Altered Structure of the Mn4Ca Cluster in the Oxygen-evolving Complex of Photosystem II by a Histidine Ligand Mutation*

The effect of replacing a histidine ligand on the properties of the oxygen-evolving complex (OEC) and the structure of the Mn4Ca cluster in Photosystem II (PSII) is studied by x-ray absorption spectroscopy using PSII core complexes from the Synechocystis sp. PCC 6803 D1 polypeptide mutant H332E. In the x-ray crystallographic structures of PSII, D1-His332 has been assigned as a direct ligand of a manganese ion, and the mutation of this histidine ligand to glutamate has been reported to prevent the advancement of the OEC beyond the S2Yz• intermediate state. The manganese K-edge (1s core electron to 4p) absorption spectrum of D1-H332E shifts to a lower energy compared with that of the native WT samples, suggesting that the electronic structure of the manganese cluster is affected by the presence of the additional negative charge on the OEC of the mutant. The extended x-ray absorption spectrum shows that the geometric structure of the cluster is altered substantially from that of the native WT state, resulting in an elongation of manganese-ligand and manganese-manganese interactions in the mutant. The strontium-H332E mutant, in which calcium is substituted by strontium, confirms that strontium (calcium) is a part of the altered cluster. The structural perturbations caused by the D1-H332E mutation are much larger than those produced by any biochemical treatment or mutation examined previously with x-ray absorption spectroscopy. The substantial structural changes provide an explanation not only for the altered properties of the D1-H332E mutant but also the importance of the histidine ligand for proper assembly of the Mn4Ca cluster.

The oxygen-evolving complex (OEC) 3 located in the Photosystem II (PSII) membrane-bound protein in plant, algae, and cyanobacteria catalyzes the water oxidation reaction (1)(2)(3)(4)(5)(6). The OEC couples the four-electron chemistry of water oxidation with the one-electron photochemistry of the reaction center by sequentially storing oxidizing equivalents through five intermediate S states (S i , where i ϭ 0 -4), before one molecule of dioxygen is evolved. The heart of the OEC consists of four manganese atoms and one calcium atom; the Mn 4 Ca cluster provides a high degree of redox and chemical flexibility so that several oxidizing equivalents can be stored during the S state cycle. Unlike inorganic catalysts, however, the uniqueness of catalytic centers in metalloenzymes arises from their protein environment. This environment provides (i) direct ligands to the cluster for maintaining the metal structure, while giving structural flexibility to the cluster at the same time, (ii) controls the redox potential of the cluster by participating in charge distribution during the catalytic reaction, and (iii) provides a hydrogen bonding network. Thus, nature has taken advantage of the special properties of the metal ions and tuned them by protein encapsulation to perform a wide variety of specific functions associated with life processes.
The assignment of the protein ligands to the Mn 4 Ca cluster has been attempted by a wide range of mutational studies. These studies have revealed the potential ligands to the Mn 4 Ca cluster. Considerable structural information has also been obtained through extensive studies by x-ray crystallography and various spectroscopic techniques over the past several years (1)(2)(3)(4)(5)7). The x-ray crystallography studies up to the current resolution of 2.9 Å (8 -11) have located the electron density associated with the water-oxidizing Mn 4 Ca cluster within the large complex of PS II peptides. Structural ambiguity remains, however, because of the limited resolution and radiation damage specifically to the Mn 4 Ca cluster (12,13). The ligand environment of the Mn 4 Ca cluster in the 3.5 and 3.0 (2.9) Å structural models also differ in a number of aspects, including the binding mode of the carboxylate ligands (i.e. bidentate versus monodentate) and the orientations of the backbone residues. These differences are likely due to the limited resolution of the diffraction data, different extents of radiation damage, and different interpretations of the electron density. Polarized x-ray spectroscopy with PSII single crystals also provides Mn 4 Ca models based on the accurate distances and the polarization characteristics of metal-metal or metal-ligand vectors within the protein environment (14). However, the assignment of the ligand environment remains tentative in this case because the proposed OEC environment obtained by this method is a simple combination of the Mn 4 Ca cluster model obtained from the polarized EXAFS and the ligand environment from the crystal structure (Fig. 1).
Despite the structural ambiguities mentioned above, all of the experimental evidence so far indicates that the D1-His 332 residue is an important direct ligand to the Mn 4 Ca cluster. In the recent x-ray crystallographic structural models of PSII, the Mn 4 Ca cluster is ligated by a single histidine residue, His 332 of the D1 polypeptide (8 -11). This assignment is supported by recent 31-34 GHz ESEEM studies of the Mn 4 Ca cluster in its S 2 state (15,16). These studies show that the nitrogen couplings observed in earlier ESEEM studies conducted at lower frequencies (9 -12 GHz) (17)(18)(19)(20) originate from a single 14 N nucleus. The higher frequency studies permit the quantitative determination of the hyperfine and nuclear quadrupolar couplings of the coupled nucleus; the higher frequency studies are consistent with direct coordination of the 14 N atom to a manganese ion in the S 2 state (15,16). On the basis of earlier 9-GHz ESEEM studies conducted with unlabeled and [ 15 N]histidine, this 14 N atom is known to originate from the nitrogen of a histidine imidazole group (19,21). Because the 14 N coupling essentially disappears in PSII core complexes from the D1-H332E mutant of Synechocystis sp. PCC 6803 (20), the 14 N nucleus has long been assigned to D1-His 332 , consistent with the x-ray structural models (5).
