Protection of photosystem I during sudden light stress depends on ferredoxin:NADP(H) reductase abundance and interactions

Abstract Plant tolerance to high light and oxidative stress is increased by overexpression of the photosynthetic enzyme Ferredoxin:NADP(H) reductase (FNR), but the specific mechanism of FNR-mediated protection remains enigmatic. It has also been reported that the localization of this enzyme within the chloroplast is related to its role in stress tolerance. Here, we dissected the impact of FNR content and location on photoinactivation of photosystem I (PSI) and photosystem II (PSII) during high light stress of Arabidopsis (Arabidopsis thaliana). The reaction center of PSII is efficiently turned over during light stress, while damage to PSI takes much longer to repair. Our results indicate a PSI sepcific effect, where efficient oxidation of the PSI primary donor (P700) upon transition from darkness to light, depends on FNR recruitment to the thylakoid membrane tether proteins: thylakoid rhodanase-like protein (TROL) and translocon at the inner envelope of chloroplasts 62 (Tic62). When these interactions were disrupted, PSI photoinactivation occurred. In contrast, there was a moderate delay in the onset of PSII damage. Based on measurements of ΔpH formation and cyclic electron flow, we propose that FNR location influences the speed at which photosynthetic control is induced, resulting in specific impact on PSI damage. Membrane tethering of FNR therefore plays a role in alleviating high light stress, by regulating electron distribution during short-term responses to light.


Introduction
Electrons excited at photosystem I (PSI) reduce ferredoxin (Fd), and the majority of these are used by the enzyme translocon at the inner envelope of chloroplasts 62 (Tic62) in the dark (Juric et al., 2009;Benz et al., 2010;Yang et al., 2016), but this association is weakened at higher pH in vitro, and in the light (Alte et al., 2010). Association of FNR with PSI is also reported (Andersen et al., 1992;Marco et al., 2019), and based on immunogold labeling it was proposed that FNR released from the Tic62/TROL tethers in the presence of light could be localized at PSI (Kramer et al., 2021). Oxidized PSI is re-reduced with electrons from plastocyanin (Hippler and Drepper, 2006), which shuttles them through the thylakoid lumen from the cytochrome b 6 f complex (Cyt b 6 f). The Cyt b 6 f is reduced in turn by plastoquinol (PQH 2 ), which is the electron acceptor at photosystem II (PSII). This plastoquinone (PQ) reduction and oxidation drives formation of DpH, which is used by the thylakoid ATPase to synthesize ATP.
Photoexcitation of PSI can alternatively be used to form DpH by returning electrons from Fd to PQ through at least two cyclic electron flow (CEF) pathways (Munekage et al., 2004). One Fd:PQ reductase is a homolog of respiratory complex I, known as the NDH (Burrows et al., 1998;Shikanai et al., 1998;Yamamoto and Shikanai, 2013), while the mechanism of the other Fd:PQ reductase, which is inhibited by antimycin A (AA), remains enigmatic (Munekage et al., 2002;Fisher and Kramer, 2014;Nawrocki et al., 2019). Proposed mechanisms involve the function of the proton-gradient regulation 5 (Pgr5) and Pgr5 like 1 (PgrL1) proteins (Munekage et al., 2002;Hertle et al., 2013) and/or direct electron donation from Fd or even FNR to PQ at the Cyt b 6 f (Zhang et al., 2001;Nawrocki et al., 2019;Sarewicz et al., 2021). FNR has also been co-purified with the Cyt b 6 f (Clark et al., 1984;Zhang et al., 2001) and interaction is reported with PgrL1 (DalCorso et al., 2008). We have recently reported that FNR binding to the thylakoid through either the TROL or Tic62 membrane tethers is necessary for CEF when Arabidopsis (Arabidopsis thaliana) plants are transferred from the dark to the light (Kramer et al., 2021), a condition in which CEF is upregulated (Joliot and Joliot, 2006).
Production of damaging reactive oxygen species (ROS) increases in the absence of acceptors at either photosystem. If PQ availability at PSII is limiting, excitation is transferred from triplet-state chlorophyll to O 2 , producing dangerous singlet oxygen ( 1 O 2 ) (Telfer et al., 1994;Krieger-Liszkay, 2005). This occurs even at very low light, and the reaction center subunit of PSII (D1) is constantly turned over to ensure functional PSII (Fufezan et al., 2002). To minimize such damage, excitation at PSII can be dissipated through nonphotochemical quenching (NPQ) mechanisms induced by acidification of the thylakoid lumen (Ruban, 2016;Ruban and Wilson, 2020). Upregulation of this thermal dissipation in vivo requires the PsbS protein and de-epoxidation of the xanthophyll pigment violaxanthin into zeaxanthin, both of which are triggered by the formation of DpH (Ware et al., 2015;Sacharz et al., 2017;Steen et al., 2020;Saccon et al., 2020aSaccon et al., , 2020b. If oxidized Fd availability at PSI is limiting, excited electrons are passed from the iron-sulfur (FeS) clusters at the stromal side of PSI to oxygen, forming superoxide (O 2 •-) (Allen and Hall, 1974), which is rapidly dismutated enzymatically to hydrogen peroxide (H 2 O 2 ) (Asada, 1999). This H 2 O 2 can be converted to hydroxyl radicals ( • OH) through Fenton chemistry, a process catalyzed by the PSI FeS clusters in their reduced state, resulting in their rapid destruction by the resulting • OH (Sonoike et al., 1995;Sonoike, 2011). Degradation of the PSI reaction center subunits quickly follows (Ivanov et al., 1998;Tjus et al., 1998) and this requires costly reconstruction of the entire photosystem, which can take days (Kudoh and Sonoike, 2002;Lima-Melo et al., 2019a, 2019b. In angiosperms, this is prevented by restricting electron donation to PSI, maintaining the P700 active center and FeS clusters in an oxidized state (Shimakawa and Miyake, 2019;Storti et al., 2020). PSI donor limitation is promoted either through downregulation/damage to PSII (Ivanov et al., 1998;Tikkanen et al., 2014;Chaux et al., 2015), or through "photosynthetic control" in which PQH 2 oxidation at the Cyt b 6 f is limited by high DpH (Tikkanen et al., 2015;Colombo et al., 2016). It has been reported that the role of CEF in photoprotection is related to enhanced generation of DpH to induce protective mechanisms, rather than providing alternative PSI acceptors (Rantala et al., 2020).
Experiments on transgenic Nicotiana tabacum (tobacco), with FNR either over-expressed or knocked down by antisense, showed that FNR abundance corresponded with tolerance to high light and oxidative stress, and considerable damage at PSII was detected when FNR contents were low (Palatnik et al., 2003;Rodriguez et al., 2007). It was hypothesized that FNR could be affording protection through its role as a PSI acceptor, preventing build-up of electrons in the thylakoid and possibly acting as a diaphorase to remove radical species. We previously found that FNR could generate considerable O 2 •during turnover, and that stress remediation pathways were upregulated in Arabidopsis overexpressing FNR (Kozuleva et al., 2016). We interpreted this as indicating the enzyme might confer stress tolerance by "priming" the plant to induce increased abundance of proteins involved in protective mechanisms, such as superoxide dismutation, prior to stress exposure. FNR:membrane complex associations are also implicated in stress responses. It has been shown that FNR is released from the thylakoid during oxidative stress (Palatnik et al., 1997), and plants with different FNR:membrane associations also vary in stress responses (Lintala et al., 2009(Lintala et al., , 2012Kozuleva et al., 2016).
There has been much recent research on PSI photoinhibition, a considerable amount relating to how processes in the post-PSI electron transfer cascade can relieve acceptor limitation and prevent damage Miyake, 2018a, 2018b;Lima-Melo et al., 2019a, 2019bLeister, 2020;Storti et al., 2020). However, there are no reports of how this might relate to FNR abundance or location, despite the fact that FNR abundance strongly corresponds to stress tolerance (Palatnik et al., 1997(Palatnik et al., , 2003Rodriguez et al., 2007;Kozuleva et al., 2016). In the work presented here, we dissect the impact of changing FNR content and sub-chloroplast location on PSI and PSII. We find that when plants with perturbed FNR:membrane interactions are exposed to sudden high light, there is specific photoinactivation of PSI. Dark recovery after illumination shows that PSI damage correlates with this inactivation. Surprisingly, over the same timescale, PSII is slightly less photodamaged than in the wild-type (WT) as measured by fluorescence quenching in the dark (qPd). Protection of PSI correlates with CEF capacity, the rate of proton gradient (DpH) generation and is AA dependent, but PsbS-mediated NPQ has a minimal effect on PSI photoprotection. It is suggested that protection at PSI may contribute to the stress tolerance conferred by FNR.

