Participation of the C-terminal Region of the D1-polypeptide in the First Steps in the Assembly of the Mn4Ca Cluster of Photosystem II*

Amino acid residue D1-Asp170 of the D1-polypeptide of photosystem II was previously shown to be implicated in the binding and oxidation of the first manganese to be assembled into the Mn4Ca cluster of the oxygen-evolving complex (OEC). According to recent x-ray crystallographic structures of photosystem II, D1-Glu333 is proposed to participate with D1-Asp170 in the coordination of Mn4 of the OEC. Other residues in the C-terminal region of the D1-polypeptide are proposed to coordinate nearby manganese of the cluster. Site-directed replacements in Synechocystis sp. PCC 6803 at D1-His332, D1-Glu333, D1-Asp342, D1-Ala344, and D1-Ser345 were examined with regard to their ability to influence the binding and oxidation of the first manganese in manganese-depleted photosystem II core complexes. Direct and indirect measurements reveal in all mutants, but most marked in D1-Glu333 replaced by His, an impaired ability of Mn2+ to reduce YZ·, indicating a reduced ability (elevated Km) compared with WT to bind and oxidize the first manganese of the OEC. The effect on the Km of these mutations is, however, considerably weaker than some of those constructed at D1-Asp170 (replacement by Asn, Ala, and Ser). These observations imply that the C-terminal residues ultimately involved in manganese coordination contribute to the high affinity binding at D1-Asp170 likely through electrostatic interactions. That these residues are far from D1-Asp170 in the primary structure of the D1-polypeptide, imply that the C terminus of the D1-polypeptide is already close to its mature conformation at the first stages of assembly of the Mn4Ca cluster.

Photosystem II (PSII) 2 of oxygenic photosynthesis is a multisubunit membrane-associated pigment-protein complex that supports light-driven oxidation of water to molecular oxygen. Recent x-ray crystallographic structures of the PSII core complex have been reported at resolutions ranging from 3.0 to 3.8 Å (1)(2)(3)(4)(5). Two of these (3)(4)(5) depict in detail the structure and coordination environment of the Mn 4 Ca cluster, responsible for water oxidation (oxygen-evolving complex or OEC). The cluster contains four manganese, whose oxidation states are thought to range from II to IV (6), depending on the state of oxidation of the OEC, and one Ca 2ϩ . The proteinaceous ligands (a total of 11-12 (5)) to the cluster are contributed by two polypeptides, D1 (PsbA) and CP43 (PsbC), with additional coordination provided by water molecules (including hydroxide), mono-, di-, and tri -oxo bridges and Cl Ϫ (7)(8)(9).
Charge separation is initiated by the reaction center redox components coordinated by polypeptides D1 and D2 (10 -12). The oxidizing equivalent is stabilized on ChlP D1 within 6 -8 ps (13,14) and is transferred to redox-active tyrosine Y Z (D1-Tyr 161 ) in tens to hundreds of nanoseconds (15,16). In tens to hundreds of microseconds, the oxidizing equivalent is transferred to the Mn 4 Ca cluster (15,16), with the localization of the cationic charge, depending on the oxidation state of the cluster and still a matter of speculation (9,17). The reducing equivalent generated in primary charge separation initially resides on a pheophytin (13,14), and is transferred to the primary quinone electron acceptor, Q A , in ϳ250 ns (18). Subsequent electron transfer to the secondary electron acceptor, Q B , occurs in the hundreds of microseconds time scale and is coupled to proton transfer (19).
Whereas there have been considerable advances in the understanding of the structure and coordination of the Mn 4 Ca cluster, the x-ray (3)(4)(5) and extended x-ray absorption fine structure (EXAFS)-derived (8,20) structures provide no information as to how the cluster is assembled. Such information has been obtained through examination of the functional consequences of site-directed mutations (21)(22)(23). For instance, Ananyev and co-workers (24) have demonstrated using a D2-Y160F mutant (D2-Tyr 160 replaced by Phe) that redox-active tyrosine Y D probably plays an indirect role in accelerating photoactivation (light-driven assembly) of the Mn 4 Ca cluster. Extensive site-directed mutagenesis and spectroscopic studies have associated D1-Asp 170 more directly with binding of the first manganese involved in the assembly of the OEC (25)(26)(27)(28). For example, Ala, Asn, and Ser mutations at this site raise the apparent dissociation constant (K m ) 50 -60-fold in manganesedepleted PSII core complexes, as measured by blockage by Mn 2ϩ of charge recombination between Q A Ϫ and the oxidized donor side (25) and by the concentration dependence of Mn 2ϩ on the rate of reduction of Y Z ⅐ (26). D1-Asp 170 is shown in the x-ray crystallographic structure to coordinate Mn4, making this manganese, therefore, the most likely to be the first one bound in the assembly process. The two most detailed x-ray structures (3)(4)(5) depict Mn4 to be also coordinated by D1-Glu 333 , with the 3.0-Å structure (5) depicting this residue as bidentate, bridging Mn4 and Mn3. As a consequence of this assignment, substantial attention in this paper has been directed at mutations constructed at this site. Some reservations, however, have been voiced regarding the accuracy of the proposed structures because of reduction of the Mn 4 Ca cluster during the collection of the x-ray diffraction data (9,29,30). Considering the resolution of the structures, and the fact that the x-ray diffraction data were collected at 100 K, it would seem unlikely that a displacement of the manganese-manganese distances of Ͻ1 Å upon reduction (30) would cause major reorganization of the relative locations of the coordinating residues. Therefore, residues D1-Asp 170 and D1-Glu 333 , although far apart in the primary structure of the D1-polypeptide, are certainly located in close proximity in the fully assembled Mn 4 Ca cluster, even if they do not coordinate the same manganese. It is at present unclear at what stage of Mn 4 Ca cluster assembly the C terminus of D1 is folded to bring the two residues together. In this paper, experiments are described that examine a possible role for D1-Glu 333 in the binding and oxidation of Mn 2ϩ at the high affinity manganese binding site, implicated in the first step of cluster assembly. Mutations at this site are compared with others in the C-terminal region of D1. The reported observations are consistent with the C terminus of the D1-polypeptide, taking on a conformation close to that depicted in the x-ray crystal structure at the very first step of the assembly process. However, D1-Glu 333 appears not to be more privileged than other coordinating D1-C-terminal carboxylate residues in facilitating the coordination of the first manganese.

