Genetically introduced hydrogen bond interactions reveal an asymmetric charge distribution on the radical cation of the special-pair chlorophyll P680

The special-pair chlorophyll (Chl) P680 in photosystem II has an extremely high redox potential (Em) to enable water oxidation in photosynthesis. Significant positive-charge localization on one of the Chl constituents, PD1 or PD2, in P680+ has been proposed to contribute to this high Em. To identify the Chl molecule on which the charge is mainly localized, we genetically introduced a hydrogen bond to the 131-keto C=O group of PD1 and PD2 by changing the nearby D1-Val-157 and D2-Val-156 residues to His, respectively. Successful hydrogen bond formation at PD1 and PD2 in the obtained D1-V157H and D2-V156H mutants, respectively, was monitored by detecting 131-keto C=O vibrations in Fourier transfer infrared (FTIR) difference spectra upon oxidation of P680 and the symmetrically located redox-active tyrosines YZ and YD, and they were simulated by quantum-chemical calculations. Analysis of the P680+/P680 FTIR difference spectra of D1-V157H and D2-V156H showed that upon P680+ formation, the 131-keto C=O frequency upshifts by a much larger extent in PD1 (23 cm−1) than in PD2 (<9 cm−1). In addition, thermoluminescence measurements revealed that the D1-V157H mutation increased the Em of P680 to a larger extent than did the D2-V156H mutation. These results, together with the previous results for the mutants of the His ligands of PD1 and PD2, lead to a definite conclusion that a charge is mainly localized to PD1 in P680+.

