The role of Ca²⁺ and protein scaffolding in the formation of nature’s water oxidizing complex

Significance

Unlike cofactor insertion into other metalloproteins, assembly of the photosynthetic water oxidation complex of photosystem II (PSII) in plants, algae, and cyanobacteria is a light-driven process that photooxidatively incorporates Mn²⁺ and Ca²⁺ ions into the protein matrix forming the catalytic Mn₄CaO₅ metal cluster. This self-assembly process is important both for de novo biogenesis and for the frequent repair of PSII due to its susceptibility to photodamage. While the basic kinetic scheme for this process was established nearly 50 y ago, the molecular details have remained enigmatic. Here we describe results on the role of inorganic and protein cofactors and integrate them with previous information to obtain an important upgrade in our understanding of this process.

Abstract

Photosynthetic O₂ evolution is catalyzed by the Mn₄CaO₅ cluster of the water oxidation complex of the photosystem II (PSII) complex. The photooxidative self-assembly of the Mn₄CaO₅ cluster, termed photoactivation, utilizes the same highly oxidizing species that drive the water oxidation in order to drive the incorporation of Mn²⁺ into the high-valence Mn₄CaO₅ cluster. This multistep process proceeds with low quantum efficiency, involves a molecular rearrangement between light-activated steps, and is prone to photoinactivation and misassembly. A sensitive polarographic technique was used to track the assembly process under flash illumination as a function of the constituent Mn²⁺ and Ca²⁺ ions in genetically engineered membranes of the cyanobacterium Synechocystis sp. PCC6803 to elucidate the action of Ca²⁺ and peripheral proteins. We show that the protein scaffolding organizing this process is allosterically modulated by the assembly protein Psb27, which together with Ca²⁺ stabilizes the intermediates of photoactivation, a feature especially evident at long intervals between photoactivating flashes. The results indicate three critical metal-binding sites: two Mn and one Ca, with occupation of the Ca site by Ca²⁺ critical for the suppression of photoinactivation. The long-observed competition between Mn²⁺ and Ca²⁺ occurs at the second Mn site, and its occupation by competing Ca²⁺ slows the rearrangement. The relatively low overall quantum efficiency of photoactivation is explained by the requirement of correct occupancy of these metal-binding sites coupled to a slow restructuring of the protein ligation environment, which are jointly necessary for the photooxidative trapping of the first stable assembly intermediate.

Photosystem II (PSII) utilizes solar energy to catalyze one of the most important and most thermodynamically demanding reactions in nature: the oxidation of water into protons and molecular oxygen. The electrons extracted from the substrate water molecules are transferred through the redox-active cofactors of the photosynthetic electron transport chain eventually to reduce the electron acceptor NADP⁺, thereby forming the primary reductant for the synthesis of biomass from CO₂ and other inorganic nutrients. Thus, the H₂O-oxidation reaction is the basis of oxygenic photosynthetic metabolism and the primary driver of biomass accumulation on the planet (1, 2) and represents a key chemical process for the development of carbon neutral energy technologies (3, 4). Natural H₂O oxidation is driven by light-induced charge separation within the PSII reaction center (RC), a 700-kDa membrane protein homodimer consisting of over 20 different subunits and ∼60 organic and inorganic cofactors (for reviews, see refs. 1, 2). The catalysis of H₂O oxidation is mediated by a metal cluster (Mn₄CaO₅) buried within the protein complex at the interface between intrinsic and extrinsic polypeptides toward the luminal surface of the photosynthetic membrane (5) (SI Appendix, Fig. S1). Photoexcitation of the multimeric chlorophyll (Chl) P₆₈₀, which functions as the primary photochemical electron donor, results in the primary charge separation on a picosecond timescale into the highly oxidizing radical cation P₆₈₀⁺ and radical anion Pheo⁻ (6). To minimize the backreaction or oxidation of Chl and/or neighboring proteins, the highly reactive P₆₈₀⁺ is rereduced (20–250 ns) by the redox active tyrosine D1-Tyr161 (Y_z) of the D1 reaction center polypeptide located on the donor side proximal to P₆₈₀. Meanwhile, the energized electron is stabilized by transfer from Pheo⁻ through quinone acceptors (Q_A and Q_B) and on through the remainder of the intersystem electron pathway leading to photosystem I. During the course of four consecutive charge separation events, the Mn cluster passes through a series of oxidant storage states (S-states) with the catalytic cycle balanced by removing four electrons from two bound water molecules with the release of O₂ and four protons.

The PSII complex is subject to incessant photodamage, and a remarkable feature is its ability to undergo self-repair. Photodamage primarily occurs within PSII, and much of the damage is localized in the D1 (PsbA) protein, which binds the main redox cofactors involved in photochemical charge separation. Efficient mechanisms have evolved to remove and replace damaged reaction center proteins and assemble them with their requisite cofactors (reviewed in ref. 7). A key step in both de novo synthesis and the repair synthesis of PSII is the assembly of the Mn₄CaO₅ core into the ligation environment of the PSII protein matrix (8–13). Referred to as photoactivation, the assembly of the Mn₄CaO₅ occurs through a series of photochemical reactions that involve the oxidation of Mn²⁺ ions using the same electron extraction pathway of the fully mature PSII (Fig. 1). Charge separation oxidizes Mn²⁺ ions to Mn^≥3+ as the oxo-bridged multinuclear metal center forms. The multistep process begins with the binding of a single Mn²⁺ ion (14), as its hydroxide (15) to a high-affinity site (HAS) involving the D1-Asp170 carboxylate moiety (16, 17). Additional Mn²⁺ ions are photooxidatively incorporated into the growing metal cluster via the oxo-bridges that are presumably derived from water ligands of the incoming metal ions. The quantum efficiency of photosynthetic H₂O oxidation in the fully functional PSII is greater than 90%. In contrast, the photoassembly of the metal cluster is remarkably inefficient with an overall quantum efficiency below 1%, despite the fact that photooxidative assembly uses the same charge separation cofactors, including the oxidized forms of the primary and secondary electron donors, P₆₈₀ and Y_Z, respectively. To account for the low quantum efficiency and for a still unresolved light-independent “dark rearrangement” that must occur between two or more light-induced charges, a so-called “two-quantum model” (Fig. 1) was developed (8). Experimental support for this kinetic model is comprehensive (reviewed in refs. 7, 18), including, for example, direct evidence tracking the assembly process demonstrating a two-quantum requirement, and that part of the overall inefficiency is due to the competition between the slow and/or inefficient assembly steps and back-reactions of the charge-separated state (19). Nevertheless, evidence regarding the structure and chemistry of the intermediates and the nature of the dark rearrangement has remained scarce.