EXPERIMENTAL PROCEDURES
The glucose-tolerant strain of the cyanobacterium Synechocystis sp. PCC 6803 (31) was used for construction of the sitedirected mutants described in this paper. All strains were grown on BG-11 medium as described by Williams (31) and Metz et al. (32). Five millimolar glucose was included in the medium to allow the propagation of mutants inactive in PSII. The site-directed mutations were constructed in psbA3 as described by Nixon and Diner (25) using a host strain (TD41) in which the psbA1, -2, and -3 genes had been previously deleted.
Initial rates of oxygen evolution were measured on whole cells of Synechocystis sp. PCC 6803 in saturating light (Ͼ610 nm) at 25°C using a thermostated Hansatech oxygen electrode. The cells were suspended in BG-11 medium containing 5 mM glucose to which 0.3 mM 2,6-dichloro-p-benzoquinone and 1 mM K 3 Fe(CN) 6 had been added. The concentration of chlorophyll present in the sample was determined spectroscopically by extracting the cell pigments by brief sonication in methanol and using an extinction coefficient of 79.24 ml mg Ϫ1 cm Ϫ1 at 665 nm (33). Typical initial rates of oxygen evolution were 490 Ϯ 36 mol of O 2 (mg Chl) Ϫ1 h Ϫ1 in strain TC35 (psbA1 Ϫ , psbA2 Ϫ , with a wild-type copy of psbA3).
PSII core complexes were isolated from cells grown in 20-liter clear plastic carboys (Nalgene Labware). The isolation procedure was performed according to the combined methods of Tang and Diner (34) and Rögner et al. (35), in that order and stored at Ϫ80°C until use. The passage of the core complexes through the hydroxylapatite column (SynChropak HAP-3, SynChrom Inc.) resulted in complete extraction of the Mn 4 Ca cluster (25).
Optical spectroscopy on PSII core complexes was performed using a flash detection spectrophotometer based on a design by Joliot et al. (36) and described in Metz et al. (32). This instrument has 1-s time resolution. Actinic flashes were provided either by a xenon flash (EG&G FX-199, 8-mm arc, 2-s width at half-height) or a Cyanosure LFDL-3 dye laser using sulforhodomaine 640 dye (Exciton Inc., 1-s width at half-height). The kinetics of charge recombination between Q A Ϫ and the oxidized PSII donor side in the presence and absence of MnCl 2 were followed at 325 nm (absorbance maximum of Q A Ϫ Ϫ Q A , (37)) as previously described (25,38), using weak detecting flashes (EG&G FX-199U) rendered monochromatic by a Jobin Yvon HL300 monochromator. The kinetics of Y Z ⅐ reduction in PSII core complexes, measured at different MnCl 2 concentrations, were performed as described by Diner and Nixon (26). All core complexes were suspended at a concentration of 10 g/ml chlorophyll in 20 mM MES-NaOH (pH 6.1), 1 mM CaCl 2 , 1 to 2 M K 3 Fe(CN) 6 .
The same instrument was also used for measurements in whole cells of the amplitude of and kinetics of relaxation of the relative quantum yield of chlorophyll fluorescence (25,38). This kinetic relaxation, following excitation with saturating actinic light flashes (EG&G FX-199), was monitored using weak detecting flashes (EG&G FX-199) rendered monochromatic by the HL300 monochromator at 422 nm. The filter sets used for the actinic flash and to protect the sample photodiode were the same as used previously (32) except that the Schott LF550 was replaced by a Schott LF470. The same technique was used to determine the relative PSII reaction center concentration per cell, which is proportional to the total variable fluorescence yield at a fixed cell concentration. The variable fluorescence is the difference in the fluorescence yield measured at a given time point after actinic illumination minus the F 0 level (Q A oxidized) measured before actinic illumination. Prior to measurements of the evolution of the variable fluorescence following actinic flash illumination, Synechocystis cells at an optical density of 0.9 cm Ϫ1 at 730 nm were preincubated for 10 min in BG-11 plus 50 mM HEPES-NaOH (pH 7.5), 0.3 mM p-benzoquinone, and 0.3 mM K 3 Fe(CN) 6 . The preincubation in the dark was long enough to allow for complete oxidation of the primary and secondary quinone electron acceptors, Q A and Q B . To measure the kinetics of charge recombination between Q A Ϫ and the oxidized donor side, 40 M DCMU was added after the 10-min dark incubation, blocking oxidation of Q A Ϫ by Q B . A single saturating actinic flash was given Ն1 min later and the relaxation of the variable fluorescence yield followed as above. Further addition of 20 mM NH 2 OH provides a stable donor electron that acts in place of the Mn 4 Ca cluster, blocking charge recombination. A flash train of 20 saturating flashes (18 Hz) then results in the stable accumulation of Q A Ϫ . The variable fluorescence yield measured under these conditions is a measure of the relative amounts of PSII on a per cell basis (39).