The special-pair chlorophyll (Chl) P680 in photosystem II has an extremely high redox potential (E m ) to enable water oxidation in photosynthesis. Significant positive-charge localization on one of the Chl constituents, P D1 or P D2 , in P680 ؉ has been proposed to contribute to this high E m . To identify the Chl molecule on which the charge is mainly localized, we genetically introduced a hydrogen bond to the 13 1 -keto C‫؍‬O group of P D1 and P D2 by changing the nearby D1-Val-157 and D2-Val-156 residues to His, respectively. Successful hydrogen bond formation at P D1 and P D2 in the obtained D1-V157H and D2-V156H mutants, respectively, was monitored by detecting 13 1 -keto C‫؍‬O vibrations in Fourier transfer infrared (FTIR) difference spectra upon oxidation of P680 and the symmetrically located redox-active tyrosines Y Z and Y D , and they were simulated by quantum-chemical calculations. Analysis of the P680 ؉ /P680 FTIR difference spectra of D1-V157H and D2-V156H showed that upon P680 ؉ formation, the 13 1 -keto C‫؍‬O frequency upshifts by a much larger extent in P D1 (23 cm ؊1 ) than in P D2 (<9 cm ؊1 ). In addition, thermoluminescence measurements revealed that the D1-V157H mutation increased the E m of P680 to a larger extent than did the D2-V156H mutation. These results, together with the previous results for the mutants of the His ligands of P D1 and P D2 , lead to a definite conclusion that a charge is mainly localized to P D1 in P680 ؉ .
Photosynthetic water oxidation performed by plants and cyanobacteria produces virtually all oxygen in the atmosphere. This reaction takes place in photosystem II (PSII) 3 protein complexes, which have the function of abstracting electrons from water using light energy releasing protons and molecular oxy-gen (1)(2)(3)(4)(5). The obtained electrons and protons are used to produce NADPH and ATP, respectively, which are utilized to synthesize sugars from CO 2 . Photochemistry in PSII starts with light-induced charge separation from the excited singlet state of monomeric chlorophyll (Chl), Chl D1 , coupled with that of the special-pair Chl, P680, to form a radical pair between P680 and a pheophytin (Pheo) electron acceptor, P680 ϩ Pheo Ϫ (6, 7). The electron is transferred from Pheo to the primary quinone electron acceptor Q A and then the secondary quinone acceptor Q B , which becomes quinol upon two-electron reduction and is released into thylakoid membranes (8). On the electron donor side, the cation radical of P680 oxidizes the redox-active tyrosine Y Z and then the Mn 4 CaO 5 cluster, where two water molecules are oxidized upon four electron transfer reactions (1)(2)(3)(4)(5).
P680 has a dimeric structure of two Chl molecules, P D1 and P D2 , that are bound to the D1 and D2 subunits, respectively ( Fig.  1). To achieve water oxidation, which has a redox potential (E m ) of 880 mV at pH 6.0, P680 needs to be a very strong oxidant. Indeed, its E m has been estimated to be ϳ1200 mV (9 -12), which is much higher than the E m values of the special pairs of other reaction centers such as P700 of photosystem I (ϳ500 mV (13)) and P870 of bacterial reaction centers (ϳ500 mV (14)). The mechanism that P680 has a high E m value has been extensively argued. It was suggested that the high E m originates from electrostatic interactions with other cofactors and proteins (15)(16)(17), a low dielectric environment (18), and localization of a positive charge on one Chl in the dimer (19,20).
As for the charge localization on the dimer, electron spin resonance (21) and Fourier transform infrared (FTIR) spectroscopy (22) showed that a spin density or a positive charge on P680 is mostly localized on one Chl in the dimer. These measurements, however, could not specify the Chl molecule, P D1 or P D2 , on which a charge is mainly distributed. Diner et al. (23) attempted to answer this question by analyzing site-directed mutants of Synechocystis sp. PCC 6803, in which His ligands of P D1 and P D2 , D1-His-198 and D2-His-197, respectively (24 -27), were replaced with other amino acids. They examined the effects of mutations by detecting changes in the visible absorption bands of P680 in difference spectra. Blue shifts of a bleaching band of P680 at 433 nm in the Soret region by up to 3 nm were observed in D1-His-198 mutants, whereas red shifts by 0.5-1.5 nm were observed in D2-His-197 mutants. In the Q y region, the bleaching peak at 672.5 nm was blue-shifted by 3 nm upon D1-H198Q mutation. From these results, they proposed This work was supported by Grants-in-aid for Japan Society for the Promotion of Science Fellows 15J10320 (to S. N.) and Scientific Research from Japan Society for the Promotion of Science 26840091 (to R. N.) and 24000018 and 24107003 (to T. N.). The authors declare that they have no conflicts of interest with the contents of this article. 1 To whom correspondence may be addressed. that the cation is stabilized primarily on P D1 (23). A similar Soret shift by D1-H198Q mutation was recently confirmed using Thermosynechococcus elongatus (28). Theoretical calculations by Saito et al. (17) and Narzi et al. (29) based on the atomic coordinates of the X-ray crystallographic structure of PSII supported the above view and predicted the asymmetric distribution of a positive charge on P680 ϩ favoring P D1 due to the lower E m of P D1 than the E m of P D2 . Although the above conclusion of charge localization on P D1 seems consistent and reasonable, interpretation of the mutation-induced perturbation of the Soret and Q y bands needs caution. It is known that the excited states of Chls and Pheos in a PSII reaction center are coupled with each other, and the absorption bands of these chromophores are considerably congested in the narrow range of each absorption region. Theoretical works by Renger and co-workers (30 -32) showed that the Q y transition of P680 is formed by the excitonic coupling of P D1 and P D2 with similar contributions. Indeed, although the experimental absorption difference spectrum in the Q y region upon P680 ϩ formation was best simulated when the positive charge is localized on P D1 , the position of the bleaching peak of P680 was very similar even when a charge was assumed to be localized on P D2 (31). The strength of the excitonic coupling between P D1 and P D2 in the Soret transition has not yet been clarified, and hence the reason for the opposite shifts by mutations of the His ligands of P D1 and P D2 is not straightforward. Thus, the mutational effects on the changes in the Q y and Soret bands have not been fully explained to draw a definite conclusion for charge localization on P D1 in P680 ϩ .
Another way to perturb the P680 property by mutation is changing the hydrogen bond interactions at the 13 1 -keto CϭO groups, which are involved in the macrocycle conjugation of Chls. The change in the hydrogen bond of the 13 1 -keto CϭO can be monitored by FTIR difference or resonance Raman spectroscopy. Indeed, for bacterial reaction centers, various sitedirected mutants changing the hydrogen bonds of the 13 1 -keto CϭO groups of bacteriochlorophylls in special pairs have been analyzed using such a spectroscopic method as well as redox potential measurement (14,(33)(34)(35), showing that hydrogen bond formation at a 13 1 -keto CϭO group increases the redox potential of a special pair. The merits of vibrational detection of the 13 1 -keto CϭO interaction are as follows. (i) (B)Chl CϭO vibrations of individual (B)Chls in a dimer are not coupled with each other and hence can be examined independently. (ii) The 13 1 -keto CϭO vibration is sensitive to a hydrogen bond interaction, and its frequency spans in a wide range depending on the strength of interaction from ϳ1710 cm Ϫ1 for free interaction to ϳ1650 cm Ϫ1 for strong hydrogen bonding (36 -38). (iii) Upon cation formation, the 13 1 -keto CϭO frequency upshifts by ϳ30 cm Ϫ1 in monomeric (B)Chl (39 -41), and the extent of the shift reflects the charge distribution on each (B)Chl in a dimer (22,34). Thus, using vibrational spectroscopy, the genetically perturbed hydrogen bond of the 13 1 -keto CϭO of each (B)Chl molecule in a dimer can be independently monitored, and the charge distribution can be examined.
Light-induced FTIR difference spectroscopy has been extensively used to investigate the structural changes and photochemical reactions in PSII at a molecular level (42)(43)(44)(45)(46)(47)(48). As for P680, light-induced FTIR difference spectra upon photo-oxidation of P680 have been measured using various PSII preparations (22, 49 -51). It has been shown that a single negative band due to neutral P680 was detected at ϳ1700 cm Ϫ1 , representing a free interaction of the 13 1 -keto CϭO group. This is consistent with the X-ray crystal structure of PSII core complexes (24 -27), where amino acid residues neighboring the 13 1 -keto CϭO bonds of P D1 and P D2 are non-hydrogen-bonding D1-Val-157 and D2-Val-156, respectively ( Fig. 1). For P680 ϩ , two positive peaks were observed at ϳ1724 and ϳ1710 cm Ϫ1 in PSII membranes and core complexes (22,50,51). From the upshifts by ϳ24 and ϳ10 cm Ϫ1 from the neutral P680 band, it was proposed that more than 70% of a charge is distributed on one Chl in P680 ϩ (22). To identify the Chl molecule that mainly possesses a positive charge on P680 ϩ , the assignment of these two bands to either P D1 or P D2 is required. Such an assignment can be achieved by introduction of a hydrogen bond to the 13 1 -keto CϭO of P D1 and P D2 .
Hydrogen bond introduction to the 13 1 -keto CϭO groups of P680 is also useful to answer a long-standing question for the origins of prominent differential signals at ϳ1700 cm Ϫ1 in FTIR difference spectra upon oxidation of symmetrically located redox-active tyrosines, Y Z and Y D (52)(53)(54)(55)(56)(57)(58). Because of the close location of Y Z and Y D to P D1 and P D2 , respectively, and the peak frequencies similar to those of the 13 1 -keto CϭO bands of neutral P680, it has been proposed that these peaks in Y Z ⅐ /Y Z and Y D ⅐ /Y D difference spectra arise from the 13 1 -keto CϭO vibrations of P D1 and P D2 , respectively, which are affected by Y Z(D) ⅐ formation (52,53,(55)(56)(57)(58). However, the possibility of the assignments to peptide CϭO vibrations could not have been excluded (52)(53)(54)(55). Hydrogen bond formation at the 13 1 -keto CϭO group of P D1 and P D2 should provide a clear answer to the Red arrows indicate the 13 1 -keto CϭO groups of P D1 and P D2 . In the D1-V157H and D2-V156H mutants, a Val residue located near P D1 (D1-V157) and P D2 (D2-V156), respectively, was replaced with His to introduce a hydrogen bond to the 13 1 -keto CϭO group.
question for the assignments of these signals. If these assignments are confirmed, the differential signals around 1700 cm Ϫ1 in the Y Z ⅐ /Y Z and Y D ⅐ /Y D spectra can be useful markers for monitoring the 13 1 -keto CϭO interactions of P D1 and P D2 .
In this study, we mutated D1-Val-157 and D2-Val-156 to His to introduce a hydrogen bond at the 13 1 -keto CϭO group of P D1 and P D2 , respectively, using a cyanobacterium Synechocystis sp. PCC 6803 (Fig. 1). The obtained D1-V157H and D2-V156H mutants were analyzed by detecting light-induced FTIR spectra of P680, Y Z , and Y D upon their light-induced oxidation. The interactions of the 13 1 -keto CϭO groups in these mutants and the frequency changes were also simulated by quantum mechanics/molecular mechanics (QM/MM) calculations based on the high-resolution X-ray crystallographic structure (26). Furthermore, the effect of mutations on the redox potential of P680 was examined using thermoluminescence (TL) measurements. The obtained data showed successful introduction of hydrogen bonds to the 13 1 -keto CϭO groups of P D1 and P D2 and provided solid evidence for the main distribution of a positive charge on P D1 in P680 ϩ .