Fig. 1.

Kinetic scheme of basic two-quantum mechanism of PSII photoactivation. Double arrows indicate light-activated processes with the quantum efficiencies Φ₁ and Φ₂ representing the first and second photooxidative events in the assembly sequence, k_A representing the dark rearrangement, and k_D representing the decay of intermediates. After the initial two Mn are photoligated, subsequent Mn appear to be added with high quantum efficiency.

Calcium is a critical cofactor in the process of H₂O oxidation by PSII because it mediates the delivery of substrate water to a Mn coordination site for oxidation and dioxygen formation (20). While an early study suggested that Ca²⁺ is not required for photoassembly of the Mn cluster (21), this was later reevaluated (22), and there is now general agreement that Ca²⁺ is vital (11, 13, 22). Nevertheless, the mechanism of how Ca²⁺ facilitates the photoassembly remains obscure. Binding of Mn²⁺ to the HAS with simultaneous binding of Ca²⁺ to an adjacent binding site facilitates the formation of the [Mn²⁺-(OH)-Ca²⁺] complex by inducing deprotonation of a water ligand of Mn²⁺. Calcium lowers the pK_a for water ligand, which is controlled by a nearby base B⁻ that serves as a primary proton acceptor with a pK_a dependent on Ca²⁺ bound to its effector site (23). The deprotonation of the intermediates and the tuning of pK_as facilitated by Cl⁻ binding is critical for assembly (24). The absence of Ca²⁺ during photoactivation leads to the formation of noncatalytic, multinuclear high-valence Mn species (“inappropriately bound Mn”) that inactivates the PSII complex (22). At the same time, high concentrations of the Ca²⁺ cofactor diminishes the efficiency of photoactivation. These results indicate that Ca²⁺ and Mn²⁺ compete for each other’s binding sites, leading to an optimality relationship in their relative concentrations during photoactivation (11–13).

Three extrinsic polypeptides, PsbO, PsbV, and PsbU, serve to enclose and stabilize the cyanobacterial water oxidation complex (WOC), whereas plants and certain eukaryotic algae possess PsbO, PsbQ, and PsbP (for review, see ref. 25). The extrinsic polypeptides prevent the reduction of the Mn cluster by exogenous reductants (26–28) and help retain the Ca²⁺ cofactor, which is otherwise prone to loss during the catalytic cycle (29, 30). However, by enclosing the WOC, accessibility of ions for the photoactivation of the Mn₄CaO₅ may be restricted. Indeed, the most efficient in vitro photoactivation procedures either explicitly or coincidentally involved the biochemical removal of the extrinsic proteins, and genetic deletion of the most evolutionarily conserved extrinsic protein, PsbO, increases the quantum efficiency of photoactivation (7). Proteomic analysis of highly intact cyanobacterial PSII revealed novel proteins, including a small protein designated Psb27, apparently associated with PSII subpopulations (31). Psb27 was identified as a lipoprotein associated with inactive PSII monomers that prevents binding of PSII extrinsic subunits (PsbO, PsbU, and PsbV) to the premature PSII (32), keeping the active site “open” and thereby maintaining a sufficient diffusion rate of Mn²⁺, Ca²⁺, and Cl⁻ ions, thus acting as a molecular chaperone for successful photoactivation (33), and does so through a specific interaction with the E-loop of CP43, which is a luminal domain of PSII that directly interacts with the Mn₄CaO₅. Overall, Psb27 appears to be strictly associated with organisms possessing PSII and is important for its assembly (34–37), yet its precise role in facilitating assembly remains to be resolved. The application of in vitro photoactivation procedures using genetically tractable cyanobacteria has not been described, and here we describe such a system and use it for the analysis of the roles that Ca²⁺ and Psb27 play in the assembly of the Mn₄CaO₅.

Results

Mn²⁺ and Ca²⁺ Competition during Photoassembly Leads to Decreased Quantum Efficiency and Yield during Photoactivation of Cyanobacterial Membranes.

We developed a procedure for isolating Mn-depleted thylakoid membranes from Synechocystis sp. PCC6803 (hereafter, Synechocystis), enabling control of the photoactivation conditions, such as pH and ion composition, as well as permitting facile genetic modification of the constituent proteins. Importantly, the membranes retain the native electron acceptor system, so that artificial electron acceptors are not necessary as in previous in vitro experimental systems. Mn-depleted membranes produced by hydroxylamine (HA) extraction showed substantial restoration of photosynthetic activity (40%) (SI Appendix, Fig. S2A and Table S1), consistent with published yields in plant preparations (see ref. 12 for quantitative analysis), and, most importantly, display similar kinetic features of photoactivation compared to plant membrane preparations and to in vivo photoactivation experiments in Synechocystis (SI Appendix, Fig. S3). We focused on the role of Ca²⁺ in photoactivation and how the extrinsic proteins modulate the demand for both Mn²⁺ and Ca²⁺. As shown previously, photoactivation requires presence of both Mn²⁺ and Ca²⁺ cations for the assembly of PSII, and postaddition of Ca²⁺ after illumination only resulted in a very small increase in the yield of active PSII (SI Appendix, Fig. S2B). Postaddition of Sr²⁺ or Mg²⁺ cations in the dark did not result in a significant yield of photoactivation. Thus, our results concur with the general conclusion that Ca²⁺ is absolutely required for the assembly process (11, 13, 22).

