombines. The third, also non-luminescent, pathway has been interpreted as direct recombination of a hole in the OEC and an electron in Q A − without the primary pair as an intermediate (Rappaport & Lavergne, 2009).
The Q band peaked at 12°C and the B band at 37°C (Fig. 2a). We applied the model of Rappaport & Lavergne (2009) for the analysis of the Q band (see also Rantamäki & Tyystjärvi, 2011). For the B band, a first-order model with one recombining component (Randall & Wilkins, 1945;Tyystjärvi & Vass, 2004) was used. The rate constant of each pathway was obtained by fitting the curve to the respective model (Table S1). The rate constant of the indirect pathway of S 2 Q A − recombination (k indirect ) and the rate constant of S 2 Q B − recombination (the reactions that may lead to 1 O 2 production) showed steep temperature dependences in the −2 to +35°C range (Fig. 2b). Such direct comparisons between the rate constant of recombination and 1 O 2 formation are justified in isolated thylakoids in which PSII reaction centers would remain essentially closed during illumination, irrespective of the temperature (Fig. S3). In leaves, however, the rate of 3 P 680 production from a recombination reaction under continuous illumination would be proportional to the rate constant of the recombination times the concentration of its substrate (a reduced quinone in this case). To measure the effect of temperature on the closure of PSII centers, we estimated the relative concentration of Q A − , [Q A − ] rel , at 5, 20, and 35°C, in pumpkin leaves using the fluorescence parameter 1 − qL (Figs 2 c, S3). However, the product k indirect × [Q A − ] rel showed a highly similar temperature dependence to the rate constant k indirect alone, much steeper than the temperature dependence of 1 O 2 production (Fig. 2d). The temperature dependence of the rate constant of S 2 Q B − recombination (inset of Fig. 2b) was also steeper than that of 1 O 2 formation; no correction for the in vivo rate of S 2 Q B − recombination was deemed necessary because Q B is a two-electron carrier and therefore the concentration of Q B − would not depend strongly on temperature in continuous light. Thermoluminescence peaks originating from the recombination of the S 3 Q A − and S 3 Q B − states are not drastically different from those related to the S 2 state (Vass & Govindjee, 1996); recombination reactions involving the S 3 state instead of S 2 would therefore not change the conclusions drawn herein. The 'direct' pathway of S 2 Q A − recombination also contributed to our experimental data (Fig. 2b), but this pathway does not have a radical pair intermediate (Rappaport & Lavergne, 2009) and therefore cannot contribute to 1 O 2 production. To summarize, the S 2/3 Q A/B − → S 1/2 Q A/B recombination reactions cannot be the main producers of 1 O 2 in thylakoids.
The temperature dependence of misses of OEC has been earlier measured from spinach thylakoids (Isgandarova et al., 2003), and a comparison shows that the temperature dependence of

Research
New Phytologist misses, especially at 10-35°C, is highly similar to that of 1 O 2 production (Figs 1, 2d). These data suggest that 1 O 2 is produced mainly via the miss-associated P 680 + Q A − recombination reaction (see Discussion section for details).
Temperature dependence of photoinhibition is positive, universal among different species, and depends on the wavelength of illumination Connections between 1 O 2 and photoinhibition observed in previous literaturefor a review, see Tyystjärvi (2013)prompted us to compare their temperature dependences. The temperature dependence of photoinhibition was measured by illuminating isolated pumpkin thylakoids with strong white light at 3-35°C. The loss of light-saturated O 2 evolution (water (H 2 O) to DMBQ), measured from aliquots of the treated suspension, was fitted to the first-order reaction equation (Fig. S4a) to obtain a raw rate constant, from which the rate constant of dark inactivation, occurring in isolated thylakoids, was subtracted to calculate the rate constant of photoinhibition k PI (Fig. S4b). To ensure that the results are not a property of isolated systems only, the temperature dependence of photoinhibition was also measured in vivo by illuminating intact pumpkin leaves or cells of the cyanobacterium Synechocystis sp. PCC 6803 in the presence of lincomycin to block PSII repair (Fig. S4c). The data show an essentially identical, positive temperature dependence of photoinhibition for pumpkin thylakoids, Synechocystis cells, and pumpkin leaves; k PI approximately doubled in the measured physiological temperature range (Fig. 3a). Furthermore, the temperature dependence of photoinhibition resembled those of misses and 1 O 2 production (Fig. 2d).
