Structure and Orientation of the Mn4Ca Cluster in Plant Photosystem II Membranes Studied by Polarized Range-extended X-ray Absorption Spectroscopy*♦

X-ray absorption spectroscopy has provided important insights into the structure and function of the Mn4Ca cluster in the oxygen-evolving complex of Photosystem II (PS II). The range of manganese extended x-ray absorption fine structure data collected from PS II until now has been, however, limited by the presence of iron in PS II. Using a crystal spectrometer with high energy resolution to detect solely the manganese Kα fluorescence, we are able to extend the extended x-ray absorption fine structure range beyond the onset of the iron absorption edge. This results in improvement in resolution of the manganese-backscatterer distances in PS II from 0.14 to 0.09Å. The high resolution data obtained from oriented spinach PS II membranes in the S1 state show that there are three di-μ-oxo-bridged manganese-manganese distances of ∼2.7 and ∼2.8Å in a 2:1 ratio and that these three manganese-manganese vectors are aligned at an average orientation of ∼60° relative to the membrane normal. Furthermore, we are able to observe the separation of the Fourier peaks corresponding to the ∼3.2Å manganese-manganese and the ∼3.4Å manganese-calcium interactions in oriented PS II samples and determine their orientation relative to the membrane normal. The average of the manganese-calcium vectors at ∼3.4Å is aligned along the membrane normal, while the ∼3.2Å manganese-manganese vector is oriented near the membrane plane. A comparison of this structural information with the proposed Mn4Ca cluster models based on spectroscopic and diffraction data provides input for refining and selecting among these models.

Photosynthesis by green plants, algae, and cyanobacteria provides essentially all of the dioxygen in the biosphere as a byproduct of the electron transfer processes utilizing water as the ultimate electron source: Water oxidation is a light-driven reaction that is catalyzed by an oxygen-evolving complex (OEC) 4 of Photosystem II (PS II) (1)(2)(3)(4). The active site of the OEC is known to be a proteinbound complex containing four manganese and one calcium atom. This complex cycles through a series of five intermediate redox states that are referred to as S states (S 0 to S 4 ) (5). The S state transitions are driven by successive light-induced oneelectron oxidations of the PS II reaction center. In each step the complex accumulates oxidizing equivalents until dioxygen is released during the spontaneous return from S 4 to S 0 .
Many of the proposed mechanisms of water oxidation depend critically on knowledge of the Mn 4 Ca cluster structure. To date, structural models of the OEC complex have been suggested based on EPR techniques (6 -9), x-ray absorption spectroscopy (XAS) (10 -14), x-ray diffraction (XRD) (15)(16)(17), and infrared spectroscopy (Fourier transform infrared) (18). The XRD studies (3.0 -3.8 Å resolution) have located the Mn 4 Ca cluster in the density map (16,17) and confirmed the presence of calcium in the OEC cluster, as had been shown previously by EPR (19 -21) and by extended x-ray absorption fine structure (EXAFS) spectroscopy (22,23). A recent XAS study showed that the OEC complex is very susceptible to reduction and disruption during x-ray exposure, under the conditions used in collecting the published XRD data (24). Consequently, the precise location of the manganese and calcium atoms has not been reliably established within active OEC centers by XRD, as acknowledged in the most recent study (17).
Manganese XAS enables a detailed analysis of the Mn 4 Ca cluster in the OEC. X-ray absorption near-edge structure (XANES) contains information on the electronic structure and changes in oxidation states of the manganese that accompany S state transitions (25). EXAFS allows for a precise determination of manganese-backscatterer distances (26) and is, furthermore, very sensitive for establishing the permissible x-ray dose (24).
Recent EXAFS studies of the Mn 4 Ca cluster of the OEC have led to the conclusion that there are three manganese-manganese vectors in the range 2.7-2.9 Å reflecting di--oxo-bridged manganese atom pairs (27), one manganese-manganese vector at 3.3 Å, and one or two manganese-calcium vectors at 3.4 Å (22,23). The ability of the EXAFS technique to determine the presence of similar backscatterers at closely separated distances, ⌬R, is dependent on ⌬k, the width of the k-space EXAFS data set (Å Ϫ1 ) (for details see supplemental data). The presence of iron, partly an integral component of the OEC and partly from adventitious sources, restricts the useful range in conventional manganese EXAFS to ϳ550 eV (⌬k ϭ 12.5 Å Ϫ1 ) above the manganese edge (iron K-edge at 7120 eV). Consequently, in PS II the manganese-manganese distance resolution is limited to ⌬r ϭ 0.14 Å, meaning that two manganese-manganese vectors should differ greater than 0.14 Å to be resolved. Improvement in manganese-backscatterer distance resolution is critical for precise structural and mechanistic studies of the OEC. Conventional EXAFS spectra of PS II samples are based on the detection of the manganese K␣ 1,2 fluorescence (ϳ5.9 keV) using a solid state detector of at best 150 -200 eV from full width at half-maximum resolution (28 -30), making it impossible to discriminate completely against the Fe K␣ 1,2 fluorescence at 6.4 keV (see Fig. 1). This limitation can be overcome by utilizing a crystal monochromator with high resolution (ϳ1 eV) for the fluorescence detection (31,32). Recently we showed that manganese EXAFS of the OEC can be collected up to ϳ1000 eV (k ϭ 16.1 Å Ϫ1 ) above the manganese K-edge (27), improving the manganese-backscatterer distance resolution to 0.09 Å. This enabled us to study the heterogeneity in the manganesemanganese distances of solution samples in the S 1 and S 2 states, providing evidence for three manganese-manganese distances of ϳ2.7 and ϳ2.8 Å present in a 2:1 ratio (27). These improvements in determining the structural parameters are important for choosing among different proposed structural models, and they provide an opportunity for investigating the changes that occur as the Mn 4 Ca catalyst cycles through the S states.