RESULTS
Physiological Characterization of Mutations-Site-directed mutant strains were constructed in psbA3 in the Synechocystis sp. PCC 6803 host strain TD41, as described by Nixon and Diner (25), and contained one of each of the following mutations: D1-His 332 was replaced by Leu; D1-Glu 333 by Ala, Cys, Gln, and His; D1-Asp 342 by Val; D1-Ala 344 by a Stop codon; and D1-Ser 345 by Pro. All of these strains lack PsbA1 and PsbA2. The wild type strains (TC31 and TC35), to which the mutant stains are compared, contain wild-type PsbA3 but lack PsbA1 and -2. TC31 and TC35 were derived from identical constructs and host strains. The site-directed mutants show depressed rates of oxygen evolution (Table 1) and are unable to grow photoautotrophically, with the exception of D1-E333Q that is very weakly photoautotrophic (this work and Ref. 40). In addition, the amounts of PSII core complex isolated biochemically from the D1-E333H, Ala, and Cys mutants were significantly lower from what could be isolated from wild type, consistent with reduced variable fluorescence yields observed in each case (see below).
Donor and Acceptor Side Electron Transfer-The fluorescence yield of chlorophyll in PSII depends upon the oxidation states of several redox cofactors. State P 680 PheoQ A , competent for charge separation, shows a low quantum yield of fluorescence (F 0 ). State P 680 PheoQ A Ϫ , formed from P 680 PheoQ A following charge separation and reduction of P 680 ϩ , shows an elevated quantum yield of fluorescence (F max ). The chargeseparated state P 680 ϩ Pheo Ϫ Q A Ϫ , formed upon excitation of P 680 PheoQ A Ϫ , is close in energy to the lowest singlet excited states of the primary donor and of the antenna chlorophylls.
The consequent long-lived excited states, which reside for the most part in the antenna, are responsible for the elevated fluorescence yield. The state P 680 ϩ PheoQ A Ϫ formed following primary and secondary charge separation of photoexcited state P 680 *PheoQ A is very much lower in energy than P 680 ϩ Pheo Ϫ Q A Ϫ and has a fluorescence yield close to F 0 . This is because P 680 ϩ quenches the fluorescence (41). Upon reduction of P 680 ϩ , but prior to the oxidation of Q A Ϫ , the fluorescence yield increases to F max (42). Upon electron transfer from Q A Ϫ to Q B or Q B Ϫ the fluorescence yield again decreases, with the rate of decrease tracking the kinetics of electron transfer between the quinones. In Synechocystis cells and under the conditions measured here, the relationship between the quantum yield of fluorescence (between F 0 and F max ) and the relative concentration of Q A Ϫ is to a first approximation linear. Monitoring the variable fluorescence yield following each of a series of saturating flashes provides a measure of the relative concentration of PSII and of the intactness of the electron donor side of the PSII reaction center. The relative concentration of the PSII centers in the various mutants was obtained by pretreating cells with 0.3 mM p-benzoquinone and 0.3 mM K 3 Fe(CN) 6 followed by illumination in the presence of 40 M DCMU and 20 mM NH 2 OH, as described under "Experimental Procedures." The results are shown in Table 1 and show a substantial difference in the center concentration between the different D1-Glu 333 mutants. Fig. 1A shows the variation in the quantum yield of fluorescence following each of a series (1.67 Hz) of five saturating actinic flashes given to whole cells of WT (TC35) Synechocystis sp. PCC 6803, pretreated with 0.3 M ferricyanide and 0.3 M p-benzoquinone in BG-11 medium plus 50 mM HEPES-NaOH (pH 7.5). The presence of these oxidants assures that the acceptor side of the reaction center is initially in the Q A Q B state in the dark. The amplitude of the variable fluorescence yield measured at 100 s after the first actinic flash for the Gln, His, and Cys mutants was determined to be 0.89, 0.45, and 0.40, respectively, compared with WT (TC35). The relative amplitudes of the variable fluorescence observed in Fig. 1 in the mutants were somewhat greater than those determined using NH 2 OH plus DCMU ( Table 1) for reasons that may have to do with the influence of the intactness of the Mn 4 Ca cluster on the a Determined in liquid growth medium or Petri plates in the absence of glucose. b Initial O 2 rates measured in saturating light. c Estimated from the total yield of variable chlorophyll fluorescence (F max Ϫ F 0 ). d K m,overall is the concentration of Mn 2ϩ that produces half-saturation of the blockage of charge recombination in previously manganese-depleted PSII core complexes (see Fig. 4). e K m1 ;K m2 are apparent dissociation constants for Mn 2ϩ in two-component hyperbolic fits of the charge recombination data (see Fig. 4). f ND, not determined.
Q A /Q A Ϫ reduction potential (43), which influences the kinetics of Q A Ϫ oxidation. The variable fluorescence and the yields of PSII core complexes were consistently higher in the case of D1-E333Q than for D1-E333H and Cys for reasons that likely have to do with Gln being a better steric match to the wild-type Glu.
