Substitution of the D1-Asn87 site in photosystem II of cyanobacteria mimics the chloride-binding characteristics of spinach photosystem II

Photoinduced water oxidation at the O2-evolving complex (OEC) of photosystem II (PSII) is a complex process involving a tetramanganese-calcium cluster that is surrounded by a hydrogen-bonded network of water molecules, chloride ions, and amino acid residues. Although the structure of the OEC has remained conserved over eons of evolution, significant differences in the chloride-binding characteristics exist between cyanobacteria and higher plants. An analysis of amino acid residues in and around the OEC has identified residue 87 in the D1 subunit as the only significant difference between PSII in cyanobacteria and higher plants. We substituted the D1-Asn87 residue in the cyanobacterium Synechocystis sp. PCC 6803 (wildtype) with alanine, present in higher plants, or with aspartic acid. We studied PSII core complexes purified from D1-N87A and D1-N87D variant strains to probe the function of the D1-Asn87 residue in the water-oxidation mechanism. EPR spectra of the S2 state and flash-induced FTIR spectra of both D1-N87A and D1-N87D PSII core complexes exhibited characteristics similar to those of wildtype Synechocystis PSII core complexes. However, flash-induced O2-evolution studies revealed a decreased cycling efficiency of the D1-N87D variant, whereas the cycling efficiency of the D1-N87A PSII variant was similar to that of wildtype PSII. Steady-state O2-evolution activity assays revealed that substitution of the D1 residue at position 87 with alanine perturbs the chloride-binding site in the proton-exit channel. These findings provide new insight into the role of the D1-Asn87 site in the water-oxidation mechanism and explain the difference in the chloride-binding properties of cyanobacterial and higher-plant PSII.

stabilizing one of the two isomers. Also, it has been observed that depletion of Cl Ϫ favors the S ϭ 5/2 spin isomer in the S 2 state of spinach PSII (10,11). This observation, however, has not been seen in cyanobacterial PSII indicating that the Cl Ϫbinding site and its associated hydrogen-bonding network affects the spin isomer distribution differently in the two species (12). Mutations of second-shell residues in cyanobacterial PSII such as D2-K317R, D1-N181A, and D1-N181S have yielded the high-spin, S ϭ 5/2, form of the S 2 state, highlighting the importance of these residues in the spin isomer equilibrium distribution (3,13). Therefore, it is of interest to focus on changes in the second-shell residues that may impact the hydrogen-bonding network.
Sequence alignment studies of the second-shell residues have revealed that the only statistically significant difference between the two species occurs at D1 position 87 (2). This site is occupied by an asparagine residue in a significant majority of cyanobacterial PSII including Synechocystis PCC 6803 as opposed to an alanine residue in spinach PSII. Furthermore, this residue is an important part of the "narrow" channel of hydrogen-bonded waters whose role is yet to be determined (2). The hydrogen-bonding network surrounding the manganese cluster is extensive; therefore, perturbation at any point might be propagated over a long distance and can affect remote sites like the Cl Ϫ -binding site which is ϳ13 Å from D1 residue 87. Fig. 1 shows the extensive hydrogen-bonding network connected to D1 residue 87 in cyanobacterial PSII obtained from Thermosynechococcus elongatus (Protein Data Bank code 3WU2). Based on these observations it is therefore hypothesized that substitution of D1 residue 87 in cyanobacterial PSII would be able to effect some of the changes that are different in the two species. The current study deals with characterization of substitutions of D1 residue 87, specifically D1-N87A and D1-N87D. In addition to understanding the role of this site, our study also has uncovered some of the fundamental aspects that are different between the two species and hence provides new insight into the mechanism of water oxidation.