Estimating PSI photoinactivation and PSII photoinhibition during illumination
In order to differentiate between impact at PSII and PSI during the illumination period, we first established high light treatment protocols. To estimate the onset of damage at PSII, we measured PSII capacity as Y(II) (Schreiber and Klughammer, 2009) and the value of photochemical qPd, which incorporates a comparison between actual and theoretical F 0 values (Ruban and Murchie, 2012;Ruban, 2017) to identify the early stages of PSII photodamage. We also defined a protocol to estimate PSI capacity during illumination. Although measurement of PSI capacity in plants following dark adaptation (Pm) is well established (Yamori et al., 2016;Miyake, 2018a, 2018b;Rantala et al., 2020), this is hard to accurately determine during illumination. We followed the protocol of Klughammer and Schreiber (2008), by briefly turning off the actinic light while applying a 10 s far-red (FR) light pulse prior to a saturating flash. Here, we call this transient P700 maximum (tPm) to help differentiate it from the Pm values calculated following prolonged dark adaptation/recovery.
To examine whether decreases in PSI capacity measured during illumination (tPm) translate into photodamage as calculated by Pm, we compared A. thaliana WT with a classic genetic model prone to PSI damage, the pgr5 mutant (Takagi and Miyake, 2018;Rantala et al., 2020;Storti et al., 2020). The tPm is relatively similar in pgr5 and the WT for the first 5 min of high light (Figure 1), as it drops from 1 to around 0.78. Thereafter, the WT stabilizes at around 0.75, while pgr5 continues by decreasing further to around 0.35. Some recovery is seen in both genotypes following 30 min in the dark (to 0.85 for the WT and 0.4 for pgr5), indicating a component of the decrease in tPm is not related to PSI damage. For this reason, throughout the text, we refer to decreases in tPm as PSI photoinactivation, as it contains a recoverable component, while decreases in Pm following dark relaxation are described as photodamage. A significant $0.4 difference between WT and pgr5 values is consistent between the final tPm and the dark recovered Pm. This confirms the predisposition of pgr5 to PSI photodamage, and indicates that differences between WT and pgr5 tPm values are predominantly due to PSI damage. Therefore, at least in these conditions, tPm can provide an estimate of real-time PSI photodamage, and this also confirms that damage to PSI (a decrease in PSI activity by $15% after dark recovery) can be achieved in WT Arabidopsis through the application of 15 min of high light.

