The Basic Properties of the Electronic Structure of the Oxygen-evolving Complex of Photosystem II Are Not Perturbed by Ca²⁺ Removal

Background: EPR/⁵⁵Mn ENDOR spectroscopy of the oxygen-evolving complex (OEC) and Mn²⁺ in Ca²⁺-depleted photosystem II.

Results: Electronic model of the Ca²⁺-depleted OEC; characterization of Mn²⁺ binding.

Conclusion: Ca²⁺ is not critical for maintaining the electronic and spatial structure of the OEC. Its removal exposes a Mn²⁺ binding site supposedly in an extrinsic subunit.

Significance: Mechanistic implications for water oxidation; Mn²⁺ in photoassembly/D1 protein repair.

Abstract

Ca²⁺ is an integral component of the Mn₄O₅Ca cluster of the oxygen-evolving complex in photosystem II (PS II). Its removal leads to the loss of the water oxidizing functionality. The S₂′ state of the Ca²⁺-depleted cluster from spinach is examined by X- and Q-band EPR and ⁵⁵Mn electron nuclear double resonance (ENDOR) spectroscopy. Spectral simulations demonstrate that upon Ca²⁺ removal, its electronic structure remains essentially unaltered, i.e. that of a manganese tetramer. No redistribution of the manganese valence states and only minor perturbation of the exchange interactions between the manganese ions were found. Interestingly, the S₂′ state in spinach PS II is very similar to the native S₂ state of Thermosynechococcus elongatus in terms of spin state energies and insensitivity to methanol addition. These results assign the Ca²⁺ a functional as opposed to a structural role in water splitting catalysis, such as (i) being essential for efficient proton-coupled electron transfer between Y_Z and the manganese cluster and/or (ii) providing an initial binding site for substrate water. Additionally, a novel ⁵⁵Mn²⁺ signal, detected by Q-band pulse EPR and ENDOR, was observed in Ca²⁺-depleted PS II. Mn²⁺ titration, monitored by ⁵⁵Mn ENDOR, revealed a specific Mn²⁺ binding site with a submicromolar K_D. Ca²⁺ titration of Mn²⁺-loaded, Ca²⁺-depleted PS II demonstrated that the site is reversibly made accessible to Mn²⁺ by Ca²⁺ depletion and reconstitution. Mn²⁺ is proposed to bind at one of the extrinsic subunits. This process is possibly relevant for the formation of the Mn₄O₅Ca cluster during photoassembly and/or D1 repair.

Introduction

The oxygen-evolving complex (OEC)⁵ of photosystem II (PS II) catalyzes the light-driven oxidation of water. The OEC contains an inorganic Mn₄O₅Ca metallocofactor that includes five μ-oxo bridge linkages and is coordinated by a framework of surrounding amino acids (1–6) in a highly defined manner that confers catalytic function. The redox-active tyrosine residue Y_Z (D1-Tyr-161) enables electron transfer from the Mn₄O₅Ca cluster to P₆₈₀^+·, the radical cation formed upon photon absorption and charge separation. The Mn₄O₅Ca cluster undergoes four successive oxidations, cycling through a series of different net valence states, referred to as the S_i states (where i = 0–4 denotes the number of oxidizing equivalents stored in the cluster). The transient state S₄ spontaneously returns to S₀ upon regaining four electrons from the two substrate water molecules, which in the process form molecular oxygen. The release of O₂ is followed by the rebinding of at least one H₂O molecule (for reviews, see Refs. 7–14).

X-ray crystallographic structures of the PS II protein complex provided an atomic picture of the structure of the OEC (1–6), identifying all amino acids that ligate the Mn₄O₅Ca cluster. The metallocofactor resembles a distorted chair, consisting of the cuboidal moiety Mn₃O₃Ca (Mn_B3Mn_C2Mn_D1),⁶ with the fourth, outer manganese ion (Mn_A4), connected to the cuboid via an additional μ-oxo bridge (O4) to one of the manganese vertices (Mn_B3). The reported cluster is likely modified due to photoreduction of the Mn^III and Mn^IV ions, such that the Mn-Mn and Mn-Ca distances seen in the x-ray structure are all elongated as compared with those derived from extended x-ray absorption fine structure (EXAFS) measurements (15). Allowing for this, the basic topology of the x-ray structure is similar to earlier literature models, including the geometry-optimized density functional theory (DFT) models of Kusunoki (16), Siegbahn (17), and the recent model of Ames et al. (18), in which the cuboid exhibits an open conformation with Mn_A4 connected to Mn_B3 via a di-μ-oxo bridge (Fig. 1).

FIGURE 1.

Stereo view of a DFT model of the Mn₄O₅Ca cluster in the S₂ state and directly ligating amino acid residues and H₂O/OH⁻ molecules (18). Amino acids, except CP43-Glu-354, are from PS II subunit D1. Manganese, calcium, nitrogen, oxygen, carbon, and hydrogen atoms are shown in purple, green, blue, red, gray, and white, respectively. Nonpolar H atoms are omitted for clarity.

The Ca²⁺ ion of the Mn₄O₅Ca cluster, which can be removed from and reconstituted into the OEC (19–21), is essential for catalytic function (19–23). The non-catalytic Ca²⁺-depleted OEC cannot complete the S state cycle, advancing only to a modified S₂ state, termed S₂′ (24, 25). The reason for this remains unclear. However, four basic explanations exist in the current literature based on the proposed role(s) for the Ca²⁺ ion during the S state cycle (for reviews, see Refs. 26–28). These include the following: (i) As an integral component of the OEC (6), the Ca²⁺ ion can be suspected to be of crucial structural importance. However, EXAFS experiments suggest that Ca²⁺ depletion leads to only a small spatial reorganization of the remnant Mn₄O₅ cluster (29). (ii) It facilitates fast one-electron transfer from the OEC to Y_Z^+· (for reviews, see Refs. 11 and 30). The formation of the S₂′ state requires long visible light illumination at temperatures ≥0 °C. This is in contrast to the native S₂ state, which can be generated via visible light illumination at −78 °C. This apparent increase in the activation energy of OEC turnover upon Ca²⁺ removal may represent a decoupling of the Y_Z^+· from the OEC, such that Ca²⁺-mediated protein conformational changes and/or H⁺ translocations associated with physiological S state transitions are blocked. (iii) It is a binding/staging site for substrate water and its deprotonation (26, 31). The kinetics of substrate water binding to the OEC are affected by biochemical exchange of Ca²⁺ with Sr²⁺, the only surrogate ion able to confer catalytic activity (19, 23, 32). It can presumably act in place of Ca²⁺ as it has approximately the same size and a similar Lewis acidity (31). This result has been interpreted as evidence for Ca²⁺ binding one of the substrate waters. Inhibition due to Ca²⁺ depletion would then reflect the loss of a substrate binding site. (iv) Although the basic structural arrangement of manganese ions in the cluster is retained upon Ca²⁺ removal, it is uncertain if their magnetic and/or electronic interactions are perturbed, which could lead to a decoupling of the cluster or a rearrangement of the manganese valence states. Thus, Ca²⁺ depletion could potentially change the redox properties as well as substrate and/or protein interactions of the complex, inhibiting catalytic function.