Growth and steady-state oxygen assay
To investigate the significance of the difference in D1 residue 87 between spinach PSII and cyanobacterial PSII, the D1-N87A and D1-N87D mutations were constructed in Synechocystis sp. PCC 6803. D1-N87A cells and D1-N87D cells exhibit photoau-totrophic growth. The doubling times of D1-N87A and D1-N87D cells were 26 and 21 h, respectively, compared with 14 h for wildtype cells (Fig. S1). The light-saturated O 2 -evolution rates of D1-N87A and D1-N87D cells were 64 Ϯ 4 and 36 Ϯ 2% compared with the rate of wildtype cells, respectively (Table  S1 and Note S1). On the basis of the maximum fluorescence yields (F max Ϫ F 0 ) of wildtype, D1-N87A, and D1-N87D cells, the PSII contents of D1-N87A and D1-N87D cells were estimated to be 73 Ϯ 5 and 36 Ϯ 4% compared with wildtype cells, respectively (Fig. S2, Table S1, and Note S1). On the basis of measurements of the kinetics of charge recombination between Q A . and the donor side of PSII following 5 s of actinic illumination in the presence of 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU), it was estimated that 10 -14% of PSII reaction centers in D1-N87A and D1-N87D cells lack Mn 4 CaO 5 clusters in vivo ( Fig. S3 and Note S2). The presence of PSII reaction centers lacking Mn 4 CaO 5 clusters implies that the Mn 4 CaO 5 cluster is assembled less efficiently or is less stable in D1-N87A and D1-N87D cells compared with wildtype.
The light-saturated rates of O 2 evolution in PSII core complexes isolated from D1-N87D and D1-N87A cells were 886 Ϯ 28 and 3100 Ϯ 56 mol of O 2 (mg of Chl) Ϫ1 h Ϫ1 , respectively, compared with 3400 Ϯ 83 mol of O 2 (mg of Chl) Ϫ1 h Ϫ1 for wildtype PSII core complexes under similar assay conditions using a buffer containing 60 mM Cl Ϫ ( Table 1). The lower PSII activity in the D1-N87D core complex can be due to inefficient assembly of the OEC, impaired S-state cycling or due to inactivation during biochemical isolation of PSII core complexes.

Cl ؊ dependence of steady-state oxygen evolution
To study the effect of Cl Ϫ on the steady-state oxygen evolution of D1-N87A and D1-N87D PSII, the PSII core complexes were suspended in a buffer containing 0.1 mM Cl Ϫ . The lightsaturated activity of D1-N87D and D1-N87A PSII under this condition is given in Table 1. Under low Cl Ϫ conditions, the oxygen-evolution activity of D1-N87D PSII decreased by 16% relative to Cl Ϫ -sufficient conditions. The Cl Ϫ dependence of D1-N87D PSII is comparable with that of wildtype Synechocystis PSII core complexes, which exhibit a decrease in activity of only 12% under similar experimental conditions. However, the Cl Ϫ dependence of D1-N87A PSII is markedly different from wildtype Synechocystis PSII, with a 56% decrease in the oxygenevolution activity in low Cl Ϫ conditions relative to Cl Ϫ -sufficient conditions; notably, this decrease is comparable with the 52% decrease observed for spinach PSII membranes (Table 1). These results demonstrate that the D1-N87A mutation modulates the Cl Ϫ -binding site, causing the Cl Ϫ dependence of PSII

D1-N87A modulates chloride binding in photosystem II
activity to be similar to spinach PSII. It is interesting, however, that the D1-N87D substitution has little effect on Cl Ϫ binding. Based on these results, we investigated the effect of these single-point mutations at D1 residue 87 on the binding affinity of chloride. The Cl Ϫ concentrations were varied from 0.1 to 5 mM, and their effect on PSII activity was measured. The observation of biphasic binding curves (Fig. 2) suggests the presence of at least two types of Cl Ϫ -binding sites: one in which the bound chloride is exchangeable with free chloride (which titrates with a binding constant of K D ) and another non-exchangeable site (which gives activity in the absence of added chloride). The binding curves were computed by fitting the data to a biphasic Michaelis-Menten equation (Equation 1), where V obs is the observed rate of oxygen evolution, V max is the maximum observed activity which is normalized to 1, f is the fraction of centers with non-exchangeable Cl Ϫ , and K D is the binding constant of the exchangeable Cl Ϫ . The measured K D values for Cl Ϫ in D1-N87A and D1-N87D PSII are 0.730 Ϯ 0.218 mM (f ϭ 0.4) and 1.35 Ϯ 0.956 mM (f ϭ 0.8), respectively, which fall within the range of K D values obtained for Cl Ϫ in spinach PSII (0.5-0.7 mM) (14 -16). These results indicate that the binding environment of the exchangeable Cl Ϫ in the two D1-Asn 87 -substituted PSII samples is similar to that in spinach PSII. However, the fraction of exchangeable Cl Ϫ is much larger in D1-N87A Synechocystis PSII and spinach PSII than in wildtype Synechocystis PSII and D1-N87D Synechocystis PSII.