FNR abundance and location are related to PSI inactivation
As a first step to investigating the relationship between FNR and photoinactivation at PSI and PSII, we analyzed the fnr1 knockout mutant of FNR1 in Arabidopsis, which has increased susceptibility to high light stress (Kozuleva et al., 2016). This was done over gradually increasing light intensity to allow induction of PSII photoprotective mechanisms Ruban, 2019, 2020). Measurements of tPm ( Figure 2A) showed greater photoinactivation of PSI in fnr1, both at low and high light intensities. The PSI quantum yield, Y(I), was decreased in fnr1 relative to the wt at lower light intensities ( Figure 2B). As expected for a mutant lacking $50% of FNR (a component involved in acceptor regeneration at PSI), the fnr1 genotype showed significantly greater acceptor limitation, Y(NA), than the WT ( Figure 2C). Surprisingly, the fnr1 genotype shows slightly less photodamage to PSII, as indicated by higher qPd and Y(II) values than WT plants (Figure 2, D and E), although only Y(II) showed statistical significance.
The fnr1 plants not only lack around 50% of the WT FNR, but also additionally show a change in the location of the pgr5 (open right-pointing triangles). tPm was measured upon illumination. Absolute Pm was determined after 30 min of dark adaptation. Data shown are mean ± SEM (n 5 3 individuals per genotype). Differences attributed by post hoc honest significant difference Tukey's test. P values are indicated as "***" 5 0.001, "**" 5 0.01, "*" 5 0.05, "." 5 0.1.
remaining enzyme (Lintala et al., 2007;Hanke et al., 2008), due to low interaction with Tic62 and TROL . To understand whether the phenotype of fnr1 is due to decreased FNR content or altered FNR interactions, we analyzed the tic62/trol double mutant, which lacks both FNR:membrane tether proteins. The tic62/trol mutant retains higher levels of FNR than the fnr1 mutant, although still less than the WT (Lintala et al., 2014). Relative to the WT, tic62/ trol and fnr1 show a similar decrease in tPm ( Figure 3A), which becomes significant at higher light intensities, although the PSI acceptor limitation is much lower in tic62/trol ( Figure  3C). Similar increases in qPd and Y(II) are also seen between the genotypes (Figure 3, D and E). To compare real-time photoinactivation at PSI and photodamage to PSII, we plotted qPd against tPm (Figures 2,F and 3,F). Although there appears to be a negative correlation between tPm and qPd, this is unlikely to be causative, because the parameters show poor correlation in the WT, fnr1 and tic62/trol genotypes (R 2 = 0.330, 0.451, and 0.512, respectively).
Both fnr1 and tic62/trol plants have disrupted FNR:tether protein interaction and decreased total FNR content. To try and distinguish the impact of FNR interactions from FNR abundance, we exploited plants in which the fnr1 mutant was complemented with genes for FNR proteins that specifically interact with TROL (ZmFNR1), Tic62 (ZmFNR2) and only show weak interaction with either tether (ZmFNR3) (Twachtmann et al., 2012;Kramer et al., 2021). Chloroplasts of these plants previously showed equivalent FNR immunostaining to the WT (Kramer et al., 2021). We aimed to use the fnr1:ZmFNR3 line to establish whether WT levels of enzyme, weakly bound to the Tic62/TROL tethers could prevent the decreased tPm seen in fnr1 and tic62/trol. Because antigenicity of iso-proteins may vary, we confirmed that NADPH dependent Fd reduction activity of leaf crude extracts was roughly equivalent between WT (0.26 mmol -1 mg protein -1 s ± 0.002 SD) and fnr1:ZmFNR3 (0.24 mmol -1 mg protein -1 s ± 0.014 SD). On increasing light intensities, the PSI parameters of fnr1:ZmFNR3 did not fully recover to WT  (I)). C, acceptor limitation at PSI (Y(NA)). D, photochemical quenching of PSII in the dark (qPd). E, Effective quantum yield of PSII (Y(II)). F, Relationship between tPm and qPd. Data shown are mean ± SEM (n = 5-7 individuals per genotype). Differences attributed by post hoc honest significant difference Tukey's test. P values are indicated as "***" 5 0.001, "**" 5 0.01, "*" 5 0.05, "." 5 0.1. levels, although this was not statistically significant, and developed significantly higher Y(II) values (Supplemental Figure  S1). This similar profile of fnr1:ZmFNR3 to tic62/trol and fnr1 indicates that membrane tether association might indeed be related to the relationship between FNR and stress tolerance. By contrast, Supplemental Figure S1 shows that fnr1:ZmFNR1 and fnr1:ZmFNR2, which interact strongly with TROL and Tic62, respectively, were almost identical to the WT in terms of tPm, Y(I), qPd, and Y(II).

