A synthetic Mn4Ca-cluster mimicking the oxygen-evolving center of photosynthesis

See allHide authors and affiliations

Science  08 May 2015:
Vol. 348, Issue 6235, pp. 690-693
DOI: 10.1126/science.aaa6550

Mimicking the oxygen evolution center

Making a synthetic analog of plant photosynthesis is a key goal for exploiting solar energy and replacing fossil fuels. Zhang et al. synthesized a manganese-calcium cluster that looks and acts like the oxygen evolution center in photosystem II (see the Perspective by Sun). The mimic structurally resembles the biological complex, with the notable exception of bridging protein ligands and water-binding sites on a dangling Mn atom. Functionally, however, the cluster's metal center readily undergoes four redox transitions, which contribute to splitting water into oxygen. This and other synthetic mimics will pave the way for developing more efficient catalysts for artificial photosynthesis.

Science, this issue p. 690; see also p. 635


Photosynthetic splitting of water into oxygen by plants, algae, and cyanobacteria is catalyzed by the oxygen-evolving center (OEC). Synthetic mimics of the OEC, which is composed of an asymmetric manganese-calcium-oxygen cluster bound to protein groups, may promote insight into the structural and chemical determinants of biological water oxidation and lead to development of superior catalysts for artificial photosynthesis. We synthesized a Mn4Ca-cluster similar to the native OEC in both the metal-oxygen core and the binding protein groups. Like the native OEC, the synthetic cluster can undergo four redox transitions and shows two magnetic resonance signals assignable to redox and structural isomerism. Comparison with previously synthesized Mn3CaO4-cubane clusters suggests that the fourth Mn ion determines redox potentials and magnetic properties of the native OEC.

The oxygen-evolving center (OEC) in photosystem II (PSII) of plants, algae, and cyanobacteria facilitates splitting of water into O2, protons, and electrons (14). Crystallographic structures (58) reveal that the core of the OEC consists of a Mn3CaO4 cubane motif and a “dangler” Mn linked via two bridging oxides, forming a distinct asymmetric Mn4Ca-cluster (Fig. 1A). This cluster is coordinated to four water molecules, one imidazole, and six carboxylate groups of the amino acid residues of the PSII polypeptides (Fig. 1C). The structure of the OEC as well as the oxidation states of the four manganese ions undergo changes during the water-oxidation reaction cycle, or S-state cycle (4, 9, 10). Spectroscopic results and computational chemistry have provided insight in reaction intermediates and mechanisms (4, 916). The lability of the protein-bound OEC prevents electrochemical characterization and controlled modification of its structure and constituents. Only molecular mimics can facilitate direct measurement and major chemical variation of redox potentials or proton binding energies (pKa values); alternative mechanistic pathways can be explored, and spectroscopic signatures of the native OEC and computational approaches can be “calibrated.”

Fig. 1 Crystal structures of the native OEC and the synthetic Mn4Ca complex I.

(A) Mn4CaO5 core of the native OEC (8). (B) Mn4CaO4 core of I. (C) Structure of the native OEC, including ligating protein side-chains and water molecules. (D) Structure of I, including all ligand groups. (E) Synthesis of I. Distances are given in angstroms; Mn, Ca, O, N, and C are shown in purple, green, orange, blue, and yellow, respectively. For clarity, all H atoms and pivalic CH3 groups are omitted.

Numerous multi-manganese complexes have been synthesized as possible mimics of the OEC (1722), some of which contained a Mn3CaO4 cluster resembling closely the cubane part of the native OEC (1921). Investigation of chemical variants proved useful for unravelling determinants of Mn oxidation potentials. Replacement of the calcium of the Mn3CaO4 cubane increased the redox potential of the MnIV3/MnIV2MnIII couple by more than 1 V (Ca2+ ≈ Sr2+ < Zn2+ < Y3+ < Sc3+ < Mn3+) (22). Calcium (20) or silver ions (21) were incorporated in the Mn4 dangling position, but mimicking the natural paragon by proper coordination of a fourth Mn ion was never attained.

We synthesized complex I, [Mn4CaO4(ButCO2)8 (ButCO2H)2(py)] (But, tert-butyl; py, pyridine), at a high yield (~50%) and starting from inexpensive commercial chemicals. Briefly, a precursor of I was synthesized by a reaction of Bun4NMnO4 (Bun, n-butyl), Mn(CH3CO2)2·(H2O)4, and Ca(CH3CO2)2·H2O (molar ratio of 4: 1: 1) in boiling acetonitrile in the presence of an excess of pivalic acid. Complex I was obtained after recrystallization of the precursor in the presence of 2% pyridine in ethyl acetate (Fig. 1E). Crystallographic characterization (Fig. 1, B and D) reveals one Mn3CaO4 cubane linked to a dangler Mn via O5, which is bridging between four metal ions (thus denoted as μ4-O). This is in line with the influential first OEC model deduced from crystallographic data at 3.5 Å resolution (6) and related mechanistic proposals (23). The asymmetric Mn4CaO4-cluster of I is similar to the native OEC regarding both structure of the Mn4Ca-oxo core and the ligating carboxylate groups (table S5). The integration of the Ca ion into the Mn3CaO4 cubane as well as its anchoring by three carboxylates bridging to Mn1, Mn2, and Mn4 is identical, as is a Mn-carboxylate chain topology (Mn1–OCO–Mn2–OCO–Mn3–OCO–Mn4). In the synthetic complex, all metal ions that are pairwise connected by a bridging oxygen are also connected by a bridging carboxylate. The same design principle applies to the biological complex, with the notable exception of the Mn1–Mn4 connectivity. There are further differences: The μ2-O4 atom of the native OEC is replaced by a bridging carboxylate group, and the terminally coordinated water molecules are replaced by more complex groups.

