Assessment of the manganese cluster’s oxidation state via photoactivation of photosystem II microcrystals

Significance

Photosynthetic water oxidation by the multisubunit membrane protein complex PSII is an important process that sustains all aerobic life on Earth by producing molecular oxygen from sunlight and water. Understanding the mechanism of this process is crucial toward advancing fundamental knowledge as well as providing a blueprint for the development of solar fuel devices. Important pieces of information required for solving the mechanism of biological water oxidation are the oxidation states of the manganese ions forming the catalytic site of water oxidation in PSII. Here, we resolve a long-standing controversy between 2 competing schools of thought, by providing a clear-cut determination of overall manganese oxidation states using a simple counting experiment.

Abstract

Knowledge of the manganese oxidation states of the oxygen-evolving Mn₄CaO₅ cluster in photosystem II (PSII) is crucial toward understanding the mechanism of biological water oxidation. There is a 4 decade long debate on this topic that historically originates from the observation of a multiline electron paramagnetic resonance (EPR) signal with effective total spin of S = 1/2 in the singly oxidized S₂ state of this cluster. This signal implies an overall oxidation state of either Mn(III)₃Mn(IV) or Mn(III)Mn(IV)₃ for the S₂ state. These 2 competing assignments are commonly known as “low oxidation (LO)” and “high oxidation (HO)” models of the Mn₄CaO₅ cluster. Recent advanced EPR and Mn K-edge X-ray spectroscopy studies converge upon the HO model. However, doubts about these assignments have been voiced, fueled especially by studies counting the number of flash-driven electron removals required for the assembly of an active Mn₄CaO₅ cluster starting from Mn(II) and Mn-free PSII. This process, known as photoactivation, appeared to support the LO model since the first oxygen is reported to evolve already after 7 flashes. In this study, we improved the quantum yield and sensitivity of the photoactivation experiment by employing PSII microcrystals that retained all protein subunits after complete manganese removal and by oxygen detection via a custom built thin-layer cell connected to a membrane inlet mass spectrometer. We demonstrate that 9 flashes by a nanosecond laser are required for the production of the first oxygen, which proves that the HO model provides the correct description of the Mn₄CaO₅ cluster’s oxidation states.

Photosynthesis is essential for life as it not only stores solar energy within chemical bonds, but also provides the molecular oxygen required for unlocking the stored energy with high efficiency. The photosynthetic reaction sequence begins with the light-driven oxidation of water that is catalyzed in the biosphere exclusively by PSII. Paradoxically, PSII is also inactivated by light exposure, necessitating a repair process that includes the disassembly and reassembly of the catalytic site for water oxidation in a PSII protein every half hour under full sunlight. Revealing the water oxidation mechanism by PSII, as well as the reassembly process of PSII, is vital for enhancing our knowledge of this key process and may provide inspiration toward the development of artificial water oxidation catalysts needed for the sustainable production of solar fuels, such as hydrogen, ammonia, or methane.

The kinetic scheme describing water oxidation by PSII is known as the Kok cycle (1). It comprises 1 stable (S₁), 3 semistable (S₀, S₂, and S₃), and 1 transient (S₄) intermediate(s) where higher numbers denote successive oxidation of the catalytic center that contains a Mn₄CaO₅ cluster. Recent serial femtosecond X-ray crystallography data provided detailed information about structural changes in the Mn₄CaO₅ cluster as it cycles through the S₀, S₁, S₂, and S₃ states (2, 3), yet the water oxidation mechanism remains controversial (4, 5). One of the long-standing discussion points is the overall oxidation state of the Mn₄CaO₅ cluster (6). Historically, the debate on the Mn oxidation states of the Mn₄CaO₅ cluster arose from the continuous wave X-band electron paramagnetic resonance (EPR) based discovery that the S₂ state gives rise to a multiline signal with an effective total spin of S = 1/2 that has 18–20 hyperfine lines (7, 8). This is a signature of a Mn(III)Mn(IV) mixed valence state within a complex with more than 2 Mn ions, which lead to the overall oxidation state assignments of either Mn(III)₃Mn(IV) or Mn(III)Mn(IV)₃ for the Mn₄CaO₅ cluster (6). As recent results confirm that each S-state advancement corresponds to a manganese centered one-electron oxidation (9), the oxidation states for all S-state intermediates in the Kok cycle can be calculated if one of the states is known. Thus, starting from the 2 options for the S₂ state, 2 competing sets of oxidation state assignment have been proposed and are commonly known as the LO or HO models. Subsequent ⁵⁵Mn electron nuclear double resonance (ENDOR) studies support the assignment of Mn(III)₃Mn(IV) for the S₀ state (10) and of Mn(III)Mn(IV)₃ for S₂ (10, 11). The HO model is, furthermore, consistent with recent ⁵⁵Mn electron–electron double resonance detected NMR (EDNMR) studies of the S₃ state that support the Mn(IV)₄ assignment for this state (12). Nonetheless, there are alternative interpretations of the S = 1/2 EPR multiline signal (13) and a ⁵⁵Mn ENDOR study (14) that supports the LO model.

