1 Introduction

In oxygenic photosynthesis, light-driven water-oxidation to molecular oxygen is carried out by the oxygen-evolving complex (OEC) of photosystem II (PSII). PSII is a multi-subunit pigment–protein complex that is located in the thylakoid membranes of green plants, algae and cyanobacteria. It is the only known biological water-splitting catalyst, and it is responsible for converting the atmosphere of the Earth from being anaerobic (2.3 billion years ago) to containing ~20% molecular oxygen [1]. The water-splitting catalyst within PSII is a Mn4O x Ca cluster (4 ≤ x ≤ 7) of not yet fully resolved structure, where x signifies the number of oxygen bridges and terminal water-derived ligands [2]. The uncertainty in the structure originates from the fact that the analysis of PSII by X-ray crystallography is still hampered by limited resolution and by specific radiation damage to the Mn4O x Ca cluster [3–5], and that extended X-ray absorption fine structure (EXAFS) spectroscopy cannot reveal the precise bridging pattern to Ca and the coordination of the surrounding ligands, including the two substrate water molecules. The surrounding protein microenvironment is important for tuning the energetics of water-splitting by allowing proton-coupled electron transfer and for regulating substrate access and product release [6–8]. One specifically important nearby protein side chain is tyrosine D1-Tyr161 (or TyrZ), since it is responsible for the electron transfer between the Mn4O x Ca cluster and P ·+680 .

P ·+680 is formed after light absorption by the antenna complexes and excitation energy transfer to the reaction center of PSII. New data indicate that the primary charge separation occurs between a chlorophyll molecule termed ChlD1 and its neighboring pheophytinD1 molecule [9]. Because of the strong coupling between the four reaction-center chlorophyll molecules [10], the radical is distributed over several pigments, including PD1 and PD2. Driven by this photochemical reaction, the Mn4O x Ca cluster undergoes a cycle composed of five distinct redox intermediates termed S i states (i = 0–4) [11]. Molecular oxygen is released during the S 3 → [S 4] → S 0 transition. Under working conditions the S 0 state is the most reduced state of the Mn4O x Ca cluster, but after sufficient dark-adaptation almost all centers are poised in the S 1 state. The S 2 and S 3 states are semi-stable and decay back to the S 1 state with t 1/2 = 3–5 min (20°C, neutral pH), while the S 0 state is oxidized by Tyr oxD of the D2 protein [12] in a much slower reaction to the S 1 state [13]. The S 4 state is a so far untrapped transient intermediate.

In spite of the above-mentioned progress in structural determination of the OEC, it is still unknown when, where and how the substrate water molecules bind to the Mn4O x Ca cluster during the S-state cycle. Because water is also required for the overall function of PSII, spectroscopic and functional studies of substrate binding need to employ either substrate analogs such as ammonia or methanol or isotope labeling (1H/2H, 16O/17O, 18O) [7, 14, 15]. Time-resolved membrane-inlet mass spectrometry experiments show that at least one substrate water molecule is bound in all S states, while both are bound in different chemical ways or environments in the S 2 and S 3 states [16–18]. Fourier-transform infrared (FTIR) spectroscopy measurements indicate that the hydrogen bond strength of one exchangeable, metal-(Mn, Ca)-bound water molecule gets more asymmetric during the S 1 → S 2 transition [19, 20]. Similarly, nuclear magnetic resonance paramagnetic relaxation enhancement (NMR-PRE or dispersion) measurements [21] were interpreted to show direct Mn-binding of at least one partially protonated water species in the S 2 state.

Further important tools for studying water-binding to Mn are (pulse) electron paramagnetic resonance (EPR) and electron-nuclear double resonance (ENDOR) spectroscopy. These techniques can detect the coupling of nuclear spins with the electron spin of S = 1/2 in the S 2 [22] and S 0 states [23–25] of the Mn4O x Ca cluster. The strength of the coupling and whether it is isotropic or anisotropic can then be used to argue whether or not 1H (I = 1/2), 2H (I = 1) or 17O (I = 5/2) are directly (protons via oxygen) bound to Mn.