The decay of the fluorescence yield, following each actinic flash in Fig. 1A, reflects the oxidation of Q A Ϫ by Q B (Q B Ϫ ) in WT (TC35). These kinetic traces show a marked oscillation of period two, rapid on odd-numbered flashes and slow on evennumbered flashes. On odd-numbered flashes, Q A Ϫ Q B is generated and rapidly transfers an electron to Q B forming Q A Q B Ϫ is then generated on even-numbered flashes. Electron transfer from Q A Ϫ Q B Ϫ to form Q A Q B ϭ is energetically unfavorable and Q A Ϫ oxidation must be coupled to the protonation of Q B Ϫ (44). Electron transfer to form Q A Q B H Ϫ (reaction 2) is therefore slowed relative to reaction 1 by the need to protonate the secondary quinone prior to or concomitant with the electron transfer (44,45). Fig. 1B shows an experiment similar to that of Fig. 1A for whole cells of the three D1-333 mutants treated as above. No period two oscillations of the variable fluorescence yield are apparent. In addition, the fluorescence yield at the earliest time point (100 s) progressively decreases with actinic flash number. This observation is in contrast to that of WT (TC35) where the initial fluorescence yields, following each actinic flash, show a damped period two oscillation around a horizontal straight line. A progressively increased quenching of fluorescence at 100 s, accompanied by an absence of period two oscillations, indicates the absence of or impairment of tertiary electron donation. In such a case, Y Z ⅐ is unable to be fully reduced in the time between a first actinic light flash and a second. As a result of the incomplete reduction of Y Z ⅐, the lifetime of P 680 ϩ is increased, generating increased amounts of P 680 ϩ PheoQ A Ϫ at 100 s following each subsequent actinic flash of the flash series. Because P 680 ϩ is a quencher of fluorescence (see above and Ref. 41), the increased concentration of P 680 ϩ PheoQ A Ϫ results in a progressive quenching. The quenching is already apparent on the second flash and more marked on the third and subsequent flashes. It is more extreme for the D1-E333H and Cys mutants, similar to what was previously observed for the most perturbed mutations constructed at D1-170 (e.g. D1-D170S and Ala (25)) and consistent with their reduced ability to evolve O 2 . The quenching is less marked for the D1-E333Q mutant, which is less impaired in its ability to evolve O 2 . These observations are similar to those reported by Chu et al. (40) for the His and Gln mutants and indicate either an impaired ability of the OEC to advance to the higher S states or a lack of assembly of the manganese cluster in some (Gln mutant) or nearly all centers (His, Cys, and Ala mutants). Kinetics of Charge Recombination-Charge recombination in PSII between the reduced acceptor and oxidized donor sides is thought to occur through P 680 ϩ . A consequence of the increased lifetime of Y Z ⅐ is that there should be an increased equilibrated concentration of P 680 ϩ , which means that the mutants should show an increase in the rate of charge recombination between Q A Ϫ and the oxidized donor side. The kinetics of Q A Ϫ oxidation by charge recombination can be followed by measuring the decay of the quantum yield of fluorescence in the presence of 40 M DCMU, which blocks forward oxidation of Q A Ϫ . Fig. 2 shows such a charge recombination experiment in whole cells of WT (TC35) and D1-E333Q. The kinetics were fit with two exponential terms plus a constant. Of the two exponential components obtained for the WT (TC35), the slower and major component gave a of 1.89 s (t1 ⁄ 2 ϭ 1.31 s, 65% of total), whereas the faster and minor component gave a of 141 ms (t1 ⁄ 2 ϭ 98 ms, 23% of total). Of the two exponential components for D1-E333Q the faster and major component gave a of 51 ms (t1 ⁄ 2 ϭ 35 ms, 45% of total) and a slower and minor component gave a of 2.63 s (t1 ⁄ 2 ϭ 1.82 s, 37% of total). The large slow component observed in the WT(TC35) reflects the recombination between Q A Ϫ and the S2 state of the OEC. The smaller percentage slow component in the D1-E333Q mutant likely represents the reduced fraction of centers that have also generated S2Q A Ϫ with an assembled Mn 4 Ca cluster. The larger faster component in the D1-E333Q mutant is consistent with an increased lifetime of Y Z ⅐ in those centers in which the Mn 4 Ca cluster is not assembled and which have not been able to oxidize a mononuclear Mn 2ϩ during the lifetime of the charge separated state. The increased Y Z ⅐ concentration in turn produces an increased equilibrated concentration of P 680 ϩ , accelerating the recombination.

Mn 2ϩ Influence on Charge Recombination of Q A
Ϫ with Y Z ⅐-One reason for an increased lifetime of Y Z ⅐ in the D1-E333Q mutant is the partial absence of or inability to oxidize manga-nese, the tertiary electron donor of PSII. Measurements of the kinetics of charge recombination in the presence of varying concentrations of MnCl 2 provide a convenient means to evaluate the ability to bind and oxidize Mn 2ϩ in PSII core complexes that had been previously depleted of manganese. In such complexes, the extent of re-oxidation of Q A Ϫ by charge recombination in the presence of Mn 2ϩ monitors how effectively Mn 2ϩ donates to Y Z ⅐ during the lifetime of the charge separated state (25). In the absence of Mn 2ϩ , Q A Ϫ is re-oxidized by recombination through P 680 with Y Z ⅐. Forward electron transfer does not occur from Q A Ϫ due to the absence of Q B in the PSII core complexes. In the presence of increasing concentrations of Mn 2ϩ , the re-oxidation of Q A Ϫ by recombination is increasingly blocked as Y Z ⅐ is reduced by Mn 2ϩ faster than it can be reduced by back reaction with Q A Ϫ . The reason for the blockage is that Mn 3ϩ is much more stable than Y Z ⅐ with regard to charge recombination. An example of how this works is shown in Fig. 3 where the lifetime of Q A Ϫ is followed at 325 nm, an absorbance maximum of the Q A Ϫ Ϫ Q A difference spectrum (37). Here we compare WT (TC35) manganese-depleted PSII core complexes in the presence of 0.25 mM EDTA, 1 M K 3 Fe(CN) 6 , and 1 mM CaCl 2 in the presence or absence of 1 mM exogenous Mn 2ϩ (MnCl 2 ). One micromolar K 3 Fe(CN) 6 is added to assure that Q A is initially oxidized but not so concentrated as to oxidize Q A Ϫ in competition with charge recombination. In the absence of Mn 2ϩ , reoxidation of Q A Ϫ is rapid and markedly biphasic (t1 ⁄ 2 of 53 ms (45%) and 2.6 s (24%)), with the rapid component arising from Y Z ⅐Q A Ϫ recombination. In the presence of 0.75 mM Mn 2ϩ (not complexed by EDTA) the reoxidation of Q A Ϫ by recombination with Y Z ⅐ is largely blocked due to Y Z ⅐ reduction by Mn 2ϩ with exponential components with t1 ⁄ 2 of 98 ms (9%) and 4.8 s (57%). The slow phase of Q A Ϫ oxidation likely comes from slow electron transfer to K 3 Fe(CN) 6 or recombination with Mn 3ϩ or a combination of the two.