The oxidation states of the Mn₄CaO₅ cluster have been studied over the past 3 decades by a large number of additional techniques. Mn K-edge X-ray absorption near edge structure (15–17) and Mn Kβ X-ray emission spectroscopy (XES) (9, 15, 18) studies by various groups unanimously support the HO model. Yet these studies are challenged by alternative interpretations (19, 20) that suggest that the available data are better suited toward the LO model.

Beyond spectroscopic methods, the oxidation state of Mn₄CaO₅ can be inferred by an inventory of the redox equivalence between the Mn²⁺ and the functional Mn₄CaO₅ cluster. Titration of reductant to functional Mn₄CaO₅ while monitoring the amount of Mn²⁺ formation (21) or the progressive shift of the first maximum in flash-induced oxygen oscillation patterns allow assessments of the overall oxidation states (22, 23). Most of these data have been interpreted to support the HO model, but the study by Kretschmann et al. (23) supports the LO model.

The converse approach is the photooxidation of exogenously supplied Mn²⁺ by PSII depleted of the Mn₄CaO₅ cluster (apo-PSII) to restore O₂-evolving functionality in a process known as photoactivation (refs. 24–33; for reviews, see refs. 34 and 35). When using single turnover flashes, the photooxidation of Mn is a one-electron oxidation event as seen, for example, in intact dark-adapted (S₁ state) PSII samples where such an illumination leads to a period 4 oscillation with the first O₂ evolution after the third flash (1, 36, 37). Thus, by counting the number of single turnover flashes required to evolve the first O₂ during photoactivation of apo-PSII, the overall oxidation state of the Mn₄CaO₅ cluster can be established. Starting from exogenously supplied Mn²⁺, O₂ evolution after 7 flashes corresponds to the LO model, while O₂ evolution after 9 flashes corresponds to the HO model (see Fig. 1). In contrast to the majority of the results presented above, previous reports of the oxidation state determination via photoactivation support the LO model (33, 35, 38). However, the light flashes used in these studies were not strictly single turnover but rather consisted of trains of short LED pulses. While these flashes, which were employed for increasing the yield of photoactivation to a level that O₂ evolution could be detected, were designed by taking independently determined kinetics of the photoactivation process into account (33), this opens the possibility of multiple oxidation’s per flash. One indication for this may be the unexpected observation of O₂ evolution below the minimum number of 7 light flashes (33).

Fig. 1.

Schematic of photoactivation of PSII based on the (A) high oxidation model and (B) low oxidation model. S₀–S₄ states are intermediate states of the Kok cycle; S₋₅ to S₋₁ states are photoactivation intermediate states. The dark stable S₁ state is labeled in bold. Each double arrow corresponds to a flash-induced one-electron oxidation.