Electron paramagnetic resonance spectroscopy
EPR experiments recording the S 1 -and S 2 -state spectra of spinach PSII membranes, and Synechocystis wildtype, D1-N87A and D1-N87D PSII core complexes were performed (S 2 -minus-S 1 difference spectra shown in Fig. 3, unsubtracted spectra are included in Fig. S4, EPR spectra of spinach PSII membranes are included in Fig. S9). EPR spectra of D1-N87A were also collected after changing the cryoprotectant to sucrose. The spectrum in the presence of sucrose is similar to the spectrum in 10% (v/v) glycerol as cyoprotectant (Fig. S10). The dark spectra corresponding to the S 1 state exhibit a cytochrome c 550 signal. Upon 200 K illumination, the OEC advances to the S 2 state, which is characterized by the presence of a g ϭ 2 multiline EPR signal. The multiline signal was observed in wildtype, D1-N87A, and D1-N87D PSII (Fig. 3). However, the S ϭ 5/2 spin isomer giving the S 2 -state g ϭ 4.1 signal that is observed under these conditions in spinach PSII membranes was absent in both D1-N87D and D1-N87A PSII. In addition, the S 2 -minus-S 1 difference spectra show a lightinduced cytochrome b 559 signal at g ϭ 3. The fraction of PSII

D1-N87A modulates chloride binding in photosystem II
centers lacking a functional OEC gives rise to this signal when the sample is illuminated at 200 K. The intensities of g ϭ 2 multiline signal of D1-N87D and D1-N87A PSII are smaller relative to wildtype PSII for a similar chlorophyll concentration. Both the weaker g ϭ 2 multiline signal and the 200 K lightinduced cytochrome b 559 signal indicate that the D1-N87D and D1-N87A PSII samples have fewer functional OECs, consistent with the lower oxygen-evolution activity observed for the EPR samples.

FTIR spectroscopy mid-frequency region
The mid-frequency FTIR difference spectra induced by four successive flashes given to wildtype and D1-N87D PSII core complexes are compared in Fig. 4 (black and red traces, respectively). The spectra that are induced by the first, second, third, and fourth flashes correspond predominantly to the S 2 -minus-S 1 , S 3 -minus-S 2 , S 0 -minus-S 3 , and S 1 -minus-S 0 FTIR difference spectra, respectively (17)(18)(19)(20).