Photoinhibition by strong nanosecond laser pulses has earlier been suggested to be caused by 1 O 2 specifically originating from S 2 Q A − and S 2 Q B − recombination reactions (Keren et al., 1997). Measurements of laser-pulse-induced photoinhibition in pumpkin thylakoids (Fig. 3b) confirmed the characteristic dependence of the photoinhibitory efficiency (per flash) on the time interval between the flashes (Keren et al., 1997). However, comparison of Figs 1-3 shows that the temperature response of the laser-pulse-induced photoinhibition does not resemble the temperature dependence of any of the S 2 Q A − or S 2 Q B − recombination pathways, indicating that photoinhibition induced with short laser pulses is not related to these recombination reactions. In addition, the temperature dependence of laser-pulse-induced photoinhibition did not resemble that of photoinhibition induced by continuous light (Fig. 3a), suggesting a different photoinhibitory mechanism.
The resemblance of the temperature dependence of photoinhibition caused by continuous high-intensity white light (Fig. 3a) with those of misses and 1 O 2 production ( Fig. 2d) suggests, in turn, that the miss-associated recombination reaction leads to the production of 1 O 2 , which then damages PSII. As 1 O 2 is not produced under UV radiation (Hideg & Vass, 1996), we tested whether the temperature dependence of photoinhibition would be lost in UV radiation. Different wavelengths of visible light were also tested. As shown previously (e.g. Hakala et al., 2005), the k PI value, when compared with photon flux density, is much higher under UV radiation than under visible light (Fig. 3c). Normalized data show that the positive temperature dependence remains, although it is somewhat milder in UV than in visible wavelengths, especially than in red light (Fig. 3d).

Photoinhibition proceeds similarly under aerobic and anaerobic conditions
Besides UV illumination, anaerobicity is a condition where photoinhibition has been previously shown to occur even though 1 O 2 is not produced. Therefore, to better understand the connection between 1 O 2 production and photoinhibition, we next illuminated thylakoids in anaerobic conditions. In this case, photoinhibition was assayed with the F V /F M fluorescence parameter. As shown by Sipka et al. (2021), F V /F M cannot be taken as a measure of the PSII quantum yield but can be used as an empirical PSII activity parameter. The experiments showed an essentially similar temperature dependence of photoinhibition under anaerobic and aerobic conditions (Fig. 4a). The action spectrum, another characteristic of the reaction mechanism, was similar for anaerobic and aerobic photoinhibition (Fig. 4b).
Photoinhibition in vitro can be reversible under anaerobic conditions, mainly because of depletion and rebinding of bicarbonate to PSII (Sundby et al., 1992). However, we did not observe any reversibility, nor did addition of bicarbonate affect photoinhibition (Fig. S5).
To further test the effect of 1 O 2 production on photoinhibition, we illuminated pumpkin thylakoids in the presence of two efficient 1 O 2 scavengers, water-soluble histidine and hydrophobic α-tocopherol, and found no effect on photoinhibition or on its temperature dependence (Fig. 4c). In the dark, these compounds did not show any clear effects either, except small protection by histidine against light-independent inactivation of PSII (Fig. S6). These results show that either photoinhibition is independent of 1 O 2 or has parallel mechanisms, some of them independent of O 2 .

New Phytologist
We also measured aerobic and anaerobic photoinhibition in the presence of DCMU, a herbicide that blocks electron transfer from Q A to Q B . Again, photoinhibition was measured with F V / F M . Interestingly, in the presence of DCMU, the temperature dependence was lost (Fig. 4d), and in the visible range the action spectrum was flatter than in the absence of DCMU (Fig. 4e). Owing to a relatively low resolution, we may have missed a previously observed peak at 670 nm (Santabarbara et al., 2001).