However, in a recent study (22) of a different mutant, D1-H332S, of Thermosynechococcus elongatus, the nitrogen coupling observed in the three-pulse 9-GHz ESEEM spectrum of wild type PSII core complexes was largely unchanged by the D1-H332S mutation. It was proposed that the nitrogen cou-pling observed in all the ESEEM studies originates from D1-His 337 rather than D1-His 332 and that the loss of the nitrogen coupling in the D1-H332E mutant (20) was caused by secondary structural perturbations introduced into PSII by the mutation. These same structural perturbations were proposed (22) to give rise to the altered S 2 state multiline EPR signal that is observed in the D1-H332E mutant (20,23) and the inability of D1-H332E cells (24) and PSII core complexes (23) to advance beyond the S 2 Y Z ⅐ state.
To study whether the failure of the catalytic cycle of D1-H332E is due to the structural changes of the OEC itself or to the secondary effect of the mutation, we applied x-ray spectroscopy in the current study. Manganese x-ray absorption spectroscopy is exquisitely sensitive to changes in manganesemanganese and manganese-calcium vectors in the Mn 4 Ca cluster of PSII and to changes in the charge density on the manganese ions. Moreover, XAS can detect the signals from manganese or calcium/strontium even when there are no other spectroscopic signatures from the cluster, because the x-ray spectroscopic signals are never silent. This property is useful in studying whether the cluster assembles in the mutant species. For example, it is possible that the cluster assembles in a modified geometry that makes it inactive and or incapable of exhibiting other spectroscopic signals. Determining the geometric and electronic structure of such altered clusters in mutants is particularly important because the changes may give us a rationale for identifying the properties of the native cluster that make it unique for catalyzing the difficult water oxidation reaction. XAS is also useful for studying the changes in the geometric and electronic structures as the altered cluster advances through the S state cycle and in determining the state at which the cycle is impaired. These structure-function correlations in the altered structures that may assemble in some mutants can be critical for understanding the native, active catalytic cluster. Accordingly, to characterize structural perturbations that may be caused by the D1-H332E mutation and to further investigate the structural role of D1-His 332 in the Mn 4 Ca cluster of PSII, we conducted a manganese x-ray absorption study of the D1-H332E mutant of Synechocystis sp. PCC 6803.

Construction of Mutant and Propagation of Cultures-
The construction of the D1-H332E mutation was described previously (20). Briefly, the mutation was introduced into the psbA-2 gene of Synechocystis sp. PCC 6803 and transformed into a host strain of Synechocystis that lacks all three psbA genes and that contains a hexahistidine tag fused to the C terminus of CP47. Solid media contained 5 mM glucose, 10 M 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU), and 5 g/ml kanamycin monosulfate. The DCMU and antibiotic were omitted from the liquid cultures. Large scale liquid cultures (each consisting of three 7-liter cultures held in glass carboys) were propagated as described previously (25). For the isolation of strontium-containing PSII core complexes, the CaCl 2 in the liquid media was replaced with an equivalent concentration of SrCl 2 (26,27). To verify the integrity of the mutant cultures that were harvested for the purification of PSII core complexes, an aliquot of each culture was set aside, and the sequence of the relevant portion of the psbA-2 gene was obtained after PCR amplification of genomic DNA (28).
Purification of PSII Core Complexes-PSII core complexes of the WT and D1-H332E mutant of Synechocystis sp. PCC 6803 retaining the manganese cluster were purified under dim green light at 4°C with nickel-nitrilotriacetic acid Superflow affinity resin (Qiagen) as described previously (27). The purification buffer consisted of 1.2 M betaine, 10% (v/v) glycerol, 50 mM MES-NaOH (pH 6.0), 20 mM CaCl 2 , 5 mM MgCl 2 , 50 mM histidine, 1 mM EDTA, and 0.03% (w/v) n-dodecyl ␤-D-maltoside. To ensure that strontium-containing PSII core complexes contained no excess strontium ions, strontium-containing PSII was purified in the presence of buffers containing CaCl 2 . The purified PSII core complexes were concentrated to ϳ1.0 mg of chlorophyll/ml by ultrafiltration and stored in liquid N 2 .
Preparation of Samples for X-ray Absorption Studies-To prepare samples for the x-ray absorption studies, the PSII core complexes were transferred to a buffer containing 1.2 M betaine, 40% (v/v) glycerol, 50 mM MES-NaOH (pH 6.0), 20 mM CaCl 2 , 5 mM MgCl 2 , and 0.03% (w/v) n-dodecyl ␤-D-maltoside by concentrating them to ϳ9 mg of chlorophyll/ml, diluting them 20-fold with a buffer containing 42% (v/v) glycerol, and then again concentrating them to ϳ9 mg of chlorophyll/ml. The resulting concentrated PSII core complexes were transferred to epoxy sample holders (40 l each) designed to fit into both EPR and x-ray liquid He cryostats. After dark adaptation for 1 h at room temperature, the samples were predominantly in the S 1 state. These samples were stored in liquid nitrogen until used. Half of the S 1 sample holders were taken out later and illuminated with a 400 W tungsten lamp at 200 K for 5 min to generate the S 2 state for the wild type S 2 state and at 273 K for 2 min for the H332E mutant sample. The illuminated samples were then stored in liquid nitrogen.
EPR Measurements-The samples were characterized by EPR spectroscopy at 8 K to ensure the quality of the samples before and after the illumination. The spectra were collected using a Varian E-109 EPR spectrometer equipped with an Air Products Helitran liquid helium cryostat. The generation of the S 2 state after the illumination was checked by the intensity of EPR multiline signal characteristic to the S 2 state. All of the samples were also checked with EPR prior to XAS measurement to see whether free Mn(II) was produced during any biochemical and sample handling processes. No detectable Mn(II) signal was observed for any of the mutant and wild type samples.