D1-N87A modulates chloride binding in photosystem II
The mid-frequency S 2 -minus-S 1 spectrum of D1-N87D PSII core complexes showed substantial changes compared with wildtype PSII throughout the mid-frequency region (upper red trace in Fig. 4). In the carbonyl stretching [(CϭO)] region, the negative feature at 1746 cm Ϫ1 was broadened slightly and shifted ϳ2 cm Ϫ1 to higher frequency. In the amide I region, the negative features at 1689 and 1680 cm Ϫ1 were replaced by positive features at 1684 and 1670 cm Ϫ1 , and the negative feature at 1629 cm Ϫ1 was replaced by a negative feature at 1641 cm Ϫ1 . These features correspond to amide I modes because they shift significantly to lower frequencies after global incorporation of 13 C (21, 22) but not appreciably after global incorporation of 15 N (21, 22). In the overlapping asymmetric carboxylate stretching [ asym (COO Ϫ )]/amide II region, the positive feature at 1586 cm Ϫ1 was shifted ϳ3 cm Ϫ1 to higher frequency and diminished in amplitude, the 1560(Ϫ), 1551(ϩ), and 1544(Ϫ) cm Ϫ1 features shifted downward, and the 1523(Ϫ) cm Ϫ1 feature was diminished. The 1587 cm Ϫ1 feature corresponds to a asym (COO Ϫ ) mode because it shifts 30 -35 cm Ϫ1 to lower frequency after global incorporation of 13 C (21-24) but is largely insensitive to the global incorporation of 15 N (3, 21-23, 25). The 1560(Ϫ), 1551(ϩ), and 1544(Ϫ) features correspond to amide II modes because they shift appreciably to lower frequencies after global incorporation of either 13 C (21-24) or 15 N (3,(21)(22)(23)25). In the symmetric carboxylate stretching [ sym (COO Ϫ )] region, the positive shoulder at 1432 cm Ϫ1 was eliminated; features at 1415(Ϫ), 1410(ϩ), and 1400(Ϫ) cm Ϫ1 were shifted ϳ2 cm Ϫ1 to higher frequencies; features at 1354(Ϫ) and 1343(ϩ) cm Ϫ1 were shifted 4 -5 cm Ϫ1 to lower frequencies; and a negative feature at 1384 cm Ϫ1 appeared.
The mid-frequency S 2 -minus-S 1 spectrum of D1-N87A PSII core complexes (Fig. S6, upper red trace) showed changes similar to those observed for D1-N87D PSII core complexes. The mid-frequency S 3 -minus-S 2 , S 0 -minus-S 3 , and S 1 -minus-S 0 difference spectra of D1-N87D PSII core complexes (Fig. 4, middle three pairs of spectra) showed changes in the amide I and overlapping asym (COO Ϫ )/amide II regions but fewer changes in sym (COO Ϫ ) regions than in the S 2 -minus-S 1 spectrum. In the S 3 -minus-S 2 spectrum, the positive feature at 1746 cm Ϫ1 shifted to 1743 cm Ϫ1 , the positive feature at 1666 cm Ϫ1 was eliminated, a negative feature at 1651 cm Ϫ1 appeared, the amplitudes of negative features near 1622 and 1612 cm Ϫ1 increased, the amplitude of the negative feature at 1568 cm Ϫ1 decreased, new derivative-shaped features appeared at 1554(ϩ)/1544(Ϫ) and 1546(Ϫ)/1506(ϩ) cm Ϫ1 , and the negative feature at 1363 cm Ϫ1 shifted to 1356 cm Ϫ1 and decreased in amplitude. In the S 0 -minus-S 3 spectrum, the positive feature at 1746 cm Ϫ1 shifted to 1748 cm Ϫ1 , the positive features at 1723 and 1709 cm Ϫ1 were eliminated, the derivative-shaped feature at 1650(Ϫ)/1641(ϩ) cm Ϫ1 shifted to 1645(Ϫ)/1635(ϩ) cm Ϫ1 , the negative feature at 1565 cm Ϫ1 was eliminated, and the derivative-shaped feature at 1358(ϩ)/1344(Ϫ) cm Ϫ1 shifted to 1355(ϩ)/1340(Ϫ) cm Ϫ1 . In the S 1 -minus-S 0 spectrum, the derivative-shaped feature at 1753(ϩ)/1745(Ϫ) cm Ϫ1 shifted to 1751(ϩ)/1743(Ϫ) cm Ϫ1 , the positive feature at 1715 cm Ϫ1 was replaced by a negative feature at 1700 cm Ϫ1 , the positive feature at 1687 cm Ϫ1 was replaced by a negative feature at 1683 cm Ϫ1 , the negative feature at 1586 cm Ϫ1 was eliminated, the positive feature at 1568 cm Ϫ1 was shifted to 1572 cm Ϫ1 , a negative feature appeared at 1507 cm Ϫ1 , the positive feature at 1421 cm Ϫ1 shifted to 1428 cm Ϫ1 , and the negative feature at 1383 cm Ϫ1 was eliminated. The mid-frequency S 3 -minus-S 2 , S 0 -minus-S 3 , and S 1 -minus-S 0 difference spectra of D1-N87A PSII core complexes (Fig. S6, middle three pairs of spectra) showed changes similar to those observed for D1-N87D PSII core complexes.

Strongly H-bonded O-H stretching region
The O-H stretching vibrations of strongly H-bonded OH groups can be observed as very broad positive features between 3200 and 2500 cm Ϫ1 (13, 26 -29). These regions of the S nminus-S nϩ1 difference spectra of D1-N87D PSII core complexes are compared with wildtype PSII core complexes in Fig.  5. In the S 2 -minus-S 1 spectrum (Fig. 5, upper left panel), the broad feature is overlain with numerous positive features that have been attributed to a mixture of C-H stretching vibrations from aliphatic groups and N-H stretching vibrations and their Fermi resonance overtones from the imidazole group(s) of one or more histidine residues (29 -31).
The D1-N87D mutation caused no substantial alterations to the broad features of any of the S n -minus-S nϩ1 spectra. The same is true for the D1-N87A mutation (Fig. S7).

Polarographic O 2 measurements
To probe the S-state cycling, flash induced polarographic O 2 -yield measurements were taken for PSII core complexes isolated from His-tagged wildtype, D1-N87A, and D1-N87D strains. The O 2 -yield patterns are shown in Fig. 7.
The high intensity of the signal corresponding to the first flash in D1-N87D-mutated PSII may be attributed to the presence of damaged centers. The four-periodic oscillation pattern was fit to a modified Kok model (42). The miss and double-hit factors thereby computed are reported in Table 2. The miss factor for D1-N87A PSII core complexes is 11%, and that for D1-N87D core complexes is 29% compared with 14% for Histagged wildtype PSII core complexes and 10% for spinach PSII (43). The percentage of double hits was small and did not vary significantly among the three types of PSII core-complexes. The oxygen-release kinetics for D1-N87D PSII core complexes was found to be similar to that of His-tagged wildtype PSII core complexes (Fig. S5).