Additional geometric information about the spatial arrangement of manganese-backscatter vectors can be obtained if oriented PS II samples, such as oriented membranes or single crystals, are used for the measurement of EXAFS dichroism with linearly polarized synchrotron x-rays. Collection of the polarized EXAFS spectra on oriented PS II membranes at different angles between the membrane normal and the x-ray electric field vector results in dichroism that depends on how the particular absorber manganese-backscatter vector is aligned with respect to the electric field of the x-ray beam. Thus, the average orientation of a particular manganese-backscatter vector relative to the membrane normal and the average number of scatterers per absorbing atom can be determined (33)(34)(35).
Previous studies on oriented native and NH 3 -treated PS II membranes were based on conventional EXAFS. Average angles relative to the membrane normal of ϳ60°for the ϳ2.7 Å vectors (di--oxo-bridged Mn 2 units) and ϳ43°for the ϳ3.3 Å vectors (superposition of mono--oxo-bridged manganesemanganese and manganese-calcium vectors) have been reported in two studies (34,35), whereas another study reported an average angle of 80 Ϯ 10°for the ϳ2.7 Å vectors without providing results for the ϳ3.3 Å vector (36). Because of the limited resolution, conventional EXAFS is not able to determine the orientations of the individual manganese-manganese and manganese-calcium vectors in the 3.2-3.4 Å region. In a complementary study, strontium K-edge polarized EXAFS of strontium-reactivated PS II membranes was used to predict the manganese-calcium orientation. It showed a lower and upper limit of 0 and 23°, respectively, for the average angle between the manganese-strontium vector(s) and the membrane normal and yielded an isotropic coordination number of manganese neighbors to strontium of either one or two (23). A recent polarized x-ray absorption spectroscopy study of PS II single crystals from cyanobacteria, using an x-ray dose below the threshold of damage, has derived feasible structures for the Mn 4 Ca cluster and the orientation of the cluster in the PS II crystal (14).
In this work, we applied range-extended EXAFS to study the dichroism of the Mn 4 Ca cluster in oriented PS II membranes from spinach chloroplasts. The study shows: (i) the separation of the manganese-manganese (ϳ3.2 Å) and manganese-calcium (ϳ3.4 Å) vectors, which allows independent analysis of FIGURE 1. Range-extended x-ray absorption spectroscopy. Left, x-ray fluorescence of manganese and iron; Above, manganese K␣ 1 and K␣ 2 fluorescence peaks, with natural line width of ϳ5 eV, split by 11 eV. The multicrystal monochromator with 1-eV resolution is tuned to the manganese K␣ 1 peak. Below, fluorescence peaks of manganese and iron as detected using germanium detector. The fluorescence peaks are convoluted with the electronic window resolution of 150 -200 eV of the germanium detector. This method of detection cannot resolve manganese K␣ 1 and K␣ 2 fluorescence peaks. Note different energy scales for the schemes shown above and below. Iron is an obligatory element in functional PS II complexes. Right, comparison of the PS II manganese K-edge EXAFS spectrum from an S 1 state PS II sample obtained with a traditional 30-element energy-discriminating germanium detector with a spectrum collected using the high resolution crystal monochromator. Use of the high resolution detector eliminates the interference of iron and removes the limit of the energy range for manganese EXAFS data collection.
their orientation relative to the membrane normal; (ii) the determination of the dichroism characteristics of the three short manganese-manganese vectors (two at 2.7 Å and one at 2.8 Å) and their orientation in the PS II membrane. These results are used to discuss the structure and orientation of the Mn 4 Ca cluster in the PS II membrane.

MATERIALS AND METHODS
Sample Preparation and Characterization-PS II samples were prepared from spinach as previously described (37). They typically contain 4 manganese per 200 -250 chlorophylls. The oxygen evolution rates for the PS II samples used in this study are between 400 and 500 mol O 2 /(mg chlorophylls ⅐ h). The membranes were resuspended in 50 mM MES buffer, pH ϭ 6.0, containing 0.4 M sucrose and 5 mM CaCl 2 and pelleted by centrifugation at 4°C (39,000 ϫ g, 1 h). One or two drops of 50 mM MES buffer were added to the pellet, and the resulting paste was painted onto Mylar tape. The PS II membranes were dried under a stream of cold nitrogen gas at 4°C in the dark for ϳ1 h, as described previously (38). This process was repeated five to seven times to generate samples with a sufficiently thick sample layer for the x-ray absorption experiment.