Recently, a crystal structure of a Mn-depleted PSII core complex from Thermosynechococcus elongatus was reported at 2.55 Å resolution that is based on a complete depletion of the Mn₄CaO₅ cluster in PSII crystals via NH₂OH and EDTA treatments (39). Importantly, the resulting apo-PSII crystals retained all 20 protein subunits, including the 3 extrinsic proteins that stabilize the Mn₄CaO₅ cluster. This preparation of highly purified and well-defined apo-PSII crystals present a new opportunity to assess the overall oxidation state of Mn₄CaO₅ by photoactivation. In this paper, we use this preparation in combination with saturating nanosecond (ns) laser pulses that fully exclude multiple oxidation’s of the reaction center’s during one flash. Highly sensitive detection of the evolved O₂ was achieved by a thin-layer membrane inlet cell (Fig. 2) that was connected to an isotope ratio mass spectrometer (IRMS). This unique experimental design enabled us to detect O₂ release after the first S₃ → [S₄] → S₀ transition reached during photoactivation. Consequently, the oxidation state determination did not require any complicated analysis beyond the straightforward count of the number of laser flashes required to observe O₂ evolution.

Fig. 2.

Schematic of the thin-layer membrane inlet mass spectrometer cuvette. The sample is confined within a 6-mm-diameter chamber that has a height of 100 µm.

Results and Discussion

Flash-induced oxygen release patterns (FIOPs) of PSII microcrystals of T. elongatus core preparations were recorded at 20 °C in the presence of the artificial electron acceptor phenyl-p-benzoquinone (PPBQ) employing a thin-layer membrane inlet mass spectrometry (MIMS) cuvette (Fig. 2) and nanoseconds duration green (532 nm) laser flashes. The FIOPs of native PSII microcrystals (blue trace in Fig. 3) exhibited the typical period 4 oscillation of dark-adapted PSII samples with maxima at the third, seventh, and 11th flashes. The damping of this oscillation can be described by a miss parameter of 10%, which demonstrates that the laser light was saturating and that the exogenous electron acceptor allowed sustaining oxygen evolution well beyond the number of flashes needed to determine the overall Mn oxidation state of the oxygen-evolving complex. The excellent oscillation is remarkable given the fact that, due to the slow mass transport of O₂ from the MIMS cuvette into the ion source of the IRMS, the laser flashes were spaced 15 s apart to resolve individual flash-induced oxygen peaks. In this context, we note that the small oxygen yield after the second flash is not due to double oxidation of a minor fraction of PSII centers (double hits) but to the extremely long S₂ and S₃ state lifetimes in the presence of PPBQ that led to an incomplete S₁ state synchronization within the 60 min dark-adaptation period after the preflash employed to oxidize tyrosine D (40–44).

Fig. 3.

(Top) Representative FIOPs of native PSII microcrystals (blue trace) and during the photoactivation of apo-PSII microcrystals using equal flash spacings of 15 s (red trace). The red trace magnified by 150 times; the blue trace offset from zero for clarity. Error bars are 95% confidence interval derived from the fitting procedure and smaller than the marker symbols. (Bottom) Original oxygen signals from the MIMS system during laser flashing of apo-PSII microcrystals incubated in an assembly buffer (red trace); apo-PSII microcrystals without Mn²⁺ and Ca²⁺ (green trace) and a native PSII microcrystal (blue trace). The dashed line: fitted oxygen peaks after each laser flash; the black line: sum of all fitted oxygen peaks. The vertical arrows indicate laser flashes. The red and green traces are magnified by 100 times. The green and blue traces were offset from zero for clarity.

The green trace obtained with Mn-free (Fig. 3) apo-PSII microcrystals in the absence of added Mn²⁺ and Ca²⁺ but with the exogenous electron acceptor PPBQ shows that, in contrast to initial trials with Clark and Joliot type electrodes, the laser pulses did not cause any flash-induced artifacts in the MIMS system. Importantly, no O₂ was produced by apo-PSII in the absence of added Mn and Ca cofactors.

On the basis of these controls, photoactivation experiments were performed by adding 10 mM MnCl₂, 400 mM CaCl₂, 5 mM NaHCO₃, and the same acceptor as above (assembly buffer) to dark-adapted apo-PSII microcrystals (no preflash treatment), which were then flashed in the same way as above. No molecular oxygen was observed in the first 8 flashes, but the ninth flash induced a clearly resolved O₂ signal indicating the successful assembly of the Mn₄CaO₅ cluster in a fraction of the centers (red trace in Fig. 3). The subsequent rise of the flash-induced O₂ signal shows the photoactivation of further PSII centers. It does not display any oscillation, indicating a significant miss factor causing mixing of S states during photoactivation.

