A security network in PSI photoprotection: regulation of photosynthetic control, NPQ and O2 photoreduction by cyclic electron flow

Cyclic electron flow (CEF) around PSI regulates acceptor-side limitations and has multiple functions in the green alga, Chlamydomonas reinhardtii. Here we draw on recent and historic literature and concentrate on its role in Photosystem I (PSI) photoprotection, outlining causes and consequences of damage to PSI and CEF’s role as an avoidance mechanism. We outline two functions of CEF in PSI photoprotection that are both linked to luminal acidification: firstly, its action on Photosystem II with non-photochemical quenching and photosynthetic control and secondly, its action in poising the stroma to overcome acceptor-side limitation by rebalancing NADPH and ATP ratios for carbon fixation.

In the early years of photosynthesis research, a cyclic photophosphorylation was described that required ferredoxin (Fd), did not evolve oxygen (O 2 ) and resulted in the accumulation of ATP (Arnon et al., 1954). From this observation, experiments performed in a variety of organisms from cyanobacteria to higher plants using a combined pharmacological and in vitro approach created a robust model for what is now referred to as cyclic electron flow (CEF; thoroughly reviewed, in Bendall and Manasse, 1995). More recently, Arabidopsis thaliana lines altered in CEF have been identified and have enriched the ways we have to study these pathways (Joet et al., 2001;Munekage et al., 2002Munekage et al., , 2004DalCorso et al., 2008). Biochemical approaches have shown that the Proton-Gradient Regulator5 (PGR5) and PGR5-Like1 (PGRL1) proteins form an interaction that results in a ferredoxin-plastoquinone reductase (FQR) activity (Hertle et al., 2013). In the unicellular, green alga, Chlamydomonas reinhardtii, this pathway and the function of these proteins is conserved (Petroutsos et al., 2009;Tolleter et al., 2011;Johnson et al., 2014). In Chlamydomonas a second type of CEF is also in operation where the mediator at the level of the PQ pool is a type-2 NADPH dehydrogenase (Desplats et al., 2009), with the nda2 mutant shown to have a phenotype in CEF (Jans et al., 2008). Here, we focus on the PGR5 pathway and work done on the Chlamydomonas mutants pgr5 and pgrl1 mutants, that both demonstrate no PGR5/PGRL1dependent CEF (Alric, 2014). Our focus is on Chlamydomonas but due to the conservation of this pathway we also make reference to work done in other photosynthetic organisms.
Cyclic electron flow is a generator of proton motive force that (i) can produce supplementary ATP to meet ATP:NADPH requirements for the Calvin Benson Bassham (CBB) cycle and the CO 2 concentrating mechanism (CCM; reviewed by Alric, 2010), and (ii) triggers regulatory mechanisms, namely nonphotochemical quenching (NPQ) and cytochrome b 6 f complex (cytb 6 f ) "photosynthetic control" (Joliot and Johnson, 2011). Its rate is highest under conditions where the stromal poise is reduced, thus PGR5-CEF has been considered as a regulator of redox homeostasis for the photosynthetic chain (Nishikawa et al., 2012). Among the phenotypes observed in CEF-altered strains of both Arabidopsis and Chlamydomonas, Photosystem I (PSI) photoinhibition arose in conditions of high light or limiting CO 2 (Munekage et al., 2002;Dang et al., 2014;Johnson et al., 2014) and fluctuating light (Suorsa et al., 2012) leading to the assignment of yet another role for PGR5-CEF. While Photosystem II (PSII) photoinhibition is frequently observed and has complex models that describe the mechanism (Murata et al., 2012), PSI photoinhibition remains poorly understood. In this work, we review the potential causes of photoinhibition that occur at the acceptor-side of PSI and the processes triggered by CEF that can contain it. For the sake of comprehensive reviewing of mechanisms involved in PSI photoprotection, other connected pathways are also introduced.