In plant preparations, optimal photoactivation, both in vitro (12, 13, 38) and in organello (11), requires an optimal Ca²⁺/Mn²⁺ ratio characterized by an excess of Ca²⁺ relative to Mn²⁺, which reflects a competition between the ions for their respective binding sites (12). To establish whether similar kinetic features operate in cyanobacteria, the Ca²⁺ concentration dependence of photoactivation was performed at two fixed Mn²⁺ concentrations, 250 or 500 µM (Fig. 2 A and B). These concentrations of Mn²⁺ should saturate its binding to the HAS (13), but should still allow observation of the predicted competition between the two cations observed in plant preparations (11–13). A small level of photoactivation can be seen in the absence of added Ca²⁺ ions, which could be explained by trace amounts of residual Mn²⁺ and Ca²⁺ ions remaining in the extracted thylakoid preparation, since the level of this activity could be minimized, but never completely eliminated, by extensive washing of membranes using Chelex-treated buffer. For all of the samples containing 250 µM Mn²⁺ in photoactivation buffer, the maximum yield of photoactivation was observed at 600–700 flashes, with the half-maximal yield at occurring with ∼150–200 flashes (Fig. 2A). In terms of maximal yield and quantum efficiency of photoactivation, the optimum Ca²⁺ concentration for photoactivation at 250 µM Mn²⁺ occurs at 10 mM, corresponding to Ca²⁺/Mn²⁺ ratio of 40:1. The decrease in O₂ evolution at lower Ca²⁺ concentrations is likely caused by the competitive binding of Mn²⁺ to the Ca²⁺ binding site. In the context of earlier findings, this would prevent the formation of a bridged species [Mn³⁺-(OH)-Ca²⁺] that is proposed to be a crucial intermediate during the assembly (23), and perhaps leading to the formation of inappropriately bound high-valence, multinuclear Mn deposits (22). To better understand the role of ion competition and test the hypothesis that Mn²⁺ binding at the Ca²⁺ site leads to photoinactivation of PSII during photoassembly, we carried out the same experiment, but doubling Mn²⁺ concentration to 500 µM (Fig. 2B). For the optimum photoactivation at 500 µM Mn²⁺ the requirement for Ca²⁺ increased two-fold. Interestingly, samples containing 20 and 40 mM of Ca²⁺ in photoactivation buffer started to show signs of photoinactivation observed earlier at low Ca²⁺ concentration at 250 µM Mn²⁺. From these results we can conclude that an excess of Mn²⁺ ions leads to the decreased yield of PSII photoactivation due to photoinactivation, while excess of Ca²⁺ ions does not have such effect.

Fig. 2.

Calcium dependence of photoactivation under sequence of single turnover flashes at 2 Hz (500-ms flash interval) of HA-extracted thylakoid membranes from WT control at 0 (black circle), 2 (blue square), 5 (purple diamond), 10 (red circle), 20 (green square), and 40 (orange diamond) mM of CaCl₂ combined with 250 μM (A) or 500 μM MnCl₂ (B). (C) Overall quantum efficiency of photoactivation (Φ_PA). (D) Quantum efficiency of inactivation (Φ_PI). Data were fit to Eq. 1 for parameter estimation. Error bars represent SD with n ≧ 3.

To analyze the role of Ca²⁺ concentration on quantum efficiency of photoactivation and photoinactivation as a function of flash number and to better understand inhibitory effect of low Ca²⁺ to Mn²⁺ ratios, data described in Fig. 2 was fit to a double exponential equation:

$${\left[A\right]}_n={\left[\text{A}\right]}_0\cdot \left(1-e^{{\mathrm{\Phi }}_{PA}\cdot n}\right)\cdot \left(e^{-{\mathrm{\Phi }}_{PI}\cdot n}\right)$$ [1]

The equation accounts for the progressively smaller pool of apocenters during the flash sequence as more centers become photoactivated (10), combined with a term that represents the loss of centers due to photoinactivation processes (12). Here, [A]_n represents the yield of active centers on the nth flash, whereas [A]₀ is the concentration of apo-PSII centers prior to the photoactivation, Φ_PA, which represents the overall efficiency of the multiphoton assembly process. The photoinactivation term, Φ_PI, represents the quantum efficiency of irreversible photodamage or, alternatively, the formation of inactive centers due to inappropriately bound Mn as a consequence of supraoptimal Mn²⁺ concentrations (22). This is also seen with varying [Mn²⁺] at fixed [Ca²⁺], where there is a strong dependence of Φ_PI on [Mn²⁺] (SI Appendix, Figs. S4 and S5). The corresponding estimates of parameters are plotted as a function of the Ca²⁺ concentration (Fig. 2 C and D). Additionally, the yields of photoactivation, indicated by the levels reached after ∼500 flashes (Fig. 2 A and B), increase with [Ca²⁺] up to 10 mM Ca²⁺ at 250 µM Mn²⁺ and 20 mM Ca²⁺ at 500 µM Mn²⁺, but higher [Ca²⁺] decreased yields, especially at 250 µM Mn²⁺. Interestingly, at 250 µM Mn²⁺ the quantum efficiency of photoactivation, Φ_PA, is relatively unaffected by Ca²⁺ concentration throughout the range tested (Fig. 2C). In contrast, at the higher (500 µM Mn²⁺) concentrations, Φ_PA is strongly dependent on Ca²⁺ availability, perhaps reflecting competitive occupation of Mn²⁺ in the Ca²⁺ effector site. As shown in Fig. 2C, at low [Ca²⁺], the Φ_PA is low, reaches a maximum at 20 mM coinciding with overall apparent optimum (Fig. 2B), and declines at higher [Ca²⁺]. Overall, higher abundance of Mn²⁺ ions inhibits the assembly through competition with Ca²⁺ (11–13).

The difference between Φ_PA being unaffected by Ca²⁺ at 250 µM Mn²⁺ versus its being affected at 500 µM in Fig. 2C is intriguing and suggests a complex interplay of binding constants. The HAS likely remains occupied by Mn²⁺ (K_D < 10 µM) (13, 14, 16), with both the 250- and 500-µM Mn²⁺ experiments, even with comparatively high [Ca²⁺]. However, at the higher [Mn²⁺] the results indicate that the Ca²⁺ effector site is substantially occupied by Mn²⁺, leading to inactivation, consistent with previous kinetic analysis (12). It is unclear at which step(s) the replacing of Ca²⁺ ion with Mn ion is inhibitory; however, we find that Ca²⁺ can even be replaced with Mn in assembled PSII, producing a light-dependent inactivation during the catalytic turnover of the S-state cycle of the intact Mn cluster (SI Appendix, Fig. S6). This fits with the observation that Ca²⁺ is more readily released in the higher S-states, presumably due to charge repulsion (29, 30). Accordingly, the replacement of Ca²⁺ by Mn²⁺ at the Ca²⁺ site results not only in a failure to advance in photoassembly, but also a greater frequency of inactivation (Fig. 2D). The optimal Ca²⁺/Mn²⁺ ratio for overall photoactivation thus reflects a balance between binding of Ca²⁺ at its effector site, which prevents photoinactivation due to inappropriate binding of Mn²⁺ (22), but not so high as to outcompete Mn²⁺ at a Mn site preventing photooxidative Mn incorporation (12–14). But which Mn site? Based upon the fact that Φ_PA at lower [Mn²⁺] is independent of [Ca²⁺] in the range tested, as well as previous estimates of Mn²⁺ affinity at the HAS, we conclude that the yield-limiting competition between Ca²⁺ at a Mn²⁺ site occurs not at the HAS, but rather at a second Mn site (SMS) involved in the photoactivation pathway. This is consistent with the results of the effect of [Ca²⁺] on the rate of the dark rearrangement step, k_A (next section).

