D1-Asn-298 in photosystem II is involved in a hydrogen-bond network near the redox-active tyrosine YZ for proton exit during water oxidation

In photosynthetic water oxidation, two water molecules are converted into one oxygen molecule and four protons at the Mn4CaO5 cluster in photosystem II (PSII) via the S-state cycle. Efficient proton exit from the catalytic site to the lumen is essential for this process. However, the exit pathways of individual protons through the PSII proteins remain to be identified. In this study, we examined the involvement of a hydrogen-bond network near the redox-active tyrosine YZ in proton transfer during the S-state cycle. We focused on spectroscopic analyses of a site-directed variant of D1-Asn-298, a residue involved in a hydrogen-bond network near YZ. We found that the D1-N298A mutant of Synechocystis sp. PCC 6803 exhibits an O2 evolution activity of ∼10% of the wild-type. D1-N298A and the wild-type D1 had very similar features of thermoluminescence glow curves and of an FTIR difference spectrum upon YZ oxidation, suggesting that the hydrogen-bonded structure of YZ and electron transfer from the Mn4CaO5 cluster to YZ were little affected by substitution. In the D1-N298A mutant, however, the flash-number dependence of delayed luminescence showed a monotonic increase without oscillation, and FTIR difference spectra of the S-state cycle indicated partial and significant inhibition of the S2 → S3 and S3 → S0 transitions, respectively. These results suggest that the D1-N298A substitution inhibits the proton transfer processes in the S2 → S3 and S3 → S0 transitions. This in turn indicates that the hydrogen-bond network near YZ can be functional as a proton transfer pathway during photosynthetic water oxidation.

Plants and cyanobacteria utilize water as an ultimate electron donor to reduce carbon dioxide in the synthesis of sugars (1). This reaction of water oxidation is performed in photosystem II (PSII) 3 protein complexes, which are embedded in thylakoid membranes. Water oxidation produces protons and molecular oxygen (2)(3)(4)(5)(6)(7)(8)(9); protons are released into the thylakoid lumen, making a proton gradient across the membrane, and are used to synthesize ATP, whereas molecular oxygen is liberated to the air to be the major oxygen source of the atmosphere.
Efficient proton exit from the catalytic site is essential in the mechanism of water oxidation. However, it remains unclear how the four protons are transferred from WOC to the lumen through the PSII proteins during the S-state cycle. The X-ray structures of PSII complexes (17,18,27,28) and theoretical calculations based on them have shown that there are several channels suitable for proton transfer around the Mn 4 CaO 5 cluster leading to the lumen (Fig. 1A) (29 -39). These channels involve a number of water molecules forming hydrogen-bond networks, as revealed by the recent high-resolution X-ray structures that resolved water oxygen atoms (17)(18)(19)(20)(21). Near the entrance of each channel, either D1-Asp-61 or Y Z is located, and a water cluster is formed between these two residues including water ligands to Mn4 (W1 and W2) and Ca 2ϩ (W3 and W4).
The channel from D1-Asp-61 to PsbO through the Cl-1 site and the D1-Glu-65/D2-Glu-312/D1-Arg-334 triad has been most extensively studied and proposed to function as a pathway for protons or substrate water (designated "Cl path" in the present report; Fig. 1A) (29 -37, 39). Theoretical calculations suggested that D1-Asp-61 plays a crucial role in proton transfer (40,41), and that the Cl Ϫ ion has a function to regulate proton transfer through this pathway (34). D1-Asp-61 is also located near the entrance of a water chain starting at O4 and leading to PsbU (designated "O4 path"; Fig. 1A), which has been proposed to function as a proton exit pathway especially in the S 0 3 S 1 transition (38,39). The recent X-ray free electron laser study for the S 3 -state showed the displacement of a water molecule near O4, suggesting the involvement of water molecules near O4 in proton transfer (21). The hydrogen-bond network near Y Z has also been suggested to be a proton pathway (16,17,42,43). Y Z form a hydrogen-bonded triad with D1-His-190 and D1-Asn-298, which is connected with a hydrogen-bond network leading to PsbV through D1-Asn-322 (designated "Y Z -Asn-298 path"; Fig. 1A). Molecular dynamics (MD) simulations (36, 39) sug-gested a tight hydrogen-bond network of this pathway, which is advantageous to proton transfer. A large water cluster interacting the Y Z /D1-His-190/D1-Asn-298 triad also forms another hydrogen-bond network through O1 and D1-Glu-329 leading to PsbV (designated "Y Z -O1 path"). MD simulation showed that this channel contains a number of mobile water molecules (33,39). It is noted that the above key amino acid residues in the hydrogen-bond networks in cyanobacterial PSII (Fig. 1A) are all conserved in higher plants whose PSII-LHCII structures were recently resolved by cryo-electron microscopy (44,45).
The finding of several channels around the Mn 4 CaO 5 cluster raises a question, which proton pathways are used for individual protons released in the S-state transitions? The most direct experimental method to answer this question is examining the effects of site-directed mutations, by which amino acid residues on putative proton channels are altered to different amino acids, on the reactions of individual S-state transitions. Many mutants of amino acid residues in putative channels have been investigated so far (46 -60). Most of them are located on the Cl path; mutations of D1-Asp-61 to Ala and Asn significantly retarded the kinetic rate of O 2 release in the S 3 3 S 0 transition (47,53,58) and changed the property of a nearby water network (57). Mutations of D2-Lys-317 and D1-Asn-181, which are ligands to Cl-1, also retarded the O 2 kinetics or decreased the efficiency of the S 3 3 S 0 transition (54,55,59). In addition, mutants of any of the D1-Glu-65/D2-Glu-312/D1-Arg-334 triad substantially inhibited the S 3 3 S 0 transition (49,52). These data suggest that the Cl path is involved in proton or water transfer at least in the S 3 3 S 0 transition, although a caution is necessary in the interpretation of results for charged residues, whose mutations could affect the redox potential of Figure 1. A, hydrogen-bond network around the Mn 4 CaO 5 cluster and putative proton transfer pathways. The X-ray free electron laser structure of Thermosynechococcus vulcanus at 1.95-Å resolution (PDB code 4UB6 (18)) was used to draw the picture. Mn, purple; Ca, orange; Cl, green; O, red; N, blue. Oxygen atoms of water molecules are colored cyan. The numbering of the Mn and oxygen atoms in the Mn 4 CaO 5 cluster and its water ligands follows the numbering in Umena et al. (17). D1-Asn-298, which was replaced with Ala in this work, is circled with a red dotted line. Magenta arrows represent putative proton pathways, which were designated Cl path, O4 path, Y Z -O1 path, and Y Z -Asn-298 path in this work. Hydrogen bonds are shown with green dotted lines. Amino acid residues with labels are on the D1 subunit except for Lys-317 and Glu-312 on the D2 subunit, and Glu-354 and Ala-411 on the CP43 subunit. B, the S-state cycle of water oxidation.