XAS Measurements-XAS was performed at the Stanford Synchrotron Radiation Laboratory on Beamline 9-3 at an electron energy of 3.0 GeV with an average current of 85-100 mA. The radiation was monochromatized by a Si(220) double-crystal monochromator. The intensity of the incident x-ray beam was monitored by a N 2 -filled ion chamber (I 0 ) in front of the sample. The monochromator energy was calibrated using a pre-edge peak of KMnO 4 (6543.3 eV) for manganese XAS and using an edge peak of strontium acetate (16,120 eV) for strontium XAS. These standards were placed between two N 2 -filled ionization chambers (I 1 and I 2 ) after the sample. The total number of x-ray photons used for the measurements was ϳ4 ϫ 10 13 photons mm Ϫ2 for the strontium XAS (16 -17 keV) and ϳ5 ϫ 10 12 photons mm Ϫ2 for the manganese XAS (6 -7 keV) and are well below the x-ray damage threshold as established previously (12). The manganese K-edge was also closely monitored for any reduction of manganese as seen by a shift in the K-edge inflection point energy. The monochromator was detuned at 6600 eV (16,200 eV for strontium) to 50% (or greater) of maximal flux to attenuate the x-ray second harmonic. The samples were kept at 10 K in a liquid helium flow cryostat to minimize radiation damage. Data reduction of the EXAFS spectra was performed using EXAFSPAK (29). Pre-edge and post-edge background were subtracted from the XAS spectra, and the results were normalized with respect to edge height. After conversion of background-corrected spectra from energy space to photoelectron wave vector (k) space (E 0 ϭ 6561.3 eV for manganese and 16,120 eV for strontium) and after k 3 -weighted, a four-domain spline function was subtracted. Curve fitting was performed using ab initio calculated phases and amplitudes from the program FEFF 8.2 (30,31). The details of the curve fitting analysis is described in the supplemental materials.

RESULTS
Manganese XANES of D1-H332E-The XAS spectra of Synechocystis sp. PCC 6803 WT and D1-H332E PSII core complexes in the dark state (S 1 ) and the illuminated (S 2 ) state are compared in Fig. 2. The WT S 1 and S 2 XANES spectra are similar to those observed in spinach PSII membrane preparations (32), in which the formal oxidation states have been assigned as Mn(III) 2 Mn(IV) 2 for the S 1 state and Mn(III)Mn(IV) 3 for the S 2 state. During the S 1 to S 2 transition, the rising edge energy shifts ϳ0.7 eV to higher energy in the WT (Fig. 2a). In its dark state (S 1 ), the D1-H332E mutant has a lower rising edge energy than the WT S 1 state and a slightly different shape as seen in the second derivatives (Fig. 2b). This is not due to the release of free Mn(II) in the mutant samples because no Mn(II) signal was observed in the EPR spectra taken prior to the XAS measurement. One possible explanation for this low energy shift in D1-H332E is that the formal oxidation state of the D1-H332E dark state is not Mn(III) 2 (IV) 2 , but lower such as (III) 3 (IV) or (II)(III)(IV) 2 . The energy shift between the WT and the D1-H332E is 1.69 eV. This is much larger than the energy shift between the S 1 to S 2 transition of the WT, but within the range of one oxidation state differences (for example, shifts of ϳ2.1 eV for the S 0 to S 1 transition and ϳ1.1 eV for the S 1 to S 2 transition have been observed in spinach PSII membrane preparations (32)). This variation in the magnitude of the energy shift corresponding to a single oxidation event demonstrates the difficulties of making formal oxidation state assignments on the basis of the XANES edge shift; the edge shape is also highly sensitive to structural changes.
In H332E PSII core complexes, only ϳ60% of the PSII centers contain manganese clusters, although all of these advance to the S 2 state in response to continuous illumination at 273 K (23). Upon illumination at 273 K, the H332E XANES spectrum shifts ϳ0.6 eV toward higher energy (Fig. 2c), indicating that manganese is oxidized during this transition in the mutant similar to the WT. The dark state of D1-H332E is EPR silent and, upon illumination, exhibits an altered multiline EPR spectrum thatsuperficiallyresemblesthatincalcium-depleted,strontium-substituted, and NH 3 -inhibited WT PSII preparations (20,23). It is therefore likely that the formal oxidation state of the D1-H332E mutant in its dark state is the same as that of the WT, i.e. Mn(III) 2 Mn(IV) 2 .