To evaluate the efficacy of electron donation by Mn 2ϩ to Y Z ⅐, the ratio of ⌬I/I at 2 s divided by ⌬I/I at 500 s was calculated as  in Ref. 25 using manganese-depleted PSII core complexes. In the absence of added Mn 2ϩ , this ratio is low (typically 0.35-0.4). The ratio increases with increasing concentrations of Mn 2ϩ , as recombination is increasingly blocked by Mn 2ϩ reduction of Y Z ⅐. By plotting the ratio as a function of the Mn 2ϩ concentration (Fig. 4), we can evaluate how well Mn 2ϩ is bound by the core complex and oxidized by Y Z ⅐ during the lifetime of the charge separated state. The Mn 2ϩ titrations in the wild-type and site-directed mutants were fit with two hyperbolic components (Y ϭ A[Mn 2ϩ ]/([Mn 2ϩ ] ϩ K m1 ) ϩ B[Mn 2ϩ ]/([Mn 2ϩ ] ϩ K m2 ) ϩ C), where K m1 and K m2 are apparent dissociation constants for the binding and oxidation of Mn 2ϩ . Table 1 Table 1) gives components of 0.044 (57%) and 9.8 (43%) M for the WT (TC31). In the case of the site-directed mutants, the dominant component is in the micromolar range with the exception of D1-E333H, which has a major component of 44 M. The submicromolar components are all of much smaller amplitude in the mutants than they are in the WT. All of the mutants clearly require higher concentrations than WT (TC31) of MnCl 2 to block charge recombination, although in detail their dependence on the Mn 2ϩ concentration does differ in each case. The elevated K m values reflect the contributions that all of these sites make to the coordination and oxidation of Mn 2ϩ at the D1-Asp 170 site, consistent with their localization in close proximity to the high affinity binding site. None of the constructed mutants produce effects of the same magnitude as some of those at D1-Asp 170 (e.g. replacement by Asn, Ala, and Ser), implying that the contributions of the wild-type residues to the binding and oxidation of the first Mn 2ϩ are indirect, likely through electrostatic effects that enhance the local concentration of Mn 2ϩ . The range in K m between the mutants would then likely reflect differences in electrostatic charge and location.
Influence of Mn 2ϩ on the Kinetics of Reduction of Tyrosine Y Z ⅐-The titration of the blockage of charge recombination by MnCl 2 is an indirect measure of the rate of reduction of Y Z ⅐ by Mn 2ϩ , as it represents the competition between two pathways for Y Z ⅐ reduction (from Mn 2ϩ and Q A Ϫ ). Although much more costly in core complexes, it is possible to measure the Y Z ⅐ lifetime directly following a saturating actinic flash by measuring the absorbance difference at 429 Ϫ 436 nm as a function of time (26). This difference is primarily due to Y Z ⅐ Ϫ Y Z and comes from electrostatic field changes associated with Y Z oxidation on the spectrum of ChlP D1 (47). By taking this difference, one also eliminates the absorbance difference contributed by P 680 ϩ Ϫ P 680 . There is also a contribution from Q A Ϫ Ϫ Q A , which was eliminated by measuring the kinetics of Q A Ϫ oxidation at 325 nm and multiplying by a correction factor of 0.45, a reflection of the extinction coefficient at 325 nm versus that at 429 Ϫ 436 nm. This value was then subtracted from the 429 Ϫ 436 nm difference to obtain the absorbance difference of Y Z ⅐ Ϫ Y Z only. An example is shown in Fig. 5, which plots the kinetics of relaxation of Y Z ⅐ Ϫ Y Z following a single saturating actinic flash in core complexes of WT (TC31) and D1-E333Q in the absence and presence of 1, 3, and 10 M MnCl 2 . The pseudo first-order rate constants were extracted from fits to a family of experiments of this type performed on both WT (TC31 and TC35) and the D1-E333Q mutant. In an earlier study on WT (TC31) (26), the data were fit out to 1 s with two exponentials plus a constant. In the present study the data were fit out to 170 ms with a single exponential plus a constant, providing a greater constraint on the fitting. The faster of the two rate constants obtained in the earlier study and the single rate constant in the present study are within a factor of two of each other. Fig. 6 shows the pseudo first-order rate constants associated with the reduction of Y Z ⅐ plotted as a function of the concentration of MnCl 2 in core complexes isolated from WT (both TC35 and TC31) and mutants D1-D170S and D1-E333Q. In the absence of added MnCl 2 , the WT strains show one rate for core complexes prepared as described under "Experimental Procedures" (62 s Ϫ1 , 73% of the total relaxation) and another much slower component (4 s Ϫ1 ) in core complexes prepared in the same way, but washed with 2.5 mM EDTA (pH 8.0). The EDTA treatment removes tightly bound adventitious Mn 2ϩ that is capable of reducing Y Z ⅐. In the absence of a tertiary electron donor, the rate of 4 s Ϫ1 is dominated by charge recombination between Q A Ϫ and Y Z , considerably faster than the rate (Յ0.14 s Ϫ1 ) of Q A Ϫ oxidation by K 3 Fe(CN) 6  The rate of Y Z ⅐ reduction in WT (TC31) was shown previously (26) to increase linearly with MnCl 2 up to a concentration of 30 M. Fig. 6 shows here for both TC31 and TC35 a secondorder rate constant of 2.6 ϫ 10 7 M Ϫ1 s Ϫ1 for Mn 2ϩ reduction of Y Z ⅐. As shown previously (26), the linear increase in the rate of reduction of Y Z ⅐ with [Mn 2ϩ ] in the D1-D170S mutant shows a much smaller second-order rate constant, here 8.0 ϫ 10 5 M Ϫ1 s Ϫ1 , indicative of a much weaker binding of Mn 2ϩ . In the case of the D1-E333Q