The Mn₄O₅Ca cluster in the S₂ state exhibits a characteristic multiline EPR signal centered at g ≈ 2 (33) that arises from an S = 1/2 ground spin state of the cluster. Under certain conditions (illumination, reactants), additional signals are observed at higher g values; in spinach, a second broad signal can be detected at g ≈ 4.1 (34, 35), attributed to an S = 5/2 spin state (36). These signals are affected by the presence of small alcohols, foremost methanol (MeOH) (37–41), which enhance the intensity of the multiline signal at the expense of the g ≈ 4.1 signal (37) (for a full discussion see Ref. 41). The Mn₄O₅ cluster in the S₂′ state also exhibits a multiline signal; however, its hyperfine splitting pattern is perturbed. It contains a larger number of resolved lines as compared with the native S₂ multiline signal, with a smaller average line spacing (5.5–6 versus 8.8 mT). The magnetic interaction between Y_Z^• and the OEC is also perturbed in Ca²⁺-depleted PS II as evidenced by changes in the tyrosine split signal of the S₂′Y_Z^• state (24, 25).

A detailed understanding of the electronic structure of the Mn₄O₅Ca cluster in the S₂ state has been developed from pulse EPR data (42–46), in particular ⁵⁵Mn electron nuclear double resonance (ENDOR). These experiments demonstrated that the four manganese ions contribute about equally to the ground electronic state of the S₂ state; i.e. all four manganese ions carry approximately the same spin density. This requirement allows an assessment of the electronic exchange interactions between the four manganese ions and the development of Mn₄ coupling schemes. These necessarily reflect the geometric structure of the OEC and allow the assignment of the individual manganese oxidation states. Our recently proposed model for the S₂ state (18) is described under “Discussion.” This scheme places the only Mn^III ion inside the cuboidal unit (Mn_D1) (see also Ref. 47) and compares favorably with information from complementary spectroscopic measurements (48–50).

Although it has not been directly observed by EPR spectroscopy, the possibility of another paramagnetic manganese species being able to bind to the Ca²⁺-depleted PS II has been suggested in an earlier study by Booth et al. (51). The additional species was suggested to be a Mn²⁺ ion that can bind specifically to a site in the protein complex that is created or becomes accessible via structural changes in the course of Ca²⁺ removal. This was based on the observation that, after equimolar amounts of Mn²⁺ ions had been added to Ca²⁺-depleted PS II, no Mn²⁺ was observed by X-band continuous wave (CW) EPR. Upon titrating Ca²⁺ ions back into these samples, Mn²⁺ was released as seen from the appearance of the six-line Mn²⁺ EPR signal.

In this work, both the spin system of the Mn₄O₅ cluster in the S₂′ state of Ca²⁺-depleted PS II and the binding of Mn²⁺ ions to this protein were studied by EPR and ENDOR spectroscopy at X- and Q-band frequencies. The results provide new insight into the role of the Ca²⁺ ion in the native OEC.

EXPERIMENTAL PROCEDURES

Sample Preparation

PS II-enriched thylakoid membranes were prepared from spinach based on the procedure of Berthold et al. (52) using detergent treatment by incubation with Triton X-100 for 15 min. All work was performed in the dark or very dim green light, and the PS II was kept at 4 °C before storage in the dark at −80 °C or in liquid N₂. Chlorophyll concentrations were determined by assays using aqueous acetone (80%) extracts (53) with updated extinction coefficients (54) using an ATI Unicam UV-visible spectrometer UV2–300.

Ca²⁺ depletion and reconstitution based on the low pH/citrate treatment method (21) was achieved as described previously (55). The final buffer used was 50 mm MES, 15 mm NaCl, 0.4 m sucrose, 1 mm EDTA, pH 6.5. Ca²⁺ removal and, as a proof for the integrity of the OEC, Ca²⁺ rebinding was confirmed both by enzymatic assays and by X-band CW EPR. The O₂ evolution rates of native PS II were ∼400 μmol O₂/mg of chlorophyll/h (see the following section). O₂ evolution rates dropped to 5–10% in Ca²⁺-depleted and were reactivated to >80% in Ca²⁺-reconstituted samples. Similar percentages of the S₂ multiline signal were observed after white light illumination with a tungsten lamp through an aqueous 5% CuSO₄ IR filter of the respective samples at 200 K for 5 min (Table 1, Fig. 2A). These numbers are consistent with previous literature reports (25, 29, 56).

TABLE 1.

Oxygen evolution activities and relative S₂ multiline EPR signal intensities of the Ca²⁺-containing native, the Ca²⁺-depleted, and the Ca²⁺-reconstituted PS II membrane preparations from spinach

Observable	Native	Ca2+-depleted	Ca2+-reconstituted
Enzymatic rates/μmol O2/mg chlorophyll/ha	390 ± 30	27 ± 1	330 ± 30
Relative enzymatic rates	100%	7 ± 0%	84 ± 8%
Relative S2 state multiline signal intensitiesb	100%	8 ± 3%	105 ± 12%

^a Determined as an average of at least 8 single measurements at a minimum of 2 different chlorophyll concentrations from 5 to 25 μg/ml.

^b Determined from the peak-to-trough distances of four prominent derivative peaks in the CW EPR spectrum (100).

FIGURE 2.

EPR and ENDOR experimental spectra (black solid traces) and simulations (red dashed traces). A, X-band CW EPR of PS II isolated from spinach. Shown are the Ca²⁺-depleted OEC poised in the S₂′ state (a) and the native (b), Ca²⁺-depleted (c), and Ca²⁺-reconstituted OECs (d) illuminated at 200 K. In the experimental spectrum, the region of the overlapping Y_D^• signal (g ≈ 2) was omitted for clarity. In a, a fourth order polynomial, and in b–d, a background signal of the resonator cavity were subtracted from the raw data. Experimental parameters: MW frequencies, 9.634 GHz (a), 9.44 GHz (b–d); MW power, 0.5 milliwatt (a), 20 milliwatts (b–d); modulation amplitude, 7.5 G (a), 15 G (b–d); time constant, 82 ms; temperature, 8 K (a), 10 K (b–d). B, X-band (a) and Q-band (b–e) Davies ENDOR of the Ca²⁺-depleted S₂′ state from spinach compared with the native and Sr²⁺-substituted S₂ states from spinach and T. elongatus. Shown are Ca²⁺-depleted Mn₄O₅ S₂′, spinach (a and b), native Mn₄O₅Ca S₂, spinach (taken from Refs. 45 and 46) (c), native Mn₄O₅Ca S₂, T. elongatus (from Ref. 49) (d), and Sr²⁺-substituted Mn₄O₅Sr S₂, T. elongatus (from Ref. 49) (e). a and b were smoothed using a 9- and 5-point moving average, respectively. b is the difference of an S₂′ state spectrum after illumination at 0 °C minus an S₁′ state spectrum after dark adaptation (supplemental Fig. S2) to remove an overlapping Mn²⁺ signal. The Mn²⁺ signal is attributed to residual Mn²⁺ ions stemming from a small fraction of damaged manganese clusters. In X-band pulse EPR (not shown) and ENDOR spectra (a), Mn²⁺ contributions were avoided by optimizing the MW pulse lengths for the S₂′ state signal of the Mn₄O₅ spin system with an S = 1/2 ground spin state. Experimental parameters: a and b, MW frequencies, 9.717 GHz (X-band), 34.033 GHz (Q-band); shot repetition rate, 5 ms; MW pulse lengths π, 12 ns (X-band), 72 ns (Q-band); τ, 200 ns (X-band), 440 ns (Q-band); magnetic fields, 380 mT (X-band), 1208 mT (Q-band); RF pulse length π_RF, 4 μs; temperature, 5 K; c–e, see Refs. 45 and 49.

Advancement of dark-adapted S₁′ state EPR samples to the S₂′ and S₂′Y_Z^• states (25) was done by illumination at 0 °C for 3 min, with 125 μm 3-(3,4-dichlorophenyl)-1,1-dimethylurea (10 mm in dimethyl sulfoxide) added to the samples advanced to the S₂′ state, which restricts the acceptor site and, thus, Y_Z to one turnover.

