Photosynthetic water oxidation at elevated dioxygen partial pressure monitored by time-resolved X-ray absorption measurements

Abstract

The atmospheric dioxygen (O₂) is produced at a tetramanganese complex bound to the proteins of photosystem II (PSII). To investigate product inhibition at elevated oxygen partial pressure (pO₂ ranging from 0.2 to 16 bar), we monitored specifically the redox reactions of the Mn complex in its catalytic S-state cycle by rapid-scan and time-resolved X-ray absorption near-edge spectroscopy (XANES) at the Mn K-edge. By using a pressure cell for X-ray measurements after laser-flash excitation of PSII particles, we found a clear pO₂ influence on the redox reactions of the Mn complex, with a similar half-effect pressure as determined (2–3 bar). However, XANES spectra and the time courses of the X-ray fluorescence collected with microsecond resolution suggested that the O₂ evolution transition itself (S₃⇒S₀+O₂) was not affected. Additional (nonstandard) oxidation of the Mn complex at high pO₂ explains our experimental findings more readily. Our results suggest that photosynthesis at ambient conditions is not limited by product inhibition of the O₂ formation step.

In oxygenic photosynthesis, plants, algae, and cyanobacteria facilitate the primary biomass formation by exploiting solar energy for driving the conversion of water (H₂O) and carbon dioxide (CO₂) into carbohydrates (C_nH_2nO_n). Dioxygen (O₂) is formed as product of the water oxidation chemistry of the photosystem II (PSII) protein complex (1–5). Photosynthetic CO₂ fixation and O₂ formation have shaped the Earth's atmosphere (6) by (i) lowering the CO₂ level to <0.04% and (ii) creation of a maximal O₂ concentration of 35% (7), ≈300 million years before the present and a current level of ≈20%.

In photosynthetic organisms, the low level of atmospheric CO₂ can limit photosynthesis, related to the activity of CO₂-fixing ribulose bisphosphate carboxylase (8, 9). It is less clear whether water oxidation by PSII is directly affected by the concentration of O₂, the immediate product of light-driven water splitting. This question was only recently addressed experimentally by Clausen and Junge (10). They exposed PSII from cyanobacteria to oxygen partial pressures (pO₂) ranging from 0.2 to 30 bar and, analyzing flash-induced near-UV absorption changes, discovered a pO₂ effect with a half-saturation pressure of only ≈2 bar (10). Subsequently, in measurements of delayed Chl fluorescence, together with Clausen and Junge we confirmed that also the donor side of the plant PSII in its native membrane environment is affected by elevated pO₂ (11). The surprisingly low half-pressure of only 2–3 bar was taken as an indication that the efficiency of the O₂ formation step could be reduced significantly by product inhibition. Such a limitation of the water oxidation chemistry may have imposed restraints on the evolution of oxygenic photosynthesis (10). Notably, the pO₂ inside of photosynthetic cells in microbial mats and bioreactors can be severalfold higher than the atmospheric level of 0.2 bar (12). Thus, inhibition of O₂ formation already at moderate pO₂ could limit photosynthetic productivity in present ecosystems and future biotechnological applications.

The catalyst of photosynthetic water oxidation is a Mn₄Ca complex bound to PSII (1–5). Tentative atomic models of the complex resulted from crystallographic and X-ray absorption data (13, 14). Frequently, the Mn complex, its ligand environment, and a nearby redox-active tyrosine residue [Y_Z, Tyr-160/161 of the D1 protein (15)] are denoted as oxygen-evolving complex (OEC). The paradigmatic functional model for discussion of the redox reactions of the OEC has been proposed already in 1970 by Kok et al. (15), accounting for the period-of-four pattern in the flash-number dependence of O₂ formation first observed by Joliot and Joliot (16). The Kok cycle involves the states S₀–S₄; the subscript indicates the number of oxidizing equivalents accumulated by the OEC (15, 17). The OEC can be driven synchronously through the S-state cycle by the application of saturating flashes of visible light. After dark adaptation, the S₁-state prevails, and each light flash induces a specific transition: 1st flash, S₁→S₂; 2nd flash, S₂→S₃; 3rd flash, S₃→S₄→S₀ + O₂; 4th flash, S₀→S₁. Increasing desynchronization of the S-state population is the result of so-called miss events caused by competition between forward electron transfer and charge recombination processes (18–21).

