Читать книгу Solar-to-Chemical Conversion - Группа авторов - Страница 35

Light harvesting, excitation energy transfer, and charge separation are functions shared by all types of photosynthetic organisms. What is special about oxygenic photosynthesis is the use of water as the ultimate electron donor by PS‐II. Water oxidation takes place at an active site, the OEC, that harbors an inorganic oxo‐bridged Mn₄Ca cluster. The cluster is readily assembled from Mn²⁺ and Ca²⁺ in solution through a process known as photoassembly [66, 117]. The OEC is successively oxidized by the Y_Z tyrosyl radical (D1‐Tyr161), storing up to four oxidizing equivalents before releasing dioxygen. The storage/catalytic cycle of the OEC is described by the Kok–Joliot cycle of S_i states (i = 0–4), with S₀ being the lowest oxidation state of the OEC and S₄ the highest state that evolves dioxygen (Figure 3.8). Among those states S₁ is the resting state of PS‐II, i.e. the state to which the enzyme reverts if left in the dark. Y_Z and the Mn cluster are in close spatial and electronic contact. Along each S_i → S_i+1 transition, the intermediates S_iY_Z^• formed when P680^•+ oxidizes Y_Z have a finite lifetime at low temperature and can be studied by electron paramagnetic resonance (EPR) spectroscopy [86, 118–125]. Studies of these intermediates provide information about the tyrosyl radical itself, its interaction with the manganese cluster, the spin state of the cluster, and changes in hydrogen bonding and protonation occurring during the S‐state transition. Storing the four oxidizing equivalents before performing the four‐electron water oxidation provides a low‐energy pathway for oxidation of water to dioxygen and avoids formation of dangerous reactive intermediates that would result from partial oxidation of the substrate. An important feature of the OEC is that electrons and protons are removed in an alternate fashion along the S‐state cycle [126–128]. This creates a redox‐leveling effect, which means that the four successive oxidative steps can take place within a narrow range of potential.

Figure 3.8 The cycle of intermediate oxidation states of the oxygen‐evolving complex. Water is assumed to bind at the S₃ state of the cluster and upon reconstitution of the S₀ state. The S₄ state is a postulated but unobserved transient intermediate that decays spontaneously to S₀ with release of dioxygen.

The geometric structure of the OEC has been the subject of speculation for a very long time [129, 130], ever since EPR studies on the S₂ state in 1981 established the presence of four antiferromagnetically interacting Mn ions giving rise to a multiline g = 2 EPR signal arising from a cluster with total spin state S = 1/2 [131]. Extended X‐ray absorption fine structure (EXAFS) studies provided increasingly detailed and accurate information about the metal–metal distances within the cluster over the next decades [130, 132–139], but no unique three‐dimensional reconstruction of this information could be achieved without additional input from crystallography [140, 141]. The appearance of the first XRD structure of PS‐II in 2001 [47] and the development of crystallographic models over the following years culminated in an atomic‐resolution model of the OEC core in 2011 [55]. A particular challenge for crystallography was the control of X‐ray radiation damage that led to reduction of the Mn ions [136, 142, 143] and compromised the quality and reliability of structural information contained in the fitted structural models [144–146]. This problem was addressed to large extent [147, 148] by the use of XFEL approaches [57], although certain structural details remain debatable [56, 149–151].

Our present view of the OEC cluster in the dark‐stable S₁ state is depicted in Figure 3.9. It is an asymmetric Mn₄CaO₅ cluster, where three Mn ions and a Ca ion form a Mn₃CaO₄ cubane unit, while a fourth Mn ion is attached externally to this cubane both by coordination to an oxo bridge of the cubane and by a fifth oxo bridge (note that the protonation state of the bridges cannot be inferred from crystallographic models). The inorganic core is mostly ligated by carboxylates provided by the D1 and CP43 proteins: D1‐Asp170, D1‐Glu189, D1‐Glu333, D1‐Asp342, CP43‐Glu354, and the C‐terminal D1‐Ala344. There is a single nitrogen‐donor ligand, D1‐His332, coordinated to Mn1. Four water‐derived ligands, i.e. H₂O or OH, are identified in the crystallographic models; two of them are attached to Mn4 (W1 and W2), and two are attached to calcium (W3 and W4).