ACCEPTOR-SIDE LIMITATION IS THE CAUSE OF PSI PHOTOINHIBITION
PSI photoinhibition was first reported in isolated chloroplasts submitted to strong light (Jones and Kok, 1966). Satoh was able to differentiate two types of damage that corresponded to damage to the two photosystems using fragmented chloroplasts. Artificial donors were used to measure the capacity of PSI to transfer electrons to terminal acceptor, NADPH. These experiments showed that the addition of the PSII inhibitor, DCMU specifically but incompletely prevented photo-inactivation of PSI (Satoh, 1970a,b). Photo-inactivation of PSI was avoided by addition of excess Fd showing that PSI photoinhibition is an acceptor-side limited phenomenon. The observation that the same group of cofactors that could enhance CEF (including Fd) were also involved in the avoidance of photoinactivation of PSI led to the discussion of CEF as a photoprotectant for PSI (Satoh, 1970c). Further studies demonstrated the destruction of PSI-bound iron-sulfur centers (F X , F A/B ) by oxidative species primarily superoxide anion radical (O •− 2 ) (Sonoike et al., 1995). Production of O •− 2 can occur within: the iron-sulfur centers of PSI, reduced Fd and stromal flavodehydrogenases (NADP+ ferredoxin dehydrogenase, glutathione reductase and monohydrate ascorbate reductase) in plant chloroplasts (discussed, in Asada, 2000). In permissive conditions, radicals are enzymatically neutralized into water, resulting in the net uptake of O 2 reported by Mehler (1951), establishing a pseudo-cyclic pathway for electrons known as the water-water cycle. When radical production exceeds detoxifying capacity, O •− 2 irreversibly damages PSI primary acceptors (F X , F A/B ) and prevents stable accumulation of P700 + in high light (Figure 1). This is because of fast charge recombination at the level of intermediary acceptors A 0/1 (Setif and Brettel, 1990). The resultant decrease in the quantity of oxidizable P700 is thus a common measurement for probing the photoinhibition of PSI.
Interestingly, no singlet oxygen ( 1 O 2 ) is produced in overexcited PSI (triplet excited state, 3 P700) because P700 is sterically screened from O 2 (Setif et al., 1981). Hence, P700 + and 3 P700 are most probably efficient quenchers of excess excitation of plant PSI as observed for cyanobacterial PSI (Schlodder et al., 2005;Shubin et al., 2008). Contrarily, 1 O 2 is the main photodamaging species produced in acceptor-side limited PSII ( 3 P680) (Durrant et al., 1990). It is remarkable that O 2 can be sensitized in PSII and not in PSI (Hideg and Vass, 1995), in other words that 1 O 2 production within PSII was not evolutionarily eliminated. Over time this may be why a signaling role has developed for 1 O 2 (Telfer, 2014) resulting in some selectivity in the degradation of PSII protein under photoinhibitory conditions. As compared to the "monolithic" architecture of PSI, the modular architecture of PSII allows for a unique degradation of damaged D1 and re-use of other subunits (extensively reviewed, in Caffarri et al., 2014) and may be another reason why PSII damage-and-repair cycle has been a target of selective pressure. On the contrary, PSI has no known molecular mechanism per se to set its turnover in tune with light intensity. The protection of PSI from photoinhibition would appear to require a set of distinctly different properties than that of PSII (Allahverdiyeva et al., 2015) which includes buffering acceptor side limitations in the stroma. Selective, irreversible photoinhibition of PSI in Chlamydomonas is observed to occur both in CEF-altered strains (Dang et al., 2014;Johnson et al., 2014;Kukuczka et al., 2014;Bergner et al., 2015) and in strains with severe acceptor side limitations such as those lacking RuBisCO (Johnson et al., 2010). Crpgr5 and Crpgrl1 strains demonstrate decreased amounts of oxidizable P700 and PSI protein measured by western hybridization after exposition to high light (Johnson et al., 2014;Kukuczka et al., 2014) and after transition from high (2%) to atmospheric concentrations of CO 2 (Dang et al., 2014). In the following sections we present CEF's role in triggering several mechanisms avoiding long-lasting limitations at the acceptor-side of PSI.