Proton transfer pathway in photosynthetic water oxidation
the Mn 4 CaO 5 cluster and hence the electron transfer reaction. In contrast, fewer mutagenesis studies have been performed for the channels near the Y Z site. Mutation of D1-Gln-165, which is hydrogen-bonded with W4 near Y Z , to Glu decreased the efficiency of the S 3 3 S 0 transition (56), whereas mutation of D1-Glu-329 on the Y Z -O1 path to Gln little affected the S-state transitions (52). Fourier transform infrared (FTIR) analysis of the S-state cycle using various mutants by Debus and co-workers (52,54,56,57) further showed the presence of an extensive hydrogen-bond network around the Mn 4 CaO 5 cluster.
D1-Asn-298 is located at a crucial position to keep a hydrogen-bond network involving water molecules in the Y Z -Asn-298 and Y Z -O1 pathways (Fig. 1A). This has also been proposed to be a key residue to determine the hydrogen-bonded structure of Y Z and the rate of its proton-coupled electron transfer through the interaction with D1-His-190 (15,16). Thus, mutation of D1-Asn-298 could affect the proton transfer from the catalytic site as well as electron transfer from the Mn 4 CaO 5 cluster to Y Z ⅐ during S-state transitions. Yamasato et al. (50) previously found that the D1-N298I mutant obtained by random mutagenesis inhibited electron flow from water to the electron acceptor side. More recently, Kuroda et al. (60) generated 19 site-directed mutants of D1-Asn-298 using Chlamydomonas reinhardtii, and showed that only seven mutants (D1-N298G, -A, -C, -M, -S, -Q, and -H) have O 2 evolution activity but with significantly low rates. However, detailed properties of the mutants of D1-Asn-298, such as specific inhibition of either electron or proton transfer in individual S-state transitions, have not yet been investigated, and hence no clear answer was obtained about the involvement of pathways near the Y Z site in the water oxidation mechanism.
In this work, we investigated whether the hydrogen-bond network near Y Z is functional or not in proton transfer during water oxidation, by constructing a D1-N298A mutant using a cyanobacterium Synechocystis sp. PCC 6803, and analyzing this mutant by means of light-induced FTIR difference spectroscopy and detection of thermoluminescence (TL) and delayed luminescence (DL). Because Asn and Ala are both non-charged amino acids and the replacement with a small Ala side chain can avoid the undesirable alterations of main chains, only the perturbation of the hydrogen-bond network is expected to occur without changes in the electrostatic interaction and protein conformations. FTIR difference spectroscopy is a powerful method for detecting the structural perturbations of amino acid residues, water molecules, and the hydrogen-bond network around the Mn 4 CaO 5 cluster, as well as monitoring the inhibition of the S-state transitions (61)(62)(63)(64)(65)(66)(67)(68)(69)(70). The results provided solid evidence for the involvement of D1-Asn-298, and hence the hydrogen-bond network near Y Z , in proton transfer in the S 2 3 S 3 and S 3 3 S 0 transitions of water oxidation.

In vivo analyses of the D1-N298A mutant
The strains of Synechocystis sp. PCC 6803 used in the present study lacked the psbA1 and psbA3 genes, and hence PsbA2 was expressed as a D1 protein (71). In addition, a histidine tag was attached to the C terminus of the CP47 protein. Thus, the strain having a PsbA2 protein with a native sequence is designated "WT*" hereafter. In contrast, the Asn-298 of PsbA2 was replaced by Ala in the D1-N298A mutant. The D1-N298A mutant grew photoautotrophically with a rate slightly slower than the WT* ( Fig. 2A). The O 2 evolution activities of WT* and D1-N298A were 600 -670 and 60 -70 mol of O 2 (mg of Chl) Ϫ1 h Ϫ1 , respectively, in cells, and 2800 -3300 and 190 -300 mol of O 2 (mg of Chl) Ϫ1 h Ϫ1 , respectively, in isolated PSII complexes. Thus, the O 2 evolution activity of the D1-N298A mutant is about 10% of WT* in both cells and PSII core complexes. This reduced activity of the D1-N298A mutant is in good agreement with that of a corresponding mutant of a green alga C. reinhardtii, which showed a 11.5% activity of WT in cells (60).
TL glow curves of PSII in the presence and absence of DCMU originate from charge recombination of S 2 Q A Ϫ (Q band) and S 2 Q B Ϫ (B band), respectively (72). WT* cells showed Q and B bands at about 14 and 34°C, respectively, and they were virtually unaffected by the D1-N298A mutation (Fig. 2B). In contrast, the flash-number dependence of DL, which was monitored at 0.8 s after each flash, showed significantly different patterns between WT* and D1-N298A cells (Fig. 2C). Whereas WT* cells showed a typical period-four oscillation of DL signals, D1-N298A cells did not show any oscillation and the DL amplitude monotonically increased with a flash number.