Manganese EXAFS of D1-H332E- Fig. 3 shows the Fourier transforms of k 3 -weighted manganese EXAFS spectra of Synechocystis sp. PCC 6803 WT and D1-H332E PSII core complexes in the dark (S 1 ) and illuminated (S 2 ) states (k-space data shown in supplemental Fig. S1). The WT spectra (Fig. 3a) are almost identical to those of the S 1 and S 2 spectra of spinach PSII membrane preparations, suggesting that the structure of the Mn 4 Ca cluster in the S 1 and S 2 states is the same in Synechocystis as in spinach. The D1-H332E mutant dark (S 1 ) state spectrum ( Fig. 3b) differs significantly from that of the WT S 1 state spectrum (Fig. 3a). The first FT peak (labeled I in Fig. 3), which arises primarily from manganese-oxygen interactions with a possible small contribution from one or two manganese-nitrogen interaction(s) in the WT, becomes more intense and shifted toward longer distance in the mutant. In addition, the second Fourier transform (FT) peak (labeled II in Fig. 3), which largely arises from the di--oxo bridged manganese-manganese interactions in the WT, decreases and merges with the third peak (labeled III in Fig. 3 and assigned predominantly to oxo-bridged manganese-manganese and manganese-calcium distances at ϳ3.3 and ϳ3.4 Å in the WT) in the mutant. These changes indicate that the D1-H332E mutation substantially alters the structure of the Mn 4 Ca cluster. The spectra obtained with three different preparations of D1-H332E PSII core complexes were identical. This reproducibility provides proof that the structure of the OEC cluster in the mutant is homogeneous and that it does not change as a function of the preparation. Additionally, we estimated that the manganese stoichiometry in the D1-H332E mutant is the same as the native PSII, i.e. four manganese ions/PSII, based on previous EPR studies (20,23) and the analysis of manganese x-ray fluorescence data (see supplemental materials).
Upon illumination, the first FT peak intensity decreases in the mutant (Fig. 3b, gray trace). Because the first FT peak of the WT is essentially unchanged upon illumination (Fig. 3a, gray trace), the substantially decreased amplitude of the first FT peak in the mutant indicates that substantial structural changes (e.g. ligand reorganization) take place during the S 1 to S 2 transition in the mutant, whereas little or no structural change takes place during this transition in WT (33).
Strontium XAS on strontium-substituted H332E-It has been suggested on the basis of biochemical studies that mutations of D1-His 332 diminish the affinity of calcium for the Mn 4 Ca cluster (34). Because the x-ray crystallographic structural models place the calcium ion and His 332 ligand on opposite sides of the Mn 4 Ca cluster, the apparently diminished affinity of calcium has been hypothesized to be a consequence of mutation-induced perturbations to the Mn 4 Ca cluster. To investigate whether the calcium ion is present in the manganese cluster of D1-H332E PSII core complexes, samples were prepared having strontium substituted for calcium, and measurements were carried out using strontium XAS (instead of calcium XAS). Many studies have shown that the calcium ion can be replaced with strontium while retaining O 2 evolving activity in PSII (26,27,35). Additionally, strontium XAS (absorption energy, ϳ16,120 eV) has several experimental advantages over calcium XAS (absorption energy, ϳ 4500 eV), such as (i) the x-ray radiation damage is less for the same number of photons incident on the sample, (ii) the attenuation of the incident and scattered x-rays is far less at the strontium x-ray absorption energy, and (iii) the fluorescence yield of strontium is higher than that of calcium (see Refs. 36 and 37).   (Fig. 4b, gray trace) shows a distinctive second FT peak (arising primarily from manganese-strontium interactions) that closely resembles the second FT peak in the WT. These data show that the manganese-strontium interaction exists in the H332E mutant and that the manganese-strontium interaction(s) are at a similar distance and amplitude to the WT. On the other hand, the first FT peak in the mutant (corresponding to strontiumoxygen interactions) is much stronger than in the WT. We also collected manganese XAS on this sample, to see whether the substitution of strontium for calcium caused any structural perturbations to the Mn 4 cluster (supplemental Fig. S2). Unlike the situation in S. elongatus, where biosynthetically incorporated strontium does not exchange with exogenous calcium ions (26), strontium/calcium exchange occurs when Synechocystis PSII core complexes containing biosynthetically incorporated strontium are purified in buffers containing calcium (27). Because strontium XAS requires the elimination of exogenous strontium ions, our PSII core complexes containing biosynthetically incorporated strontium were purified in calciumcontaining buffers. Therefore, only ϳ75% of the strontium-H332E PSII core complexes contained strontium, whereas the remainder contained calcium (27). This heterogeneity is not an issue for carrying out strontium XAS measurements, because only the strontium-H332E fraction is probed. For manganese XAS on this sample, on the other hand, both strontium-H332E and calcium-H332E fractions contribute to the spectrum, and we cannot distinguish their individual contributions (supplemental Fig. S2). In the manganese EXAFS shown in supplemental Fig. S2, the spectra of the calcium-H332E and strontium (calcium)-H332E PSII samples are almost identical up to the FT peak II region, but the FT peak III region differs significantly. This difference might be explained as a consequence of a larger contribution of the strontium H332E fraction, which results in a distance elongation (typically, actual distances of 3.4 Å for  manganese-calcium and 3.5 Å for manganese-strontium) and increase in the intensity (caused by heavier atom contribution) around the peak III region. Although further detailed structural information cannot be extracted from the manganese XAS on the strontium-H332E sample, the geometry of the Mn 4 cluster in the strontium-containing H332E sample seems to be very similar to that in the calcium-containing H332E sample.
Manganese EXAFS Curve Fitting of the WT S 1 State Spectrum-Manganese EXAFS curve fits were carried out for the k 3 -weighted EXAFS spectra of the WT S 1 state and the D1-H332E mutant dark state (S 1 ). Fitting results for the WT S 1 state for the manganese-nearest neighbor interactions are given in Table 1, in which R, n, and 2 show actual distance (Å) (note that the Fourier transforms only show apparent distances), coordination number, and EXAFS Debye-Waller factor (Å 2 ), respectively. The n values are defined as the total number of absorber-backscatterer vectors divided by the number of absorber atoms per OEC. The R factor (R f , %) shows the goodness of the fit.