For Ca²⁺ and Mn²⁺ titration experiments, dark-adapted Ca²⁺-depleted PS II membranes were rebuffered in EDTA-free buffer by three cycles of dilution, centrifugation at 39,000 × g for 15 min, and resuspension using 50 mm MES, 15 mm NaCl, 5 mm MgCl₂, 0.4 m sucrose, pH 6.5. The final concentration of PS II reaction centers (RCs) in the samples was 28 ± 3 μm based on a chlorophyll concentration of 6.3 ± 0.8 mg ml⁻¹ and assuming 250 chlorophylls/RC (57) after 15 min Triton X-100 treatment. The samples were incubated with known amounts of Mn²⁺ ranging from 0 to 4 eq per RC for 2 h. For the Ca²⁺ titration, samples containing 0.8 added eq of Mn²⁺ were incubated with known amounts of Ca²⁺ between 0 and 2400 eq for one additional hour. Mn²⁺ and Ca²⁺ ions were added from stock solutions of their chlorides.

Oxygen Evolution Measurements

Steady state PS II enzyme activity at 25 °C was determined by polarographic measurement of the O₂ concentration in a PS II-containing assay mixture using a Clark-type Hansatech oxygen electrode with a high sensitivity Teflon membrane under continuous illumination with a tungsten lamp through an aqueous 5% CuSO₄ IR filter. The assay medium was the buffer of the samples lacking EDTA and with 5 mm MgCl₂ and 0.2 mm phenyl-p-benzoquinone (20 mm in dimethyl sulfoxide) added as an electron acceptor.

EPR/ENDOR Spectroscopy

X-band CW EPR spectra were recorded on a Bruker ELEXSYS E500 spectrometer equipped with an ESR900 liquid helium flow cryostat and an ITC503 helium flow temperature controller (Oxford Instruments Ltd.). X-band pulse experiments were performed with a Bruker ESP 380E spectrometer equipped with a dielectric ring resonator, an Oxford ITC liquid helium flow system, and a temperature controller. Q-band pulse experiments were performed using a Bruker ELEXSYS E580 spectrometer equipped with a laboratory-built cylindrical ENDOR resonator (58), a CF935 cryostat, and an ITC5025 temperature controller (Oxford Instruments Ltd.). Field-swept electron spin echo (ESE)-detected experiments were performed at Q-band frequencies using the pulse sequence π/2-τ- π/2-τ-echo with π = 72 ns and τ = 440 ns. For ⁵⁵Mn Davies ENDOR, the pulse sequence was π-π_RF-T-π/2-τ- π/2-τ-echo, with π = 12 ns (X-band), 72 ns (Q-band), or 16 ns (Q-band Mn²⁺ titration/quantification), π_RF = 4 μs (X-, Q-band) or 4.5 μs (Q-band Mn²⁺ titration/quantification), T = 3.4 μs (X-, Q-band) or 1 μs (Q-band Mn²⁺ titration/quantification), and τ = 200 ns (X-band), 440 ns (Q-Band), or 320 ns (Q-band Mn²⁺ titration/quantification). The radio frequency (RF) was varied randomly in the desired range, and the RF pulses were amplified by an ENI 5100L amplifier. Except for Mn²⁺ titration/quantification, ⁵⁵Mn Davies ENDOR spectra were collected using a home-built computer console with SpecMan control software (59) coupled to an SMT02 external RF pulse generator.

EPR/ENDOR Spectral Simulations

Simulations of EPR and ⁵⁵Mn ENDOR spectra were performed numerically using the EasySpin software package (60). The fitting procedures employed a least squares minimization routine. All tensors were set to be collinear. The Ca²⁺-depleted Mn₄O₅ cluster in the S₂′ state was treated as an effective electronic spin S = 1/2 ground state coupled to the four ⁵⁵Mn nuclei, described by the following spin Hamiltonians for the EPR (Equation 1) and ⁵⁵Mn ENDOR (Equation 2) spectra.

The EPR spectrum was calculated using second order perturbation theory, neglecting nuclear Zeeman terms and forbidden transitions. The ⁵⁵Mn ENDOR spectra were calculated exactly, including nuclear Zeeman terms and considering all transitions. For the monomeric Mn²⁺ ion (S = 5/2, I = 5/2) bound to the Ca²⁺-depleted PS II, the following spin Hamiltonian was solved exactly both for the ESE and ENDOR spectra:

For details on the simulation procedure and the theoretical background, see Refs. 46, 49, and 61.

Temperature-dependent CW EPR Signal Intensity

The temperature was calibrated using a thermometer in place of the sample in the EPR tube. To guarantee that the actual unsaturated intensity I₁ of the S₂′ state modified multiline, as the ground state signal, was measured at all temperatures, the saturation behavior was studied at the lowest temperature employed. As a result, the non-saturating microwave (MW) power of 0.1 milliwatt was used throughout. The intensities I₁ of the derivative signals were measured by means of the heights of 19 peaks throughout the spectral range, thereby minimizing statistical errors and contributions of underlying broader signals, such as from cytochrome b₅₅₉ and the semiquinone-iron complex. How the ground-to-first excited state energy difference Δ is determined from the temperature dependence of I₁ is outlined in the supplemental data.

Quantification of the Relative Concentrations of PS II-bound Mn²⁺ and Hexaquo-Mn²⁺

The Mn²⁺ species in Ca²⁺ and Mn²⁺ titration samples were quantified by means of their Q-band ⁵⁵Mn Davies ENDOR spectra in two ways, and the results were averaged. (i) The relative contributions of the spectra from the pure Mn²⁺ species needed to reproduce the spectra from the various titration points were determined. The spectra from Mn²⁺ already present in the Ca²⁺-depleted PS II samples without the addition of Mn²⁺ ions and from 40 μm MnCl₂ dissolved in the titration buffer represented PS II-bound and hexaquo-Mn²⁺, respectively. (ii) The relative amplitudes of the ⁵⁵Mn ENDOR m_S = −3/2 transitions, which appear in different RF ranges characteristic for the two Mn²⁺ species, were quantified in the regions of 353–376 MHz for PS II-bound Mn²⁺ and 390–395 MHz for hexaquo-Mn²⁺.

RESULTS

EPR and ⁵⁵Mn ENDOR of the Ca²⁺-depleted Mn₄O₅ Cluster in the S₂′ State

The characteristic modified multiline CW EPR signal (24, 25) was observed for Ca²⁺-depleted PS II samples poised in the S₂′ state. It is centered at g ≈ 2 and spans the magnetic field range from ∼260 to ∼430 mT, resolving at least 27 hyperfine interaction (HFI) lines with an average peak-to-peak spacing of ∼6 mT (Fig. 2A). The central HFI lines are superimposed by the signal of the stable tyrosyl radical Y_D^• centered at g ≈ 2, which is not depicted for clarity of presentation. The broad underlying signal of the reduced Q_A^−·Fe²⁺ complex (62) contributes in the 350–375-mT region (24, 25, 29).

Traces a and b in Fig. 2B show the X- and Q-band Davies ENDOR spectra of the S₂′ state recorded at 5 K and magnetic fields of 380 and 1208 mT, respectively. The ⁵⁵Mn ENDOR spectrum of the Mn₄O₅ cluster in the S₂′ state is essentially invariant across the corresponding EPR signal envelope (supplemental Fig. S1). It is ∼130 MHz wide, extending over a range from ∼60 to ∼190 MHz. As compared with the ⁵⁵Mn ENDOR spectrum of the native S₂ state (Fig. 2Bc, supplemental Fig. S1), the Ca²⁺-depleted S₂′ state spectrum is broader. The edges of the spectrum change up to 10 MHz, resulting in a ∼20 and ∼10 MHz increase in the width of the X- and Q-band ⁵⁵Mn ENDOR spectra, respectively, as compared with the Ca²⁺-containing S₂ state of spinach PS II (42, 45, 46). The Q-band spectrum of the S₂′ state exhibits five clearly resolved peaks, as also seen for the native S₂ state spectrum from Thermosynechococcus elongatus (Fig. 2Bd); however, their positions differ slightly.