Figure 3.9 The Mn₄CaO₅ cluster and its protein pocket in the dark‐stable S₁ state as revealed by protein crystallography (PDB ID: 3WU2, a), and a scheme showing the commonly used labeling of the ions comprising the inorganic core.

The second coordination sphere of the OEC contains the redox‐active tyrosine and its hydrogen‐bonded histidine partner D1‐His190. The tyrosine hydrogen‐bonds directly with one of the Ca‐bound waters and hence is in close interaction with the cluster. Additional residues such as D1‐His337, CP43‐Arg357, D1‐Asp61, and D2‐Lys317, as well as a functionally required chloride ion, interact with the inorganic core and its ligands mostly via hydrogen bonds. These residues play important roles in regulating properties of the cluster and its ligands [152–157], such as the magnetic interaction between specific pairs of Mn ions and local pK_a values of various groups, and may influence or directly participate in proton translocation. An important additional aspect of the local environment of the OEC is the system of water channels and hydrogen‐bonding networks that surround it. These channels and networks are crucial for connecting the active site of water oxidation to the solvent‐exposed surface of the protein and play critical roles in substrate delivery, proton transfer, and product release [158–173].

There is a strong but complex connection between the geometric and electronic structure of the OEC. This connection is key for deciphering the structure of the other S‐states and, eventually, for understanding the mechanism of biological water oxidation [7]. In the following, some of the currently most well‐supported ideas about the geometric and electronic structure of the other S‐states will be presented, with the caveat that there exist significant open questions and ambiguities about many of the specifics [7, 8].

A central question concerns the oxidation states of the Mn ions, their distribution within the cluster, and how they change along the S_i–S_i+1 transitions. Important information on the electronic structure of the cluster can be obtained from magnetic resonance methods, as well as from XAS and XES [128, 174–183]. Structural interpretations of such data can in turn be achieved by spectroscopy‐oriented quantum chemical methods [146, 150, 179, 184–193] that have been extensively benchmarked for high‐valent manganese systems [150, 179, 186, 194–198] and additionally incorporate geometric information from EXAFS and crystallographic models. The dominant view is that the Mn oxidation states evolve from Mn(III)₃Mn(IV) in the S₀ state to Mn(III)₂Mn(IV)₂ in S₁, Mn(III)Mn(IV)₃ in S₂, and Mn(IV)₄ in S₃. This assignment is called the “high oxidation state scheme” [199] as opposed to the low oxidation state hypothesis [200–205] that assigns two more electrons to the Mn ions, with oxidation states ranging from Mn(III)₃Mn(II) in S₀ to Mn(III)₂Mn(IV)₂ in S₃. EPR spectroscopy has helped to identify spin states of all observable intermediates (S = 1/2 for S₀ [206–213], S = 0 for S₁ with a low‐lying S = 1 state [212–217], two forms of S₂ with S = 1/2 and S ≥ 5/2 [218–225], and S = 3 for S₃ [226–228]) but cannot uniquely assign absolute oxidation states. ⁵⁵Mn electron nuclear double resonance (ENDOR) studies of the S₂ state first demonstrated the “3+1” Y‐shaped configuration of the cluster [176, 229], while ⁵⁵Mn hyperfine coupling parameters supported the Mn(III)Mn(IV)₃ oxidation state assignment for S₂ [176, 230, 231] and confirmed the absence of Mn(II) in the S₀ state [230, 232]. Electron–electron double resonance (ELDOR) detected nuclear magnetic resonance experiments (EDNMR) of the S₃ state [228] demonstrated that all Mn ions are similar and isotropic, consistent with the Mn(IV)₄ oxidation state assignment. X‐ray absorption and emission spectroscopies broadly agree with the results of magnetic resonance spectroscopies regarding oxidation states and localization of oxidation events. X‐ray absorption near‐edge spectroscopy (XANES) shows a shift of the Mn K‐edge to higher energies with each S‐state transition, consistent with successive Mn‐based oxidation [181, 233, 234] and a change in coordination in the S₂ → S₃ transition [235]. XES studies observe changes in the Kβ′, Kβ_1,3, and Kα lines. Recent time‐resolved studies of Kβ_1,3 emission spectra of the OEC at room temperature that included comparisons with reference compounds support the high oxidation state assignment described above as well as Mn(III)–Mn(IV) oxidation in the S₂ → S₃ transition [182], while room‐temperature Kα XES studies that similarly compared data on the OEC with those on synthetic compounds in different oxidation states confirmed that the OEC reaches the Mn(IV)₄ oxidation level in the S₃ state [183]. The same assignment of oxidation states is supported by independent studies of photoactivation of PS‐II microcrystals, which measure the number of flash‐driven electron removals required for assembly of an active manganese cofactor from Mn(II) and the Mn‐free enzyme [236].