CEF TRIGGERS FAST QUENCHING, PHOTOSYNTHETIC CONTROL AND PSII PHOTOINHIBITION RESULTING IN PSI PHOTOPROTECTION
As already suggested (Sonoike, 2011), non-photochemical quenching (NPQ) of PSII avoids excessive electron flow to PSI via linear electron flow (LEF) to prevent photoinhibition. CEF limits electrons entering the thylakoid chain because it prompts both excitation-dependent quenching (qE) and indirectly PSII photoinhibition (qI), thus avoiding overflow to PSI. Acidification of the lumen triggers qE (Briantais et al., 1979) and occurs during CEF due to coupling of electron transfer and proton translocation in the cytb 6 f. Since both LEF and CEF pass through cytb 6 f, the exact contribution of CEF to the formation of a qE is hard to determine but an altered ability to develop qE is observed in Crpgr5 and Crpgrl1 strains (Tolleter et al., 2011;Dang et al., 2014;Johnson et al., 2014;Kukuczka et al., 2014) concomitantly with PSI photoinhibition (Dang et al., 2014;Johnson et al., 2014). This FIGURE 1 | Acceptor-side limitation and excess electron flow promotes CEF or in its absence leads to the irreversible damage of PSI centers. The linear electron flow coming from PSII (gray dashed arrow) is the source of electrons for the PSI reaction center (P700 + /P700) that transfers electrons from the chlorophyll excited state (P700 * ) and subsequently delivers to downstream acceptors within PSI (FX, FA, FB iron-sulfur centers) then to stromal electron carriers (ferredoxin, FNR, NADP + ) (light gray arrow). When CO2 fixation decreases, acceptor-side limitation gradually leads to accumulation of NADPH and overreduction of stromal and PSI electron carriers (light red arrow). In this case electrons are redirected to O2 either at the level of NADPH without the production of reactive oxygen species (ROS; gray dashed arrow) or produce the very reactive superoxide anion radical (O •− 2 ) at the level of FX, FA, FB, Fd, and FNR with a rate exceeding the detoxification process. Thus O •− 2 will irreversibly destroys the centers (red arrows) resulting in an inability to oxidize *P700 and on a longer time scale the degradation of the entire PSI complex. Preventing this scenario, cyclic electron flow triggers downregulation of linear electron flow at the site of PSII and cytb6f by enhancing proton accumulation in the lumen.
is also consistent both with the failure to acidify the lumen under short saturating illumination in Atpgr5 plants (Suorsa et al., 2012) and reduced growth of Crpgrl1 strains in fluctuating light (Dang et al., 2014). A recent report challenging the effects of rapid quenching of PSII in PSI photoprotection showed that an absence of qE (in Atnpq4 mutants lacking the PsbS protein that induces qE in higher plants, Li et al., 2000) does not have a dramatic effect on P700 oxidation kinetics at any light regime as opposed to Atpgr5 mutants where steady state oxidation of P700 is abolished . Partial compensatory mechanisms may, however, act between CEF and qE as double mutant strain Crpgrl1npq4, (lacking both CEF and the LHCSR3 protein that acts as the activator for qE in Chlamydomonas, Peers et al., 2009), are particularly susceptible to PSI photoinhibition in comparison to the simple Crpgrl1 mutant (Kukuczka et al., 2014;Bergner et al., 2015). This may coincide with a PSI-photoprotective role recently proposed for LHCSR3 via its association with the PSI antenna system under "state 2 conditions" (Allorent et al., 2013;Bergner et al., 2015) but here LHCSR3-dependent quenching of LHCIs and/or PSI-bound LHCIIs has not been strictly established. The argument against would be that the quenching of PSI antenna is irrelevant given the harmlessness of *P700 and furthermore photo-oxidation events at PSI (measured on isolated complexes in vitro) have been shown to take place after photoinhibition is completed, chlorophyll oxidation being preceded by irreversible carotenoid oxidation (Santabarbara, 2006). For now, the literature would suggest that qE is a first level of photoprotection in photoinhibitory conditions and rapidly protects not only PSII but also PSI by reducing electron flow (Figure 2).