Discussion
Both D1-N87A and D1-N87D Synechocystis sp. PCC 6803 cells exhibit photoautotrophic growth. The S-state cycling efficiency of the OEC can further be quantified by the miss parameter (␣), which denotes the probability that the OEC fails to advance its oxidation state following a photochemical charge separation. The D1-N87A and wildtype PSII core complexes exhibit similar miss factors supporting their efficient S-state cycling. However, the miss factor for the D1-N87D PSII cores is twice that of the wildtype PSII cores. Further, the higher intensity of the first-flash bare-platinum electrode signal from D1-N87D PSII core complexes is indicative of damaged centers. Hence, the effect of D1-N87D mutation can be attributed to inefficient turnover or impairment in the stability of the OEC caused by perturbations in the hydrogen-bonding network. The observation that the D1-N87D, but not the D1-N87A, substitution impairs the S-state cycling illustrates that an Ala or Asn residue at D1 residue 87 is needed for efficient S-state cycling. Fig. 1 shows the hydrogen-bonded network of water and amino acids around the D1-Asn 87 residue in cyanobacterial PSII. We hypothesized that the change from D1-Asn 87 in cyanobacteria to D1-A87 in spinach would perturb the hydrogenbonded network around Water 1 (W1) and Water 2 (W2) bound to Mn4 in the OEC, thereby biasing the equilibrium distribution in the S 2 state toward the (S ϭ 5/2) g ϭ 4.1 EPR signal conformation. Hence, it was expected that, unlike wildtype cyanobacterial PSII, the S 2 -state EPR spectra of D1-N87A PSII would exhibit an S 2 -state g ϭ 4.1 EPR signal resembling the S 2 -state EPR spectrum of spinach PSII. However, the S 2 -state EPR spectra of both D1-N87A and the D1-N87D PSII core complexes are similar to wildtype PSII, i.e. neither exhibits an S 2 -state g ϭ 4.1 EPR signal. These results may be reflective of subtle differences in the second-shell amino acid residues and hydrogen-bonded water molecules between spinach and cyanobacterial PSII that are not reflected in the primary amino acid sequences, which may favor the presence of the g ϭ 4.1 EPR signal conformation in the S 2 state for spinach PSII. Further, differences in the extrinsic subunits between the two species may also contribute to the greater stability of the S ϭ 5/2 conformation in spinach PSII relative to cyanobacterial PSII.
We next investigated whether D1-Asn 87 -substituted PSII exhibits other properties like spinach PSII. Because it has been found that the oxygen-evolution activity of spinach PSII is more sensitive to Cl Ϫ depletion than cyanobacterial PSII, the Cl Ϫ effect on the steady-state oxygen-evolution activity of D1-N87A PSII and D1-N87D PSII was studied. Cl Ϫ was depleted from the PSII core complexes by diffusion into a Cl Ϫ - Figure 5. Comparison of the FTIR difference spectra of wildtype (black) and D1-N87D (red) Synechocystis PSII core complexes between 3100 and 2150 cm ؊1 in response to four successive flash illuminations applied at 0°C. The data were collected simultaneously with that shown in Fig. 4. The spectra of the D1-N87D samples were multiplied vertically as in Fig. 4 and were shifted vertically to coincide approximately at 2500 cm Ϫ1 .

D1-N87A modulates chloride binding in photosystem II
deficient medium. This technique has been previously used to remove Cl Ϫ from spinach PSII; however, it has been found to be less effective for removal of Cl Ϫ from cyanobacterial PSII, which has been attributed to a higher affinity of cyanobacterial PSII for Cl Ϫ (3, 44 -46). We find that Cl Ϫ depletion affects the oxygen-evolution activity of D1-N87A PSII to a similar extent as spinach PSII. However, D1-N87D PSII exhibits a Cl Ϫ effect like that of wildtype cyanobacterial PSII (3). This can be rationalized by the fact that a non-polar amino acid residue like alanine will affect the hydrogen-bonding network surrounding the OEC to a greater extent than a polar residue like aspartic acid. Hence, we conclude that D1 residue 87 plays an important role in the network of hydrogen-bonded waters and amino acid residues such that the effect of a single point mutation at D1 residue 87 can be translated to the Cl Ϫ -binding site ϳ13 Å away, as shown in Fig. 1. Further, the finding that the only statistically significant difference in the amino acid sequences of cyanobacterial and plant PSII around the OEC occurs at D1 residue 87 (2) can be correlated with the different Cl Ϫ -binding properties of PSII from the two species.