The chemistry of photosynthetic water oxidation still is not completely understood (1, 5). Four semistable intermediate states of the reaction cycle (S₀, S₁, S₂, S₃) are well characterized. Identification of further, transiently formed reaction intermediates may pave the road toward a more complete understanding of the reaction mechanism. Recently, we have shown that formation of the S₄-state involves a reaction intermediate that likely is formed by a deprotonation step (22, 23). These and other findings facilitated an extension of the classical S-state cycle model (17, 24). On the basis of their O₂ pressure experiments, Clausen and Junge have proposed that in the S₄→S₀ transition a reaction intermediate is formed where water is partially oxidized to the peroxide level, and the Mn complex concomitantly is reduced to a level equivalent to the S₂-state (S₂*, see Fig. 1). The proposed S₂*-state differs from the intermediate described in ref. 22, as clarified elsewhere (24–26).

Fig. 1.

Hypothetical energy-level diagram for a thermodynamic O₂ backpressure effect on the S₄→S₀ transition. The Gibbs free energy, G_P, of the final product state (S₀ + O₂) increases by ≈60 meV per 10-fold pO₂ increase. At pO₂ > 10 bar, the energy of a peroxidic reaction intermediate was predicted to be higher than G_P, implying almost complete inhibition of O₂ formation (10). The energy of the S₂* reaction intermediate has been proposed to be above (10) or below (11) the initial Y_Z^oxS₃-state. In the present work, validity of the above scheme is scrutinized and eventually rejected.

The mechanistic importance of a peroxidic reaction intermediate prompted us to investigate the influence of an elevated pO₂ on the redox reactions by rapid-scan and time-resolved X-ray absorption spectroscopy (XAS) at the Mn K-edge (22, 27–29), where the spectral changes exclusively reflect structure and oxidation state changes of the Mn complex of PSII (30–32).

Results

X-Ray Photoreduction at Elevated pO₂.

For X-ray absorption measurements (at 20 ± 1 °C), the PSII samples were placed in a cell sustaining O₂ pressures up to 20 bar [supporting information (SI) Materials and Methods and Fig. S1], and the S-state transitions of the Mn complex were induced by nanosecond laser flashes. Modification of the oxidation state of the Mn complex by X-ray irradiation (33–35) was studied at various pO₂ levels. No specific O₂ effect on X-ray photoreduction was found (SI Materials and Methods and Fig. S2). For all of the following experiments, the X-ray intensities were adjusted such that the Mn oxidation state remained unchanged during measurements (negligible Mn photoreduction).

X-Ray Absorption Near-Edge Structure (XANES) Spectra at Atmospheric and Elevated pO₂.

XANES spectra were recorded immediately after excitation of PSII samples in the pressure cell by 0–6 laser flashes (Fig. 2 A and B). At ambient pO₂ (0.2 bar), similar shifts of the K-edge to higher energies on flashes 1 and 2 were detected as previously (32, 36). They are attributable to oxidation of the Mn complex in the S₁→S₂ and S₂→S₃ transitions. After the 3rd flash, the edge shifted to lower energies because of Mn reduction by electrons from water in the O₂ evolution step (S₃→S₄→S₀) (Fig. 2A). The corresponding edge energies are depicted in Fig. 3A (filled circles). At a pO₂ of 11 bar, shifts of the K-edge energy also were observed (Fig. 2B), but at increasing flash numbers their magnitudes differed from those at ambient pO₂, suggesting that the redox chemistry of the Mn complex was altered by an elevated pO₂ (Figs. 2B and 3B).

Fig. 2.

XANES spectra of the Mn complex measured after the application of 0–4 laser flashes to PSII samples at pO₂ of 0.2 bar (A) and 11 bar (B). Spectra are the average of 2 (0.2 bar) and 4 (11 bar) sets of independently measured spectra. (Insets) Expanded half-height region of the K-edges. (C) Calculated spectra of pure S-states (color code as indicated), obtained by deconvolution of the spectra in A and B using parameters listed in Table 1 [lines, 0.2 bar; triangles, 11 bar O₂ (a); dots, 11 bar O₂ (b)] and in the legend of Fig. 3. (Inset) Corresponding difference spectra.