FTIR difference spectrum upon Y Z oxidation
Light-induced FTIR difference spectra upon Y Z oxidation (Y Z ⅐ /Y Z difference spectra) were obtained using manganesedepleted PSII core complexes of WT* and D1-N298A (Fig. 3). Characteristic bands of the Y Z ⅐ /Y Z spectra were virtually iden-

Proton transfer pathway in photosynthetic water oxidation
tical between the two species. The prominent positive band at 1512 cm Ϫ1 has been assigned to the CO stretching vibration of the oxidized Y Z ⅐ radical, whereas a negative band at 1258 -1257 cm Ϫ1 has been assigned to the coupled CO stretching/COH bending vibration of neutral Y Z (16,73). A differential signal at 1707/1700 cm Ϫ1 and a negative peak at 1113-1110 cm Ϫ1 have been attributed to the keto C ϭ O vibration of P D1 in P680 affected by Y Z oxidation (71) and the imidazole CN stretch of D1-His-190 hydrogen bonded with Y Z (73), respectively. Bands in the 1680 -1630 cm Ϫ1 region, and a band at 1541 cm Ϫ1 can be assigned to amide I and II vibrations, respectively, representing the perturbations of protein backbones. Except for changes in other minor bands and some intensity changes in the amide I and II bands, a notable change by the mutation was observed only in the appearance of a differential signal at 1244/1234 cm Ϫ1 in the spectrum of D1-N298A, although the origin of this signal is unknown at the present stage. The absence of significant changes in the characteristic bands upon the D1-N298A mutation indicates that the hydrogen-bonded interaction of Y Z was virtually unaffected by this mutation.

Flash-induced FTIR difference spectra of S-state transitions
Light-induced FTIR difference spectra of the S-state cycle of WOC were measured by applying four flashes on the O 2 -evolving PSII core complexes from WT* and D1-N298A. The first-, second-, third-, and fourth-flash spectra of the WT* PSII complexes in a moderately hydrated film (Fig. 4A, black lines) represent the structural changes upon the S 1 3 S 2 , S 2 3 S 3 , S 3 3 S 0 , and S 0 3 S 1 transitions, respectively. The spectral features were very similar to those of previously reported spectra (52, 54 -57, 74 -81). Bands at 1700 -1600 cm Ϫ1 mainly arise from the amide I vibrations (CO stretching vibrations of backbone amides), representing the perturbations of polypeptide chains near the Mn 4 CaO 5 cluster (78,82). Characteristic bands at 1450 -1350 cm Ϫ1 are due to the symmetric COO Ϫ stretching vibrations (78, 80 -85), whereas the coupled asymmetric COO Ϫ stretching bands appear at 1600 -1500 cm Ϫ1 , overlapping the amide II bands (NH bending and CN stretching vibrations of backbone amides) (78,82). The changes of these COO Ϫ bands during the S-state cycle are attributed to the structural changes of carboxylate residues around the Mn 4 CaO 5 cluster (80,81,(83)(84)(85). The obtained spectra showed characteristic features depending on the flash number, especially in the symmetric COO Ϫ region around 1400 cm Ϫ1 , reflecting high efficiencies of the S-state transitions.
Spectra of the S-state cycle of the D1-N298A mutant at the first to fourth flashes are also shown in Fig. 4A (red lines) superimposing those of WT* (black lines). These spectra of D1-N298A were scaled using a factor to adjust the intensity of

Proton transfer pathway in photosynthetic water oxidation
the CN band of ferricyanide at 2116 cm Ϫ1 at the first-flash (not shown) to that of WT * , representing the same extent of an electron flow in PSII in the S 1 3 S 2 transition. Although the firstflash spectrum due to the S 1 3 S 2 transition showed overall features similar to those of WT*, some differences were observed in several bands (Fig. 4A, a). In the symmetric COO Ϫ region, band intensities at 1417(Ϫ) and 1364(ϩ) cm Ϫ1 were significantly reduced. In addition, bands at 1652(ϩ), 1587(ϩ), 1543(Ϫ), 1523(Ϫ), and 1508(ϩ) cm Ϫ1 in the amide I and the asymmetric COO Ϫ /amide II regions decreased their intensities. Furthermore, a negative band at 1747 cm Ϫ1 , which has been assigned to the C ϭ O stretching vibration of an unidentified COOH group involved in a hydrogen-bond network around the Mn 4 CaO 5 cluster (52,56), diminished and is replaced by a positive band at 1745 cm Ϫ1 (Fig. 4B, a). In contrast, spectral features of the second-flash spectra of D1-N298A were very similar to those of WT* (Fig. 4Ab). However, an overall intensity in the COO Ϫ region (1450 -1300 cm Ϫ1 ) was ϳ60% of WT*. A small positive peak at 1745 cm Ϫ1 due to a COOH group, which has a different origin from the negative band at 1747 cm Ϫ1 at the first flash (56), also diminished ( Fig. 4B, b).
The most striking change in the spectra of the D1-N298A mutant is that the third-and fourth-flash spectra showed very small intensities (Fig. 4A, c and d), indicating that the S 3 3 S 0 transition was significantly inhibited in this mutant. Assuming that the FTIR spectrum of the S 3 3 S 0 transition in the D1-N298A mutant has an identical feature with that of WT*, spectral fitting in the COO Ϫ region of the third-flash spectrum of the mutant using the second-and third-flash spectra of WT* (76) provided the ϳ20% contribution of the S 3 3 S 0 transition at the third flash. With the ϳ60% contribution of the S 2 3 S 3 transition at the second flash, it follows that the efficiency of the S 3 3 S 0 transition in the D1-N298A mutant is ϳ30% of WT*. Previously, the D2-K317R mutant also showed similar intensity decreases in the third-and fourth-flash FTIR spectra in a moderately hydrated film (55), but the intensities recovered in a solution sample. It was suggested that partial dehydration reduced the efficiency of the S 3 3 S 0 transition in the K317R mutant (55). To examine the similar dehydration effect, FTIR spectra of the S-state cycle were measured in solution samples of WT* and D1-N298A (Fig. 5). All the spectral changes by the D1-N298A mutation mentioned above were also observed in the solution samples, indicating that these changes are not ascribed to partial dehydration but are attributed to direct effects of the D1-N298A mutation irrespective of sample conditions.
The mutation effects were also observed in water vibrations obtained in hydrated PSII films. In the high frequency region of 3700 -3500 cm Ϫ1 (Fig. 6A), weakly hydrogen-bonded OH vibrations of water molecules (77, 86, 87) showed a differential signal at 3607/3588 cm Ϫ1 and a negative band at 3662 cm Ϫ1 in the first-flash spectrum of WT*, analogously to the previous reports (57,77,80,86,87). In the D1-N298A mutant, the negative peak at 3662 cm Ϫ1 diminished, whereas another negative peak at 3588 cm Ϫ1 seems unchanged (Fig. 6A, a). At the second flash, a broad negative band at ϳ3620 cm Ϫ1 in WT* (57, 77) seemed slightly downshifted upon the D1-N298A mutation (Fig. 6A, b).
In the region of vibrations of polarizable protons in strong hydrogen bonds (3100 -2200 cm Ϫ1 ) (77, 87), a broad positive feature with several sharp peaks was observed in the first-flash spectrum of WT* (Fig. 6B, a). This band feature was attributed to the OH stretching vibrations of water molecules around the Mn 4 CaO 5 cluster (87), overlapping the NH stretching vibration with Fermi resonance peaks of a His side chain (D1-His-337)