The direct interpretation of crystallographic models, supported by quantum chemical calculations, indicates that in the S₁ state the terminal Mn1 and Mn4 ions are present as Mn(III), with their Jahn–Teller axes aligned almost collinearly along O5. The precise protonation of the model has not been definitively assigned, with the protonation state of O5 (O²⁻ or OH⁻) and W2 (H₂O or OH⁻) remaining uncertain [145, 146, 150, 151, 237, 238]. The possibility of crystallographically unresolved structural heterogeneity in the S₁ state is also discussed [149, 239–241], which would not be unlikely given the spectroscopic heterogeneity reported both in the S₁ state and in the S₁Y_Z^• intermediate [118, 119, 242–246]. The preceding S₀ state has one more Mn(III) ion compared with S₁, and this has been assigned to Mn3, making Mn2 the only Mn(IV) ion of the cluster in S₀. The most likely protonation state assignment involves a hydroxy for O5, provided this bridge is unprotonated in S₁ [150, 247, 248], while a protonated O4 bridge in S₀ [249] is less likely according to spectroscopy [248].

A widely accepted structural/electronic model for the S₂ state posits the presence of two valence (redox) isomers, i.e. two geometrically similar forms with different distribution of oxidation states among the Mn ions [188, 250, 251] (Figure 3.10). This is based on quantum chemical calculations of exchange coupling constants, spin states, and ⁵⁵Mn hyperfine coupling parameters that first proposed explicit connections between modified crystallographic models and electronic structure data from magnetic resonance spectroscopies [188]. The two valence isomers differ in the position of the unique Mn(III) ion of the S₂ state, either at Mn1 (“open cubane” isomer S₂^A) or at Mn4 (“closed cubane” isomer, S₂^B). The different valence distribution has two important consequences: (i) the connectivity within the cluster is slightly different, as the central O5 bridge is more tightly bound to the Mn(IV) ion in each case rather than to the Mn(III) that exhibits a clear Jahn–Teller elongation axis in the O5 direction, and (ii) the exchange coupling topology is different in each isomer, resulting in different total spin states and related spectroscopic properties. Thus, the S₂^A isomer has a spin S = 1/2 ground state, whereas the S₂^B isomer has a spin S = 5/2 ground state. These correspond exactly to the two observed EPR signals of the S₂ state at g = 2 and g = 4.1, respectively. Additionally, the computed ⁵⁵Mn hyperfine coupling constants for the S₂^A model agree very well with the experimental constants measured for the S = 1/2 signal [146, 188]. Finally, the two quantum chemically derived valence isomers are almost isoenergetic and interconvertible over a low barrier, in direct analogy with the two EPR signals being interconvertible and having a small energy separation [252]. Although other possibilities are discussed in the literature [253, 254], no other interpretation satisfies all of the above experimental constraints.

Figure 3.10 Proposed models for the inorganic core in the S₂ state of the OEC, with the first coordination sphere mostly omitted for clarity. The different magnetic topologies of the two valence isomers, as expressed through the pairwise exchange coupling constants J_ij (values shown in cm⁻¹; J < 0 is antiferromagnetic coupling), lead to different total spin states and g values for the corresponding EPR signals.