mutant, the dependence on [Mn 2ϩ ] is different from the other two cases, showing a generally sigmoidal dependence with small effects on rate at 1 and 3 M MnCl 2 . The rates and percent contribution obtained from the fits for the D1-E333Q mutant in the presence of 0, 1, 3, and 10 M MnCl 2 are 45 (39%), 47 (76%), 71 (75%), and 358 s Ϫ1 (77%), respectively. Like the initial observation attributed to adventitious Mn 2ϩ binding (in the absence of added MnCl 2 ), the D1-E333Q mutant shows an intermediate dependence on [Mn 2ϩ ], weaker than WT and stronger than that of D1-D170S. This dependence mirrors that of the effect of Mn 2ϩ on charge recombination (Fig. 4). Both sets of experiments reflect a weakening of Mn 2ϩ binding and oxidation in the D1-Glu 333 mutants, consistent with participation of D1-Glu 333 and the other C-terminal coordinating residues in the binding and oxidation of Mn 2ϩ in the earliest steps of assembly of the Mn 4 Ca cluster but at a level substantially weaker than that attributed to D1-Asp 170 .

DISCUSSION
We have previously shown that D1-Asp 170 is implicated in binding and oxidation of the first manganese of the OEC. Mutations at this site not only affect the affinity of Mn 2ϩ for the reaction center, but they affect as well the coordination envi-  ronment of the Mn 3ϩ or Mn 4ϩ (depending on the mutation) generated upon illumination at a concentration of 1.2 Mn 2ϩ / center (28). The 3.0-(5) and 3.5-Å (4) x-ray crystallographic structures of PSII depict D1-Asp 170 as a ligand in the fully assembled manganese complex, coordinating Mn4, implying that this manganese is the first to be assembled into the Mn 4 Ca cluster. Also depicted coordinating this manganese is D1-Glu 333 , a residue located far from D1-Asp 170 in the primary structure of the D1-polypeptide. The x-ray structure shows the C-terminal region of the D1-polypeptide to be folded in such a way as to bring these two residues together to permit simultaneous coordination of the same manganese. Even if there were some error in the x-ray structures regarding the details of cluster co-coordination (see Introduction and below), it is certain that these residues must be in close proximity. The experiments presented here allow us to distinguish between two cases: 1) where the conformation of the C-terminal region is already close to its mature form prior to or during the binding and oxidation of Mn 2ϩ , bringing D1-Asp 170 and some D1 C-terminal residues close enough to cooperatively contribute to the binding and oxidation of the first manganese, and 2) where the conformation of the C-terminal region is established following the binding and oxidation of one or more manganese and where the D1 C-terminal residues do not contribute at all to the coordination of the first Mn.
PSII Donor Side Electron Transfer and Charge Recombination-Examination of site-directed mutations constructed at D1-333 to replace Glu with Gln, His, and Cys and at other sites in the C-terminal region, allows one to discriminate between these possibilities. Experiments in whole cells of the D1-Glu 333 mutants show different levels of impairment in oxygen evolution relative to WT (Table 1), a progressive fluorescence quenching observed following each of a series of five actinic flashes and a loss of the oscillation of period two in the kinetic relaxation following each actinic flash (Fig. 1). These observations imply a loss of tertiary electron donation in PSII, consistent either with a reduced affinity for Mn 2ϩ binding and oxidation resulting in an impaired ability to assemble the Mn 4 Ca cluster, or with an energetically unfavorable oxidation of manganese with or without an assembled Mn 4 Ca cluster. Measurements of charge recombination in the presence of DCMU in whole cells of the D1-E333Q mutant (Fig. 2) show a dominant rapid kinetic phase with a t1 ⁄ 2 of 35 ms (45% of total). Measurements of the kinetics of charge recombination in whole cells of WT (TC35) and mutant D1-D170S show major components with t1 ⁄ 2 of 1.31 s (65%) and 15 ms (87%), respectively. It is thus likely that the differences in the rates of recombination between these different strains reflect differences in the ability of the oxidizing equivalent to be shared with a tertiary manganese. In the case of the WT this is predominantly the S1 state of the OEC, which is oxidized to S2. In the most extreme case of D1-D170S there is probably no tertiary electron donor manganese at all and the charge recombination occurs between Y Z ⅐ and Q A Ϫ through P 680 . That charge recombination between Q A Ϫ and P 680 ϩ occurs with a t1 ⁄ 2 of 1 ms (32) implies an equilibrium constant of ϳ14 for the reaction P 680 ϩ Y Z ϭ P 680 Y Z ⅐(H ϩ ). The case of D1-E333Q is intermediate between the two, with a slow phase (t1 ⁄ 2 ϭ 1.82 s, 37%) likely reflecting that fraction of centers that has successfully assembled a functional Mn 4 Ca complex (this work and Refs. 40 and 48) and that has advanced to the S2 state. The slight slowing of this component relative to that of WT may reflect either the influence of the mutation on the rate of S2Q A Ϫ recombination or some contribution from Q A Ϫ Mn 3ϩ recombination in those centers that have not assembled the Mn 4 Ca cluster, but which have been able to bind and oxidize Mn 2ϩ during the lifetime of the charge separated state. The dominant fast phase (t1 ⁄ 2 ϭ 35 ms, 45% of total) in the D1-E333Q mutants very