Structural changes correlated with magnetic spin state isomorphism in the S2 state of the Mn4CaO5 cluster in the oxygen-evolving complex of photosystem II

Mn XAS indicating different structures in the spin isomers of the S2 state of PSII


Introduction
In oxygenic photosynthesis, light-driven water oxidation to molecular oxygen is carried out by the oxygen-evolving complex (OEC) in photosystem II (PSII). PSII is a multisubunit protein complex in the thylakoid membrane of plants, algae, and cyanobacteria. 1,2 The OEC consists of four oxo-bridged Mn atoms and one Ca atom (Mn 4 CaO 5 ) ligated to the D1 and CP43 subunits by carboxylate and histidine ligands. 3,4 During water oxidation, the Mn 4 CaO 5 complex cycles through ve intermediate states, collectively called the S states, labeled S 0 -S 4 in the Kok cycle. 5 S 0 is the most reduced state while S 1 , S 2 and S 3 represent sequentially higher oxidation states in the OEC. O 2 is released during the S 3 / [S 4 ] / S 0 transition, where S 4 is a transient state. Thus, the Mn 4 CaO 5 cluster accumulates four charges before the release of O 2 .
The oxidation state of each S-state has been formally assigned as Mn III 3 Mn IV for S 0 , Mn III 2 Mn IV 2 for S 1 , Mn III Mn IV 3 for S 2 , and Mn IV 4 for S 3 . 6-10 We note that there has been debate regarding the oxidation state assignment of the S 3 state (i.e. whether it is formally Mn IV 4 or Mn III Mn IV 3 with charge delocalized on the ligands), 8,11,12 and the current view from several experiments point more towards the formal oxidation state of Mn IV 4 . However, formal oxidation states does not necessarily coincide with effective number of electrons in the metal valence shells because of important factors like metalligand covalency. 13,14 A recent resonant inelastic X-ray scattering spectroscopy (RIXS) study indicate increasing delocalization of positive charge on to the ligands during the S-state transitions. 15 Among the S-states, the S 2 state is the most studied state due to the presence of rich EPR signals and nearly 100% conversion by illumination starting from the dark stable S 1 state. The subsequent S 2 to S 3 state transition is accompanied by noticeable Mn-Mn distance changes, 16 and several factors such as Cadepletion, 17 site-specic mutations, 18 and chemical treatments (for example, with uoride) 19 are known to block this advance. The requirement for a structural change, and its susceptibility to many chemical and biochemical treatments, makes S 2 to S 3 transition one of the critical steps for water oxidation reaction during the S-state cycle.
In recent studies, [20][21][22][23][24] the isomorphism observed in the S 2 state has been suggested to be of importance in relation to the formation of the S 3 state, where the chemical environment is prepared for the O-O bond formation to occur in the following steps. The presence of such chemical exibility within the same OEC redox state (i.e. S-state) may play an important role in the catalytic process, for example, by providing a low energy barrier for the water exchange process. In the current study, we investigate isomorphism in the S 2 state using Mn K-edge X-ray absorption, both XANES and EXAFS, and emission spectroscopy, and further discuss the mechanistic implication of such isomorphous states to the catalytic function of the OEC.
In the S 2 state, two types of EPR signals have been assigned to the Mn cluster. The multiline signal (MLS) centered at g ¼ 2 (S 2 -g2), exhibiting at least 18 partially resolved hyperne lines at X-band ($9 GHz), is a low spin (S total ¼ 1/2, i.e. Mn III /Mn IV and Mn IV /Mn IV are antiferromagnetically-coupled, respectively) ground state. 9,[25][26][27][28][29][30][31][32][33] Another broad featureless EPR signal at g $ 4.1 (S 2 -g4), attributed to a higher spin multiplicity (S total ¼ 5/2, i.e. ferromagnetically-coupled three Mn IV with antiferromagnetically-coupled one Mn III ) ground state, is also observed under different experimental conditions. [34][35][36][37][38][39][40] The high spin (S total ¼ 5/2, called HS S 2 or S 2 -g4 in the text) and low spin (S total ¼ 1/2, called LS S 2 or S 2 -g2 in the text) forms in the S 2 state are interrelated, on the basis of the observation of amplitude conversion of the S 2 -g4 EPR signal to the S 2 -g2 EPR signal. 34,[41][42][43] The distribution of high spin and low spin species, g values and hyperne coupling values of these spin state changes are sensitive to several parameters, such as (a) species (higher-plant, thermophile or non-thermophile cyanobacterial PSII), (b) the presence of chemical additives like alcohol (methanol or ethanol), sucrose and glycerol (oen used as a cryo-protectant) in the sample, (c) substitution of the native Ca 2+ in the OEC (Ca 2+ -PSII) by Sr (Sr 2+ -PSII), and (d) halide substitution in PSII with Br À or I À replacing the Cl À of the native state. A detailed discussion of these studies can be found in several reviews. 32,44,45 Briey, in samples illuminated at 195 K, both S 2 -g2 and S 2 -g4 signals are observed in the presence of sucrose, while with glycerol, ethylene glycol, or ethanol, the MLS is enhanced and the S 2 -g4 EPR signal is suppressed. 41 Some treatments such as (c) and (d) stabilize the HS S 2 form in the presence of the LS S 2 form. Illumination by near-infrared (NIR) light at low temperature ($150 K) has been shown to convert the S 2 -g2 form to the S 2 -g4 form without further advancement of S-state of the OEC. 39,41 Subsequent annealing in the dark at 200 K converts the S 2 -g4 form back to the S 2 -g2 form, 34 showing that these two forms are interconvertible. Both S 2 -g2 and S 2 -g4 forms show similar oscillation patterns around the S state cycle. 41 PSII samples treated with NH 3 , F À , NO 3 À , or I À or when Ca 2+ is replaced by Sr 2+ have been reported to show an enhanced S 2 -g4 signal with the line widths and g values being slightly different. 46 The S-state transitions focused on the S 1 -S 2 -S 3 steps are summarized in Scheme 1.