Fig. 3.

Mn K-edge energies (filled circles) determined from spectra as in Fig. 2 (A, 0.2 bar; B, 11 bar O₂). Error bars represent SD values from 2 (at 0.2 bar) and 4 (at 11 bar) datasets. The lines were obtained by simulation of the flash patterns using the parameters listed in Table 1 [in B: solid line (a), dashed line (b)]. An S₁-state edge energy of 6,551.5 eV and edge shifts as determined in (27, 28) were used in the simulations. (C and D) Filled circles represent the magnitude of the millisecond rise in the X-ray absorption at 6,552.5 eV (Fig. 4) caused by Mn reduction in the S₃→S₄→S₀ transition. Edge energy estimates derived from the X-ray transients in Fig. 4 are shown for comparison in A and B (open triangles, assuming direct proportionality between edge shift and X-ray fluorescence change for excitation at 6,552.5 eV; normalization to equal first-flash response). Lines in C and D show simulations with parameters in Table 1 [in D: (a) solid line, (b) dashed line].

Time-Resolved X-Ray Experiments at elevated pO₂.

The changes in the XANES spectra in Fig. 2 merely reflect the net effect of Mn oxidation and reduction because by illumination with a sequence of Laser flashes, the Mn complex is not driven fully synchronously through its reaction cycle. Among other things, because of incomplete turnover of PSII (miss events), “S-state scrambling” occurs on higher flash numbers. For example, the 3rd flash induces not only Mn reduction and O₂ formation (S₃→S₄→S₀), but also Mn oxidation in a sizable fraction of the PSII population (mostly S₂→S₃ transition). Time-resolved X-ray absorption measurements on dark-adapted PSII samples facilitate disentangling of the oxidation and reduction processes.

For a sequence of 10 laser flashes, time courses of the X-ray fluorescence (F_X-ray) at constant X-ray excitation energy (6,552.5 eV) were recorded (Fig. 4) (22, 29). A decrease in the fluorescence level after the flash indicates oxidation of the Mn complex, an increase reflects Mn reduction. [We note that the magnitude of ΔF_X-ray is S-state-dependent; for excitation at 6,552.5 eV, it is larger in S₁→S₂ than in S₂→S₃ (36–38).] At 0.2 bar O₂, Mn reduction was absent on flash 1 and small on flash 2, whereas at 13 bar O₂, significant Mn reduction occurred already on flash 2. The average fluorescence level at flash numbers >3 was lower than that at 0.2 bar, suggesting a higher mean K-edge energy, in agreement with the XANES spectra. Furthermore, the K-edge shifts estimated on basis of the X-ray transients in Fig. 4 and from the XANES spectra compare well (Fig. 3, open triangles and filled circles).

Fig. 4.

Time courses of the X-ray fluorescence at pO₂ of 0.2 bar (Left) and 13 bar (Right) as induced by the first 8 of a train of 10 laser flashes applied to dark-adapted PSII samples (excitation energy of 6,552.5 eV). Approximately 500 (0.2 bar) and 1,500 (13 bar) transients were averaged and are displayed at a resolution of 50 μs per data point. The 13 bar transients were obtained by averaging measurements for pO₂ values ranging from 11 to 16 bar. Data were normalized such that the amplitude of the response to the 1st flash was similar for the 0.2 and the 13 bar datasets. Smooth lines represent simulations by a biexponential function using a halftime of 1.2 ms to account for Mn reduction (positively directed changes) and variable halftimes for Mn oxidation (negatively directed signals). For selected transients the contributions from Mn oxidation (open arrows) and Mn reduction (filled arrows) are indicated.