Aside from ATP production and qE, high pmf downregulates LEF (Rumberg et al., 1968) at the site of PQH 2 oxidation: originally called "back pressure, " (Stiehl and Witt, 1969), and now known as "photosynthetic control. " Comparison of the Cr∆rbcL mutant against the double mutant Cr∆rbcL pgr5 or wildtype (WT) clearly shows the effects of photosynthetic control imposed by CEF witnessed by a strong increase in chlorophyll fluorescence in the single mutant indicative of a gradually decreasing electron flow, while the double mutant has fluorescence kinetics that resemble the WT (Johnson et al., 2014). The mutant lacking RuBisCO is an important genetic tool to observe the effects of an absence of CO 2 fixation on CEF because it is difficult to modulate CO 2 much below atmospheric concentrations in Chlamydomonas due to the efficiency of the carbon concentrating mechanism (CCM): thus this double mutant gives us a window on the mechanisms of CEF as if the WT were under strong CO 2 -limited conditions. As important as qE in fluctuating light, photosynthetic control is established when a rapid response to FIGURE 2 | Cyclic electron flow promotes proton accumulation in the lumen and triggers regulatory mechanisms that can protect PSI from photoinhibition. Under constraining conditions, electrons are recycled from the acceptor-side of PSI by PGR5-CEF that results in a rapid acidification of the lumen. This promotes (i) the energy-dependent quenching of PSII antennas (qE) and (ii) photosynthetic control at the level of cytb6f that exerts reducing pressure on PSII to provoke a controlled photoinhibition (qI). These mechanisms result in a decrease of electron flow to PSI. Regulation of ATPsynthase conductivity by protons, electrochemical gradient partitioning and O2 photoreductive pathways produce ∆pH, producing ATP and contributing to the recycling of NADPH. Extra ATP produced by CEF is used by the Calvin-Benson-Bassham (CBB) cycle to assimilate CO2 and contributes to the regeneration of NADP + . Decreasing linear electron flow or increasing the sinks downstream of PSI avoids the over-reduction of ferredoxin (Fd) and PSI centers.
severe acceptor side limiting conditions is required to buffer a sudden burst of electron flow toward PSI (Suorsa et al., 2012). Moreover, while qE is not constitutive but inducible in Chlamydomonas, on the contrary to plants (Peers et al., 2009), photosynthetic control is likely to be crucial in the very first hours of exposition to drastic conditions. As a secondary consequence of photosynthetic control, reducing pressure increases on the Q B site of PSII so that PSII centers remain in a closed state longer. Overexcited chlorophyll ( 3 P680) activates O 2 into 1 O 2 triggering photoinhibition of PSII (reviewed, in Sonoike, 2011;Murata et al., 2012). Highlighting the key role of pmf in control of linear electron transfer, PSI was shown more susceptible to photoinhibition than PSII in nigericin-infiltrated leaves where control at the level of cytb 6 f could not develop (Joliot and Johnson, 2011). Moreover, photoinhibition of PSI in Atpgr5 could be avoided by modulating PSII turnover with the addition of the protein translation inhibitor lincomycin (Tikkanen et al., 2014). Thus, PSII photoinhibition mediated by photosynthetic control is a secondary level of photoprotection in drastic photoinhibitory conditions that exceed qE dissipation capacity: it indirectly but effectively acts as a shunt to avoid sustained PSI acceptor-side limitations (Figure 2).
In Chlamydomonas, the pgr5 mutation combined with an absence of chloroplast ATPsynthase results in a less photosensitive phenotype than the ATPase mutant alone, where light sensitivity has been attributed to luminal over-acidification (Johnson et al., 2014). This observation shows that photosynthetic control relying on CEF actively contributes to decreasing the luminal pH and supports previous work (Rott et al., 2011). In higher plants, triggering of low luminal pH has also been correlated with changes in conductivity of the ATPsynthase to protons (Kanazawa and Kramer, 2002) and also to partitioning of the proton motive force between its osmotic (or concentration gradient, ∆pH) and electrical (∆Ψ) component (Avenson et al., 2005). ATPsynthase conductivity to protons is increased in Atpgr5 (Avenson et al., 2005;Wang et al., 2015) with similar observations seen in knocked-down PGR5 rice lines (Nishikawa et al., 2012). This may be ruled by the concentration of substrate for ATP production, i.e., ADP and phosphate (P i ): in spinach thylakoids, artificially decreasing P i levels resulted in lower ATPase conductivity and a lower luminal pH, thus promoting qE (Takizawa et al., 2008). These observations show the metabolic interconnections between ATP, CEF and ATPsynthase. As already suggested (Shikanai, 2014), further studies should be done to explain the acceptorside limitation occurring in strains affected