Thermoluminescence
TL measurements were performed using cells of the four strains in the presence of DCMU. The TL glow curves obtained after illumination at Ϫ20°C provided the so-called Q band (59), which originates from S 2 Q A Ϫ recombination (Fig. 2). The features of the glow curves of the D1-WT (Fig. 2a, black line) and D2-WT (b, black line) cells were virtually identical, showing a Q band at 11°C. In contrast, the D1-V157H mutation (Fig. 2a, red line) upshifted the Q band by 17°C to give a peak at 28°C with an increased intensity by a factor of 2.05, whereas D2-V156H mutation (b, blue line) induced only a small upshift by 2°C showing a peak at 14°C with a slight intensity increase by a factor of 1.22 (Table 1). 3A shows light-induced P680 ϩ /P680 FTIR difference spectra (1800 -1100 cm Ϫ1 ) of the PSII core complexes from the D1-V157H (trace a, red line) and D2-V156H (trace b, blue line) mutants in comparison with the spectra of D1-WT (trace a, black line) and D2-WT (trace b, black line), respectively. Positive and negative bands represent the cationic and neutral forms of P680, respectively. The spectra of D1-WT and D2-WT were virtually identical, indicating that there is no effect of deletion of the psbA1 and psbA3 genes in D1-WT and of the psbD2 gene in D2-WT on the P680 interaction. These WT spectra are very similar to the previously reported P680 ϩ /P680 FTIR difference spectra of cyanobacterial PSII core complexes and spinach PSII membranes (22,50,51). Prominent bands in the 1730 -1690-cm Ϫ1 region and minor bands in the 1760 -1730-cm Ϫ1 region have been assigned to the 13 1 -keto CϭO and 13 2 -ester CϭO stretching vibrations, respectively, of P680, although medium intensity bands at 1620 -1100 cm Ϫ1 have been attributed to its chlorin ring vibrations (22, 36 -41, 49 -51). In addition, prominent bands at 1690 -1620 cm Ϫ1 can be assigned to the amide I vibrations due to the CO stretches of backbone amides (60), which appeared by perturbations of polypeptide main chains upon cation formation on P680. The overall features of the P680 ϩ /P680 spectra of the D1-V157H (Fig. 3A, trace a, red line) and D2-V156H (Fig. 3A, trace b, blue line) mutants were very similar to those of the WT species (traces a and b, black lines). In particular, bands of the chlorin ring vibra-

Figure 2. TL glow curves of wild-type and mutant cells in the presence of DCMU due to S 2 Q A
؊ recombination. a, D1-WT (black) and D1-V157H (red). b, D2-WT (black) and D2-V156H (blue). tions (1620 -1100 cm Ϫ1 ) were virtually unaffected by mutations. In contrast, some specific changes were observed in the 13 1 -keto CϭO region (1730 -1690 cm Ϫ1 ). The expanded view of the 13 1 -keto CϭO region of the P680 ϩ /P680 spectra is shown in Fig. 3B. In the WT spectra (Fig.  3B, traces a and b, black lines), a prominent negative band at 1697 cm Ϫ1 and two positive bands at 1726 and 1708 cm Ϫ1 have been assigned to the 13 1 -keto CϭO vibrations of neutral and cationic forms of P680, respectively (22). The single band at 1697 cm Ϫ1 was interpreted as the overlap of the two CϭO bands from P D1 and P D2 , both of which are free from hydrogen bond interactions. In contrast, the split bands at 1726 and 1708 cm Ϫ1 of P680 ϩ were attributed to the asymmetric distribution of a positive charge on P D1 and P D2 (22). The D1-V157H spectrum (Fig. 3B, trace a, red line) also exhibited two positive bands at 1726 and 1708 cm Ϫ1 but changed their relative intensities. The 1726-cm Ϫ1 band was weakened, whereas that at 1708 cm Ϫ1 was strengthened. In addition, there was an intensity increase around 1720 cm Ϫ1 most likely due to the appearance of a new band. In neutral P680, the negative band at 1697 cm Ϫ1 slightly downshifted to 1695 cm Ϫ1 with a weakened intensity.

Table 1 Peak temperatures and relative intensities of TL bands (S 2 Q
The D2-V156H spectrum (Fig. 3B, trace b, blue line) showed a much clearer change. The negative band at 1697 cm Ϫ1 of WT disappeared, leaving a half-intensity band at 1703 cm Ϫ1 . Instead, a new band appeared at 1680 cm Ϫ1 superimposing an original band at 1682 cm Ϫ1 . This change is best interpreted as that one of the bands that contribute to the intensity at 1697 cm Ϫ1 downshifted by ϳ17 cm Ϫ1 to 1680 cm Ϫ1 , whereas another band was left at 1703 cm Ϫ1 . Along with this change, a positive shoulder at 1708 cm Ϫ1 seemed to downshift by ϳ19 cm Ϫ1 to 1689 cm Ϫ1 . In contrast, the main positive band at 1726 cm Ϫ1 was virtually unchanged upon the D2-V156H mutation.    4B, a, red line), concomitantly with some intensity decrease. In contrast, there was no change in these peaks in the D2-V156H spectrum (Fig. 4B, b, blue line). Fig. 5A shows light-induced Y D ⅐ /Y D FTIR difference spectra of PSII complexes from D1-WT (a, black line), D1-V157H (a, red line), D2-WT (b, black line), and D2-V156H (b, blue line) in the 1800 -1200-cm Ϫ1 region. Overall spectral features were very similar among the spectra, including the CO band of oxidized Y D ⅐ at 1503 cm Ϫ1 and the CO/COH band of reduced Y D at 1251 cm Ϫ1 (52,53,61). However, a considerable change was observed in the 13 1 -keto CϭO region of P680 in the D2-V156H spectrum (Fig. 5B, b). The 1702/1695-cm Ϫ1 peaks in D2-WT (Fig. 5B, b, black line) completely disappeared, and instead a large positive peak appeared at 1676 cm Ϫ1 . A corresponding negative peak may be a dip at 1684 cm Ϫ1 , whose intensity is relatively small probably because an original positive peak at 1688 cm Ϫ1 overlaps. This spectral change indicates that the 1702/1695-cm Ϫ1 peaks downshifted by 18 -19 cm Ϫ1 to 1684/ 1676 cm Ϫ1 upon the D2-V156H mutation. In sharp contrast, the corresponding peaks at 1703/1695 cm Ϫ1 in D1-WT ( Fig.  5B, a, black line) were unchanged upon the D1-V157H mutation ( Fig. 5B, a, red line).

QM/MM calculations of the 13 1 -keto C‫؍‬O frequencies of P680 models
To rationalize the observed changes in the 13 1 -keto CϭO frequencies of P680 by mutations, we performed QM/MM calculations of the P680 models of WT and the mutants. The QM region of WT consists of P D1 , P D2 , D1-His-198, D2-His-197, D1-Val-157, and D2-Val-156, whereas chromophores (Chl D1 , Chl D2 , Pheo D1 , and Pheo D2 ), amino acid residues, and water molecules within 10 Å from P D1 and P D2 were assigned to the MM region (Fig. 6). For the models of the D1-V157H and D2-V156H mutants, the side chains of D1-Val-157 and D2-Val-156, respectively, were replaced with a His side chain. The optimized geometries of the QM region of these P680 models are shown in Fig. 7, and the calculated 13 1 -keto CϭO frequencies are presented in Table 2 together with the values of the hydrogen bond distances and angles. In both of the mutant models,  the His side chain was hydrogen-bonded to the 13 1 -keto CϭO oxygen at the imidazole N -H (proximal NH). The structure that has a hydrogen bond at the N -H (distal NH) was not converged in the D1-V157H model, and it provided a less stable form than the N -H hydrogen-bonded form with a higher energy by 10.8 kcal/mol in the D2-V156H model. In the WT model, the 13 1 -keto CϭO frequency of P D2 was calculated to be lower by 9 cm Ϫ1 than that of P D1 (1692 cm Ϫ1 versus 1703 cm Ϫ1 ), representing an asymmetric P680 structure. In the D1-V157H model, the His NH formed a rather weak hydrogen bond with the 13 1 -keto CϭO of P D1 with a hydrogen bond distance (H⅐⅐⅐O) of 2.61 Å and an angle (N-H⅐⅐⅐O) of 96.4° (  Table 2). Indeed, only a small downshift by 4 cm Ϫ1 of the 13 1keto CϭO frequency of P D1 was estimated. In contrast, the D2-V156H model formed a relatively strong hydrogen bond between the His and the 13 1 -keto CϭO of P D2 with a distance (H⅐⅐⅐O) of 1.94 Å and an angle of 124.8°, resulting in a large downshift of the 13 1 -keto CϭO frequency by 17 cm Ϫ1 (Table  2). It is notable that in these mutant models, the 13 1 -keto CϭO frequency of the other side of Chl without mutation (i.e. P D2 in D1-V157H and P D1 in D2-V156H) was unaffected, indicative of the independence of the 13 1 -keto CϭO vibrations of P D1 and P D2 .