Proton transfer pathway in photosynthetic water oxidation
hydrogen-bonded with the O3 oxo bridge of the Mn 4 CaO 5 cluster (88). The intensities of these bands were weakened upon the D1-N298A mutation. In the second-flash spectra (Fig. 6B,  b), WT* exhibited broad positive features at 2681 and 2456 cm Ϫ1 (57,77), whereas a new positive band appeared at 2907 cm Ϫ1 upon the D1-N298A mutation leaving the broad features.

Discussion
D1-Asn-298 is a crucial amino acid residue in retaining a hydrogen-bond network near Y Z . It supports a water chain that leads from W4 to PsbV through D1-Asn-191 and D1-Asn-322 (Y Z -Asn-298 path), and a water cluster that interacts with O1 of the Mn 4 CaO 5 cluster and D1-Glu-329 leading to PsbV (Y Z -O1 path) (Fig. 1A), both of which have been candidates of proton/ water pathways (31-33, 35-37, 39). Thus, mutation of D1-Asn-298 to non-hydrogen bonding Ala is expected to significantly perturb the structures of these hydrogen-bond networks by reorganizing nearby water molecules. In addition, D1-Asn-298 is a key amino acid residue in determining the hydrogenbonded structure of Y Z . The amide C ϭ O of the Asn-298 side chain accepts an hydrogen bond from the N-H of D1-His-190, whereas the N of His-190 functions as a hydrogen-bond acceptor of the OH of Y Z (15)(16)(17)(18). This hydrogen-bonded structure of the Y Z /D1-His-190/D1-Asn-298 triad realizes rapid proton-coupled electron transfer of Y Z , in which upon oxidation by P680 ϩ , the proton of Y Z is immediately shifted to D1-His-190 forming a HisH ϩ cation (15,16). The D1-N298A mutation hence could also perturb the hydrogen-bonded structure of Y Z and its reaction. In contrast, mutation from Asn to Ala, both of which are non-charged amino acids, should have little electrostatic effect on Y Z and the Mn 4 CaO 5 cluster to change their redox potentials. Furthermore, a relatively remote location (Ͼ10 Å) of D1-Asn-298 from the Mn 4 CaO 5 cluster without any direct interactions even with the first shell ligands makes less likely the possibility that its mutation has a direct effect on the catalytic reaction of water oxidation.
We examined the effect of the D1-N298A mutation on the hydrogen-bonded structure of Y Z by comparing the Y Z ⅐ /Y Z FTIR difference spectrum of the manganese-depleted PSII core complex from the D1-N298A mutant with that from WT*. The results showed that they had very similar features (Fig. 3). In particular, the vibrational frequencies of the CO stretching band of Y Z ⅐ (1512 cm Ϫ1 ) and the CO stretching/COH bending band of neutral Y Z (1257 cm Ϫ1 ), which are sensitive to changes in hydrogen-bonding interactions, were identical within 1 cm Ϫ1 , indicating that the hydrogen-bonded interaction of Y Z was little affected by this mutation. It is presumed that the replacement of Asn with a smaller Ala produced enough space for a new water molecule to bind to the N of D1-His-190 instead of the Asn C ϭ O. Thus, the hydrogen-bonding pattern of the Y Z /D1-His-190/D1-Asn-298 triad was probably unchanged in the Y Z /D1-His-190/H 2 O triad in the mutant, although the possibility remains that in intact PSII with the Mn 4 CaO 5 cluster, the mutation has some effect on the Y Z interaction through the hydrogen-bond network around the Mn 4 CaO 5 cluster.
The reaction process of the S 1 3 S 2 transition is thought to be simple electron transfer from the Mn 4 CaO 5 cluster to Y Z ⅐ .
Hence, if electron transfer from the Mn 4 CaO 5 cluster to Y Z ⅐ is impaired by the D1-N298A mutation, the efficiency of the S 1 3 S 2 transition should also be affected. The S 2 /S 1 FTIR difference spectra of WT* and the D1-N298A mutant at the first flash (Figs. 4A, a, and 5a) showed similar overall features but with intensity changes and slight frequency shifts in some bands. At the second flash, the spectrum of D1-N298A showed very similar features to the spectrum of WT* representing the S 2 3 S 3 transition (Figs. 4A, b, and 5b), although the mutant spectrum had a smaller overall intensity (ϳ60% of WT*). The absence of the features of the S 2 /S 1 spectrum in the second-flash spectrum of the mutant (e.g. 1432 cm Ϫ1 peak characteristic of the firstflash spectrum was not involved in the second-flash spectrum) indicates that the S 1 3 S 2 efficiency in D1-N298A is as high as that in WT*. In addition, the peak temperatures of the TL glow curves at 14 and 34°C arising from charge recombination of S 2 Q A Ϫ (Q band) and S 2 Q B Ϫ (B band), respectively, little changed upon the D1-N298A mutation (Fig. 2B). This indicates that the redox potential of the S 1 3 S 2 transition of the Mn 4 CaO 5 cluster, which should regulate the electron transfer reaction, is virtually unchanged by this mutation. These observations suggest that the D1-N298A mutation little affected the electron transfer from the Mn 4 CaO 5 cluster to Y Z ⅐ at least during the S 1 3 S 2 transition.
In contrast to unaffected electron transfer, the hydrogenbond network around the Mn 4 CaO 5 cluster was clearly affected by the D1-N298A mutation, as revealed by the following observations. (i) The symmetric COO Ϫ region (1450 -1300 cm Ϫ1 ) in the S 2 /S 1 FTIR difference spectra showed characteristic changes, such as intensity decreases of the 1417 and 1364 cm Ϫ1 bands by the mutation (Figs. 4A, a, and 5a). These bands were recently assigned to the symmetric COO Ϫ vibrations of D1-Asp-342/D1-Asp-61 and D1-Asp-170/D1-Glu-333/D1-Asp-342/D1-Glu-189, respectively, by quantum mechanics/ molecular mechanics (QM/MM) calculations (85), suggesting slight perturbations of these carboxylate groups by the mutation. (ii) A negative band at 1747 cm Ϫ1 in the S 2 /S 1 spectrum and a positive band at 1745 cm Ϫ1 in the S 3 /S 2 spectrum (Fig.  4B), which have been assigned to the CO stretching vibrations of unidentified COOH groups (52,56), diminished in the spectra of the D1-N298A mutant, indicating the pK a changes of these COOH groups were located on a common hydrogenbond network involving D1-Asn-298. (iii) The intensities of the amide II bands in the 1590 -1500 cm Ϫ1 region together with an amide I band at 1652 cm Ϫ1 in the S 2 /S 1 spectrum significantly decreased (Figs. 4A, a, and 5a), suggesting the perturbation of the backbone interactions near D1-Asn-298. (iv) A negative band at 3662 cm Ϫ1 in the weakly hydrogen-bonded OH region of water (57,77,87) in the S 2 /S 1 spectrum diminished and a broader negative band at 3620 cm Ϫ1 in the S 3 /S 2 spectrum (57, 77) seemed slightly downshifted (Fig. 6A). (v) The intensity of the feature in the 3000 -2600 cm Ϫ1 region, which have been assigned to strongly hydrogen-bonded OH vibrations of water molecules (77,87) overlapping the NH vibration of protonated D1-His-337 (88), in the S 2 /S 1 difference spectrum decreased by the D1-N298A mutation (Fig. 6B, a). In addition, in the similar region in the S 3 /S 2 spectrum, a positive band newly appeared at 2907 cm Ϫ1 (Fig. 6B, b). These spectral changes in (iv) and (v)