Recently, density functional theory (DFT) calculations by two groups suggest theoretical structural models corresponding to the two spin states 21,47 and conclude that the two spin states are almost isoenergetic. Ab initio molecular dynamics simulations by Bovi et al. showed that these two states could interconvert over a low barrier (DG # of 10.6 kcal mol À1 ). 48 In proposed models by Pantazis et al., 21 the two spin states arise from a different location of Mn III ; for LS S 2 , Mn III is located in the corner of the cubane motif (Mn D1 ), while for HS S 2 , it is located at the tail Mn A4 (see Scheme 1). They along with a few other studies suggest that such isomorphism makes O 5 unique, and that O 5 may be a likely candidate for the slow-exchanging water in the S 2 state. 3,4,22,[49][50][51][52][53][54] Previously, Liang et al. performed an XAS study on the HS S 2 state. 55 In their study, the authors concluded that HS S 2 state is different from S 1 and LS S 2 . They observe that the low spin S 2 state showed a positive K edge shi compared to high spin state and an elongation of one of the Mn-Mn bond distances from 2.73 to 2.85Å. 55 In our current study, with improved data quality and the structural information available for the OEC S 1 state from X-ray diffraction, 4 we gain a detailed structural insight that will help us in understanding the mechanistic detail of the S 2 to S 3 transition.
In this study, we used X-ray absorption (XAS) and X-ray emission spectroscopy (XES) to study the nature of the two spin states in the S 2 state. The possible structural changes are analysed based on the geometry obtained from the 1.95Å resolution crystal structure of the S 1 state. 4 We discuss the structural and electronic structural differences of the two spin states, and its relation to the functional role in S 2 to S 3 transition and subsequently during the water oxidation reaction.

Materials and method
Preparation of PSII membranes PSII-enriched membrane fragments were prepared under dim green light from spinach leaves according to Berthold et al. 56 PSII membranes were resuspended to a chlorophyll (Chl) concentration of 8 mg Chl per mL in a buffer containing 30% (v/v) glycerol, 50 mM MES-NaOH (pH 6.0), 5 mM MgCl 2 , 5 mM CaCl 2 , 15 mM NaCl and stored at À80 C until used. All samples were measured in this buffer. Oxygen-evolution activity of 400-500 mmol of O 2 per mg of Chl per h was observed. The oxygenevolution activity was measured in a buffer with 50 mM MES-NaOH (pH 6.0), 10 mM MgCl 2 , 5 mM CaCl 2 and 15 mM NaCl at 25 C under saturating light and in the presence of 0.5 mM phenyl-p-benzoquinone (pPBQ) as electron acceptor.
The samples for X-ray studies were prepared by mounting PSII membranes pellets (chlorophyll concentrations in these samples ranged from 20 to 25 mg mL À1 ) directly onto the Lucite sample holders, with a hollowed compartment (dimensions of 2.1 Â 0.3 Â 0.15 cm) backed by a piece of mylar tape. All illuminations, EPR, and X-ray measurements were performed directly on samples mounted in these holders.

Generation of the S-states by illumination
All the sample preparations as described above were performed in the dark or with dim green light at 4 C to poise the PSII centers in the S 1 state and then the samples were frozen in liquid nitrogen. The HS S 2 and LS S 2 states were generated by light illumination at 140 AE 1 K or 195 K. Prior to illumination, dark-adapted samples were equilibrated for 3 min at 140 AE 1 K or 195 K. For 195 K illumination, the temperature was maintained in a dry ice/ethanol bath in an unsilvered dewar, and samples were continuously illuminated for 10 min using a 400 W tungsten lamp, with a 7 cm path of 5% CuSO 4 as a heat and IR light lter. For 140 K illumination, the temperature was maintained with a continuous stream of liquid nitrogen-cooled nitrogen gas. Samples were continuously illuminated for 10 min using a 400 W tungsten lamp, with a 7 cm path of water as a heat lter. The temperature was monitored throughout the illumination period with a copper-constantan thermocouple. Aer illumination, samples were frozen in liquid nitrogen within 1-2 seconds. The 2 ash data used in this study was collected previously. The S 3 spectra were deconvoluted using the protocol established previously. 57

EPR spectroscopy
Low-temperature X-band EPR spectra were recorded using a Varian E109 EPR spectrometer equipped with a Model 102 Microwave bridge. Sample temperature was maintained at 8 K using an Air Products LTR liquid helium cryostat. The following spectrometer conditions were used: microwave frequency, 9.22 GHz; eld modulation amplitude, 32 G at 100 kHz; microwave power, 20 mW. The EPR signals were quantitated by adding the peak-to-trough amplitudes of S 2 -g4 or four of the downeld hyperne lines of the S 2 -g2 MLS, respectively.