Discussion
In this study, we identified the Chl molecule (P D1 or P D2 ) in which a positive charge is mainly localized on P680 ϩ by analyzing the site-directed mutants, D1-V157H and D2-V156H, of Synechocystis sp. PCC 6803, which are expected to introduce a hydrogen bond interaction to the 13 1 -keto CϭO group of P D1 and P D2 , respectively (Fig. 1). Both mutants showed clear changes in TL glow curves (Fig. 2) and the 13 1 -keto CϭO stretching region of the P680 ϩ /P680 FTIR difference spectra (Fig. 3), indicating successful hydrogen bond formation to P680. In addition, the D1-V157H mutation induced a downshift of the 1707/1699-cm Ϫ1 differential signal in the Y Z ⅐ /Y Z FTIR difference spectrum (Fig. 4, trace a) by 4 cm Ϫ1 without any change in the Y D ⅐ /Y D spectrum (Fig. 5, trace a), whereas D2-V156H mutation induced a downshift of the similar differential signal at 1703/1695 cm Ϫ1 in the Y D ⅐ /Y D spectrum by 19 cm Ϫ1 (Fig. 5, trace b) without a change in the Y Z ⅐ /Y Z spectrum. These observations in the Y Z ⅐ /Y Z and Y D ⅐ /Y D spectra confirmed the hydrogen bond formation at P D1 in D1-V157H and at P D2 in D2-V156H. In addition, the observations provided definite assignments of the differential signals at ϳ1700 cm Ϫ1 in the Y Z ⅐ /Y Z and Y D ⅐ /Y D spectra to the 13 1 -keto CϭO vibrations of P D1 and P D2 , respectively, excluding the possibility of the assignments to the peptide CϭO vibrations (53)(54)(55).
It has been shown that upon oxidation of Y Z , its proton is transferred to the neighboring D1-His-190 that becomes a protonated cation (57,(62)(63)(64), whereas a proton from Y D is released to the bulk (58,65), and hence a positive charge is not accumulated around Y D . Thus, the similar band shifts of the 13 1 -keto CϭO of P D1 and P D2 by Y Z ⅐ and Y D ⅐ formation, respectively, may be caused by a common change in the protein environment around the CϭO group induced by tyrosine radical formation rather than an electrochromic effect. In any case, it is now clear that the differential signals at ϳ1700 cm Ϫ1 in the Y Z ⅐ /Y Z and Y D ⅐ /Y D spectra are useful markers to independently monitor the 13 1 -keto CϭO vibrations of P D1 and P D2 , respectively, in neutral P680 without interference of the bands of P680 ϩ . Because in both cases the frequency difference between the two peaks of the differential signal (8 cm Ϫ1 ) is similar to the width of a 13 1 -keto CϭO band (for example, the full width of half-maximum of the 13 1 -keto CϭO band at 1726 cm Ϫ1 is   Table 3). The lower P D2 frequency by 4 cm Ϫ1 than P D1 may reflect an asymmetric P680 structure or an asymmetric protein environment around the 13 1 -keto CϭO groups. Upon D1-V157H mutation, the differential signal around 1703 (1707/1699) cm Ϫ1 in the Y Z ⅐ /Y Z difference spectrum downshifted by ϳ3 cm Ϫ1 to that around 1700 (1703/1697) cm Ϫ1 (Fig. 4B, trace a; Table 3). In contrast, the D2-V156H mutation induced a large downshift of the differential signal by ϳ19 cm Ϫ1 from 1699 (1702/1695) to 1680 (1684/1676) cm Ϫ1 (Fig. 5B, trace b; Table 3). Thus, P D2 forms a relatively strong hydrogen bond with the introduced His in the D2-V156H mutant, whereas P D1 forms only a weak hydrogen bond with the His in D1-V157H. These frequency shifts of the 13 1 -keto CϭO vibrations are consistent with the mutation-induced changes in the 13 1 -keto CϭO bands of P680 in the P680 ϩ /P680 difference spectra (Fig. 3B). In the D2-V156H mutant (Fig. 3B, trace b), the negative band at 1697 cm Ϫ1 largely downshifted to 1680 cm Ϫ1 , where a differential signal was found in the Y D ⅐ /Y D spectrum (Fig. 5B, trace b), leaving a band at 1703 cm Ϫ1 , where there was a differential signal in the Y Z ⅐ /Y Z spectrum (Fig. 4B, trace a). Thus, the downshifted band at 1680 cm Ϫ1 can be assigned to the 13 1 -keto CϭO of P D2 in this mutant, although the remaining band at 1703 cm Ϫ1 is assigned to that of P D1 (Table 3). In contrast, in the P680 ϩ /P680 spectrum of the D1-V157H mutant, the frequency of a negative peak at 1697 cm Ϫ1 only slightly downshifted to 1695 cm Ϫ1 concomitantly with an intensity increase in the neighboring positive shoulder at 1708 cm Ϫ1 (Fig. 3B, trace a). This change is consistent with a 3-cm Ϫ1 downshift of the P D1 band at 1703 cm Ϫ1 observed in the Y Z ⅐ /Y Z difference spectrum upon D1-V157H mutation (Fig. 4B,  trace a).
The cause of the asymmetric hydrogen bond formation, i.e. weak and strong hydrogen bonding at P D1 and P D2 , respectively, by introduction of His residues replacing the Val residues (Table 3) was investigated by QM/MM calculations for the P680 models of the D1-V157H and D2-V156H mutants based on the high-resolution (1.9 Å) X-ray crystal structure of WT ( Fig. 7; Table 2) (26). It was found that the space around the 13 1 -keto CϭO group of P D1 is too narrow to fit the His side chain, and only a weak hydrogen bond with a long hydrogen bond distance (O⅐⅐⅐H) of 2.61 Å and a non-proper angle of 96.4°w as formed in D1-V157H ( Fig. 7B; Table 2). A small downshift of the CϭO frequency by 4 cm Ϫ1 was calculated, reproducing the experimental downshift of 3 cm Ϫ1 . In contrast, a His side chain fits better to the space around the 13 1 -keto CϭO of P D2 . A much stronger hydrogen bond with His was formed in D2-V156H with a shorter hydrogen bond distance of 1.94 Å and a better angle of 124.8°( Fig. 7C; Table 2), providing a 17-cm Ϫ1 downshift of the CϭO frequency, which reproduced the experimental downshift of 19 cm Ϫ1 . This difference in the space around the 13 1 -keto CϭO group between P D1 and P D2 is reflected by a shorter distance between the oxygen atom of the 13 1 -keto CϭO and the C␣ of the Val residue for P D1 (5.3 Å) than that for P D2 (5.8 Å) in the X-ray structure of WT (26). Thus, the asymmetric protein environment of P680 is the primary cause for the difference between the hydrogen bond strengths in the two mutants that introduced a His side chain for Val. In WT, the 13 1 -keto CϭO frequency of P D2 was calculated to be lower by 9 cm Ϫ1 than that of P D1 (Table 2), reproducing the experimental tendency that showed a lower frequency by 4 cm Ϫ1 (Table 3). It was also shown that the CϭO frequencies of the non-perturbed Chls in the mutants, i.e. P D2 in D1-V157H and P D1 in D2-V156H, were unchanged, reinforcing the independency of the two 13 1 -keto CϭO vibrations of P680.
In monomeric Chl, cation formation generally upshifts the 13 1 -keto CϭO frequency by ϳ30 cm Ϫ1 (39 -41). In a Chl dimer, the extent of upshift becomes smaller depending on the ratio of charge distribution between two Chls (22,34). In the P680 ϩ /P680 difference spectrum of WT (Fig. 3B, black line), two positive bands due to the 13 1 -keto CϭO vibrations of P680 ϩ were observed at 1726 and 1708 cm Ϫ1 , which were upshifted from a negative band at 1697 cm Ϫ1 consisting of two bands at 1703 and 1699 cm Ϫ1 (Table 3). The lower frequency of the 1697-cm Ϫ1 peak than the frequencies of the two constituents suggests that at least one of the two bands upshifts by a smaller extent than the bandwidth upon P680 oxidation, most probably forming a weak positive band at 1708 cm Ϫ1 . Hence, the original frequency of the latter band should also be slightly