Proton transfer pathway in photosynthetic water oxidation
represent some perturbations in the hydrogen-bonded structure of a water network near the Mn 4 CaO 5 cluster by the D1-N298A mutation.
Similar changes in the symmetric COO Ϫ , COOH, and amide II regions in FTIR difference spectra have been previously observed in site-directed mutants of D1-Asp-61, D1-Glu-65, D2-Lys-317, D2-Glu-312, and D1-Arg-334 on the Cl path, and those of D1-Gln-165 near Y Z and D1-Glu-329 on the Y Z -O1 path (Fig. 1A) (52, 54 -57). In addition, the similar decrease in the weakly hydrogen-bonded OH band at ϳ3662 cm Ϫ1 and the broad feature in the 3000 -2400 cm Ϫ1 region in the S 2 /S 1 difference spectrum was observed in the D1-D61A mutant (57). Indeed, the contribution of W1, which directly interacts with D1-Asp-61, to the bands at ϳ3662 and 3000 -2400 cm Ϫ1 was suggested by the QM/MM calculation (87). Recently, diminishing of the ϳ3662 cm Ϫ1 band in the S 2 /S 1 spectrum was also observed in PSII in which Ca 2ϩ is substituted with Sr 2ϩ (89). All of the above observations, therefore, indicate that a hydrogenbond network structurally coupled with the Mn 4 CaO 5 cluster is widely extended across the channels near Cl and Y Z involving D1-Asn-298.
The S 3 /S 2 FTIR difference spectrum at the second flash decreased its overall intensity to ϳ60% of WT* by the D1-N298A mutation (Figs. 4A, b, and 5b). Furthermore, the FTIR spectra at the third and fourth flashes showed very small intensities (Figs. 4A and 5). These observations were identical between samples in hydrated films (Fig. 4) and solutions (Fig. 5). These FTIR data of isolated PSII core complexes are consistent with the in vivo data of DL measurements using D1-N298A cells, in which the DL intensity monotonically increased with the flash number and reached near maximum by 4 or 5 flashes, in contrast to WT* cells that showed a clear period-four oscillation (Fig. 2C). In addition, the 10-fold decrease in the O 2 evolution activity by the D1-N298A mutation was identical between cells and PSII core complexes. These results indicate that the intensity changes in the FTIR spectra by mutation are ascribed to neither the core preparations nor the measurement conditions, but represent the real changes in efficiencies in the S-state transitions. It is thus concluded that the efficiency of the S 2 3 S 3 transition is lowered to ϳ60% and the S 3 3 S 0 transition is significantly inhibited upon the D1-N298A mutation.
It is generally accepted that one and two protons are released in the S 2 3 S 3 and S 3 3 S 0 transitions, respectively (24 -26). With the absence of the prominent effect on the electron transfer from the Mn 4 CaO 5 cluster to Y Z ⅐ in the S 1 3 S 2 transition, together with the unlikeliness of direct involvement of D1-Asn-298 in the catalytic reaction at the Mn 4 CaO 5 cluster because of its remote distance, the observed efficiency decreases in the S 2 3 S 3 and S 3 3 S 0 transitions suggest that the proton transfer from the catalytic site are partially and significantly inhibited in the S 2 3 S 3 and S 3 3 S 0 transitions, respectively, in the D1-N298A mutant. Alternative possibilities, however, cannot be excluded at the present stage that electron transfer reactions are affected in the S-state transitions other than S 1 (56), and D1-E329Q (52), have been examined to investigate the effects of mutations on the S-state transitions. However, none of these mutants apparently blocked the S 2 3 S 3 transition. Although some changes in the symmetric COO Ϫ stretching region of the second-flash FTIR spectra in mutants at the D1-Glu-65/D2-Glu-312/D1-Arg-334 triad were interpreted to be due to a decreased S 2 3 S 3 efficiency (52,56), electrostatic effects by mutations of charged amino acid residues have not been identified. In addition, although the D1-D61N mutation slowed the S 2 3 S 3 transition by a factor of 2-4, the S 1 3 S 2 transition was also slowed by a similar extent, and hence this retardation was suggested to be caused by the increase in the redox potential of the Mn 4 CaO 5 cluster (47,53). The partial inhibition of the S 2 3 S 3 efficiency in the D1-N298A mutant observed in our study suggests that when the main proton pathway was blocked by this mutation, another pathway was used for the proton exit in the S 2 3 S 3 transition or this transition originally uses multiple proton pathways. Such flexible proton transfer through different pathways may be realized by a widely extended hydrogenbond network involving several different pathways around the Mn 4 CaO 5 cluster mentioned above.
Because significant inhibition by the D1-N298A mutation was observed in the S 3 3 S 0 transition, at least one of two protons released in this transition most likely uses the pathway near Y Z perturbed by this mutation. The other proton may use the Cl path because mutations at D1-Asp-61, D2-Lys-317, D1-Asn-181, and the D1-Glu-65/D2-Glu-312/D1-Arg-334 triad substantially lowered the efficiency of this transition or significantly retarded the O 2 evolution kinetics.
A tight hydrogen-bond network is advantageous to rapid proton transfer using the Grotthuss mechanism rather than a network of mobile water molecules. Previous MD simulation by Ogata et al. (36) showed that water molecules in the Y Z -Asn-298 path ("path 1" in Ref. 36) showed only small fluctuations. In addition, Sakashita et al. (39) recently showed by MD simulation of water accessibility that water molecules near D1-Asn-298 are less exchangeable in contrast to exchangeable water molecules in the Y Z -O1 path and the Cl path ("O1-water chain" and "Glu-65/Glu-312 channel," respectively, in Ref. 39), and suggested the involvement of the hydrogen-bond network near D1-Asn-298 in proton transfer. In the D1-N298A mutant, the tight hydrogen-bond network near D1-Asn-298 should be disrupted and may be replaced by a loose water network, which would retard the proton transfer from the catalytic site. Although this mutation may also loosen the structure of a water cluster near O1 in the Y Z -O1 path, the effect would be minor because water molecules in this channel are originally mobile as shown in previous MD simulations (33,35,36,39). In contrast to proton transfer, channels involving mobile water molecules are advantageous to water transfer. Although insertion of substrate water to the catalytic site has been suggested to occur in the S 2 3 S 3 and S 3 3 S 0 transitions (40, 42, 43, 66, 76, 90 -93), the loosened water network possibly produced by the D1-N298A mutation would not inhibit water insertion. Thus, the N298A-induced inhibition of the S 2 3 S 3 and S 3 3 S 0 tran-