It has been suggested that at elevated pO₂, the 3rd flash does not induce a 4-electron reduction of the Mn complex, but a 2-electron reduction resulting in a peroxidic intermediate (S₂*) (10, 11). According to this rationale, a diminished amplitude of the Mn reduction transient on flash 3 may be expected at high pO₂. However, as clearly visible in Fig. 4 Right, flash 3, the magnitude of the Mn reduction (millisecond increase marked by a filled arrow) was not diminished at 13 bar O₂, suggesting that rather than a 2-electron reduction, the usual 4-electron reduction of Mn was induced by the 3rd flash. Furthermore, the X-ray transients show that on higher flash numbers also at 13 bar O₂, each flash induces Mn oxidation and reduction; the average magnitude of the flash-induced transients is not smaller than at 0.2 bar (Fig. 5B). Thus, inhibition of the light-triggered redox reactions by accumulation of an S₂* intermediate can be excluded. We feel that the X-ray transients of Fig. 4 cannot be reconciled in a reasonably straightforward way with the proposed S₂* formation in a major fraction of PSII (10, 11).

Fig. 5.

Half-effect O₂ pressure and rate of Mn reduction. (A) Pressure dependence of the O₂ effect on PSII and (B) averaged X-ray transients collected at 0.2 and 13 bar. (A) Amplitudes of the millisecond phases were determined for the X-ray transients induced by the 2nd and 3rd flashes; the ratio of these amplitudes is shown (filled circles; error bars denote SD). The line represents an asymptotic simulation with a pO₂^1/2 of 2.9 bar. (B) Transients represent the average of the transients on the first 10 flashes applied to dark-adapted PSII. Smooth lines denote biexponential simulations. The signal rise associated with Mn reduction in the S₃→S₄→S₀ transition was simulated with an identical halftime of 1.2 ms for transients collected at pO₂ of 0.2 and 13 bar. The initial decrease was approximated by an exponential function with halftimes of Mn oxidation of 105 ± 20 μs (0.2 bar) and 120 ± 20 μs (13 bar).

What causes the differences between the X-ray data collected at atmospheric and elevated O₂ pressures? The time-resolved data indicates, for elevated pO₂, significant Mn reduction and hence O₂ formation already on the 2nd flash (Fig. 4). Simulations on the basis of Kok's S-state model (15, 36) of the data in Fig. 4 were used to characterize and quantify the O₂ effects.

Simulations of the Flash Number Dependence in the X-Ray Data.

Curve fitting of the X-ray transients in Fig. 4 (smooth lines) yielded the amplitudes caused by Mn reduction (t½ ≈1.2 ms). The quaternary oscillations of (i) the millisecond amplitude (Fig. 3 C and D), (ii) the K-edge energy (Fig. 3 A and B), and the complete XANES spectra (Fig. 2C) can be described by similar sets of transition parameters for all 0.2-bar data, on the one hand, and similar parameter sets for all high-pO₂ data (Table 1). The parameters for the 0.2-bar data are close to determined values (36). Two simulation approaches with alternative parameter sets described the high-pO₂ data similarly well (Table 1).

Table 1.

Simulation parameters (in percentage)

	pO2, bar	Initial S0, S1, S2, S3	Miss factor	Double hits	Non-QB centers
XANES spectra	0.2	10, 80, 10, 0	15	0	15
	11(a)	10, 80, 10, 0	7	14	15
	11(b)	10, 60, 30, 0	15	55	15
X-ray time scans	0.2	0, 90, 10, 0	12	0	10
	13(a)	0, 90, 10, 0	9	14	10
	13(b)	0, 60, 40, 0	12	35	10

The simulation results are shown as lines in Fig. 3. Values revealing pronounced changes at elevated pO₂ are marked bold. (a) Uniform double hit probability on all four S-transitions. (b) Apparent double hits only on the S₀→S₁ transition.

“Apparent” double-hits on all S-state transitions (Fig. 3 B and D; solid lines).

The classical double-hits (15) were excluded because of the use of nanosecond laser flashes. An option to rationalize an apparent double-hit at high pO₂ is rapid formation of reactive oxygen species (ROS) (39–41) that then oxidize the Mn complex. The comparably large magnitude of Mn reduction in the time-resolved X-ray data (Fig. 4; sum of reduction phases on the first 4 flashes) and the large edge-shift on flash 1 (S₁→S₂) in the K-edge spectra (Fig. 2) are in agreement with surplus flash-induced Mn oxidation.

Increased initial S₂ population in the samples plus a large apparent double-hit probability in the S₀→S₁ transition only (Fig. 3 B and D; dashed lines).