-Keto C‫؍‬O frequencies (cm ؊1 ) of P D1 and P D2 detected in P680 ؉ /P680, Y Z ⅐ /Y Z , and Y D ⅐ /Y D FTIR difference spectra
a A middle frequency of the two peaks in a differential signal is shown. b An apparent peak frequency is shown, which is slightly shifted from the true frequency by a band overlap. c A shift from the WT value is given in parentheses.
lower than 1708 cm Ϫ1 . The presence of the two bands in P680 ϩ was previously attributed to the localization of 70 -80% of a charge on the Chl showing the higher-frequency band at 1726 cm Ϫ1 (22). The assignments of the two 13 1 -keto CϭO bands of P680 ϩ were achieved by inspection of the P680 ϩ /P680 spectra of the D1-V157H and D2-V156H mutants (Fig. 3). In the D2-V156H mutant, the 1708-cm Ϫ1 band downshifted to ϳ1689 cm Ϫ1 by ϳ19 cm Ϫ1 , the same extent as the shift of the neutral form, whereas the 1726-cm Ϫ1 band was unchanged (Fig. 3B, trace b). In contrast, in the D1-V157H mutant, the 1726-cm Ϫ1 band decreased the intensity concomitantly with the appearance of a broad feature around 1720 cm Ϫ1 , indicative of a downshift by ϳ6 cm Ϫ1 of the 13 1 -keto CϭO band in partial centers, whereas the frequency of the 1708-cm Ϫ1 band was little affected. The observation of the remaining intensity at 1726 cm Ϫ1 and a broad feature of the band around 1720 cm Ϫ1 may be attributed to an unstable hydrogen bonding interaction between the introduced His and the P D1 CϭO, which could cause destruction of the weak hydrogen bond in some centers. The above observations in the two mutants provide clear assignments of the 1726-and 1708-cm Ϫ1 bands to the 13 1 -keto CϭO vibrations of P D1 and P D2 , respectively, in P680 ϩ . It is therefore concluded that a positive charge is mainly localized on P D1 , providing a larger upshift. This conclusion is consistent with the mutation effects on the TL glow curves due to S 2 Q A Ϫ recombination in the presence of DCMU ( Fig. 2; Table 1). The D1-V157H mutant largely upshifted the peak temperature from 11 to 28°C by 17°C concomitantly with a significant intensity increase by a factor of 2.05 (Fig. 2a). In contrast, the D2-V156H mutant upshifts the peak temperature only by 3°C with a small intensity increase by a factor of 1.22 (Fig. 2b). It is known that the S 2 Q A Ϫ recombination takes place mainly through a relaxation pathway via P680 ϩ Pheo Ϫ (66,67), and the quantum yield of emission through P680* is very small (3% in WT (68)). Thus, the peak temperature of the TL band mainly reflects ⌬G between P680 ϩ Pheo Ϫ and S 2 Q A Ϫ , whereas the TL intensity reflects ⌬G between P680* and P680 ϩ Pheo Ϫ (66, 67). These energy gaps are expressed using the redox potentials of the components shown in Equations 1 and 2, where F is the Faraday constant, and ⌬G(P680*-P680) is the transition energy of P680 that has a constant value of 1.8 eV. In the D1-V157H and D2-V156H mutants that perturbed the hydrogen-bonding property of P680, E m (P680/P680 ϩ ) is expected to be mainly affected. Thus, the observed upshifts of the peak temperature, which represent the increases in ⌬G(P680 ϩ Pheo Ϫ -S 2 Q A Ϫ ), indicate the upshifts of E m (P680/ P680 ϩ ), to a larger extent in D1-V157H than in D2-V156H. The mutation-induced change in ⌬G(P680*-P680 ϩ Pheo Ϫ ), ⌬⌬G(P680*-P680 ϩ Pheo Ϫ ), which corresponds to the change in E m (P680/P680 ϩ ), ⌬E m (P680/P680 ϩ ), according to Equation 2, is expressed using the ratio of the TL intensity of the mutant (IЈ) to that of WT (I) (67) as shown in Equation 3, where T m is the peak temperature. Using Equation 3 and the observed intensity increase by mutation, ⌬E m (P680/P680 ϩ ) was estimated to be ϩ18.1 Ϯ 0.7 and ϩ4.9 Ϯ 0.4 mV in D1-V157H and D2-V156H, respectively (Table 1), in agreement with the tendency predicted by the shifts of the peak temperature. It is noted that the obtained ⌬E m values are based on a rather simplified assumption in the thermoluminescence theory, and hence they may be considered as approximate estimations. Nevertheless, the larger E m change by the perturbation of P D1 , despite much weaker hydrogen bonding at P D1 than P D2 , is consistent with the FTIR conclusion that a positive charge is mainly distributed to P D1 in P680 ϩ . In addition, the 18-mV increase is consistent with the weak hydrogen bond in D1-V157H, taking into account the 60 -80-mV increase in E m (P/ P ϩ ) by formation of a proper hydrogen bond with His at the 13 1 -keto CϭO of bacteriochlorophyll observed in the studies of bacterial reaction centers (14,34).
The conclusion in this study, i.e. charge localization on P D1 in P680 ϩ , obtained using the mutants introducing hydrogen bonds to P D1 and P D2 basically agrees with that of the previous study by Diner et al. (23), which used the mutants of the His ligands of these Chls. They analyzed the mutants by detecting the changes in the Soret and Q y bands. Although excitonic couplings among the chromophores were not considered in their analysis, later theoretical studies by Renger and co-workers (30 -32) showed that the excited states of P D1 and P D2 are appreciably coupled with each other. It was suggested that P D1 and P D2 , which have site energies at 666 nm in the Q y transitions, form two delocalized exciton states at ϳ675 and ϳ658 nm (31). The experimental observation of a blue shift of the Q y bleach at 672.5 nm by 3 nm upon D1-H198Q mutation (23) was well simulated by a blue shift of the site energy of P D1 by 8 nm (31). Because the experimental data of the Q y shifts by D2-His-197 mutations were not reported (23), the contribution of P D2 to the bleaching band was not revealed experimentally. In the Soret region, Diner et al. (23) observed blue shifts by up to 3 nm upon D1-His-198 mutations and red shifts by 0.5-1.5 nm upon D2-His-197 mutations. This opposite displacement in the D2-His-197 mutants was explained by the position of the P D2 band at a longer wavelength (436 nm) than the P D1 band (433 nm) and the increased contribution of the P D2 band to the difference spectrum due to the change in the E m of P D2 . However, the excitonic couplings of the Soret transitions have not been theoretically clarified, and hence the assumption of a weak coupling between P D1 and P D2 was not guaranteed. In contrast, in our study, the 13 1 -keto CϭO vibrations of P D1 and P D2 , which were detected by FTIR spectroscopy, are independent of each other, and a genetically introduced hydrogen bond changed only the CϭO frequency of the perturbed Chl (Table 3). Therefore, our results for the perturbations of the 13 1 -keto CϭO vibrations, together with the results for the perturbations of the electronic transitions by Diner et al. (23), lead to a definite conclusion that a positive charge mainly resides on P D1 .
Localization of a positive charge on P D1 is crucial in the physiological function of P680. A density functional theory (DFT) calculation of a P680 model previously estimated that charge localization on one Chl upshifts the E m (P680/P680 ϩ ) by ϳ140 mV compared with the charge-delocalized state (20). This E m increase is essential for the capability of Y Z oxidation, whose E m value has been estimated to be lower than that of P680 by ϳ110 mV (69). Indeed, in the isolated PSII reaction center complexes (D1-D2-Cyt b 559 ), where the electronic state of P680 ϩ is significantly altered and a positive charge is rather delocalized over P D1 and P D2 (22), P680 ϩ does not have a function of Y Z oxidation anymore. In addition to providing a driving force of Y Z oxidation by the high E m , the presence of a charge on P D1 may be advantageous to the electron transfer rate from Y Z to P680 ϩ due to the closer distance from Y Z to P D1 than to P D2 (the edge-to-edge distance of the conjugated systems is 8 and 17 Å for Y Z -P D1 and Y Z -P D2 , respectively (26)). Thus, the faster rate of Y Z oxidation (30 ns to 50 s) (70,71) than the rate of Y D oxidation (ϳms) (72,73) by P680 ϩ could be related to this asymmetric charge distribution, in addition to the difference in the proton transfer mechanism, i.e. Y Z shifts a proton only to the neighboring D1-His-190 (57,(62)(63)(64), whereas a proton from Y D is transferred to the bulk through a long proton pathway (58,65).