PSI photoinactivation is partly independent of acceptor limitation
To further probe the relationship between stress tolerance and FNR:membrane tether interactions, we selected three specific genotypes for further experiments: WT; the tic62/ trol mutant, which lacks FNR:membrane tether interactions; and the fnr1:ZmFNR1 line, which contains WT levels of FNR and increased FNR:TROL interaction (Kramer et al., 2021).
Interaction of FNR with TROL is stronger than Tic62 (Alte et al., 2010), so fnr1:ZmFNR1 has exaggerated FNR:tether binding in comparison to the WT. These genotypes were then subjected to a sudden application of high light (as in Figure 1) to promote rapid PSI damage. Figure 4A shows that tPm values were significantly lower in the tic62/trol mutant over the high light treatment ( Figure 4A). This is despite the fact that Y(NA) values indicate acceptor limitation does not vary between WT and tic62/trol ( Figure 4C). In contrast to the pgr5 mutant ( Figure  1), PSI inactivation occurred rapidly during the initial part of the treatment. Importantly, the statistically significant $0.1 difference between WT and tic62/trol tPm values over illumination is also seen in the Pm values, measured after a 30min dark recovery. This indicates that the difference between WT and tic62/trol tPm values is due to increased photodamage at PSI. Although FNR content in the tic62/trol mutant is greater than the fnr1 mutant, it is still decreased in comparison to the WT (Lintala et al., 2014). As a further  (I)). C, Acceptor limitation at PSI (Y(NA)). D, qPd. E, Effective quantum yield at PSII (Y(II)). F, Relationship between tPm and qPd. Data shown are means ± SEM (n = 5-7 individuals per genotype). Differences attributed by post hoc honest significant difference Tukey. P values indicated by "***" 5 0.001, "**" 5 0.01, "*" 5 0.05, "." 5 0.1.
control to distinguish between the impact of decreased total FNR dose and decreased FNR:Tic62/TROL interactions, we also tested the fnr1 mutant line complemented with maize FNR3, which has a low affinity for TROL and Tic62. Interestingly, treatment of fnr1:ZmFNR3 with sudden high light resulted in significantly decreased tPm values relative to WT within 1 min ( Figure 4D). As in the case of the tic62/ trol mutant, there was no difference in PSI acceptor limitation as measured by Y(NA) for either fnr1:ZmFNR3 ( Figure  4F) or the fnr1 mutant (Supplemental Figure S2). This confirms that the impact of FNR:tether interactions on PSI photodamage is at least partly independent of acceptor limitation.

FNR abundance and location are important for CEF and DpH generation
Apart from preventing acceptor limitation at PSI, another mechanism by which FNR could prevent photoinactivation is through a role in CEF, to restrict electron supply to PSI.
We have recently shown that short-term induction of CEF on transfer from dark to light is dependent on FNR:tether interactions (Kramer et al., 2021). This data showed that the fnr1 mutant lost CEF capacity, and that this could be rescued by expression of ZmFNR1 (strong TROL binding) but not ZmFNR3 (weak tether interactions). We, therefore, compared relative CEF capacity of WT and the tic62/trol mutant following either dark adaptation, or light acclimation (Figure Timing of changes in light intensity is indicated above the graph: the gray bar indicates a step-wise increase to first 543, then 692 mmol photon m -2 s -1 (1 min each). The white bar indicates continuous illumination at 1,385 mmol photon m -2 s -1 , and the black bar indicates a dark relaxation period. A and D, tPm was measured over illumination. Absolute Pm was determined after 30 min of dark adaptation. B and E, Effective quantum yield of PSI (Y(I)). C and F, Acceptor limitation at PSI (Y(NA)). A-C, Data shown are means ± SEM (n 5 3 individuals per genotype). Significant differences were attributed by post-hoc honestly significant difference Tukey Test and P values 50.05 are indicated as the following letters: "a" for WT, fnr1:ZmFNR1 6 ¼ tic62/trol; "b" for WT 6 ¼ tic62/trol. D-F, Differences from WT attributed by Student's t test. P values indicated by "***"50.001, "**"50.01, "*"50.05, "."50.1. 5A). CEF is elevated in the WT after dark adaptation as previously reported (Kramer et al., 2021), but not in the tic62/ trol mutant. CEF has the capacity to accelerate generation of DpH and we therefore compared the rate at which DpH is generated at the onset of light. Initially, 312 mmol photon m -2 s -1 was used, and we confirmed that this light intensity was saturating by repeating the experiment at 586 mmol photon m -2 s -1 ( Figure 5B). Dark-adapted chloroplast preparations from WT, fnr1:Zm-FNR1, and tic62/trol, had equivalent F v /F m values of 0.74 ± 0.01, 0.72 ± 0.04, and 0.70 ± 0.03, respectively, indicating that chloroplasts were of equivalent quality. We found no significant difference between the amplitude of fluorescence quenching of the three genotypes. However, we observed a $30% increase in the half time to reach maximum fluorescence quenching in tic62/trol compared to the WT and fnr1:ZmFNR1 ( Figure 5C) at both irradiance intensities. As the same genotype shows an inability to upregulate CEF on dark adaptation, caused by disruption of FNR:tether interactions, it is highly likely that the decreased rate of DpH development upon transitions from darkness to light is due to a missing contribution of CEF to proton pumping. To further investigate whether an inability to upregulate CEF on transition to higher light results in the decreased tPm values measured in the tic62/trol and fnr1:ZmFNR3 plants, we examined the impact of the classical CEF inhibitor AA on tPm. Figure 5D shows that infiltration of leaves with 5 mM AA is sufficient to eliminate the difference in tPm between WT and fnr1:ZmFNR3, while the difference between WT and tic62/trol ceases to be significant. Infiltration with methyl viologen (MV), which should compete with FNR for reduced electrons, also eliminates this difference but only on a longer timescale.
Faster development of DpH might accelerate PSI oxidation, and therefore protection, by inducing PsbS-dependent NPQ to limit the flux of electrons from PSII (Tikkanen et al., 2014;Chaux et al., 2015). We, therefore, compared how NPQ induction over increasing light intensities influenced inactivation of PSI (tPm). Figure 6 shows that there is no significant difference in NPQ values between the genotypes, and that tPm does not correlate with NPQ development over increasing light intensity. To further evaluate whether PsbS-dependent NPQ might play a role in protecting PSI from photoinactivation in Arabidopsis, we also measured genotypes with enhanced NPQ (L17) (Li et al., 2002), or compromised NPQ (npq4) (Havaux and Niyogi, 1999; Figure  6), confirming that, under these conditions at least, there is no relationship between NPQ and tPm. We further treated these genotypes with the sudden high light protocol (Supplemental Figure S3, C-E) which resulted in similar tPm and final Pm values for WT, L17, and npq4. This result suggests that NPQ does not limit PSI re-reduction in either condition measured here.
Any mechanistic role of FNR in CEF remains unclear. To compare the impact of FNR:Tic62/TROL interaction with the classical pathways of CEF, we used the same protocol to check whether a similar response could be measured in pgr5 (a mutant of the Pgr5/PgrL1 pathway, as shown in Figure 1) and crr2 (an assembly factor mutant of the NDH complex). In Figure 7, neither of these mutants show a similar pattern of PSI inactivation to tic62/trol, fnr1, or fnr1:ZmFNR3 plants. On sudden high light illumination of pgr5, tPm is initially higher than in WT and then gradually decreases over the time course, to much lower values than other genotypes ( Figure 7A), while crr2 also shows higher initial tPm values than WT, but does not decrease below WT values. Y(NA) values indicate increased acceptor limitation relative to wt for both pgr5 and crr2 ( Figure 7C). Interestingly, when these genotypes are subject to an increasing light protocol (Supplemental Figure S4) crr2 plants present higher tPm values than the WT up to light intensities of 1,096 mmol photons m -2 s -1 .