The binding of water can be most directly probed by demonstrating the coupling of water-exchangeable oxygens to the electron spin of the Mn4O x Ca cluster. Initially it was attempted to detect line broadenings of the S 2 multiline signal caused by the hyperfine coupling of 17O [26]. However, the effects were rather small. Similarly, electron spin echo envelope modulation (ESEEM) studies were performed after H 162 O/H 172 O exchange with varying results [27, 28]. Recently, we employed 17O hyperfine sublevel correlation (HYSCORE) spectroscopy, which clearly revealed the coupling of one (type) of 17O with the paramagnetic Mn4O x Ca cluster (S 2 state). The determined coupling strength of 10–11 MHz is indicative of a water-exchangeable Mn–O–Mn bridge [29]. This site likely corresponds to an earlier reported low-frequency FTIR signal at 606 cm−1 that is sensitive to H 162 O/H 182 O-exchange and which was also assigned to a bridged oxo species [30, 31].

Somewhat more indirectly water binding can be probed via the observed changes in the hyperfine coupling of protons after 1H/2H-exchange. These types of experiments can give vital information regarding the protonation state of Mn-bound substrate molecules/atoms. Therefore, X-band continuous-wave (CW) and Mims ENDOR and ESEEM spectroscopy in conjunction with 1H2O/2H2O (D2O) exchange have been employed by several groups [32–38]. They all reveal several hyperfine interactions (HFI) in the order of 2–5 MHz between 1H and the Mn4O x Ca cluster (or correspondingly weaker once for 2H). On the basis of these findings all but one study [34] conclude that at least one hydroxo or water is (or maybe) directly bound to Mn in the S 2 state.

Here, we employ multifrequency pulse EPR techniques to specifically address the question whether the exchangeable oxygen detected by 17O HYSCORE is protonated.

2 Materials and Methods

2.1 PSII (BBY) Sample Preparation

Photosystem II membrane fragments (BBY) were isolated from spinach according to the method of Berthold and co-workers [39] with minor modifications. The oxygen-evolving activity of the PSII preparations at saturating light intensities was about 500 μmol (O2) mg (Chl)−1 h−1.

2.2 1H2O/2H2O (D2O)-Substituted BBY Samples

The dark-adapted samples (S 1 state) were suspended in buffer containing 0.4 M sucrose, 25 mM 2-(N-morpholino)ethanesulfonic acid (MES)–NaOH (pH 6.1), 15 mM NaCl, and 3 mM MgCl2. The samples were washed two times with 2H2O-enriched (>95%) buffer of identical composition. Finally, the samples were concentrated by 20 min centrifugation at 4°C in calibrated 4 mm X-band and 3 mm Q-band EPR tubes to a final concentration of 20–30 mg Chl/ml. The total incubation time in 2H2O was about 90 min. All samples contained ~3% (v/v) methanol.

2.3 Advancement of PSII Samples from S 1 to S 2 State

The PSII samples in the S 1 state were advanced to the S 2 state by 10–20 min continuous illumination at 200 K (dry ice/ethanol bath) using two 250 W halogen lamps, from which the infrared and ultraviolet contributions were largely removed by the following filters: 8 cm water, 2 cm CuSO4 solution (5% w/v), Schott KG 3 (2 mm) and Schott GG 445 (2 mm). The final light intensity at the sample level was ~0.5 W/cm2.