Bond-valence sum (BVS) analysis of I suggests oxidation states of Mn13+, Mn24+, Mn34+, and Mn43+ (Fig. 1B and table S2), in complete analogy to the OEC in its S1 state (Fig. 1A) (8, 10). The Mn−Mn distances of ~2.75 Å (table S3) match x-ray absorption fine-structure (XAFS) (4, 10) and recent x-ray free electron laser (XFEL) (8) results on the S1-state OEC. Because of its close structural similarity, our synthetic model supports that the generally shorter interatomic distances found in the recent XFEL study (8) [shorter than in (7)] are chemically reasonable for a CaMnIII2MnIV2O21N1 complex (21 first-sphere O-ligands in the native OEC, including O-Glu189, versus 22 in I).

One informative difference between I and the native OEC is the coordination of O5, the μ4-O atom bridging between three Mn and a single Ca ion, likely playing a decisive role in photosynthetic water oxidation (7, 8, 12, 14, 15, 24, 25). The Mn3/4–O5 bond lengths (1.85/1.85 Å) are shorter in I than in the native OEC (2.2/2.3 Å). Even more remarkable, in the native OEC the Mn1–O5 distance is so long (2.7 Å) that Mn1 becomes effectively five-coordinated, with an open coordination site that may be essential for coordination of a “substrate” water molecule upon oxidation of Mn4 from the +3 to the +4 state (8, 14, 15). Comparison of I and the OEC clarifies the origin of this prominent structural difference: In the synthetic complex, Mn1 and Mn4 are clamped together by a single bridging carboxylate, whereas in the OEC, they are spaced by ligation of two neighboring protein residues, His332 and Glu333, to Mn1, Mn3 and Mn4. Only the His332-Glu333 dyad breaks the rule that each pair of μ-oxo–bridged metal ions also is μ-carboxylato–bridged. We therefore propose that in the protein, the absence of μ-carboxylato–bridging between Mn1 and Mn4, possibly supported by a spacing effect of the His332-Glu333 dyad, allows for formation of the functionally crucial open coordination site at the Mn1 ion.

The cyclic voltammogram (CV) of I dissolved in 1,2-dichloroethane (Fig. 2) shows that five redox states are accessible. Thus, the synthetic complex can accumulate oxidation equivalents in close analogy to the native OEC (4, 9, 10). Although the native OEC is embedded in a protein environment that is closely mimicked in complex I, there are also water molecules close to the native metal-oxo core, which might render direct comparison of redox potentials problematic. Yet, the voltammograms of I in 1,2-dichloroethane and acetonitrile are similar (fig. S8), suggesting that the redox potentials do not depend strongly on solvent polarity. Addition of trace amounts of water (≤ 1%) to I in acetonitrile could modify the complex through ligand exchange (fig. S12) and results in obscured CV oxidation and reduction waves, possibly relating to a superimposed catalytic wave (figs. S9 and S10).

Fig. 2 S-state cycle of the native OEC and CV of I.

Potentials versus NHE; scan rate of 100 mV/s. Native OEC is in green, peak positions of CV waves are in red, and estimated midpoint potentials are in blue. The likely oxidation states of the four Mn ions in various S-states of the native OEC and plausible oxidation states of I are indicated with black numbers. A CV of I in acetonitrile and further details are provided in fig. S8.

The as-synthesized complex is in the S1-state. The S1→S2 transition of I is detected at a redox potential [~0.8 V versus normal hydrogen electrode (NHE)], which is close to the estimated potential of the corresponding OEC redox transition (≥ 0.9 V) (3, 26) but is remarkably different from that of the previously Mn3CaO4 complex without a dangling Mn ion, indicating that the fourth Mn ion could play a crucial role in tuning the redox potential of the Mn4Ca-cluster (1921). Before and after dissolution in organic solvent, complex I exhibits a broad parallel-mode electron paramagnetic resonance (EPR) signal (fig. S3), which resembles the g value and width of the S1-state signal of the native OEC.