XAS measurements
X-ray absorption spectra were collected at the Stanford Synchrotron Radiation Lightsource (SSRL) on beamline 7-3 at an electron energy of 3.0 GeV and an average current of 500 mA. The intensity of the incident X-rays was monitored by a N 2 -lled ion chamber (I 0 ) in front of the sample. The slit in front of the I 0 detector was closed to a vertical size of 2.5 mm and a horizontal size of 14 mm. The radiation was monochromatized by a Si (220) double-crystal monochromator. The total photon ux on the sample was limited to $3 Â 10 6 photons per mm 2 , which was determined to be non-damaging on the basis of detailed radiation-damage studies. 16,[58][59][60] We compared consecutive XAS scans from each sample and detected no shi in the K-edge energy over rst ve scans at each spot (Fig. S1 †). The samples were protected from the beam during the monochromator movements between different energy positions by a shutter that was synchronized with the scan program. The samples were kept at 8 K in a He atmosphere at ambient pressure by using an Oxford CF-1208 continuous-ow liquid He cryostat. Data were recorded as uorescence excitation spectra by using a germanium 30-element energy-resolving detector (Canberra Electronics). For Mn XAS, energy was calibrated by the pre-edge peak of KMnO 4 (6543.3 eV), which was placed between two N 2 -lled ionization chambers (I 1 and I 2 ) aer the sample.
Data reduction of the EXAFS spectra was performed using SIXPAK. 61 Pre-edge and post-edge backgrounds were subtracted from the XAS spectra, and the results were normalized with respect to edge height. Background removal in k-space was achieved through a ve-domain cubic spline. Curve tting was performed with Artemis and IFEFFIT soware using ab initiocalculated phases and amplitudes from the program FEFF 8.2. 62,63 EXAFS curve-tting procedure is described in detail in the ESI. † Mn XANES pre-edge spectra were t using EDG_FIT in EXAFSPAK. 64 The XANES inection point energy (IPE) was extracted from zero crossing of the second derivative in the energy region between 6550 eV and 6554 eV.

XES measurements
X-ray emission spectra were collected at SSRL on beamline 6-2. The beamline monochromator, using two cryogenically cooled Si crystals in (111) reection, was used to set the incident photon energy to 10.4 keV. The X-ray beam was focused to 0.45 (V) Â 0.45 (H) mm (fwhm) by means of vertical and horizontal focusing mirrors. The X-ray ux at 10.4 keV was $1 Â 10 13 photons per s per mm 2 . During the measurement, samples were kept at 10 K in a continuous ow liquid helium cryostat (Oxford Instruments CF1208) under helium exchange gas atmosphere. Emission spectra were recorded by means of a high-resolution crystal-array spectrometer, using the 440 reection of 7 spherically bent Si(110) crystals (100 mm diameter, 1 m radius of curvature), aligned on intersecting Rowland circles. 65 An energyresolving Si dri detector (Vortex) was positioned at the focus of the 7 diffracting elements. A helium-lled polyethylene bag was placed between the cryostat and the spectrometer to minimize signal attenuation due to air absorption. Each energy point in the spectra was collected at a fresh sample spot. The maximum exposure time at each spot was 2.5 seconds and the signal was read out in bins of 50 ms duration. At rst, a time-scan at a single emission energy was carried out for each S-state to check the onset time of radiation-induced changes of the signal intensity. No changes were observed at least for the rst 1.5 s, and therefore the rst 20 bins (equivalent to 1 s) were averaged for the nal spectra. The signal intensity from each sample spot was normalized by the emission signal intensity recorded at 6491.5 eV within 7 s from the same sample spot, aer going through all the fresh spots. Fig. S2 † shows the XES spectra aer rst 20 bins (equivalent to 1 s) and from bin 11-30 (0.5-1.5 s) of the 140 K NIR illuminated sample. We see no damage till 1.5 s of data collection.