The predominant contribution to the FT peak I is from manganese-oxygens at r ϭ ϳ1.87 Å. Although one or two manganese-nitrogen interactions from the histidine residue(s) contribute to this region, it cannot be distinguished in the data because of the minor contribution (one of ϳ24 manganeseligand interactions at similar distances). Therefore, manganese-oxygen and manganese-nitrogen interactions were treated together in the current curve fitting.
The second and third peak regions (Fig. 3a) are due mainly to the di--oxo bridged manganese-manganese (ϳ 2.7 Å) and mono--oxo-bridged manganese-manganese (ϳ3.3 Å) interactions. The presence of manganese-calcium interactions (3.4 Å) in addition to the predominant mono--oxo bridged manganese-manganese interaction in the third peak has previously been demonstrated on the basis of calcium XAS experiments conducted with spinach PSII preparations (strontium XAS experiments on strontium-substituted PSII show four manganese-strontium interactions between ϳ3.5 and 4.0 Å; see "Strongtium EXAFS Curve Fitting of the D1-H332E Spectra") (37). Therefore, manganese-calcium interactions are included in the fit. In Table 1, the most preferable structural motif that supports the vector components of the recent EXAFS studies (the ratio of di--oxo bridged manganese-manganese:mono--oxo bridged manganese-manganese:manganese-oxygen-cal-cium interactions ϭ 3:1:x, where x ϭ 2 or 3; supplemental Fig.  S3, A-D) was used for fitting the WT S 1 state spectrum. The contribution of three manganese-manganese interactions and their distance heterogeneities (2.7-2.8 Å) becomes clear only when the polarized spectra or the range-extended EXAFS data are used for the fit (38,39). Therefore, this heterogeneity was not considered in this study. Including manganese-carbon interactions, which arise from the second shell carbons at ϳ3.2 Å from manganese, slightly improves the fitting quality. Although the coordination number n ϭ 2.75 for the manganese-carbon interactions was used through the fits, based on the ligand model reported in the 2.9 Å crystal structure (11), this number remains highly ambiguous because of the uncertainly in the ligand binding modes (for example, whether the binding modes of the carboxylate ligands are bidentate or monodentate). Nevertheless, the Debye-Waller factor ( 2 ) of this absorber-backscatterer path is large (ϳ0.018 Å 2 ), indicating that the distance distribution of the manganese-carbon interactions is high, and its contribution to the EXAFS spectrum has only a minor effect. In addition, the inclusion of this path does not noticeably change the fitting parameters for the other paths.
We have also considered two shell contributions to the first FT peak region: one part from the manganese-bridging oxygen and the other from the manganese-terminal oxygen. We assumed that the manganese-bridging oxygens have shorter distances (ϳ1.8 Å) than those of the manganese-terminal oxygens (1.9 Å) on the basis of model compound studies (40,41). No obvious improvement was observed in the two-shell fit (not shown). Fits 2 and 4 ( Table 1) are shown in Fig. 5a.
Manganese EXAFS Curve Fitting of the D1-H332E Spectra- Table 2 summarizes the curve fitting results of the D1-H332E mutant in its dark-stable state (S 1 ). In Fig. 3b, the increased intensity and the longer distance shift of the first FT peak of the D1-H332E sample compared with that of the WT indicate that the manganese-ligand distances are more uniform in the mutant. The one-shell fit of the first peak gives a manganeseligand distance of 1.89 Å in the mutant, whereas it is ϳ1.87 Å in the WT, suggesting that an elongation of the manganese-ligand interactions occurs in the mutant (see also supplemental Table  S1).
The averaged manganese-manganese distance (peak II) is also longer (ϳ2.77 Å) in the mutant compared with the WT a The manganese-oxygen (ϳ1.8 Å) coordination number was fixed to n ϭ 6.0 for simplicity, assuming hexa-coordination for all four manganese atoms. Note that if one manganese atom out of four is penta-coordinate as suggested by the literature (53), the n value becomes 5.75. However, this does not cause a noticeable effect on the fit quality. b n values are defined as the total number of absorber-backscatterer vectors divided by the number of absorber atoms per OEC. For example, n ϭ 1.5 implies that there are three manganese-manganese ϳ2.7 Å interactions. c The manganese-carbon (ϳ3.2 Å) coordination number was fixed to n ϭ 2.75, based on the ligand model of the 2.9 Å crystal structure (11). Each manganese atom sees approximately two to four carbons of carboxylate groups and an imidazole ring, although these numbers remain somewhat ambiguous at the 2.9 Å structure because of its moderate resolution. However, the contribution of this path to the fitting quality is minor due to the high Debye-Waller factor.
(ϳ2.72 Å). On the other hand, the intensity of the peak II region is decreased significantly in the mutant. This could be either due to the increased distance heterogeneity in the ϳ 2.7 Å region or due to the reduced number of di--oxo bridged man-ganese-manganese interactions in the mutant. To test these two hypotheses, n of the manganese-manganese ϳ 2.7 Å interaction was fixed either to n ϭ 1 (two manganese-manganese interactions), or to n ϭ 1.5 (three manganese-manganese interactions) in the fit (3:1:x and 2:1:x, or 3:2:x and 2:2:x). The result between two or three manganese-manganese 2.7 Å interactions was not conclusive because the fitting quality is similar within the error of the method (for example, Fit 1 compared with Fit 7 in Table 2). The result, however, shows that the averaged di-oxo-bridged manganese-manganese distance is ϳ0.05 Å elongated in the mutant. If there are three di--oxo bridged manganese-manganese interactions, the distance heterogeneity is much higher than that of the WT based on the decreased intensity of peak II. The presence of the manganese-calcium (strontium) interactions was confirmed by the strontium XAS data obtained with the strontium-substituted H332E PSII core complexes, described below. Therefore, the manganese-calcium interaction was included in all the fits. The mono--oxo bridged manganese-manganese interaction (ϳ 3.2 Å) seems to be necessary for maintaining a reasonable fit quality, and elimination of this interaction made fit quality worse (not shown). This implies that the unclear peak III region in the mutant spectrum (Fig. 3b) is a consequence of the elongation of peak II components.