2.4 EPR Measurements

The measurements of X- and Q-band pulse EPR, τ-dependence/two-dimensional three-pulse (2D-3P) ESEEM, 2H-HYSCORE and 1H/2H Mims ENDOR at 4.2 K were performed on Bruker ELEXSYS E-580 X- and Q-band pulse EPR spectrometers equipped with superXFT and superQFT bridges, respectively, and with Oxford-900 liquid helium cryostats and ITC-503 temperature controllers. The 1H/2H Mims ENDOR data were collected with Spec-Man software [40] as described in Refs. [41–43]. The corresponding pulse sequences and EPR settings are given in the figure legends. The microwave frequencies in the pulse X- and pulse Q-band measurements were 9.71 and 33.9 GHz, respectively.

3 Results

3.1 X-band Pulse EPR Measurements in the S 2 State

Figure 1 shows the X-band light-minus-dark S 2 state electron spin echo (ESE)-detected field-swept difference spectrum of our concentrated (20–30 mg Chl/ml) spinach PSII membrane sample after washing into 2H2O. The hyperfine lines of the S 2 multiline signal can be discerned as steps. This shows that the S 2 state can be generated in this sample. The vertical bar and the arrow mark the field positions where the ESEEM/HYSCORE and 1H/2H ENDOR measurements were performed (see below).

Fig. 1
figure 1

Light-minus-dark ESE-detected field-swept X-band EPR spectrum of the S 2 state in PSII membrane fragments at 4.2 K. The sample was highly enriched (>95%) in 2H2O and had a Chl concentration of about 20–30 mg/ml. The used pulse sequence is a π/2–τ–π pulse, where π/2 = 12 ns, τ = 128 ns, and shots repetition time (SRT) = 5 ms with microwave frequency of 9.70 GHz. The displayed signal represents the average of four scans. The absorption of the Tyr ·D radical was removed for the presentation. The vertical bar and the arrow indicate the field positions where the Mims ENDOR and ESEEM/HYSCORE were measured in Figs. 2, 3 and 4, respectively

Full size image

3.2 X- and Q-Band 1H/2H Mims ENDOR

It was reported previously in the literature that the 2H signal originating from the exchangeable 2H can be detected by the X-band 2H CW and Mims ENDOR, and in two-pulse and three-pulse (2P/3P) ESEEM experiments (see Sect. 1). Here we employ 1H/2H Mims ENDOR, a sensitive technique to probe at X- and Q-band frequencies directly the weak HFI of highly concentrated BBY samples in buffer either enriched in 1H2O or 2H2O. The obtained results are displayed in Fig. 2. In the 1H2O sample, the 1H signal is centered at X-band at the 1H Larmor frequency of 14.26 MHz (B 0 = 335 mT). Additionally two weak shoulders (see the arrows) were observed (gray spectrum in Fig. 2). No signal above background was detected in the region of 0–10 MHz. After the 1H2O/2H2O substitution, the 2H signal was detected at a radio frequency (rf) position of around 2.19 MHz (the 2H Larmor frequency at B 0 = 335 mT) and a width of ~1.5 MHz (black spectrum in Fig. 2). This signal was broadened without revealing a clear splitting, which in principle would be centered at ν 2H (2.19 MHz) with a splitting distance of A, i.e., ν 2H ± A/2. The appearance of the new signal after 1H2O/2H2O substitution demonstrates, in agreement with the previous reports, that there are water-exchangeable protons in the vicinity of Mn in the S 2 state. Additionally, also the signal at 14.26 MHz is still observed, which originates from non-exchangeable protons that couple with the Mn4O x Ca cluster and other underlying EPR signals like Tyr ·D . Since we know from time-resolved membrane-inlet mass spectrometry that substrate water exchanges completely within a few seconds [16, 44], the latter signal cannot originate from substrate water protons.