This simulation approach could correspond to slow dark oxidation of the S₁-state (before flash 1) and more rapid dark oxidation of the S₀-state (between flashes 3 and 4). To assess S₂ formation by dark oxidation of the S₁-state (at high pO₂), we used EPR spectroscopy. In dark-adapted PSII samples that had been exposed to 15 bar O₂ for 30 min, the typical EPR multiline signal of the S₂-state (42) was not detectable (Fig. S3). Thus, dark oxidation resulting in formation of the normal S₂-state can be excluded (but not the generation of an EPR-silent modified S₂-state).

Assuming the absence of any specific influence of elevated pO₂ on the S₃→S₄→S₀ transition and by using the altered parameters discussed above, the XANES spectra of the pure S-states were calculated. The resulting high-pO₂ spectra, within the precision of the experiment, are identical to the low-pO₂ spectra of the pure S-states (Fig. 2C). Accordingly, the XANES spectra at high pO₂ can be explained without invoking any S₂* formation.

We emphasize that all simulations involving S₂* formation in a sizeable fraction of PSII (≥10%), followed either by an S₂*→S₂ or S₂*→S₀ transition in the dark or a blockage in the S₂*-state, consistently failed to describe the experimental results (data not shown). The above simulations support the conclusion already drawn from visual inspection of the X-ray transients of Fig. 4, namely the absence of a high-pO₂ influence on the S₃→S₄→S₀ transition itself. The high-pO₂ effect detected here is explainable by additional oxidations of the Mn complex that are either light-induced (1st simulation approach) or proceed in the dark (2nd approach).

O₂ Half-Saturation Pressure.

Time-resolved experiments were performed at pO₂ values ranging from 0.2 to 16 bar (time courses not shown), and the millisecond amplitudes of Mn reduction were determined for each flash. Then, the ratio of millisecond amplitudes on flashes 2 and 3 was calculated (F3/F2; Fig. 5A) as a relative measure of the alterations in the S-state distribution. The pronounced decrease in F3/F2 with increasing pO₂ was described by a pO₂^1/2 of 2.9 bar (Fig. 5A, line); saturation occurred at ≈10 bar. Thus, the pO₂ of ≥10 bar in the XAS measurements discussed above was sufficient to saturate the O₂ effects. A similar pO₂^1/2 value has been observed in previous O₂ experiments on different types of PSII samples (10, 11).

Halftimes of Mn Oxidation and Reduction.

The halftimes of Mn oxidation and reduction were determined from the time-resolved X-ray data (Fig. 5B). To improve the signal-to-noise ratio, the transients of the first 10 flashes (for flashes 1–8, see Fig. 4) were averaged and subjected to a biexponential simulation (lines). The amplitude of the oxidation–reduction transient is slightly increased at high pO₂, possibly explainable by additional, nonstandard oxidations of the Mn complex shortly after flash excitation (model a in Table 1). By using at 13 bar and 0.2 bar an identical halftime of Mn reduction of 1.2 ± 0.1 ms, a value that is typical for PSII (10, 11), convincing simulations were obtained. Furthermore, an apparent mean halftime of Mn oxidation of ≈110 μs (Fig. 5B, see legend) was derived, which also was identical at 0.2 and 13 bar pO₂ (within the error range). In conclusion, the X-ray transients suggest unchanged halftimes of Mn reduction in the O₂ formation step and of Mn oxidation at high pO₂.

Discussion

Already at moderately elevated pO₂ (half-effect pressure of 2–3 bar), the redox chemistry of the PSII manganese complex is modified, as found in refs. 10 and 11. However, the X-ray absorption data do not indicate inhibition of the oxygen evolution transition (i.e., S₃→S₄→S₀+O₂). Specifically, the formation of a partially reduced state of the Mn complex resembling the S₂-state (S₂*) in a major fraction of PSII (5) cannot be reconciled with the experimental findings of the present investigation. In conclusion, the thermodynamic product inhibition illustrated in Fig. 1 is not observed.