Construction of a D1-V157H site-directed mutant
Site-directed mutagenesis of the psbA2 gene encoding the D1 subunit in Synechocystis sp. PCC 6803 was performed following the method of Chu et al. (74) with some modifications. A psbA triple-deletion strain (⌬psbA1/⌬psbA2/⌬psbA3) was generated from the Synechocystis sp. PCC 6803 47-H strain, which has a six-histidine tag at the C terminus of the CP47 subunit and was selected by co-expression of the kanamycin (Km) resistance gene (75). The entire coding regions of psbA1 (slr1181) and psbA3 (sll1867) were replaced with the chloramphenicol (Cm) and erythromycin (Em) resistance genes, respectively, and the coding region of psbA2 (slr1311), except for the first 447 bp, was replaced with the gentamycin (Gm) resistance gene (76). The psbA triple-deletion strain was isolated and maintained on BG-11 agar plates containing 5 g/ml Km, 5 g/ml Cm, 5 g/ml Em, 2.5 g/ml Gm, 5 mM glucose, and 10 M 3-(3,4)dichlorophenyl-1,1-dimethylurea (DCMU) under a continuous low-light condition using white fluorescent lamps. The genotypes of the strains were confirmed by PCR analysis.
The coding region of psbA2, except for the first 2 bp, together with 630 bp downstream of the TAA stop codon in the genomic DNA was amplified by PCR using the following six primers, including five restriction enzyme sites: psbA2-1, 5Ј-GGA-TCTTCCAGAGATGAATTCCAACGACTCTCCAACA-GCG-3Ј (EcoRI site underlined); psbA2-2, 5Ј-CATATTGGA-AGATCAGTCGACCAAAGTAGCCGTGGGCG-3Ј (SalI site underlined); psbA2-3, 5Ј-CGGCTACTTTGGTCGACTGAT-CTTCCAATATGCTTCTTTCAAC-3Ј (SalI site underlined); psbA2-4, 5Ј-TATTCAGTTGGCATTGGATCCTCGAGTT-AACCGTTGACAGCAGGAGC-3Ј (BamHI and XhoI sites underlined); psbA2-5, 5Ј-GCTGTCAACGGTTAACTCGAG-GATCCAATGCCAACTGAATAATCTGCAAA-3Ј (XhoI and BamHI sites underlined); psbA2-6, 5Ј-CTGCCGTTCGACGA-TACTAGTGCGAGGGCAATCATCAATTCC-3Ј (SpeI site underlined). The PCR fragments were cloned in the pMD19-T vector (Takara) using the In-Fusion HD cloning kit (Clontech). The resultant plasmid was designated as pRN102. The Sm resistance gene was inserted into pRN102 at the BamHI site in the same direction as the psbA2 gene. This plasmid, designated as pRN123, was used as a parental vector for site-directed mutagenesis. The wild-type control strain of the D1 subunit (D1-WT) was obtained by transforming the psbA triple-deletion strain with pRN123. The D1-V157H mutation was introduced into pRN102 by replacing a GTA codon at the target site with a CAT codon using inverse PCR. The DNA fragment, which was obtained by digestion using EcoRI and SalI, was inserted into pRN123 at the corresponding site. The resultant plasmid was introduced into the psbA triple-deletion strain. The D1-WT and D1-V157H strains were isolated and maintained on BG-11 agar plates containing 5 g/ml Km, 5 g/ml Cm, 5 g/ml Em, and 5 g/ml Sm in the presence of 5 mM glucose and 10 M DCMU under a continuous low-light condition. The genotype of the D1-V157H mutant was confirmed by PCR analysis and DNA sequencing. No trace of the wild-type psbA2 gene was detected in all cultures of the mutant strain.