Fits corresponding to an increase in the manganese-manganese ϳ3.3 Å interactions to two as compared with one in the WT, namely 3:2:x and 3:1:x, or 2:2:x and 2:1:x, were also considered without much improvement in the quality of the fit. The addition of the second shell manganese-carbon (ligand) ϳ3.1 Å interactions slightly improved the R factor. However, the contribution of this absorber-backscatterer path has only a minor impact on the modeling of the Mn 4 Ca cluster because of its high Debye-Waller factor, and it is not included in Table 2. The representative fits (Fits 2 and 8) are shown in Fig. 5b.
Strontium EXAFS Curve Fitting of the D1-H332E Spectra-Fitting results for the strontium XAS on strontium-H332E are given in Table 3. Previously, calcium EXAFS of native PS II and strontium EXAFS of strontium-reactivated spinach PSII membranes showed the proximity of calcium to manganese at 3.4 Å, and strontium to manganese at 3.5 Å in the dark-stable S 1 state (36,42,43). A more recent study on strontium-substituted PSII core complexes from S. elongatus shows strontium to be prox-  Table 1 for the WT and Fits 2 and 8 for the D1-H332E in Table 2 are shown in the figure. imate to all four of the manganese in the Mn 4 cluster in all S states (37). The study shows that there are two to three strontium-manganese interactions in which strontium and manganese are likely to be bridged by -oxo groups ϳ3.5 Å, whereas there are two or one longer interactions of ϳ3.9 Å. The fitting results of the published strontium-substituted PSII core complexes from S. elongatus is shown in supplemental Table S2. FT peak I in Fig. 4b is best simulated by one shell of seven to eight oxygen atoms at r ϭ 2.56 Å. For convenience, the n number of the strontium-oxygen 2.5 Å interaction was fixed to 8 in Table 3. The Debye-Waller factor of this peak is much smaller (0.005) in the mutant than in the WT (supplemental Table S2). Assuming that the coordination number of strontium is the same in the WT and mutant samples, the result suggests that the strontium-oxygen distance is more uniform in the mutant compared with the WT. On the other hand, FT peak II, which is best fit with two to three shorter strontium-manganese interactions and one to two longer strontium-manganese interactions (37) in the strontium-WT PSII core complexes, appeared with the same intensity in the two samples, suggesting that the manganese-strontium binding modes are similar between the WT and the mutant. Fits 1 and 2 are shown in Fig. 6. The R f value of Fit 1 is slightly better than others and preferable, suggesting that there are three short (ϳ3.5 Å) and one long (ϳ 3.9 Å) strontium-manganese interactions similar to that seen in the WT.

DISCUSSION
Structural Changes of the OEC in the H332E Mutant-The D1-H332E mutant of Synechocystis sp. PCC 6803 has previ-ously been characterized with chlorophyll fluorescence (24,34) and thermoluminescence (24) measurements in intact cells, and with optical absorption (23), EPR (20,23), and ESEEM (20) measurements in PSII core complexes. These data show that, in the mutant, the temperature threshold for forming the S 2 state is ϳ100 K higher than in WT, that the quantum yield for forming the S 2 state is very low, corresponding to a dramatic slowing of electron transfer from the manganese cluster to Y Z ⅐ , that the S 2 /S 1 midpoint potential is substantially decreased, that manganese oxidation beyond the S 2 Y Z ⅐ state is blocked, and that the S 2 state multiline EPR signal is altered, exhibiting more hyperfine lines and narrower splittings than in WT. In addition, D1-H332E PSII core complexes exhibit no S 1 state multiline EPR signal (23). These data were interpreted as showing that although the manganese cluster is assembled in the altered ligand environment, the mutation perturbs the structure of the Mn 4 Ca cluster and the network of hydrogen bonds that facilitates the S 2 to S 3 transition (20,23,24). Compared with the energy scale of EPR or IR spectroscopy, the energy scale of XAS is much larger (1 eV shift in the edge corresponds to 8066 cm Ϫ1 ). Thus, changes in XANES would indicate much larger changes in the electronic/geometric structure that would also be reflected in the substantial changes in the EXAFS. In the current study, the XAS data show that the D1-H332E mutation substantially alters the structure of the Mn 4 Ca cluster in both the S 1 and S 2 states, resulting in an elongation of manganeseligand and di--oxo bridged manganese-manganese interactions in the mutant. These structural alterations are far greater than those caused by any mutation 4 or biochemical treatment examined previously by x-ray absorption spectroscopy, including calcium depletion (44), strontium/calcium exchange (45), and NH 3 inhibition (46).
The lower rising edge energy, and the changes in the shape in the XANES spectrum of the D1-H332E PSII core complexes in its dark-stable (S 1 ) state could originate from a change in manganese oxidation state or from a change in charge density on the manganese cluster caused by a change in cluster structure, ligand symmetry, or ligand exchange. Because dark-adapted D1-H332E PSII core complexes (EPR silent) produce a multiline EPR signal upon illumination, which is evidence for a paramagnetically coupled four-manganese cluster, we conclude that the formal oxidation state of dark-adapted D1-H332E PSII centers is the same as that of the WT, i.e. Mn(III) 2 Mn(IV) 2 . Therefore, the XANES shift to lower energy compared with WT probably results from a combination of (i) 4 J. Yano, R. J. Service, V. K. Yachandra, and R. J. Debus, unpublished data.  Table 3 are shown in the figure.