The X-band CW EPR and X- and Q-band ⁵⁵Mn Davies ENDOR spectra were simultaneously simulated using the spin Hamiltonian formalism (for details see “Experimental Procedures” and Refs. 46 and 49). In these simulations, the S₂′ state Mn₄O₅ cluster is treated as an effective S = 1/2 electronic spin state coupled to the four ⁵⁵Mn nuclei, the same as for the native S₂ state (41, 42, 46, 48, 49, 63–66). This approach requires the ground electronic spin state to be well separated from higher states, as is experimentally observed (see the following section). The simulations reproduce all the major spectral features of the EPR and ⁵⁵Mn ENDOR spectra (Fig. 2, dashed red traces).

The isotropic and anisotropic values of the fitted effective ⁵⁵Mn HFI tensors A_i (i = 1–4) are given in Table 2. For means of comparison, the numbers for the native S₂ state from spinach (46) and the native and Sr²⁺-substituted S₂ states from T. elongatus (49) are also listed. A full set of G and HFI tensor components is listed in supplemental Table S1. As seen for the Mn₄O₅Ca/Sr clusters, four effective HFI tensors are required to simulate the Mn₄O₅ cluster spectra. Their magnitudes are on the order seen for mono- and dimeric Mn³⁺ and Mn⁴⁺ complexes. Hence, their individual spin projection coefficients ρ_i must be on the order of 1 (see Ref. 49). In contrast to preliminary simulations of the S₂′ spectra (67) or others on the S₂ state from spinach PS II (46), the HFI tensors were not constrained to axial symmetry. However, as was found previously in simulations of the Mn₄O₅Ca/Sr clusters in T. elongatus (49), the tensors nevertheless show a considerable degree of axial symmetry. Moreover, these four OEC clusters show the same geometries of their HFI tensors, with larger axial than equatorial tensor components (A_aniso < 0) for A₂ − A₄ and vice versa for the largest HFI A₁ (A_aniso > 0).

TABLE 2.

Isotropic and anisotropic values of the effective ⁵⁵Mn HFI tensors A_i (i = 1–4) used for the simulations of the X- and Q-band EPR and ENDOR spectra of the Ca²⁺-depleted PS II from spinach in the S₂′ state (Fig. 2) and for the S₂ states of native spinach PS II (46) and native and Sr²⁺-substituted PS II from T. elongatus (49)

Species	OEC state	Tensor component	A1	A2	A3	A4
			MHz	MHz	MHz	MHz
Spinach	-Ca2+ S2′	isoa	311	234	202	171
		anisob	72	−84	−38	−59
	Ca2+ S2	iso	298	248	205	193
		aniso	35	−40	−60	−70
T. elongatus	Ca2+ S2	iso	312	251	208	191
		aniso	55	−40	−48	−108
	Sr2+ S2	iso	332	243	203	173
		aniso	59	−37	−30	−56

^a The isotropic A_i,iso (i = 1–4) values are the averages of the principal values: A_i,iso = (A_i,x + A_i,y + A_i,z)/3.

^b The anisotropy in the A_i values is expressed as the difference A_i,aniso = A_⊥ − A_‖ between the equatorial and axial components of the tensor. The equatorial and axial A_i values are defined as A_i,⊥ = (A_i,x + A_i,y)/2, A_i,‖ = A_i,z.

Spin State Energies of the Ca²⁺-depleted Mn₄O₅ Cluster in the S₂′ State

The energy difference Δ of the paramagnetic ground spin state and the first exited state was estimated from the temperature dependence of the unsaturated X-band CW modified multiline signal of the Ca²⁺-depleted S₂′ state. The measured intensities I₁ of the derivative signal at a series of temperatures are depicted in a Curie plot versus 1/T in Fig. 3. This relation is approximately linear over the measured range from 14.4 to 5.5 K and extrapolates to 0 for T → ∞. This Curie behavior of the temperature dependence indicates that the Ca²⁺-depleted S₂′ state features an S = 1/2 ground spin state energetically well separated from states of higher spin multiplicity. The temperature dependence of the S₂′ modified multiline signal can be reproduced reasonably well with Δ ≥ 35 cm⁻¹ corresponding to J_eff ≥ 12 cm⁻¹ (see “Experimental Procedures”). This relatively large separation from states of higher spin multiplicity allows the S₂′ state Mn₄O₅ spin system to be treated in the strong exchange limit, i.e. as an effective S = 1/2 spin state, as assumed in the previous section.

FIGURE 3.

Curie plot showing the dependence of the intensity I₁ of the modified multiline derivative signal of the Ca²⁺-depleted S₂′ state on the inverse temperature T. The error of the x-values comes from the calibration of the actual temperature at the sample position (see “Experimental Procedures”). The curves are simulations of the Curie temperature dependence over a range of Δ values on the basis of Equation S1 in the supplemental data and the simplified electron 2-spin coupling scheme for the OEC outlined under “Experimental Procedures.” Experimental parameters: MW frequency, 9.437 GHz; MW power, 0.1 milliwatt; modulation amplitude, 0.75 mT; time constant, 82 ms; temperatures, 5.5, 6.3, 7.3, 8.7, and 14.4 K.

EPR and ⁵⁵Mn ENDOR of a Specifically Bound Mn²⁺ Ion

The Ca²⁺-depleted PS II preparations exhibit an additional EPR and ENDOR signal in all accessible S′ states that is not present in native PS II samples. At 5 K, Q-band ESE-detected field sweep EPR spectra of the dark-adapted Ca²⁺-depleted PS II preparations (S₁′ state), in which the Mn₄O₅ cluster does not show a perpendicular mode EPR signal, displayed a broad EPR signal centered at g = 1.99 with a full width at half-maximum of ∼63 mT (Fig. 4, inset). A corresponding signal was not observed using CW X-band EPR spectroscopy; the signal is probably too broad to be discerned from the base-line drift in the CW EPR experiment (51). Q-band Davies ENDOR spectra were recorded at several magnetic fields in the RF frequency range of 30 to 400 MHz (Fig. 4). The ⁵⁵Mn ENDOR spectra are dominated by two broad peaks between 100–195 MHz and another line centered at ∼370 MHz. The two lines at 100–195 MHz are dependent on the magnetic field and shift to higher frequencies with increasing magnetic field. The spectra also contain sharp proton signals, one centered at the ¹H Larmor frequency (∼50 MHz) and a strongly coupled one at ∼75 MHz with decreasing amplitude at increasing field positions. Its partner at low frequency (∼25 MHz) lies outside the spectral range. No further low frequency signals were detected for this species using either ENDOR or electron spin echo envelope modulation (ESEEM).

FIGURE 4.

Q-band pulse ESE-detected field-swept EPR (inset) and Davies ENDOR experimental spectra (black solid traces) and simulations (Sim.; red dashed traces) of the Mn²⁺ ion bound to Ca²⁺-depleted PS II isolated from spinach and poised in the S₁′ state. In the EPR spectrum, the region of the overlapping Y_D^• EPR signal (g ≈ 2) is not displayed for clarity and was omitted in the simulations. The arrows indicate the four magnetic fields at which the ENDOR spectra were measured. Experimental parameters: MW frequency, 34.07 GHz; shot repetition rate, 5 ms; MW pulse length π, 72 ns; τ, 440 ns; Davies ENDOR, magnetic fields, 1195, 1208, 1224, 1260 mT (top to bottom); RF pulse length π_RF, 4 μs; temperature, 5 K.