D1-N87A modulates chloride binding in photosystem II
meets the narrow channel at D1-Asp 61 , and this channel also connects to the O4 -oxo ligand of the OEC (49). Our finding that D1-Asn 87 , a narrow-channel residue, modulates the Cl Ϫbinding site indicates that the residues in the narrow channel modulate the functioning of the broad channel. Thus, there must be an extensive hydrogen-bonding network that interconnects the two channels.
To further probe the effects of the D1-Asn 87 substitutions on the structure and hydrogen-bonding interactions around the OEC, FTIR characterization of D1-N87D and D1-N87A PSII was performed. The D1-N87D and D1-N87A substitutions alter amide and carboxylate vibrational modes throughout the mid-frequency regions of the S nϩ1 -minus-S n FTIR difference spectra. Substitutions of other residues that participate in the networks of hydrogen bonds that surround the Mn 4 CaO 5 cluster also cause numerous alterations in the mid-frequency regions of the S nϩ1 -minus-S n FTIR difference spectra. Examples are D1-E65A (50) (57), produce little or no changes to any of S nϩ1 -minus-S n FTIR difference spectra. The changes to amide I, amide II, asym (COO Ϫ ) and sym (COO Ϫ ) modes in the S nϩ1 -minus-S n spectra produced by the D1-N87D and D1-N87A substitutions are, therefore, consistent with the location of D1-Asn 87 in the network of hydrogen bonds forming a water entry channel leading to the Mn 4 CaO 5 cluster (58).
Features in the (CϭO) regions of the S nϩ1 -minus-S n FTIR difference spectra probe the extensive networks of hydrogen bonds surrounding the Mn 4 CaO 5 cluster (50,51,59). The shifts of the features in this region, e.g. between 1748 and 1743 cm Ϫ1 , produced by the D1-N87D and D1-N87A substitutions are also consistent with D1-Asn 87 participating in a network of hydrogen bonds that forms part of a water entry channel.
On the basis of quantum mechanics/molecular mechanicsbased simulations (27), the broad positive feature observed between 3200 and 2500 cm Ϫ1 in the S 2 -minus-S 1 spectrum has been assigned to the coupled O-H stretching vibrations of strongly hydrogen-bonded water molecules in a network that links D1-Asp 61 with the Ca 2ϩ ion and tyrosine Z (27). The feature is positive because the hydrogen bonds of the water molecules in this network are strengthened by the positive charge that develops on the Mn 4 CaO 5 cluster during the S 1 -to-S 2 transition (27). This feature is dominated by the O-H stretching vibrations of Water 1 (W1) and Water 2 (W2) bound to Mn4 in the OEC (27) and is eliminated by the D1-D61A mutation (26). The broad positive features in these regions of the S 3 -minus-S 2 , S 0 -minus-S 3 , and S 1 -minus-S 0 spectra have been assigned to strongly hydrogen bonded O-H stretching vibrations of water molecules in networks of hydrogen bonds that are highly polarizable (28,29). That the D1-N87D and D1-N87A mutations produce little change in these regions in any of the S nϩ1 -minus-S n difference spectra implies that D1-Asn 87 has limited influence on the strongly hydrogen-bonded water molecules that give rise to the broad features observed in these regions.
On the basis of quantum mechanics/molecular mechanicsbased simulations (27), the features between 3700 and 3500 cm Ϫ1 in the S 2 -minus-S 1 difference spectrum have been assigned to the coupled O-H stretching vibrations of weakly hydrogen-bonded water molecules in the network of hydrogen bonds that links D1-Asp 61 with the Ca 2ϩ ion and tyrosine Z (27). The features observed in the other S nϩ1 -minus-S n spectra in this region are presumed to have the same origin (29,38). Consequently, these features contain contributions from multiple water molecules. In the S 2 -minus-S 1 spectra of D1-N87D and D1-N87A PSII core complexes, the positive feature at 3619 cm Ϫ1 was eliminated. In the S 3 -minus-S 2 spectra, the intensity of the broad negative feature centered at 3600 cm Ϫ1 was diminished. These alterations are consistent with D1-Asn 87 causing slight perturbations to this network.
In summary, our study provides new evidence on the function of the amino acid residue at D1 residue 87 in spinach and Synechocystis PSII. This residue helps in efficient S-state cycling of the OEC. The D1-N87A substitution in cyanobacteria, which replicates D1 residue 87 in spinach, alters the Cl Ϫ -binding properties such that D1-N87A PSII mimics spinach PSII. This indicates that D1 residue 87 modulates the effect of Cl Ϫ on the PSII activity by perturbing the hydrogen-bonding network across the narrow and broad channels that interconnect D1 residue 87 with the Cl Ϫ -binding site.