The high-pO₂ effects in the X-ray data can be rationalized by Mn oxidation by ROS formed upon flash illumination (39–41, 43) and/or dark oxidation of the S₀ and S₁-state, the 2 low-potential S-states (5). Either option is sufficient to model adequately the X-ray data. For additional flash-induced oxidations, a plausible mechanistic scenario is that oversaturating laser flashes at high pO₂ result in especially pronounced chlorophyll triplet (³Chl) and singlet oxygen (¹O₂) formation (39, 43). The high-oxidation potential of ¹O₂, its extreme reactivity, and a diffusion length on the order of 100 Å could facilitate direct or indirect oxidation of the Mn complex, possibly resulting in superoxide formation (39–41).

The O₂/H₂O midpoint potential (E_m) at elevated pO₂ (≈0.95 V at 10 bar and pH 6.2) is close to the estimated E_m of the Mn complex in its S₁- and S₀-states, but likely too low for dark oxidation of the S₂- and S₃-states (5). Thus, dark oxidation of the lower S-states may be thermodynamically feasible at high pO₂; it could be indirect, involving other cofactors and/or ROS. Our finding that high-pO₂ exposure does not result in formation of the typical S₂-state EPR signal disfavors dark oxidation of S₁ to a normal S₂-state. However, formation of a modified S₂-state cannot be excluded. Further studies are required to clarify the mode of O₂ interaction with the Mn complex.

Clausen and Junge (10) have discovered the pO₂ influence on the redox reactions of the PSII Mn complex by means of analysis of flash-induced UV absorption changes. Subsequently we investigated, in cooperation with Clausen and Junge, the pO₂ effect by time-resolved measurements of delayed chlorophyll fluorescence (DF) (11). In both studies it was suggested that at saturating pO₂ (>10 bar), dioxygen formation is blocked because of formation of a peroxidic S₂* intermediate (10). Our present results rather exclude S₂* formation in a sizeable fraction of PSII. The 2 spectroscopic methods used in refs. 10 and 11 do not reflect specifically the oxidation state changes of the Mn complex. The resulting ambiguity in combination with a modeling approach that did not include the possibility of additional, nonstandard oxidations may explain the diverging conclusions on the pO₂ effect in refs. 10 and 11.

Differences in the results of the present and previous studies also could originate from unequal experimental conditions. Specifically, the reactions of ROS with PSII may have been different in the current and previous investigations. They could depend on experimental details such as sample thickness (here only ≈300 μm, thus possibly particularly rapid pO₂ equilibration) and PSII concentration (here much higher than in the solution studies). Whether the thermodynamics of the water oxidation cycle of the intact Mn complex itself depends critically on these experimental details is still unclear but, in our opinion, unlikely.

Even at a pO₂ >10 bar, we have not obtained any indication for O₂ product inhibition of the S₃→S₄→S₀ transition itself. An increase in pO₂ from 0.2 to 10 bar corresponds to an increase in G_P, the Gibbs free energy of the product state (Y_Z⁰S₀+O₂), by

We conservatively estimate that any inhibition of the S₃→S₀ transition that affected >30% of the PSII would have been resolvable. This implies that also at pO₂ >10 bar, the free-energy level, G_I, of an intermediate in the S₃→S₄→S₀ transition lies above G_P by a value that corresponds to an equilibrium constant of ≈3. Thus, we obtain for ΔG_I−P (= G_I − G_P):

In conclusion, we estimate that the free energy of the product state (Y_Z⁰S₀+O₂ at pH 6.2) is ≥130 meV below the free-energy level of any intermediate in the S₃→S₄→S₀ transition.

The basic reaction cycle of the Mn complex discussed elsewhere (17, 24) implies that O₂ formation in the S₄→S₀ transition is associated with the release of 1 proton. Consequently, the free energy of the product state is predicted to rise upon acidification by ≈60 meV per pH unit. In intact organisms, the pH in the thylakoid lumen drops upon illumination to values between 5 and 6 (44). At pH 5, G_P thus may be by ≈70 meV higher than at pH 6.2, resulting in a decrease of the minimal ΔG_I−P to ≈60 meV. This figure implies that at atmospheric pO₂ even in the presence of a luminal acidification, a sizable O₂ product inhibition is unlikely.