These EPR and ⁵⁵Mn ENDOR signals can be readily assigned to high spin Mn²⁺ with S = 5/2, although their appearance is different from the spectra typically associated with Mn²⁺ complexes (see Discussion and Fig. 5A). Simultaneous simulations of the EPR and of four ENDOR spectra at different magnetic fields (Fig. 4, dashed red traces) are consistent with this assignment. They reproduce both the spectral breadth and line shape of the EPR absorption signal and the peaks in the four ⁵⁵Mn ENDOR spectra. Besides a near-isotropic G tensor (principal values 1.983, 1.996, 2.002), the simulations yielded an almost isotropic HFI tensor with the principal components A_x = 256 MHz, A_y = 260 MHz and A_z = 257 MHz, resulting in an isotropic average A_iso of 258 MHz. In addition, the simulations required a large fine structure parameter D = −2355 MHz with a pronounced rhombicity η = E/D = −0.38 of the zero-field splitting (ZFS). It is noted that the predominant contribution to the width of the EPR and ENDOR signals is the large and rhombic ZFS interaction (more information on the effect of the ZFS can be found in the supplemental data and Ref. 61). Hence, considering the good agreement of the measured and calculated EPR and ENDOR signals (Fig. 4), the optimized fine structure parameter D can be considered robust; i.e. a single set of D and E values is sufficient to rationalize the data. The fact that the inclusion of a distribution of the ZFS parameters is not required indicates that there are only small site-to-site inhomogeneities of the Mn²⁺ ligand sphere. Therefore, we propose that the Mn²⁺ ion is bound to one specific site in Ca²⁺-depleted PS II.

FIGURE 5.

Mn²⁺ and Ca²⁺ titrations monitored by Q-band ⁵⁵Mn ENDOR. A, Q-band ⁵⁵Mn Davies ENDOR spectra of dark-adapted Ca²⁺-depleted PS II samples (S₁′ state) with 0.16, 1.2, and 4.0 eq (black, red, and blue trace) of Mn²⁺ ions added relative to the number of PS II RCs and of 40 μm MnCl₂ (corresponding to 1.6 eq) dissolved in the same buffer used for the PS II titration experiments. For the titration curve, see supplemental Fig. S3. The spectra were smoothed using a 5-point moving average. B, titration of Ca²⁺-depleted PS II samples containing 1 eq of Mn²⁺ ions with respect to the PS II RCs with Ca²⁺. The relative ⁵⁵Mn ENDOR signal amplitudes of Mn²⁺ ions bound to the PS II protein complex (black squares) and hexaquo-Mn²⁺ in solution (red circles), quantified as described under “Experimental Procedures,” are plotted against the equivalents of Ca²⁺ ions added to the samples. The concentrations of both Mn²⁺ species as a function of added Ca²⁺ were reproduced by a sigmoid fit curve (solid lines). The concentration of RCs in the samples was 25 ± 3 μm. Experimental parameters: MW frequency, 34.03 GHz; shot repetition rate, 5 ms; MW pulse length π, 16 ns; τ, 320 ns; magnetic field, 1224 mT; RF pulse length π_RF, 4.5 μs; temperature, 5 K.

Mn²⁺ and Ca²⁺ Titration Experiments

To further investigate the Mn²⁺ species described in the previous section, Mn²⁺ and Ca²⁺ titration experiments of Ca²⁺-depleted PS II samples were performed, monitoring the CW EPR and ENDOR signal described above.

Mn²⁺/Ca²⁺ Titration Monitored by CW EPR

Mn²⁺ ions were added to Ca²⁺-depleted PS II samples and the characteristic S₂′Y_Z^• state split signal, S₂′ multiline signal, and hexaquo-Mn²⁺ signal (not shown) were measured. The addition of ≤0.8 eq of Mn²⁺ ions relative to the number of PS II RCs did not quantitatively alter the three signals. The Mn²⁺ ions added are CW EPR-silent, as seen in the study of Booth et al. (51), which is consistent with a protein-bound Mn²⁺ species. In addition, this species does not cause any line broadening or even splitting of the signals from the OEC or the tyrosyl radicals. The addition of ≥0.8 equivalents of Mn²⁺ ions resulted in the appearance of the hexaquo-Mn²⁺ signal. The subsequent addition of Ca²⁺ to Ca²⁺-depleted, Mn²⁺-loaded PS II samples led to a loss of the S₂′Y_Z^• state split signal and of the multiline signal, as the Ca²⁺-reconstituted Mn₄O₅Ca cluster can proceed beyond the S₂′ state upon illumination. A concomitant increase of the Mn²⁺ six-line signal was observed due to the release of the PS II-bound Mn²⁺ into solution (51).

Mn²⁺/Ca²⁺ Titration Monitored by ⁵⁵Mn ENDOR

Mn²⁺ binding was also directly monitored by Q-band ENDOR. The concentrations of PS II-bound and solubilized Mn²⁺ ions in each sample were quantified by means of the relative amplitudes of their characteristic ⁵⁵Mn ENDOR signals (Fig. 5A; for the titration curve, see supplemental Fig. S3). Without the addition of MnCl₂, dark-adapted Ca²⁺-depleted PS II (S₁′ state) always displayed the PS II-bound Mn²⁺ signal shown in Fig. 4. The addition of ∼0.8 eq of MnCl₂ led to a 4–5-fold increase of this signal with only little free hexaquo-Mn²⁺ (15 ± 4%) present at the same time. This suggests that ∼20% of RCs contain a bound Mn²⁺ before exogenous addition of MnCl₂ so that in the end a total of ∼1 eq Mn²⁺ is in the sample. The basal Mn²⁺ is likely derived from centers damaged during the Ca²⁺ depletion procedure and nominally corresponds to the loss of ∼5% Mn₄O₅(Ca) clusters. The high occupancy of the Mn²⁺ site suggests that it is of high affinity, with a dissociation constant K_D that is too small to be determined here. From the employed concentrations of the binding partner, K_D is expected to be in the submicromolar/nanomolar range. It is also noted that the addition of the chelating agent EDTA did not remove or alter the appearance of the bound Mn²⁺ signal, consistent with the protein site having a high affinity for Mn²⁺.

An additional Ca²⁺ titration was performed on the fully Mn²⁺-loaded Ca²⁺-depleted PS II (+0.8 eq of MnCl₂, i.e. a final ratio of 1 Mn²⁺ ion per PS II RC). The Ca²⁺ concentrations ranged from 0 to 2400 eq Ca²⁺ per RC (0–60 mm). In Fig. 5B, the relative concentrations of the two Mn²⁺ species (PS II-bound and solubilized) are plotted against the equivalents of Ca²⁺ ions added. This behavior could be reproduced by a sigmoid curve with a half-saturation value of 700 Ca²⁺ ions per RC. This value is similar to 1200 eq of Ca²⁺ reported in Booth et al. (51). The difference may be due to the Ca²⁺ depletion method used, the low pH/citrate treatment in this study versus a NaCl salt wash (24) in the study of Booth et al. (51). Their differing effects on the extrinsic PS II subunit composition could alter the Ca²⁺ binding kinetics (see Refs. 24 and 51).