Proton transfer pathway in photosynthetic water oxidation
sitions observed in the present study is more likely caused by the impairment of proton transfer rather than water transfer. One possible mechanism of proton transfer through the Y Z site has been proposed from FTIR and QM/MM analyses (16,70). QM/MM calculations showed that upon Y Z oxidation, the network of nearby water molecules is rearranged and one water molecule moves toward D1-His-190. The proton that was shifted from Y Z to the N of D1-His-190 has a large polarizability as revealed by a broad feature at ϳ2800 cm Ϫ1 in a Y Z ⅐ /Y Z FTIR spectrum (16). It was thus proposed that this proton between Y Z ⅐ and HisH ϩ may have a chance to hop to the nearby water, probably using NH bending vibration, followed by immediate proton transfer through a water chain supported by D1-Asn-298. An alternative proposal was proton transfer from W4, a ligand to Ca 2ϩ , or a nearby water to the water chain near D1-Asn-298. The transferred proton from D1-His-190 or W4 will be immediately replenished by a proton from a substrate water through a water network between Y Z and Mn4 (Fig. 1A) using coupled OH stretching vibrations extended to several water molecules (87). Note that this proton transfer mechanism is different from the previous so-called hydrogen abstraction model (94), in which the proton of Y Z is first released to the bulk upon its oxidization. The driving force of the above proton transfer process is charge repulsion between the protonated cation of D1-His-190 and the excess positive charge on the Mn 4 CaO 5 cluster in the S 2 and S 3 states. Also, proton release needs to take place before electron transfer in the S 2 3 S 3 and S 3 3 S 0 transitions to decrease the redox potential of the Mn 4 CaO 5 cluster (42). Thus, this proton transfer mechanism through the Y Z site can be functional in the S 2 3 S 3 transition and for the first proton in the S 3 3 S 0 transition. Indeed, proton transfer as a lag phase (ϳ30 s) before electron transfer (42) or as a rate-limiting step (ϳ350 s) coupled to electron transfer (43) have been proposed in the S 2 3 S 3 transition, whereas a lag phase (ϳ200 s) before the electron transfer/O 2 evolution phase in the S 3 3 S 0 transfer has been attributed to a proton transfer process (58,(95)(96)(97). These putative proton transfer processes during the S 2 3 S 3 and S 3 3 S 0 transitions take place in the similar time regime of tens or hundreds of microseconds (42,43,58,(95)(96)(97), suggesting that they are performed by a similar proton transfer mechanism. However, the extent of inhibition by the D1-Asn-298 mutation was different between these transitions: the S 2 3 S 3 transition is only partially inhibited, whereas the S 3 3 S 0 transition is more significantly blocked. This difference could be related to different hydrogenbonded structures around the Mn 4 CaO 5 cluster between the S 2 -and S 3 -states and their changes upon mutation, which are revealed by a broad positive feature in the strongly hydrogenbonded OH region at 2800 -2400 cm Ϫ1 in the S 3 /S 2 FTIR difference spectrum (57,77) and a change in the mutant such as the appearance of a positive feature at 2907 cm Ϫ1 (Fig. 6B, b).
In conclusion, spectroscopic analyses of the D1-N298A mutant suggest that the hydrogen-bond network near Y Z is functional in proton transfer during water oxidation, especially in the S 2 3 S 3 and S 3 3 S 0 transitions. D1-Asn-298 plays a pivotal role in forming a tight hydrogen-bond network near Y Z , which is crucial for efficient proton transfer. At the present stage, however, possibilities other than proton transfer as an inhibited process by the mutation cannot be fully excluded. Further studies using this mutant as well as other site-directed mutants of residues on putative proton pathways, in combination with spectroscopic analyses such as time-resolved infrared spectroscopy, are necessary for identifying proton release pathways for four individual protons in the S-state cycle. Such studies of proton transfer are crucial for full understanding of the mechanism of photosynthetic water oxidation.