Calcium Stabilizes Intermediates of Photoactivation and Prolongs the Dark Rearrangement Time.

To test the hypothesis that Ca²⁺ influences the stability of the assembly intermediates, photoactivation was performed varying [Ca²⁺] and the time interval between photoactivating xenon light flashes (Fig. 3). According to the two-quantum model of photoactivation (Fig. 1) the initial photooxidation of Mn²⁺ (state “A”) produces the first unstable intermediate “B,” followed by a light-independent “rearrangement” step, leading to the formation of a second unstable intermediate “C,” which is a configuration capable of productively utilizing the second light quantum to form the first stable intermediate, “D” (8). Short intervals produce low yields since not enough time has elapsed for completion of the dark rearrangement. Long intervals between the charge separations allow the decay of intermediates resulting in low yields of active centers (8, 9). This results in a bell-shaped curve recording the yield of photoactivation due to a fixed number of flashes given at different flash intervals from 50 to 10,000 ms (Fig. 3). Estimations of the parameters under each condition (Table 1) describing the dark rearrangement, k_A, and the decay of intermediates, k_D, were determined by deriving kinetic parameters from the rising and falling slopes of the bell-shaped curve in plots of photoactivation as a function of the flash interval curve fitted to Eq. 2:

$${\left[A\right]}_n=\left[k_A/k_D-k_A\right]\times {\left[A\right]}_O\times \left(e^{-k_{A}td}-e^{-k_{D}td}\right)$$ [2]

The overall shape of the resultant bell-shaped curves closely resembles previous results using higher plant material with estimated values for dark rearrangement (k_A, B➝C), and the decay of intermediates (k_D, B and C) are in the range of 200–400 ms and 7–14 s, respectively, in our experiments (Table 1). These values are slower than with spinach PSII preparations, but in the same proportions with the dark rearrangement, k_A, typically 10-fold faster than the rate of decay, k_D (reviewed in ref. 18). These differences may be attributed to the different source of experimental material and/or choice of buffers (e.g., the sucrose concentration is much higher than typically used). Depending upon the [Ca²⁺], different optima for maximal yields of photoactivation were observed with increasing Ca²⁺ giving increasingly longer optima for the length of the dark period between flashes (Fig. 3). Photoactivation with 40 mM Ca²⁺ enhanced yields at the longest interval tested (10 s), again suggesting that Ca²⁺ stabilizes the intermediates of photoactivation. Accordingly, the fit values for the decay of intermediates k_D, are shifted to slower rates of decay (Table 1). The stabilizing effect of Ca²⁺ on photoactivation intermediates contrasts with the existence of the optimal [Ca²⁺]/[Mn²⁺] (Fig. 2) ratio for the net yield under repetitive flashing at a constant interval (500 ms). The flash interval experiment also provides estimates for the second parameter in the two-quantum model, the dark rearrangement, k_A, (B➝C, Fig. 1). This is the time needed between flashes before subsequent flashes become productive, which is experimentally reflected by diminished yields of photoactivation at short flash intervals (Fig. 3). Remarkably, the rearrangement takes longer time to complete at high [Ca²⁺] concentrations (Table 1).

Fig. 3.

Photoactivation yields as a function of flash interval at different calcium concentrations. Sequences of 150 Xe flashes applied at different flash intervals to HA-extracted thylakoid membranes from WT control and oxygen yields measured on a bare platinum electrode. Plots correspond to samples containing 2 (blue circle), 10 (red square), and 40 (black diamond) mM of CaCl₂ a fixed [Mn²⁺] = 250 μM. Error bars represent SD with n ≧ 3. Data were fit to Eq. 2 to estimate the dark rearrangement constant, k_A, and the decay of photoactivation intermediates “B and C,” k_D (see Fig. 1 for kinetic model and Table 1 for estimated values for k_A and k_D).

Table 1.

Dark rearrangement, k_A, and decay of intermediates, k_D, parameters characteristic of photoactivation of membranes from WT, ∆psbO, and 27OE strains

Strain/[Ca2+] (mM Ca2+)	kA (s−1) (t1/2 [ms])	kD (s−1) (t1/2 [s])	Fit quality, r
WT control, 2	2.4 (288)	0.134 (5.17)	0.99
WT control, 10 mM Ca2+	2.8 (244)	0.095 (7.29)	0.98
WT control, 40 mM Ca2+	1.6 (433)	0.051 (13.5)	0.99
∆psbO, 2 mM Ca2+	1.9 (364)	0.104 (6.66)	0.99
∆psbO, 10 mM Ca2+	2.2 (315)	0.072 (9.62)	0.99
∆psbO, 40 mM Ca2+	1.5 (462)	0.049 (14.1)	0.99
27OE, 2 mM Ca2+	1.5 (462)	0.130 (5.33)	0.98
27OE, 10 mM Ca2+	1.4 (495)	0.085 (8.15)	0.99
27OE, 40 mM Ca2+	1.8 (385)	0.027 (25.6)	0.99

Data were fit to Eq. 2 to estimate the dark rearrangement constant, k_A, and the decay of photoactivation intermediates “B and C," k_D (see Fig. 1 for model). Error bars represent standard deviation n ≧ 3.