DISCUSSION

Location of the Mn²⁺ Binding Site

Based on the observations described above (see “Results”), a preliminary assignment can be made as to where the binding site of the Mn²⁺ ion is located. No strong magnetic interaction was observed between the Mn²⁺ ion and the Ca²⁺-depleted Mn₄O₅ cluster or the tyrosyl radical Y_Z^• in the form of a broadening or splitting of the corresponding EPR signals. Thus, Mn²⁺ binding directly to the Ca²⁺ site of the OEC can be excluded. This Mn²⁺ ion must be at least 10 Å away not to be detectable via dipolar magnetic interaction. A similar argument holds for Y_D^• (D2-Tyr-160), as it also displays an unperturbed EPR lineshape when Mn²⁺ is bound. These “exclusion zones” are indicated by green and violet spheres in Fig. 6A. There is, however, a long range dipolar interaction between the Mn²⁺ ion and Y_D^• as evidenced by the relaxation enhancement of its EPR signal (51). Being smaller than the enhancement resulting from the Mn₄O₅Ca cluster in the S₂ state suggests a weaker Mn²⁺-Y_D^• interaction and thus a longer distance than the 31 Å measured between the cluster and Y_D (6).

FIGURE 6.

Ca²⁺ and potential Mn²⁺ binding sites in cyanobacterial PS II crystals. A, PS II crystal structure from T. vulcanus (6) (PDB accession number 3ARC) highlighting the Ca²⁺ ions (black spheres) as well as a Ca²⁺ binding site found in PS II from T. elongatus (gray sphere) and their distances to the paramagnetic entities Mn₄O₅Ca cluster, Y_Z^•, and Y_D^•. The 10 Å spheres around the latter indicate the approximate region in which a bound Mn²⁺ would cause a splitting of their EPR signals and thus can be excluded to contain the Mn²⁺ binding site. B and C, ligand environments of the Ca²⁺ ions in the extrinsic PsbO proteins from T. vulcanus and T. elongatus (T.e.) (5), respectively. Oxygen, nitrogen, and carbon atoms are shown in red, blue, and yellow, respectively. Differences between the PsbO proteins of these cyanobacterial species and from higher plant spinach are displayed by a sequence alignment in supplemental Fig. S4. All distances are in Å.

The binding and titration behavior can either be rationalized by a significant allosteric effect of Ca²⁺ on the Mn²⁺ site, or Mn²⁺ binding could take place directly at a depleted Ca²⁺ site. The recent crystal structure (6) of PS II from Thermosynechococcus vulcanus exhibits three additional Ca²⁺ sites at distances greater than 30 Å from the paramagnetic species monitored, i.e. the Mn₄O₅(Ca) cluster, Y_Z^•, and Y_D^• (Fig. 6A). In the structure of PS II from T. elongatus, a different Ca²⁺ site in PsbO has been identified (4, 5, 68), not found in the T. vulcanus crystals. All these Ca²⁺ sites are located on the lumenal/donor side of PS II in the subunit CP47, the cytochrome b₅₅₉ subunit β (PsbF), and the extrinsic protein PsbO, and are solvent-accessible. It is not clear, however, whether Ca²⁺ binding at these sites is solely a crystallization artifact under the conditions used or of physiological relevance. With the exception of the two sites in PsbO, the Ca²⁺ sites appear to be of low affinity, as the Ca²⁺ ions are ligated to a large part by H₂O and glycerol. In contrast, the two Ca²⁺ sites seen in the PsbO possess at least three ligands from amino acid side chains (Fig. 6, B and C) and thus are potentially of high affinity. In the homologous PsbO from spinach, which has also been reported to bind Ca²⁺ (69–71), Asn-197 and Val-198 of the binding motif in Fig. 6B correspond to the conserved residues Ser-286 and Val-287, whereas there is no equivalent for Thr-135. Glu-81, Glu-140, and His-257 in the other binding motif (Fig. 6C) correspond to Glu-146, Glu-205, and Glu-317 (for a sequence alignment, see supplemental Fig. S4). Mn²⁺ binding to PsbO has indeed been demonstrated previously in isolated PsbO from higher plants (72–74). As in the present study and Ref. 51, protein-bound Mn²⁺ did not show a CW EPR signal, but a six-line signal was observed after denaturation of the protein (73). PsbO was reported to show carbonic anhydrase activity, which was maximal in the presence of Mn²⁺ (74).

The magnetic properties of the Mn²⁺ ion provide information about the immediate ligand environment in this binding pocket. The D and E values of Mn²⁺ complexes of higher symmetry, such as Mn²⁺-EDTA and hexaquo-Mn²⁺, are significantly smaller than those for the PS II-bound Mn²⁺ described here (see supplemental data and Ref. 61). The large and highly rhombic ZFS reflects an asymmetric coordination sphere. Both the 7- and the 5-fold coordination geometries of the Ca²⁺ ions in PsbO from the two cyanobacterial species exhibit considerable asymmetry (Fig. 6, B and C). In addition, the large proton coupling seen also suggests the Mn²⁺ ion to have at least one water ligand. The absence of any smaller coupling, such as from ¹⁴N (not shown), indicates that the Mn²⁺ ion does not bind to a N-containing ligand residue like histidine. Thus, the absence of a (visible) water and the presence of His-257 as ligands of the Ca²⁺ ion in T. elongatus PsbO (Fig. 6C) favor Mn²⁺ binding to the Ca²⁺ site in PsbO identified in the T. vulcanus crystal structure (Fig. 6B).

PS II from higher plants exhibits an extrinsic subunit composition different from that of the cyanobacterial system. Higher plant lumenal PsbP has been reported to be capable of binding Mn²⁺ stoichiometrically (75, 76). Similar to Ca²⁺-depleted PS II in this study and in Ref. 51, isolated PsbP loaded with Mn²⁺ did not show a Mn²⁺ X-band CW EPR signal unless it was denatured. A bound Mn²⁺ could be detected by high field EPR spectroscopy and distinguished from non-specifically attached Mn²⁺, similar to the present study. It is noted though that the binding constant reported in Bondarava et al. (76) is probably incorrect; for a discussion, see Ref. 77. Moreover, the Mn²⁺ ion in PsbP could be (partially) replaced by Ca²⁺, and Zn²⁺ has been found to bind at one of the two proposed Mn²⁺ sites in PsbP crystals from spinach (PDB accession number 2VU4) and its cyanobacterial homologue CyanoP (78). Mn²⁺ bound to the PsbP would be at least 30 Å from either the OEC or Y_D, again consistent with the distance constraints identified above. Thus, PsbP could also contain the putative site of specific Mn²⁺ binding in Ca²⁺-depleted higher plant PS II.

The physiological role of the putative Mn²⁺ binding site is the delivery of Mn²⁺ to the OEC during photoassembly and/or the storage of Mn²⁺ during the damage/repair cycle of the D1 protein (see Refs. 79–82). Ca²⁺ is essential for photoactivation of the OEC. It was suggested to bind at a site within the PS II complex, which leads to a conformational change of the protein pocket where the OEC is assembled (i.e. the C terminus of D1). Thus, it appears reasonable that in the absence of Ca²⁺, it is favorable for the PS II supercomplex to sequester in a site Mn²⁺ that can be rapidly delivered upon an increase in Ca²⁺ concentration. In this scenario the lumenal Ca²⁺ concentration would be a signaling mechanism for OEC assembly and repair.

Spectral Properties of the Mn₄O₅ Cluster in the S₂′ State Compared with Other S₂ State Systems

The appearance of the ⁵⁵Mn ENDOR spectra, the fitted ⁵⁵Mn HFI tensors, and the ground-to-first excited state energy separation of the Ca²⁺-depleted S₂′ state all fall within the natural spectral variations observed for the native S₂ states in different species (41). This demonstrates that the basic electronic and thus also spatial structure of the Mn₄O₅ cluster remains intact upon Ca²⁺ removal. This confirms and further refines observations on the interatomic distances of the manganese ions from earlier EXAFS experiments (29).