Photoinhibition proceeded faster in anaerobic than in aerobic conditions (Fig. 4b). To test whether accumulation of C-centered radicals would explain this result, we illuminated pumpkin thylakoids in the presence of the radical probe POBN and found more radicals after anaerobic than after aerobic illumination (Fig. 4f). However, accumulation of radicals did not show a clear temperature dependence at 5-35°C in the absence of DCMU (Fig. 4f), contrary to photoinhibition. Both in the presence and absence of O 2 , DCMU strongly suppressed radical accumulation but, curiously, also imposed a positive temperature dependence (Fig. 4f). Accumulation of C-centered radicals was negligible in the dark at 20°C, and the signal, www.newphytologist.com obtained by illumination, remained stable in the dark at 5-35°C (Fig. S7). A connection between misses and photoinhibition was further probed by comparing the pH dependences of the two phenomena. The k PI value decreased by one-third from pH 6.8 to pH 8.2 (Fig. S8), whereas misses show little pH dependence in this range (Messinger & Renger, 1994). However, Davletshina & Semin (2020) have suggested that the pH dependence of photoinhibition may reflect pH-dependent changes in the structure of the OEC. These changes may affect photoinhibition without affecting the miss rate.

Temperature of dark incubation affects fluorescence-based estimations of photoinhibition
Finally, we assayed photoinhibition in pumpkin leaves both with F V /F M and O 2 evolution. A similar positive temperature dependence was obtained irrespective of the assay method (Fig. S9a). However, if the 30 min dark incubation before the F V /F M measurement was conducted at 5°C (the illumination temperature), photoinhibition appeared to proceed faster than when the dark incubation was done at 22°C (Fig. S9b). Thus, dark-incubation temperature may greatly affect the extent of the observed decline in F V /F M .

Discussion
Miss-associated recombination of P 680 + Q A − is responsible for singlet oxygen production of photosystem II In plants, chloroplasts are the most important producers of 1 O 2 in the light (Hideg et al., 2002;Prasad et al., 2018), and the potential of 3 P 680 for the 1 O 2 production has been clear for many years (Telfer et al., 1994). 3 P 680 is produced by charge recombination reactions, but the importance of different recombination pathways has not been known. It has been suggested that S n Q A/ B − → S n−1 Q A/B reactions produce enough 3 P 680 and subsequently 1 O 2 to inactivate PSII in weak light or during illumination with short laser pulses (Keren et al., 1997;Vass, 2011). The recombination of P 680 + Pheo − has been suggested to be important in 1 O 2 production in strong light when Q A is mostly reduced (Vass, 2011;Rehman et al., 2013). Our data show that these recombination reactions cannot significantly contribute to 1 O 2 production; the temperature dependences of the slow recombinations are too steep and that of the rapid recombination of the primary radical pair is too flat to account for the observed temperature dependence of 1 O 2 formation by thylakoid membranes (Figs 1, 2). Furthermore, 3 P 680 is short-lived in the presence of Q A − (Hillmann et al., 1995;Santabarbara et al., 2003). The miss-associated recombination of P 680 + Q A − is the only reaction with a temperature response similar to the observed 1 O 2 production (Figs 1, 2), and therefore the data strongly suggest that 1 O 2 is produced in a reaction between O 2 and 3 P 680 , where 3 P 680 is formed by the miss-associated recombination of P 680 Fig. 5a). This reaction, like all recombination reactions of PSII (except for the recombination of the primary pair), is expected to have a high triplet yield because of the lack of spin correlation of the reactants.

Energetics of misses, recombination reactions, and singlet oxygen formation
The miss-associated recombination of S 1 P 680 + Q A − would obviously proceed via pathways equivalent to the reducing side of the indirect and excitonic pathways of S 2 Q A − recombination (Rappaport & Lavergne, 2009); that is, via the S 1 P 680 + Pheo − Q A intermediate. Assuming that the pre-exponential factor s takes similar values in P 680 + Q A − recombination as in S 2 Q A − recombination (Table S1), and knowing that the Gibbs energy change from P 680 (Dau & Zaharieva, 2009), the time constant of the recombination reaction P 680 + PheoQ A − → P 680 + Pheo − Q A (which may lead to 3 P 680 ) would be 191 μs at 25°C (see Calculations in Methods S1), in agreement with the time constant of 100-200 μs, measured by Renger & Wolff (1976).