Our results suggest that the rate of photosynthetic water oxidation is not seriously limited by thermodynamic product inhibition of the O₂ evolution transition. Growth limitations at elevated pO₂ may root in diverse oxidative processes (40) but presumably not in inhibition of the O₂ formation step. It remains an open question whether the insensitivity of the O₂ formation step to product inhibition reported here has always been a feature of PSII or has evolved concomitantly with the increase in the atmospheric pO₂.

Materials and Methods

XAS samples of high-activity PSII membrane particles from spinach (45) were prepared by pipetting PSII suspensions (10 μL, ≈12 mg of Chl mL⁻¹, containing 100 μM phenyl-p-benzoquinone (PPBQ) as an electron acceptor) into each of the 20 cavities of the sample holders (29) to be inserted into the pressure cell. Samples were partially dehydrated (loss of ≈20% of the initial water content) by drying in a desiccator for 1 h over silica gel at 0.5 bar of air pressure at 4 °C and then frozen in liquid nitrogen. Compared with previous studies (22, 36), this procedure resulted in a higher water content of the samples ensuring (i) that prolonged exposure to an atmosphere of dry O₂ gas did not cause any dehydration detrimental to sample integrity and (ii) an almost isotropic orientation of the PSII membrane particles so that magic-angle excitation was not required and XAS could be performed at an excitation angle of 45°. To maximize the S₁-state content, each sample was illuminated by a single saturating flash and then dark-adapted for 1 h during the drying period. Approximately 300 stripes with 20 sample cavities each were prepared (≈6,000 PSII samples in total).

X-ray experiments at the Mn K-edge were carried out at the undulator beamline ID26 at the European Synchrotron Radiation Facility at Grenoble, France, employing developed techniques for time-resolved XAS measurements (22, 29). The 3 undulators of the beamline were adjusted such that an approximately flat intensity profile of the exciting X-ray beam resulted in the region of the Mn K-edge (maximal X-ray flux ≈10¹³ photons s⁻¹, spot size 1 × 1.5 mm², attenuated by stacked Al foils whenever required). A newly constructed pressure cell sustaining O₂ pressures up to 20 bar was used (see SI Materials and Methods). For measurements at elevated pO₂, (1) the cell was flushed (cap slightly unscrewed) for 10 s with O₂ gas of highest purity (2). The cap of the cell was closed, and the pressure was adjusted to the desired value for 1 min (3). The pressure was slowly released and then again adjusted to the desired level (4). Samples were incubated at room temperature for 10–20 min before measurements. After preincubation, the duration of measurements on the 20 samples of 1 holder was ≈10 min (total time of samples under pressure of 20–30 min). For measurements with control samples (ambient air atmosphere), an identical timing protocol was used.

PSII samples residing in the pressure cell were excited by a laser (Continuum Inlite-II, 532 nm, pulses of 5 ns and 50 mJ spaced by 700 ms); neighboring samples were shielded by a shutter. The X-ray fluorescence was detected by a large-area scintillation detector shielded against scattered X-rays by 10 μm of Cr foil and visible light by 2 μm of Al foil.

XANES spectra at the Mn K-edge were recorded within 1 s (irradiation time limited by a shutter, τ_open < 1 ms) by using the rapid-scan mode of ID26 (27, 28). Samples were illuminated by 0–6 laser flashes; last flash ≈200 ms before the start of the monochromator scan. For a single XANES spectrum, ≈60 scans were averaged, each collected on a fresh sample. Spectra were normalized to an amplitude of 1.28 at 6,564 eV (22, 29).

In all time-resolved X-ray experiments with microsecond resolution, fluorescence transients were recorded for excitation at 6,552.5 eV. The samples were irradiated by the X-rays within a time interval of 5 ms before and 15 ms after each laser flash. For 1 transient, 500–1,000 measurements were collected, each on a fresh PSII sample, and averaged.

XANES spectra collected after application of 0–3 laser flashes were corrected for imperfect S-state advancement in the S-state cycle to yield the spectra of the pure S-states (36). For this deconvolution (and all simulations referred to in Results), we used a specific EXCEL spreadsheet that allowed for the joint simulation (least-squares curve-fitting) of the experimentally determined flash-number dependence of the K-edge energy (from XANES spectra) and the amplitude of the millisecond reduction phase of Mn in the O₂-evolving step (from exponential simulations of the time-resolved X-ray data).