The paint-and-dry cycles produce one-dimensionally ordered samples with a preferred orientation of the PS II membrane normal perpendicular to the substrate surface. The extent of orientation (mosaic spread, which is the half-width of the Gaussian distribution of the angle of the membrane normal to the substrate normal in the PS II samples) was assessed from the angle dependence of the Tyr D ox and cytochrome b 559 EPR signals (see supplemental data). X-band EPR spectroscopy was performed with a Varian E-109 spectrometer, a standard TE 102 cavity, and an Air Products liquid helium cryostat. The samples used in this study displayed a mosaic spread of 15-20°. After drying the samples, their integrity was assayed by monitoring the amount of S 2 multiline signal formed upon sample illumination at 195 K. The amplitude of the manganese signal was the same as that obtained from randomly oriented membranes at a similar concentration. Manganese K-edge XANES spectra of oriented samples can be used to reconstruct the solution XANES spectrum, which is very sensitive to the manganese oxidation state and damaged PS II centers containing Mn 2ϩ . The two spectra are indistinguishable, indicating the intact state of the oriented samples.
Data Collection-The x-ray spectra were recorded on the BioCAT undulator beamline 18-ID at the Advanced Photon Source (Argonne, IL). The energy of the incident x-rays was selected by means of a nitrogen-cooled silicon double-crystal monochromator at (111) orientation, yielding ϳ1-eV resolution. The monochromator energy was calibrated using the preedge peak energy of KMnO 4 at 6543.3 eV. Higher harmonics from the monochromator were rejected by the focusing mirror. The incident beam intensity was set to ϳ4 ϫ 10 12 photons/s (ϳ60% of the flux available at 18-ID) at a beam size of 1 ϫ 2 mm 2 . This allowed us to perform fast EXAFS scans in continuous mode before the onset of radiation damage; 15 s per sweep, energy range 6500 to 7500 eV in 1 eV increments, one sweep per spot on sample, 15-20 different spots per sample depending on orientation, ϳ100 samples per orientation. The EXAFS scan parameters were chosen subsequent to and on the basis of a radiation damage study of the samples. XANES spectra were collected under identical conditions (number of photons, time and temperature that were used for subsequent EXAFS measurements), and the inflection point energy of the XANES spectra was monitored for any shifts to establish the safe x-ray dose (24). Radiation damage measurements were determined for both 15 and 75°orientations of the samples used in the study and were repeated each time we had x-ray beamtime at the synchrotron sources to account for any changes in the beam characteristics. A second 15-s sweep of the EXAFS for some samples was collected and the spectra were unchanged, providing additional confirmation of the absence of radiation damage. To avoid unnecessary sample exposure, a beam shutter was automatically inserted when data were not being collected. The manganese K␣ fluorescence was detected by four spherically bent germanium analyzers (8.9 cm diameter, 85 cm radius of curvature) using the (333) Bragg reflection in a Rowland geometry. The analyzer energy was tuned to the manganese K␣ 1 peak at 5899 eV at a Bragg angle of 74.84°. A nitrogen-cooled solid state (germanium) detector was placed at the common focus of the four crystals on the intersecting Rowland circles. The analyzer bandwidth of 0.8 eV was determined by measuring the elastically scattered peak.
Experimental procedures and limitations for measuring range-extended EXAFS past multiple K-or L-edges, and the design and operation of the spectrometer have been described previously (32,39). All samples were measured below 10 K in a liquid He cooled cryostat (Oxford CF1208).
Data Analysis-For each EXAFS scan, the energy was calibrated using the KMnO 4 pre-edge reference peak (6543.3 eV), and the intensity was normalized by I 0 before averaging. Approximately 1000 scans were averaged for each orientation of PS II membranes relative to the x-ray e-vector with a custom Matlab program. Data reduction of the EXAFS spectra was done as described previously (10,40). Curve fitting was performed using ab initio calculated phases and amplitudes from the FEFF8 program from the University of Washington (41). These phases and amplitudes were used in the EXAFS Equation 1, which is described below and contains a sinusoidal function that gives the distance and an amplitude function that contains information about the scattering atom and the number of such neighboring atoms.
The neighboring atoms to the central atom(s) are divided into j shells, with all atoms with the same atomic number and distance from the central atom grouped into a single shell. Within each shell, the coordination number N j denotes the number of neighboring atoms in shell j at a distance of R j from the central atom, i. f effj is the ab initio amplitude function for shell j, and the Debye-Waller term e Ϫ2j 2 k 2 accounts for damping due to both static and thermal disorder in absorber-back-scatterer distances. The mean free path term e Ϫ2Rj/j(k) reflects losses due to inelastic scattering, where j (k) is the electron mean free path. The oscillations in the EXAFS spectrum are reflected in the sinusoidal term sin(2kR j ϩ ␣ ij (k)), where ␣ ij (k) is the ab initio phase function for shell j. This sinusoidal term shows the direct relation between the frequency of the EXAFS oscillations in k-space and the absorber-backscatterer distance. The EXAFS equation (Equation 1) was used to fit the experimental Fourier isolates using N, R, and 2 as variable parameters. Fit details and evaluation of fit qualities are given in the supplemental data.