Our analyses of the preceding results leads us to conclude that at high [Ca²⁺], assembly occurs, albeit suboptimally, due to competition between metals for their respective sites as shown before (11–13). However, we now find that high [Ca²⁺] also allows assembly at long flash intervals due to stabilization of intermediates “B” and/or “C” (Fig. 1). On the other hand, at low [Ca²⁺], photoactivation yields are also decreased (Fig. 3), at least partly due to photoinactivation when incorrectly Mn²⁺ occupies the Ca²⁺ site. The protective function of Ca²⁺ is important also in PSII centers that already have been assembled (SI Appendix, Fig. S6) consistent with the proposed “gate-keeper” function observed with intact PSII preparations (28, 39). Thus, we conclude that Ca²⁺ is important for assembly because it stabilizes intermediate “B” and/or “C” and that it prevents inactivation and failure to advance due to inappropriate binding of Mn²⁺ into the Ca²⁺ site during assembly (12, 22). Besides its gate-keeping role, Ca²⁺ can also block the necessary binding of the second Mn²⁺ especially at very high [Ca²⁺]/[Mn²⁺] ratios. This accounts for the slowing of the very slow dark rearrangement (B➝C). While sufficient concentrations of Ca²⁺ increase the chance of successful formation of intermediate “C,” an excess of Ca²⁺ competes with the binding of the second Mn²⁺ at the SMS, thereby delaying the time before the second Mn²⁺ ion can occupy its site for photooxidation. This competitive inhibition extends the time before the rearranged state can be trapped (C⇒D) until the competing Ca²⁺ is replaced with Mn²⁺ enabling the photooxidative formation of stable intermediate “D.” This, in turn, suggests that the dark rearrangement consists of a molecular reorganization (e.g., conformational change) that is only fruitful if a second Mn²⁺ bound at its correct site.

Role of PsbO and Psb27 during Photoactivation.

The PSII assembly cofactor protein, Psb27, which appears to facilitate diffusional access to the WOC assembly site (32, 33, 37), does so by interacting with the E-loop of CP43 to allosterically modify its conformation (36, 40), which is hinged (18) in a way that opens the WOC for greater diffusion into the sites of cofactor binding (33). Since the natural steady-state abundance of Psb27 in the cell likely evolved to cope with assembly and repair of only a fraction of centers from the total population of PSII in the cell, then there may not be enough copies of Psb27 to stoichiometrically service all PSII centers upon quantitative removal of the Mn cluster experimentally produced by HA extraction. Similarly, diffusion of Mn²⁺ and Ca²⁺ ion could also be limited by the presence of the extrinsic proteins, impeding photoassembly as previously suggested from in vivo experiments with mutants (33, 41). To address these considerations, the photoactivation of two Synechocystis mutants, ∆psbO lacking PsbO (42), and a strain overexpressing Psb27 (27OE) (SI Appendix, Fig. S7), were investigated. Importantly, the more open configurations of the WOC in both strains enabled a net recovery of O₂ evolution under continuous illumination that was more than 50% higher than in the wild-type (WT) control (SI Appendix, Table S1). Additionally, the optimal [Ca²⁺]/[Mn²⁺] ratio for both mutants appears to be about twice as high compared to the WT control (80/1 vs. 40/1) indicating alterations of the donor side polypeptide structure have a differential effect with the demand for Ca²⁺ being higher and/or a lower than the demand for Mn²⁺ (Fig. 4). Interestingly, both mutants showed somewhat lower quantum efficiency, Φ_PA, compared to WT control; however, at higher [Ca²⁺], the yield continued increasing without saturation or decline through the entire flash sequence, suggesting that while the demand for Ca²⁺ was higher, fewer centers were lost to photoinactivation (compare with WT in Fig. 2). Accordingly, ∆psbO and 27OE appeared less prone to photoinactivation (Φ_PI) compared to the WT control (Fig. 4 and SI Appendix, Figs. S8–S11) suggesting the open configuration diminishes the tendency for Mn²⁺ to occupy the Ca²⁺ effector site, but with the optimum [Ca²⁺] shifted higher. These observations are consistent with increased diffusion of the metal cofactors inside apo-PSII with proportionally increased exchange rates for Ca²⁺ at its effector site. Neither mutant showed a strong correlation between [Ca²⁺] and quantum efficiency of photoactivation (Fig. 4C), which is similar to WT control preparations at the lower, 250 µM [Mn²⁺] tested, but not the higher, 500 µM [Mn²⁺]. Photoinactivation rate (Fig. 4D) is negligible for both mutant strains. With increased diffusional access, Ca²⁺ appears to more effectively prevent the inhibitory effect of Mn²⁺ on photoassembly. Notably, ∆psbO and 27OE show different Mn-dependent photoactivation behavior compared to WT control. While 27OE has a requirement similar to the WT control for Mn²⁺ ions, with the optimum at 250 µM at 10 mM Ca²⁺, ∆psbO appears to have a lower requirement for available Mn²⁺ and shows maximal yield of photoactivation in the range of concentrations from 50 to 500 µM (SI Appendix, Figs. S8–S11).

Fig. 4.

Calcium dependence of photoactivation under sequence of single turnover flashes at 2 Hz (500-ms flash interval) of HA-extracted thylakoid membranes from ∆psbO (A) and 27OE (B) at 0 (black circle), 2 (blue square), 5 (purple diamond), 10 (red circle), 20 (green square), and 40 (orange diamond) mM of CaCl₂ combined with 250 μM MnCl₂. (C) Overall quantum efficiency of photoactivation (Φ_PA). (D) Quantum efficiency of inactivation (Φ_PI) in ∆psbO (black circle) and 27OE (red square) membranes, respectively. Data were fit to Eq. 1 for parameter estimation. Error bars represent SD with n ≧ 3.

As with the WT, the flash interval experiment shows that at longer intervals between flashes, Ca²⁺ stabilizes assembly intermediates (Fig. 5) as reflected in the slower decay constant, k_D (Table 1). However, 27OE decreases in k_D overall, further slowing the decay with the increase of [Ca²⁺], suggesting a possible role for Psb27 in stabilizing the photoactivation intermediates “B” and/or “C” (Fig. 1), perhaps by stabilizing the binding of Ca²⁺ at its effector site. Additionally, the dark rearrangement, k_A, is generally slower than the WT for both ∆psbO and 27OE (Table 1), although higher [Ca²⁺] does not further slow the dark rearrangement in 27OE as it does for WT and ∆psbO. Apparently, both high [Ca²⁺] and the more open configuration of apo-PSII leads to a slower dark rearrangement, k_A (Table 1), yet promotes better exchange of Ca²⁺ and prevents the decay of the photoactivation intermediates (Table 1). The increased stabilization of intermediates in 27OE in comparison with ∆psbO accords with the differences in the final yields of active centers as a function of flash number at different [Ca²⁺] (Fig. 4A vs. Fig. 4B). Note, for example, that the 10 mM Ca²⁺ reaction in 27OE produces higher yields compared to ∆psbO, closely approaching the yields at the optimal 20 mM [Ca²⁺]. Taken together, the results suggest that Psb27 enhances the stabilization of intermediates with accessibility of the site of cluster assembly, especially for Ca²⁺ ions, thereby shifting the optimal ratio of Mn²⁺ and Ca²⁺. As discussed, given the known characteristics of the interaction of Psb27 with apo-PSII (36), this facilitation of photoactivation likely occurs allosterically via its interaction with the luminal E-loop of CP43, which, besides D1, provides a ligand to the mature Mn₄CaO₅.