The S₂ → S₃ transition is the most complex among the observable transitions of the catalytic cycle and the subject of active research. The nature of the S₃ state itself remains contentious [255], particularly after recent XFEL models of PS‐II were interpreted in terms of mutually incompatible valence states and structural forms [58, 61, 63]. A plausible scenario that is well supported by quantum chemical calculations on realistic models of the OEC and is maximally consistent with spectroscopic data on the electronic structure of the cluster is presented in Figure 3.11. It suggests that the presence of two valence isomers in the S₂ state plays a functional role as part of a gating mechanism [7, 191, 256]. The essential features of this mechanism are as follows: (i) formation of the tyrosyl radical, i.e. the S₂Y_Z^• intermediate, causes a reorientation of the dipole moment of the OEC toward Asp61 [257], which can act as proton acceptor [51, 152, 181, 258–260]; (ii) deprotonation of the cluster is required for the OEC to progress past the S₂Y_Z^• form [191]; and (iii) the deprotonated S₂^A isomer cannot progress to the S₃ state, but the deprotonated S₂^B is predicted to be so unstable in the presence of Y_Z^• that the cluster spontaneously reduces the tyrosyl radical before completion of a Mn1 → Mn4 O5 bridge shift to yield an all‐Mn(IV) S₃ state species, S₃^B [191]. This has an unusual five‐coordinate Mn(IV) ion at Mn4 that correlates with the high total spin of S = 6, the unusually high local zero‐field splitting of the five‐coordinate Mn(IV) ion [261], and the ability to absorb in the NIR [191]. These properties can explain a whole list of observations regarding the S₃ state that would otherwise be incomprehensible. These include the presence of different populations that give or do not give signals in the EPR [228, 242] and that absorb or do not absorb in the NIR [262–265]. This species can subsequently bind water either internally via the calcium ion [250] or externally [191, 266] through a channel associated with methanol and ammonia interaction with the OEC [157, 169, 170, 267–270] to give rise to additional isomers, all assignable to the S₃ state and all featuring four Mn(IV) ions [191, 255, 271, 272].

Figure 3.11 The S₂ → S₃ transition according to Retegan et al. [191] and possible isovalent Mn(IV)₄ components of the S₃ state; the superscript “W” indicates binding of an additional water ligand.

Alternative ideas for the S₃ state include formation of an oxyl radical as opposed to Mn‐centered oxidation [273, 274] and onset of O—O bond formation as a peroxo or superoxo unit [58, 275–277]. These ideas are consistent to some extent with at least one of the available XFEL crystallographic models of the S₃ state, but not with the bulk of spectroscopic information that requires Mn‐based oxidation in the S₂ → S₃ transition [255] or with the most widely accepted interpretations of substrate exchange kinetics [278–280].

The uncertainty about the composition and nature of the S₃ state translates into uncertainty about the nature of the subsequent steps that remain experimentally unresolved and include formation of the active species after the final light‐driven oxidation, formation of the O—O bond, release of dioxygen, and reconstitution of the S₀ state. A well‐known radical‐based mechanism for O—O bond formation has been proposed by Siegbahn and is based on structure S₃^A,W of Figure 3.11 [281–288]. It involves ligand‐based oxidation in the S₄ state to yield a terminal oxyl radical at Mn1 that couples with the oxo bridge that connects Mn3, Mn4, and Ca to form the O—O bond (Figure 3.12a). The same type of oxyl–oxo coupling can take place starting with the S₃^B,W isomer. An alternative to the above mechanism was proposed by Shoji et al. [290] and also assumes S₃^A,W to be the active species, but involves initiation of O—O bond formation at the S₃Y_Z^• intermediate coupled with intramolecular proton transfer. This circumvents the need to invoke an actual S₄ intermediate because the O—O bond is created with concomitant Mn(IV) reduction to Mn(III) before the tyrosyl radical of the S₃Y_Z^• intermediate is reduced by the inorganic core [290]. It is noted that formation of a terminal oxyl radical [291–293] in the hypothetical S₄ state is connected to the high‐spin octahedral Mn(IV) ion in the S₃ state, since formation of a genuine Mn(V) ion is unfavorable in such ligand field [289]. However, if water binding is not required for advancement of the OEC from the S₃ to the S₄ state, then the S₃^B model of Figure 3.11 can also be a candidate for catalytic progression. In this case, Krewald et al. [289] demonstrated that Mn4 remains five‐coordinate and forms a genuine Mn4(V)‐oxo species, with two unpaired electrons localized on the high‐spin Mn4(V) ion and no spin density on the equatorial oxo group [289]. This allows O—O bond formation to occur via genuine nucleophilic coupling that might occur synchronously with water binding to Mn4 (Figure 3.12b) [289], while the formation of three Jahn–Teller axes pointing simultaneously toward the newly formed O₂ unit would contribute to its irreversible expulsion from the active site. The mechanism of Krewald et al. provides access to thermodynamically favorable even‐electron water oxidation [294, 295], potentially to a genuine single‐step 4‐electron transformation, and entirely avoids formation of potentially harmful radical intermediates [289]. Finally, it is worth mentioning an idea proposed by Zhang and Sun, as yet unsupported by quantum chemical calculations, according to which a redox isomerization in the highest state of the cycle could create a cluster with a highly oxidized Mn(VII) center and two terminal oxo groups that would couple to yield dioxygen [296]. Detailed discussions of these and other alternative hypotheses for the mechanism of biological O—O bond formation are available in recent literature [7, 255, 274–277, 287, 288, 290, 297–301].