The spatial resolution in EXAFS is inversely related to the spectral range. Several formulas can be found in the EXAFS literature describing the resolution limits of the method, such as ⌬R⌬k Ϸ 1, ⌬Rk max ϭ /2 and ⌬R⌬k ϭ /2 (40,42); for more details see the supplemental data (40).
For a detailed explanation of the theory of polarized EXAFS see the supplemental data. Angle is the angle between the x-ray e-vector and the membrane normal, and denotes the relative orientation of the manganese-backscatterer (manganese-manganese or manganese-calcium) vector of interest to the membrane normal.

RESULTS
Manganese X-ray Absorption Spectra-X-ray absorption manganese K-edge spectra of oriented membranes in the S 1 state are shown in Fig. 2. Data were collected for two orientations in which the sample normal is placed at either 15 or 75°to the direction of the e-vector of the polarized x-rays. The edge positions and post-edge shape exhibit a marked angle dependence, as reported previously for oriented PS II membranes (34). Dichroism of the manganese K-edge spectra is even more obvious in the second-derivative spectra (Fig. 2, bottom). The powder manganese XANES spectrum created from the spectra collected from oriented membranes at two different orientations is identical to that obtained from a frozen solution sample. This result provides independent confirmation that the oriented samples are intact and not damaged by x-rays.
The k 3 -weighted EXAFS spectra of the S 1 state of PS II oriented with the membrane normal at 15 or 75°to the x-ray e-vector are shown in Fig. 3. The spectra are distinctly dichroic. The region of the photoelectron wavevector from 3.5 to 11.5 Å Ϫ1 (denoted by the dashed line in Fig. 3), which is accessible by conventional EXAFS, agrees with results published earlier for oriented PS II membranes in the S 1 state (34). Clear differences can be seen in the spectra between the two orientations. Fig. 4A shows the Fourier transforms of the range-extended EXAFS (k 3 -weighted) at 15 or 75°. The Fourier transforms exhibit well defined peaks, labeled I, II, IIIA, and IIIB, corresponding to the shells of backscatterers at different "apparent" distances, RЈ, from the manganese absorber. The apparent distance is shorter than the actual distance due to a phase shift induced by the interaction of the given absorber-scatterer pair with the photoelectron. For comparison, the Fourier transforms of the range-extended EXAFS spectra of both orientations, but truncated at 11.5 Å Ϫ1 , are shown in Fig. 4B. Significant improvement in spectral resolution is observed for the range-extended EXAFS data (Fig. 4A). Increased spectral resolution reveals the orientation dependence of peaks II and IIIA and IIIB. The intensity of peak II, which consists of three manganese-manganese distances at 2.7 and 2.8 Å (see below) changes significantly between 15 and 75°, with higher intensity at 75°. Peak III shows a complex nature containing at least two peaks, IIIA and IIIB, with distances of 3.2 and 3.4 Å, respec-  tively. Peak IIIA is more intense at 75°but has a decreased intensity at 15°. Peak IIIB is more intense at 15°but is close to the noise level at 75°.
Previous calcium EXAFS studies of native PS II (22), strontium EXAFS of strontium-substituted PS II (23), and manganese EXAFS of calcium-depleted PS II (43) have demonstrated the contribution of both manganese-manganese (ϳ3.2 Å) and manganese-calcium (ϳ3.4 Å) vectors to Fourier transform (FT) peak III. The current observation (Fig. 4A) shows the distinct manganese-manganese and manganese-calcium vectors contributing to peak III at different distances and orientations. An assignment of peak IIIB to the manganese-calcium vectors can be made, taking into account the longer distance and different dichroism (oriented predominantly along the membrane normal). The dichroic behavior of peak IIIB (see below) is similar to that reported previously for the manganese-strontium vectors (23). In a like manner peak IIIA, which has a maximum at a shorter distance, can be assigned to the manganese-manganese vector. The intensity of this peak remains sizable also at the 15°o rientation, and further analysis (see below) is needed to evaluate the orientation of this vector relative to the membrane normal.
The significant decrease of the half-width of the EXAFS FT peaks obtained in the range-extended EXAFS experiments (compare Fig. 4, A and B) results in good separation of peaks I and II. When phase correction is applied in the Fourier transform of the range-extended EXAFS (k 3 -weighted) at 15 or 75°, an additional peak, termed IЈ, is seen in Fig. 4C. Manganese-oxygen or manganese-chlorine interactions may be expected at a distance of ϳ2.2 Å (44), although any assignment of this feature must await further studies. Range-extended EXAFS on PS II preparations with bromine substituted for chlorine has the potential for clarifying the involvement of the halide cofactor in the OEC.
The Fourier transforms shown in Fig. 4, A and C, provide the basis for drawing qualitative conclusions about possible distances in the manganese-backscatterer pairs and their preferential orientations relative to the membrane normal. Reliable quantitative results can be obtained by fitting the experimental data using the EXAFS Equation 1, as described under "Materials and Methods" and in the supplemental data. The assignment of each peak and a detailed analysis of the actual distance and orientation of each vector are described below. We will concentrate on FT peaks II and III, because these are from manganese-manganese and manganese-calcium backscattering and provide the most reliable information about the Mn 4 Ca structure and orientation.