Fig. 5.

Photoactivation yields as a function of flash interval at different calcium concentrations. Sequences of 150 Xe flashes applied at different flash intervals to HA-extracted thylakoid membranes from ∆psbO (A) and 27OE (B) with O₂ yields measured on a bare platinum electrode. Plots correspond to samples containing 2 (blue circle), 10 (red square), and 40 (black diamond) mM of CaCl₂ at a fixed [Mn²⁺] = 250 μM. Error bars represent SD with n ≧ 3. Data were fit to Eq. 2 to estimate the dark rearrangement constant, k_A, and the decay of photoactivation intermediates “B and C,” k_D (see Fig. 1 for kinetic model and Table 1 for estimated values for k_A and k_D).

Discussion

How Ca²⁺ Stabilizes Photoactivation Intermediates yet Retards the Dark Rearrangement.

Our results show that Ca²⁺ stabilizes the assembly intermediates of photoactivation, a feature especially evident at long intervals between photoactivating flashes (Figs. 3 and 5). Moreover, the results also show that excess Ca²⁺ slows down the already very slow dark rearrangement (B➝C), an effect that is enhanced by the more open configuration of the apo-PSII assembly site in the mutants where greater ion exchange occurs (Figs. 3 and 5 and Table 1). According to the two-quantum model of photoactivation (8), a stable intermediate “D” is formed by two light-dependent steps separated by a light-independent rearrangement (Fig. 1). The first light-activated step (A⇒B) involves photooxidation of a single Mn²⁺ ion (8, 10, 12, 14) bound as hydroxide [(Mn)-OH⁺] (15) at the HAS and occurs with high quantum efficiency (14, 16, 43, 44). The concurrent binding of Ca²⁺ at its adjacent effector site does not affect the affinity of the binding of this Mn²⁺, but upon photooxidation of Mn²⁺ in the presence of bound Ca²⁺, the bridging species, [Mn³⁺-(OH⁻)-Ca²⁺] ⇌ [Mn³⁺-(O^2-)-Ca²⁺], is produced (23), which is thought to facilitate the next steps. This rapid initial photooxidation (A⇒B) is followed by a remarkably slow (100–400 ms) rearrangement (B➝C), that involves a protein conformational change and/or ion relocation. Only then can the second charge separation become effective (C⇒D), photooxidizing a second Mn²⁺, and thereby trapping the first stable intermediate “D” (8). Current models suggest this first stable intermediate, “D,” is a binuclear (Mn³⁺)₂-(di-μ-oxo) bridged structure (9, 45), possibly corresponding to the binuclear di-μ-oxo bridged structure produced by partial disassembly of the intact Mn₄CaO₅ using reducing agents (26, 28) or thermal (45) treatment.

Based upon the findings that Φ_PA at lower [Mn²⁺] is independent of [Ca²⁺] (Fig. 2C) combined with estimates of Mn²⁺ affinity at the HAS (K_D < 10 μM) (13, 14, 16), we infer that the competition between Ca²⁺ for a critical Mn²⁺ site occurs not at the HAS, but rather at an SMS involved in the photoactivation pathway and that excess Ca²⁺, while protecting from photoinactivation, inhibits assembly due to occupation of the SMS preventing photooxidation of the second Mn²⁺ (C⇒D). The present experiments cannot discriminate the identity of the SMS, although it is reasonable to suppose the SMS includes ligands that form with the other Mn1-Mn3 positions (SI Appendix, Fig. S1B, orange spheres), and that corresponds to the site estimated to have K_D∼50 μM from steady-state photoactivation experiments (12). It also does not exclude the possibility, for example, that the Mn³⁺ ion initially produced from Mn²⁺ at the HAS (A⇒B) relocates to the SMS as part of the dark rearrangement (B➝C), according to the “translocation model” (7, 46).

The stabilization of intermediates (B, C) by Ca²⁺ is most simply explained in the same way that Ca²⁺ prevents the photoinactivation of apo-PSII. Photoinactivation is due to incorrect occupation of the Ca²⁺ effector site (CAS) by Mn²⁺ evident at suboptimal Ca²⁺/Mn²⁺ ratios, which can even occur in intact PSII (SI Appendix, Fig. S6) in line with the proposed gate-keeper function of Ca²⁺ in the assembled Mn₄CaO₅ (26, 28, 39). Furthermore, Ca²⁺ tends to be lost from its effector site due to the electrostatic repulsion occurring upon oxidation of Y_Z (30) and/or associated pK_a changes (29). Although the mechanism of Mn-dependent inactivation is not clear, occupation of the CAS by Mn²⁺ correlates with photoinactivation, as well as “misses” due to the requirement for Ca²⁺ to form the necessary bridging species with Mn at the HAS needed for advancement through the assembly sequence (23). Accordingly, the occupation of the CAS by Ca²⁺ is important throughout the duration of the dark rearrangement (B➝C). The observation that both ∆psbO and 27OE each exhibit minimal photoinactivation suggests that ion exchange at the Ca²⁺ site is rapid in the more open configuration, and this enables centers that have lost the Ca²⁺ ion to rapidly reacquire a replacement.

Besides stabilizing intermediates, Ca²⁺ also slows the dark rearrangement (Figs. 3 and 5 and Table 1). At supraoptimal concentrations, Ca²⁺ appears to block the necessary binding of the second Mn²⁺, at the second Mn-binding site, SMS, as already noted. While sufficient concentrations of Ca²⁺ increase the chance of successful formation of intermediate “C,” an excess of Ca²⁺ competes with the binding of the second Mn²⁺-binding site, thereby delaying the time before the second Mn²⁺ ion can occupy the SMS for photooxidation, forming the first stable intermediate “D,” which is predicted to be a binuclear (Mn³⁺)₂-(di-μ-oxo) bridged structure (9, 45). However, if Ca²⁺ competitively occupies the second Mn²⁺ site, then the rearranged configuration cannot be productively converted due to the absence of the second Mn²⁺ needed for the photooxidative second light step (C⇒D). This proposed model suggests that the dark rearrangement consists of a molecular reorganization (e.g., conformational change and/or ion relocation) that is only ready for photooxidative stabilization (C⇒D), if a second Mn²⁺ bound at its correct site, the SMS.