The Ca²⁺-depleted S₂′ state from spinach resembles the native S₂ state from T. elongatus with regard to the spin state energies. Upon removal of the Ca²⁺ ion, Δ increases to ≥35 cm⁻¹, which is much larger than for the native spinach S₂ state (Δ = 3–6 cm⁻¹) but more similar to T. elongatus (Δ = 12–25 cm⁻¹) (41, 83, 84). In intact spinach PS II, the energy ladder is sensitive to MeOH addition. The mechanism by which MeOH binding perturbs the electronic structure of the S₂ state was recently discussed in Su et al. (41). In the model proposed, MeOH binding to the OEC increases the electronic coupling of the pending manganese (Mn_A4) to the cuboidal (Mn_B3,Mn_C2,Mn_D1) unit. It is this effective coupling that defines the ground-to-first excited state energy difference Δ of the S₂ state. Ca²⁺ depletion appears to have the same effect. However, the addition of MeOH did not modify the appearance of the S₂′ state ESE and ENDOR spectra (not shown). It is emphasized though that this effect is of the same size as that of the variation between species and thus is unlikely to be of physiological significance.

Electronic Structure/Exchange Coupling Scheme of the Ca²⁺-depleted Mn₄O₅ Cluster in the S₂′ State

To further rationalize the spectral results from the Ca²⁺-depleted Mn₄O₅ cluster, a spin coupling scheme for the S₂′ state was developed. It was constructed to meet the following requirements: (i) a ground state of spin multiplicity S = 1/2, (ii) a ground-to-first excited state energy difference Δ ≈ 35 cm⁻¹_, (iii) spin projection factors |ρ_i| ≈ 1 for all four manganese electronic spins, and (iv) intrinsic ZFS constants d_i of the manganese ions that lie within the range found for mono- and dimeric model complexes, i.e. 1 cm⁻¹ < |d| < 5 cm⁻¹ for Mn^III and |d| < 0.1 cm⁻¹ for Mn^IV ions in an octahedral ligand environment (see Refs. 18, 47, and 49). The inferred structural (29) and spectral similarity of the native and the Ca²⁺-depleted manganese cluster suggest that the spin coupling scheme for the native S₂ state (Fig. 7, c = 1) (18), in which Mn_D1 is the Mn^III ion, can be used as a starting point. Calculated on the basis of the refined model of the OEC in the latest crystal structure (6), the basic arrangement of this scheme is in accordance with the spatial organization as described by Siegbahn and our group (17, 18, 47, 85), in which Mn_B3, Mn_C2, and Mn_D1 form a trimeric core unit connected to Mn_A4 by a di-μ-oxo bridge via Mn_B3 (Fig. 1). Thus, this scheme represents an extension of the (3 + 1)- or Y-coupling schemes, proposed earlier in EPR spectroscopic studies (42, 46, 47, 49), where J_A4-C2 = J_A4-D1 = 0.

FIGURE 7.

Model for the electronic structure of the OEC in the native S₂ and Ca²⁺-depleted S₂′ states calculated based on a refined DFT structure of the OEC (18) in the latest crystal structure (6). A–D label the manganese ions in their respective oxidation state, and the numbers give the pair-wise exchange coupling J_ij between the electronic spins of the Mn^III and Mn^IV ions in cm⁻¹. The constant c is 1 in the originally derived model but differs for the various clusters and conditions, such as the presence or absence of MeOH. The S₂′ state can be described by the scheme with c = 1.65.

The coupling topology fulfills criteria (i) and (iii) as ground spin state multiplicity and spin projection factors are the same for the two states, S₂ and S₂′. In contrast, their ground-to-first excited state energy differences Δ and effective ⁵⁵Mn HFI tensors A_i, relevant for (ii) and (iv), are different. Thus, the Δ = 10.5 cm⁻¹ calculated for the S₂ state coupling scheme also differs from the experimental Δ ≥ 35 cm⁻¹ determined for the S₂′ state. Correlations between the exchange coupling scheme and this energy difference have been investigated in previous studies (41, 47). One mechanism by which Δ is influenced directly was shown to be the strength and the sign of the exchange coupling between Mn_A4 and the trimeric unit comprising Mn_B3, Mn_C2, and Mn_D1. An increase or decrease in the magnitudes of the coupling constant J_A4-B3 results in a larger or smaller energy gap, respectively. As the monomer-trimer joint is in the vicinity of a possible binding site of a MeOH molecule, this rationalizes the effect of MeOH on the electronic structure of the Mn₄O₅Ca cluster in the native S₂ state (41). For the Ca²⁺-depleted S₂′ state, the coupling of Mn_A4 to the trimeric unit was varied by multiplying the respective exchange coupling constants J_A4-B3, J_A4-C2, and J_A4-C2 by a factor c (Fig. 7). It can be readily calculated that with c = 1.65 (J_A4-B3 = −46 cm⁻¹, J_A4-C2 = 7 cm⁻¹, and J_A4-D1 = 10 cm⁻¹), Δ is 35 cm⁻¹ and thus in the desired range.

For testing whether the obtained model also reproduces reasonable estimates for the intrinsic ZFS values d_i of the Mn^III and Mn^IV ions, a brief description on how those can be assessed based on the inferred coupling scheme and the fitted effective HFI tensors is given in the supplemental data. Because of their inherently small ZFSs, the d_i values of the three Mn^IV ions can be assumed to be 0 cm⁻¹ for the calculations of the intrinsic HFI tensors a_i from the fitted effective A_i and the computed ρ_i tensors. Mn^III ions generally exhibit an absolute isotropic HFI value |a_iso| in the range between 165 and 225 MHz and considerable anisotropy defined as the difference a_i,aniso = |a_‖| − |a_⊥| between the absolute values. Mn^IV ions tend to exhibit slightly larger isotropic HFI values (|a_iso| = 187–253 MHz) and only small intrinsic HFI anisotropies (|a_iso| < ∼30 MHz) (see Ref. 49). For the Ca²⁺-depleted S₂′ state, a ZFS value d_D1 of the Mn_D1^III ion in the range of −2.24 to −2.31 cm⁻¹ yields a_i tensors consistent with the valence states of the individual manganese ions. An optimized ZFS value d_D1 = −2.27 cm⁻¹ leads to the spin projection and intrinsic HFI tensors ρ_i and a_i listed in Table 3. In terms of the intrinsic isotropic and anisotropic HFI values, the calculated numbers match the prerequisites as found in the literature very well. As the ZFS d_D1 = −2.24 to −2.31 cm⁻¹ lies in the range usually found for Mn^III ions (1 cm⁻¹ < |d| < 5 cm⁻¹), the developed model fulfills the four essential criteria imposed.

TABLE 3.

Calculated spin projection tensor components ρ_⊥ and ρ_‖, intrinsic ⁵⁵Mn HFI tensor components a_⊥ and a_‖, and isotropic and anisotropic intrinsic HFI values a_iso and a_aniso for the Mn ions of the OEC in the Ca²⁺-depleted S₂′ state on the basis of the electronic exchange coupling scheme in Fig. 7 with c = 1.65 and intrinsic ZFS values d_A4 = d_B3 = d_C2 = 0 cm⁻¹ for the Mn^IV ions and d_D1 = −2.27 cm⁻¹ for the Mn_D1^III ion

Manganese ion	ρ⊥	ρ‖	a⊥a	a‖a	aisob	aanisoc
			MHz	MHz	MHz	MHz
MnA4 (MnIV)	1.03	1.25	197	230	208	33
MnB3 (MnIV)	−0.81	−1.09	187	190	188	3
MnC2 (MnIV)	−0.87	−1.21	220	188	209	−31
MnD1 (MnIII)	1.66	2.04	202	123	175	−79

^a The equatorial and axial a_i values are defined as: a_⊥ = (a_x + a_y)/2, a_‖ = a_z.