Curve Fitting of EXAFS FT Peak II-Fits of Fourier peak II were carried out both separately and in conjunction with peak III (peaks II ϩ III). The large differences in amplitude and envelope shapes of the oscillation of the peak II (and III) isolates at the two different orientations reflect the dichroism observed in the Fourier transform amplitudes (Fig. 5). The quality of the fits is judged using the fit error parameters ⌽ and ⑀ 2 ; to know how they are determined see the supplemental data (note that ⑀ 2 is normalized to the number of fit parameters). Fit error parameters reflect the deviation between the simulations and isolates. Fits 1-4 in Table 1 show the results from fitting one and two manganese-manganese shells to peak II at the 15 and 75°orientations of the S 1 state. Addition of the second manganese-manganese shell (fits 3 and 4 in Table 1) results in considerable improvement of the fit quality. With two different manganesemanganese distances at 2.7 and 2.8 Å, the fit error ⌽ and ⑀ 2 decreased by ϳ40% for the two-shell fit relative to the one-shell fit. In our previous range-extended EXAFS study with isotropic solution samples we concluded that peak II is best interpreted as consisting of 2.7 Å and 2.8 Å manganese-manganese vectors (di--oxo-bridged manganese-manganese moieties) with a 2:1 ratio in both S 1 and S 2 states (27). Our present results on oriented samples support this conclusion.
The N app values are higher for the 75°orientation compared with those for the 15°orientation. The orientation dependence of data extracted with a one-shell fit of Peak II (Tables 1 and . Fourier transforms of the EXAFS spectra from oriented PS II membranes. A, FT of manganese K-edge EXAFS spectra (Fig. 3) from oriented PS II membrane samples in the S 1 state obtained with a high resolution spectrometer (range-extended EXAFS) at orientations of 15°(solid line) and 75°(dashed line) of the membrane normal with respect to the x-ray e-vector. The k range was 3.5-15.2 Å Ϫ1 . Fourier peaks in A and B appear at an apparent distance RЈ that is shorter than the actual distance R by ϳ0.5 Å due to a phase shift. B, Fourier transforms of the same data as in A, but the k range was truncated at 11.5 Å Ϫ1 for comparison with earlier published polarized EXAFS data obtained with conventional EXAFS. C, same as in A, but a phase correction was done using the EXAFSPAK suite of programs by Drs. Graham George and Ingrid Pickering (Stanford Synchrotron Radiation Laboratory), which results in the conversion of the apparent distance RЈ into an approximate real distance R.
supplemental Table S1, fits 1 and 2) results in N iso ϭ 1.3 Ϯ 0.3, which corresponds to two or three manganese-manganese interactions at an average angle of ͗͘ ϭ 63 Ϯ 5°, for a 2.74 Ϯ 0.02 Å manganese-manganese vector. Note that two manganese-manganese vectors correspond to N iso ϭ 1.0, which is at the lower border of the error bar; taking into account that the EXAFS technique tends to underestimate N values, the N iso ϭ 1.3 Ϯ 0.3 obtained favors three manganese-manganese interactions (expected N iso ϭ 1.5). The angle is in agreement with that reported earlier based on conventional polarized EXAFS data (34,35). The new range-extended data show that peak II contains interactions at 2.7 and 2.8 Å, which can be analyzed separately (Table 1, fits 3 and 4). Fig. 6 shows linear plots of N app derived from fits 3 and 4 in Table 1 (solid squares) against 3cos 2 Ϫ1 (see supplemental Equation S6). Third points (open squares) were obtained from extended EXAFS measurements of isotropic PS II S 1 in solution (see supplemental Table S1). Linear fits using only two data points from oriented samples (solid lines) or using three data points including the isotropic values (dashed lines) are nearly identical and result in the same N iso and ͗͘ values. Supplemental Fig. S7 shows more traditional polar plots of the N app derived from fits 1 to 4 in Table 1 and supplemental Table S1 and plotted with respect to the detection angle (). Analysis of the orientation dependence of the 2.72 Ϯ 0.02 Å manganese-manganese vector results in N iso ϭ 0.88 Ϯ 0.2 (two manganese-manganese interactions) at an average angle ͗͘ ϭ 61 Ϯ 5°, with respect to the membrane normal. The 2.83 Ϯ 0.02 Å manganese-manganese vector exhibits N iso ϭ 0.46 Ϯ 0.12 (one manganese-manganese interaction) at an angle ͗͘ ϭ 64 Ϯ 10°with respect to the membrane normal.