A miss of the OEC occurs because the S-state does not always advance, not because the recombination of P 680 + Q A − occasionally wins competition with normal advancement of the S-state (Pham et al., 2019). The simplest mechanism by which misses can originate from the oxidizing side of PSII is that the OEC has a miss-prone state (OEC miss ), characterized by a slow S n → S n+1 transition. Thus, the temperature dependence of 1 O 2 formation reflects the activation energies of a multistep reaction beginning with the reaction OEC normal → OEC miss , which, based on analysis of the results of Isgandarova et al. (2003), has an E a of 0.224 eV (Fig. 5b). This E a is the enthalpy difference between the transition state of the reaction OEC normal → OEC miss and OEC normal . If we assume that the miss factor is 8% at 25°C and all misses occur in the S 2 → S 3 transition, then the equilibrium constant of the reaction OEC normal → OEC miss is 0.32. As K eq ¼ e ÀΔG r =k b T , where ΔG r is the Gibbs energy change of the reaction, this further implies that OEC miss is 0.03 eV above OEC normal .
After a charge separation and reduction of Q A , a multistep reaction forming 1 O 2 consists of the reaction OEC normal → OEC miss , recombination P 680 + Q A − → P 680 Q A , formation of 3 P 680 , and a reaction between O 2 and 3 P 680 (Fig. 5a). An effective E a of a multistep reaction is calculated by adding k b T and the sum of the standard-state enthalpies (subtracting those of the reactants) of all intermediates and transition states of the reaction, where each enthalpy value is multiplied by its degree of rate control (DRC) (Mao & Campbell, 2019). The radical pair P 680 + Pheo − represents the transition state of the P 680 + Q A − → P 680 Q A recombination reaction, which implies that the E a of the recombination reaction is the same as the redox potential difference of the Pheo/Pheo − and Q A /Q A − pairs, 0.33 eV (Dau & Zaharieva, 2009). A slightly smaller E a of 1 O 2 formation, 0.31 eV (Fig. 5b), indicates that contributions from the formation of 3 P 680 and the reaction between O 2 and 3 P 680 to the E a are close to zero. Indeed, formation of 3 P 680 occurs even at 10 K (Lendzian et al., 2003), indicating that the triplet formation does not depend on temperature at 2-35°C. 1 O 2 production by New Phytologist (2023)

Research
New Phytologist illuminated rose bengal, in turn, has an activation energy of only 0.055 eV (Fig. S2b). This E a belongs to a complicated reaction in which the conversion of 1 O 2 to O 2 via collision of H 2 O competes with the reaction of 1 O 2 with histidine. We now assume that the products DRC × E a for the reactions between 3 P 680 and O 2 , followed by the reaction between 1 O 2 and histidine, are negligible because these reactions occur after 1 O 2 formation and because the E a contribution of this series of steps would be comparable to the small E a measured for the production and detection of 1 O 2 in the rose bengal system. With these assumptions, direct application of the model of Mao & Campbell (2019) for the measured E a of 1 O 2 formation results in E a ( 1 O 2 ) = k b T + DRC1 × 0.224 eV + DRC2 × 0.03 eV + DRC3 × 0.33 eV = 0.31 eV, where DRC1 belongs to the moving of the reaction coordinate from OEC normal to the transition state of the reaction OEC normal → OEC miss , DRC2 belongs to the formation of OEC miss from the transition state, and DRC3 belongs to the transitional step P 680 We further assume that DRC2 is zero, and, taking into account that DRC1 + DRC2 + DRC3 = 1, we get DRC1 = 0.39 and DRC3 = 0.61 (calculations in Methods S1). Thus, the rate of 1 O 2 formation is controlled both by the frequency of misses and by the rate of the miss-associated recombination reaction.