Computational details
The optimizations were carried out using Gaussian 09 (ref. 66) and ONIOM calculation. 67 The initial structure was based on a previous study on the OEC in the S 2 state. 68 S ¼ 13/2 spin state was used so that the oxidation states of Mn ions in the S 2 state is Mn III Mn IV 3 . The Mn oxidation states were determined by the Mulliken's spin population analysis. The high layer of ONIOM calculation was assigned to the Mn 4 CaO 5 cluster and the ligands (Asp170, Glu189, His332, Glu333, His337, Asp342, Ala344, Glu354, Arg357, W1-W4 and other water ligands). Notations for each residue are similar to those in the PDB-data (3ARC). 3 The low layer of the ONIOM was assigned to the residues within 40Å radius of Ca in the Mn 4 CaO 5 cluster. The high layer was calculated with wB97XD DFT functional, 69 LanL2DZ basis sets for metals (Mn, Ca) and 6-31G(d) for other atoms (H, C, N, O). The low layer was calculated with Amber force eld. 70

EPR characterization
EPR spectra from the spinach PSII S 2 states in 30% glycerol buffer are shown in Fig. 1. Illumination of PSII membranes at 195 K results in the formation of the S 2 MLS. Under these illumination conditions, the dominant feature is the S 2 MLS that corresponds to the total spin (S total ) of 1/2 that arises from exchange interaction of one high-spin Mn III and three high-spin Mn IV , as has been intensively studied in the past. 9,25-33 While a weak, broad peak is also present in the region around g ¼ 4 ( Fig. 1 ((a) minus dark)), the small intensity of the signal shows that this species is nearly absent under our experimental conditions. When PSII membrane samples are illuminated at low temperature (140 K) in the absence of an IR lter, the photogeneration of the broad S 2 -g4 signal is observed, with a small S 2 -g2 signal ( Fig. 1 ((b) minus dark)). The amount of the S 2 -g2 in the sample illuminated at 140 K is approximately 20% of the intensity of S 2 -g2 signal from 195 K illuminated sample. The transition from the HS S 2 to LS S 2 occurs by increasing the temperature, which is supported by the reduction of the S 2 -g4 EPR signal and the increase of the S 2 -g2 signal when the 140 K NIR illuminated sample (S 2 -g4 dominant) is annealed to 200 K. To shows that there is interconversion between the S 2 -g4 and S 2 -g2 species by temperature, EPR data of the annealed sample were collected at 8 K. The S 2 -g2 signal of the annealed sample increased up to 70% level of the 195 K illuminated sample while the S 2 -g4 signal decreases down to $30% (Fig. 2). O 2 activity of the g2 rich and g4 rich spinach PSII It is known that PSII samples from spinach in the S 2 state in glycerol buffer have a dominant S 2 -g2 signal with only a trace of the S 2 -g4 signal for 195 K illumination, while the PSII in sucrose buffer have both S 2 -g2 signal and S 2 -g4 signal in almost 50 : 50 ratio. 41 We observe a similar trend in the EPR spectra of these samples are shown in Fig. 1 (glycerol) and Fig. S3 (sucrose) in the ESI. † To check the activity of the S 2 -g4-rich and S 2 -g2-rich PSII samples, the O 2 evolution activity of both samples are compared by dividing the same batch of PSII thylakoid samples Fig. 1 EPR spectra of PSII samples in glycerol illuminated for 10 minutes at (a) 195 K (blue) (b) 140 K with NIR (red) along with corresponding dark (grey) EPR spectra. The difference spectra are between the spectra after illumination and the spectra of the same darkadapted sample. The large intensity from Y D c in each spectrum has been removed for clarity ($3200 G). Spectrometer condition: microwave frequency, 9.22 GHz; field modulation amplitude, 32 G at 100 KHz; microwave power, 20 mW. The spectra are collected at 8 K.
into two parts and transferring one part into glycerol buffer, and the other into sucrose buffer [50 mM MES-NaOH (pH 6.0), 5 mM MgCl 2 , 5 mM CaCl 2 and 15 mM NaCl, 0.4 M sucrose]. The O 2 activity was very similar between the two samples, giving rates of 420 AE 10 mmol of O 2 per mg of Chl per h in glycerol buffer and 408 AE 10 mmol of O 2 per mg of Chl per h in sucrose buffer. These measurements were performed with three different sample preparations. The results shows that the number of the active centers is more or less the same in the two samples, while the fraction of centers that can be cryo-trapped in the S 2 -g4 or the S 2 -g2 spin states is signicantly different, depending on the buffer conditions.