The O₂ yield after the ninth flash was ∼0.1% of that obtained from native PSII microcrystals after the third flash, indicating that the S₁ state population was 0.1% after 6 photoactivation flashes. This implies that the average quantum yield of photoactivation for the first 6 flashes is close to 32% (SI Appendix), i.e., higher than the typically reported quantum yield of 1% based on noncrystalline Mn₄CaO₅ depleted PSII preparations (24, 25, 27, 28, 34). This comparison also allows the estimation that the absolute O₂ yield induced by the ninth flash was about 15 nM, i.e., well above the estimated oxygen detection limit of 3 nM in the current experimental setup (SI Appendix, Figs. S1 and S2). We note that the observation of O₂ evolution after 9 flashes can be made consistent with the LO model if the O₂ patterns of photoactivation are shifted by 2 electrons due to the presence of auxiliary electron donors that donate electrons more efficiently to ${\text{Y}}_{\mathrm{Z}}^{\text{ox}}$ or P680⁺ than Mn. The most likely auxiliary electron donors are tyrosine D, Y_D, and cytochrome b 559, cyt b₅₅₉. However, this scenario is considered highly unlikely due to the presence of a high affinity Mn²⁺ binding site at or near the assembled Mn₄CaO₅ cluster involving the D1-A170 and D1-E333 residues (29, 45–47). Transient variable fluorescence (48, 49) and EPR measurements (49, 50) during single flash experiments on Mn depleted PSII samples have demonstrated that electron donation from this specifically bound Mn²⁺ to ${\text{Y}}_{\mathrm{Z}}^{\text{ox}}$ is very efficient. This is expected given the much longer distances (>30 Å) of Y_D or cyt b₅₅₉ as compared to Mn²⁺ at the high affinity binding site (7 Å). Moreover, the rate of Y_Z oxidation by P680⁺ at pH < 7 is, at least, 2 orders of magnitude faster compared to the oxidation of Y_D or cyt b₅₅₉ based on optical and EPR measurements in Mn depleted PSII samples (51, 52).

According to the two-quantum model of photoactivation (25, 34), a second photon is required to stabilize the intermediate formed after the first flash; the optimal flash spacing between the first and the second flash is reported to be between 0.5 and 1 s. Thus, the long dark time (15 s) between flashes used above may allow significant back reaction of the first photooxidized intermediate, potentially reducing the O₂ yield that may originate from the seventh flash to below our detection limit. In the results presented in Fig. 4, we, therefore, applied a modified flash sequence in which a preflash train was used with 0.5 s spacing (green), followed by a dark time of 240 s before a second flash train (15 s spacing, red) was applied. This modified flash sequence should allow optimal formation of the 2-photon stabilized intermediate by minimizing the amount of back reaction thereby allowing the detection of O₂ evolution from the seventh flash (if any). For apo-PSII microcrystal suspensions that were subjected to 2 preflashes (2PF), the total number of flashes (sum of preflashes and flashes) required to evolve O₂ was, however, still 9, further supporting the HO model (Fig. 4). This test additionally excludes the scenario where residual NH₂OH that remained in the samples after 10 washing steps, calculated to be below 0.1 μM (SI Appendix), leads to back reactions of photoactivation intermediates between the dark times of the flashes. We further confirmed this observation by examining the sum of flashes required to evolve O₂ from apo-PSII that were subjected to 0PF to 8PF. Between 0PF and 6PF, in total, 9 laser flashes were required in each case to evolved O₂ and the O₂ yields induced by the ninth flash were similar in each case (blue symbols in Fig. 4). At 6PF, 7PF, and 8PF, the number of additional laser flashes (red) required to evolve O₂ remained constant at 3 flashes, while the O₂ yield increased with each additional preflash. These observations can be rationalized by the formation of the S₁ state after 6PF and of the S₂ and S₃ states after 7PF and 8PF, which decay back to the dark stable S₁ during the 240 s dark time via electron donation by, for example, tyrosine Y_D (see below). Therefore, 3 additional flashes were required in each case to complete the first catalytic cycle to evolve O₂, and the larger number of total flashes led to the expected increased photoactivation yields (compare to Fig. 3). EPR experiments showed that, in the dark-adapted apo-PSII samples, indeed, 94% of Y_D was reduced due to the NH₂OH treatment (SI Appendix, Fig. S4). Thus, the data in Fig. 3 provide additional strong evidence that the S₂ state was formed for the first time after 7 flashes and that the first O₂ was produced after 9 flashes during the first S₃ → [S₄] → S₀ transition.