How Does Psb27 Facilitate Photoactivation?

It has been shown that Psb27 is the vital player in PSII repair and assembly of the Mn₄CaO₅ cluster of PSII that provides greater accessibility to the site of Mn-cluster assembly (32, 33, 36, 47–49). We find that Psb27 facilitates the photoactivation of the WOC in a more complex manner than simply displacing extrinsic polypeptides from apo-PSII. Both ∆psbO and 27OE have increased access of Mn²⁺ and Ca²⁺ to the apo-PSII consistent with expectations (reviewed in ref. 25), and also both mutations produce shifts toward higher optimal [Ca²⁺]/[Mn²⁺] ratios. In the absence of extrinsic proteins, light induces the loss of Ca²⁺ from its binding site in an S-state dependent manner (29, 30), and one of the main functions of the extrinsic proteins is to retain the ion (50, 51). Thus, the shift to higher optimal [Ca²⁺]/[Mn²⁺] ratios in the mutants is likely because in the absence of retention of Ca²⁺ in the vicinity of the assembly site, a relatively higher [Ca²⁺] is required. However, if the function of Psb27 only increased exchangeability of Ca²⁺, then ∆psbO and 27OE would have similar phenotypes in regard to photoactivation. This was not the case, and 27OE provides additional support of photoactivation: 27OE exhibits a remarkable stabilization at high [Ca²⁺], yet does not exhibit a proportionally dramatic increase in the dark rearrangement time at the highest [Ca²⁺] as seen with ∆psbO (Table 1). Additionally, it has a greater ability to sustain high yields of photoactivation at moderate [Ca²⁺] compared to ∆psbO (Fig. 4). Chemical crosslinking indicates that Psb27 docks to the outer face (distal to the Mn-assembly site) of the E-loop of CP43 and exerts its effects, including the weakened binding of extrinsic proteins, allosterically (36), and the strength of this interaction changes during the proteolytic processing of the C terminus of the D1 protein (40). This suggests that the Psb27–E-loop interaction stabilizes a structural arrangement that 1) enhances the selectivity of the second Mn²⁺ photooxidation site vis-á-vis the competitive binding of Ca²⁺ at the SMS, 2) stabilizes the binding of Ca²⁺ to the CAS, and 3) maintains an open configuration that enables rapid rebinding of Ca²⁺ if the ion is lost from the Ca-effector site and/or facilitates the exchange of metals in malformed metal centers.

What Is the Dark Rearrangement?

The utilization of single turnover flashes during the flash interval experiment ensures that the dark rearrangement (B➝C) estimates the rate, k_A, of a molecular process proposed to be a conformational change (22, 38, 41, 52) or the relocation of a bound ion (7, 46). Once the rearrangement has occurred, the still labile configuration can be stabilized by a second quantum causing the photooxidation of the second Mn²⁺ at the SMS, to produce the first stable intermediate, “D” (Fig. 1). Based upon the observation that Psb27 allosterically modulates the assembly of the Mn₄CaO₅, we suppose that the mobility of the E-loop is part of the dark rearrangement process. The E-loop is a globular domain situated between transmembrane helices 5 and 6 of CP43 and directly interacts with the assembled Mn₄CaO₅, including a bridging carboxylate ligand to Mn2 and Mn3 via CP43-Glu354 (5). Moreover, it directly contacts the C-terminal domain of the D1-protein, which contains amino acids coordinating the Mn₄CaO₅ and notably provides a ligand to the Ca²⁺ and Mn1 via the carboxyl group of its C terminus. Thus, movements of the E-loop during photoactivation are likely coupled to movements of the C-terminal domain, and vice versa (18). A recent cryo-EM structure resolution of apo-PSII shows alternative structural arrangements of the apo-WOC, including the E-loop displaced away from the HAS, which is itself rearranged compared the coordination environment of the assembled Mn₄CaO₅ (53), and a recent time-resolved AFM study has demonstrated that the E-loop undergoes hinged stochastic fluctuations (54). The increased accessibility in ∆psbO cannot be explained by steric covering of the apo-WOC, and instead, the open configuration must relate to the fact that PsbO forms a structural bridge between the E-loop and other parts of the WOC including the D1 C-terminal domain (SI Appendix, Fig. S1A). Thus, the more open configuration due to the loss of PsbO is likely due to enhanced mobility of the WOC. The position of the E-loop may be stabilized by Psb27, so that it still may fluctuate, but in a range that optimizes assembly, in contrast to the loss of PsbO, where the more open configuration is evident, but not the other kinetic features, which are absent. However, the predicted mobility of the apo-WOC alone does not explain the conformational rearrangement because the rate, k_A, slows down rather than speeds up in the mutants, and it is strongly affected by [Ca²⁺]. So, while conformational mobility of the E-loop and the associated D1-carboxyl terminus is a reasonable explanation for the more open structure, how excess Ca²⁺ would also slow the rearrangement process is not obvious. This is where the SMS may factor in: if the dark rearrangement corresponds to conformational fluctuations alternating between the open and closed configurations, then a specific closed conformation may be necessary for Mn²⁺ ligation and/or photooxidation at the SMS. Indeed, the actual formation SMS ligation environment may be contingent upon the rearrangement. Furthermore, if the specific closed conformation is only infrequently realized during the course of conformational fluctuations, as suggested by the dwell times of the open and closed conformations (54), then competitive occupation of the SMS by Ca²⁺ would delay the time until Mn²⁺ reoccupies the SMS and the specific conformation is revisited during the course of the conformational fluctuations. In this model, the rearrangement rate reflects the frequency of occurrence that intermediate, “B”-containing Mn³⁺ has conformationally rearranged, the intrinsic protein conformational rearrangement time, and the SMS site is actually occupied by a Mn²⁺ ion (Fig. 6).

Fig. 6.

Schematic model of the three main metal-binding sites discussed and a minimal model of photoactivation with proposed conformational change and ion exchange accounting for the results. Brackets around individual states A, B, and C indicate alternate protein conformations with similar bound ion states (blue text in figure). Dark decay (k_D) of states B and C to A indicated with arrows with dotted lines. Dark rearrangement (B➞C, solid arrow below) is proposed to include ion binding and conformational changes of the CP43 e-loop and D1-carboxyterminal domain. Only the productive pathway involving binding of Mn²⁺ to the SMS indicated by C^SMS:Mn during dark rearrangement is shown, but the not the nonproductive Ca²⁺ binding at the SMS (SI Appendix, Fig. S1).