Mn K-edge spectra Fig. 3a shows the Mn XANES spectra of the two S 2 spin states (HS S 2 and LS S 2 ), together with S 1 and S 3 states. In the presence of glycerol as a cryo-protectant, the majority of the PSII centers are in the LS S 2 state when illuminated at 195 K, as observed in the EPR spectra (see Fig. 1 ((a) minus dark)). On the other hand, illumination at 140 K generates a large fraction of the PSII centers in the HS S 2 state. Using the estimated ratio of S 2 -g2 MLS intensity between the samples under the two illumination conditions (195 K illuminated vs. the 140 K illuminated samples), it is inferred that a minor fraction ($20%) of LS S 2 state is present in the samples illuminated at 140 K. The corresponding amount of LS S 2 state spectrum was subtracted from the XAS spectrum of 140 K NIR illuminated sample to obtain the pure HS S 2 XAS spectrum. This is based on the assumption that 195 K illuminated and 140 K NIR illuminated samples consist of a linear combination of the HS and LS S 2 states. The pure LS S 2 XAS spectra were also obtained by the same method. The untreated XAS spectra from the 195 K illuminated and the 140 K NIR illuminated samples are shown in Fig. S4 in the ESI. † Interestingly, the XANES rising edge position of the HS S 2 state is slightly but noticeably lower in energy than that of the LS S 2 state as shown in Fig. 3a. While the edge positions of LS S 2 and S 3 states are very close, their spectral shapes are not exactly the same. This difference is more clearly seen in the 2 nd derivative spectra (Fig. 3a bottom). The inection point energy obtained from the 2 nd derivative XANES spectra are, 6552.11 eV (S 1 ), 6552.89 (S 2 HS), 6553.44 (S 2 LS), and 6553.71 (S 3 ). It is oen difficult to compare these numbers with literature values due to different procedures for generation of the 2 nd derivative spectra. Therefore, we compared XANES spectra of all S-states treated in the same way, to eliminate any ambiguity that may arise from such data treatment. We further note that the inection point energy could be a possible indicator of Mn charge density, although multiple-scattering effects in the XANES region could mask such changes when the structural changes are accompanied by oxidation state changes. For this reason, we cautiously state that the edge shi observed in the HS and LS S 2 state suggests a change in charge density of Mn in these two states. The HS S 2 state might be slightly lower in the effective positive charge density on Mn compared to the LS S 2 state. We conrmed that there is no indication of Mn II being released during the 140 K NIR illumination by monitoring the presence of the Mn II EPR signal since such an effect will also lower the Mn XANES edge position. Another potential cause for lowering the Mn edge position is the presence of a fraction of the S 1 state in the HS S 2 sample due to the low temp. illumination. We excluded this possibility based on the results of the annealing experiments (Fig. 2). Aer annealing the S 2 -g4 sample to 200 K, the multiline (g ¼ 2) spectra increased to 70% compared to the S 2 -g2 state spectra. On the other hand, the g ¼ 4 part of the annealed spectra was reduced to 30% of the S 2 -g4 spectra.
In addition, we investigated the Mn XANES pre-edge peaks of the two states, as it serves as another indicator of the effective charge density. The pre-edge spectra are slightly, but noticeably different in the LS S 2 and HS S 2 states (Fig. 3a inset). The preedge spectra were t with a pseudo-Voigt line with a 1 : 1 ratio of Lorentzian and Gaussian functions and the peak area was compared between these two states. The number of the pre-edge components and their positions were estimated by the 2 nd derivative spectra. The area of the pre-edge peak was 0.22 for LS S 2 and 0.24 for HS S 2 ( Fig. S5 and Table S1 in the ESI †). 71 While the slightly larger pre-edge area observed in HS S 2 may indicate a more distorted ligand environment in this state as compared to LS S 2 , the difference is rather small for drawing any concrete conclusions. Fig. 3b shows the EXAFS spectra of the two S 2 spin states, together with the S 1 and S 3 state spectra. A comparison of the HS and LS S 2 state spectra shows noticeable differences in the 2 nd Fourier transform (FT) peak width and intensity as well as the 3 rd FT peak intensity, which is signicantly higher in the HS S 2 state spectrum. The 2 nd FT peak corresponds to the di-m-oxo bridged Mn-Mn interactions around 2.7Å and the 3 rd FT peak arises from the contribution of mono-m-oxo bridged Mn-Mn and Mn-Ca interactions around 3.3Å. Such differences are also visible in the EXAFS oscillation in the k-space spectra (Fig. 3b  inset). Furthermore, both HS and LS S 2 spectra differ from the S 3 state spectrum, suggesting that the structural geometries in these three states are not the same. Detailed EXAFS analysis is discussed in the next section.

Mn EXAFS curve tting
Mn EXAFS curve tting of the HS and LS S 2 states were carried out to extract structural parameters of the Mn cluster in these states. Descriptions of the parameters used are provided in the ESI. † Fig. 4 shows t results, and the t parameters are summarized in Table 1. Structures for LS and HS S 2 state have been proposed previously based on EPR and quantum chemical calculations, 21 and we therefore used those as starting structural models for tting the EXAFS data.
The LS S 2 state ts well with the proposed open cubane-like structure. 16,21 In this structure, there are three short Mn-Mn interactions around 2.7-2.8Å and one long Mn-Mn interaction around 3.3Å (LS S 2 -t #1 in Fig. 4  In this model, the numbers of short and long Mn-Mn interactions and Mn-Ca interactions in the HS S 2 state remain the same as in the LS S 2 model. Therefore, the same parameters obtained from the LS S 2 t (LS S 2 -t #1) were used as starting parameters (HS S 2 -t #1). In the experimental spectrum of HS S 2 , the 3rd FT peak intensity increases noticeably, while the 2nd FT peak becomes narrower than that of the S 2 LS spectrum. In HS S 2 (t #1), the atomic distances remained similar to the initial parameters. The weaker FT peak II intensity and the stronger FT peak III intensity were compensated by Debye Waller factors. While FT peak II could be t with three Mn-Mn interactions with an average distance of 2.73Å, the presence of a longer $2.8Å Mn-Mn interaction was not preferable. This observation suggests that a complete cubane with Mn D1 , Mn C2 , Mn B3 , and Ca cannot be formed in this state; as such a structure typically will have high distance heterogeneity in the range of 2.7 to 2.8Å. We also tested a hypothetical model where one of the three Mn-Mn short distances elongates, as it would in the presence of a mono-m-oxo-like bridge, i.e. giving a short (2.7-2.8 A) and long ($3.3Å) Mn-Mn distances ratio of 1 : 1 (HS S 2 -t #2). The t quality was improved by 50% for this model. In a later section, we further discuss (a) whether such a structure is possible, (b) the interconversion between HS and LS form in the S 2 state, and (c) the relation of the two S 2 isomers to the formation of the S 3 state.