Fig. 2
figure 2

1H/2H Mims ENDOR of concentrated PSII membranes from spinach suspended either in 2H2O-enriched (black) or in 1H2O-buffer (gray). The data were recorded with a pulse sequence π/2–τ–π/2–t rf–π/2–τ-echo, where π/2 = 200 ns, τ = 420 ns, the length of rf pulse t rf was 25 μs in both X- and Q-band measurements. The Larmor frequencies of 2H in X-band (B 0 = 335 mT, 9.70 GHz) and Q-band (B 0 = 1220 mT, 33.93 GHz) are 2.19 and 7.974 MHz, respectively. For 1H, it is 14.26 MHz (B 0 = 335 mT) in X-band. The inset spectrum was measured at Q-band (2H2O-enriched sample). The arrows indicate the positions of 1H-signals from exchangeable protons

Full size image

To improve the resolution of the 2H signal, Mims ENDOR was measured also at Q-band frequency (34 GHz). While the signal moved as expected to the 2H Larmor frequency at Q-band of 7.9 MHz, the signal remained unresolved with a width of ~1.5 MHz despite trying various different τ values to avoid the blind-spot behavior in Mims ENDOR (see Sect. 3.3).

For the same element, it is known that the HFIs of different isotopes have a good proportion to the corresponding nuclear magnetogyric ratio, i.e., g 1H/g 2H, with g 1H = 5.5856948 and g 2H = 0.8574388288. Therefore, a scrutinized inspection was conducted on the difference between the 1H X-band Mims ENDOR obtained in the 1H2O- and 2H2O-enriched samples. Figure 2 shows that some exchangeable 1H signals splitting with HFI of A 1H = 2.5–4 MHz disappeared after the 2H2O substitute (see arrows). With respect to the ratio g 1H/g 2H = 6.5144, these exchanged 1H signals give rise to the corresponding 2H signals with the HFI of A 2H = 0.4–0.6 MHz, which are equal to approximately half of the width (1.5 MHz) of the broad 2H signal (the two black spectra in Fig. 2). These data demonstrated that the HFIs between the 2H and the Mn4O x Ca cluster are fairly weak (see also Sect. 4). Thus, 3P-ESEEM and 2H-HYSCORE were used in the following measurements to further characterize the nature of this coupling.

3.3 X-band 2D-3P ESEEM in the 2H2O-Enriched Samples

In practice it is often found that there is a similar blind-spot behavior in 3P-ESEEM and HYSCORE experiments. At blind-spots no nuclear coherence is generated by the microwave pulses. To avoid these conditions in subsequent HYSCORE experiments, the τ-dependence/2D-3P ESEEM (2D-3P ESEEM) was recorded for a 2H2O-enriched sample at the magnetic field of B 0 = 355 mT. The data obtained are displayed in Fig. 3. They display both a weak 1H signal (because of high 2H2O enrichment) and a strong 2H signal centered at 15.1 and 2.3 MHz, respectively. Both signals show the discussed above blind-spot behavior with oscillating signal intensity, when τ increased from 118 to 800 ns. Figure 3 also shows weak 14N signals at 3–8 MHz. The origin of the 14N signals is still not clear, but it is known that the Mn4O x Ca cluster has one direct histidine ligand [45].

Fig. 3
figure 3

2D-3P ESEEM of concentrated PSII membranes from spinach suspended in 2H2O-enriched buffer. The spectra were recorded with the pulse sequence π/2–τ–π/2–T–π/2–τ-echo, where π/2 = 24 ns and τ was varied from 108 to 888 ns in 20 ns steps. T was varied from 48 to 6192 ns in 24 ns steps. Four-phase cycling with 50 shots/point and a shot repetition time of 5 ms was employed. The magnetic field was set to B 0 = 355 mT. The Larmor frequencies of 1H and 2H are 15.115 and 2.32 MHz, respectively. After Fourier transformation, the projections on the top and on the right show the 1D spectrum in the frequency domain, and the oscillation of the signal amplitude as a function of τ, respectively