in PGR5-CEF in the light of the scenario proposed by Kramer and coworkers for qE regulation. regulates acceptor-side limitation in the absence of reactions for consumption of NADPH. Rubisco-less mutants but also CBB cycle mutants and those affected in starch metabolism show a strong increase in CEF or in CEF-dependent photosynthetic control that results in a repressed rate of LEF (Livingston et al., 2010;Johnson and Alric, 2012;Johnson et al., 2014;Krishnan et al., 2015). When the CBB cycle is an insufficient sink for reducing power, O 2 photoreduction pathways may work in conjunction with CEF to protect PSI. On the other hand, under non-acceptor side limited conditions (steady state high light and/or high CO 2 ) an absence of CEF in Crpgrl1 did not result in photoinhibition to PSI and this capacity to acclimate was shown to be due to a sustained dependence on O 2 photoreduction pathways (Dang et al., 2014). While photorespiration is a minimal process in green algae due to the CCM, other important sinks exist for reducing equivalents downstream of PSI that terminate on O 2 . These include: (i) export of reducing power to the respiratory chain to stimulate oxidative phosphorylation in the mitochondria, (ii) ROS-producing ("Mehler") reactions with a concomitant increase in detoxifying enzymes and (iii) ROS-independent ("Mehler-like") NADPH:O 2 oxidoreduction probably by flavodiiron proteins (FLV; Peltier et al., 2010). Mechanisms (ii) and (iii) dually generate proton gradient and thus ATP Elli, 1995, 1996) and regenerate NADP + thus avoiding PSI acceptor-side limitations, and over-expression of FLV proteins 1 and 3 in cyanobacteria have been observed to stabilize PSI under fluctuating light (Allahverdiyeva et al., 2013). Mechanism (i) mitochondrial cooperation, also generates ATP but the ATP is probably not reshuttled back into the chloroplast, its major role would be thus to regenerate oxidized NADP + (Figure 2). Radmer and Kok (1976) first observed the potential for O 2 to replace CO 2 fixation during a light-to-dark transition or in the presence of CBB cycle inhibitors. The role of such an acceptor side activity within the chloroplast such as Mehler (O 2 reduction) or hydrogenase (H + reduction) would enable Chlamydomonas cells to reoxidise the electron transport chain in the light, convincingly shown after anaerobic incubation (Forti et al., 2005;Ghysels et al., 2013). In the Crpgr5 ∆rbcL, lacking both CO 2 fixation and CEF, O 2 photoreduction rates can completely compensate for CO 2 fixation resulting in WT O 2 evolution levels (Johnson et al., 2014). Similarly, in a detached leaf assay addition of antimycin A provokes both production of H 2 O 2 and a strong sustained malate dehydrogenase activity resulting in high rates of mitochondrial O 2 uptake (Fridlyand et al., 1998). While very removed from the steady-state metabolic flow observed in WT strains under standard conditions, these experimental observations provide us with the maximal rates for the different pathways, and suggest possible compensatory reactions. It would appear that CEF down regulates ATP-independent O 2 reducing pathways and up regulates ATP-dependent CO 2 reduction by CBB cycle. Therefore, CEF can be seen as limiting ROS production under acceptor side limitations. Furthermore, it has been suggested that an interplay between CEF and O 2 photoreduction acts as a buffer to poise electron flow toward carbon fixation (Backhausen et al., 2000). The action of H 2 O 2 as an activator of NDH CEF in Arabidopsis provides further evidence that O 2 photoreduction pathways and CEF are working in tandem (Strand et al., 2015). The model that emerges is that regulation of temporary excesses of reductant at the acceptor side of PSI is controlled by an interplay between CEF, the Mehler reaction, FLV proteins and the malate valve with another level of control exerted by redox regulators such as thioredoxins (Scheibe and Dietz, 2012). These pathways likely form a set of communicating reactions that can rebalance NADPH/NADP + ratios and avoid PSI photoinhibition.

CONCLUDING REMARKS
While the study of mutants reveals to us the limitations of a system, the complete photosynthetic apparatus is perfectly able to acclimate to both light and changing redox conditions with CEF and its protective role over PSI placed centrally as a regulator of this flexibility. Further understanding of PSI photoinhibition, proposed to be a major determinant in crop productivity (Tikkanen et al., 2014), may allow the rational modification of photosynthesis to improve the efficiency of plant crops and the production of renewable algal biomass.

ACKNOWLEDGMENTS
This work was supported by grants from the Agence Nationale de la Recherche (ChloroPaths: ANR-14-CE05-0041-01) and the CEA Tech Department of Commissariat des Energies Atomiques et Energies Alternatives (CEA) for FC grant. We thank Jean Alric for constructive discussions.