Mn Kb 1,3 /Kb 0 XES
XES Kb 1,3 /Kb 0 transitions provide complementary information to XANES, by probing the Mn 3p to 1s emission process that is sensitive to the number of unpaired 3d electrons through 3d/3p spin exchange interactions. We measured the Kb 1,3 XES spectra of the LS and HS S 2 states. We observe a slight shi in the Kb 1,3 emission spectra between the LS and HS S 2 states (Fig. 5). Fig. S6 † shows raw and smoothed data for the Kb 1,3 /Kb 0 XES transitions for LS and HS S 2 states along with the residual plot. The spectra were smoothed using a sum of nonlinear lineshapes. We observe that the spectrum of the LS S 2 state is at a slightly lower energy than the HS S 2 state, as becomes evident in the difference spectra of the LS and HS S 2 states (Fig. 5). With an increase in the oxidation state of Mn, fewer unpaired 3d valence electrons can interact with the 3p hole, leading to a decrease in the magnitude of 3p-3d exchange interaction, which results in the Kb 1,3 emission spectra shiing to a lower energy. Hence, the LS S 2 state might have slightly higher effective positive charge density on Mn compared to the HS S 2 . This is in agreement with the changes observed in the XANES spectra reported in the earlier section.

Discussion
The nature of the two isomers in the S 2 state of higher plant PSII was investigated using X-ray spectroscopy with the support of EPR spectroscopy. We have observed the XAS (XANES and EXAFS) and XES spectral changes between the HS and LS S 2 species, and here we discuss possible structural models and the transition phenomena, with a comparison to the proposed models in the literature.
First, the fact that the population of the LS and HS S 2 species seems to shi depending on the buffer conditions implies that some variation of the XANES edge positions for the S 2 state shown in the literature may contain this effect since this kind of isomorphism was known but not differentiated until recently.  Table 1.  Nevertheless, the LS S 2 state should be the dominant spin state in the literature studies, when glycerol buffer is used. 7,11,16,72 Structural models of the high-spin and low-spin S 2 states The differences observed in the XAS spectra provide evidence for the different electronic structure and the metal-metal atomic distances in the HS and LS S 2 states. The HS and LS structural models that involve interconversion of the Mn III position in these two species have been proposed by Pantazis et al. 21 and Isobe et al. 47 based on EPR results and from quantum chemical calculations. In the LS S 2 , Mn III is located at the Mn D1 position that is ligated to His332 of D1 chain, while in the HS S 2 state Mn III is at the Mn A4 position of the cluster, which has two water ligands (Fig. S7 †). The EXAFS curve tting results match with the result that the LS S 2 state is an open cubane-like structure in which Mn III is at the Mn D1 site, as previously suggested. While in general EXAFS is not a technique that can be used to conclusively point to a single, unique, structural model, 73 the observed peak intensity change in the FT EXAFS spectra is clear evidence of the structural differences between the two spin states. The EXAFS curve tting results suggests that the ratio of short and long Mn-Mn interactions may be different in the HS and LS S 2 forms. The HS S 2 form could exhibit two short and two long Mn-Mn interactions, in which the central oxygen (O 5 ) is required to be nearly at the center between Mn A4 and Mn D1 . Fig. 6 shows the possible structural changes of the HS and LS S 2 states, based on the EXAFS observation. In these models, we kept the formal oxidation state assignment of each Mn as suggested by Pantazis et al. 21 and Isobe et al., 47 in which Mn III is located at the Mn D1 in LS S 2 form, while it is at the Mn A4 in HS form. Within our current knowledge, it is reasonable to think that the S total ¼ 1/2 being formed with anti-ferromagnetically-coupled Mn III and Mn IV along with two anti-ferromagnetically-coupled Mn IV , and S total ¼ 5/2 being formed with three ferro-magnetically coupled Mn IV in the cubane moiety with anti-ferromagnetically-coupled Mn III at Mn A4 . The XANES edge position of the HS S 2 appears slightly lower than that of the LS form, suggesting that the effective charge density of the HS form may be lower than that of LS. This observation is also supported by the Mn Kb 1,3 XES results. As formal oxidation state and number of unpaired spins should be the same between HS and LS S 2 state (although the total number of spin differs due to exchange coupling of the four Mn), one speculation is that the different protonation states of the ligand oxygen or geometry of the cluster in these two states shis the effective charge density on Mn. If the protonation state of a ligand oxygen is different, O 5 located between Mn D1 and Mn A4 is one possible candidate that could weaken two Mn-Mn interactions and therefore result in two long (>3.0Å) Mn-Mn interactions. The deprotonated O 5 in the LS S 2 form is conrmed by EXAFS that shows all three distances, Mn A4 -Mn B3 , Mn B3 -Mn C2 , Mn C2 -Mn D1 , to be around $2.74Å. 16 On the contrary, if O 5 is protonated in the HS form, both Mn 1D -Mn 3B and Mn 4D -Mn 3B will be elongated. The difference in oxo-bridge protonation state may affect the effective charge density on Mn. However, the S 2 structure with protonated O 5 in the  Mn III,IV,IV,IV   4 oxidation state is expected to be energetically much higher ($25 to 30 kcal) compared to the deprotonated O 5 (Fig. 7). This is also observed in a recent theoretical study by Krewald et al. 10 Thus, this S 2 state cannot exist as a stable form, unless there are other factors that stabilize such a structure. Hence, the difference in the geometry between the two states may be the reason for the shi in effective charge density in the HS S 2 state.