Construction of mutant strains and extraction of PSII
The D1-N87A and D1-N87D mutations were constructed in the psbA-2 gene of Synechocystis sp. PCC 6803 and transformed into a host strain of Synechocystis that lacks all three psbA genes and contains a hexahistidine tag fused to the C terminus of CP47. The control wildtype strain used in this study was constructed in identical fashion as the mutants, but with a transforming plasmid that carried no mutation (53). Single colonies were selected for their ability to grow on solid medium containing 5 g/ml kanamycin monosulfate and 20 g/ml gentamycin sulfate (53). D1-N87A cells were grown on solid BG-11 plates in the absence of glucose. The antibiotics were omitted from liquid cultures. The cells from small-scale liquid cultures were propagated in three 7-liter carboys as described previously (26) and bubbled with 1% CO 2 in air. The cells were grown with warm white fluorescent lighting at 28°C (140 E m Ϫ2 s Ϫ1 ). The PSII extraction and purification were done under dim light conditions at 4°C using a nickel-nitrilotriacetic acid super flow affinity resin (Qiagen) as described previously (60). The purified PSII core complexes were concentrated to ϳ1 mg of Chl/ml and stored in a buffer solution containing 1.2 M betaine, 10% (v/v) glycerol, 50 mM MES-NaOH (pH 6.0), 20 mM CaCl 2 , 5 mM MgCl 2 , 50 mM histidine, 1 mM EDTA, and 0.03% (w/v) n-dodecyl ␤-D-maltoside at Ϫ80°C. Polymerase chain reaction amplification and sequencing of genomic DNA in the region of the psbA gene that contains the D1-Asn 87 codon were performed on aliquots of each culture that were harvested to verify the integrity of the mutant cultures (53). No trace of the wildtype codon was detected in any of the mutant cultures. Spinach PSII membranes were prepared as described previously (61,62) and suspended to D1-N87A modulates chloride binding in photosystem II final chlorophyll concentrations of 5.0 -7.0 mg ml Ϫ1 in 20 mM MES-NaOH (pH 6.0), 15 mM NaCl, and 30% (w/v) ethylene glycol.

Growth rate measurements
For measuring growth rates, wildtype, D1-N87A, and D1-N87D cells were grown photosynthetically in 100 ml of liquid BG-11 medium supplemented in the absence of glucose at 28°C with a light intensity of 140 E m Ϫ2 s Ϫ1 . Carbon dioxide was not bubbled at this stage. Growth was recorded by measuring the absorbance of the cultures at 730 nm using a Beckman DU-640 UV-visible spectrophotometer.

Analysis of amino acid residues
D1 sequences from Cyanobase and NCBI were compiled. For higher plants, 11 species were analyzed, all of which have Ala 87 . For green algae, 59 species were analyzed of which 58 have Asn 87 and 1 has Ala 87 . For cyanobacteria, 94 sequences (some cyanobacteria have more than one D1 isoform) were analyzed, of which 81 have Asn 87 and 13 have Ala 87 .

Steady-state oxygen assay
PSII isolated from spinach, and wildtype, D1-N87A, and D1-N87D Synechocystis cells were stirred in low-chloride buffer ([Cl Ϫ ] ϭ 0.1 mM) in the dark for 2 min prior to each oxygenevolution assay. The Cl Ϫ -dependent assays were performed by adding an appropriate volume of Cl Ϫ from a stock solution in a buffer containing 1 M sucrose, 5 mM Ca 2ϩ [using Ca(OH) 2 ], and MES (pH 6.5). Oxygen evolution was monitored with a Clark-type electrode, and the oxygen-assay chamber was maintained at 25°C using a temperature-controlled water bath. The samples were illuminated with an Oriel 1000 W tungsten-halogen lamp. 250 M 2-phenyl-1,4-benzoquinone with 1 mM K 3 FeCN 6 was used as the electron acceptor, and 5 g of Chl was used in each assay. The oxygen evolution of whole cells was carried out using a Clark electrode in the presence of 2 mM 2,6-dichloro-p-benzoquinone with 2 mM K 3 FeCN 6 as electron acceptors (50).