Figure 3.12 Two selected scenarios for the nature of the S₄ state and O—O bond formation from the computational literature: (a) formation of a Mn1(IV)‐oxyl group in the S₄ state is followed by odd‐electron radical oxyl–oxo coupling [285], and (b) formation of a five‐coordinate high‐spin Mn4(V)‐oxo is followed by intramolecular nucleophilic coupling with concerted water binding [289]. Thick lines indicate direction of Jahn–Teller axes of Mn(III) ions.

It should be clear from the above that despite enormous strides, several aspects of the biological system, including its exact atomistic structure and crucial mechanistic details, remain incompletely understood for the later steps of the catalytic cycle. In the effort to better understand the natural water oxidation catalyst, a major target has been the synthesis of molecular mimics that reproduce structural and electronic properties of the OEC [302]. Following a long history in the development of oligonuclear manganese model complexes [303–305], the past decade has witnessed seminal achievements with the synthesis of manganese–calcium clusters that closely mimic the stoichiometry, metal oxidation states, and bonding topology of the OEC [306–318]. Landmark reports by the groups of Agapie [307] and Christou [308] established access to Mn(IV)₃CaO₄ cubanes, whose magnetic properties (ferromagnetic coupling to a total S = 9/2 state) mirror those of the cuboidal subunit of the OEC in specific states [198]. Zhang et al. [314] subsequently achieved the synthesis of a complex with a Mn₄CaO₄ core that reproduces the arrangement of metal ions of the OEC and has oxidation states equivalent to the S₁ state of the OEC, Mn(III)₂Mn(IV)₂. Moreover, the complex can be oxidized and produces spectroscopic signatures similar to those of the natural system [237, 314]. Extensions of this work include variants of the original complex with exchangeable solvent molecules [316].

Molecular biomimetic complexes are indispensable for elucidating structure–property correlations of relevance to the OEC, and they are a valuable source of insight into how specific geometric or electronic features of a polynuclear manganese cluster affect its overall properties and function [319–323]. At the same time, it should be acknowledged that structural mimics of the OEC have not been linked so far to appreciable water oxidizing activity. Water oxidation has been known for heterogeneous manganese oxides [324–328], but as far as molecular systems are concerned, manganese complexes reported to catalyze oxygen evolution are typically not direct mimics of the OEC, while their performance lags far behind noble metal molecular or solid‐state catalysts [324–328]. There is undoubtedly vast unexplored potential for the development of biomimetic manganese‐based molecular water oxidation catalysts. However, our current understanding of the biological system strongly indicates that its catalytic ability is not simply encoded in the structure of the inorganic cluster of the OEC, but depends critically on the protein matrix that both fine‐tunes the properties of the cluster and performs crucial functions in terms of managing proton‐coupled electron transfer and regulating the flow of substrate and product. Therefore, it is conceivable that any small‐molecule mimic of the OEC, although useful as structural and electronic analog of the biological active site, is destined to fail as a practical water oxidation catalyst because it will not be able to reproduce the functionality that is taken care of by the PS‐II enzyme as a whole. Promising approaches that would be arguably more suitable for large‐scale realization of artificial photosynthesis are discussed in subsequent chapters.