As shown in Fig. 3a, the EXAFS spectra of HS and LS S 2 states and the S 3 state are all different. This implies that the atomic  Our group has proposed a S 3 state structure to comprise a closed-cubane structural motif, based on the EXAFS studies (Fig. 6) that showed an elongation of Mn-Mn distances within the cubane-motif. The rationale of our proposal is from inorganic model compound studies where the elongation of metal-metal distances is observed when the cubane is formed. 74,75 A similar elongation of the metal-metal distances is visible in the S 3 EXAFS spectrum (i.e. while all three di-m-oxo Mn-Mn distances are $2.74Å in the S 2 state, it is more distributed over the range of 2.72-2.82Å in the S 3 state). 16 In the current study, we observed that the HS S 2 EXAFS is different from that of S 3 state, suggesting that the geometries of HS S 2 and S 3 state structures are likely different. One possibility is that, as depicted in Fig. 6, in LS S 2 O 5 is bound to Mn A4 , while in HS S 2 it is more or less equidistant from Mn A4 and Mn D1 . In S 3 state, the O 5 position is shied to Mn D1 . However, an uncertainty remains if the S 3 state has a large heterogeneity (EPR active and EPR silent species), as suggested by Boussac et al. 45 and Cox et al. 20 based on EPR studies. Then the EXAFS spectrum under our experimental condition could be a mixture of the EPR active and the EPR silent species. Further studies of this potential heterogeneity are necessary. Also, we cannot eliminate the possibility of the inserted water model suggested by Cox et al. 20 and Siegbahn et al., 52 if the elongation of the metal-metal distances in the S 3 state occurs by the expansion of the open-cubane moiety due to the effect of newly inserted water into the open Mn1 site (Fig. 6).
Transition process between the S 1 , S 2 , to S 3 states Currently, the radiation-damage-free dark state structure published by Suga et al. with 1.95Å resolution 4 serves as the most reliable foundation for considering possible distance changes in the higher S-states. Mn-Mn distances and number of interactions in the crystal structure matches reasonably well with the structural parameters obtained from earlier EXAFS studies of the S 1 state (i.e. there are three short 2.7-2.8Å Mn-Mn interactions and one long $3.3Å Mn-Mn interaction in addition to three Mn-Ca interactions). 16,59,60 One of the remaining uncertainties in the S 1 structure, however, is the protonation state of O 5 . Suga et al. suggest O 5 to be protonated, based on the long Mn A4 -O 5 and Mn D1 -O 5 distances 4 in which the O 5 position was obtained from the omit map. However, there is another explanation that proposes a deprotonated O 5 in the S 1 state. In the following section, we discuss these two possibilities, case (1) for protonated O 5 and case (2) for deprotonated O 5 in the S 1 state, in relation to the S 2 state formation (Fig. 8).
In case (1), the O 5 proton needs to move away from the OEC upon S 1 to S 2 (via 200 K illumination or RT laser ash) transition, as it is most likely that O 5 is deprotonated in the LS S 2 form. If the protonation state is different between HS and LS S 2 states, one possible reason for this to occur is that O 5 is protonated in the S 1 state, and remains as it is in the HS S 2 (case (1), S 2 HS (a) in Fig. 8). Small structural or chemical changes that are required for the proton motion could be prohibited under the illumination condition at 140 K with NIR illumination while going to the 'native' S 2 state may require an illumination at higher temperature. As previously discussed, the S 2 structure with protonated O 5 is energetically much higher than the deprotonated one (Fig. 7), which makes this model highly unlikely. Moreover, it is difficult to rationalize an observed reversibility between LS S 2 and HS S 2 form with the O 5 -protonated model. Therefore, the O 5 proton in the S 1 state needs to move to a nearby ligand in the S 1 to HS S 2 transition (case (1), S 2 HS (b) in Fig. 8). The EXAFS curve tting results in this study suggest that there could be two elongated Mn-Mn interactions longer than 3Å, and two di-m-oxo bridged Mn-Mn interactions at $2.7Å. Such a structure is different from the one proposed from the theoretical studies. 21,47 Although we cannot completely rule out a model that contains three 2.7Å Mn-Mn and one 3.3Å Mn-Mn, a complete cubane formation at HS S 2 state seems to be less likely from the current EXAFS data due to decreased heterogeneity around 2.7Å interactions. The structural differences in HS and LS S 2 states could be reasoned by O 5 position moving closer to Mn 1D upon oxidation of Mn 1D from Mn III to Mn IV during the S 1 to HS S 2 state transition, but Mn 1D -O 5 is weakly bound with a distance longer than 2Å. In case (2) in which O 5 is deprotonated in the S 1 state, a similar argument to case (1) is applicable (Fig. 8).