EPR measurements
EPR samples of WT and D1-Asn 87 -substituted Synechocystis PSII were prepared in 50 mM MES-NaOH (pH 6.5), 10% (v/v) glycerol, 1.2 M betaine, 20 mM CaCl 2 , 5 mM MgCl 2 , and 1 mM EDTA and were concentrated to 1 mg of Chl/ml using Amicon centrifugal cells having a 100-kDa cutoff. EPR samples of spinach PSII were prepared as described previously (9). The dark scan corresponding to the S 1 state was recorded. Then the S 2 state was generated by illuminating the sample with a red LED (623 nm) in a 200 K acetone/dry ice bath, and the S 2 -state spectrum was recorded. The measurements were performed using a Bruker ELEXSYS E500 spectrometer equipped with a SHQ resonator and an Oxford ESR-900 continuous flow cryostat at 7.5 K. The EPR parameters used for recording the spectra are as follows: microwave frequency, 9.38 GHz; modulation frequency, 100 kHz; modulation amplitude, 19.95 G; microwave power, 5 milliwatt; sweep time, 84 s; conversion time, 41 ms; and time constant, 82 ms. Each spectrum is the average of two scans.

FTIR measurements
Purified PSII core complexes were exchanged into FTIR analysis buffer (40 mM sucrose, 10 mM MES-NaOH, pH 6.0, 5 mM CaCl 2 , 5 mM NaCl, and 0.06% (w/v) n-dodecyl ␤-D-maltoside), concentrated to ϳ3.3 mg of Chl/ml, mixed with 1/10 volume of fresh 100 mM potassium ferricyanide (dissolved in water), spread to a diameter of ϳ13 mm in the center of a 25 ϫ 2-mm diameter BaF 2 window, and dried lightly under a stream of dry nitrogen gas as described previously (26). Sample concentrations were adjusted so that the absolute absorbance of the amide I band at 1657 cm Ϫ1 was 0.7-1.1. To maintain the sample at 95% relative humidity in the FTIR sample compartment, six 1-l droplets of a solution of 40% (v/v) glycerol in water were spotted around the periphery of the window, not touching the sample (29). Sealed samples were allowed to equilibrate at 0°C in darkness in the FTIR sample compartment for 1.5 h, given 6 pre-flashes, and allowed to dark adapt for an additional 30 min (59). Mid-frequency FTIR spectra were recorded with a Bruker Vertex 70 spectrometer (Bruker Optics, Billerica, MA) as described previously (59).
After dark adaptation, six successive flashes were applied with an interval of 13 s between each. Two single-beam spectra were recorded before the first flash, and one single-beam spectrum was recorded starting 0.33 s after the first and subsequent flashes (each single-beam spectrum consisted of 100 scans). The 0.33-s delay was incorporated to allow for the oxidation of Q A . by the ferricyanide. To obtain difference spectra corresponding to successive S-state transitions, the single-beam spectrum that was recorded after the n th flash was divided by the single-beam spectrum that was recorded immediately before the n th flash, and the ratio was converted to units of absorption. To estimate the background noise level, the second pre-flash single-beam spectrum was divided by the first, and the ratio was converted to units of absorption. The sample was dark-adapted for 30 min, and then the cycle was repeated. The cycle was repeated 15 times for each sample, and the difference spectra recorded with multiple samples were averaged.

Polarographic oxygen measurements
Flash-induced O 2 yields were measured polarographically using a bare platinum electrode with a silver counter electrode poised at Ϫ700 mV versus the normal hydrogen electrode. An EG&G xenon flash lamp controlled by an Arduino Uno (Ivrea, Italy) board provided periodic flashes interspersed by a delay time of 1 s. The measurements were performed in a buffer containing 1 M sucrose, 10 mM CaCl 2 , 200 mM NaCl, and 50 mM MES-NaOH (pH 6.50) as described previously (3,43). 2,6-Dichloro-p-benzoquinone and K 3 FeCN 6 were also added to final concentrations of 500 M and 1 mM, respectively, before taking measurements. The resulting period 4 oscillations were analytically fit to the VZAD model using the BOBYQA nonlinear optimization algorithm (42).