likely arises from those centers that have been unable to assemble the Mn 4 Ca cluster. This phase, two times slower than what is observed for charge recombination in D1-D170S, may reflect the influence of higher affinity binding of Mn 2ϩ in D1-E333Q relative to D1-D170S on the reduction potential of Y Z ⅐/Y Z or of P680 ϩ /P680. Chu et al. (40) reported a sensitivity of the rate of charge recombination and of the rate of oxygen evolution to the intensity of light used to grow the cells in a number of D1-Glu 333 mutants. We did not observe such sensitivity to light here (Table 1) as the rate of O 2 evolution was consistently higher in D1-E333Q cells cultivated at 50 E/m 2 s (29% of WT (TC35)) than at 2 E/m 2 s (16% of WT (TC35)). The difference in O 2 evolution rate was likely in large part due to a difference in the PSII center content in the cells at the two different light intensities (60 and 43% of WT, respectively). Furthermore, the rate of O 2 evolution observed in the D1-E333Q cells grown in carboys was close to that observed in dim light, implying that the core complexes isolated from such cells would not have been subject to light-induced damage. No differences were seen in O 2 evolution rates or in PSII content per cell in WT (TC35 cells) cultivated at the same two light intensities and in carboys. The differences observed here compared with Chu et al. (40) may reflect in part the fact that the mutations here were constructed in psbA3 and in Chu et al. (40) in psbA2. These two genes show some differences in light regulation of transcription (49).
Blockage of Charge Recombination by Mn 2ϩ -Two sets of experiments were performed using PSII core complexes to examine the binding and oxidation of the first manganese of the OEC. Measurements of the ability of Mn 2ϩ to block charge recombination between Q A Ϫ and the donor side indicate an increased K m,overall in PSII core complexes isolated from both the D1-E333Q and the D1-E333H mutants ( Table 1). The concentration of Mn 2ϩ that blocks half of the centers is 0.3, 2.5, and 20 M, respectively, for WT (TC31 and TC35), D1-E333Q, and D1-E333H. These are compared with K m,overall of 1.3, 3.0, 3.4, and 1.7 M for the D1-H332L, D1-D342V, D1-A344Stop, and D1-S345P mutant strains. The detailed fit of the concentration dependence to two hyperbolic binding components with dissociation constants K m1 and K m2 is shown in Table 1. The WT strains are clearly different from the mutant strains, with the latter showing a much less substantial tight binding component. Clearly the binding, although heterogeneous for all strains, is weaker for all of the C-terminal mutants as compared with WT. The replacement of D1-Glu 333 with Gln, however, does not appear to be any more detrimental to the binding and oxidation of the first manganese than are mutations that effect other carboxylate residues, D1-D342V, D1-A344Stop, and D1-S345P (non-C-terminal processing strain (46)). These observations imply that D1-Glu 333 does not contribute to the same extent as D1-Asp 170 to the formation of a high affinity binding site for the first manganese in OEC assembly. This residue therefore probably does not participate with D1-Asp 170 in direct coordination of the first manganese. However, it does, along with the other carboxylate residues, contribute to enhanced Mn 2ϩ binding, the origin of which may lie in the formation of an anionic space near D1-Asp 170 in which Mn 2ϩ is concentrated relative to its concentration in the bulk phase. D1-E333H is an outlier in that it shows a large component with rather weak binding. This effect could be due to the influence of steric interference by or a positive charge on the imidazole ring of histidine in the binding of Mn 2ϩ to the D1-Asp 170 site, although we cannot exclude some influence of photodamage in this strain as suggested by Chu et al. (40). However, as pointed out above, this latter possibility seems unlikely in light of the fact that the cells, from which the core complexes were isolated, were grown in carboys.
Influence of Mn 2ϩ on the Kinetics of Reduction of Tyrosine Y Z ⅐-Measurements of the rates of reduction of Y Z ⅐ (Fig. 6) indicate a difference between the PSII core complexes even prior to the addition of MnCl 2 . The first-order rate constants observed before MnCl 2 addition are 62 (73%), 45 (39%), and 1.4 s Ϫ1 (80%) for the PSII core complexes isolated from the WT (TC35), D1-E333Q, and D1-D170S strains, respectively. An EDTA wash drops the rate in WT (TC31) to a value (4 s Ϫ1 ) similar to that observed in D1-D170S (1.4 s Ϫ1 ). Such adventitious Mn 2ϩ is likely due to trace contamination of the reagents during the preparation of the core complexes. The adventitious Mn 2ϩ is very tightly bound prior to the exciting flash. In the case of D1-D170S there is no pre-bound Mn 2Ϫ and all of the Mn 2ϩ binds after the flash and very weakly. The adventitiously bound Mn 2ϩ is clearly able to influence the lifetime of Y Z ⅐, with the amount bound and/or its ability to donate to Y Z ⅐ influenced by the nature of the mutation. The sensitivity of WT (TC31) to the EDTA wash and the ability to increase the percent contribution of the 45 s Ϫ1 first-order rate constant upon addition of 1 M MnCl 2 to the D1-E333Q core complexes does imply that the extent of occupancy of a binding site of varying affinity does differ between the different strains even before the addition of MnCl 2 .