Altered Mn 4 Ca Cluster in a PSII Mutant
increased negative charge density on manganese caused by, for example, replacing a neutral histidyl nitrogen with a negative carboxylate oxygen and (ii) accompanying structural changes of the Mn 4 Ca cluster. The presence of a calcium (strontium) ion in the manganese cluster of D1-H332E is confirmed by the similar intensity of the manganese-strontium interactions observed in the strontium EXAFS of the mutant. The manganese EXAFS curve fitting result suggests that the Mn 4 Ca cluster in D1-H332E contains two to three di--oxo bridged manganese-manganese units, one to two mono--oxo bridged manganese-manganese unit, and two to three ϳ 3.5 Å manganese-calcium interactions. In the one-shell fit, the averaged distance of di--oxo-bridged manganese-manganese interaction in the mutant is changed from ϳ2.72 Å (WT) to ϳ2.77 Å (D1-H332E) (Tables 1 and 2). A possible reason for this elongation is a protonation of the oxo-bridge (Fig. 7a), presumably to restore the charge balance because of the replacement of a neutral histidine ligand by a negatively charged glutamate residue. Another is a ligand structural change, where the replacement of the histidine residue by a glutamate triggers a previously bridging bidendate carboxylate changing its coordination mode from bidentate to monodentate (Fig. 7b). In the case of protonation of an oxo-bridge, we expect an approximately 0.1 Å elongation of the manganese-manganese distance (47). In the case of a carboxylate changing its coordination mode, removal of a bidentate-type bridge could elongate the di--oxo bridged manganese-manganese distance also by 0.1 Å, such as is observed between the di--oxo/-carboxylato and the di--oxo species (48). The proposed scenarios for the distance elongation suggested above could also occur by replacing His 332 by another proximal carboxylate residue or water molecule.
Regarding the structural motif of the altered Mn 4 Ca cluster in the mutant, the following two possibilities are considered. If there are three di--oxo-bridged manganese-manganese interactions, as in models having (3:1:x) or (3:2:x) configurations (shown in supplemental Fig. S3), the distance heterogeneity would be greater in D1-H332E than in WT in addition to the manganese-manganese distance being elongated. In recent polarized XAS and range-extended XAS studies, the presence of three di--oxo-bridged manganese-manganese interactions having 2.7 Å:2.8 Å distances in a 2:1 ratio has been suggested (39,49). In the D1-H332E mutant, we would expect an elongation (to Ͼ2.8 Å) of the longer component and/or an increased fraction of the longer component: in either case, peak II will be shifted to a longer distance, and the intensity will be decreased. If there are two di--oxo-bridged manganese-manganese interactions as in the model having (2:1:x) or (2:2:x) configurations in supplemental Fig. S3, one or both of the distance(s) would be elongated in the mutant.
Accompanied by the elongation and weak intensity of the ϳ2.7 Å manganese-manganese interaction (peak II), there is a strong manganese-ligand FT peak I in the D1-H332E dark state. This suggests a more symmetric manganese-oxygen ligand environment, which could happen if an oxo-bridge is broken, ligands are ligated as monodentate instead of bidentate, or water is ligated instead of carboxylate in the altered OEC (Fig.  7b).
In the strontium XAS experiments, a stronger strontiumoxygen FT peak was observed in D1-H332E than in the WT, just as a stronger manganese-ligand peak was observed in the manganese XAS of the mutant. This is likely due to a more uniform strontium-oxygen environment and elongation of the strontium-oxygen distances in the mutant. Although typical strontium-oxygen (carboxylate) and strontium-oxygen (water) distances are both within the range of 2.5-2.7 Å and therefore cannot be discriminated on the basis of the distance, a more symmetric strontium-oxygen environment may occur, for example, when a carboxylate residue ligated to strontium is replaced with a water molecule. One possibility is that the mutation-induced alteration of the Mn 4 cluster structure causes the rearrangement of the ligands around the strontium (calcium) and that a water molecule ligates to the Mn 4 cluster in place of a carboxylate group. On the other hand, the strontiummanganese interactions are similar in the mutant and WT. On the basis of these observations, we consider the following two cases for the manganese-strontium (calcium) interactions: (i) the manganese-strontium (calcium) binding modes are the same in the WT and the mutant, but the hydrogen bonding network is changed; for example, more water molecules ligate to strontium (calcium) instead of carboxylate oxygen atom(s); or (ii) the manganese-strontium (calcium) binding mode is altered, but the number of the manganese-strontium (calcium) interactions remains the same.