Full size image

3.4 X-band 2H-HYSCORE in the 2H2O-Enriched Samples

In HYSCORE measurements, signals may be resolved in two separate quadrants based on the relation of |ν| > |A/2| and |ν| < |A/2|, respectively, where ν is the Larmor frequency of the nucleus under investigation and A is the HFI between the nuclear spin and the electron spin (of Mn4O x Ca cluster). Figure 4 shows only the (++) quadrant of our 2H-HYSCORE spectrum measured with τ = 252 ns, since no signal was detectable in the (−+) quadrant. The absence of a single in the (−+) quadrant is in sharp contrast to our recent 17O HYSCORE data and shows that there are no strong HFIs between 2H and the Mn4O x Ca cluster in the S 2 state. In the (++) quadrant (Fig. 4), a strong 2H signal is seen that is centered at the 2H Larmor frequency of 2.32 MHz at B 0 = 355 mT. The broad line shape indicates the presence of a weak, unresolved HFI. Similar results were observed in other HYSCORE measurements performed at several other τ values (128, 156, 188, 216, 300, and 616 ns). All these results indicated that the HFI between the 2H and the Mn4O x Ca cluster is too weak to be resolved by HYSCORE. Clearly no strong or intermediate HFI between exchangeable protons and the Mn4O x Ca cluster are present.

Fig. 4
figure 4

X-band 2H-HYSCORE spectrum of concentrated PSII membranes from spinach suspended in 2H2O-enriched buffer. The spectrum was recorded with a pulse sequence π/2–τ–π/2–t 1–π–t 2–π/2–τ-echo with π/2 = 24 ns. The times t 1 and t 2 were varied from 60 to 6,720 ns in 24 ns steps. Four-phase cycling with 50 shots/point and a shot repetition time of 5 ms was employed. Other parameters: B 0 = 355 mT, τ = 256 ns and the microwave frequency was 9.70 GHz

Full size image

4 Discussion

4.1 Protonation State of Bound Substrate Water in the S 2 State

In our previous study, a 10-MHz HFI between one bulk water-exchangeable oxygen with the electron spin on the Mn4O x Ca cluster was detected in the S 2 state of PSII by the 17O-HYSCORE. This showed that in the S 2 state there is very likely only one (type of) Mn-ligated oxygen that can be readily exchanged with bulk water (note of caution: in EPR the absence of a signal is no absolute proof for the absence of an interaction). In the present study, we endeavored to probe the protonation state of this bound ‘water’ in the S 2 state by employing isotope labeling with 2H2O and several pulse EPR techniques at X- and Q-band frequencies. These techniques are sensitive to HFIs between 1H or 2H and the paramagnetic Mn4O x Ca cluster. In our report we focus on 2H couplings, since they represent bulk water exchangeable protons—a minimal requirement for protons of substrate water.

The coupling strength of a deuterium bound to this exchangeable oxygen can be estimated on the basis of simple considerations as follows: it is well established that the Mn–O distance is typically 1.8–2 Å in a bridge, and ~2.2 Å to a terminal water ligand [46–48]. The O–1H and O–2H distances are ~1 Å. Therefore, the maximal distance between the 1H/2H and the Mn ion(s) via the oxygen is 2.8–3.2 Å. Since the Fermi contact interaction is proportional to the g-factors of 17O (0.758) and 2H (0.857), and inversely proportional to the cubed through-bond distance (d 3), a minimal isotropic hyperfine interaction A 2H of about 3 MHz can be roughly estimated based on:

$$ A_{{2{\text{H}}}} = A_{{17{\text{O}}}} \times {\frac{{d^{3}_{(170{\text{-}} {\text{Mn}})} }}{{d^{3}_{(2 {\text{H}}{\text{-}} {\text{Mn}})} }}} \times {\frac{{g_{{ 2 {\text{H}}}} }}{{g_{{ 1 7 {\text{O}}}} }}} $$