To trap the S2 state of I, we used [Fe(phen)3](PF6)3 (phen, 1,10-phenanthroline) as a chemical oxidant with a redox potential of +1.1 V versus NHE. Two S2-state signals are observed in the perpendicular-mode EPR spectrum (Fig. 3): (i) a multiline EPR signal centered at g value of 2.0, with a width of ~1600 G and more than 20 hyperfine peaks, and (ii) a signal centered at a g value greater than 4, with a width of ~500 G and without resolved hyperfine structure. The two EPR signals observed are similar to the well-known g = 2.0 and g ≥ 4.1 EPR signals, both observed in the S2-state of the native OEC (9, 27, 28). Both EPR signals of I exhibit a linear Curie-plot behavior (fig. S6), suggesting that both are ground-state signals, in full analogy to the native OEC, in which the two EPR signals and their respective spin states (g = 2 signal, S = 1/2; g = 4.1/4.9 signals, S ≥ 3/2) have been proposed to arise from two OEC conformers differing in the localization of the Mn3+ ion in the S2-state (Mn13+Mn44+ versus Mn14+Mn43+) (28).

Fig. 3 EPR spectrum of I in its MnIV3MnIII state (S2-state) at 7K.

The broad multiline signal (>20 lines) centered close to g = 2 as well as EPR signals at higher g values (g > 4) are assignable to the Mn complex in different spin states. For clarity, the signal at g = ~3 range from [Fe(phen)3]3+ (figs. S4 and S7) was replaced with a dashed line.

The possibility of structural conformers potentially affecting the exchange coupling between Mn ions is suggested by subtle structural difference between the two monomers resolved in the crystallographic unit cell of I. Specifically, the nonidentical Mn1-Mn4 and Ca-O5 distances (table S4) may be of relevance. These differences were determined for the crystallographically characterized S1-state cluster. At present, it is still open whether and how these differences are amplified for the dissolved complex or through Mn oxidation in the S1→S2 transition. A similarly close analogy of the magnetic properties has never been observed for cubane-type Mn3CaO4 complexes (1921) nor for any other multinuclear Mn complex. The observation of a low-spin multiline signal in I contrasts the high-spin state of Mn3CaO4 cubane complexes (20, 29), supporting that the fourth Mn ion is crucial regarding the magnetic properties of the heterobimetallic complexes. Complex I and future variants could become valuable model systems in experimental and theoretical investigations aiming at insight in the magnetic properties of the biological complex, in which the spin localization has been suggested to be of relevance in the O-O bond formation chemistry (14).

Each of the four redox transition of I increases the potential for the subsequent oxidation step—for example, from 0.8 V (S1→S2) to 1.25 V (S2→S3) (Fig. 2). In the native OEC, the S1→S2 transition is not coupled to a charge-compensating proton release (11), and its redox potential increases (10) to an unknown level. On the basis of the general similarity of the S1→S2 transition in I and the native OEC, we propose that in both, a potential increase of similar magnitude may occur. This potential increase in the S1→S2 transition can explain why in PSII, a charge-compensating and redox-potential–lowering deprotonation of the OEC—likely coupled to further chemical changes―precedes Mn oxidation in the S2→S3 (13) and S3→S4 transition (30). In the native OEC, redox leveling through deprotonation reduces the overpotential need for accumulation of oxidizing equivalents (1, 10). In the absence of efficient redox leveling, the overpotential need would be prohibitively high, as shown by the behavior of I in organic solvents (Fig. 2 and fig. S8).

Although I is similar in many ways to native OEC, it still deviates in several functionally crucial aspects, including the water-binding sites at Mn4 and the Ca ion. We consider the synthesis of I as a first step for synthesis of further variants that mimic native OEC even closer. The synthetic flexibility is illustrated by related complexes II and III (Fig. 4), which have terminal ligands at the Ca ion or at Mn4 that are exchanged by alternative neutral ligands. In the native OEC, these terminal ligand sites are occupied by water molecules. The binding sites of the readily exchangeable neutral ligands on Ca and the dangler Mn in the described synthetic complexes may serve as potential positions to create water binding sites that resemble closely the respective sites of the native OEC. Last, the complexes described here can serve as references for calculation of vibrational spectra (fig. S13) and absolute OEC redox potentials; the latter is pressingly needed for verification of computational models (12, 14).

Fig. 4 Structures of complexes with exchanged terminal ligands.

The complexes differ from I by ligand exchange either on (A) Ca (II) or (B) Mn4 (III).

Supplementary Materials

Materials and Methods

Figs. S1 to S13

Tables S1 to S8

References (3135)

References and Notes

  1. Acknowledgments: This work was supported by the National Natural Science Foundation of China (nos. 20973186, 31070216, 21076049, and 91427303), Chinese Academy of Sciences (KJCX2-YW-W25), Japan Society for the Promotion of Science (no. 24000018/Shen), and by the Deutsche Forschungsgemeinschaft (CRC 1078, A4/Dau). We thank A. Zouni and W. Junge for their suggestions. Crystallographic data have been deposited at the Cambridge Crystallographic Database Centre, and the deposition numbers are 1042697(I), 1042515 (II), and 1042514 (III).

Stay Connected to Science

Navigate This Article