D1-His 332 Is a Ligand of the Mn 4 Ca Cluster-As noted above, the structural perturbations caused by the D1-H332E mutation are substantially larger than those produced by any biochemical treatment or mutation examined previously with XAS, includ-ing the extraction of calcium from the Mn 4 Ca cluster. Whereas calcium depletion, strontium/calcium exchange, and NH 3 inhibition all alter the S 2 state multiline EPR signal in the same manner as the D1-H332E mutation, manganese-EXAFS studies show that these treatments alter the structure of the Mn 4 Ca cluster either negligibly (in the case of strontium substitution) or to much lesser extents than the D1-H332E mutation (in the case of calcium extraction). Only in the case of NH 3 inhibition, where NH 3 is proposed to enter the coordination sphere of the Mn 4 Ca cluster as a amido bridge between two manganese ions (50), was elongation of a manganese-manganese distance observed (from 2.72 to 2.87 Å) (46). One significant difference of the mutation compared with the above treatments is that the Mn 4 Ca cluster must necessarily be assembled in a different ligand environment in the mutant. In the WT, the D1 and CP43 protein subunits of PSII provide ligands to the OEC: the current view of the assembly process is that the first two Mn 2ϩ ions are bound and photo-oxidized into Mn 3ϩ during their assembly into the apoprotein, and then the binding of the subsequent Mn 2ϩ ions occurs (51,52). Whether the ligands that are required for the assembly are the same as the proposed direct ligands to the Mn 4 Ca cluster remains a question. Nevertheless, the fact that the Mn 4 Ca cluster can assemble in the H332E mutant but in an altered structure suggests that a different ligand environment has modified the Mn 4 Ca cluster. On the basis of the earlier manganese-EXAFS studies of calcium-depleted, strontium/calcium-exchanged, and NH 3 -inhibited PSII preparations, it is difficult to imagine how the D1-H332E mutation could cause the substantial structural perturbations to the Mn 4 Ca cluster observed in our manganese-EXAFS data if D1-His 332 is not directly coordinated to a manganese ion. Therefore, we conclude that our data are in support of the x-ray crystallographic models depicting D1-His 332 as a ligand to a manganese ion. Upon mutation of D1-His 332 to glutamate, the missing imidazole ligand must be replaced by another ligand/residue, such as (i) a glutamate carboxylate, (ii) another carboxylate residue, (iii) a water molecule, or (iv) another histidine ligand. Another possibility is that no new residue is ligated to the manganese, and the coordination number therefore changes from six to five. Alternatively, a combination of those possibilities could occur if the reorganization of several ligands is triggered by the D1-H332E mutation. The manganese XANES spectrum of D1-H332E helps to distinguish between these possibilities: as discussed above, its shift to lower energy in the S 1 dark state, indicative of an increased negative charge density on the Mn 4 Ca cluster compared with WT, implies that the imidazole group of D1-His 332 is likely replaced by D1-Glu 332 or by another nearby carboxylate group.
Comparison of D1-H332E with Other Mutants-As described in the Introduction, there are two ESEEM studies in the literature on D1-H332 mutants from Synechocystis sp. PCC 6803 (D1-H332E, same as in this study) (19,21) and Thermosynechococcus elongatus (D1-H332S/Q) (22). Although both studies agree that D1-His 332 is a ligand of manganese, there are differences in the assignment of the spin-echo signals attributed to manganese-histidine ligand nitrogen atoms: one providing support for D1-His 332 and the other preferring D1-His 337 , which is not a direct ligand of manganese in the current crystal structures (9,10). It is suggested by Sugiura et al. (22) that (i) all of the published ESEEM studies have detected D1-His 337 instead of D1-His 332 and (ii) the structural perturbations caused by the D1-H332E mutation eliminate the nitrogen couplings of D1-His 337 (for a more detailed description see supplemental materials). Clarification of the source of the histidyl nitrogen coupling in PSII will require additional ESEEM studies of the S 2 state Mn 4 Ca cluster in D1-His 332 and D1-His 337 mutants. The current XAS study on the D1-H332E mutant from Synechocystis sp. PCC 6803 study cannot directly resolve all the questions posed by the ESEEM studies. However, the XAS study points to the conclusion that D1-His 332 directly coordinates to a manganese ion in the Mn 4 Ca cluster, and the altered EPR signal is a consequence of the altered OEC structure. The substantial changes seen in the structure of the manganese cluster in the mutant points to the consequences of the different ligand environment that influences the assembly and the importance of His 332 in Synechocystis sp. PCC 6803. The several differences of the D1-H332S and D1-H332Q mutants compared with D1-H332E: such as (i) D1-H332S and D1-H332Q evolve O 2 , whereas D1-H332E does not; (ii) D1-H332S and D1-H332Q exhibit a normal S 2 EPR mulitiline signal, whereas D1-H332E does not; and (iii) D1-H332S and D1-H332Q shows small changes in the S 3 state, whereas D1-H332E cannot advance beyond the S 2 Y Z ⅐ state, make the D1-H332S and D1-H332Q mutants attractive candidates for future XAS studies. Such XAS studies may also shed light on the differences between the mutants in their assembly/stability and coordination compensation/mutant rescue properties.
Concluding Remarks-On the basis of XANES and EXAFS data, we show that the D1-H332E mutation substantially alters the structure of the Mn 4 Ca cluster in both the S 1 and S 2 states, resulting in an elongation of manganese-ligand and manganese-manganese interactions in the mutant. These structural perturbations are larger than those produced by any biochemical treatment or mutation examined previously with x-ray absorption spectroscopy. Because it is difficult to imagine how the D1-H332E mutation could cause these substantial structural perturbations if D1-His 332 is not directly coordinated to a manganese ion, we conclude that D1-His 332 ligates a manganese ion, in support of the current crystallographic structural models. In the D1-H332E mutant, Mn 4 Ca cluster can be assembled but in an altered manner. This cluster advances one oxidation state (dark to illuminated state) by absorbing light, but the altered OEC structure prevents the complete catalytic cycle of the water splitting. This suggests that the first oxidation step (S 1 to S 2 ) takes place relatively easily, whereas the second step (S 2 to S 3 ) is critical and structurally more demanding.