In contrast to this expectation, all results obtained in this study for the 2H2O-enriched samples show just a broadened 2H signal of ~1.5 MHz width without resolved splitting according to ν 2H ± A 2H/2. This indicates that the coupling between the deuterons and the electron spin on the Mn4O x Ca cluster is rather weak. The strength of this HFI can be estimated indirectly from the observation of Fig. 2 that in the Mims ENDOR, the occurrence of the 2H signal after 1H/2H exchange leads to a small decrease of the 1H signal (see arrows in Fig. 2) that have a HFI of A 1H = 2.5–4 MHz. For an element, the HFIs of different isotopes are, to a good approximation, proportional to their corresponding nuclear magnetogyric ratios (i.e., g 1H = 5.5856948 and g 2H = 0.8574388288). Due to g 1H/g 2H = 6.5144, these exchanged 1H signals give rise to the corresponding 2H signals with the HFI of A 2H = 0.4–0.6 MHz, which are 30–40% of the full width (1.5 MHz) of the observed 2H signal.

We therefore conclude that in agreement with our previous proposal the water-exchangeable oxygen of the Mn4O x Ca cluster represents one fully deprotonated substrate ‘water’ that is likely bonded as a bridge in a Mn-bis μ2-oxo-Mn unit of the Mn4O x Ca cluster in the S 2 state. However, binding as mono-μ-oxo or in a Mn-μ3-oxo-μ2-oxo-Mn unit or even binding as terminal Mn=O may not be ruled out entirely on the basis of just these measurements without further model studies and detailed theoretical calculations. It is noted that a terminal Mn=O is disfavored by X-ray absorption near-edge structure experiments on model complexes [49].

Our previous 17O data and the current 1H/2H results do not provide evidence for a second Mn-bound water molecule in the S 2 state. Therefore, if a second substrate water molecule is bound in the OEC in the S 2 state, as suggested by time-resolved membrane-inlet mass spectrometry, this water is likely coordinated to Ca2+ or to protein ligands. On the other hand, our data are also consistent with a recent proposal by Siegbahn [50] that suggests that the second water molecule binds during the structural rearrangement that occurs in the S 2 → S 3 transition.

The above conclusions are in agreement with two previous findings. Dismukes and coworkers [34] reported three weak hyperfine interactions of 4.5, 2.9 and 1.5 MHz and a very weak isotropic coupling of about 0.5 MHz as revealed by 1H ENDOR spectroscopy. On the basis of a spin-coupled point-pair model Dismukes et al. [34] estimated that the minimal distance to the nearest exchangeable 1H is 3.65 Å. On this basis it was concluded that the Mn4O x Ca cluster is basically ‘dry’ with only weak H bonds to oxo-bridges, and that all water ligands directly bound to Mn must be fully deprotonated. A possible role of Ca in water binding was discussed [34]. However, on the basis of similar data and using a more complex theoretical modeling Fiege et al. [35] did not fully exclude direct water-binding to Mn.

A second line of support comes from a pioneering low-frequency FTIR study by Babcock and coworkers [30]. In this study, a 606 cm−1 mode was found in the S 2 state that corresponds to a mode at 625 cm−1 in the S 1 state. This mode was clearly downshifted after exchange with H 182 O to 596 cm−1. This exchange was complete in less than 30 min. By comparison to existing data on model compounds this mode was concluded to most likely arise from the exchange of only one oxo within a Mn–bis-μ2-oxo–Mn unit or possibly within a Mn–μ3-oxo-μ2-oxo–Mn unit. The latter suggestion might be able to explain why only one of the oxos is exchangeable. Furthermore, Babcock et al. [30] showed that the frequency of this oxo is upshifted to 618 cm−1 by Ca2+/Sr2+ exchange, but that it was unaffected by 40Ca2+/44Ca2+ exchange. This was taken as evidence that Ca2+ is not directly involved in binding of this oxo, but that the exchange of Ca2+ against Sr2+ affects the binding of this oxo indirectly, for example, via structural changes of the cluster or its protein microenvironment. This finding will be important later on in the discussion on substrate water exchange rates.

However, there are also apparent conflicts with several recent reports that propose the binding of protonated water to the Mn4O x Ca cluster in the S 2 state. These concerns are addressed in the following sections.