In conclusion, the initial stages of photoactivation appear to involve three critical metal-binding sites: two Mn (HAS, SMS) and one Ca (CAS). These sites are utilized for photooxidative metal cluster formation in a process that involves structural rearrangements of the protein scaffolding that templates the assembly. Occupation of the CAS by Ca²⁺ is critical for stabilizing assembly intermediates during the dark rearrangement (B➝C). Incorrect occupation of the CAS by Mn²⁺, e.g., at suboptimal Ca/Mn ratios, results in photochemical misses and a photoinactivation process that probably corresponds to the Mn²⁺ photooxidation pathway leading of high-valence, but misassembled Mn (22). The long-observed competition between Mn²⁺ and Ca²⁺ appears to occur at the SMS, and the occupation of this site by competing Ca²⁺ ions slows the dark rearrangement. The results are most easily rationalized in a model where the dark rearrangement involves the conformational mobility of the WOC domain, which fluctuates in the absence of the Mn₄CaO₅ (54) (Fig. 6). In this model, the trapping of the first stable state “D” occurs when three conditions are met simultaneously: 1) the rearranging WOC assumes an optimal structural geometry, 2) the metal-binding sites are appropriately occupied, and 3) photochemistry generates an oxidant to oxidize the second Mn²⁺ (C⇒D). According to this model, the accessory protein Psb27 allosterically constrains structural fluctuations in the apo-WOC in a way that optimizes the assembly by stabilizing the intermediates and the promoting productive occupancy of the metal-binding sites by their cognate ions.

Materials and Methods

Strains and Growth Conditions.

The glucose-tolerant Synechocystis sp. PCC6803 control strain (WT control) expressing only the WT psbA2 gene and having a hexa-histidine tag fused to the carboxyl-terminus of CP47 was maintained in BG-11 medium, as described previously (55). Experimental cultures were grown in flat 1L tissue culture flasks in 800 mL BG-11 media buffered with 10 mM Hepes-NaOH pH 8.0 (HBG-11) supplemented with 5 mM glucose (Sigma) under a PFD (photon flux density) of ∼80 µmol m⁻² s⁻¹ at 30 °C. Cultures were bubbled with filter-sterilized air. Light intensity measurements were made with a Walz light meter (Germany).

Mutant Strain Generation.

The ∆psbO deletion mutant was constructed previously and involved replacement of the psbO coding sequence with an Sp^r/Sm^r antibiotic resistance gene (42). For overexpression of Psb27 gene in Synechocystis, the chromosomal locus comprising the ORF slr1645 (Psb27) was amplified by PCR (SI Appendix, Table S2) using Herculase II Phusion DNA polymerase (Agilent) and integrated at a neutral site within the Synechocystis genome between ORFs slr1169 and slr1285 (SI Appendix, Fig. S12). Transformation was carried as described previously using selective agar plates containing 5 mM glucose, 10 µM DCMU, and 12.5 µg/mL spectinomycin followed by further selection at 25 µg/mL spectinomycin (56).

Preparation of Mn-Depleted Thylakoid Membranes.

Cultures (2.4 L) were harvested in early stationary phase (OD_750nm ∼1.8–2.2) and exhibited variable Chl fluorescence ([F_m − F₀]/F₀) values >0.5, as measured with a Photon Systems Instruments (PSI) Fluorometer FL 3500. Cells were collected by centrifugation at 25 °C at 6,000 g (Sorvall, F-9 rotor) for 15 min and gently resuspended with a minimal volume of H20BG-11(same as growth media, but with 20 mM Hepes-NaOH pH 8.0) using a paintbrush. The cell suspension volume was expanded to ∼200 mL with additional H20BG-11. The washed cells were centrifuged again at 25 °C at 10,200 g for 5 min (Sorvall, F-14 rotor). The pelleted cells were resuspended as before in H20BG-11 medium, and the suspension was adjusted to a Chl concentration of 100 μg mL⁻¹ Chl. Hydroxylamine was added to 1 mM from a freshly prepared 400 mM stock, and the treated suspension was incubated for 12 min in the darkness with rotary agitation (200 rpm) at room temperature. Maintaining complete darkness, the HA-extracted cells were then washed by resuspension with 100 mL H20BG-11 and rotary agitation (200 rpm) for 5 min before pelleting again. This washing step was repeated four more times with the aim of depleting residual HA. Finally, the cells were resuspended in 120 mL of breaking buffer (1.2 M betaine, 50 mM MES-NaOH [pH 6.0], 10% [vol/vol] glycerol, and 5 mM MgCl₂) that was prepared with ultrapure reagents and Chelex-100 (Bio-Rad)–treated solutions and incubated in the dark on ice for 1 h. Cells were pelleted at 10,200 g for 5 min and resuspended to a total volume of 14 mL. Prior to cell breakage, 1 mM benzamidine, 1 mM ε-amino–n-caproic acid, 1 mM phenylmethylsulfonyl fluoride, and 0.05 mg/mL DNase I were added. Cells were broken by four cycles of 5 s ON and 5 min OFF in an ice-cooled glass bead homogenizer (Bead-Beater, BioSpecProducts, USA). After breakage the sample was centrifuged at 3,600 g for 10 min to pellet unbroken cells and cell debris. Thylakoids were obtained from the supernatant cell homogenate by ultracentrifugation (35 min at 40,000 rpm in a Beckman Ti45 rotor), and thylakoid-containing pellets were suspended in Chelex-100–treated breaking buffer to a concentration of 1.0–1.5 mg of Chl mL⁻¹. Concentrated thylakoid membranes were flash-frozen as 100 µL aliquots in liquid nitrogen and stored at −80 °C.

Photoactivation of HA-Extracted Membranes.

HA-extracted membranes were photoactivated either in suspension for subsequent assay for restoration of O₂ evolution detected using a Clark-type electrode or directly on a bare platinum electrode that permits the centrifugal deposition of samples upon the electrode surface (16). Flash illumination under each mode of photoactivation was provided using an EG&G xenon flash lamp. The buffer was supplemented with various concentrations of cations CaCl₂, MnCl₂, SrCl₂, and MgCl₂ to examine their role in photoactivation for the role of cations on photoactivation. See SI Appendix, Supplementary Materials for details for data collection and data processing.

Data Availability.

All study data are included in the article and supporting information.