In our earlier study, the very high affinity site (K m Ͻ 1 M) appeared in WT to account for little more than one-third of the Y Z ⅐ relaxation in the absence of added MnCl 2 . In the present study (PSII core complexes from strain TC35), the percent occupancy of the high affinity site is higher (73%), although the second-order rate constant calculated using data from both TC31 and TC35 (2.6 ϫ 10 7 M Ϫ1 s Ϫ1 ) is similar and only slightly lower than in the earlier study (4 to 4.8 ϫ 10 7 M Ϫ1 s Ϫ1 ) (26) because of the manner in which the kinetics were fit. The Mn 2ϩ titrated in the second-order process for Y Z ⅐ reduction binds subsequently to the actininc flash at a rate that increases with the Mn 2ϩ concentration, coming on and off many times prior to its oxidation. The earlier data were fit out to 1 s with two exponentials plus a constant, whereas the present data were fit out to 170 ms with a single exponential plus a constant. In the case of the D1-D170S mutant, the kinetic data measured earlier (26) were refit with a single exponential out to 1 s giving a second-order rate constant of 8.0 ϫ 10 5 M Ϫ1 s Ϫ1 , only slightly larger than the 5.5 ϫ 10 5 M Ϫ1 s Ϫ1 measured earlier. This comparison places the second-order rate constant for the D1-D170S some 33-80-fold smaller than that of WT (TC31 and TC35). In the case of the D1-E333Q mutant, the rate constant observed for Y Z ⅐ reduction in the absence of added MnCl 2 changes little upon the addition of 1 M MnCl 2 (45 s Ϫ1 ) although its percent contribution increases. The pseudo firstorder rate constant, however, increases to slightly less than that of WT upon increasing the Mn 2ϩ concentration to 10 M, consistent with the idea that a very high affinity binding component is affected by this and other C-terminal mutations.
Modeling of Early Manganese Coordination-The site-directed mutations, described in this and in previous work, that have the most impact on the K m for binding and oxidation of Mn 2ϩ in Mn-depleted core complexes are D1-D170S, Ala, and Asn. D1-D170E and Cys are the most conservative mutations with the least impact on K m . The presence of the carboxylate group or at least an anionic charge at this site, therefore appears to contribute to the Mn 2ϩ affinity. The consequence of the replacement of D1-Glu 333 with Gln or His is much smaller, with the former quite similar to D1-D342V, D1-A344Stop, and D1-S345P. The C-terminal mutations therefore fall into a second class of weaker effects on first Mn 2ϩ binding and oxidation. The simplest model to explain these observations is one in which D1-Asp 170 is a key component of the high affinity Mn 2ϩ binding site. The C-terminal residues mentioned above all contribute, but more weakly (D1-His 332 most weakly), to the high affinity binding site, most likely through their contributions of anionic charges that increase the local concentration of Mn 2ϩ available to the high affinity binding site at D1-Asp 170 (Scheme 1). That all of these mutations influence the binding behavior of the first Mn 2ϩ indicates that they are near neighbors in the three-dimensional environment of D1-Asp 170 despite their distance in the primary sequence of D1. The difference in behavior of D1-E333Q and His could mean that the latter residue is particularly close to D1-Asp 170 inhibiting sterically or electrostatically the binding process.
One could argue that the C-terminal mutations have some influence on the conformation of the high affinity binding site, a view somewhat different from the electrostatic model described above. However, both an electrostatic model and a conformational one imply that the C terminus of the D1-polypeptide assumes a conformation that enhances the high affinity binding of the first manganese to be assembled into the OEC. This conformation must place the C-terminal region in the vicinity of D1-Asp 170 at the very beginning of the binding process in a conformation close to that observed in the three-SCHEME 1. Schematic showing carboxylate-containing D1-C-terminal residues that enhance the local concentration of Mn 2؉ close to the high affinity Mn 2؉ binding site at D1-Asp 170 . dimensional x-ray structures where the Mn 4 Ca cluster is fully assembled.
Recent x-ray absorption spectroscopy studies have shown that, under conditions of x-ray diffraction data collection on PSII crystals, the production of solvated electrons significantly reduces the manganese complex, largely to the Mn(II) state (29,30). Consequently, reservations have been expressed regarding the accuracy of the described coordination environments of the component manganese and calcium of the OEC in the x-ray crystal structures (9,29,30). The coordination environments as depicted in the x-ray structures may therefore be somewhat different from those that actually exist in the more oxidized dark stable S0 and S1 states of the OEC. The 3-Å resolution of the best x-ray structure still allows for some uncertainty in the ligand assignment. These uncertainties in the precise location of the C-terminal manganese-coordinating resides do not affect the conclusions arrived at in this paper that only require a close proximity of these residues to D1-Asp 170 prior to or during coordination of the first manganese to be assembled into the cluster, most likely what ultimately becomes Mn4 of the fully assembled cluster. Whereas the present results do not require a co-coordination of the first bound manganese by D1-Asp 170 and D1-Glu 333 and may even argue against it, they do not exclude such co-coordination from existing in the fully assembled Mn 4 Ca cluster as depicted in the 3-Å crystal structure (5).