Construction of a D2-V156H site-directed mutant
Site-directed mutagenesis of the psbD1 gene encoding the D2 subunit in Synechocystis sp. PCC 6803 was performed according to the method of Tang et al. (77) with some modifications. A psbD1/psbD2/psbC-deletion strain (⌬psbD1/ ⌬psbD2/⌬psbC) was generated from the Synechocystis sp. PCC 6803 47-H strain (75). The entire coding regions of psbD1 (sll0849) and psbC (sll0851) were replaced with the Em resistance gene, although that of psbD2 (slr0927) was replaced with the Cm resistance gene. The psbD1/psbD2/psbC-deletion strain was isolated and maintained on BG-11 agar plates containing 5 g/ml Km, 5 g/ml Cm, 5 g/ml Em, 5 mM glucose, and 10 M DCMU under a continuous low-light condition. The genotype of the strain was confirmed by PCR analysis.
The coding region of psbD1 and psbC together with the 585 bp upstream of the ATG start codon of psbD1 and the 589 bp downstream of the TAG stop codon of psbC in the genomic DNA was amplified by PCR using the following six primers, including five restriction enzyme sites: psbD1-1, 5Ј-GGA-TCTTCCAGAGATCTGCAGAACTACTCAGTAACTCC-CGCCG-3Ј (PstI site underlined); psbD1-2, 5Ј-CACCAAGCA-AAACCAGTCGACGGAAGGTCACGTCCCCC-3Ј (SalI site underlined); psbD1-3, 5Ј-GGGACGTGACCTTCCGTCG-ACTGGTTTTGCTTGGTGGTCG-3Ј (SalI site underlined); psbD1-4, 5Ј-CTTTGCAAAATCAGAGGATCCTCGAGCT-AGTCGAGGTCAGGCATGAAC-3Ј (BamHI and XhoI sites underlined); psbD1-5, 5Ј-CCTGACCTCGACTAGCTCGA-GGATCCTCTGATTTTGCAAAGGTTTTGC-3Ј (XhoI and BamHI sites underlined); psbD1-6, 5Ј-CTGCCGTTCGACGA-TACTAGTGCCATTAAAGAATTGGCTAAAGAAGC-3Ј (SpeI site underlined). The PCR fragments were cloned in the pMD19-T vector (TaKaRa) using the In-Fusion HD cloning kit (Clontech), and the resultant plasmid was designated as pRN105. The Sm resistance gene was inserted into the pRN105 at the BamHI site in the same direction as the psbD1 and psbC genes. This plasmid, designated as pRN126, was used as a parental vector for site-directed mutagenesis. A wild-type control strain of the D2 subunit (D2-WT) was obtained by transforming the psbD1/psbD2/psbC-deletion strain with pRN126. D2-V156H mutation was introduced into pRN105 by replacing a GTC codon at the target site with a CAC codon using inverse PCR. The DNA fragment, which was obtained by digestion using XbaI and SalI, was inserted into pRN126 at the corresponding site. The resultant plasmid was introduced into the psbD1/psbD2/psbC-deletion strain. The D2-WT and D2-V156H strains were isolated and maintained on BG-11 agar plates containing 5 g/ml Km, 5 g/ml Cm, and 5 g/ml Sm in the presence of 5 mM glucose and 10 M DCMU under a continuous low-light condition. The genotype of the D2-V156H mutant was confirmed by PCR analysis and DNA sequencing. No trace of the wild-type psbD1 gene was detected in all cultures of the mutant strain.

Cell growth
WT and mutant cells on the agar plate were inoculated into 40 ml of BG-11 medium (78) supplemented with 4 mM Hepes-NaOH (pH 7.5) and 5 g/ml Km/Cm/Em/Sm and Km/Cm/Sm for D1-WT/D1-V157H and D2-WT/D2-V156H, respectively, and were grown photoautotrophically by bubbling with air containing 1% (v/v) CO 2 at 30°C under continuous illumination (20 mol photons m Ϫ2 s Ϫ1 ) using white fluorescent lamps. Cells at this stage were used for the measurements of O 2 evolution and TL glow curves. For PSII core preparation, cells were further grown in an 8-liter culture bottle without antibiotics under the photoautotrophic growth condition mentioned above. Cells cultured in six bottles (total volume of 48 liters) were used for preparation of PSII core complexes from each strain.

Preparation of PSII core complexes
PSII core complexes were purified according to the method by Sakurai et al. (79) with some modifications. Harvested cells were washed once with a buffer (pH 6.0) containing 50 mM Mes-NaOH, 5 mM CaCl 2 , 10 mM MgCl 2 , and 25% (w/v) glycerol (buffer A). The cells in buffer A were disrupted by glass beads as described previously (57), and then the lysate was diluted with an equal volume of a buffer (pH 6.0) containing 40 mM Mes-NaOH, 5 mM CaCl 2 , and 10 mM MgCl 2 . Unbroken cells were removed by centrifugation at 2000 ϫ g for 5 min, and then the supernatant was centrifuged at 48,000 ϫ g for 20 min, providing thylakoid membranes as resultant pellets. Thylakoids suspended in buffer A were solubilized with 1% (w/v) n-dodecyl ␤-D-maltoside (DM) at a Chl concentration of 1.0 mg/ml by stirring for 10 min on ice. After centrifugation at 27,000 ϫ g for 15 min, the resultant supernatant was applied to a Ni 2ϩ affinity column equilibrated with buffer A containing 0.04% DM (buffer B). The column was washed with 1 volume of buffer B containing 5 mM L-histidine and further washed with buffer B (pH 6.0) containing 100 mM NaCl and 100 mM imidazole-HCl until the eluate became colorless. PSII core complexes were eluted with buffer B containing 50 mM L-histidine and then concentrated by ultrafiltration (Vivaspin 20, Sartorius Stedim, 100-kDa molecular mass cutoff).
Manganese depletion from PSII was performed by treatment of PSII core complexes (0.35 mg Chl/ml) with 10 mM NH 2 OH in a buffer (pH 6.5) containing 20 mM Mes-NaOH, 5 mM NaCl, and 0.03% DM (buffer C) for 1 h on ice in the dark. The sample was then washed with buffer C and concentrated to ϳ3.5 mg of Chl/ml using ultrafiltration (Vivaspin 500, Sartorius Stedim, 100-kDa molecular mass cutoff). For Q A depletion, PSII core complexes in a buffer (pH 6.5) containing 200 mM Mes-NaOH, 5 mM NaCl, and 0.03% DM were treated with 100 mM sodium dithionite and 30 M benzyl viologen (80), followed by dark incubation for 18 h at 4°C. The manganese cluster was also removed during this treatment. The resultant Q A -depleted PSII complexes were washed with buffer C and concentrated to ϳ3.5 mg of Chl/ml using ultrafiltration.