Construction of site-directed mutant
The wild-type control strain of the D1 subunit (WT*) and the D1-N298A strain were constructed in Synechocystis sp. PCC 6803 with an analogous method of Nagao et al. (71). Plasmid pRN123, which involved the coding region of psbA2, was used as a parental vector for site-directed mutagenesis. A host Synechocystis strain, which was transformed with pRN123 to obtain the WT* strain, lacked all of the three psbA genes (⌬psbA1/⌬psbA2/⌬psbA3) and contains a His 6 tag attached to the C terminus of CP47 (71). Mutation of D1-Asn-298 to Ala was introduced into pRN123 by replacing an AAC codon at a target site with a GCC codon, and the resultant plasmid was introduced into the host ⌬psbA1/⌬psbA2/⌬psbA3 strain.
The WT* and D1-N298A strains were maintained on BG-11 (98) agar plates containing 5 g ml Ϫ1 of kanamycin (Km), 5 g ml Ϫ1 of chloramphenicol (Cm), 5 g ml Ϫ1 of erythromycin (Em), and 5 g ml Ϫ1 of spectinomycin (Sm) in the presence of 5 mM glucose and 10 M 3-(3,4)-dichlorophenyl-1,1-dimethylurea (DCMU) under a continuous low-light condition. The genotype of the D1-N298A mutant was confirmed by PCR analysis and DNA sequencing in both cases of in vivo analyses and preparation of PSII core complexes for FTIR measurements. No trace of the wild-type psbA2 gene was detected in any cultures of the D1-N298A strain.

Cell growth
WT* and D1-N298A cells were grown photoautotrophically in 40 ml of a BG-11 medium (98) supplemented with 4 mM Hepes-NaOH (pH 7.5) and 5 g ml Ϫ1 Km/Cm/Em/Sm by bubbling with air containing 1% (v/v) CO 2 at 30°C under continuous illumination (20 mol of photons m Ϫ2 s Ϫ1 ) by white fluorescence lamps. The cell density was monitored as an optical density at 730 nm (A 730 ) using a spectrophotometer (Shimadzu UV-3100PC). For in vivo analyses, cells grown to the concentration of A 730 ϭ 0.5-1.0 were inoculated into 40 ml of a fresh BG-11 medium without antibiotics (A 730 ϭ ϳ0.1), and were cultured under the same condition. Log-phase cells were collected by centrifugation (1000 ϫ g for 5 min at 25°C) and suspended in a BG-11 medium. For PSII preparation, cells were grown in an 8-liter culture bottle without antibiotics under the photoautotrophic growth condition. 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
O 2 -evolving PSII core complexes were purified using the method by Nagao et al. (71) with a minor modification. Thyla-Proton transfer pathway in photosynthetic water oxidation koid membranes suspended in 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) were solubilized with 1% (w/v) n-dodecyl ␤-D-maltoside (DM) at a Chl concentration of 1.0 mg ml Ϫ1 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 followed by further washing with buffer B. PSII complexes were eluted with buffer B containing 50 mM L-histidine and then concentrated by ultrafiltration (Vivaspin 20, 100 kDa MWCO, Sartorius Stedim). The isolated PSII complexes were stored in liquid nitrogen.

Measurement of O 2 evolution activity
The O 2 evolution activity was measured using a Clark-type oxygen electrode at 30°C under a saturating light condition. The activities of cell samples (10 g of Chl) were measured in BG-11 medium in the presence of 1 mM DCBQ and 1 mM potassium ferricyanide as electron acceptors, whereas those of PSII complexes (3.5-5 g of Chl) were measured in a buffer (pH 6.0) containing 1 M sucrose, 5 mM CaCl 2 , 10 mM NaCl, and 50 mM Mes-NaOH in the presence of 0.1 mM DCBQ and 4 mM potassium ferricyanide as electron acceptors.