Photoinhibition has three parallel mechanisms
Our data conclusively show that the temperature dependence of photoinhibition is positive both in vivo and in vitro. The result agrees with several earlier studies (Tyystjärvi et al., 1994;Lazarova et al., 2014;Ueno et al., 2016;Mattila et al., 2020) but contrasts with the findings of Tsonev & Hikosaka (2003) and Kornyeyev et al. (2003), who found a strong negative temperature dependence. The difference might suggest that, in our experiments, excitation pressure (suggested to be the cause of fast photoinhibition at low temperatures) was similar at all temperatures. Though this actually is true for the in vitro results (Fig.  S3), the PSII yield of pumpkin leaves showed a clear negative dependence on temperature, indicating decreased excitation pressure at higher temperatures (Fig. 2c). Furthermore, no connection between excitation pressure and k PI at different temperatures was found in our earlier study (Mattila et al., 2020). A possible reason why Tsonev & Hikosaka (2003) and Kornyeyev et al. (2003) observed a negative temperature dependence for photoinhibition is that they used the Chl fluorescence parameter F V / F M for quantification of PSII activity. A low temperature during dark incubation between illumination and fluorescence measurement can affect the results by slowing the relaxation of nonphotochemical fluorescence quenching (Fig. S9); Tsonev & Hikosaka (2003) indeed performed the dark incubation and photoinhibition treatment at the same temperature.
1 O 2 has often been suggested to function as a causal agent of photoinhibition (Vass, 2011;Tyystjärvi, 2013), but the occurrence of photoinhibition under UV radiation and in anaerobic conditions where 1 O 2 is not formed indicates that parallel mechanisms must function. We will first treat the visible-light-specific photoinhibition mechanisms as one combined mechanism functioning parallel with another mechanism that is fully responsible for photoinhibition under UV radiation. The apparent E a of photoinhibition, in the presence of two parallel pathways, is the weighted sum, calculated as E a (total) = (k 1 E 1 + k 2 E 2 )/(k 1 + k 2 ), where k i and E i are the rate constant and E a of reaction i (i = 1, 2), respectively. As k PI is proportional to photon flux density (Tyystjärvi & Aro, 1996), we can simplify the equation by the normalization k 1 + k 2 = 1. Pheo − ) recombination pathway or equivalent recombination from the miss configuration, but the miss recombination (red arrows) is much faster and has a temperature dependence matching singlet O 2 ( 1 O 2 ) production. (b) Activation energy of photoinhibition in pumpkin thylakoids (th.), pumpkin leaves (lf.), and Synechocystis cells (S.), obtained under illumination in white (W), blue (B), or red (R) light and in ultraviolet (UV)-A, UV-B, or UV-C radiation, measured using O 2 evolution as a PSII activity assay. Activation energy of photoinhibition in aerobic (air; as a control) and anaerobic (anaero) conditions, quantified using the fluorescence parameter F V /F M as a PSII activity assay (sparsely hatched bars). Activation energy of 1 O 2 production, measured with a histidine-based method. Activation energy of the miss factor of spinach thylakoids (M; densely hatched bars), calculated from published data (Isgandarova et al., 2003) We assume that the UV mechanism is triggered by light absorption by the Mn ions of OEC (Hakala et al., 2005) and first calculate the contribution of the Mn mechanism in visible light. Absorbance values of Mn complexes decrease with wavelength (e.g. Horner et al., 1999), and the contribution of the Mn mechanism is likely to be negligible in long-wavelength (red) visible light (Hakala et al., 2005;Ohnishi et al., 2005). We therefore postulate that photoinhibition in red light is entirely caused by mechanism(s) dependent on the light absorption by Chls, whose combination consequently must have an apparent E a of 0.46 eV (Fig. 5b). If photoinhibition of thylakoids in white light (E a 0.20 eV) is a linear combination of the UV mechanism (E a 0.12 eV in UV-A) and the mechanism(s) functioning in red light, then the Mn mechanism contributes 76% in white and 65% in blue light. For more details, see Calculations in Methods S1. Such a high contribution of a mechanism that is independent of Chl absorption agrees with the relatively small protective effect of nonphotochemical quenching of Chl excitations (Sarvikas et al., 2006;Havurinne & Tyystjärvi, 2017), although it can also be explained by assuming involvement of uncoupled Chls in photoinhibition (Santabarbara et al., 2001). The mechanistic details of the Mn mechanism, except for the release of Mn ions from OEC (Hakala et al., 2005), are still obscure, and therefore a dependence from O 2 or involvement of reactive O 2 species in the Mn mechanism cannot be excluded. However, photoinhibition induced by UV-A radiation was only weakly affected by removal of O 2 (Fig. 4b).