FNR impact on PSI photoinactivation upon fluctuating light
It has been frequently reported that fluctuating light exacerbates the damage to PSI caused by perturbation of CEF (Suorsa et al., 2012(Suorsa et al., , 2013Kono et al., 2014;Huang et al., 2018;Nikkanen et al., 2018), and we therefore repeated our experiment using a fluctuating light environment. This comprised three phases of high light (1385 lmol photon m -2 s -1 ) intensity illumination alternating with total darkness ( Figure  8). Following the first light treatment, tPm values were lower for the tic62/trol mutant throughout the illumination protocol ( Figure 8A). When measured during illumination periods, genotype had no effect on PSI yield and acceptor limitation ( Figure 8B, Figure 8C). There were significantly lower Y(I) and higher Y(NA) values for tic62/trol following the dark steps. The final Pm values were lower after this fluctuating light routine ( Figure 8A) relative to those obtained after continuous high light ( Figure 4A) in all the genotypes. Interestingly, the dark recovery of Pm only decreased by $9% in fnr1:Zm-FNR1 which, unlike the WT, retained a significant difference from tic62/trol.

Discussion
We hypothesized that, as FNR is a critical component of the post-PSI electron transfer cascade, its abundance, and/or location on the membrane (close to PSI), might contribute to PSI protection. This could be analogous to the impact of the flavodiiron proteins Alboresi et al., 2019), or photorespiration (Allahverdiyeva et al., 2011;Hanawa et al., 2017;Takagi et al., 2017), providing excess sinks to maintain P700 in a protected, oxidized state. Because differences in electron transport induced by altered FNR content or location are most pronounced in the early stages of illumination following dark adaptation (Twachtmann et al., 2012;Kramer et al., 2021), it was critical to estimate PSI inactivation during the early stages of illumination, for which we used the tPm parameter (Figure 1). This revealed that, over the relatively short high light stress applied here, FNR content and membrane association specifically  Figure 5 The impact of :FNR:tether interactions on CEF and generation of DpH. Genotypes are: WT (black), fnr1:ZmFNR1 (red) and tic62/trol (cyan). A, Comparison of the maximum rate of leaf CEF as a function of total electron flow. Values calculated from ECS measurements following a 20 s high light pulse (actinic light for total electron flow, FR light for CEF). Means of measurements on three to five individuals per genotype ± SEM. Light acclimation was for 5 min at 150 mmol photons m -2 s -1 . B, DpH generation by isolated chloroplasts, measured by 9-AA fluorescence quenching. Onset of actinic irradiance is indicated by upward arrow. Irradiance at 312 (upper graph) and 586 mmol photon m -2 s -1 (lower graph). Traces are averages of three independent experiments. C, Rate of DpH generation, calculated as t 1/2 on illumination at the indicated light intensities.
Values are means ± SEM of experiments on three independent chloroplast preparations for each genotype. Differences attributed by post-hoc honestly significant difference Tukey's test for half times for rate of DpH generation and Student t test for CEF analysis. P values are indicated as "***"50.001, "**"50.01, "*"50.05, "."50.1. D, Impact of inhibitors on tPm determination following sudden illumination at 1,095 mmol photon m -2 s -1 . Individual leaves were dark-adapted for 40 min and infiltrated with buffer, 5 mM AA or 250 mM MV, prior to illumination and measurement of tPm at the indicated times. Data are means ± SEM (n 5 3 individuals per genotype). Differences from WT values attributed by Student's t test. P value indicated by "***"50.001, "**"50.01, "*"50.05. impact on photoinactivation of PSI. Interestingly, the increased PSI photoinactivation in fnr1, fnr1:ZmFNR3, and tic62/trol mutants occurs at the same time as a small, but significant improvement in PSII protection.
PSI acceptor limitation is associated with O 2 •generation (Takahashi and Asada, 1988;Kozuleva et al., 2014), and FNR can also generate O 2 •during catalysis (Kozuleva et al., 2016). Perhaps surprisingly, detection of light-dependent free radical evolution in antisense tobacco leaves indicated that decreased FNR content resulted in no difference in O 2 •-, but rather an increase in 1 O 2 (Palatnik et al., 2003), a radical normally associated with PSII damage. In addition, spin-trapping of O 2 •produced by chloroplasts from Arabidopsis single trol knockout plants was previously shown to be lower than in WT in both light and dark (Vojta et al., 2015). The results in tobacco prompted the suggestion that PSII was acceptor limited, due to back up of electrons through the whole electron transport chain (Palatnik et al., 2003). This might not be contradictory to our data, as the light stress applied to tobacco FNR antisense plants prior to measurement was for much longer time scales than used in our study. Our data (Figures 2-4 and 8; Supplemental Figures S1 and S2) indicate that the FNR associated PSI photoinactivation occurs early in illumination. In the absence of correct FNR:tether interaction, poor initial regulation of events around PSI on illumination, or an increase in light intensity, may therefore cause a temporary increase in O 2 •-(and therefore • OH) production relative to the WT. PSI damage would then progress from the destruction of FeS clusters to protein subunits at a small percentage of PSI complexes, resulting in 1 O 2 production from triplet state chlorophyll at damaged PSI (Sonoike et al., 1995;Tjus et al., 1998). PSII is highly sensitive to 1 O 2 (Krieger-Liszkay, 2005) and so a small amount of PSI damage could result in considerable PSII damage (Sonoike, 2011). The small, short-term improvement in PSII tolerance to high light stress seen in plants with disrupted FNR:tether interactions might have a number of causes. These include secondary effects to decrease light harvesting or increase PQ acceptor availability at PSII in response to the direct impact at PSI, and are beyond the scope of this study.