Measurement of O 2 evolution activity
O 2 evolution activity was measured using a Clark-type oxygen electrode at 30°C. Cells (10 g of Chl) were suspended in BG11 medium containing 1 mM 2,6-dichloro-p-benzoquinone and 1 mM potassium ferricyanide as electron acceptors, and O 2 evolution was recorded upon illumination by saturating light.

TL measurements
WT and mutant cells were centrifuged at 1000 ϫ g for 5 min at 25°C and suspended in fresh BG-11 medium (0.25 mg of Chl/ml). The cells were then exposed to continuous light (200 mol photons m Ϫ2 s Ϫ1 ; ϳ16 milliwatts cm Ϫ2 at the sample point) from a white fluorescent lamp for 30 s at 30°C, followed by incubation at this temperature for 5 min in the dark. TL measurements were performed using a laboratory-built apparatus as described previously (81). A cell suspension (70 l) in the presence of 50 M DCMU was loaded onto a piece of filter paper and illuminated with continuous white light (ϳ55 milliwatts cm Ϫ2 at the sample point) from a halogen lamp (MEJIRO PRECISION PHL-150) for 10 s at Ϫ20°C. The sample was quickly cooled down and then warmed at a rate of 40°C/min to record a TL glow curve.

FTIR measurements
Light-induced FTIR difference spectra were recorded using a Bruker VERTEX 80 spectrophotometer equipped with an MCT detector (InfraRed D313-L) at 4 cm Ϫ1 resolution. A Ge filter to cut IR light at Ͼ2200 cm Ϫ1 (Andover, 4.50ILP-25) was placed in the IR path in front of the sample to improve the signal-to-noise ratios of spectra as well as to block a He-Ne laser beam from the interferometer. P680 ϩ /P680 FTIR spectra were measured following the method described previously (22). An aliquot (3.5-5 l) of the Q A -depleted PSII suspension was mixed with 1 l of 500 mM potassium ferricyanide and 1 l of 10 mM SiMo on a CaF 2 plate (13 mm in diameter). The sample was lightly dried under N 2 gas flow and covered with another CaF 2 plate with 0.7 l of water. The sample temperature was adjusted to 250 K in a liquid-N 2 cryostat (Oxford, model DN1704) using a tempera-ture controller (Oxford, model ITC-5). Single-beam spectra with two scans (1-s accumulation) were recorded under dark and during illumination. This measurement was repeated 5000 times, and average single-beam spectra were used to calculate a P680 ϩ /P680 difference spectrum as a light-minus-dark difference. Light illumination was performed with a continuous-wave beam at 661 nm (ϳ38 milliwatts cm Ϫ2 at the sample point) from a diode laser (L4660S-90-TE, Micro Laser Systems).
Y Z ⅐ /Y Z FTIR spectra were measured as described previously (57). An aliquot (4 -5 l) of the manganese-depleted PSII sample was mixed with 1 l of 100 mM potassium ferricyanide on a BaF 2 plate (13 mm in diameter). The sample was lightly dried under N 2 gas flow and covered with another BaF 2 plate with 0.7 l of water. The sample temperature was adjusted to 250 K in the cryostat. Single-beam spectra with 50 scans (25-s accumulation) were recorded twice before and once after single-flash illumination. The sample was then dark-adapted for 225 s. This measurement scheme was repeated 80 times, and average spectra were used to calculate a Y Z ⅐ /Y Z difference spectrum as lightminus-dark difference and a base line as a dark-minus-dark difference representing a noise level. Flash illumination was performed with a Q-switched Nd:YAG laser (Quanta-Ray INDI-40-10; 532 nm, ϳ7 ns full width at half-maximum) with a power of ϳ7 mJ pulse Ϫ1 cm Ϫ2 at the sample point.
Y D ⅐ /Y D FTIR spectra were measured with a method described previously (58). An aliquot (3 l) of the manganesedepleted PSII was mixed with 1 l of 20 mM potassium ferricyanide and 1 l of 20 mM potassium ferrocyanide on a BaF 2 plate (25 ϫ 25 mm). The sample was lightly dried under N 2 gas flow. The resultant sample film in an oval shape (6 ϫ 9 mm) was moderately hydrated by sealing the cell using another BaF 2 plate and a silicone spacer (0.5 mm in thickness) enclosing 2 l of 40% (v/v) glycerol solution without touching the sample (82). The sample temperature was adjusted to 10°C by circulating cold water in a copper holder. Single-beam spectra with 100 scans (50-s accumulation) were recorded twice before and once after five flashes (1 Hz) from the Nd:YAG laser. The sample was then dark-adapted for 750 s. This measurement scheme was repeated 50 times, and average single-beam spectra were used to calculate a Y D ⅐ /Y D difference spectrum as a light-minus-dark difference and a base line as a dark-minus-dark difference representing a noise level.

QM/MM calculation of P680 models
QM/MM calculations were performed following the method described previously (83,84). The initial coordinates of PSII models were obtained from the X-ray structure at a 1.9 Å resolution (Protein Data Bank code 3ARC) (26). In addition to P D1 and P D2 , cofactors (Chl D1 , Chl D2 , Pheo D1 , and Pheo D2 ), amino acid residues, and water molecules located within 10 Å from the heavy atoms of P D1 and P D2 (including the phytol chains) were extracted from the X-ray structure (the whole QM/MM region is shown in Fig. 6). Hydrogen atoms of amino acid residues were generated and optimized using the AMBER force field (85), whereas those of Chl and Pheo molecules were originally produced and optimized by the DFT method with the B3LYP functional using 6 -31G(d) as a basis set. Atomic charges of these molecules were also calculated as electrostatic potentials. QM/MM calculations were performed using the two layer ONIOM method (86) with the electronic embedding scheme in the Gaussian 09 program package (87). The QM region consists of P D1 and P D2 (without the phytol chains), and the side chains of D1-His-198, D2-His-197, D1-Val-157, and D2-Val-156 (Fig.  7A). For the models of D1-V157H and D2-V156H mutants, D1-Val-157 and D2-Val-156, respectively, were replaced with a His side chain (Fig. 7, B and C). As an initial structure, a His side chain (neutral N H or N H form) was positioned so that the NH group interacts with the 13 1 -keto CϭO of P D1(D2) . In the QM/MM geometry optimization, the coordinates of the QM region were fully relaxed, whereas those of the MM region were fixed. Geometry optimization and normal mode calculations of the QM region were performed using the DFT method at the B3LYP/6 -31G(d) level. Calculated 13 1 -keto CϭO frequencies were scaled using a scaling factor of 0.9416 to adjust the calculated value of the P D1 CϭO (1808.6 cm Ϫ1 ) to the experimental one of 1703 cm Ϫ1 obtained from the Y Z ⅐ /Y Z difference spectrum (Table 2).