The fact that visible light induces photoinhibition in anaerobic conditions indicates that a 1 O 2 -independent mechanism must exist. The Mn mechanism cannot explain all such photoinhibition because the E a of anaerobic photoinhibition in visible light is 0.26 eV whereas that of the UV-active mechanism is 0.12 eV. The effect of O 2 on the rate of photoinhibition in UV radiation and visible light can be used to estimate the importance of the Mn mechanism in anaerobic conditions in visible light. The ratio k PI (anaerobic)/k PI (aerobic) is, on average, 3.45 in visible light but only 1.2 at 312 nm (Fig. 4b), suggesting that anaerobicity does not boost the Mn mechanism and thus that all increase in the rate of photoinhibition due to lack of O 2 is accounted by an O 2independent visible-light-specific mechanism. Following this assumption, the Mn mechanism, accounting for 76% of visiblelight photoinhibition in aerobic conditions, only contributes by 22% in anaerobic conditions. Now, the E a of the O 2independent visible-light mechanism, functioning in parallel to the Mn mechanism, becomes (0.26 eV − 0.22 × 0.12 eV)/ 0.78 = 0.30 eV, which is somewhat but not drastically higher than the E a of the misses (see Calculations in Methods S1), suggesting a causal relationship between misses and the O 2independent mechanism. We suggest that the O 2 -independent mechanism is the classical donor-side photoinhibition (Callahan & Cheniae, 1985;Chen et al., 1992;Jegerschöld & Styring, 1996), in which P 680 + , if long-lived, commits a harmful oxidation in PSII. Misses, by prolonging the lifetime of P 680 + by barring electron flow from OEC, may trigger this reaction in healthy PSII. If the miss-associated recombination has a time constant of 191 μs while electron transfer from Q A − to Q B takes 500 μs, then the miss mechanism would prolong the lifetime of P 680 + in 38% of the cases because electron transfer from Q A − to Q B occurs before the miss-associated P 680 + Q A − recombination. Thus, electron transfer from Q A − to Q B after a miss would lock PSII to the P 680 + state until the missed OEC finally advances the S-state.
The enhanced formation of C-centered radicals in anaerobic conditions (Fig. 4f) may suggest that oxidation of PSII proteins by P 680 + is linked to anaerobic photoinhibition. In line with this suggestion, DCMU strongly suppressed radical formation. However, anaerobic photoinhibition is not suppressed by DCMU, indicating that not all protein oxidation events lead to loss of PSII activity; DCMU may alter the probabilities of different oxidation events. The lack of matching temperature dependence in radical formation and photoinhibition responses (Fig. 4d,f) confirms that the relationship between photoinhibition and formation of protein radicals is not straightforward.
The relative rates of the 1 O 2 -dependent and O 2 -independent visible-light-specific mechanisms in the presence of O 2 cannot be estimated from the present data, but the E a of their combination, 0.46 eV, and that of the O 2 -independent reaction, 0.30 eV, imply that the E a of the 1 O 2 -dependent mechanism is ≥ 0.46 eV. The strong photoprotective effect of carotenoids (Jahns et al., 2000;Hakkila et al., 2013) suggests that the 1 O 2dependent mechanism is the major contributor among the Chldependent mechanisms. Owing to the high E a value of 1 O 2dependent photoinhibition, the relative contribution of this mechanism is expected to increase with temperature. This may explain why studies with cyanobacteria and algae that are cultivated and treated with high light in their cultivation temperature often yield results supporting the importance of 1 O 2 (Jahns et al., 2000;Fufezan et al., 2007;Hakkila et al., 2013;Treves et al., 2016), whereas the results of the present study (mostly conducted on plant thylakoids) suggest a large contribution of the Mn mechanism that has a low E a .
Temperature dependence may not exactly reflect E a for PSII charge recombination reactions with tunneling character (Moser et al., 2005). The good fit of the thermoluminescence data ( Fig. 2a) may suggest that E a is not damped, but the accuracy of the photoinhibition data does not allow estimation of dampening of the activation. Furthermore, we cannot exclude the possibility that the matching temperature dependences are fortuitous. In particular, 1 O 2 formation by uncoupled Chls (Santabarbara et al., 2001(Santabarbara et al., , 2002(Santabarbara et al., , 2003 might have a temperature dependence matching that measured for 1 O 2 (Fig. 1).

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.         Methods S1 Calculations.