^b The isotropic a_iso values are the averages of the individual components of the tensor a_iso = (a_x + a_y + a_z)/3.

^c The anisotropy of the a tensor is expressed as the difference a_aniso = a_‖ − a_⊥ between the parallel and perpendicular tensor components.

Structural Implications of the Zero-field Splitting d_D1 of the Mn_D1^III Ion

The removal of the Ca²⁺ ion from the spinach OEC is found to result in a significant change of d_D1 from −1.2 cm⁻¹ (41) to −2.2 to −2.3 cm⁻¹. This perturbation is larger than for the Ca²⁺/Sr²⁺ replacement in PS II from T. elongatus. For these systems, the intrinsic ZFS values of the Mn_D1^III ion are relatively similar (Ca²⁺, d_D1 = −1.3 cm⁻¹; Sr²⁺, d_D1 = −1.2 cm⁻¹) (49). It is, however, noted that the signs of the d_D1 and of the HFI anisotropy of the Mn^III ion do not change between the Ca²⁺-depleted S₂′ and the Ca²⁺-containing S₂ state. These parameters can be related to the ligand sphere of the Mn^III ion (86–88). Negative numbers for d_D1 and a_D1,aniso correspond to a ⁵B_1g ground state, obtained in the cases of square pyramidal 5-coordinate or tetragonally elongated 6-coordinate ligand geometries. This suggests the coordination sphere of the Mn_D1^III for the S₂′ and S₂ states to be similar. However, the increase in the magnitude of d_D1 upon Ca²⁺ removal does indicate modifications of the precise binding mode, e.g. altered ligand distances and angles. One possible mechanism for altering d_D1 is protonation of one of the μ-oxo bridges ligating the Ca²⁺ ion (Fig. 1) as a means of overall charge compensation of the cluster upon Ca²⁺ removal. It is known from model complexes that Mn-Mn distances are elongated upon protonation of Mn-O-Mn bridges (89). However, within the trimeric cuboidal unit, this lengthening could be strongly impaired for the Mn_C2-Mn_D1 distance. The fitted averaged distance of the Mn-Mn interactions at 2.7–2.8 Å from EXAFS on Ca²⁺-depleted PS II samples (29), however, does not allow for a conclusive assessment. Also, glutamate 189 of the D1 protein (D1-Glu-189), which directly coordinates the Mn_D1^III (6, 17, 47) and potentially also the Ca²⁺ ion (18), could be reoriented upon Ca²⁺ depletion leading to a distortion of the coordination sphere and thus an altered d_D1.

In the latest crystal structure, all four manganese ions are 6-coordinate (6). This, however, requires the O5 μ-oxo bridge to be a ligand of Mn_A4, Mn_B3, and Mn_D1, engendering very long Mn-O5 bond distances well outside the range seen in model complexes (see Ref. 18) and by EXAFS spectroscopy of the Mn₄O₅Ca cluster in PS II (90, 91). In most geometry-optimized DFT structures, such as those proposed by Siegbahn and our group (17, 18, 47), the position of O5 is significantly altered (Fig. 1). The O5 shifts toward the Mn_A4, forming a genuine μ-oxo bridge between Mn_A4 and Mn_B3, and results in Mn_D1 having an open coordination site. In this case, in the Ca²⁺-depleted S₂′ state, Glu-189 might function as a bidentate ligand in a then tetragonally elongated 6-coordinate Mn_D1^III ligand sphere, leading to the observed change of d_D1.

Alternatively, the absence of Ca²⁺ may allow this open site to be occupied by a water molecule in the S₂′ state (Fig. 8) forming a sixth ligand to Mn_D1. The Mn_D1-bound water molecule is the second substrate water in the mechanism proposed by Siegbahn (17), which potentially binds during the S₂-to-S₃ transition. Thus, within this model, one of the roles of Ca²⁺ in the active cluster would be to prevent the second substrate from binding too early in the reaction cycle (25, 92). This activity would presumably avoid detrimental side product formation (reactive oxygen species) and lead to single product (O₂) formation. Consistent with this role for the Ca²⁺ ion is the known S state dependence of the affinity of Ca²⁺ to this site (93). It drops significantly in the S₃ state, suggesting that in this state Ca²⁺ is less tightly bound, having a more flexible ligand sphere that potentially allows greater solvent access to the Mn_D1 ion.

FIGURE 8.

Scheme of the native Mn₄O₅Ca cluster in the S₂ state and the Ca²⁺-depleted S₂′ state represented by a hypothesized Mn₄O₅ cluster. In the putative S₂′ state, the fast exchanging substrate water is already bound to Mn_D1^III, filling the space of the Ca²⁺ ion. W_s and W_f denote the slowly and fast exchanging substrate waters, respectively (96, 99).

Besides μ-oxo bridge protonation, the loss of two positive charges is likely to be compensated by protonation of amino acid residues ligating the Ca²⁺ ion in the intact cluster. Other possibilities are the replacement of Ca²⁺ by monovalent Na⁺ in the samples or the absence of complete charge compensation, leaving the Mn₄O₅ cluster with an additional negative charge. It is evident that any of these modifications could have a critical effect on the catalytic capabilities of the cluster, especially with regard to proton-coupled electron transfer to Y_Z. In light of the proposed deprotonation sequence 1,0,1,2 for the individual oxidation steps starting from S₀ (94), this would explain the Mn₄O₅ cluster being able to advance to S₂′ but not from S₂′Y_Z^• to S₃′.

Conclusions

This study demonstrates that Ca²⁺ is not required for conferring the critical electronic properties to the Mn₄O₅Ca cluster. This also confirms that Ca²⁺ is not essential for structural maintenance of the OEC. Its presence or absence does not affect the position of the only Mn^III ion of the cluster in the S₂/S₂′ state (Mn_D1), and the contribution of the four manganese ions to the electronic states S₂ and S₂′ does not differ considerably. Thus, the necessity for Ca²⁺ in water splitting catalysis must be due to another functional role of the Ca²⁺ ion.

Although the exact mechanism of inhibition upon Ca²⁺ removal is still unclear, two models can be considered in terms of the two basic catalytic mechanisms proposed in the literature. (i) For mechanisms that involve O-O bond formation between a Ca²⁺-bound and a manganese-bound substrate water (be it a terminal ligand Mn^V = O or a μ-oxo bridge) (11, 95–97), inhibition due to Ca²⁺ depletion is readily explained. The enzyme is inactive, as it has lost a substrate binding site. It should be noted though that this model provides no rationale for the fact that the catalytic cycle is blocked at the stage of S₂′Y_Z^•. (ii) Instead, O-O bond formation has been proposed to follow a mechanism that results in the coupling of substrates bound to two manganese sites (be it between two terminal bound Mn-O ligands or involving a μ-oxo bridge via oxyl radical coupling) (10, 14, 16, 17, 98). Then, inhibition due to Ca²⁺ removal probably represents a secondary effect where the Ca²⁺ ion is critical for maintaining the H-bond network between Y_Z and the manganese cluster (6, 11, 30) as opposed (or in addition) to perturbation of substrate binding. Thus, Ca²⁺ removal would disable proton-coupled electron transfer during the S₂′Y_Z^•-to-S₃′ transition, preventing substrate deprotonation and concomitant oxidation of Mn_D1^III. Therefore, the elucidation of the mechanistic role of the Ca²⁺ ion in the OEC is tightly linked to understanding the mechanism of photosynthetic water splitting.