1H ENDOR studies by Kawamori and coworkers [32] and by Yamada et al. [38] report several 1H signals at frequencies of 2–4 MHz that are partly exchangeable against bulk 2H2O at relatively slow time scales (3–24 h). Point dipole approximations yield relatively short distances of about 2.7–3.3 Å. This was taken as evidence for the binding of protonated substrate water to Mn. Similarly Britt and coworkers [36] using 1H/2H ESEEM spectroscopy reported several classes of exchangeable deuterons. The most interesting one was concluded to comprise two 2H with an isotropic coupling of 0.45 MHz and an anisotropic coupling of 0.64 MHz. Again a point dipole approximation was employed to estimate a distance of 2.67 Å, which together with the isotropic coupling was taken as evidence for the binding of one H2O or two OH− to Mn. Since 55Mn ENDOR data reveal that the electron spin of the S 2 state is distributed over all four Mn ions [51], the dipole approximation is likely a too simple model and therefore thus calculated distances cannot be taken as reliable indicators for the binding of protonated water to Mn. It should be noted that the magnitude of the reported 1H and 2H couplings compares well with the estimated ranges in this report.

Nugent, Evans and coworkers [27] performed two 17O ESEEM studies. In the first study [27], they concluded that the Mn4O x Ca cluster does not contain any exchangeable oxygen since no 17O HFI was found. In the second study [28], a 17O coupling was reported for the central part of the S 2 multiline signal and the binding of water (not hydroxyl) to a quasi-axial MnIII center was proposed. However, the reported couplings are not consistent with our later 17O HYSCORE study [29].

With the help of FTIR spectroscopy at higher frequencies Noguchi and Sugiura [20] found two modes in the S 2–S 1 difference spectrum in the weakly H-bonded OH region, which appeared at 3617 and 3588 cm−1. These modes were sensitive to 2H2O/1H2O exchange and to H 162 O/H 182 O substitution, and thereby provide strong evidence for a water molecule that changes its hydrogen bond strength upon S state turnover. It was suggested that this water molecule is coupled to metal (Ca or Mn) [19], however, no strong evidence, except for the S-state dependence, was presented for this. It is therefore possible, that this water molecule is not Mn-bound, but either bound to Ca or to an amino acid in the vicinity of the Mn4O x Ca cluster. Both latter options would be consistent with our findings.

Kimura et al. [31] determined the full S state dependence of the low-frequency FTIR signals. In this detailed study three classes of oxygen and hydrogen interactions were observed. Class one was only sensitive to 16O/18O exchange. This class corresponds to the 606 cm−1 mode of Chu et al. [30] discussed above and was also assigned to a Mn–O–Mn moiety. The second class was only sensitive to 1H/2H exchange and was assigned to exchangeable hydrogens of side chains or of the protein backbone. The class III signals were sensitive to both 16O/18O and 1H/2H exchange. These signals therefore belong very likely to water molecules or hydroxide ions in the vicinity of the Mn4O x Ca cluster. The authors [30] favor direct binding of this potential substrate ‘water’ to Mn. It seems likely that the same moiety gives rise to the class III signals and to the high frequency once observed by Noguchi and Sugiura [20]. However, it appears that the suggestion of direct Mn binding is largely based on the cited previous suggestions by ESEEM and ENDOR. Indeed, the authors [30] do not exclude water/hydroxide binding to Ca or H-bridging to a Mn–O–Mn bridge.

NMR-PRE experiments originally performed by Wydrzynski et al. [52] and later by Sharp and coworkers [21] show that the relaxation rate of the protons in the bulk water is dependent on the oxidation state of the Mn4O x Ca cluster. This suggests that there is at least one bound water or hydroxo directly coordinated to Mn, so that this relaxation enhancement by the paramagnetic Mn4O x Ca cluster can be transmitted via exchange reactions to the bulk phase. However, the theory for this effect is complex and straightforward conclusions cannot be made on the basis of the current data and their analysis.