Curve Fitting of EXAFS FT Peak III-Curve fitting results for peak III are shown in Table 2. Analysis of peak III has been problematic in EXAFS studies of PS II because of the relatively weak intensity and correspondingly low signal-to-noise ratio. As mentioned above, peak III was found to exhibit dichroism and to consist of at least two different manganese-backscatterer vectors (34,35). Both of these conclusions are strengthened by the new range-extended EXAFS measurements. Combinations of manganese, calcium, carbon, and oxygen were tested to fit peak III (10,34,40). Distances of 3.1-3.4 Å are reported between manganese atoms bridged by a single 2 -or 3 -oxo unit. In addition, for carboxylate-or histidine-derived ligands, a highly disordered shell of carbon atoms at 2.9 -3.3 Å may be expected from the next-nearest neighbor atoms to the metal (46 -51). However, attempts to include manganese-light atom (oxygen or carbon) shells into peak III fits result in increased fit errors (data not shown). On the basis of differences in apparent distances (RЈ), dichroism for peak IIIA and IIIB, and previous studies of oriented strontium-reactivated PS II membranes (23), we conclude that the manganese-calcium vector contributes mainly at 15°and the manganese-manganese vector at 75°. One-shell fits presented in Table 2 (rows 1 and 2) agree well FIGURE 5. Fourier isolates of peak II and III from oriented PS II membranes. Fourier isolates of peak II (top) and peak III (bottom) of one-dimensionally oriented PS II samples in the S 1 state with the membrane normal at 15°and 75°with respect to the x-ray e-vector. The data are k 3 -weighted from 3.5 to 15.2 Å Ϫ1 (see Fig. 4A). For the back transform the individual Fourier peaks II and III were isolated by applying a Hamming window to the first and last 15% of the chosen range, leaving the middle 70% untouched. For typical isolates the range was ϳ0.7 Å for peak II and ϳ0.8 -0.9 Å for peak III.  MARCH 9, 2007 • VOLUME 282 • NUMBER 10

Structure and Orientation of the Mn 4 Ca Cluster in Photosystem II
with the above assignment and support a longer distance for the manganese-calcium vector compared with that for the manganese-manganese vector. Estimation of the minor contributions of the manganese-calcium vector at 75°orientation and the manganese-manganese vector at 15°orientation is required to determine the orientations of those vectors relative to the membrane normal, but isolating peaks of weak intensities results in unreliable two-shell fits. As described in an earlier study (23), we can either assume an N app Ϸ 0 for contributions close to the noise level or estimate the upper limit of the N app based on proportionality of the reduced amplitude of FT peak IIIA or IIIB to the peak maximum at the same RЈ in the complementary orientation. When the measured amplitude was close to the calculated noise level (between 4 and 10 Å in the FTs), as for the manganese-calcium interaction, the noise level in the FT spectrum was used to estimate the upper limit of N app . The upper limits of the N app for manganese-manganese and manganese-calcium interactions are listed in Table 2. Fig. 7 shows N app for the manganese-manganese component of Peak III derived from Table 2 plotted against 3cos 2 Ϫ1, as was done for Peak II in Fig. 6. A third point (open squares) was obtained from extended EXAFS measurements of isotropic PS II S 1 in solution (supplemental Table  S2). A linear fit using only two data points from oriented samples (solid lines) and a fit using three data points including the isotropic values (dashed lines) are similar and, within experimental error, result in similar N iso and ͗͘ values. The orientation dependence of N app for the manganese-manganese 3.2 Ϯ 0.02 Å vector (see Table 2) results in N iso ϭ 0.39 Ϯ 0.1 and ͗͘ ϭ 70°. These values are consistent with a single manganese-manganese 3.2 Å vector, for which the expected value of N iso is 0.5.
For the manganese-calcium 3.4 Å distance, the expected value for N iso is 0.25 for one manganese-calcium vector or 0.50 for two vectors. If we assume that N app ϭ 0 at 75°, the angle dependence of N app for this component results in N iso ϭ 0.29 Ϯ 0.09 for ͗͘ ϭ 0°. This value favors one manganese-calcium interaction at 3.40 Ϯ 0.02 Å; however, taking into account the error range and previous data of Cinco et al. (22) the possibility of two interactions at this distance cannot be excluded (22). The upper limit of this angle was estimated to be 18°in this work and 23°by Cinco et al. (23) for the strontium-manganese interaction. Despite relatively high uncertainty, we can conclude that this vector is aligned near to the membrane normal. (12,13,33,52,53) and EPR studies (6,7,54,55), it is known that a major structural motif of the OEC is the di--oxo-bridged Mn 2 unit. Conventional EXAFS studies could not settle the question of whether there are two or three di--oxo-bridged manganese-manganese moieties in the native S 1 and S 2 states (40,53,56). The uncertainty in the determination of the numbers of absorberbackscatterer vectors by EXAFS has prevented a clear solution to this problem. However, conventional EXAFS studies more clearly show that there is heterogeneity in the manganese-manganese distances in the range of 2.7-2.85 Å for the S 0 state (40), for samples poised in the S 2 g ϭ 4.1 state (57), as well as for the chemically modified F Ϫ -(58) and the NH 3 -treated (35) S 2 state samples. For those states with two manganese-manganese distances, the number of di--oxo-bridged manganese-manganese interactions can be determined with a higher accuracy, considering that only an integral number of each interaction is allowed. Re-evaluation of earlier results by Robblee et al. (40) demonstrated that they are more consistent with a 2:1 ratio, with the shorter distance predominating, supporting OEC models containing three 2.7-2.85 Å manganese-manganese FIGURE 6. Linear plot of polarized manganese EXAFS data from Fourier peak II. Linear plots of the x-ray absorption dichroism of Peak II for oriented PS II samples in the S 1 state. The N app values (solid squares) are derived from two-shell curve fits