We expected that the main impact of FNR abundance and/or location on PSI damage would relate to acceptor limitation. However, our results indicate that the effect is more complex. Although acceptor limitation (as measured by Y(NA)) is elevated on initial increases in light intensities for the fnr1 mutant (Figure 2), this is not the case for the tic62/trol mutant (Figure 3, Supplemental Figure S1) neither for tic62/trol or fnr1:ZmFNR3 on sudden high light (Figure 4). Moreover, infiltration of leaves prior to measurement with MV, which eliminates acceptor limitation at PSI, does not immediately diminish the difference in tPm between these genotypes and the WT (Figure 5), suggesting that acceptor limitation is not the predominant cause of their increased PSI photoinactivation. The inhibition by MV later in the time course presumably arises from superoxide production, and it is interesting that it does not immediately impact on whatever protective function FNR is performing in sudden high light. This may be related to reports that MV is actually a relatively weak inhibitor of electron transfer by Fd between PSI and FNR (Setif, 2015), meaning that under the conditions used here sufficient reduced Fd is probably still available to FNR.
The increasing light experiments (Figures 2 and 3) were designed to distinguish damage at PSII from PSI in conditions where NPQ was able to develop, while the sudden light increase (Figure 4) tests PSI activity when electron flux from PSII is maximum. The stronger binding of FNR to the Tic62/TROL tether proteins in WT and fnr1:ZmFNR1 lines at the onset of the high light period is therefore a more likely explanation for their higher tPm values than in fnr1, tic62/ trol or fnr1:ZmFNR3, rather than differences in FNR abundance. The data indicate that, without NPQ, PSI acceptors are overwhelmed irrespective of any differences in FNR content seen between our genotypes. We included the fnr1:ZmFNR1 plants in our study to examine whether tighter binding to membrane tethers might enhance PSI protection. In most experiments, this was not the case, but intriguingly Timing of changes in light intensity is indicated above the graph: the gray bar indicates a step-wise increase to first 543, then 692 mmol photon m -2 s -1 (1 min each), the white bar indicates continuous illumination at 1,385 mmol photon m -2 s -1 , and the black bar indicates a dark relaxation period. A, tPm was measured over illumination and absolute Pm was determined after 30 min of dark adaptation. B, Effective quantum yield of PSI (Y(I)). C, acceptor limitation at PSI (Y(NA)). Data shown are means ± SEM (n 5 3 individuals per genotype). Significant differences were attributed by post-hoc honestly significant difference Tukey's test and P values under 0.05 are indicated by are indicated as the following letters: "a" for pgr5 and crr2 6 ¼ WT; "b" for pgr5 and WT 6 ¼ crr2; "c" for pgr5 6 ¼ crr2; "d" for pgr5 6 ¼ WT and crr2; "e" for pgr5 6 ¼ WT.  (I)). C, acceptor limitation at PSI (Y(NA)). Data shown are means ± SEM (n 5 3 individuals per genotype). Significant differences were attributed by post-hoc honestly significant difference Tukey's test and are indicated by the following letters: "a" for fnr1:Zm-FNR1 and WT 6 ¼ tic62/trol; "b" for fnr1:Zm-FNR1 6 ¼ tic62/trol. fnr1:ZmFNR1 did show slightly higher Pm values than the WT following dark recovery from the fluctuating light protocol ( Figure 8).
As we previously reported an impact of FNR location on CEF during the dark to light transition for the fnr1 and fnr1:ZmFNR3 lines (in which FNR:membrane tether interactions are disrupted) (Kramer et al., 2021), we confirmed that this phenotype was also present in tic62/trol mutants ( Figure  5A). The absence of this CEF might be expected to result in slower generation of DpH, which is indeed the case for this genotype ( Figure 5B). We could not detect a significant difference in steady-state DpH, which might be expected as it is subject to extensive regulation (Tikkanen et al., 2015;Colombo et al., 2016;Yang et al., 2018). Many chloroplast protective mechanisms are dependent on DpH, including NPQ (Johnson and Ruban, 2011) and photosynthetic control . We found no evidence for a relationship between NPQ and tPm or Y(I) in our genotypes ( Figure  7; Supplemental Figure S3) and plants with enhanced or attenuated NPQ also showed minimum impact on PSI photoinactivation. This confirms previous reports that PSII downregulation has little effect on PSI protection (Tikkanen et al., 2015;Miyake, 2018a, 2018b).
An inability to regulate electron donor supply by photosynthetic control has been proposed to cause the extreme PSI damage reported in pgr5 mutants Miyake, 2018a, 2018b;Lima-Melo et al., 2019a, 2019b and confirmed here (Figures 1 and 7). In this work, we did not compare the dark-adapted and light acclimated CEF capacity of pgr5 and crr2 plants, but the same technique has previously been used to show that, while NDH mutants are relatively unaffected, dark-adapted Arabidopsis pgr5 mutants have decreased total CEF capacity, as measured following 120 s illumination . As the tic62/trol plants show lower rates of initial DpH generation ( Figure 5, B and C), it seems likely that a slower induction of photosynthetic control might also be responsible for the PSI inactivation associated with decreased FNR content and membrane tether interaction.
Interestingly, during the first 5 min of high light stress, the tPm values for pgr5 plants are higher than or equivalent to WT (Figure 1), and only decrease drastically thereafter. Those of tic62/trol and fnr1:ZmFNR3 (Figure 4), and fnr1 (Supplemental Figure S2) show the opposite trend, with an immediate drop relative to WT, after which the relative difference between the genotypes changes little. This is consistent with disruption of a process in the initial stages of illumination, such as the slower induction of DpH seen in Figure 5, B and C. FNR has been found in complex with PGR5 in algae (Iwai et al., 2010) and angiosperm PGRL1 was also reported to interact with FNR (DalCorso et al., 2008). AA is the canonical inhibitor of FQR-dependent CEF, and evidence supports a role for PGR5 in its mechanism of action (Sugimoto et al., 2013), while AA also abolishes FNR-dependent differences in tPm ( Figure 5D). Although our data do not rule out these proteins acting in concert, they suggest that the mechanism disrupting CEF induction during the dark to light transition is different between the pgr5 mutant and the fnr1, tic62/trol, or fnr1:ZmFNR3 genotypes.
One possible explanation for our data is therefore that, in addition to acceptor limitation, the stress phenotype associated with decreased FNR content and tether association could be due to increased donor arrival through slower induction of photosynthetic control. This would explain the specific, short-term inactivation of PSI seen in the fnr1 and tic62/trol genotypes on exposure to high light.