We therefore conclude that there is no, to our knowledge, compelling evidence in the literature that demonstrates binding of water or hydroxide to Mn in the S 2 state of PSII. In contrast, three independent experiments strongly favor the conclusion that there is one exchangeable oxo group in the S 2 state, which therefore likely represents one of the two substrate water molecules. Substrate water exchange experiments have demonstrated that the two substrate molecules bind in different chemical environments because they exchange in the S 2 and S 3 states with different rates and activation energies [16, 44]. Accordingly, the two substrate water molecules were denoted as fast-exchanging (Wf) and slow-exchanging (Ws) [53]. It is now interesting to speculate if the proposed μ2-oxo-bridge corresponds to Wf or Ws.

Since the other substrate molecule (not detectable in this study) is either bound to Ca or the protein, it is very likely that the exchangeable μ2-oxo-bridge can be identified with Ws. This assignment may be thought to be in conflict with the finding that the slow phase of water exchange is more strongly affected by Ca2+/Sr2+ exchange than the fast phase [54]. First, one has to note that the fast phase of exchange was only resolved in these experiments in the S 3 state and not in the S 2 state. This is important since EXAFS and other experiments indicate a significant structural change between the S 2 and S 3 state within the Mn4O x Ca cluster [55, 56]. Second, as summarized above, Chu et al. [30] showed that the FTIR mode of the oxo-bridge is sensitive to Ca2+/Sr2+ exchange, without direct binding of the oxo to Ca. This observed frequency shift demonstrates the sensitivity of the proposed Ws to Ca2+/Sr2+ exchange. The smaller effect on the fast phase may then be explained by the assumption that this exchange rate is not limited by binding to Ca2+/Sr2+, but by other factors such as, for example, diffusion through the protein to the active site. This would be consistent with the smaller activation energy that is coupled with this exchange compared to the slow exchange rate. Furthermore, our assignment of Ws is consistent with a theoretical study by Siegbahn and coworkers [57]. By employing density functional theory calculations they estimated the energetic barriers for terminal water or hydroxo versus oxo-bridges within a bis-μ2-oxo-bridged Mn dimer. While a direct comparison of this simple model to the exchange processes within the OEC is rather complicated, the data are best in agreement with Ws being a bridging oxygen, while the exchange of Wf is limited by diffusion through the protein. However, another theoretical study implies that within the PSII complex water bound to Ca2+ exchanges, in contrast to what is known from solutions, more slowly than a terminal Mn(IV)=O species [58].

On the basis of this discussion a picture of water binding in the S 2 state arises, in which Ws is a fully deprotonated substrate water that is bound directly to the Mn, probably in form of a μ2-oxo-bridge between MnA and MnB, while Wf probably corresponds to the water observed by Noguchi and Sugiura [20] and is therefore fully protonated and either bound to Ca or to a nearby amino acid. This conclusion is illustrated in Fig. 5 on the basis of a recent model of the Mn4O x Ca cluster (however, if the calculations by Batista and coworkers [58] are correct, then the assignment of Wf and Ws would have to be reversed). The reason why only one oxo-bridge can be exchanged, and why this rate is faster than expected on the basis of some recent mass spectrometry-based experiments [59, 60] is closely coupled to the not yet fully resolved structure of the OEC and therefore remains to be established in future studies.

Fig. 5
figure 5

Proposed substrate ‘water’ binding modes and sites at the S 2 state of the Mn4O x Ca. W s and W f represent the slow- and the fast-exchanging substrate ‘water’ molecules as determined by time-resolved membrane-inlet mass spectrometry [16, 44]. The four Mn ions are denoted as A, B, C, D, while the oxygen ions are represented by dark spheres. The structural model employed is based on single-crystal EXAFS spectroscopy [2] and theoretical calculations [61]

Full size image