Fig. 4.

(Top) Laser flash spacing’s used for photoactivation of apo-PSII microcrystals suspension by various combinations of tightly spaced preflashes (green line) and monitoring flashes (red line) with 15-s spacing. The dark time of 240 s was employed to allow the back reaction of the S₂ and S₃ states to S₁. (Bottom) Total number of flashes required to observe the first O₂ evolution (left axis) during the photoactivation of apo-PSII crystals with various combinations of preflashes and monitoring flashes (see Top). The yields of the first O₂ peaks, plotted in blue, are the averages of 2 repeat measurements, and the error bars are SDs.

To summarize, in contrast to previous photoactivation studies that employed complicated LED pulse trains as flashes, our data obtained with single turnover ns-laser pulses provides strong support for the HO state model of the Mn₄CaO₅ cluster. Together with the spectroscopic evidence for the HO model, this provides the required firm basis for deriving the mechanism of water oxidation. Furthermore, this study demonstrates the feasibility of the stepwise photoactivation of the Mn₄CaO₅ cluster depleted PSII microcrystals, which will allow detailed time resolved structural characterization of photoactivation intermediates using serial femtosecond X-ray crystallography.

Materials and Methods

T. elongatus cells were cultivated, and PSII core complexes purified as described previously (39). Intact PSII microcrystals with a size between 20 and 80 µm were generated according to literature methods (53). Apo-PSII microcrystals were obtained as in ref. 39 by incubating intact PSII microcrystals for 30 min with 50 mM NH₂OH and 50 mM EDTA, but the crystals were, subsequently, washed 8 times and dialyzed 2 times to achieve rigorous removal of NH₂OH and EDTA. The estimated residual NH₂OH and EDTA concentrations are no more than 0.1 µM (SI Appendix). EPR spectra of the apo-PSII microcrystal suspension in the g = 2.0 region were completely devoid of features associated with Mn²⁺ before and after heat treatment at 363 K to release any residual Mn, confirming the complete absence of Mn in apo-PSII preparations (SI Appendix, Fig. S3).

The apo-PSII microcrystals were stored in 0.1 M MES pH 6.5, 0.1 M NH₄Cl, and 20% PEG 5000 (MIMS buffer). All manipulations of apo-PSII microcrystals were performed in complete darkness with the aid of an IR camera and 940 nm illumination. Apo-PSII microcrystal suspensions (0.5 ± 0.15 mM) in an assembly buffer (MIMS buffer with 10 mM MnCl₂, 400 mM CaCl₂, 5 mM NaHCO₃, and 0.3 mM PPBQ, 6% H₂¹⁸O) were confined within a 6 mm diameter chamber that is 100 µm thick (thin-layer MIMS cuvette; Fig. 2) and allowed to degas for 4 min. Saturating single turnover flashes were provided using a 10 ns duration laser (532 nm and 60 mJ/cm² per pulse). The thin-layer MIMS cuvette was connected to the source vacuum of a Thermo Delta XP IRMS via a dry ice/ethanol cooling trap (54). Oxygen evolution was monitored via the m/z 34 signal from the IRMS. The oxygen yield for each flash-induced peak was calculated using a multipeak fitting procedure implemented in the software Igor Pro-6.3 (WaveMetrics) (for details see the SI Appendix).

Data Availability Statement.

All data discussed in the paper are available at http://uu.diva-portal.org under accession number diva2:1368663.