TL and DL measurements
TL and DL were measured using a laboratory-built apparatus, as described previously (99). Before measurements, cells suspended in BG-11 medium (250 g of Chl ml Ϫ1 ) were exposed to white continuous light (200 mol of photons m Ϫ2 s Ϫ1 ; ϳ16 milliwatt cm Ϫ2 at the sample point) for 30 s at 30°C, followed by incubation at this temperature for 5 min in the dark. A cell suspension (70 l) was then loaded onto a piece of filter paper. For detection of a TL B band (S 2 Q B Ϫ recombination), the sample was illuminated by a single saturating flash from a Xe lamp (SL-230S, Sugawara) at 5°C, whereas for a TL Q band (S 2 Q A Ϫ recombination), the sample in the presence of 50 M DCMU was illuminated with continuous white light (ϳ55 milliwatt cm Ϫ2 at the sample point) from a halogen lamp (MEJIRO PRECISION PHL-150) for 10 s at Ϫ20°C. The illuminated sample was quickly cooled down and then warmed at a rate of 40°C min Ϫ1 to record a TL glow curve.
For DL measurement, the sample was illuminated by a series of saturating Xe flashes (1 Hz) at 25°C, and DL emission upon each flash was recorded. The DL amplitude at 0.8 s after each flash was plotted against a flush number. The measurement was performed once for each sample, and the data obtained using three different samples were averaged.

FTIR measurements
Light-induced FTIR difference spectra were recorded using a Bruker IFS-66/S spectrophotometer equipped with an MCT detector (InfraRed D313-L) at 4 cm Ϫ1 resolution. Flash illumination was performed by a Q-switched Nd:YAG laser (Quanta-Ray GCR-130; 532 nm, ϳ7 ns full width at half-maximum) with a power of ϳ7 mJ pulse Ϫ1 cm Ϫ2 at a sample point.
FTIR measurements of the S-state transitions in PSII core complexes were performed following the previous methods (26,74,76,77). PSII complexes were washed with a buffer (pH 6.0) containing 10 mM Mes-NaOH, 5 mM NaCl, 5 mM CaCl 2 , 40 mM sucrose, and 0.06% DM (buffer C) and concentrated to ϳ2.5 mg of Chl ml Ϫ1 by ultrafiltration (Apollo 7-ml High-Performance Centrifugal Concentrators, 150 kDa MWCO, Orbital Biosciences). Two sample types, solutions and moderately hydrated films, were used for measurements. For solution samples, 10 l of the PSII solution (ϳ2.5 mg of Chl ml Ϫ1 ) in buffer C mixed with 1 l of 100 mM potassium ferricyanide (total volume: 11 l) was lightly dried on a CaF 2 plate (25 mm in diameter) under N 2 gas flow. The resultant sample was then mixed with 1 l of Milli-Q water, and sandwiched with another CaF 2 plate with a circular groove (14-mm inner diameter; 1-mm width). The sample cell was sealed with silicone grease in the outer part of the groove, where a tiny piece of aluminum foil was placed as a spacer. For preparation of a hydrated film, 4.5 l of PSII solution (ϳ2.5 mg of Chl ml Ϫ1 ) in buffer C mixed with additional 5.5 l of buffer C and 1 l of 100 mM potassium ferricyanide (total volume: 11 l) was dried on a CaF 2 plate (25 ϫ 25 mm) in a circle shape (8 mm in diameter) under N 2 gas flow. The resultant sample was sealed using another CaF 2 plate and a silicone spacer (0.5 mm in thickness), enclosing 2 l of 40% (v/v) glycerol solution (95% relative humidity) without touching the sample to moderately hydrate the film (76). For both types of samples, the sample temperature was kept at 10°C by circulating cold water through a copper holder.
In the measurement of a moderately hydrated film, two preflashes with a 1-s interval were first applied to the sample followed by dark adaptation for 15 min to synchronize all centers to the S 1 state. Four flashes were then applied with intervals of 10 s; a single-beam spectrum with 20 scans (10-s scan) was measured twice before the first flash and once after individual flashes. The whole measurement scheme was repeated 12 times, and the spectra were averaged using five different samples (total 1200 scans). The measurement of a solution sample was performed using the same scheme, but with the dark interval of 10 min. The measurement was repeated 24 times using one sample, and the spectra were averaged (total 480 scans). To obtain better signal-to-noise ratios in the 3700 -2200 cm Ϫ1 region of the spectra in the S 1 3 S 2 and S 2 3 S 3 transitions, measurements by two flashes were performed using the hydrated films. In this case, preflashes except for the first ones were omitted because the S 2 -and S 3 -states returned back to the S 1 -state during dark adaptation for 15 min. Other conditions were the same as the four-flash measurements of the hydrated films. The measurement was repeated 120 times, and the spectra using four samples were averaged (total 8400 scans). These average spectra were used to calculate flash-induced spectra of the S-state transitions together with a dark-minus-dark spectrum representing a noise level. Y Z ⅐ /Y Z FTIR difference spectra were measured following the method described previously (16) with a slight modification. Mn depletion was performed by 10 mM NH 2 OH treatment for 1 h on ice, followed by washing four times with a buffer (pH 6.5) containing 20 mM Mes-NaOH, 5 mM NaCl, and 0.06% DM using ultrafiltration (Vivaspin 500, 100 kDa MWCO, Sartorius Stedim), which finally concentrated the sample to 2.5 mg of Chl ml Ϫ1 . An aliquot (8 l) of the sample solution, which was mixed Proton transfer pathway in photosynthetic water oxidation with 1 l of 100 mM potassium ferricyanide, was dried on a BaF 2 plate (13 mm in diameter) under a N 2 gas flow, and then sandwiched with another BaF 2 plate together with 0.85 l of Milli-Q water. The sample temperature was adjusted to 250 K in a cryostat (Oxford DN1704). Single-beam spectra with 50 scans (25-s scan) were recorded twice before and once after a single flash. This measurement was repeated 90 and 160 times for WT* and D1-N298A, respectively, with a dark interval of 225 s, and the average spectra were used to calculate a Y Z ⅐ /Y Z spectrum together with a dark-minus-dark spectrum representing a noise level.