The question arises whether the HS S 2 state populated by 140 K NIR illumination, that we observed in this study, is the same as the S 2 state with enhanced g ¼ 4 signals that is observed at higher temperature illumination (200 K illumination or laser ash at room temperature) under certain buffer conditions (e.g. with sucrose buffer) or additional chemical treatments (e.g. with a substitution of Ca 2+ by Sr 2+ ). Moreover, whether the HS and LS S 2 species is populated under physiological conditions and such heterogeneity plays a role in the S-state advancement and the catalytic reaction still remains unanswered. While this question requires the room temperature study, here we speculate if the g ¼ 4 species made under different conditions are always the same. It has been suggested that any changes that disturb the hydrogen bond network around the OEC inuence the electronic properties of Mn A4 through the W1 and/or W2 ligands that are ligated to this Mn. 44 There is a trend that any changes in the electron donating effect in ligands of Mn A4 favor this site to have a lower Mn oxidation state (i.e. Mn III ). 44 The high-resolution crystal structure of the T. vulcanus PSII 3 has shown that the hydrogen bond network around OEC is extended from W1 and W2 of Mn A4 to D1-D61 (Fig. S7 †). Pokhrel et al. suggested that this hydrogen bonding network of W1 and W2 determines the equilibrium between the g ¼ 2 and g ¼ 4 forms of the OEC in the S 2 state. 44 As a consequence, we can speculate that the HS S 2 EPR signal around the g ¼ 4 region changes and its exact g-value (i.e. the zero-eld splitting) is highly affected by such effects. The fact that we observe changes in HS S 2 state from LS S 2 state implies that the OEC structure that causes the g $ 4 EPR signal, whether by buffer contents or chemical treatments, is different from that of the g ¼ 2 species. A question still remains if all the g ¼ 4 species observed in the EPR spectra represents the same structure. This may not be the case if some of the g ¼ 4 signals under certain conditions are not from the S total ¼ 5/2 ground This journal is © The Royal Society of Chemistry 2016 state, but from the excited state of a different ground state spin conguration, thus these signals may represent other ground state spin congurations. For example, the temperaturedependent EPR signal from a weakly-coupled Mn(III/IV) dimer core has been reported in a model system. 76 In this case, the 16-line spectrum of the ground state (S total ¼ 1/2) and the g $ 5 spectrum from the excited state (S total ¼ 3/2) are observed. No observation of temperature dependence when going from S 2 -g2 to S 2 -g4 EPR signals in PSII suggest that both the signals studied here arise from the ground state. Also, since the proposed models available are based on EPR measurements that are all measured at low temperature, the S 2 species that exist in room temperature still remains unknown.
In the current study, we use spinach PSII to investigate the S 2 -g2 and S 2 -g4 states. Unlike spinach PSII in which the S 2 -g4 signal is observed under certain buffer conditions, there are additional EPR signals along with the S 2 -g4 signal in Synechococcus PSII wild type. In these samples, g ¼ 6 to 10 signals are also observed when samples are IR illuminated, and the pure g ¼ 4 signal is only observed when the native Ca 2+ or Cl À is substituted. Nevertheless, the intensity of these low EPR eld signals are weak, and it suggests that S 2 -g2 state is the dominant species in the wild type. The reason for such species dependence is not known, as the crystal structure is only available for the Synechococcus PSII. However, it is likely due to small differences in the hydrogen-bonding network that extends from W1 and W2 of the OEC to the water channel leads to subtle differences in the electronic structure.

Conclusions
We have investigated the structure and the electronic structure of the two spin-isomers in the PSII S 2 intermediate states. The XAS data suggests different structural congurations for the HS and LS S 2 states. It also suggests that their structures are different from the subsequent S 3 state. Whether the HS S 2 state serves as an intermediate state between the LS S 2 and the S 3 state as proposed from theoretical modeling is still an open question, and the high-resolution crystal structure of these intermediate states, possibly at the room temperature is necessary to resolve it. Despite such noticeable structural differences during the catalytic pathway due to the likely modication of the hydrogenbonding network, the O 2 evolution activity remains similar for both spin forms. This implies a certain exibility of the OEC in its geometric and electronic structure under physiological conditions, although one state may be more preferable than the other.