Plant material and growth conditions
Unless otherwise indicated, plants were grown in 12 h light at 21 C, 12 h dark at 18 C, and in 0.12 L pots on a 6:6:1 ratio of John Innes No. 3 soil, Levington M3 potting compost and perlite (Scotts UK, Ipswich, UK). Arabidopsis (A. thaliana) genotypes ecotype Columbia (WT) and fnr1 were as described in (Hanke et al., 2008), Arabidopsis plants expressing Zea mays FNR1, FNR2, and FNR3 in the fnr1 background were as described previously (Kramer et al., 2021) and tic62/ trol (Lintala et al., 2014) was a kind gift of Paula Mulo. pgr5 (Munekage et al., 2002)) and crr2 (Hashimoto et al., 2003) were a kind gift from Toshiharu Shikanai. The npq4 (Havaux and Niyogi, 1999) and L17 (Li et al., 2002) were also used in this work. Unless stated otherwise all measurements were carried out on 6-week-old plants that showed no signs of inflorescence or visible anthocyanin accumulation. Plants were dark-adapted for 45 min before each measurement and F v /F m was on average 0.78 ± 0.002 with no significant difference between genotypes.

Chlorophyll fluorescence and P700 absorption measurements and parameters
Chlorophyll a fluorescence and P700 oxidation measurements were performed simultaneously using a DUAL-KLAS-NIR-PAM (Walz, Effeltrich, Germany). In this method, the P700 redox state is monitored by deconvoluting the four absorption signals by differential model plots (Klughammer and Schreiber, 2008).
An assessment of the true quantity of photoinactivated PSII, as photochemical qPd, was calculated as described by Ruban and Murchie (2012).
where Fm 0 is the maximum fluorescence yield after exposure to actinic light, Fo 0 act. and Fo 0 calc. are the actual and theoretical minimum fluorescence yields measured in the dark after actinic light treatment, respectively. Fo 0 calc. is determined according to the equation of Oxborough and Baker (1997), as defined below. Fo 0 calc : ¼ 1=ð1=Fo À 1=Fm þ 1=Fm 0 Þ At lower actinic light levels, Fo 0 act. %Fo 0 calc. and qPd = 1. However, as the intensity of actinic light rises, Fo 0 act. also increases, which is an indicator of photoinactivation. The divergence between Fo 0 act. and Fo 0 calc. is due to permanent closure of RCs and causes qPd51.
The effective 'quantum yield' of PSII (Y(II) is undermined by two processes induced by excessive light: NPQ and photoinactivation (qPd). The effect of each upon Y(II) was therefore calculated as YðIIÞ ¼ ðqPd Á Fv=FmÞ=½1 þ ð1 À Fv=FmÞ Á NPQ Fv/Fm represents the maximum Y(II), which is calculated as (Fm-Fo)/Fm, with Fo and Fm being the minimum and maximum fluorescence yields, respectively. The theoretical Y(II) for a given NPQ is calculated with maximal values of Fv/Fm and qPd as constants. Damage to PSI is usually quantified as a change in the maximum photo-oxidisible fraction of PSI (Pm), as measured following dark adaptation. According to Klughammer and Schreiber (2008), Pm can be determined during illumination by (1) measuring the maximal signal level induced by a saturating pulse (oxidisable P700), and (2) measuring the signal level in the dark several hundred milliseconds after the saturating pulse (baseline level; P700 fully reduced). In this work, we have combined these approaches to estimate activity of PSI during illumination. Actinic light is turned off, and far-red light is applied for 10 s before measurement of Pm with a saturating pulse. To avoid confusion, we refer to measurements following dark adaptation/recovery as Pm (photodamage), while values taken during illumination, from which PSI capacity partially recovers (Figure 1), are termed "transient Pm" or tPm, and reflect steady-state inactivation. Both Pm and tPm measurements were normalized to an initial Pm measured following 45-min dark adaptation prior to illumination. Y(I) and Y(NA) were calculated from the P700 traces according to published protocols (Klughammer and Schreiber, 2008). Typical WT chlorophyll fluorescence and P700 traces are shown in Supplemental Figure S5. For vacuum infiltration experiments, leaves of dark-adapted Arabidopsis plants were individually vacuum infiltrated by placing a whole leaf inside a syringe barrel containing 20 mL of buffer (330 mM sorbitol, 20 mM HEPES, pH 7) with or without AA (5 mM) or MV (250 mM). The narrow syringe opening was blocked and the plunger was used to gently draw air out of the tissue until the leaf lost buoyancy, at which point it was immediately measured.