of FT peak II (Fig. 4A and Table 1) and are plotted against 3cos 2 Ϫ1 (see supplemental Equation S6). Best fits are shown for the different manganese-manganese vectors as solid lines. Additional third points (open squares) were obtained from extended EXAFS measurements of isotropic PS II S 1 in solution (see supplemental Table S1). Fits with three data points including the isotropic values (dashed lines) are very close to those obtained using only oriented sample data. interactions. For the native S 1 and S 2 states, with less distance heterogeneity, the results of the conventional EXAFS were still not conclusive (40). The recent range-extended EXAFS study of PS II in solution provides evidence supporting the presence of three di--oxobridged manganese-manganese vectors in the native S 1 and S 2 states (27). The conclusion is based on the fits of the Fourier peak II and II ϩ III isolates, which demonstrated that: (i) two distinct manganese-manganese vectors contribute to peak II (2.7 and 2.8 Å); (ii) there is an unequal distribution of the coordination numbers of N 1 (2.7 Å) and N 2 (2.8 Å), which would be consistent with the presence of three di--oxo-bridged manganese-manganese moieties; (iii) the fit clearly improved when the N 1 /N 2 ratio is close to 2:1 with N tot Ϸ 1.5. The difference between the two di--oxo-bridged manganese-manganese distances is approximately 0.1 Å for both the S 1 and the S 2 state, which explains why the traditional EXAFS study with a distance resolution of 0.14 Å was unable to reveal such distance heterogeneity.

FT Peak II; Three Short Manganese-Manganese Interactions in the OEC-From a number of EXAFS
The current polarized EXAFS data from oriented PS II membranes in the S 1 state support the conclusions summarized above for the S 1 state in solution. The fit qualities for peak II shown in Table 1 demonstrate significant improvement if two manganese-manganese vectors are introduced. The x-ray absorption linear dichroism from oriented PS II membranes demonstrates that the average manganese-manganese (ϳ2.7 Å) vector and manganese-manganese (ϳ2.8 Å) vector both have similar orientation of ϳ60°to the membrane normal, as summarized in Table 3. The averaged (ϳ2.7-2.8 Å) vector is oriented at ϳ63°, which is the same as the value reported earlier from conventional EXAFS studies with oriented PS II membranes (34).
Several possible reasons for manganese-manganese distance heterogeneity can be suggested: 1) differences in the redox states of manganese atoms resulting in the 2.7 Å and 2.8 Å manganese-manganese vectors; 2) differences in types of -oxo bridges ( 2 -oxo versus 3 -oxo) connecting manganese atoms; 3) protonation of the -oxo bridge. Protonation of the -oxo bridge lowers the manganese-oxygen bond order, which causes an increase in the manganese-manganese distance. Studies of model compounds support these possible reasons for distance heterogeneity (59 -61).

FT Peak III; Orientation of the Long Manganese-Manganese and Manganese-Calcium Vectors Relative to the Membrane
Normal-Fits to EXAFS data allow consideration of some relevant questions about the chemical nature of backscatterers contributing to peak III and, now, about the orientation of those manganese-backscatterer vectors relative to the membrane normal. Calcium was included in the fit combination because it has been implicated as a structural element of the OEC through O 2 evolution activity (19,45,(62)(63)(64)(65)(66), EPR (20), and calcium-and strontium-EXAFS experiments (22,23). In conventional EXAFS experiments it was noticed previously that the addition of the manganese-calcium vector to the manganese-manganese long interaction improves fit qualities. The evidence that peak III cannot be a result of only manganese-calcium interactions came from a study in which the manganese EXAFS spectrum of calcium-depleted PS II showed a peak III with decreased intensity (43). High resolution range-extended EXAFS data on oriented PS II membranes provide new experimental support for the conclusion that peak III contains both manganese-manganese and manganese-calcium interactions; the combination of oriented preparations and range-extended EXAFS allows the FIGURE 7. Linear plot of polarized manganese EXAFS data from Fourier peak III. Linear plots of the x-ray absorption dichroism of the manganesemanganese (3.20 Ϯ 0.02 Å) (black) and manganese-calcium (3.40 Ϯ 0.02 Å) (gray) vectors for oriented PS II samples in the S 1 state. The N app values were derived from one-shell fits (solid squares) of FT peak III ( Fig. 4A and Table 2) and are plotted with respect to 3cos 2 Ϫ1 (estimation of N app for manganesemanganese vector at ϭ 15°and for manganese-calcium vector at ϭ 75°is described in the text). The best fits of N app versus to supplemental Equation S6 are shown for the manganese-manganese and manganese-calcium vectors (solid line) taking into account the experimentally determined mosaic spread of ⍀ ϭ 20°. Additional third points (open squares) were obtained from range-extended EXAFS measurements of PS II S 1 solution (see supplemental  Table S2). Fits with three data points including the isotropic values (dashed lines) are close to that obtained using only oriented sample data. The results from Figs. 6 and 7 and the polar plots in the supporting information are summarized in Table 3.  MARCH 9, 2007 • VOLUME 282 • NUMBER 10 two types of interactions to be resolved. Information about the relative orientation of the manganese-strontium vectors was reported in the earlier study of strontium-reconstituted, oriented PS II membranes (23). Results of analysis of the x-ray absorption linear dichroism from peak III of the oriented PS II samples are summarized in Table 3. The orientation of the averaged manganese-calcium vector is similar to that reported for strontium-reconstituted, oriented PS II membranes (23). The manganese-manganese (ϳ3.2 Å) vector is oriented more toward the membrane plane. The average of both vectors is in agreement with the results from conventional EXAFS: ϳ43°for the ϳ3.3 Å vectors (combination of mono--oxo-bridged manganese-manganese and manganese-calcium vectors) (34).