Chloroplast extraction and proton gradient determination
The protocol was performed basically as described previously (Saccon et al., 2020a(Saccon et al., , 2020b. Briefly, Protoplasts were obtained from the mesophyll layer of dark-adapted leaves from Arabidopsis adult plants. To expose the chloroplastenriched leaf layer to the enzyme solution, the lower epidermis was removed with adhesive tape. Leaves were then floated for 1 h on a solution containing 0.4 M Mannitol, 20 mM KCl, 20 mM MES, 10 mM CaCl 2 , 0.1% bovine serum albumin (BSA) (pH 5.5) in the presence of 1.5% cellulase Onozuka TM R-10 and 0.4% Macerozyme R-10 (Yakult, from Serva, Heidelberg, Germany). The solution was then filtered through a layer of muslin cloth and centrifuged twice (3 min, 100 rcf, 4 C). The obtained protoplasts were resuspended in reaction buffer containing 0.5 sorbitol, 20 mM HEPES, 20 mM MES, 20 mM Na-citrate, 10 mM NaHCO 3 , 15 mM MgCl 2 , 0.1% BSA (pH 8). Their intactness was checked with a bright-field optical microscope and Fv/Fm average values were confirmed.
DpH was determined from the measurement of 9-aminoacridine (9-AA) fluorescence using the Dual-ENADPH and Dual-DNADPH modules for the Dual-PAM-100 fluorimeter (Walz, Effeltrich, Germany). Intact chloroplasts (35 lM chlorophyll) were suspended in reaction buffer (as above) in the presence of 5 lM 9-AA. The chloroplasts were treated using 312 or 586 lmolphotonm -2 s -1 actinic light for 5 min, followed by 5 min of darkness. The 9-AA quenching traces were normalized to (fluorescence before illumination -fluorescence after illumination)/fluorescence before illumination. Half time of DpH induction (t 1/2 ) was taken as the time for the fluorescence level to reach half of the maximum value of quenched fluorescence upon illumination.

CEF measurements
Measurements were performed basically as described previously (Kramer et al., 2021). In brief, plants were dark incubated for 30 min before transfer into actinic light at 150 lmolphotonm -2 s -1 . After 20 s and 5 min of illumination, LEF and CEF were measured by following the relaxation kinetics of the carotenoid electrochromic bandshift at 520 nm (corrected with the bandshift at 546 nm). Saturating FR light (k 4 720 nm), was used to fully excite PSI with minimum excitation of PSII and calculate the CEF only signal. Electron flow was estimated from the amplitude of the electrochromic shift (ECS) signal upon excitation with a saturating single turnover flash (5 ns laser pulse). Total electron flow was measured with an actinic flash of 1,100 lmol photon m -2 s -1 while CEF was measured with only FR light at the maximum setting (estimated as 1,400 lmol photon m -2 s -1 by the manufacturer). LEF was calculated by subtraction of CEF from total electron flow.

FNR activity
Crude leaf extracts were prepared from 6-week-old plants by grinding with sand at 4 C in 50mM Tris-HCl pH 7.5, 100 mM NaCl, 2 mM MgCl 2 , 1 mM pefabloc, 0.1% w/v polyvinylpolypyrrolidone. The supernatant from a 2 min, 2,000 g centrifugation at 4 C was taken. FNR in this sample was solubilized by the addition of Triton X-100 to 1% (v/v), before further centrifugation at 15,000 g at 4 C for 5 min. FNR activity of this supernatant, equivalent to 30 mg total protein, was measured in the dark using a cytochrome c reduction assay in the presence of 20 mM Arabidopsis Fd 2 and an NADPH regeneration system as described previously (Hanke et al., 2005).

Statistics
The t test Student's, analysis of variance, and post-hoc analysis Tukey's tests were also performed in R version 3.5.3 (R_Core_Team, 2019).

Supplemental data
The following materials are available in the online version of this article.
Supplemental Figure S1. The impact of FNR:tether interaction on PSI and PSII photoinactivation during increasing light intensity.
Supplemental Figure S2. Additional experiments on the impact of FNR:tether associations on PSI and PSII photoinactivation during high light exposure in Arabidopsis.
Supplemental Figure S3. Impact of altered NPQ on PSI inactivation over increasing light (A and B) and during sudden high light stress (C-E) in Arabidopsis.
Supplemental Figure S4. Impact of two different CEF pathways on photoinactivation of PSI and PSII during a gradual increase of light intensity.
Supplemental Figure S5. Example traces of the initial section of a typical PSII chlorophyll fluorescence (red trace) and P700 absorption (blue trace) simultaneous measurement.