Structure and Orientation of the Mn 4 Ca Cluster in PS II-
The structural information about the Mn 4 Ca cluster in the S 1 state, based on polarized EXAFS data, is summarized in Table 3. Topological models for the Mn 4 cluster compatible with the EXAFS data (27) and containing three short 2.7-2.8 Å manga-nese-manganese vectors and one manganese-manganese interaction at 3.2 Å are shown in Fig. 8A. Previously, dichroism characteristics of only the averages of the short manganesemanganese vectors at ϳ2.7 Å and the long manganesemanganese and manganese-calcium interactions at ϳ3.3 Å were known (34). Those data were not sufficient to restrict the possible orientations of the proposed models with respect to the PS II membrane or the PS II protein frame. With the new range-extended EXAFS data, we can for the first time determine independently the angle dependence for three different (2.8 Å, averaged 2.7 Å, and 3.2 Å) manganese-manganese vectors relative to the membrane normal. This allows us to impose additional restrictions on proposed structural models. The importance of the results in Table 3 is that they restrict the angles of the manganese-manganese or manganese-calcium vectors relative to the membrane normal for all three manganese-manganese vectors simultaneously.
Knowledge of the angles between the membrane normal and each of three different vectors involving manganesemanganese interactions allows us to determine whether one or more orientations for any particular model are consistent with the dichroism data. For this purpose we used the averaged 2.7, 2.8, and 3.2 Å manganese-manganese vectors, and we illustrate this approach for each of the models in Fig. 8A; the orientations shown in Fig. 8B are in agreement with the dichroism measurements. We emphasize that there may be other models that can be tested in this manner (this would include structural and optical isomers of listed models). There are two possibilities for the placement of the 2.8 Å manganese-manganese vector for Model I and three possibilities for Model II. As structures Ia and Ib and IIa, IIb, and IIc in Fig. 8B demonstrate, different placement of the 2.8 Å manganese-manganese vectors results in rather small changes in the orientation of models, as follows from the close values of the angle of 2.8 Å manganese-manganese vector and averaged angle of two 2.7 Å manganese-manganese vectors to the membrane normal (ϳ60°, Table 3). Uncertainty in the angle between the 3.2 Å manganese-manganese vector and membrane normal (Ͼ70, Table 3) results in a subset of model orientations as this angle changes within the determined range; however, this does not produce dramatic changes in the model orientations. Model III has a high degree of rotational freedom for the 3.2 Å manganese-oxygen-manganese mono--oxo unit that results in multiple solutions; in Fig. 8B we show only one such example. There are many possibilities for the placement of calcium in the models shown in Fig. 8B, and the average orientation of the manganese-calcium vector can best be described to be within a cone about the membrane normal, as shown in Fig.  8C. Using the results of polarized EXAFS (Table 3), the range of possible orientations of the models to the membrane normal can be dramatically reduced as illustrated in Fig. 8B. However, the following uncertainties remain: (i) rotational ambiguity in the membrane plane, which is always present for one-dimensionally oriented samples such as oriented membranes and (ii) multiple possibilities for calcium coordination; present data do not allow us to distinguish clearly whether there is one or two manganese-calcium interactions FIGURE 8. Cluster models compatible with polarized range-extended EXAFS. A, models for the Mn 4 Ca cluster compatible with the range-extended manganese EXAFS data with three short 2.7-2.8 Å manganese-manganese distances and one longer manganese-manganese distance at 3.2 Å. B, Mn 4 models developed from Fig. 8A topological core structures and their proposed orientation relative to the membrane normal consistent with polarized range-extended EXAFS ( Table 3). Note that in the membrane plane there is a rotational ambiguity which is always present for one-dimensionally oriented samples such as layered membranes. We emphasize that there may be other models that can be tested in this manner (this would include structural and optical isomers of listed models). C, the orientation of the average manganese-calcium vector in relation to the 3.2 Å manganese-manganese vector. The cones represent a range for the average manganese-calcium vector(s) along the membrane normal (ϳ18°), and the 3.2 Å manganese-manganese vector toward the membrane plane (ϳ20°), respectively.